We have an "older" IBM disk array that we can no longer get new drives for. We can get reconditioned drives, but we would prefer using new drives. It uses Seagate ST3400832NS disks. However, when we try to install a new drive, it is recognized but is reported as not supported.

I contacted a data recovery software vendor to ask if they had any ideas for how to make the drive work. Their response was that the serial interface at this link might provide the access/interface required to make this happen:

http://hddguru.com/content/en/articles/ ... schematic/

Since the drive reports additional characters in the model number (ST3400832NS IBM...), I wanted to first try to see if I could change this.

I've built the circuit and am now talking to the drive. Changing the serial number using the TMOS commands was easy enough, but changing the model number is apparently not going to be so straightforward (as expected).

My question is this: Can anyone help me figure out where the model number information is stored, and how to change it using the TMOS interface/commands?

Any help would be MUCH appreciated.

Thanks in advance!

-- Rob

I believe you would need to use "Set Stuff" command to alter the model name. However, there might be more than just a model+serial numbers. Anyway, the only way to find out is to try it first

PS. You mentioned a company, is it A-FF/Atola Technology?

_________________
Dmitry

I got everything hooked up last night, changed the model number (setstuff worked perfectly, thank you!), inserted the drive into the array and... I got a new error

The error now is: Unable to read a valid signature sector from the drive.

Any thoughts on how to figure out what/where this signature is? My guess is that I am way out my league with this one... but I've been out of my league for some time now

Thanks a million!!!

P.S. You guessed it... Atola has helped me GREATLY, I can't tell you how much I appreciate it!!!

I have found this document:
ftp://ftp.software.ibm.com/systems/supp ... 642-01.pdf

Here is a quote:


It seems that those hard drives should have some sectors pre-filled with some specific data. The best way to find out is to get a new "certified" hard drive and just make a sector-by-sector copy of it to the patched drive.

Also, there might be an HPA (host protected area) set on those hard drives with some specific information. To find out if those drives have HPA activated, you can use MHDD or capacity restore tool (make sure you do not click Restore ). In case they do have HPA set, you would need to temporary remove it (with MHDD: hpa command), dump its contents to a file (with MHDD: TOF command), then copy that file to the destination drive (with MHDD: FF command), and finally set HPA on the destination drive to match the source drive (MHDD: hpa command).

However, it could be something other than HPA (I do not really think they used DCO since it does not make much sense, but who knows...). They could use some specific vendor commands to access some sectors from the firmware area. In this case it would be somewhat hard to find out what it is... but nothing impossible of course

_________________
Dmitry

Hi Maysoft,
Thanks for your informative post , I would like to know How and Which Motherborads can write information on Hard disk drives? Since i am struggling with Forensic disk cloning , this is very important for me.
I request you to kindly go through my post http://forum.hddguru.com/hard-disk-drives-repair-and-data-recovery-f1/salvationdata-data-recovery-mate-review-t8063.html and give me your valuable suggestions / opinion.
Best Regards
Hddbug

Thank you for the "next step" tips. I am looking forward to trying these things. Unfortunately, the only certified drives I have are currently in production and are used by critical systems. I'll figure out some time to do it, but it may be a while. Worst case, I'll be able to do it when we run out of space I'll post again when I know more.

Thanks again!!!!

Hello

I know this thread is awfully old - but has anyone been able to solve this problem?

We have 20 of these disks with all the official IBM stickers (Part Nr, FRU Nr), Firmware, Tray - still to no use, they do not qualify as 'certified' in our DS4100 / EXP100 combo.

I am quite sure the signature is missing but before I start searching, maybe someone has useful tips...

Regards

P@

100% "signature is missing "

_________________
All went well until I plugged the drive in.

This is where it gets really interesting...

I took out one of the certified drives, to check for HPA and eventually make a sector copy. The disk had previously been marked 'defective' by the controller. I also took out one of the uncertified drives - as destination.

First thing, there doesn't seem to be a HPA. Apparently the signature is in a 'normal' sector. So far so good.

To attach the drives to a SATA controller, I had to take both out of their respective tray, which by the way acts as a SATA to FC converter.

I then did some read-only checks on the certified drive, including a full sector scan, to find out expectedly that it was still OK, even though the controller of the DS4100 had marked it as defective.

Then I decided I'd rather try to rebuild the RAID Volume first, freeing up the Hot Spare and then do a sector copy of the Hot Spare.

So I put the certified disk back in a tray and inserted it in the DS4100. The Controller marked it as 'optimal' but also as... 'uncertified'. I was startled at first because I thought I'd only done read-only checks. Then I realized the obvious - I'd swapped the two trays. So I swapped them back and now the disk is 'certified' and 'optimal' again, and the Controller is currently doing a Copy-Back.

Apparently, each disk is 'mated' to its tray, in other words the signature also contains some sort of link to an unique ID of the tray.

So we are in possession of what seems to be 20 original IBM disks, 20 original IBM trays and probably no way to ever get them to work the way they were intended to. Even if I could be sure that the 20 disks actually belong to the 20 trays, that would still leave me with 210 possibilities, if my maths serve me right.

Thanks IBM. What a pita...

FYI this "drive lock-in" isn't an IBM decision; they just OEM'd this array family from LSI's array division (a.k.a. Engenio), as do many of the other "top-tier" computer system manufacturers.

There are several possible reasons for those RAID controllers to mark a drive as "failed", so the fact that you can read from the "failed" drive isn't a surprise (in fact it's quite common). You'd need to check the RAID controller log from the time when it originally marked the drive as "failed", to understand what problem the controller thinks it detected. If that original issue isn't found & understood, then replacing the drive won't prevent the same thing happening again.

Thanks for the answer.

To my understanding, IBM sold this, labeled as IBM, so they do have direct responsibility.

I do understand drive locking, licenses for Flash Copy and Partitions, you name it. Features and quality assurance don't come for free, I'm OK with that.

What does p155 me off is that all disks and trays obviously are IBM original (as confirmed by an IBM technician), they are in working order, yet still we won't be able to use them. And as this storage unit is quite old, it's not like you can easily get some more drives from IBM anymore.

Sure, the broker who sold them has primary responsibility for this, still I'd have expected original IBM parts to be working in a corresponding IBM storage unit.

Regarding the failed drive, the log isn't very informative, but I'd expect some time-out. As seen with RAIDs from many suppliers (IBM, Dell, HP/Compaq), reinserting the disk is always my first step and most of the time it does the trick, at least until the next time-out.

I politely disagree. IBM didn't say that raw drives from another source (even with the same model number as those already in the array) would work in that array. The spare parts for that array sold by IBM (and the other OEMs of those LSI arrays) are, as you're finding out, specific matched drives & trays, with the drive signature written to them. That's what get sold as official spares & additional drives. Anything else that you get, from other sources, isn't IBM's problem IMHO.

It seems that you've got a kit of drive parts & trays and, as you're finding out, that's not going to work in that type of array.


I understand that you're unhappy, but this is what happens when not buying original parts - sometimes these "lock-in" schemes will bite.


If you bought the drives from a broker, and they said the drives would work in that array, then I completely agree that's a contractual problem for the broker to resolve for you. But that's a bit different from your blanket "IBM must take responsibility" statement that you started with IMHO.


And that was exactly my point - the underlying issue won't get fixed in that scenario. Finding the cause of such timeouts (if that's what the log shows) is usually possible (and even fun) in my experience, though not necessarily easy or quick.

Oh wow, now this is sort of going astray...

What I meant with responsibility is only the decision to sell the storage unit with drive lock-in. As opposed to LSI's responsibility. Nothing more, of course. Again, I respect that decision, but definitely wouldn't want to blame it on the OEM manufacturer.

Now I would still have expected Drives and Trays with original IBM FRU to somehow work... But let's leave it at that. Anyway, it just won't happen.

I'd certainly love to be able to find the cause of the timeout - but do not have neither that kind of expertise, sadly, nor the time.

Anyway, thanks for yours!

P@

P.S.: Vulcan as Avro, Gatling, Kawasaki, Star Trek or something totally different?

You can erase the whole drive and keep the same drive tray and you will have an "uncertified" error also

_________________
All went well until I plugged the drive in.

Yes, absolutely, since erasing the whole drive would also erase the drive signature area. So the result after erasure is an "uncertified" drive, even if the drive was working fine and was "certified" before the erasure. Been there, done that...

Obviously there are workarounds, but they assume you've got a drive with a valid signature to start with, and it seems that the OP does not have that.

Hello again

We certainly do have drives with valid signatures - 8 of them. 7 being in a currently working RAID and the 8th being the Hot Spare.

We bought an expansion unit and 20 more drives at a broker to enhance our storage and that's where trouble started...

As stated, I could do a full sector copy of the hot spare to one of the 20 drives, but I certainly don't expect it to work. I should probably then try to enhance the signature with a reference to the actual tray (I'd actually bet on the IEEE Addr), but where and how... I might still give it a try one of these days...

My previous post was in response to guru (so I was not expecting it to be directly useful to you ) and it was regarding the effect of erasing drives (which companies sometimes insist on before disposal etc.).

I just had a thought: Perhaps that is what happened to destroy the signatures on those disks you bought from the broker? However even knowing that answer won't help with any "fix".

I was not saying that having a valid signature on other disks (i.e. those 8 which are currently in use) will help you to put one onto disks which are currently without a valid signature (i.e. those 20 that you bought from the broker) - it won't help.

FYI not all SATA disks report an IEEE number (a.k.a. WWN) in their Identify Device response data - it's optional. Last time I checked, the drive signature related to at least the drive's s/n (which all SATA disks do report) and the s/n of the interposer card (and possibly something else too), as I mentioned before.

I hope that helps to clear up any confusion after you read my previous comment to guru.

Thanks for clearing things up - I certainly read to fast.

I concur with your conclusions, the broker wiped the disks before reselling them, actually making them useless. Of course he won't take them back, actually he doesn't even answer my mails.

I shall soon receive 2 original disks from IBM. As stated before, we get some timeout on at least one of the productive disks, strangely not when in production, but recurrently when 'collecting drive data' (which obviously I won't try out again). So I'll be happy to have some more spares around.

Also I'll certainly give the working spares a good look before inserting them...

Do you by chance happen to remember any more details regarding the drive signature (for instance approx where it is located?).

Thanks in advance

Hi,
I am up to a similar problem, we have 3 failed drives in a DS4100 and I had to delete one array to assign hot spare drives. We cannot find any "certified" drives. Did anyone find a workaround to this problem.

Regards,
Mihail

Just FYI, as I explained in my first comment on March 1st above, depending on the type (and timing) of the "failure" reported for these drives, it is possible that the problem is not with the drives at all. In that case, you would be wasting your time / effort / money by trying to replace the drives, if the real problem is actually elsewhere.