Raid 5 ( HELP DESPERATE ) non raid drives

[maciekmiszkiel]

....I know I'll be told to not trust RAID as a backup option but unfortunately I did.

I have two Raid arrays in place. Raid 1 for the OS and Raid 5 for storage. Everything was working fine until yesterday. Windows updates took place, my system rebooted and I had no longer RAID 5 array working. When I went into the BIOS my setting for the controller was changed from RAID to IDE. I changed it back to RAID hoping that it would work. No change. I went into the controllers config and 2 out of 4 disks in Raid 5 array are saying that they are not members of the array anymore. I have no option to add them back to the array. I can delete or create a new one or set the disks to non raid like the other two. Apart from that my OS took a dive too. I am not sure what is the right procedure here. So here come my questions and I am open to your suggestions.

1. Am I going to hurt my Raid 5 array by reinstalling the OS on my Raid 1 ?
2. If not should I disconnect the Raid 5 drives before I install the OS again?
3. What are my options for recovering the data from Raid 5 ? Procedures ?

Thanks again.

 

[Cr0nJ0b]

I implore you to take a deep breath...walk away from your PC for an hour or so, and DON'T DO ANYTHING!

If you are like me and still want to get you data back you most likely can. But, again, if you are like me, you might be in a bit of a panic mode right now...and that will not help things at all.

 

[talion83]

Installing an OS on the RAID 1 will have no effect on the RAID 5 in a HW raid situation.

If you are nervous you may choose the wrong drive, then you can disconnect the RAID 5 - otherwise as mentioned it shouldn't have any effect; the OS sees and will treat your RAID just like it would standard drives, so if it sees "two" drives and you select one to use, it uses that one.

For recovery options - it will partially depending upon the RAID controller (which in this case sounds like your motherboard). You should be able to create a new RAID 5, choose the same disks, and it will rebuild based upon the existing meta data. But I don't want to say this WILL work on your systems RAID controller.

 

[Cr0nJ0b]

So if you are back from your walk, let's try to fix things up.

First, we need to know what kind of raid array it was. Intel? what chipset and motherboard? what controller? Software RAID hardware RAID?

I see that it's RAID 5, 4 disks.

Do you know if the drives are good?

DO NOT WRITE ANYTHING TO THE DRIVES. be careful not to load windows on a primary drive while these drives are connected, because you might stamp the partition.

Let's assume you have a standard Intel setup like I do, with a RAID 5 set using the latest Intel Chipset and RST drivers.

You can rebuild the set, BUT...the drives have to be connected to the EXACT SAME CHANNELS! this is key. I screwed this up when it happened to me. If the drives are placed in the exact same order on the same channels, then you can create a new raid array, the same as the original and recover all of the data...unless you have a drive fault for wrote to the drives.

Step 1: get your OS back online. You should be able to boot to the mirrored set without issues. If you need to reload the OS for some reason, do that as a pre-step. UNPLUG YOUR RAID 5 DRIVES!!!. Once you can boot up the OS, do the patches and get it all shiny. This will take a while...but stay calm...you will get your data back.

Step 2: Add your RAID 5 drives back. I would delete the RAID 5 Array and then do a quick test of each drive with the manufacturers drive test. This will make sure that you don't have any horrendous drive issues. If you have the time do full tests, but that will take a very long time...and I'm guessing you are impatient.

Answer 1: No you won't hurt your raid 5 array more than already have. I'm not sure what data you have on the array or how that was incorporated into your OS but it's all recoverable. The caveat here is that if you have the drives plugged in when you reload, Win 7 might stamp the front of your set or a drive in your set with some of it's own metadata and your DO NOT WANT THIS! that's why you unplug the drives until the OS is loaded.

Answer: Yes you can break and rebuild a raid 5 set without killing your data. It seems that Intel will setup the drives in the exact same original state when you rebuild the drives, as long as you don't mess with their port position. If you did mess with the cables, Like I did when this happened to me, then there is another step that you can take to figure things out...but lets hope you didn't.

Step 3: If you did things right and put the drives back in the right order and built the RAID 5 set again with the same options (I used the default block size etc), then you should be able to see the drives in the OS under the disk management utility. DO NOT INITIALIZE OR FORMAT THIS PARTITION! (that should be obvious).

Step 3: Get a partition recovery program and search the disks for a partition. I'm at work now, so I don't have the names of the programs, but there are a couple out there that can recover. TestDisk is one that i used...there are others.

I used this link to get me started with the recovery.

http://www.overclock.net/t/478557/howto-recover-intel-raid-non-member-disk-error

If everything goes right and you choose the right options in test disk, you should be able to reboot and see the drive like before....if that's the case BACK IT UP! I would also recommend testing some files...Open some of the AVI files (I know you have them) and mount some ISOs etc. look at and PIctures you may have etc.

I went through this process and it only paritially worked for me...because I had some more issues. But after a week or so, I resolved everything and got back 90% of all the data. If I knew then what I know now, it would have taken me a day or so, but that's life.

feel free to post more questions. I will monitor this thread for the next few days.

 

[rsutoratosu]

You might want to look into r studio

http://www.r-tt.com/Articles/RAID_Recovery_Presentation/

It has the abililty to construct a virtual raid volume off failed raid 5 drives and let you extract data.

 

[maciekmiszkiel]

Guys, thanks a lot for the replies. I'll try it tonight and see what comes out of it.

My config:
intel i5 760 1st gen 1156 socket
asus maximus III gene
8gb ram

Since it's the asus board I think it uses the latest intel rapid storage technology.

 

[exdeath]

Vendor for the controller should have software to diagnose and confirm broken RAID drives and restore missing member signatures, etc.

 

[maciekmiszkiel]

Vendor for the controller should have software to diagnose and confirm broken RAID drives and restore missing member signatures, etc.

That would be Intel. From what I've heard they are not really helpful.

 

[maciekmiszkiel]

So if you are back from your walk, let's try to fix things up.

First, we need to know what kind of raid array it was. Intel? what chipset and motherboard? what controller? Software RAID hardware RAID?

I see that it's RAID 5, 4 disks.

Do you know if the drives are good?

DO NOT WRITE ANYTHING TO THE DRIVES. be careful not to load windows on a primary drive while these drives are connected, because you might stamp the partition.

Let's assume you have a standard Intel setup like I do, with a RAID 5 set using the latest Intel Chipset and RST drivers.

You can rebuild the set, BUT...the drives have to be connected to the EXACT SAME CHANNELS! this is key. I screwed this up when it happened to me. If the drives are placed in the exact same order on the same channels, then you can create a new raid array, the same as the original and recover all of the data...unless you have a drive fault for wrote to the drives.

Step 1: get your OS back online. You should be able to boot to the mirrored set without issues. If you need to reload the OS for some reason, do that as a pre-step. UNPLUG YOUR RAID 5 DRIVES!!!. Once you can boot up the OS, do the patches and get it all shiny. This will take a while...but stay calm...you will get your data back.

Step 2: Add your RAID 5 drives back. I would delete the RAID 5 Array and then do a quick test of each drive with the manufacturers drive test. This will make sure that you don't have any horrendous drive issues. If you have the time do full tests, but that will take a very long time...and I'm guessing you are impatient.

Answer 1: No you won't hurt your raid 5 array more than already have. I'm not sure what data you have on the array or how that was incorporated into your OS but it's all recoverable. The caveat here is that if you have the drives plugged in when you reload, Win 7 might stamp the front of your set or a drive in your set with some of it's own metadata and your DO NOT WANT THIS! that's why you unplug the drives until the OS is loaded.

Answer: Yes you can break and rebuild a raid 5 set without killing your data. It seems that Intel will setup the drives in the exact same original state when you rebuild the drives, as long as you don't mess with their port position. If you did mess with the cables, Like I did when this happened to me, then there is another step that you can take to figure things out...but lets hope you didn't.

Step 3: If you did things right and put the drives back in the right order and built the RAID 5 set again with the same options (I used the default block size etc), then you should be able to see the drives in the OS under the disk management utility. DO NOT INITIALIZE OR FORMAT THIS PARTITION! (that should be obvious).

Step 3: Get a partition recovery program and search the disks for a partition. I'm at work now, so I don't have the names of the programs, but there are a couple out there that can recover. TestDisk is one that i used...there are others.

I used this link to get me started with the recovery.

http://www.overclock.net/t/478557/howto-recover-intel-raid-non-member-disk-error

If everything goes right and you choose the right options in test disk, you should be able to reboot and see the drive like before....if that's the case BACK IT UP! I would also recommend testing some files...Open some of the AVI files (I know you have them) and mount some ISOs etc. look at and PIctures you may have etc.

I went through this process and it only paritially worked for me...because I had some more issues. But after a week or so, I resolved everything and got back 90% of all the data. If I knew then what I know now, it would have taken me a day or so, but that's life.

feel free to post more questions. I will monitor this thread for the next few days.

I do not remember the block size settings and likes. I would have to check on that tonight.

Hardware raid. As for the disks, I have not mounted them to a different machine and checked if they contain anything. I will do that tonight as well. Wouldn't they be marked as "failed" if they were though ?


I guess I'll start from reinstalling the OS and then once there, I'll start the applying your steps.

 

[Cr0nJ0b]

sometimes if your drive is failing it will kick itself out during the boot up in BIOS, and the controller will see that you are missing a drive. Since you are getting non-members, that means that likely the controller decided that the signature on the disks is not valid or readable or something. that's the part you need to fix first.

You have hardware RAID based on the Intel Chipset. They use the same code or close versions of it in lots of their motherboard chipsets, so they are all very close cousins. That's why you can move a set from one Intel Chipset to another in most cases.

You don't need to move the drives to another system, just make sure that you don't mess around with the order and cables...that is Critical...if i wasn't clear earlier.

If it were me, I'd burn a bootable WD, Seagate etc disk util DVD and just run a quick scan to rule out a hosed drive.

With Intel, you can delete the set and recreate it as many times as you like. As long as you use the same order and settings the data will sit safely where it was. The partition recovery part is the bit that puts the overlay data structure back on it so you can see your data and the volume again.

If you don't remember your block size and settings I would guess 99% that you used the default...so stick with that. It won't hurt it as long as you don't write to the disk.

In my case, I put the drives back in the wrong order and "successfully" recovered the partition...only to find that my pictures looked like a Picasso...that was my tip that I got the drive order wrong.

Like I said, I went to school on this subject when i lost my set last time, and where there is a will there is a way to recover you data...that is unless you totally lose a drive...in that case you're pretty much dead.

 

[maciekmiszkiel]

So tonight I managed to recreate my Raid 1 array. Installed OS on it and everything seems to be working fine. I have unplugged all 4 drives that were in Raid 5 array during OS setup. These are the pictures of what I see in my disk management utility and Intel Rapid Storage Manager.

http://dl.dropbox.com/u/14167065/winview.JPG

http://dl.dropbox.com/u/14167065/winview2.JPG

The second two pictures are of what I see in my Raid controllers Bios utility.

http://dl.dropbox.com/u/14167065/biosview.JPG

http://dl.dropbox.com/u/14167065/biosview2.JPG

...so Cr0nJ0b, you were saying that I should delete the Raid 5 array in the Bios utility even though it is telling me that my data will be lost ? If so, after I do that and boot into windows I should see 4 separate volumes/drives. At that point you recommended running a diagnostic utility on those drives and then creating the Raid 5 array again ? I apologize that I'm asking so many questions, but English is not my first language. I am making sure that I understand every step you mentioned correctly.

I checked and my drives are connected to the same ports like they originally were.

 

[Cr0nJ0b]

...so Cr0nJ0b, you were saying that I should delete the Raid 5 array in the Bios utility even though it is telling me that my data will be lost ? If so, after I do that and boot into windows I should see 4 separate volumes/drives. At that point you recommended running a diagnostic utility on those drives and then creating the Raid 5 array again ? I apologize that I'm asking so many questions, but English is not my first language. I am making sure that I understand every step you mentioned correctly.

I checked and my drives are connected to the same ports like they originally were.

The fast answer is YES. I know it's scary to see the big red letters that say your data will be lost, and in fact it will...but as you know...it already is lost...effectively. When you recreate your raid group, you will be putting the headers back on that tell the raid controller how to look for data on the drive. When you do this you would be and will be wiping out the partition header...which is already inaccessible. What you get is a new raid set that looks clean. That's where the partition recovery software comes in. I had a lot of funkiness when I did my recovery, probably because of the chipset and BIOS I was using, but it all worked eventually. BTW, in my case my raid set was RAID 0, but it's the same principal.

If you are really, too scared to try this...which I would understand...there are some other tools on the market that would allow you to break the RAID 5 set into independent drives and then have the software piece things together...but in that case, you will be in the same shape with a partition table that is lost.

 

[Purpose]

Ignore everything that has been said so far.
Get testdisk and follow the step-by-step and your data will be returned to perfect working order.
http://www.cgsecurity.org/wiki/TestDisk_Step_By_Step

I've done this on three occasions now with my raid5 array, and it has worked every single time.

The key thing here is the disks *must* be plugged into the same port that they were plugged into when you built the array the very first time.

Rebuilding the array in the manner that is described by Testdisk will NOT damage your data if you plug them in wrong, you will just have to keep repeating it with the drives plugged in, in different combinations until you get it right.

 

[maciekmiszkiel]

Ignore everything that has been said so far.
Get testdisk and follow the step-by-step and your data will be returned to perfect working order.
http://www.cgsecurity.org/wiki/TestDisk_Step_By_Step

I've done this on three occasions now with my raid5 array, and it has worked every single time.

The key thing here is the disks *must* be plugged into the same port that they were plugged into when you built the array the very first time.

Rebuilding the array in the manner that is described by Testdisk will NOT damage your data if you plug them in wrong, you will just have to keep repeating it with the drives plugged in, in different combinations until you get it right.

Are you telling me that I should not delete my Raid 5 array then before I run TestDisk ?

 

[maciekmiszkiel]

I guess I have to also purchase enough disks to offload the recovered data as well. After reading the TestDisk instructions it looks like you cannot just put it back on that Raid 5 array right away.

 

[Cr0nJ0b]

Ignore everything that has been said so far.
Get testdisk and follow the step-by-step and your data will be returned to perfect working order.
http://www.cgsecurity.org/wiki/TestDisk_Step_By_Step

I've done this on three occasions now with my raid5 array, and it has worked every single time.

The key thing here is the disks *must* be plugged into the same port that they were plugged into when you built the array the very first time.

Rebuilding the array in the manner that is described by Testdisk will NOT damage your data if you plug them in wrong, you will just have to keep repeating it with the drives plugged in, in different combinations until you get it right.

Probably the most unhelpful post I've seen to date. Don't listen to the people offering valid advice, and RTFM...thanks.

How exactly do you use Test Disk without first putting the drives back into a raid array? the drives are non-members, so how do you fix that? If you have a really good idea of how to help, then I sugget you respond with something concrete and specific. I took considerable acception to the "ignore everything you've read so far", especially when you then go on to parrot what I've already stated with less information and less support.

thanks for nothing.

 

[Cr0nJ0b]

I guess I have to also purchase enough disks to offload the recovered data as well. After reading the TestDisk instructions it looks like you cannot just put it back on that Raid 5 array right away.

You should have a backup target already. If you don't you need one. I'm not sure how big your array is, but I would guess it's probably 3TB usable. If it's bigger you should really look at using something like a NAS device as a central backup...it would be hard to backup 4TB+ to a single drive. If it's smaller, then buy a single external drive eSATA if you can or a big internal drive and backup to that.

What you want to do in the end is recover the partition, copy the data to the back drive, delete the partition and the raid set, create a new set and partition, then copy the data back. It will take some time, but in the end you will also have a complete backup.

 

[maciekmiszkiel]

You should have a backup target already. If you don't you need one. I'm not sure how big your array is, but I would guess it's probably 3TB usable. If it's bigger you should really look at using something like a NAS device as a central backup...it would be hard to backup 4TB+ to a single drive. If it's smaller, then buy a single external drive eSATA if you can or a big internal drive and backup to that.

What you want to do in the end is recover the partition, copy the data to the back drive, delete the partition and the raid set, create a new set and partition, then copy the data back. It will take some time, but in the end you will also have a complete backup.

Well...looks like I'll have to shell out some money for storage and drives then. I don't have that in place yet.

So to recap it. At this point I should Delete my Raid 5, then recreate it, boot into Windows and run TestDisk.

 

[maciekmiszkiel]

Cr0nJ0b: I have just purchased a NAS with 4tb of space.

I just had a few more questions before I proceed with what you proposed. After you saw my screenshots from win disk management utility, does everything look normal so far ?

Am I supposed to see 3 disks, one of them not initialized and 2 online ?

 

[maciekmiszkiel]

I went onto the source page you have pulled the info that you provided me with from it. There is one thing that concerns me. One of the posters there had very similar setup as mine. I'll be fixing this issue on the same machine that the issue has happened.

If I delete the Raid 5 array and then recreate it in Bios utility, after I boot to windows on that machine my intel rapid storage will initialize like it did on his box. Does it mean that I'll be loosing all that data once it initializes ?

 

[Cr0nJ0b]

Cr0nJ0b: I have just purchased a NAS with 4tb of space.

I just had a few more questions before I proceed with what you proposed. After you saw my screenshots from win disk management utility, does everything look normal so far ?

Am I supposed to see 3 disks, one of them not initialized and 2 online ?

Okay, now that I'm looking at all the pics side by side, I get it a little better and can see why the capacity for backup is an issue...

Everything looks like what I would expect for your situation. Drives 3 and 4 are showing up as non-member disks, which means that the controller thinks that they are not part of the original set. This could happen for a number of reasons, in my case it was an OS installation gone wrong that stamped the front of one of my drives partition sectors and messed with the raid sig. It's telling me that your device is gone.

I think the reason that the disk manager is seeing those 6TB drives is that it's looking at each drive and each one has part of the 6TB partition....

What I would do is delete the other disks in the bios so that you have 4 disks that are not members of a raid volume, then recreate the volume using the defaults (name isn't important call it what you like).

Once you do that you can boot the OS.

I'm not 100% on the next part to be honest. so here's what I belive. When you boot up, and go into disk manager you should be prompted that there is a disk (say disk 4) that needs to be initialized. It will ask you if you want to initialize the disk. I would say no the first time and see if testdisk can find it anyway. if you cancel that window you aren't effecting the drives and it should show up as a single blank bar with 6TB.

If Testdisk can't see it at that point with a scan, you may need to initialize it, but DON'T format it...for sure DO NOT FORMAT the drive.

When I did this (8-10 times) on the same set, I deleted and rebuild the volume numerous times. I never recall the Intel utility doing anying in windows other than reporting that a set exists. You could build the set there, but never did that. I built the set at the bios layer and they used testdisk.

 

[maciekmiszkiel]

Okay, now that I'm looking at all the pics side by side, I get it a little better and can see why the capacity for backup is an issue...

Everything looks like what I would expect for your situation. Drives 3 and 4 are showing up as non-member disks, which means that the controller thinks that they are not part of the original set. This could happen for a number of reasons, in my case it was an OS installation gone wrong that stamped the front of one of my drives partition sectors and messed with the raid sig. It's telling me that your device is gone.

I think the reason that the disk manager is seeing those 6TB drives is that it's looking at each drive and each one has part of the 6TB partition....

What I would do is delete the other disks in the bios so that you have 4 disks that are not members of a raid volume, then recreate the volume using the defaults (name isn't important call it what you like).

Once you do that you can boot the OS.

I'm not 100% on the next part to be honest. so here's what I belive. When you boot up, and go into disk manager you should be prompted that there is a disk (say disk 4) that needs to be initialized. It will ask you if you want to initialize the disk. I would say no the first time and see if testdisk can find it anyway. if you cancel that window you aren't effecting the drives and it should show up as a single blank bar with 6TB.

If Testdisk can't see it at that point with a scan, you may need to initialize it, but DON'T format it...for sure DO NOT FORMAT the drive.

When I did this (8-10 times) on the same set, I deleted and rebuild the volume numerous times. I never recall the Intel utility doing anying in windows other than reporting that a set exists. You could build the set there, but never did that. I built the set at the bios layer and they used testdisk.

Please take a look at the picture again. Disk 3 and 4 are actually members of the raid 5 array, that's broken and 0 and 1 are members of raid 1. I think you meant disk 2 and 5 for deletion. The problem is there is no option in the controlers bios utility for delete. I can only delete or create raid volumes but cannot adjust them thus I have no idea how to proceed.

 

[Cr0nJ0b]

Yes I meant 2 and 5. You can only delete member disks. When you delete the volume set you will turn the member disk to non-member disks. You will in the end have 4 non-member disks that you will use to recreate the array.

 

[maciekmiszkiel]

How do we know that the initialization won't start automatically upon the raid volume recreation, even before I boot up to the OS ? Aren't we trying to prevent that to happen before I run disktest ?

 

[Nothinman]

I would boot a Linux Live CD and see if dmraid can start the array. Since it's a software RAID array there should be a signature on each disk telling it which array, what type, number, etc for all of the members and dmraid might be able to startup the array so you can copy everything off and recreate it. No guarantees, but it's free and worth a try.