Recovery of RAID 5 with 2 healthy disks "unassigned" due to BIOS fault.

Toggle sidebar Toggle sidebar
W

Werecat

Honorable
Sep 18, 2012
12
0
10,510
I have a Gigabyte GA-Z77X-UD5H with a RAID 5 using Intel Rapid Storage Technology and 4 1.5 TB Samsung SpinPoint F2EG HD154UI drives.
Today upon starting the computer the powerswitch was stuck, causing the computer to cycle on/reboot a couple of times until i got the switch unstuck. That caused the dualbios to recover the second bios, which i guess caused the following mistake.
When i tried to boot the computer initially i couldn't find my SSD boot disk, which i fixed by correctly assigning boot sequence. Furthermore i assigned the SATA raidcontroller to RAID.
When i enter the Intel Rapid Storage Technology menu (before booting) i find that my RAID 5 volume has status "failed", but notice that all four Samsungs are connected, but only two are member disks. The remaining two are non-RAID disks, but show up as healthy.

I haven't dared to delete the RAID volume to create a new spanning the original four drives, as i'm afraid it will delete all data.

So is it an option to "fool" it and create a "new" volume or will it delete the old raid volume (and data) overwriting it with the new?

And if not, does anyone have a solution to add the two healthy disks (that still contain the data from the RAID volume) to the RAID volume so as to rebuild it with no data loss. (It is worth mentioning that i tried from the Intel Rapid Storage Technology applikation in windows, to set the disks as present, which doesnt work because they are present as non-RAID disks.)

Any suggestions will be greatly appreciated.
 
Sort by date Sort by votes
R

Rookie_MIB

Distinguished
Jul 27, 2013
1,970
10
20,965
I know that when working with a Linux software raid, you have options to 'assume clean' which skips a resync/rebuild. The scary thing is that if it were just a single disk, you could just rebuild and it would be fine. With two 'bad' disks though - it would be very problematic.

However, some searching around found this post which is very similar to the situation you're in, and a solution was posted here:

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

Original post was here (about halfway down the page).

https://communities.intel.com/message/12050

Good luck...
 
Upvote 0 Downvote
M

marko55

Honorable
Nov 29, 2015
800
0
11,660
I've had this same behavior immediately after upgrading the BIOS on an X99 board. Thankfully it was a fresh build so there was no user data on the RAID-5 array but 2/4 drives lost membership. Nothing I tried worked and just ended up recreating the array.

This promotes one of the benefits of having a dedicated RAID card. Even if your mobo catches on fire you can simply forklift your RAID card & the drives to another mobo & all your RAID configs & data are intact.
 
Upvote 0 Downvote
W

Werecat

Honorable
Sep 18, 2012
12
0
10,510


Thank you. I haven't "post-n-forgetted". But the searches are taking quite a drag. I followed the extremeoverclocking.com guide (though i now regret doing step 1 (Reset both HDs to non-member using Intel BIOS utility), but more on that later.

Testdisk shows all drives and the "non-existant" Raid drive.
Image2.jpg


When i analyze the individual drives i get these results (for each of the individual drives):
Image1.jpg


For the complete Raid drive i get (sorry no screen-dump) two drives with filetables and an error 82% through the scan (indicating no filetable on at least on of the remaining drives. Now thats when i started to regret doing step 1, thinking that if i had not erased the two raid drives in the bios, maybe i would be having my data now...

I've tried the Runtime raid reconstructor (https://www.runtime.org/raid.htm) - free version.
Image3.jpg

And it cant find any significant filesystem, though it clued me on to the fact that i used a block size of 4096 (dunno why) when i created the raid initially.

And thats me back to the testdisk scan with a new block size.
Image4.jpg


(50 cylinders in 30 secs = approx 11.4 hours..)



Neither the gigabyte bios nor the intel raid "sub-bios" had the option of changing the drives to member drives.

I will be back with news, i am not giving up on my data so easily. Any suggestions will still be very much appreciated.

 
Upvote 0 Downvote
P

Pamhad

Reputable
Jan 14, 2016
3
0
4,510


Did you succeed is recovering your Raid? Exactly the same thing happened to me today - faulty on/off switch and exactly the same symptoms.

I haven't done anything but panic as yet.

 
Upvote 0 Downvote
P

Pamhad

Reputable
Jan 14, 2016
3
0
4,510
I didn't dare do any thing for days, I have about 8 TB of data on the RAID5 and I am ashamed to say while I had backups of some stuff from last September, a lot of it wasn't backed up at all... This morning I gritted my teeth and started using the instructions on this thread https://communities.intel.com/thread/3351?start=30&tstart=0 with one small change - before I started, I uninstalled the Microsoft Rapid Storage Technology software as my system disk isn't part of the RAID and I didn't want it to start initialising the new RAID.

Well blow me down with a feather if it didn't work! I followed the steps (fortunately I had not detached any disks so no worry about the order) and Testdisk found my old RAID array straight away. With trepidation I clicked the Write button, rebooted and all of my data was there! In total it took maybe 5 mins.

So now my hands have stopped shaking, my heart has stopped beating at about a million beats a minute and I am doing backups. Fortunately I have sufficient spare disks around to make a complete copy of everything and I will be setting up an automated nightly backup as soon as I have finished!
 
Upvote 0 Downvote
R

Rookie_MIB

Distinguished
Jul 27, 2013
1,970
10
20,965
Glad it's all fixed. Funny thing is that there are so many problems that have to do with the freaking metadata for raid arrays which should be relatively easy to fix - yet it's not always clear.

When I recently tried to upgrade my NAS (freeNAS, 5x2TB Raid5) to a newer version, it tried to import the array and failed. So I went back to revert to the old freeNAS, and in the process of trying to import, it updated that metadata to v1.3 from v1.2. Well, old freeNAS didn't know what to do with v1.3 metadata, so basically I was hosed and lost about 5.5TB of movie/tv shows, drivers, software, all kinds of stuff.

So - went to straight mdadm raid5 on a CentOS build. Running solid as a rock, although much like you, I was ready to cry at losing all that data. Nice thing about mdadm raid is that it is very robust and mature with lots of options for recovering things.
 
Upvote 0 Downvote
W

Werecat

Honorable
Sep 18, 2012
12
0
10,510
Thank you for your replies.
I've been fiddling around with settings and testdisk, but have so far nothing to show for it. Been getting som disappointing data though.
24v1vh4.jpg

And this one...
vo9xro.jpg

And almost gave up..
2u9sspt.jpg


But the you guys started to suggest checking R-studio out, so i did that. And by God it could find my files, but (there's always a but).
2nki6c3.jpg

2 years!! I'm not that desperate to get all my old games and savegames back. Happy to spend the 79.95$ :sad:

Or am i doing this wrong. I can't figure out if R-studios can repair the file structure, instead of just rebuilding the files to another drive?

When i select the partition with my files, this is the information available.
2vrzjgn.jpg


How do i use it to rebuild the files structure instead of using 2 years on rebuilding the files and copying them to another drive?
 
Upvote 0 Downvote
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom