Active ARCHIVE processes which require I/O operations on save volumes (i.e. EXPORT, IMPORT, LIST, RESTORE and SAVE) and which have been interrupted (e.g. due to a system crash) can be restarted via the PROCESS statement. Instead of having to begin again from the beginning, they may resume at a defined position during the run.
This is possible only if the ARCHIVE run was started with PARAM RESTART=YES. In this case, checkpoints are written into the checkpoint file (ARCHIVE.CHKPT), with the help of which the run can be resumed.
Only the commands which were being processed at the time of the interruption are restarted. This means that:
the command must have been accepted as valid and the save or restore operation must have been started;
all subsequent commands are ignored and you must enter them again;
the environment for the restart must be exactly the same as at the time of the interruption (for example, you may not change the pubset or the user ID if a subsequent restart is to be successful).
In order to permit a restart of an ARCHIVE run which consists of several subtasks, of which one is aborted, all other subtasks are terminated immediately.
A restart without a directory file is pointless after an event such as a system crash or a device error. The file being processed prior to the interruption cannot be saved correctly if the EOF markers could not be written in the interrupted save operation.
This file cannot be logged. Message ARC0060 is issued.
Figure 5: Restarting ARCHIVE processes