I have more than a year's worth of backlog on reading the replies, i think i stopped getting eMails for new replies and then didn't keep track of it anymore.
However, i wanted to post an update on my "failed" 870 EVO 4TB that caused me to write this thread. As a reminder:
Funny thing though: I did not RMA my drive (the failed one from the first post) yet, to see how it develops, and so far, it hasn't developed any new bad sectors. Maybe the newest firmware SVT02B6Q prevents that somehow. I filled the drive to the brim a couple times (0 bytes free space), but it seems that once you have deleted the files which contained the bad sectors, those bad sectors are indeed mapped to intact ones and are not written to again. This is how i never got another bad file to this date, and the bad sector count remained the same at 329 for that drive.
I will RMA it eventually, but i thought this is an interesting observation. I purposely use this drive more heavily than before (of course only for files i can easily download again), but ever since i deleted the bad files, the drive is behaving. Bad sectors seem reallocated to good ones and no new bad ones developing.
So yes, i kept using this failed SSD the whole two years since i posted this thread.
This was the initial state:
I should note that "Uncorrectable Error Count" and "ECC Error Rate" both went to 4819 shortly thereafter, as i was trying to pull my data off the drive and it got stuck on the files with the bad sectors. Once i had moved off all the data from the drive, and deleted the files i couldn't read out anymore (the ones driving up the two aforementioned values), those bad sectors got remapped. And with the newer firmware, it didn't develop any new ones.
Fast-forward to present day:
The same drive, but with 54 TBW more and loads more hours of runtime. None of the values that are critical for this error have increased from then until now. As i explained, shortly after the first screenshot, "Uncorrectable Error Count" and "ECC Error Rate" both went to 4819, so they were already like that two years ago.
I can honestly say, i didn't spare this drive, i used it for downloads, extracting files, i filled it completely (0 Bytes free) many times, i operated with less than 2 GB free for a long time, downloading stuff into the same 2 GB of free space over and over. And i didn't have a single problem again.
The POR (Power-on Reset) Recovery Count is completely benign, it just means that the PC shut off hard (no Windows shutdown), which for example happens several times during or after a BIOS update, this is nothing too important, just a log entry. The mysterious "FC" value at the bottom rose by a grand total of eight, still not sure what it means, but doesn't seem to be too important, all the files seem to be ok still. I have copied many TB off the drive and never had another read error.
Now, don't get me wrong, the original fault was horrendous, i would have never expected it from such a drive. It is inexcusable, and the handling of the issue from Samsung is poor, they are just silently doing RMAs and no public word about it, at least i didn't see anything. So i guess we all agree on that.
What i just want to say is, if someone doesn't want to or can't do an RMA of the drive, mine has fully recovered, it seems just as good as the unaffected drives now. With the five years of warranty, i can still RMA it when the time is right.
In the meantime, i have purchased several other Samsung drives, a 990 PRO 1TB as my main boot drive, an 870 QVO 8TB for storage, and another three 870 QVO 4TB. So i haven't given up on Samsung just yet. Although i am not nearly as enthusiastic about them, and i also have a WD Black SN850X 4TB now.
BTW, a family member's SanDisk Ultra 3D suffers from a known problem which was first observed on the identical WD Blue 3D, that the oldest files become extremely slow to read out because the controller fails to refresh them (
article in German,
Google-translated). He is currently reading out his files over the span of weeks, at under 2 MB/s! We estimated that he needs about 100 hours to read out the remaining files. At least the files themselves all seem to be ok...