Nagios Support Forum

Since quite some time i am using NagiosXI as main central and use multiple (remote network) nagios core machines to create a distributed monitoring system. Some time ago i asked for mod gearman support which wasn't available at that time. However since its now supported i do have some questions.

1. At the moment i am using NRDP to send the data from the local nagios core machines to the central. Do i have to disable this sending to use mod gearman?
2. Will there be a debian version of the (worker) installer script?
3. If i read correctly it should be possible to have all configs on the main central and use the core machines as a sort dumb terminals. Is this a correct assumption? I am looking for a way to only configure a machine 1x instead of 2x that i have to do now.
4. Seeing as i am using a distributed system at the moment (and thus recieving passive checks). Would i benefit from using mod gearman?
5. How can i handle the dummy checks in nagiosxi. At the moment i have a extra nagios core installation which handles the dummy checks and sends them passively to the main central. Because i work with servicechecks on hostgroups i havent found an easy way to solve this on the nagiosxi installation.

Thanks for the answers.

1. At the moment i am using NRDP to send the data from the local nagios core machines to the central. Do i have to disable this sending to use mod gearman?

There's nothing that says you have to. Mod Gearman is a way to distribute active checks. Since NRDP is all passive, you could leave all of your current passive checks in place if you wanted.

2. Will there be a debian version of the (worker) installer script?

We don't have plans for this at the moment...

3. If i read correctly it should be possible to have all configs on the main central and use the core machines as a sort dumb terminals. Is this a correct assumption? I am looking for a way to only configure a machine 1x instead of 2x that i have to do now.

This is the key difference between active and passive checks. Active check configuration is all done on the Nagios server, while passive checks require some configuration on the remote machine when changes need to be made.

4. Seeing as i am using a distributed system at the moment (and thus recieving passive checks). Would i benefit from using mod gearman?

Centralized config management is about the only reason I can think of for switching over. In my opinion, if it's not broken, don't fix it. It'd be a pretty big job to switch everything over, so you'll have to evaluate if the central management is worth it.

5. How can i handle the dummy checks in nagiosxi. At the moment i have a extra nagios core installation which handles the dummy checks and sends them passively to the main central. Because i work with servicechecks on hostgroups i havent found an easy way to solve this on the nagiosxi installation.

I'm not really sure what you mean by dummy checks. If you've currently got Core servers doing the bulk of the work, and then forwarding those results up to XI, that changes my remark for item #4. If the dummy servers are doing their own active checks by hostgroups, then you could probably just setup those machines are gearman workers pretty easily since they're already configured to check by hostgroup.

Thanks for the answers.

With the dummy checks i mean the following. Some hosts arent pingable (non-managable switches as example). Since we use a general group that does pinging these hosts will give criticals. So what we did was make a service check named the same and use as check a dummy ok check. This works fine on the core machines because there i can add a !host to the service. On the nagioxi server i dont have that option though to ignore a host in a group but still have him get the same services.

I hope this is understandable.

So, if I am understanding this right.. what you really want to do is distribute those ping checks across all services in a host group? Or are you looking to distribute all checks possible, on to mod gearman workers?

Well, i would like to have the best configurable solution since now i have to do everything 2x (on the core and on the central). I know mod_gearman will be tricky to setup at first but afterwards it should save time i presume.

As i explained before the only problem i am facing with the central are the dummy checks but i am reasonable sure i can work around those with a new way of grouping (because of the single configuration possibility).

The other problem will be the debian core installs we are using. Because of this i will have to create my own installation script to get this to work.

There is a workaround to !host exclusions with the Core Config Manager. If you have a service->hostgroup assignment, as long as you don't have any service->host relationships for that service, you can actually add the following as a custom variable, and it will be recognized correctly with Core.

host_name !hostA,!hostB,!hostC

I assume i would add this to the service itself?

I was thinking of creating a parent -> child tree where the main parent for example would be named Firewalls and have no services on it. Then create 2 childs where 1 would have the ping service and the other have the dummy ping service.

The only downside from this would be more hostgroups to view although we hardly use them anyway.

I assume i would add this to the service itself?

Yes, that's correct.