mlflowtoprom

A simple Rust service that receives MLflow metrics and exposes them in Prometheus format.

Usage

1. Send metrics (from TensorFlow/MLflow):

   curl -X POST http://localhost:8080/api/v2.0/mlflow/runs/MY_RUN_ID/log-metrics \
     -H "Content-Type: application/json" \
     -d '{"metrics": [{"key": "accuracy", "value": 0.95, "timestamp": 123, "step": 10}]}'
   

2. Scrape metrics (Prometheus):

   curl http://localhost:8080/metrics
   

   Output:

   # HELP mlflow_MY_RUN_ID_accuracy Metric from MLflow run MY_RUN_ID
   # TYPE mlflow_MY_RUN_ID_accuracy gauge
   mlflow_MY_RUN_ID_accuracy 0.95
   

3. Health checks (Kubernetes):

   curl http://localhost:8080/healthz  # Liveness probe
   curl http://localhost:8080/ready   # Readiness probe
   

Run

cargo run

Server listens on 0.0.0.0:8080.

Endpoints

Method	Endpoint	Description
POST	/api/v2.0/mlflow/runs/{run_id}/log-metrics	Accept MLflow metrics
GET	/metrics	Prometheus metrics
GET	/healthz	Kubernetes liveness
GET	/ready	Kubernetes readiness

Docker (optional)

FROM rust:1.70
WORKDIR /app
COPY . .
RUN cargo build --release
CMD ["./target/release/mlflowtoprom"]