[BUG] fix persistent HNSW parameter migration

chromadb/db/mixins/sysdb.py

@@ -19,6 +19,7 @@
     UniqueConstraintError,
 )
 from chromadb.db.system import SysDB
+from chromadb.segment.impl.vector.hnsw_params import PersistentHnswParams
 from chromadb.telemetry.opentelemetry import (
     add_attributes_to_current_span,
     OpenTelemetryClient,
@@ -776,7 +777,12 @@ def _insert_config_from_legacy_params(
         collections_t = Table("collections")
 
         # Get any existing HNSW params from the metadata
-        hnsw_metadata_params = HnswParams.extract(metadata or {})
+        metadata = metadata or {}
+        if metadata.get("hnsw:batch_size") or metadata.get("hnsw:sync_threshold"):
+            hnsw_metadata_params = PersistentHnswParams.extract(metadata)
+        else:
+            hnsw_metadata_params = HnswParams.extract(metadata)
+
         hnsw_configuration = HNSWConfigurationInternal.from_legacy_params(
             hnsw_metadata_params  # type: ignore[arg-type]
         )

chromadb/test/property/strategies.py

@@ -281,7 +281,7 @@ def collections(
     with_hnsw_params: bool = False,
     has_embeddings: Optional[bool] = None,
     has_documents: Optional[bool] = None,
-    with_persistent_hnsw_params: bool = False,
+    with_persistent_hnsw_params: st.SearchStrategy[bool] = st.just(False),
     max_hnsw_batch_size: int = 2000,
     max_hnsw_sync_threshold: int = 2000,
 ) -> Collection:
@@ -294,16 +294,18 @@ def collections(
     dimension = draw(st.integers(min_value=2, max_value=2048))
     dtype = draw(st.sampled_from(float_types))
 
-    if with_persistent_hnsw_params and not with_hnsw_params:
+    use_persistent_hnsw_params = draw(with_persistent_hnsw_params)
+
+    if use_persistent_hnsw_params and not with_hnsw_params:
         raise ValueError(
-            "with_hnsw_params requires with_persistent_hnsw_params to be true"
+            "with_persistent_hnsw_params requires with_hnsw_params to be true"
         )
 
     if with_hnsw_params:
         if metadata is None:
             metadata = {}
         metadata.update(test_hnsw_config)
-        if with_persistent_hnsw_params:
+        if use_persistent_hnsw_params:
             metadata["hnsw:batch_size"] = draw(
                 st.integers(min_value=3, max_value=max_hnsw_batch_size)
             )

chromadb/test/property/test_cross_version_persist.py

@@ -272,7 +272,12 @@ def persist_generated_data_with_old_version(
 
 # Since we can't pickle the embedding function, we always generate record sets with embeddings
 collection_st: st.SearchStrategy[strategies.Collection] = st.shared(
-    strategies.collections(with_hnsw_params=True, has_embeddings=True), key="coll"
+    strategies.collections(
+        with_hnsw_params=True,
+        has_embeddings=True,
+        with_persistent_hnsw_params=st.booleans(),
+    ),
+    key="coll",
 )
 
 

chromadb/test/property/test_persist.py

@@ -61,7 +61,7 @@ def settings(request: pytest.FixtureRequest) -> Generator[Settings, None, None]:
 collection_st = st.shared(
     strategies.collections(
         with_hnsw_params=True,
-        with_persistent_hnsw_params=True,
+        with_persistent_hnsw_params=st.just(True),
         # Makes it more likely to find persist-related bugs (by default these are set to 2000).
         max_hnsw_batch_size=10,
         max_hnsw_sync_threshold=10,