When running queries, Teradata Fabric connectors create temporary tables (or views) on the target database. You can control where these temporary objects are created by configuring one of the following NVP properties Temporary Database Name, Temporary Schema Name, Temporary Schema, or Database Name.
If a Teradata Fabric query is stopped, the connector attempts to cleanup these temporary objects as part of the rollback process. However, if the cleanup fails (or does not happen) – for example, due to a Teradata Fabric software restart – a manual cleanup may be required.
The temporary tables may contain metadata or customer data based on the query. If Temporary Database Name or Temporary Schema Name is not configured, Teradata Fabric uses the database provided in the NVP Database Name, Schema Name, or the default database/schema of the logged-in user.
When a manual cleanup is needed, locate the relevant tables either in the database/schema defined in the Temporary Database Name, Temporary Schema Name, or default Database/Schema.
- BigQuery connector: Tables start with temp_
- Hive connector: Temporary views are named import_query-uuid_view
- All connectors:
- For export clause (FT) queries: User-defined table names are appended with the query UUID, for example, xyz_query-uuid
- For RTO remote join queries: Names follow the format Temp_sess_session-id, for example, Temp_sess_1018_req_12_0_2_0