[leafnode-list] Re: Leafnode not getting new groups

[Antoine Levitt <smeuuh@xxxxxxx>]

27/05/11 12:33, Matthias Andree
> Am 27.05.2011 01:32, schrieb Antoine Levitt:
>> 27/05/11 01:17, Matthias Andree
>
>>> - what file system is used for your /var/spool/news?
>>
>> Ext4, it's not a weird NTFS or anything.
>
> Right. mtime should work on the native Linux and BSD file systems.
>
>>> - is there anything that modifies, touches, or otherwise change the
>>> news.gwene.org "mtime" stat() attribute?  Leafnode touches the file to
>>> record what time it had last used for "NEWGROUPS", and if the mtime gets
>>> bumped (increased) behind leafnode's back, that's why it fails.
>>
>> I don't think so. Only weird thing I can think of would be a suspend
>> right in the middle of a leafnode check. Oh, and we got a time change
>> march 27 over here, and it's not entirely impossible that troubles began
>> around that date.
>
> Daylight Savings Time changes shouldn't matter if handled properly,
> because the GMT/UTC that is used to check for groups is continous and
> doesn't leap ahead one hour.
>
> The code also doesn't look like it were sensitive to suspends that cause
> connection timeouts on the server.  The mtime is bumped (like with the
> "touch" command) only after leafnode has downloaded all new groups and
> descriptions from the server.
>
> But I see that there are bugs in the error handling, depending on what
> leafnode has last tried it doesn't detect the connection loss and still
> touches the time, so the suspend may have caused it.  Good catch I'd
> say. :-)
>
>>> fetchnews -nvveD131 -S news.gwene.org | grep '^>NEWGROUPS'
>>
>> antoine@lambda:~$ sudo fetchnews -nvveD131 -S news.gwene.org 2>&1 | grep
-C4
>> '^>NEWGROUPS'
>>> DATE
>> <111 20110526232643
>> check_date: news.gwene.org: server time 1306452403, our time 1306452403
>> news.gwene.org: checking for new newsgroups
>>> NEWGROUPS 110526 232628 GMT
>> <231 New newsgroups follow
>> <.
>> news.gwene.org: found 0 new newsgroups
>> news.gwene.org: feedtype == NNTP.
>>
>> Not sure how to understand this ...
>
> Well, the first items pertain to checking if client (leafnode) and
> server (INN) time are in synch (they are in your case), and to warn the
> user if they are too far apart.
>
> The second item requests the "groups added since 2011 May 26 23:26:28
> UTC" -- this should be the time of the last successful fetchnews
> connection to news.gmane.org.

Ah right, I thought those were deltas of unix timestamps or something,
didn't occur to me they might just be dates represented this way :)

>
>>> After you've looked at all that, the recovery is to either simply run
>>> "fetchnews -f" to fetch all group lists from all servers, or,
>>> lighter-weight, to remove the last:news.gwene.org* files.  Either voids
>>> some way to understand what has caused this.
>>
>> I'll wait before doing this then.
>
> It's now safe to run. I know where to look now, and what to look for.
> Thanks for the report. :)

Cool, thanks for the help!
-- 
_______________________________________________
leafnode-list mailing list
leafnode-list@xxxxxxxxxxxxxxxxxxxxxxxxxxxx
https://www.dt.e-technik.uni-dortmund.de/mailman/listinfo/leafnode-list
http://leafnode.sourceforge.net/