>> I'm (still) running NetBSD 1.4.2 on a SPARCstation 20.
> 
> I was running NetBSD 1.4.2 on a SS20 for quite some time; but I was using
> NTP v4 (4.0.99k, to be precise) and a locally attached GPS device (with
> some kernel fixes for the the kernel pps support; see pr kern/13072).
> If you don't want to upgrade to a more recent NTP, you might enable
> statistics, and see if that explains anything.

How do I do that?

> Your drift file claims
> that your system clock is just about perfect (error is 0.000), which looks
> a little suspicious to me.  What does "ntpq -p" say after the ntp daemon
> has been running a while (several hours)?

Well, this looks pretty bad:

[5:33] isolar:/tmp % ntpq -p
     remote           refid      st t when poll reach   delay   offset    disp
==============================================================================
 LOCAL(1)        LOCAL(1)         8 -    -   64    0     0.00    0.000 16000.0
 workmachine.jpl 0.0.0.0         16 -    -   64    0     0.00    0.000 16000.0
 ftp06.apple.com 0.0.0.0         16 -    -   64    0     0.00    0.000 16000.0

[5:33] isolar:/tmp % ntpq -c associations
ind assID status  conf reach auth condition  last_event cnt
===========================================================
  1 50820  8000   yes    no
  2 50821  8000   yes    no
  3 50822  8000   yes    no

I see regular polls of these two machines, why would they be banished to
Stratum 16 (and, presumably, ignored)?  Why are they considered "no"[t]
reachable?

I ktrace'd the running xntpd and I see the polls, and occasional calls to
a "ntp_adjtime" system call that always returns a "5" error code:

 13911 xntpd	CALL	ntp_adjtime(0xeffffa30)
 13911 xntpd	RET	ntp_adjtime 5

I have a pretty simple ntp.conf:

-------
server          workmachine.jpl.nasa.gov.        version 2
server          time.apple.com

driftfile       /var/db/ntp.drift

restrict default notrust nomodify noserve
restrict N.N.N.N mask 255.255.255.0		# Local network
restrict 127.0.0.1                              # localhost

keys            /var/db/ntp.keys
requestkey      *****

# local hardware clock used as last-resort backup if network fails
server  127.127.1.1            # local clock treated as stratum 8
fudge   127.127.1.1 stratum 8
-------

Why won't my xntpd listen to the 2 servers it regularly polls?

Apologies for the non-SPARC-port-specific nature, but "xntpd" and I just
have never really played well together in the NetBSD/SPARC sandbox.  :-)

	- Greg