[LEAPSECS] more poison in the NTP pool

[mike cook]

Le 02/09/2012 19:49, Ask Bjørn Hansen a écrit :

> On Sep 2, 2012, at 9:00 AM, leapsecs-request at leapsecond.com wrote:

>

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

>> With a notable percentage of the NTP servers always indicating leap

>> it's another instance of "Don't believe everything the internet says."

> While true, I think that's missing the point Warner was making: If people running a time server can't reliably get the time server itself to manage the leap seconds properly, what are the odds really that everyone else will get it right – ever.

>

> (As a side note I am planning to have the NTP Pool system monitor the leap flag, though it's not clear how much that will really help: https://github.com/abh/ntppool/issues/55 )

Abusive servers could be removed from the pool. It won't help those 
already using them, but could prevent infection from spreading to new 
clients with old versions of ntpd. Just as a test I started up a server 
with 5 of the mavericks as configured servers and verified that they 
were propagating the bad leap indicators. My version of ntp did not get 
infected, and is not propagating bad packets to downstream clients.

-- 
Les chiens aboient, et la caravane passe.