[ntp:questions] Loopstats updated less frequently than expected

David Mills mills at udel.edu
Sun May 24 15:44:46 UTC 2009


David,

As it says in very many places, the clock filter algorithm selects the 
minium delay sample of the last eight samples. Studies back to the early 
1980s show these are by far the best ones. Sometimes, but not always, 
this results in seven out of eight samples being discarded. The 
discipline loop paramters are specifically selected to oversample by a 
factor of two relative to the Nyquist rate, even if seven out of eitght 
samples are lost. Now, would you rather the algorithm use more samples 
even if they are not optimal?

Dave

David J Taylor wrote:

>Folks,
>
>I have the following entries in ntpd.conf:
>
>server 192.168.0.2  iburst maxpoll 6
>server 192.168.0.7  maxpoll 6
>server 0.uk.pool.ntp.org minpoll 10
>
>.. and the following as the most recent loopstats:
>
>54974 37901.029 -0.000209922 13.840 0.000108471 0.007233 6
>54974 38229.033 0.000346890 13.846 0.000221473 0.007179 6
>54974 38754.041 -0.000217131 13.840 0.000287548 0.007132 6
>54974 38954.047 -0.000091306 13.839 0.000272631 0.006682 6
>
>I note that the times between these four entries are:
>
>328s
>525s
>200s
>
>As I have maxpoll set to 6, I would have expected entries no more than 64s 
>apart.  What am I failing to understand here?  The poll values reported in 
>ntpq -p are 64, 64 and 1024 - as expected.
>
>Thanks,
>David 
>
>_______________________________________________
>questions mailing list
>questions at lists.ntp.org
>https://lists.ntp.org/mailman/listinfo/questions
>  
>




More information about the questions mailing list