Your Java service is running. The dashboard looks normal. But some telemetry never arrives, and nothing clearly tells you why.

This session goes inside OpenTelemetry Java to examine how telemetry can fail in less obvious ways: asynchronous metric collection, configuration that does not behave as expected, and HTTP sending paths where reliability depends on details below application instrumentation. We will use real OpenTelemetry Java contribution work to show how to trace a suspicious symptom through SDK code, tests, configuration, and data flow.

This is not an OpenTelemetry introduction, a dashboard tour, or a vendor observability pitch. It is a Developer Tools session about debugging the tool that developers rely on to debug everything else.

Attendees will learn a practical way to reason about missing or unexpected telemetry, identify which layer to inspect, read unfamiliar Java infrastructure code, and turn a failure hypothesis into a focused test and fix.