How does cleanup work?

The scheduler cleans up the data set of a job in one of the following ways:

At the time the job is submitted by the controller, a step is added to the JCL before the existing steps. The name of the added step is EQQCLEAN. It uses the list of data sets to be deleted, cataloged, or uncataloged as input. This list is obtained from the history of previous runs. To get the history, the controller requests the job log from the data store and then merges the history in the local repository of the controller.
A single-step job, created by the controller, is executed before the job run.

If automatic job recovery is used, cleanup and automatic recovery cooperate to ensure that cleanup actions are either completed, or revoked before any automatic recovery action is started for the operation. For information about automatic job recovery, see Automatic recovery of jobs and started tasks.

Note that a user exit, EQQUXCAT, is available to be called by EQQCLEAN to discard specific cleanup actions at job run time. For details, see the Customization and Tuning manual.