[skip ci] update segment_size to segment_row_count (#3068)

* [skip ci] update segment_size to segment_row_count
Signed-off-by: Nzw <zw@milvus.io>

* [skip ci] remove metric type in collection params
Signed-off-by: Nzw <zw@milvus.io>
Co-authored-by: Nzw <zw@milvus.io>
Co-authored-by: NWang XiangYu <xy.wang@zilliz.com>

tests/milvus_python_test/collection/test_collection_count.py

@@ -15,7 +15,7 @@ dim = 128
 tag = "tag"
 collection_id = "count_collection"
 add_interval_time = 3
-segment_size = 10
+segment_row_count = 5000
 default_fields = gen_default_fields() 
 entities = gen_entities(nb)
 raw_vectors, binary_entities = gen_binary_entities(nb)
@@ -31,7 +31,7 @@ class TestCollectionCount:
         scope="function",
         params=[
             1,
-            5000,
+            4000,
             6001
         ],
     )
@@ -186,7 +186,7 @@ class TestCollectionCountIP:
         scope="function",
         params=[
             1,
-            5000,
+            4000,
             6001
         ],
     )
@@ -341,7 +341,7 @@ class TestCollectionCountBinary:
         scope="function",
         params=[
             1,
-            5000,
+            4000,
             6001
         ],
     )
@@ -507,7 +507,7 @@ class TestCollectionMultiCollections:
         scope="function",
         params=[
             1,
-            5000,
+            4000,
             6001
         ],
     )
@@ -564,7 +564,8 @@ class TestCollectionMultiCollections:
             res = connect.count_entities(collection_list[i])
             assert res == insert_count
 
-    def test_collection_count_multi_collections_binary(self, connect, jac_collection, insert_count):
+    # TODO:
+    def _test_collection_count_multi_collections_binary(self, connect, jac_collection, insert_count):
         '''
         target: test collection rows_count is correct or not with multiple collections of JACCARD
         method: create collection and add entities in it,
@@ -587,7 +588,8 @@ class TestCollectionMultiCollections:
             res = connect.count_entities(collection_list[i])
             assert res == insert_count
 
-    def test_collection_count_multi_collections_mix(self, connect):
+    # TODO:
+    def _test_collection_count_multi_collections_mix(self, connect):
         '''
         target: test collection rows_count is correct or not with multiple collections of JACCARD
         method: create collection and add entities in it,

tests/milvus_python_test/collection/test_collection_logic.py

@@ -7,9 +7,9 @@ from multiprocessing import Process
 from utils import *
 
 dim = 128
-default_segment_size = 1024
+default_segment_row_count = 100000
 drop_collection_interval_time = 3
-segment_size = 10
+segment_row_count = 5000
 collection_id = "logic"
 vectors = gen_vectors(100, dim)
 default_fields = gen_default_fields() 

tests/milvus_python_test/collection/test_collection_stats.py

@@ -7,7 +7,7 @@ import pytest
 from utils import *
 
 dim = 128
-segment_size = 10
+segment_row_count = 5000
 nprobe = 1
 top_k = 1
 epsilon = 0.0001

tests/milvus_python_test/collection/test_create_collection.py

@@ -13,9 +13,9 @@ from utils import *
 nb = 1
 dim = 128
 collection_id = "create_collection"
-default_segment_size = 1024
+default_segment_row_count = 100000
 drop_collection_interval_time = 3
-segment_size = 10
+segment_row_count = 5000
 default_fields = gen_default_fields() 
 entities = gen_entities(nb)
 
@@ -42,9 +42,9 @@ class TestCreateCollection:
 
     @pytest.fixture(
         scope="function",
-        params=gen_segment_sizes()
+        params=gen_segment_row_counts()
     )
-    def get_segment_size(self, request):
+    def get_segment_row_count(self, request):
         yield request.param
 
     def test_create_collection_fields(self, connect, get_filter_field, get_vector_field):
@@ -59,7 +59,7 @@ class TestCreateCollection:
         collection_name = gen_unique_str(collection_id)
         fields = {
                 "fields": [filter_field, vector_field],
-                "segment_size": segment_size
+                "segment_row_count": segment_row_count
         }
         logging.getLogger().info(fields)
         connect.create_collection(collection_name, fields)
@@ -77,20 +77,20 @@ class TestCreateCollection:
         collection_name = gen_unique_str(collection_id)
         fields = {
                 "fields": [filter_field, vector_field],
-                "segment_size": segment_size
+                "segment_row_count": segment_row_count
         }
         connect.create_collection(collection_name, fields)
         assert connect.has_collection(collection_name)
         
-    def test_create_collection_segment_size(self, connect, get_segment_size):
+    def test_create_collection_segment_row_count(self, connect, get_segment_row_count):
         '''
         target: test create normal collection with different fields
-        method: create collection with diff segment_size
+        method: create collection with diff segment_row_count
         expected: no exception raised
         '''
         collection_name = gen_unique_str(collection_id)
         fields = copy.deepcopy(default_fields)
-        fields["segment_size"] = get_segment_size
+        fields["segment_row_count"] = get_segment_row_count
         connect.create_collection(collection_name, fields)
         assert connect.has_collection(collection_name)
 
@@ -197,7 +197,7 @@ class TestCreateCollectionInvalid(object):
         scope="function",
         params=gen_invalid_ints()
     )
-    def get_segment_size(self, request):
+    def get_segment_row_count(self, request):
         yield request.param
 
     @pytest.fixture(
@@ -222,20 +222,20 @@ class TestCreateCollectionInvalid(object):
         yield request.param
 
     @pytest.mark.level(2)
-    def test_create_collection_with_invalid_segment_size(self, connect, get_segment_size):
+    def test_create_collection_with_invalid_segment_row_count(self, connect, get_segment_row_count):
         collection_name = gen_unique_str()
         fields = copy.deepcopy(default_fields)
-        fields["segment_size"] = get_segment_size
+        fields["segment_row_count"] = get_segment_row_count
         with pytest.raises(Exception) as e:
             connect.create_collection(collection_name, fields)
 
-    @pytest.mark.level(2)
-    def test_create_collection_with_invalid_metric_type(self, connect, get_metric_type):
-        collection_name = gen_unique_str()
-        fields = copy.deepcopy(default_fields)
-        fields["fields"][-1]["params"]["metric_type"] = get_metric_type
-        with pytest.raises(Exception) as e:
-            connect.create_collection(collection_name, fields)
+    # @pytest.mark.level(2)
+    # def test_create_collection_with_invalid_metric_type(self, connect, get_metric_type):
+    #     collection_name = gen_unique_str()
+    #     fields = copy.deepcopy(default_fields)
+    #     fields["fields"][-1]["params"]["metric_type"] = get_metric_type
+    #     with pytest.raises(Exception) as e:
+    #         connect.create_collection(collection_name, fields)
 
     @pytest.mark.level(2)
     def test_create_collection_with_invalid_dimension(self, connect, get_dim):
@@ -285,34 +285,33 @@ class TestCreateCollectionInvalid(object):
         with pytest.raises(Exception) as e:
             connect.create_collection(collection_name, fields)
 
-    def test_create_collection_no_segment_size(self, connect):
-        '''
-        target: test create collection with no segment_size params
-        method: create collection with corrent params
-        expected: use default default_segment_size
-        '''
-        collection_name = gen_unique_str(collection_id)
-        fields = copy.deepcopy(default_fields)
-        fields.pop("segment_size")
-        connect.create_collection(collection_name, fields)
-        res = connect.get_collection_info(collection_name)
-        logging.getLogger().info(res)
-        assert res["segment_size"] == default_segment_size
-
-    # TODO:
-    def _test_create_collection_no_metric_type(self, connect):
+    def test_create_collection_no_segment_row_count(self, connect):
         '''
-        target: test create collection with no metric_type params
+        target: test create collection with no segment_row_count params
         method: create collection with corrent params
-        expected: use default L2
+        expected: use default default_segment_row_count
         '''
         collection_name = gen_unique_str(collection_id)
         fields = copy.deepcopy(default_fields)
-        fields["fields"][-1]["params"].pop("metric_type")
+        fields.pop("segment_row_count")
         connect.create_collection(collection_name, fields)
         res = connect.get_collection_info(collection_name)
         logging.getLogger().info(res)
-        assert res["metric_type"] == "L2"
+        assert res["segment_row_count"] == default_segment_row_count
+
+    # def _test_create_collection_no_metric_type(self, connect):
+    #     '''
+    #     target: test create collection with no metric_type params
+    #     method: create collection with corrent params
+    #     expected: use default L2
+    #     '''
+    #     collection_name = gen_unique_str(collection_id)
+    #     fields = copy.deepcopy(default_fields)
+    #     fields["fields"][-1]["params"].pop("metric_type")
+    #     connect.create_collection(collection_name, fields)
+    #     res = connect.get_collection_info(collection_name)
+    #     logging.getLogger().info(res)
+    #     assert res["metric_type"] == "L2"
 
     # TODO: assert exception
     def test_create_collection_limit_fields(self, connect):

tests/milvus_python_test/collection/test_get_collection_info.py

@@ -9,7 +9,7 @@ from utils import *
 
 collection_id = "info"
 default_fields = gen_default_fields() 
-segment_size = 10
+segment_row_count = 5000
 
 
 class TestInfoBase:
@@ -30,9 +30,9 @@ class TestInfoBase:
 
     @pytest.fixture(
         scope="function",
-        params=gen_segment_sizes()
+        params=gen_segment_row_counts()
     )
-    def get_segment_size(self, request):
+    def get_segment_row_count(self, request):
         yield request.param
 
     """
@@ -53,7 +53,7 @@ class TestInfoBase:
         collection_name = gen_unique_str(collection_id)
         fields = {
                 "fields": [filter_field, vector_field],
-                "segment_size": segment_size
+                "segment_row_count": segment_row_count
         }
         connect.create_collection(collection_name, fields)
         res = connect.get_collection_info(collection_name)
@@ -64,15 +64,15 @@ class TestInfoBase:
         # assert dimension
 
     # TODO
-    def test_create_collection_segment_size(self, connect, get_segment_size):
+    def test_create_collection_segment_row_count(self, connect, get_segment_row_count):
         '''
         target: test create normal collection with different fields
-        method: create collection with diff segment_size
+        method: create collection with diff segment_row_count
         expected: no exception raised
         '''
         collection_name = gen_unique_str(collection_id)
         fields = copy.deepcopy(default_fields)
-        fields["segment_size"] = get_segment_size
+        fields["segment_row_count"] = get_segment_row_count
         connect.create_collection(collection_name, fields)
         # assert segment size
 
@@ -141,7 +141,7 @@ class TestInfoBase:
         collection_name = gen_unique_str(collection_id)
         fields = {
                 "fields": [filter_field, vector_field],
-                "segment_size": segment_size
+                "segment_row_count": segment_row_count
         }
         connect.create_collection(collection_name, fields)
         # insert
@@ -153,15 +153,15 @@ class TestInfoBase:
         # assert dimension
 
     # TODO
-    def test_create_collection_segment_size_after_insert(self, connect, get_segment_size):
+    def test_create_collection_segment_row_count_after_insert(self, connect, get_segment_row_count):
         '''
         target: test create normal collection with different fields
-        method: create collection with diff segment_size
+        method: create collection with diff segment_row_count
         expected: no exception raised
         '''
         collection_name = gen_unique_str(collection_id)
         fields = copy.deepcopy(default_fields)
-        fields["segment_size"] = get_segment_size
+        fields["segment_row_count"] = get_segment_row_count
         connect.create_collection(collection_name, fields)
         # insert
         # assert segment size

tests/milvus_python_test/entity/test_delete.py

@@ -10,7 +10,7 @@ from utils import *
 
 
 dim = 128
-segment_size = 10
+segment_row_count = 5000
 collection_id = "test_delete"
 DELETE_TIMEOUT = 60
 tag = "1970-01-01"

tests/milvus_python_test/entity/test_get_entity_by_id.py

@@ -11,7 +11,7 @@ from utils import *
 
 
 dim = 128
-segment_size = 10
+segment_row_count = 5000
 collection_id = "test_get"
 DELETE_TIMEOUT = 60
 tag = "1970-01-01"

tests/milvus_python_test/entity/test_insert.py

@@ -9,7 +9,7 @@ from milvus import DataType
 from utils import *
 
 dim = 128
-segment_size = 10
+segment_row_count = 5000
 collection_id = "test_insert"
 ADD_TIMEOUT = 60
 tag = "1970-01-01"
@@ -209,7 +209,7 @@ class TestInsertBase:
         collection_name = gen_unique_str("test_collection")
         fields = {
             "fields": [filter_field, vector_field],
-            "segment_size": segment_size
+            "segment_row_count": segment_row_count
         }
         connect.create_collection(collection_name, fields)
         ids = [i for i in range(nb)]
@@ -283,7 +283,7 @@ class TestInsertBase:
         collection_name = gen_unique_str("test_collection")
         fields = {
             "fields": [filter_field, vector_field],
-            "segment_size": segment_size
+            "segment_row_count": segment_row_count
         }
         connect.create_collection(collection_name, fields)
         entities = gen_entities_by_fields(fields["fields"], nb, dim)

tests/milvus_python_test/entity/test_list_id_in_segment.py

@@ -8,7 +8,7 @@ import pytest
 from utils import *
 
 dim = 128
-segment_size = 100
+segment_row_count = 100000
 nb = 6000
 tag = "1970-01-01"
 field_name = "float_vector"

tests/milvus_python_test/entity/test_search.py

@@ -11,7 +11,7 @@ from milvus import DataType
 from utils import *
 
 dim = 128
-segment_size = 10
+segment_row_count = 5000
 top_k_limit = 2048
 collection_id = "search"
 tag = "1970-01-01"

tests/milvus_python_test/test_compact.py

@@ -13,7 +13,7 @@ nprobe = 1
 top_k = 1
 tag = "1970-01-01"
 nb = 6000
-segment_size = 10
+segment_row_count = 5000
 entity = gen_entities(1)
 entities = gen_entities(nb)
 raw_vector, binary_entity = gen_binary_entities(1)

tests/milvus_python_test/test_flush.py

@@ -7,7 +7,7 @@ import pytest
 from utils import *
 
 dim = 128
-segment_size = 10
+segment_row_count = 5000
 index_file_size = 10
 collection_id = "test_flush"
 DELETE_TIMEOUT = 60
@@ -155,7 +155,7 @@ class TestFlushBase:
         collection_new = gen_unique_str("test_flush")
         fields = {
             "fields": [filter_field, vector_field],
-            "segment_size": segment_size
+            "segment_row_count": segment_row_count
         }
         connect.create_collection(collection_new, fields)
         connect.create_partition(collection, tag)

tests/milvus_python_test/test_partition.py

@@ -9,7 +9,7 @@ from utils import *
 
 
 dim = 128
-segment_size = 10
+segment_row_count = 5000
 collection_id = "partition"
 nprobe = 1
 tag = "1970-01-01"

tests/milvus_python_test/test_wal.py

@@ -8,7 +8,7 @@ from utils import *
 
 dim = 128
 collection_id = "test_wal"
-segment_size = 10
+segment_row_count = 5000
 WAL_TIMEOUT = 60
 tag = "1970-01-01"
 insert_interval_time = 1.5

tests/milvus_python_test/utils.py

@@ -16,7 +16,7 @@ epsilon = 0.000001
 default_flush_interval = 1
 big_flush_interval = 1000
 dimension = 128
-segment_size = 10
+segment_row_count = 5000
 
 # TODO:
 all_index_types = [
@@ -199,14 +199,9 @@ def gen_single_filter_fields():
 
 def gen_single_vector_fields():
     fields = []
-    for metric_type in ['HAMMING', 'IP', 'JACCARD', 'L2', 'SUBSTRUCTURE', 'SUPERSTRUCTURE', 'TANIMOTO']:
-        for data_type in [DataType.FLOAT_VECTOR, DataType.BINARY_VECTOR]:
-            if metric_type in ["L2", "IP"] and data_type == DataType.BINARY_VECTOR:
-                continue
-            if metric_type not in ["L2", "IP"] and data_type == DataType.FLOAT_VECTOR:
-                continue
-            field = {"field": data_type.name, "type": data_type, "params": {"metric_type": metric_type, "dim": dimension}}
-            fields.append(field)
+    for data_type in [DataType.FLOAT_VECTOR, DataType.BINARY_VECTOR]:
+        field = {"field": data_type.name, "type": data_type, "params": {"dim": dimension}}
+        fields.append(field)
     return fields
 
 
@@ -215,9 +210,9 @@ def gen_default_fields():
         "fields": [
             {"field": "int64", "type": DataType.INT64},
             {"field": "float", "type": DataType.FLOAT},
-            {"field": "vector", "type": DataType.FLOAT_VECTOR, "params": {"metric_type": "L2", "dim": dimension}}
+            {"field": "vector", "type": DataType.FLOAT_VECTOR, "params": {"dim": dimension}}
         ],
-        "segment_size": segment_size
+        "segment_row_count": segment_row_count
     }
     return default_fields
 
@@ -311,14 +306,14 @@ def add_vector_field(entities, is_normal=False):
     return entities
 
 
-def update_fields_metric_type(fields, metric_type):
-    tmp_fields = copy.deepcopy(fields)
-    if metric_type in ["L2", "IP"]:
-        tmp_fields["fields"][-1]["type"] = DataType.FLOAT_VECTOR
-    else:
-        tmp_fields["fields"][-1]["type"] = DataType.BINARY_VECTOR
-    tmp_fields["fields"][-1]["params"]["metric_type"] = metric_type
-    return tmp_fields
+# def update_fields_metric_type(fields, metric_type):
+#     tmp_fields = copy.deepcopy(fields)
+#     if metric_type in ["L2", "IP"]:
+#         tmp_fields["fields"][-1]["type"] = DataType.FLOAT_VECTOR
+#     else:
+#         tmp_fields["fields"][-1]["type"] = DataType.BINARY_VECTOR
+#     tmp_fields["fields"][-1]["params"]["metric_type"] = metric_type
+#     return tmp_fields
 
 
 def remove_field(entities):
@@ -363,7 +358,7 @@ def add_vector_field(nb, dimension=dimension):
     return field_name
         
 
-def gen_segment_sizes():
+def gen_segment_row_counts():
     sizes = [
             1,
             2,