Procedure for fixing multicast problems

[Bill Nickless]

In light of the fact that IP multicast routing is entirely dependant on
attempted use to debug and fix, I would like to suggest the following
elements of a procedure to follow tomorrow (Wednesday, 15 September 1999).

- 8:00 AM EDT (7:00 AM CDT, etc): All sites that are theoretically
  capable of doing native IP multicast audio/video will attempt to do 
  native IP multicast.  The network folks will work on the following sites,
  in order, to debug and fix IP multicast:

  Boston University (video and audio, especially to ANL)
  NCSA ACCESS-DC (video and audio)
  Argonne
  Utah
  Kentucky

  These sites are known to require bridging until additional
  network enhancements are made that cannot be done in realtime
  tomorrow.  Accordingly, we should "triage" them for later work.

  Moscow
  NCSA in Illinois
  New Mexico
  Maui High Performance Computing Center

- 8:45 AM EDT (7:45 AM CDT, etc): An inventory of sites will be taken,
  and those sites that are unable to participate in IP multicast audio/
  video due to routing problems will be instructed through the AG MUD
  which bridges to use.

Thereafter, at any break 30 minutes or longer, sites who are having native
IP multicast problems should *stop using the bridges* and attempt to use
native IP multicast.  15 minutes before showtime would be the "drop dead"
time for the network people to stop debugging multicast and to assist in
inventory of sites to switch back to the bridges as necessary.

According to http://chautauqua.bu.edu/chautauqua/day_two.html the scheduled
breaks will be at these times (EDT):

   10:15 am: Break (30 minutes)
   11:45 am: Lunch (75 minutes)
   3:00 pm: Break (30 minutes)

During these breaks, it would be very useful to have network engineers for
the various sites standing by, preferably reachable in real-time
(telephone, AG MUD) to fix problems in real time.  I will be happy to start
a teleconference for network engineer debugging discussions.  Access Grid
node operators ALSO need to be on-line in the AG MUD to report status as
the debugging progresses.

I fully intend to be online at the Argonne Access Grid node before 8:00 AM
EDT, and am looking forward to working with all of you on this.
===
Bill Nickless    http://www.mcs.anl.gov/people/nickless      +1 630 252 7390
PGP:0E 0F 16 80 C5 B1 69 52 E1 44 1A A5 0E 1B 74 F7     nickless at mcs.anl.gov