VM migration state on host failure

[sergei56]

Hello.
We have a cluster of 4 nodes 6.4-4.
Iscsi shared storage.
HA is configured for VM.
I simulate shutting down one node by quickly pulling the network cables out of it.
The VM automatically starts on the other host.
But after I go to it, I get windows notification about an unexpected shutdown.
The agent and drivers are installed on the machine.
Is this normal behavior?
It turns out that the VM crashes and everything that the user did is lost.
I just thought that the VM should remember the state and just start on another node.
The documentation is clearly not written about this.
Please tell me if it should be so.

 

[dietmar]

I simulate shutting down one node by quickly pulling the network cables out of it.
The VM automatically starts on the other host.
But after I go to it, I get windows notification about an unexpected shutdown.
The agent and drivers are installed on the machine.
Is this normal behavior?

Yes. The VM on the unreachable node gets "fenced" (killed) before we can start it on another node (To make sure only one node writes to the shared storage).

 

[sergei56]

Thanks for the quick response.