You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Re-validate immutable bootc/rpm-ostree status scenarios on a current image #838
Blocked re-validation of two @pending immutable-OS scenarios from #833 that need a live QEMU bootc VM to re-run.
These two scenarios in tests/common/features/common_immutable.feature cite blockers that predate recent bootc releases; status population in a fresh QEMU install has likely improved, but nothing can re-check them without a manual boot:
No RPM packages are layered on the base image — rpm-ostree status --json reportedly fails in a fresh QEMU bootc install, so the test cannot distinguish "no layers" from "no answer".
bootc status shows a pinned image — bootc status reportedly not reliably populated in a fresh QEMU install.
Reproduce / re-validate
Boot a current ghcr.io/projectbluefin/bluefin:testing (or :latest) image in QEMU, SSH in as the test user, and run manually:
rpm-ostree status --json
bootc status --json
If both populate correctly, remove the @pending tag from each scenario and re-run the common suite. If either still fails, re-anchor that scenario's @pending comment to a fresh issue and leave it pending with the specific failure captured.
Originally flagged in #833 (no tracking issue existed for these two blockers).
Re-validation of these two @pending immutable-OS scenarios requires booting a current Bluefin image (e.g. ghcr.io/projectbluefin/bluefin:testing) in QEMU and running rpm-ostree status --json / bootc status --json, then either un-pending the scenarios (if status populates) or re-anchoring the @pending comment to a fresh issue (if it does not). That decision cannot be made from this environment:
No boot capability here — no /dev/kvm, no qemu-system-*, no virsh/libvirt, no docker/podman, and no pre-pulled Bluefin image. rpm-ostree status / bootc status read booted-ostree runtime state, so a container pull is not a substitute for a live boot. The full image boot only happens in the separate e2e.yml workflow (triggered by projectbluefin/lab against a target-image); pr-validate.yml only lints, checks quarantine age, and checks the coverage snapshot — it never boots an image.
Un-pending without that boot would violate Re-validate immutable bootc/rpm-ostree status scenarios on a current image #838's explicit "re-validate on a current image" requirement (a tag-only PR has no evidence behind it), and it is also not cleanly shippable right now: un-pending changes the common suite counts, which forces regenerating the coverage snapshot in docs/skills/test-authoring/suite-map/SKILL.md, and the runtime-gate pattern this repo uses (@requires_* in before_scenario) needs tests/common/features/environment.py. Both files are owned by open PRs chore: remove bluefinctl test coverage #822 (DIRTY) and test(smoke): skip @vm_only scenarios in container lanes #740 (suite-map/SKILL.md), so a disjoint PR can't be opened until they land.
Waiting on: a current Bluefin image booted in the lab (projectbluefin/lab build/VM) so the two scenarios can be run and the pass/fail decision made. Once that's available (and #822/#740 have merged, freeing environment.py and the coverage snapshot), the fix is minimal: remove @pending from the two scenarios and regenerate the coverage snapshot.
Summary: The requested changes are prepared so the two image checks can be tried on a freshly started system. They are not yet verified on a running image, and repository rules prevent opening a second pull request that changes the same file as an existing one.
What I need from you: Resolve the overlapping pull request, then open and run the prepared change before merging it.
Prepared branch: Karibhou:hive/issue-838-immutable-validation, commit d397ece4. It removes the two pending tags and regenerates the coverage snapshot. Open PR #1076 also modifies docs/skills/test-authoring/suite-map/SKILL.md; repository policy requires waiting for it to land or closing the overlap before opening another PR.
Local checks: common-suite Behave dry run, Ruff, coverage snapshot check, documentation validation and skill index check passed. The live run is still required: dispatch manual.yml with suites=common, test_ref set to the PR head, and image=dakota:testing pinned by digest, as approved in the issue; do not merge before it passes.
The PR body says this is partial/non-closing work (Refs/Part of or leave-open wording), so Hive will not add a closing keyword. GitHub will not show its issue-list PR icon without a closing keyword; the stage comment is the link.
Re-validated on a current image: manual run 42923513204 (ghcr.io/projectbluefin/utah:testing, common, with the #1138 mount fix) booted and both scenarios passed: 'No RPM packages are layered on the base image' and 'bootc status shows a pinned image'. common-a 52 passed, common-b 36 passed, no scenario failures. Nothing left to do here.
Activity
test(common): activate podman user socket scenario and track immutabl…
test(common): activate podman user socket scenario and track immutabl…
test(common): activate podman user socket scenario and track immutabl…
kylerankin commentedon Sep 21, 2026
Re-validation of these two
@pendingimmutable-OS scenarios requires booting a current Bluefin image (e.g.ghcr.io/projectbluefin/bluefin:testing) in QEMU and runningrpm-ostree status --json/bootc status --json, then either un-pending the scenarios (if status populates) or re-anchoring the@pendingcomment to a fresh issue (if it does not). That decision cannot be made from this environment:/dev/kvm, noqemu-system-*, novirsh/libvirt, nodocker/podman, and no pre-pulled Bluefin image.rpm-ostree status/bootc statusread booted-ostree runtime state, so a container pull is not a substitute for a live boot. The full image boot only happens in the separatee2e.ymlworkflow (triggered byprojectbluefin/labagainst atarget-image);pr-validate.ymlonly lints, checks quarantine age, and checks the coverage snapshot — it never boots an image.commonsuite counts, which forces regenerating the coverage snapshot indocs/skills/test-authoring/suite-map/SKILL.md, and the runtime-gate pattern this repo uses (@requires_*inbefore_scenario) needstests/common/features/environment.py. Both files are owned by open PRs chore: remove bluefinctl test coverage #822 (DIRTY) and test(smoke): skip @vm_only scenarios in container lanes #740 (suite-map/SKILL.md), so a disjoint PR can't be opened until they land.Waiting on: a current Bluefin image booted in the lab (
projectbluefin/labbuild/VM) so the two scenarios can be run and the pass/fail decision made. Once that's available (and #822/#740 have merged, freeingenvironment.pyand the coverage snapshot), the fix is minimal: remove@pendingfrom the two scenarios and regenerate the coverage snapshot.— hive: backend=pi model=lemonade/Ornith-1.5-35B-A3B-GGUF-Q6_K
🐝 Hive Agent:
contributor| SHA:db329ee6hivecommons-hive commentedon Oct 22, 2026
Unblocked: the
blockedlabel was applied by the hive without a blocked-on marker, so nothing could lift it. This issue is actionable again.64 remaining items
Karibhou commentedon Nov 6, 2026
Summary: The requested changes are prepared so the two image checks can be tried on a freshly started system. They are not yet verified on a running image, and repository rules prevent opening a second pull request that changes the same file as an existing one.
What I need from you: Resolve the overlapping pull request, then open and run the prepared change before merging it.
Prepared branch:
Karibhou:hive/issue-838-immutable-validation, commitd397ece4. It removes the two pending tags and regenerates the coverage snapshot. Open PR #1076 also modifiesdocs/skills/test-authoring/suite-map/SKILL.md; repository policy requires waiting for it to land or closing the overlap before opening another PR.Local checks: common-suite Behave dry run, Ruff, coverage snapshot check, documentation validation and skill index check passed. The live run is still required: dispatch
manual.ymlwithsuites=common,test_refset to the PR head, andimage=dakota:testingpinned by digest, as approved in the issue; do not merge before it passes.— hive: backend=omp model=openai-codex/gpt-6.1-terra
🐝 Hive Agent:
contributor| SHA:d397ece4docs: drop immutability checks from common pending note
hivecommons-hive commentedon Nov 9, 2026
PR #1116 covers this issue: #1116
The PR body says this is partial/non-closing work (
Refs/Part ofor leave-open wording), so Hive will not add a closing keyword. GitHub will not show its issue-list PR icon without a closing keyword; the stage comment is the link.test: prepare two system checks for validation
Danathar commentedon Nov 11, 2026
Re-validated on a current image: manual run 42923513204 (ghcr.io/projectbluefin/utah:testing, common, with the #1138 mount fix) booted and both scenarios passed: 'No RPM packages are layered on the base image' and 'bootc status shows a pinned image'. common-a 52 passed, common-b 36 passed, no scenario failures. Nothing left to do here.
ci: boot the suites a pull request affects before it merges (#1142)