severity 557636 normal
thanks
On Mon, Nov 23, 2009 at 02:09:18PM +0100, Bastian Blank wrote:
> Package: e2fsprogs
> Version: 1.41.9-1
>
> fsck.ext3 enforces synchronized clocks, but does not pull anything in to
> actually make sure that they are.
This only applies in virtualized environments, because...
> In most virtualized environments the clock is set from the host on
> startup but independent after that.
In a non-virtualized environment, synchronized clocks aren't an issue
at all. I would thus consider this to be a virtualization problem.
It's not appropriate to require ntp to be running on non-virtualized
machines, but it's a darned good idea if you are running virtualized
machines in productions, since often the timer interrupts aren't as
exact and time has a tendency to drift far more than OS's running on
bare metal.
> Because this can also happen easily for the _supported_ configuration
> with a system clock saved in local time (see /etc/defaults/rcS), I
> though about considering this as breaks the whole system.
That's a separate issue, and that situation has been solved. There
are kernel patches that have been pulled in by Ubuntu, and which are
in 2.6.32 that solve the problem with the last write time getting set
when the journal gets replayed (i.e., on an unclean shutdown).
There might also be a problem if you reboot during the transition
between summer time (daylight savings time) and normal time, but
that's a fundamental problem with having the hardware clock tick
localtime, and is why having your local hardware clock tick localtime
is fundamentally bug and a Really Bad Idea.
- Ted
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
