Path: utzoo!news-server.csri.toronto.edu!cs.utexas.edu!usc!zaphod.mps.ohio-state.edu!tut.cis.ohio-state.edu!ucbvax!parc.xerox.com!deering
From: deering@parc.xerox.com (Steve Deering)
Newsgroups: comp.protocols.time.ntp
Subject: Re: ntp & Sparcstation running 4.0.1 -- loss of sync
Message-ID: <91Mar9.130434pst.3143@skylark.parc.xerox.com>
Date: 9 Mar 91 21:04:32 GMT
References: <91Mar8.102724pst.16508@alpha.xerox.com>
Sender: daemon@ucbvax.BERKELEY.EDU
Distribution: inet
Organization: The Internet
Lines: 32

> Running ntpd (manual page dated 15 June 1989) on a Sun4 sparcstation
> running SunOS 4.0.1c I can get in sync with another host and get the
> offset down to something reasonable (< 100 ms) with a dispersion in the
> thousands.  Then the dispersion suddenly goes out to 64000 and we are
> out of sync for a couple of minutes.  This cycle repeats about every 5
> minutes.  The offset often goes out to as much as 5s.  Someone suggested
> that this may be a know problem (with solution): any ideas?
>
>     Alan Young.

This is a consequence of one or more known problems with the sparcstation
timekeeping code, which I don't have enough time to describe right now.
(If the NTP mailing list is archived somewhere, you might be able to dig
up some old messages giving the gory details.)  But I will tell you what
works for us at Xerox PARC and on Dartnet: we use the xntpd daemon from
the Univ. of Toronto on our sparcstations.  Included in the xntpd release
is a program called "tickadj" that must be run at boot time (before running
xntpd) as follows:

			tickadj -As -t 9999

See the tickadj man page for an explanation.  With this setup, we usually
see offsets less than 1 msec and dispersions less than 2 on our sparcstations
(models 1, 1+ and 2).

One more thing: if you are using the the sparcstation screen for console
output, make sure you are running a window system on that screen, rather
than the plain ROM monitor.  Output via the ROM routines can cause many
clock interrupts to be lost.

Steve Deering
deering@xerox.com