CURRENT_MEETING_REPORT_



Reported by John Moy/Proteon

OSPF Minutes

The OSPF Working Group met Thursday, March 14th at the St.  Louis IETF.

The main topic of dicussion was the OSPF MIB. Fred Baker and Rob Coltun
had published the OSPF MIB as an Internet Draft, and had received a
number of comments, especially from the Network Management Directorate.
Their chief comment was that there were too many variables.  So the
Working Group went through the MIB variable by variable, to see which
variables could be combined and which could be discarded.  We ended up
reducing the 127 original variables to 101.

A number of other things were also discussed:


   o Fred Baker pointed out that, for all MIB groups that can be
     dynamically created/destroyed, there needs to be an extra variable
     to accomplish deletion.  This will add to the size of the MIB.
   o Some time was spent in discussing how to do route filtering in OSPF
     at area boundaries.  This can be done through selective
     non-advertisement of certain area ranges.  John Moy pointed out
     that this is dangerous to do in transit areas.  It was decided that
     since this type of route filtering is not covered in the OSPF spec,
     it would not be mentioned in the OSPF MIB either.
   o Dino Farinacci mentioned a problem he encountered where an OSPF
     router (call it router X) received an external LSA originated by
     router Y which listed one of X's IP addresses as the forwarding
     address.  After performing its routing table calculation, router X
     ended up pointing at itself for the destination.  It was agreed the
     the spec should say that, when performing the routing table
     calculation, those external LSAs whose forwarding address is one of
     the router's own addresses should be ignored.  [Editor's note:  The
     very existence of such an advertisement probably indicates a
     problem in the exchange of information between OSPF and other
     protocols such as RIP/EGP].
   o We agreed that all configuration items in the MIB will have
     read/write access, but that the MIB will say that in the absence of
     adequate authentication the variables can be implemented as
     read-only.


There was not enough time at the meeting for examination of the OSPF
Trap MIB that Rob Coltun has created.  We plan on discussing this

                                   1






document via electronic mail.

Attendees

Fred Baker               fbaker@emerald.acc.com
Rob Coltun               rcoltun@trantor.umd.edu
Dino Farinacci           dino@3com.com
Jeffrey Honig            jch@devvax.tn.cornell.edu
Ann Kerr
Donald Merritt           don@brl.mil
John Moy                 jmoy@proteon.com
Michael Reilly           reilly@pa.dec.com
Glenn Trewitt            trewitt@pa.dec.com



                                   2