Version 1.22
v1.22.0
Section titled “v1.22.0”New Containers
Section titled “New Containers”- Caddy (
caddy): alternative reverse proxy to Traefik. Disabletraefikand enablecaddyin the container map to switch. Caddy routes the same containers (proxy: true) and uses the same settings (reverse_proxy.yml): Let’s Encrypt wildcard SSL through the DNS challenge, the IP allowlist,hmsdocker_proxy_security_hardening, Authentik and Tinyauth sign-ins, external hosts, the Homepage widget, the Plex certificate script and the CrowdSec bouncer. Disabled by default - CrowdSec (
crowdsec): reads Traefik’s access logs, detects abusive traffic and blocks those IPs on every route through Traefik (ports80and443), using the CrowdSec bouncer plugin. Requires Traefik or Caddy (CrowdSec with Caddy), and turns on their access logs while enabled. Uses the Docker Socket Proxy whensocket-proxyis enabled. Disabled by default - NetBird (
netbird): mesh VPN client that joins the server to a NetBird network, on NetBird Cloud or a self-hosted management server. Sethmsdocker_netbird_setup_keyininventory/group_vars/all/netbird.yml. Disabled by default
All variables now start with hmsdocker_
Section titled “All variables now start with hmsdocker_”Every setting was renamed to start with hmsdocker_, the role’s name, so the names are consistent and cannot clash with other roles. For example:
| Before | Now |
|---|---|
hms_docker_domain |
hmsdocker_domain |
hms_docker_container_map |
hmsdocker_container_map |
traefik_log_level |
hmsdocker_traefik_log_level |
container_uid |
hmsdocker_container_uid |
plex_claim_token |
hmsdocker_plex_claim_token |
The full list is in migrations/variable_renames.yml.
make updaterenames the variables in yourinventory/group_vars/all/*.ymlfiles for you, including where one variable is used in the value of another- If you are updating from v1.20 or older, run
make updatea second time (ormake migrate-config) to rename them. The first run still uses the old update steps - The old names keep working. While any are in use, the playbook lists them in one warning at the start of the run
- Variables passed with
-eare renamed too, for example-e debug_mode=trueis now-e hmsdocker_debug_mode=true
Reverse proxy settings
Section titled “Reverse proxy settings”The settings Traefik and Caddy share no longer have traefik in their name, and they moved from traefik.yml to a new defaults file, reverse_proxy.yml. Run make update to rename them in your inventory and move them from inventory/group_vars/all/traefik.yml to inventory/group_vars/all/reverse_proxy.yml. Until then the old names keep working, and are listed in the same warning. If a setting is in both files, make update leaves it and warns, because the value in traefik.yml is the one used.
| Old name | New name |
|---|---|
traefik_ssl_enabled |
hmsdocker_ssl_enabled |
traefik_ssl_use_letsencrypt_staging_url |
hmsdocker_ssl_letsencrypt_staging |
traefik_ssl_letsencrypt_email |
hmsdocker_ssl_letsencrypt_email |
traefik_ssl_sans |
hmsdocker_ssl_sans |
traefik_ssl_dns_provider_environment_vars |
hmsdocker_ssl_dns_provider_environment_vars |
traefik_ssl_dns_resolver_1, traefik_ssl_dns_resolver_2 |
hmsdocker_ssl_dns_resolver_1, hmsdocker_ssl_dns_resolver_2 |
traefik_security_hardening |
hmsdocker_proxy_security_hardening |
traefik_subnet_allow_list |
hmsdocker_proxy_allow_list |
traefik_ext_hosts_enabled, traefik_ext_hosts_list |
hmsdocker_proxy_ext_hosts_enabled, hmsdocker_proxy_ext_hosts_list |
traefik_enable_access_logs |
hmsdocker_proxy_access_logs |
traefik_verify_endpoints |
hmsdocker_proxy_verify_endpoints |
The Traefik-only settings keep traefik in their name and stay in traefik.yml: traefik_ssl_dns_provider_code, traefik_ssl_dns_provider_zone and traefik_log_level are now hmsdocker_traefik_ssl_dns_provider_code, hmsdocker_traefik_ssl_dns_provider_zone and hmsdocker_traefik_log_level.
The traefik key of each container in the container map is now proxy (whether the container is reachable through Traefik or Caddy). make update renames it in inventory/group_vars/all/container_map.yml (and any other file there that defines the container map) and keeps your values. Until then, entries that still use traefik keep working, with a deprecation warning. A container map passed with -e/--extra-vars as hmsdocker_container_map must be renamed by hand: the playbook stops if it still uses traefik.
make update now saves your config to backups/inventory-<date> before changing it. Going back to the stable release with make switch-stable before this version is released keeps the new names, which the stable release does not know. Restore that backup afterwards, see Switching back to stable.
A setting under an old name now carries over to the new name with exactly the same value, on every Ansible version. Before, values such as false, no or an empty list were dropped when a setting was renamed, so the default of the new name applied, and Ansible older than 2.19 turned a value such as the text "true" into True.
New variables
Section titled “New variables”reverse_proxy.yml:
hmsdocker_proxy_trusted_proxies(default[]): proxies in front of Traefik or Caddy, such as a load balancer. The reverse proxy reads the client IP from theirX-Forwarded-Forheader, so its access log (and CrowdSec) shows the real client IP. See Proxies in front of Traefik
crowdsec.yml:
hmsdocker_crowdsec_caddy_bouncer_version(defaultv0.15.0): the CrowdSec bouncer module built into the Caddy image
caddy.yml (new):
hmsdocker_caddy_version(default"2.11.9"),hmsdocker_caddy_log_level(defaultINFO)hmsdocker_caddy_dns_provider(default:hmsdocker_traefik_ssl_dns_provider_code) andhmsdocker_caddy_dns_provider_config(default{env.CF_DNS_API_TOKEN}): the caddy-dns provider and its settings for SSL. See Caddy SSL Certificateshmsdocker_caddy_dns_provider_module,hmsdocker_caddy_extra_modules: modules built into the Caddy imagehmsdocker_caddy_custom_config_location(default<apps path>/caddy/custom): folder of your own*.caddysites
Changes
Section titled “Changes”- New installs enable Seerr instead of Overseerr. Existing container maps are not changed
- The playbook warns when Overseerr or Jellyseerr is enabled. Both are pending removal, see the v1.16 notes for moving your data to Seerr
authentik_old_pg_userandauthentik_old_pg_dbare no longer supported. If you set them, the playbook stops before changing anything and tells you to move the values tohmsdocker_authentik_pg_userandhmsdocker_authentik_pg_db- The upgrade prompts for versions older than v0.2 were removed
- The Tracearr container (
tracearr) setsTIMESCALEDB_AUTO_UPDATE=true, so Tracearr updates the TimescaleDB extension automatically
- Speedtest Tracker: with
authentik: trueonspeedtest-tracker, the Traefik router referenced a nonexistentauthentik-proxy-<project>-speedtest-midwaremiddleware and failed. It now referencesauthentik-proxy-<project>-speedtest-tracker-midware, the middleware the Authentik outpost creates - Tinyauth: the warning that
hmsdocker_tinyauth_protected_containersentries are not protected no longer appears while thetinyauthcontainer is disabled, or for entries whose container is disabled. When it does appear, it names the reason. The same goes for external hosts withtinyauth: yes - Plex SSL certificate script: the PKCS12 file given to Plex now includes the intermediate certificates along with the server certificate, so clients that validate the full chain no longer reject the connection. The script is now owned by
root, since it runs from root’s crontab, and requirescryptography>=36 make install-reqsnow installs theansible.posixandansible.utilscollections. The playbook uses both, but they were only present when the fullansiblepackage was installed- The credentials of an additional CIFS share are no longer shown in the playbook output
- Notifiarr no longer fails with an undefined variable when it is enabled without Traefik
Internal
Section titled “Internal”- Sonarr, Radarr and Lidarr share one bootstrap task file, and their compose services, the Traefik and Homepage labels of every container, the Authentik outposts and the media share mounts are each written from one shared template or task file instead of copies. The generated files are the same as before
ansible-lintnow runs in CI, and the playbook test also runs when the Makefile, migrations or CI variables change