To: Sami Dalouche <skoobi@free.fr>
cc: pgsql-performance <pgsql-performance@postgresql.org>
Subject: Re: [Fwd: Re: Outer joins and Seq scans] 
In-reply-to: <1193865739.23774.2.camel@samlaptop> 
References: <1193865739.23774.2.camel@samlaptop>
Comments: In-reply-to Sami Dalouche <skoobi@free.fr>
	message dated "Wed, 31 Oct 2007 22:22:19 +0100"
Date: Wed, 31 Oct 2007 18:05:16 -0400
Message-ID: <21628.1193868316@sss.pgh.pa.us>
From: Tom Lane <tgl@sss.pgh.pa.us>

Sami Dalouche <skoobi@free.fr> writes:
> -- For some reason, my message doesn't seem to go through the mailing
> list, so I am trying without any attachment

Please don't do that, at least not that way.  These explain outputs have
been line-wrapped to the point of utter unreadability.

The main problem looks to me that you're trying to do a 25-way join.
You'll want to increase join_collapse_limit and maybe fool with the
geqo parameters.  I fear you won't get a plan in a sane amount of time
if you try to do the full query as a single exhaustive search.  You
can either raise join_collapse_limit all the way and trust geqo to
find a decent plan repeatably (not a real safe assumption unfortunately)
or raise both join_collapse_limit and geqo_threshold to some
intermediate level and hope that a slightly wider partial plan search
will find the plan you need.

It's also possible that you're just stuck and the outer join is
inherently harder to execute.  I didn't study the query closely enough
to see if it's joining to any left join right-hand-sides, or anything
else that would forbid picking a nice join order.

			regards, tom lane