Eugen COCA wrote: > David, > > you probably started mtrg after the server was in good working > condition. That's why I said, if you are so kind, to reboot the > computer and see the initial start-up (after the initial STEP > correction). The same problem may appear when the GPS receiver will > lose the fix for several hours, or the room temperature changes with > several degrees. I do not know if we could obtain microsecond > precision in such extreme conditions. > > Are there other experiences already made ?
OK, I suppose I need to know how the system might behave after a reboot! You will see the results here during the course of the day. http://www.david-taylor.myby.co.uk/mrtg/daily_ntp.html Before the initial step, the offset from the internet ntp servers on that system was about 1400ms (1.4s). Immediately after the initial step, the server synced to an internet source, and showed the offset of the GPS/PPS source to be -7ms. After about 5 minutes, the GPS/PPS source had been selected as the sync source, and showed an offset of -0.007ms. I've added the results of running occasional ntpq -p over the first five minutes or so at the end of this Web page: http://www.david-taylor.myby.co.uk/ntp/FreeBSD-GPS-PPS.htm Thanks for the pointer to your own graphs. With only 1ms resolution, it's difficult to judge what is happening. You might be better to record microseconds instead of milliseconds. Whilst I do get a transient, it's microseconds, not milliseconds. Are you sure your GPS has a good view of the sky? Cheers, David _______________________________________________ questions mailing list [email protected] https://lists.ntp.isc.org/mailman/listinfo/questions
