[LEAPSECS] Leap Sec vs Y2K

[Daniel R. Tobias]

On 11 Dec 2010 at 0:40, Paul Sheer wrote:


> At the ISP I consult for there are about 20 servers serving 60,000

> customers. Their clocks routinely go out of sync and it doesn't affect

> service.


I have a script that dumps the timestamps of each of a number of 
servers where I work; this is a recent result:

Fri Dec 10 23:07:12 EST 2010
Fri Dec 10 23:07:11 EST 2010
Fri Dec 10 23:07:06 EST 2010
Fri Dec 10 23:07:11 EST 2010
Fri Dec 10 23:07:12 EST 2010
Fri Dec 10 23:07:15 EST 2010
Fri Dec 10 23:07:16 EST 2010
Fri Dec 10 23:07:15 EST 2010
Fri Dec 10 23:07:15 EST 2010
Fri Dec 10 23:07:15 EST 2010
Fri Dec 10 23:07:12 EST 2010
Fri Dec 10 23:07:12 EST 2010
Fri Dec 10 23:07:13 EST 2010
Fri Dec 10 23:07:13 EST 2010
Fri Dec 10 23:07:12 EST 2010
Fri Dec 10 23:07:13 EST 2010
Fri Dec 10 23:07:13 EST 2010 

They're more in sync than they were at various times in the past, due 
in part to my own nagging of the sysadmin to get automatic time 
syncing in place consistently (it sometimes takes a good deal of 
fighting to get them to do that, as they are always deathly afraid of 
it corrupting the database or something, due to the time being set 
backward while it's running; this has always proved unfounded, as the 
time syncing is now fully in place even on the database servers with 
no ill effect).  Still, there's a good deal of drift in between auto-
syncs, leading to a ten second gap between the most extreme outliers, 
though most servers are within a five second span.

A leap second will be lost in the noise here, resulting merely in the 
next regular adjustment going one second more (or less) than it would 
have otherwise, with a different absolute magnitude on each server 
depending on its own drift.

-- 
== Dan ==
Dan's Mail Format Site: http://mailformat.dan.info/
Dan's Web Tips: http://webtips.dan.info/
Dan's Domain Site: http://domains.dan.info/