[ntp:questions] Keeping NTP Honest Redux

Unruh unruh-spam at physics.ubc.ca
Mon Nov 9 13:20:02 UTC 2009


David Woolley <david at ex.djwhome.demon.invalid> writes:

>David J Taylor wrote:

>>  http://www.satsignal.eu/mrtg/performance_narvik.php

>How much of that offset do you attribute to actual clock error?

>If most of it, that graph makes a strong case for a low maxpoll, or a 
>different algorithm.

a) This is windows.
b) Except for the abberation at 12, the offset seems to track the disk temp.
However, it is not the offset which is important but the rate. It would be good
to plot the rate as well (from loopstats) since it is the rate not the offset
that the temp changes. mind you, since that rate is ntp's idea of the rate and it
takes a while to change it ( which is why the offset is not always zero) this may
not track that well.




More information about the questions mailing list