Add distributed tracing for analyzing performance

[mmurto]

Optionally configurable distributed tracing would be great to help debug slower requests.

[mnonnenmacher]

@mmurto Would the traceId entry added in #875 fulfill your requirements? Given that the MDC propagation works everywhere, all log lines caused by an API request should now contain the traceID.

[mmurto]

@mmurto Would the traceId entry added in #875 fulfill your requirements? Given that the MDC propagation works everywhere, all log lines caused by an API request should now contain the traceID.

Trace ID is a the first step there, thanks! Tracing also requires instrumenting functions (either automatically or manually) to create spans that use the same trace ID for the same call, and setting up the tracing provider that sends the traces to a tracing backend, like Grafana Tempo.

[mnonnenmacher]

Is this something that the server has to actively support or something that users could configure in their infrastructure without code changes?

[mmurto]

Is this something that the server has to actively support or something that users could configure in their infrastructure without code changes?

Code changes to the server are needed.