Kubernetes v1.34 introduces an alpha feature providing insight into the health of specialized hardware like GPUs used in Pods. This new functionality, built upon KEP-4680, extends Dynamic Resource Allocation (DRA) drivers. Device health status is now reported directly into a Pod's status field via DRA. Exposing device health helps users diagnose issues in stateful applications or long-running jobs quickly. This allows operators and developers to identify underlying hardware failures. This is done through a new gRPC health service defined in the dra-health API group. The Kubelet integrates with DRA drivers, storing health updates in a cache. Pod status is updated when a device’s health changes, reflecting in the allocatedResourcesStatus field. This allows users to identify failed hardware by inspecting a Pod’s status using kubectl. To use the feature, the ResourceHealthStatus feature gate needs to be enabled. Planned enhancements include detailed health messages and configurable timeouts. The Kubernetes community aims to improve hardware failure handling and welcomes user feedback.