Message-ID: <42A93E91.5070502@archonet.com>
Date: Fri, 10 Jun 2005 08:17:37 +0100
From: Richard Huxton <dev@archonet.com>
User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050317)
MIME-Version: 1.0
To: Yann Michel <yann-postgresql@spline.de>
Cc: pgsql-performance@postgresql.org
Subject: Re: postgresql.conf runtime statistics default
References: <20050609100852.GA16538@zoom.spline.inf.fu-berlin.de>
	<42A83FFA.3030209@archonet.com>
	<20050610045702.GA19866@zoom.spline.inf.fu-berlin.de>
In-Reply-To: <20050610045702.GA19866@zoom.spline.inf.fu-berlin.de>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

Yann Michel wrote:
> Hi,
> 
> On Thu, Jun 09, 2005 at 02:11:22PM +0100, Richard Huxton wrote:
> 
>>>To my question: I found the parameter "stats_reset_on_server_start"
>>>which is set to true by default. Why did you choose this (and not false)
>>>and what are the impacts of changeing it to false? I mean, as long as I
>>>understood it, each query or statements generates some statistic data
>>>which is used by the optimizer (or anything equal) later on. So in my
>>>oppinion, wouldn't it be better so set this parameter to false and to
>>>enable a kind of a "startup reset_stats" option?
>>
>>This is administrator statistics (e.g. number of disk blocks read from 
>>this index) not planner statistics. You're right - it would be foolish 
>>to throw away planner stats.
> 
> 
> So what is best to set this parameter to and when? As I read this
> parameter is documented within the section "16.4.7.2. Query and Index
> Statistics Collector" so I guess it is better to set it to false as
> described above. Or am I wrong?

It depends on whether you want to know how much activity your 
tables/indexes have received *ever* or since you last restarted. If you 
altered your database schema, added/removed indexes or changed 
hardware/configuration then you might want to reset the counts to zero 
to more easily see the effect of the new setup.

-- 
   Richard Huxton
   Archonet Ltd