On each node, the operating system will broadcast a distress message when it detects that the system is about to fail or is being shut down. Cluster resource services will respond to a distress message broadcast on the cluster communications network.
Examples of when a node would initiate a distress message include, but are not limited to, ending all subsystems or those in which cluster jobs or exit programs reside, when a delayed power down is requested from the Hardware Management Console (HMC), or when UPS battery is drained while operating on backup power.
It the failing node is a primary node, cluster resource services will initiate a failover. If the node is not the primary node, the node is ended and no longer participates in the cluster until user action is taken.