[PLUG] cisco 678 DSL latency issues

[Rogan Creswick]

On Tue, Aug 05, 2003 at 08:40:57AM -0700, Russell Evans wrote:
> Have you compared the Windows setup to the Linux setup? 
> 
> ifconfig ethX  vs ipconfig /all

The ip info (ip, gateway, netmask, dns) is the same accross OS's. I'm
following up on another lead from Unicom or I'd reboot to double check
right now.

> netstat -rn	vs I can't remember if windows netstat will give you the route
> table or the flags needed if it does.

this looks to be the same output as route -n, and looks normal (to me):

Kernel IP routing table
Destination       Gateway         Genmask         ...
216.104.67.224    0.0.0.0         255.255.255.224  U     0    0    0   eth0
0.0.0.0           216.104.67.225  0.0.0.0          UG    0    0    0   eth0

The numbers correspond to my networki(216.104.67.224),
mask(255.255.255.224) and gateway(216.104.67.225) as they should.

> What do your IP tables look like? 
> 
> iptables -L

The current box I'm on (my laptop) dosen't have iptables. (never
thought I needed it on the laptop). So I don't think this is an issue.

> Using Ethereal to look at the traffic might give you an idea of what the issue
> is.  

Not currently installed, and there is very little traffic (netstat
-vatu shows no established connections when sitting idle, as expected
for a linux laptop). 

> 
> This is a long shot, but
> 
> cat /proc/sys/net/ipv4/ip_local_port_range
> 
> If the fisrt number is 1024 do an 

first number is big, (32768).

The current theory is that my nics won't negotiate 10/100 correctly..

-Rogan

> 
> echo 1025-4999 > /proc/sys/net/ipv4/ip_local_port_range
> 
> I had a 675, route mode not bridging, doing nat and it would map port 1024 on
> the inside ethernet  to port zero on the DSL interface. It would be really odd
> if this was an issue with your set up, but its something to try if the normal
> issues aren't what is dinking your set up.
> 
> Thank you
> Russell
> 
> 
> On Tue, 5 Aug 2003 07:57:30 -0700, Rogan Creswick said:
> 
> > On Mon, Aug 04, 2003 at 11:34:28PM -0700, Russell Evans wrote:
> >  > Are you pinging with name or address? If name then try IP address. 
> >  
> >  It makes no difference.  Pinging my Qwest gateway by ip in Linux
> >  results in 2000ms pings, while pinging the same ip in Windows takes
> >  only ~60ms. I've also tried two different dns servers on both sides.
> >  
> >  I get the same results with engr.orst.edu (by name and ip).
> >  
> >  -Rogan
> >  
> >  > 
> >  > Thank you
> >  > Russell
> >  > 
> >  > 
> >  > On Mon, 4 Aug 2003 21:58:51 -0700, Rogan Creswick said:
> >  > 
> >  > >  I'm running bridged with a cisco 678, 640/256, static ip.  If the
> >  > >  cisco is plugged into a windows box everything works great, ping times
> >  > >  are arround 90ms.  However, booting into Linux causes ping times to
> >  > >  jump to 2000ms, connectivity and bandwith are fine, but latency is
> >  > >  horrid (I can download a kernel at 40k/s or better, but it takes
> >  > >  2000ms for the xfer to start).
> >  > 
> >  > 
> >  > 
> >  > _______________________________________________
> >  > PLUG mailing list
> >  > PLUG at lists.pdxlinux.org
> >  > http://lists.pdxlinux.org/mailman/listinfo/plug
> >  
> >  _______________________________________________
> >  PLUG mailing list
> >  PLUG at lists.pdxlinux.org
> >  http://lists.pdxlinux.org/mailman/listinfo/plug
> >  
> >  
> 
> 
> _______________________________________________
> PLUG mailing list
> PLUG at lists.pdxlinux.org
> http://lists.pdxlinux.org/mailman/listinfo/plug