Yike whats this [Undownloaded article 9563] 180 of them

Yike whats this [Undownloaded article 9563] 180 of them

[Harry Putnam]

I just saw something I never saw before in my comp.lang.perl.moderated
group. This group is under agent control, and I just cvs'ed yesterday.

I happened to open it with a C-u <RET> and I found 189 messages that
say:
[...]
 1@  31-Dec [   0: Gnus Agent          ] [Undownloaded article 9561]
 1@  31-Dec [   0: Gnus Agent          ] [Undownloaded article 9562]
 1@  31-Dec [   0: Gnus Agent          ] [Undownloaded article 9563]
[...]

Note the old date.  And I've never seen this notation in a summary
buffer before. Gnus Agent listing out undownloaded articles like that.

The very odd thing is that this group was begun fresh Jan 9 and has a
predicate of `true'.  There should be no undownloaded articles.
Further these file numbers do not in fact exist.  Also, they are not
known to the .overview file, ditto .agentview.  Nor do they exit in
.newsrc.eld.  Agent.lib/history doesn't know about them either.

However I did find a huge list in agent.lib/history about a different
group that looks like:

[...]
No-Message-ID-in-article	730859	alt.solaris.x86 59576 
No-Message-ID-in-article	730859	alt.solaris.x86 59577 
No-Message-ID-in-article	730859	alt.solaris.x86 59578 
[...]

Why are article numbers differernt than file names there? And is it
likely there really weren't message ids in hundreds of messages.

Far as I know, my server hasn't undergone any major changes.
(newsguy.com)

Anyone know what is happening here?

Re: Yike whats this [Undownloaded article 9563] 180 of them

[ShengHuo ZHU]

Harry Putnam <reader@newsguy.com> writes:

> I just saw something I never saw before in my comp.lang.perl.moderated
> group. This group is under agent control, and I just cvs'ed yesterday.
>
> I happened to open it with a C-u <RET> and I found 189 messages that
> say:
> [...]
>  1@  31-Dec [   0: Gnus Agent          ] [Undownloaded article 9561]
>  1@  31-Dec [   0: Gnus Agent          ] [Undownloaded article 9562]
>  1@  31-Dec [   0: Gnus Agent          ] [Undownloaded article 9563]
> [...]
>
> Note the old date.  And I've never seen this notation in a summary
> buffer before. Gnus Agent listing out undownloaded articles like that.

Gnus Agent knows nothing about these articles, but these articles
possibly exist on the server. Therefore, nnagent inserts such bogus
lines.  The date means nothing.

> The very odd thing is that this group was begun fresh Jan 9 and has a
> predicate of `true'.  There should be no undownloaded articles.
> Further these file numbers do not in fact exist.  Also, they are not
> known to the .overview file, ditto .agentview.  Nor do they exit in
> .newsrc.eld.  Agent.lib/history doesn't know about them either.

If you see those articles in plugged mode, but see such lines when you
reenter the group in unplugged mode, then it is a bug in Gnus Agent.

> However I did find a huge list in agent.lib/history about a different
> group that looks like:
>
> [...]
> No-Message-ID-in-article	730859	alt.solaris.x86 59576 
> No-Message-ID-in-article	730859	alt.solaris.x86 59577 
> No-Message-ID-in-article	730859	alt.solaris.x86 59578 
> [...]
>
> Why are article numbers differernt than file names there? And is it
> likely there really weren't message ids in hundreds of messages.

File names? The second numbers are article numbers, i.e. file
names. Could you check whether the articles really have no
Message-ID's?

ShengHuo

Re: Yike whats this [Undownloaded article 9563] 180 of them

[Harry Putnam]

ShengHuo ZHU <zsh@cs.rochester.edu> writes:

[...]

>> [...]
>> No-Message-ID-in-article	730859	alt.solaris.x86 59576 
>> No-Message-ID-in-article	730859	alt.solaris.x86 59577 
>> No-Message-ID-in-article	730859	alt.solaris.x86 59578 
>> [...]
>>
>> Why are article numbers differernt than file names there? And is it
>> likely there really weren't message ids in hundreds of messages.
>
> File names? The second numbers are article numbers, i.e. file
> names. Could you check whether the articles really have no
> Message-ID's?

What is the first number? 

Another post should appear momentarily detailing a little more. Namely
I indadvertantly ended up with the % mark on one of those marked:
   [Undownloaded article 9735]   
The agent dutifully downloaded it (it wasn't present before) now it is
present on disk but the summary buffer still says:
 [...]    [Undownloaded article 9735] 
I've posted some of the headers.

But it should be noted that gnus here hardly ever goes plugged as you
asked about.  I get everthing in batch mode. The running instance of
gnus rarely goes plugged... only for some test etc.  This setup has
been in existence for quite a long time and has worked across many cvs
upgrades ... probably for more than a year.  May this have some effect
in this instance?  Maybe due to recent changes in gnus-agent?

Since this occured I've marked them all for download ..now waiting for
my batch (cron job) to get them.

Re: Yike whats this [Undownloaded article 9563] 180 of them

[Harry Putnam]

Harry Putnam <reader@newsguy.com> writes:
[...]

Further notes:

> Since this occured I've marked them all for download ..now waiting for
> my batch (cron job) to get them.

[ -ed HP referrence `them' means the messages reported as marked:
 [Undownloaded article 9735]]

Agent has downloaded them all now .. They all appear in the buffer
exactly as they did before only now the download mark is still set:
 % [incorrect date]                     [Undownloaded article NUMBER] 

Still bearing the `%' mark although they have been downloaded and
marked read as before.... This can't be the desired behavior...

Re: Yike whats this [Undownloaded article 9563] 180 of them

[ShengHuo ZHU]

Harry Putnam <reader@newsguy.com> writes:

> ShengHuo ZHU <zsh@cs.rochester.edu> writes:
>
> [...]
>
>>> [...]
>>> No-Message-ID-in-article	730859	alt.solaris.x86 59576 
>>> No-Message-ID-in-article	730859	alt.solaris.x86 59577 
>>> No-Message-ID-in-article	730859	alt.solaris.x86 59578 
>>> [...]
>>>
>>> Why are article numbers differernt than file names there? And is it
>>> likely there really weren't message ids in hundreds of messages.
>>
>> File names? The second numbers are article numbers, i.e. file
>> names. Could you check whether the articles really have no
>> Message-ID's?
>
> What is the first number? 

The first number is the time when the article is downloaded.

> Another post should appear momentarily detailing a little more. Namely
> I indadvertantly ended up with the % mark on one of those marked:
>    [Undownloaded article 9735]   
> The agent dutifully downloaded it (it wasn't present before) now it is
> present on disk but the summary buffer still says:
>  [...]    [Undownloaded article 9735] 
> I've posted some of the headers.

If it is not included in the .overview file, the .overview file is
hosed.

> But it should be noted that gnus here hardly ever goes plugged as you
> asked about.  I get everthing in batch mode. The running instance of
> gnus rarely goes plugged... only for some test etc.  This setup has
> been in existence for quite a long time and has worked across many cvs
> upgrades ... probably for more than a year.  May this have some effect
> in this instance?  Maybe due to recent changes in gnus-agent?

The changes are adding such lines for articles with undownloaded
headers.  Before the changes, it was difficult to notice that some
articles were not downloaded, because those articles are not shown but
marked as read as if they didn't exist on the server.

ShengHuo

Re: Yike whats this [Undownloaded article 9563] 180 of them

[Harry Putnam]

ShengHuo ZHU <zsh@cs.rochester.edu> writes:

> File names? The second numbers are article numbers, i.e. file
> names. Could you check whether the articles really have no
> Message-ID's?

They are all zero length files.

Re: Yike whats this [Undownloaded article 9563] 180 of them

[Harry Putnam]

Harry Putnam <reader@newsguy.com> writes:

> [...]
>  1@  31-Dec [   0: Gnus Agent          ] [Undownloaded article 9561]
>  1@  31-Dec [   0: Gnus Agent          ] [Undownloaded article 9562]
>  1@  31-Dec [   0: Gnus Agent          ] [Undownloaded article 9563]
> [...]
>
> Note the old date.  And I've never seen this notation in a summary
> buffer before. Gnus Agent listing out undownloaded articles like that.
>
> The very odd thing is that this group was begun fresh Jan 9 and has a
> predicate of `true'.  There should be no undownloaded articles.
> Further these file numbers do not in fact exist.  Also, they are not
> known to the .overview file, ditto .agentview.  Nor do they exit in
> .newsrc.eld.  Agent.lib/history doesn't know about them either.

Gack .. another one just popped up:
I think I ended up with the % mark on one of those I mentioned and it got
downloaded now.  Now the summary buffer says:
 1R  31-Dec [   0: Gnus Agent          ] [Undownloaded article 9735]

But there is really a message there now with these (abbrieviated)
headers: 

From: e.mergl@bawue.de (Edmund Mergl)
Subject: Re: PL_perl_destruct_level = 1 -> segmentation fault
Newsgroups: comp.lang.perl.moderated
Date: 8 Jan 2002 08:26:07 -0800
Message-ID: <7f70c711.0201080826.8d26ccd@posting.google.com>
[...]
Xref: 	spln comp.lang.perl.moderated:9735

This seems like a pretty bad problem... I verified that the agent
settings are as reported.  In afact I have only one predicate.
Default is set to true.  GE settings look like:

("comp.lang.perl.moderated" 3
 ((1 . 9875))
 ((seen
   (9782 . 9872)
   (9874 . 9875))
  (download 9735)
  (dormant 9735 9826 9828 9831 9838
	   (9855 . 9856)
	   (9869 . 9872)
	   9874)
  (reply 9828 9872 9875)
  (tick 9875)))

Re: Yike whats this [Undownloaded article 9563] 180 of them

[ShengHuo ZHU]

Harry Putnam <reader@newsguy.com> writes:

> Harry Putnam <reader@newsguy.com> writes:
>
>> [...]
>>  1@  31-Dec [   0: Gnus Agent          ] [Undownloaded article 9561]
>>  1@  31-Dec [   0: Gnus Agent          ] [Undownloaded article 9562]
>>  1@  31-Dec [   0: Gnus Agent          ] [Undownloaded article 9563]
>> [...]
>>
>> Note the old date.  And I've never seen this notation in a summary
>> buffer before. Gnus Agent listing out undownloaded articles like that.
>>
>> The very odd thing is that this group was begun fresh Jan 9 and has a
>> predicate of `true'.  There should be no undownloaded articles.
>> Further these file numbers do not in fact exist.  Also, they are not
>> known to the .overview file, ditto .agentview.  Nor do they exit in
>> .newsrc.eld.  Agent.lib/history doesn't know about them either.
>
> Gack .. another one just popped up:
> I think I ended up with the % mark on one of those I mentioned and it got
> downloaded now.  Now the summary buffer says:
>  1R  31-Dec [   0: Gnus Agent          ] [Undownloaded article 9735]
>
> But there is really a message there now with these (abbrieviated)
> headers: 
>
> From: e.mergl@bawue.de (Edmund Mergl)
> Subject: Re: PL_perl_destruct_level = 1 -> segmentation fault
> Newsgroups: comp.lang.perl.moderated
> Date: 8 Jan 2002 08:26:07 -0800
> Message-ID: <7f70c711.0201080826.8d26ccd@posting.google.com>
> [...]
> Xref: 	spln comp.lang.perl.moderated:9735
>
> This seems like a pretty bad problem... I verified that the agent
> settings are as reported.  In afact I have only one predicate.
> Default is set to true.  GE settings look like:

I guess .agentview and/or .overview are hosed. Are there more than one
entries of 9735?  If so, remove redundant ones or use

   C-u M-x gnus-agent-regenerate RET

ShengHuo

Re: Yike whats this [Undownloaded article 9563] 180 of them

[Henrik Enberg]

Harry Putnam <reader@newsguy.com> writes:

> I just saw something I never saw before in my comp.lang.perl.moderated
> group. This group is under agent control, and I just cvs'ed yesterday.

[...]

>  1@  31-Dec [   0: Gnus Agent          ] [Undownloaded article 9561]
>  1@  31-Dec [   0: Gnus Agent          ] [Undownloaded article 9562]
>  1@  31-Dec [   0: Gnus Agent          ] [Undownloaded article 9563]

I get this too (with date 1070-01-01).  As far as I can tell articles
that turn up like this are the ones that expired from my agent, but are
still left on my newsserver.

Henrik
-- 
Will the highways on the Internet become more few?
		-- George W. Bush