MS-474 - Create index hang if use branch-0.3.1 server config

Former-commit-id: 6b8e462fcfbfc2bbba93c32851058d1d2c8b1cda

cpp/CHANGELOG.md

@@ -93,7 +93,11 @@ Please mark all change in change log and use the ticket from JIRA.
 - MS-487 - Define metric type in CreateTable
 - MS-488 - Improve code format in scheduler
 - MS-495 - cmake: integrated knowhere
+- MS-496 - Change the top_k limitation from 1024 to 2048
+- MS-502 - Update tasktable_test in scheduler
+- MS-504 - Update node_test in scheduler
 - MS-505 - Install core unit test and add to coverage
+- MS-508 - Update normal_test in scheduler
 
 ## New Feature
 - MS-343 - Implement ResourceMgr

cpp/src/scheduler/Scheduler.cpp

@@ -108,6 +108,7 @@ void
 Scheduler::OnFinishTask(const EventPtr &event) {
 }
 
+// TODO: refactor the function
 void
 Scheduler::OnLoadCompleted(const EventPtr &event) {
     auto load_completed_event = std::static_pointer_cast<LoadCompletedEvent>(event);
@@ -120,18 +121,23 @@ Scheduler::OnLoadCompleted(const EventPtr &event) {
                 if (not resource->HasExecutor() && load_completed_event->task_table_item_->Move()) {
                     auto task = load_completed_event->task_table_item_->task;
                     auto search_task = std::static_pointer_cast<XSearchTask>(task);
-                    auto location = search_task->index_engine_->GetLocation();
                     bool moved = false;
 
-                    for (auto i = 0; i < res_mgr_.lock()->GetNumGpuResource(); ++i) {
+                    // to support test task, REFACTOR
-                        auto index = zilliz::milvus::cache::GpuCacheMgr::GetInstance(i)->GetIndex(location);
+                    if (auto index_engine = search_task->index_engine_) {
-                        if (index != nullptr) {
+                        auto location = index_engine->GetLocation();
-                            moved = true;
+
-                            auto dest_resource = res_mgr_.lock()->GetResource(ResourceType::GPU, i);
+                        for (auto i = 0; i < res_mgr_.lock()->GetNumGpuResource(); ++i) {
-                            Action::PushTaskToResource(load_completed_event->task_table_item_->task, dest_resource);
+                            auto index = zilliz::milvus::cache::GpuCacheMgr::GetInstance(i)->GetIndex(location);
-                            break;
+                            if (index != nullptr) {
+                                moved = true;
+                                auto dest_resource = res_mgr_.lock()->GetResource(ResourceType::GPU, i);
+                                Action::PushTaskToResource(load_completed_event->task_table_item_->task, dest_resource);
+                                break;
+                            }
                         }
                     }
+
                     if (not moved) {
                         Action::PushTaskToNeighbourRandomly(task, resource);
                     }
@@ -147,7 +153,7 @@ Scheduler::OnLoadCompleted(const EventPtr &event) {
                     // step 1: calculate shortest path per resource, from disk to compute resource
                     auto compute_resources = res_mgr_.lock()->GetComputeResource();
                     std::vector<std::vector<std::string>> paths;
-                    std::vector<uint64_t > transport_costs;
+                    std::vector<uint64_t> transport_costs;
                     for (auto &res : compute_resources) {
                         std::vector<std::string> path;
                         uint64_t transport_cost = ShortestPath(self, res, res_mgr_.lock(), path);
@@ -176,7 +182,7 @@ Scheduler::OnLoadCompleted(const EventPtr &event) {
                     task->path() = task_path;
                 }
 
-                if(self->name() == task->path().Last()) {
+                if (self->name() == task->path().Last()) {
                     self->WakeupLoader();
                 } else {
                     auto next_res_name = task->path().Next();

cpp/src/scheduler/Scheduler.h

@@ -21,6 +21,7 @@ namespace milvus {
 namespace engine {
 
 
+// TODO: refactor, not friendly to unittest, logical in framework code
 class Scheduler {
 public:
     explicit

cpp/src/scheduler/TaskTable.cpp

@@ -136,7 +136,7 @@ std::vector<uint64_t>
 TaskTable::PickToLoad(uint64_t limit) {
     std::vector<uint64_t> indexes;
     bool cross = false;
-    for (uint64_t i = last_finish_, count = 0; i < table_.size() && count < limit; ++i) {
+    for (uint64_t i = last_finish_ + 1, count = 0; i < table_.size() && count < limit; ++i) {
         if (not cross && table_[i]->IsFinish()) {
             last_finish_ = i;
         } else if (table_[i]->state == TaskTableItemState::START) {
@@ -152,7 +152,7 @@ std::vector<uint64_t>
 TaskTable::PickToExecute(uint64_t limit) {
     std::vector<uint64_t> indexes;
     bool cross = false;
-    for (uint64_t i = last_finish_, count = 0; i < table_.size() && count < limit; ++i) {
+    for (uint64_t i = last_finish_ + 1, count = 0; i < table_.size() && count < limit; ++i) {
         if (not cross && table_[i]->IsFinish()) {
             last_finish_ = i;
         } else if (table_[i]->state == TaskTableItemState::LOADED) {
@@ -200,15 +200,15 @@ TaskTable::Get(uint64_t index) {
     return table_[index];
 }
 
-void
+//void
-TaskTable::Clear() {
+//TaskTable::Clear() {
-// find first task is NOT (done or moved), erase from begin to it;
+//// find first task is NOT (done or moved), erase from begin to it;
-//        auto iterator = table_.begin();
+////        auto iterator = table_.begin();
-//        while (iterator->state == TaskTableItemState::EXECUTED or
+////        while (iterator->state == TaskTableItemState::EXECUTED or
-//            iterator->state == TaskTableItemState::MOVED)
+////            iterator->state == TaskTableItemState::MOVED)
-//            iterator++;
+////            iterator++;
-//        table_.erase(table_.begin(), iterator);
+////        table_.erase(table_.begin(), iterator);
-}
+//}
 
 
 std::string

cpp/src/scheduler/TaskTable.h

@@ -40,10 +40,10 @@ struct TaskTimestamp {
 };
 
 struct TaskTableItem {
-    TaskTableItem() : id(0), state(TaskTableItemState::INVALID), mutex() {}
+    TaskTableItem() : id(0), task(nullptr), state(TaskTableItemState::INVALID), mutex() {}
 
-    TaskTableItem(const TaskTableItem &src)
+    TaskTableItem(const TaskTableItem &src) = delete;
-        : id(src.id), state(src.state), mutex() {}
+    TaskTableItem(TaskTableItem &&) = delete;
 
     uint64_t id; // auto increment from 0;
     TaskPtr task; // the task;
@@ -114,8 +114,8 @@ public:
      * Remove sequence task which is DONE or MOVED from front;
      * Called by ?
      */
-    void
+//    void
-    Clear();
+//    Clear();
 
     /*
      * Return true if task table empty, otherwise false;
@@ -229,7 +229,9 @@ private:
     std::function<void(void)> subscriber_ = nullptr;
 
     // cache last finish avoid Pick task from begin always
-    uint64_t last_finish_ = 0;
+    // pick from (last_finish_ + 1)
+    // init with -1, pick from (last_finish_ + 1) = 0
+    uint64_t last_finish_ = -1;
 };
 
 

cpp/src/scheduler/resource/Node.cpp

@@ -17,27 +17,6 @@ Node::Node() {
     id_ = counter++;
 }
 
-void Node::DelNeighbour(const NeighbourNodePtr &neighbour_ptr) {
-    std::lock_guard<std::mutex> lk(mutex_);
-    if (auto s = neighbour_ptr.lock()) {
-        auto search = neighbours_.find(s->id_);
-        if (search != neighbours_.end()) {
-            neighbours_.erase(search);
-        }
-    }
-}
-
-bool Node::IsNeighbour(const NeighbourNodePtr &neighbour_ptr) {
-    std::lock_guard<std::mutex> lk(mutex_);
-    if (auto s = neighbour_ptr.lock()) {
-        auto search = neighbours_.find(s->id_);
-        if (search != neighbours_.end()) {
-            return true;
-        }
-    }
-    return false;
-}
-
 std::vector<Neighbour> Node::GetNeighbours() {
     std::lock_guard<std::mutex> lk(mutex_);
     std::vector<Neighbour> ret;
@@ -48,8 +27,13 @@ std::vector<Neighbour> Node::GetNeighbours() {
 }
 
 std::string Node::Dump() {
-    // TODO(linxj): what's that?
+    std::stringstream ss;
-    return std::__cxx11::string();
+    ss << "<Node, id=" << std::to_string(id_) << ">::neighbours:" << std::endl;
+    for (auto &neighbour : neighbours_) {
+        ss << "\t<Neighbour, id=" << std::to_string(neighbour.first);
+        ss << ", connection: " << neighbour.second.connection.Dump() << ">" << std::endl;
+    }
+    return ss.str();
 }
 
 void Node::AddNeighbour(const NeighbourNodePtr &neighbour_node, Connection &connection) {

cpp/src/scheduler/resource/Node.h

@@ -37,12 +37,6 @@ public:
     void
     AddNeighbour(const NeighbourNodePtr &neighbour_node, Connection &connection);
 
-    void
-    DelNeighbour(const NeighbourNodePtr &neighbour_ptr);
-
-    bool
-    IsNeighbour(const NeighbourNodePtr& neighbour_ptr);
-
     std::vector<Neighbour>
     GetNeighbours();
 

cpp/src/scheduler/task/SearchTask.cpp

@@ -83,11 +83,14 @@ CollectFileMetrics(int file_type, size_t file_size) {
 
 XSearchTask::XSearchTask(TableFileSchemaPtr file)
     : Task(TaskType::SearchTask), file_(file) {
-    index_engine_ = EngineFactory::Build(file_->dimension_,
+    if (file_) {
-                                         file_->location_,
+        index_engine_ = EngineFactory::Build(file_->dimension_,
-                                         (EngineType) file_->engine_type_,
+                                             file_->location_,
-                                         (MetricType) file_->metric_type_,
+                                             (EngineType) file_->engine_type_,
-                                         file_->nlist_);
+                                             (MetricType) file_->metric_type_,
+                                             file_->nlist_);
+    }
+
 }
 
 void
@@ -103,6 +106,8 @@ XSearchTask::Load(LoadType type, uint8_t device_id) {
             index_engine_->CopyToCpu();
         } else {
             // TODO: exception
+            std::string msg = "Wrong load type";
+            ENGINE_LOG_ERROR << msg;
         }
     } catch (std::exception &ex) {
         //typical error: out of disk space or permition denied
@@ -147,17 +152,17 @@ XSearchTask::Execute() {
     server::CollectDurationMetrics metrics(index_type_);
 
     std::vector<long> output_ids;
-    std::vector<float> output_distence;
+    std::vector<float> output_distance;
     for (auto &context : search_contexts_) {
         //step 1: allocate memory
         auto inner_k = context->topk();
         auto nprobe = context->nprobe();
         output_ids.resize(inner_k * context->nq());
-        output_distence.resize(inner_k * context->nq());
+        output_distance.resize(inner_k * context->nq());
 
         try {
             //step 2: search
-            index_engine_->Search(context->nq(), context->vectors(), inner_k, nprobe, output_distence.data(),
+            index_engine_->Search(context->nq(), context->vectors(), inner_k, nprobe, output_distance.data(),
                                   output_ids.data());
 
             double span = rc.RecordSection("do search for context:" + context->Identity());
@@ -167,12 +172,12 @@ XSearchTask::Execute() {
             //step 3: cluster result
             SearchContext::ResultSet result_set;
             auto spec_k = index_engine_->Count() < context->topk() ? index_engine_->Count() : context->topk();
-            XSearchTask::ClusterResult(output_ids, output_distence, context->nq(), spec_k, result_set);
+            XSearchTask::ClusterResult(output_ids, output_distance, context->nq(), spec_k, result_set);
 
             span = rc.RecordSection("cluster result for context:" + context->Identity());
             context->AccumReduceCost(span);
 
-            //step 4: pick up topk result
+            // step 4: pick up topk result
             XSearchTask::TopkResult(result_set, inner_k, metric_l2, context->GetResult());
 
             span = rc.RecordSection("reduce topk for context:" + context->Identity());
@@ -194,13 +199,13 @@ XSearchTask::Execute() {
 }
 
 Status XSearchTask::ClusterResult(const std::vector<long> &output_ids,
-                                  const std::vector<float> &output_distence,
+                                  const std::vector<float> &output_distance,
                                   uint64_t nq,
                                   uint64_t topk,
                                   SearchContext::ResultSet &result_set) {
-    if (output_ids.size() < nq * topk || output_distence.size() < nq * topk) {
+    if (output_ids.size() < nq * topk || output_distance.size() < nq * topk) {
         std::string msg = "Invalid id array size: " + std::to_string(output_ids.size()) +
-            " distance array size: " + std::to_string(output_distence.size());
+            " distance array size: " + std::to_string(output_distance.size());
         ENGINE_LOG_ERROR << msg;
         return Status(DB_ERROR, msg);
     }
@@ -217,7 +222,7 @@ Status XSearchTask::ClusterResult(const std::vector<long> &output_ids,
                 if (output_ids[index] < 0) {
                     continue;
                 }
-                id_distance.push_back(std::make_pair(output_ids[index], output_distence[index]));
+                id_distance.push_back(std::make_pair(output_ids[index], output_distance[index]));
             }
             result_set[i] = id_distance;
         }

cpp/src/scheduler/task/SearchTask.h

@@ -12,6 +12,7 @@ namespace zilliz {
 namespace milvus {
 namespace engine {
 
+// TODO: rewrite
 class XSearchTask : public Task {
 public:
     explicit

cpp/src/scheduler/task/Task.h

@@ -34,6 +34,7 @@ class Task;
 
 using TaskPtr = std::shared_ptr<Task>;
 
+// TODO: re-design
 class Task {
 public:
     explicit

cpp/src/scheduler/task/TestTask.cpp

@@ -13,7 +13,7 @@ namespace milvus {
 namespace engine {
 
 
-TestTask::TestTask(TableFileSchemaPtr& file) : XSearchTask(file) {}
+TestTask::TestTask(TableFileSchemaPtr &file) : XSearchTask(file) {}
 
 void
 TestTask::Load(LoadType type, uint8_t device_id) {
@@ -22,9 +22,12 @@ TestTask::Load(LoadType type, uint8_t device_id) {
 
 void
 TestTask::Execute() {
-    std::lock_guard<std::mutex> lock(mutex_);
+    {
-    exec_count_++;
+        std::lock_guard<std::mutex> lock(mutex_);
-    done_ = true;
+        exec_count_++;
+        done_ = true;
+    }
+    cv_.notify_one();
 }
 
 void

cpp/src/server/grpc_impl/GrpcRequestTask.cpp

@@ -646,7 +646,7 @@ SearchTask::OnExecute() {
                    search_param_->query_record_array(i).vector_data().data(),
                    table_info.dimension_ * sizeof(float));
         }
-        rc.ElapseFromBegin("prepare vector data");
+        rc.RecordSection("prepare vector data");
 
         //step 6: search vectors
         engine::QueryResults results;
@@ -669,7 +669,7 @@ SearchTask::OnExecute() {
         ProfilerStop();
 #endif
 
-        rc.ElapseFromBegin("search vectors from engine");
+        rc.RecordSection("search vectors from engine");
         if (!stat.ok()) {
             return SetError(DB_META_TRANSACTION_FAILED, stat.ToString());
         }
@@ -684,8 +684,6 @@ SearchTask::OnExecute() {
             return SetError(SERVER_ILLEGAL_SEARCH_RESULT, msg);
         }
 
-        rc.ElapseFromBegin("do search");
-
         //step 7: construct result array
         for (auto &result : results) {
             ::milvus::grpc::TopKQueryResult *topk_query_result = topk_result_list->add_topk_query_result();
@@ -697,7 +695,7 @@ SearchTask::OnExecute() {
         }
 
         //step 8: print time cost percent
-        double span_result = rc.RecordSection("construct result");
+        rc.RecordSection("construct result and send");
         rc.ElapseFromBegin("totally cost");
 
 
@@ -969,4 +967,4 @@ DropIndexTask::OnExecute() {
 }
 }
 }
-}
+}

cpp/src/utils/ValidationUtil.cpp

@@ -94,7 +94,7 @@ ValidationUtil::ValidateTableIndexMetricType(int32_t metric_type) {
 
 ErrorCode
 ValidationUtil::ValidateSearchTopk(int64_t top_k, const engine::meta::TableSchema& table_schema) {
-    if (top_k <= 0 || top_k > 1024) {
+    if (top_k <= 0 || top_k > 2048) {
         return SERVER_INVALID_TOPK;
     }
 

cpp/src/wrapper/knowhere/vec_impl.cpp

@@ -103,6 +103,7 @@ ErrorCode VecIndexImpl::Search(const long &nq, const float *xq, float *dist, lon
         // TODO(linxj): avoid copy here.
         memcpy(ids, p_ids, sizeof(int64_t) * nq * k);
         memcpy(dist, p_dist, sizeof(float) * nq * k);
+
     } catch (KnowhereException &e) {
         WRAPPER_LOG_ERROR << e.what();
         return KNOWHERE_UNEXPECTED_ERROR;

cpp/src/wrapper/knowhere/vec_index.cpp

@@ -14,6 +14,8 @@
 #include "vec_impl.h"
 #include "wrapper_log.h"
 
+#include <cuda.h>
+
 
 namespace zilliz {
 namespace milvus {
@@ -246,11 +248,13 @@ void ParameterValidation(const IndexType &type, Config &cfg) {
         case IndexType::FAISS_IVFSQ8_GPU:
         case IndexType::FAISS_IVFFLAT_GPU:
         case IndexType::FAISS_IVFPQ_GPU: {
+            //search on GPU
             if (cfg.get_with_default("nprobe", 0) != 0) {
                 auto nprobe = cfg["nprobe"].as<int>();
                 if (nprobe > GPU_MAX_NRPOBE) {
-                    WRAPPER_LOG_WARNING << "When search with GPU, nprobe shoud be no more than " << GPU_MAX_NRPOBE << ", but you passed " << nprobe
+                    WRAPPER_LOG_WARNING << "When search with GPU, nprobe shoud be no more than " << GPU_MAX_NRPOBE
-                                      << ". Search with " << GPU_MAX_NRPOBE << " instead";
+                                        << ", but you passed " << nprobe
+                                        << ". Search with " << GPU_MAX_NRPOBE << " instead";
                     cfg.insert_or_assign("nprobe", GPU_MAX_NRPOBE);
                 }
             }

cpp/src/wrapper/knowhere/vec_index.h

@@ -14,8 +14,6 @@
 #include "knowhere/common/config.h"
 #include "knowhere/common/binary_set.h"
 
-#include "cuda.h"
-
 
 namespace zilliz {
 namespace milvus {
@@ -62,7 +60,7 @@ class VecIndex {
                              long *ids,
                              const Config &cfg = Config()) = 0;
 
-    virtual VecIndexPtr CopyToGpu(const int64_t& device_id,
+    virtual VecIndexPtr CopyToGpu(const int64_t &device_id,
                                   const Config &cfg = Config()) = 0;
 
     virtual VecIndexPtr CopyToCpu(const Config &cfg = Config()) = 0;
@@ -86,16 +84,16 @@ extern ErrorCode write_index(VecIndexPtr index, const std::string &location);
 
 extern VecIndexPtr read_index(const std::string &location);
 
-extern VecIndexPtr GetVecIndexFactory(const IndexType &type, const Config& cfg = Config());
+extern VecIndexPtr GetVecIndexFactory(const IndexType &type, const Config &cfg = Config());
 
 extern VecIndexPtr LoadVecIndex(const IndexType &index_type, const zilliz::knowhere::BinarySet &index_binary);
 
-extern void AutoGenParams(const IndexType& type, const long& size, Config& cfg);
+extern void AutoGenParams(const IndexType &type, const long &size, Config &cfg);
 
-extern void ParameterValidation(const IndexType& type, Config& cfg);
+extern void ParameterValidation(const IndexType &type, Config &cfg);
 
-extern IndexType ConvertToCpuIndexType(const IndexType& type);
+extern IndexType ConvertToCpuIndexType(const IndexType &type);
-extern IndexType ConvertToGpuIndexType(const IndexType& type);
+extern IndexType ConvertToGpuIndexType(const IndexType &type);
 
 }
 }

cpp/unittest/scheduler/node_test.cpp

@@ -11,58 +11,71 @@ protected:
         node1_ = std::make_shared<Node>();
         node2_ = std::make_shared<Node>();
         node3_ = std::make_shared<Node>();
-        node4_ = std::make_shared<Node>();
+        isolated_node1_ = std::make_shared<Node>();
+        isolated_node2_ = std::make_shared<Node>();
 
         auto pcie = Connection("PCIe", 11.0);
 
         node1_->AddNeighbour(node2_, pcie);
+        node1_->AddNeighbour(node3_, pcie);
         node2_->AddNeighbour(node1_, pcie);
     }
 
     NodePtr node1_;
     NodePtr node2_;
     NodePtr node3_;
-    NodePtr node4_;
+    NodePtr isolated_node1_;
+    NodePtr isolated_node2_;
 };
 
 TEST_F(NodeTest, add_neighbour) {
-    ASSERT_EQ(node3_->GetNeighbours().size(), 0);
+    ASSERT_EQ(isolated_node1_->GetNeighbours().size(), 0);
-    ASSERT_EQ(node4_->GetNeighbours().size(), 0);
+    ASSERT_EQ(isolated_node2_->GetNeighbours().size(), 0);
     auto pcie = Connection("PCIe", 11.0);
-    node3_->AddNeighbour(node4_, pcie);
+    isolated_node1_->AddNeighbour(isolated_node2_, pcie);
-    node4_->AddNeighbour(node3_, pcie);
+    ASSERT_EQ(isolated_node1_->GetNeighbours().size(), 1);
-    ASSERT_EQ(node3_->GetNeighbours().size(), 1);
+    ASSERT_EQ(isolated_node2_->GetNeighbours().size(), 0);
-    ASSERT_EQ(node4_->GetNeighbours().size(), 1);
 }
 
-TEST_F(NodeTest, del_neighbour) {
+TEST_F(NodeTest, repeat_add_neighbour) {
-    ASSERT_EQ(node1_->GetNeighbours().size(), 1);
+    ASSERT_EQ(isolated_node1_->GetNeighbours().size(), 0);
-    ASSERT_EQ(node2_->GetNeighbours().size(), 1);
+    ASSERT_EQ(isolated_node2_->GetNeighbours().size(), 0);
-    ASSERT_EQ(node3_->GetNeighbours().size(), 0);
+    auto pcie = Connection("PCIe", 11.0);
-    node1_->DelNeighbour(node2_);
+    isolated_node1_->AddNeighbour(isolated_node2_, pcie);
-    node2_->DelNeighbour(node2_);
+    isolated_node1_->AddNeighbour(isolated_node2_, pcie);
-    node3_->DelNeighbour(node2_);
+    ASSERT_EQ(isolated_node1_->GetNeighbours().size(), 1);
-    ASSERT_EQ(node1_->GetNeighbours().size(), 0);
+    ASSERT_EQ(isolated_node2_->GetNeighbours().size(), 0);
-    ASSERT_EQ(node2_->GetNeighbours().size(), 1);
-    ASSERT_EQ(node3_->GetNeighbours().size(), 0);
-}
-
-TEST_F(NodeTest, is_neighbour) {
-    ASSERT_TRUE(node1_->IsNeighbour(node2_));
-    ASSERT_TRUE(node2_->IsNeighbour(node1_));
-
-    ASSERT_FALSE(node1_->IsNeighbour(node3_));
-    ASSERT_FALSE(node2_->IsNeighbour(node3_));
-    ASSERT_FALSE(node3_->IsNeighbour(node1_));
-    ASSERT_FALSE(node3_->IsNeighbour(node2_));
 }
 
 TEST_F(NodeTest, get_neighbours) {
-    auto node1_neighbours = node1_->GetNeighbours();
+    {
-    ASSERT_EQ(node1_neighbours.size(), 1);
+        bool n2 = false, n3 = false;
-    ASSERT_EQ(node1_neighbours[0].neighbour_node.lock(), node2_);
+        auto node1_neighbours = node1_->GetNeighbours();
+        ASSERT_EQ(node1_neighbours.size(), 2);
+        for (auto &n : node1_neighbours) {
+            if (n.neighbour_node.lock() == node2_) n2 = true;
+            if (n.neighbour_node.lock() == node3_) n3 = true;
+        }
+        ASSERT_TRUE(n2);
+        ASSERT_TRUE(n3);
+    }
 
-    auto node2_neighbours = node2_->GetNeighbours();
+    {
-    ASSERT_EQ(node2_neighbours.size(), 1);
+        auto node2_neighbours = node2_->GetNeighbours();
-    ASSERT_EQ(node2_neighbours[0].neighbour_node.lock(), node1_);
+        ASSERT_EQ(node2_neighbours.size(), 1);
+        ASSERT_EQ(node2_neighbours[0].neighbour_node.lock(), node1_);
+    }
+
+    {
+        auto node3_neighbours = node3_->GetNeighbours();
+        ASSERT_EQ(node3_neighbours.size(), 0);
+    }
+}
+
+TEST_F(NodeTest, dump) {
+    std::cout << node1_->Dump();
+    ASSERT_FALSE(node1_->Dump().empty());
+
+    std::cout << node2_->Dump();
+    ASSERT_FALSE(node2_->Dump().empty());
 }

cpp/unittest/scheduler/normal_test.cpp

@@ -2,6 +2,7 @@
 #include "scheduler/ResourceMgr.h"
 #include "scheduler/Scheduler.h"
 #include "scheduler/task/TestTask.h"
+#include "scheduler/tasklabel/DefaultLabel.h"
 #include "scheduler/SchedInst.h"
 #include "utils/Log.h"
 #include <gtest/gtest.h>
@@ -9,48 +10,44 @@
 
 using namespace zilliz::milvus::engine;
 
-TEST(normal_test, test1) {
+
+TEST(normal_test, inst_test) {
     // ResourceMgr only compose resources, provide unified event
-//    auto res_mgr = std::make_shared<ResourceMgr>();
     auto res_mgr = ResMgrInst::GetInstance();
-    auto disk = res_mgr->Add(ResourceFactory::Create("disk", "ssd", true, false));
+
-    auto cpu = res_mgr->Add(ResourceFactory::Create("cpu", "CPU", 0));
+    res_mgr->Add(ResourceFactory::Create("disk", "DISK", 0, true, false));
-    auto gpu1 = res_mgr->Add(ResourceFactory::Create("gpu", "gpu0", false, false));
+    res_mgr->Add(ResourceFactory::Create("cpu", "CPU", 0, true, true));
-    auto gpu2 = res_mgr->Add(ResourceFactory::Create("gpu", "gpu2", false, false));
 
     auto IO = Connection("IO", 500.0);
-    auto PCIE = Connection("IO", 11000.0);
+    res_mgr->Connect("disk", "cpu", IO);
-    res_mgr->Connect(disk, cpu, IO);
+
-    res_mgr->Connect(cpu, gpu1, PCIE);
+    auto scheduler = SchedInst::GetInstance();
-    res_mgr->Connect(cpu, gpu2, PCIE);
 
     res_mgr->Start();
-
-//    auto scheduler = new Scheduler(res_mgr);
-    auto scheduler = SchedInst::GetInstance();
     scheduler->Start();
 
     const uint64_t NUM_TASK = 1000;
     std::vector<std::shared_ptr<TestTask>> tasks;
     TableFileSchemaPtr dummy = nullptr;
 
-    for (uint64_t i = 0; i < NUM_TASK; ++i) {
+    auto disks = res_mgr->GetDiskResources();
-        if (auto observe = disk.lock()) {
+    ASSERT_FALSE(disks.empty());
+    if (auto observe = disks[0].lock()) {
+        for (uint64_t i = 0; i < NUM_TASK; ++i) {
             auto task = std::make_shared<TestTask>(dummy);
+            task->label() = std::make_shared<DefaultLabel>();
             tasks.push_back(task);
             observe->task_table().Put(task);
         }
     }
 
-    sleep(1);
+    for (auto &task : tasks) {
+        task->Wait();
+        ASSERT_EQ(task->load_count_, 1);
+        ASSERT_EQ(task->exec_count_, 1);
+    }
 
     scheduler->Stop();
     res_mgr->Stop();
 
-    auto pcpu = cpu.lock();
-    for (uint64_t i = 0; i < NUM_TASK; ++i) {
-        auto task = std::static_pointer_cast<TestTask>(pcpu->task_table()[i]->task);
-        ASSERT_EQ(task->load_count_, 1);
-        ASSERT_EQ(task->exec_count_, 1);
-    }
 }

cpp/unittest/scheduler/tasktable_test.cpp

@@ -5,30 +5,37 @@
 
 using namespace zilliz::milvus::engine;
 
+
+/************ TaskTableBaseTest ************/
+
 class TaskTableItemTest : public ::testing::Test {
 protected:
     void
     SetUp() override {
-        item1_.id = 0;
+        std::vector<TaskTableItemState> states{
-        item1_.state = TaskTableItemState::MOVED;
+            TaskTableItemState::INVALID,
-        item1_.priority = 10;
+            TaskTableItemState::START,
+            TaskTableItemState::LOADING,
+            TaskTableItemState::LOADED,
+            TaskTableItemState::EXECUTING,
+            TaskTableItemState::EXECUTED,
+            TaskTableItemState::MOVING,
+            TaskTableItemState::MOVED};
+        for (auto &state : states) {
+            auto item = std::make_shared<TaskTableItem>();
+            item->state = state;
+            items_.emplace_back(item);
+        }
     }
 
     TaskTableItem default_;
-    TaskTableItem item1_;
+    std::vector<TaskTableItemPtr> items_;
 };
 
 TEST_F(TaskTableItemTest, construct) {
     ASSERT_EQ(default_.id, 0);
+    ASSERT_EQ(default_.task, nullptr);
     ASSERT_EQ(default_.state, TaskTableItemState::INVALID);
-    ASSERT_EQ(default_.priority, 0);
-}
-
-TEST_F(TaskTableItemTest, copy) {
-    TaskTableItem another(item1_);
-    ASSERT_EQ(another.id, item1_.id);
-    ASSERT_EQ(another.state, item1_.state);
-    ASSERT_EQ(another.priority, item1_.priority);
 }
 
 TEST_F(TaskTableItemTest, destruct) {
@@ -36,6 +43,107 @@ TEST_F(TaskTableItemTest, destruct) {
     delete p_item;
 }
 
+TEST_F(TaskTableItemTest, is_finish) {
+    for (auto &item : items_) {
+        if (item->state == TaskTableItemState::EXECUTED
+            || item->state == TaskTableItemState::MOVED) {
+            ASSERT_TRUE(item->IsFinish());
+        } else {
+            ASSERT_FALSE(item->IsFinish());
+        }
+    }
+}
+
+TEST_F(TaskTableItemTest, dump) {
+    for (auto &item : items_) {
+        ASSERT_FALSE(item->Dump().empty());
+    }
+}
+
+TEST_F(TaskTableItemTest, load) {
+    for (auto &item : items_) {
+        auto before_state = item->state;
+        auto ret = item->Load();
+        if (before_state == TaskTableItemState::START) {
+            ASSERT_TRUE(ret);
+            ASSERT_EQ(item->state, TaskTableItemState::LOADING);
+        } else {
+            ASSERT_FALSE(ret);
+            ASSERT_EQ(item->state, before_state);
+        }
+    }
+}
+
+TEST_F(TaskTableItemTest, loaded) {
+    for (auto &item : items_) {
+        auto before_state = item->state;
+        auto ret = item->Loaded();
+        if (before_state == TaskTableItemState::LOADING) {
+            ASSERT_TRUE(ret);
+            ASSERT_EQ(item->state, TaskTableItemState::LOADED);
+        } else {
+            ASSERT_FALSE(ret);
+            ASSERT_EQ(item->state, before_state);
+        }
+    }
+}
+
+TEST_F(TaskTableItemTest, execute) {
+    for (auto &item : items_) {
+        auto before_state = item->state;
+        auto ret = item->Execute();
+        if (before_state == TaskTableItemState::LOADED) {
+            ASSERT_TRUE(ret);
+            ASSERT_EQ(item->state, TaskTableItemState::EXECUTING);
+        } else {
+            ASSERT_FALSE(ret);
+            ASSERT_EQ(item->state, before_state);
+        }
+    }
+}
+
+
+TEST_F(TaskTableItemTest, executed) {
+    for (auto &item : items_) {
+        auto before_state = item->state;
+        auto ret = item->Executed();
+        if (before_state == TaskTableItemState::EXECUTING) {
+            ASSERT_TRUE(ret);
+            ASSERT_EQ(item->state, TaskTableItemState::EXECUTED);
+        } else {
+            ASSERT_FALSE(ret);
+            ASSERT_EQ(item->state, before_state);
+        }
+    }
+}
+
+TEST_F(TaskTableItemTest, move) {
+    for (auto &item : items_) {
+        auto before_state = item->state;
+        auto ret = item->Move();
+        if (before_state == TaskTableItemState::LOADED) {
+            ASSERT_TRUE(ret);
+            ASSERT_EQ(item->state, TaskTableItemState::MOVING);
+        } else {
+            ASSERT_FALSE(ret);
+            ASSERT_EQ(item->state, before_state);
+        }
+    }
+}
+
+TEST_F(TaskTableItemTest, moved) {
+    for (auto &item : items_) {
+        auto before_state = item->state;
+        auto ret = item->Moved();
+        if (before_state == TaskTableItemState::MOVING) {
+            ASSERT_TRUE(ret);
+            ASSERT_EQ(item->state, TaskTableItemState::MOVED);
+        } else {
+            ASSERT_FALSE(ret);
+            ASSERT_EQ(item->state, before_state);
+        }
+    }
+}
 
 /************ TaskTableBaseTest ************/
 
@@ -55,6 +163,16 @@ protected:
     TaskTable empty_table_;
 };
 
+TEST_F(TaskTableBaseTest, subscriber) {
+    bool flag = false;
+    auto callback = [&]() {
+        flag = true;
+    };
+    empty_table_.RegisterSubscriber(callback);
+    empty_table_.Put(task1_);
+    ASSERT_TRUE(flag);
+}
+
 
 TEST_F(TaskTableBaseTest, put_task) {
     empty_table_.Put(task1_);
@@ -78,6 +196,125 @@ TEST_F(TaskTableBaseTest, put_empty_batch) {
     empty_table_.Put(tasks);
 }
 
+TEST_F(TaskTableBaseTest, empty) {
+    ASSERT_TRUE(empty_table_.Empty());
+    empty_table_.Put(task1_);
+    ASSERT_FALSE(empty_table_.Empty());
+}
+
+TEST_F(TaskTableBaseTest, size) {
+    ASSERT_EQ(empty_table_.Size(), 0);
+    empty_table_.Put(task1_);
+    ASSERT_EQ(empty_table_.Size(), 1);
+}
+
+TEST_F(TaskTableBaseTest, operator_) {
+    empty_table_.Put(task1_);
+    ASSERT_EQ(empty_table_.Get(0), empty_table_[0]);
+}
+
+TEST_F(TaskTableBaseTest, pick_to_load) {
+    const size_t NUM_TASKS = 10;
+    for (size_t i = 0; i < NUM_TASKS; ++i) {
+        empty_table_.Put(task1_);
+    }
+    empty_table_[0]->state = TaskTableItemState::MOVED;
+    empty_table_[1]->state = TaskTableItemState::EXECUTED;
+
+    auto indexes = empty_table_.PickToLoad(1);
+    ASSERT_EQ(indexes.size(), 1);
+    ASSERT_EQ(indexes[0], 2);
+}
+
+TEST_F(TaskTableBaseTest, pick_to_load_limit) {
+    const size_t NUM_TASKS = 10;
+    for (size_t i = 0; i < NUM_TASKS; ++i) {
+        empty_table_.Put(task1_);
+    }
+    empty_table_[0]->state = TaskTableItemState::MOVED;
+    empty_table_[1]->state = TaskTableItemState::EXECUTED;
+
+    auto indexes = empty_table_.PickToLoad(3);
+    ASSERT_EQ(indexes.size(), 3);
+    ASSERT_EQ(indexes[0], 2);
+    ASSERT_EQ(indexes[1], 3);
+    ASSERT_EQ(indexes[2], 4);
+}
+
+TEST_F(TaskTableBaseTest, pick_to_load_cache) {
+    const size_t NUM_TASKS = 10;
+    for (size_t i = 0; i < NUM_TASKS; ++i) {
+        empty_table_.Put(task1_);
+    }
+    empty_table_[0]->state = TaskTableItemState::MOVED;
+    empty_table_[1]->state = TaskTableItemState::EXECUTED;
+
+    // first pick, non-cache
+    auto indexes = empty_table_.PickToLoad(1);
+    ASSERT_EQ(indexes.size(), 1);
+    ASSERT_EQ(indexes[0], 2);
+
+    // second pick, iterate from 2
+    // invalid state change
+    empty_table_[1]->state = TaskTableItemState::START;
+    indexes = empty_table_.PickToLoad(1);
+    ASSERT_EQ(indexes.size(), 1);
+    ASSERT_EQ(indexes[0], 2);
+}
+
+TEST_F(TaskTableBaseTest, pick_to_execute) {
+    const size_t NUM_TASKS = 10;
+    for (size_t i = 0; i < NUM_TASKS; ++i) {
+        empty_table_.Put(task1_);
+    }
+    empty_table_[0]->state = TaskTableItemState::MOVED;
+    empty_table_[1]->state = TaskTableItemState::EXECUTED;
+    empty_table_[2]->state = TaskTableItemState::LOADED;
+
+    auto indexes = empty_table_.PickToExecute(1);
+    ASSERT_EQ(indexes.size(), 1);
+    ASSERT_EQ(indexes[0], 2);
+}
+
+TEST_F(TaskTableBaseTest, pick_to_execute_limit) {
+    const size_t NUM_TASKS = 10;
+    for (size_t i = 0; i < NUM_TASKS; ++i) {
+        empty_table_.Put(task1_);
+    }
+    empty_table_[0]->state = TaskTableItemState::MOVED;
+    empty_table_[1]->state = TaskTableItemState::EXECUTED;
+    empty_table_[2]->state = TaskTableItemState::LOADED;
+    empty_table_[3]->state = TaskTableItemState::LOADED;
+
+    auto indexes = empty_table_.PickToExecute(3);
+    ASSERT_EQ(indexes.size(), 2);
+    ASSERT_EQ(indexes[0], 2);
+    ASSERT_EQ(indexes[1], 3);
+}
+
+TEST_F(TaskTableBaseTest, pick_to_execute_cache) {
+    const size_t NUM_TASKS = 10;
+    for (size_t i = 0; i < NUM_TASKS; ++i) {
+        empty_table_.Put(task1_);
+    }
+    empty_table_[0]->state = TaskTableItemState::MOVED;
+    empty_table_[1]->state = TaskTableItemState::EXECUTED;
+    empty_table_[2]->state = TaskTableItemState::LOADED;
+
+    // first pick, non-cache
+    auto indexes = empty_table_.PickToExecute(1);
+    ASSERT_EQ(indexes.size(), 1);
+    ASSERT_EQ(indexes[0], 2);
+
+    // second pick, iterate from 2
+    // invalid state change
+    empty_table_[1]->state = TaskTableItemState::START;
+    indexes = empty_table_.PickToExecute(1);
+    ASSERT_EQ(indexes.size(), 1);
+    ASSERT_EQ(indexes[0], 2);
+}
+
+
 /************ TaskTableAdvanceTest ************/
 
 class TaskTableAdvanceTest : public ::testing::Test {
@@ -104,25 +341,116 @@ protected:
 };
 
 TEST_F(TaskTableAdvanceTest, load) {
-    table1_.Load(1);
+    std::vector<TaskTableItemState> before_state;
-    table1_.Loaded(2);
+    for (auto &task : table1_) {
+        before_state.push_back(task->state);
+    }
 
-    ASSERT_EQ(table1_.Get(1)->state, TaskTableItemState::LOADING);
+    for (size_t i = 0; i < table1_.Size(); ++i) {
-    ASSERT_EQ(table1_.Get(2)->state, TaskTableItemState::LOADED);
+        table1_.Load(i);
+    }
+
+    for (size_t i = 0; i < table1_.Size(); ++i) {
+        if (before_state[i] == TaskTableItemState::START) {
+            ASSERT_EQ(table1_.Get(i)->state, TaskTableItemState::LOADING);
+        } else {
+            ASSERT_EQ(table1_.Get(i)->state, before_state[i]);
+        }
+    }
+}
+
+TEST_F(TaskTableAdvanceTest, loaded) {
+    std::vector<TaskTableItemState> before_state;
+    for (auto &task : table1_) {
+        before_state.push_back(task->state);
+    }
+
+    for (size_t i = 0; i < table1_.Size(); ++i) {
+        table1_.Loaded(i);
+    }
+
+    for (size_t i = 0; i < table1_.Size(); ++i) {
+        if (before_state[i] == TaskTableItemState::LOADING) {
+            ASSERT_EQ(table1_.Get(i)->state, TaskTableItemState::LOADED);
+        } else {
+            ASSERT_EQ(table1_.Get(i)->state, before_state[i]);
+        }
+    }
 }
 
 TEST_F(TaskTableAdvanceTest, execute) {
-    table1_.Execute(3);
+    std::vector<TaskTableItemState> before_state;
-    table1_.Executed(4);
+    for (auto &task : table1_) {
+        before_state.push_back(task->state);
+    }
 
-    ASSERT_EQ(table1_.Get(3)->state, TaskTableItemState::EXECUTING);
+    for (size_t i = 0; i < table1_.Size(); ++i) {
-    ASSERT_EQ(table1_.Get(4)->state, TaskTableItemState::EXECUTED);
+        table1_.Execute(i);
+    }
+
+    for (size_t i = 0; i < table1_.Size(); ++i) {
+        if (before_state[i] == TaskTableItemState::LOADED) {
+            ASSERT_EQ(table1_.Get(i)->state, TaskTableItemState::EXECUTING);
+        } else {
+            ASSERT_EQ(table1_.Get(i)->state, before_state[i]);
+        }
+    }
+}
+
+TEST_F(TaskTableAdvanceTest, executed) {
+    std::vector<TaskTableItemState> before_state;
+    for (auto &task : table1_) {
+        before_state.push_back(task->state);
+    }
+
+    for (size_t i = 0; i < table1_.Size(); ++i) {
+        table1_.Executed(i);
+    }
+
+    for (size_t i = 0; i < table1_.Size(); ++i) {
+        if (before_state[i] == TaskTableItemState::EXECUTING) {
+            ASSERT_EQ(table1_.Get(i)->state, TaskTableItemState::EXECUTED);
+        } else {
+            ASSERT_EQ(table1_.Get(i)->state, before_state[i]);
+        }
+    }
 }
 
 TEST_F(TaskTableAdvanceTest, move) {
-    table1_.Move(3);
+    std::vector<TaskTableItemState> before_state;
-    table1_.Moved(6);
+    for (auto &task : table1_) {
+        before_state.push_back(task->state);
+    }
 
-    ASSERT_EQ(table1_.Get(3)->state, TaskTableItemState::MOVING);
+    for (size_t i = 0; i < table1_.Size(); ++i) {
-    ASSERT_EQ(table1_.Get(6)->state, TaskTableItemState::MOVED);
+        table1_.Move(i);
+    }
+
+    for (size_t i = 0; i < table1_.Size(); ++i) {
+        if (before_state[i] == TaskTableItemState::LOADED) {
+            ASSERT_EQ(table1_.Get(i)->state, TaskTableItemState::MOVING);
+        } else {
+            ASSERT_EQ(table1_.Get(i)->state, before_state[i]);
+        }
+    }
 }
+
+TEST_F(TaskTableAdvanceTest, moved) {
+    std::vector<TaskTableItemState> before_state;
+    for (auto &task : table1_) {
+        before_state.push_back(task->state);
+    }
+
+    for (size_t i = 0; i < table1_.Size(); ++i) {
+        table1_.Moved(i);
+    }
+
+    for (size_t i = 0; i < table1_.Size(); ++i) {
+        if (before_state[i] == TaskTableItemState::MOVING) {
+            ASSERT_EQ(table1_.Get(i)->state, TaskTableItemState::MOVED);
+        } else {
+            ASSERT_EQ(table1_.Get(i)->state, before_state[i]);
+        }
+    }
+}
+