[ntp:questions] ntpdate refusing ntp.nmi.nl
David Woolley
david at ex.djwhome.demon.co.uk.invalid
Sat Jan 10 00:00:53 UTC 2009
Folkert van Heusden wrote:
>
> Ok so the root dispersion is not too high if I understand you correctly
> (well it would be better if it were lower)
The root dispersion is consistent with having a very old reference time.
The reference time is too old, which means the root dispersion is too
high. It is the high root distance (which includes the root dispersion)
that is causing the server to be rejected.
> The systems I checked that sync to this server (and others, of course)
> seem to ignore it:
>
> *GENERIC(0) .DCFa. 0 l 55 64 337 0.000 -2.421 2.390
> +adsl.remco.org .GPS. 1 u 504 1024 377 19.581 3.159 0.188
> ...
> ntp.nmi.nl .IRIG. 1 u 535 1024 377 12.146 52.565 0.044
>
>
> The reason that I keep on going on this system is that I would like to
> have them get it to also work with ntpdate and such and not only the
> windows implementation. The reasons for them to fix it I come up with is
It is not working with nptd, either. There is no character in front of
it, which means that it is being rejected. If you do rv for the
association, you will see a code indicating that the the distance is too
high.
> now:
> - root dispersion too high
and therefore root distance is too high, and therefore it is rejected as
a source.
> - offset too high; 52ms (tried from ADSL, cable and SDSL connected
The very large root distance would make 52ms highly acceptable, if the
clock weren't being rejected. 52ms is well within the, nearly, 4,000ms
uncertainty.
In reality, this figure could either represent that it really has been
unsynchronised for three days, but is drifting at rather less than 15ppm
(and 15ppm is really rather pessimistic), or that you have a calibration
error on your PPS input.
> systems, synced to GPS+PPS, DCF77, MSF and systems on the internet)
> - reference time too old
DCF is not good for very high accuracy time.
Someone mentioned that this machine is only intended for synchronising
Windows. I don't believe that w32time performs a root distance check on
its servers.
More information about the questions
mailing list