On Mon, 2006-10-30 at 20:40 -0500, Tom Lane wrote:
> "Simon Riggs" <simon@2ndquadrant.com> writes:
> > That was understood; in the above example I agree you need to flush. If
> > you don't pass a truncation point, you don't need to flush whether or
> > not you actually truncate. So we don't need to flush *every* time,
>
> OK, but does that actually do much of anything for your performance
> complaint? Just after GlobalXmin has passed a truncation point, *every*
> vacuum the system does will start performing a flush-n-fsync, which
> seems like exactly what you didn't like. If the syncs were spread out
> in time for different rels then maybe this idea would help, but AFAICS
> they won't be.
Makes sense, so we shouldn't do it that way after all.
Are you OK with the other patches I've submitted? My understanding was
that you're gonna have a look at those and this general area? I don't
want to hold up the release because of a PITR patch.
Feedback welcome ;-)
--
Simon Riggs
EnterpriseDB http://www.enterprisedb.com