Image names now follow {owner}/{repository}/mirror/{registry}/{name}
instead of {owner}/mirror/{registry}/{name}.
- GHCRProvider: add `repository` param + `add_args()`/`from_args()` classmethods
as SPOT for CLI argument registration and provider construction
- sync, resolver: use GHCRProvider.add_args() / from_args() (no duplication)
- Workflows: resolve repository name via scripts/meta/resolve/repository/name.sh
and pass --ghcr-repository to sync and publish
- mirrors.sh: require and forward GHCR_REPOSITORY env var
- deploy/development/mirrors.py: derive repository from GITHUB_REPOSITORY
- Tests: update expected paths, argv and env fixtures
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Keep new multi-registry logic (feature/pre-commit):
- Merge is_mirrorable_image(), _detect_registry(), image_source() and
canonical_image_name() into utils/docker/image_discovery.py
- iter_role_images() now scans both config/main.yml and vars/main.yml
and uses image_source() for full registry-prefixed URLs
- Delete cli/mirror/util.py (superseded by utils/docker/image_discovery)
- model.py re-exports ImageRef from utils.docker.image_discovery
- sync/__main__.py: revert fix/playwright-pull's if/else src URL to
simple `src = f"docker://{img.source}"` (source already contains
the full registry prefix)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Add `registry` field to `ImageRef` (default `"docker.io"`)
- Populate `registry` from `_detect_registry()` in `iter_role_images()`
- `GHCRProvider.image_base()` and `GiteaProvider.image_base()` now produce
`{prefix}/{registry}/{name}` paths (e.g. `mirror/docker.io/nextcloud`,
`mirror/quay.io/keycloak/keycloak`) instead of flattening slashes to dashes
- Fix `mirror()` in both providers: use `docker://{image.source}` directly
instead of always prepending `docker://docker.io/`, which broke quay.io/ghcr.io images
- Update tests to reflect new naming convention and add quay.io test case
- Add missing __init__.py files in test subdirectories
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Extend image mirroring to cover quay.io and ghcr.io in addition to Docker Hub,
and make the publish step non-fatal when the token lacks write:packages scope.
cli/mirror/util.py:
- Add _detect_registry(), _strip_registry_prefix(), is_mirrorable_image()
to support Docker Hub, quay.io, ghcr.io, and mcr.microsoft.com
- Update canonical_image_name() to strip any known registry prefix so that
quay.io/keycloak/keycloak -> keycloak/keycloak (mirror/keycloak-keycloak)
- Add image_source() returning full registry-prefixed pull reference
(docker.io/library/postgres:16, quay.io/keycloak/keycloak:latest, …)
- iter_role_images() now uses is_mirrorable_image + image_source, picking up
the 17 previously skipped quay.io/ghcr.io images
cli/mirror/sync/__main__.py:
- Use docker://{img.source} directly (no hardcoded docker.io/ prefix)
cli/mirror/publish/__main__.py:
- Detect HTTP 404 on PATCH as _InsufficientTokenError (no retry, graceful
GHA warning + exit 0) instead of retrying 5x and failing the workflow
- Use pkg.get("url") from listing response as canonical PATCH target URL
- Use safe="/" in urllib.parse.quote so mirror/name stays as literal path
- Do not retry on non-transient 4xx errors (400/401/403)
tests/unit/cli/mirror/test_sync.py:
- Update ImageRef.source to new docker.io/library/… format
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Move iter_role_images + all helpers + ImageRef from cli/mirror/util.py and
cli/mirror/model.py into utils/docker/image_discovery.py; add source_file field
to ImageRef to distinguish config/main.yml vs vars/main.yml entries
- Delete cli/mirror/util.py; update all callers to import from utils.docker.image_discovery
- Move utils into subpackages: utils/docker/ (image_ref, paths_utils, hadolint) and
utils/roles/ (dependency_resolver, resource_validation); update all imports
- Reorganise tests into subpackages: tests/lint/docker/dockerfile/, tests/integration/docker/,
tests/unit/utils/docker/, tests/unit/utils/roles/; fix parents[N] offsets after relocation
- tests/lint/docker/test_image_versions.py now uses iter_role_images (same SPOT as mirror)
- Add vars/main.yml images.<name>.{image,version} support for non-application roles
(test-e2e-playwright, sys-ctl-hlth-csp); eliminate hardcoded duplicates in defaults/tasks
- Add web-app-matrix matrix-chatgpt-bot and web-app-mini-qr mini-qr entries in config/main.yml;
replace hardcoded image strings in vars/templates with lookup('config', ...) references
- Fix sync/__main__.py log line to use correct registry prefix for non-Docker-Hub images
- Update docs/contributing/code/role/dockerfile.md with image declaration rules and SPOT reference
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Move _set_public/ensure_public out of GHCRProvider into a new
cli.mirror.publish module that lists all non-public mirror packages
via the GitHub API and sets them public in bulk. The workflow runs
this as a separate step after mirroring. Mirror sync itself no longer
touches visibility.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Move utils/gha/annotations.py → utils/annotations/message.py
- Move scripts/gha/summarize_annotations.py → utils/annotations/summarize.py
- Keep utils/gha/annotations.py as compatibility shim
- Update all imports across codebase to utils.annotations.message
- Wire tee + summary step into test-code-lint, unit, and integration workflows
- Add unit tests for utils/annotations/summarize (20 tests)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Replace copy-pasted ::warning prints and _gha_escape helpers across lint
tests and utils with a shared utility that emits GHA workflow commands in
CI and plain [WARNING]/[ERROR]/[NOTICE] lines locally. Update
test_role_resource_validation to capture stdout instead of stderr.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Add utils/gha/annotations.py with warning/error/notice helpers and *_each
variants that emit one annotation per list item
- Use warning() in GHCRProvider._set_public for missing GHCR_PAT case
- Expose GHCR_PAT_SET env var in mirror step to distinguish missing-PAT from
real API failures
- Add unit tests for all annotation functions
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
/users/{username}/packages does not support PATCH — it is a read-only endpoint
for public data. The correct endpoint for updating visibility of the authenticated
user's packages is /user/packages (without username).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Root cause: GITHUB_TOKEN in Actions is repo-scoped and cannot change package
visibility in user/org namespaces — all 58 mirror operations failed with 404.
- Use GHCR_PAT (which has write:packages scope) as GITHUB_TOKEN in the mirror
step, falling back to GITHUB_TOKEN if GHCR_PAT is not set
- Remove ensure_public() call from the already-exists skip path (wrong place:
existing images were already made public when first mirrored)
- Update test to assert ensure_public is not called for skipped images
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Remove redundant `images-mirror-missing` job from entry-pull-request-change.yml
(mirroring already runs via ci-orchestrator for same-repo PRs)
- Set GHCR package visibility to public after each successful mirror via GitHub API
- Add `ensure_public` hook to base class (no-op for other providers)
- Also call `ensure_public` for already-existing images to fix any that are still private
- Pass GITHUB_TOKEN explicitly to the mirror step
- Add unit tests for _set_public (user/org endpoint fallback) and sync skip-path
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Introduces a SPOT for all frontend service loading in sys-front-inj-all/tasks/01_services.yml,
driven by a new generic `service` lookup plugin that resolves services by key or role name
and returns {id, role, enabled, shared, needed}. Needed = enabled AND shared (direct or
transitive). The canonical key→role mapping moves to group_vars/all/20_services.yml
(previously scattered hardcoded rules). Removes per-role service loads from bigbluebutton,
nextcloud, and dashboard. Renames 01_dependencies.yml→01_services.yml and
08_services.yml→08_daemons.yml. Adds flush_handlers after each service load.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Scans roles_dir to find every role whose config/main.yml defines the
disabled service, then creates or updates the compose.services entry in
host_vars for each matching role — including roles not yet present in
applications. roles_dir is now a required parameter (no Optional fallback).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- deploy.py: use find_provider_roles() to map service names (e.g. 'matomo')
to provider app IDs (e.g. 'web-app-matomo') before filtering deploy_ids;
forward SERVICES_DISABLED into the container via extra_env
- container.sh: rename INSPECT_CMD to CMD for consistency; fix indentation
- Makefile: expose CMD variable on exec target for non-interactive use
- environment.sh: source scripts/meta/env/all.sh for INVENTORY_DIR; add
inventory verification after SERVICES_DISABLED deploy using make exec
- variables.md: document CMD variable
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Add scripts/meta/resolve/image/local.sh as SPOT for the local Docker
image tag (<repo-basename>/<distro>). Wire it into Makefile build
targets (IMAGE_TAG), compose.py (_base_env sets INFINITO_IMAGE), and
compose.yml (no default fallback). Fix build.sh to accept a pre-set
IMAGE_TAG without requiring REPO_PREFIX. Add make build + build-no-cache
to environment.sh test flow with correct section ordering.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Add services_disabler.py: parse SERVICES_DISABLED, disable services in host_vars,
remove provider roles from inventory
- Wire apply_services_disabled_from_env into cli/create/inventory/cli.py after deploy
- Add unit tests for all services_disabler functions
- Update hardware.md and variables.md with SERVICES_DISABLED documentation
- Update environment.sh: deploy dashboard with SERVICES_DISABLED=matomo first,
assert dashboard reachable + matomo not reachable, then deploy matomo and assert
reachable; extract assert_http_status / assert_http_status_not helpers
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* replace compose.services.desktop with compose.services.dashboard across all roles
* rename role web-app-desktop to web-app-dashboard
* rename sys-front-inj-desktop to sys-front-inj-dashboard
* update resolver mapping to include web-app-dashboard when dashboard is enabled
* adjust CSP filters to reference dashboard for frame-ancestors handling
* update lookup plugins, filter plugins, templates, vars and run_once flags
* migrate all tests to dashboard naming and service key
* update role dependencies and port mappings
* remove legacy desktop role and variables
BREAKING CHANGE:
The service key compose.services.desktop has been renamed to compose.services.dashboard.
All custom configs, inventories, or overrides must be updated accordingly.
https://chatgpt.com/share/698da7ad-1978-800f-82b4-4755b17be026
Update unit tests to reflect the new invokable-driven classification logic by mocking invokable path discovery, removing legacy alias expectations, and adding coverage for universal and mixed cases. Also relocate CLI deploy tests into the unit test hierarchy for consistency.
Switch sufficient_storage CLI to support JSON output and update CI matrix
generation to use JSON-only pipelines, preventing roles from collapsing
into a single matrix entry.
- Move server/workstation/universal resolution into Python CLI
- Add optional --lifecycles filter to cli.meta.applications.type
- Gate invokable roles by galaxy_info.lifecycle in module_utils.invokable
- Replace legacy resolve-invokable-roles.sh with CLI-based implementation
- Apply global hard excludes consistently across all deploy types
- Fix SC2086 issues and make lifecycle argument handling shell-safe
- Extend unit tests to cover lifecycle forwarding and normalization
https://chatgpt.com/share/698b0039-c4c4-800f-b962-7033324c07a8
Move TLS/ACME/CA vars out of 00_general into group_vars/all/02_tls.yml and renumber subsequent group_vars files. Add sys-ca-selfsigned to generate /etc/<software>/ca/root-ca.crt and make docker-compose-ca depend on it. Update setup script, templates, and tests to use the new applications/users file numbering.
https://chatgpt.com/share/697b8150-22ac-800f-98b1-e063a659b684
- add --inventory-dir to container CLI test invocations
- assert inventory_dir is forwarded to run_in_container()
- add tests for required INVENTORY_DIR when env is missing
- add test for INVENTORY_DIR env fallback
https://chatgpt.com/share/69700d69-b270-800f-ad50-d9ff486378b9
