How can "observability" (logs, metrics, traces) complement traditional testing in production environments, e.g. via "testing in production" or canary releases?

A Observability allows teams to detect issues that only manifest under real production conditions (scale, data, integrations) which pre-production testing cannot fully replicate, enabling fast detection and rollback B Observability replaces the need for any pre-release testing entirely C Observability tools can only be used after a project is fully decommissioned D Canary releases require disabling all monitoring during rollout