SCIENTIFIC-LINUX-USERS Archives

December 2010

SCIENTIFIC-LINUX-USERS@LISTSERV.FNAL.GOV

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Pann McCuaig <[log in to unmask]>
Reply To:
Pann McCuaig <[log in to unmask]>
Date:
Wed, 22 Dec 2010 14:42:51 -0500
Content-Type:
text/plain
Parts/Attachments:
text/plain (75 lines)
On Wed, Dec 22, 2010 at 13:00, Phong Nguyen wrote:

> Have you tried determining what's in mdadm.conf in the initrd file? It
> might be getting some incorrect assembly instructions for md0.

This may well be the issue. Other than /etc/modprobe.conf (which is
obvious from the mkinitrd man page) and /etc/mdadm.conf, what does the
initrd file look at (or where can I find out)?


> On 22 Dec 2010, at 1203, Pann McCuaig wrote:
> 
> > Greetings SL fans,
> > 
> > Sorry for the length of this post, but I'm hoping someone can come to my
> > rescue and want to provide sufficient context.
> > 
> > Recently I had a hard drive failure on a Sun X4600 box running SL4.8.
> > The box has four drives; the drive that failed was /boot (only). The
> > other three drives make up /dev/md0.
> > 
> > The /boot drive was not backed up (headsmack!).
> > 
> > I have created a rescue USB stick based on the System Rescue CD, which
> > boots via grub. I can boot the System Rescue CD successfully, and it
> > sees /dev/md0, which I can then mount and read from and write to.
> > 
> > I have attempted to create a kernel and initrd image to add to the USB
> > stick that will boot the box as if the kernel and initrd image were on
> > the failed (and now removed) boot drive.
> > 
> > I built the kernel and initrd image on a box similar to the box with the
> > failed hard drive. I replaced /etc/modprobe.conf with the file from the
> > target server, and then did 'yum install kernel-largesmp'. I copied all
> > the resulting kernel-related files from /boot to the USB stick, as well
> > as the appropriate directory from /lib/modules.
> > 
> > I then restored the helper box to its original state.
> > 
> > I booted the System Rescue CD on the target system and copied the
> > /lib/modules directory into place on /dev/md0. I fixed up grub/menu.lst
> > to have a stanza to boot the newly created SL4 kernel. I rebooted the
> > box, and everything seemed to be going swimmingly, until . . .
> > 
> > . . . the booted kernel seems unable to build /dev/md0 and the boot
> > process fails.
> > 
> > In the original configuration, the boot drive was /dev/sda, and the
> > drives making up the soft RAID partition were /dev/sdb, dev/sdc, and
> > /dev/sdd.
> > 
> > The System Rescue CD detects the USB stick as /dev/sda and the three SAS
> > drives as sdb, sdc, and sdd. All is well.
> > 
> > It's not clear to me what is going awry with the SL kernel, but as
> > the boot verbiage scrolls by, I see /dev/sdc referenced twice, and no
> > reference to /dev/sdd. When the kernel attempts to assemble /dev/md0, it
> > uses /dev/sda, /dev/sdb, and /dev/sdc and this fails and /dev/md0 cannot
> > be mounted and the kernel panics.
> > 
> > Help, please. Suggestions? Thanks.
> > 
> > BTW, I've put both SL4.8 Disc One, and the SL4.8 Live CD on a bootable
> > USB stick; both boot successfully, but I was unable to find a way to
> > make either one recognize /dev/md0, much less "rescue" me.
> > 
> > Cheers,
> > Pann

-- 
Pann McCuaig <[log in to unmask]>                212-854-8689
Systems Coordinator, Economics Department, Columbia University
Department Computing Resources:
               http://www.columbia.edu/cu/economics/computing/

ATOM RSS1 RSS2