[ntp:questions] Re: How to handle leap second condition correctly?

David L. Mills mills at udel.edu
Sat Jun 11 21:24:01 UTC 2005


Harlan,

See page 10 of the draft RFC. It clearly states the stratum in the 
client packet should be set to zero, not fifteen. Reivew rfc1305 for the 
mapping between packet stratum and the internal value used in the NTP 
reference implementations.

I was very careful in stating the expectations for servers and clients. 
If you have trouble with the text, please honk.

Dave

Harlan Stenn wrote:

> Lemme try part of this again.
> 
> 
>>Are you possibly trying to say that an SNTP client should not also act
>>as a server?
> 
> 
> Yes, and I will change that to MUST NOT.
> 
> 
>>Whilst true, this is only a SHOULD NOT, not a MUST NOT,
>>and Windows violates this rule anyway, which means it has become 
>>unenforceable (ignoring that Windows has more serious protocol
>>violations).
> 
> 
> By my read of the spec, SNTP is allowed 2 modes of operation:
> 
> - at the "leaf" level with no direct refclock connection
> - at stratum 1 with a direct refclock connection
> 
> sntp is incredibly lightweight and will believe the first answer it
> receives.
> 
> The spec also says that one "side" cannot (is unable to) determine if
> the other side is SNTP or NTP.
> 
> Therefore, if one has an sntp leaf node, I believe that leaf node should?
> must? advertise itself at stratum 15.
> 
> I also believe that the SNTP (draft) RFCs need some work in this area, as
> we are having this discussion.
> 
> H



More information about the questions mailing list