[Pacemaker] migration fix for ocf:heartbeat:Xen

[Dejan Muhamedagic]

On Tue, Aug 23, 2011 at 03:54:35PM +0000, Daugherity, Andrew W wrote:
> > Message: 7
> > Date: Thu, 11 Aug 2011 21:07:00 +0000
> > From: "Daugherity, Andrew W" <adaugherity at tamu.edu>
> > To: "pacemaker at oss.clusterlabs.org" <pacemaker at oss.clusterlabs.org>
> > Subject: [Pacemaker] migration fix for ocf:heartbeat:Xen
> > Message-ID: <93B5E618-AD19-4993-8066-CB4F8E4EF322 at tamu.edu>
> > Content-Type: text/plain; charset="us-ascii"
> > 
> > I have discovered that sometimes when migrating a VM, the migration itself will succeed, but the migrate_from call on the target node will fail, as apparently the status hasn't settled down yet.  This is more likely to happen when stopping pacemaker on a node, causing all its VMs to migrate away.  Migration succeeds, but then (sometimes) the status call in migrate_from fails, and the VM is unnecessarily stopped and started.  Note that it is NOT a timeout problem, as the migrate_from operation (which only checks status) takes less than a second.
> > 
> > I noticed the VirtualDomain RA does a loop rather than just checking the status once as the Xen RA does, so I patched a similar thing into the Xen RA, and that solved my problem.
> (patch/logs snipped)
> 
> 
> No comments?  What does it take to get this patch accepted?  I'd much rather use the mainline version than have to reapply my patch after every HAE update.  I guess I could open an SR with Novell but this is ultimately an upstream issue.

Indeed it is an upstream issue, though it would get from Novell
upstream as well. It's just that sometimes upstream is out of
free cycles, at least temporarily. Sorry for that.

Cheers,

Dejan

> Andrew Daugherity
> Systems Analyst
> Division of Research, Texas A&M University
> adaugherity at tamu.edu
> 
> 
> _______________________________________________
> 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://developerbugs.linux-foundation.org/enter_bug.cgi?product=Pacemaker