Linux SSD blacklist

PreguntoYo · 03-26-2019, 05:09 PM

Hello:

Today the smartd daemon mailed me that a partition of my RAID 1 array is in trouble:

Code:

This message was generated by the smartd daemon running on:

   host name:  deathstar
   DNS domain: example.org

The following warning/error was logged by the smartd daemon:

Device: /dev/sdc [SAT], 4 Currently unreadable (pending) sectors

Device info:
WDC WD10EFRX-68FYTN0, S/N:WD-WCC4J0UV33RT, WWN:5-0014ee-2b75082cd, FW:82.00A82, 1.00 TB

For details see host's SYSLOG.

You can also use the smartctl utility for further investigation.
Another message will be sent in 24 hours if the problem persists.

Uuugh, this is bad. That partition (/dev/sdc2) forms an array with /dev/sdb2.

Today I felt like "experimenting" with Linux and first moved the array data from /dev/sdc to the newly created /dev/sdb2. Then repartitioned /dev/sdc and moved data back to the new /dev/sdc2. Uh-oh.

The content inside the partitions is LUKS encripted. I wonder if I've damaged my files in the process. OK, some questions:

I have some old HDs here which I could use as spare, but this drive wasn't very old, and I'm afraid something similar will soon happen with any of them. I would like to try an SSD. I've heard that some SSDs have errors in the firmware which make them Linux incompatible.

I've googled for consumer SATA SSDs (I'd have enough with 256 GiB) and Samsung appears in most searches, but it also seems to have horrible reputation among Linux users.

Is there some updated list of Linux incompatible SSDs which I should be careful about before buying?.
How could I guess which files where affected by those 4 unreadable sectors?

Thanks, any help appreciated.

jefro · 03-26-2019, 07:30 PM

"list of Linux incompatible SSDs" My guess is that the SSD will perform basically the same under any OS if used in similar settings.

"How could I guess which files where affected by those 4 unreadable sectors" You won't like this answer but your only choice would be to have some reference like backup or checksums. It may be possible to run some file check to identify the state of the files and just do a test run and report the files marked bad.

Others may have better ideas.

rknichols · 03-26-2019, 07:43 PM

Any of those unreadable sectors would have caused an I/O error when you moved the data to another disk. If you told the copying program to ignore I/O errors, then that information is lost. Now that you have copied data back to sdc, any of those bad sectors that have been rewritten will have been remapped to spare sectors, so there is no longer any way (other than some very sensitive timing tests) to determine which sectors were bad.

capt ron · 03-27-2019, 04:35 PM

Regarding Samsung SSDs, I've run a number of them going back to the 840 series on Linux machines without problem. Currently using 1TB 970 Pro NVME m.2 and 850 Evo SATA devices on this box. The m.2 is the boot device and has hosted both Windows 10 and Ubuntu Linux, although at present I've shuffled the Windows installation off to a VMWare virtual machine because I don't use it very often

First I've heard Samsung SSDs were horrible on Linux. I hope mine don't read this.