setup_hadoop_distribution.yml

---
- name: Check if hadoop in installed
  stat: "path=/home/{{ ansible_user }}/hadoop-{{ hadoop_version }}/bin"
  register: hadoop_installation_path

- name: Download hadoop distribution in master if not installed in host
  get_url:
    url: "https://dlcdn.apache.org/hadoop/common/hadoop-{{ hadoop_version }}/hadoop-{{ hadoop_version }}.tar.gz"
    dest: "/home/{{ ansible_user }}/hadoop-{{ hadoop_version }}.tar.gz"
    checksum: "sha512:{{ hadoop_sha512_checksum }}"
  delegate_to: master
  when: hadoop_installation_path.stat.isdir is not defined or not hadoop_installation_path.stat.isdir
  throttle: 1

- name: Sync hadoop from master to worker nodes
  ansible.posix.synchronize:
    src: "/home/{{ ansible_user }}/hadoop-{{ hadoop_version }}.tar.gz"
    dest: "/home/{{ ansible_user }}/hadoop-{{ hadoop_version }}.tar.gz"
    mode: push
  delegate_to: master
  when:
    - inventory_hostname != 'master'
    - hadoop_installation_path.stat.isdir is not defined or not hadoop_installation_path.stat.isdir
  become: no

- name: Decompress the hadoop distribution
  unarchive:
    src: "/home/{{ ansible_user }}/hadoop-{{ hadoop_version }}.tar.gz"
    dest: "/home/{{ ansible_user }}"
    remote_src: yes
    creates: "/home/{{ ansible_user }}/hadoop-{{ hadoop_version }}/bin"
  register: installation_task

- name: Remove local copy of hadoop download if successfully installed
  file:
    path: "/home/{{ ansible_user }}/hadoop-{{ hadoop_version }}.tar.gz"
    state: absent
  when: installation_task is succeeded or installation_task is skipped

- name: Symlink decompressed hadoop directory to hadoop home
  file:
    src: "/home/{{ ansible_user }}/hadoop-{{ hadoop_version }}"
    dest: "{{ hadoop_home }}"
    state: link