Minutes of RIPv2 Meeting in Amsterdam

["Acee Lindem - [e-mail]" <[e-mail]>]

 Regarding routing domains, if there was a requirement for this construct
 in the RIPv2 internet draft it doesn't seem that removing it satisfies
 that requirement. However, I don't have the justification to argue this
 point.

 Regarding the MIB, I really think the indexing for unnumbered interfaces
 should be handled in the same manner as in the OSPF mib (i.e., with an
 additional qualifier (e.g. INDEX { ospfIfIpAddress, ospfAddressLessIf }).

 Additionally the following items need to be clarified for MIB implmentors:

  1.  The rip2GlobalRouteChanges object should be qualified to indicate that a
      refresh of the route's age does not constitute a route change.

  2.  The rip2IfConfDefaultMetric object is not well explained and at least on
      the surface looks like a remnant of an implementation.  If this implemen-
      tation dependent object is to be included than why not support the
      objects equivalent to the more general and useful GateD metric-in and
      metric-out interface parameters?

  3.  The rip2IfConfReceive object should support a value indicating not to
      listen to rip updates in order to be consistent with the "doNotSend"
      value for the rip2IfConfSend object.

  4.  The RFC must be more specific as to what constitutes an active peer.
      For example, an active peer could be defined as one on an accessible
      network which has been heard from in the last 180 seconds.

  5.  Consideration should be given to coordinating rip2PeerLastUpdate
      and rip2PeerVersion. In other words, why replace rip2PeerLastUpdate
      with rip2PeerLastPacket or define rip2PeerVersion to be the version
      of the last update packet.

Thanks,