Kubernetes v1.35 blocks kubelet on cgroup v1 nodes; one command checks your fleet

Starting with Kubernetes v1.35, the kubelet setting failCgroupV1 defaults to true, so kubelets refuse to start on nodes still running cgroup v1. Affected nodes go NotReady, and on control-plane nodes the API server and etcd static pods may also fail to return after a reboot. kubeadm's SystemVerification preflight check flags this during init, join and upgrade, and the author advises against bypassing it with --ignore-preflight-errors. The article provides a kubectl debug command that runs stat on each node's cgroup filesystem to tell cgroup v1 (tmpfs) from v2 (cgroup2fs), and notes that the chroot /host step is required for accurate results. Common older distributions such as CentOS 7, RHEL 7 and 8, Ubuntu 18.04 and 20.04, and Amazon Linux 2 default to cgroup v1, making self-managed kubeadm clusters on long-lived VMs most at risk.
This is an AI-generated summary. ShortSingh links to the original source for the complete article.
Discussion (0)
Log in to join the discussion and vote.
Log in