Nagios Support Forum

I've been using a mix of NSCient for windows & nrpe for Linux/Unix servers
nsclient has been pretty straightforward with the download & install of their msi. NRPE however has been problematic in that internet access is required to compile & install - unless you can find an rpm for the OS version on that server.

NCPA seems to have come a long way , so I was wondering if this is the way forward? What can the other agents do that NCPA can't and how much development effort is Nagios putting into NCPA ?
I see that nsclient is also providing a version now for Linux.

The other pain point is the deployment of the agents. Windows isn't too bad as we can use SCCM if available or use Global Domain policies. Again , it's the Unix/Linux servers that are difficult to deploy.
I like the idea of NRDS, however this is only useful for Passive monitoring.

I'm planning a new Nagios Installation for 500+ servers (50/50 Windows & Unix) so am looking for a pathway to move towards for the agent deployment & installation.
Is NCPA the preferred solution, or NSClient for Window/Linux. If NRPE has features that isn't matched by the other agents, then maybe a mix of NSClient & NRPE as I have now?

Also interested to know if any development is happening with NRDS to allow the deployment of these agents for use with Active monitoring?

regards... Fred

Fred Kroeger wrote:What can the other agents do that NCPA can't and how much development effort is Nagios putting into NCPA ?

NRDS is a good example of something that NSClient++ does better than NCPA currently. Pushing plugins to remote machines to make deployment easier is one of the features that'll be included in a future NCPA release, but is currently native to NSClient++.

Other than that, there isn't much. The main benefit of NSClient++ is that it's been around for so dang long there's oodles of documentation out there for implementing pretty much every sort of language/interpreter/platform with it. The main benefit of NCPA is that you can come here and get a response to a problem/question relatively quickly.

Fred Kroeger wrote:I like the idea of NRDS, however this is only useful for Passive monitoring.

You could use the NRDS platform for the sole purpose of pushing plugins to your Linux boxes. It'd be a bit hacky because you'd have to disable the passive check bit, but you could use NRDS for something like that.

Fred Kroeger wrote:Is NCPA the preferred solution, or NSClient for Window/Linux. If NRPE has features that isn't matched by the other agents, then maybe a mix of NSClient & NRPE as I have now?

Unofficially, NCPA is the "preferred solution". The wizards and other Nagios XI end of things aren't 100% there yet since we need to support legacy setups, though. There are also some very very old systems that may have trouble supporting NCPA but can handle NRPE or NSClient++ just fine. Windows 2003 comes to mind since the version of psutil we build NCPA with will simply not work on Windows 2003.

Since we do it in-house and it's not beholden to the off-hours of an open source community, you can yell at us about a problem and it's much more likely to get fixed. You can also come here with specific architectural gripes or desired features and get the typical 24-hour turnaround on an actual human answering your concerns. I mention all of this because the NSClient++ community (just as one example) isn't quite as prolific as this one.

Fred Kroeger wrote:Also interested to know if any development is happening with NRDS to allow the deployment of these agents for use with Active monitoring?

A lot of the current NRDS related item in Nagios XI will probably be re-worked to use NCPA in the future, but don't quote me on that. As they exist currently, there's just cron jobs (or scheduled tasks) with a dirt simple script that hasn't been updated in a while. Updating NCPA to handle this a little more intelligently is the end goal.

Did @mcapra's post help you? Please let us know if you have any additional questions.

Just my 5c
=> NSClient has real-time eventlog monitoring. Not having to actively check for event with id x and source y is just awesome and saves a lot of Nagios server resources.
=> NSClient aggregates data and averages values like CPU, Memory, while NCPA returns the current value (@Nagios, please correct me if I'm wrong)
=> NSCLient is written in C++, which is more performant then Python
=> The NSClient community is much larger then NCPA.. (and I personally think even most Nagios admins are still using NSClient for Windows, correct me if I'm wrong)

Nagios just prefers NCPA, as for them it's important to not be dependent on one client which is not their own. If someday there would a giant exploit in NSClient and Mr Medin decides to stop supporting it, at least they have an alternative.

Grtz

Thanks for the assist, @WillemDH!