[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[leafnode-list] Re: placeholder articles after running applyfilter
On Thu, Oct 09 2003, Matthias Andree wrote:
> Reiner Steib writes:
>> BTW, I got lots of "Leafnode placeholder" articles in Gnus. Those
>> groups are not new, but they didn't contain any articles at that time.
Sorry for not including anything useful debug info upto now. I hope
this one contains enough info. I didn't change debugmode (is 153 now)
in leafnode yet. Please tell me if I should do this (which items?)
for a further analysis.
> Is your leafnode agent-ized in Gnus?
No (gnus-agent's value is nil globally).
> What does ls -la on the group folder (/var/spool/news/gmane/blah/mumble)
> print?
/var/spool/news/gmane/science/physics/whatsnew$ ll -a
total 8
drwxr-xr-x 3 [...] 4096 Sep 7 19:17 ..
-rw-r--r-- 1 [...] 0 Oct 3 23:10 .last.posting
drwxr-xr-x 2 [...] 4096 Oct 8 15:49 .
(JFTR: this was before applying the 2003-10-09 patch.)
> When has texpire been run for the last time?
Oct 7 03:05 (after that, I switched the cron job off in order to
better be able to reproduce the other bug I reported.)
> Is the group listed in /var/spool/news/interesting.groups?
Yes.
/var/spool/news/interesting.groups$ ll *whats*
-rw-rw-r-- 1 [...] 0 Sep 7 18:41 gmane.science.physics.whatsnew
> Are you running with delaybody?
No.
> Please try this patch. If the problem persists, please show an NNTP
> trace (might be a Gnus bug, its XOVER handling doesn't seem too robust).
>
> One more question: are you running Gnus with suppression of duplicate
> articles enabled?
Yes.
> I do, so I may not see the "lots of" -- all those place holders have
> a constant (though group-specific) Message-ID.
By "lots of" I meant "in lots of different groups". I'm subscribed to
many announcement and other low-traffic groups and my expiry time is
quite short (expire = 4) compared to the posting frequency in these
groups.
> The last hunk of the patch is the most important one, without it, you
> cannot request a pseudo article with XOVER n, for instance. Makes me
> wonder where the others come from though...
I upgraded to today's (20031009a) snapshot, which AFAICS already
includes this patch. I watched the groups *texinfo*, which have no
articles in leafnode's spool.
*Summary*: I need to run applyfilter and fetchnews to reproduce the
new placeholder articles now. It seem that applyfilter messes up the
article counts and fetchnews tries to correct this, leading to an
increase of the article counts.
Details follow...
Below, I ran the function _list_it_ before/after each step. This
function does the following:
_list_it_()
{
date
cd /var/spool/news/gmane/comp/tex/texinfo
find . -ls
grep -e texinfo /var/spool/news/leaf.node/groupinfo
cd /var/spool/news/interesting.groups/
ls -l *texinfo*
}
* Before running applyfilter:
$ _list_it_
Thu Oct 9 13:35:28 CEST 2003
[...]drwxr-xr-x 4 [...] 4096 May 4 20:49 .
[...]drwxr-xr-x 2 [...] 4096 Oct 2 22:21 ./general
[...]-rw-r--r-- 1 [...] 0 Sep 29 11:15 ./general/.last.posting
[...]drwxr-xr-x 2 [...] 4096 Oct 3 03:05 ./pretest
[...]-rw-r--r-- 1 [...] 0 Sep 29 11:15 ./pretest/.last.posting
gmane.comp.tex.texinfo.bugs m 1 1 1037977483
gmane.comp.tex.texinfo.general m 74 74 1037977483
gmane.comp.tex.texinfo.pretest m 39 39 1052071490
-rw-rw-r-- 1 [...] 0 Apr 7 2003 gmane.comp.tex.texinfo.general
-rw-rw-r-- 1 [...] 0 May 4 20:33 gmane.comp.tex.texinfo.pretest
* Then I ran applyfilter for all texinfo groups: "cd
interesting.groups; for i in gmane.comp.tex.texinfo.*;do echo
$i;applyfilter $i;done
$ _list_it_
Thu Oct 9 13:36:10 CEST 2003
[...]drwxr-xr-x 4 [...] 4096 May 4 20:49 .
[...]drwxr-xr-x 2 [...] 4096 Oct 2 22:21 ./general
[...]-rw-r--r-- 1 [...] 0 Sep 29 11:15 ./general/.last.posting
[...]drwxr-xr-x 2 [...] 4096 Oct 3 03:05 ./pretest
[...]-rw-r--r-- 1 [...] 0 Sep 29 11:15 ./pretest/.last.posting
gmane.comp.tex.texinfo.bugs m 1 1 1037977483
gmane.comp.tex.texinfo.general m 74 75 1037977483
gmane.comp.tex.texinfo.pretest m 39 40 1052071490
-rw-rw-r-- 1 [...] 0 Apr 7 2003 gmane.comp.tex.texinfo.general
-rw-rw-r-- 1 [...] 0 May 4 20:33 gmane.comp.tex.texinfo.pretest
Does the part "74 75" or "39 40" make sense? Shouldn't it be "high
low"?
$ echo 'LIST ACTIVE gmane.comp.tex.texinfo.general'|nc -w1 localhost nntp
200 Leafnode NNTP daemon, version 2.0.0.alpha20031009a at lumberjack.physik.uni-ulm.de
215 Newsgroups in form "group high low flags".
gmane.comp.tex.texinfo.general 0000000074 0000000075 m
high < low? Strange.
* [time passes...] Cron ran fetchnews in the meantime. Now I get the
placeholder articles again!
$ _list_it_
Thu Oct 9 15:47:42 CEST 2003
[...]drwxr-xr-x 4 [...] 4096 May 4 20:49 .
[...]drwxr-xr-x 2 [...] 4096 Oct 2 22:21 ./general
[...]-rw-r--r-- 1 [...] 0 Sep 29 11:15 ./general/.last.posting
[...]drwxr-xr-x 2 [...] 4096 Oct 3 03:05 ./pretest
[...]-rw-r--r-- 1 [...] 0 Sep 29 11:15 ./pretest/.last.posting
gmane.comp.tex.texinfo.bugs m 1 1 1037977483
gmane.comp.tex.texinfo.general m 75 75 1037977483
gmane.comp.tex.texinfo.pretest m 40 40 1052071490
-rw-rw-r-- 1 [...] 0 Apr 7 2003 gmane.comp.tex.texinfo.general
-rw-rw-r-- 1 [...] 0 May 4 20:33 gmane.comp.tex.texinfo.pretest
fetchnews seems to try to correct the wrong counts leading to an
increase of the article count, i.e. a placeholder article is displayed
as the client (Gnus) sends...
,----[ Hitting `g' in Gnus (`gnus-group-get-new-news') ]
| 20031009T154601.476 localhost MODE READER
| 20031009T154601.499 localhost LIST ACTIVE de.sci.physik
| [...]
| 20031009T154601.502 localhost LIST ACTIVE gmane.comp.tex.texinfo.pretest
| 20031009T154601.502 localhost LIST ACTIVE gmane.comp.tex.texinfo.general
`----
,----[ Hitting `RET' in Gnus (`gnus-topic-select-group') ]
| 20031009T160227.652 localhost GROUP gmane.comp.tex.texinfo.general
| 20031009T160227.655 localhost HEAD 75
`----
Output taken from the *nntp-log* buffer in Gnus (setq
nntp-record-commands t).
Bye, Reiner.
--
,,,
(o o)
---ooO-(_)-Ooo--- PGP key available via WWW http://rsteib.home.pages.de/
--
leafnode-list@xxxxxxxxxxxxxxxxxxxxxxxxxxxx -- mailing list for leafnode
To unsubscribe, send mail with "unsubscribe" in the subject to the list