[SOLVED] mdadm - quantum superblock?

[sampablokuper]

Hi folks,

I'm having a problem with a RAID device. I just drafted a forum post about it to start this thread here at linuxforums.org, but when I clicked the "Submit New Thread" button, for some reason I got the following error:

The following errors occurred with your submission:

1. You are only allowed to post URLs to other sites after you have made 15 posts or more.

which is odd, because my post didn't contain any URLs.

So, THIS post that I'm writing now is about two problems. The first problem is that linuxforums.org is a bit broken, because it's claiming my post contained URLs when it didn't.

The second problem is the one to do with my RAID device rejecting the --assemble command. The details of that are in my original draft, which I have resorted to attaching to this post as a text file, to work around the linuxforums.org bug mentioned above.

All help gratefully received!

Sam

[oz]

Hello and welcome!

No, the forums aren't broken, but you can check this thread for a work around on the 15-posts issue, and other information about how to use the forums:

http://www.linuxforums.org/forum/com...ng-forums.html

[sampablokuper]

Ah, thank you! Right, here's the original post, with the offending 'at' symbols removed:
--

Some time ago, when I was learning to use mdadm (which, I guess, I still am; I certainly wouldn't say I've mastered it; you'll see why), I created a RAID 5 device from 3 JBOD (linear) devices. One of those linear devices' constituent drives seems to have died, but I don't see why that should be a problem for the readability of the RAID 5 array as a whole. Anyhow, here's what I'm getting at the terminal:

% mdadm --detail /dev/md0                                      
/dev/md0:
        Version : 0.90
  Creation Time : Sun Oct 19 20:58:07 2008
     Raid Level : linear
     Array Size : 449326912 (428.51 GiB 460.11 GB)
   Raid Devices : 2
  Total Devices : 2
Preferred Minor : 0
    Persistence : Superblock is persistent

    Update Time : Sun Dec 28 03:20:15 2008
          State : clean
 Active Devices : 2
Working Devices : 2
 Failed Devices : 0
  Spare Devices : 0

       Rounding : 64K

           UUID : 9ea504f3:f6185a40:320bb4ed:de379393
         Events : 0.3

    Number   Major   Minor   RaidDevice State
       0       8        0        0      active sync   /dev/sda
       1       8       48        1      active sync   /dev/sdd

% mdadm --assemble --force --verbose /dev/md9 /dev/md0 /dev/md2
mdadm: looking for devices for /dev/md9
mdadm: cannot open device /dev/md0: Device or resource busy
mdadm: /dev/md0 has no superblock - assembly aborted

I'm puzzled why /dev/md0 reports its details with the --detail command, including giving details of its superblock, and yet apparently has no superblock when the --assemble command is run moments later.

If I repeat those commands, the same thing happens consistently: this isn't an intermittent bug.

Any suggestions on how I can re-assemble /dev/md9 gratefully received!

[zanerock]

Usually this means the arrays are assembled, but not active. Check mdstat:

cat /proc/mdstat

If you see the arrays/devices listed, then stop them:

mdadm --stop /dev/md0

Then run the assemble again.

Regards,
Z

[sampablokuper]

Thanks for your reply, Zane.

Usually this means the arrays are assembled, but not active. Check mdstat:

cat /proc/mdstat

Sure enough:

% cat /proc/mdstat
Personalities : [linear] [multipath] [raid0] [raid1] [raid6] [raid5] [raid4] [raid10] 
md9 : inactive md0[0](S) md2[2](S)
      918188736 blocks
       
md1 : inactive sdf[1](S)
      390711296 blocks
       
md2 : active linear sdc[0] sde[1]
      468861952 blocks 64k rounding
      
md0 : active linear sda[0] sdd[1]
      449326912 blocks 64k rounding
      
unused devices: <none>

If you see the arrays/devices listed, then stop them:

mdadm --stop /dev/md0

Ah, unfortunately...

% mdadm --stop /dev/md0
mdadm: failed to stop array /dev/md0: Device or resource busy
Perhaps a running process, mounted filesystem or active volume group?

Now, this is a bit odd, because:

% umount /dev/md0      
umount: /dev/md0: not mounted

Further suggestions welcome!

[zanerock]

I think the problem is md0 is being used in md9 even though md9 isn't running.

But...

Stop md9, then re-assemble it. If md0 is working, no need to stop it. I wasn't paying close attention and used md0 is my first example as illustrative.

[sampablokuper]

D'oh! Seems I'm a bit dozy today. Thanks again!