Notes, files, ... around the deployment of JupyterHub instances with various math software

So far those are some notes we took with Thierry Dumont while playing around with JupyterHub and Sage.

Basic installation on Ubuntu 15.04

• Follow the instructions on https://github.com/jupyter/jupyterhub, replacing pip by sudo pip3 in each command

• sudo pip3 install zmq complains with:

  build/temp.linux-x86_64-3.4/scratch/vers.c:4:17: fatal error: zmq.h: Aucun fichier ou dossier de ce type
  #include "zmq.h"
  

  but seems happy after ...

Installing and configuring a jupyterhub with single user docker spawner

• Ugrade to Docker 1.6 and to a recent Docker Python client:

  curl -sSL https://get.docker.com/ | sh
  pip3 install --upgrade docker.py
  

  Rationale:

  • The Python Docker client shiped with Ubuntu 15.04 (package docker-python) is not recent enough for jupyterhub/dockerspawner; so it needs to be upgraded with:

    pip3 install --upgrade docker.py

  • This recent docker client requires Docker 1.6 while Ubuntu 15.04 ships Docker 1.5.

  To test the compatibility between Docker and the Docker Python client:

  sudo python3
  from docker import Client
  c = Client(base_url='unix://var/run/docker.sock')
  c.containers()
  

• Follow the instructions on https://github.com/jupyter/dockerspawner:

  git clone [email protected]:jupyter/dockerspawner.git	
  
  cd dockerspawner
  sudo pip3 install -r requirements.txt
  sudo python3 setup.py install
  
  --> does not work for me (td) on Ubuntu 14.04 LTS:
  cd dockerspawner
  python3 setup.py install
  <---- seems ok.
  

• Get the docker image (virtual size: 2.9 Go!)

  sudo docker pull jupyter/singleuser
  

• Create a JupyterHub configuration file:

  jupyterhub --generate-config
  

• Edit jupyterhub_config.py and add:

  c.JupyterHub.spawner_class = 'dockerspawner.DockerSpawner'
  

• Variant for systemuser:

  sudo docker pull jupyter/systemuser
  
  c.JupyterHub.spawner_class = 'dockerspawner.DockerSpawner'
  

• Current status:

  Once authenticated on jupyterhub, and redirected to the jupyter
  web server spawned within the singleuser docker container, we
  get a "505 internal server error".
  

Adding the Sage kernel into a system-wide jupyter

This is a quick hack to achieve it if you have run sage -notebook=ipython:

cp -r ~/.sage/ipython-3.1.0/kernels/* /usr/local/share/jupyter/kernels/

Change ownership of /usr/local/share/jupyter/kernels/ to root:root



Actually, kernels/* are:

doc: a symbolic link to  /usr/local/sage/src/doc/output/html/en

logo-64x64.png:  a symbolic link to /usr/local/sage/local/share/sage/ext/notebook-ipython/logo-64x64.png

logo.svg a symbolic link to /usr/local/sage/local/share/sage/ext/notebook-ipython/logo.svg

Change "/usr/local/sage" to the correct path.

kernel.json:  an example:

{"display_name": "Sage 6.8.beta2", "argv": ["/usr/local/sage/sage",
"-python", "-m", "sage.repl.ipython_kernel", "-f",
"{connection_file}"]}moi@kepler:/usr/local

Documentation for "kernels" in Ipython3 is at:
https://ipython.org/ipython-doc/dev/development/kernels.html