> I've been receiving quite a few of these lately:
>
>  Error  occured  while processing  file  (5944)  from TARIM@TRBOUN:  "Sorry,
>  validity check failed. Try an IBM NOTE command. The problem might be due to
>  inaccurate  MAILER tables  at  your  node". File  has  been transferred  to
>  [log in to unmask]
>
> I am assuming  that the 'MAILER tables' it  refers to is ours.  All of these
> originated from sites running CDC, [log in to unmask]  I have looked at our
> MAILER table and the entries for them do exist:
>
>  Outgoing -
>       TRBOUN     TRBOUN    ?  DEFRT  1
>       CALSTATE   CALSTATE  ?  DEFRT  1
>       GRATHUN1   GRATHUN1  ?  DEFRT  1
 
This is an a problem that CALSTATE has been dealing with for some time.  As
a CDC CYBER shop, my knowledge of IBM NJE protocols is limited, but here is
my understanding of the problem:
 
There are, apparently, three or four places within the NJE header that indicate
the source user name: among these are origin remote name (ORN) and owner user
name (OUN).  It is the nature of the CDC NJE implementation to set the ORN
to the user who sent the mail; however, the OUN is set to the user name of
the user under which the mailer job is running: SYSTEMX.  SYSTEMX is akin to
root on a Unix system.  The UMASS mailer software running on CYBERs executes
from SYSTEMX, hence there is one field that does not match the others.
 
Near then end of 1988, Cornell came on line as an INTERBIT, and this problem
caused some rather nasty mail looping.  As it turned out, CORNELLA and CORNELLC
had some local code which incorrectly altered the source user name in NJE
transmissions.  The evidence for this was that the interactive progress
messages which indicate source user and node, present node, and ultimate
destination user and node had the source user changed from the actual sender
to SYSTEMX (taking SYSTEMX from the OUN field).  Thus, any incorrectly addressed
mail that passed through Cornell was rejected to [log in to unmask]  (Actually,
it was rejected to SYSTEMXA@CALSTATE due to a bona fide bug in the CDC NJE
emulation software).  Replies to messages passing through Cornell were sent
NOT to the message sender, but to SYSTEMX.  Cornell has since removed the local
code (which was in RSCS) and all was fine.
 
Now, node UICVM seems to have caught the same problem.  Foolishly, I discarded
the message sent to me by the RSCS man at Cornell which identified the errant
code and the fix.  This was before UICVM.  I am working on tracking down the
code indentifier.
 
Node UCLAVM has a similar problem based on the same cause, but the mailer
software their is either not XMAILER or has some hefty local mods.  I forget
which.  It rejects mail to unknown users back to the OUN, SYSTEMX.  But, mail
getting through to an existing user will reply to the actual sender in ORN.
 
Also, for the longest time, UWAVM kept rejecting mail from CALSTATE with a
message saying "user AT CALSTATE DOES NOT MATCH RSCS ORIGIN OF user" (or
whatever the CHARLATAN response was).  The interesting thing about this
rejection was that "user" was the actual sending user name, not SYSTEMX.  In
fact, in the rejections from UWAVM, "SYSTEMX" showed up nowhere.  It was not
until the Cornell problem that I figured that SYSTEMX was the probable cause.
 
I mention these cases as much for background purposes as I do to introduce my
next point:  It appears to me that the BITNET IBM population has no real
consensus on the function of all those source fields.  This has kept me from
applying any local code at CALSTATE (everybody hates having to modify vendor-
supplied software :-))  Although I have no direct evidence, I suspect that the
implementors of NJEF (the NJE emulation software) followed the letter of the
law when writing the code:  CDC may have funny character sets, but they DO
adhere to standards quite well.  (CDC BASIC and Pascal are hideously standard.
Just try converting GENROUTES to standard Pascal... :-<)
 
Anyways, I would VERY MUCH like to solve the "CDC problem."  Those of you who
have had problems, could you please correspond with me via POSTMAS@CALSTATE?
Also, anyone who can be of assistance, especially in the NJE/RSCS area, feel
free to correspond with [log in to unmask]  POSTMAS@CALSTATE is me, as well
as 1GTLEJS@CALSTATE; the latter is my personal account, the former is the
official consultant's address.  By the way, if any mail to either of the
two accounts is rejected due to mailbox full, wait a day or two and resend it;
I accidentally sent out a couple of global signoffs from these accounts which
are not the CALSTATE NAD... rejections are still trickling in.
 
I'll sumarize the results to this and other appropriate lists.
 
 
Ed Skochinski                      (213) 985-9614.
Operating Systems Support
The California State University
 
An additional note:  any mail sent to any address@CALSTATE beginning with the
7 letters "postmas" will get to the postmaster account.  I wont be in the
office until July 5th.