[ntp:questions] Bug 2314, Clock jitter reads zero in the loopstats file

Mischanko, Edward T Edward.Mischanko at arcelormittal.com
Thu Jan 10 05:44:04 UTC 2013


I checked my home FreeBSD 8.2 NTP server for the clk_jitter reads zero bug, and yes, I see it too when using PPS kernel discipline.

server 127.127.22.1 flag3 1

http://www.eecis.udel.edu/~mills/ntp/html/drivers/driver22.html

So, I need to understand what "Controls the kernel PPS discipline" means and what the differences in code are between disabled, (default), and enabled are.  I also need to understand the importance of the clock jitter statistic; I notice that system jitter is still reported with kernel PPS discipline enabled.  If system jitter is less --yet to be determined--with kernel PPS discipline enabled, wouldn't that imply that kernel PPS would be prefered, regardless of reported clock jitter?  Again, I have been mostly concerned with the clock offset and the stability of the frequency.  I need to understand the importance and differences in both clock jitter and system jitter?

Regards,
Ed

 

> -----Original Message-----
> From: Mischanko, Edward T
> Sent: Wednesday, January 09, 2013 1:34 PM
> To: 'David J Taylor'
> Subject: RE: Raspberry Pi Linux distributions w/ NTP
> 
> David,
> 
> I'm glad I was able to suggest something that is helpful.  As to
> what it means....  I am guessing that something in the Linux
> kernel is not quite right when it comes to NTP.  What that
> something is, I don't know.  Is it possible to submit a bug to
> the Linux community as to how they are implementing PPS support
> in their kernel?
> 
> Regards,
> Ed
> 
> > -----Original Message-----
> > From: David J Taylor [mailto:david-taylor at blueyonder.co.uk]
> > Sent: Wednesday, January 09, 2013 10:40 AM
> > To: Mischanko, Edward T
> > Subject: Re: Raspberry Pi Linux distributions w/ NTP
> >
> > OK, Ed.  I changed RasPi-1 at 14:20 UTC,  Please let me know
> if
> > you see any
> > difference....
> >
> >   http://www.satsignal.eu/mrtg/performance_ntp.php
> >
> > Cheers,
> > David
> > ==============================
> >
> > Interestingly, clk_jitter is now being reported as 0.004 ms,
> and
> > the zero
> > jitter in bug 2314 is not present in the loopstats.  It seems
> to
> > be
> > stabilising at 0.003815 ms.
> >
> > Cheers,
> > David
> > --
> > SatSignal Software - Quality software written to your
> > requirements
> > Web: http://www.satsignal.eu
> > Email: david-taylor at blueyonder.co.uk



More information about the questions mailing list