tech-userlevel archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: ntpd loose 10s

Emmanuel Dreyfus <> writes:

> From time to time, nagios raise me an alert for a xen DOMU 
> (NetBSD 6.0/i386, Xen 4.1.3) because time suddently shifted of 10s.

How is that measured, and could it be wrong (long asymmetric delay on
measurement, perhaps)?

> ntpd still considers itself in sync (see status below). And the

ntpd will only declare itself out of sync after it has enough data.  if
the clock was behaving, it may be polling at 1024s interval.

> problem resolves itself as it appeared. Anyone has an idea of what 
> can be going on?

I don't follow 'resolves itself'.

> # ntpdc 
> ntpdc> kerninfo
> pll offset:           -0.00603435 s
> pll frequency:        -103.546 ppm
> maximum error:        3.79263 s
> estimated error:      0.0141 s
> status:               6001  pll nano mode=fll
> pll time constant:    10
> precision:            1e-09 s
> frequency tolerance:  496 ppm

> # date; rdate -p
> Thu Jan 10 10:10:17 CET 2013
> Thu Jan 10 10:10:05 2013

so that's showing it 12 seconds ahead.

What happens if you just let this be?  Does the time go back to right,
with or without an ntp log message?

I would also suggest writing something that logs time/offset and then
graph that.

Attachment: pgpyGqSNW5wnF.pgp
Description: PGP signature

Home | Main Index | Thread Index | Old Index