Dependency with two parents

Good morning,

I want to make a dependency between my satellite and two WAN connections.

My setup:

The active Master checks the cluster (satellite reachable) and the two WAN connections.
My target is to built a depedency to disable check and notification for the cluster, if only both wan connections lost.

Sounds like a job for check_multi.

2 Likes

check_icmp should be able to handle this and ping both addresses by adding a second one to the -H parameter and

-m
    number of alive hosts required for success
check_icmp -H 8.8.8.8 9.2.5.99 10.10.10.10 -m 2
./check_icmp -H 8.8.8.8 9.2.5.99 10.10.10.10 -m 2
OK - 8.8.8.8: rta 9,166ms, lost 0% :: 9.2.5.99: rta nan, lost 100% :: 10.10.10.10: rta 19,070ms, lost 0%
|8.8.8.8rta=9,166ms;200,000;500,000;0; 8.8.8.8pl=0%;40;80;; 8.8.8.8rtmax=9,549ms;;;; 8.8.8.8rtmin=8,910ms;;;; 9.2.5.99rta=0,000ms;200,000;500,000;0; 9.2.5.99pl=100%;40;80;; 9.2.5.99rtmax=0,000ms;;;; 9.2.5.99rtmin=0,000ms;;;; 10.10.10.10rta=19,070ms;200,000;500,000;0; 10.10.10.10pl=0%;40;80;; 10.10.10.10rtmax=20,432ms;;;; 10.10.10.10rtmin=18,516ms;;;;

But there is an issue with check_icmp if it come from the nagios-plugins package (on CentOS at least), that the calculation of the min alive is not working. I already opened a github issue, but no response until now: https://github.com/nagios-plugins/nagios-plugins/issues/404

Therefore on my CentOS system I switched to check_mping https://exchange.icinga.com/exchange/Ping%20of%20multiple%20Hosts%20via%20icmp%20protocol

1 Like

Or business process logic, or the DSL cluster checks. The multi parents discussion is still going on on GitHub, I am out of ideas for now.

3 Likes

I’d go for the business process module, too.

1 Like