SpinRite

[CrazyTeeka]

Kind of ironic that it failed at 97%. To the best of my knowledge, it was not a hard failure and I assume that there is no data of any value to worry about.

https://forums.grc.com/threads/ssd-has- ... line.1385/

Kind of ironic that the user ran a level 3 on that SSD, and then the drive went offline. Funny how SR cannot recover from this, but real tools like RapidSpar or PC-3000 would just repower drive and continue on to 100%.

It just blows my mind how users just throw SR at every drive expecting some kind of magic.

[lcoughey]

If their objective was to recover the data, it likely would have only taken a few minutes with ddrescue. With a log, they could just pickup where they left off, either by repowering the computer or simply reconnecting the power cable on the drive...in comparrison to SpinRite without the hardware ability to repower.

If their objective was to "fix" the drive errors and speed, a 10 second secure erase would have done that.

For people to pay for a program like SpinRite which takes longer and has more risk than a free program, it will never cease to amaze me.

[lcoughey]

With Steve following me, it didn't take long for him to comment.

[Joep]

With Steve following me, it didn't take long for him to comment.

Hahaha, he's following you and Franc and no one else!

[Joep]

If 2 we get either nothing (as drive response is binary: either (corrected) data or error + no data).

Actually, there is at least one well known exception involving "moLD" or "241A" sectors in Seagate Rosewood drives.

https://groups.google.com/g/datarecover ... 5iuWTVCAAJ
https://groups.google.com/group/datarec ... 0.1&view=1 (sector dump)

I can't begin to imagine what SpinRite would do with it.

Sure, this is true. But if we if look at how it's supposed to work drive either responds with error + no data or data. If drive responds without error but with incorrect data this is an exception you almost can not prepare for.

[fzabkar]

I've been trying to get Steve to come clean on what "Dynastat" actually does in modern drives, ie any drive produced since SR v6.0. However, he has been avoiding the question.

AFAICT, SR v6.0 (copyright 2004) predates the ATA 48-bit SCT READ LONG command, which was introduced some time after 2005, so SR v6.0 cannot support drives larger than 137GB (128GiB). Moreover, SCT READ LONG was already obsolete by 2013, perhaps even years earlier, so any foundation for Dynastat's "statistical reassembly" of an unreadable sector disappeared more than a decade ago. Worse still, AF drives have 4KB physical sectors (4096 data bytes + 100 ECC bytes), so these cannot be handled by SCT READ LONG.

To me, Dynastat is little more than a placebo.

[CrazyTeeka]

If only there was a way to spy on the comms between SpinRite and the Drive. If such a tool were to be created it would be neat.

[fzabkar]

You can use GSmartControl to view the drive's error log. The log will contain the last 5 ATA commands prior to, and including, the one that caused the error.

Do you still have the Seagate test drive?

[CrazyTeeka]

Do you still have the Seagate test drive?

No, but I can always grab more drives off eBay.

I was going to try to find some Fujitsu/WD (Marvell)/Samsung/Toshiba drives with bad sectors, so I can clear the G-List myself. But the tool has no support for Seagate drives and newer WD drives which appear to have changed the vendor specific commands...

table.jpg (51.47 KiB) Viewed 66084 times

[Joep]

I've been trying to get Steve to come clean on what "Dynastat" actually does in modern drives, ie any drive produced since SR v6.0. However, he has been avoiding the question.

AFAICT, SR v6.0 (copyright 2004) predates the ATA 48-bit SCT READ LONG command, which was introduced some time after 2005, so SR v6.0 cannot support drives larger than 137GB (128GiB). Moreover, SCT READ LONG was already obsolete by 2013, perhaps even years earlier, so any foundation for Dynastat's "statistical reassembly" of an unreadable sector disappeared more than a decade ago. Worse still, AF drives have 4KB physical sectors (4096 data bytes + 100 ECC bytes), so these cannot be handled by SCT READ LONG.

To me, Dynastat is little more than a placebo.

Yes. So it can not be really doing much else than reading same sector upto 1000, 2000 (?) times. And each read results in error returned or a read the drive itself seems good. And it appears it just keeps retrying even after the drive returned a valid read.

It then keeps comparing the good reads which should all be identical .. In fact if they're not then only reasonable conclusion is that the drive returns garbage.