Question 4TB WD hard disk activity light always on, partitions visible but cannot mount, no matter the OS or port (USB/SATA) ?

Dec 21, 2020
18
1
15
0
Hi, been using computers for over 20 years, first time this happens to me and I'm at a loss here, so I hope somebody has new ideas for me to try :-/

4TB Western Digital Blue hard disk (SATA) only 15 months old. Last year, I upgraded from a 3TB WD Blue (approx. 4 years old, never had an issue with it) to get a little more storage space and to rotate HDD so that it lasts longer... it looks like it didn't! But it did work flawlessly until 2 days ago: I was using the computer and an app reported it couldn't write to that hard disk. I rebooted the computer and it never rebooted ever again (with that drive attached). Restarting without that HDD attached, the computer boots absolutely fine.

NO WEIRD NOISE: The HDD does not make any weird noise at all, no spin up weirdness, no grinding, no clicking.

SPIN UP AND DETECTION IS QUICK: Spin up is fast, the 2 partitions on the HDD show up in Linux no problem, not even slow to detect at all. I CANNOT mount any of those partitions in Ubuntu (most recent stable build).

ACTIVITY LIGHT IS ALWAYS ON, HDD WORKING 100% OF THE TIME: If the HDD is attached to any port (tried all SATA ports and 2 different USB ports via an external HDD enclosure) to ANY OS (Ubuntu, Win7, Win10), activity light immediately goes to 100% solid and every 3 seconds I can hear it read or write a tiny bit (like 0.01 second) and every minute, I can hear it read/write a little longer like 0.1 second. This happens without me interacting with the disk at all, the partition does not even need to be mounted.

RE-ATTACHING HDD TO MAIN RIG SPAWNED CHKDSK: After trying with an USB enclosure, I re-attached the HDD to its original location on the internal SATA port and powered on the computer. CHKDSK spawned and it has been scanning what appears to be that HDD for 40 hours now. It is at 72% and I can see it fixing some index errors. It is slow since the beginning but I could see approx. 100 files scanned per second at the beginning of the scan. Now at 72% I can see approx. 100 files scanned per hour. It is still progressing, but at this rate, it's going to take months to finish. I wonder if this is just a couple bad sectors in a row and then it will get faster once it's done with those sectors...

Since I cannot seem to be able to mount the partitions on that drive, I wonder how I could copy files onto another 4TB disk that I have handy here. I understand Ubuntu has a disk image command in the DISKS app but when I tried this, it was copying at a rate of approx. 40KB/s and it said there were 1 year 10 months left... like... wtf... If I could mount the drive, I could at least just grab the most important files first and leave the rest there for a year to copy (lol)

Any ideas for me to try gentlemen? (and/or gentlewomen)

Thank you for reading
 
Dec 21, 2020
18
1
15
0
Warranty replace, and recover the data from your backup.
I don't make 4TB mirrors every day, so I would prefer a real solution. Not some generic "When was the last time you made a backup" sentence that implies you are syncing 24/7, when it is practically impossible IRL.
 

USAFRet

Titan
Moderator
Mar 16, 2013
137,029
6,906
166,190
21,053
I don't make 4TB mirrors every day, so I would prefer a real solution. Not some generic "When was the last time you made a backup" sentence that implies you are syncing 24/7, when it is practically impossible IRL.
Not impossible at all. Not in the least.
All my house systems backup every night. Only the changes since yesterday. Takes but a minute or two, all automated.


But, to your problem...
HDDs are physical devices. As such, they can and do fail. Sometimes when they are very very new.
If you're seeing issues across multiple OS's (Ubuntu, Win 7, Win 10)...then that would indicate a problem with the actual drive.

No, this is NOT the answer anyone wants to hear. But it is a likely issue with this particular drive.
 

USAFRet

Titan
Moderator
Mar 16, 2013
137,029
6,906
166,190
21,053
This outlines my personal backup routine. A bit modified since this was wriiten, but the basics still hold.
It connects to my QNAP NAS, but could easily be to an external drive.

 
Dec 21, 2020
18
1
15
0
Re:backups everyday; It really depends on what you do every day with your hard drive. Some days I can get 100,000 files changed, others only 100. Sometimes 1TB changes, sometimes 100MB. Also, I cannot use internet backups because my upload speed is 80KB/s (max, so when we sleep and all devices are turned off, basically). Yes 80 KILOBYTES a second, you read that correctly. Backing up a single day changes for one hard drive over internet would literally take weeks, where I live. Thanks for your input, but this is not a solution to my issue. I need to get some of the files back.

CHDSK has processed approx. 700 files since I first posted this thread, so it is progressing faster (extremely slowly, but still faster) than this morning. It is fixing errors in index $i30 approx. 2 per hour and about 250 files scanned per hour.

Looking for other suggestions, anyone?
 
Last edited:

DSzymborski

Polypheme
Moderator
The suggestion is to back things up properly. Sorry, but that's the answer. 24/7 isn't necessary for most, but daily certainly is easy and it doesn't necessarily have to be online.

You should at least have a local backup on a separate hard drive. And modern backup programs are smart, they know how to only update the files that have changed since last backup or even, only the parts of the files that have changed.
 
Dec 21, 2020
18
1
15
0
I do have backups, but I don't have every single file. I'm not asking for a solution 1 week ago. I'm asking for a solution to the problem today and that is get some of the files out of the hard drive that is NOT failed yet.

It's like calling an ambulance and having them telling you that the solution to your broken arm is that you should have worn a seatbelt. That is not what the wrecked person is calling an ambulance for.

But thanks for reading and for your copy/paste of the preaching that was already done by the previous member.

Now, anybody with a potential solution to the problem? Any expert in disk imaging, forensic or special software that I could try to salvage at least files that are missing from my last backup? Law enforcement do it with broken hard drives, so there must be some things I could try?
 
Last edited:
Dec 21, 2020
18
1
15
0
If you can't mount it, then it HAS already failed.
Sorry, I should have been more clear: It has bad sectors for sure, yes, but it has NOT COMPLETELY FAILED YET. It is very slow, but CHKDSK has been fixing indexes non stop, it has fixed 2 more and has scanned more than a thousand files since your last reply, so surely this drive still is readable?
 

DSzymborski

Polypheme
Moderator
I do have backups, but I don't have every single file. I'm not asking for a solution 1 week ago. I'm asking for a solution to the problem today and that is get some of the files out of the hard drive that is NOT failed yet.

It's like calling an ambulance and having them telling you that the solution to your broken arm is that you should have worn a seatbelt. That is not what the wrecked person is calling an ambulance for.

But thanks for reading and for your copy/paste of the preaching that was already done by the previous member.

Now, anybody with a potential solution to the problem? Any expert in disk imaging, forensic or special software that I could try to salvage at least files that are missing from my last backup? Law enforcement do it with broken hard drives, so there must be some things I could try?
This isn't like repairing a broken arm. If the hard drive isn't functioning, it's more like repairing death. Find me a productive solution for that.
 
Dec 21, 2020
18
1
15
0
SEE, now you're talking, thank you! That's the kind of advice that I'm looking for in my situation TODAY.

Now, have you tried ddrescue personally? Any experience on how to not have it run for 1 year and 10 months? Do you know an actual command that I could run for ignoring bad sectors and just quickly grab anything it can? Any way to actually tell it to grab only this or that folder on the drive and ignore anything else? Will this command keep the NTFS directory structure so I know where any of those 1.4 million files go in the directory tree? Can I do that command using Ubuntu (most recent stable build) live USB distro? If not, then any USB live distro you could pinpoint that fits on a 4GB USB stick?

Thank you for your understanding.

I read on the ddrescue manual now "It is best that the device or partition to be rescued is not mounted at all, not even read-only." and that is precisely the kind of thing I'm looking for since mounting times out after a minute with an unknown error. I am almost sure Windows has set the drive "dirty" to help prevent any other OS to mount and further damage its contents. This may be why the OS on which CHKDSK has been running for 2 days on the main computer's seem the only OS that can actually scan actual files on those partitions. The partitions may just be set to a special state that prevent other OS'es to mount them. Tough I have tried mounting it with the "ro" flag under Ubuntu and it still wouldn't mount. But CHKDSK seems to be able to scan all the files, slowly but surely.
 
Last edited:

USAFRet

Titan
Moderator
Mar 16, 2013
137,029
6,906
166,190
21,053

img

Dec 27, 2020
13
0
10
0
Hi!

Typical WD and Seagate HDD disease is a slowdown (named "slow responding issue" - usually it has one or more weak heads and the resulting unstable / bad sectors that are diligently collected by the firmware and this slows down the drive - i.e. overrides the firmware to stop write / read operations because sector list data cannot be updated) the first sign is the dying of the drive: weak heads constantly developing bad sectors - if one or more of them in the disk's Service Area - named SA, where the neccessary informations stored to up & running the drive itself). When they start to slow down noticeably it means their end, because under the hood, only the "chewing gum" holds them together: they are full of bad sectors, but the firmware can still replace them from the spare area for a while. In this case (using home methods), your drive can't stand a full user data area backup. If the drive has not stopped completely, special software developed for data recovery can be used to disable the weak heads and turn off realtime error-correcting background processes, thus speeding up the drive, then the other heads can make a Bit-perfect disk image copy of the sectors covered by them at normal speed, and then by switching the faulty heads back on, one can also try to save it's sectors. Thus, by putting as little stress on the drive as possible, there may be a minimal chance of recovering the data without breaking it and replacing the head block (because that would make it significantly more expensive - unfortunately, the heads have often been degraded to such an extent that the replacement cannot be swept away).

HDDSuperclone (it runs in Ubuntu too) has script to manage WD's slow responding issue.

As USAFRet written before, you have to have a copy of the disk first, then you can culling through the files.
 
Dec 21, 2020
18
1
15
0
Yes, I have learned and used GNU ddrescue for 5 days now. I started the process the right way by dumping into a disk image on a larger drive, along with a map file to be able to resume in case of a power outage. The disk image will allow me to duplicate it to a third drive at any step of the process and do all sorts of experiments on it, namely try ntfsfix under Linux and if it fails to mount, then CHKDSK under Windows. I will be attempting this soon-ish.

GNU ddrescue has salvaged about 80% of the drive in the first pass (copying) and that pass should be done by tomorrow evening. I am not scraping or trimming any bad sector yet. There is a very significant amount of them (surprisingly for a new-ish drive) and for the first 2TB of the drive, there were a COUPLE bad sectors here and there, but mostly just recoverable DATA, it was copying often at 20MB/s other times at 10MB/s.

Then there is a surface of about 900MB that it only copies at 1.5MB/s and in that area, it stalls ddrescue every 40MB for 3-4 seconds, then it resumes copying. What's weird here is that I sat for 10 minutes and watched ddrescue's status table and it really seems like there is a bad sector once every 40MB. It's always this amount of scraped DATA then 1-3 read errors. When you think about it, it almost seems like the platter is cracked from the edge towards the center of the disk or something, because 40MB is close to 1 track, if I'm not mistaken.

Anyway, we'll see how it ends. All I want is to see what I can quickly get out of it, as it may not even be worth it in my case to spend weeks on trying to get back any DATA from bad sectors. We'll see.

After ddrescue, I wanted to duplicate the salvaged disk image to play with repair commands and whatnot and it went well for the first 1TB then it slowed to a crawl and would literally take weeks to finish copying that single 4TB disk image file onto a new drive that normally copies at 180MB/s :(

VERY LARGE FILE DUPLICATION SLOW UNDER LINUX SOLUTION: I want to add for all readers coming here in the same situation that Linux has always had (and still has as of December 2020) issues with copying very large files, no matter if it is done using "cp" or "rsync" (I tried both and the result was the same). Perhaps the most shouted situation is copying big files from HDD to USB stick or vice and versa but it seems like it can happen from HDD to HDD as well.
What happens is it will start copying quickly like 120MB/s and as the copy reaches something like 500GB, it will then slow down to a crawl like 5MB/s which will take weeks to finish copying. We're talking about completely sane drives, no bad sector or anything. This issue is known to happen only to some brands of motherboards and/or drives (IMO, this is probably more related to the buggy Linux support for some drive controllers on some motherboards) and has been posted over and over for years on Linux development boards but developers have not fixed this issue yet apparently.
This makes it very time consuming under Linux to duplicate a 4TB disk image and it's completely out of question to attempt ANYTHING with the original disk image that took 1 week to scrape. ANOTHER problem was that the 4TB disk image would not entirely fit onto a 4TB NTFS partition, it was missing a COUPLE OF MEGABYTES to fit. Linux had no problem making it fit for some reason, I think it cheated and used non-standard block size because when I attached the drive to Windows, it would see TWO 2TB partitions instead of just one 4TB partition.
Anyway, I resolved both issues by shutting down Linux and attaching both drives to a Windows rig and then I used the "diskpart" command line program to format as exFAT which allowed a HIGHER BLOCK SIZE than NTFS and this was absolutely essential, because the 4TB disk image that ddrescue produced could NOT FIT onto a standard Windows NTFS partition (max 64KB block size).
  • Attach a temporary drive (4TB) to SATA.
  • Attach the original disk image drive (6TB) to SATA.
  • Destroy all partitions and make 1 full spanning partition on the 4TB drive using Windows Disk Management console.
  • Launch CMD.EXE with admin privileges.
Code:
diskpart
select disk 0 (obviously, make sure this is your temporary drive)
select partition 1
format fs=exfat unit=256k quick
  • In Windows Explorer, copy the disk image file onto the temporary 4TB drive as you would any file normally. This only took 6 hours (!!!) under Windows at 182MB/s, while Linux was still struggling at the 25% mark after 14 hours!
 
Last edited:
Dec 21, 2020
18
1
15
0
So, after 4 weeks, it's now at the 80.67% mark and has been very, very slowly progressing in the 80.6% mark for more than 1 week (the dd rig took some days off since December to do some other tasks and 1 day a week I would clone a full copy to a third drive so I could browse it to get some files back on my main rig). You see, although there aren't a ton of read errors, it is only copying at 1800 bytes/s (yes 1.8KB/s) with an estimation of 13 years to 100%. Every day, it seems to slow down further an average of 50-100 bytes/s.

So it's now clear I have to turn to firmware tweaks in order to speed things up, even if it means potentially damaging the drive (which will never, ever be used after the recovery anyway).

@img I am now exploring your suggestion of HDDSuperClone. Your link does not seem to work anymore but I have found the author's LiveCD images repository here and I am now attempting to boot the current version: https://drive.google.com/drive/folders/0B48xJNqot6ABUjNxa2VaME5INUE

I am thinking of trying this method for clearing the drive's internal relocation list: https://superuser.com/a/1120599

I have nothing to lose, because by the year 2034, I don't think this hard drive will still be spinning, not to mention the PSU, motherboard or perhaps even myself for the matter XD

EDIT: This is what I'm dealing with here under HDDSuperClone. It is aligned with what I've seemed to experience under GNU ddrescue. It would copy at 100MB/s for a couple seconds, then it enters that super slow area (the gray spots in the picture). There is zero (or very occasional) read error. It almost looks like one of the heads has an issue and is reading SUPER SLOW (like 2KB/s). I tried both WD patch mod 02 and 32 but the result is the same. So a bad head or a cracked platter perhaps? BTW, the drive in question is a WD40EZRZ which I believe has 4 platters and 8 heads. (EDIT: It is 6 heads after checking on WD warranty status page)


 
Last edited:
If you enter your drive's serial number into WD's warranty checker, it should tell you the drive's family, RPM and number of heads:

https://support.wdc.com/warranty/warrantystatus.aspx

BTW, CHKDSK is a bad thing to do (in repair mode) against a physically failing drive. Even if you successfully clone your drive, you will then most likely need to recover from the damage that CHKDSK has already done.

I suspect that your drive may be one of the later models where the "slow responding" fix cannot be applied, at least not by HDDSuperTool or HDDSuperClone.
 
Last edited:
Dec 21, 2020
18
1
15
0
@fzabkar I understand and I have not initiated CHKDSK. When I rebooted my computer in mid December, all of a sudden, Windows triggered a CHKDSK and I let it sit over night until I figured something was definitely wrong with the drive. I am not doing anything to the drive, it is not even mounted at all since then. I only "dd" blocks from it and then mess with the disk image it makes. BTW, I could recover most of my files already as the partition table and the directory structure seemingly was untouched by the 1 out of 6 sector that's "broken" on the drive.

I followed the link you provided fzabkar, entered my drive's serial number and I see "6HD STD" (FYI, exact model displayed there is WD40EZRZ-00GXCB0) so it's 6 heads and now it's 100% obvious that 1 of the 6 heads is broken, that's why 1 sector out of 6 is EXTREMELY slow to read (if the DATA returned is even valid, this I should verify next week, otherwise I'll be waiting for years for nothing)

Now wondering if I could just do a heads swap with another WD drive? Is this something that can even be done or are heads not even compatible between models of the same brand?
 
Last edited:
A head swap needs to be done in a "clean cabinet" and requires specialist tools. You also need to understand the requirements for selecting a matching donor. It's not something that a DIY-er would attempt. AIUI you will be facing a cost of $500 minimum, if you were to send the drive away.

ISTM that if one of the heads is totally dead, the most you will recover would be 83%, assuming that the data density is the same for each surface. I hope you are not hearing any bad noises, as that might be indicative of head-to-disc contact.
 
Last edited:
Dec 21, 2020
18
1
15
0
A head swap needs to be done in a "clean cabinet" and requires specialist tools. It's not something that a DIY-er would attempt.
It's either that with a risk of losing whatever is not yet recovered (19% which some of it definitely consists of unused space on the drive) or I wait 13 years (without any interruption) and see what comes out of it. To me, the choice is quite easy to make. I want to reiterate that I already got 80.67% back on a disk image that resides on another drive and that most of my files were recovered in this 80% already.
 
Reactions: fzabkar
There is nothing harmful on that site.

HDD Oracle is hated by the data recovery business, so there are people who constantly try to discredit and undermine it. That particular URL links to an article that was written by me. In fact Google's "safe browsing" bot even flags my FreeBasic source code (ie plain text) as unsafe.

As for the programs, many are designed to access the hard drive at the firmware level, so this may trigger false positives.
 

ASK THE COMMUNITY