[ntp:questions] Will ntpd tell me if it encounters problems?

Martin Burnicki martin.burnicki at meinberg.de
Mon Apr 16 12:46:11 UTC 2007


gluino wrote:
> I have some general ntp-newbie questions, that I guess I should ask
> here:
> In what unit is the "when" column of ntpq -p?
In seconds. It means the number of seconds after the last poll, and when
"when" reaches the nimber of "poll" then ntpd sends the next query to the
associated time source.

> Using meinberg's windows binaries, how can I tell whether my clock is
> being smoothly slewed as opposed to being stepped? (Does "time reset
> -0.140676s" in the log mean it changed in a single step? why didn't it
> slew?)

Time is normally slewed, unless the difference exceeds 128 milliseconds, in
which case it is stepped. And right, "time reset" in the log indicates the
time has been stepped.

Please keep in mind "the Meinberg binaries" are based on the standard NTP
sources which we have only compiled and put into a GUI installer in order
to simplify installation under Windows.

>>From what I understand, NTP continually estimates the rate error of my
> motherboard clock, writing to the driftfile, so that the correction
> can be continuously applied. Does this mean the red PPM line of
> monitor's stats graph converges to zero? (after-correction ppm) or
> does it converge to the ppm error of my hardware clock?

It converges to the frequency error of your motherboard's crystal, so it
indicates the frequency error you would observe if the frequency was _not_
actually corrected by ntpd.

Please note this value is basically always the same for a given board, but
may vary a bit with environmental temperature changes.

> What happens if the computer is powered-off for a few days, and then
> turned on after a few days?
> I'm guessing the drift correction does not work when the computer is
> off?

Of course there's no correction if the PC is off. the drift file just
preserves the last known value across reboots in order to speed up initial


Martin Burnicki

Meinberg Funkuhren
Bad Pyrmont

More information about the questions mailing list