Operators have no way to see relay health at a glance. Checking events per second, active connections, or DB/Redis status requires SSH access and reading raw logs.
- Add cluster-aware in-memory counters for EPS, accepted/rejected events, and active WebSocket connections
- Add lightweight DB and Redis health checks
- Expose metrics via GET /admin/metrics as a Server-Sent Events (SSE) stream for live updates
- Track CPU and memory load
- Avoid heavy DB polling, use in-memory counters where possible
Operators have no way to see relay health at a glance. Checking events per second, active connections, or DB/Redis status requires SSH access and reading raw logs.