VMs powered down after datastore connectivity loss

[Newb449]

Apologies for the newbie question.

I have 2 ESXi4 hosts connected to an iSCSI SAN via a switch. I have vSphere server and HA but not DRS. The switch was rebooted unexpectedly, which caused a break between the hosts and the storage where the VMs reside. I expected the VMs to crash but it appears they powered off in an orderly fashion.

After the switch reboot, the VMs powered up automatically (because this was set to do so in my vSphere client)

So was the powering off a function of the start up/shutdown settings in my vSphere client? I didn’t see anything in the logs pertaining to HA and I didn’t think this was a function of HA.

[AndreTheGiant · Accepted solution]

Check VMware HA properties (in cluster settings).

You will see that isolation response is set to shutdown the VM.

Isolation is when a host see that he cannot ping nobody.

If you restart the switch all hosts get in isolation mode.

You can change the timeout with HA advanced settings:

http://kb.vmware.com/kb/1006421

Andre

[Newb449]

So this is a HA feature! I thought host isolation was if the host couldn't reach the VM via its TCP I/P NIC. So if I set this to leave powered on the VMs would have remained running (although the VMs would have been unreachable since the same switch handles the VM IP network.

[AndreTheGiant]

So if I set this to leave powered on the VMs would have remained running

Yes, and is a good setting if you have a single physical switch.

Andre