How is the drift value in ntp.drift calculated?

ryandrk@hotmail.com (Drk Ryan) writes:

> How is the drift value in ntp.drift calculated?

Its PPM (parts per million): Set your clock, let it run freely, then
compare with a reference clock. divide the time error by time passed
and multiply by one million. Add a sign according to taste. I thinks
that's about it.

Or was your question more specific?

Regards,
Ulrich

Ulrich Windl <Ulrich.Windl@RZ.Uni-Regensburg.DE> wrote in message news:<m31xv0k7td.fsf@pc5234.rz.uni-regensburg.de>...
> ryandrk@hotmail.com (Drk Ryan) writes:
> 
> > How is the drift value in ntp.drift calculated?
> 
> Its PPM (parts per million): Set your clock, let it run freely, then
> compare with a reference clock. divide the time error by time passed
> and multiply by one million. Add a sign according to taste. I thinks
> that's about it.
> 
> Or was your question more specific?

No I think that was it...thanks. 
So basically drift = offset/interval 

> 
> Regards,
> Ulrich

ryandrk@hotmail.com (Drk Ryan) wrote in message news:<de98eac0.0309011245.2cb3cf86@posting.google.com>...
> Ulrich Windl <Ulrich.Windl@RZ.Uni-Regensburg.DE> wrote in message news:<m31xv0k7td.fsf@pc5234.rz.uni-regensburg.de>...
> > ryandrk@hotmail.com (Drk Ryan) writes:
> > 
> > > How is the drift value in ntp.drift calculated?
> > 
> > Its PPM (parts per million): Set your clock, let it run freely, then
> > compare with a reference clock. divide the time error by time passed
> > and multiply by one million. Add a sign according to taste. I thinks
> > that's about it.
> > 

Where a clock is being disciplined by ntpd then at the end of the
correction interval the time error shoud be very close (time is a
moving target) to zero. In this case should the value in ntp.drift not
be close to zero also?

Drk wrote:
> ryandrk@hotmail.com (Drk Ryan) wrote in message news:<de98eac0.0309011245.2cb3cf86@posting.google.com>...
>> Ulrich Windl <Ulrich.Windl@RZ.Uni-Regensburg.DE> wrote in message news:<m31xv0k7td.fsf@pc5234.rz.uni-regensburg.de>...
>> > ryandrk@hotmail.com (Drk Ryan) writes:
>> > 
>> > > How is the drift value in ntp.drift calculated?
>> > 
>> > Its PPM (parts per million): Set your clock, let it run freely, then
>> > compare with a reference clock. divide the time error by time passed
>> > and multiply by one million. Add a sign according to taste. I thinks
>> > that's about it.
>> > 
> 
> Where a clock is being disciplined by ntpd then at the end of the
> correction interval the time error shoud be very close (time is a
> moving target) to zero. In this case should the value in ntp.drift not
> be close to zero also?

The value in ntp.drift a the systematic frequency error of the system
clock (i.e. the clock permanently loses time or gains time). If
ntp.drift contains 0.000, your clock perfectly keeps time (except for
random errors).

Marc

-- 
  _ _    Marc A. Donges                      +49 721 6904-2130
  'v'    Klosterweg 28 / E110
 /   \   76131 Karlsruhe          PGP-Key(RSA): 1024R/429D9719
  W W                      http://www.hadiko.de/~marc/marc.asc