
Kubernetes v1.37, themed "Garhwal" to honor the Himalayan roots of release lead Dipesh Rawat, introduces 67 enhancements focused on workload efficiency and cluster operations. Key updates include the beta release of Horizontal Pod Autoscaler scale-to-zero for event-driven batch processing, gang scheduling for AI training workloads, and workload-aware preemption to optimize resource utilization. The release also marks the transition of the metrics API to stable and the introduction of KAML for improved manifest readability. Critical deprecations require immediate attention from cluster operators, specifically the removal of IPVS mode in kube-proxy, the discontinuation of KubeDNS in favor of CoreDNS, and the restriction of static pod access to secrets and config maps. These changes collectively aim to modernize scheduling, enhance security, and streamline day-two operations for complex, high-performance computing environments.
Sign in to continue reading, translating and more.
Open full episode in Podwise