Skip to content

Multi-node Kind cluster fails to start on Ubuntu with Docker cgroup v2 #4030

@gauravhalnawar1011

Description

@gauravhalnawar1011

When trying to create a multi-node Kind cluster on Ubuntu (Docker using cgroup v2, systemd driver), the control-plane node fails to initialize. The logs indicate:

ERROR: failed to create cluster: failed to init node with kubeadm: command "docker exec --privileged kubeadm init ..." failed with error: exit status 1
[...]
your configuration file uses a deprecated API spec: "kubeadm.k8s.io/v1beta3"

Environment:

Ubuntu 22.04 (or 20.04 with Docker cgroup v2)

Docker 24+

Kind v0.30.0

Kubernetes node image: kindest/node:v1.33.4

Multi-node config: 1 control-plane + 2 workers

containerd inside Kind nodes uses default configuration

Steps to reproduce:

Create a Kind config with multiple nodes (1 control-plane + 2 workers).

Run:

kind create cluster --name test-cluster --image kindest/node:v1.33.4 --config config.yaml

Observed behavior:

Control-plane container fails to initialize.

kubeadm logs show deprecated API warnings and kubelet never becomes healthy.

All nodes are deleted automatically.

Expected behavior:

Control-plane node should start successfully.

All worker nodes should join the cluster.

Additional notes:

Single-node clusters work fine.

Adding containerdConfigPatches with SystemdCgroup = true resolves the issue.

Metadata

Metadata

Assignees

No one assigned

    Labels

    kind/bugCategorizes issue or PR as related to a bug.

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions