[LEAPSECS] more poison in the NTP pool

[Warner Losh]

On Sep 1, 2012, at 8:05 AM, Rob Seaman wrote:


> On Aug 31, 2012, at 11:58 PM, David Malone wrote:

> 

>> I have 4-year versions of the "fraction of servers advertising

>> leapbits" graph here:

>> 

>> 	http://www.maths.tcd.ie/~dwmalone/time/leap2012/ntp-leapbits.png

>> 

>> It seems from this that there are always a few people advertising

>> leap seconds when they shouldn't be...

> 

> Fascinating...so it's more an adjustment in pool pH than the more dramatic "poison".  The community is simply paying closer attention now due to all hoopla, while David was previously prudently monitoring the temporal pH with leapbit litmus.


Does kinda show that it isn't trivial to get this right all the time...

Also shows the wisdom of filtering the LEAP indicator except on Dec 31 and Jun 30th that the reference implementation did for ages...  Helps to contain the damage.

Warner