I want to make a setup, when workload is distributed between two sites are high availability mechanism protects from one site failure, restarting vms on another site.

Scenario I use:

1. Poweroff hosted engine host, and then host with highly-available vm

2. Engine comes back on another site (which is fine).

3. What I expect, engine finds, that there's a host not available with HA vms, and restart these vms on another host. But, I only see in logs, that engine reports, that it cannot connect to that host (java.net.NoRouteToHostException: No route to host).

Whel I poweroff just server with app vm, High Availability works correctly, performs fencing and restarts vm on another host.

I use release 4.1 from RedHat.

Has anyone faced similar issue? And, if yes, how could that be toubleshooted?

Thank you in advance,

Sergei.