> I'm seeing this in my logs about 30 times a day, from what I gather  =

> from a little googling this is not normal behavior. The system in  =

> question is a GX260 (ICH4 i865) using TSC for time counters. Should I=
  =

> make like what others have reported and switch to i8254 and erase the=
  =

> drift file? I have one stratum 1 and 4 stratum 2 servers in my ntp  =

> conf file.
>
> Jul  9 04:02:09 pylon ntpd[8624]: kernel time sync enabled 6001
> Jul  9 04:53:22 pylon ntpd[8624]: kernel time sync enabled 2001
> Jul  9 06:01:37 pylon ntpd[8624]: kernel time sync enabled 6001
> Jul  9 06:35:45 pylon ntpd[8624]: kernel time sync enabled 2001
> Jul  9 07:26:58 pylon ntpd[8624]: kernel time sync enabled 6001
> Jul  9 07:44:02 pylon ntpd[8624]: kernel time sync enabled 2001
> Jul  9 08:18:13 pylon ntpd[8624]: kernel time sync enabled 6001
> ...

This is your ntpd going from

  status:               2001  pll nano
to
  status:               6001  pll nano mode=3Dfll

as shown by "ntpdc -c kern".  In other words, it's not able to
maintain the frequency-locked loop mode over time.  I suspect the
reason might be that either the readings from your primary
synchronization source or perhaps more precise, those readings
compared to the local clock wobble around a bit.

As to whether this is normal, well, I'm seeing it on some of my
systems as well...  Previously, before timecounters, I seem to
remember that I've only seen ntpd get to "pll" mode.

Regards,

- H=E5vard