[ntp:questions] Re: logs or stats showing when NTP made changes to the local clock

Joseph Harvell jharvell at dogpad.net
Tue Sep 19 22:53:50 UTC 2006


David:

Thanks for clarifying this.  I guess I should say I am *only* interested
in step corrections, not the ongoing slew corrections you are describing.

David Woolley wrote:
> In article <t_WPg.3489$6S3.2474 at newssvr25.news.prodigy.net>,
> Joseph Harvell <jharvell at dogpad.net> wrote:
> 
>> Subject: logs or stats showing when NTP made changes to the local clock
> 
> I suspect a common misunderstanding here.  On a better platform, the 
> combination of ntpd and the kernel update the clock frequency on every
> timer tick.  Even with ntpd version 3, running without kernel support, 
> the tweaking was done every four seconds.
> 
> If things are running smoothly, the parameters that control this are 
> updated every time a packet is returned from an eligible server.
> 
>> Can ntpd be made to log this?  Can I infer it from the stats?  I am
>> particularly interested in step corrections.
> 
> Whilst the error recovery steps get included in the system log, recording
> the fine tuning would put an excessive load on the system.
> 
> (The common misunderstanding is that ntpd periodically corrects the time,
> when it actually continuously corrects the frequency.)




More information about the questions mailing list