Re: [leafnode-list] Backtrace of hung fetchnews

[Pierre Abbat <phma@xxxxxxxxxxxxx>]

On Friday 08 February 2002 20:52, Matthias Andree wrote:
> Pierre Abbat <phma@xxxxxxxxxxxxx> writes:
> > This is fetchnews 1.9.18, with the patch I posted previously. fetchnews
> > connected to cserver.vadim.com.pl, neither posted nor fetched anything,
> > connected to news.mhogaming.com, and hung.
> >
> > #7  0x8052e56 in getaline (f=0x83b6ed0) at miscutil.c:698
> > #8  0x804e85e in nntpreply () at nntputil.c:138
> > #9  0x804ec0c in nntpconnect (upstream=0x8061b90) at nntputil.c:238
> > #10 0x804e41a in main (argc=1, argv=0xbffffcec) at ./fetchnews.c:1612
>
> Looks like either the upstream server did not answer after accepting the
> connection (may be caused by full or defective disk drives on their
> site) or it disconnected but leafnode didn't notice.

Figured out one more clue and a workaround. nntpconnect() sets an alarm, 
which apparently somehow got lost. I therefore killed fetchnews -ALRM and it 
resumed connecting to the next news server. The workaround is to killall 
-ALRM fetchnews before starting fetchnews in my cronjob.

phma

-- 
leafnode-list@xxxxxxxxxxxxxxxxxxxxxxxxxxxx -- mailing list for leafnode
To unsubscribe, send mail with "unsubscribe" in the subject to the list