An Elastic Beanstalk environment running recent Amazon Linux instances shows no memory metrics in the Beanstalk console. The application has degraded performance suspected to be memory-related. How should the developer collect memory usage data for the instances?

The Amazon CloudWatch agent is required to collect detailed memory usage metrics from EC2 instances, including those managed by Elastic Beanstalk. By default, CloudWatch only collects basic metrics like CPU utilization. Installing and configuring the agent on the instances allows it to gather memory data and send it to CloudWatch. Configuring the CloudWatch agent to push logs to CloudWatch Logs using port 443 is for log collection, not metrics. Adding configuration in the Elastic Beanstalk .ebextensions directory is the correct method to automate the agent's installation and configuration, but it's not the direct action of enabling memory tracking itself. Creating a CloudWatch dashboard is for displaying metrics, not for collecting them.