Now after just an hour the drift is -30, and all offsets are near zero.
Same server, same network, what happened??
You stated tht the drift file said 500.0 before. Maybe that is what the
designers of your distro stuck in there. If they did it is complete
incompetence. ntp would almost certainly start to adjust the rate to
bring it to greater than 500 and at that point ntpd goes nutty. Since
you erased the drift file, ntpd has to try to adjust the frequency.
Now, it may also be that if you have cpu frequency shifting for power
reasons, that ntpd naturally came to the conclusion at some point that
the drift was over 500PPM causing your problems.
No, ntp.drift is not fixed. If I delete it it's recreated by ntpd after
a while.
I'm monitoring ntp.drift on this server. It's now around 62, changing
slightly every now and then. I will definitely check the CPU stepping,
hopefully that would explain why it was 500 before.
Another server, which gave frequency error before, is now synced after
restarting the service. But ntp.drift is 450, which I guess shows
something wrong with the clock / CPU stepping and it's likely to fail
again soon. In fact, the offset has raised over the last few hours.
Antonio
_______________________________________________
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions