[ntp:questions] Two servers, one client : one problem

Steve Kostecke kostecke at ntp.org
Tue May 3 17:41:19 UTC 2011


On 2011-05-03, Yann I. <yann.frm at gmail.com> wrote:

> I have some problems about the configuration of ntp servers and
> client. You'll find the configurations used for the test and some
> outputs (of "ntpq -p" for example). First of all, the both servers
> are standalone and are not synchronized with an external ntp server.

They need a common time reference. Or one needs to follow the other.

> Later, they'll be connected to external ntp server. Later...

Then you need to configure one to poll the other.

> The servers and client don't use iptable.

The ntpq peer billbords you've posted show that the ntpds are able to
poll each other (note the non-zero values in the reach, delay, and
offset columns).

> My problem is the following : I don't understand why the client can't
> be synchronized with one of the servers. The servers are seen as a
> "falsetickers"... The "ntpq associations" shows me a "reject" but I
> don't understand the reason...

[snip]

>   remote      refid   st t when poll reach  delay   offset jitter
>
>==================================================================
> xsrv01-cba   LOCAL(0)  5 u   10   64  377   0.135  267.274  4.308
> xsrv02-yil   LOCAL(0)  5 u   59   64  377   0.127  -56.638  0.427

The problem is that xsrv01 and xsrv02 are 323.885 milliseconds apart.

If you were using NTP as it was designed (i.e. with a common time
reference) this would not be an issue.

-- 
Steve Kostecke <kostecke at ntp.org>
NTP Public Services Project - http://support.ntp.org/




More information about the questions mailing list