We've been testing Zimbra on CentOS 5 and drbd for a couple of weeks and finally deployed it. Once we deployed it, and started migrating mail over (using imapsync) the system would simply stop the Zimbra processes or freeze completely.
I've increased the swap because I believe the system was running out of available memory (one the mailboxes I was migrating was 2gigs).
Needless to say, now Zimbra refuses to start because of a botched mysql database.
I'm reinstalling now, but should this ever happen again, what's the recovery procedure?