DEV Community

loading...
Cover image for Using Docker Containers In Jobs - GitHub Actions

Using Docker Containers In Jobs - GitHub Actions

mihinduranasinghe profile image Mihindu Ranasinghe Updated on ・4 min read

👉 Prerequisites

If you are new to GitHub Actions, I suggest you to read these articles


As we discussed in previous articles, we can run our workflows in virtual machines by using operating systems that GitHub Actions provides as GitHub Hosted Runners.

Using Docker Container

We can use a docker container which will be installed on the the virtual machine & then instead of running directly on virtual machine, our workflow will runs on a particular container.

This can be an image from Docker Hub.


Docker Hub

Docker Hub is the platform which allows users to create, manage, and deliver their container applications. Docker Hub is the largest library of container images.

Link: Docker Hub

Alt Text

Once you click on an image (For example Node), you can see all the dockerfile links with the respective tags.

Alt Text


How to use ?

You have to provide the,
publisher-dockerhub-username : docker-image-tag with the image tag under the "container:"

  • We can not use containers on windows and mac-os virtual machines.
  • We can use containers only on ubuntu machines.

name: workflow-name
on: push

jobs:
  job-name:
    runs-on: ubuntu-latest
    container:
      image: dockerhub-username:docker-image-tag

Enter fullscreen mode Exit fullscreen mode

Example:


name: container
on: push

jobs:
  node-docker:
    runs-on: ubuntu-latest
    container:
      image: node:14.15.0-alpine3.12

Enter fullscreen mode Exit fullscreen mode

Also you can provide some other parameters like environment variables, ports & options...etc


name: container
on: push

jobs:
  node-docker:
    runs-on: ubuntu-latest
    container:
      image: node:14.15.0-alpine3.12

     #env:
     #  ENV1: Available to this container only

     #ports: To expose our container

     #options: --cpus 1 --host

Enter fullscreen mode Exit fullscreen mode

So in next figure you can confirm that steps are running inside the container we initialized with dockerhub image, instead of running directly on the virtual machine.


name: container
on: push

jobs:
  node-docker:
    runs-on: ubuntu-latest
    container:
      image: node:14.15.0-alpine3.12
    steps:
         - name: Log the node version
           run: |
             node -v
             cat /etc/os-release  


Enter fullscreen mode Exit fullscreen mode

Here we echo the node version of the image and some operating system information of the docker container.

Output:

Alt Text


We can also specify multiple containers to run as services.

If you have an application to run multiple services, you might need multiple images to run in our virtual machine. Normally we use Docker-Compose to do something like that, **but in GitHub Actions we have a similar way to run this type of applications.

Running multiple docker containers in a job.

Instead of having container here, we can provide "services: "
We can give the services as docker-images under the "services:" and they run as separate containers in our virtual machine.


name: workflow-name
on: push

jobs:
  job-name:
    runs-on: ubuntu-latest
    services:
      service-name-1:
        image: dockerhub-username:docker-image-1
        ports:
           - 3001:3000

      service-name-2:
        image: dockerhub-username:docker-image-2
        ports:
           - 4001:4000

    steps:
      - name: Get from service-1
        run: http://localhost:3001/api/users

      - name: Get from service-2
        run: http://localhost:4001/api/items


      - name: post a user example
        run: "curl -x POST http://localhost:3001/api/user -H 'Content-Type: application/json' -d '{\"username"\: \"hello"\, \"address"\: "sampleaddress"}' "


Enter fullscreen mode Exit fullscreen mode

Here it creates two containers for service-1 & service-2 inside our virtual machine

In order to communicate between those two containers, we can use the service-name as the host name in the app.


Running docker containers in individual steps.

In GitHub Actions, we can use different containers in different steps as well.

  • We can use "with:" key to provide inputs to this container in the particular step

  • We can give an entrypoint in with: key and override the entrypoint of the dockerfile

  • To find some entrypoints of the echo, node ...etc use this commands

Ex 1:

type -a  echo
Enter fullscreen mode Exit fullscreen mode
  • Output - Entrypoint of echo in ubuntu:
echo is  /bin/echo
Enter fullscreen mode Exit fullscreen mode

Ex 2 :

type -a  node
Enter fullscreen mode Exit fullscreen mode
  • Output - Entrypoint of node in ubuntu:
node is  usr/local/bin/echo
Enter fullscreen mode Exit fullscreen mode

Use this "type -a " command in ubuntu to find some Entrypoints of the installed tools & softwares.

Example:


name: container
on: push

jobs:
  node-docker:
    runs-on: ubuntu-latest
    container:
      image: node:14.15.0-alpine3.12
      steps:
         - name: Log the parent container node version
           run: |
             node -v 
           # Echo the node version of the parent container


         - name: Log the step container node version
           uses: docker://node:12.14.1-alpine3.10
           with:
             entrypoint: usr/local/bin/echo
             args: -v
            # Echo the node version of this step container based on this 12.14.1-alpine3.10 image


         - name: Step with docker to echo a string
           uses: docker://node:12.14.1-alpine3.10
           with:
             entrypoint: /bin/echo
             args: Hello Docker !


Enter fullscreen mode Exit fullscreen mode

Push the workflow file into the repository & check the Actions window of the repository

Outputs:

v14.15.0
v12.14.1
Hello Docker !

Enter fullscreen mode Exit fullscreen mode

👉 What's Next?

Thank You

Hope you all enjoyed and learned something from this. Let me know your comments and suggestions in the discussion section.

👉 Visit me - https://mihinduranasinghe.com/

Discussion (0)

pic
Editor guide