upgrade_fabric.yml

---

- hosts: switches
  user: cumulus
  serial: 1

  tasks:

  #########
  #### check if the switch needs to be upgraded or not

  - name: Register the OS version
    shell: grep "VERSION_ID=" /etc/os-release | cut -d "=" -f 2
    register: current_version
    tags: register

  - name: Check if switch needs to be upgraded or not
    block:
      - debug:
         msg: "Switch needs to be upgraded from {{ current_version.stdout }} to {{ target_version }}"


      #########
      #### Check if the switch is doing L2 or only L3

      - name: Check if the switch is doing L2 or only L3
        shell: grep "clagd-enable" /etc/network/interfaces.d/interfaces | awk '{print $2}'
        register: clagd_enabled
        tags: register


      #########
      #### pre-flight checks applied to all the fabric switches

      # Extecute the consistency checks on all the switches
      - include_tasks: fabric_consistency.yml


      #########
      #### Stop critical services

      - pause:
          prompt: "Press any key to start the upgrade"

      - name: Stop L2 services
        block:
          - debug:
             msg: "Will stop L2 services"

          - name: Register the clagd state
            shell: /usr/bin/clagctl | grep "Our Priority" | awk '{ print $NF }'
            register: clagd_state
            tags: register

          - name: Set clagd role as secondary
            command: /usr/bin/clagctl priority 8200
            become: true
            when: clagd_state.stdout == "primary"

          - name: Print clagd state
            debug:
              msg: "{{ clagd_state.stdout }}"

          - name: Stop clagd service
            systemd: state=stopped name=clagd
            become: true

        when: clagd_enabled.stdout == "yes"

      - name: Stop FRR service
        systemd: state=stopped name=frr
        become: true

      - name: Shutdown all the ports
        script: files/shutdown_ports.sh
        become: true

      - pause:
          prompt: "Validate that everything properly converged before doing the upgrade"

      #########
      #### Start update

      - name: Upgrade system
        become: true
        apt:
          update_cache: yes
          autoremove: yes
          upgrade: dist
          allow_unauthenticated: yes
          dpkg_options: 'force-confold'


      #########
      ### Reboot system

      - name: reboot system
        shell: sleep 2 && shutdown -r now "Ansible package updates triggered"
        async: 1
        poll: 0
        become: true
        ignore_errors: true

      - pause:
          seconds: 5

      - name: waiting for switch to come back
        become: false
        wait_for_connection:
          timeout: 300


      #########
      ### Ensure that system rebooted properly

      - name: get system uptime
        command: awk -F . '{print $1}' /proc/uptime
        register: uptime
        become: false

      - name: verify a reboot was actually initiated
        # machine should have started after it has been rebooted
        fail:
          msg: "Uptime is too high for reboot to have worked: {{ uptime.stdout }}s"
        when: uptime.stdout|int > 300


      #########
      ### post-flight checks

      - name: POST register the OS version after upgrading
        shell: grep "VERSION_ID=" /etc/os-release | cut -d "\"" -f 2
        register: os_version_after
        tags: register

      - name: POST system version before and after upgrading
        debug:
          msg: "before: {{ current_version.stdout }}, after: {{ os_version_after.stdout }}"

      # Wait a bit for everything to converge
      - pause:
          seconds: 40

      # Execute the consistency checks on all the switches
      - include_tasks: fabric_consistency.yml

    when: current_version.stdout is version(target_version, '<')

  - name: Check if switch needs to be upgraded or not
    debug:
      msg: "Switch is already running version {{ current_version.stdout }}"
    when: current_version.stdout is version(target_version, '=')