realaravinth/debian-mirror-gitlab
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity
debian-mirror-gitlab/doc/user/project/pages/getting_started_part_four.md
Pirate Praveen 45981217aa New upstream version 11.10.8+dfsg
2019-07-07 11:18:12 +05:30

11 KiB
Raw Blame History

last_updated author author_gitlab level article_type date
2018-02-16 Marcia Ramos marcia intermediate user guide 2017-02-22

Creating and Tweaking GitLab CI/CD for GitLab Pages

To get started with GitLab Pages, you can use one of the project templates, a .gitlab-ci.yml template, or fork an existing example project. Therefore, you don't need to understand all the ins and odds of GitLab CI/CD to get your site deployed. Still, there are cases where you want to write your own script or tweak an existing one. This document guides you through this process.

This guide also provides a general overview and clear introduction for getting familiar with the .gitlab-ci.yml file and writing one for the first time.

GitLab CI/CD serves numerous purposes, to build, test, and deploy your app from GitLab through Continuous Integration, Continuous Delivery, and Continuous Deployment methods. You will need it to build your website with GitLab Pages, and deploy it to the Pages server.

To implement GitLab CI/CD, the first thing you need is a configuration file called .gitlab-ci.yml placed at your website's root directory.

What this file actually does is telling the GitLab Runner to run scripts as you would do from the command line. The Runner acts as your terminal. GitLab CI/CD tells the Runner which commands to run. Both are built-in in GitLab, and you don't need to set up anything for them to work.

Explaining every detail of GitLab CI/CD and GitLab Runner is out of the scope of this guide, but we'll need to understand just a few things to be able to write our own .gitlab-ci.yml or tweak an existing one. It's an Yaml file, with its own syntax. You can always check your CI syntax with the GitLab CI Lint Tool.

Practical example

Let's consider you have a Jekyll site. To build it locally, you would open your terminal, and run jekyll build. Of course, before building it, you had to install Jekyll in your computer. For that, you had to open your terminal and run gem install jekyll. Right? GitLab CI + GitLab Runner do the same thing. But you need to write in the .gitlab-ci.yml the script you want to run so GitLab Runner will do it for you. It looks more complicated then it is. What you need to tell the Runner:

$ gem install jekyll
$ jekyll build

Script

To transpose this script to Yaml, it would be like this:

script:
  - gem install jekyll
  - jekyll build

Job

So far so good. Now, each script, in GitLab is organized by a job, which is a bunch of scripts and settings you want to apply to that specific task.

job:
  script:
  - gem install jekyll
  - jekyll build

For GitLab Pages, this job has a specific name, called pages, which tells the Runner you want that task to deploy your website with GitLab Pages:

pages:
  script:
  - gem install jekyll
  - jekyll build

The public directory

We also need to tell Jekyll where do you want the website to build, and GitLab Pages will only consider files in a directory called public. To do that with Jekyll, we need to add a flag specifying the destination (-d) of the built website: jekyll build -d public. Of course, we need to tell this to our Runner:

pages:
  script:
  - gem install jekyll
  - jekyll build -d public

Artifacts

We also need to tell the Runner that this job generates artifacts, which is the site built by Jekyll. Where are these artifacts stored? In the public directory:

pages:
  script:
  - gem install jekyll
  - jekyll build -d public
  artifacts:
    paths:
    - public

The script above would be enough to build your Jekyll site with GitLab Pages. But, from Jekyll 3.4.0 on, its default template originated by jekyll new project requires Bundler to install Jekyll dependencies and the default theme. To adjust our script to meet these new requirements, we only need to install and build Jekyll with Bundler:

pages:
  script:
  - bundle install
  - bundle exec jekyll build -d public
  artifacts:
    paths:
    - public

That's it! A .gitlab-ci.yml with the content above would deploy your Jekyll 3.4.0 site with GitLab Pages. This is the minimum configuration for our example. On the steps below, we'll refine the script by adding extra options to our GitLab CI.

Artifacts will be automatically deleted once GitLab Pages got deployed. You can preserve artifacts for limited time by specifying the expiry time.

Image

At this point, you probably ask yourself: "okay, but to install Jekyll I need Ruby. Where is Ruby on that script?". The answer is simple: the first thing GitLab Runner will look for in your .gitlab-ci.yml is a Docker image specifying what do you need in your container to run that script:

image: ruby:2.3

pages:
  script:
  - bundle install
  - bundle exec jekyll build -d public
  artifacts:
    paths:
    - public

In this case, you're telling the Runner to pull this image, which contains Ruby 2.3 as part of its file system. When you don't specify this image in your configuration, the Runner will use a default image, which is Ruby 2.1.

If your SSG needs NodeJS to build, you'll need to specify which image you want to use, and this image should contain NodeJS as part of its file system. E.g., for a Hexo site, you can use image: node:4.2.2.

Note: We're not trying to explain what a Docker image is, we just need to introduce the concept with a minimum viable explanation. To know more about Docker images, please visit their website or take a look at a summarized explanation here.

Let's go a little further.

Branching

If you use GitLab as a version control platform, you will have your branching strategy to work on your project. Meaning, you will have other branches in your project, but you'll want only pushes to the default branch (usually master) to be deployed to your website. To do that, we need to add another line to our CI, telling the Runner to only perform that job called pages on the master branch only:

image: ruby:2.3

pages:
  script:
  - bundle install
  - bundle exec jekyll build -d public
  artifacts:
    paths:
    - public
  only:
  - master

Stages

Another interesting concept to keep in mind are build stages. Your web app can pass through a lot of tests and other tasks until it's deployed to staging or production environments. There are three default stages on GitLab CI: build, test, and deploy. To specify which stage your job is running, simply add another line to your CI:

image: ruby:2.3

pages:
  stage: deploy
  script:
  - bundle install
  - bundle exec jekyll build -d public
  artifacts:
    paths:
    - public
  only:
  - master

You might ask yourself: "why should I bother with stages at all?" Well, let's say you want to be able to test your script and check the built site before deploying your site to production. You want to run the test exactly as your script will do when you push to master. It's simple, let's add another task (job) to our CI, telling it to test every push to other branches, except the master branch:

image: ruby:2.3

pages:
  stage: deploy
  script:
  - bundle install
  - bundle exec jekyll build -d public
  artifacts:
    paths:
    - public
  only:
  - master

test:
  stage: test
  script:
  - bundle install
  - bundle exec jekyll build -d test
  artifacts:
    paths:
    - test
  except:
  - master

The test job is running on the stage test, Jekyll will build the site in a directory called test, and this job will affect all the branches except master.

The best benefit of applying stages to different jobs is that every job in the same stage builds in parallel. So, if your web app needs more than one test before being deployed, you can run all your test at the same time, it's not necessary to wait one test to finish to run the other. Of course, this is just a brief introduction of GitLab CI and GitLab Runner, which are tools much more powerful than that. This is what you need to be able to create and tweak your builds for your GitLab Pages site.

Before Script

To avoid running the same script multiple times across your jobs, you can add the parameter before_script, in which you specify which commands you want to run for every single job. In our example, notice that we run bundle install for both jobs, pages and test. We don't need to repeat it:

image: ruby:2.3

before_script:
  - bundle install

pages:
  stage: deploy
  script:
  - bundle exec jekyll build -d public
  artifacts:
    paths:
    - public
  only:
  - master

test:
  stage: test
  script:
  - bundle exec jekyll build -d test
  artifacts:
    paths:
    - test
  except:
  - master

Caching Dependencies

If you want to cache the installation files for your projects dependencies, for building faster, you can use the parameter cache. For this example, we'll cache Jekyll dependencies in a vendor directory when we run bundle install:

image: ruby:2.3

cache:
  paths:
  - vendor/

before_script:
  - bundle install --path vendor

pages:
  stage: deploy
  script:
  - bundle exec jekyll build -d public
  artifacts:
    paths:
    - public
  only:
  - master

test:
  stage: test
  script:
  - bundle exec jekyll build -d test
  artifacts:
    paths:
    - test
  except:
  - master

For this specific case, we need to exclude /vendor from Jekyll _config.yml file, otherwise Jekyll will understand it as a regular directory to build together with the site:

exclude:
  - vendor

There we go! Now our GitLab CI not only builds our website, but also continuously test pushes to feature-branches, caches dependencies installed with Bundler, and continuously deploy every push to the master branch.

Advanced GitLab CI for GitLab Pages

What you can do with GitLab CI is pretty much up to your creativity. Once you get used to it, you start creating awesome scripts that automate most of tasks you'd do manually in the past. Read through the documentation of GitLab CI to understand how to go even further on your scripts.