On Apr 6, 2006, at 12:03 PM, Fredrik Tolf wrote:
On Thu, 2006-04-06 at 11:51 -0500, John Jolet wrote:
sounds like you're using device mapper as a module...compile it into
the kernel. module loading happens too late
Thanks for your reply, but I seriously doubt that to be the problem,
since /etc/rcscripts/addons/lvm-start explicitly contains the
following
code:
if [[ -e /proc/modules ]] && \
! grep -qs 'device-mapper' /proc/{devices,misc}
then
modprobe dm-mod &>/dev/null
fi
Also, if the dm module wasn't loaded, I think that is what the LVM
tools
would complain at, not at not finding the PVs, right?
Fredrik Tolf
well, i've not had this problem, so i'm speculating...but if for
whatever reason the dm code isn't resident, it would, in fact,
complain about not being able to find pvs...pvcreate is the first
step to creating a volume group. i can't recall right now what
happens if you try to run pvcreate without the dm extensions loaded.
I did do that once apon a time. maybe it let me do that, but died on
the lv creation. modprobe will fail on my system, because i always
compile that into the kernel....but i can't say at what stage of the
boot process lvm-start gets called. i'm probably way off base....but
if the /etc/rcscripts/addons/lvm-start script resides on an lv, or
calls a binary that's say in /usr, which on my system is on an lv, it
would naturally fail at that point....
--
gentoo-user@gentoo.org mailing list
