CPULOAD reporting issue

[gmackey]

Hello,

I'm running the latest version of NSClient++ (0.5.0.62) and have run into an odd issue with check_nt CPULOAD. Nagios XI specifies the timeframe in minutes, with a default of 5. However, NSClient++ 0.5.0.62 interprets that timeframe as being in seconds and reports as such. I need the 5-minute average to be a 5-minute average, not a 5-second average. Same for my 60-minute CPU monitors that are currently reporting 60-second averages due to this issue. Have you guys seen this problem before? I haven't had any other issues that I've noticed with this version of NSClient++ and I have it running on about 60 servers so far. These are all Windows servers, mostly 2012 R2 and some 2008 R2. I haven't tried older NSClient++ versions and I'd like to stick with this latest stable version of 0.5.x if possible.

Here's the command I'm using:

check_nt -v CPULOAD -l 5,80,90

That command returns the 5-second average, not 5-minute average, on these systems. I've verified this using a load generator. I can't specify "5m" since Nagios XI says it's invalid. Any thoughts? I can change the value to 300 but then the email notifications will say "(300 minute average)" and that would be a little confusing. Thanks!

-Greg

[tgriep]

If the NSCLient Agent is returning the load for 5 seconds instead of minutes using the check_nt, then the NSClient Agent has a bug and the author will have to fix the issue.
You could setup the NSClient to use NRPE and run the check that way.
For some examples, you can view this link.
https://docs.nsclient.org/reference/win ... #check_cpu

[gmackey]

I've confirmed that NSClient++ has been this way for years and will not be changing. I've made the decision to convert all check_nt services to check_nrpe since I wish to use a supported/maintained version of NSClient++ as well as a supported/maintained protocol.

https://docs.nsclient.org/faq/#53-i-use-check_nt-and

Check_nt is NOT a good protocol and is considered abandon-ware as no one updates the check_nt command any longer. NSClient++ supports it only for legacy reasons. There is generally no reason to use check_nt as check_nrpe and check_nscp can achieve the same thing.

Please feel free to lock this thread. Thanks.