A processor leaves the XCS network when HIPLEX MSCF is terminated. HIPLEX MSCF is terminated when the MSCF subsystem is terminated in the course of the BS2000 shutdown procedure, or can be terminated explicitly by issuing the STOP-SUBSYSTEM MSCF,SUBSYSTEM-PARAMETER='FORCED=YES' command. When this command is executed, the entire HIPLEX MSCF functionality of the processor is cleared down and all imported shared pubsets are exported. The remaining (local) functionality of the processor is retained.
HIPLEX MSCF is terminated in a number of phases. In the first phase, the user tasks still using XCS functionality are terminated with CANCEL-JOB. In addition, UTM applications protected against CANCEL-JOB will be disposed to terminate by sending a corresponding shutdown announcement. Network exit is initiated at the latest with the expiration of the period defined by the MSCF configuration parameter USER-TERM-LIMIT. If the processor is the last sharer in the XCS network, the network is cleared down (termination). Otherwise, the processor leaves the network and the partner processors remaining in the network disconnect the processor from their network (leave reconfiguration). The registered functions are terminated in succession on the processor. The functions first of all terminate throughout the network (global termination) and then clear down their local resources (local termination). The network clear-down concludes when the global termination of the last function is completed. When a processor leaves the network, its exit is not completed until the partner processors have also disconnected this processor from the network. Following the network exit and the local termination of the functions, the MSCF connections are cleared down and the MSCF subsystem is terminated.
Notes
The exit from the XCS network is coordinated with the partners. The exiting processor only disconnects from the network in uncoordinated mode if cooperation between the network sharers has been disrupted due the loss of a connection or some other error. Following a crash, the XCS network must then be recovered by the remaining sharers in the network by means of a fail reconfiguration.
While a processor is leaving the network, no other processor can be added to the network or leave the network. Such operations will be delayed until the network exit has concluded.
If the network is in a reconfiguration phase, the exit of the processor (and hence also the shutdown) is delayed until the reconfiguration is completed. However, if the reconfiguration lasts longer than 10 minutes, the processor disconnects from the network (message MCS1003, reason 14).