Runs

Prev Next

Available in VPC.

A Run represents the execution of a pipeline. Because a Run stores the experiment settings used when executing a pipeline, it can be rerun at any time. From the Runs interface, you can view details for any selected Run. The available details include the pipeline's execution graph, execution status, step-by-step inputs and outputs, execution logs, and metadata.

mlxp_console_runs03_ko

Component Description
Pipeline Select Select the pipeline to run by clicking Choose button.
Pipeline Version Select Select a version of the chosen pipeline by clicking Choose button.
Run name Enter a name for the Run.
Description Enter a description for the Run.
Experiment Select Select the experiment for the Run by clicking Choose button.
Service Account Enter the service account for the pod that runs the Run. If omitted, MLXP uses the default value.
Run Type Select One-off for a single execution or Recurring to configure a schedule.
Pipeline Root Enter artifact storage information if needed. If the pipeline root is already set for the namespace, this field is optional (see Pipeline Root Documentation link)
Run Parameters Enter input parameters for the Run if required. If default values exist, they are automatically populated.

mlxp_console_runs01_ko

Component Description
Create Run Configure settings to start a new Run.
Archive Change selected Runs to Archived status.
Active tab Shows all Runs that have run or are currently running.
Archived tab A place to move unnecessary or excessive Runs from the Active tab. You must archive a Run before deleting it.

mlxp_console_runs02_ko

Component Description
Retry Restart the Run.
Clone run Duplicate all configuration of the Run and create a new one.
Termniate Delete the Run.
Graph tab View the execution status or results for each step.
Detail tab View detailed properties of the Run.
Pipeline Spec tab View the YAML specification used to execute the Run.