This post is about the error we face in golden gate. Here we have the condition when we face this issue, symptoms, cause, and solution of the error.

Applies to:
Oracle GoldenGate – Version: 11.1.1.1.0 and later [Release: 11.1.1 and later ]
Information in this document applies to any platform.

Symptoms
Extract abends with below error:

ERROR OGG-01028 Record position (SeqNo: 336, RBA: 1040, SCN: 0.40418944 (40418944)) is beyond end of recovery (SeqNo: 335, RBA: 9016952, SCN: 0.40255602 (40255602), Timestamp: 2011-06-21 10:57:03.000000).

Cause
The extract may have hit a bug which can occur while processing a 0-length record while the extract is enabled for Bounded Recovery.

 

SOLUTION:

There are two methods to fix it.

1. Restart the extract ( most of the time restart fix the issue)
2. If a restart is not helping, then add the below parameter in extract and restart.

As a workaround, please use the below extract parameter and start the extract which will turn off the Bounded recovery. When BR is off, the Extract will do a normal recovery, so it needs all the archive logs file starting from recovery checkpoint of the extract.

BR BROFF

The only time BR is used is upon restart of extract, and only if there were any long running transactions that were persisted. If all of the transactions in the workload are of short duration (less than the BR interval default of 4 hrs), then even upon restart, standard recovery is active instead of bounded recovery.

And regarding transactions, we are only concerned with ones that modify data. There may, of course, be batch jobs which run and take a long time, so these could fall into the long duration category that’s mentioned above, but even in this case, as long as the redo logs/archive logs are available for standard recovery, the restart time should not be much longer with SR (standard recovery) than BR.

Once the extract has passed by the problematic RBA, this parameter can be removed.The permanent fix is planned for OGG v11.2.1.

Tagged:

About The Author

Leave a Reply