checkcpu ERROR: Could not get data for 20m NSCA

[comrade]

using nsclient.ini for report cpu usage for nagios server
modules
CheckDisk = 1
CheckSystem = 1
NSCAClient = 1
Scheduler = 1

settings/scheduler/schedules
my_cpu_check =checkcpu warn=80 crit=90 time=20m time=5m time=30s ShowAll=long

nagios display
ERROR: Could not get data for 20m perhaps we don't collect data this far back?
when using on windows XP pro SP3.(the same on 3 tested computers, 0.3 with old config successfully display cpu usage on the same computers)

same nsclient.ini on windows 7 SP1 work corectly.

[mickem]

There was an update to 0.4.0 build 174? which fixed an issue regarding the default buffer size which build do you have of NSClient++?

Michael Medin

[comrade]

i tested 0.4.0 build 174 and 0.4.1.1 and result is the same
on windows XP pro sp3 32bit
ERROR: Could not get data for 20m perhaps we don't collect data this far back?

[comrade]

tested with version 0.4.1.2 but checkcpu show the same message on windows XP PRO 32bit then reporting to NSCA server
ERROR: Could not get data for 20m perhaps we don't collect data this far back?

[mickem]

Maybe I should have been more clear about this before but 0.4.1 is not "the new build of 0.4.0".
The idea is that 0.4.0 and 0.4.1 will live in parallell untill 0.4.1 is "ready".
Thus allowing for more frequent updates and smaller "jump".

So 0.4.0 is the "new version" (still).
0.4.1 is an experimental branch which has experimental features so dont install unless you want to give feedback and/or try out some of the new features.

Currently 0.4.1 has:

• real encryption for NRPE, NSCA, check_nt
• an improved version of the the new NSCP protocol

And in the next few weeks:

• NRDP support as well as check_mk.

To return back to you issue :)
I shall look into it this weekend and hopefully have a solution after that :)

Michael Medin

[mickem]

please validate if you have any other errors in the log:
Out of the box config with enable checksystem gives me:

...
l ce\simple_client.hpp:80   OK:OK: 20m: average load 0%, 5m: average load 0%, 30s: average load 1%
l ce\simple_client.hpp:82    Performance data: '20m'=0%;80;90 '5m'=0%;80;90 '30s'=1%;80;90

Michael Medin

[comrade]

I install 0.4.0.175 on windows XP PRO 32bit but the same error for CPU
errors in log:
2012-06-13 08:41:22: l:..\..\..\trunk\service\NSClient++.cpp:385: NSClient++ 0,4,0,175 2012-06-12 w32 booting...
2012-06-13 08:41:23: e:..\..\..\..\trunk\modules\CheckSystem\PDHCollector.cpp:115: Failed to open performance counters: \4\30: PdhAddCounter? failed: -1073738823: The specified counter could not be found.

2012-06-13 08:41:23: e:..\..\..\..\trunk\modules\CheckSystem\PDHCollector.cpp:152: No performance counters were found we will not wait for the end instead...
2012-06-13 08:42:04: e:..\..\..\..\trunk\modules\CheckSystem\PDHCollector.cpp:237: Failed to get CPU value: <UN ATTACHED>: No value has been collected yet

[mickem]

Aha, then there are two reasons for this:

1. Counters not working (see FAQ)
2. Strange language (set counters manually)

Michael Medin

[comrade]

Counters from Xp system

[comrade]

windows 7 counters

[comrade]

counter working, and with 0.3.9 checkcpu also work o winxp.
Default windows language is Lithuanian, no language packs.
I added results from command
nscp client --module CheckSystem --exec listpdh
the same config file work on win7 x64.
counter exist on both computer
\Processor(_Total)\% Processor Time
Wich shoul look command for manualy check cpu load?

[mickem]

I shall update so the error message contains the counter whit the problem

[mickem]

this has been resolved in 0.4.1 (recent builds)

Michael Medin