Developed a container termination metrics feature for the kubernetes/kubernetes repository, enabling improved visibility into container exit patterns and supporting error-rate calculations. The work focused on implementing the terminated_containers_total metric in kubelet using Go, with careful attention to metrics instrumentation and Kubernetes best practices. To prevent cardinality explosion, the metric used low-cardinality labels such as container_type, exit_code, and reason, while a robust status manager ensured that each termination transition was recorded exactly once. Comprehensive unit and node-level end-to-end tests validated metrics exposure, enhancing observability for abnormal container exits and providing actionable data for reliability and incident response.