A DevOps engineer is supporting early-stage development for a developer platform that runs on Amazon Elastic Kubernetes Service (Amazon EKS). Recently, the platform has experienced an increased rate…

While CloudWatch dashboards are useful, default EKS service metrics often provide high-level cluster health, lacking the detailed, container-level performance and diagnostic data needed to pinpoint the reasons for container restart failures.

AWS CloudTrail logs API calls made to AWS services, primarily for auditing and security purposes, and does not provide runtime diagnostic information or performance metrics essential for troubleshooting container-level issues like restarts.

AWS CloudTrail Insights detects unusual API activity patterns in CloudTrail logs, focusing on control plane anomalies rather than application-level performance or container runtime diagnostics relevant to container restart failures.

Amazon CloudWatch Container Insights is specifically designed for monitoring containerized applications and Amazon EKS clusters, automatically collecting metrics and logs related to CPU, memory, disk, network, and container restart counts. This granular diagnostic information is crucial for identifying the root causes of container restart failures and effectively troubleshooting issues within the platform.