Module 3 Lesson 5: Job Control (Rules)

You don't always want every job to run.

• You don't want to "Deploy to Production" when you are just fixing a typo in a README.
• You don't want to run expensive "Load Tests" on every tiny commit.
• rules gives you the logic to control your pipeline.

1. The Modern Way: rules

rules is an "If/Then" system for jobs.

deploy-prod:
  stage: deploy
  script:
    - echo "Deploying..."
  rules:
    # ONLY run if we are on the 'main' branch
    - if: $CI_COMMIT_BRANCH == "main"
      when: manual      # Creates a button you must click!
    # Otherwise, do not include this job in the pipeline
    - when: never

---

2. Common Scenarios

A. Run only on Merge Requests

test-mr:
  rules:
    - if: $CI_PIPELINE_SOURCE == "merge_request_event"

B. Run when a specific file changes

build-docs:
  rules:
    - changes:
        - docs/**/*

(This is great for saving time; if you only changed code, don't waste time rebuilding the documentation).

---

3. The allow_failure Key

Sometimes you have a "New" test that is flaky. You don't want it to turn the whole pipeline red.

experimental-test:
  script: ./new-tests.sh
  allow_failure: true

The job will turn Orange (Warning) if it fails, but the pipeline will still continue to the next stage.

---

4. only and except (Legacy)

In older tutorials, you will see only: [main] instead of rules.

• Advice: GitLab is moving away from only/except. Always use rules for new projects; it is much more powerful and flexible.

---

Exercise: The Condition Lab

1. Create a job prod-deploy that only runs on the main branch and is manual.
2. Create a job beta-deploy that only runs on any branch EXCEPT main.
3. Modify your README.md. Did the beta-deploy job run?
4. Create a "Merge Request" (Pull Request) for your branch. Did the jobs change?
5. Why is the when: manual setting essential for a "Continuous Delivery" (not deployment) workflow?

---

Summary

You have completed Module 3: Building Your First Pipeline. You can now pass data, protect secrets, speed up builds with cache, and control exactly when your automation takes action.

Next Module: Pro level: Module 4: Advanced Pipeline Orchestration.