[ntp:questions] basic questions about the leapsecond

David L. Mills mills at udel.edu
Tue Dec 16 15:51:32 UTC 2008


Martin,

I've had a really hard time with the state of the release/dev versions, 
but I don't manage that. The release version has a date of August this 
year, but the protocol module is dated August 2006. I spent a whole year 
beginning in June 2007 making major changes, fixes and improvements, but 
they have not appeared in the current release. All I can say is that my 
comments apply to the development version and that the release version 
is on a different maintenance track. I assume, but cannot confirm, that 
the docuemtation in the release version applies specifically to that 
version. The documentaiton in the current development version is on the 
web and is regularly updated (as of last night in fact). The 
documentation in other places may or may not conform to reality.

Dave

Martin Burnicki wrote:

>Dave,
>
>David L. Mills wrote:
>  
>
>>Guys,
>>
>>See http://www.eecis.udel.edu/~mills/ntp/html/ntpd.html#leap for what
>>actually is in the implementation. 
>>    
>>
>
>... which is valid for the current developer version, NTP 4.2.5p[something],
>but must not necessarily match the current (or earlier) stable versions,
>i.e. NTP 4.2.4p5 and earlier.
>
>  
>
>>As visivle here, the Spectracom 
>>(GPS/WWVB) driver, ACTS telephone modem driver and WWV audio driver do
>>correctly display leap information. The Meinberg GPS receiver, EndRun
>>CDMA receiver and audio IRIG driver do not display leapsecond
>>information.  
>>    
>>
>
>... yet. Of course the Meinberg GPS receiver is aware of the upcoming leap
>second, and it will set the leap second announcement flag a certain
>interval before the time the leap second is actually inserted.
>
>The interval depends on the firmware version. It is 59 minutes in older
>firmware versions (due to compatibility with the German longwave
>transmitter DCF77) and has been extended in recent versions to 1 day if the
>Uni Erlangen string format is used.
>
>  
>
>>The USNO and NIST servers do not show leap bits in the VME 
>>or ACTS associations, but that might not be significant until the day
>>before the leap. Our servers pogo.udel.edu and rackety.udel.edu are
>>correctly configure and the leap warning bits in the driver associations
>>correctly set. To check a server near you, run ntpq, as and rv commands
>>for the system peer and verify the leap bits are set to 01,
>>
>>As mentioned in the documentation, it is not necessary to run Autokey to
>>download the leapseconds file from NIST. It is availabe via FTP from the
>>pub directory and leap-seconds file.
>>    
>>
>
>Again, AFAIK this is true only for the developer version of NTP. 
>
>I guess most NTP servers run a stable version, however, which (again AFAIK)
>requires autokey to be enabled in order to evaluate the leap second file.
>Unless I'm totally missing something the stable versions up to the current
>one do not even provide a way to download that file by itself but need
>manual intervention to copy the file to the right location.
>
>As I've already posted earlier in this NG people should tell to which
>version they are referring if they explayin things which have undergone
>basic changes like the leap second handling.
>
>Martin
>  
>




More information about the questions mailing list