[ntp:questions] Synchronizing contrived time

mike cook michael.cook at sfr.fr
Tue Mar 11 20:42:46 UTC 2014


I have been trying to reproduce this but don't get the same messages. From the source it looks like the local clock driver returned a stepped system counter  So for me, it looks like something is resetting the system clock outside ntpd. Could the server be running any other process which could do that? Is ntpd running on a multi cpu or multi core system, or more in line with what you are seeing, in a virtual machine?


Le 11 mars 2014 à 21:30, Amit Dor-Shifer a écrit :

> Hi William. The conf file appears in the post (I've cat-ed b4 executing
> ntpd), and is exactly what I was suggested here initially.
> Thanks,
> Amit
> On 12/03/2014 5:45 AM, "William Unruh" <unruh at invalid.ca> wrote:
> 
>> On 2014-03-11, Amit Dor-Shifer <amit.dor.shifer at gmail.com> wrote:
>>> Hi.
>>> 
>>> With the configuration I've been advised here, I'm still not able to
>>> maintaine a skewed time on the server. ntpd does execute, yet a few
>> seconds
>>> after its initialization, it resets the time skew I've generated (I.E,
>> even
>>> b4 the client had a chance to query it)
>> 
>> ??? It has nothing to "reset" with respect to. why do you not post your
>> whole ntp.conf file here for the server?
>> 
>> 
>>> 
>>> $ sudo tail -f /var/log/messages
>>> #...snip
>>> $
>>> $
>>> $ date
>>> Tue Mar 11 20:43:03 EST 2014
>>> $ cat ~/test_ntpd/ntp.conf
>>> server 127.127.1.0 minpoll 4 maxpoll 4
>>> $ sudo date -s "$(date) - 50 minute"
>>> Tue Mar 11 19:53:30 EST 2014
>>> $ date
>>> Tue Mar 11 19:53:35 EST 2014
>>> $ sudo ntpd -g -f ~/test_ntpd/ntp.conf
>>> sudo: timestamp too far in the future: Mar 11 20:43:30 2014
>>> 
>>> Mar 11 19:54:06  ntpd[31080]: ntpd 4.2.6p5 at 1.2349-o Sat Nov 23 18:21:48
>> UTC
>>> 2013 (1)
>>> $ Mar 11 19:54:06  ntpd[31081]: proto: precision = 1.495 usec
>>> Mar 11 19:54:06  ntpd[31081]: 0.0.0.0 c01d 0d kern kernel time sync
>> enabled
>>> Mar 11 19:54:06  ntpd[31081]: format error frequency file
>>> /home/amit/test_ntpd/ntp.conf
>>> Mar 11 19:54:06  ntpd[31081]: Listen and drop on 0 v4wildcard 0.0.0.0 UDP
>>> 123
>>> Mar 11 19:54:06  ntpd[31081]: Listen and drop on 1 v6wildcard :: UDP 123
>>> Mar 11 19:54:06  ntpd[31081]: Listen normally on 2 lo 127.0.0.1 UDP 123
>>> Mar 11 19:54:06  ntpd[31081]: Listen normally on 3 eth0 10.250.129.79 UDP
>>> 123
>>> Mar 11 19:54:06  ntpd[31081]: Listen normally on 4 lo ::1 UDP 123
>>> #...snip
>>> Mar 11 19:54:06  ntpd[31081]: peers refreshed
>>> Mar 11 19:54:06  ntpd[31081]: Listening on routing socket on fd #22 for
>>> interface updates
>>> 
>>> $
>>> $
>>> $
>>> $ date
>>> Tue Mar 11 19:54:09 EST 2014
>>> $
>>> $
>>> $
>>> $
>>> $ Mar 11 19:54:16  ntpd[31081]: 0.0.0.0 c016 06 restart
>>> Mar 11 19:54:16  ntpd[31081]: 0.0.0.0 c012 02 freq_set kernel -25.992 PPM
>>> 
>>> $
>>> $ date
>>> Tue Mar 11 19:54:21 EST 2014
>>> $ Mar 11 20:44:23  ntpd[31081]: 0.0.0.0 c61c 0c clock_step +3000.023162 s
>>> Mar 11 20:44:23  ntpd[31081]: 0.0.0.0 c614 04 freq_mode
>>> Mar 11 20:44:24  ntpd[31081]: 0.0.0.0 c618 08 no_sys_peer
>>> 
>>> $
>>> $
>>> $
>>> $ date
>>> Tue Mar 11 20:44:29 EST 2014
>>> 
>>> David, when you were suggesting before that I didn't configure the
>> correct
>>> local clock driver - is that achieved via ntp.conf? Could you refer me to
>>> an example?
>>> Thanks
>>> 
>>> 
>>> On Tue, Mar 11, 2014 at 6:12 AM, William Unruh <unruh at invalid.ca> wrote:
>>> 
>>>> On 2014-03-10, Amit Dor-Shifer <amit.dor.shifer at gmail.com> wrote:
>>>>> Thanks.
>>>>> Would this then be the reference implementation?
>>>>> 
>>>>> [user at centos test_ntpd]$ rpm -qf $(which ntpd)
>>>>> ntp-4.2.6p5-1.el6.centos.x86_64
>>>>> [user at centos test_ntpd]$ cat ~/test_ntpd/ntp.conf
>>>>> server 127.127.1.0 minpoll 4 maxpoll 4
>>>>> [user at centos test_ntpd]$ ps -ef|grep ntp
>>>>> user     13411 12236  0 22:03 pts/0    00:00:00 grep ntp
>>>>> 
>>>>> # moving clock back some 50 mins
>>>> 
>>>> AAAAAARGH. No do not do that. ntpd is horrible at handlking sudden
>>>> changes in time. It is like testing how good a car is by driving it off
>>>> a cliff. It was never designed to fly.
>>>> 
>>>> 
>>>>> 
>>>> 
>>>> _______________________________________________
>>>> questions mailing list
>>>> questions at lists.ntp.org
>>>> http://lists.ntp.org/listinfo/questions
>>>> 
>> 
>> _______________________________________________
>> questions mailing list
>> questions at lists.ntp.org
>> http://lists.ntp.org/listinfo/questions
>> 
> _______________________________________________
> questions mailing list
> questions at lists.ntp.org
> http://lists.ntp.org/listinfo/questions


More information about the questions mailing list