Start a Conversation

This post is more than 5 years old

Solved!

Go to Solution

8604

March 8th, 2016 07:00

Force Failed RAID Back Online

I work at a data recovery company (www.data-medics.com) and was recently contacted regarding recovering data from a failed EMC Clariion AX4-5i unit.  Here's the situation:

The OS drives are still in a healthy working array.  The data portion was configured in a RAID 6 with 8 500GB SATA drives.  One drive went offline and went unnoticed for a long time.  A second drive went offline (and was noticed), then during the rebuild, a third drive went down.  The latter two drives to go offline were both just affected by bad sectors and were imaged with only a few KB of unread sectors.  So we have more than enough disks to go online.

So here's my questions:

1. Is there a way to force the data array back online using the disks?  Even changing serial numbers of the clones isn't a problem, we're a professional data recovery lab and can do that in a few minutes.  But, I assume some metadata needs to be modified to un-mark the members from being "failed".

2. We can virtually build the array, however it doesn't appear there's any known file system on there.  I assume it's EMC's proprietary FLARE filesystem.  Is there any way that we can extract files that way, assuming that we can destripe the data from the array to a single image file?

3. Is there even a file system on the data array, or is it all on the OS part of the unit?

We deal with a ton of RAID arrays, but these EMC units are pretty foreign to us.

Thanks a million in advance for any guidance you can provide here.

18 Posts

April 12th, 2016 11:00

Hi Folks, sorry about the confusion when it comes to EMC drives.  I have gone to my sources and found that Jared is correct in that these SATA drives cannot be reformatted at 520.  It does indeed use the raw 512 format. The AX4-5 system software converts to 520 on the fly. The drive assembly is therefore labeled as 520 for use with our arrays.....but the raw drive itself is at 512.    

Additionally the back-end on an AX4-5 is SAS so EMC uses the raw SATA drive and couples it with a SAS bridge card to create a SAS drive assembly.

Hope this helps!

Dan

4.5K Posts

March 8th, 2016 09:00

Which specific drives failed. The OS drives are the first four in the enclosure - disks 0-3. These are divided into two sections - the Flare OS part and the User part. If the array is still booting that would indicate that the Flare portion of the drives are still working, so it's probably the User portion of the drives that suffered the errors.

It's probably beyond the capabilities of this forum to help with this issue as engineering normally needs to get involved with this type of issue (more than two disks fail in a Raid 6).

You may need to contact EMC support. If the customer has a service contact with EMC you can open a service call using the array serial number. If no contact, you could still open a case with EMC on the AX4 serial number but it would still need to go to engineering for resolution and I'm sure that even that would be possible.

glen

1.4K Posts

April 4th, 2016 04:00

So here's my questions:

1. Is there a way to force the data array back online using the disks?  Even changing serial numbers of the clones isn't a problem, we're a professional data recovery lab and can do that in a few minutes.  But, I assume some metadata needs to be modified to un-mark the members from being "failed".

Yes. You can try bringing 3rd Drive Online. You may try Cold Storage method, changing paddle card etc.

2. We can virtually build the array, however it doesn't appear there's any known file system on there.  I assume it's EMC's proprietary FLARE filesystem.  Is there any way that we can extract files that way, assuming that we can destripe the data from the array to a single image file?

EMC AX-4 runs on Windows 2003 Server Edition. Yes, its a customized Windows OS but the things you want to do and change needs to be done using fcli but AFAIK it doesn't do destriping of the data from the broken array to a single image file or multiple file. Nor does AX-4 has capability like HP's Entry level SAN i.e. HP MSA where you can inform your SAN to trust failed drive as Good drive for DR purposes.

3. Is there even a file system on the data array, or is it all on the OS part of the unit?

Refer Answer 2.

You may need to contact EMC support. If the customer has a service contact with EMC you can open a service call using the array serial number. If no contact, you could still open a case with EMC on the AX4 serial number but it would still need to go to engineering for resolution and I'm sure that even that would be possible.

I don't think so glen. EMC Technical Support and Sustaining do not do Data Recovery and I highly doubt these array are even in active Contract! If such was the case then 3rd Party Data Recovery Team may not have been contacted in the first place!

195 Posts

April 5th, 2016 08:00

Also:  520 byte sectors...not 512.

1 Rookie

 • 

4 Posts

April 5th, 2016 08:00

In this case it was 512 byte sectors.  They are just SATA drives not SAS, so they aren't even capable of 520 byte sector size.

No, EMC support was absolutely no help which is why they were seeking our services in the first place.

We ended up just using a consultant for this case who specializes in EMC.  Case is just finishing up copying data now.

1.4K Posts

April 6th, 2016 02:00

No, EMC support was absolutely no help which is why they were seeking our services in the first place.

They are right at their place. EMC Technical Support is a break/fix support NOT a data recovery support that too for EOSL SAN.

4.5K Posts

April 7th, 2016 08:00

FYI - all the disks used in the EMC arrays use the 520 byte sectors - the disks are prepped in manufacturing at EMC and reformatted to 520 from 512. This if for all the drives - Fiber channal, ATA, SATA, SAS. Flare understands this and expects the disks to be formatted that way.

glen

1 Rookie

 • 

4 Posts

April 7th, 2016 11:00

Nope, you're wrong about that.  SATA drives are generally completely incapable of doing 520 byte sectors, with the exception of some Nearline SATA drives.  But this unit was just using old junk consumer grade Seagate desktop drives which I confirmed were in fact 512 byte formatted.  The extra 8 bytes is just a checksum of the 512 bytes anyway, so doesn't actually affect the data.

Don't forget we are a data recovery firm and all we do is work on hard drives.  Just not usually EMC.

4.5K Posts

April 11th, 2016 09:00

Sorry to disagree, but EMC formats all the mid-range array disks (those used in the CLARiiON prodcuts which includes the AX-Series) at 520 bytes per sector.

glen

1 Rookie

 • 

4 Posts

April 11th, 2016 10:00

Sorry, but you're just plain wrong. Consumer grade SATA drives cannot be formatted to 520 byte sectors, it's impossible.  That's only their SCSI and SAS drives which are 520.  SCSI and SAS enterprise drives can be formatted to 512, 520 and even 528 byte sector sizes.  SATA only has 512, 4K, and 512e (emulated 512 used on 4k advanced format).

Do a little research before you go calling out people as "wrong",  You're making a fool of yourself.

4.5K Posts

April 12th, 2016 11:00

Jared - I stand corrected - the documentation that was available was confusing for the SATA drives for the AX4 - Dan was kind enough to track this down for us.

glen

No Events found!

Top