okd/playbooks/common/openshift-master/restart.yml

---
- include: ../openshift-cluster/evaluate_groups.yml

- name: Validate configuration for rolling restart
  hosts: oo_masters_to_config
  roles:
  - openshift_facts
  tasks:
  - fail:
      msg: "openshift_rolling_restart_mode must be set to either 'services' or 'system'"
    when: openshift_rolling_restart_mode is defined and openshift_rolling_restart_mode not in ["services", "system"]
  - openshift_facts:
      role: "{{ item.role }}"
      local_facts: "{{ item.local_facts }}"
    with_items:
      - role: common
        local_facts:
          rolling_restart_mode: "{{ openshift_rolling_restart_mode | default('services') }}"
      - role: master
        local_facts:
          cluster_method: "{{ openshift_master_cluster_method | default(None) }}"

# Creating a temp file on localhost, we then check each system that will
# be rebooted to see if that file exists, if so we know we're running
# ansible on a machine that needs a reboot, and we need to error out.
- name: Create temp file on localhost
  hosts: localhost
  connection: local
  become: no
  gather_facts: no
  tasks:
  - local_action: command mktemp
    register: mktemp
    changed_when: false

- name: Check if temp file exists on any masters
  hosts: oo_masters_to_config
  tasks:
  - stat: path="{{ hostvars.localhost.mktemp.stdout }}"
    register: exists
    changed_when: false

- name: Cleanup temp file on localhost
  hosts: localhost
  connection: local
  become: no
  gather_facts: no
  tasks:
  - file: path="{{ hostvars.localhost.mktemp.stdout }}" state=absent
    changed_when: false

- name: Warn if restarting the system where ansible is running
  hosts: oo_masters_to_config
  tasks:
  - pause:
      prompt: >
        Warning: Running playbook from a host that will be restarted!
        Press CTRL+C and A to abort playbook execution. You may
        continue by pressing ENTER but the playbook will stop
        executing after this system has been restarted and services
        must be verified manually. To only restart services, set
        openshift_master_rolling_restart_mode=services in host
        inventory and relaunch the playbook.
    when: exists.stat.exists and openshift.common.rolling_restart_mode == 'system'
  - set_fact:
      current_host: "{{ exists.stat.exists }}"
    when: openshift.common.rolling_restart_mode == 'system'

- name: Determine which masters are currently active
  hosts: oo_masters_to_config
  any_errors_fatal: true
  tasks:
  - name: Check master service status
    command: >
      systemctl is-active {{ openshift.common.service_type }}-master
    register: active_check_output
    when: openshift.master.cluster_method | default(None) == 'pacemaker'
    failed_when: false
    changed_when: false
  - set_fact:
      is_active: "{{ active_check_output.stdout == 'active' }}"
    when: openshift.master.cluster_method | default(None) == 'pacemaker'

- name: Evaluate master groups
  hosts: localhost
  become: no
  tasks:
  - fail:
      msg: >
        Did not receive active status from any masters. Please verify pacemaker cluster.
    when: "{{ hostvars[groups.oo_first_master.0].openshift.master.cluster_method | default(None) == 'pacemaker' and 'True' not in (hostvars
              | oo_select_keys(groups['oo_masters_to_config'])
              | oo_collect('is_active')
              | list) }}"
  - name: Evaluate oo_active_masters
    add_host:
      name: "{{ item }}"
      groups: oo_active_masters
      ansible_ssh_user: "{{ g_ssh_user | default(omit) }}"
      ansible_become: "{{ g_sudo | default(omit) }}"
    with_items: "{{ groups.oo_masters_to_config | default([]) }}"
    when: (hostvars[item]['is_active'] | default(false)) | bool
  - name: Evaluate oo_current_masters
    add_host:
      name: "{{ item }}"
      groups: oo_current_masters
      ansible_ssh_user: "{{ g_ssh_user | default(omit) }}"
      ansible_become: "{{ g_sudo | default(omit) }}"
    with_items: "{{ groups.oo_masters_to_config | default([]) }}"
    when: (hostvars[item]['current_host'] | default(false)) | bool

- name: Validate pacemaker cluster
  hosts: oo_active_masters
  tasks:
  - name: Retrieve pcs status
    command: pcs status
    register: pcs_status_output
    changed_when: false
  - fail:
      msg: >
        Pacemaker cluster validation failed. One or more nodes are not online.
    when: not (pcs_status_output.stdout | validate_pcs_cluster(groups.oo_masters_to_config)) | bool

- name: Restart masters
  hosts: oo_masters_to_config:!oo_active_masters:!oo_current_masters
  vars:
    openshift_master_ha: "{{ groups.oo_masters_to_config | length > 1 }}"
  serial: 1
  tasks:
  - include: restart_hosts.yml
    when: openshift.common.rolling_restart_mode == 'system'
  - include: restart_services.yml
    when: openshift.common.rolling_restart_mode == 'services'

- name: Restart active masters
  hosts: oo_active_masters
  serial: 1
  tasks:
  - include: restart_hosts_pacemaker.yml
    when: openshift.common.rolling_restart_mode == 'system'
  - include: restart_services_pacemaker.yml
    when: openshift.common.rolling_restart_mode == 'services'

- name: Restart current masters
  hosts: oo_current_masters
  serial: 1
  tasks:
  - include: restart_hosts.yml
    when: openshift.common.rolling_restart_mode == 'system'
  - include: restart_services.yml
    when: openshift.common.rolling_restart_mode == 'services'