[ntp:questions] Keeping NTP Honest

Unruh unruh-spam at physics.ubc.ca
Sun Jul 12 20:28:24 UTC 2009


"Richard B. Gilbert" <rgilbert88 at comcast.net> writes:

>Evandro Menezes wrote:
>> On Jul 10, 2:40 pm, Unruh <unruh-s... at physics.ubc.ca> wrote:
>>> And you have at least a 1/5 chance that IT is the bad server. What do
>>> you do then?
>> 
>> Well, bar a false ticker, when it's ignored, it would keep NTP in
>> shorter poll periods.  IOW, "honest".
>> 
>> I wonder though if the right thing would be to configure 6 servers
>> with half of them limited to 64s polling...
>> 
>> Thanks.

>Four, five, or seven are the "magic" numbers which allow you to survive 
>the failure of one, two, or three servers respectively.  I suppose these 
>numbers could be extended to cover the failures of still more servers 
>but at some point you've got to say "this is ridiculous" and stop.

>Failure, in this context, can mean anything from not responding to 
>responding with the wrong year; e.g. suddenly it's 2023.  Such things 
>should not happen but they do; the last NTP Survey found a server 
>responding with the wrong year!

>Tinkering with MINPOLL and MAXPOLL is generally a bad idea.  Ntpd will 
>adjust its polling interval to the optimum value for the conditions then 
>obtaining.

Optimal? Optimizing what?




More information about the questions mailing list