Start (and Stop) a Pipeline Execution - Amazon SageMaker

Start (and Stop) a Pipeline Execution

You can start and stop a pipeline execution in the Amazon SageMaker Studio console. For information about how to view a list of pipeline executions, see View a Pipeline.

To start and stop a pipeline execution in the Amazon SageMaker Studio console, complete the following steps based on whether you use Studio or Studio Classic.

Studio
To start a pipeline execution
  1. Open the SageMaker Studio console by following the instructions in Launch Amazon SageMaker Studio.

  2. In the left navigation pane, select Pipelines.

  3. (Optional) To filter the list of pipelines by name, enter a full or partial pipeline name in the search field.

  4. Select a pipeline name. The Executions page opens and displays a list of pipeline executions.

  5. You can create an execution from either the Executions or Graph pages. To create an execution from the Executions page, choose Create. To create an execution from the Graph page, choose Graph to the left of the executions table and then Create execution in the top right of the DAG.

  6. Enter or update the following required information:

    • Name – A name unique to your account in the AWS Region.

    • Description – An optional description for your execution.

    • ProcessingInstanceType – The Amazon EC2 instance type to use for the processing job.

    • TrainingInstanceType – The Amazon EC2 instance type to use for the training job

    • InputData – The Amazon S3 URI to the input data.

    • PreprocessScript – The Amazon S3 URI to the preprocessing script.

    • EvaluateScript – The Amazon S3 URI to the model evaluation script.

    • AccuracyConditionThreshold – The threshold of model accuracy to achieve to register the model into the registry.

    • ModelGroup – The registry into which to register the model.

    • MaximumParallelTrainingJobs – The maximum number of training jobs to run in parallel.

    • MaximumTrainingJobs – The maximum number of training jobs to run.

  7. Choose Create.

To stop a pipeline execution
  1. In the left navigation pane, select Pipelines.

  2. (Optional) To filter the list of pipelines by name, enter a full or partial pipeline name in the search field.

  3. Select a pipeline name. The Executions page opens and displays a list of pipeline executions.

  4. Select the execution to stop.

  5. Choose Stop.

To resume a stopped pipeline execution
  1. In the left navigation pane, select Pipelines.

  2. (Optional) To filter the list of pipelines by name, enter a full or partial pipeline name in the search field.

  3. Select a pipeline name. The Executions page opens and displays a list of pipeline executions.

  4. Select the execution to resume.

  5. Choose Resume.

Studio Classic
To start, stop, or resume a pipeline execution
  1. Sign in to Amazon SageMaker Studio Classic. For more information, see Launch Amazon SageMaker Studio Classic.

  2. In the Studio Classic sidebar, choose the Home icon ( ).

  3. Select Pipelines from the menu.

  4. To narrow the list of pipelines by name, enter a full or partial pipeline name in the search field.

  5. Select a pipeline name.

  6. From the Executions or Graph tab in the execution list, choose Create execution.

  7. Enter or update the following required information:

    • Name – Must be unique to your account in the AWS Region.

    • ProcessingInstanceCount – The number of instances to use for processing.

    • ModelApprovalStatus – For your convenience.

    • InputDataUrl – The Amazon S3 URI of the input data.

  8. Choose Start.

  • To see details of the execution or to stop the execution, choose View details on the status banner.

    • To stop the execution, choose Stop on the status banner.

    • To resume the execution from where it was stopped, choose Resume on the status banner.

Note

If your pipeline fails, the status banner will show Failed status. After troubleshooting the failed step, choose Retry on the status banner to resume running the pipeline from that step.

For a list of registered models, see Automate MLOps with SageMaker Projects.