Santa Fe Minutes

[Frank Kastenholz <[e-mail]>]

Here are the meeting minutes that I mailed to Megan at CNRI
for inclusion in the proceedings.

Frank



Ethernet MIB Working Group
Meeting of 21 November 1991
IETF Meeting, Santa Fe, New Mexico

Chairman: Frank Kastenholz

The Ethernet MIB Working Group met at the Santa Fe IETF
meeting on Thursday, 21 November 1991.

The first order of business was to review the administrative
issues surrounding the working group:

1. The charter and the "Pax Davin" were reviews and discussed
   at length. The fact that we will strictly adhere to the
   rules of charter and the "Pax Davin" was openly discussed
   and agreed to by all present.

2. The working group decided to try to have the documents
   required by the charter ready for publication at the
   same time as the MIB is put forward for Draft Standard
   status. This would be in about 6 months. These two 
   documents are:
   A) An explanation for te assignment of MANDATORY
	  status to optional 802.3 variables, and
   B) An implementation report on the MIB variables.

3. A mailing list will need to be created. 
   [This has been done -- [e-mail]].

After discussing adminstrative issues, the working group turned
its attention to the MIB itself. The following items were
discussed (thanks to Anil Rijsinghani of DEC for his notes of
the meeting). Any changes to the MIB will be made to the 
version to be put forward for DRAFT STANDARD status.

1) The working group discussed the text in the MIB which allows
   an implementation to return 0 for counters for which the
   underlying events are not counted. It was realized that this
   wording makes it impossible to disambiguate the two cases of
   not implementing a counter and and 0 occurances of the underlying
   event.

   The working group discussed the issue and a vote was taken on
   it. The group decided to remove the offending text from the document.
   The working group realized that for variables to which this text
   applies, there are four alternatives which should apply:
   a) Delete the variable from the MIB entirely as its utility
	  has not been demonstated by wide implementation experience,
   b) Move the variable into a separate optional MIB Group,
   c) The implementor must figure out some way to support the 
	  variable, and
   d) The implementor would not implement the variable, return
	  noSuchName errors whenever the variable is accessed and not
	  claim compliance to the MIB.

2) The possibility of an 802.3 specific (as opposed to DIX Ethernet
   specific or common to both) mib group was discussed. It was decided
   to continue this discussion on the mailing list.

3) The dot3StatsExcessiveDeferrals object is implemented in
   only one chip out of 14 studied by Anil Rijsinghani.
   It should either be made optional or removed from the MIB.
   The other mandatory objects are implementable with commonly
   available chips and supporting software.

   This will be considered for further study on the mailing
   list.

4) The TDR definition in the MIB is not sufficient, given that IEEE 
   802.3 does not define this object.  It does not describe the two
   conditions under which the object is defined, and how to
   distinguish between them (short and open cable faults).
   Anil Rijsinghani will contribute text to clarify this.