Re: [leafnode-list] Can't get leafnode to work with news servers that work with clients.

[Matthias Andree <matthias.andree@xxxxxx>]

Adam Funk schrieb am 2005-06-07:

> On a Debian testing system I recently installed the leafnode
> 1.11.2.rel-1 package.  I have been trying to get it to work with my
> ISP's two news servers, text.news.blueyonder.co.uk (no binaries) and
> news.blueyonder.co.uk, both of which work fine with the KNode and Pan
> news clients.
> 
> I seem to have two problems.
> 
> 1. The servers time out while leafnode is "considering" articles to
>    download.  For example:

Please see TROUBLESHOOTING in the README file for instructions on how to
get proper debugging output, which should be obtained from syslog (you
may need to edit syslog.conf and reload syslogd first!) rather than the
screen:
I need a few lines of commands and replies before the "Server
disconnection..." line, you may need to use "debugmode=2" rather than
"debugmode=1".

If and only if you have configured a filterfile= or other filters, you
can try to set "article_despite_filter=1" in your config file and see if
that helps. IF it helps, blueyonder's servers are royally broken.

> text.news.blueyonder.co.uk: reading server info
> from /var/spool/news/leaf.node/text.news.blueyonder.co.uk
> Merging in /var/spool/news/leaf.node/text.news.blueyonder.co.uk~ from
> previous run
> alt.religion.kibology: considering articles 48065 - 48078
> alt.religion.kibology: 13 articles fetched, 1 killed
> news.answers: no new articles
> news.announce.newusers: no new articles
> news.newusers.questions: no new articles
> news.groups.questions: no new articles
> news.software.readers: considering articles 21161 - 21162
> news.software.readers: 2 articles fetched, 0 killed
> alt.test: considering articles 267867 - 268263

Leafnode attempts to fetch an article ("ARTICLE") or article header
("HEAD") here and the server goes away before sending a reply code.
The "considering" phase is complete, although this is not logged
separately. I can tell this from the message "Server disconnection or
timeout before article could be retrieved #1" though.

> 2. The processes started by the hourly fetchnews cron job stay alive
>    and hold the lock file for a *long* time and prevent subsequent
>    jobs from starting.  They don't appear to do anything according to
>    `top` and find `/var/spool/news -mmin -5` (for example).

Please get me a backtrace of the hanging process, 9489 in this case. The
FAQ that should have shipped as /usr/share/doc/leafnode/FAQ.txt.gz has
detailed information on how to do that.

If the backtrace contains hex addresses and no line numbers, try instead
to trace the server conversation with ethereal and send me the last
few lines of the reassembled TCP stream.

> Here's the customized part of /etc/news/leafnode/config:
> 
> server = text.news.blueyonder.co.uk
> expire = 25
> timeout = 25
> initialfetch = 200
> noxover = 1
> 
> server = news.blueyonder.co.uk
> expire = 25
> timeout = 25
> initialfetch = 200
> noxover = 1
> 
> debugmode = 1
> timeout_fetchnews = 60

Looks sane. Make sure that syslog.conf is capturing news.debug output
and check the log file, it has more information than the screen print.

HTH,

-- 
Matthias Andree
-- 
_______________________________________________
leafnode-list mailing list
leafnode-list@xxxxxxxxxxxxxxxxxxxxxxxxxxxx
http://www.dt.e-technik.uni-dortmund.de/mailman/listinfo/leafnode-list
http://leafnode.sourceforge.net/