Feature request: Observability (metrics)

[xBlaz3kx]

Since I can see there are many open issues with connection/compatibility issues and race conditions, maybe it would make sense to implement some metrics for central systems. I think it would be helpful to have real-time insights into a central system and do it in a standardized way.

There are two things that we could instrument: the websocket connection and the OCPPj layer. I'm particularly interested in the OCPPj layer instrumentation, as It would be useful in debugging OCPP-related issues, e.g. compatibility between a Charge point and a Central System/CPMS.

As to what format to expose the metrics in, I would suggest OpenTelemetry over Prometheus metrics, since it is being set as an industry standard, and you can easily convert/map from one format to another (using the OTel Collector for example).

I was thinking about:

• Currenctly active connections/clients (gauge)
• Connection ping/pong response time (per client)
• Message rate (inbound/outbound)

---

• Response success rate (inbound/outbound, per clientId, per message type, with error reason)
• Request/response time (per client, message type)

Let me know your thoughs.