[ntp:questions] Severe time drift for Undisciplined Local Clock
Richard B. gilbert
rgilbert88 at comcast.net
Wed Apr 11 02:04:36 UTC 2007
deon at nulan.co.za wrote:
> Hi All
>
> Thanks for the tremendous response I have received. I have bowed to the
> collective wisdom and have located a Windows XP machine at each of our
> 200+ sites which we have configured to serve as a Time server. It is
> using the onboard RTC as the source.
>
> I still have a problem though.
>
> I can ping the server and Linux sees it but the time is not synched. I
> have run ntpq and the rv command indicates that the server is rejected.
> Why I cannot establish. The complete log files are attached. Anyone
> else which can offer a reason? BTW, have tried from 2 different servers
> and get the same results so either I am doing something wrong or XP has
> a problem. (The only problem I am aware of on XP is when it is used to
> synch to External Clock Sources).
>
>
> I am calling ntpd like this:
>
> ntpd -A -l /var/log/ntp/main.log
>
> My /etc/ntp.conf file has one line:
>
> server 192.168.222.100
>
> Any help appreciated!
>
> Deon
>
>
>
> ------------------------------------------------------------------------
>
> ntpq> pe
> remote refid st t when poll reach delay offset jitter
> ==============================================================================
> 192.168.222.100 .LOCL. 1 u 55 64 377 0.327 -418554 147.741
> ntpq> as
>
<snip>
Try starting ntpd with the -g option. That should set the time
unconditionally. The offset is HUGE. So is the jitter. How long did
you wait, after starting ntpd, before you took that ntpq display?
Normally you should wait at least thirty minutes.
Note that the display shows that NO server has been selected as the
synchronization source. IOW, you gave it only one source and ntpd
believes that that source is not suitable. The high value for jitter
might have something to do with that.
Using the "iburst" keyword should give you faster startup when you get
your other problems resolved; e.g.
server 192.168.222.100 iburst
More information about the questions
mailing list