[ntpwg] KISS codes

[Danny Mayer]

anthony.flavin at bt.com wrote:
> Danny,
> 
> I don't see how this would work. Especially in the case of SNTP. The
> problem you site is where the sheer number of devices polling the server
> swamped it. I believe that at the moment an NTP server only stores the
> past 600 which is unlikely to be enough to determine who the trouble
> makers are, so unless you are proposing to store the last several thousand
> requests, how are you going to determine who to send the codes to?
> 

I disagree. If the code already knows enough to send out any of these
KISS codes then you have the right to expect the clients to examine the
KISS codes. The fact that the reference implementation may only keep 600
past client addresses is a red herring. The fact that you get back KISS
codes in the first place says that it's doable. One such use might be to
set up an address blacklist which doesn't require existing client
information.

However, I'm not referring to the implementation, I'm referring to the
protocol. If you receive a KISS code, you should examine it to see what
it says and take action based on what it tells you.

Danny