When an application is running and the database crashes:
- Database sessions and session ids are preserved, and can continue to be used.
- All outstanding database transactions are aborted and all of the databases involved are rolled back to the state they would have been in if the transactions had not begun.
- All requests that have not completed are aborted and any work they have performed is rolled back.
- All spool files, including those databases retained for completed requests, are discarded.
- Applications already waiting for a response before the crash/recovery occurs are notified.
- Applications submitting a request during the crash and recovery process may be notified.
- Applications later using the abort, fetch, rewind, or end request function for a Teradata SQL request aborted by the crash/recovery process, or submitting a new request in a transaction that was in progress when the crash occurred and thus was aborted in the recovery process, will be notified.
- If a restart occurs during a logoff request, CLI automatically resubmits the request.