[ntp:questions] Re: What went wrong with the leap second

Serge Bets serge.bets at NOSPAM.laposte.invalid
Sat Jan 21 15:31:43 UTC 2006

Hello Ulrich,

 On Friday, January 20, 2006 at 15:01:21 +0100, Ulrich Windl wrote:

> Another good question is how NTP will behave on different references
> showing different leap bits: Will it absolutely trust the current sync
> peer, or will it make a majority decision?

The ntp daemon will take the bitwise OR of all sources leap bits. Though
I don't know what exactly means "all" here: Really all reachable, all
survivors, all candidates, or such. A single leap=01 source wins over
any number of leap=00 refclocks, servers, and NIST leapsecond files.

> Also, can a bit mistakenly set via one reference be cleared by another
> reference? (Or malicuosly be reset?)

No. But I have a plan to change that, permitting to a valid NIST
leapsecond file to master daemon's leap bits.

Serge point Bets arobase laposte point net

More information about the questions mailing list