Cluster responses to failures
As the name indicates, the purpose of Microsoft Failover Clustering is to provide failover capabilities. Whenever the cluster service on a node is not reachable within a five second window, a failover event is triggered. There are two ways for a node to become unavailable. The first is a simple crash. This could be a blue-screen error or a power outage. The second is network isolation. In this state, the host is still up, but it is not available.
Another failover trigger is the failure of a monitored service, as described in Chapter 10, Maintaining and Monitoring a Hyper-V Server Cluster. In this case, the cluster's first response is to try to restart the guest on the same node. Its second response is to move it to another node, much as it would if that guest's host became isolated.
The way that the cluster responds to a failover will vary based on the condition and configuration of protected resources. The cluster nodes are constantly exchanging configuration...
