Retry Pipeline Deployments

Updated 1 month ago by Rashmi Nanda Sahoo

Pipeline execution might fail for many reasons, such as infrastructure changes or changes to resource access. In such cases, rerunning an entire Pipeline can be costly and time-consuming.

Harness provides an option to resume Pipeline executions from any executed Stage or from the failed Stage. These options enable you to quickly rerun stages after you identify the cause of the failure. 

Retrying a Pipeline or Stage is different from rerunning a Pipeline or Stage. When you rerun, you can select new values for Runtime Inputs. When you retry a Pipeline or Stage, you are running the Pipeline or Stage exactly as it was run before. See Run Specific Stages in Pipeline.

Harness provides an option to resume Pipeline executions from any previously executed Stage or from the failed Stage.

In this topic:

Before You Begin

Limitations

  • You can retry Pipelines when they are in Failed, Aborted, Expired, or Rejected status.
  • You cannot retry a Stage's execution if it was run successfully. You need to rerun the Pipeline and select Stage. This will be a new run of the Stage. To rerun a successful Stage in a Pipeline, click Rerun in the execution. For more information, see Rerun Stage.
  • You cannot change mandatory settings, parameters, or conditions when you retry.

Review: Serial and Parallel Stages

Stages can be added to Pipelines serially and in parallel. Here is an example that shows both:

How you run and retry Stages is different depending on whether you are retrying a serial or a parallel Stage.

Option: Retry Serial Stages

This topic assumes you have a Harness Project set up. If not, see Create Organizations and Projects.

You can create a Pipeline from CI and CD module in your Project, and then add Stages for any module.

Click on the failed deployment that you want to retry and click Retry Failed Pipeline.

The Retry Pipeline settings appear.

Choose the failed Stage or any previous Stage to retry the Pipeline. The Runtime Inputs for the selected and later stages are automatically filled from the previous execution. You can modify these Runtime Inputs while retrying the Pipeline.

Option: Retry Parallel Stages

When there are parallel Stages in your Pipeline and one of the parallel Stages fails, you can execute the failed Stage or all the parallel Stages.

Let's take an example of a Pipeline that has parallel Stages.

If Stage pstage1 fails, you can either execute all the parallel Stages or only the failed Stage.

All the previous values are populated for the stages. You can keep them or modify them as needed.

See Also


Please Provide Feedback