On Wed, 21 Jul 2004 23:17:47 -0700, Shahar wrote:
> Juha Laiho <(E-Mail Removed)> wrote in message news:<cdmb95$ggd$(E-Mail Removed)-int>...
>> (E-Mail Removed) (Shahar) said:
>> >is it possible?
>> >
>> >I searched and tried ebtables w/o any luck so far.
>> >
>> >I have a bridge over a 3 port junction.
>> >I want to redirect a packet incoming from eth0 to eth2
>> >without changing it's mac address.
>>
>> MAC? Source or destination? Both? Anyway, doesn't sound likely -- the
>> MAC address is specific to a network segment, and thus should be set
>> to the new value each time a packet traverses from one segment to another.
>>
>> Could you describe in more detail (and in a larger context) what it is
>> that you're doing?
>
> dnat.
>
> i have a linux box with 3 nics.
>
> nic1 connected to a router
> nic2 connected to box2 (10.0.0.2)
> nic3 connected to box3 (10.0.0.50)
>
> I want to push all(well...) packet incoming from nic1
> to nic2 with the original mac address(i.e. no DNAT) so they will arrive
> to box2's nic card.
>
> ebtables dnat the packet...
> and writing/manipulating kernel is not an option at the moment.
Check out some of the posts to/from "Captain Beefheart". He seems to be
doing something simliar to you, and I believe has arrived at exactly this
point: doing routing, without doing (d)nat. My understanding (but I don't
currently use it) is that if you don't run iptables, there is no (d)nat?
The routing is just done "straight up" meaning your local addresses go out
the other side without being munged to look like they're from one IP.
BTW, I'm not clear what you're trying to do? Diagnostics? If you're just
trying to diagnose something why don't you use tcpdump or something, and
put your interface into promiscuous mode. It almost looks like you're
trying to setup to do "man in the middle address spoofing" or a "hijack"?
The reason the other poster asked what you were trying to do, was to try
to determine what you were really trying to do and why. Sometimes people
get fixated on mechanisms and get embroiled in technical complexities when
what they really want is easily obtainable by some other simpler means.
You still haven't explained what your application is... or why...
--
Juhan Leemet
Logicognosis, Inc.