Switch CI detection to GITHUB_ACTIONS and disable compose builds via
INFINITO_NO_BUILD=1. The deploy workflows now rely on pulling the prebuilt
CI image instead of triggering local Docker builds.
- Remove legacy UP_BUILD handling
- Enable --no-build in compose when running in real CI
- Keep local/act builds working as before
https://chatgpt.com/share/69666e0b-f08c-800f-93fd-611eced4b100
- Rename _deploy-matrix.yml -> _deploy-per-app.yml and drop unused inputs
(tested_lifecycles, only_distro, final_exclude_re).
- Keep one runner per app (matrix.app) and run all distros inside that job.
- Makefile: run deploy tests through act workflow_dispatch (no serial script).
- Add scripts/ci/discover.sh and scripts/ci/deploy-one-app.sh as small SRP helpers.
- Update reusable workflow callers to use the renamed file.
https://chatgpt.com/share/69659ff3-1414-800f-8652-5298fcf55398
- Add reusable _deploy-matrix workflow with discovery + per-app deploy across distros
- Enforce serial execution via strategy.max-parallel=1 and act --concurrent-jobs 1
- Support filters: mode (server|workstation|rest), only_app, only_distro, include/exclude regex
- Refactor test runner: remove discovery, require --app, deploy single app with deps only
- Generalize discover script to discover-apps.sh with INCLUDE_RE/EXCLUDE_RE + lifecycle gating
- Update Makefile targets to drive workflow_dispatch inputs for local act runs
https://chatgpt.com/share/69659ff3-1414-800f-8652-5298fcf55398
Apply TESTED_LIFECYCLES filtering directly in the deploy test runner so roles
with unsupported lifecycles (e.g. pre-alpha) are excluded consistently for
both workstation and server workflows.
This prevents roles like desk-copyq (lifecycle: pre-alpha) from being deployed
in CI unless explicitly enabled via TESTED_LIFECYCLES.
https://chatgpt.com/share/6963e316-0a50-800f-a895-8828269a0ca3
- Replace direct docker compose up with python-based orchestrator (Compose.up)
- Centralize stack startup, systemd readiness wait, and init logic in Compose.up
- Run container init via scripts/docker/entry.sh instead of make target
- Remove duplicate systemd wait in deploy test runner
- Add cli.deploy.test.up entrypoint for make up integration
https://chatgpt.com/share/6962a2c8-0f50-800f-8bfb-2efd16adaf40
Ensure stdout is captured when calling run_after_resolution inside the infinito
container so declared galaxy_info.run_after deps (e.g. matomo, mailu) are
correctly detected and added to deploy_ids.
https://chatgpt.com/share/69617fbb-5af4-800f-848d-edeb2c1125a8
Add a systemd readiness gate to the CI compose wrapper to avoid a race
where systemctl is called before DBus/systemd is fully ready.
This prevents sporadic failures like:
"System has not been booted with systemd as init system (PID 1)".
No Dockerfile or Ansible role changes required.
https://chatgpt.com/share/696045b5-bd08-800f-acd9-e3a4e2e23e52
Resolve run_after dependencies via Compose.exec inside the infinito
CI container instead of invoking the resolver on the host.
This ensures:
- consistent runtime environment (PYTHONPATH, installed code)
- correct resolution when local and container sources differ
- alignment with other compose-based CI execution paths
https://chatgpt.com/share/69603903-d158-800f-815d-c506e3a3a44a
Avoid returning from finally in the test runner, which previously
overrode non-zero exit codes when --keep-stack-on-failure was set.
The compose stack is now conditionally kept without affecting the
actual test result.
https://chatgpt.com/share/696032fd-57d8-800f-bcc8-a0edf7b829a4
- Restore lifecycle-based role exclusion via cli.meta.roles.lifecycle_filter
- Make tested lifecycles configurable through TESTED_LIFECYCLES env
- Apply lifecycle gating consistently for workstation and server tests
- Fix minor bug in invokable parsing (avoid shadowing built-in list)
- Improve logging with lifecycle exclusion details
https://chatgpt.com/share/695fbcf4-2148-800f-8685-56ee73a9bae0
- Print full raw inventory file when application_id validation fails
- Fix invalid reason detection to use in_roles / in_inventory flags
- Simplify CI debugging for per-app dedicated deploy tests
https://chatgpt.com/share/695fb5d2-801c-800f-a1c8-06c2bcace450
* Avoid GitHub Actions matrix job limits by looping distros inside the per-app job
* Simplify discover output emission and pin checkout/upload-artifact to stable versions
* Tidy long log lines and minor formatting in deploy/test helpers
https://chatgpt.com/share/695fa9b9-4460-800f-bcc3-d566b83649b2
* Add `cli.deploy.test` module to orchestrate compose-based CI deploy tests
* Support server per-app runs (including run_after dependencies) and workstation selection
* Add optional image build flags (--no-cache, --missing) and stack retention on failure
* Simplify `scripts/tests/deploy.sh` to a thin wrapper that forwards args to the Python runner
https://chatgpt.com/share/695f9e22-6e9c-800f-9a50-65f553619af2
This change removes --network=host usage, introduces DOCKER_IN_CONTAINER
for reliable environment detection, and replaces hardcoded localhost
bindings with an explicit DOCKER_BIND_HOST abstraction to avoid nftables
issues in CI/DiD environments.
https://chatgpt.com/share/69497616-7b00-800f-8fa3-ddc6eb9c5e98
- Mount /tmp/gh-action into the CI container so Docker can access run-scoped paths
- Scope OpenResty www and nginx volumes to GITHUB_RUN_ID when running in container
- Avoid leaking host /etc/nginx and /var/www into CI runs
- Remove redundant preflight
https://chatgpt.com/share/694918d0-53b4-800f-93a7-ec82946f3e24
- Move OpenResty restart logic into included handler tasks
- Use block/rescue to dump container logs on failure
- Tolerate restarting state during pre-flight validation
- Wait up to 2 minutes for container to reach running state
- Add explicit change detection for docker restart
- Enable assert mode in container deploy CLI
https://chatgpt.com/share/6948fa05-1970-800f-a010-59c8c3588713
Create a dummy authorized_keys file inside the CI container inventory
after inventory generation and before ansible-playbook execution.
This prevents sys-bkp-provider-user from failing on missing file lookups
while keeping the role logic unchanged and the deploy tests deterministic.
* Remove legacy **main** re-export wrappers to avoid `runpy` "already in sys.modules" warnings
* Introduce `cli.deploy.container.command` as the stable implementation module
* Reduce `cli.deploy.container.__main__` to a minimal entrypoint delegating to `command.main`
* Re-export container helpers from package `__init__` without importing `__main__`
* Update unit tests to import/patch the new command module and keep CLI dispatch stable
https://chatgpt.com/share/69484a22-60dc-800f-a23d-068d7c45fe37
Stop importing __main__ from package __init__.py to prevent runpy RuntimeWarning when using `python -m`.
- Move CLI logic into command.py modules
- Keep __main__.py as thin wrappers
- Update unit tests to import command modules directly
https://chatgpt.com/share/6947fb8e-550c-800f-8323-e36f92cb5bba
- Split dedicated deploy logic into focused modules (apps, modes, paths, proc, runner)
- Ensure all make targets run from the repository root via explicit cwd handling
- Fix "make: No rule to make target" errors caused by incorrect working directory
- Slim down __main__ to a thin CLI entrypoint
- Rework unit tests into one file per module and expand coverage
https://chatgpt.com/share/69472632-9740-800f-962c-bcb4474963d6
- Switch Makefile targets to direct module execution (cli.build.*)
- Remove obsolete flags and normalize tree invocation
- Minor formatting cleanups for readability and test stability
- Keep inventory and include paths consistent after package layout migration
https://chatgpt.com/share/6946eac4-eb48-800f-becb-f08420ab03a6
- Remove clean-keep-logs target and enforce git-repo guard in Makefile
- Always use `make clean` during deploy cleanup with live output
- Add shared subprocess runner for consistent stdout/stderr handling
- Align unit tests with simplified cleanup semantics
- Fix .dockerignore to ignore build artifacts and Ruff cache correctly
https://chatgpt.com/share/6945784a-daac-800f-bc02-d7e8d3087376
- Install Docker CLI in the Docker image (distro-aware)
- Switch deploy container runner from Docker-in-Docker to host /var/run/docker.sock
with clearer diagnostics when docker/permissions are missing
- Rewrite scripts/tests/deploy.sh to build a local infinito-<distro> image via Make
and run deployment strictly against that local tag (no pkgmgr/GHCR pulls)
https://chatgpt.com/share/694513c7-daec-800f-a9b5-c8fb25274958
- Introduce a dedicated Python virtualenv (deps target) and run all setup scripts through it
- Fix missing PyYAML errors in clean, CI, and Nix environments
- Refactor build defaults into cli/setup for clearer semantics
- Make setup deterministic and independent from system Python
- Replace early Makefile shell expansion with runtime evaluation
- Rename messy-test to test-messy and update deploy logic and tests accordingly
- Keep setup and test targets consistent across Makefile, CLI, and unit tests
https://chatgpt.com/share/693de226-00ac-800f-8cbd-06552b2f283c
- Replace inline shell scripts with Python-driven orchestration
- Introduce inner dockerd started via detached docker run
- Add docker exec–based inventory creation and dedicated deploy execution
- Add automatic vault password generation for CI
- Update GitHub Actions workflow to use new container deploy wrapper
- Add complete unit test suite for container deploy behavior
- Fix CLI argument forwarding and ensure single docker run call
Conversation reference:
https://chatgpt.com/share/6931c45d-4e40-800f-852f-6c9b1f7dc281
kevinveenbirkenbach