[leafnode-list] Re: possible bug in leafnode-2.0.0.alpha?

[Theodore Heise]

On Mon, 24 Oct 2005, Matthias Andree wrote:
> Theodore Heise <theo at heise.nu> writes:
>
>> I'm running leafnode-2.0.0.alpha20050810a in Slackware 9.1...
>> Yesterday, I read a post in alt.cellular.verizon that had added a 
>> crossposting to comp.dcom.xdls with follow-up set to the latter 
>> group.  I wrote a follow-up to that post, allowing slrn to honor 
>> the follow-up setting and post only to comp.dcom.xdsl:
>>
>>   2)	Message-ID: <slrndjmnkh.1eq.theo at linus.heise.nu>
>>
>> ...comp.dcom.xdsl was not in my interesting or subscribed groups, 
>> so I then added it to my slrn groups, and opened the leafnode 
>> placeholder.  My follow-up has now appeared on other servers, and 
>> I have posted a follow-up to it from a different account:
>>
>>   3)	Message-ID: <slrndjo94c.eon.theise at worf.netins.net>
>>
>> However, my first post (2) does not appear on my own news spool.

> Sorry I couldn't get to this earlier. If your expiry time is lower 
> than 30 days, just forget the whole story because we cannot debug 
> any more. (Even if it's shorter we may be unable to debug).

Not at all a problem, its just cool you're so consistent at 
supporting this great package.  Here's the relevant bit from my 
config:

[/etc/leafnode/config]
...
## Unread discussion threads will be deleted after this many days if
## you don't define special expire times. Mandatory.
expire = 60
...


> Can you find the message.id/???/* file with the Message-ID and 
> show me what "ls -l" gives you for that file?

root at linus:~# ls -l /var/spool/news/message.id/<slrndjmnkh.1eq.theo at linus.heise.nu>
-su: syntax error near unexpected token `newline'


> Can you grep your logs for the Message-ID of (2)?

I grepped the spool and found it (see following), and used that as 
input for the "ls -l" above.  My logs from early October have all 
fallen off the end of what logrotate keeps, but I could probably dig 
them out from backups if this would be helpful.

root at linus:~# grep -r slrndjmnkh.1eq.theo at linus.heise.nu 
/var/spool/news/message.id/* /var/spool/news/message.id/521/<slrndjo94c.eon.theise at worf.netins.net>:References: <1127679591.892165.259860 at g49g2000cwa.googlegroups.com> <11jegi4k52qf53a at corp.supernews.com> <dh7s7k$beq$1 at ratbert.glorb.com> <slrndjmnkh.1eq.theo at linus.heise.nu>


> Perhaps this works too:
>
> find /var/spool/news -name '*slrndjmnkh.1eq.theo*' -ls
>
> If it shows some output, I'd like to know which. If it doesn't show
> output, I'm at a loss for now and I'll have to reproduce this.

Nothing is returned for this find command.


> My current suspicion is that the message-ID file from the post is 
> still there, but the article isn't copied into the group directory 
> as the group isn't subscribed or existent. The Message-ID file 
> however blocks the download.

I'm not familiar with the detailed workings of leafnode/fetchnews, 
but this seems reasonable.


> Suggestions you can try in random order, let me know which work 
> and which don't:
>
> 1. run "fetchnews -x 100" to try to fetch the previous 100 articles too.

root at linus:~# fetchnews -x 100 -N comp.dcom.xdsl

This doesn't seem to make any difference, the article is still 
reported by slrn as not available.  As I think about it, wouldn't 
this only have the possibility of working if my upstream server 
retains articles since before this one was posted?  I don't think 
this is so in this case.


> 2. run "texpire after the post" and subscribe to the group 
> *before* running fetchnews for the next time

I'm not sure I understand this exactly, but is this a suggested 
workaround for dealing with the situation the next time it happens? 
If it's something I should try for the case I originally described, 
I'm afraid I don't understand it.


> I wonder if a locally originated post should automatically (1) 
> create or (2) create and subscribe to an unsubscribed remote 
> group. Suggestions, with rationale?

Seems to me that (2) would be preferable.  Anyone posting to a new 
group has already implied they have an interest in that group.  If 
the new group is indicated by a specific follow-up header, anyone 
using a decent newsreader should see this when they post and be able 
to make a decision at that time whether or not they are interested 
in that group.

Matthias, thanks for looking into this.  Let me know what else I can 
do to help dig into it, if anything.

Ted

-- 
Theodore (Ted) Heise     <theo at heise.nu>     Bloomington, IN, USA