[ntp:questions] Re: offset values from ntpq and ntptrace

Richard B. Gilbert rgilbert88 at comcast.net
Thu Mar 30 15:17:13 UTC 2006


Tanguy.ropitault at gmail.com wrote:

> Ok, so if I understand correctly, if I want to monitor the performance
> of NTP on a client (e.g its accuracy), I have to use the estimated
> errors field in statfile because It seems to be the same value given by
> ntptime? And is this value  the estimated error compared to the
> reference clock? or compared to the server'clock? I know it's quite the
> same but I want to know.
> And one thing is very odd, when you look ntp documentation, all
> monitoring for ntp are done with the offset value, so everyone is
> wrong? It's not usefull tu use offset value? And there is still a
> question that is confusing my mind... in some documentation I read that
> precision over a wan should be few ms... what does this mean compared
> to the estimated errors? is it the same... I know that accuracy and
> precision are different things but i don't really see... maybe someone
> have a good example???
> 

Precision is a measure of how finely your clock can slice time.  It's 
the smallest possible difference between two successive readings of the 
clock.  Precision is usually specified as a negative power of two.  On 
Sun Solaris you get a precision of -21 or 1 microsecond.  On VMS, it's 
-10 or ten milliseconds.

Accuracy is the measure of the difference between your clock and UTC. 
Accuracy can be a little difficult to determine. . . .  You would have 
to know the exact UTC time at some instant and compare that value with 
your clock. . . .




More information about the questions mailing list