[ntp:questions] Tighter regulation?

unruh unruh at invalid.ca
Mon May 20 23:01:39 UTC 2013


On 2013-05-20, unruh <unruh at invalid.ca> wrote:
> On 2013-05-20, David Woolley <david at ex.djwhome.demon.invalid> wrote:
>> Mischanko, Edward T wrote:
>>
>>> Does anyone know what setting can be changed that will cause
>>> tighter regulation of the offset.  My goal is to get clk_wander to
>>
>> Why do you want to track network propagation delay changes, at the 
>> expense of accurate time keeping?
>>
>>> equal as close to zero as possible more often.  I would also like
>>> to see the frequency adjusted with every change in offset data;
>>> it currently does not appear to do that; it seems to be random.
>>
>> I think you are seeing the effects of the 8 stage minimum delay filter. 
>>   Defeating this is likely to increase jitter against true time, as it 
>> will cause the frequency to be adjusted based on low quality measurement 
>> samples, increasing the variability of the frequency.
>
> Well if the prop delays were wildly varying, the network jitter can be
> small, and ntpd will still throw out 7 of 8.

The grammer of that sentence is impossible. What I meant was:
Well, that is correct if the propagation delays were varying wildly.
However, the network jitter can be small, and ntpd will still throw out
about 7 of 8 (the lest delay of the last 8 measurements), wasting a
precious resource-- measurements. 

>
>>
>> Note that the time constants for the frequency adjustment loops are 
>> significantly larger than 8 times the sample interval.



More information about the questions mailing list