Re: Backup hot-standby database.

From: Robert Treat <rob(at)xzilla(dot)net>
To: Stephen Rees <srees(at)pandora(dot)com>
Cc: pgsql-admin(at)postgresql(dot)org
Subject: Re: Backup hot-standby database.
Date: 2011-03-21 15:49:15
Message-ID: AANLkTi=yJun+gVppAV2Un5K8OXrynt3_pB4s2M=1cqhr@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-admin

On Fri, Mar 18, 2011 at 4:55 PM, Stephen Rees <srees(at)pandora(dot)com> wrote:
> Robert,
>
> Thank you for reply. I had the wrong end of the stick regarding pg_dump and
> hot-standby.
> I will take a look at omnipitr, as you suggest.
>
> Per your comment
>>
>> You have to stop replay while you are doing the dumps like this
>
> how do I stop, then resume, replay with both the master and hot-standby
> available throughout?
>

If you are using WAL file based replication, you need some logic in
your restore script that will enable it to stop feeding xlog segments
into the slave (think "if pause file exists, return, else cp xlog
file"). This would leave the slave available, just with no updates
coming in.

If you are using streaming, I think it's much harder. There are some
new function to pause and resume streaming WAL coming in 9.1, it might
be possible to back-patch those, but we haven't looked at it yet.

Robert Treat
play: xzilla.net
work: omniti.com
hiring: l42.org/lg

In response to

Browse pgsql-admin by date

  From Date Subject
Next Message Paul Gross 2011-03-21 16:08:35 Restarting recovery mode with pg_standby after failure
Previous Message Tapio Pitkäranta 2011-03-21 11:53:07 Re: Memory limits and PostgreSQL 8.4 - Over 63 GB of memory for DB?