Skip to content

[quality] test: unit coverage for files/os/sysupdate.d transfers (60-uki, 70-k0s) + release-artifact name contract - #52

Closed
hivecommons-hive[bot] wants to merge 1 commit into
mainfrom
quality/test-sysupdate-transfers
Closed

hivecommons-hive[bot] wants to merge 1 commit into
mainfrom
quality/test-sysupdate-transfers

Conversation

@hivecommons-hive

Copy link
Copy Markdown
Contributor

Test Improvement

New file: tests/unit/test_sysupdate_transfers.py (23 cases). No production files touched.

Claimed ground (exact)

  • Files under test: files/os/sysupdate.d/60-uki.transfer, files/os/sysupdate.d/70-k0s.transfer, files/os/sysupdate.d/50-root.transfer (source-side only), files/k0s/sysext/extension-release.k0s
  • Cross-referenced (read-only, no assertions on changed logic): artifact-name lines under elements/**/*.bst
  • New test file only: tests/unit/test_sysupdate_transfers.py

Gap this closes

tests/unit/test_repart_layout.py pins only the target side of 50-root.transfer against the installer repart config. 60-uki.transfer and 70-k0s.transfer had zero coverage, and no test pinned the source MatchPattern of any transfer to an artifact the build actually emits. Renaming a release asset in elements/oci/*.bst breaks OTA silently: systemd-sysupdate finds no candidate and reports the system up to date forever.

What is asserted

  • every transfer parses and carries [Transfer]/[Source]/[Target]; filenames are NN-lowercase.transfer with unique ordering prefixes
  • every [Source] is Type=url-file pointing at this repo's own https release feed (origin-drift guard)
  • every source MatchPattern corresponds to an artifact some element emits: bluefin-server-ddi-<v>.raw.zst, bluefin-server-<v>.efi, k0s-<v>.raw.zst
  • UKI: Type=file, Path=/efi/EFI/Linux, .efi extension preserved, source/target names agree
  • k0s sysext: Type=regular-file, Path=/var/lib/extensions, Mode=0644, stable CurrentSymlink=k0s.raw, and target is the decompressed form of the downloaded .zst
  • extension-release.k0s filename and NAME= match the merged image name k0s, and ID=_any (a mismatch makes systemd-sysext reject the merge at boot)

Verification

  • python3 -m pytest tests/unit -q → 100 passed, 1 xfailed (23 new)
  • mutation check: renaming CurrentSymlink and the k0s asset pattern fails 4 of the new tests, then passes again on revert

Disjointness

Only other open PR on this repo is #27 (elements/installer/installer-stack.bst, elements/oci/bluefin-server-installer.bst, docs/skills/ddi-installer.md) — no overlap with files/os/sysupdate.d/, files/k0s/sysext/, or tests/. The element artifact-name lookup here scans elements/** generically rather than pinning bluefin-server-installer.bst line numbers, so #27 cannot conflict.


Filed by quality agent (hold-gated mode). Human review required — do not merge, do not remove hold.

— hive: agent=quality backend=copilot model=claude-opus-5

Adds tests/unit/test_sysupdate_transfers.py: 23 cases covering
60-uki.transfer and 70-k0s.transfer, which had zero coverage, plus the
source-side artifact-name contract for all three transfers.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: sec-check[bot] <sec-check[bot]@users.noreply.github.com>
@hivecommons-hive hivecommons-hive Bot added the hold Indicates that a PR should not merge because someone has issued a /hold command. label Sep 7, 2026
@castrojo

castrojo commented Sep 7, 2026

Copy link
Copy Markdown
Contributor

Merged into main via #59. Thanks!

@castrojo castrojo closed this Sep 7, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

hold Indicates that a PR should not merge because someone has issued a /hold command.

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant