From: Tom Lane <tgl@sss.pgh.pa.us>
To: Jeff Janes <jeff.janes@gmail.com>
cc: Guillaume Smet <guillaume.smet@gmail.com>,
 pgsql-performance@postgresql.org
Subject: Re: Any idea on how to improve the statistics estimates for this
 plan?
In-reply-to: 
 <CAMkU=1x9GoZb_6HQf3sF=mN6NEJhwRFsJMO1AwJBh0nDk34sJQ@mail.gmail.com>
References: 
 <CALt0+o9EdN9HDWuzNO2-+Y4XWU6OT03jzvHdzOCrHsLa5dgyMQ@mail.gmail.com>
	<CAMkU=1x9GoZb_6HQf3sF=mN6NEJhwRFsJMO1AwJBh0nDk34sJQ@mail.gmail.com>
Comments: In-reply-to Jeff Janes <jeff.janes@gmail.com>
	message dated "Fri, 07 Dec 2012 18:32:15 -0800"
Date: Fri, 07 Dec 2012 23:16:45 -0500
Message-ID: <2904.1354940205@sss.pgh.pa.us>
Precedence: bulk
Sender: pgsql-performance-owner@postgresql.org

Jeff Janes <jeff.janes@gmail.com> writes:
> The trivial answer to "why" is that it thinks that the vast majority
> of the 33914 rows from the hash join will find no partners in r, but
> in fact each has about 10 partner in r.  Why does it think that?

I'm wondering if maybe the vast majority of the rows indeed have no join
partners, but there are a small number with a large number of partners.
The statistics might miss these, if so.

			regards, tom lane


-- 
Sent via pgsql-performance mailing list (pgsql-performance@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-performance