Table of Contents
A process can be discontinued by any of a variety of causes, including system commands, program faults, or resource limits. In most situations, this discontinuation is permanent. The system does not attempt to continue execution of the process or restart it from the beginning. The system removes the process stack, process information block (PIB), task attribute block (TAB), and any temporary files that the process was using. Only the permanent files used by the process are preserved and reflect all changes made by the process before it terminated.
However, in certain cases you can cause various aspects of a process to be saved for later restarting. This section discusses the following methods of saving and restarting processes:
-
Work Flow Language (WFL) job restarts. The system automatically stores information about WFL jobs that enables the jobs to restart after a halt/load. Further, you can terminate and restart a WFL job manually with the RESTART (Restart Jobs) system command.
-
Checkpoint facility. This facility stores a complete copy of a process and its data items so that the process can be re-created later from a given point in its execution.
-
CHECKPOINTARRAY procedure. ALGOL and NEWP processes can use this procedure to copy an array to disk or recover the array from disk. CHECKPOINTARRAY has fewer restrictions than the checkpoint facility and thus is suitable for wider use.
An additional restart method is the RESTART task attribute, which restarts a process that was terminated due to a fault. For further information about the RESTART attribute, refer to Retrying a Failed Task in Determining Process History.
