A vSphere master is considered to be in a failed state when it is unable to communicate with vCenter Server. This condition is not always reported correctly by vCenter Server, and at times, it can be generated falsely. The actual condition can be caused by a vSphere HA master when it is not able to reach to its heartbeat datastores. It is resolved automatically if a vSphere host is in a temporary failed state.
The possible cause could be a failure in accessing the heartbeat datastores. The heartbeat datastores' reachability can also be caused by network errors. You can apply the network and storage troubleshooting skills you have learned in previous chapters to resolve vSphere host in a failed state.