On Tue, 3 Aug 2021, Nir Soffer wrote:
On Tue, Aug 3, 2021 at 5:51 PM Sketch <ovirt(a)rednsx.org>
wrote:
>
> On the 4.3 cluster, migration works fine with any storage backend. On
> 4.4, migration works against gluster or NFS, but fails when the VM is
> hosted on POSIX cephfs.
What do you mean by "fails"?
What is the failing operation (move disk when vm is running or not?)
and how does it fail?
Sorry, I guess I didn't explain the issue well enough. Moving disks
between ceph and gluster works fine, even while the VM is running.
> It appears that the VM fails to start on the new host, but
it's not
> obvious why from the logs. Can anyone shed some light or suggest further
> debugging?
What doesn't work is live migration of running VMs between hosts running
4.4.7 (or 4.4.6 before I updated) when their disks are on ceph. It
appears that vdsm attempts to launch the VM on the destination host, and
it either fails to start or dies right after starting (not entirely clear
from the logs). Then the running VM gets paused due to a storage error.