From: Gaetano Mendola <mendola@bigfoot.com>
Subject: Re: Memory usage on subselect
Date: Wed, 26 May 2004 18:29:42 +0200
Organization: PYRENET Midi-pyrenees Provider
Lines: 39
Message-ID: <40B4C5F6.3010700@bigfoot.com>
References: <20040522175419.GB54296@dyatel.antar.bryansk.ru>
	<24527.1085329698@sss.pgh.pa.us>
	<20040523183249.GC54296@dyatel.antar.bryansk.ru>
	<D11F1F08-AD5F-11D8-9250-000A958E367A@llgc.org.uk>
	<10740.1085405857@sss.pgh.pa.us>
	<9ADF76EC-AF2C-11D8-9250-000A958E367A@llgc.org.uk>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
To: Dan Field <dof@llgc.org.uk>
User-Agent: Mozilla Thunderbird 0.6 (Windows/20040502)
In-Reply-To: <9ADF76EC-AF2C-11D8-9250-000A958E367A@llgc.org.uk>
To: pgsql-sql@postgresql.org

Dan Field wrote:

> 
> On 24 May 2004, at 14:37, Tom Lane wrote:
> 
>> Dan Field <dof@llgc.org.uk> writes:
>>
>>> I have a similar problem with just one of my queries (although it isn't
>>> a sub select):
>>
>>
>> You really ought to vacuum and/or analyze occasionally.  The EXPLAIN
>> results show that the planner hasn't got any non-default statistics
>> for any of these tables.
> 
> 
> Wow, thanks for that. I'd been pulling my hair out for a couple of days 
> wondering where I was going wrong.
> I went from 45 second queries down to sub second query lengths  after a 
> simple vacuum full analyze.
> 
> I've now added nightly and monthly cron jobs to do this for me in future.
> 
> Out of curiosity, why is this deemed a DBA task rather than an automated 
> postgres task?
> 
> Once again, many thanks.

You have to use the pg_autovacuum demon.

Run the vacuum full and the reindex once in a week.


Regards
Gaetano Mendola