>
>Why does mail  distributed by LISTSERV to some European subscribers
>(DGOGWDG1) take as long as 7 days from YORKVM1?  Mail sent directly
>to the local MAILER  (not through LISTSERV) reach the same European
>destination  within a few hours.  ...
 
   I am sure someone will chime in with the details, but I noticed that
our LISTSERV has not received distributions from BITNIC's LISTSERV for
long periods of time (eg. 18 hours).  BITNIC is the "closest" LISTSERV
to the east for most of our distributions.  The latest problems seem to
have started this past weekend (I think - time flies too quickly for me!).
 
   As I understand it Cornell had some problems with VM/XA which affected
both RSCS and some other links (see note from Nick Gimbrone dated 23 Oct 88,
20:09 EDT which just got here this morning (26 Oct 88) since LINKFAIL itself
is on BITNIC... ;-)
 
   This has happened before.  It seems silly to overwork the small
(4361?) BITNIC machine with all the distributed LISTSERV traffic
crossing the net.  I think BITNIC may have peered lists so it might
have to stay on the backbone but maybe the CUNYVM-BITNIC link could be
given a high weight to encourage use of LISTSERVs on the various
"sides" of the PSU-CUNY axis.  I know Eric modified the weights on the
link to Chile so stuff wouldn't take the "long way" around...  ;-)
 
   Anyway, a listserv which must be shut down for long periods of time
seems to stretch the definition that Eric provides of a "LISTSERV backbone"
in that section of the LISTINST MEMO.
 
   It seems to me that having a LISTSERV server on each "leg" out from
the hub might be a workable way to minimize traffic without the
"hub(s)" having to have LISTSERV backbone machines themselves.  I am
thinking here that everyone uses DIST2 (right?) for all lists (eg.
Mail-via=DIST2 and not distribute).  Or, maybe the LISTSERV at PSUVM
would assume the role for many items.
 
   At any rate, maybe BITNIC-CUNYVM should no longer be a "weightless" link!
Maybe it would be better to call that an "international" link.  ;-)
 
   I would like to find out more about the current situation which is delaying
mail on lists.  Remember that if your list is DIST2 it is easy to get a
map of the path all the mail from your node will take!
 
   For example, on our AILIST redistribution (DIST2, about 340 subscribers)
the item goes to LISTSERV@BITNIC (except for a few local/direct distributions).
The 2nd hop is to BYUADMIN, FRMOP11, PSUVM, YALEVM and with terminal hops from
BITNIC to JPNKEKVM, NUSVM, and ORION.
 
   Of course, if BITNIC were given a higher weight one of the 2ndary
sites may become our target...  Would have to look at the weights to
figure that out or would have to simulate it.  (Gee - is there a
parameter on a DIST2 Debug=yes job for substituting weights for that
one run?  ;-) Some have said that PSU is really the topological hub of
"bitnet" anyway so maybe that would assume the role that BITNIC's LISTSERV
has now...
 
   Marty (where IS that flame retardant shield?)
 
-----------
Marty Hoag
ND Higher Education Computer Network    US Mail: NDSU Computer Center
Phone: (701)-237-8639                            PO Box 5164 / UCCS
Bitnet:   NU021172@NDSUVM1    (NOTE 0 = ZERO)    Fargo, ND  58105
Internet: [log in to unmask]
UUCP:    ...!psuvax1!NDSUVM1.BITNET!NU021172