TOPIC
Reliability
These are the same questions found in the library, grouped here by the capability each one mainly examines.
129 questions
Questions 101–129 of 129 on this page.
A · Professional problem solving and implementation12
- Users see response times double, but model inference time has barely changed. How would you break down the request-processing stages, locate the bottleneck, and limit impact?
- To reduce inference cost, the team caches answers by prompt text. User permissions, knowledge versions, and regional rules may differ. How would you define cache keys and invalidation?
- A migration across millions of records cannot fit in one transaction. The business cannot expose a partially completed state. How would you design batched writes, visibility, and recovery?
- A hosted model upgrades without notice. Standard health checks pass, but classification boundaries and output length shift. How would you build safeguards?
- You need retries and timeouts for an async function. On final failure, callers still need each attempt’s error and whether side effects occurred. How would you design the result?
- CPU averages 35%, but the request queue keeps growing and threads wait on external I/O. How would you identify the real capacity bottleneck?
- Database storage grows while cleanup scans cannot reclaim space; reporting transactions run for hours. How would you diagnose and govern this?
- Instance health checks verify only process liveness, while user requests fail on a critical dependency timeout. Adding every dependency to probes may remove all instances. How would you design this?
- Multiple regions accept the same request independently. Idempotency records have a TTL and machine clocks may differ by seconds. How would you prevent duplicate writes at expiry boundaries?
- A live index needs a new shard key as data grows quickly and writes cannot stop. How would you plan old and new indexes and decide when to cut over?
- A multi-tenant query platform shares a database pool. One tenant submits many complex reports and increases latency for others. How would you isolate and meter usage?
- Product data lives in application, CDN, and client caches. After a price update, users see different values. How would you define refresh, versioning, and invalidation semantics?
No questions in this section match your target. ·
D · Delivery and ongoing operations12
- A critical dependency vulnerability needs prompt remediation, but the patch may conflict with the current release candidate; delay prolongs exposure, while immediate rollout could disrupt service. How would you plan validation, temporary mitigation, and release authorization?
- A consumer failure leaves a queue hours behind; after repair, many old messages will replay into billing, notifications, and analytics, which have different capacities and side effects. How would you control recovery rate and protect consumers?
- After a core dependency outage, most functionality has returned, but errors remain above baseline and upstream backlogs keep growing. The business wants full traffic restored quickly. How would you set recovery criteria and traffic ramp-up?
- A failure went unnoticed because monitoring missed a critical user path until support escalated it. The review calls for more monitoring, but the team must show it detects issues earlier without creating excessive noise. How would you validate the change?
- After a release, login, payment, and notification services fail in succession. Each team believes its incident is most urgent, and some alerts may share one root cause. How would you coordinate response, prioritize, and avoid interference?
- An outage did not meet the threshold for a major customer incident, but affected a small group and exposed process gaps. How would you decide whether and how much to publish?
- Upstream events are delayed by hours, while downstream reports still show success and customers make incorrect decisions. How would you design freshness states and notifications?
- The business depends on a third-party API with no formal SLA and inconsistent support. How would you set internal continuity goals and fallback options?
- Monitoring automatically restarts unhealthy services, but this incident originates in a shared data layer and mass restarts could amplify connection storms. How would you bound automation?
- A third-party service has recovered, but callbacks may have been lost during the outage. Customer-visible state differs from the platform record. How would you confirm impact and communicate?
- The outage is resolved, but some customers still have incorrect balances and promised compensation is pending. How would you define closure and continuing ownership?
- A release breaks reads, but the previous version cannot read the new data format. How would you contain impact and compare rollback, forward fix, and a compatibility layer?
No questions in this section match your target. ·
E · Collaboration, communication, and influence2
- A downstream team wants near-real-time responses and high availability, but platform capacity, cost, and failure modes are not discussed. How would you negotiate?
- Engineering wants an iteration to reduce recurring failures, while the business fears delaying customer features. How would you negotiate goals, investment, and returns?
No questions in this section match your target. ·
G · Learning, adaptation, and sustained growth2
- A mobile release you own prevents some users from launching the app, and the team is rolling back. How would you recover, communicate, and improve under pressure?
- A deployment nearly writes sensitive data to public logs, but an automated check blocks it and no customer is affected. Some say there is no incident to review. How would you handle it?
No questions in this section match your target. ·