> When a device fails to successfully fence a node we try the next device > configured, or if no device is left we fall back to the current > procedure and wait until we got the lrm agent lock of the node. Do you think it is a good idea to merge both fencing mechanisms?