Hi, I was wondering if anyone has considered or discussed eDirectory
database consistency when taking snapshots? We use a product called
Veeam for backup and replication which relies on VMWare snapshots. The
developers of Veeam stress that it is important to bring databases to a
transaction consistent state before taking a snapshot in VMWare (by
stopping or suspending the database). Windows has Volume Shadow Copy
Service (VSS) to do this natively, but Linux does not, so in VMWare the
recommended practice is to use -pre-freeze/post-thaw scripts- that stop
databases automatically before and after taking snapshots. If we do not
stop/suspend the database, we would get a crash-consistent snapshot
(without WMWare quiescence) or a file system consistent snapshot (with
VMWare quiescence). These two levels of consistency are insufficient
and may cause corruption in databases when reverting to these snapshots.
Thus, the goal for databases is to use transaction-consistent
snapshots, which can be achieved by stopping or suspending the database
before snapping.

Since I understand that eDirectory uses FLAIM databases, is there a
recommended way to stop it or should we simply use the "rcndsd stop"
command? Also, since we backup and replicate nightly, is it harmful to
stop/restart eDirectory twice nightly when Veeam takes snapshots of our
VM's? Thanks for insight that anyone may have, as there seems to be
very little documentation concerning this topic.

bhnguyen's Profile: https://forums.netiq.com/member.php?userid=4430
View this thread: https://forums.netiq.com/showthread.php?t=46927