Name specific patterns: circuit breakers, retry with backoff, graceful degradation. Show you think about what the user sees when a dependency is down.
Strong answers cover: circuit breakers, retries with exponential backoff and jitter, timeouts, fallback responses, graceful degradation, dead letter queues, and idempotent operations. Best candidates discuss the difference between transient and permanent failures, and how they decide which errors to retry versus surface to the user.
Tests production-readiness mindset. Developers who only handle the happy path create fragile systems. Those who design for failure build reliable ones. Ask: "What happens to your system if the payment provider is down for 30 minutes?"