Intel Raid ISM not recognized, mdadm can't assemble

[Bill64]

Hi there,

Centos 7.3.

Have a BIOS RAID 1 created on an X99P-SLI.
It worked with Centos 6.
C7 doesn't recognise it. If I run:

Code: Select all

# mdadm --assemble --scan –verbose

mdadm: looking for devices for further assembly
mdadm: no recogniseable superblock on /dev/dm-2
mdadm: no recogniseable superblock on /dev/dm-1
mdadm: no recogniseable superblock on /dev/dm-0
mdadm: Cannot assemble mbr metadata on /dev/sdg2
mdadm: no recogniseable superblock on /dev/sdg1
mdadm: No OROM/EFI properties for /dev/sdg
mdadm: no RAID superblock on /dev/sdg
mdadm: Cannot assemble mbr metadata on /dev/sdf2
mdadm: no recogniseable superblock on /dev/sdf1
mdadm: No OROM/EFI properties for /dev/sdf
mdadm: no RAID superblock on /dev/sdf
mdadm: No arrays found in config file or automatically

Which is confusing because both drives seem to be healthy:

Code: Select all

# mdadm -E /dev/sdf
/dev/sdf:
          Magic : Intel Raid ISM Cfg Sig.
        Version : 1.1.00
    Orig Family : b3611c18
         Family : b3611c18
     Generation : 00005a13
     Attributes : All supported
           UUID : b8c4cae4:035d4d8f:efc2d16c:cc6beb19
       Checksum : 274c5983 correct
    MPB Sectors : 1
          Disks : 2
   RAID Devices : 1

  Disk00 Serial : WD-WMATV4408793
          State : active
             Id : 00000004
    Usable Size : 1953519880 (931.51 GiB 1000.20 GB)

[Volume1]:
           UUID : 55d64cc1:d79a3c04:5ab6ca1b:60bf5a2b
     RAID Level : 1
        Members : 2
          Slots : [UU]
    Failed disk : none
      This Slot : 0
     Array Size : 1953519616 (931.51 GiB 1000.20 GB)
   Per Dev Size : 1953519880 (931.51 GiB 1000.20 GB)
  Sector Offset : 0
    Num Stripes : 7630936
     Chunk Size : 64 KiB
       Reserved : 0
  Migrate State : idle
      Map State : normal
    Dirty State : clean

  Disk01 Serial : WD-WMATV4388773
          State : active
             Id : 00000005
    Usable Size : 1953519880 (931.51 GiB 1000.20 GB)


# mdadm -E /dev/sdg
/dev/sdg:
          Magic : Intel Raid ISM Cfg Sig.
        Version : 1.1.00
    Orig Family : b3611c18
         Family : b3611c18
     Generation : 00005a13
     Attributes : All supported
           UUID : b8c4cae4:035d4d8f:efc2d16c:cc6beb19
       Checksum : 274c5983 correct
    MPB Sectors : 1
          Disks : 2
   RAID Devices : 1

  Disk01 Serial : WD-WMATV4388773
          State : active
             Id : 00000005
    Usable Size : 1953519880 (931.51 GiB 1000.20 GB)

[Volume1]:
           UUID : 55d64cc1:d79a3c04:5ab6ca1b:60bf5a2b
     RAID Level : 1
        Members : 2
          Slots : [UU]
    Failed disk : none
      This Slot : 1
     Array Size : 1953519616 (931.51 GiB 1000.20 GB)
   Per Dev Size : 1953519880 (931.51 GiB 1000.20 GB)
  Sector Offset : 0
    Num Stripes : 7630936
     Chunk Size : 64 KiB
       Reserved : 0
  Migrate State : idle
      Map State : normal
    Dirty State : clean

  Disk00 Serial : WD-WMATV4408793
          State : active
             Id : 00000004
    Usable Size : 1953519880 (931.51 GiB 1000.20 GB)

Am I doing something wrong?
What am I supposed to do here?

Many thanks,
Bill

[hunter86_bg]

Have you tried to point to "mdamd" which are the components of your raid?
If it can't find them , this doesn't mean it can't assemble them:

ASSEMBLE MODE
Usage: mdadm --assemble md-device options-and-component-devices...

You can pass by name, uuid,super-minor

P.S.: It won't be a bad idea to create a backup of those disks, just to be on the safe side.

[Bill64]

Thanks for the reply hunter86_bg

Running:

Code: Select all

# mdadm --assemble /dev/md0 --uuid b8c4cae4-035d-4d8f-efc2-d16ccc6beb19

does nothing. It takes about a second and a half then an empty prompt. No error or any other message. And md0 does not appear in /dev.
I run it as su -.

Am I doing something wrong?

Thanks,
Bill

[hunter86_bg]

I suspect that its either a kernel or mdadm issue.

Can you check the solution proposed in this Arch Forum and share your experience.

Don't forget to backup!

[Bill64]

Thanks for the reply hunter86_bg

... share your experience.

Shouldn't have asked that ... but if you did:
Absolutely Awful!

I went to that page and understood, update the software.

Code: Select all

sudo pacman -Syu   # update to latest

I did. After some lengthy download and reboot I received a "Something went wrong, log out and try again" screen. Tried it several times - same error. Luckily, Centos programmers know they stuff it up regularly so there is a way to boot the previous install.

The rest of it is just gobbledygook to me:

Code: Select all

sudo vim /usr/lib/udev/rules.d/64-md-raid-assembly.rules   # and prepend '/usr/bin/env IMSM_NO_PLATFORM=1 ' to usages of '/usr/bin/mdadm'
sudo mkinitcpio -p linux   # rebuild your init image

Would someone care to translate it to Centos commands please?

Also

Code: Select all

https://bugs.archlinux.org/task/46295

says

You can use IMSM_NO_PLATFORM environment variable (see man mdadm)

I looked at the mdadm man but there is no mention of IMSM_NO_PLATFORM.
I see, the above commands are setting this environment var. but still, it doesn't make sense how to do it.

Also, I understand, there supposed to be an mdadm.conf file in /etc. There isn't one.

Thanks
Bill

[TrevorH]

Is there any data on the array that you need? If not, perhaps you should just try recreating it at the BIOS level. I know there have been some changes in mdadm that made it refuse to assemble arrays where there was an underlying error in the RAID metadata maintained by the BIOS RAID functionality and recreating it would fix those if they are present.

[hunter86_bg]

So, have you tried to replace in "/usr/lib/udev/rules.d/63-md-raid-arrays.rules" ?
From:

Code: Select all

IMPORT{program}="/usr/sbin/mdadm --detail --export $devnode"

To:

Code: Select all

IMPORT{program}="/bin/env IMSM_NO_PLATFORM=1 /usr/sbin/mdadm --detail --export $devnode"

And then update the initramfs:

Code: Select all

cp /boot/initramfs-$(uname -r).img /boot/initramfs-$(uname -r).img.$(date +%m-%d-%H%M%S).bak && dracut -f -v

[Bill64]

Thanks for the replies.

@TrevorH. It was created with the BIOS. See further down.

@hunter86_bg. I did what you said.
Editing the /usr/lib/udev/rules.d/63-md-raid-arrays.rules was successful.
The cp /boot/initramfs-$(uname -r).img /boot/initramfs-$(uname -r).img.$(date +%m-%d-%H%M%S).bak && dracut -f -v command installed several things but not 'busybox' and skipped several others.

Just for completeness, commands are run as 'su -' .

The GUI become unstable. Windows irregularly freeze - can't move them - Terminal and Mozilla - the main bar at the top works then freezes, can't select any application. Double clicking the 'Home' dir is unresponsive.
First it all works then - it seems - after running the mdadm command in the terminal they GUI freezes. Have to confirm if this is actually the case.
Still, a window can be selected and I can enter a command in the terminal.

Re-booted several times - same behaviour.
If the terminal is open, I can enter a command. Tried

Code: Select all

# mdadm --assemble --scan --verbose

result is the same, can't find any arrays.

I tried what TrevorH suggested. Deleted the array in the BIOS, save, reboot. Then re-create it, save, boot.
Same result. mdadm can't find the drives. There are two individual disks presented under the 'Home' dir but not mounted.

The GUI is still unworkable. I'll wait with reinstalling CentOS till you reply. In case you want to check/do something.

Thanks,
Bill

[hunter86_bg]

I'm sorry to hear that it ddin't work.As I have no more ideas - you have to choose whether to roll back to CentOS6 (as the raid worked there) or to create a software raid from scratch:

1) If you have a backup and want to move to the CentOS 7 -> remove the raid from the BIOS, run a bootable DVD/USB - > Troubleshooting and create a software raid via mdadm.
2) You have no recent backup or you have decided to roll back to CentOS 6-> Install CentOS6 as you have confirmed that the raid did work. Than make a backup and decide if you wish to take the first option or to stay with the current install.

[Bill64]

Thanks again for the efforts!

After lots of reading it turns out, mdadm is the wrong approach here; dmraid is to be used here:

Code: Select all

# dmraid -ay

assembles the array and it can be mounted, no problem.

The question now is, how do I 'automate' invoking this - and the mounting of the array - during boot.
I guess, mounting the array would be in /etc/fstab once the 'block device' is created.
But how to do that during the boot, escapes me.

Guess a script could be written/run once logged into the GUI, it's just that is so ... aaammm ... is there a more elegant solution?

There is no system on this array, only data.

Many thanks,
Bill