[ntp:questions] ntpd oddness

John Oliver joliver at john-oliver.net
Wed Apr 2 19:43:19 UTC 2008


On Wed, 02 Apr 2008 09:58:38 +0200, Martin Burnicki wrote:
> What's the output of:
>
> date -u; date

[root at 0123456789-VCS ~]# date -u; date
Wed Apr  2 20:29:25 UTC 2008
Wed Apr  2 20:29:25 GMT 2008

>> [joliver at 0123456789-VCS ~]$ /sbin/hwclock
>> Tue 01 Apr 2008 09:05:39 PM GMT  -0.323329 seconds
>> [joliver at 0123456789-VCS ~]$ sudo /sbin/hwclock --systohc
>> [joliver at 0123456789-VCS ~]$ /sbin/hwclock
>> Tue 01 Apr 2008 09:05:52 PM GMT  -0.776568 seconds
>
> Normally the system time is only initialized from the hwclock (the RTC chip
> on the mainboard) at boot time, and when the system shuts down properely
> the current system time should be written back to the hwclock. In most
> cases you see this in the console messages.
>
> If you have a dual/multi boot system then you must take care that all
> operating systems assume the RTC to run at the same time, i.e. either local
> time or UTC.

There is only one OS on the host(s) in question.

> If the system is Linux only I'd suggest you configure your Linux system such
> that the RTC chip keeps UTC time only. If then the time is not correct
> after a reboot your on-board battery may be low.

There are no reboots.  The above example is one command after another...
read the hwclock, set it from the system time, and then immediately read
it again.  I don't see how this could be a hardware issue, unless this
chip is specifically programmed to always add/subtract 12 hours from the
time it's set to, which I rather doubt :-)

-- 
* John Oliver                              http://www.john-oliver.net/ *

-- 
Posted via a free Usenet account from http://www.teranews.com




More information about the questions mailing list