Siebel Data Warehouse Installation and Administration Guide > DAC Quick Start >

Handling ETL Failures with the DAC

This section includes the following topics:

• When the Execution of an Execution Plan Fails
• In Case of Abnormal Termination of the DAC Server
• Discarding the Current Run Execution Plan
• Failure of Aggregator Transformation Tasks with Sorted Input
• Performing a Full Load Again

When the Execution of an Execution Plan Fails

When an execution plan is executed, if a task fails, the status of the tasks that are dependent on the failed task is changed to Stopped. While tasks are still running, the execution plan's status is Running. When all the tasks have been run, and if one or more tasks have failed, the execution plan's status is changed to Failed.

You can check the tasks that have failed in the Current Run tab of the Execute view, fix the problems, and then requeue the failed tasks by changing the status to Queued. You can then restart the ETL. All the tasks will then be rerun. You can also manually run a task, change its status to Completed, and then restart the ETL. Tasks with a Completed status are skipped.

CAUTION:  The DAC server does not validate tasks that have been run manually.

To restart a Failed ETL, click Run Now from the Current Run tab of the Execute view.

In Case of Abnormal Termination of the DAC Server

If the DAC server fails during the execution of the ETL, the status of the ETL execution will remain as Running. When the DAC server is started again, it will automatically run the ETL if the Auto Start ETL system preference is set to True. If the same system preference is set to False, when the server restarts, it will set the correct status as Failed. In order to execute the ETL from the point of failure, submit the request to the server again.

The DAC server will automatically terminate if it looses connection to the DAC repository.

Discarding the Current Run Execution Plan

You can discard an execution plan that failed by navigating to the Current Run tab, right-clicking on the execution plan and changing its status to Mark as Completed. This will force the run status to be updated as Completed. When you submit a request for another run, the DAC server creates another instance of it.

CAUTION:  Perform this procedure in a development or testing environment only, since it might leave the data in an inconsistent state, causing you to have to reload all of the data.

Failure of Aggregator Transformation Tasks with Sorted Input

Tasks that use Informatica Aggregator transformation can fail when the Sorted Input option is active. The tasks SDE_DTLFORECASTFACT and SDE_COSTLIST are examples of tasks that can fail in such a situation.

To prevent such tasks from failing, in Informatica Designer, navigate to Mapping Designer, open the corresponding mapping, and in the Aggregator transformation, remove the check from the Sorted Input check box.

Performing a Full Load Again

In test environments, after running several ETLs, you might want to perform a full load again. You do this by accessing Tools > ETL Management > Reset Data Warehouse.

CAUTION:  This procedure will cause the DAC server to perform a full load the next time an ETL is run, which will truncate data from the target tables. Therefore, this action should be used only in development and testing environments. You should not perform this procedure in a production environment unless you plan to rebuild the data warehouse in its entirety.