update-k8s replaces machine.disks with nil

[AngellusMortis]

Bug Report

Description

I was trying to update the config for kube-proxy to enable metrics and hit this.

Running a upgrade-k8s in place terminates in the following error if machine.disks was left as the default value ([]).

I was able to work around it by manually setting disks to nil and then doing an apply/reboot before the upgrade. But it is still a bug.

Applied configuration with a reboot: this config change can't be applied in immediate mode
diff:   &v1alpha1.Config{
        ConfigVersion: "v1alpha1",
        ConfigDebug:   &false,
        ConfigPersist: &true,
        MachineConfig: &v1alpha1.MachineConfig{
                ... // 7 identical fields
                MachinePods:    nil,
                MachineNetwork: &{NetworkHostname: "cluster-1", NetworkInterfaces: {&{DeviceInterface: "enp130s0", DeviceDHCP: &true, DeviceVIPConfig: &{SharedIP: "192.168.2.250"}}}, NameServers: {"192.168.1.10"}},
-               MachineDisks:   []*v1alpha1.MachineDisk{},
+               MachineDisks:   nil,
                MachineInstall: &{InstallDisk: "/dev/sda", InstallImage: "factory.talos.dev/installer/daa836c9267d89003d0314e32d120d49d5c7"..., InstallWipe: &true},
                MachineFiles:   {&{FileContent: "[plugins]\n[plugins.\"io.containerd.grpc.v1.cri\"]\n    [plugins.\"io"..., FilePath: "/etc/cri/conf.d/20-customization.part", FileOp: "create"}},
                ... // 13 identical fields
        },
        ClusterConfig: &{ClusterID: "RnYo6TKYEQlkZfre3Hsx6d8HImh5KlbFEBCZ43xx3qo=", ClusterSecret: "yyoJjNTSdLWSzk4Pq/XgKcmavaYJ0bxecMHNAgVWcO8=", ControlPlane: &{Endpoint: &{URL: &{Scheme: "https", Host: "192.168.2.250:6443"}}}, ClusterName: "mortis", ...},
  }

Logs

Environment

• Talos version: 1.7.6
• Kubernetes version: 1.30.3
• Platform: