What is a common mistake with load balancing and health in Ingress?

Relying only on pod existence rather than meaningful readiness can send traffic to an application that cannot serve requests.