[icinga-users] Using cluster-zone check for host state

Markus Joosten markus.joosten at plumbe.de
Mon Nov 30 17:47:46 CET 2015

> From:Per-Henrik Lundblom <ph at whatever.nu>
> Sent: Monday 30th November 2015 15:58
> To: icinga-users at lists.icinga.org
> Subject: Re: [icinga-users] Using cluster-zone check for host state
> * Per-Henrik Lundblom <ph at whatever.nu> [151124 13:52]:
> > Instead I want to use the cluster transport as a way to detect host
> > state. Therefore I change check_command for the hosts client[12] to
> > "cluster-zone" which doesn't work. I assume this is because the host
> > state check_command is run on each host, not on monitor1. Thus, when the
> > cluster-zone check fails on either client, it is actually detected but
> > it is not communicated to monitor1 and the client host state remains UP.
> Judging from the lack of replies, I assume it is impossible to contol
> the host state check_commands in this way?
Checks are always executed from an endpoint in the respective zone the host and service objects are defined in.
Therefore you would have to create additional host objects for your clients in your *master* zone. (This also applies to checks via ping, since you need to execute the checks from your master, and not from the client!)
These additional hosts in your master zone could then use "cluster-zone" as their check-command.

The cluster-zone check defined on your endpoints in the satellite zones are still useful, since you also might want to check the connectivity from the satellites, in case your master goes down.


More information about the icinga-users mailing list