|\_ B A C K U P N I N J A /()/ `\| a silent flower blossom death strike to lost data. Backupninja allows you to coordinate system backup by dropping a few simple configuration files into /etc/backup.d/. Most programs you might use for making backups don't have their own configuration file format. Backupninja provides a centralized way to configure and coordinate many different backup utilities. Features: - easy to read ini style configuration files. - you can drop in scripts to handle new types of backups. - backup actions can be scheduled - you can choose when status report emails are mailed to you (always, on warning, on error, never). - console-based wizard (ninjahelper) makes it easy to create backup action configuration files. - passwords are never sent via the command line to helper programs. - works with Linux-Vservers (http://linux-vserver.org/) Backup types: - secure, remote, incremental filesytem backup (via rdiff-backup). incremental data is compressed. permissions are retained even with an unpriviledged backup user. - backup of mysql databases (via mysqlhotcopy and mysqldump). - backup of ldap databases (via slapcat and ldapsearch). - basic system and hardware info - encrypted remote backups (via duplicity). - backup of subversion repositories. The following options are available: -h, --help This usage message -d, --debug Run in debug mode, where all log messages are output to the current shell. -f, --conffile FILE Use FILE for the main configuration instead of /etc/backupninja.conf -t, --test Test run mode. This will test if the backup could run, without actually preforming any backups. For example, it will attempt to authenticate or test that ssh keys are set correctly. -n, --now Perform actions now, instead of when they might be scheduled. No output will be created unless also run with -d. --run FILE Runs the specified action FILE (e.g. one of the /etc/backup.d/ files). Also puts backupninja in debug mode. CONFIGURATION FILES =================== The general configuration file is /etc/backupninja.conf. In this file you can set the log level and change the default directory locations. You can force a different general configuration file with "backupninja -f /path/to/conf". To preform the actual backup, backupninja processes each configuration file in /etc/backup.d according to the file's suffix: .sh -- run this file as a shell script. .rdiff -- this is a configuration for rdiff-backup .maildir -- this is a configuration to backup maildirs .dup -- this is a configuration for duplicity .mysql -- mysql backup configuration .ldap -- ldap backup configuration .sys -- general system reports .svn -- subversion repository backups Support for additional configuration types can be added by dropping bash scripts with the name of the suffix into /usr/share/backupninja. The configuration files are processed in alphabetical order. However, it is suggested that you name the config files in "sysvinit style." For example: 00-disabled.ldap 10-runthisfirst.sh 20-runthisnext.mysql 90-runthislast.rdiff Typically, you will put a '.rdiff' config file last, so that any database dumps you make are included in the filesystem backup. Configurations files which begin with 0 (zero) are skipped. Unless otherwise specified, the config file format is "ini style." For example: # this is a comment [fishes] fish = red fish = blue [fruit] apple = yes pear = no thanks \ i will not have a pear. SCHEDULING ========== By default, each configuration file is processed everyday at 01:00 (1 AM). This can be changed by specifying the 'when' option in a config file. For example: when = sundays at 02:00 when = 30th at 22 when = 30 at 22:00 when = everyday at 01 <-- the default when = Tuesday at 05:00 A configuration file will be processed at the time(s) specified by the "when" option. If multiple "when" options are present, then they all apply. If two configurations files are scheduled to run in the same hour, then we fall back on the alphabetical ordering specified above. If two configurations files are scheduled close to one another in time, it is possible to have multiple copies of backupninja running if the first instance is not finished before the next one starts. These values for 'when' are equivalent: when = tuesday at 05:30 when = TUESDAYS at 05 These values for 'when' are invalid: when = tuesday at 2am when = tuesday at 2 when = tues at 02 REAL WORLD USAGE ================ Backupninja can be used to impliment whatever backup strategy you choose. It is intended, however, to be used like so: (1) First, databases are safely copied or exported to /var/backups. Typically, you cannot make a file backup of a database while it is in use, hence the need to use special tools to make a safe copy or export into /var/backups. (2) Then, vital parts of the file system, including /var/backups, are nightly pushed to a remote, off-site, hard disk (using rdiff-backup). The local user is root, but the remote user is not priviledged. Hopefully, the remote filesystem is encrypted. There are many different backup strategies out there, including "pull style", magnetic tape, rsync + hard links, etc. We believe that the strategy outlined above is the way to go because: (1) hard disks are very cheap these days, (2) pull style backups are no good, because then the backup server must have root on the production server, and (3) rdiff-backup is more space efficient and featureful than using rsync + hard links. SSH KEYS ======== In order for rdiff-backup to sync files over ssh unattended, you must create ssh keys on the source server and copy the public key to the remote user's authorized keys file. For example: root@srchost# ssh-keygen -t dsa root@srchost# ssh-copy-id -i /root/.ssh/id_dsa.pub backup@desthost Now, you should be able to ssh from user 'root' on srchost to user 'backup' on desthost without specifying a password. Note: when prompted for a password by ssh-keygen, just leave it blank by hitting return. INSTALLATION ============ Requirements: apt-get install bash gawk Recommended: apt-get install rdiff-backup gzip hwinfo Files: /usr/sbin/backupninja -- main script /etc/cron.d/backupninja -- runs main script nightly /etc/logrotate.d/backupninja -- rotates backupninja.log /etc/backup.d/ -- directory for configuration files /etc/backupninja.conf -- general options /usr/share/backupninja -- handler scripts which do the actual work Installation: There is no install script, but you just need to move files to the correct locations. All files should be owned by root. # tar xvzf backupninja.tar.gz # cd backupninja # mv backupninja /usr/sbin/backupninja # mv etc/logrotate.d/backupninja /etc/logrotate.d/backupninja # mv etc/cron.d/backupninja /etc/cron.d/backupninja # mkdir /etc/backup.d/ # mv etc/backupninja.conf /etc/backupninja.conf # mv handlers /usr/share/backupninja VSERVERS ======== If you are using Linux-Vservers (http://linux-vserver.org/) there are some special capabilities that different handlers have to make vserver backups easier. Set the variable "vservers" to be "yes" in /etc/backupninja.conf and see the example configuration files for each handler to configure the vserver specific variables. Additional vserver variables that can be configured in /etc/backupninja.conf. but probably don't need to be changed: VSERVERINFO (default: /usr/sbin/vserver-info) VSERVER (default: /usr/sbin/vserver) VROOTDIR (default: `$VSERVERINFO info SYSINFO |grep vserver-Rootdir | awk '{print $2}'; fi`) NINJAHELPER =========== Ninjahelper is an additional script which will walk you through the process of configuring backupninja. Ninjahelper has a menu driven curses based interface (using dialog). To add an additional 'wizard' to ninjahelper, follow these steps: (1) to add a helper for the handler "blue", create the file blue.helper in the directory where the handlers live. (ie /usr/share/backupninja). (2) next, you need to add your helper to the global HELPERS variable and define the main function for your helper (the function name is always _wizard). for example, blue.helper: HELPERS="$HELPERS blue:description_of_this_helper blue_wizard() { ... do work here ... } (3) check the examples of the included helpers to see how they are written. The dialog functions are defined in easydialog.sh.