NTP problem - Clock too fast for NTP to keep up?

[jdow]

From: "Gene Heskett" <gene.heskett at verizon.net>

> On Thursday 10 February 2005 20:49, jdow wrote:
> >From: "Gene Heskett" <gene.heskett at verizon.net>
> >
> >> On Thursday 10 February 2005 17:47, Peter Kiem wrote:
> >> >>>There is also a cpuspeed ticked on in ntsysv so could that be
> >> >>> an issue?
> >> >>
> >> >> Its a possibility, can you turn it off?  OTOH, I'd certainly
> >> >> think that if it was being called, it would be shown to be
> >> >> loaded according to an lsmod.  But I'd still turn it off and
> >> >> reboot just to check.
> >> >
> >> >Disabled it and rebooted but it didn't make a difference :(
> >>
> >> Thats what I was araid of.  JoAnne talked me into turning off the
> >> apic in my kernel, and handheld me into getting ntp to run (I
> >> think) and I've set my tickadj back to the default of 10000, but I
> >> think thats going to have to come down, its a couple of seconds
> >> ahead of my watch in about 20 minutes.
> >>
> >> Do you have apic on in your kernel build?
> >
> >Give it some time, Gene. It seems to be far enough off it may take
> >a day or so to get down to 2ms type errors. (And if you have ntpdate
> >running please be nice to yourself and turn it off. It will screw
> >up the feedback loop ntp uses to average out bad data.)
> 
> It is, and I assume the same caveat holds for fooling with tickadj 
> and/or its server list?

It probably would not do anything meaningful. A virtual machine is not
really all that perfect an analog of a true piece of hardware.

{^_^}