CODEWITHSUNDEEP
ansible
Joey Yi Zhao
Joey Yi Zhao

Reputation: 42484

How to create new system service by ansible-playbook

I have created a script to start/stop my application. Now I want to add it as a centos system service. First I created a task to create a link from my script to /etc/init.d/service_name as below.

---

- name: create startup link
  file: src={{ cooltoo_service_script }} dest={{ cooltoo_service_init }} state=link

After create the service, I want to add it to system service. The command used to do that is "chkconfig --add service_name". I wonder whether there is a ansible module to do that instead of hardcoded the command in ansible-playbook file. I have looked at this page http://docs.ansible.com/ansible/service_module.html but it only shows how to manage a service not create a new one.

Upvotes: 34

Views: 89326

Answers (4)

Cameron Kerr
Cameron Kerr

Reputation: 1875

The 'service' module supports an 'enabled' argument.

Here's an example part of a playbook, which I will freely admit does look like a newbie attempt. This assumes RHEL/CentOS 6.x, which uses SysV, not systemd.

  - name: install rhel sysv supervisord init script
    copy: src=etc/rc.d/init.d/supervisord dest=/etc/rc.d/init.d/supervisord owner=root group=root mode=0755

  - name: install rhel sysv supervisord sysconfig
    copy: src=etc/sysconfig/supervisord dest=/etc/sysconfig/supervisord owner=root group=root mode=0640

  - name: enable sysv supervisord service
    service: name=supervisord enabled=yes

  - name: start supervisord
    service: name=supervisord state=started

IMPORTANT A lot of custom init scripts WILL FAIL with Ansible and SysV init; the reason being that the 'status' option (service supervisord status) needs to a return an LSB-compliant return code. Otherwise, Ansible will not know if a service is up or down, and idempotency will fail (restart will still work because that is unconditional)

Here's part of a script, which I've just rewritten to make use of the 'status' function within /etc/init.d/functions (you'll notice this same pattern in other Red Hat provided init-scripts in /etc/init.d/

status)
    /usr/bin/supervisorctl $OPTIONS status
    status -p $PIDFILE supervisord
    # The 'status' option should return one of the LSB-defined return-codes,
    # in particular, return-code 3 should mean that the service is not
    # currently running. This is particularly important for Ansible's 'service'
    # module, as without this behaviour it won't know if a service is up or down.
    RETVAL=$?
    ;;

Reference: http://refspecs.linuxfoundation.org/LSB_5.0.0/LSB-Core-generic/LSB-Core-generic/iniscrptact.html

If the status action is requested, the init script will return the following exit status codes.

0 program is running or service is OK 1 program is dead and /var/run pid file exists 2 program is dead and /var/lock lock file exists 3 program is not running 4 program or service status is unknown 5-99 reserved for future LSB use 100-149 reserved for distribution use 150-199 reserved for application use 200-254 reserved

Upvotes: 20

StockB
StockB

Reputation: 774

For RedHat/CentOS 7 (using systemd/systemctl), the equivalent of chkconfig --add ${SERVICE_NAME} is systemctl daemon-reload [via fedoraproject.org].

Then, using the systemd module of Ansible 2.2 or greater, you may start a service with a preceding systemctl daemon-reload like so [via docs.ansible.com]:

# Example action to restart service cron on centos, in all cases, also issue daemon-reload to pick up config changes
- systemd:
    state: restarted
    daemon_reload: yes
    name: crond

Based on my experience, the daemon_reload parameter can also be used within the generic service module as well, although it isn't documented, and might fail on non-systemd systems:

- service:
    state: restarted
    daemon_reload: yes
    name: crond

Upvotes: 5

JosephKumarMichael
JosephKumarMichael

Reputation: 1031

The below code snippet will create Service in CentOS 7.

Code

Tasks

/tasks/main.yml

- name: TeamCity | Create environment file
  template: src=teamcity.env.j2 dest=/etc/sysconfig/teamcity
- name: TeamCity | Create Unit file
  template: src=teamcity.service.j2 dest=/lib/systemd/system/teamcity.service mode=644
  notify:
    - reload systemctl
- name: TeamCity | Start teamcity
  service: name=teamcity.service state=started enabled=yes

Templates

/templates/teamcity.service.j2

[Unit]
Description=JetBrains TeamCity
Requires=network.target
After=syslog.target network.target
[Service]
Type=forking
EnvironmentFile=/etc/sysconfig/teamcity
ExecStart={{teamcity.installation_path}}/bin/teamcity-server.sh start
ExecStop={{teamcity.installation_path}}/bin/teamcity-server.sh stop
User=teamcity
PIDFile={{teamcity.installation_path}}/teamcity.pid
Environment="TEAMCITY_PID_FILE_PATH={{teamcity.installation_path}}/teamcity.pid"
[Install]
WantedBy=multi-user.target

\templates\teamcity.env.j2

TEAMCITY_DATA_PATH="{{ teamcity.data_path }}"

Handlers

\handlers\main.yml

- name: reload systemctl
  command: systemctl daemon-reload

Reference :

Upvotes: 64

udondan
udondan

Reputation: 59989

Indeed the service module only manages already registered services as you have figured out. To my knowledge there is no module to register a service.

Are you aware this step can be skipped with some modifications to your init.d script? If the script follows those rules you can just use the service module to enable/start the service.

Upvotes: 6

Related Questions