Message-ID: <4D357FB6.60005@2ndquadrant.com>
Date: Tue, 18 Jan 2011 06:55:34 -0500
From: Greg Smith <greg@2ndquadrant.com>
User-Agent: Thunderbird 2.0.0.24 (X11/20100317)
MIME-Version: 1.0
To: Josh Berkus <josh@agliodbs.com>
CC: Tom Lane <tgl@sss.pgh.pa.us>, Robert Haas <robertmhaas@gmail.com>,
	Magnus Hagander <magnus@hagander.net>, pgsql-hackers@postgresql.org
Subject: Re: auto-sizing wal_buffers
References: <AANLkTik3mZ3sTfbHQW-1B44NRMkUzRsMnPb8g-HtttVv@mail.gmail.com>
	<AANLkTimQ3+osgAEH4meS97FmLN-4evSTS2W6fctov=jY@mail.gmail.com>
	<AANLkTinhU9EEPGM29S+KH=qK4t-mLhBBNAaLCGQSx-4y@mail.gmail.com>
	<20767.1294964709@sss.pgh.pa.us> <4D314406.3000703@2ndquadrant.com>
	<4D3220E3.20804@agliodbs.com>
In-Reply-To: <4D3220E3.20804@agliodbs.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit

Josh Berkus wrote:
> I think we can be more specific on that last sentence; is there even any
> *theoretical* benefit to settings above 16MB, the size of a WAL segment?
>  Certainly there have been no test results to show any.
>   

There was the set Marti just reminded about.  The old wording suggested 
big enough to fix a single transaction was big enough, and that let to 
many people being confused and setting this parameter way too low.  
Since it's possible I'm wrong about 16MB being the upper limit, I didn't 
want the wording to specifically rule out people testing that size to 
see what happens.  We believe  there's never any advantage due to the 
forced wal segment switch, but having test results to the contrary 
floating around keeps me from being too aggressive in how the wording 
there goes.

-- 
Greg Smith   2ndQuadrant US    greg@2ndQuadrant.com   Baltimore, MD
PostgreSQL Training, Services, and 24x7 Support  www.2ndQuadrant.us
"PostgreSQL 9.0 High Performance": http://www.2ndQuadrant.com/books