when it comes to VM High Availability it is highly recommended to fence the error host.
This technically means that the error host is going to get shut down (in best case “powered off”) by something like ipmi. Is that right so far?
In the “fence_host.sh” script the following example is listed:
fence_ilo -a $FENCE_IP -l -p
The man page for fence_ilo says that the default action of this command is “reboot”.
The action can be set to “off” with the “-o” parameter:
fence_ilo -a $FENCE_IP -l -p -o
This is the way it should be used and a reboot is not the desired behavior here, right?
Thanks for a short feedback!