Failure testing - what am I missing?

[mvalpreda]

I set up a 2 node Hyper-V cluster in a lab with nested hypervisors. Live migration works great, now I wanted to do some failure testing by simulating a node failure. Hard powered off the host where the VM was running. The VM shows as 'Unmonitored' and the VM is non-responsive. After about 3-5 minutes, the VM is back on the host that is up, but the VM has been reset/power cycled. In watching this https://www.youtube.com/watch?v=a0ynM4tRnhg video on YouTube, the presenter's VM just keeps going.....that is what I am looking to do.

Did I miss something in the VM config? Something else I need to configure on the Hyper-V host?

[Boris (staff)]

At 14:52 of that footage, the speaker confirms the VM is running, but after having rebooted (uptime for that VM is shown as 48 seconds only).

In fact, you are not missing anything. With Windows Failover Cluster, you have no fault tolerance (like what VMware offers), it's just high availability, which in turn means your VMs will be brought up on the partner host whenever their host fails.

[mvalpreda]

Thanks for that. I was setting it up on my side, watching the video, etc. Must have missed that part! Looked up and saw the VM running and thought I was doing something wrong.

Assuming there is nothing in the Hyper-V world, regardless of number of hosts, that would keep VMs up if a node went down hard?

[Boris (staff)]

To the best of my knowledge, in the Hyper-V world there is nothing offering fault tolerance functionality at the level of VMs. There are some third party solutions though, which you could explore.