Debian Bug report logs - #549260
/etc/init.d/{halt,reboot} need Required-Start: $all

version graph

Package: initscripts; Maintainer for initscripts is Debian sysvinit maintainers <pkg-sysvinit-devel@lists.alioth.debian.org>; Source for initscripts is src:sysvinit.

Reported by: Rik Theys <rik.theys@gmail.com>

Date: Thu, 1 Oct 2009 21:51:01 UTC

Severity: important

Found in version sysvinit/2.87dsf-10

Reply or subscribe to this bug.

Toggle useless messages

View this report as an mbox folder, status mbox, maintainer mbox


Report forwarded to debian-bugs-dist@lists.debian.org, Petter Reinholdtsen <pere@debian.org>:
Bug#549260; Package insserv. (Thu, 01 Oct 2009 21:51:04 GMT) Full text and rfc822 format available.

Acknowledgement sent to Rik Theys <rik.theys@gmail.com>:
New Bug report received and forwarded. Copy sent to Petter Reinholdtsen <pere@debian.org>. (Thu, 01 Oct 2009 21:51:04 GMT) Full text and rfc822 format available.

Message #5 received at submit@bugs.debian.org (full text, mbox):

From: Rik Theys <rik.theys@gmail.com>
To: Debian Bug Tracking System <submit@bugs.debian.org>
Subject: insserv: no longer unmounts (root) filesystem on shutdown/reboot
Date: Thu, 01 Oct 2009 23:44:02 +0200
Package: insserv
Version: 1.12.0-13
Severity: critical
Justification: causes serious data loss

Subject: insserv: system no longer unmounts filesystems on reboot/shutdown
Package: insserv
Version: 1.12.0-13
Justification: causes non-serious data loss
Severity: grave


After switching to insserv, my system no longer seems to unmount the (local) filesystems on shutdown and reboot.
Each time the system comes back up, it spawns a recovery shell and asks to fsck the root filesystem.


-- System Information:
Debian Release: squeeze/sid
  APT prefers testing
  APT policy: (500, 'testing'), (500, 'stable')
Architecture: amd64 (x86_64)

Kernel: Linux 2.6.30-1-amd64 (SMP w/2 CPU cores)
Locale: LANG=en_US.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8)
Shell: /bin/sh linked to /bin/dash

Versions of packages insserv depends on:
ii  libc6                         2.9-25     GNU C Library: Shared libraries

insserv recommends no packages.

Versions of packages insserv suggests:
pn  bootchart                     <none>     (no description available)

-- no debconf information

-- System Information:
Debian Release: squeeze/sid
  APT prefers testing
  APT policy: (500, 'testing'), (500, 'stable')
Architecture: amd64 (x86_64)

Kernel: Linux 2.6.30-1-amd64 (SMP w/2 CPU cores)
Locale: LANG=en_US.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8)
Shell: /bin/sh linked to /bin/dash

Versions of packages insserv depends on:
ii  libc6                         2.9-25     GNU C Library: Shared libraries

insserv recommends no packages.

Versions of packages insserv suggests:
pn  bootchart                     <none>     (no description available)

-- no debconf information




Information forwarded to debian-bugs-dist@lists.debian.org, Petter Reinholdtsen <pere@debian.org>:
Bug#549260; Package insserv. (Thu, 01 Oct 2009 22:12:05 GMT) Full text and rfc822 format available.

Acknowledgement sent to Petter Reinholdtsen <pere@hungry.com>:
Extra info received and forwarded to list. Copy sent to Petter Reinholdtsen <pere@debian.org>. (Thu, 01 Oct 2009 22:12:05 GMT) Full text and rfc822 format available.

Message #10 received at 549260@bugs.debian.org (full text, mbox):

From: Petter Reinholdtsen <pere@hungry.com>
To: Rik Theys <rik.theys@gmail.com>, 549260@bugs.debian.org
Subject: Re: Bug#549260: insserv: no longer unmounts (root) filesystem on shutdown/reboot
Date: Fri, 2 Oct 2009 00:05:43 +0200
[Rik Theys]
> After switching to insserv, my system no longer seems to unmount the
> (local) filesystems on shutdown and reboot.  Each time the system
> comes back up, it spawns a recovery shell and asks to fsck the root
> filesystem.

Can you provide the output from 'ls /etc/rcS.d/ /etc/rc6.d/'?

This sound a bit like the problem reported in #543375.  Could it be
the same problem?  Do you use the default Debian kernel, or a
home-made one?

Happy hacking,
-- 
Petter Reinholdtsen




Information forwarded to debian-bugs-dist@lists.debian.org, Petter Reinholdtsen <pere@debian.org>:
Bug#549260; Package insserv. (Sat, 03 Oct 2009 18:36:03 GMT) Full text and rfc822 format available.

Acknowledgement sent to Petter Reinholdtsen <pere@hungry.com>:
Extra info received and forwarded to list. Copy sent to Petter Reinholdtsen <pere@debian.org>. (Sat, 03 Oct 2009 18:36:03 GMT) Full text and rfc822 format available.

Message #15 received at 549260@bugs.debian.org (full text, mbox):

From: Petter Reinholdtsen <pere@hungry.com>
To: 549260@bugs.debian.org, Rik Theys <rik.theys@gmail.com>
Subject: Re: Bug#549260: insserv: no longer unmounts (root) filesystem on shutdown/reboot
Date: Sat, 3 Oct 2009 20:26:23 +0200
severity 549260 important
thanks

I'm unable to reproduce this in my test environment, so without more
information, I assume it only affect users with some combination of
packages with incorrect init.d script headers or is actually a problem
with a different package.  Because of this, I reduce the severity to
important.  Please provide output from the ls command I asked for, and
check if the problem is the util-linux one.

Happy hacking,
-- 
Petter Reinholdtsen




Severity set to 'important' from 'critical' Request was from Petter Reinholdtsen <pere@hungry.com> to control@bugs.debian.org. (Sat, 03 Oct 2009 18:36:04 GMT) Full text and rfc822 format available.

Information forwarded to debian-bugs-dist@lists.debian.org, Petter Reinholdtsen <pere@debian.org>:
Bug#549260; Package insserv. (Sat, 03 Oct 2009 21:15:15 GMT) Full text and rfc822 format available.

Acknowledgement sent to Rik Theys <rik.theys@gmail.com>:
Extra info received and forwarded to list. Copy sent to Petter Reinholdtsen <pere@debian.org>. (Sat, 03 Oct 2009 21:15:15 GMT) Full text and rfc822 format available.

Message #22 received at 549260@bugs.debian.org (full text, mbox):

From: Rik Theys <rik.theys@gmail.com>
To: Petter Reinholdtsen <pere@hungry.com>
Cc: 549260@bugs.debian.org
Subject: Re: Bug#549260: insserv: no longer unmounts (root) filesystem on shutdown/reboot
Date: Sat, 3 Oct 2009 23:06:17 +0200
[Message part 1 (text/plain, inline)]
Hi,

The kernel used is the default current squeeze kernel.

Here's the output from the ls command:

/etc/rc6.d/:
K01alsa-utils    K01fuse         K01network-manager  K01postfix
K03mysql-ndb      S01umountnfs.sh      S16auditd
K01anacron       K01irqbalance   K01ntp              K01rsync
K04mysql-ndb-mgm  S01umountroot        S16networking
K01atd           K01joystick     K01nvidia-glx       K01saned
K06hwclock.sh     S09cryptdisks-early  S17wpa-ifupdown
K01bluetooth     K01kdm          K01nvidia-kernel    K01stunnel4
README            S10lvm2              S21urandom
K01cpufrequtils  K01kvm          K01openbsd-inetd    K01uml-utilities
S01reboot         S11cryptdisks
K01cron          K01laptop-mode  K01openvpn          K02avahi-daemon
S01sendsigs       S13ifupdown
K01dirmngr       K01libvirt-bin  K01pcscd            K02mysql
S01umountfs       S14rsyslog

/etc/rcS.d/:
README               S07hdparm            S10ifupdown-clean
S13ifupdown               S16pppd-dns               S21alsa-utils
S01mountkernfs.sh    S07hostname.sh       S10lvm2
S13mountall.sh            S16procps                 S21bootmisc.sh
S02udev              S07hwclockfirst.sh   S10module-init-tools
S14mountall-bootclean.sh  S16x11-common             S21fuse
S03mountdevsubfs.sh  S07pcmciautils       S10mtab.sh
S15mountoverflowtmp       S17mountnfs.sh            S21screen-cleanup
S04bootlogd          S08checkroot.sh      S10udev-mtab
S16auditd                 S18mountnfs-bootclean.sh  S21urandom
S05keymap.sh         S09cryptdisks-early  S11cryptdisks
S16networking             S19console-screen.sh      S22stop-bootlogd-single
S06keyboard-setup    S10hwclock.sh        S12checkfs.sh
S16policykit              S20console-setup

I'll check and see if it's the same bug as #543375.

Regards,

Rik


On Fri, Oct 2, 2009 at 12:05 AM, Petter Reinholdtsen <pere@hungry.com>wrote:

> [Rik Theys]
> > After switching to insserv, my system no longer seems to unmount the
> > (local) filesystems on shutdown and reboot.  Each time the system
> > comes back up, it spawns a recovery shell and asks to fsck the root
> > filesystem.
>
> Can you provide the output from 'ls /etc/rcS.d/ /etc/rc6.d/'?
>
> This sound a bit like the problem reported in #543375.  Could it be
> the same problem?  Do you use the default Debian kernel, or a
> home-made one?
>
> Happy hacking,
> --
> Petter Reinholdtsen
>



-- 

Greetings,

Rik
[Message part 2 (text/html, inline)]

Information forwarded to debian-bugs-dist@lists.debian.org, Petter Reinholdtsen <pere@debian.org>:
Bug#549260; Package insserv. (Sat, 03 Oct 2009 21:33:06 GMT) Full text and rfc822 format available.

Acknowledgement sent to Petter Reinholdtsen <pere@hungry.com>:
Extra info received and forwarded to list. Copy sent to Petter Reinholdtsen <pere@debian.org>. (Sat, 03 Oct 2009 21:33:06 GMT) Full text and rfc822 format available.

Message #27 received at 549260@bugs.debian.org (full text, mbox):

From: Petter Reinholdtsen <pere@hungry.com>
To: Rik Theys <rik.theys@gmail.com>, 549260@bugs.debian.org
Subject: Re: Bug#549260: insserv: no longer unmounts (root) filesystem on shutdown/reboot
Date: Sat, 3 Oct 2009 23:18:24 +0200
[Rik Theys]
> Here's the output from the ls command:
> 
> /etc/rc6.d/:
> K01alsa-utils    K01fuse         K01network-manager  K01postfix K03mysql-ndb      S01umountnfs.sh      S16auditd
> K01anacron       K01irqbalance   K01ntp              K01rsync   K04mysql-ndb-mgm  S01umountroot        S16networking
> K01atd           K01joystick     K01nvidia-glx       K01saned   K06hwclock.sh     S09cryptdisks-early  S17wpa-ifupdown
> K01bluetooth     K01kdm          K01nvidia-kernel    K01stunnel4 README           S10lvm2              S21urandom
> K01cpufrequtils  K01kvm          K01openbsd-inetd    K01uml-utilities S01reboot         S11cryptdisks K01cron          K01laptop-mode  K01openvpn          K02avahi-daemon S01sendsigs       S13ifupdown
> K01dirmngr       K01libvirt-bin  K01pcscd            K02mysql S01umountfs       S14rsyslog

This ordering is very strange.  How did you enable dependency based
boot sequencing?  Was this done using the insserv or sysv-rc package
scripts, or some other way?  When done the normal way, all S* symlinks
in rc0.d and rc6.d would be changed to K* symlinks to reflect how
scripts are actually called in these runlevels.

Happy hacking,
--
Petter Reinholdtsen




Information forwarded to debian-bugs-dist@lists.debian.org, Petter Reinholdtsen <pere@debian.org>:
Bug#549260; Package insserv. (Sun, 04 Oct 2009 05:57:06 GMT) Full text and rfc822 format available.

Acknowledgement sent to Petter Reinholdtsen <pere@hungry.com>:
Extra info received and forwarded to list. Copy sent to Petter Reinholdtsen <pere@debian.org>. (Sun, 04 Oct 2009 05:57:06 GMT) Full text and rfc822 format available.

Message #32 received at 549260@bugs.debian.org (full text, mbox):

From: Petter Reinholdtsen <pere@hungry.com>
To: Rik Theys <rik.theys@gmail.com>
Cc: 549260@bugs.debian.org
Subject: Re: Bug#549260: insserv: no longer unmounts (root) filesystem on shutdown/reboot
Date: Sun, 4 Oct 2009 07:51:40 +0200
[Rik Theys]
> I believe /the inserv installation complained it was not possible to enable
> dependency based boot because there were 3 scripts that didn't support it.
> 
> I removed the init scripts in question and then ran
> 
> dpkg-reconfigure insserv
> insserv
> 
> (if I remember correctly).
> 
> How can I reconfigure insserv to make the correct changes?

Right.  Then I understand what happened.  The insserv call reordered
the boot and shutdown scripts but the migration was not done as it was
intended by the package scripts of sysv-rc or insserv (depending on
version), so the shutdown sequences became wrong.  Where did you read
that you should run insserv manually?  I ask because I want to track
down and update any inaccurate documentation.

> I've now run dpkg-reconfigure -plow sysv-rc and it has activated dependency
> based boot and renamed the links in rc0.d

Good.

> The issue with X/kdm remains: it is started on boot and I see the
> nvidia logo appear but then I get the console again and X is no
> longer running. If I login and restart kdm, X comes up and stays up.
>
> Any ideas?

I would guess some related init.d script is missing start
dependencies, but it is hard to give a more concrete proposal without
the output from 'ls /etc/rcS.d /etc/rc2.d'.

Happy hacking,
-- 
Petter Reinholdtsen




Information forwarded to debian-bugs-dist@lists.debian.org, Petter Reinholdtsen <pere@debian.org>:
Bug#549260; Package insserv. (Tue, 06 Oct 2009 06:42:02 GMT) Full text and rfc822 format available.

Acknowledgement sent to Rik Theys <rik.theys@gmail.com>:
Extra info received and forwarded to list. Copy sent to Petter Reinholdtsen <pere@debian.org>. (Tue, 06 Oct 2009 06:42:03 GMT) Full text and rfc822 format available.

Message #37 received at 549260@bugs.debian.org (full text, mbox):

From: Rik Theys <rik.theys@gmail.com>
To: 549260@bugs.debian.org
Subject: Fwd: Bug#549260: insserv: no longer unmounts (root) filesystem on shutdown/reboot
Date: Tue, 6 Oct 2009 08:21:05 +0200
[Message part 1 (text/plain, inline)]
forgot to CC the bug report.

---------- Forwarded message ----------
From: Rik Theys <rik.theys@gmail.com>
Date: Mon, Oct 5, 2009 at 11:11 PM
Subject: Re: Bug#549260: insserv: no longer unmounts (root) filesystem on
shutdown/reboot
To: Petter Reinholdtsen <pere@hungry.com>




On Sun, Oct 4, 2009 at 7:51 AM, Petter Reinholdtsen <pere@hungry.com> wrote:

>
> Where did you read
> that you should run insserv manually?  I ask because I want to track
> down and update any inaccurate documentation.
>
> I didn't read this anywhere, I just tried it after reading the insserv man
page. I also couldn't find any info in /usr/share/doc/insserv.

> I've now run dpkg-reconfigure -plow sysv-rc and it has activated
> dependency
> > based boot and renamed the links in rc0.d
>
> Good.
>
> > The issue with X/kdm remains: it is started on boot and I see the
> > nvidia logo appear but then I get the console again and X is no
> > longer running. If I login and restart kdm, X comes up and stays up.
> >
> > Any ideas?
>
> I would guess some related init.d script is missing start
> dependencies, but it is hard to give a more concrete proposal without
> the output from 'ls /etc/rcS.d /etc/rc2.d'.
>
> Here's the output from the ls command:

 /etc/rc2.d:
README           S01nvidia-glx     S02anacron       S02irqbalance
S02rsync         S03hal          S04mysql            S06rc.local
S01acpi-support  S01nvidia-kernel  S02apmd          S02mcelog
S02ssh           S03mysql-ndb    S04network-manager  S06rmnologin
S01dirmngr       S01rsyslog        S02atd           S02mysql-ndb-mgm
S02stunnel4      S03pulseaudio   S04saned            S06stop-bootlogd
S01joystick      S01sudo           S02cpufrequtils  S02ntp
S02sysstat       S04cups         S05bootlogs
S01kvm           S01uml-utilities  S02cron          S02openbsd-inetd
S03avahi-daemon  S04kdm          S05postfix
S01loadcpufreq   S02acpid          S02dbus          S02openvpn
S03bluetooth     S04libvirt-bin  S06laptop-mode

/etc/rcS.d:
README               S07hdparm            S10ifupdown-clean
S13ifupdown               S16pppd-dns               S21alsa-utils
S01mountkernfs.sh    S07hostname.sh       S10lvm2
S13mountall.sh            S16procps                 S21bootmisc.sh
S02udev              S07hwclockfirst.sh   S10module-init-tools
S14mountall-bootclean.sh  S16x11-common             S21fuse
S03mountdevsubfs.sh  S07pcmciautils       S10mtab.sh
S15mountoverflowtmp       S17mountnfs.sh            S21screen-cleanup
S04bootlogd          S08checkroot.sh      S10udev-mtab
S16auditd                 S18mountnfs-bootclean.sh  S21urandom
S05keymap.sh         S09cryptdisks-early  S11cryptdisks
S16networking             S19console-screen.sh      S22stop-bootlogd-single
S06keyboard-setup    S10hwclock.sh        S12checkfs.sh
S16policykit              S20console-setup

Looking at the X log file, the log file from the failed startup stops where
the working X log continues to process input devices from HAL. Although hal
is started before kdm. Maybe it doesn't start fast enough?

Regards,

Rik





-- 

Greetings,

Rik
[Message part 2 (text/html, inline)]

Information forwarded to debian-bugs-dist@lists.debian.org, Petter Reinholdtsen <pere@debian.org>:
Bug#549260; Package insserv. (Tue, 06 Oct 2009 10:24:03 GMT) Full text and rfc822 format available.

Acknowledgement sent to Petter Reinholdtsen <pere@hungry.com>:
Extra info received and forwarded to list. Copy sent to Petter Reinholdtsen <pere@debian.org>. (Tue, 06 Oct 2009 10:24:05 GMT) Full text and rfc822 format available.

Message #42 received at 549260@bugs.debian.org (full text, mbox):

From: Petter Reinholdtsen <pere@hungry.com>
To: Rik Theys <rik.theys@gmail.com>, 549260@bugs.debian.org
Subject: Re: Bug#549260: Fwd: Bug#549260: insserv: no longer unmounts (root) filesystem on shutdown/reboot
Date: Tue, 6 Oct 2009 11:39:01 +0200
[Rik Theys]
>  /etc/rc2.d:

Reformattet for easier reading:

S01acpi-support
S01dirmngr
S01joystick
S01kvm
S01loadcpufreq
S01nvidia-glx
S01nvidia-kernel
S01rsyslog
S01sudo
S01uml-utilities
S02acpid
S02anacron
S02apmd
S02atd
S02cpufrequtils
S02cron
S02dbus
S02irqbalance
S02mcelog
S02mysql-ndb-mgm
S02ntp
S02openbsd-inetd
S02openvpn
S02rsync
S02ssh
S02stunnel4
S02sysstat
S03avahi-daemon
S03bluetooth
S03hal
S03mysql-ndb
S03pulseaudio
S04cups
S04kdm
S04libvirt-bin
S04mysql
S04network-manager
S04saned
S05bootlogs
S05postfix
S06laptop-mode
S06rc.local
S06rmnologin
S06stop-bootlogd

This ordering is correct as far as I know.  hal starts before kdm, and
nothing starting after kdm should affect its behaviour.

> Looking at the X log file, the log file from the failed startup
> stops where the working X log continues to process input devices
> from HAL. Although hal is started before kdm. Maybe it doesn't start
> fast enough?

That might very well be the case.  I suspect we might discover this
kind of problems more when we speed up the boot, ie latent race
conditions which have been hidden because the old boot was slow.

Does it help to add a sleep command to the end of the hal script?  For
the booting to work, hal need to be fully operational before its
init.d script exits, as the later scripts may expect hal to be
working.

I suspect this bug should be about document the requirement of
enabling insserv, and that a new BTS report should be created to
adressed the failing kdm issue.

Happy hacking,
-- 
Petter Reinholdtsen




Information forwarded to debian-bugs-dist@lists.debian.org, Petter Reinholdtsen <pere@debian.org>:
Bug#549260; Package insserv. (Tue, 06 Oct 2009 13:45:06 GMT) Full text and rfc822 format available.

Acknowledgement sent to Rik Theys <rik.theys@gmail.com>:
Extra info received and forwarded to list. Copy sent to Petter Reinholdtsen <pere@debian.org>. (Tue, 06 Oct 2009 13:45:06 GMT) Full text and rfc822 format available.

Message #47 received at 549260@bugs.debian.org (full text, mbox):

From: Rik Theys <rik.theys@gmail.com>
To: Petter Reinholdtsen <pere@hungry.com>
Cc: 549260@bugs.debian.org
Subject: Re: Bug#549260: Fwd: Bug#549260: insserv: no longer unmounts (root) filesystem on shutdown/reboot
Date: Tue, 6 Oct 2009 15:40:18 +0200
[Message part 1 (text/plain, inline)]
Hi,

On Tue, Oct 6, 2009 at 11:39 AM, Petter Reinholdtsen <pere@hungry.com>wrote:

> This ordering is correct as far as I know.  hal starts before kdm, and
> nothing starting after kdm should affect its behaviour.
>
> > Looking at the X log file, the log file from the failed startup
> > stops where the working X log continues to process input devices
> > from HAL. Although hal is started before kdm. Maybe it doesn't start
> > fast enough?
>
> That might very well be the case.  I suspect we might discover this
> kind of problems more when we speed up the boot, ie latent race
> conditions which have been hidden because the old boot was slow.
>
> Does it help to add a sleep command to the end of the hal script?  For
> the booting to work, hal need to be fully operational before its
> init.d script exits, as the later scripts may expect hal to be
> working.
>

I'll test that later this week.


>
> I suspect this bug should be about document the requirement of
> enabling insserv, and that a new BTS report should be created to
> adressed the failing kdm issue.
>

OK. Can you clone and retitle this bug, or should I file a new bug report?

Regards,

Rik
[Message part 2 (text/html, inline)]

Information forwarded to debian-bugs-dist@lists.debian.org, Petter Reinholdtsen <pere@debian.org>:
Bug#549260; Package insserv. (Tue, 06 Oct 2009 20:12:07 GMT) Full text and rfc822 format available.

Acknowledgement sent to Rik Theys <rik.theys@gmail.com>:
Extra info received and forwarded to list. Copy sent to Petter Reinholdtsen <pere@debian.org>. (Tue, 06 Oct 2009 20:12:07 GMT) Full text and rfc822 format available.

Message #52 received at 549260@bugs.debian.org (full text, mbox):

From: Rik Theys <rik.theys@gmail.com>
To: Petter Reinholdtsen <pere@hungry.com>
Cc: 549260@bugs.debian.org
Subject: Re: Bug#549260: Fwd: Bug#549260: insserv: no longer unmounts (root) filesystem on shutdown/reboot
Date: Tue, 6 Oct 2009 22:05:47 +0200
On Tuesday 06 October 2009 11:39:01 Petter Reinholdtsen wrote:
> > Looking at the X log file, the log file from the failed startup
> > stops where the working X log continues to process input devices
> > from HAL. Although hal is started before kdm. Maybe it doesn't start
> > fast enough?
> 
> That might very well be the case.  I suspect we might discover this
> kind of problems more when we speed up the boot, ie latent race
> conditions which have been hidden because the old boot was slow.
> 
> Does it help to add a sleep command to the end of the hal script?  For
> the booting to work, hal need to be fully operational before its
> init.d script exits, as the later scripts may expect hal to be
> working.

Adding a sleep 10 at the end of the hal init script doesn't fix the issue.
Manually reordering the kdm init script in rc2.d so it starts after 
networkmanager seems to fix the problem.

My working rc2.d (I named the kdm symlink aakdm):

rik@jupiter:~$ ls -1 /etc/rc2.d
README                         
S01acpi-support                
S01dirmngr                     
S01joystick                    
S01kvm                         
S01loadcpufreq                 
S01nvidia-glx
S01nvidia-kernel
S01rsyslog
S01sudo
S01uml-utilities
S02acpid
S02anacron
S02apmd
S02atd
S02cpufrequtils
S02cron
S02dbus
S02irqbalance
S02mcelog
S02mysql-ndb-mgm
S02ntp
S02openbsd-inetd
S02openvpn
S02rsync
S02ssh
S02stunnel4
S02sysstat
S03avahi-daemon
S03bluetooth
S03hal
S03mysql-ndb
S03pulseaudio
S04cups
S04libvirt-bin
S04mysql
S04network-manager
S04saned
S05aakdm
S05bootlogs
S05postfix
S06laptop-mode
S06rc.local
S06rmnologin
S06stop-bootlogd

Rik




Information forwarded to debian-bugs-dist@lists.debian.org, Petter Reinholdtsen <pere@debian.org>:
Bug#549260; Package insserv. (Fri, 23 Apr 2010 16:21:03 GMT) Full text and rfc822 format available.

Acknowledgement sent to "Chanoch (Ken) Bloom" <kbloom@gmail.com>:
Extra info received and forwarded to list. Copy sent to Petter Reinholdtsen <pere@debian.org>. (Fri, 23 Apr 2010 16:21:03 GMT) Full text and rfc822 format available.

Message #57 received at 549260@bugs.debian.org (full text, mbox):

From: "Chanoch (Ken) Bloom" <kbloom@gmail.com>
To: 549260@bugs.debian.org
Subject: Problem is in rc6.d
Date: Fri, 23 Apr 2010 11:04:29 -0500
The problem is with the initscripts
/etc/init.d/reboot and /etc/init.d/halt

/etc/init.d/reboot has no dependencies, so insserv assigns it to
S01reboot (in runlevel 6). It therefore alphabetizes before all of the
other scripts (particularly S01umountfs, S01umountnfs, S01umountroot)
and the computer reboots before its filesystems are unmounted.

/etc/init.d/reboot likewise has no dependencies, so insserv assigns it
to S01halt (in runlevel 0). The same symptoms ensue.

--Ken

-- 
Chanoch (Ken) Bloom. PhD candidate. Linguistic Cognition Laboratory.
Department of Computer Science. Illinois Institute of Technology.
http://www.iit.edu/~kbloom1/




Information forwarded to debian-bugs-dist@lists.debian.org, Petter Reinholdtsen <pere@debian.org>:
Bug#549260; Package insserv. (Fri, 23 Apr 2010 16:21:05 GMT) Full text and rfc822 format available.

Acknowledgement sent to "Chanoch (Ken) Bloom" <kbloom@gmail.com>:
Extra info received and forwarded to list. Copy sent to Petter Reinholdtsen <pere@debian.org>. (Fri, 23 Apr 2010 16:21:05 GMT) Full text and rfc822 format available.

Message #62 received at 549260@bugs.debian.org (full text, mbox):

From: "Chanoch (Ken) Bloom" <kbloom@gmail.com>
To: 549260@bugs.debian.org, control@bugs.debian.org
Subject: Fix:
Date: Fri, 23 Apr 2010 11:17:16 -0500
reassign 549260 initscripts 2.87dsf-10
retitle 549260 /etc/init.d/{halt,reboot} need Required-Start: $all
affects 549260 insserv
thanks

The fix for this problem is to add Required-Start: $all to
/etc/init.d/halt and /etc/init.d/reboot so that they run after
/etc/init.d/umountfs, /etc/init.d/umountroot, and any other scripts
that are supposed to be started to clean things up at shutdown.

As these scripts are in initscripts, I'm reassigning there.

-- 
Chanoch (Ken) Bloom. PhD candidate. Linguistic Cognition Laboratory.
Department of Computer Science. Illinois Institute of Technology.
http://www.iit.edu/~kbloom1/




Bug reassigned from package 'insserv' to 'initscripts'. Request was from "Chanoch (Ken) Bloom" <kbloom@gmail.com> to control@bugs.debian.org. (Fri, 23 Apr 2010 16:30:03 GMT) Full text and rfc822 format available.

Bug No longer marked as found in versions insserv/1.12.0-13. Request was from "Chanoch (Ken) Bloom" <kbloom@gmail.com> to control@bugs.debian.org. (Fri, 23 Apr 2010 16:30:03 GMT) Full text and rfc822 format available.

Bug Marked as found in versions sysvinit/2.87dsf-10. Request was from "Chanoch (Ken) Bloom" <kbloom@gmail.com> to control@bugs.debian.org. (Fri, 23 Apr 2010 16:30:04 GMT) Full text and rfc822 format available.

Changed Bug title to '/etc/init.d/{halt,reboot} need Required-Start: $all' from 'insserv: no longer unmounts (root) filesystem on shutdown/reboot' Request was from "Chanoch (Ken) Bloom" <kbloom@gmail.com> to control@bugs.debian.org. (Fri, 23 Apr 2010 16:30:05 GMT) Full text and rfc822 format available.

Added indication that 549260 affects insserv Request was from "Chanoch (Ken) Bloom" <kbloom@gmail.com> to control@bugs.debian.org. (Fri, 23 Apr 2010 16:30:05 GMT) Full text and rfc822 format available.

Information forwarded to debian-bugs-dist@lists.debian.org, Debian sysvinit maintainers <pkg-sysvinit-devel@lists.alioth.debian.org>:
Bug#549260; Package initscripts. (Fri, 23 Apr 2010 17:09:03 GMT) Full text and rfc822 format available.

Acknowledgement sent to Petter Reinholdtsen <pere@hungry.com>:
Extra info received and forwarded to list. Copy sent to Debian sysvinit maintainers <pkg-sysvinit-devel@lists.alioth.debian.org>. (Fri, 23 Apr 2010 17:09:03 GMT) Full text and rfc822 format available.

Message #77 received at 549260@bugs.debian.org (full text, mbox):

From: Petter Reinholdtsen <pere@hungry.com>
To: "Chanoch (Ken) Bloom" <kbloom@gmail.com>, 549260@bugs.debian.org
Subject: Re: Bug#549260: Fix:
Date: Fri, 23 Apr 2010 19:06:41 +0200
[Chanoch (Ken) Bloom]
> The fix for this problem is to add Required-Start: $all to
> /etc/init.d/halt and /etc/init.d/reboot so that they run after
> /etc/init.d/umountfs, /etc/init.d/umountroot, and any other scripts
> that are supposed to be started to clean things up at shutdown.

Actually, that is the wrong fix.  The correct fix is to rename all S*
files in rc0.d and rc6.d to K*, and rerun update-rc.d to correctly
order the shutdown scripts.

How was this machine migrated to dependency based boot sequencing?
The process can not have been done correctly, as the sysv-rc migration
script should have taken care of this renaming.

Happy hacking,
-- 
Petter Reinholdtsen




Information forwarded to debian-bugs-dist@lists.debian.org, Debian sysvinit maintainers <pkg-sysvinit-devel@lists.alioth.debian.org>:
Bug#549260; Package initscripts. (Mon, 13 Dec 2010 21:57:06 GMT) Full text and rfc822 format available.

Acknowledgement sent to Steven Chamberlain <steven@pyro.eu.org>:
Extra info received and forwarded to list. Copy sent to Debian sysvinit maintainers <pkg-sysvinit-devel@lists.alioth.debian.org>. (Mon, 13 Dec 2010 21:57:07 GMT) Full text and rfc822 format available.

Message #82 received at 549260@bugs.debian.org (full text, mbox):

From: Steven Chamberlain <steven@pyro.eu.org>
To: 549260@bugs.debian.org
Cc: pere@hungry.com, kbloom@gmail.com
Subject: Re: Bug#549260: insserv: no longer unmounts (root) filesystem on shutdown/reboot
Date: Mon, 13 Dec 2010 21:52:18 +0000
Hi,

I just reproduced this bug on a Debian lenny machine.  For some silly 
reason I decided to do the following:

# apt-get install insserv

# insserv
insserv: warning: current start runlevel(s) (0) of script `halt' 
overwrites defaults (empty).
insserv: warning: current start runlevel(s) (0 6) of script `umountfs' 
overwrites defaults (empty).
insserv: warning: current start runlevel(s) (6) of script `reboot' 
overwrites defaults (empty).
insserv: warning: current start runlevel(s) (0 6) of script `sendsigs' 
overwrites defaults (empty).
insserv: warning: current start runlevel(s) (0 6) of script `umountroot' 
overwrites defaults (empty).
insserv: warning: current start runlevel(s) (0 6) of script 
`umountnfs.sh' overwrites defaults (empty).

# ls /etc/rc0.d
K01acct
K01apcupsd
K01arpwatch
K01atd
K01edac
K01etc-setserial
K01exim4
K01fancontrol
K01ipmievd
K01mdadm
K01mrd6
K01nfs-kernel-server
K01ntp
K01openbsd-inetd
K01radvd
K01rng-tools
K01rsync
K01setserial
K01smartmontools
K01tftpd-hpa
K01vz
K01watchdog
K01wd_keepalive
K01xinetd
K02nfs-common
K03rsyslog
K03syslog-ng
K04hwclock.sh
K05mountoverflowtmp
README
S01halt
S01sendsigs
S01umountfs
S01umountnfs.sh
S01umountroot
S07mdadm-raid
S09cryptdisks-early
S10ifupdown
S10lvm2
S11cryptdisks
S16networking
S16urandom
S17portmap


It's lucky I checked to see what the result was of running insserv, 
before rebooting or shutting down the machine.  I suspect it would have 
caused a RAID desync followed by a fsck and who knows what else.

Adding Required-Start: $all to /etc/init.d/{halt,reboot} and re-running 
insserv *appeared* to fix the most obvious issue.  But I think I know 
why that's wrong;  the other initscripts should instead have the 
appropriate Required-Stop entries to make sure they get run before their 
dependencies (eg. local/remote filesystems, syslog).  And then it would 
be *implied* that 'halt' or 'reboot' must run last.

But some initscripts in lenny perhaps predate dependency-based booting 
so not all of them have the appropriate entries there.  This means that 
the ordering of my startup scripts is probably messed up now too. 
Fortunately I keep backups.

I checked, and upgrading to the squeeze versions of insserv, sysvinit, 
sysv-rc and initscripts does correctly rename the contents of 
/etc/rc{0,6}.d from S* to K*.  To completely fix the mess I made by 
running 'insserv' without arguments on a lenny system, I think a full 
'apt-get dist-upgrade' to squeeze might work.  In fact it may be a good 
time for me to upgrade to squeeze anyway.

I think the advice here would be 'DON'T RUN INSSERV WITHOUT ARGUMENTS! 
At least on a lenny install'.  But how many Linux commands ultimately 
lead to data loss / filesystem corruption by running them without any 
arguments, especially without using something like a '--force' option?

Regards,
-- 
Steven Chamberlain
steven@pyro.eu.org




Send a report that this bug log contains spam.


Debian bug tracking system administrator <owner@bugs.debian.org>. Last modified: Mon Apr 21 16:42:03 2014; Machine Name: buxtehude.debian.org

Debian Bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.