Received: from localhost (unknown [200.46.204.187])
	by postgresql.org (Postfix) with ESMTP id AB39A2E0F8F
	for <pgsql-performance-postgresql.org@postgresql.org>;
	Wed, 14 Nov 2007 18:46:30 -0400 (AST)
Received: from postgresql.org ([200.46.204.71])
	by localhost (mx1.hub.org [200.46.204.187]) (amavisd-maia, port 10024)
	with ESMTP id 41903-03 for
	<pgsql-performance-postgresql.org@postgresql.org>;
	Wed, 14 Nov 2007 18:46:22 -0400 (AST)
Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130])
	by postgresql.org (Postfix) with ESMTP id 35EB52E1C5B
	for <pgsql-performance@postgresql.org>;
	Wed, 14 Nov 2007 18:46:26 -0400 (AST)
Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1])
	by sss.pgh.pa.us (8.14.1/8.14.1) with ESMTP id lAEMkOIE017000;
	Wed, 14 Nov 2007 17:46:24 -0500 (EST)
To: Russell Smith <mr-russ@pws.com.au>
cc: Jean-David Beyer <jeandavid8@verizon.net>,
 pgsql-performance@postgresql.org
Subject: Re: Curious about dead rows. 
In-reply-to: <473B6C9D.1020804@pws.com.au> 
References: <4735DBDB.5050208@verizon.net> <16627.1194714491@sss.pgh.pa.us>
	<4735FA9F.1080805@verizon.net>
	<b42b73150711131628q1469df3el6f557e10de07a05@mail.gmail.com>
	<473A5CC0.3060003@verizon.net>
	<b42b73150711131936l65b97b9ya7953383673b2710@mail.gmail.com>
	<473AE63D.7090305@verizon.net>
	<20071114152249.GA16041@crankycanuck.ca>
	<473B2A1F.30804@verizon.net> <473B6C9D.1020804@pws.com.au>
Comments: In-reply-to Russell Smith <mr-russ@pws.com.au>
	message dated "Thu, 15 Nov 2007 08:46:05 +1100"
Date: Wed, 14 Nov 2007 17:46:24 -0500
Message-ID: <16999.1195080384@sss.pgh.pa.us>
From: Tom Lane <tgl@sss.pgh.pa.us>
X-Virus-Scanned: Maia Mailguard 1.0.1
X-Archive-Number: 200711/226
X-Sequence-Number: 27641

Russell Smith <mr-russ@pws.com.au> writes:
> It is possible that analyze is not getting the number of dead rows right?

Hah, I think you are on to something.  ANALYZE is telling the truth
about how many "dead" rows it saw, but its notion of "dead" is "not good
according to SnapshotNow".  Thus, rows inserted by a not-yet-committed
transaction would be counted as dead.  So if these are background
auto-analyzes being done in parallel with inserting transactions that
run for awhile, seeing a few not-yet-committed rows would be
unsurprising.

I wonder if that is worth fixing?  I'm not especially concerned about
the cosmetic aspect of it, but if we mistakenly launch an autovacuum
on the strength of an inflated estimate of dead rows, that could be
costly.

			regards, tom lane