TLDR Tech / Practical
Kubernetes Transitions to cgroup v2: Key Changes
Kubernetes is phasing out cgroup v1 in favor of cgroup v2, with the failCgroupV1 setting becoming mandatory true in version 1.35. This transition introduces unified resource management, per-container Out-Of-Memory (OOM) killing, Pressure Stall Information (PSI) metrics, and support for Memory Quality of Service (QoS) and dynamic scaling. System administrators must update their Linux nodes to use cgroup v2 and configure the correct cgroup drivers prior to upgrading their Kubernetes clusters.
Read the full story