NSClient++ loses communication with Nagios

[davidlac]

Dear,
NSclient works fine for my check services but for no reasons, nsclient stops to communicate with Nagios.
On Nagios console i've got the following message: State : Unknown : CHECK_NRPE: Error receiving data from daemon.
On Nsclient Log: error:modules\NRPEListener\NRPEListener.cpp:289: Unauthorize access from: xxx.xxx.xxx.xxx

My Nagios IP xxx.xxx.xxx.xxx is declared in nsc.ini.
allowed_hosts=xxx.xxx.xxx.xxx
My version of nsclient is V0.3.8.75

Restart Nsclient service on W2k3, W2K, W2k8 or XP resolves this PB.
Any idea why nsclient needs to be restarted sometimes?

[mickem]

Could you provide debug logs from NSClient++?

[davidlac]

Hello,
My debug logs from nsclient++:
debugbefore.txt = debug when nsclient failed to communicate. Error error:modules\NRPEListener\NRPEListener.cpp:289: Unauthorize access from: xxx.xxx.xxx.xxx
debugafter.txt = debug after nsclient service restarted. Communication OK

[mickem]

Well, what I need is the debug log *when* it stops working.
Easiest way is to enable FileLogger module and set debug=true...

Michael Medin

[davidlac]

Ok I enable debug mode on several clients.
I'll send you results soon.

[davidlac]

Here a log of nsclient with debug mode enable.
The host agent communication failed since nsclient started at 2012-02-06 22:41:40
After investigation, the server restarted, so nsclient started too.
Does it happen for other users after system restart?

[mickem]

This was very usefull...
I have had reports (sporadic) from people who claim just this.
Out of curiously xxx.xxx.xxx.xxx is that you editing out the ip or is that the actual message?

But I think this could be due to service dependencies actually it could be that we need to start nsclient++ "after" network is online I shall investigate and hopefully you can help me test this...

Michael Medin

[mickem]

Also are you using a hostname or IP address for in the allowed hosts file?

And could you also try setting:

[NRPE]
; ...
cache_allowed_hosts=0

This will prevent the client to not caching the allowed hosts list meaning it might be reevaluated when the network stack is up...

Michael Medin

[davidlac]

I edit IP by xxx.xxx.xxx.xxx.
I detected this bug several months ago and I created dependence on DNS with nsclient services because my allowed_hosts in nsc.ini uses dns name. Without this configuration all of nsclient agent bugs after host restart. With this dependence only a few of them have a problem. (50 of 2050)
I'll try cache_allowed_hosts=0

[davidlac]

Hello,
I configured cache_allowed_hosts=0 on 4 clients with this problem. Next reboot is planned on 02/23/2012.
I'll check and tell you if it works on march.
Thanks for advices

[mickem]

Added dependency to Tcpip on the service install which in conjunction with the cache_allowed_hosts=0 flag hopefully solves this issue. If not please reopen...

Michael Medin