Exporting Multiple Docker Images Alongside Docker-Compose

It is possible to export a Docker container for use on another machine.

However, what if you have two containers locally that are connected with docker-compose, and you want to export them together?

Unfortunately, you can’t quite do that. However, exporting them both together with your compose file is not that difficult.

Exporting

Let’s look at an example. Say I have a project/ directory, with two subdirectories: project/frontend/ and project/backend/, and each has its own Dockerfile. A project/docker-compose.yml that connects them together might look like this:

version: "3.2"
services:
  backend:
    build: ./backend
    volumes:
      - ./backend:/app/backend
    ports:
      - "8000:8000"
    stdin_open: true
    tty: true
    command: python3 manage.py runserver 0.0.0.0:8000
  frontend:
    build: ./frontend
    volumes:
      - ./frontend:/app
      # One-way volume to use node_modules from inside image
      - /app/node_modules
    ports:
      - "3000:3000"
    environment:
      - NODE_ENV=development
    depends_on:
      - backend
    stdin_open: true
    command: yarn start

Here, docker-compose has defined several important aspects about the project, including where the code is located. If I run docker-compose up, it will start both environments. It will refer to the two Dockerfiles in their subdirectories, and then will create containers accordingly. If you open Docker Desktop, you should see the two containers running together under a section named after the folder your .yml file is in.

For example. My directory is called canoe_docker.

Now you can export the running containers to .tar files. To do so, first look at the running processes with docker ps. For example, this is what my terminal prints out:

CONTAINER ID        IMAGE                   COMMAND                  CREATED             STATUS              PORTS                    NAMES
1b7cf2821d4e        canoe_docker_frontend   "docker-entrypoint.s…"   4 hours ago         Up 5 minutes        0.0.0.0:3000->3000/tcp   canoe_docker_frontend_1
902987a44d0a        canoe_docker_backend    "python3 manage.py r…"   4 hours ago         Up 5 minutes        0.0.0.0:8000->8000/tcp   canoe_docker_backend_1

Then commit them with the following commands. Make sure to change it to your own container_id, and your own name for the image.

docker commit 902987a44d0a my_backend
docker commit 1b7cf2821d4e my_frontend

Next, save them to a tarball (this will take a while). The tarball can have any name you like.

docker save my_backend > backend.tar
docker save my_frontend > frontend.tar

Finally, you need a new docker-compose.yml. This only needs to be changed to reflect that you are no longer using directories, but Docker images. We will change the build argument, and remove the volumes.

version: "3.2"
services:
  backend:
    image: my_backend
    ports:
      - "8000:8000"
    stdin_open: true
    tty: true
    command: python3 manage.py runserver 0.0.0.0:8000
  frontend:
    image: my_frontend
    ports:
      - "3000:3000"
    environment:
      - NODE_ENV=development
    depends_on:
      - backend
    stdin_open: true
    command: yarn start

So now you have three files to export:

backend.tar
frontend.tar
docker-compose.yml  # This should be the updated one.

Importing

Once you get your files to a new machine, spinning up Docker is trivial! Make sure that you are in the right directory, then load the tarballs into Docker (this will take a loooong time):

docker load --input frontend.tar && docker load --input backend.tar

Finally, run them.

docker-compose up