[ntp:questions] NTP makes a time jump

unruh unruh at invalid.ca
Mon Jul 8 17:52:22 UTC 2013


On 2013-07-08, John Hasler <jhasler at newsguy.com> wrote:
> unruh writes:
>> lets say one computer has a natural offset of 200PPM and another is
>> -100PPM. When both are slewing at 500PPM, one is effectively slewing
>> at 300 PPM and the other at 600, which goes out by 50ms in about an
>> hour.
>
> Therefor you should tinker with the slew rate of the master, limiting it
> so that none of the slaves ever have to slew to track it.

?? Do you mean none of the slaves ever have to jump to track it? Since
all ntpd does ever is either jump or slew.
I do not know if you can tinker with the max slew rate. If the time is
really off by 5 min say, ntpd will (if you tell it never to jump) peg
its slew rate to 500PPM until it gets close. If your other machine's
natural drift rate is very different from that of the server, they could
have trouble keeping up. Now, if you could make the max slew rate of the
server smaller than the smallest difference between drift and 500PPM
then they could keep up. But how does one limit the slew rate of the
server to say 200PPM without recompiling?



More information about the questions mailing list