Re: [rtcweb] Stephen Farrell's Discuss on draft-ietf-rtcweb-transports-14: (with DISCUSS and COMMENT)

[Harald Alvestrand <[e-mail]>]

Den 03. aug. 2016 16:51, skrev Stephen Farrell:
> Stephen Farrell has entered the following ballot position for
> draft-ietf-rtcweb-transports-14: Discuss
> 
> When responding, please keep the subject line intact and reply to all
> email addresses included in the To and CC lines. (Feel free to cut this
> introductory paragraph, however.)
> 
> 
> Please refer to https://www.ietf.org/iesg/statement/discuss-criteria.html
> for more information about IESG DISCUSS and COMMENT positions.
> 
> 
> The document, along with other ballot positions, can be found here:
> https://datatracker.ietf.org/doc/draft-ietf-rtcweb-transports/
> 
> 
> 
> ----------------------------------------------------------------------
> DISCUSS:
> ----------------------------------------------------------------------
> 
> 
> I'd like to briefly chat about one aspect of this...
> 
> Section 3.4: Allowing configuration of STUN/TURN servers
> from JS makes it easier for a calling server to track a
> user's call meta-data, if the JS supplied configuration is
> e.g. always preferred.

I'm a bit confused by this remark.

Yes, doing "pc = new RTCPeerConnection({iceservers: {urls:
"turn:trackserver.net"}) will cause packets to be sent to
"trackserver.net" whenever the PC attempts to open a connection.

What is the tracking attack that is facilitated by this ability, which
isn't more simply facilitated by some variant of <img
src="http://trackserver.net"> or its Javascript equivalent?

>  Shouldn't a browser prefer locally
> configured servers if those exist and can be used?  Or are
> there other things to be said about which STUN/TURN
> servers to use when there are multiple choices? (Apologies
> if this is handled by ICE already, I forget;-)

I have understood the lists of servers to be additive: That the browser,
if provided with two lists, would attempt all of them. It would then be
up to corporate policy to block access to all but the corporate,
locally-configured server, if that's their desire.

> 
> 
> ----------------------------------------------------------------------
> COMMENT:
> ----------------------------------------------------------------------
> 
> 
> - 3.1: Would "TURN/TLS" not be better than "TURN/SSL"?

Yes. I have changed this in
https://github.com/rtcweb-wg/rtcweb-transport/pull/35.

> 
> - 3.2: Would s/or its TURN server/or the peer's TURN
> server/ be better?

Yes, it would. I'll change this.

Added to https://github.com/rtcweb-wg/rtcweb-transport/pull/31/files
since that PR was changing the next line of the draft.