Thanks to a yeoman effort by Walker Griggs (graduating senior and kubernetes savant), we are trying out a new service based on jupyterhub notebooks.

Currently, it supports minimal python 3 notebooks as well as text files through containers orchestrated via kubernetes with storage provided by a glusterfs federation. The entire system is built off of our local cloud instance using our production HPC resources.

Give it a try at: https://jupyter.colby.edu. Login with your Colby credentials, you start with a 10gig allocation, but please make sure to back your work up as this is a testing environment.

 

You can add python packages by opening a terminal, and typing

 

pip install --user <packagename>

by using –user the pip install puts the package in you 10gig of storage, which will persist when the container is recreated at next login or after 3 hours of inaacativity. The downside is that any running kernel won’t know about the install. To overcome this simply click on the running tab on the main window:

find the notebook you want to notice the installed package, click shutdown. Then reopen the notebook from the files menu. A new kernel will be instantiated and it will notice the addions you have made that are installed in your home directory.

 

 

If you have any comments/insights/complaints, send us a not at acits_research@colby.edu