Re: Hot Standby remaining issues

On Tue, 2009-12-01 at 20:26 +0200, Heikki Linnakangas wrote:
> Simon Riggs wrote:
> > commit 02c3eadb766201db084b668daa271db4a900adc9
> > Author: Simon Riggs <sriggs(at)ebony(dot)(none)>
> > Date: Sat Nov 28 06:23:33 2009 +0000
> >
> > Added wal_standby_info GUC to turn RM_STANDBY_ID messages on/off.
> > Various comments added also.
> >
>
> This patch makes it unsafe to start hot standby mode from a shutdown
> checkpoint, because we don't know if wal_standby_info was enabled in the
> master.

> I still don't think we need the GUC.

If that's a good plan we can remove it in late beta. Let's keep it there
for now.

> But for future-proofing, perhaps we
> should add a flag to shutdown checkpoint records, indicating whether
> it's safe to start hot standby from it. That way, if we decide to add a
> GUC like that at a later stage, we don't need to change the on-disk format.

OK, I understand this now. Taken me a while, even though obvious now I
see.

--
Simon Riggs www.2ndQuadrant.com