Received: from mx1.redhat.com (ext-mx09.extmail.prod.ext.phx2.redhat.com [10.5.110.13]) by int-mx05.intmail.prod.int.phx2.redhat.com (8.13.8/8.13.8) with ESMTP id o435cx7T014425 for ; Mon, 3 May 2010 01:38:59 -0400 Received: from maude.comedia.it (maude.comedia.it [77.93.254.181]) by mx1.redhat.com (8.13.8/8.13.8) with ESMTP id o435cmpI016044 for ; Mon, 3 May 2010 01:38:49 -0400 Received: from localhost (localhost [127.0.0.1]) by maude.comedia.it (Postfix) with ESMTP id A080B86F28 for ; Mon, 3 May 2010 07:38:47 +0200 (CEST) Received: from maude.comedia.it ([127.0.0.1]) by localhost (maude.comedia.it [127.0.0.1]) (amavisd-new, port 10025) with LMTP id XBCgaApjgMXj for ; Mon, 3 May 2010 07:38:42 +0200 (CEST) Date: Mon, 3 May 2010 07:38:42 +0200 From: Luca Berra Message-ID: <20100503053842.GA25863@maude.comedia.it> References: <4BDDE17F.7090300@strike.wu.ac.at> MIME-Version: 1.0 Content-Disposition: inline In-Reply-To: <4BDDE17F.7090300@strike.wu.ac.at> Subject: Re: [linux-lvm] logical volumes and raid alignment Reply-To: LVM general discussion and development List-Id: LVM general discussion and development List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , List-Id: Content-Type: text/plain; charset="us-ascii"; format="flowed" Content-Transfer-Encoding: 7bit To: linux-lvm@redhat.com On Sun, May 02, 2010 at 10:33:03PM +0200, Alexander 'Leo' Bergolth wrote: >Hi! > >I am using LVM on a hardware RAID 5 and I'd like to tune the alignment >of lvm extents to match the raid stripe size. > >My raid chunk size is 64k and I am using 5 data-bearing discs (raid 5 >with 6 discs), so the raid stripe size should be 64k * 5 = 320k. > >I have already managed to properly align the partition to a multiple of >this value and I am able to align the beginning of the first extent by >specifying --dataalignment 320k to pvcreate. > >But unfortunately the --physicalextentsize must be a power of 2, so I >cannot use a multiple of 320k as the physical extent size. > >What are your experiences with tuning and properly aligning lvm + raid? >Does anyone know a workaround? I had a similar issue, and best thing i came up with was creating volume group with PE Size == chunk_size, and create/extend all logical volumes to a multiple of data_disks PE. L. -- Luca Berra -- bluca@comedia.it Communication Media & Services S.r.l. /"\ \ / ASCII RIBBON CAMPAIGN X AGAINST HTML MAIL / \