Jupyter Starter

Set up a VirtualBox VM with a Jupyter Labs server and basic Python3 Data Science libraries.

Requirements

Building the VM requires VirtualBox and Vagrant

sudo apt-get install virtualbox

gets you the former. http://vagrant.io gets you the latter.

Building the VM

vagrant up

Depending on various factors (moon, tailwind, internet provider) provisioning and starting the VM takes between 8 and 20 minutes on my laptop.

Edit Vagrantfile to customize which packages get installed at VM build time.

Running the Jupyter Labs server

With a VM running,

vagrant ssh
cd /vagrant
./runserver

The latter will report out a URL that includes a login token. Copy that URL and paste it into a browser to get to the Jupyter UI.

By default, runserver will use the directory ./notebooks for notebooks. This directory is in .gitignore, so that you can keep it in a separate git repository. If you would rather keep everything in one repo, remove the entry in .gitignore.

To specify an alternate notebook directory, use

./runserver path/to/notebooks

Stopping the Jupyter Labs server

From a browser, select File / Shut Down.

Or, from inside the VM, ^C stops the server.

Stopping and restarting the Virtual Machine

From outside the VM, vagrant suspect will suspend a VM, and vagrant resume will reactivate it.

Alternatively, vagrant halt to stop it entirely, and vagrant up to bring it back to life.

Do one of these before you reboot your laptop/server. VirtualBox is fairly intolerant of being stopped suddenly, and you might have to vagrant destroy / vagrant up to provision a fresh VM.

License

None. There's nothing original here.