Implement Metrics Endpoint

[dominiquekleeven]

Introduce endpoints for service observability:

• A Health Check endpoint for basic service status
• A Prometheus Metrics endpoint for monitoring and alerting

---

Tasks

• Implement a /health endpoint to verify that the service is running
• Add a /metrics endpoint compatible with Prometheus scraping
• Determine and implement additional Prometheus metrics beyond FastAPI defaults (e.g. auth failures, endpoint-specific metrics, forecasting/training metrics)

---

Considerations

• Should the health check endpoint be publicly accessible?
  This needs to be decided. In many deployments, health checks are available publicly or to a load balancer but not necessarily to anyone else.
• Ensure that /metrics is secured or restricted to monitoring tools only.

[dominiquekleeven]

https://github.com/trallnag/prometheus-fastapi-instrumentator

Single line to expose fast api metrics, easy to add custom metrics as well.

Metrics to consider:

Training / Forecasting count and duration

Historical datapoint count, regressor count, regressor historical datapoint count as these impact performance.

And standard data such as memory usage, python processes