DOP-C02 Practice Question 93: StackSets Operation Preferences
AWS Certified DevOps Engineer – Professional (DOP-C02) · Test 10 · question 3 of 10
A StackSet deploys a security baseline to 60 accounts across three Regions. A DevOps engineer must update the StackSet so that: a bad template is caught early and the rollout stops before it reaches most accounts, the update in each Region finishes before the next Region starts, and throughput within a Region stays high enough that all its accounts finish quickly once the template is validated as safe.
Which operation preferences settings satisfy all three goals?
Show answer and explanation
Answer:
A small FailureToleranceCount stops the operation after only a handful of account failures, which catches a broken template early instead of letting it spread. RegionConcurrencyType set to SEQUENTIAL with a RegionOrder guarantees one Region completes (or the operation halts) before the next Region begins, satisfying the ordering requirement. Throughput depends on ConcurrencyMode. Under the default, STRICT_FAILURE_TOLERANCE, concurrency can be at most one more than the failure tolerance, so a tolerance of 1 would cap each Region at 2 accounts at a time no matter how high MaxConcurrentPercentage is. SOFT_FAILURE_TOLERANCE decouples the two: the high MaxConcurrentPercentage then lets most of a Region's accounts update in parallel, while the operation still stops once the failure tolerance is exceeded. Together these settings catch failures early, enforce Region-by-Region ordering, and still update quickly within a Region.
Option B sets FailureTolerancePercentage to 100, which means the operation tolerates every account failing and never stops early, the opposite of catching a bad template. MaxConcurrentCount of 1 also serializes every single account, which defeats the throughput goal.
Option C runs all three Regions in parallel, so a bad template can reach accounts in all three Regions before any tolerance threshold is hit, violating the Region-ordering requirement. A MaxConcurrentPercentage of 10 also throttles throughput far below what "finish quickly" implies.
Option D omits both failure tolerance settings, so the tolerance defaults to zero. Under the default STRICT_FAILURE_TOLERANCE mode, concurrency can then be at most one more than that tolerance, so the operation does not update every account in a Region at once as the option intends. Its stated goal of updating accounts "regardless of failures" also contradicts the requirement to stop a bad template early.
Reference: StackSetOperationPreferences