Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!umich!samsung!brutus.cs.uiuc.edu!psuvax1!psuvm!TECMTYVM!POSTMAST
From: POSTMAST@TECMTYVM.BITNET (Juan M. Courcoul)
Newsgroups: bit.listserv.nodmgt-l
Subject: Re:  Information for node CWUVAXEN is incorrect
Message-ID: <900206.192650.EDT.POSTMAST@TECMTYVM>
Date: 6 Feb 90 23:26:50 GMT
Sender: Node Management Discussion <NODMGT-L@BITNIC>
Reply-To: Node Management Discussion <NODMGT-L@BITNIC>
Organization: Instituto Tecnologico y de Estudios Superiores de Monterrey
Lines: 40
Approved: NETNEWS@PSUVM Gateway
Comments: cc: NETSERV Maintainers List <NETSRV-M@HEARN.BITNET>
In-Reply-To:  Message of Tue, 6 Feb 90 16:41:10 EST from <RAF@NIHCU>

Mulling over the problem regarding CWUVAXEN and asking ALL the nodes to
hand-edit their routing tables, I'd like to propose the following:

Given that:

a) BITNIC's NETSERV still hasn't received the VERS9002 update.

b) A routing table generation request cancels any previous one that
   hasn't been carried out.

And assuming that:

c) The distribution path of the VERS9002 update in BITNET/CREN starts at
   BITNIC and works on down the rest of the NETSERVs.

d) The node management procedures regarding the routing tables generated
   and distributed via NETSERV in EARN have been long since ironed out
   and there is a better control there, given that there are fewer nodes.

It would seem logical to send the corrected VERS9003 update ASAP so that
both this and VERS9002 would arrive to the servers within a few hours of
each other and be applied before midnight of the same day, when the table
generation process is triggered. If my line of reasoning is correct, there
should only occur one table generation, with the last (VERS9003) version
of BITEARN NODES in place.

Granted that it may be difficult to ensure that this sequencing occurs,
given the state of the link queues, perhaps all of us NETSERV controllers
can help by putting our servers offline until both updates arrive, sort them
in the requisite order AND THEN reactivate the servers. (This may mean
LOGGING OFF a NETSERV for a day or two).

It seems much more feasible to control the operation of the 15 to 20 NETSERV
servers in BITNET/EARN over a period of a couple of days, than have the
more than 3,100 nodes carry out the hand-modification of their tables.

Hopefully, this suggestion will arrive BEFORE the update... ;-)

Juan M. Courcoul
Mexico NETSERV Controller