[ClusterLabs] Antw: Coming in Pacemaker 2.0.3: Year 2038 compatibility
Ulrich.Windl at rz.uni-regensburg.de
Mon Oct 7 07:06:18 EDT 2019
>>> Ken Gaillot <kgaillot at redhat.com> schrieb am 17.09.2019 um 06:38 in
<a55bd11ed4f480324ff6ce27e3b68e0e10bd43fb.camel at redhat.com>:
> Hi all,
> I wanted to highlight a feature of the next Pacemaker release: it will
> be ready for the Year 2038.
> I'm sure most people reading this are familiar with the problem.
> Representing epoch timestamps (seconds since 1970‑01‑01) as signed 32‑
> bit integers will overflow at 2038‑01‑19 03:14:07 UTC, wreaking havoc
> and bringing about the collapse of civilization (or at least some
> embedded systems).
> Most OSes are ready at the kernel and C library levels on 64‑bit CPU
> architectures. (There's still a lot of work to be done for filesystems
> and applications, and 32‑bit architectures may never be fixed.)
Amazingly, I had been asked about 32-bit and 2038 recently, so maybe you want
to see See also: https://en.wikipedia.org/wiki/Year_2038_problem
> Until now, Pacemaker has not been Y2038‑ready, storing
> timestamps insufficiently in memory and the CIB. This is expected to be
> fully remedied in 2.0.3. So go ahead, set up that rule to put your
> cluster in standby at 10:30 p.m. March 15, 2040. :‑)
> I'm planning to start the 2.0.3 release cycle in about a month.
> This change will not be backported to the 1.1 series, which is expected
> to be end‑of‑life sometime in the next few years.
> Ken Gaillot <kgaillot at redhat.com>
> Manage your subscription:
> ClusterLabs home: https://www.clusterlabs.org/
More information about the Users