Perc 5i Drive change

Just for the records

now the OS recognises the raid as a RAW partition

while the Dell UI recognises it as a healthy VD with the same name/RAID level/strip size as before.

One more question please: What does E mean in "Virtual Disk 0 : RAID 5 :1.14432E7 MB"

Well, fortunately I had read inside here some advise NOT to force online a bad drive.

Fisrt I changed it with a new one and then forced all drives to ONLINE.

In between more and more drives went OFFLINE as I pulled wrong cables.

I also did tried to import foreign config (inside the windows graphical UI) but all found foreign configs where the same

"I then changed everything to On line state"

This is where you went wrong.  You should have stopped and asked for help when you discovered you pulled the cable for the wrong drive.

When you say you "changed everything to online state" ... did you force online ALL drives, including the failed one?  

As soon as you pulled the cable on the wrong drive, the RAID 5 went offline.  What you should have done, is put the disk back in, boot to CTRL-R, and Import the Foreign Configuration.  Forcing online that drive alone may or may not have been the problem (unlikely to have caused it), but when [if] you forced online the originally failed drive, then you forced the controller to use its configuration and data, which was most likely out of date - and probably corrupt too ... the controller would have been unable to properly reconcile the mismatched data and would have just scrambled the data in the process.

If you MUST get the data back, then call a data recovery service ... you are looking at $1000-2000 PER DISK in a RAID array for recovery.  Otherwise, just wipe it out, configure/initialize it again, and start over.  Initializing will destroy what little may be left of your data.

Wish I had better news for you.  I'd ask if you have a backup, but the downside to having several TB's of storage is finding a sufficient method to back it up, so I'd be inclined to guess you don't have one.

Yes, the hardware recognizes the array as healthy, but it doesn't have to care about the actual file system, which is obviously corrupted - the degree of which remains to be seen, but is likely quite high, if not complete - it only cares about the storage base on which the file system is laid.

e = scientific notation:

1.14432e7 MB = 11443200 MB = 11,443,200 MB = 11,443 GB = 11.4 TB (the usable size of a 7-disk RAID 5 after the bit math)

"Data-recovery didn't yield hopeful results"

You need to call a data recovery professional, like Drive Savers or OnTrack.  DYI software is often too hard or ineffective.

OK, forget the failed drive, perhaps I can copy it to another drive exactly the same.

Let's say Imy PC was stolen but I found my drives.

I buy another perc 5/i adapter and try to re-build the RAID.

Will my data be there ? Or initialization destroys data ?

"I mix the cables I pull everything out or I delete the RAID. The data remains unaffected?"

I don't understand what you mean by this.

"Is there a way to re-build the RAID and get the data back?"

In my experience, and from what you have said occurred, no, the data on the array has been scrambled/corrupted beyond your ability to recover anything without ADVANCED recovery methods ... only a professional can tell you if there is a chance to recover anything at this point.  Calling is free, recovery is very expensive.

hello again.

Data-recovery didn't yield hopeful results...

I am trying to find a post-catastrophy solution.

If I take away the failed drive and manage to copy it sector-by-sector to another healthy drive

would I have any hope to re-build the RAID ?

Cluster size is the same, the hardware remains the same, the drives are in the right order, what steps should be taken?

Should I first initialize and then try Data-recovery ?

Let's put it another way.

I mix the cables I pull everything out or I delete the RAID.

The data remains unaffected? Is there a way to re-build the RAID and get the data back ?

At this point, your failed drive isn't the problem.  However, in your hypothetical situation, setting everything about your actual situation aside ...

You could buy another PERC 5/i adapter, attach the drives, import the foreign configuration, and have a perfectly healthy array.  There would be no need to "rebuild" a failed drive, unless one was failed before putting the drives on the new controller.  The array would be there, and would be healthy - and if your data were healthy and intact, then it would also be there without issue.  You ONLY initialize when you CREATE a RAID array (if selected), which DOES destroy data.

Consider this:

A seven-disk RAID 5 - for every 6 data bits that are written/stored on the drives, 1 parity bit is written ('p').  The parity bit is used to recreate the missing data on a drive should it fail (RAID 5 allows ONLY for a single disk to fail - one disk's worth of parity is not enough to cover for a two-disk failure scenario).

Let's say that 'a' is a single file written to the RAID array and the numbers are chunks of that data written across the drives ... each chunk holds part of that file.  In this example, each letter will represent one word (file) and each number will represent one letter (chunk).  The '-' is the parity bit ... the controller uses mathematics, based on the original file (word), and generates a value that can help it to calculate the missing data in any single position, should a disk fail (sorry, feline should be feline):

Now, let's say disk 4 fails:

agr_es + parity = agrees
com_os + parity = combos
fel_ne + parity = feline
papers = papers (missing parity, file whole, no need to calculate)

Because of the above, the RAID controller can continue running as if nothing happened (albeit with degraded performance), and when you put in a new drive to rebuild, it writes the missing data it has calculated permanently to the new disk.

Now, let's say, that instead of "rebuilding" disk 4, we replace it with a blank disk, and force it online.  Forcing it online causes the controller to happily assign chunk addresses to the new disk (a4,b4,c4, dp), so that the array structure is sound, HOWEVER, because it was forced online and not rebuilt, the appropriate data values that belong in those chunks, that the controller cold have calculated and written there during a rebuild, were NOT written to the new disk, so the data currently stored in those chunk addresses on that drive are accessed (on a blank drive, they are blank/empty).  So, you end up with this:

Now, image that instead of the word 'agree', let's say that it is part of your partition table, your Windows system files, the registry, file table, file headers, or file data ... they are now unreadable and worthless.  Only a professional can undo this type of damage - and not always - and even when they can, it is often incomplete at best (notice that 'papers' would be recoverable, but in this example, that is only because it is so small - consider larger files that contain dozens or hundreds of chunks and parity bits).

Putting in the original disk and forcing it online "may" repair some files, but would likely now break others, (and would make professional data recovery more difficult).

I really wish I had better news for you - or some super secret trick to try, but what you described as having happened is exactly what I have described in this example.

wow!

perfect!

Really educating! Thanks

But in this example, if I finally find the original disk4 (or a copy of it) couldn't I tell the controller, "hey, this is the real order"

I assume I then have to force everything offline and back online?