[LEAPSECS] telescope systems saw the leap second

Zefram zefram at fysh.org
Sun Jul 1 14:21:14 EDT 2012

Poul-Henning Kamp wrote:

>It means that Daniel Gambis wasn't even close to the mark

It's been a good advertisement for the position that if there are to
be leaps then we should be leaping most months (usually in alternating
directions). The Linux glitches are not just an example of a rarely-used
system concealing bugs: a major factor in them has been the uncertainty
about which kernels are susceptible, which is itself a product of the
rarity of leap seconds. Leaps need to happen reasonably frequently or
not at all.

(I personally have entirely escaped the glitches. Two Linux 2.6.32,
one 3.2.0, and one 3.4.4, all ticked through it with exemplary behaviour.
Got some nice logs of the kernel and ntpd aspects of the system clocks.
Every NTP server involved seems to have stepped correctly too.)


More information about the LEAPSECS mailing list