Using pause module the perform reboot during playbook

[Gregory Shulov]

Hi All, 

I am trying to update kernel in playbook and reboot the server without existing playbook.

Here is the example:

# Temprorary hack that fixes mismatched kernel and kernel module versions

- name: "Restart VM to Load new kernel"

  action: command /sbin/reboot

  only_if: "${foo.stdout} != 0"

  

- name: "Pause playbook and wait for VM to boot"

  action: pause minutes=2

  only_if: "${foo.stdout} != 0"

Here is the output:

  

TASK: [Restart VM to Load new kernel] ********************* 

changed: [1.1.1.10]

TASK: [Pause playbook and wait for VM to boot] ********************* 

(^C-c = continue early, ^C-a = abort)

[1.1.1.10]

Pausing for 120 seconds

fatal: [1.1.1.10] => Failed to open session: SSH session not active

Is it even possible ?

[Jeroen Hoekx]

Hello,

On 21 October 2012 14:08, Gregory Shulov <gregory...@gmail.com> wrote:

> I am trying to update kernel in playbook and reboot the server without
> existing playbook.
>
> Here is the example:
>
> # Temprorary hack that fixes mismatched kernel and kernel module versions
> - name: "Restart VM to Load new kernel"
> action: command /sbin/reboot
> only_if: "${foo.stdout} != 0"
>
>
> - name: "Pause playbook and wait for VM to boot"
> action: pause minutes=2
> only_if: "${foo.stdout} != 0"

What you need is the wait_for module.

action: command /sbin/reboot
local_action: wait_for host=${inventory_hostname} port=22 state=stopped
local_action: wait_for host=${inventory_hostname} port=22

This should wait until SSH is available. If that does not work
reliably, the host is down before the first wait_for launches for
example, you can remove the first wait_for and add a 'delay' parameter
to the second one.

Greetings,

Jeroen

[Geoffrey Widdel]

This does not work...please test your code before posting.

[Christopher O'Connell]

Here's a snippet from a playbook I use to update CentOS machines to the CentOSPlus kernel including a reboot (if necessary)

- name: update the kernel
  yum: name=kernel* state=latest
  register: kernelup
- name: reboot the system
  command: reboot
  when: kernelup.changed
- name: wait for ssh to come back up
  local_action: wait_for host={{ ansible_fqdn }} port=22 delay=120 timeout=900 state=started
  when: kernelup.changed

The nice thing about this is that it *won't* reboot the machine if the kernel was already up to date.

(Obviously, in this example the osplus repo needs to have been enabled previosly, using lineinfile or by sending a custom repo definition file. Nonetheless, these commands should work fairly well).

All the best,

~ Christopher