cimon-io / ansible-role-systemd-service Goto Github PK

In the service template as shown below, the ExecReload in the for loop should reference exec_reload instead of the given array.

Correct format:

{% if systemd_service[item].exec_reload is defined %}
{% if systemd_service[item].exec_reload is not string %}
{% for exec_reload in systemd_service[item].exec_reload -%}
ExecReload={{ exec_reload }}
{% endfor %}
{% else %}
ExecReload={{ systemd_service[item].exec_reload }}
{% endif -%}
{% endif -%}

Hello,

First of all, thank you for this project!

I would like to request that user services are supported.
I am starting to work with rootless Podman and would like to use your role to create my service definitions.

I think all that would be needed is to specify the user systemd location for the service file rather than /etc and to do a user side daemon-reload.

The following role invocation using Ansible 2.9.10 on Debian Stretch controller:

- name: >
    Install systemd service unit to generate /dev/loop nodes on startup.
    (rc.local is disabled in Debian Buster in lieu of systemd.)
  become: true
  hosts: engineering_services_jenkins_cluster
  vars:
    - systemd_service:
        create-loop-devices:
          description: Oneshot service unit to create loop devices at boot.
          enabled: Yes
          exec_start: "/bin/bash -lc '/usr/local/bin/create-loop-devices.sh'"
          user: root
          group: root
          type: oneshot
          restart: no
  roles:
    - role: cimon-io.systemd_service

Results in the following service unit on the controlled (Debian Buster) machine:

[Unit]
Description="Oneshot service unit to create loop devices at boot."

[Service]
Type=oneshot
RestartSec=30
Restart=False
User=root
Group=root
ExecStart=/bin/bash -lc '/usr/local/bin/create-loop-devices.sh'

which results in the following syslog entry in the controlled machine:

systemd[1]: /etc/systemd/system/create-loop-devices.service:6: Failed to parse service restart specifier, ignoring: False

Expected behavior is that the default "restart: no" would result in a valid "restart" setting in the systemd service file. The man page says that Restart:

Takes one of no, on-success, on-failure, on-abnormal, on-watchdog, on-abort, or always.

As a test, I changed my role to not specify a "restart" parameter, and then, and only then did my service actually run.

Your syntax is much nicer than the tumf version. We have some CentOS 7 boxes so it'd be nice to have support for that. :-)

Kind of cruel to forget one of the major players in systemd. I know, it's trivial to fix. Just thought I'd mention it. But I must congratulate you on the work you did to make this work. It's a major job. I will try it soon.