[Enhancement] support phased schedule (#47868)

Signed-off-by: stdpain <[email protected]>
(cherry picked from commit d7ad29e)

# Conflicts:
#	be/src/exec/pipeline/scan/olap_scan_context.cpp
#	fe/fe-core/src/main/java/com/starrocks/qe/QeProcessorImpl.java

be/src/exec/CMakeLists.txt

@@ -21,6 +21,7 @@ set(LIBRARY_OUTPUT_PATH "${BUILD_DIR}/src/exec")
 set(EXECUTABLE_OUTPUT_PATH "${BUILD_DIR}/src/exec")
 
 set(EXEC_FILES
+    capture_version_node.cpp
     data_sink.cpp
     empty_set_node.cpp
     exec_node.cpp
@@ -166,6 +167,7 @@ set(EXEC_FILES
     sorting/sort_column.cpp
     sorting/sort_permute.cpp
     connector_scan_node.cpp
+    pipeline/capture_version_operator.cpp
     pipeline/exchange/exchange_merge_sort_source_operator.cpp
     pipeline/exchange/exchange_parallel_merge_source_operator.cpp
     pipeline/exchange/exchange_sink_operator.cpp

be/src/exec/capture_version_node.cpp

@@ -0,0 +1,36 @@
+// Copyright 2021-present StarRocks, Inc. All rights reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     https://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+#include "exec/capture_version_node.h"
+
+#include "exec/pipeline/capture_version_operator.h"
+#include "exec/pipeline/pipeline_builder.h"
+
+namespace starrocks {
+
+pipeline::OpFactories CaptureVersionNode::decompose_to_pipeline(pipeline::PipelineBuilderContext* context) {
+    return OpFactories{std::make_shared<pipeline::CaptureVersionOpFactory>(context->next_operator_id(), id())};
+}
+
+StatusOr<pipeline::MorselQueueFactoryPtr> CaptureVersionNode::scan_range_to_morsel_queue_factory(
+        const std::vector<TScanRangeParams>& scan_ranges) {
+    pipeline::Morsels morsels;
+    for (const auto& scan_range : scan_ranges) {
+        morsels.emplace_back(std::make_unique<pipeline::ScanMorsel>(id(), scan_range));
+    }
+    auto morsel_queue = std::make_unique<pipeline::FixedMorselQueue>(std::move(morsels));
+    return std::make_unique<pipeline::SharedMorselQueueFactory>(std::move(morsel_queue), 1);
+}
+
+} // namespace starrocks

be/src/exec/capture_version_node.h

@@ -0,0 +1,32 @@
+// Copyright 2021-present StarRocks, Inc. All rights reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     https://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+#pragma once
+
+#include "exec/exec_node.h"
+#include "exec/scan_node.h"
+
+namespace starrocks {
+class CaptureVersionNode final : public ExecNode {
+public:
+    CaptureVersionNode(ObjectPool* pool, const TPlanNode& tnode, const DescriptorTbl& descs)
+            : ExecNode(pool, tnode, descs) {}
+    ~CaptureVersionNode() override = default;
+
+    StatusOr<pipeline::MorselQueueFactoryPtr> scan_range_to_morsel_queue_factory(
+            const std::vector<TScanRangeParams>& global_scan_ranges);
+
+    pipeline::OpFactories decompose_to_pipeline(pipeline::PipelineBuilderContext* context) override;
+};
+} // namespace starrocks

be/src/exec/exec_node.cpp

@@ -50,6 +50,7 @@
 #include "exec/aggregate/distinct_streaming_node.h"
 #include "exec/analytic_node.h"
 #include "exec/assert_num_rows_node.h"
+#include "exec/capture_version_node.h"
 #include "exec/connector_scan_node.h"
 #include "exec/cross_join_node.h"
 #include "exec/dict_decode_node.h"
@@ -74,6 +75,7 @@
 #include "exec/union_node.h"
 #include "exprs/dictionary_get_expr.h"
 #include "exprs/expr_context.h"
+#include "gen_cpp/PlanNodes_types.h"
 #include "gutil/strings/substitute.h"
 #include "runtime/descriptors.h"
 #include "runtime/exec_env.h"
@@ -564,6 +566,10 @@ Status ExecNode::create_vectorized_node(starrocks::RuntimeState* state, starrock
         *node = pool->add(new StreamAggregateNode(pool, tnode, descs));
         return Status::OK();
     }
+    case TPlanNodeType::CAPTURE_VERSION_NODE: {
+        *node = pool->add(new CaptureVersionNode(pool, tnode, descs));
+        return Status::OK();
+    }
     default:
         return Status::InternalError(strings::Substitute("Vectorized engine not support node: $0", tnode.node_type));
     }

be/src/exec/olap_scan_node.cpp

@@ -746,6 +746,23 @@ StatusOr<TabletSharedPtr> OlapScanNode::get_tablet(const TInternalScanRange* sca
     return tablet;
 }
 
+StatusOr<std::vector<RowsetSharedPtr>> OlapScanNode::capture_tablet_rowsets(const TabletSharedPtr& tablet,
+                                                                            const TInternalScanRange* scan_range) {
+    std::vector<RowsetSharedPtr> rowsets;
+    if (scan_range->__isset.gtid) {
+        std::shared_lock l(tablet->get_header_lock());
+        RETURN_IF_ERROR(tablet->capture_consistent_rowsets(scan_range->gtid, &rowsets));
+        Rowset::acquire_readers(rowsets);
+    } else {
+        int64_t version = strtoul(scan_range->version.c_str(), nullptr, 10);
+        // Capture row sets of this version tablet.
+        std::shared_lock l(tablet->get_header_lock());
+        RETURN_IF_ERROR(tablet->capture_consistent_rowsets(Version(0, version), &rowsets));
+        Rowset::acquire_readers(rowsets);
+    }
+    return rowsets;
+}
+
 int OlapScanNode::estimated_max_concurrent_chunks() const {
     // We temporarily assume that the memory tried in the storage layer
     // is the same size as the chunk_size * _estimated_scan_row_bytes.

be/src/exec/olap_scan_node.h

@@ -87,6 +87,9 @@ class OlapScanNode final : public starrocks::ScanNode {
     int estimated_max_concurrent_chunks() const;
 
     static StatusOr<TabletSharedPtr> get_tablet(const TInternalScanRange* scan_range);
+    static StatusOr<std::vector<RowsetSharedPtr>> capture_tablet_rowsets(const TabletSharedPtr& tablet,
+                                                                         const TInternalScanRange* scan_range);
+
     static int compute_priority(int32_t num_submitted_tasks);
 
     int io_tasks_per_scan_operator() const override {

be/src/exec/pipeline/capture_version_operator.cpp

@@ -0,0 +1,52 @@
+// Copyright 2021-present StarRocks, Inc. All rights reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     https://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+#include "exec/pipeline/capture_version_operator.h"
+
+#include "common/logging.h"
+#include "exec/olap_scan_node.h"
+#include "runtime/runtime_state.h"
+#include "storage/rowset/rowset.h"
+#include "util/defer_op.h"
+
+namespace starrocks::pipeline {
+Status CaptureVersionOperator::prepare(RuntimeState* state) {
+    RETURN_IF_ERROR(SourceOperator::prepare(state));
+    _capture_tablet_rowsets_timer = ADD_TIMER(_unique_metrics, "CaptureTabletRowsetsTime");
+
+    return Status::OK();
+}
+
+StatusOr<ChunkPtr> CaptureVersionOperator::pull_chunk(RuntimeState* state) {
+    // do capture rowset here
+    auto defer = DeferOp([this]() { _is_finished = true; });
+    std::vector<std::vector<RowsetSharedPtr>> tablet_rowsets;
+
+    {
+        SCOPED_TIMER(_capture_tablet_rowsets_timer);
+        auto scan_rages = _morsel_queue->prepare_olap_scan_ranges();
+        std::vector<std::vector<RowsetSharedPtr>> tablet_rowsets;
+        VLOG_QUERY << "capture rowset scan_ranges nums:" << scan_rages.size();
+        for (auto& scan_range : scan_rages) {
+            ASSIGN_OR_RETURN(TabletSharedPtr tablet, OlapScanNode::get_tablet(scan_range));
+            ASSIGN_OR_RETURN(auto rowset, OlapScanNode::capture_tablet_rowsets(tablet, scan_range));
+            tablet_rowsets.emplace_back(std::move(rowset));
+        }
+        //
+        _rowset_release_guard = MultiRowsetReleaseGuard(std::move(tablet_rowsets), adopt_acquire_t{});
+    }
+    return nullptr;
+}
+
+} // namespace starrocks::pipeline

be/src/exec/pipeline/capture_version_operator.h

@@ -0,0 +1,60 @@
+// Copyright 2021-present StarRocks, Inc. All rights reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     https://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+#pragma once
+
+#include "exec/pipeline/source_operator.h"
+#include "util/runtime_profile.h"
+
+namespace starrocks::pipeline {
+
+// CaptureVersion returns an empty result set.
+class CaptureVersionOperator final : public SourceOperator {
+public:
+    CaptureVersionOperator(OperatorFactory* factory, int32_t id, int32_t plan_node_id, int32_t driver_sequence)
+            : SourceOperator(factory, id, "capture_version", plan_node_id, false, driver_sequence) {}
+
+    ~CaptureVersionOperator() override = default;
+
+    Status prepare(RuntimeState* state) override;
+
+    bool has_output() const override { return !_is_finished; }
+
+    bool is_finished() const override { return _is_finished; }
+
+    StatusOr<ChunkPtr> pull_chunk(RuntimeState* state) override;
+
+private:
+    MultiRowsetReleaseGuard _rowset_release_guard;
+    RuntimeProfile::Counter* _capture_tablet_rowsets_timer = nullptr;
+    bool _is_finished{};
+};
+
+class CaptureVersionOpFactory final : public SourceOperatorFactory {
+public:
+    CaptureVersionOpFactory(int32_t id, int32_t plan_node_id)
+            : SourceOperatorFactory(id, "capture_version", plan_node_id) {}
+
+    ~CaptureVersionOpFactory() override = default;
+
+    OperatorPtr create(int32_t degree_of_parallelism, int32_t driver_sequence) override {
+        return std::make_shared<CaptureVersionOperator>(this, _id, _plan_node_id, driver_sequence);
+    }
+
+    bool with_morsels() const override { return true; }
+
+    SourceOperatorFactory::AdaptiveState adaptive_initial_state() const override { return AdaptiveState::ACTIVE; }
+};
+
+} // namespace starrocks::pipeline

be/src/exec/pipeline/fragment_context.cpp

@@ -14,15 +14,20 @@
 
 #include "exec/pipeline/fragment_context.h"
 
+#include <memory>
+
 #include "exec/data_sink.h"
 #include "exec/pipeline/group_execution/execution_group.h"
 #include "exec/pipeline/pipeline_driver_executor.h"
 #include "exec/pipeline/stream_pipeline_driver.h"
 #include "exec/workgroup/work_group.h"
+#include "runtime/client_cache.h"
 #include "runtime/data_stream_mgr.h"
 #include "runtime/exec_env.h"
 #include "runtime/stream_load/stream_load_context.h"
 #include "runtime/stream_load/transaction_mgr.h"
+#include "util/threadpool.h"
+#include "util/thrift_rpc_helper.h"
 #include "util/time.h"
 
 namespace starrocks::pipeline {
@@ -85,6 +90,37 @@ void FragmentContext::count_down_execution_group(size_t val) {
     auto status = final_status();
     _workgroup->executors()->driver_executor()->report_exec_state(query_ctx, this, status, true, true);
 
+    if (_report_when_finish) {
+        /// TODO: report fragment finish to BE coordinator
+        TReportFragmentFinishResponse res;
+        TReportFragmentFinishParams params;
+        params.__set_query_id(query_id());
+        params.__set_fragment_instance_id(fragment_instance_id());
+        // params.query_id = query_id();
+        // params.fragment_instance_id = fragment_instance_id();
+        const auto& fe_addr = state->fragment_ctx()->fe_addr();
+
+        class RpcRunnable : public Runnable {
+        public:
+            RpcRunnable(const TNetworkAddress& fe_addr, const TReportFragmentFinishResponse& res,
+                        const TReportFragmentFinishParams& params)
+                    : fe_addr(fe_addr), res(res), params(params) {}
+            const TNetworkAddress fe_addr;
+            TReportFragmentFinishResponse res;
+            const TReportFragmentFinishParams params;
+
+            void run() override {
+                (void)ThriftRpcHelper::rpc<FrontendServiceClient>(
+                        fe_addr.hostname, fe_addr.port,
+                        [&](FrontendServiceConnection& client) { client->reportFragmentFinish(res, params); });
+            }
+        };
+        //
+        std::shared_ptr<Runnable> runnable;
+        runnable = std::make_shared<RpcRunnable>(fe_addr, res, params);
+        (void)state->exec_env()->streaming_load_thread_pool()->submit(runnable);
+    }
+
     destroy_pass_through_chunk_buffer();
 
     query_ctx->count_down_fragments();

be/src/exec/pipeline/fragment_context.h

@@ -169,6 +169,8 @@ class FragmentContext {
     bool enable_group_execution() const { return _enable_group_execution; }
     void set_enable_group_execution(bool enable_group_execution) { _enable_group_execution = enable_group_execution; }
 
+    void set_report_when_finish(bool report) { _report_when_finish = report; }
+
 private:
     bool _enable_group_execution = false;
     // Id of this query
@@ -227,6 +229,8 @@ class FragmentContext {
 
     RuntimeProfile::Counter* _jit_counter = nullptr;
     RuntimeProfile::Counter* _jit_timer = nullptr;
+
+    bool _report_when_finish{};
 };
 
 class FragmentContextManager {

be/src/exec/pipeline/fragment_executor.cpp

@@ -18,6 +18,7 @@
 #include <unordered_map>
 
 #include "common/config.h"
+#include "exec/capture_version_node.h"
 #include "exec/cross_join_node.h"
 #include "exec/exchange_node.h"
 #include "exec/exec_node.h"
@@ -103,6 +104,9 @@ Status FragmentExecutor::_prepare_query_ctx(ExecEnv* exec_env, const UnifiedExec
     }
 
     _query_ctx = exec_env->query_context_mgr()->get_or_register(query_id);
+    if (_query_ctx == nullptr) {
+        return Status::Cancelled("Query has been cancelled");
+    }
     _query_ctx->set_exec_env(exec_env);
     if (params.__isset.instances_number) {
         _query_ctx->set_total_fragments(params.instances_number);
@@ -253,6 +257,8 @@ Status FragmentExecutor::_prepare_runtime_state(ExecEnv* exec_env, const Unified
         exec_env->runtime_filter_worker()->open_query(query_id, query_options, *runtime_filter_params, true);
     }
     _fragment_ctx->prepare_pass_through_chunk_buffer();
+    _fragment_ctx->set_report_when_finish(request.unique().params.__isset.report_when_finish &&
+                                          request.unique().params.report_when_finish);
 
     auto* obj_pool = runtime_state->obj_pool();
     // Set up desc tbl
@@ -505,6 +511,15 @@ Status FragmentExecutor::_prepare_exec_plan(ExecEnv* exec_env, const UnifiedExec
         }
     }
 
+    std::vector<ExecNode*> capture_version_nodes;
+    plan->collect_nodes(TPlanNodeType::CAPTURE_VERSION_NODE, &capture_version_nodes);
+    for (auto* node : capture_version_nodes) {
+        const std::vector<TScanRangeParams>& scan_ranges = request.scan_ranges_of_node(node->id());
+        ASSIGN_OR_RETURN(auto morsel_queue_factory,
+                         down_cast<CaptureVersionNode*>(node)->scan_range_to_morsel_queue_factory(scan_ranges));
+        morsel_queue_factories.emplace(node->id(), std::move(morsel_queue_factory));
+    }
+
     if (_wg && _wg->big_query_scan_rows_limit() > 0) {
         if (logical_scan_limit >= 0 && logical_scan_limit <= _wg->big_query_scan_rows_limit()) {
             _query_ctx->set_scan_limit(std::max(_wg->big_query_scan_rows_limit(), physical_scan_limit));
@@ -747,7 +762,8 @@ Status FragmentExecutor::prepare(ExecEnv* exec_env, const TExecPlanFragmentParam
             LOG(WARNING) << "Prepare fragment failed: " << print_id(request.common().params.query_id)
                          << " fragment_instance_id=" << print_id(request.fragment_instance_id())
                          << " is_stream_pipeline=" << request.is_stream_pipeline()
-                         << " backend_num=" << request.backend_num() << " fragment= " << request.common().fragment;
+                         << " backend_num=" << request.backend_num();
+            VLOG_QUERY << "Prepare fragment failed fragment=" << request.common().fragment;
         }
     });