Severe xover corruption problems on INN2.2.1

[Hans Lambermont]

Katsuhiro Kondou wrote:

> In article <199909301340.PAA25240 at garg.mpn.cp.philips.com>,
> 	Hans.Lambermont at nl.origin-it.com (Hans Lambermont) wrote;
> } We have severe xover corruption problems on INN2.2.1 production systems.
> 
> As for corrupted result of xover, there can be two reason.
> One is corrupted .overview.index in the group and another
> is corrupted offset of overview data in token.  How about
> the result of 'sm -o token'?  If it seems to be ok, then
> .overview.index is broken.

Thanks for your time again; here's some of my digging:

news.software.nntp: article 84470

xover 84470
84470   e.be>           1130    10      Xref: news.eur.cis.philips.com nl.comp.games:52211 save game mission 4 van nod!!!  "mathieu" <china.nature at online.be> Tue, 5 Oct 1999 16:25:17 +0200  <7td10c$7ng$1 at trex.antw.online.be> 734    6Xref: news.eur.cis.philips.com nl.comp.games:52212 Re: Staatsschuld 550 miljard versneld

head 84470
221 84470 <yl3dvrdpgg.fsf at windlord.stanford.edu> head
Path: news.eur.cis.philips.com!newsfeed2.news.nl.uu.net!sun4nl!newsfeed.berkeley
.edu!newsfeed.stanford.edu!nntp.stanford.edu!not-for-mail
From: Russ Allbery <rra at stanford.edu>
Newsgroups: news.software.nntp
Subject: Re: INN maxed? Time to add a Diablo router?
Date: 04 Oct 1999 10:07:43 -0700
Organization: The Eyrie
Lines: 14
Message-ID: <yl3dvrdpgg.fsf at windlord.stanford.edu>
References: <foo-0210991412200001 at 192.168.2.3> <yl3dvtp0xb.fsf at windlord.stanford .edu> <foo-0210992103530001 at 192.168.2.3> <yl3dvtnjuv.fsf at windlord.stanford.edu> <brad-0310991300490001 at dialup995.brussels2.skynet.be> <slrn7veimb.irn.niels at liquid.tpb.net> <brad-0410991311360001 at brad.techos.skynet.be>
NNTP-Posting-Host: windlord.stanford.edu
X-Newsreader: Gnus v5.4.66/Emacs 19.34
Xref: news.eur.cis.philips.com news.software.nntp:84470

grephistory 'yl3dvrdpgg.fsf at windlord.stanford.edu'
@0302523230390000000000022D1B000000140D00B3D79800EB010000@

sm -o @0302523230390000000000022D1B000000140D00B3D79800EB010000@ 


this sm -o prints 2 blank lines. That's not good. So we experience
corrupted offset of the overview data in this token ? How should we fix
this ? (and prevent it from happening again ;)

Any help appreciated.

regards,
   Hans Lambermont