Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 6 Jul 2009 19:37:01 -0400
From:      Louis Mamakos <louie@transsys.com>
To:        Christian Weisgerber <naddy@mips.inka.de>
Cc:        freebsd-net@freebsd.org
Subject:   Re: NTP - default /etc/ntp.conf
Message-ID:  <20090706233701.GA30824@ringworld.transsys.com>
In-Reply-To: <h2t1pr$28me$1@lorvorc.mips.inka.de>
References:  <20090702084608.GA59311@roberto-al.eurocontrol.fr> <200907020955.aa73037@walton.maths.tcd.ie> <h2t1pr$28me$1@lorvorc.mips.inka.de>

next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, Jul 06, 2009 at 02:30:19PM +0000, Christian Weisgerber wrote:
> David Malone <dwmalone@maths.tcd.ie> wrote:
> 
> > (Also, we probably really want people to run in orphan mode rather
> > than local clock mode, but we can wait a little longer until orphan
> > mode is more commonly deployed, IMHO...)
> 
> I didn't know about orphan mode, so I had to try it right away.
> 
> $ cat /etc/ntp.conf
> server 127.127.8.0 mode 14
> fudge 127.127.8.0 time1 0.236
> tos orphan 5
> $ ntpq -p
>      remote           refid      st t when poll reach   delay   offset  jitter
> ==============================================================================
> *GENERIC(0)      .DCFa.           0 l  27h   64    0    0.000   -1.255   0.793
> 
> Shouldn't ntpd have figured out by now that the clock is gone (I
> unplugged it yesterday) and have switched into orphan mode?
> 
> -- 
> Christian "naddy" Weisgerber                          naddy@mips.inka.de

It seems like orphan mode is something that you'd run on an ensemble
of local machines to ensure that they continue to be synchronized with
each other because that's deemed important for some application need.
I don't understand why you'd go to all this trouble on a single host
to simulate NTP clock synchronization when in fact the local clock
isn't synchronized to anything?

If you're concerned about keeping the local clock synchronized to UTC
in the event that your NTP process is partitioned from it's external
NTP peers, then you don't need to do anything.  Once NTP has
determined what you're clock's intrinsic frequency error is, the local
clock model will contnue to apply this frequency correction even in
the absence of external peers providing offset/delay samples.  (This
is the magic number squirreled away in /var/db/ntp.drift, or wherever
it gets put these days.)

louie




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20090706233701.GA30824>