1 |
On Sat, 24 Dec 2005 19:33:26 -0500 |
2 |
Philip Webb <purslow@×××××××××.ca> wrote: |
3 |
|
4 |
> There will be a leap second between 051231 235959 & 060101 000000 . |
5 |
> Does anyone know how the time servers used by NTP handle this ? |
6 |
> Is it just left to the local machine to realise it's 1 sec fast |
7 |
> & adjust over a few hours or does something else alert it to correct |
8 |
> things ? If the former, it could create problems for those running |
9 |
> experiments; if the latter, does anyone know how it is done ? |
10 |
> The last leap second was 1998/9 , |
11 |
|
12 |
According to the RFC there has always been a mechanism, but it was not |
13 |
originally automatic. In RFC958, http://rfc.net/rfc958.html, there is |
14 |
this reference |
15 |
|
16 |
5.4. Leap Seconds |
17 |
|
18 |
A standard mechanism to effect leap-second correction is not a |
19 |
part of this specification. It is expected that the Leap |
20 |
Indicator bits would be set by hand in the primary reference |
21 |
clocks, then trickle down to all other clocks in the network, |
22 |
which would execute the correction at the specified time and reset |
23 |
the bits. |
24 |
|
25 |
The newer NTP standard is now RFC1305, http://rfc.net/rfc1305.html. |
26 |
>From my brief look it seems that the leap seconds are listed in a |
27 |
file. The server then sets the leap indicator when needed. |
28 |
|
29 |
> before NTP was widely used. |
30 |
|
31 |
Not sure about that, but even the later standard is now 13 years old. |
32 |
The original is 7 years older than that. I'd be surprised if a lot of |
33 |
permanently connected sites, eg Universities, haven't been using it for |
34 |
a long time now. However most home users wouldn't have had much use |
35 |
for it before the growth of lower rate broadband services. |
36 |
|
37 |
-- |
38 |
Ian. |
39 |
|
40 |
EOM |
41 |
|
42 |
-- |
43 |
gentoo-user@g.o mailing list |