Re: Vacuum Question

Tom Lane wrote:
>
> Ed Loehr <eloehr(at)austin(dot)rr(dot)com> writes:
> >>>>> % while test 1; do echo -n "`date`: "; psql -d pdb -c "vacuum analyze;
> >>>>> select count(*) from foo;"; sleep 3; done
> >>>>>
> >>>>> This seems to consistently crash after the first vacuum with the
> >>>>> following message:
> >>
> >> This is a known gotcha that's got nothing to do with any sort of
> >> concurrency. You can't safely send a VACUUM followed by anything
> >> else in a single query string.
>
> > Well, I thought that select count(*) might've been causing a problem, so
> > I experimented without it and found the same problem. Doesn't seem to
> > happen with 'vacuum'or 'vacuum analyze foo', only 'vacuum analyze'...
>
> I can't reproduce any problem with just a "vacuum" (with or without
> analyze) and no following command.
>
> I did, however, notice that very occasionally the inserting process
> would spit out weird error messages like "Function '(int4)' does not
> exist" and null probin for procedure 481". This seems
> to be due to VACUUM (on system tables) causing syscache entries to be
> flushed at unexpected times. I've committed patches for the two cases
> I observed, but there may be more lurking...

Yes, I was getting a similar Function error message, though I never saw
the init_fcache message. And it appeared the backend would crash
coincidentally with the function error message. If the patch can be
applied to 7.0 (beta3), and you don't mind posting it, I could test it
here...

Regards,
Ed Loehr