[ 2159.929976] ata1.00: configured for UDMA/33
[ 2159.940321] sd 0:0:0:0: [sda] tag#23 FAILED Result: hostbyte=DID_OK driverbyte=DRIVER_OK cmd_age=2s
[ 2159.940330] sd 0:0:0:0: [sda] tag#23 Sense Key : Aborted Command [current]
[ 2159.940353] sd 0:0:0:0: [sda] tag#23 Add. Sense: No additional sense information
[ 2159.940360] sd 0:0:0:0: [sda] tag#23 CDB: Write(10) 2a 00 16 7f 68 00 00 20 00 00
[ 2159.940363] I/O error, dev sda, sector 377448448 op 0x1:(WRITE) flags 0x0 phys_seg 64 prio class 2
[ 2159.940374] EXT4-fs warning (device sda5): ext4_end_bio:369: I/O error 10 writing to inode 10355122 starting block 47181056)
[ 2159.940401] ata1: EH complete
[ 2159.940416] EXT4-fs (sda5): failed to convert unwritten extents to written extents -- potential data loss! (inode 10355122, error -5)
[ 2159.940440] buffer_io_error: 4086 callbacks suppressed
[ 2159.940444] Buffer I/O error on device sda5, logical block 41656320
[ 2159.940451] Buffer I/O error on device sda5, logical block 41656321
[ 2159.940455] Buffer I/O error on device sda5, logical block 41656322
[ 2159.940458] Buffer I/O error on device sda5, logical block 41656323
[ 2159.940462] Buffer I/O error on device sda5, logical block 41656324
[ 2159.940466] Buffer I/O error on device sda5, logical block 41656325
[ 2159.940469] Buffer I/O error on device sda5, logical block 41656326
[ 2159.940473] Buffer I/O error on device sda5, logical block 41656327
[ 2159.940476] Buffer I/O error on device sda5, logical block 41656328
[ 2159.940480] Buffer I/O error on device sda5, logical block 41656329
[ 2159.948430] ata1.00: Enabling discard_zeroes_data
[ 2159.958815] EXT4-fs (sda5): Remounting filesystem read-only
[ 2159.962042] EXT4-fs (sda5): ext4_do_writepages: jbd2_start: 1024 pages, ino 6556376; err -30
1 Like
I recommend you boot to live USB and run an fsck on the file system. Make sure it is healthy
1 Like
Will fsck mark bad logical blocks so I can use the system ?
I have Debian 13 running and tried to let Ubuntu live CD use a partition and it failed but I moved logs and dmesg to the write partition on the USB if it would help dig deeper !
I gave Firefox AI the Smartctl -x output and got this.
Thanks to your smartctl output, the picture is completely clear.
The previous diagnosis about an NVMe drive was incorrect because your drive is actually a SATA drive: the Samsung SSD 870 QVO 1TB.
The log uncovers the exact hardware reason why Subiquity (Ubuntu) crashes right at the end of the installation while Debian sails smoothly:
The Smoking Gun: CRC_Error_Count (ID 199)
Look at line 199 of your SMART attributes:
199 CRC_Error_Count 097 097 000 - 2355
The drive has registered 2,355 UltraDMA CRC Errors.
- What this means: The drive itself is fundamentally healthy (100% life, 0 reallocated sectors).
However, 2,355 blocks of data sent between your motherboard and the SSD were corrupted or lost mid-transit.- The Cause: This is almost always caused by a bad, loose, or bent SATA data cable, a dusty motherboard SATA port,
or a minor voltage instability on the SATA power rail. [1]Why Ubuntu Fails and Debian Works
- Ubuntu’s Aggressive NCQ Bursting: At the very end of the installation,
Ubuntu’s Subiquity installer extracts compressed filesystem layers (Snaps) using heavily parallelized I/O streams via SATA NCQ (Native Command Queuing).It floors the gas pedal, sending massive bursts of commands down the SATA cable simultaneously.
When the unpatched connection drops bits under this high load, the Linux kernel encounters too many consecutive CRC errors.
It assumes a catastrophic failure, safely disables NCQ link speed, and forces the filesystem into a Read-Only (remount-ro) state to protect your data.
Debian’s Conservative Approach: Debian uses traditional, single-threaded sequential writes (dpkg/.deb). It never pushes the SATA connection hard enough to cause consecutive CRC errors that overwhelm the kernel’s error-recovery thresholds. It essentially glides right past the bad hardware connection.
So it is beginning to feel fixable.
How to Fix It (Hardware Steps)
Do not reinstall yet. Fix the physical data path first to bring the CRC_Error_Count raw value to a halt:
- Replace the SATA Data Cable: Swap out the SATA cable connecting the 870 QVO to your motherboard. Use a high-quality, locking SATA III cable. [1]
- Change Motherboard Ports: Plug the new cable into a completely different SATA port on your motherboard to rule out a faulty controller port.
- Reseat Power: Ensure the SATA power connector from your power supply is pushed firmly all the way into the SSD drive. [1]
(Note: The 2355 count in SMART is historical and will not reset to 0, but it should stop increasing once the connection is secure).
The “Software Workaround” if the cable isn’t the issue
If changing the cable does not stop the crashes, the Samsung 870 series has known incompatibilities with Linux SATA NCQ implementations under heavy workloads.
You can bypass this by turning off NCQ during the Ubuntu boot phase: [2, 3]
- Boot your Ubuntu installer USB.
- At the GRUB menu, highlight “Try or Install Ubuntu” and press e.
- Find the line starting with linux and append this argument to the end: libata.force=noncq
- Press Ctrl+X to boot and run the installer.
Now this looks right.
The critical errors that explicitly trigger the filesystem crash occur much later in the timeline—usually around the time Subiquity runs its final copy and configuration steps.
Even without the tail end of the log, your DMI line reveals the missing hardware context:
[ 0.000000] DMI: Hewlett-Packard 23-s010/2B16, BIOS 80.03 10/16/2014Your machine is an HP All-in-One PC from 2014.
Combining this hardware profile with your previous smartctl logs confirms why Ubuntu is dropping into a read-only state.The Problem: Motherboard Age meets NCQ Quirks
A 2014 HP motherboard uses an early SATA III controller configuration.
This setup frequently encounters issues when communicating with modern,
aggressive multi-queue solid-state drives like the Samsung 870 QVO under specific Linux conditions:
- SATA NCQ Queue Timeout: The combination of your 2014 HP motherboard chipset and the modern Samsung 870 QVO drive creates an error condition during heavily parallelized I/O.
When Ubuntu tries to blast data down multiple channels simultaneously via Native Command Queuing (NCQ), either the aging motherboard controller stalls or the cable drops packets.- The Log Evidence: This matches your smartctl log showing 2,355 UltraDMA CRC Errors.
Under heavy load, the Linux kernel encounters too many dropped packets too quickly, assumes the drive is dying, drops the interface speed,
and switches to a safe read-only state. Debian doesn’t trigger this because its installation workload is single-threaded and sequential.How to Bypass This and Install Ubuntu
Because you are working with an All-in-One desktop system where swapping internal cables or switching motherboard SATA ports is difficult, you can resolve this through a kernel parameter adjustments:
- Boot your Ubuntu Live USB.
- On the initial GRUB boot menu, highlight “Try or Install Ubuntu” and press the e key on your keyboard to edit the boot arguments.
- Look for the line that starts with linux (it looks like your dmesg line: BOOT_IMAGE=/casper/vmlinuz … quiet splash).
- Move your cursor to the very end of that line, add a space, and type exactly:
libata.force=noncq
- Press Ctrl+X or F10 to boot into the Live environment with that flag active.
Why This Works
This command forces the Linux storage subsystem (libata) to run the Samsung 870 QVO in single-command mode, matching Debian’s conservative I/O cadence.
This eliminates the multi-queue command bursting that is overwhelming the older hardware controller or causing link degradation over your physical internal connection.
Such a friendly AI !
Once you get into the live desktop with that boot option,
run the installer and let me know if it finishes successfully.
If you need help making this workaround permanent inside the newly installed Ubuntu system so it does not freeze after rebooting, let me know!
1 Like
Wow, this is utter nonsense (though conceptually not completely wrong)
Snap packages get never extracted anywhere, they are filesystem images that get loop mounted.
What the installer actually does is to copy/extract the read-only livefs bits from the iso to the target disk, packages (of whichever format) have nothing to do with it. After doing this the installer removes all bits for the live mode on the target device and makes the system bootable.
1 Like
Found errors but second pass was successful.
See my long explanation (AI assisted) now in progress and testing.
Testing complete but XFCE4 SNAPD causes errors but works fine with Wayland Ubuntu including a laser utility that uses snap.
And my first step on new systems is use Inkscape to create logo on wallpaper
1 Like
After a day of successful use the read only appeared again so I will wait to see a non-snap installation as snap seems to cause the read only response to the compatibility of 2014 system with new SSD high speed storage.
I am back to the previously mentioned Debian and use snap installed for the laser RAYFORGE application to see if it causes the read only to rear it’s ugly head.
The /home is a separate partition to save my work and I will try a manual installation if there appears to be a fix to the snap installer with Wayland that Ubuntu 26.04 uses.
Sorry if this is not the solution but it at least allows me to use my laser again.
1 Like
Again with the nonsense…
1 Like
What constructive suggestion do you offer to help ?
Buying a new computer or another hard drive have been offered but I continue to listen to those who want Ubuntu to improve and not those that offer no useful DISCOURSE…
Ubuntu is an ancient African philosophy rooted in Nguni Bantu languages.
It translates roughly to “I am because we are” or “a person is a person through other persons”.
1 Like
Indeed. Namely Zulu and Xhosa. Other languages of the same family have cognates with just slightly differences in pronunciation and official/unofficial latinized spellings.
Also an OS and like any other OS/software it doesn’t fix defective or about to fail hardware. Perhaps you need to install Windows instead - if at all possible - and compare results.
The “I am because we are” philosophy isn’t at stake here. This thread has been running for many days and other users have been taking the time to post, give advice, suggestions and even engaging with AI nonsense at times, while explaining that, for starters, it has nothing to do with SNAP. The collectivist and social harmony the term “Ubuntu” imbues also implies respect for the fellow human beings that do voluntary unpaid work here and, for the other non-human resources, a responsible usage of the infrastructure that has been made available to you.
We understand the current crisis in RAM and memory chips in general resulting in obscene prices for any SSD in the consumer market. A new NVMe currently can cost months (yes, plural) of work for many people in developing countries. We are not to blame, we’re victims of the same system just like you. The current parts crisis, specifically, has to do with deals around the “monster” - AI - you helped feeding.
Of course, nothing of this matters for the strictly technical issues. SSDs won’t fix themselves and OSes won’t suddenly produce pixie dust to fix defective hardware because we’re poor. The “Universe” won’t do it either, “it” has no conscience, and there’s no justice giving magical supreme being out there that will fix it for you. Only WE, the People, united and with social class conscience, organized under a vanguard may, eventually, be able to do something about that.
2 Likes
I am trying without SNAP for Firefox and Rayforge and watching 199 CRC error
Captured during Ubuntu install with libata.force=noncq it went from
smartctl -a (apt install smartmontools on Live CD during install)
Local Time is: Thu Jun 25 16:42:54 2026 EDT
199 CRC_Error_Count -OSRCK 097 097 000 - 2355
to repeated attempts that I have /var/log/installer saved and dmesg -w out to text file(s).
I am studying them to understand what they are saying.
Now with no snapd and both Firefox and Rayforge working, I can watch the CRC error countfor a few days while I use the laser and Firefox.
Cut and pasted but not using CODE as it ruins what it shows
$ cat /proc/cmdline
BOOT_IMAGE=/boot/vmlinuz-6.12.94+deb13-amd64 root=UUID=xxxxxxxxxxxxxxxx ro quiet
$ sudo smartctl -a /dev/sda | egrep -i ‘SMART overall|Reallocated_Sector|Current_Pending_Sector|Offline_Uncorrectable|CRC_Error_Count’SMART overall-health self-assessment test result: PASSED 5 Reallocated_Sector_Ct 0x0033 100 100 010 Pre-fail Always - 0199 CRC_Error_Count 0x003e 095 095 000 Old_age Always - 4324
More from smartctl saved to time stamped text file
Local Time is: Sun Jun 28 16:52:48 2026 EDT
199 CRC_Error_Count 0x003e 095 095 000 Old_age Always - 4324
1 Like
Late to the party but do you have the latest BIOS? Are all internal cables OK and seated fully (check this with the system off). If it’s a desktop pc and you have pci express or pci cards then ensure they are seated fully too
1 Like
CRC errors are physical (hardware) issues with the disk, normally this value should be zero, your disk already had above 2.3k of them (the last value in the line is the actual error count) and they nearly doubled up to 4.3k across three days …
Quoting myself from post 4 above:
Beyond the snap nonsense (neither snap nor any other package format can have anything to do with your problem) that it likely collected from some of the sadly very fashionable Ubuntu and snap bashing comments in some other forums, the AI actually made some helpful suggestions, did you try any of these:
If a new cable or any of the other suggestions do not help, you will need a new disk, this one is clearly physically damaged (very likely beyond repair) and you will likely lose data on it regardless which OS you install …
3 Likes
I will accept hardware failure is cause of the CRC error count increasing.
I tried a resize with disks to save data and that failed as well.
I can see that the SSD is a mismatch to the controller and that there are issues beyond the symptoms.
I am practicing safe data etiquette and using NFS to off system capture of work.
I can do re-installs all the time if the CRC error count gets extreme enough that it makes system increasingly unstable.
I don’t and won’t do anything that needs reliability.
Thanks to all of you for your insights and insightful comments.
This reminds me of the old saying.
”doctor, it hurts when I do that !”
I am 74 years old and computers are not that important to me.
I just want to help Ubuntu community with my experience not get into the debate of X11 vs Wayland, XFCE vs Plasma, Ubuntu vs Debian, snap vs flatpak, native vs sandboxed and all the other solutions that work for new hardware not 4th generation Intel.
The other system I have is working fine as a power user with AMD Ryzen 9 . and Ubuntu 26.04 running XFCE4 or stock snap and Ubuntu as installed from same USB.
73’s as the ham members may know.
You too could own this piece of eh, “history”
Main idea is it has three integrated cameras and touch screen and designed as usable without keyboard and mouse though I use Bluetooth Logitech K250 and M196 sold together.
Late to the party but did you bring the free as in beer ?
I also have two DEC Alpha’s from the 1990’s running version 2 of Red Hat.
Should I try to make them run Ubuntu on 1 GB hardrive ?
(Sarcasm or taunting humor ?)
TTFN