[ntp:questions] Very large jitter and offsets on GPS ref clock after upgrade to "p5"

Thomas Laus lausts at acm.org
Sat Jan 7 14:52:27 UTC 2012


On 2012-01-06, Chris Albertson <albertson.chris at gmail.com> wrote:
> I just upgraded to?4.2.6p5 at 1.2349-o and I'm getting some odd results.
> NTPD has been running for about 12 hours not on a Linux system. ?The
> following three "peers" displays were taken about one minute?apart.
> How do I go about debugging this? ? Where to start? ? The PPS coming
> from the Oncore has less jitter on it than I can measure with an
> HP5328 counter. ?I shows a period of exactly 1,000,000 uSec.  ?I look
> at "clockstats" and
>
>
> I know you are going to ask so here are the config files
>
> MODE 4
> LAT    33 51 54.315
> LONG -118 23 01.782
> HT     25.56 M
> DELAY 50 NS
> ASSERT
> SHMEM /var/log/ntpstats/ONCORE
> POSN3D
> MASK 0
>
> ****************************
Cris:

The Oncore driver has had a few recent syntax changes and it is possible that
your PPS input is not being seen or works different than it had in the
past.

Here is my Oncore configuration file.

HARDPPS
PPS_CAPTUREASSERT
MODE 1
LON -84.2017844758
LAT 40.7762210511
HT 223.445
MASK 20
DELAY 92.1 ns

Look at the information presented with the ntpq -c kern command:


associd=0 status=0415 leap_none, sync_uhf_radio, 1 event, clock_sync,
pll offset:            -0.002089
pll frequency:         -10.0246
maximum error:         0.002239
estimated error:       1e-06
kernel status:         pll ppsfreq ppstime ppssignal nano
pll time constant:     4
precision:             1e-06
frequency tolerance:   495.911
pps frequency:         -10.0246
pps stability:         0.0134583
pps jitter:            0.002
calibration interval   256
calibration cycles:    4614
jitter exceeded:       3690
stability exceeded:    0
calibration errors:    8

The keywords are ppsfreq, ppstime, ppssignal and nano.  That should
tell you if pps is working on your system.

Tom

-- 
Public Keys:
PGP KeyID = 0x5F22FDC1
GnuPG KeyID = 0x620836CF



More information about the questions mailing list