[ntp:questions] Re: ntpdate steps 86400 seconds

Tom Smith smith at cag.zko.hp.com
Fri Sep 30 12:47:08 UTC 2005


Tom Smith wrote:
> K.Dinh.P wrote:
> 
>> "John Pettitt" <jpp at cloudview.com> wrote in message
>>
>>> K.Dinh.P wrote:
>>>
>>>> I have a question concerning ntpdate. I synchronize to a timeserver 
>>>> using
>>>> crontab with ntpdate v4.1.1. The syslog is shown below:
>>>>
>>>>
>>>
>>> [snip]
>>>
>>> Why are you using ntpdate instead of ntpd?
>>
>>
>>
>> Beacuse the precision with ntpd is too large on the single board 
>> hardware I´m using.
>>
> 
> I don't quite understand that.
> 
> However, you say you are synchronizing a time SERVER by running ntpdate.
> Would that be a time server that is supplying time to other systems using
> ntpd? It won't work, and the odd behavior you are seeing may well be
> precisely because you are running ntpdate in the way you are.
> 
> -Tom

By the way, I assume these are successive operations:

Sep 23 23:59:00 xxx ntpdate[7167]: adjust time server 10.168.214.100 offset
0.000305 sec

Sep 24 00:00:01 xxx ntpdate[7176]: adjust time server 10.168.214.100
offset -0.008321 sec

Sep 25 00:01:00 xxx ntpdate[7226]: step time server 10.168.214.100 offset
86400.000773 sec

I assume you realize that 86400 seconds is 24 hours and that you've
noted that your date has indeed changed by a day just after midnight.
I suspect that the server may be changing back by a day 10 minutes later.
You might start by asking which date was correct and then go on to find out
what that server is doing each day at midnight (and at 10 minutes after).
Maybe some similarly poor method of time "synchronization"?




More information about the questions mailing list