Core (innd) with 20000627

Bettina Fink laura at hydrophil.de
Fri Jun 30 14:29:37 UTC 2000


On Fri, 30 Jun 2000, Heath Kehoe wrote:

> One thing you could try (if you havn't already) is:
>  1) kill all of the news processes
>  2) delete the __db.* files in /news/spool/overview
>  3) run "ovdb_recover -f" (as the news user)
>  4) rc.news

Did that and the overview looks ok for now. But that didn't
solve the problem with the articles that got accepted twice ...
I think the complete history, spool and overview is mangled.

BTW, I saw that several times in the past even when everything
was ok: After a "rc.news stop", there are processes that won't
terminate.

11457 pts/2    S      0:00 -bash
13600 pts/2    S      0:00 tail -f log/news.notice
11428 pts/1    S      0:00 -bash
13695 pts/1    R      0:00 ps xf
24280 ?        S      3:21 db_checkpoint -p 1 -k 2000

This time it was only one (db_checkpoint), but I've seen more
and different processes. Seen with db-3.0.55 as well as with
db-3.1.14.

> This will only buy you some time until it gets corrupted again.
> I recommend that you delete your overview and downgrade to
> berkeleydb-2.7.7.

I'll do that.

Bye,
Bettina




More information about the inn-workers mailing list