Simple base image for a static Jupyter notebook server.

This is meant to run under vagrant, so first do:

$ vagrant up

Then, login to the image:

$ vagrant ssh
$ cd /vagrant

The first time you provision the VM, you should also pull some of the base docker images:

$ docker pull ubuntu
$ docker pull jupyter/tmpnb
$ docker pull jupyter/configurable-http-proxy

Next, build the image:

docker build -t odewahn/jupyter-kernel .

You can start everything like this:

$ ./start.sh

Then, go to your browser and hit:

http://localhost:8000/

This spins up a new container and redirects you to it. It will look something like this:

http://localhost:8000/user-RgpNmthp8Oqr/

The websocket URL is then just this:

ws://localhost:8000/user-RgpNmthp8Oqr/

If you want to hack on just the kernel code (i.e., the container that actually starts a single notebook kernel)

From the /vagrant directory, start test container like this:

docker run -it -p 8888:8888 -v $(pwd):/usr/src  -w /usr/src odewahn/jupyter-kernel /bin/bash

This will map the code volume in the container all the way through to your host directory.

Then, run the program:

python jupyter-kernel.py

Go to localhost:8888 to see the output of the single kernel. It should just give a status code, like this:

{"status": "ok"}