Tue, 16 Oct 90 18:27:34 SET
|
Sorry for x-posting ...
The recent problems with VM-UTIL "motivated" me to do some investigations. For
all who don't know the TREARN peer of VM-UTIL will be un-peered because of
problems with jo-jo'ed mailings (which fortunately were rejected at the other
peer).
I wont comment on this very problem. All I did was sending some RELEASE
commands to various LISTxxRx's and invetigate some list headers. Well, there
are at least 5 different kinds of listservers around:
LISTSERV 1.6e - current LISTSERV release
LISTSERV 1.nx - backlevel LISTSERVs
LISTEARN 1.1 - soon-to-come (?) LISTEARN release
LISTEARN 1.0 - current relase
LISTSERV 1.5o maintained by EARN
The reason why I'm bringing this to your attention is that not only the code
differs but also the network-related control files. The current mixture can
indeed lead to funny results:
A subscription request for a peered list is in the worst case forwarded to
three different peers depending on the LISTxxRx I chosse to send the request
to. Whilst this problem is known to pop up in a transition phase (usually less
than one week) it is now a permanent one.
To give a more lively example:
AWIIMC11 is 1 node downstream from AEARN. LISTSERV@AWIIMC11 (running LISTSERV
1.6e) insists that my nearest host for VMUTIL is DEARN. LISTSERV@AEARN
(running LISTEARN) is absolutely convinced that TREARN is the best placement.
While this may only be some insignificant nuisance keep in mind that
DISTRIBUTE uses the same view of the network ....
Well, I don't want to start another war but the facts are that this makes
(slowly but steady) peering of lists where LISTSERVs and LISTEARNs are
involved more or less ridiculous - EXPLODE will be a two-step action, first
explode the list on - let's say - a LISTSERV and after the movement explode it
again on the target LISTEARNs or v.v.
'Nuff said for know ....
Christian
|
|
|