From: Heikki Linnakangas <heikki@enterprisedb.com>
To: Pomarede Nicolas <npomarede@corp.free.fr>
Cc: pgsql-performance@postgresql.org
Subject: Re: truncate a table instead of vaccum full when count(*) is 0
Date: Tue, 08 May 2007 11:59:48 +0100
Message-ID: <46405824.9010000@enterprisedb.com> (raw)
In-Reply-To: <Pine.LNX.4.64.0705081242570.22289@localhost>
References: <Pine.LNX.4.64.0705081125140.20675@localhost>
<46404BEA.5060009@enterprisedb.com>
<Pine.LNX.4.64.0705081214250.22289@localhost>
<464052DB.30206@enterprisedb.com>
<Pine.LNX.4.64.0705081242570.22289@localhost>
Pomarede Nicolas wrote:
> There's not too much simultaneous transaction on the database, most of
> the time it shouldn't exceed one minute (worst case). Except, as I need
> to run a vacuum analyze on the whole database every day, it now takes 8
> hours to do the vacuum (I changed vacuum values to be a little slower
> instead of taking too much i/o and making the base unusable, because
> with default vacuum values it takes 3-4 hours of high i/o usage (total
> base is 20 GB) ).
>
> So, at this time, the complete vacuum is running, and vacuuming only the
> spool table gives all dead rows are currently not removable (which is
> normal).
Oh, I see. I know you don't want to upgrade, but that was changed in
8.2. Vacuum now ignores concurrent vacuums in the oldest xid
calculation, so the long-running vacuum won't stop the vacuum on the
spool table from removing dead rows.
--
Heikki Linnakangas
EnterpriseDB http://www.enterprisedb.com
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Reply to all the recipients using the --to and --cc options:
reply via email
To: pgsql-performance@postgresql.org
Cc: heikki@enterprisedb.com, npomarede@corp.free.fr
Subject: Re: truncate a table instead of vaccum full when count(*) is 0
In-Reply-To: <46405824.9010000@enterprisedb.com>
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox