[ntp:questions] Clock skew changes drastically between reboots

Richard B. gilbert rgilbert88 at comcast.net
Mon Apr 2 13:00:50 UTC 2007


Spoon wrote:
> Harlan Stenn wrote:
> 
>> Consider plotting the value.  If you are really looking at at 12 hour 
>> cycle,
>> it could be related to room temp (or CPU temp) based on AM/PM stuff.
> 
> 
> I don't think it is caused by temperature.
> 
> Consider this experiment:
> 
> I started ntpd Friday afternoon.
> Here is the tail of the log file.
> One line per hour.
> Line 1: Apr 1 00:00:54
> Line N: Apr 2 10:00:54
> 
> ntpd: offset 0.001188 sec freq 6.789 ppm error 0.000358 poll 10
> ntpd: offset 0.002901 sec freq 6.790 ppm error 0.000755 poll 10
> ntpd: offset 0.004828 sec freq 7.384 ppm error 0.000051 poll 10
> ntpd: offset 0.003921 sec freq 7.864 ppm error 0.000879 poll 10
> ntpd: offset -0.001503 sec freq 7.991 ppm error 0.001083 poll 10
> ntpd: offset -0.003009 sec freq 7.991 ppm error 0.001505 poll 10
> ntpd: offset -0.007434 sec freq 6.531 ppm error 0.000336 poll 10
> ntpd: offset -0.004766 sec freq 5.755 ppm error 0.001196 poll 10
> ntpd: offset 0.002509 sec freq 5.756 ppm error 0.000872 poll 10
> ntpd: offset 0.005521 sec freq 6.205 ppm error 0.000692 poll 10
> ntpd: offset 0.006017 sec freq 6.944 ppm error 0.001040 poll 10
> ntpd: offset 0.005880 sec freq 6.948 ppm error 0.000194 poll 10
> ntpd: offset 0.004049 sec freq 8.087 ppm error 0.000968 poll 10
> ntpd: offset -0.000532 sec freq 8.087 ppm error 0.001370 poll 10
> ntpd: offset -0.003859 sec freq 7.231 ppm error 0.000423 poll 10
> ntpd: offset -0.003713 sec freq 6.777 ppm error 0.000697 poll 10
> ntpd: offset -0.001831 sec freq 6.775 ppm error 0.001047 poll 10
> ntpd: offset 0.000496 sec freq 6.657 ppm error 0.001292 poll 10
> ntpd: offset 0.001386 sec freq 6.658 ppm error 0.000338 poll 10
> ntpd: offset 0.002866 sec freq 7.010 ppm error 0.000635 poll 10
> ntpd: offset 0.002626 sec freq 7.012 ppm error 0.000285 poll 10
> ntpd: offset 0.001566 sec freq 7.014 ppm error 0.000034 poll 10
> ntpd: offset 0.001261 sec freq 7.015 ppm error 0.000430 poll 10
> ntpd: offset 0.001342 sec freq 7.016 ppm error 0.000604 poll 10
> ntpd: offset 0.001609 sec freq 7.017 ppm error 0.000133 poll 10
> ntpd: offset 0.001987 sec freq 7.145 ppm error 0.000445 poll 10
> ntpd: offset 0.002616 sec freq 7.467 ppm error 0.000466 poll 10
> ntpd: offset 0.001823 sec freq 7.689 ppm error 0.000438 poll 10
> ntpd: offset -0.001790 sec freq 7.754 ppm error 0.000954 poll 10
> ntpd: offset -0.003357 sec freq 7.751 ppm error 0.000725 poll 10
> ntpd: offset -0.004273 sec freq 7.749 ppm error 0.000485 poll 10
> ntpd: offset -0.003576 sec freq 7.119 ppm error 0.000606 poll 9
> ntpd: offset -0.002520 sec freq 7.113 ppm error 0.000399 poll 9
> ntpd: offset -0.001714 sec freq 7.105 ppm error 0.000060 poll 9
> ntpd: offset -0.001317 sec freq 6.830 ppm error 0.000306 poll 10
> 
> The frequency offset varies between 5.7 and 8.1 ppm
> 
> I rebooted the computer at 10:38:26
> 
> I removed the drift file and started ntpd.
> Here's what it reports now:
> 20 lines per hour
> Line 1: Apr 2 10:46:33
> Line N: Apr 2 12:01:33
> 
> ntpd: offset 0.000000 sec freq 0.000 ppm error 0.005181 poll 6
> ntpd: offset 0.000000 sec freq 0.000 ppm error 0.020572 poll 6
> ntpd: offset 0.000000 sec freq 0.000 ppm error 0.009341 poll 6
> ntpd: offset 0.032652 sec freq 38.008 ppm error 0.005221 poll 6
> ntpd: offset 0.032652 sec freq 38.008 ppm error 0.005221 poll 6
> ntpd: offset 0.023276 sec freq 39.111 ppm error 0.005418 poll 6
> ntpd: offset 0.010924 sec freq 39.194 ppm error 0.006991 poll 6
> ntpd: offset 0.007166 sec freq 39.329 ppm error 0.004073 poll 7
> ntpd: offset 0.007166 sec freq 39.329 ppm error 0.004073 poll 7
> ntpd: offset 0.007166 sec freq 39.329 ppm error 0.004073 poll 7
> ntpd: offset 0.001360 sec freq 39.348 ppm error 0.003408 poll 7
> ntpd: offset -0.000522 sec freq 39.344 ppm error 0.002619 poll 7
> ntpd: offset -0.000522 sec freq 39.344 ppm error 0.002619 poll 7
> ntpd: offset -0.000522 sec freq 39.344 ppm error 0.002619 poll 7
> ntpd: offset -0.002840 sec freq 39.307 ppm error 0.001394 poll 7
> ntpd: offset -0.002840 sec freq 39.307 ppm error 0.001394 poll 7
> 
> The frequency offset varies between 38 and 39.4 ppm
> 
> In other words, the frequency offset of my system clock has changed by 
> more than 30 ppm across reboots!
> 
> Temperature has remained approximately constant in the server room.
> 
> I am thoroughly confused :-(

Do you really need to reboot?  Microsoft is the only vendor of Operating 
Systems that seem to require regular reboots.  Most, if not all, of the 
others seem to be able to run for weeks, months or years between reboots.




More information about the questions mailing list