Integration Testing of a Dockerized Angular App

We use docker to run our Protractor AngularJS integration tests. In this blog post I will explain our setup step by step and show how docker can be used to run tests.

By Johannes Schüth, on June 22, 2016


We are using the Protractor testing framework in order to test our Gentics Mesh AngularJS Admin UI webapp. Protractor is a great tool for end to end / integration testing of AngularJS Apps since it provides APIs which makes testing those Apps very easy.

Installation: npm install -g protractor

Selenium Node

Protractor uses Selenium in order to test the application.

The easiest way to setup a selenium node is to use the webdriver-manager tool but it is also possible to use selenium docker images.

Installation: npm install -g webdriver-manager

Our tests use the Firefox selenium node but it is also very easy to run tests using chrome.

The command docker run -d -p 4444:4444 selenium/standalone-firefox will spawn a new docker container which runs the selenium node.

Application Under Test

Next we need our AngularJS application. In our case this is Gentics Mesh. You can start the Gentics Mesh Demo Docker Container via docker run -p 8080:8080 gentics/mesh-demo. The started container will run the Gentics Mesh backend server and also serve the admin UI.


We have two test cases:

The login test will log in using the default demo credentials and verify that the demo project is being displayed.


The demo test will just access the demo endpoint which displays a simple AngularJS dummy webapp which displays the demo contents.


Both tests use browser.params in order to access params we specify during protractor startup.

The main protractor configuration file is currently set up to run both tests using firefox.


Finally it is time to execute the protractor tests.

You can run the tests locally using a local selenium node which can be started using the webdriver-manager start command. Once the selenium node and your Gentics Mesh application is ready you can invoke the protractor tests:

Combine all the things

Of course you could now glue ALL the pieces together using <$INSERT_YOUR_TOOL_HERE> here. In our case we use docker-compose and a little bit of bash for this.

I create a GitHub gist which contains all sources.

git clone protractor-docker-example
cd protractor-docker-example


Installation: described here

The /docker-compose.yml file contains three services:

  • tester: An image which contains the protractor tests. This will also execute these tests.

  • selenium: The selenium node which will be used by the tester service.

  • mesh: The application which will be tested. The selenium service must be able to access it.

The entrypoint value is customized in order to insert a bash script which will block execution until the needed service is ready. The /api/v1/admin/status endpoint is polled in order to verify that mesh is ready to accept requests. The selenium hub url is checked in a similar way.

A very simple docker file which adds the protractor configuration and test files to a new image which we will use to run our tests.


Test Runner

The test runner script is used to startup the docker-compose setup and wait until the tests have been executed. The test results will be printed to stdout and the exitcode of protractor will be passed along.

The following script is an adaptation of the script which was written by Harrison Harnisch. He also wrote an excellent blog post on integration testing using docker.


Example Output:

Finally all that is needed to do is to hook the script up to a continuous integration process. We use Jenkins CI and just needed to add another stage to our Jenkinsfile which describes our build pipeline in order to execute the tests. I would also recommend that you use environment variables within your docker-compose.yml file in order to configure and control which version of your App will be started.

Title Photo by Wesley Caribe via unsplash.

Thanks for reading our blog! Try our headless CMS Gentics Mesh for free!

Free Download