Hyper-V Program Manager
A Microsoft support engineer recently contacted me with an interesting problem. They were working with a customer who had an application that was having problems. They had been trying to diagnose the problem – and had gotten to the point where they had decided that they wanted to get a crash dump from the virtual machine and analyze it.
The problem they were then hitting was that the virtual machine was clustered, and whenever they would try to gather a crash dump, Windows Failover Clustering would detect that the guest operating system had crashed and would restart the virtual machine on another node in the cluster.
Normally, this is exactly what you want to happen. But in this case it was a problem.
Fortunately, there is a relatively simple way to disable this specific functionality. What you need to do is:
You should be careful about changing this setting though. As once you do, clustering will no longer respond to actual guest operating crashes either.