The premise
A request rarely stays inside one service. The hard part is preserving enough context to explain its path after it becomes asynchronous.
Working through it
Propagating a correlation identifier across API calls, messages, and background jobs creates a shared handle for logs, traces, and operational questions. It should be treated as part of the system contract, not an afterthought.
Current takeaway
Tracing starts with consistent context propagation. Tools become far more useful once the system provides a reliable thread to follow.