[ntp:questions] Root dispersion: understanding and possibly reducing it

Neil Green ncguk at hotmail.co.uk
Tue Dec 1 11:57:15 UTC 2015


Hi all,

I have a stratum 1 ntp server running on Raspbian on a Raspberry Pi (a Zero, though this applies to any model Raspberry Pi) and I would like to see if it's possible to reduce my rootdisp, which fluctuates between 1.000 and 1.400-ish. It never drops below 1.000.

My ntp.conf:

---

driftfile /var/lib/ntp/ntp.drift
logfile /var/log/ntp.log
leapfile /home/time/leap-seconds.3645216000

disable kernel

server 80.5.182.144 iburst
server 89.16.173.64 iburst
server 143.210.16.201 iburst
server 217.114.59.3 iburst
server 158.43.128.66 iburst
server 129.215.42.240 iburst

server 127.127.20.0 mode 24 minpoll 4 maxpoll 4 iburst prefer
fudge 127.127.20.0 flag1 1 time1 0.000 time2 0.400 flag3 0 refid RPIZ

restrict -4 default kod notrap nomodify nopeer noquery

restrict 127.0.0.1

---

Output of 'ntpq -crv -pn; /usr/local/sbin/ntptime; ntpq -c clockvar':

associd=0 status=0415 leap_none, sync_uhf_radio, 1 event, clock_sync,
version="ntpd 4.3.84 at 1.2483-o Fri Nov 27 18:47:32 UTC 2015 (1)",
processor="armv6l", system="Linux/4.1.13+", leap=00, stratum=1,
precision=-20, rootdelay=0.000, rootdisp=1.225, refid=RPIZ,
reftime=da080321.6796e9cb  Tue, Dec  1 2015 11:29:37.404,
clock=da080330.c7965550  Tue, Dec  1 2015 11:29:52.779, peer=39168, tc=4,
mintc=3, offset=0.003835, frequency=-11.387, sys_jitter=0.002954,
clk_jitter=0.001, clk_wander=0.000, tai=36, leapsec=201507010000,
expire=201512280000
     remote           refid      st t when poll reach   delay   offset  jitter
==============================================================================
*80.5.182.144    10.178.1.162     2 u    4   64  377   12.301    2.254   1.227
-89.16.173.64    94.125.132.7     3 u   55   64  377   18.123    4.441   0.659
-143.210.16.201  158.43.192.66    2 u   38   64  377   21.984    5.279   2.965
-217.114.59.3    33.117.170.50    2 u   12   64  377   19.178    5.351   0.850
+158.43.128.66   193.67.79.202    2 u   20   64  377   15.751    2.079   0.488
+129.215.42.240  129.215.64.32    2 u   36   64  377   23.160    1.119   1.005
o127.127.20.0    .RPIZ.           0 l   15   16  377    0.000    0.004   0.003
ntp_gettime() returns code 5 (ERROR)
  time da080330.f6137f38  Tue, Dec  1 2015 11:29:52.961, (.961235914),
  maximum error 16000000 us, estimated error 16000000 us, TAI offset 0
ntp_adjtime() returns code 5 (ERROR)
  modes 0x0 (),
  offset 0.000 us, frequency 0.000 ppm, interval 1 s,
  maximum error 16000000 us, estimated error 16000000 us,
  status 0x2041 (PLL,UNSYNC,NANO),
  time constant 3, precision 0.001 us, tolerance 500 ppm,
associd=0 status=0000 no events, clk_unspec,
device="NMEA GPS Clock", timecode="$GPZDA,112952.00,01,12,2015,00,00*6E",
poll=4119, noreply=0, badformat=0, baddata=0, fudgetime2=400.000,
stratum=0, refid=RPIZ, flags=1

---

Rootdisp is described as "total dispersion to the primary reference clock" and that dispersion "Represents the maximum error of the local clock relative to the reference clock". So is the rootdisp merely the disparity between the local computer's clock and the reference time (in this case GPS/PPS)?

If so, can the rootdisp be reduced and how? Also, does it matter in the slightest in the context of serving precise time?

Thanks for any input, explanations etc.


More information about the questions mailing list