On Mon, 2007-02-26 at 18:14 -0500, Tom Lane wrote:
> "Simon Riggs" <simon@2ndquadrant.com> writes:
> > A prototype patch is posted to -patches, which is WORK IN PROGRESS.
> > [This patch matches discussion thread on -hackers.]
>
> What does this accomplish other than adding syntactic sugar over a
> feature that really doesn't work well anyway? I don't see any point
> in encouraging people to use commit_delay in its present form. If we
> had a portable solution for millisecond-or-so waits then maybe it would
> work ...
This patch doesn't intend to implement group commit. I've changed the
meaning of commit_delay, sorry if that confuses.
You and I discussed this in Toronto actually, IIRC. The best way to
describe this proposal is deferred fsync, so perhaps a different
parameter commit_fsync_delay would be more appropriate.
Bruce has requested this feature many times from me, so I thought it
about time to publish.
The key point is that COMMIT NOWAIT doesn't wait for group commit to
return, it just doesn't wait at all - leaving someone else to flush WAL.
It's much better than fsync=off.
--
Simon Riggs
EnterpriseDB http://www.enterprisedb.com