cloud-init failed execution / qemu agent exec status exitcode 2 when using files[].contentFrom

[miloaissatu]

What steps did you take and what happened:

When adding files to bootstrap configtemplate or controlplane definition using contentFrom secret or configMap, new machines fail to provision with machine status indicating.

* InfrastructureReady: cloud-init failed execution
* NodeHealthy: Waiting for ProxmoxMachine to report spec.providerID

The VM itself starts up fine, cloud-init runs to completion and it joins the cluster as far as kubernetes is concerned. The files loaded regularly and via contentFrom secret and configmap are all there on the VM.

I'm not sure how to troubleshoot further to determine what the qemu-agent exit code of 2 means or what its doing at that point that leads the the machine status being marked as 'failed' seemingly before cloud-init starts of while it's still running. The capmox controller log and kubectl get -o yaml outputs for the machine and proxmoxmachine are attached.

capmox.log

kubectl_outputs.txt

What did you expect to happen:
Able to use files[].contentFrom.{secret/configMap} successfully for bootstrap and controlplane providers.

Anything else you would like to add:
I'm using caprke2 https://github.com/rancher/cluster-api-provider-rke2 / https://caprke2.docs.rancher.com/. I haven't tried kubeadm with contentFrom. https://cluster-api.sigs.k8s.io/tasks/bootstrap/kubeadm-bootstrap/#additional-features

Environment:

• Proxmox version: 8.4.11
• Cluster-api-provider-proxmox version: v0.7.4
• Kubernetes version: (use kubectl version): Rancher kubernetes v1.33.5+rke2r1
• OS (e.g. from /etc/os-release): Rocky Linux 10.0 (Red Quartz). I also tried with Ubuntu 24.04.3 LTS (Noble Numbat), no difference.

[tril0byte]

try setting ProxmoxMachineTemplate spec.template.spec.checks.skipCloudInitStatus: true and rebuilding the cluster

I had the same problem which was preventing my cluster from reconciling, and skipCloudInitStatus fixed the overall cluster status for me, although the Machines are still not Ready waiting on the Proxmox Provider ID, it no longer blocks progress.

[miloaissatu]

do you know why it happens? i tried looking through the code but couldn't work out why it was getting that non-0 exit from the api seemingly before cloud-init even starts.

[mcbenjemaa]

Which OS distribution are you using?

[miloaissatu]

rocky10

[tril0byte]

I am not using cloud-init. I'm using Talos 'nocloud' images. I found the mention of skipCloudInitStatus at https://github.com/ionos-cloud/cluster-api-provider-proxmox/blob/main/docs/Usage.md?plain=1#L303 which seems to be part of a Flatcar Linux section