Issue of docker client in build container not reaching a daemon

[RomainMendez]

Hi everyone, I'm troubleshooting my Drone CI pipeline using the given example, and I've found that my build container is trying to reach localhost. And while I observe the containers being brought up on the machine (and exposing ports 2375 and 2376), the "DOCKER_IP" value in "command.sh" is empty I'm not sure how to fix this.

[RomainMendez]

type: docker
name: default

steps:
  - name: integration testing
    image: quay.io/testcontainers/dind-drone-plugin
    environment:
      CI_WORKSPACE: "/drone/src"
      INTEGRATION_TEST: "YES"
    settings:
      # This specifies the command that should be executed to perform build, test and
      #  integration tests. Not to be confused with Drone's `command`:
      cmd: ./run_test.sh
      # This image will run the cmd with your build steps
      build_image: python:3.11.7-bookworm
      # Not mandatory; enables pre-fetching of images in parallel with the build, so may save 
      #  time:
      prefetch_images:
        - "tomsquest/docker-radicale:latest"
      # Not mandatory; sets up image name 'aliases' by pulling from one registry and tagging
      #  as a different name. Intended as a simplistic mechanism for using a private registry 
      #  rather than Docker Hub for a known set of images. Accepts a dictionary of
      #  private registry image name to the Docker Hub image that it is a substitute for.
      #  Note that all images are pulled synchronously before the build starts, so this is
      #  inefficient if any unnecessary images are listed.
      image_aliases:
        tomsquest/docker-radicale:latest: tomsquest/docker-radicale:latest
    volumes:
      - name: dockersock
        path: /var/run

# Specify docker:dind as a service
services:
- name: docker
  image: docker:dind
  privileged: true
  volumes:
  - name: dockersock
    path: /var/run

volumes:
- name: dockersock
  temp: {}

[RomainMendez]

So the IP field is empty, I've checked the machine under which it is running and I see a docker0 socket existing.

[RomainMendez]

After more debugging, it appears to be a bug in how testcontainers uses the daemon, as using the docker pluging directly seems to work just fine. There's an issue about this topic on the main repo

[RomainMendez]

So I found the issue, docker:dind by default uses TLS nowadays, so the plugin should in the doc force the usage of TLS and add the certs from the service, should I add a PR ?

[animavitis]

U do not have to use this plugin at all. Same this can be done by this:

kind: pipeline
name: default
steps:
  - name: integration testing
    image: python:3.11.7-bookworm
    privileged: true
    commands:
      - ./run_test.sh
    volumes:
      - name: docker
        path: /var/run/docker.sock

volumes:
  - name: docker
    host:
      path: /var/run/docker.sock

[rnorth]

Sorry for the extensive delay - this repo is not under active maintenance.

As per #32 I'm proposing that this repository be formally archived; if you need to continue using it you're welcome to fork, but it's become unviable for me or the Testcontainers team to maintain.