Use Docker to build Docker images

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

You can use GitLab CI/CD with Docker to create Docker images. For example, you can create a Docker image of your application, test it, and push it to a container registry.

To run Docker commands in your CI/CD jobs, you must configure GitLab Runner to support docker commands. This method requires privileged mode.

If you want to build Docker images without enabling privileged mode on the runner, you can use a Docker alternative.

Enable Docker commands in your CI/CD jobs

To enable Docker commands for your CI/CD jobs, you can use:

Use the shell executor

To include Docker commands in your CI/CD jobs, you can configure your runner to use the shell executor. In this configuration, the gitlab-runner user runs the Docker commands, but needs permission to do so.

  1. Install GitLab Runner.

  2. Register a runner. Select the shell executor. For example: 

    sudo gitlab-runner register -n \
  --url "https://gitlab.com/" \
  --registration-token REGISTRATION_TOKEN \
  --executor shell \
  --description "My Runner"

  3. On the server where GitLab Runner is installed, install Docker Engine. View a list of supported platforms.

  4. Add the gitlab-runner user to the docker group: 

    sudo usermod -aG docker gitlab-runner

  5. Verify that gitlab-runner has access to Docker: 

    sudo -u gitlab-runner -H docker info

  6. In GitLab, add docker info to .gitlab-ci.yml to verify that Docker is working: 

    default:
  before_script:
    - docker info

build_image:
  script:
    - docker build -t my-docker-image .
    - docker run my-docker-image /script/to/run/tests

You can now use docker commands (and install Docker Compose if needed).

When you add gitlab-runner to the docker group, you effectively grant gitlab-runner full root permissions. For more information, see security of the docker group.

Use Docker-in-Docker

“Docker-in-Docker” (dind) means:

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

You should use Docker-in-Docker with TLS enabled, which is supported by GitLab.com instance runners.

You should 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 can cause incompatibility problems when new versions are released.

Use the Docker executor with Docker-in-Docker

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

Docker-in-Docker with TLS enabled in the Docker executor

The Docker daemon supports connections over TLS. TLS is the default in Docker 19.03.12 and later.

caution
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" \
  --docker-image "docker:24.0.5" \
  --docker-privileged \
  --docker-volumes "/certs/client"
    • This command registers a new runner to use the docker:24.0.5 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"
    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. You should include the docker:24.0.5-dind service: 

    default:
  image: docker:24.0.5
  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/ee/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
  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.

Assuming that the runner’s config.toml is similar to: 

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

You can now use docker in the job script. You should include the docker:24.0.5-dind service: 

default:
  image: docker:24.0.5
  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/ee/ci/docker/using_docker_images.html#accessing-the-services
  #
  # If you're using GitLab Runner 12.7 or earlier with the Kubernetes executor and Kubernetes 1.6 or earlier,
  # the variable must be set to tcp://localhost:2375 because of how the
  # Kubernetes executor connects services to the job container
  # DOCKER_HOST: tcp://localhost:2375
  #
  DOCKER_HOST: tcp://docker:2375
  #
  # This instructs Docker not to start over TLS.
  DOCKER_TLS_CERTDIR: ""

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

Use the Kubernetes executor with Docker-in-Docker

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

Docker-in-Docker with TLS enabled in Kubernetes

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:
  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. You can now use docker in the job script. You should include the docker:24.0.5-dind service: 

    default:
  image: docker:24.0.5
  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.
  DOCKER_HOST: tcp://docker:2376
  #
  # The 'docker' hostname is the alias of the service container as described at
  # https://docs.gitlab.com/ee/ci/services/#accessing-the-services.
  # If you're using GitLab Runner 12.7 or earlier with the Kubernetes executor and Kubernetes 1.6 or earlier,
  # the variable must be set to tcp://localhost:2376 because of how the
  # Kubernetes executor connects services to the job container
  # DOCKER_HOST: tcp://localhost:2376
  #
  # 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
  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 example above 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:
  config: |
    [[runners]]
      [runners.kubernetes]
        image = "ubuntu:20.04"
        privileged = true

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

    default:
  image: docker:24.0.5
  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.
  DOCKER_HOST: tcp://docker:2375
  #
  # The 'docker' hostname is the alias of the service container as described at
  # https://docs.gitlab.com/ee/ci/services/#accessing-the-services.
  # If you're using GitLab Runner 12.7 or earlier with the Kubernetes executor and Kubernetes 1.6 or earlier,
  # the variable must be set to tcp://localhost:2376 because of how the
  # Kubernetes executor connects services to the job container
  # DOCKER_HOST: tcp://localhost:2376
  #
  # This instructs Docker not to start over TLS.
  DOCKER_TLS_CERTDIR: ""
build:
  stage: build
  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:

  • The 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

Use the Docker executor with Docker socket binding

To use Docker commands in your CI/CD jobs, you can bind-mount /var/run/docker.sock into the container. Docker is then available in the context of the image.

If you bind the Docker socket you can’t use docker:24.0.5-dind as a service. Volume bindings also affect services, making them incompatible.

To make Docker available in the context of the image, you need to mount /var/run/docker.sock into the launched containers. To do this with the Docker executor, add "/var/run/docker.sock:/var/run/docker.sock" to the Volumes in the [runners.docker] section.

Your configuration should look similar to this example: 

[[runners]]
  url = "https://gitlab.com/"
  token = RUNNER_TOKEN
  executor = "docker"
  [runners.docker]
    tls_verify = false
    image = "docker:24.0.5"
    privileged = false
    disable_cache = false
    volumes = ["/var/run/docker.sock:/var/run/docker.sock", "/cache"]
  [runners.cache]
    Insecure = false

To mount /var/run/docker.sock while registering your runner, include the following options: 

sudo gitlab-runner register -n \
  --url "https://gitlab.com/" \
  --registration-token REGISTRATION_TOKEN \
  --executor docker \
  --description "My Docker Runner" \
  --docker-image "docker:24.0.5" \
  --docker-volumes /var/run/docker.sock:/var/run/docker.sock

For complex Docker-in-Docker setups like Code Quality checks using Code Climate, you must match host and container paths for proper execution. For more details, see Improve Code Quality performance with private runners.

Enable registry mirror for docker:dind service

When the Docker daemon starts inside the service container, it uses the default configuration. You might want to configure a registry mirror for performance improvements and to ensure you do not exceed Docker Hub rate limits.

The service in the .gitlab-ci.yml file

You can append extra CLI flags to the dind service to set the registry mirror: 

services:
  - name: docker:24.0.5-dind
    command: ["--registry-mirror", "https://registry-mirror.example.com"]  # Specify the registry mirror to use
The service in the GitLab Runner configuration file

If you are a GitLab Runner administrator, you can specify the command to configure the registry mirror for the Docker daemon. The dind service must be defined for the Docker or Kubernetes executor.

Docker: 

[[runners]]
  ...
  executor = "docker"
  [runners.docker]
    ...
    privileged = true
    [[runners.docker.services]]
      name = "docker:24.0.5-dind"
      command = ["--registry-mirror", "https://registry-mirror.example.com"]

Kubernetes: 

[[runners]]
  ...
  name = "kubernetes"
  [runners.kubernetes]
    ...
    privileged = true
    [[runners.kubernetes.services]]
      name = "docker:24.0.5-dind"
      command = ["--registry-mirror", "https://registry-mirror.example.com"]
The Docker executor in the GitLab Runner configuration file

If you are a GitLab Runner administrator, you can use the mirror for every dind service. Update the configuration to specify a volume mount.

For example, if you have a /opt/docker/daemon.json file with the following content: 

{
  "registry-mirrors": [
    "https://registry-mirror.example.com"
  ]
}

Update the config.toml file to mount the file to /etc/docker/daemon.json. This mounts the file for every container created by GitLab Runner. The configuration is detected by the dind service. 

[[runners]]
  ...
  executor = "docker"
  [runners.docker]
    image = "alpine:3.12"
    privileged = true
    volumes = ["/opt/docker/daemon.json:/etc/docker/daemon.json:ro"]
The Kubernetes executor in the GitLab Runner configuration file

If you are a GitLab Runner administrator, you can use the mirror for every dind service. Update the configuration to specify a ConfigMap volume mount.

For example, if you have a /tmp/daemon.json file with the following content: 

{
  "registry-mirrors": [
    "https://registry-mirror.example.com"
  ]
}

Create a ConfigMap with the content of this file. You can do this with a command like: 

kubectl create configmap docker-daemon --namespace gitlab-runner --from-file /tmp/daemon.json
note
You must use the namespace that the Kubernetes executor for GitLab Runner uses to create job pods.

After the ConfigMap is created, you can update the config.toml file to mount the file to /etc/docker/daemon.json. This update mounts the file for every container created by GitLab Runner. The dind service detects this configuration. 

[[runners]]
  ...
  executor = "kubernetes"
  [runners.kubernetes]
    image = "alpine:3.12"
    privileged = true
    [[runners.kubernetes.volumes.config_map]]
      name = "docker-daemon"
      mount_path = "/etc/docker/daemon.json"
      sub_path = "daemon.json"

Known issues with Docker socket binding

When you use Docker socket binding, you avoid running Docker in privileged mode. However, the implications of this method are:

  • By sharing the Docker daemon, you effectively disable all the container’s security mechanisms and expose your host to privilege escalation. This can cause container breakout. For example, if a project ran docker rm -f $(docker ps -a -q), it would remove the GitLab Runner containers.
  • Concurrent jobs might not work. If your tests create containers with specific names, they might conflict with each other.
  • Any containers created by Docker commands are siblings of the runner, rather than children of the runner. This might cause complications for your workflow.

  • Sharing files and directories from the source repository into containers might not work as expected. Volume mounting is done in the context of the host machine, not the build container. For example: 

     docker run --rm -t -i -v $(pwd)/src:/home/app/src test-image:latest run_app_tests

You do not need to include the docker:24.0.5-dind service, like you do when you use the Docker-in-Docker executor: 

default:
  image: docker:24.0.5
  before_script:
    - docker info

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

Authenticate with registry in Docker-in-Docker

When you use Docker-in-Docker, the standard authentication methods do not work, because a fresh Docker daemon is started with the service. You should authenticate with registry.

Make Docker-in-Docker builds faster with Docker layer caching

When using Docker-in-Docker, Docker downloads all layers of your image every time you create a build. You can make your builds faster with Docker layer caching.

Use the OverlayFS driver

note
The instance runners on GitLab.com use the overlay2 driver by default.

By default, when using docker:dind, Docker uses the vfs storage driver, which copies the file system on every run. You can avoid this disk-intensive operation by using a different driver, for example overlay2.

Requirements

  1. Ensure a recent kernel is used, preferably >= 4.2.

  2. Check whether the overlay module is loaded: 

    sudo lsmod | grep overlay

    If you see no result, then the module is not loaded. To load the module, use: 

    sudo modprobe overlay

    If the module loaded, you must make sure the module loads on reboot. On Ubuntu systems, do this by adding the following line to /etc/modules: 

    overlay

Use the OverlayFS driver per project

You can enable the driver for each project individually by using the DOCKER_DRIVER CI/CD variable in .gitlab-ci.yml: 

variables:
  DOCKER_DRIVER: overlay2

Use the OverlayFS driver for every project

If you use your own runners, you can enable the driver for every project by setting the DOCKER_DRIVER environment variable in the [[runners]] section of the config.toml file: 

environment = ["DOCKER_DRIVER=overlay2"]

If you’re running multiple runners, you must modify all configuration files.

Read more about the runner configuration and using the OverlayFS storage driver.

Docker alternatives

To build Docker images without enabling privileged mode on the runner, you can use one of these alternatives:

Buildah example

To use Buildah with GitLab CI/CD, you need a runner with one of the following executors:

In this example, you use Buildah to:

  1. Build a Docker image.
  2. Push it to GitLab container registry.

In the last step, Buildah uses the Dockerfile under the root directory of the project to build the Docker image. Finally, it pushes the image to the project’s container registry: 

build:
  stage: build
  image: quay.io/buildah/stable
  variables:
    # Use vfs with buildah. Docker offers overlayfs as a default, but Buildah
    # cannot stack overlayfs on top of another overlayfs filesystem.
    STORAGE_DRIVER: vfs
    # Write all image metadata in the docker format, not the standard OCI format.
    # Newer versions of docker can handle the OCI format, but older versions, like
    # the one shipped with Fedora 30, cannot handle the format.
    BUILDAH_FORMAT: docker
    FQ_IMAGE_NAME: "$CI_REGISTRY_IMAGE/test"
  before_script:
    # GitLab container registry credentials taken from the
    # [predefined CI/CD variables](../variables/index.md#predefined-cicd-variables)
    # to authenticate to the registry.
    - echo "$CI_REGISTRY_PASSWORD" | buildah login -u "$CI_REGISTRY_USER" --password-stdin $CI_REGISTRY
  script:
    - buildah images
    - buildah build -t $FQ_IMAGE_NAME
    - buildah images
    - buildah push $FQ_IMAGE_NAME

If you are using GitLab Runner Operator deployed to an OpenShift cluster, try the tutorial for using Buildah to build images in rootless container.

Use the GitLab container registry

After you’ve built a Docker image, you can push it to the GitLab container registry.

Troubleshooting

Error: docker: Cannot connect to the Docker daemon at tcp://docker:2375

This error is common when you are using Docker-in-Docker v19.03 or later: 

docker: Cannot connect to the Docker daemon at tcp://docker:2375. Is the docker daemon running?

This error occurs because Docker starts on TLS automatically.

This error can also occur with the Kubernetes executor when attempts are made to access the Docker-in-Docker service before it has fully started up. For a more detailed explanation, see issue 27215.

Docker no such host error

You might get an error that says docker: error during connect: Post https://docker:2376/v1.40/containers/create: dial tcp: lookup docker on x.x.x.x:53: no such host.

This issue can occur when the service’s image name includes a registry hostname. For example: 

default:
  image: docker:24.0.5
  services:
    - registry.hub.docker.com/library/docker:24.0.5-dind

A service’s hostname is derived from the full image name. However, the shorter service hostname docker is expected. To allow service resolution and access, add an explicit alias for the service name docker: 

default:
  image: docker:24.0.5
  services:
    - name: registry.hub.docker.com/library/docker:24.0.5-dind
      alias: docker

Error: Cannot connect to the Docker daemon at unix:///var/run/docker.sock

You might get the following error when trying to run a docker command to access a dind service: 

$ docker ps
Cannot connect to the Docker daemon at unix:///var/run/docker.sock. Is the docker daemon running?

Make sure your job has defined these environment variables:

  • DOCKER_HOST
  • DOCKER_TLS_CERTDIR (optional)
  • DOCKER_TLS_VERIFY (optional)

You may also want to update the image that provides the Docker client. For example, the docker/compose images are obsolete and should be replaced with docker.

As described in runner issue 30944, this error can happen if your job previously relied on environment variables derived from the deprecated Docker --link parameter, such as DOCKER_PORT_2375_TCP. Your job fails with this error if:

Error: unauthorized: incorrect username or password

This error appears when you use the deprecated variable, CI_BUILD_TOKEN: 

Error response from daemon: Get "https://registry-1.docker.io/v2/": unauthorized: incorrect username or password

To prevent users from receiving this error, you should:

  • Use CI_JOB_TOKEN instead.
  • Change from gitlab-ci-token/CI_BUILD_TOKEN to $CI_REGISTRY_USER/$CI_REGISTRY_PASSWORD.

Error during connect: no such host

This error appears when the dind service has failed to start: 

error during connect: Post "https://docker:2376/v1.24/auth": dial tcp: lookup docker on 127.0.0.11:53: no such host

Check the job log to see if mount: permission denied (are you root?) appears. For example: 

Service container logs:
2023-08-01T16:04:09.541703572Z Certificate request self-signature ok
2023-08-01T16:04:09.541770852Z subject=CN = docker:dind server
2023-08-01T16:04:09.556183222Z /certs/server/cert.pem: OK
2023-08-01T16:04:10.641128729Z Certificate request self-signature ok
2023-08-01T16:04:10.641173149Z subject=CN = docker:dind client
2023-08-01T16:04:10.656089908Z /certs/client/cert.pem: OK
2023-08-01T16:04:10.659571093Z ip: can't find device 'ip_tables'
2023-08-01T16:04:10.660872131Z modprobe: can't change directory to '/lib/modules': No such file or directory
2023-08-01T16:04:10.664620455Z mount: permission denied (are you root?)
2023-08-01T16:04:10.664692175Z Could not mount /sys/kernel/security.
2023-08-01T16:04:10.664703615Z AppArmor detection and --privileged mode might break.
2023-08-01T16:04:10.665952353Z mount: permission denied (are you root?)

This indicates the GitLab Runner does not have permission to start the dind service:

  1. Check that privileged = true is set in the config.toml.
  2. Make sure the CI job has the right Runner tags to use these privileged runners.

Error: cgroups: cgroup mountpoint does not exist: unknown

There is a known incompatibility introduced by Docker Engine 20.10.

When the host uses Docker Engine 20.10 or newer, then the docker:dind service in a version older than 20.10 does not work as expected.

While the service itself will start without problems, trying to build the container image results in the error: 

cgroups: cgroup mountpoint does not exist: unknown

To resolve this issue, update the docker:dind container to version at least 20.10.x, for example docker:24.0.5-dind.

The opposite configuration (docker:24.0.5-dind service and Docker Engine on the host in version 19.06.x or older) works without problems. For the best strategy, you should to frequently test and update job environment versions to the newest. This brings new features, improved security and - for this specific case - makes the upgrade on the underlying Docker Engine on the runner’s host transparent for the job.