innd 2.3 hangon after an news.daily

[ab at icsmedia.de]

> >Apr 17 11:14:19 papenburg-lnx ovdb_recover[408]: OVDB: log_get: read: Input/output error
> >
> 
> Those "Input/output error" messages (errno == EIO) probably mean you've
> got bad blocks on your disk; or some other hardware trouble.
>
Nice to hear, but a program haven't to core.
The ovdb_recover run into a segmention fault, and the innd into an endless loop.
That is the Bug. I have on that machine 128M Ram, 1G HD free,
559 newsgroups and a history with 200M. And no msgs from the kernel about bad-blocks or
less memory.

And :-) another Bug. If I call a line like that:
cat text_file_with_new_groups | xargs -i{} ../bin/ctlinnd newgroup {}
wc  text_file_with_new_groups => 1000

After a few groups I got this result:
---
Apr 18 10:33:15 papenburg-lnx overchan[30187]: OVDB: ov00021: Too many open files in system
Apr 18 10:33:15 papenburg-lnx overchan[30187]: OVDB: open_db_file failed: Too many open files in syste
---

With two eventualities, the ctlinnd can't found the libc or the innd dies without
a message.


Axel