Issues with attempting to deploy a Talos v1.12.2 cluster via Omni and Proxmox.

[macmandr197]

Hello! As titled, I'm attempting to create a cluster via Omni using cluster templates. Within this cluster template, I would like Talos/Omni to handle the node name, and do not care to specify my own hostnames. Typically, if I have done this in the past, I would not specify anything. I understand now, however, that a "hostnameconfig" patch is required within the cluster template, especially if providing a static hostname, etc.

For whatever reason, even with the hostnameconfig patch explicitly defined, the node's network.hostname is still configured which results in this error: configuration validation failed: 1 error occurred: * static hostname is already set in v1alpha1 config even though I do not believe I am explicitly setting a hostname anywhere?

I have explicitly specified that the hostnameconfig auto value should be set to stable so that Omni will let the talos machine config generate itself. This template is applied as follows (I'll only provide the control plane patch for now to save space)

cluster template

---
kind: Cluster
name: labber
labels:
  cluster-id: "1"
kubernetes:
  version: v1.35.0
talos:
  version: v1.12.2
patches:
  - name: disable-default-cni
    inline:
      cluster:
        network:
          cni:
            name: none
        proxy:
          disabled: true
---
kind: ControlPlane
machineClass:
  name: proxmox-control-plane
  size: 1
systemExtensions:
  - siderolabs/iscsi-tools
  - siderolabs/nfsd
  - siderolabs/qemu-guest-agent
  - siderolabs/util-linux-tools
patches:
  - name: stable-auto-hostname
    inline:
      apiVersion: v1alpha1
      kind: HostnameConfig
      auto: stable
  - file: iac/omni/cluster-template/patches/control-plane.yaml
---
kind: Workers
name: workers
machineClass:
  name: proxmox-worker
  size: 3
systemExtensions:
  - siderolabs/iscsi-tools
  - siderolabs/nfsd
  - siderolabs/qemu-guest-agent 
  - siderolabs/util-linux-tools
patches:
  - name: stable-auto-hostname
    inline:
      apiVersion: v1alpha1
      kind: HostnameConfig
      auto: stable
  - name: worker-labels
    inline:
      machine:
        nodeLabels:
          node-role.kubernetes.io/worker: ""
  - file: iac/omni/cluster-template/patches/worker.yaml
  - name: longhorn-storage
    inline:
      machine:
        kubelet:
          extraMounts:
            - destination: /var/lib/longhorn
              type: bind
              source: /var/local/longhorn
              options:
                - bind
                - rshared
                - rw
---
kind: Workers
name: gpu-workers
machineClass:
  name: proxmox-intel-gpu-worker
  size: 0
systemExtensions:
  - siderolabs/iscsi-tools
  - siderolabs/nfsd
  - siderolabs/qemu-guest-agent
  - siderolabs/util-linux-tools
  - siderolabs/i915
  - siderolabs/intel-ucode
  - siderolabs/xe
patches:
  - name: stable-auto-hostname
    inline:
      apiVersion: v1alpha1
      kind: HostnameConfig
      auto: stable
  - name: worker-labels
    inline:
      machine:
        nodeLabels:
          node-role.kubernetes.io/worker: ""
  - file: iac/omni/cluster-template/patches/intel-gpu-worker.yaml
  - name: longhorn-storage
    inline:
      machine:
        kubelet:
          extraMounts:
            - destination: /var/lib/longhorn
              type: bind
              source: /var/local/longhorn
              options:
                - bind
                - rshared
                - rw

control-plane patch

machine:
  network:
    nameservers:
      - *.*.*.*
      - 8.8.8.8
    interfaces:
      - deviceSelector:
          physical: true
        dhcp: true
        # VIP for control plane (shared across all control plane nodes)
        vip:
          ip: *.*.*.*
  features:
    hostDNS:
      enabled: true
      forwardKubeDNSToHost: false

[macmandr197]

FWIW, I am also using the Proxmox provider, with the machine class as specified below. I do not believe Proxmox is interfering with Omni/Talos' ability to set a hostname?

metadata:
  namespace: default
  type: MachineClasses.omni.sidero.dev
  id: proxmox-control-plane
spec:
  matchlabels: []
  autoprovision:
    providerid: proxmox
    kernelargs: []
    metavalues: []
    providerdata: |
      cores: 2
      sockets: 1
      memory: 16000
      disk_size: 50
      network_bridge: {redacted}
      storage_selector: name == "pool-1"
      disk_ssd: true
      disk_discard: true
      disk_iothread: true
      disk_cache: writeback
      disk_aio: io_uring
      grpctunnel: 0

[smira]

Please report it here: https://github.com/siderolabs/omni-infra-provider-proxmox/issues

It's a Proxmox provider issue, it is known, but there is no fix (yet).

[Unix4ever]

Posting the issue in Proxmox repo for visibility.
Was fixed already