Re: undo tablespace keeps growing, why can't I reclaim space?

@hpuxrac: I don't follow the logic either...You're right, I should have
been more specific. The complete batch comprises of three transactions. Those 3 transactions consume 3g in PAT. That same batch w/ the same three transactions w/ the same data and indexes, everything the same except on PROD, comsumes 41G.

The only difference is, that during the run on production, I increased undo, because it was runningout of space. As per a contribution in another thread in by mr. Howard J Rogers, I have learned that, when there is space left in the undo tablespace, Oracle will claim it.

@Arun Mathur: thanks for your suggestion, but, again, I know how to
avoid the problem; direct path insert should do the job. It reportedly generates not as much undo.

What's done, is done. I cannot rerun the batch for quite some time, but I would like to know the mechanics and internals used by Oracle when it decides to reclaim space from existing undo segments.

Perhaps a deep look into the good old concepts manual could shed some light on this issue. Received on Mon Aug 15 2005 - 11:14:05 CDT