RD Parameter

Purpose

The RD (Restart Definition) parameter specifies how a job should be restarted if it fails or is canceled and needs to be resubmitted. It provides control over both automatic step restart and checkpoint restart capabilities, allowing you to determine whether a job can be restarted at all, and if so, how that restart should be handled.

Syntax

JOB Statement Format

jcl

1
//jobname JOB parameters,RD=value

Parameter Values

value

Value	Description
R	Requests restart at the beginning of the job (first step) if the job fails
RNC	Requests restart at the beginning of the job, but any checkpoints are to be ignored
NC	Suppresses automatic step restart and checkpoint restart, even if requested by the RESTART parameter
NR	Suppresses automatic step restart but allows checkpoint restart if requested by the RESTART parameter

Note:

If you omit the RD parameter, no automatic restart is requested. However, step restart and checkpoint restart can still be requested using the RESTART parameter.

Usage Examples

Basic Job Restart

jcl

1
2
//TESTJOB JOB (ACCT),'JOHN DOE',CLASS=A,
//         MSGCLASS=X,RD=R

This job will be restarted from the beginning if it fails or is canceled.

Restart Without Checkpoints

jcl

1
2
//TESTJOB JOB (ACCT),'JOHN DOE',CLASS=A,
//         MSGCLASS=X,RD=RNC

This job will be restarted from the beginning if it fails, ignoring any checkpoints taken.

No Restart or Checkpoints

jcl

1
2
//TESTJOB JOB (ACCT),'JOHN DOE',CLASS=A,
//         MSGCLASS=X,RD=NC

This job cannot be restarted automatically and will not use checkpoints, even if specified in the RESTART parameter.

Checkpoints Only

jcl

1
2
//TESTJOB JOB (ACCT),'JOHN DOE',CLASS=A,
//         MSGCLASS=X,RD=NR

This job will not restart automatically but can be restarted from checkpoints if specified in the RESTART parameter.

Restart Concepts

Types of Restart

Automatic Step Restart

When a job step terminates abnormally, the system can automatically restart the job from the beginning of the failing step. This restart occurs without user intervention when the following conditions are met:

RD=R or RD=RNC is specified on the JOB statement
The failing step was not already being rerun
The job is eligible for restart (not in a HOLD state)
Operator allows the restart when prompted

Checkpoint Restart

A program can establish checkpoints during its execution by issuing the CHKPT macro. If the program fails after establishing checkpoints, it can be restarted from the last checkpoint rather than from the beginning of the step. The RD parameter controls whether these checkpoints are used:

RD=R allows checkpoint restart
RD=RNC suppresses checkpoint restart
RD=NC suppresses checkpoint restart
RD=NR allows checkpoint restart

Restart Process Flow

When a job or step abnormally terminates, the system examines the RD parameter
Based on the RD parameter value, the system decides whether to initiate automatic restart or use checkpoints
The operator is notified and can choose to allow or prevent the restart
If restart proceeds, the system positions input data sets at the appropriate points
Output data sets created by the job before the restart are handled according to disposition parameters

Relationship with Other JCL Parameters

Interaction with RESTART Parameter

The RESTART parameter on the JOB statement specifies where to restart a job. The RD parameter controls whether this restart capability is available:

RESTART Usage	RD Impact
`RESTART=*`	RD=NC prevents restart; RD=NR, R, or RNC allows restart
`RESTART=(stepname)`	RD=NC prevents restart; RD=NR, R, or RNC allows restart at the specified step
`RESTART=(stepname,checkid)`	RD=NC or RD=RNC prevents checkpoint restart; RD=R or RD=NR allows checkpoint restart

Interaction with SYSCHK DD Statement

When requesting a checkpoint restart, you must include a SYSCHK DD statement to identify the checkpoint dataset:

jcl

1
//SYSCHK   DD  DSN=checkpoint.dataset.name,DISP=OLD

The SYSCHK DD statement refers to the dataset containing checkpoint records created by the CHKPT macro.

Practical Applications

When to Use Each RD Value

Value	Best Used When
R	Long-running jobs that should be restarted automatically if they fail Jobs with built-in checkpoint capability that need automatic recovery Production jobs that must complete despite system failures
RNC	Jobs that should restart from the beginning regardless of checkpoints When checkpoint data might be unreliable or outdated When full reprocessing is safer than partial reprocessing
NC	One-time processing jobs that should not be restarted Testing and debugging scenarios where retries are not wanted Jobs where partial processing could cause data inconsistency
NR	Jobs that should only be restarted manually from checkpoints When operator inspection is required before restart Debugging scenarios where checkpoint restart is needed but automatic restart is not

Common Scenarios

Batch Update Jobs: Use RD=R for critical batch update jobs to ensure they complete processing
Data Extraction Jobs: Use RD=RNC when extracting data to ensure complete extraction without partial updates
Report Generation: Use RD=NC for regular reports that can simply be rerun if they fail
Database Maintenance: Use RD=NR to allow checkpoint restart only after manual verification of database integrity

Considerations and Limitations

Data Set Integrity

When a job is restarted, data set integrity is a critical concern:

Input data sets: Must be repositioned correctly for restart
Output data sets: May contain partial or duplicate data after restart
NEW data sets: Should use DISP=(NEW,CATLG,DELETE) or DISP=(NEW,CATLG,KEEP) to ensure proper handling
MOD data sets: May contain duplicate data after restart if not properly managed

Performance Impact

Creating checkpoints has a performance impact on the running job. Each checkpoint operation:

Writes job state information to the checkpoint dataset
May force I/O buffers to be written to disk
Increases job execution time proportionally to checkpoint frequency

Storage Requirements

Checkpoint datasets require disk space to store:

Program status information
Open dataset position information
Contents of the program's storage areas

Best Practices

Document restart strategy for each production job
Test restart capability before implementing in production
Consider data integrity when designing restart options
Use checkpoints judiciously - too many can impact performance, too few can result in excessive reprocessing
Coordinate with operations staff to ensure they understand restart options for each job
Review restart logs to identify patterns or recurring issues
Maintain checkpoint datasets with appropriate retention periods
Consider using RD=NC during testing to prevent unintended restarts

Related Concepts

JOB Statement - The statement where RD is specified
RESTART Parameter - Specifies the step or checkpoint ID at which to restart a job
SYSCHK DD Statement - Required for checkpoint restart
DISP Parameter - Controls dataset disposition during normal and abnormal termination