fsck cannot be ignored with plymouth

[Martin Erik Werner]

Binary package hint: plymouth

PROBLEM

There seems to be no way to skip an fsck with plymouth.

TEST CASE

1. sudo touch /forcefsck && sudo reboot

2. While fsck is in progress, press c

3. press i for ignore

* Plymouth locks up and reboot is required

OBSERVATIONS

Currently, and in my case, to be able to get to the cancel screen at all, c must be pressed almost immediately as plymouth splash is initiated, due to Bug #571707 which makes plymouth stall otherwise.
If using the patched packages available for said bug, it seems to be possible to cancel at any point during the fsck (there is still a considerable delay between the keypress and the cancel screen though), but the issue in this bug remains.

Message when c is pressed says "Serious error was encounterd";
cancelling an fsck does not inherently cause errors, does it? This is a case of very bad wording to me.

Trying to skip mounting (pressing s) after cancelling fsck locks up plymouth with the message "waiting for /tmp" Is this the right way to handle this? Should the option to skip mounting the root filesystem even exist? (Now Bug #578184)

Dropping to maintenance shell with m works as expected, but only if plymouth hasn't already locked up by any of the other options

---

Architecture: i386
CurrentDmesg:
 [ 77.543932] vboxdrv: exports duplicate symbol RTMemExecFree (owned by vboxguest)
 [ 78.010602] vboxvfs: Successfully loaded version 3.1.6_OSE (interface 0x00010004)
 [ 85.195014] end_request: I/O error, dev fd0, sector 0
 [ 85.308195] end_request: I/O error, dev fd0, sector 0
 [ 85.668090] eth1: no IPv6 routers present
DefaultPlymouth: /lib/plymouth/themes/ubuntu-logo/ubuntu-logo.plymouth
DistroRelease: Ubuntu 10.04
InstallationMedia: Ubuntu 10.04 "Lucid Lynx" - Alpha i386 (20091209)
Lsusb: Error: command ['lsusb'] failed with exit code 1:
MachineType: innotek GmbH VirtualBox
Package: plymouth 0.8.2-2ubuntu2
PackageArchitecture: i386
ProcCmdLine: BOOT_IMAGE=/boot/vmlinuz-2.6.32-22-generic root=UUID=c3804ba1-72aa-4d8c-88c3-1bb81f2619f6 ro quiet splash
ProcEnviron:
 LANG=en_GB.utf8
 SHELL=/bin/bash
ProcFB: 0 VGA16 VGA
ProcVersionSignature: Ubuntu 2.6.32-22.33-generic 2.6.32.11+drm33.2
Tags: lucid
TextPlymouth: /lib/plymouth/themes/ubuntu-text/ubuntu-text.plymouth
Uname: Linux 2.6.32-22-generic i686
UserGroups: adm admin cdrom dialout lpadmin plugdev sambashare
dmi.bios.date: 12/01/2006
dmi.bios.vendor: innotek GmbH
dmi.bios.version: VirtualBox
dmi.modalias: dmi:bvninnotekGmbH:bvrVirtualBox:bd12/01/2006:svninnotekGmbH:pnVirtualBox:pvr1.2:
dmi.product.name: VirtualBox
dmi.product.version: 1.2
dmi.sys.vendor: innotek GmbH

[Martin Erik Werner (arand)]

apport information

[Martin Erik Werner (arand)]

apport information

[Martin Erik Werner (arand)]

apport information

[Martin Erik Werner (arand)]

apport information

[Martin Erik Werner (arand)]

apport information

[Martin Erik Werner (arand)]

apport information

[Martin Erik Werner (arand)]

apport information

[Martin Erik Werner (arand)]

apport information

[Martin Erik Werner (arand)]

Confirming as per https://bugs.edge.launchpad.net/ubuntu/+source/plymouth/+bug/571707/comments/83

[Martin Erik Werner (arand)]

When running fsck manually (recovery mode -> service rsyslog stop; screen; plymouthd; plymouth --show-splash; mount -o ro,remount /; mountall --force-fsck --no-events -v) I get this output [screenshot] when I switch over to plymouth and press c.

[yeti (utu)]

comments, fwiw:
1.#577485 re showfsck is related.
2.my installation says it will fsck every 20 boots.
3.no fsck has ever shown any errors.
4.why are we doing so often an fsck on a journalled fs, anyway?

[Colin Watson (cjwatson)]

Accepted mountall into lucid-proposed, the package will build now and be available in a few hours. Please test and give feedback here. See https://wiki.ubuntu.com/Testing/EnableProposed for documentation how to enable and use -proposed. Thank you in advance!

[Martin Erik Werner (arand)]

Ignoring seems to work as it should with proposed update.
(I had one random occurance where plymouth hung after i was pressed and the interactive section had been cleared, but I have been unable to reproduce this, and otherwise it has worked exactly as expected)

"Skip mounting" issue is still unchanged (is this relevant? Is it a wontfix? Should it be addressed in a new bug?)

The wording is unchanged, but again, should this be another bug?

Marking as verification-done since the main issue is indeed gone

[Launchpad Janitor (janitor)]

This bug was fixed in the package mountall - 2.15

---------------
mountall (2.15) lucid-proposed; urgency=low

  [ Scott James Remnant ]
  * Fix an obvious thinko error that meant that the "I"gnore fsck error key
    for a "hard" failure was ignored.
  * When cancelling filesystem checks, only cancel those that are actually
    checking filesystems; otherwise those that are merely verifying the
    superblock will return an "unrecoverable error" rather than "cancelled".
    LP: #577331.

  [ Steve Langasek ]
  * Only send plymouth a progress update when there's actual progress to
    report; otherwise we flood plymouthd with redundant events, and the
    progress will spin for minutes after the fsck itself is finished. Thanks
    to Tero Mononen and Anders Kaseorg for the patch. LP: #571707.
 -- Steve Langasek <email address hidden> Sun, 09 May 2010 01:04:24 +0200

[Steve Langasek (vorlon)]

copied from lucid-proposed to maverick.

[Steve Langasek (vorlon)]

arand,

With the mountall in proposed, you should not be getting the 'serious error was encountered' message after pressing 'c'. Are you saying that you still get this message?

The mountall code is subtle and quick to anger, so other issues should certainly be filed as separate bug reports and taken one step at a time; but if you're still getting "serious error" on fsck cancel, that implies bug #577331 is not fixed, which is one of the ones supposed to be fixed with this SRU.

[Martin Erik Werner (arand)]

Yes, for me I still get the message "serious errors were found while checking the disk drive for /" whenever I cancel the fsck.
The error messages on the screenshot in comment #10 are still the same and present if I cancel and switch to the plymouth VT (arrowkey left).

I am using mountall 2.15 from -proposed.

[Martin Erik Werner (arand)]

Side note.
Concerning the habit of plymouth to happily suggest skip mounting the root FS I've put down separately Bug #578184

[Steve Langasek (vorlon)]

Ok, so you seem to get the specific error "Inode bitmap not loaded while setting block group checksum info". This is an error from fsck, and not a bug in mountall - it sounds like you may genuinely have an error on your root filesystem, because cancelling fsck should not result in this error being returned. If you let fsck run, does it finish successfully? If you then run fsck again on the next reboot and cancel it, do you again get the error?

[Martin Erik Werner (arand)]

yes, and yes.
Also tested on a fresh lucid->maverick machine, and same issue is seen there.

It could be isolated to virtualbox, since that is the only thing I'm using. Although I don't see why it would...

[Benjamin Kay (benkay)]

Using mountall from lucid-proposed, I also still get the "Serious errors were found..." message when I press c to cancel the fsck. I am not testing on VirtualBox -- this is happening on my actual hardware. The situation is somewhat improved, since pressing i to ignore the error message now resumes normal startup (before, pressing i did nothing). Ideally, though, "Serious errors..." should not be displayed at all on cancellation per Bug #578184.

[Launchpad Janitor (janitor)]

This bug was fixed in the package mountall - 2.15

---------------
mountall (2.15) lucid-proposed; urgency=low

  [ Scott James Remnant ]
  * Fix an obvious thinko error that meant that the "I"gnore fsck error key
    for a "hard" failure was ignored.
  * When cancelling filesystem checks, only cancel those that are actually
    checking filesystems; otherwise those that are merely verifying the
    superblock will return an "unrecoverable error" rather than "cancelled".
    LP: #577331.

  [ Steve Langasek ]
  * Only send plymouth a progress update when there's actual progress to
    report; otherwise we flood plymouthd with redundant events, and the
    progress will spin for minutes after the fsck itself is finished. Thanks
    to Tero Mononen and Anders Kaseorg for the patch. LP: #571707.
 -- Steve Langasek <email address hidden> Sun, 09 May 2010 01:04:24 +0200

[Martin Erik Werner (arand)]

Should this bug be reopened? Or should the fsck error/fail on user cancellation be considered a separate issue, since it is in fact possible to ignore the fsck with the current fix?

[Martin Erik Werner (arand)]

Also running on Maverick with the same issue, another virtualbox.

This is what is printed to the plymouth VT (similar to above screenshot):
"
fsck from util-linux-ng 2.17.2
User cancelled filesystem checks
/dev/sda1: e2fsck canceled
fsck.ext4: Inode bitmap not loaded while setting block group checksum info
mountall: fsck / [207] terminated with status 8
mountall: Unrecoverable fsck error: /
"

If running fsck from a liveCD, cancelling it will produce the same error:
"
^C/dev/sda1: e2fsck canceled
fsck.ext4: Inode bitmap not loaded while setting block group checksum info
~$ echo $?
8
"

Is this error from fsck correct? Is the problem the error in fsck, or is it the way mountall/plymouth interprets the error as a serious failiure?

Letting the fsck run through correctly (in which case no errors are reported) does not affect this in any way.

[yeti (utu)]

I think this bug may be a byproduct of the bug described in 540801.
A work-around for 540801 confirmed by many is proposed there.
I'd be interested to know if anyone else sees the (or a) connection.
I haven't tried that fix myself, because I don't understand it well enough.

[Steve Langasek (vorlon)]

On Mon, May 17, 2010 at 08:25:25PM -0000, yeti wrote:
> I think this bug may be a byproduct of the bug described in 540801.

It is not.

--
Steve Langasek Give me a lever long enough and a Free OS
Debian Developer to set it on, and I can move the world.
Ubuntu Developer http://www.debian.org/
<email address hidden> <email address hidden>

[Steve Langasek (vorlon)]

On Sun, May 16, 2010 at 02:47:18PM -0000, arand wrote:
> If running fsck from a liveCD, cancelling it will produce the same error:
> "
> ^C/dev/sda1: e2fsck canceled
> fsck.ext4: Inode bitmap not loaded while setting block group checksum info
> ~$ echo $?
> 8
> "

> Is this error from fsck correct? Is the problem the error in fsck, or is
> it the way mountall/plymouth interprets the error as a serious failiure?

That's not a problem in mountall/plymouth. If the error is incorrect, then
it's a bug in e2fsprogs; please open a separate bug report against that
package.

--
Steve Langasek Give me a lever long enough and a Free OS
Debian Developer to set it on, and I can move the world.
Ubuntu Developer http://www.debian.org/
<email address hidden> <email address hidden>

[Martin Erik Werner (arand)]

Reported Bug #582035 against e2fsprogs.

The bug here should be considered fixed then I take it.

[Royston F. Pickering (trustroy)]

Cannot mount usb drives or cd drives