[ntp:questions] Re: Leap bits, when to set?

Arnold arnold at hacktic.nl
Wed Dec 28 14:12:35 UTC 2005


On 23-12-05 02:42, David L. Mills wrote:
> The kernel modifications were after rfc-1305 on NTPv3 was published. I
> described what the current ntpd does. The issue is not "processing"them,
> but passing them along to dependent clients, who may have an poll
> interval of over a day. Presumably, the current rewrite for NTPv4 will
> make these points wickedly clear.

I am sorry, but from this, it is not clear to me, when NTP-servers
should send the NTP-leap second indication 'over the line' to NTP-clients.

>From RFC-1305 and the article on
http://www.eecis.udel.edu/~mills/leap.html ("The NTP leap bits are set
by the protocol on the day of insertion") I figure that an NTP-server
should only send replies with the NTP leap bits set on 31 Dec 2005.
Other people claim that an NTP-server should send replies with the NTP
leap bits set from 1 Dec 2005 to 31 Dec 2005.

So my question is: is an NTP-server that sends replies to clients *with*
the NTP leap bits set in the period 1 Dec 2005 to 30 Dec 2005 (days on
which no leap second occurs), sending valid information compliant with
RFC-1305? Or is such a server (invalidly) telling me that each day of
December 2005 has a leap second?
(For this question, I don't care about 'processing' by the kernel, etc.)

   Arnold &:-)




More information about the questions mailing list