Add pre-checks in create index task

Signed-off-by: sunby <bingyi.sun@zilliz.com>

internal/core/src/indexbuilder/IndexWrapper.cpp

@@ -34,9 +34,6 @@ IndexWrapper::IndexWrapper(const char* serialized_type_params, const char* seria
     auto mode = get_config_by_name<std::string>("index_mode");
     auto index_mode = mode.has_value() ? mode_map[mode.value()] : knowhere::IndexMode::MODE_CPU;
 
-    auto index_type = get_index_type();
-    auto metric_type = get_metric_type();
-    AssertInfo(!is_unsupported(index_type, metric_type), index_type + " doesn't support metric: " + metric_type);
     index_ = knowhere::VecIndexFactory::GetInstance().CreateVecIndex(get_index_type(), index_mode);
     Assert(index_ != nullptr);
 }
@@ -266,21 +263,6 @@ IndexWrapper::get_index_type() {
     return type.has_value() ? type.value() : knowhere::IndexEnum::INDEX_FAISS_IVFPQ;
 }
 
-std::string
-IndexWrapper::get_metric_type() {
-    auto type = get_config_by_name<std::string>(knowhere::Metric::TYPE);
-    if (type.has_value()) {
-        return type.value();
-    } else {
-        auto index_type = get_index_type();
-        if (is_in_bin_list(index_type)) {
-            return knowhere::Metric::JACCARD;
-        } else {
-            return knowhere::Metric::L2;
-        }
-    }
-}
-
 std::unique_ptr<IndexWrapper::QueryResult>
 IndexWrapper::Query(const knowhere::DatasetPtr& dataset) {
     return std::move(QueryImpl(dataset, config_));

internal/core/src/indexbuilder/IndexWrapper.h

@@ -59,9 +59,6 @@ class IndexWrapper {
     std::string
     get_index_type();
 
-    std::string
-    get_metric_type();
-
     template <typename T>
     std::optional<T>
     get_config_by_name(std::string name);

internal/core/src/indexbuilder/utils.h

@@ -14,7 +14,6 @@
 #include <vector>
 #include <string>
 #include <algorithm>
-#include <tuple>
 
 #include "index/knowhere/knowhere/index/IndexType.h"
 
@@ -58,14 +57,6 @@ Need_BuildAll_list() {
     return ret;
 }
 
-std::vector<std::tuple<std::string, std::string>>
-unsupported_index_combinations() {
-    static std::vector<std::tuple<std::string, std::string>> ret{
-        std::make_tuple(std::string(knowhere::IndexEnum::INDEX_FAISS_BIN_IVFFLAT), std::string(knowhere::Metric::L2)),
-    };
-    return ret;
-}
-
 template <typename T>
 bool
 is_in_list(const T& t, std::function<std::vector<T>()> list_func) {
@@ -93,11 +84,5 @@ is_in_need_id_list(const milvus::knowhere::IndexType& index_type) {
     return is_in_list<std::string>(index_type, Need_ID_List);
 }
 
-bool
-is_unsupported(const milvus::knowhere::IndexType& index_type, const milvus::knowhere::MetricType& metric_type) {
-    return is_in_list<std::tuple<std::string, std::string>>(std::make_tuple(index_type, metric_type),
-                                                            unsupported_index_combinations);
-}
-
 }  // namespace indexbuilder
 }  // namespace milvus

internal/indexbuilder/index.go

@@ -14,7 +14,6 @@ package indexbuilder
 import "C"
 import (
 	"errors"
-	"fmt"
 	"strconv"
 	"unsafe"
 
@@ -143,8 +142,6 @@ func (index *CIndex) Delete() error {
 }
 
 func NewCIndex(typeParams, indexParams map[string]string) (Index, error) {
-	fmt.Println("NNNNNNNNNNNNNNNNNNNNNNNNNNN typeParams: ", typeParams)
-	fmt.Println("NNNNNNNNNNNNNNNNNNNNNNNNNNN indexParams: ", indexParams)
 	protoTypeParams := &indexcgopb.TypeParams{
 		Params: make([]*commonpb.KeyValuePair, 0),
 	}
@@ -171,14 +168,10 @@ func NewCIndex(typeParams, indexParams map[string]string) (Index, error) {
 					CIndex* res_index);
 	*/
 	var indexPtr C.CIndex
-	fmt.Println("before create index ........................................")
 	status := C.CreateIndex(typeParamsPointer, indexParamsPointer, &indexPtr)
-	fmt.Println("after create index ........................................")
 	errorCode := status.error_code
-	fmt.Println("EEEEEEEEEEEEEEEEEEEEEEEEEE error code: ", errorCode)
 	if errorCode != 0 {
 		errorMsg := C.GoString(status.error_msg)
-		fmt.Println("EEEEEEEEEEEEEEEEEEEEEEEEEE error msg: ", errorMsg)
 		defer C.free(unsafe.Pointer(status.error_msg))
 		return nil, errors.New(" failed, C runtime error detected, error code = " + strconv.Itoa(int(errorCode)) + ", error msg = " + errorMsg)
 	}

internal/indexbuilder/task.go

@@ -2,7 +2,6 @@ package indexbuilder
 
 import (
 	"context"
-	"fmt"
 	"log"
 	"strconv"
 	"time"
@@ -172,12 +171,10 @@ func (it *IndexBuildTask) Execute() error {
 		indexParams[key] = value
 	}
 
-	fmt.Println("before NewCIndex ..........................")
 	it.index, err = NewCIndex(typeParams, indexParams)
 	if err != nil {
 		return err
 	}
-	fmt.Println("after NewCIndex ..........................")
 
 	getKeyByPathNaive := func(path string) string {
 		// splitElements := strings.Split(path, "/")
@@ -226,7 +223,6 @@ func (it *IndexBuildTask) Execute() error {
 
 	for _, value := range insertData.Data {
 		// TODO: BinaryVectorFieldData
-		fmt.Println("before build index ..................................")
 		floatVectorFieldData, fOk := value.(*storage.FloatVectorFieldData)
 		if fOk {
 			err = it.index.BuildFloatVecIndexWithoutIds(floatVectorFieldData.Data)
@@ -242,15 +238,12 @@ func (it *IndexBuildTask) Execute() error {
 				return err
 			}
 		}
-		fmt.Println("after build index ..................................")
 
 		if !fOk && !bOk {
 			return errors.New("we expect FloatVectorFieldData or BinaryVectorFieldData")
 		}
 
-		fmt.Println("before serialize .............................................")
 		indexBlobs, err := it.index.Serialize()
-		fmt.Println("after serialize .............................................")
 		if err != nil {
 			return err
 		}

internal/master/index_task.go

@@ -24,11 +24,6 @@ func (task *createIndexTask) Ts() (Timestamp, error) {
 }
 
 func (task *createIndexTask) Execute() error {
-	// modify schema
-	if err := task.mt.UpdateFieldIndexParams(task.req.CollectionName, task.req.FieldName, task.req.ExtraParams); err != nil {
-		return err
-	}
-	// check if closed segment has the same index build history
 	collMeta, err := task.mt.GetCollectionByName(task.req.CollectionName)
 	if err != nil {
 		return err
@@ -44,6 +39,20 @@ func (task *createIndexTask) Execute() error {
 		return fmt.Errorf("can not find field name %s", task.req.FieldName)
 	}
 
+	// pre checks
+	isIndexable, err := task.mt.IsIndexable(collMeta.ID, fieldID)
+	if err != nil {
+		return err
+	}
+	if !isIndexable {
+		return fmt.Errorf("field %s is not vector", task.req.FieldName)
+	}
+
+	// modify schema
+	if err := task.mt.UpdateFieldIndexParams(task.req.CollectionName, task.req.FieldName, task.req.ExtraParams); err != nil {
+		return err
+	}
+	// check if closed segment has the same index build history
 	for _, segID := range collMeta.SegmentIDs {
 		segMeta, err := task.mt.GetSegmentByID(segID)
 		if err != nil {

internal/proxy/proxy_test.go

@@ -488,7 +488,9 @@ func TestProxy_CreateIndex(t *testing.T) {
 		go func(group *sync.WaitGroup) {
 			defer group.Done()
 			createCollection(t, collName)
-			createIndex(t, collName, fieldName)
+			if i%2 == 0 {
+				createIndex(t, collName, fieldName)
+			}
 			dropCollection(t, collName)
 			// dropIndex(t, collectionName, fieldName, indexName)
 		}(&wg)
@@ -510,7 +512,9 @@ func TestProxy_DescribeIndex(t *testing.T) {
 		go func(group *sync.WaitGroup) {
 			defer group.Done()
 			createCollection(t, collName)
-			createIndex(t, collName, fieldName)
+			if i%2 == 0 {
+				createIndex(t, collName, fieldName)
+			}
 			req := &servicepb.DescribeIndexRequest{
 				CollectionName: collName,
 				FieldName:      fieldName,
@@ -539,7 +543,9 @@ func TestProxy_DescribeIndexProgress(t *testing.T) {
 		go func(group *sync.WaitGroup) {
 			defer group.Done()
 			createCollection(t, collName)
-			createIndex(t, collName, fieldName)
+			if i%2 == 0 {
+				createIndex(t, collName, fieldName)
+			}
 			req := &servicepb.DescribeIndexProgressRequest{
 				CollectionName: collName,
 				FieldName:      fieldName,