distributed-tracing

Covers following a single request across service boundaries — context propagation, span and attribute design, sampling that keeps the traces worth keeping, and using traces to find where latency actually accumulates. Use this whenever the user debugs a slow request touching multiple services, instru