Create index failed caused by server crashed (#2736)

Signed-off-by: Nshengjun.li <shengjun.li@zilliz.com>
Co-authored-by: Ngroot <yhmo@zeronedata.com>

CHANGELOG.md

@@ -14,6 +14,7 @@ Please mark all changes in change log and use the issue from GitHub
 -   \#2598 Fix Milvus docker image report illegal instruction
 -   \#2617 Fix HNSW and RNSG index files size
 -   \#2637 Suit the range of HNSW parameters
+-   \#2642 Create index failed and server crashed
 -   \#2649 Search parameter of annoy has conflict with document
 -   \#2690 Remove body parser in show-partitions endpoints
 

core/src/index/knowhere/knowhere/index/vector_index/IndexBinaryIDMAP.cpp

@@ -112,6 +112,22 @@ BinaryIDMAP::QueryById(const DatasetPtr& dataset_ptr, const Config& config) {
 }
 #endif
 
+int64_t
+BinaryIDMAP::Count() {
+    if (!index_) {
+        KNOWHERE_THROW_MSG("index not initialize");
+    }
+    return index_->ntotal;
+}
+
+int64_t
+BinaryIDMAP::Dim() {
+    if (!index_) {
+        KNOWHERE_THROW_MSG("index not initialize");
+    }
+    return index_->d;
+}
+
 void
 BinaryIDMAP::Add(const DatasetPtr& dataset_ptr, const Config& config) {
     if (!index_) {

core/src/index/knowhere/knowhere/index/vector_index/IndexBinaryIDMAP.h

@@ -56,14 +56,10 @@ class BinaryIDMAP : public VecIndex, public FaissBaseBinaryIndex {
 #endif
 
     int64_t
-    Count() override {
-        return index_->ntotal;
-    }
+    Count() override;
 
     int64_t
-    Dim() override {
-        return index_->d;
-    }
+    Dim() override;
 
     int64_t
     IndexSize() override {

core/src/index/knowhere/knowhere/index/vector_index/IndexBinaryIVF.cpp

@@ -129,6 +129,22 @@ BinaryIVF::QueryById(const DatasetPtr& dataset_ptr, const Config& config) {
 }
 #endif
 
+int64_t
+BinaryIVF::Count() {
+    if (!index_) {
+        KNOWHERE_THROW_MSG("index not initialize");
+    }
+    return index_->ntotal;
+}
+
+int64_t
+BinaryIVF::Dim() {
+    if (!index_) {
+        KNOWHERE_THROW_MSG("index not initialize");
+    }
+    return index_->d;
+}
+
 void
 BinaryIVF::Train(const DatasetPtr& dataset_ptr, const Config& config) {
     GETTENSORWITHIDS(dataset_ptr)

core/src/index/knowhere/knowhere/index/vector_index/IndexBinaryIVF.h

@@ -68,14 +68,10 @@ class BinaryIVF : public VecIndex, public FaissBaseBinaryIndex {
 #endif
 
     int64_t
-    Count() override {
-        return index_->ntotal;
-    }
+    Count() override;
 
     int64_t
-    Dim() override {
-        return index_->d;
-    }
+    Dim() override;
 
 #if 0
     DatasetPtr

core/src/index/knowhere/knowhere/index/vector_index/IndexIDMAP.cpp

@@ -142,6 +142,22 @@ IDMAP::QueryById(const DatasetPtr& dataset_ptr, const Config& config) {
 }
 #endif
 
+int64_t
+IDMAP::Count() {
+    if (!index_) {
+        KNOWHERE_THROW_MSG("index not initialize");
+    }
+    return index_->ntotal;
+}
+
+int64_t
+IDMAP::Dim() {
+    if (!index_) {
+        KNOWHERE_THROW_MSG("index not initialize");
+    }
+    return index_->d;
+}
+
 VecIndexPtr
 IDMAP::CopyCpuToGpu(const int64_t device_id, const Config& config) {
 #ifdef MILVUS_GPU_VERSION

core/src/index/knowhere/knowhere/index/vector_index/IndexIDMAP.h

@@ -54,14 +54,10 @@ class IDMAP : public VecIndex, public FaissBaseIndex {
 #endif
 
     int64_t
-    Count() override {
-        return index_->ntotal;
-    }
+    Count() override;
 
     int64_t
-    Dim() override {
-        return index_->d;
-    }
+    Dim() override;
 
     int64_t
     IndexSize() override {

core/src/index/knowhere/knowhere/index/vector_index/IndexIVF.cpp

@@ -217,6 +217,22 @@ IVF::GetVectorById(const DatasetPtr& dataset_ptr, const Config& config) {
 }
 #endif
 
+int64_t
+IVF::Count() {
+    if (!index_) {
+        KNOWHERE_THROW_MSG("index not initialize");
+    }
+    return index_->ntotal;
+}
+
+int64_t
+IVF::Dim() {
+    if (!index_) {
+        KNOWHERE_THROW_MSG("index not initialize");
+    }
+    return index_->d;
+}
+
 void
 IVF::Seal() {
     if (!index_ || !index_->is_trained) {

core/src/index/knowhere/knowhere/index/vector_index/IndexIVF.h

@@ -59,14 +59,10 @@ class IVF : public VecIndex, public FaissBaseIndex {
 #endif
 
     int64_t
-    Count() override {
-        return index_->ntotal;
-    }
+    Count() override;
 
     int64_t
-    Dim() override {
-        return index_->d;
-    }
+    Dim() override;
 
 #if 0
     DatasetPtr

core/src/index/knowhere/knowhere/index/vector_index/IndexNSG.cpp

@@ -149,11 +149,17 @@ NSG::Train(const DatasetPtr& dataset_ptr, const Config& config) {
 
 int64_t
 NSG::Count() {
+    if (!index_) {
+        KNOWHERE_THROW_MSG("index not initialize");
+    }
     return index_->ntotal;
 }
 
 int64_t
 NSG::Dim() {
+    if (!index_) {
+        KNOWHERE_THROW_MSG("index not initialize");
+    }
     return index_->dimension;
 }
 

core/src/index/knowhere/knowhere/index/vector_index/IndexSPTAG.cpp

@@ -200,11 +200,17 @@ CPUSPTAGRNG::Query(const DatasetPtr& dataset_ptr, const Config& config) {
 
 int64_t
 CPUSPTAGRNG::Count() {
+    if (!index_ptr_) {
+        KNOWHERE_THROW_MSG("index not initialize");
+    }
     return index_ptr_->GetNumSamples();
 }
 
 int64_t
 CPUSPTAGRNG::Dim() {
+    if (!index_ptr_) {
+        KNOWHERE_THROW_MSG("index not initialize");
+    }
     return index_ptr_->GetFeatureDim();
 }
 

core/src/index/knowhere/knowhere/index/vector_offset_index/IndexIVF_NM.cpp

@@ -335,5 +335,21 @@ IVF_NM::SealImpl() {
 #endif
 }
 
+int64_t
+IVF_NM::Count() {
+    if (!index_) {
+        KNOWHERE_THROW_MSG("index not initialize");
+    }
+    return index_->ntotal;
+}
+
+int64_t
+IVF_NM::Dim() {
+    if (!index_) {
+        KNOWHERE_THROW_MSG("index not initialize");
+    }
+    return index_->d;
+}
+
 }  // namespace knowhere
 }  // namespace milvus

core/src/index/knowhere/knowhere/index/vector_offset_index/IndexIVF_NM.h

@@ -59,14 +59,10 @@ class IVF_NM : public VecIndex, public OffsetBaseIndex {
 #endif
 
     int64_t
-    Count() override {
-        return index_->ntotal;
-    }
+    Count() override;
 
     int64_t
-    Dim() override {
-        return index_->d;
-    }
+    Dim() override;
 
 #if 0
     DatasetPtr

core/src/index/knowhere/knowhere/index/vector_offset_index/IndexNSG_NM.cpp

@@ -151,11 +151,17 @@ NSG_NM::Train(const DatasetPtr& dataset_ptr, const Config& config) {
 
 int64_t
 NSG_NM::Count() {
+    if (!index_) {
+        KNOWHERE_THROW_MSG("index not initialize");
+    }
     return index_->ntotal;
 }
 
 int64_t
 NSG_NM::Dim() {
+    if (!index_) {
+        KNOWHERE_THROW_MSG("index not initialize");
+    }
     return index_->dimension;
 }
 

core/src/index/thirdparty/hnswlib/hnswalg_nm.h

@@ -46,9 +46,6 @@ namespace hnswlib {
             dist_func_param_ = s->get_dist_func_param();
             M_ = M;
             maxM_ = M_;
-            cmli_cnt_ = 0;
-            cmli_cnt2_ = 0;
-            cmli_time_ = 0.0;
             maxM0_ = M_ * 2;
             ef_construction_ = std::max(ef_construction,M_);
             ef_ = 10;
@@ -124,9 +121,6 @@ namespace hnswlib {
         size_t M_;
         size_t maxM_;
         size_t maxM0_;
-        size_t cmli_cnt_;
-        size_t cmli_cnt2_;
-        double cmli_time_;
         size_t ef_construction_;
 
         double mult_, revSize_;
@@ -497,13 +491,8 @@ namespace hnswlib {
                                              dist_func_param_), data[j]);
                     }
 
-                    if (candidates.size() >= Mcurmax) cmli_cnt2_ ++;
                     size_t indx = 0;
-                    auto t0 = std::chrono::high_resolution_clock::now();
                     getNeighborsByHeuristic2(candidates, Mcurmax, data, indx, pdata);
-                    auto t1 = std::chrono::high_resolution_clock::now();
-                    cmli_time_ += (double)std::chrono::duration_cast<std::chrono::milliseconds>( t1 - t0 ).count();
-                    cmli_cnt_ ++;
 
 //                while (candidates.size() > 0) {
 //                    data[indx] = candidates.top().second;