[Nut-upsuser] Shutdown order for servers

[Manuel Wolfshant]

On September 18, 2020 11:02:48 AM GMT+03:00, Kirill Nikonorov 
<knikonor at gmail.com> wrote:
>Hi Manuel,
>
>thank you for pointing this. I know about upsshed but I have no idea
>how to
>let dependable server to discover that another server was shut down.
>Any
>examples of such scripts?
>
The nut server will trigger ( via nut client ) upssched on the slaves. 
Since I usually approach most things bottom up, I guess that for me the 
easiest method would probably be to use arping or ping (something along 
ping -c1 $target; echo $? ) to verify ( from the script launched by 
upssched ) if the monitored servers are still up. But I am sure that 
there are a ton of ways to test, even doing an application-level ( L7 ) 
verification rather than a L2 or L3 as I suggested above. Not to mention 
that dedicated apps exist, of course.
It's up to you to pick the method most appropriate for your use case.


wolfy

>чт, 17 сент. 2020 г. в 21:00, Manuel Wolfshant
><wolfy at nobugconsulting.ro>:
>
>> On September 17, 2020 6:09:18 PM GMT+03:00, Kirill Nikonorov <
>> knikonor at gmail.com> wrote:
>> >Hello to All!
>> >It might be not exactly NUT question but implementation.
>> >I have a rack with one UPS and several servers getting power from
>this
>> >device. One server is a master and other are slaves. All systems are
>> >running Debian 4.15.3 kernel and APC UPS. Everything works fine but
>> >what I
>> >need is to make an order for shutting down servers as they have
>> >different
>> >roles.
>> >Is it possible to configure SQL and VM servers to go down first,
>then
>> >mail
>> >server, then NAS and UPS master as the last one after all of them?
>> >I see that in case of power outage all servers receive signal from
>> >master
>> >simultaneously. May I change the order somehow?
>>
>>
>> Yes, you xan do that.. You will need to write a specific script on
>each
>> aerver and run it via nut's client upssched directive
>>
>> Wolfy
>> >