[ntp:questions] Leap seconds

David L. Mills mills at udel.edu
Tue Jan 25 02:23:16 UTC 2005


As long as the primary servers as a group correctly interpret the radio 
timecode and do recognize the leap epoch, and if kernel provisions are 
available, the clock will not be stepped at the leap epoch. See the 
executive summary www.eecis.udel.edu/~mills/leap.html, which is also 
linked from the NTP project page.


Alain wrote:
> Enrique Perez-Terron escreveu:
> [...]
>> Please correct me if I am wrong.
>> This raises the question what will time(NULL) return at
>> 23:59:58, 23:59:59, 23:59:60, 00:00:00, and what will gettimeofday()
>> return - and when will it return? (I have seen a proposal to have
>> gettimeofday() sleep until 00:00:00 if called after 23:59:59.9999 but
>> before 00:00:00.)
>> Excuse my ignorance, does ntp count the time only in standard calendar
>> ways, or does it have a notion of seconds since an epoch?  In case of
>> the latter, does this number increase monotonically at a constant rate?
>> Is this time available for applications outside ntpd?
> It looks like the "safest way on critical systems" and "I don care much 
> but want to be on the safe side" would be not to step and just skew that 
> extra/missing second. Would the "-x" option do that?
> Would that be ok for a typical small-business where ntp is only used for 
> loging intrusion detection and 1s is good enough?
> Alain

More information about the questions mailing list