Control plane adding/scaling problem #2502
Replies: 3 comments 1 reply
-
First of all, running an even number of the control plane nodes is strongly discouraged. You should always run an odd number of control plane nodes (1, 3, 5...) so that you don't run into issues with the etcd quorum. The second problem that I see is that there are no control plane components for the second control plane node. For example, there's only one kube-apiserver, kube-controller-manager... The third problem is that the Can you please do the following:
Please make sure to redact the secrets when posting logs if there are any. |
Beta Was this translation helpful? Give feedback.
-
@xmudrii Hi, First of all, running an even number of the control plane nodes is strongly discouraged. You should always run an odd number of control plane nodes (1, 3, 5...) so that you don't run into issues with the etcd quorum. You're right. The development version has one cluster and one machine, but we tested whether we would be able to scale control plane to high availability after the project was completed. Of course, the number of two machines was wrong. We tested on 3 but the problem is the same.
|
Beta Was this translation helpful? Give feedback.
-
@xmudrii I found the cause. But not solution.
It seems that changing the CNI to cilium or kubeProxyReplacement causes the problem. Should I change anything else in the configuration? |
Beta Was this translation helpful? Give feedback.
-
Hi,
We deploy KubeOne cluster (Hetzner) with one "control plane" and one "worker node".
terraform.tfvars
kubeone.yaml
kubectl get nodes
kubectl get pods -A
How we scale control plane?
terraform output -json > tf.json kubeone apply --manifest kubeone.yaml --credentials credentials.yaml --tfjson tf.json
Errors occur after this step.
Pods it's going to crash
Beta Was this translation helpful? Give feedback.
All reactions