Merge pull request qdrant#238 from qdrant/enable-index-while-uploading

Enable indexing while uploading

Author: xzfc

engine/clients/qdrant/configure.py

@@ -69,18 +69,16 @@ def recreate(self, dataset: Dataset, collection_params):
         if not set(payload_index_params.keys()).issubset(dataset.config.schema.keys()):
             raise ValueError("payload_index_params are not found in dataset schema")
 
+        optimizers_config = self.collection_params.setdefault("optimizers_config", {})
+        # By default, disable index building while uploading
+        optimizers_config.setdefault("max_optimization_threads", 0)
+
         self.client.recreate_collection(
             collection_name=QDRANT_COLLECTION_NAME,
             **vectors_config,
             **self.collection_params
         )
-        self.client.update_collection(
-            collection_name=QDRANT_COLLECTION_NAME,
-            optimizer_config=rest.OptimizersConfigDiff(
-                # indexing_threshold=10000000,
-                max_optimization_threads=0,
-            ),
-        )
+
         for field_name, field_type in dataset.config.schema.items():
             if field_type in ["keyword", "uuid"]:
                 is_tenant = payload_index_params.get(field_name, {}).get(

engine/clients/qdrant/upload.py

@@ -58,14 +58,17 @@ def upload_batch(cls, batch: List[Record]):
 
     @classmethod
     def post_upload(cls, _distance):
-        cls.client.update_collection(
-            collection_name=QDRANT_COLLECTION_NAME,
-            optimizer_config=OptimizersConfigDiff(
-                # indexing_threshold=10_000,
-                # Set to a high number to not apply limits, already limited by CPU budget
-                max_optimization_threads=100_000,
-            ),
-        )
+        # If index building is disabled through the collection settings, enable it
+        collection = cls.client.get_collection(collection_name=QDRANT_COLLECTION_NAME)
+        if collection.config.optimizer_config.max_optimization_threads == 0:
+            cls.client.update_collection(
+                collection_name=QDRANT_COLLECTION_NAME,
+                optimizer_config=OptimizersConfigDiff(
+                    # indexing_threshold=10_000,
+                    # Set to a high number to not apply limits, already limited by CPU budget
+                    max_optimization_threads=100_000,
+                ),
+            )
 
         cls.wait_collection_green()
         return {}

experiments/configurations/qdrant-on-disk.json

@@ -17,7 +17,10 @@
     "engine": "qdrant",
     "connection_params": {},
     "collection_params": {
-      "optimizers_config": { "default_segment_number": 17 },
+      "optimizers_config": {
+        "default_segment_number": 17,
+        "max_optimization_threads": null
+      },
       "quantization_config": { "scalar": {"type": "int8", "quantile": 0.99, "always_ram": false} },
       "vectors_config": { "on_disk": true },
       "hnsw_config": { "on_disk": true, "m": 0, "payload_m": 16 },
@@ -36,7 +39,10 @@
     "engine": "qdrant",
     "connection_params": {},
     "collection_params": {
-      "optimizers_config": { "default_segment_number": 17 },
+      "optimizers_config": {
+        "default_segment_number": 17,
+        "max_optimization_threads": null
+      },
       "quantization_config": { "scalar": {"type": "int8", "quantile": 0.99, "always_ram": false} },
       "vectors_config": { "on_disk": true },
       "hnsw_config": { "on_disk": true, "m": 0, "payload_m": 16 },

experiments/configurations/qdrant-single-node.json

@@ -29,7 +29,8 @@
       "optimizers_config": {
         "max_segment_size": 1000000,
         "default_segment_number": 3,
-        "memmap_threshold": 10000000
+        "memmap_threshold": 10000000,
+        "max_optimization_threads": null
       }
     },
     "search_params": [
@@ -63,7 +64,8 @@
       "optimizers_config": {
         "max_segment_size": 1000000,
         "default_segment_number": 3,
-        "memmap_threshold": 10000000
+        "memmap_threshold": 10000000,
+        "max_optimization_threads": null
       }
     },
     "search_params": [
@@ -93,7 +95,8 @@
       "optimizers_config": {
         "max_segment_size": 1000000,
         "default_segment_number": 3,
-        "memmap_threshold": 10000000
+        "memmap_threshold": 10000000,
+        "max_optimization_threads": null
       }
     },
     "search_params": [
@@ -120,7 +123,8 @@
       "optimizers_config": {
         "max_segment_size": 1000000,
         "default_segment_number": 3,
-        "memmap_threshold": 10000000
+        "memmap_threshold": 10000000,
+        "max_optimization_threads": null
       }
     },
     "search_params": [