A. While many of VMware's features in ESX and vSphere are fantastic additions to the virtual environment, some still have me scratching my head. One of those is VM Monitoring, and it's configured at the point you create an ESX cluster with the VMware HA feature turned on.
Figure: VM Monitoring
VM Monitoring is different than HA's host monitoring in that the item being watched is the individual virtual machine (VM) rather than the ESX host. With VM Monitoring enabled, VMs that miss a set of VMware Tools heartbeats will be automatically restarted by vSphere. The number and interval of heartbeats can be configured for the cluster and individual VMs. The figure above shows the wizard page where you'll find its configuration as you create a new cluster.
VM Monitoring, in this expert's opinion, can be a dangerous thing. Its heartbeats are sourced from the VMware Tools. Those tools have, at various points in the past, had their own problems, from the occasional need to upgrade to the rare (but complete) failure of the tools on one or more VMs.
When that happens and heartbeats can't be heard by vSphere, you're in for a reboot. This can happen even if the VM or its installed applications aren't experiencing a problem. Call me crazy, but I don't particularly like some external force deciding when my VMs need a reboot. I leave this feature turned off. My advice is that you do the same.
That said, I'm always open to alternative opinions. Do you have a counterexample, one that shows where VM Monitoring has saved the day? If so, comment below and let us know!
Catch up with @ConcentratdGreg on Twitter!