HerokuConda

A Dockerfile for a python3 miniconda setup tested on heroku based on continuumio/miniconda3 This container will run a miniconda Python 3 jupyter notebook with pandas scipy matplotlib and scikit-learn

---

Build and deploy

Deploying a docker images requires both docker and the heroku cli to be installed

To deploy docker images you need to login into the heroku registry (see heroku docs)

docker login --username=_ --password=$(heroku auth:token) registry.heroku.com

Next build and deploy your app, change <heroku_app_name> to your heroku app name

docker build . -t registry.heroku.com/<heroku_app_name>/web
docker push registry.heroku.com/<heroku_app_name>/web

Running locally

because locally you do not have the default environment variable PORT, you need to set it yourself

docker run -tip 8888:8888 -e PORT=8888 registry.heroku.com/<heroku_app_name>/web

---

note: deploying again destroys the file system, export your work before updating your app