Re: Clue on 2.0.33 crashes

Ulrich Windl (ulrich.windl@rz.uni-regensburg.de)
Mon, 23 Feb 1998 09:47:39 +0100


On 20 Feb 98 at 16:14, Hubert Mantel wrote:

> Hi,
>
> On Fri, 20 Feb 1998, Doug Ledford wrote:
>
> > Jos Vos wrote:
> > >
> > > Thomas Schenk wrote:
> > >
> > > > And I would add that we have found that prior to every freeze that our
> > > > systems have experienced under 2.0.33, we find a syslog message from xntpd
> > > > as follows:
> > > >
> > > > Feb 11 09:18:46 acme xntpd[151]: kernel pll status change 89
> > > >
> > > > These messages do not occur immediately prior to the crash, but we have
> > > > theorized that this status change is a precursor to the lockup and the
> > > > number of lockups that we have seen has been greatly reduced since we
> > > > stopped using xntpd and started using ntpdate with the -b option in a
> > > > cron job to sync the clocks.

I think I have answered such a question before. For i386 architecture
no problem arises. Really! For other architectures I don't know,
because the state as of 2.0.33 seems inconsistent. If you add one of
my recent PPSkits I tried to make the implementation consistent
again. I have never tried that, but still I don't believe that the
system crashes because of that.

BTW: I have posted an article in comp.protocols.time.ntp about
"kernel pll status change 89"...

> >
> > Hmmmm, I'm running xntpd on about 5 systems and I've never had a lock nor
> > seen this message. Of course, that could be merely luck. The real question
> > in my mind is what is that message all about. There were significant
> > updates to the kernel time code in 2.0.32 or 2.0.33 (can't remember which).

First of all the message is NOT from the kernel, and secondly it
depends on your ntp.conf file. If you have a verbose "logconfig" at
the beginning of the file, you might see it.

The message indicates that that the time state has been detected to
be instable; usually right after a restart or time step. Can also
indicate network congestion or very high machine loads.

> > It seems (by memory) that the changes were written by either Ulrich
> > {Wendel,Drepper} (SP?). Maybe the author would comment?

``Windl, like the wind plus an "ell"'' ;-)

>
> We're running xntpd on all of our machines. IIRC Ulrich Windl is not
> subscribed to the kernel list.

Right, but I'm answereing non-SPAM mail.

Regards
Ulrich
P.S. I hope I forgot nobody to reply to.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu