- Dependencies
- Role Variables
- The Caddyfile
- Whether to template the Caddyfile on each run
- The OS to download caddy for
- The version of Caddy to use
- Auto update Caddy?
- Additional Available Packages
- Use
setcap
? - Use systemd capabilities controls
- Add additional environment variables or files
- Use additional CLI arguments
- Use a GitHub OAuth token to request the list of caddy releases
- Example Playbooks
- Developing
- Debugging
- Contributing
This role installs and configures the caddy web server. The user can specify any http configuration parameters they wish to apply their site. Any number of sites can be added with configurations of your choice.
None
See Caddyfile docs. Notice the |
used to include a multi-line string. You may set caddy_conf_filename
to config.json
to use json format.
default:
caddy_conf_filename: Caddyfile
caddy_config: |
https://localhost:2020
respond "Hello, world!"
If you wish to use a template for the config you can do this:
caddy_config: "{{ lookup('template', 'templates/Caddyfile.j2') }}"
By default the Caddyfile is templated on each run. By setting this variable you can ensure the file is created on the first run but never updated after.
caddy_config_update: true
default:
caddy_os: linux
default:
caddy_version: ''
This option cannot be used together with caddy_packages
because that option prevents the downloads from using Github, meaning no older versions are available.
The version of Caddy will only be changed after the first run of the role if caddy_update
is set to true
, otherwise the version of Caddy will remain as the one selected on first run of the role.
default:
caddy_update: true
Changing this variable will reinstall Caddy with the new packages if caddy_update
is enabled. Check https://caddyserver.com/download for available packages.
This causes the builds to be downloaded from https://caddyserver.com rather than using the github releases. This service is provided for free by the Caddy maintainers and if you rely on it you should consider donating. The capacity of this service is limited so if you use this role to manage Caddy across many hosts it is recommended to use a different method.
default:
caddy_packages: []
This allows Caddy to open a low port (under 1024 - e.g. 80, 443).
default:
caddy_setcap: true
default:
caddy_systemd_capabilities_enabled: false
caddy_systemd_capabilities: "CAP_NET_BIND_SERVICE"
NOTE: This feature requires systemd v229 or newer and might be needed in addition to caddy_setcap: yes
.
Supported:
- Debian 9 (stretch)
- Fedora 25
- Ubuntu 16.04 (xenial)
RHEL/CentOS has no release that supports systemd capability controls at this time.
Add environment variables to the systemd script.
default:
caddy_environment_variables: {}
Example usage:
caddy_environment_variables:
FOO: bar
SECONDVAR: spam
Add environment files to the systemd script.
default:
caddy_environment_files: []
Example usage:
caddy_environment_files:
- /etc/default/caddy_additional_env_file
default:
caddy_additional_args: ""
Example for LetsEncrypt staging:
caddy_additional_args: "-ca https://acme-staging.api.letsencrypt.org/directory"
This role uses the GitHub releases list to check when a new version is available. GitHub has some fairly agressive rate-limiting which can cause failures. You can set your GitHub token to increase the limits for yourself when running the role (e.g. if deploying many servers behind a NAT or running this role repeatedly as part of a CI process).
default:
caddy_github_token: ""
---
- hosts: all
become: yes
roles:
- role: caddy_ansible.caddy_ansible
caddy_config: |
files.example.com
encode gzip
file_server browse {
root /home/caddy/
}
Example with DigitalOcean DNS for TLS:
---
- hosts: all
roles:
- role: caddy_ansible.caddy_ansible
caddy_environment_variables:
DO_AUTH_TOKEN: "your-token-here"
caddy_systemd_capabilities_enabled: true
caddy_systemd_network_dependency: false
caddy_packages: ["github.com/caddy-dns/lego-deprecated"]
caddy_config: |
nextcloud.example.com {
log
reverse_proxy https://localhost:8080 {
header_up Host {http.request.host}
header_up X-Real-IP {http.request.remote.host}
header_up X-Forwarded-For {http.request.remote.host}
header_up X-Forwarded-Port {http.request.port}
header_up X-Forwarded-Proto {http.request.scheme}
}
tls [email protected] {
dns lego_deprecated digitalocean
}
}
It is recommended to create a virtualenv for development and then install all requirements for tests:
python3 -m pip install -U requirements.txt
python3 -m pip install -U ansible ansible-lint yamllint molecule-plugins[docker,lint] pytest pytest-testinfra
After doing so you can run the molecule tests with:
PY_COLORS=1 ANSIBLE_FORCE_COLOR=1 MOLECULE_DISTRO=ubuntu2004 molecule test
Or run the alternate playbook for testing when a specific version is requested:
PY_COLORS=1 ANSIBLE_FORCE_COLOR=1 MOLECULE_DISTRO=ubuntu2004 MOLECULE_PLAYBOOK=converge-version.yml molecule test
If the service fails to start you can figure out why by looking at the output of Caddy.
systemctl status caddy -l
If something doesn't seem right, open an issue!
Pull requests are welcome. Please test your changes beforehand with vagrant:
vagrant up
vagrant provision # (since it already provisioned there should be no changes here)
vagrant destroy