USB external Hard disk - May be LVM
fc-cornette at insight.rr.com
Tue Mar 22 03:33:51 UTC 2005
Antonio Montagnani wrote:
> Jim Cornette ha scritto/wrote il giorno/on 20/03/2005 11:13:
> Tnx Jim
> As now it is too late to start this task, let me resume what I am aiming
> to and what is my experience...
> 1) I am trying to set up a Fedora on an external USB hard disk
> 2) I do want that the system will start Fedora immediately if the
> external disk is connected (otherwise Windows should run)
> During installation I decided not to install Grub as my goal is to have
> a system booting Linux when the external HD i sconnected otherwise
> Windows must be started. No choices to the user....and I want to be
> connect to any system (mainly laptops) and use Linux
I toyed around a bit with an Industrial computer that was capable of
booting from USB hard drives,usb-cdroms or usb-floppies once and the
installation booted up to a point. I think the installation was RHL 7.3
and used devicenames vs labels. It kernel panicked, but at least it
started to boot.
For your attempted setup, if BIOS supports usb booting devices, you
might choose the usb-hard-drive as the booting device before the hard
drive. In theory and with my limited success with booting from a usb
hard drive, it *might* work.
I believe with labels (new default) instead of device names, grub should
start from the boot loader in the MBR of the USB hard disk. The labels
should allow every partition to find its mountpoint. If you cannot set
your boot device order in BIOS, it would be a tough thing to get to
work. I believe you would still have to run the line included in your
reply below for the usb modules to be available for when the kernel
loads and assumes control of the USB-HD.
> Therefore I installed Fedora on this HD issuing the command:
> Linux expert noacpi
> When I rebooted the system, of course it didn't boot (USB booting is
> I tried to follow instructions at www.simonf.com/usb but I got lost.
> When I issued the command
> mkinitrd --preload=ehci-hcd --preload=usb-storage --preload=scsi_mod
> --preload=sd_mod /boot/usbinitrd.img 2.6.9-1.667
I can see somewhat what they are attempting. I didn't try to get these
usb related modules to be included in the kernel.
I asked on another list why this option was not available during
instalations before. I believe erratic operation or other reliability
issues were given as the reasons for the method not to be ofered during
> But the system didn't boot !!!
> So I decided to check what was in the /boot directory and I didn't find
> the usbinitrd.img file.
Try unmounting the /boot partition and see if the intird is in the
directory used as a mounting point for the boot partition. If the boot
partition was not mounted when you ran mkintird, it would be in the
/boot directory when the /boot partition is not mounted. I found that in
rescue mode a few times, the boot partition was not mounted when I
chrooted to /mnt/sysimage. I believe I ran rpm to install a kernel on a
usb drive that I could only get functional when chrooting from a usb
drive and the kernel was installed in the boot directory when the /boot
partition was not mounted. (I didn't think to check if /boot was mounted.)
> I suppose that I will have to issue the modified command mkinitrd.
> At this point I realized that I couldn't see other directories..
This happened to me for the disk that I referred to. I tried googling
for the references to mount an lvm with little luck. I then checked a
webmail account of mine and found the mail describing how to mount an
lvm. (initialize, then make mountable)
> What is your suggestion?? to play a fresh installation with standard
> partitions (how do you manage them???) or making them active following
> your instructions?
I wouldn't trust my earlier directions to get a system active. It was
luck that it worked well enough to get at my data. I believe that
dev-mapper (Not sure of spelling or actual functions for this. But I
realize it controls the lvms in some way.)
You must realize that the computer has it in for you. The irrefutable
proof of this is that the computer always does what you tell it to do.
More information about the users