NTP fails synchronization with server at startup
Ed Greshko
Ed.Greshko at greshko.com
Thu Dec 6 00:00:50 UTC 2007
Paul Smith wrote:
> On Dec 5, 2007 11:40 PM, Paul Smith <phhs80 at gmail.com> wrote:
>>>>>>> Another thing. Are you sure that ntp isn't doing it's stuff, even
>>>>>>> though the bootup shows a fail. Before you do the ntpd restart, run
>>>>>>> the following as user.
>>>>>>>
>>>>>>> /usr/sbin/ntpq
>>>>>>> then type pe, which will give you some info on which servers ntp is
>>>>>>> trying to connect to, and how successfull it is being. You can keep
>>>>>>> typing pe at intervals, which will show ntp's progress at reaching a
>>>>>>> point where a time server is being used as a "sys peer". The server
>>>>>>> being used will be prefixed by a "*". Other useable servers will be
>>>>>>> prefixed by a "+" "candidat". To quit ntpq type q.
>>>>>> Thanks, Nigel. In fact,
>>>>>> # /usr/sbin/ntpq
>>>>>> ntpq> pe
>>>>>> No association ID's returned
>>>>>> ntpq> pe
>>>>>> remote refid st t when poll reach delay offset
>>>>>> jitter
>>>>>> =======================================================================
>>>>>> ==== === clock-a.develoo 192.12.19.20 2 u 28 64 3 190.143
>>>>>> 438.261 9.311 ntpq>
>>>>> Well it appears to have a connection to the timeserver here, and often
>>>>> takes a while before the timeserver is accepted as a system peer. Then an
>>>>> "*" will appear before clock-a.develoo. Your reach is showing as 3, and
>>>>> will gradually progress until it reaches 377, but this can take some
>>>>> time.
>>>>>
>>>>>> i.e., when I run pe after a while, I get the above, but the first time
>>>>>> I run pe, I get
>>>>>>
>>>>>> 'No association ID's returned'
>>>>> That usually indicates that ntp cannot contact the timeserver, no network
>>>>> connection, or the timeserver is not accessable.
>>>>>
>>>>>> Can I be sure that ntp is running now and synchronizing with a ntp
>>>>>> server?
>>>>> It appears to be running, but I think you have a problem in only having
>>>>> one timeserver available.
>>>>>
>>>>>> Paul
>>>>> Paul. I'd still suggest that you add more timeservers to your
>>>>> /etc/ntp.conf. Try the 3 that I am using. I know they are not the closest
>>>>> to you, but they have been reliable for me. As I mentioned earlier, make
>>>>> sure that everything in /etc/ntp.conf is commented out, except the
>>>>> driftfile line, comment out also your present server, and add the ones
>>>>> I've listed below. Save the changes, restart the ntp daemon, and rerun
>>>>> /usr/sbin/ntpq. Type pe every minute or so, and see how it progresses.
>>>>>
>>>>> server ntp.obspm.fr
>>>>> server ntp.kamino.fr
>>>>> server ntp2.belbone.be
>>>>>
>>>>> Is this just the one machine you have connected to the Internet, or are
>>>>> you on a LAN with other machines that are also using ntp to get their
>>>>> time from Internet timeservers?
>>>> Thanks again, Nigel. Does it seem that it is working now?
>>>>
>>>> # /usr/sbin/ntpq
>>>> ntpq> pe
>>>> remote refid st t when poll reach delay offset
>>>> jitter
>>>> ===========================================================================
>>>> === *syrte8.obspm.fr 134.157.254.19 2 u 27 64 377 62.413 132.612
>>>> 16.037 +ns1.kamino.fr 193.52.184.106 2 u 20 64 377 85.748
>>>> 119.231 10.125 +ntp2.belbone.be 195.13.23.6 2 u 54 64 377
>>>> 69.566 104.344 12.046 ntpq>
>>>>
>>>> No, I am directly connected to the Internet, with no LAN in between.
>>>>
>>>> Paul
>>> That looks fine, and just what I'd expect to see.
>> I have just noticed that at booting, the Network Manager daemon is
>> loaded after the ntp one. This may be the cause of the problem.
>
> How can I configure my computer in order to have ntp being launched
> after the network manager daemon?
I told you that already.....
In /etc/rc5.d (assuming you are booting to run level 5) move S58ntpd to
SXXntpd where XX is higher than the network manager daemon's number.
There may be a GUI to do it....but I don't recall at the moment.
Ed
More information about the fedora-list
mailing list