CI Setup for application
Let’s say we have a Node.js API that retrieves a list of books in a database. We can create a pipeline that pushes our code through three phases: build, test and deploy. A pipeline is a group of steps that are grouped by similar characteristics. With those phases our pipeline is defined by three types:
Project Pipeline
Continuous Integration Pipeline
Deploy Pipeline
The Project Pipeline installs dependencies, runs linters and any scripts that deal with the code. The Continuous Integration Pipeline runs automated tests and builds a distributed version of the code. Finally, the Deploy Pipeline deploys code to a designated cloud provider and environment.
The steps that the three pipelines execute are called jobs. When you group a series of jobs by those characteristics it is called stages. Jobs are the basic building block for pipelines. They can be grouped together in stages and stages can be grouped together into pipelines. Here’s an example hierarchy of jobs, stages, and pipelines:
In this hierarchy, all three components are considered three different pipelines. The main bullets — build, test, and deploy are stages and each bullet under those sections are jobs. Let’s break this out into a GitLab CI/CD yamlfile.
Using GitLab CI/CD
To use GitLab CI/CD, create a file called .gitlab-ci.yml at the root of the project in your GitLab repository and add the following yaml:
As I mentioned earlier, GitLab CI/CD uses Runners to execute pipelines. We can define which operating system and predefined libraries we would want our Runner to be based off by using the image directive. In our instance, we will be using the latest version of Node.js for our Runner. The stagesdirective allows us to predefine a stage for the entire configuration. Jobs will be executed based off of the order listed in the stages directive. To learn more about stages you can view it here. The before_script directive is used to run a command before all jobs.
Now let’s start with our job dedicated to the Build stage. We are going to call this job build-min-code. In this job we want it to install dependencies and minify the code. We can start this off with using the script directive. The scriptdirective is a shell script that gets executed within the Runner. Then we are going to assign this job to the “build” stage. To assign a job to a stage, use the stagedirective.
Now we have a job associated with our Build stage and we are going to do that for our Test stage. Our test job is going to be called run-unit-testand we are going to use the npm script in our API to run a test npm test.
Finally, we are going to add a job to handle our Deploy stage: deploy-production, deploy-staging. In this instance, we are going to have two different jobs for deployment (staging and production). These jobs will reflect the same layout as our previous jobs but with a small change. Currently, all of our jobs are automatically set to be triggered on any code push or branch. We do not want to have that for when we deploy our code to staging and production. To prevent that from happening we use the only directive. The only directive defines the names of branches and tags for which the job will run. The job will look like the following:
In our deploy-staging job, the Runner will only execute it if there was a change to the develop branch and for deploy-production the masterbranch. Here is a screenshot below that shows a code push made to the master branch.
From this image all three stages and jobs are triggered with the exception of deploy-staging since the code push was to themaster branch. GitLab CI/CD comes with an intuitive interface to help show what jobs and stages are running and what errors are occurring in the midst of the build. Below is the final version of the .gitlab-ci.yaml file. If you wish to test this out yourself, here is the link to the example application.
Last updated
