svn commit: r326095 - head/usr.sbin/bsdinstall/scripts

[Ian Lepore]

On Sat, 2017-11-25 at 22:09 +1100, Bruce Evans wrote:
> On Fri, 24 Nov 2017, Ian Lepore wrote:
> 
> [...]
> 
> Is there any use for ntp as a client if you have an atomic clock?  Just to
> validate both it and ntpd?
> 
> Bruce

Ntpd treats the atomic clock as one of its peers...  the clock emits a
PPS pulse and serves as a refclock peer.  There generally also needs to
be a network peer, or another type of refclock such as a gps receiver,
to tell ntpd how to number the seconds.  Some atomic clocks are able to
number the seconds as well (such as a Microsemi 5071a cesium), but
that's not a feature that's used often in my experience.

When the frequency output of the atomic clock directly drives the
kernel timecounter (obviously using some custom hardware), ntpd isn't
very good at being just a network protocol server.  It really wants to
steer the kernel clock based on phase/frequency measurements from some
peer.  There are dregs of old "LOCKCLOCK" support code in ntpd which is
supposed to handle the "clock is externally steered" case, but when I
tried to configure and build it to run that way last year, I had no
success (which could be due to poor documentation and my
misunderstanding of it, or could just be the support has bitrotted).

-- Ian