Raid5 boot failure after grow

Hello,

I'm using Fedora 15 and I added an additional disk to my raid5 setup
and grew the array to encompass it. It all worked fine and rebuilt and
everything. However, now when I boot my computer I am dropped to a
shell, stating "dracut" (I wonder what that means). Anyways, when in
here I can cat /etc/mdadm.conf and it appears to be an old
configuration file. It lists only 3 raid devices in that
configuration. I now have 4.

If I run mdadm --assemble --scan after removing the config file, it
will automatically find my array and all is well. But this doesn't
solve the fact that it doesn't boot to it.

dmesg reports "no root device
'block:/dev/disk/by-uuid/933ac67-98f1-481f-934e-c9af492ba28c ' found"

This is particularly interesting, because blkid lists /dev/md127
(after I've managed to mount it), as that same UUID.

fyi, I use GRUB to boot to a /boot partition, which is then supposed
to mount the array. Is there anything obvious that I have missed? Do I
have to rebuild the initrd, or something like that? (since it seems
the fallback shell doesn't preserve any changes to /etc/mdadm.conf).

Let me know if more information is needed.

--
Shaun Reich,
KDE Software Developer (kde.org)
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html

Re: Raid5 boot failure after grow

On 07/19/2011 06:43 PM, Shaun Reich wrote:
> Hello,
>
> I'm using Fedora 15 and I added an additional disk to my raid5 setup
> and grew the array to encompass it. It all worked fine and rebuilt and
> everything. However, now when I boot my computer I am dropped to a
> shell, stating "dracut" (I wonder what that means). Anyways, when in
> here I can cat /etc/mdadm.conf and it appears to be an old
> configuration file. It lists only 3 raid devices in that
> configuration. I now have 4.

dracut is the "new" cross-distribution initramfs environment. I'm using it in my gentoo boxes.

> If I run mdadm --assemble --scan after removing the config file, it
> will automatically find my array and all is well. But this doesn't
> solve the fact that it doesn't boot to it.
>
> dmesg reports "no root device
> 'block:/dev/disk/by-uuid/933ac67-98f1-481f-934e-c9af492ba28c ' found"
>
> This is particularly interesting, because blkid lists /dev/md127
> (after I've managed to mount it), as that same UUID.
>
> fyi, I use GRUB to boot to a /boot partition, which is then supposed
> to mount the array. Is there anything obvious that I have missed? Do I
> have to rebuild the initrd, or something like that? (since it seems
> the fallback shell doesn't preserve any changes to /etc/mdadm.conf).

Yes, your initramfs needs a current copy of your mdadm.conf file, if it has any at all.

HTH,

Phil
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html

Re: Raid5 boot failure after grow

Do you happen to know how to rebuild it when I can't boot into it?
I've tried booting into a livecd, mounting the raid volume and the
/boot, running 'dracut --mdadmconf "initramfs-$(uname -r).img" $(uname
-r)', copying the generated initramfs into the separate /boot
partition, and making grub try to boot it, but it simply is not
working.

It seems to be because it's actually using the livecd's kernel and
mods, which is rather unfortunate and would probably be why it is
unable to boot. I've also tried simply doing it from the livecd (with
the proper mdadm.conf) and then copying that file (from the livecd),
into /boot. That at least gets me from a blank blinking cursor, to a
message about a failed modprobe attempt for /lib/modules, prior to
falling back to a shell.

Ideas?

--
Shaun Reich,
KDE Software Developer (kde.org)
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html

Re: Raid5 boot failure after grow

On 07/20/2011 12:39 PM, Shaun Reich wrote:
> Do you happen to know how to rebuild it when I can't boot into it?
> I've tried booting into a livecd, mounting the raid volume and the
> /boot, running 'dracut --mdadmconf "initramfs-$(uname -r).img" $(uname
> -r)', copying the generated initramfs into the separate /boot
> partition, and making grub try to boot it, but it simply is not
> working.

I'm not familiar with Fedora, so I don't know what tools it has on the liveCD. Look for a tool that will mount everything for you then put you in a chroot. Then run the appropriate update-initramfs utility within the chroot.

> It seems to be because it's actually using the livecd's kernel and
> mods, which is rather unfortunate and would probably be why it is
> unable to boot. I've also tried simply doing it from the livecd (with
> the proper mdadm.conf) and then copying that file (from the livecd),
> into /boot. That at least gets me from a blank blinking cursor, to a
> message about a failed modprobe attempt for /lib/modules, prior to
> falling back to a shell.
>
> Ideas?

HTH,

Phil
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html

Re: Raid5 boot failure after grow

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi,
did you actually chroot into the grown array before updating your initramfs image?

BTW what is on the raid5? Is it the / fs?

Cheers,
- -N

On 07/20/2011 07:39 PM, Shaun Reich wrote:
> Do you happen to know how to rebuild it when I can't boot into it?
> I've tried booting into a livecd, mounting the raid volume and the
> /boot, running 'dracut --mdadmconf "initramfs-$(uname -r).img" $(uname
> -r)', copying the generated initramfs into the separate /boot
> partition, and making grub try to boot it, but it simply is not
> working.
>
> It seems to be because it's actually using the livecd's kernel and
> mods, which is rather unfortunate and would probably be why it is
> unable to boot. I've also tried simply doing it from the livecd (with
> the proper mdadm.conf) and then copying that file (from the livecd),
> into /boot. That at least gets me from a blank blinking cursor, to a
> message about a failed modprobe attempt for /lib/modules, prior to
> falling back to a shell.
>
> Ideas?
>
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iQEcBAEBAgAGBQJOJ/osAAoJEDFLYVOGGjgXkJQH/2HSfCiRMpTqCy4EPzKE udoi
Wa06t862E0HXJuTauHFmtRAYvFV82G7TaRxSPznSDrKsit7FTMPSdzSWGUf9 JDHO
miFSPaENdB/H6qV1FA1yYX3KcARaihAlqbDX2OcMFaLFr5krRpy8FZ8gBL45 3JYw
SlYl2/UP3q8r+TuobCuy8bwWV3Xv8Iw6UO0SiYZYZAELaASRcnNnnFw+wb21 VZI7
HKItkpO6mk/UkQjgbGykSOQjLLBSFUx9MLU1FI+nOwDnlzyyb5/Jnv3zUjWh OC7K
lBIz3Sq7E2Jp6XKqZ9kKtTxebRxMybeEsUWJ3GTwQ5HrUryywZ4uGuntP9lW cJY=
=nxFq
-----END PGP SIGNATURE-----
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html

Re: Raid5 boot failure after grow

Yeah, / is on it. I just managed to fix it by running chroot and
dracut --mdadmconf. My issue was that I was using dracut --mdadmconf
$(uname -r)...which was trying to use the livecd's running kernel.
Right when I specified manually the newest kernel for my raid system,
it worked. Thanks.

--
Shaun Reich,
KDE Software Developer (kde.org)
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html