Skip to content

Latest commit

 

History

History
59 lines (40 loc) · 2.51 KB

How-to-setup-a-JupyterHub-server-on-Azure-for-using-Devito-in-a-classroom-environment.md

File metadata and controls

59 lines (40 loc) · 2.51 KB
Raw

The purpose of this workflow is to set up a JupyterHub server on Azure for using Devito in a classroom environment. In this setup, we prepare a server where each user logs-in with GitHub credentials and has a PWD where Devito is installed. This solution delivers access to shared resources.

A lot of resources and how-to tutorials on how to set up a JupyterHub for your application on Azure can be found out there.

What we do:

Organizer/Educator workflow

Step 1: Setup TLJH on our VM. We follow the link here to set up our VM: http://tljh.jupyter.org/en/latest/install/azure.html Setup includes: Installing The Littlest JupyterHub¶, adding users, install conda / pip packages for all users

  • Note here: Depending on the number of users expected to connect we may need around 560M of disk size per user. So it is recommended that we increase disk size. (e.g. for around 20 users, 11G were occupied)

What we installed for the full Devito experience:

sudo apt-get install texlive-full # (notebooks rendering)
sudo apt-get install mpich libmpich-dev # (for those brave for MPI)

All the rest were already in the JupyterHub

Step 2: Add user authentication (via Github)

In order to add GitHub authentication for the new users we follow the approach presented here: http://tljh.jupyter.org/en/latest/howto/auth/github.html

(To reset authentication follow: http://tljh.jupyter.org/en/latest/howto/auth/firstuse.html#howto-auth-firstuse)

Step 3: Add the init-setup script for each user to the end of /etc/skel/.bashrc.

# e.g. for Transform 2020
git clone https://github.com/devitocodes/devito.git # Clone latest Devito master
git clone https://github.com/devitocodes/transform2020.git # Clone Devito tutorial
cd devito && pip install --user -e . # Install Devito
pip install --user matplotlib # Matplotlib is needed for tutorials

Also have look: http://tljh.jupyter.org/en/latest/howto/content/share-data.html if you want to add a priori any data files.

Attendee / Student workflow

Each new user connects to the public IP, authenticates with Github credentials, and opens a new terminal from the JupyterHub environment.

Open: http://<PublicIP> and authenticate using GIT credentials
New -> Terminal

When the terminal is open, the script from /etc/skel/.bashrc is triggered and installs latest Devito master on user PWD. User can then enjoy the notebooks.

Keep in mind:

  • We have seen company-firewall protected laptops not being able to pop-up the terminal window.
  • User install with pip install --user ..pkg-name..