BalticGrid sites need a policy for running regular backup of the critical data. Those data include:

• site configuration data

• system logs for tracking any security issues

• accounting logs for tracking the site usage

Here we propose a solution that is aimed to help the site administrators to run and manage the data for backup.

Idea

There are two scripts:

• one for storing files in backup called grid-backup
  

• one for restoring files called grid-restore
  

in a config file you provide the list of files and directories that need to be backuped. In a command line parameter you provide the destination where to store the data and how long you wish to preserve that data.

you provide the name of the file or directory you want to restore and destination where to put restored data.

The scripts are just wrappers to rdff-backup they are based on.

Installation:

1. Download the scripts:
   http://www.balticgrid.org/Grid_Operations/technicalguides/Files/grid-backup
   http://www.balticgrid.org/Grid_Operations/technicalguides/Files/grid-restore
2. Make sure, that the files are executable:
   #chmod u+x grid-backup grid-restore
   
3. Install rdiff-backup tool
   either manualy - see: http://rdiff-backup.nongnu.org/ (python installation is also required)
   or:
   #apt-get install rdiff-backup
   if above command don't works you probably need to add apt repository:
   #echo "rpm http://linuxsoft.cern.ch dag/redhat/el3/en/i386 dag" >> /etc/apt/sources.list.d/dag.list
   #apt-get update
   

Configuration:

#grid-backup --create-config-file

a config file and choose sections according to your needs.

eg. vi /etc/grid-backup

Usage:

There are two functionalities: backup and restore.

Backup script can also delete files older than given limit after backup.

#grid-backup DESTINATION [--delete-older-files|-d time]

eg.:

#grid-backup /backup -d 3M5D1h (remove older than 3 Months 5 Days 1 hour)

#grid-restore hostname.net::/remote

see: man rdiff-backup, form more time format information

grid-restore BACKUP_DIR/FILE[|DIR]_TO_RECOVER RECOVERED_FILE[DIR]_NAME [time(=now default)]

eg.:

#grid-restore /backup/var/log/messages ./messages 2M3D (restore file as it was 2 months and 3 days ago)

#grid-restore hostname.net::/remote/directory /etc/directory 2005-11-09T12:43:53

(

If you are absolutely sure, that you want to remove a target first, then restore a file, you may add a --force parameter at very end. (beware: hidden "rm -rf")

Crontab & do NOT forget

(about reading logfile)

#crontab -e

add a line like this for running backup daily:

4 1 * * *      /root/grid-backup/grid-backup backup@secure-host.somewhere.net::/site_backup_dir

Note, that if you want to make remote backups in crontabs, you need to configure no-passwd ssh authentification. How to do it? See eg. (TODO link).

TODOs:

in text