DomainKey-Signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma;
	h=mime-version:sender:in-reply-to:references:date
	:x-google-sender-auth:message-id:subject:from:to:cc:content-type
	:content-transfer-encoding;
	b=Yy6Zg5Jk8Q8oyjfLcViLdFYa9fWoxgGb0rYFSMAeYYOAsBnal9M/Up3Wa7snBGCU0W
	SUk7YmNCZLimkddzcHmp6fiVKahJ1OFY+DVQpQE6pDw4iUC/OcMdpc8Za3FNBSIIUw9T
	ISdjgyCIrpRQktcNARrMDwZSxdIYHxLXHAaGc=
MIME-Version: 1.0
In-Reply-To: <49C3FABF.3000305@dunslane.net>
References: <7d10d2df0903111227g2e40e742g5992cba5158dd4d2@mail.gmail.com>
	<1d4e0c10903111659x6c96e6cau33ee121ee29a48a@mail.gmail.com>
	<29255.1236822351@sss.pgh.pa.us>
	<7d10d2df0903120638s3908aadbm5f285220fc738843@mail.gmail.com>
	<7d10d2df0903201247r7ff71d1bsf655176a86ca5be4@mail.gmail.com>
	<49C3FABF.3000305@dunslane.net>
Date: Fri, 20 Mar 2009 22:15:01 +0000
Message-ID: <7d10d2df0903201515l7e443adfjabb9d8399fc19059@mail.gmail.com>
Subject: Re: Full statement logging problematic on larger machines?
From: Frank Joerdens <frank@joerdens.de>
To: Andrew Dunstan <andrew@dunslane.net>
Cc: Tom Lane <tgl@sss.pgh.pa.us>, Guillaume Smet <guillaume.smet@gmail.com>,
	pgsql-performance@postgresql.org, Nic Ferrier <nic@woome.com>,
	Avleen Vig <avleen@woome.com>, Mike Rogers <mike.rogers@woome.com>
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit

On Fri, Mar 20, 2009 at 8:21 PM, Andrew Dunstan <andrew@dunslane.net> wrote:
>
>
> Frank Joerdens wrote:
>>
>> On Thu, Mar 12, 2009 at 1:38 PM, Frank Joerdens <frank@joerdens.de> wrote:
>>
>>>
>>> On Thu, Mar 12, 2009 at 1:45 AM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>>> [...]
>>>
>>>>
>>>> You could try changing _IOLBF
>>>> to _IOFBF near the head of postmaster/syslogger.c and see if that helps.
[...]
>> Can you explain again what the extent of multiplexed messages I'll
>> have to expect is? What exactly is the substance of the tradeoff?
>> Occasionally the server will write the same line twice? Don't really
>> follow why ...
[...]
> I don't believe changing this will result in any multiplexing. The
> multiplexing problem was solved in 8.3 by the introduction of the chunking
> protocol between backends and the syslogger, and changing the output
> buffering of the syslogger should not cause a multiplexing problem, since
> it's a single process.

Hum, we're still on 8.2 - last attempt to upgrade before xmas was
unsuccessful; we had to roll back due to not fully understood
performance issues. We think we nailed the root cause(s) those though
and will make another better planned effort to upgrade before March is
out.

Oh well, maybe this all means we shouldn't try to get this running on
8.2 and just tackle the issue again after the upgrade ...

Cheers,

Frank