We have a two node cluster (2012 R2), and since 6 weeks we have the problem that live migration does no longer work.
Every time one of the nodes goes offline, the live-migration get stuck due to the virtual machine beeing stuck in "stopping" state.
Only a reboot of the node allows to the vm to recover to an working state. (The other node cannot shutdown, because it cannot migrate all vms sitting on it.
This happens no matter it its a manual invoked live-migration or live-migration due to shutdown / maintenance mode.
I managed to workaround the problem by setting (Get-Cluster).DrainOnShutdown to 0 - Then the cluster performs a quick migration in case of shutdown - which works.
However, once the node is back online it tries to move the vms back - using live migration - and gets stuck again. (It doesn't matter which node is acting as source, nor does the VM-Instance make any difference)
I read different topics about this issue, but none of the found solutions finally worked:
- I disabled VMQ for the nics of the heartbeat-network
- I installed an update for 2012 r2 on the nodes which states to "fix" a similiar issue: https://support.microsoft.com/en-us/kb/2958264
- I completely disabled the Hearbeat-Network and tried all the other networks, which leads to the same issue.
- I removed one node from the cluster, and added it again. Afterwards I removed the other node from the cluster and added it again.
In another post i read, that somebody encountered this issue due to some MPIO-Configuration-Problems (ISCSI-Storage) - However the MPIO seems to work as expected on both nodes.