[ntp:questions] Re: is there a way to "lock" the drift frequency

Terje Mathisen terje.mathisen at hda.hydro.com
Mon Nov 17 07:25:33 UTC 2003


wayne wrote:

> In <bp8i08$f3t$1 at osl016lin.hda.hydro.com> Terje Mathisen <terje.mathisen at hda.hydro.com> writes:
> 
> 
>>It would seem possible to patch & recompile to only allow some
>>relatively small range around the initial ntp.drift value.
> 
> 
> It is my understanding that there are a fair amount of subtle reasons
> for the way ntp does the things it does.  So, would such a patch be a
> good idea?

No, you would definitely blow up quite often, in the form of 
clock-stepping when regular slew operations should suffice.

I.e., I'm _not_ reccomending that you actually do this!

OTOH, you did claim that your clock _in reality_ was quite stable, 
varying very little around your normal ntp.drift value, right?

Assuming this is correct, you could indeed make a "broken" ntpd 
specifically for your machine, which clamped the slew rate.

Terje

-- 
- <Terje.Mathisen at hda.hydro.com>
"almost all programming can be viewed as an exercise in caching"




More information about the questions mailing list