Amazon AWS Certified Solutions Architect - Associate SAA-C03 Bundle

The option Allow TCP 2049 between approved client and mount-target security groups addresses a separate condition: linux clients lose file access because EFS mount-target security groups reject NFS. NFS network access must be explicitly authorized. The actual requirement is a data pipeline writes temporary state to EFS and assumes it is immune to accidental deletion. Recommended corrective action: Add backup/version/recovery procedures for shared filesystem content. Regional HA does not undo logical file deletion.

Move shared files to Amazon EFS with appropriate mount targets would help if an application stores shared assets only on one EC2 root volume. Regional EFS can provide shared NFS across instances. That is not the symptom described here: a data pipeline writes temporary state to EFS and assumes it is immune to accidental deletion. The matching response is Add backup/version/recovery procedures for shared filesystem content, because regional HA does not undo logical file deletion.

Configure EFS mount targets in required Availability Zones solves a different operational problem where eFS clients in a second AZ have no local mount target and fail during AZ loss. AZ-local mount targets provide regional client access resilience. It does not directly address a data pipeline writes temporary state to EFS and assumes it is immune to accidental deletion. The case calls for Add backup/version/recovery procedures for shared filesystem content, since regional HA does not undo logical file deletion.

Use the load balancer's managed DNS name or Route 53 alias would help if an application uses fixed IPs for an internal ALB and fails after scaling. Managed load-balancer IPs are not stable integration identifiers. That is not the symptom described here: a single interface endpoint in one subnet is used by all application AZs. The matching response is Create interface endpoint ENIs in required AZ subnets and configure DNS appropriately, because interface endpoint placement influences AZ path availability.

For the alternative failure a database fails over but applications cache the old resolved IP indefinitely, the control Honor DNS TTL and reconnect to managed database endpoints is relevant: endpoint failover depends on DNS refresh behavior. Here the evidence instead is a single interface endpoint in one subnet is used by all application AZs. The selected technical mechanism is Create interface endpoint ENIs in required AZ subnets and configure DNS appropriately: interface endpoint placement influences AZ path availability.

Configure redundant Route 53 Resolver endpoints and routing solves a different operational problem where private VPC DNS resolvers fail to resolve on-prem names after route changes. Hybrid DNS needs resilient resolver ingress/egress architecture. It does not directly address a single interface endpoint in one subnet is used by all application AZs. The case calls for Create interface endpoint ENIs in required AZ subnets and configure DNS appropriately, since interface endpoint placement influences AZ path availability.

For the alternative failure a branch reaches only one Direct Connect connection during provider failure, the control Use resilient redundant connections with suitable BGP preference is relevant: diverse hybrid connectivity reduces physical link dependency. Here the evidence instead is an application uses fixed IPs for an internal ALB and fails after scaling. The selected technical mechanism is Use the load balancer's managed DNS name or Route 53 alias: managed load-balancer IPs are not stable integration identifiers.

The competing option Create interface endpoint ENIs in required AZ subnets and configure DNS appropriately would be useful when a single interface endpoint in one subnet is used by all application AZs. Interface endpoint placement influences AZ path availability. The present issue is an application uses fixed IPs for an internal ALB and fails after scaling; it is more directly addressed by Use the load balancer's managed DNS name or Route 53 alias. Managed load-balancer IPs are not stable integration identifiers.

The option Configure redundant Route 53 Resolver endpoints and routing addresses a separate condition: private VPC DNS resolvers fail to resolve on-prem names after route changes. Hybrid DNS needs resilient resolver ingress/egress architecture. The actual requirement is an application uses fixed IPs for an internal ALB and fails after scaling. Recommended corrective action: Use the load balancer's managed DNS name or Route 53 alias. Managed load-balancer IPs are not stable integration identifiers.

The competing option Use resilient redundant connections with suitable BGP preference would be useful when a branch reaches only one Direct Connect connection during provider failure. Diverse hybrid connectivity reduces physical link dependency. The present issue is private VPC DNS resolvers fail to resolve on-prem names after route changes; it is more directly addressed by Configure redundant Route 53 Resolver endpoints and routing. Hybrid DNS needs resilient resolver ingress/egress architecture.

The option Create interface endpoint ENIs in required AZ subnets and configure DNS appropriately addresses a separate condition: a single interface endpoint in one subnet is used by all application AZs. Interface endpoint placement influences AZ path availability. The actual requirement is private VPC DNS resolvers fail to resolve on-prem names after route changes. Recommended corrective action: Configure redundant Route 53 Resolver endpoints and routing. Hybrid DNS needs resilient resolver ingress/egress architecture.

Use the load balancer's managed DNS name or Route 53 alias would help if an application uses fixed IPs for an internal ALB and fails after scaling. Managed load-balancer IPs are not stable integration identifiers. That is not the symptom described here: private VPC DNS resolvers fail to resolve on-prem names after route changes. The matching response is Configure redundant Route 53 Resolver endpoints and routing, because hybrid DNS needs resilient resolver ingress/egress architecture.

The competing option Create interface endpoint ENIs in required AZ subnets and configure DNS appropriately would be useful when a single interface endpoint in one subnet is used by all application AZs. Interface endpoint placement influences AZ path availability. The present issue is a database fails over but applications cache the old resolved IP indefinitely; it is more directly addressed by Honor DNS TTL and reconnect to managed database endpoints. Endpoint failover depends on DNS refresh behavior.

The option Configure redundant Route 53 Resolver endpoints and routing addresses a separate condition: private VPC DNS resolvers fail to resolve on-prem names after route changes. Hybrid DNS needs resilient resolver ingress/egress architecture. The actual requirement is a database fails over but applications cache the old resolved IP indefinitely. Recommended corrective action: Honor DNS TTL and reconnect to managed database endpoints. Endpoint failover depends on DNS refresh behavior.

For the alternative failure a branch reaches only one Direct Connect connection during provider failure, the control Use resilient redundant connections with suitable BGP preference is relevant: diverse hybrid connectivity reduces physical link dependency. Here the evidence instead is a database fails over but applications cache the old resolved IP indefinitely. The selected technical mechanism is Honor DNS TTL and reconnect to managed database endpoints: endpoint failover depends on DNS refresh behavior.

The option Create interface endpoint ENIs in required AZ subnets and configure DNS appropriately addresses a separate condition: a single interface endpoint in one subnet is used by all application AZs. Interface endpoint placement influences AZ path availability. The actual requirement is a branch reaches only one Direct Connect connection during provider failure. Recommended corrective action: Use resilient redundant connections with suitable BGP preference. Diverse hybrid connectivity reduces physical link dependency.

Use the load balancer's managed DNS name or Route 53 alias would help if an application uses fixed IPs for an internal ALB and fails after scaling. Managed load-balancer IPs are not stable integration identifiers. That is not the symptom described here: a branch reaches only one Direct Connect connection during provider failure. The matching response is Use resilient redundant connections with suitable BGP preference, because diverse hybrid connectivity reduces physical link dependency.

Configure redundant Route 53 Resolver endpoints and routing solves a different operational problem where private VPC DNS resolvers fail to resolve on-prem names after route changes. Hybrid DNS needs resilient resolver ingress/egress architecture. It does not directly address a branch reaches only one Direct Connect connection during provider failure. The case calls for Use resilient redundant connections with suitable BGP preference, since diverse hybrid connectivity reduces physical link dependency.

Move terminal failures to a DLQ with controlled redrive would help if messages that exhaust retries require operator review without blocking all orders. DLQ handling isolates errors while healthy work continues. That is not the symptom described here: a write succeeds but the caller times out and retries the same order. The matching response is Use idempotency keys and atomic duplicate detection on order storage, because retries must not double-commit business actions.

For the alternative failure a Step Functions activity fails after a partial external side effect, the control Use compensation logic and recorded step state before retrying is relevant: workflow retries cannot automatically undo external effects. Here the evidence instead is a write succeeds but the caller times out and retries the same order. The selected technical mechanism is Use idempotency keys and atomic duplicate detection on order storage: retries must not double-commit business actions.

Make the write idempotent and acknowledge only after durable success solves a different operational problem where a queue consumer crashes after writing a DynamoDB item but before acknowledging the message. At-least-once redelivery may repeat successful processing. It does not directly address a write succeeds but the caller times out and retries the same order. The case calls for Use idempotency keys and atomic duplicate detection on order storage, since retries must not double-commit business actions.

For the alternative failure messages that exhaust retries require operator review without blocking all orders, the control Move terminal failures to a DLQ with controlled redrive is relevant: dLQ handling isolates errors while healthy work continues. Here the evidence instead is a Step Functions activity fails after a partial external side effect. The selected technical mechanism is Use compensation logic and recorded step state before retrying: workflow retries cannot automatically undo external effects.

The competing option Use idempotency keys and atomic duplicate detection on order storage would be useful when a write succeeds but the caller times out and retries the same order. Retries must not double-commit business actions. The present issue is a Step Functions activity fails after a partial external side effect; it is more directly addressed by Use compensation logic and recorded step state before retrying. Workflow retries cannot automatically undo external effects.

The option Add bounded exponential backoff and jitter in replay logic addresses a separate condition: a downstream API returns rate-limit errors during recovery replay. Controlled retries prevent overload during recovery. The actual requirement is a Step Functions activity fails after a partial external side effect. Recommended corrective action: Use compensation logic and recorded step state before retrying. Workflow retries cannot automatically undo external effects.

The competing option Add bounded exponential backoff and jitter in replay logic would be useful when a downstream API returns rate-limit errors during recovery replay. Controlled retries prevent overload during recovery. The present issue is a queue consumer crashes after writing a DynamoDB item but before acknowledging the message; it is more directly addressed by Make the write idempotent and acknowledge only after durable success. At-least-once redelivery may repeat successful processing.

The option Use compensation logic and recorded step state before retrying addresses a separate condition: a Step Functions activity fails after a partial external side effect. Workflow retries cannot automatically undo external effects. The actual requirement is a queue consumer crashes after writing a DynamoDB item but before acknowledging the message. Recommended corrective action: Make the write idempotent and acknowledge only after durable success. At-least-once redelivery may repeat successful processing.

Use idempotency keys and atomic duplicate detection on order storage would help if a write succeeds but the caller times out and retries the same order. Retries must not double-commit business actions. That is not the symptom described here: a queue consumer crashes after writing a DynamoDB item but before acknowledging the message. The matching response is Make the write idempotent and acknowledge only after durable success, because at-least-once redelivery may repeat successful processing.

The competing option Use idempotency keys and atomic duplicate detection on order storage would be useful when a write succeeds but the caller times out and retries the same order. Retries must not double-commit business actions. The present issue is a downstream API returns rate-limit errors during recovery replay; it is more directly addressed by Add bounded exponential backoff and jitter in replay logic. Controlled retries prevent overload during recovery.

The option Use compensation logic and recorded step state before retrying addresses a separate condition: a Step Functions activity fails after a partial external side effect. Workflow retries cannot automatically undo external effects. The actual requirement is a downstream API returns rate-limit errors during recovery replay. Recommended corrective action: Add bounded exponential backoff and jitter in replay logic. Controlled retries prevent overload during recovery.

For the alternative failure messages that exhaust retries require operator review without blocking all orders, the control Move terminal failures to a DLQ with controlled redrive is relevant: dLQ handling isolates errors while healthy work continues. Here the evidence instead is a downstream API returns rate-limit errors during recovery replay. The selected technical mechanism is Add bounded exponential backoff and jitter in replay logic: controlled retries prevent overload during recovery.

The option Make the write idempotent and acknowledge only after durable success addresses a separate condition: a queue consumer crashes after writing a DynamoDB item but before acknowledging the message. At-least-once redelivery may repeat successful processing. The actual requirement is messages that exhaust retries require operator review without blocking all orders. Recommended corrective action: Move terminal failures to a DLQ with controlled redrive. DLQ handling isolates errors while healthy work continues.

Use compensation logic and recorded step state before retrying would help if a Step Functions activity fails after a partial external side effect. Workflow retries cannot automatically undo external effects. That is not the symptom described here: messages that exhaust retries require operator review without blocking all orders. The matching response is Move terminal failures to a DLQ with controlled redrive, because dLQ handling isolates errors while healthy work continues.

For the alternative failure a write succeeds but the caller times out and retries the same order, the control Use idempotency keys and atomic duplicate detection on order storage is relevant: retries must not double-commit business actions. Here the evidence instead is messages that exhaust retries require operator review without blocking all orders. The selected technical mechanism is Move terminal failures to a DLQ with controlled redrive: dLQ handling isolates errors while healthy work continues.