On 12/12/2011 5:22 PM, Joe Wulf wrote:
>>> On 11/29/2011 1:42 PM, Pete Ashdown wrote:
>>>> Is there anything I can do to decrease the convergence time?
>>> 
>>> Little or nothing! NTPD can, and sometimes does, take ten hours
>>> to
reach "steady state". It needs about thirty minutes to find a
>>> reasonable facsimile of the correct time. For the next nine
>>> hours
and thirty minutes, it will refine that value until it's as good as it's
>>> going to get.
> 
> An IT tech in my world would be fired, probably on the spot, for
giving the above answer as justification for why getting a system
operational takes so long (and we don't have near enough people as it is).
> 

This is a mischaracterization of what's going on. ntpd is able to get a
relatively accurate time within seconds of starting up. Exactly how
quickly depends on a variety of factors. Everything after those few
seconds is to bring the clock frequency to a steady state where both
jitter and delay is as small as ntpd can make it. ntpd goes through a
nunmber of different stages:
1) get the time right;
2) adjust the frequency of the clock so that it stays on time;
3) make smaller adjustments as it refines the frequency even more;
4) a steady state where adjustments are infrequent and small;

Pete was wrong to claim that it takes 30 minutes to get through step 1.
Usually it in the order of 4-7 seconds as long as you use iburst on the
server or pool lines in the config file. If you don't do that it takes a
lot longer. Steve Kostecke has some numbers on how long this takes.
It is the last stage that can take a long time and it can get perturbed
by things like the CPU heating up because you are running backups, heavy
CPU load, the network drops, heavy network traffic etc.

I describe the time to reach stage 4 as the "settling time". It's how
long it takes to reach what is more or less the steady state. It's not
currently something that is currently directly measured though it could
be.The stages before that are geared toward bringing the clock to this
final stage but the clock is already pretty accurate at that point.
Think of it like a spring, move one end too far and it will bounced back
in the other direction rather quickly. Too little and it will take a
long time to get to where it needs to be. The real gory details are in
papers that you can find at http://www.ntp.org/.

Danny
_______________________________________________
questions mailing list
[email protected]
http://lists.ntp.org/listinfo/questions

Reply via email to