initscripts in jaunty does not umount root filesystem jfs

Bug #302871 reported by Graziano
4
Affects Status Importance Assigned to Milestone
sysvinit (Ubuntu)
Invalid
Undecided
Unassigned

Bug Description

Binary package hint: initscripts

Description: Ubuntu jaunty (development branch)
Release: 9.04
initscripts:
  Installed: 2.86.ds1-61ubuntu1
  Candidate: 2.86.ds1-61ubuntu1
  Version table:
 *** 2.86.ds1-61ubuntu1 0
        500 http://it.archive.ubuntu.com jaunty/main Packages
        100 /var/lib/dpkg/status
jfsutils:
  Installed: 1.1.12-2
  Candidate: 1.1.12-2
  Version table:
 *** 1.1.12-2 0
        500 http://it.archive.ubuntu.com jaunty/main Packages
        100 /var/lib/dpkg/status

Using jaunty, with jfs root partition. I expect that at shutdown AFTER all processes TERM and KILL all filesystems shall cleanly umount and flush journals to disk BEFORE poweroff.

Shutdown with intrepid was Ok. I noticed at boot a filesystem check is ALWAYS needed on root filesystem. Found that at shutdown root filesystem is NOT unmounted (reported as busy).
I have to startup always with AC power to force fsck on my laptop, or fsck by hand at boot (root is mounted read-only and startup procedure ends up in big mess...).

Revision history for this message
Albert Damen (albrt) wrote :

Can you please attach a syslog containing a shutdown sequence?
If you do it quickly after you started your system it will be in /var/log/syslog. If you do it later the log may have rotated to /var/log/syslog.0

Changed in sysvinit:
status: New → Incomplete
Revision history for this message
Graziano (graziano-giuliani-gmail) wrote :
Download full text (3.5 KiB)

This is the relevant part after shutdown:

Nov 29 11:44:05 portatile init: tty4 main process (4639) killed by TERM signal
Nov 29 11:44:05 portatile init: tty5 main process (4640) killed by TERM signal
Nov 29 11:44:05 portatile init: tty2 main process (4646) killed by TERM signal
Nov 29 11:44:05 portatile init: tty3 main process (4647) killed by TERM signal
Nov 29 11:44:05 portatile init: tty6 main process (4648) killed by TERM signal
Nov 29 11:44:05 portatile init: tty1 main process (5816) killed by TERM signal
Nov 29 11:44:05 portatile bonobo-activation-server (graziano-12412): could not a
ssociate with desktop session: Failed to connect to socket /tmp/dbus-hVR4aLYLAl:
 Connection refused
Nov 29 11:44:06 portatile spamd[5076]: spamd: server killed by SIGTERM, shutting
 down
Nov 29 11:44:09 portatile kernel: [ 3381.550257] ip6_tables: (C) 2000-2006 Netfi
lter Core Team
Nov 29 11:44:10 portatile bonobo-activation-server (graziano-12594): could not a
ssociate with desktop session: Failed to connect to socket /tmp/dbus-hVR4aLYLAl:
 Connection refused
Nov 29 11:44:10 portatile avahi-daemon[5028]: Got SIGTERM, quitting.
Nov 29 11:44:10 portatile avahi-daemon[5028]: Leaving mDNS multicast group on in
terface wlan0.IPv4 with address 192.168.1.37.
Nov 29 11:44:14 portatile NetworkManager: <WARN> nm_signal_handler(): Caught si
gnal 15, shutting down normally.
Nov 29 11:44:14 portatile kernel: klogd 1.5.0, ---------- state change ---------
-
Nov 29 11:44:14 portatile kernel: Kernel logging re-started after SIGSTOP.
Nov 29 11:44:14 portatile ntpd[6350]: ntpd exiting on signal 15
Nov 29 11:44:14 portatile acpid: exiting
Nov 29 11:44:14 portatile ntfs-3g[4270]: Unmounting /dev/sda1 (WinXP)
Nov 29 11:44:14 portatile NetworkManager: <info> (eth0): now unmanaged
Nov 29 11:44:14 portatile NetworkManager: <info> (eth0): device state change: 2
 -> 1
Nov 29 11:44:14 portatile NetworkManager: <info> (eth0): cleaning up...
Nov 29 11:44:14 portatile NetworkManager: <info> (eth0): taking down device.
Nov 29 11:44:14 portatile bluetoothd[5511]: bridge pan0 removed
Nov 29 11:44:14 portatile bluetoothd[5511]: Stopping SDP server
Nov 29 11:44:14 portatile bluetoothd[5511]: Exit
Nov 29 11:44:14 portatile kernel: klogd 1.5.0, ---------- state change ---------
-
Nov 29 11:44:14 portatile kernel: Kernel logging re-started after SIGSTOP.
Nov 29 11:44:16 portatile kernel: klogd 1.5.0, ---------- state change ---------
-
Nov 29 11:44:16 portatile kernel: Kernel logging re-started after SIGSTOP.
Nov 29 11:44:16 portatile NetworkManager: <info> Unmanaged Device found; state
CONNECTED forced. (see http://bugs.launchpad.net/bugs/191889)
Nov 29 11:44:16 portatile NetworkManager: <info> (wlan0): now unmanaged
Nov 29 11:44:16 portatile NetworkManager: <info> (wlan0): device state change:
8 -> 1
Nov 29 11:44:16 portatile NetworkManager: <info> (wlan0): deactivating device (
reason: 36).
Nov 29 11:44:16 portatile NetworkManager: <info> wlan0: canceled DHCP transacti
on, dhcp client pid 6199
Nov 29 11:44:16 portatile NetworkManager: <info> (wlan0): removing resolv.conf
from /sbin/resolvconf
Nov 29 11:44:16 portatile NetworkManager: <info> (wlan0): cleanin...

Read more...

Revision history for this message
Albert Damen (albrt) wrote :

The lines "Kernel logging re-started after SIGSTOP" are strange.
I have the same messages (with / on ext3), also seeing / is busy when I shutdown or reboot. It seems klogd and syslogd are not terminated anymore, and keep the file system busy.
As a work-around, I have created symlinks in /etc/rc{0,6}.d to terminate klogd and syslogd, and now I can shutdown and reboot without problems. Also, the log lines "Kernel logging re-started after SIGSTOP" have stopped.

Can you please create those symlinks and verify if that solves the issue for you as well?
You can do this with the commands:
$ cd /etc/rc0.d
$ sudo ln -s ../init.d/klogd K89klogd
$ sudo ln -s ../init.d/sysklogd K90sysklogd
$ cd /etc/rc6.d
$ sudo ln -s ../init.d/klogd K89klogd
$ sudo ln -s ../init.d/sysklogd K90sysklogd

Then try a few shutdowns and reboots and see if it helps. Please report back your results.

Revision history for this message
Graziano (graziano-giuliani-gmail) wrote :

That made the trick for me also.
Why are klogd and syslogd not terminated anymore? Debug "feature"?
This bug is to be considered unrelated to both jfs and initscripts, and should be reported in some way to system logging daemons devels.

Thanks to Albert for pointing this out.

Closed for me.

Revision history for this message
Graziano (graziano-giuliani-gmail) wrote :
Revision history for this message
Albert Damen (albrt) wrote :

Thanks for the confirmation Graziano. I will add a proposal for a fix in the new bug.

For future reference, if you find a bug affects a different package then you thought first, you can change the package in the existing bug report. You can do that by pressing the down arrow in the yellow bar, under affects.
I will close this bug as invalid.

Changed in sysvinit:
status: Incomplete → Invalid
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.