Observability Standard

Every service emits structured logs with timestamp, severity, service, environment, correlation identifier, and safe error context. Logs exclude secrets and sensitive payloads.

Services expose health and readiness signals meaningful to Kubernetes. Metrics cover request volume, errors, latency, saturation, external dependency behavior, and domain-specific outcomes. Distributed tracing is adopted for high-value cross-service paths as the platform matures.

A release is operationally ready when dashboards and alerts can distinguish application failure, dependency failure, capacity pressure, and deployment regression. Alerting is actionable and owned; noisy alerts are defects.