How would you answer an interview scenario involving Kubernetes observability and troubleshooting?

A Deployment has intermittent 503 errors after a rollout. Explain how you would investigate Pods, readiness, Services, ingress, events, resource pressure, and application traces.