[svlug] Odd networking problem

[James Sparenberg]

On Mon, 2005-11-28 at 12:02 +0000, Eric N. Valor wrote:
> All:
> 
> I'm having a strange issue I've never seen before in many years of
> administering Linux systems.  
> 
> I've set up a simple SSH server using Debian Sarge and a custom kernel
> based on 2.6.14.2.  I can SSH into the box with no problem, but after a
> random (and short) amount of time the connection is dropped and
> reconnections are refused for another random (and short) amount of
> time.  
> 
> The SSH config is the stock Sarge config.  I've tried running telnet,
> and I get the same exact problem which leads me to believe it's not an
> SSH config issue.  I can still ping the system and nmap still reports
> open telnet and SSH connections during the "timeout".  There is no
> firewall of any kind, either hardware or onboard IPTables.
> 
> Nothing interesting appears in syslog, kern, auth, etc.
> 
> Any thoughts?

Just in the form of questions.  Might help, Might not.

1.  If you ssh in 2 times (before the drop) do they both drop
simultaneously or separately. If it's separate then we do need to look
at timeouts being set, If it's simultaneous then the likely hood is
driver related.

2.  Does it do this regardless of the box you come from (eliminate one
end) If any box is dropped in the same manor then it's the box in
question if only one box gets dropped.  It's the problem.

3.  If #1 is simultaneous then does it do this with  the Stock Sarge
kernel.  (Understanding that some hardware etc might not work, but for
testing all we really need is ssh and networking to work right) If it
does drop, then it could well be hardware (bad nic) If it doesn't drop
then I'd look at the nic drivers in 2.6.14 as being the cause of the
problem.   

James