VM HA (VM Monitoring - Reaction to Datastore failure)

[Mouhamad]

Hello there,

2 ESXi hosts connected to a shared storage. We have HA enabled with default settings and "VM Monitoring = Disabled".

In case one of the ESXi hosts running VM1 lost access to Datastore1 that hosts VM1, does HA kick in and reboot that machine on the other host?

If not, will it do when VM monitoring is enabled? and if we have more than 10 hosts, how does it know on which the relative datastore is still active, I mean on which host?

Regards,

[Cato Grace]

VM monitoring will restart a VM if HA stops detecting VMtools heartbeats as well as network and storage IO from the VM.

If a host loses access to a datastore as far as HA is concerned it depends on what version of vSphere is in use and how that access is lost (APD or PDL - see details here VMware KB: Permanent Device Loss (PDL) and All-Paths-Down (APD) in vSphere 5.x ).  If running vSphere 6, HA now has a feature (VM Component Protection) that will restart all VMs impacted by an APD or PDL on a non-impacted host. If using prior versions of vSphere even with VM monitoring enabled user involvement will be required as VM monitoring only restarts the VM (not on a new host) (vSphere 5.1 Storage Enhancements – Part 4: All Paths Down (APD) | CormacHogan.com).

Does this answer your questions?