[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [leafnode-list] How can i suppress this message (Alarm clock)?

On Tue, 14 Sep 2004, Michael Rulov wrote:

> 16:28 home/$ fetchnews >/dev/null 2>@1
> Alarm clock

Short answer: not easily.

Try setting timeout=60 (or longer, the figure is in seconds) for each of
your servers in /etc/leafnode/config.

Long answer:

fetchnews uses the alarm() syscall to request an interrupt 30 seconds
(default time, configured as above) after starting a read from the
network, in order not to hand for hours if the upstream server isn't
responding but isn't hanging up either. This will send a SIGALRM, which
_should_ be caught by fetchnews. Apparently, this does not work. One
cause may be that I still haven't migrated _all_ signal() syscalls to
sigaction() and the mix may cause problems. I'll convert all signal()
calls to sigaction() and hope that will remedy your problem.

You can also try to run fetchnews with verbose logging (try -vvv) to see
where it catches the problem.

HTH for now, more later,
leafnode-list mailing list