Samsung 870 EVO - Beware, certain batches prone to failure!

[80251]

In Samsung Magician I have no FC value (252 base 10) in the SMART values for my 870 QVO 4 TiB. Total LBA's Written in the last SMART value indicated in Samsung Magician's SMART for all three of my Samsung SSD's (870 QVO and 2 850 Pros).

 

[E.S]

SMART-Parameter "FC" increased by one again today.
It may continue to increase.
Very disappointing.

It is my speculation, but after writing data, some problem may occur after several thousand hours.

 

[CiTay]

One one of my unaffected drives it went from 1210 two weeks ago to 1222 now. Still no bad sectors. For me, it can do whatever it wants, as long as there are no new bad sectors later.

 

[MEMANIAsama]

Chiming in as this is one of the first forums I found when googling my issue to figure out what was going on. At first I assumed it was my HDD I've been using for ages but was shocked to find out it was my SSD I bought in January 2022. I have a 870 EVO 2TB SSD (MZ-77E2T0B/AM). As you can see by the screenshot I've got those dreaded little red circles in Samsung Magician in Uncorrectable Error Count and ECC Error Rate. Not sure particularly on the manufacture date of my device, haven't checked. I have a firmware update I can do, but I assume that won't magically fix my SSD, the damage is done. I've got to call Samsung directly on their 800 number (blah) to request an RMA because their website allowing me to submit a repair request ticket for it for some reason. Is there anything else I need to provide them or be prepared for?

 

[E.S]

Please read the forum from the beginning before asking questions.
You may also want to read this forum.

2 x Samsung EVO 870 4 TB nach knapp einem Jahr mit defekten Sektoren

Hallo liebes Forum! Ich stehe momentan ein bisschen auf der Leitung, denn ich habe ein kurioses Problem, das ich so noch nie bei einen meiner SSD angetroffen habe. Seit gut 10 Jahren verwende ich Samsung SSD, derzeit ua 2 x Samsung EVO 870 4 TB (neben einer 980 Pro und einer älteren 860 Pro)...

www.computerbase.de

 

[MEMANIAsama]

You may also want to read this forum.

2 x Samsung EVO 870 4 TB nach knapp einem Jahr mit defekten Sektoren

Hallo liebes Forum! Ich stehe momentan ein bisschen auf der Leitung, denn ich habe ein kurioses Problem, das ich so noch nie bei einen meiner SSD angetroffen habe. Seit gut 10 Jahren verwende ich Samsung SSD, derzeit ua 2 x Samsung EVO 870 4 TB (neben einer 980 Pro und einer älteren 860 Pro)...

www.computerbase.de

Ah, that's not in a language I know so not much help for me unfortunately. Thanks though. I've read through the forum but can't make heads or tails of some stuff (this isn't an area I'm too familiar or understanding of.) It looks like a Crucial is one of my next best bet for a SATA SSD even though that had its own share of problems?

 

[Waldorf]

i ve seen most data centers like blaze have reported more than once, that its usually the controller that dies, roughly 25% after 2y of use.
since i build for friends/family, i go through a couple of drives per year, either because of storage need, or because a drive didn't perform as expected (+5GB/s), and lots of them Samsungs, especially early ones.
yet i had none of the fail, incl the 870 im using for almost 2y noe, as i wanted mlc for backup drive, but 2 crucials and one OCZ went from working to unreadable within a week.
then again, i back up important stuff to 3 drives in 2 locations, and never buy the same brand/drive twice, so i dont end up like a customer coming to the shop with 4 dead HDDs from his raid (2006)..

 

[E.S]

One one of my unaffected drives it went from 1210 two weeks ago to 1222 now. Still no bad sectors. For me, it can do whatever it wants, as long as there are no new bad sectors later.

What I'm worried about is that the value that should be counted in SMART ID "05" (reallocated sector count) is replaced with SMART ID "FC" to deceive it? That's what it means.
If so, the continued increase in "FC" is a problem.

 

[CiTay]

I have a firmware update I can do, but I assume that won't magically fix my SSD, the damage is done.

The damage to some of your data is done (those files that are partly stored in bad sectors), but the SSD can be brought back to a fully usable state. I've been using my affected SSD for over 10 months now and the original bad sectors have been remapped with good ones, as well as no new ones developing (probably thanks to the firmware update). Of course, i don't fully trust this SSD anymore, and i shall RMA it eventually, but the SSD can recover from the problem.

Of course there are still many inexcusable things with this failure, for example that it happens without much use, that it makes some files unreadable, that it happens on what's supposed to be the best SATA SSD model. But once you get all your intact files off the drive and delete the couple affected files that can't be fully read out anymore, then the drive should remap the bad sectors and you will have the full capacity worth of good sectors again. And so far the newest firmware has prevented any new bad sectors. But yeah, i won't hold it against you if you RMA even with this information.

 

[Waldorf]

Not sure why some here to expect consumer grade hw not to fail (at all).
Ignoring that this might have been preventable in some form (by Samsung),
anything just close to +99% reliability will cost multiple times more, one reason mil/space stuff isnt cheap.

 

[CiTay]

Not sure why some here to expect consumer grade hw not to fail (at all).
Ignoring that this might have been preventable in some form (by Samsung),
anything just close to +99% reliability will cost multiple times more, one reason mil/space stuff isnt cheap.

You see, Samsung SSDs have been statistically the most reliable component inside a PC, with failure rates well below 1%: https://www.pugetsystems.com/labs/articles/Samsung-SATA-SSD-s-Amazing-Reliability-772/

So when the latest and greatest SATA SSD of them - which costs a pretty penny BTW, i think i paid around 390€ for each of my three 870 EVO 4TB - suddenly can have this kind of fault, without any kind of warning or depending on how much you used it etc., just somewhat depening on certain batches, then this is unexpected, in the truest sense of the word. I absolutely expect these SSDs not to fail, overall. Single ones, yes, you can always get one bad one of anything. But look at this thread, faults left, right and center, and not a word from Samsung about it.

 

[MEMANIAsama]

The damage to some of your data is done (those files that are partly stored in bad sectors), but the SSD can be brought back to a fully usable state. I've been using my affected SSD for over 10 months now and the original bad sectors have been remapped with good ones, as well as no new ones developing (probably thanks to the firmware update). Of course, i don't fully trust this SSD anymore, and i shall RMA it eventually, but the SSD can recover from the problem.

This is good to know. Then I will do the firmware update and hope for the best I suppose. I reached out to Samsung Memory Services and they're telling me it's from a bad SATA cable. Which doesn't seem to reflect anything I've read here, it also doesn't really solve all the bad sectors I have on this thing. I may end up getting Crucial as a secondary and have this one be a back-up, especially if they try to give me the run around on getting an RMA.

But once you get all your intact files off the drive and delete the couple affected files that can't be fully read out anymore, then the drive should remap the bad sectors and you will have the full capacity worth of good sectors again.

Is this as simple as copying everything over to another drive and then formatting this one? Sorry, this is my first real SSD so I'm in foreign territory.

 

[80251]

Yeah Samsung it's all down to bad SATA cables. All those people who reported premature Samsung 870 EVO failures (here and on newegg and on amazon) all have bad SATA cables -- as if an end user can't test for that independently.

 

[Count von Schwalbe]

Is this as simple as copying everything over to another drive and then formatting this one? Sorry, this is my first real SSD so I'm in foreign territory.

Yup, same as any other drive. Make sure you update firmware before formatting and thank your lucky stars that the format table wasn't stored in a block that went bad.

 

[CiTay]

Is this as simple as copying everything over to another drive and then formatting this one? Sorry, this is my first real SSD so I'm in foreign territory.

More or less. Copy everything over, and you will see that a couple files just can't be read out (i think i used FreeFileSync to keep track of them), those you will just have to delete later. Theoretically it's possible to save the intact parts of them, but i didn't bother, those files weren't so important (i have a backup of the important ones). When it's all said and done, the SSD should have remapped the bad sectors with good ones. Each SSD comes with a whole bunch of reserve sectors, because it's inevitable that a few sectors go bad (usually way later than here, because it tends to depend on the amount of times they've been written to). So when that remapping has been done, every last Byte of space can be written to and read from normally again. I literally filled my drive up to the last Byte several times, and so far, no broken files. Still leaves a sour taste in your mouth and i won't put anything important on there.

 

[E.S]

RMA probably won't fix the problem.

Maybe Just replace it with the new firmware 870evo.
The new firmware has a problem with increasing SMART ID "FC".
And there is a possibility that the value that should be counted with SMART ID "05" (the number of reassigned sectors) is replaced with SMART ID "FC" to deceive.

 

[CiTay]

And there is a possibility that the value that should be counted with SMART ID "05" (the number of reassigned sectors) is replaced with SMART ID "FC" to deceive.

We will watch that value, but even if it had something to do with bad sectors, it wouldn't be too relevant to the end user, as long as the files stored on the drive never get affected. If the drive can prevent that intelligently, and prevent it for the whole guaranteed TBW or more (most SSDs can write almost twice their TBW before any drive failures) as well as 5+ years, then it would behave exactly as expected and everyone would be happy.

 

[MEMANIAsama]

RMA probably won't fix the problem.

Right but at least I can start with a "fresh" slate with firmware setup at the start, and I'm owed one. Still, not a bad time to get that Crucial 2TB, certainly not getting a 2nd Samsung as I originally intended. Screw 'em.

 

[E.S]

We will watch that value, but even if it had something to do with bad sectors, it wouldn't be too relevant to the end user, as long as the files stored on the drive never get affected. If the drive can prevent that intelligently, and prevent it for the whole guaranteed TBW or more (most SSDs can write almost twice their TBW before any drive failures) as well as 5+ years, then it would behave exactly as expected and everyone would be happy.

SMART ID "05" (the number of reallocated sector count) is important. Increasing this is a big problem. Because it is the bad sectors that are relocated.
Also, the relocation area is limited, and when it runs out, it's lifespan.

 

[Watermelon5]

This explains a lot. Someone brought a very expensive Alienware laptop to me a few months ago with two of these in RAID. Both were totally dead. Apparently it just happened overnight. Production date was right in the middle of this range.

 

[80251]

8 months of silence from Samsung on what is obviously a systemic problem.

 

[disenter]

I think it's not limited to 2Tb & 4Tb versions. I have a 870 EVO 1tb that has died as well. My nas wouldn't touch it, SMART info shows uncorrectable error count at 249 where 99 is max, and ECC error also showing the same number, and marking both 'critical'. I was able to repartition it, and it will accept a 'quick' format, but fails a long format, and doesn't want to be written to at all. Only bought it August '21, and manufactured January 2021.
I've sent in to Samsung about it, but can't seem to find anywhere to register it. Hopefully the proof of purchase, and export of smart data from Samsung Magician is enough for them to replace it!

 

[E.S]

The date of manufacture probably doesn't matter.
The problem occurs with old firmware.
Also, the new firmware has another problem as mentioned above.

 

[Waldorf]

@CiTay
<1 does not equal 0, and its still a consumer product, no matter what reliability it has/had,
so i will never bet on its "life", even if it was a drive with zero fails.
we dont even know (iirc) what the cause is, might be a part that isnt even made by samsung.

my important data is backed up (3 times), everything else is a pain in the butt to get but will just take time to dl/copy etc,
so my only issue would be the time it takes for rma and have a new drive back,
major reason i preferred OCZ and Crucial drives in the past (offering to ship new drive right away),
and why i will never go past 1TB for nand, less data to lose/recover.

as example: Porsche is the most reliable sports car maker, its products costing multiple factors more than this,
yet i will never assume it will be a 100% reliable.

 

[CiTay]

we dont even know (iirc) what the cause is, might be a part that isnt even made by samsung.

Samsung has full vertical integration, so there isn't a part that doesn't come from Samsung, i mean the NAND and controller and firmware are all Samsung.

Anyway, i've already posted why i expected better from Samsung.
Yes, one Porsche can also fail. But you certainly wouldn't expect them to launch a new model and after a year there are failures all around the world, with no word from Porsche about it. So...