[ntp:questions] Why does ntp keep changing my conf file?

unruh unruh at wormhole.physics.ubc.ca
Thu Sep 23 11:59:50 UTC 2010


On 2010-09-22, Daniel Havey <dhavey at yahoo.com> wrote:
> I had a feeling it would come down to this.  Now the mystery has been solved.  The ntpdate script is much faster than ntpd.  That is why it is recommended for use rather than ntpd.  I agree with them.  A day or even 10 hours is totally unacceptable.  This is an emulab like testbed.  We bring nodes into experiments, wipe them out, re-configure, and reboot frequently.  Even 10 minutes, would be extremely annoying.  So if I reboot my nodes 6 times in a day then I must spend an hour waiting for ntpd to do it's voodoo?

ntp is for disciplining a local clock so that it keeps the best time and
time rate. Your solution has the local clock running in a sawtooth
fashion-- advancing at a different rate than 1 sec / sec and then
jumping. Those jumps are predictable and noisy. If that is OK for you,
go ahead and use ntpdate. Just do not let your machines out into the
wild. 
If you want your machines to keep good time over long periods use ntpd.
If not, then do what you want. 
(Note as I said if you run linux or bsd, chrony is far fater at
converging to the right rate and time and also disciplines the clock
much better than does ntp. But if you are running windows that is
irrelevant. )


 
>
> My UCSB webserver is where I would post a graph if I wanted to put a link here to the graph (which I don't anymore).
>
> --- On Wed, 9/22/10, unruh <unruh at wormhole.physics.ubc.ca> wrote:
>
>> From: unruh <unruh at wormhole.physics.ubc.ca>
>> Subject: Re: [ntp:questions] Why does ntp keep changing my conf file?
>> To: questions at lists.ntp.org
>> Date: Wednesday, September 22, 2010, 3:02 PM
>> On 2010-09-22, Daniel Havey <dhavey at yahoo.com>
>> wrote:
>> > I guess I can't show the graph, since I cannot reach
>> my UCSB webserver from here.
>> 
>> What does your web server have to do with it?
>> 
>> >
>> > [dhavey at node1 ~/MANET]$ ntpq -npc rv
>> >? ? ? remote? ? ? ?
>> ???refid? ? ? st t when poll
>> reach???delay???offset?
>> jitter
>> >==============================================================================
>> > *192.168.1.10? ?
>> 130.221.24.100???2
>> u???23? 256? 377? ?
>> 0.168???17.224???1.515
>> > assID=0 status=0644 leap_none, sync_ntp, 4 events,
>> event_peer/strat_chg,
>> > version="ntpd 4.2.2p1 at 1.1570-o Sat Dec 19 00:58:16 UTC
>> 2009 (1)",
>> > processor="i686", system="Linux/2.6.18-164.11.1.el5",
>> leap=00,
>> > stratum=3, precision=-20, rootdelay=4.211,
>> rootdispersion=84.326,
>> > peer=53715, refid=192.168.1.10,
>> > reftime=d044e85b.b36d9436? Wed, Sep 22 2010
>> 13:40:59.700, poll=8,
>> > clock=d044eb73.1a72e7e7? Wed, Sep 22 2010
>> 13:54:11.103, state=4,
>> > offset=17.224, frequency=135.852, jitter=1.638,
>> noise=7.224,
>> > stability=6.210, tai=0
>> >
>> > It's going the wrong way.? It should be reducing
>> the offset and it is making it worse.
>> >
>> 
>> No, it should not. The offset will increase because it has
>> the wrong
>> value for the frequency drift. Eventually it will converge
>> to the right
>> value and the offset will then decreease until you have
>> offsets of about
>> .1ms.
>> NOte that if you used chrony, the process would take about
>> 10 min. With
>> ntp it will take 5-10 hours. It is a feature.
>> 
>> 
>> 
>> >
>> >
>> > --- On Wed, 9/22/10, Chuck Swiger <cswiger at mac.com>
>> wrote:
>> >
>> >> From: Chuck Swiger <cswiger at mac.com>
>> >> Subject: Re: [ntp:questions] Why does ntp keep
>> changing my conf file?
>> >> To: dhavey at yahoo.com
>> >> Cc: questions at lists.ntp.org
>> >> Date: Wednesday, September 22, 2010, 1:36 PM
>> >> On Sep 22, 2010, at 1:18 PM, Daniel
>> >> Havey wrote:
>> >> > The bloody thing made it worse!? Look at the
>> >> graph.? It was doing just fine at 0.002147 second
>> >> offset, then I started ntpd and look what it
>> does.?
>> >> This is definitely not very good for a testbed.
>> >> 
>> >> What graph?? You'll need to mail out a link,
>> since
>> >> most mailing lists eat attachments (they've got to
>> be fed
>> >> somehow, I guess)....
>> >> 
>> >> Also, ntpq output like the following is helpful
>> for
>> >> debugging issues:
>> >> 
>> >> % ntpq -npc rv
>> >> ? ???remote? ? ? ?
>> >> ???refid? ? ? st t when poll
>> >> reach???delay???offset?
>> >> jitter
>> >>
>> ==============================================================================
>> >> -66.250.45.2? ???192.5.41.40?
>> >> ? ? 2 u???22? 512?
>> >> 377???26.389? ?
>> >> 7.667???0.498
>> >> *18.26.4.105? ???.GPS.? ?
>> >> ? ? ? ? 1 u???54?
>> >> 512?
>> >> 377???24.070???-0.925???0.458
>> >> +128.59.59.177???128.59.39.48?
>> >> ???2 u???42? 512?
>> >> 377???26.898? ?
>> >> 0.408???0.033
>> >> +17.254.0.27? ???17.72.133.54?
>> >> ???2 u???48? 512?
>> >> 377???54.681? ?
>> >> 0.587???0.106
>> >>? 24.103.228.242? 18.26.4.105? ? ? 2
>> >> u???37? 512?
>> >> 377???33.338???-2.185???0.726
>> >> associd=0 status=0644 leap_none, sync_ntp, 4
>> events,
>> >> freq_mode,
>> >> version="ntpd 4.2.4p5-a Mon Sep 20 14:30:32 EDT
>> 2010 (1)",
>> >> processor="i386", system="FreeBSD/6.4-STABLE",
>> leap=00,
>> >> stratum=2,
>> >> precision=-19, rootdelay=24.070,
>> rootdispersion=26.037,
>> >> peer=43124,
>> >> refid=18.26.4.105,
>> >> reftime=d044e485.9adbf554? Wed, Sep 22 2010
>> >> 16:24:37.604, poll=9,
>> >> clock=d044e6bd.18405690? Wed, Sep 22 2010
>> >> 16:34:05.094, state=4,
>> >> offset=-0.239, frequency=19.470, jitter=1.043,
>> >> noise=0.380,
>> >> stability=0.019, tai=0
>> >> 
>> >> Regards,
>> >> -- 
>> >> -Chuck
>> >> 
>> >> 
>> >
>> >
>> >? ? ???=
>> 
>> _______________________________________________
>> questions mailing list
>> questions at lists.ntp.org
>> http://lists.ntp.org/listinfo/questions
>> 
>
>
>       =




More information about the questions mailing list