Previous Topic: Recovering a Large Number of FilesNext Topic: Recovery Procedures from Database File I/O Errors


Recovery Procedures After a Warmstart Failure

Before You Begin

Before you begin the recovery process, determine why the warmstart failed. Start by checking any shutdown or warmstart messages. The failure could be due to:

Corrective Action

If the failure is due to:

Change

Action

Changes in the DMCL and a timestamp mismatch is detected

Warmstart the system using the prior version of the DMCL load module

Note: The warmstart failure could have been avoided through the use of change tracking by the CV.

Changes in the startup JCL

Correct the JCL and restart the system

Note: The warmstart failure could have been avoided through the use of change tracking by the CV.

Software maintenance

Back out the maintenance and restart the system

Disabling change tracking

Warmstart the system specifying an IGNORE_SYSTRK_DMCL SYSIDMS parameter and ensure that the correct DMCL load module and startup JCL is used

Note: For more information about implementing change tracking, see "Change Tracking" in the CA IDMS System Operations Guide.

Steps

In the unlikely event that hardware or software problems prevent the warmstart process from recovering the database, follow these steps:

Action

Statement

Offload all journal files

ARCHIVE JOURNAL with the FULL option to offload all full journal files. This should be followed by an ARCHIVE JOURNAL with the READ option to offload the journal that was active when the abnormal system failure occurred.

Check for incomplete InDoubt transactions.

Be sure to include all archive files created since the last quiesce point.

PRINT JOURNAL and FIX ARCHIVE

If incomplete InDoubt transactions exist, complete them manually by creating a new archive file.

For more information, see 21.4.2, “Incomplete Transactions and Manual Recovery”.

FIX ARCHIVE using manual recovery control file input to complete the InDoubt transactions.

Recover the transactions that were active at the time of the system failure (that is, abended transactions)

ROLLBACK with the ACTIVE option

Unlock the areas that were not accessed during the rollback process. The ROLLBACK statement identifies what areas it unlocked.

UNLOCK

Reinitialize the journal files

FORMAT JOURNAL

Data Sharing Considerations

If a member of a data sharing group is unable to warmstart and manual recovery must be undertaken, any shared area that was being updated by the failing member must be quiesced in all other members of the data sharing group before the ROLLBACK utility is executed. To quiesce the area, change its status to OFFLINE or TRANSIENT RETRIEVAL. Do not use the DCMT QUIESCE command to quiesce the area.

Note: For additional data sharing considerations, see 21.11, "Data Sharing Recovery Considerations".