Re: OT: Incoming/Outgoing route of IP subnet
Alexander Steinert, 2003-Apr-14 13:36 +0200:
> > > OK, let's assume we have
> > > an IP subnet N,
> > > a host H in N,
> > > a host G in N and
> > > a host A outside of N.
> > > Packets from H to A go via G, since H is configured as gateway on H.
> > > How save is it to assume that all packets from A to H (the other way
> > > round) go via G?
> > Hmm... Not really safe at all since the specs say that the routes do
> > not need to be symmetrical. Since they are not required to be you can
> > almost certainly assume that at times they will not be symmetrical.
> > Therefore an assumption like that will certainly break something at
> > some time in the future.
> > Could you give us a hint? What are you planning? Usually I find that
> > specific questions like this are often the result of other more
> > general questions and the latter are the more interesting ones.
> I'm implementing an TCP/IP/ETH stack on a small device and wondered if I
> can simplify the ARP part and spare RAM by the assumption sketched
> above. One could remember the MAC src address of an incoming frame and
> use it as the MAC dest address for answering frames. No ARP requests
> would be necessary, only ARP answers.
So, your described network looks like:
- I assume G is H's default gateway
- I assume A's network has more than host A
1. If there is no other path from A to H, then all packets from A to
H will pass through G.
2. For your ARP simplification, you could give G a static ARP entry
for host A in G as well as any other static routes you add to G.
3. You could allow G to use a broadcast MAC everytime it doesn't have
a MAC for an IP, then learn on return traffic.
Jeff Coppock Systems Engineer
Diggin' Debian Admin and User