I recently purchased a third hard drive, and decided to do a linear RAID with my second one (mounting at /home). I packed away all my data, ran:

mkraid /dev/md0
mdadm --create --verbose /dev/md0 --level=linear --raid-devices=2 /dev/hda1 /dev/hdb1

my /etc/raidtab looks like:

raiddev /dev/md0
raid-level linear
nr-raid-disks 2
chunk-size 32
persistent-superblock 1
device /dev/hda1
raid-disk 0
device /dev/hdb1
raid-disk 1

I made a partition, set up a filesystem, added it to /etc/fstab, and loaded my data back on, and when I rebooted, /dev/md0 didn't mount.

The catch is, running:

raidstart /dev/md0
mount /dev/md0

mounts my drive just fine! RAID is compiled right into my kernel (not as a module), so I'm hoping someone might have some insight as to why it's not automounting at boot.

--Nate

Did you remember to change the partition type to 0xfd using fdisk?
From the Software-RAID-HOWTO

I thought I did, but upon running fdisk, it says:

Warning: invalid flag 0x0000 of partition table 4 will be corrected by w(rite)

But then, the same thing pops up for my other, non-RAID drives... which doesn't seem right.

also, if I haven't run raidstart yet, all it says is:

Unable to read /dev/md0

I'm really lost here.

Hi ,

Any way Partion 4 mus be extended partition, did you made it extended????

so, I figured maybe I took a misstep somewhere, and maybe I should start from scratch... so I copied off my data again, ran raidstop, and then ran:

mkraid /dev/md0

it said that the drives were already part of a RAID array, so I ran:

mkraid /dev/md0 -R

and then:

mdadm --create --verbose /dev/md0 --level=linear --raid-devices=2 /dev/hda1 /dev/hdb1

to which it replied:

mdadm: Cannot open /dev/hda1: Device or resource busy
mdadm: Cannot open /dev/hdb1: Device or resource busy
mdadm: create aborted

my /proc/mdstat shows:

md0 : active linear hdb1[1] hda1[0]
507926912 blocks 32k rounding

what's going on here? why can't I run mdadm on my new array?

mkraid is equivalent to mdadm --create. You need to do one or the other.

okay, so in light of that, I just ran:

mkraid /dev/md0 -R
cfdisk /dev/md0 (I started with a fresh Linux Raid Autodetect partition)
mkfs.ext3 /dev/md0
reboot

during boot, it says:

md: Autodetecting RAID arrays.
md: autorun ...
md: ... autorun DONE.

...

/sbin/e2fsck: Invalid argument while trying to open /dev/md0
/dev/md0:
The superblock could not be read or does not describe a correct ext2
filesystem. If the device is valid and it really contains an ext2
filesystem (and not swap or ufs or something else), then the superblock
is corrupt, and you might try runing e2fsck with an alternate superblock:
e2fsck -b 8193 <device>

...

Mounting non-root local filesystems:
EXT3-fs: unable to read superblock
mount: wrong fs type, bad option, bad superblock on /dev/md0,
missing codepage or other error


and then, it's unsurprisingly not there when I log in... but if I run raidstart and mount, it starts up and mounts just fine and dandy.

I might also add that upon running cfdisk again, it shows the type as "free space" - not as linux raid autodetect. what on earth??

after much deliberating, I realized it was an order-of-operations issue!

instead of running:

mkraid /dev/md0 -R
cfdisk /dev/md0

I ran:

cfdisk /dev/hda
cfdisk /dev/hdb
mkraid /dev/md0 -R

and it worked!

mkraid implies raidstart. As per the howto, you need to change the partition type when the raid device is not running (see 'NOTE' in my first post). By the way, my version of raidtools doesn't mention an 'R' option to mkraid (mkraid -R).

-R means "really-force" on mine... I suppose to keep people from typing -f by mistake?

I guess my confusion arose from the fact that I wasn't supposed to partition the array - I was supposed to partition each drive.

Ah, ok. It sounds familiar now. It's just not mentioned in the man page. Anyway, yeah, fdisk wasn't writing to the MBR.