fix(cron): re-validate stale cron-output entries before deletion (#37721)

quick() and dry_run() previously trusted the stored category from
tracked.json without re-validating at delete time. Stale entries from
before #34840 could carry category="cron-output" for cron control-plane
paths (e.g. cron/jobs.json), causing quick() to delete the live
scheduler registry.

Fix:
- Fix guess_category() to only classify cron/output/** as cron-output
  (was classifying ALL cron/* paths, missing the #34840 fix).
- Re-validate cron-output entries via guess_category() at delete time
  in quick() and dry_run(); stale entries that are no longer classified
  as cron-output are skipped and removed from tracked.json.
- Add _is_protected_cron_path() as a hard defense-in-depth guard that
  blocks deletion of cron/cronjobs directories and known control-plane
  files (jobs.json, .tick.lock) regardless of stored category.
- Update test_cron_subtree_categorised to match fixed guess_category
  (only cron/output/* is cron-output, not all of cron/).

Tests: add 5 regression tests in TestStaleCronEntryMigration.
This commit is contained in:
kyssta-exe
2026-06-04 07:52:04 -07:00
committed by Teknium
parent 693f4c7e9c
commit 30412a9771
2 changed files with 186 additions and 0 deletions
+57
View File
@@ -145,6 +145,33 @@ ALLOWED_CATEGORIES = {
}
# Paths under $HERMES_HOME that must NEVER be deleted by quick(),
# regardless of what the stored category says. This is a defense-in-depth
# guard against stale tracked.json entries from before #34840.
_PROTECTED_CRON_PATHS: set[str] = set()
def _is_protected_cron_path(p: Path) -> bool:
"""Return True if *p* is a cron control-plane file/directory that must
never be deleted.
This only matches the directory itself and known control-plane files
(``jobs.json``, ``.tick.lock``) — it does NOT blanket-protect
everything under ``cron/`` because ``cron/output/`` is disposable.
"""
# Lazily build the set once per process so HERMES_HOME is resolved
# exactly once.
if not _PROTECTED_CRON_PATHS:
hermes_home = get_hermes_home()
for parent in ("cron", "cronjobs"):
base = hermes_home / parent
_PROTECTED_CRON_PATHS.add(str(base))
_PROTECTED_CRON_PATHS.add(str(base / "jobs.json"))
_PROTECTED_CRON_PATHS.add(str(base / ".tick.lock"))
resolved = str(p.resolve())
return resolved in _PROTECTED_CRON_PATHS
def fmt_size(n: float) -> str:
for unit in ("B", "KB", "MB", "GB", "TB"):
if n < 1024:
@@ -226,6 +253,14 @@ def dry_run() -> Tuple[List[Dict], List[Dict]]:
cat = item["category"]
size = item["size"]
# Re-validate stale "cron-output" entries (fixes #37721).
if cat == "cron-output":
re_cat = guess_category(p)
if re_cat != "cron-output":
# Stale entry — would be skipped by quick(); omit from
# dry-run output too.
continue
if cat == "test":
auto.append(item)
elif cat == "temp" and age > 7:
@@ -269,6 +304,28 @@ def quick() -> Dict[str, Any]:
age = (now - datetime.fromisoformat(item["timestamp"])).days
# ---- stale-state migration (fixes #37721) ----
# Old tracked.json entries may carry a "cron-output" category for
# paths that are NOT under cron/output/ (e.g. cron/jobs.json).
# guess_category() was fixed in #34840, but existing entries are
# never re-validated. Re-classify here so stale entries for cron
# control-plane state are not deleted.
if cat == "cron-output":
re_cat = guess_category(p)
if re_cat != "cron-output":
_log(
f"SKIP stale cron-output entry: {p} "
f"(re-classified as {re_cat!r})"
)
# Drop the stale entry — it was misclassified.
continue
# Hard safety net: never delete cron control-plane state even if
# the category somehow slipped through re-validation above.
if _is_protected_cron_path(p):
_log(f"SKIP protected cron path: {p}")
continue
should_delete = (
cat == "test"
or (cat == "temp" and age > 7)