Hello,

in RHCS we have the fencing concept for targets very similar to the oVirt ones: avoid data corruption and also fast react to problematic hosts situations.

The implementation is quite similar to the oVirt one, with several fencing agents sometimes common, like fence_ipmilan.

In RHCS documentation there is a chapter describing how to configure hosts so that they don't react to power button pressure. This guarantees that failover is as fast as possible, and also that the same host to be fenced, could create more damages if it reacts and begins the shutdown procedure instead of simply powering off.

See:

https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/8/html/configuring_and_managing_high_availability_clusters/assembly_configuring-fencing-configuring-and-managing-high-availability-clusters#proc_configuring-acpi-for-fence-devices-configuring-fencing

It seems to me that oVirt / RHV lack this feature.

Inside oVirt NGN and also RHVH-H, the /etc/systemd/logind.conf file is not configured with an entry of type

HandlePowerKey=ignore

So in some cases of tests I'm doing, I see in its virtual console that the to-be-fenced host begins its OS shutdown flow when it detects the power button pressure. Typically after 2-3 seconds the system powers off and then on again, but I also saw about 10 seconds delays in one case.

I have opened a case (number 03002278) for my RHV products, but I would also like to get your comments here if I'm wrong with my considerations.

Gianluca