Intel H10 512GB SSD + 32GB Optane NVMe troubles - Is there a trick?

[ThomasH]

I'm on my 4th recent Intel H10 512GB SSD + 32GB Optane NVMe HBRPEKNX0202AH drive, and I think all, or at least 3 of 4 had a very similar issue where the drive seems to read 1,000,215,216 sectors very but in which the file system is totally screwed and deep scanning produces basically no user data results. I've read them in:

1) A USB-NVMe adapter
2) direct-connected to an NVMe port on an intel 8th gen with suggested optane drivers installed (I think),
3) in the original laptop (booting from a R-Studio emergency usb boot drive)(although the customer in this current case had flashed/updated the BIOS, so I don't have original config),
4) with Deepspar DPI NVMe attachment.

However, the file system seems like it's either damaged, missing a lot, or perhaps a fresh failed Windows install.
Maybe the customers are accidentally reinstalling windows, potentially over an encrypted original install, but I worry this might be different.

I worry that perhaps the 32GB optane part is storing data that is crucial, but I've read contradicting things about whether this is just a cache that would have duplicated files vs added on.
I'm not sure if this is something the drive handles on its own, or if it's managed through BIOS/Optane drivers.

There are some Intel Optane programs that can be installed, and I've seen photos that show that software recognizing the 32GB optane part separately, but at least in my attempts, I was never able to see even a known-good donor as anything but a single disk, even after installing the Intel-suggested RAID drivers and software.

Has anyone else encountered these and figured out if there is a special way to access these that might present a different outcome?
I worry I'm missing something since it's a unique drive, and since the results are odd, despite the drives seeming to read all sectors with corrupted file system.
But, it could be (and seems to be I guess), just coincidence and the customers or the way the drives are failing is just ruining the data / failing Windows reinstall/repairs, and the original partitions were probably encrypted... or maybe the TRIM kills the old data (though scrolling through hex does show huge remnants of non-zero data in my examples, so probably encryption.)

[ThomasH]

One extra peculiarity that I'm 98% sure about is that both 'bad' drives I currently have initially detected in the motherboard's secondary m.2 socket, but after installing the RST Optane Windows drivers and something that updated my motherboard's bios (from intel, not the mb manufacturer), neither of the 2 'bad' drives even detect any more in that port, yet the good donor I have still does.
This makes me wonder if it actually is something that's failed on the ssd that is only observable (such that it stops detection) once the optane-enabled drivers and bios is installed.
Note: the 'bad' drives still detect via the other means (usb, and pcie nvme adapters).
or
This could just be a wild goose chase, I don't know.

[gord]

Hello,

Have same case in now, imaged 100% fine and looks like new OS or something was installed,
Have you found out what's causing this.

Thanks
Gord

[ThomasH]

For what it's worth, the Intel H10 SSDs do indeed seem to require a special adapter capable of reading both channels on the drive. (Probably easiest to find a laptop capable of reading both channels of the device (or better-yet, get the customer's original laptop so you don't have to change anything)). (A motherboard with Optane compatibility isn't enough. It must be specifically compatible with Optane Memory H series.)
"A system that is Intel® Optane™ Memory Ready doesn't mean that it supports the Intel® Optane™ memory H Series."
See: https://www.intel.com/content/www/us/en ... orage.html
and https://www.asipartner.com/resources/in ... quirements

I don't suppose anyone has or knows of an adapter capable of this? I may order a compatible laptop just to be able to try to test these. I've had at least 5-8 of them by now.
A compounding factor that I think is occurring on some that I've seen are that customers have tried to hook them to up to NVMe-USB adapters and let chkdisk try to fix things, which really has thrown a wrench into the process, especially since it's a SSD.
I have one in this state in now, and even deep scans pull basically nothing from it (without access to the optane portion so far).

[300ddr]

Been dealing with the same issues lately and would love some clarity.

Oddly, depending on if I use a PCIe->PCI adapter, or direct to PCIe slot on mobo, and depending on which random motherboard I'm using, the drive appears differently when using a Linux boot disk and UFS. Only one of several PCs I tested in my shop can see the cache partition (27GB). Note that it only sees the 27GB partition and not the other one. But, I still haven't been able to build very good structure even after cloning this partition and using UFS to Build Intel RAID (it's a relatively new option that I've only got to show up using UFS Linux). You can do (I believe) the same thing in UFS Windows by making a Span RAID with the 27GB cache partition on top. But, the User folder, if visible at all, still seems to be missing files (or many are corrupt). I've tried making the RAID with the "bottom" drive being a DDI clone (476GB), UFS clone (using a computer that sees the 467GB partition), etc.

It seems most of these drives also have Bitlocker enabled, so look out for that (sometimes a partition scan in UFS can find it).

TL/DR: same boat, and don't really know an answer. I think buying a computer that uses one of these SSDs may help find the answer. I'd be very interested to boot to Linux and check UFS to see if the Intel Cache partition AND the "big" partition are both found separately.

[300ddr]

Just got another today. IDs as 476gb on DDI-USB Stabilizer with adapter and DDI-PCIe (and Linux UFS on one of my PCs). No raw files found. No bitlocker metadata found. And, when using my computer that HAS seen the 27GB Intel Cache partition on other Optane SSDs, nothing appears at all. My theory is that the cache NAND has failed and the bitlocker metadata was on the cache.

[ThomasH]

TL;DR: I finally recovered one of these properly.
Success seems to require:
1) a newish motherboard with an m.2 slot capable of bifurcation.
2) if Intel Rapid Storage Technology is not enabled in your new motherboard bios, it will detect as 2 drives but not combine them, and no software I had would combine them either. You might want this first just to make separate clones of both the optane and regular ssd, in case something goes wrong in the next step...though maybe that's a waste... not sure.
3) Turn on / enable Intel Rapid Storage Technology in the motherboard BIOS. Then when resetting/rebooting, the system seems to auto-combine or do something that then allows the drive to be properly seen.
4) (I didn't have to install any extra Windows drivers for intel RST or Optane in my case.) --This would allow some extra options and control, but I'm not sure if they'd be safe or ever needed.



I finally got another that this time had not been tampered with prior to coming in by the customer or a computer shop using a regular reader.
I think what happens with these sometimes is that people use a regular NVMe reader and try to recover in a way that alters the original / chkdsk them, which will ruin or diminish chances (at least in a few of them that I've seen).

This one was indeed using the H10 Optane + SSD combined inside the laptop, as any scans on the drive (either part of it) wouldn't detect any real data.
I built a new Intel 12 series system that I confirmed has m.2 slots capable of bifurcation. This is required, and I don't think any adapter is going to help unless it is capable of bifurcation...and unless that adapter is specially built, that I don't think that would be the case.

I installed the H10 drive into an m.2 slot on this new motherboard and powered on. Oddly enough, it only detected the 512GB SSD and not the Optane 32GB drive on first boot. Repowered, changing nothing, and it found both drives this time
Ok, at least it saw both drives. I cloned both into .img files just in case, but nothing I had, including UFS Explorer (which mentions being able to work with Optane) was able to see the data, even after scanning.
After finishing the clones, I went ahead and, in the Motherboard BIOS of my system, I selected the "Intel Rapid Storage Technology" option to "ON" (it was off--which is why it was able to see both drives, but not combine them).
Once it saved that option and reset/repowered I saw a screen come up indicating it was doing something with cache It was too fast to read and was done pretty quick. (I was glad I'd made those clones of the separate drives just in case). After maybe 20-30 seconds of whatever checking/combining/un-caching it was doing, Windows booted and I had access to the customer data automatically. It was Bitlocked, but in the way the DE / R-studio automatically finds the key.

Hope this helps someone else. I'm happy to finally figure this out and confirm the fix.
I'd only recovered 1 or 2 others of these H10 drives (out of about 8-10 total), and those were only recovered either because it wasn't really using the Optane, or maybe 1 that wasn't bitlocked that allowed for at least partial recovery + RAW type recovery.

[300ddr]

Interesting, thanks for sharing!

[osity]

i received a HBRPEKNX0202AH drive in today for recovery and it can be seen by R-Studio as a 476GB. I don't see the 32GB portion.
Was there actually anything wrong with the drive you received?

I can aceess the folder structure without issue but the user profile is missing. What's happening with the user files here?

[300ddr]

>I can aceess the folder structure without issue but the user profile is missing. What's happening with the user files here?

This is the typical issue.

[Roman_TS]

We got such drive on research. It has the same architecture as the Kingston mSATA with PS3108 where two drives (CPU+NAND array) are located on the same PCB but using the same interface (but different contacts in data area).

H10 is similar. Two drives are there - one is Intel Optain + it's own Intel controller (32GB Optain memory) + SM controller and 512-1024GB of NAND Memory.

To read this drive correctly, need to use a special adapter which will take into account different data lines. After imaging of both drives need to reconstruct RAID0 in DE RAID. Structure is similar to Apple Fusion.

[osity]

So the user files are located in a RAID array thats made up of the 32GB optane memory + 512GB NAND?

It only RAIDS the user profile and leaves the rest unRAIDed on the 512GB?

[Roman_TS]

32Gb array is using for caching, NTFS File System files storing, partially - frequently used user files and some other things including the partitions map. We are researching it right now, pretty interesting thing. We are going to collect all information and release an article.

[osity]

Source of the post This is the typical issue.

Right, but is that a/the malfunction of a portion of the drive or is it becuase of the way we are trying to access it?

[300ddr]

I think the former, but hoping the latter for future recoveries

I have one machine that sees the cache portion (only), and the times I've rebuilt the optane raid in UFS, the User files are still never recoverable (or, maybe once, but that's it).