Off-line Rollback Technical Concepts

Method

The batch recovery process implements user journals to mark the beginning and end of transactions.  The processing defers the clearing of all update transaction files until the entire batch stream has completed, and the successful/unsuccessful completion of jobs has been determined.  The delay of file clears is not used for batch reporting programs, since these programs can also run concurrently with the online applications.  If a report program fails and is subsequently rolled back, the reports must be resubmitted.

Processing

The Off-line Rollback Control program (RM100C) provides control for the off-line recovery stream.  The stream flow is as follows:

1.   The Recovery Management Off-line Transaction Check program (RM100E) is called. This program is described in detail in the Program Descriptions section.

a.   The program checks the Recovery Management Sync Point File for failed off-line programs.  The Sync Point File contains the beginning (BE) and ending (EN) journal entries that mark the boundaries for each job.  This file also contains the recovered error (RE) journal entries that signify that a job has been recovered previously.

b.   Any job that is found to have failed is placed into the Off-line Rollback Limit File (RM100AP).

2.   The next step of the rollback procedure is determined.

a.   If no failures are detected, processing continued at step 6.

b.   If failures are detected, a screen is displayed with a list of failed transactions.

(1)  If F13 (Return Without Recovery) is pressed, processing continues at step 6.

(2)  If F22 (Continue Recovery) is pressed, processing continues at step 3.

3.   The Recovery Management Off-line Rollback Extract program (RM120C) is called.

a.   The program displays journal entries for the off-line programs that were determined to have failed (these are found by reading RM100AP). 

b.   These entries are placed in the Off-line Rollback File (RMJRNLP).  There are three types of journal entries that are rolled back:  update (UB), add (PT), and delete (DL).

4.   The Recovery Management Off-line Remove Journal Change program (RM140C) is called.

a.   The program removes journal changes for each failed off-line job that is found in RM100AP.  The changes, within the specified boundaries, are removed from all files that are being journaled.  The boundaries for the remove journal change command are taken from RM100AP and are determined in step 1.

5.   The Recovery Management Off-line Send Recovered Error Message program (RM160C) is called.

a.   The program writes a recovered error (RE) journal entry for each failed off-line job that is found in RM100AP.

b.   If a batch update program completed normally, then the input transaction files for the program are cleared.

c.   If a batch update program fails, then audit trail and transaction register files are cleared for the program.

6.   The next step of the rollback procedure is determined.

a.   If no jobs failed, then processing continues at step 8.

b.   If there are failed jobs, then processing continues at step 7.

7.   The Recovery Management Off-line Rollback Print program (RM180L) is called.  This program is described in detail in Program Descriptions section.  The program reads the Off-line Rollback File (RMJRNLP) and produces a report of all rolled back journal entries by file.

8.   The Recovery Management work files (RMJRNLP and RMSYNCP) are cleared, and processing ends.