Date: Sun, 21 Dec 2008 21:46:48 -0300
From: Alvaro Herrera <alvherre@commandprompt.com>
To: Michael Monnerie <michael.monnerie@is.it-management.at>
Cc: pgsql-admin@postgresql.org
Subject: Re: reindexdb dying with SIGPIPE on 8.2.5
Message-ID: <20081222004648.GC8720@alvh.no-ip.org>
References: <1218035421.19154.24.camel@oberschatzi.cff>
	<200812190311.58087@zmi.at> <20081219030959.GM5447@alvh.no-ip.org>
	<200812212121.05678@zmi.at>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <200812212121.05678@zmi.at>
User-Agent: Mutt/1.5.18 (2008-05-17)

Michael Monnerie wrote:
> On Freitag 19 Dezember 2008 Alvaro Herrera wrote:
> > You don't need to vacuum or reindex after a cluster; it removes all
> > dead rows (except those that vacuum won't remove anyway)
> 
> OK, but I do not cluster all tables, as it only makes sense for tables 
> which get read in some sequential (based on that index) way. For 
> completely random accessed tables I skip cluster, and rely on 
> vacuum/reindex.

Right -- I'm just saying that if you cluster a table, you don't need to
vacuum it.  Of course, you need to vacuum all the other tables.

>  Also, I think pg_toast tables cannot be clustered, right?

Not directly, but if you cluster a plain table, its toast table is
clustered too.

-- 
Alvaro Herrera                                http://www.CommandPrompt.com/
The PostgreSQL Company - Command Prompt, Inc.