[Nagiosplug-devel] check_http: Socket timeout after 10 second s (but webserver is up! )

Christoph Stotz stotz at logo.de
Sat Feb 8 13:57:04 UTC 2003


Hello again,

thanks to Pauls great help, everything is working now. I used the patch
located at

http://www.myipv6.de/patches/vlan/3c59x.c-8021q.patch

In order to be able to apply it I removed some comment lines. Then the patch
worked
perfectly.


Thanks again & cheers!

Christoph


> -----Original Message-----
> From: Christoph Stotz 
> Sent: Samstag, 8. Februar 2003 16:56
> To: 'Subhendu Ghosh'; 'pb at bateauknowledge.nl'; 'karl at debisschop.net'
> Cc: 'nagiosplug-devel at lists.sourceforge.net'
> Subject: RE: [Nagiosplug-devel] check_http: Socket timeout 
> after 10 seconds (but webserver is up! )
> 
> 
> Hello,
> 
> first of all thank you very much for trying to help. 
> Unfortunately I found out right know, that the problem I am 
> experiencing is not directly caused by the check_http plugin. 
> The whole trouble is caused by the VLAN (802.1q) feature I am 
> using on the monitoring server. For those of you who do not 
> know what that is, you might want to have a look to the 
> following two links. There it's explained quite well:
> 
ftp://ftp.netlab.ohio-state.edu/pub/jain/courses/cis788-97/virtual_lans/inde
x.htm
http://www.candelatech.com/~greear/vlan.html

The short version of the explanation is: using VLAN "tagging" you are able
to have several "virtual" ethernet devices connected by a single physical
ethernet device (of course sharing the available bandwidth). In my case this
allows me to monitor different trunks of my network directly without having
to pass by a firewall or router. The alternative to that is to mount
multiple NIC's (in my case multiple quad-NICs).

If you go and have a look to the second link I just posted in this message,
you will read

"MTU problems exist for many ethernet drivers. Other than that, things seem
fairly stable! "

Well - that is exactly what I am experiencing. The whole thing is stable,
but I am experiencing a MTU-Problem. As soon as the webserver I want to
check with check_http is located on a VLAN trunk and returning more than
1476 bytes of data to the check_http, the packet gets either malformed or
fragmented. Due to an issue with the ethernet kernel driver, those returning
fragments are not put together properly and therefore check_http times out
with a Socket error.

So, this is an issue for the VLAN mailing list rather than the
nagiosplug-devel mailing list. If you experience this problem, the VLAN
mailing list located at

http://www.wanfear.com/mailman/listinfo/vlan

might be a good point to start. By the way I am using a 3c905C-TX/TX-M
[Tornado] rev 78 NIC with the 3c59x module version LK1.1.16 (19 July 2001).


If I solve the issue, I will let you know how.


Thanks again.


Kind Regards

Christoph Stotz




More information about the Nagiosplugin-devel mailing list