PRIMECLUSTER forcibly stops a failed node by the shutdown facility and prevents the conflict of user resources.
If stopping other nodes forcibly, use the option hardware such as RCI or XSCF.
Therefore, if an error occurs in some machines, which the option hardware is not made redundant, the failed node cannot be forcibly stopped and the service may not be switched.
Even in this situation, by using the Remote Power Distribution Unit, the node's power and redundant power are cut and the node can be forcibly stopped for sure. Then, the cluster system is switched.
Using the Remote Power Distribution Unit is optional.
When using the Remote Power Distribution Unit, control it by the RPDU shutdown agent of the shutdown facility to check the status of the power supply to the node and cut the power.
