I think this is probably a good idea. I will try running the tests that have been recommended to hopefully figure out what's going on, but unless we find something that can be fixed it seems like replacing the SSD would provide the best stability rather than trying to force this one to work.
File system keeps corrupting (need help)
- Thread starter Jac857
- Start date
-
- Tags
- file system ssd
Thank you, I have attached the file below and I will use the article you provided to hopefully figure out what any of this means.Overall test passed but a ton of info is omitted from the report, smartctl command you need is:
It will create smart.txt file in your home dir, attach it here, it's more useful than screenshot.Bash:sudo smartctl -Ax /dev/nvme0n1 > ~/smart.txt
Here is an article from my bookmarks about how to interpret smart report:
Thank you for the suggestions
You are correct in suggesting that the file check repairing command is not needed if the fsck command doesn't return an error.
On the kernel panic, the message shows that the kernel can't find the root filesystem. In that case the first thing to try is to tell the kernel where the root filesystem is, for example, add the kernel option:
root=/dev/<root>
where "root" is the root device shown in the
lsblk output on your machine. In post #10 above, one can identify the root partition by the symbol "/" in the MOUNTPOINTS column, so for that machine the device name would be: /dev/nvme0n1p3. You have to get it correct from your machine.If unsure how to add a kernel option, perhaps check here: https://askubuntu.com/questions/19486/how-do-i-add-a-kernel-boot-parameter
If that works, then the machine should boot and then you can mount any other partitions you need to run. If it doesn't work, there's some other issue which will need to be attended to.
Last edited:
I added the kernel option pointing to my root drive and attempted to boot with the setting, and ended up in BusyBox with file system errors. I ran a manual fsck, which repaired corruption in what seems to be my Firefox cache, after which I was able to exit and end up on my desktop. Do you think this is something I will be able to mostly solve with my current hardware, or would I have a more stable system by getting a different SSD? I appreciate the help.
Edit: I edited the GRUB to make the kernel option permanent and used sudo update-grub to apply the changes
Last edited:
CaffeineAddict
Well-Known Member
Not sure what's happening but the output is same as with previous command, on my system it gives much more info with same version of smarctl.
until you verify that the ssd is physically ok you can run into the same issues over and over. that is why I said use S.M.A.R.T. on the drive and verify that it is in good working order. otherwise you will just go in circles.
Aside from the other answers, I'm gonna trhow in my 2 cents - just FYI, so you know this for the future: there are cases where bad blocks won't appear in diagnostic tools + the S.M.A.R.T. function doesn't always work (or properly) with SSDs. It was originally intended for hard discs (you know - spinnin' discs, needles, weighs a ton). Sometimes this SMART function glitches out when used on SSDs: it reads one of my SSDs has 48 bad sectors and yet that SSD behaves perfectly well + a few Spyware tools (Spyware 10) said it was in perfect health.
So far the best test for SSDs I have personally found (in Linux) is to copy a large file (50-60 GB or more) from another storage to the device you want tested using rsync. If the write speed you see is twice lower than usual, that's a clear sign the target device is declining and/or is deffective in so many ways that won't appear in diagnostic tools. But if the writing speed is the usual, then that means the I/O is OK.
I do that test with 2 bash scripts and it takes seconds but IDK if you're this advanced to use such scripts.
you can have bad sectors and everything works fine. Does not mean the SMART was incorrect, it means the drive is compensating for the issue. It does mean the drive is going down. Yes SMART has issues at times with ssd but it is reading information from the chipset of the drive. Much like when the mechanic plugs his computer into your car, he can see much of what is going on and has gone on. ignoring errors on SMART is like seeing the check engine light on your car but saying it is fine because it still runs. I would love an update to SMART that does better with ssd and nvme but don't ignore what it says just because you personally do not see an issue. The drives are designed to compensate for issues rather than just outright fail.
Last edited:
I had a look at your smartctl output and the second line looks bad to me:
Media and Data Integrity Errors: 1Error Information Log Entries: 10,013Doing the same on my machine here (1600 power-on-hours, i.e. more than yours), shows no errors. Just to be sure, you can repeat the command by explicitly specifying NVME explicitly
smartctl -AX -d nvme /dev/nvme123 but I don't think it will show something else. More important, have a look at the error log itself:nvme error-log /dev/nvme123 It will be a long list of the same error. When the drive can recover an error, it is stated (my log also shows a long list of recovered problems). Perhaps the log gives an idea. smartctl can read the error log as well, but I don't know the command off-head.
For NVME typical problems can be temperature related. I did not catch if that was already mentioned. The drives are smart, but sustained high temperature are a problem for them. Switching its slot, heatsink, making sure some air passes over it (on a desktop) may help.
Similar threads
