Start a Conversation

Unsolved

A

3 Posts

3639

February 22nd, 2019 01:00

Restoring RAID5 when foreign configuration has offline disk

(Edited because of inadvertent previous incomplete post)

 We have a RAID5 array composed of three 4Tb drives, slots 0-2, with no hot spares. The controller is a PERC H310 Mini.

 When drive 2 died the array entered degraded mode, but functioning normally otherwise. We then substituted the dead drive 2 for a new one, with the intent to fully restore the array. Since the new one was initially labeled as JBOD, we added it into the array with

perccli64 /c0/v0 start migrate type=raid5 option=add disk=32:2

and the migration started. However, drive 1 seems to have failed during (or right after) the migration, and now we have the following situation (see attached photos):

  • The VD configuration in the controller is cleared.
  • The foreign configuration has drive 0 online, but drive 1 offline. Drive 2 is missing, but that is kind of expected.

BIOS_View.jpg

So, my questions are:

  1. Perccli will not allow me to restore the foreign configuration because one of the drives is labeled offline. Is there a way to force perccli to import the foreign configuration?  That would be my preferred option. I would then set the drive 1 online, rebuild the array and then substitute drive 1.
  2. If 1. is not possible, maybe I can retag the array (at present, drives 0 and 1 are foreign UGood). What are the array parameters that I must ensure to be equal to the old ones (I have the output of the functioning perccli /c0/v0 show all available)?
  3. If I have to retag, do I retag with only drives 0 and 1 present, in order to create a degraded RAID5 and then add drive 2 and rebuild? Or do I retag with the three drives, remove drive 2, go into degraded and then add drive 2 and rebuild?

The data in the array is quite important and unfortunately we don't have access to backups, so I want to make sure I don't do anything foolish.

Thanks in advance!

4 Operator

 • 

2.9K Posts

February 22nd, 2019 08:00

Hello,

If the data is that critical, then the absolute best thing you can do is engage a data recovery service like DriveSavers or Ace Data Recovery. I've been told that DriveSavers offers a discount, if you tell them that Dell referred you. Ace could be the same. I would hesitate to retag, because if it doesn't work, you have the potential to really mess up your data badly, so you may want to weigh your options and risk.

1.  perccli /cx[/ex]/sx set online should be what you are looking for. The CLI reference guide is where I found that. You can get said guide here: https://topics-cdn.dell.com/pdf/dell-sas-hba-12gbps_reference-guide_en-us.pdf

2. When creating a virtual disk, I use the PERC configuration utility, like in your picture. You need to use the same drives, same VD size, same stripe size, and Secure VD settings to have a chance at a successful retag. The command arguments should be in that reference guide.

3. You'll need to retag with all three drives, then perform any replacements.

Just because it's worth repeating though, if this data is critical, you really would be best served with a data recovery service. A retag is a high-risk option and I've seen data lost about as often as I've seen it recovered. That having been said, when I talk to folks, it's generally because something broke or didn't go right. No one calls, chats, or posts when there's no issue. You could be fine, but those companies do offer free consultations and I've been impressed with both when they show up to our campus.

3 Posts

February 23rd, 2019 08:00

Hi Dell-DylanJ,

Thanks for your reply. Do you know of any company in Spain that will offer discounts in data recovery from a Dell referral?


1.  perccli /cx[/ex]/sx set online should be what you are looking for. The CLI reference guide is where I found that. You can get said guide here: https://topics-cdn.dell.com/pdf/dell-sas-hba-12gbps_reference-guide_en-us.pdf

I had already tried that, but the command failed. I guess it works only for offline drives in the active configuration, but it can't change the foreign configuration.


2. When creating a virtual disk, I use the PERC configuration utility, like in your picture. You need to use the same drives, same VD size, same stripe size, and Secure VD settings to have a chance at a successful retag. The command arguments should be in that reference guide.

3. You'll need to retag with all three drives, then perform any replacements.

Thanks again, this really helps. I'll think I'll try to get a couple of spares to first clone the drives, retag the clones and then, if that doesn't work, send the originals out for recovery.

4 Operator

 • 

2.9K Posts

February 25th, 2019 07:00

That's probably the best plan. As for any recovery companies in Spain, I don't know what would be available. Hopefully, things just go well with the cloning process and make all that unnecessary.

3 Posts

February 28th, 2019 16:00

 Hi,

 I've finally gotten the clones, and I am ready to try the retag. However, I would feel safer if I could check that the data is accessible in some form. For that, I've tried several RAID recovery products that attempt to rebuild the array via software, but none of them could autodetect the underlying mapping / block order, eg 1.2.P;3.P.4;P.5.6 .

 Does anyone know what would be the typical block order for a three-drive RAID5 ? In case it helps, relevant lines in the controller log are

T53: ARs configured: 1
T53: Ar 0: size=1d1b00000, activeLd=1, numPd=3, seq=0, powerSave=0 powerSaveInProgress=0 PDs: 41(20/0) 01(20/1) 41(41/255)
T53: LD  0: L=5  SS=128  Size=3a3600000  NL=448  Status=0  DT=0    BT=0  , Encr=0, prop=[ID=00,dcp=00,ccp=00,ap=0,dc=0,dbgi=0,S=0|0,dps=00,cps=fe]
T53:        span 0: sBlk=00000000, nBlk=1d1b00000, ar=0

 

 Also I would like to make a backup of the array metadata before retagging. Where in the disk would that be located?

 Thanks!

 Xavier

No Events found!

Top