How to successfully restart a network without reboot over SSH?

Question

In Ubuntu 14.04 neither sudo service networking restart nor sudo /etc/init.d/networking restart do anything any more. They also both exit with code 1. Something has obviously changed (or half changed) but I can't find what. This obviously causes problems with remote network reconfiguration and tools like Ansible.

$ sudo bash -x /usr/sbin/service networking restart
[sudo] password for wirehive:
++ basename /usr/sbin/service
+ VERSION='service ver. 0.91-ubuntu1'
++ basename /usr/sbin/service
+ USAGE='Usage: service < option > | --status-all | [ service_name [ command | --full-restart ] ]'
+ SERVICE=
+ ACTION=
+ SERVICEDIR=/etc/init.d
+ OPTIONS=
+ '[' 2 -eq 0 ']'
+ cd /
+ '[' 2 -gt 0 ']'
+ case "${1}" in
+ '[' -z '' -a 2 -eq 1 -a networking = --status-all ']'
+ '[' 2 -eq 2 -a restart = --full-restart ']'
+ '[' -z '' ']'
+ SERVICE=networking
+ shift
+ '[' 1 -gt 0 ']'
+ case "${1}" in
+ '[' -z networking -a 1 -eq 1 -a restart = --status-all ']'
+ '[' 1 -eq 2 -a '' = --full-restart ']'
+ '[' -z networking ']'
+ '[' -z '' ']'
+ ACTION=restart
+ shift
+ '[' 0 -gt 0 ']'
+ '[' -r /etc/init/networking.conf ']'
+ which initctl
+ grep -q upstart
+ initctl version
+ case "${ACTION}" in
+ stop networking
stop: Job failed while stopping
+ :
+ exec start networking
start: Job is already running: networking

What is the correct way to restart networking in Ubuntu 14.04 Server remotely?

Solution 1

Turns out this is a "feature". The only supported way of restarting an interface in Ubuntu Server is sudo ifdown eth0 && sudo ifup eth0

https://bugs.launchpad.net/ubuntu/+source/ifupdown/+bug/1301015

Solution 2

ifdown, ifup didn't work for me (likely SSH connection timeout before the second command). What did work was:

sudo service network-manager restart

This was on a 14.04 ubuntu-desktop system.

Solution 3

What's changed is they don't want you "bouncing" the network anymore. stopping, and starting, still work. restart no longer works. I just "solved" this "problem", that is to say I got the old behavior back. To revert to prior behavior: Take a 13.10 /etc/init/networking.conf file and replace the 14.04 file with it. (edit: clarified which replaces which)

The process looks like this:

(Just before this, I configured my /etc/network/interfaces for eth1 on a 192.168.117.x address)

"/etc/network/interfaces" 16L, 413C written  
root@1404-Anode:~# service networking restart  
stop: Job failed while stopping  
start: Job is already running: networking  
root@1404-Anode:~# echo "hmm, wth?"
hmm, wth?
root@1404-Anode:~# ifconfig  
eth0      Link encap:Ethernet  HWaddr 00:0c:29:d6:a8:19  
          inet addr:192.168.115.105  Bcast:192.168.115.255  Mask:255.255.255.0  
          inet6 addr: 2002:4077:9050:1234:a08c:29c1:ce9b:a57b/64 Scope:Global  
          inet6 addr: fe80::20c:29ff:fed6:a819/64 Scope:Link  
          inet6 addr: 2002:4077:9050:1234:20c:29ff:fed6:a819/64 Scope:Global  
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1  
          RX packets:646 errors:0 dropped:0 overruns:0 frame:0  
          TX packets:531 errors:0 dropped:0 overruns:0 carrier:0  
          collisions:0 txqueuelen:1000  
          RX bytes:58748 (58.7 KB)  TX bytes:75465 (75.4 KB)  
(lo removed here)  
root@1404-Anode:~# cd /etc/init
root@1404-Anode:/etc/init# diff networking.conf.1310 networking.conf.1404  
13c13  
<       and (stopped udevtrigger or container)) or runlevel [2345]  
---  
>     and (stopped udevtrigger or container)) or runlevel [2345] or stopped  networking   >RESULT=failed PROCESS=post-stop EXIT_STATUS=100  
16a17,20  
>     if [ "$UPSTART_EVENTS" = "stopped" ] && [ "$UPSTART_JOB" = "networking" ] && [ "$EXIT_STATUS" = "100" ]; then  
>         exit 0  
>     fi  
>  
21a26,31  
>     if [ -z "$UPSTART_STOP_EVENTS" ]; then  
>        echo "Stopping or restarting the networking job is not supported."  
>         echo "Use ifdown & ifup to reconfigure desired interface."  
>         exit 100  
>     fi  
root@1404-Anode:/etc/init#  

Do the same for /etc/init.d/networking script, which is what the /etc/init/networking.conf file references/calls.

root@1404-Anode:/etc/init# cp networking.conf.1310 networking.conf  
root@1404-Anode:/etc/init# cd ../init.d  
root@1404-Anode:/etc/init.d# diff networking.1404 networking.1310  
15d14  
< STATEDIR="$RUN_DIR/state"
21a21,27
> # Make sure that it's clear to the user that they shouldn't use this
> # script under upstart
> if init_is_upstart; then
>       echo "ERROR: Calling a sysvinit script on a system using upstart isn't supported. Please use the 'service' command instead."  
>       exit 1
> fi
>
52,54d57
<       if ! chown root:netdev "$RUN_DIR" ; then
<           log_warning_msg "can't chown $RUN_DIR"
<       fi
160,162d162
<       if init_is_upstart; then
<               exit 1
<       fi
166c166
<       state=$(ifquery --state)
---
>       state=$(cat /run/network/ifstate)
root@1404-Anode:/etc/init.d# cp networking.1310 networking
root@1404-Anode:/etc/init.d# service networking restart
networking stop/waiting
networking start/running
root@1404-Anode:/etc/init.d# ifconfig
eth0      Link encap:Ethernet  HWaddr 00:0c:29:d6:a8:19
          inet addr:192.168.115.105  Bcast:192.168.115.255  Mask:255.255.255.0
          inet6 addr: 2002:4077:9050:1234:a08c:29c1:ce9b:a57b/64 Scope:Global
          inet6 addr: fe80::20c:29ff:fed6:a819/64 Scope:Link
          inet6 addr: 2002:4077:9050:1234:20c:29ff:fed6:a819/64 Scope:Global
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:3398 errors:0 dropped:0 overruns:0 frame:0
          TX packets:2545 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000
          RX bytes:318654 (318.6 KB)  TX bytes:418804 (418.8 KB)

eth1      Link encap:Ethernet  HWaddr 00:0c:29:d6:a8:23
          inet addr:192.168.117.105  Bcast:192.168.117.255  Mask:255.255.255.0
          inet6 addr: fe80::20c:29ff:fed6:a823/64 Scope:Link
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:98 errors:0 dropped:58 overruns:0 frame:0
          TX packets:15 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000
          RX bytes:20055 (20.0 KB)  TX bytes:1226 (1.2 KB)
(lo removed)
root@1404-Anode:/etc/init.d# echo "hah, it works! *8^)"
Hah, it works! *8^)
root@1404-Anode:/etc/init.d#

(note: the steps where I scp'd the 1310 versions into the 1404 system are omitted for brevity). 

Obviously there's a reason they put a defensive exit in there, but they don't bother to really output what's going on very well.

An entry does go into /var/log/upstart/networking.log when you try, that looks like:

Stopping or restarting the networking job is not supported.  
Use ifdown & ifup to reconfigure desired interface.  

But they really could/should have output that as the dialog message when you try service networking restart. ah well. figured it out and even a old-way work around.

EDIT: I have found this causes an unintentional triggering of the script controlled by /etc/init/failsafe.conf which is undesirable as it causes a 120 second timeout delay in every boot up... as well as perhaps masking actual misconfigurations/network issues which this delay's appearance would indicate, but it's already showing all the time. (e.g. An unplugged cable, which was allowing access to a network file share mapped in /etc/fstab, for instance)

In any case, I will figure out what is causing this always hitting the timeout, and post a fix when I find it.

Solution 4

In reference to kvm-user420's response, I've gone ahead and setup a script to replace Ubuntu 14.04's networking scripts with that of Ubuntu 13.10

You can find it here: https://github.com/metral/restore_networking

Enjoy!

View more solutions

Comments

[chili555]

Do you really want to restart networking or just restart an interface such as eth0? In any case, it will probably break the ssh connection.

[goo]

Since both are shell scripts, gain information with sudo bash -x /usr/sbin/service networking restart and sudo bash -x /etc/init.d/networking restart

[SimonJGreen]

@chili555 what I want to happen is for changes in /etc/network/interfaces to take effect. Prior to these scripts stopping working SSH connections survive this restart.

[SimonJGreen]

@waltinator additional info added to question, thanks for the suggestion

[Admin]

@kvm-user420 can you please upload your conf files. It doesn't work for me and my networking script was a bit different (in Ubuntu 14.04).

[Philippe Paré]

FWIW on 15.04 sudo service networking restart seems to work OK for me...

[Julia Anne Jacobs]

Make sure you run 'sudo ifdown eth0 && sudo ifup eth0' (note the second sudo) if you're not running this command as root

[meawoppl]

Is this ok to do if connected via ssh? Ie, will both commands run before the disconnect?

[u2n]

Accepted answer didn't work for me; this one did. (Fresh install of 14.04.1)

[gerlos]

Can confirm that this works for me on Trusty (Ubuntu 14.04).

[dieuwe]

The "ifdown/ifup" answer did not work for me either on Ubuntu 14.04, but this did.

[Swiss]

The accepted answer is applicable for Ubuntu 12.04 and lower. This answer is acceptable for 14.04.

[0xF2]

the wisdom of modifying a core OS behavior is debatable.

[Dig]

Yes I am using 14.04.1 and the accepted answer did NOT work for me either. The good news is that 'sudo service network-manager restart' works a treat for me. Thank you all involved.

[Andrew]

This worked for me. Hopefully, one day Upstart will just go away ;-)

[polarise]

In my case, running sudo service networking restart seemed to have messed things up. So I had to run sudo ifdown eth0 && sudo ifup eth0 twice. Yes, twice.

[Xdg]

@ExceptionSlayer: No, they didn't work for me. I had to power down server physically and turn it on again (Ubuntu 14.04).

[YahyaE]

It worked on 14.04 LTS.

[Isaac]

Isn't using a ; instead of && a safer option?

[flickerfly]

Why did they remove the ability to restart? I'd like to know that before reinstating it.

[flickerfly]

Presence of network-manager depends on your install version. I believe the server version doesn't have it, but the desktop one does. They are very different approaches to network configuration.

[flickerfly]

That bug report is quite a read! Sounds like there are a notable amount of cases not easily satisfied by this. Thanks

[Nam G VU]

Not work for 15.10

[Dan]

This unfortunately doesn't work from Ansible, the first ifdown kills the connection and the 2nd command is never run. Borrowing from a stack overflow answer, executing the command in a separate shell will work, e.g. command: bash -c "ifdown eth0 && ifup eth0". Source: stackoverflow.com/questions/28715003/…

[ihadanny]

worked for me on 15.10\

[Dale Anderson]

@0xF2 - you mean questionable.

[0xF2]

@DaleAnderson that too ;-)