[ntp:questions] Using both Server and Peer in ntp.conf

Richard B. gilbert rgilbert88 at comcast.net
Thu Feb 8 22:10:57 UTC 2007


Uwe Klein wrote:
> Hal Murray wrote:
> 
>>> How many minutes, hours, or days of "holdover" with reasonably 
>>> correct time you may get depends on the quality of the local clock, 
>>> the stability of the temperature, the phase of the moon and the whims 
>>> of the gods!
>>
>>
>>
>> We can at least do a bit of arithmetic.
>>
>> There are 86400 seconds in a day.  If your clock is off
>> by 11 ppm, it will drift 1 second per day.
>>
>> If your machine is in an airconditioned machine room, the
>> average temperature is probably pretty stable.  I'd guess
>> a second per week would be conservative.
>>
>> If your machine is in an environment with poor temperature
>> regulation, the day to day average temperature is probably
>> the important consideration.
>>
> Yes it works that way, I have systems ( embedded, linux + RT )
> running in a reasonably climatised room that have to do exactly
> 18000 sample cycles per 30min interval. When I have fixed
> the timer value to make it fit I can leave it alone for quite
> some time : weeks, month,..
> 
> But I get an imediate error when the gas tank for the heating is
> empty ;-)
> 
> To return to my question:
> Does ntpd follow this deterioration in any way?
> 
> uwe
> 

What do you mean by "follow"?

Ntpd will synchronize with a time source if a suitable source is 
available.  If none is available the most recent frequency correction 
value continues to be used. If the internal temperature of the machine 
remains constant, the clock should remain accurate for many hours.  If 
the temperature changes the clock will drift.




More information about the questions mailing list