[ntp:questions] Re: Are these Windows XP clients beyond hope?

Neil Trotter neil_trotter at hotmail.com
Mon Sep 19 04:52:57 UTC 2005


In article <432E39F0.3040303 at ntp.isc.org>, Danny Mayer said:

> Neil Trotter wrote:
> > 
> > 4. List "nearest" servers first.
> > 
> > (To be truthful, I may have made #4 up, as it just seems to make sense -
> > but I *think* I remember reading this).
> 
> It makes no difference. That's not how the selection algorithms work. It 
> will look at all of them initially and then drop servers that don't make 
> the cut.

Thanks for clarifying.

> > I administer 2 small networks, both consisting of a mix of Fedora Linux
> > servers and Windows 2K/XP workstations, all x86 hardware.
> > 
> > The Linux servers poll internet time sources only.  Windows clients run
> > Heiko's build 1354, and poll local linux servers and the internet.
> > Windows clocks are not intended to be polled on our networks (that's
> > just my rule).
> > 
> > All machines are keeping good time except 2 of the windows XP clients at
> > one of the sites, and these are drifting wildly.  Since all the config
> > files are similar, and 2 other XP clients on the same network (with
> > identical config files) are staying sync'd, is this just a case of
> > hopelessly bad hardware clocks, or is there something else I can try?
> 
> I suspect that there's something wrong with those XP clients. I develop 
> on XP and have seen no such issues. It's not the code, but if any 
> multimedia applications are being run you may be losing interrupts and 
> that causes problems with the clock.

The machines are idle - 97% "system idle process", 3% vnc (me), +around
30 regular processes/services which register 0% cpu.  So no apparent
reason to be losing interrupts.

Thanks, Danny.


-- 
Neil
Alternative Network Directory
http://www.and-world.com
http://forum.and-world.com




More information about the questions mailing list