Jupyterhub on azure

This repository contains scripts for automatic creation of customised JupyterHub instances running on Azure cloud designed for multi-user classes. This doesn't (yet) use docker or elastic scaling. It's just a plan old server in the cloud -- albeit one that we can set up and tear down at will.

People can use it to set up their own servers -- no need to speak to the IT department if they don't want to.

Setting Up a new server

• Clone this repository
• Modify create_key.sh and run it once. This is for the SSL support
• Moidfy install.sh for the commands that will be run on the VM
• Modify and run create_vm.sh from your local machine. This creates the Azure VM

Useful sysadmin notes when using the resulting server

Adding accounts to sudoers list

When running your course, you may have classroom assistants or other trusted users who you may want to give full sudo access to. To add training_user2 to the sudoers list for example:

sudo usermod -aG sudo training_user2

You'll get the following error messages

sent invalidate(passwd) request, exiting
sent invalidate(group) request, exiting
sent invalidate(passwd) request, exiting
sent invalidate(group) request, exiting

These are nothing to worry about

Stopping and starting the JupyterHub service

The JupyterHub service can be stopped, started etc with the following commands

sudo systemctl daemon-reload
sudo systemctl start jupyterhub
sudo systemctl stop jupyterhub
sudo systemctl restart jupyterhub
sudo systemctl status jupyterhub

JupyterHub cloud installs elsewhere

If you don't like how this one works, you may like one of the following

The Data Science VM has JupyterHub pre-installed (and JupyterLab on the Ubuntu DSVM) – https://docs.microsoft.com/en-us/azure/machine-learning/data-science-virtual-machine/overview

Azure Lab Services - https://azure.microsoft.com/en-us/services/lab-services/