A zoning-map microservice listens on its port but deadlocks and stops handling requests. Extra replicas alone do not clear the stuck processes. Which probe should restart the unhealthy containers?
Select an answer to reveal the explanation.
Short Explanation
Liveness asks "should we kill this and start fresh?" A deadlock that still holds the port needs that restart—not more replicas of the same stuck box. Do not point liveness at flaky collaborator checks, or healthy Pods restart during every upstream blip.
Full Explanation
Liveness decides whether the container should be killed and restarted. A deadlock that still accepts the TCP port is the classic case: readiness would only hide it from the Service, while the bad process keeps running. Avoid wiring liveness to dependency health—temporary collaborator failures then restart healthy Pods. Use liveness for process health; use readiness for traffic admission.