[ntp:questions] Re: NTP for dummies

David L. Mills mills at udel.edu
Sat Oct 1 20:43:04 UTC 2005


I didn't write the documentation or rfc1305 or NTP project page or the 
book (in press) for dummies and dummies shouldn't read them. I wrote 
those documents to reveal how and why the contraption works and how to 
manipulate every little thing in excruciating detail. Folks who do read 
them are certainly not dummies.

Your tone suggests I offended you and likely other dummies with dreary 
boring prose. Guilty as charged. Please, please, don't read those 
documents. Read the really fine user-friendly documents at www.ntp.org 
written for real people, including the faq, twitchy and related helpful 
tips. Better not peek at the NTP project page either; that infruriates 
dummies even more. Above all, dummies should not read the book. Read the 
following instead:

Rybaczyk, J. Expert Network Time Protocol: an experience in time. 
Springer-Verlag, New York, 2005, 153 pp.


Richard B. Gilbert wrote:
> Cam wrote:
>> Hello list,
>> I've been alternately reading the NTP documentation and banging my 
>> head on the desk, so please excuse any intemperance and/or incoherence.
>> I'm glad my car's manual wasn't written by the same people who wrote 
>> the NTP documentation or it would start with an exhaustive treatment 
>> of the carnot cycle and then roar off into advanced thermodynamics 
>> when all I wanted to know was how often to change the d**n oil. Pauses 
>> to bang head on desk several more times. Apparently the idea of 
>> starting with simple examples ("hello world") and working up to 
>> complex examples ("program to prove the four-color theorem") didn't 
>> occur to them; they want to prove the theorem right away.
>> But I digress. I have a computer, lets call it MASTER, which has time 
>> of day that I'm happy with. I have a bunch of other computers, all on 
>> the same subnet, and I want them to set their clocks to match MASTER. 
>> That's it. Sounds like the making of a real simple example. I believe 
>> this can be done because the NTP pages make reference to an 
>> "Undisciplined Local Clock" but as with all the documentation it 
>> assumes you are already an expert so no simple example is given. Whack 
>> whack.
>> So, finally, the question is: Does anybody have a link to a web page 
>> that gives some simple examples (eg "to sync from machine do 
>> this", "to setup a local undiciplined server do that")? If so it would 
>> be greatly appreciated.
>> Thanks
>> Cam Farnell
>> ps I've already R'd the F'ing M or at least made a serious attempt at 
>> it. I don't want to know every arcane detail of NTP in the known 
>> universe; I want to set up a *really* simple system.
>> _______________________________________________
>> questions mailing list
>> questions at lists.ntp.isc.org
>> https://lists.ntp.isc.org/mailman/listinfo/questions
> Install ntpd on all computers involved.  How is left as an exercise for 
> the student.
> On MASTER, create /etc/ntp.conf.  It should contain, at a minimum:
> #
> # Declare the local clock to be the clock of last resort.
> # It will be used to serve time in the absence of any other.
> #
> server              # Local clock, unit 0
> fudge stratum 10
> On the clients, create /etc/ntp.conf containing:
> server MASTER
> YOU will be responsible for keeping the time on MASTER correct!  
> MASTER's clock WILL drift.  It may drift badly.  YOU must check the time 
> periodically and adjust MASTER's clock.  The clients should stay in 
> synch with MASTER.  If, after a couple of months of neglect, you find 
> that all the clocks are 27 minutes slow, you have no one to blame but 
> yourself.  YOU HAVE BEEN WARNED!!!!  If you have to make large 
> corrections  the clients will take a long time to catch up.  If you have 
> to make a correction greater than 1024 seconds (about 17 minutes) all 
> the ntpd on all the clients will panic and exit.
> The above is a bare bones (well gnawed) configuration that no right 
> thinking person would tolerate but it should work.
> Far better would be to have MASTER served by either four internet 
> servers, or a hardware reference clock (GPS timing receiver, WWV 
> receiver, WWVB receiver, CHU receiver, etc.)  That should not only keep 
> everybody synched up but also provide time correct to within 10 
> milliseconds or better.

More information about the questions mailing list