From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mx2.redhat.com (mx2.redhat.com [10.255.15.25]) by int-mx2.corp.redhat.com (8.13.1/8.13.1) with ESMTP id kBG1mOXl025822 for ; Fri, 15 Dec 2006 20:48:25 -0500 Received: from mail.cnr.colostate.edu (clavin.cnr.colostate.edu [129.82.112.67]) by mx2.redhat.com (8.12.11.20060308/8.12.11) with ESMTP id kBG1mOo2011655 for ; Fri, 15 Dec 2006 20:48:24 -0500 Received: from nrelsmtp.nrel.colostate.edu (newman [129.82.112.30]) by mail.cnr.colostate.edu (8.13.7/8.13.7) with ESMTP id kBG1mHv8013767 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=FAIL) for ; Fri, 15 Dec 2006 18:48:17 -0700 Received: from [129.82.104.126] (donbot [129.82.104.126]) by nrelsmtp.nrel.colostate.edu (8.13.7/8.13.7) with ESMTP id kBG1mF3D013762 for ; Fri, 15 Dec 2006 18:48:16 -0700 Message-ID: <4583505F.1050605@nrel.colostate.edu> Date: Fri, 15 Dec 2006 18:48:15 -0700 From: "Ty! Boyack" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Subject: [linux-lvm] Control of which devices are used for striping (or mirroring) 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" To: LVM general discussion and development Folks, Is there anyway to control which devices a logical volume gets striped across? The reason for the question is that we have a situation with multiple raid arrays that can be expanded. but present new luns when they are expanded. The first array is partially populated and gives us LUN 0. The second array is also partially populated and also gives us LUN 0. We put both of those luns (as PVs) in a volume group, and then build our LVs striped across both LUNS for performance. When we add equal disks to both arrays, we get a new LUN 1 on each array. I would expect to add these new volums into the save VG. If I make a new LV saying --stripe 2, I don't want it to end up striping across LUN 0 and LUN 1 on the same array, since that will only hurt performance. Naturally, the same issue occurs with mirroring. If I mirror on LUN0 and LUN1 of the same array, and that array goes bad, then the mirror isn't doing a whole lot of good ;-) Even better than controlling which devices get mapped at lvcreate time, I'd love it if there was a way to create a "grouping" of PVs that LVM knew were really the same device, and therefore provide no benefit to striping or mirroring. Any way to do that? I'm certainly open to suggestions about striping or mirroring outside of LVM, but this will be used in a clustered environment, so I'd like to keep as much as possible in (C)LVM or find cluster-safe tools to work with outside of LVM. Thanks, -Ty! -- -===========================- Ty! Boyack NREL Unix Network Manager ty@nrel.colostate.edu (970) 491-1186 -===========================-