[SOLVED] KVM Migration Fails after node upgrade to 3.3

[lweidig]

We recently decided to migrate our Proxmox servers from 3.2 to 3.3. As we have done in the past we used live migration and moved all the running servers off one of the devices, ran the apt-get update; apt-get dist-upgrade and then rebooted the machine. Started moving back the guests and had no issues with all of the containers. However, our KVM based machines give us the errors:

Oct 01 11:50:48 starting migration of VM 113 to node 'xxxx' (nnn.nnn.nnn.nnn)
Oct 01 11:50:48 copying disk images
Oct 01 11:50:48 starting VM 113 on remote node 'xxx'
Oct 01 11:50:53 starting ssh migration tunnel
Oct 01 11:50:54 starting online/live migration on localhost:60000
Oct 01 11:50:54 migrate_set_speed: 8589934592
Oct 01 11:50:54 migrate_set_downtime: 0.1
Oct 01 11:50:56 ERROR: online migrate failure - aborting
Oct 01 11:50:56 aborting phase 2 - cleanup resources
Oct 01 11:50:56 migrate_cancel
Oct 01 11:51:01 ERROR: migration finished with problems (duration 00:00:15)
TASK ERROR: migration problems

This is a Windows 2012 server that we have running. I did see the thread: ProxMox 3.2 to 3.3, impressions and a problem which I suspect is intertwined with the issue that we are seeing, but I did not want to post into that and clutter up the issue. That one seems more about losing licensing based on underlying hardware changes.

I need to get the machines migrated back to the upgraded server or at least be given the assurance they will restart if I shut them down, upgrade the server they are on, reboot and then try migration. Would prefer not to risk the servers not coming back up under this second option though and find a way to move them live.

I assume there are additional logs/information to provide, but not exactly sure what is all needed.

Thanks!

 

[kotakomputer]

For Windows Server after 2003, including Windows Server 2008 and 2012 I suggest you use OFFLINE migration. Shutdown the VM first.

 

[lweidig]

For Windows Server after 2003, including Windows Server 2008 and 2012 I suggest you use OFFLINE migration. Shutdown the VM first.

Thanks for the reply. So you have moved systems back over to a 3.3 system from a 3.2 system offline? We have really never had issues with online migration in the past, but I will give this a try. Will have to wait until early hours of the morning as the servers are critical and cannot be brought down during the day.

 

[lweidig]

Thanks for the reply. So you have moved systems back over to a 3.3 system from a 3.2 system offline? We have really never had issues with online migration in the past, but I will give this a try. Will have to wait until early hours of the morning as the servers are critical and cannot be brought down during the day.

I can confirm that this did work for me and now all nodes are running 3.3! I also have to say that the HTML5 console is AWESOME - great job!