Question 310
An Auto Scaling group must allow a newly launched EC2 instance to register with an external license broker and download large application assets before it is considered available. Normal EC2 launch completion happens too early. Which feature permits a controlled pause in the launch lifecycle?
- Configure only EC2 user data that starts the license-registration job, without delaying the Auto Scaling instance transition to InService.
- Use an Auto Scaling lifecycle hook with a bounded heartbeat and completion action after bootstrap checks.
- Increase only the minimum number of running instances without adding any bootstrap gating.
- Set the ALB healthy threshold to one response and assume it checks the off-host license broker registration before adding an instance.
Explanation
Correct Answer
B:Lifecycle hooks pause transitions and allow initialization tasks to complete before an instance enters service.
Incorrect Answers
A:User data can run bootstrap commands but cannot by itself guarantee Auto Scaling waits for external readiness before completing launch.
C:More capacity does not prevent a newly launched but uninitialized instance from entering service.
D:An ALB health probe may help with readiness but does not establish a controlled Auto Scaling lifecycle pause and completion action for the bootstrap process.
Question 311
A web service runs predictable morning load spikes. EC2 instance startup and application initialization take several minutes, so reactive scale-out starts too slowly even though capacity is available. The team wants preinitialized instances kept ready to enter service. Which Auto Scaling feature fits?
- Increase the Auto Scaling group maximum capacity without launching any preinitialized instances ahead of the spike.
- Use an EC2 Auto Scaling warm pool with appropriate stopped or running instance state and verified initialization.
- Use only an Auto Scaling group maximum-capacity increase and assume instances are preinitialized before any scale event.
- Use a target tracking scale-out policy only after demand has exceeded threshold for several minutes, without any prewarmed capacity.
Explanation
Correct Answer
B:Warm pools preserve preinitialized capacity, reducing the time needed to serve sudden demand compared with cold EC2 startup.
Incorrect Answers
A:A larger upper limit permits later instance launches but does not reduce startup lag for required preinitialization.
C:A higher maximum permits more launches but does not shorten application bootstrap of a newly started instance.
D:Reactive tracking can scale eventually, but does not solve the warm-up delay that exceeds the predictable request spike ramp.
Question 312
A reservations API stores items in DynamoDB. Two workers update the same reservation and occasionally overwrite each other because neither verifies the version it read. The business needs optimistic concurrency without a global distributed lock. Which write condition should be used?
- Increase read capacity units and assume faster reads guarantee no concurrent writer changes the value.
- Use DynamoDB DAX cached reads as the final uniqueness check without any conditional write.
- Use a conditional update that compares the expected version number and increments it atomically on success.
- Perform a separate read and write with no conditional expression and assume the interval between calls cannot contain another write.
Explanation
Correct Answer
C:DynamoDB conditional writes can reject stale updates so concurrent writers do not silently overwrite a newer state.
Incorrect Answers
A:Capacity may reduce latency but cannot provide version-based write isolation.
B:A cache does not enforce atomic optimistic concurrency at the backing table.
D:Read-then-write is not atomic and allows the lost-update race already observed.
Question 313
A reservation API already uses DynamoDB TransactWriteItems to update inventory and create reservation records atomically. During flash sales, some calls return TransactionCanceledException. Investigators need to distinguish a genuine ConditionalCheckFailed rejection for exhausted inventory from a transient transaction conflict before deciding which requests are safe to retry. What diagnostic approach should the team implement?
- Examine transaction cancellation reasons in request-item order and transaction-conflict metrics, retry only transient conflicts with bounded backoff while treating failed stock conditions as business rejections.
- Retry every TransactionCanceledException indefinitely, including ConditionalCheckFailed on a depleted inventory item, until an order is forced through.
- Inspect only aggregate ConsumedCapacity returned by successful DynamoDB requests and assume it distinguishes all conditional and conflict cancellations.
- Read the items again using TransactGetItems and assume the read results reveal the precise cancellation reason of the failed prior write transaction.
Explanation
Correct Answer
A:DynamoDB exposes per-item cancellation reason codes and transaction conflict signals, allowing the application to distinguish contention from an expected failed condition and avoid blindly retrying out-of-stock orders.
Incorrect Answers
B:A failed stock condition represents a business rule, not necessarily a transient failure; indefinite retries neither add inventory nor preserve truthful order rejection.
C:Consumed capacity alone does not identify which item failed a condition versus a concurrent transaction conflict; the cancellation reasons supply that distinction.
D:A new consistent transaction read can show the current state but cannot substitute for the previous write request’s cancellation reason codes and conflict telemetry.
Question 314
A fleet of queue workers is unable to keep up with seasonal customer jobs. CPU remains low, but approximate number of visible SQS messages and age of oldest message rise while consumers are healthy. What scaling metric should guide the worker fleet?
- Disable SQS message metrics and scale out only after customers complain about overdue jobs.
- Double the queue message retention period and assume it processes messages at a higher rate automatically.
- Use queue backlog per worker or oldest-message age with target tracking or a controlled step scaling policy.
- Use an EC2 CPU target tracking policy even though the observed fleet CPU remains low while queue age and visible backlog grow.
Explanation
Correct Answer
C:Queue demand, rather than idle CPU, captures the throughput pressure of unprocessed work and supports responsive worker capacity.
Incorrect Answers
A:Removing backlog telemetry leaves demand unobserved until service objectives are already breached.
B:Retention affects how long messages remain available, not how quickly workers consume them.
D:Low CPU provides no actionable scale signal for the measured queue-processing demand and overdue messages.
Question 315
A company cannot risk EC2 capacity unavailability during a critical quarterly settlement window. It already knows instance type, Region and target Availability Zone, and must reserve capacity without committing to a year of usage discount. Which EC2 capacity option is relevant?
- Use EC2 On-Demand Capacity Reservations for the eligible instance attributes and target Availability Zone.
- Purchase only a Savings Plan, assuming its hourly spend commitment guarantees AZ-specific physical launch capacity.
- Create an Auto Scaling policy with a larger desired count and assume AWS cannot reject any instance launch.
- Purchase a Compute Savings Plan commitment sized to the settlement window and assume discounts guarantee an available instance in the specified AZ.
Explanation
Correct Answer
A:Capacity Reservations hold matching instance capacity in the selected AZ independently of long-term price-discount commitments.
Incorrect Answers
B:Savings Plans provide discounts but do not reserve exact Availability Zone capacity.
C:Desired capacity is a request, not a guarantee of available matching EC2 capacity.
D:Savings Plans discount eligible spend but do not hold physical capacity for compatible EC2 instance attributes in a particular AZ.
Question 316
A batch cluster uses EC2 Spot instances and periodically receives capacity rebalance recommendations shortly before interruption. Jobs checkpoint to S3. Which Auto Scaling behavior should be configured to improve replacement timing?
- Ignore rebalance recommendations and wait for every Spot instance to disappear before requesting replacements.
- Persist checkpoint files only on the interrupted EC2 instance’s EBS root volume without verifying what happens when Auto Scaling terminates the instance.
- Disable Capacity Rebalancing and rely solely on a scheduled replacement after the Spot instance termination notice.
- Enable Capacity Rebalancing where supported and coordinate instance replacement with interruption-aware checkpoint and drain actions.
Explanation
Correct Answer
D:Capacity Rebalancing can proactively replace at-risk Spot instances before they are interrupted, reducing work disruption.
Incorrect Answers
A:Delaying replacement loses usable lead time and raises interruption-related capacity gaps.
B:An Auto Scaling termination can remove or detach root storage and cannot guarantee the interrupted task checkpoint remains recoverable independently.
C:Without proactive rebalance, replacement requests lose the lead time available from rebalance recommendations and may create capacity gaps.
Question 317
A payments ECS service on Fargate has an API container and an essential local authorization-proxy sidecar in each task. During scaling, the API starts first and fails because the proxy is still initializing; the proxy later passes its container health check. The service does not need a traffic rollback change. Which task-definition change prevents the API from starting prematurely?
- Set the API container dependsOn condition to HEALTHY for the proxy, define a meaningful proxy container health check, and allow sufficient startup time on a supported Fargate platform.
- Use a dependsOn START condition for the proxy, which guarantees the proxy has passed its functional health check before the API starts.
- Increase only the ALB target-group deregistration delay; that controls traffic draining from old targets, not startup order of containers inside a task.
- Increase service desired count and remove the proxy health check so both containers start without checking their dependency.
Explanation
Correct Answer
A:ECS supports container startup dependencies; HEALTHY waits for the dependency container to pass its defined health check before starting its dependent container. This addresses the observed ordering race.
Incorrect Answers
B:START establishes that the proxy container started, not that its authentication listener is healthy, so the API can still race and fail on startup.
C:Target deregistration delay affects termination of existing targets during rollout and cannot ensure a sidecar has initialized before the API container begins.
D:More tasks reproduce the same per-task proxy readiness race; removing the health check eliminates the evidence needed for a HEALTHY dependency.
Question 318
A producer submits business events to an EventBridge bus. A downstream API target is throttled and EventBridge exhausts supported retries, but the business must retain failed delivery context for later repair. Which target configuration should be added?
- Configure an EventBridge Archive and assume any event that failed a specific rule-target delivery is automatically isolated and retried as a failed target invocation.
- Change only event bus name while keeping target retry exhaustion unrecorded.
- Configure appropriate EventBridge target retry policy and dead-letter queue with alerts and controlled replay.
- Disable target delivery error metrics and accept the event loss as successful publishing.
Explanation
Correct Answer
C:An EventBridge target DLQ captures eligible delivery failures after retries instead of silently losing business events.
Incorrect Answers
A:Archive/replay can preserve selected published events but is distinct from automatic per-target dead-lettering when a rule exhausts delivery retries.
B:Bus naming does not capture failed deliveries or create a replay queue.
D:A successful PutEvents call does not prove every target delivery succeeded.
Question 319
An AWS Step Functions workflow runs a payment risk check against a rate-limited partner API. It should retry transient 429 errors with bounded exponential backoff, but route permanent validation failures into a compensating branch. Which state-machine features apply?
- Put every API call into one Lambda invocation with an unlimited execution duration and in-memory failure state.
- Configure an AWS Step Functions Wait state alone to delay every API call, without Retry/Catch transitions for transient and permanent error classes.
- Define Retry behavior for expected transient errors and Catch transitions for persistent or nonretryable outcomes, with explicit compensation.
- Retry all errors forever with no backoff and never emit a terminal failure for malformed payloads.
Explanation
Correct Answer
C:Step Functions distinguishes retriable failures from terminal paths and can orchestrate cleanup or compensation after exhausted retries.
Incorrect Answers
A:Lambda execution is time-bounded and ephemeral, not a durable orchestrator for extended conditional retries.
B:Wait supplies time delays but does not classify error names or direct permanent validation failures into the specified compensation branch.
D:Unbounded retries amplify partner throttling and cannot resolve permanent business validation failures.
Question 320
A Kinesis Data Streams workload reports one hot shard while other shards are idle. Producers use a constant partition key for all customers and consumer processing keeps falling behind despite additional total shard capacity. Which design change addresses the skew?
- Choose partition keys that spread independent records across hash ranges and assess resharding or enhanced fanout as appropriate.
- Keep using the same constant partition key and simply add many unused shards, assuming one key automatically spans them.
- Switch Kinesis into on-demand capacity mode but retain a single constant partition key for nearly all new records.
- Decrease stream retention so the hot shard can process its current incoming writes faster.
Explanation
Correct Answer
A:Kinesis assigns records to shards based on partition key; a constant key concentrates traffic regardless of spare capacity elsewhere.
Incorrect Answers
B:A given partition key maps to a shard hash range and will not distribute itself merely because more shards exist.
C:On-demand can respond to distributed shard demand but does not force one hot partition key to spread across unrelated shard hash ranges.
D:Shorter retention does not remove concentrated ingest throughput on the selected shard.
AWS Certified Solutions Architect - Professional SAP-C02 Practice Tests