Lucid boot failed to complete after fsck

I also noticed that on the Lucid install that does work I get the following message
when I forcechk that partition :

"Your disk needs to be checked for errors, this may take some time"
                     "press C to skip this test"

On the Lucid install that fails I only get:
"Checking disk 1 of 1 (71% complete)"

I never see the press C to skip.

Both installs have the same mountall and esfsck version numbers.

Something similar happened to me, the filesystem stopped at 78% for over 20 minutes. I couldn't figure out how to get to another vty or skip (as vmc notes, the "press C to skip" message was not visible), so I rebooted. Things booted up normally then.

On Fri, 2010-04-02 at 20:33 +0000, vmc wrote:

> Both installs have the same mountall and esfsck version numbers.
>
What are the version numbers of mountall and plymouth?

Scott
--
Scott James Remnant
<email address hidden>

>What are the version numbers of mountall and plymouth?

mountall version 2.8
e2fsck version 1.41.11

Vern

I have the same problem - fschk seems to hang (in my case at different %-stages).
Noticeably - The scan process has already started before the number of disks to be checked is suddenly increased from "foo bar...% 1 of 1" to "foo bar...% 1 of 2" !
The option to interupt with C is visible (all the time) - but the system seems to reach a point during the scan where it is not responding any more.
I am also not able to switch to any tty and have to use the reset-button - the following reboot works fine, without any fs checks

off topic:
During bootup I always get the message:
"Your disk needs to be checked for errors, this may take some time"
but no disks are listed (that should be checked) - in this case, the system boots flawlessly without any file system checks
this may have something to do with the fact that I got "soft-resets fails"-messages on bootup since karmic but thats just a stupid guess..

hardware / platform:
- Gigabyte MA790GP-UD4H (AMD 790GX / SB750 Chipset), Bios v. F7A, SATA set to AHCI Mode
- two SATA Drives:
sda - SuperTalent UltraDrive GX2 64GB (SSD)
sdb - WD Green Power 500GB something..
- AMD64 lucid including all updates

my packet version:
- e2fslibs and e2fsprogs: 1.41.11-1ubuntu1
- plymouth: 0.8.1-4

damn I forgot:
- version of mountall is 2.10

Any hints on other attachments that may help is appreciated

I used todays daily-live APR3, to install with same results.

'sudo touch /forcefsck' , and rebooting.

Now mountall version = 2.10
            e2fsck version = 1.41.11

I'm sorry.

plymouth version = 0.8.1-4

Confirmed
+ every boot a short glims of the message "Your disk needs to be checked for errors, this may take some time"
+ same " hangs" after a 'sudo touch /forcechk' until I boot in recovery mode
Working on a 66bit with ext4 and the option data=writeback

I installed todays, Apt4, daily-live. Same problem.

What I did notice is a difference between aptitude version and mountall version.

aptitude show mountall = Version 2.10
mountall --version = 2.8

Having the same issue. The checker stops at 2 points for me. With 1 HDD, it stops at 71%, with 2 HDD it stops at 80%.

No, way to skip the checker by pressing C

I had the same issue as well 2 days ago. I was stuck at 47% . I couldn't boot into recovery mode either - kept getting something about needing to run fsck manually.

I had to reinstall since I didn't have my live USB with me and my CD drive doesn't work. The only USB drive I had at the time was 256MB and didn't have enough space to make a new USB. And the Ubuntu minimal install wasn't out for Lucid yet :(

Next morning I got my 8GB live USB drive and reinstalled. I'm dreading the next time this happens!

got this yesterday:

first boot: hang with fsck at 71 %
2d boot hang at 92 %
third boot boots well

since, all other boots are normal (but fsck not called at time, wait and see next time)

This sounds rather worrying and should be fixed before the RC. Is sysvinit really the right component?

@amano, That's a good question. As you can see, it "hinted" at sysvinit.

aptitude search e2fsck
p e2fsck-static

But package e2fsck-static is not installed.

Since Scott James Remnant replied, I would think he would have changed the package, if it was wrong.

Now that I think about it, I don't think fsck package is the problem. mountall is.

If you issue a 'sudo touch /forcefsck', after it fails to complete, your left with forcefsck file still intact.

Its mountall duty to remove it. It can't because it doesn't finish.

====
mountall.conf:
exec mountall --daemon $force_fsck $fsck_fix
end script

post-stop script
    rm -f /forcefsck 2>dev/null || true <<never gets here. <<< It needs to get here.

My main desktop just hit this very problem today. at 70% it stops and nothing else happens.

this has raid-1 disks so I'm afraid of booing from another source and having to do all that manually.

This happened to me on Saturday after performing a large update. It stopped at 73% during disk checking. After forcefully shutting down my laptop by holding the power button and rebooting, system booted up normally.

Yes, also the same freeze with the "natural" unforced bootcheck..........

Same for me on a Dell MIni 9; whether forced or "natural", fsck hangs at 70% for a short time, then advances to 71% and the system promptly freezes altogether.

If I shut down with the off button and boot into the recovery mode of the same (latest) kernel, fsck completes and is followed by a lot of error messages about Plymouth failing, then back to the recovery menu. From there normal boot is possible.

Could this be related?

https://bugs.launchpad.net/ubuntu/+source/mountall/+bug/487744

I don't know when this mountall fix was released - might it have broken something?
I'm pretty sure this behaviour is something that started within the last few days.

did a little bit of experimenting:
removed splash from grub boot
and updated grub
forced a next boot check
and rebooted
the boot ended with a black tty screen with a blancking curser
control alt f1 opened a new tty with a login screen
start x / logged in
and everything worked fine

the control alt f1 tric did not work with splash

Isn't that bug 553745 ?

bugs out after removing nvidia* !

Before, I was stopped with fsck @71% and just access to #2.
Problem with gdm-start/stop

I tried to remove nvidia* and reboot : display problem detected by ubuntu, corrected and reboot : it works.

But no 3D effect n_n

Is the bug comes from nvidia driver ?

I doubt if the problem is with nVidia - I had (or have? - it happened to me once so far) this bug on a Dell Mini 10v netbook with Intel graphics.

I understand ; I just reinstall nvidia-185 driver and bug re-happened.
This is my experiment ; maybe a way to find the problem ? :)

This problem has hit me twice recently. The first time (iirc) I was able to ctrl-alt-del at the Plymouth screen, and upon reboot it briefly flashed the fsck message again but booted successfully. The second time was just this morning (with a fully up to date system), and I had to ctrl-alt-f2 to get to a TTY and log into a shell. A "ps -ef | grep fsck" indicated no instances of it hanging around, so it appears that fsck is not to blame. I then did a "sudo reboot" and once again it managed to boot successfully.

Just happened to me as well (stuck at 75% complete). However I am able to ssh into the machine. This happened after I put an previous karmic xorg.conf back into /etc/X11/ to test an nvidia-96 problem (#523108 & #539196) and rebooted.

2.6.32-19-generic (updated as of this morning)
$ apt-cache policy e2fsprogs
e2fsprogs:
  Installed: 1.41.11-1ubuntu1
  Candidate: 1.41.11-1ubuntu1
$ apt-cache policy plymouth
plymouth:
  Installed: 0.8.1-4ubuntu1
  Candidate: 0.8.1-4ubuntu1
$ apt-cache policy mountall
mountall:
  Installed: 2.10
  Candidate: 2.10
$ apt-cache policy nvidia-96
nvidia-96:
  Installed: 96.43.14-0ubuntu11
  Candidate: 96.43.14-0ubuntu11

No instance of fsck running. Nothing obvious in dmseg, xorg.0.log, nothing new in /var/log/fsck.
Tail of .xsession-errors shows:
** (gnome-panel:1870): WARNING **: Failed to send buffer

** (gnome-panel:1870): WARNING **: Failed to send buffer
The application 'gnome-panel' lost its connection to the display :0.0;
most likely the X server was shut down or you killed/destroyed
the application.

so I'm thinking it may be related to the karmic xorg.conf file that I used. I'll try rebooting to see if any change (leaving the xorg.conf file in place).

The xorg.conf doesn't appear to have been the problem. Rebooted leaving the xorg.conf in place & no problem now. So much for that theory.

'sudo touch /forcefsk' and reboot results in "75% complete" again. Moved the xorg.conf file completely out & reboot - does the same (75% complete), so it's repeatable. I can only reboot completely after removing the /forcefsk file.

Could ureadahead have something to do with this?
I noticed this on my laptop when fooling around with ureadahead.

After testing in a vm I can do the following:
sudo touch /forcefsck
reboot.
fsck stalls at 70% for a sec, and the screen jumps directly to the desktop.
ls / shows forcefsck still exists for about 40 secs.

After removing the pack file for ureadahead and touching /forcefsck again...rebooting gets me a disk check to 70% and the screen goes blank, and it doesn't seem to ever bring up the desktop.
host cpu usage is idle.
I can ctrl-alt-f1 and log in.
/var/lib/ureadahead/pack was created.
/forcefsck is never removed

next reboot, disk check appears again (because forcefsck wasn't removed), and the desktop is loaded. /forcefsck is removed.

It is possible that the above users have been experiencing issues because many updates trigger a reprofile (such as installing/removing the nvidia driver)

also after a successful reboot, I get a crash report for plymouthd (#553745)

Nevermind about ureadahead. I uninstalled it and got the same behavior.

Have the same problem here, gets stuck at 71%.

Same thing on a Dell M1330, sticking at 90%.
This is a real showstopper bug.

Same thing on Lenovo T400 2767-R9U

After using 'sudo touch /forcefsck' trying to solve a different issue, my Acer AOD250 froze on 71% as well. Had to boot into Recovery Mode to escape.

Is it possible to make this bug "critcal"? Consequently freezes (in)directly caused by a fscheck at boot voor many users..

Reassigning to mountall for now, I seriously doubt sysvinit has anything to do with this.

I have made the following observations:

Although I am unable to cancel the disk checks by pressing 'C', when the plymouth-animation (red white dots) is running, fsck is running in the background doing the disk checks
The progress indicator percentage seems to increase while the disk check is running. When the disk check finishes, the progress indicator does not increase any more, gets stuck at some percentage value and the plymouth-animation stops and shows all dots as red
When the disk check is in progress, I am able to get to a VT by pressing Ctrl+Alt+F1. This yields a VT with a blinking cursor
When the disk check finishes (plymouth-animation stops) I am unable to get to a VT by pressing Ctrl+Alt+F1
Ctrl+Alt+Delete works regardless of whether the disk check is running or has finished

I have a system with a 1TB hard drive, which takes fsck about 30 minutes to check. When it is time for the disk to be checked, I am greeted with the "Your disk needs to be checked"-message at boot up. Pressing 'C' to cancel the check does nothing, and if I restart my computer (before the disk check has had a chance to finish) I am greeted with the same message and the same inability to cancel the check.
However; if I don't reboot my machine I can hear that the disk is being accessed, and the HDD-activity-LED on my computer case is turned on. If I wait the ~30 minutes that it takes for the disk check to finish, I can hear that the disk stops being accessed, and the HDD-LED turns off. When this happens, the progress indicator gets stuck at some percentage value, the plymouth-animation stops, and I cannot get to a terminal by pressing Ctrl+Alt+F1. Ctrl+Alt+Delete functions however, and when I restart my computer after waiting for - what seems to be - the disk check ending, I am not met by the "Your disk needs to be checked"-message and the computer boots normally.

So it seems that fsck indeed is running in the background, but pressing 'C' does not cancel the disk check, and when the disk check finishes the boot isn't resumed and plymouth animation stops.

The same behaviour is observed when I issue the command "sudo touch /forcefsck", but when the disk check (presumably) finishes (the plymouth animation stops), the file 'forcefsck' isn't removed and the disk check is started again on the next boot. If I boot up from an USB stick and remove the 'forcefsck'-file manually, the system boots up normally.

So the bug seems to be two-fold:

1. The disk check(s) cannot be cancelled by pressing 'C'
2. The system does not continue to boot after the disk check(s) finish

Similar experience In Lucid with Dell Mini-9, EXT4. But FSCK does not finish or continue in the background and it freezes at 70%. Pressing "C" does not cancel at any point in the check but I can get to TTY and remove forcefsck. I have yet to get fsck to complete a full check but I have not tried from a Live USB stick or CD yet.

I can confirm what Rune Svendsen describes!
the updates of udisks, libparted0, mountall (and other updates from the last few days if involved at all) - may have changed the percentage level I am able to get to (-> now 97%, before there was no chance to get higher than 71%) but the result is still the same - the whole system freezes completely, no chance to switch to TTY anymore and the plymouth animation stops entirely with red dots. It is even impossible to en-/disable numlock anymore...

This is kind of getting serious since nautilus is getting slower and slower changing directories, which has come off for a few days now

Same here. However, I can reboot by pressing ctrl+alt+del.

It seems like the fsck goes up to 70% and then it just goes up in 1% stages until it freezes completely. That's weird. Also, it always says 1 of 1 disk, then 1 of 2 disks and after that, 1 of 3 disks. Always cheking 3 disks? That's weird.

I also have this bug with 70% and freeze

Same here freeze at 71 % ..

Just to confirm that this is happening to me as well (physical machines, as well as virtual machines), with all machines updated as of today (Apr 13, 2010).

This freeze also happened to me on the Ubuntu Netbook Remix 10.04, on a asus eee 901 with ssd and ext2 file system.
If the system was not properly shutdown, I think it is recommended to run fsck.
To solve the problem I booted from the Ubuntu USB disk and used the console to run fsck:

Code:

sudo fsck /dev/<your partition>

After this it booted ok again.

possible duplicates 561317 ? 561312 ? 559761 ? 555753 ? 554737 ? 549824 ? 538810 ?

no matter how you initiate fsck, be it by 'touch /forcefsck'' or by manually increasing mount count by 'tune2fs -C <no>' fsckk either silently aborts or with latest kernel update to 2.6.32-20 interrupst boot process ending with a dead system.
Checked with root filesystems ext3 and ext4.

Se also https://bugs.launchpad.net/bugs/538810 which dates back 1 month.

After creating /forcefsck all I get is this at startup by removing the "quiet splash" options in grub (screenshot attached). And it stays there. Video seems to be dead, I can't even change VT but a control+alt+del seems to work ok (at least the system reboots).

Notice I can't boot the system anymore until I enter recovery mode, which boots ok, and then reboot again.

To understand the screenshot messages:
/dev/sda2 is /
/dev/sda3 is /var
/dev/sda5 is /tmp
AlmacenSATA (/dev/sdb1) is /home

How can we debug what's happening in a better way?

You might want to check

https://bugs.launchpad.net/ubuntu/+source/plymouth/+bug/554737

I think it's what is talked about here.
Same symptoms.

Yes indeed; looks as if this might well be the same bug. Good news, if so, as there seems to be some progress being made.

Further to my comments #28, #29, and #30: to confirm, the issue appears to be unrelated to nvidia, xorg.conf, etc. Just had the same happen on my HP G60-530US Notebook. That machine is using:
*-display:0
             description: VGA compatible controller
             product: Mobile 4 Series Chipset Integrated Graphics Controller
             vendor: Intel Corporation

This machine also stopped at 71%. ssh'ing into the machine (as I did on the other) and looking at the processes shows:
$ ps -e
  PID TTY TIME CMD
    1 ? 00:00:00 init
    2 ? 00:00:00 kthreadd
    3 ? 00:00:00 migration/0
    4 ? 00:00:00 ksoftirqd/0
    5 ? 00:00:00 watchdog/0
    6 ? 00:00:00 migration/1
    7 ? 00:00:00 ksoftirqd/1
    8 ? 00:00:00 watchdog/1
    9 ? 00:00:00 events/0
   10 ? 00:00:00 events/1
   11 ? 00:00:00 cpuset
   12 ? 00:00:00 khelper
   13 ? 00:00:00 netns
   14 ? 00:00:00 async/mgr
   15 ? 00:00:00 pm
   17 ? 00:00:00 sync_supers
   18 ? 00:00:00 bdi-default
   19 ? 00:00:00 kintegrityd/0
   20 ? 00:00:00 kintegrityd/1
   21 ? 00:00:00 kblockd/0
   22 ? 00:00:00 kblockd/1
   23 ? 00:00:00 kacpid
   24 ? 00:00:00 kacpi_notify
   25 ? 00:00:00 kacpi_hotplug
   26 ? 00:00:00 ata/0
   27 ? 00:00:00 ata/1
   28 ? 00:00:00 ata_aux
   29 ? 00:00:00 ksuspend_usbd
   30 ? 00:00:00 khubd
   31 ? 00:00:00 kseriod
   32 ? 00:00:00 kmmcd
   35 ? 00:00:00 khungtaskd
   36 ? 00:00:00 kswapd0
   37 ? 00:00:00 ksmd
   38 ? 00:00:00 aio/0
   39 ? 00:00:00 aio/1
   40 ? 00:00:00 ecryptfs-kthrea
   41 ? 00:00:00 crypto/0
   42 ? 00:00:00 crypto/1
   45 ? 00:00:00 pciehpd
   54 ? 00:00:00 kstriped
   55 ? 00:00:00 kmpathd/0
   56 ? 00:00:00 kmpathd/1
   57 ? 00:00:00 kmpath_handlerd
   58 ? 00:00:00 ksnapd
   59 ? 00:00:00 kondemand/0
   60 ? 00:00:00 kondemand/1
   61 ? 00:00:00 kconservative/0
   62 ? 00:00:00 kconservative/1
  283 ? 00:00:00 scsi_eh_0
  284 ? 00:00:00 scsi_eh_1
  285 ? 00:00:00 scsi_eh_2
  286 ? 00:00:00 scsi_eh_3
  287 ? 00:00:00 scsi_eh_4
  288 ? 00:00:00 scsi_eh_5
  301 ? 00:00:00 usbhid_resumer
  320 ? 00:00:00 jbd2/sda7-8
  321 ? 00:00:00 ext4-dio-unwrit
  322 ? 00:00:00 ext4-dio-unwrit
  337 ? 00:00:05 plymouthd
  355 ? 00:00:00 flush-8:0
  358 ? 00:00:07 mountall
  381 ? 00:00:00 upstart-udev-br
  384 ? 00:00:00 udevd
  537 ? 00:00:00 kpsmoused
  705 ? 00:00:00 iwlagn
  706 ? 00:00:00 phy0
  719 ? 00:00:00 i915
  765 ? 00:00:00 hd-audio0
  818 ? 00:00:00 smbd
  828 ? 00:00:00 rsyslogd
  831 ? 00:00:00 sshd
  832 ? 00:00:00 dbus-daemon
  843 ? 00:00:00 smbd
  862 ? 00:00:00 gdm-binary
  863 ? 00:00:00 NetworkManager
  868 ? 00:00:00 avahi-daemon
  870 ? 00:00:00 avahi-daemon
  871 ? 00:00:00...

Latest mountall update in lucid has fixed bug

When I upgraded to Ubuntu 10.4 (Lucid), fsck began taking too long to complete. I once let it run for two hours before stopping it at 91% complete.

As Neil wrote on 2010-05-15. Updating mountall eliminated these problems for me.

Using Synaptic I marked mountall for upgrade. This updated mountall from version 2.14 to version 2.15.

Now my fsck takes about five minutes, the same time it took under Ubuntu 9.04.