A: There are two steps required to get notified when metrics are no longer coming thru from a resource:

  • Create a metric that "aggregates and counts" an existing metric that you expect to be coming thru every minute
  • Create an alert where this new metric is "0"

Specifically, let's take a Windows VM resource.  We expect that a CPU Utilization metric is always coming thru, during every monitoring cycle (unlike for example Event Logs).  We want to be notified when CPU did not come thru, thus indicating a possible issue with the server or the agent on it

1. Define an aggregate metric over CPUTime:

2. Define an alert that compares this new metric to 0.  Consider adding a "Sustained timeout" to this alert to ensure that you receive notification only when metrics are not coming thru for a certain amount of time