GitLab Pipelines Basics

Introduction

Quick rundown of basics for GitLab pipelines. Pretty simple, you organize jobs into stages. The stages are ran consecutively, suppose you have the following stages

stages:
  - A
  - B
  - C

 job1:
   stage: A

 job2:
   stage: A

 job3:
   stage: B

 job4:
   stage: C

All jobs in stage A must be successful before any jobs in stage B can be ran. And a similar process applies for stage C. Now if you want to specify order within a stage you can use either needs: [job1, job2, ...] or dependencies: [job1, job2, ...]. The major difference between the two is use needs if you want to control the order of execution but use dependencies to control the order the artifacts of jobs are downloaded. I seem to use needs more often, but ymmv.

Workflows

Workflows control when a pipeline runs while rules control when a job runs.

workflow:
  rules:
    # Run pipeline for merge requests
    - if: '$CI_PIPELINE_SOURCE == "merge_request_event"'
      when: always
    # Run pipeline for for tags
    - if: '$CI_COMMIT_TAG' # I think you can use branch name t\infty
      when: always
    # If you want, but it is common to have pipelines run for merges to the default branch
    - if: '$CI_COMMIT_BRANCH == main' # can also use regex or $CI_DEFAULT_BRANCH
      when: always
    # This is generally recommended, but note that this denies run for
    # everything other than the stuff you specified above
    - when: never

 # ... Rest of file is ommited for conciseness ...

 # Make this job have stricter requirements
 make-commit:
   stage: last
   scripts: # make commit to manifest repo or whatever
   rules:
     # Only run on pushes to dev branch
     - if: '$CI_COMMIT_BRANCH == "dev" && $CI_PIPLINE_SOURCE == push'
       when: on_success
     - when: never