Previously good drives suddenly reporting wrong sizes

[SecCon]

So lets try starting with the basics.

It is a LSI 3108 in my Huawei server with space for 14 drives, two of which SFF.
I use Avago MSM 17.05.06.00 for all configurations.

I had 2 x 3TB Seagate drives confed as RAID1 . Checks out. Had it for a few months. Never a problem. They are on Port 0 and Port 1.
I add 2 x 3TB Toshiba drives to add some redundancy and storage and try to reconf them all to RAID5. Those are on Port 10 and 11.

As I try to make the RAID5 I see that the result is 5,5GB of space, instead of expected 9TB.

At the same time, looking in to the drive specs, somehow the "old" Seagate drives now report 2,520GB (GIGABYTE) instead of the previous normal 2,7 something TB. Both does.

All drives report 512B sector sizes, so while looking at that it does not seem to be the main issue. I looked at them using a few tools (Windows environment) and CrystalDiskInfo is able to drill through the RAID controller and report the disks SMART and all that, it also gives the same disk size as the Raid MSM software.

Tried booting in to SeaTools X on the Server but they don't see past the Raid controller. Loaded up the drives on a laptop with a USB - HDD dockingstation and booted from SeaTools there, and it can not identify any immediate obvious issues. It says capacity is 2,5GB/ 2,7GB depending on GB or GiB. I think LBA is screwed up somehow.

Any ideas as to what cause two perfectly good drives to suddenly misbehave?

 

[i386]

As I try to make the RAID5 I see that the result is 5,5GB of space, instead of expected 9TB.

Are you sure it's raid 5 and not 10?

 

[SecCon]

Yes.

So now checked both drives.

Both are Barracuda ST3000DM008

Both report (GB/GiB) 2,71 / 2,52 in SeaTools X for Linux. I find no way to alter that, despite running scans and "fix all".
Both have rev. 1.04.
None have available FW updates.
Both have been running good for excess of 13,000 hours.
Both have Max LBA of 5284783
Both have Phys. Sector set to 512
Both have Log. Sector set to 512

 

[i386]

Both have Max LBA of 5284783

Can you acces the hdds and post what the labels on the hdds say?

 

[SecCon]

 

[i386]

I was hoping that there would be the number of lbas on it

Both have Phys. Sector set to 512
Both have Log. Sector set to 512

Labels say sata af (advanced format aka 4k sectors).
Either the labels are wrong or the tools/controller report bs ._.

 

[SecCon]

LBA is reported as 5284783 , what I wrote above.
I guess I might have to check that vs specs on the Seagate site.

https://www.seagate.com/www-content/datasheets/pdfs/3-5-barracuda-3tbDS1900-10-1710US-en_US.pdf

https://www.seagate.com/www-content/product-content/barracuda-fam/barracuda-new/en-us/docs/100804187a.pdf

I never had advanced format 4k sectors conf working, probably due to mixed drives.

Just baffled at the lack of options to actually reset the drives to factory. I checked the raid controller in "bios" offline mode, and there is nothing I can do about the drives there except creating and deleting raid volumes and mark them for removal.

Maybe boot up a storcli session could do something. Since I know have them loose, can you run storcli vs usb connected drives? And if so, would I have better options to reset them?

 

[SecCon]

Been through a bunch of 3TB drives that I was not really planning on using, but considering I have a few slots to spare on the server I was reviewing how to use them as "misc" storage drives, testing and such.

As it is I have 6 drives that are either showing 750GB space, which should be fixable with MBR to GPT conversion back and forth, and some are stuck on showing a few GB of available storage, as posted above.

I have tried several different ways to restore their capacity but have failed on all, despite using:

- diskpart clean,
- mbr > gpt (all are of course GPT to begin with),
- HD Sentinel detects the error, but do not fix it
- Crystaldiskinfo has no tools
- Acronis Partition software do not detect error and do not do anything to fix anything
- Random disk / partition software may detect error but wont say if its gonna fix it if buying a license.

Might boot up Gparted to see if that detects or does anything, but from what I see on the website there is nothing obviously related.

 

[SecCon]

Stilll not solved. Raised the matter in another tech related forum and at WD community since some WD drives are also affected.

Some people give me the standard answer about windows reporting 2,7 TB vs 3,0 TB and try to explain that to me, they don't read the entire post and probably think that when I write 2,7 GB I probably just typed a typo... annoying crap.

 

[SecCon]

Changed the thread title.

 

[jabuzzard]

That sounds to me like the sectors on the drive that holds the firmware have gone bad. When that happens they start reporting the wrong size and often the wrong model numbers. Basically, the firmware on a flash chip is just enough to spin the drive up and load the proper firmware of the spinning platters into RAM. If those sectors go bad then the drive is toast. In the past, I can understand that it saved a few cents on the flash chip. In 2023 that's no longer the case as the smallest flash chips you can buy more than comfortably hold the entire firmware for the drive.

 

[SecCon]

Been messing a bit with DD and its extremely logical syntax [not!] but no closer to a solution. Some drives are of course to old to be relevant, but some others not that much.

 

[SecCon]

So I have a solution for this, or rather, at least it solved some, if not all, of my drives. I am not ready with it yet, but before doing a writeup I point to this thread at Superuser involving a malware situation, which is not my scenario, and a link to Wikipedia article containing further information.

SuperUser post

Wikipedia "Tech" involved and potential solutions

I will get back with a lengthier post, but I already gave enough info for others to have a chance at solving it.