Yuvi’s Guide to Building JupyterHub images#
This work is
licensed under a Creative Commons Attribution
4.0 International License.
Welcome to my first ever ‘book’ of any sort :) This is an attempt to extract knowledge that seems to be stuck in my head on building Docker images of various sorts for use with JupyterHubs. These are somewhat opinionated, but have worked well across various communities & institutions for many years now. Even if you do not want to adopt any of it wholesale, I hope there are enough parts you can pick and choose to use from here.
While the focus of this book is on building images for use with JupyterHub, everything here applies 100% to building images that work on mybinder.org or any other BinderHubs.
Scope#
The goal of this book is two-fold:
Help you pick the method of docker image building that is right for you, without requiring you to be an expert in containers.
Provide detailed how-to guides on using these methods to build and maintain images for interactive datascience use.
If you are already an expert in container technology, and the images you build will be maintained by others who are also experts in container technology, maybe this book is not for you.
Methods covered in this book#
The 4 methods I want to cover in this book are:
Using repo2docker
Using an unmodified community curated image, such as jupyter docker-stacks, pangeo images or the rocker images.
Inheriting from and modifying a community curated base image (an extension of (2))
Building your own image from scratch, with a
Dockerfile.
As with life, there are trade-offs in making this choice. This book’s primary objective is to help you make this choice, and follow-through with it.