Hello!
  I have a question about the bridge patch
bridge-nf-0.0.7-against-2.4.19.diff.

I am using RH 8.0,but with the linux-2.4.20 kernel.  I
applied the patch and yippie skip, all was good with
the world.  Using brctl I was able to bring up the
machine in a bridge fashion.  I have set up iptables
to basically do no denial rules, but instead am
QUEUEing packets to a userspace queue.  This queue is
designed to mangle packets between 2 machines with the
queue installed.  It mangles the TCP/UDP header and
payload (so the ether and ip is ok) in a way that the
other side know how to de-mangle.  In testing, I
basically have a data pumper and a listener/absorber
sending a few meg on the 100BaseT connection.

The configuration where I have 2 normal (non-bridge)
machines connected works great, hostA mangling packets
sent to hostB (and hostB demangles), but I have a
serious problem when bridging with the above patch is
used (although I have no firewall denials, i need to
set up QUEUE rules on specific interfaces).  My
thruput is destroyed, in what appears to be a constant
re-transmission of a packet after about 20 or so
packets being delivered.  

I guess my first question: is there anything in the
patch that would be put off by the fact that I mangle
the TCP/UDP portion of the header?
If not, are there thought on where I could start
examining this problem?  

-Scott





__________________________________________________
Do you Yahoo!?
Yahoo! Platinum - Watch CBS' NCAA March Madness, live on your desktop!
http://platinum.yahoo.com
_______________________________________________
Bridge mailing list
[EMAIL PROTECTED]
http://www.math.leidenuniv.nl/mailman/listinfo/bridge

Reply via email to