You want to make sure that the cluster is controlling the states of your
VM's to avoid erratic behavior caused by the VM config conflicting with the
cluster config. One of the major problems is with resources failing over and
coming online extremely slow.
You should set the VM start action to "Nothing" and the stop action to
"Save". When you make these changes make sure to refresh the VM config from
Don't shutdown VM guest from within the guest themselves. This will lead to
the VM simply failing-over and coming back online on the other node. When you
need to shutdown a host down, simply quick migrate to the surviving host. If
you need to shutdown the surviving host, then shut down the VM's from
failover clustering on the surviving node. Although, this is not required as
the host will take the guest down for you as part of its shutdown process.
When the host come back online the cluster service will bring the VM back
"Adin Sabic" wrote:
> When using clustering, I read that the VM's should not be configured to
> automatically start or shutdown as the clustering services will mess things
> up. I
> understand that if you are shutting a node down, the VM's should be migrated
> to the other node.
> What is the behaviour and how should things be configured when both nodes
> have to be shutdown at the same time (e.g. UPS shutdown event)? I need the
> nodes and all VM's to shutdown cleanly and automatically start on reboot.