I have a stored proc that do a very large update. Sometimes the job failed with error ORA-30036 Unable to extend segment by 8 in undo tablespace 'undotbs2'
But after a few hours, we reran the job and it completed successfully.
I checked and found undotbs2 already has AUTOEXTENSIBLE set to YES, and size is 3 GB, so I guess the undo tablespace already has pretty decent size there, and has automatic space management already turned on.
My question is, why does it complete successfully after we rerun it? Is it because there were other transactions using undotbs2 at the same time? For this error, Oracle mentions "An alternative is to wait until active transactions to commit.", does "active transactions" refer to other transactions/sql that were happened to run besides the stored proc?
Oracle version is 11.2.0.1.0
Thank you