IBM LoadLeveler for AIX 5L: Using and Administering

checkpoint

Indicates if a job is able to be checkpointed.

Checkpointing a job is a way of saving the state of the job so that if the job does not complete it can be restarted from the saved state rather than starting the job from the beginning.

The syntax is:

checkpoint = interval | yes | no

Where:

interval

Specifies that LoadLeveler will automatically checkpoint your program at preset intervals. The time interval is specified by the settings in the MIN_CKPT_INTERVAL and MAX_CKPT_INTERVAL keywords in the configuration file. Since a job with a setting of interval is considered checkpointable, you can initiate a checkpoint using any method in addition to the automatic checkpoint. The difference between interval and yes is that interval enables LoadLeveler to automatically take checkpoints on the specified intervals while the value yes does not enable that ability.

yes

Enables a job step to be checkpointed. With this setting, a checkpoint can be initiated either under the control of an application or by a method external to the application. With a setting of yes, LoadLeveler will not checkpoint on the intervals specified by the MIN_CKPT_INTERVAL and MAX_CKPT_INTERVAL keywords in the configuration file. The difference between yes and interval is that interval enables LoadLeveler to automatically take checkpoints on the specified intervals while the value yes does not enable that ability.

no

The step cannot be checkpointed. This is the default.

If you specify an invalid value for the checkpoint keyword, an error message is generated and the job is not submitted.

For example, if a checkpoint is initiated from within the application but checkpoints are not to be taken automatically by LoadLeveler you can use:

checkpoint = yes

For detailed information on checkpointing, see Step 14: Enable checkpointing.