Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Update on Samsung SSD Reliability

pugetsystems.com

Post reply on HN
Print view View original

231–240 of 243 posts

Re: Update on Samsung SSD Reliability

#231

Post by gowld »

ThePowerOfFuet wrote: post #230
lazyweb wrote: post #211

Earlier quoted context omitted.

Similar experience for me. Four 2TB 980pro in RAID-Z2. Since introducing the drives in August 2021 I've had to replace them five times. I think none of the original SSDs are left. Only between 5 - 25TB written on average. Usually individual uncorrectable errors caught by ZFS, but one drive just straight up died. I keep a cold spare these days. On the bright side, if the cycle continues, I'll never run out of warranty…

> On the bright side, if the cycle continues, I'll never run out of warranty from Samsung or the vendor. The warranty is from the date of sale of the original unit. Replacing one doesn't reset the warranty end date.

If this really becomes an issue due to repeated failure, lemon law activates for a refund.

Re: Update on Samsung SSD Reliability

#232

Post by efreak »

renonce wrote: post #187
wahern wrote: post #138

Earlier quoted context omitted.

> as well as trying to overprice SLC out of production. Is it even possible to buy SLC drives any more? For the past 5+ years the only outlet I've been able to find that even advertise SLC is https://www.delkin.com/ , and you need to speak to sales to even get a price. I just assumed they and any other similar suppliers bought giant lots of chips at the tail end of SLC production and jack up the price on every new or…

Many TLC drives have SLC cache so if you limit usage of the drive you may effectively use it in SLC-only mode. This does not say anything about its endurance however. If you want endurance, better mirror your data to HDD.

Show me software that can mirror an SSD to HDD in realtime without the "backup" being affected by the slower write speeds. I was looking for a way to do this a few years ago, and couldn't find anything. I'd be very happy if I could pull my SSD out, change my boot drive, and boot into my existing OS without any issues. My understanding is that existing solutions for this delay write confirmation until both drives are complete, negating the speed advantage of the SSD.

Re: Update on Samsung SSD Reliability

#233

Post by efreak »

spyder wrote: post #144
adgjlsfhk1 wrote: post #34

Earlier quoted context omitted.

While it's true that MLC is mostly dead, you might want to consider a higher capacity TLC ssd. If you double your capacity, you double the endurance since SSD endurance is in drive writes per day, and a bigger SSD will likely have a bigger SLC cache to help with the write speed.

Yep, you can even adjust the overprovisioning manually (at least for Samsung). So if you need more endurance (and improved random write performance) just buy bigger capacity and increase the over provisioning allocation. Found this good summary about it with graphs showing the impact of it: https://www.atpinc.com/blog/over-provisioning-ssd-benefits-e...

As long as you're using a filesystem Samsung understands? When I overprovisioned my 860, windows showed that the partition didn't fill the drive. When I decreased the size of the partition in windows, Samsung Magician showed that the overprovisioning had increased. While I'm not sure how the drive itself deals with overprovisioning, Magician certainly detects it based on the size of the partition.

Re: Update on Samsung SSD Reliability

#234

Post by Marsymars »

rlpb wrote: post #36
mrtksn wrote: post #15

Earlier quoted context omitted.

This sounds like an SLA agreement, its very unlikely you'll get that for 100 bucks. Even if this manufacturer somehow perfected their process and have zero defects, they are still acquiring a 10 years liability for 100 dollars of revenue.

APC sell surge protectors with equipment protection insurance for less than $100. Apparently, it's possible even for products sold at $100.

Well they sure market this, but it seems pretty difficult to collect. For the insurance to apply you have to register within 10 days of purchase with a list of what will be plugged into the the surge protector. In case of damage, you have to pay to ship the surge protector to APC. If they determine it's been damage by power line transients, then you need to ship all damaged equipment somewhere to be evaluated. If it's determined that the equipment was damaged by power line transients, then APC chooses whether they'll pay for repairs or for current fair market value.

Re: Update on Samsung SSD Reliability

#235

Post by othersteve »

SV_BubbleTime wrote: post #221
deadbeeves wrote: post #166

Earlier quoted context omitted.

Yes, the non-raw values are always reported with higher meaning better. The raw values are the raw measurements/counts and they each mean different things.

Is the number the percentage of a max 100%?

You did the right thing by replacing the Seagate ST2000DM008. I do data recovery professionally and that's not one of my favorite drives! Lots of issues IMHO.

Re: Update on Samsung SSD Reliability

#236

Post by acabal »

zargon wrote: post #180
acabal wrote: post #32

Earlier quoted context omitted.

Indeed! Fortunately the RAID was also backed-up offsite. But, the entire process was shocking in several ways. At least Samsung was fairly speedy with the RMAs and it was basically no-questions-asked... because I imagine they're getting tons of these mailed back to them.

Did they require the failed drives to be returned? Did the secure erase function work on a failed drive? I guess this is a good reason for full disk encryption even on machines where I'm satisfied with their physical security.

Yes, they sent a prepaid label to return the bad drives. I did a secure erase both via Linux command line and via their proprietary software and it seemed to work. The whole process took about 3 weeks per drive.

Re: Update on Samsung SSD Reliability

#237

Post by deadbeeves »

SV_BubbleTime wrote: post #221
deadbeeves wrote: post #166

Earlier quoted context omitted.

Yes, the non-raw values are always reported with higher meaning better. The raw values are the raw measurements/counts and they each mean different things.

Is the number the percentage of a max 100%?

No, sometimes it's in the 0-200 range. I believe the device always reports the maximum and minimum possible values.

Re: Update on Samsung SSD Reliability

#238

Post by jwatt »

YoumuChan wrote: post #151
Aardwolf wrote: post #55

Earlier quoted context omitted.

I now found and followed this: https://blog.quindorian.org/2021/05/firmware-update-samsung-... And it seems to have worked. After extracting this updater tool and running it, smartctl kept showing the old firmware version (3B2QGXA7), but after reboot it now shows the new version (5B2QGXA7). I took the risk of running this while the OS (Archlinux) was running with the disk mounted (this is the OS install disk), and at…

The command can be simplified as isoinfo -R -i xxx.iso -x /initrd | gzip -dc | cpio -idv --no-absolute-filenames "root/fumagician*" if you don't want to go through the mounting and extracting everything.

For me `isoinfo` doesn't like the format of the .iso file ("CD-ROM is NOT in ISO 9660 format"). Using bsdtar worked though:

  bsdtar xOf xxx.iso initrd | gzip -dc | cpio -idv --no-absolute-filenames "root/fumagician*"
(And of course then running `root/fumagician/fumagician` in either case.)

Re: Update on Samsung SSD Reliability

#239

Post by renonce »

efreak wrote: post #232
renonce wrote: post #187

Earlier quoted context omitted.

Many TLC drives have SLC cache so if you limit usage of the drive you may effectively use it in SLC-only mode. This does not say anything about its endurance however. If you want endurance, better mirror your data to HDD.

Show me software that can mirror an SSD to HDD in realtime without the "backup" being affected by the slower write speeds. I was looking for a way to do this a few years ago, and couldn't find anything. I'd be very happy if I could pull my SSD out, change my boot drive, and boot into my existing OS without any issues. My understanding is that existing solutions for this delay write confirmation until both drives are…

Check out Syncthing. I personally run two instances of syncthing, synchronization happens very fast and as long as there isn't high write volumes it syncs pretty quickly. If you don't need realtime backup, rsync'ing in a loop would work.

Re: Update on Samsung SSD Reliability

#240

Post by AdamJacobMuller »

sebazzz wrote: post #217
mvdwoord wrote: post #197

Earlier quoted context omitted.

sounds ideal for a public terminal ;)

There used to be a PCI module (iirc) that would assist in that. I believe it was called a Bo(u)rne Again module. It was used in school computers etc.

There was an IDE dongle which did this. It was at a school I was visiting (but not my regular school) and it was pretty cool because it allowed them to be pretty liberal with permissions on the computers since you couldn't really do anything terrible as a reboot would fix everything and they powered them off every night.
Post reply on HN