Kern Sibbald
Traduit de l'anglais par Ludovic Strappazon
13 octobre 2008
Ce manuel documente la version 2.4.3 (10 October 2008)
de Bacula mais est en cours de traduction.
Copyright ©1999-2007, Free Software Foundation Europe e.V.
Permission is granted to copy, distribute and/or modify this document under the terms of the
GNU Free Documentation License, Version 1.2 published by the Free Software Foundation;
with no Invariant Sections, no Front-Cover Texts, and no Back-Cover Texts.
A copy of the license is included in the section entitled "GNU Free Documentation License".
Permission vous est accordée de copier, distribuer et/ou modifier ce document
selon les termes de
la GNU Free Documentation License, Version 1.2 publiée par la Free Software
Foundation,
sans Sections Inaltérables, et sans Textes de Couverture-Avant et sans
Textes de Couverture-Arrière.
Une copie de la licence est incluse dans la section intitulée "GNU Free
Documentation License".
Bacula est un jeu de programmes qui vous permet (ou à l'administrateur système) de faire des sauvegardes, restaurations, et vérifications des données d'un ordinateur sur un réseau hétérogène. Bacula peut fonctionner complètement sur un seul ordinateur. Il est capable de sauvegarder sur des supports variés, y compris disques et cartouches.
En termes techniques, il s'agit d'un programme de sauvegarde Client/Serveur. Bacula est relativement facile d'utilisation et efficace, tout en offant de nombreuses fonctions avancées de gestion de stockage qui facilitent la recherche et la restauration de fichiers perdus ou endommagés. Grâce à sa conception modulaire, Bacula est échelonnable depuis le simple système constitué d'un ordinateur, jusqu'au système de plusieurs centaines d'ordinateurs disséminés sur un vaste réseau.
Si vous utilisez actuellement un programme tel que tar, dump, ou bru pour sauvegarder vos données, et aimeriez une solution réseau, plus de flexibilité, ou les commodités d'un catalogue, Bacula vous procurera certainement les fonctions supplémentaires que vous recherchez. Cependant, si vous avez peu d'expérience des systèmes Unix ou si vous n'avez pas l'expérience d'un système de sauvegarde sophistiqué, nous ne vous recommandons pas l'utilisation de Bacula, car il est beaucoup plus difficile à installer et utiliser que tar ou dump.
Si vous attendez de Bacula qu'il se comporte comme les programmes simples mentionnés ci-dessus et qu'il écrive sur toute cartouche insérée dans le lecteur, vous éprouverez des difficultés à travailler avec Bacula. Bacula est conçu pour protéger vos données en suivant les règles que vous spécifiez, ce qui signifie que la réutilisation d'une cartouche ne se fera qu'en dernier ressort. Il est possible de "contraindre" Bacula à écraser toute cartouche dans le lecteur, mais il est plus facile et plus efficace d'utiliser un outil plus basique pour ce genre d'opérations.
Si vous utilisez Amanda et aimeriez un programme de sauvegarde qui peut écrire sur plusieurs volumes (qui ne soit pas limité par la capacité de vos cartouches), Bacula peut certainement satisfaire vos besoins, d'autant que plusieurs de nos utilisateurs estiment que Bacula est plus simple à installer et utiliser que d'autres programmes équivalents.
Si vous utilisez actuellement un logiciel commercial sophistiqué tel que Legato Networker, ARCserveIT, Arkeia, ou PerfectBackup+, vous pourriez être interessé par Bacula qui fournit la plupart de leurs fonctions et qui est un logiciel libre sous licence GNU version 2.
Bacula est constitué des cinq composants ou services majeurs suivants :
(remerciements à Aristedes Maniatis pour ce schéma et le suivant)
Les trois bases de données actuellement supportées (MySQL, PostgreSQL, ou SQLite) fournissent de nombreuses fonctions telles l'indexation rapide, requêtes arbitraires, et sécurité. Bien que nous prévoyions de supporter d'autres bases de données SQL majeures, l'implémentation actuelle s'interface seulement avec MySQL, PostgreSQL, et SQLite. Pour plus de détails consultez le document sur la conception des Services Catalogue .
Les RPMs pour MySQL et PostgreSQL font partie de la distribution Red Hat. Sinon, il est tout à fait aisé de les construire à partir des sources. Consultez le chapitre Installer et configurer MySQL ou Installer et configurer PostgreSQL de ce document pour plus de détails. Pour plus d'informations sur MySQL et PostgreSQL, consultez www.mysql.com ou www.postgresql.org.
Configurer et construire SQLite est encore plus facile. Pour les détails de configuration de SQLite, consultez le chapitre Installer et Configurer SQLite de ce document.
Pour réaliser avec succès les opérations de sauvegarde et restauration, les quatre services suivants doivent être configurés et lancés : le Director Daemon, le File Daemon, le Storage Daemon et MySQL, PostgreSQL ou SQLite.
Pour que Bacula comprenne votre système, quels clients vous voulez sauvegarder et comment, vous devez créer un certain nombre de fichiers de configuration. La suite brosse un tableau de ces opérations.
Bacula est en constante évolution, par conséquent, ce manuel ne sera pas toujours en accord avec le code. Si un objet de ce manuel est précédé d'un astérisque (*), cela signifie que cette fonctionnalité particulière n'est pas implémentée. S'il est précédé d'un signe plus (+), cela indique que la fonction est peut-être partiellement implémentée.
Si vous lisez la version de ce manuel fournie avec les sources de Bacula, le paragraphe ci-dessus reste vrai. En revanche, si vous lisez la version en ligne : www.bacula.org/manual, veuillez garder à l'esprit que cette version décrit la version courante de développement de Bacula (celle du CVS) qui peut contenir des fonctionnalités qui n'existent pas dans la version "officielle". De même, il est généralement un peu à la traîne derrière le code.
Pour faire fonctionner Bacula rapidement, nous vous recommandons de commencer par parcourir la section Terminologie ci-dessous, de passer rapidement en revue le chapitre suivant intitulé L'état actuel de Bacula, puis le Guide de démarrage rapide de Bacula, qui vous donnera une vue d'ensemble de la mise en oeuvre de Bacula . Après quoi vous devriez poursuivre avec le chapitre sur L'installation de Bacula, puis le chapitre Comment configurer Bacula, et finalement, le chapitre Exécuter Bacula.
Pour faciliter la communication autour de ce projet, nous fournissons ici les définitions de la terminologie que nous utilisons.
La fonction Catalogue est de celles qui distinguent Bacula de simples programmes de sauvegarde et archivage tels que dump et tar.
Verify peut aussi être utilisé pour s'assurer que les données les plus récemment écrites sur un volume sont cohérentes avec ce qui figure dans le catalogue (c-à-d il compare les attributs de fichiers), ou encore, confronter le contenu du volume aux fichiers originaux sur le disque.
La période de rétention des fichiers détermine la durée pendant laquelle les enregistrements concernant les fichiers seront gardés dans le catalogue. Cette période est importante car le volume des enregistrements relatifs aux fichiers occupe, de loin, le plus d'espace dans la base de données. Par conséquent, vous devez vous assurer qu'un "élagage" (NDT : pruning) régulier de ces enregistrements est effectué. (Voir la commande retention de la Console pour plus de détails sur ce sujet).
La période de rétention des jobs est la durée pendant laquelle les enregistrements relatifs aux jobs seront conservés dans le catalogue. Notez que tous les enregistrements relatifs aux fichiers sont attachés aux jobs qui ont sauvegardé ces fichiers. Les enregistrements relatifs aux fichiers peuvent être purgés tout en conservant ceux relatifs aux jobs. Dans ce cas, l'information concernant les jobs exécutés restera disponible, mais pas les détails des fichiers sauvegardés. Normalement, lorsqu'un job est purgé, tous les enregistrements concernant les fichiers qu'il a sauvegardé le sont aussi.
La période de rétention des volumes est la durée minimale de conservation d'un volume avant qu'il ne soit réutilisé. Bacula n'effacera, en principe, jamais un volume qui contient la seule copie de sauvegarde d'un fichier. Dans les conditions idéales, le catalogue maintiendrait les entrées pour tous les fichiers sauvegardés pour tous les volumes courants. Une fois qu'un volume est écrasé, les fichiers qui étaient sauvegardés dessus sont automatiquement effacés du catalogue. Cependant, s'il y a un très gros pool de volumes ou si un volume n'est jamais écrasé, le catalogue pourrait devenir énorme. Pour maintenir le catalogue dans des proportions gérables, les informations de sauvegarde devraient être supprimées après une période de rétention des fichiers définie.
Bacula est un programme de sauvegarde, restauration et vérification, ce n'est pas un système complet de disaster recovery à lui seul, mais il peut en être une partie clef si vous planifiez soigneusement et suivez les instructions incluses dans le chapitre Plan de reprise d'activité avec Bacula de ce manuel.
Avec la planification appropriée, telle que décrite dans le chapitre sur le plan de reprise d'activité, Bacula peut devenir la pièce centrale de votre plan de reprise d'activité. Par exemple, si vous avez créé un(e) disque(tte) boot d'urgence et un(e) disque(tte) de secours Bacula pour sauvegarder les informations de partitionnement courantes de votre disque dur, et maintenu un jeu de sauvegardes complet de votre système, il est possible de reconstruire complètement votre système "depuis le métal brut" (NDT : From bare metal) .
Si vous avez utilisé la directive WriteBootstrap dans votre job ou quelque autre moyen pour sauvegarder un fichier bootstrap valide, vous pourrez l'utiliser pour extraire les fichiers nécessaires (sans utiliser le catalogue et sans chercher manuellement les fichiers à restaurer).
Le diagramme fonctionnel suivant montre les interactions typiques entre les
services Bacula pour un job de type sauvegarde. Chaque bloc représente en
général un processus séparé (normalement un daemon). En
général, le director surveille le flux d'informations. Il maintient aussi
le catalogue.
En d'autres termes, ce qui est et ce qui n'est pas actuellement implémenté et fonctionnel.
Le problème n'existe pas sur les noyaux 2.6.
Même si votre lecteur est dans la liste ci dessous, vérifiez le Chapitre Test des Lecteurs Bandes de ce manuel pour les procédures que vous pouvez utiliser pour vérifier si votre lecteur de bandes est susceptible de fonctionner avec Bacula. Si votre lecteur est en mode bloc fixe, il peut paraître fonctionner avec Bacula jusqu'à ce que vous essayiez de restaurer et que Bacula tente de se positionner sur la bande. Seuls la procédure ci-dessus et vos propres tests peuvent vous garantir un fonctionnement correct.
Il est très difficile de fournir une liste de lecteurs de bandes supportés, ou de lecteurs qui sont connus pour fonctionner avec Bacula en raison du peu de retours de la part des usagers. (par conséquent, si vous utilisez Bacula sur un lecteur qui ne figure pas dans la liste, merci de nous le faire savoir). Selon les informations provenant de nos utilisateurs, les lecteurs suivants sont connus pour fonctionner avec Bacula. Un trait d'union dans une colonne signifie "inconnu" :
OS | Fabr. | Media | Modèle | Capacité |
- | ADIC | DLT | Adic Scalar 100 DLT | 100GB |
- | ADIC | DLT | Adic Fastor 22 DLT | - |
- | - | DDS | Compaq DDS 2,3,4 | - |
- | Exabyte | - | Lecteurs Exabyte de moins de dix ans | - |
- | Exabyte | - | Exabyte VXA drives | - |
- | HP | Travan 4 | Colorado T4000S | - |
- | HP | DLT | HP DLT drives | - |
- | HP | LTO | HP LTO Ultrium drives | - |
FreeBSD 4.10 RELEASE | HP | DAT | HP StorageWorks DAT72i | - |
- | Overland | LTO | LoaderXpress LTO | - |
- | Overland | - | Neo2000 | - |
- | OnStream | - | OnStream drives (see below) | - |
- | Quantum | DLT | DLT-8000 | 40/80GB |
Linux | Seagate | DDS-4 | Scorpio 40 | 20/40GB |
FreeBSD 4.9 STABLE | Seagate | DDS-4 | STA2401LW | 20/40GB |
FreeBSD 5.2.1 pthreads patched RELEASE | Seagate | AIT-1 | STA1701W | 35/70GB |
Linux | Sony | DDS-2,3,4 | - | 4-40GB |
Linux | Tandberg | - | Tandbert MLR3 | - |
FreeBSD | Tandberg | - | Tandberg SLR6 | - |
Solaris | Tandberg | - | Tandberg SLR75 | - |
Linux Gentoo | ADIC | - | IBM Ultrium LTO I | 100/200 Go |
Une liste des Librarires supportées figure dans le chapitre librairies (autochangers) de ce document, où vous trouverez d'autres lecteurs de bandes qui fonctionnent avec Bacula.
Auparavant les lecteurs de bandes OnStream IDE-SCSI ne fonctionnaient pas avec Bacula. A partir de la version 1.33 de Bacula et de la version 0.9.14 du pilote noyau ou supérieur,ce lecteur est supporté. Consultez le chapitre de test car vous devez le configurer pour fonctionner en mode blocs de taille fixe.
Les lecteurs QIC sont connus pour avoir nombre de particularités (taille de blocs fixe, et un EOF plutôt que deux pour terminer la bande). En conséquence, vous devrez être particulièrement attentif à sa configuration pour le faire fonctionner avec Bacula.
A moins que vous n'ayez appliqué un correctif sur la bibliothèque pthreads de votre système FreeBSD, vous perdrez des données quand Bacula aura rempli une bande et passera à la suivante. La raison en est que les bibliothèques pthreads sans correctifs échouent à retourner un état d'alerte à Bacula signalant que la fin de bande est proche. Consultez le chapitre test des lecteurs de bandes de ce manuel pour d'importantes informations concernant la configuration de votre lecteur de bande pour qu'il soit compatible avec Bacula.
Pour des informations sur les libraries (autochangeurs) supportées, allez voir la section Libraries connues pour fonctionner avec Bacula du chapitre Librairies supportées de ce manuel.
Ci-dessous, vous trouverez une table de spécifications de cartouches DLT et LTO qui vous permettra de vous faire une idée de la capacité et de la rapidité des lecteurs et cartouches modernes. Les capacités reportées ici sont les natives, sans compression. Tous les lecteurs modernes pratiquent la compression matérielle, et les fabricants affichent souvent une capacité compressée avec un ratio de 2:1. Le facteur de compression réel dépend principalement des données sauvegardées, mais je pense qu'un ratio 1,5:1 est beaucoup plus raisonnable (autrement dit, multipliez la valeur de la table par 1,5 pour obtenir une estimation grossière de la capacité compressée). Les taux de transfert sont arrondis au GB/hr le plus proche. Les valeurs sont fournies par les divers fabricants.
Le type de media est la désignation du fabricant, vous n'êtes pas obligé de l'utiliser (mais vous devriez...) dans vos ressources de configuration Bacula.
Type de media | Type de lecteur | Capacité des media | Taux de transfert |
DDS-1 | DAT | 2 GB | ?? GB/hr |
DDS-2 | DAT | 4 GB | ?? GB/hr |
DDS-3 | DAT | 12 GB | 5.4 GB/hr |
Travan 40 | Travan | 20 GB | ?? GB/hr |
DDS-4 | DAT | 20 GB | 11 GB/hr |
VXA-1 | Exabyte | 33 GB | 11 GB/hr |
DAT-72 | DAT | 36 GB | 13 GB/hr |
DLT IV | DLT8000 | 40 GB | 22 GB/hr |
VXA-2 | Exabyte | 80 GB | 22 GB/hr |
Half-high Ultrum 1 | LTO 1 | 100 GB | 27 GB/hr |
Ultrium 1 | LTO 1 | 100 GB | 54 GB/hr |
Super DLT 1 | SDLT 220 | 110 GB | 40 GB/hr |
VXA-3 | Exabyte | 160 GB | 43 GB/hr |
Super DLT I | SDLT 320 | 160 GB | 58 GB/hr |
Ultrium 2 | LTO 2 | 200 GB | 108 GB/hr |
Super DLT II | SDLT 600 | 300 GB | 127 GB/hr |
VXA-4 | Exabyte | 320 GB | 86 GB/hr |
Ultrium 3 | LTO 3 | 400 GB | 216 GB/hr |
Si vous êtes comme moi, vous voulez faire fonctionner Bacula immédiatement pour en avoir un aperçu, puis, plus tard, vous reviendrez en arrière pour lire et connaître tous les détails. C'est exactement ce que ce chapitre se propose d'accomplir : vous faire avancer rapidement sans tous les détails. Si vous voulez sauter la section sur les Pools, Volumes et Labels, vous pourrez toujours y revenir, mais s'il vous plaît, veuillez lire ce chapitre jusqu'à la fin, et en particulier suivre les instructions pour tester votre lecteur de bandes.
Nous supposons que vous êtes parvenus à construire et installer Bacula, sinon, vous devriez d'abord jeter un oeil aux Prérequis (système) puis au chapitre Compiler et installer Bacula de ce manuel.
Afin de rendre Bacula aussi flexible que possible, les directives lui sont données en plusieurs endroits. L'instruction principale est la ressource Job, qui définit un job. Un job de type sauvegarde consiste en général en un FileSet, un client, un Schedule pour un ou plusieurs niveaux ou horaires de sauvegardes, un Pool, ainsi que des instructions additionnelles. Un autre point de vue est de considérer le FileSet comme "Que sauvegarder ?", le Client comme "Qui sauvegarder ?", le Schedule comme "Quand sauvegarder ?" et le Pool comme "Où sauvegarder ?" (c'est à dire, "Sur quel volume ?)
Typiquement, une combinaison FileSet/Client aura un job correspondant. La plupart des directives, telles que les FileSets, Pools, Schedules, peuvent être mélangées et partagées entre les jobs. Ainsi, vous pouvez avoir deux définitions (ressources) de jobs sauvegardant différents serveurs et utilisant les mêmes Schedule, FileSet (sauvegardant donc les mêmes répertoires sur les deux machines) et peut-être même les mêmes Pools. Le Schedule définira quel type de sauvegarde sera exécuté et à quel moment (par exemple les Full le mercredi, les incrémentales le reste de la semaine), et lorsque plus d'un job utilise le même Schedule la priorité du job détermine lequel démarre en premier. Si vous avez de nombreux jobs, vous devriez utiliser la directive JobDefs, qui vous permet de définir des paramètres par défaut pour vos jobs, qui peuvent être changés au sein de la ressource Job, mais qui vous évitent de réécrire les mêmes paramètres pour chaque job. En plus des FileSets que vous voulez sauvegarder, vous devriez aussi avoir un job qui sauvegarde votre catalogue.
Enfin, sachez qu'en plus des jobs de type Backup, vous pouvez aussi utiliser des jobs de type restore, verify, admin, qui ont chacun des exigences différentes.
Si vous avez utilisé un programme tel que tar pour sauvegarder votre système, les notions de Pools, Volumes et labels peuvent vous sembler un peu confuses au premier abord. Un Volume est un simple support physique (cartouche, ou simple fichier) sur lequel Bacula écrit vos données de sauvegarde. Les Pools regroupent les Volumes de sorte qu'une sauvegarde n'est pas restreinte à la capacité d'un unique Volume. Par conséquent, plutôt que de nommer explicitement les Volumes dans votre Job, vous spécifiez un Pool, et Bacula se chargera de sélectionner le prochain Volume utilisable du Pool et vous demandera de le monter.
Bien que les options de base soient spécifiées dans la ressource Pool du fichier de configuration du Director, le Pool réel est géré par le Catalogue Bacula. Il contient les informations de la ressource Pool (bacula-dir.conf) ainsi que les informations concernant tous les Volumes qui ont été ajoutés au Pool. L'ajout de Volumes se fait en principe manuellement depuis la Console grâce à la commande label.
Pour chaque Volume, Bacula gère une quantité d'informations considérable telles que les première et dernière date et heure d'écriture, le nombre de fichiers sur le Volume, le nombre de bytes sur le Volume, le nombre de montages, etc.
Pour que Bacula puisse lire ou écrire sur un Volume physique, celui-ci doit avoir reçu un étiquettage logiciel afin que Bacula soit assuré que le bon Volume est monté. Ceci s'effectue en principe manuellement depuis la Console grâce à la commande label.
Les étapes de création de Pool, ajout de Volumes à ce Pool, et écriture d'étiquettes logicielles sur les Volumes, peuvent sembler pénibles au premier abord, mais en fait, elles sont tout à fait simples à franchir, et elles vous permettent d'utiliser plusieurs Volumes (plutôt que d'être limité à la capacité d'un seul). Les Pools vous procurent aussi une flexibilité importante pour votre politique de sauvegarde. Par exemple, vous pouvez avoir un Pool de Volumes "Daily" pour vos sauvegardes Incrémentales et un Pool de Volumes "Weekly" pour vos sauvegardes complètes (Full). En spécifiant le Pool approprié dans les Jobs de sauvegarde quotidiens et hebdomadaires, vous garantissez qu'aucun Job quotidien n'écrira jamais sur un Volume du Pool réservé aux sauvegardes hebdomadaire et vice versa, et Bacula vous dira quelle cartouche est requise, et quand.
Pour plus de détails concernant les Pools, consultez la section Ressource Pool du chapitre sur la configuration du Director, ou poursuivez votre lecture, nous reviendrons plus tard sur ce sujet.
Après avoir exécuté la commande ./configure ad hoc, make et make install, si c'est la première fois que vous exécutez
Bacula, vous devez créer des fichiers de configuration valides pour le
Director, le File Daemon, le Storage Daemon et le programme Console. Si vous
avez suivi nos recommandations, des fichiers de configuration par défaut
ainsi que les binaires des daemons seront situés dans votre
répertoire d'installation. Dans tous les cas les binaires se trouvent dans
le répertoire que vous avez spécifié au niveau de l'option --
sbindir de la commande ./configure, et les fichiers de configuration
se trouvent dans le répertoire que vous avez spécifié au niveau de
l'option --
sysconfdir.
Lors des paramétrages initiaux de Bacula, il vous faudra investir un peu de temps pour modifier les fichiers de configuration par défaut afin de les adapter à votre environnement. Ceci peut nécessiter de redémarrer Bacula plusieurs fois jusqu'à ce que tout fonctionne correctement. Ne cédez pas au désespoir. Une fois que vous aurez créé vos fichiers de configuration, vous n'aurez que rarement besoin de les changer et de redémarrer Bacula. Le gros du travail consistera à changer la cartouche quand elle est pleine.
Le programme console est utilisé par l'administrateur pour interagir avec le Director et pour arrêter et démarrer manuellement des jobs, ou encore pour obtenir des informations sur les jobs en cours d'exécution ou programmés.
Le fichier de configuration de la Console se trouve dans le répertoire que
vous avez spécifié au niveau de l'option --
sysconfdir de la commande
./configure et par défaut se nomme console.conf.
Si vous avez choisi de construire la Console GNOME avec l'option --
enable-gnome, vous y trouverez également son fichier de configuration par
défaut, nommé gnome-console.conf.
Il en va de même pour la console wxWidgets, qui est construite par l'option
--
enable-bwx-console, et le nom du fichier de configuration par défaut
est, dans ce cas, bwx-console.conf.
Normalement, pour les nouveaux utilisateurs, aucune modification n'est requise pour ces fichiers. Les réglages par défaut sont raisonnables.
Le programme Monitor est typiquement une icône dans la barre des tâches. Cependant, lorsque l'icône est étendue en une fenètre, l'administrateur ou l'utilisateur peut obtenir des informations concernant le Director ou l'état des sauvegardes sur la machine locale ou n'importe quelle autre daemon Bacula configuré.
L'image montre le tray-monitor configuré pour trois daemons. En cliquant sur les boutons radio dans le coin en haut à gauche de l'image, vous pouvez voir l'état de chacun des daemons. L'image montre l'état du Storage Daemon (MainSD) sélectionné.
Le fichier de configuration du Monitor se trouve dans le répertoire
spécifié au niveau de l'option --
sysconfdir de la commande ./configure et se nomme par défaut tray-monitor.conf. En principe,
pour les nouveaux utilisateurs, il suffit de changer les permissions de ce
fichier pour permettre aux utilisateurs non-root d'exécuter le Monitor, en
effet cette application doit être exécuté par le même utilisateur que
l'environnement graphique (n'oubliez pas de donner aux non-root le droit
d'exécuter bacula-tray-monitor). Ceci ne constitue pas une faille de
sécurité tant que vous utilisez les réglages par défaut.
Le File Daemon, est le programme qui s'exécute sur chaque machine cliente. A la demande du Director, il détermine les fichiers à sauvegarder et les expédie au Storage Daemon.
Le fichier de configuration du File Daemon se trouve dans le répertoire
spécifié au niveau de l'option --
sysconfdir de la commande ./configure et se nomme par défaut bacula-fd.conf. Normalement, pour
les nouveaux utilisateurs, aucune modification n'est requise pour ce fichier.
Les réglages par défaut sont raisonnables. Cependant, si vous envisagez de
sauvegarder plus d'une machine, il vous faudra installer le File Daemon avec
un fichier de configuration spécifique sur chaque machine à sauvegarder.
Les informations concernant chaque File Daemon doivent apparaître dans le
fichier de configuration du Director.
Le director est le programme central qui contrôle tous les autres daemons. Il planifie et surveille les jobs à exécuter.
Le fichier de configuration du Director se trouve dans le répertoire
spécifié au niveau de l'option --
sysconfdir de la commande ./configure et se nomme par défaut bacula-dir.conf.
En général, la seule modification nécessaire consiste à faire en sorte que la directive Include de la Ressource FileSet contienne au moins une ligne avec un nom de fichier ou de répertoire valide à sauvegarder.
Si vous ne possédez pas de lecteur DLT, vous voudrez probablement modifier la ressource Storage pour donner un nom plus représentatif de votre périphérique de stockage. Vous pouvez toujours utiliser les noms existants puisque vous êtes libre de les assigner arbitrairement, mais ils doivent s'accorder avec les noms correspondants dans le fichier de configuration du Storage Daemon.
Vous pouvez aussi changer l'adresse électronique pour les notifications vers votre propre adresse e-mail plutôt que vers celle de root (configuration par défaut).
Enfin, si vous avez plusieurs systèmes à sauvegarder, il vous faudra spécifier un File Daemon (ou client) pour chaque système sauvegardé, précisant ses nom, adresse et mot de passe. Nous estimons que baptiser vos daemons du nom de vos systèmes suffixés avec -fd aide beaucoup à corriger les erreurs. Ainsi, si votre système est foobaz, vous nommerez le daemon foobaz-fd. Pour le Director, vous pourriez utiliser foobaz-dir, et foobaz-sd pour le Storage Daemon. Chacun de vos composants de Bacula doit avoir un nom unique Si vous les nommez tous à l'identique, en plus de ne jamais savoir quel daemon envoie quel message, s'ils partagent le même répertoire de travail (working directory), les noms de fichiers temporaires des daemons ne seront pas uniques et vous aurez d'étranges erreurs.
Le Storage Daemon est responsable, sur demande du Director, de la réception des données en provenance des File Daemons, et de leur écriture sur le medium de stockage, ou, dans le cas d'une restauration, de trouver les données pour les envoyer vers le File Daemon.
Le fichier de configuration du Storage Daemon se trouve dans le répertoire
spécifié au niveau de l'option --
sysconfdir de la commande ./configure et se nomme par défaut bacula-sd.conf. Modifiez ce
fichier pour accorder les noms de périphériques de stockage à ceux que
vous possédez. Si le processus d'installation a convenablement détecté
votre système, elles seront déjà correctement réglées. Ces
ressources de stockage "Name" et "Media Type" doivent être les mêmes
que leurs correspondantes du fichier de configuration du Director bacula-dir.conf. Si vous souhaitez sauvegarder vers un fichier plutôt que
sur des bandes, la ressource Device doit pointer vers un répertoire où des
fichiers seront créés en guise de Volumes lorque vous étiquetterez
(label) vos Volumes.
Vous pouvez tester la validité syntaxique de vos fichiers de configuration, afficher tout message d'erreur et terminer. Par exemple, en supposant que vous avez installé vos binaires et fichiers de configuration dans le même répertoire,
cd <installation-directory> ./bacula-dir -t -c bacula-dir.conf ./bacula-fd -t -c bacula-fd.conf ./bacula-sd -t -c bacula-sd.conf ./bconsole -t -c bconsole.conf ./gnome-console -t -c gnome-console.conf ./bwx-console -t -c wx-console.conf su <normal user> -c "./bacula-tray-monitor -t -c tray-monitor.conf"
testera le fichier de configuration de chacun des principaux programmes. Si le fichier de configuration est correct, le programme se termine sans rien afficher. Veuillez noter que selon les options de configuration que vous avez choisies, il se peut qu'aucune des commandes ci-dessus ne soit valable sur votre système. Si vous avez installé les binaires dans les répertoires traditionnels d'Unix plutôt que dans un simple répertoire, il vous faudra modifier les commandes ci-dessus en conséquence (pas de "./" devant les commandes, et un chemin devant les fichiers de configuration).
Avant de gaspiller votre temps avec Bacula pour finalement constater qu'il ne fonctionne pas avec votre lecteur de bandes, veuillez s'il vous plaît lire le chapitre btape -- Tester votre lecteur de bandes de ce manuel. Si vous possédez un lecteur standard SCSI moderne sur un Linux ou un Solaris, fort probablement, il fonctionnera, mais mieux vaut tester que d'être déçu. Pour FreeBSD (et probablement les autres xBSD), la lecture du chapitre mentionné ci-dessus est un devoir. Pour FreeBSD, consultez aussi The FreeBSD Diary pour une description détaillée de la méthode pour faire fonctionner Bacula sur votre système. De plus, les utilisateurs de versions de FreeBSD antérieures à 4.9-STABLE datée du lundi 29 décembre 2003 15:18:01 UTC qui prévoient d'utiliser des lecteurs de bandes sont invités à lire le fichier platforms/freebsd/pthreads-fix.txt du répertoire principal de Bacula, qui contient d'importantes informations sur la compatibilité de Bacula avec leur système.
La nouvelle librairie pthreads /lib/tls installée par défaut sur les systèmes Red Hat récents (kernels 2.4.x) est défectueuse. Vous devez la supprimer ou la renommer, puis rebooter votre système avant d'exécuter Bacula, faute de quoi, après environ une semaine de fonctionnement, Bacula se bloquera pour de longues périodes, voire définitivement. Veuillez consulter le chapitre Systèmes supportés pour plus d'informations sur ce problème.
Ce problème n'existe plus avec les noyaux 2.6.
La partie la plus importante de l'exécution de Bacula est probablement la capacité de restaurer les fichiers. Si vous n'avez pas essayé de récupérer des fichiers au moins une fois, vous subirez une bien plus forte pression le jour où vous devrez réellement le faire, et serez enclin à commettre des erreurs que vous n'auriez pas commises si vous aviez déjà essayé.
Pour avoir rapidement une bonne idée de la façon d'utiliser Bacula, nous vous recommandons fortement de suivre les exemples du chapitre exécuter Bacula de ce manuel, où vous trouverez des informations détaillées sur l'exécution de Bacula.
Si vous utilisez le bacula-dir.conf par défaut ou une variante, vous constaterez qu'il récupère toutes les sorties de Bacula dans un fichier. Pour éviter que ce fichier ne croisse sans limites, nous vous recommandons de copier le fichier logrotate depuis scripts/logrotate vers /etc/logrotate.d/bacula. Ainsi les fichiers de logs subiront une rotation mensuelle et seront conservés pour une durée maximum de cinq mois. Vous pouvez éditer ce fichier pour adapter la rotation à votre convenance.
Si vous avez l'intention d'utiliser Bacula en tant qu'outil de disaster recovery plutôt que comme un simple programme pour restaurer les fichiers perdus, vous serez intéressé par le chapitre Plan de reprise d'activité avec Bacula de ce manuel.
De toute façon, vous êtes fortement invité à tester soigneusement la restauration de quelques fichiers que vous aurez préalablement sauvegardés, plutôt que d'attendre qu'un désastre ne frappe. Ainsi, vous serez préparé.
En général, il vous faudra les sources de la version courante de Bacula, et si vous souhaitez exécuter un client Windows, vous aurez besoin de la version binaire du client Bacula pour Windows. Par ailleurs, Bacula a besoin de certains paquetages externes (tels SQLite, MySQL ou PostgreSQL) pour compiler correctement en accord avec les options que vous aurez choisies. Pour vous simplifier la tâche, nous avons combiné plusieurs de ces programmes dans deux paquetages depkgs (paquetages de dépendances). Ceci peut vous simplifier la vie en vous fournissant tous les paquets nécessaires plutôt que de vous contraindre à les trouver sur la Toile, les charger et installer.
Si vous faites une mise à jour de Bacula, vous devriez d'abord lire attentivement les ReleaseNotes de toutes les versions entre votre version installée et celle vers laquelle vous souhaitez mettre à jour. Si la base de données du catalogue a été mise à jour (c'est presque toujours le cas Ć chaque nouvelle version majeure), vous devrez soit réinitialiser votre base de données et repartir de zéro, soit en sauvegarder une copie au format ASCII avant de procéder à sa mise à jour. Ceci est normalement fait lorsque Bacula est compilé et installé par :
cd <installed-scripts-dir> (default /etc/bacula) ./update_bacula_tables
Ce script de mise à jour peut aussi être trouvé dans le répertoire src/cats des sources de Bacula.
S'il y a eu plusieurs mises à jour de la base de données entre votre version et celle vers laquelle vous souhaitez évoluer, il faudra appliquer chaque script de mise à jour de base de données. Vous pouvez trouver tous les anciens scripts de mise à jour dans le répertoire upgradedb des sources de Bacula. Il vous faudra éditer ces scripts pour qu'ils correspondent à votre configuration. Le script final, s'il y en a un, sera dans le répertoire src/cats comme indiqué dans la ReleaseNote.
Si vous migrez d'une version majeure vers une autre, vous devrez remplacer tous vos composants (daemons) en même temps car, généralement, le protocole inter-daemons aura changé. Par contre, entre deux versions mineures d'une même majeure (par exemple les versions 1.32.x), à moins d'un bug, le protocole inter-daemons ne changera pas. Si cela vous semble confus, lisez simplement les ReleaseNotes très attentivement, elles signaleront si les daemons doivent être mis à jour simultanément.
Enfin, notez qu'il n'est généralement pas nécessaire d'utiliser make uninstall avant de procéder à une mise à jour. En fait, si vous le faites vous effacerez probablement vos fichiers de configuration, ce qui pourrait être désastreux. La procédure normale de mise à jour est simplement :
./configure (your options) make make install
En principe, aucun de vos fichiers .conf ou .sql ne devrait être écrasé, et vous devez exĆ©cuter les deux commandes make et make install. make install sans un make prĆ©alable ne fonctionnera pas.
Pour plus d'informations sur les mises à jour, veuillez consulter la partie Upgrading Bacula Versions du chapitre Astuces de ce manuel
Comme nous l'évoquions plus haut, nous avons combiné une série de programmes dont Bacula peut avoir besoin dans les paquets depkgs et depkgs1. Vous pouvez, bien sur, obtenir les paquets les plus récents directement des auteurs. Le fichier README dans chaque paquet indique où les trouver. Pourtant, il faut noter que nous avons testé la compatibilité des paquets contenus dans les fichiers depkgs avec Bacula.
Vous pouvez, bien sur, obtenir les dernieres versions de ces paquetages de leurs auteurs. Les références nécessaires figurent dans le README de chaque paquet. Quoi qu'il en soit, soyez conscient du fait que nous avons testé la compatibilité des paquetages des fichiers depkgs.
Typiquement, un paquetage de dépendances sera nommé depkgs-ddMMMyy.tar.gz et depkgs1-ddMMMyy.tar.gz où dd est le jour où n'ous l'avons publié, MMM l'abbréviation du mois et yy l'année. Par exemple: depkgs-07Apr02.tar.gz. Pour installer et construire ce paquetage (s'il est requis), vous devez:
La composition exacte des paquetages de dépendance est susceptible de changer de temps en temps, voici sa composition actuelle :
Paquets externes | depkgs | depkgs1 |
SQLite | X | - |
SQLite3 | X | - |
mtx | X | - |
readline | - | X |
pthreads | - | - |
zlib | - | - |
wxWidgets | - | - |
Notez que certains de ces paquets sont de taille respectable, si bien que l'étape de compilation peut prendre un certain temps. Les instructions ci-dessous construiront tous les paquets contenus dans le répertoire. Cependant, la compilation de Bacula, ne prendra que les morceaux dont Bacula a effectivement besoin.
Une alternative consiste à ne construire que les paquets nécessaires. Par exemple,
cd bacula/depkgs make sqlite
configurera et construira SQLite et seulement SQLite.
Vous devriez construire les paquets requis parmi depkgs et/ou depkgs1 avant de configurer et compiler Bacula car Bacula en aura besoin dès la compilation.
Même si vous n'utilisez pas SQLite, vous pourriez trouver le paquet depkgs pratique pour construire mtx car le programme tapeinfo qui vient avec peut souvent vous fournir de précieuses informations sur vos lecteurs de bandes SCSI (e.g. compression, taille min/max des blocks,...).
Le paquet depkgs-win32 est obsolète à partir de la version 1.39 de Bacula. Il était autrefois utilisé pour compiler le client natif Win32 qui est désormais construit sur Linux grâce à un mécanisme de compilation croisée. Tous les outils et librairies tierces sont automatiquement téléchargées par l'exécution de scripts apropriés. Lisez le fichier src/win32/README.mingw32 pour plus de détails.
Veuillez consulter la section Systèmes supportés du chapitre Démarrer avec Bacula de ce manuel.
L'installation basique est plutôt simple.
Notez que si vous avez dejà MySQL ou PostgreSQL sur votre système vous pouvez sauter cette phase pourvu que vous ayez construit "the thread safe libraries'' et que vous ayez déjà installé les rpms additionnels sus-mentionnés.
make distclean
afin d'être certain de repartir de zéro et d'éviter d'avoir un mélange avec vos premières options. C'est nécessaire parce que ./configure met en cache une bonne partie des informations. make distclean est aussi recommandé si vous déplacez vos fichiers source d'une machine à une autre. Si make distclean échoue, ignorez-le et continuez.
Si vous obtenez des erreurs durant le linking dans le répertoire du
Storage Daemon (/etc/stored), c'est probablement parce que vous avez chargé
la librairie statique sur votre système. J'ai remarqué ce problème sur
un Solaris. Pour le corriger, assurez-vous de ne pas avoir ajouté l'option
--
enable-static-tools à la commande ./configure.
Si vous ignorez cette étape (make) et poursuivez immédiatement avec make install, vous commettez deux erreurs sérieuses : d'abord, votre installation va échouer car Bacula a besoin d'un make avant un make install ; ensuite, vous vous privez de la possibilité de vous assurer qu'il n'y a aucune erreur avant de commencer à écrire les fichiers dans vos répertoires système.
make uninstall make distclean ./configure (vos-nouvelles-options) make make install
Si tout se passe bien, ./configure déterminera correctement votre système et configurera correctement le code source. Actuellement, FreeBSD, Linux (RedHat), et Solaris sont supportés. Des utilisateurs rapportent que le client Bacula fonctionne sur MacOS X 10.3 tant que le support readline est désactivé.
Si vous installez Bacula sur plusieurs systèmes identiques, vous pouvez simplement transférer le répertoire des sources vers ces autres systèmes et faire un "make install''. Cependant s'il y a des différences dans les librairies, ou les versions de systèmes, ou si vous voulez installer sur un système différent, vous devriez recommencer à partir de l'archive tar compressée originale. Si vous transférez un répertoire de sources où vous avez déjà exécuté la commande ./configure, vous DEVEZ faire:
make distclean
avant d'exécuter à nouveau ./configure. Ceci est rendu nécessaire par l'outil GNU autoconf qui met la configuration en cache, de sorte que si vous réutilisez la configuration d'une machine Linux sur un Solaris, vous pouvez être certain que votre compilation échouera. Pour l'éviter, comme mentionné plus haut, recommencez depuis l'archive tar, ou faites un "make distclean''.
En général, vous voudrez probablement sophistiquer votre configure pour vous assurer que tous les modules que vous souhaitez soient construits et que tout soit placé dans les bons répertoires.
Par exemple, sur Fedora, RedHat ou SuSE, on pourrait utiliser ceci:
CFLAGS="-g -Wall" \ ./configure \ --sbindir=$HOME/bacula/bin \ --sysconfdir=$HOME/bacula/bin \ --with-pid-dir=$HOME/bacula/bin/working \ --with-subsys-dir=$HOME/bacula/bin/working \ --with-mysql \ --with-working-dir=$HOME/bacula/bin/working \ --with-dump-email=$USER
Notez: l'avantage de cette configuration pour commencer, est que tout sera mis dans un seul répertoire, que vous pourrez ensuite supprimer une fois que vous aurez exécuté les exemples du prochain chapitre, et appris comment fonctionne Bacula. De plus, ceci peut être installé et exécuté sans être root.
Pour le confort des développeurs, j'ai ajouté un script defaultconfig au répertoire examples. Il contient les réglages que vous devriez normalement utiliser, et chaque développeur/utilisateur devrait le modifier pour l'accorder à ses besoins. Vous trouverez d'autres exemples dans ce répertoire.
Les options --
enable-conio ou --
enable-readline sont utiles car
elles confèrent un historique de lignes de commandes et des capacités
d'édition à la Console. Si vous avez inclus l'une ou l'autre option, l'un
des deux paquets termcap ou ncurses sera nécessaire pour
compiler. Sur la plupart des systèmes, y compris RedHat et SuSE, vous
devriez inclure le paquet ncurses. Si Le processus de configuration de
Bacula le détecte, il l'utilisera plutôt que la librairie termcap.
Sur certains systèmes, tels que SUSE, la librairie termcap n'est
pas dans le répertoire standard des librairies par conséquent, l'option
devrait être désactivée ou vous aurez un message tel que:
/usr/lib/gcc-lib/i586-suse-linux/3.3.1/.../ld: cannot find -ltermcap collect2: ld returned 1 exit status
lors de la compilation de la Console Bacula. Dans ce cas, il vous faudra placer la variable d'environnement LDFLAGS avant de compiler.
export LDFLAGS="-L/usr/lib/termcap"
Les mêmes contraintes de librairies s'appliquent si vous souhaitez utiliser les sous-programmes readlines pour l'édition des lignes de commande et l'historique, ou si vous utilisez une librairie MySQL qui requiert le chiffrement. Dans ce dernier cas, vous pouvez exporter les librairies additionnelles comme indiqué ci-dessus ou, alternativement, les inclure directement en paramètres de la commande ./configure comme ci-dessous :
LDFLAGS="-lssl -lcyrpto" \ ./configure <vos-options>
Veuillez noter que sur certains systèmes tels que Mandriva, readline tend
à "avaler'' l'invite de commandes, ce qui le rend totalement inutile. Si
cela vous arrive, utilisez l'option "disable'', ou si vous utilisez une
version postérieure à 1.33 essayez --
enable-conio pour utiliser une
alternative à readline intégrée. Il vous faudra tout de même termcap
ou ncurses, mais il est peu probable que le paquetage conio gobe vos
invites de commandes.
Readline n'est plus supporté depuis la version 1.34. Le code reste disponible, et si des utilisateurs soumettent des patches, je serai heureux de les appliquer. Cependant, étant donné que chaque version de readline semble incompatible avec les précédentes, et qu'il y a des différences significatives entre les systèmes, je ne puis plus me permettre de le supporter.
Avant de construire Bacula, vous devez décider si vous voulez utiliser SQLite, MySQL ou PostgreSQL. Si vous n'avez pas déjà MySQL ou PostgreSQL sur votre machine, nous vous recommandons de démarrer avec SQLite. Ceci vous facilitera beaucoup l'installation car SQLite est compilé dans Bacula et ne requiert aucune administration. SQLite fonctionne bien et sied bien aux petites et moyennes configurations (maximum 10-20 machines). Cependant, il nous faut signaler que plusieurs utilisateurs ont subi des corruptions inexpliquées de leur catalogue SQLite. C'est pourquoi nous recommandons de choisir MySQL ou PostgreSQL pour une utilisation en production.
Si vous souhaitez utiliser MySQL pour votre catalogue Bacula, consultez le chapitre Installer et Configurer MySQL de ce manuel. Vous devrez installer MySQL avant de poursuivre avec la configuration de Bacula. MySQL est une base de données de haute qualité très efficace et qui convient pour des configurations de toutes tailles. MySQL est légèrement plus complexe à installer et administrer que SQLite en raison de ses nombreuses fonctions sophistiquées telles que userids et mots de passe. MySQL fonctionne en tant que processus distinct, est réellement une solution professionnelle et peut prendre en charge des bases de données de dimension quelconque.
Si vous souhaitez utiliser PostgreSQL pour votre catalogue Bacula, consultez le chapitre Installer et Configurer PostgreSQL de ce manuel. Vous devrez installer PostgreSQL avant de poursuivre avec la configuration de Bacula. PostgreSQL est très similaire à MySQL bien que tendant à être un peu plus conforme à SQL92. PostgreSQL possède beaucoup plus de fonctions avancées telles que les transactions, les procédures stockées, etc. PostgreSQL requiert une certaine connaissance pour son installation et sa maintenance.
Si vous souhaitez utiliser SQLite pour votre catalogue Bacula, consultez le chapitre Installer et Configurer SQLite de ce manuel.
Il y a de nombreuses options et d'importantes considérations données ci-dessous que vous pouvez passer pour le moment si vous n'avez eu aucun problème lors de la compilation de Bacula avec une configuration simplifiée comme celles montrées plus haut.
Si le processus ./configure ne parvient pas à trouver les librairies spécifiques (par exemple libintl), assurez vous que le paquetage approprié est installé sur votre système. S'il est installé dans un répertoire non standard (au moins pour Bacula), il existe dans la plupart des cas une option parmi celles énumérées ci-dessous (ou avec "./configure --help") qui vous permettra de spécifier un répertoire de recherche. D'autres options vous permettent de désactiver certaines fonctionnalités (par exemple --disable-nls).
Si vous souhaitez vous jeter à l'eau, nous vous conseillons de passer directement au chapitre suivant, et d'exécuter le jeu d'exemples. Il vous apprendra beaucoup sur Bacula, et un Bacula de test peut être installé dans un unique répertoire (pour une destruction aisée) et exécuté sans être root. Revenez lire les détails de ce chapitre si vous avez un quelconque problème avec les exemples, ou lorsque vous voudrez effectuer une installation réelle.
TAQUET MISE A JOUR
Les options en ligne de commande suivantes sont disponibles pour configure afin d'adapter votre installation à vos besoins.
Par défaut, Bacula installe une simple page de manuel dans /usr/share/man. Si vous voulez qu'elle soit installée ailleurs, utilisez cette options pour spécifier le chemin voulu. Notez que les principaux documents Bacula en HTML et PDF sont dans une archive tar distincte des sources de distribution de Bacula.
--
disable-static-tools.
--
enable-client-only décrite plus loin est aussi intéressante
pour compiler un simple client sans les autres parties du programme.
Pour lier un binaire statique, l'éditeur de liens a besoin des versions statiques de toutes les librairies utilisées, aussi les utilisateurs rencontrent fréquemment des erreurs d'édition de liens à l'utilisation de cette option. La première chose à faire est de s'assurer d'avoir la librairie glibc statiquement liée sur votre système. Ensuite, il faut s'assurer de ne pas utiliser les options --openssl ou --with-python de la commande configure, car elle requierent des librairies supplémentaires. Vous devriez pouvoir activer ces options, mais il vous faudra charger les librairies statiques additionnelles correspondantes.
Pour lier un binaire statique, l'éditeur de liens a besoin des versions statiques de toutes les librairies utilisées, aussi les utilisateurs rencontrent fréquemment des erreurs d'édition de liens à l'utilisation de cette option. La première chose à faire est de s'assurer d'avoir la librairie glibc statiquement liée sur votre système. Ensuite, il faut s'assurer de ne pas utiliser les options --openssl ou --with-python de la commande configure, car elle requierent des librairies supplémentaires. Vous devriez pouvoir activer ces options, mais il vous faudra charger les librairies statiques additionnelles correspondantes.
Pour lier un binaire statique, l'éditeur de liens a besoin des versions statiques de toutes les librairies utilisées, aussi les utilisateurs rencontrent fréquemment des erreurs d'édition de liens à l'utilisation de cette option. La première chose à faire est de s'assurer d'avoir la librairie glibc statiquement liée sur votre système. Ensuite, il faut s'assurer de ne pas utiliser les options --openssl ou --with-python de la commande configure, car elle requierent des librairies supplémentaires. Vous devriez pouvoir activer ces options, mais il vous faudra charger les librairies statiques additionnelles correspondantes.
Pour lier un binaire statique, l'éditeur de liens a besoin des versions statiques de toutes les librairies utilisées, aussi les utilisateurs rencontrent fréquemment des erreurs d'édition de liens à l'utilisation de cette option. La première chose à faire est de s'assurer d'avoir la librairie glibc statiquement liée sur votre système. Ensuite, il faut s'assurer de ne pas utiliser les options --openssl ou --with-python de la commande configure, car elle requierent des librairies supplémentaires. Vous devriez pouvoir activer ces options, mais il vous faudra charger les librairies statiques additionnelles correspondantes.
Pour lier un binaire statique, l'éditeur de liens a besoin des versions statiques de toutes les librairies utilisées, aussi les utilisateurs rencontrent fréquemment des erreurs d'édition de liens à l'utilisation de cette option. La première chose à faire est de s'assurer d'avoir la librairie glibc statiquement liée sur votre système. Ensuite, il faut s'assurer de ne pas utiliser les options --openssl ou --with-python de la commande configure, car elle requierent des librairies supplémentaires. Vous devriez pouvoir activer ces options, mais il vous faudra charger les librairies statiques additionnelles correspondantes.
--
disable-largefile.
Voyez aussi la note ci-dessous, après le paragraphe --with-postgreSQL
Voyez aussi la note ci-dessous, après le paragraphe --with-postgreSQL
Voyez aussi la note ci-dessous, après le paragraphe --with-postgreSQL
Notez que pour que Bacula soit configuré correctement, vous devez spécifier l'une des quatre options de bases de données supportées : --with-sqlite, --with-sqlite3, --with-mysql, ou --with-postgresql, faute de quoi ./configure échouera.
--
with-readline n'est pas renseignée, readline sera désactivé. Cette
option affecte la compilation de Bacula. Readline fournit le programme
Console avec un historique des lignes de commandes et des capacités
d'édition. Readline n'est désormais plus supporté, ce qui signifie que
vous l'utilisez à vos risques et périls
Pour plus d'informations sur la configuration et les tests de TCP wrappers, consultez la section Configurer et Tester TCP Wrappers du chapitre sur la sécurité.
Sur SuSE, les librairies libwrappers requises pour lier Bacula appartiennent au paquet tcpd-devel. Sur RedHat, le paquet se nomme tcp_wrappers.
--
with-baseport permet d'assigner automatiquement trois ports consécutifs
à partir du port de base spécifié. Vous pouvez aussi changer les
numéros de ports dans les fichiers de configuration. Cependant, vous devez
prendre garde à ce que les numéros de ports se correspondent fidèlement
dans chacun des trois fichiers de configuration. Le port de base par défaut
est 9101, ce qui assigne les ports 9101 à 9103. Ces ports (9101, 9102 et
9103) ont été officiellement assigné à Bacula par l'IANA. Cette
option n'est utilisée que pour modifier les fichiers de configuration de
Bacula. Vous pouvez à tout moment faire cette modification en éditant
directement ces fichiers.
Notez: de nombreuses options supplémentaires vous sont présentées
lorsque vous entrez ./configure --
help, mais elles ne sont pas
implémentées.
Pour la plupart des systèmes, nous recommandons de commencer avec les options suivantes :
./configure \ --enable-smartalloc \ --sbindir=$HOME/bacula/bin \ --sysconfdir=$HOME/bacula/bin \ --with-pid-dir=$HOME/bacula/bin/working \ --with-subsys-dir=$HOME/bacula/bin/working \ --with-mysql=$HOME/mysql \ --with-working-dir=$HOME/bacula/working
Si vous souhaitez installer Bacula dans un répertoire d'installation
plutôt que de l'exécuter depuis le répertoire de compilation, (comme le
feront les développeurs la plupart du temps), vous devriez aussi inclure les
options --
sbindir et --
sysconfdir avec les chemins appropriés. Aucune n'est
nécessaire si vous ne vous servez pas de "make install'', comme c'est le
cas pour la plupart des travaux de développement. Le processus
d'installation va créer les répertoires sbindir et sysconfdir s'ils
n'existent pas, mais il ne créera pas les répertoires pid-dir, subsys-dir
ni working-dir, aussi assurez vous qu'ils existent avant de lancer Bacula.
L'exemple ci-dessous montre la façon de procéder de Kern.
Avec SQLite:
CFLAGS="-g -Wall" ./configure \ --sbindir=$HOME/bacula/bin \ --sysconfdir=$HOME/bacula/bin \ --enable-smartalloc \ --with-sqlite=$HOME/bacula/depkgs/sqlite \ --with-working-dir=$HOME/bacula/working \ --with-pid-dir=$HOME/bacula/bin/working \ --with-subsys-dir=$HOME/bacula/bin/working \ --enable-gnome \ --enable-conio
ou
CFLAGS="-g -Wall" ./configure \ --sbindir=$HOME/bacula/bin \ --sysconfdir=$HOME/bacula/bin \ --enable-smartalloc \ --with-mysql=$HOME/mysql \ --with-working-dir=$HOME/bacula/working --with-pid-dir=$HOME/bacula/bin/working \ --with-subsys-dir=$HOME/bacula/bin/working --enable-gnome \ --enable-conio
ou une installation RedHat complètement traditionnelle :
CFLAGS="-g -Wall" ./configure \ --prefix=/usr \ --sbindir=/usr/sbin \ --sysconfdir=/etc/bacula \ --with-scriptdir=/etc/bacula \ --enable-smartalloc \ --enable-gnome \ --with-mysql \ --with-working-dir=/var/bacula \ --with-pid-dir=$HOME/var/run \ --enable-conio
Notez que Bacula suppose que les répertoires /var/bacula, /var/run et /var/lock/subsys existent, ils ne seront pas crées par le processus d'installation.
D'autre part, avec gcc 4.0.1 20050727 (Red Hat 4.0.1-5) sur processeur AMD64 et sous CentOS4 64 bits, un bug du compilateur génère du code erroné qui conduit Bacula à des erreurs de segmentation. Typiquement, vous le rencontrerez d'abord avec le Storage Daemon. La solution consiste à s'assurer que Bacula est compilé sans optimisation (normalement -O2)
Pour installer Bacula depuis les sources, il vous faudra les paquetages suivants sur votre système (ils ne sont pas installés par défaut) : libiconv, gcc 3.3.2, stdc++, libgcc ( pour les librairies stdc++ and gcc_s ), make 3.8 ou plus récent.
Il vous faudra probablement aussi ajouter /usr/local/bin et /usr/css/bin à PATH pour ar.
#!/bin/sh CFLAGS="-g" ./configure \ --sbindir=$HOME/bacula/bin \ --sysconfdir=$HOME/bacula/bin \ --with-mysql=$HOME/mysql \ --enable-smartalloc \ --with-pid-dir=$HOME/bacula/bin/working \ --with-subsys-dir=$HOME/bacula/bin/working \ --with-working-dir=$HOME/bacula/working
Comme mentionné ci-dessus, le processus d'installation va créer les répertoires sbindir et sysconfdir s'ils n'existent pas, mais il ne créera pas les répertoires pid-dir, subsys-dir ni working-dir, aussi assurez vous qu'ils existent avant de lancer Bacula.
Notez que vous pouvez aussi avoir besoin des paquetages suivants pour installer Bacula depuis les sources :
SUNWbinutils, SUNWarc, SUNWhea, SUNWGcc, SUNWGnutls SUNWGnutls-devel SUNWGmake SUNWgccruntime SUNWlibgcrypt SUNWzlib SUNWzlibs SUNWbinutilsS SUNWGmakeS SUNWlibm export PATH=/usr/bin::/usr/ccs/bin:/etc:/usr/openwin/bin:/usr/local/bin:/usr/sfw/bin:/opt/sfw/bin:/usr/ucb:/usr/sbin
Veuillez consulter: The FreeBSD Diary pour une description détaillée de la méthode pour faire fonctionner Bacula sur votre système. De plus, les utilisateurs de versions de FreeBSD antérieures à la 4.9-STABLE du lundi 29 décembre 2003 15:18:01 qui envisagent d'utiliser des lecteurs de bandes doivent consulter le chapitre Tester son lecteur de bandes de ce manuel pour d'importantes informations sur la configuration des lecteurs pour qu'ils soient compatibles avec Bacula.
Si vous utilisez Bacula avec MySQL, vous devriez prendre garde à compiler MySQL avec les threads natifs de FreeBSD plutôt qu'avec ceux de Linux, car c'est avec ceux là qu'est compilé Bacula et le mélange des deux ne fonctionnera probablement pas.
Pour installer la version binaire Win32 du File Daemon, consultez le chapitre Installation sur systèmes Win32 de ce document.
A partir de la version 1.34, Bacula n'utilise plus CYGWIN pour le client Win32. Il est cependant encore compilé sous un environnement CYGWIN -- Bien que vous puissiez probablement le faire avec seulement VC Studio. Si vous souhaitez compiler le client Win32 depuis les sources, il vous faudra Microsoft C++ version 6.0 ou supérieur. Dans les versions de Bacula antérieures à la 1.33, CYGWIN était utilisé.
Notez qu'en dépit du fait que la plupart des éléments de Bacula puissent compiler sur les systèmes Windows, la seule partie que nous avons testée et utilisée est le File Daemon.
Finalement, vous devriez suivre les instructions d'installation de la section Win32 Installation sur systèmes Win32 de ce document en occultant la partie qui décrit la décompression de la version binaire.
Voici le script que j'utilise pour compiler sur mes machines Linux de "production'':
#!/bin/sh # This is Kern's configure script for Bacula CFLAGS="-g -Wall" \ ./configure \ --sbindir=$HOME/bacula/bin \ --sysconfdir=$HOME/bacula/bin \ --enable-smartalloc \ --enable-gnome \ --with-pid-dir=$HOME/bacula/bin/working \ --with-subsys-dir=$HOME/bacula/bin/working \ --with-mysql=$HOME/mysql \ --with-working-dir=$HOME/bacula/bin/working \ --with-dump-email=$USER \ --with-smtp-host=mail.your-site.com \ --with-baseport=9101 exit 0
Notez que je fixe le port de base à 9101, ce qui signifie que Bacula
utilisera le port 9101 pour la console Director, le port 9102 pour le File
Daemon, et le 9103 pour le Storage Daemon. Ces ports devraient être
disponibles sur tous les systèmes étant donné qu'ils ont été
officiellement attribués à Bacula par l'IANA (Internet Assigned Numbers
Authority). Nous recommandons fortement de n'utiliser que ces ports pour
éviter tout conflit avec d'autres programmes. Ceci est en fait la
configuration par défaut si vous n'utilisez pas l'option --
with-baseport.
Vous pouvez aussi insérer les entrées suivantes dans votre fichier /etc/services de façon à rendre les connections de Bacula plus aisées à repérer (i.e. netstat -a):
bacula-dir 9101/tcp bacula-fd 9102/tcp bacula-sd 9103/tcp
Avant de personnaliser vos fichiers de configuration, vous voudrez installer Bacula dans son répertoire définitif. tapez simplement:
make install
Si vous avez précédemment installé Bacula, les anciens binaires seront écrasés, mais les anciens fichiers de configuration resteront inchangés, et les "nouveaux'' recevront l'extension .new. Généralement, si vous avez déjà installé et exécuté Bacula, vous préfèrerez supprimer ou ignorer les fichiers de configuration avec l'extension .new
Si vous exécutez le Director et le Storage Daemon sur une machine et si vous voulez sauvegarder une autre machine, vous devez avoir un File Daemon sur cette machine. Si la machine et le système sont identiques, vous pouvez simplement copier le binaire du File Daemon bacula-fd ainsi que son fichier de configuration bacula-fd.conf, puis modifier le nom et le mot de passe dans bacula-fd.conf de façon à rendre ce fichier unique. Veillez à faire les modifications correspondantes dans le fichier de configuration du Director (bacula-dir.conf).
Si les architectures, les systèmes, ou les versions de systèmes diffèrent, il vous faudra compiler un File Daemon sur la machine cliente. Pour ce faire, vous pouvez utiliser la même commande ./configure que celle utilisée pour construire le programme principal, soit en partant d'une copie fraiche du répertoire des sources, soit en utilisant make distclean avant de lancer ./configure.
Le File Daemon n'ayant pas d'accès au catalogue, vous pouvez supprimer les
option --
with-mysql ou --
with-sqlite. Ajoutez l'option --
enable-client-only. Ceci va compiler seulement les librairies et programmes
clients, et donc éviter d'avoir à installer telle ou telle base de
données. Lancez make avec cette configuration, et seul le client sera
compilé.
Si vous souhaitez que vos daemons soient lancés automatiquement au démarrage de votre système (une bonne idée !), une étape supplémentaire est requise. D'abord, le processus ./configure doit reconnaître votre système -- ce qui signifie que ce doit être une plate-forme supportée et non inconnue, puis vous devez installer les fichiers dépendants de la plate-forme comme suit :
(devenez root) make install-autostart
Notez que la fonction d'autodémarrage n'est implémentée que pour les systèmes que nous supportons officiellement (actuellement FreeBSD, RedHat Linux, et Solaris), et n'a été pleinement testée que sur RedHat Linux.
make install-autostart installe les scripts de démarrage apropriés ainsi que les liens symboliques nécessaires. Sur RedHat Linux, Ces scripts résident dans /etc/rc.d/init.d/bacula-dir /etc/rc.d/init.d/bacula-fd, et /etc/rc.d/init.d/bacula-sd. Toutefois, leur localisation exacte dépend de votre système d'exploitation.
Si vous n'installez que le File Daemon, tapez:
make install-autostart-fd
Pour recompiler tout exécutable, tapez
make
dans le répertoire correspondant.. Afin d'éliminer tous les objets et binaires (y compris les fichiers temporaires nommés 1,2 ou 3 qu'utilise Kern), tapez
make clean
Pour un nettoyage exhaustif en vue de distribution, entrez:
make distclean
Notez que cette commande supprime les Makefiles. Elle est en principe lancée depuis la racine du répertoire des sources pour les préparer à la distribution. Pour revenir de cet état, vous devez réexécuter la commande ./configure à la racine des sources puisque tous les Makefiles ont été détruits.
Pour ajouter un nouveau fichier dans un sous-répertoire, éditez Makefile.in dans ce sous-répertoire, puis faites un simple make. Dans la plupart des cas, le make reconstruira le Makefile à partir du nouveau Makefile.in. Dans certains cas, il peut être nécessaire d'exécuter make une deuxième fois. Dans les cas extrèmes, remontez à la racine des sources et entrez make Makefiles.
Pour ajouter des dépendances:
make depend
La commande make depend insère les fichiers d'en-têtes de dépendances aux Makefile et Makefile.in pour chaque fichier objet. Cette commande devrait être lancée dans chaque répertoire où vous modifiez les dépendances. En principe, il suffit de l'exécuter lorsque vous ajoutez ou supprimez des sources ou fichiers d'en-têtes. make depend est invoqué automatiquement durant le processus de configuration.
Pour installer:
make install
En principe, vous n'utilisez pas cette commande si vous êtes en train de développer Bacula, mais si vous vous apprétez à l'exécuter pour sauvegarder vos systèmes.
Après avoir lancé make install, les fichiers suivants seront installés sur votre système (à peu de choses près). La liste exacte des fichiers installés et leur localisation dépend de votre commande c./configure (e.g. gnome-console et gnome-console.conf ne sont pas installés si vous ne configurez pas GNOME. De même, si vous utilisez SQLite plutôt que MySQL, certains fichiers seront différents.
bacula bacula-dir bacula-dir.conf bacula-fd bacula-fd.conf bacula-sd bacula-sd.conf bacula-tray-monitor tray-monitor.conf bextract bls bscan btape btraceback btraceback.gdb bconsole bconsole.conf create_mysql_database dbcheck delete_catalog_backup drop_bacula_tables drop_mysql_tables fd gnome-console gnome-console.conf make_bacula_tables make_catalog_backup make_mysql_tables mtx-changer query.sql bsmtp startmysql stopmysql bwx-console bwx-console.conf
Le Tray Monitor est déjà installé si vous avez utilisé l'option --
enable-tray-monitor de la commande configure et exécuté make
install.
Comme vous n'exécutez pas votre environnement graphique en tant que root (si vous le faites, vous devriez changer cette mauvaise habitude), n'oubliez pas d'autoriser votre utilisateur à lire tray-monitor.conf, et exécuter bacula-tray-monitor (ceci ne constitue pas une faille de sécurité).
Puis, connectez vous à votre environnement graphique (KDE, Gnome, ou autre), lancez bacula-tray-monitor avec votre utilisateur et observez si l'icone d'une cartouche apparaît quelque part sur l'écran, usuellement dans la barre des tâches. Sinon, suivez les instructions suivantes relatives à votre gestionnaire de fenêtres.
System tray, ou zone de notification si vous utilisez la terminologie GNOME, est supporté par GNOME depuis la version 2.2. Pour l'activer, faites un click droit sur un de vos espaces de travail, ouvrez le menu Ajouter à ce bureau, puis Utilitaire et enfin, cliquez sur Zone de notification. (NDT: A valider)
System tray est supporté par KDE depuis la version 3.1. Pour l'activer, faites un click droit sur la barre de tâches, ouvrez le menu Ajouter, puis Applet, enfin cliquez sur System Tray.
Lisez la documentation pour savoir si votre gestionnaire de fenêtres supporte le standard systemtray de FreeDesktop, et comment l'activer le cas échéant.
Consultez le chapitre Configurer Bacula de ce manuel pour les instructions de configuration de Bacula.
Si vous suivez les instructions de ce chapitre, vous aurez couvert la majorité des problèmes qui peuvent survenir. Il va sans dire que si vous vous apercevez un jour que nous avons omis un point important, votre retour sera le bienvenu et enrichira ce document pour le bénéfice de tous.
Dans ce qui suit, nous supposons que vous avez installé Bacula, que vous l'avez plus ou moins compris, que vous avez au moins travaillé sur le chapitre "Une brève documentation", et que vous avez installé une configuration de production basique. Si ce n'est pas le cas, faites-le avant de revenir ici. La suite est une sorte de checklist d'actions à ne pas négliger, avec de brèves explications sur les raisons de nos recommandations. Vous trouverez les détails ailleurs dans le manuel. L'ordre suivi est peu ou prou celui que vous suivrez en installant un système de sauvegardes en production (utilisez cette liste même si vous êtes déjà en production !).
Bien que les points évoqués ci-dessous ne soient pas critiques, nous les recommandons à votre attention, car ils peuvent vous éviter des problèmes.
Si vous devez absolument implémenter une politique de sauvegardes dans laquelle vous écrivez chaque soir sur une cartouche différente que vous exportez le lendemain matin hors site, nous vous proposons quelques conseils :
Ce chapitre vous guidera à travers les étapes nécessaires pour exécuter Bacula. Pour cela, nous supposons que vous avez installé Bacula, peut être dans un simple répertoire comme le décrit le chapitre précédent, auquel cas vous pouvez exécuter Bacula sans être root pour ces tests. Nous supposons d'autre part que vous n'avez pas modifié les fichiers de configuration. Dans le cas contraire, nous vous recommandons de désinstaller Bacula et de le réinstaller sans rien modifier. Les exemples de ce chapitre utilisent les fichiers de configuration par défaut, et créent les volumes dans le répertoire /tmp de votre disque. De plus, les données sauvegardées seront celle du répertoire des sources de Bacula où vous l'avez compilé. Par conséquent, tous les daemons peuvent être exécutés sans les droits root pour ces tests. Notez bien qu'en production, vos File Daemons devront être exécutés en tant que root. Voyez le chapitre sur la sécurité pour plus d'informations sur ce sujet.
Voici les étapes que nous suivrons :
Chacune de ces étapes est décrite en détail ci-dessous.
Avant d'utiliser Bacula pour la première fois en production, nous vous recommandons d'exécuter la commande test du programme btape ainsi qu'il est décrit dans le chapitre Programmes utilitaires de ce manuel. Ce programme vous aidera à vous assurer que votre lecteur de bandes fonctionne correctement avec Bacula. Si vous avez un lecteur moderne de marque HP, Sony, ou Quantum DDS ou DLT qui fonctionne sous Linux ou Solaris, vous pouvez probablement vous dispenser de faire ce test car Bacula est bien testé avec ces lecteurs et ces systèmes. Dans tous les autres cas, vous êtes fortement encouragé à exécuter les tests avant de poursuivre. btape dispose aussi d'une commande fill qui tente de reproduire le comportement de Bacula lorsqu'il remplit une cartouche et qu'il poursuit son écriture sur la suivante. Vous devriez songer à faire ce test, sachez cependant qu'il peut être long (environ 4 heures sur mon lecteur) de remplir une cartouche de haute capacité.
Si vous utilisez MySQL ou PostgreSQL pour votre catalogue Bacula, vous devez démarrer la base de données avant d'essayer de lancer un job pour éviter d'obtenir des messages d'erreur au démarrage de Bacula. J'utilise les scripts startmysql et stopmysql pour démarrer mon MySQL local. Notez que si vous utilisez SQLite, vous n'aurez pas à utiliser startmysql ou stopmysql. Si vous utilisez ceci en production, vous souhaiterez probablement trouver un moyen pour démarrer automatiquement MySQL ou PostgreSQL après chaque redémarrage du système.
Si vous utilisez SQLite (c'est à dire, si vous avez spécifié l'option
--
with-sqlite=xxx de la commande ./configure, vous n'avez rien à faire.
SQLite est démarrée automatiquement par Bacula.
Que vous ayez compilé Bacula depuis les sources ou que vous ayez installé les rpms, tapez simplement :
./bacula start
dans votre répertoire d'installation pour démarrer les trois daemons.
Le script bacula lance le Storage Daemon, le File Daemon et le Director Daemon, qui tournent tous trois en tant que daemons en tâche de fond. Si vous utilisez la fonction de démarrage automatique de Bacula, vous pouvez, au choix, lancer les trois daemons lors du démarrage, ou au contraire les lancer individuellement avec les scripts bacula-dir, bacula-fd, et bacula-sd usuellement situés dans /etc/init.d, bien que leur localisation effective soit dépendante du système d'exploitation.
Notez que seul le File Daemon a été porté sur les systèmes Windows, et qu'il doit être démarré différamment. Veuillez consulter le chapitre La version Windows de Bacula de ce manuel.
Les paquetages rpm configurent les daemons pour qu'ils s'exécutent en tant qu'utilisateur root et en tant que groupe bacula. Le processus d'installation rpm se charge de créer le groupe bacula s'il n'existe pas sur le système. Tout utilisateur ajouté au groupe bacula hérite de l'accès aux fichiers créés par les daemons. Pour modifier ce comportement, éditez les scripts de démarrage des daemons :
puis redémarrez-les.
Le chapitre installation de ce manuel indique comment installer les scripts de démarrage automatique des daemons.
Pour communiquer avec le Director et pour s'enquérir de l'état de Bacula ou de jobs en cours d'exécution, tapez simplement :
./bconsole
dans le répertoire de plus haut niveau.
Si vous avez installé la console GNOME et utilisé l'option --
enable-gnome
de la commande configure, vous pouvez aussi utiliser la console GNOME en tapant :
./gnome-console
Vous pouvez aussi utiliser le programme wxWidgets bwx-console.
Pour simplifier, nous ne décrirons ici que le programme ./bconsole. La plus grande partie de ce qui est décrit ici s'applique aussi aux programmes ./gnome-console et bwx-console.
La commande ./bconsole lance le programme Console, qui se connecte au Director. Bacula étant un programme réseau, vous pouvez utiliser la Console depuis n'importe quelle machine de votre réseau. Cependant, la plupart du temps le Console est exécutée sur la même machine que le Director. En principe, la Console devrait produire un affichage similaire à :
[kern@polymatou bin]$ ./bconsole Connecting to Director lpmatou:9101 1000 OK: HeadMan Version: 1.30 (28 April 2003) *
L'astérisque est l'invite de commande de la console.
Tapez help pour obtenir la liste des commandes disponibles :
*help Command Description ======= =========== add add media to a pool autodisplay autodisplay [on/off] -- console messages automount automount [on/off] -- after label cancel cancel job=nnn -- cancel a job create create DB Pool from resource delete delete [pool=<pool-name> | media volume=<volume-name>] estimate performs FileSet estimate debug=1 give full listing exit exit = quit help print this command label label a tape list list [pools | jobs | jobtotals | media <pool> | files jobid=<nn>]; from catalog llist full or long list like list command messages messages mount mount <storage-name> prune prune expired records from catalog purge purge records from catalog query query catalog quit quit relabel relabel a tape release release <storage-name> restore restore files run run <job-name> setdebug sets debug level show show (resource records) [jobs | pools | ... | all] sqlquery use SQL to query catalog status status [storage | client]=<name> time print current time unmount unmount <storage-name> update update Volume or Pool use use catalog xxx var does variable expansion version print Director version wait wait until no jobs are running *
Pour plus de détails sur les commandes de la console, consultez le chapitre Console de ce manuel.
A ce stade, nous supposons que vous avez :
--
your-options
En outre, nous supposons pour le moment que vous utilisez les fichiers de configuration par défaut.
Maintenant, entrez les commandes suivantes :
show filesets
Vous devriez obtenir quelque chose comme :
FileSet: name=Full Set O M N I /home/kern/bacula/regress/build N E /proc E /tmp E /.journal E /.fsck N FileSet: name=Catalog O M N I /home/kern/bacula/regress/working/bacula.sql N
Il s'agit d'un FileSet prédéfini qui sauvegardera le répertoire des sources de Bacula. Les noms de répertoires qui seront réellement affichés devraient correspondre à votre configuration. Dans une perspective de tests, nous avons choisi un répertoire de taille et de complexité modérée (environ 40 Mo). Le FileSet Catalog est utilisé pour sauvegarder le catalogue Bacula et nous ne nous y attarderons pas pour le moment. Les entrées I sont les fichiers ou répertoires qui seront inclus dans la sauvegarde, tandis que les entrées E sont ceux qui en seront exclus, quand aux entrées O, ce sont les options spécifiées pour ce FileSet. Vous pouvez changer ce qui est sauvegardé en modifiant la ligne File = de la ressource FileSet.
Il est maintenant temps de lancer votre première sauvegarde. Nous allons sauvegarder votre répertoire sources de Bacula vers un volume File dans votre répertoire /tmp afin de vous montrer combien c'est facile. Saisissez :
status dir
Vous devriez obtenir :
rufus-dir Version: 1.30 (28 April 2003) Daemon started 28-Apr-2003 14:03, 0 Jobs run. Console connected at 28-Apr-2003 14:03 No jobs are running. Level Type Scheduled Name ================================================================= Incremental Backup 29-Apr-2003 01:05 Client1 Full Backup 29-Apr-2003 01:10 BackupCatalog ====
Où les dates et le nom du Director seront différents et en accord avec votre installation. Ceci montre qu'une sauvegarde incrémentale est planifiée pour le job Client1 à 1h05, et qu'une sauvegarde full est planifiée pour le job BackupCatalog à 1h10. Vous devriez remplacer le nom Client1 par celui de votre machine, sinon vous risquez la confusion lorsque vous ajouterez de nouveaux clients. Pour ma machine réelle, j'utilise Rufus plutôt que Client1.
A présent, tapez :
status client
Vous devriez obtenir :
The defined Client resources are: 1: rufus-fd Item 1 selected automatically. Connecting to Client rufus-fd at rufus:8102 rufus-fd Version: 1.30 (28 April 2003) Daemon started 28-Apr-2003 14:03, 0 Jobs run. Director connected at: 28-Apr-2003 14:14 No jobs running. ====
Dans ce cas, le client se nomme rufus-fd, votre nom sera différent, mais la ligne qui débute par rufus-fd Version... est produite par votre File Daemon, nous sommes donc maintenant surs qu'il fonctionne.
Finalement, faites de même pour votre Storage Daemon :
status storage
Vous devriez obtenir :
The defined Storage resources are: 1: File Item 1 selected automatically. Connecting to Storage daemon File at rufus:8103 rufus-sd Version: 1.30 (28 April 2003) Daemon started 28-Apr-2003 14:03, 0 Jobs run. Device /tmp is not open. No jobs running. ====
Vous noterez que le périphérique du Storage Daemon par défaut est nommé File et qu'il utilise le périphérique /tmp, qui n'est actuellement pas ouvert.
Maintenant, lancez un job :
run
Vous devriez obtenir :
Using default Catalog name=MyCatalog DB=bacula A job name must be specified. The defined Job resources are: 1: Client1 2: BackupCatalog 3: RestoreFiles Select Job resource (1-3):
Ici, Bacula affiche la liste des trois différents jobs que vous pouvez exécuter. Choisissez le numéro 1 et validez (entrée).
Vous devriez obtenir :
Run Backup job JobName: Client1 FileSet: Full Set Level: Incremental Client: rufus-fd Storage: File Pool: Default When: 2003-04-28 14:18:57 OK to run? (yes/mod/no):
Prenez un peu de temps pour examiner cet affichage et le comprendre. Il vous est demandé de valider, modifier ou annuler l'exécution d'un job nommé Client1 avec le FileSet Full Set que nous avons affiché plus haut en incrémental sur votre client rufus, utilisant le périphérique de stockage File et le pool Default à la date indiquée sur la ligne "When".
Nous avons le choix de valider (yes), modifier un ou plusieurs des paramètres ci-dessus (mod), ou de ne pas exécuter le job (no).
Validez l'exécution du job (yes), vous devriez immédiatement obtenir l'invite de commande de la console (un astérisque). Après quelques minutes, la commande messages devrait produire un résultat tel que :
28-Apr-2003 14:22 rufus-dir: Last FULL backup time not found. Doing FULL backup. 28-Apr-2003 14:22 rufus-dir: Start Backup JobId 1, Job=Client1.2003-04-28_14.22.33 28-Apr-2003 14:22 rufus-sd: Job Client1.2003-04-28_14.22.33 waiting. Cannot find any appendable volumes. Please use the "label" command to create a new Volume for: Storage: FileStorage Media type: File Pool: Default
Le premier message signale qu'aucune sauvegarde full n'a jamais été faite, et que par conséquent Bacula élève votre incrémentale en une Full (ce comportement est normal). Le second message indique que le job a démarré avec le JobId 1 et le troisième message vous informe que Bacula ne peut trouver aucun volume dans le pool Default sur lequel écrire les données du job. Ceci est normal, car nous n'avons encore créé (ou étiqueté) aucun volume. Bacula vous fournit tous les détails concernant le volume dont il a besoin.
A ce point, le job est bloqué en attente d'un volume. Vous pouvez le vérifier en utilisant la commande status dir. Pour continuer, vous devez créer un volume sur lequel Bacula pourra écrire. Voici la manipulation :
label
Bacula devrait afficher :
The defined Storage resources are: 1: File Item 1 selected automatically. Enter new Volume name:
Entrez un nom commençant par une lettre et ne contenant que des chiffres et des lettres (périodes, tirets et souligné "_" sont aussi autorisés). Par exemple entrez TestVolume001, vous devriez obtenir :
Defined Pools: 1: Default Item 1 selected automatically. Connecting to Storage daemon File at rufus:8103 ... Sending label command for Volume "TestVolume001" Slot 0 ... 3000 OK label. Volume=TestVolume001 Device=/tmp Catalog record for Volume "TestVolume002", Slot 0 successfully created. Requesting mount FileStorage ... 3001 OK mount. Device=/tmp
Finalement, tapez la commande messages, vous devriez obtenir quelque chose comme :
28-Apr-2003 14:30 rufus-sd: Wrote label to prelabeled Volume "TestVolume001" on device /tmp 28-Apr-2003 14:30 rufus-dir: Bacula 1.30 (28Apr03): 28-Apr-2003 14:30 JobId: 1 Job: Client1.2003-04-28_14.22.33 FileSet: Full Set Backup Level: Full Client: rufus-fd Start time: 28-Apr-2003 14:22 End time: 28-Apr-2003 14:30 Files Written: 1,444 Bytes Written: 38,988,877 Rate: 81.2 KB/s Software Compression: None Volume names(s): TestVolume001 Volume Session Id: 1 Volume Session Time: 1051531381 Last Volume Bytes: 39,072,359 FD termination status: OK SD termination status: OK Termination: Backup OK 28-Apr-2003 14:30 rufus-dir: Begin pruning Jobs. 28-Apr-2003 14:30 rufus-dir: No Jobs found to prune. 28-Apr-2003 14:30 rufus-dir: Begin pruning Files. 28-Apr-2003 14:30 rufus-dir: No Files found to prune. 28-Apr-2003 14:30 rufus-dir: End auto prune.
Si rien ne se passe dans l'immédiat, vous pouvez continuer de rentrer la commande messages jusqu'à ce que le job se termine, ou utiliser la commande autodisplay on afin que les messages soient affichés dès-qu'ils sont disponibles.
si vous faites ls -l dans votre répertoire /tmp, vous verrez l'élément suivant :
-rw-r----- 1 kern kern 39072153 Apr 28 14:30 TestVolume001
Il s'agit du volume File que vous venez juste d'écrire, et qui contient toutes les données du job que vous venez d'exécuter. Si vous exécutez d'autres jobs, il seront ajoutés à la suite de ce volume, à moins que vous n'ayez spécifié un autre comportement.
Vous vous demandez peut-être s'il va vous falloir étiqueter vous même chaque volume que Bacula sera amené à utiliser. La réponse, en ce qui concerne les volumes disque tels que celui que nous avons utilisé, est non. Il est possible de paramétrer Bacula pour qu'il créée lui même les volumes. En revanche, pour les volumes de type cartouche, il vous faudra très probablement étiqueter chaque volume que vous voulez utiliser.
Si vous souhaitez en rester là, saisissez simplement quit dans la console, puis stoppez Bacula avec ./bacula stop. Pour nettoyer votre installation des résultats de vos tests, supprimez le fichier /tmp/TestVolume001, et réinitialisez votre catalogue en utilisant :
./drop_bacula_tables ./make_bacula_tables
Notez bien que ceci supprimera toutes les informations concernant les jobs précédemment exécutés et que, si c'est sans doute ce que vous souhaitez faire en fin de phase de test, ce n'est généralement pas une opération souhaitable en utilisation normale.
Si vous souhaitez essayer de restaurer les fichiers que vous venez de sauvegarder, lisez la section suivante.
Si vous avez utilisé la configuration par défaut et sauvegardé les sources de Bacula comme dans la démonstration ci-dessus, vous pouvez restaurer les fichiers sauvegardés en saisissant les commandes suivantes dans la Console :
restore all
Vous obtiendrez :
First you select one or more JobIds that contain files to be restored. You will be presented several methods of specifying the JobIds. Then you will be allowed to select which files from those JobIds are to be restored. To select the JobIds, you have the following choices: 1: List last 20 Jobs run 2: List Jobs where a given File is saved 3: Enter list of comma separated JobIds to select 4: Enter SQL list command 5: Select the most recent backup for a client 6: Select backup for a client before a specified time 7: Enter a list of files to restore 8: Enter a list of files to restore before a specified time 9: Find the JobIds of the most recent backup for a client 10: Find the JobIds for a backup for a client before a specified time 11: Enter a list of directories to restore for found JobIds 12: Cancel Select item: (1-12):
Comme vous pouvez le constater, les options sont nombreuses, mais pour l'instant, choisissez l'option 5 afin de sélectionner la dernière sauvegarde effectuée. Vous obtiendrez :
Defined Clients: 1: rufus-fd Item 1 selected automatically. The defined FileSet resources are: 1: 1 Full Set 2003-04-28 14:22:33 Item 1 selected automatically. +-------+-------+----------+---------------------+---------------+ | JobId | Level | JobFiles | StartTime | VolumeName | +-------+-------+----------+---------------------+---------------+ | 1 | F | 1444 | 2003-04-28 14:22:33 | TestVolume002 | +-------+-------+----------+---------------------+---------------+ You have selected the following JobId: 1 Building directory tree for JobId 1 ... 1 Job inserted into the tree and marked for extraction. The defined Storage resources are: 1: File Item 1 selected automatically. You are now entering file selection mode where you add and remove files to be restored. All files are initially added. Enter "done" to leave this mode. cwd is: / $
(J'ai tronqué l'affichage à droite par soucis de lisibilité.) Comme vous pouvez le constater au début de cet affichage, Bacula connaît vos clients, et puisque vous n'en avez qu'un, il est automatiquement sélectionné. Il en va de même pour le FileSet. Bacula produit alors une liste de tous les jobs qui constituent la sauvegarde courante. Dans le cas présent, il n'y en a qu'un. Notez que le Storage Daemon est aussi sélectionné automatiquement. Bacula est maintenant en mesure de produire une arborescence à partir de tous les fichiers qui ont été sauvegardés (il s'agit d'une représentation en mémoire de votre système de fichiers). A ce stade, vous pouvez utiliser les commandes cd , ls et dir pour naviguer dans l'arborescence et voir quels fichiers peuvent être restaurés. Par exemple, si je saisis cd /home/kern/bacula/bacula-1.30 suivi de dir, j'obtiens la liste de tous les fichiers du répertoire source de Bacula. Pour plus d'information sur ce sujet, veuillez consulter le chapitre La commande Restore.
Pour quitter, tapez simplement :
done
Vous obtiendrez :
Bootstrap records written to /home/kern/bacula/testbin/working/restore.bsr The restore job will require the following Volumes: TestVolume001 1444 files selected to restore. Run Restore job JobName: RestoreFiles Bootstrap: /home/kern/bacula/testbin/working/restore.bsr Where: /tmp/bacula-restores Replace: always FileSet: Full Set Client: rufus-fd Storage: File JobId: *None* When: 2005-04-28 14:53:54 OK to run? (yes/mod/no):
Si vous acceptez (yes), vos fichiers seront restaurés vers le répertoire /tmp/bacula-restores. Si vous préférez restaurer les fichiers à leurs emplacements d'origine, vous devez utiliser l'option mod et régler explicitement le paramètre Where à vide ou "/". Nous vous conseillons de poursuivre avec yes. Après quelques instants, la commande messages devrait produire la liste des fichiers restaurés, ainsi qu'un résumé du job qui devrait ressembler à ceci :
28-Apr-2005 14:56 rufus-dir: Bacula 1.30 (28Apr03): 28-Apr-2003 14:56 JobId: 2 Job: RestoreFiles.2005-04-28_14.56.06 Client: rufus-fd Start time: 28-Apr-2005 14:56 End time: 28-Apr-2005 14:56 Files Restored: 1,444 Bytes Restored: 38,816,381 Rate: 9704.1 KB/s FD termination status: OK Termination: Restore OK 28-Apr-2005 14:56 rufus-dir: Begin pruning Jobs. 28-Apr-2005 14:56 rufus-dir: No Jobs found to prune. 28-Apr-2005 14:56 rufus-dir: Begin pruning Files. 28-Apr-2005 14:56 rufus-dir: No Files found to prune. 28-Apr-2005 14:56 rufus-dir: End auto prune.
Après avoir quitté la Console, vous pouvez examiner les fichiers dans le répertoire /tmp/bacula-restores, il contient l'arborescence avec tous vos fichiers. Supprimez-le après avoir vérifié :
rm -rf /tmp/bacula-restore
Saisissez simplement la commande quit.
Si vous êtes parvenus à faire fonctionner tous les exemples ci-dessus, vous êtes sans doute prèt à ajouter un nouveau client (File Daemon), c'est à dire une seconde machine que vous souhaitez sauvegarder. La seule chose à installer sur la nouvelle machine est le binaire bacula-fd (ou bacula-fd.exe pour Windows) et son fichier de configuration bacula-fd.conf. Vous pouvez démarrer en copiant le fichier précédemment créé moyennant une modification mineure pour l'adapter au nouveau client : changez le nom de File Daemon (rufus-fd dans l'exemple ci-dessus) en le nom que vous avez choisi pour le nouveau client. Le mieux est d'utiliser le nom de la machine. Par exemple :
... # # "Global" File daemon configuration specifications # FileDaemon { # this is me Name = rufus-fd FDport = 9102 # where we listen for the director WorkingDirectory = /home/kern/bacula/working Pid Directory = /var/run } ...
devient :
... # # "Global" File daemon configuration specifications # FileDaemon { # this is me Name = matou-fd FDport = 9102 # where we listen for the director WorkingDirectory = /home/kern/bacula/working Pid Directory = /var/run } ...
Où rufus-fd est devenu matou-fd (je ne montre qu'une partie du fichier). Le choix des noms vous appartient. Pour l'instant, je vous recommande de ne rien changer d'autre. Plus tard, vous changerez le mot de passe.
Installez cette configuration sur votre seconde machine. Il vous faut maintenant ajouter quelques lignes à votre bacula-dir.conf pour définir le nouveau File Daemon. En vous basant sur l'exemple initial qui devrait être installé sur votre système, ajoutez les lignes suivantes (essentiellement, une copie des lignes existantes avec seulement les noms modifiés) à votre bacula-dir.conf :
# # Define the main nightly save backup job # By default, this job will back up to disk in /tmp Job { Name = "Matou" Type = Backup Client = matou-fd FileSet = "Full Set" Schedule = "WeeklyCycle" Storage = File Messages = Standard Pool = Default Write Bootstrap = "/home/kern/bacula/working/matou.bsr" } # Client (File Services) to backup Client { Name = matou-fd Address = matou FDPort = 9102 Catalog = MyCatalog Password = "xxxxx" # password for File Retention = 30d # 30 days Job Retention = 180d # six months AutoPrune = yes # Prune expired Jobs/Files }
Assurez-vous que le paramètre Address de la ressource Storage a pour valeur le nom pleinement qualifié et non quelque chose comme "localhost". L'adresse spécifiée est envoyée au client et doit être un nom pleinement qualifié. Si vous utilisez "localhost", l'adresse du Storage Daemon ne sera pas résolue correctement, il en résultera un timeout lorsque le File Daemon échouera à connecter le Storage Daemon.
Il n'y a rien d'autre à faire. J'ai copié les ressources existantes pour créer un second job (Matou) pour sauvegarder le second client (matou-fd). le client se nomme matou-fd et le job Matou, le fichier bootstrap est modifié mais tout le reste est inchangé. Ceci signifie que Matou sera sauvegardé avec la même planification sur les mêmes cartouches. Vous pourrez changer ceci plus tard, pour le moment, restons simples.
La seconde modification consiste en l'ajout d'une nouvelle ressource Client qui définit matou-fd et qui a l'adresse correcte matou (mais dans la vraie vie, vous pouvez avoir besoin d'un nom pleinement qualifié ou d'une adresse IP. J'ai aussi conservé le même mot de passe (xxxxx dans l'exemple).
A ce stade, il suffit de redémarrer Bacula pour qu'il prenne en compte vos modifications. L'invite que vous avez vu plus haut devrait maintenant inclure la nouvelle machine.
Pour une utilisation en production vous voudrez probablement utiliser plusieurs pools et différentes planifications. Il vous appartient de faire les adaptations qui seyent à vos besoins. Dans tous les cas, n'oubliez pas de changer les mots de passe dans les fichiers de configuration du Director et du Client pour des raisons de sécurité.
Vous trouverez des astuces importantes concernant le changement des noms et mots de passe, ainsi qu'un diagramme décrivant leurs correspondances dans la section Erreurs d'authentification du chapitre FAQ de ce manuel.
rufus-sd: block.c:337 === Write error errno=28: ERR=No space left on device
Ceci indique que Bacula a reçu une erreur d'écriture à cause de la carouche pleine. Bacula va maintenant rechercher une cartouche utilisable dans le pool spécifié pour le job. Dans la situation idéale, vous avez réglé correctement vos rétentions et spécifié que vos cartouches peuvent être recyclées automatiquement. Dans ce cas, Bacula recycle automatiquement vos cartouches sorties de rétention et est en mesure de réécrire dessus. Pour plus d'informations sur le recyclage, veuillez consulter le chapitre Recyclage de ce manuel. Si vous constatez que vos cartouches ne sont pas recyclées correctement, consultez la section sur le Recyclage manuel du chapitre Recyclage.
Si comme moi, vous avez un très grand nombre de cartouches que vous étiquetez avec la date de première écriture, si vous n'avez pas réglé vos périodes de rétention, Bacula ne trouvera pas de cartouche dans le pool et il vous enverra un message tel que :
rufus-sd: Job kernsave.2002-09-19.10:50:48 waiting. Cannot find any appendable volumes. Please use the "label" command to create a new Volume for: Storage: SDT-10000 Media type: DDS-4 Pool: Default
Ce message sera répété une heure plus tard, puis deux heures plus tard et ainsi de suite en doublant à chaque fois l'intervalle à concurrence d'un jour jusqu'à ce que vous créiez un volume.
Que faire dans cette situation ?
La réponse est simple : d'abord, fermez le lecteur à l'aide de la commande unmount du programme Console. Si vous n'avez qu'un lecteur, il sera sélectionné automatiquement, sinon assurez-vous de démonter celui spécifié dans le message (dans ce cas STD-10000).
Ensuite, retirez la cartouche du lecteur et insérez-en une vierge. Notez que sur certains lecteurs anciens, il peut être nécessaire d'écrire une marque de fin de fichier (mt -f /dev/nst0 weof) pour éviter que le lecteur ne déroule toute la cartouche lorsque Bacula tente de lire le label. (NDT : j'ai un doute, la vo dit : "to prevent the drive from running away when Bacula attempts to read the label.")
Finalement, utilisez la commande label dans la console pour écrire un label sur le nouveau volume. la commande label va contacter le Storage Daemon pour qu'il écrive l'étiquette logicielle. Si cette opération se termine correctement, le nouveau volume est ajouté au pool et la commande mount est envoyée au Storage Daemon. Voyez les sections précédentes de ce chapitre pour plus de détails sur l'étiquetage des cartouches.
Bacula peut maintenant poursuivre le job et continuer d'écrire les données sauvegardées sur le nouveau volume.
Si Bacula cycle sur un pool de volumes, au lieu du message ci-dessus "Cannot find any appendable volumes.", Bacula peut vous demander de monter un volume particulier. Dans ce cas, essayez de le satisfaire. Si, pour quelque raison, vous n'avez plus le volume, vous pouvez monter n'importe quel autre volume du pool, pourvu qu'il soit utilisable, Bacula l'utilisera. La commande list volumes du programme Console permet de déterminer les volumes utilisables et ceux qui ne le sont pas.
Si, comme moi, vous avez paramétré correctement vos périodes de rétention, mais n'avez plus aucun volume libre, vous pouvez ré-étiqueter et ré-utiliser un volume comme suit :
Pour ré-étiqueter manuellement le volume, suivez les étapes supplémentaire ci-dessous :
La plupart des commandes citées ci-dessus, à l'exception de list, vous invitent à compléter la liste des arguments fournis si vous vous contentez d'entrer le nom de la commande.
Si vous voulez débugger la sortie des daemons en cours d'exécution, lancez-les, depuis le répertoire d'installation, comme suit :
./bacula start -d100
Cette possibilité peut vous fournir une aide précieuse si vos daemons ne démarrent pas correctement. Normalement, la sortie des daemons est dirigée vers le périphérique NULL, avec un niveau de débuggage supérieur à zéro, elle est dirigée vers le terminal de lancement.
Pour stopper les trois daemons, tapez simplement :
./bacula stop
dans le répertoire d'installation.
L'exécution de bacula stop peut signaler des pids non trouvés. C'est Ok, spécialement si l'un des bacula stop est mort, ce qui est très rare.
Pour faire une sauvegarde complète (Full) du système, chaque File Daemon doit être exécuté en tant que root afin d'avoir les permissions requises pour accéder à tous les fichiers. Les autres daemons n'ont pas besoin des privilèges root. Cependant, le Storage Daemon doit être capable d'accéder aux lecteurs, ce qui Sur beaucoup de systèmes, n'est possible que pour root. Vous pouvez, au choix, exécuter le Storage Daemon en tant que root, ou changer les permissions sur les lecteurs pour autoriser les accès non-root. MySQL et PostgreSQL peuvent être installés et exécutés avec un userid quelconque, les privilèges root ne sont pas requis.
Lorsque vous lancez les daemons Bacula, le Storage Daemon tente d'ouvrir tous les périphériques de stockage définis et de vérifier le volumes courrament montés. Il n'accepte aucune connection de la console tant que tous les périphériques n'ont pas été vérifiés. Une cartouche qui a été utilisé précédemment doit être rembobinée, ce qui, sur certain lecteurs, peut prendre plusieurs minutes. Par conséquent, vous devriez faire preuve d'un peu de patience lorsue vous tentez de contacter le Storage Daemon pour la première fois après le lancement de Bacula. Si vous avez un accès visuel à votre lecteur, celui-ci devrait être prèt à l'emploi lorsque son témoin lumineux cesse de clignoter.
Les mêmes considérations s'appliquent si vous avez monté une cartouche vierge dans un lecteur tels qu'un HP DLT. Il peut s'écouler une à deux minutes avant que le lecteur se rende compte que la cartouche est vierge. Si vous tentez de la monter pendant cette période, il est probable que vous aller geler votre pilote SCSI (c'est le cas sur mon système RedHat). Par conséquent, nous vous enjoignons une fois encore à être patient lors de l'insertion de cartouches vierges. Laissez le lecteur s'initialiser avant de tenter d'y accéder.
Si l'un ou plusieurs de vos File Daemons rencontre des difficultés à se connecter au Storage Daemon, c'est très probablement que vous n'avez pas utilisé un nom pleinement qualifié pour la directive Address de la ressource Storage du fichier de configuration du Director. Le résolveur de la machine cliente (celle qui exécute le FD) doit être capable de résoudre le nom que vous avez spécifié dans cette directive en une adresse IP. Un exemple d'adresse ne fonctionnant pas est localhost. Un exemple qui pourrait fonctionner : megalon. Un exemple qui a encore plus de chances de fonctionner : magalon.mydomain.com. Sur les systèmes Win32, si vous ne disposez pas d'un bon résolveur (c'est souvent le cas sur Win98), vous pouvez essayer en utilisant une adresse IP plutôt qu'un nom.
Si votre adresse est correcte, assurez vous qu'aucun autre programme n'utilise le port 9103 sur la machine qui héberge le Storage Daemon. Les numéros de ports de Bacula sont autorisés par l'IANA, et ne devraient donc pas être utilisés par d'autres programmes, mais il semble que certaines imprimantes HP les utilisent. Exécutez la commande netstat -a sur la machine qui héberge le Storage Daemon pour déterminer qui utilise le port 9103 (utilisé pour les communications du FD vers le SD).
Chacun des trois daemons (Director, File, Storage) acceptent quelques options sur la ligne de commande. En général, chacun d'entre eux, de même que le programme Console, admet les otpions suivantes :
Le Director a les options spécifiques suivantes :
Le File Daemon les options spécifiques suivantes :
Le Storage Daemon n'a pas d'options spécifiques.
Le programme Console n'a pas d'options spécifiques.
La création de pool est automatique au démarrage de Bacula, aussi si vous comprenez déjà le concept de pools et leur fonctionnement, vous pouvez passer à la section suivante.
Lorsque vous exécutez un job, Bacula doit déterminer quel volume utiliser pour sauvegarder le FileSet. Plutôt que de spécifier un volume directement, vous spécifiez l'ensemble de volumes dans lequel vous autorisez Bacula à puiser lorsqu'il lui faut un volume pour écrire les données sauvegardées. Dès lors, Bacula se charge de sélectionner le premier volume utilisable dans le pool approprié au périphérique que vous avez spécifié pour le job exécuté. Lorsqu'un volume est plein, Bacula change son VolStatus de Append en Full, et utilise le volume suivant, et ainsi de de suite. S'il n'y a pas de volume utilisable, Bacula envoie un message à l'opérateur pour réclamer la création d'un volume approprié.
Bacula garde trace des noms de pools, des volumes contenus dans les pools, et de plusieurs caractéristiques de chacun de ces volumes.
Lorsque Bacula démarre, il s'assure que toutes les définitions de ressources Pool ont été enregistrées dans le catalogue. Vous pouvez le vérifier avec la commande :
list pools
du programme Console, qui devrait produire quelque chose comme :
*list pools Using default Catalog name=MySQL DB=bacula +--------+---------+---------+---------+----------+-------------+ | PoolId | Name | NumVols | MaxVols | PoolType | LabelFormat | +--------+---------+---------+---------+----------+-------------+ | 1 | Default | 3 | 0 | Backup | * | | 2 | File | 12 | 12 | Backup | File | +--------+---------+---------+---------+----------+-------------+ *
Si vous tentez de créer un pool existant, Bacula affiche :
Error: Pool Default already exists. Once created, you may use the {\bf update} command to modify many of the values in the Pool record.
Bacula exige que chaque volume comporte une étiquette (NDT : label) logicielle. Il existe plusieurs stratégies pour étiqueter les volumes. Celle que j'utilise consiste à les étiqueter à l'aide du programme Console au fur et à mesure qu'ils sont requis par Bacula. Ainsi, lorsqu'il a besoin d'un volume qu'il ne trouve pas dans son catalogue, Bacula m'envoie un e-mail pour m'enjoindre à ajouter un volume au pool. J'utilise alors la commande label dans la console pour étiqueter un nouveau volume et le définir dans le catalogue, après quoi Bacula est en mesure de l'utiliser. Alternativement, je peux utiliser la commande relabel pour ré-étiquter un volume qui n'est plus utilisé, pourvu qu'il ait le VolStatus Purged.
Une autre stratégie consiste à étiqueter un ensemble de volumes, et à les utiliser au fur et à mesure que Bacula les réclame. C'est le plus souvent ce qui est fait lorsque vous cyclez sur un groupe de volumes, par exemple avec une librairie. Pour plus de détails sur le recyclage, veuillez consulter le chapitre Recyclage automatique des volumes de ce manuel.
Si vous exécutez un job Bacula alors que vous n'avez pas de volumes étiquetés dans le pool concerné, Bacula vous en informe, et vous pouvez les créer "à la volée". Dans mon cas, j'étiquette mes cartouches avec la date, par exemple : DLT-18April02. Voyez ci-dessous pour plus de détails sur l'usage de la commande label.
L'étiquetage des volumes se fait, en principe, avec le programme Console.
Si Bacula annonce que vous ne pouvez étiqueter une cartouche au motif qu'elle porte déjà une étiquette, démontez-la avec la commande unmount, puis recommencez avec une cartouche vierge.
Etand donné que le support de stockage physique est différent pour chaque périphérique, la commande label vous propose une liste de ressources Storage définies telle que celle-ci :
The defined Storage resources are: 1: File 2: 8mmDrive 3: DLTDrive 4: SDT-10000 Select Storage resource (1-4):
A ce stade, vous devriez avoir une cartouche vierge dans votre lecteur d'un type correspondant à la ressource Storage que vous avez sélectionné.
Bacula vous demande le nom du volume :
Enter new Volume name:
S'il proteste :
Media record for Volume xxxx already exists.
Cela signifie que le nom de volume xxxx que vous avez entré existe dèjà dans le catalogue. Vous pouvez afficher la liste des média définis avec la commande list media. Notez que la colonne LastWritten a ici été tronquée pour permettre un affichage propre.
+---------------+---------+--------+----------------+-----/~/-+------------+-----+ | VolumeName | MediaTyp| VolStat| VolBytes | LastWri | VolReten | Recy| +---------------+---------+--------+----------------+---------+------------+-----+ | DLTVol0002 | DLT8000 | Purged | 56,128,042,217 | 2001-10 | 31,536,000 | 0 | | DLT-07Oct2001 | DLT8000 | Full | 56,172,030,586 | 2001-11 | 31,536,000 | 0 | | DLT-08Nov2001 | DLT8000 | Full | 55,691,684,216 | 2001-12 | 31,536,000 | 0 | | DLT-01Dec2001 | DLT8000 | Full | 55,162,215,866 | 2001-12 | 31,536,000 | 0 | | DLT-28Dec2001 | DLT8000 | Full | 57,888,007,042 | 2002-01 | 31,536,000 | 0 | | DLT-20Jan2002 | DLT8000 | Full | 57,003,507,308 | 2002-02 | 31,536,000 | 0 | | DLT-16Feb2002 | DLT8000 | Full | 55,772,630,824 | 2002-03 | 31,536,000 | 0 | | DLT-12Mar2002 | DLT8000 | Full | 50,666,320,453 | 1970-01 | 31,536,000 | 0 | | DLT-27Mar2002 | DLT8000 | Full | 57,592,952,309 | 2002-04 | 31,536,000 | 0 | | DLT-15Apr2002 | DLT8000 | Full | 57,190,864,185 | 2002-05 | 31,536,000 | 0 | | DLT-04May2002 | DLT8000 | Full | 60,486,677,724 | 2002-05 | 31,536,000 | 0 | | DLT-26May02 | DLT8000 | Append | 1,336,699,620 | 2002-05 | 31,536,000 | 1 | +---------------+---------+--------+----------------+-----/~/-+------------+-----+
Une fois que Bacula a vérifié que le volume n'existe pas encore, il vous demande le pool dans lequel vous souhaitez que le volume soit créé. S'il n'existe qu'un pool, il est sélectionné automatiquement.
Si la cartouche est étiquetée correctement, un enregistrement de volume est aussi créé dans le pool. Ainsi, le nom du volume et tous ses attributs apparaîtront lorque vous afficherez les volumes du pool. De plus, le volume est disponible pour les sauvegardes, pourvu que le MediaType coØincide avec celui requis par le Storage Daemon.
Lorsque vous avez étiqueté la cartouche, vous n'avez répondu qu'à quelques questions la concernant -- principalement son nom, et éventuellement le Slot. Cependant, un enregistrement de volume dans le catalogue (connu au niveau interne en tant qu'enregistrement Media) contient un certain nombre d'attributs. La plupart d'entre eux sont renseignés selon les valeurs par défaut qui ont été définies lors de la création du pool (au trement dit, le pool comporte la plupart des attributs par défaut utilisés lors de la création d'un volume).
Il est aussi possible d'ajouter des media aux pools sans les étiqueter physiquement. C'est la fonction de la commande add. Pour plus d'informations, veuillez consulterle chapitre Console de ce manuel.
Lors de son démarrage, chacun des programmes qui composent Bacula lit un fichier de configuration spécifié sur la ligne de commande, ou par défaut bacula-dir.conf, bacula-fd.conf, bacula-sd.conf, ou console.conf pour le Director Daemon, le File Daemon, le Storage Daemon, et le programme Console respectivement.
Chaque service (Director, Client, Storage, Console) possède son propre fichier de configuration qui contient un groupe de directives. Dans la suite, nous désignerons ces groupes de directives par le mot Ressource. Les ressources son très similaires d'un service à l'autre, mais peuvent contenir des directives différentes selon les services. Par exemple, dans le fichier de configuration du Director, la ressource Director définit le nom du Director, quelques paramètres généraux du Director et son mot de passe. Dans le fichier de configuration du File Daemon, la ressource Director spécifie les Directors autorisés à utiliser le File Daemon.
Avant de lancer Bacula pour la première fois, vous devez adapter chaque fichier de configuration. Des fichiers de configuration auront été créés par le processus d'installation, mais il doivent être modifiés pour correspondre à votre système. Le schéma suivant donne une vue globale des ressources.
(Remerciements à Aristedes Maniatis pour ce schéma.)
Vous pouvez vérifier que les fichiers de configuration sont lisibles, y compris les caractères étrangers en examinant votre variable d'environnement bf LANG. Celle-ci doit se terminer par .UTF-8. Par exemple en_US.UTF-8. La syntaxe exacte pour la définir peut varier fortement d'unsystème d'exploitation à l'autre.
Bacula considère que tous les noms de fichiers sont au format UTF-8 sur les machines Unix et Linux. Sur les machines Windows, ils sont en UTF-16 et sont automatiquement convertis en UTF-8.
Bien qu'il ne soit pas nécessaire de connaître le détail de toutes les directives possibles, une connaissance basique des ressources Bacula est indispensable. Chaque directive contenue dans une ressource (entre accollades) est composé d'un mot-clef suivi du signe "=", suivi d'une ou plusieurs valeurs. Le mot clef doit être l'un de ceux connus par Bacula, et peut comporter des caractères majuscules et minuscules ainsi que des espaces.
Chaque définition de ressource doit comporter la directive Name, et peut optionnellement comporter la directive Description. La directive Name est utilisée pour identifier de façon unique la ressource. La directive Description sera utilisé lors de l'affichage pour offrir une identification plus aisée de la ressource. Par exemple :
Director { Name = "MyDir" Description = "Main Bacula Director" WorkingDirectory = "$HOME/bacula/bin/working" }
Définit la ressource Director avec le nom "MyDir'' et le répertoire de travail $HOME/bacula/bin/working. En général, si vous voulez utiliser des espaces dans le nom à droite du signe "=", vous devez l'entourer de doubles quotes. Sinon, les quotes ne sont généralement pas requises car une fois définies, les chaînes quotées et non quotées sont toutes équivalentes.
Lors de la lecture d'un fichier de configuration, les lignes blanches sont ignorées, et tout ce qui suit le caractère dièse (#) jusqu'à la fin de la ligne est considéré comme commentaire. Un point virgule (;) indique la fin logique d'une ligne et tout ce qui suit le point virgule est considéré comme le paramètre suivant. Un paramètre qui apparaît seul sur une ligne ne nécessite pas de point virgule, vous ne verrez donc pas beaucoup de points virgule dans les exemples de ce manuel.
La casse (majuscules/minuscules) et les espaces sont totalement ignorées dans les mots-clef des directives des ressources (la partie à gauche du signe "='').
A l'intérieur des mots-clef (à gauche du signe "=''), les espaces ne sont pas significatives. Ainsi, les mots-clef : name, Name, et N a m e sont tous identiques.
Les espaces après le signe "='' et avant le premier caractère de la valeur son ignorées.
En général, les espaces à l'intérieur d'une valeur sont significatives (non ignorées), et si la valeur est un nom, vous devez l'encadrer de doubles quotes pour que l'espace soit acceptée. Les noms peuvent contenir jusqu'à 127 caractères. Actuellement, un nom peut contenir n'importe quel caractère ASCII. A l'intérieur d'une chaîne quotée, tout caractère précédé d'un backslash (\) est pris tel quel (utile pour insérer des backslashes et doubles quotes (")).
Veuillez cependant noter que les noms de ressource Bacula ainsi que certains autres noms (par exemple les noms de volumes) ne doivent contenir que des lettres (y compris accentuées ISO), nombres, et une poignée de caractères spéciaux (espace, souligné, ...). Tous les autres caratères et ponctuations sont prohibés.
Si vous souhaitez éclater votre fichier de configuration en fichiers plus petits, l'inclusion est possible avec la syntaxe @NomDeFichier où nom de fichier est le chemin absolu vers un le fichier à inclure. Toute donnée primitive peut être remplacée par une spécification @NomDeFichier. Par exemple
Director { Name=@xxx est valide et substituera le fichier xxx à @xxx, mais Director { Name@xxx = something n'est pas valide puisque @xxx apparaît au milieu d'un (???mot clef/directive/...???)
Lorsqu'il parcourt les enregistrements de ressource, Bacula classe les données selon les types enumérés ci-dessous. En première lecture, cette liste peut vous paraître accablante, mais en réalité, tout y est d'une logique élégante et directe.
Toute abbréviation des ces modificateurs est aussi autorisée (i.e. seconds peut être spécifié par sec ou s. Une spécification m sera interprétée en tant que mois.
La spécification d'une durée peut comporter autant de couples nombre/modificateur que vous le souhaitez. Par exemple:
1 week 2 days 3 hours 10 mins 1 month 2 days 30 sec
sont des spécifications valides (à partir de la version 1.35.1).
Note! Dans les versions de Bacula antérieures à 1.31, le modificateur était optionnel. Il est désormais obligatoire.
La table suivante énumère tous les types de ressource de la version courante de Bacula. Elle montre quelle ressource doit être définie pour chaque service (daemon). Les fichiers de configuration par défaut contiennent au moins un exemple de chaque ressource permise, aussi ne soyez pas angoissé à l'idée d'avoir à créer toutes ces directives ex nihilo.
Resource | Director | Client | Storage | Console |
Autochanger | No | No | Yes | No |
Catalog | Oui | Non | Non | Non |
Client | Oui | Oui | Non | Non |
Console | Oui | Non | Non | Oui |
Device | Non | Non | Oui | Non |
Director | Oui | Oui | Oui | Oui |
FileSet | Oui | Non | Non | Non |
Job | Oui | Non | Non | Non |
JobDefs | Oui | Non | Non | Non |
Message | Oui | Oui | Oui | Non |
Pool | Oui | Non | Non | Non |
Schedule | Oui | Non | Non | Non |
Storage | Oui | Non | Oui | Non |
Pour qu'un daemon puisse en contacter un autre, il lui faut s'authentifier avec un mot de passe. Dans la plupart des cas, le mot de passe est associé à un nom particulier, de sorte que nom et mot de passe doivent correspondre.
Les fichiers de configuration par défaut sont automatiquement définis avec des autorisations correctes et des mots de passe aléatoires. Si vous modifiez ces fichiers, vous devrez être attentif à leur cohérence.
Voici un schéma des correspondances que doivent respecter les couples ''nom/mot de passe" des différents fichiers de configurations.
Dans la colonne de gauche, vous trouverez les ressources Director, Storage et Client, avec leurs noms et mots de passe -- ils sont tous dans le fichier bacula-dir.conf. Dans la colonne de droite figurent les valeurs correspondantes dans les fichiers de configuration de la Console, du Storage Daemon (SD) et du File Daemon (FD).
Veuillez noter que l'adresse fd-sd, qui apparaît dans la ressource Storage du Director, précédée d'une atérisque est transmise au File Daemon sous forme symbolique. Le File Daemon la résout alors en une adresse IP. Pour cette raison, vous devez utiliser soit une adresse IP, soit une adresse pleinement qualifiée. Une adresse telle que localhost, n'étant pas pleinement qualifiée, sera résolue par le File Daemon en l'hôte local du File Daemon, ce qui n'est probablement pas le résultat désiré. Le mot de passe utilisé par le File Daemon pour autoriser la communication avec le Storage Daemon est temporaire et unique pour chaque job. Il n'est spécifié dans aucun fichier de configuration.
Les détails de chaque ressource et des directives permises sont décrits dans les chapitres suivants.
Les fichiers de configuration suivants doivent être définis:
Parmi tous les fichiers de configuration requis pour exécuter Bacula, celui du Director est le plus compliqué, et c'est celui que vous modifierez le plus souvent, en ajoutant des clients ou en modifiant les FileSets.
Pour une discussion générale concernant les fichiers et ressources ainsi que les types de données reconnus par Bacula, veuillez consulter le chapitre Configuration de ce manuel.
Les types de ressources du Director sont :
Job, JobDefs, Client, Storage, Catalog, Schedule, FileSet, Pool, Director, et Messages. Nous les présentons ici dans l'ordre le plus logique (relativement au fichier de configuration du Director) :
La ressource Director définit les attributs du Director exécuté sur le réseau. Dans l'implémentation actuelle, il n'y a qu'une ressource Director, mais la réalisation finale contiendra plusieurs Directors pour maintenir la redondance de la base des indexes et média.
Si vous avez spécifié un utilisateur et/ou un groupe pour le Director lors de la configuration avec les options --with-dir-user et/ou --with-dir-group de la commande ./configure, le répertoire de travail Working Directory doit appartenir doit appartenir à ce groupe et à cet utilisateur.
Typiquement, sur les systèmes Linux, vous utiliserez ici /var/run. Si vous n'installez pas Bacula dans les répertoires système, vous pouvez utiliser le répertoire de travail Working Directory défini plus haut. Cette directive est requise.
Où <nombre> est le nombre maximal de jobs qui peuvent être exécutés simultanément par le Director. La valeur par défaut est 1, mais vous pouvez utiliser une valeur plus grande. Notez que le format des volumes devient beaucoup plus compliqué avec plusieurs jobs exécutés simultanément. De ce fait, les restaurations peuvent prendre beaucoup plus de temps si Bacula doit faire le tri parmi les segments entremélés de ces jobs. Ceci peut être évité en s'arrangeant pour que chacun des jobs exécutés simultanément écrive sur un volume distinct. Une autre possibilité consiste à utiliser le data spooling : les données seront d'abord "spoolées" sur disque simultanément, ensuite les fichiers "spool" seront écrits séquentiellement sur le volume.
Dans certains cas, des directives telles que Maximum Volume Jobs ne sont pas correctement synchronisées avec le nombre de jobs simultanés, et des problèmes de synchronisation subtils peuvent survenir, aussi des tests minutieux sont recommandés.
Actuellement, il n'y a aucun paramètre de configuration pour régler ou limiter le nombre de connections par console. Un maximum de cinq connection simultanées est autorisé.
Pour plus de détails concernant l'exécution simultanée de plusieurs jobs, consultez la partie Exécution simultanée de plusieurs jobs du chapitre Astuces de ce manuel.
DirAddresses = { ip = { addr = 1.2.3.4; port = 1205;} ipv4 = { addr = 1.2.3.4; port = http;} ipv6 = { addr = 1.2.3.4; port = 1205; } ip = { addr = 1.2.3.4 port = 1205 } ip = { addr = 1.2.3.4 } ip = { addr = 201:220:222::2 } ip = { addr = bluedot.thun.net } }
où "ip", "ip4", "ip6", "addr", et "port sont les mots clef. Notez que les adresses peuvent être spécifiées sous forme de quadruplets pointés, ou suivant la notation à doubles points IPv6, ou encore sous forme de nom symbolique (seulement pour la spécification ip). D'autre part, le port peut être spécifié par un nombre, ou par une valeur mnémonique du fichier /etc/services. Si un port n'est pas précisé, celui par défaut sera utilisé. Si une section ip est spécifiée, la résolution peut être faite soit par IPv4, soit par IPv6. Si ip4 est spécifié, seules les résolutions IPv4 seront permises. Il en va de même avec ip6.
Notez que si vous utilisez la directive DirAddresses, vous ne devez utiliser ni la directive DirPort, ni la directive DirAddress dans la même ressource.
Voici un exemple d'une ressource Director valide :
Director { Name = HeadMan WorkingDirectory = "$HOME/bacula/bin/working" Password = UA_password PidDirectory = "$HOME/bacula/bin/working" QueryFile = "$HOME/bacula/bin/query.sql" Messages = Standard }
La ressource Job définit un Job (sauvegarde, restauration,...) que Bacula doit exécuter. Chaque définition de ressource Job contient le nom d'un client, la liste des éléments à sauvegarder (FileSet), la planification (Schedule) pour ce Job, le lieu où sauvegarder ces données (Storage Device) et quel groupe de media utiliser (Pool). En effet, chaque ressource Job doit répondre aux questions : "Quoi ?", "Où ?", "Quand ?" et "Comment ?" soit, respectivement Fileset, Storage, Schedule, Type et Niveau (Sauvegarde/Restauration - Full/Differentielle/Incrémentale). Notez que le FileSet doit être spécifié lors des restaurations pour des raisons historiques, mais il n'est plus utilisé.
Un seul type (Backup, Restore, ...) peut être spécifié pour un Job donné. Si vous voulez sauvegarder plusieurs FileSets sur le même client, vous devez définir un Job pour chacun d'entre eux.
Lors de l'exécution d'un Job, son nom unique est composé du nom que vous avez spécifié ici suffixé avec la date et l'heure de sa planification. Cette directive est requise.
Pour un job de type Backup le niveau doit être l'un des suivants :
Si toutes les conditions ci-dessus ne sont pas réalisées, le Director augmentera la sauvegarde incrémentale en une sauvegarde Full. Dans le cas contraire, la sauvegarde incrémentale sera effectuée normalement.
Le File Daemon (Client) détermine les fichiers à sauvegarder pour une incrémentale par comparaison de l'heure de démarrage du Job précédent (Full, Différentiel ou Incrémental) avec les dates de dernière modification de chaque fichier (st_mtime) et de ses attributs (st_ctime). Si le fichier ou ses attributs ont changés depuis cette date de démarrage, alors le fichier sera sauvegardé.
Veuillez noter que certains logiciels anti-virus peuvent modifier la date
st_time lors de leurs opérations de scan. Ainsi, si l'antivirus modifie
la date d'accès (st_atime), qui n'est pas utilisée par Bacula, il
provoquera une modification du st_ctime et conduira Bacula à sauvegarder
les fichiers concernés lors des incrémentales et différentielles. Dans le
cas de l'antivirus Sophos, vous pouvez éviter cet inconvénient en utilisant
l'option --
no-reset-atime. Pour les autres logiciels, voyez
leurs manuels.
Lorsque Bacula effectue une sauvegarde incrémentale, tous les fichiers modifiés présents sur le système sont sauvegardés. Cependant, tout fichier supprimé depuis la dernière Full demeure dans le catalogue, ce qui signifie que si vous effectuez une restauration à partir de sauvegardes incrémentales (et de la Full associée), les fichiers supprimés depuis la dernière Full seront aussi restaurés. Ces fichiers n'apparaîtront plus dans le catalogue après avoir fait une nouvelle sauvegarde Full. Le processus pour supprimer ces fichiers du catalogue lors d'une incrémentale ralentirait fortement les sauvegardes incrémentales. Il n'est actuellement pas implémenté dans Bacula.
De plus, si vous déplacez un répertoire plutôt que de le copier, les fichiers qu'il contient voient leurs dates de dernière modification (st_mtime) et de dernier accès (st_ctime) inchangés. Par conséquent, ces fichiers ne seront probablement sauvegardés par aucune incrémentale ou différentielle, puisque ces dernières ne se réfèrent qu'à ces indicateurs. Aussi, il est préférable de copier un dossier avant de supprimer l'original plutôt que de le déplacer, si vous voulez qu'il soit correctement sauvegardé.
Si toutes les conditions ci-dessus ne sont pas réalisées, le Director augmentera la sauvegarde différentielle en une sauvegarde Full. Dans le cas contraire, la sauvegarde différentielle sera effectuée normalement.
Le File Daemon (Client) détermine les fichiers à sauvegarder pour une différentielle par comparaison de l'heure de démarrage de la dernière sauvegarde Full avec les dates de dernière modification de chaque fichier (st_mtime) et de ses attributs (st_ctime). Si le fichier ou ses attributs ont changés depuis cette date de démarrage, alors le fichier sera sauvegardé. La date de démarrage utilisée est affiché après le Since du rapport de Job. Dans de rares cas, certains fichiers sont sauvegardés deux fois à cause de l'utilisation de la date de démarrage de la sauvegarde précédente, mais ceci assure qu'aucun changement n'est perdu. Comme pour les incrémentales, vous devriez vous assurer que les horloges de votre serveur Bacula et de vos clients sont synchronisées, ou aussi proches que possible, pour éviter le risque d'omission d'un fichier. Notez qu'à partir de la version 1.33, Bacula effectue automatiquement ces ajustements de sorte que les horloges utilisées par Bacula soient synchrones.
Veuillez noter que certains logiciels anti-virus peuvent modifier la date
st_time lors de leurs opérations de scan. Ainsi, si l'antivirus modifie
la date d'accès (st_atime), qui n'est pas utilisée par Bacula, il
provoquera une modification du st_ctime et conduira Bacula à sauvegarder
les fichiers concernés lors des incrémentales et différentielles. Dans le
cas de l'antivirus Sophos, vous pouvez éviter cet inconvénient en utilisant
l'option --
no-reset-atime. Pour les autres logiciels, voyez
leurs manuels.
Lorsque Bacula effectue une sauvegarde différentielle, tous les fichiers modifiés présents sur le système sont sauvegardés. Cependant, tout fichier supprimé depuis la dernière Full demeure dans le catalogue, ce qui signifie que si vous effectuez une restauration à partir de sauvegardes différentielles (et de la Full associée), les fichiers supprimés depuis la dernière Full seront aussi restaurés. Ces fichiers n'apparaîtront plus dans le catalogue après avoir fait une nouvelle sauvegarde Full. Le processus pour supprimer ces fichiers du catalogue lors d'une incrémentale ralentirait fortement les sauvegardes différentielles. Il n'est actuellement pas implémenté dans Bacula, mais plannifié pour une future version de Bacula.
Comme noté ci-dessus, si vous déplacez un répertoire plutôt que de le copier, les fichiers qu'il contient voient leurs dates de dernière modification (st_mtime) et de dernier accès (st_ctime) inchangés. Par conséquent, ces fichiers ne seront probablement sauvegardés par aucune incrémentale ou différentielle, puisque ces dernières ne se réfèrent qu'à ces indicateurs. Aussi, il est préférable de copier un dossier avant de supprimer l'original plutôt que de le déplacer, si vous voulez qu'il soit correctement sauvegardé.
Régulièrement, quelqu'un demande à quoi servent les sauvegardes différentielles du moment que les incrémentales récupèrent tous les fichiers modifiés. Il existe plusieurs réponses à cette question, mais la plus importante à mes yeux est de combiner toutes les incrémentales et différentielles depuis la dernière full en une seule différentielle. Ceci a deux effets : 1. La redondance. 2. Plus important, la réduction du nombre de volumes requis pour faire une restauration en éliminant la nécessité de lire tous les volumes des précédentes incrémentales depuis la dernière full.
Pour un Job de type Restore, aucun niveau ne doit être spécifié.
Pour un Job de type Verify, le niveau peut être l'un des suivants :
Attention ! Si vous exécutez deux jobs Verify Catalog simultanément sur le même client, les résultats seront probablement erronnés. En effet, Verify Catalog modifie le catalogue lors de son exécution afin de détecter les nouveaux fichiers.
Attention ! Si vous exécutez deux jobs Verify VolumeToCatalog simultanément sur le même client, les résultats seront probablement erronnés. En effet, Verify VolumeToCatalog modifie le catalogue lors de son exécution afin de détecter les nouveaux fichiers.
Cette commande peut se révéler très utile si vous avez des problèmes de disque car elle comparera l'état actuel de votre disque avec la dernière sauvegarde valide, qui peut remonter à plusieurs jobs.
Notez que l'implémentation actuelle (1.32c) n'identifie pas les fichiers qui ont été supprimés.
Si vous utilisez la commande Restore dans la console pour lancer une restauration, le fichier bootstrap sera créé automatiquement à partir des fichiers que vous avez sélectionnés pour la restauration.
Pour plus de détails concernant les fichiers bootstrap, veuillez consulter le chapitre Restaurer des fichiers avec le fichier Bootstrap de ce manuel.
En utilisant cette fonction, vous aurez constamment un fichier bootstrap capable de recouvrer l'état le plus récent de votre système. Le fichier bootstrap devrait être écrit sur un disque monté sur une autre machine, de sorte que vous puissiez en disposer immédiatement en cas de défaillance de votre disque dur. Une alternative consiste à copier le fichier sur une autre machine après chaque mise à jour.
Si la spécification de fichier bootstrap débute par une barre verticale (|), Bacula considère la spécification comme un nom de programme vers lequel les les enregistrement bootstrap seront redirigés. Ce peut être, par exemple, un script qui vous envoie par e-mail les enregistrements bootstrap.
A partir de la version 1.40, Bacula effectue des character substitution comme dans une directive RunScript. Pour gérer automatiquement vos fichiers bootstrap, vous pouvez utiliser ceci dans vos JobDefs :
JobDefs { Write Bootstrap = "%c_%n.bsr" ... }
Pour plus de détails sur l'utilisation de fichiers bootstrap, veuillez consulter le chapitre intitulé Le Fichier Bootstrap de ce manuel.
Si cette directive est désactivée, le Storage daemon privilégiera les lecteurs inutilisés. Ce mode de fonctionnement peut être très utile pour ces sites avec de nombreux lecteurs qui où il peut être préférable de maximiser le flux des sauvegardes au prix d'une utilisation d'un plus grand nombre de lecteurs et de cartouches. Afin d'optimiser l'utilisation de plusieurs lecteurs, vous voudrez probablement lancer chacun de vos jobs l'un après l'autre avec un intervalle de 5 secondes environ. Ceci aidera à assurer que chaque nuit, le même lecteur (volume) est sélectionné pour le même job. Autrement, lors d'une restauration, vous pourriez trouver vos fichiers dispersés sur beaucoup plus de volumes que nécessaire.
La commande spécifiée est exécutée en tant que programme externe avant le
lancement du job. Cette directive est optionnelle.
Vous disposez des options suivantes :
Options | Valeur | Defaut | Informations |
Runs On Success | Yes/No | Yes | La commande est exécutée si le job s'est terminé avec succès |
Runs On Failure | Yes/No | No | La commande est exécutée si le job a échoué |
Runs On Client | Yes/No | Yes | La commande est exécutée sur le client |
Runs When | Before|After|Always | Never | Le moment où la commande doit être exécutée |
Abort Job On Error | Yes/No | Yes | Abandonne le job si le script retourne un résultat non nul |
Command | Le chemin vers votre script |
D'autre part, la chaîne bf command est parcourue puis envoyée vers la fonction execvp(), ce qui signifie que le chemin de la commande est recherché pour son exécution, mais qu'il n'y a aucune interprétation shell. Par conséquent, si vous voulez utiliser des commandes complexes ou toute fonctionnalité du shell telle que la redirection, vous devez appeler un script shell où vous mettrez vos commandes.
Avant de soumettre la commande spécifiée au système d'exploitation, Bacula effectue les substitutions suivantes :
%% = % %c = Nom du client %d = Nom du Director %e = Statut de sortie du job %i = JobId %j = Nom unique du job %l = Niveau du job %n = Nom du job %s = Temps Depuis (NDT : Since Time) %t = Type de job (Backup,...) %v = Nom de volume
Le code de statut de fin de job peut prendre les valeurs suivantes :
Aussi, si vous l'utilisez dans une ligne de commande, il vous faudra l'encadrer de quotes.
Vous disposez des raccourcis suivants :
Mot clef | RunsOnSuccess | RunsOnFailure | AbortJobOnError | Runs On Client | RunsWhen |
Run Before Job | Yes | No | Before | ||
Run After Job | Yes | No | No | After | |
Run After Failed Job | No | Yes | No | After | |
Client Run Before Job | Yes | Yes | Before | ||
Client Run After Job | Yes | No | Yes | After | |
Client Run After Failed Job | No | Yes | Yes | After |
Exemple :
RunScript { RunsWhen = Before AbortJobOnError = No Command = "/etc/init.d/apache stop" } RunScript { RunsWhen = After RunsOnFailure = yes Command = "/etc/init.d/apache start" }
Considérations particulières à Windows D'autre part, pour les clients Windows à partir de la version 1.33, notez bien que vous devez fournir un chemin correct pour votre script, et que le script peut avoir l'extension .com, .exe, ou .bat. Si vous spécifiez un chemin, vous devez aussi spécifier l'extension complète. Les commandes à la façon d'Unix ne fonctionneront pas, à moins que vous n'ayez installé et correctement configuré Cygwin en plus (et séparément) de Bacula.
La commande peut être n'importe quel programme reconnu par cmd.exe ou command.com comme un fichier exécutable. Spécifier une extension de fichier exécutable est optionnel, à moins qu'il y ait une ambiguïté (par exemple ls.bat, ls.exe).
Bacula cherche la commande dans le répertoire "System %Path%" (Dans la boîte de dialogue des variables d'environnement vous avez les variables "système" et "utilisateurs". Si bacula-fd fonctionne en tant que service, seules les variables d'environnement systèmes sont accessibles.)
Les variables d'environnement système peuvent être invoquées avec la syntaxe %var% et utilisées comme portion du nom de la commande ou des arguments.
Lorsque la spécification du chemin absolu d'un exécutable ou le nom de l'exécutable contient des espaces ou des caractères spéciaux, ils doivent être quotés. Il en va de même pour les arguments.
ClientRunBeforeJob = "\"C:/Program Files/Software Vendor/Executable\" /arg1 /arg2 \"foo bar\""
Les caractères spéciaux &()[]{}^=;!'+,`û devront être quotés s'ils font partie d'un nom de fichier ou d'un argument.
If someone is logged in a blank ``command'' window running the commands will be present during the execution of the command.
Quelques suggestions de Phil Stracchino pour l'exécution sur les machines Win32 avec le File Daemon Win32 natif :
ClientRunBeforeJob = ``c:/bacula/bin/systemstate.bat''
plutôt qu'au format DOS/Windows :
ClientRunBeforeJob = ``c:\bacula\bin\systemstate.bat'' INCORRECT
L'exemple suivant d'utilisation de la directive Client Run Before Job a été soumis
par un utilisateur :
Vous pourriez écrire un script shell pour sauvegarder une base DB2 dans un FIFO. Voici
le script en question :
#!/bin/sh # ===== backupdb.sh DIR=/u01/mercuryd mkfifo $DIR/dbpipe db2 BACKUP DATABASE mercuryd TO $DIR/dbpipe WITHOUT PROMPTING & sleep 1
La ligne suivante dans la ressource Job du fichier bacula-dir.conf :
Client Run Before Job = "su - mercuryd -c \"/u01/mercuryd/backupdb.sh '%t' '%l'\""
Lorsque le job est exécuté, vous obtiendrez un message de sortie du script annonçant que la sauvegarde a démarré. Même si la commande est exécutée en arrière plan avec &, le job bloquera jusqu'à la commande "db2 BACKUP DATABASE", et la sauvegarde se fige.
Pour remédier à cette situation, la ligne "db2 BACKUP DATABASE" devrait être modifiée en :
db2 BACKUP DATABASE mercuryd TO $DIR/dbpipe WITHOUT PROMPTING > $DIR/backup.log 2>&1 < /dev/null &
Il est important de rediriger l'entrée et la sortie d'une commande en arrière plan vers /dev/null pour éviter le bloquage du script.
Run Before Job = "echo test"est équivalent á :
RunScript { Command = "echo test" RunsOnClient = No RunsWhen = Before }
Lutz Kittler a fait remarquer que ceci peut être un moyen aisé pour modifier vos schedules pour les vacances. Par exemple, supposons que vous fassiez habituellement des sauvegardes Full le vendredi, mais que jeudi et vendredi soient fériés. Pour éviter d'avoir à changer les cartouches entre jeudi et vendredi alors que personne n'est au bureau, vous pouvez créer un RunBeforeJob qui retourne un statut non nul jeudi et zéro les autres jours. Ainsi, le job de jeudi ne sera pas exécuté, et la cartouche que vous avez inséré mercredi sera disponible pour la Full de vendredi.
Un exemple d'utilisation de cette directive est donné au chapitre Astuces de ce manuel. Lisez le paragraphe Run After Failed Job si vous voulez exécuter une commande lorqu'un job se termine avec un statut anormal.
RunScript Command = "echo test" RunsWhen = After RunsOnFailure = yes RunsOnClient = no RunsOnSuccess = yes # default, you can drop this line
Le chapitre Trucs et astuces de ce manuel propose un exemple d'utilisation de cette directive.
Cette spécification peut se révéler utile pour les pc portables ainsi que pour toutes les machines qui ne sont pas connectées au réseau en permanence.
run = "Nightly-backup level=%s since=\"%s\" storage=DDS-4"Un job cloné ne démarrera pas de nouveaux clones, aussi il n'est pas possible de les cascader.
La priorité affecte seulement les jobs en file d'attente, et non les jobs déja en cours d'exécution. Si un ou plusieurs jobs de priorité 2 sont déjà en cours d'exécution, et si un nouveau job est programmé avec la priorité 1, les jobs en cours d'exécution doivent se terminer pour que le job de priorité 1 puisse démarrer.
La priorité par défaut est 10.
Si vous voulez exécutez plusieurs jobs simultanés, vous devriez garder les points suivants à l'esprit :
Si vous avez plusieurs jobs de priorités différentes, il est préférable de ne pas les démarrer exactement à la même heure, car Bacula doit les examiner un à la fois. Si, par hazard, Bacula commence par traiter un job de priorité inférieure, il sera exécuté avant votre job de priorité élevé. Pour éviter cette situation, démarrez l'un quelconque des jobs de priorité élevée quelques secondes avant ceux de basse priorité. Ainsi, vous serez assuré que Bacula examine les jobs dans l'ordre voulu et que votre schéma de priorités sera respecté.
Cette directive devrait être activée lors de l'écriture sur des périphérique qui requièrent un montage (par exemple, les DVDs), afin de vous assurer que le fichier partition courant, celui qui contient les données de ce job, est envoyé vers le périphérique, et qu'aucune donnée n'est laissée dans le fichier temporaire sur le disque dur. Quoi qu'il en soit, avec certains supports tels que les DVD+R et DVD-R, beaucoup d'espace (environ 10 Mb) est perdu à chaque fois qu'un fichier partition est écrit. Aussi, si vous exécutez plusieurs jobs à la suite, vous devriez régler cette directive à no pour tous ces jobs sauf le dernier, pour éviter un gaspillage important d'espace, tout en ayant la certitude que les données sont bien écrites sur le médium lorsque tous les jobs sont achevés.
Cette directive est ignorée avec les bandes et les périphériques FIFO.
Voici un exemple de définition de ressource Job valide.
Job { Name = "Minou" Type = Backup Level = Incremental # default Client = Minou FileSet="Minou Full Set" Storage = DLTDrive Pool = Default Schedule = "MinouWeeklyCycle" Messages = Standard }
La ressource Jobdefs admet toutes les directives qui peuvent apparaître dans une ressource Job. Une ressource Jobdefs ne créé en aucun cas un Job, son rôle est de pouvoir être désignée dans une ressource Job comme un ensemble de paramètres par défaut. Ceci permet de définir plusieurs jobs similaires avec concision, en ne mentionnant, pour chaque job, que les différences avec les valeurs par défaut spécifiées dans la ressource Jobdefs.
La ressource Schedule offre un moyen pour planifier automatiquement un Job, mais aussi la possibilité de surcharger les paramètres par défaut de Level, Pool, Storage, et Messages ressources. Si une ressource Schedule n'est pas spécifiée dans un job, ce job ne peut être exécuté que manuellement. En général, vous spécifierez une action et le moment de son lancement.
La directive Job-overrides permet d'outrepasser les spécifications de Level, Storage, Messages et Pool écrites dans la ressource Job. De plus, les spécifications FullPool, DifferentialPool et IncrementalPool permettent de passer outre les spécification de Pool, en accord avec le niveau (level) effectif d'exécution du job.
L'utilisation de surcharges permet de peaufiner le paramétrage d'un job particulier. Par exemple, vous pourriez surcharger une spécification Messages qui enverrait vos logs de backups vers un fichier, de façon à ce qu'ils vous soient envoyés par mails pour les Fulls hebdomadaires ou mensuelles.
Les directives Job-overrides sont spécifiées en tant que mot-clef=valeur où le mot-clef est l'un des suivants : Level, Storage, Messages, Pool, FullPool, DifferentialPool ou IncrementalPool, et la valeur est définie selon le format adapté à la directive. Vous pouvez spécifier plusieurs surcharges Job-overrides en une seule directive Run en les séparant par des espaces ou des trailing comas (traduction ?). Par exemple :
Spécifie l'utilisation du Pool nommé Full si le job est une sauvegarde Full, ou s'il a été élevé en Full bien qu'ayant été lancé en tant que différentiel ou incrémental.
Date-time-specification Détermine la planification d'exécution du job. La spécification est une répétition, et, par défaut, Bacula est paramétré pour exécuter un job au début de chaque heure de chaque jour de chaque semaine de chaque mois de chaque année. Ce n'est probablement pas ce que vous souhaitez, aussi vous devez préciser ou limiter les moments où vous souhaitez voir vos jobs exécutés. Toute spécification est supposée cyclique et servira à limiter le cycle par défaut. Ceci se fait en spécifiant des masques ou des horaires, jours de la semaine, jours du mois, semaines du mois, semaines de l'année et mois de l'année où vous voulez exécuter le job. En combinant ces possibilités, vous pouvez définir une planification qui se répète à presque n'importe quelle fréquence.
Concrètement, vous devez définir les mois, jour, heure et minute où le job est à exécuter. Parmis ces quatre objets, le jour est particulier en ce qu'il peut spécifier un jour du mois (1,2,...31) ou de la semaine (Monday, Tuesday,...Sunday). Enfin, vous pouvez aussi spécifier un jour de la semaine pour restreindre la planification à la première, deuxième, troisième, quatrième ou cinquième semaine du mois.
Par exemple, si vous spécifiez seulement un jour de la semaine, disons Mardi, le job sera exécuté toutes les heures de chaque mardi de chaque mois. La raison en est que les paramètres Mois et Heure sont restés à leurs valeurs par défaut : chaque mois et chaque heure.
Notez que, par défaut, sans autre spécification, votre job s'exécutera au début de chaque heure. Si vous souhaitez que votre job s'exécute plus souvent qu'une fois par heure, il vous faudra définir plusieurs spécifications run avec pour chacune une minut différente.
Les dates et horaires d'exécutions des jobs peuvent être spécifiés comme suit, en pseudo-BNF :
<void-keyword> = on <at-keyword> = at <week-keyword> = 1st | 2nd | 3rd | 4th | 5th | first | second | third | forth | fifth <wday-keyword> = sun | mon | tue | wed | thu | fri | sat | sunday | monday | tuesday | wednesday | thursday | friday | saturday <week-of-year-keyword> = w00 | w01 | ... w52 | w53 <month-keyword> = jan | feb | mar | apr | may | jun | jul | aug | sep | oct | nov | dec | january | february | ... | december <daily-keyword> = daily <weekly-keyword> = weekly <monthly-keyword> = monthly <hourly-keyword> = hourly <digit> = 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 0 <number> = <digit> | <digit><number> <12hour> = 0 | 1 | 2 | ... 12 <hour> = 0 | 1 | 2 | ... 23 <minute> = 0 | 1 | 2 | ... 59 <day> = 1 | 2 | ... 31 <time> = <hour>:<minute> | <12hour>:<minute>am | <12hour>:<minute>pm <time-spec> = <at-keyword> <time> | <hourly-keyword> <date-keyword> = <void-keyword> <weekly-keyword> <day-range> = <day>-<day> <month-range> = <month-keyword>-<month-keyword> <wday-range> = <wday-keyword>-<wday-keyword> <range> = <day-range> | <month-range> | <wday-range> <date> = <date-keyword> | <day> | <range> <date-spec> = <date> | <date-spec> <day-spec> = <day> | <wday-keyword> | <day-range> | <wday-range> | <week-keyword> <wday-keyword> <day-range> | <wday-range> | <daily-keyword> <month-spec> = <month-keyword> | <month-range> | <monthly-keyword> <date-time-spec> = <month-spec> <day-spec> <time-spec>
Notez que les spécifications de semaine et d'année suivent les définitions ISO standard de semaine et année, où la semaine 1 est la semaine qui contient le premier jeudi de l'année, ou alternativement, la semaine qui contient le quatrième jour de janvier. Les semaines sont numérotées w01 à w53. w00 est pour Bacula la semaine qui précède la première semaine ISO (c'est à dire celle qui contient les quelques premiers jours de l'année si aucun n'est un jeudi). w00 n'est pas définie dans les spécifications ISO. Une semaine commence le Lundi et se termine le Dimanche.
Voici un exemple de ressource Schedule nommée WeeklyCycle qui exécute un job de niveau Full chaque Dimanche à 1h05 et un job de niveau incrémental du Lundi au Samedi à 1h05 :
Schedule { Name = "WeeklyCycle" Run = Level=Full sun at 1:05 Run = Level=Incremental mon-sat at 1:05 }
Voici un exemple de cycle mensuel :
Schedule { Name = "MonthlyCycle" Run = Level=Full Pool=Monthly 1st sun at 1:05 Run = Level=Differential 2nd-5th sun at 1:05 Run = Level=Incremental Pool=Daily mon-sat at 1:05 }
Le premier de chaque mois :
Schedule { Name = "First" Run = Level=Full on 1 at 1:05 Run = Level=Incremental on 2-31 at 1:05 }
Toutes les dix minutes :
Schedule { Name = "TenMinutes" Run = Level=Full hourly at 0:05 Run = Level=Full hourly at 0:15 Run = Level=Full hourly at 0:25 Run = Level=Full hourly at 0:35 Run = Level=Full hourly at 0:45 Run = Level=Full hourly at 0:55 }
Pour chacun de vos schedules, vous pouvez visualiser le masque associé grâce à la commande show schedules du programme Console. Notez que le bit masque est "zero based", et que Dimanche est le premier jour de la semaine (bit 0)
La ressource FileSet définit les fichiers à inclure dans une sauvegarde. Pour chaque job de type sauvegarde, il est nécessaire de définir au moins une ressource FileSet. Un FileSet consiste en une liste de fichiers ou répertoires à inclure, une liste de fichiers ou répertoires à exclure, et diverses options de sauvegardes telles que compression, chiffrement et signatures qui doivent être appliquées à chaque fichier.
Toute modification de la liste des fichiers inclus provoque la création par Bacula d'un nouveau FileSet (défini par le nom et la somme de contrôle MD5 du contenu du paragraphe Include). Chaque fois qu'un nouveau FileSet est créé, Bacula s'assure que la première sauvegarde est une Full.
Le nombre de ressources Include par FileSet n'est pas limité, chacune ayant sa propre liste de répertoires et/ou fichiers à sauvegarder et ses propres paramètres définis par une ou plusieurs ressources Options. La liste de fichiers file-list consiste en un nom de fichier ou répertoire par ligne. Les noms de répertoire doivent être spécifiés sans slash final.
Vous devez toujours spécifier des chemins absolus pour tout fichier ou répertoire que vous placez dans un FileSet. De plus, sur les machines Windows, vous devez toujours préfixer le répertoire ou nom de fichier d'une spécification de disque (par exemple : c:/xxx) en utilisant le séparateur de répertoire Unix (slash /).
Le comportement par défaut de Bacula en ce qui concerne le traitement des répertoires est de descendre récursivement dans chaque répertoire et de sauvegarder tous les fichiers et sous-répertoires. Par défaut, Bacula ne suit pas les systèmes de fichiers transverses (en terminologie Unix, les points de montage). Ceci signifie que si vous spécifiez la partition racine ( par exemple /), Bacula sauvegardera seulement la partition racine, et aucun des systèmes de fichiers montés. De façon analogue, sur les systèmes Windows, vous devez expliciter chacun des disques que vous souhaitez sauvegarder (par exemple c:/ et d:/...). De plus, au moins pour les systèmes Windows, il sera la plupart du temps nécessaire d'encadrer chaque spécification de doubles quotes, particulièrement si le nom du répertoire (ou du fichier) comporte des espaces. La commande df des systèmes Unix vous fournira la liste des répertoires qu'il vous faudra spécifier pour tout sauvegarder. Voyez ci-dessous pour un exemple.
Soyez attentif à ne pas inclure un répertoire deux fois, car il serait sauvegardé deux fois, ce qui gaspillerait l'espace sur votre périphérique de sauvegarde. Cette erreur est facile à commettre. Par exemple :
Include { File = / File = /usr Options { compression=GZIP } }
Sur un système Unix où /usr est un sous répertoire (plutôt qu'un système de fichiers monté), cette ressource Include sauvegarderait /usr deux fois. Dans ce cas, sur les versions antérieures à 1.32f-5-09Mar04, en raison d'un bug, vous ne pourriez restaurer les fichiers liés physiquement sauvegardés deux fois.
Si vous avez utilisé des versions de Bacula antérieures à 1.34.3, vous noterez ces modifications dans la syntaxe des FileSets :
La ressource Options est optionnelle, mais lorsqu'elle est spécifiée, elle doit contenir une liste de lignes "mot-clef=valeur" relatives aux options à appliquer à la liste de fichiers/répertoires. Plusieurs ressources Options peuvent être spécifiées l'une après l'autre. Lorsqu'un fichier se trouve dans un dossier spécifié, les options sont appliquées au nom de fichier pour savoir s'il doit être sauvegardé, et comment. Les ressources Options sont appliquées dans l'ordre où elles apparaîssent dans le FileSet jusqu'à ce qu'il y en ait une qui corresponde. Une ressource Options qui ne contient pas de directive wild (spécification de caractère joker, voir ci-dessous) est considérée comme concernant tous les fichiers. Il est important de bien comprendre ceci, car une fois que Bacula a déterminé que des Options s'appliquent à un fichier donné, ce fichier sera sauvegardé sans tenir compte d'aucunes des éventuelles autres ressources Options. Ceci signifie que toute ressource Options avec caractères joker doit apparaître avant une ressource Options sans caractères joker.
Si, pour quelque raison, Bacula applique toutes les ressources Options à un fichier sans qu'aucune ne corresponde (en général à cause de caractères joker qui ne correspondent pas), par défaut Bacula sauvegardera le fichier. Ceci est assez logique si vous considérez la situation sans options, où vous souhaitez que tout soit sauvegardé. De plus, dans le cas ou aucune correspondance n'est trouvée, Bacula utilise les options de la dernière ressource Options. Par conséquent, si vous souhaitez définir un jeu d'options par défaut, vous devriez les placer dans la dernière ressource Options.
Les directives disponibles pour les ressources Options sont les suivantes :
La compression logicielle est particulièrement intéressante lorsque vous sauvegardez sur disque, et peut être d'un grand secours si vous avez un ordinateur rapide mais un réseau lent.
La spécification GZIP utilise le niveau de compression six par défaut (i.e. GZIP est équivalent à GZIP6). Si vous voulez utiliser un niveau différent (de 1 à 9), vous pouvez le spécifier en ajoutant le numéro du niveau voulu à la fin du mot GZIP, sans espace. Ainsi, compression=GZIP1 désigne la compression la moins efficace, mais l'algorithme le plus rapide, tandis que compression=GZIP9 est le niveau de compression le plus élevé, mais requière plus de puissance de calcul. Selon la documentation GZIP, les niveaux de compression supérieurs à 6 ne procurent généralement que peu de compression supplémentaire alors qu'ils sont plutôt exigeants en puissance de calcul.
Le jeu d'options pins5 (qui compare les bits de permissions, les inodes, les nombres de liens, la taille des fichiers et les signatures MD5) est très utile pour des jobs de type verify de niveaux Level=Catalog ou Level=DiskToCatalog.
Restrictions: Bacula lit les fichiers dans des tampons de 32K. Si le tampon entier est rempli de zéros, il sera traité en tant que bloc clairsemé, et ne sera pas écrit sur la cartouche. En revanche, si une partie quelconque du tampon est non-nulle, le tampon sera intégralement copié sur la cartouche, avec éventuellement des secteurs de disque (généralement 4098 octets) entièrement nuls. La détection par Bacula des blocs clairsemés a lieu sur des blocs de 32K plutôt que sur des blocs de taille déterminée par le système. Si quelqu'un considère ceci comme un réelle problème, merci d'envoyer une demande de modification en exposant les raisons. Ce code est apparu avec la version 1.27 de Bacula.
Si vous n'êtes pas familier avec les notions de fichiers clairsemés, prenons pour exemple un fichier où vous écrivez 512 octets à l'adresse 0, puis 512 octets à l'adresse 1 million. Le système d'exploitation n'allouera que deux blocs, et rien n'est alloué pour l'espace vide. Pourtant, lorsque vous lisez le fichier clairsemé, le système retourne tous les zéros comme si l'espace était alloué, et si vous sauvegardez un tel fichier, vous utiliserez beaucoup d'espace sur le volume pour écrire des zéros. Pire encore, lorsque vous restaurez ce fichier à son emplacement initial, tous les emplacements précédemment vides seront cette fois alloués, occupant ainsi beaucoup plus d'espace disque. En activant l'option sparse, Bacula recherchera spécifiquement l'espace vide dans les fichiers afin d'éviter ces inconvénients. Le prix à payer est que Bacula doit d'abord examiner chaque bloc lu avant de l'écrire. Sur un système lent, ceci peut-être important. Si vous suspectez certains de vos fichiers d'être clairsemés, vous devriez mesurer les performances et gains d'espace avec et sans l'options, ou ne l'activer que pour les fichiers effectivement clairsemés.
<file-list> est une liste de répertoires et/ou noms de fichiers spécifiés avec la directive File =. Pour inclure des noms contenant des espaces, entourez-les de guillemets (doubles quotes).
Il existe quelques notations particulières pour spécifier des fichiers et répertoires dans une liste de fichiers file-list. Les voici :
Par exemple :
Include { Options { signature = SHA1 } File = "|sh -c 'df -l | grep \"^/dev/hd[ab]\" | grep -v \".*/tmp\" \ | awk \"{print \\$6}\"'" }
produira une liste de toutes les partitions locales sur un système RedHat Linux. Notez que la ligne si dessus a été coupée, mais devrait normalement être écrite sur une seule ligne. Quoter est un réel problème car vous devez d'une part le faire pour Bacula - ce qui consiste à précéder tout \ et tout '' avec un \ - et d'autre part pour les commandes shell. En définitive, il est probablement plus aisé d'exécuter un petit fichier tel que :
Include { Options { signature=MD5 } File = "|my_partitions" }
où le fichier my_partitions contient :
#!/bin/sh df -l | grep "^/dev/hd[ab]" | grep -v ".*/tmp" \ | awk "{print \$6}"
Si la barre verticale (|) devant "my_partitions" est précédée d'une barre oblique (\), le programme sera exécuté sur la machine cliente plutôt que sur la machine hébergeant le Director -- (ceci est implémenté, mais n'est pas complètement testé, et a été rapporté fonctionner sous Windows). Veuillez noter que si le nom de fichier est donné entre quotes, vous devrez utiliser deux barres obliques. Voci un exemple, fourni par John Donagher, qui sauvegarde toutes les partitions UFS locales sur un système distant :
FileSet { Name = "All local partitions" Include { Options { signature=SHA1; onefs=yes; } File = "\\|bash -c \"df -klF ufs | tail +2 | awk '{print \$6}'\"" } }
Notez que deux barres obliques \ sont requises après les doubles quotes (l'une préserve l'autre). Si vous utilisez Linux, changez simplement ufs en ext3 (ou votre système de fichiers préféré) et l'affaire sera dans le sac.
Si vous précédez le signe "inférieur" (<) d'une barre oblique \<, le fichier mentionné sera lu sur la machine cliente au lieu de celle hébergeant le Director. Veullez noter que si le nom de fichier est donné entre quotes, il vous faudra utiliser deux barres obliques.
Include { Options { signature=MD5; sparse=yes } File = /dev/hd6 }
va sauvegarder les données de la partition /dev/hd6.
will backup the data in device /dev/hd6.
Ludovic Strappazon a fait remarquer que cette fonction pouvait servir à sauvegarder un disque Microsoft Windows. Il suffit de booter avec un Linux Rescue Disk, puis de charger un client Bacula statiquement lié comme décrit dans le chapitre Disaster Recovery avec Bacula de ce manuel. Sauvegardez alors la partition complète. En cas de désastre, vous pouvez alors restaurer la partition désirée en bootant une fois encore sur le Linux Rescue Disk et en utilisant le client Bacula statiquement lié.
Include { Options { signature=SHA1 readfifo=yes } File = /home/abc/fifo }
si /home/abc/fifo est un FIFO, Bacula va l'ouvrir, le lire, et stocker toutes les données ainsi obtenues sur le volume. Notez qu'il faut que vous ayez un processus qui écrit sur le FIFO, faute de quoi Bacula restera en suspens, et abandonnera au bout d'une minute pour passer au fichier suivant. Les données lues peuvent être de nature quelconque puisque Bacula les traite comme un flux.
Cette fonction est un excellent moyen de faire une sauvegarde "à chaud" d'une très grosse base de données. Vous pouvez utiliser la directive RunBeforeJob pour créer le FIFO et démarrer un programme qui lit dynamiquement votre base de données et l'écrit sur le FIFO. Bacula l'écrira alors sur le volume.
Lors de l'opération de restauration, l'inverse se produit : après que Bacula ait créé le FIFO, s'il y avait des données stockées par son biais (inutile de les lister explicitement ni d'ajouter aucune option), elles seront renvoyées vers le FIFO. Par conséquent, s'il existe un tel FIFO à restaurer, vous devez vous assurer qu'il y a un programme lecteur ou Bacula se bloquera et passera au fichier suivant après une minute.
Voici un exemple de définition de ressource FileSet valide. Notez que le premier Include insère le contenu du fichier /etc/backup.list lors du démarrage de Bacula (i.e. le @).
FileSet { Name = "Full Set" Include { Options { Compression=GZIP signature=SHA1 Sparse = yes } File = @/etc/backup.list } Include { Options { wild = *.o Exclude = yes } File = /root/myfile File = /usr/lib/another_file } }
Notez que dans l'exemple ci-dessus, tous les fichiers mentionnés dans /etc/backup.list seront compressé avec GZIP, qu'une signature SHA1 sera calculée sur le contenu des fichiers (leurs données), et que la prise en charge particulière des fichiers clairsemés (sparse) s'appliquera.
Les deux répertoires /root/myfile et /usr/lib/another_file seront aussi sauvegardés sans aucune option, mais tous les fichiers à extension .o de ces répertoires seront exlus de la sauvegarde.
Supposons que vous vouliez sauvegarder tout sauf /tmp sur votre système. La commande df vous fournit le résultat suivant :
[kern@rufus k]$ df Filesystem 1k-blocks Used Available Use% Mounted on /dev/hda5 5044156 439232 4348692 10% / /dev/hda1 62193 4935 54047 9% /boot /dev/hda9 20161172 5524660 13612372 29% /home /dev/hda2 62217 6843 52161 12% /rescue /dev/hda8 5044156 42548 4745376 1% /tmp /dev/hda6 5044156 2613132 2174792 55% /usr none 127708 0 127708 0% /dev/shm //minimatou/c$ 14099200 9895424 4203776 71% /mnt/mmatou lmatou:/ 1554264 215884 1258056 15% /mnt/matou lmatou:/home 2478140 1589952 760072 68% /mnt/matou/home lmatou:/usr 1981000 1199960 678628 64% /mnt/matou/usr lpmatou:/ 995116 484112 459596 52% /mnt/pmatou lpmatou:/home 19222656 2787880 15458228 16% /mnt/pmatou/home lpmatou:/usr 2478140 2038764 311260 87% /mnt/pmatou/usr deuter:/ 4806936 97684 4465064 3% /mnt/deuter deuter:/home 4806904 280100 4282620 7% /mnt/deuter/home deuter:/files 44133352 27652876 14238608 67% /mnt/deuter/files
Si vous vous contentez de spécifier / dans votre liste d'inclusions, Bacula ne sauvegardera que le système de fichiers /dev/hda5. Pour sauvegarder tous vos systèmes de fichiers sans inclure les systèmes de fichiers montés Samba ou NFS et en excluant /tmp, /proc, .journal, et .autofsck, que vous ne voulez ni sauvegarder ni restaurer, vous pouvez utiliser ce qui suit :
FileSet { Name = Include_example Include { Options { wild = /proc wild = /tmp wild = \.journal wild = \.autofsck exclude = yes } File = / File = /boot File = /home File = /rescue File = /usr } }
/tmp étant sur son propre système de fichiers et n'étant pas explicitement nommé dans la liste d'inclusion, il n'est pas nécessaire de le spécifier dans la liste d'exclusion. Cependant, il peut être préférable de le faire malgré tout par souci de clarté et au cas où il ne serait plus sur sa propre partition après un remplacement de disques.
Ayez conscience qu'il peut être très dangereux de permettre à Bacula de traverser ou changer de système de fichiers au gré des ppoints de montage. Par exemple, avec ce qui suit :
FileSet { Name = "Bad example" Include { Options { onefs=no } File = /mnt/matou } }
vous sauvegardez une partition NFS montée (/mnt/matou), et puisque onefs est désactivée, Bacula traverse les systèmes de fichiers. Si jamais /mnt/matou contient lui même un point de montage où le système de fichiers de la machine sauvegardée est monté, ce qui est souvent le cas, vous vous retrouvez pris dans un boucle récursive, et la sauvegarde ne se terminera jamais.
Le FileSet suivant sauvegarde une partition raw :
FileSet { Name = "RawPartition" Include { Options { sparse=yes } File = /dev/hda2 } }
Lorsque vous sauvegardez et restaurez une partition raw, vous devriez vous assurer qu'aucun autre processus, y compris le système, n'écrit sur cette partition. En guise de précaution, nous recommandons ardemment de ne sauvegarder en mode raw que des partitions non montées, ou montées en lecture seule. Ceci peut être fait si nécessaire avec la directive RunBeforeJob.
Si vous saisissez des noms de fichiers Windows, les chemins des répertoires devraient être précédés de double-points (comme dans "c:"). Cependant, les séparateurs de champs doivent être spécifiés selon la convention Unix (c'est à dire, la barre oblique avant : "/"). Si vous souhaitez inclure une apostrophe dans un nom de fichier, précédez-la d'une barre oblique arrière (\\). Par exemple, vous pourriez utiliser ce qui suit pour sauvegarder le répertoire "My Documents" d'une machine Windows :
FileSet { Name = "Windows Set" Include { Options { wild = *.obj wild = *.exe exclude = yes } File = "c:/My Documents" } }
Pour que les listes d'exclusion fonctionnent correctement sous Windows, vous devez observer les règles suivante :
Merci à Thiago Lima pour nous avoir résumé les points ci-dessus. Si vous rencontrez des difficultés pour faire fonctionner vos listes d'inclusion ou d'exclusion, songez à utiliser la commande estimate job=xxx listing documentée dans le chapitre Console chapter de ce manuel.
Sur les systèmes Win32, si vous déplacez un répertoire ou si vous renommez un fichier de la liste à sauvegarder, et si une Full a déjà eu lieu, Bacula ne saura reconnaître qu'il existe de nouveaux fichiers à sauvegarder lors d'une incrémentale ou d'une différentielle (faites-en le reproche à Microsoft, pas à moi !). Pour pallier à ce problème, veuillez copier tout répertoire ou fichier de la zone sauvegardée. Si vous ne disposez pas de suffisamment d'espace disque, déplacez-les, mais lancez alors une sauvegarde Full.
Vous pouvez aussi inclure des noms de fichiers ou chemins absolus, en plus de l'utilisation de caractères jokers et de la directive Exclude=yes dans les ressources Options comme exposé ci-dessus, en ajoutant simplement les fichiers à exclure dans une ressource Exclude du FileSet. Par exemple :
FileSet { Name = Exclusion_example Include { Options { Signature = SHA1 } File = / File = /boot File = /home File = /rescue File = /usr } Exclude { File = /proc File = /tmp File = .journal File = .autofsck } }
Cet exemple est une contribution de Phil Stracchino :
This is my Windows 2000 fileset: FileSet { Name = "Windows 2000 Full Set" Include { Options { signature=MD5 } File = c:/ } Exclude { # Most of these files are excluded not because we don't want # them, but because Win2K won't allow them to be backed up # except via proprietary Win32 API calls. File = "/Documents and Settings/*/Application Data/*/Profiles/ */*/Cache/*" File = "/Documents and Settings/*/Local Settings/Application Data/ Microsoft/Windows/[Uu][Ss][Rr][Cc][Ll][Aa][Ss][Ss].*" File = "/Documents and Settings/*/[Nn][Tt][Uu][Ss][Ee][Rr].*" File = "/Documents and Settings/*/Cookies/*" File = "/Documents and Settings/*/Local Settings/History/*" File = "/Documents and Settings/*/Local Settings/ Temporary Internet Files/*" File = "/Documents and Settings/*/Local Settings/Temp/*" File = "/WINNT/CSC" File = "/WINNT/security/logs/scepol.log" File = "/WINNT/system32/config/*" File = "/WINNT/msdownld.tmp/*" File = "/WINNT/Internet Logs/*" File = "/WINNT/$Nt*Uninstall*" File = "/WINNT/Temp/*" File = "/temp/*" File = "/tmp/*" File = "/pagefile.sys" } }
Remarque : les trois lignes coupées de cette liste d'exclusion ne l'ont été que pour des motifs de mise en page, elles doivent en réalité être écrites sur une seule ligne.
L'ancienne Ressource FileSet des versions antérieures à 1.34.3 est obsolète, mais fonctionne encore. Nous vous encourageons à utiliser la nouvelle forme, car le code correspondant sera supprimé à partir de la version 1.37.
Si vous voulez vous faire une idée précise de ce qui sera effectivement sauvegardé par un FileSet, ou si vous voulez vous assurer de l'efficacité d'une liste d'exclusion, vous pouvez utiliser la commande estimate du programme Console. Voyez la commande estimate dans le chapitre Console de ce manuel.
Les noms de fichiers NTFS contenant des caractères Unicode (i.e. > 0xFF) ne peuvent, pour le moment, être nommé eplicitement. Vous devez inclure de tels fichiers en désignant un répertoire de niveau supérieur ou une lettre de disque ne contenant pas de caractère Unicode.
La ressource Client définit les attributs des clients sauvegardés par ce Director. Il faut une ressource Client par machine sauvegardée.
Les enregistrements File peuvent en fait être conservés pour une période inférieure à celle affectée à cette directive dans les cas où vous avez spécifié des périodes plus courtes pour les directives Job Retention ou Volume Retention. La plus courte des trois prend le pas sur les autres. Les durées peuvent être exprimées en secondes, minutes, heures, jours, semaines, mois, trimestres ou années. Consultez le chapitre Adapter les fichiers de configuration de ce manuel pour plus de détails sur les spécifications de durées.
La valeur par défaut est de 60 jours.
Si un enregistrement de Job est sélectionné pour élagage, tous les enregistrements File et JobMedia associés seront aussi élagués du catalogue, sans qu'il ne soit tenu compte de la période File Retention définie. Par conséquent, vous utiliserez, en principe, une période File Retention inférieure à la période Job retention. La période Job retention peut en fait s'avérer inférieure à la valeur que vous avez spécifiée si vous avez affecté une valeur inférieure à la directive Volume Retention dans la ressource Pool. Les périodes Job retention et Volume retention sont appliquées indépendamment, la plus petite prend le pas sur l'autre.
Les durées peuvent être exprimées en secondes, minutes, heures, jours, semaines, mois, trimestres ou années. Consultez le chapitre Adapter les fichiers de configuration de ce manuel pour plus de détails sur les spécifications de durées.
La valeur par défaut est 180 jours.
Voici un exemple d'une définition de ressource Client valide :
Client { Name = Minimatou FDAddress = minimatou Catalog = MySQL Password = very_good }
La ressource Storage définit les Storage Daemons disponibles pour le Director.
Actuellement, Bacula n'autorise qu'un seul type de media. Par conséquent, si vous disposez d'un lecteur qui en supporte plusieurs, vous pouvez utiliser une chaîne unique pour désigner les volumes de l'un ou l'autre type, par exemple Media Type = DDS-3-4 pour les types DDS-3 et DDS-4, mais ces volumes ne seront montés que sur les lecteurs spécifiés comme acceptant les deux types : DDS-3-4
Si vous voulez contraindre Bacula à utiliser un seul Storage Daemon ou un seul lecteur, vous devez spécifier un Media Type unique pour ce lecteur. C'est un point important qui devrait être bien compris. Notez que ceci s'applique également aux volumes disque. Si vous définissez plus d'une ressource Device disque dans votre fichier de configuration du Storage Daemon, les volumes sur ces deux devices seront en fait incompatibles car l'un ne pourra être monté sur l'autre puisqu'ils se trouvent dans des répertoires différents. C'est pourquoi vous devriez probablement plutôt utiliser deux Media Types distincts pour vos deux devices disque (même si vous pensez à eux comme ayant l'un et l'autre le type File).
Vous trouverez pus de détails à ce sujet dans le chapitre Gestion des volumes : fondements (NDT:basic volumes management) de ce manuel.
Le MediaType spécifié ici doit correspondre au MediaType spécifié dans la ressource Device du fichier de configuration du Storage Daemon. Cette directive est requise, et est utilisée par le Director et le Storage Daemon pour s'assurer qu'un volume sélectionné automatiquement dans un Pool correspond à un périphérique physique. Si un Storage Daemon gère plusieurs périphériques (par exemple, s'il écrit sur plusieurs volumes de type File sur différentes partitions), cette directive vous permet de préciser exactement quel périphérique utiliser.
Comme mentionné ci-dessus, la valeur spécifiée dans la ressource Storage du Director doit s'accorder avec celle spécifiée dans la ressource Device du fichier de configuration du Storage Daemon. Ceci représente aussi un contrôle supplémentaire pour assurer que vous n'essayez pas d'écrire les données destinées à un lecteur DLT sur un lecteur 8mm.
Pour que la robotique soit utilisée, vous devez aussi spécifier Autochanger = yes dans la ressource La Ressource Device du fichier de configuration du Storage Daemon, ainsi que d'autres paramètres importants du Storage Daemon. Vous trouverez plus d'information à ce sujet dans le chapitre Utiliser une librairie de ce manuel.
Alors qu'il est possible de définir des nombres maximum de jobs simultanés supérieurs à 1 dans les ressource Director, Job et Client, vous devriez porter une attention particulière au paramétrage de cette directive pour le Storage Daemon. En conservant la valeur 1, vous évitez que deux jobs écrivent simultanément sur le même volume ce qui, quoique supporté, n'est pas recommandé actuellement.
Voici un exemple de ressource Storage valide :
# Definition of tape storage device Storage { Name = DLTDrive Address = lpmatou Password = storage_password # password for Storage daemon Device = "HP DLT 80" # same as Device in Storage daemon Media Type = DLT8000 # same as MediaType in Storage daemon }
Un autre aspect important d'un pool est qu'il contient des attributs par défaut (Nombre maximum de jobs, période de rétention, drapeau de recyclage,...) qui sont conférés à tout volume lui appartenant lors de sa création. Ceci vous évite d'avoir à répondre à un grand nombre de questions lorsque vous étiquettez (label) un nouveau volume. Chacun de ces attributs peut ensuite être modifié sur chaque volume individuellement avec la commande update du programme Console. Notez que vous devez préciser explicitement quel pool est à utiliser avec chaque job. Bacula ne recherche pas automatiquement le pool correct.
Dans la plupart des installations de Bacula, toutes les sauvegardes de toutes les machines vont vers un unique jeu de volumes. Dans ce cas, vous n'utiliserez probablement que le pool par défaut Default. Si votre stratégie de sauvegarde vous impose à monter chaque jour une cartouche différente, vous voudrez probablement définir des pools distincts pour chaque jour. Pour plus d'informations à ce sujet, consultez le chapitre Stratégies de sauvegarde de ce manuel.
Pour utiliser un pool, vous devez suivre trois étapes :
D'abord, le pool doit être défini dans le fichier de configuration du Director. Ensuite le pool doit être enregistré dans le catalogue. Ceci est fait automatiquement par le Director à chaque fois qu'il démarre, ou peut être réalisé manuellement à l'aide de la commande create du programme Console. Enfin, si vous modifiez la définition du pool dans le fichier de configuration du Director et redémarrez Bacula, le pool sera mis à jour automatiquement, ce qui peut aussi être réalisé manuellement avec la commande update pool du programme Console pour rafraichir l'image du catalogue. C'est cette image du catalogue plutôt que l'image de la ressource du Director qui est utilisée pour les attributs de volume par défaut. Notez que pour que le pool soit automatiquement créé ou mis à jour, il doit être référencé explicitement par une ressource Job.
Ensuite le médium physique doit être étiquetté. L'étiquettage peut être réalisé soit par la commande label du programme Console, soit en utilisant le programme btape. La méthode à privilégier est la première.
Finallement, vous devez ajouter des noms de volumes (et leurs attributs) au pool. Pour que les volumes soient utilisés par Bacula, ils doivent être du même Media Type que l'Archive Device spécifiée pour le job. (Autrement dit, si vous vous apprétez à sauvegarder vers un lecteur DLT, le pool doit contenir des volumes DLT, puisque des volumes 8mm ne peuvent être montés sur un lecteur DLT). Le Media Type revêt une importance particulière si vous sauvegardez vers des fichiers. Lorsque vous exécutez un job, vous devez explicitement préciser le pool. Bacula sélectionne dès lors automatiquement le prochain volume du pool à utiliser, en s'assurant que le Media Type de tout volume sélectionné est bien celui requis par la ressource Storage spécifiée pour le job.
Si vous utilisez la commande label du programme Console pour étiquetter les volumes, il sont automatiquement ajoutés au pool, aussi cette dernière étape n'est généralement pas requise.
Il est aussi possible d'ajouter des volumes au catalogue sans avoir explicitement étiquetté les volumes physiques. Ceci s'effectue avec la commande add du programme Console.
Comme mentionné plus haut, à chaque démarrage, Bacula examine tous les pools associés à chaque catalogue, et si un enregistrement n'existe pas encore, il est créé à partir de la définition du pool dans la ressource. Bacula devrait probablement effectuer un update pool si vous modifiez la définition du pool mais, actuellement, vous devez le faire manuellement avec la commande update pool du programme Console.
La ressource Pool définie dans le fichier de configuration du Director peut contenir les directives suivantes :
La valeur définie par cette directive dans le fichier de configuration du Director est la valeur par défaut utilisée lorsqu'un nouveau volume est créé. Modifier la valeur dans le fichier de configuration ne changera pas ce qui est stocké sur le volume. Pour changer cette valeur pour un volume existant, vous devez utiliser la commande update du programme Console.
Notez que la valeur définie par cette directive dans le fichier de configuration du Director est la valeur par défaut utilisée lors de la création d'un volume. Une fois le volumes créé, modifier la valeur dans le fichier de configuration ne changera pas ce qui est stocké sur le volume. Pour changer cette valeur pour un volume existant, vous devez utiliser la commande update du programme Console.
La valeur définie par cette directive dans le fichier de configuration du Director est la valeur par défaut utilisée lors de la création d'un volume. Une fois le volumes créé, modifier la valeur dans le fichier de configuration ne changera pas ce qui est stocké sur le volume. Pour changer cette valeur pour un volume existant, vous devez utiliser la commande update du programme Console.
La valeur définie par cette directive dans le fichier de configuration du Director est la valeur par défaut utilisée lors de la création d'un volume. Une fois le volumes créé, modifier la valeur dans le fichier de configuration ne changera pas ce qui est stocké sur le volume. Pour changer cette valeur pour un volume existant, vous devez utiliser la commande update du programme Console.
Vous pourriez utiliser cette directive, par exemple, si vous avez un volume dédié aux sauvegardes incrémentales, et un volume dédié aux Fulls hebdomadaires. Une fois que la Full est passée, vous pouvez préférer utiliser un autre volume pour les incrémentales. Ceci peut être accompli en règlant la période Volume Use Duration à six jours pour les volumes des incrémentales. Autrement dit, celui-ci sera utilisé durant les six jours qui suivent une full, puis un autre volume d'incrémentales sera utilisé. Veillez à utiliser des périodes relativement courtes telles que 23 heures, ou vous pourriez placer Bacula en situation de devoir attendre tout un week-end le montage d'une cartouche par l'opérateur pour pouvoir terminer une sauvegarde.
Ce n'est qu'à la fin d'un job que Cette valeur est controlée, et que le statut du volume est éventuellement changé en Used, ce qui signifie que bien que la période Volume Use Duration puisse avoir expiré, l'entrée correspondante du catalogue ne sera pas mise à jour jusqu'à ce que le prochain job utilisant ce volume soit exécuté.
La valeur définie par cette directive dans le fichier de configuration du Director est la valeur par défaut utilisée lors de la création d'un volume. Une fois le volumes créé, modifier la valeur dans le fichier de configuration ne changera pas ce qui est stocké sur le volume. Pour changer cette valeur pour un volume existant, vous devez utiliser la commande update volume du programme Console.
Il est important de savoir que lorsque la période Volume Retention expire, Bacula ne recycle pas systématiquement le volume concerné. Il tente de conserver les données intactes aussi longtemps que possible avant d'écrire sur ce volume.
La valeur par défaut est 365 jours. Notez que cette directive règle la valeur par défaut pour chaque enregistrement de volume du catalogue lorsque le volume est créé. Cette valeur du catalogue peut ensuite être modifiée avec la commande update du programme Console.
En définissant plusieurs pools avec différentes périodes de rétention (volume retention), vous pouvez efficacement gérer vos cartouches avec, par exemple un pool de cartouches recyclé chaque semaine, un autre recyclé chaque mois, et ainsi de suite. Cependant, il faut bien garder à l'esprit que si votre période de rétention Volume Retention est trop courte, il peut arriver que votre dernière sauvegarde Full valide soit supprimée, de sorte que vous n'ayez plus une sauvegarde complète de votre système, et que votre prochaine incrémentale soit élevée en une Full. Par conséquent, la valeur minimum de la période Volume Retention devrait être au moins le double de l'intervalle séparant vos Fulls. Autrement dit, pour des Fulls mensuelles, la période Volume Retention devrait être supérieure ou égale à deux mois.
Notez que la valeur définie par cette directive dans le fichier de configuration du Director est la valeur par défaut utilisée lors de la création d'un volume. Une fois le volumes créé, modifier la valeur dans le fichier de configuration ne changera pas ce qui est stocké sur le volume. Pour changer cette valeur pour un volume existant, vous devez utiliser la commande update du programme Console.
Notez que la valeur définie par cette directive dans le fichier de configuration du Director est la valeur par défaut utilisée lors de la création d'un volume. Une fois le volumes créé, modifier la valeur dans le fichier de configuration ne changera pas ce qui est stocké sur le volume. Pour changer cette valeur pour un volume existant, vous devez utiliser la commande update du programme Console.
Cette directive peut être utile si vous avez un nombre fixe de volumes dans un pool et que vous voulez cycler sur ces volumes après avoir spécifié les périodes de rétention qui conviennent.
Cette directive peut être utile si vous avez un nombre fixe de volumes dans un pool et que vous voulez cycler sur ces volumes après avoir spécifié les périodes de rétention qui élaguent les volumes avant que vous n'ayez terminé le cycle sur les volumes.
Cette directive peut être utile si vous avez un nombre fixe de volumes dans un pool et que vous voulez cycler sur ces volumes lorsque tous les volumes sont pleins, sans avoir à vous soucier de paramétrer les périodes de rétentions qui conviendraient. Cependant, en l'utilisant, vous courrez le risque de perdre toutes vos données.
Soyez conscient que Purge Oldest Volume ne fait aucun cas d'aucune période de rétention. Si vous activez cette directive alors que vous ne possédez qu'un seul volume, ce volume sera systématiquement écrasé tout de suite après avoir été rempli ! Aussi, assurez vous au moins d'avoir un nombre décent de volumes dans votre pool avant d'exécuter un job. Si vous voulez que les périodes de rétention soient prises en compte, n'utilisez pas cette directive. Pour spécifier une période de rétention, utilisez la directive Volume Retention (voir ci dessus).
Je recommande fortement de ne pas utiliser cette directive, car il est certain que tôt ou tard, Bacula recyclera un volume contenant des données valides et récentes. La valeur par défaut est no
Le format devrait être spécifié entre guillemets, et consiste en une chaîne de lettres, chiffres et caractères spéciaux tiret (-), souligné (_), double point (:) et point (.), qui sont considérés valides pour un nom de volume.
De plus, le format peut comporter des caractères variables qui seront substituées par un algorithme complexe, ce qui permet de créer des noms de volumes avec plusieurs formats différents. En tous les cas, le processus d'expansion des variables doit aboutir au jeu de caractères définis comme légaux dans le dernier paragraphe. Généralement, ces caractères variables commencent par un signe dollar ($) ou un crochet droit ([). Si vous spécifiez des caractères variables vous devriez toujours les encadrer de guillemets. Pour plus de détails sur ce sujet, veuillez consulter le chapitre Variable Expansion de ce manuel.
Si aucun caractère variable n'est découvert dans la chaîne, le nom de volume sera constitué de la chaîne format suffixée du nombre de volumes dans le pool plus un, au format 4 chiffres et avec des zéros en tête. Par exemple, avec Label Format = ''File-``, les volumes seront nommés File-0001, File-0002, ...
Exception faite des variables spécifiques aux jobs, vous pouvez tester votre LabelFormat en utilisant la section var command du chapitre Console de ce manuel.
Dans la plupart des cas, vous devriez encadrer la spécification de format (la partie à droite du signe égale) entre guillemets. Notez que cette directive est obsolète et qu'elle est remplacée, à partir de la version 1.37 par un script Python pour la création des noms de volumes.
Pour qu'un pool puisse être utilisé lors d'une sauvegarde, il faut qu'il lui soit associé au moins un volume. Les volumes sont créés et affectés aux pools avec les commandes label ou add du programme Bacula Console. Outre l'affectation du volume au pool (c'est à dire son référencement dans le catalogue), le volume physique doit recevoir une étiquette logicielle valide pour que Bacula l'accepte. Ceci peut être réalisé automatiquement grāce à la commande label. D'autre part, Bacula peut effectuer cette opération automatiquement si l'instruction lui en est donné, mais cette fonctionnalité n'est pas encore pleinement implémentée.
Voici un exemple d'une définition de ressource Pool valide :
Pool { Name = Default Pool Type = Backup }
La ressource Catalog précise quel catalogue utiliser pour le job courant. Actuellement, Bacula ne peut utiliser qu'un type de serveur de bases de données défini lors de sa configuration : SQLite, MySQL, PostgreSQL. En revanche, vous pouvez utiliser autant de catalogues que vous le souhaitez. Par exemple, vous pouvez avoir un catalogue par client, ou encore un catalogue pour les sauvegardes, un autre pour les jobs de type Verify et un troisième pour les restaurations.
Voici un exemple d'une définition de ressource Catalog valide :
Catalog { Name = SQLite dbname = bacula; user = bacula; password = "" # no password = no security }
en voici une deuxième pour un catalogue sur une autre machine :
Catalog { Name = MySQL dbname = bacula user = bacula password = "" DB Address = remote.acme.com DB Port = 1234 }
Pour les détails sur la ressource Messages, veuillez consulter le chapitre La ressource Messages de ce manuel.
A partir de la version 1.33 de Bacula, l'administrateur dispose de trois sortes de consoles différentes pour interagir avec le Director. Ces trois types de consoles comportent trois niveaux de sécurité différents.
La ressource Console est optionnelle. Les directives suivantes sont permises dans le fichier de configuration du Director.
JobACL = kernsave, "Backup client 1", "Backup client 2" JobACL = "RestoreFiles"
Avec cette spécificaton, la console peut accéder aux ressources du Director pour les quatre jobs désignés par la directive JobACL, et uniquement à eux.
En plus des différents noms de ressources du Director et de commandes, le mot-clef spécial *all* peut être spécifié dans chacune des directives ACL ci-dessus. Sa présence signifie que toute ressource ou commande (pourvu qu'elle soit appropriée à la directive) est acceptée. Pour un exemple de fichier de configuration, voyez le chapitre Configuration de la console de ce manuel
# # Default Bacula Director Configuration file # # The only thing that MUST be changed is to add one or more # file or directory names in the Include directive of the # FileSet resource. # # For Bacula release 1.15 (5 March 2002) -- redhat # # You might also want to change the default email address # from root to your address. See the "mail" and "operator" # directives in the Messages resource. # Director { # define myself Name = rufus-dir QueryFile = "/home/kern/bacula/bin/query.sql" WorkingDirectory = "/home/kern/bacula/bin/working" PidDirectory = "/home/kern/bacula/bin/working" Password = "XkSfzu/Cf/wX4L8Zh4G4/yhCbpLcz3YVdmVoQvU3EyF/" } # Define the backup Job Job { Name = "NightlySave" Type = Backup Level = Incremental # default Client=rufus-fd FileSet="Full Set" Schedule = "WeeklyCycle" Storage = DLTDrive Messages = Standard Pool = Default } Job { Name = "Restore" Type = Restore Client=rufus-fd FileSet="Full Set" Where = /tmp/bacula-restores Storage = DLTDrive Messages = Standard Pool = Default } # List of files to be backed up FileSet { Name = "Full Set" Include { Options { signature=SHA1 } # # Put your list of files here, one per line or include an # external list with: # # @file-name # # Note: / backs up everything File = / } Exclude { } } # When to do the backups Schedule { Name = "WeeklyCycle" Run = Full sun at 1:05 Run = Incremental mon-sat at 1:05 } # Client (File Services) to backup Client { Name = rufus-fd Address = rufus Catalog = MyCatalog Password = "MQk6lVinz4GG2hdIZk1dsKE/LxMZGo6znMHiD7t7vzF+" File Retention = 60d # sixty day file retention Job Retention = 1y # 1 year Job retention AutoPrune = yes # Auto apply retention periods } # Definition of DLT tape storage device Storage { Name = DLTDrive Address = rufus Password = "jMeWZvfikUHvt3kzKVVPpQ0ccmV6emPnF2cPYFdhLApQ" Device = "HP DLT 80" # same as Device in Storage daemon Media Type = DLT8000 # same as MediaType in Storage daemon } # Definition for a DLT autochanger device Storage { Name = Autochanger Address = rufus Password = "jMeWZvfikUHvt3kzKVVPpQ0ccmV6emPnF2cPYFdhLApQ" Device = "Autochanger" # same as Device in Storage daemon Media Type = DLT-8000 # Different from DLTDrive Autochanger = yes } # Definition of DDS tape storage device Storage { Name = SDT-10000 Address = rufus Password = "jMeWZvfikUHvt3kzKVVPpQ0ccmV6emPnF2cPYFdhLApQ" Device = SDT-10000 # same as Device in Storage daemon Media Type = DDS-4 # same as MediaType in Storage daemon } # Definition of 8mm tape storage device Storage { Name = "8mmDrive" Address = rufus Password = "jMeWZvfikUHvt3kzKVVPpQ0ccmV6emPnF2cPYFdhLApQ" Device = "Exabyte 8mm" MediaType = "8mm" } # Definition of file storage device Storage { Name = File Address = rufus Password = "jMeWZvfikUHvt3kzKVVPpQ0ccmV6emPnF2cPYFdhLApQ" Device = FileStorage Media Type = File } # Generic catalog service Catalog { Name = MyCatalog dbname = bacula; user = bacula; password = "" } # Reasonable message delivery -- send most everything to # the email address and to the console Messages { Name = Standard mail = root@localhost = all, !skipped, !terminate operator = root@localhost = mount console = all, !skipped, !saved } # Default pool definition Pool { Name = Default Pool Type = Backup AutoPrune = yes Recycle = yes } # # Restricted console used by tray-monitor to get the status of the director # Console { Name = Monitor Password = "GN0uRo7PTUmlMbqrJ2Gr1p0fk0HQJTxwnFyE4WSST3MWZseR" CommandACL = status, .status }
La configuration du Client (ou File Daemon) est l'une des plus simples à effectuer. En principe, vous n'aurez rien à modifier au fichier par défaut en dehors du nom du Client afin d'identifier clairement les messages d'erreur.
Pour un discours général sur les fichiers de configuration et ressources incluant les types de données reconnues par Bacula, veuillez consulter le chapitre Configuration de ce manuel. Les ressources suivantes doivent être définies :
La ressource Client (ou File Daemon) définit le nom du Client (en tant que client du Director), ainsi que le port sur lequel le Client écoute les connections du Director.
Sur les systèmes Win32, vous pouvez, dans certaines circonstances, être amené à spécifier une lettre de disque au niveau de la spécification du répertoire. Aussi, assurez-vous que ce répertoire est bien accessible en écriture par l'utilisateur SYSTEM, faute de quoi les restaurations peuvent échouer (le fichier bootstrap transféré au File Daemon par le Director est temporairement placé dans ce répertoire avant d'être transmis au Storage Daemon).
Typiquement, sur les systèmes Linux, vous utiliserez la valeur /var/run pour cette directive. Si vous n'installez pas Bacula dans les répertoires du système, vous pouvez utiliser le Working Directory tel que défini ci-dessus.
Si vous continuez de recevoir de messages broken pipe malgré Heartbeat Interval, et si vous utilisez Windows, vous devriez envisager de mettre à jour votre pilote ethernet. Il s'agit d'un problème connu des pilotes NVidia NForce 3 (4.4.2 17/05/2004). Vous pouvez aussi essayer la procédure suivante suggérée par Thomas Simmons pour les machines Win32 :
Démarrer > Panneau de configuration > connections réseau
Faites un click droit sur connection pour l'adaptateur nvidia et sélectionnez "propriétés". Sous l'onglet "Général", cliquez "Configurer...". Sous l'onglet "Avancé", désactivez l'option "Checksum Offload" et cliquez "Ok" pour sauvegarder la modification.
L'absence de communication, ou des interruptions dans les communications peuvent aussi être causées par des firewalls Linux si vous avez une règle qui limite les connections ou le trafic.
Right click the connection for the nvidia adapter and select properties. Under the General tab, click "Configure...". Under the Advanced tab set "Checksum Offload" to disabled and click OK to save the change.
Lack of communications, or communications that get interrupted can also be caused by Linux firewalls where you have a rule that throttles connections or traffic.
FDAddresses = { ip = { addr = 1.2.3.4; port = 1205; } ipv4 = { addr = 1.2.3.4; port = http; } ipv6 = { addr = 1.2.3.4; port = 1205; } ip = { addr = 1.2.3.4 port = 1205 } ip = { addr = 1.2.3.4 } ip = { addr = 201:220:222::2 } ip = { addr = bluedot.thun.net } }
où ip, ip4, ip6, addr, et port sont tous des mots-clef. Notez que les adresses peuvent être spécifiées aussi bien en quadruplets pointés qu'en notation IPv6 double pointée ou avec des noms symboliques (seulement pour la la spécification d'ip). De même, les ports peuvent être spécifié sous forme de nombres, ou avec les valeurs mnémoniques du fichier /etc/services. Si un port n'est pas spécifié, la valeur par défaut est utilisée. Si une section ip est spécifiée, la résolution peut s'effectuer avec IPv4 et IPv6. Si ip4 est spécifié, alors seule la résolution IPv4 est permise, il en va de même avec ip6.
Voici un exemple d'une définition de ressource Client valide :
Client { # this is me Name = rufus-fd WorkingDirectory = $HOME/bacula/bin/working Pid Directory = $HOME/bacula/bin/working }
Veuillez noter que si ce Director est utilisé à des fins de surveillance, nous vous recommandons fortement de mettre cette directive à yes pour éviter de sérieux problèmes de sécurité.
Ainsi, plusieurs Directors peuvent être autorisés à utiliser les services de ce Client. Chaque Director aura un nom différent, et, en principe, un mot de passe différent.
Voici un exemple d'une définition de ressource Director valide :
# # List Directors who are permitted to contact the File daemon # Director { Name = HeadMan Password = very_good # password HeadMan must supply } Director { Name = Worker Password = not_as_good Monitor = Yes }
Il doit y avoir au moins une ressource Messages définie dans le fichier de configuration du Client.
Voici un exemple de fichier de configuration de Client :
# # Default Bacula File Daemon Configuration file # # For Bacula release 1.35.2 (16 August 2004) -- gentoo 1.4.16 # # There is not much to change here except perhaps to # set the Director's name and File daemon's name # to something more appropriate for your site. # # # List Directors who are permitted to contact this File daemon # Director { Name = rufus-dir Password = "/LqPRkX++saVyQE7w7mmiFg/qxYc1kufww6FEyY/47jU" } # # Restricted Director, used by tray-monitor to get the # status of the file daemon # Director { Name = rufus-mon Password = "FYpq4yyI1y562EMS35bA0J0QC0M2L3t5cZObxT3XQxgxppTn" Monitor = yes } # # "Global" File daemon configuration specifications # FileDaemon { # this is me Name = rufus-fd WorkingDirectory = $HOME/bacula/bin/working Pid Directory = $HOME/bacula/bin/working } # Send all messages except skipped files back to Director Messages { Name = Standard director = rufus-dir = all, !skipped }
Des exemples de directives de ressources device connues pour fonctionner pour beaucoup de lecteurs de bandes communs peuvent être trouvés dans le répertoire : <bacula-src>/examples/devices. La plupart seront énumérés ici.
Pour une discussion générale concernant les fichiers de configuration de Bacula, les ressources et les types de données reconnus, veuillez consulter le chapitre Configuration de ce manuel. Les définitions de ressources Storage suivantes doivent être définies :
En général, les propriétés spécifiées au niveau de la ressource Storage définissent des propriétés globales du Storage Daemon. Chaque fichier de configuration de Storage Daemon doit avoir sa propre définition de ressource Storage.
Typiquement, sur les systèmes Linux, vous utiliserez ici /var/run. Si vous n'installez pas Bacula dans les répertoires système, vous pouvez utiliser le répertoire de travail Working Directory défini plus haut. Cette directive est requise.
SDAddresses = { ip = { addr = 1.2.3.4; port = 1205; } ipv4 = { addr = 1.2.3.4; port = http; } ipv6 = { addr = 1.2.3.4; port = 1205; } ip = { addr = 1.2.3.4 port = 1205 } ip = { addr = 1.2.3.4 } ip = { addr = 201:220:222::2 } ip = { addr = bluedot.thun.net } }
où "ip", "ip4", "ip6", "addr", et "port" sont des mots-clef. Notez que les adresses peuvent être spécifiées sous forme de quadruplets pointés, de nom symboliques (uniquement dans la spécification "ip") ou en notation IPv6 à double points. Le port peut quand à lui être spécifié par son numéro, ou par sa valeur mnémonique du fichier /etc/services. Si un port n'est pas spécifié, la valeur par défaut est utilisée. Si une section ip est spécifiée, la résolution peut être réalisée par ipv4 ou ipv6. En revanche, si ip4 ou ip6 est spécifiée, seule la résolution correspondante fonctionne.
Vous pouvez, avec ces directives, remplacer les valeurs des directives SDPort et SDAddress montrées ci-dessous.
Voici une définition typique d'une ressource Storage du Storage Daemon :
# # "Global" Storage daemon configuration specifications appear # under the Storage resource. # Storage { Name = "Storage daemon" Address = localhost WorkingDirectory = "~/bacula/working" Pid Directory = "~/bacula/working" }
La ressource Director spécifie le nom du Director qui est autorisé à utiliser les services du Storage Daemon. Il peut exister plusieurs ressources Director. Le nom et le mot de passe du Director doivent s'accorder avec leurs homologues dans le fichier de configuration du Storage Daemon.
Si ce Director est utilisé par un superviseur, nous vous recommandons fortement d'activer cette directive pour éviter de sérieux problèmes de sécurité.
Voici un exemple d'une définition de ressource Director valide :
Director { Name = MainDirector Password = my_secret_password }
La ressource Device spécifie les détails de chaque périphérique (en général, un lecteur de bandes) qui peut être utilisé par le Storage Daemon. Un Storage Daemon peut disposer de plusieurs ressources Device. En général, les propriétés spécifiées dans la ressource Device sont spécifiques au périphérique.
Comme mentionné plus haut,Archive Device est, en principe, le nom d'un lecteur de bandes, mais vous pouvez tout aussi bien spécifier le chemin absolu vers un répertoire existant. Dans ce cas, Bacula utilisera un fichier pour stocker les données dans le répertoire spécifié, le nom de fichier utilisé sera celui du volume tel que spécifié dans le catalogue. Si vous souhaitez écrire dans plusieurs répertoires (dans le but de répartir la charge sur plusieurs disques), vous devez définir deux ressources Device, chacune comportant une Archive Device avec un répertoire différent.
Une troisième possibilité consiste à spécifier le nom d'un FIFO. Un FIFO est un fichier spécial qui connecte deux programmes via la mémoire du noyau. Si vous spécifiez un FIFO en guise d'Archive Device, vous devez avoir un programme qui lit ce que Bacula écrit dans le FIFO. Lorsque le Storage Daemon démarre le job, il attend que le programme lecteur commence à lire pendant un délai maximal de de MaximumOpenWait secondes, au delà duquel le job est terminé. Par conséquent, il est préférable de lancer le programme lecteur au début du job, par exemple grâce à la directive RunBeforeJob. Pour ce type de périphérique, vous ne devez jamais spécifier AlwaysOpen, puisque vous voulez que le Storage Daemon ne l'ouvre que lorsqu'un job démarre, aussi veuillez attribuer explicitement la valeur No à cette directive. Puisqu'un FIFO est un périphérique à sens unique, Bacula ne tente pas d'en lire le label, il se contente d'y écrire. Pour créer un volume FIFO dans le catalogue, utilisez la commande add plutôt que la commande label afin d'éviter de tenter d'écrire un label.
Lors d'une opération de restauration, si l'Archive Device est un FIFO, Bacula tente de lire le FIFO, aussi vous devez avoir un programme externe qui écrit dans le FIFO. Bacula attend que ce programme commence à écrire pendant un délai maximal de MaximumOpenWait secondes après quoi il termine le job. Comme mentionné ci-dessus, vous pouvez utiliser la directive RunBeforeJob pour lancer ce programme auteur dès le début du job.
La directive Archive Device est requise.
La directive Device Type n'est pas requise, et si elle n'est pas spécifiée, Bacula tentera de deviner cette information selon la spécification Archive Device fournie. Il existe plusieurs avantages à spécifier explicitement le type de périphérique. D'abord, sur certains systèmes, les périphériques bloc et caractère ont le même type, ce qui signifie que sur ces systèmes, Bacula est probablement incapable de deviner qu'un périphérique est un DVD. Ensuite, si vous spécifiez explicitement le type de périphérique, le point de montage n'a pas besoin d'être défini jusqu'à ce que le périphérique soit ouvert. C'est le cas de la plupart des périphériques amovibles tels que les USB montés par le daemon HAL. Au contraire, si le type de périphérique n'est pas spécifié explicitement, le point de montage doit exister dès le démarrage du Storage Daemon.
Cette directive est apparue avec la version 1.38.6 de Bacula.
Même si les noms que vous assignez sont arbitraires, vous devriez les choisir avec circonspection, car le Media Type est utilisé pour déterminer le périphérique de stockage à sélectionner lors d'une restauration. Ainsi, vous devriez certainement utiliser le même Media Type pour tous les lecteurs dont les cartouches sont interchangeables. Ce n'est généralement pas un problème si vous n'avez qu'un Storage Daemon, mais c'en est un avec plusieurs Storage Daemon, surtout s'ils utilisent des média incompatibles.
Si, par exemple, vous spécifiez le Media Type "DDS-4", Bacula pourra lors de restaurations sélectionner tout Storage Daemon qui supporte les "DDS-4". Si vous avez une librairie, vous voudrez peut-être baptiser son Media Type d'un nom qui lui soit unique, à moins que vous souhaitiez pouvoir utiliser ses volumes dans d'autres lecteurs. Vous devriez aussi vous assurer d'avoir des noms de Media Type uniques si les media ne sont pas compatibles d'un lecteur à l'autre. Cette spécification est requise pour tous les périphériques.
Enfin, si vous utilisez le stockage sur disque, sachez que chaque ressource Device a généralement un point de montage (ou répertoire) différent. Afin que Bacula puisse sélectionner correctement la ressource Device à utiliser, chacun doit avoir un Media Type distinct.
Changer Command = "/path/mtx-changer %c %o %S %a %d"
Et vous installerez le programme mtx sur votre système (paquetage tiers). Un exemple de cette commande figure dans le fichier de configuration par défaut du Storage Daemon, bacula-sd.conf. Pour plus de détails concernant les substitutions de caractères qui peuvent être utilisées pour configurer votre librairie, veuillez consulter le chapitre sur l'utilisation des Librairies. Les utilisateurs de FreeBSD voudront probablement jeter un oeil aux quelques scripts fournis dans le répertoire examples/autochangers.
Notez que vous pouvez trouver un usage à cette commande sans nécessairement posséder une librairie. L'exemple ci-dessous utilise le programme tapeinfo qui vient avec le paquet mtx mais peut être utilisé avec n'importe quel lecteur. Vous devrez tout de même spécifier une directive Changer Device dans votre ressource Device (voir ci-dessus) afin que le périphérique SCSI générique puisse être édité dans la commande (avec %c).
Voici un exemple qui affiche les alertes en provenance du lecteur dans les rapports de jobs :
Alert Command = "sh -c 'tapeinfo -f %c | grep TapeAlert'"
Et un exemple de ce qui peut en sortir lorqu'il y a un problème :
bacula-sd Alert: TapeAlert[32]: Interface: Problem with SCSI interface between tape drive and initiator.
Si vous avez spécifié Always Open = yes (comme recommandé) et si vous voulez utiliser le lecteur pour autre chose, libérez-le simplement avec la commande unmount dans la console Bacula. N'oubliez-pas ensuite de remonter le lecteur avec la commande mount afin que Bacula soit prèt à prendre en charge le prochain job planifié.
Pour le stockage sur disque (File Storage), cette directive est ignorée. Dans le cas d'un stockage FIFO, vous devez mettre cette directive à No.
Notez bien que si vous mettez cette directive à No, Bacula libère le lecteur entre chaque job, obligeant le lecteur à rembobiner la cartouche, et à replacer la bande à la fin de la zone de données, ce qui peut prendre beaucoup de temps.
Pour contraindre la taille des blocs à être fixe, comme c'est le cas de certains périphériques à accès séquentiel, stipulez des tailles de blocs minimum Minimum block size et maximum Maximum block size identiques. Le paramétrage par défaut est 0 pour les deux directives et la taille de bloc par défaut est de 64 512 octets.
Par exemple, si vous souhaitez fixer la taille des blocs à 100K octets, spécifiez :
Minimum block size = 100K Maximum block size = 100KNotez que si vous spécifiez une taille de blocs fixe comme ci-dessus, le lecteur doit être réglé soit en mode "taille de blocs variable", soit en mode "taille de blocs fixe" avec impérativement la même taille de blocs fixe que celle spécifiée dans Bacula (ce paramètre se règle généralement au niveau du lecteur avec mt), faute de quoi vous aurez des erreurs à la relecture de vos cartouches.
Si vous voulez que votre taille de blocs soit variable mais comprise entre 64 Ko et 200 Ko, spécifiez :
Minimum block size = 64K Maximum blocksize = 200K
Si aucune valeur n'est spécifiée (ou si la valeur spécifiée est 0), le Storage Daemon utilise la valeur par défaut de 64 512 octets.
Le réglage par défaut de cette directive est Yes. Cette option est utilisée lors de l'écriture à la suite d'une cartouche, pour s'assurer que les données précédemment écrites ne seront pas corrompues. Nous vous recommandons, si vous avez un lecteur non-standard ou inhabituel, de le tester avec le programme btape pour vérifier s'il supporte ou non cette fonction. Tous les lecteurs modernes (au delà de 1998) la supportent.
La valeur par défaut de cette directive est Yes.
(NDT : Paragraphe à revoir VO ci dessous) If No, the default, no special action is taken by Bacula with the End of Medium (end of tape) is reached because the tape will be positioned after the last EOF tape mark, and Bacula can append to the tape as desired. However, on some systems, such as FreeBSD, when Bacula reads the End of Medium (end of tape), the tape will be positioned after the second EOF tape mark (two successive EOF marks indicated End of Medium). If Bacula appends from that point, all the appended data will be lost. The solution for such systems is to specify BSF at EOM which causes Bacula to backspace over the second EOF mark. Determination of whether or not you need this directive is done using the test command in the btape program.
Si vous utilisez un noyau Linux 2.6, ou un système tel que FreeBSD ou Solaris, la directive Offline On Unmount abandonnera votre lecteur sans cartouche, et Bacula incapable de l'utiliser. Pour plus d'informations sur ce problème, consultez la section description de Offline On Unmount dans le chapitre sur les tests de lecteurs.
La valeur par défaut a été choisie relativement importante, mais pas trop, au cas ou vous transmettriez vos données via Internet. Il est clair que sur un réseau local rapide, vous pouvez augmenter cette valeur et améliorer les performances. Par exemple, certains utilisateurs ont obtenu des facteurs d'accélération de l'ordre de 5 à 10 en utilisant un tampon réseau initial de 65 536 octets. La plupart des utilisateurs indiquent que des valeurs plus grandes ne semblent pas améliorer les performances. Si vous voulez améliorer la viteese de vos sauvegardes, cette directive est probablement le meilleur endroit pour expérimenter. Vous voudrez probablement effectuer les modifications correspondantes dans les fichiers de configuration de chacun des File Daemons.
Si le périphérique requiert un montage, l'ordre de montage est transmis lorsque cette valeur est atteinte. Dans ce cas, vous devez vous assurer d'avor suffisament d'espace dans votre répertoire tampon, faute de quoi vos données resteront dans le répertoire tampon.
Cette directive est ignorée pour les lecteurs de bandes et les FIFO.
Toutes les directives décrites dans cette section sont implémentées dans Bacula à partir de la version 1.37.
A partir de la version 1.39.5, les directives "Requires Mount", "Mount Point", "Mount Command", et "Unmount Command" s'appliquent aux systèmes de fichiers amovibles tels que les périhériques USB, et plus seulement aux DVDs.
La plupart du temps, vous le définirez ainsi :
Mount Command = "/bin/mount -t iso9660 -o ro %a %m"
La plupart du temps, vous le définirez ainsi :
Unmount Command = "/bin/umount %m"
Pour un DVD, vous utiliserez la plupart du temps le script fourni dvd-handler comme suit :
Command that must be executed to write a part to the device. Before the command is executed, %a is replaced with the Archive Device, %m with the Mount Point, %e is replaced with 1 if we are writing the first part, and with 0 otherwise, and %v with the current part filename.
For a DVD, you will most frequently specify the Bacula supplied dvd-handler script as follows:
Write Part Command = "/path/dvd-handler %a write %e %v"
Où /path est le chemin vers votre répertoire de scripts, et dvd-handler est le script fourni avec Bacula. Cette commande est déjà présente quoique commentée dans le fichier de configuration du Storage Daemon. Pour l'utiliser, il vous suffit de supprimer le caractère #.
Pour un DVD, vous utiliserez la plupart du temps le script fourni dvd-handler comme suit :
Free Space Command = "/path/dvd-handler %a free"
Où /path est le chemin vers votre répertoire de scripts, et dvd-handler est le script fourni avec Bacula. Si vous voulez spécifier votre propre commande, examinez le code de dvd-handler afin de voir le type de retour attendu par Bacula. Cette commande est déjà présente quoique commentée dans le fichier de configuration du Storage Daemon. Pour l'utiliser, il vous suffit de supprimer le caractère #.
Si vous n'utilisez pas cette directive, Bacula s'attendra à ce qu'il y ait toujours de la place dur le périphérique.
La ressource Autochanger supporte les librairies à un ou plusieurs lecteurs en regroupant une ou plusieurs ressources Device en une unité nommée Autochanger dans Bacula (souvent désignée en tant que librairie de bandes par les constructeurs). Si vous possédez une librairie, et si vous voulez qu'elle fonctionne correctement, vous devez avoir une ressource Autochanger dans le fichier de configuration de votre Storage Daemon, et les directives Storage de votre Director doivent se référer au nom de la ressource Autochanger si elles sont supposées utiliser la librairie. Dans les versions antérieures à 1.38.0, les directives Storage du Director se référaient directement aux ressources Device qui étaient des librairies. Désormais, ce type de référence directe ne fonctionne plus avec les librairies.
Voici un exemple de définition de ressource Autochanger valide :
Autochanger { Name = "DDS-4-changer" Device = DDS-4-1, DDS-4-2, DDS-4-3 Changer Device = /dev/sg0 Changer Command = "/etc/bacula/mtx-changer %c %o %S %a %d" } Device { Name = "DDS-4-1" Drive Index = 0 Autochanger = yes ... } Device { Name = "DDS-4-2" Drive Index = 1 Autochanger = yes ... Device { Name = "DDS-4-3" Drive Index = 2 Autochanger = yes Autoselect = no ... }
Notez l'importance de la directive Autochanger = yes dans chaque définition de périphérique appartenant à une librairie. Un périphérique ne devrait pas être défini comme appartenant à plusieurs librairies. Aussi, votre directive Device dans la ressource Storage du Director devrait comporter le nom de la ressource Autochanger plutôt que le nom de l'un des lecteurs.
Si vous avez un lecteur qui appartient physiquement à une librairie mais que vous ne souhaitez pas que Bacula puisse l'utiliser automatiquement (par exemple, si vous voulez le réserver pour les restaurations) vous pouvez utiliser la directive :
Autoselect = no
à la ressource Device de ce lecteur. Dans ce cas, Bacula ne le sélectionnera pas automatiquement en accédant à la librairie. Vous pouvez encore utiliser le lecteur en le désignant par son nom de ressource device plutôt que par celui de la ressource Autochanger. Un exemple d'une telle définition est montré ci-dessus pour le lecteur DDS-4-3, qui ne sera pas sélectionné si le nom DDS-4-changer est utilisé dans une ressource Storage, mais le sera si DDS-4-3 est utilisé.
Pour une description de la ressource Messages, veuillez consulter le chapitre La ressource Messages de ce manuel.
Voici un exemple de fichier de configuration du Storage Daemon :
# # Default Bacula Storage Daemon Configuration file # # For Bacula release 1.37.2 (07 July 2005) -- gentoo 1.4.16 # # You may need to change the name of your tape drive # on the "Archive Device" directive in the Device # resource. If you change the Name and/or the # "Media Type" in the Device resource, please ensure # that bacula-dir.conf has corresponding changes. # Storage { # definition of myself Name = rufus-sd Address = rufus WorkingDirectory = "$HOME/bacula/bin/working" Pid Directory = "$HOME/bacula/bin/working" Maximum Concurrent Jobs = 20 } # # List Directors who are permitted to contact Storage daemon # Director { Name = rufus-dir Password = "ZF9Ctf5PQoWCPkmR3s4atCB0usUPg+vWWyIo2VS5ti6k" } # # Restricted Director, used by tray-monitor to get the # status of the storage daemon # Director { Name = rufus-mon Password = "9usxgc307dMbe7jbD16v0PXlhD64UVasIDD0DH2WAujcDsc6" Monitor = yes } # # Devices supported by this Storage daemon # To connect, the Director's bacula-dir.conf must have the # same Name and MediaType. # Autochanger { Name = Autochanger Device = Drive-1 Device = Drive-2 Changer Command = "/home/kern/bacula/bin/mtx-changer %c %o %S %a %d" Changer Device = /dev/sg0 } Device { Name = Drive-1 # Drive Index = 0 Media Type = DLT-8000 Archive Device = /dev/nst0 AutomaticMount = yes; # when device opened, read it AlwaysOpen = yes; RemovableMedia = yes; RandomAccess = no; AutoChanger = yes Alert Command = "sh -c 'tapeinfo -f %c |grep TapeAlert|cat'" } Device { Name = Drive-2 # Drive Index = 1 Media Type = DLT-8000 Archive Device = /dev/nst1 AutomaticMount = yes; # when device opened, read it AlwaysOpen = yes; RemovableMedia = yes; RandomAccess = no; AutoChanger = yes Alert Command = "sh -c 'tapeinfo -f %c |grep TapeAlert|cat'" } Device { Name = "HP DLT 80" Media Type = DLT8000 Archive Device = /dev/nst0 AutomaticMount = yes; # when device opened, read it AlwaysOpen = yes; RemovableMedia = yes; } #Device { # Name = SDT-7000 # # Media Type = DDS-2 # Archive Device = /dev/nst0 # AutomaticMount = yes; # when device opened, read it # AlwaysOpen = yes; # RemovableMedia = yes; #} #Device { # Name = Floppy # Media Type = Floppy # Archive Device = /mnt/floppy # RemovableMedia = yes; # Random Access = Yes; # AutomaticMount = yes; # when device opened, read it # AlwaysOpen = no; #} #Device { # Name = FileStorage # Media Type = File # Archive Device = /tmp # LabelMedia = yes; # lets Bacula label unlabeled media # Random Access = Yes; # AutomaticMount = yes; # when device opened, read it # RemovableMedia = no; # AlwaysOpen = no; #} #Device { # Name = "NEC ND-1300A" # Media Type = DVD # Archive Device = /dev/hda # LabelMedia = yes; # lets Bacula label unlabeled media # Random Access = Yes; # AutomaticMount = yes; # when device opened, read it # RemovableMedia = yes; # AlwaysOpen = no; # MaximumPartSize = 800M; # RequiresMount = yes; # MountPoint = /mnt/cdrom; # MountCommand = "/bin/mount -t iso9660 -o ro %a %m"; # UnmountCommand = "/bin/umount %m"; # SpoolDirectory = /tmp/backup; # WritePartCommand = "/etc/bacula/dvd-handler %a write %e %v" # FreeSpaceCommand = "/etc/bacula/dvd-handler %a free" #} # # A very old Exabyte with no end of media detection # #Device { # Name = "Exabyte 8mm" # Media Type = "8mm" # Archive Device = /dev/nst0 # Hardware end of medium = No; # AutomaticMount = yes; # when device opened, read it # AlwaysOpen = Yes; # RemovableMedia = yes; #} # # Send all messages to the Director, # mount messages also are sent to the email address # Messages { Name = Standard director = rufus-dir = all operator = root = mount }
La ressource Messages définit la façon dont les messages doivent être construits et vers quelles destinations ils doivent être transmis.
Bien que chaque daemon intègre un gestionnaire de messages pleinement fonctionnel, vous choisirez certainement de centraliser les messages appropriés des File Daemons et du Storage Daemon vers le Director. Ainsi, tous les messages associés à un job donné peuvent être combinés et envoyés en un simple courrier électronique vers l'utilisateur, ou enregistré dans quelque fichier de logs.
Chaque message généré par un daemon Bacula possède un type associé tel que INFO, WARNING, ERROR, FATAL, etc. La ressource Messages vous permet de stipuler les types de messages que vous voulez voir, et où les envoyer. De plus, un message peut être expédié vers plusieurs destinations. Par exemple, vous pouvez faire en sorte quque tous les messages d'erreur soient consignés dans un fichier de logs tout en vous étant envoyés par courrier élecronique. En définissant plusieurs ressources Messages, vous pouvez profiter de différents modes de prise en charge pour chaque type de job (par exemple, selon qu'il s'agit d'une full ou d'un incrémentale).
En général, les messages sont attachés à un job et sont inclus dans le rapport de job. Il existe de rares situations où ce n'est pas possible, par exemple lorsqu'aucun job n'est en cours d'exécution, ou si une erreur de communication se produit entre un daemon et le Director. Dans ce genre de situations, le message demeure dans le système et devrait être purgé à la fin job suivant. Cependant, comme de tels messages ne sont pas attachés à un job, tous ceux qui sont envoyés par courrier électronique sont envoyés à /usr/lib/sendmail. Si sur votre système, comme c'est le cas de FreeBSD, sendmail réside en un autre emplacement, veillez à le lier depuis l'emplacement ci-dessus.
Les enregistrements contenus dans une ressource Messages consistent en une spécification de destination suivie d'une liste de types de messages message-types au format :
ou, pour ces destinations qui nécessitent de spécifier une adresse (e-mail, par exemple) :
où destination est l'un des mots-clef prédéfinis qui précise où le message doit être expédié (stdout, file, ...), message-type est l'un des mots-clef prédéfinis qui précise le type de messages généré par Bacula (ERROR, WARNING, FATAL, ...) et address varie selon le mot clef destination mais peut typiquement être une adresse de courrier électronique ou un nom de fichier.
Voici la liste des directives disponibles pour définir des ressources Messages :
mail -s "Bacula Message" <recipients> Dans de nombreusx cas, selon votre machine, cette commande peut ne pas fonctionner. La directive MailCommand vous permet de stipuler précisément la façon d'envoyer vos courrier électroniques. Lors de l'exécution de la commande command, spécifiée entre guillemets, les substitutions suivantes sont effectuées :
Voici la commande que j'utilise (Kern) :
mailcommand = "/home/kern/bacula/bin/bsmtp -h mail.example.com -f \"\(Bacula\) %r\" -s \"Bacula: %t %e of %c %l\" %r"
Notez que la commande entière devrait apparaître sur une seulle ligne plutôt que découpée comme ici pour des raisons de présentation.
Le programme bsmtp est fourni en tant que partie de Bacula. Pour plus de détails, consultez la section bsmtp -- Personnaliser l'envoi de vos message par courrier électronique. Testez soigneusement toute commande mailcommand pour vous assurer que votre passerelle bsmtp accepte le format d'adressage que vous utilisez. Certains programmes tels Exim peut se montrer très sélectif en ce qui concerne les format autorisés, particulièrement en ce qui concerne le champ "from".
Où la destination peut être l'une des suivantes :
Où address dépend de la destination, qui peut être l'une des suivantes :
syslog = all, !skipped, !saved
Voici un exemple d'une définition de ressource Messages valide, où tous les messages sont envoyés par courrier électronique à enforcement@sec.com à l'exception de ceux concernant les fichiers explicitement exclus (skipped), et des messages d'arrêt de daemon (terminate). De plus, tous les messages de type mount sont envoyés à l'opérateur (courrier à enforcement@sec.com). Enfin, tous les messages autres que ceux relatifs aux fichiers explicitement exclus et aux fichiers sauvegardés sont envoyés vers la console :
Messages { Name = Standard mail = enforcement@sec.com = all, !skipped, !terminate operator = enforcement@sec.com = mount console = all, !skipped, !saved }
A l'exception de l'adresse électronique (modifiée pour éviter le spam), voici la ressource Message du Director de Kern. Notez que les commandes mailcommand et operatorcommand sont sur une seule ligne et non coupées comme ici pour des besoins de mise en page.
Messages { Name = Standard mailcommand = "bacula/bin/bsmtp -h mail.example.com \ -f \"\(Bacula\) %r\" -s \"Bacula: %t %e of %c %l\" %r" operatorcommand = "bacula/bin/bsmtp -h mail.example.com \ -f \"\(Bacula\) %r\" -s \"Bacula: Intervention needed \ for %j\" %r" MailOnError = security@example.com = all, !skipped, \ !terminate append = "bacula/bin/log" = all, !skipped, !terminate operator = security@example.com = mount console = all, !skipped, !saved }
Le fichier de configuration de la console est le plus simple de tous, et en général, vous n'aurez rien d'autre à changer que le mot de passe. Il contient simplement les informations nécessaires pour se connecter au(x) Director(s).
Pour une discussion générale sur les fichiers de configuration et leurs ressources ainsi que les types de données reconnus par Bacula, consultez le chapitre Configuration de ce manuel.
Les ressources Console suivantes doivent être définies :
La ressource Director définit les attributs du Director exécuté sur le réseau. Vous pouvez avoir plusieurs ressources Director dans un seul fichier de configuration de console. Dans ce cas, le programme Console vous demandera de choisir à son lancement le Director vous voulez contacter.
--
with-base-port de la commande configure. Ce
port doit être identique à celui spécifié par la directive DIRport de la
ressource Director du fichier de configuration du Director.
La valeur par défaut est 9101, aussi cet directive n'est normalement pas
spécifiée.
Voici un exemple réel :
Director { Name = HeadMan address = rufus.cats.com password = xyz1erploit }
Cette ressource n'est disponible que dans la version GNOME de la console. Elle vous permet de définir les fontes que vous voulez utiliser pour l'affichage dans la fenêtre principale.
Font = "LucidaTypewriter 9"
Merci à Phil Stracchino d'avoir fourni le code pour cette fonctionnalité.
Un autre exemple :
ConsoleFont { Name = Default Font = "Monospace 10" }
Depuis la version 1.33 de Bacula, il existe trois types de consoles différents avec lesquels l'administrateur ou l'utilisateur peut interagir avec le Director. Ces trois types de consoles comportent trois niveaux de sécurité.
Ce second type de console ne possède, par défaut, aucun privilège. Pour lui en accorder, vous devez les spécifier explicitement dans la ressource Console du Director.
Ainsi, vous pouvez avoir plusieurs consoles avec des noms et mots de passe distincts et avec, pour chacune, des privilèges différents. Ces consoles ne peuvent, par défaut, absolument rien faire. C'est vous qui leur accordez des privilèges, ou plutôt l'accès à des commandes et ressources en spécifiant des listes de contrôle d'accès dans la ressource Console du Director. L'administrateur peut ainsi contrĆ“ler finement les actions permises Ć des consoles (ou utilisateurs) particuliers.
La ressource Console est optionnelle. Cependant, si elle est spécifiée, vous pouvez utiliser les ACLs (listes de contrôle d'accès) dans le fichier de configuration du Director pour restreindre une console particulière ou un utilisateur pour qu'ils ne voient que les informations relatives à ses jobs ou à sa machine cliente.
Vous pouvez spécifier autant de ressources Console que vous voulez dans le fichier de configuration de la console. Cependant, si vous avez plusieurs ressources Director (c'est à dire si vous vous connectez à plusieurs directors distincts), vous pouvez associer l'une de ces ressources Console à une ressource Director particulière. Ainsi, lorsque vous choisirez un Director particulier, la configuration de console appropriée sera utilisée. Consultez le paragraphe ci-dessous sur la directive "Director" de la ressource Console pour plus d'informations sur le sujet.
L'exemple ci-dessous nous a été fournit par Phil Stracchino. Si nous définissons ce qui suit dans le fichier bconsole.conf (ou bwx-console.conf) d'un utilisateur :
Director { Name = MyDirector DIRport = 9101 Address = myserver Password = "XXXXXXXXXXX" # no, really. this is not obfuscation. } Console { Name = restricted-user Password = "UntrustedUser" }
Où le mot de passe dans la section Director est délibérément incorrect, et la console a reçu le nom restricted-client. Si d'autre part, dans le fichier de configuration du Director (inaccessible diectement à l'utilisateur), nous définissons :
Console { Name = restricted-user Password = "UntrustedUser" JobACL = "Restricted Client Save" ClientACL = restricted-client StorageACL = main-storage ScheduleACL = *all* PoolACL = *all* FileSetACL = "Restricted Client's FileSet" CatalogACL = DefaultCatalog CommandACL = run }
Alors, l'utilisateur de connectant au Director depuis cette console sera connecté en tant que restricted-client, et il ne sera en mesure que de voir ou accéder au job nommé Restricted Client Save, avec le stockage main-storage, n'importe quel planification ou pool, un FileSet nommé Restricted Client's File, un catalogue nommé DefaultCatalog, et la seule commande qu'il pourra utiliser dans la console sera la commande run. En d'autres termes, cet utilisateur est plutôt limité dans ce qu'il peut voir ou faire avec Bacula.
Voici un exemple d'un fichier de configuration de bconsole qui permet à plusieurs Directors avec différentes Consoles selon le Director :
Director { Name = MyDirector DIRport = 9101 Address = myserver Password = "XXXXXXXXXXX" # no, really. this is not obfuscation. } Director { Name = SecondDirector DIRport = 9101 Address = secondserver Password = "XXXXXXXXXXX" # no, really. this is not obfuscation. } Console { Name = restricted-user Password = "UntrustedUser" Director = MyDirector } Console { Name = restricted-user Password = "A different UntrustedUser" Director = SecondDirector }
Le second Director, référencé en tant que "secondserver" pourrait ressembler à ceci :
Console { Name = restricted-user Password = "A different UntrustedUser" JobACL = "Restricted Client Save" ClientACL = restricted-client StorageACL = second-storage ScheduleACL = *all* PoolACL = *all* FileSetACL = "Restricted Client's FileSet" CatalogACL = RestrictedCatalog CommandACL = run, restore WhereACL = "/" }
Pour plus de détails concernant la console et ses commandes, consultez le chapitre La console Bacula de ce manuel.
Voici un exemple de fichier de configuration de console :
# # Bacula Console Configuration File # Director { Name = HeadMan address = "my_machine.my_domain.com" Password = Console_password }
The Monitor configuration file is a stripped down version of the Director configuration file, mixed with a Console configuration file. It simply contains the information necessary to contact Directors, Clients, and Storage daemons you want to monitor.
For a general discussion of configuration file and resources including the data types recognized by Bacula, please see the Configuration chapter of this manual.
The following Monitor Resource definition must be defined:
The Monitor resource defines the attributes of the Monitor running on the network. The parameters you define here must be configured as a Director resource in Clients and Storages configuration files, and as a Console resource in Directors configuration files.
The Director resource defines the attributes of the Directors that are monitored by this Monitor.
As you are not permitted to define a Password in this resource, to avoid obtaining full Director privileges, you must create a Console resource in the Director's configuration file, using the Console Name and Password defined in the Monitor resource. To avoid security problems, you should configure this Console resource to allow access to no other daemons, and permit the use of only two commands: status and .status (see below for an example).
You may have multiple Director resource specifications in a single Monitor configuration file.
--
with-base-port option of the ./configure command. This port must be
identical to the DIRport specified in the Director resource of
the
Director's configuration file. The
default is 9101 so this record is not normally specified.
The Client resource defines the attributes of the Clients that are monitored by this Monitor.
You must create a Director resource in the Client's configuration file, using the Director Name defined in the Monitor resource. To avoid security problems, you should set the Monitor directive to Yes in this Director resource.
You may have multiple Director resource specifications in a single Monitor configuration file.
The Storage resource defines the attributes of the Storages that are monitored by this Monitor.
You must create a Director resource in the Storage's configuration file, using the Director Name defined in the Monitor resource. To avoid security problems, you should set the Monitor directive to Yes in this Director resource.
You may have multiple Director resource specifications in a single Monitor configuration file.
There is no security problem in relaxing the permissions on tray-monitor.conf as long as FD, SD and DIR are configured properly, so the passwords contained in this file only gives access to the status of the daemons. It could be a security problem if you consider the status information as potentially dangereous (I don't think it is the case).
Concerning Director's configuration:
In tray-monitor.conf, the password in the Monitor resource must point to
a restricted console in bacula-dir.conf (see the documentation). So, if
you use this password with bconsole, you'll only have access to the
status of the director (commands status and .status).
It could be a security problem if there is a bug in the ACL code of the
director.
Concerning File and Storage Daemons' configuration:
In tray-monitor.conf, the Name in the Monitor resource must point to a
Director resource in bacula-fd/sd.conf, with the Monitor directive set
to Yes (once again, see the documentation).
It could be a security problem if there is a bug in the code which check
if a command is valid for a Monitor (this is very unlikely as the code
is pretty simple).
An example Tray Monitor configuration file might be the following:
# # Bacula Tray Monitor Configuration File # Monitor { Name = rufus-mon # password for Directors Password = "GN0uRo7PTUmlMbqrJ2Gr1p0fk0HQJTxwnFyE4WSST3MWZseR" RefreshInterval = 10 seconds } Client { Name = rufus-fd Address = rufus FDPort = 9102 # password for FileDaemon Password = "FYpq4yyI1y562EMS35bA0J0QC0M2L3t5cZObxT3XQxgxppTn" } Storage { Name = rufus-sd Address = rufus SDPort = 9103 # password for StorageDaemon Password = "9usxgc307dMbe7jbD16v0PXlhD64UVasIDD0DH2WAujcDsc6" } Director { Name = rufus-dir DIRport = 9101 address = rufus }
Click here to see the full example.
# # Restricted Director, used by tray-monitor to get the # status of the file daemon # Director { Name = rufus-mon Password = "FYpq4yyI1y562EMS35bA0J0QC0M2L3t5cZObxT3XQxgxppTn" Monitor = yes }
Click here to see the full example.
# # Restricted Director, used by tray-monitor to get the # status of the storage daemon # Director { Name = rufus-mon Password = "9usxgc307dMbe7jbD16v0PXlhD64UVasIDD0DH2WAujcDsc6" Monitor = yes }
Click here to see the full example.
# # Restricted console used by tray-monitor to get the status of the director # Console { Name = Monitor Password = "GN0uRo7PTUmlMbqrJ2Gr1p0fk0HQJTxwnFyE4WSST3MWZseR" CommandACL = status, .status }
La console Bacula (parfois désignée "Agent utilisateur") est un programme qui permet à l'utilisateur autorisé ou à l'administrateur système d'interagir avec le Director.
Actuellement, la console Bacula existe en deux versions : une interface shell (façon TTY), et une interface graphique GNOME. Avec la console Bacula, vous pouvez déterminer l'état d'un job particulier, examiner le contenu du catalogue et effectuer certaines manipulations de cartouches.
Il existe d'autre part un programme nommé bwx-console, bâtie avec wxWidgets qui offre une interface graphique aux opérations de restauration.
Etant donné que la Console interagit avec le Director au travers du réseau, il n'est pas nécessaire que les deux programmes résident sur la même machine.
Bacula a besoin d'un minimum de retour de la Console afin de pouvoir utiliser plus d'une cartouche. En effet, lorsqu'il en réclame une nouvelle, il attend jusqu'à ce qu'un opérateur lui indique, via la Console, qu'une nouvelle cartouche est montée.
Lors de son lancement, la Console lit le fichier de configuration standard nommé bconsole.conf (ou gnome-console.conf dans le cas de la version GNOME) Ce fichier définit une configuration par défaut de la Console et, à l'heure actuelle, la seule ressource définie est la ressource Director, qui informe la Console du nom et de l'adresse du Director. Pour plus d'informations sur la configuration de la Console, voyez le chapitre Configurer la Console de ce manuel.
Le programme Console admet les options suivantes :
Usage: bconsole [-s] [-c config_file] [-d debug_level] -c <file> set configuration file to file -dnn set debug level to nn -n no conio -s no signals -t test - read configuration and exit -? print this message.
Après son démarrage, la Console est en attente de vos commandes, ce qui est indiqué par une astérisque (*) (ce n'est pas le cas dans la version GNOME où vous saisissez vos commandes dans la boite texte en bas de l'écran). Vous pouvez, pour toutes les commandes, vous contenter d'entrer le nom de la commande, la Console se chargera de vous demander les arguments nécessaires, mais dans la plupart des cas, vous pouvez entrer les commandes suivies de leurs arguments. Le format général est :
<command> <keyword1>[=<argument1>] <keyword2>[=<argument2>] ...
Où command est l'une des commandes énumérées ci-dessous, keyword est l'un des mots-clef énumérés ci-dessous (usuellement suivi d'un argument), et argument est la valeur du mot-clef. La commande peut être abrégée jusqu'à sa plus courte abréviation unique. Si deux commandes commencent par les mêmes lettres, c'est celle qui apparaît en tête dans la liste fournie par la commande help qui sera sélectionnée si votre abréviation est ambigüe. Aucun des mots-clef suivant la commande ne peut être abrégé.
Par exemple :
list files jobid=23
énumère les fichiers sauvegardés par le job de JobId 23.
Cette autre commande :
show pools
affiche toutes les ressources Pool.
Normalement, le programme Console se termine si vous saisissez quit ou exit. Cependant, il il attend jusq"à ce que le Director ait pris en compte la commande, ce qui peut prendre du temps si ce dernier est déjà occupé à une tâche longue (par exemple, un élagage du catalogue). Si vous voulez quitter la Console immédiatement, utilisez la commande .quit.
Il n'existe actuellement aucun moyen d'interrompre une commande de la Console une fois lancée (Ctrl-C ne marche pas). En revanche, à l'invite d'une commande vous demandant de choisir parmi plusieurs possibilités, vous pouvez annuler la commande en entrant un point (.), vous serez dans la plupart des cas ramené à l'invite principal, ou à l'invite précédente, dans le cas de choix imbriqués. En quelques endroits, comme celui où l'on vous demande un nom de volume, le point sera pris pour la réponse (Bacula pensera que vous voulez nommer votre volume "."). Dans cette situation, vous serez la plupart du temps en mesure d'annuler à l'invite suivante.
jobid=536
Notez que cette liste est probablement incomplète, car le processus de création est toujours en cours. Il se peut aussi qu'elle ne soit pas dans l'ordre alphabétique.
Les commandes suivantes sont actuellement implémentées :
Lorsque l'affichage automatique est désactivé, vous devez explicitement en demander l'affichage avec la commande messages.
Une fois qu'un job est marqué "A supprimer", il peut se passer quelques instants (en général, moins d'une minute) avant qu'il se termine, en fonction des opérations en cours.
Si, au lancement d'un job, Bacula détermine qu'il n'y a pas de pool enregistré dans le catalogue, mais qu'il existe une ressource Pool pour le pool approprié, alors il le créé pour vous. Si vous voulez le voir apparaître immédiatement dans le catalogue, utilisez cette commande pour forcer sa création immédiate.
Voici la forme complète de cette commande :
delete pool=\lt{}pool-name\gt{}
supprime un pool du catalogue.
delete volume=\lt{}volume-name\gt{} pool=\lt{}pool-name\gt{}
supprime du catalogue un volume du pool spécifié.
delete JobId=\lt{}job-id\gt{} JobId=\lt{}job-id2\gt{} ...
supprime du catalogue le job spécifié.
delete Job JobId=n,m,o-r,t ...
supprime les job de JobIds m,n,o,p,q,r et t (où m,n,... sont, bien sur, des nombres). Ainsi, la commande "delete jobid" accepte les listes et les plages de jobids.
Optionnellement, vous pouvez ajouter le mot-clef listing, auquel cas tous les fichiers à sauvegarder seront affichés. Notez qu'un tel affichage peut prendre un certain temps s'il s'agit d'une grosse sauvegarde. Voici la forme complète de cette commande :
estimate job=\lt{}job-name\gt{} listing client=\lt{}client-name\gt{} fileset=\lt{}fileset-name\gt{} level=\lt{}level-name\gt{}
La spécification du job est suffisante, mais vous pouvez aussi passer outre le client, le FileSet et/ou le niveau en les spécifiant sur la ligne de commande.
Par exemple, vous pourriez faire ceci :
@output /tmp/listing estimate job=NightlySave listing level=Incremental @output
ce qui produirait une liste complète de tous les fichiers à sauvegarder pour le job NightlySave au cours d'une sauvegarde incrémentale, et qui consignerait cette liste dans le fichier /tmp/listing. Notez que l'Ć©valuation produite par cette commande se base sur les tailles de fichiers contenues dans l'objet "rĆ©pertoire", aussi l'estimation peut ĆŖtre trĆØs Ć©loignĆ©e de la rĆ©alitĆ© si vous avez des fichiers creux (NDT : sparse files) sur votre systĆØme. Ce type de fichiers se rencontre souvent sur les systĆØmes 64 bits avec certains systĆØmes de fichiers. Le volume obtenu par l'Ć©valuation est celui que sauvegardera Bacula si l'option sparse est dĆ©sactivĆ©e. Il n'y a actuellement aucun moyen d'Ć©valuer le volume de ce qui serait sauvegardĆ© avec l'option sparse activĆ©e.
label storage=\lt{}storage-name\gt{} volume=\lt{}volume-name\gt{} slot=\lt{}slot\gt{}
Si vous omettez l'un quelconque des arguments, il vous sera réclamé. Le type de média est automatiquement récupéré de la ressource Storage. Une fois que les informations requises sont réunies, la Console contacte le Storage Daemon spécifié et lui ordonne d'étiqueter la cartouche spécifiée. Si l'étiquetage s'effectue correctement, la Console créé un nouvel enregistrement dans le catalogue pour le volume dans le pool approprié.
Les noms de volumes ne doivent contenir que des lettres, chiffres et les caractères spéciaux tiret (-), sous-ligné (_), double-point (:), et point (.). Tous les autres caractères, y compris l'espace, sont illégaux. Cette restriction vise à assurer une bonne lisibilité des noms de volumes pour réduire le risque d'erreurs humaines.
Notez que lors de l'étiquetage d'une cartouche vierge, Bacula obtient des erreurs read I/O error lorqu'il tente de vérifier si la cartouche a déjà un label. Si vous voulez éviter ce genre de message, placez un indicateur de fin de fichier sur votre cartouche avant son étiquetage :
mt rewind mt weof
La commande label peut échouer pour plusieurs raisons :
Il existe deux moyens pour ré-étiqueter un volume qui porte déjà une étiquette Bacula. La méthode brutale consiste à écrire une marque de fin de fichier sur la cartouche vec la commande du système d'exploitation mt, quelque chose dans ce style :
mt -f /dev/st0 rewind mt -f /dev/st0 weof
puis d'utiliser la commande label pour ajouter une nouvelle étiquette. Cette méthode peut cependant laisser des traces de l'ancien volume dans le catalogue.
Il est préférable d'utiliser la commande relabel décrite ci-dessous sur un volume purgé (automatiquement ou avec la commande purge).
Si votre librairie comporte un lecteur de codes barres, vous pouvez étiqueter tous les volumes qu'elle contient en utilisant la commande label barcodes. En effet, après le lancement de cette commande, Bacula monte chaque cartouche l'une après l'autre et l'étiquette du nom de son code barres. simultanément, l'enregistrement approprié est créé dans le catalogue. Toute cartouche dont le code barres commence par les mêmes caractères que ceux spécifiés par la directive "CleaningPrefix" de la ressource Pool du director est considérée comme une cartouche de nettoyage et ne reçoit donc pas d'étiquette, bien qu'une entrée dans le catalogue lui soit dédiée. Par exemple avec :
Pool { Name ... Cleaning Prefix = "CLN" }
tout slot contenant une cartouche de code barres CLNxxxxx sera traitée en tant que cartouche de nettoyage et ne sera jamais montée. Notez que la forme complète de la commande est :
update storage=xxx pool=yyy slots=1-5,10 barcodes
list jobs list jobid=<id> (affiche le jobid id) list ujobid=<unique job name> (affiche le job dont le nom unique est <unique job name>) list job=<job-name> (Affiche tous les jobs dont le nom est "job-name") list jobname=<job-name> (voir ci-dessus) Dans cette commande, vous pouvez ajouter "limit=nn" pour limiter la sortie \`a nn jobs. list jobmedia list jobmedia jobid=<id> list jobmedia job=<job-name> list files jobid=<id> list files job=<job-name> list pools list clients list jobtotals list volumes list volumes jobid=<id> list volumes pool=<pool-name> list volumes job=<job-name> list volume=<volume-name> list nextvolume job=<job-name> list nextvol job=<job-name> list nextvol job=<job-name> days=nnn
Ce que font la plupart des commandes ci-dessus devrait être plus ou moins évident. En général, si vous ne spécifiez pas tous les arguments requis, la Console vous sollicitera pour les arguments manquants.
La commande list nextvol affiche le nom du volume qui dera utilisé par le job spécifié. Soyez conscient que le prochain volume utilisé pour un job dépend de nombreux facteurs dont le temps, et les autres jobs qui seront exécutés avant celui spécifié, qui peuvent remplir une cartouche qui était vide au moment de l'exécution de list nextvol. Aussi, considérez la réponse fournie par cette commande comme une bonne estimation plutôt que comme une réponse définitive. De plus, cette commande a certains effets de bord : étant donné qu'elle exécute le même algorithme qu'un job, elle est susceptible de purger ou recycler un volume. Par défaut, le job spécifié doit être exécuté dans les deux jours ou aucun volume ne sera trouvé. Vous pouvez cependant spécifier jusqu'à 50 jours en avant avec la directive days=nnn. Si, par exemple, un vendredi, vous voulez savoir quel volume sera requis lundi pour le job MyJob, utilisez list nextvol job=MyJob days=3.
Si vous souhaitez ajouter vos propres commandes pour interroger le catalogue, vous pouvez les placer dans le fichier query.sql. Cela demande quelques connaissances du langage SQL. Voyez le paragraphe sur la commande query ci-dessous pour plus d'informations. Voyez aussi le paragraphe sur la commande llist qui permet l'affichage complet des informations du catalogue.
Voici un exemple d'affichage produit par la commande list pools :
+------+---------+---------+---------+----------+-------------+ | PoId | Name | NumVols | MaxVols | PoolType | LabelFormat | +------+---------+---------+---------+----------+-------------+ | 1 | Default | 0 | 0 | Backup | * | | 2 | Recycle | 0 | 8 | Backup | File | +------+---------+---------+---------+----------+-------------+
Comme mentionné précédemment, la commande list affiche des informations du catalogue. Certais éléments sont ajoutés dans le catalogue dès le démarrage de Bacula, mais en général, la plupart ne le sont que lorsqu'ils sont utilisés pour la première fois. C'est le cas des clients, des jobs, etc.
Bacula créé une entrée relative à un nouveau client dans le catalogue la première fois que vous exécutéz un job pour ce client. L'entrée est créée que le job aboutisse ou qu'il échoue, mais il doit au moins démarrer. Lorsque le client est contacté, des informations supplémentaires sont récupérées du client (le résultat d'un "uname -a") et ajoutées au catalogue. Un status n'entraîne pas l'enregistrement dans le catalogue.
Si vous voulez visualiser les ressources Client disponibles dans votre catalogue, utilisez la commande show clients.
Si, au lieu du list pools de l'exemple précédent, vous saisissez llist pools, vous obtiendrez un affichage de ce genre :
PoolId: 1 Name: Default NumVols: 0 MaxVols: 0 UseOnce: 0 UseCatalog: 1 AcceptAnyVolume: 1 VolRetention: 1,296,000 VolUseDuration: 86,400 MaxVolJobs: 0 MaxVolBytes: 0 AutoPrune: 0 Recycle: 1 PoolType: Backup LabelFormat: * PoolId: 2 Name: Recycle NumVols: 0 MaxVols: 8 UseOnce: 0 UseCatalog: 1 AcceptAnyVolume: 1 VolRetention: 3,600 VolUseDuration: 3,600 MaxVolJobs: 1 MaxVolBytes: 0 AutoPrune: 0 Recycle: 1 PoolType: Backup LabelFormat: File
La commande mount est utilisée pour obtenir de Bacula qu'il lise un volume chargé dans un lecteur. C'est un moyen d'indiquer à Bacula que vous avez chargé une cartouche qu'il doit examiner. Cette commande n'est utilisée que lorsque Bacula a demandé votre intervention pour charger un lecteur vide, ou lorsque vous avez explicitement démonté un volume avec la commande unmount dans la Console, ce qui provoque la fermeture du lecteur. Si vous avez une librairie, vous ne ferez pas opérer Bacula dessus avec la commande mount. Voici les différentes formes de cette commande :
mount storage=<storage-name>
mount [ jobid=<id> | job=<job-name> ]
Si vous avez spécifié Automatic Mount = yes dans la ressource Device du Storage Daemon, Alors Bacula pourra accéder automatiquement au volume, à moins que vous ne l'ayez explicitement démonté (unmount) dans la Console.
La commande python restart réinitialise l'interpréteur Python du Director. Ceci peut être très utile pour effectuer des tests, car une fois que le Director est lancé, et l'interpréteur Python initialisé, il n'y a pas d'autre moyen de lui faire intégrer des modifications du script de démarrage DirStartUp.py. Pour plus de détails sur l'écriture de scripts Python, consultez le chapitre Ecrire des scripts Python.
prune files|jobs|volume client=<client-name> volume=<volume-name>
Pour qu'un volume soit élagué, son VolStatus doit être Full, Used, ou Append, faute de quoi l'élagage sera sans effet.
purge files jobid=<jobid>|job=<job-name>|client=<client-name>
purge jobs client=<client-name> (of all jobs)
purge volume|volume=<vol-name> (of all jobs)
Pour qu'un volume puisse être purgé, son VolStatus doit être Full, Used, ou Append, faute de quoi la purge sera sans effet.
relabel storage=<storage-name> oldvolume=<old-volume-name> volume=<newvolume-name>
Si vous omettez l'un quelconque des arguments, la console vous sollicitera pour obtenir les informations manquantes. Pour qu'un volume puisse être ré-étiqueté, il doit figurer dans le catalogue, et avoir le statut Purged ou Recycle. Cette situation peut se présenter automatiquement par l'application des périodes de rétention, ou vous pouvez l'obtenir par une purge explicite du volume.
Une fois que le volume a été physiquement ré-étiqueté, les données qu'il contenait sont définitivement et irrémédiablement perdues.
release storage=<storage-name>
Après cette commande, le lecteur est gardé à l'état ouvert par Bacula (sauf si l'option Always Open est désactivée dans la configuration du Storage Daemon), et il ne peut donc être utilisé par un autre programme. Toutefois, il est possible, avec certains lecteurs, de changer la cartouche à ce stade. Lors du prochain job, Bacula saura relire l'étiquette de la cartouche pour savoir laquelle est montée. Si vous voulez être en mesure d'utiliser le lecteur avec un autre programme, par exemple mt, vous devez uiliser la commande unmount pour que Bacula le libère complètement.
Bien qu'il soit possible de recharger la configuration du Director à la volée, alors même que des jobs sont en cours d'exécution, il faut garder à l'esprit que c'est une opération complexe, qui n'est pas dénuée d'effets de bords. C'est pourquoi il est recommandé, si vous êtes amené à utiliser la commande reload, de redémarrer le Director dès que vous en aurez l'opportunité.
restore storage=<storage-name> client=<client-name> where=<path> pool=<pool-name> fileset=<fileset-name> select current all done
Où l'option current, si elle est spécifiée, indique à la commande restore de sélectionner automatiquement la sauvegarde la plus récente (sinon, vous serez sollicité à ce sujet). L'option all, si elle est spécifiée, indique à la commande restore de restaurer tous les fichiers (sinon, vous serez sollicité à ce sujet). Pour plus de détails concernant la commande restore, consultez le chapitre Restaurations avec Bacula.
run job=<job-name> client=<client-name> fileset=<FileSet-name> level=<level-keyword> storage=<storage-name> where=<directory-prefix> when=<universal-time-specification> yes
Toute information omise quoique requise fait l'objet d'une liste de sélection, et avant le lancement du job, un bilan des paramètres vous est présenté avec options d'accord, refus et modification, à moins que vous ayez spécifié yes, auquel cas le job est immédiatement envoyé vers le planificateur.
Sur mon système, j'obtiens ce qui suit lorsque je lance la commande run :
A job name must be specified. The defined Job resources are: 1: Matou 2: Polymatou 3: Rufus 4: Minimatou 5: Minou 6: PmatouVerify 7: MatouVerify 8: RufusVerify 9: Watchdog Select Job resource (1-9):
Si je choisis le numéro 5, j'obtiens :
Run Backup job JobName: Minou FileSet: Minou Full Set Level: Incremental Client: Minou Storage: DLTDrive Pool: Default When: 2003-04-23 17:08:18 OK to run? (yes/mod/no):
Si maintenant j'entre yes, le job est exécuté. Si je choisis mod, voici les otpions qui me sont proposées :
Parameters to modify: 1: Level 2: Storage 3: Job 4: FileSet 5: Client 6: When 7: Pool Select parameter to modify (1-7):
Vous pouvez, si vous le souhaitez, démarrer un job plus tard, en utilisant le paramètre When. Pour cela, faites le choix mod, puis sélectionnez When (no. 6) et enfin, saisissez l'heure et le jour de lancement désirés au format AAAA-MM-JJ HH:MM:SS.
Cette commande est utilisée pour paramétrer le niveau de débuggage de chaque daemon. Voici la forme complète de cette commande.
setdebug level=nn [trace=0/1 client=<client-name> | dir | director | storage=<storage-name> | all]
Si le paramètre de traçage est actif (trace=1), alors le daemon est placé en mode traçage, ce qui signifie que toutes les informations de débuggage sont envoyées vers le fichier bacula.trace dans le répertoire courant du daemon. En principe, ce n'est nécessaire que pour le débuggage des clients Win32 où les informations ne peuvent être envoyées vers un terminal ou redirigées vers un fichier. en mode traçage, chaque message de débuggage est ajouté au fichier, que vous devez supprimer explicitement lorsque vous avez fini.
Cette commande vous permet d'interroger directement le catalogue. Notez que vous devriez savoir exactement ce que vous faites en utilisant cette commande, car vous pouvez endommager sérieusement votre catalogue. Consultez le paragraphe relatif à la commande query qui offre un moyen à la fois plus simple et plus sur de saisir des requêtes SQL.
En fonction du moteur de base de données que vous utilisez (MySQL, PostgreSQL ou SQLite), vous disposerez de commandes quelque peu différentes. Pour plus de détails, référez-vous aux documentations de MySQL, PostgreSQL ou SQLite.
status [all | dir=<dir-name> | director | client=<client-name> | storage=<storage-name> | days=nnn]
Si vous entrez status dir, la Console énumère tous les jobs en cours d'exécution, un résumé des jobs planifié pour exécution au cours des prochaines 24 heures incluant le volume qui sera probablement utilisé, et donne la liste des dix derniers jobs exécutés avec leurs états. Gardez à l'esprit les deux éléments suivants :
Dans la liste des jobs en cours d'exécutions, vous pouvez trouver ce type d'informations :
2507 Catalog MatouVerify.2004-03-13_05.05.02 is waiting execution 5349 Full CatalogBackup.2004-03-13_01.10.00 is waiting for higher priority jobs to finish 5348 Differe Minou.2004-03-13_01.05.09 is waiting on max Storage jobs 5343 Full Rufus.2004-03-13_01.05.04 is running
La liste ci-dessus indique que le job de JobId 5343 (Rufus) est en cours. Le job de JobId 5348 (Minou) est en attente de la fin du job 5343 qui utilise la même ressource Storage, ce qui provoque le "waiting on max Storage jobs". Le job de JobId 5349 a une priorité inférieure à celle de tous les autres jobs, aussi, il est en attente de la fin de jobs de priorités supérieures. Finalement, le job de jobId 2508 (MatouVerify) est en attente ("waiting execution") car un seul job peut être exécuté en même temps.
Si vous faites un status dir, Bacula affiche par défaut les premières occurrences de tous les jobs prévus pour exécution aujourd'hui et demain. Si vous voulez voir les jobs prévus pour les trois prochains jours, (Si, par exemple vendredi, vous voulez voir les premières occurrences des cartouches à utiliser vendredi, samedi, dimanche et lundi), vous pouvez ajouter l'option days=3. Notez, days=0 montre les premières occurrences des jobs planifiés pour être exécutés aujourd'hui seulement. Si vous avez plusieurs exécutions planifiées, pour chaque job, seule la première occurrence sera affiché pour la période spécifiée.
Si votre job paraît bloqué, vous pouvez avoir une idée générale du problème en utilisant status dir, et obtenir une information plus spécifique avec status storage=xxx. Par exemple, si j'utilise cette dernière commande sur un système inoccupé, j'obtiens :
status storage=File Connecting to Storage daemon File at 192.168.68.112:8103 rufus-sd Version: 1.39.6 (24 March 2006) i686-pc-linux-gnu redhat (Stentz) Daemon started 26-Mar-06 11:06, 0 Jobs run since started. Running Jobs: No Jobs running. ==== Jobs waiting to reserve a drive: ==== Terminated Jobs: JobId Level Files Bytes Status Finished Name ====================================================================== 59 Full 234 4,417,599 OK 15-Jan-06 11:54 kernsave ==== Device status: utochanger "DDS-4-changer" with devices: "DDS-4" (/dev/nst0) Device "DDS-4" (/dev/nst0) is mounted with Volume="TestVolume002" Pool="*unknown*" Slot 2 is loaded in drive 0. Total Bytes Read=0 Blocks Read=0 Bytes/block=0 Positioned at File=0 Block=0 Device "Dummy" is not open or does not exist. No DEVICE structure. Device "DVD-Writer" (/dev/hdc) is not open. Device "File" (/tmp) is not open. ==== In Use Volume status: ====
Ce qui révèle qu'aucun job n'est en cours d'exécution, et qu'aucun des périphériques n'est en cours d'utilisation. Si je démonte la librairie (unmount), qui ne sera plus utilisée dans cet exemple, et que je lance un job qui utilise le stockage File, le job se bloque. Si je demande à nouveau status storage=xxx, j'obtiens :
status storage=File ... Device status: Autochanger "DDS-4-changer" with devices: "DDS-4" (/dev/nst0) Device "DDS-4" (/dev/nst0) is not open. Device is BLOCKED. User unmounted. Drive 0 is not loaded. Device "Dummy" is not open or does not exist. No DEVICE structure. Device "DVD-Writer" (/dev/hdc) is not open. Device "File" (/tmp) is not open. Device is BLOCKED waiting for media. ==== ...
Il devrait maintenant être clair que si un job nécessitant la librairie est exécuté, il bloquera en raison du démontage de cette dernière par l'utilisateur. Mais le problème pour le job que j'ai lancé avec le périphérique "File" est que le périphérique est bloqué en attente d'un media : Bacula a besoin que vous étiquetiez un volume.
unmount storage=<storage-name> unmount [ jobid=<id> | job=<job-name> ]
media, volume, pool, slots
Dans le cas de la mise à jour d'un volume, vous serez interrogé sur le paramètre que vous voulez modifier. Voici ces paramètres :
Volume Status Volume Retention Period Volume Use Duration Maximum Volume Jobs Maximum Volume Files Maximum Volume Bytes Recycle Flag Slot InChanger Flag Pool Volume Files Volume from Pool All Volumes from Pool
Pour le paramètre slot, update slots, Bacula obtient une liste de tous les slots et de leurs codes barres du Storage Daemon, pour chaque code barres trouvé, le slot est mis à jour dans l'enregistrement Media du catalogue. C'est très pratique si vous déplacez des cartouches dans la librairie, ou si vous changez des magasins de cartouches. Dans la foulée, le drapeau InChanger est aussi mis à jour.Ceci permet à BAcula de savoir quels cartouches sont effectivement dans la librairie.
Si vous n'avez pas de lecteur de codes barres, vous pouvez faire la même chose depuis la version 1.33 grâce à la commande update slots scan. Le mot-clef scan ordonne à Bacula de monter physiquement chaque cartouche afin de lire son VolumeName.
Pour le paramètre Pool, update pool, Bacula déplace le volume de son pool courant vers le pool spécifié.
Pour les parmètres Volume from Pool et All Volumes from Pool, les valeurs suivantes sont mises à jour depuis l'enregistrement de pool : Recycle, VolRetention, VolUseDuration, MaxVolJobs, MaxVolFiles, and MaxVolBytes.
Voici la forme complète de la commande update :
update volume=xxx pool=yyy slots volstatus=xxx VolRetention=ddd VolUse=ddd MaxVolJobs=nnn MaxVolBytes=nnn Recycle=yes|no slot=nnn enabled=n
use <database-name>
Les requêtes suivantes sont actuellement (version 1.24) disponibles :
Available queries: 1: List Job totals: 2: List where a file is saved: 3: List where the most recent copies of a file are saved: 4: List total files/bytes by Job: 5: List total files/bytes by Volume: 6: List last 20 Full Backups for a Client: 7: List Volumes used by selected JobId: 8: List Volumes to Restore All Files: 9: List where a File is saved: Choose a query (1-9):
wait [jobid=nn] [jobuid=unique id] [job=job name]
Voici une liste de commandes préfixées d'un point (.). Elles ont pour vocation d'être utilisées soit dans des programmes batch, soit par des interfaces graphiques. Elles ne sont, en principe, pas utilisées en mode interactif. Une fois que le développement d'interfaces graphiques aura démarré, cette liste s'accroîtra considérablement.
.backups job=xxx list backups for specified job .defaults client=xxx fileset=yyy list defaults for specified client .die cause the Director to segment fault (for debugging) .dir when in tree mode prints the equivalent to the dir command, but with fields separated by commas rather than spaces. .jobs list all job names .levels list all levels .filesets list all fileset names .clients list all client names .pools list all pool names .types list job types .msgs return any queued messages .messages get quick messages .help help command output .quit quit .status get status output .exit quit
Normalement, toutes les commandes saisies dans la Console sont immédiatement transmises au Director, qui peut résider sur une autre machine, afin d'y être exécutées. Il existe cependant quelques commandes, toutes précédées du caractère arobase (@), qui ne sont pas envoyées au Director, mais directement interprétées par la Console. Notez que seule la Console tty implémente ces commandes, et non la Console GNOME. En voici la liste :
@output /dev/null commands ... @output
./bconsole -c ./bconsole.conf <<END_OF_DATA unmount storage=DDS-4 quit END_OF_DATA
A l'exécution de ce fichier, le périphérique DDS-4 est démonté. Vous pouvez, si vous le souhaitez, exécuter cette tâche lors d'un job avec les directives RunBeforeJob ou RunAfterJob.
Il est aussi possible d'exécuter la Console à partir de l'entrée d'un fichier contenant les commandes comme suit :
./bconsole -c ./bconsole.conf <filename
où le fichier nommé filename contient un ensemble quelconque de commandes de la Console.
Voici un exemple réel, issu des tests de régression de Bacula. Il étiquette un volume (sur disque) exécute une sauvegarde puis une restauration des fichiers sauvegardés.
bin/bconsole -c bin/bconsole.conf <<END_OF_DATA @output /dev/null messages @output /tmp/log1.out label volume=TestVolume001 run job=Client1 yes wait messages @# @# now do a restore @# @output /tmp/log2.out restore current all yes wait messages @output quit END_OF_DATA
Les données issues de la sauvegarde sont envoyées vers /tmp/log1.out et celles de la restaurations vers /tmp/log2.out. Pour vérifier que les opérations se sont déroulées correctement, les fichiers de sorties sont contrôlés avec :
grep "^Termination: *Backup OK" /tmp/log1.out backupstat=$? grep "^Termination: *Restore OK" /tmp/log2.out restorestat=$?
Si vous avez utilisé la commande label pour étiqueter un volume, alors celui-ci est automatiquement ajouté au pool, et vous n'avez pas besoin de le faire manuellement.
Une alternative consiste à ajouter plusieurs volumes à un pool sans les étiqueter préalablement. Vous devrez alors les étiqueter plus tard, lorsque Bacula en aura besoin.
Avant d'ajouter un volume à un pool, vous devez connaître les informations suivantes :
Par exemple, pour ajouter un media à un pool, vous utiliseriez les commandes suivantes dans la Console :
*add Enter name of Pool to add Volumes to: Default Enter the Media Type: DLT8000 Enter number of Media volumes to create. Max=1000: 10 Enter base volume name: Save Enter the starting number: 1 10 Volumes created in pool Default *
Pour voir ce que vous avez ajouté, tapez :
*list media pool=Default +-------+----------+---------+---------+-------+------------------+ | MedId | VolumeNa | MediaTyp| VolStat | Bytes | LastWritten | +-------+----------+---------+---------+-------+------------------+ | 11 | Save0001 | DLT8000 | Append | 0 | 0000-00-00 00:00 | | 12 | Save0002 | DLT8000 | Append | 0 | 0000-00-00 00:00 | | 13 | Save0003 | DLT8000 | Append | 0 | 0000-00-00 00:00 | | 14 | Save0004 | DLT8000 | Append | 0 | 0000-00-00 00:00 | | 15 | Save0005 | DLT8000 | Append | 0 | 0000-00-00 00:00 | | 16 | Save0006 | DLT8000 | Append | 0 | 0000-00-00 00:00 | | 17 | Save0007 | DLT8000 | Append | 0 | 0000-00-00 00:00 | | 18 | Save0008 | DLT8000 | Append | 0 | 0000-00-00 00:00 | | 19 | Save0009 | DLT8000 | Append | 0 | 0000-00-00 00:00 | | 20 | Save0010 | DLT8000 | Append | 0 | 0000-00-00 00:00 | +-------+----------+---------+---------+-------+------------------+ *
Notez que la Console a automatiquement ajouté un numéro au nom de volume de base que vous avez spécifié ("Save" dans ce cas). Si vous ne souhaitez pas ce comportement, répondez simplement 0 (zéro) à la queston "Enter number of Media volumes to create . Max=1000:" et un seul volume sera créé avec le nom exact que vous avez spécifié.
Nous allons maintenant décrire la restauration de fichiers avec la commande restore de la Console, qui est le mode de restauration recommandé. Il existe cependant un programme indépendant nommé bextract, qui permet lui aussi de restaurer des fichiers. Pour plus d'informations sur ce programme, consultez le chapitre Programmes utilitaires Bacula de ce manuel. Vous y trouverez aussi des informations sur le programme bls qui sert à produire une liste du contenu de vos volumes, et sur le programme bscan qui vous sera utilie si vous voulez restaurer les enregistrements du catalogue relatifs à un ancien volume qui n'y figure plus.
En général, pour restaurer un fichier ou un ensemble de fichiers, vous devez exécuter un job de type restore, par conséquent, vous devez prédéfinir un tel job dans le fichier de configuration de votre Director. Les paramètres (Client, FileSet,...) que vous définissez ici ne sont pas importants, Bacula les ajustera automatiquement lors de l'utilisation de restore.
Bacula étant un programme réseau, il vous appartient de vous assurer que vous avez sélectionné le bon client et le bon disque dur pour recevoir la restauration. Bacula peut sauvegarder le client A et restaurer ses fichiers sur le client B, pourvu que leurs systèmes ne soient pas trop différents au niveau de leurs structures de fichiers. Par défaut, Bacula restaure les données sur leur client d'origine, mais pas à leur emplacement d'origine : dans le répertoire /tmp/bacula-restores. Vous pouvez modifier ces valeurs par défaut lorsque la commande restore vous demande confirmation d'exécution du job en choisissant l'option mod.
Ceci est réalisé par la commande restore de la Console. Vous sélectionnez d'abord le type de restauration souhaitée ce qui entraîne la sélection des JobIds requis et la construction d'une arborescence interne à Bacula contenant les enregistrements de fichiers des JobIds sélectionnés. A ce stade, le processus de restauration entre dans un mode où vous pouvez naviguer interactivement dans l'arborescence des fichiers disponibles pour restauration et sélectionner ceux que vous voulez restaurer. Ce mode est similaire au programme de sélection de fichier interactif standard d'Unix restore.
Si vos fichiers ont été élagués, la commande restore sera dans l'incapacité de les trouver. Voyez ci-dessous pour plus de détails sur ce cas de figure.
Dans la Console, après avoir saisi restore, le menu suivant vous est présenté :
First you select one or more JobIds that contain files to be restored. You will be presented several methods of specifying the JobIds. Then you will be allowed to select which files from those JobIds are to be restored. To select the JobIds, you have the following choices: 1: List last 20 Jobs run 2: List Jobs where a given File is saved 3: Enter list of comma separated JobIds to select 4: Enter SQL list command 5: Select the most recent backup for a client 6: Select backup for a client before a specified time 7: Enter a list of files to restore 8: Enter a list of files to restore before a specified time 9: Find the JobIds of the most recent backup for a client 10: Find the JobIds for a backup for a client before a specified time 11: Enter a list of directories to restore for found JobIds 12: Cancel Select item: (1-12):
Notez que ce processus de sélection automatique ne sélectionnera jamais un job qui a échoué (terminé avec un statut d'erreur). Si vous disposez d'un tel job dont vous voulez extraire des fichiers, vous devez eplicitement entrer son JobId au niveau du choix 3 et choisir les fichiers à restaurer.
Si certains de jobs requis pour la restauration ont eu leurs enregistrements de fichiers élagués, la restauration sera incomplète. Bacula ne détecte pas, pour l'instant, cette condition. Vous pouvez cependant la contrôler en examinant attentivement la liste des jobs sélectionnés et affichés par Bacula. Si vous trouvez des jobs dont le champ JobFiles est à zéro alors que ces fichiers auraient dû être sauvegardés, alors vous pouvez vous attendre à des problèmes.
Si tous les enregistrements de fichiers ont été élagués, Bacula constatera qu'il n'y a aucune référence à aucun fichier pour le JobIds sélectionnés et vous en informera, et vous proposera de faire une restauration complète (non sélective) de ces JobIds. Ceci est possible car Bacula sait encore où commencent les données sur les volumes, même s'il ne sait plus où sont les fichiers individuellement.
A titre d'exemple, supposons que nous sélectionnions l'option 5 (restaurer à l'état le plus récent). Bacula vous demande alors le client désiré ce qui, sur mon système, se manifeste ainsi :
Defined clients: 1: Rufus 2: Matou 3: Polymatou 4: Minimatou 5: Minou 6: MatouVerify 7: PmatouVerify 8: RufusVerify 9: Watchdog Select Client (File daemon) resource (1-9):
Si vous n'avez qu'un client, il est automatiquement sélectionné. Dans le cas présent, j'entre Rufus pour sélectionner ce client. Bacula a maintenant connaître le FileSet à restaurer, aussi il affiche :
The defined FileSet resources are: 1: Full Set 2: Kerns Files Select FileSet resource (1-2):
J'opte pour le choix 1, ma sauvegarde full. En principe, vous n'aurez qu'un FileSet pour chaque job, et si vos machines de ressemblent (mêmes systèmes), vous pouvez n'avoir qu'un seul FileSet pour tous vos clients.
A ce stade, Bacula détient toutes les informations dont il a besoin pour trouver le jeu de sauvegardes le plus récent. Il va maintenant interroger le cataloguie, ce qui peut prendre un peu de temps, et afficher quelque chose comme :
+-------+------+----------+-------------+-------------+------+-------+---------- --+ | JobId | Levl | JobFiles | StartTime | VolumeName | File | SesId | VolSesTime | +-------+------+----------+-------------+-------------+------+-------+---------- --+ | 1,792 | F | 128,374 | 08-03 01:58 | DLT-19Jul02 | 67 | 18 | 1028042998 | | 1,792 | F | 128,374 | 08-03 01:58 | DLT-04Aug02 | 0 | 18 | 1028042998 | | 1,797 | I | 254 | 08-04 13:53 | DLT-04Aug02 | 5 | 23 | 1028042998 | | 1,798 | I | 15 | 08-05 01:05 | DLT-04Aug02 | 6 | 24 | 1028042998 | +-------+------+----------+-------------+-------------+------+-------+---------- --+ You have selected the following JobId: 1792,1792,1797 Building directory tree for JobId 1792 ... Building directory tree for JobId 1797 ... Building directory tree for JobId 1798 ... cwd is: / $
(Certaines colonnes sont tromquées pour des nécessités de mise en page).
Selon le nombre de JobFiles pour chaque JobId, la construction de l'arborescence peut prendre un certain temps. Si vous constatez que tous les JobFiles sont à zéro, vos fichiers ont probalement été élagués et vous ne pourrez pas sélectionner les fichiers individuellement : vous devrez restaurer tout ou rien.
Dans notre exemple, Bacula a trouvé quatre jobs qui comprennent la sauvegarde la plus récente du client et du FileSet spécifiés. Deux des jobs ont le même JobId car le job a écrit sur deux volumes différents. Le troisième est une incrémentale qui n'a sauvegardé que 254 fichier sur les 128 374 de la full. Le quatrième est aussi une incrémentale, et n'a sauvegardé que 15 fichiers.
Maintenant Bacula insère ces jobs dans l'arborescence, sans en marquer aucun pour restauration par défaut. Il vous indique le nombre de fichiers dans l'arbre, et vous informe que le répertoire de travail courant (cwd) est /. Finalement, Bacula vous invite avec le signe ($) à saisir des commandes pour vous déplacer dans l'arborescence, et sélectionner des fichiers.
Si vous voulez que tous les fichiers de l'arbre soient marqués pour restauration à sa construction, tapez restore all.
Plutôt que de choisir l'option 5 du premier menu (sélectionner la sauvegarde la plus récente pour un client), si nous avions choisi l'option 3 (Entrer une liste de JobIds à sélectionner), et si nous avions saisi 1792,1797,1798, nous serions arrivés au même point.
Il faut noter un point si vous saisissez manuellement les JobIds : vous devez les entrer dans l'ordre où ils ont été exécutés (en général, l'ordre croissant. Si vous les sasissez dans un ordre différent, vous courrez le risque de ne pas version la plus récente d'un fichier sauvegardé plusieurs fois si celui-ci a été sauvegardé dans plusieurs jobs.
Entre vos JobIds directement peut aussi vous permettre de restaurer depuis un job qui a écrit des données sur les volumes mais qui s'est terminé en erreur.
Dans le mode sélection de fichiers, vous pouvez utiliser help ou une question (?) pour produire un résumé des commandes disponibles :
Command Description ======= =========== cd change current directory count count marked files in and below the cd dir long list current directory, wildcards allowed done leave file selection mode estimate estimate restore size exit same as done command find find files, wildcards allowed help print help ls list current directory, wildcards allowed lsmark list the marked files in and below the cd mark mark dir/file to be restored recursively in dirs markdir mark directory name to be restored (no files) pwd print current working directory unmark unmark dir/file to be restored recursively in dir unmarkdir unmark directory name only no recursion quit quit and do not do restore ? print help
Par défaut, aucun fichier n'est sélectionné pour restauration (sauf si vous avez ajouté all à la ligne de commande). Si, à ce stade, vous voulez tout restaurer, vous devriez saisir mark *, puis done, Bacula écrira alors les données bootstrap dans un fichier et sollicitera votre approbation pour démarrer la restauration.
Si vous n'utilisez pas mark *, vous commencez avec une sélection vide. Vous pouvez simplement regarder et marquer (mark) les fichiers et/ou répertoires qui vous intéressent. Il est aisé de commettre une erreur dans ces opérations, et la gestion des erreurs dans Bacula n'est pas parfaite, aussi contrôlez votre travail avec la commande ls ou dir pour voir quels fichiers ont été sélectionnés. Les fichiers sélectionnés sont précédés d'une astérisque.
Pour contrôler ce qui est marqué et ce qui ne l'est pas utilisez la commande count qui affiche :
128401 total files. 128401 marked to be restored.
Chacune des commandes ci-dessus sera expliqué plus en détail dans la prochaine section. Poursuivons avec notre exemple, en validant la restauration de tous les fichiers. En saisissant done, Bacula affiche :
Bootstrap records written to /home/kern/bacula/working/restore.bsr The restore job will require the following Volumes: DLT-19Jul02 DLT-04Aug02 128401 files selected to restore. Run Restore job JobName: kernsrestore Bootstrap: /home/kern/bacula/working/restore.bsr Where: /tmp/bacula-restores Replace: always FileSet: Kerns Files Client: Rufus Storage: SDT-10000 JobId: *None* OK to run? (yes/mod/no):
Examinez chaque élément attentivement pour vous assurer que tout est conforme à ce que vous souhaitez. En particulier, vérifiez la ligne where, qui vous indique dans quelle partie du système de fichiers vos données seront restaurées, et quel client va les recevoir (par défaut, les restaurations ont lieu sur le client d'origine). Ces paramètres n'auront pas forcément les bonnes valeurs, mais vous pouvez les modifier à l'aide de la commande mod et en vous laissant guider par l'invite de la console.
L'affichage ci-dessus suppose que vous ayez défini une ressource Job de type restore dans le fichier de configuration de votre Director. en principe, vous n'en n'aurez besoin que d'une, car, par nature, une restauration est une opération essentiellement manuelle. A l'aide de la Console, vous pourrez modifier le job Restore pour faire ce que vous voulez qu'il fasse.
Un exemple de ressource Job de type restore est donné plus bas.
Pour en revenir à notre exemple, en plus de vérifier le client, il est sage de vérifier que le périphérique de stockage choisi par Bacula est le bon. Bien que le FileSet soit présenté, il est en fait ignoré dans la restauration. Le processus de restauration choisit ses fichiers en lisant le fichier bootstrap, et restaure tous les fichiers associés au JobId considéré si ce fichier n'est pas spécifié.
Enfin, avant de lancer la restauration, notez que le lieu par défaut pour les fichiers restaurés n'est pas leur emplacement d'origine mais le répertoire /tmp/bacula-restores. Vous pouvez modifier cette valeur par défaut dans le fichier de configuration du Director, ou avec l'option mod. Si vous voulez restaurer les fichiers à leurs emplacements d'origine, modifiez l'option where : spécifiez la racine (/Ā ou rien du tout).
Si vous entrez maintenant yes, Bacula lance la restauration. le Storage Daemon va d'abord requérir le volume DLT-19Jul02, puis le DLT-04Aug02 une fois qu'il aura extrait les fichiers requis du premier.
Si vous n'avez qu'un petit nombre de fichiers à restaurer dont vous connaissez les noms, vous pouvez, aux choix, placer ces noms dans un fichier qui sera lu par Bacula, ou saisir les noms un par un. Les noms de fichier doivent inclure le chemin absolu. Les caractères jokers ne peuvent être utilisés.
Pour saisir la liste, choisissez l'option 7 dans le menu de la commande restore :
To select the JobIds, you have the following choices: 1: List last 20 Jobs run 2: List Jobs where a given File is saved 3: Enter list of comma separated JobIds to select 4: Enter SQL list command 5: Select the most recent backup for a client 6: Select backup for a client before a specified time 7: Enter a list of files to restore 8: Enter a list of files to restore before a specified time 9: Find the JobIds of the most recent backup for a client 10: Find the JobIds for a backup for a client before a specified time 11: Enter a list of directories to restore for found JobIds 12: Cancel Select item: (1-12):
Vous êtes alors invité à préciser le client :
Defined Clients: 1: Timmy 2: Tibs 3: Rufus Select the Client (1-3): 3
Si vous n'avez qu'un client, il est sélectionné automatiquement. Finalement, Bacula vous demande d'entrer un nom de fichier :
Enter filename:
Vous pouvez, à ce stade, saisir le chemin absolu et le nom du fichier :
Enter filename: /home/kern/bacula/k/Makefile.in Enter filename:
Si Bacula ne peut en trouver aucune copie, il affiche ce qui suit :
Enter filename: junk filename No database record found for: junk filename Enter filename:
Si vous souhaitez que Bacula récupère la liste des fichiers à restaurer depuis un fichier, rédigez ce fichier et donnez lui un nom commençant par le signe moins (<) et saisissez-le ici. Lorsque vous avez entré tous les noms de fichiers, validez une ligne vide. Bacula écrit maintenant le fichier bootstrap, vous indique les cartouches qui seront utilisées, et vous propose de valider la restauration :
Enter filename: Automatically selected Storage: DDS-4 Bootstrap records written to /home/kern/bacula/working/restore.bsr The restore job will require the following Volumes: test1 1 file selected to restore. Run Restore job JobName: kernsrestore Bootstrap: /home/kern/bacula/working/restore.bsr Where: /tmp/bacula-restores Replace: always FileSet: Kerns Files Client: Rufus Storage: DDS-4 When: 2003-09-11 10:20:53 Priority: 10 OK to run? (yes/mod/no):
Il est possible d'automatiser la sélection des fichiers en plaçant votre liste de fichiers dans, part exemple, /tmp/file-list, puis en utilisant la commande suivante :
restore client=Rufus file=</tmp/file-list
Si, en modifiant les paramètres du job restauration, vous constatez que Bacula vous demande d'entrer un numéro de job, c'est vous n'avez pour l'instant spécifié ni numéro de job, ni fichier bootstrap. Entrez simplement zéro pour pouvoir continuer et sélectionner une autre option à modifier.
Si tout ce qui précède vous a semblé compliqué, vous admettrez certainement que ce n'est vraiment pas le cas après quelques essais. Il est possible de faire tout ce qui vient d'être vu en utilisant la ligne de commande, à l'exception de la sélection du FileSet. Voici une telle ligne de commande :
restore client=Rufus select current all done yes
Le spécification client=Rufus sélectionne automatiquement le client Rufus, l'option current précise que vous voulez une restauration à l'état le plus récent possible, et le yes élude l'invite finale yes/mod/no et exécute directement la restauration.
Voici la liste des arguments de la ligne de commandes :
Selon la façon dont vous restaurez, vous pouvez ou non restaurer les attributs de fichiers à leur état initial. Voici quelques uns des problèmes auxquels vous pouvez être confrontés, et, pour les restaurations sur la machine d'origine, comment les éviter.
Pour éviter ce problème, vous devriez créer le répertoire alternatif avant de lancer la restauration. Bacula ne changera pas les attributs de ce répertoire, du moment que ce n'est pas l'un des répertoires à restaurer.
Le répertoire de restauration par défaut est /tmp/bacula-restores/, qui devient /tmp/bacula-restores/e/ si vous restaurez depuis le disque E. Aussi, assurez-vous que ce répertoire existe avant de lancer la restauration, ou utilisez l'option mod pour sélectionner un répertoire destination existant.
Certains utilisateurs ont signalé des problèmes en restaurant des fichiers qui participent à Active Directory. Ils ont aussi rapporté que le changement de l'Id utilisateur sous lequel est exécuté Bacula de SYSTEM en un Id d'administrateur du domaine résout le problème.
Restaurer des fichiers est généralement beaucoup plus lent que de les sauvegarder, ce pour plusieurs raisons. La première est que lors d'une sauvegarde, la cartouche est normalement déjà positionnée, Bacula n'a qu'à écrire dessus. D'autre part, les restaurations étant si rares (par rapport aux sauvegardes), Bacula ne garde dans le catalogue que l'emplacement sur la cartouche du premier fichier et du premier bloc pour chaque job, et non l'emplacement de chaque fichier, ce qui occuperait trop de place dans le catalogue.
Bacula se place d'abord sur la bonne marque de fichier sur la cartouche, puis sur le bloc correct, puis lit séquentiellement chaque enregistrement jusqu'à trouver ceux correspondant aux fichier que vous voulez restaurer. Une fois ces fichiers restaurés, Bacula cesse de lire la cartouche.
Enfin, au lieu de simplement lire un fichier comme pour une sauvegarde, Bacula doit, lors d'une restauration, créer les fichiers, tandis que le système d'exploitation doit, de son coté, allouer de l'espace disque pour ces fichiers restaurés.
Pour toutes ces raisons, le processus de restauration est généralement beaucoup plus lent que celui de sauvegarde (une restauration peut prendre trois fois plus de temps que la sauvegarde).
Les problèmes que les utilisateurs rencontrent le plus souvent lors des restaurations sont des messages d'erreurs tels que :
04-Jan 00:33 z217-sd: RestoreFiles.2005-01-04_00.31.04 Error: block.c:868 Volume data error at 20:0! Short block of 512 bytes on device /dev/tape discarded.
ou
04-Jan 00:33 z217-sd: RestoreFiles.2005-01-04_00.31.04 Error: block.c:264 Volume data error at 20:0! Wanted ID: "BB02", got ".". Buffer discarded.
Ces deux types de messages indiquent que vous avez probablement utilisé votre lecteur en mode "blocs de taille fixe" plutôt qu'en mode "blocs de taille variable". Le mode "blocs de taille fixe" fonctionne avec tout programme qui lit les cartouches séquentiellement tel que tar, cependant Bacula repositionne la bande suivant les blocs lors des restaurations, ce qui lui permet d'améliorer les performances en restauration de plusieurs ordres de grandeur lorsqu'il s'agit de restaurer quelques fichiers isolés. Il existe plusieurs moyens pour vous tirer de ce mauvais pas.
Tentez-les l'un après l'autre, en rétablissant votre ressource Device après chacun des tests :
Il existe de multiples causes qui peuvent être à l'origine de messages d'alerte ou d'erreurs lors des restaurations. Les plus courantes sont les suivantes :
Si en revanche la taille du fichier restaurée est plus petite, vous devriez pousser vos investigations du coté d'un éventuel problème de bande et contrôler les rapports de Bacula ainsi que votre système de journalisation.
Job { Name = "RestoreFiles" Type = Restore Client = Any-client FileSet = "Any-FileSet" Storage = Any-storage Where = /tmp/bacula-restores Messages = Standard Pool = Default }Si la directive Where n'est pas précisée, les fichiers sont restaurés à leur emplacement d'origine.
Une fois que vous avez sélectionné les jobs à restaurer et après que Bacula a créé l'arborescence des répertoires en mémoire, vous entrez dans le mode de sélection de fichiers, ce qui est rappelé par l'invite ($). Dans ce mode, vous pouvez utiliser les commandes énumérées plus haut. Vous pouvez naviguer dans l'arborescence avec la commande cd tout comme vous le feriez dans un système de fichiers. Lorsque vous êtes dans un répertoire, vous pouvez en sélectionner des fichiers ou répertoires pour restauration. Par défaut, aucun fichier n'est sélectionné. Si vous souhaitez au contraire partir d'une situation où tous les fichiers sont sélectionnés, saisissez simplement : cd / et mark *. Sinon, sélectionnez vos fichiers avec la commande mark. Les commandes disponibles sont :
Notez que sur les systèmes Windows, les différents disques (c:, d:, ...) sont traités comme des répertoires dans l'arborescence. Une conséquence est que vous devez saisir cd c: ou éventuellement cd C: pour descendre dans le premier répertoire.
Après exécution, la commande mark affiche un bref résumé :
No files marked.
si aucun fichier n'a été sélectionné, ou :
nn files marked.
si certains fichiers ont été sélectionnés.
Je présente ici quelques-uns des problèmes qui peuvent survenir et rendre les opérations de restaurations plus difficiles, et quelques idées pour se sortir de ces situations délicates. Des informations plus spécifiques aux restaurations complètes des clients ou du serveur sont fournies dans le chapitre Disaster recovery avec Bacula.
Dans le cas contraire, vous devrez restaurer le catalogue.
./drop_bacula_tables ./make_bacula_tablesAprès réinitialisation de la base, vous devriez pouvoir démarrer Bacula. Si vous tentez maintenat d'utiliser la commande run, ça ne marchera pas puisque le catalogue est vierge. Cependant, vous pouvez exécuter manuellement un job et spécifier votre fichier bootstrap. Pour cela, utilisez la commande run et choisissez un job de type retore. Si vous utilisez le bacula-dir.conf par défaut, il s'agit de RestoreFiles. Vous devriez obtenir quelque chose comme :
Run Restore job JobName: RestoreFiles Bootstrap: /home/kern/bacula/working/restore.bsr Where: /tmp/bacula-restores Replace: always FileSet: Full Set Client: rufus-fd Storage: File When: 2005-07-10 17:33:40 Catalog: MyCatalog Priority: 10 OK to run? (yes/mod/no):Plusieurs paramètres seront différents dans votre cas. Vous aller modifier (commande mod) le paramètre Bootstrap pour qu'il pointe vers votre fichier bootstrap, et vous assurer que les autres paramètres sont corrects. Notez que le FileSet est ignoré lorque vous utilisez un fichier bootstrap. Une fois que vous avez fixé tous les bons paramètres, exévutez le job, vous devriez récupérer la sauvegarde de votre catalogue. Il vous reste alors à régénérer votre base de données à partir du fichier de sauvegarde ASCII.
22-Apr 10:22 HeadMan: Start Backup JobId 7510, Job=CatalogBackup.2005-04-22_01.10.0 22-Apr 10:23 HeadMan: Bacula 1.37.14 (21Apr05): 22-Apr-2005 10:23:06 JobId: 7510 Job: CatalogBackup.2005-04-22_01.10.00 Backup Level: Full Client: Polymatou FileSet: "CatalogFile" 2003-04-10 01:24:01 Pool: "Default" Storage: "DLTDrive" Start time: 22-Apr-2005 10:21:00 End time: 22-Apr-2005 10:23:06 FD Files Written: 1 SD Files Written: 1 FD Bytes Written: 210,739,395 SD Bytes Written: 210,739,521 Rate: 1672.5 KB/s Software Compression: None Volume name(s): DLT-22Apr05 Volume Session Id: 11 Volume Session Time: 1114075126 Last Volume Bytes: 1,428,240,465 Non-fatal FD errors: 0 SD Errors: 0 FD termination status: OK SD termination status: OK Termination: Backup OKAvec ces informations, vous pouvez créer manuellement un fichier bootstrap et embrayer sur les instructions données plus haut pour restaurer le catalogue. Un fichier bootstrap reconstruit pour le job ci-dessus ressemblerait à ceci :
Volume="DLT-22Apr05" VolSessionId=11 VolSessionTime=1114075126 FileIndex=1-1
où les valeurs Volume name, Volume Session IdĀ , et Volume Session Time conformes à celles du rapport de job ont été introduites. Notez aussi le FileIndex de valeur 1, ce sera toujours le cas pourvu qu'un seul fichier ait été sauvegardé par le job.
L'inconvénient de ce fichier bootstrap par rapport à celui créé automatiquement lorsque vous le spécifiez est qu'il ne comporte aucune spécification File ou Block, aussi Bacula doit examiner toutes les données du volume pour trouver le fichier requis. Un fichier bootstrap complètement renseigné ressemblerait à ceci :
Volume="DLT-22Apr05" VolSessionId=11 VolSessionTime=1114075126 VolFile=118-118 VolBlock=0-4053 FileIndex=1-1
1 Job 0 Fileset ne restaure rien.
llist jobid=120 JobId: 120 Job: save.2005-12-05_18.27.33 Job.Name: save PurgedFiles: 0 Type: B Level: F Job.ClientId: 1 Client.Name: Rufus JobStatus: T SchedTime: 2005-12-05 18:27:32 StartTime: 2005-12-05 18:27:35 EndTime: 2005-12-05 18:27:37 JobTDate: 1133803657 VolSessionId: 1 VolSessionTime: 1133803624 JobFiles: 236 JobErrors: 0 JobMissingFiles: 0 Job.PoolId: 4 Pool.Name: Full Job.FileSetId: 1 FileSet.FileSet: BackupSetVous pouvez alors déterminer le(s) volume(s) avec :
sql select VolumeName from JobMedia,Media where JobId=1 and JobMedia.MediaId=Media.MediaId;Finalement, vous pouvez créer un fichier bootstrap iavec ces informations comme décrit plus haut.
A partir de la version 1.38.0, lorsque vous entrez un jobId après avoir fait le choix 3, Bacula v ous demande si vous voulez restaurer tous les fichiers du job, collecte pour vous les informations requises et écrit le fichier bootstrap.
./bls -j -V DLT-22Apr05 /dev/nst0pourrait produire ceci :
bls: butil.c:258 Using device: "/dev/nst0" for reading. 21-Jul 18:34 bls: Ready to read from volume "DLT-22Apr05" on device "DLTDrive" (/dev/nst0). Volume Record: File:blk=0:0 SessId=11 SessTime=1114075126 JobId=0 DataLen=164 ... Begin Job Session Record: File:blk=118:0 SessId=11 SessTime=1114075126 JobId=7510 Job=CatalogBackup.2005-04-22_01.10.0 Date=22-Apr-2005 10:21:00 Level=F Type=B End Job Session Record: File:blk=118:4053 SessId=11 SessTime=1114075126 JobId=7510 Date=22-Apr-2005 10:23:06 Level=F Type=B Files=1 Bytes=210,739,395 Errors=0 Status=T ... 21-Jul 18:34 bls: End of Volume at file 201 on device "DLTDrive" (/dev/nst0), Volume "DLT-22Apr05" 21-Jul 18:34 bls: End of all volumes.Bien sur, il y aura de nombreuses autres informations affichées, nous n'avons reproduit ici que les essentielles. D'après les informations sur le début (Begin job Session Record et sur la fin End Job Session Record du job, vous pouvez écrire un fichier bootstrap comme indiqué plus haut.
Pour rechercher le JobId d'une sauvegarde d'un fichier donné, choisissez l'option 2.
Vous pouvez aussi utiliser la commande query pour trouver l'information :
*query Available queries: 1: List Job totals: 2: List up to 20 places where a File is saved regardless of the directory: 3: List where the most recent copies of a file are saved: 4: List last 20 Full Backups for a Client: 5: List all backups for a Client after a specified time 6: List all backups for a Client 7: List Volume Attributes for a selected Volume: 8: List Volumes used by selected JobId: 9: List Volumes to Restore All Files: 10: List Pool Attributes for a selected Pool: 11: List total files/bytes by Job: 12: List total files/bytes by Volume: 13: List Files for a selected JobId: 14: List Jobs stored in a selected MediaId: 15: List Jobs stored for a given Volume name: Choose a query (1-15):
This document briefly describes the GUI programs that work with Bacula. The GUI programs that are currently available are:
For more information on bimagemgr, please see below.
For more information on gnome-console, please consult the Console Chapter of this manual.
For more information, please see Configuring the Monitor Program chapter this manual.
bimagemgr is a utility for those who backup to disk volumes in order to commit them to CDR disk, rather than tapes. It is a web based interface written in Perl and is used to monitor when a volume file needs to be burned to disk. It requires:
SQLite databases and DVD burning are not supported by bimagemgr at this time, but both are planned for future releases.
Please see the README file in the bimagemgr directory of the distribution for instructions.
Calling the program in your web browser, e.g. http://localhost/cgi-bin/bimagemgr.pl will produce a display as shown below in Figure 1. The program will query the bacula database and display all volume files with the date last written and the date last burned to disk. If a volume needs to be burned (last written is newer than last burn date) a "Burn" button will be displayed in the rightmost column.
Place a blank CDR disk in your recorder and click the "Burn" button. This will cause a pop up window as shown in Figure 2 to display the burn progress.
When the burn finishes the pop up window will display the results of cdrecord as shown in Figure 3. Close the pop up window and refresh the main window. The last burn date will be updated and the "Burn" button for that volume will disappear. Should you have a failed burn you can reset the last burn date of that volume by clicking its "Reset" link.
In the bottom row of the main display window are two more buttons labeled "Burn Catalog" and "Blank CDRW". "Burn Catalog" will place a copy of your bacula catalog on a disk. If you use CDRW disks rather than CDR then "Blank CDRW" allows you to erase the disk before re-burning it. Regularly committing your backup volume files and your catalog to disk with bimagemgr ensures that you can rebuild easily in the event of some disaster on the bacula server itself.
bimagemgr is a utility for those who backup to disk volumes in order to commit them to CDR disk, rather than tapes. It is a web based interface written in Perl and is used to monitor when a volume file needs to be burned to disk. It requires:
DVD burning is not supported by bimagemgr at this time, but both are planned for future releases.
Installation from tarball: 1. Examine the Makefile and adjust it to your configuration if needed. 2. Edit config.pm to fit your configuration. 3. Do 'make install' as root. 4. Edit httpd.conf and change the Timeout value. The web server must not time out and close the connection before the burn process is finished. The exact value needed may vary depending upon your cd recorder speed and whether you are burning on the bacula server on on another machine across your network. In my case I set it to 1000 seconds. Restart httpd. 5. Make sure that cdrecord is setuid root.
Installation from rpm package: 1. Install the rpm package for your platform. 2. Edit /cgi-bin/config.pm to fit your configuration. 3. Edit httpd.conf and change the Timeout value. The web server must not time out and close the connection before the burn process is finished. The exact value needed may vary depending upon your cd recorder speed and whether you are burning on the bacula server on on another machine across your network. In my case I set it to 1000 seconds. Restart httpd. 4. Make sure that cdrecord is setuid root.
For bacula systems less than 1.36: 1. Edit the configuration section of config.pm to fit your configuration. 2. Run /etc/bacula/create_cdimage_table.pl from a console on your bacula server (as root) to add the CDImage table to your bacula database.
Accessing the Volume files: The Volume files by default have permissions 640 and can only be read by root. The recommended approach to this is as follows (and only works if bimagemgr and apache are running on the same host as bacula.
For bacula-1.34 or 1.36 installed from tarball - 1. Create a new user group bacula and add the user apache to the group for Red Hat or Mandrake systems. For SuSE systems add the user wwwrun to the bacula group. 2. Change ownership of all of your Volume files to root.bacula 3. Edit the /etc/bacula/bacula startup script and set SD_USER=root and SD_GROUP=bacula. Restart bacula.
Note: step 3 should also be done in /etc/init.d/bacula-sd but released versions of this file prior to 1.36 do not support it. In that case it would be necessary after a reboot of the server to execute '/etc/bacula/bacula restart'.
For bacula-1.38 installed from tarball - 1. Your configure statement should include: --with-dir-user=bacula --with-dir-group=bacula --with-sd-user=bacula --with-sd-group=disk --with-fd-user=root --with-fd-group=bacula 2. Add the user apache to the bacula group for Red Hat or Mandrake systems. For SuSE systems add the user wwwrun to the bacula group. 3. Check/change ownership of all of your Volume files to root.bacula
For bacula-1.36 or bacula-1.38 installed from rpm - 1. Add the user apache to the group bacula for Red Hat or Mandrake systems. For SuSE systems add the user wwwrun to the bacula group. 2. Check/change ownership of all of your Volume files to root.bacula
bimagemgr installed from rpm > 1.38.9 will add the web server user to the bacula group in a post install script. Be sure to edit the configuration information in config.pm after installation of rpm package.
bimagemgr will now be able to read the Volume files but they are still not world readable.
If you are running bimagemgr on another host (not recommended) then you will need to change the permissions on all of your backup volume files to 644 in order to access them via nfs share or other means. This approach should only be taken if you are sure of the security of your environment as it exposes the backup Volume files to world read.
Calling the program in your web browser, e.g. http://localhost/cgi-bin/bimagemgr.pl will produce a display as shown below in Figure 1. The program will query the bacula database and display all volume files with the date last written and the date last burned to disk. If a volume needs to be burned (last written is newer than last burn date) a "Burn" button will be displayed in the rightmost column.
Place a blank CDR disk in your recorder and click the "Burn" button. This will cause a pop up window as shown in Figure 2 to display the burn progress.
When the burn finishes the pop up window will display the results of cdrecord as shown in Figure 3. Close the pop up window and refresh the main window. The last burn date will be updated and the "Burn" button for that volume will disappear. Should you have a failed burn you can reset the last burn date of that volume by clicking its "Reset" link.
In the bottom row of the main display window are two more buttons labeled "Burn Catalog" and "Blank CDRW". "Burn Catalog" will place a copy of your bacula catalog on a disk. If you use CDRW disks rather than CDR then "Blank CDRW" allows you to erase the disk before re-burning it. Regularly committing your backup volume files and your catalog to disk with bimagemgr ensures that you can rebuild easily in the event of some disaster on the bacula server itself.
Without proper setup and maintenance, your Catalog may continue to grow indefinitely as you run Jobs and backup Files. How fast the size of your Catalog grows depends on the number of Jobs you run and how many files they backup. By deleting records within the database, you can make space available for the new records that will be added during the next Job. By constantly deleting old expired records (dates older than the Retention period), your database size will remain constant.
If you started with the default configuration files, they already contain reasonable defaults for a small number of machines (less than 5), so if you fall into that case, catalog maintenance will not be urgent if you have a few hundred megabytes of disk space free. Whatever the case may be, some knowledge of retention periods will be useful.
Bacula uses three Retention periods: the File Retention period, the Job Retention period, and the Volume Retention period. Of these three, the File Retention period is by far the most important in determining how large your database will become.
The File Retention and the Job Retention are specified in each Client resource as is shown below. The Volume Retention period is specified in the Pool resource, and the details are given in the next chapter of this manual.
Since File records in the database account for probably 80 percent of the size of the database, you should carefully determine exactly what File Retention period you need. Once the File records have been removed from the database, you will no longer be able to restore individual files in a Job. However, with Bacula version 1.37 and later, as long as the Job record still exists, you will be able to restore all files in the job.
Retention periods are specified in seconds, but as a convenience, there are a number of modifiers that permit easy specification in terms of minutes, hours, days, weeks, months, quarters, or years on the record. See the Configuration chapter of this manual for additional details of modifier specification.
The default File retention period is 60 days.
As mentioned above, once the File records are removed from the database, you will no longer be able to restore individual files from the Job. However, as long as the Job record remains in the database, you will be able to restore all the files backuped for the Job (on version 1.37 and later). As a consequence, it is generally a good idea to retain the Job records much longer than the File records.
The retention period is specified in seconds, but as a convenience, there are a number of modifiers that permit easy specification in terms of minutes, hours, days, weeks, months, quarters, or years. See the Configuration chapter of this manual for additional details of modifier specification.
The default Job Retention period is 180 days.
If you turn this off by setting it to no, your Catalog will grow each time you run a Job.
Over time, as noted above, your database will tend to grow. I've noticed that even though Bacula regularly prunes files, MySQL does not effectively use the space, and instead continues growing. To avoid this, from time to time, you must compact your database. Normally, large commercial database such as Oracle have commands that will compact a database to reclaim wasted file space. MySQL has the OPTIMIZE TABLE command that you can use, and SQLite version 2.8.4 and greater has the VACUUM command. We leave it to you to explore the utility of the OPTIMIZE TABLE command in MySQL.
All database programs have some means of writing the database out in ASCII format and then reloading it. Doing so will re-create the database from scratch producing a compacted result, so below, we show you how you can do this for MySQL, PostgreSQL and SQLite.
For a MySQL database, you could write the Bacula database as an ASCII file (bacula.sql) then reload it by doing the following:
mysqldump -f --opt bacula > bacula.sql mysql bacula < bacula.sql rm -f bacula.sql
Depending on the size of your database, this will take more or less time and a fair amount of disk space. For example, if I cd to the location of the MySQL Bacula database (typically /opt/mysql/var or something similar) and enter:
du bacula
I get 620,644 which means there are that many blocks containing 1024 bytes each or approximately 635 MB of data. After doing the msqldump, I had a bacula.sql file that had 174,356 blocks, and after doing the mysql command to recreate the database, I ended up with a total of 210,464 blocks rather than the original 629,644. In other words, the compressed version of the database took approximately one third of the space of the database that had been in use for about a year.
As a consequence, I suggest you monitor the size of your database and from time to time (once every 6 months or year), compress it.
If you find that you are getting errors writing to your MySQL database, or Bacula hangs each time it tries to access the database, you should consider running MySQL's database check and repair routines. The program you need to run depends on the type of database indexing you are using. If you are using the default, you will probably want to use myisamchk. For more details on how to do this, please consult the MySQL document at: http://www.mysql.com/doc/en/Repair.html.
If the errors you are getting are simply SQL warnings, then you might try running dbcheck before (or possibly after) using the MySQL database repair program. It can clean up many of the orphaned record problems, and certain other inconsistencies in the Bacula database.
The same considerations apply that are indicated above for MySQL. That is, consult the PostgreSQL documents for how to repair the database, and also consider using Bacula's dbcheck program if the conditions are reasonable for using (see above).
There are a considerable number of ways each of the databases can be tuned to improve the performance. Going from an untuned database to one that is properly tuned can make a difference of a factor of 100 or more in the time to insert or search for records.
For each of the databases, you may get significant improvements by adding additional indexes. The comments in the Bacula make_xxx_tables give some indications as to what indexes may be appropriate.
For MySQL, what seems to be very important is to use the examine the my.cnf file. You may obtain significant performances by switching to the my-large.cnf or my-huge.cnf files that come with the MySQL source code.
For SQLite3, one significant factor in improving the performance is to ensure that there is a "PRAGMA synchronous = NORMAL;" statement. This reduces the number of times that the database flushes the in memory cache to disk. There are other settings for this PRAGMA that can give even further performance improvements at the risk of a database corruption if your system crashes.
For PostgreSQL, you might want to consider turning fsync off. Of course doing so can cause corrupted databases in the even of a machine crash. There are many different ways that you can tune PostgreSQL, the following document discusses a few of them: http://www.varlena.com/varlena/GeneralBits/Tidbits/perf.html.
There is also a PostgreSQL FAQ question number 3.3 that may answer some of your questions about how to improve performance of the PostgreSQL engine: http://www.postgresql.org/docs/faqs.FAQ.html#3.3.
Over time, as noted above, your database will tend to grow. I've noticed that even though Bacula regularly prunes files, PostgreSQL has a VACUUM command that will compact your database for you. Alternatively you may want to use the vacuumdb command, which can be run from a cron job.
All database programs have some means of writing the database out in ASCII format and then reloading it. Doing so will re-create the database from scratch producing a compacted result, so below, we show you how you can do this for PostgreSQL.
For a PostgreSQL database, you could write the Bacula database as an ASCII file (bacula.sql) then reload it by doing the following:
pg_dump bacula > bacula.sql cat bacula.sql | psql bacula rm -f bacula.sql
Depending on the size of your database, this will take more or less time and a fair amount of disk space. For example, you can cd to the location of the Bacula database (typically /usr/local/pgsql/data or possible /var/lib/pgsql/data) and check the size.
First please read the previous section that explains why it is necessary to compress a database. SQLite version 2.8.4 and greater have the Vacuum command for compacting the database.
cd {\bf working-directory} echo 'vacuum;' | sqlite bacula.db
As an alternative, you can use the following commands, adapted to your system:
cd {\bf working-directory} echo '.dump' | sqlite bacula.db > bacula.sql rm -f bacula.db sqlite bacula.db < bacula.sql rm -f bacula.sql
Where working-directory is the directory that you specified in the Director's configuration file. Note, in the case of SQLite, it is necessary to completely delete (rm) the old database before creating a new compressed version.
You may begin using Bacula with SQLite then later find that you want to switch to MySQL for any of a number of reasons: SQLite tends to use more disk than MySQL, SQLite apparently does not handle database sizes greater than 2GBytes, ... Several users have done so by first producing an ASCII "dump" of the SQLite database, then creating the MySQL tables with the create_mysql_tables script that comes with Bacula, and finally feeding the SQLite dump into MySQL using the -f command line option to continue past the errors that are generated by the DDL statements that SQLite's dump creates. Of course, you could edit the dump and remove the offending statements. Otherwise, MySQL accepts the SQL produced by SQLite.
If ever the machine on which your Bacula database crashes, and you need to restore from backup tapes, one of your first priorities will probably be to recover the database. Although Bacula will happily backup your catalog database if it is specified in the FileSet, this is not a very good way to do it, because the database will be saved while Bacula is modifying it. Thus the database may be in an instable state. Worse yet, you will backup the database before all the Bacula updates have been applied.
To resolve these problems, you need to backup the database after all the backup jobs have been run. In addition, you will want to make a copy while Bacula is not modifying it. To do so, you can use two scripts provided in the release make_catalog_backup and delete_catalog_backup. These files will be automatically generated along with all the other Bacula scripts. The first script will make an ASCII copy of your Bacula database into bacula.sql in the working directory you specified in your configuration, and the second will delete the bacula.sql file.
The basic sequence of events to make this work correctly is as follows:
Assuming that you start all your nightly backup jobs at 1:05 am (and that they run one after another), you can do the catalog backup with the following additional Director configuration statements:
# Backup the catalog database (after the nightly save) Job { Name = "BackupCatalog" Type = Backup Client=rufus-fd FileSet="Catalog" Schedule = "WeeklyCycleAfterBackup" Storage = DLTDrive Messages = Standard Pool = Default RunBeforeJob = "/home/kern/bacula/bin/make_catalog_backup" RunAfterJob = "/home/kern/bacula/bin/delete_catalog_backup" Write Bootstrap = "/home/kern/bacula/working/BackupCatalog.bsr" } # This schedule does the catalog. It starts after the WeeklyCycle Schedule { Name = "WeeklyCycleAfterBackup Run = Full sun-sat at 1:10 } # This is the backup of the catalog FileSet { Name = "Catalog" Include = signature=MD5 { @working_directory@/bacula.sql } }
Be sure to write a bootstrap file as in the above example. It is preferable to write or copy the bootstrap file to another computer. It will allow you to quickly recover the database backup should that be necessary. If you do not have a bootstrap file, it is still possible to recover your database backup, but it will be more work and take longer.
If you are running a database in production mode on your machine, Bacula will happily backup the files, but if the database is in use while Bacula is reading it, you may back it up in an unstable state.
The best solution is to shutdown your database before backing it up, or use some tool specific to your database to make a valid live copy perhaps by dumping the database in ASCII format. I am not a database expert, so I cannot provide you advice on how to do this, but if you are unsure about how to backup your database, you might try visiting the Backup Central site, which has been renamed Storage Mountain (www.backupcentral.com). In particular, their Free Backup and Recovery Software page has links to scripts that show you how to shutdown and backup most major databases.
As mentioned above, if you do not do automatic pruning, your Catalog will grow each time you run a Job. Normally, you should decide how long you want File records to be maintained in the Catalog and set the File Retention period to that time. Then you can either wait and see how big your Catalog gets or make a calculation assuming approximately 154 bytes for each File saved and knowing the number of Files that are saved during each backup and the number of Clients you backup.
For example, suppose you do a backup of two systems, each with 100,000 files. Suppose further that you do a Full backup weekly and an Incremental every day, and that the Incremental backup typically saves 4,000 files. The size of your database after a month can roughly be calculated as:
Size = 154 * No. Systems * (100,000 * 4 + 10,000 * 26)
where we have assumed 4 weeks in a month and 26 incremental backups per month. This would give the following:
Size = 154 * 2 * (100,000 * 4 + 10,000 * 26) or Size = 308 * (400,000 + 260,000) or Size = 203,280,000 bytes
So for the above two systems, we should expect to have a database size of approximately 200 Megabytes. Of course, this will vary according to how many files are actually backed up.
Below are some statistics for a MySQL database containing Job records for five Clients beginning September 2001 through May 2002 (8.5 months) and File records for the last 80 days. (Older File records have been pruned). For these systems, only the user files and system files that change are backed up. The core part of the system is assumed to be easily reloaded from the RedHat rpms.
In the list below, the files (corresponding to Bacula Tables) with the extension .MYD contain the data records whereas files with the extension .MYI contain indexes.
You will note that the File records (containing the file attributes) make up the large bulk of the number of records as well as the space used (459 Mega Bytes including the indexes). As a consequence, the most important Retention period will be the File Retention period. A quick calculation shows that for each File that is saved, the database grows by approximately 150 bytes.
Size in Bytes Records File ============ ========= =========== 168 5 Client.MYD 3,072 Client.MYI 344,394,684 3,080,191 File.MYD 115,280,896 File.MYI 2,590,316 106,902 Filename.MYD 3,026,944 Filename.MYI 184 4 FileSet.MYD 2,048 FileSet.MYI 49,062 1,326 JobMedia.MYD 30,720 JobMedia.MYI 141,752 1,378 Job.MYD 13,312 Job.MYI 1,004 11 Media.MYD 3,072 Media.MYI 1,299,512 22,233 Path.MYD 581,632 Path.MYI 36 1 Pool.MYD 3,072 Pool.MYI 5 1 Version.MYD 1,024 Version.MYI
This database has a total size of approximately 450 Megabytes.
If we were using SQLite, the determination of the total database size would be much easier since it is a single file, but we would have less insight to the size of the individual tables as we have in this case.
Note, SQLite databases may be as much as 50% larger than MySQL databases due to the fact that all data is stored as ASCII strings. That is even binary integers are stored as ASCII strings, and this seems to increase the space needed.
By default, once Bacula starts writing a Volume, it can append to the volume, but it will not overwrite the existing data thus destroying it. However when Bacula recycles a Volume, the Volume becomes available for being reused, and Bacula can at some later time over write the previous contents of that Volume. Thus all previous data will be lost. If the Volume is a tape, the tape will be rewritten from the beginning. If the Volume is a disk file, the file will be truncated before being rewritten.
You may not want Bacula to automatically recycle (reuse) tapes. This would require a large number of tapes though, and in such a case, it is possible to manually recycle tapes. For more on manual recycling, see the section entitled Manually Recycling Volumes below in this chapter.
Most people prefer to have a Pool of tapes that are used for daily backups and recycled once a week, another Pool of tapes that are used for Full backups once a week and recycled monthly, and finally a Pool of tapes that are used once a month and recycled after a year or two. With a scheme like this, the number of tapes in your pool or pools remains constant.
By properly defining your Volume Pools with appropriate Retention periods, Bacula can manage the recycling (such as defined above) automatically.
Automatic recycling of Volumes is controlled by three records in the Pool resource definition in the Director's configuration file. These three records are:
Automatic recycling of Volumes is performed by Bacula only when it wants a new Volume and no appendable Volumes are available in the Pool. It will then search the Pool for any Volumes with the Recycle flag set and whose Volume Status is Full. At that point, the recycling occurs in two steps. The first is that the Catalog for a Volume must be purged of all Jobs and Files contained on that Volume, and the second step is the actual recycling of the Volume. The Volume will be purged if the VolumeRetention period has expired. When a Volume is marked as Purged, it means that no Catalog records reference that Volume, and the Volume can be recycled. Until recycling actually occurs, the Volume data remains intact. If no Volumes can be found for recycling for any of the reasons stated above, Bacula will request operator intervention (i.e. it will ask you to label a new volume).
A key point mentioned above, that can be a source of frustration, is that Bacula will only recycle purged Volumes if there is no other appendable Volume available, otherwise, it will always write to an appendable Volume before recycling even if there are Volume marked as Purged. This preserves your data as long as possible. So, if you wish to "force" Bacula to use a purged Volume, you must first ensure that no other Volume in the Pool is marked Append. If necessary, you can manually set a volume to Full. The reason for this is that Bacula wants to preserve the data on your old tapes (even though purged from the catalog) as long as absolutely possible before overwriting it.
As Bacula writes files to tape, it keeps a list of files, jobs, and volumes in a database called the catalog. Among other things, the database helps Bacula to decide which files to back up in an incremental or differential backup, and helps you locate files on past backups when you want to restore something. However, the catalog will grow larger and larger as time goes on, and eventually it can become unacceptably large.
Bacula's process for removing entries from the catalog is called Pruning. The default is Automatic Pruning, which means that once an entry reaches a certain age (e.g. 30 days old) it is removed from the catalog. Once a job has been pruned, you can still restore it from the backup tape, but one additional step is required: scanning the volume with bscan. The alternative to Automatic Pruning is Manual Pruning, in which you explicitly tell Bacula to erase the catalog entries for a volume. You'd usually do this when you want to reuse a Bacula volume, because there's no point in keeping a list of files that USED TO BE on a tape. Or, if the catalog is starting to get too big, you could prune the oldest jobs to save space. Manual pruning is done with the prune command in the console. (thanks to Bryce Denney for the above explanation).
There are three pruning durations. All apply to catalog database records and not to the actual data in a Volume. The pruning (or retention) durations are for: Volumes (Media records), Jobs (Job records), and Files (File records). The durations inter-depend a bit because if Bacula prunes a Volume, it automatically removes all the Job records, and all the File records. Also when a Job record is pruned, all the File records for that Job are also pruned (deleted) from the catalog.
Having the File records in the database means that you can examine all the files backed up for a particular Job. They take the most space in the catalog (probably 90-95% of the total). When the File records are pruned, the Job records can remain, and you can still examine what Jobs ran, but not the details of the Files backed up. In addition, without the File records, you cannot use the Console restore command to restore the files.
When a Job record is pruned, the Volume (Media record) for that Job can still remain in the database, and if you do a "list volumes", you will see the volume information, but the Job records (and its File records) will no longer be available.
In each case, pruning removes information about where older files are, but it also prevents the catalog from growing to be too large. You choose the retention periods in function of how many files you are backing up and the time periods you want to keep those records online, and the size of the database. You can always re-insert the records (with 98% of the original data) by using "bscan" to scan in a whole Volume or any part of the volume that you want.
By setting AutoPrune to yes you will permit Bacula to automatically prune all Volumes in the Pool when a Job needs another Volume. Volume pruning means removing records from the catalog. It does not shrink the size of the Volume or affect the Volume data until the Volume gets overwritten. When a Job requests another volume and there are no Volumes with Volume Status Append available, Bacula will begin volume pruning. This means that all Jobs that are older than the VolumeRetention period will be pruned from every Volume that has Volume Status Full or Used and has Recycle set to yes. Pruning consists of deleting the corresponding Job, File, and JobMedia records from the catalog database. No change to the physical data on the Volume occurs during the pruning process. When all files are pruned from a Volume (i.e. no records in the catalog), the Volume will be marked as Purged implying that no Jobs remain on the volume. The Pool records that control the pruning are described below.
When this time period expires, and if AutoPrune is set to yes, and a new Volume is needed, but no appendable Volume is available, Bacula will prune (remove) Job records that are older than the specified Volume Retention period.
The Volume Retention period takes precedence over any Job Retention period you have specified in the Client resource. It should also be noted, that the Volume Retention period is obtained by reading the Catalog Database Media record rather than the Pool resource record. This means that if you change the VolumeRetention in the Pool resource record, you must ensure that the corresponding change is made in the catalog by using the update pool command. Doing so will insure that any new Volumes will be created with the changed Volume Retention period. Any existing Volumes will have their own copy of the Volume Retention period that can only be changed on a Volume by Volume basis using the update volume command.
When all file catalog entries are removed from the volume, its VolStatus is set to Purged. The files remain physically on the Volume until the volume is overwritten.
Retention periods are specified in seconds, minutes, hours, days, weeks, months, quarters, or years on the record. See the Configuration chapter of this manual for additional details of time specification.
The default is 1 year.
It is also possible to "force" pruning of all Volumes in the Pool associated with a Job by adding Prune Files = yes to the Job resource.
After all Volumes of a Pool have been pruned (as mentioned above, this happens when a Job needs a new Volume and no appendable Volumes are available), Bacula will look for the oldest Volume that is Purged (all Jobs and Files expired), and if the Recycle flag is on (Recycle=yes) for that Volume, Bacula will relabel it and write new data on it.
The full algorithm that Bacula uses when it needs a new Volume is:
The above occurs when Bacula has finished writing a Volume or when no Volume is present in the drive.
On the other hand, if you have inserted a different Volume after the last job, and Bacula recognizes the Volume as valid, it will request authorization from the Director to use this Volume. In this case, if you have set Recycle Current Volume = yes and the Volume is marked as Used or Full, Bacula will prune the volume and if all jobs were removed during the pruning (respecting the retention periods), the Volume will be recycled and used. The recycling algorithm in this case is:
This permits users to manually change the Volume every day and load tapes in an order different from what is in the catalog, and if the volume does not contain a current copy of your backup data, it will be used.
Each Volume inherits the Recycle status (yes or no) from the Pool resource record when the Media record is created (normally when the Volume is labeled). This Recycle status is stored in the Media record of the Catalog. Using the Console program, you may subsequently change the Recycle status for each Volume. For example in the following output from list volumes:
+----------+-------+--------+---------+------------+--------+-----+ | VolumeNa | Media | VolSta | VolByte | LastWritte | VolRet | Rec | +----------+-------+--------+---------+------------+--------+-----+ | File0001 | File | Full | 4190055 | 2002-05-25 | 14400 | 1 | | File0002 | File | Full | 1896460 | 2002-05-26 | 14400 | 1 | | File0003 | File | Full | 1896460 | 2002-05-26 | 14400 | 1 | | File0004 | File | Full | 1896460 | 2002-05-26 | 14400 | 1 | | File0005 | File | Full | 1896460 | 2002-05-26 | 14400 | 1 | | File0006 | File | Full | 1896460 | 2002-05-26 | 14400 | 1 | | File0007 | File | Purged | 1896466 | 2002-05-26 | 14400 | 1 | +----------+-------+--------+---------+------------+--------+-----+
all the volumes are marked as recyclable, and the last Volume, File0007 has been purged, so it may be immediately recycled. The other volumes are all marked recyclable and when their Volume Retention period (14400 seconds or 4 hours) expires, they will be eligible for pruning, and possibly recycling. Even though Volume File0007 has been purged, all the data on the Volume is still recoverable. A purged Volume simply means that there are no entries in the Catalog. Even if the Volume Status is changed to Recycle, the data on the Volume will be recoverable. The data is lost only when the Volume is re-labeled and re-written.
To modify Volume File0001 so that it cannot be recycled, you use the update volume pool=File command in the console program, or simply update and Bacula will prompt you for the information.
+----------+------+-------+---------+-------------+-------+-----+ | VolumeNa | Media| VolSta| VolByte | LastWritten | VolRet| Rec | +----------+------+-------+---------+-------------+-------+-----+ | File0001 | File | Full | 4190055 | 2002-05-25 | 14400 | 0 | | File0002 | File | Full | 1897236 | 2002-05-26 | 14400 | 1 | | File0003 | File | Full | 1896460 | 2002-05-26 | 14400 | 1 | | File0004 | File | Full | 1896460 | 2002-05-26 | 14400 | 1 | | File0005 | File | Full | 1896460 | 2002-05-26 | 14400 | 1 | | File0006 | File | Full | 1896460 | 2002-05-26 | 14400 | 1 | | File0007 | File | Purged| 1896466 | 2002-05-26 | 14400 | 1 | +----------+------+-------+---------+-------------+-------+-----+
In this case, File0001 will never be automatically recycled. The same effect can be achieved by setting the Volume Status to Read-Only.
Most people will want Bacula to fill a tape and when it is full, a new tape will be mounted, and so on. However, as an extreme example, it is possible for Bacula to write on a single tape, and every night to rewrite it. To get this to work, you must do two things: first, set the VolumeRetention to less than your save period (one day), and the second item is to make Bacula mark the tape as full after using it once. This is done using UseVolumeOnce = yes. If this latter record is not used and the tape is not full after the first time it is written, Bacula will simply append to the tape and eventually request another volume. Using the tape only once, forces the tape to be marked Full after each use, and the next time Bacula runs, it will recycle the tape.
An example Pool resource that does this is:
Pool { Name = DDS-4 Use Volume Once = yes Pool Type = Backup AutoPrune = yes VolumeRetention = 12h # expire after 12 hours Recycle = yes }
This example is meant to show you how one could define a fixed set of volumes that Bacula will rotate through on a regular schedule. There are an infinite number of such schemes, all of which have various advantages and disadvantages.
We start with the following assumptions:
We start the system by doing a Full save to one of the weekly volumes or one of the monthly volumes. The next morning, we remove the tape and insert a Daily tape. Friday evening, we remove the Daily tape and insert the next tape in the Weekly series. Monday, we remove the Weekly tape and re-insert the Daily tape. On the first Friday of the next month, we insert the next Monthly tape in the series rather than a Weekly tape, then continue. When a Daily tape finally fills up, Bacula will request the next one in the series, and the next day when you notice the email message, you will mount it and Bacula will finish the unfinished incremental backup.
What does this give? Well, at any point, you will have the last complete Full save plus several Incremental saves. For any given file you want to recover (or your whole system), you will have a copy of that file every day for at least the last 14 days. For older versions, you will have at least 3 and probably 4 Friday full saves of that file, and going back further, you will have a copy of that file made on the beginning of the month for at least a year.
So you have copies of any file (or your whole system) for at least a year, but as you go back in time, the time between copies increases from daily to weekly to monthly.
What would the Bacula configuration look like to implement such a scheme?
Schedule { Name = "NightlySave" Run = Level=Full Pool=Monthly 1st sat at 03:05 Run = Level=Full Pool=Weekly 2nd-5th sat at 03:05 Run = Level=Incremental Pool=Daily tue-fri at 03:05 } Job { Name = "NightlySave" Type = Backup Level = Full Client = LocalMachine FileSet = "File Set" Messages = Standard Storage = DDS-4 Pool = Daily Schedule = "NightlySave" } # Definition of file storage device Storage { Name = DDS-4 Address = localhost SDPort = 9103 Password = XXXXXXXXXXXXX Device = FileStorage Media Type = 8mm } FileSet { Name = "File Set" Include = signature=MD5 { fffffffffffffffff } Exclude = { *.o } } Pool { Name = Daily Pool Type = Backup AutoPrune = yes VolumeRetention = 10d # recycle in 10 days Maximum Volumes = 10 Recycle = yes } Pool { Name = Weekly Use Volume Once = yes Pool Type = Backup AutoPrune = yes VolumeRetention = 30d # recycle in 30 days (default) Recycle = yes } Pool { Name = Monthly Use Volume Once = yes Pool Type = Backup AutoPrune = yes VolumeRetention = 365d # recycle in 1 year Recycle = yes }
Perhaps the best way to understand the various resource records that come into play during automatic pruning and recycling is to run a Job that goes through the whole cycle. If you add the following resources to your Director's configuration file:
Schedule { Name = "30 minute cycle" Run = Level=Full Pool=File Messages=Standard Storage=File hourly at 0:05 Run = Level=Full Pool=File Messages=Standard Storage=File hourly at 0:35 } Job { Name = "Filetest" Type = Backup Level = Full Client=XXXXXXXXXX FileSet="Test Files" Messages = Standard Storage = File Pool = File Schedule = "30 minute cycle" } # Definition of file storage device Storage { Name = File Address = XXXXXXXXXXX SDPort = 9103 Password = XXXXXXXXXXXXX Device = FileStorage Media Type = File } FileSet { Name = "Test Files" Include = signature=MD5 { fffffffffffffffff } Exclude = { *.o } } Pool { Name = File Use Volume Once = yes Pool Type = Backup LabelFormat = "File" AutoPrune = yes VolumeRetention = 4h Maximum Volumes = 12 Recycle = yes }
Where you will need to replace the ffffffffff's by the appropriate files to be saved for your configuration. For the FileSet Include, choose a directory that has one or two megabytes maximum since there will probably be approximately 8 copies of the directory that Bacula will cycle through.
In addition, you will need to add the following to your Storage daemon's configuration file:
Device { Name = FileStorage Media Type = File Archive Device = /tmp LabelMedia = yes; Random Access = Yes; AutomaticMount = yes; RemovableMedia = no; AlwaysOpen = no; }
With the above resources, Bacula will start a Job every half hour that saves a copy of the directory you chose to /tmp/File0001 ... /tmp/File0012. After 4 hours, Bacula will start recycling the backup Volumes (/tmp/File0001 ...). You should see this happening in the output produced. Bacula will automatically create the Volumes (Files) the first time it uses them.
To turn it off, either delete all the resources you've added, or simply comment out the Schedule record in the Job resource.
Although automatic recycling of Volumes is implemented in version 1.20 and later (see the Automatic Recycling of Volumes chapter of this manual), you may want to manually force reuse (recycling) of a Volume.
Assuming that you want to keep the Volume name, but you simply want to write new data on the tape, the steps to take are:
Once the Volume is marked Purged, it will be recycled the next time a Volume is needed.
If you wish to reuse the tape by giving it a new name, follow the following steps:
Please note that the relabel command applies only to tape Volumes.
For Bacula versions prior to 1.30 or to manually relabel the Volume, use the instructions below:
mt -f /dev/nst0 rewind mt -f /dev/nst0 weof
where you replace /dev/nst0 with the appropriate device name on your system.
Please be aware that the delete command can be dangerous. Once it is done, to recover the File records, you must either restore your database as it was before the delete command, or use the bscan utility program to scan the tape and recreate the database entries.
This chapter presents most all the features needed to do Volume management. Most of the concepts apply equally well to both tape and disk Volumes. However, the chapter was originally written to explain backing up to disk, so you will see it is slanted in that direction, but all the directives presented here apply equally well whether your volume is disk or tape.
If you have a lot of hard disk storage or you absolutely must have your backups run within a small time window, you may want to direct Bacula to backup to disk Volumes rather than tape Volumes. This chapter is intended to give you some of the options that are available to you so that you can manage either disk or tape volumes.
Getting Bacula to write to disk rather than tape in the simplest case is rather easy. In the Storage daemon's configuration file, you simply define an Archive Device to be a directory. For example, if you want your disk backups to go into the directory /home/bacula/backups, you could use the following:
Device { Name = FileBackup Media Type = File Archive Device = /home/bacula/backups Random Access = Yes; AutomaticMount = yes; RemovableMedia = no; AlwaysOpen = no; }
Assuming you have the appropriate Storage resource in your Director's configuration file that references the above Device resource,
Storage { Name = FileStorage Address = ... Password = ... Device = FileBackup Media Type = File }
Bacula will then write the archive to the file /home/bacula/backups/<volume-name> where <volume-name> is the volume name of a Volume defined in the Pool. For example, if you have labeled a Volume named Vol001, Bacula will write to the file /home/bacula/backups/Vol001. Although you can later move the archive file to another directory, you should not rename it or it will become unreadable by Bacula. This is because each archive has the filename as part of the internal label, and the internal label must agree with the system filename before Bacula will use it.
Although this is quite simple, there are a number of problems. The first is that unless you specify otherwise, Bacula will always write to the same volume until you run out of disk space. This problem is addressed below.
In addition, if you want to use concurrent jobs that write to several different volumes at the same time, you will need to understand a number of other details. An example of such a configuration is given at the end of this chapter under Concurrent Disk Jobs.
Some of the options you have, all of which are specified in the Pool record, are:
UseVolumeOnce = yes.
Maximum Volume Jobs = nnn.
Maximum Volume Bytes = mmmm.
Volume Use Duration = ttt.
Note that although you probably would not want to limit the number of bytes on a tape as you would on a disk Volume, the other options can be very useful in limiting the time Bacula will use a particular Volume (be it tape or disk). For example, the above directives can allow you to ensure that you rotate through a set of daily Volumes if you wish.
As mentioned above, each of those directives is specified in the Pool or Pools that you use for your Volumes. In the case of Maximum Volume Job, Maximum Volume Bytes, and Volume Use Duration, you can actually specify the desired value on a Volume by Volume basis. The value specified in the Pool record becomes the default when labeling new Volumes. Once a Volume has been created, it gets its own copy of the Pool defaults, and subsequently changing the Pool will have no effect on existing Volumes. You can either manually change the Volume values, or refresh them from the Pool defaults using the update volume command in the Console. As an example of the use of one of the above, suppose your Pool resource contains:
Pool { Name = File Pool Type = Backup Volume Use Duration = 23h }
then if you run a backup once a day (every 24 hours), Bacula will use a new Volume for each backup, because each Volume it writes can only be used for 23 hours after the first write. Note, setting the use duration to 23 hours is not a very good solution for tapes unless you have someone on-site during the weekends, because Bacula will want a new Volume and no one will be present to mount it, so no weekend backups will be done until Monday morning.
Use of the above records brings up another problem -- that of labeling your Volumes. For automated disk backup, you can either manually label each of your Volumes, or you can have Bacula automatically label new Volumes when they are needed. While, the automatic Volume labeling in version 1.30 and prior is a bit simplistic, but it does allow for automation, the features added in version 1.31 permit automatic creation of a wide variety of labels including information from environment variables and special Bacula Counter variables. In version 1.37 and later, it is probably much better to use Python scripting and the NewVolume event since generating Volume labels in a Python script is much easier than trying to figure out Counter variables. See the Python Scripting chapter of this manual for more details.
Please note that automatic Volume labeling can also be used with tapes, but it is not nearly so practical since the tapes must be pre-mounted. This requires some user interaction. Automatic labeling from templates does NOT work with autochangers since Bacula will not access unknown slots. There are several methods of labeling all volumes in an autochanger magazine. For more information on this, please see the Autochanger chapter of this manual.
Automatic Volume labeling is enabled by making a change to both the Pool resource (Director) and to the Device resource (Storage daemon) shown above. In the case of the Pool resource, you must provide Bacula with a label format that it will use to create new names. In the simplest form, the label format is simply the Volume name, to which Bacula will append a four digit number. This number starts at 0001 and is incremented for each Volume the pool contains. Thus if you modify your Pool resource to be:
Pool { Name = File Pool Type = Backup Volume Use Duration = 23h LabelFormat = "Vol" }
Bacula will create Volume names Vol0001, Vol0002, and so on when new Volumes are needed. Much more complex and elaborate labels can be created using variable expansion defined in the Variable Expansion chapter of this manual.
The second change that is necessary to make automatic labeling work is to give the Storage daemon permission to automatically label Volumes. Do so by adding LabelMedia = yes to the Device resource as follows:
Device { Name = File Media Type = File Archive Device = /home/bacula/backups Random Access = Yes; AutomaticMount = yes; RemovableMedia = no; AlwaysOpen = no; LabelMedia = yes }
You can find more details of the Label Format Pool record in Label Format description of the Pool resource records.
Automatic labeling discussed above brings up the problem of Volume management. With the above scheme, a new Volume will be created every day. If you have not specified Retention periods, your Catalog will continue to fill keeping track of all the files Bacula has backed up, and this procedure will create one new archive file (Volume) every day.
The tools Bacula gives you to help automatically manage these problems are the following:
The first three records (File Retention, Job Retention, and AutoPrune) determine the amount of time that Job and File records will remain in your Catalog, and they are discussed in detail in the Automatic Volume Recycling chapter of this manual.
Volume Retention, AutoPrune, and Recycle determine how long Bacula will keep your Volumes before reusing them, and they are also discussed in detail in the Automatic Volume Recycling chapter of this manual.
The Maximum Volumes record can also be used in conjunction with the Volume Retention period to limit the total number of archive Volumes (files) that Bacula will create. By setting an appropriate Volume Retention period, a Volume will be purged just before it is needed and thus Bacula can cycle through a fixed set of Volumes. Cycling through a fixed set of Volumes can also be done by setting Recycle Oldest Volume = yes or Recycle Current Volume = yes. In this case, when Bacula needs a new Volume, it will prune the specified volume.
Now suppose you want to use multiple Pools, which means multiple Volumes, or suppose you want each client to have its own Volume and perhaps its own directory such as /home/bacula/client1 and /home/bacula/client2 ... With the single Storage and Device definition above, neither of these two is possible. Why? Because Bacula disk storage follows the same rules as tape devices. Only one Volume can be mounted on any Device at any time. If you want to simultaneously write multiple Volumes, you will need multiple Device resources in your bacula-sd.conf file, and thus multiple Storage resources in your bacula-dir.conf.
OK, so now you should understand that you need multiple Device definitions in the case of different directorys or different Pools, but you also need to know that the catalog data that Bacula keeps contains only the Media Type and not the specific storage device. This permits a tape for example to be re-read on any compatible tape drive. The compatibility being determined by the Media Type. The same applies to disk storage. Since a volume that is written by a Device in say directory /home/bacula/backups cannot be read by a Device with an Archive Device definition of /home/bacula/client1, you will not be able to restore all your files if you give both those devices Media Type = File. During the restore, Bacula will simply choose the first available device, which may not be the correct one. If this is confusing, just remember that the Directory has only the Media Type and the Volume name. It does not know the Archive Device (or the full path) that is specified in the Storage daemon. Thus you must explicitly tie your Volumes to the correct Device by using the Media Type.
The example shown below shows a case where there are two clients, each using its own Pool and storing their Volumes in different directories.
The following example is not very practical, but can be used to demonstrate the proof of concept in a relatively short period of time. The example consists of a two clients that are backed up to a set of 12 archive files (Volumes) for each client into different directories on the Storage maching. Each Volume is used (written) only once, and there are four Full saves done every hour (so the whole thing cycles around after three hours).
What is key here is that each physical device on the Storage daemon has a different Media Type. This allows the Director to choose the correct device for restores ...
The Director's configuration file is as follows:
Director { Name = my-dir QueryFile = "~/bacula/bin/query.sql" PidDirectory = "~/bacula/working" WorkingDirectory = "~/bacula/working" Password = dir_password } Schedule { Name = "FourPerHour" Run = Level=Full hourly at 0:05 Run = Level=Full hourly at 0:20 Run = Level=Full hourly at 0:35 Run = Level=Full hourly at 0:50 } Job { Name = "RecycleExample" Type = Backup Level = Full Client = Rufus FileSet= "Example FileSet" Messages = Standard Storage = FileStorage Pool = Recycle Schedule = FourPerHour } Job { Name = "RecycleExample2" Type = Backup Level = Full Client = Roxie FileSet= "Example FileSet" Messages = Standard Storage = FileStorage1 Pool = Recycle1 Schedule = FourPerHour } FileSet { Name = "Example FileSet" Include = compression=GZIP signature=SHA1 { /home/kern/bacula/bin } } Client { Name = Rufus Address = rufus Catalog = BackupDB Password = client_password } Client { Name = Roxie Address = roxie Catalog = BackupDB Password = client1_password } Storage { Name = FileStorage Address = rufus Password = local_storage_password Device = RecycleDir Media Type = File } Storage { Name = FileStorage1 Address = rufus Password = local_storage_password Device = RecycleDir1 Media Type = File1 } Catalog { Name = BackupDB dbname = bacula; user = bacula; password = "" } Messages { Name = Standard ... } Pool { Name = Recycle Use Volume Once = yes Pool Type = Backup LabelFormat = "Recycle-" AutoPrune = yes VolumeRetention = 2h Maximum Volumes = 12 Recycle = yes } Pool { Name = Recycle1 Use Volume Once = yes Pool Type = Backup LabelFormat = "Recycle1-" AutoPrune = yes VolumeRetention = 2h Maximum Volumes = 12 Recycle = yes }
and the Storage daemon's configuration file is:
Storage { Name = my-sd WorkingDirectory = "~/bacula/working" Pid Directory = "~/bacula/working" MaximumConcurrentJobs = 10 } Director { Name = my-dir Password = local_storage_password } Device { Name = RecycleDir Media Type = File Archive Device = /home/bacula/backups LabelMedia = yes; Random Access = Yes; AutomaticMount = yes; RemovableMedia = no; AlwaysOpen = no; } Device { Name = RecycleDir1 Media Type = File1 Archive Device = /home/bacula/backups1 LabelMedia = yes; Random Access = Yes; AutomaticMount = yes; RemovableMedia = no; AlwaysOpen = no; } Messages { Name = Standard director = my-dir = all }
With a little bit of work, you can change the above example into a weekly or monthly cycle (take care about the amount of archive disk space used).
Bacula can, of course, use multiple disks, but in general, each disk must be a separate Device specification in the Storage daemon's conf file, and you must then select what clients to backup to each disk. You will also want to give each Device specification a different Media Type so that during a restore, Bacula will be able to find the appropriate drive.
The situation is a bit more complicated if you want to treat two different physical disk drives (or partitions) logically as a single drive, which Bacula does not directly support. However, it is possible to back up your data to multiple disks as if they were a single drive by linking the Volumes from the first disk to the second disk.
For example, assume that you have two disks named /disk1 and /disk2. If you then create a standard Storage daemon Device resource for backing up to the first disk, it will look like the following:
Device { Name = client1 Media Type = File Archive Device = /disk1 LabelMedia = yes; Random Access = Yes; AutomaticMount = yes; RemovableMedia = no; AlwaysOpen = no; }
Since there is no way to get the above Device resource to reference both /disk1 and /disk2 we do it by pre-creating Volumes on /disk2 with the following:
ln -s /disk2/Disk2-vol001 /disk1/Disk2-vol001 ln -s /disk2/Disk2-vol002 /disk1/Disk2-vol002 ln -s /disk2/Disk2-vol003 /disk1/Disk2-vol003 ...
At this point, you can label the Volumes as Volume Disk2-vol001, Disk2-vol002, ... and Bacula will use them as if they were on /disk1 but actually write the data to /disk2. The only minor inconvenience with this method is that you must explicitly name the disks and cannot use automatic labeling unless you arrange to have the labels exactly match the links you have created.
An important thing to know is that Bacula treats disks like tape drives as much as it can. This means that you can only have a single Volume mounted at one time on a disk as defined in your Device resource in the Storage daemon's conf file. You can have multiple concurrent jobs running that all write to the one Volume that is being used, but if you want to have multiple concurrent jobs that are writting to separate disks drives (or partitions), you will need to define separate Device resources for each one, exactly as you would do for two different tape drives. There is one fundamental difference, however. The Volumes that you creat on the two drives cannot be easily exchanged as they can for a tape drive, because they are physically resident (already mounted in a sense) on the particular drive. As a consequence, you will probably want to give them different Media Types so that Bacula can distinguish what Device resource to use during a restore. An example would be the following:
Device { Name = Disk1 Media Type = File1 Archive Device = /disk1 LabelMedia = yes; Random Access = Yes; AutomaticMount = yes; RemovableMedia = no; AlwaysOpen = no; } Device { Name = Disk2 Media Type = File2 Archive Device = /disk2 LabelMedia = yes; Random Access = Yes; AutomaticMount = yes; RemovableMedia = no; AlwaysOpen = no; }
With the above device definitions, you can run two concurrent jobs each writing at the same time, one to /disk2 and the other to /disk2. The fact that you have given them different Media Types will allow Bacula to quickly choose the correct Storage resource in the Director when doing a restore.
If we take the above example and add a second Client, here are a few considerations:
In this example, we have two clients, each with a different Pool and a different number of archive files retained. They also write to different directories with different Volume labeling.
The Director's configuration file is as follows:
Director { Name = my-dir QueryFile = "~/bacula/bin/query.sql" PidDirectory = "~/bacula/working" WorkingDirectory = "~/bacula/working" Password = dir_password } # Basic weekly schedule Schedule { Name = "WeeklySchedule" Run = Level=Full fri at 1:30 Run = Level=Incremental sat-thu at 1:30 } FileSet { Name = "Example FileSet" Include = compression=GZIP signature=SHA1 { /home/kern/bacula/bin } } Job { Name = "Backup-client1" Type = Backup Level = Full Client = client1 FileSet= "Example FileSet" Messages = Standard Storage = File1 Pool = client1 Schedule = "WeeklySchedule" } Job { Name = "Backup-client2" Type = Backup Level = Full Client = client2 FileSet= "Example FileSet" Messages = Standard Storage = File2 Pool = client2 Schedule = "WeeklySchedule" } Client { Name = client1 Address = client1 Catalog = BackupDB Password = client1_password File Retention = 7d } Client { Name = client2 Address = client2 Catalog = BackupDB Password = client2_password } # Two Storage definitions with differen Media Types # permits different directories Storage { Name = File1 Address = rufus Password = local_storage_password Device = client1 Media Type = File1 } Storage { Name = File2 Address = rufus Password = local_storage_password Device = client2 Media Type = File2 } Catalog { Name = BackupDB dbname = bacula; user = bacula; password = "" } Messages { Name = Standard ... } # Two pools permits different cycling periods and Volume names # Cycle through 15 Volumes (two weeks) Pool { Name = client1 Use Volume Once = yes Pool Type = Backup LabelFormat = "Client1-" AutoPrune = yes VolumeRetention = 13d Maximum Volumes = 15 Recycle = yes } # Cycle through 8 Volumes (1 week) Pool { Name = client2 Use Volume Once = yes Pool Type = Backup LabelFormat = "Client2-" AutoPrune = yes VolumeRetention = 6d Maximum Volumes = 8 Recycle = yes }
and the Storage daemon's configuration file is:
Storage { Name = my-sd WorkingDirectory = "~/bacula/working" Pid Directory = "~/bacula/working" MaximumConcurrentJobs = 10 } Director { Name = my-dir Password = local_storage_password } # Archive directory for Client1 Device { Name = client1 Media Type = File1 Archive Device = /home/bacula/client1 LabelMedia = yes; Random Access = Yes; AutomaticMount = yes; RemovableMedia = no; AlwaysOpen = no; } # Archive directory for Client2 Device { Name = client2 Media Type = File2 Archive Device = /home/bacula/client2 LabelMedia = yes; Random Access = Yes; AutomaticMount = yes; RemovableMedia = no; AlwaysOpen = no; } Messages { Name = Standard director = my-dir = all }
Bacula allows you to specify that you want to write to DVD. However, this feature is implemented only in version 1.37 or later. You may in fact write to DVD+RW, DVD+R, DVD-R, or DVD-RW media. The actual process used by Bacula is to first write the image to a spool directory, then when the Volume reaches a certain size or, at your option, at the end of a Job, Bacula will transfer the image from the spool directory to the DVD. The actual work of transferring the image is done by a script dvd-handler, and the heart of that script is a program called growisofs which allows creating or adding to a DVD ISO filesystem.
You must have dvd+rw-tools loaded on your system for DVD writing to work. Please note that the original dvd+rw-tools package does NOT work with Bacula. You must apply a patch which can be found in the patches directory of Bacula sources with the name dvd+rw-tools-5.21.4.10.8.bacula.patch.
The fact that Bacula cannot use the OS to write directly to the DVD makes the whole process a bit more error prone than writing to a disk or a tape, but nevertheless, it does work if you use some care to set it up properly. However, at the current time (28 October 2005) we still consider this code to be experimental and of BETA quality. As a consequence, please do careful testing before relying on DVD backups in production.
The remainder of this chapter explains the various directives that you can use to control the DVD writing.
The following directives are added to the Storage daemon's Device resource.
Most frequently, you will define it as follows:
Mount Command = "/bin/mount -t iso9660 -o ro %a %m"
Most frequently, you will define it as follows:
Unmount Command = "/bin/umount %m"
For a DVD, you will most frequently specify the Bacula supplied dvd-handler script as follows:
Write Part Command = "/path/dvd-handler %a write %e %v"
Where /path is the path to your scripts install directory, and dvd-handler is the Bacula supplied script file. This command will already be present, but commented out, in the default bacula-sd.conf file. To use it, simply remove the comment (#) symbol.
For a DVD, you will most frequently specify the Bacula supplied dvd-handler script as follows:
Free Space Command = "/path/dvd-handler %a free"
Where /path is the path to your scripts install directory, and dvd-freespace is the Bacula supplied script file. If you want to specify your own command, please look at the code in dvd-handler to see what output Bacula expects from this command. This command will already be present, but commented out, in the default bacula-sd.conf file. To use it, simply remove the comment (#) symbol.
If you do not set it, Bacula will expect there is always free space on the device.
In addition to the directives specified above, you must also specify the other standard Device resource directives. Please see the sample DVD Device resource in the default bacula-sd.conf file. Be sure to specify the raw device name for Archive Device. It should be a name such as /dev/cdrom or /media/cdrecorder or /dev/dvd depending on your system. It will not be a name such as /mnt/cdrom.
The following directives are added to the Director's Job resource.
It should be set to yes when writing to devices that require a mount (for example DVD), so you are sure that the current part, containing this job's data, is written to the device, and that no data is left in the temporary file on the hard disk. However, on some media, like DVD+R and DVD-R, a lot of space (about 10Mb) is lost everytime a part is written. So, if you run several jobs each after another, you could set this directive to no for all jobs, except the last one, to avoid wasting too much space, but to ensure that the data is written to the medium when all jobs are finished.
This directive is ignored for devices other than DVDs.
To retrieve the current mode of a DVD-RW, run:
dvd+rw-mediainfo /dev/xxxwhere you replace xxx with your DVD device name.
Mounted Media line should give you the information.
To set the device to Restricted Overwrite mode, run:
dvd+rw-format /dev/xxxIf you want to set it back to the default Incremental Sequential mode, run:
dvd+rw-format -blank /dev/xxx
dd if=/dev/zero bs=1024 count=512 | growisofs -Z /dev/xxx=/dev/fd/0Then, try to mount the device, if it cannot be mounted, it will be considered as blank by Bacula, if it can be mounted, try a full blank (see below).
growisofs -Z /dev/xxx=/dev/zerowhere you replace xxx with your DVD device name. However, note that this blanks the whole DVD, which takes quite a long time (16 minutes on mine).
If you manage 5 or 10 machines and have a nice tape backup, you don't need Pools, and you may wonder what they are good for. In this chapter, you will see that Pools can help you optimize disk storage space. The same techniques can be applied to a shop that has multiple tape drives, or that wants to mount various different Volumes to meet their needs.
The rest of this chapter will give an example involving backup to disk Volumes, but most of the information applies equally well to tape Volumes.
A site that I administer (a charitable organization) had a tape DDS-3 tape drive that was failing. The exact reason for the failure is still unknown. Worse yet, their full backup size is about 15GB whereas the capacity of their broken DDS-3 was at best 8GB (rated 6/12). A new DDS-4 tape drive and the necessary cassettes was more expensive than their budget could handle.
They want to maintain 6 months of backup data, and be able to access the old files on a daily basis for a week, a weekly basis for a month, then monthly for 6 months. In addition, offsite capability was not needed (well perhaps it really is, but it was never used). Their daily changes amount to about 300MB on the average, or about 2GB per week.
As a consequence, the total volume of data they need to keep to meet their needs is about 100GB (15GB x 6 + 2GB x 5 + 0.3 x 7) = 102.1GB.
The chosen solution was to buy a 120GB hard disk for next to nothing -- far less than 1/10th the price of a tape drive and the cassettes to handle the same amount of data, and to have Bacula write to disk files.
The rest of this chapter will explain how to setup Bacula so that it would automatically manage a set of disk files with the minimum intervention on my part. The system has been running since 22 January 2004 until today (08 April 2004) with no intervention. Since we have not yet crossed the six month boundary, we still lack some data to be sure the system performs as desired.
Getting Bacula to write to disk rather than tape in the simplest case is rather easy, and is documented in the previous chapter. In addition, all the directives discussed here are explained in that chapter. We'll leave it to you to look at the details there. If you haven't read it and are not familiar with Pools, you probably should at least read it once quickly for the ideas before continuing here.
One needs to consider about what happens if we have only a single large Bacula Volume defined on our hard disk. Everything works fine until the Volume fills, then Bacula will ask you to mount a new Volume. This same problem applies to the use of tape Volumes if your tape fills. Being a hard disk and the only one you have, this will be a bit of a problem. It should be obvious that it is better to use a number of smaller Volumes and arrange for Bacula to automatically recycle them so that the disk storage space can be reused. The other problem with a single Volume, is that at the current time (1.34.0) Bacula does not seek within a disk Volume, so restoring a single file can take more time than one would expect.
As mentioned, the solution is to have multiple Volumes, or files on the disk. To do so, we need to limit the use and thus the size of a single Volume, by time, by number of jobs, or by size. Any of these would work, but we chose to limit the use of a single Volume by putting a single job in each Volume with the exception of Volumes containing Incremental backup where there will be 6 jobs (a week's worth of data) per volume. The details of this will be discussed shortly.
The next problem to resolve is recycling of Volumes. As you noted from above, the requirements are to be able to restore monthly for 6 months, weekly for a month, and daily for a week. So to simplify things, why not do a Full save once a month, a Differential save once a week, and Incremental saves daily. Now since each of these different kinds of saves needs to remain valid for differing periods, the simplest way to do this (and possibly the only) is to have a separate Pool for each backup type.
The decision was to use three Pools: one for Full saves, one for Differential saves, and one for Incremental saves, and each would have a different number of volumes and a different Retention period to accomplish the requirements.
Putting a single Full backup on each Volume, will require six Full save Volumes, and a retention period of six months. The Pool needed to do that is:
Pool { Name = Full-Pool Pool Type = Backup Recycle = yes AutoPrune = yes Volume Retention = 6 months Accept Any Volume = yes Maximum Volume Jobs = 1 Label Format = Full- Maximum Volumes = 6 }
Since these are disk Volumes, no space is lost by having separate Volumes for each backup (done once a month in this case). The items to note are the retention period of six months (i.e. they are recycled after 6 months), that there is one job per volume (Maximum Volume Jobs = 1), the volumes will be labeled Full-0001, ... Full-0006 automatically. One could have labeled these manual from the start, but why not use the features of Bacula.
For the Differential backup Pool, we choose a retention period of a bit longer than a month and ensure that there is at least one Volume for each of the maximum of five weeks in a month. So the following works:
Pool { Name = Diff-Pool Pool Type = Backup Recycle = yes AutoPrune = yes Volume Retention = 40 days Accept Any Volume = yes Maximum Volume Jobs = 1 Label Format = Diff- Maximum Volumes = 6 }
As you can see, the Differential Pool can grow to a maximum of six volumes, and the Volumes are retained 40 days and thereafter they can be recycled. Finally there is one job per volume. This, of course, could be tightened up a lot, but the expense here is a few GB which is not too serious.
Finally, here is the resource for the Incremental Pool:
Pool { Name = Inc-Pool Pool Type = Backup Recycle = yes AutoPrune = yes Volume Retention = 20 days Accept Any Volume = yes Maximum Volume Jobs = 6 Label Format = Inc- Maximum Volumes = 5 }
We keep the data for 20 days rather than just a week as the needs require. To reduce the proliferation of volume names, we keep a week's worth of data (6 incremental backups) in each Volume. In practice, the retention period should be set to just a bit more than a week and keep only two or three volumes instead of five. Again, the lost is very little and as the system reaches the full steady state, we can adjust these values so that the total disk usage doesn't exceed the disk capacity.
The following example shows you the actual files used, with only a few minor modifications to simplify things.
The Director's configuration file is as follows:
Director { # define myself Name = bacula-dir DIRport = 9101 QueryFile = "/home/bacula/bin/query.sql" WorkingDirectory = "/home/bacula/working" PidDirectory = "/home/bacula/working" Maximum Concurrent Jobs = 1 Password = " " Messages = Standard } # By default, this job will back up to disk in /tmp Job { Name = client Type = Backup Client = client-fd FileSet = "Full Set" Schedule = "WeeklyCycle" Storage = File Messages = Standard Pool = Default Full Backup Pool = Full-Pool Incremental Backup Pool = Inc-Pool Differential Backup Pool = Diff-Pool Write Bootstrap = "/home/bacula/working/client.bsr" Priority = 10 } # List of files to be backed up FileSet { Name = "Full Set" Include = signature=SHA1 compression=GZIP9 { / /usr /home } Exclude = { /proc /tmp /.journal /.fsck } } Schedule { Name = "WeeklyCycle" Run = Full 1st sun at 1:05 Run = Differential 2nd-5th sun at 1:05 Run = Incremental mon-sat at 1:05 } Client { Name = client-fd Address = client FDPort = 9102 Catalog = MyCatalog Password = " " AutoPrune = yes # Prune expired Jobs/Files Job Retention = 6 months File Retention = 60 days } Storage { Name = File Address = localhost SDPort = 9103 Password = " " Device = FileStorage Media Type = File } Catalog { Name = MyCatalog dbname = bacula; user = bacula; password = "" } Pool { Name = Full-Pool Pool Type = Backup Recycle = yes # automatically recycle Volumes AutoPrune = yes # Prune expired volumes Volume Retention = 6 months Accept Any Volume = yes # write on any volume in the pool Maximum Volume Jobs = 1 Label Format = Full- Maximum Volumes = 6 } Pool { Name = Inc-Pool Pool Type = Backup Recycle = yes # automatically recycle Volumes AutoPrune = yes # Prune expired volumes Volume Retention = 20 days Accept Any Volume = yes Maximum Volume Jobs = 6 Label Format = Inc- Maximum Volumes = 5 } Pool { Name = Diff-Pool Pool Type = Backup Recycle = yes AutoPrune = yes Volume Retention = 40 days Accept Any Volume = yes Maximum Volume Jobs = 1 Label Format = Diff- Maximum Volumes = 6 } Messages { Name = Standard mailcommand = "bsmtp -h mail.domain.com -f \"\(Bacula\) %r\" -s \"Bacula: %t %e of %c %l\" %r" operatorcommand = "bsmtp -h mail.domain.com -f \"\(Bacula\) %r\" -s \"Bacula: Intervention needed for %j\" %r" mail = root@domain.com = all, !skipped operator = root@domain.com = mount console = all, !skipped, !saved append = "/home/bacula/bin/log" = all, !skipped }
and the Storage daemon's configuration file is:
Storage { # definition of myself Name = bacula-sd SDPort = 9103 # Director's port WorkingDirectory = "/home/bacula/working" Pid Directory = "/home/bacula/working" } Director { Name = bacula-dir Password = " " } Device { Name = FileStorage Media Type = File Archive Device = /files/bacula LabelMedia = yes; # lets Bacula label unlabeled media Random Access = Yes; AutomaticMount = yes; # when device opened, read it RemovableMedia = no; AlwaysOpen = no; } Messages { Name = Standard director = bacula-dir = all }
Although Recycling and Backing Up to Disk Volume have been discussed in previous chapters, this chapter is meant to give you an overall view of possible backup strategies and to explain their advantages and disadvantages.
Probably the simplest strategy is to back everything up to a single tape and insert a new (or recycled) tape when it fills and Bacula requests a new one.
This system is very simple. When the tape fills and Bacula requests a new tape, you unmount the tape from the Console program, insert a new tape and label it. In most cases after the label, Bacula will automatically mount the tape and resume the backup. Otherwise, you simply mount the tape.
Using this strategy, one typically does a Full backup once a week followed by daily Incremental backups. To minimize the amount of data written to the tape, one can do (as I do) a Full backup once a month on the first Sunday of the month, a Differential backup on the 2nd-5th Sunday of the month, and incremental backups the rest of the week.
If you use the strategy presented above, Bacula will ask you to change the tape, and you will unmount it and then remount it when you have inserted the new tape.
If you do not wish to interact with Bacula to change each tape, there are several ways to get Bacula to release the tape:
#!/bin/sh /full-path/console -c /full-path/console.conf <<END_OF_DATA release storage=your-storage-name END_OF_DATA
In this example, you would have AlwaysOpen=yes, but the release command would tell Bacula to rewind the tape and on the next job assume the tape has changed. This strategy may not work on some systems, or on autochangers because Bacula will still keep the drive open.
#!/bin/sh /full-path/console -c /full-path/console.conf <\<END_OF_DATA unmount storage=your-storage-name END_OF_DATA # the following is a shell command mt eject /full-path/console -c /full-path/console.conf <<END_OF_DATA mount storage=your-storage-name END_OF_DATA
This scheme is quite different from the one mentioned above in that a Full backup is done to a different tape every day of the week. Generally, the backup will cycle continuously through 5 or 6 tapes each week. Variations are to use a different tape each Friday, and possibly at the beginning of the month. Thus if backups are done Monday through Friday only, you need only 5 tapes, and by having two Friday tapes, you need a total of 6 tapes. Many sites run this way, or using modifications of it based on two week cycles or longer.
The simplest way to "force" Bacula to use a different tape each day is to define a different Pool for each day of the the week a backup is done. In addition, you will need to specify appropriate Job and File retention periods so that Bacula will relabel and overwrite the tape each week rather than appending to it. Nic Bellamy has supplied an actual working model of this which we include here.
What is important is to create a different Pool for each day of the week, and on the run statement in the Schedule, to specify which Pool is to be used. He has one Schedule that accomplishes this, and a second Schedule that does the same thing for the Catalog backup run each day after the main backup (Priorities were not available when this script was written). In addition, he uses a Max Start Delay of 22 hours so that if the wrong tape is premounted by the operator, the job will be automatically canceled, and the backup cycle will re-synchronize the next day. He has named his Friday Pool WeeklyPool because in that Pool, he wishes to have several tapes to be able to restore to a time older than one week.
And finally, in his Storage daemon's Device resource, he has Automatic Mount = yes and Always Open = No. This is necessary for the tape ejection to work in his end_of_backup.sh script below.
For example, his bacula-dir.conf file looks like the following:
# /etc/bacula/bacula-dir.conf # # Bacula Director Configuration file # Director { Name = ServerName DIRport = 9101 QueryFile = "/etc/bacula/query.sql" WorkingDirectory = "/var/lib/bacula" PidDirectory = "/var/run" SubSysDirectory = "/var/lock/subsys" Maximum Concurrent Jobs = 1 Password = "console-pass" Messages = Standard } # # Define the main nightly save backup job # Job { Name = "NightlySave" Type = Backup Client = ServerName FileSet = "Full Set" Schedule = "WeeklyCycle" Storage = Tape Messages = Standard Pool = Default Write Bootstrap = "/var/lib/bacula/NightlySave.bsr" Max Start Delay = 22h } # Backup the catalog database (after the nightly save) Job { Name = "BackupCatalog" Type = Backup Client = ServerName FileSet = "Catalog" Schedule = "WeeklyCycleAfterBackup" Storage = Tape Messages = Standard Pool = Default # This creates an ASCII copy of the catalog RunBeforeJob = "/usr/lib/bacula/make_catalog_backup -u bacula" # This deletes the copy of the catalog, and ejects the tape RunAfterJob = "/etc/bacula/end_of_backup.sh" Write Bootstrap = "/var/lib/bacula/BackupCatalog.bsr" Max Start Delay = 22h } # Standard Restore template, changed by Console program Job { Name = "RestoreFiles" Type = Restore Client = ServerName FileSet = "Full Set" Storage = Tape Messages = Standard Pool = Default Where = /tmp/bacula-restores } # List of files to be backed up FileSet { Name = "Full Set" Include = signature=MD5 { / /data } Exclude = { /proc /tmp /.journal } } # # When to do the backups # Schedule { Name = "WeeklyCycle" Run = Level=Full Pool=MondayPool Monday at 8:00pm Run = Level=Full Pool=TuesdayPool Tuesday at 8:00pm Run = Level=Full Pool=WednesdayPool Wednesday at 8:00pm Run = Level=Full Pool=ThursdayPool Thursday at 8:00pm Run = Level=Full Pool=WeeklyPool Friday at 8:00pm } # This does the catalog. It starts after the WeeklyCycle Schedule { Name = "WeeklyCycleAfterBackup" Run = Level=Full Pool=MondayPool Monday at 8:15pm Run = Level=Full Pool=TuesdayPool Tuesday at 8:15pm Run = Level=Full Pool=WednesdayPool Wednesday at 8:15pm Run = Level=Full Pool=ThursdayPool Thursday at 8:15pm Run = Level=Full Pool=WeeklyPool Friday at 8:15pm } # This is the backup of the catalog FileSet { Name = "Catalog" Include = signature=MD5 { /var/lib/bacula/bacula.sql } } # Client (File Services) to backup Client { Name = ServerName Address = dionysus FDPort = 9102 Catalog = MyCatalog Password = "client-pass" File Retention = 30d Job Retention = 30d AutoPrune = yes } # Definition of file storage device Storage { Name = Tape Address = dionysus SDPort = 9103 Password = "storage-pass" Device = Tandberg Media Type = MLR1 } # Generic catalog service Catalog { Name = MyCatalog dbname = bacula; user = bacula; password = "" } # Reasonable message delivery -- send almost all to email address # and to the console Messages { Name = Standard mailcommand = "/usr/sbin/bsmtp -h localhost -f \"\(Bacula\) %r\" -s \"Bacula: %t %e of %c %l\" %r" operatorcommand = "/usr/sbin/bsmtp -h localhost -f \"\(Bacula\) %r\" -s \"Bacula: Intervention needed for %j\" %r" mail = root@localhost = all, !skipped operator = root@localhost = mount console = all, !skipped, !saved append = "/var/lib/bacula/log" = all, !skipped } # Pool definitions # # Default Pool for jobs, but will hold no actual volumes Pool { Name = Default Pool Type = Backup } Pool { Name = MondayPool Pool Type = Backup Recycle = yes AutoPrune = yes Volume Retention = 6d Accept Any Volume = yes Maximum Volume Jobs = 2 } Pool { Name = TuesdayPool Pool Type = Backup Recycle = yes AutoPrune = yes Volume Retention = 6d Accept Any Volume = yes Maximum Volume Jobs = 2 } Pool { Name = WednesdayPool Pool Type = Backup Recycle = yes AutoPrune = yes Volume Retention = 6d Accept Any Volume = yes Maximum Volume Jobs = 2 } Pool { Name = ThursdayPool Pool Type = Backup Recycle = yes AutoPrune = yes Volume Retention = 6d Accept Any Volume = yes Maximum Volume Jobs = 2 } Pool { Name = WeeklyPool Pool Type = Backup Recycle = yes AutoPrune = yes Volume Retention = 12d Accept Any Volume = yes Maximum Volume Jobs = 2 } # EOF
Note, the mailcommand and operatorcommand should be on a single line each. They were split to preserve the proper page width. In order to get Bacula to release the tape after the nightly backup, he uses a RunAfterJob script that deletes the ASCII copy of the database back and then rewinds and ejects the tape. The following is a copy of end_of_backup.sh
#! /bin/sh /usr/lib/bacula/delete_catalog_backup mt rewind mt eject exit 0
Finally, if you list his Volumes, you get something like the following:
*list media Using default Catalog name=MyCatalog DB=bacula Pool: WeeklyPool +-----+-----------+-------+--------+-----------+-----------------+-------+------+ | MeId| VolumeName| MedTyp| VolStat| VolBytes | LastWritten | VolRet| Recyc| +-----+-----------+-------+--------+-----------+-----------------+-------+------+ | 5 | Friday_1 | MLR1 | Used | 2157171998| 2003-07-11 20:20| 103680| 1 | | 6 | Friday_2 | MLR1 | Append | 0 | 0 | 103680| 1 | +-----+-----------+-------+--------+-----------+-----------------+-------+------+ Pool: MondayPool +-----+-----------+-------+--------+-----------+-----------------+-------+------+ | MeId| VolumeName| MedTyp| VolStat| VolBytes | LastWritten | VolRet| Recyc| +-----+-----------+-------+--------+-----------+-----------------+-------+------+ | 2 | Monday | MLR1 | Used | 2260942092| 2003-07-14 20:20| 518400| 1 | +-----+-----------+-------+--------+-----------+-----------------+-------+------+ Pool: TuesdayPool +-----+-----------+-------+--------+-----------+-----------------+-------+------+ | MeId| VolumeName| MedTyp| VolStat| VolBytes | LastWritten | VolRet| Recyc| +-----+-----------+-------+--------+-----------+-----------------+-------+------+ | 3 | Tuesday | MLR1 | Used | 2268180300| 2003-07-15 20:20| 518400| 1 | +-----+-----------+-------+--------+-----------+-----------------+-------+------+ Pool: WednesdayPool +-----+-----------+-------+--------+-----------+-----------------+-------+------+ | MeId| VolumeName| MedTyp| VolStat| VolBytes | LastWritten | VolRet| Recyc| +-----+-----------+-------+--------+-----------+-----------------+-------+------+ | 4 | Wednesday | MLR1 | Used | 2138871127| 2003-07-09 20:2 | 518400| 1 | +-----+-----------+-------+--------+-----------+-----------------+-------+------+ Pool: ThursdayPool +-----+-----------+-------+--------+-----------+-----------------+-------+------+ | MeId| VolumeName| MedTyp| VolStat| VolBytes | LastWritten | VolRet| Recyc| +-----+-----------+-------+--------+-----------+-----------------+-------+------+ | 1 | Thursday | MLR1 | Used | 2146276461| 2003-07-10 20:50| 518400| 1 | +-----+-----------+-------+--------+-----------+-----------------+-------+------+ Pool: Default No results to list.
Note, I have truncated a number of the columns so that the information fits on the width of a page.
Bacula supporte les librairies pour les opérations de lecture et écriture. Plusieurs conditions sont requises pour que Bacula puisse utiliser une librairie. Celles-ci sont expliquées en détail ci-dessous. Mais voyons d'abord la liste de ces conditions :
Dans les versions ultérieures à 1.37, la nouvelle directive Autochanger resource permet de grouper les ressources Device pour créer des librairies avec plusieurs lecteurs. Si vous avez une librairie, vous devez utiliser cette ressource.
Bacula utilise son propre script mtx-changer pour interagir avec un programme qui effectue réellement les changement de cartouches. Ainsi, mtx-changer peut être adapté pour fonctionner avec n'importe quel programme de prise en chgarge de librairie. La version actuelle de mtx-changer fonctionne avec le programme mtx . Cependant, des utilisateurs de FreeBSD ont réalisé un script, disponible dans le répertoire examples/autochangers, qui permet à Bacula de fonctionner avec le programme chio.
Bacula supporte aussi les librairies équipées de lecteurs de codes barres. Ce support inclut deux commandes de la console Bacula : label barcodes et update slots. Pour plus de détails au sujet de ces commandes, voyez la section "Support des lecteurs de codes barres" plus loin.
Le support des librairies dans Bacula n'inclue pas, pour le moment, la gestion du nettoyage des lecteurs, ni celle des bacs de cartouches ou des silos.
Le support des librairies à un ou plusieurs lecteurs requiert la ressource Autochanger resource.
En principe, si mtx fonctionne correctement avec votre librairie, ce n'est qu'une question d'adaptation du script mtx-changer pour que Bacula s'interface correctement avec la librairie. Vous pouvez trouver une liste des librairies supportées par mtx en suivant le lien suivant : http://mtx.opensource-sw.net/compatibility.php. Le site officiel du projet mtx se trouve ici : http://mtx.opensource-sw.net/.
Si vous avez des difficultés, veuillez utiliser la commande auto du programme btape pour tester le fonctionnement de votre librairie avec Bacula. Lorsque Bacula fonctionne, souvenez vous que pour beaucoup de distributions (par exemple FreeBSD, Debian,...), le Storage Daemon est exécuté en tant que bacula.tape plutôt que root.root, aussi vous devrez vous assurer que le Storage Daemon dispose de droits suffisants pour accéder à la librairie.
Sous Linux, vous pouvez lire le fichier /proc/scsi/scsi :
cat /proc/scsi/scsi
pour connaître vos périphériques SCSI. Vous pouvez aussi examiner les fichiers /proc/scsi/sg/device_hdr et /proc/scsi/sg/devices :
footnotesize
cat /proc/scsi/sg/device_hdr /proc/scsi/sg/devices
pour déterminer comment spécifier leur nom de périphérique (/dev/sg0 pour le premier, /dev/sg1 pour le second, ...) au niveau de la directive Changer Device
Pour des informations plus détaillées sur le sujet, veuillez consulter la section Linux SCSI Tricks du chapitre sur les tests de lecteurs de ce manuel.
Sous FreeBSD, vous disposez de la commande :
camcontrol devlist
pour afficher la liste des périphériques SCSI ainsi que le /dev/passn que vous utiliserez pour renseigner la directive Changer Device
Assurez-vous que votre Storage Daemon dispose bien des privilèges requis pour accéder à ce périphérique.
L'astuce suivante, destinée aux utilisateurs de FreeBSD, provient de Danny Butroyd. Au redémarrage, Bacula n'aura PLUS les permissions requises pour contrôler le périphérique /dev/pass0. Pour vous affanchir de cette difficulté, éditez le fichier /etc/devfs.conf et ajoutez lui ceci :
own pass0 root:bacula perm pass0 0666 own nsa0.0 root:bacula perm nsa0.0 0666
Nous avons ainsi donné au groupe Bacula la permission d'écrire sur le périphérique nsa0.0. Pour activer ces modifications, exécutez : /etc/rc.d/devfs restart
Vous n'aurez plus à modifier les permissions sur ces périphériques pour que Bacula continue d'utiliser la librairie après un redémarrage.
Veuillez lire les sections ci-dessous pour bien comprendre comment les librairies fonctionnent avec Bacula. Bien que nous fournissions un script mtx-changer par défaut, il se peut que votre librairie nécessite quelques aménagements de ce script. Si vous voulez voir des exemples de fichiers de configuration et de scripts, jetez un oeil au répertoire <bacula-src>/examples/devices où vous trouverez un exemple de ressource Device Bacula : HP-autoloader.conf ainsi que plusieurs scripts mtx-changer modifiés pour fonctionner avec diverses librairies.
Pour utiliser convenablement une librairie, Bacula doit savoir quel volume se trouve dans quel slot de la librairie. Les slots sont les emplacements où sont rangées les cartouches lorsqu'elles ne sont pas dans un lecteur. Bacula numérote ces slots de un jusqu'au nombre de cartouches contenues dans la librairie.
Bacula n'utilisera pas automatiquement une cartouche présente dans la librairie si elle ne porte pas d'étiquette (label) Bacula et si son numéro de slot n'est pas référencé dans le catalogue. Vous devez, à l'aide de la console, assigner un slot à chaque cartouche présente dans la librairie. Cette information est conservée dans le catalogue avec les autres données relatives au volume. Si le slot n'est pas précisé, ou s'il est égal à zéro, alors Bacula ne tentera pas d'utiliser la librairie, même si tous les enregistrements de configuration sont présents. De même, la commande mount de la console Bacula ne provoque pas non plus l'utilisation de la librairie, mais se contente d'ordonner à Bacula de lire toute cartouche éventuellement présente dans le lecteur.
Vous pouvez contrôler le numéro de slot et le drapeau InChanger avec la commande :
list Volumes
dans la console.
Certaines librairies comportent plusieurs périphériques de lecture/éctriture (lecteurs). La nouvelle ressource Autochanger apparue avec la version 1.37 vous permet de grouper des ressources Devices (représentant chacune un lecteur). Le Director est toujours en mesure d'adresser directement un lecteur, mais ce faisant, il outrepasse le fonctionnement propre aux groupements de lecteurs. Il est préférable que la Ressource Storage du Director définisse une ressource Autochanger, permettant ainsi au Storage Daemon de s'assurer qu'un seul lecteur à la fois utilise le script mtx-changer, et que deux lecteurs ne tentent pas de lire le même volume.
Les librairies à lecteurs multiples nécessitent d'utiliser la directive Drive Index dans la ressource Device du Storage Daemon. Les lecteurs sont numérotés à partir de zéro, ce qui constitue la valeur par défaut. Pour utiliser un deuxième lecteur dans une librairie, vous devez définir une seconde ressource Device et lui attribuer le Drive Index 1. En général, le second périphérique aura le même Changer Device (canal de contrôle) que le premier, mais une Archive Device différente.
Par défaut, les jobs Bacula préfèrent écrire sur un volume déjà monté. Si vous avez une librairie avec plusieurs lecteurs, et si vous souhaitez que Bacula écrive sur plusieurs volumes du même pool en même temps, vous devez désactiver la directive Prefer Mounted Volumes dans la ressource Job du Director. Ainsi le Storage Daemon pourra maximiser l'usage des lecteurs.
La configuration des librairies s'effectue dans Bacula au niveau de le ressource Device du Storage Daemon. Quatre directives permettent de définir l'usage de la librairie par Bacula : Autochanger, Changer Device, Changer Command et Maximum Changer Wait
Ces quatre directives sont décrites en détail ci-dessous. Notez cependant que les directives Changer Device et Changer Command ne sont pas requises dans la ressource Device si elles figurent dans la ressource Autochanger.
Sur FreeBSD, on spécifiera typiquement Changer Device = /dev/pass0 ou Changer Device = /dev/passn.
Sur Solaris, ce sera Changer Device = /dev/rdsk.
Assurez vous que votre Storage Daemon possède les permissions d'accéder à ce périphérique.
%% = % %a = archive device name %c = changer device name %d = changer drive index base 0 %f = Client's name %j = Job name %o = command (loaded, load, or unload) %s = Slot base 0 %S = Slot base 1 %v = Volume name
Voici un exemple d'utilisation de mtx avec le script mtx-changer :
Changer Command = "/etc/bacula/mtx-changer %c %o %S %a %d"
Où vous devrez adapter le chemin /etc/bacula pour qu'il co'incide à la réalité de votre installation. Les détails des trois commandes (loaded, load, unload) utilisées par Bacula ainsi que la sortie qui en est attendue sont donnés dans la section Interface entre Bacula et les librairies ci-dessous.
Au delà de ce délai, le programme de chargement est tué et Bacula sollicite l'intervention d'un opérateur.
Device Index = 1
Pour utiliser le second lecteur, vous devez avoir une seconde définition de ressource Device dans le fichier bacula-sd.conf. Voyez la section concernant les lecteurs multiples plus haut dans ce chapitre pour plus de plus amples informations.
De plus, pour un fonctionnement correct de la librairie, vous devez définir une ressource Autochanger.
La ressource Autochanger supporte les librairies à un ou plusieurs lecteurs en regroupant une ou plusieurs ressources Device en une unité nommée Autochanger dans Bacula (souvent désignée en tant que librairie de bandes par les constructeurs). Si vous possédez une librairie, et si vous voulez qu'elle fonctionne correctement, vous devez avoir une ressource Autochanger dans le fichier de configuration de votre Storage Daemon, et les directives Storage de votre Director doivent se référer au nom de la ressource Autochanger si elles sont supposées utiliser la librairie. Dans les versions antérieures à 1.38.0, les directives Storage du Director se référaient directement aux ressources Device qui étaient des librairies. Désormais, ce type de référence directe ne fonctionne plus avec les librairies.
Voici un exemple de définition de ressource Autochanger valide :
Autochanger { Name = "DDS-4-changer" Device = DDS-4-1, DDS-4-2, DDS-4-3 Changer Device = /dev/sg0 Changer Command = "/etc/bacula/mtx-changer %c %o %S %a %d" } Device { Name = "DDS-4-1" Drive Index = 0 Autochanger = yes ... } Device { Name = "DDS-4-2" Drive Index = 1 Autochanger = yes ... Device { Name = "DDS-4-3" Drive Index = 2 Autochanger = yes Autoselect = no ... }
Notez l'importance de la directive Autochanger = yes dans chaque définition de périphérique appartenant à une librairie. Un périphérique ne devrait pas être défini comme appartenant à plusieurs librairies. Aussi, votre directive Device dans la ressource Storage du Director devrait comporter le nom de la ressource Autochanger plutôt que le nom de l'un des lecteurs.
Si vous avez un lecteur qui appartient physiquement à une librairie mais que vous ne souhaitez pas que Bacula puisse l'utiliser automatiquement (par exemple, si vous voulez le réserver pour les restaurations) vous pouvez utiliser la directive :
Autoselect = no
à la ressource Device de ce lecteur. Dans ce cas, Bacula ne le sélectionnera pas automatiquement en accédant à la librairie. Vous pouvez encore utiliser le lecteur en le désignant par son nom de ressource device plutôt que par celui de la ressource Autochanger. Un exemple d'une telle définition est montré ci-dessus pour le lecteur DDS-4-3, qui ne sera pas sélectionné si le nom DDS-4-changer est utilisé dans une ressource Storage, mais le sera si DDS-4-3 est utilisé.
Les deux ressource suivantes implémentent une librairie :
Autochanger { Name = "Autochanger" Device = DDS-4 Changer Device = /dev/sg0 Changer Command = "/etc/bacula/mtx-changer %c %o %S %a %d" } Device { Name = DDS-4 Media Type = DDS-4 Archive Device = /dev/nst0 # Normal archive device Autochanger = yes LabelMedia = no; AutomaticMount = yes; AlwaysOpen = yes; }
où vous adapterez les directives Archive Device, Changer Device et Changer Command pour qu'elles conviennent à votre système.
Les ressources suivantes implémentent une librairie multi-lecteurs :
Autochanger { Name = "Autochanger" Device = Drive-1, Drive-2 Changer Device = /dev/sg0 Changer Command = "/etc/bacula/mtx-changer %c %o %S %a %d" } Device { Name = Drive-1 Drive Index = 0 Media Type = DDS-4 Archive Device = /dev/nst0 # Normal archive device Autochanger = yes LabelMedia = no; AutomaticMount = yes; AlwaysOpen = yes; } Device { Name = Drive-2 Drive Index = 1 Media Type = DDS-4 Archive Device = /dev/nst1 # Normal archive device Autochanger = yes LabelMedia = no; AutomaticMount = yes; AlwaysOpen = yes; }
où vous adapterez les directives Archive Device, Changer Device et Changer Command pour qu'elles conviennent à votre système.
Si vous utilisez la directive Autochanger = yes à la ressource Storage du fichier de configuration de votre Director, la console Bacula vous demandera automatiquement le numéro de slot lors de l'utilisation des commandes add ou label pour ce périphérique de stockage. Si votre script mtx-changer est correctement installé, Bacula chargera la bonne cartouche à l'exécution de la commande label.
Vous devez aussi spécifier Autochanger = yes dans la ressource Device du Storage Daemon ainsi que nous l'avons décrit plus haut pour que la librairie soit utilisée. Veuillez consulter la section Ressource Storage dans le chapitre sur la configuration du Director pour plus de détails sur ce sujet.
Ainsi, toutes les phases de l'utilisation des cartouches peuvent être intégralement automatisées. Il est aussi possible de paramétrer ou modifier la valeur du slot en utilisant le sous-menu Volume Parameters de la commande update de la console.
Même si tous les paramètres ci-dessus sont correctement spécifiés, Bacula ne tentera d'accéder à la librairie que s'il existe un slot non-nul parmi les volumes enregistrés dans le catalogue.
Si votre librairie est équipée d'un lecteur de codes barres, vous pouvez étiqueter vos volumes l'un après l'autre en utilisant la commande label barcodes. Bacula montera et étiquettera chaque cartouche porteuse d'un code barres contenue dans la librairie avec le nom spécifié par le code barres. L'enregistrement aproprié sera aussi créé dans le catalogue. Toute cartouche dont le code barres commence par les caractères spécifiés par la directive Cleaning Prefix est considérée comme une cartouche de nettoyage, et ne sera pas étiquetée. Par exemple, avec :
Pool { Name ... Cleaning Prefix = "CLN" }
toute cartouche de code barres CLNxxxx sera traitée en tant que cartouche de nettoyage, et ne sera pas montée.
unmount (changez vos cartouches et/ou ex\'ecutez mtx) mount
Si vous omettez de faire "unmount" avant de telles changements, Bacula ne saura plus ce qui est dans la librairie, et ce qui n'y est pas, et peut même cesser de fonctionner parce qu'il s'attend à avoir le contrôle exclusif de la librairie tandis quie le lecteur est monté.
Notez que les volumes doivent être pré-étiquetés pour pouvoir être utilisés automatiquement dans la librairie lors d'une sauvegarde. Si vous ne disposez pas d'un lecteur de code barres, ceci se fait manuellement, ou à l'aide d'un script.
Si vous avez plusieurs magasins ou si vous insérez ou retirez des cartouches d'un magasin, vous devriez en informer Bacula. Ainsi, Bacula sera en mesure d'utiliser préférentiellement des cartouches qu'il sait être dans la librairie, prévenant ainsi des interventions humaines inutiles.
Si votre librairie est équipée d'un lecteur de codes barres, il est aisé de tenir Bacula informé : chaque fois que vous changez un magasin, ajoutez ou prélevez une cartouche, faites simplement :
unmount (remove magazine) (insert new magazine) update slots mount
dans la console. Avec cette commande, Bacula se renseigne auprès de la librairie pour connaître les volumes qu'elle contient. Ceci ne nécessite pas d'accéder aux volumes car la librairie se charge de faire son inventaire lors de sa mise sous tension. Bacula s'assure alors que tout volume présent dans la librairie est marqué présent dans le catalogue et que tout volume absent de la librairie est marqué absent dans le catalogue. En outre, les numéros de slots des volumes sont corrigés dans le catalogue s'ils sont inexacts.
Si vous ne disposez pas d'un lecteur de codes barres, vous avez plusieurs alternatives :
update slots scan
qui ordonne à Bacula de lire l'étiquette (label) de chacune des cartouches dans la librairie par montage successif, et de mettre à jour les informations (Slot, drapeau InChanger) dans le catalogue. Cette méthode est efficace, mais prend du temps pour charger chaque cartouche et en lire l'étiquette.
Vous pouvez simuler un lecteur de codes barres dans votre librairie en faisant en sorte que le script mtx-changer retourne les informations que retournerait une librairie avec lecteur de codes barres. Pour cela, commentez la ligne ci-dessous dans le "case" aux alentours de la ligne 99 :
${MTX} -f $ctl status | grep " *Storage Element [0-9]*:.*Full" | awk "{print \$3 \$4}" | sed "s/Full *\(:VolumeTag=\)*//"
en ajoutant un # au début de cette ligne (vous pouvez aussi supprimer la ligne). A sa place, ajoutez une nouvelle ligne dont le rôle est d'imprimer le contenu d'un fichier. Par exemple :
cat /etc/bacula/changer.volumes
Le nom du fichier est libre, mais assurez vous d'utiliser un chemin absolu. Le contenu du fichier doit avoir le format :
1:Volume1 2:Volume2 3:Volume3 ...
Où 1, 2, 3 sont les numéros de slots et Volume1, Volume2, Volume3 sont les noms de volumes dans ces slots. Vous pouvez utiliser plusieurs fichiers représentant les contenus de plusieurs magasins, ainsi, lorsque vous changez de magasin, contentez vous de copier le contenu du fichier associé dans le fichier /etc/bacula/changer.volumes. Il n'est pas utile de stopper et redémarrer Bacula lors d'un changement de magasins, mettez simplement les bonnes valeurs dans le fichier avant de lancer la commande update slots. Votre librairie apparaîtra à Bacula comme équipée d'un lecteur de codes barres.
Si vous ne changez qu'une cartouche, vous ne voulez peut-être pas passer au crible tous vos volumes, c'est pourquoi la commande update slots (de même que la commande update slots scan) possède la forme additionnelle :
update slots=n1,n2,n3-n4, ...
où le mot-clef scan peut éventuellement être ajouté. n1, n2, n3-n4 représentent respectivement les numéros et la plage de slots que vous souhaitez mettre à jour.
Cette forme est particulièrement utile si vous voulez utiliser "scan" (couteux en temps) en restrignant l'opération à quelques slots.
Par exemple, si vous lancez la commande :
update slots=1,6 scan
Bacula va charger le volume du slot 6, lire son étiquette logicielle (label) et mettre à jour le catalogue, avant de faire de même avec la cartouche du slot 6. Avec la commande :
update slots=1-3,6
il va lire les codes barres des volumes dans les slots 1,2,3 et 6, et faire les mises à jour approriées dans le catalogue. Si vous n'avez pas de lecteur de codes barres, ni n'en simulez comme décrit plus haut, la commande ci-dessus ne trouvera aucun nom de volume et ne fera donc rien.
Si vous rencontrez des problèmes sur FreeBSD lorsque Bacula tente de sélectionner une cartouche, et si le message est Device not configured, c'est parce que FreeBSD a fait disparaître le fichier de périphérique /dev/nsa1 lorsqu'il n'y avait plus de cartouche montée dans le lecteur. Par conséquent, Bacula ne peut ouvrir le périphérique. Une solution consiste à charger une cartouche avant le lancement de Bacula. Ce problème est corrigé dans les versions de Bacula ultérieures à 1.32f-5.
Veuillez consulter le chapitre Tester votre lecteur de ce manuel pour d'importantes informations sur votre lecteur avant de passer au test de la librairie.
Avant d'essayer d'utiliser une librairie avec Bacula, il est préférable de vérifier "à la main" que le bras robotisé fonctionne. Pour ce faire, utilisez les commandes suivantes (à supposer que le script mtx-changer est installé dans /etc/bacula/mtx-changer) :
Cette commande devrait afficher :
1: 2: 3: ...
soit un numéro suivi de deux points pour chacun des slots occupé dans la librairie. Si votre librairie a un lecteur de codes barres, celui-ci sera affiché après les deux points. Si un message d'erreur s'affiche, vous devez résoudre le problème (par exemple, essayez un autre nom de périphérique si /dev/sg0 n'est pas le bon. PAr exemple, sur FreeBSD c'est généralement /dev/pass2).
Cette commande devrait retourner le nombre de slots de votre librairie.
Si une cartouche est chargée, cette commande devrait la décharger.
Si vous avez une cartouche dans le slot 3, elle sera chargée dans le slot de lecture (0).
devrait afficher "3"
Une fois que toutes les commandes ci-dessus fonctionnent correctement, Bacula devrait être capable d'utiliser la librairie, pourvu que votre configuration comporte la bonne commande Changer Command. A ce stade, il ne peut subsister qu'un problème : si votre librairie requiert un certain délai pour charger la cartouche après l'exécution de la commande. Immédiatement après avoir obtenu le retour du script mtx-changer, Bacula commande le rembobinage et la lecture de la bande. S'il obtient une erreur I/O, vous devriez probablement insérer une pause (sleep 20) après la commande mtx, mais prenez soin de terminer le script avec un code de sortie 0 en ajoutant exit 0 après toute commande que vous ajoutez au script, car Bacula contrôle le code de sortie du script qui devrait être 0 si tout s'est bien passé.
Vous pouvez tester si vous avez ou non besoin d'une telle pause en exécutant le script suivant :
#!/bin/sh /etc/bacula/mtx-changer /dev/sg0 unload /etc/bacula/mtx-changer /dev/sg0 load 3 mt -f /dev/st0 rewind mt -f /dev/st0 weof
S'il fonctionne correctement, vous n'êtes sans doute pas concerné par ce problème. Sinon, commencez par ajouter sleep 30 voire sleep 60 juste après la commande "/etc/bacula/mtx-changer /dev/sg0 load 3". Si ça marche, vous pouvez alors intégrer cette pause dans le script mtx-changer afin qu'elle soit effective lorsque Bacula est exécuté.
Quelques rares librairies exigent l'éjection de la cartouche avant de pouvoir la décharger. Dan ce cas, la commande /etc/bacula/mtx-changer /dev/sg0 load 3 ne fonctionne jamais, quel que soit la durée de la pause. Si vous pensez avoir ce problème, insérez une commande "eject" juste avant la commande /etc/bacula/mtx-changer /dev/sg0 unload :
#!/bin/sh /etc/bacula/mtx-changer /dev/sg0 unload mt -f /dev/st0 offline /etc/bacula/mtx-changer /dev/sg0 load 3 mt -f /dev/st0 rewind mt -f /dev/st0 weof
Naturellement, si vous avez besoin de la commande offline, vous devriez l'intégrer au script mtx-changer, en n'oubliant pas de préserver le code de sortie du script par l'ajout de exit 0.
Comme indiqué précédemment, plusieurs scripts qui implémentent ces fonctions sont regroupés dans <bacula-source>/examples/devices, ils peuvent vous inspirer pour faire en sorte que le votre fonctionne.
Si Bacula affiche "Rewind error on /dev/nst0. ERR=Input/output error." vous avez probablement besoin d'accroître la pause dans le script mtx-changer
Supposons que vous ayez convenablement défini les directives Device du Storage Daemon, et que vous ayez ajouté la directive Autochanger = yes dans la ressource Storage de votre fichier bacula-dir.conf.
Maintenant, alimentez votre librairie avec quelques cartouches vierges.
Que faire pour que Bacula accède à ces cartouches ?
Une stratégie consiste à pré-étiqueter chacune des cartouches. Pour cela, démarrez Bacula, puis utilisez la commande label dans la console :
./console Connecting to Director rufus:8101 1000 OK: rufus-dir Version: 1.26 (4 October 2002) *label
l'affichage devrait être :
Using default Catalog name=BackupDB DB=bacula The defined Storage resources are: 1: Autochanger 2: File Select Storage resource (1-2): 1
Choisissez la librairie (choix 1), vous obtenez :
Enter new Volume name: TestVolume1 Enter slot (0 for none): 1
Ici saisissez TestVolume1 en guise de nom, et 1 pour le slot. On vous demande alors :
Defined Pools: 1: Default 2: File Select the Pool (1-2): 1
Sélectionnez le pool Default (ce qui est fait automatiquement si vous n'avez que celui-là). Bacula poursuit en déchargeant toute cartouche chargée, en chargeant celle du slot 1 et en l'étiquetant. Dans cet exemple, le lecteur était vide, il en résulte l'affichage :
Connecting to Storage daemon Autochanger at localhost:9103 ... Sending label command ... 3903 Issuing autochanger "load slot 1" command. 3000 OK label. Volume=TestVolume1 Device=/dev/nst0 Media record for Volume=TestVolume1 successfully created. Requesting mount Autochanger ... 3001 Device /dev/nst0 is mounted with Volume TestVolume1 You have messages. *
Vous pouvez continuer à étiqueter les autres volumes, les messages changeront légèrement du fait qu'il y aura cette fois une cartouche à décharger avant de charger la suivante.
Une fois que tous vos volumes sont étiquetés, Bacula est en mesure de les charger lorsqu'il en a besoin.
Pour "voir" votre étiquetage, saisissez la commande list volumes dans la console, vous devriez obtenir quelque chose comme :
*{\bf list volumes} Using default Catalog name=BackupDB DB=bacula Defined Pools: 1: Default 2: File Select the Pool (1-2): 1 +-------+----------+--------+---------+-------+--------+----------+-------+------+ | MedId | VolName | MedTyp | VolStat | Bites | LstWrt | VolReten | Recyc | Slot | +-------+----------+--------+---------+-------+--------+----------+-------+------+ | 1 | TestVol1 | DDS-4 | Append | 0 | 0 | 30672000 | 0 | 1 | | 2 | TestVol2 | DDS-4 | Append | 0 | 0 | 30672000 | 0 | 2 | | 3 | TestVol3 | DDS-4 | Append | 0 | 0 | 30672000 | 0 | 3 | | ... | +-------+----------+--------+---------+-------+--------+----------+-------+------+
Bacula utilise les codes barres à travers deux commandes de la console : label barcodes et update slots.
La commande label barcodes ordonne à Bacula de lire tous les codes barres de toutes les cartouches présentes dans la librairie à l'aide de la commande mtx-changer list. Les cartouches sont ensuite montées l'une après l'autre pour être étiqueté du nom de leur code barres.
La commande update slots commence par obtenir du script mtx-changer la liste des cartouches et de leurs codes barres. Ensuite, il confronte chacune des valeurs du catalogues à cette liste afin de le mettre à jour. Notez que si un volume ne figure pas dans le catalogue, il n'y a rien a faire. Cette commande est utile pour synchroniser Bacula avec le contenu de la librairie si vous avez changé de magasin ou déplacé des cartouches.
La directive Cleaning Prefix peut être utilisée dans la ressource Pool pour définir un préfixe de nom de volume qui, s'il correspond au code barres d'un volume confère à ce volume le statut (VolStatus) Cleaning. Ceci évite que Bacula tente d'écrire sur une cartouche de nettoyage.
Bacula appelle le script mtx-changer que vous spécifiez au niveau de la directive Changer Command. En principe, ce sera le script mtx-changer que nous fournissons, mais ce pourrait être n'importe quel programme qui implémente les commandes loaded, load, unload, list, et slots qu'utilise Bacula. Voici le format sous lequel ces commandes doivent retourner les informations :
- Currently the changer commands used are: loaded -- retourne le num\'ero du slot d'origine de la cartouche charg\'ee, avec pour base 1 et 0 pour le lecteur. load -- charge la cartouche du slot sp\'ecifi\'e dans le lecteur.(notez que certains mat\'eriels requi\`erent une pause de 30 secondes apr\`es cette commande) unload -- d\'echarge le lecteur (la cartouche retourne dans son slot d'origine). list -- retourne une ligne par cartouche pr\'esente dans la librairie au format <slot>:<barcode> o\`u {\bf slot} est un entier non-nul repr\'esentant le num\'ero du slot, et {\bf barcode} est le code barres, s'il existe et si la librairie le prend en charge, associ\'e \`a la cartouche (dans le cas contraire, le champ "barcode" est laiss\'e blanc. slots -- retourne le nombre total de slots dans la librairie.
Bacula contrôle le code de sortie du programme appelé. Si ce code est 0, les informations sont acceptées. Dans le cas contraire, elles sont ignorées et le lecteur est traité comme s'il n'était pas dans une librairie.
J'hésite à qualifier ces librairies de "supportées", car les seules en ma possession et que je peux tester sont une HP SureStore DAT40X6 et une Overland PowerLoader LTO-2. Toutes les autres librairies citées ici ont été raportées comme fonctionnant avec Bacula par des utilisateurs. Notez que dans la colonne Capacité/Slot, je précise la capacité compressée par cartouche (ou slot).
OS | Fabr. | Media | Modèle | Slots | Cap/Slot |
Linux | Adic | DDS-3 | Adic 1200G | 12 | - |
Linux | Adic | DLT | FastStore 4000 | 7 | 20GB |
Linux | Adic | LTO-1/2, SDLT 320 | Adic Scalar 24 | 24 | 100GB |
Linux | Adic | LTO-2 | Adic FastStor 2, Sun Storedge L8 | 8 | 200GB |
- | CA-VM | ?? | Tape | ?? | ?? |
Linux | Dell | DLT VI,LTO-2 | PowerVault 122T/132T/136T | - | 100GB |
Linux | Dell | LTO-2 | PowerVault 124T | - | 200GB |
- | DFSMS | ?? | VM RMM | - | ?? |
Linux | Exabyte | VXA2 | VXA PacketLoader 1x10 2U | 10 | 80/160GB |
- | Exabyte | LTO | Magnum 1x7 LTO Tape Auotloader | 7 | 200/400GB |
Linux Gentoo 1.4 | Exabyte | AIT-2 | 215A | 15 (2 drives) | 50GB |
Linux | HP | DDS-4 | SureStore DAT-40X6 | 6 | 40GB |
Linux | HP | Ultrium-2/LTO | MSL 6000/ 60030/ 5052 | 28 | 200/400GB |
- | HP | DLT | A4853 DLT | 30 | 40/70GB |
Linux | HP (Compaq) | DLT VI | Compaq TL-895 | 96+4 import export | 35/70GB |
z/VM | IBM | ?? | IBM Tape Manager | - | ?? |
z/VM | IBM | ?? | native tape | - | ?? |
Linux | IBM | LTO | IBM 3581 Ultrium Tape Loader | 7 | 200/400GB |
SuSE 9.0 | IBM | LTO | IBM 3581 Ultrium Tape Loader | 7 | 200/400GB |
FreeBSD 5.4 | IBM | DLT | IBM 3502-R14 -- rebranded ATL L-500 | 14 | 35/70GB |
Debian | Overland | LTO | Overland LoaderXpress LTO/DLT8000 | 10-19 | 40-100GB |
Fedora | Overland | LTO | Overland PowerLoader LTO-2 | 10-19 | 200/400GB |
FreeBSD 5.4-Stable | Overland | LTO-2 | Overland Powerloader tape | 17 | 100GB |
- | Overland | LTO | Overland Neo2000 LTO | 26-30 | 100GB |
- | Quantum | ?? | Super Loader | ?? | ?? |
FreeBSD 4.9 | QUALSTAR TLS-4210 (Qualstar) | AIT1: 36GB, AIT2: 50GB all uncomp | QUALSTAR TLS-4210 | 12 | AIT1: 36GB, AIT2: 50GB all uncomp |
Linux | Skydata | DLT | ATL-L200 | 8 | 40/80 |
- | Sony | DDS-4 | TSL-11000 | 8 | 40GB |
Linux | Sony | AIT-2 | LIB-304(SDX-500C) | ? | 200GB |
Linux | Sony | AIT-3 | LIB-D81) | ? | 200GB |
FreeBSD 4.9-STABLE | Sony | AIT-1 | TSL-SA300C | 4 | 45/70GB |
- | Storagetek | DLT | Timberwolf DLT | 6 | 40/70 |
- | Storagetek | ?? | ACSLS | ?? | ?? |
Solaris | Sun | 4mm DLT | Sun Desktop Archive Python 29279 | 4 | 20GB |
Linux | Tandberg | DLT VI | VS 640 | 8? | 35/70GB |
Linux 2.6.x | Tandberg Data | SLR100 | SLR100 Autoloader | 8 | 50/100GB |
Bacula allows you to specify that you want the Storage daemon to initially write your data to disk and then subsequently to tape. This serves several important purposes.
Data spooling is exactly that "spooling". It is not a way to first write a "backup" to a disk file and then to a tape. When the backup has only been spooled to disk, it is not complete yet and cannot be restored until it is written to tape. In a future version, Bacula will support writing a backup to disk then later Migrating or Copying it to a tape.
The remainder of this chapter explains the various directives that you can use in the spooling process.
The following directives can be used to control data spooling.
SpoolData = yes|no
SpoolData = yes|no
Maximum Spool Size = size Where size is a the maximum spool size for all jobs specified in bytes.
Maximum Job Spool Size = size Where size is the maximum spool file size for a single job specified in bytes.
Spool Directory = directory
Please be very careful to exclude the spool directory from any backup, otherwise, your job will write enormous amounts of data to the Volume, and most probably terminate in error. This is because in attempting to backup the spool file, the backup data will be written a second time to the spool file, and so on ad infinitum.
Another advice is to always specify the maximum spool size so that your disk doesn't completely fill up. In principle, data spooling will properly detect a full disk, and despool data allowing the job to continue. However, attribute spooling is not so kind to the user. If the disk on which attributes are being spooled fills, the job will be canceled. In addition, if your working directory is on the same partition as the spool directory, then Bacula jobs will fail possibly in bizarre ways when the spool fills.
You may be asking what Python is and why a scripting language is needed in Bacula. The answer to the first question is that Python is an Object Oriented scripting language with features similar to those found in Perl, but the syntax of the language is much cleaner and simpler. The answer to why have scripting in Bacula is to give the user more control over the whole backup process. Probably the simplest example is when Bacula needs a new Volume name, with a scripting language such as Python, you can generate any name you want, based on the current state of Bacula.
Python must be enabled during the configuration process by adding
a --
with-python, and possibly specifying an alternate
directory if your Python is not installed in a standard system
location. If you are using RPMs you will need the python-devel package
installed.
When Python is configured, it becomes an integral part of Bacula and runs in Bacula's address space, so even though it is an interpreted language, it is very efficient.
When the Director starts, it looks to see if you have a Scripts Directory Directive defined (normal default /etc/bacula/scripts, if so, it looks in that directory for a file named DirStartUp.py. If it is found, Bacula will pass this file to Python for execution. The Scripts Directory is a new directive that you add to the Director resource of your bacula-dir.conf file.
Note: Bacula does not install Python scripts by default because these scripts are for you to program. This means that with a default installation with Python enabled, Bacula will print the following error message:
09-Jun 15:14 bacula-dir: ERROR in pythonlib.c:131 Could not import Python script /etc/bacula/scripts/DirStartUp. Python disabled.
The source code directory examples/python contains sample scripts for DirStartUp.py, SDStartUp.py, and FDStartUp.py that you might want to use as a starting point. Normally, your scripts directory (at least where you store the Python scripts) should be writable by Bacula, because Python will attempt to write a compiled version of the scripts (e.g. DirStartUp.pyc) back to that directory.
When starting with the sample scripts, you can delete any part that you will not need, but you should keep all the Bacula Event and Job Event definitions. If you do not want a particular event, simply replace the existing code with a noop = 1.
There are four Python objects that you will need to work with:
The first thing the startup script must do is to define what global Bacula events (daemon events), it wants to see. This is done by creating a Bacula Events class, instantiating it, then passing it to the set_events method. There are three possible events.
Access to the Bacula variables and methods is done with:
import bacula
The following are the read-only attributes provided by the bacula object.
A simple definition of the Bacula Events Class might be the following:
import sys, bacula class BaculaEvents: def JobStart(self, job): ...
Then to instantiate the class and pass it to Bacula, you would do:
bacula.set_events(BaculaEvents()) # register Bacula Events wanted
And at that point, each time a Job is started, your BaculaEvents JobStart method will be called.
Now to actually do anything with a Job, you must define which Job events you want to see, and this is done by defining a JobEvents class containing the methods you want called. Each method name corresponds to one of the Job Events that Bacula will generate.
A simple Job Events class might look like the following:
class JobEvents: def NewVolume(self, job): ...
Here, your JobEvents class method NewVolume will be called each time the Job needs a new Volume name. To actually register the events defined in your class with the Job, you must instantiate the JobEvents class and set it in the Job set_events variable. Note, this is a bit different from how you registered the Bacula events. The registration process must be done in the Bacula JobStart event (your method). So, you would modify Bacula Events (not the Job events) as follows:
import sys, bacula class BaculaEvents: def JobStart(self, job): events = JobEvents() # create instance of Job class job.set_events(events) # register Job events desired ...
When a job event is triggered, the appropriate event definition is called in the JobEvents class. This is the means by which your Python script or code gets control. Once it has control, it may read job attributes, or set them. See below for a list of read-only attributes, and those that are writable.
In addition, the Bacula job obbject in the Director has a number of methods (subroutines) that can be called. They are:
The following attributes are read/write within the Director for the job object.
The following read-only attributes are available within the Director for the job object.
The following write-only attributes are available within the Director:
There is a new Console command named python. It takes a single argument restart. Example:
python restart
This command restarts the Python interpreter in the Director. This can be useful when you are modifying the DirStartUp script, because normally Python will cache it, and thus the script will be read one time.
If you are having problems loading DirStartUp.py, you will probably not get any error messages because Bacula can only print Python error messages after the Python interpreter is started. However, you may be able to see the error messages by starting Bacula in a shell window with the -d1 option on the command line. That should cause the Python error messages to be printed in the shell window.
If you are getting error messages such as the following when loading DirStartUp.py:
Traceback (most recent call last): File "/etc/bacula/scripts/DirStartUp.py", line 6, in ? import time, sys, bacula ImportError: /usr/lib/python2.3/lib-dynload/timemodule.so: undefined symbol: PyInt_FromLong bacula-dir: pythonlib.c:134 Python Import error.
It is because the DirStartUp script is calling a dynamically loaded module (timemodule.so in the above case) that then tries to use Python functions exported from the Python interpreter (in this case PyInt_FromLong). The way Bacula is currently linked with Python does not permit this. The solution to the problem is to put such functions (in this case the import of time into a separate Python script, which will do your calculations and return the values you want. Then call (not import) this script from the Bacula DirStartUp.py script, and it all should work as you expect.
An example script for the Director startup file is provided in examples/python/DirStartup.py as follows:
# # Bacula Python interface script for the Director # # You must import both sys and bacula import sys, bacula # This is the list of Bacula daemon events that you # can receive. class BaculaEvents: def __init__(self): # Called here when a new Bacula Events class is # is created. Normally not used noop = 1 def JobStart(self, job): """ Called here when a new job is started. If you want to do anything with the Job, you must register events you want to receive. """ events = JobEvents() # create instance of Job class events.job = job # save Bacula's job pointer job.set_events(events) # register events desired sys.stderr = events # send error output to Bacula sys.stdout = events # send stdout to Bacula jobid = job.JobId; client = job.Client numvols = job.NumVols job.JobReport="Python Dir JobStart: JobId=%d Client=%s NumVols=%d\n" % (jobid,client,numvols) # Bacula Job is going to terminate def JobEnd(self, job): jobid = job.JobId client = job.Client job.JobReport="Python Dir JobEnd output: JobId=%d Client=%s.\n" % (jobid, client) # Called here when the Bacula daemon is going to exit def Exit(self, job): print "Daemon exiting." bacula.set_events(BaculaEvents()) # register daemon events desired """ These are the Job events that you can receive. """ class JobEvents: def __init__(self): # Called here when you instantiate the Job. Not # normally used noop = 1 def JobInit: # Called when the job is first scheduled noop = 1 def JobRun: # Called just before running the job after initializing # This is the point to change most Job parameters. # It is equivalent to the JobRunBefore point. noop = 1 def NewVolume(self, job): # Called when Bacula wants a new Volume name. The Volume # name returned, if any, must be stored in job.VolumeName jobid = job.JobId client = job.Client numvol = job.NumVols volname = "TestA-001" job.JobReport = "JobId=%d Client=%s NumVols=%d VolumeName=%s" % (jobid, client, numvol,volname) job.JobReport="Python before New Volume set for Job.\n" job.VolumeName=volname def VolumePurged(self, job): # Called when a Volume is purged. The Volume name can be referenced # with job.VolumeName noop = 1
Bacula supports ANSI or IBM tape labels as long as you enable it. In fact, with the proper configuration, you can force Bacula to require ANSI or IBM labels.
Bacula can create an ANSI or IBM label, but if Check Labels is enabled (see below), Bacula will look for an existing label, and if it is found, it will keep the label. Consequently, you can label the tapes with programs other than Bacula, and Bacula will recognize and support them.
Even though Bacula will recognize and write ANSI and IBM labels, it always writes its own tape labels as well.
When using ANSI or IBM tape labeling, you must restrict your Volume names to a maximum of 6 characters.
If you have labeled your Volumes outside of Bacula, then the ANSI/IBM label will be recognized by Bacula only if you have created the HDR1 label with BACULA.DATA in the Filename field (starting with character 5). If Bacula writes the labels, it will use this information to recognize the tape as a Bacula tape. This allows ANSI/IBM labeled tapes to be used at sites with multiple machines and multiple backup programs.
These are questions that have been submitted over time by the Bacula users.
Please also see the bugs section of this document for a list of known bugs and solutions.
Bacula has been my only backup tool for over four years backing up 5 machines nightly (3 Linux boxes running RedHat, a WinXP machine, and a WinNT machine).
There are a number of reasons for this stability.
During the authorization process, the Storage daemon and File daemon also require that the Director authenticates itself, so both ends require the other to have the correct name and password.
If you have edited the conf files and modified any name or any password, and you are getting authentication errors, then your best bet is to go back to the original conf files generated by the Bacula installation process. Make only the absolutely necessary modifications to these files -- e.g. add the correct email address. Then follow the instructions in the Running Bacula chapter of this manual. You will run a backup to disk and a restore. Only when that works, should you begin customization of the conf files.
Another reason that you can get authentication errors is if you are running Multiple Concurrent Jobs in the Director, but you have not set them in the File daemon or the Storage daemon. Once you reach their limit, they will reject the connection producing authentication (or connection) errors.
If you are having problems connecting to a Windows machine that previously worked, you might try restarting the Bacula service since Windows frequently encounters networking connection problems.
Some users report that authentication fails if there is not a proper reverse DNS lookup entry for the machine. This seems to be a requirement of gethostbyname(), which is what Bacula uses to translate names into IP addresses. If you cannot add a reverse DNS entry, or you don't know how to do so, you can avoid the problem by specifying an IP address rather than a machine name in the appropriate Bacula conf file.
Here is a picture that indicates what names/passwords in which files/Resources must match up:
In the left column, you will find the Director, Storage, and Client resources, with their names and passwords -- these are all in bacula-dir.conf. The right column is where the corresponding values should be found in the Console, Storage daemon (SD), and File daemon (FD) configuration files.
Another thing to check is to ensure that the Bacula component you are trying to access has Maximum Concurrent Jobs set large enough to handle each of the Jobs and the Console that want to connect simultaneously. Once the maximum connections has been reached, each Bacula component will reject all new connections.
Finally, make sure you have no hosts.allow or hosts.deny file that is not permitting access to the site trying to connect.
cd <bacula-source>/src/cats ./drop_mysql_tables ./make_mysql_tables
If you are using SQLite, do the following:
Delete bacula.db from your working directory. cd <bacula-source>/src/cats ./drop_sqlite_tables ./make_sqlite_tables
Then write an EOF on each tape you used with Bacula using:
mt -f /dev/st0 rewind mt -f /dev/st0 weof
where you need to adjust the device name for your system.
If you have previously done an unmount command, all Storage daemon sessions (jobs) will be completely blocked from using the drive unmounted, so be sure to do a mount after your unmount. If in doubt, do a second mount, it won't cause any harm.
For the first problem, see the next FAQ question. For the second problem, please review the Windows Installation instructions in this manual.
To see what is going on when the File daemon starts on Windows, do the following:
Start a DOS shell Window. cd c:\bacula\bin bacula-fd -d100 -c c:\bacula\bin\bacula-fd.conf
This will cause the FD to write a file bacula.trace in the current directory, which you can examine and thereby determine the problem.
[When I Start the Console, the Error Messages Fly By. How can I see them? ] Either use a shell window with a scroll bar, or use the gnome-console. In any case, you probably should be logging all output to a file, and then you can simply view the file using an editor or the less program. To log all output, I have the following in my Director's Message resource definition:
append = "/home/kern/bacula/bin/log" = all, !skipped
Obviously you will want to change the filename to be appropriate for your system.
[I didn't realize that the backups were not working on my Windows Client. What should I do? ] You should be sending yourself an email message for each job. This will avoid the possibility of not knowing about a failed backup. To do so put something like:
Mail = yourname@yourdomain = all, !skipped
in your Director's message resource. You should then receive one email for each Job that ran. When you are comfortable with what is going on (it took me 9 months), you might change that to:
MailOnError = yourname@yourdomain = all, !skipped
then you only get email messages when a Job errors as is the case for your Windows machine.
You should also be logging the Director's messages, please see the previous FAQ for how to do so.
If you want to read a document that pertains only to a specific version, please use the one distributed in the source code.
Another reason why Bacula may not find a suitable Full backup is that every time you change the FileSet, Bacula will require a new Full backup. This is necessary to ensure that all files are properly backed up in the case where you have added more files to the FileSet. Beginning with version 1.31, the FileSets are also dated when they are created, and this date is displayed with the name when you are listing or selecting a FileSet. For more on backup levels see below.
Note that due to limitations Win32 path and filenames cannot exceed 260 characters. By using Win32 Unicode functions, we will remove this restriction in later versions of Bacula.
The second feature, which gives a lot of power and flexibility to Bacula is the Bootstrap record definition.
The third unique feature, which is currently (1.30) unimplemented, and thus can be called vaporware :-), is Base level saves. When implemented, this will enormously reduce tape usage.
The most common problem is that you have not specified a fully qualified email address and your bsmtp server is rejecting the mail. The next most common problem is that your bsmtp server doesn't like the syntax on the From part of the message. For more details on this and other problems, please see the Getting Email Notification to Work section of the Tips chapter of this manual. The section Getting Notified of Job Completion of the Tips chapter may also be useful. For more information on the bsmtp mail program, please see bsmtp in the Volume Utility Tools chapter of this manual.
When Bacula creates a Media record (Volume), it uses many default values from the Pool record. If you subsequently change the Pool record, the new values will be used as a default for the next Volume that is created, but if you want the new values to apply to existing Volumes, you must manually update the Volume Catalog entry using the update volume command in the Console program.
Bacula also has compression code, which is normally used only when backing up to file Volumes. There are two conditions for this "software" to become enabled.
If the library is found by Bacula during the ./configure it will be mentioned in the config.out line by:
ZLIB support: yes
[Bacula is Asking for a New Tape After 2 GB of Data but My Tape holds 33 GB. Why?] There are several reasons why Bacula will request a new tape.
If after reading the above mentioned section, you believe that Bacula is not correctly handling the level (Differential/Incremental), please send us the following information for analysis:
The above information can allow us to analyze what happened, without it, there is not much we can do.
There are several things you can do to improve the situation.
SD Connect Timeout = 5 min
in the FileDaemon resource.
bacula-dir -c bacula-dir.conf ... 0>\&1 2>\&1 >/dev/null
and likewise for the other daemons.
For example, I keep a 30 day retention period for my Files to keep my catalog from getting too big, but I keep my tapes for a minimum of one year, just in case.
llist Volume=xxx
If it doesn't have the right value, you can use:
update Volume=xxx
to change it.
If it is neither of the above, please submit a bug report at bugs.bacula.org.
Another solution might be to run the daemon with the debug option by:
Start a DOS shell Window. cd c:\bacula\bin bacula-fd -d100 -c c:\bacula\bin\bacula-fd.conf
This will cause the FD to write a file bacula.trace in the current directory, which you can examine to determine the problem.
In at least one case, the problem has been a bad driver for a Win32 NVidia NForce 3 ethernet card with driver (4.4.2 17/05/2004). In this case, a good driver is (4.8.2.0 06/04/2005). Moral of the story, make sure you have the latest ethernet drivers loaded, or use the following workaround as suggested by Thomas Simmons for Win32 machines:
Browse to: Start > Control Panel > Network Connections
Right click the connection for the nvidia adapter and select properties. Under the General tab, click "Configure...". Under the Advanced tab set "Checksum Offload" to disabled and click OK to save the change.
Lack of communications, or communications that get interrupted can also be caused by Linux firewalls where you have a rule that throttles connections or traffic. For example, if you have:
iptables -t filter -A INPUT -m limit --limit 3/second --limit-burst 3 -j DROP
you will want to add the following rules before the above rule:
iptables -t filter -A INPUT --dport 9101 -j ACCEPT iptables -t filter -A INPUT --dport 9102 -j ACCEPT iptables -t filter -A INPUT --dport 9103 -j ACCEPTThis will ensure that any Bacula traffic will not get terminated because of high usage rates.
In fact, you do not tell Bacula what tapes to use. It is the inverse. Bacula tells you want tapes it wants. You put tapes at its disposition and it chooses.
Now, if you *really* want to be tricky and try to tell Bacula what to do, it will be reasonable if for example you mount a valid tape that it can use on a drive, it will most likely go ahead and use it. It also has a documented algorithm for choosing tapes -- but you are asking for problems ...
So, the trick is to invert your concept of things and put Bacula in charge of handling the tapes. Once you do that, you will be fine. If you want to anticipate what it is going to do, you can generally figure it out correctly and get what you want.
If you start with the idea that you are going to force or tell Bacula to use particular tapes or you insist on trying to run in that kind of mode, you will probably not be too happy.
I don't want to worry about what tape has what data. That is what Bacula is designed for.
If you have an application where you *really* need to remove a tape each day and insert a new one, it can be done the directives exist to accomplish that. In such a case, one little "trick" to knowing what tape Bacula will want at 2am while you are asleep is to run a tiny job at 4pm while you are still at work that backs up say one directory, or even one file. You will quickly find out what tape it wants, and you can mount it before you go home ...
There are a number of example scripts for various things that can be found in the example subdirectory and its subdirectories of the Bacula source distribution.
The first thing to do before upgrading from one version to another is to ensure that you don't overwrite or delete your production (current) version of Bacula until you have tested that the new version works.
If you have installed Bacula into a single directory, this is simple: simply make a copy of your Bacula directory.
If you have done a more typical Unix installation where the binaries are placed in one directory and the configuration files are placed in another, then the simplest way is to configure your new Bacula to go into a single file. Alternatively, make copies of all your binaries and especially your conf files.
Whatever your situation may be (one of the two just described), you should probably start with the defaultconf script that can be found in the examples subdirectory. Copy this script to the main Bacula directory, modify it as necessary (there should not need to be many modifications), configure Bacula, build it, install it, then stop your production Bacula, copy all the *.conf files from your production Bacula directory to the test Bacula directory, start the test version, and run a few test backups. If all seems good, then you can proceed to install the new Bacula in place of or possibly over the old Bacula.
When installing a new Bacula you need not worry about losing the changes you made to your configuration files as the installation process will not overwrite them providing that you do not do a make uninstall.
If the new version of Bacula requires an upgrade to the database, you can upgrade it with the script update_bacula_tables, which will be installed in your scripts directory (default /etc/bacula), or alternatively, you can find it in the <bacula-source>/src/cats directory.
One of the first things you should do is to ensure that you are being properly notified of the status of each Job run by Bacula, or at a minimum of each Job that terminates with an error.
Until you are completely comfortable with Bacula, we recommend that you send an email to yourself for each Job that is run. This is most easily accomplished by adding an email notification address in the Messages resource of your Director's configuration file. An email is automatically configured in the default configuration files, but you must ensure that the default root address is replaced by your email address.
For examples of how I (Kern) configure my system, please take a look at the .conf files found in the examples sub-directory. We recommend the following configuration (where you change the paths and email address to correspond to your setup). Note, the mailcommand and operatorcommand should be on a single line. They were split here for presentation:
Messages { Name = Standard mailcommand = "/home/bacula/bin/bsmtp -h localhost -f \"\(Bacula\) %r\" -s \"Bacula: %t %e of %c %l\" %r" operatorcommand = "/home/bacula/bin/bsmtp -h localhost -f \"\(Bacula\) %r\" -s \"Bacula: Intervention needed for %j\" %r" Mail = your-email-address = all, !skipped, !terminate append = "/home/bacula/bin/log" = all, !skipped, !terminate operator = your-email-address = mount console = all, !skipped, !saved }
You will need to ensure that the /home/bacula/bin path on the mailcommand and the operatorcommand lines point to your Bacula binary directory where the bsmtp program will be installed. You will also want to ensure that the your-email-address is replaced by your email address, and finally, you will also need to ensure that the /home/bacula/bin/log points to the file where you want to log all messages.
With the above Messages resource, you will be notified by email of every Job that ran, all the output will be appended to the log file you specify, all output will be directed to the console program, and all mount messages will be emailed to you. Note, some messages will be sent to multiple destinations.
The form of the mailcommand is a bit complicated, but it allows you to distinguish whether the Job terminated in error or terminated normally. Please see the Mail Command section of the Messages Resource chapter of this manual for the details of the substitution characters used above.
Once you are totally comfortable with Bacula as I am, or if you have a large number of nightly Jobs as I do (eight), you will probably want to change the Mail command to Mail On Error which will generate an email message only if the Job terminates in error. If the Job terminates normally, no email message will be sent, but the output will still be appended to the log file as well as sent to the Console program.
The section above describes how to get email notification of job status. Occasionally, however, users have problems receiving any email at all. In that case, the things to check are the following:
director = director-name = all
mailcommand = "mail -s test your@domain.com"
mailcommand = "/home/bacula/bin/bsmtp -f \"root@localhost\" %r"
If like me, you have setup Bacula so that email is sent only when a Job has errors, as described in the previous section of this chapter, inevitably, one day, something will go wrong and Bacula can stall. This could be because Bacula crashes, which is vary rare, or more likely the network has caused Bacula to hang for some unknown reason.
To avoid this, you can use the RunAfterJob command in the Job resource to schedule a Job nightly, or weekly that simply emails you a message saying that Bacula is still running. For example, I have setup the following Job in my Director's configuration file:
Schedule { Name = "Watchdog" Run = Level=Full sun-sat at 6:05 } Job { Name = "Watchdog" Type = Admin Client=Watchdog FileSet="Verify Set" Messages = Standard Storage = DLTDrive Pool = Default Schedule = "Watchdog" RunAfterJob = "/home/kern/bacula/bin/watchdog %c %d" } Client { Name = Watchdog Address = rufus FDPort = 9102 Catalog = Verify Password = "" File Retention = 1day Job Retention = 1 month AutoPrune = yes }
Where I established a schedule to run the Job nightly. The Job itself is type Admin which means that it doesn't actually do anything, and I've defined a FileSet, Pool, Storage, and Client, all of which are not really used (and probably don't need to be specified). The key aspect of this Job is the command:
RunAfterJob = "/home/kern/bacula/bin/watchdog %c %d"
which runs my "watchdog" script. As an example, I have added the Job codes %c and %d which will cause the Client name and the Director's name to be passed to the script. For example, if the Client's name is Watchdog and the Director's name is main-dir then referencing $1 in the script would get Watchdog and referencing $2 would get main-dir. In this case, having the script know the Client and Director's name is not really useful, but in other situations it may be.
You can put anything in the watchdog script. In my case, I like to monitor the size of my catalog to be sure that Bacula is really pruning it. The following is my watchdog script:
#!/bin/sh cd /home/kern/mysql/var/bacula du . * | /home/kern/bacula/bin/bsmtp \ -f "\(Bacula\) abuse@whitehouse.com" -h mail.yyyy.com \ -s "Bacula running" abuse@whitehouse.com
If you just wish to send yourself a message, you can do it with:
#!/bin/sh cd /home/kern/mysql/var/bacula /home/kern/bacula/bin/bsmtp \ -f "\(Bacula\) abuse@whitehouse.com" -h mail.yyyy.com \ -s "Bacula running" abuse@whitehouse.com <<END-OF-DATA Bacula is still running!!! END-OF-DATA
By using a WriteBootstrap record in each of your Director's Job resources, you can constantly maintain a bootstrap file that will enable you to recover the state of your system as of the last backup without having the Bacula catalog. This permits you to more easily recover from a disaster that destroys your Bacula catalog.
When a Job resource has a WriteBootstrap record, Bacula will maintain the designated file (normally on another system but mounted by NSF) with up to date information necessary to restore your system. For example, in my Director's configuration file, I have the following record:
Write Bootstrap = "/mnt/deuter/files/backup/client-name.bsr"
where I replace client-name by the actual name of the client that is being backed up. Thus, Bacula automatically maintains one file for each of my clients. The necessary bootstrap information is appended to this file during each Incremental backup, and the file is totally rewritten during each Full backup.
Note, one disadvantage of writing to an NFS mounted volume as I do is that if the other machine goes down, the OS will wait forever on the fopen() call that Bacula makes. As a consequence, Bacula will completely stall until the machine exporting the NSF mounts comes back up. A possible solution to this problem was provided by Andrew Hilborne, and consists of using the soft option instead of the hard option when mounting the NFS volume, which is typically done in /etc/fstab/. The NFS documentation explains these options in detail. However, I found that with the soft option NFS disconnected frequently causing even more problems.
If you are starting off in the middle of a cycle (i.e. with Incremental backups) rather than at the beginning (with a Full backup), the bootstrap file will not be immediately valid as it must always have the information from a Full backup as the first record. If you wish to synchronize your bootstrap file immediately, you can do so by running a restore command for the client and selecting a full restore, but when the restore command asks for confirmation to run the restore Job, you simply reply no, then copy the bootstrap file that was written to the location specified on the Write Bootstrap record. The restore bootstrap file can be found in restore.bsr in the working directory that you defined. In the example given below for the client rufus, my input is shown in bold. Note, the JobId output has been partially truncated to fit on the page here:
(in the Console program) *{\bf restore} First you select one or more JobIds that contain files to be restored. You will then be presented several methods of specifying the JobIds. Then you will be allowed to select which files from those JobIds are to be restored. To select the JobIds, you have the following choices: 1: List last 20 Jobs run 2: List Jobs where a given File is saved 3: Enter list of JobIds to select 4: Enter SQL list command 5: Select the most recent backup for a client 6: Cancel Select item: (1-6): {\bf 5} The defined Client resources are: 1: Minimatou 2: Rufus 3: Timmy Select Client (File daemon) resource (1-3): {\bf 2} The defined FileSet resources are: 1: Kerns Files Item 1 selected automatically. +-------+------+-------+---------+---------+------+-------+------------+ | JobId | Levl | Files | StrtTim | VolName | File | SesId | VolSesTime | +-------+------+-------+---------+---------+------+-------+------------+ | 2 | F | 84 | ... | test1 | 0 | 1 | 1035645259 | +-------+------+-------+---------+---------+------+-------+------------+ You have selected the following JobId: 2 Building directory tree for JobId 2 ... The defined Storage resources are: 1: File Item 1 selected automatically. You are now entering file selection mode where you add and remove files to be restored. All files are initially added. Enter "done" to leave this mode. cwd is: / $ {\bf done} 84 files selected to restore. Run Restore job JobName: kernsrestore Bootstrap: /home/kern/bacula/working/restore.bsr Where: /tmp/bacula-restores FileSet: Kerns Files Client: Rufus Storage: File JobId: *None* OK to run? (yes/mod/no): {\bf no} {\bf quit} (in a shell window) {\bf cp ../working/restore.bsr /mnt/deuter/files/backup/rufus.bsr}
Bacula keeps a count of the number of files on each Volume in its Catalog database so that before appending to a tape, it can verify that the number of files are correct, and thus prevent overwriting valid data. If the Director or the Storage daemon crashes before the job has completed, the tape will contain one more file than is noted in the Catalog, and the next time you attempt to use the same Volume, Bacula will reject it due to a mismatch between the physical tape (Volume) and the catalog.
The easiest solution to this problem is to label a new tape and start fresh. If you wish to continue appending to the current tape, you can do so by using the update command in the console program to change the Volume Files entry in the catalog. A typical sequence of events would go like the following:
- Bacula crashes - You restart Bacula
Bacula then prints:
17-Jan-2003 16:45 rufus-dir: Start Backup JobId 13, Job=kernsave.2003-01-17_16.45.46 17-Jan-2003 16:45 rufus-sd: Volume test01 previously written, moving to end of data. 17-Jan-2003 16:46 rufus-sd: kernsave.2003-01-17_16.45.46 Error: I cannot write on this volume because: The number of files mismatch! Volume=11 Catalog=10 17-Jan-2003 16:46 rufus-sd: Job kernsave.2003-01-17_16.45.46 waiting. Cannot find any appendable volumes. Please use the "label" command to create a new Volume for: Storage: SDT-10000 Media type: DDS-4 Pool: Default
(note, lines wrapped for presentation) The key here is the line that reads:
The number of files mismatch! Volume=11 Catalog=10
It says that Bacula found eleven files on the volume, but that the catalog says there should be ten. When you see this, you can be reasonably sure that the SD was interrupted while writing before it had a chance to update the catalog. As a consequence, you can just modify the catalog count to eleven, and even if the catalog contains references to files saved in file 11, everything will be OK and nothing will be lost. Note that if the SD had written several file marks to the volume, the difference between the Volume cound and the Catalog count could be larger than one, but this is unusual.
If on the other hand the catalog is marked as having more files than Bacula found on the tape, you need to consider the possible negative consequences of modifying the catalog. Please see below for a more complete discussion of this.
Continuing with the example of Volume = 11 Catalog = 10, to enable to Bacula to append to the tape, you do the following:
{\bf update} Update choice: 1: Volume parameters 2: Pool from resource 3: Slots from autochanger Choose catalog item to update (1-3): {\bf 1} Defined Pools: 1: Default 2: File Select the Pool (1-2): +-------+---------+--------+---------+-----------+------+----------+------+-----+ | MedId | VolName | MedTyp | VolStat | VolBytes | Last | VolReten | Recy | Slt | +-------+---------+--------+---------+-----------+------+----------+------+-----+ | 1 | test01 | DDS-4 | Error | 352427156 | ... | 31536000 | 1 | 0 | +-------+---------+--------+---------+-----------+------+----------+------+-----+ Enter MediaId or Volume name: {\bf 1}
(note table output truncated for presentation) First, you chose to update the Volume parameters by entering a 1. In the volume listing that follows, notice how the VolStatus is Error. We will correct that after changing the Volume Files. Continuing, you respond 1,
Updating Volume "test01" Parameters to modify: 1: Volume Status 2: Volume Retention Period 3: Volume Use Duration 4: Maximum Volume Jobs 5: Maximum Volume Files 6: Maximum Volume Bytes 7: Recycle Flag 8: Slot 9: Volume Files 10: Pool 11: Done Select parameter to modify (1-11): {\bf 9} Warning changing Volume Files can result in loss of data on your Volume Current Volume Files is: 10 Enter new number of Files for Volume: {\bf 11} New Volume Files is: 11 Updating Volume "test01" Parameters to modify: 1: Volume Status 2: Volume Retention Period 3: Volume Use Duration 4: Maximum Volume Jobs 5: Maximum Volume Files 6: Maximum Volume Bytes 7: Recycle Flag 8: Slot 9: Volume Files 10: Pool 11: Done Select parameter to modify (1-10): {\bf 1}
Here, you have selected 9 in order to update the Volume Files, then you changed it from 10 to 11, and you now answer 1 to change the Volume Status.
Current Volume status is: Error Possible Values are: 1: Append 2: Archive 3: Disabled 4: Full 5: Used 6: Read-Only Choose new Volume Status (1-6): {\bf 1} New Volume status is: Append Updating Volume "test01" Parameters to modify: 1: Volume Status 2: Volume Retention Period 3: Volume Use Duration 4: Maximum Volume Jobs 5: Maximum Volume Files 6: Maximum Volume Bytes 7: Recycle Flag 8: Slot 9: Volume Files 10: Pool 11: Done Select parameter to modify (1-11): {\bf 11} Selection done.
At this point, you have changed the Volume Files from 10 to 11 to account for the last file that was written but not updated in the database, and you changed the Volume Status back to Append.
This was a lot of words to describe something quite simple.
The Volume Files option exists only in version 1.29 and later, and you should be careful using it. Generally, if you set the value to that which Bacula said is on the tape, you will be OK, especially if the value is one more than what is in the catalog.
Now lets consider the case:
The number of files mismatch! Volume=10 Catalog=12
Here the Bacula found fewer files on the volume than what is marked in the catalog. Now, in this case, you should hesitate a lot before modifying the count in the catalog, because if you force the catalog from 12 to 10, Bacula will start writing after the file 10 on the tape, possibly overwriting valid data, and if you ever try to restore any of the files that the catalog has marked as saved on Files 11 and 12, all chaos will break out. In this case, you will probably be better off using a new tape. In fact, you might want to see what files the catalog claims are actually stored on that Volume, and back them up to another tape and recycle this tape.
Only the File daemon needs to run with root permission (so that it can access all files). As a consequence, you may run your Director, Storage daemon, and MySQL or PostgreSQL database server as non-root processes. Version 1.30 has the -u and the -g options that allow you to specify a userid and groupid on the command line to be used after Bacula starts.
As of version 1.33, thanks to Dan Langille, it is easier to configure the Bacula Director and Storage daemon to run as non-root.
You should protect the Bacula port addresses (normally 9101, 9102, and 9103) from outside access by a firewall or other means of protection to prevent unauthorized use of your daemons.
You should ensure that the configuration files are not world readable since they contain passwords that allow access to the daemons. Anyone who can access the Director using a console program can restore any file from a backup Volume.
You should protect your Catalog database. If you are using SQLite, make sure
that the working directory is readable only by root (or your Bacula userid),
and ensure that bacula.db has permissions -rw-r--
r--
(i.e. 640) or
more strict. If you are using MySQL or PostgreSQL, please note that the Bacula
setup procedure leaves the database open to anyone. At a minimum, you should
assign the user bacula a userid and add it to your Director's
configuration file in the appropriate Catalog resource.
If you normally change tapes every day or at least every Friday, but Thursday is a holiday, you can use a trick proposed by Lutz Kittler to ensure that no job runs on Thursday so that you can insert Friday's tape and be sure it will be used on Friday. To do so, define a RunJobBefore script that normally returns zero, so that the Bacula job will normally continue. You can then modify the script to return non-zero on any day when you do not want Bacula to run the job.
If you have an autochanger but it does not support barcodes, using a "trick" you can make Bacula automatically label all the volumes in your autochanger's magazine.
First create a file containing one line for each slot in your autochanger that has a tape to be labeled. The line will contain the slot number a colon (:) then the Volume name you want to use. For example, create a file named volume-list, which contains:
1:Volume001 2:TestVolume02 5:LastVolume
The records do not need to be in any order and you don't need to mention all the slots. Normally, you will have a consistent set of Volume names and a sequential set of numbers for each slot you want labeled. In the example above, I've left out slots 3 and 4 just as an example. Now, modify your mtx-changer script and comment out all the lines in the list) case by putting a # in column 1. Then add the following two lines:
cat <absolute-path>/volume-list exit 0
so that the whole case looks like:
list) # # commented out lines cat <absolute-path>/volume-list exit 0 ;;
where you replace <absolute-path> with the full path to the volume-list file. Then using the console, you enter the following command:
label barcodes
and Bacula will proceed to mount the autochanger Volumes in the list and label them with the Volume names you have supplied. Bacula will think that the list was provided by the autochanger barcodes, but in reality, it was you who supplied the <barcodes>.
If it seems to work, when it finishes, enter:
list volumes
and you should see all the volumes nicely created.
You may want to backup laptops or portables that are not always connected to the network. If you are using DHCP to assign an IP address to those machines when they connect, you will need to use the Dynamic Update capability of DNS to assign a name to those machines that can be used in the Address field of the Client resource in the Director's conf file.
At some point, you may want to be absent for a week or two and you want to make sure Bacula has enough tape left so that the backups will complete. You start by doing a list volumes in the Console program:
{\bf list volumes} Using default Catalog name=BackupDB DB=bacula Pool: Default +---------+---------------+-----------+-----------+----------------+- | MediaId | VolumeName | MediaType | VolStatus | VolBytes | +---------+---------------+-----------+-----------+----------------+- | 23 | DLT-30Nov02 | DLT8000 | Full | 54,739,278,128 | | 24 | DLT-21Dec02 | DLT8000 | Full | 56,331,524,629 | | 25 | DLT-11Jan03 | DLT8000 | Full | 67,863,514,895 | | 26 | DLT-02Feb03 | DLT8000 | Full | 63,439,314,216 | | 27 | DLT-03Mar03 | DLT8000 | Full | 66,022,754,598 | | 28 | DLT-04Apr03 | DLT8000 | Full | 60,792,559,924 | | 29 | DLT-28Apr03 | DLT8000 | Full | 62,072,494,063 | | 30 | DLT-17May03 | DLT8000 | Full | 65,901,767,839 | | 31 | DLT-07Jun03 | DLT8000 | Used | 56,558,490,015 | | 32 | DLT-28Jun03 | DLT8000 | Full | 64,274,871,265 | | 33 | DLT-19Jul03 | DLT8000 | Full | 64,648,749,480 | | 34 | DLT-08Aug03 | DLT8000 | Full | 64,293,941,255 | | 35 | DLT-24Aug03 | DLT8000 | Append | 9,999,216,782 | +---------+---------------+-----------+-----------+----------------+
Note, I have truncated the output for presentation purposes. What is significant, is that I can see that my current tape has almost 10 Gbytes of data, and that the average amount of data I get on my tapes is about 60 Gbytes. So if I go on vacation now, I don't need to worry about tape capacity (at least not for short absences).
Equally significant is the fact that I did go on vacation the 28th of June 2003, and when I did the list volumes command, my current tape at that time, DLT-07Jun03 MediaId 31, had 56.5 Gbytes written. I could see that the tape would fill shortly. Consequently, I manually marked it as Used and replaced it with a fresh tape that I labeled as DLT-28Jun03, thus assuring myself that the backups would all complete without my intervention.
This tip was submitted by Marc Brueckner who wasn't sure of the case of some of his files on Win32, which is case insensitive. The problem is that Bacula thinks that /UNIMPORTANT FILES is different from /Unimportant Files. Marc was aware that the file exclusion permits wild-cards. So, he specified:
"/[Uu][Nn][Ii][Mm][Pp][Oo][Rr][Tt][Aa][Nn][Tt] [Ff][Ii][Ll][Ee][Ss]"
As a consequence, the above exclude works for files of any case.
Please note that this works only in Bacula Exclude statement and not in Include.
This tip also comes from Marc Brueckner. (Note, this tip is probably outdated by the addition of ClientRunBeforJob and ClientRunAfterJob Job records, but the technique still could be useful.) First I thought the "Run Before Job" statement in the Job-resource is for executing a script on the remote machine(the machine to be backed up). It could be usefull to execute scripts on the remote machine e.g. for stopping databases or other services while doing the backup. (Of course I have to start the services again when the backup has finished) I found the following solution: Bacula could execute scrips on the remote machine by using ssh. The authentication is done automatically using a private key. First You have to generate a keypair. I've done this by:
ssh-keygen -b 4096 -t dsa -f Bacula_key
This statement may take a little time to run. It creates a public/private key pair with no passphrase. You could save the keys in /etc/bacula. Now you have two new files : Bacula_key which contains the private key and Bacula_key.pub which contains the public key.
Now you have to append the Bacula_key.pub file to the file authorized_keys in the \root\.ssh directory of the remote machine. Then you have to add (or uncomment) the line
AuthorizedKeysFile %h/.ssh/authorized_keys
to the sshd_config file on the remote machine. Where the %h stands for the home-directory of the user (root in this case).
Assuming that your sshd is already running on the remote machine, you can now enter the following on the machine where Bacula runs:
ssh -i Bacula_key -l root "ls -la"
This should execute the "ls -la" command on the remote machine.
Now you could add lines like the following to your Director's conf file:
... Run Before Job = ssh -i /etc/bacula/Bacula_key 192.168.1.1 \ "/etc/init.d/database stop" Run After Job = ssh -i /etc/bacula/Bacula_key 192.168.1.1 \ "/etc/init.d/database start" ...
Even though Bacula version 1.32 has a ClientRunBeforeJob, the ssh method still could be useful for updating all the Bacula clients on several remote machines in a single script.
This tip comes from Phil Stracchino.
If you decide to blow away your catalog and start over, the simplest way to re-add all your prelabeled tapes with a minimum of fuss (provided you don't care about the data on the tapes) is to add the tape labels using the console add command, then go into the catalog and manually set the VolStatus of every tape to Recycle.
The SQL command to do this is very simple:
update Media set VolStatus = "Recycle";
Bacula will then ignore the data already stored on the tapes and just re-use each tape without further objection.
This tip comes from Volker Sauer.
Note, this tip was given prior to implementation of ACLs in Bacula (version 1.34.5). It is left here because dumping/displaying ACLs can still be useful in testing/verifying that Bacula is backing up and restoring your ACLs properly. Please see the aclsupport FileSet option in the configuration chapter of this manual.
For example, you could dump the ACLs to a file with a script similar to the following:
#!/bin/sh BACKUP_DIRS="/foo /bar" STORE_ACL=/root/acl-backup umask 077 for i in $BACKUP_DIRS; do cd $i /usr/bin/getfacl -R --skip-base .>$STORE_ACL/${i//\//_} done
Then use Bacula to backup /root/acl-backup.
The ACLs could be restored using Bacula to the /root/acl-backup file, then restored to your system using:
setfacl --restore/root/acl-backup
This tip was provided by Alexander Kuehn.
Bacula is a really nice backup program except that the manual tape changing requires user interaction with the bacula console.
Fortunately I can fix this. NOTE!!! This suggestion applies for people who do *NOT* have tape autochangers and must change tapes manually.!!!!!
Bacula supports a variety of tape changers through the use of mtx-changer scripts/programs. This highly flexible approach allowed me to create this shell script which does the following: Whenever a new tape is required it sends a mail to the operator to insert the new tape. Then it waits until a tape has been inserted, sends a mail again to say thank you and let's bacula continue it's backup. So you can schedule and run backups without ever having to log on or see the console. To make the whole thing work you need to create a Device resource which looks something like this ("Archive Device", "Maximum Changer Wait", "Media Type" and "Label media" may have different values):
Device { Name=DDS3 Archive Device = # use yours not mine! ;)/dev/nsa0 Changer Device = # not really required/dev/nsa0 Changer Command = "# use this (maybe change the path)! /usr/local/bin/mtx-changer %o %a %S" Maximum Changer Wait = 3d # 3 days in seconds AutomaticMount = yes; # mount on start AlwaysOpen = yes; # keep device locked Media Type = DDS3 # it's just a name RemovableMedia = yes; # Offline On Unmount = Yes; # keep this too Label media = Yes; # }
As the script has to emulate the complete wisdom of a mtx-changer it has an internal "database" containing where which tape is stored, you can see this on the following line:
labels="VOL-0001 VOL-0002 VOL-0003 VOL-0004 VOL-0005 VOL-0006 VOL-0007 VOL-0008 VOL-0009 VOL-0010 VOL-0011 VOL-0012"
The above should be all on one line, and it effectivly tells Bacula that volume "VOL-0001" is located in slot 1 (which is our lowest slot), that volume "VOL-0002" is located in slot 2 and so on.. The script also maintains a logfile (/var/log/mtx.log) where you can monitor its operation.
Bacula can run multiple concurrent jobs, but the default configuration files are not set to do so. Using the Maximum Concurrent Jobs directive, you have a lot of control over how many jobs can run at the same time, and which jobs can run simultaneously. The downside is that it can be a bit tricky to set it up for the first time as you need to set the concurrency in at least five different places.
The Director, the File daemon, and the Storage daemon each have a Maximum Concurrent Jobs directive that determines overall number of concurrent jobs the daemon will run. The default is one for the Director and ten for both the File daemon and the Storage daemon, so assuming you will not be running more than ten concurrent jobs, the only changes that are needed are in the Director's conf file (bacula-dir.conf).
Within the Director's configuration file, Maximum Concurrent Jobs can be set in the Direct, Job, Client, and Storage resources. Each one must be set properly, according to your needs, otherwise your jobs may be run one at a time.
For example, if you want two different jobs to run simultaneously backing up the same Client to the same Storage device, they will run concurrentl only if you have set Maximum Concurrent Jobs greater than one in the Director resource, the Client resource, and the Storage resource in bacula-dir.conf.
We recommend that you carefully test your multiple concurrent backup including doing thorough restore testing before you put it into production.
Below is a super stripped down bacula-dir.conf file showing you the four places where the the file has been modified to allow the same job NightlySave to run up to four times concurrently. The change to the Job resource is not necessary if you want different Jobs to run at the same time, which is the normal case.
# # Bacula Director Configuration file -- bacula-dir.conf # Director { Name = rufus-dir Maximum Concurrent Jobs = 4 ... } Job { Name = "NightlySave" Maximum Concurrent Jobs = 4 Client = rufus-fd Storage = File ... } Client { Name = rufus-fd Maximum Concurrent Jobs = 4 ... } Storage { Name = File Maximum Concurrent Jobs = 4 ... }
This document describes the utility programs written to aid Bacula users and developers in dealing with Volumes external to Bacula.
Starting with version 1.27, each of the following programs requires a valid Storage daemon configuration file (actually, the only part of the configuration file that these programs need is the Device resource definitions). This permits the programs to find the configuration parameters for your archive device (generally a tape drive). By default, they read bacula-sd.conf in the current directory, but you may specify a different configuration file using the -c option.
Each of these programs require a device-name where the Volume can be found. In the case of a tape, this is the physical device name such as /dev/nst0 or /dev/rmt/0ubn depending on your system. For the program to work, it must find the identical name in the Device resource of the configuration file. See below for specifying Volume names.
Please note that if you have Bacula running and you ant to use one of these programs, you will either need to stop the Storage daemon, or unmount any tape drive you want to use, otherwise the drive will busy because Bacula is using it.
If you are attempting to read or write an archive file rather than a tape, the device-name should be the full path to the archive location including the filename. The filename (last part of the specification) will be stripped and used as the Volume name, and the path (first part before the filename) must have the same entry in the configuration file. So, the path is equivalent to the archive device name, and the filename is equivalent to the volume name.
In general, you must specify the Volume name to each of the programs below (with the exception of btape). The best method to do so is to specify a bootstrap file on the command line with the -b option. As part of the bootstrap file, you will then specify the Volume name or Volume names if more than one volume is needed. For example, suppose you want to read tapes tape1 and tape2. First construct a bootstrap file named say, list.bsr which contains:
Volume=test1|test2
where each Volume is separated by a vertical bar. Then simply use:
./bls -b list.bsr /dev/nst0
In the case of Bacula Volumes that are on files, you may simply append volumes as follows:
./bls /tmp/test1\|test2
where the backslash (\) was necessary as a shell escape to permit entering the vertical bar (|).
And finally, if you feel that specifying a Volume name is a bit complicated with a bootstrap file, you can use the -V option (on all programs except bcopy) to specify one or more Volume names separated by the vertical bar (|). For example,
./bls -V Vol001 /dev/nst0
You may also specify an asterisk (*) to indicate that the program should accept any volume. For example:
./bls -V* /dev/nst0
bls can be used to do an ls type listing of a Bacula tape or file. It is called:
Usage: bls [options] <device-name> -b <file> specify a bootstrap file -c <file> specify a config file -d <level> specify debug level -e <file> exclude list -i <file> include list -j list jobs -k list blocks (no j or k option) list saved files -L dump label -p proceed inspite of errors -v be verbose -V specify Volume names (separated by |) -? print this message
For example, to list the contents of a tape:
./bls -V Volume-name /dev/nst0
Or to list the contents of a file:
./bls /tmp/Volume-name or ./bls -V Volume-name /tmp
Note that, in the case of a file, the Volume name becomes the filename, so in the above example, you will replace the xxx with the name of the volume (file) you wrote.
Normally if no options are specified, bls will produce the equivalent output to the ls -l command for each file on the tape. Using other options listed above, it is possible to display only the Job records, only the tape blocks, etc. For example:
./bls /tmp/File002 bls: butil.c:148 Using device: /tmp drwxrwxr-x 3 k k 4096 02-10-19 21:08 /home/kern/bacula/k/src/dird/ drwxrwxr-x 2 k k 4096 02-10-10 18:59 /home/kern/bacula/k/src/dird/CVS/ -rw-rw-r-- 1 k k 54 02-07-06 18:02 /home/kern/bacula/k/src/dird/CVS/Root -rw-rw-r-- 1 k k 16 02-07-06 18:02 /home/kern/bacula/k/src/dird/CVS/Repository -rw-rw-r-- 1 k k 1783 02-10-10 18:59 /home/kern/bacula/k/src/dird/CVS/Entries -rw-rw-r-- 1 k k 97506 02-10-18 21:07 /home/kern/bacula/k/src/dird/Makefile -rw-r--r-- 1 k k 3513 02-10-18 21:02 /home/kern/bacula/k/src/dird/Makefile.in -rw-rw-r-- 1 k k 4669 02-07-06 18:02 /home/kern/bacula/k/src/dird/README-config -rw-r--r-- 1 k k 4391 02-09-14 16:51 /home/kern/bacula/k/src/dird/authenticate.c -rw-r--r-- 1 k k 3609 02-07-07 16:41 /home/kern/bacula/k/src/dird/autoprune.c -rw-rw-r-- 1 k k 4418 02-10-18 21:03 /home/kern/bacula/k/src/dird/bacula-dir.conf ... -rw-rw-r-- 1 k k 83 02-08-31 19:19 /home/kern/bacula/k/src/dird/.cvsignore bls: Got EOF on device /tmp 84 files found.
If you are listing a Volume to determine what Jobs to restore, normally the -j option provides you with most of what you will need as long as you don't have multiple clients. For example,
./bls -j -V Test1 -c stored.conf DDS-4 bls: butil.c:258 Using device: "DDS-4" for reading. 11-Jul 11:54 bls: Ready to read from volume "Test1" on device "DDS-4" (/dev/nst0). Volume Record: File:blk=0:1 SessId=4 SessTime=1121074625 JobId=0 DataLen=165 Begin Job Session Record: File:blk=0:2 SessId=4 SessTime=1121074625 JobId=1 Level=F Type=B Begin Job Session Record: File:blk=0:3 SessId=5 SessTime=1121074625 JobId=5 Level=F Type=B Begin Job Session Record: File:blk=0:6 SessId=3 SessTime=1121074625 JobId=2 Level=F Type=B Begin Job Session Record: File:blk=0:13 SessId=2 SessTime=1121074625 JobId=4 Level=F Type=B End Job Session Record: File:blk=0:99 SessId=3 SessTime=1121074625 JobId=2 Level=F Type=B Files=168 Bytes=1,732,978 Errors=0 Status=T End Job Session Record: File:blk=0:101 SessId=2 SessTime=1121074625 JobId=4 Level=F Type=B Files=168 Bytes=1,732,978 Errors=0 Status=T End Job Session Record: File:blk=0:108 SessId=5 SessTime=1121074625 JobId=5 Level=F Type=B Files=168 Bytes=1,732,978 Errors=0 Status=T End Job Session Record: File:blk=0:109 SessId=4 SessTime=1121074625 JobId=1 Level=F Type=B Files=168 Bytes=1,732,978 Errors=0 Status=T 11-Jul 11:54 bls: End of Volume at file 1 on device "DDS-4" (/dev/nst0), Volume "Test1" 11-Jul 11:54 bls: End of all volumes.
shows a full save followed by two incremental saves.
Adding the -v option will display virtually all information that is available for each record:
Normally, except for debugging purposes, you will not need to list Bacula blocks (the "primitive" unit of Bacula data on the Volume). However, you can do so with:
./bls -k /tmp/File002 bls: butil.c:148 Using device: /tmp Block: 1 size=64512 Block: 2 size=64512 ... Block: 65 size=64512 Block: 66 size=19195 bls: Got EOF on device /tmp End of File on device
By adding the -v option, you can get more information, which can be useful in knowing what sessions were written to the volume:
./bls -k -v /tmp/File002 Volume Label: Id : Bacula 0.9 mortal VerNo : 10 VolName : File002 PrevVolName : VolFile : 0 LabelType : VOL_LABEL LabelSize : 147 PoolName : Default MediaType : File PoolType : Backup HostName : Date label written: 2002-10-19 at 21:16 Block: 1 blen=64512 First rec FI=VOL_LABEL SessId=1 SessTim=1035062102 Strm=0 rlen=147 Block: 2 blen=64512 First rec FI=6 SessId=1 SessTim=1035062102 Strm=DATA rlen=4087 Block: 3 blen=64512 First rec FI=12 SessId=1 SessTim=1035062102 Strm=DATA rlen=5902 Block: 4 blen=64512 First rec FI=19 SessId=1 SessTim=1035062102 Strm=DATA rlen=28382 ... Block: 65 blen=64512 First rec FI=83 SessId=1 SessTim=1035062102 Strm=DATA rlen=1873 Block: 66 blen=19195 First rec FI=83 SessId=1 SessTim=1035062102 Strm=DATA rlen=2973 bls: Got EOF on device /tmp End of File on device
Armed with the SessionId and the SessionTime, you can extract just about anything.
If you want to know even more, add a second -v to the command line to get a dump of every record in every block.
./bls -k -v -v /tmp/File002 bls: block.c:79 Dump block 80f8ad0: size=64512 BlkNum=1 Hdrcksum=b1bdfd6d cksum=b1bdfd6d bls: block.c:92 Rec: VId=1 VT=1035062102 FI=VOL_LABEL Strm=0 len=147 p=80f8b40 bls: block.c:92 Rec: VId=1 VT=1035062102 FI=SOS_LABEL Strm=-7 len=122 p=80f8be7 bls: block.c:92 Rec: VId=1 VT=1035062102 FI=1 Strm=UATTR len=86 p=80f8c75 bls: block.c:92 Rec: VId=1 VT=1035062102 FI=2 Strm=UATTR len=90 p=80f8cdf bls: block.c:92 Rec: VId=1 VT=1035062102 FI=3 Strm=UATTR len=92 p=80f8d4d bls: block.c:92 Rec: VId=1 VT=1035062102 FI=3 Strm=DATA len=54 p=80f8dbd bls: block.c:92 Rec: VId=1 VT=1035062102 FI=3 Strm=MD5 len=16 p=80f8e07 bls: block.c:92 Rec: VId=1 VT=1035062102 FI=4 Strm=UATTR len=98 p=80f8e2b bls: block.c:92 Rec: VId=1 VT=1035062102 FI=4 Strm=DATA len=16 p=80f8ea1 bls: block.c:92 Rec: VId=1 VT=1035062102 FI=4 Strm=MD5 len=16 p=80f8ec5 bls: block.c:92 Rec: VId=1 VT=1035062102 FI=5 Strm=UATTR len=96 p=80f8ee9 bls: block.c:92 Rec: VId=1 VT=1035062102 FI=5 Strm=DATA len=1783 p=80f8f5d bls: block.c:92 Rec: VId=1 VT=1035062102 FI=5 Strm=MD5 len=16 p=80f9668 bls: block.c:92 Rec: VId=1 VT=1035062102 FI=6 Strm=UATTR len=95 p=80f968c bls: block.c:92 Rec: VId=1 VT=1035062102 FI=6 Strm=DATA len=32768 p=80f96ff bls: block.c:92 Rec: VId=1 VT=1035062102 FI=6 Strm=DATA len=32768 p=8101713 bls: block.c:79 Dump block 80f8ad0: size=64512 BlkNum=2 Hdrcksum=9acc1e7f cksum=9acc1e7f bls: block.c:92 Rec: VId=1 VT=1035062102 FI=6 Strm=contDATA len=4087 p=80f8b40 bls: block.c:92 Rec: VId=1 VT=1035062102 FI=6 Strm=DATA len=31970 p=80f9b4b bls: block.c:92 Rec: VId=1 VT=1035062102 FI=6 Strm=MD5 len=16 p=8101841 ...
Normally, you will restore files by running a Restore Job from the Console program. However, bextract can be used to extract a single file or a list of files from a Bacula tape or file. In fact, bextract can be a useful tool to restore files to an empty system assuming you are able to boot, you have statically linked bextract and you have an appropriate bootstrap file.
It is called:
Usage: bextract [-d debug_level] <device-name> <directory-to-store-files> -b <file> specify a bootstrap file -dnn set debug level to nn -e <file> exclude list -i <file> include list -p proceed inspite of I/O errors -V specify Volume names (separated by |) -? print this message
where device-name is the Archive Device (raw device name or full filename) of the device to be read, and directory-to-store-files is a path prefix to prepend to all the files restored.
NOTE: On Windows systems, if you specify a prefix of say d:/tmp, any file that would have been restored to c:/My Documents will be restored to d:/tmp/My Documents. That is, the original drive specification will be stripped. If no prefix is specified, the file will be restored to the original drive.
Using the -e option, you can specify a file containing a list of files to be excluded. Wildcards can be used in the exclusion list. This option will normally be used in conjunction with the -i option (see below). Both the -e and the -i options may be specified at the same time as the -b option. The bootstrap filters will be applied first, then the include list, then the exclude list.
Likewise, and probably more importantly, with the -i option, you can specify a file that contains a list (one file per line) of files and directories to include to be restored. The list must contain the full filename with the path. If you specify a path name only, all files and subdirectories of that path will be restored. If you specify a line containing only the filename (e.g. my-file.txt) it probably will not be extracted because you have not specified the full path.
For example, if the file include-list contains:
/home/kern/bacula /usr/local/bin
Then the command:
./bextract -i include-list -V Volume /dev/nst0 /tmp
will restore from the Bacula archive /dev/nst0 all files and directories in the backup from /home/kern/bacula and from /usr/local/bin. The restored files will be placed in a file of the original name under the directory /tmp (i.e. /tmp/home/kern/bacula/... and /tmp/usr/local/bin/...).
The -b option is used to specify a bootstrap file containing the information needed to restore precisely the files you want. Specifying a bootstrap file is optional but recommended because it gives you the most control over which files will be restored. For more details on the bootstrap file, please see Restoring Files with the Bootstrap File chapter of this document. Note, you may also use a bootstrap file produced by the restore command. For example:
./bextract -b bootstrap-file /dev/nst0 /tmp
The bootstrap file allows detailed specification of what files you want restored (extracted). You may specify a bootstrap file and include and/or exclude files at the same time. The bootstrap conditions will first be applied, and then each file record seen will be compared to the include and exclude lists.
If you wish to extract files that span several Volumes, you can specify the Volume names in the bootstrap file or you may specify the Volume names on the command line by separating them with a vertical bar. See the section above under the bls program entitled Listing Multiple Volumes for more information. The same techniques apply equally well to the bextract program.
The bscan program can be used to re-create a database (catalog) from the backup information written to one or more Volumes. This is normally needed only if one or more Volumes have been pruned or purged from your catalog so that the records on the Volume are no longer in the catalog.
With some care, it can also be used to synchronize your existing catalog with a Volume. Although we have never seen a case of bscan damaging a catalog, since bscan modifies your catalog, we recommend that you do a simple ASCII backup of your database before running bscan just to be sure. See Compacting Your Database.
bscan can also be useful in a disaster recovery situation, after the loss of a hard disk, if you do not have a valid bootstrap file for reloading your system, or if a Volume has been recycled but not overwritten, you can use bscan to re-create your database, which can then be used to restore your system or a file to its previous state.
It is called:
Usage: bscan [options] <bacula-archive> -b bootstrap specify a bootstrap file -c <file> specify configuration file -d <nn> set debug level to nn -m update media info in database -n <name> specify the database name (default bacula) -u <user> specify database user name (default bacula) -P <password> specify database password (default none) -h <host> specify database host (default NULL) -p proceed inspite of I/O errors -r list records -s synchronize or store in database -v verbose -V <Volumes> specify Volume names (separated by |) -w <dir> specify working directory (default from conf file) -? print this message
If you are using MySQL or PostgreSQL, there is no need to supply a working directory since in that case, bscan knows where the databases are. However, if you have provided security on your database, you may need to supply either the database name (-b option), the user name (-u option), and/or the password (-p) options.
As an example, let's suppose that you did a backup to Volumes "Vol001" and "Vol002", then sometime later all records of one or both those Volumes were pruned or purged from the database. By using bscan you can recreate the catalog entries for those Volumes and then use the restore command in the Console to restore whatever you want. A command something like:
bscan -c bacula-sd.conf -v -V Vol001\|Vol002 /dev/nst0
will give you an idea of what is going to happen without changing your catalog. Of course, you may need to change the path to the Storage daemon's conf file, the Volume name, and your tape (or disk) device name. This command must read the entire tape, so if it has a lot of data, it may take a long time, and thus you might want to immediately use the command listed below. Note, if you are writing to a disk file, replace the device name with the path to the directory that contains the Volumes. This must correspond to the Archive Device in the conf file.
Then to actually write or store the records in the catalog, add the -s option as follows:
bscan -s -m -c bacula-sd.conf -v -V Vol001\|Vol002 /dev/nst0
When writing to the database, if bscan finds existing records, it will generally either update them if something is wrong or leave them alone. Thus if the Volumes you are scanning are all or partially in the catalog already, no harm will be done to that existing data. Any missing data will simply be added.
If you have multiple tapes, you should scan them with:
bscan -s -m -c bacula-sd.conf -v -V Vol001\|Vol002\|Vol003 /dev/nst0
You should, always try to specify the tapes in the order they are written. However, bscan can handle scanning tapes that are not sequential. Any incomplete records at the end of the tape will simply be ignored in that case. If you are simply reparing an existing catalog, this may be OK, but if you are creating a new catalog from scratch, it will leave your database in an incorrect state. If you do not specify all necessary Volumes on a single bscan command, bscan will not be able to correctly restore the records that span two volumes. In other words, it is much better to specify two or three volumes on a single bscan command rather than run bscan two or three times, each with a single volume.
Note, the restoration process using bscan is not identical to the original creation of the catalog data. This is because certain non-essential data such as volume reads, volume mounts, etc is not stored on the Volume, and thus is not restored by bscan. The results of bscanning are, however, perfectly valid, and will permit restoration of any or all the files in the catalog using the normal Bacula console commands.
If you wish to compare the contents of a Volume to an existing catalog without changing the catalog, you can safely do so if and only if you do not specify either the -m or the -s options. However, at this time (Bacula version 1.26), the comparison routines are not as good or as thorough as they should be, so we don't particularly recommend this mode other than for testing.
This is the mode for which bscan is most useful. You can either bscan into a freshly created catalog, or directly into your existing catalog (after having made an ASCII copy as described above). Normally, you should start with a freshly created catalog that contains no data.
Starting with a single Volume named TestVolume1, you run a command such as:
./bscan -V TestVolume1 -v -s -m -c bacula-sd.conf /dev/nst0
If there is more than one volume, simply append it to the first one separating it with a vertical bar. You may need to precede the vertical bar with a forward slash escape the shell -- e.g. TestVolume1\|TestVolume2 . The -v option was added for verbose output (this can be omitted if desired). The -s option that tells bscan to store information in the database. The physical device name /dev/nst0 is specified after all the options.
For example, after having done a full backup of a directory, then two incrementals, I reinitialized the SQLite database as described above, and using the bootstrap.bsr file noted above, I entered the following command:
./bscan -b bootstrap.bsr -v -s -c bacula-sd.conf /dev/nst0
which produced the following output:
bscan: bscan.c:182 Using Database: bacula, User: bacula bscan: bscan.c:673 Created Pool record for Pool: Default bscan: bscan.c:271 Pool type "Backup" is OK. bscan: bscan.c:632 Created Media record for Volume: TestVolume1 bscan: bscan.c:298 Media type "DDS-4" is OK. bscan: bscan.c:307 VOL_LABEL: OK for Volume: TestVolume1 bscan: bscan.c:693 Created Client record for Client: Rufus bscan: bscan.c:769 Created new JobId=1 record for original JobId=2 bscan: bscan.c:717 Created FileSet record "Kerns Files" bscan: bscan.c:819 Updated Job termination record for new JobId=1 bscan: bscan.c:905 Created JobMedia record JobId 1, MediaId 1 bscan: Got EOF on device /dev/nst0 bscan: bscan.c:693 Created Client record for Client: Rufus bscan: bscan.c:769 Created new JobId=2 record for original JobId=3 bscan: bscan.c:708 Fileset "Kerns Files" already exists. bscan: bscan.c:819 Updated Job termination record for new JobId=2 bscan: bscan.c:905 Created JobMedia record JobId 2, MediaId 1 bscan: Got EOF on device /dev/nst0 bscan: bscan.c:693 Created Client record for Client: Rufus bscan: bscan.c:769 Created new JobId=3 record for original JobId=4 bscan: bscan.c:708 Fileset "Kerns Files" already exists. bscan: bscan.c:819 Updated Job termination record for new JobId=3 bscan: bscan.c:905 Created JobMedia record JobId 3, MediaId 1 bscan: Got EOF on device /dev/nst0 bscan: bscan.c:652 Updated Media record at end of Volume: TestVolume1 bscan: bscan.c:428 End of Volume. VolFiles=3 VolBlocks=57 VolBytes=10,027,437
The key points to note are that bscan prints a line when each major record is created. Due to the volume of output, it does not print a line for each file record unless you supply the -v option twice or more on the command line.
In the case of a Job record, the new JobId will not normally be the same as the original Jobid. For example, for the first JobId above, the new JobId is 1, but the original JobId is 2. This is nothing to be concerned about as it is the normal nature of databases. bscan will keep everything straight.
Although bscan claims that it created a Client record for Client: Rufus three times, it was actually only created the first time. This is normal.
You will also notice that it read an end of file after each Job (Got EOF on device ...). Finally the last line gives the total statistics for the bscan.
If you had added a second -v option to the command line, Bacula would have been even more verbose, dumping virtually all the details of each Job record it encountered.
Now if you start Bacula and enter a list jobs command to the console program, you will get:
+-------+----------+------------------+------+-----+----------+----------+---------+ | JobId | Name | StartTime | Type | Lvl | JobFiles | JobBytes | JobStat | +-------+----------+------------------+------+-----+----------+----------+---------+ | 1 | kernsave | 2002-10-07 14:59 | B | F | 84 | 4180207 | T | | 2 | kernsave | 2002-10-07 15:00 | B | I | 15 | 2170314 | T | | 3 | kernsave | 2002-10-07 15:01 | B | I | 33 | 3662184 | T | +-------+----------+------------------+------+-----+----------+----------+---------+
which corresponds virtually identically with what the database contained before it was re-initialized and restored with bscan. All the Jobs and Files found on the tape are restored including most of the Media record. The Volume (Media) records restored will be marked as Full so that they cannot be rewritten without operator intervention.
It should be noted that bscan cannot restore a database to the exact condition it was in previously because a lot of the less important information contained in the database is not saved to the tape. Nevertheless, the reconstruction is sufficiently complete, that you can run restore against it and get valid results.
If the Storage daemon crashes during a backup Job, the catalog will not be properly updated for the Volume being used at the time of the crash. This means that the Storage daemon will have written say 20 files on the tape, but the catalog record for the Volume indicates only 19 files.
Bacula refuses to write on a tape that contains a different number of files from what is in the catalog. To correct this situation, you may run a bscan with the -m option (but without the -s option) to update only the final Media record for the Volumes read.
If you use bscan to enter the contents of the Volume into an existing catalog, you should be aware that the records you entered may be immediately pruned during the next job, particularly if the Volume is very old or had been previously purged. To avoid this, after running bscan, you can manually set the volume status (VolStatus) to Read-Only by using the update command in the catalog. This will allow you to restore from the volume without having it immediately purged. When you have restored and backed up the data, you can reset the VolStatus to Used and the Volume will be purged from the catalog.
The bcopy program can be used to copy one Bacula archive file to another. For example, you may copy a tape to a file, a file to a tape, a file to a file, or a tape to a tape. For tape to tape, you will need two tape drives. (a later version is planned that will buffer it to disk). In the process of making the copy, no record of the information written to the new Volume is stored in the catalog. This means that the new Volume, though it contains valid backup data, cannot be accessed directly from existing catalog entries. If you wish to be able to use the Volume with the Console restore command, for example, you must first bscan the new Volume into the catalog.
Usage: bcopy [-d debug_level] <input-archive> <output-archive> -b bootstrap specify a bootstrap file -c <file> specify configuration file -dnn set debug level to nn -i specify input Volume names (separated by |) -o specify output Volume names (separated by |) -p proceed inspite of I/O errors -v verbose -w dir specify working directory (default /tmp) -? print this message
By using a bootstrap file, you can copy parts of a Bacula archive file to another archive.
One of the objectives of this program is to be able to recover as much data as possible from a damaged tape. However, the current version does not yet have this feature.
As this is a new program, any feedback on its use would be appreciated. In addition, I only have a single tape drive, so I have never been able to test this program with two tape drives.
This program permits a number of elementary tape operations via a tty command interface. The test command, described below, can be very useful for testing older tape drive compatibility problems. Aside from initial testing of tape drive compatibility with Bacula, btape will be mostly used by developers writing new tape drivers.
btape can be dangerous to use with existing Bacula tapes because it will relabel a tape or write on the tape if so requested regardless that the tape may contain valuable data, so please be careful and use it only on blank tapes.
To work properly, btape needs to read the Storage daemon's configuration file. As a default, it will look for bacula-sd.conf in the current directory. If your configuration file is elsewhere, please use the -c option to specify where.
The physical device name must be specified on the command line, and this same device name must be present in the Storage daemon's configuration file read by btape
Usage: btape [-c config_file] [-d debug_level] [device_name] -c <file> set configuration file to file -dnn set debug level to nn -s turn off signals -t open the default tape device -? print this message.
An important reason for this program is to ensure that a Storage daemon configuration file is defined so that Bacula will correctly read and write tapes.
It is highly recommended that you run the test command before running your first Bacula job to ensure that the parameters you have defined for your storage device (tape drive) will permit Bacula to function properly. You only need to mount a blank tape, enter the command, and the output should be reasonably self explanatory. Please see the Tape Testing Chapter of this manual for the details.
The full list of commands are:
Command Description ======= =========== bsf backspace file bsr backspace record cap list device capabilities clear clear tape errors eod go to end of Bacula data for append test General test Bacula tape functions eom go to the physical end of medium fill fill tape, write onto second volume unfill read filled tape fsf forward space a file fsr forward space a record help print this command label write a Bacula label to the tape load load a tape quit quit btape rd read tape readlabel read and print the Bacula tape label rectest test record handling functions rewind rewind the tape scan read tape block by block to EOT and report status print tape status test test a tape for compatibility with Bacula weof write an EOF on the tape wr write a single record of 2048 bytes
The most useful commands are:
The readlabel command can be used to display the details of a Bacula tape label. This can be useful if the physical tape label was lost or damaged.
In the event that you want to relabel a Bacula, you can simply use the label command which will write over any existing label. However, please note for labeling tapes, we recommend that you use the label command in the Console program since it will never overwrite a valid Bacula tape.
The following programs are general utility programs and in general do not need a configuration file nor a device name.
bsmtp is a simple mail transport program that permits more flexibility than the standard mail programs typically found on Unix systems. It can even be used on Windows machines.
It is called:
Usage: bsmtp [-f from] [-h mailhost] [-s subject] [-c copy] [recipient ...] -c set the Cc: field -dnn set debug level to nn -f set the From: field -h use mailhost:port as the bsmtp server -s set the Subject: field -? print this message.
If the -f option is not specified, bsmtp will use your userid. If the option is not specified bsmtp will use the value in the environment variable bsmtpSERVER or if there is none localhost. By default port 25 is used.
recipients is a space separated list of email recipients.
The body of the email message is read from standard input.
An example of the use of bsmtp would be to put the following statement in the Messages resource of your bacula-dir.conf file. Note, these commands should appear on a single line each.
mailcommand = "/home/bacula/bin/bsmtp -h mail.domain.com -f \"\(Bacula\) %r\" -s \"Bacula: %t %e of %c %l\" %r" operatorcommand = "/home/bacula/bin/bsmtp -h mail.domain.com -f \"\(Bacula\) %r\" -s \"Bacula: Intervention needed for %j\" %r"
Where you replace /home/bacula/bin with the path to your Bacula binary directory, and you replace mail.domain.com with the fully qualified name of your bsmtp (email) server, which normally listens on port 25. For more details on the substitution characters (e.g. %r) used in the above line, please see the documentation of the MailCommand in the Messages Resource chapter of this manual.
It is HIGHLY recommended that you test one or two cases by hand to make sure that the mailhost that you specified is correct and that it will accept your email requests. Since bsmtp always uses a TCP connection rather than writing in the spool file, you may find that your from address is being rejected because it does not contain a valid domain, or because your message is caught in your spam filtering rules. Generally, you should specify a fully qualified domain name in the from field, and depending on whether your bsmtp gateway is Exim or Sendmail, you may need to modify the syntax of the from part of the message. Please test.
When running bsmtp by hand, you will need to terminate the message by entering a ctl-d in column 1 of the last line.
dbcheck is a simple program that will search for inconsistencies in your database, and optionally fix them. The dbcheck program can be found in the <bacula-source>/src/tools directory of the source distribution. Though it is built with the make process, it is not normally "installed".
It is called:
Usage: dbcheck [-c config] [-C catalog name] [-d debug_level] [] -b batch mode -C catalog name in the director conf file -c director conf filename -dnn set debug level to nn -f fix inconsistencies -v verbose -? print this message
If the -c option is given with the Director's conf file, there is no need to enter any of the command line arguments, in particular the working directory as dbcheck will read them from the file.
If the -f option is specified, dbcheck will repair (fix) the inconsistencies it finds. Otherwise, it will report only.
If the -b option is specified, dbcheck will run in batch mode, and it will proceed to examine and fix (if -f is set) all programmed inconsistency checks. If the -b option is not specified, dbcheck will enter interactive mode and prompt with the following:
Hello, this is the database check/correct program. Please select the function you want to perform. 1) Toggle modify database flag 2) Toggle verbose flag 3) Repair bad Filename records 4) Repair bad Path records 5) Eliminate duplicate Filename records 6) Eliminate duplicate Path records 7) Eliminate orphaned Jobmedia records 8) Eliminate orphaned File records 9) Eliminate orphaned Path records 10) Eliminate orphaned Filename records 11) Eliminate orphaned FileSet records 12) Eliminate orphaned Client records 13) Eliminate orphaned Job records 14) Eliminate all Admin records 15) Eliminate all Restore records 16) All (3-15) 17) Quit Select function number:
By entering 1 or 2, you can toggle the modify database flag (-f option) and the verbose flag (-v). It can be helpful and reassuring to turn off the modify database flag, then select one or more of the consistency checks (items 3 through 9) to see what will be done, then toggle the modify flag on and re-run the check.
The inconsistencies examined are the following:
During standard purging (or pruning) of Job records, Bacula does not check for orphaned Filename records. As a consequence, over a period of time, old unused Filename records will accumulate and use space in your database. This check will eliminate them. It is strongly recommended that you run this check at least once a year, and for large database (more than 200 Megabytes), it is probably better to run this once every 6 months.
testfind permits listing of files using the same search engine that is used for the Include resource in Job resources. Note, much of the functionality of this program (listing of files to be included) is present in the estimate command in the Console program.
The original use of testfind was to ensure that Bacula's file search engine was correct and to print some statistics on file name and path length. However, you may find it useful to see what bacula would do with a given Include resource. The testfind program can be found in the <bacula-source>/src/tools directory of the source distribution. Though it is built with the make process, it is not normally "installed".
It is called:
Usage: testfind [-d debug_level] [-] [pattern1 ...] -a print extended attributes (Win32 debug) -dnn set debug level to nn - read pattern(s) from stdin -? print this message. Patterns are used for file inclusion -- normally directories. Debug level>= 1 prints each file found. Debug level>= 10 prints path/file for catalog. Errors are always printed. Files/paths truncated is a number with len> 255. Truncation is only in the catalog.
Where a pattern is any filename specification that is valid within an Include resource definition. If none is specified, / (the root directory) is assumed. For example:
./testfind /bin
Would print the following:
Dir: /bin Reg: /bin/bash Lnk: /bin/bash2 -> bash Lnk: /bin/sh -> bash Reg: /bin/cpio Reg: /bin/ed Lnk: /bin/red -> ed Reg: /bin/chgrp ... Reg: /bin/ipcalc Reg: /bin/usleep Reg: /bin/aumix-minimal Reg: /bin/mt Lnka: /bin/gawk-3.1.0 -> /bin/gawk Reg: /bin/pgawk Total files : 85 Max file length: 13 Max path length: 5 Files truncated: 0 Paths truncated: 0
Even though testfind uses the same search engine as Bacula, each directory to be listed, must be entered as a separate command line entry or entered one line at a time to standard input if the - option was specified.
Specifying a debug level of one (i.e. -d1) on the command line will cause testfind to print the raw filenames without showing the Bacula internal file type, or the link (if any). Debug levels of 10 or greater cause the filename and the path to be separated using the same algorithm that is used when putting filenames into the Catalog database.
This chapter is concerned with testing and configuring your tape drive to make sure that it will work properly with Bacula using the btape program.
In general, you should follow the following steps to get your tape drive to work with Bacula. Start with a tape mounted in your drive. If you have an autochanger, load a tape into the drive. We use /dev/nst0 as the tape drive name, you will need to adapt it according to your system.
Do not proceed to the next item until you have succeeded with the previous one.
mt -f /dev/nst0 rewind tar cvf /dev/nst0 . mt -f /dev/nst0 rewind tar tvf /dev/nst0
./btape -c bacula-sd.conf /dev/nst0 test
It isn't necessary to run the autochanger part of the test at this time, but do not go past this point until the basic test succeeds. If you do have an autochanger, please be sure to read the Autochanger chapter of this manual.
restore select all done yes
Do a diff on the restored directory to ensure it is identical to the original directory.
./btape -c bacula-sd.conf /dev/nst0 auto
Adjust your autochanger as necessary to ensure that it works correctly. See the Autochanger chapter of this manual for a complete discussion of testing your autochanger.
If you have reached this point, you stand a good chance of having everything work. If you get into trouble at any point, carefully read the documentation given below. If you cannot get past some point, ask the bacula-users email list, but specify which of the steps you have successfully completed. In particular, you may want to look at the Tips for Resolving Problems section below.
With version 2.6 of the Linux kernel, if there is no tape in the drive, the OS will wait 2 minutes (default) then return a failure, and consequently, Bacula version 1.36 and below will fail the job. This is important to keep in mind, because if you use and option such as Offline on Unmount = yes, there will be a point when there is no tape in the drive, and if another job starts or if Bacula asks the operator to mount a tape, when Bacula attempts to open the drive (about a 20 minute delay), it will fail and Bacula will fail the job.
In version 1.38.x, the Bacula code partially gets around this problem -- at least in the initial open of the drive. However, functions like Polling the drive do not work correctly if there is no tape in the drive. Providing you do not use Offline on Unmount = yes, you should not experience job failures as mentioned above. If you do experience such failures, you can also increase the Maximum Open Wait time interval, which will give you more time to mount the next tape before the job is failed.
Starting with version 1.27, each of the tape utility programs including the btape program requires a valid Storage daemon configuration file (actually, the only part of the configuration file that btape needs is the Device resource definitions). This permits btape to find the configuration parameters for your archive device (generally a tape drive). Without those parameters, the testing and utility programs do not know how to properly read and write your drive. By default, they use bacula-sd.conf in the current directory, but you may specify a different configuration file using the -c option.
btape device-name where the Volume can be found. In the case of a tape, this is the physical device name such as /dev/nst0 or /dev/rmt/0ubn depending on your system that you specify on the Archive Device directive. For the program to work, it must find the identical name in the Device resource of the configuration file. If the name is not found in the list of phsical names, the utility program will compare the name you entered to the Device names (rather than the Archive device names). See below for specifying Volume names.
If you are attempting to read or write an archive file rather than a tape, the device-name should be the full path to the archive location including the filename. The filename (last part of the specification) will be stripped and used as the Volume name, and the path (first part before the filename) must have the same entry in the configuration file. So, the path is equivalent to the archive device name, and the filename is equivalent to the volume name.
This program permits a number of elementary tape operations via a tty command interface. The test command, described below, can be very useful for testing tape drive compatibility problems. Aside from initial testing of tape drive compatibility with Bacula, btape will be mostly used by developers writing new tape drivers.
btape can be dangerous to use with existing Bacula tapes because it will relabel a tape or write on the tape if so requested regardless of whether or not the tape contains valuable data, so please be careful and use it only on blank tapes.
To work properly, btape needs to read the Storage daemon's configuration file. As a default, it will look for bacula-sd.conf in the current directory. If your configuration file is elsewhere, please use the -c option to specify where.
The physical device name or the Device resource name must be specified on the command line, and this same device name must be present in the Storage daemon's configuration file read by btape
Usage: btape [options] device_name -b <file> specify bootstrap file -c <file> set configuration file to file -d <nn> set debug level to nn -p proceed inspite of I/O errors -s turn off signals -v be verbose -? print this message.
An important reason for this program is to ensure that a Storage daemon configuration file is defined so that Bacula will correctly read and write tapes.
It is highly recommended that you run the test command before running your first Bacula job to ensure that the parameters you have defined for your storage device (tape drive) will permit Bacula to function properly. You only need to mount a blank tape, enter the command, and the output should be reasonably self explanatory. For example:
(ensure that Bacula is not running) ./btape -c /usr/bin/bacula/bacula-sd.conf /dev/nst0
The output will be:
Tape block granularity is 1024 bytes. btape: btape.c:376 Using device: /dev/nst0 *
Enter the test command:
test
The output produced should be something similar to the following: I've cut the listing short because it is frequently updated to have new tests.
=== Append files test === This test is essential to Bacula. I'm going to write one record in file 0, two records in file 1, and three records in file 2 btape: btape.c:387 Rewound /dev/nst0 btape: btape.c:855 Wrote one record of 64412 bytes. btape: btape.c:857 Wrote block to device. btape: btape.c:410 Wrote EOF to /dev/nst0 btape: btape.c:855 Wrote one record of 64412 bytes. btape: btape.c:857 Wrote block to device. btape: btape.c:855 Wrote one record of 64412 bytes. btape: btape.c:857 Wrote block to device. btape: btape.c:410 Wrote EOF to /dev/nst0 btape: btape.c:855 Wrote one record of 64412 bytes. btape: btape.c:857 Wrote block to device. btape: btape.c:855 Wrote one record of 64412 bytes. btape: btape.c:857 Wrote block to device. btape: btape.c:855 Wrote one record of 64412 bytes. btape: btape.c:857 Wrote block to device. btape: btape.c:410 Wrote EOF to /dev/nst0 btape: btape.c:387 Rewound /dev/nst0 btape: btape.c:693 Now moving to end of media. btape: btape.c:427 Moved to end of media We should be in file 3. I am at file 3. This is correct! Now the important part, I am going to attempt to append to the tape. ... === End Append files test ===
If you do not successfully complete the above test, please resolve the problem(s) before attempting to use Bacula. Depending on your tape drive, the test may recommend that you add certain records to your configuration. We strongly recommend that you do so and then re-run the above test to insure it works the first time.
Some of the suggestions it provides for resolving the problems may or may not be useful. If at all possible avoid using fixed blocking. If the test suddenly starts to print a long series of:
Got EOF on tape. Got EOF on tape. ...
then almost certainly, you are running your drive in fixed block mode rather than variable block mode. Please see below for help on resolving that.
For FreeBSD users, please see the notes below for doing further testing of your tape drive.
You can find out what SCSI devices you have by doing:
cat /proc/scsi/scsi
For example, I get the following:
Attached devices: Host: scsi2 Channel: 00 Id: 01 Lun: 00 Vendor: HP Model: C5713A Rev: H107 Type: Sequential-Access ANSI SCSI revision: 02 Host: scsi2 Channel: 00 Id: 04 Lun: 00 Vendor: SONY Model: SDT-10000 Rev: 0110 Type: Sequential-Access ANSI SCSI revision: 02
The above represents first an autochanger and second a simple tape drive. The HP changer (the first entry) uses the same SCSI channel for data and for control, so in Bacula, you would use:
Archive Device = /dev/nst0 Changer Device = /dev/sg0
If you want to remove the SDT-10000 device, you can do so as root with:
echo "scsi remove-single-device 2 0 4 0">/proc/scsi/scsi
and you can put add it back with:
echo "scsi add-single-device 2 0 4 0">/proc/scsi/scsi
where the 2 0 4 0 are the Host, Channel, Id, and Lun as seen on the output from cat /proc/scsi/scsi. Note, the Channel must be specified as numeric.
Below is a slightly more complicated output, which is a single autochanger with two drives, and which operates the changer on a different channel from from the drives:
Attached devices: Host: scsi0 Channel: 00 Id: 00 Lun: 00 Vendor: ATA Model: WDC WD1600JD-75H Rev: 08.0 Type: Direct-Access ANSI SCSI revision: 05 Host: scsi2 Channel: 00 Id: 04 Lun: 00 Vendor: HP Model: Ultrium 2-SCSI Rev: F6CH Type: Sequential-Access ANSI SCSI revision: 03 Host: scsi2 Channel: 00 Id: 05 Lun: 00 Vendor: HP Model: Ultrium 2-SCSI Rev: F6CH Type: Sequential-Access ANSI SCSI revision: 03 Host: scsi2 Channel: 00 Id: 06 Lun: 00 Vendor: OVERLAND Model: LXB Rev: 0106 Type: Medium Changer ANSI SCSI revision: 02
The above tape drives are accessed on /dev/nst0 and /dev/nst1, while the control channel for those two drives is /dev/sg3.
If you are getting error messages such as:
Volume data error at 0:1! Wanted block-id: "BB02", got "". Buffer discarded
It is very likely that Bacula has tried to do block positioning and ended up at an invalid block. This can happen if your tape drive is in fixed block mode while Bacula's default is variable blocks. Note that in such cases, Bacula is perfectly able to write to your Volumes (tapes), but cannot position to read them.
There are two possible solutions.
mt -f /dev/nst0 defblksize 0
or whatever is appropriate on your system.
Block Positioning = no
to the Device resource. This is not the recommended procedure because it can enormously slow down recovery of files, but it may help where all else fails. This directive is available in version 1.35.5 or later (and not yet tested).
If you are getting error messages such as:
Volume data error at 0:0! Block checksum mismatch in block=0 len=32625 calc=345678 blk=123456
You are getting tape read errors, and this is most likely due to one of the following things:
If you get an error message such as:
dev open failed: dev.c:265 stored: unable to open device /dev/nst0:> ERR=No such device or address
the first time you run a job, it is most likely due to the fact that you specified the incorrect device name on your Archive Device.
If Bacula works fine with your drive, then all off a sudden you get error messages similar to the one shown above, it is quite possible that your driver module is being removed because the kernel deems it idle. This is done via crontab with the use of rmmod -a. To fix the problem, you can remove this entry from crontab, or you can manually modprob your driver module (or add it to the local startup script). Thanks to Alan Brown for this tip.
When Bacula moves to the end of the medium, it normally uses the ioctl(MTEOM) function. Then Bacula uses the ioctl(MTIOCGET) function to retrieve the current file position from the mt_fileno field. Some SCSI tape drivers will use a fast means of seeking to the end of the medium and in doing so, they will not know the current file position and hence return a -1. As a consequence, if you get "This is NOT correct!" in the positioning tests, this may be the cause. You must correct this condition in order for Bacula to work.
There are two possible solutions to the above problem of incorrect file number:
Hardware End of File = no
This will cause Bacula to use the MTFSF request to seek to the end of the medium, and Bacula will keep track of the file number itself.
Bacula's preferred method of working with tape drives (sequential devices) is to run in variable block mode, and this is what is set by default. You should first ensure that your tape drive is set for variable block mode (see below).
If your tape drive is in fixed block mode and you have told Bacula to use different fixed block sizes or variable block sizes (default), you will get errors when Bacula attempts to forward space to the correct block (the kernel driver's idea of tape blocks will not correspond to Bacula's).
All modern tape drives support variable tape blocks, but some older drives (in particular the QIC drives) as well as the ATAPI ide-scsi driver run only in fixed block mode. The Travan tape drives also apparently must run in fixed block mode (to be confirmed).
Even in variable block mode, with the exception of the first record on the second or subsequent volume of a multi-volume backup, Bacula will write blocks of a fixed size. However, in reading a tape, Bacula will assume that for each read request, exactly one block from the tape will be transferred. This the most common way that tape drives work and is well supported by Bacula.
Drives that run in fixed block mode can cause serious problems for Bacula if the drive's block size does not correspond exactly to Bacula's block size. In fixed block size mode, drivers may transmit a partial block or multiple blocks for a single read request. From Bacula's point of view, this destroys the concept of tape blocks. It is much better to run in variable block mode, and almost all modern drives (the OnStream is an exception) run in variable block mode. In order for Bacula to run in fixed block mode, you must include the following records in the Storage daemon's Device resource definition:
Minimum Block Size = nnn Maximum Block Size = nnn
where nnn must be the same for both records and must be identical to the driver's fixed block size.
We recommend that you avoid this configuration if at all possible by using variable block sizes.
If you must run with fixed size blocks, make sure they are not 512 bytes. This is too small and the overhead that Bacula has with each record will become excessive. If at all possible set any fixed block size to something like 64,512 bytes or possibly 32,768 if 64,512 is too large for your drive. See below for the details on checking and setting the default drive block size.
To recover files from tapes written in fixed block mode, see below.
If you have a modern SCSI tape drive and you are having problems with the test command as noted above, it may be that some program has set one or more of your SCSI driver's options to non-default values. For example, if your driver is set to work in SysV manner, Bacula will not work correctly because it expects BSD behavior. To reset your tape drive to the default values, you can try the following, but ONLY if you have a SCSI tape drive on a Linux system:
become super user mt -f /dev/nst0 rewind mt -f /dev/nst0 stoptions buffer-writes async-writes read-ahead
The above commands will clear all options and then set those specified. None of the specified options are required by Bacula, but a number of other options such as SysV behavior must not be set. Bacula does not support SysV tape behavior. On systems other than Linux, you will need to consult your mt man pages or documentation to figure out how to do the same thing. This should not really be necessary though -- for example, on both Linux and Solaris systems, the default tape driver options are compatible with Bacula.
You may also want to ensure that no prior program has set the default block size, as happened to one user, by explicitly turning it off with:
mt -f /dev/nst0 defblksize 0
If you would like to know what options you have set before making any of the changes noted above, you can now view them on Linux systems, thanks to a tip provided by Willem Riede. Do the following:
become super user mt -f /dev/nst0 stsetoptions 0 grep st0 /var/log/messages
and you will get output that looks something like the following:
kernel: st0: Mode 0 options: buffer writes: 1, async writes: 1, read ahead: 1 kernel: st0: can bsr: 0, two FMs: 0, fast mteom: 0, auto lock: 0, kernel: st0: defs for wr: 0, no block limits: 0, partitions: 0, s2 log: 0 kernel: st0: sysv: 0 nowait: 0
Note, I have chopped off the beginning of the line with the date and machine name for presentation purposes.
Some people find that the above settings only last until the next reboot, so please check this otherwise you may have unexpected problems.
Beginning with Bacula version 1.35.8, if Bacula detects that you are running in variable block mode, it will attempt to set your drive appropriately. All OSes permit setting variable block mode, but some OSes do not permit setting the other modes that Bacula needs to function properly.
As far as I can tell, there is no way with the mt program to check if your tape hardware compression is turned on or off. You can, however, turn it on by using (on Linux):
become super user mt -f /dev/nst0 defcompression 1
and of course, if you use a zero instead of the one at the end, you will turn it off.
If you have built the mtx program in the depkgs package, you can use tapeinfo to get quite a bit of information about your tape drive even if it is not an autochanger. This program is called using the SCSI control device. On Linux for tape drive /dev/nst0, this is usually /dev/sg0, while on FreeBSD for /dev/nsa0, the control device is often /dev/pass2. For example on my DDS-4 drive (/dev/nst0), I get the following:
tapeinfo -f /dev/sg0 Product Type: Tape Drive Vendor ID: 'HP ' Product ID: 'C5713A ' Revision: 'H107' Attached Changer: No MinBlock:1 MaxBlock:16777215 SCSI ID: 5 SCSI LUN: 0 Ready: yes BufferedMode: yes Medium Type: Not Loaded Density Code: 0x26 BlockSize: 0
where the DataCompEnabled: yes means that tape hardware compression is turned on. You can turn it on and off (yes|no) by using the mt commands given above. Also, this output will tell you if the BlockSize is non-zero and hence set for a particular block size. Bacula is not likely to work in such a situation because it will normally attempt to write blocks of 64,512 bytes, except the last block of the job which will generally be shorter. The first thing to try is setting the default block size to zero using the mt -f /dev/nst0 defblksize 0 command as shown above. On FreeBSD, this would be something like: mt -f /dev/nsa0 blocksize 0.
On some operating systems with some tape drives, the amount of data that can be written to the tape and whether or not compression is enabled is determined by the density usually the mt -f /dev/nst0 setdensity xxx command. Often mt -f /dev/nst0 status will print out the current density code that is used with the drive. Most systems, but unfortunately not all, set the density to the maximum by default. On some systems, you can also get a list of all available density codes with: mt -f /dev/nst0 densities or a similar mt command. Note, for DLT and SDLT devices, no-compression versus compression is very often controlled by the density code. On FreeBSD systems, the compression mode is set using mt -f /dev/nsa0 comp xxx where xxx is the mode you want. In general, see man mt for the options available on your system.
If your tape drive requires fixed block sizes (very unusual), you can use the following records:
Minimum Block Size = nnn Maximum Block Size = nnn
in your Storage daemon's Device resource to force Bacula to write fixed size blocks (where you sent nnn to be the same for both of the above records). This should be done only if your drive does not support variable block sizes, or you have some other strong reasons for using fixed block sizes. As mentioned above, a small fixed block size of 512 or 1024 bytes will be very inefficient. Try to set any fixed block size to something like 64,512 bytes or larger if your drive will support it.
Also, note that the Medium Type field of the output of tapeinfo reports Not Loaded, which is not correct. As a consequence, you should ignore that field as well as the Attached Changer field.
To recover files from tapes written in fixed block mode, see below.
On most FreeBSD systems such as 4.9 and most tape drives, Bacula should run with:
mt \ -f \ /dev/nsa0 \ seteotmodel \ 2 mt \ -f \ /dev/nsa0 \ blocksize \ 0 mt \ -f \ /dev/nsa0 \ comp \ enable
You might want to put those commands in a startup script to make sure your tape driver is properly initialized before running Bacula.
Then according to what the btape test command returns, you will probably need to set the following (see below for an alternative):
Hardware End of Medium = no BSF at EOM = yes Backward Space Record = no Backward Space File = no Fast Forward Space File = no TWO EOF = yes
Then be sure to run some append tests with Bacula where you start and stop Bacula between appending to the tape, or use btape version 1.35.1 or greater, which includes simulation of stopping/restarting Bacula.
Please see the file platforms/freebsd/pthreads-fix.txt in the main Bacula directory concerning important information concerning compatibility of Bacula and your system. A much more optimal Device configuration is shown below, but does not work with all tape drives. Please test carefully before putting either into production.
Note, for FreeBSD 4.10-RELEASE, using a Sony TSL11000 L100 DDS4 with an autochanger set to variable block size and DCLZ compression, Brian McDonald reports that to get Bacula to append correctly between Bacula executions, the correct values to use are:
mt \ -f \ /dev/nsa0 \ seteotmodel \ 1 mt \ -f \ /dev/nsa0 \ blocksize \ 0 mt \ -f \ /dev/nsa0 \ comp \ enable
and
Hardware End of Medium = no BSF at EOM = no Backward Space Record = no Backward Space File = no Fast Forward Space File = yes TWO EOF = no
This has been confirmed by several other people using different hardware. This configuration is the preferred one because it uses one EOF and no backspacing at the end of the tape, which works much more efficiently and reliably with modern tape drives.
Finally, here is a Device configuration that Danny Butroyd reports to work correctly with the Overland Powerloader tape library using LT0-2 and FreeBSD 5.4-Stable:
# Overland Powerloader LT02 - 17 slots single drive Device { Name = Powerloader Media Type = LT0-2 Archive Device = /dev/nsa0 AutomaticMount = yes; AlwaysOpen = yes; RemovableMedia = yes; RandomAccess = no; Changer Command = "/usr/local/sbin/mtx-changer %c %o %S %a %d" Changer Device = /dev/pass2 AutoChanger = yes Alert Command = "sh -c 'tapeinfo -f %c |grep TapeAlert|cat'" # FreeBSD Specific Settings Offline On Unmount = no Hardware End of Medium = no BSF at EOM = yes Backward Space Record = no Fast Forward Space File = no TWO EOF = yes }
On FreeBSD, you can do a camcontrol devlist as root to determine what drives and autochangers you have. For example,
undef# camcontrol devlist at scbus0 target 2 lun 0 (pass0,sa0) at scbus0 target 4 lun 0 (pass1,sa1) at scbus0 target 4 lun 1 (pass2)
from the above, you can determine that there is a tape drive on /dev/sa0 and another on /dev/sa1 in addition since there is a second line for the drive on /dev/sa1, you know can assume that it is the control device for the autochanger (i.e. /dev/pass2). It is also the control device name to use when invoking the tapeinfo program. E.g.
tapeinfo -f /dev/pass2
Bacula version 1.33 (not 1.32x) is now working and ready for testing with the OnStream kernel osst driver version 0.9.14 or above. Osst is available from: http://sourceforge.net/projects/osst/.
To make Bacula work you must first load the new driver then, as root, do:
mt -f /dev/nosst0 defblksize 32768
Also you must add the following to your Device resource in your Storage daemon's conf file:
Minimum Block Size = 32768 Maximum Block Size = 32768
Here is a Device specification provided by Michel Meyers that is known to work:
Device { Name = "Onstream DI-30" Media Type = "ADR-30" Archive Device = /dev/nosst0 Minimum Block Size = 32768 Maximum Block Size = 32768 Hardware End of Medium = yes BSF at EOM = no Backward Space File = yes Fast Forward Space File = yes Two EOF = no AutomaticMount = yes AlwaysOpen = yes Removable Media = yes }
Because there are often problems with certain tape drives or systems when end of tape conditions occur, btape has a special command fill that causes it to write random data to a tape until the tape fills. It then writes at least one more Bacula block to a second tape. Finally, it reads back both tapes to ensure that the data has been written in a way that Bacula can recover it. Note, there is also a single tape option as noted below, which you should use rather than the two tape test. See below for more details.
This can be an extremely time consuming process (here it is about 6 hours) to fill a full tape. Note, that btape writes random data to the tape when it is filling it. This has two consequences: 1. it takes a bit longer to generate the data, especially on slow CPUs. 2. the total amount of data is approximately the real physical capacity of your tape, regardless of whether or not the tape drive compression is on or off. This is because random data does not compress very much.
To begin this test, you enter the fill command and follow the instructions. There are two options: the simple single tape option and the multiple tape option. Please use only the simple single tape option because the multiple tape option still doesn't work totally correctly. If the single tape option does not succeed, you should correct the problem before using Bacula.
If you have been previously running your tape drive in fixed block mode (default 512) and Bacula with variable blocks (default), then in version 1.32f-x and 1.34 and above, Bacula will fail to recover files because it does block spacing, and because the block sizes don't agree between your tape drive and Bacula it will not work.
The long term solution is to run your drive in variable block mode as described above. However, if you have written tapes using fixed block sizes, this can be a bit of a pain. The solution to the problem is: while you are doing a restore command using a tape written in fixed block size, ensure that your drive is set to the fixed block size used while the tape was written. Then when doing the restore command in the Console program, do not answer the prompt yes/mod/no. Instead, edit the bootstrap file (the location is listed in the prompt) using any ASCII editor. Remove all VolBlock lines in the file. When the file is re-written, answer the question, and Bacula will run without using block positioning, and it should recover your files.
SCSI tapes may either be written in variable or fixed block sizes. Newer drives support both modes, but some drives such as the QIC devices always use fixed block sizes. Bacula attempts to fill and write complete blocks (default 65K), so that in normal mode (variable block size), Bacula will always write blocks of the same size except the last block of a Job. If Bacula is configured to write fixed block sizes, it will pad the last block of the Job to the correct size. Bacula expects variable tape block size drives to behave as follows: Each write to the drive results in a single record being written to the tape. Each read returns a single record. If you request less bytes than are in the record, only those number of bytes will be returned, but the entire logical record will have been read (the next read will retrieve the next record). Thus data from a single write is always returned in a single read, and sequentially written records are returned by sequential reads.
Bacula expects fixed block size tape drives to behave as follows: If a write length is greater than the physical block size of the drive, the write will be written as two blocks each of the fixed physical size. This single write may become multiple physical records on the tape. (This is not a good situation). According to the documentation, one may never write an amount of data that is not the exact multiple of the blocksize (it is not specified if an error occurs or if the the last record is padded). When reading, it is my understanding that each read request reads one physical record from the tape. Due to the complications of fixed block size tape drives, you should avoid them if possible with Bacula, or you must be ABSOLUTELY certain that you use fixed block sizes within Bacula that correspond to the physical block size of the tape drive. This will ensure that Bacula has a one to one correspondence between what it writes and the physical record on the tape.
Please note that Bacula will not function correctly if it writes a block and that block is split into two or more physical records on the tape. Bacula assumes that each write causes a single record to be written, and that it can sequentially recover each of the blocks it has written by using the same number of sequential reads as it had written.
Linux does support both SCSI SPACE Filemarks and End-of-data: When MTEOM is called in MT_ST_FAST_MTEOM mode, SCSI SPACE Filemarks with count = 8388607 is used. In the other case, SCSI SPACE End-of-data is used. There is no real slow mode like in Solaris - I just expect, that for older tape drives Filemarks may be slower than End-of-data, but not so much as in Solaris slow mode. File number is tracked for MTEOM just without MT_ST_FAST_MTEOM - when MT_ST_FAST_MTEOM is used, it is not.
FreeBSD does support both SCSI SPACE Filemarks and End-of-data, but when MTEOD (MTEOM) is called, SCSI SPACE End-of-data is always used. FreeBSD never use SCSI SPACE Filemarks for MTEOD. File number is never tracked for MTOED.
If I use Yes, OS must not use SCSI SPACE End-of-data, because Bacula expects, that there is tracked file number, which is not supported by SCSI specification. Instead, the OS have to use SCSI SPACE Filemarks.
If I use No, an action depends on Fast Forward Space File.
Considering Hardware End of Medium = no and Fast Forward Space File = no When I set the two to no, file positioning was very slow on my LTO-3:
HEOM = no, FFSF = no: ~ 10 - 100 minutes \end{verbatime} while even with {\bf Hardware End of Medium = no} but {\bf Fast Forward Space File = yes}, the time is 10 to 100 times faster. \begin{verbatim} HEOM = no, FFSF = yes: ~ 1 minute
If you are getting errors such as:
3992 Bad autochanger "load slot 1, drive 1": ERR=Child exited with code 1.
and you are running your Storage daemon as non-root, then most likely you are having permissions problems with the control channel. Running as root, set permissions on /dev/sgX so that the userid and group of your Storage daemon can access the device. You need to ensure that you all access to the proper control device, and if you don't have any SCSI disk drives (including SATA drives), you might want to change the permissions on /dev/sg*.
If you are running on a Linux system, and you have a set of working configuration files, it is very unlikely that Bacula will crash. As with all software, however, it is inevitable that someday, it may crash, particularly if you are running on another operating system or using a new or unusual feature.
This chapter explains what you should do if one of the three Bacula daemons (Director, File, Storage) crashes.
Each of the three Bacula daemons has a built-in exception handler which, in case of an error, will attempt to produce a traceback. If successful the traceback will be emailed to you.
For this to work, you need to ensure that a few things are setup correctly on your system:
If all the above conditions are met, the daemon that crashes will produce a traceback report and email it to you. If the above conditions are not true, you can either run the debugger by hand as described below, or you may be able to correct the problems by editing the btraceback file. I recommend not spending too much time on trying to get the traceback to work as it can be very difficult.
The changes that might be needed are to add a correct path to the gdb program, correct the path to the btraceback.gdb file, change the mail program or its path, or change your email address. The key line in the btraceback file is:
gdb -quiet -batch -x /home/kern/bacula/bin/btraceback.gdb \ $1 $2 2>\&1 | mail -s "Bacula traceback" your-address@xxx.com
Since each daemon has the same traceback code, a single btraceback file is sufficient if you are running more than one daemon on a machine.
To "manually" test the traceback feature, you simply start Bacula then obtain the PID of the main daemon thread (there are multiple threads). Unfortunately, the output had to be split to fit on this page:
[kern@rufus kern]$ ps fax --columns 132 | grep bacula-dir 2103 ? S 0:00 /home/kern/bacula/k/src/dird/bacula-dir -c /home/kern/bacula/k/src/dird/dird.conf 2104 ? S 0:00 \_ /home/kern/bacula/k/src/dird/bacula-dir -c /home/kern/bacula/k/src/dird/dird.conf 2106 ? S 0:00 \_ /home/kern/bacula/k/src/dird/bacula-dir -c /home/kern/bacula/k/src/dird/dird.conf 2105 ? S 0:00 \_ /home/kern/bacula/k/src/dird/bacula-dir -c /home/kern/bacula/k/src/dird/dird.conf
which in this case is 2103. Then while Bacula is running, you call the program giving it the path to the Bacula executable and the PID. In this case, it is:
./btraceback /home/kern/bacula/k/src/dird 2103
It should produce an email showing you the current state of the daemon (in this case the Director), and then exit leaving Bacula running as if nothing happened. If this is not the case, you will need to correct the problem by modifying the btraceback script.
Typical problems might be that gdb is not on the default path. Fix this by specifying the full path to it in the btraceback file. Another common problem is that the mail program doesn't work or is not on the default path. On some systems, it is preferable to use Mail rather than mail.
It should be possible to produce a similar traceback on systems other than Linux, either using gdb or some other debugger. Solaris with gdb loaded works quite fine. On other systems, you will need to modify the btraceback program to invoke the correct debugger, and possibly correct the btraceback.gdb script to have appropriate commands for your debugger. If anyone succeeds in making this work with another debugger, please send us a copy of what you modified.
If for some reason you cannot get the automatic traceback, or if you want to interactively examine the variable contents after a crash, you can run Bacula under the debugger. Assuming you want to run the Storage daemon under the debugger (the technique is the same for the other daemons, only the name changes), you would do the following:
kill -15 PID
where you replace PID by the actual value.
gdb ./bacula-sd
run -s -f -c ./bacula-sd.conf
You may replace the ./bacula-sd.conf with the full path to the Storage daemon's configuration file.
thread apply all bt
After that you can issue any debugging command.
Each of the daemons normally has debug compiled into the program, but disabled. There are two ways to enable the debug output. One is to add the -d nnn option on the command line when starting the debugger. The nnn is the debug level, and generally anything between 50 and 200 is reasonable. The higher the number, the more output is produced. The output is written to standard output.
The second way of getting debug output is to dynamically turn it on using the Console using the setdebug command. The full syntax of the command is:
setdebug level=nnn client=client-name storage=storage-name dir
If none of the options are given, the command will prompt you. You can selectively turn on/off debugging in any or all the daemons (i.e. it is not necessary to specify all the components of the above command).
At the current time only the File daemon or Client program has been tested on Windows. As a consequence, when we speak of the Windows version of Bacula below, we are referring to the File daemon only.
The Windows version of the Bacula File daemon has been tested on Win98, WinMe, WinNT, and Win2000 systems. We have coded to support Win95, but no longer have a system for testing. The Windows version of Bacula is a native Win32 port, but there are very few source code changes to the Unix code, which means that the Windows version is for the most part running code that has long proved stable on Unix systems. When running, it is perfectly integrated with Windows and displays its icon in the system icon tray, and provides a system tray menu to obtain additional information on how Bacula is running (status and events dialog boxes). If so desired, it can also be stopped by using the system tray menu, though this should normally never be necessary.
Once installed Bacula normally runs as a system service. This means that it is immediately started by the operating system when the system is booted, and runs in the background even if there is no user logged into the system.
Normally, you will install the Windows version of Bacula from the binaries. This install is standard Windows .exe that runs an install wizard using the NSIS Free Software installer, so if you have already installed Windows software, it should be very familiar to you.
If you have a previous version Cygwin of Bacula (1.32 or lower) installed, you should stop the service, uninstall it, and remove the Bacula installation directory possibly saving your bacula-fd.conf file for use with the new version you will install. The new native version of Bacula has far fewer files than the old Cygwin version, so it is better to start with a clean directory.
Finally, proceed with the installation.
winbacula-1.xx.0.exe
Also, if you do not wish to see the full listing of all files restored in the job output after running a restore job, you can add , !restored to the director directive in the Messages resource.
That should complete the installation process. When the Bacula File Server is
ready to serve files, an icon representing a
cassette (or tape) will appear in the system tray
; right click on it and a menu will appear.
The Events item is currently unimplemented, by selecting the Status item, you can verify whether any jobs are running or not.
When the Bacula File Server begins saving files, the color of the holes in the
cassette icon will change from white to green ,
and if there is an error, the holes in the cassette icon will change to red
.
If you are using remote desktop connections between your windows boxes, be warned that that tray icon does not always appear. It will always be visible when you log into the console, but the remote desktop may not display it.
After installing Bacula and before running it, you should check the contents of c:\bacula\bin\bacula-fd.conf to ensure that it corresponds to your configuration.
Finally, but pulling up the Task Manager (ctl-alt-del), verify that Bacula is running as a process (not an Application) with User Name SYSTEM. If this is not the case, you probably have not installed Bacula while running as Administrator, and hence it will be unlikely that Bacula can access all the system files.
Once Bacula has been installed, it can be uninstalled using the standard Windows Add/Remove Programs dialog found on the Control panel.
The most likely source of problems is authentication when the Director attempts to connect to the File daemon that you installed. This can occur if the names and the passwords defined in the File daemon's configuration file c:\bacula\bin\bacula-fd.conf on the Windows machine do not match with the names and the passwords in the Director's configuration file bacula-dir.conf located on your Unix/Linux server.
More specifically, the password found in the Client resource in the Director's configuration file must be the same as the password in the Director resource of the File daemon's configuration file. In addition, the name of the Director resource in the File daemon's configuration file must be the same as the name in the Director resource of the Director's configuration file.
It is a bit hard to explain in words, but if you understand that a Director normally has multiple Clients and a Client (or File daemon) may permit access by multiple Directors, you can see that the names and the passwords on both sides must match for proper authentication.
One user had serious problems with the configuration file until he realized that the Unix end of line conventions were used and Bacula wanted them in Windows format. This has not been confirmed though.
Running Unix like programs on Windows machines is a bit frustrating because the Windows command line shell (DOS Window) is rather primitive. As a consequence, it is not generally possible to see the debug information and certain error messages that Bacula prints. With a bit of work, however, it is possible. When everything else fails and you want to see what is going on, try the following:
Start a DOS shell Window. cd c:\bacula\bin bacula-fd -t >out type out
The -t option will cause Bacula to read the configuration file, print any error messages and then exit. the > redirects the output to the file named out, which you can list with the type command.
If something is going wrong later, or you want to run Bacula with a debug option, you might try starting it as:
bacula-fd -d 100 >out
In this case, Bacula will run until you explicitly stop it, which will give you a chance to connect to it from your Unix/Linux server. In later versions of Bacula (1.34 on, I think), when you start the File daemon in debug mode it can write the output to a trace file bacula.trace in the current directory. To enable this, before running a job, use the console, and enter:
trace on
then run the job, and once you have terminated the File daemon, you will find the debug output in the bacula.trace file.
In addition, you should look in the System Applications log on the Control Panel to find any Windows errors that Bacula got during the startup process.
Finally, due to the above problems, when you turn on debugging, and specify trace=1 on a setdebug command in the Console, Bacula will write the debug information to the file bacula.trace in the directory from which Bacula is executing.
If any applications are running during the backup and they have files opened exclusively, Bacula will not be able to backup those files, so be sure you close your applications (or tell your users to close their applications) before the backup. Fortunately, most Microsoft applications do not open files exclusively so that they can be backed up. However, you will need to experiment. In any case, if Bacula cannot open the file, it will print an error message, so you will always know which files were not backed up. For version 1.37.25 and greater, see the section below on Volume Shadow Copy Service.
During backup, Bacula doesn't know about the system registry, so you will either need to write it out to an ASCII file using regedit /e or use a program specifically designed to make a copy or backup the registry.
In Bacula version 1.31 and later, we use Windows backup API calls by default. Typical of Windows, programming these special BackupRead and BackupWrite calls is a real nightmare of complications. The end result gives some distinct advantages and some disadvantages.
First, the advantages are that on WinNT/2K/XP systems, the security and ownership information is now backed up. In addition, with the exception of files in exclusive use by another program (a major disaster for backup programs on Windows), Bacula can now access all system files. This means that when you restore files, the security and ownership information will be restored on WinNT/2K/XP along with the data.
The disadvantage of the Windows backup API calls is that it produces non-portable backups. That is files and their data that are backed up on WinNT using the native API calls (BackupRead/BackupWrite) cannot be restored on Win95/98/Me or Unix systems. In principle, a file backed up on WinNT can be restored on WinXP, but this remains to be seen in practice (not yet tested). In addition, the stand-alone tools such as bls and bextract cannot be used to retrieve the data for those files because those tools are not available on Windows. All restores must use the Bacula restore command. This restriction is mentioned for completeness, but in practice should not create any problems.
As a default, Bacula backs up Windows systems using the Windows API calls. If you want to backup data on a WinNT/2K/XP system and restore it on a Unix/Win95/98/Me system, we have provided a special portable option that backs up the data in a portable fashion by using portable API calls. See the portable option on the Include statement in a FileSet resource in the Director's configuration chapter for the details on setting this option. However, using the portable option means you may have permissions problems accessing files, and none of the security and ownership information will be backed up or restored. The file data can, however, be restored on any system.
You should always be able to restore any file backed up on Unix or Win95/98/Me to any other system. On some systems, such as WinNT/2K/XP, you may have to reset the ownership of such restored files. Any file backed up on WinNT/2K/XP should in principle be able to be restored to a similar system (i.e. WinNT/2K/XP), however, I am unsure of the consequences if the owner information and accounts are not identical on both systems. Bacula will not let you restore files backed up on WinNT/2K/XP to any other system (i.e. Unix Win95/98/Me) if you have used the defaults.
Finally, if you specify the portable=yes option on the files you back up. Bacula will be able to restore them on any other system. However, any WinNT/2K/XP specific security and ownership information will be lost.
The following matrix will give you an idea of what you can expect. Thanks to Marc Brueckner for doing the tests:
+
Backup OS | Restore OS | Results |
WinMe | WinMe | Works |
WinMe | WinNT | Works (SYSTEM permissions) |
WinMe | WinXP | Works (SYSTEM permissions) |
WinMe | Linux | Works (SYSTEM permissions) |
WinXP | WinXP | Works |
WinXP | WinNT | Works (all files OK, but got "The data is invalid" message) |
WinXP | WinMe | Error: Win32 data stream not supported. |
WinXP | WinMe | Works if Portable=yes specified during backup. |
WinXP | Linux | Error: Win32 data stream not supported. |
WinXP | Linux | Works if Portable=yes specified during backup. |
WinNT | WinNT | Works |
WinNT | WinXP | Works |
WinNT | WinMe | Error: Win32 data stream not supported. |
WinNT | WinMe | Works if Portable=yes specified during backup. |
WinNT | Linux | Error: Win32 data stream not supported. |
WinNT | Linux | Works if Portable=yes specified during backup. |
Linux | Linux | Works |
Linux | WinNT | Works (SYSTEM permissions) |
Linux | WinMe | Works |
Linux | WinXP | Works (SYSTEM permissions) |
Microsoft added VSS to Windows XP and Windows 2003. From the perspective of a backup-solution for Windows, this is an extremely important step. VSS allows Bacula to backup open files and even to interact with applications like RDBMS to produce consistent file copies. VSS aware applications are called VSS Writers, they register with the OS so that when Bacula wants to do a Snapshot, the OS will notify the register Writer programs, which may then create a consistent state in their application, which will be backed up. Examples for these writers are "MSDE" (Microsoft database engine), "Event Log Writer", "Registry Writer" plus 3rd party-writers. If you have a non-vss aware application (e.g. SQL Anywhere or probably MySQL), a shadow copy is still generated and the open files can be backed up, but there is no guarantee that the file is consistent.
Bacula produces a message from each of the registered writer programs when it is doing a VSS backup so you know which ones are correctly backed up.
Bacula supports VSS on both Windows 2003 and Windows XP. Technically Bacula creates a shadow copy as soon as the backup process starts. It does then backup all files from the shadow copy and destroys the shadow copy after the backup process. Please have in mind, that VSS creates a snapshot and thus backs up the system at the state it had when starting the backup. It will disregard file changes which occur during the backup process.
VSS can be turned on by placing an
Enable VSS = yes
in your FileSet resource.
Important Note!! Under the current implementation, you may only run a single job at a time in any Win32 File daemon that has VSS active. Running multiple simultanous jobs in the File daemon will most likely cause jobs to fail. This restriction does not apply to the Director, Storage daemons, or any File daemons not running VSS.
The VSS aware File daemon has the letters VSS on the signon line that it produces when contacted by the console. For example:
Tibs-fd Version: 1.37.32 (22 July 2005) VSS Windows XP MVS NT 5.1.2600the VSS is shown in the line above. This only means that the File daemon is capable of doing VSS not that VSS is turned on for a particular backup. There are two ways of telling if VSS is actually turned on during a backup. The first is to look at the status output for a job, e.g.:
Running Jobs: JobId 1 Job NightlySave.2005-07-23_13.25.45 is running. VSS Backup Job started: 23-Jul-05 13:25 Files=70,113 Bytes=3,987,180,650 Bytes/sec=3,244,247 Files Examined=75,021 Processing file: c:/Documents and Settings/kern/My Documents/My Pictures/Misc1/Sans titre - 39.pdd SDReadSeqNo=5 fd=352Here, you see under Running Jobs that JobId 1 is "VSS Backup Job started ..." This means that VSS is enabled for that job. If VSS is not enabled, it will simply show "Backup Job started ..." without the letters VSS.
The second way to know that the job was backed up with VSS is to look at the Job Report, which will look something like the following:
23-Jul 13:25 rufus-dir: Start Backup JobId 1, Job=NightlySave.2005-07-23_13.25.45 23-Jul 13:26 rufus-sd: Wrote label to prelabeled Volume "TestVolume001" on device "DDS-4" (/dev/nst0) 23-Jul 13:26 rufus-sd: Spooling data ... 23-Jul 13:26 Tibs: Generate VSS snapshots. Driver="VSS WinXP", Drive(s)="C" 23-Jul 13:26 Tibs: VSS Writer: "MSDEWriter", State: 1 (VSS_WS_STABLE) 23-Jul 13:26 Tibs: VSS Writer: "Microsoft Writer (Bootable State)", State: 1 (VSS_WS_STABLE) 23-Jul 13:26 Tibs: VSS Writer: "WMI Writer", State: 1 (VSS_WS_STABLE) 23-Jul 13:26 Tibs: VSS Writer: "Microsoft Writer (Service State)", State: 1 (VSS_WS_STABLE)In the above Job Report listing, you see that the VSS snapshot was generated for drive C (if other drives are backed up, they will be listed on the Drive(s)="C" You also see the reports from each of the writer program. Here they all report VSS_WS_STABLE, which means that you will get a consistent snapshot of the data handled by that writer.
If you turn on the firewalling feature on Windows (default in WinXP SP2), you are likely to find that the Bacula ports are blocked and you cannot communicate to the other daemons. This can be deactivated through the Security Notification dialog, which is apparently somewhere in the Security Center. I don't have this on my computer, so I cannot give the exact details.
The command:
netsh firewall set opmode disable
is purported to disable the firewall, but this command is not accepted on my WinXP Home machine.
If you want to see if the File daemon has properly opened the port and is listening, you can enter the following command in a shell window:
netstat -an | findstr 910[123]
We don't currently have a good solution for disaster recovery on Windows as we do on Linux. The main piece lacking is a Windows boot floppy or a Windows boot CD. Microsoft releases a Windows Pre-installation Environment (WinPE) that could possibly work, but we have not investigated it. This means that until someone figures out the correct procedure, you must restore the OS from the installation disks, then you can load a Bacula client and restore files. Please don't count on using bextract to extract files from your backup tapes during a disaster recovery unless you have backed up those files using the portable option. bextract does not run on Windows, and the normal way Bacula saves files using the Windows API prevents the files from being restored on a Unix machine. Once you have an operational Windows OS loaded, you can run the File daemon and restore your user files.
Please see Disaster Recovery of Win32 Systems for the latest suggestion, which looks very promising.
It looks like Bart PE Builder, which creates a Windows PE (Pre-installation Environment) Boot-CD, may be just what is needed to build a complete disaster recovery system for Win32. This distribution can be found at http://www.nu2.nu/pebuilder/ .
If you restore files backed up from WinNT/XP/2K to an alternate directory, Bacula may need to create some higher level directories that were not saved (or restored). In this case, the File daemon will create them under the SYSTEM account because that is the account that Bacula runs under as a service. As of version 1.32f-3, Bacula creates these files with full access permission. However, there may be cases where you have problems accessing those files even if you run as administrator. In principle, Microsoft supplies you with the way to cease the ownership of those files and thus change the permissions. However, a much better solution to working with and changing Win32 permissions is the program SetACL, which can be found at http://setacl.sourceforge.net/ .
If you have not installed Bacula while running as Administrator and if Bacula is not running as a Process with the userid (User Name) SYSTEM, then it is very unlikely that it will have sufficient permission to access all your files.
Some users have experienced problems restoring files that participate in the Active Directory. They also report that changing the userid under which Bacula (bacula-fd.exe) runs, from SYSTEM to a Domain Admin userid, resolves the problem.
The following solution was provided by Dan Langille <dan at langille in the dot org domain>. The steps are performed using Windows 2000 Server but they should apply to most Win32 platforms. The procedure outlines how to deal with a problem which arises when a restore creates a top-level new directory. In this example, "top-level" means something like c:\src, not c:\tmp\src where c:\tmp already exists. If a restore job specifies / as the Where: value, this problem will arise.
The problem appears as a directory which cannot be browsed with Windows Explorer. The symptoms include the following message when you try to click on that directory:
If you encounter this message, the following steps will change the permissions to allow full access.
You should see something like this:
With the above procedure, you should now have full control over your restored directory.
In addition to the above methods of changing permissions, there is a Microsoft program named cacls that can perform similar functions.
A suggestion by Damian Coutts using Microsoft's NTBackup utility in conjunction with Bacula should permit a full restore of any damaged system files on Win2K/XP. His suggestion is to do an NTBackup of the critical system state prior to running a Bacula backup with the following command:
ntbackup backup systemstate /F c:\systemstate.bkf
The backup is the command, the systemstate says to backup only the system state and not all the user files, and the /F c:\systemstate.bkf specifies where to write the state file. this file must then be saved and restored by Bacula.
To restore the system state, you first reload a base operating system if the OS is damaged, otherwise, this is not necessary, then you would use Bacula to restore all the damaged or lost user's files and to recover the c:\systemstate.bkf file. Finally if there are any damaged or missing system files or registry problems, you run NTBackup and catalogue the system statefile, and then select it for restore. The documentation says you can't run a command line restore of the systemstate.
To the best of my knowledge, this has not yet been tested. If you test it, please report your results to the Bacula email list.
Please see the Director's Configuration chapter of this manual for important considerations on how to specify Windows paths in Bacula FileSet Include and Exclude directives.
Bacula versions prior to 1.37.28 do not support Windows Unicode filenames. As of that version, both bconsole and bwx-console support Windows Unicode filenames. There may still be some problems with multiple byte characters (e.g. Chinese, ...) where it is a two byte character but the displayed character is not two characters wide.
Path/filenames longer than 260 characters are not supported. This may be possible in a future version.
These options are not normally seen or used by the user, and are documented here only for information purposes. At the current time, to change the default options, you must either manually run Bacula or you must manually edit the system registry and modify the appropriate entries.
In order to avoid option clashes between the options necessary for Bacula to run on Windows and the standard Bacula options, all Windows
specific options are signaled with a forward slash character (/), while as
usual, the standard Bacula options are signaled with a minus (-), or a minus
minus (--
). All the standard Bacula options can be used on the Windows
version. In addition, the following Windows only options are implemented:
It is important to note that under normal circumstances the user should never need to use these options as they are normally handled by the system automatically once Bacula is installed. However, you may note these options in some of the .bat files that have been created for your use.
Some users like to shutdown their windows machines after a backup using a Client Run After Job directive. If you want to do something similar, you might take the shutdown program from the apcupsd project or one from the Sysinternals project.
When disaster strikes, you must have a plan, and you must have prepared in advance otherwise the work of recovering your system and your files will be considerably greater. For example, if you have not previously saved the partitioning information for your hard disk, how can you properly rebuild it if the disk must be replaced?
Unfortunately, many of the steps one must take before and immediately after a disaster are very operating system dependent. As a consequence, this chapter will discuss in detail disaster recovery (also called Bare Metal Recovery) for Linux and Solaris. For Solaris, the procedures are still quite manual. For FreeBSD the same procedures may be used but they are not yet developed. For Win32, no luck. Apparently an "emergency boot" disk allowing access to the full system API without interference does not exist.
Here are a few important considerations concerning disaster recovery that you should take into account before a disaster strikes.
The remainder of this section concerns recovering a Linux computer, and parts of it relate to the Red Hat version of Linux. The Solaris procedures can be found below under the Solaris Bare Metal Recovery section of this chapter.
Previously Bacula supported a floppy rescue disk. This code has been removed in 1.37.40 and later.
A so called "Bare Metal" recovery is one where you start with an empty hard disk and you restore your machine. There are also cases where you may lose a file or a directory and want it restored. Please see the previous chapter for more details for those cases.
Bare Metal Recovery assumes that you have the following items for your system:
In addition, to the above assumptions, the following conditions or restrictions apply:
To build the Bacula Rescue CDROM, you must get a copy of the rescue files. In version 1.37 and later, they are separate from the Bacula source. One place you can find the rescue files is in the Source Forge Bacula CVS module named rescue.
Please read the README file in the main directory of the Rescue source code. Before using it, you must run configure and specify the location of the Bacula source code (not necessary if installed from rpms). This permits the build of the rescue disk to automatically create a statically linked Bacula File daemon.
You will find the necessary scripts in linux/cdrom subdirectory of the rescue source code. If you installed the bacula rpm package the scripts will be found in the /etc/bacula/rescue/cdrom directory.
Before you can do a Bare Metal recovery, you must create a Bacula Rescue CDROM, which will contain everything you need to begin recovery. This assumes that you will have your Director and Storage daemon running on a different machine. If you want to recover a machine where the Director and/or the database were previously running things will be much more complicated.
The primary goals of the Bacula rescue CD are:
One of the main of the advantages of a Bacula Rescue CDROM is that it contains a bootable copy of your system, so you should be familiar with it.
You should probably make a new rescue CDROM each time you make any major updates to your kernel, and every time you upgrade a major version of Bacula.
The whole process with the exception of burning the CDROM is done with the following commands:
(Build a working version of Bacula in the bacula-source directory) cd <bacula-source> ./configure (your options) make cd <bacula-rescue-source> ./configure --with-bacula=<path-to-bacula-source> cd linux/cdrom su (become root) make all
The above instructions were for building the rescue CDROM from a bacula-rescue release. You will note that you need to do a separate ./configure in the rescue source directory and that you need to provide it the path to the Bacula source so that it can build a statically linked File daemon.
For users of the bacula-rescue rpm the static bacula-fd has already been built and placed in /etc/bacula/rescue/cdrom/bin/ along with a symbolic link to your /etc/bacula/bacula-fd.conf file. Rpm users only need to do the second step:
cd /etc/bacula/rescue/cdrom su (become root) make all
At this point, if the scripts are successful, they should have done the following things:
Once this is accomplished, you need only burn it into a CDROM. This can be done directly from the makefile with:
make burn
However, you may need to modify the Makefile to properly specify your CD burner as the detection process is complicated especially if you have two CDROMs or do not have cdrecord loaded on your system. Users of the rescue rpm package should definitely examine the Makefile since it was configured on the host used to produce the rpm package. If you find that the make burn does not work for you, try doing a:
make scan
and use the output of that to modify the Makefile accordingly.
The "make all" that you did above actually does the equivalent to the following:
make kernel make binaries make bacula make iso
If you wish, you can modify what you put on the CDROM and redo any part of the make that you wish. For example, if you want to add a new directory, you might do the first three makes, then add a new directory to the CDROM, and finally do a "make iso". Please see the README file in the rescue/linux/cdrom or /etc/bacula/rescue/cdromdirectory for instructions on changing the contents of the CDROM.
At the current time, the size of the CDROM is about 50MB (compressed to about 20MB), so there is quite a bit more room for additional programs. Keep in mind that when this CDROM is booted, *everything* is in memory, so the total size cannot exceed your memory size, and even then you will need some reserve memory for running programs, ...
You may also use:
make copy-static-bacula
which is similar to a "make all" except that instead of trying to build the static File daemon from the Bacula source, it will assume that you have already installed a static copy of the FD in the sbindir location (usually /usr/local/sbin), and will copy it from there.
Finally, if you want to be completely responsible for getting your own FD binary on the disk, you can do the following:
cd linux/cdrom touch rpm_release make kernel make binaries make bacula (add your own Bacula FD to the bacula/bin directory) make iso rm -f rpm_release
The rpm_release file prevents the "make bacula" from attempting to build or copy a File daemon, so that you can do it before the "make iso" step. Once "make iso" is run, you can no longer add anything to the in-memory part of the image. You can still add files to the cdtree directory, and when you do a "make burn" they will be written to the CDROM. However, to access them, you must be able to mount the CDROM after booting it, then copy them into memory.
You can put multiple systems on the same rescue CD if you wish. This is because the information that is specific to your OS will be stored in the /bacula-hostname directory, where hostname is the name of the host on which you are building the CD. Suppose for example, you have two systems. One named client1 and one named client2. Assume also that your CD burner is on client1, and that is the machine we start on, and that we can ssh into client2 and also client2's disks are mounted on client1.
ssh client2 cd <bacula-source> ./configure (your options) make cd <bacula-rescue-source> ./configure --with-bacula=<path-to-bacula-source> cd linux/cdrom su (become root) make all exit
Again, for rpm package users the above command set would be:
ssh client2 cd /etc/bacula/rescue/cdrom su (enter root password) make bacula exit
Thus we have just built a Bacula rescue directory on client2. Now, on client1, we copy the appropriate directory to two places (explained below), then build an ISO and burn it:
cd <bacula-source> ./configure (your options) make cd <bacula-rescue-source> ./configure --with-bacula=<path-to-bacula-source> cd linux/cdrom su (become root) c=/mnt/client2/home/user/bacula/rescue/linux/cdrom cp -a $c/roottree/bacula-client2 roottree cp -a $c/roottree/bacula-client2 cdtree make all make burn exit
And with the rpm package:
cd /etc/bacula/rescue/cdrom su (enter root password) c=/mnt/client2/etc/bacula/rescue/cdrom cp -a $c/roottree/bacula-client2 roottree cp -a $c/roottree/bacula-client2 cdtree make all make burn exit
In summary, with the above commands, we first build a Bacula directory on client2 in roottree/bacula-client2, then we copied the bacula-client2 directory into the client1's roottree so it is available in memory after booting, and we also copied it into the cdtree so it will also be on the CD as a separate directory and thus can be read without booting the CDROM. Then we made and burned the CDROM for client1, which of course, contains the client2 data.
Now, let's assume that your hard disk has just died and that you have replaced it with an new identical drive. In addition, we assume that you have:
This is a relatively simple case, and later in this chapter, as time permits, we will discuss how you might recover from a situation where the machine that crashes is your main Bacula server (i.e. has the Director, the Catalog, and the Storage daemon).
You will take the following steps to get your system back up and running:
Now for the details ...
When the CDROM boots, you will be presented with a script that looks like:
Welcome to the Bacula Rescue Disk 1.1.0 To proceed, press the <ENTER> key or type "linux <runlevel>" linux 1 -> shell linux 2 -> login (default if ENTER pressed) linux 3 -> network started and login (network not working yet) linux debug -> print debug during boot then login
Normally, at this point, you simply press ENTER. However, you may supply options for the boot if you wish.
Once it has booted, you will be requested to login something like:
Welcome to the Bacula Rescue CDROM 2.4.21-15.0.4.EL #1 Wed Aug 4 03:08:03 EDT 2004 Please login using root and your root password ... RescueCD login:
Note, you must enter the root password for the system on which you loaded the kernel or on which you did the build of the CDROM. Once you are logged in, you will be in the home directory for root, and you can proceed to examine your system.
The complete Bacula rescue part of the CD will be in the directory: /bacula-hostname, where hostname is replaced by the name of the host machine on which you did the build for the CDROM. This naming procedure allows you to put multiple restore environments for each of your machines on a single CDROM if you so wish to do. Please see the README document in the rescue/linux/cdrom directory for more information on adding to the CDROM.
At this point, you should bring up your network. Normally, this is quite simple and requires just a few commands. Please cd into the /bacula-hostname directory before continuing. To simplify your task, we have created a script that should work in most cases by typing:
cd /bacula-hostname ./start_network
You can test it by pinging another machine, or pinging your broken machine machine from another machine. Do not proceed until your network is up.
Assuming that your hard disk crashed and needs repartitioning, proceed with:
./partition.hda
If you have multiple disks, do the same for each of them. For SCSI disks, the repartition script will be named: partition.sda. If the script complains about the disk being in use, simply go back and redo the df command and umount commands until you no longer have your hard disk mounted. Note, in many cases, if your hard disk was seriously damaged or a new one installed, it will not automatically be mounted. If it is mounted, it is because the emergency kernel found one or more possibly valid partitions.
If for some reason this procedure does not work, you can use the information in partition.hda to re-partition your disks by hand using fdisk.
If you have repartitioned your hard disk, you must format it appropriately. The formatting script will put back swap partitions, normal Unix partitions (ext2) and journaled partitions (ext3) as well as Reiser partitions (rei). Do so by entering for each disk:
./format.hda
The format script will ask you if you want a block check done. We recommend to answer yes, but realize that for very large disks this can take hours.
Once the disks are partitioned and formatted, you can remount them with the mount_drives script. All your drives must be mounted for Bacula to be able to access them. Run the script as follows:
./mount_drives df
The df command will tell you if the drives are mounted. If not, re-run the script again. It isn't always easy to figure out and create the mount points and the mounts in the proper order, so repeating the ./mount_drives command will not cause any harm and will most likely work the second time. If not, correct it by hand before continuing.
If you have booted with a Bacula Rescue CDROM, your statically linked Bacula File daemon and the bacula-fd.conf file will be in the /bacula-hostname/bin directory. Make sure bacula-fd and bacula-fd.conf are both there.
Edit the Bacula configuration file, create the working/pid/subsys directory if you haven't already done so above, and start Bacula. Before starting Bacula, you will need to move it and bacula-fd.conf from /bacula-hostname/bin, to the /mnt/disk/tmp directory so that it will be on your hard disk. Then start it with the following command:
chroot /mnt/disk /tmp/bacula-fd -c /tmp/bacula-fd.conf
The above command starts the Bacula File daemon with the proper root disk location (i.e. /mnt/disk/tmp. If Bacula does not start, correct the problem and start it. You can check if it is running by entering:
ps fax
You can kill Bacula by entering:
kill -TERM <pid>
where pid is the first number printed in front of the first occurrence of bacula-fd in the ps fax command.
Now, you should be able to use another computer with Bacula installed to check the status by entering:
status client=xxxx
into the Console program, where xxxx is the name of the client you are restoring.
One common problem is that your bacula-dir.conf may contain machine addresses that are not properly resolved on the stripped down system to be restored because it is not running DNS. This is particularly true for the address in the Storage resource of the Director, which may be very well resolved on the Director's machine, but not on the machine being restored and running the File daemon. In that case, be prepared to edit bacula-dir.conf to replace the name of the Storage daemon's domain name with its IP address.
On the computer that is running the Director, you now run a restore command and select the files to be restored (normally everything), but before starting the restore, there is one final change you must make using the mod option. You must change the Where directory to be the root by using the mod option just before running the job and selecting Where. Set it to:
/
then run the restore.
You might be tempted to avoid using chroot and running Bacula directly and then using a Where to specify a destination of /mnt/disk. This is possible, however, the current version of Bacula always restores files to the new location, and thus any soft links that have been specified with absolute paths will end up with /mnt/disk prefixed to them. In general this is not fatal to getting your system running, but be aware that you will have to fix these links if you do not use chroot.
At this point, the restore should have finished with no errors, and all your files will be restored. One last task remains and that is to write a new boot sector so that your machine will boot. For lilo, you enter the following command:
./run_lilo
If you are using grub instead of lilo, you must enter the following:
./run_grub
Note, I've had quite a number of problems with grub because it is rather complicated and not designed to install easily under a simplified system. So, if you experience errors or end up unexpectedly in a chroot shell, simply exit back to the normal shell and type in the appropriate commands from the run_grub script by hand until you get it to install. When you run the run_grub script, it will print the commands that you should manually enter if that is necessary.
First unmount all your hard disks, otherwise they will not be cleanly shutdown, then reboot your machine by entering exit until you get to the main prompt then enter Ctrl-d. Once back to the main CDROM prompt, you will need to turn the power off, then back on to your machine to get it to reboot.
If everything went well, you should now be back up and running. If not, re-insert the emergency boot CDROM, boot, and figure out what is wrong.
Above, we considered how to recover a client machine where a valid Bacula server was running on another machine. However, what happens if your server goes down and you no longer have a running Director, Catalog, or Storage daemon? There are several solutions:
The first option, is very difficult because it requires you to have created a static version of the Director and the Storage daemon as well as the Catalog. If the Catalog uses MySQL or PostgreSQL, this may or may not be possible. In addition, to loading all these programs on a bare system (quite possible), you will need to make sure you have a valid driver for your tape drive.
The second suggestion is probably a much simpler solution, and one I have done myself. To do so, you might want to consider the following steps:
Since every flavor and every release of Linux is different, there are likely to be some small difficulties with the scripts, so please be prepared to edit them in a minimal environment. A rudimentary knowledge of vi is very useful. Also, these scripts do not do everything. You will need to reformat Windows partitions by hand, for example.
Getting the boot loader back can be a problem if you are using grub because it is so complicated. If all else fails, reboot your system from your floppy but using the restored disk image, then proceed to a reinstallation of grub (looking at the run-grub script can help). By contrast, lilo is a piece of cake.
The same basic techniques described above also apply to FreeBSD. Although we don't yet have a fully automated procedure, Alex Torres Molina has provided us with the following instructions with a few additions from Jesse Guardiani and Dan Langille:
The same basic techniques described above apply to Solaris:
However, during the recovery phase, the boot and disk preparation procedures are different:
Once the disk is partitioned, formatted and mounted, you can continue with bringing up the network and reloading Bacula.
As mentioned above, before a disaster strikes, you should prepare the information needed in the case of problems. To do so, in the rescue/solaris subdirectory enter:
su ./getdiskinfo ./make_rescue_disk
The getdiskinfo script will, as in the case of Linux described above, create a subdirectory diskinfo containing the output from several system utilities. In addition, it will contain the output from the SysAudit program as described in Curtis Preston's book. This file diskinfo/sysaudit.bsi will contain the disk partitioning information that will allow you to manually follow the procedures in the "Unix Backup & Recovery" book to repartition and format your hard disk. In addition, the getdiskinfo script will create a start_network script.
Once you have your disks repartitioned and formatted, do the following:
When a pre-1.30 version of Bacula restores a directory, it first must create the directory, then it populates the directory with its files and subdirectories. The act of creating the files and subdirectories updates both the modification and access times associated with the directory itself. As a consequence, all modification and access times of all directories will be updated to the time of the restore.
This has been corrected in Bacula version 1.30 and later. The directory modification and access times are reset to the value saved in the backup after all the files and subdirectories have been restored. This has been tested and verified on normal restore operations, but not verified during a bare metal recovery.
If any of you look closely at the bootstrap file that is produced and used for the restore (I sure do), you will probably notice that the FileIndex item does not include all the files saved to the tape. This is because in some instances there are duplicates (especially in the case of an Incremental save), and in such circumstances, Bacula restores only the last of multiple copies of a file or directory.
Due to open system files, and registry problems, Bacula cannot save and restore a complete Win2K/XP/NT environment.
A suggestion by Damian Coutts using Microsoft's NTBackup utility in conjunction with Bacula should permit a Full bare metal restore of Win2K/XP (and possibly NT systems). His suggestion is to do an NTBackup of the critical system state prior to running a Bacula backup with the following command:
ntbackup backup systemstate /F c:\systemstate.bkf
The backup is the command, the systemstate says to backup only the system state and not all the user files, and the /F c:\systemstate.bkf specifies where to write the state file. this file must then be saved and restored by Bacula.
To restore the system state, you first reload a base operating system, then you would use Bacula to restore all the users files and to recover the c:\systemstate.bkf file, and finally, run NTBackup and catalogue the system statefile, and then select it for restore. The documentation says you can't run a command line restore of the systemstate.
This procedure has been confirmed to work by Ludovic Strappazon -- many thanks!
A new tool is provided in the form of a bacula plugin for the BartPE rescue CD. BartPE is a self-contained WindowsXP boot CD which you can make using the PeBuilder tools available at http://www.nu2.nu/pebuilder/ and a valid Windows XP SP1 CDROM. The plugin is provided as a zip archive. Unzip the file and copy the bacula directory into the plugin directory of your BartPE installation. Edit the configuration files to suit your installation and build your CD according to the instructions at Bart's site. This will permit you to boot from the cd, configure and start networking, start the bacula file client and access your director with the console program. The programs menu on the booted CD contains entries to install the file client service, start the file client service, and start the WX-Console. You can also open a command line window and CD Programs\Bacula and run the command line console bconsole.
Bacula versions after 1.31 should properly restore ownership and permissions on all WinNT/XP/2K systems. If you do experience problems, generally in restores to alternate directories because higher level directories were not backed up by Bacula, you can correct any problems with the SetACL available under the GPL license at: http://sourceforge.net/projects/setacl/.
Ludovic Strappazon has suggested an interesting way to backup and restore complete Win32 partitions. Simply boot your Win32 system with a Linux Rescue disk as described above for Linux, install a statically linked Bacula, and backup any of the raw partitions you want. Then to restore the system, you simply restore the raw partition or partitions. Here is the email that Ludovic recently sent on that subject:
I've just finished testing my brand new cd LFS/Bacula with a raw Bacula backup and restore of my portable. I can't resist sending you the results: look at the rates !!! hunt-dir: Start Backup JobId 100, Job=HuntBackup.2003-04-17_12.58.26 hunt-dir: Bacula 1.30 (14Apr03): 17-Apr-2003 13:14 JobId: 100 Job: HuntBackup.2003-04-17_12.58.26 FileSet: RawPartition Backup Level: Full Client: sauvegarde-fd Start time: 17-Apr-2003 12:58 End time: 17-Apr-2003 13:14 Files Written: 1 Bytes Written: 10,058,586,272 Rate: 10734.9 KB/s Software Compression: None Volume names(s): 000103 Volume Session Id: 2 Volume Session Time: 1050576790 Last Volume Bytes: 10,080,883,520 FD termination status: OK SD termination status: OK Termination: Backup OK hunt-dir: Begin pruning Jobs. hunt-dir: No Jobs found to prune. hunt-dir: Begin pruning Files. hunt-dir: No Files found to prune. hunt-dir: End auto prune. hunt-dir: Start Restore Job RestoreFilesHunt.2003-04-17_13.21.44 hunt-sd: Forward spacing to file 1. hunt-dir: Bacula 1.30 (14Apr03): 17-Apr-2003 13:54 JobId: 101 Job: RestoreFilesHunt.2003-04-17_13.21.44 Client: sauvegarde-fd Start time: 17-Apr-2003 13:21 End time: 17-Apr-2003 13:54 Files Restored: 1 Bytes Restored: 10,056,130,560 Rate: 5073.7 KB/s FD termination status: OK Termination: Restore OK hunt-dir: Begin pruning Jobs. hunt-dir: No Jobs found to prune. hunt-dir: Begin pruning Files. hunt-dir: No Files found to prune. hunt-dir: End auto prune.
If for some reason you want to do a Full restore to a system that has a working kernel (not recommended), you will need to take care not to overwrite the following files:
/etc/grub.conf /etc/X11/Conf /etc/fstab /etc/mtab /lib/modules /usr/modules /usr/X11R6 /etc/modules.conf
Many thanks to Charles Curley who wrote Linux Complete Backup and Recovery HOWTO for the The Linux Documentation Project. This is an excellent document on how to do Bare Metal Recovery on Linux systems, and it was this document that made me realize that Bacula could do the same thing.
You can find quite a few additional resources, both commercial and free at Storage Mountain, formerly known as Backup Central.
And finally, the O'Reilly book, "Unix Backup & Recovery" by W. Curtis Preston covers virtually every backup and recovery topic including bare metal recovery for a large range of Unix systems.
Bacula TLS (Transport Layer Security) is built-in network encryption code to provide secure network transport similar to that offered by stunnel or ssh. The Bacula code was written by Landon Fuller.
Supported features of this code include:
This document will refer to both "server" and "client" contexts. These terms refer to the accepting and initiating peer, respectively.
Diffie-Hellman anonymous ciphers are not supported by this code. The use of DH anonymous ciphers increases the code complexity and places explicit trust upon the two-way CRAM-MD5 implementation. CRAM-MD5 is subject to known plaintext attacks, and it should be considered considerably less secure than PKI certificate-based authentication.
Appropriate autoconf macros have been added to detect and use OpenSSL
if enabled on the ./configure line with --
enable-openssl
To generate the parameter file, you may use openssl:
openssl dhparam -out dh1024.pem -5 1024
You may create a self-signed certificate for use with the Bacula TLS that will permit you to make it function, but will not allow certificate validation. The .pem file containing both the certificate and the key valid for 10 years can be made with the following:
openssl req -new -x509 -nodes -out bacula.pem -keyout bacula.pem -days 3650
The above script will ask you a number of questions. You may simply answer each of them by entering a return, or if you wish you may enter your own data.
An alternative is to generate your self-signed certificates with TinyCA, which has a very nice Graphical User Interface. TinyCA can be found at http://tinyca.sm-zone.net/.
The process of getting a certificate that is signed by a CA is quite a bit more complicated. You can purchase one from quite a number of PKI vendors, but that is not at all necessary for use with Bacula. To get a CA signed certificate, you will either need to find a friend that has setup his own CA or to become a CA yourself, and thus you can sign all your own certificates. The book OpenSSL by John Viega, Matt Mesier & Pravir Chandra from O'Reilly explains how to do it, or you can read the documentation provided in the Open-source PKI Book project at Source Forge: http://ospkibook.sourceforge.net/docs/OSPKI-2.4.7/OSPKI-html/ospki-book.htm. Note, this link may change.
Landon has supplied us with the TLS portions of his configuration files, which should help you setting up your own.
bacula-dir.conf
Director { # define myself Name = backup1-dir ... TLS Require = yes TLS Verify Peer = yes TLS Allowed CN = "bacula@backup1.example.com" TLS Allowed CN = "administrator@example.com" TLS CA Certificate File = /usr/local/etc/ssl/ca.pem # This is a server certificate, used for incoming # console connections. TLS Certificate = /usr/local/etc/ssl/backup1/cert.pem TLS Key = /usr/local/etc/ssl/backup1/key.pem } Storage { Name = File Address = backup1.example.com ... TLS Require = yes TLS CA Certificate File = /usr/local/etc/ssl/ca.pem # This is a client certificate, used by the director to # connect to the storage daemon TLS Certificate = /usr/local/etc/ssl/bacula@backup1/cert.pem TLS Key = /usr/local/etc/ssl/bacula@backup1/key.pem }
bacula-fd.conf
Director { Name = backup1-dir ... TLS Require = yes TLS Verify Peer = yes # Allow only the Director to connect TLS Allowed CN = "bacula@backup1.example.com" TLS CA Certificate File = /usr/local/etc/ssl/ca.pem\ # This is a server certificate. It is used by connecting # directors to verify the authenticity of this file daemon TLS Certificate = /usr/local/etc/ssl/server1/cert.pem TLS Key = /usr/local/etc/ssl/server1/key.pem }
bacula-sd.conf
Storage { # definition of myself Name = backup1-sd ... # These TLS configuration options are used for incoming # file daemon connections. Director TLS settings are handled # below. TLS Require = yes # Peer certificate is not required/requested -- peer validity # is verified by the storage connection cookie provided to the # File Daemon by the director. TLS Verify Peer = no TLS CA Certificate File = /usr/local/etc/ssl/ca.pem # This is a server certificate. It is used by connecting # file daemons to verify the authenticity of this storage daemon TLS Certificate = /usr/local/etc/ssl/backup1/cert.pem TLS Key = /usr/local/etc/ssl/backup1/key.pem } # # List Directors who are permitted to contact Storage daemon # Director { Name = backup1-dir ... TLS Require = yes # Require the connecting director to provide a certificate # with the matching CN. TLS Verify Peer = yes TLS Allowed CN = "bacula@backup1.example.com" TLS CA Certificate File = /usr/local/etc/ssl/ca.pem # This is a server certificate. It is used by the connecting # director to verify the authenticity of this storage daemon TLS Certificate = /usr/local/etc/ssl/backup1/cert.pem TLS Key = /usr/local/etc/ssl/backup1/key.pem }
Depuis les tous premiers stades de Bacula (janvier 2000) jusqu'à aujourd'hui (Décembre 2005), Bacula a connu deux formats majeurs d'écriture sur les cartouches. Le second format a été introduit dans la version 1.27 en novembre 2002, et n'a pas changé depuis. En principe, Bacula devrait encore pouvoir lire le format d'origine, mais j'avoue ne pas avoir essayé depuis longtemps...
Bien que le format des cartouches soit fixé, les types de données qui peuvent être écrites sur les cartouches sont extensibles, ce qui nous a permis d'ajouter de nouvelles fonctionnalités telles que les ACLs, les données Win32, les données chiffrées... Naturellement, une ancienne version de Bacula ne saurait lire des nouveaux flux de données, mais chaque nouvelle version de Bacula est en principe capable de lire les anciens flux.
Si vous voulez être absolument certain de pouvoir lire vos vieilles cartouches, vous devriez :
1. Essayer de lire les vieilles cartouches de temps en temps, une fois par an par exemple.
2. Conserver une copie statiquement liée de chaque version de Bacula que vous avez utilisée en production. Ainsi, si pour quelque raison nous venions à abandonner la compatibilité avec les anciens formats de cartouches, vous pourriez toujours remettre en service une vieille copie de Bacula...
Le second point est probablement excessif, en toute rigueur, il pourrait vous sauver un jour.
Les TCP Wrappers sont implémentés si vous les activez lors de la
configuration (./configure :--:
with-tcp-wrappers). Avec ce code activé, vous
pourrez contrôler qui peut accéder à vos daemons. Ce contrôle
est obtenu par la modification du fichier /etc/hosts.allow. Le nom de
programme qu'utilise Bacula pour appliquer ces restrictions est celui
que vous avez spécifié dans le fichier de configuration du daemon.
Vous ne devez pas utiliser l'option twist dans votre /etc/hosts.allow car elle stopperait les daemons Bacula lorsqu'une
connection est refusée.
Le nom exact du paquet requis pour compiler avec le support TCP wrappers dépend du système. Il s'agit, par exemple, de tcpd-devel sur SuSE, et de tcp_wrappers sur RedHat.
Dan Langille a fourni les informations suivantes concernant la configuration et les tests de TCP Wrappers avec Bacula.
Si vous lisez hosts_options(5), vous verrez une option nommée twist. Cette option remplace le processus courant par une instance de la commande shell spécifiée. Voici un exemple typique de son utilisation :
ALL : ALL \ : severity auth.info \ : twist /bin/echo "Vous n'\^etes pas autoris\'e \`a utiliser %d depuis %h."
Le code libwrap tente d'éviter twist s'il est exécuté dans un processus résident. Il en résulte que le processus (e.g. bacula-fd, bacula-sd, bacula-dir) sera stoppé si la première connection à son port provoque l'invocation de l'option twist. Le risque est qu'une attaque provoque l'arrêt des daemons. Cette situation est évitée si votre fichier /etc/hosts.allow contient un jeu de règles approprié. L'exemple suivant est suffisant :
undef-fd : localhost : allow undef-sd : localhost : allow undef-dir : localhost : allow undef-fd : ALL : deny undef-sd : ALL : deny undef-dir : ALL : deny
Vous devez accorder les noms des daemons à ceux spécifiés dans leurs fichiers de configuration respectifs. Ce ne sont, en général, pas les noms des fichiers binaires des daemons. Il n'est pas possible d'utiliser les noms des binaires car plusieurs daemons peuvent être exécutés sur une machine avec des fichiers de configuration distincts.
Dans ces exemples, le Director est undef-dir, le Storage Daemon est undef-sd, et le File Daemon est undef-fd. Ajustez ces noms pour qu'ils conviennent à votre configuration. L'exemple de règles ci-dessus suppose que SD, FD et DIR sont tous sur la même machine. Si vous avez un client FD distant, il vous suffira de placer le jeu de règles suivant sur ce client :
undef-fd : director.example.org : allow undef-fd : ALL : deny
Où director.example.org est l'hôte qui contactera le client (i.e. la machine sur laquelle le Bacula Director tourne). L'usage de "ALL : deny" assure que l'option twist (si présente) n'est pas invoquée. Pour tester correctement votre configuration, démarrez le(s) daemon(s), puis essayez de vous y connecter depuis une adresse IP qui devrait être capable de le faire. Vous devriez voir quelque chose comme :
$ telnet undef 9103 Trying 192.168.0.56... Connected to undef.example.org. Escape character is '^]'. Connection closed by foreign host. $
C'est la réponse correcte. Si vous voyez ceci :
$ telnet undef 9103 Trying 192.168.0.56... Connected to undef.example.org. Escape character is '^]'. You are not welcome to use undef-sd from xeon.example.org. Connection closed by foreign host. $
Alors, twist a été invoquée, et votre configuration est incorrecte. vous devez ajouter la directive "deny". Il est important de noter que vos tests doivent inclure le redémarrage des daemons après chaque tentative de connexion. Vous pouvez aussi tcpdchk(8) et tcpdmatch(8) pour valider jeu de règles /etc/hosts.allow. Voici un test simple avec tcpdmatch :
$ tcpdmatch undef-dir xeon.example.org warning: undef-dir: no such process name in /etc/inetd.conf client: hostname xeon.example.org client: address 192.168.0.18 server: process undef-dir matched: /etc/hosts.allow line 40 option: allow access: granted
Si vous exécutez Bacula en tant que standalone daemon, les avertissements ci-dessus peuvent être ignorés sans scrupules. Voici un exemple qui révèle que "deny" fait defaut à vos règles, et que l'option twist a été invoquée.
$ tcpdmatch undef-dir 10.0.0.1 warning: undef-dir: no such process name in /etc/inetd.conf client: address 10.0.0.1 server: process undef-dir matched: /etc/hosts.allow line 91 option: severity auth.info option: twist /bin/echo "You are not welcome to use undef-dir from 10.0.0.1." access: delegated
Voici quelques recommandations de Dan Languille :
C'est une bonne idée d'exécuter vos daemons avec des privilèges aussi faibles que possible. En d'autres termes, si vous pouvez, n'exécutez pas d'applications en tant que root si elle n'ont pas besoin d'être exécutées en tant que root. Le Storage Daemon et le Director Daemon n'ont pas besoin d'être exécutés en tant que root. Le File Daemon en a besoin pour accéder à l'ensemble des fichiers du système. Pour vous passer des privilèges root, il vous faut créer un utilisateur et un groupe. Choisir bacula pour l'un et l'autre me semble une bonne idée.
Le port FreeBSD crée cet utilisateur et ce groupe pour vous. (En fait, au moment ou j'écris ces lignes, ce n'est pas encore le cas, mais ça le sera bientôt). Voici à quoi ressemblent ces entrées sur mon portable FreeBSD :
bacula:*:1002:1002::0:0:Bacul Daemon:/var/db/bacula:/sbin/nologin
J'ai utilisé vipw pour créer ces entrées. J'ai utilisé un User ID et un Group ID disponibles sur mon système : 1002.
J'ai aussi créé un groupe dans /etc/group:
bacula:*:1002:
L'utilisateur bacula, contrairement au daemon Bacula, aura un répertoire dédié (home directory) : /var/db/bacula qui est le répertoire standard pour le catalogue de Bacula.
A présent, vous avez un utilisateur et un groupe bacula, et vous pouvez sécuriser le répertoire dédié de bacula en utilisant cette commande :
chown -R bacula:bacula /var/db/bacula/
Celle-ci assure que seul l'utilisateur bacula peut accéder à ce répertoire. Elle signifie aussi que si nous exécutons le Director et le Storage Daemon en tant que bacula, ces daemons auront aussi des accès restreints. Ce ne serait pas le cas s'ils étaient exécutés en tant que root.
Il est important de noter que le Storage Daemon a vraiment besoin d'appartenir au groupe operator pour un accès normal aux lecteurs de bandes. (au moins sur FreeBSD, c'est ainsi que les choses sont configurées par défaut). De tels périphériques sont en principe attribués à root:operator. Il est plus facile et moins dangereux de faire de bacula un membre de ce groupe que de jouer avec les permissions du système.
Démarrer les daemons bacula
Pour démarrer les daemons bacula sur FreeBSD, utilisez la commande :
/usr/local/etc/rc.d/bacula.sh start
Pour vous assurer que tous fonctionnent :
$ ps auwx | grep bacula root\ 63416\ 0.0\ 0.3\ 2040 1172\ ??\ Ss\ 4:09PM 0:00.01 /usr/local/sbin/bacula-sd -v -c /usr/local/etc/bacula-sd.conf root\ 63418\ 0.0\ 0.3\ 1856 1036\ ??\ Ss\ 4:09PM 0:00.00 /usr/local/sbin/bacula-fd -v -c /usr/local/etc/bacula-fd.conf root\ 63422\ 0.0\ 0.4\ 2360 1440\ ??\ Ss\ 4:09PM 0:00.00 /usr/local/sbin/bacula-dir -v -c /usr/local/etc/bacula-dir.conf
If you have a firewall or a DMZ installed on your computer, you may experience difficulties contacting one or more of the Clients to back them up. This is especially true if you are trying to backup a Client across the Internet.
If you are attempting to do this, the sequence of network events in Bacula to do a backup are the following:
Console -> DIR:9101 DIR -> SD:9103 DIR -> FD:9102 FD -> SD:9103
Where it should be obvious that DIR represents the Director, FD the File daemon or client, and SD the Storage daemon. The numbers that follow those names are the standard ports used by Bacula, and the -> represents the left side making a connection to the right side (i.e. the right side is the "server" or is listening on the specified port), and the left side is the "client" who initiates the conversation.
Note, port 9103 serves both the Director and the File daemon, each having its own independent connection.
If you are running iptables, you might add something like:
-A FW-1-INPUT -m state --state NEW -m tcp -p tcp --dport 9101:9103 -j ACCEPT
on your server, and
-A FW-1-INPUT -m state --state NEW -m tcp -p tcp --dport 9102 -j ACCEPT
on your client. In both cases, I assume that the machine is allowed to initiate connections on any port. If not, you will need to allow outgoing connections on ports 9102 and 9103 on your server and 9103 on your client. Thanks to Raymond Norton for this tip.
Jesse Guardiani's solution for his network for this problem, in his own words, is:
My bacula server is on the 192.168.1.0/24 network at IP address 192.168.1.52. For the sake of discussion we will refer to this network as the 'internal' network because it connects to the internet through a NAT'd firewall. We will call the network on the public (internet) side of the NAT'd firewall the 'external' network. Also, for the sake of discussion we will call my bacula server:
server.int.mydomain.tld
when a fully qualified domain name is required, or simply:
server
if a hostname is adequate. We will call the various bacula daemons running on the server.int.mydomain.tld machine:
server-fd server-sd server-dir
In addition, I have two clients that I want to back up with Bacula. The first client is on the internal network. Its fully qualified domain name is:
private1.int.mydomain.tld
And its hostname is:
private1
This machine is a client and therefore runs just one bacula daemon:
private1-fd
The second client is on the external network. Its fully qualified domain name is:
public1.mydomain.tld
And its hostname is:
public1
This machine also runs just one bacula daemon:
public1-fd
Finally, I have a NAT firewall/gateway with two network interfaces. The first interface is on the internal network and serves as a gateway to the internet for all the machines attached to the internal network (For example, server.int.mydomain.tld and private1.int.mydomain.tld). The second interface is on the external (internet) network. The external interface has been assigned the name:
firewall.mydomain.tld
Remember:
*.int.mydomain.tld = internal network *.mydomain.tld = external network
server-sd manages a 4 tape AIT autoloader. All of my backups are written to server-sd. I have just *one* Device resource in my server-sd.conf file:
Device { Name = "autochanger1"; Media Type = AIT-1; Archive Device = /dev/nrsa1; Changer Device = /dev/ch0; Changer Command = "/usr/local/sbin/chio-bacula %c %o %S %a"; Label Media = yes; AutoChanger = yes; AutomaticMount = yes; # when device opened, read it AlwaysOpen = yes; Hardware End of Medium = No Fast Forward Space File = No BSF at EOM = yes }
(note, please see the Tape Testing chapter of this manual for important FreeBSD information.) However, I have *two* Storage resources in my server-dir.conf file:
Storage { Name = "autochanger1-int" # Storage device for backing up Address = server.int.mydomain.tld SDPort = 9103 Password = "mysecretpassword" Device = "autochanger1" Media Type = AIT-1 Autochanger = yes } Storage { Name = "autochanger1-ext" # Storage device for backing up Address = firewall.mydomain.tld SDPort = 9103 Password = "mysecretpassword" Device = "autochanger1" Media Type = AIT-1 Autochanger = yes }
Note that BOTH of the above server-dir.conf Storage resources use the same 'autochanger1' Device resource from server-sd.conf.
My backup jobs run consecutively, one after the other, so only one of the above Storage resources is being used by Bacula file daemons at any given time. I don't know if this would cause problems at a site that runs more than one backup in parallel to a single tape device.
In addition to the above, I have two Client resources defined in server-dir.conf:
Client { Name = private1-fd Address = private1.int.mydomain.tld FDPort = 9102 Catalog = MyCatalog Password = "mysecretpassword" # password for FileDaemon } Client { Name = public1-fd Address = public1.mydomain.tld FDPort = 9102 Catalog = MyCatalog Password = "mysecretpassword" # password for FileDaemon }
And finally, to tie it all together, I have two Job resources defined in server-dir.conf:
Job { Name = "Private1-Backup" Type = Backup Client = private1-fd FileSet = "Private1" Schedule = "WeeklyCycle" Storage = "autochanger1-int" Messages = Standard Pool = "Weekly" Write Bootstrap = "/var/db/bacula/Private1-Backup.bsr" Priority = 12 } Job { Name = "Public1-Backup" Type = Backup Client = public1-fd FileSet = "Public1" Schedule = "WeeklyCycle" Storage = "autochanger1-ext" Messages = Standard Pool = "Weekly" Write Bootstrap = "/var/db/bacula/Public1-Backup.bsr" Priority = 13 }
It is important to notice that because the 'Private1-Backup' Job is intended to back up a machine on the internal network it uses the 'autochanger1-int' Storage resource. On the other hand, the 'Public1-Backup' Job is intended to back up a machine on the external network, so it uses the 'autochanger1-ext' Storage resource.
I have left the Pool, Catalog, Messages, FileSet, Schedule, and Director resources out of the above server-dir.conf examples because they are not pertinent to the discussion.
If I want to run a backup of private1.int.mydomain.tld and store that backup using server-sd then my understanding of the order of events is this:
Alternatively, if I want to run a backup of public1.mydomain.tld and store that backup using server-sd then my understanding of the order of events is this:
In order for the above 'Public1-Backup' Job to succeed, firewall.mydomain.tld:9103 MUST be forwarded using the firewall's configuration software to server.int.mydomain.tld:9103. Some firewalls call this 'Server Publication'. Others may call it 'Port Forwarding'.
Also, if you have denial of service rate limiting in your firewall, this too can cause Bacula disconnects since Bacula can at times use very high access rates. To avoid this, you should implement default accept rules for the Bacula ports involved before the rate limiting rules.
Finally, if you have a Windows machine, it will most likely by default disallow connections to the Bacula Windows File daemon. See the Windows chapter of this manual for additional details.
Since Bacula maintains a catalog of files, their attributes, and either SHA1 or MD5 signatures, it can be an ideal tool for improving computer security. This is done by making a snapshot of your system files with a Verify Job and then checking the current state of your system against the snapshot, on a regular basis (e.g. nightly).
The first step is to set up a Verify Job and to run it with:
Level = InitCatalog
The InitCatalog level tells Bacula simply to get the information on the specified files and to put it into the catalog. That is your database is initialized and no comparison is done. The InitCatalog is normally run one time manually.
Thereafter, you will run a Verify Job on a daily (or whatever) basis with:
Level = Catalog
The Level = Catalog level tells Bacula to compare the current state of the files on the Client to the last InitCatalog that is stored in the catalog and to report any differences. See the example below for the format of the output.
You decide what files you want to form your "snapshot" by specifying them in a FileSet resource, and normally, they will be system files that do not change, or that only certain features change.
Then you decide what attributes of each file you want compared by specifying comparison options on the Include statements that you use in the FileSet resource of your Catalog Jobs.
In the discussion that follows, we will make reference to the Verify Configuration Example that is included below in the A Verify Configuration Example section. You might want to look it over now to get an idea of what it does.
The main elements consist of adding a schedule, which will normally be run daily, or perhaps more often. This is provided by the VerifyCycle Schedule, which runs at 5:05 in the morning every day.
Then you must define a Job, much as is done below. We recommend that the Job name contain the name of your machine as well as the word Verify or Check. In our example, we named it MatouVerify. This will permit you to easily identify your job when running it from the Console.
You will notice that most records of the Job are quite standard, but that the FileSet resource contains verify=pins1 option in addition to the standard signature=SHA1 option. If you don't want SHA1 signature comparison, and we cannot imagine why not, you can drop the signature=SHA1 and none will be computed nor stored in the catalog. Or alternatively, you can use verify=pins5 and signature=MD5, which will use the MD5 hash algorithm. The MD5 hash computes faster than SHA1, but is cryptographically less secure.
The verify=pins1 is ignored during the InitCatalog Job, but is used during the subsequent Catalog Jobs to specify what attributes of the files should be compared to those found in the catalog. pins1 is a reasonable set to begin with, but you may want to look at the details of these and other options. They can be found in the FileSet Resource section of this manual. Briefly, however, the p of the pins1 tells Verify to compare the permissions bits, the i is to compare inodes, the n causes comparison of the number of links, the s compares the file size, and the 1 compares the SHA1 checksums (this requires the signature=SHA1 option to have been set also).
You must also specify the Client and the Catalog resources for your Verify job, but you probably already have them created for your client and do not need to recreate them, they are included in the example below for completeness.
As mentioned above, you will need to have a FileSet resource for the Verify job, which will have the additional verify=pins1 option. You will want to take some care in defining the list of files to be included in your FileSet. Basically, you will want to include all system (or other) files that should not change on your system. If you select files, such as log files or mail files, which are constantly changing, your automatic Verify job will be constantly finding differences. The objective in forming the FileSet is to choose all unchanging important system files. Then if any of those files has changed, you will be notified, and you can determine if it changed because you loaded a new package, or because someone has broken into your computer and modified your files. The example below shows a list of files that I use on my Red Hat 7.3 system. Since I didn't spend a lot of time working on it, it probably is missing a few important files (if you find one, please send it to me). On the other hand, as long as I don't load any new packages, none of these files change during normal operation of the system.
The first thing you will want to do is to run an InitCatalog level Verify Job. This will initialize the catalog to contain the file information that will later be used as a basis for comparisons with the actual file system, thus allowing you to detect any changes (and possible intrusions into your system).
The easiest way to run the InitCatalog is manually with the console program by simply entering run. You will be presented with a list of Jobs that can be run, and you will choose the one that corresponds to your Verify Job, MatouVerify in this example.
The defined Job resources are: 1: MatouVerify 2: kernsrestore 3: Filetest 4: kernsave Select Job resource (1-4): 1
Next, the console program will show you the basic parameters of the Job and ask you:
Run Verify job JobName: MatouVerify FileSet: Verify Set Level: Catalog Client: MatouVerify Storage: DLTDrive OK to run? (yes/mod/no): mod
Here, you want to respond mod to modify the parameters because the Level is by default set to Catalog and we want to run an InitCatalog Job. After responding mod, the console will ask:
Parameters to modify: 1: Job 2: Level 3: FileSet 4: Client 5: Storage Select parameter to modify (1-5): 2
you should select number 2 to modify the Level, and it will display:
Levels: 1: Initialize Catalog 2: Verify from Catalog 3: Verify Volume 4: Verify Volume Data Select level (1-4): 1
Choose item 1, and you will see the final display:
Run Verify job JobName: MatouVerify FileSet: Verify Set Level: Initcatalog Client: MatouVerify Storage: DLTDrive OK to run? (yes/mod/no): yes
at which point you respond yes, and the Job will begin.
Thereafter the Job will automatically start according to the schedule you have defined. If you wish to immediately verify it, you can simply run a Verify Catalog which will be the default. No differences should be found.
If you have setup your messages correctly, you should be notified if there are any differences and exactly what they are. For example, below is the email received after doing an update of OpenSSH:
HeadMan: Start Verify JobId 83 Job=RufusVerify.2002-06-25.21:41:05 HeadMan: Verifying against Init JobId 70 run 2002-06-21 18:58:51 HeadMan: File: /etc/pam.d/sshd HeadMan: st_ino differ. Cat: 4674b File: 46765 HeadMan: File: /etc/rc.d/init.d/sshd HeadMan: st_ino differ. Cat: 56230 File: 56231 HeadMan: File: /etc/ssh/ssh_config HeadMan: st_ino differ. Cat: 81317 File: 8131b HeadMan: st_size differ. Cat: 1202 File: 1297 HeadMan: SHA1 differs. HeadMan: File: /etc/ssh/sshd_config HeadMan: st_ino differ. Cat: 81398 File: 81325 HeadMan: st_size differ. Cat: 1182 File: 1579 HeadMan: SHA1 differs. HeadMan: File: /etc/ssh/ssh_config.rpmnew HeadMan: st_ino differ. Cat: 812dd File: 812b3 HeadMan: st_size differ. Cat: 1167 File: 1114 HeadMan: SHA1 differs. HeadMan: File: /etc/ssh/sshd_config.rpmnew HeadMan: st_ino differ. Cat: 81397 File: 812dd HeadMan: st_size differ. Cat: 2528 File: 2407 HeadMan: SHA1 differs. HeadMan: File: /etc/ssh/moduli HeadMan: st_ino differ. Cat: 812b3 File: 812ab HeadMan: File: /usr/bin/scp HeadMan: st_ino differ. Cat: 5e07e File: 5e343 HeadMan: st_size differ. Cat: 26728 File: 26952 HeadMan: SHA1 differs. HeadMan: File: /usr/bin/ssh-keygen HeadMan: st_ino differ. Cat: 5df1d File: 5e07e HeadMan: st_size differ. Cat: 80488 File: 84648 HeadMan: SHA1 differs. HeadMan: File: /usr/bin/sftp HeadMan: st_ino differ. Cat: 5e2e8 File: 5df1d HeadMan: st_size differ. Cat: 46952 File: 46984 HeadMan: SHA1 differs. HeadMan: File: /usr/bin/slogin HeadMan: st_ino differ. Cat: 5e359 File: 5e2e8 HeadMan: File: /usr/bin/ssh HeadMan: st_mode differ. Cat: 89ed File: 81ed HeadMan: st_ino differ. Cat: 5e35a File: 5e359 HeadMan: st_size differ. Cat: 219932 File: 234440 HeadMan: SHA1 differs. HeadMan: File: /usr/bin/ssh-add HeadMan: st_ino differ. Cat: 5e35b File: 5e35a HeadMan: st_size differ. Cat: 76328 File: 81448 HeadMan: SHA1 differs. HeadMan: File: /usr/bin/ssh-agent HeadMan: st_ino differ. Cat: 5e35c File: 5e35b HeadMan: st_size differ. Cat: 43208 File: 47368 HeadMan: SHA1 differs. HeadMan: File: /usr/bin/ssh-keyscan HeadMan: st_ino differ. Cat: 5e35d File: 5e96a HeadMan: st_size differ. Cat: 139272 File: 151560 HeadMan: SHA1 differs. HeadMan: 25-Jun-2002 21:41 JobId: 83 Job: RufusVerify.2002-06-25.21:41:05 FileSet: Verify Set Verify Level: Catalog Client: RufusVerify Start time: 25-Jun-2002 21:41 End time: 25-Jun-2002 21:41 Files Examined: 4,258 Termination: Verify Differences
At this point, it was obvious that these files were modified during installation of the RPMs. If you want to be super safe, you should run a Verify Level=Catalog immediately before installing new software to verify that there are no differences, then run a Verify Level=InitCatalog immediately after the installation.
To keep the above email from being sent every night when the Verify Job runs, we simply re-run the Verify Job setting the level to InitCatalog (as we did above in the very beginning). This will re-establish the current state of the system as your new basis for future comparisons. Take care that you don't do an InitCatalog after someone has placed a Trojan horse on your system!
If you have included in your FileSet a file that is changed by the normal operation of your system, you will get false matches, and you will need to modify the FileSet to exclude that file (or not to Include it), and then re-run the InitCatalog.
The FileSet that is shown below is what I use on my Red Hat 7.3 system. With a bit more thought, you can probably add quite a number of additional files that should be monitored.
Schedule { Name = "VerifyCycle" Run = Level=Catalog sun-sat at 5:05 } Job { Name = "MatouVerify" Type = Verify Level = Catalog # default level Client = MatouVerify FileSet = "Verify Set" Messages = Standard Storage = DLTDrive Pool = Default Schedule = "VerifyCycle" } # # The list of files in this FileSet should be carefully # chosen. This is a good starting point. # FileSet { Name = "Verify Set" Include { Options { verify=pins1 signature=SHA1 } File = /boot File = /bin File = /sbin File = /usr/bin File = /lib File = /root/.ssh File = /home/kern/.ssh File = /var/named File = /etc/sysconfig File = /etc/ssh File = /etc/security File = /etc/exports File = /etc/rc.d/init.d File = /etc/sendmail.cf File = /etc/sysctl.conf File = /etc/services File = /etc/xinetd.d File = /etc/hosts.allow File = /etc/hosts.deny File = /etc/hosts File = /etc/modules.conf File = /etc/named.conf File = /etc/pam.d File = /etc/resolv.conf } Exclude = { } P Client { Name = MatouVerify Address = lmatou Catalog = Bacula Password = "" File Retention = 80d # 80 days Job Retention = 1y # one year AutoPrune = yes # Prune expired Jobs/Files } Catalog { Name = Bacula dbname = verify; user = bacula; password = "" }
%define rh7 0
and edit it to read
%define rh7 1
Alternately you may pass the define on the command line when calling rpmbuild:
rpmbuild -ba --define "build_rh7 1" bacula.spec rpmbuild --rebuild --define build_rh7 1" bacula-x.x.x-x.src.rpm
%define mysql 0 OR %define mysql4 0 OR %define mysql5 0
to
%define mysql 1 OR %define mysql4 1 OR %define mysql5 1
in the spec file directly or pass it to rpmbuild on the command line:
rpmbuild -ba --define "build_rh7 1" --define "build_mysql 1" bacula.spec rpmbuild -ba --define "build_rh7 1" --define "build_mysql4 1" bacula.spec rpmbuild -ba --define "build_rh7 1" --define "build_mysql5 1" bacula.spec
chmod -R 777 /usr/src/redhat chmod -R 777 /usr/src/RPM chmod -R 777 /usr/src/packages
If you are working on a shared system where you can not use the method above then you need to recreate the appropriate above directory tree with all of its subdirectories inside your home directory. Then create a file named
.rpmmacros
in your home directory (or edit the file if it already exists) and add the following line:
%_topdir /home/myuser/redhat
Another handy directive for the .rpmmacros file if you wish to suppress the creation of debug rpm packages is:
%debug_package %{nil}
chown bacula.bacula /var/bacula/* chown root.bacula /var/bacula/bacula-fd.9102.state chown bacula.disk /var/bacula/bacula-sd.9103.state
Further, if you are using File storage volumes rather than tapes those files will also need to have ownership set to user bacula and group bacula.
Build with one of these 3 commands: rpmbuild --rebuild \ --define "build_rhel4 1" \ --define "build_sqlite 1" \ bacula-1.38.3-1.src.rpm rpmbuild --rebuild \ --define "build_rhel4 1" \ --define "build_postgresql 1" \ bacula-1.38.3-1.src.rpm rpmbuild --rebuild \ --define "build_rhel4 1" \ --define "build_mysql4 1" \ bacula-1.38.3-1.src.rpm For CentOS substitute '--define "build_centos4 1"' in place of rhel4. For 64 bit support add '--define "build_x86_64 1"'
Red Hat builds --define "build_rh7 1" --define "build_rh8 1" --define "build_rh9 1" Fedora Core build --define "build_fc1 1" --define "build_fc3 1" --define "build_fc4 1" --define "build_fc5 1" --define "build_fc6 1" --define "build_fc7 1" Whitebox Enterprise build --define "build_wb3 1" Red Hat Enterprise builds --define "build_rhel3 1" --define "build_rhel4 1" CentOS build --define "build_centos3 1" --define "build_centos4 1" Scientific Linux build --define "build_sl3 1" --define "build_sl4 1" SuSE build --define "build_su9 1" --define "build_su10 1" --define "build_su102 1" Mandrake 10.x build --define "build_mdk 1" Mandriva build --define "build_mdv 1" MySQL support: for mysql 3.23.x support define this --define "build_mysql 1" if using mysql 4.x define this, currently: Mandrake 10.x, Mandriva 2006.0, SuSE 9.x & 10.0, FC4 & RHEL4 --define "build_mysql4 1" if using mysql 5.x define this, currently: SuSE 10.1 & FC5 --define "build_mysql5 1" PostgreSQL support: --define "build_postgresql 1" Sqlite support: --define "build_sqlite 1" Build the client rpm only in place of one of the above database full builds: --define "build_client_only 1" X86-64 support: --define "build_x86_64 1" Supress build of bgnome-console: --define "nobuild_gconsole 1" Build the WXWindows console: requires wxGTK >= 2.6 --define "build_wxconsole 1" Build the Bacula Administration Tool: requires QT >= 4 --define "build_bat 1" Build python scripting support: --define "build_python 1" Modify the Packager tag for third party packages: --define "contrib_packager Your Name <youremail@site.org>"
chown bacula:bacula /var/baculaNote: as of 1.38.8 /var/bacula is installed root:bacula with permissions 770.
The information in this chapter is provided so that you may either create your own bootstrap files, or so that you can edit a bootstrap file produced by Bacula. However, normally the bootstrap file will be automatically created for you during the restore command in the Console program, or by using a Write Bootstrap record in your Backup Jobs, and thus you will never need to know the details of this file.
The bootstrap file contains ASCII information that permits precise specification of what files should be restored. It is a relatively compact form of specifying the information, is human readable, and can be edited with any text editor.
The general format of a bootstrap file is:
<keyword>= <value>
Where each keyword and the value specify which files to restore. More precisely the keyword and their values serve to limit which files will be restored and thus act as a filter. The absence of a keyword means that all records will be accepted.
Blank lines and lines beginning with a pound sign (#) in the bootstrap file are ignored.
There are keywords which permit filtering by Volume, Client, Job, FileIndex, Session Id, Session Time, ...
The more keywords that are specified, the more selective the specification of which files to restore will be. In fact, each keyword is ANDed with other keywords that may be present.
For example,
Volume = Test-001 VolSessionId = 1 VolSessionTime = 108927638
directs the Storage daemon (or the bextract program) to restore only those files on Volume Test-001 AND having VolumeSessionId equal to one AND having VolumeSession time equal to 108927638.
The full set of permitted keywords presented in the order in which they are matched against the Volume records are:
This for a given Volume, the triple VolSessionId, VolSessionTime, and FileIndex uniquely identifies a file stored on the Volume. Multiple copies of the same file may be stored on the same Volume, but for each file, the triple VolSessionId, VolSessionTime, and FileIndex will be unique. This triple is stored in the Catalog database for each file.
The Volume record is a bit special in that it must be the first record. The other keyword records may appear in any order and any number following a Volume record.
Multiple Volume records may be specified in the same bootstrap file, but each one starts a new set of filter criteria for the Volume.
In processing the bootstrap file within the current Volume, each filter specified by a keyword is ANDed with the next. Thus,
Volume = Test-01 Client = "My machine" FileIndex = 1
will match records on Volume Test-01 AND Client records for My machine AND FileIndex equal to one.
Multiple occurrences of the same record are ORed together. Thus,
Volume = Test-01 Client = "My machine" Client = "Backup machine" FileIndex = 1
will match records on Volume Test-01 AND (Client records for My machine OR Backup machine) AND FileIndex equal to one.
For integer values, you may supply a range or a list, and for all other values except Volumes, you may specify a list. A list is equivalent to multiple records of the same keyword. For example,
Volume = Test-01 Client = "My machine", "Backup machine" FileIndex = 1-20, 35
will match records on Volume Test-01 AND (Client records for My machine OR Backup machine) AND (FileIndex 1 OR 2 OR 3 ... OR 20 OR 35).
As previously mentioned above, there may be multiple Volume records in the same bootstrap file. Each new Volume definition begins a new set of filter conditions that apply to that Volume and will be ORed with any other Volume definitions.
As an example, suppose we query for the current set of tapes to restore all files on Client Rufus using the query command in the console program:
Using default Catalog name=MySQL DB=bacula *query Available queries: 1: List Job totals: 2: List where a file is saved: 3: List where the most recent copies of a file are saved: 4: List total files/bytes by Job: 5: List total files/bytes by Volume: 6: List last 10 Full Backups for a Client: 7: List Volumes used by selected JobId: 8: List Volumes to Restore All Files: Choose a query (1-8): 8 Enter Client Name: Rufus +-------+------------------+------------+-----------+----------+------------+ | JobId | StartTime | VolumeName | StartFile | VolSesId | VolSesTime | +-------+------------------+------------+-----------+----------+------------+ | 154 | 2002-05-30 12:08 | test-02 | 0 | 1 | 1022753312 | | 202 | 2002-06-15 10:16 | test-02 | 0 | 2 | 1024128917 | | 203 | 2002-06-15 11:12 | test-02 | 3 | 1 | 1024132350 | | 204 | 2002-06-18 08:11 | test-02 | 4 | 1 | 1024380678 | +-------+------------------+------------+-----------+----------+------------+
The output shows us that there are four Jobs that must be restored. The first one is a Full backup, and the following three are all Incremental backups.
The following bootstrap file will restore those files:
Volume=test-02 VolSessionId=1 VolSessionTime=1022753312 Volume=test-02 VolSessionId=2 VolSessionTime=1024128917 Volume=test-02 VolSessionId=1 VolSessionTime=1024132350 Volume=test-02 VolSessionId=1 VolSessionTime=1024380678
As a final example, assume that the initial Full save spanned two Volumes. The output from query might look like:
+-------+------------------+------------+-----------+----------+------------+ | JobId | StartTime | VolumeName | StartFile | VolSesId | VolSesTime | +-------+------------------+------------+-----------+----------+------------+ | 242 | 2002-06-25 16:50 | File0003 | 0 | 1 | 1025016612 | | 242 | 2002-06-25 16:50 | File0004 | 0 | 1 | 1025016612 | | 243 | 2002-06-25 16:52 | File0005 | 0 | 2 | 1025016612 | | 246 | 2002-06-25 19:19 | File0006 | 0 | 2 | 1025025494 | +-------+------------------+------------+-----------+----------+------------+
and the following bootstrap file would restore those files:
Volume=File0003 VolSessionId=1 VolSessionTime=1025016612 Volume=File0004 VolSessionId=1 VolSessionTime=1025016612 Volume=File0005 VolSessionId=2 VolSessionTime=1025016612 Volume=File0006 VolSessionId=2 VolSessionTime=1025025494
One thing that is probably worth knowing: the bootstrap files that are generated automatically at the end of the job are not as optimized as those generated by the restore command. This is because the ones created at the end of the file, contain all files written to the Volume for that job. As a consequence, all the files saved to an Incremental or Differential job will be restored first by the Full save, then by any Incremental or Differential saves.
When the bootstrap file is generated for the restore command, only one copy (the most recent) of each file is restored.
So if you have spare cycles on your machine, you could optimize the bootstrap files by doing the following:
./console restore client=xxx select all no quit Backup bootstrap file.
The above will not work if you have multiple FileSets because that will be an extra prompt. However, the restore client=xxx select all builds the in-memory tree, selecting everything and creates the bootstrap file.
The no answers the Do you want to run this (yes/mod/no) question.
If you want to extract or copy a single Job, you can do it by selecting by JobId (code not tested) or better yet, if you know the VolSessionTime and the VolSessionId (printed on Job report and in Catalog), specifying this is by far the best. Using the VolSessionTime and VolSessionId is the way Bacula does restores. A bsr file might look like the following:
Volume="Vol001" VolSessionId=10 VolSessionTime=1080847820
If you know how many files are backed up (on the job report), you can enormously speed up the selection by adding (let's assume there are 157 files):
FileIndex=1-157 Count=157
Finally, if you know the File number where the Job starts, you can also cause bcopy to forward space to the right file without reading every record:
VolFile=20
There is nothing magic or complicated about a BSR file. Parsing it and properly applying it within Bacula *is* magic, but you don't need to worry about that.
If you want to see a *real* bsr file, simply fire up the restore command in the console program, select something, then answer no when it prompts to run the job. Then look at the file restore.bsr in your working directory.
If you use the ./configure --
with-mysql=mysql-directory statement for
configuring Bacula, you will need MySQL version 3.23.53 or later
installed in the mysql-directory.
Bacula has been tested on MySQL version 4.1.12 and works providing
you are running it in the default installation that is compatible
with MySQL 3.23.x. If you are using one of the new modes such
as ANSI/ISO compatibility, you may experience problems.
If MySQL is installed in the standard system location, you need only enter
--
with-mysql since the configure program will search all the
standard locations. If you install MySQL in your home directory or some
other non-standard directory, you will need to provide the full path to it.
Installing and Configuring MySQL is not difficult but can be confusing the first time. As a consequence, below, we list the steps that we used to install it on our machines. Please note that our configuration leaves MySQL without any user passwords. This may be an undesirable situation if you have other users on your system.
Beginning with Bacula version 1.31, the thread safe version of the
MySQL client library is used, and hence you must add the --
enable-thread-safe-client option to the ./configure as shown below:
tar xvfz mysql-filename
Note, the above command requires GNU tar. If you do not have GNU tar, a command such as:
zcat mysql-filename | tar xvf -
will probably accomplish the same thing.
where you replace mysql-source-directory with the directory name where you put the MySQL source code.
--
enable-thread-safe-client --
prefix=mysql-directory
where you replace mysql-directory with the directory name where you want to install mysql. Normally for system wide use this is /usr/local/mysql. In my case, I use kern/mysql.
This takes a bit of time.
This will put all the necessary binaries, libraries and support files into the mysql-directory that you specified above.
This will create the necessary MySQL databases for controlling user access. Note, this script can also be found in the bin directory in the installation directory
The MySQL client library mysqlclient requires the gzip compression library libz.a or libz.so. If you are using rpm packages, these libraries are in the libz-devel package. On Debian systems, you will need to load the zlib1g-dev package. If you are not using rpms or debs, you will need to find the appropriate package for your system.
At this point, you should return to completing the installation of Bacula. Later after Bacula is installed, come back to this chapter to complete the installation. Please note, the installation files used in the second phase of the MySQL installation are created during the Bacula Installation.
At this point, you should have built and installed MySQL, or already have a running MySQL, and you should have configured, built and installed Bacula. If not, please complete these items before proceeding.
Please note that the ./configure used to build Bacula will need to
include --
with-mysql=mysql-directory, where mysql-directory is the
directory name that you specified on the ./configure command for configuring
MySQL. This is needed so that Bacula can find the necessary include headers
and library files for interfacing to MySQL.
Bacula will install scripts for manipulating the database (create, delete, make tables etc) into the main installation directory. These files will be of the form *_bacula_* (e.g. create_bacula_database). These files are also available in the <bacula-src>/src/cats directory after running ./configure. If you inspect create_bacula_database, you will see that it calls create_mysql_database. The *_bacula_* files are provided for convenience. It doesn't matter what database you have chosen; create_bacula_database will always create your database.
Now you will create the Bacula MySQL database and the tables that Bacula uses.
--
prefix option. This can be important to
know if you want to make a special backup of the Bacula database or to
check its size.
Each of the three scripts (grant_mysql_privileges, create_mysql_database and make_mysql_tables) allows the addition of a command line argument. This can be useful for specifying the user and or password. For example, you might need to add -u root to the command line to have sufficient privilege to create the Bacula tables.
To take a closer look at the access privileges that you have setup with the above, you can do:
mysql-directory/bin/mysql -u root mysql select * from user;
After you have done some initial testing with Bacula, you will probably want to re-initialize the catalog database and throw away all the test Jobs that you ran. To do so, you can do the following:
cd <install-directory> ./drop_mysql_tables ./make_mysql_tables
Please note that all information in the database will be lost and you will be starting from scratch. If you have written on any Volumes, you must write an end of file mark on the volume so that Bacula can reuse it. Do so with:
(stop Bacula or unmount the drive) mt -f /dev/nst0 rewind mt -f /dev/nst0 weof
Where you should replace /dev/nst0 with the appropriate tape drive device name for your machine.
After configuring Bacula with
./configure --
enable-thread-safe-client --
prefix=<mysql-directory>
where <mysql-directory> is in my case /home/kern/mysql, you may
have to configure the loader so that it can find the MySQL shared libraries.
If you have previously followed this procedure and later add the --
enable-thread-safe-client options, you will need to rerun the ldconfig program shown below. If you put MySQL in a standard place such as
/usr/lib or /usr/local/lib this will not be necessary, but in my
case it is. The description that follows is Linux specific. For other
operating systems, please consult your manuals on how to do the same thing:
First edit: /etc/ld.so.conf and add a new line to the end of the file with the name of the mysql-directory. In my case, it is:
/home/kern/mysql/lib/mysql then rebuild the loader's cache with:
/sbin/ldconfig If you upgrade to a new version of MySQL, the shared library names will probably change, and you must re-run the /sbin/ldconfig command so that the runtime loader can find them.
Alternatively, your system my have a loader environment variable that can be set. For example, on a Solaris system where I do not have root permission, I use:
LD_LIBRARY_PATH=/home/kern/mysql/lib/mysql
Finally, if you have encryption enabled in MySQL, you may need to add -lssl -lcrypto to the link. In that case, you can either export the appropriate LDFLAGS definition, or alternatively, you can include them directly on the ./configure line as in:
LDFLAGS="-lssl -lcyrpto" \ ./configure \ <your-options>
mysql mysql-devel
This will be the same with most other package managers too.
Attention !!! Si vous envisagez d'utiliser PostgreSQL, vous devriez être conscient de la philosophie des mises à jour de PostgreSQL qui peut être déstabilisant dans un environnement de production. En gros, pour chaque mise à jour vers une version majeure, vous devez exporter vos bases de données au format ASCII, faire la mise à jour, et recharger vos bases de données. Ceci est dû au à des mises à jour fréquentes du "format de données" d'une version à l'autre, et aucun outil n'est fourni pour effectuer la conversion automatiquement. Si vous omettez d'exporter vos bases au format ASCII, elles peuvent devenir complètement inutiles si aucun des nouveaux outils ne peut y accéder en raison d'un changement de format, auquel cas le serveur PostgreSQL sera dans l'incapacité de démarrer.
Si vous avez utilisé l'option ./configure
--
with-postgresql=PostgreSQL-Directory pour configurer Bacula, vous
avez besoin d'installer la version 7.3 ou supérieure de PostgreSQL.
ATTENTION! Les versions préalables à la 7.3 ne fonctionnent pas avec
Bacula. Si PostgreSQL est installé dans ses répertoires sandards, seule
l'option --
with-postgresql est nécessaire, le programme de
configuration scrutant tous les répertoires standards. Si PostgreSQL est
installé dans votre répertoire de travail ou dans un répertoire
atypique, il faut préciser l'option --
with-postgresql suivie du
répertoire ad hoc.
Installer et configurer PostgreSQL n'est pas compliqué mais peut être déroutant la première fois. Si vous préférez, vous pouvez utiliser le paquet de votre distribution. Les paquets binaires sont disponibles sur la plupart des mirroirs de PostgreSQL.
Si vous préférez installer PostgreSQL à partir des sources, nous vous recommandons de suivre les instructions de la documentation PostgreSQL.
Si vous utilisez PostgreSQL pour FreeBSD, cet article vous sera peut être utile. Même si vous n'utilisez pas FreeBSD, l'article contient des informations utiles à la configuration et au paramétrage de PostgreSQL.
Après l'installation de PostgreSQL, terminez l'installation de Bacula. Ensuite, quand Bacula sera installé, reprenez ce chapitre pour terminer l'installation. Notez que les fichiers d'installation utilisés dans cette seconde phase de l'installation de PostgreSQL sont créés durant l'installation de Bacula.
Si vous en êtes là, vous avez construit et installé PostgreSQL, ou vous aviez déjà un serveur PostgreSQL existant et vous avez configuré et installé Bacula. Dans le cas contraire, nous vous invitons à le faire avant de poursuivre.
Notez bien que la commande ./configure utilisée pour
construire Bacula nécessite d'ajouter l'option --
with-postgresql=repertoire_de_PostgreSQL, où repertoire_de_PostgreSQL spécifie le chemin de PostgreSQL indiqué à
la commande ./configure. (si vous n'avez pas spécifié de répertoire ou
si PostgreSQL est installé dans son répertoire par défaut, cette option
n'est pas nécessaire). Cette option est nécessaire pour que Bacula puisse
trouver les fichiers d'en-tête et les librairies d'interface à PostgreSQL.
Bacula installe les scripts pour la gestion de la base de données (créer, détruire, créer les tables, etc.) dans le répertoire principal de l'installation. Ces fichiers sont de la forme *_bacula_* (par exemple create_bacula_database). Ces fichiers sont également disponibles dans le répertoire <bacula-src>/src/cats après que la commande ./configure ait été lancée. Si vous consultez le fichier create_bacula_database, vous verrez qu'il fait appel à create_postgresql_database. Les fichiers *_bacula_* sont fournis pour faciliter les choses. Peu importe la base de données choisie, create_bacula_database créera la base de données.
Maintenant vous allez créer la base de données PostgreSQL et les tables utilisées par Bacula. On présume dans la suite que votre serveur PostgreSQL fonctionne. Vous devez exécuter les différentes étapes ci-dessous en tant qu'utilisateur autorisé à créer des bases. Ceci peut être fait avec l'utilisateur PostgreSQL (sur la plupart des systèmes il s'agit de pgsql. NDT: sur debian il s'agit de postgres)
Ce répertoire contient le catalogue des routines d'interfaces.
Ce script créé le catalogue bacula PostgreSQL. S'il échoue, c'est probablement que vous n'avez pas les droits requis sur la base de données. Sur la plupart des systèmes, le propriétaire de la base de données est pgsql, et sur d'autres tels que RedHat ou Fedora, c'est postgres. Vous pouvez déterminer lequel en examinant le fichier /etc/passwd. Pour créer un nouvel utilisateur avec votre nom ou le nom bacula, vous pouvez faire ce qui suit :
su (entrez le mot de passe root) su pgsql (ou postgres) createuser kern (ou peut-\^etre bacula) Shall the new user be allowed to create databases? (y/n) y Shall the new user be allowed to create more new users? (y/n) (choisissez ce que vous voulez) exit
A ce stade, vous devriez pouvoir exécuter la commande ./create_bacula_database
Créée les tables utilisées par Bacula.
Créée l'utilisateur de la base de données bacula avec des droits d'accès restreints. Vous pouvez modifier ce script pour cadrer avec votre propre configuration. Attention, cette base n'est pas protégée par un mot de passe.
Chacun de ces scripts (create_bacula_database, make_bacula_tables et grant_bacula_privileges) permet l'ajout d'arguments en ligne de commande. Ceci peut être utile pour spécifier le nom de l'utilisateur. Par exemple, vous pouvez avoir besoin d'ajouter -h nom_d_hote à la ligne de commande pour spécifier le serveur de base de données distant.
Pour avoir un bon aperçu des droits d'accès que vous avez spécifié vous pouvez utiliser la commande
repertoire_de_PostgreSQL/bin/psql --command \\dp bacula
J'ai rencontré un problème de permissions avec le mot de passe. J'ai finalement du modifier mon fichier pg_hba.conf (situé dans /var/lib/pgsql/data sur ma machine) :
de local all all ident sameuser vers local all all trust sameuser
Ceci a résolu le problème pour moi, mais ce n'est pas pas forcément une bonne chose du point de vue de la sécurité, mais j'ai ainsi pu exécuter mes scripts de régression sans mot de passe.
Un moyen plus sécurisé pour l'authentification auprès de la base de données consiste à utiliser le hachage MD5 des mots de passe. Pour cela, éditez les fichier pg_hba.conf, et ajoutez ajoutez ce qui suit juste avant les lignes "local" et "host" existantes :
local bacula bacula md5
Puis redémarrez le daemon Postgres (la plupart du temps, avec "/etc/init.d/postgresql restart") pour activer cette nouvelle règle d'authentification.
Ensuite, en tant qu'administrateur Postgres (connectez-vous en tant qu'utilisateur postgres ou en utilisant su pour devenir root, puis su postgres), ajoutez un mot de passe à la base de données bacula pour l'utilisateur bacula avec les commandes suivantes :
\$ psql bacula bacula=# alter user bacula with password 'secret'; ALTER USER bacula=# \\q
Enfin, il vous faudra ajouter ce mot de passe en deux endroits du fichier bacula-dir.conf : au niveau de la ressource Catalog et au niveau de la directive RunBeforeJob de la ressource Job BackupCatalog. Avec les mots de passe en place, ces deux lignes devraient ressembler à ceci :
dbname = bacula; user = bacula; password = "secret" ... and ... RunBeforeJob = "/etc/make_catalog_backup bacula bacula secret"
Naturellement, vous devriez choisir un meilleur mot de passe, et vous assurer que le fichier bacula-dir.conf qui contient ce mot de passe n'est lisible que par root.
Même avec ces restrictions, il reste un problème de sécurité avec cette approche : sur certaines plateformes, la variable d'environnement utilisée pour soumettre le mot de passe à Postgres est disponible pour tout utilisateur local du système. Pour supprimer ce problème, l'équipe Postgres a décrété obsolète ce mécanisme de passage de mot de passe par variable d'environnement et recommande d'utiliser un fichier .pgpass. Pour utiliser ce mécanisme, créez un fichier nommé .pgpass vcontenant une simple ligne :
localhost:5432:bacula:bacula:secret
Ce fichier devrait être copié dans les répertoires personnels (NDT : home directories) de tous les comptes susceptibles d'avoir besoin d'accéder à la base de données : typiquement, il s'agit de root, bacula et tout utilisateur de la console Bacula. Les fichiers doivent appartenir aux utilisateur et groupe correspondant : root:root pour la copie dans root, etc. Les permissions doivent être positionnées à 600 pour limiter l'accès au propriétaire du fichier.
Après avoir fait un certain nombre de tests avec Bacula, vous aurez très certainement envie de nettoyer le catalogue des sauvegardes et faire disparaître tous les travaux de tests que vous avez lancés. Pour ce faire, vous pouvez exécuter les commandes suivantes:
cd <r\'epertoire_d_installation> ./drop_bacula_tables ./make_bacula_tables ./grant_bacula_privileges
Attention! Toutes les informations contenues dans cette base seront perdues et vous repartirez de zéro. Si vous avez écrit sur certains volumes (média de sauvegarde), vous devrez écrire une marque de fin de fichier (EOF) sur chacun d'eux afin que Bacula puisse les réutiliser. Pour ce faire:
(arr\^eter Baula ou demonter les volumes) mt -f /dev/nst0 rewind mt -f /dev/nst0 weof
où vous devrez remplacer /dev/nst0 par le chemin approprié de votre lecteur de sauvegarde.
postgresql postgresql-devel
Il en va de même avec la plupart des gestionnaires de paquetages.
La procédure de migration présentée ici à fonctionné pour Norm Dressler <ndressler at dinmar dot com>
Ce process a été testé en utilisant les versions suivantes des différents logiciels:
ATTENTION! Par précaution, réalisez une sauvegarde complète de vos systèmes avant de procéder à cette migration.
mysqldump -f -t -n >bacula-backup.dmp
local all all trust host all all 127.0.0.1 255.255.255.255 trust ATTENTION: vous devez red\'emmarer PostgreSQL si vous faites des changements dans ce fichier.
./create_postgresql_database ./make_postgresql_tables ./grant_postgresql_privileges
psql -Ubacula bacula
Vous ne devriez avoir aucune erreur.
psql -Ubacula bacula <bacula-backup.dmp>
psql -Ubacula bacula SELECT SETVAL('basefiles_baseid_seq', (SELECT MAX(baseid) FROM basefiles)); SELECT SETVAL('client_clientid_seq', (SELECT MAX(clientid) FROM client)); SELECT SETVAL('file_fileid_seq', (SELECT MAX(fileid) FROM file)); SELECT SETVAL('filename_filenameid_seq', (SELECT MAX(filenameid) FROM filename)); SELECT SETVAL('fileset_filesetid_seq', (SELECT MAX(filesetid) FROM fileset)); SELECT SETVAL('job_jobid_seq', (SELECT MAX(jobid) FROM job)); SELECT SETVAL('jobmedia_jobmediaid_seq', (SELECT MAX(jobmediaid) FROM jobmedia)); SELECT SETVAL('media_mediaid_seq', (SELECT MAX(mediaid) FROM media)); SELECT SETVAL('path_pathid_seq', (SELECT MAX(pathid) FROM path)); SELECT SETVAL('pool_poolid_seq', (SELECT MAX(poolid) FROM pool));
If you upgrade PostgreSQL, you must reconfigure, rebuild, and re-install Bacula otherwise you are likely to get bizarre failures. If you to modify the bacula.spec file to account for the new PostgreSQL version. You can do so by rebuilding from the source rpm. To do so, you may need install from rpms and you upgrade PostgreSQL, you must also rebuild Bacula.
Tous mes remerciements à Dan Languille pour l'écriture du driver PostgreSQL qui deviendra très certainement la base de données la plus réputée utilisable avec Bacula
If you use the ./configure --
with-sqlite statement for configuring Bacula, you will need SQLite version 2.8.16 or later installed. Our standard
location (for the moment) for SQLite is in the dependency package depkgs/sqlite-2.8.16. Please note that the version will be updated as new
versions are available and tested.
You may install and use SQLite version 3.x with Bacula by using:
./configure --
with-sqlite3. You should ensure that
when the database is created that you have used
PRAGMA synchronous = NORMAL;otherwiset SQLite version 3.x is 4 to 10 times slower than version 2.8.16.
Installing and Configuring is quite easy.
tar xvfz depkgs.tar.gz
Note, the above command requires GNU tar. If you do not have GNU tar, a command such as:
zcat depkgs.tar.gz | tar xvf -
will probably accomplish the same thing.
At this point, you should return to completing the installation of Bacula.
Please note that the ./configure used to build Bacula will need to
include --
with-sqlite.
This phase is done after you have run the ./configure command to configure Bacula.
Bacula will install scripts for manipulating the database (create, delete, make tables etc) into the main installation directory. These files will be of the form *_bacula_* (e.g. create_bacula_database). These files are also available in the <bacula-src>/src/cats directory after running ./configure. If you inspect create_bacula_database, you will see that it calls create_sqlite_database. The *_bacula_* files are provided for convenience. It doesn't matter what database you have chosen; create_bacula_database will always create your database.
At this point, you can create the SQLite database and tables:
This directory contains the Bacula catalog interface routines.
This script creates the SQLite database as well as the tables used by Bacula. This script will be automatically setup by the ./configure program to create a database named bacula.db in Bacula's working directory.
If you have followed the above steps, this will all happen automatically and the SQLite libraries will be linked into Bacula.
We have much less "production" experience using SQLite than using MySQL. SQLite has performed flawlessly for us in all our testing. However, several users have reported corrupted databases while using SQLite. For that reason, we do not recommend it for production use.
If Bacula crashes with the following type of error when it is started:
Using default Catalog name=MyCatalog DB=bacula Could not open database "bacula". sqlite.c:151 Unable to open Database=/var/lib/bacula/bacula.db. ERR=malformed database schema - unable to open a temporary database file for storing temporary tables
this is most likely caused by the fact that some versions of SQLite attempt to create a temporary file in the current directory. If that fails, because Bacula does not have write permission on the current directory, then you may get this errr. The solution is to start Bacula in a current directory where it has write permission.
After you have done some initial testing with Bacula, you will probably want to re-initialize the catalog database and throw away all the test Jobs that you ran. To do so, you can do the following:
cd <install-directory> ./drop_sqlite_tables ./make_sqlite_tables
Please note that all information in the database will be lost and you will be starting from scratch. If you have written on any Volumes, you must write an end of file mark on the volume so that Bacula can reuse it. Do so with:
(stop Bacula or unmount the drive) mt -f /dev/nst0 rewind mt -f /dev/nst0 weof
Where you should replace /dev/nst0 with the appropriate tape drive device name for your machine.
Previously it was intended to be used primarily by Bacula developers for testing; although SQLite is also a good choice for this. We do not recommend its use in general.
This database is simplistic in that it consists entirely of Bacula's internal structures appended sequentially to a file. Consequently, it is in most cases inappropriate for sites with many clients or systems with large numbers of files, or long-term production environments.
Below, you will find a table comparing the features available with SQLite and MySQL and with the internal Bacula database. At the current time, you cannot dynamically switch from one to the other, but must rebuild the Bacula source code. If you wish to experiment with both, it is possible to build both versions of Bacula and install them into separate directories.
Feature | SQLite or MySQL | Bacula |
Job Record | Yes | Yes |
Media Record | Yes | Yes |
FileName Record | Yes | No |
File Record | Yes | No |
FileSet Record | Yes | Yes |
Pool Record | Yes | Yes |
Client Record | Yes | Yes |
JobMedia Record | Yes | Yes |
List Job Records | Yes | Yes |
List Media Records | Yes | Yes |
List Pool Records | Yes | Yes |
List JobMedia Records | Yes | Yes |
Delete Pool Record | Yes | Yes |
Delete Media Record | Yes | Yes |
Update Pool Record | Yes | Yes |
Implement Verify | Yes | No |
MD5 Signatures | Yes | No |
In addition, since there is no SQL available, the Console commands: sqlquery, query, retention, and any other command that directly uses SQL are not available with the Internal database.
There are a number of different licenses that are used in Bacula. If you have a printed copy of this manual, the details of each of the licenses referred to in this chapter can be found in the online version of the manual at http://www.bacula.org.
The GNU Free Documentation License (FDL) is used for this manual, which is a free and open license. This means that you may freely reproduce it and even make changes to it. However, rather than distribute your own version of this manual, we would much prefer if you would send any corrections or changes to the Bacula project.
The most recent version of the manual can always be found online at http://www.bacula.org.
The vast bulk of the source code is released under the GNU General Public License version 2..
Most of this code is copyrighted: Copyright ©2000-2007 Free Software Foundation Europe e.V.
Portions may be copyrighted by other people (ATT, the Free Software Foundation, ...). These files are released under the GPL license.
Some of the Bacula library source code is released under the GNU Lesser General Public License. This permits third parties to use these parts of our code in their proprietary programs to interface to Bacula.
Some of the Bacula code, or code that Bacula references, has been released to the public domain. E.g. md5.c, SQLite.
Bacula® is a registered trademark of Kern Sibbald.
We have trademarked the Bacula name to ensure that any program using the name Bacula will be exactly compatible with the program that we have released. The use of the name Bacula is restricted to software systems that agree exactly with the program presented here.
http://www.bacula.org/FLA-bacula.en.pdf
and should be filled out then sent to:
Free Software Foundation Europe
Freedom Task Force
Sumatrastrasse 25
8006 Zürich
Switzerland
Please note that the above address is different from the officially registered office mentioned in the document. When you send in such a complete document, please notify me: kern at sibbald dot com.
NO WARRANTY
BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING, REPAIR OR CORRECTION.
IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES.
The purpose of this License is to make a manual, textbook, or other functional and useful document "free" in the sense of freedom: to assure everyone the effective freedom to copy and redistribute it, with or without modifying it, either commercially or noncommercially. Secondarily, this License preserves for the author and publisher a way to get credit for their work, while not being considered responsible for modifications made by others.
This License is a kind of "copyleft", which means that derivative works of the document must themselves be free in the same sense. It complements the GNU General Public License, which is a copyleft license designed for free software.
We have designed this License in order to use it for manuals for free software, because free software needs free documentation: a free program should come with manuals providing the same freedoms that the software does. But this License is not limited to software manuals; it can be used for any textual work, regardless of subject matter or whether it is published as a printed book. We recommend this License principally for works whose purpose is instruction or reference.
This License applies to any manual or other work, in any medium, that contains a notice placed by the copyright holder saying it can be distributed under the terms of this License. Such a notice grants a world-wide, royalty-free license, unlimited in duration, to use that work under the conditions stated herein. The "Document", below, refers to any such manual or work. Any member of the public is a licensee, and is addressed as "you". You accept the license if you copy, modify or distribute the work in a way requiring permission under copyright law.
A "Modified Version" of the Document means any work containing the Document or a portion of it, either copied verbatim, or with modifications and/or translated into another language.
A "Secondary Section" is a named appendix or a front-matter section of the Document that deals exclusively with the relationship of the publishers or authors of the Document to the Document's overall subject (or to related matters) and contains nothing that could fall directly within that overall subject. (Thus, if the Document is in part a textbook of mathematics, a Secondary Section may not explain any mathematics.) The relationship could be a matter of historical connection with the subject or with related matters, or of legal, commercial, philosophical, ethical or political position regarding them.
The "Invariant Sections" are certain Secondary Sections whose titles are designated, as being those of Invariant Sections, in the notice that says that the Document is released under this License. If a section does not fit the above definition of Secondary then it is not allowed to be designated as Invariant. The Document may contain zero Invariant Sections. If the Document does not identify any Invariant Sections then there are none.
The "Cover Texts" are certain short passages of text that are listed, as Front-Cover Texts or Back-Cover Texts, in the notice that says that the Document is released under this License. A Front-Cover Text may be at most 5 words, and a Back-Cover Text may be at most 25 words.
A "Transparent" copy of the Document means a machine-readable copy, represented in a format whose specification is available to the general public, that is suitable for revising the document straightforwardly with generic text editors or (for images composed of pixels) generic paint programs or (for drawings) some widely available drawing editor, and that is suitable for input to text formatters or for automatic translation to a variety of formats suitable for input to text formatters. A copy made in an otherwise Transparent file format whose markup, or absence of markup, has been arranged to thwart or discourage subsequent modification by readers is not Transparent. An image format is not Transparent if used for any substantial amount of text. A copy that is not "Transparent" is called "Opaque".
Examples of suitable formats for Transparent copies include plain ASCII without markup, Texinfo input format, LaTeX input format, SGML or XML using a publicly available DTD, and standard-conforming simple HTML, PostScript or PDF designed for human modification. Examples of transparent image formats include PNG, XCF and JPG. Opaque formats include proprietary formats that can be read and edited only by proprietary word processors, SGML or XML for which the DTD and/or processing tools are not generally available, and the machine-generated HTML, PostScript or PDF produced by some word processors for output purposes only.
The "Title Page" means, for a printed book, the title page itself, plus such following pages as are needed to hold, legibly, the material this License requires to appear in the title page. For works in formats which do not have any title page as such, "Title Page" means the text near the most prominent appearance of the work's title, preceding the beginning of the body of the text.
A section "Entitled XYZ" means a named subunit of the Document whose title either is precisely XYZ or contains XYZ in parentheses following text that translates XYZ in another language. (Here XYZ stands for a specific section name mentioned below, such as "Acknowledgements", "Dedications", "Endorsements", or "History".) To "Preserve the Title" of such a section when you modify the Document means that it remains a section "Entitled XYZ" according to this definition.
The Document may include Warranty Disclaimers next to the notice which states that this License applies to the Document. These Warranty Disclaimers are considered to be included by reference in this License, but only as regards disclaiming warranties: any other implication that these Warranty Disclaimers may have is void and has no effect on the meaning of this License.
You may copy and distribute the Document in any medium, either commercially or noncommercially, provided that this License, the copyright notices, and the license notice saying this License applies to the Document are reproduced in all copies, and that you add no other conditions whatsoever to those of this License. You may not use technical measures to obstruct or control the reading or further copying of the copies you make or distribute. However, you may accept compensation in exchange for copies. If you distribute a large enough number of copies you must also follow the conditions in section 3.
You may also lend copies, under the same conditions stated above, and you may publicly display copies.
If you publish printed copies (or copies in media that commonly have printed covers) of the Document, numbering more than 100, and the Document's license notice requires Cover Texts, you must enclose the copies in covers that carry, clearly and legibly, all these Cover Texts: Front-Cover Texts on the front cover, and Back-Cover Texts on the back cover. Both covers must also clearly and legibly identify you as the publisher of these copies. The front cover must present the full title with all words of the title equally prominent and visible. You may add other material on the covers in addition. Copying with changes limited to the covers, as long as they preserve the title of the Document and satisfy these conditions, can be treated as verbatim copying in other respects.
If the required texts for either cover are too voluminous to fit legibly, you should put the first ones listed (as many as fit reasonably) on the actual cover, and continue the rest onto adjacent pages.
If you publish or distribute Opaque copies of the Document numbering more than 100, you must either include a machine-readable Transparent copy along with each Opaque copy, or state in or with each Opaque copy a computer-network location from which the general network-using public has access to download using public-standard network protocols a complete Transparent copy of the Document, free of added material. If you use the latter option, you must take reasonably prudent steps, when you begin distribution of Opaque copies in quantity, to ensure that this Transparent copy will remain thus accessible at the stated location until at least one year after the last time you distribute an Opaque copy (directly or through your agents or retailers) of that edition to the public.
It is requested, but not required, that you contact the authors of the Document well before redistributing any large number of copies, to give them a chance to provide you with an updated version of the Document.
You may copy and distribute a Modified Version of the Document under the conditions of sections 2 and 3 above, provided that you release the Modified Version under precisely this License, with the Modified Version filling the role of the Document, thus licensing distribution and modification of the Modified Version to whoever possesses a copy of it. In addition, you must do these things in the Modified Version:
If the Modified Version includes new front-matter sections or appendices that qualify as Secondary Sections and contain no material copied from the Document, you may at your option designate some or all of these sections as invariant. To do this, add their titles to the list of Invariant Sections in the Modified Version's license notice. These titles must be distinct from any other section titles.
You may add a section Entitled "Endorsements", provided it contains nothing but endorsements of your Modified Version by various parties--for example, statements of peer review or that the text has been approved by an organization as the authoritative definition of a standard.
You may add a passage of up to five words as a Front-Cover Text, and a passage of up to 25 words as a Back-Cover Text, to the end of the list of Cover Texts in the Modified Version. Only one passage of Front-Cover Text and one of Back-Cover Text may be added by (or through arrangements made by) any one entity. If the Document already includes a cover text for the same cover, previously added by you or by arrangement made by the same entity you are acting on behalf of, you may not add another; but you may replace the old one, on explicit permission from the previous publisher that added the old one.
The author(s) and publisher(s) of the Document do not by this License give permission to use their names for publicity for or to assert or imply endorsement of any Modified Version.
You may combine the Document with other documents released under this License, under the terms defined in section 4 above for modified versions, provided that you include in the combination all of the Invariant Sections of all of the original documents, unmodified, and list them all as Invariant Sections of your combined work in its license notice, and that you preserve all their Warranty Disclaimers.
The combined work need only contain one copy of this License, and multiple identical Invariant Sections may be replaced with a single copy. If there are multiple Invariant Sections with the same name but different contents, make the title of each such section unique by adding at the end of it, in parentheses, the name of the original author or publisher of that section if known, or else a unique number. Make the same adjustment to the section titles in the list of Invariant Sections in the license notice of the combined work.
In the combination, you must combine any sections Entitled "History" in the various original documents, forming one section Entitled "History"; likewise combine any sections Entitled "Acknowledgements", and any sections Entitled "Dedications". You must delete all sections Entitled "Endorsements".
You may make a collection consisting of the Document and other documents released under this License, and replace the individual copies of this License in the various documents with a single copy that is included in the collection, provided that you follow the rules of this License for verbatim copying of each of the documents in all other respects.
You may extract a single document from such a collection, and distribute it individually under this License, provided you insert a copy of this License into the extracted document, and follow this License in all other respects regarding verbatim copying of that document.
A compilation of the Document or its derivatives with other separate and independent documents or works, in or on a volume of a storage or distribution medium, is called an "aggregate" if the copyright resulting from the compilation is not used to limit the legal rights of the compilation's users beyond what the individual works permit. When the Document is included in an aggregate, this License does not apply to the other works in the aggregate which are not themselves derivative works of the Document.
If the Cover Text requirement of section 3 is applicable to these copies of the Document, then if the Document is less than one half of the entire aggregate, the Document's Cover Texts may be placed on covers that bracket the Document within the aggregate, or the electronic equivalent of covers if the Document is in electronic form. Otherwise they must appear on printed covers that bracket the whole aggregate.
Translation is considered a kind of modification, so you may distribute translations of the Document under the terms of section 4. Replacing Invariant Sections with translations requires special permission from their copyright holders, but you may include translations of some or all Invariant Sections in addition to the original versions of these Invariant Sections. You may include a translation of this License, and all the license notices in the Document, and any Warranty Disclaimers, provided that you also include the original English version of this License and the original versions of those notices and disclaimers. In case of a disagreement between the translation and the original version of this License or a notice or disclaimer, the original version will prevail.
If a section in the Document is Entitled "Acknowledgements", "Dedications", or "History", the requirement (section 4) to Preserve its Title (section 1) will typically require changing the actual title.
You may not copy, modify, sublicense, or distribute the Document except as expressly provided for under this License. Any other attempt to copy, modify, sublicense or distribute the Document is void, and will automatically terminate your rights under this License. However, parties who have received copies, or rights, from you under this License will not have their licenses terminated so long as such parties remain in full compliance.
The Free Software Foundation may publish new, revised versions of the GNU Free Documentation License from time to time. Such new versions will be similar in spirit to the present version, but may differ in detail to address new problems or concerns. See http://www.gnu.org/copyleft/.
Each version of the License is given a distinguishing version number. If the Document specifies that a particular numbered version of this License "or any later version" applies to it, you have the option of following the terms and conditions either of that specified version or of any later version that has been published (not as a draft) by the Free Software Foundation. If the Document does not specify a version number of this License, you may choose any version ever published (not as a draft) by the Free Software Foundation.
To use this License in a document you have written, include a copy of the License in the document and put the following copyright and license notices just after the title page:
Copyright ©YEAR YOUR NAME. Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation; with no Invariant Sections, no Front-Cover Texts, and no Back-Cover Texts. A copy of the license is included in the section entitled "GNU Free Documentation License".
If you have Invariant Sections, Front-Cover Texts and Back-Cover Texts, replace the "with...Texts." line with this:
with the Invariant Sections being LIST THEIR TITLES, with the Front-Cover Texts being LIST, and with the Back-Cover Texts being LIST.
If you have Invariant Sections without Cover Texts, or some other combination of the three, merge those two alternatives to suit the situation.
If your document contains nontrivial examples of program code, we recommend releasing these examples in parallel under your choice of free software license, such as the GNU General Public License, to permit their use in free software.
Disclaimer
This is an unofficial translation of the GNU Free Documentation License into French. It was not published by the Free Software Foundation, and does not legally define the distribution terms for documentation that uses the GNU FDL--only the original English text of the GNU FDL does that. However, we hope that this translation will help French speakers understand the GNU FDL better.
Avertissment
Ceci est une traduction française non officielle de la Licence de documentation libre GNU. Elle n'a pas été publiée par la Free Software Foundation, et ne fixe pas légalement les conditions de redistribution des documents qui l'utilisent -- seul le texte original en anglais le fait. Nous espérons toutefois que cette traduction aidera les francophones à mieux comprendre la FDL GNU.
L'objet de cette Licence est de rendre tout manuel,
livre ou autre document écrit "libre"
au sens de la liberté d'utilisation, à savoir :
assurer à chacun la liberté effective de le copier ou
de le redistribuer, avec ou sans modifications, commercialement ou
non. En outre, cette Licence garantit à l'auteur et à
l'éditeur la reconnaissance de leur travail, sans qu'ils
soient pour autant considérés comme responsables des
modifications réalisées par des tiers.
Cette Licence est une sorte de "copyleft",
ce qui signifie que les travaux dérivés du document
d'origine sont eux-mêmes "libres" selon
les mêmes termes. Elle complète la Licence Publique
Générale GNU, qui est également une Licence
copyleft, conçue pour les logiciels libres.
Nous avons conçu cette Licence pour la
documentation des logiciels libres, car les logiciels libres ont
besoin d'une documentation elle-même libre : un logiciel
libre doit être accompagné d'un manuel garantissant les
mêmes libertés que celles accordées par le
logiciel lui-même. Mais cette Licence n'est pas limitée
aux seuls manuels des logiciels ; elle peut être utilisée
pour tous les documents écrits, sans distinction particulière
relative au sujet traité ou au mode de publication. Nous
recommandons l'usage de cette Licence principalement pour les travaux
destinés à des fins d'enseignement ou devant servir de
documents de référence.
Cette Licence couvre tout manuel ou tout autre travail écrit contenant une notice de copyright autorisant la redistribution selon les termes de cette Licence. Le mot "Document" se réfère ci-après à un tel manuel ou travail. Toute personne en est par définition concessionnaire et est référencée ci-après par le terme "vous". Vous acceptez la licence si vous copiez, modifiez, ou redistribuez le travail en sorte que la permission sous la loi copyright est nécessaire.
Une "Version modifiée" du Document désigne tout travail en contenant la totalité ou seulement une portion de celui-ci, copiée mot pour mot, modifiée et/ou traduite dans une autre langue.
Une "Section secondaire" désigne une annexe au Document, ou toute information indiquant les rapports entre l'auteur ou l'éditeur et le sujet (ou tout autre sujet connexe) du document, sans toutefois être en rapport direct avec le sujet lui-même (par exemple, si le Document est un manuel de mathématiques, une Section secondaire ne traitera d'aucune notion mathématique). Cette section peut contenir des informations relatives à l'historique du Document, des sources documentaires, des dispositions légales, commerciales, philosophiques, ou des positions éthiques ou politiques susceptibles de concerner le sujet traité.
Les "Sections inaltérables" sont des sections secondaires considérées comme ne pouvant être modifiées et citées comme telles dans la notice légale qui place le Document sous cette Licence.
Les "Textes de couverture" sont les textes courts situés sur les pages de couverture avant et arrière du Document, et cités comme tels dans la mention légale de ce Document. Un Text de Couverture-Avant peut consister au maximum de 5 mots, et us Text de Couvertur-Arrière peut consister au maximum de 25 mots.
Le terme "Copie transparente" désigne une version numérique du Document représentée dans un format dont les spécifications sont publiquement disponibles et dont le contenu peut être visualisé et édité directement et immédiatement par un éditeur de texte quelconque, ou (pour les images composées de pixels) par un programme de traitement d'images quelconque, ou (pour les dessins) par un éditeur de dessins courant. Ce format doit pouvoir être accepté directement ou être convertible facilement dans des formats utilisables directement par des logiciels de formatage de texte. Une copie publiée dans un quelconque format numérique ouvert mais dont la structure a été conçue dans le but exprès de prévenir les modifications ultérieures du Document ou dans le but d'en décourager les lecteurs n'est pas considérée comme une Copie Transparente. Une copie qui n'est pas "Transparente" est considérée, par opposition, comme "Opaque".
Le format de fichier texte codé en ASCII
générique et n'utilisant pas de balises, les formats de
fichiers Texinfo ou LaTeX, les formats de fichiers SGML ou XML
utilisant une DTD publiquement accessible, ainsi que les formats de
fichiers HTML simple et standard, écrits de telle sorte qu'ils
sont modifiables sans outil spécifique, sont des exemples de
formats acceptables pour la réalisation de Copies
Transparentes. Les formats suivants sont opaques : PostScript, PDF,
formats de fichiers propriétaires qui ne peuvent être
visualisés ou édités que par des traitements de
textes propriétaires, SGML et XML utilisant des DTD et/ou des
outils de formatage qui ne sont pas disponibles publiquement, et du
code HTML généré par une machine à l'aide
d'un traitement de texte quelconque et dans le seul but de la
génération d'un format de sortie.
La "Page de titre" désigne, pour les ouvrages imprimés, la page de titre elle-même, ainsi que les pages supplémentaires nécessaires pour fournir clairement les informations dont cette Licence impose la présence sur la page de titre. Pour les travaux n'ayant pas de Page de titre comme décrit ci-dessus, la "Page de titre" désigne le texte qui s'apparente le plus au titre du document et situé avant le texte principal.
Une section "Intitulèe XYZ" désigne une partie du Document nommè dont le title est prècisemment XYZ ou contient XYZ entre parenthèses après texte traduit dans une autre langue. (Ici XYZ signifie le nom specifique d'une section mentionnèe dessous, par exemple "Remerciements", "Dédicaces", "Approbations", ou "Historique".)
"Concerver le Title" d'une telle section quand vous modifier le the Document signifie qu'il reste une section "Intitulèe XYZ" selon cette definition.
Le Document peut contenir Warranty Disclaimers près de la mentionne que cette License applique au Document. Ces Warranty Disclaimers sont considerès être incluent par rèference dans cette License, mais seulement concernant les Warranty Disclaimers: toute autre implication que ces Warrenty Disclaimers peuvent avoir est nul et n'a pas d'effet sur le interpretation de cette License.
Vous pouvez copier et distribuer le Document sur tout type de support, commercialement ou non, à condition que cette Licence, la notice de copyright et la notice de la Licence indiquant que cette Licence s'applique à ce Document soient reproduits dans toutes les copies, et que vous n'y ajoutiez aucune condition restrictive supplémentaire. Vous ne pouvez pas utiliser un quelconque moyen technique visant à empêcher ou à contrôler la lecture ou la reproduction ultérieure des copies que vous avez créées ou distribuées. Toutefois, vous pouvez solliciter une rétribution en échange des copies. Si vous distribuez une grande quantité de copies, référez-vous aux dispositions de la section 3.
Si vous publiez des copies imprimées de ce
Document à plus de 100 exemplaires et que la Licence du
Document indique la présence de Textes de
couverture, vous devez fournir une couverture pour chaque copie,
qui présente les Textes de couverture des première
et dernière pages de couverture du Document. Les
première et dernière pages de couverture doivent
également vous identifier clairement et sans ambiguité
comme étant l'éditeur de ces copies. La première
page de couverture doit comporter le titre du Document en mots
d'importance et de visibilité égales. Vous pouvez
ajouter des informations complémentaires sur les pages de
couverture. Les copies du Document dont seule la couverture a
été modifiée peuvent être considérées
comme des copies conformes, à condition que le titre du
Document soit préservé et que les conditions
indiquées précédemment soient respectées.
Si les textes devant se trouver sur la couverture sont
trop importants pour y tenir de manière claire, vous pouvez ne
placer que les premiers sur la première page et placer les
suivants sur les pages consécutives.
Si vous publiez plus de 100 Copies opaques du
Document, vous devez soit fournir une Copie transparente
pour chaque Copie opaque, soit préciser ou fournir avec
chaque Copie opaque une adresse réseau publiquement
accessible d'une Copie transparente et complète du
Document, sans aucun ajout ou modification, et à
laquelle tout le monde peut accéder en téléchargement
anonyme et sans frais, selon des protocoles réseau communs et
standards. Si vous choisissez cette dernière option, vous
devez prendre les dispositions nécessaires, dans la limite du
raisonnable, afin de garantir l'accès non restrictif à
la Copie transparente durant une année pleine après
la diffusion publique de la dernière Copie opaque
(directement ou <I>via</I> vos revendeurs).
Nous recommandons, mais ce n'est pas obligatoire, que
vous contactiez l'auteur du Document suffisamment tôt
avant toute publication d'un grand nombre de copies, afin de lui
permettre de vous donner une version à jour du Document.
Vous pouvez copier et distribuer une Version
modifiée du Document en respectant les conditions
des sections 2 et 3 précédentes, à condition de
placer cette Version modifiée sous la présente
Licence, dans laquelle le terme "Document"
doit être remplacé par les termes "Version
modifiée", donnant ainsi l'autorisation de
redistribuer et de modifier cette Version modifiée à
quiconque en possède une copie. De plus, vous devez effectuer
les actions suivantes dans la Version modifiée :
Si la Version modifiée contient de
nouvelles sections préliminaires ou de nouvelles annexes
considérées comme des Sections secondaires et
que celles-ci ne contiennent aucun élément copié
à partir du Document, vous pouvez à votre convenance en
désigner une ou plusieurs comme étant des Sections
inaltérables. Pour ce faire, ajoutez leurs titres dans la
liste des Sections inaltérables au sein de la notice de
Licence de la version Modifiée. Ces titres doivent êtres
distincts des titres des autres sections.
Vous pouvez ajouter une section nommée
"Approbations" à condition que
ces approbations ne concernent que les modifications ayant donné
naissance à la Version modifiée (par exemple,
comptes rendus de revue du document ou acceptation du texte par une
organisation le reconnaissant comme étant la définition
d'un standard).
Vous pouvez ajouter un passage comprenant jusqu'à
cinq mots en première page de couverture, et jusqu'à
vingt-cinq mots en dernière page de couverture, à la
liste des Textes de couverture de la Version modifiée.
Il n'est autorisé d'ajouter qu'un seul passage en première
et en dernière pages de couverture par personne ou groupe de
personnes ou organisation ayant contribué à la
modification du Document. Si le Document comporte déjà
un passage sur la même couverture, ajouté en votre nom
ou au nom de l'organisation au nom de laquelle vous agissez, vous ne
pouvez pas ajouter de passage supplémentaire ; mais vous
pouvez remplacer un ancien passage si vous avez expressément
obtenu l'autorisation de l'éditeur de celui-ci.
Cette Licence ne vous donne pas le droit d'utiliser le
nom des auteurs et des éditeurs de ce Document à
des fins publicitaires ou pour prétendre à
l'approbation d'une Version modifiée.
Vous pouvez fusionner le Document avec d'autres
documents soumis à cette Licence, suivant les spécifications
de la section 4 pour les Versions modifiées, à
condition d'inclure dans le document résultant toutes les
Sections inaltérables des documents originaux sans
modification, et de toutes les lister dans la liste des Sections
inaltérables de la notice de Licence du d<SPAN STYLE="font-weight: medium">ocument</SPAN>
résultant de la fusion.
Le document résultant de la fusion n'a besoin
que d'une seule copie de cette Licence, et les Sections
inaltérables existant en multiples exemplaires peuvent
être remplacées par une copie unique. S'il existe
plusieurs Sections inaltérables portant le même
nom mais de contenu différent, rendez unique le titre de
chaque section en ajoutant, à la fin de celui-ci, entre
parenthèses, le nom de l'auteur ou de l'éditeur
d'origine, ou, à défaut, un numéro unique. Les
mêmes modifications doivent être réalisées
dans la liste des Sections inaltérables de la notice de
Licence du document final.
Dans le document résultant de la fusion, vous
devez rassembler en une seule toutes les sections "Historique"
des documents d'origine. De même, vous devez rassembler les
sections "Remerciements" et "Dédicaces".
Vous devez supprimer toutes les sections "Approbations".
Vous pouvez créer un regroupement de documents
comprenant le Document et d'autres documents soumis à
cette Licence, et remplacer les copies individuelles de cette Licence
des différents documents par une unique copie incluse dans le
regroupement de documents, à condition de respecter pour
chacun de ces documents l'ensemble des règles de cette Licence
concernant les copies conformes.
Vous pouvez extraire un document d'un tel regroupement
et le distribuer individuellement sous couvert de cette Licence, à
condition d'y inclure une copie de cette Licence et d'en respecter
l'ensemble des règles concernant les copies conformes.
La compilation du Document ou de ses dérivés
avec d'autres documents ou travaux séparés et
indépendants sur un support de stockage ou sur un média
de distribution quelconque ne représente pas une Version
modifiée du Document tant qu'aucun copyright n'est
déposé pour cette compilation. Une telle compilation
est appelée "agrégat" et cette
Licence ne s'applique pas aux autres travaux indépendants
compilés avec le Document s'ils ne sont pas eux-mêmes
des travaux dérivés du Document.
Si les exigences de la section 3 concernant les Textes
de couverture sont applicables à ces copies du Document,
et si le Document représente un volume inférieur
à un quart du volume total de l'agrégat, les Textes
de couverture du Document peuvent être placés
sur des pages de couverture qui n'encadrent que le Document au
sein de l'agrégat. Dans le cas contraire, ils doivent
apparaître sur les pages de couverture de l'agrégat
complet.
La traduction est considérée comme une
forme de modification, vous pouvez donc distribuer les traductions du
Document selon les termes de la section 4. Vous devez obtenir
l'autorisation spéciale des auteurs des Sections
inaltérables pour les remplacer par des traductions, mais
vous pouvez inclure les traductions des Sections inaltérables
en plus des textes originaux. Vous pouvez inclure une traduction de
cette Licence à condition d'inclure également la
version originale en anglais. En cas de contradiction entre la
traduction et la version originale en anglais, c'est cette dernière
qui prévaut.
Vous ne pouvez pas copier, modifier, sous-licencier ou
distribuer le Document autrement que selon les termes de cette
Licence. Tout autre acte de copie, modification, sous-Licence ou
distribution du Document est sans objet et vous prive
automatiquement des droits que cette Licence vous accorde. En
revanche, les personnes qui ont reçu de votre part des copies
ou les droits sur le document sous couvert de cette Licence ne voient
pas leurs droits révoqués tant qu'elles en respectent
les principes.
La Free Software Foundation peut publier de temps en
temps de nouvelles versions révisées de cette Licence.
Ces nouvelles versions seront semblables à la présente
version dans l'esprit, mais pourront différer sur des points
particuliers en fonction de nouvelles questions ou nouveaux
problèmes. Voyez http:
www.gnu.org/copyleft
pour plus de détails.
Chaque version de cette Licence est dotée d'un
numéro de version distinct. Si un Document spécifie
un numéro de version particulier de cette Licence, et porte la
mention "ou toute autre version ultérieure",
vous pouvez choisir de suivre les termes de la version spécifiée
ou ceux de n'importe quelle version ultérieure publiée
par la Free Software Foundation. Si aucun numéro de version
n'est spécifié, vous pouvez choisir n'importe quelle
version officielle publiée par la Free Sofware Foundation.
Pour utiliser cette Licence avec un document que vous
avez écrit, incorporez une copie du texte de cette Licence en
anglais et placez le texte ci-dessous juste après la page de
titre :
Copyright ©ANNÉE VOTRE NOM Permission vous est donnée de copier, distribuer et/ou modifier ce document selon les termes de la Licence GNU Free Documentation License, Version 1.2 ou ultérieure publiée par la Free Software Foundation; sans Sections Inaltèrables, sans Texts de Couverture-Avant, et sans Texts de Couverture-Arrière. Une copie de cette Licence est incluse dans la section appelée GNU Free Documentation License.
Si votre Document contient de Sections Inaltérables, Texts de Couverture-Avant et Texts de Couverture-Arrière remplacez la ligne "sans Sections Inaltèrables ..." avec ceci:
avec les Sections Inaltèrables suivants LISTE DES TITRES DES SECTIONS INALTÉRABLE, avec les Texts de Couverture-Avant suivants LIST, et avec les Texts de Couverture-Arrière suivants LIST.
Si vous avec Sections Inaltèrables sans Texts de Couverture-Avant, ou une autre combinaison des trois, combinez les deux alternatives selon pour convenir.
Si votre Document contient
des exemples non triviaux de code programme, nous recommandons de
distribuer ces exemples en parallèle sous Licence GNU General
Public License, qui permet leur usage dans les logiciels libres.
Pour GNU FDL Version 1.2 (novembre 2002):
Version 1.2.0 FR (Kern Sibbald, juillet 2006)
Traduit de HTML à LaTeX
Version 2, June 1991
Copyright (C) 1989, 1991 Free Software Foundation, Inc. 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA Everyone is permitted to copy and distribute verbatim copies of this license document, but changing it is not allowed.
The licenses for most software are designed to take away your freedom to share
and change it. By contrast, the GNU General Public License is intended to
guarantee your freedom to share and change free software--
to make sure the
software is free for all its users. This General Public License applies to
most of the Free Software Foundation's software and to any other program whose
authors commit to using it. (Some other Free Software Foundation software is
covered by the GNU Library General Public License instead.) You can apply it
to your programs, too.
When we speak of free software, we are referring to freedom, not price. Our General Public Licenses are designed to make sure that you have the freedom to distribute copies of free software (and charge for this service if you wish), that you receive source code or can get it if you want it, that you can change the software or use pieces of it in new free programs; and that you know you can do these things.
To protect your rights, we need to make restrictions that forbid anyone to deny you these rights or to ask you to surrender the rights. These restrictions translate to certain responsibilities for you if you distribute copies of the software, or if you modify it.
For example, if you distribute copies of such a program, whether gratis or for a fee, you must give the recipients all the rights that you have. You must make sure that they, too, receive or can get the source code. And you must show them these terms so they know their rights.
We protect your rights with two steps: (1) copyright the software, and (2) offer you this license which gives you legal permission to copy, distribute and/or modify the software.
Also, for each author's protection and ours, we want to make certain that everyone understands that there is no warranty for this free software. If the software is modified by someone else and passed on, we want its recipients to know that what they have is not the original, so that any problems introduced by others will not reflect on the original authors' reputations.
Finally, any free program is threatened constantly by software patents. We wish to avoid the danger that redistributors of a free program will individually obtain patent licenses, in effect making the program proprietary. To prevent this, we have made it clear that any patent must be licensed for everyone's free use or not licensed at all.
The precise terms and conditions for copying, distribution and modification follow.
TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
0. This License applies to any program or other work which contains a notice placed by the copyright holder saying it may be distributed under the terms of this General Public License. The ``Program'', below, refers to any such program or work, and a ``work based on the Program'' means either the Program or any derivative work under copyright law: that is to say, a work containing the Program or a portion of it, either verbatim or with modifications and/or translated into another language. (Hereinafter, translation is included without limitation in the term ``modification''.) Each licensee is addressed as ``you''.
Activities other than copying, distribution and modification are not covered by this License; they are outside its scope. The act of running the Program is not restricted, and the output from the Program is covered only if its contents constitute a work based on the Program (independent of having been made by running the Program). Whether that is true depends on what the Program does.
1. You may copy and distribute verbatim copies of the Program's source code as you receive it, in any medium, provided that you conspicuously and appropriately publish on each copy an appropriate copyright notice and disclaimer of warranty; keep intact all the notices that refer to this License and to the absence of any warranty; and give any other recipients of the Program a copy of this License along with the Program.
You may charge a fee for the physical act of transferring a copy, and you may at your option offer warranty protection in exchange for a fee.
2. You may modify your copy or copies of the Program or any portion of it, thus forming a work based on the Program, and copy and distribute such modifications or work under the terms of Section 1 above, provided that you also meet all of these conditions:
These requirements apply to the modified work as a whole. If identifiable sections of that work are not derived from the Program, and can be reasonably considered independent and separate works in themselves, then this License, and its terms, do not apply to those sections when you distribute them as separate works. But when you distribute the same sections as part of a whole which is a work based on the Program, the distribution of the whole must be on the terms of this License, whose permissions for other licensees extend to the entire whole, and thus to each and every part regardless of who wrote it.
Thus, it is not the intent of this section to claim rights or contest your rights to work written entirely by you; rather, the intent is to exercise the right to control the distribution of derivative or collective works based on the Program.
In addition, mere aggregation of another work not based on the Program with the Program (or with a work based on the Program) on a volume of a storage or distribution medium does not bring the other work under the scope of this License.
3. You may copy and distribute the Program (or a work based on it, under Section 2) in object code or executable form under the terms of Sections 1 and 2 above provided that you also do one of the following:
The source code for a work means the preferred form of the work for making modifications to it. For an executable work, complete source code means all the source code for all modules it contains, plus any associated interface definition files, plus the scripts used to control compilation and installation of the executable. However, as a special exception, the source code distributed need not include anything that is normally distributed (in either source or binary form) with the major components (compiler, kernel, and so on) of the operating system on which the executable runs, unless that component itself accompanies the executable.
If distribution of executable or object code is made by offering access to copy from a designated place, then offering equivalent access to copy the source code from the same place counts as distribution of the source code, even though third parties are not compelled to copy the source along with the object code.
4. You may not copy, modify, sublicense, or distribute the Program except as expressly provided under this License. Any attempt otherwise to copy, modify, sublicense or distribute the Program is void, and will automatically terminate your rights under this License. However, parties who have received copies, or rights, from you under this License will not have their licenses terminated so long as such parties remain in full compliance.
5. You are not required to accept this License, since you have not signed it. However, nothing else grants you permission to modify or distribute the Program or its derivative works. These actions are prohibited by law if you do not accept this License. Therefore, by modifying or distributing the Program (or any work based on the Program), you indicate your acceptance of this License to do so, and all its terms and conditions for copying, distributing or modifying the Program or works based on it.
6. Each time you redistribute the Program (or any work based on the Program), the recipient automatically receives a license from the original licensor to copy, distribute or modify the Program subject to these terms and conditions. You may not impose any further restrictions on the recipients' exercise of the rights granted herein. You are not responsible for enforcing compliance by third parties to this License.
7. If, as a consequence of a court judgment or allegation of patent infringement or for any other reason (not limited to patent issues), conditions are imposed on you (whether by court order, agreement or otherwise) that contradict the conditions of this License, they do not excuse you from the conditions of this License. If you cannot distribute so as to satisfy simultaneously your obligations under this License and any other pertinent obligations, then as a consequence you may not distribute the Program at all. For example, if a patent license would not permit royalty-free redistribution of the Program by all those who receive copies directly or indirectly through you, then the only way you could satisfy both it and this License would be to refrain entirely from distribution of the Program.
If any portion of this section is held invalid or unenforceable under any particular circumstance, the balance of the section is intended to apply and the section as a whole is intended to apply in other circumstances.
It is not the purpose of this section to induce you to infringe any patents or other property right claims or to contest validity of any such claims; this section has the sole purpose of protecting the integrity of the free software distribution system, which is implemented by public license practices. Many people have made generous contributions to the wide range of software distributed through that system in reliance on consistent application of that system; it is up to the author/donor to decide if he or she is willing to distribute software through any other system and a licensee cannot impose that choice.
This section is intended to make thoroughly clear what is believed to be a consequence of the rest of this License.
8. If the distribution and/or use of the Program is restricted in certain countries either by patents or by copyrighted interfaces, the original copyright holder who places the Program under this License may add an explicit geographical distribution limitation excluding those countries, so that distribution is permitted only in or among countries not thus excluded. In such case, this License incorporates the limitation as if written in the body of this License.
9. The Free Software Foundation may publish revised and/or new versions of the General Public License from time to time. Such new versions will be similar in spirit to the present version, but may differ in detail to address new problems or concerns.
Each version is given a distinguishing version number. If the Program specifies a version number of this License which applies to it and ``any later version'', you have the option of following the terms and conditions either of that version or of any later version published by the Free Software Foundation. If the Program does not specify a version number of this License, you may choose any version ever published by the Free Software Foundation.
10. If you wish to incorporate parts of the Program into other free programs whose distribution conditions are different, write to the author to ask for permission. For software which is copyrighted by the Free Software Foundation, write to the Free Software Foundation; we sometimes make exceptions for this. Our decision will be guided by the two goals of preserving the free status of all derivatives of our free software and of promoting the sharing and reuse of software generally.
NO WARRANTY
11. BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM ``AS IS'' WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING, REPAIR OR CORRECTION.
12. IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES.
END OF TERMS AND CONDITIONS
If you develop a new program, and you want it to be of the greatest possible use to the public, the best way to achieve this is to make it free software which everyone can redistribute and change under these terms.
To do so, attach the following notices to the program. It is safest to attach them to the start of each source file to most effectively convey the exclusion of warranty; and each file should have at least the ``copyright'' line and a pointer to where the full notice is found.
{\it one line to give the program's name and an idea of what it does.} Copyright (C) {\it yyyy} {\it name of author} This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
Also add information on how to contact you by electronic and paper mail.
If the program is interactive, make it output a short notice like this when it starts in an interactive mode:
Gnomovision version 69, Copyright (C) {\it year} {\it name of author} Gnomovision comes with ABSOLUTELY NO WARRANTY; for details type `show w'. This is free software, and you are welcome to redistribute it under certain conditions; type `show c' for details.
The hypothetical commands `show w' and `show c' should show the
appropriate parts of the General Public License. Of course, the commands you
use may be called something other than `show w' and `show c'; they
could even be mouse-clicks or menu items--
whatever suits your program.
You should also get your employer (if you work as a programmer) or your school, if any, to sign a ``copyright disclaimer'' for the program, if necessary. Here is a sample; alter the names:
Yoyodyne, Inc., hereby disclaims all copyright interest in the program `Gnomovision' (which makes passes at compilers) written by James Hacker. {\it signature of Ty Coon}, 1 April 1989 Ty Coon, President of Vice
This General Public License does not permit incorporating your program into proprietary programs. If your program is a subroutine library, you may consider it more useful to permit linking proprietary applications with the library. If this is what you want to do, use the GNU Library General Public License instead of this License. Return to GNU's home page.
FSF & GNU inquiries & questions to gnu@gnu.org. Other ways to contact the FSF.
Comments on these web pages to webmasters@www.gnu.org, send other questions to gnu@gnu.org.
Copyright notice above. Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111, USA
Updated: 3 Jan 2000 rms
image of a Philosophical GNU [ English | Japanese ]
This GNU Lesser General Public License counts as the successor of the GNU Library General Public License. For an explanation of why this change was necessary, read the Why you shouldn't use the Lesser GPL for your next library article.
Version 2.1, February 1999
Copyright (C) 1991, 1999 Free Software Foundation, Inc. 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA Everyone is permitted to copy and distribute verbatim copies of this license document, but changing it is not allowed. [This is the first released version of the Lesser GPL. It also counts as the successor of the GNU Library Public License, version 2, hence the version number 2.1.]
The licenses for most software are designed to take away your freedom to share
and change it. By contrast, the GNU General Public Licenses are intended to
guarantee your freedom to share and change free software--
to make sure the
software is free for all its users.
This license, the Lesser General Public License, applies to some specially
designated software packages--
typically libraries--
of the Free Software
Foundation and other authors who decide to use it. You can use it too, but we
suggest you first think carefully about whether this license or the ordinary
General Public License is the better strategy to use in any particular case,
based on the explanations below.
When we speak of free software, we are referring to freedom of use, not price. Our General Public Licenses are designed to make sure that you have the freedom to distribute copies of free software (and charge for this service if you wish); that you receive source code or can get it if you want it; that you can change the software and use pieces of it in new free programs; and that you are informed that you can do these things.
To protect your rights, we need to make restrictions that forbid distributors to deny you these rights or to ask you to surrender these rights. These restrictions translate to certain responsibilities for you if you distribute copies of the library or if you modify it.
For example, if you distribute copies of the library, whether gratis or for a fee, you must give the recipients all the rights that we gave you. You must make sure that they, too, receive or can get the source code. If you link other code with the library, you must provide complete object files to the recipients, so that they can relink them with the library after making changes to the library and recompiling it. And you must show them these terms so they know their rights.
We protect your rights with a two-step method: (1) we copyright the library, and (2) we offer you this license, which gives you legal permission to copy, distribute and/or modify the library.
To protect each distributor, we want to make it very clear that there is no warranty for the free library. Also, if the library is modified by someone else and passed on, the recipients should know that what they have is not the original version, so that the original author's reputation will not be affected by problems that might be introduced by others.
Finally, software patents pose a constant threat to the existence of any free program. We wish to make sure that a company cannot effectively restrict the users of a free program by obtaining a restrictive license from a patent holder. Therefore, we insist that any patent license obtained for a version of the library must be consistent with the full freedom of use specified in this license.
Most GNU software, including some libraries, is covered by the ordinary GNU General Public License. This license, the GNU Lesser General Public License, applies to certain designated libraries, and is quite different from the ordinary General Public License. We use this license for certain libraries in order to permit linking those libraries into non-free programs.
When a program is linked with a library, whether statically or using a shared library, the combination of the two is legally speaking a combined work, a derivative of the original library. The ordinary General Public License therefore permits such linking only if the entire combination fits its criteria of freedom. The Lesser General Public License permits more lax criteria for linking other code with the library.
We call this license the ``Lesser'' General Public License because it does Less to protect the user's freedom than the ordinary General Public License. It also provides other free software developers Less of an advantage over competing non-free programs. These disadvantages are the reason we use the ordinary General Public License for many libraries. However, the Lesser license provides advantages in certain special circumstances.
For example, on rare occasions, there may be a special need to encourage the widest possible use of a certain library, so that it becomes a de-facto standard. To achieve this, non-free programs must be allowed to use the library. A more frequent case is that a free library does the same job as widely used non-free libraries. In this case, there is little to gain by limiting the free library to free software only, so we use the Lesser General Public License.
In other cases, permission to use a particular library in non-free programs enables a greater number of people to use a large body of free software. For example, permission to use the GNU C Library in non-free programs enables many more people to use the whole GNU operating system, as well as its variant, the GNU/Linux operating system.
Although the Lesser General Public License is Less protective of the users' freedom, it does ensure that the user of a program that is linked with the Library has the freedom and the wherewithal to run that program using a modified version of the Library.
The precise terms and conditions for copying, distribution and modification follow. Pay close attention to the difference between a ``work based on the library'' and a ``work that uses the library''. The former contains code derived from the library, whereas the latter must be combined with the library in order to run.
TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
0. This License Agreement applies to any software library or other program which contains a notice placed by the copyright holder or other authorized party saying it may be distributed under the terms of this Lesser General Public License (also called ``this License''). Each licensee is addressed as ``you''.
A ``library'' means a collection of software functions and/or data prepared so as to be conveniently linked with application programs (which use some of those functions and data) to form executables.
The ``Library'', below, refers to any such software library or work which has been distributed under these terms. A ``work based on the Library'' means either the Library or any derivative work under copyright law: that is to say, a work containing the Library or a portion of it, either verbatim or with modifications and/or translated straightforwardly into another language. (Hereinafter, translation is included without limitation in the term ``modification''.)
``Source code'' for a work means the preferred form of the work for making modifications to it. For a library, complete source code means all the source code for all modules it contains, plus any associated interface definition files, plus the scripts used to control compilation and installation of the library.
Activities other than copying, distribution and modification are not covered by this License; they are outside its scope. The act of running a program using the Library is not restricted, and output from such a program is covered only if its contents constitute a work based on the Library (independent of the use of the Library in a tool for writing it). Whether that is true depends on what the Library does and what the program that uses the Library does.
1. You may copy and distribute verbatim copies of the Library's complete source code as you receive it, in any medium, provided that you conspicuously and appropriately publish on each copy an appropriate copyright notice and disclaimer of warranty; keep intact all the notices that refer to this License and to the absence of any warranty; and distribute a copy of this License along with the Library.
You may charge a fee for the physical act of transferring a copy, and you may at your option offer warranty protection in exchange for a fee.
2. You may modify your copy or copies of the Library or any portion of it, thus forming a work based on the Library, and copy and distribute such modifications or work under the terms of Section 1 above, provided that you also meet all of these conditions:
(For example, a function in a library to compute square roots has a purpose that is entirely well-defined independent of the application. Therefore, Subsection 2d requires that any application-supplied function or table used by this function must be optional: if the application does not supply it, the square root function must still compute square roots.)
These requirements apply to the modified work as a whole. If identifiable sections of that work are not derived from the Library, and can be reasonably considered independent and separate works in themselves, then this License, and its terms, do not apply to those sections when you distribute them as separate works. But when you distribute the same sections as part of a whole which is a work based on the Library, the distribution of the whole must be on the terms of this License, whose permissions for other licensees extend to the entire whole, and thus to each and every part regardless of who wrote it.
Thus, it is not the intent of this section to claim rights or contest your rights to work written entirely by you; rather, the intent is to exercise the right to control the distribution of derivative or collective works based on the Library.
In addition, mere aggregation of another work not based on the Library with the Library (or with a work based on the Library) on a volume of a storage or distribution medium does not bring the other work under the scope of this License.
3. You may opt to apply the terms of the ordinary GNU General Public License instead of this License to a given copy of the Library. To do this, you must alter all the notices that refer to this License, so that they refer to the ordinary GNU General Public License, version 2, instead of to this License. (If a newer version than version 2 of the ordinary GNU General Public License has appeared, then you can specify that version instead if you wish.) Do not make any other change in these notices.
Once this change is made in a given copy, it is irreversible for that copy, so the ordinary GNU General Public License applies to all subsequent copies and derivative works made from that copy.
This option is useful when you wish to copy part of the code of the Library into a program that is not a library.
4. You may copy and distribute the Library (or a portion or derivative of it, under Section 2) in object code or executable form under the terms of Sections 1 and 2 above provided that you accompany it with the complete corresponding machine-readable source code, which must be distributed under the terms of Sections 1 and 2 above on a medium customarily used for software interchange.
If distribution of object code is made by offering access to copy from a designated place, then offering equivalent access to copy the source code from the same place satisfies the requirement to distribute the source code, even though third parties are not compelled to copy the source along with the object code.
5. A program that contains no derivative of any portion of the Library, but is designed to work with the Library by being compiled or linked with it, is called a ``work that uses the Library''. Such a work, in isolation, is not a derivative work of the Library, and therefore falls outside the scope of this License.
However, linking a ``work that uses the Library'' with the Library creates an executable that is a derivative of the Library (because it contains portions of the Library), rather than a ``work that uses the library''. The executable is therefore covered by this License. Section 6 states terms for distribution of such executables.
When a ``work that uses the Library'' uses material from a header file that is part of the Library, the object code for the work may be a derivative work of the Library even though the source code is not. Whether this is true is especially significant if the work can be linked without the Library, or if the work is itself a library. The threshold for this to be true is not precisely defined by law.
If such an object file uses only numerical parameters, data structure layouts and accessors, and small macros and small inline functions (ten lines or less in length), then the use of the object file is unrestricted, regardless of whether it is legally a derivative work. (Executables containing this object code plus portions of the Library will still fall under Section 6.)
Otherwise, if the work is a derivative of the Library, you may distribute the object code for the work under the terms of Section 6. Any executables containing that work also fall under Section 6, whether or not they are linked directly with the Library itself.
6. As an exception to the Sections above, you may also combine or link a ``work that uses the Library'' with the Library to produce a work containing portions of the Library, and distribute that work under terms of your choice, provided that the terms permit modification of the work for the customer's own use and reverse engineering for debugging such modifications.
You must give prominent notice with each copy of the work that the Library is used in it and that the Library and its use are covered by this License. You must supply a copy of this License. If the work during execution displays copyright notices, you must include the copyright notice for the Library among them, as well as a reference directing the user to the copy of this License. Also, you must do one of these things:
For an executable, the required form of the ``work that uses the Library'' must include any data and utility programs needed for reproducing the executable from it. However, as a special exception, the materials to be distributed need not include anything that is normally distributed (in either source or binary form) with the major components (compiler, kernel, and so on) of the operating system on which the executable runs, unless that component itself accompanies the executable.
It may happen that this requirement contradicts the license restrictions of other proprietary libraries that do not normally accompany the operating system. Such a contradiction means you cannot use both them and the Library together in an executable that you distribute.
7. You may place library facilities that are a work based on the Library side-by-side in a single library together with other library facilities not covered by this License, and distribute such a combined library, provided that the separate distribution of the work based on the Library and of the other library facilities is otherwise permitted, and provided that you do these two things:
8. You may not copy, modify, sublicense, link with, or distribute the Library except as expressly provided under this License. Any attempt otherwise to copy, modify, sublicense, link with, or distribute the Library is void, and will automatically terminate your rights under this License. However, parties who have received copies, or rights, from you under this License will not have their licenses terminated so long as such parties remain in full compliance.
9. You are not required to accept this License, since you have not signed it. However, nothing else grants you permission to modify or distribute the Library or its derivative works. These actions are prohibited by law if you do not accept this License. Therefore, by modifying or distributing the Library (or any work based on the Library), you indicate your acceptance of this License to do so, and all its terms and conditions for copying, distributing or modifying the Library or works based on it.
10. Each time you redistribute the Library (or any work based on the Library), the recipient automatically receives a license from the original licensor to copy, distribute, link with or modify the Library subject to these terms and conditions. You may not impose any further restrictions on the recipients' exercise of the rights granted herein. You are not responsible for enforcing compliance by third parties with this License.
11. If, as a consequence of a court judgment or allegation of patent infringement or for any other reason (not limited to patent issues), conditions are imposed on you (whether by court order, agreement or otherwise) that contradict the conditions of this License, they do not excuse you from the conditions of this License. If you cannot distribute so as to satisfy simultaneously your obligations under this License and any other pertinent obligations, then as a consequence you may not distribute the Library at all. For example, if a patent license would not permit royalty-free redistribution of the Library by all those who receive copies directly or indirectly through you, then the only way you could satisfy both it and this License would be to refrain entirely from distribution of the Library.
If any portion of this section is held invalid or unenforceable under any particular circumstance, the balance of the section is intended to apply, and the section as a whole is intended to apply in other circumstances.
It is not the purpose of this section to induce you to infringe any patents or other property right claims or to contest validity of any such claims; this section has the sole purpose of protecting the integrity of the free software distribution system which is implemented by public license practices. Many people have made generous contributions to the wide range of software distributed through that system in reliance on consistent application of that system; it is up to the author/donor to decide if he or she is willing to distribute software through any other system and a licensee cannot impose that choice.
This section is intended to make thoroughly clear what is believed to be a consequence of the rest of this License.
12. If the distribution and/or use of the Library is restricted in certain countries either by patents or by copyrighted interfaces, the original copyright holder who places the Library under this License may add an explicit geographical distribution limitation excluding those countries, so that distribution is permitted only in or among countries not thus excluded. In such case, this License incorporates the limitation as if written in the body of this License.
13. The Free Software Foundation may publish revised and/or new versions of the Lesser General Public License from time to time. Such new versions will be similar in spirit to the present version, but may differ in detail to address new problems or concerns.
Each version is given a distinguishing version number. If the Library specifies a version number of this License which applies to it and ``any later version'', you have the option of following the terms and conditions either of that version or of any later version published by the Free Software Foundation. If the Library does not specify a license version number, you may choose any version ever published by the Free Software Foundation.
14. If you wish to incorporate parts of the Library into other free programs whose distribution conditions are incompatible with these, write to the author to ask for permission. For software which is copyrighted by the Free Software Foundation, write to the Free Software Foundation; we sometimes make exceptions for this. Our decision will be guided by the two goals of preserving the free status of all derivatives of our free software and of promoting the sharing and reuse of software generally.
NO WARRANTY
15. BECAUSE THE LIBRARY IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY FOR THE LIBRARY, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES PROVIDE THE LIBRARY ``AS IS'' WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE LIBRARY IS WITH YOU. SHOULD THE LIBRARY PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING, REPAIR OR CORRECTION.
16. IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR REDISTRIBUTE THE LIBRARY AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE USE OR INABILITY TO USE THE LIBRARY (INCLUDING BUT NOT LIMITED TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD PARTIES OR A FAILURE OF THE LIBRARY TO OPERATE WITH ANY OTHER SOFTWARE), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES.
END OF TERMS AND CONDITIONS
If you develop a new library, and you want it to be of the greatest possible use to the public, we recommend making it free software that everyone can redistribute and change. You can do so by permitting redistribution under these terms (or, alternatively, under the terms of the ordinary General Public License).
To apply these terms, attach the following notices to the library. It is safest to attach them to the start of each source file to most effectively convey the exclusion of warranty; and each file should have at least the ``copyright'' line and a pointer to where the full notice is found.
{\it one line to give the library's name and an idea of what it does.} Copyright (C) {\it year} {\it name of author} This library is free software; you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation; either version 2.1 of the License, or (at your option) any later version. This library is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more details. You should have received a copy of the GNU Lesser General Public License along with this library; if not, write to the Free Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
Also add information on how to contact you by electronic and paper mail.
You should also get your employer (if you work as a programmer) or your school, if any, to sign a ``copyright disclaimer'' for the library, if necessary. Here is a sample; alter the names:
Yoyodyne, Inc., hereby disclaims all copyright interest in the library `Frob' (a library for tweaking knobs) written by James Random Hacker. {\it signature of Ty Coon}, 1 April 1990 Ty Coon, President of Vice
That's all there is to it! Return to GNU's home page.
FSF & GNU inquiries & questions to gnu@gnu.org. Other ways to contact the FSF.
Comments on these web pages to webmasters@www.gnu.org, send other questions to gnu@gnu.org.
Copyright notice above. Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111, USA
Updated: 27 Nov 2000 paulv
Once a new major version of Bacula is released, the Bacula users will vote on a list of new features. This vote is used as the main element determining what new features will be implemented for the next version. Generally, the development time for a new release is between 4 to 9 months.
For the current list of project, please see the projects page in the CVS at: http://cvs.sourceforge.net/viewcvs.py/*checkout*/bacula/bacula/projects see the projects file in the main source directory. The projects file is updated approximately once every six months.
Separately from the project list, Kern maintains a current list of tasks as well as ideas, feature requests, and occassionally design notes. This list is updated roughly weekly (sometimes more often). For a current list of tasks you can see kernstodo in the Source Forge CVS at http://cvs.sourceforge.net/viewcvs.py/*checkout*/bacula/bacula/kernstodo.
Thanks to Richard Stallman for starting the Free Software movement and for bringing us gcc and all the other GNU tools as well as the GPL license.
Thanks to Linus Torvalds for bringing us Linux.
Thanks to all the Free Software programmers. Without being able to peek at your code, and in some cases, take parts of it, this project would have been much more difficult.
Thanks to John Walker for suggesting this project, giving it a name, contributing software he has written, and for his programming efforts on Bacula as well as having acted as a constant sounding board and source of ideas.
Thanks to the apcupsd project where I started my Free Software efforts, and from which I was able to borrow some ideas and code that I had written.
Special thanks to D. Scott Barninger for writing the bacula RPM spec file, building all the RPM files and loading them onto Source Forge. This has been a tremendous help.
Many thanks to Karl Cunningham for converting the manual from html format to LaTeX. It was a major effort flawlessly done that will benefit the Bacula users for many years to come. Thanks Karl.
Thanks to Dan Langille for the incredible amount of testing he did on FreeBSD. His perseverance is truly remarkable. Thanks also for the many contributions he has made to improve Bacula (pthreads patch for FreeBSD, improved start/stop script and addition of Bacula userid and group, stunnel, ...), his continuing support of Bacula users. He also wrote the PostgreSQL driver for Bacula and has been a big help in correcting the SQL.
Thanks to multiple other Bacula Packagers who make and release packages for different platforms for Bacula.
Thanks to Christopher Hull for developing the native Win32 Bacula emulation code and for contributing it to the Bacula project.
Thanks to Robert Nelson for bringing our Win32 implementation up to par with all the same features that exist in the Unix/Linux versions. In addition, he has ported the Director and Storage daemon to Win32!
Thanks to Thorsten Engel for his excellent knowledge of Win32 systems, and for making the Win32 File daemon Unicode compatible, as well as making the Win32 File daemon interface to Microsoft's Volume Shadow Copy (VSS). These two are big pluses for Bacula!
Thanks to Landon Fuller for writing both the communications and the data encryption code for Bacula.
Thanks to Arno Lehmann for his excellent and infatigable help and advice to users.
Thanks to all the Bacula users, especially those of you who have contributed ideas, bug reports, patches, and new features.
Bacula can be enabled with data encryption and/or communications encryption. If this is the case, you will be including OpenSSL code that that contains cryptographic software written by Eric Young (eay@cryptsoft.com) and also software written by Tim Hudson (tjh@cryptsoft.com).
The Bat (Bacula Administration Tool) graphs are based in part on the work of the Qwt project (http://qwt.sf.net).
The original variable expansion code used in the LabelFormat comes from the Open Source Software Project (www.ossp.org). It has been adapted and extended for use in Bacula. This code is now deprecated.
There have been numerous people over the years who have contributed ideas, code, and help to the Bacula project. The file AUTHORS in the main source release file contains a list of contributors. For all those who I have left out, please send me a reminder, and in any case, thanks for your contribution.
Thanks to the Free Software Foundation Europe e.V. for assuming the responsibilities of protecting the Bacula copyright.
Certain words and/or products are Copyrighted or Trademarked such as Windows (by Microsoft). Since they are numerous, and we are not necessarily aware of the details of each, we don't try to list them here. However, we acknowledge all such Copyrights and Trademarks, and if any copyright or trademark holder wishes a specific acknowledgment, notify us, and we will be happy to add it where appropriate.
Well fortunately there are not too many bugs, but thanks to Dan Langille, we have a bugs database where bugs are reported. Generally, when a bug is fixed, a patch for the currently released version will be attached to the bug report.
The directory patches in the current SVN always contains a list of the patches that have been created for the previously released version of Bacula. In addition, the file patches-version-number in the patches directory contains a summary of each of the patches.
A "raw" list of the current task list and known issues can be found in kernstodo in the main Bacula source directory.
Please note that as of version 1.37, the Variable Expansion is deprecated and replaced by Python scripting (not yet documented).
Variable expansion is somewhat similar to Unix shell variable expansion. Currently (version 1.31), it is used only in format labels, but in the future, it will most likely be used in more places.
This is basically a string expansion capability that permits referencing variables, indexing arrays, conditional replacement of variables, case conversion, substring selection, regular expression matching and replacement, character class replacement, padding strings, repeated expansion in a user controlled loop, support of arithmetic expressions in the loop start, step and end conditions, and recursive expansion.
When using variable expansion characters in a Volume Label Format record, the format should always be enclosed in double quotes (").
For example, ${HOME} will be replaced by your home directory as defined in the environment. If you have defined the variable xxx to be Test, then the reference ${xxx:p/7/Y/r} will right pad the contents of xxx to a length of seven characters filling with the character Y giving YYYTest.
Within Bacula, there are three main classes of variables with some minor variations within the classes. The classes are:
Since the syntax is quite extensive, below, you will find the pseudo BNF. The special characters have the following meaning:
::= definition ( ) grouping if the parens are not quoted | separates alternatives '/' literal / (or any other character) CAPS a character or character sequence * preceding item can be repeated zero or more times ? preceding item can appear zero or one time + preceding item must appear one or more times
And the pseudo BNF describing the syntax is:
input ::= ( TEXT | variable | INDEX_OPEN input INDEX_CLOSE (loop_limits)? )* variable ::= DELIM_INIT (name|expression) name ::= (NAME_CHARS)+ expression ::= DELIM_OPEN (name|variable)+ (INDEX_OPEN num_exp INDEX_CLOSE)? (':' command)* DELIM_CLOSE command ::= '-' (TEXT_EXP|variable)+ | '+' (TEXT_EXP|variable)+ | 'o' NUMBER ('-'|',') (NUMBER)? | '#' | '*' (TEXT_EXP|variable)+ | 's' '/' (TEXT_PATTERN)+ '/' (variable|TEXT_SUBST)* '/' ('m'|'g'|'i'|'t')* | 'y' '/' (variable|TEXT_SUBST)+ '/' (variable|TEXT_SUBST)* '/' | 'p' '/' NUMBER '/' (variable|TEXT_SUBST)* '/' ('r'|'l'|'c') | '%' (name|variable)+ ('(' (TEXT_ARGS)? ')')? | 'l' | 'u' num_exp ::= operand | operand ('+'|'-'|'*'|'/'|'%') num_exp operand ::= ('+'|'-')? NUMBER | INDEX_MARK | '(' num_exp ')' | variable loop_limits ::= DELIM_OPEN (num_exp)? ',' (num_exp)? (',' (num_exp)?)? DELIM_CLOSE NUMBER ::= ('0'|...|'9')+ TEXT_PATTERN::= (^('/'))+ TEXT_SUBST ::= (^(DELIM_INIT|'/'))+ TEXT_ARGS ::= (^(DELIM_INIT|')'))+ TEXT_EXP ::= (^(DELIM_INIT|DELIM_CLOSE|':'|'+'))+ TEXT ::= (^(DELIM_INIT|INDEX_OPEN|INDEX_CLOSE))+ DELIM_INIT ::= '$' DELIM_OPEN ::= '{' DELIM_CLOSE ::= '}' INDEX_OPEN ::= '[' INDEX_CLOSE ::= ']' INDEX_MARK ::= '#' NAME_CHARS ::= 'a'|...|'z'|'A'|...|'Z'|'0'|...|'9'
The items listed in command above, which always follow a colon (:) have the following meanings:
- perform substitution if variable is empty + perform substitution if variable is not empty o cut out substring of the variable value # length of the variable value * substitute empty string if the variable value is not empty, otherwise substitute the trailing parameter s regular expression search and replace. The trailing options are: m = multiline, i = case insensitive, g = global, t = plain text (no regexp) y transpose characters from class A to class B p pad variable to l = left, r = right or c = center, with second value. % special function call (none implemented) l lower case the variable value u upper case the variable value
The loop_limits are start, step, and end values.
A counter variable name followed immediately by a plus (+) will cause the counter to be incremented by one.
To create an ISO date:
DLT-${Year}-${Month:p/2/0/r}-${Day:p/2/0/r}
on 20 June 2003 would give DLT-2003-06-20
If you set the environment variable mon to
January|February|March|April|May|... File-${mon[${Month}]}/${Day}/${Year}
on the first of March would give File-March/1/2003
Prior to version 1.37, Bacula did not have built-in communications encryption. Please see the TLS chapter if you are using Bacula 1.37 or greater.
Without too much effort, it is possible to encrypt the communications between any of the daemons. This chapter will show you how to use stunnel to encrypt communications to your client programs. We assume the Director and the Storage daemon are running on one machine that will be called server and the Client or File daemon is running on a different machine called client. Although the details may be slightly different, the same principles apply whether you are encrypting between Unix, Linux, or Win32 machines. This example was developed between two Linux machines running stunnel version 4.04-4 on a Red Hat Enterprise 3.0 system.
First, you must know that with the standard Bacula configuration, the Director will contact the File daemon on port 9102. The File daemon then contacts the Storage daemon using the address and port parameters supplied by the Director. The standard port used will be 9103. This is the typical server/client view of the world, the File daemon is a server to the Director (i.e. listens for the Director to contact it), and the Storage daemon is a server to the File daemon.
The encryption is accomplished between the Director and the File daemon by using an stunnel on the Director's machine (server) to encrypt the data and to contact an stunnel on the File daemon's machine (client), which decrypts the data and passes it to the client.
Between the File daemon and the Storage daemon, we use an stunnel on the File daemon's machine to encrypt the data and another stunnel on the Storage daemon's machine to decrypt the data.
As a consequence, there are actually four copies of stunnel running, two on the server and two on the client. This may sound a bit complicated, but it really isn't. To accomplish this, we will need to construct four separate conf files for stunnel, and we will need to make some minor modifications to the Director's conf file. None of the other conf files need to be changed.
Since pictures usually help a lot, here is an overview of what we will be doing. Don't worry about all the details of the port numbers and such for the moment.
File daemon (client): stunnel-fd1.conf |===========| Port 29102 >----| Stunnel 1 |-----> Port 9102 |===========| stunnel-fd2.conf |===========| Port 9103 >----| Stunnel 2 |-----> server:29103 |===========| Director (server): stunnel-dir.conf |===========| Port 29102 >----| Stunnel 3 |-----> client:29102 |===========| stunnel-sd.conf |===========| Port 29103 >----| Stunnel 4 |-----> 9103 |===========|
In order for stunnel to function as a server, which it does in our diagram for Stunnel 1 and Stunnel 4, you must have a certificate and the key. It is possible to keep the two in separate files, but normally, you keep them in one single .pem file. You may create this certificate yourself in which case, it will be self-signed, or you may have it signed by a CA.
If you want your clients to verify that the server is in fact valid (Stunnel 2 and Stunnel 3), you will need to have the server certificates signed by a CA (Certificate Authority), and you will need to have the CA's public certificate (contains the CA's public key).
Having a CA signed certificate is highly recommended if you are using your client across the Internet, otherwise you are exposed to the man in the middle attack and hence loss of your data.
See below for how to create a self-signed certificate.
To simplify things a bit, let's for the moment consider only the data channel. That is the connection between the File daemon and the Storage daemon, which takes place on port 9103. In fact, in a minimalist solution, this is the only connection that needs to be encrypted, because it is the one that transports your data. The connection between the Director and the File daemon is simply a control channel used to start the job and get the job status.
Normally the File daemon will contact the Storage daemon on port 9103 (supplied by the Director), so we need an stunnel that listens on port 9103 on the File daemon's machine, encrypts the data and sends it to the Storage daemon. This is depicted by Stunnel 2 above. Note that this stunnel is listening on port 9103 and sending to server:29103. We use port 29103 on the server because if we would send the data to port 9103, it would go directly to the Storage daemon, which doesn't understand encrypted data. On the server machine, we run Stunnel 4, which listens on port 29103, decrypts the data and sends it to the Storage daemon, which is listening on port 9103.
The Storage resource of the bacula-dir.conf normally looks something like the following:
Storage { Name = File Address = server SDPort = 9103 Password = storage_password Device = File Media Type = File }
Notice that this is running on the server machine, and it points the File daemon back to server:9103, which is where our Storage daemon is listening. We modify this to be:
Storage { Name = File Address = localhost SDPort = 9103 Password = storage_password Device = File Media Type = File }
This causes the File daemon to send the data to the stunnel running on localhost (the client machine). We could have used client as the address as well.
In the diagram above, we see above Stunnel 2 that we use stunnel-fd2.conf on the client. A pretty much minimal config file would look like the following:
client = yes [29103] accept = localhost:9103 connect = server:29103
The above config file does encrypt the data but it does not require a certificate, so it is subject to the man in the middle attack. The file I actually used, stunnel-fd2.conf, looked like this:
# # Stunnel conf for Bacula client -> SD # pid = /home/kern/bacula/bin/working/stunnel.pid # # A cert is not mandatory here. If verify=2, a # cert signed by a CA must be specified, and # either CAfile or CApath must point to the CA's # cert # cert = /home/kern/stunnel/stunnel.pem CAfile = /home/kern/ssl/cacert.pem verify = 2 client = yes # debug = 7 # foreground = yes [29103] accept = localhost:9103 connect = server:29103
You will notice that I specified a pid file location because I ran stunnel under my own userid so I could not use the default, which requires root permission. I also specified a certificate that I have as well as verify level 2 so that the certificate is required and verified, and I must supply the location of the CA (Certificate Authority) certificate so that the stunnel certificate can be verified. Finally, you will see that there are two lines commented out, which when enabled, produce a lot of nice debug info in the command window.
If you do not have a signed certificate (stunnel.pem), you need to delete the cert, CAfile, and verify lines.
Note that the stunnel.pem, is actually a private key and a certificate in a single file. These two can be kept and specified individually, but keeping them in one file is more convenient.
The config file, stunnel-sd.conf, needed for Stunnel 4 on the server machine is:
# # Bacula stunnel conf for Storage daemon # pid = /home/kern/bacula/bin/working/stunnel.pid # # A cert is mandatory here, it may be self signed # If it is self signed, the client may not use # verify # cert = /home/kern/stunnel/stunnel.pem client = no # debug = 7 # foreground = yes [29103] accept = 29103 connect = 9103
It will most likely be the simplest to implement the Data Channel encryption in the following order:
stunnel stunnel-sd.conf
stunnel stunnel-fd2.conf
The Job control channel is between the Director and the File daemon, and as mentioned above, it is not really necessary to encrypt, but it is good practice to encrypt it as well. The two stunnels that are used in this case will be Stunnel 1 and Stunnel 3 in the diagram above. Stunnel 3 on the server might normally listen on port 9102, but if you have a local File daemon, this will not work, so we make it listen on port 29102. It then sends the data to client:29102. Again we use port 29102 so that the stunnel on the client machine can decrypt the data before passing it on to port 9102 where the File daemon is listening.
We need to modify the standard Client resource, which would normally look something like:
Client { Name = client-fd Address = client FDPort = 9102 Catalog = BackupDB Password = "xxx" }
to be:
Client { Name = client-fd Address = localhost FDPort = 29102 Catalog = BackupDB Password = "xxx" }
This will cause the Director to send the control information to localhost:29102 instead of directly to the client.
The stunnel config file, stunnel-dir.conf, for the Director's machine would look like the following:
# # Bacula stunnel conf for the Directory to contact a client # pid = /home/kern/bacula/bin/working/stunnel.pid # # A cert is not mandatory here. If verify=2, a # cert signed by a CA must be specified, and # either CAfile or CApath must point to the CA's # cert # cert = /home/kern/stunnel/stunnel.pem CAfile = /home/kern/ssl/cacert.pem verify = 2 client = yes # debug = 7 # foreground = yes [29102] accept = localhost:29102 connect = client:29102
and the config file, stunnel-fd1.conf, needed to run stunnel on the Client would be:
# # Bacula stunnel conf for the Directory to contact a client # pid = /home/kern/bacula/bin/working/stunnel.pid # # A cert is not mandatory here. If verify=2, a # cert signed by a CA must be specified, and # either CAfile or CApath must point to the CA's # cert # cert = /home/kern/stunnel/stunnel.pem CAfile = /home/kern/ssl/cacert.pem verify = 2 client = yes # debug = 7 # foreground = yes [29102] accept = localhost:29102 connect = client:29102
It will most likely be the simplest to implement the Control Channel encryption in the following order:
stunnel stunnel-dir.conf
stunnel stunnel-fd1.conf
On the client machine, you can just duplicate the setup that you have on the first client file for file and it should work fine.
In the bacula-dir.conf file, you will want to create a second client pretty much identical to how you did for the first one, but the port number must be unique. We previously used:
Client { Name = client-fd Address = localhost FDPort = 29102 Catalog = BackupDB Password = "xxx" }
so for the second client, we will, of course, have a different name, and we will also need a different port. Remember that we used port 29103 for the Storage daemon, so for the second client, we can use port 29104, and the Client resource would look like:
Client { Name = client2-fd Address = localhost FDPort = 29104 Catalog = BackupDB Password = "yyy" }
Now, fortunately, we do not need a third stunnel to on the Director's machine, we can just add the new port to the config file, stunnel-dir.conf, to make:
# # Bacula stunnel conf for the Directory to contact a client # pid = /home/kern/bacula/bin/working/stunnel.pid # # A cert is not mandatory here. If verify=2, a # cert signed by a CA must be specified, and # either CAfile or CApath must point to the CA's # cert # cert = /home/kern/stunnel/stunnel.pem CAfile = /home/kern/ssl/cacert.pem verify = 2 client = yes # debug = 7 # foreground = yes [29102] accept = localhost:29102 connect = client:29102 [29104] accept = localhost:29102 connect = client2:29102
There are no changes necessary to the Storage daemon or the other stunnel so that this new client can talk to our Storage daemon.
You may create a self-signed certificate for use with stunnel that will permit you to make it function, but will not allow certificate validation. The .pem file containing both the certificate and the key can be made with the following, which I put in a file named makepem:
#!/bin/sh # # Simple shell script to make a .pem file that can be used # with stunnel and Bacula # OPENSSL=openssl umask 77 PEM1="/bin/mktemp openssl.XXXXXX" PEM2="/bin/mktemp openssl.XXXXXX" ${OPENSSL} req -newkey rsa:1024 -keyout $PEM1 -nodes \ -x509 -days 365 -out $PEM2 cat $PEM1 > stunnel.pem echo "" >>stunnel.pem cat $PEM2 >>stunnel.pem rm $PEM1 $PEM2
The above script will ask you a number of questions. You may simply answer each of them by entering a return, or if you wish you may enter your own data.
The process of getting a certificate that is signed by a CA is quite a bit more complicated. You can purchase one from quite a number of PKI vendors, but that is not at all necessary for use with Bacula. To get a CA signed certificate, you will either need to find a friend that has setup his own CA or to become a CA yourself, and thus you can sign all your own certificates. The book OpenSSL by John Viega, Matt Mesier & Pravir Chandra from O'Reilly explains how to do it, or you can read the documentation provided in the Open-source PKI Book project at Source Forge: http://ospkibook.sourceforge.net/docs/OSPKI-2.4.7/OSPKI-html/ospki-book.htm. Note, this link may change.
Please see the script ssh-tunnel.sh in the examples directory. It was contributed by Stephan Holl.
This document was generated using the LaTeX2HTML translator Version 2002-2-1 (1.71)
Copyright © 1993, 1994, 1995, 1996,
Nikos Drakos,
Computer Based Learning Unit, University of Leeds.
Copyright © 1997, 1998, 1999,
Ross Moore,
Mathematics Department, Macquarie University, Sydney.
The command line arguments were:
latex2html -white -no_subdir -split 0 -toc_stars -white -notransparent -init_file latex2html-init.pl bacula
The translation was initiated by on 2008-10-13