z/OS UNIX System Services Planning
Previous topic | Next topic | Contents | Contact z/OS | Library | PDF


Successful shutdowns

z/OS UNIX System Services Planning
GA32-0884-00

The shutdown succeeds only if all non-permanent z/OS UNIX processes end, all permanent processes are successfully checkpointed, and if all physical file systems are successfully quiesced. Otherwise, the shutdown request will fail. Nonpermanent processes within jobs that are not prepared for shutdown cause the shutdown request to fail. These jobs are identified in messages so that you can force these jobs to end. At this point, because most processes have been ended, you should force the hung jobs to end and then try the shutdown again. Because some jobs might have ended abnormally, JES spool resources might have accumulated for these jobs; you will have to purge them using commands such as $POJOBQ,READY.

At each phase of shutdown, it is possible that there could be a stall where no shutdown activity is occurring. That situation could cause the shutdown to hang. If such a situation is detected, the shutdown will wait approximately six minutes for the stall to resolve itself. If the stall does not resolve itself by then, the shutdown request will fail.

Because some resources are tied to components outside of the scope of the kernel (shared memory, mmap, shared libraries, for example), you must end any application that is using any of these resources before z/OS UNIX can be ended, including applications that are registered as permanent.

Tip: Because the F OMVS,SHUTDOWN support encompasses the existing support in the F BPXOINIT,SHUTDOWN= command, you do not need to issue F BPXOINIT,SHUTDOWN before using F OMVS,SHUTDOWN. If F OMVS,SHUTDOWN fails, z/OS UNIX services are reenabled whether or not a F BPXOINIT,SHUTDOWN= was done prior to the F OMVS,SHUTDOWN command. An F BPXOINIT command of any kind issued when OMVS is shut down is ignored.

Guideline: Use F OMVS,SHUTDOWN carefully because this method will take down other system address spaces. As a result, some system-wide resources might not be completely cleaned up during a shutdown and restart. Do not use this command to shut down and restart the z/OS UNIX environment on a frequent basis. (If you do so, you will eventually have to do a reIPL.) An example of a system-wide resource that can be consumed due to the shutdown are non-reusable ASIDs. If colony address spaces are being used, a non-reusable ASID will be consumed for each colony address space that is shut down. For this reason, installations should plan on increasing the value set for the RSVNONR= parameter in the IEASYSxx parmlib member to account for the consumption of non-reusable ASIDs due to each shut down of OMVS. If this value is not increased, the installation might receive an error message after shutting down the system multiple times.

If a shutdown does not succeed, a critical z/OS® UNIX resource might not have been available during the shutdown, due to a prior system problem, such as latch contention. If a resource such as the z/OS UNIX file system MOUNT latch could not be obtained, the shutdown request to likely to stall and then fail.

Go to the previous page Go to the next page




Copyright IBM Corporation 1990, 2014