Veritas-bu

[Veritas-bu] DB backup tapes and DR

2006-12-22 16:07:25
Subject: [Veritas-bu] DB backup tapes and DR
From: bob944 at attglobal.net (bob944)
Date: Fri, 22 Dec 2006 16:07:25 -0500
> Ok I have a DR question for everyone. I have 3 sites. All 3 of them
> have master servers. All sites do tape vaulting and send their tapes
> to the main site for DR storage of a week and then are sent back. 
> However, the db backup tapes for netbackup do not get vaulted out.
> I tried this once and when the db backup job ran the next day for
> the one remote site it complained about not having its netbackup
> tape in the robot. So my question is, How does everyone else
> handle remote sites that have masters and db backup tapes?  Our
> main site backs up to disk so that is not a problem but the
> smaller remote sites keep their tapes onsite. Any
> ideas or suggestions?

Greg - did this ever get answered?  I didn't see it when I was catching
up.

Regarding:
> However, the db backup tapes for netbackup do not get vaulted out.
> I tried this once and when the db backup job ran the next day for
> the one remote site it complained about not having its netbackup
> tape in the robot.

It shouldn't have been ejected, unless you have a process to eject it
and replace it, physically, and update the bpsyncinfo data in the
configure catalog backups interface.

You can it differently, but the way Vault has always worked is that it
produces a _new_ catalog backup tape--you don't eject either of the two
offline-catalog-backup tapes.  This part of the sequence is 

o  duplicate images (optional) to Duplicates (historical, not required)
pool
o  duplicate (actually, make a new one) the catalog backup to
NBDB_Duplicates (again, the old 3.1 name for the pool)
o  eject everything in Duplicates and NBDB_Duplicates
o  inject everything returned from offsite

The new catalog backup your produce will also have the information about
the duplicatations that happen just before you produce it, as well.

Note that the offline catalog backup process does _not_ draw tapes from
your scratch pool as normal backups do.  You must seed the
NBDB_Duplicates pool with tapes or Vault's catalog backup step will
fail.  As there is nothing with less value than an older catalog backup
tape, if you have a daily run from your offsite vendor, and handle tapes
individually (rather than in cases), set the retention on
NBDB_Duplicates to, say, 7 days (no matter what the retention of the
other tapes you send out) and have more than seven tapes in that pool.
That covers days/weekends/holidays/mistakes when one doesn't get brought
back for a few days and you don't have to worry about it again.