Ross Boswell writes:
> Hi
>
> Thanks for your super work on nntpcache. I think it IS the
> best thing since sliced blue vein (cheese).
>
> I am running version 0.92.2UL on a Linux box serving tin 1.3, and
> I have discovered two problems --
>
> 1) Selecting a group frequently causes tin to freeze indefinitely,
> displaying (for example):
>
> Group comp.os.linux.hardware ('q' to quit)...
>
> Sending a -HUPCL to the nntpcache child process converts the tin display to:
>
> Connection to news server has timed out. Reconnect? (y/n): y
>
> and responding "yes" usually proceeds fine for that group, but the next
> group selection is likely to give the same problem.
>
> 2) The first selection of an article usually displays garbage text (looks
> like a buffer not cleared from the previous article displayed).
> Deselecting and reselecting fixes the display.
>
> These may be known bugs -- I appreciate I AM at the bleeding edge.
I'm seeing this as well as 1 other bug :
(BTW i'm using linux 2.0.0 on this machine)
I had 2 newsfeeds, from here, in France. All was fine (except the 2
previous bugs).
I added a third, slow newsfeed from USA. (well slow from
here). Sometime later, nntcached died, saying
Aug 28 15:16:49 popeye /usr/local/sbin/nntpcached[230]:
sockets.c:133:Bad file number: fd too high or low (-1 vs 266)
I think this has to do with fd limits (256 ?) in Linux ...
I removed the newsfeed from USA, and all is fine. Is there a way to
limit the fd limit for nntpcache ?
Yann.
--
_ _ ___ _________________________________
| | || . \ | Fac des sciences de Nantes. |
\ /| | | _ | Tel : 40373069 / 40374945 |
|_| |___/<_> | ydupont@sciences.univ-nantes.fr |
---------------------------------
