Minutes for NTP at interim-2016-ntp-2
minutes-interim-2016-ntp-2-1

Meeting Minutes Network Time Protocol (ntp) WG
Title Minutes for NTP at interim-2016-ntp-2
State Active
Other versions plain text
Last updated 2016-04-01

Meeting Minutes
minutes-interim-2016-ntp-2

   NTP/TICTOC WG Interim Meeting
10 March 2016

Karen reviewed the Note Well and asked that WebEx chat, rather than
jabber, be used for during-meeting IM.

Denis reported on the status of the BCP.  Harlan and Dieter have been
helping, with Denis still leading the editing.  The doc is available
in github for those who want to edit it directly.  He still needs help
on load balancing.  Greg Dowd had volunteered before to write that
text and reiterated his willingness today.  Karen wants to get this
shipped and is concerned about the (lack of) security content, since
initial impetus was for security issues.  Denis will post a list of
what's needed - he thinks the doc could use more time to "simmer" and
suggested pushing the deadline a few months.  We'll plan to have a
conversation re: deadline in April; Denis and others will be prepared
in April to talk re: path/timeline to publication.

Dieter reports that he's pleased with the WGLC on NTS.  Two open
topics: selection of MAC algorithm and how NTS cooperates with NTP's
symmetric mode.  The first one is close to resolved.  The second one
is impeded by the poor description of NTP symmetric mode in other
docs.  Dieter would like a pointer to a document that better describes
symmetric mode.  Failing that, he'll work with Danny and Harlan to try
to nail this down.  Kristof Teichel expressed frustration re: how hard
it is to add security bits to an under-specified protocol.  Danny
Meyer reiterated that he requested a clarification that NTS does not
validate timestamps.  Harlan reiterates that we've only been working
on NTS for unicast, not broadcast or multicast NTP.  Karen requested
that Harlan post something to the list on this, just to have it on the
record.

Karen asked Richard to forward to the NTP list the email he sent to
IANA re: early allocation.  Karen doesn't expect any changes to that
request.  Richard pointed out that the main reason for an early
allocation was an OpenSSL deadline (~two weeks from now).  He's gotten
automated acknowledgement from IANA, but no human acknowledgement -
the process is opaque.

Harlan points out that "the IANA document" may need some work, generally.  
As an example, autoconf has some allocations that aren't listed in the IANA 
document.  He thinks reserving 1-7 may not be appropriate, as autokey is 
deprecated.  Karen asked Harlan to identify the document he's talking about. 
(He later identified it as the IANA NTP registry, available at 
https://www.iana.org/assignments/ntp-parameters/ntp-parameters.xhtml)

Karen reported on some other document status.  [I missed some of this.]

1588 MIB has been updated; it's on the IESG telechat next Thursday.

1588-over-mpls waiting for shepherd write up from Yakov.

We're expecting an AD change in April; Suresh will be our new AD.
Still planning to propose combining NTP and TICTOC.  Charter work has
not yet happened.


Karen started a discussion re: how to proceed with further
clarifications and fixing existing documentation problems, with the
example being the checksum and general extension docs.  Karen thinks
we don't have resources for a full NTPv5, but it may make sense to
collect to requirements for v5.

Harlan is aware of things NTPv4 can't do; he thinks a version 5 is
needed to fix some of these.  But, absent more resources, he thinks
"design by committee" is doomed to failure.  He proposes starting with
implementation and then writing RFC based on the implementation.
Danny and Harlan observe that this is how NTP has evolved before.

Sam Weiler excoriated Harlan for spurning the IETF, positing that
documenting things before or contemporaneously with implementation
need not be more burdensome than the implement-first,
document-over-the-following-decades approach taken to date.  The
example of the poorly-specified symmetric mode, which is currently a
stumbling block for NTS, as above, was offered an example of the pain
inflicited by the implement-first approach.

Danny isn't sure what can be shoehorned into v4 v. what really needs a
v5.

This discussion was tabled 'til April.  Karen would appreciate thoughts 
on how to frame this conversation.  Karen hopes to have a report from the 
other NTS implementer (Martin).

Danny points out that he and Harlan are working on a new MAC extension
field doc.

Harlan says he's working on a draft on Suggested REFID.

Next meeting @ IETF 95 in Buenos Aries.