Re: [Ice] Tsvart last call review of draft-ietf-ice-pac-03

[Justin Uberti <[e-mail]>]

On Mon, Jan 20, 2020 at 10:42 PM Yoshifumi Nishida <[e-mail]>
wrote:

> Hi Justin,
>
> Thanks for the response.
> I put my comments in lines.
>
> On Sun, Jan 19, 2020 at 7:06 PM Justin Uberti <[e-mail]> wrote:
>
>>
>>
>> On Sun, Jan 19, 2020 at 12:56 AM Yoshifumi Nishida via Datatracker <
>> [e-mail]> wrote:
>>
>>> Reviewer: Yoshifumi Nishida
>>> Review result: Almost Ready
>>>
>>> This document has been reviewed as part of the transport area review
>>> team's
>>> ongoing effort to review key IETF documents. These comments were written
>>> primarily for the transport area directors, but are copied to the
>>> document's
>>> authors and WG to allow them to address any issues raised and also to
>>> the IETF
>>> discussion list for information.
>>>
>>> When done at the time of IETF Last Call, the authors should consider this
>>> review as part of the last-call comments they receive. Please always CC
>>> [e-mail] if you reply to or forward this review.
>>>
>>> Summary:
>>>    This document is straightforward and almost ready for publication,
>>>    but it will be better to clarify the following points.
>>>
>>> 1: How to calculate the PAC timer is not very clear to me.
>>>    Does this draft recommend to use the equation described in Section
>>> 14.3 of
>>>    RFC8445 or are there other ways? I think this would be better to be
>>>    clarified.
>>>
>>
>> Yes, the equation in 14.3 coupled with the STUN backoff guidance in
>> https://tools.ietf.org/html/rfc5389#section-7.2, although these values
>> are just recommendations. The point here is to say that whatever is used
>> for the check timeouts should also be used for the PAC timer.
>>
>
> Right. I was just wondering that it might be useful to mention the
> recommended value can be calculated from the equation.
> If the readers know what'll be the recommandation value, I think they can
> have some ideas about whether their values are conservative or aggressive.
>
>
>>
>>> 2: I presume this draft only focuses on UDP candidates, but I think
>>> clarifying
>>> it would be useful.
>>>    I am also wondering how to treat PAC timer if agents have a mix of
>>> TCP and
>>>    UDP candidates.
>>>
>>
>> The guidance here applies to both UDP and TCP candidates. It would not be
>> unheard of for a server to only offer TCP candidates, and the client to
>> offer zero candidates, as in S 3.1.
>>
>
> I see. But, in this case, will there be a need to update RFC6544?
> Also, if we set PAC timer around 500 msec but establishing a TCP
> connection takes longer than it, should it be considered failed or not?
>
> Given RTO floor of 500 ms and exponential backoff per 5389, the PAC timer
will typically be around 30 seconds. Perhaps a note to this effect would
clarify this and point #1.