Fixing Drupal SimpleTest issues inside Docker Containers

5th May 2017

TL;DR You need to include the name of your web server container as the --url option to run-scripts.php.

I’ve been a Drupal VM user for a long time, but lately I’ve been using a combination Drupal VM and Docker for my local development environment. There were a couple of issues preventing me from completely switching to Docker - one of which being that when I tried running of my Simpletest tests, a lot of them would fail where they would pass when run within Drupal VM.

Here’s an excerpt from my docker-compose.yml file:

services:
  php:
    image: wodby/drupal-php:5.6
    volumes:
      - ./repo:/var/www/html

  nginx:
    image: wodby/drupal-nginx:7-1.10
    environment:
      NGINX_BACKEND_HOST: php
      NGINX_SERVER_ROOT: /var/www/html/web
    ports:
      - "80:80"
   volumes_from:
      - php
...

Nginx and PHP-FPM are running in separate containers, the volumes are shared across both and the Nginx backend is set to use the php container.

This is the command that I was using to run the tests:

$ docker-compose run --rm \
  -w /var/www/html/web \
  php \
  php scripts/run-tests.sh \
    --php /usr/local/bin/php \
    --class OverrideNodeOptionsTestCase

This creates a new instance of the php container, sets the working directory to my Drupal root and runs Drupal’s run-tests.sh script with some arguments. In this case, I'm running the OverrideNodeOptionsTestCase class for the override_node_options tests. Once complete, the container is deleted because of the --rm option.

This resulted in 60 of the 112 tests failing, whereas they all passed when run within a Drupal VM instance.

Test summary
------------

Override node options 62 passes, 60 fails, 29 exceptions, and 17 debug messages

Test run duration: 2 min 25 sec

Running the tests again with the--verbose option, I saw this message appear in the output below some of the failing tests:

simplexml_import_dom(): Invalid Nodetype to import

Update: I later found that https://www.drupal.org/docs/7/testing/running-tests-through-command-line#troubleshooting references this error message, but I didn’t see this page within my original search.

After checking that I had all of the required PHP extensions installed, I ran docker-compose exec php bash to connect to the php container and ran curl http://localhost to check the output. Rather than seeing the HTML for the site, I got this error message:

curl: (7) Failed to connect to localhost port 80: Connection refused

Whereas curl http://nginx returns the HTML for the page, so included it with the --url option to run-tests.sh, and this resulted in my tests all passing.

$ docker-compose run --rm \
  -w /var/www/html/web \
  php \
  php scripts/run-tests.sh \
    --php /usr/local/bin/php \
    --url http://nginx \
    --class OverrideNodeOptionsTestCase
Test summary
------------

Override node options 121 passes, 0 fails, 0 exceptions, and 34 debug messages

Test run duration: 2 min 31 sec

Note: In this example I have separate nginx and php containers, but I've tried and had the same issue when running Nginx and PHP-FPM in the same container - e.g. called app - and still needed to add --url http://app in order for the tests to run successfully.

I don’t know if this issue is macOS specfic (I know that Drupal CI is based on Docker, and I don’t know if it’s an issue) but I’m going to test also on my Ubuntu Desktop environment and investigate further and also compare the test run times for Docker in macOS, Docker in Ubuntu and within Drupal VM. I’m also going to test this with PHPUnit tests with Drupal 8.

Questions? Comments? I’m @opdavies on Twitter.

About the Author

Picture of Oliver

Oliver Davies is a Web Developer, System Administrator and Drupal specialist based in the UK. He is a Senior Developer at Microserve and also provides freelance consultancy services for Drupal, Symfony and Laravel applications and Linux servers.

Availability

  • Currently no spare part-time capacity.
  • Currently no spare full-time capacity.