I have an APD situation that I need help with - again!
If our SAN fails (testing here) the we see the following:
- On all hosts, misc.APDTimeout is set to 30 seconds
- On failing the SAN, the VM with vCenter running on it lost connection and then once the 30 second timeout was reached, shut down.
- The vCenter event log showed that an APD state was indeed logged. Once the timeout was reached, it fast-failed the VM IO as expected and then the VM was reset.
- As it was unable to find the config file it could not restart and so remained off.
This is the ideal behavior for our VMs - don't have time to explain why, and so I am trying to find out why the vCenter VM behaves this way when none of the other VMs do. We DO NOT want them to keep running in an APD state. Redundancy within our applications negates the reasons to keep them up.
<edit>
I've been looking at the Windows 2008 Server Task Manager and the ability to react to events. Is it possible to get the guest OS to receive events from the ESXi host? In addition, how could I send 2 snmp traps from esxi - one to the VMs and one to our event collection server? Curious about thse.
</edit>
I'd really appreciate some advice or insights on this.
