[ntp:questions] ***SPAM*** Re: Monitoring NTP - nptq -p

Danny Mayer mayer at ntp.isc.org
Mon Mar 26 11:57:49 UTC 2007


Spoon wrote:
> Richard B. gilbert wrote:
>> Spoon wrote:
>>> Keith E. Brandt, M.D. wrote:
>>>
>>>> 'Poll' is the polling interval in seconds, which seems to vary quite 
>>>> a bit on my system, so here's the first real question - how does ntp 
>>>> pick the polling interval (a pointer to the docs is fine, I just 
>>>> haven't been able to uncover it yet).
>>> Have you played with minpoll and maxpoll?
>>>
>>> http://www.eecis.udel.edu/~mills/ntp/html/confopt.html
>>>
>>> minpoll minpoll
>>> maxpoll maxpoll
>>>
>>>     These options specify the minimum and maximum poll intervals for 
>>> NTP messages, in seconds as a power of two. The maximum poll interval 
>>> defaults to 10 (1,024 s), but can be increased by the maxpoll option 
>>> to an upper limit of 17 (36.4 h). The minimum poll interval defaults 
>>> to 6 (64 s), but can be decreased by the minpoll option to a lower 
>>> limit of 4 (16 s). These option are valid only with the server and 
>>> peer commands.
>>>
>>> I don't know under what conditions ntpd changes the polling interval.
>> Playing with MINPOLL and MAXPOLL is NOT recommended!!
>>
>> Ntpd varies the polling interval between MINPOLL and MAXPOLL in order to 
>> best respond to existing conditions.  The math is over my head but it 
>> has something to do with the "Allan Intercept" q.v.
>>
>> My oversimplified English explanation is that short polling intervals 
>> allow large errors to be corrected quickly while the longest polling 
>> interval allows small errors to be corrected very accurately.
> 
> I would have thought that short polling intervals are always better,
> ignoring traffic overhead issues:
> 
> If the current "correct" interval should have been e.g. 64 seconds
> instead of 16 seconds, just ignore 3 out of 4 replies.
> 
> Where is the flaw in my logic?

While you might think so at first glance in practice it isn't so. Dave
Mills has spent many years researching this and if you really want to
understand why this isn't so you should read his papers.

If you understand some physics you might think of the clock oscillating
back and forth much like a spring does when stretched between two points
and set to oscillate. Then think about how you might get the spring to
oscillate at just one frequency. This stuff is harder than you think and
Dave has the work to prove that your suggestion is the wrong answer.

Danny



More information about the questions mailing list