Had an interesting problem recently where a rollback segment datafile became corrupted. For this instance I'm running v8.1.7 on an NT4.0 server, this instance is used for DSS.
We were performing a load and the server locked up. When we rebooted the server and restarted the database, the database recovery could not finish. This was due to the corrupted rollback segment. Now we couldn't take the rollback segment offline as it had active transactions to rollback, and the rollback could not complete as the datafile was corrupt.
Due to time constraints I could not save this off to another server to play with, I recovered the database with the previous nights cold backup and the data load continued. While I wish I could have kept this around to experiment with I did not have the time or a spare server large enough for this. Has anybody else out there had a similar problem? And if so how else did you resolve it? Thanks.