Observability

Observability interviews test whether you can understand production behavior from signals—not whether you memorized every Grafana panel. Strong answers connect logs, metrics, traces, alerting, and tools to user impact, SLOs, and incident response.

This domain spans logging (structured events, levels, correlation, PII safety), metrics (RED, USE, business KPIs, Prometheus basics), tracing (spans, trace IDs, OpenTelemetry), alerting (SLO, SLA, error budget, fatigue), and common tools (Grafana, Prometheus, Loki, Datadog, Sentry, New Relic).

On interviews: for a checkout outage, walk through which SLI you watch, which logs and traces you query, how correlation IDs tie frontend errors to API logs, and when you page vs ticket.

Common pitfalls: metrics without user-visible SLIs; logs with PII; tracing broken at queues; paging on log volume; buying APM without instrumentation standards.

The trade-off is observability coverage versus cost, cardinality, privacy, and on-call sustainability.

Checklist:

• Define golden signals per critical journey.
• Correlate logs, metrics, and traces with shared IDs.
• Alert on symptoms and SLO burn.
• Use child topics for signal-type depth.