Docker environment with MLflow tracking server.
To run MLflow tracking server you can use prepared docker-compose.yml file orDockerfile.
This option additionally creates container with PostgreSQL database. You shouldset environment variables invariables section or use.env file. List ofrequired variable can be found in.env.default file.
If you don't need PostgreSQL database, it will be good option to choose Dockerfileoption. Just build and run. Don't forget about required variables – pass it asarguments.
| Name | Required | Description |
|---|
| ARTIFACT_ROOT | yes | Value for MLflow server --default-artifact-root param |
| HOST | yes | Value for MLflow server --host param |
| BACKEND_STORE_URI | yes | Value for MLflow server --backend-store-uri param |
| GOOGLE_APPLICATION_CREDENTIALS | no | Credentials for Google Cloud |
| POSTGRES_USER | yes – for docker-compose run option | PostgreSQL username |
| POSTGRES_PASSWORD | yes – for docker-compose run option | PostgreSQL password |
| POSTGRES_DB | yes – for docker-compose run option | PostgreSQL database name |