> kernel : REJECTED PACKET: IN=br0 PHYSIN=eth0 OUT=br0 PHYSOUT=eth1
> SRC=10.1.2.10 DST=10.1.2.11 LEN=57 TOS=0x10 PERC=0x40 TTL=127 ID=44775 DF
> PROTO=TCP SPT=1790 DPT=25 WINDOW=64172 RES=0x00 ACK PSH UGRP=0
>
> Why does it say that the packet came in on eth0, that interface in
> connected to my ISP (in the little diagram I drew earlier)?

I think your 10.1.2.10 box is sending messages to 10.1.2.11 using the default 
gateway. The reason is shown below.

> I also started tcpdump on eth0 and eth1 on the bridge computer, and on
> computer 2. This is the arp messages I got from tcpdump.
>
> Bridge computer eth0:
> arp who-has 10.1.2.11 tell 10.1.2.10
> arp reply 10.1.2.11 is-at 0:0:c:7:ac:21
>
> Bridge computer eth1:
> arp who-has 10.1.2.11 tell 10.1.2.10
> arp reply 10.1.2.11 is-at 0:0:c:7:ac:21
>
> Computer 2:
> arp who-has 10.1.2.11 tell 10.1.2.10
> arp reply 10.1.2.11 is-at 0:50:da:92:2c:84
>
>
> The hardware address 0:0:c:7:ac:21 is the default gw, the correct hw
> address for computer 2 is 0:50:da:92:2c:84
>
> Why does the bridge report that 10.1.2.11 is on the hw address of the
> default gw?

The default gateway knows how to get to 10.1.2.11 and tells this with an arp 
reply. IMHO, I don't think that is correct behaviour from that default 
gateway (it's a local address).
I think you're best off stopping those ARP requests from going on the eth0 
wire. You can use ebtables for this :) Probably the only ARP requests that 
should go onto the eth0 wire are those for the default gateway.

-- 
cheers,
Bart

_______________________________________________
Bridge mailing list
[EMAIL PROTECTED]
http://www.math.leidenuniv.nl/mailman/listinfo/bridge

Reply via email to