[ntp:hackers] Fwd: UTC Leap Second at 2005-12-31T23:59:60Z, Risks Thereof

Richard S. Shuford shuford at aics.net
Thu Sep 1 21:39:33 UTC 2005


Folks,

Maybe you've seen this, but I forward it in case not.

 ...RSS

///////////////////////////////////////////////////////////////////////////////

.Organization: University of California, Berkeley
.Date: Sun, 28 Aug 2005 21:14:57 PDT
.From: RISKS List Owner <risko at csl.sri.com>
.Subject: Risks Digest 24.02
.
.RISKS-LIST: Risks-Forum Digest  Sunday 28 August 2005  Volume 24 : Issue 02
.ACM FORUM ON RISKS TO THE PUBLIC IN COMPUTERS AND RELATED SYSTEMS (comp.risks)
.Peter G. Neumann, moderator, chmn ACM Committee on Computers and Public Policy
.This issue is archived as http://catless.ncl.ac.uk/Risks/24.02.html

 Date: Mon, 22 Aug 2005 18:58:22 -0700 (PDT)
 From: Dave Glicksberg <davidg at bourbaki.jpl.nasa.gov>
 Subject: Risks of First UTC Leap Second in 7 Years

  [Originally submitted 2005-07-07, but lost in the shuffle.  PGN]

The International Earth Rotation Service (IERS, http://www.iers.org) just
announced a UTC leap second for the end of 2005, specifically at
2005-12-31T23:59:60Z (see http://hpiers.obspm.fr/eoppc/bul/bulc/bulletinc.dat).
The previous leap second was 7 years before, at 1998-12-31T23:59:60Z, which
was before Y2K!  In contrast, from UTC's inception in 1972 through 1998, leap
seconds were fairly common, occurring every 0.5 to 2.5 years.

UTC is the basis for civil and military timekeeping worldwide.  It is
transmitted in coded radio time signals like WWV, and it is used by Russia's
navigation satellites GLONASS (http://www.glonass-center.ru/stime.html),
which therefore must accommodate leap seconds.  However, GPS satellites use
a continuous timescale that does NOT have leap seconds.

THE RISKS?

* In the 7 years since the last leap second, maintainers of systems and
  software that are UTC-aware may have forgotten how to properly handle
  a leap second, whether it is done manually or automatically (e.g., by
  synchronization with WWV, or with time servers that properly handle the
  leap second).

* Newer systems and software have never encountered a leap second, unless
  via thorough testing.  Some systems may have omitted consideration of
  leap seconds altogether!

* Potential downtime or errors due to the need to do a manual update, or
  due to incorrect automatic updating.

* Consequences of forgetting that the leap second occurs simultaneously
  around the world, regardless of local time zone.  In New York, the leap
  second will occur at 7PM (actually, 18:59:60) on New Year's Eve, and in
  Moscow, it will occur at 3AM (02:59:60) New Year's Day.

Dave Glicksberg -- glicksbergd AT eh see em DOT oh are gee -- MY OPINIONS ONLY

///////////////////////////////////////////////////////////////////////////////



More information about the hackers mailing list