SUMMARY: Can't understand these xntp Error Messages

From: Jochen Bern (
Date: Wed Jun 28 1995 - 11:16:16 CDT

I even haven't seen my Request come back to me, and the Problem is still
manifest at this Moment, but I think I've gotten Information enough to
cure and explain it, so I summarize right now.

Technical Answer: Run xntp's tickadj Program on Boot (Options -As).
Philosophical Answer: Learn to use nroff. :-)

I wrote:
> I've set up xntp lately (version 3.1) and keep seeing Error Messages
> that I don't really understand. I designated two Machines as Servers
> (aurelia and penthesilea) [that] keep getting Lines like
> aurelia xntpd[28263]: Previous time adjustment didn't complete
> Likewise, the other Hosts which attempt to do a
> '/usr/local/etc/randomized_ntpdate -s aurelia penthesilea'
> hourly from cron cause a
> antiope ntpdate[27368]: no server suitable for synchronization found
[All this on 4.1.3_U1B]

Technicalities first:

> aurelia xntpd[28263]: Previous time adjustment didn't complete

This is caused by having the OS still trying to sync the Unix Clock to
the Hardware Clock, which is of course counterproductive when running
xntp. Upon Installation of xntp, one should use the util/tickadj Program
to check out the Kernel Variables tickadj and dosynctodr. In a Nutshell:
On every Hardware Clock Tick, the Kernel adds 'tick' Microseconds to the
Unix Clock; If adjusting the Clock by Skew (date -a, adjtime(2) and
Friends), tick will be raised or lowered by tickadj; dosynctodr != 0
triggers regular Syncs to the Hardware Clock. The usual Value of tickadj
(tick/10, set in /sys/*/param.c) is WAY to high when running xntp, as
all Adjustments are forced to be a Multiple of it. dosynctodr can't be
set except by adb and xntp's tickadj as far as I can tell.

Since I don't want to adb obscure Stuff into some Kernels (which I'm bound
to have forgotten next Time we compile Kernels), I copied xntp's Program
util/tickadj to /usr/local/etc and added a Line "/usr/local/etc/tickadj -As"
(set tickadj to Value deemed optional and dosynctodr to Zero) immediately
preceding the Call to xntpd to /etc/rc.local.

The Error Messages on the Clients:

> antiope ntpdate[27368]: no server suitable for synchronization found

are merely a Consequence of both Servers having Problems, which makes them
tell the Clients "don't trust me" when queried.

All this is described in doc/, but not having much Experience
with nroff, I tried to nroff -man it (instead of nroff -me), making the
Output hard to read, and especially chopping the Header which would have
pointed me to that important Detail.

On another Note, I've been told that 3.1 is pretty outdated; 3.4t is
the current one.

Thanks to Tim Pointing <> for having the technical
Information right on Top of his Head.

                                                                J. Bern

  /\  /""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""\
 /  \/ (7Bit,Size Limit!) | P.O. Box 1203 | Ham:  \/\
/ J. \ (8Bit,SUN Att.) | D-54202 Trier | DD0KZ /  \
\Bern/ No Finger etc.; Use Mail (Subj. "##" for Autoreply List) and \  /
 \  /\ WWW. /\/
  \/  \____________________________________________________________/

This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:10:28 CDT