Health Monitoring
This document describes the health endpoint and what to monitor.
Backend health endpoint
The backend exposes GET /health for liveness checks.
Response: JSON with basic status (e.g. {"status": "ok"}).
Use for:
- Load balancer health checks
- Kubernetes liveness/readiness probes
- Container orchestration
What to monitor
| Component | Check | Notes |
|---|---|---|
| Backend | GET /health |
Returns 200 when app is running |
| PostgreSQL | Connection from backend | Backend will fail if DB is down |
| Redis | Optional | If used, backend may degrade without it |
| Data360 MCP | Backend can reach MCP URL | Chat may fail for data tools if MCP is down |
| LLM provider | Backend can reach provider | Chat will fail if provider is unreachable |
Minimal health response
The default /health endpoint does not check database or Redis connectivity. It only confirms the FastAPI app is running. For stricter readiness, you can add custom checks (e.g. DB ping, Redis ping) to the health route.
Logging
Application errors and key events are logged. Use LOG_LEVEL to control verbosity. See Logging for details.