z/OS DFSMSdfp Checkpoint/Restart
Previous topic | Next topic | Contents | Contact z/OS | Library | PDF


Operator Considerations

z/OS DFSMSdfp Checkpoint/Restart
SC23-6862-00

If a step requests automatic restart and is eligible for restart, the system displays the following message to request authorization for the restart:
   xxIEF225D   SHOULD  jobname.stepname.procstepname
                                   [checkid.] RESTART
Checkid appears in the message only if restart at a checkpoint is requested. It contains from 1 to 16 characters and identifies the checkpoint entry to be used to perform the restart. The operator must reply to the request for authorization as follows:
   REPLY xx,{‘YES’|‘NO’|‘HOLD’}

YES authorizes the restart, HOLD postpones it, and NO prohibits it.

If it is not clear that the restart should proceed, the operator should reply HOLD to the authorization message. Later, if restart should be allowed, initiate the restart by using the RELEASE command, which will be the same as an initial YES reply. If the decision is to deny the restart authorization, the operator can cancel the job in the HOLD queue. However, that HOLD, as well as YES, causes special disposition processing to occur during the abnormal termination. This processing keeps all OLD (or MOD) data sets and deletes all NEW data sets if a step restart is requested and keeps all data sets if a checkpoint/restart is requested. If you decide to disallow the restart but want to allow normal disposition (as requested on the job's DD statements) of data sets that were kept, release the job, wait until restart has begun, and then cancel the job.

If a job is canceled in the HOLD queue, any paging space allocated to a VIO data set is not released until the system is reinitialized. To avoid this situation, the operator should release the job, wait until the restart begins, and then cancel the job.

Using the VARY and UNLOAD commands, before the operator replies YES, and after the authorization request, may cause the system's volume and device configuration during a restart execution of the job to be different from what it is during the original execution of the job. The operator may eliminate use of defective volumes and devices.

The ability to use a different volume at restart usually exists only in the case of a NEW data set on a nonspecific volume that is not open at the checkpoint. If a checkpoint/restart is performed, the data set must not be open at the checkpoint. The ability to use a different device does not apply to the device or devices containing the system residence (SYSRES), SPOOL, or PAGE packs. If a checkpoint/restart is to be performed and a data set is open at the checkpoint, the same type of device must be allocated to the data set during both the original and restart executions.

Note: When an initiator selects a job for automatic step restart and the job is reinterpreted, no message is issued to the operator regarding virtual storage requirements, because its execution is not location dependent.

If ADDRSPC=REAL is specified on the JOB or EXEC statement, the restart may be delayed by the system waiting for the allocation of storage. If another job is using the required storage, you do not receive a message; you receive only a delay. Enter a DISPLAY A command to see if a system task or another job is using the storage required by the job with a nonpageable region. You can stop or cancel the conflicting task or job.

Go to the previous page Go to the next page




Copyright IBM Corporation 1990, 2014