long delay at shutdown/reboot - network-manager doesn't close correctly

Status changed to 'Confirmed' because the bug affects multiple users.

I can confirm this bug.

After this:
"killing all remaining processes ... [fail]"
there is this warning:
"nmdispatcher.action caught signal 15 shutting down ..."

This happens with all my three different (64-bit, 32-bit, Nvidia, AMD, Intel) setups.

Also confirming the workaround
adding the line:
stop on runlevel [6]
into the file /etc/init/network-manager.conf
fixes this issue.

Confirm the same issue on oneiric i386 logged as gnome-classic

* Asking all remaining processes to terminate... [OK]
* Killing all remaining processes... [fail]

asking for remaining processes seems to hang (maybe a timeout)

If I insert "stop on runlevel [6]" into /etc/init/network-manager.conf on a Lenovo S10-3t,
I get many error messages after Killing all remaining..:

modem-manager[n]: <info> Modem Manager (version 0.5) starting...
modem-manager[n]: Could not get the system bus. Make sure the message bus daemon is running! Message: Failed to connect to socket /var/run/dbus/system_bus_socket: Connection refused

I inserted this code into /etc/init/network-manager.conf and it didn't change my shutdown dialogue or speed. I'm running Lubuntu on a Dell Latitude i686 with a SSD, and had the same problem with Xubuntu on my machine prior to this install.

I got the same errors as Harald after trying the above. It was still very slow to shutdown and sometimes didn't shutdown at all.

The only way I could previously get it to shutdown reliably was with a "sudo shutdown -h now".

I have now found if I remove the network manager it completely fixes it. I now have no errors and quick shutdown.

sudo apt-get remove network-manager

If I re-install the above network manager, the problem comes back. So that is definitely the cause in my case.

I have installed "wicd" as a replacement network manager. I did this before removing the Ubuntu supplied network manager.

I am using 11.10 that has been upgraded from 11.04. I have wired ethernet.

In my case replacing line:
stop on stopping dbus
into:
stop on runlevel [06]
into the file /etc/init/network-manager.conf
fixes this problem.

I have the same issue - running Ubuntu 11.10 on a Dell Latitude D830.

The noted work-around above fixes the problem on my installation.

Retraction of my previous comment. I did implement the workaround and it worked once... Ever since I'm having the same issue again.

I had the same issue but I fixed It after adding this line stop on runlevel [6] into the file /etc/init/network-manager.conf.
I've never had this issue since I added the line.

any news on this bug?
if necessary I could make some more tests.. it would be nice to a slick shutdown/restart in Precise!

Lucazade -
you may want to take a look at this report, probably the same issue, a different approach to 'working around' in lieu of any official fix which is not happening as of yet.

Bug 903825

thanks mc3man for the suggestion..
yes, i believe it is the same issue. I'll try that workaround to confirm the bug report.

Changing /etc/init/network-manager.conf as mentioned here did not fully solve the problem for me.
Instead, it led to network-manager complaining about a missing messaging bus (DBus?) during shutdown (which was still too long ;) ).

The workaround mentioned in Bug 903825, however, seems to work perfectly.

Haven't tested yet the workaround 903825 but I can say the workaround I've posted in this bug report works on 3 of the 4 machines I use.

Is there a tag for Quality check to add to the bug report? something like the One Hundred Papercut?
An LTS release with this bug is really sad.

Not to spam this report, but for users having heavy troubles with network-manager, an other solution is:
- purging network-manager
- installing wicd

wicd works so smootly

Dino we are lucky that we have alternatives, and wicd is a good one, but we have to fix
this network manager issue also because it is a regression.

This fixed it for me (I didn't alter the conf)
sudo mv /etc/rc6.d/S35networking /etc/rc6.d/S15networking
sudo mv /etc/rc0.d/S35networking /etc/rc0.d/S15networking

Taken from solution here;
https://bugs.launchpad.net/ubuntu/+source/netbase/+bug/903825

See as well: http://askubuntu.com/questions/67939/why-does-ubuntu-11-10-start-and-shut-down-so-slowly

#19 also did it for me. It is quite annoying to see Ubuntu shutdown almost as long as Windows, so please fix this issue! It is not a big deal changing the filenames, so please update as fast as possible.

#19 works for me too.

Add me to the #19 success list ;)

This bug has been reported on the Ubuntu ISO testing tracker.

A list of all reports related to this bug can be found here:
http://iso.qa.ubuntu.com/qatracker/reports/bugs/869635

So, fixing up the package assignments. There's changes needed in three packages for this, and I'll push the updates shortly:

- NetworkManager should write pid files to /run/sendsigs.omit.d for processes it spawns (dhclient, dnsmasq). These processes will already be stopped by NetworkManager (and otherwise would get respawned during shutdown).

- ModemManager looks like it needs a small update to the upstart file to "stop on stopped network-manager", rather than "stop on stopping network-manager", to avoid it getting respawned via dbus if it's stopped too quickly before NetworkManager has a change to shutdown.

- wpasupplicant should write a pid file to /run/sendsigs.omit.d when spawned via dbus.

On Precise i386, each time i shutdown the system in verbose mode, i can quickly see 2 lines complaining about "failed":

- check for active services (or so) ---> failed
- desactivating swap ----> failed

these lines are always in the 5 latest ones before real shutdown. Dont know if its important or not and related to this bug or not.

This bug was fixed in the package modemmanager - 0.5.1.96+git201202081807.635fce1-0ubuntu2

---------------
modemmanager (0.5.1.96+git201202081807.635fce1-0ubuntu2) precise; urgency=low

  * debian/ubuntu/modemmanager.upstart: fix stop condition to make sure we stop
    only when NetworkManager is stopped as much as possible, to avoid being
    respawned by it via DBus. (LP: #869635, #919071)
 -- Mathieu Trudel-Lapierre <email address hidden> Thu, 16 Feb 2012 10:43:54 -0500

This bug was fixed in the package wpasupplicant - 0.7.3-6ubuntu1

---------------
wpasupplicant (0.7.3-6ubuntu1) precise; urgency=low

  * debian/patches/dbus-activation-cmdline.patch: have wpasupplicant create
    a pid file in /run/sendsigs.omit.d when activated by DBus. (LP: #869635)
 -- Mathieu Trudel-Lapierre <email address hidden> Thu, 16 Feb 2012 16:21:29 -0500

Oopsie, there's also a fix needed for dhclient to actually allow this particular file to be used; adding another task...

Status changed to 'Confirmed' because the bug affects multiple users.

This bug was fixed in the package network-manager - 0.9.2.0+git201202161854.8572ecf-0ubuntu1

---------------
network-manager (0.9.2.0+git201202161854.8572ecf-0ubuntu1) precise; urgency=low

  * upstream snapshot 2012-02-16 18:54:47 (GMT)
    + 8572ecfd7cc823097ff80de54ed6d7e17059c4fb
  * debian/patches/nl3-default-ip6-route.patch: refreshed.
  * debian/patches/manage-privacy-extensions.patch: allow NM to manage privacy
    extensions on a per-connection basis.
  * debian/libnm-util2.symbols: add/update symbols:
    + nm_setting_ip6_config_get_enable_ip6_privacy@Base
    + nm_setting_802_1x_ck_format_get_type@Base
    + nm_setting_802_1x_ck_scheme_get_type@Base
    + nm_setting_bond_add_option@Base
    + nm_setting_bond_get_num_options@Base
    + nm_setting_bond_get_option@Base
    + nm_setting_bond_get_option_by_name@Base
    + nm_setting_bond_remove_option@Base
    + nm_setting_compare_flags_get_type@Base
    + nm_setting_diff_result_get_type@Base
    + nm_setting_gsm_network_band_get_type@Base
    + nm_setting_gsm_network_type_get_type@Base
    + nm_setting_hash_flags_get_type@Base
    + nm_setting_secret_flags_get_type@Base
    + nm_utils_security_type_get_type@Base
    + nm_wep_key_type_get_type@Base
  * debian/libnm-glib4.symbols: add new symbols:
    + nm_client_permission_get_type@Base
    + nm_client_permission_result_get_type@Base
    + nm_secret_agent_get_secrets_flags_get_type@Base
    + nm_wimax_nsp_network_type_get_type@Base
  * debian/patches/add_sendsigs_omissions.patch: move NM pid files for dhclient
    and dnsmasq (and also dhcpcd and named if they were used) to the
    /run/sendsigs.omit.d directory and with a slightly different name. This
    will let us skip over those processes as shutdown since they will be killed
    by NetworkManager when its upstart job stops. (LP: #869635)
  * debian/rules: re-enable tests; run them inside dbus-test-runner.
  * debian/control: Build-Depends on dbus-test-runner for tests.
 -- Mathieu Trudel-Lapierre <email address hidden> Thu, 16 Feb 2012 16:01:07 -0500

This bug was fixed in the package isc-dhcp - 4.1.1-P1-17ubuntu13

---------------
isc-dhcp (4.1.1-P1-17ubuntu13) precise; urgency=low

  * debian/apparmor-profile.dhclient: allow the new pid file for dhclient which
    NetworkManager will use. (LP: #869635)
 -- Mathieu Trudel-Lapierre <email address hidden> Thu, 16 Feb 2012 17:01:19 -0500

Will these fixed backported to oneiric?

Status changed to 'Confirmed' because the bug affects multiple users.

Guess it could be SRU'd to oneiric.

Benjamin, can you confirm similar changes will solve the issues?

A duplicate of this bug (903825) includes a fix which worked perfectly for me in Oneiric:

The solution:

I renamed the links in /etc/rc0.d and rc6.d:

- S31umountnfs.sh to S05umountnfs.sh
- S35networking to S15networking

this bug was solved in Ubuntu but it still present in Kubuntu 12.04.
I had to apply my first workaround (present in original post) to get
a fast shutdown/restart.

Is it possible to fix this permanently also for kubuntu?

For anyone for whom the above solutions didn't work, passing "reboot=pci" as described in this page solved the problem for me http://serverfault.com/questions/126571/system-hangs-while-rebooting-on-debian
this appears to be related to acpi issues in the bios; the solutions above did not work for me.
(Disclaimer: this was on linux-mint (no, not the debian version), not ubuntu. However the bug was identical).

I am on Kubuntu 12.04 and it also affects me still. Any fix for Kubuntu?

actually this is not a bug. I have found the solution to this problem.

You just need to make sure that you have safely removes or eject all the removable drives (USB falsh disk, External Harddisk, etc) before shutting down. Because linux still reads it.

Mouse and external CD/DVD room doesn't have problems.

This way the message ( linux name : xxx tty1 modemmanager(805): caught signal (15)) won't show up at all.

So you don't need to edit conf, remove modemmanager or anything at all.

not a bug? so what is it? a feature??!?
you called it as well "problem" :)

btw the bug is present also without any usb device connected.

@Erie

im sorry to say that your comment is fully wrong here:

- ive no external device plugged on usb
- the mouse is plugged via an adaptor on ps2
- the problem of modemmanger delay only exist with network-manager & network-manager-gnome installed. If they are removed, and wicd installed instead, then the issue is gone.

That is on RR i386 logged as gnome-classic (without effects)

As you can see, that issue is still not fixed:
https://bugs.launchpad.net/ubuntu/+source/modemmanager/+bug/1024868

Whats the point to add a "oneiric" tag is a few days before its EOL ? https://wiki.ubuntu.com/Releases
and all the related packages are set "Fix Released".

I added it because in a few days oneiric will reach EOL, so this bug would automatically be closed

See also LP: #1073433.

Oneiric is now EOL; so new bug might be reported if that trouble still exist.

bug is still here.. but i'm not.. i'm on arch!
so tired to fix this bug in every ubuntu release by hand :(

The bug still exists in ubuntu 13.04

Compaq Evo D5PS
Linux Mint 15 Cinnamon

This happened only once, and it was when shutting down after installing from a USB, also having no internet connection. Just downloaded the iso yesterday using Unetbootin in Windows.

modem-manager (1500) caught signal 15 shutting down

I still have the same issue even on 13.04 with modemmanager 0.6.0.0 and network-manager 0.9.8.0. None of the workarounds worked for me. It’s really annoying.

I am resurrecting this thread, because I am living with this bug since 12.10 (and I am now on 13.10). None of the "workarounds" work *sigh*

I am facing this issue still on 14.10 . I have all the updates pushed by Ubuntu/Canonical. Earlier when I faced this issue, I removed the modem-manager and it worked like charm. This was like 2 weeks back. Now from last 3-4 days the issue is again. I am going to try one of these now:

1) adding the line:
stop on runlevel [6]
into the file /etc/init/network-manager.conf

- purging network-manager
- installing wicd
2)