yt-dlp-dags/ansible/playbook-worker.yml

---
- name: Deploy Airflow Workers
  hosts: airflow_workers
  gather_facts: yes
  vars_files:
    - "{{ inventory_dir }}/group_vars/all/generated_vars.yml"
    - "{{ inventory_dir }}/group_vars/all/vault.yml"
  pre_tasks:
    - name: Announce worker deployment
      debug:
        msg: "Starting deployment for Airflow Worker: {{ inventory_hostname }} ({{ ansible_user }}@{{ ansible_host }})"

    - name: Configure system timezone
      # Ensures all services and logs on this node use a consistent timezone.
      community.general.timezone:
        name: "{{ host_timezone }}"
      become: yes

    - name: Install NTP for time synchronization
      ansible.builtin.apt:
        name: ntp
        state: present
      become: yes

    - name: Ensure NTP service is started and enabled
      ansible.builtin.service:
        name: ntp
        state: started
        enabled: yes
      become: yes

    - name: Set deploy_group to a valid single group name
      set_fact:
        deploy_group: "ytdl"

    - name: Ensure deploy group exists
      group:
        name: "{{ deploy_group }}"
        state: present
      become: yes

    - name: Ensure deploy user exists
      user:
        name: "{{ ansible_user }}"
        group: "{{ deploy_group }}"
        state: present
      become: yes

    - name: Validate deploy_group variable
      ansible.builtin.assert:
        that:
          - deploy_group is defined
          - deploy_group is string
          - "',' not in deploy_group"
          - "' ' not in deploy_group"
        fail_msg: "The 'deploy_group' variable ('{{ deploy_group }}') must be a single, valid group name. It should not contain commas or spaces."

    - name: Check for swapfile
      stat:
        path: /swapfile
      register: swap_file
      become: yes

    - name: Create 8GB swapfile
      command: fallocate -l 8G /swapfile
      when: not swap_file.stat.exists
      become: yes

    - name: Set swapfile permissions
      file:
        path: /swapfile
        mode: '0600'
      when: not swap_file.stat.exists
      become: yes

    - name: Make swap
      command: mkswap /swapfile
      when: not swap_file.stat.exists
      become: yes

    - name: Check current swap status
      command: swapon --show
      register: swap_status
      changed_when: false
      become: yes

    - name: Enable swap
      command: swapon /swapfile
      when: "'/swapfile' not in swap_status.stdout"
      become: yes

    - name: Add swapfile to fstab
      lineinfile:
        path: /etc/fstab
        regexp: '^/swapfile'
        line: '/swapfile none swap sw 0 0'
        state: present
      become: yes

    - name: Get GID of the deploy group
      getent:
        database: group
        key: "{{ deploy_group }}"
      register: deploy_group_info
      become: yes

    - name: Set deploy_group_gid fact
      set_fact:
        deploy_group_gid: "{{ deploy_group_info.ansible_facts.getent_group[deploy_group][1] }}"
      when: deploy_group_info.ansible_facts.getent_group is defined and deploy_group in deploy_group_info.ansible_facts.getent_group

    - name: Ensure deploy_group_gid is set to a valid value
      set_fact:
        deploy_group_gid: "0"
      when: deploy_group_gid is not defined or deploy_group_gid == ""

    - name: Configure system limits
      copy:
        src: "configs/etc/sysctl.d/99-system-limits.conf"
        dest: "/etc/sysctl.d/99-system-limits.conf"
        owner: root
        group: root
        mode: '0644'
      become: yes
      register: limits_sysctl_config_copy

    - name: Apply sysctl settings for system limits
      command: sysctl --system
      become: yes
      when: limits_sysctl_config_copy.changed

    - name: Create logs directory structure relative to deployment
      file:
        path: "./logs/yt-dlp-ops/communication_logs"
        state: directory
        mode: '0755'
        owner: "{{ ansible_user }}"
        group: "{{ deploy_group }}"
      become: yes

    - name: Build local Docker images (e.g., camoufox)
      ansible.builtin.command: >
        docker compose --project-directory . -f configs/docker-compose-ytdlp-ops.yaml build
      args:
        chdir: "{{ airflow_worker_dir }}"
      become: yes
      become_user: "{{ ansible_user }}"
      register: docker_build_result
      changed_when: "'Building' in docker_build_result.stdout or 'writing image' in docker_build_result.stdout"

    - name: Pull pre-built Docker images for ytdlp-ops services
      ansible.builtin.command: >
        docker compose --project-directory . -f configs/docker-compose-ytdlp-ops.yaml pull --ignore-buildable
      args:
        chdir: "{{ airflow_worker_dir }}"
      become: yes
      become_user: "{{ ansible_user }}"
      register: docker_pull_result
      retries: 3
      delay: 10
      changed_when: "'Pulling' in docker_pull_result.stdout or 'Downloaded' in docker_pull_result.stdout"

    - name: Show docker pull output
      ansible.builtin.debug:
        var: docker_pull_result.stdout_lines
      when: docker_pull_result.changed

    - name: Ensure Airflow project directory is writable by the container user (UID 50000)
      ansible.builtin.file:
        path: "{{ airflow_worker_dir }}"
        owner: 50000
        group: 50000
      become: yes

    - name: Ensure Airflow subdirectories are writable by the container user (UID 50000)
      ansible.builtin.file:
        path: "{{ item }}"
        owner: 50000
        group: 50000
        recurse: yes
        state: directory
      loop:
        - "{{ airflow_worker_dir }}/dags"
        - "{{ airflow_worker_dir }}/logs"
        - "{{ airflow_worker_dir }}/plugins"
        - "{{ airflow_worker_dir }}/config"
      become: yes

    - name: Create .dockerignore on worker to exclude runtime data from build context
      ansible.builtin.copy:
        dest: "{{ airflow_worker_dir }}/.dockerignore"
        content: |
          # Exclude build artifacts and virtual environments
          __pycache__/
          *.pyc
          *.pyo
          .venv/
          venv/

          # Exclude sensitive information
          .env
          .vault_pass

          # Exclude local development and OS-specific files
          .DS_Store
          .idea/
          *.swp

          # Exclude large directories with runtime data that should not be in the image
          logs/
          downloadfiles/
          addfiles/
          *downloads/
          postgres-data/
          redis-data/
          minio-data/
        owner: "{{ ansible_user }}"
        group: "{{ deploy_group }}"
        mode: '0644'
      become: yes

  tasks:
    - name: Install pipx
      ansible.builtin.apt:
        name: pipx
        state: present
      become: yes

    - name: Install Glances for system monitoring
      ansible.builtin.command: pipx install glances[all]
      args:
        creates: "{{ ansible_env.HOME }}/.local/bin/glances"
      become: yes
      become_user: "{{ ansible_user }}"

    # Include Docker health check
    - name: Include Docker health check tasks
      include_tasks: tasks/docker_health_check.yml

  roles:
    - ytdlp-worker