Kubernetes Health Checks Confirm Process Status, Not Business Outcomes

Kubernetes liveness, readiness, and startup probes are designed to answer narrow questions about process state — such as whether a container accepts traffic or has finished initializing — not whether the application is fulfilling its actual business purpose. A deployment can show all green health checks while a critical workflow like checkout silently fails, because probes never execute real customer-facing code paths. This gap exists by design: the orchestrator manages process lifecycles and has no mechanism to understand or verify business logic. The failure lies not in Kubernetes itself but in the assumption by architects that a green dashboard means business objectives are being met. Common false-positive patterns include readiness probes that pass while downstream dependencies are unreachable, database connection checks that succeed despite replication lag, and backup jobs that exit cleanly with unusable output.