On Thu, 2008-12-18 at 08:29 +0200, Heikki Linnakangas wrote:
> Right, this is certainly not a new problem. It's not even a new problem
> in the context of replication or hot standby, because we already have
> the problem with PITR and file-based log shipping.
>
> Also, it's not just a problem *during* the recovery. The index is just
> as corrupt after the recovery has finished.
Agreed.
> I think we should just leave it alone for 8.4, and fix it properly in a
> future relase by implementing WAL-logging for hash indexes.
You really think we should just leave it alone? That gives me less work,
so I will accept that if you think so. Gives me the shivers though.
-- Simon Riggs www.2ndQuadrant.comPostgreSQL Training, Services and Support