MLflow-Docker-S3

Features

• MLflow in Docker container
• Mysql Docker container for MLflow tracking data
• Minio browser(https://min.io/) Docker container for artifacts.
• Nginx proxy Docker container for MLflow UI

How to setup

1. Clone the Repo

2. Update .env file with required details

3. Start the Setup by this one line:

   $ docker-compose up -d

4. Open up http://localhost:5000 for MlFlow, and http://localhost:9000 for S3 bucket (MLflow artifacts) with credentials from .env file

5. Configure MLflow client-side

For running mlflow files we need various environment variables set on the client side. To generate them use the script ./bashrc_install.sh, which installs it on your system.

$ ./bashrc_install.sh
[ OK ] Successfully installed environment variables into your .bashrc!

The script installs this variables: AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY, MLFLOW_S3_ENDPOINT_URL, MLFLOW_TRACKING_URI. All of them are needed to use mlflow from the client-side.

6. Test the MLflow setup for tracking and Artifacts in S3

python mlflow_tracking.py

References

• mlflow-docker - Production ready docker-compose configuration for ML Flow with Mysql and Minio S3 Topics
• deploy-mlflow-with-docker-compose - Track your machine learning experiences with MLflow easily deployed thanks to docker-compose