I have an oracle database 11g, there is a daily process that I run that is filling up the undo tablespace exponentially to 3.6 GB before the used space comes down to below 10 MB and the tablespace size will remain at 3.6 GB way above the initial size of 500 MB. I have 3 redo log groups sized at 100 MB each. My question is what leads to that exponential space usage in my undo tablespace? Please assist.
I recommend that you periodically query which users are consuming the space in the UNDO tablespace. Perhaps you can run your query from cron. A sample query is at blog.contractoracle.com/…/undo-tablespace-usage.html or you can Google the words “undo tablespace usage”.