Your problem is likely the oracle (transactions) are still holing on to and UNDO segment and oracle want to right itself.
>>They [offline undo segments] don't show a status of needing recovery.
I have not been in the exact position, but in a very similar one...sadly several times.
I would like to see more of the ORA-00600 or other ORA- error details.
I have my own notes titled;
Drop Rollback or UNDO Tablspace With Active / Corrupt / "NEEDS RECOVERY" Segments
I will send them to you directly.
But in a basically it goes like this;
- The Issue: ORA- errors
- UNDO/RBS Seem OK!?: select segment_name, status from dba_rollback_segs;
- Edit init.ora to Comment UNDO/RBS parameters
- UNDO/RBS Issue Obvious now!: select segment_name, status from dba_rollback_segs;
- IF YOU CAN *NOT* OPEN THE DATABASE: Edit init.ora to Comment UNDO/RBS parameters & ADD "_smu_debug_mode", event 10015
- startup Again
- View event="10015 trace file for corrupted rollback/undo segments
- NOW FIX CORRUPTED SEGMENTS
- Edit init.ora to "force" Rollback or UNDO offline: _OFFLINE_ROLLBACK_SEGMENTS or _CORRUPTED_ROLLBACK_SEGMENTS
- UNDO/RBS Issue Is Real (bad)!: select segment_name, status from dba_rollback_segs;
- Drop Rollback or UNDO Segments:
- UNDO/RBS All Gone...Easy and Simple to Drop UNDO/RBS Tablespace.
- Drop The Rollback or UNDO Tablespace
- RE-Create The Rollback or UNDO Tablespace
- Edit init.ora to Comment _OFFLINE_ROLLBACK_SEGMENTS= and UNcomment "undo_", "rbs" parameters.
- UNDO/RBS Issue GONE!: select segment_name, status from dba_rollback_segs;
I will past the whole doc in another post but not sure if it will make it through?
From: email@example.com on behalf of Strickland, Mark
Sent: Wed 10/26/2005 6:19 PM
Subject: Trying to drop old undo tablespace that needs recovery
I've opened a TAR, but I wanted to also solicit ideas from the list. I
inherited a database that has an old undo tablespace "UNDO2" that had a
datafile that has needed recovery since August 9th. I don't have
archived logs going back that far. The tablespace isn't being used. A
new undo tablespace called "UNDO" was created sometime back and is being
used. The database has been shutdown and restarted successfully since
the problem started. Every night when the tablespaces are put in backup
mode, errors are thrown in the alert log for the undo2 tablespace
because the file that needs recovery is offline. The tablespace has
offline undo segments. They don't show a status of needing recovery. I
researched it in Metalink and believed that I could offline drop the
datafiles in that old tablespace and then drop the tablespace. I was
able to execute the alter database offline drop statements successfully,
but when I tried to drop the tablespace, I got an ORA-00600. Anybody
got any ideas?