The vSphere HA agent on a host is in the Agent Unreachable state for a minute or more. User intervention might be required to resolve this situation.
vSphere HA reports that an agent is in the Agent Unreachable state when the agent for the host cannot be contacted by the master host or by vCenter Server. Consequently, vSphere HA is not able to monitor the virtual machines on the host and might not restart them after a failure.
A vSphere HA agent can be in the Agent Unreachable state for several reasons. This condition most often indicates that a networking problem is preventing vCenter Server or the master host from contacting the agent on the host, or that all hosts in the cluster have failed. This condition can also indicate the unlikely situation that vSphere HA was disabled and then re-enabled on the cluster while vCenter Server could not communicate with the vSphere HA agent on the host, or that the ESXi host agent on the host has failed, and the watchdog process was unable to restart it. In any of these cases, a failover event is not triggered when a host goes into the Unreachable state.
Determine if vCenter Server is reporting the host as not responding. If so, there is a networking problem, an ESXi host agent failure, or a total cluster failure. After the condition is resolved, vSphere HA should work correctly. If not, reconfigure vSphere HA on the host. Similarly, if vCenter Server reports the hosts are responding but a host's state is Agent Unreachable, reconfigure vSphere HA on that host.