[ntp:questions] NTP servers redundancy

unruh unruh at wormhole.physics.ubc.ca
Wed Jan 20 21:48:42 UTC 2010


On 2010-01-20, Brian Utterback <brian.utterback at sun.com> wrote:
> Rob wrote:
>
>> This is not relevant when the two servers return the same time within
>> the margin that one considers acceptable.
>
> You are incorrect. It is relevant. In practice, no two servers EVER
> have exactly the same time. And each time NTP decides to switch to the
> other server, it is also going to adjust the clock frequency to
> compensate for the perceived drift, even though the clock frequency
> may have already be correct.  It is thus likely for clock hopping to
> cause a feedback loop that will take the system clock outside of the
> interval between the two clocks. With the correct timing on the
> hopping, it can go very far outside the interval.

A problem with the ntpd protocol.

>
> Brian Utterback




More information about the questions mailing list