arjanv/2.5 device-mapper lvm2

Bill Rugolsky Jr. brugolsky at telemetry-investments.com
Wed Dec 10 02:20:27 UTC 2003


On Tue, Dec 09, 2003 at 08:17:22PM -0500, Jeremy Katz wrote:
> On Tue, 2003-12-09 at 16:19, Joe Orton wrote:
> > I tried this on my FC1 system with a lvm rootfs, challenges were:
> > 
> > 1. lvm2 conflicts with lvm; I gave up and copy and pasted the stuff
> > into the initrd to get a kernel which booted into rc.sysinit
> 
> Yeah, I forgot about this...  Need to decide what I want to do there. 
> Two sets of identically named man pages for different commands is
> annoying :/

Luca Berra's lvm2 package for Mandrake handles this nicely using
/etc/alternatives:

   http://percy.comedia.it/~bluca/cooker/lvm2/

A simple expedient is to create an lvm-man subpackage.

BTW, I made a stab at all of this back in January.  You can find
my ancient packages here,

   ftp://67.83.203.127/pub/lvm
 
where I went through the effort of repackaging lvm as lvm1 and turning
it into a multi-call binary like lvm2, and separating our lvm*-links
and lvm*-man. (I did not think of /etc/alternatives at the time. :-/)
Setting up the links is useful for legacy scripts, etc.  I also built
/sbin/lvm1.static and /sbin/lvm2.static with diet libc, for use in initrds
(which was a pain, due to header file issues).  I'd like to get update
packages building with klibc, and create a "mkinitramfs."  [Is there an
initramfs "plan" for supporting udev, nfsroot, md/dm/lvm2, crypto, etc.?]

> > 2. rc.sysinit fails in fsck unable to find the device:
> > 
> > fsck.ext3: No such device or address while trying to open /dev/mainvg/rootfs
> > Possibly non-existent or swap device?
> > 
> > I wasn't sure how to fix that (initscripts-4.73).

The stumbling block that I hit back in January when switching back and
forth between LVM1 and LVM2 (and which hosed my root-on-LVM config),
was caused by not removing the old /dev/<vg> on the root fs prior to
invoking the vgscan in rc.sysinit.

I put it aside and moved on to other things when it became apparent that
device mapper was not yet stable enough (as Arjan had warned me).

Hope this helps.

Regards,

	Bill Rugolsky





More information about the test mailing list