All EC2 instances in an Auto Scaling group behind an Application Load Balancer have stopped serving traffic and are failing HTTP target group health checks. The application process is not running and system logs show many out-of-memory errors. To improve resilience against a possible memory leak and enable monitoring/alerts, which combination of actions should you take? (Choose two.)

Modifying the Auto Scaling group to replace instances failing load balancer health checks ensures that unhealthy instances, like those with memory leaks, are automatically terminated and replaced with new, healthy ones. This directly addresses the resilience requirement by maintaining application availability. Installing and configuring the CloudWatch agent on EC2 instances is necessary to collect detailed memory utilization metrics, as EC2's default metrics do not include memory. Creating an alarm for high memory usage with an SNS topic provides proactive monitoring and alerts, allowing you to detect and respond to potential memory leaks before they cause widespread outages. Reducing HealthCheckIntervalSeconds would make checks more frequent but wouldn't solve the underlying memory leak or provide monitoring. Changing health checks to TCP would only verify port reachability, not application health, and would mask the problem. Enabling an available memory consumption metric in the CloudWatch console is not possible for EC2 instances without the CloudWatch agent; default EC2 metrics do not include memory utilization.