During a recent power supply failure, the boards on 3 disks in my 4-member ICH9r RAID-5 array literally fried. I believe that the data on the platters is ok, that it's simply the boards that are finished.
Losing the data would be a disaster.
After the smoke had cleared, in a panicked craze, I pulled the drives out of the system but forgot to note which drive corresponded to which port. Now I am trying to reconstruct the array by imaging the individual members to new disks by exchanging the good disk boards I have (one from the good drive and one from a spare) for the burnt ones, one at a time.
This is leading to some confusing observations and questions:
1. What does a single array member look like to the Windows disk manager? Two of the disks show up as "healthy" partitions and two show up as unallocated space. Is this because only two disks carry the information needed to initialize the array or is it a sign that the data may be damaged?
2. Even sector-by-sector images of the old disks do not show up as array members. Does ICH9r rely on a drive's serial number to identify its position in an array?
3. If the drive serial number is important, then one option is to create a new array with the new disks, then clone the old data to the new array. If I do this, will the ICH9r controller "remember" the serial numbers of the old array, just in case I need to try restoring from the original disks (e.g. if I find a replacement drive board?)
4. I will pay good money for a Maxtor/Seagate 250GB 3.5" SATA Drive, P/N 9FV132-327, Firmware 3.AAC (no other FW works).
From what I have read, most Motherboard RAID systems store drive and array identification data in a RAID SuperBlock, this SuperBlock is located beyond the last accessable data sector; something like RaidReconstructor from Runtime.org can read the SuperBlock and reconstruct the original data for you.
If needed, you can also get hard drive spares from hdd-parts.com
I have 4 raids on my machine with 2 more being added this month so i have been thru 'the disaster' more than once and gained a lot of experience with data recovery. I hope i can help you with yours.
Firstly **** DONT PANIC **** there are some good solutions out there.
I recently had a raid 5 fail and refuse to rebuild with a new drive installed.
I have tried many of the expensive, and free, recovery software tools out there and found only one i would trust entirely. Zero Assumption Recovery.
The thing that makes it stand out is that you simply make the HDDs into standard non raid unformatted drives (very easy). This means the 4 Hdds are no longer anything to do with any raid at all.
ZAR runs and asks you raid type and which physical HDDs to use (note that as they dont have drive letters assigned it uses a simple way to pick them). If you dont know the stripe size, or any other parameters, including which order the HDDs were assigned by the raid ZAR will run analysis and determine them for you.
Then you run the lengthy recovery stage and it will offer a list of partitions to choose from. a few minutes later you get a normal explorer type menu and you can recover the data to a backup disk(s).
You end up with your original raided data back on non raid ntfs/fat32 drives ready to be re-copied to the raid once its fixed.
I have recovered before even when 2 of 4 drives have shown as unallocated.
I'm currently using file scavenger to do the same job. i'm rebuilding the array in a system with windows installed on an other drive. it is scanning ALL the files (1TB worth data). Good news is that I can see my Raid 0 files. They are lsited fine. Bad news is that it takes hours to complete. not to mention I need to copy the data afterward.
I hope it works and it wont let me down with corrupt files or so..
I eventually managed to repair one of the drives by simply soldering a short wire across the fried voltage regulator. This worked long enough to rebuild the array to a new disk--and then failed permanently 5 minutes after my data was back!