Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 16 Dec 1999 18:51:27 +0100
From:      Poul-Henning Kamp <phk@critter.freebsd.dk>
To:        mjacob@feral.com
Cc:        Ollivier Robert <roberto@eurocontrol.fr>, "FreeBSD Current Users' list" <freebsd-current@FreeBSD.ORG>
Subject:   Re: HEADSUP: ntp4 to replace xntpd 
Message-ID:  <16818.945366687@critter.freebsd.dk>
In-Reply-To: Your message of "Thu, 16 Dec 1999 09:34:50 PST." <Pine.BSF.4.05.9912160934230.23583-100000@semuta.feral.com> 

next in thread | previous in thread | raw e-mail | index | archive | help
In message <Pine.BSF.4.05.9912160934230.23583-100000@semuta.feral.com>, Matthew
 Jacob writes:
>
>Huh? What about the impact on all ntp.conf files? Or is this seamless?

I was just about to start to compose an email with some info on this
one when you email arrived.

/etc/ntp.conf is the same unless you have a refclock.  If you have
a refclock you need to revisit your setup.

About NTPv4 in general I can say that a lot of things have changed under
the hood.

Between the two of us Dave Mills and I have managed to get the
"nanokernel" to act sensibly in the domain inside +/- 1usec which
the old one didn't.  (See http://gps.freebsd.dk for what kind of
performance this can result in, given appropriate hardware).

A lot of changes in the "control" code in ntpd has also changed,
and as far as I can tell for the better all around.

There is support for *very* long poll intervals (18hours I belive
a lot cheaper if you sync by dialup) and burst mode which is more
suitable for dial-on-demand kind of lines.

There is a new API for PPS kind of signals (I'm also partly 
guilty here, but I must admit that I think the result is far
too rococco for my taste).

One thing which is new, is the initial synchronization:  it can look
broken, but it isn't.  You will likely see something like this:

	1. Ntpd starts

	2. Home in on some server, steps the clock to zero offset.
	   in the process we loose sync again.

	3. Catch sync again.  Very light and slow adaptation of
	   frequency the clock slowly drifts off to +/- 128msec.

	4. Clock is steped again and frequency set to slope of the
	   drift from step 3.

	5. Business as usual.

Depending on your clock step three can take hours to complete,
the better your clock the longer it takes.  Dave Mills old rule 
of thumb applies even more than before:  Only tweak your ntp.conf
right before lunch.  That gives it 24hours to settle before you
fiddle it again.

You can find much more about NTPv4 on www.ntp.org

--
Poul-Henning Kamp             FreeBSD coreteam member
phk@FreeBSD.ORG               "Real hackers run -current on their laptop."
FreeBSD -- It will take a long time before progress goes too far!


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-current" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?16818.945366687>