[ntp:questions] Very rapid polling

gary.limanapier at elisys.co.uk gary.limanapier at elisys.co.uk
Sun Feb 15 09:49:29 UTC 2009


On Feb 15, 3:06 am, Unruh <unruh-s... at physics.ubc.ca> wrote:
> Steve Kostecke <koste... at ntp.org> writes:
> >On 2009-02-14, Unruh <unruh-s... at physics.ubc.ca> wrote:
> >> gary.limanap... at elisys.co.uk writes:
>
> >>>The only way we can stop it is to disable W32Time process on the
> >>>Tablets, but this then leaves them unsynchronised. Still searching for
> >>>a solution?
>
> >> So this definitely sounds like a bug of some kind in the Windows
> >> server's ntp code. Are they running the ntpd reference implimentation
> >> or what ntp software are they running?
> >w32time ... see the quoted text from the OP.
>
> It is w32 time on the tablets, but they are not the problem, the server is.
> It is the server, he said, which pumps out  packets trying to change the
> polling interval of the clients  every 300ms. Thus it
> is the server's program which one needs to look into. At least that was how
> I read the post.
> Note that one could surely install ntpd onto the tablets instead of
> W32Time.
> to keep them synchronized.

We are running W32Time on all machines including the servers. The
server itself obtains its own time independantly via a hardware clock.
This is used to update the CMOS clock from which W32Time then serves
out to the Clients. I will take up your suggestion and look into using
ntpd to see if there is any difference.




More information about the questions mailing list