This message was deleted.

Did not experience. Are you experiencing it on all the nodes? Check rancher logs if something is wrong on the management side. Also i noted that a cluster update will overwrite some of your changes you did on rancher-managed workloads (such as the ingress controller arguments, in my case)

yes on all the nodes. apparently the health check endpoint liveness for kube-proxy is not accessible

http-get <http://localhost:10256/healthz> delay=10s timeout=15s period=10s #success=1 #failure=8

.

You can't explore the cluster to check pods logs, correct?

i found the problem. ip_vs module was not present in the kernel. i added it in all nodes and it's working again.

Oh, running your nodes on KVM? 😄

yes on proxmox vm 😅

you found this topic, correct? 😄 https://github.com/rancher/rke2/issues/4416

no this one https://github.com/rancher/rancher/issues/13769#issuecomment-548166140

Well, nice one. Kubenetwork issues are a effin pain. Happy you managed to find the solution.