Use Docker-in-Docker

  • Tier: Free, Premium, Ultimate
  • Offering: GitLab.com, GitLab Self-Managed, GitLab Dedicated

Docker-in-Docker (dind) means your registered runner uses the Docker executor or the Kubernetes executor. The executor uses a container image of Docker, provided by Docker, to run your CI/CD jobs.

The Docker image includes all of the docker tools and can run the job script in context of the image in privileged mode.

Always pin a specific version of the image, like docker:24.0.5. If you use a tag like docker:latest, you have no control over which version is used. This action can cause incompatibility problems when new versions are released.

Use with Docker executor

You can use the Docker executor to run jobs in a Docker container.

The Docker daemon supports connections over TLS. Use TLS when possible. TLS is the default in Docker 19.03.12 and later and is supported by GitLab.com instance runners.

This task enables --docker-privileged, which effectively disables the container’s security mechanisms and exposes your host to privilege escalation. This action can cause container breakout. For more information, see runtime privilege and Linux capabilities.

To use Docker-in-Docker with TLS enabled:

  1. Install GitLab Runner.

  2. Register GitLab Runner from the command line. Use docker and privileged mode:

    sudo gitlab-runner register -n \
  --url "https://gitlab.com/" \
  --registration-token REGISTRATION_TOKEN \
  --executor docker \
  --description "My Docker Runner" \
  --tag-list "tls-docker-runner" \
  --docker-image "docker:24.0.5-cli" \
  --docker-privileged \
  --docker-volumes "/certs/client"
    • This command registers a new runner to use the docker:24.0.5-cli image (if none is specified at the job level). To start the build and service containers, it uses the privileged mode. If you want to use Docker-in-Docker, you must always use privileged = true in your Docker containers.
    • This command mounts /certs/client for the service and build container, which is needed for the Docker client to use the certificates in that directory. For more information, see the Docker image documentation.

    The previous command creates a config.toml entry similar to the following example:

    [[runners]]
  url = "https://gitlab.com/"
  token = TOKEN
  executor = "docker"
  [runners.docker]
    tls_verify = false
    image = "docker:24.0.5-cli"
    privileged = true
    disable_cache = false
    volumes = ["/certs/client", "/cache"]
  [runners.cache]
    [runners.cache.s3]
    [runners.cache.gcs]

  3. You can now use docker in the job script. Include the docker:24.0.5-dind service:

    default:
  image: docker:24.0.5-cli
  services:
    - docker:24.0.5-dind
  before_script:
    - docker info

variables:
  # When you use the dind service, you must instruct Docker to talk with
  # the daemon started inside of the service. The daemon is available
  # with a network connection instead of the default
  # /var/run/docker.sock socket. Docker 19.03 does this automatically
  # by setting the DOCKER_HOST in
  # https://github.com/docker-library/docker/blob/d45051476babc297257df490d22cbd806f1b11e4/19.03/docker-entrypoint.sh#L23-L29
  #
  # The 'docker' hostname is the alias of the service container as described at
  # https://docs.gitlab.com/ci/services/#accessing-the-services.
  #
  # Specify to Docker where to create the certificates. Docker
  # creates them automatically on boot, and creates
  # `/certs/client` to share between the service and job
  # container, thanks to volume mount from config.toml
  DOCKER_TLS_CERTDIR: "/certs"

build:
  stage: build
  tags:
    - tls-docker-runner
  script:
    - docker build -t my-docker-image .
    - docker run my-docker-image /script/to/run/tests

Docker-in-Docker with TLS disabled in the Docker executor

Sometimes there are legitimate reasons to disable TLS. For example, you have no control over the GitLab Runner configuration that you are using.

  1. Register GitLab Runner from command line. Use docker and privileged mode:

    sudo gitlab-runner register -n \
  --url "https://gitlab.com/" \
  --registration-token REGISTRATION_TOKEN \
  --executor docker \
  --description "My Docker Runner" \
  --tag-list "no-tls-docker-runner" \
  --docker-image "docker:24.0.5-cli" \
  --docker-privileged

    The previous command creates a config.toml entry similar to the following example:

    [[runners]]
  url = "https://gitlab.com/"
  token = TOKEN
  executor = "docker"
  [runners.docker]
    tls_verify = false
    image = "docker:24.0.5-cli"
    privileged = true
    disable_cache = false
    volumes = ["/cache"]
  [runners.cache]
    [runners.cache.s3]
    [runners.cache.gcs]

  2. Include the docker:24.0.5-dind service in the job script:

    default:
  image: docker:24.0.5-cli
  services:
    - docker:24.0.5-dind
  before_script:
    - docker info

variables:
  # When using dind service, you must instruct docker to talk with the
  # daemon started inside of the service. The daemon is available with
  # a network connection instead of the default /var/run/docker.sock socket.
  #
  # The 'docker' hostname is the alias of the service container as described at
  # https://docs.gitlab.com/ci/services/#accessing-the-services
  #
  DOCKER_HOST: tcp://docker:2375
  #
  # This instructs Docker not to start over TLS.
  DOCKER_TLS_CERTDIR: ""

build:
  stage: build
  tags:
    - no-tls-docker-runner
  script:
    - docker build -t my-docker-image .
    - docker run my-docker-image /script/to/run/tests

Use a Unix socket on a shared volume between Docker-in-Docker and build container

Directories defined in volumes = ["/certs/client", "/cache"] in the Docker-in-Docker with TLS enabled in the Docker executor approach are persistent between builds. If multiple CI/CD jobs using a Docker executor runner have Docker-in-Docker services enabled, then each job writes to the directory path. This approach might result in a conflict.

To address this conflict, use a Unix socket on a volume shared between the Docker-in-Docker service and the build container. This approach improves performance and establishes a secure connection between the service and client.

The following is a sample config.toml with temporary volume shared between build and service containers:

[[runners]]
  url = "https://gitlab.com/"
  token = TOKEN
  executor = "docker"
  [runners.docker]
    image = "docker:24.0.5-cli"
    privileged = true
    volumes = ["/runner/services/docker"] # Temporary volume shared between build and service containers.

The Docker-in-Docker service creates a docker.sock. The Docker client connects to docker.sock through a Docker Unix socket volume.

job:
  variables:
    # This variable is shared by both the DinD service and Docker client.
    # For the service, it will instruct DinD to create `docker.sock` here.
    # For the client, it tells the Docker client which Docker Unix socket to connect to.
    DOCKER_HOST: "unix:///runner/services/docker/docker.sock"
  services:
    - docker:24.0.5-dind
  image: docker:24.0.5-cli
  script:
    - docker version

Docker-in-Docker with proxy enabled in the Docker executor

You might need to configure proxy settings to use the docker push command.

For more information, see Proxy settings when using dind service.

Use with Kubernetes executor

You can use the Kubernetes executor to run jobs in a Docker container.

To use Docker-in-Docker with TLS enabled in Kubernetes:

  1. Using the Helm chart, update the values.yml file to specify a volume mount.

    runners:
  tags: "tls-dind-kubernetes-runner"
  config: |
    [[runners]]
      [runners.kubernetes]
        image = "ubuntu:20.04"
        privileged = true
      [[runners.kubernetes.volumes.empty_dir]]
        name = "docker-certs"
        mount_path = "/certs/client"
        medium = "Memory"

  2. Include the docker:24.0.5-dind service in the job:

    default:
  image: docker:24.0.5-cli
  services:
    - name: docker:24.0.5-dind
      variables:
        HEALTHCHECK_TCP_PORT: "2376"
  before_script:
    - docker info

variables:
  # When using dind service, you must instruct Docker to talk with
  # the daemon started inside of the service. The daemon is available
  # with a network connection instead of the default
  # /var/run/docker.sock socket.
  DOCKER_HOST: tcp://docker:2376
  #
  # The 'docker' hostname is the alias of the service container as described at
  # https://docs.gitlab.com/ci/services/#accessing-the-services.
  #
  # Specify to Docker where to create the certificates. Docker
  # creates them automatically on boot, and creates
  # `/certs/client` to share between the service and job
  # container, thanks to volume mount from config.toml
  DOCKER_TLS_CERTDIR: "/certs"
  # These are usually specified by the entrypoint, however the
  # Kubernetes executor doesn't run entrypoints
  # https://gitlab.com/gitlab-org/gitlab-runner/-/issues/4125
  DOCKER_TLS_VERIFY: 1
  DOCKER_CERT_PATH: "$DOCKER_TLS_CERTDIR/client"

build:
  stage: build
  tags:
    - tls-dind-kubernetes-runner
  script:
    - docker build -t my-docker-image .
    - docker run my-docker-image /script/to/run/tests

Docker-in-Docker with TLS disabled in Kubernetes

To use Docker-in-Docker with TLS disabled in Kubernetes, you must adapt the previous example to:

  • Remove the [[runners.kubernetes.volumes.empty_dir]] section from the values.yml file.
  • Change the port from 2376 to 2375 with DOCKER_HOST: tcp://docker:2375.
  • Instruct Docker to start with TLS disabled with DOCKER_TLS_CERTDIR: "".

For example:

  1. Using the Helm chart, update the values.yml file:

    runners:
  tags: "no-tls-dind-kubernetes-runner"
  config: |
    [[runners]]
      [runners.kubernetes]
        image = "ubuntu:20.04"
        privileged = true

  2. You can now use docker in the job script. Include the docker:24.0.5-dind service:

    default:
  image: docker:24.0.5-cli
  services:
    - name: docker:24.0.5-dind
      variables:
        HEALTHCHECK_TCP_PORT: "2375"
  before_script:
    - docker info

variables:
  # When using dind service, you must instruct Docker to talk with
  # the daemon started inside of the service. The daemon is available
  # with a network connection instead of the default
  # /var/run/docker.sock socket.
  DOCKER_HOST: tcp://docker:2375
  #
  # The 'docker' hostname is the alias of the service container as described at
  # https://docs.gitlab.com/ci/services/#accessing-the-services.
  #
  # This instructs Docker not to start over TLS.
  DOCKER_TLS_CERTDIR: ""
build:
  stage: build
  tags:
    - no-tls-dind-kubernetes-runner
  script:
    - docker build -t my-docker-image .
    - docker run my-docker-image /script/to/run/tests

Known issues with Docker-in-Docker

Docker-in-Docker is the recommended configuration, but you should be aware of the following issues:

  • docker-compose command: This command is not available in this configuration by default. To use docker-compose in your job scripts, follow the Docker Compose installation instructions.

  • Cache: Each job runs in a new environment. Because every build gets its own instance of the Docker engine, concurrent jobs do not cause conflicts. However, jobs can be slower because there’s no caching of layers. See Docker layer caching.

  • Storage drivers: By default, earlier versions of Docker use the vfs storage driver, which copies the file system for each job. Docker 17.09 and later use --storage-driver overlay2, which is the recommended storage driver. See Using the OverlayFS driver for details.

  • Root file system: Because the docker:24.0.5-dind container and the runner container do not share their root file system, you can use the job’s working directory as a mount point for child containers. For example, if you have files you want to share with a child container, you could create a subdirectory under /builds/$CI_PROJECT_PATH and use it as your mount point. For a more detailed explanation, see issue #41227.

    variables:
  MOUNT_POINT: /builds/$CI_PROJECT_PATH/mnt
script:
  - mkdir -p "$MOUNT_POINT"
  - docker run -v "$MOUNT_POINT:/mnt" my-docker-image