Difference between revisions of "Repository backup 5.x"

Revision as of 19:09, 25 January 2010

OpenKM backup depends on repository configuration. By default, all documents are stored in $JBOSS_HOME/repository. This default folder can be changed using the property repository.home in the file OpenKM.cfg. If you want to make a backup, you should keep a copy of this folder. Is also recommended to backup files located at $JBOSS_HOME/server/default/data/hypersonic because there are several databases which handle important data:

• OKMActivity: Stores the user activity log. All the actions performed by an user in OpenKM are logged. Only the administrator user can access this log and search for specific actions and users. This database can be very big if there are many users and make an intensive use of the application.

• OKMAuth: This database is used to manage authentication. All the users, roles and passwords are here, so take care of it. If this database is damaged, you won't be able to log into OpenKM.

• OKMDashboard: Is used to keep the visited items of the dashboard.

• OKMWorkflow: Is used to store the process definitions and process instance data. Also is used to manage task and node transitions.

It is recommended to stop JBoss before start the backup to prevent file modifications in the middle of the process because can make the backup useless. The backup process should have these steps:

• Stop JBoss
• Backup important folders
• Start JBoss

In Unix / Linux you can backup the repository folder running these commands from $JBOSS_HOME. First stop JBoss:

$ ./bin/shutdown.sh -S

Once JBoss is stopped (you can see the JBoss log to ensure it has stopped) , make the backup:

$ tar czf repository.tgz repository

To check the backup, rename the repository folder:

$ mv repository repository-old

Restore the backup:

$ tar xzf repository.tgz

And start JBoss:

$ ./bin/run.sh

This backup script uses rsync to minimize network load and creates incremental backups, preserving last four backups.

#!/bin/bash
#
## BEGIN CONFIG ##
HOST=$(uname -n)
FILES="/home/openkm"
## END CONFIG ##
echo -e "### BEGIN: $(date +"%x %X") ###\n"

# Stop JBoss
/etc/init.d/jboss stop
while [ "$(ps -ef | grep java | grep jboss | wc -l)" -gt "0" ]; do
  sleep 5; echo ".";
done

# Copy to backup server
ssh backup@server "cd $HOST; rm -rf backup.3; mv backup.2 backup.3; mv backup.1
backup.2; mv backup.0 backup.1"
rsync -apzhR --stats --delete --exclude=*~ --delete-excluded --link-
dest="/home/backup/$HOST/backup.1" $FILES backup@server:/home/backup/
$HOST/backup.0

# Start JBoss
/etc/init.d/jboss start
echo -e "\n### END: $(date +"%x %X") ###"

For more info, read http://www.mikerubel.org/computers/rsync_snapshots/.

Due to Jackrabbit flexibility you can configure your OpenKM installation to store the documents in a database, for example. In this case you have to modify the backup procedure. Anyway the $JBOSS_HOME/repository folder should be backup because it contains repository medatada.

There is an alternative method for backups: you can export the whole repository from OpenKM administration. The folder structure will be re-created in the local filesystem, and all document on them but you will lose the document history and other metadata like Property Groups.

Ensure you have enough free disk space before perform this operation.