Automatic restart

If an IMS cold start is being done and the previous shutdown was abnormal, RCF will automatically invoke Fast Path Recovery Utility. In both recovery modes:

1. TSSAFR1 and the DBRC utility DSPURX00 are invoked directly as subtasks.

2. A return code from TSSAFR1 of 4 or 12 indicates that batch backouts are required.

3. Any required batch backout jobs are submitted, using the RCFBBO member from the data set specified on the RCFJCL DD statement.

The number of recovery threads (for example, DEDB data output files created by TSSAFR1) is based on the value specified for the #DEDB parameter in the user-generated RCFALLOC module. A value greater than 1 implies multi-thread DEDB recovery.

Fast Path Recovery Utility processing can be performed using the following methods:

• Single-thread DEDB recovery

• Multi-thread DEDB recovery

Single-thread DEDB recovery

Note the following items when single-thread DEDB recovery is used:

• The #DEDB parameter in the RCFALLOC module is 1.

• TSSAFR1 has created a single DEDB recovery data output file. RCF directly invokes subtasks to perform the SORT, TSSAFR2 and DBRC steps of Fast Path Recovery Utility processing. RCF notifies the MTO of this processing via RCF25xx messages.

• If DEDB recovery problems are encountered, RCF informs the MTO and pauses to allow manual recovery execution.

Multi-thread DEDB recovery

Note the following items when multi-thread DEDB recovery is used:

• The #DEDB parameter in the RCFALLOC module is greater than 1.

• TSSAFR1 has created multiple DEDB data output files. In this mode, the RCFJCL DD statement is required. RCF uses the RCFFRU member from this data set to submit multiple jobs to perform the SORT, TSSAFR2 and DBRC steps of Fast Path Recovery Utility processing. Refer to Sample skeleton JCL procedures.

• RCF builds an internal table entry for each job. When all jobs have been submitted, RCF issues a WTOR message and pauses until the MTO replies to indicate that the jobs have completed. RCF verifies that all recovery threads completed successfully. If a recovery thread has not completed, RCF informs the MTO and pauses to allow manual intervention. The MTO can override this condition and force RCF processing to continue.

• If any job submission errors are encountered, RCF pauses to allow the MTO to perform the processing manually.