Jaeger Distributed Tracing Specialist

Trace context propagation maintains span continuity across async boundaries by extracting and injecting correlation IDs into message headers and event payloads. Use OpenTelemetry's context API to capture the active span before queuing work, then restore it when the consumer processes the message, ensuring end-to-end observability in Jaeger.

Enable automatic HTTP instrumentation during service initialization using HuleEdu conventions, then manually create spans for background tasks with standardized naming (e.g., 'api.submit_request', 'llm_provider.openai.call'). Attach relevant attributes and propagate trace context to maintain correlation across all service boundaries.

Missing spans typically indicate lost trace context during async handoffs or queue transitions. Check that correlation IDs are propagated in Kafka headers and Redis payloads, verify span naming follows conventions, and confirm the OTLP exporter is correctly initialized. Link traces with logs and metrics to pinpoint context loss.

Yes. Jaeger tracing supports both Kafka and Redis by capturing trace context before publishing events and restoring it in consumers. Use span propagation headers in Kafka messages and context storage in Redis to maintain trace chains across queue-based workflows and ensure observability for event-driven architectures.

Use hierarchical span names like 'api.submit_request' and 'llm_provider.openai.call' to organize operations by domain. Attach standardized attributes including service name, operation type, and outcome. This convention improves trace readability in Jaeger UI and enables consistent filtering and alerting across services.