Distinguish between Tyk internal errors and APIs errors

Hello Tyk community,

I know that Tyk exposes Prometheus metrics for APIs, but how can we monitor that Tyk gateway, grpc server or pump themselves are not having errors? Do those components expose their own internal metrics?

Thank you!

If I am not wrong, I think our instrumentation documentation would be of help.

There is also a discussion on the forums about Open Telemetry and Distributed Tracing you may find useful

How can instrumentation give insights about errors from internals of the gateway or pump or grpc server? I need to distinguish API errors (generated from downstream services) from errors in the gateway or pump or from grpc server plugins. Also, I was more thinking about Prometheus metric I can set alerts on.

hi @scelentano! our upcoming OpenTelemetry integration (planed to be released end of September) will help you to quickly troubleshoot any issues - and help you understand where the issue comes from: upstream service, rate-limiting, authentication, middleware, …

You can analyse OTel distributed traces in any observability tools or alternatively generate metrics from the traces to export into Prometheus using the OpenTelemetry collector.

Here is an example of a distributed trace in Jaeger:

Does it help with what you have in mind?