LoadBalancer service got stuck with message: Service is ready:Load balancer is being provisioned

[serhiynovos]

I installed Harvester 1.2.1 and connected Rancher 2.8.0 and then created RKE2 cluster with Cilium CNI.

I created IP pool with a few available IP addresses and Virtual Machine load balancer is working fine. Then I deployed a service which has LoadBalancer service and it got stuck with the next message "Service is ready:Load balancer is being provisioned"

On Harvester side I see LB for this service is created and active but it does not show which IP address is used (but in IP pool I see one more IP address was occupied)

[brandond]

I don't believe this is the correct place for your issue. Please reopen this issue against harvester, as you are using the harvester load-balancer controller.

[pck]

I had a similar problem, and switching from Cilium to Calico CNI didn't help.
Rancher provisioned RKE2 with 3 × "etcd + control" + 2 × worker nodes and the kube-system/kube-vip daemonset was not running anywhere. (I think because of a NoExecute-taint on nodes with etcd.) Reconfiguring the cluster as 3 × etc + 2 × control + 2 × worker solved my issue.

[serhiynovos]

@pck
Please refer to this issue harvester/harvester#4891

The issue that kube-vip DeamonSet did have toleration to run on etcd nodes only control plane. This comment solves the problem harvester/harvester#4891 (comment)