kube-vip manifest #9243
-
|
Can you tell me where the manifest definition for kube-vip is stored and how to change it for newly created EKSA clusters? |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment 1 reply
-
|
In EKS Anywhere, kube-vip is deployed as a static Pod on each control-plane node. Where the kube-vip manifest lives at runtime: /etc/kubernetes/manifests/kube-vip.yaml The kubelet watches this directory and automatically starts any static Pod defined there. kube-vip is launched this way on every control-plane node. How EKS Anywhere generates it: During cluster creation, EKS Anywhere renders the kube-vip static-pod manifest based on values from your Cluster specification. The most important input is the control plane endpoint: spec: This endpoint host becomes the virtual IP (or hostname) that kube-vip advertises and that kubeadm uses for the API server. How to change it for newly created EKS Anywhere clusters: Supported and recommended:
Not officially supported / fragile:
If you need persistent customization of kube-vip flags (interface selection, ARP/BGP settings, etc.) across all newly created clusters, the practical options are:
Summary: kube-vip runs as a static Pod. |
Beta Was this translation helpful? Give feedback.
In EKS Anywhere, kube-vip is deployed as a static Pod on each control-plane node.
That means the manifest that actually runs kube-vip is stored directly on the node filesystem, not in Kubernetes objects.
Where the kube-vip manifest lives at runtime:
/etc/kubernetes/manifests/kube-vip.yaml
The kubelet watches this directory and automatically starts any static Pod defined there. kube-vip is launched this way on every control-plane node.
How EKS Anywhere generates it:
During cluster creation, EKS Anywhere renders the kube-vip static-pod manifest based on values from your Cluster specification. The most important input is the control plane endpoint:
spec:
controlPlaneConfiguration:
endpoint:
…