Re: [pcp] How can I make a ~stateless forwarding PCP proxy again?

[Markus Stenberg <[e-mail]>]

On 5.5.2014, at 23.38, Reinaldo Penno (repenno) <[e-mail]> wrote:
> "Getting back to my previous question, now that I think of it, actually, the IP address of the client is in the THIRD_PARTY option. Only thing that isn’t explicitly stored somewhere is the address client contacts the proxy on (it probably has 1-N addresses per M interfaces)."
> 
> Yes, but I believe in the current back-to-back proxy proposal (newest draft) the third party option would/should be completely removed by the proxy before forwarding a message.


Well, it actually behaves the same way what I said - it _does_ (in firewall case) insert THIRD_PARTY options on way to server and back. Only difference is that in my case, I don’t even want to think of being a NAT that proxies requests onward (I’m sure you need lots of state at that point and the content of the current draft seems focused mostly on that).

I wonder if someone’s implemented the back-to-back draft yet? It looks like a lot of work. The minimal state design I sketched is about 500 LoC of C[1] (with quite short line lengths too, debug messages, etc that I wrote today). Perhaps it would be a bit more if I actually bother to send ANNOUNCEs some day, but not much.

And that’s all you really need, given

a) you don’t want to cascade that proxy (you can’t have multiple THIRD_PARTY options)

b) you can live with ‘someone else’ having the control of server selection and PCP-forwarder being just dumb element following orders.

Perhaps I’ll write a draft on my thoughts on homenet + PCP angle for the next IETF (sigh, too full calendar, but one day of hacking this was amusing enough).

Cheers,

-Markus

[1] https://github.com/fingon/minimalist-pcproxy