RAID 0 Array disappeared

[Kr1z]

After a restart the other day, my first drive in the array starting showing up as "Non-Raid Disk". After messing around with the BIOS I could not get it to boot back into Windows, and now both my drives are showing up as "Non-Raid Disks" in the ich10r ultility, which doesn't let me do anything beside create another array, and it prompting to delete everything, which I obviously choose not to do. I been reading up on the ich10r controller, and added a third non raid HDD to install Windows on, and finally got the Intel Matrix Storage Mananger installed.

Both the disks in my array show up in Windows and it seems like the data is still on there but the raid boot configuration is messed up and now I'm attempting to rebuild the same RAID configuration, but I ran into some trouble as its asking me to delete the data on the second disk which was in my array.

I'm just lost on where to go with this anymore...
How can I rebuild a RAID array without deleting any information.
The array was spilt into a 500GB partiation and a 1330GB partiation over two drives both using a 128 stripe size.

 

[omniscence]

Did you try to recreate the RAID0 array in the bios? I vaguely remember that just recreated a RAID0 that failed without being prompted to overwrite a disk. That happend after I accidently booted the system with one of the SATA cables detached which caused the BIOS to disassemble the RAID0. To be completely sure you should make a device level backup of both disks before you overwrite something. On the other hand irrecoverable data should never be placed on a RAID0.

 

[Kr1z]

Did you try to recreate the RAID0 array in the bios? I vaguely remember that just recreated a RAID0 that failed without being prompted to overwrite a disk. That happend after I accidently booted the system with one of the SATA cables detached which caused the BIOS to disassemble the RAID0. To be completely sure you should make a device level backup of both disks before you overwrite something. On the other hand irrecoverable data should never be placed on a RAID0.

This is what my BIOS shows as options.

 

[sub.mesa]

If the data on your RAID is truely important, i can assist you with FreeBSD RAID0 recovery. But you can also boot Ubuntu Linux and see if it can mount your volume, but since some disks have missing metadata it likely will not be auto-detected properly.

The trick of deleting the volume and re-creating it could work. Note that if it was RAID5 this procedure would be really dangerous. It will only work if the exact same parameters are used: stripesize, raid level and disk order. This last properly is often unknown to the user, and re-creating an array can sometimes yield a different disk order, depending on the implementation. That means the array would be corrupted when you write to it, and show as partially corrupt when reading it. Either way, be careful if this important data. It's easy to make recoverable data unrecoverable by user actions.

You just had a regular 3-disk RAID0, right?

 

[Kr1z]

If the data on your RAID is truely important, i can assist you with FreeBSD RAID0 recovery. But you can also boot Ubuntu Linux and see if it can mount your volume, but since some disks have missing metadata it likely will not be auto-detected properly.

The trick of deleting the volume and re-creating it could work. Note that if it was RAID5 this procedure would be really dangerous. It will only work if the exact same parameters are used: stripesize, raid level and disk order. This last properly is often unknown to the user, and re-creating an array can sometimes yield a different disk order, depending on the implementation. That means the array would be corrupted when you write to it, and show as partially corrupt when reading it. Either way, be careful if this important data. It's easy to make recoverable data unrecoverable by user actions.

You just had a regular 3-disk RAID0, right?

That would awesome if you can assist me, yea it was pretty important information, I had been trying to get rid of the raid 0 array, but I never got the chance to buy a single 2tb hdd to transfer everything over.

I had a regular 2 disk raid 0 the 2 1tb samsung ones, that 80 gb drive is another drive I threw in yesterday to install windows to get back online.

 

[omniscence]

I think just recreating should work. There is no reason for a RAID0 implementation to overwrite anything except a superblock. Like submesa suggested, using an ubuntu live cd to reassamble the array can work. The support for Intel ICH raid in mdadm is rather complete.

If you have really important data on the drives I strongly advise doing a backup first, even if you have to buy a 2 TB drive. As you said, you will need another drive in any case to get your data off the RAID0.

 

[sub.mesa]

FreeBSD recovery procedure is read-only; not a byte will chance on your physical drives.

To use this method, i need to login to SSH to a running FreeBSD system with your RAID disks connected. That means you need a separate system disk with FreeBSD installed.

Perhaps the simplest method is to write my Mesa binary image to a pendrive, as it contains a full FreeBSD distribution capable of recovering your data. Instructions are on this page:
http://submesa.com/mesa

If you have FreeBSD up and running, you need to forward port 22 SSH (TCP) to the IP-address of the FreeBSD box. It uses DHCP by default, and you can view the IP with 'ifconfig' after being logged in with root (password: mesa).

If you get a 'mountroot' prompt, then enter "ufs:/dev/label/MESA-8.1" and hit enter, note: this is case sensitive.

 

[Kr1z]

If the data on your RAID is truely important, i can assist you with FreeBSD RAID0 recovery. But you can also boot Ubuntu Linux and see if it can mount your volume, but since some disks have missing metadata it likely will not be auto-detected properly.

The trick of deleting the volume and re-creating it could work. Note that if it was RAID5 this procedure would be really dangerous. It will only work if the exact same parameters are used: stripesize, raid level and disk order. This last properly is often unknown to the user, and re-creating an array can sometimes yield a different disk order, depending on the implementation. That means the array would be corrupted when you write to it, and show as partially corrupt when reading it. Either way, be careful if this important data. It's easy to make recoverable data unrecoverable by user actions.

You just had a regular 3-disk RAID0, right?

I actually went and downloaded Ubuntu and to my suprise it showed up.

After that I decided to restart and I noticed my 2 RAID array started to boot up again. I was in Windows and restarted again and it disapeared... Ran Ubuntu again then restarted then my raid array popped up again and started to boot, and this time I let it to the checkdisk scan and it seems to have fixed itself.

Anyways I ordered another 1 TB hdd to move over all my files, I am done with RAID after this huge scare.

Thanks for the help, Ubuntu saved my life, and after messing around in it for a while its a really cool OS.

 

[sub.mesa]

Hey, that sounds great! Apparently there was enough metadata left to activate your array and grant Ubuntu access. Not sure what caused you to be able to boot into Windows again. But hey it worked, now backup your data ASAP!

I also recommend to check the SMART data of each HDD, and look for:
- Current Pending Sector (bad sectors UNFIXED; very dangerous; should be zero)
- Reallocted Sector Count (bad sectors fixed and no longer cause trouble)
- UDMA CRC Error Count (errors due to bad cabling or EMI)

Download HDTune and click the Health tab to display this kind of information. Look for the last column, labeled as "data" for the actual value for the above mentioned properties; not the threshold values themselves.

I highly recommend to setup a good backup solution. With current high capacity disks with low price per GB, there is no real reason not to, especially if the data to backup actually isn't that large.