[HECnet] Static IPs for everyone!

[Johnny Billquist]

On 2016-11-07 11:16, Sampsa Laine wrote:
>
>> On 7 Nov 2016, at 12:06, Johnny Billquist <bqt at softjar.se> wrote:
>>
>> Of course they get dumped. The bridge is designed that way. You cannot add a reandom member to the bridge and have it working.
>>
>> Both sides always have to agree in order for the bridge to work.
>>
>> If you move your end to a new address, then the other end also needs to update its config.
>>
>> So, if you move your end of the bridge to a new address, you need to tell me, so that I also update my config with which address you have.
>
> I know, I did that - in fact I set up two transit points but the problem is that for some reason the ORIGINAL UDP IP is sent along with the packet, not the IP address of the node doing the resending of the packet.
>
> It’s not your fault, I hope you didn’t read it as that - it’s definitely the correct behaviour on part of the bridge and probably actually probably the correct behaviour for socat..
>
>
> Basically, in summary, this was the setup
>
>
> NODE A:4711	-> 	TRANSIT:4711 	-> 	NODE B:4711
> NODE B:4711	->	TRANSIT:4712	->	NODE A:4711
>
> Node A is expecting packets from TRANSIT:4712, but sees packets from NODE B:4711
> Node B in turn expects them from TRANSIT:4711, but sees packets from NODE A:4711.
>
> So end result = dump the weird looking packets.

So, what you did was that you configured NODE A and NODE B to talk to 
TRANSIT. But how do you expect the packets to actually go by TRANSIT? 
Are you messing with the routing tables?

You need TRANSIT to be a proper relay, meaning you either play NAT, or 
else you run a bridge on that machine.

	Johnny

-- 
Johnny Billquist                  || "I'm on a bus
                                   ||  on a psychedelic trip
email: bqt at softjar.se             ||  Reading murder books
pdp is alive!                     ||  tryin' to stay hip" - B. Idol