This commit updates multiple roles to ensure compatibility with Ansible 2.20.
Several include paths and task-loading mechanisms required adjustments,
as Ansible 2.20 applies stricter evaluation rules for complex Jinja expressions
and no longer resolves certain relative include paths the way Ansible 2.18 did.
Key changes:
- Replaced legacy once_finalize.yml and once_flag.yml with the new structure
under tasks/utils/once/finalize.yml and tasks/utils/once/flag.yml.
- Updated all include_tasks statements to use 'path_join' with playbook_dir,
ensuring deterministic and absolute file resolution across roles.
- Fixed all network helper includes by converting direct relative paths such as
'roles/docker-compose/tasks/utils/network.yml' to proper Jinja-evaluated paths.
- Normalized MATOMO_* variable names for consistency with the updated variable
scope behavior in Ansible 2.20.
- Removed deprecated patterns that were implicitly supported in Ansible 2.18
but break under the more strict variable and path resolution model in 2.20.
These changes are part of the full migration step required to ensure the
infinito-nexus roles remain stable, deterministic, and forward-compatible with
Ansible 2.20.
Details of the discussion and reasoning can be found in this conversation:
https://chatgpt.com/share/69300a8d-24d4-800f-bec0-e895a695618a
### Overview
This commit introduces a broad set of improvements across the defaults
generator, credential creation subsystem, inventory creation workflow,
and InventoryManager core logic.
### Major Changes
- Support empty or config/main.yml in defaults generator and ensure that
applications with empty configs are still included in defaults_applications.
- Add '--snippet' and '--allow-empty-plain' modes to create/credentials.py
with non-destructive merging and correct plain-secret handling.
- Ensure empty strings for 'plain' credentials are never encrypted.
- Update InventoryManager to fully support allow_empty_plain and prevent
accidental overwriting or encrypting existing VaultScalar or dict values.
- Add full-size implementation of cli/create/inventory.py including
dynamic inventory building, role filtering, host_vars management, and
parallelised credential snippet generation.
- Fix schemas (Magento, Nextcloud, OAuth2-Proxy, keyboard-color, etc.) to
align with the new credential model and avoid test failures.
- Improve get_app_conf consistency by ensuring credentials.* paths are
always resolvable for applications even when config/main.yml is empty.
### Added Test Coverage
- Unit tests for defaults generator handling empty configs.
- Full test suite for create/inventory.py including merge logic and
vault-safe host_vars loading.
- Extensive tests for InventoryManager: plain-secret behavior,
vault handling, and recursion logic.
- Update or remove outdated tests referencing old schema behaviour.
### Context
This commit is associated with a refactoring and debugging session documented here:
https://chatgpt.com/share/692ec0e1-5018-800f-b568-d09a53e9d0ee
- Add CSP3 support for style/script: include -elem and -attr directives
- Base (style-src, script-src) now unions elem/attr (CSP2/Safari fallback)
- Respect explicit base disables (e.g. style-src.unsafe-inline: false)
- Hashes only when 'unsafe-inline' absent in the final base tokens
- Nginx: set CSP only for HTML/worker via header_filter_by_lua_block; drop for subresources
- Remove per-location header_filter; keep body_filter only
- Update app role flags to *-attr where appropriate; extend desktop CSS sources
- Add comprehensive unit tests for union/explicit-disable/no-mirror-back
Ref: https://chatgpt.com/share/68f87a0a-cebc-800f-bb3e-8c8ab4dee8ee
This update introduces conditional Redis support for Gitea, allowing connection
to either a local or centralized Redis instance depending on configuration.
Includes resource limits for the Redis service and corresponding environment
variables for cache, session, and queue backends.
Reference: ChatGPT conversation on centralized vs per-app Redis architecture (2025-10-15).
https://chatgpt.com/share/68ef5930-49c8-800f-b6b8-069e6fefda01
- Converted group_vars/all/13_ldap.yml from lower-case to ALL-CAPS nested keys.
- Updated all roles, tasks, templates, and filter_plugins to reference LDAP.* instead of ldap.*.
- Fixed Keycloak JSON templates to properly quote Jinja variables.
- Adjusted svc-db-openldap filter plugins and unit tests to handle new LDAP structure.
- Updated integration test to only check uniqueness of TOP-LEVEL ALL-CAPS constants, ignoring nested keys.
See: https://chatgpt.com/share/68b01017-efe0-800f-a508-7d7e2f1c8c8d
- removed obsolete role `srv-web-7-7-dns-records` (README, meta, tasks)
- updated Gitea role to use `sys-dns-cloudflare-records` with explicit record vars
- updated web-opt-rdr-www role to use new DNS role with zone detection (`to_zone`)
- added REDIRECT_WWW_FLAVOR var to support "edge" flavor selection
- replaced CERTBOT_DNS_API_TOKEN with CLOUDFLARE_API_TOKEN everywhere
- introduced generic sys-dns-cloudflare-records role for managing DNS records
- added sys-dns-hetzner-rdns role with both Cloud (hcloud) and Robot API flavors
- updated Mailu role to:
- generate DKIM before DNS setup
- delegate DNS + rDNS records to the new generic roles
- removed legacy per-role Cloudflare vars (MAILU_CLOUDFLARE_API_TOKEN)
- extended group vars with HOSTING_PROVIDER for rDNS flavor decision
- added hetzner.hcloud collection to requirements
This consolidates DNS management into reusable roles,
supports both Cloudflare and Hetzner providers,
and standardizes variable naming across the project.
