If a job terminates abnormally, the checkpoint/restart facilities
allow you to restart the job, as follows:
- Automatic step restart, that is, restart by the system from the
beginning of a job step.
- Automatic checkpoint restart, that is, restart by the system
from a checkpoint within a job step.
- Deferred step restart, that is, restart at a later time from the
beginning of a job step.
- Deferred checkpoint restart, that is, restart at a later time
from a checkpoint within a job step.
Restarts are controlled by:
- RD parameters on JOB and EXEC statements. (Restart is not supported
for started tasks; do not use the RD parameter on the JOB statement
for a started task.)
- Checkpoints, if written. Each time a CHKPT macro is executed,
a checkpoint is written.
- The job journal, which is only required for an automatic restart.
In a JES3 system, the programmer can code a JOURNAL parameter on the
JES3 //*MAIN statement to control whether JES3 creates a journal
for the job.
- In deferred restarts, a RESTART parameter on the JOB statement
for the restarting job and a SYSCHK DD statement to identify the data
set containing the checkpoint written in response to the CHKPT macro.
(Restart is not supported for started tasks; do not use the RESTART
parameter on the JOB statement for a started task.)
Use of Restart <
Either form of restart saves having to execute the job from its
beginning. If the job is long, restarting can save a lot of time
and computer resources.
For more information about restarting, see z/OS DFSMSdfp Checkpoint/Restart.
Examples:
//J1 JOB ,'B. MORRISON',RD=RNC
//J2 JOB ,'H. MORRILL'
//S1 EXEC PGM=TESTING,RD=R
//S2 EXEC PGM=TESTED,RD=NC