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

[FISHTACO]

Anyone know if this batch of 870 Evo are effected with issues? (MZ-77E2T0B/AM) https://www.amazon.ca/gp/product/B08QB93S6R/ref=ppx_yo_dt_b_asin_title_o00_s00?ie=UTF8&psc=1

 

[Potomac]

It's like casino games, you may win, you may lose,
only Samsung knows the real reliability of current 870 EVO SSDs production (quality control process, flash memory modules, quality of the controller, of the firmware).

There is a risk, some amazon customers had problems :

Started to Fail Within Months
Reviewed in Canada on September 1, 2022
Capacity : 1TBVerified Purchase

A few months after purchase I noticed that some of my data became unreadable when doing my backup. Taking a look at the Magician utility revealed that the drive was encountering an awful lot of ECC errors. Since the return window was already over and apparently there's no proper warranty in Canada for the drive either (that's what "Canada Version" means) I decided to keep it around to see whether it's just some blocks that were damaged from the get go.

It seems like that's not the case as I recently noticed some old data that haven't been touched manually has started getting corrupted as well, so it seems like the drive's blocks are either just gradually breaking down or the firmware doesn't verify when it moves something around internally.

Such poor quality control and support for one of the most expensive consumer brands is just appalling. We'll see we how the RMA through Amazon goes once the replacement arrives, which is not going to be a Samsung one for sure.

You still have 5 years warranty, possibility of quick RMA if things go bad.

If you want 100% reliability : buy pro series from Samsung, but it's more expensive.
And I suggest no to store important data on 870 EVO SSD, just put the OS, programs, games, things you can easily reinstall if the SSD crashes,

And use a HDD for storing important data (better for data archiving than SSDs), RAID systems can be also a good strategy.

 

[chrcoluk]

Some thoughts, I wonder if anyone with failures has volume shadow copy enabled? The recent discovery in my thread that when thats enabled Microsoft might defrag the SSD automatically.

 

[kapone32]

It's like casino games, you may win, you may lose,
only Samsung knows the real reliability of current 870 EVO SSDs production (quality control process, flash memory modules, quality of the controller, of the firmware).

There is a risk, some amazon customers had problems :



You still have 5 years warranty, possibility of quick RMA if things go bad.

If you want 100% reliability : buy pro series from Samsung, but it's more expensive.
And I suggest no to store important data on 870 EVO SSD, just put the OS, programs, games, things you can easily reinstall if the SSD crashes,

And use a HDD for storing important data (better for data archiving than SSDs), RAID systems can be also a good strategy.

Or just buy a different budget drive from a "premium" brand.

Some thoughts, I wonder if anyone with failures has volume shadow copy enabled? The recent discovery in my thread that when thats enabled Microsoft might defrag the SSD automatically.

I can't believe that this thread is from January. Samsung seem to suffer from the same issue OCZ and Corsair 1st Gen drives suffered from or those thumb drives from Kingston. It is obvious that they are either using unstable NAND or the controller is poorly insulated. I would never buy one of these. Give me a 660P from Intel (much maligned) every day before I would get one of these drives.

 

[Daytrader]

Some thoughts, I wonder if anyone with failures has volume shadow copy enabled? The recent discovery in my thread that when thats enabled Microsoft might defrag the SSD automatically.

Well 99.9% of people will be using system restore in windows, so i doubt that has anything to do with it ?

 

[chrcoluk]

Well 99.9% of people will be using system restore in windows, so i doubt that has anything to do with it ?

It wouldnt be the root cause but might accelerate the failure state?

I think by default Win10 and Win11 have SR off now.

 

[Daytrader]

It wouldnt be the root cause but might accelerate the failure state?

I think by default Win10 and Win11 have SR off now.

Did not know that, my system been going with system restore on since 2015, when i built my PC, did not know it was disabled by default now, thx

 

[chrcoluk]

Did not know that, my system been going with system restore on since 2015, when i built my PC, did not know it was disabled by default now, thx

Well double check as I am not 100% sure, but I kind of remember having to turn it on manually on Win10. If it is on by default, it will only be for the C: drive, as it was only ever on by default for the OS drive only.

 

[Daytrader]

Well double check as I am not 100% sure, but I kind of remember having to turn it on manually on Win10. If it is on by default, it will only be for the C: drive, as it was only ever on by default for the OS drive only.

Yeh good point, my two 4TB ssd's were both not os drives, so would not matter to me anyway, and still failed..

 

[Mussels]

Some thoughts, I wonder if anyone with failures has volume shadow copy enabled? The recent discovery in my thread that when thats enabled Microsoft might defrag the SSD automatically.

And what thread was that?

While individual files may get defragged, thats not the same as the entire drive

system restore is on by default, but only on the C: Drive and it's not write heavy - it's almost the opposite (same as shadow copies)
What they do is an NTFS feature where a new file of the same name doesnt over-write the old one, it goes to free space instead, and the old ones are marked as a 'can be deleted, but use empty space first'

So if space is needed they're removed and used, but if not your shadow copy/SR is available from the last time it was written - it was made invisible, not actually deleted

 

[chrcoluk]

It may be a partial defrag but still a defrag, I did measure the writes during the defrag process, and posted it in the thread.
Read it carefully. I watched it carefully, it did multiple free space consolidation passes, and was about 20 minutes of fairly heavy writing. I posted the exact measured writes taken from SMART in the thread.

Windows keeps defragging one of my SSDs

This started since last windows update, screenshot attached, note it has correctly identified the SSD, so doesnt think its a spindle. All my other SSDs only trim, this one is doing an actual defrag. I keep stopping it but then it keeps starting again, unless I kill the defrag service. If I...

www.techpowerup.com

 

[Mussels]

It may be a partial defrag but still a defrag, I did measure the writes during the defrag process, and posted it in the thread.
Read it carefully. I watched it carefully, it did multiple free space consolidation passes, and was about 20 minutes of fairly heavy writing. I posted the exact measured writes taken from SMART in the thread.

Windows keeps defragging one of my SSDs

This started since last windows update, screenshot attached, note it has correctly identified the SSD, so doesnt think its a spindle. All my other SSDs only trim, this one is doing an actual defrag. I keep stopping it but then it keeps starting again, unless I kill the defrag service. If I...

www.techpowerup.com

replied in the thread

 

[mplayerMuPDF]

Or just buy a different budget drive from a "premium" brand.


I can't believe that this thread is from January. Samsung seem to suffer from the same issue OCZ and Corsair 1st Gen drives suffered from or those thumb drives from Kingston. It is obvious that they are either using unstable NAND or the controller is poorly insulated. I would never buy one of these. Give me a 660P from Intel (much maligned) every day before I would get one of these drives.

It is sad how far Samsung SSDs have fallen. The 860 Evo is such a good drive. I have three of them, two of which I bought myself, one 500GB 2.5" in the laptop I am typing this one, one 2.5" 250 GB I was gifted that is in my Llano ProBook and an M.2 250 GB that is in a damaged laptop that I have been trying to sell but I will probably salvage it from that computer and use it in a new system that I will commission later this month. That said, I will never buy a (post-870 Evo) Samsung SSD again after this disaster and kudos to the TPU community for exposing this.

I actually for some years thought that Kingston was awesome as as a kid I had one of their DataTraveler (USB 2.0) drives (containing the only copy of a big homework assignment) and it survived a journey in the washing machine, which saved my little ass. However, lately I have read some less than great stories about them.

Anyway, for SSDs I now prefer Crucial/Micron and Sandisk and I am not afraid to buy used ones (from non-sketchy sellers, of course). I don't need the latest and greatest and am happy to reduce the amount of e-waste. I would never consider all the exotic brands; I see no reason to gamble on them. SK Hynix is also blacklisted for me as they don't seem to deal well with power loss at all, which is a big no no for me.

 

[80251]

@mplayerMuPDF
Have you had any SK Hynix SSD's fail? I bought their 1 TiB P31 M.2 SSD which has held up fine so far (fingers crossed).

 

[mplayerMuPDF]

@mplayerMuPDF
Have you had any SK Hynix SSD's fail? I bought their 1 TiB P31 M.2 SSD which has held up fine so far (fingers crossed).

I have never owned an SK Hynix SSD but I was referring to https://www.tomshardware.com/news/sk-hynix-sabrent-rocket-ssds-data-loss

 

[DZMBA]

A few days ago I came across a comment on reddit mentioning this issue, & I have four 2TB Samsung 870 Evos, two purchased in Feb & two in June. So I looked....
At the time I hadn't notice anything off yet, but apparently the batch from Feb had been blowing up the error logs for several months now. It wasn't until I actually tried getting data off them did the "Can't read from source file or disk" & "Cyclic Redundancy Check" user facing errors start.

The first image is what I posted in that reddit thread just 5 days ago. Those error rates will jump up quite a bit as I try to get data off. The second shows my current situation after diving in. I offloaded nearly all the data I could from the storage pool, but I still don't have enough redundancy to handle a double failure occurring in my largest drives, that are Samsung no less! I replaced old Crucial M4s that I thought weren't as reliable but were nothing but reliable for 7-8yrs. Samsung.... I don't know what to say. And I have 2 more of them that are just 2 months away from 7 months since purchase!

Here is the current state.

Manufactured October & November of 2021

 

[Daytrader]

Your shipped feb 2022 drive, what date was made ?

 

[chewie198]

A few days ago I came across a comment on reddit mentioning this issue, & I have four 2TB Samsung 870 Evos, two purchased in Feb & two in June. So I looked....
At the time I hadn't notice anything off yet, but apparently the batch from Feb had been blowing up the error logs for several months now. It wasn't until I actually tried getting data off them did the "Can't read from source file or disk" & "Cyclic Redundancy Check" user facing errors start.

The first image is what I posted in that reddit thread just 5 days ago. Those error rates will jump up quite a bit as I try to get data off. The second shows my current situation after diving in. I offloaded nearly all the data I could from the storage pool, but I still don't have enough redundancy to handle a double failure occurring in my largest drives, that are Samsung no less! I replaced old Crucial M4s that I thought weren't as reliable but were nothing but reliable for 7-8yrs. Samsung.... I don't know what to say. And I have 2 more of them that are just 2 months away from 7 months since purchase!

Here is the current state.

I'm in the same boat, I have six 4TB drives running on a TrueNAS SCALE RAID-Z2 pool with triple redundancy, but the errors have become so prolific that I've nearly lost the pool a few times. It took me a while to discover that the problem was actually the drives too - there were so many failures that I assumed something spanning all of the drives like the software, controller, cabling, or power delivery must have failed and so I wasted some time troubleshooting the wrong issues. Fortunately simply re-reading from the drive usually seems to allow it to quickly re-silver with no lost data, and this is for a small business with multiple pool backups, but it's pretty frustrating to end up with potential downtime here because so many of the drives are failing and it happened all at once across nearly all the drives. Out of six drives, two were manufactured mid-2021 and two were from January of this year, and at least one of the drives from this year is also failing. As soon as I can pull the drives out of the server I'll check which firmware they're running. I suspect that the faulty SVT01B6Q firmware was still being used in January at least, which leaves me with potentially 6 RMAs to deal with while hoping the remainder doesn't just self-destruct while waiting for the replacements. I have to say, this is potentially one of the worst failure modes I've seen on any storage drive, not just Samsung, because it has the potential to render multiple layers of server redundancy moot when you have entire clusters of drives failing simultaneously.

 

[shovenose]

Yeah the fact that it starts with silent corruption vs just failure is concerning. At least with ZFS you should be safe.

 

[Daytrader]

Hope someone has passed on this thread to Samsung ?

 

[80251]

@Daytrader
The whole idea that Samsung isn't aware of the failures that are happening across their entire 870 EVO series doesn't seem believable. There are plenty of reviews of 870 EVO's on newegg and amazon that mention these catastrophic failures.

It would be nice if Samsung would:
a. own up to the problems with their 870 EVO series
b. stop manufacturing said SSD's and/or recall the entire series

@chewie198
Those 4 TiB 870 EVO's aren't cheap either. What a clusterf*%k.

 

[Daytrader]

@Daytrader
The whole idea that Samsung isn't aware of the failures that are happening across their entire 870 EVO series doesn't seem believable. There are plenty of reviews of 870 EVO's on newegg and amazon that mention these catastrophic failures.

It would be nice if Samsung would:
a. own up to the problems with their 870 EVO series
b. stop manufacturing said SSD's and/or recall the entire series

@chewie198
Those 4 TiB 870 EVO's aren't cheap either. What a clusterf*%k.

yup i paid almost £700 for my two

 

[mplayerMuPDF]

@Daytrader
The whole idea that Samsung isn't aware of the failures that are happening across their entire 870 EVO series doesn't seem believable. There are plenty of reviews of 870 EVO's on newegg and amazon that mention these catastrophic failures.

It would be nice if Samsung would:
a. own up to the problems with their 870 EVO series
b. stop manufacturing said SSD's and/or recall the entire series

@chewie198
Those 4 TiB 870 EVO's aren't cheap either. What a clusterf*%k.

I mean, Samsung overall is quite a shady company so with that in mind it is perhaps not too surprising. Best thing we consumers can do is spread the word and recommend buying from solid brands that use solid components (and for me personally that also means no Phison controller, at the very least for SATA drives). There are so many good options out there today: Crucial/Micron, Intel, Sandisk, WD (personally not a great fan of them in general) and HP. I consider ADATA to be questionable and I don't know why you would go with brands like Corsair, Mushkin, PNY, Sabrent, Kingston etc. when you have the options above but if you want to gamble or are a megafan of one those brands, be my guest.

 

[DZMBA]

Your shipped feb 2022 drive, what date was made ?

I finally got the drives pulled after sacrificing all movies, games, & downloads (had to delete the volume stuck on "not enough redundancy remaining to repair the virtual disk" to successfully complete the "Preparing for Removal").

(I've already registered the drives so not too sure it matters if I show the serials)

• S6PNNJ0RB17997K was manufactured 2021.11 (November 2021). This is the really bad one.
• S620NJ0RA04653P was manufactured 2021.10 (October 2021)

I'm currently running tests on just the 2021.10. SAMSUNG Magician isn't able to recognize the drives through the LSI SAS controller. Which is a huge pain, my PC is stuffed in a hidden area wedged behind the desk and wall, so digging into it requires moving everything & re-organizing cables.
Well, while moving the drives to the motherboard controller, the SATA cable to the 2021.11 must not have got plugged in all the way. I was shoving cables blind not wanting to also remove the GPU or HDD cage. So it looks like I'm only testing one at a time.

As of now, I've tested the 2021.10 (the better one), & no issues have been found!?? It's passed every test SAMSUNG Magician has to offer. I've even avoided updating the firmware bcus I wanted to see if doing it would change the behavior. The Reallocated Sector Count did not increase, but Uncorrectable Errors did from 71 to 166. I've since filled the drive with my backup data, and plan to later test if I can read it back & compare it bit-by-bit without issue.

UPDATE: That one SATA cable is just bad I guess and they're seemingly both passing. I've only done the short tests for the 2021.11 one. Got a few more hours left on the full scan but I expected it to instantly blow up.

 

[Daytrader]

On my bad drive, i could not copy alot of the data to another drive, so i knew they were bad, i had alot of uncorrectable error count, like 10698, rma accepted straight away.