Guaranteed Restore Point: Missing Automatic Expiration for Upgrade Rollback

When performing an in-place Oracle database upgrade, using a guaranteed restore point provides a mechanism to roll back if the process fails or results in regressions. Unlike standard flashback points, a guaranteed point prevents the database from purging the required undo data, ensuring the state is preserved regardless of the undo retention settings. The primary uncertainty lies in the management of the undo tablespace during the long-running upgrade process. If the upgrade involves significant DML or schema changes, the undo tablespace may grow rapidly because the guaranteed point cannot reclaim space segments. There is no built-in automated trigger to drop the guaranteed restore point once the upgrade is verified as successful, potentially leading to disk space exhaustion. How does Oracle handle undo tablespace growth when the MAXSIZE limit is reached while a guaranteed restore point is active during an upgrade? What are the recommended strategies for ensuring the undo tablespace does not fail the upgrade process without compromising the rollback capability?