Forespell data science stack
This repo contains (1) a Dockerfile with Jupyter notebook data science stack whose image is built by Docker Hub when master is updated, and (2) a script to launch the Docker container on a powerful AWS EC2 spot instance on https://notebook.forespell.com with an attached EBS volume for data persistence.
- Install the AWS CLI with
pip install awscli. Install your AWS credentials under
- Create an EFS in the region
eu-west-1in the AWS console and give assign it the
- Add an inbound rule to the
defaultsecurity group in the AWS console to allow traffic from port 443.
- Set the environment variable
CLOUDFLARE_API_KEY. On Linux, use
~/.bashrc, on macOS use
~/.bash_profileand on Windows use
IPython.lib.passwd()to generate a hash of the notebook password, and update the
NOTEBOOK_PASSWORDvariable with it.
Starting and stopping the notebook server
To start the notebook server, simply run
notebook.sh on Linux and macOS, or
notebook.bat on Windows. Then go to https://notebook.forespell.com. To add an exception for the self-signed certificate, see http://superuser.com/questions/632059/how-to-add-a-self-signed-certificate-as-an-exception-in-chrome.
To stop the notebook server, simply run
docker-machine rm supercomputer.