Refurbish partially failed hard drive?

Question

A friend gave me an external hard drive that had been performing terribly to see what I can diagnose and fix.

First step was shucking the drive to eliminate the USB controller as a variable. Then I chucked it into an old PC and ran chkdsk. It took 2 full days (>40 hrs runtime), but it eventually completed with ~8000 bad sectors. The event viewer additionally reports controller issues - The driver detected a controller error on \Device\Ide\IdePort4.

I did run the HDD manufacturer's diagnostic tool (it only detected the drive when the enclosure controller was attached, but eh). It reported an error and stopped, and there is no documentation on the internet that says what the error code means.

Copying is nearly impossible as the transfer starts and the device starts thrashing immediately, rw heads clicking, while transfer is stalled. I do however suspect that the problem is only limited to a certain area of the hard drive.

So I wonder - is there some application out there that will map the problem areas and then create a partition around them or fill that area with hidden files or some tool to mark all slow areas as bad blocks or any other method to recover some functionality?

Answer 1

So I wonder - is there some application out there that will map the problem areas and then create a partition around them or fill that area with hidden files or some tool to mark all slow areas as bad blocks or any other method to recover some functionality?

ddrescue will provide you with that information about unreadable areas with its log file. Creating a partition around defective areas is not a smart concept but it shows up irregularly on this site. Why splitting the space of hard drive in potentially small unsuable areas when FAT and NTFS file systems can store information about bad sectors? Invoking chkdsk with the right parameters does the job.

Unless the defective drive is the last one available and assuming you work on the European Space Station and the pizza shuttle has already left for the day there is no need to waste time on broken drives.

Answer 2

Modern hard drives reserve some fraction of their space as replacement blocks. New drives typically have a small number of bad blocks, which, when found, are replaced out of this pool.

As a drive ages, a few more blocks may go bad and are replaced.

Eventually, the drive may collect damage caused by debris floating around in the drive. When this occurs, it will rapidly gain new bad blocks and the replacement pool will be quickly exhausted. The SMART algorithms and internal diagnostics will pick up on this before the pool is exhausted and make warnings available to the operating system and the BIOS before the drive fails -- although this is not always relayed to the user.

By the time that tools like chkdsk are needed to find bad blocks, the drive has exhausted all of its replacement pool and the drive is well on the way to its final death and data loss is inevitable. Attempts to "repair" the drive with chkdsk will only make things worse as it will spread the debris around inside the drive and cause new scrapes on the surface until even the internal calibration tracks are damaged and the drive will then fail to even initialize properly on power up.

The drive you have is basically now bricked. It is long past being fixable. Any "refurbishing" would basically need to be done at the factory and would involve a thorough cleaning and replacement of the internal magnetic media and likely the drive heads as well, and probably most of the rest of the internal active parts too.

Answer 3

This is obviously not a new problem and people have thought about the same solutions I had in mind before.

Some research revealed the following resources:

Repartition Bad Drive [Windows Program] - does almost exactly what I wanted on the tin. Will scan your drive and create partitions around the bad areas. Can be configured with min partition size and buffers, to avoid making too many partitions. Creates MBR partitions and RAW volumes. Says IDE mode is faster than AHCI. When I ran the program, it ended up a bit buggy, with numbers going out of bounds (>100% progress, negative sector counts), but the run ended successfully.

HDDScan [Windows program] - can create a list of bad LBAs.

ddrescue [Linux program] - r2d3 says:

ddrescue will provide you with that information about unreadable areas with its log file.

Basic and dynamic disks [Article] - Since you might end up with many scattered partitions, you might need to convert your disk to a dynamic disk, to allow one volume to span across multiple non-contiguous partitions. Also explains differences between MBR and GPT.

Hide Data in Bad Blocks [Article] - A security article, that nonetheless deals with exactly the techniques required to isolate bad blocks manually. Written from a linux perspective. Can be combined with any method that gives you a list of your bad blocks.

---

Initially, based on the pattern of chkdsk activity (lower numbered sectors being slow to scan, while being fast at the end) I suspected that the latter sectors were in a fully usable state. Ultimately the drive I have ended up completely bricked, as the pattern of the RW head audibly thrashing/clicking extended to all areas of the drive when I tried to apply the techniques above. I suspect it's a controller issue or a head crash or something more global at this point.