NagiosXI as Poller & Collector

[sbaviswa]

Hi

Would like to have some documents to configure NagiosXI into poller & collector.
The scenario:
There will be a central nagiosxi server & DB server working as collector + nagiosxi pollers in all the branch datacenters.
The pollers must poll the metrics of devices under their scope. Later the centralized collector must receive all the metrics and stored centrally.

Any help is welcome.

Regards
Viswa

[tmcdonald]

It actually sounds like Nagios Fusion might be a better fit:

https://www.nagios.com/downloads/nagios-fusion/

We developed it basically for the same reason, to be a central collector/view for all monitored objects.

[sbaviswa]

Thanks for the reply.

The idea of fusion is just to collate the views/dashboards of all associated nagiosxi instances. Which means the data resides only on the distributed nodes & not centrally.

Our customer would like to have a true distributed setup where the data is collected centrally from all the other distributed nodes. Hence ruggedness in terms of db replication, etc can be done in the central location.

I can quote an example like how FullyAutomatedNagios operates --> A distributed nagios core installations with a central centureon server & db

If I use fusion (incase nagiosxi does not suite for poller-collector configurations), then all the distributed nodes must be built will ha & db replication. Is that correct?

regards - viswa

[karthek]

Then you should have a look at DNX method.

https://assets.nagios.com/downloads/gen ... utions.pdf

I bet this is what you are looking for.

#####EDIT#####

Few more solutions:

https://assets.nagios.com/downloads/nag ... ios_XI.pdf
https://labs.consol.de/nagios/mod-gearm ... od_gearman

[sbaviswa]

DNX is an alternate method of mod_gearman. It just spreads the checks across different worker machines.

To my understanding :
1. DNX & mod_gearman are for distributing the load of the nagios server (in terms of checks processing)
2. MNTOS is for combining the tactical overviews of multiple nagios instances
3. fusion is for combining the views/ dashboard of multiple nagios instances

None of the method I understand can be configured as a poller-collector setup in a truely distributed environment.

So any suggestions as I mentioned earlier is welcome.

[eloyd]

Mod_gearman does exactly what you're asking. From the document referenced:

Mod_Gearman and its worker processes intercept checks made by Nagios and run them externally, process them, and send them back to the Nagios server

It does not hold the results of the check, and it is not able to be queried or display stats or anything else - it just does the work and reports back to Nagios where the results are processed, notifications are sent, and reports generated.

Otherwise, you are, indeed, talking about multiple Nagios servers, which you could then use DNX to send back to a central server for reporting and alerting.

[sbaviswa]

@ eloyd ---> can the DNX worker process servers be located in a different datacenter, get the local monitored data and send it to the central nagios server located somewhere else.
Request you to see my previous post..
To my understanding (correct me if I am wrong) the checks gets dispersed (intercepted) to the DNX servers from the nagios server. Hence all monitored data will first reach to nagios server and then spread.
But my requirement is that all monitoring to be done locally at the respective datacenters and the data must be pushed to the central server at specified intervals.

[eloyd]

Here's how I would do it:

• Assume you have two data centers (just for this example) plus one central location.
• Set up a Nagios XI server at each data center, plus one at your central location.
• Ensure NRDP is enabled on the central Nagios server.
• Each data center Nagios server performs all checks local to each data center.
• Duplicate all hosts and services from data center Nagios servers on the central server as passive checks.
• Set event handlers on each data center Nagios server checks to use NRDP to send the check result back to the central Nagios server simultaneously.
• (Optionally, you could use cron to read /usr/local/nagios/var/nagios.log and duplicate sending commands that way)

You now have three Nagios servers, two of which are performing work and sending alerts in data centers, and a third which is simply collecting all service data in real time as passive checks, but not doing anything with it except being able to be used for reports, etc.

[sbaviswa]

Super eloyd... I think this sounds interesting.
Give me some time to try this method in the test environment and give feedback on how it works.

BTW in this case the central server can be considered as a DR location too right...

[eloyd]

The central server would, by definition, have the same service checks on it as the data center servers, they would just be defined as passive. You would have to force them to check in a DR scenario, or else you would have to define them as active but disabled and then enable them in a DR scenario.