Message-ID: <491DB1E9.6030500@archonet.com>
Date: Fri, 14 Nov 2008 17:14:17 +0000
From: Richard Huxton <dev@archonet.com>
User-Agent: Thunderbird 2.0.0.17 (X11/20080914)
MIME-Version: 1.0
To: Patrice Beliveau <pbeliveau@avior.ca>
CC: pgsql-performance@postgresql.org
Subject: Re: Difference in query plan
References: <491DB061.4020002@avior.ca>
In-Reply-To: <491DB061.4020002@avior.ca>
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit

Patrice Beliveau wrote:
> Thanks,
> 
> I'm already doing a vacuum full every night on all database, but the
> REINDEX fix it and now it's working fine

Are you sure it was the REINDEX? The plan was using a sequential scan.

> But this raise a question
> 
> 1) This table is cleared every night and recomputed, does this mean that
> I should REINDEX every night also

Looks like you should. Or drop the indexes, load the data, re-create the
indexes, that can be quicker.

> 2) Why this thing didn't happen in the other schema

Have you re-loaded schema1 more often? It might even be the particular
order that rows are loaded - a btree can become "unbalanced" sometimes.

-- 
  Richard Huxton
  Archonet Ltd