4 Commits

Include renames

Author	SHA1	Message	Date
Lukas GreveandClaude Opus 5.5	f7b6623b25	ci: smoke-test built images; build on labeled PRs, weekly and on demand ... smoke-test.sh checks an image from build-image.sh in two steps: - static, read-only: if the image declares SELinux, no kernel command line (/etc/kernel/cmdline, ESP and /boot loader entries) may carry selinux=0 and files must be labeled. This catches images built on a host with SELinux disabled, where anaconda inherits the host's selinux=0 and overrides the kickstart's `selinux --enforcing` (CI run 237); - boot: KVM with -snapshot, no network, waits for qemu-guest-agent to answer, so it does not depend on a serial console in the image. Verified on the runner: the run-237 image fails the static check, a fresh build passes both (agent answered after 27s). build-image.yaml now smoke-tests each edition before compressing it, and besides v*.*.* tags also runs: - on PRs into main labeled `build-image` (the review gate: Gitea does not let authors approve their own PRs), re-running on new pushes while the label stays, only for image-affecting paths, and not when an unrelated label is added; - weekly (Mondays 03:00), to catch upstream Fedora breakage; - via workflow_dispatch. Non-tag runs keep the images as 7-day artifacts. README updated, including the stale "every push to main" sentence. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>	2026-09-23 20:04:32 +02:00
Lukas GreveandClaude Opus 5.5	41eb18cc70	ci: build images only on release tags, not on every push to main ... A full build of both editions takes ~25 min on the fedora:host runner (run 237: 10:46-11:10), too costly to repeat on every push to main. build-image.yaml now triggers only on v*.*.* tags; the tag path already publishes the image and kickstart as a release, so the main-only upload-artifact step is dropped. Recipes are still linted and validated on every push by ci.yml. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>	2026-09-23 13:42:39 +02:00
Lukas GreveandClaude Sonnet 5	b8cd2966f4	build-image: serialize host-wide with flock, clean up stale mounts/devices ... Gitea runs each matrix entry as its own task, so build-image.yaml's `max-parallel: 1` never serialized the two editions -- both builds started at once on the fedora:host runner and corrupted each other's global anaconda/livemedia-creator state (/mnt/sysroot, /mnt/sysimage, loop and device-mapper devices), surfacing as bootctl failing with "Couldn't find EFI system partition" even though the dish's EFI partition was fine. build-image.sh now takes a host-wide flock around livemedia-creator, and under the lock first clears any stale mount/dm/loop device a previously failed build left behind. Verified on the runner: both editions launched concurrently now run serially under the lock and build successfully. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>	2026-09-23 12:46:08 +02:00
Lukas GreveandClaude Sonnet 5	67316d9852	cook: two default dishes (phyllomeos, phyllomeos-headless) + guest/experimental tiers; CI builds raw images ... Manifest groups get a tier; only `default` is generated (make all TIER=..., --tier). Default: phyllomeos (GNOME + virt-manager) and phyllomeos-headless, Fedora 44, systemd-boot, CPU-agnostic hypervisor ingredient (hypervisor_type: any). guest tier: guest-server, guest-desktop. experimental: biosboot/grub, encrypted, rawhide. Generator clears stale recipes/dishes before writing. build-image.sh: --dish and --tier; deploy.sh: --tier. build-iso.yaml replaced by build-image.yaml: raw images for both editions on the fedora:host runner (TMPDIR=/var/tmp, max-parallel 1), attached with the flattened kickstarts to tagged releases. Live edition no longer built. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>	2026-09-20 11:35:28 +02:00