feat: drop blanket recreates, ACME-DNS knobs, notify_push override

- Drop `recreate: always` from collabora/drawio/homarr/opencloud/traefik
  handlers and the authentik_outpost_ldap start task. `up -d` with
  `state: present` already recreates exactly the services whose
  compose definition changed; the blanket recreate was forcing
  restarts even when nothing relevant moved.
- Rewrite the `*_domains` Traefik Host loop to the `Host(\`a\`) ||
  Host(\`b\`)` form across authentik/collabora/garage/nextcloud so the
  rule still matches when traefik can't normalize the comma-form into
  the same canonical shape.
- Traefik: add `traefik_acme_tcp_only` (sets LEGO_EXPERIMENTAL_DNS_TCP_ONLY)
  and `traefik_acme_disable_ans_checks` (disables lego's authoritative-NS
  propagation check) for environments where the DNS path between the
  traefik container and the zone's nameservers is constrained.
- Traefik DMZ collector: two-step merge so a `traefik_dmz_exposed_services`
  entry that sets its own `backend_host` wins over the host fallback;
  lets a route target an internal FQDN covered by the backend cert's
  SANs instead of the raw IP.
- Nextcloud: add `nextcloud_notify_push_domain` override for the
  `occ notify_push:setup` call so the setup check can hit an internal
  FQDN instead of hairpinning through the DMZ. Push router now matches
  every entry in `nextcloud_domains`.
- Nextcloud: also %2F-escape slashes in the postgres user/password
  inside the notify_push DATABASE_URL.

roles/authentik/templates/docker-compose.yml.j2

@@ -48,7 +48,7 @@ services:
     labels:
       - traefik.enable=true
       - traefik.docker.network={{ authentik_traefik_network }}
-      - traefik.http.routers.{{ authentik_service_name }}.rule=Host({% for d in authentik_domains %}`{{ d }}`{% if not loop.last %}, {% endif %}{% endfor %})
+      - traefik.http.routers.{{ authentik_service_name }}.rule={% for d in authentik_domains %}Host(`{{ d }}`){% if not loop.last %} || {% endif %}{% endfor +%}
 {% if authentik_use_ssl %}
       - traefik.http.routers.{{ authentik_service_name }}.entrypoints=websecure
       - traefik.http.routers.{{ authentik_service_name }}.tls=true

roles/collabora/handlers/main.yml

@@ -5,4 +5,4 @@
 - name: restart collabora
   community.docker.docker_compose_v2:
     project_src: "{{ collabora_docker_compose_dir }}"
-    state: restarted
+    state: present

roles/collabora/templates/docker-compose.yml.j2

@@ -20,7 +20,7 @@ services:
     labels:
       - traefik.enable=true
       - traefik.docker.network={{ collabora_traefik_network }}
-      - traefik.http.routers.{{ collabora_service_name }}.rule=Host({% for d in collabora_domains %}`{{ d }}`{% if not loop.last %}, {% endif %}{% endfor %})
+      - traefik.http.routers.{{ collabora_service_name }}.rule={% for d in collabora_domains %}Host(`{{ d }}`){% if not loop.last %} || {% endif %}{% endfor +%}
       - traefik.http.services.{{ collabora_service_name }}.loadbalancer.server.port={{ collabora_port }}
 {% if collabora_use_ssl %}
       - traefik.http.routers.{{ collabora_service_name }}.entrypoints=websecure

roles/drawio/handlers/main.yml

@@ -5,4 +5,4 @@
 - name: restart drawio
   community.docker.docker_compose_v2:
     project_src: "{{ drawio_docker_compose_dir }}"
-    state: restarted
+    state: present

roles/garage/templates/docker-compose.yml.j2

@@ -14,7 +14,7 @@ services:
       - traefik.enable=true
       - traefik.docker.network={{ garage_traefik_network }}
       # S3 API endpoint
-      - traefik.http.routers.{{ garage_service_name }}.rule=Host({% for d in garage_s3_domains %}`{{ d }}`{% if not loop.last %}, {% endif %}{% endfor %})
+      - traefik.http.routers.{{ garage_service_name }}.rule={% for d in garage_s3_domains %}Host(`{{ d }}`){% if not loop.last %} || {% endif %}{% endfor +%}
 {% if garage_use_ssl %}
       - traefik.http.routers.{{ garage_service_name }}.entrypoints=websecure
       - traefik.http.routers.{{ garage_service_name }}.tls=true

roles/homarr/handlers/main.yml

@@ -5,4 +5,4 @@
 - name: restart homarr
   community.docker.docker_compose_v2:
     project_src: "{{ homarr_docker_compose_dir }}"
-    state: restarted
+    state: present

roles/nextcloud/defaults/main.yml

@@ -65,6 +65,12 @@ nextcloud_trusted_proxies: "172.16.0.0/12"
 # File locking and real-time push notifications
 nextcloud_enable_notify_push: false
 nextcloud_notify_push_image: "icewind1991/notify_push:1.3.1"
+# Domain used when calling `occ notify_push:setup`. Defaults to the
+# first nextcloud_domains entry (the canonical public name). Override
+# with an internal FQDN to avoid hairpinning the setup check through
+# the DMZ; the FQDN must also be in nextcloud_domains so the push
+# router matches it.
+# nextcloud_notify_push_domain: "cloud.int.example.com"
 
 # Non-default apps to install and enable
 nextcloud_apps_to_install:

roles/nextcloud/tasks/notify_push.yml

@@ -5,4 +5,4 @@
 - name: Configure notify_push base endpoint
   community.docker.docker_container_exec:
     container: "{{ nextcloud_docker_compose_dir | basename }}-nextcloud-1"
-    command: php /var/www/html/occ notify_push:setup https://{{ nextcloud_domains[0] }}/push
+    command: php /var/www/html/occ notify_push:setup https://{{ nextcloud_notify_push_domain | default(nextcloud_domains[0]) }}/push

roles/nextcloud/templates/docker-compose.yml.j2

@@ -35,7 +35,7 @@ services:
     labels:
       - traefik.enable=true
       - traefik.docker.network={{ nextcloud_traefik_network }}
-      - traefik.http.routers.{{ nextcloud_service_name }}.rule=Host({% for d in nextcloud_domains %}`{{ d }}`{% if not loop.last %}, {% endif %}{% endfor %})
+      - traefik.http.routers.{{ nextcloud_service_name }}.rule={% for d in nextcloud_domains %}Host(`{{ d }}`){% if not loop.last %} || {% endif %}{% endfor +%}
 {% if nextcloud_use_ssl %}
       - traefik.http.routers.{{ nextcloud_service_name }}.entrypoints=websecure
       - traefik.http.routers.{{ nextcloud_service_name }}.tls=true
@@ -136,7 +136,7 @@ services:
     environment:
       PORT: "7867"
       REDIS_URL: "redis://redis:6379"
-      DATABASE_URL: "postgres://{{ nextcloud_postgres_user | urlencode }}:{{ nextcloud_postgres_password | urlencode }}@db:5432/{{ nextcloud_postgres_db }}"
+      DATABASE_URL: "postgres://{{ nextcloud_postgres_user | urlencode | replace('/', '%2F') }}:{{ nextcloud_postgres_password | urlencode | replace('/', '%2F') }}@db:5432/{{ nextcloud_postgres_db }}"
       DATABASE_PREFIX: "oc_"
       NEXTCLOUD_URL: "http://nginx"
     networks:
@@ -145,7 +145,7 @@ services:
     labels:
       - traefik.enable=true
       - traefik.docker.network={{ nextcloud_traefik_network }}
-      - traefik.http.routers.{{ nextcloud_service_name }}-push.rule=Host(`{{ nextcloud_domains[0] }}`) && PathPrefix(`/push`)
+      - traefik.http.routers.{{ nextcloud_service_name }}-push.rule=({% for d in nextcloud_domains %}Host(`{{ d }}`){% if not loop.last %} || {% endif %}{% endfor %}) && PathPrefix(`/push`)
       - traefik.http.services.{{ nextcloud_service_name }}-push.loadbalancer.server.port=7867
 {% if nextcloud_use_ssl %}
       - traefik.http.routers.{{ nextcloud_service_name }}-push.entrypoints=websecure

roles/opencloud/handlers/main.yml

@@ -5,4 +5,4 @@
 - name: restart opencloud
   community.docker.docker_compose_v2:
     project_src: "{{ opencloud_docker_compose_dir }}"
-    state: restarted
+    state: present

roles/traefik/defaults/main.yml

@@ -33,6 +33,18 @@ traefik_acme_tsig_secret: ""  # TSIG secret
 traefik_acme_propagation_timeout: "120"
 traefik_acme_polling_interval: "2"
 traefik_acme_ttl: "60"
+# Force lego's DNS lookups (SOA resolution, propagation checks) onto
+# TCP instead of UDP. Useful when container egress can reach the
+# nameserver on TCP/53 but UDP/53 is blocked or unreliable. Sets the
+# upstream env var LEGO_EXPERIMENTAL_DNS_TCP_ONLY=true on the
+# traefik container.
+traefik_acme_tcp_only: false
+# Disable lego's propagation check against the zone's authoritative
+# nameservers. Use when the SOA-listed NS hostname resolves to an
+# address that isn't reachable from this traefik host (e.g. a DMZ
+# box that can only see the internal NS IP, not the public one).
+# lego still polls via the configured `resolvers:` list.
+traefik_acme_disable_ans_checks: false
 
 # Self-signed certificate configuration (for vagrant/testing)
 traefik_selfsigned_cert_dir: "{{ docker_volume_dir }}/certs"

roles/traefik/handlers/main.yml

@@ -5,4 +5,4 @@
 - name: restart traefik
   community.docker.docker_compose_v2:
     project_src: "{{ docker_compose_dir }}"
-    state: restarted
+    state: present

roles/traefik/tasks/main.yml

@@ -9,7 +9,18 @@
 
 - name: Build service registry from backend servers (DMZ mode)
   set_fact:
-    proxied_services: "{{ proxied_services | default([]) + hostvars[item].traefik_dmz_exposed_services | default([]) | map('combine', {'backend_host': hostvars[item].ansible_host | default(item)}) | list }}"
+    # Two-step merge so a service entry's own `backend_host` wins:
+    # entries that set it pass through unchanged, entries that don't
+    # get the backend host's ansible_host as fallback. The override
+    # lets a route target an internal FQDN covered by the backend
+    # cert's SANs instead of the raw IP (which would fail backend
+    # TLS verification at the proxy hop).
+    proxied_services: >-
+      {{
+        proxied_services | default([])
+        + (hostvars[item].traefik_dmz_exposed_services | default([]) | selectattr('backend_host', 'defined') | list)
+        + (hostvars[item].traefik_dmz_exposed_services | default([]) | rejectattr('backend_host', 'defined') | map('combine', {'backend_host': hostvars[item].ansible_host | default(item)}) | list)
+      }}
   loop: "{{ _backend_servers | default([]) }}"
   when: traefik_mode == 'dmz'
 

roles/traefik/templates/docker-compose.yml.j2

@@ -12,6 +12,9 @@ services:
       RFC2136_PROPAGATION_TIMEOUT: "{{ traefik_acme_propagation_timeout }}"
       RFC2136_POLLING_INTERVAL: "{{ traefik_acme_polling_interval }}"
       RFC2136_TTL: "{{ traefik_acme_ttl }}"
+{% if traefik_acme_tcp_only | default(false) %}
+      LEGO_EXPERIMENTAL_DNS_TCP_ONLY: "true"
+{% endif %}
 {% endif %}
     ports:
       - "80:80"

roles/traefik/templates/traefik.yml.j2

@@ -48,6 +48,10 @@ certificatesResolvers:
         provider: rfc2136
         resolvers:
           - "{{ traefik_acme_dns_nameserver }}"
+{% if traefik_acme_disable_ans_checks | default(false) %}
+        propagation:
+          disableANSChecks: true
+{% endif %}
 {% endif %}
 
 {% if traefik_use_ssl %}