Brian T. Brunner <[EMAIL PROTECTED]> wrote: >What I'm reading between the various posts is that the abusive clients >can't be expected to behave to any known rules, so feeding them good >time, bad time, or fixed time is equally unproductive... they continue to >hammer the server.
I think sending bad time is the best solution. My assumption is that a person installs a time setting program with the assumption that it will make the time more accurate. They then will completely ignore it unless: 1. They are contacted by someone telling them of a problem 2. Their computer ceases to keep good time. Since we can't count on (1) happening very often we should concentrate on (2), ie we should reply to abusive clients with the wrong time (when I say wrong the hour, day, month and year should all be wrong [1] ) so their time setting program will set their computer to the wrong time. They may notice this and discard the program. To avoid getting a bad reputation the writers of the program will improve it to ensure it doesn't get marked as abusive. >Solution: put the time servers behind a packet-dropping firewall, >as has been suggested by others, so I'm out of helpful ideas for this thread. That one is known not to work since they will just assume packet loss and retry even more. [1] Perhaps everyone agrees on a specific bad time to send or an offset of say 17220322 seconds (6 months, 18ish days, 7 hours, 25 minutes, 22 seconds, with a 2 day, day of week difference). -- Simon J. Lyall | Very Busy | Web: http://www.darkmere.gen.nz/ "Inside me Im Screaming, Nobody pays any attention." | eMT. "Those who sacrifice sound quality for hard disk space deserve neither." _______________________________________________ questions mailing list [email protected] https://lists.ntp.isc.org/mailman/listinfo/questions
