logging-tracing-design

Component logging + tracing — structured JSON log schema (timestamp RFC 3339, level, service, env, version, trace_id, span_id, tenant_id, error), OpenTelemetry spans + attributes (HTTP/DB/messaging/rpc/exception semantic conventions) + events + links, W3C Trace Context propagation (traceparent + tracestate + baggage) over HTTP/gRPC/async headers, OTLP export to collector (avoid vendor lock-in), head+tail sampling (100% errors + slow, 1-10% happy), log↔trace linking, PII redaction fail-closed, retention tiers (hot 24-72h / warm 30d / cold 90d-7y).