Debugging Snapshot too old

[ocp2b]

I need to debug a problem one of our older databases is having (Oracle 7.3.4). A user is getting the snapshot too old error message. Is there a way to go through the trace file or create a trace file so that I can analyze what is happening? I'm not familiar with 7.3.4. I realize the quick fix is to increase the redo log file size or add another redo log file, but I suspect that won't solve the problem. I suspect that the user is issuing a fetch/select for update and is leaving the transaction open for an unreasonable amount of time.

Any assistance is greatly appreciated.

[sonofsita]

A user is getting the snapshot too old error message

means problem with RBS.. increase the size of RBS not the tablespace..I think it is ORA-01555 error.

The snapshot is too old because there aren't enough Rollback segments.

[pando]

Originally posted by ocp2b
quick fix is to increase the redo log file size or add another redo log file

sorry but that is incorrect, there is nothing to do with redo logs ora-1555. It is your RBS settings, you either

1. increase RBS extent sizes
2. increase minextents
3. increase optimal in you RBS
4. increase number of RBS

it happenes becase the extent which contains the before image of a transaction has been overwritten by another transaction (it is overwritten because the previous transaction has finished and freed the extent) and a long query cannot find that consistent before image in the RBS