Cluster configuration : Change state when agent is not connected

Hello !

I am using icinga with one master and many agent (almost 300 agent with 20 services per agent). All the monitoring is based on my cmdb.

I have a question,

Sometimes some servers are going to die (after 30 minutes), but the master can still join them (hostalive / ping) during this time. All the services check are not coming to the master so there is a lag and I have the message “Remote Icinga instance HOSTNAME is not connected to master” on all services (20) on the host with state UNKNOWN.

So my problem is I dont get any critical (only after 30 mimutes when the server is really dead because of the hostalive check). I would like to know if it’s possible to change the state when I have a message “Remote icinga is not connected …” to critical instead of unknown ? Or maybe, of all services are in stage unknown then change host state to critical ?

Of course if you have a better solution I would love to know it !

Thank you :slight_smile:

Hi & welcome to the icinga community,

just replace hostalive with cluster-zone as check command.

2 Likes

I will sleep less stupid tonight, it’s working.

Thank you ! :slight_smile: