[ntp:questions] Time offset / Jitter under FreeBSD

Bob bobsjunkmail at bellsouth.net
Sun Dec 2 19:13:20 UTC 2007


"Garrett Wollman" <wollman at bimajority.org> wrote in message 
news:fiurcr$cva$1 at grapevine.csail.mit.edu...
> In article <fitlnh$phn$1 at registered.motzarella.org>,
> Bob <bobsjunkmail at bellsouth.net> wrote:
>
>>I installed FreeBSD on a drive, and booted my windows machine from it. The
>>FreeBSD port of ntpd shows a offsets in the 20-60 ms range. Issues with
>>jitter, also.
>
> How long did you leave it running?  What drift value was reported?
>
> My stratum-0, running FreeBSD 6.2, consistently reports offsets to
> other stratum-0s within +/- 5 ms, and to its refclock within a couple
> of microseconds with the nanokernel and kernel PPS.  My stratum-1s,
> likewise.  These are all running the stock ntpd, which is 4.2.0-a.
>
> $ ntpdc -c kerni ntp-0
> pll offset:           1.314e-06 s
> pll frequency:        68.119 ppm
> maximum error:        0.012947 s
> estimated error:      7e-06 s
> status:               2107  pll ppsfreq ppstime ppssignal nano
> pll time constant:    6
> precision:            1e-09 s
> frequency tolerance:  496 ppm
> pps frequency:        68.119 ppm
> pps stability:        0.022 ppm
> pps jitter:           1.142e-06 s
> calibration interval: 256 s
> calibration cycles:   85694
> jitter exceeded:      506860
> stability exceeded:   0
> calibration errors:   3

For grins, I decided to start it with a -N. That does seem to make a 
difference. Someone else suggested getting the latest distro - 4.2.4.p4 - no 
difference. The compile was a learning experience in itself, as I've just 
started messing with BSD. Offset was apparent on all three computers I tried 
booting my install on. As far as how long it was running....  a few hours. 
My issue was that the offset was getting larger with time. The Meinberg 
version on the very same hardware under Win XP is consistently sub 
millisecond. I'll eventually be feeding it 1PPS directly.  What is "....with 
the nanokernel and kernel PPS...."? When I compiled the 4.2.4.p4, I used 
their defaults, and I've wound up with a 1.3 meg binary - quite a bit larger 
than the one that is the default in BSD 6.2. I'm assuming that has 
everything under the sun in it.

n2qew# ntpdc -c kerni
pll offset:           0.00176014 s
pll frequency:        9.271 ppm
maximum error:        0.082631 s
estimated error:      0.000428 s
status:               2001  pll nano
pll time constant:    4
precision:            1e-09 s
frequency tolerance:  496 ppm

n2qew# ntpq -p
     remote           refid      st t when poll reach   delay   offset 
jitter
==============================================================================
-tesla.fireduck. 63.192.96.10     2 u   52   64  377   34.693    0.110 
1.032
+vine.desynched. 132.163.4.101    2 u   54   64  377   36.271    1.617 
0.493
-server1-a.your. 64.202.112.75    2 u   53   64  377   56.687    9.209 
1.072
+207.46.197.32   138.23.180.126   3 u   56   64  377  106.590    2.940 
0.880
*10.33.90.10     68.216.79.113    2 u    7   16  377    0.211    2.887 
0.091 





More information about the questions mailing list