Re: [gentoo-user] LVM boot problem

[Boyd Stephen Smith Jr.]

On Thursday 06 April 2006 11:44, Fredrik Tolf <fredrik@×××××××××.com> wrote 
about '[gentoo-user] LVM boot problem':
> I'm having a bit of a problem with LVM2 on Gentoo. See, I have a
> computer with a couple of hard drives in an LVM, and when it boots, I
> can see it loading the driver modules for the controller cards properly
> (I've added them to /etc/modules.autoload.d), but then when it comes to
> initializing the LVM, it complains that it cannot find all the PVs to
> initialize the VG. In fact, it complains about _all_ the PVs, as though
> it found none.

Is / on LVM?  Are you using an initrd / initramfs?  Does your early 
userspace start a /dev daemon?  Does your early userspace have a different 
lvm.conf then your final userspace?

> Later, when the system is booted, I can log in as root and just run
> "vgchange -ay" to get the VG up. Note that I don't have to run pvscan
> first, which I'm assuming has to mean that it has somehow managed to
> find the PVs somewhere in the meantime.

Are all of your pvs standard disks or disk partitions?  What modules are 
being loaded by the gentoo /etc/modules.autoload system?  What is your 
RC_VOLUME_ORDER in /etc/conf.d/rc?  Does 
your /lib/rcscripts/addons/lvm-start.sh 
contain "/sbin/vgscan --mknodes --ignorelockingfailure >/dev/null"?

> This problem is quite annoying, since it means that I have to restart
> NFS and boot the system without the network cable plugged in (if it was,
> the NFS clients would try to reconnect and get handles to the wrong
> filesystem, which makes them mark the mount points as stale).

You might investigate the mpt= /etc/exports option to only export 
filesystems that have been successfully mounted, at least temporarily.

-- 
"If there's one thing we've established over the years,
it's that the vast majority of our users don't have the slightest
clue what's best for them in terms of package stability."
-- Gentoo Developer Ciaran McCreesh