Multi-Account, Multi-Cluster EKS Architecture Quiz
This quiz tests your understanding of Multi-Account, Multi-Cluster EKS Architecture.
- What is a valid validation objective for two-cluster A/B EKS?
- A) Fixing a single-AZ EKS control plane
- B) Testing containment of upgrade/add-on/webhook failures to one cluster
- C) Automatically covering every Region outage
- D) Guaranteeing lower cost
Show Answer
Answer: B) Testing containment of upgrade/add-on/webhook failures to one cluster
Explanation: Cluster-specific failure containment can justify the design. Shared VPC, DNS, Account, and data dependencies require separate analysis.
- Which statement correctly describes ARC zonal shift's fail-safe behavior?
- A) Traffic is blocked immediately for any workload once a zonal shift occurs
- B) If a workload's endpoints exist only in the impaired AZ, EKS keeps sending traffic to that AZ anyway
- C) Every workload is automatically relocated to another AZ
- D) This behavior only occurs on EKS Fargate
Show Answer
Answer: B) If a workload's endpoints exist only in the impaired AZ, EKS keeps sending traffic to that AZ anyway
Explanation: This is the fail-safe behavior — a workload deployed to only a single AZ is not protected by zonal shift. Test N-1 load and single-AZ exceptions; this does not mean a practice run always stops that workload.
- What does the EC2 link-local 1,024 packet/sec allowance imply?
- A) Check combined DNS/IMDS/NTP traffic and actual DNS metrics
- B) Unlimited capacity reserved exclusively for DNS
- C) A 100 Gbps-per-AZ limit
- D) Proof that CoreDNS is the most common failure
Show Answer
Answer: A) Check combined DNS/IMDS/NTP traffic and actual DNS metrics
Explanation: Review the shared link-local allowance and VPC DNS ENI limit. Verify linklocal_allowance_exceeded and DNS latency instead of inferring cause from Pod density alone.
- What should be checked when disabling ALB target-group cross-zone balancing?
- A) Guaranteed ALB cross-zone transfer savings
- B) Per-AZ capacity, distinguishing empty-AZ 503 from unhealthy-target failover
- C) Automatic cluster restarts
- D) All targets become healthy
Show Answer
Answer: B) Per-AZ capacity, distinguishing empty-AZ 503 from unhealthy-target failover
Explanation: Target stickiness and Lambda targets have restrictions; empty and unhealthy targets differ. ALB cross-zone regional data transfer has no additional transfer charge.