Re: Two ethernet interfaces puzzle
From: Jason Bourne (j_bourne_treadstone_at_hotmail.com)
Date: 08/10/04
- Next message: Kevin Nathan: "Re: windowmaker"
- Previous message: Mazuego: "vmware-any-any79 problem!"
- In reply to: steve: "Re: Two ethernet interfaces puzzle"
- Next in thread: Gonzalo: "Re: Two ethernet interfaces puzzle"
- Reply: Gonzalo: "Re: Two ethernet interfaces puzzle"
- Reply: steve: "Re: Two ethernet interfaces puzzle"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Mon, 09 Aug 2004 19:45:39 -0400
steve wrote:
[snip]
> what you can do is move your wireless router to a different subnet, (
> possibly 192.168.2.x) then route the packets from there.
>
Actually, when he first tries this it will more than likely behave the same
as in the OP. However, once it is configured in this manner he may then have
the ability to "route add" the subnet as a unique route. I know this will
work with other operating systems, but haven't verified it with Linux yet.
For example, Windows has a wrapper that encapsulates the interfaces and if
one goes down will automatically make attempts to move on to using the next
route in the routing table.
>
> I had actually taken this discussion to the networking newsgroup about a
> year ago, and asked why this was not possible, as i should be allowed as
> many network cards on the same subnet as my pci slots allow.
>
> Steve
I have not read the rfc's in a few years now myself, but even though they
are somewhat dated make for informative reading. The thing with the rfc's
is that they are guidelines which are open to interpretation by whoever is
developing for any particular OS. I am not quite sure that there is any
specific rule that says you cannot have more than one interface in the same
subnet.
However, the problem arises when you stick routing into the picture.
Subnetting in TCP/IP is the doorway to routing and switching. I have worked
in data communications since before TCP/IP and personal computers were even
in existence. I have worked in various Novell, Arcnet, Token ring with
Banyan Vines, Windows NT with Ethernet, etc, and my experience has always
been 2 nics in the same subnet equals problems of some form or fashion. I
have seen instances of infinite loops that even the Spanning Tree
algorithms could not break.
My first thoughts when trying to troubleshoot something of this nature is to
fire up a protocol analyzer, get a capture of the live traffic, then pick
it apart in an attempt to understand what the heck is going on. This means
packet after packet, frame by frame. Once you have a handle on what the
bits are doing you may then be able to figure out what needs to get changed
so they'll then behave in the manner you wish them to.
I am interested in his situation as it has the potential for learning about
how Linux behaves. I'm currently too involved with preliminary testing of
Oracle 10g on Linux for an upcoming migration from MS-SQL to actually
simulate this for my own curiosity (just not enough hours in the day). I'm
hoping he'll report back on what he finds because if he is successful we
all have the opportunity to learn something. :-)
-Jason
-- PGP Fingerprint: CA481E501CCDAEA6DC28940488833E0887E064D7
- Next message: Kevin Nathan: "Re: windowmaker"
- Previous message: Mazuego: "vmware-any-any79 problem!"
- In reply to: steve: "Re: Two ethernet interfaces puzzle"
- Next in thread: Gonzalo: "Re: Two ethernet interfaces puzzle"
- Reply: Gonzalo: "Re: Two ethernet interfaces puzzle"
- Reply: steve: "Re: Two ethernet interfaces puzzle"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
|