[ntp:questions] Windows built-in SNTP/NTP clients

David L. Mills mills at udel.edu
Wed Aug 13 16:56:04 UTC 2008


Neither does Windows implement the mode-6 protocol nor does it conform 
to the basic protocol. See the reference implementation documentation 
about Windows issues. Also see the alternative workaround in ntp_proto.c.

As the author of rfc1305 I say you misquote me. The mode-6 control and 
monitoring protocol is an integral component of the specification; the 
mode-7 protocol is intended as propietary. In any case the mode-6 
protocol was defined and implmented well before SNMP.

An NTP MIB has been implemented by some manufacturers and another 
proposed by the NTP working group, but neither is supported by the 
reference implementation. Either MIB might be appropriate for management 
purposes, but for complete monitoring and performance evalutation the 
limitations of current SNMP clients make the mode-6 protocol necessary.


David J Taylor wrote:
> Ryan Malayter wrote:
>>On Tue, Aug 12, 2008 at 2:21 AM, David J Taylor
>><david-taylor at blueyonder.neither-this-bit.nor-this-bit.co.uk> wrote:
>>>I just checked Vista (Home Premium) and the built-in NTP does not
>>>respond to NTPQ requests.  Does the one in Windows 2003 or 2008?
>>>Can any implementation which fails to respond to NTPQ commands be
>>>called "decent"?
>>NTP mode 6 packets are an OPTIONAL part of the NTP specification in
>>RFC-1305, and RFC-1305 even goes so far as to suggest that mode 6
>>packets be used only when other management and monitoring tools are
>>unavailable. Microsoft provides other management tools (w32tm, group
>>policy, event viewer, etc.), although they are barely sufficient.
>>As far as I know, the reference implementation and its derivatives are
>>the *only* implementations supported by ntpq. Some of its functions
>>are even version-specific and not mentioned in any RFC, right?
>>So asking you question the other way is just as valid: "Can any
>>implementation which fails to respond to "w32tm" commands be called
>>"decent"? Sounds silly.
> Thanks for your input, Ryan.  I'm really thinking of the management 
> environment, where you have a single monitor system checking on a host of 
> different clients.  I guess the "correct" approach there would be if all 
> the clients responded to SNMP requests, rather than using a proprietary 
> protocol.  For the moment, though, I would want an NTP client which could 
> be monitored by NTPQ (although only the offset is of routine interest to 
> me).
> Cheers,
> David 

More information about the questions mailing list