Interview Prep Guide

Support Engineer Production Troubleshooting Interview Questions

Production support questions covering APIs, HTTP, logs, metrics, SQL diagnosis, escalation, privacy, incident response, and customer communication.

Applied Support Engineering Questions

  1. How do DNS, TCP, TLS, HTTP, and application errors differ during diagnosis?

    Diagnose in layers: name resolution, connection, certificate negotiation, protocol response, then application behavior, using evidence appropriate to each boundary.

  2. How do correlation IDs improve a support investigation?

    A correlation or request ID links one user action across gateway, service, queue, and database logs without relying only on timestamps or personal data.

  3. When is a read-only SQL query useful in production support?

    It can verify record state, relationships, timing, and scope when authorized, using bounded indexed queries through audited read-only access.

  4. How would you reproduce and document an API problem?

    Capture the endpoint, method, sanitized inputs, authentication context, time, environment, expected and actual response, request ID, and the smallest safe reproducible request.

Advanced Production Support Questions

  1. How do severity and priority differ in support operations?

    Severity describes technical and user impact; priority reflects the response order after considering severity, reach, contractual obligations, workarounds, and business context.

  2. How would you investigate a problem affecting only one tenant?

    Compare tenant configuration, entitlements, data shape, region, feature flags, permissions, recent changes, and request traces against a healthy tenant without crossing privacy boundaries.

  3. How should support handle rate limits and intermittent latency?

    Identify which boundary is limiting, examine response headers and percentiles, confirm request volume and retries, preserve idempotency, and coordinate capacity or usage changes.

  4. How do you protect customer data while collecting diagnostic evidence?

    Minimize collection, redact secrets and personal data, use approved secure channels, restrict access and retention, and record consent or purpose where required.

Support Engineering Scenarios

  1. Customers report intermittent 502 errors, but health checks stay green. What do you inspect?

    Correlate failures by time, route, instance, region, and dependency; inspect gateway and upstream timing, connection pools, resets, deploys, and percentile traces.

  2. A customer reports data missing, but you cannot reproduce it. What do you do next?

    Clarify exact filters, timezone, permissions, record IDs, timestamps, and workflow; trace the record through ingestion, storage, processing, and presentation.

  3. Two systems show different values for the same customer record. How do you investigate?

    Define the source of truth and field semantics, compare update times and lineage, inspect sync failures and transformations, then reconcile without overwriting valid newer data.

  4. How do you communicate during and after a major customer-facing incident?

    During the incident, provide impact, scope, known facts, mitigations, and the next update time; afterward, document root cause, recovery, prevention, and owned actions.