Introduction

To override a GitLab CI/CD pipeline job, you can achieve this by defining a job with the same name in your local .gitlab-ci.yml file as the job you wish to override. GitLab's CI/CD configuration merging mechanism ensures that the configuration defined in your local .gitlab-ci.yml takes precedence over any included or inherited configurations when job names are identical.

Here's how this works:

• Identify the job to override:

  Locate the job in your included or template CI/CD configuration that you want to modify.

• Create a job with the same name:

  In your project's .gitlab-ci.yml file, define a new job with the exact same name as the target job.

• Specify the desired changes:

  Within this new job definition, include the specific configurations you want to override. This could involve:

  • Changing the script: Modify the commands executed by the job.
  • Adjusting variables: Override or add new CI/CD variables specific to that job.
  • Modifying rules or only/except: Control when the job runs.
  • Updating stage or tags: Assign the job to a different stage or runner.
  • Adding before_script or after_script: Introduce commands to run before or after the main script.

Example:

If you have an included template with a job named build_app:

# .gitlab-ci-template.yml
build_app:
  stage: build
  script:
    - echo "Building application from template"
    - make build

You can override it in your project's .gitlab-ci.yml by defining a job with the same name:

# .gitlab-ci.yml
include:
  - local: .gitlab-ci-template.yml

build_app:
  script:
    - echo "Building application with custom logic"
    - custom_build_script.sh
  variables:
    CUSTOM_VAR: "true"

In this example, the build_app job in .gitlab-ci.yml will override the script and add a new variable to the build_app job defined in .gitlab-ci-template.yml. The stage will be inherited from the template unless explicitly overridden.

Conclusion

As you can see, it's quite simple to achieve a job definition override in gitlab. This allows for a templated job defition to be used and re-used in as many jobs as necessary as well as the flexibility to re-define the said job when there is a specific use case that requires it.