Can I use the porter on baremetal k8s to advertise kube api server?

[wangxin311]

There is circular dependency in metallb.

For example, if we use MetalLB to create and advertise the LB IP for api server, however kubelet cannot talk to the control plane until MetalLB has started and configured the LB IP, But MetalLB cannot start until kubelet can talk to the control plane and discover that it should be running the pod!

This issue is not specifically for metallb, wondering if porter has the same issue? Thanks

[duanjiong]

Why kubelet has to connect to external ip, why not cluster ip?

[wangxin311]

There is a user case for HA Kubernetes cluster which deployed on Baremetal to advertise Kube api server IP address, some solutions like HA proxy is working but that for L2 and many people seeking for a way to advertise it via BGP, it’s pretty common actually and I believe anyone who is going to deploy HA k8s cluster will facing the same issue.

[duanjiong]

You may need to set up a VIP to be able to route to the first master node, and then you can continue with the rest of the operation after the cluster is up.

[duanjiong]

You can refer to this https://kubernetes.io/docs/setup/production-environment/tools/kubeadm/high-availability/#stacked-control-plane-and-etcd-nodes.

[wangxin311]

This is not related on how to spin up HA cluster.

Here are some discussions regarding of the issues I am talking about on metallb repo, FYI.

metallb/metallb#168

[immanuelfodor]

If you are still looking for a HA cluster, I created a small, low-resource project that solves the problem, please take a look at https://github.com/immanuelfodor/kube-karp