[time-nuts] Does GPS time reception work everywhere all ofthetime?

[Michael Conlen]

Absolutely. I've found four to be best. That leaves you with three in the
event of a failure, but if you go to five or beyond often NTP will switch
from one time server to another which causes it to think the time has
radically changed, and the clock goes faster/slower/faster/slower for a bit.


There is the option of having a preferred server which will minimize this if
you happen to know of one that should be particularly good, then NTP will
generally stick to that one unless it has good reason to switch.

NTP servers switching it's sense of what the correct time is randomly is the
biggest headache with applications I've dealt with since the other servers
getting time from your stratum 2 server may not adjust their sense of the
correct time in anything representing unison.

Now if I could just get the datacenter to let me put an antenna on their
roof...

--
Mike

On Wed, Sep 1, 2010 at 12:21 PM, Bob Camp <lists at rtty.us> wrote:

> Hi
>
> Indeed, there's not much you can do to figure out which of the two is
> correct. This is well explained in the NTP doc's. My concern was that
> people
> drop back to 1 or 2 servers from 6 or 8 and hop from one issue to another.
>
> Bob
>
> -----Original Message-----
> From: time-nuts-bounces at febo.com [mailto:time-nuts-bounces at febo.com] On
> Behalf Of Michael Conlen
> Sent: Tuesday, August 31, 2010 9:46 AM
> To: Discussion of precise time and frequency measurement
> Subject: Re: [time-nuts] Does GPS time reception work everywhere all
> ofthetime?
>
> If it has one it has no choice but to accept that the one server has the
> right time, if it has two it has no way to tell which is most likely to be
> closest to correct. If it has three or more it can make value judgements.
>
> --
> Mike
>
> On Aug 31, 2010, at 9:12 AM, Bob Camp wrote:
>
> > Hi
> >
> > On the other hand, NTP gets bothered if it has only 2 independent
> partners.
> > It's quite happy with one, but two seems to bother it. Three or four
> seems
> > to be a "sweet spot".
> >
> > -------------------
> >
> > Still beating on the email problem... seems to be a router somewhere in
> Ohio
> > or Illinois that's the issue. May get fixed some day, until then no
> incoming
> > email.
> >
> > Bob
> >
> > =============
> > -----Original Message-----
> > From: time-nuts-bounces at febo.com [mailto:time-nuts-bounces at febo.com] On
> > Behalf Of Poul-Henning Kamp
> > Sent: Tuesday, August 31, 2010 4:58 AM
> > To: Discussion of precise time and frequency measurement
> > Subject: Re: [time-nuts] Does GPS time reception work everywhere all of
> > thetime?
> >
> > In message <4C7CC2F1.3060604 at orange.fr>, mike cook writes:
> >
> >> of references about how to do this in a belt and braces manner. How far
> > you go (5 independent clocks should be sufficient) depends on how much
> >> your downtime costs.
> >
> > Let me just add a bit of caution here:  The NTP software gets confused
> > if it has too many servers to select from.
> >
> > I would _never_ advice configuring more than 5 ntp servers for any
> > machine, unless some of them deliberately are made unavailable and
> > only used for monitoring/sanity-check.
> >
> > --
> > Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
> > phk at FreeBSD.ORG         | TCP/IP since RFC 956
> > FreeBSD committer       | BSD since 4.3-tahoe
> > Never attribute to malice what can adequately be explained by
> incompetence.
> >
> > _______________________________________________
> > time-nuts mailing list -- time-nuts at febo.com
> > To unsubscribe, go to
> > https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> > and follow the instructions there.
> >
> >
> >
> >
> > _______________________________________________
> > time-nuts mailing list -- time-nuts at febo.com
> > To unsubscribe, go to
> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> > and follow the instructions there.
>
>
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to
> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.
>
>
>
>
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to
> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.
>