Found a workaround.
Changed the fence agent type from "ilo4" to "ipmilan" then added
"lanplus=1,power_wait=30" (without quotes) to options.
Now the host can be fenced successfully and all HA VMs in that host will
be restarted in another hosts.
Did a small experiment with power_wait
I'm running an oVirt hosted-engine environment on 3 hosts. To test VMs'
HA functionality, I shutdown host02's link, where one of my HA VMs is
running on, using this command:
2016-05-10 09:59:19 ICT [root@host02 ~]# ip link set bond0 down
Few seconds later, an attempt to fence host02 was
Mail list logo