[ovirt-users] Several VM's stuck in Migration

[Andrew Wippler]

I was in the process of upgrading from 3.5.5 to 3.6.4 when it happened. I am sure there were version mismatches.

I think what generated this issue was that I edited the cluster the host belonged to while it was going down for maintenance. This caused the tasks tab to show multiple tasks of the host going down for maintenance. They all eventually went into en error state and the task kept repeating itself until I rebooted the host.

Andrew Wippler | Sr. Systems Administrator | Lancaster Baptist Church | Dr. Paul Chappell, Pastor

________________________________
From: Michal Skrivanek <michal.skrivanek at redhat.com>
Sent: Friday, April 01, 2016 1:00 AM
To: Andrew Wippler
Cc: users at ovirt.org
Subject: Re: [ovirt-users] Several VM's stuck in Migration


On 31 Mar 2016, at 19:48, Andrew Wippler <Andrew.Wippler at lancasterbaptist.org<mailto:Andrew.Wippler at lancasterbaptist.org>> wrote:

Mark,

Hopefully you have resolved your stuck VM issue by now (nearly two months later). I came across this email from a google search. I was having the same exact problem with VMs being stuck in migration mode. The trigger to the VM migration issue was that I was setting my Host into maintenance mode. It migrated half of the active VMs successfully and the other half were stuck.

Hi,
It would be great to get logs from the system, or at least describe the specifics (anything special? gluster? old version?)
otherwise no one will ever be able to help;)

Thanks,
michal


There were two things I had to do to fix all of my stuck VMs:
1. I shut down the host (long enough for oVirt to fence it and notice that it was down). I then powered it on and "confirmed host as rebooted." This reset the active VM count on the host and turned on my HA VMs on other hosts.
2. For the one VM that was still stuck after a host reboot, I had to login to postgres and set the status to 0 on that VM. This allowed me to shut down the migrating VM and start it back up. (I do not think I would have had to do this step if I did not alter the database entry for this VM before rebooting the host.)

For step 2, I used the methods described here: https://access.redhat.com/discussions/688043 and as seen below:
ssh ovirt-engine
su postgres
psql
\connect engine
SELECT vm_guid status FROM vm_static WHERE vm_name='XXXX';
UPDATE vm_dynamic SET status=0 WHERE vm_guid='xxxxxxxx';
UPDATE vm_dynamic SET run_on_vds = null  WHERE vm_guid='xxxxxxxx';
\q
(I was then able to shut down the VM)

--
Andrew Wippler <andrew.wippler at lancasterbaptist.org> | Sr. Systems Administrator | Lancaster Baptist Church

_______________________________________________
Users mailing list
Users at ovirt.org<mailto:Users at ovirt.org>
http://lists.ovirt.org/mailman/listinfo/users

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ovirt.org/pipermail/users/attachments/20160401/d1ca60da/attachment-0001.html>