On Tue, 2006-02-14 at 15:55 -0500, Tom Lane wrote:
> Simon Riggs <simon@2ndquadrant.com> writes:
> > On Tue, 2006-02-14 at 13:20 -0400, Tom Lane wrote:
> >> add missing vacuum_delay_point() call in btvacuumcleanup.
>
> > Cool and Interesting. That might explain some pretty dire performance
> > numbers from last week while running auto vacuum. Performance was
> > flat-lining for a while. Still need to investigate further though.
>
> Yeah, the missing delay would result in a spike in I/O demand from
> vacuum (auto or otherwise) while processing a big index, if you had
> vacuum delay configured. GIST had the same problem, too.
Perhaps if vacuum_delay_point() contained a timer check, we'd be able to
see if any gap between vacuum delays was more than the actual delay
itself. It would be nice to know they are all gone, forever.
Best Regards, Simon Riggs