kaufi95/jupyter

Jupyterhub setup running with DockerSpwaner and nbgrader installed to singleuser servers

Important information

• The nbgrader extension is enabled by default for all users. This can be configured in the nbgrader_config.py file. This will be done at a later point.
• Sometimes the IP_ADDRESS variable in the docker-compose.yaml file needs to be changed to the IP address of the JupyterHub container instance. This is a known issue and will be fixed at a later point. The IP address can be found by running docker network inspect jupyterhub-network in the terminal when the hub is running or you can look it up on docker desktop. The IP address also needs to be changed in the nbgrader_config.py file of the singleuser image. The image needs to be removed from docker registry and be rebuilt afterwards!

Additional information

• The docker-compose.yaml contains two services:
  • lab: This is a singleuser JupyterLab instance including nbgrader extension
  • hub: This is the JupyterHub container instance, which is used to spawn the singleuser instances

The singleuser image is used to spawn new singleuser instances.

Requirements

• Docker

Installation

1. Clone this repository
2. Build the singleuser image by running this command: docker build -t jupyterlab-singleuser ./docker/lab
3. Start up the system by running: docker-compose up

Test workflow

This workflow shows the setup of the system and how it can be used. The workflow is based on the following scenario:

• An admin creates a course and adds an instructor to it
• The instructor adds students to the course
• The instructor creates an assignment and releases it
• The students fetch the assignment, complete it and submit it
• The instructor collects and grades the assignment and releases the feedback

1. run docker-compose up
2. go to localhost:8000
3. sign up a new users with usernames admin1, instructor1, user1, user2, user3
4. login with the admin1 user
5. create a new course by opening a terminal in the container and type cm create_course SP
6. add the instructor to the course by typing cm add_instructor SP instructor1 -f instructor1 -l instructor1 -e instructor1@mail
7. login with the instructor1 user
8. add the students to the course by typing cm add_student SP user1 -f user1 -l user1 -e user1@mail and cm add_student SP user2 -f user2 -l user2 -e user2@mail in a terminal in the container
9. create a file called nbgrader_config.py in the root directory and insert c.CourseDirectory.course_id = "SP"
10. create a folder called source. Insert a folder called SP and add a notebook to it
11. open formgrader and generate and release the assignment
12. login with the users user1 and user2 and fetch, complete and save the assignments
13. login with the user user3 -> this user should not be able to see or fetch the assignment
14. run docker exec -it jupyter-user1 /bin/bash and type rm -rf /home/user1/SP/.ipynb_checkpoints
15. run docker exec -it jupyter-user2 /bin/bash and type rm -rf /home/user2/SP/.ipynb_checkpoints
16. submit the assignments
17. login with the instructor1 user
18. collect and grade the assignment and generate and release the feedback
19. logout with all users
20. run docker compose down

Notes

To view the courses, assigned instructors and students, login with the admin1 user and head to the following links:

• http://localhost:8000/services/ngshare/courses
• http://localhost:8000/services/ngshare/instructors/SP
• http://localhost:8000/services/ngshare/students/SP

To see other available endpoints, head to http://localhost:8000/services/ngshare/ (not all endpoints are GET-endpoints)

TODO

• Fix issue with .ipynb_checkpoint files, so nbgrader does not throw an error when fetching assignments (LibreTexts/ngshare#161)
• Fix issue with fetching feedbach
• Fix issue with IP_ADDRESS variable in docker-compose.yaml file
• Fix issue with ip address in nbgrader_config.py file
• Fix issue with nbgrader extension being enabled by default for all users
• Fix error log (admin1)
• Remove GRANT_SUDO, so new packages can't be installed
• Restrict other ways to cheat / communicate with other students (pip, ...)
• Clean up code