Skip Failed Datums

The errCmd parameter enables you to fail a datum without failing the whole job.
Before you read this section, make sure that you understand such concepts as Datum and Pipeline.

When HPE Machine Learning Data Management processes your data, it breaks it up into units of computation called datums. Each datum is processed separately. In a basic pipeline configuration, a failed datum results in a failed job. However, in some cases, you might not need all datums to consider a job successful. If your downstream pipelines can be run on only the successful datums instead of needing all the datums to be successful, HPE Machine Learning Data Management can mark some datums as recovered which means that they failed with a non-critical error, but the successful datums will be processed.

To configure a condition under which you want your failed datums not to fail the whole job, you can add your custom error code in errCmd and errStdin fields in your pipeline specification.

For example, your DAG consists of two pipelines:

  • The pipeline 1 cleans the data.
  • The pipeline 2 trains your model by using the data from the first pipeline.

That means that the second pipeline takes the results of the first pipeline from its output repository and uses that data to train a model. In some cases, you might not need all the datums in the first pipeline to be successful to run the second pipeline.

The following diagram describes how HPE Machine Learning Data Management transformation and error code work:

err_cmd logic

Here is what is happening in the diagram above:

  1. HPE Machine Learning Data Management executes the transformation code that you defined in the cmd field against your datums.
  2. If a datum is processed without errors, HPE Machine Learning Data Management marks it as processed.
  3. If a datum fails, HPE Machine Learning Data Management executes your error code (errCmd) on that datum.
  4. If the code in errCmd successfully runs on the skipped datum, HPE Machine Learning Data Management marks the skipped datum as recovered. The datum is in a failed state and, therefore, the pipeline does not put it into the output repository, but successful datums continue onto the next step in your DAG.
  5. If the errCmd code fails on the skipped datum, the datum is marked as failed, and, consequently, the job is marked as failed.

You can view the processed, skipped, and recovered datums in the PROGRESS field in the output of the pachctl list job -p <pipeline name> command:

Datums in progress

HPE Machine Learning Data Management writes only processed datums of successful jobs to the output commit so that these datums can be processed by downstream pipelines. For example, in your first pipeline, HPE Machine Learning Data Management processes three datums. If one of the datums is marked as recovered and two others are successfully processed, only these two successful datums are used in the next pipeline.

If you want to let the job proceed with only the successful datums being written to the output, set "errCmd" : ["true"]. The failed datums, which are “recovered” by errCmd in this way, will be retried on the next job, just as failed datums.