[news.software.b] B news duplicate articles

cathyf@Rice.EDU (Catherine A. Foulston) (08/03/89)

Help!  As many people have observed in news.admin, a bunch of old
articles got dumped out on the net.  I don't understand why the
history mechanism is not causing these articles to be rejected.

o I am running B News 2.11 patchlevel 17 and nntp 1.5.5.
o I am running expire nightly and rebuilding once a week.  Articles
  are supposed to stay around for 2 weeks or less depending on the
  newsgroup.  History is supposed to be retained for 30 days for
  ALL articles.  At least this is what I think I am doing with expire.
  Maybe I don't understand it.  What options am I supposed to give
  expire to make it do this?
o I recently converted from the old usg-style history to using ndbm.
  At that time I *completely* rebuilt the history file from scratch.

o There are two copies of most of the duplicates in the directory.
  Doing a diff on them shows that only the path is different.
o Only the LATER of the two copies appears in the history file.

o Most recently the problem has been with the large batch of old news
  that everybody got that was dated July 22.  But I have had this
  problem before (even before I converted to ndbm).

Maybe I just don't understand how expire or history are supposed to
work.  If anyone can explain this to me or suggest possible sources
of this problem, I would appreciate it greatly.  Many thanks in
advance.  (P.S:  I already RTFM--I guess I don't understand it.)

regards,

cathy :)     As usual, email replies are probably better than posts
Cathy Foulston   =||=   cathyf@rice.edu   =||=   {backbone?}!rice!cathyf
Disclaimer:  I'm a student--I don't even KNOW Rice University's opinion.

tale@pawl.rpi.edu (David C Lawrence) (08/03/89)

In <413@brazos.Rice.edu> cathyf@Rice.EDU (Catherine A. Foulston) writes:
CAF> o Most recently the problem has been with the large batch of old news
CAF>   that everybody got that was dated July 22.

Don't let the 22 July date fool you.  As many of the message-ids
reveal (as do the contents of some mor topical articles) the messages
are really from June.

Dave
--
 (setq mail '("tale@pawl.rpi.edu" "tale@itsgw.rpi.edu" "tale@rpitsmts.bitnet"))

tneff@bfmny0.UUCP (Tom Neff) (08/04/89)

I think the articles are really from well before Jul 22.  So although
this date may fall within your "history.*" range, the articles themselves
won't have matching IDs to spot and force a silent cancel.
-- 
"We walked on the moon --	((	Tom Neff
	you be polite"		 )) 	tneff@bfmny0.UU.NET

allbery@nc386.UUCP (Brandon S. Allbery) (08/08/89)

Uh, if I remember correctly (ha!) the last time this happened some site had
inserted tabs into the Message-IDs.  (Of course, that bug was supposed to have
been fixed....)

Has anyone looked to see what the message-IDs have in common, if anything?

++Brandon
-- 
Brandon S. Allbery, moderator of comp.sources.misc	     allbery@NCoast.ORG
uunet!hal.cwru.edu!ncoast!allbery		    ncoast!allbery@hal.cwru.edu
"Why do trans-atlantic transfers take so long?"
"Electrons don't swim very fast."  -john@minster.york.ac.uk and whh@PacBell.COM