Upgrades and reboots - question

[Patrick Galbraith]

Hi all!

We've been working on a playbook that does a dist upgrade and then performs a reboot. This has been problematic and results in a failure return status that sometimes means it worked and sometimes not. What is the trick to reboots -- being able to smoothly reconnect after a reboot signal?

---

- name: reboot the server

  command: shutdown -r now "Ansible updates triggered"

  async: 0

  poll: 0

  ignore_errors: true

  tags:

    - reboot

- name: waiting for server to come back

  sudo: false

  local_action: >

    wait_for:

      host: {{ inventory_hostname }}

      timeout: 300

      delay: 0

      port: 22

      state: started

  tags:

    - reboot

Is there a technique that is commonly used that results in better plays that result in success?

Regards,

Patrick 

[J Hawkesworth]

Just a guess but you might want to have a few seconds of delay set in your wait_for so that there is time for network services and sshd to come back up before ansible goes looking for it.

[Azul Inho]

This works on my machine,

- name: wait for server to come back

  local_action: shell while true; do echo "Waiting ..." ; ssh -o ConnectTimeout=5 -o BatchMode=yes  {{ inventory_hostname }} pwd ; [ $? -eq 0 ] && break || sleep 5; done

  sudo: false

 

[John Batty]

Your server may still be in the process of shutting down when your "wait_for" is executed.  So the wait_for could make a connection and think the server is back up when it is actually on its way down...

Try setting the delay parameter in the wait_for (e.g to 15) to ensure that the server is down/rebooting before waif_for starts checking.

John