[ntp:questions] Windows - Seven Days Later

Jerry Baker jerry at novalid.invalid
Thu Oct 7 22:06:05 UTC 2004


W. D. wrote:
> At 15:30 10/7/2004, Jerry Baker, wrote:
>>I have made several postings about how NTP is having a real problem 
>>remaining stable. I now have a little more than a weeks worth of data 
>>graphed with MRTG available. Perhaps I am expecting too much from NTP, 
>>but it looks to me like there is something critically wrong with NTP's 
>>operation on the Windows platform (SNTP clients with no stepping keep 
>>better time).
>>
>>Take a look at http://jerbaker.dhs.org/ntp/
> 
> Hey Jerry,
> 
> I am willing to take your word that you are having problems.
> 
> I just got a box working running NTP.  It is keeping time 
> extremely accurately.  I am running FreeBSD on an older
> computer.  It's possible to get microsecond accuracy on an
> old 486, if you have a GPS antenna.  The operating 
> system is important.  I recommend FreeBSD.  
> 
> I am putting together a How To to install NTP on FreeBSD.
> It will be ready in a day or so.  Let me know if you are
> interested.
> 
> So, I recommend you grab an old 486, load it with FreeBSD
> (http://www.US-Webmasters.com/FreeBSD/Install/) and install
> NTP.  

I would really prefer to just fix whatever it is that is broken in NTP. 
I am more interested in tracking down this issue than I am in the 
accuracy of the time itself. There is obviously something haywire with 
NTP on Windows, and I would like to get to the bottom of it. I do not 
have the expertise on my own though, which is why I am posting here. My 
suspicion now is that there is something about thread scheduling on 
Windows that does not work quite the way someone expected when they 
ported the code. NTP appears to be *extremely* sensitive to CPU usage 
(even just editing and saving text files).



More information about the questions mailing list