Hi,

At 9:55 -0600 6-2-2001, Frederick Bruckman wrote:
>I found that with NetBSD-1.4.3 and ntp-4.?, it sort of syncs, it's
>just that disk activity makes it run so slow that the correction is
>pegged at 500ppm, and it almost never catches up. It's a shame,
>because the real-time clock is very accurate, and only loses a couple
>of seconds a year. On a diskless machine, or perhaps one with a small
>read-only disk, NTP might actually work well.

My experiences with ntp aren't that bad. In fact, I was surprised not to
see the often reported notorious time-loss on my Centris 650. Instead,
ntpd occasionally logs only small corrections (few tenths of a second).
Only during heavy disk activity (moving /usr/pkgsrc to another disk, full
compile of a kernel, etc.), I see lines like the following in the syslog:

   Feb  1 15:43:15 macbsd ntpd[147]: time reset 0.239779 s
   Feb  1 16:14:19 macbsd ntpd[147]: time reset 3.545726 s
   Feb  1 16:33:51 macbsd ntpd[147]: time reset 3.434815 s
   Feb  1 16:56:31 macbsd ntpd[147]: time reset -0.318382 s

My /etc/ntp.conf contains no local clock, only the server lines pointing
to our university's time servers. /var/db/ntp.drift contains "196.323".
This is with 1.5, Jan.5 snapshot, almost -current kernel.

Best wishes,

        Nico
-- 
Dr. N.J.R. van Eikema Hommes   Computer-Chemie-Centrum
hommes@chemie.uni-erlangen.de  Universitaet Erlangen-Nuernberg
Phone:   +49-(0)9131-8526532   Naegelsbachstrasse 25
FAX:     +49-(0)9131-8526566   D-91052 Erlangen, Germany