WIndows 7 asks to format drive after RAID5 OCE

[kruuth]

Hello everyone. This is my first post here and I'm at the end of my rope. Here is my situation:

I have a Highpoint Rocketraid 2680SGL card running my array. Initially I had 4 1tb drives. Last Friday I added a fifth drive, and, followed the instructions from Highpoint, went through the steps to expand the array.

Now, according to windows, I have a drive with one 3tb partition, and then 1tb of empty space. When I click on the drive, it asks if I want to format it.

Is there something that I need to do in order to get this working? I need to get the files off the old drive at least. I was under the impression that OCE was supposed to allow you to expand the volume without destroying the data that was on it.

 

[MaxBurn]

A RAID expansion is one of the more risky operations that you can do on an array. Do you have a backup before you did it? Does the controller think the array is healthy?

 

[kruuth]

The controller thinks the array is just fine. Oh, it changed the name of the array from Raid_5_0 to Raid_5_1.

 

[kruuth]

Any ideas? I'm told this card uses fake raid not HW raid.

 

[ddrueding]

The 3TB partition with 1TB free indicates that the capacity expansion worked at least that far, with the only step left being to go into disk amanagement and resize the partition. But you should still be able to get to your data right now.

1. What does your RAID controller say about the array, is it healthy? Are all the drives listed as connected?
2. What does Windows say about the disks in device manager? Everything look good there?

RAID5 OCE is one of the riskiest things you can do, and they all advise to have a backup of your data. Of course, if you had the spare storage handy to keep a full backup, you wouldn't have bothered with the OCE in the first place.

 

[kruuth]

That's reassuring then. How do I do that? Is there something I can do in windows to just mount the volume? I've never done anything like this before.

The 3TB partition with 1TB free indicates that the capacity expansion worked at least that far, with the only step left being to go into disk amanagement and resize the partition. But you should still be able to get to your data right now.

1. What does your RAID controller say about the array, is it healthy? Are all the drives listed as connected?
2. What does Windows say about the disks in device manager? Everything look good there?

RAID5 OCE is one of the riskiest things you can do, and they all advise to have a backup of your data. Of course, if you had the spare storage handy to keep a full backup, you wouldn't have bothered with the OCE in the first place.

 

[kruuth]

Sorry I should have said more. Here are the answers to your questions:

1. The raid controller said maintenance during the migration. After that it is reporting that everything is fine. All five drives are listed as connected and healthy.
2. Windows seems to see it fine. When I open the drive in disk management, I see the 4tb volume, with a 3tb raw partition and 1tb of unused space.

 

[ddrueding]

So it sounds like everything is working except for some possible data corruption. In "My Computer" you can see the drive letter, but when you double-click on it Windows says it isn't formatted?

We are nearing the end of my knowledge when we get to filesystems and their integrity. At this point I would try some kind of scandisk or a data recovery program like GetDataBack.

 

[kruuth]

That is correct ddruedling. There is a drive there, but windows thinks that it is unformatted. What scandisk are you referring to?

 

[ddrueding]

It seems many people in your situation have had luck with a program called EaseUS. I've never tried it myself. If you have another array of sufficient capacity somewhere, I would just use GetDataBack to recover what you can and then reformat.

 

[kruuth]

Not really. Is there somewhere online I can backup arouns 2tb worth of data briefly?

 

[ddrueding]

That amount of data will take ages just to get online and restore from. I've heard of people purchasing a drive locally and then returning it, but I can't recommend that move on principle. But I won't judge if that is what you decide to do

 

[MaxBurn]

Keep the drive for backups.

 

[kruuth]

I noticed that some files aren't recovering so hot....

 

[Mercutio]

Many moons ago, I recovered about 1200GB of data from a RAID 5 failure. Even though the recovery was perfect in that I got all of my files back, I still saw minor corruption in recovered audio and video files.
The chances of having additional hard errors during a RAID recovery on a large volume are actually pretty high, so this might not be entirely surprising.

 

[kruuth]

There's no way to fix that? Some videos that I have are not easily replaceable.

 

[Mercutio]

Data recovery is a precious gift. You should be happy you're getting back anything at all and go forth a disk-crash fearing, backup-having man. Or lady. Whatever.

 

[ddrueding]

For RAID, there is nothing better than RAID-1. But there is no substitute to backups. A full backup not associated with the original in any way.

 

[Mercutio]

Also: RAID1 is not a backup.

 

[mubs]

As drive capacities go up, UREs (Unrecoverable Errors) also are, spelling disaster during Raid recovery processes. The highest capacities I've stumbled on for Raid drives these days are 900 GB (this is for Dell NAS systems). Of course manufacturers (WD, Hitachi, Seagate) list drives up to 4 TB as "Enterprise Grade" drives.

I'd like to hear from the likes of Handy and others who work with humongous amounts of corporate data as to how they are dealing with this situation.

 

[Howell]

Lots of spindles.

 

[ddrueding]

I've been hearing for years that UREs would prevent RAID levels with checksums from working, with everyone going back to RAID-10 and more backups. I've always been a fan of the performance benefits of writing to a RAID-10 array, and with capacities getting huge and costs plummeting it isn't that big a deal.

 

[Mercutio]

Well... yes and no. You have to bear in mind that there are costs just for having the port that the disk sits on, and the costs for the chassis, and the power and the labor and and and... And somewhere there's a beancounter who knows exactly what all those costs are.

And in a RAID10 or RAID50, you're being pretty liberal with all those resources. It's not actually a bad thing if you've been given the budget and mandate to do it, but damned near everything in IT falls back on "Do as much as you can with as little as possible."

 

[Stereodude]

I've been hearing for years that UREs would prevent RAID levels with checksums from working, with everyone going back to RAID-10 and more backups. I've always been a fan of the performance benefits of writing to a RAID-10 array, and with capacities getting huge and costs plummeting it isn't that big a deal.

Sort of a hear no evil see no evil approach? In RAID-10 do you even know you got an URE? At least in RAID-5 you've got parity that can detect data corruption / validate integrity. RAID-1 just assumes everything is good and when a drive drops and you rebuild the array it just mirrors the working drive with no verification that the data being mirrored is "good".

 

[Mercutio]

I can't say that I spend much time looking at disk controller logs unless I see a hiccup someplace. I suspect dd's in about the same boat.

 

[Chewy509]

We're I'm working now, I've got just under 50TB of available storage (and most of it is full), we've migrated away from hardware based RAID onto ZFS storage pools for the file servers, simply due to silent corruption that was not picked up on the underlying hardware (RAID50 config). Combined with weekly scrubs and automatic email notification (one of the great features of FMA on solaris/openindiana) we now know about corruption even before anyone notices. And with RAIDZ2, the corruption is usually corrected without fault. How often do we find a fault - rarely (2 instances in the last 3 months on the SATA based storage arrays, 0 instances on the SAS based storage arrays).

(This is just for our file servers, the ESXi servers still use traditional hardware RAID).

 

[mubs]

Sort of a hear no evil see no evil approach? In RAID-10 do you even know you got an URE? At least in RAID-5 you've got parity that can detect data corruption / validate integrity. RAID-1 just assumes everything is good and when a drive drops and you rebuild the array it just mirrors the working drive with no verification that the data being mirrored is "good".

Yeah, but a Raid-5 rebuild will stop when an URE is encountered. In Raid-10, this may be a "silent error", but at least the rest of the data is recovered. Pick your problem.

 

[Mercutio]

We're I'm working now, I've got just under 50TB of available storage (and most of it is full), we've migrated away from hardware based RAID onto ZFS storage pools for the file servers, simply due to silent corruption that was not picked up on the underlying hardware (RAID50 config).

The amount of stuff sitting around my apartment is in the same ballpark, actually. I've been tempted with the idea of moving to Server 2012 and Storage Pools for performance reasons rather than zPools like I'm presently using, but I'm mildly terrified that the underlying technology would not catch hard errors on my drives. I do have real backups, but that doesn't mean I want to use them.

 

[Chewy509]

Unfortunately, in the work I'm doing now, data integrity on the storage arrays comes above all else, including performance.

 

[Mercutio]

Well if you want to see something hilarious, watch how long it takes to copy 33 million 100kB files to a RAIDZ2l.

 

[LunarMist]

Unfortunately, in the work I'm doing now, data integrity on the storage arrays comes above all else, including performance.

Why is that unfortunate? How can it be otherwise for data?