CURRENT_MEETING_REPORT_
Reported by Kent England/BBN
Minutes of the Network Joint Management Working Group (NJM)
Canonical Trouble Mailbox
What if we all defined a mailbox named ``trouble@your.net'' for receipt
of network trouble reports?
What if we all defined a finger-name ``noc@your.net'' so users could
receive a small bit of information about our respective NOCs?
There was some general discussion of DNS records needed for this. Many
nets don't have a machine called ``your.net'', but use MX records. Some
don't use MX records for ``your.net'', but instead use a machine named
``noc.your.net'' or similar. Finger requires an address record, but an
alias record could provide some indirection for ``your.net''
Vikas Aggrawal, JvNCnet, noted that he had posted a note on the
namedroppers list discussing this. [Vikas, repost to njm?]
So what names should we use? Joe Ragland, CONCERT, doesn't like ``noc''
and Carol Ward from WestNet doesn't like ``trouble''. Gene suggested
the name ``net-trouble@noc.your.net'' for the canonical mailbox, but
this is too long for finger, so use ``noc@noc.your.net'' for finger.
Dan Long, NEARnet, will add a field to his NOC PhoneBook entry for nets
to include the preferred mailbox name.
Then Gene raised the issue of DNS inverse address-to-name lookups (PTR
records). Gene suggested that all router interfaces should inverse
lookup to a descriptive name, and, in addition, the host zero address
should invert to something descriptive [of what?]. Vince Fuller noted
that DNS already maps host zero to the gateway name. This is left as an
unresolved issue.
John Curran, NEARnet, noted upcoming Responsible Person DNS records that
we will find useful. Gene Hastings reminded us all to fill out
information for Dan Long's NOC PhoneBook entries. Dan said he would
send a note to njm with changes to the entry information.
Gene raised the issue of ``nsr'' mailing list usage, and suggested a new
Merit list for discussion of internet woes called ``internet-
ops@merit.edu''. Ittai Hershman, ANS, said there had been some
discussion in Merit/ANS about moving discussions from nsr to nwg or
Gene's suggested new list. The concern is about too large a group of
readers engaging in too much traffic of discussion and diluting the
quality of the nsr list for operational people. John Curran noted that
individuals will use whatever list they can find to report trouble and
we need to educate users as these misdirections are corrected. Ittai
noted that this had been discussed this morning in UCP in the context of
1
machine parseable messages, but these lists don't scale well. Dale
Johnson, Merit, noted that this issue parallels the issue of usage of
ietf list and that we need to create ``nsr-discuss'' as has been
suggested for ietf.
Joe Ragland said that one reason we need this list is to relieve Merit
of unrelated traffic. FARNET members need an independent channel for
communication. Cathy Witbrodt [sp?], ESnet, asked if users aren't going
to use this list and Ittai said UCP has addressed this concern. Dan
Long noted that service providers need a discussion list for themselves.
We agreed to use the existing nwg mailing list as an ``nsr-discuss''
list and to use njm as a meta problem discussion list. [So nsr remains
the channel for Merit to send out announcements, nwg is for discussion
of operational problems, and njm is for meta discussions about ops.
-kwe]
Tricks of the Trade
Vince Fuller, BARRNET, noted that he is tired of reports to BARRNET from
users that say that BARRNET is broken, when in actuality these users are
simply unable to traceroute across the MILNET. How can users be made
more aware of the limitations of traceroute?
Jordan Becker asked if every AS has a reliable host for pinging and
tracing? Dan Long noted he will include such an entry in his NOC
PhoneBook.
What about test servers to test telnet, mail, etc? Dan Long noted the
success of the NEARnet mail bouncer [bouncer@nic.near.net] as a very
useful tool for site contacts to use to test mailers. This automatic
bouncer has reduced the workload on NEARnet operations and analyst staff
tremendously and is seen as a very valuable service [almost free].
How do we associate network numbers to the AS announcing them? Jessica
Yu of Merit noted that they have this file at Merit [net.now?] Ittai
noted that new tools for getting and updating this information are under
development [ANS?] and we should be hearing more about this in the near
future.
Cathy, ESnet, noted that she builds router access control lists from
this file and Merit should announce changes to this file format in
advance to avoid Cathy having problems like the day she lost NEARnet
when it went over to the T3 backbone and the file format changed.
Gene noted that Van Jacobson [everyone genuflect] has a new path
characteristics analysis tool that analyzes paths per hop.
John Curran noted that the NNSC is doing another issue of the Internet
Managers PhoneBook on paper and electronically.
Dial-ups and Serial Port Servers
Dave OLeary, SURANet, noted that he had sent a note to regional-techs
2
asking for information on dial-up service and hadn't gotten much
response. This started a discussion of dial-up servers, or serial port
servers. Gene Hastings noted that PSC has a NetBlazer dial-up with SLIP
for schools to use. Lines are shared.
Dave OLeary asked about the Livingston product and Brian Lloyd and Vince
Fuller seemed to know most about this new product.
There was some discussion about the difficulty of using a NetBlazer as a
router. Seems NetBlazers don't do dynamic routing very well. And then
there is the difficulty of tracking hosts amongst serial ports.
Milo Medin, NSI ``bureaucrat'', noted that dial-up servers could use
OSPF and advertise host routes in order to solve the host tracking
problem.
The question was asked about who all sells serial port servers and the
list looks like:
o Cisco TRouter (NEARnet uses)
o Xylogics Annex
o NetBlazer
o Xyplex has something
o NAT cheap router (see Vince Fuller for more info)
Other Business
As we seemed to have used up all our time, the other agenda items were
deferred.
Attendees
Vikas Aggarwal aggarwal@jvnc.net
Jordan Becker becker@nis.ans.net
John Curran
Tom Easterday tom@cic.net
Kent England kwe@bbn.com
Vince Fuller vaf@stanford.edu
Jack Hahn hahn@sura.net
Eugene Hastings hastings@psc.edu
Ittai Hershman ittai@nis.ans.net
Dale Johnson dsj@merit.edu
James Jokl jaj@virginia.edu
Dan Jordt danj@nwnet.net
Mark Knopper mak@merit.edu
Walter Lazear lazear@gateway.mitre.org
Daniel Long long@nic.near.net
Milo Medin medin@nsipo.nasa.gov
Donald Morris morris@ucar.edu
David O'Leary oleary@sura.net
Marsha Perrott mlp+@andrew.cmu.edu
Joe Ragland jrr@concert.net
3
Ron Roberts roberts@jessica.stanford.edu
Harri Salminen hks@funet.fi
Tom Sandoski tom@concert.net
Bernhard Stockman boss@sunet.se
Roxanne Streeter streeter@nsipo.nasa.gov
Claudio Topolcic topolcic@nri.reston.va.us
Carol Ward cward@spot.colorado.edu
Cathy Wittbrodt cjw@nersc.gov
Jessica Yu jyy@merit.edu
4