[comp.mail.misc] collapse of EARN

emv@math.lsa.umich.edu (Edward Vielmetti) (07/29/90)

FYI.  If you are expecting mail from somewhere in EARN-land, be patient.

--Ed

Edward Vielmetti, U of Michigan math dept <emv@math.lsa.umich.edu>
comp.archives moderator

From: ERIC@SEARN.BITNET (Eric Thomas)
Newsgroups: bit.listserv.linkfail
Subject: Serious network problems in Europe
Date: 27 Jul 90 14:29:29 GMT
Reply-To: Eric Thomas <ERIC@SEARN.BITNET>
Comments: cc: LSTSRV-L@SEARN.BITNET, EARN-NOG@FRMOP11.BITNET

EARN is presently  suffering from a number of  serious problems affecting
the delivery of files to european users:

1. The 3705 controller at CEARN is  "dead". Austria, India and the UK are
   disconnected, and the BSC line to Germany is also affected. I have not
   been able to contact anybody at CERN  who could take action on this; I
   expect  the problem  to be  solved by  monday, but  it is  unlikely to
   happen before that.

2. LISTSERV@CEARN presently has 2,000 files  in its reader (this has been
   steadily increasing during  the week), which means  very long delivery
   delays for distribution lists.

3. LISTSERV@FRULM11 had  over 6,000 files in its reader  last friday; the
   backlog is now almost gone, but the FRULM11 spool is full with a queue
   of something like  1,600 files, the smallest one  being 2,000 records.
   This is the reason why this server is often not logged on, and this is
   why the  distribution of  large postings to  EARN is  basically halted
   until the situation improves. A 64k  line for FRULM11 has already been
   ordered, but it will probably not be delivered before october.

One of the causes of the last  2 problems is the "duplicate list posting"
problem that we started experiencing about  3 weeks ago. A broken RSCS or
MAILER at  some site  which we  have unfortunately not  yet been  able to
identify yet is sometimes resending  an unpredictable amount of copies of
the very same mail message (from 2  to 200); you can imagine what happens
when  these messages  hit a  popular distribution  list like  INFO-VAX. A
LISTSERV fix for release 1.6e (16E-003O FIX from LISTSERV@SEARN) was made
available to  prevent the  distribution of  such duplicate  postings, but
those which are already "in the  pipe" cannot be eliminated; in addition,
not  all  servers are  running  with  this  fix  yet, and  a  significant
proportion of EARN servers do not run release 1.6.

Another cause  is a swiss  node administrator who generated  around 1,000
netwide signoff jobs  for deleted student accounts. A third  cause is the
increasing  popularity of  the  BITFTP server,  which unfortunately  does
nothing  to help  the RSCS  backlogs in  the north  of France.  There are
probably other causes which I have not yet found out about.

  Eric