Message-ID: 
 <3c7898fc579b37bfec2f5ff971df7532.squirrel@mail.brad.office.comodo.net>
In-Reply-To: <25082.1269800427@sss.pgh.pa.us>
References: <201003281841.41865.mike.williams@comodo.com>
	<25082.1269800427@sss.pgh.pa.us>
Date: Sun, 28 Mar 2010 20:29:52 +0100
Subject: Re: ERROR: could not open segment 1 of relation
	1663/743352/743420 (target block 6407642): No such file or directory
From: "Mike Williams" <mike.williams@comodo.com>
To: "Tom Lane" <tgl@sss.pgh.pa.us>
Cc: pgsql-admin@postgresql.org
User-Agent: SquirrelMail/1.4.19
MIME-Version: 1.0
Content-Type: text/plain;charset=iso-8859-1
Content-Transfer-Encoding: 8bit
Importance: Normal

On Sun, March 28, 2010 7:20 pm, Tom Lane wrote:
> Mike Williams <mike.williams@comodo.com> writes:
>> ERROR:  could not open segment 1 of relation 1663/743352/743420 (target
>> block
>> 2171336): No such file or directory
>> CONTEXT:  SQL statement "SELECT day FROM min_stats WHERE day<= $1  AND
>> unique_ip=-1 ORDER BY day DESC LIMIT 1"
>> PL/pgSQL function "update_unique_ips" line 13 at SQL statement


> Wow.  The symptom seems to be what you'd get from a corrupt tuple
> pointer in an index, but it's hard to believe that such a thing would
> persist over reindexes and even reloads.  Is the "target block" number
> always the same, or does that change?  Can you reproduce the problem on
> a different machine?  What platform are you running on?

While it's always pleasing to find an error condition that makes an expert
go "Wow", it is also quite troubling! :)

The "target block" stays the same for extended periods, but does change.
Oddly it's been quite for a while now, half an hour or so ago it had been
going nearly every minute for a good hour. For the last 30 times the block
was 11387925, before then 6634333 once, then 15520041 for 4 times, 6667190
before that for a while.
And just as I'm about to hit send, 3 more errors. All block 14528541.

Where does the 1663 number come from?
<datadir/base>/743352/743420 is a real file, but I don't have any file 1663.

The server is Gentoo, and I don't have an identical server to use. This
machines HA partner is staying at 8.3.5 until we're completely and
entirely sure 8.3.10 works right.

-- 
Mike Williams