fix(scanner): retain raw enumeration quantum

Co-Authored-By: heihutu <heihutu@gmail.com>

Co-Authored-By: zhi22915 <qiuzgang@gmail.com>

scripts/diagnose_scanner_enumeration_restart.py

@@ -85,15 +85,20 @@ def validate_recoverable_quantum(reports, *, objects, budget, require_converged)
         raise ValueError("no scanner restart reports were produced")
     previous = None
     made_enumeration_progress = False
+    made_raw_page_commit_progress = False
     made_classification_progress = False
     made_durable_progress = False
     for index, report in enumerate(reports):
         validate_report(report, round_number=index, pid=report["pid"], objects=objects, budget=budget)
+        if report["raw_page_index_parent"] == "bucket" and report["raw_page_index_committed_entries"] > 0:
+            made_raw_page_commit_progress = True
         if previous is not None:
             if report["objects_before"] != previous["objects_retained"]:
                 raise ValueError("durable retained coverage did not survive process restart")
             if report["objects_retained"] < previous["objects_retained"]:
                 raise ValueError("durable retained coverage regressed across restart")
+            if replays_raw_window(previous, report):
+                raise ValueError("raw enumeration window replayed without durable coverage")
             if (report["raw_page_index_parent"] == previous["raw_page_index_parent"]
                     and report["raw_page_index_committed_entries"] < previous["raw_page_index_committed_entries"]
                     and not previous["raw_page_index_complete"]):
@@ -104,6 +109,8 @@ def validate_recoverable_quantum(reports, *, objects, budget, require_converged)
         previous = report
     if not made_enumeration_progress:
         raise ValueError("restart proof did not exercise raw enumeration")
+    if not made_raw_page_commit_progress:
+        raise ValueError("restart proof did not commit a durable raw enumeration page")
     if not made_classification_progress:
         raise ValueError("restart proof did not exercise object classification")
     if not made_durable_progress:

scripts/test_diagnose_scanner_enumeration_restart.py

@@ -140,6 +140,15 @@ class ReportTests(unittest.TestCase):
         advanced = dict(current, objects_retained=1)
         self.assertFalse(replays_raw_window(previous, advanced))
 
+    def test_recoverable_quantum_rejects_replayed_raw_window(self):
+        previous = self.report()
+        previous.update(objects_retained=0, versions_retained=0, bytes_retained=0,
+                        objects_processed=0, snapshot_complete=False, outcome="partial")
+        current = dict(previous, round=1, pid=124, objects_before=0)
+
+        with self.assertRaisesRegex(ValueError, "raw enumeration window replayed"):
+            validate_recoverable_quantum([previous, current], objects=4, budget=16, require_converged=False)
+
     def test_recoverable_quantum_requires_three_stage_progress_and_convergence(self):
         first = self.report()
         first.update(round=0, pid=123, raw_entries=2, raw_page_index_committed_entries=2,
@@ -193,6 +202,15 @@ class ReportTests(unittest.TestCase):
         with self.assertRaisesRegex(ValueError, "object classification"):
             validate_recoverable_quantum([report], objects=4, budget=16, require_converged=False)
 
+    def test_recoverable_quantum_rejects_missing_raw_page_commit(self):
+        report = self.report()
+        report.update(snapshot_complete=False, outcome="partial",
+                      raw_page_index_committed_entries=0, raw_page_index_indexed_entries=1,
+                      objects_retained=1, versions_retained=1, bytes_retained=1)
+
+        with self.assertRaisesRegex(ValueError, "durable raw enumeration page"):
+            validate_recoverable_quantum([report], objects=4, budget=16, require_converged=False)
+
 
 if __name__ == "__main__":
     unittest.main()