unable to live migrate VM between cluster

[MarkjT]

Hi,

We are running Proxmox 4.4 (Understand this is an older version) which has 9 nodes running ceph. Everything has been working fine until we noticed today that we are unable to live migrate VM's. When we try to migrate, the task starts but produces an exit code 1 and exit code 255.

What i have been able to work out is if we power off a VM, it will migrate to another node and will start. After i have done this, the VM will then live migrate as normal.

Has anyone experienced this before, is there a known fix or cause for this issue ? There have been no changes made so a bit confused why i am now seeing this. It will be a bit of a pain to shut down VM's, migrate them while they are off and start them back up and then live migrate them back.

Any help would be appreciated.

Thanks.

 

[dcsapak]

is the node where you want to live migrate to a lower version? if yes, this is expected behaviour since live migration only works reliable from old -> new
also having the same version on all nodes helps with live migration

 

[MarkjT]

Thank you for the reply, yes, all the nodes are at the same version. All the nodes have been running for months with out any issues at all. We were able to live migrate with out any issues until we tried yesterday. Nothing has changed which is really odd. It is also odd that we discovered if you power down a vm, the migrate it to another node and start it, we are then able to live migrate the VM with errors - any thoughts why this combination would fix it ? There are not any real errors in the logs from what i can see which is quite puzzling ! Like i said, all has been working flawlessly up until we tried to live migrate one yesterday.

Any thoughts ?

Thank you !

 

[dcsapak]

Any thoughts ?

no not really besides some unknown bug in the old qemu version maybe? but since 4.x is EOL anyway and you should upgrade, i am not sure it make sense to put much time in debugging