[ntp:questions] poll interval - Clarification! - RFC compliance question

Steve Kostecke kostecke at ntp.org
Mon Jun 16 20:28:03 UTC 2008


On 2008-06-16, Unruh <unruh-spam at physics.ubc.ca> wrote:

> anton.a.persson at ericsson.com (Anton Persson A) writes:
>
>>We are NOT implementing the NTP protocol, we are using the reference
>>implementation straight out of <unnamed commercial embedded Linux
>>distribution>. I thought that was clear from the syntax of my
>>configuration and the fact that we use ntpq/ntpdc... ;-)
>
> The distros may change ntp. Ie, they are usually NOT the reference
> implimentation.

It is not uncommon for OSes to patch software to resolve build issues
(e.g. fix compilation problems, adjust for non-standard tool-chains,
customize file locations, etc.) 

> I think that on a lost lock, ntp first decreases the poll interval, on
> the theory that the loss was temporary, and that the next time it will
> be there. If that does not work, then the poll interval is increased.
> But I think the min and max poll are still honoured. Ie with a maxpoll
> of 10 it does not wait for 8 days (poll 16) before trying again.

You may wish to read the pertinent portion of the documentation for the
OP's version of NTP:

http://doc.ntp.org/4.2.4/ntpd.html#poll

-- 
Steve Kostecke <kostecke at ntp.org>
NTP Public Services Project - http://support.ntp.org/




More information about the questions mailing list