fix: prevent fresh+reconsolidation item name collision in CEFS consolidation (#2059)

Co-authored-by: mattgodbolt-molty <mattgodbolt-molty@users.noreply.github.com>

Author: mattgodbolt-molty

bin/lib/cefs/consolidation.py

@@ -683,11 +683,25 @@ def prepare_consolidation_items(
     """
     items_for_consolidation = []
     subdir_mapping = {}
+    seen_subdir_names: dict[str, str] = {}  # subdir_name -> item.name, for duplicate detection
 
     for item in group:
         # Use the installable name as subdirectory name (sanitized for filesystem)
         subdir_name = sanitize_path_for_filename(Path(item.name))
 
+        # Guard against duplicate subdir names within the same group. This can happen when a
+        # fresh item and a reconsolidation item have the same installable name — both produce
+        # the same sanitized subdir_name and parallel workers would race to write to the same
+        # extraction directory (OSError: Directory not empty).
+        if subdir_name in seen_subdir_names:
+            raise ValueError(
+                f"Duplicate subdir name '{subdir_name}' in consolidation group: "
+                f"'{item.name}' (from_reconsolidation={item.from_reconsolidation}) conflicts with "
+                f"'{seen_subdir_names[subdir_name]}'. "
+                f"Fresh items and reconsolidation items with the same name must not appear in the same group."
+            )
+        seen_subdir_names[subdir_name] = item.name
+
         # For reconsolidation items, we already have the extraction path
         if item.from_reconsolidation:
             extraction_path = item.extraction_path

bin/lib/cli/cefs.py

@@ -596,8 +596,18 @@ def consolidate(
 
         if recon_candidates:
             _LOGGER.info("Found %d items from consolidated images for reconsolidation", len(recon_candidates))
-            # Reconsolidation candidates are already in the right format
-            cefs_items.extend(recon_candidates)
+            # Filter out reconsolidation candidates whose name already appears as a fresh item.
+            # If both a fresh item and a reconsolidation item have the same name they would
+            # produce the same sanitized subdir_name and race when extracted in parallel.
+            fresh_names = {item.name for item in cefs_items}
+            filtered_recon = [c for c in recon_candidates if c.name not in fresh_names]
+            skipped = len(recon_candidates) - len(filtered_recon)
+            if skipped:
+                _LOGGER.info(
+                    "Skipping %d reconsolidation candidate(s) whose name already appears as a fresh item",
+                    skipped,
+                )
+            cefs_items.extend(filtered_recon)
 
     if not cefs_items:
         _LOGGER.warning("No CEFS items found matching filter: %s", " ".join(filter_) if filter_ else "all")

bin/test/cefs/consolidation_test.py

@@ -862,3 +862,50 @@ def test_gather_reconsolidation_candidates(tmp_path):
     assert len(candidates_filtered) == 2
     filtered_names = {c.name for c in candidates_filtered}
     assert filtered_names == {"tools/small 1.0.0", "tools/small 2.0.0"}
+
+
+def test_duplicate_subdir_names_in_consolidation_group_are_rejected(tmp_path):
+    """Regression test: fresh item + reconsolidation item with the same name must not appear
+    in the same consolidation group.
+
+    Production failure (2026-04-10, run #24232582644):
+      OSError: [Errno 39] Directory not empty:
+        .tmp_extract_2d3e5ffc/compilers_swift_static-sdk_sdk-0-1-0_6.2.4
+        -> compilers_swift_static-sdk_sdk-0-1-0_6.2.4
+
+    Both 'compilers/swift/static-sdk/sdk-0-1-0 6.2.4' (fresh install) and a
+    reconsolidation candidate from an existing consolidated image produced the
+    same sanitized subdir_name. Parallel workers raced to write the same
+    directory, and the second found it non-empty.
+    """
+    mount_point = Path("/cefs")
+
+    fresh_path = tmp_path / "swift-static-sdk-6.2.4"
+    cefs_target = mount_point / "7e" / "7e37bea4f759832456d69f12_swift-6.2.4-static-sdk"
+    fresh_path.symlink_to(cefs_target)
+
+    recon_path = tmp_path / "swift-static-sdk-6.2.4-recon"
+
+    group = [
+        ConsolidationCandidate(
+            name="compilers/swift/static-sdk/sdk-0-1-0 6.2.4",
+            nfs_path=fresh_path,
+            squashfs_path=Path("/efs/cefs-images/7e/7e37bea4f759832456d69f12_swift-6.2.4-static-sdk.sqfs"),
+            size=238_000_000,
+            from_reconsolidation=False,
+        ),
+        ConsolidationCandidate(
+            name="compilers/swift/static-sdk/sdk-0-1-0 6.2.4",  # same name as above
+            nfs_path=recon_path,
+            squashfs_path=Path("/efs/cefs-images/26/26ccf7a2fd1f9d4667de2307_consolidated.sqfs"),
+            size=50_000_000,
+            extraction_path=Path("compilers_swift_static-sdk_sdk-0-1-0_6.2.4"),
+            from_reconsolidation=True,
+        ),
+    ]
+
+    # Both items sanitize to the same subdir_name: 'compilers_swift_static-sdk_sdk-0-1-0_6.2.4'.
+    # prepare_consolidation_items must detect this and raise ValueError — if it silently
+    # returned both, parallel workers would race to extract to the same directory.
+    with pytest.raises(ValueError, match="Duplicate subdir name"):
+        prepare_consolidation_items(group, mount_point)