Problem with socket timeouts. (Dru)

[Stanley Hopcroft]

Dear Sir,

I am writing to thank you for your letter and say

On Sun, Nov 14, 2004 at 08:09:48PM -0800, nagios-users-request at lists.sourceforge.net wrote:
> 
> Message: 4
> Date: Mon, 15 Nov 2004 16:27:03 +1300
> From: Dru <andru at treshna.com>
> Subject: Re: [Nagios-users] Problem with socket timeouts.
> 
> 
> > Dru wrote:
> >
> >> I changed it to 50 seconds and it still timed out.
> >
> >
> > This sounds like a firewall issue. You realise you need to open up all 
> > the firewalls for the checks to run through, right?
> 
> There is a firewall but its outside the network. I am just trying to 
> mointor servers running on the local machine. It could
> be passing the wrong $HOST in possible.



> Is there a way to get the $HOST 
> details?

>  It'll be great if the website outputted the full
> debugging output from running the plugin.

Unfortunately, no,

The plugins (such as this) are intended to be light weight: intensive 
debugging code is minimised to increase the plugin performance.

You can do this with a plugin wrapper that logs stuff useful to 
debugging in a file. Doing so has been described in these lists before 
(eg

http://article.gmane.org/gmane.network.nagios.user/23653/match=plugin+wrapper

)

> Like execting plugin 
> $fullpath/$pluginname -H $host etc so i know exactly why its screwing 
> up.   Does anyone thats work on nagios code know if this is easy to 
> enable in regard to logging?
>

You can add any amount of debugging hooks in the wrapper eg in a sHell 
wrapper

echo $@ (to dump all the arguments)
echo $0 (make sure program name is correct)
nohup tcpdump -w somefile.pcap host <-H option value> port 25 &
etc etc 

 
> I saw other people having issues on google but they said they recompiled 
> from source and problem went away. I just dont get it.
>

Most people I suspect are not. check_smtp has been pretty solid for 
years.

Rejoice, you have a repeatable problem and the list will help you sort 
it out. It's not as if you need to wait for <famous name> service or 
anything.

Another thing you can try is run tcpdump while Nag is running, filtering 
for the destination and see if there is any traffic going to where you 
expect it to be going, and if so, what happened to it.

Has the Nagios server an IP address that is resolvable by the target ? 
If not, this can play havoc with all sorts of tcp servers (mail, ssh etc 
that look up the client address). -- Stupid question, since runs from 
the CLI Ok.

Does the Nag host have multiple interfaces and your check connections 
come from an interface you don't expect ? tcpdump will tell you. --  
Stupid again since CLI is Ok.

If dig -x <nag server address> (or nslookup) doesn't return a name on 
the target, there is no way check_smtp or any socket based check (except 
maybe check_htp) will work.

Lstly, if this gets to be an intractable problem, think about having a 
test Nag configured with debug (--enable-DEBUG[1-4] to configure) with a 
stripped config running in foreground mode. This is an easy way to  deal 
with problems like this.

HTH,

Yours sincerely.
 

-- 
Stanley Hopcroft

Network specialist, IT Infrastructure
IP Australia
Ph: (02) 6283 3189  Fax: (02) 6281 1353
PO Box 200 Woden  ACT 2606
http://www.ipaustralia.gov.au
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: disclaimer.txt
URL: <https://www.monitoring-lists.org/archive/users/attachments/20041115/9456ee70/attachment.txt>