Automated/unattended bare metal installation using USB

[mangkoran]

I would like to install Talos on bare metal local host completely (or maybe mostly) unattended e.g. plug USB to host and turn on the host and Talos should configure/install itself to the specified install disk in the host without the need for talosctl apply-config from client. I have searched Talos docs and GitHub repo and it seems there are 2 possible ways:

1. Using metal ISO and metal-iso disk volume
2. Using nocloud ISO and CIDATA disk volume

I tried with 2 USB drives just for testing purpose as I would like to only use the first USB drive for Talos installation. The first USB drive I used Ventoy to create partition to store the Talos ISO and then create 2 additional partitions using Windows "Disk Management" with the label metal-iso and CIDATA. The second USB drive I format the entire drive with metal-iso label. I have put the required config in each partition. However I still couldn't make it work using either ways. I even cannot do talosctl xxx if I boot the host with metal ISO as it said error getting disks: rpc error: code = Unavailable desc = connection error: desc = "transport: Error while dialing: dial tcp 192.168.2.235:50000: connect: connection refused".

I compared how Talos and Arch sees the USB drive to check if the USB drives have correctly partitioned and labeled.

Talos v1.9.2 nocloud

❯ talosctl -n 192.168.2.235 disks --insecure
DEV            MODEL                        SERIAL           TYPE      UUID   WWID                                               MODALIAS      NAME   SIZE     BUS_PATH                                                                                   SUBSYSTEM          READ_ONLY   SYSTEM_DISK
/dev/loop0     -                            -                UNKNOWN   -      -                                                  -             -      4.1 kB   /virtual                                                                                   /sys/class/block   *
/dev/loop1     -                            -                UNKNOWN   -      -                                                  -             -      4.1 kB   /virtual                                                                                   /sys/class/block   *
/dev/loop2     -                            -                UNKNOWN   -      -                                                  -             -      74 MB    /virtual                                                                                   /sys/class/block   *
/dev/nvme0n1   SAMSUNG MZVLB512HBJQ-00A00   S5EGNE0NA01531   NVME      -      eui.0025388a0103e6d3                               -             -      512 GB   /pci0000:00/0000:00:02.3/0000:04:00.0/nvme                                                 /sys/class/block
/dev/sda       SanDisk 3.2Gen1              -                HDD       -      -                                                  scsi:t-0x00   -      62 GB    /pci0000:00/0000:00:08.1/0000:05:00.3/usb2/2-2/2-2:1.0/host0/target0:0:0/0:0:0:0           /sys/class/block
/dev/sdb       SanDisk 3.2Gen1              -                HDD       -      t10. USB\0\0\0\0 SanDisk 3.2Gen10001212803202420   scsi:t-0x00   -      62 GB    /pci0000:00/0000:00:08.1/0000:05:00.4/usb4/4-2/4-2.4/4-2.4:1.0/host3/target3:0:0/3:0:0:0   /sys/class/block

❯ talosctl -n 192.168.2.235 get discoveredvolumes --insecure
NODE   NAMESPACE   TYPE               ID        VERSION   TYPE        SIZE     DISCOVERED   LABEL     PARTITIONLABEL
       runtime     DiscoveredVolume   loop2     1         disk        74 MB    squashfs
       runtime     DiscoveredVolume   nvme0n1   1         disk        512 GB
       runtime     DiscoveredVolume   sda       1         disk        62 GB    gpt
       runtime     DiscoveredVolume   sda1      1         partition   61 GB                           Ventoy
       runtime     DiscoveredVolume   sda2      1         partition   34 MB    vfat         VTOYEFI   VTOYEFI
       runtime     DiscoveredVolume   sda3      1         partition   67 MB                           Basic data partition
       runtime     DiscoveredVolume   sda4      1         partition   65 MB                           Basic data partition
       runtime     DiscoveredVolume   sdb       1         disk        62 GB

Arch

root@archiso ~ # lsblk -f
NAME       FSTYPE   FSVER            LABEL       UUID                                 FSAVAIL FSUSE% MOUNTPOINTS
loop0      squashfs 4.0                                                                     0   100% /run/archiso/airootfs
sda
├─sda1     exfat    1.0              Ventoy      4E21-0000
│ └─ventoy iso9660  Joliet Extension ARCH_202501 2025-01-01-08-45-10-00
├─sda2     vfat     FAT16            VTOYEFI     223C-F3F8
├─sda3     exfat    1.0              CIDATA      EA61-F0E0
└─sda4     exfat    1.0              metal-iso   9273-9E2C
sdb
└─sdb1     exfat    1.0              metal-iso   9E2D-7A2B
nvme0n1

root@archiso ~ # blkid
/dev/sda4: LABEL="metal-iso" UUID="9273-9E2C" BLOCK_SIZE="512" TYPE="exfat" PARTLABEL="Basic data partition" PARTUUID="48b7d789-4f89-4d64-91cb-14cf8f9b9e8f"
/dev/sda3: LABEL="CIDATA" UUID="EA61-F0E0" BLOCK_SIZE="512" TYPE="exfat" PARTLABEL="Basic data partition" PARTUUID="1a453317-9a96-41a1-aeb7-68e1ed56dd5d"
/dev/sda2: SEC_TYPE="msdos" LABEL_FATBOOT="VTOYEFI" LABEL="VTOYEFI" UUID="223C-F3F8" BLOCK_SIZE="512" TYPE="vfat" PARTLABEL="VTOYEFI" PARTUUID="068d9c30-898c-4ef6-aa79-c669dd056dc6"
/dev/sda1: LABEL="Ventoy" UUID="4E21-0000" BLOCK_SIZE="512" TYPE="exfat" PARTLABEL="Ventoy" PARTUUID="a52ff3fc-4c9e-4898-8c6f-43cf75a06508"
/dev/mapper/ventoy: BLOCK_SIZE="2048" UUID="2025-01-01-08-45-10-00" LABEL="ARCH_202501" TYPE="iso9660" PTUUID="d736165e" PTTYPE="dos"
/dev/sdb1: LABEL="metal-iso" UUID="9E2D-7A2B" BLOCK_SIZE="512" TYPE="exfat" PARTUUID="9e08b970-01"
/dev/loop0: BLOCK_SIZE="1048576" TYPE="squashfs"

It seems the USB drive LABEL is seen differently in Talos as it doesn't show some partition labels. In Arch the partition LABEL is correctly detected (CIDATA, metal-iso).

Am I doing it incorrectly? Please let me know if I need to provide additional item.

[smira]

If you could provide the full image of the device, or a way to reproduce it with command-line, we can look into it.

Also what version of Talos are you using?