Boot Problem after Crash

[Tony Abou-Assaleh]

Lennart Sorensen wrote:
> On Fri, Mar 07, 2008 at 04:56:21PM -0400, Tony Abou-Assaleh wrote:
>> Lennart Sorensen wrote:
>>> On Fri, Mar 07, 2008 at 01:39:18PM -0400, Tony Abou-Assaleh wrote:
>>>> Hi tlugers,
>>>>
>>>> My PSU was toasted. I took the HDD out and installed it on another PC. 
>>>> The file system had problems but after e2fsck it appears stable. No data 
>>>> was lost as far as I can tell, but I couldn't boot into by Ubuntu 7.10 
>>>> (xubunutu) Linux.
>>>>
>>>> I booted using a live CD and went to rescue mode. I updated/upgraded 
>>>> packages using apt-get, executed update-initramfs, and all seemed well. 
>>>> I can even start apache and sshd from this rescue shell.
>>>>
>>>> When I try to boot from the HDD I consistently get the same thing: I get 
>>>> the BusyBox initramfs shell and I don't know how to go past that. 
>>>> Nothing is mounted.
>>>>
>>>> I have the boot partition on /dev/sda1 and the root partition on raid1 
>>>> volume. I am using only a single drive from the raid array.
>>>>
>>>> Any ideas why I'm getting the initramfs prompt on boot and how to get 
>>>> past that?
>>> Perhaps the new machine enumerates the disks differently so it doesn't
>>> know where to look for the root.
>>>
>>> What do you get from 'cat /proc/partitions' in the initramfs shell?
>> --
>> major	minor	#blocks	name
>> 8	0	xxx	sda
>> 8	1	xxx	sda1
>> 8	2	xxx	sda2
>> 9	0	xxx	md0
>> --
>>
>> I omitted the block counts. sda1 is the boot partition, and sda2 is the 
>> raid container. When I run the same command from the recovery shell, I 
>> get 2 additional entries:
>>
>> --
>> 253	0	xxx	dm-0
>> 253	1	xxx	dm-1
>> --
>>
>> Where dm-0 is the swap partition and dm-1 is the root partition.
>>
>>> How about /proc/mdstat?
>> --
>> Personalities : [raid1]
>> md0 : active raid1 sda2[0]
>> 	xxx blocks [2/1] [U_]
> 
> So one disk in the raid failed to appear?

Yes.

>> unused devices: <none>
>> --
>>
>>> Perhaps you can find out what root is called and change the boot loader
>>> to use root= whatever that is.
>>>
>>> I do everything in the boot loader and fstab by UUID these days just to
>>> avoid this kind of hassle.
>> I changed it to use root=UUID=xxx in the grub menu.lst, same thing.
> 
> Can you mount the root partitions from initramfs and then pivotroot or
> whatever they call it now to it and continue the boot?

When I did lvm scan, it showed my root and swap volumes as inactive. 
After I activated them, I was able to mount them. I tried to chroot to 
the root but that didn't go too well because none the libraries required 
by many standard utilities were loaded. So I don't know what to do next 
from initramfs prompt.

>>> The other option is that your initramfs only loaded the driver modules
>>> needed by the old machine and not the ones for the new one, although
>>> most I have seen recently (in debian at least) try to load pretty much
>>> everything.  Is the new machine perhaps too new or simply not supported
>>> by your linux version?
>> I can boot fine from a live CD, so the machine is linux compatible. I 
>> was able to connect to the Internet, start apache and sshd manually, and 
>> connect to them from another machine.
>>
>> Also I ran update-initramfs on the new machine, so again that's not 
>> likely to be the problem.
>>
>> It looks like the raid volume is recognized during the boot sequence, 
>> but the partitions within it are not. Any ideas?
> 
> I have never used partitions on raid.  I always run LVM on raid.  I know
> how that works.  LVM is much more flexible than partitions, so why use
> partitions on raid?

My bad. I am actually using LVM and they're volumes, not partitions.

Cheers,

TAA

-- 
Tony Abou-Assaleh
Email:    taa-HInyCGIudOg at public.gmane.org
Web site: http://tony.abou-assaleh.net
--
The Toronto Linux Users Group.      Meetings: http://gtalug.org/
TLUG requests: Linux topics, No HTML, wrap text below 80 columns
How to UNSUBSCRIBE: http://gtalug.org/wiki/Mailing_lists