[SOLVED]Is my SSD repairable ?

[kajosim]

Running Mageia 9 Plasma on my main stationary PC which is 7 years old.
As usual I only boot up once a day and let i stand idle when I am not using it (it is not sleeping or hibernating)
I have several TABs in my Dolphin windows and use Plasma "start up session from last session" so my Dolphins have the same TABs as last time.

When I returned to the PC this afternoon my Dolphins had lost all previous TABs.
I tried to create new TABs and rebooted , again all TAbs were missing.

Immediately after login the CPU was running 100% on 1 core and pressing ESCAPE had no effect.
Wondering what was going on, I ran dmesg .

I tried to reset Plasma "start up session from last session" and rebooted .
I ran dmesg once more giving the same result. ie e2fsck is run on sda6 which is my root partion which includes /home
the sda is a Kingston SSD (7 years)
the sda1 (UEFI Bios) should I "manually" do a fsck , but do not know how ? use a Live Mageia on DVD ?

All my "permanent" documents/mail/files are stored on a partition on a HDD . The partition and most used directories are sym-linked to my /home .
Later I got a second HDD for more space.
So all my data is safe without sda

So even when e2fsck is run on sda6 then this will be repeated next time PC boots.

But otherwise it boots without problem and what I have tested is working apart from Dolphin loosing its TAB configuration.

excerpts from dmesg :

Code: Select all

[    3.784049] EXT4-fs (sda6): unmounting filesystem 73928837-b11e-4758-812a-c7f74f280412.
[    3.810498] dracut: Checking ext4: /dev/disk/by-uuid/73928837-b11e-4758-812a-c7f74f280412
[    3.810937] dracut: issuing e2fsck -a /dev/disk/by-uuid/73928837-b11e-4758-812a-c7f74f280412
[    3.826633] dracut: /dev/disk/by-uuid/73928837-b11e-4758-812a-c7f74f280412: clean, 467049/6406144 files, 5029002/25599748 blocks
[    3.833681] dracut: Mounting /dev/disk/by-uuid/73928837-b11e-4758-812a-c7f74f280412 with -o rw,noatime,ro
[    3.899892] EXT4-fs (sda6): mounted filesystem 73928837-b11e-4758-812a-c7f74f280412 ro with ordered data mode. Quota mode: none.
[    3.973133] dracut: Mounted root filesystem /dev/sda6
[    4.074689] dracut: Switching root
[    4.183658] systemd[1]: RTC configured in localtime, applying delta of 60 minutes to system time.

[    6.334438] FAT-fs (sda1): Volume was not properly unmounted. Some data may be corrupt. Please run fsck.

[    6.738713] systemd-journald[579]: /var/log/journal/a6964f3a229743e58f5927f5d4b31bf3/system.journal: Monotonic clock jumped backwards relative to last journal entry, rotating.

Should I consider sda as unrepairable or can I fsck sda1 and full repair of sda6 ?

Best regards

[doktor5000]

For sda6 during boot the fsck was clean, so it seems to be OK from a filesystem perspective. Also your dmesg does not show anything bad related to the filesystem, you might need to add full journal logs.
What you should do is either run

Code: Select all

smartctl -a /dev/sda

as root to check the health for the SSD. You could also use something like gsmartcontrol or gnome-disks for more convenience, it shows the same information.
You're mostly looking for Wear_Range_Delta for a Kingston SSD, but also Media_Wearout_Indicator, Wear_Leveling_Count, SSD_Life_Left
but also for unrecoverable read errors which shows if the error correction already triggered and if there are unrecoverable errors - if there are, you should replace the SSD probably.

the sda is a Kingston SSD (7 years)
the sda1 (UEFI Bios) should I "manually" do a fsck , but do not know how ? use a Live Mageia on DVD ?

You can ignore that, as long as it boots as that happens quite often for the UEFI partition.

[filip]

Apart from great doc's advice I'm thinking something else as a cause. IIRC default Mageia fstab doesn't set autorepair on FAT partitions. It seems that your boot partition has FAT FS flag dirty set. If you unmount it, fsck it and then mount it back it should be fine.

[kajosim]

Thanks for your answers.
For safety I first made a complete backup of my data (approx 4TB) .
installed gsmartcontrol .

excerpt from smartctl -a /dev/sda

Code: Select all

ID# ATTRIBUTE_NAME          FLAG     VALUE WORST THRESH TYPE      UPDATED  WHEN_FAILED RAW_VALUE
  1 Raw_Read_Error_Rate     0x000a   100   100   000    Old_age   Always       -       0
  9 Power_On_Hours          0x0012   100   100   000    Old_age   Always       -       42575
 12 Power_Cycle_Count       0x0012   100   100   000    Old_age   Always       -       1425
169 Bad_Block_Rate          0x0000   100   100   000    Old_age   Offline      -       7
192 Unsafe_Shutdown_Count   0x0012   100   100   000    Old_age   Always       -       54
231 SSD_Life_Left           0x0013   100   100   000    Pre-fail  Always       -       95
233 Flash_Writes_GiB        0x0013   100   100   000    Pre-fail  Always       -       8616
241 Lifetime_Writes_GiB     0x0012   100   100   000    Old_age   Always       -       5624


Running gsmartcontrol it shows similar results and no problems.
Running the Extended Self-Test shows also no error .

While booting I was able to login while there still was a lot of disk activity. At least once when I waited until all disk activiy had ended to login then all Dolphin TABs were present.
But at last when I booted I only got an complete empty desktop - no Dolphin nor Konsole windows .
So I decided there must be a problem with the SSD.

I rely daily on my stationary PC so I had already bought a replacement PC but it turned out not to have enough power on any USB connectors to reliably restore data fra a 4TB external disk which at times required more than 2 Watt.
So I decided to buy a new SSD for my 7 year old PC. Both the old and the new SSD uses SATA connection and not using the M2 connector.

So I made a complete new installation of Mageia 9 with updates and installing (most of) the same programs which was on the old installation so I am back to normal and everything is working.

I have an USB-to-SATA connector and connected the old SSD to one of my laptops. Installed gsmartcontrol
Running gsmartcontrol it shows similar results as before and no problems.
Running the Extended Self-Test shows also no error .

Before I would repartition the old SSD I made a backup of the home-partition (ext4 formatted) on the SSD.
During copying I got 3 "could not read" errors :
home/.xine.session.0 , home/.cache/ibus/dbus-XLqeu9vr , home/.cache/ibus/dbus-jgVp4sSE was not readable.

CONCLUSION :
So my 7 year old Kingston SSD do have read errors on "disk" even though both smartctl and gsmartcontrol with Extended Self-Test show no problems.
The Extended Self-Test is build into the SSD so it is not capable to "look" at itself from an external point of view.

I plan to do the following test :
Format the "home" partition as VFAT, copy files to it and see whether there is any read errors, and maybe see how MSWIN behaves.
Other suggestions

Best regards.

[doktor5000]

You only provided a small excerpt of smart data.

CONCLUSION :
So my 7 year old Kingston SSD do have read errors on "disk" even though both smartctl and gsmartcontrol with Extended Self-Test show no problems.
The Extended Self-Test is build into the SSD so it is not capable to "look" at itself from an external point of view.

Those are two completely different things. While the SSD may be fine from a physical/hardware perspective, the filesystem might still be corrupted (logical perspective).

[kajosim]

Thanks.

a complete smartctl log is attached.

My impression is that the file system got more and more corrupt . In the end no windows (Dolphin , konsole) at all after boot The desktop was completely empty .

How can that happen if the SSD may be fine from a physical/hardware perspective ?
What else can make eg home/.cache/ibus/dbus-jgVp4sSE non readable ?

A wild guess is that a gamma-ray has hit the PC .

I will do some more testing to see whether a new formatting will remove the not readabe.

Regards

[morgano]

Other reasons may be transient problems in other parts. I have had "rouge" RAM, CPU, disk cable, main board transiently misbehaving, but I did not understand what until later when problem escalated so I could see what.
Temperature also affect most components.
For example I first changed RAM because it tested bad, but later when running RAM test using all CPU cores, I noticed "memory faults" detected only for two cores, so CPU fault really...
RAM faults can on some mainboards be detected. But CPU faults we are is not as protected from.
And you can imagine what happens if processed filesystem gets written to disk.

So one thing I suggest is to perform RAM test overnight with the option use all cores.

[doktor5000]

How can that happen if the SSD may be fine from a physical/hardware perspective ?
What else can make eg home/.cache/ibus/dbus-jgVp4sSE non readable ?

As you didn't post the actual message this could also be a normal permission issue if those files are only readable by the root user for example.

From the smart data the drive seems fine so far at 95% health and with not even 6TBW well below the designed 80 TBW (terabytes written, shown as Lifetime_Writes_GiB in smart data) spec.

I'd do a full fsck from a live DVD and would continue using it, if you do regular backups.

[kajosim]

Thanks for all answers.

I did a RAM test but forgot about "use all CPU" a whole night. It showed no error.

Installed a new SSD and reinstalled Mageia 9 , updates and extra programs.
And now it boot up with the same windows and TABs as at the last logout.

Short while ago before the new SSD I was in Firefox not able to log into one specific homepage with username and password. This web server has lot of users so some times it takes some time to login. Therefore I thought it might just be an overload issue. But after a new SSD and reinstall it is still the same, I do not get logged in. It might be a Firefox problem so still in Plasma I chose to run the (Gnome) Web browser which gave the same problem. Then logging into the Gnome desktop using the Web browser, still same problem.
I also tried on a laptop using Mageia 9 , Plasma , Firefox and here there is no problem, it works. I rely un using this program so it is good I am still able to use it.
This is my primary PC and I rely on it everyday. As it is now I do not trust my old PC anymore whatever may be the problem.

Morgano may be right about all sorts of possible problems and Doktor5000 about the old SSD still be fully functionel and OK.
After 7 years all Morgano's problems may be possible.
When I have time left I will do some tests on the old SSD and if OK make some use of it. As doktor5000 said it should have many years left.

So after the above I went and bought a new PC with a 1TB SSD. The shop did not have a 6TB HDD so I got a 4TB HDD which I replaced for a 6TB in the 1 year old PC. This 6TB is put into the new PC.
Installed Mageia 9 and updated. At the moment I am using a backup for copying all data files from the first HDD of the old PC.
Soon will the second 6TB HDD of the old PC be unmounted and moved to the new PC so that I do not need to copy the content of this one.
At the same time I will make a final copy of the Thunderbird mail database on the old PC and on the new PC "push it under" so my mails are continuing just as it was a short time before.

[morgano]

I have a "funny" issue in that on one of my five+ laptops i suddenly could not log onto my bank in Firefox.
Also tested a fresh desktop user login.
Some months and FF versions later it works on that computer but not on another laptop.
Both always works using Falkon. (except the site warns me it might not work), and always in FF on all other computers.

When I began programming, I thought programs lasted eternally, only hardware wore down.
Now I have learned that hardware usually lasts far longer than software...

Whatever. There are ways to install Firefox, FirefoxESR, Chromiun and more alternatives as flatpaks, as well as running i.e Falkon. Did you try if they work for your use case?
https://wiki.mageia.org/en/Flatpak

[kajosim]

END OF STORY :

As I wrote : after installing a new SSD everything worked fine -- for a short while.
Then it gradually became worse and in the end the BIOS was not able to see the SSD.
So the 7 year old PC got salvaged.
The 16GB RAM and a SATA DVD-drive was moved to a low specs PC I bought 6 years ago. This is now the only PC I bought from new which has physical space for a DVD drive.

Thanks to all for comments and information