How to dial with ORA-01555 Snapshot too old and SDM what to do

  • KM03660817
  • 30-Jun-2020
  • 30-Jun-2020

Summary

Error raised when the amount of rows on OLTP has increased a lot cuasing this issue.

Question

We want to explain how to deal with this error. This documentation only applies to Oracle database.
There isn’t a recipe to fix it at once, but the purpose of this document is to give you a couple of clues to take in mind as part of troubleshooting, with high probability to get the issue fixed.
Most common cause is:
Customer is running a BF or doing a relocate and during the processing, Oracle sent fail to SDM saying this Ora-01555, meaning, a lot of data is being moved but the rollback segment or undo segment now is filled.
This block the process because SDM is sending still data to Oracle for archive, breaking the process of the BF.

Answer

How to deal with:
First you have to analyze what is happening, specially talking about rows amount, most of time, customer ran BF sometime ago and now execute it again getting that error.
Then it seems to be that, there are a lot of records more on OLTP today than last time BF ran, other point is, maybe on DB side DBA reduced rollback segment.
You have to attach this from two fronts:
·         From WC using following screen to modify values, screenshot where to modify is:
·         From Oracle:
o   Modify storage parameters and configuration parameters, this belongs to your dba, so, you have to explain him about the issue and work together based on following information.
What to modify as try:
·         On WC (WebConsole) set:
·         Combined=5000
·         combined Statement count=1000
·         D mov batch size = 5000000
·         Parallel workers=10
·         Selection batch size = 500000
Note:
This paremters are a good base where to start, however, you must work with your DBA looking for fit values allowing you to process avoid this error.
 
Read help on each parameter..
 
·         On Oracle side:
o   Increasing rollback segment or undo segment
o   Check for redo log and share pool size, if DB is not set automatic about growing.