HetznerTerra/ansible/site.yml

---
- name: Bootstrap Kubernetes cluster
  hosts: cluster
  become: true
  gather_facts: true

  pre_tasks:
    - name: Wait for SSH
      wait_for_connection:
        delay: 10
        timeout: 300

  roles:
    - common

- name: Setup primary control plane
  hosts: control_plane[0]
  become: true

  vars:
    k3s_primary: true
    k3s_token: "{{ lookup('password', '/dev/null length=32 chars=ascii_letters,digits') }}"
    k3s_primary_private_ip: "{{ k3s_private_ip }}"
    k3s_primary_public_ip: "{{ ansible_host }}"
    k3s_primary_ip: "{{ k3s_private_ip }}"
    k3s_node_ip: "{{ k3s_private_ip }}"
    # kube_api_endpoint is set in inventory group_vars

  roles:
    - k3s-server

- name: Get join info from primary
  hosts: control_plane[0]
  become: true
  tasks:
    - name: Fetch node token
      command: cat /var/lib/rancher/k3s/server/node-token
      register: node_token
      changed_when: false

    - name: Set join token fact
      set_fact:
        k3s_token: "{{ node_token.stdout }}"
        k3s_primary_private_ip: "{{ k3s_private_ip }}"
        k3s_primary_public_ip: "{{ ansible_host }}"

    - name: Fetch kubeconfig
      fetch:
        src: /etc/rancher/k3s/k3s.yaml
        dest: ../outputs/kubeconfig
        flat: true

- name: Bootstrap addon prerequisite secrets
  hosts: control_plane[0]
  become: true

  roles:
    - addon-secrets-bootstrap

- name: Deploy Hetzner CCM (required for workers with external cloud provider)
  hosts: control_plane[0]
  become: true

  roles:
    - ccm-deploy

- name: Setup secondary control planes
  hosts: control_plane[1:]
  become: true

  vars:
    k3s_primary: false
    k3s_token: "{{ hostvars[groups['control_plane'][0]]['k3s_token'] }}"
    k3s_primary_ip: "{{ hostvars[groups['control_plane'][0]]['k3s_primary_private_ip'] }}"
    k3s_primary_public_ip: "{{ hostvars[groups['control_plane'][0]]['k3s_primary_public_ip'] }}"
    k3s_node_ip: "{{ k3s_private_ip }}"
    # Use Load Balancer for HA - all control planes join via LB endpoint
    k3s_join_endpoint: "{{ kube_api_endpoint | default(hostvars[groups['control_plane'][0]]['k3s_primary_private_ip']) }}"

  roles:
    - k3s-server

- name: Setup workers
  hosts: workers
  become: true

  vars:
    k3s_token: "{{ hostvars[groups['control_plane'][0]]['k3s_token'] }}"
    # Use Load Balancer for HA - workers join via LB endpoint
    k3s_server_url: "https://{{ kube_api_endpoint | default(hostvars[groups['control_plane'][0]]['k3s_primary_private_ip']) }}:6443"
    k3s_node_ip: "{{ k3s_private_ip }}"

  roles:
    - k3s-agent

- name: Deploy observability stack
  hosts: control_plane[0]
  become: true

  roles:
    - role: observability
      when: not (observability_gitops_enabled | default(true) | bool)

- name: Provision Grafana content
  hosts: control_plane[0]
  become: true

  roles:
    - role: observability-content
      when: not (observability_gitops_enabled | default(true) | bool)

- name: Bootstrap Doppler access for External Secrets
  hosts: control_plane[0]
  become: true

  roles:
    - doppler-bootstrap

- name: Detect existing Tailscale service proxies
  hosts: control_plane[0]
  become: true
  tasks:
    - name: Check for current Tailscale service hostnames
      command: kubectl get svc -A -o jsonpath='{range .items[*]}{.metadata.annotations.tailscale\.com/hostname}{"\n"}{end}'
      register: existing_tailscale_hostnames
      changed_when: false
      failed_when: false

- name: Clean up stale Tailscale devices
  hosts: localhost
  connection: local
  vars:
    tailscale_reserved_hostnames:
      - rancher
      - grafana
      - prometheus
      - flux
  tasks:
    - name: Delete stale devices only before service proxies exist
      include_role:
        name: tailscale-cleanup
      when: >-
        hostvars[groups['control_plane'][0]].existing_tailscale_hostnames.stdout_lines | default([])
        | intersect(tailscale_reserved_hostnames)
        | length == 0

- name: Finalize
  hosts: localhost
  connection: local
  tasks:
    - name: Update kubeconfig server address
      command: |
        sed -i 's/127.0.0.1/{{ hostvars[groups["control_plane"][0]]["ansible_host"] }}/g' ../outputs/kubeconfig
      changed_when: true

    - name: Display success message
      debug:
        msg: |
          Cluster setup complete!
          Control planes: {{ groups['control_plane'] | length }}
          Workers: {{ groups['workers'] | length }}
          To access the cluster:
            export KUBECONFIG={{ playbook_dir }}/../outputs/kubeconfig
            kubectl get nodes