[ntp:questions] NTP.ORG MEINBERG KEEP TIME ACCURATE to 10MS

unruh unruh at invalid.ca
Wed Nov 14 16:34:37 UTC 2012


On 2012-11-14, David Woolley <david at ex.djwhome.demon.invalid> wrote:
> David Taylor wrote:
>
>> 
>> The secret has been to make a tight connection between the Windows PCs 
>> and the "best available" stratum-1 server by fixing the polling at 32 
>> seconds.  In your case, this may correspond to a line like:
>> 
>>  server 10.1.126.204 iburst minpoll 5 maxpoll 5
>> 
>
> Especially if the stratum one server isn't local, setting a low maxpoll 
> will cause the time to track propagation asymmetries in your network 
> connection.  You will get low offset, but relatively large time errors, 
> which could easily exceed the offsets.

The time will always track (suitably averaged) propagation assymetries.
There is nothing that ntp or ntpd can do about that. You might mean
short timescale assymetries. But a far more critical issue is that a
long timescale does not allow ntp to track oscillator rate changes due
to for example temperature changes. This I think is a far more worrysome
issue than propagation delays ( unless you are at the end of a busy adsl
connection).



More information about the questions mailing list