[ntp:hackers] smearing the leap second
stenn at ntp.org
Thu Jul 9 21:48:42 UTC 2015
In theory I agree with you. But reality bites.
We still see huge numbers of people running very old versions of NTP.
How do you propose to get lots of devices to get upgraded kernels in
order to properly handle a leap second?
If they *do* properly handle the leap second, how do you get lots of
instances of *appliations* to properly handle what looks to them like a
backward time step?
Mike S writes:
> NTP shouldn't be involved in "smearing" time, at all. The only reason
> for smeared time is because of systems (e.g. POSIX) which don't properly
> support leap seconds. Any smearing should be done by the OS, not NTP,
> which should strictly follow UTC. To do otherwise is to deliberately
> break NTP.
> Even then, NTP and parts of RFC 5905 are fundamentally flawed even in
> that regard - it's supposed to count leap seconds with a "monotonically
> increasing" timescale, but in practice has the same flaw as POSIX. It's
> simply not a well designed (or documented for that matter, search for
> "leap" in the RFC) protocol.
> hackers mailing list
> hackers at lists.ntp.org
More information about the hackers