NCPA and processes check.

[tonyk]

Hi,

I have been banging my head into the wall with the following problem:

My goal is to make a processes service check with ncpa that excludes one or some processes.

I can make it work on the command line as follows:

nagios@nagios4:~$ /usr/lib64/nagios/plugins/check_ncpa.py -q 'name=^((?!kworker).)*$,match=regex' -T 15 -H server -t token -M processes
OK: Process count for processes named ^((?!kworker).)*$ was 148 | 'process_count'=148;;; 'cpu'=22.64%;;; 'memory'=81.24%;;; 'memory_vms'=45.0GB;;; 'memory_rss'=13.51GB;;;
Processes Matched
PID: Name: Username: Exe: Memory: CPU
-----------------------------------
1: systemd: root: 0.03 % (VMS 0.20 GB, RSS 0.00 GB): 0.00 %
2: kthreadd: root: 0.0 % (VMS 0.00 GB, RSS 0.00 GB): 0.00 %
.
.


But I can't make it work in nagios because of the "!" which is the delimiter in nagios.

Any ideas?

[benjaminsmith]

Hello Tony,

Please post the check command from your nagios config file that is not working for us to review. Thanks.

[tonyk]

Well I have tried different things, but this is the basic one:

define service {
check_period 24X7
active_checks_enabled 1
notification_interval 0
notification_period 24x7
check_command check_ncpa_processes!xxxxxxx!processes -q 'name=^((?!kworker).)*$,match=regex'
host_name host
check_interval 7
process_perf_data 1
service_description Check_Processes_test
max_check_attempts 3
notification_options w,c,r,u
}

define command {
command_line $USER1$/check_ncpa.py -T $ARG3$ -H $HOSTADDRESS$ -t $ARG1$ -M $ARG2$
command_name check_ncpa_processes
}

And the output for this is:

No data received on stdout
Plugin output (standard error)
/bin/sh: -c: line 0: syntax error near unexpected token `)'
/bin/sh: -c: line 0: `/usr/lib64/nagios/plugins/check_ncpa.py -T kworker).)*$,match=regex' -H host -t xxxxxxxx -M processes -q 'name=^((?'

Which I totally understand, because the delimiter is "!". But I don't know how to work around this.

Actually I just found a way. I have to add the -q 'name=^((?!kworker).)*$,match=regex' directly to the command. That should work...

define command {
command_line $USER1$/check_ncpa.py -T $ARG3$ -H $HOSTADDRESS$ -t $ARG1$ -M $ARG2$ -q 'name=^((?!kworker).)*$,match=regex'
command_name check_ncpa_processes_exclude
}

But it doesn't. I use adagios and if i try to test the command with the check plugin command it works ok:

OK: Process count for processes named ^((?!kworker).)*$ was 156 | 'process_count'=156;;; 'cpu'=24.0%;;; 'memory'=77.18%;;; 'memory_vms'=66.07GB;;; 'memory_rss'=12.82GB;;;
Processes Matched
PID: Name: Username: Exe: Memory: CPU
-----------------------------------
1: systemd: root: 0.03 % (VMS 0.20 GB, RSS 0.00 GB): 0.00 %
2: kthreadd: root: 0.0 % (VMS 0.00 GB, RSS 0.00 GB): 0.00 %
6: ksoftirqd/0: root: 0.0 % (VMS 0.00 GB, RSS 0.00 GB): 0.00 %
7: migration/0: root: 0.0 % (VMS 0.00 GB, RSS 0.00 GB): 0.00 %
8: rcu_bh: root: 0.0 % (VMS 0.00 GB, RSS 0.00 GB): 0.00 %
9: rcu_sched: root: 0.0 % (VMS 0.00 GB, RSS 0.00 GB): 0.01 %
...

Adn if I run the command shown by adagios - Status - Service - Information - Check Command:

/usr/lib64/nagios/plugins/check_ncpa.py -T 15 -H host -t xxxxxx -M processes -q 'name=^((?!kworker).)*$,match=regex'

That runs ok and shows the correct output as above.

But when it is running in nagios itself, the regexp does not seem to work as the output is:

OK: Process count for processes named ^((?!kworker).)*$ was 0

Darn...Of course this would be easily solved if there was a match=negate possibility in check_ncpa.py itself...

[mbellerue]

/usr/lib64/nagios/plugins/check_ncpa.py -T 15 -H host -t xxxxxx -M processes -q 'name=^((?!kworker).)*$,match=regex'

/usr/local/nagios/plugins/check_ncpa.py -H 192.168.145.90 -t 'ASecretToken' -M 'processes' -q 'name=^((?!kworker).)*$,match=regex'

So this is strange, I've got this working, though granted it's Nagios XI.

Of course this would be easily solved if there was a match=negate possibility in check_ncpa.py itself...

I can throw that in as a feature request. However, for the moment, would the negate plugin be useful here? Here's the documentation, it's Nagios XI branded, but I believe negate ships with Core.
https://assets.nagios.com/downloads/nag ... ios-XI.pdf

[tonyk]

I don't see how the negate plugin woiuld help here. I do use it for some plugins, but it is only useful to change the state output to something else.

I did however manage to make a workaround for this problem.

I checked the ncpa_listener.log for the problem and noticed, that if I run the check through nagios I get an extra "%24" in the end of the GET http line. If i run it from the command line it doesn't and the check works.

DId not find out how to fix that directly so I tried to do this with a wrapper script:

Code: Select all

#!/bin/sh
/usr/lib64/nagios/plugins/check_ncpa.py -q 'name=^((?!kworke).)*$,match=regex' $* 
exit

And changing the command to use this instead of check_ncpa.py. And finally it is working!!!

[scottwilkerson]

And finally it is working!!!

Great!

Glad to see you have it working!

Locking thread