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

[Koozwad]

So a bit of an update: I was able to browse and recover a significant portion of my files with testdisk. Unfortunately, the one file I was looking for that was saved to my Desktop is missing in testdisk. Not sure if there is anything else I can do to recover it as it seems to have just disappeared.

I'm planning on shipping the drive to Samsung tomorrow, so hopefully its not too late to RMA!

I'm glad that you were able to save most of the files! Have you checked 'C:\Users\%yourusername%\Desktop' for the missing file?

These drives have 5 years warranty last time I checked, and it was a widespread and well-known issue.

 

[wvcaudill2]

I'm glad that you were able to save most of the files! Have you checked 'C:\Users\%yourusername%\Desktop' for the missing file?

These drives have 5 years warranty last time I checked, and it was a widespread and well-known issue.

Yea that was the first place I checked but the file is missing.

it’s just odd that the one file I’m looking for is also the only one not there.

It was a secured 7zip archive file. Perhaps it was only a shortcut saved on the desktop but even then I would think it would be visible.

 

[SPDIF]

Again, backup that thing, and after that you can do anything with any program to search that file. RMA it today, or Samsung can do difficult when you wait to long!
You have been warned.

 

[wvcaudill2]

Again, backup that thing, and after that you can do anything with any program to search that file. RMA it today, or Samsung can do difficult when you wait to long!
You have been warned.

Unfortunately, I was not able to get a good clone. Every time I tried (and with every tool I tried) my computer would completely freeze. I left it in that state overnight and it never unfroze. Hence why I was trying to hunt for specific files I would have liked to recover.

At this point, I'm not sure what else can be done so I am planning on shipping the drive to Samsung today.

It's rather unfortunate that the one file I saved to my Desktop is the one file I cannot locate. The odds of that being the case are so low.

 

[SPDIF]

Not much you can do at this point, you may be unlucky that that one file just happens to be in a bad block in the NAND where the data is lost because of the bad condition.

Where there is no data left anymore, you can't get out anything.
You can't invent data if there is nothing left to read.

 

[CiTay]

RMA it today, or Samsung can do difficult when you wait to long!

So you're suggesting it's down to the day? Wonder what they're gonna say when i eventually RMA mine then, 2.5+ years after the failure...

 

[SPDIF]

So you're suggesting it's down to the day? Wonder what they're gonna say when i eventually RMA mine then, 2.5+ years after the failure.

In many cases they did make it very difficult for me, but okay everyone for his own i guess.

 

[Koozwad]

In many cases they did make it very difficult for me, but okay everyone for his own i guess.

I waited almost 3 years to return mine, but I did it via a webshop. Got a new one immediately(so fast that I even doubt they tested it), probably because it's such a common problem with the early versions. Didn't even send them any evidence that it was a widespread issue.

For anyone RMAing it, you could try it first via the place you bought it from - maybe they'll be less difficult about it!

 

[dockalacek]

Hey,

what's the situation today? Is it still dangerous/risky to get an 870 EVO, or was it just a certain low quality batch back then, and now everything is fine? A friend asked me if I would reinstall Windows 10 on his mom's laptop, who complains that the laptop is slow. Presumably the machine still contains the old hard disk. I think buying a SATA SSD would be like throwing live water on it. Are 870 EVOs nowadays trouble-free, or is it still worth checking after purchase to make sure it has the latest firmware, and it can still die after a few months? Is it still better to avoid the 870 outright? We are interested in the 256GB version. Thanks for any replies. Otherwise I'm still thinking about the Kingston KC600 256 GB, has anyone bought it recently? Has there been a change in chips or controller?

 

[mlwmlw]

Just had this start with a drive two days ago! Was in my hot swap bay and randomly disconnected. Bought a new hot swap bay. Same problem. Stick in another drive, I have no problem.

Manufacture Date: 2023.06

The other drive I put in is also an 870 evo. Works with no issues.

In windows event viewer, I always see:

1. "Disk 4 has been surprise removed"
2. The CMigrationService service terminated unexpectedly. It has done this 1 time(s). (is this related?)
3. The driver detected a controller error on \Device\Harddisk5\DR5. (remember, hot swap, i tried it in different bays too)
4. The IO operation at logical block address 0x19a13f18 for Disk 4 (PDO name: \Device\0000003d) was retried.
5. "An error was detected on device \Device\Harddisk8\DR8 during a paging operation."

I'm going to buy all new sata cables. What else should I try before throwing in the towel?

BTW, guys is there any way to buy these drives somewhere with a cheap, long warranty? Samsung makes me ship it there, if I remember correctly, at my expense. I've had amazon warranties where I don't pay a single thing. I don't want to buy the expensive amazon data protection plan, only a warranty. I can handle my own redundancy and backup.

If I'm going to have drives fail randomly I'd love a warranty on all of the future ones I buy.

This is only the 2nd SSD I've had fail my whole life, 13+ years with SSDs, well over 20 (I have 8 in my PC right now, all 870s with one 860, and I've had many pcs in the past). The only other SSD I had fail was a PNY CS900. I only ever bought one of those, never going to buy another.

But yeah statistically I'm reasonably happy with the samsung evos. With HDD's, I don't know what it was, them getting beat up in shipping I suppose, ordering online I got multiple dead HDD's delivered over the years.

So, who would you recommend to buy these from with a cheap extended warranty, that I don't need to cover data?

Sucks to be shelling out $100 on a drive to replace it in a few months to a few years.

 

[CiTay]

I'm going to buy all new sata cables. What else should I try before throwing in the towel?

Get all your data off the drive (well, as much as you can, some files probably can't be fully read out anymore because they are saved where there's a bad sector), update its firmware. Once the SSD can remap the bad sectors with fresh sectors, you should be back in business. Or you RMA it, Samsung will probably have you do a full scan with Samsung Magician.

 

[chrcoluk]

It is easy to imagine that the controller has gone bad and could be swapped. It would fit the symptom of not being recognized by the BIOS.

One problem is the mapping, knowing the order of data to be read back and so forth, it might be possible but I expect its very complicated.

 

[SPDIF]

I'm going to buy all new sata cables. What else should I try before throwing in the towel?

Just backup this thing if possible and RMA back to Sammy. There's no firmware updates available, only for some old stock SVT01 SSD. SVT03 Firmware is only for new SSD buyed with new NAND. No other SSD with SVT01 or 02 can be updated to SVT03.

Even if bad sectors can be repaired, better RMA it, you want to trust your data to an 870EVO that has bad repaired sector's on it? It's faulty and no firmware update or repair can change this. Better RMA it and get a fresh new one.

Repair is NO repair by all means, they are just replaced with spare sector's. But the bad blocks are still there! So it will further degrade sooner or later.

A good healthy 870EVO should show this screen and nothing else!

As long as the points marked are green, you are good to go.

Even the old HDD from earlier could not be repaired when it had bad sector's on it's platters. Then and even now they advice to replace the disk ASAP and backup everything!
Why does everyone think that an SSD is another story? Bad is bad and must be replaced when it has defective blocks or sectors.
A good working SSD should have NO bad blocks until it's EOL, or Wear Leveling Count is close to 100 percent utilization. As with HDD, there are no exceptions for this rule.

 

[CiTay]

Repair is NO repair by all means, they are just replaced with spare sector's. But the bad blocks are still there! So it will further degrade sooner or later.

There are some misconceptions here. Every (!) SSD has bad sectors from factory. If you wanted an SSD with no bad sectors, they would have to ask at least 10x the current price, because they have to throw away at least 10x the NAND chips while doing validation during production and manufacturing.

No, you have a bunch of bad blocks from factory, but this is then taken as the "zero point" from factory. Meaning, these are the ones already remapped from factory, the user gets to to see zero "Runtime Bad Blocks". At least Samsung hide the factory bad blocks completely. Other brands are more open:

This is a Sandisk SSD which will openly admit the factory bad blocks (1830 here, it was like that from the second i first ran it), and the "Grown bad blocks" which came later (none here).

Furthermore, your statement that "it will further degrade sooner or later", i cannot confirm either. I am using my failed 870 EVO 4TB for over two years now after the initial remapping of the bad sectors, with zero new bad sectors having appeared, despite torturing it and filling it regularly.

A good working SSD should have NO bad blocks until it's EOL, or Wear Leveling Count is close to 100 percent utilization. As with HDD, there are no exceptions for this rule.

This is categorically false. You can read here this extensive and very well-made analysis of SSD longevity, Google-translated: https://3dnews-ru.translate.goog/93..._sl=ru&_x_tr_tl=en&_x_tr_hl=en&_x_tr_pto=wapp

Here is an example, a lowly Crucial BX500, officially 80 TBW endurance. They seriously understated the TBW value, in reality the NAND flash used in this older SSD can do maybe 30x that or more! Here is from an SSD endurance test, you can see that their SSD shows 0% (because they have written way more than 80 TB to it), yet it still works fine:

Conclusion: Even with remapped sectors, and even if CrystalDiskInfo freaks out about the health status (which is mostly calculated from the SSD's wear indicators, which are adjusted to an often highly conservative TBW), the SSD can still run fine. Not always of course, but according to tests such as the above, easily often enough to invalidate your statement, i'm afraid.

Mind you, i'm not here trying to defend Samsung. I've said this before. If i wanted to defend Samsung, i wouldn't have posted this thread. But i will not accept blanket statements which are incorrect. We have to be quite precise when we give any advice.

 

[SPDIF]

I am not here to defend Samsung to, there are many other good brands available today. But added bad blocks during life time is a bad sign, even if it where an old HDD, i have enough experience over 30 years selling this stuff.

I totally agree with what you're saying. But I'm afraid it doesn't matter at all here. We are talking about bad blocks that suddenly occur with a new SSD. This is also not accepted with an old HDD.
Nobody wants a new HDD with bad sectors, so you shouldn't accept this with a new SSD.

And you hear the same explanation everywhere, you should replace a bad hard drive as quickly as possible and make a backup of it. So you think we should just accept this as normal? I do not think so. An SSD may not show bad blocks when new, even after use for a while. Point other line. You don't want a repair on your SSD at all. That's why I said sooner or later, so take a backup and buy a new one if your SSD or HDD shows bad sectors.

Bad blocks on an SSD that are repaired with spare blocks are basically the same as a car with oil pressure problems, you add some fresh oil and you just remove the light. So no more problems?

 

[Sora]

an SSD will not ship to you new with any remapped sectors in the smart stats, this data is wiped for normalization prior to boxing and sale.

 

[SPDIF]

They ask what to do, so to stay on the topic it's simple, RMA back and get a new one at home, no discussion. And there are other good brands available now, some even better then Samsung! But the 870EVO stays the most wanted SSD in the SATA segment.

 

[xorbe]

I flashed my 870 EVO drives before I ever saw any problems. Are they still doomed?

 

[SPDIF]

No, all problems where fixed long time already, Samsung does not want to cut it's own flesh anymore. I have not see any SSD come back as before here. They are reliable again now just as it was before the pandemic.

 

[defjukie]

Been aware of this thread for the last year as I have a 2021 cira 870 EVO that I am using as a game drive (no primary copies of important data per se).

Got a CRC error trying to play a particular map of on UT2k4 (installed the game it when I got the drive in 2021 so maybe it was always corrupted as I don't remember playing that map yet) yesterday.

So I checked the drive's SMART info in Samsung Magician and indeed errors have appeared (it was all perfect/zeroes last I checked a few months ago, just like my old 850 EVO still is).

My drive otherwise seems to be working fine and I have no issues with other games. A Short Scan in Samsung Magician produced no errors either.

Should I just wait and see for it to actually fail (or not) and not bother with the hassle of a potential rejected RMA with Samsung at this stage*? Thank you.

*I would otherwise return it to the retailer but I have since moved a bit far from them, though still am in the same country, i.e. Aus.

 

[RJARRRPCGP]

Been aware of this thread for the last year as I have a 2021 cira 870 EVO that I am using as a game drive (no primary copies of important data per se).

Got a CRC error trying to play a particular map of on UT2k4 (installed the game it when I got the drive in 2021 so maybe it was always corrupted as I don't remember playing that map yet) yesterday.

So I checked the drive's SMART info in Samsung Magician and indeed errors have appeared (it was all perfect/zeroes last I checked a few months ago, just like my old 850 EVO still is).

View attachment 343108

My drive otherwise seems to be working fine and I have no issues with other games. A Short Scan in Samsung Magician produced no errors either.

Should I just wait and see for it to actually fail (or not) and not bother with the hassle of a potential rejected RMA with Samsung at this stage*? Thank you.

*I would otherwise return it to the retailer but I have since moved a bit far from them, though still am in the same country, i.e. Aus.

Yep, that drive's unquestionably doomed!

 

[SPDIF]

Yep, that drive is toasted... RMA it back to SAMSUNG ASAP and take a backup before you loose data. This drive has Long Covid in it's veins.

Healthy drive should look like this;

 

[Koozwad]

My drive otherwise seems to be working fine and I have no issues with other games. A Short Scan in Samsung Magician produced no errors either.

Should I just wait and see for it to actually fail (or not) and not bother with the hassle of a potential rejected RMA with Samsung at this stage*? Thank you.

*I would otherwise return it to the retailer but I have since moved a bit far from them, though still am in the same country, i.e. Aus.

I would try making an RMA request with the retailer first, as, depending on the retailer they can be a lot less difficult about it and send a new drive back much faster. I recently did the same with a 3+ year failed drive and received one pretty much instantly after they received it. Though your case may be different as the drive works fine otherwise. In my case the drive was mostly unusable(literally). Tell them your reason(s) first and see how they react.

 

[SPDIF]

In any way RMA it, better you get a new one, and don't have to think about it anymore.

 

[SilentCartographer]

In the last week or so I've had one of these go bad.

870 EVO 1TB
Manufactured 2021.06.11
Purchased 2021.09

Reallocations started rising. Uncorrectables then shot up, especially after a surface scan with magician.

The main array in that box has redundancy but I'd forgotten to setup a backup routine for this disk which hosts two VDISKS for a guest VM. Not heavily driven, low overall writes. FW update silenced the metrics for a few days then we want to SMART progressively getting worse. ddrescue was able to pull working copies from it to a newer EVO 1TB. Looks like there was some loss but so far it's not proven itself to be impacting. Might be some cache data on the 2nd vdisk that will turn out to be bad.

Now to talk to Samsung.