ssd failure

hevidevi · 2021-10-10 13:12:11

Had a ssd die on me today, a crucial 240GB.

Not sure why, it died when i was downloading an iso via uget-gtk downloader, i had put the connections to 7 for simultaneous connections and was getting some really fast speeds like 200 MB per second which may be too fast, so all i can think of is that my ssd could not handle the writes or piping or something, maybe it was just on it way out though, only a year old though. Got the input/output error as soon as the uget-gtk program crashed.

The disk completely disappeared, just does not show up in the bios so im pretty certain its kaput.

I had backups so all good.

Anyone ever had this happen?

Last edited by hevidevi (2021-10-10 13:14:14)

durham · 2021-10-10 13:59:10

I've heard tales of exactly as it happened to you: sudden complete failure without warning. As I am running on a SSD, this also worries me.

Did you run with discards on or, if using LVM run fstrim every once in a while (say, once a week)?

hevidevi · 2021-10-10 14:15:27

durham wrote:

I've heard tales of exactly as it happened to you: sudden complete failure without warning. As I am running on a SSD, this also worries me.
Did you run with discards on or, if using LVM run fstrim every once in a while (say, once a week)?

No i did not use discards on, only noatime for kernel parameters. Not using lvm either, just plain old ext4 encrypted root.

I was of the opinion that the wiki was right in saying this in regards to the discard option.

https://wiki.debian.org/SSDOptimization

The "discard" options is not needed if your SSD has enough overprovisioning (spare space) or you leave (unpartitioned) free space on the SSD.

steve_v · 2021-10-11 03:45:43

hevidevi wrote:

Anyone ever had this happen?

Yup, and it was a Crucial as well. Overheated and died under a sustained write workload despite being in a fan-cooled drive cage that didn't exceed 25C ambient.

hevidevi · 2021-10-11 10:58:52

steve_v wrote:

hevidevi wrote:
Anyone ever had this happen?
Yup, and it was a Crucial as well. Overheated and died under a sustained write workload despite being in a fan-cooled drive cage that didn't exceed 25C ambient.

Well that sucks.

I will have to keep an eye on the ssd temperature on the next one i buy. Im currently using an old spin drive that is about 8 year old so that will probably pack it in soon too. But atleast i got 8 years out of it not 1.

steve_v · 2021-10-12 04:46:10

hevidevi wrote:

I will have to keep an eye on the ssd temperature on the next one i buy.

To be fair, it was one of the budget BX series. They're cheap for a reason and the specs are mostly lies.
I already had a good rant about them here, so I won't repeat it.

hevidevi · 2021-10-12 09:19:07

steve_v wrote:

hevidevi wrote:
I will have to keep an eye on the ssd temperature on the next one i buy.
To be fair, it was one of the budget BX series. They're cheap for a reason and the specs are mostly lies.
I already had a good rant about them here, so I won't repeat it.

Yeah mine was the same series as well. Might pay to spend a bit more next time, any recommendations? Im thinking possibly WD or maybe Samsung.

Ive got a gigabyte ssd in my laptop but im pretty certain that is about the same quality of crucial as it was just as cheap, i think i paid around $50 for both drives.

steve_v · 2021-10-12 10:14:35

hevidevi wrote:

any recommendations? Im thinking possibly WD or maybe Samsung.

I'm running 5 samsungs, 4 kingstons, 2 intels, an ancient OCZ, and a couple of random (micron OEMs IIRC) drives in various applications ATM, and I haven't had any problems with any of them... Besides a couple of the kingstons griping about being out of write endurance, but then they've been doing that for about 6 months now and they're still working fine.
That's a very small sample size of course, but the samsungs and kingstons regularly get the same kind of thrashing that killed those crucial drives, and they don't even blink.

I don't know anything about WD SSDs since I've never owned one, and as far as Gigabyte goes... TBH I'm pretty biased there. I've considered anything Gigabyte to be pure refuse since about 2004, and I don't know if quality has improved in the intervening years because I haven't bought from them since.

If I had to pick, I'd probably say Samsung or Intel of you want performance, and Kingston if you want dirt cheap without being completely horrible. But I'd also say take with a large grain of salt, because I don't have any real data to back it up.

James1138 · 2021-10-12 19:12:02

I found several things that may help in future. Setting automatic TRIM job to daily https://easylinuxtipsproject.blogspot.c … d.html#ID6 and setting the browser (Firefox or Chrome) to use memory cache more and write cache less.

James
Indiana

hevidevi · 2021-10-13 12:38:28

steve_v wrote:

hevidevi wrote:
any recommendations? Im thinking possibly WD or maybe Samsung.
I'm running 5 samsungs, 4 kingstons, 2 intels, an ancient OCZ, and a couple of random (micron OEMs IIRC) drives in various applications ATM, and I haven't had any problems with any of them... Besides a couple of the kingstons griping about being out of write endurance, but then they've been doing that for about 6 months now and they're still working fine.
That's a very small sample size of course, but the samsungs and kingstons regularly get the same kind of thrashing that killed those crucial drives, and they don't even blink.
I don't know anything about WD SSDs since I've never owned one, and as far as Gigabyte goes... TBH I'm pretty biased there. I've considered anything Gigabyte to be pure refuse since about 2004, and I don't know if quality has improved in the intervening years because I haven't bought from them since.
If I had to pick, I'd probably say Samsung or Intel of you want performance, and Kingston if you want dirt cheap without being completely horrible. But I'd also say take with a large grain of salt, because I don't have any real data to back it up.

Thanks for the detailed reply. I completely understand what you mean by grain of salt, everyone's experience can differ. I will most likely get a samsung, see how that goes and get a backup drive in a kingston.

hevidevi · 2021-10-13 12:46:58

James1138 wrote:

I found several things that may help in future. Setting automatic TRIM job to daily https://easylinuxtipsproject.blogspot.c … d.html#ID6 and setting the browser (Firefox or Chrome) to use memory cache more and write cache less.
James
Indiana

Good ideas James, thanks.

golinux · 2021-10-13 13:53:49

Spinning rust ftw . . .

Cheerful Charlie · 2022-02-25 20:21:36

I was running Mageia on a hard disk. That hard disk crashed. I had downloaded the new Chimaera 4.0. Loaded that on another hard disk. That crashed. I am running in crippled mode. A trip to the hospital put all on hold. But I have a new Crucial 1 TB SSD coming. If it stores data, it eventually dies. I have a monstrous big case with 10 5 1/4" bays and I use carts so, I will be transitioning away from hard disks. Get used to SSDs dying suddenly. That is the way of the SSD tribe. Running OSs from a SSD is hard on SSDs. Since a 500 GB Crucial SSD with a 5 year warranty is only $60, I will do that. I will put /home et al on the 1 TB, and set up another 500 TB SSD for timeshift and saving daily driver VM images. All SSDs are ephemeral and you have to expect that and work with the problem. When your OS boot SSD eventually starts dying, replace it with a new cheap but quality SSD and copy backups over, update GRUB and continue on.

In the bad old days of 80 GB hard drives, I used to have problems from time to time, but usually got a lot more out of a hard disk with fsck -c- -c -y, a deep bad block check that did wonders. But that was an all night affair. It will take 2 weeks or more to do a 2 TB hard disk that way.

Life is hard and you better have good backups.

Last edited by Cheerful Charlie (2022-02-25 20:41:32)

Head_on_a_Stick · 2022-02-26 09:56:52

James1138 wrote:

Setting automatic TRIM job to daily https://easylinuxtipsproject.blogspot.c … d.html#ID6

Daily is probably excessive and anyway that guide is for systemd. If you are using systemd then the util-linux package already supplies a (weekly) fstrim.timer unit file that can be enabled without having to create your own version.

For my Alpine system I have this in /etc/periodic/weekly/fstrim:

#!/bin/sh
/usr/bin/fstrim --listed-in /etc/fstab:/proc/self/mountinfo --verbose --quiet-unsupported

EDIT: for Devuan I think the appropriate location would be /etc/cron.weekly/fstrim. Not sure though.

Last edited by Head_on_a_Stick (2022-02-26 10:02:55)

Camtaf · 2022-02-26 10:24:05

Hmm, I've just swapped over most of mine to SSDs, I thought they had improve performance these days, compared to when first introduced, but I don't have anything really important, & I do keep recent backups, so I'll likely be OK.

I still have some HDDs, & these are what I backup too, (usually have 2 copies, different disks), I'd hate to have to replace all my music & downloaded stuff.

The officially official Devuan Forum!

#1 2021-10-10 13:12:11

ssd failure

#2 2021-10-10 13:59:10

Re: ssd failure

#3 2021-10-10 14:15:27

Re: ssd failure

#4 2021-10-11 03:45:43

Re: ssd failure

#5 2021-10-11 10:58:52

Re: ssd failure

#6 2021-10-12 04:46:10

Re: ssd failure

#7 2021-10-12 09:19:07

Re: ssd failure

#8 2021-10-12 10:14:35

Re: ssd failure

#9 2021-10-12 19:12:02

Re: ssd failure

#10 2021-10-13 12:38:28

Re: ssd failure

#11 2021-10-13 12:46:58

Re: ssd failure

#12 2021-10-13 13:53:49

Re: ssd failure

#13 2022-02-25 20:21:36

Re: ssd failure

#14 2022-02-26 09:56:52

Re: ssd failure

#15 2022-02-26 10:24:05

Re: ssd failure

Board footer