Re: [Ntp] Roman Danyliw's No Objection on draft-ietf-ntp-chronos-20: (with COMMENT)

[Neta R S <[e-mail]>]

Hi Roman,

I added the architectural and operational details you requested in the
draft.
However, I can upload it only on July 22nd, when the submission tool will
be opened.

Many thanks,
Neta



On Tue, Jul 18, 2023, 8:40 AM Roman Danyliw <[e-mail]> wrote:

> Hi!
>
>
>
> I’ve really appreciated all of the clarifications that have been provided
> in this thread to explain the interaction between Khronos and ntpd.  My
> recommendation and substance of my key COMMENT points is that these
> architectural and operational details should be added to the draft.
>
>
>
> Roman
>
>
>
> *From:* Neta R S <[e-mail]>
> *Sent:* Monday, July 17, 2023 11:12 PM
> *To:* Danny Mayer <[e-mail]>
> *Cc:* Roman Danyliw <[e-mail]>; The IESG <[e-mail]>;
> [e-mail]; [e-mail]; NTP WG <[e-mail]>;
> Dieter Sibold <[e-mail]>; Karen O'Donoghue <
> [e-mail]>
> *Subject:* Re: [Ntp] Roman Danyliw's No Objection on
> draft-ietf-ntp-chronos-20: (with COMMENT)
>
>
>
> Hi,
>
>
>
> Fixing the time can be done by Khronos slowly in small steps by calling
> the clock discipline routine like ntpd.
>
> During this time the ntpd can be stopped (or ignored from the OS side).
>
>
>
> I'm sure that in NTF, with your help, we will find the best way.
>
>
>
> Best,
>
> Neta
>
>
>
> On Sun, Jul 16, 2023, 1:19 PM Danny Mayer <[e-mail]> wrote:
>
>
>
> On 7/14/23 11:09 PM, Neta R S wrote:
>
> Hi Roman,
>
>
>
> Thanks for your feedback.
>
> Please see me reply below (in blue).
>
>
>
> Thanks,
>
> Neta
>
>
>
> On Wed, Jul 12, 2023 at 3:22 PM Roman Danyliw via Datatracker <
> [e-mail]> wrote:
>
>
>
> -- Per Warren’s ballot, when “Khronos” detects malfeasance, how does it
> interact with OS or what new behavior does it introduce to a NTP client?
>
> Khronos is an algorithm designed by this draft to be implemented alongside
> any given NTP client (such as NTPv4). Khronos implementation should be
> capable of monitoring the local time in a machine and if needed fixing it.
> We leave this as implementation detail whether any specific implementation
> adaptations are required to pair Khronos with a given NTP client in
> different operating systems.
>
>
>
> I don't see how Khronos can fix the local time without adversely affecting
> the NTP client running on the system. Changing the local time is likely to
> cause it to exit at a minimum and possibly have other bad side effects to
> not just the NTP client but also other applications running on that system.
>
> Danny
>
>