Frans Grotepass wrote:

> Is it possible to change the ntp.conf to include similar functionality to the
> -g option?
> _______________________________________________
> questions mailing list
> questions@lists.ntp.isc.org
> https://lists.ntp.isc.org/mailman/listinfo/questions
>

It's almost certainly possible! Why do you believe that doing so would
be a wise use of scarce resources?

On Tuesday 04 July 2006 14:05, Richard B. Gilbert wrote:
> Frans Grotepass wrote:
> > Is it possible to change the ntp.conf to include similar functionality to
> > the -g option?
> > _______________________________________________
> > questions mailing list
> > questions@lists.ntp.isc.org
> > https://lists.ntp.isc.org/mailman/listinfo/questions
>
> It's almost certainly possible!

That is what I thought

> Why do you believe that doing so would
> be a wise use of scarce resources?

I don't understand your question. The -g option will make the clock skip to
the server time if it is off by a large amount. We build embedded systems
that are shipped around the world. They include an ntp client that will
synchronize with a server. If the internal clock is way off, NTP simply dies
if -g is not used. With -g the startup does one jump and from there,
continues with clockskew. How does this hamper my resources?

Best regards,
_______________________________________________
questions mailing list
questions@lists.ntp.isc.org
https://lists.ntp.isc.org/mailman/listinfo/questions

Frans Grotepass wrote:

> On Tuesday 04 July 2006 14:05, Richard B. Gilbert wrote:
>
>>Frans Grotepass wrote:
>>
>>>Is it possible to change the ntp.conf to include similar functionality to
>>>the -g option?
>>>_______________________________________________
>>>questions mailing list
>>>questions@lists.ntp.isc.org
>>>https://lists.ntp.isc.org/mailman/listinfo/questions
>>
>>It's almost certainly possible!
>
>
> That is what I thought
>
>
>>Why do you believe that doing so would
>>be a wise use of scarce resources?
>
>
> I don't understand your question. The -g option will make the clock skip to
> the server time if it is off by a large amount. We build embedded systems
> that are shipped around the world. They include an ntp client that will
> synchronize with a server. If the internal clock is way off, NTP simply dies
> if -g is not used. With -g the startup does one jump and from there,
> continues with clockskew. How does this hamper my resources?
>
> Best regards,
> _______________________________________________
> questions mailing list
> questions@lists.ntp.isc.org
> https://lists.ntp.isc.org/mailman/listinfo/questions
>

It uses the developer's time! They are all unpaid volunteers, working
in their spare time. While you might find such a feature useful, most
would not.

If you want to start ntpd with the -g option and lack a CLI to do it
with, you could patch the code yourself. I believe it would be far
simpler to make the code believe it had found a -g in the startup
command than to make it parse ntp.conf to find the option.

Frans Grotepass wrote:
> On Tuesday 04 July 2006 14:05, Richard B. Gilbert wrote:
>> Frans Grotepass wrote:
>> Why do you believe that doing so would
>> be a wise use of scarce resources?

> continues with clockskew. How does this hamper my resources?

By "scarce resources" I believe he meant "scare programming resources on
the NTP project" - as in "Why do we have to make this change, test this
change, document this change, and support this change, when you could
just pass -g as one of the elements of argv[] you pass to the main()
routine of the embedded NTP client and be done with it."

I've done that sort of thing myself on embedded projects - just pass the
options you need on the fake command line you pass to the embedded program.
_______________________________________________
questions mailing list
questions@lists.ntp.isc.org
https://lists.ntp.isc.org/mailman/listinfo/questions

See the miscopt.html page - if you really want to do what you think you want
to do, see the 'tinker panic' section.

It might be better overall to use some sort of "nanny" script that watches
ntpd and restarts it if needed, along with producing a log entry.

There are other (usually very rare) reasons why ntpd could exit, and a nanny
script will catch those cases, too.

HB

Frans Grotepass wrote:
> On Tuesday 04 July 2006 14:05, Richard B. Gilbert wrote:
>
>> Why do you believe that doing so would
>> be a wise use of scarce resources?
>
> I don't understand your question. The -g option will make the clock skip to
> the server time if it is off by a large amount. We build embedded systems
> that are shipped around the world. They include an ntp client that will
> synchronize with a server. If the internal clock is way off, NTP simply dies
> if -g is not used. With -g the startup does one jump and from there,
> continues with clockskew. How does this hamper my resources?
>

The quick way to deal with this is to change the cmd_args.c to set it on
even without the -g switch. Or do you still want it to be optional?

Danny
_______________________________________________
questions mailing list
questions@lists.ntp.isc.org
https://lists.ntp.isc.org/mailman/listinfo/questions

Danny wrote:

> The quick way to deal with this is to change the cmd_args.c to set it on
> even without the -g switch. Or do you still want it to be optional?

Why go to that much trouble and why recommend a solution that "hides" the
behavior? I already posted the answer, which is to look at whatever html
page is in the docs that says how to do this trivally, with the current
resources, in a way which does not "hide" the behavior in an undocumented
code change.

H