Re: ntpdate steps 86400 seconds

K.D.P wrote: 
Yes I´m saying that ntpd doesn´t get any better than 200ms on my hardware.
My test on ntpd and ntpdate doesn´t result in that your saying about ntpdate
being less accuracy.


You may be confused by the output of "ntpq -p". The units are in
milliseconds, not seconds. "0.200" means 200 microseconds.

If ntpd were really only keeping an offset of 200 milliseconds,
it would be continually stepping the time back to a zero offset.
Any offset greater than 128 milliseconds for a few polling cycles
causes ntpd to step the time rather than slew it. Note also that
you have to run ntpd for a little while before it stabilizes and
begins to keep up with the server (computes its characteristic
drift rate).

On the other hand, given that the server you are using is itself
stepping the time rather than maintaining a smooth monotonically
increasing time base, it may not be surprising that ntpd didn't
work well when using that server (or one like it).

-Tom
.