Chronoverse propagates distributed trace context across synchronous and asynchronous boundaries.
HTTP entry
The server middleware extracts incoming context, starts the request span, and keeps trace identity available for downstream calls and structured logging.
gRPC boundaries
Client interceptors inject trace metadata and server middleware extracts it before handling the RPC. Domain spans therefore remain connected to the original HTTP request or worker operation.
Kafka boundaries
Producers inject trace context into record headers. Consumers extract that context before creating processing spans, connecting an asynchronous event back to the transaction or worker that published it.
Worker-created work
When a worker emits a follow-up event, the current processing context is injected so the trace continues through multi-stage workflow, execution, logging, and analytics paths.
Export
Services and workers export OpenTelemetry traces, metrics, and logs to the bundled LGTM collector over OTLP gRPC.
Trace context helps explain a distributed operation. Replay safety, outbox state, and leases remain the authoritative mechanisms for correctness.