mount : / is busy, because of lightdm

[blitzter47]

Hello,

I have Ubuntu 12.04
When I do shutdown/reboot normally, I got, at the end, the error

...
mount : / is busy
* Will now halt

Then when Ubuntu boots, shell indicates that there's a recovery on hard drive where Ubuntu system resides:

[    7.252288] EXT4-fs (sdb5): INFO: recovery required on readonly filesystem
[    7.252353] EXT4-fs (sdb5): write access will be enabled during recovery
[   11.201583] EXT4-fs (sdb5): recovery complete
[   11.203425] EXT4-fs (sdb5): mounted filesystem with ordered data mode. Opts: (null)

But, if I stop lightdm through Terminal before shutdown/reboot, the error messages disappear.
So in terminal:

sudo stop lightdm

then log in through tty for entering shutdown/reboot command, the "mount : / is busy" error disappears before powering computer off, and booting Ubuntu doesn't have to do recovery.

I suppose that there's permission problem for terminating lightdm process during shutdown/reboot process or something wrong while terminating lightdm process (and maybe child processes also)

How to investigate/debug shutdown process step by step or solve permission problem while terminating processes? Do you have any other idea?

Thanks.
By the way, to bypass this error, I used Terminal or created a script to execute the following commands for shutdowns and reboots:

sudo stop lightdm & sudo reboot
sudo stop lightdm & sudo shutdown -h now

[blitzter47]

Fixed.

Summary
It seems that there's a root's abnormal process (in my case) "/sbin/plymouthd --mode=shutdown" that doesn't terminate at all and is still active while shutting down/rebooting, causing unclean unmount. But if I manually terminate lightdm (or gdm) first, then reboot, system unmount will be clean, without problem.

Of course, it can be not only plymouthd, but also any other application/process that hangs (eg. mysql, apache, etc.) Just try manual diagnostic by yourself in "Methodology" section to see what's wrong.

Symptoms
When shutting down/rebooting, (without "splash", nor "quiet" parameters in grub), in console, it displays

[...]
mount: / is busy
* Will now halt

Then when booting Ubuntu (without "splash," nor "quiet" parameters in grub), it displays recovery message:

[ 7.252288] EXT4-fs (sdb5): INFO: recovery required on readonly filesystem
[ 7.252353] EXT4-fs (sdb5): write access will be enabled during recovery
[ 11.201583] EXT4-fs (sdb5): recovery complete
[ 11.203425] EXT4-fs (sdb5): mounted filesystem with ordered data mode. Opts: (null)

Methodology
The source of problem is known by creating bash script manually, containing code:

#! /bin/sh

ps -Af >> process.log

executed during reboot (or shut down), in /etc/rc6.d (reboot) directory, placed before "S60umountroot" script file (in numerical order). After reboot, "process.log" file is created in hard drive root / directory, which will log the problematic "/sbin/plymouthd --mode=shutdown" process, that is not supposed to be present, at the right side, at the end of file, like this (in bold):

UID PID PPID C STIME TTY TIME CMD
root 1 0 0 00:28 ? 00:00:01 /sbin/init
root 2 0 0 00:28 ? 00:00:00 [kthreadd]
root 3 2 0 00:28 ? 00:00:00 [ksoftirqd/0]
root 5 2 0 00:28 ? 00:00:00 [kworker/u:0]
root 6 2 0 00:28 ? 00:00:00 [migration/0]
root 7 2 0 00:28 ? 00:00:00 [watchdog/0]
root 8 2 0 00:28 ? 00:00:00 [cpuset]
root 9 2 0 00:28 ? 00:00:00 [khelper]
root 10 2 0 00:28 ? 00:00:00 [kdevtmpfs]
root 11 2 0 00:28 ? 00:00:00 [netns]
root 12 2 0 00:28 ? 00:00:00 [sync_supers]
root 13 2 0 00:28 ? 00:00:00 [bdi-default]
root 14 2 0 00:28 ? 00:00:00 [kintegrityd]
root 15 2 0 00:28 ? 00:00:00 [kblockd]
root 16 2 0 00:28 ? 00:00:00 [ata_sff]
root 17 2 0 00:28 ? 00:00:00 [khubd]
root 18 2 0 00:28 ? 00:00:00 [md]
root 19 2 0 00:28 ? 00:00:00 [kworker/u:1]
root 20 2 0 00:28 ? 00:00:00 [kworker/0:1]
root 21 2 0 00:28 ? 00:00:00 [khungtaskd]
root 22 2 0 00:28 ? 00:00:00 [kswapd0]
root 23 2 0 00:28 ? 00:00:00 [ksmd]
root 24 2 0 00:28 ? 00:00:00 [khugepaged]
root 25 2 0 00:28 ? 00:00:00 [fsnotify_mark]
root 26 2 0 00:28 ? 00:00:00 [ecryptfs-kthrea]
root 27 2 0 00:28 ? 00:00:00 [crypto]
root 35 2 0 00:28 ? 00:00:00 [kthrotld]
root 37 2 0 00:28 ? 00:00:00 [scsi_eh_0]
root 38 2 0 00:28 ? 00:00:00 [scsi_eh_1]
root 41 2 0 00:28 ? 00:00:00 [kworker/0:2]
root 61 2 0 00:28 ? 00:00:00 [devfreq_wq]
root 202 2 0 00:28 ? 00:00:00 [ttm_swap]
root 252 2 0 00:28 ? 00:00:00 [jbd2/sda7-8]
root 253 2 0 00:28 ? 00:00:00 [ext4-dio-unwrit]
root 773 2 0 00:28 ? 00:00:00 [krfcommd]
root 1139 2 0 00:28 ? 00:00:00 [flush-8:0]
root 5886 2 0 00:33 ? 00:00:00 [kworker/0:0]
root 8221 1 0 00:35 ? 00:00:00 /bin/sh /etc/init.d/rc 6
root 8255 1 3 00:35 ? 00:00:00 /sbin/plymouthd --mode=shutdown #should not be present for normal shut down/reboot
root 8467 8221 0 00:35 ? 00:00:00 /bin/sh /etc/rc6.d/S60perso-script stop
root 8471 8467 0 00:35 ? 00:00:00 ps -Af

If you would like to check that yourself (all with sudo):

• Do following renaming in /etc/rc6.d (reboot) directory:
  ##original filename ----->--------renamed to --------->------customized filename##
  S60 umountroot-------------->------renamed to ---------->---------S61umountroot

• Then create bash script (containing above code) (with executable attribute) before "S61umountroot" by naming it "S60perso-script" in same directory.
  
  By the way, our script is after "S59cryptdisks-early" file (according to S## order number).
• Then reboot and look at process.log file in hard drive root /, which will contain "/sbin/plymouthd --mode=shutdown" at the right side, at the end of file, like above example.

My solutions (need sudo)
The best way, I think, is to kill plymouthd just before system is unmounting.

So still with the early manually created bash script "S60perso-script" in /etc/rc6.d (reboot) directory, replace the file content by :

#! /bin/sh
### BEGIN INIT INFO
# Provides:          
# Required-Start:
# Required-Stop:
# Should-Stop:       
# Default-Start:
# Default-Stop:      0 6
# Short-Description: Kill freezing plymouthd
### END INIT INFO

PATH=/usr/lib/lightdm/lightdm:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games

case "$1" in
  start)
    
    killall -9 plymouthd
    
    ;;
  restart|reload|force-reload)
    
    ;;
  stop)
    
    killall -9 plymouthd
    
    ;;
  *)

    ;;
esac

:

Then do same thing in /etc/rc0.d (shut down) directory (with renaming "S60umountroot" to "S61umountroot"). Then you can create symbolic link or copy our script from /etc/rc6.d to /etc/rc0.d.

So same file, with same content, same filename and executable attribute, in /etc/rc6.d and /etc/rc0.d.
So killing plymouthd command will execute not only for reboots, but also for shut downs.

Alternative solution (which can unfortunately add delay displaying shut down splash screen or console message):
To avoid computer to invoke plymouthd during shut down/reboot. just edit (with sudo) /etc/init/plymouth.conf file by adding # to comment the following line (line 21):

***Final result file***
[...]
21|        #exec /sbin/plymouthd --mode=shutdown

Then save.