[Pacemaker] LVM restarts after SLES upgrade

[Dejan Muhamedagic]

Hi,

On Thu, Apr 26, 2012 at 12:46:35PM +0200, Andreas Kurz wrote:
> On 04/25/2012 11:00 AM, Frank Meier wrote:
> > Am 24.04.2012 17:53, schrieb pacemaker-request at oss.clusterlabs.org:
> > 
> >> Message: 2
> >> Date: Tue, 24 Apr 2012 15:58:53 +0000
> >> From: "Daugherity, Andrew W" <adaugherity at tamu.edu>
> >> To: "<pacemaker at oss.clusterlabs.org>" <pacemaker at oss.clusterlabs.org>
> >> Subject: Re: [Pacemaker] LVM restarts after SLES upgrade
> >> Message-ID: <114AD516-3DA6-43E1-8D15-F5D9D3EAAAD1 at tamu.edu>
> >> Content-Type: text/plain; charset="us-ascii"
> >>
> >> On Apr 24, 2012, at 4:28 AM, <pacemaker-request at oss.clusterlabs.org>
> >>  <pacemaker-request at oss.clusterlabs.org> wrote:
> >>
> >>> Date: Tue, 24 Apr 2012 09:34:12 +0000
> >>> From: emmanuel segura <emi2fast at gmail.com>
> >>> Message-ID:
> >>>       <CAE7pJ3DGvkzMR1d9HNm76s8wtrAj0t1==31RnafHe-YEgTV06Q at mail.gmail.com>
> >>>
> >>> Hello Frank
> >>>
> >>> Maybe this it's not the probelem, but i see this constrain wrong from
> >>> my point of view
> >>> =============================================================
> >>> order o-Testclustervm inf: c-xen-vg-fs vm-clusterTest
> >>> order o-clvmglue-xenvgfs inf: c-clvm-glue c-xen-vg-fs
> >>> =============================================================
> >>> to be
> >>> =============================================================
> >>> order o-clvmglue-xenvgfs inf: c-clvm-glue c-xen-vg-fs
> >>> order o-Testclustervm inf: c-xen-vg-fs vm-clusterTest
> >>> =============================================================
> >>
> >> How is that any different?  Both sets of order constraints are identical, and look correct.  Changing the order you add them in makes no difference, as the rules are evaluated as a set, and the crm shell will reorder them in alphabetical (ASCIIbetical, actually) order anyway.
> >>
> >>
> >>> 2012/4/24, Frank Meier <frank.meier at hr-group.de>:
> >>>> Every time the vgdisplay -v TestXenVG is hanging(ca.2min)
> >>>>
> >>>> I see two of this peocesses:
> >>>> /bin/sh /usr/lib/ocf/resource.d//heartbeat/LVM monitor
> >>>> /bin/sh /usr/lib/ocf/resource.d//heartbeat/LVM monitor
> >>>> is this OK, or have we a race condition?
> >>
> >> Frank, I see you have multipath in your LVM config.  Have you tried it with multipath disabled?  I wonder if this isn't a pacemaker/corosync problem but rather a lower-level storage problem.  Still, whatever the cause, it doesn't fill me with confidence about upgrading to SLES 11 SP2... I guess it's time to bring up that test cluster I've been meaning to build.
> >>
> >> -Andrew
> >>
> > 
> > Hi,
> > 
> > yes, I've tested now without multipathd, but the problem exist furthermore.
> 
> You already found this thread?
> 
> http://lists.linux-ha.org/pipermail/linux-ha/2011-November/044267.html

A solution for that has been pushed upstream. It's not yet
available as a maintenance update to the SLE customers, I think,
but it would be good to try it out.

Thanks,

Dejan

> .... there was also another discussion I can't find atm regarding
> possible tunings like i/o scheduler and lvm filter changes.
> 
> Regards,
> Andreas
> 
> -- 
> Need help with Pacemaker?
> http://www.hastexo.com/now
> 
> 



> _______________________________________________
> Pacemaker mailing list: Pacemaker at oss.clusterlabs.org
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
> 
> Project Home: http://www.clusterlabs.org
> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
> Bugs: http://bugs.clusterlabs.org