[ntp:questions] Re: offset of two acts servers
David L. Mills
mills at udel.edu
Fri Feb 3 14:40:17 UTC 2006
rtxo,
Verily the ACTS driver does comply with the NIST delay calibration
scheme. Note that different calls to the NIST services do show rather
different delays, presumably due to different routing in the public
telephone network. However, the delay compensation scheme works very
well. From the delays you show for other Internet servers, the
difference, probably due to assymetric routing, is not suprising.
Dave
rtxo wrote:
>
> Hi Folks--
>
> I've setup a host with gps and wwv audio and the wwv audio gets down
> to within 100 microseconds of the gps routinely. As an experiment, I
> added two external ntp servers to this configuration. Both of these acts
> external servers are "close," being within 20 millisecond ping times.
>
> Oddly, one of the acts hosts is consistently 3-5 milliseconds behind,
> and the other 3-5 milliseconds ahead:
>
> gnu at smidge.wraith.sf.ca.us[244] ntpq -p -n
> remote refid st t when poll reach delay offset
> jitter
> ==============================================================================
>
> *127.127.20.0 .GPS. 0 l 31 64 377 0.000 -0.001
> 0.008
> +127.127.36.0 .WV20. 0 l 34 64 376 0.000 -0.082
> 0.027
> x69.25.96.13 .ACTS. 1 u 804 1024 377 17.315 -4.407
> 0.358
> x207.126.98.204 .ACTS. 1 u 572 1024 377 14.020 3.622
> 0.184
>
> Quite curious. The jitter on the network connection to these hosts
> typically
> is less than a millisecond, although I've seen some Big Combers roll in,
> the
> internet being what it is.
>
> I wonder whats going on with these hosts. The refclock_acts driver code has
> a comment that claims the acts adjusts for the round trip delay.
>
> ../Steven
More information about the questions
mailing list