Running Jupyter notebooks inside ShinyProxy

Screenshots

ShinyProxy can run Jupyter notebooks using one of their official Docker images. However, you can adapt these Docker images, such that links inside the notebook are not opened in a new tab. See the official documentation. In addition, you may want to mount a volume for persisting your notebooks. In that case you need to tell the image to use the correct permissions for the volume. Both changes are applied in the openanalytics/shinyproxy-jupyter-datascience image. This image is built from the Dockerfile in this repo.

Building the Docker image

To pull the image made in this repository from Docker Hub, use

sudo docker pull openanalytics/shinyproxy-jupyter-datascience

The relevant Docker Hub repository can be found at https://hub.docker.com/r/openanalytics/shinyproxy-jupyter-datascience

To build the image from the Dockerfile, navigate into the root directory of this repository and run

sudo docker build -t openanalytics/shinyproxy-jupyter-datascience .

ShinyProxy Configuration

Note: ShinyProxy 2.6.0 or later is required for running Jupyter notebooks.

Create a ShinyProxy configuration file (see application.yml for a complete file)

proxy:
  specs:
    - id: jupyter-notebook-lab
      display-name: Jupyter Notebook Lab
      description: Jupyter Notebook running in lab mode.
      container-cmd: ["start-notebook.sh", "--NotebookApp.token=''", "--NotebookApp.base_url=#{proxy.getRuntimeValue('SHINYPROXY_PUBLIC_PATH')}"]
      container-image: openanalytics/shinyproxy-jupyter-datascience
      container-volumes: [ "/tmp/jupyter/#{proxy.userId}/work:/home/jovyan/work"]
      port: 8888
      websocket-reconnection-mode: None
      target-path: "#{proxy.getRuntimeValue('SHINYPROXY_PUBLIC_PATH')}"

Note: this will mount /tmp/jupyter/#{proxy.userId} as the workspace for storing the Jupyter notebooks. You can save your notebooks in the work directory. The next time you start the Jupyter application (i.e. after logging out), your files are available again.

Screenshots

(c) Copyright Open Analytics NV, 2021-2024.