Beware what SSDs you use...

[John Larkin]

On Fri, 18 Aug 2023 06:56:45 GMT, Jan Panteltje <alien@comet.invalid>
wrote:

Western Digital, SanDisk Extreme SSDs don’t store data safely, lawsuit says
The suit is seeking class-action certification.
https://arstechnica.com/gadgets/2023/08/lawsuit-takes-western-digital-to-task-over-sandisk-ssds-allegedly-erasing-data/

I think that bit densities have been pushed so hard that charge
storage per bit is marginal, which is fixed by massive error
correction. A little process variation could wreck that.

Hard drives also push bit density to the limits, and do many
corrections, but for some reason that works better.

 

[boB]

On Mon, 21 Aug 2023 21:01:23 -0700, Don Y
<blockedofcourse@foo.invalid> wrote:

On 8/21/2023 8:11 PM, Don Y wrote:
Spinning rust seems to give more of a \"heads up\" to coming
failures (unless the motor/actuator fails).

[Having said that, I\'ve only encountered two disk failures in
40+ years]

While *I* have only had two drive failures, I designed a box
to sanitize and characterize disks for reuse. One of the things
it does is monitor performance of the drive *while* exercising
and sanitizing it to determine if \"reuse\" is realistic.

Lots of disks from businesses see oodles of PoHs!

This SSD became unbootable but was basically readable externally after
the crash. The data from the logs was not there though.

So I put Ubuntu Linux on it and it worked but I had a problem with
that Linux crashing a day later. Assumed the crash was the problem.
Maybe I will take it apart to look inside now. It\'s only 256 GB
anyway

boB

 

[Don Y]

On 8/23/2023 1:12 PM, boB wrote:

Lots of disks from businesses see oodles of PoHs!

This SSD became unbootable but was basically readable externally after
the crash. The data from the logs was not there though.

One of the magnetic disks that I had fail suffered a similar fate.
I scratched my head wondering how it could still be intact -- yet
not boot. But, it was only a 160G drive so not worth the trouble.

The other drive was a laptop drive that I\'d cobbled into a
diskless workstation. It kept wanting to spin down. And, cron(8)
kept spinning it right back up. Over and over again (as the box
ran 24/7/365). This proved too much for the drive.

[I\'ve a similar drive in a similar box but don\'t let it spin down
and its run reliably for a couple of years, continuously]

So I put Ubuntu Linux on it and it worked but I had a problem with
that Linux crashing a day later. Assumed the crash was the problem.
Maybe I will take it apart to look inside now. It\'s only 256 GB
anyway

When media was expensive, it made sense to see if there was
some \"correctable\" cause to try to salvage an \"investment\".
35? years ago, I used to run with external SCSI drives;
I bought a dozen 4G (!) drives at $1K/ea.

A drive crashed on me, unexpectedly. I replaced it with
it\'s clone -- and THAT crashed! (Um, no... I don\'t believe
*two* drives crashing!)

Before I resorted to the third copy of the data (on MO media),
I did some digging and discovered there was a bug in the SCSI
driver for the (new -- at that time) version of FreeBSD that
I was running.

Rolled back to an earlier version. Reformatted the drives
and reloaded the data from the MO disks.

Now, my sanitizer has a set of adaptive criteria that it uses
to decide which drives are worth reusing and which should just
be scrapped. The *last* thing you want to do is put an
unreliable drive into service and have to \"support\" it!