Incident Response with Metrics, Logs, and Traces
Incident response in an interview starts with customer impact, clear ownership, and a reversible mitigation before root-cause diagnosis. This interactive lesson works a storefront checkout latency incident using metrics, logs, and traces. Covers: incident severity, incident commander and service owner roles, rollback and feature flags, deployment versus dependency hypotheses, success checks, blameless follow-up. Practice: What should I do with these signals. Prerequisites: Observability basics.
A dashboard shows the fictional peak-season storefront has a sharp p95 latency increase on /checkout, more load-balancer 5xx responses, and normal application-task CPU. An interviewer asks for the first three moves. A strong answer does not name commands first. It states customer impact, assigns control, and chooses a reversible mitigation.
Customer impact, ownership, and mitigation
Let’s begin with a concise incident narrative: “Checkout ...