From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mx1.redhat.com (ext-mx14.extmail.prod.ext.phx2.redhat.com [10.5.110.19]) by int-mx01.intmail.prod.int.phx2.redhat.com (8.13.8/8.13.8) with ESMTP id qBEIXCcZ000567 for ; Fri, 14 Dec 2012 13:33:12 -0500 Received: from mail.bmsi.com (www.bmsi.com [24.248.44.156]) by mx1.redhat.com (8.14.4/8.14.4) with ESMTP id qBEIXBwo002764 for ; Fri, 14 Dec 2012 13:33:11 -0500 Received: from sdg.bmsi.com (sdg.bmsi.com [192.168.9.34] (may be forged)) (authenticated bits=0) by mail.bmsi.com (8.14.3/8.14.3) with ESMTP id qBEIX9xC004362 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO) for ; Fri, 14 Dec 2012 13:33:10 -0500 Message-ID: <50CB70E5.6010002@bmsi.com> Date: Fri, 14 Dec 2012 13:33:09 -0500 From: Stuart D Gathman MIME-Version: 1.0 References: <50C9F6AF.50908@redhat.com> <50CA0492.1050906@redhat.com> In-Reply-To: <50CA0492.1050906@redhat.com> Content-Transfer-Encoding: quoted-printable Subject: Re: [linux-lvm] Convert striped lv to single pv on another disk? 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="iso-8859-1"; format="flowed" To: linux-lvm@redhat.com On 12/13/2012 11:38 AM, Bryn M. Reeves expounded in part: > On 13/12/12 16:11, Gianluca Cecchi wrote: >> I found in another thread that the resulting LV with pvmove remains a >> striped one on only one disk as far as its structure is concerned... >> with degraded performance... >> Have you an example of pvmove command options to test to execute based >> on my test? >> So that I can verify with my loop devices? > Yes, you're right. You will end up with both stripes still present but > on a single disk. > > The best I could do using pvmove was to move individual extents around > (!) to make the LV contiguous on-disk. Of course, this doesn't remove > the striping and inflates the number of segments since each pair of > striped extents effectively becomes a segment. > > The lvconvert command also still accepts a -i parameter but doesn't > do anything useful with it and doesn't report an error. > Could you create an asymmetrical mirror - where one leg is striped and=20 the other is linear? Then you can just sync the mirror and delete the=20 striped leg. I could even see operating with such a mirror - call it=20 RAID =EF=BF=BD. Reads would prefer the striped side, and the mirror is the= re in=20 case one of the stripes fails, while requiring only 1 additional drive=20 and degrading only write performance. RAID10 over the 3 drives is probably better, though.