[ntp:questions] Why does ntp keep changing my conf file?

unruh unruh at wormhole.physics.ubc.ca
Mon Sep 20 15:20:52 UTC 2010


On 2010-09-17, David Woolley <david at ex.djwhome.demon.invalid> wrote:
> Daniel Havey wrote:
>
>> So, you think that a PC clock will drift 20-50ms in 5 seconds?  Seems like a lot, but whatever.  Let me see if I've got this right, you tell me I might get say synchronization of ~10ms with ntpd running on a lan with everybody on the same switch or perhaps one switch away?
>> 
>
> On its own no.  With the help of ntpdate, yes.  I wouldn't actually 
> expect to reach this level very often, but I'd certainly expect several 
> milliseconds.  You will get these large steps because you are using 
> individual measurements, which are very vulnerable to scheduling delays 
> and network propagation delays, which ntpd, itself, will average out (it 
> will effectively low pass filter the jitter out of the measurements).
>
> If you seriously need very tight synchronisation, you should use ntpd or 
> possibly chrony, with a pulse per second source parallel wired to all 
> the machines.  In extreme cases, you may need to make sure the 
> transmission lines from the PPS source are or equal length.

Since it is a foot per nanosecond, unless the machines are more than
about a third of KM apart, the lengths are not an issue. (if they are
that far apart, be careful in using proper cables and line drivers to
remove reflections). The computers will tend to need about 1usec to
respond to interrupts and read the system clock. 




More information about the questions mailing list