How to remove raid 5 array of 5 SAS and use only use one SAS at a time without writing any thing to disc for recovery

Question

I have HP server ML370 g5 with 8 SAS, c drive 1 72 gb raid 0, d drive 2 72 gb raid 0, f drive 5 146 gb raid 5.

2 of 5 sas drive has got bad sectors and raid 5 is showing status failed.

now i want to remove all this 5 SAS and put 1 by 1 in any of the bay to make its image for data recovery purpose without writing anything to the drive. how should i proceed. i also want to keep drive c and d intact. also is it possible if i put all this 5 drives in the bay with the same sequense will it recognise the raid 5 array ( i read some where its smart controller..just curious)

many thanks in advance.

Answer 1

How do you intend to do this data recovery? if you plan to do it yourself it would be helpful to know how so we can clarify the method needed exactly. If it's a third party company that's going to recover it for you then I'd suggest you just leave things as they are. You're aware aren't you that a single disk won't have a mountable file system on it right?

So onto the actual question, as the controller and disk each have their raid configuration stored on them the issue you get with trying to read each disks 'raw' is that the controller will always want to see them as an array, if you try to reconfigure the controller to see each disk as a separate disk it'll just wipe each disk individually and you'll lose everything.

If you really want to read each disk raw I'd suggest you connect each disk, individually, to the most basic, stock, disk controller you can buy - no raid features, nothing, just a cheapo one. This way the disk will be seen by the machine as essentially a blank disk - it's almost certainly how any recovery software would want you to see the disks.

I hope you don't mind but I'm going to favourite this question as next time I need to justify to someone why they shouldn't use R5 I will be able to fall back on your own painful experiences. I hope you manage to recover your data, presumably you didn't have a backup?

Answer 2

HP use a different parity calculation to most other systems. It is nicknamed 'Delayed Parity' and the parity stripes are larger than the data stripes so you really need to know what you are doing.

If the data is important, consider getting a quote from a company that knows ho HP/Compaq RIAD works with SAS drives.

Answer 3

You give some, but not complete details about the disk configuration. In particular, you have 8 disks, of which at least two have some of the C: drive data on them in RAID0, at least two have some of the D: drive data on them in RAID0, and at least 5 others have some of the F: drive data on them in RAID5. You then mention that two disks of the RAID5 configured F: have failed (bad sectors).

What isn't clear is:

• Which disks (it would help if you numbered them, since you have 8) participated in C: and D:'s RAID0 arrays, and were any of those disks failed?
• How many disks were in the RAID array?
• What happened to either the 8th disk (assuming C and D shared two drives), the 6, 7, and 8th disks (assuuming the RAID0 shared with the RAID5), or where is the extra disk (assuming C and D didn't share drives)?
• Where is drive E?

Short answers to some of your questions:

• If you lose any drive in a RAID0 array, you are probably not going to recover much data from it.
• If you lose more than one drive from a RAID5 array, you are probably not going to recover much from it.
• Hardware raid probably means you will want professional help recovering it.
• You can probably get data off of some of the disks... then what?

Answer 4

The way I would approach this is by using a "dumb" controller for doing the copy. For example, in a similar situation I had with SATA drives, I used a USB to SATA adapter to make copies of the raw discs. You definitely want to make pristine copies of the discs before you start trying to reconstruct the data from them, just in case there's a mistake. But smart RAID controllers can get in your way with this. So, get a dumb SAS controller to make the copy would be my recommendation.

Then you'll need to piece together the layout of the data on the disc. The only time I've done this it was with a RAID-1 array, so it was very easy -- I just needed to find out the offset of the data within the drive. In your case you are going to need to figure out the striping, and you'll probably also need to figure out the parity (since you likely had one disc fail, and then some time later another disc, you can't just rely on the non-parity data, you'll have to reconstruct from the parity data).

Unfortunately, this may be fairly hard. You may want to see if the controller vendor has low-level layout documentation, but a friend who had a RAID-5 array fail was told that this information is proprietary and they couldn't give it out.

Have you contacted a data recovery house to see if they can help? It may be cost-effective to have them work on it, as some of them I've heard are versed with the low-level layouts of different RAID controllers data and can recover from these sorts of situations.

Not to rub salt in the wound, but obviously this seems to point out a couple of weaknesses in your operations: a lack of backups and not having array monitoring that told you when the first drive failed. Be sure those issues get addressed in the future.