Re: Perfomance decreasing - Mailing list pgsql-general

From Erwin Lansing
Subject Re: Perfomance decreasing
Date
Msg-id 20010820212301.E88482@mail.droso.net
Whole thread Raw
In response to Perfomance decreasing  (Alexander Loginov <sas@mplik.ru>)
List pgsql-general
On Mon, Aug 20, 2001 at 02:41:05PM -0400, wsheldah@lexmark.com wrote:
>
>
> Does it help if you drop and recreate the indexes, in addition to the vacuuming
> you're doing now?  I think this was suggested not long ago on this list.

I reduced the number of times vacuum was run with analyze, and run a
normal vacuum twice a day. The files are no longer growing since. I'm
leaving for hollidays now, I'll investigate more when I return next
week.

/erwin
>
>
>
>
> Erwin Lansing <erwin%lansing.dk@interlock.lexmark.com> on 08/14/2001 04:38:59 AM
>
> To:   pgsql-general%postgresql.org@interlock.lexmark.com
> cc:    (bcc: Wesley Sheldahl/Lex/Lexmark)
> Subject:  Re: [GENERAL] Perfomance decreasing
>
>
> On Tue, Aug 14, 2001 at 02:06:40PM +0600, Alexander Loginov wrote:
> > Hello.
> >
> >        I have a question about perfomance.
> >        I'm running PostgreSQL 7.1.2 at FreeBSD 4.3.
> >
> >        For  the first 1-2 days of running perfomance is excellent. But
> >        after  that,  speed  began  to  decrease.  And  after a week of
> >        operation, perfomance  falls  8-10  times, than at first day of
> >        using.
> >
> >        I'm  doing  vacuum  periodically  (once a hour), but perfomance
> >        still falls down.
> >
> >        After that I dump database as text file, make dropdb & createdb
> >        and  after  that,  restore  database from dump -> Perfomance is
> >        excellent again (for 1-2 days).
> >
> >        Why this situation occures? May be I must use "VACUUM ANALYSE"
> >        instead of VACUUM?
> >
>
> I have actually the same problem, also FreeBSD 4.3, pgsql 7.1.2. I do
> use VACUUM ANALYSE quite often. The problem in the end gets that bad
> that perl-jobs cannot perform any SELECTs, or at least they stop
> returning results before dbi times out. So far I have tracked the
> problem down to the size of the database in the filesystem, where
> problems start occurring when it exceeds 1,4 Gb. A
> dump/drop/create/restore reduces files size to approx. 350 Mb.
>
> Any pointers would be helpful as a weekly dump/restore is not quite
> optimal :)
>
> /erwin
>
> --
> Erwin Lansing       --        http://droso.org
> "You've got mail"
>
> ---------------------------(end of broadcast)---------------------------
> TIP 4: Don't 'kill -9' the postmaster
>
>
>
--
Erwin Lansing         --         http://droso.org
"You've got mail"

pgsql-general by date:

Previous
From: Andrew Gould
Date:
Subject: Re: Problems with UPDATE
Next
From: Doug McNaught
Date:
Subject: Re: How do you recover a postgres db?