On Wed, 30 Apr 2003, Bruce Momjian wrote:
> Medi Montaseri wrote:
> > I am using PG 7.2 with fsync truned on but still need a more immediate
> > write thru.
> > Does PG uses the fsync (or open_sync, etc) just for the WAL or for
> > tables as well ?
>
> Just for WAL.
>
> > I understand the consequence of limiting buffer cache but my application
> > cares more about
> > integrity than performance. I'd like to be able to write the data to the
> > physical storage ASAP,
> > or ideally 100% in sync mode.
> >
> > I have found that by reducing the checkpoint parameters I can write the
> > data to disk sooner.
> > For example
> >
> > checkpoint_segments = 1
> > checkpoint_timeout = 30
> >
> > What else do I have to work with....?
>
> PostgreSQL is already 100% reliable (pull plug and see) so I don't see
> any value to changing those parameters.
What about commit_siblings and commit_delay? I haven't really played a
lot with those. I would think increasing the commit delay and the number
of siblings should give better, but slightly bursty performance.