> Roger, June;  I don't think we're losing any mail; one of the costs of
> a distributed list system is that every mail 'stops' more often, and
> there's more chance of mail getting into SIFO rather than FIFO order; so
> that when a discussion is started with a *very long* mail, as it crawls its
> way through the servers folks answer it right away with short files that may
> get to those of us out in the sticks a lot sooner...
> /jenny
> (geez, that was all one sentence, wasn't it? sorry... been reading a lot
> of Faulkner lately :-)
 
I am aware of the shortest job first queuing.  It does appear that there may be
a bottleneck between me and UIUCVMD, the server that I am on.  One of the
messages about ARPA digests arrived here about 12 hours after it had been sent.
However, the message that started the discussion still has not arrived, nor has
the message that I complained about not getting several days ago.
 
I just tried a test of sending a short file from here to UIUCVMD and UGA.  The
path to UIUCVMD seems to be NIHCU-UMDD-UMDB-GWUVM-PSUVM-OHSTVMA-UICVM-UIUCVMD.
The file (one line) waited longest at UICVM, next longest at PSUVM.  The path
to UGA seems to be
NIHCU-UMDD-UMDB-GWUVM-PSUVM-OHSTVMA-UCCCVM1-UTKVM1-UTCVM-UGA.  The UGA file
arrived first, even though it was sent later.  Of course, I am sending in the
opposite direction from LISTSERV mail, delays may be different the other way
and it was a one line file.
 
PHIL@UIUCVMD was nice enough to send me a week's worth of messages from this
list.  I found at least 3 that had never arrived here, the two that I had
mentioned and one other.
 
On the subject of the ARPA digests, today I just received my third copy of
Info-IBMPC Vol. 5, No. 101.  It is not unusual for me to receive multiple
copies, although I don't always get more than one.  I asked the digest editor
about this once.  He said that if something interrupted the process of sending
the digests to BITNET, they would start over again from the beginning.