reboot_handler

Handle reboot notification, reboot remote UNIX/Linux machines and wait them back.

This handler listens a message of "reboot system", provides variables to skip reboot or skip the confirmation before rebooting.

Requirements

Ansible >= 2.2, we are listen the notifications
jinja2 >= 2.6
permissions to become root
remote machines must be rebootable via shutdown or reboot commmands
the connections of remote machines must be in ssh, smart, paramiko or funcd

Role Variables

Variable skip_reboot, default True. This variable should be defined in host/group vars. The reboot action is dangerous in many situations, so we will not reboot the machine unless the skip_root for a inventory_hostname is set to True.

Variable reboot_confirm, default True. if set to True, the handler will trigger a prompt once for a batch, the prompting message contains the list of rebooting machines in this batch. Press Ctrl+C and A to abort the playbook, or press Enter key to continue.

Dependencies

role gzm55.require_implicity_localhost
role gzm55.require_disabe_become
role gzm55.local_id_plugin
role gzm55.require_become_root

Example Playbook

- hosts: servers
  roles:
  - { role: user.reboot_handler, reboot_confirm: True }
  tasks:
  - name: change system
    raw: 'true'
    notify: "reboot system"

Developing

The testing progress is based on molecule with vagrant driver which is not supported by travis-ci. So in each release, we have to run molecule test on a developing machine, and then galaxy import manually.

TODO: use wait_for_connection module (ansible >= 2.3) to wait machines comming back.

License

BSD

gzm55 / ansible-reboot-handler-role