[Top] [All Lists]

Re: Problems mounting an ext3 filesystem with XFS kernel

To: Linux XFS Mailing List <linux-xfs@xxxxxxxxxxx>
Subject: Re: Problems mounting an ext3 filesystem with XFS kernel
From: Lucas Barbuto <lucas@xxxxxxxxx>
Date: Wed, 29 Jan 2003 09:54:37 +1100
In-reply-to: <>
References: <20030127232221.GA21514@xxxxxxxxx> <>
Reply-to: Lucas Barbuto <lucas@xxxxxxxxx>
Sender: linux-xfs-bounce@xxxxxxxxxxx
User-agent: Mutt/1.4i
Hi Seth,

On Tue, Jan 28, 2003 at 09:46:39AM +0100, Seth Mos wrote:
> At 10:22 28-1-2003 +1100, Lucas Barbuto wrote:
> > Yes.  It's /dev/md1.
> not /dev/md0 ??

No, I named my RAID devices after the underlying partitions they consist
of, ie, hde1 and hdg1 map to md1, to avoid confusing myself.

> That means that linux detects the disk in a different way then your bios 
> does. You can correct this with a line in your lilo.conf.
> It might be related to your booting problem. If you say that the boot drive 
> is hda, which would correspond with what the bios thinks. However when the 
> kernel boots the first disk is detected as hde and it can get confused at 
> this point.

The boot device (as specified in LiLo) is /dev/md1.  I have started a
similar thread on linux-raid as I've realised that this is a RAID
problem rather than an XFS problem.  There I received the advice that
the kernel is booting off /dev/hde1 or /dev/hdg1 and not actually off
the RAID device and that the RAID is not being constructed before the
kernel tries to mount the root device (md2), hence crashing ensues.
This makes sense to me, I haven't been able to fix it though.
> My lilo.conf contains.
> boot=/dev/hde
> map=/boot/map
> install=/boot/boot.b
> prompt
> timeout=50
> linear
> default=2418-rht-1.2
> vga=0x430
> message=/boot/message
> image=/boot/vmlinuz-2.4.18-19-xfs-1.2pre5
>         label=2418-rht-1.2
>         read-only
>         root=/dev/hde1
> This machine has the onboard IDE deactivated as well as your system and 
> thus hda does not exist so I changed the boot line.
> Subsequent lilo does not give me a line saying it's confused about where 
> the disk should be.

Yeah, that's similar to what mine was, but I was still getting the
warning message.

> You could check either the kernel config or a initrd image file in your 
> /boot/ If that's where debian places the kernels.

I checked, not using initrd.

> You can always check though. I forgot to compile in the md drivers once and 
> put a box 150Km (~100Miles) away out of service.

Hmm, well this box is under my desk at the moment, but it's headed for a
data centre so I need to get it right :).

> If I understand what you are trying to do you want to format the / 
> partition with ext3 and leave the rest XFS?
> If that is so format /dev/md1 with ext2
> mkfs.ext2 /dev/md1
> Create the ext3 journal ontop of it.
> tune2fs -j /dev/md1
> Now you should be able to mount the / as ext3.

Yeah, that's right, I used 'mke2fs -j /dev/md1' to create an ext3, I
think this does the same thing you suggested, but in one line.

Anyhow, I've since given up on the onboard RAID controller and I've put
my system disks back in the primary and secondary IDE slots because I've
done it that way before and it's worked.  I'll just use the onboard RAID
for an extra data disk or two.

Problem is, now I'm trying to get the system to boot with the root RAID
in degraded mode with a boot disk but it's telling me:

Invalid compressed format (err=1)

 -- System halted


Thanks for your help.



<Prev in Thread] Current Thread [Next in Thread>