If a z/OS® image fails, the DFSMShsm host on that system also fails, but all of the recall requests that originated on that host remain intact on the CRQ. The coupling facility notifies the remaining connected hosts of the failure. The requests that originated on the failed host remain eligible for selection. In-process requests on the failed host remain on the queue and are made available for other hosts to restart them.
If the failing host is processing a request from ML2 tape, then all requests from that tape cannot be selected until the failing host is restarted. The tape is marked as "in-use" by the failing host, and that indication will not be reset until the failing host restarts. If the falling host is not restarted for some time, enter the LIST HOST(hostid) RESET command, where hostid is the DFSMShsm host id of the host that has failed.