[ntp:questions] Re: Question on abusive clients.

Richard B. Gilbert rgilbert88 at comcast.net
Thu Dec 29 20:23:49 UTC 2005

Brian T. Brunner wrote:

>After reading, I wondered: It appears that the timestamp field is 
>present in the KoD packet, would it be possible to track which IP has 
>been given the KoD, and the timestamp of when that was decided, 
>then  when another packet comes in, fill the timestamp field with the 
>time the KoD was decided?
>Cost: 64 bits per system that has been KoD'd, a reply to each KoD packet.
>Effect: system that has been KoD'd sees the time holding still, making
>the server a false-ticker to that client almost instantly.
>Assumption on my part: false-tickers get labeled as such by the client that 
>has concluded the fact, then that client subsequently leaves that server alone.
>Brian Brunner
>brian.t.brunner at gai-tronics.com

I believe that once a server is configured, it is polled whether or not 
it's a false ticker.  Its time will be evaluated each time it is polled 
and the time will be discarded as long as the server remains a false 
ticker.  If a former false ticker corrects itself, its time will be 
considered and, if the NTP algorithms so decide, it will be used.

Administrators who are paying attention will cease to use servers that 
offer poor quality time.

More information about the questions mailing list