Date: Tue, 10 Oct 2006 20:19:53 +0200
From: Tobias Brox <tobias@nordicbet.com>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: Tobias Brox <tobias@nordicbet.com>,
	pgsql-performance@postgresql.org
Subject: Re: long running transactions
Message-ID: <20061010181953.GA7584@oppetid.no>
References: <25404.1160497420@sss.pgh.pa.us>
 <20061010163913.GA8097@oppetid.no>
	<20061010164142.GA5360@oppetid.no> <25631.1160498572@sss.pgh.pa.us>
	<20061010170639.GB5360@oppetid.no> <25953.1160500192@sss.pgh.pa.us>
	<20061010171304.GC5360@oppetid.no> <26123.1160500707@sss.pgh.pa.us>
	<20061010174955.GA284@oppetid.no> <528.1160503495@sss.pgh.pa.us>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <528.1160503495@sss.pgh.pa.us>
Organization: Group Nordicbet
User-Agent: Mutt/1.5.9i

[Tom Lane - Tue at 02:04:55PM -0400]
> > It seems stuck, has had the same transid for a long while, and the
> > number of undeletable dead rows in our tables are increasing.
> 
> Perhaps you have overly aggressive vacuum cost delay settings?

Perhaps, though I wouldn't expect it to sleep in the middle of a
transaction - and also, it really did seem to me that it's doing work
rather than only sleeping.  

The transaction id for the vacuum process is the same now as when I
wrote the previous email, and the number of dead unremovable rows have
increased steadily.

The settings in effect are:

autovacuum_vacuum_cost_delay = 500
autovacuum_vacuum_cost_limit = 200