Nagios Support Forum

I'm familiar with adding plugins to unix hosts using NRPE, but up until now we've never had to do this with our windows servers. Our windows servers are using NSClient++, and we'd like to add a plugin to check total % of CPU utilization for specific processes, but I have no idea how to do that on the host side.

For example, with NRPE I know that the plugin needs to be added to the directory where all our other plugins are and that a command definition need to be added to the nrpe.cfg file. I'm not sure what the equivalent for nsclient would be since I've never had to do that before. Just need a nudge in the right direction.

You should be able to use performance counters for this task:
https://www.nsclient.org/nscp/wiki/Chec ... eckCounter You will need to enable NSClientListener.dll in your nsclient config if you have not done so already.

Our windows servers are not using nrpe currently. I found a plugin on exchange that looks like it'll do what we want it to, but have no idea how to get nsclient to work with it.

http://exchange.nagios.org/directory/Pl ... ed/details

You could try using the check_nt perf counter check:

snapon_admin wrote:Our windows servers are not using nrpe currently.

You will need to enable nrpe to run external scripts, so maybe try to use the check_nt command above.

Oh ok, I see now. Alright so the check_nt perf counter check, I would like to see if that will do what we want rather than worry about implementing nrpe and using a different plugin. When I run the command you specified the output, no matter what process I try, is always '0'. What exactly is this checking? Also, one other question; how does this check handle multiple processes with the same name? Is there a way to see if any process of a specific name (ours has 9 running that are the same name) is using more than say 50% CPU?

Additional processes seem to be indicated with a processname, processname#1, processname#2, etc. You can see all of the named processes that are available by opening up perfmon, selecting Performance Monitor, and adding a new counter for Process, % User Time and finally selecting a process if you wish. As for the data abrist had you show, that is % user time, or the amount of cpu usage that process is using within user space, if it were to make a kernel space call, that would be under % Priviledged Time.

Ok, so for multiple processes that'd look something like this?
Also, is there anyway to do this and see the memory usage?

What is the output of that? If it looks correct chances are it should be. We can take a look for you. As far as memory usage goes you can find memory based counters in the windows performance monitor as well. Are you asking about seeing a process's memory usage?

Not sure if the output is correct or not. I get '0' no matter what number I use, but of the 9 processes only 2 of them are actually above 0% CPU right now (one is 2, the other is 4), so for all I know it could be correct. As for the memory question yes, I would like to see how much RAM a process is using, either in MB/GB or percentage.

Did you add all of the performance counters, that you are trying to monitor (Performance Monitor->Add) and restart the NSClient++ service?

If you get a "0" (zero), maybe your command is incorrect (you may need to do some escaping).

As for the memory, I believe you can use the "working set" counter, for example: