This is a two part question. First, I'm wondering how an ESXi host decides the order in which vms are shutdown when the "Virtual Machine Startup/Shutdown" settings are not configured? If I shutdown a host with active vms, I know they will be shutdown before the host goes off, but is the order random?
Second, according to the documentation the "Default shutdown delay" is "the maximum time the ESX host will wait for a shutdown command to complete. If the Virtual Machine does not shutdown within the delay time specified a power off is issued to the Virtual Machine." So this will only come into play if a vm locks up during shutdown or is just slow to shutdown, correct?
I have been testing the process here http://www.techhead.co.uk/how-to-automatically-shut-down-vmware-esxi-gracefully-during-power-failure... to shutdown a free ESXi server using APC PCNS. I've found the UPS does not get the shutdown command from the Windows vm with PCNS installed unless I crank the "Default shutdown delay" setting up to 900s. With the default setting of 120s what I think is happening is the host shuts the PCNS vm down before it can send the UPS the shutdown command. This doesn't really make sense though if the "Default shutdown delay" only comes into play when a vm locks up.
Can anyone else make sense of this?
And once I get this working on an unlicensed ESXi host I want to set it up on our licensed HA/DRS cluster (ie part one of my question). I know that the "proper" way to use PCNS on licensed hosts is to use a vMA but I've been down that path before and had a bad experience so if I can use a Windows vm to accomplish the same thing that's what I'll do
Regards,
Jeff
