Re: [lwip-devel] Question regarding MDNS defending name.

[Jens Nielsen]

Hi

Nice to hear more work is being pitched into mDNS :)  Did you also 
intend to post your work back to lwip?

As far as I know...
> However, after some time, it receives a packet whose contents indicates that
> it's name/service are in conflict.
This would be another node sending a response or announce with our 
name/service but different rdata, right?

>  From the RFC 6762, Section 9, the immediate action would be to return to
> PROBING. However, I am confused, bcz at this stage, since the node have
> already confirmed its records being unique, it should instead send a
> RESPONSE rather than restart the probing as suggested
But since the conflict packet is a response/announce you can't send a 
response to that? Just because probing is completed doesn't mean it's 
100% guaranteed to be unique, maybe a switch was down so both nodes 
passed probing on each part of the network? Going back to probing seems 
like an appropriate action to me, I recall the other mDNS implementation 
I was testing against did that when my probing implementation misbehaved 
and I announced a name that was already in use...

> , and it's in
> contradiction with Section 8.1
> "*In order to allow services to announce their presence without
>     unreasonable delay, the time window for probing is intentionally set
>     quite short.  As a result of this, from the time the first probe
>     packet is sent, another device on the network using that name has
>     just 750 ms to respond to defend its name.*"
I'm not sure I understand the contradiction? That part is about not 
delaying a response to a probe (as it should when responding to normal 
queries, as described in Section 6)

Btw, just to make sure we aren't doing the same thing, you're aware 
there are a set of patches for probing on savannah here? 
https://savannah.nongnu.org/patch/?9555

Best regards
Jens