[SOLVED] systemd stuck 5 min at shutdown (netcfg multi-profile daemon)

stqn · 2012-09-04 21:09:36

Hi, I’m trying to make systemd work and one of my problems is that shutdown hangs for five minutes (really) just after displaying:

[  OK  ] Stopped target Remote File Systems

The next step is apparently “Netcfg multi-profile daemon”.

I followed the instructions to get systemd’s shutdown log, which confirmed that netcfg.service is the culprit, but doesn’t explain what it is waiting for:

(…)
[  228.955139] systemd[1]: openntpd.service: cgroup is empty
[  228.955211] systemd[1]: openntpd.service changed stop-sigterm -> dead
[  228.955266] systemd[1]: Job openntpd.service/stop finished, result=done
[  228.955330] systemd[1]: network.target changed active -> dead
[  228.955333] systemd[1]: Job network.target/stop finished, result=done
[  228.955367] systemd[1]: About to execute: /usr/bin/netcfg-daemon stop
[  228.964029] systemd[1]: Forked /usr/bin/netcfg-daemon as 960
[  228.964072] systemd[1]: netcfg.service changed exited -> stop

(…)

(some debug output from my wireless driver…)

[  229.023442] systemd[1]: Received SIGCHLD from PID 960 (netcfg-daemon).
[  229.023468] systemd[1]: Got SIGCHLD for process 960 (netcfg-daemon)
[  229.023563] systemd[1]: Child 960 died (code=exited, status=0/SUCCESS)
[  229.023567] systemd[1]: Child 960 belongs to netcfg.service
[  229.023574] systemd[1]: netcfg.service: control process exited, code=exited status=0
[  229.023755] systemd[1]: netcfg.service got final SIGCHLD for state stop
[  229.023932] systemd[1]: netcfg.service changed stop -> stop-sigterm
[  229.023980] systemd[1]: Got SIGCHLD for process 993 (wireless)
[  229.024024] systemd[1]: Child 993 died (code=killed, status=9/KILL)

(…)

(some debug output from my wireless driver…)

[  229.024587] systemd[1]: Received SIGCHLD from PID 497 (wpa_supplicant).
[  229.024609] systemd[1]: Got SIGCHLD for process 497 (wpa_supplicant)
[  229.024657] systemd[1]: Child 497 died (code=exited, status=0/SUCCESS)

(…)

(230-231: some debug output from my wireless driver…)

[  319.022963] systemd[1]: netcfg.service stopping timed out. Killing.
[  319.023222] systemd[1]: netcfg.service changed stop-sigterm -> stop-sigkill
[  319.023236] systemd[1]: Running GC...
[  409.022260] systemd[1]: netcfg.service still around after SIGKILL. Ignoring.
[  409.022318] rtw_cmd_thread: leaving... check & free all cmd_obj resources
[  409.022323] rtw_cmd_thread: leaving... call up terminate_cmdthread_sema
[  409.022491] systemd[1]: netcfg.service changed stop-sigkill -> final-sigterm
[  499.021524] systemd[1]: netcfg.service stopping timed out (2). Killing.
[  499.021700] systemd[1]: netcfg.service changed final-sigterm -> failed
[  499.022079] systemd[1]: Job netcfg.service/stop finished, result=done
[  499.022229] systemd[1]: Unit netcfg.service entered failed state.
[  499.022244] systemd[1]: basic.target changed active -> dead

(…)

My wireless driver is 8192cu from the AUR.

/etc/conf.d/netcfg:

NETWORKS=(@myssid)
WIRED_INTERFACE="eth0"
WIRELESS_INTERFACE="wlan0"

/etc/network.d/myssid:

CONNECTION='wireless'
DESCRIPTION='A simple WPA encrypted wireless connection using a static IP'
INTERFACE='wlan0'
SECURITY='wpa'
ESSID='my ssid'
KEY='my key'
IP='static'
ADDR='192.168.0.5'
GATEWAY='192.168.0.254'
DNS=('127.0.0.1')

(comments removed from both files)

Any idea?

Last edited by stqn (2012-11-03 18:07:46)

valr · 2012-09-05 11:22:30

Maybe the netcfg debug option can provide more info ? https://wiki.archlinux.org/index.php/Netcfg#Debugging

stqn · 2012-09-05 19:01:14

Thanks valr. For now I think I solved it by copying /lib/systemd/system/netcfg.service to /etc/systemd/system/ and adding “TimeoutStopSec=1” to the “[Service]” section:

[Unit]
Description=Netcfg multi-profile daemon
Before=network.target
Wants=network.target

[Service]
Type=oneshot
RemainAfterExit=yes
ExecStart=/usr/bin/netcfg-daemon start
ExecStop=/usr/bin/netcfg-daemon stop
TimeoutStopSec=1

[Install]
WantedBy=multi-user.target

This way systemd doesn’t wait 90 s before trying to kill whatever it thinks must be stopped and doesn’t want to exit… but only 1 s.

jouke · 2012-09-05 19:20:35

This smells like a bug in systemd. It looks like it is trying to kill some forked-off process, while Type=oneshot should tell it there is no such thing. A better fix than through TimeoutStopSec=1 would be to add KillMode=none on that place. If this turns out to be a good fix and current systemd versions are still buggy, I will put this in the next netcfg release.

Start your testing!

stqn · 2012-09-05 23:59:03

I tried “KillMode=none” but got the 4 min 30 s wait again.

PS:
systemd 189-4
initscripts 2012.08.3-2
linux 3.5.3-1

Edit: Looks like the only process started by netcfg is wpa_supplicant:

$ systemctl status netcfg
netcfg.service - Netcfg multi-profile daemon
	  Loaded: loaded (/usr/lib/systemd/system/netcfg.service; disabled)
	  Active: active (exited) since Thu, 06 Sep 2012 15:09:48 +0200; 17min ago
	 Process: 327 ExecStart=/usr/bin/netcfg-daemon start (code=exited, status=0/SUCCESS)
	  CGroup: name=systemd:/system/netcfg.service
		  └ 486 wpa_supplicant -B -P /run/wpa_supplicant_wlan0.pid...

According to my log above, wpa_supplicant exits quickly and cleanly, so I don’t know what systemd is trying to kill!

Last edited by stqn (2012-09-06 13:33:42)

daidai67 · 2012-09-06 16:31:22

Just to say I got the exact same problem. Very strange that a "systemctl stop" is instantaneous contrary to the shutdown sequence

stqn · 2012-09-08 14:05:14

I don’t undertand what’s going on; now even with my original fix (TimeoutStopSec=1), systemd still hangs! The log looks similar to previously.

My last update of systemd, initscripts and netcfg is from 2012-09-03.

Going back to initscripts for now.

lahwaacz · 2012-09-08 16:32:57

The 'KillMode=none' fix works fine for me.

drrossum · 2012-09-12 10:57:21

Having the same issue here. In the log I find:

Sep 10 16:45:38 idefix shutdown[1083]: shutting down for system reboot
...
Sep 10 16:47:08 idefix systemd[1]: netcfg.service stopping timed out. Killing.
Sep 10 16:47:08 idefix systemd[1]: Unit netcfg.service entered failed state.
Sep 10 16:47:08 idefix systemd[1]: Shutting down.

I have the same wait time if I do 'systemctl stop'. The netcfg setup is fine, using netcfg interactively works flawlessly.

Snark1994 · 2012-09-15 18:55:06

jouke wrote:

This smells like a bug in systemd. It looks like it is trying to kill some forked-off process, while Type=oneshot should tell it there is no such thing. A better fix than through TimeoutStopSec=1 would be to add KillMode=none on that place. If this turns out to be a good fix and current systemd versions are still buggy, I will put this in the next netcfg release.
Start your testing!

jouke's fix seems to have worked for me - it was a slightly intermittent problem before, but I haven't had it in the week or so since I applied his fix.

jouke · 2012-10-06 09:18:21

The fix finally landed in [core] today. I wonder: was it still needed for recent systemd versions?
For reference: https://bugs.freedesktop.org/show_bug.cgi?id=54041

tomk · 2012-10-13 12:24:27

jouke, this issue still affects netcfg@.service, and adding KillMode=none fixes it. Could you put it in the next release?

jrussell · 2012-10-25 14:54:07

I still have this problem, system is 100% up-to-date - have not edited any .service files

boomshalek · 2012-11-03 16:36:29

I am running netcfg-3.0-1 and the service has the KillMode=none now per default in it, but I still get stuck ... any news ?

EDIT: I am only connecting through eth0 with dhcp through netcfg@ethernet-dhcp.service

Last edited by boomshalek (2012-11-03 16:39:07)

stqn · 2012-11-03 18:24:48

Problem solved here with an up to date system (as of this morning) and the official netcfg service unit. I also changed a few of my startup files (.xinitrc, .zprofile) as per the latest wiki documentation, which might have helped.

boomshalek and jrussel, the only advice I can give is to check the systemd and any other relevant wiki page (your DE, DM, xinitrc…) to see if your configuration needs some fixes.

Arch Linux

#1 2012-09-04 21:09:36

[SOLVED] systemd stuck 5 min at shutdown (netcfg multi-profile daemon)

#2 2012-09-05 11:22:30

Re: [SOLVED] systemd stuck 5 min at shutdown (netcfg multi-profile daemon)

#3 2012-09-05 19:01:14

Re: [SOLVED] systemd stuck 5 min at shutdown (netcfg multi-profile daemon)

#4 2012-09-05 19:20:35

Re: [SOLVED] systemd stuck 5 min at shutdown (netcfg multi-profile daemon)

#5 2012-09-05 23:59:03

Re: [SOLVED] systemd stuck 5 min at shutdown (netcfg multi-profile daemon)

#6 2012-09-06 16:31:22

Re: [SOLVED] systemd stuck 5 min at shutdown (netcfg multi-profile daemon)

#7 2012-09-08 14:05:14

Re: [SOLVED] systemd stuck 5 min at shutdown (netcfg multi-profile daemon)

#8 2012-09-08 16:32:57

Re: [SOLVED] systemd stuck 5 min at shutdown (netcfg multi-profile daemon)

#9 2012-09-12 10:57:21

Re: [SOLVED] systemd stuck 5 min at shutdown (netcfg multi-profile daemon)

#10 2012-09-15 18:55:06

Re: [SOLVED] systemd stuck 5 min at shutdown (netcfg multi-profile daemon)

#11 2012-10-06 09:18:21

Re: [SOLVED] systemd stuck 5 min at shutdown (netcfg multi-profile daemon)

#12 2012-10-13 12:24:27

Re: [SOLVED] systemd stuck 5 min at shutdown (netcfg multi-profile daemon)

#13 2012-10-25 14:54:07

Re: [SOLVED] systemd stuck 5 min at shutdown (netcfg multi-profile daemon)

#14 2012-11-03 16:36:29

Re: [SOLVED] systemd stuck 5 min at shutdown (netcfg multi-profile daemon)

#15 2012-11-03 18:24:48

Re: [SOLVED] systemd stuck 5 min at shutdown (netcfg multi-profile daemon)

Board footer