1  Based on this excellent guide: https://datavirke.dk/posts/bare-metal-kubernetes-part-1-talos-on-hetzner/
 2  
 3  First, run the workflow for building the Talos image and note the link to the artifact: https://github.com/superfishial/abyss/actions/workflows/build-talos-image.yaml
 4  
 5  The convention is to set the hostname of the worker nodes to `nereid-$index` where the index starts from 1
 6  
 7  ```bash
 8  # Use a version older than latest if you want to test an upgrade
 9  export TALOS_VERSION=v1.6.1
10  # Keep in mind that device paths might change between reboots so either
11  # double-check the device path, or use one of the udev paths instead.
12  export TARGET_DISK="/dev/nvme0n1"
13  
14  # OPTIONAL in case you need to stop RAID
15  mdadm --stop /dev/md/*
16  # wipe drives
17  wipefs -a /dev/nvme*n1
18  wipefs -a /dev/sd*
19  # Download the artifacts from our action run
20  # Update <GITHUB_TOKEN> with the output of `gh auth token`
21  # Update <ARTIFACT_ID> with the last section of the url of the artifact of the action run
22  wget --header "Authorization: Bearer <GITHUB_TOKEN>" -O images.zip https://api.github.com/repos/superfishial/abyss/actions/artifacts/<ARTIFACT_ID>/zip
23  unzip images.zip
24  unxz metal-amd64.raw.xz
25  # Write the raw disk image directly to the hard drive.
26  dd if=metal-amd64.raw of=$TARGET_DISK && sync
27  
28  mount LABEL=BOOT /mnt
29  sed -i 's/console=ttyS0//g' /mnt/grub/grub.cfg
30  # Remove the console=ttyS0 from the kernel parameters for both entries
31  # Based on https://github.com/siderolabs/talos/discussions/7465
32  
33  shutdown -r now
34  ```
35  
36  Get configuration from `neptune` terraform project via: `terraform output -raw talos_worker_machine_configuration > ../worker-configuration.yaml`
37  For each of the nodes:
38  
39  - change the `machine.network.hostname` to include the index of the node such as `nereid-1` and `nereid-2`
40  - run `talosctl apply-config --insecure -n 88.198.53.24 --file worker-configuration.yaml`