debian-mirror-gitlab/doc/ci/examples/deployment

stage	group	info	type
Release	Release	To determine the technical writer assigned to the Stage/Group associated with this page, see https://about.gitlab.com/handbook/engineering/ux/technical-writing/#assignments	tutorial

Using Dpl as deployment tool

Dpl (pronounced like the letters D-P-L) is a deploy tool made for continuous deployment that's developed and used by Travis CI, but can also be used with GitLab CI/CD.

Dpl can be used to deploy to any of the supported providers.

Requirements

To use Dpl you need at least Ruby 1.9.3 with ability to install gems.

Basic usage

Dpl can be installed on any machine with:

gem install dpl

This allows you to test all commands from your local terminal, rather than having to test it on a CI server.

If you don't have Ruby installed you can do it on Debian-compatible Linux with:

apt-get update
apt-get install ruby-dev

The Dpl provides support for vast number of services, including: Heroku, Cloud Foundry, AWS/S3, and more. To use it simply define provider and any additional parameters required by the provider.

For example if you want to use it to deploy your application to Heroku, you need to specify heroku as provider, specify api_key and app. All possible parameters can be found in the Heroku API section.

staging:
  stage: deploy
  script:
    - gem install dpl
    - dpl --provider=heroku --app=my-app-staging --api_key=$HEROKU_STAGING_API_KEY

In the above example we use Dpl to deploy my-app-staging to Heroku server with API key stored in HEROKU_STAGING_API_KEY secure variable.

To use different provider take a look at long list of Supported Providers.

Using Dpl with Docker

In most cases, you configured GitLab Runner to use your server's shell commands. This means that all commands are run in the context of local user (e.g. gitlab_runner or gitlab_ci_multi_runner). It also means that most probably in your Docker container you don't have the Ruby runtime installed. You must install it:

staging:
  stage: deploy
  script:
    - apt-get update -yq
    - apt-get install -y ruby-dev
    - gem install dpl
    - dpl --provider=heroku --app=my-app-staging --api_key=$HEROKU_STAGING_API_KEY
  only:
    - master

The first line apt-get update -yq updates the list of available packages, where second apt-get install -y ruby-dev installs the Ruby runtime on system. The above example is valid for all Debian-compatible systems.

Usage in staging and production

It's pretty common in the development workflow to have staging (development) and production environments

Let's consider the following example: we would like to deploy the master branch to staging and all tags to the production environment. The final .gitlab-ci.yml for that setup would look like this:

staging:
  stage: deploy
  script:
    - gem install dpl
    - dpl --provider=heroku --app=my-app-staging --api_key=$HEROKU_STAGING_API_KEY
  only:
    - master

production:
  stage: deploy
  script:
    - gem install dpl
    - dpl --provider=heroku --app=my-app-production --api_key=$HEROKU_PRODUCTION_API_KEY
  only:
    - tags

We created two deploy jobs that are executed on different events:

1. staging is executed for all commits that were pushed to master branch,
2. production is executed for all pushed tags.

We also use two secure variables:

1. HEROKU_STAGING_API_KEY - Heroku API key used to deploy staging app,
2. HEROKU_PRODUCTION_API_KEY - Heroku API key used to deploy production app.

Storing API keys

To add secure variables, navigate to your project's Settings > CI/CD > Variables. The variables that are defined in the project settings are sent along with the build script to the runner. The secure variables are stored out of the repository. Never store secrets in your project's .gitlab-ci.yml. It is also important that the secret's value is hidden in the job log.

You access added variable by prefixing it's name with $ (on non-Windows runners) or % (for Windows Batch runners):

1. $VARIABLE - use it for non-Windows runners
2. %VARIABLE% - use it for Windows Batch runners

Read more about the CI/CD variables.