Blog Engineering Implementing .gitlab-ci.yml
June 8, 2015
3 min read

Implementing .gitlab-ci.yml

We replacing GitLab CI jobs with a .gitlab-ci.yml and wanted to share the details of that process with you and would love to hear what you think.

Blog fallback hero

We wrote about why we're replacing GitLab CI jobs with a .gitlab-ci.yml file. As we've started on implementing this large change, we wanted to share the details of that process with you and would love to hear what you think.

To recap the previous article: currently you are required to write out your CI jobs in GitLab CI's interface. We're replacing this with a single file .gitlab-ci.yml, that you place in the root of your repository.

Schema change

Currently, on a push to GitLab, GitLab sends a web-hook to the CI Coordinator. The coordinator creates a build based on the jobs that are defined in its UI, which can then be executed by the connected Runners.

In the new schema, GitLab sends the web-hook and the .gitlab-ci.yml contents to the CI Coordinator, which creates builds based on the yml file. In turn, these builds are executed by the Runners as before.

Migrating to new style

Keeping two different ways of doing things would be a strain on development and support, not to mention confusing. So we're not just deprecating the old style of defining jobs, we're removing it entirely and will migrate existing jobs.

Upon upgrading your existing jobs defined in the GitLab CI Coordinator will be converted into a YAML file with the new syntax. You can download this file at any time from the project settings.

When the GitLab webhook triggers and doesn't transmit the content from .gitlab-ci.yml, the coordinator will use the converted YAML file instead.

This makes migrating to the new style very easy. You can start by simply copy-pasting the contents of the converted YAML file to the root of your repository. Existing projects will continue to build successfully, yet new projects do not have the option to use anything else.

An example .gitlab-ci.yml

To get an idea of how the .gitlab-ci.yml will look, we've prepared an example for a Ruby on Rails project (such as GitLab itself). Of course, this is due to change as we're still working on this.

# Refs to skip
skip_refs: “deploy*”

# Run before each script

# Refs to skip
skip_refs: “deploy*”

# Run before each script
  - export PATH=$HOME/bin:/usr/local/bin:/usr/bin:/bin
  - gem install bundler
  - cp config/database.yml.mysql config/database.yml
  - cp config/gitlab.yml.example config/gitlab.yml
  - touch log/application.log
  - touch log/test.log
  - bundle install --without postgres production --jobs $(nproc)
  - “bundle exec rake db:create RAILS_ENV=test”

# Parallel jobs, each line is a parallel build
  - script: “rake spec”
    runner: “ruby,postgres”
    name: “Rspec”
  - script: “rake spinach”
    runner: “ruby,mysql”
    name: “Spinach”
    tags: true
    branches: false

# Parallel deploy jobs
  - “cap deploy production”
  - “cap deploy staging”


Dmitriy and Sytse spend some time thinking about file syntax. Scripting should be simple and memorable. Thats why we come with better proposal:

  - gem install bundler
  - bundle install
  - bundle exec rake db:create

  test: "rake spec"
    - ruby
    - postgres
    - branches

  test: "rake spinach"
    - ruby
    - mysql
    - tags

  deploy: "cap deploy stating"
    - capistrano
    - debian
    - stable

    - cap deploy production
    - cap notify
    - capistrano
    - debian
    - master
    - /^deploy-.*$/


GitLab is nothing without its community. Contribute or follow the development in the GitLab CI repository.

We want to hear from you

Enjoyed reading this blog post or have questions or feedback? Share your thoughts by creating a new topic in the GitLab community forum. Share your feedback

Ready to get started?

See what your team could do with a unified DevSecOps Platform.

Get free trial

New to GitLab and not sure where to start?

Get started guide

Learn about what GitLab can do for your team

Talk to an expert