Solutions for Western Digital HDD Slowdown/Failing

[imported_waterloo]

Hi,


Would like to seek advice on the state of a Western Digital (Black?) 2 TB HDD (roughly 2 years old) which seems to be failing.
My storage system is set up such that an SSD is the primary drive while I have the WD hard drive and another 1 TB Samsung drive (5 years old) as secondary drives. I've installed Windows 7 on the SSD while games and other programs (Chrome, Desktop data etc) are installed or allocated on the Western Digital.

The symptoms on the drive started 2 days ago and are as follows:
- The first thing that was wrong was my desktop icons took ages to load (previously was instant, now takes up to 1 min plus)
- Maybe 5x significantly slower opening of some folders (the files within takes time to display properly), compared to opening files and folders on the Samsung (more or less instant, the WD loaded instantly in the past as well)
- Tried loading some games, takes significantly more time to load but do load in the end (e.g. Far Cry 4 from main menu in to the game)
- Steam games unable to update, receives Disk Write Error
- Chkdsk automatically started on Bootup for the WD drive and kept running everytime my computer reboots


I've tried the following fixes:
1. In cmd.exe, fsutil dirty query d: and found that my D drive was "Dirty"
2. Ran chkdsk /f /r d: to do a full chkdsk scan, discovered and restored roughly 4-5 files then it booted back to windows
3. Ran Malwarebytes, no malware discovered
4. Ran HDDScan as diagnosis test and from what I remember there is 200 Current Pending Sector Count and around 200 Uncorrectable Sector Count (if I didn't recall wrongly)
5. I've tried copying large chunks of data (2x 200gb plus files and folders) from the WD to the Samsung and the transfer speeds seemed fine (lowest was 25mb/s while the highest went up to 70mb/s). However, there were a handful of files which showed the error message "unable to read from source file or disk" and could not be copied over (roughly 5-6 small files)
6. The hard drive was still slow thus I did a couple more chkdsk scans ( /f only ), but doesn't fix the slow speeds and no bad sectors are reported


I'm thinking of trying the following:
1. Run HDTune to find out if there are bad sectors
2. Use a software like HDD Regenerator or Spinrite to 'repair' the drive?
3. Copy all data and format the WD drive wiping out all data completely (this would be my least preferred solution though and I'm not sure if I should still use it afterwards)

Any ideas for what I can do next? Any help is greatly appreciated!
(Thanks for reading this through it is a large chunk of information)

 

[Soulkeeper]

Post your full smart log
"200 Current Pending Sector Count and around 200 Uncorrectable Sector Count" <-- sounds like bad news
Backup anything important if you havn't already.

 

[eton975]

Backup as you laid out then RMA.

 

[imported_waterloo]

Hi thanks for the quick replies!

I'll post up my full SMART log after work. But in your opinion the drive is likely to be failing? Or it could be other reasons (i.e. fixable by a reformat). I'm not sure if I could still RMA as I've lost the invoice for the HDD

 

[Captain_WD]

~snip~

Hey there

I'm sorry to hear about the WD Black drive. I would also suggest backing up everything important to another drive and then perform some more tests before applying for a RMA procedure. In order to be eligible for it you'd need to run WD Data Lifeguard Diagnostic tool and see if the drive fails either the quick or the extended tests. Here's a link to the tool: http://products.wdc.com/support/kb.ashx?id=lJ9Hhw
You would get the best info about your drive's condition from the drive's raw values of the S.M.A.R.T. status. Feel free to post the RAW values here so we can give you further info about the drive's condition.
Good job on the troubleshooting part so far.
Unfortunately bad sectors is something that cannot be permanently fixed. They can be isolated for a while but eventually more and more will appear. Physical bad sectors and actual damage on the drive's platter which forms small air pockets. Those pockets cause the read/write head to bounce up and down whenever it passes over them and eventually lands elsewhere on the platter, causing more damage.
Post back with some results,

Captain_WD.

 

[CiPHER]

Post your full smart log
"200 Current Pending Sector Count and around 200 Uncorrectable Sector Count" <-- sounds like bad news

Those most likely are normalised value. 200 probably is the best value for that drive, which could correlate with 0 bad sectors. Only the raw value is relevant in this case.

So TS, please post the full SMART data. SMART is always the first step to diagnose your harddrive.

Good job on the troubleshooting part so far.

Not really - he may very well have destroyed all evidence of his past problems by diagnosing this drive. Sadly, this is what many people do: they use chkdsk, surface read applications and the drive manufacturer tools. But this may fix the actual unreadable sector and in this case the Current Pending Sector is subtracted, destroying evidence of the bad sector having occurred in the past. If the sector was physically damaged, it will be replaced and you can still see some evidence in the form of Reallocated Sector Count. But in most cases (90%+ today) the sector is not physically damaged but is simply unreadable due to too many bit errors.

Harddrives are designed to generated bad sectors this way, because their ECC errorcorrection is not sufficient to combat the problem of bit errors. The manufacturer specifies this as uBER: uncorrectable Bit-Error Rate. Consumer-grade drives are rated for 10^-14 uBER which correlates to one bad sector per day (100% duty cycle) to once per few months with a more viable duty cycle. This is only true when testing many drives over a long period though, so individual drives can vary greatly. But with these kind of bad sectors, the drive itself is not damaged. Harddrives are designed this way, so the manufacturers make more money selling uBER 10^-16 drives to Enterprise users, which generate up to 100 times less bad sectors due to insufficient error correction.

Unfortunately bad sectors is something that cannot be permanently fixed. They can be isolated for a while but eventually more and more will appear.

If you see the Reallocated Sector Count increasing steadily, this may be true. But in most cases it concerns uBER bad sectors and that is normal. So normal that the manufacturer specifies how often this occurs - on average.

 

[Soulkeeper]

Those most likely are normalised value. 200 probably is the best value for that drive, which could correlate with 0 bad sectors. Only the raw value is relevant in this case.

So TS, please post the full SMART data. SMART is always the first step to diagnose your harddrive.


Not really - he may very well have destroyed all evidence of his past problems by diagnosing this drive. Sadly, this is what many people do: they use chkdsk, surface read applications and the drive manufacturer tools. But this may fix the actual unreadable sector and in this case the Current Pending Sector is subtracted, destroying evidence of the bad sector having occurred in the past. If the sector was physically damaged, it will be replaced and you can still see some evidence in the form of Reallocated Sector Count. But in most cases (90%+ today) the sector is not physically damaged but is simply unreadable due to too many bit errors.

Harddrives are designed to generated bad sectors this way, because their ECC errorcorrection is not sufficient to combat the problem of bit errors. The manufacturer specifies this as uBER: uncorrectable Bit-Error Rate. Consumer-grade drives are rated for 10^-14 uBER which correlates to one bad sector per day (100% duty cycle) to once per few months with a more viable duty cycle. This is only true when testing many drives over a long period though, so individual drives can vary greatly. But with these kind of bad sectors, the drive itself is not damaged. Harddrives are designed this way, so the manufacturers make more money selling uBER 10^-16 drives to Enterprise users, which generate up to 100 times less bad sectors due to insufficient error correction.


If you see the Reallocated Sector Count increasing steadily, this may be true. But in most cases it concerns uBER bad sectors and that is normal. So normal that the manufacturer specifies how often this occurs - on average.

Yes, I know. That's why I suggested he post his complete smart log.
I'm not new to hardware and havn't been for many years.
I just tend not to go into massive detail when making suggestions to people. It's easier for me to maintain my life outside of forums if I don't become too invested in my posts. Besides there are a lot of other people helping as well, like you

 

[Captain_WD]

~snip~

Point taken Posting the results and the tool that OP used would help figuring out if the 200 value is the RAW or the normalized one (most probably the latter) and to actually see what the program reports.
The speed of which the Reallocated Sector Count changes indeed can serve as a tool to check if those are physical bad sectors or simple bit errors. Would you recommend filling the drive with 0s and 1s several times to reset it and fix potential errors?

Captain_WD.

 

[CiPHER]

The speed of which the Reallocated Sector Count changes indeed can serve as a tool to check if those are physical bad sectors or simple bit errors.

Well, if there is no physical damage but the sector was unreadable due to bit errors exceeding the ECC error correcting capability of the drive (uBER bad sectors), then Reallocated Sector Count is not increased at all since the sector is not swapped with a reserve one.

So in this case, overwriting the sector will NOT cause the sector to be swapped for a reserve one - it will simply stay in use. This means Current Pending Sector can be 0, Reallocated Sector Count can be 0 so any evidence of there ever being a problem is simply gone! This is why i recommend querying the SMART before doing anything else like utilities that try to check the harddrive for errors. Since doing so might 'destroy' the evidence of bad sectors and you will never know the exact cause of the problem.

If you first query the SMART and see Current Pending Sector has a raw value that is not zero, and after the zero-write / long format you see it as 0 again, but Reallocated Sector Count also stays at zero and/or did not increase in value, then this means the sector was not physically damaged but simply unreadable. This is what i call 'uBER bad sectors' since it relates to the uBER specification that manufacutrers specify for their drives. Usually 10^-14 for consumer-grade drives and up to 10^-16 for enterprise drives.

Because uBER is relative to the capacity/performance of the drive that scales with data density, the problem of uBER is increasing each time the data density is increased. So with 1GB harddrives the uBER was not a problem at all and about 99%+ of all bad sectors were bad sectors with physical damage that needed to be replaced with reserve sectors. But today with very high data density harddrives (1TB+) and the uBER being the same, the problem is much bigger and in fact i estimate about 90% of all bad sectors on consumer-grade harddrives are uBER bad sectors.

They are just as dangerous for older filesystems like NTFS as regular bad sectors, but will simply disappear when overwritten. No evidence of their existence can be found in the SMART attributes, only prior to overwriting them. Sometimes, simply trying to read the sector - which triggers error recovery - can in fact solve the issue. But in most cases the uBER bad sector simply stays there until overwritten and after overwritten all evidence is gone.

Would you recommend filling the drive with 0s and 1s several times to reset it and fix potential errors?

Refreshing the drive surface can prevent so called 'weak sectors' from becoming 'bad sectors'. The difference is that weak sectors have bit errors but can still be read due to ECC bitcorrection. But once the number of bit errors deteriorates, it can exceed the capability of ECC and at that point it will become a bad sector (Current Pending Sector).

Some harddrives have built-in mechanism to detect these. It is called BGMS - Background Media Scanning. This will read portions of the harddrive when the drive is idle and attempt to find weak sectors and overwrite them before they become unreadable. Seagate has drives that do this. The downside is that the harddrive is active even when you are not using it. So higher power consumption and more heat is a result; 8W versus 5.5W idle in my case for the Seagate 7200.14 series.

There are also utilities that can do this. Spinrite for example, though obsolete today, has the ability to refresh the contents of the harddrive by reading and then writing the exact same contents back to the drive. This is useful if the drive does not have BGMS and you do not write/modify the contents of the drive often. So if you only read from a drive and do not overwrite the contents for multiple years, this is recommended.

But other utilities like 'dd' under Linux/BSD or a simple defrag under Windows will achieve the same effect.

In some cases, you may find bad sectors (Current Pending Sector) in areas of the drive that are not in use by the filesystem. For example you have 40% free space on your harddrive, the free space is not used for a long time, maybe multiple years. Those sectors may deteriorate and become (uBER) bad sectors. This is a result of 'retention' - where sectors that are not used for a long time will degrade and introduce bit errors that can cause them to become unreadable. However, this is no problem at all because these sectors are not in use by the filesystem and when the filesystem wants to take them in use by writing to them, the old contents are discarded and once overwritten the sector works perfectly again. So the problem of these bad sectors will disappear automatically.

Bad sectors are only dangerous if you use a legacy filesystem of the 1st or 2nd generation (FAT, NTFS, HFS, Ext2/3/4, UFS1/2, JFS, XFS) and those sectors are actually in use by the filesystem. In all other cases, the bad sector will disappear once they are put in use again.

Modern filesystems of the 3rd generation (ZFS, Btrfs, ReFS) have multiple mechanisms to prevent bad sectors from damaging the filesystem, even on single disks without redundancy. This is very much needed because of uBER becoming a bigger problem as harddrives get bigger and bigger. Besides, it is not acceptable that in 2015 your filesystem becomes corrupt because of one tiny spot of your harddrive cannot be read. C'mon, we want reliable storage in this era! It is not rocket science...

 

[RecoveryForce]

This is a common issue with newer western digital drives. The firmware has an issues that causes slow response which is triggered by bad sectors. During the data recovery process, we can clear the logs that are causing the slow reading symptoms, but it doesn't do anything to prevent it from happening again. But, our goal is strictly to get the data off the drive.

I don't know of any "cheap" programs for you to fix the firmware issues yourself. Copy the files off the drive and send it back for RMA replacement. If you find that copying is too unstable, you may be wiser to get a full sector-by-sector clone with a program like ddrescue that can handle bad sectors and has a log so that you never have to read the same sector twice.