Re: Checkpoint spikes

Heikki Linnakangas wrote:
> I wonder how common this issue is? When we implemented spreading of the
> write phase, we had long discussions about spreading out the fsyncs too,
> but in the end it wasn't done. Perhaps it is time to revisit that now
> that 8.3 has been out for some time and people have experience with the
> load-distributed checkpoints.
>
Circa 8.2, I ran into checkpoint problems all the time. With the
spreading logic in 8.3, properly setup, the worst case is so improved
that I usually find something else more pressing to tune, rather than
worry about the exact details of the sync process. It seems to have hit
the "good enough" point where it's hard to justify time for further
improvements when there are other things to work on. I'd still like to
see spread fsync happen one day, just hasn't been a priority for any
systems I have to improve lately.

--
Greg Smith 2ndQuadrant Baltimore, MD
PostgreSQL Training, Services and Support
greg(at)2ndQuadrant(dot)com www.2ndQuadrant.com