Data Mover provides table-level restarts. A table-level restart indicates that prior to the job failure, all of the table rows were copied successfully. Data Mover will not copy the table rows again if the job is restarted. However, if the job fails at any point while copying the rows for a table, Data Mover copies all of the table rows again when the job is restarted.
Data Mover jobs can only be restarted using the restart command under the following conditions:
- The job failed while running. The job has a status of FAILED when viewed with the status or list_jobs commands.
- The job was stopped with the stop command. The job has a status of USER_CANCELLED when viewed with the status or list_jobs commands.
In either case, do not run the cleanup command before restarting the job, or all work that the job did will be lost. Also, resolve whatever issued caused the job failure before restarting the job.
After the issue is resolved, you can restart the job with the datamove restart -job_name <job name> command.
Restart Actions
Data Mover takes different actions when restarting a job, depending on the step in which the job failed. Actions are described in the following table:
| Step | Description | Data Mover Action During Restart |
|---|---|---|
| CLEAN_UP | Added to restart jobs. Tasks for cleaning up remnants of a failed job process. | Runs the entire restart job again. |
| MOVE_DEFINITION_BEFORE_LOAD | Tasks for setting up staging tables. | Finishes performing setup tasks that were not completed, then runs the rest of the job. |
| MOVE_TABLE_DATA | Teradata ARC, Teradata PT API, or Teradata JDBC tasks used to copy the table. | If using Teradata ARC:
If using Teradata PT API or Teradata JDBC:
|
| RESOLVE_TABLE_AFTER_LOAD | Tasks such as performing INSERT or SELECT statements from staging to the target table, and adding indexes and triggers. | Completes tasks that were not completed, then runs the rest of the job. |
| HARD_DELETE | Tasks to delete rows from the target table based on information in the primary key table. | Completes tasks that were not completed, then runs the rest of the job. |
| COPY_STATISTICS | Tasks to copy statistics. | Completes tasks that were not completed, then runs the rest of the job. |
| MOVE_JOURNAL_DATA | Teradata ARC tasks to copy journal data. |
|
| ROW_COUNT_VALIDATION | Number of copied rows for each table, ensuring rows in source and target tables match. Normally the last step of a job. | Completes tasks that were not completed, then runs the rest of the job. |
Additional Information
- The restart feature does not contain any logic to resolve the root issue of why the job failed. Resolve such issues before restarting the job.
- For jobs using Teradata ARC and its online archiving feature, if a job fails during the MOVE_DATA_STEP, running the restart command causes the online archive to be restarted. When completed, the target table reflects the condition of the source table at the point-in-time when the job was restarted and not at the point-in-time when the job was originally started.
- When performing restarts, Data Mover does not gather any new information about the source or target, but runs the remainder of the job as originally intended. Data Mover does not adapt the restart job to any changes that might have occurred since the job was originally created. In some cases, it might be necessary to run the cleanup command and create a new job if the original job plan is out of date.
- If the job uses Teradata ARC and fails in the MOVE_TABLE_DATA or MOVE_JOURNAL_DATA steps, the original Teradata ARC scripts are regenerated to copy only the uncompleted objects. If the Teradata ARC scripts contain user-added changes, these changes are not carried over to the restart Teradata ARC scripts.
- When a job is restarted, the status command output reflects only the status of the work being done for the restart job.