Date: Wed, 18 May 2005 00:57:26 +0200
From: "Steinar H. Gunderson" <sgunderson@bigfoot.com>
To: pgsql-performance@postgresql.org
Subject: Re: Is there any other way to do this?
Message-ID: <20050517225726.GA24185@uio.no>
Mail-Followup-To: pgsql-performance@postgresql.org
References: <428A770C.1030107@kencast.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: <428A770C.1030107@kencast.com>
User-Agent: Mutt/1.5.9i

On Tue, May 17, 2005 at 06:58:20PM -0400, Wei Weng wrote:
> This time it worked! But VACUUM FULL requires an exclusive lock on the 
> table which I don't really want to grant. So my question is: why is VACUUM 
> ANALYZE didn't do the job? Is there any setting I can tweak to make a 
> VACUUM without granting a exclusive lock?

You just didn't vacuum often enough. Plain VACUUM (with ANALYZE or not) only
deletes dead rows, it does not reclaim the space used for them (and thus does
not compress the remaining ones into fewer pages, so they take less time to
scan). If you simply VACUUM regularily (try autovacuum from contrib, it will
probably be useful) the problem simply will never be as bad as you describe
here, and you won't need to VACUUM FULL.

/* Steinar */
-- 
Homepage: http://www.sesse.net/