Infrastructure (as Code) for DIY-ML-P
Everything needed to setup new version of DIY-ML-P.
TODO:
- Scripts for installation of Docker, docker-compose and other dependencies
- Scripts for creating folder structure
- Scripts for
- copying dagster
workspaces.yaml
template - creating a new feature store
- copying dagster
Requirements
- python & pip
- git
- dvc
- docker
- docker-compose
Folder Structure / Environment Variables
Prod environment assumes:
~/mlflow/backend/mlflow.sqlite
~/mlflow/artifacts/
~/dagster
Local environment assumes:
../../mlflow/backend/mlflow.sqlite
../../mlflow/artifacts/
../../dagster
Change as needed.
Usage
- dagster: exposed on port 8000
- mlflow: exposed on port 8001
Start:
docker-compose -f docker-compose.<ENV>.yaml up -d
Stop:
docker-compose -f docker-compose.<ENV>.yaml down