[ntp:questions] ntpd

Dave Hart hart at ntp.org
Wed Aug 3 10:12:41 UTC 2011

On Wed, Aug 3, 2011 at 08:32 UTC, Valera <valera.sigaev at developonbox.ru> wrote:
> No. It's real pc. But as I clearly understand sntp-client shows other
> offset:
> vvs at sigaev:~$ ~/work/ntp-4.2.6p3/sntp/sntp
>       2 Aug 19:59:25 sntp[18295]: Started sntp 2011-08-02 19:59:25.052161
> (-0300) +0.315752 ± 0.018951 secs

That's the same 315 millisecond offset (sntp is reporting seconds,
where ntpq is reporting milliseconds.  Clearly, ntpd 4.2.6p3 is not
working correctly for, or it would be able to bring the offset much
lower, typically lower than the delay to the source (18 milliseconds

> > > Btw: After that I downloaded development version 4.2.7... and this
> > > version also synchronizes pc fine.

> > We are close to releasing 4.2.6p4, please consider repeating your test
> > with 4.2.6p4-RC1.  There's still time to fix bugs in it before 4.2.6p4
> > is released.

> I'll be waiting.

I'm afraid we're having a problem communicating.  If you wait for
4.2.6p4 to be released, we will have missed the opportunity to fix the
problem you are seeing with 4.2.6p3.  You write that ntpd 4.2.7 is
able to synchronize the clock well (I presume to an offset much lower
than 315 msec).  You show ntpq and sntp output indicating ntpd 4.2.6p3
is synchronized, but has far too large an offset.  I am asking you to
try 4.2.6p4-RC1, to see if it also has problems, because if it does, I
would like to try to fix them before 4.2.6p4 is released.  You can
download it from:




I'm happy to hear 4.2.7 is working well for you, but I want our
upcoming 4.2.6p4 to work comparably well.

Dave Hart

More information about the questions mailing list