A CloudFormation update fails, and rollback also fails because an externally changed resource cannot return to its previous configuration. The stack is UPDATE_ROLLBACK_FAILED. Operations fixes the resource problem. What should the engineer do next?

1. Run DetectStackDrift and expect drift detection to change the stack back to UPDATE_ROLLBACK_COMPLETE.
2. Skip every resource during rollback even though the original failure is fixed.
3. Run UpdateStack immediately without completing the failed rollback.
4. Run ContinueUpdateRollback and monitor the stack until rollback completes before attempting another update.