So I have an Icinga 2 server with the web module and have deployed a host (ubuntu 18) with the director module.
I would like to monitor the memory, but I always get the same error message as above. When the check command is run should it be installed on the master as on the agent? In this the plugin check_mem.pl is only installed on the master and the agent connects to the master. I suppose check_mem.pl plugin should also be in the agent’s plugin directory, is this correct?
This is a screenshot from the webinterface of icinga 2. Where does the mem check command look for the file check_mem.pl because it is on both the Master and the agent.
Yes this is what I meant.
The check needs to exist on that host and I hope this is also the host, that this service is attached to as check_mem is a local check.
I didn’t meant the Object inspection of the CheckCommand but the Service.
I open them directly form the service in the Montioring module of Icingaweb2 and it looks like this:
Last question, if a check like the ping4 is performed on the Master server to see if a host is online, how does it do this to a host that is on another network? Do you then have to do portforwarding to that host on the client’s firewall for example? Isn’t the idea if you work with an agent that regardless of where the host is located comes to log in to the master and all I have to do is do a portforward on my own firewall that points to the master?
this way I don’t have to ask the client to change anything on their firewall but the client comes to our firewall, or am I seeing all this wrong?
Yes, to make ping work, you need functioning routing and firewalls allowing it.
You can change the host check to use the something like “Agent Health”.
In my setup this is a service that runs:
Do you run it from a satellite or from the host itself?
If you don’t run it from outside the host, the host can go down and you will not notice or you will get swamped by notifications that check from the outside like SSL/TLS, SMB/CIFS, HTTP and other checks.
Using a satellite would be the preferred method to check hosts in a other network I guess.
Also using ping in stead of a agent health check would be preferred as the host isn’t necessarily down if the icinga service isn’t running.