For jobs that take a large number of checkpoints with LOCK, the checkpoint number counts up to 999 and then recycles to 1 (leaving zero undisturbed). When this recycling occurs, previous checkpoint files are lost as new ones using the same numbers are created.

If a temporary disk file is open at a checkpoint, it is locked under the CP directory. If it is subsequently locked by the program, the name is changed to the current file title. At restart time, the file is sought only under the CP directory, resulting in a no-file condition. To avoid this condition, all files that are to be locked eventually should be opened with the file attribute PROTECTION assigned the value SAVE. To remove the file, it must be closed with PURGE. True temporary files, which are never locked, do not have this problem. All data files must be on the same medium as at the checkpoint, but need not be on the same units or the same locations on disk or disk pack. They must retain the same characteristics, such as blocking. The checkpoint/restart system makes no attempt to restore the contents of a file to their state at the time of the checkpoint; the file is merely repositioned. At this time, volume numbers are not verified.