problems with news.individual.net (seen with leafnode)

[Robert Marshall]

I use news.individual.net as a usenet provider (well one of several!)
and have leafnode installed as a local cache. Recently I've been seeing
old articles being pulled across as new. I think this has happened
around 3 times since the start of Jan.

I've just looked at the log files when the problem occurs and I'm seeing

gnu.emacs.bug: considering 20380 articles 89294 - 109673, using XOVER
gnu.emacs.bug: XOVER: 20376 seen, 2544 I have, 17832 filtered, 0 to get
gnu.emacs.bug: all articles already there
...
gnu.emacs.help: considering 9351 articles 200910 - 210260, using XOVER
gnu.emacs.help: XOVER: 9343 seen, 679 I have, 8589 filtered, 75 to get
gnu.emacs.help: will fetch 75 articles
gnu.emacs.help: 75 articles fetched (to 189357), 0 killed
news.individual.net: gnu.emacs.sources last seen article was 200910, server now has 14946 - 15248
news.individual.net: switched upstream servers for gnu.emacs.sources? upstream bug? 200910 > 15248
gnu.emacs.sources: considering 303 articles 14946 - 15248, using XOVER
gnu.emacs.sources: XOVER: 226 seen, 17 I have, 209 filtered, 0 to get
gnu.emacs.sources: all articles already there
news.individual.net: gnu.emacs.vm.bug last seen article was 14946, server now has 10752 - 10760
news.individual.net: switched upstream servers for gnu.emacs.vm.bug? upstream bug? 14946 > 10760
gnu.emacs.vm.bug: considering 9 articles 10752 - 10760, using XOVER
gnu.emacs.vm.bug: XOVER: 9 seen, 0 I have, 9 filtered, 0 to get
...
rec.music.classical.recordings: considering 41488 articles 1238905 - 1280392, using XOVER
rec.music.classical.recordings: XOVER: 41092 seen, 3336 I have, 37233 filtered, 523 to get
rec.music.classical.recordings: will fetch 523 articles
rec.music.classical.recordings: 523 articles fetched (to 805155), 0 killed

it looks as if some radical article renumbering is happening at the server end? But
not in all the above. In the fetchnews before this one I get:

timeout reading.
news.individual.net NNTP server disconnected or timed out while waiting for response
gnu.emacs.bug: considering 5 articles 230 - 234, using XOVER
Unknown reply to XOVER command: 211 20380 89294 109673 gnu.emacs.bug
news.individual.net: cannot parse reply to "GROUP gnu.emacs.gnus": "420 No such article"

It's interesting that the first newsgroup where the problem occurs is
the same one where the timeout occurred in the earlier run. Maybe
leafnode isn't recovering properly from the connect error? Timeouts also
appear to happen frequently (if as with rmc.recordings it's searching
through 40000 articles and retrieving the newer ones maybe that's not
surprising!).

Is anyone else seeing these problems? - I've looked at the n.i.n website
and can't see anywhere there where they report server problems.

In leafnode I had a filter set so that articles older than 50 days
weren't retrieved, I've just changed it to 20 to see if that masks the
problem.

Robert
--
La grenouille songe..dans son château d'eau
Links and things http://rmstar.blogspot.com/