On Tue, 2002-05-28 at 15:43, Curt Sampson wrote:
> On Tue, 28 May 2002, Tom Lane wrote:
>
> > Stephen Robert Norris <srn@commsecure.com.au> writes:
> > > One big difference, though, is that with the vacuum problem, the CPU
> > > used is almost all (99%) system time; loading up the db with lots of
> > > queries increases user time mostly, with little system time...
> >
> > Hmm, that's a curious point; leaves one wondering about possible kernel
> > bugs.
>
> If it turns out to be so, this would not be the first problem I've
> heard about in the Linux scheduler. (It was notoriously bad for years.)
>
> I'd suggest a good test would be to try this on a BSD machine and
> see if the problem exists there, too. That will at least tell you
> if it's Postgres or Linux.
>
> cjs
> --
> Curt Sampson <cjs@cynic.net> +81 90 7737 2974 http://www.netbsd.org
> Don't you know, in this new Dark Age, we're all light. --XTC
Can someone with access to a BSD machine try this? It's pretty easy to
set up, a simple loop to open a few hundred connections and a 1-line
shell script. It doesn't seem to matter what's in the database...
I'm going to try the O(1) scheduler patch for the linux kernel and see
if that helps...
Stephen