Subject: Re: Backup/disaster recovery and bandwidth (long)
Mime-Version: 1.0 (Apple Message framework v1084)
Content-Type: text/plain; charset=us-ascii
From: Scott Ribe <scott_ribe@elevated-dev.com>
In-Reply-To: <da3b784d-86db-4b60-a6cd-149efaa58c0e@zimbra.int.journyx.com>
Date: Wed, 25 Apr 2012 11:23:12 -0600
Cc: pgsql-admin@postgresql.org
Content-Transfer-Encoding: quoted-printable
Message-Id: <8D77E0D1-4AA7-4892-9508-58B7F21DDB18@elevated-dev.com>
References: <da3b784d-86db-4b60-a6cd-149efaa58c0e@zimbra.int.journyx.com>
To: Scott Whitney <swhitney@journyx.com>,
 Scott Whitney <scott@journyx.com>

On Apr 25, 2012, at 10:11 AM, Scott Whitney wrote:

> I believe, then, that when I restart server #3 (the standby who is =
replicating), he'll say "oh, geez, I was down, let me catch up on all =
that crap that happened while I was out of the loop," he'll replay the =
WAL files that were written while he was down, and then he'll catch back =
up.
>=20
> Does this sound like a viable option? Or does someone have additional =
suggestions?

Perfectly viable. However, the WAL files must exist for this to happen. =
So you need to set wal_keep_segments appropriately, or set up WAL =
archiving.

Note that you could even provide a more up-to-date database for your =
people to work with. If the testbed is nearly up to date, then an rsync =
to update it would take very little time. So you could shut down the =
replica, rsync, and bring the replica back up.


--=20
Scott Ribe
scott_ribe@elevated-dev.com
http://www.elevated-dev.com/
(303) 722-0567 voice