Hi! I've provisioned an RKE2 Cluster using Rancher with Harvester. I changed the deployed rke2-ingress-nginx-controller to setup a service of type Loadbalancer and In Harvester also defined a Loadbalancer in the correct namespace where the Healthchecks are working. The external-ip of the service rke2-ingress-nginx-controller remains in state "<pending>", what am i missing / how can i fix this?

what do you mean you “defined a loadbalancer”. If you set the Service Type to LoadBalancer, you need a LoadBalancer controller running in the cluster to actually bind the service to the LB provider. It sounds like you want to use Harvester for that; have you deployed the Harvester cloud provider and lb controller to this cluster?

Yes, i have deployed the Harvester cloud provider but i just don't understand what else is missing. I also haven't been able to find any docs regarding this. The Logs of "harvester-cloud-provider" are as followed:

E0415 13:39:06.651313       1 controller.go:289] error processing service kube-system/rke2-ingress-nginx-controller (will retry): failed to ensure load balancer: update load balancer IP of service kube-system/rke2-ingress-nginx-controller failed, error: timeout waiting for IP address, last error:%!w(<nil>)
I0415 13:39:06.651610       1 event.go:294] "Event occurred" object="kube-system/rke2-ingress-nginx-controller" fieldPath="" kind="Service" apiVersion="v1" type="Warning" reason="SyncLoadBalancerFailed" message="Error syncing load balancer: failed to ensure load balancer: update load balancer IP of service kube-system/rke2-ingress-nginx-controller failed, error: timeout waiting for IP address, last error:%!w(<nil>)"

In Harvester i can also see that a LoadBalancer is created every 5 minutes and disapears again with the following message:

No matched IPPool with requirement &{Network:default/segdck2 Project: Namespace:abcd Cluster:abcdef-e}

That error from Harvester seems pretty descriptive. Do you have an ippool that meets those requirements?

Yes, i do have a pool that matches Network and Project but haven't found a way to match the Cluster. I would also prefer using DHCP instead of setting up a static pool.