Problems with RAID build in ubuntu Server 10.04

[James Dugger]

Matt, Eric thanks for your thoughts.  Sorry for the confusion my mdadm code
reference was wrong the device names actually updated from first install to
reboot.  The boot drive is the smaller 500GB HDD (not a part of the raid).
Ubuntu renamed it to sde from sda after reboot. I found an ARRAY device call
for md3 in the mdadm.conf file.  Once I renamed the mdadm.conf file and
rebooted, a cat /proc/mdstat showed no raid devices active.  I then zeroed
the drives using dd and started over. I built a new raid10 with the device
name md4 (just to be sure).  So far the mdstat recognizes the array and
--examine shows no errors in the array.

Thanks again for you thoughts.

On Fri, Oct 21, 2011 at 11:06 AM, Matt Graham <danceswithcrows at usa.net>wrote:

> From: Eric Shubert <ejs at shubes.net>
> > On 10/19/2011 01:06 AM, James Dugger wrote:
> >> fdisk -l gives the following:
> >> /dev/sda1 1 121601 976760001 83 Linux
> >> /dev/sdb1 1 121601 976760001 83 Linux
> >> /dev/sdc1 1 121601 976760001 83 Linux
> >> /dev/sdd1 1 121601 976760001 83 Linux
>
> >> /dev/sde1 * 1 32 248832 83 Linux
> >> /dev/sde2 32 60802 488134657 5 Extended
> >> /dev/sde5 32 60802 488134656 8e Linux LVM
>
> So sde contains /boot and / , while you'd like to have sda..sdd contain the
> RAID.  This should work, but for some reason mdadm says:
>
> >> 0 0 8 49 0 active sync /dev/sdd1
> >> 1 1 8 33 1 active sync /dev/sdc1
> >> 2 2 8 17 2 active sync /dev/sdb1
> >> 3 3 8 65 3 active sync /dev/sde1
>
> I'm very surprised anything's working at all if it's trying to use sde1 as
> a
> component of the RAID.
>
> >> Notice the md3 device at the bottom of the fdisk print out.
>
> md devices can contain partition tables if you really want them to.  Some
> people do this; I wouldn't.
>
> > Looks like from the --examine that the device assignments (/dev/sd?)
> > have moved around since the array was created (sda belongs to an array
> > consisting of d,c,b,e).
>
> Hm.  I thought mdadm went by UUIDs within the RAID superblocks, not
> partition
> names.
>
> > Have a look at:
> > # ls -l /dev/disk/by_id
> > and it'll show which drives are assigned to which /dev/sd? letter.
> >
> > Then (w/out rebooting) take another crack at clearing things out, with:
> > # mdadm --zero-superblock ...
>
> > Then re-create/build the array.
>
> This may not work properly if mdadm has the superblocks tangled up.  You
> could
> zorch the superblocks yourself with dd, something like "dd if=/dev/zero
> of=/dev/sda bs=1M seek=1013760" , repeated for each of the disks.  (That's
> "seek 990G out and start writing zeroes", it'll take a lot less time than
> dding /dev/zero over the entire 1T disk.)  Then stop md3 (if possible
> without
> rebooting), then recreate it, using the right options for RAID10 and the
> right
> disk names.  I'd change the partition types of the softRAID components to
> 0xfd
> too, just because that makes it a little clearer as to what's going on, but
> that may be old-school or deprecated now or something.
>
> Have a rescue CD handy if your /boot has been eaten by this mdadm
> misadventure.  That's workaroundable, just not usually all that fun.
>
> --
> Matt G / Dances With Crows
> The Crow202 Blog:  http://crow202.org/wordpress/
> There is no Darkness in Eternity/But only Light too dim for us to see
>
> ---------------------------------------------------
> PLUG-discuss mailing list - PLUG-discuss at lists.plug.phoenix.az.us
> To subscribe, unsubscribe, or to change your mail settings:
> http://lists.PLUG.phoenix.az.us/mailman/listinfo/plug-discuss
>



-- 
James
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.PLUG.phoenix.az.us/pipermail/plug-discuss/attachments/20111022/38fa0d12/attachment.html>