The explanation is simple:
 
1.  UCHCECVM  installs  LISTSERV.  Their server  appears  to  be  working
   perfectly.
 
2. They send me  mail explaining that they are the  central node of Chile
   and would like  to join the backbone. This sounds  quite reasonable to
   me,  and,  although  they  only   recently  installed  the  code,  the
   maintainer seems  to be quite competent  (from what I can  gather). At
   least,  he  seems  proficient  with   REXX  and  assembler,  which  is
   unfortunately  not always  the case,  and sounds  like he  has several
   years of experience with VM. So, I add them to the backbone.
 
3.  For some  reason, I  had sort  of implicitly  assumed that  Chile was
   connected to  Mexico. It never  occured to me  that they could  have a
   direct line to the States, and that putting them on the backbone would
   therefore cause US  stuff to go through them (the  Mexico LISTSERV was
   already on the backbone, so I figured it wouldn't change anything).
 
4.  Gonzalo contacted  me  to say  that he  was  suddenly having  serious
   problems with  LISTSERV, and wanted  to be removed from  the backbone.
   This was  something that I  couldn't do in less  than one week  or so,
   since the  CEARN-MOP queue was around  2500 (this was the  famous week
   where FRMOP22 was  down 37.7% of the time). Remember  that PEERS NAMES
   is  > 1700  records.  Gonzalo's problem  was  that, apparently,  files
   coming from RSCSV2 (his network RSCS) were processed with an origin of
   'RSCSV2@UCHCECVM', which  caused of course  some problems. To  avoid a
   possible loss  of mail files,  he decided  to stop LISTSERV  until the
   problem was solved. However I didn't get all this information at once,
   it took me  several questions-and-answers to understand  the nature of
   the problem.  To me it  looked like  his LISTSERV had  been configured
   with the  wrong RSCS userid,  which ought to  take some 30  seconds to
   fix. During this time, Gonzalo made traces over traces of the REXX and
   assembler code, in an attempt to determine why it was not working.
 
From what I understand they were running a non-Crosswell mailer which was
retagging  the output  files (and/or  using  diag X'D4'  to change  their
origin?), and for some mysterious reason  it did not cause any problem at
the beginning.
 
In any case this was all my fault, not Gonzalo's. I should have seen that
Chile was connected to the States,  and changed the weights before adding
them to the  backbone. Gonzalo did his  best to prevent the  loss of mail
files, which meant taking LISTSERV down  for a while and spending as much
time as he could tracking down the problem.
 
  Eric