[time] variables reported by ntpq
Tue Nov 14 07:52:15 UTC 2006
>> I searched 'frequency' in RFC 1305 but I did not find such a
>> I manage 7 ntp servers and this variable may vary from -255.7 to
>> on them.
> Frequency is actually the clock frequency error, measured in PPM (parts
> million). Usually not more than a few tens of PPM.
Oh, I see.
>> Another undefined variable is noise. Nor the manual neither the RFC
>> does mention it. (It's value is in the 37-200 range and quite
> Maybe you mean jitter or stability?
> I don't see anything called noise from readvars
> in ntpq, if you could cut and paste showing where the phrase occurs I
$ ntpq -c rl localhost
assID=0 status=0644 leap_none, sync_ntp, 4 events, event_peer/strat_chg,
version="ntpd 4.2.0a at 1:4.2.0a+stable-2-r Fri Aug 26 10:30:12 UTC 2005 (1)"?,
processor="i686", system="Linux/2.6.17", leap=00, stratum=2,
precision=-20, rootdelay=0.508, rootdispersion=28.292, peer=9428,
reftime=c903ecee.571ae2da Tue, Nov 14 2006 8:21:18.340, poll=9,
clock=0xc903ef95.b221ea35, state=4, offset=-0.026, frequency=-255.776,
noise=0.348, jitter=0.340, stability=3.180
> we could help you. 37 to 200 ms might be typical jitters for a
> network connection. I think different versions of ntpq use different
> for jitter, some using ms and some using seconds.
My jitter above is definitely in ms comparing to to the output of
"ntpdc -c sysinfo localhost":
root distance: 0.00050 s
root dispersion: 0.03027 s
jitter: 0.000381 s
stability: 3.180 ppm
broadcastdelay: 0.003998 s
authdelay: 0.000001 s
I've attached a graph of noise. (The measurement unit is just guessed.)
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 11710 bytes
Url : http://fortytwo.ch/mailman/pipermail/timekeepers/attachments/20061114/3f8b57c7/attachment-0001.obj
More information about the pool