When to Code a Checkpoint

Using the CHKPT macro instruction, you can code a checkpoint within your program or at end-of-volume. The checkpoint routine records information about certain system information, user storage (including user programs in user storage), and the data sets used by the step executing the CHKPT macro. Recorded information about data sets includes the following:

For all data sets, the checkpoint routine records the information which can be coded on a DD statement, such as the device type and volume serial numbers. The contents of the step's job file control blocks (JFCBs) and job file control block extensions (JFCBEs) are also recorded, as well as the contents of the generation data group name tables (GDGNT).
For data sets which are (1) open at the checkpoint, (2) are processed on either magnetic tape or direct access devices, (3) and use the BSAM, QSAM, BPAM, VSAM, or EXCP access methods, the checkpoint routine records the information needed to reposition the data set if restart occurs at a checkpoint.

If a checkpoint is taken and an output data set is extended onto a second direct access volume because end-of-volume occurred on the first volume and there was no more space available on the volume, or because the data set contained 16 extents, restart subsequently occurs at that checkpoint and the system does not delete the extension of the data set.

Note that, when a step using the universal character set (UCS) feature is restarted, the system does not determine if the UCS buffer or forms control buffer (FCB) buffer is properly loaded, nor does it alert the operator to the UCS or FCB requirements of the step.