Press Ctrl+C to skip file checks on every boot.

[makem2]

I am using 24.04 with Ext4 file systems and every time I boot I get this option which suggest filesystem errors.

I tried to use fsck from recovery mode with the following result:

/lib/recovery-mode/recovery-menu: line 80: /etc/default/rxS: No such file or directory
fsck from util-linux 2.39.3
dev/nvme2nip5 is mounted.
e2fsck: Cannot continue, aborting

Contents of /lib/recovery-mode:

  GNU nano 7.2                           recovery-menu                                      if [ "$choice" = "resume" ]; then
    box_text=$(eval_gettext "You are now going to exit the recovery mode and continue the>
If that's the case, simply reboot from the login screen and then perform a standard boot.>
    whiptail --msgbox "$box_text" 12 70
    clear
    touch /run/friendly_recovery.resume
    systemctl daemon-reload
    systemctl --no-block isolate default.target
    exit
  fi


  /lib/recovery-mode/options/$choice test mode >/dev/null 2>&1
  retval=$?


  # Hack for the fsck case (needs to be cosidered read/write only when
  # in read-only mode and read-only only when in read/write mode)
  if [ "$choice" = "fsck" ] && [ "$READONLY" = "false" ]; then
    retval=1
  fi


  case "$retval" in
    0)
      # 0 => requires read/write
      if [ "$READONLY" = "true" ]; then
        box_text=$(eval_gettext "Continuing will remount your / filesystem in read/write >
Do you wish to continue?")
        whiptail --yesno "$box_text" 10 70 || continue


        if [ "$choice" = "fsck" ]; then
            FSCHECK="true"
        fi


        . /etc/default/rcS
        if [ -d /run/systemd/system ]; then
            [ "$FSCKFIX" = "yes" ] && fsck_mode="-y" || fsck_mode='-a'
            [ "$FSCHECK" = "true" ] || [ -f /forcefsck ] && fsck $fsck_mode
            systemctl start --job-mode=ignore-dependencies systemd-remount-fs.service
            mount -a
        else
            [ "$FSCHECK" = "true" ] || [ -f /forcefsck ] && force_fsck="--force-fsck"
            [ "$FSCKFIX" = "yes" ] && fsck_fix="--fsck-fix"
            mountall $force_fsck $fsck_fix --no-events
        fi
        rm -f /forcefsck


        if [ "$choice" = "fsck" ]; then
          echo ""
          echo $(eval_gettext "Finished, please press ENTER")
          read TMP
        fi


        READONLY=false
      fi
    ;;


    1)
      # 1 => read-only only
      if [ "$READONLY" = "false" ]; then
        box_text=$(eval_gettext "The option you selected requires your filesystem to be in read-only mode. Unfortunately another option you selected earlier, made you exit this mode.
The easiest way of getting back in read-only mode is to reboot your system.")
        whiptail --msgbox "$box_text" 12 70
        continue
      fi
    ;;


    2)
      # 2 => works in all cases
      # nothing to do
    ;;
  esac


  export READONLY
  /lib/recovery-mode/options/$choice
done

makem@makem-22:~$ sudo grub-install --version[sudo] password for makem: 
grub-install (GRUB) 2.12-1ubuntu7
makem@makem-22:~$

Searching the net for this missing file . /etc/default/rcS returns:

. /etc/default/rcs
/etc/default/rcS is a configuration file in older Ubuntu releases (pre-Systemd) that controls the behavior of the /etc/rcS</span>.d directory, which contains scripts executed during the boot process. This file is no longer relevant in Ubuntu 17.04 and later, as Systemd replaced Upstart and Initscripts.

That suggests that there is a problem elsewhere. Can I get assistance please?

[TheFu]

dev/nvme2nip5 is mounted.
e2fsck: Cannot continue, aborting

is the important error.

We cannot fsck any file system that is current mounted. One solution is to boot from any Ubuntu install ISO, enter Try Ubuntu mode, then run the fsck.

In theory, at the grub menu, we can pick "Advanced" and enter the recovery console. There should be an option to fsck all file systems. For simple installs, I'd expect that to work, but if the install is complicated with other file systems or encryption, it may not work.

We can force an fsck every xyz days on our file systems at boot. This is controlled using the tune2fs tool. I don't recall what the default is, but remember it wasn't what I wanted. It is set once per file system. It is on the file system, not necessarily on the partition, so be certain you point to the device file holding the file system. This matters if you use LVM, for example.

[hifron]

If there is Windows present I would expect Windows partition corruption due caching of its partition(in Windows) as it is set by default when access from Ubuntu(btw Windows could also access Ubuntu via Total Commander plugin )...

Ubuntu has some program this way but is not enabled by default to precache programs at start.

So if there is UEFI fat partition for both systems, it could be one of the problem, but don't think so so eagarly, maybe there is better answer...

And why the corruption? Because different drivers different behaviour(fat) and that is(or could be) also the ext4 way if different kernel means different bugs and lower driver version is unsupported(like booting data partition from newer Ubuntu in older Ubuntu install)

maybe yes, maybe not I don't know if this is the problem...

[makem2]

dev/nvme2nip5 is mounted.
e2fsck: Cannot continue, aborting

is the important error.

We cannot fsck any file system that is current mounted. One solution is to boot from any Ubuntu install ISO, enter Try Ubuntu mode, then run the fsck.

In theory, at the grub menu, we can pick "Advanced" and enter the recovery console. There should be an option to fsck all file systems. For simple installs, I'd expect that to work, but if the install is complicated with other file systems or encryption, it may not work.

We can force an fsck every xyz days on our file systems at boot. This is controlled using the tune2fs tool. I don't recall what the default is, but remember it wasn't what I wanted. It is set once per file system. It is on the file system, not necessarily on the partition, so be certain you point to the device file holding the file system. This matters if you use LVM, for example.

I do have a Windows dual boot system.

I chose at grub menu, to pick "Advanced" and enter the recovery console. There was an option to fsck all file systems. It did not work.

I will use the Ubuntu ISO method thanks. However I think it may be worth doing a check without repair first. I forget the syntax for that but will find it.

Edit: sudo fsck -N /dev/sdx find errors sudo fsck -n /dev/sdx to print errors fsck -y /dev/sdx fix errors .

[1fallen]

You mean this in Advanced boot:

sudo fsck -f /

BTW I see fsck at times as well on a one week old install of 24.04.

What about smart-tools, have you checked that first.

[makem2]

You mean this in Advanced boot:

sudo fsck -f /

BTW I see fsck at times as well on a one week old install of 24.04.

What about smart-tools, have you checked that first.

No, I mean I chose Advanced Options followed by Recovery Mode. It didn't work.

[makem2]

I used a 20.04 live usb:

xubuntu@xubuntu:~$ sudo fsck.ext4 -n /dev/nvme1n1p5
e2fsck 1.45.5 (07-Jan-2020)
/dev/nvme1n1p5: clean, 282818/2632672 files, 6332453/10530304 blocks
xubuntu@xubuntu:~$ sudo fsck.ext4 -n /dev/nvme1n1p6
e2fsck 1.45.5 (07-Jan-2020)
/dev/nvme1n1p6: clean, 462013/5496832 files, 15392844/21972736 blocks
xubuntu@xubuntu:~$ sudo fsck.ext4 -n /dev/nvme1n1p7
e2fsck 1.45.5 (07-Jan-2020)
/dev/nvme1n1p7 has unsupported feature(s): FEATURE_C12
e2fsck: Get a newer version of e2fsck!


games-steam: ********** WARNING: Filesystem still has errors **********


xubuntu@xubuntu:~$

So, the OS is ok but partitions, well 1 of them is not. Seems I need a newer version of e2fsck!. Maybe my current 24.04 has it, we will see. At least I can check them from my running system. games-steam has a windows connection

makem@makem-22:~$ sudo fsck.ext4 -n /dev/nvme1n1p7
[sudo] password for makem: 
e2fsck 1.47.0 (5-Feb-2023)
Warning!  /dev/nvme1n1p7 is mounted.
Warning: skipping journal recovery because doing a read-only filesystem check.
games-steam: clean, 361/10911744 files, 11727348/43633920 blocks
makem@makem-22:~$ sudo fsck.ext4 -n /dev/nvme1n1p7
e2fsck 1.47.0 (5-Feb-2023)
games-steam: clean, 361/10911744 files, 11727348/43633920 blocks
makem@makem-22:~$

Windows found errors on /data drive, offered to check and fix. After running said no errors found!

Ubuntu still has same 'check file system' error though. Now I don't know where to go?

[TheFu]

Warning!  /dev/nvme1n1p7 is mounted.

That needs to be fixed before you can run fsck. If you used a file manager and clicked on the partition, some file managers will mount it.

Also, if the system has been touched by 24.04, you need to use a 24.04 ISO. Tools built in 2020 can't possibly know what tools and data in 2024 will need.

I've never needed to specify the exact file system program. I let the fsck program figure out which file system specific tool should be used.

I'd use AFTER ensuring it isn't mounted and has a native Linux file system.

sudo fsck -y  /dev/nvme1n1p7

[makem2]

Warning!  /dev/nvme1n1p7 is mounted.

That needs to be fixed before you can run fsck. If you used a file manager and clicked on the partition, some file managers will mount it.

Also, if the system has been touched by 24.04, you need to use a 24.04 ISO. Tools built in 2020 can't possibly know what tools and data in 2024 will need.

I've never needed to specify the exact file system program. I let the fsck program figure out which file system specific tool should be used.

I'd use AFTER ensuring it isn't mounted and has a native Linux file system.

sudo fsck -y  /dev/nvme1n1p7

The partition which seems to be causing the problem is NTFS which means that windows should deal with it. All ext4 partitions are clean. Two were checked via the live USB and the remaining one after unmounting i. The only others were the windows OS which it checked and found to be clean if you can believe the result which had an error which was not found when repairing by windows.

Do you mean that it is still worth repeating with a 24.04 live USB the ext4 clean checks?

I think I need to find out about smart-tools as maybe that can deal with both the ext4 and NTFS systems.

[TheFu]

SMART tools, there are many, don't know about file systems at all or partitions. They are useful in looking at what the drive knows about itself.

However, for nvme storage, you shouldn't use smartctl. Need to use the NVMe tool instead. It is conveniently named nvme. Regardless, both the smartctl and nvme programs aren't pre-installed, so they need to be installed on the system to be used - if you are running in a Try Ubuntu environment (off an install ISO), then you will still install them, but they will not be there the next time you boot from that same ISO. ISOs are write-once. No updates that we do when running an ISO will be retained.