[ntp:questions] How to get combined clock offset

David Woolley david at djwhome.demon.co.uk
Wed May 11 20:18:27 UTC 2005

In article <d5rhjk$6f9$1 at dewey.udel.edu>, David L. Mills <mills at udel.edu> wrote:

> Your idea of propagating the server offsets upstrata has been suggested 
> before. The problem is, the client gets the immediate downstratum time 

I think you were reading more into what I wrote than I intended.  I was
really trying to say that offset is probably not the measure that he
wanted, but that if he did want to use such a measure, what he would
want would be the estimated error from the reference clock, not that
from the immediate upstream server.

I strongly suspect though, that he's using the local clock on the
immediate upstream server as his reference clock (hence "the server", is
the root server and is the reference clock), as, although unsupported,
that seems the most common configuration used by people asking
questions here.  That's also why I stressed UTC.

If there was an implied proposal, it was more to do with the issue that
is quite often possible for a human observer to make a better guess
at the true offset during startup or after a loop disturbance (heating 
going on etc.) than the daemon and there may be ways of automatically
getting below the phase and frequency noise quicker than at present.
They may, of course, make other situations worse.

More information about the questions mailing list