Skip to content

Multi-container applications with Docker Compose

Docker Compose is an extension of Docker which allows you to run multiple containers synchronously and in communication with one another.

Compose allows you to define and run a multi-container service using a Dockerfile and a docker-compose.yml.

Installing Compose

Compose v2 runs as a Docker CLI plugin: you type docker compose (with a space). It is included with Docker Desktop on macOS and Windows. On Linux, install the docker-compose-plugin package from Docker's package repository, for example on Ubuntu:

sudo apt-get install docker-compose-plugin
docker compose version

The older standalone docker-compose (v1) command is no longer maintained, so the commands on this page use docker compose.

Main advantages of Docker compose include:

  • Your applications can be defined in a YAML file where all the same options required in docker run are now defined (reproducibility).
  • It allows you to manage your application(s) as a single entity rather than dealing with starting individual containers (simplicity).

Note

For the sake of this example, either create your own Dockerfile or use the same Jupyter SciPy Notebook as in the Advanced Section

Creating a docker-compose.yml

Let's now create a Docker Compose .yml that calls Jupyter Notebook and RStudio

1. Create a folder shared_data in your current directory:

mkdir shared_data

2. Create an empty docker-compose.yml file (e.g., touch docker-compose.yml) and paste the following lines. Compose v2 ignores the top-level version: key found in older examples, so it is omitted here.

# All available services
services:

  # Computation
  jupyter:
    container_name: "jupyter_notebook"
    image: "jupyter/minimal-notebook"
    restart: "always"
    environment:
      - JUPYTER_TOKEN=mytoken
    user: root
    volumes:
      - ./data:/home/jovyan/work/
    ports:
      - 8888:8888

  rstudio:
    container_name: "rstudio"
    image: "rocker/rstudio"
    restart: "always"
    environment:
      - DISABLE_AUTH=true
    volumes:
      - ./data:/home/rstudio
    ports:
      - 8787:8787

3. Run both Jupyter Lab and RStudio using docker compose up instead of docker run.

Note

Handling containers with Docker Compose is fairly simple

docker compose up

attaches the volumes, opens ports, and starts the container

docker compose down

destroys the container

A brief explanation of docker-compose.yml is as below:

  • The web service builds from the Dockerfile in the current directory.
  • Forwards the container's exposed port to port 8888 on the host.
  • Mounts the project directory on the host to /work or /rstudio inside the container (allowing you to modify code without having to rebuild the image).
  • restart: always means that it will restart whenever it fails.

Running, shutting down, and restarting with docker compose

Run the containers with

$ docker compose up -d

To stop a running docker compose session, either press CTRL + C or use the command:

docker compose down

The above command removes containers, networks, volumes and images created by docker compose up.

To restart a container, use the command 

docker compose restart

restart will restart the Compose services without taking into account changes one may have made to the yml or environment.

Example using Docker Compose: WebODM

Warning

For the purpose of these following examples it is not suggested to use GitHub Codespaces.

Web Open Drone Map (WebODM)

OpenDroneMap is an open source photogrammetry toolkit to process aerial imagery into maps and 3D models running on command line. WebODM (Web OpenDroneMap) is an extension of ODM running on multiple Docker Containers provinding a user friendly web interface for easy visualization.

To use WebODM:

Prerequisites

WebODM requires docker and Docker Compose to function. Additionally, if you are on Windows, users will be required to have the Docker Windows Application installed as well as having the WSL2 (Windows Subsystem for Linux) operational.

  1. Ensure your machine is up to date: sudo apt-get update
  2. Clone the WebODM repository: git clone https://github.com/OpenDroneMap/WebODM --config core.autocrlf=input --depth 1
  3. Move into the WebODM folder: cd WebODM
  4. Run WebODM: sudo ./webodm.sh start
  5. The necessary docker images will be downloaded (~2 minutes) and WebODM will be accessible through http://localhost:8000/

Note

You will be asked to create an account as a formality. Add any username and a password and select Create Account.

6. Download example data: git clone https://github.com/OpenDroneMap/odm_data_aukerman.git. This git repository contains 77 HD images usable for WebODM. For other examples refer to ODMData.

7. In the WebODM portal, click on Select Images and GCP, navigate to odm_data_aukerman/images and select between 20-50 images (16 is the absolute minimum, whilst 32 is the suggested minimum).

webodm_1

8. WebODM will process the uploaded images (~5-10 minutes); upon completion, click View Map.

webodm_2

9. A map will open; you can click on 3D (bottom right) to see the 3D rendered model generated.

webodm_3

Adapted from CyVerse Container Camp (last source update 2022-05-12), CC BY 4.0. Spotted a problem? Open an issue.