How to concatit ansible-playbook on a server that was turned off?

Imagine such a scenario.
We have 1000 servers. All you need to update your version of Nginx (say security bugfix).
But we have one of the servers is off right now because he burned down the power unit.
Updated version of ansible nginx playbook and run it on all servers. Yeah, we all pre-tested on test environment.
Accordingly, nginx will be updated on the 999 servers, and one is not (it is off).
The power supply was repaired, the server is enabled, and HERE it is IMPORTANT, it is the old non-updated nginx.
It's good that I remember that it is not updated nginx. After switching on I can re-lose playbook. BUT it's done by hand. What if I have 1,000 was off 10 and they vkluchaetsia at different times by different employees?

I want the automatics!

The crux of the matter is this: as for starting the server automatically to play on it last actual playbook and only after that to put it into operation?

For example, we have ha-proxy is configured healt check and if the node is unavailable, then requests for her not Sluts.
And if available, requests for it are. In the example above, from the point of view ha-proxy walking gcd is quite in order. Responds to requests normally. Only here the version of nginx there leaky.
To nginx please do not cling. Updated could be absolutely any software or library. This is the essence of the matter does not change.

Who as solves this problem?

Or maybe I've got the whole approach wrong? If Yes, then tell me in what direction to look?
July 9th 19 at 13:34
3 answers
July 9th 19 at 13:36
ansible-pull will not work?
July 9th 19 at 13:38
Ansible will compare what is already on the server and if execution does not require a job to miss. Just run it again.
It is clear. But I wrote that I do not want hands to start again. What if the server will include in a week? I'll forget it between switching off and on of being updated. Want to eliminate the human factor. Moreover, many of the staff and the possibility of human error increases.

As I wrote the question - I repeat again "I want-automatic" - Taya_Wintheiser88 commented on July 9th 19 at 13:41
: Ansible is not a monitoring system. He has no idea when you got there, woke up the server. - Jaqueline.Parisian commented on July 9th 19 at 13:44
: Put Ansible locally, it will be the simplest option. Consider taking on a Central server latest script and run. - Jaqueline.Parisian commented on July 9th 19 at 13:47
: Yes, I do this just thought. But torment vague doubts that there are some other generally accepted way. - Taya_Wintheiser88 commented on July 9th 19 at 13:50
Judging by the answers, otherwise nothing. - Jaqueline.Parisian commented on July 9th 19 at 13:53
July 9th 19 at 13:40
Well run playbook for the crown. Or a third-party solution to fence with a ping from the server.
I'm interested in - are there common third-party solution for the given scenario. Doesn't anybody have such problems? How it all turned out? - Taya_Wintheiser88 commented on July 9th 19 at 13:43
: well, you can still pool ansible to deploy to the nodes , described in of the dock
but essentially the same path through the crowns , only on the nod - Jaqueline.Parisian commented on July 9th 19 at 13:46

Find more questions by tags AnsibleSystem administrationOperating systems