Corrupted expireover database

George Lindholm George.Lindholm at ubc.ca
Tue Aug 31 05:15:47 UTC 1999


Well, it happened to me again. I rebuilt the overview data from
scratch on the weekend, and by the Monday morning purge I had a
corrupted group entry, with the accompanying core file from expireover.
The machine is now my production news server (may Odin have mercy on
my soul -:) so I can't experiment as I used to in order to track down
the problem. Any ideas?

   George

> -----Original Message-----
> From: inn-workers-bounce at isc.org [mailto:inn-workers-bounce at isc.org]On
> Behalf Of Katsuhiro Kondou
> Sent: Tuesday, August 24, 1999 19:30
> To: inn-workers at isc.org
> Subject: Re: Overview space needs?
>
>
> In article <37C3253E.331A370E at ubc.ca>,
> 	George Lindholm <George.Lindholm at ubc.ca> wrote;
>
> } Here are the active entries and the GROUPentry record for the groups
> } (what's the base field?) in case that sheds some light on the problem:
>
> Base is the available lowest number for that index.
> This is useful if adding overview data is for under
> low number.  It should be less than low.
>
> } alt.spam 0000048928 0000047516 y
> } (gdb) p *ge
> } $1 = {hash = {hash = "Ò¦übê)\024KÐ\223\002>¡~\237Ü"}, alias = {
> }     hash = '\000' <repeats 15 times>}, high = 48928, low = 47516,
> }   base = 48073, count = 499, flag = 121, deleted = 0, indexinode =
> } 397210,
> }   next = {recno = -1}}
>
> Can you put trace lines like:
>
> if (ge->low > ge->base) {syslog(L_ERROR, "tradindexed: %d
> ge->low(%d) > ge->base(%d)", __LINE__, ge->low, ge->base); abort();}
>
> That line can be inserted where ge->low or ge->base
> is changed.  And you'll see how that problem comes.
> --
> Katsuhiro Kondou
>
>



More information about the inn-workers mailing list