Nagios distributed monitoring

[whitest]

Is here stable and powerful solution to build distributed monitoring system?
I have 5 servers (1 master and 4 slaves). Slaves send chack results via mod_gearman to master. But Germand works not stable. I also tried nscaweb. It's not good also.
When Nagios will have built-in mechanisms to speak with each otter???

[jdalrymple]

When Nagios will have built-in mechanisms to speak with each otter???

Probably never. The core would have to be rewritten.

We have many customers performing in excess of 20,000 service checks on a single host using today's available distribution methods. We have not yet found a limit to our scalability using today's methods. Perhaps someone has, but if they have they didn't tell us.

If you have found that limit can you please elaborate so we can understand better what it takes to overwhelm today's distribution methods?

[whitest]

jdalrymple, TY for you response!
My problem is not limit. My problem is stable of distribution solutions
Germand often hungs up. Could you please to recommend me your favourite solution?

[tmcdonald]

What end goal are you looking to achieve? Are you seeing a high load on the Nagios machine and wishing to split off checks for performance reasons? Is there a network architecture that requires you to have gearman workers in different parts of the network? Or do you just want to see the checks from multiple Nagios machines in one interface?

[whitest]

Network architecture is the reason to use several nagios servers. Each nagios-server perform cheks and notifies in own network-region. There is main-nagios server in Head-office. Main nagios receives results of checks from region-negioses. So main-nagios display status of entire network. Main nagios hasn't access to hosts in regions. It's network limitations.
I don't use geaman-wokrkers. I use only send_gearman (on region-nagioses) as replacement of NSCA.

[tmcdonald]

We developed a product called Nagios Fusion for just this sort of thing:

https://www.nagios.com/products/nagiosfusion

We offer a free, fully-functional 60-day trial on all of our products. I would strongly suggest taking a look since it was designed for exactly what you need.

[whitest]

tmcdonald, thank you! I'll try.

[klajosh2]

tmcdonald,

I am also intersted in this topic.
So do I understand correctly that Nagiosfusion is the answer of distributed monitoring from Nagios?
So I have a main nagios core server which dispatch the checks to nagiosfusion instances? (same way as mod_gearman does)
Or is this the opposite? Many nagios core instances and nagiosfusion controls the execution of the checks?

thanks,

klajosh

[abrist]

So do I understand correctly that Nagiosfusion is the answer of distributed monitoring from Nagios?

Yes and no. Consider fusion as a aggregate view of multiple nagios servers. Distributed checks are more of a mod_gearman thing, though, with enough nagios boxes, fusion could be seen as the central view of your nagios nodes.

So I have a main nagios core server which dispatch the checks to nagiosfusion instances? (same way as mod_gearman does)

No, it is just a central view for the status results for all your nagios servers.

Or is this the opposite? Many nagios core instances and nagiosfusion controls the execution of the checks?

Nope, It is just a view, each nagios box must be configured individually and each schedule and run their own checks.

[klajosh2]

abrist:
Thank you for your reply. So as a conclusion Nagiosfusion is not for me.
I have also 5 mod_gearman poller/collector (call whatever you want) in 4
different geographic location communicating with Nagios core as a main server.

It works more or less fine but sometimes I notice not explainable errors like
delays in checks. I think the Nagios core's scheduling mechanism is not
prepared to work with such type of configuration.

On the another hand I like the way how I can centrally configure the host/service checks in one
place.