Worker nodes cannot join managed Kubernetes version <1.32

[lllamnyp]

If a managed Kubernetes is created with .spec.version == 1.31, it does not complete installation, because Kubevirt's worker nodes never join the cluster and are continuously recreated every 10-15 minutes.

Environment

• Cozystack version 0.37.2
• Provider: on-prem.

To Reproduce Steps to reproduce the behavior:

kubectl -n tenant-mytenant create -f- <<EOF
apiVersion: apps.cozystack.io/v1alpha1
kind: Kubernetes
metadata:
  name: test
spec:
  version: "1.31"
EOF

Expected behaviour The worker nodes join the cluster.

Actual behaviour kubectl get nodes returns an empty list, VMs for worker nodes are continuously recreated.

Logs

$ kubectl logs virt-launcher-kubernetes-test-md0-4c72q-j7chc-85z5n guest-console-log
...
[   15.018046] cloud-init[1119]: error execution phase preflight: unable to fetch the kubeadm-config ConfigMap: this version of kubeadm only supports deploying clusters with the control plane version >= 1.32.0. Current version: v1.31.10
...

[dosubot[bot]]

This is expected behavior in Cozystack 0.37.2. The error from kubeadm shows that the worker node image uses a kubeadm version that only supports joining clusters with a control plane version >= 1.32.0. Cozystack's managed Kubernetes clusters officially support versions from 1.28 to 1.33, but the default and recommended version is 1.32 or higher, and the worker node images are built with the latest kubeadm, which enforces this minimum version requirement source.

There are no configuration options or workarounds documented to allow worker nodes to join clusters with Kubernetes 1.31. To avoid continuous VM recreation and join failures, use Kubernetes version 1.32 or newer for managed clusters in Cozystack 0.37.2 source.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}