Newgroup hashtable (NGH)

Russ Allbery rra at stanford.edu
Mon Mar 5 02:29:47 UTC 2001


Frans Dogge <fransd at worldonline.nl> writes:

> I ran into a problem yesterday, that flabbergasted me completely. The
> expireover cored, and innd just stopped. There where no error messages
> anywhere, innd just stopped, and would not start again. With the help of
> a programmer we where able to debug the core file, and found that the
> problem was with our active file. We have been using this file for about
> 4 years now, so it's big. It is so big that the Newgrouphashtable (NGH)
> which has a default value of 2048, was unable to contain it. We
> increased the size to 4096, and recompiled inn, and the problem was
> solved.

That's very odd... while the hash table currently used by innd/ng.c is
definitely too small for most servers, it uses chaining (after a fashion)
for hash collisions and therefore should handle an active file of
arbitrary size (just not very efficiently).  I don't see how that could
cause innd to segfault or why increasing the size would help this.

Do you still have the backtrace from the core file?

-- 
Russ Allbery (rra at stanford.edu)             <http://www.eyrie.org/~eagle/>


More information about the inn-workers mailing list