Change acceleration for manual jobs

  • Release version: Washingtondc
  • Updated August 1, 2024
  • 1 minute to read
  • Enable change tracking for the pipeline in the tool record page in DevOps Change Velocity.

    The GitLab job under change control must have these instructions for the pipeline execution to be resumed or canceled via the change request:
    • when: manual
    • allow_failure: false
    For example:
    
    deploy:
      stage: deploy
      tags:
        - local-runner1
      when: manual
      allow_failure: false
      script:
        - echo 'Deploy'
    Note:
    For when:manual based changes, for a change request to get created at a certain stage, all the previous stages must complete successfully. If any of the previous stages is not invoked or not successful, even though there is no dependency of the current stage on its immediate previous stage, a change request will not get created in ServiceNow.

    GitLab pipeline parallel stages is supported with GitLab Docker Image. For more details, see GitLab pipelines with parallel jobs and Implement custom actions for pipelines using generic Docker container image.

    Refer to the CI/CD pipeline configuration reference for more information on how to configure a GitLab job.

    Additional considerations:
    • If allow_failure is set to true, the pipeline continues even when the change is rejected.
    • A user with the appropriate role access in GitLab can unblock and continue a pipeline regardless of the change request state.
    Table 1. GitLab change acceleration behavior
    Manual execution Change acceleration in step Change request approved Result
    Yes Yes N/A

    If the manual job is under change control, the change is automatically created.

    Yes

    The manual job is automatically executed.

    No

    The manual job is automatically rejected/failed.

    No N/A

    The manual job waits for manual intervention from the pipeline owner via the GitLab UI (default behavior).

    No Yes N/A

    The change request is not created.

    Note:
    Parallel jobs are displayed sequentially, based on the order in which the jobs are queued for execution.