'Mark bad cells' ..? Any tools to manage declining SSDs..?

[Truman HW]

Ive heard references to sectors being marked as 'bad', etc.  (spinning drives) presumably managed by the Firmware. It stands to reason that there exists some analogous such process by which Flash drives do this, as well. Is this strictly something that's done autonomously..? Or can it be done manually, as well, and if so, how would I go about having a program scan and mark sectors / cells to be skipped on an SSD..?

The drive seems to work normally and I haven't had a chance to RMA it yet ... (all critical data is backed up, obviously) ... Still, I don't want files (or in this case a sparse images) I've tried to write to have errors due to the OS seeing the media as healthy. 

I've yet to have any previous data written to the drive later discovered as corrupted, etc. ... 

I wouldn't be surprised if this is much more complicated subject than it would appear at first thought. 

Thanks in advance for any help or knowledge you're willing to share. 

DETAILS:

TASK:          Writing a Sparse Image (OS)

OS Ver.:       10.13.6 and 10.14.6        

Role:          Target drive

Model:         Evo 970 (2TB)              

Program:       Carbon Copy Cloner

[compos mentis]

Defects are managed transparently by the drive. They appear in the SMART report as Reallocated Sectors or Bad Blocks or Retired Blocks.

If a known "dodgy" sector refuses to be reallocated, then in a hard drive one can force this sector to be seen as "pseudo-uncorrectable" by way of the ATA WRITE UNCORRECTABLE command. However, this wouldn't make sense for an SSD as the LBA of a bad sector is subject to change via wear levelling.

[Fraser Corrance]

Why are you writing data to a EVO970? Seems like an odd choice for a target drive. If you are trying to write a MAC OS to a 970 using one of those NVMe/Apple PCIe adapters that may be part of the problem. Some computers do not play well with some NVMe drives. 

I would not rely on the OS to tell you when a hard drive is going bad. I have also seen many of the manufactures tests, like ASD and Dell diagnostic test, say a hard drive passed testing when in fact it had serious issues. Have you checked the NVMe drive's SMART data using Linux? I would also suggest using something like Bad Blocks in Linux to perform a sector test on the drive and then check the SMART data again to see if any of the values that would indicate failure have changed at all. Obviously, don't run a sector test on a drive that you need to recover the data from. In cases like that, I would use DDRescue to image the NVMe in place of a sector test. If the drive fails during imaging, at least you will have a partial image of the drive to recover data from. 

Just some ideas for ya... ;-)

Fraser

[Truman HW]

I have virtually the EXACT same question and virtually the same exact equipment driving my curiosity...

Evo 970 // Retinas

And I also want a means to verify SSD reliability.

Here's where I'm at so far:

Compos: 

I believe these three companies offer demos of DriveDX, Tech Tool Pro and Smart Utility... and here're some screenshots. 

DON'T use the shorter adapters ... use the full length adapters (see attached picture). 

Amazon: ST-NGFF2013-C -- THIS is the GOOD model Sintech SSD Adapter

The B variant isn't as reliable.

Pictures of the OSX apps I've used and some brief notes on each:

1. Smart Utility

a. Provides a quick overview that's simple to read.

2. Tech Tool Pro

a. Scans media for errors 

b. Reports spare threshold used // available

3. DriveDX

a. Clearly shows the number of TB written but lacks other info (media scan + Spare threshold, etc.) 

[Fraser Corrance]

I rarely every uses Mac to test drives because the tools availible are a little too basic for my taste. In my opinion, Linux is far better suited for this purpose. I have seen Tech Tool and even ASD pass drives that were most certainly failing. There are a couple of Linux apps that I use frequently and would highly recommend. Gymager is a handy program for imaging drives but it also will give you far more extensive SMART data than DriveDX will. Bad Blocks is a Linux command line tool that is useful for running read/write sector tests. 

Fraser