[ntp:questions] ntpd wedged again

A C agcarver+ntp at acarver.net
Tue Feb 7 19:31:58 UTC 2012


On 2/7/2012 11:19, Ron Frazier (NTP) wrote:
> Hi A C,
>
> I will admit that I've only skimmed your NTPD losing sync thread
> briefly. I just wanted to share this loopstats file where my USB (no
> PPS) GPS occasionally goes crazy and has offsets almost an order of
> magnitude larger than normal. IE, normal offsets are in the 15 ms range.
> Then, they jump up to around 100 ms. Usually, I stop NTPD, unplug and
> replug the USB plug for the GPS, and restart NTPD, which fixes the
> problem. However, coincidentally, this time, I let it run. After a
> number of hours, the problem fixed itself. It's very strange. I don't
> know if my problem and yours are related in any way or not. I'm running
> Windows, with the main NTP programs from Dave Hart's site:
>
> ntp-4.2.7p249-win-x86-bin
>
> Here's a link to the loopstats file:
>
> http://dl.dropbox.com/u/9879631/loopstats.20120206p08-1-part%20crazy-above-priority

As soon as I get the C99-snprintf fixed I'll start collecting data again 
and post it if I see something go haywire.  In this case I was able to 
verify that the GPS itself was stable (o-scope on PPS).  I was ignoring 
the GPS data itself and using only PPS with the time coming from other 
servers.  All of the offsets went crazy at the same time, both on the 
PPS and on any network server.  These offsets were several seconds in 
size and they ramped up and down like the clock was spinning out of 
control. The clock would eventually be out of sync by more than a 
minute.  If the snprintf fix wasn't compiled in this time that might 
explain why, but I'm not sure yet.  I'll find out soon,


More information about the questions mailing list