NTP WG Meeting Minutes 
1300-1500, 8 November 2006 

Karen O'Donoghue, co-chair of the NTP WG, called the 
meeting to order. Dave Marlow volunteered to take 
minutes, and Jim Martin agreed to act as jabber scribe. 
Brian Haberman (NTP WG co-chair) participated remotely 
via audio stream and jabber. 

Slides: NTP WG Agenda and Administrivia (ntp-0.pdf)

Karen addressed basic administrative issues including blue 
sheets and agenda bashing. The milestone updates for the 
NTP WG were discussed. The Scope and Requirements 
document has been deferred because it was determined not 
to be needed to complete the initial charter goals of 
documenting NTPv4. The NTPv4 Protocol Specification 
and the Algorithm document have been merged into the 
"Network Time Protocol and Algorithms Specification 
Version 4. The MIB document remains unchanged. 
Because the PDF experiment was not approved it was 
decided to publish the documents in ASCII with the source 
for in XML 

*	NTPv4 Protocol Specification 	(Jack Burbank) 

Slides: NTPv4 Protocol Draft (ntp-1.pdf)
Relevant Draft: draft-ietf-ntp-ntpv4-proto-03.txt

The unification of protocol and algorithms drafts into a 
single draft was based on Dave Mill's University of 
Delaware reference report.  There were numerous changes 
to the original University of Delaware report including the 
addition of IPv6, security, and IANA considerations.  
Material was pulled in from the earlier drafts including 
information on the Kiss of Death and frame formats.  An 
early proto-04 was distributed to the list which had a 
number of editorial cleanups and the addition of normative 
language (i.e. MUST, 
MAY, and SHOULD).  The plan is to keep the early proto-
04 open for two weeks asking the Working Group to 
provide comments during this time period after which an 
updated protocol draft will be submitted for Working 
Group last call. 

*	NTPv4 MIB	(Chris Elliott)

Chris Elliott provided a detailed review of the MIB to the 
Working Group's email list prior to the meeting.  An 
updated MIB draft is in preparation.  A number of Chris' 
comments have restructured the MIB using the conventions 
currently used.    

A new object, ntpSrvStatusTime, was proposed by Chris to 
express the current 
time within the box.  Chris felt that this object is needed not 
only for NTP but for SNMP in general.   Chris wants a 
reasonably accurate (milliseconds scale) using a 64 bit 
counter.  Integers in SNMP are limited to 32 bits, and this 
is why the object has been proposed as a counter.  The units 
used by this counter are still to be worked out, but Chris 
initially made it the NTP time (time since the year 1900 
with seconds in the upper 32 bits and fractions of a second 
in the other 32 bits).  Chris believed that this would meet 
all accuracy requirements.  A jabber comment asked why 
not use UTC instead of generalized time?  Someone 
pointed out that the UTC time scale is not monotonic at the 
time leap seconds are inserted. Brian Haberman, through 
jabber, said that UTC is the way to go if this is possible.  
Chris will look further into the issues associated with the 
format of this object and propose a resolution. 

Additional new objects (ntpSrvHeartbeatInterval and 
ntpSrvNotifBits) provide an ability to send periodic 
notifications.  Karen asked Chris to fix the time object soon 
and to try and keep the MIB document on the same 
schedule as the Protocol document.

After a final discussion on the schedule for getting these 
two documents to WG Last Call, the meeting was 
adjourned.