[prev in list] [next in list] [prev in thread] [next in thread] 

List:       timekeepers
Subject:    [time] New semi-abusive client type noticed over the last day
From:       shoppa () trailing-edge ! com (Tim Shoppa)
Date:       2005-01-12 0:04:27
Message-ID: 41E45B62.nailGZ511O5IB () mini-me ! trailing-edge ! com
[Download RAW message or body]

> OK I see how the code works
> Between 2 and 8 samples received gets aggressive timing (5 second
> polling) then it backs off to a value between 30 and 300 seconds based
> on the offset (1/2 second = 30 second polling, 50 ms or less = 300 
> second polling).

Rather, why the code *doesn't* work.  That has to be the most ass-backward
polling algorithm I've ever seen.  If there's a lot of network delay,
(introducing a large offset, especially if the delay is assymetrical),
then it hammers even harder on that particular host.  I was a little
nonchalant about openntp before, but now I see: It is the ntp client
that is Most Evil.

Tim.

[prev in list] [next in list] [prev in thread] [next in thread]