Re: Dealing with isolation state mismatches

On 15 May 2014, at 14:48, Martin Thomson <martin.thomson@gmail.com> wrote:

> On 15 May 2014 01:17, tim panton <thp@westhawk.co.uk> wrote:
>> I think you are assuming that the only signalling between A and B is the SDP.
> 
> No, I'm simply assuming that the stupid-simple application might want
> to avoid this sort of error.
> 
>> There is a question as to how we indicate DTLS setup failures, but isn’t that covered elsewhere?
> 
> Honestly, I hadn't considered it, I soft of assumed that someone had
> already dealt with that.  You are right though, it's a problem.
> Fingerprint mismatches, this stuff.
> 
> Chairs, issue over here.  Let's talk about this next week.
> 

And here's a weird example I'm debugging:

I have 2 possible paths: direct and relay - both pass ICE tests - but the direct path 
has an MTU bug where the DTLS server hello fragments don't get re-assembled correctly,
but the relay path seems to work fine. However the relay route isn't used because the test is ICE based.
What's more I don't seem to get any error messages I can see, it just doesn't pass audio.

It's likely a DTLS or network bug that is causing the issue, but it would be handy to get an
indication of DTLS negotiation failure and even better if the stack tried alternate paths that are less preferred at the ICE level.

T.

Received on Thursday, 15 May 2014 14:05:19 UTC