Virtual machine monitoring uses VMware Tools to send heartbeats to the host its running on. If a heartbeat is not received within the configured threshold, then the VM is deemed as unresponsive/hung and then subsequently restarted. The threshold/sensitivity for heartbeating is configurable:

• High: Checks for a heartbeat every 30 seconds and reset every 1 hour
• Medium: Checks for a heartbeat every 60 seconds and reset every 24 hour 
• Low: Checks for a heartbeat every 120 seconds and reset every 7 days 

In order to avoid repetitive restarts and false positives, for any level of sensitivity three resets can be initiated per VM in a given reset period. Another key aspect is while heartbeats are not received, the network and storage activity of the VM is validated prior to initiating a reset. If any storage or network activity has been observed in the last 120 seconds, the VM will not be reset. This is done to ensure that the issue is not specific to VMware tools but the actual guest OS.

While the previous predefined configurations can be chosen from, a user is also at liberty to set his own criterion for VM monitoring. The same logic applies to application monitoring, however specific SDKs need to be configured to monitor application heartbeats instead of VMware tool heartbeats.