Just to report that the method outlined above does work even for more complex RAID set ups (in my case two different RAID volumes - one RAID10 and one RAID5 - spanning the same disks).
As mentioned, some means of backing up and restoring the RAID metadata stored on the disks would be a welcome addition to the software (and a less scary way of recovering from this problem).
Incidently, this happened to me shortly after upgrading to 8.9 - I don't know if this problem is related to the other issues people are seeing with 8.9?
This happened to me recently on a new install after Windows bluescreened. Only one drive was removed from the array so I was able to have the Intel software reconstruct the last drive. However, reconstructing the 1TB drive took almost 4 days during which my computer was nearly unusable due to the constant and apparently high priority disk activity. Intel really needs to address this issue. (And seriously... 4 days? It takes less than one for other software to perform a multi-pass wipe on this size drive.)
Thanks to the one that gave the "solution"
I can't try it because I have more factors to take into account and since I had backup, I will restore to a new drive.
Anyway I would like to comment here and there about the ICH RAID.
We did a 5disk RAID 5 of 750GB each, and bought a SPARE. The computer has a UPS to avoid damage due to the lack of battery backup.
After some time, a drive failed, but it didn't failed as INTEL planned it to fail. It just got some damaged sectors. We noticed just by chance, since trying to uncompressing a tarball was giving errors.
We checked the MD5 of the tarball many times, and each time it was different.And so was the MD5 for older files. The disk, when cold, was reading OK.
Intel Matrix Storage was showing that the members and the volume were OK.
Intel Chat Support incorrectly blamed the Power Supply. I contacted again, when the PSU was changed. They said that it was OK for the ICH to report good Volume and Members when there was no disk missing. But I ask... how on earth a RAID 5 is not telling the user that the readings are wrong?
One interesting thing about this, is that when updating the BIOS to 8.9 all members said in red "Incompatible" but booting (in another drive which is not part of the array) could read/write the RAID and not be able to enable the Write-Back Cache.
Putting the whole array into another exact same motherboard with the old BIOS, all members were OK. And had the same behaviour. Putting the drives back to the old motherboard with the 8.9 BIOS, all members OK again.
So after many times of trying to verify and repair and backing up data (wrong, but, something is better than nothing), I had to unplug the drives and connect them to an ICH (not R) that doesn't recognize them as RAID, since any other ICH set as AHCI will see the RAID member as such instead of a single drive.
Then I could see the SMART data from the drive. Which is hidden when using RAID, and Intel does not provide a tool that allows to see there or to warn the user.
After that, one drive was kicked, and since I don't know the drive order and one is kicked and another is failed, I've decided not to try the steps above and just recover the weekly backups.
I'm writing here just to warn everyone, that the ICH will only help you if one drive is DISCONNECTED from the RAID. But for every other issue that might occur with your data, ICH will not help.
Thank you adgriffi and to adamsap on extremeoverclocking.com
The solution you posted also saved my raid 5 array.
I lost 2 drives out of a 4 disk set after the CMOS was reset on my abit IP 35 Pro after installing new memory.
I followed your steps and was able to restore the partition and save my data.
Hmm it looks like the solution to my problem is here aswell. However a few things are different for me. I have a RAID 5 array of 3 drives, 2 are no longer seen as RAID members. I asume this doesn't really matter. The bigger problem however is that I am unable to boot with the RAID drives in their original position. It hangs on loading windows. Its odd because the RAID drives are non-bootable and my OS is on an intel ssd.
When I change the order in which the ssd and the hard drives are connected so that the ssd comes first I can boot again. But that way the solution provided here will not work. It is possible it will boot after I reset the remaining RAID drive to non-RAID but i'm not sure what's causing the problem. Anyone have an idea why my pc will not boot with the RAID drives connected before the boot drive (as it always has been) ?
I also don't have a clue what the order of the RAID 5 discs was, apart form which ports they were connected to. But i'll just try untill I find the right combo. 6 possible combo's is not that much anyway.
My suggestion would be to try to boot with all the RAID drives "disconnected" (physically) and after boot, connected them again.
Also, for you and the rest of the people that may come here. I've found some days ago a FREE? software tool called "Raid2Raid"
It's from DiskInternals so I believe it's safe (in the virus/ads/etc way) and I have not tried it yet. Seems to be free, and there is a paid tool "Raid Recovery"
Message was edited by: tecnica like 10 times
Ye I already tried that. Reset the member disk to nonraid and rebuild the raid. It would boot normally after this. I recovered my lost partition aswell with testdisk. However, when I rebooted after this checkdisk from windows started running (for like 4 hours). When it was finaly done I found that about 60% of my data had made it. At least thats what I thought. When I tried if the files were still working I found out that EVERYTHING is now corrupt and unusuable. In other words, all my data is still lost. The whole 2TB
Its really strange. I have no idea how this happened. The hard drives are not broken (I checked).
With each member independently connected in non-RAID mode, or in a motherboard that doesn't support Intel RAID, check the SMART data for each drive.
You can use SpeedFan to do so. Today drives are self-healing and that means that they will not report any error to the controller or OS. Sometimes you will find that there are numbers on SMART (I can't remember the names of the fields) that indicate that the drive DID or is about to move a cluster somewhere else because it's defective. And that some others are pending. Those pending corrections will happen the day the drive can read enough data to correct the error.
But that may never happen. So if you scan the drive surface everything will look OK to the OS and your Surface Scanning software while there are broken files. That's what happened to me. And ICH is not able to detect that at all. That renders it useless. And the worst matter is that your files will corrupt and you will not notice unless you see broken files or garbled data or run MD5 periodically over the files.
many thanks for your posting.
did a bios update of my gigabyte board and after reboot 2 disks of my 5 disk raid 5 had the state "non raid members". felt slightly panic!
ask google, found your posting, followed your instructions and saved my ***.
thanks a million
I have the exact same problem and followed every steps... BUT, when I get to step 6 (rebooting vista), my old Raid 5 partition still doesn't appear... and worse, the Intel matrix utility tries to initialize it...
For some reason, my array is still seen as "uninitialized" like it's brand new one... Even though I ran the testdisk program, found my old Raid partition, and am even able to manualy read and copy old files... I did not forget to ask testdrive to "write" the partition data....
Anyone would have n idea of what I'm missing?
This is an old thread, but if somebody is still out there who can help me, I'm experiencing this very same crisis. I reset CMOS on my Gigabyte EX58-UD5 and lost two members of the RAID. I have an OS drive independent of a 4GB RAID 5 (five, 1TB disks) and it boots just fine. Upon finding so many success stories of this Rebuild/TestDisk solution, I decided to give it a try. I reset all disks to non-raid, then rebuilt an identical array without problems. The one thing that scares me is that when I boot into Windows, the Intel Storage Matrix 8.8 software automatically starts to "Initialize" the array. Now, from what I can gather, the initialization within the Intel software re-calculates parity accross all stripes and is not "supposed" to erase any data. When you repaired your array, did you allow the Intel software to complete initialization before going into TestDisk to restore partition tables? My fear is that is if the software thinks the drives are empty, it will overwrite existing sectors with meaningless parity data. Even with Windows 7 in safe mode, the initialization continues. I've shut the PC down until I can get further guidance on this. One more time, my question: Is it safe to allow Intel Storage Matrix to initialize the new array prior to running TestDisk?
Thank you adgriffi!
You saved all my data.
I have 3 x 1.5TB drives in RAID5 on a ASUS P6T V2 deluxe X58 ICH10 based board.
I went to check my PC one morning after leaving it on all night and windows wasn't responding at all was like it was dead, so I power cycle it and the bios comes up saying a long the words of "overclocking failed, hit F1 for setup defaults or F2 to change settings" I just hit F1 to load setup defaults thinking I don't need the overclocking even though it has been rock solid for 10 months.
I boot into windows (I have my main OS of Windows 7 on OCZ 60G SSD) and it keeps popping up and asking me if if I want to format "D:", thats weird I thought so I load up file explorer and I have lost my 3TB raid 5 setup.
I go back into the bios and find that Raid is turned off and the regular IDE mode is on. I turn raid back on and see the intel matrix controller raid stuff during boot up (ctrl+I) saying 2 of my 3 drives that are normally in raid 5 are "non-Raid" member drives.
I followed the instructions resetting them and rebuilding it exactly like it was, I had my RAID5 on 128kb stripe before and after it worked, I noticed that the default is 64kb. Luckily the menu of the Intel Matrix raid thing showed my only member drive as having 128kb or else I would of just assumed it was 64kb.
After booting into windows I loaded up the TestDisk.
I used TestDisk as the example shown and I got all my data back. 3 Cheers to TestDisk as well.
I was originally quite angry at Intel/Bios ppl/Microsoft because the whole thing broke so easily, the whole point of my Raid5 setup was so that my data would be robust, I didn't think rebooting my PC and choose "F1" for bios defaults would cause me to loose my data (if I didn't find this thread).
Now I got it all back I am more even tempered, I think one of the core problems here is how resetting your MB bios can greatly affect your Intel Raid configuration, (changing it from Raid mode to simple IDE mode), and ending up with 2 "non-Raid" member drives?
I don't know if this is what caused the whole problem but it can't help, I think it would be safer if these things were completely seperated some how and the regular MB bios shouldn't have such power over if the Intel Raid configuration exists or not.
Anyway I am just so happy I got all my data back, I thought I was completely stuffed for a while there...
Has anyone seen the New IT company which has just been launched they have brilliant prices and offer home users and Businesses excellent and competitive rates. This is going to be big in the UK only £9.99 a month for remote support, telephone support and if necessary attendance at your premises for there GOLD package . Check out their web site.
Dr Fix Pc Ltd