[ntp:questions] Re: Slew mode check for XNTP

David Woolley david at djwhome.demon.co.uk
Wed Jul 26 22:45:20 UTC 2006


In article <a8e7dcda0607260549j3687bf39t25719705cd15a28c at mail.gmail.com>,
sasi05 at gmail.com (Sasikumar Jayaraman) wrote:

> I am running XNTP version 4.2.0 as per ntpdc -c version output.

There's no such version.  However the other information suggests
that you are actually running version 4.2.0 of *ntpd*.

> C:\>ntpdc -c version

This gives the version of the ntpdc program.  To get the version of ntpd,
you would need to use ntpq to issue an rv 0 command.

> C:\>ntpdate -B ntp01-ams

ntpdate is not ntpd.

> Looking for host ntp01-ams and service ntp
> host found : ntp01-ams.lab.net
> 
> The -b option is required by ntpdate on Windows NT platforms

I believe this is because Unix has a system call which says slew the
clock until you have accumulated this many ticks of correction, whereas
there is no equivalent system call on Windows.  Windows does have a
system call to change the effective clock frequency, which is available
to nptd, and, in fact must be used when it is not in error recovery 
mode, but is not suitable for ntpdate, as it would mean that ntpdate
couldn't exit until the slew completed.

> Also I tried changing the registry setting to include  -x and -g option.

For ntpd version 4.2.0, -x increases the step threshold from 128ms to
10 minutes.  If your time error was more than 10 minutes, you would still
get a step.

                    case 'x':
                        clock_max = 600;
                        break;

You would need to use a tinker command to completely disable stepping.

> Can I conclude that slew mode operation is not available for Windows version
> of  XNTP?

No, and given that stepping is the special case of clock adjustment, I don't
think a more detailed examination of the NT specific parts of the code will
indicate any special handling for NT, but without the time to read all
the code, or an NT to try it on, I can't be 100% sure.




More information about the questions mailing list