service module broken by installation of systemd on ubuntu

[FauxFaux]

Issue Type: Bug Report

Ansible Version: ansible 1.7.2+dfsg-1 (utopic package)

Environment: Ubuntu Utopic (14.10, currently beta) amd64, both, simple install (bash, ssh-keys, ...)

Summary:

If you install the systemd package on Ubuntu (Utopic / 14.10), the service module stops working:

Steps To Reproduce:

$ ansible test-foo -m service -a 'name=redis-server state=stopped enabled=no'

test-foo | FAILED >> {
    "failed": true, 
    "msg": "failure 1 running systemctl show for 'redis-server.service': Failed to get D-Bus connection: No connection to service manager.\n"
}

Expected Results:

It believes that, just because systemctl is on the path, and there's a file named /lib/systemd/system/redis-server.service, that the redis-server package is managed by systemd. However, the system is using upstart for service management, and I believe Ubuntu will be doing so for a long time.

Ironically, systemd is installed because upstart recommends it:

$ dpkg -S $(which systemctl)
systemd: /bin/systemctl
$ aptitude why systemd
i   libdbus-1-3    Recommends dbus                    
i A dbus           Recommends libpam-systemd          
i A libpam-systemd Depends    systemd (= 208-8ubuntu7)
$ aptitude why libdbus-1-3
i   upstart-bin Depends libdbus-1-3 (>= 1.2.16)
$ apt-cache show upstart | fgrep -m 1 Prior
Priority: required

Perhaps checking systemctl actually works could work?

$ systemctl 2>/dev/null || echo "not working"
not working

Maybe check what init is?

% grep -o 'init=[^ ]*' /proc/cmdline || readlink -f /sbin/init
/sbin/upstart

I don't have any systems with working systemd so I can't say what else might work.

[Jmainguy]

Hey,

I am unable to reproduce your issue, I installed ubuntu 14.10 x64, with ssh-server, nothing else. My ansible host is Centos7 with ansible 1.7.2, my target was ubuntu 14.10.

I used the following playbook

---

• name: Install systemd
  apt: name=systemd state=latest
• name: Install server-redis
  apt: name=redis-server state=latest
• name: Restart redis-server
  service: name=redis-server state=restarted

ansible-playbook -i hosts systemd.yml

PLAY [systemd test] ***********************************************************

GATHERING FACTS ***************************************************************
ok: [ubuntu1410.soh.re]

TASK: [systemd | Install systemd] *********************************************
ok: [ubuntu1410.soh.re]

TASK: [systemd | Install server-redis] ****************************************
ok: [ubuntu1410.soh.re]

TASK: [systemd | Restart redis-server] ****************************************
changed: [ubuntu1410.soh.re]

PLAY RECAP ********************************************************************
ubuntu1410.soh.re : ok=4 changed=1 unreachable=0 failed=0

[FauxFaux]

Looks like it's only the enabled=no parameter, it indeed works fine for just restarting (which I probably knew, had I bothered to look at the handlers/ instead of just the tasks/).

Reproduction:

ubuntu@wolf ~$ lxc-create -t download -n jmain -- -d ubuntu -r utopic -a amd64
Using image from local cache
Unpacking the rootfs

---
You just created an Ubuntu container (release=utopic, arch=amd64, variant=default)
The default username/password is: ubuntu / ubuntu
To gain root privileges, please use sudo.

ubuntu@wolf ~$ lxc-start -d -n jmain

ubuntu@wolf ~$ ssh jmain
ubuntu@jmain's password: 
Welcome to...
ubuntu@jmain:~$ sudo apt-get update
[sudo] password for ubuntu: 
Ign http://archive.ubuntu.com utopic InRelease
...
ubuntu@jmain:~$ sudo apt-get install python
Reading package lists... Done
...

ubuntu@wolf ~$ ssh-copy-id jmain
/usr/bin/ssh-copy-id: INFO: attempting...
..

ubuntu@wolf ~$ cat jmain
jmain

ubuntu@wolf ~$ cat jmain.yml 
- hosts: all
  tasks:
  - name: Install systemd
    apt: name=systemd state=latest
    sudo: yes

  - name: Install server-redis
    apt: name=redis-server state=latest
    sudo: yes

  - name: Restart redis-server
    service: name=redis-server state=stopped enabled=no
    sudo: yes

ubuntu@wolf ~$ ansible-playbook -i jmain -s -K jmain.yml
sudo password: 

PLAY [all] ******************************************************************** 

GATHERING FACTS *************************************************************** 
ok: [jmain]

TASK: [Install systemd] ******************************************************* 
changed: [jmain]

TASK: [Install server-redis] ************************************************** 
changed: [jmain]

TASK: [Restart redis-server] ************************************************** 
failed: [jmain] => {"failed": true}
msg: failure 1 running systemctl show for 'redis-server.service': Failed to get D-Bus connection: No connection to service manager.


FATAL: all hosts have already failed -- aborting

PLAY RECAP ******************************************************************** 
           to retry, use: --limit @/home/ubuntu/jmain.retry

jmain                      : ok=3    changed=2    unreachable=0    failed=1   

ubuntu@wolf ~$ 

[dkaufman-irise]

Same problem, different services (bind,ssh,ntp), and "enabled=yes", control machine is a mac, client is running utopic. Running ansible from git checkout, today:

failed: [MY_CLIENT] => {"failed": true}
msg: failure 1 running systemctl show for 'ssh.service': Failed to get D-Bus connection: No connection to service manager.

state= works fine, but enabled= is broken.

[Jmainguy]

Confirmed broken using the very latest service from devel.
Target server is Ubuntu 14.10

---

- name: Install systemd
  apt: name=systemd state=latest

- name: Install server-redis
  apt: name=redis-server state=latest

- name: Restart redis-server
  service: name=redis-server state=restarted enabled=no

PLAY [systemd test] *********************************************************** 

GATHERING FACTS *************************************************************** 
ok: [ubuntu1410.soh.re]

TASK: [systemd | Install systemd] ********************************************* 
ok: [ubuntu1410.soh.re]

TASK: [systemd | Install server-redis] **************************************** 
ok: [ubuntu1410.soh.re]

TASK: [systemd | Restart redis-server] **************************************** 
failed: [ubuntu1410.soh.re] => {"failed": true}
msg: failure 1 running systemctl show for 'redis-server.service': Failed to get D-Bus connection: No connection to service manager.


FATAL: all hosts have already failed -- aborting

PLAY RECAP ******************************************************************** 
           to retry, use: --limit @/root/systemd.retry

ubuntu1410.soh.re          : ok=3    changed=0    unreachable=0    failed=1   

[Jmainguy]

Used @FauxFaux suggestions to create PR #232 which fixed my test hosts, targets centos7 and ubuntu1410.

[nafg]

Is there a workaround?

[Jmainguy]

Yes, I wrote a fix, #232 is the fix, we just need Ansible to merge the code. In the meantime, you can grab this code.

[nafg]

I would call that a fix not a workaround :)

On Wed Nov 05 2014 at 11:20:06 PM Jonathan Mainguy notifications@github.com
wrote:

Yes, I wrote a fix, #232
#232 is the fix, we
just need Ansible to merge the code. In the meantime, you can grab this
code.

Reply to this email directly or view it on GitHub
#161 (comment)
.

[pitabwire]

Just to help someone who may still be struggling with this before the fix is available in pip:

Uninstall systemd
From : https://wiki.ubuntu.com/systemd#Warning.21_Experimental_code
Ubuntu 14.10 does not work well with systemd and is set for development version of Ubuntu 15.04