[ntp:questions] Everything works! -- but am I doing more than I need to?

John Ackermann N8UR jra at febo.com
Wed Nov 26 01:30:46 UTC 2003

Thanks for several helpful responses to my question yesterday re the Oncore 
driver on a FreeBSD system.  I'm now up and running, but am still somewhat 
baffled by the various bits of documentation and the relationship between 
the Oncore driver, the PPS driver, and the kernel PPS discipline.  I'm 
running ntpd version 4.1.1c-rc3 from the FreeBSD ports collection.

Here's the operative part of my /etc/ntp.conf file:

	pps /dev/oncore.pps.0 assert hardpps
	server prefer
	fudge flag3 1

And here's what ntpq -p shows:

	+GPS_ONCORE(0)	.GPS.            0 l    5   16  377    0.000    0.000 
	oPPS(0)			.PPS.            0 l   13   64  377    0.000    0.003   0.003

My question is how many of those lines in /etc/ntp.conf are needed.

I've seen references that seem to indicate that the server 
prefer" line is all you need, but without setting flag3 it doesn't appear 
that the kernel PPS discipline is started.

I've seen references that say that the Oncore driver "handles all the PPS 
stuff" but without the "server" line, I never got the PPS 
driver to appear in the ntpq -p output.

I've seen references that indicate (I think) that you don't want to use the 
kernel PPS discipline because ntpd does a better job.  So, I'm not sure if 
I *want* the PPS driver to be there...

So, what to do?  What setup will yield optimum timekeeping performance for 
an Oncore hooked to a FreeBSD ntp box (I'm using FreeBSD on this dedicated 
system because of the numerous comments on the list over time that Linux 
PPS is a box of horrors; I can take a hint :-) ).

For what it's worth, I've also attached the results of ntptime at the end 
of this message.  While we're at it, do the number of "jitter exceeded" and 
"errors" shown seem reasonable?  The system has been running for about 8 



$ ntptime
ntp_gettime() returns code 0 (OK)
  time c36e7ff3.fecd4a40  Tue, Nov 25 2003 20:27:15.995, (.995320576),
  maximum error 33873 us, estimated error 2 us, TAI offset 0
ntp_adjtime() returns code 0 (OK)
  modes 0x0 (),
  offset -2.973 us, frequency 95.808 ppm, interval 256 s,
  maximum error 33873 us, estimated error 2 us,
  time constant 6, precision 0.001 us, tolerance 496 ppm,
  pps frequency 95.808 ppm, stability 0.015 ppm, jitter 1.072 us,
  intervals 169, jitter exceeded 68, stability exceeded 0, errors 7.

More information about the questions mailing list