Prerequesits

• Docker installed
• Jupyter installed

1 Build image

Required files to build the image:

• Dockerfile
• requirements.txt

docker build -t <name of image> .

2 Run jupyter image

Run built image with command

docker run -it --name <name of container> -p 8888:8888 <name image>

if there is no local image with the name, the Docker daemon searches and downloads image e.g. "jupyter/base-notebook" from Docker hub

3 Run Jupyter notebook

run Jupyter notebook e.g. through VS Code or in browser with http://localhost:8888/
(enter server ID / token, which is in Terminal output when container is started, is needed)

• interaction between container and jupyter notebooks
• new libraries can be installed either on
  • image, rebuild of image and container required
  • container, with jupyter notebook

4 Restart container next time

(1) restart container

docker start <containerID e.g. from Docker Desktop>

(2) get Jupyter URL from logs

docker logs <containerID e.g. from Docker Desktop>

(3) use Jupyter notebook with libraries

5 Optional: Dev containers

usage

• develop and execute code inside a container
• the file devcontainer.json tells the VS Code Dev Container extension how to create and access Dev Containers

commandpalette

• Dev Containers: Add Dev Container Configuration Files creates a new devcontainer.json template file and an optional Dockerfile, which both can be further adjusted
• Dev Containers: New Dev Container builds and starts new Dev container
• Dev Containers: Rebuild and Reopen in Container

Remarks

• specify libraries in requirements.txt
• build container again after changes
• cheat sheet for Docker CLI commands https://docs.docker.com/get-started/docker_cheatsheet.pdf