Message-ID: <46954233.4050807@p-dw.com>
Date: Wed, 11 Jul 2007 22:48:51 +0200
From: Patric de Waha <lists@p-dw.com>
User-Agent: Thunderbird 1.5.0.12 (Windows/20070509)
MIME-Version: 1.0
CC: pgsql-performance@postgresql.org
Subject: Re: Two questions.. shared_buffers and long reader issue
References: <4694F8C5.2040005@p-dw.com> <27288.1184173062@sss.pgh.pa.us>
In-Reply-To: <27288.1184173062@sss.pgh.pa.us>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

Ok thanks.

iostat confirmed it's an IO bottleneck.
Will add some discs to the RAID unit.

Used 4 Raptor discs in Raid 10 until now.


best regards,
    patric


Tom Lane wrote:
> Patric de Waha <lists@p-dw.com> writes:
>   
>>    Postgres is running on a dedicated server  P4 DualCore, 4 Gig Ram.
>>     
>
> When you don't even mention your disk hardware, that's a bad sign.
> In a database server the disk is usually more important than the CPU.
>
>   
>>    Why do long readers influence the rest of the transactions in such a 
>> heavy way?
>>    Any configuration changes which can help here?
>>    Is it a disc-IO bottleneck thing?
>>     
>
> Very possibly.  Have you spent any time watching "vmstat 1" output
> to get a sense of whether your I/O is saturated?
>
>   
>> WAL files are located on another disc than the dbase itself.
>>     
>
> That's good, but it only relates to update performance not SELECT
> performance.
>
>   
>> effective_cache_size = 5000     
>>     
>
> That's way too small for a 4G machine.  You could probably stand to
> boost maintenance_work_mem too.  However, neither of these have any
> immediate relationship to your problem.
>
> 			regards, tom lane
>