Been poking at clusterctl and trying to get it to work on Proxmox for a while, with not much luck. The first VM provisions, gets online, and then seems to get stuck somewhere.

It most likely is an issue with my template, since I added support for Rocky Linux 10 to the image-builder repo locally, including validate- and build- targets. The template builds fine, and clones fine, but then stalls.

In the interest of moving forward, maybe I just use some Terraform and Ansible to deploy/provision k8s VMs in a repeatable manner.

#homelab #proxmox #kubernetes #infrastructure #capmox #imagebuilder #terraform

Did you install the cloud provider interface in the workload cluster? It needs to be working for things to progress from "Provisioned" to "Running"

@heatsink
#ClusterAPI #Proxmox #Kubernetes #K8s #CPI

@praxiscode since I’m on Proxmox, I think you’re talking about Capmox? If so, Capmox is up, running, and connected.
@heatsink capmox handles the provisioning of the vms from the capi side. The cpi takes care of setting the providerId in the workload, which capi detects

Deploy this in your workload, via ClusterResourceSet.

https://github.com/k8s-proxmox/cloud-provider-proxmox

@heatsink

GitHub - k8s-proxmox/cloud-provider-proxmox: Kubernetes Cloud Provider for Proxmox VE

Kubernetes Cloud Provider for Proxmox VE. Contribute to k8s-proxmox/cloud-provider-proxmox development by creating an account on GitHub.

GitHub
Capi uses the workload's kubeconfig to retrieve the node providerId values and correlate them to the machine objects.
@heatsink