[ntp:questions] Is this normal?

unruh unruh at wormhole.physics.ubc.ca
Tue May 11 14:58:40 UTC 2010

On 2010-05-10, Russell, David <David.Russell at dowjones.com> wrote:
> --_004_C430DA47D51D0C4B940F6928BAE21F4626A3F15C21SBKMXSMB06win_
> Content-Type: text/plain; charset="us-ascii"
> I am trying to verify the correct implementation of NTP on a device that is configured to poll a GPS based dedicated time server on the local area network.  The round trip times to this stratum 1 vary between 3.3 and 3.4ms.

Wow. that is a looooong time. If it is really on the same local network,
it should be much shorter than that. What is causing the delays-- bad
router/switch, or is the local network a modem controlled network?

> I have plotted the offset vs drift in the included jpg.  The drift rate varies between -3 and -3.2 us/s and the offset is +/- 200us.  Note that the first part of the graph is when NTP started.
> My questions are that given that a device has excellent network access to a good stratum 1 server should the drift rate vary this much?  Can I infer anything from the data regarding proper computations in the code, timestamping of the packets or even the system clock?

Sorry, that can in no way be called excellent network access. It is very
poor network access. With the huge variablility of the offset, one would
expect variability of the rate. What poll interval are you using on the
Did you write the implimentation of ntp that you are asking whether it
works correctly?  And what is this "device"?


More information about the questions mailing list