When you create a Windows Server 2008 Failover Cluster, with 3 nodes or more, and start running virtual machines on those nodes, what happens when one of those nodes fails?

Well, as you’d expect, the VMs on the now-down-physical-node, reboot onto another available node in the cluster (providing you haven’t been silly and used all the resource already on the other nodes!).  For me, the question is, is there any logic behind where those VMs restart if you have more than 3 nodes still remaining?

This KB article goes some way to answering that question: “Failover behaviour on clusters of three or more nodes”.

There are 4 possible scenarios:

  1. There is a node or resource failure and the Preferred Owner List is set.
  2. There is a node or resource failure and the Preferred Owner List is not set.
  3. Administrator manually moves group to "Best Possible" and the Preferred Owner List is set.
  4. Administrator manually moves group to "Best Possible" and the Preferred Owner List is not set

For those of you who haven’t configured clustering in 2008, basically, you can right click the properties of a VM, and choose a preferred owner (node) for which you want this VM to primarily reside.  Should that node fail, the VM won’t sulk and say “you’ve taken away my preferred node so I won’t restart anywhere else” – it’ll simply restart on another available node, however, you can also specify failback, which means when the preferred node does come back online, the VM will migrate back to it’s favourite home.

The KB document details each of the 4 scenarios listed above.  It’s not going to be to everyone’s taste, but some people may find it useful.

Here’s the link again: “Failover behaviour on clusters of three or more nodes