News › ai-infrastructure
By Zayden R., August 28, 2026
Kubernetes 1.37 promotes the metrics.k8s.io API to stable, providing reliable CPU and memory metrics for nodes and pods. This long-awaited stability ensures consistent performance data for developers and operators.
Kubernetes version 1.37 has officially elevated the metrics.k8s.io API to stable status, a change that offers developers and cluster operators a reliable source of CPU and memory usage metrics for nodes and pods. This API, now in its stable form, remains crucial for commands like kubectl top and for resource-metrics-based autoscaling.
The transition to a stable API version, metrics.k8s.io/v1, means that engineers can count on the stability guarantees that accompany such a release. This is particularly significant for those who rely on the HorizontalPodAutoscaler (HPA) and other clients using this API in production environments. Originally introduced in alpha with Kubernetes v1.6 and later as a beta in v1.8, the API's stable release marks a significant step forward for Kubernetes users who have been utilizing these metrics without changes over the years.
Technically, the v1 API mirrors the v1beta1 version in terms of resource types and fields, focusing on NodeMetrics for node-level data and PodMetrics for pod-level data, including a detailed breakdown by container. This simplicity is intentional, designed to provide essential metrics without overcomplicating the data pipeline.
Cluster operators will appreciate that no additional feature gates are required to activate this feature. The Metrics API is delivered via the API aggregation layer, often implemented through tools like metrics-server. It's advisable to ensure that your chosen implementation can serve the v1.metrics.k8s.io API. During this transition period, continuing to support both v1 and v1beta1 versions is recommended to maintain compatibility with older clients.
Additionally, Kubernetes 1.37 brings advancements in workload-aware scheduling and cluster networking, catering to the growing demands of AI and machine learning workloads. These new capabilities will help organizations optimize the deployment and scaling of complex applications, reflecting Kubernetes' ongoing adaptation to modern computing needs.
The Linux Camp teaches these topics as hands-on labs on real virtual machines, verified as you type.