This also happened to me, but it happened to me because my server was using 
four of the Dutch Surfnet stratum 1 servers as a source, who were all 
announcing the leap second, and the other two servers I had configured did not. 
But because the Surfnet servers are Stratum 1, I think NTPd chose to believe 
them instead of the servers that were not announcing the leap second.

Tom

________________________________
From: [email protected] 
[[email protected]] on behalf of César de Tassis Filho 
[[email protected]]
Sent: 03 January 2013 16:33
To: David J Taylor
Cc: [email protected]
Subject: Re: [Pool] why some ntp servers tried to insert a leap second 2 
daysago?

Hello.

This happened to one of my servers (as you can see here [1] or here [2]), but I 
don't know if it was caused by a bug in my Debian Linux or if my NTPd trusted a 
wrong source.

César

[1] - http://i.imgur.com/uvbep.png
[2] - http://www.pool.ntp.org/scores/50.115.174.192


On Thu, Jan 3, 2013 at 12:07 PM, David J Taylor 
<[email protected]<mailto:[email protected]>> wrote:
From: Franck Martin


We noticed several ntp servers tried to insert a leap second on 31st December 
2012.

There was no leap second insertion scheduled for that date. 
http://en.wikipedia.org/wiki/Leap_second#Insertion_of_leap_seconds and 
ftp://hpiers.obspm.fr/iers/bul/bulc/bulletinc.dat

The linux kernel seems to still have bugs regarding this, that ripples in user 
space. 
http://it.slashdot.org/story/12/07/01/1920217/leap-second-bug-causes-crashes 
which lead then to a high power usage in some data centers:
http://www.datacenterknowledge.com/archives/2012/07/03/hosts-leap-second-caused-spike-power-usage/

Nevertheless, why so many ntp servers tried to insert a leap second?

PS: Will we have to call John MacClane 
(http://en.wikipedia.org/wiki/Live_Free_or_Die_Hard#Plot) in June 2013? ;)
=======================================================

Franck, did you happen to notice which version of NTP those servers were 
running?  Current versions of NTP require a majority vote before adding a 
leap-second, whereas earlier versions only needed one server's flag set to mark 
a leap-second.  I recall this is from a few years back, now, though.

The Windows, FreeBSD and Linux systems here were unaffected, and are all 
running a recent development branch of NTP.

Cheers,
David
--
SatSignal Software - Quality software written to your requirements
Web: http://www.satsignal.eu
Email: [email protected]<mailto:[email protected]>
_______________________________________________
pool mailing list
[email protected]<mailto:[email protected]>
http://lists.ntp.org/listinfo/pool

_______________________________________________
pool mailing list
[email protected]
http://lists.ntp.org/listinfo/pool

Reply via email to