[news.software.nntp] peculiar problem with NNTPXMIT/QUCDN NNTPRCVR

DOUG@ysub.ysu.edu (Doug Sewell) (03/13/91)

We run a receive-only NNTP program (the NNTPRCVR program written for
PSU NetNews by Andy Hooper - hooper@qucdn.queensu.ca), with IBM FAL
TCP 1.2.2.  Two different Unix hosts send us news via NNTPXMIT - one
is running 1.5.10, the other (?) - they're other universities.

There has been no change to our TCPIP or NNTP software or configuration
in the last few weeks (sigh... I hoped that might turn up something).
I admit only a minimal understanding of the Unix NNTPXMIT program.

Our receiver program will show a host connect.  Some articles may or
may not be sent, then an article in the queue would stall the connection.
Ultimately (five minutes), the connection times out and is closed at
our end.  This is a journal of one connection:

  0 Incoming connection: xxxxxxxxxxx..2023
  0 S: 201 YSUB.YSU.EDU NNTP receiver ready, no posting.
  0 R: IHAVE <34708@nigel.ee.udel.edu>
  0 S: 335 Send the article please.
  0 Inactivity timeout                  << five minutes after '335'
  0 S: 503 Inactivity timeout, goodbye!
  0 Closing
  0 0 articles received 0 refused
  0 Incomplete article discarded, 0 lines
  0 ConnectionStateChange: Receiving only
  0 ConnectionStateChange: Nonexistent  << about 3 min after the last.

(An aside: there's something peculiar about this particular article.
I telneted to another site's NNTP port and tried to read it with
'article <34708@nigel.ee.udel.edu>' and the connection just sat
there, rather than displaying it.  Retrieving news from that site
via NNTPXFER stalls, too - I'd suspect on that article - but when I
reset the NEWNEWS time to skip that article, everything runs fine.
Also, trying to mail that article from this feed site to me fails.)

When a new connection comes in from the same site, any articles that
were received in the connection that timed out would be re-sent.  Most
peculiar is that occasionally the order the re-sent articles are sent
in changes!

When this occurs, we can still receive from the other feed site without
any difficulties.  Deleting the queue file on the transmitting site
seems to solve the problem (probably by removing the offensive article
from the outgoing list).

This happened to us at the end of January from one feed-site.

Friday (3/8), it happened to us from a different feed-site, and later
with the same site that it happened in January.  Our connectivity with
these two sites has been awful since that day - usually resumed by
editing or deleting the queue.

I and my contacts at the feed-sites are stumped right now.  We're all
saying our software hasn't changed, etc.

Any suggestions ?
--
Doug Sewell, Tech Support, Computer Center,         doug@ysub.bitnet
Youngstown State University, Youngstown,  OH 44555  doug@ysub.ysu.edu
Here comes Margaret, Mr. Wilson.  Boy, can she be a pest!