Question : Hyper-V machines in Failover Cluster: can Live Migrate, but cannot Move or Quick Move between cluster nodes.

Hi,

I've just deployed a Failover Cluster running 4 Hyper-V machines on two identical machines Windows Server 2008 R2 x64.

Those machines are running perfectly and I'm able to Live Migrate them between the two cluster nodes, with no problems at all.

The Live Migration is great, but it supposed to be used for planned maintenance only. It is *not* suitable for "unplanned" situations, in which, for example, one of the cluster nodes crashes.

So, I'm trying to simulate an unplanned situation, by doing a Move or Quick Move from one cluster to the other. However, when I try to do it, the Hyper-V machine saves its state, but usually it won't came back online on the other cluster node, and its state is flagged as failed.

To put it back online, I have to manually delete the saved state and then start it again. You have to agree that in an "unplanned situation", I want all that to happen automatically. That's what Failover is used for, right?

So, my questions is: what could be cause the Live Migrate to work, but the Move or Quick Move to fail on all Hyper-V machines?

When I look on the Event details for the failure, the error message I get is:

'Virtual Machine Web01' failed to start.

'Web01' failed to restore. (Virtual machine ID 7DA2F229-7C02-4FA6-905A-C05305A04FEF)

'Web01' is unable to restore RAM contents at address 0. (Virtual machine ID 7DA2F229-7C02-4FA6-905A-C05305A04FEF)

Since this is an urgent matter, I'm assigning 500 points to this.

Thanks in advance for your help.

Best regards,

Helder

Answer : Hyper-V machines in Failover Cluster: can Live Migrate, but cannot Move or Quick Move between cluster nodes.

To be highly available, each volume created on a LUN that is to be connected to a VM is attached to _all_ nodes in the cluster.

We use Node 1 to configure the volumes, format them NTFS, and then set them OFFLINE. Once all of the volumes have been set up and set to offline on Node 1, then refresh the storage view on all other nodes and ignore all warning messages.

You will then have a bunch of volumes available but set to offline. When running the cluster validation wizard, it will scan those volumes and indicate that they are properly configured to be highly available to _all nodes_ in the cluster.

After creating the cluster you enable CSV in the root of the FCM and then enable all volumes as CSVs.

Rule of thumb:
 - # of volumes = # of VMs + 1 (2-3 nodes)
     - The +1 is the volume used for the shared configuration files for _all_ VMs. So, its size needs to be VM RAM #GB * #VMs + 1GB. So, 4 VMs max with 8GB each would mean a shared volume of about 35GB dedicated to this purpose.
 - # of volumes = # of VMS + 2 (3+ nodes)
     - Above rule applies with the extra volume being only 512MB or 1GB for Quorum data that will be needed.

Philip