Restarting jobs

If a member fails, and jobs in execution are recovered and requeued for automatic restart (either by a warm start or the $E MEMBER command), those jobs that have journals are given affinity only to the failed member. If the failed member is unavailable, the operator can change affinity with the $T J command to attempt restart on another member.

You can also use automatic restart management to automatically restart batch jobs and started tasks (both are considered jobs). If the job is registered with automatic restart management, automatic restart management restarts jobs when either:

An executing job unexpectedly ends
The system on which a job is executing unexpectedly ends or leaves the sysplex.