A derived and lightened version of BinderHub for private repositories hosted on gitlab without kubernetes.

What is Booklab

Booklab allow you to BUILD and RUN a docker image using GitLab repository, then CONNECT with JupyterHub and that allows users to interact with the code and environment within a live JupyterHub instance.

Why Booklab ?

Binder is a wonderful tool but limited to public GitHub repository and associated with kubernetes in fact mainly google cloud. So an entry ticket to expensive for a small but smart jupyter on demand service.

How it works ?

Simple is beautiful

A web app supported by flask and the oauth plugin, authenticate and configure jupyterhub setup via redis A runner, checkout source code, build the docker image (repo2docker) and run jupyter. The philosophy is 1 runner > 1 notebook. To scale booklab, increase number of runners.

Booklab used Redis and Træfik to setup the service.

Booklab workflow for one click deployement :

1. The home page display the list of your project on gitlab configured server (via an oauth request)
2. You click on the repository to use (path + branch) and go to the deploy page
3. BL create a pair of ssh key, push the public version on gitlab for the repository selected and the private key in redis.
4. BL set the status of path:branch with the flag 'todo'
5. A runner select atomically a path:branch to deploy. Retrieve the private ssh key, clone the repo, build the docker image and run it.
6. Traefik detect a new label instanciate by docker container and configure the route http://pathbranch.local
7. The deploy page has displayed the logs in real time during step 3 to 6 and now redirect to the new notebook

Redis structure :

TYPE	key	description
SET	runners	set of runner (token) declared
STR	heartbeat:	expire each 30 seconds, and set by runner in waiting mode
HASH	status	key : value : current deployment status
LIST	log::	deployment logs
STR	token::	jupyter token authentication
HASH	conf	key:value shared from the webapp to the runner

Configuration is done with config.yml

Running a development instance:

Define the FQDN and TMPDIR environment variables, e.g.

export FQDN=localhost
export TMPDIR=/tmp

in the config.yaml file, define the values for gitlab/host, gitlab/consumer_key, gitlab/consumer_secret

Then start an instance with docker-compose:

docker-compose up

License:

GNU AFFERO GENERAL PUBLIC LICENSE

https://www.gnu.org/licenses/agpl-3.0.txt

Install:

1- Configure your DNS with the wildcard : *.example.com 2- Set TMPDIR and FQDN environment variable (ex TMPDIR=/tmp and FQDN=example.com) 3- docker network create traefik 4- docker-compose up

To improve number of runner: docker-compose scale runner=20

Credits :

• Author: Sebastien Campion sebastien.campion@inria.fr
• Font : Bangers