JupyterDockerSpawnerOnSRC

Deploy JupyterHub with DockerSpawner on SURF Research Cloud (SRC).

The JupyterHub Docker Spawner enables JupyterHub to spawn single user notebook servers in Docker containers.

This document illustrates how to create a component, a catalog item and a workspace on SRC.

Create a component

Follow the SRC Create a component instructions and in the step Define component script and source, set the source as https://github.com/RS-DAT/JupyterDockerSpawnerOnSRC.git.

See each item below for more technical information about the component:

• research-cloud-plugin.yml for a list of packages that are installed.
• jupyterhub_config.py.j2 for a list of configs for jupyterhub and docker spawner.

Create a catalog item

Follow the SRC Create a catalog item and choose the components as below:

• SRC-OS
• SRC-CO
• SRC-Ngnix
• SRC-External plugin
• the JupyterDockerSpawnerOnSRC component created in the previous step

Note: When choosing the Operating system in the step Cloud provider, choose Ubuntu 20, because the latest version of Ubuntu does not support iRODS that is needed for SRC-CO component.

Create a workspace

Follow the SRC Start a simple workspace and choose the item that is created in the previous step.

Docker images and containers

The docker image can be set in the jupyterhub config file. After logging to Jupyter server, the docker image is pulled and a container starts running. The volumes /data and /scratch are mounted within the user home directory (~/data and ~/scratch) inside the container. If the user does not stop the server, the container continues running until a default timeout is reached (see Jupyter parameters here). Otherwise, the container will be stopped (Status Exited). Next time the user login, the same container will be used. So, the data and packages are preserved across sessions. Note that, the container and images will stay on the workspace. Only when the Jupyter service is stopped, the containers will be removed.

Remove images or containers

You can log in to the workspace using ssh and use docker command with sudo to delete images/containers.

Credits

Most of the playbooks in this repository have been taken, inspired or adapted from the ones in the RS-DAT JupyterDaskOnSRC and in the eWaterCycle infra.