2007/272 Project Clearview: IPMP Rearchitecture (Commitment)

[Peter Memishian]

 > >    * Section 3.17: To resolve a conflict with the new architecture, IPMP
 > >      Singleton no longer allows an IP data address to double as an IP test
 > >      address.  More generally, this section has been expanded to cover
 > >      additional issues we (the Sun Cluster and Clearview teams) discovered
 > >      while bringing a Clearview IPMP cluster up.
 > 
 > Does this project obsolete 2002/278 (Singleton IPMP Group) and

It mostly obsoletes 2002/278 (the idea of a Singleton IPMP group remains,
but there's no sharing of data and test addresses).  Oddly, the only
interfaces exported by that case are the IP_DONTFAILOVER_IF and
IPV6_DONTFAILOVER_IF ioctls; those are in the removed interface list for
this case (see ipmp-20q.txt)

 > 2002/763 (Contract for use of SIOCGLIFGROUPNAME)?

Yes, for the SIOCSLIFGROUPNAME part.

 > Is there a new contract to replace PSARC/2002/763-01?

We were planning on providing a new contract once Sun Cluster is actually
supported on Nevada and we (the IPMP team and Sun Cluster team) are in a
position to do thorough testing and discover exactly what they need.

 > >    * Section 5.18: To ensure IPMP-unaware MIB applications won't trip over
 > >      IPMP test address information, a synthetic EXPER_IP_AND_TESTHIDDEN
 > >      MIB level has been added.
 > 
 > This will remove information that used to show up in the MIB, and thus
 > cause a small loss of observability for IPMP machines managed by SNMP.
 > 
 > I think we should have ARC advice added to the opinion: management
 > needs to be advised that while this is an improvement in accuracy of
 > the data, it's also a regression, and that we need to have updates to
 > the SNMP implementation to support IPMP (perhaps via a private MIB).

OK.

--
meem