https://github.com/scionoftech/mlflow-docker-s3
MLflow setup using Docker and AWS S3
https://github.com/scionoftech/mlflow-docker-s3
mlflow mlflow-docker mlflow-server mlflow-tracking-server
Last synced: 7 months ago
JSON representation
MLflow setup using Docker and AWS S3
- Host: GitHub
- URL: https://github.com/scionoftech/mlflow-docker-s3
- Owner: scionoftech
- Created: 2021-10-04T08:19:45.000Z (over 4 years ago)
- Default Branch: main
- Last Pushed: 2021-10-04T09:06:05.000Z (over 4 years ago)
- Last Synced: 2024-12-27T17:23:38.639Z (over 1 year ago)
- Topics: mlflow, mlflow-docker, mlflow-server, mlflow-tracking-server
- Language: Shell
- Homepage:
- Size: 5.86 KB
- Stars: 2
- Watchers: 2
- Forks: 1
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# 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:
```shell
$ 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
```shell
python mlflow_tracking.py
```
### References
- mlflow-docker - [Production ready docker-compose configuration for ML Flow with Mysql and Minio S3 Topics](https://github.com/Toumash/mlflow-docker)
- deploy-mlflow-with-docker-compose - [Track your machine learning experiences with MLflow easily deployed thanks to docker-compose](https://towardsdatascience.com/deploy-mlflow-with-docker-compose-8059f16b6039)