Intel Matrix storage failure by just unplugging the drive?

[mattropolis]

I have an Intel 965OT motherboard (genuine Intel brand) and am running the embeded matrix storage on RAID 0 with 2 SATA seagate 400gb hd's. This was working great for a long time. I also had a single 200gb drive in the machine not in the raid set. So the 800gb total was in one set (bootable c: w/ xp on it), and the 200gb was by itself (d

I wanted to run an experiment, so I unplugged the raid. Unfortunately, I unplugged one of the 400's and the 200, not the 2 400's. Boot and do the quick dope-slap at the list of discovered HD's on BIOS post. I shut it off and plug the right ones in - didn't even really let it boot. But now the raid doesn't work. Even thought the drives are all plugged back in and nothing has changed on the drives, the matrix storage thing tells me that the RAID set is unbootable and "failed". What the heck? If I just unplug a raid drive and and then plug it in again, it fails and won't let me get my data back!?! Anybody have any ideas on how to get my raid back? It sounds like the raid thinks a drive failed and just needs to be told "no, its ok - try to recover". The bios RAID offers fun things like create/delete/exit which is no help. Can I just delete and recreate the raid with the same drives and not lose the data? Any help would be greatly appreciated - I don't want to have to reconstruct this data...

Thanks

 

[sruane]

Can I just delete and recreate the raid with the same drives and not lose the data?

No. Failed is failed.

 

[mattropolis]

It seems like there should be some way to tell the raid to 'reset'/ignore the last failure and try again. I mean, the data has absolutely not been changed, it's just the motherboard thinking it's dead. Could I do a bios/board reset via jumper and plug the drives back in for 'rediscovery'?

 

[sruane]

It doesn't work that way. The drives array membership is written to the drive itself.

I suggest you email Intel. I emailed them with a Matrix Raid question I had and they were very helpful.

 

[JMecc]

You should be able to recover the data using Active Undelete (http://www.active-undelete.com) .This will be time-consuming & will still require windows reinstall but will at least get your data back.

That's scary though that striped raid could go down that easy.

Jo

 

[sruane]

You might want to triplecheck that the drives are plugged in correctly - especially the power connectors. SATA power connectors fall off pretty easily and you are using electric drives.

 

[mattropolis]

Thanks - I just mailed Intel, we'll see what happens.

Interesting. So the 'failure' data is written to the drive. There has to be recovery software out there that might be able to recover the data. I can boot to the standalone drive, and it shows my 800gb set as 'unpartitioned' so it does 'appear' to windows. I kept pretty good backups, so all I need is a few files. Any tools out there for this sort of thing? I see Runtime Software has some raid recovery tools - any others?

 

[mattropolis]

You might want to triplecheck that the drives are plugged in correctly - especially the power connectors. SATA power connectors fall off pretty easily and you are using electric drives.

Yeah, did that one once in the process.

No, they're all plugged in because the bios sees all 3 drives at Post. In fact, if I boot up in windows, I actually see 800gb as unpartitioned space (I'm booting from the 200gb spare drive). it's right..there...but...i..can't...quite...touch...it...

 

[sruane]

Do you have RAID enabled in the BIOS? You know, that option where you select IDE, AHCI or RAID?

 

[JMecc]

So you can boot to windows now (or are you doing this from another drive to try to access this set)?
Jo

 

[mattropolis]

Yes, RAID is enabled. I tried switching to AHCI and back, but no luck. What is the difference between AHCI and RAID anyway?

 

[mattropolis]

So you can boot to windows now (or are you doing this from another drive to try to access this set)?

I'm using a spare 200gb drive with XP on it - not in the raid. The raid is two 400gb SATA drives. So there are 3 in it total.

 

[sruane]

Yes, RAID is enabled. I tried switching to AHCI and back, but no luck. What is the difference between AHCI and RAID anyway?

Advanced Host Control Interface is single-drive SATA. Otherwise, to use a SATA drive you will me emulating IDE. AHCI supports command queuing.

Redundant Array of Inexpensive Disks allows you to span physical volumes and includes the features of AHCI

 

[cyberlander]

Hi,

I am having the exact same issue. Did you ever find a way to get your drives running in the RAID again or recover the data?

Thanks

 

[Guest]

make sure you plugged the SATA drives into the same sata ports on the mobo. Also try to do a CTL+I I think to get into your RAID manager during boot up. There should be a repair function in there.

 

[jookeer]

I've got a strange problem with ICH9R and 6 RAID5 disks:

In Matrix it says that disk 1-5 are members, and the 6'th disk is missing. That disk is instead listed as a "non RAID" disk. If I right click on this one the only option is "Reset to non RAID" BUT it is already listed as "non RAID". Aaarrg...

The status of the volume is "Failed" although only one of 6 disks are missing and that disk is a former member in the array.

So, is there anyway I can change the status of that disk so it will be forced as a member into the array again, all the data is intact and all disks are new and no physical errors whatsoever.

In the RAID-BIOS this disk is listed as "Spare", the other 5 disks are members in a RAID 5 array, and volume status is "Failed".

When talking to Intel web support they say if volume status is "Failed" they cannot help me in a way that will recover my data.

So, please, anyone knows how to deal with this in a fairly secure way?

 

[pankov]

Guys,
I've found a solution to this "sudden drop-out" problem that Intel Matrix Storage had.

My exact problem was that I had two disks in mixed RAID0-RAID1 configuration for both speed and safety. While I was updating the BIOS of my motherboard (Gigabyte GA-P35-DS4 ICH9R) the matrix storage decided to drop-out one the disks and label it as "Non-Raid" which led to Degraded RAID1 and FAILED RAID0. AS my OS (Windows XP + Windows 7 dual boot) was on the RAID0 I felt "toasted". I did a lot of reading the other day and found a solution that helped me NOT loose ANY data at all
This post here "showed me the light"
http://forums.extremeoverclocking.com/showpost.php?p=3329132&postcount=6

As the OP wrote, deleting RAID arrays in the BIOS (O-ROM) doesn't actually delete your data - only the metadata that tells the Matrix Storage what arrays are configured. It also clears the MBR and partition tables but with a tool like TestDisk you can easily fix this.

This absolutely saved my day (and a couple of nights to be precise).

I hope this helps someone else too

 

[mamw93]

You shouldn't have done that. When you unplug one of the drives you corrupt the RAID. It will be hard to get back the striping without any data loss. All I can suggest is to get a PCI RAID card and put the drives on that so that this doesn't happen again. The thing about using RAID on motherboard headers is that they are easily corrupted and are EXTREMELY hard to get back.

 

[pankov]

mamw93,
who are you talking to? If it's me I have to say that you are wrong. If you unplug a drive from a raid0 array and you don't write any data to the other drive(s) you don't loose any data ... yet. If you put back the drive back your data will be just fine. The problem with Intel Matrix Storage is that it doesn't allow you to easily reattach the drive if it had decided that it's not part of the array anymore. Then you have only one option - to recreate the ABSOLUTELY EXACT SAME array. This way the real data is not changed - only the metadata of the array configuration is rewritten.

 

[sub mesa]

True, but there is another way. Re-creating the array should work, provided the RAID driver does not do things like zero-writing which is very uncommon i guess. It deleting the MBR is bad, anyway. If it like zero-writes the first 20MB you also loose filesystem metadata which can hinder recovery. But in many onboard RAID setups you cannot define the exact disk order that will be used, the RAID driver may re-create the array in the wrong way.

A safer bet would be using software RAID under Linux or BSD to recover without writing anything to the disks. This would allow trial and error without further loss of data. I successfully recovered multiple onboard RAID with this method, where the RAID array was broken but the data was still there and could be successfully recovered.

 

[BushLin]

Thanks for the replies in this post, the re-create array and testdisk combo sorted out a "failed" array due to pushing limits on an overclock.

Didn't want to risk setting the missing disk as spare in the Storage Manager in case that didn't work properly, anyone tried this?

 

[Guest]

Guys,
I've found a solution to this "sudden drop-out" problem that Intel Matrix Storage had.

My exact problem was that I had two disks in mixed RAID0-RAID1 configuration for both speed and safety. While I was updating the BIOS of my motherboard (Gigabyte GA-P35-DS4 ICH9R) the matrix storage decided to drop-out one the disks and label it as "Non-Raid" which led to Degraded RAID1 and FAILED RAID0. AS my OS (Windows XP + Windows 7 dual boot) was on the RAID0 I felt "toasted". I did a lot of reading the other day and found a solution that helped me NOT loose ANY data at all
This post here "showed me the light"
http://forums.extremeoverclocking.com/showpost.php?p=3329132&postcount=6

As the OP wrote, deleting RAID arrays in the BIOS (O-ROM) doesn't actually delete your data - only the metadata that tells the Matrix Storage what arrays are configured. It also clears the MBR and partition tables but with a tool like TestDisk you can easily fix this.

This absolutely saved my day (and a couple of nights to be precise).

I hope this helps someone else too

You sir, saved my bacon with this post! I am running almost an identical configuration to yours (mixed RAID 0 and 1 on 2 disks). Last night, I hibernated my PC, as I always do, and when I turned it on this morning, one of the disks was reported missing and the RAID 1 degraded, RAID 0 failed. I was stumped since the BIOS saw the disk just fine and a full scan with the Western Digital diag tools reported both of them healthy. I also noticed that, after running the WD tools (but not actually changing anything), the missing disk was mysteriously being detected again by the RAID controller, but as a non-RAID disk. I followed the instructions in the link you supplied and BAM!...both partitions are working again! Many thanks, good sir!

For what it's worth, I'm in the process of transferring all of my data to a backup drive and will be deleting both RAID arrays. I've had nothing but problems with them from day 1 and this wake-up call was pretty much the last straw for me. I really have no faith in Intel's crappy implementation and the speed boost seemed to degrade very quickly after I had everything on there, although I fault that mostly to the fact that I was running in mixed mode.

 

[Guest]

I have fixed this problem twice now, I have a raid 5 on the intel matrix and the first time, I'm not sure what the issue was that caused it, but this time, I believe that i accdentally pulled one of the sata cables off one of the HDs. At any rate, the first time, I called Intel who told me I was screwed. Of course, I did not like this and mentioned that the idea of having Raid 5 was so this would not happen.

At any rate, the fix for me was simply to put in a Windows XP Pro disk and begin as if I was installing XP (the server has SBS 2003 on it) I hit r as if I were going into recovery console, then quit the install and the machine then booted. Not entirely sure why, maybe refreshed the f6 drivers?

 

[snoozeroo]

Guys,
I've found a solution to this "sudden drop-out" problem that Intel Matrix Storage had.

http://forums.extremeoverclocking.com/showpost.php?p=3329132&postcount=6

I hope this helps someone else too

I did for me! Thanks for posting the link!