On Mon, 5 Apr 2004, Brian Fahrlander wrote:
> On Mon, 2004-04-05 at 11:33, Tom 'Needs A Hat' Mitchell wrote:
>
> > At the end of an strace of rdate -s time.nist.gov I see:
> >
> > 5863 connect(3, {sa_family=AF_INET, sin_port=htons(37), sin_addr=inet_addr("192.43.244.18")}, 16) = ? ERESTARTSYS (To be restarted)
> > 5863 --- SIGALRM (Alarm clock) @ 0 (0) ---
> > 5863 +++ killed by SIGALRM +++
> >
> > I suspect that there is a time-out built into rdate that is getting
> > triggered. Since this the weekend that daylight saving time switched
> > here I bet the site was being hit bunches.
> >
> > I guess I could look at the source and ... naw guessing
> > is too much fun. ;-)
>
> Good man! You found a reason where once there was none. Kudos.
>
> Will the source, though, say why they picked "Alarm clock" as a
> diagnostic instead of "Time source overloaded" or "Time not available"?
Got me curious, so I poked around a little bit.
The message appears to come from the SIGALRM implementer. (The string
"Alarm clock" appears nowhere in the rdate binary.) SIGALRM is a
general-purpose POSIX timeout signal--see alarm(2). It can be used, for
example, to implement the sleep(3) function and other things. So the fact
that you are getting the message from a clock-related program is a mere
coincidence.
>
> Hackers. Good or bad, they're always a couple of degrees askew from
> the rest of the world. :>
Bless 'em!
--
Matthew Saltzman
Clemson University Math Sciences
mjs AT clemson DOT edu
http://www.math.clemson.edu/~mjs
