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

[Chrispy_]

Oh shit, 970;

i r dum:

At least I can sleep soundly tonight.

 

[Pwnstix]

Just got my replacement in. Made in Korea, fab date of 2022 03 03. S/N: S6BBNS0T10****

Started the whole process in the 3rd week of April (~19th). TTS received it on the 21st and Samsung support got back with me early the next week, shipped my replacement (according to UPS tracking info) on April 27. The whole process took around two weeks.

Hopefully this one holds up. I'm staying subscribed to this thread and will check back now and then. Until later, good luck to everyone!

edit: just got it installed. I wonder if it's a refurb, because I see a Power-on Count, POR Recovery Count, and some Total LBAs Written. Also not showing any new firmware available through Magician.

 

[Blaeza]

I've got 2 870 evos but they are only 1tb. Am I safe?

 

[ThrashZone]

I've got 2 870 evos but they are only 1tb. Am I safe?

Hi,
Read the op and run the tests, but yes some 1tb have also been effected.

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

Certain 870 EVO 4TB and 2TB drives are affected by early failures where they develop uncorrectable errors and some data just cannot be read from them anymore. This seems to primarily affect drives produced in January/February 2021. For example, i have three 870 EVO 4TB, only one is affected (so...

www.techpowerup.com

 

[Blaeza]

Looks like I'm all good. Just did the checks and seems my drives are all in order. Noice.

 

[chrcoluk]

My laptop has a nasty data corruption problem but there is no SMART errors, and the LBA test passes (albeit slower than it should). So potentially my drive is in the process of failing. But not sure.

At the very least it looks like I am going to have to recover an older backup, the registry managing VSS is corrupt which is a real nasty problem.

-- Restored C: drive from 5 day old backup and issues have gone so was corrupted registry, will keep doing backups and keep an eye on things.

 

[unspecified_error]

Howdy, folks. I just found my own 870 EVO 1TB to have this problem. Many thanks to the OP for bringing it to light. I was just checking another failed SSD (an 850 EVO) here on my PC using Samsung Magician when I noted the red 'Critical' SMART warnings on the 870. A quick search on the subject brought me to this thread. I started backing the data up immediately but ran into problems then as corrupted files ultimately halted the usual Windows backup process. As the backup process progressed though I could see the drive health % plummet in CrystalDiskInfo and watched the Reallocated Sector Count climb as it failed to read numerous sectors. In the end I just did a manual folder-by-folder, file-by-file copy skipping the unreadable files to get the most important stuff off. I don't think I've lost too much that isn't replaceable thankfully but there was quite a few unreadable files in there - 10's or 100's I'd say, though maybe only equating to a low percentage of the overall stored data (around 650 GB). It'll be going back to the supplier for RMA on Monday.

 

[unspecified_error]

Just an update to add more details. Of interest, prior to RMA'ing the drive I wiped it using Minitool's wipe facility (randomised 0/1 write). After the wipe completed I ran full scan and short/extended SMART tests using Samsung Magician once more with the result being that this time they passed/showed clear, with SM stating the drive to be in good condition. I gathered this to be because the to-date 369 failed sectors spanning the entirety of the drive had now been fully recognised and reallocated, but I'm not entirely sure, though the raw SMART info from SM and Crystal Disk Info did still indicate the Reallocated Sector Count, Uncorrectable Error Count, and ECC Error Rate to be high ("Critical" in SM), so thankfully it did still show the faulting issue (to help make the RMA process easier). I wouldn't like to trust it to hold data from this point though as I guess another ~369 sectors could go tits-up in the next ~8 months use again. I don't know how many reallocatable sectors the drive has in total, but I assumed around 1024 judging by the numbers I saw, particularly as the CDI health percentage dropped as the reallocated sector count increased during data recovery. Anyway, the RMA went painlessly and I got it replaced with a 2022-01 model which hopefully won't have this issue. I'll be keeping a closer eye on things from now on though to be sure. Regards.

 

[Captain Danger]

Is shipping back the drive to Samsung a must/requirement for RMA or did anyone get replacement without that?

Do they create a shipping label for it, and to which country exactly?

My 1 TB 870 EVO is still doing fine but I am concerned about it in the longer run as manufacturing date suggests it is from bad batch. My country doesn't allow any computer items/electronics to be sent outside the country for RMA through DHL/FedEx etc - they have banned any such export. So I wanted to know if it will be possible for me to get a replacement without sending them the drive if need be.

 

[unspecified_error]

Is shipping back the drive to Samsung a must/requirement for RMA or did anyone get replacement without that?

I just returned mine to the retail store I bought it from. They have all the details on record of when it was purchased. Presumably they'll fire it back to their suppliers, and ultimately Samsung or some official Samsung representative along the line will take responsibility.

 

[xorbe]

I mirrored my drive, upgraded the firmware, overwrote the whole drive with random data, then restored the mirror, and trimmed the drive. Will just use it as my Steam library drive I guess. It only had 1 failed LBA beforehand.

 

[Zeromis]

Just had two of these 2TB drives show the same symptoms. One was installed in the summer last year and the other I got in Autumn. That said they both have a manufacturing date in April. Luckily they both didn't have much important on, was just media files and games for the most part, and anything I didn't want to lose was backed up on a external anyway. Still quite annoying though and something I really wouldn't expect.

I just happened to be doing a routine check up of my drives and noticed one of my drives had around 5 uncorrectable errors, the other one showed 0 at that point. Was a bit worried but left it a bit to see if that rose, and it did by a few. Following that I did checks via the Samsung software (during this time I installed the new firmware too) and also chkdsk and the number sky rocketed on both drives. Also found some random files which were no longer readable. Fortunately not too many corrupted files, was just a few games easily fixed by a integrity check, think I managed to at least catch this reasonably early. Since then I've copied all my files off the drives and onto a new MX500 4TB SSD while I sort the situation out. Did see that the MX500 was having issues too, but think that must have been fixed in recent models as I bought one late 2020 with a newer firmware and it's been perfect since then - which lead me to buy this 4TB(that and a really good discount lol).

Interestingly the one I installed last summer actually had less written to it, only around 2TB, and yet the issue had progressed further on that one. With around 6 more reallocated sectors totalling just under 40 and around 5000 in uncorrectable errors. In comparison the Autumn one had around 5TB written to and it had around 1500 uncorrectable errors.

Honestly it's a good job I checked really, this seems like something that's really going to sneak up on people. You might not notice it until it's too late. It's kinda put me off buying any Samsung SSDs in the future really. Shame really as I used to trust them having had a number of their drives with no problems and good performance.

 

[chrcoluk]

Its a nasty situation, reviewers cant realistically test what happens when a device is been used as a daily driver for several months, even those endurance tests done some years back were not realistic as they only emulated erase cycles but not time or power off data retention either.

I think one way to push these manufacturers in to been more honest is if media sites were prepared to expand on educating to people to the problems by publishing articles, e.g. I did offer my MX500 to Wizzard for testing but he wasnt interested, would I blacklist either Crucial or Samsung from future purchases though? Probably not as then I would eventually run out of vendors products to buy. Every now and then people make bad products. The problem is though because of the "pretending it doesnt exist" attitude us consumers have no idea if this problem still exists on drives been manufactured today, and if it has been fixed, at which point was it fixed? We are in the dark. The MX500 situation we know at least the newer revisions are fixed, but because there was no model number change you still run the risk of buying a first revision when ordering one. Its these issues that create the distrust in the companies.

When I next buy an SSD I may consider Hynix as I think what both these companies have done is wrong.

 

[Lake_Cities_6]

November 2021, I purchased a 1TB 870 EVO and a 2TB 870 QVO. Both made in Korea. QVO date is 9-11-21, EVO date is 10-24-21. Both drives have been in low write/heavy read situations. QVO is fine, EVO just started throwing red blocks in Magician and critical alerts in SMART, which led me to discovering this thread. The problematic 870 EVO was only loaded (once) with 311 GB. As of today, it has endured a total of 320 GB written. 0.3 TB written. In 7 mos I've only re-written 9 gb. Creampuff assignment.

Whatever the problem is, it is hardly impacted or influenced by TBW.

As I was watching the full scan run, the first pass revealed only 2 red blocks, but on other blocks the scan often stalled and churned for awhile, before turning blue with the scan continuing on. With each additional scan pass, more red blocks began appearing in those spots where the previous scan took an extra long pause. Blocks were going from blue (good) to red (bad) right before my eyes, from one pass to the next.

I have one additional unopened 1TB 870 EVO that is within 2 days of its return window. Made in Thailand and a date of 2-7-22. I am returning it to the store tomorrow.

 

[imeem1]

Today I noticed my whole system was hanging and unresponsive like I never seen b4. IDK how to read SMART data but this doesn't seem good. Samsun Magician started showing failing LBA and the software itself became unresponsive as well even after closing and opening it again. I bought my SSD is March 2021. Made in January 17th 2021. Data written is 4.796 TB.

What do I need to prove to Samsung in order to RMA my SSD?

 

[Captain Danger]

What do I need to prove to Samsung in order to RMA my SSD?

Showing them the Reallocated Sector Count > 0 should be good enough as someone stated earlier in the thread in Reply #85.

 

[CiTay]

IDK how to read SMART data but this doesn't seem good.

Like i said in the original post, you have all the telltale signs of being fully affected.

1) Elevated "Reallocated Sector Count", "Used Reserve Block" and "Runtime Bad Block" count
2) Non-zero "Uncorrectable Error Count" and "ECC Error Rate", and especially if those two keep rising when you read/write files
(they usually rise a lot when you try to copy all data off it in preparation for RMA)

When you do the RMA, Samsung will tell you exactly what they need. Keep us posted about the progress in here if you can, i'd appreciate it.
Normally there should be no problems at all getting an RMA with your drive's symptoms.

 

[Zeromis]

Today I noticed my whole system was hanging and unresponsive like I never seen b4. IDK how to read SMART data but this doesn't seem good. Samsun Magician started showing failing LBA and the software itself became unresponsive as well even after closing and opening it again. I bought my SSD is March 2021. Made in January 17th 2021. Data written is 4.796 TB.

What do I need to prove to Samsung in order to RMA my SSD?

View attachment 251055


View attachment 251156

I sent them a screenshot of the SMART page in Samsung Magician and a screenshot of the diagnostic scan in Samsung Magician showing up a lot of LBA errors in my opening email. After that they just asked me to send a bunch of photos of the physical drive and a receipt to start the RMA process.

Unfortunately I could only get the receipt for one of the two drives since the other one was a unused disk from work that got given to me. Consequently that one I'm just using to store steam games from now on which I can easily re-download if the issue shows again - so far I've not seen any further issues with it though. The reallocated blocks/errors are unchanged. I assume the firmware probably did fix the underlying problem but it can't really revert the damage already done, ie the lost blocks.

 

[imeem1]

Like i said in the original post, you have all the telltale signs of being fully affected.

1) Elevated "Reallocated Sector Count", "Used Reserve Block" and "Runtime Bad Block" count
2) Non-zero "Uncorrectable Error Count" and "ECC Error Rate", and especially if those two keep rising when you read/write files
(they usually rise a lot when you try to copy all data off it in preparation for RMA)

When you do the RMA, Samsung will tell you exactly what they need. Keep us posted about the progress in here if you can, i'd appreciate it.
Normally there should be no problems at all getting an RMA with your drive's symptoms.

o drives since the other one was a unused disk from work that got given to me. Consequently that one I'm just using to store steam games from now on which I can easily re-download if the issue shows again - so far I've not seen any further issues with it though. The reallocated blocks/errors are unchanged. I assume t

so I submitted a RMA request on Samsung Canada website and I mentioned my symptoms from above + the SMART results (no screenshots needed). They gave me a ticket number and someone was suppose to contact me. I got impatient so I called 1-2 days after and I quoted my ticket number. Guy told me to send picture of receipt + the SSD. Then he provided me a letter that basically says Samsung doesn't deal with warranties directly and told me to go to my store of purchase and show them the letter.

at the store, they run some test and they gave me a brand new drive. manufacture date is July 29, 2021. I asked him if there was a 2022 one, but this was the last one.

 

[Zeromis]

so I submitted a RMA request on Samsung Canada website and I mentioned my symptoms from above + the SMART results (no screenshots needed). They gave me a ticket number and someone was suppose to contact me. I got impatient so I called 1-2 days after and I quoted my ticket number. Guy told me to send picture of receipt + the SSD. Then he provided me a letter that basically says Samsung doesn't deal with warranties directly and told me to go to my store of purchase and show them the letter.

at the store, they run some test and they gave me a brand new drive. manufacture date is July 29, 2021. I asked him if there was a 2022 one, but this was the last one.

Admittedly in the case of the EU Samsung SSD support I had to contact a specific email address for warranty support since Samsung themselves don't seem to handle it. That said it wasn't too difficult to get it, just a quick contact of their online chat. Not sure why they don't just put it up on their website though.

 

[imeem1]

is there a consensus that we should update firmware from SVT01B6Q to SVT02B6Q?

 

[CiTay]

Yes, definitely update your firmware.

 

[chrcoluk]

is there a consensus that we should update firmware from SVT01B6Q to SVT02B6Q?

Logic says yes, but I tend to refuse when there is no changelog. So i havent lol. Bad logic I know.

My second reason is I suspect they may accelerate the erase cycles on the new firmware to get round whatever problems are occurring although people are still reporting broken drives on the new firmware, so dont know.

 

[CiTay]

They're not stupid at Samsung. They will have knowledge of this fault (of course they don't publically admit it, which history will show if it's stupid or clever). So perhaps the new firmware does something to prevent it in some cases. I much rather have the drive wear out a bit faster than to have it develop bad sectors which can't be read at all anymore.

 

[Zeromis]

Logic says yes, but I tend to refuse when there is no changelog. So i havent lol. Bad logic I know.

My second reason is I suspect they may accelerate the erase cycles on the new firmware to get round whatever problems are occurring although people are still reporting broken drives on the new firmware, so dont know.

Well the firmware can't fix the damage already done, so if they updated the firmware after the problem already manifested the drive would still be dead even if the firmware did manage to fix the underlying cause.

So far I've not had any more issues showing since installing the firmware and identifying all the previous errors/bad sectors. It's not been too long though, so guess we'll see long term.