[ntp:questions] Re: NTP sync problems
David L. Mills
mills at udel.edu
Thu Dec 16 14:46:53 UTC 2004
There might be some misconception here. The mitigation algorithm
establishes a bound on the offset of any server with respect to
roundtrip delay and dispersion. This is a matter of physics and theory.
When multiple servers are involved, the algorithm computes an
intersection interval within which the server clocks can be considered
valid. A possibly correct time must be contained in this intersection.
If the intersection is empty, a correct time cannot be established. This
is independent of the stratum.
Terje Mathisen wrote:
> Brad Knowles wrote:
>> At 12:43 PM -0600 2004-12-15, Robert Rati wrote:
>>> This is what spurred my original question which
>>> was, shouldn't NTP sync to the server(s) with the lower stratum
>>> regardless of how different the time is that is being served from
>>> higher stratum servers?
>> Because this is not the way the NTP algorithms were designed to
>> work. If you want to use just one time server to which everyone
>> syncs, that should work -- until such time as the primary goes down or
>> is unreachable.
> If you then wants a secondary, higher-stratum, server as backup, said
> server should be configured to also be a client of the primary server.
> That way they will agree as long as both are up, and if the primary goes
> away, the secondary server will start out with the same time, drop down
> a few stratum levels, but still server time.
More information about the questions