If your system has a period of minimal activity (such as at 1:00 AM) you can do the backups without taking the system offline. If you perform this step every night, it is usually possible to restore without problems. The worst case scenario would be that the Lucene directory was caught adding an entry when the snapshot was taken. This can be remedied by re-building the search index from the portlet.
Take the system down on a scheduled basis.
During the down period:
Take a snapshot of the database
Make a backup copy of the Lucene index, default location is:
/opt/icecore/luceneserver/liferay.com
Make a backup of the file repositories that are in use, for example:
/home/icecoredata/filerepository
Restart your ICEcore system.
NOTE:This approach is the most conservative. It guarantees a consistent snapshot of your whole system. However, doing a backup of the file repository can take a long time depending on your backup approach.
Take the system down on a scheduled basis.
During the down period:
Take a snapshot of the database
Make a backup copy of the Lucene index
Bring the system back online.
After the system is running again, start the backup of the file repositories.
NOTE:This approach minimizes the down time. However, it is possible that during a restore after a catastrophic failure the files may be out of synchronization with the metadata. This does cause any problems when running the system, but users may find that slightly newer versions of the files exist in the restored system than were present at the time of the database snapshot. This is usually a good thing.