From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mx1.redhat.com (mx1.redhat.com [172.16.48.31]) by int-mx1.corp.redhat.com (8.11.6/8.11.6) with SMTP id i1H0X8i28634 for ; Mon, 16 Feb 2004 19:33:08 -0500 Received: from office.labsysgrp.com (wsip-68-14-253-125.ph.ph.cox.net [68.14.253.125]) by mx1.redhat.com (8.11.6/8.11.6) with SMTP id i1H0X8b28007 for ; Mon, 16 Feb 2004 19:33:08 -0500 Message-ID: <40316114.30207@backtobasicsmgmt.com> From: "Kevin P. Fleming" MIME-Version: 1.0 Subject: Re: [linux-lvm] Re: IO scheduler, queue depth, nr_requests References: <20040216131609.GA21974@cistron.nl> <20040216133047.GA9330@suse.de> In-Reply-To: <20040216133047.GA9330@suse.de> Content-Transfer-Encoding: 7bit Sender: linux-lvm-admin@redhat.com Errors-To: linux-lvm-admin@redhat.com Reply-To: linux-lvm@redhat.com List-Help: List-Post: List-Subscribe: , List-Id: LVM general discussion and development List-Unsubscribe: , List-Archive: Date: Mon Feb 16 19:32:00 2004 List-Id: Content-Type: text/plain; charset="us-ascii"; format="flowed" To: linux-lvm@redhat.com Cc: Miquel van Smoorenburg Jens Axboe wrote: > On Mon, Feb 16 2004, Miquel van Smoorenburg wrote: > >>Hello, >> >> as you might have seen from the linux-kernel mailinglist >>I have been testing for months now with a fileserver set up to >>use XFS over LVM2 on a 3ware RAID5 controller. >> I too have been fighting lower-than-expected performance with this identical combination, other than using a P4 2.8GHz instead of 3.0GHz, and 1GB RAM instead of 2GB. My 3ware 8506-8 has six Seagate Barracuda SATA drives behind it, which are very fast on their own :-) Try as I might, with the 2.6 kernel I have not been able to generate a bonnie++ or iozone run that provided performance better than a single disk, even with the 3ware card in JBOD, RAID-0 or RAID-1 mode. I had been conversing with 3ware about it, but to no real effect given than they don't support 2.6 yet. >>By fiddling about today I just found that changing >>/sys/block/sda/queue/nr_requests from 128 to something above >>the queue depth of the 3ware controller (256 doesn't work, >>384 and up do) also fixes the problem. I will try this as well this evening and report my results.