[ntp:questions] Local time sources

Jaap Winius jwinius at umrk.nl
Mon May 10 19:14:22 UTC 2010

On Mon, 10 May 2010 07:58:37 +0100, David J Taylor wrote:

> I suspect that depends on the precision required.  If you /need/ the
> precision which the GPS provides, and an inaccurate time is
> unacceptable, then add no other sources.  If serving time at a reduced
> accuracy is acceptable, add four or five external servers for
> redundancy.  In addition to redundancy, another advantage might be that
> you get a check that you aren't working on the wrong edge of the GPS
> pulse, by looking at the output of ntpq -p, and seeing that the offset
> of the backup sources is reasonable.

Okay, I think this answers my question. I've been running a Garmin GPS 18 
LVC for a few years now...


What this page does not say is that my configuration also includes a 
DFC77 receiver. This long-wave radio time signal receiver is an order of 
magnitude less precise than the GPS receiver, but it has already proved 
its value several times in terms of redundancy: twice when the GPS fell 
from the skylight following a suction-cup failure, and once when gpsd 
failed to function properly after an update. That's great, but the point 
is that there's not much else to it.



More information about the questions mailing list