From: Mark Kirkwood <markir@paradise.net.nz>
To: Laszlo Nagy <gandalf@designaproduct.biz>
Cc: pgsql-performance@postgresql.org
Subject: Re: Poor performance on seq scan
Date: Wed, 13 Sep 2006 12:14:52 +1200
Message-ID: <45074D7C.8000907@paradise.net.nz> (raw)
In-Reply-To: <45072B61.3080506@designaproduct.biz>
References: <450684EC.4090304@designaproduct.biz>
<17034.1158079926@sss.pgh.pa.us>
<45072130.6020508@modgraph-usa.com>
<45072B61.3080506@designaproduct.biz>
Laszlo Nagy wrote:
> Craig A. James wrote:
>>
>> There IS a bug for SATA disk drives in some versions of the Linux
>> kernel. On a lark I ran some of the I/O tests in this thread, and
>> much to my surprise discovered my write speed was 6 MB/sec ... ouch!
>> On an identical machine, different kernel, the write speed was 54 MB/sec.
> My disks are running in SATA150 mode. Whatever it means.
>
> I'm using FreeBSD, and not just because it dynamically alters the
> priority of long running processes. :-)
>
I dunno if this has been suggested, but try changing the sysctl
vfs.read_max. The default is 8 and results in horrible RAID performance
(having said that, not sure if RAID1 is effected, only striped RAID
levels...), anyway try 16 or 32 and see if you seq IO rate improves at
all (tho the underlying problem does look like a poor SATA
chipset/driver combination).
I also found that building your ufs2 filesystems with 32K blocks and 4K
fragments improved sequential performance considerably (even for 8K reads).
Cheers
Mark
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Reply to all the recipients using the --to and --cc options:
reply via email
To: pgsql-performance@postgresql.org
Cc: markir@paradise.net.nz, gandalf@designaproduct.biz
Subject: Re: Poor performance on seq scan
In-Reply-To: <45074D7C.8000907@paradise.net.nz>
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox