Blog post: Dynamic image building #461
Description
Title
Dynamically build user environments on a JupyterHub: A community product update
Authors
- NEED TO FIGURE THIS OUT
Category
service
Tags
- open-source
- jupyter
- education
- geoscience
- product
- cloud
- bioscience
- hiring
- funding
Custom Tags
- jupyterhub
- binder
Blog Post Content
Summary
- We've developed an experimental feature to let users build user environments on the fly from their JupyterHub.
- This brings the environment generation and sharing functionality of Binder to a JupyterHub.
- It's still in-development and we're rolling it out to communities soon. Ask us if you want to try it out.
Context
- Binder allows you to generate and share a user image, but it doesn't have the concept of user identity or persistent storage.
- JupyterHub gives you a persistent space for your cloud workflows, but historically requires an admin to define the environment for everybody.
- This brings the best of both worlds together. Hub administrators will let their users create and share their own hub environments.
- This allows communities to define their environments in github repositories (similar to Binder) so that any jupyterhub with this feature can use it.
Where to learn more
- Builds on
jupyterhub-fancy-profilesproject - Involved a lot of contributions to the repo2docker documentation for environment configuration
- See 2i2c user docs here: https://docs.2i2c.org/user/environment/dynamic-imagebuilding/
Acknowledgements
- GESIS for original funding
- VEDA for fancy profiles
- Find attribution from historical posts and add
Additional Notes
Link to our historical blog posts about this so others can see its evolution over time.