[nSLUG] Who spent midnight watching for the leap second?

gnwiii at gmail.com gnwiii at gmail.com
Mon Jan 2 12:27:23 AST 2006

On 1/2/06, Donald Teed <donald.teed at gmail.com> wrote:
> I googled for a bit of education on that, as I saw something in
> /var/log/messages
> that the kernel updated the clock.
> Clock: inserting leap second 23:59:60 UTC

I have the same message for 2.6.12-1.1381_FC3smp and 2.6.14-1.1653_FC4
kernels.  There are 2 issues: 1) NTP clients/servers failing to deal with
leap second, and 2) kernel involvement in adjusting the system clock.

It seemed pretty weird that the kernel would know about the leap
> second.  It turns out that ntp knows how to talk to the
> Linux kernel.  I don't think there is anything this fancy happening
> when ntp otherwise changes the system time.

Normally, NTP sets the drift value so that system time gradually
adjusts to a more correct value.  A leap second is simple a change
in the mapping from NTP time to system time.

Information on how to deal with leap seconds was available
at: http://www.eecis.udel.edu/~mills/leap.html

It seems that not everyone has followed the recommendations, so some
people saw time go backwards.  The other source of problems
was with certain reference clocks and or drivers.  I expect most reference
sources also implement the leap second, so extra work is needed so NTP
knows the difference.

George N. White III <aa056 at chebucto.ns.ca>
Head of St. Margarets Bay, Nova Scotia

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://nslug.ns.ca/mailman/private/nslug/attachments/20060102/f51085c9/attachment-0001.html>

More information about the nSLUG mailing list