> Hello
>
> I have similar problem with vacuum on 8.1
>
> I have 256M table. pgstattuple reports 128M free. I stopped vacuum
> after 1hour (maintenance_work_mem = 160M). I had not more time.
>
I test it on 8.3 with random data. Vacuum from 190M to 94M neded
30sec. It's much better. It isn't 100% comparable, but it is one from
more arguments for upgrade.
Regards
Pavel Stehule
>
>
> 2007/7/10, Tom Lane <tgl@sss.pgh.pa.us>:
> > Brad Nicholson <bnichols@ca.afilias.info> writes:
> > > On Tue, 2007-07-10 at 11:19 -0400, Tom Lane wrote:
> > >> Oh, I forgot to mention --- you did check that vacuum_mem is set to
> > >> a pretty high value, no? Else you might be doing a lot more
> > >> btbulkdelete scans than you need to.
> >
> > > What would you define as high for 7.4? I bumped it up to ~ 245mbs
> >
> > That sounds like plenty --- you only need 6 bytes per dead tuple,
> > so that should be enough to handle all your 15-20M dead tuples in
> > one scan.
> >
> > How big is this index again?
> >
> > regards, tom lane
> >
> > ---------------------------(end of broadcast)---------------------------
> > TIP 4: Have you searched our list archives?
> >
> > http://archives.postgresql.org/
> >
>