test(tiers): check for a leftover probe under the lock (#1758)

[JarryShaw]

Please follow the guide below

• You will be asked some questions, please read them carefully and answer honestly

• Put an x into all the boxes [ ] relevant to your pull request (like that [x])

• Use Preview tab to see how your pull request will actually look like

• Searched for similar pull requests

• Followed the coding style (make pylint, make mypy, make isort)

• make test passes, and a test case covers the change. Narrow runs only, listed below.

• Added a changelog entry under docs/source/changelog/ and regenerated CHANGELOG.md, if the change is user-visible: N/A — centralised in docs(changelog): shared 1.5.0 changelog — long-lived, merges last (#610, #616, #617, #618, #620) #657

What is the purpose of your pull request?

Tick the commit type your subject line carries.

• fix — corrects a defect
• feat — adds a feature
• perf — changes performance, not behaviour
• refactor — changes neither behaviour nor performance
• test — tests only
• docs — documentation only
• ci — workflows or build tooling
• release — bumps the version or rolls up a distribution
• chore — anything else

---

Description

Closes #1758. XdistSubprocessTests.setUp in tests/test_tier_guard_xdist.py checked for a leftover probe file before taking the exclusive _tier_guard_lock. Under --dist load, a worker could see another worker's live probe and fail with "already exists".

• Fix: the existence check now runs after lock.__enter__(). The lock release is registered before the assert, so a failing check still releases the lock. The unlink is registered after it, so cleanup still unlinks while the lock is held. ProbeWorkerCrashTests (test(tiers): keep the xdist probe out of recursive collection (#1756) #1757) already did this and is unchanged.
• Test: ProbeLeftoverCheckTests redirects TESTS_ROOT and the lock path into a temporary directory, so it needs neither xdist nor git.
  • It holds the lock with the probe present and asserts that setUp blocks instead of failing, then that it proceeds once the probe is removed and the lock released. On the old ordering it fails with setUp did not wait for the lock: [... already exists ...].
  • A second test asserts that the lock is released when a real leftover makes the check fail.

Tests:

• The two tier-guard modules plus test_unit_tests_queue.py give 154 passed and 3 skipped (the xdist classes skip in the repo venv).
• Under -n 3 --dist load in an xdist venv, they give 119 passed on each of 5 runs.

[JarryShaw]

Verdict: GOOD TO GO at 3033fe4ad. A Sonnet cross-review checked 8c3ea8339 (author: Opus). Rebasing onto 1f5766c28 after #1757 merged left the patch identical (git range-diff: 8c3ea8339 = 3033fe4ad).

• Lock released on every path:
  • skipTest returns before the lock is taken;
  • a failed check releases it, because unittest runs doCleanups after a setUp failure;
  • the test body and cleanup release it.
• The test catches the bug. Mutated copies of the code fail it in each case:
  • old ordering: setUp did not wait for the lock;
  • release registered after the assert: setUp left the exclusive lock held;
  • check dropped: 'already exists' not found.
• No flakes: the fixed code passed 5 reruns, then 40 runs under 8 busy-loop processes.
• test(tiers): keep the xdist probe out of recursive collection (#1756) #1757 untouched: this PR does not change test(tiers): keep the xdist probe out of recursive collection (#1756) #1757's collect_ignore fix or ProbeWorkerCrashTests.
• My own check, on the rebased head: the two tier-guard modules plus test_unit_tests_queue.py give 154 passed and 3 skipped, with pcapkit.__file__ printed inside the PR's worktree.

[github-actions]

Coverage: 89.59% (unit tier, Python 3.15, 3033fe4ad, Unit Tests run success)

Package	Statements	Missed	Branches	Partial	Cover
pcapkit (top level)	104	3	20	3	95.16%
pcapkit/const	18862	954	2368	868	91.06%
pcapkit/corekit	2461	69	780	35	96.54%
pcapkit/dumpkit	276	3	100	3	98.40%
pcapkit/foundation	2915	164	1016	34	93.59%
pcapkit/interface	138	7	52	5	93.68%
pcapkit/protocols	17184	224	4574	189	98.08%
pcapkit/toolkit	872	177	298	8	77.18%
pcapkit/utilities	512	2	158	4	99.10%
pcapkit/vendor	4417	2342	1010	157	43.38%

Per-file detail: the coverage-html artifact of this run.