I think this would work fine with the iptables patch (for 2.4), if you would
do the counting in the mangle/PREROUTING chain.. ipchains is fundamentally
inflexible, so I don't have much hope that it will be possible to get that
working.


On Fri, Dec 07, 2001 at 04:03:31AM +0100, blinky wrote:

> beware: this mail is maybe written in a confusing way and because of my not that 
>good english bad to understand! ;-)
> 
> 
> hi there!
> 
> i just discussed with a fried if it is possible to account ip-based traffic with a 
>bridge.
> i would be happy if you can give me a hint how to do the thingy i'll try to describe 
>you now:
> 
> image an uplink port #1. this port handles all traffic from and to the internet.
> on port #2 (configured as monitoring port for #1) i want to attach a linux-box that 
>should act as traffic accounting server.
> on the other ports there are some servers.
> 
> now i want to figure out how many traffic passed from the server at (e.g.) port #7 
>to the internet and vice versa. local traffic (like port #7 -> port #5) shouldn't get 
>counted.
> 
> i thought about having the linux-box on port #2 acting as a bridge (to get the 
>network-card into promiscous mode and have the kernel count the bytes ) and and use 
>ipchains [with the ipchains-bridging-patch] to count the amount of bytes with rules 
>like
> 
>    10.0.0.10, port *  ->  any
>    any -> 10.0.0.10, port *
> 
> 
> the problem is that the server with ip# 10.0.0.10 is physically on port #7 and not 
>on the other side of the bridge and so the kernel doesn't bridge the packets. is 
>there a way to have it bridging the packets into nirvana (/dev/null or a network-card 
>without any cable connected to it)? means: is there a way to  tell the kernel that 
>every MAC-address exists on the "no_cable_connected_to_it" network-card? this way the 
>kernel would do forwarding for every packet and the bytes would appear in ipchains. 
>it doesn't matter that the forwarded (=bridged) packets doesn't arrive anywhere since 
>they are duplicates (via the mirror-port) anyway. the only import thingy is that it 
>is possible to count the traffic.
> 
> 
> thanks for reading and any help! :-)
> 
> -
> daniel
> 
_______________________________________________
Bridge mailing list
[EMAIL PROTECTED]
http://www.math.leidenuniv.nl/mailman/listinfo/bridge

Reply via email to