Hard Site Reliability Engineer questions

32 questions · 20 points each for a correct first answer.

1. 1.How much budget does a burn rate use?error-budgets, alertinghard
2. 2.Retry amplificationretries, cascading-failureshard
3. 3.Combining percentiles from many serverslatency, monitoringhard
4. 4.Little's Law for capacitycapacity-planninghard
5. 5.Tail latency with fan-outlatency, distributed-systemshard
6. 6.Throttled but not busykubernetes, linux-performancehard
7. 7.Handling overloadload-shedding, cascading-failureshard
8. 8.Running a safe chaos experimentchaos-engineering, resiliencehard
9. 9.Why a short window in burn-rate alertsslos, alertinghard
10. 10.Coordinated omissionload-testing, k6, latencyhard
11. 11.Gaps in a rate() graphpromql, prometheushard
12. 12.Misleading histogram quantilesprometheus, latency, slishard
13. 13.Outage that outlives its triggerdistributed-systems, resiliencehard
14. 14.A hot cache key expirescaching, distributed-systemshard
15. 15.Wasted work after timeoutstimeouts, distributed-systemshard
16. 16.Go service OOMKilled under loadgo, kubernetes, memoryhard
17. 17.Two leaders after a GC pausedistributed-systems, consensushard
18. 18.DNS failover that did not fail overdns, multi-regionhard
19. 19.Refactoring Terraform without destroyterraform, infrastructure-as-codehard
20. 20.Double pages from HA Prometheusprometheus, alertmanager, high-availabilityhard
21. 21.Broken tail-based samplingopentelemetry, tracinghard
22. 22.Alerting on queue lagslis, alerting, messaginghard
23. 23.Random packet drops on a busy nodelinux, networking, kuberneteshard
24. 24.Listen queue overflowslinux, networking, performancehard
25. 25.SLOs for a low-traffic serviceslos, alertinghard
26. 26.Where to measure the SLIslis, monitoringhard
27. 27.Rollback blocked by a schema changedeployments, databaseshard
28. 28.A surprise network billcost-optimization, kubernetes, networkinghard
29. 29.Emergency access to productionsecurity, on-callhard
30. 30.What to drop when overloadedresilience, load-sheddinghard
31. 31.Slow requests, idle CPUlinux-performance, ebpfhard
32. 32.Which pods are evicted first?kubernetes, capacity-planninghard