On Tue, 16 Apr 2002, Dmitry Tkach wrote:
> > vacuum analyze;
>
> Yeah... I did that yesterday... It took about 24 hours (!) to run... And
> I don't see any difference in the behaviour :-(
>
> >
> > explain select * from a order by id limit 1;
>
>
> Aha... This is an interesting thing to try. Thanks.
> Unfortunately, I can't try that right now either! :-(
> I am running 'vacuum full' on that table (out of despare), and, despite
> what I have repeatedly heard about
> vaccums in 7.2, my 'explain' command seems to be hopelessly waiting on a
> lock, created by that vacuum
> :-(
> Is it supposed to be that way.
Unfortunately, yes. VACUUM FULL is effectively the 7.1 and earlier
vacuum. Normal vacuum wouldn't have helped you in this case since it
wouldn't compress the table (IIRC it only moves tuples within a block
rather than between them, so all the dead blocks at the beginning are
still there).