Event Command runs on Master and Satellite

Icinga 2
1

Hi togehter,
after checking the log i will find out that an event command runs on the master and the satellite.
But i need this event commad only on the master.
The log on the satellite shows me this line:

[2021-01-14 13:56:37 +0100] warning/PluginEventTask: Event command for object ‘xxxx-server.de!GENESYS_SERVER_ALARM’ (PID: 10880, arguments: ‘/usr/lib/nagios/plugins/eventhandler/notification_flagfile.pl’ ‘-h’ ‘xxxx-server.de’ ‘-n’ ‘GENESYS_SERVER_ALARM’ ‘-s’ ‘CRITICAL’) terminated with exit code 128, output: execvpe(/usr/lib/nagios/plugins/eventhandler/notification_flagfile.pl) failed: No such file or directory

I’d created the eventhandler in director:

eventhandler
eventhandler743×385 10.9 KB

object EventCommand "set-notification-flag" {
    import "plugin-event-command"
    command = [
        "/usr/lib/nagios/plugins/eventhandler/notification_flagfile.pl"
    ]
    arguments += {
        "-h" = "$host.name$"
        "-n" = "$service.name$"
        "-s" = "$service.state$"
    }
    vars.rpc_service_name = "$service.vars.rpc_service_name$"
    vars.rpc_service_state_id = "$service.state_id$"
}

Then i added the new eventhandler to the service template:
template_passive

template Service "Passive Service Volatile mit Notification Flag Template" {
    check_command = "trap-passive"
    max_check_attempts = "1"
    check_period = "timeperiod-24x7"
    check_interval = 1d
    enable_notifications = true
    enable_active_checks = true
    enable_passive_checks = true
    enable_event_handler = true
    enable_flapping = false
    enable_perfdata = false
    event_command = "set-notification-flag"
    volatile = true
    vars.enable_trap_notification = true
}

At last the service:
service

How can i configure the eventhandler that the script will only run on ther master zone?

ICINGA is running on Debian 9:

  • ICINGA Web: 2.8.2
  • director: master
  • Icinga: r2.12.3-1

Regards
Sascha

2

This is impossible at the moment, eventhandler will always be executed where the check is executed.

3

Hi Dirk
Thanks for you reply and the links.
The documentation is clear, but i do not understand why the master node run the same eventhandler.
I‘m using the eventhandler since last year and the master works like i want. Today i‘d checked the logs on the satelite and found the error messages.
Regards
Sascha

4

Unfortunately it is a simple run on the agent endpoint if the check_endpoint is set and run on parents if not. It should be at least run in the zone the service is part of for consistent behaviour which would exactly not what you want, but it is not implemented like this. :frowning: It also runs on all of them perhaps resulting in multiple restarts of a service or even more bad things, so we can only hope the developers will find a proper solution in the future. So it is totally on Icinga and not you or your configuration.

5

Hi Dirk,
ok - then I hope that the developers will soon process the open ticket. I believe that an ICINGA setup with multiple zones is not uncommon. Thanks again for the clarification

6

I also have this problem right now with an event handler, which is configured in the global-templates zone. The check I am running is executed by an satellite. It is a check by_ssh, so there is no check_endpoint defined. The satellite connects to the client via SSH and executes a plugin there.
My thought based on the docs was, the event_command is only running on the satellite, because the satellite executes the check. But unfortunately both the master and satellite execute the event_command, what leads to a double launch of a recovery job in AWX (triggered by the AWX API). My dirty workaround is only having the event_command script on the master. The satellite throws an error in the logs with “file not found”.
Are there any plans to give the event commands a little bit love? :slight_smile:

Cheers,
Marcus