In my infrastructure I have servers, which periodically do some heavy work. Let's say, that every 10 minutes the cpu utilization is 100% and it lasts no longer than 5 minutes. If the servers work in this state for longer than 15 minutes, then I want to be notified, otherwise peak utilization for a short period of time is normal.

Because of that, I created a special host class in
Code:
$SHINKEN_ETC/packs/company_packs/template.cfg
which looks like this:

Code:
(...)
define host {
        name                            linux_1min_15tries_every1min
        use                             linux
        max_check_attempts              15
        check_interval                  1
        retry_interval                  1
        flap_detection_enabled          0
        register                        0
}
(...)
Then I use it in
Code:
$SHINKEN_ETC/hosts/company_hosts.cfg
like this:

Code:
(...)
define host{
        use                     linux_1min_15tries_every1min
        contact_groups          admins
        host_name               serv1
        address                 10.10.16.1
        }
(...)
Meanwhile the host goes into CRITICAL state in Shinken every time the utilization peaks. It looks like it reaches the HARD state after the first check. Also, the counter doesn't ever go up. It always stays at 1/15 lvl. Any ideas how to fix that?

Shinken version is 1.4.2

Screenshot from 2015-08-07 16-41-12.jpg