Coverage Report

Created: 2026-05-28 09:54

next uncovered line (L), next uncovered region (R), next uncovered branch (B)
be/src/exec/pipeline/pipeline_fragment_context.cpp
Line
Count
Source
1
// Licensed to the Apache Software Foundation (ASF) under one
2
// or more contributor license agreements.  See the NOTICE file
3
// distributed with this work for additional information
4
// regarding copyright ownership.  The ASF licenses this file
5
// to you under the Apache License, Version 2.0 (the
6
// "License"); you may not use this file except in compliance
7
// with the License.  You may obtain a copy of the License at
8
//
9
//   http://www.apache.org/licenses/LICENSE-2.0
10
//
11
// Unless required by applicable law or agreed to in writing,
12
// software distributed under the License is distributed on an
13
// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
14
// KIND, either express or implied.  See the License for the
15
// specific language governing permissions and limitations
16
// under the License.
17
18
#include "exec/pipeline/pipeline_fragment_context.h"
19
20
#include <gen_cpp/DataSinks_types.h>
21
#include <gen_cpp/FrontendService.h>
22
#include <gen_cpp/FrontendService_types.h>
23
#include <gen_cpp/PaloInternalService_types.h>
24
#include <gen_cpp/PlanNodes_types.h>
25
#include <pthread.h>
26
27
#include <algorithm>
28
#include <cstdlib>
29
// IWYU pragma: no_include <bits/chrono.h>
30
#include <fmt/format.h>
31
#include <thrift/Thrift.h>
32
#include <thrift/protocol/TDebugProtocol.h>
33
#include <thrift/transport/TTransportException.h>
34
35
#include <chrono> // IWYU pragma: keep
36
#include <map>
37
#include <memory>
38
#include <ostream>
39
#include <utility>
40
41
#include "cloud/config.h"
42
#include "common/cast_set.h"
43
#include "common/config.h"
44
#include "common/exception.h"
45
#include "common/logging.h"
46
#include "common/status.h"
47
#include "exec/exchange/local_exchange_sink_operator.h"
48
#include "exec/exchange/local_exchange_source_operator.h"
49
#include "exec/exchange/local_exchanger.h"
50
#include "exec/exchange/vdata_stream_mgr.h"
51
#include "exec/operator/aggregation_sink_operator.h"
52
#include "exec/operator/aggregation_source_operator.h"
53
#include "exec/operator/analytic_sink_operator.h"
54
#include "exec/operator/analytic_source_operator.h"
55
#include "exec/operator/assert_num_rows_operator.h"
56
#include "exec/operator/blackhole_sink_operator.h"
57
#include "exec/operator/bucketed_aggregation_sink_operator.h"
58
#include "exec/operator/bucketed_aggregation_source_operator.h"
59
#include "exec/operator/cache_sink_operator.h"
60
#include "exec/operator/cache_source_operator.h"
61
#include "exec/operator/datagen_operator.h"
62
#include "exec/operator/dict_sink_operator.h"
63
#include "exec/operator/distinct_streaming_aggregation_operator.h"
64
#include "exec/operator/empty_set_operator.h"
65
#include "exec/operator/exchange_sink_operator.h"
66
#include "exec/operator/exchange_source_operator.h"
67
#include "exec/operator/file_scan_operator.h"
68
#include "exec/operator/group_commit_block_sink_operator.h"
69
#include "exec/operator/group_commit_scan_operator.h"
70
#include "exec/operator/hashjoin_build_sink.h"
71
#include "exec/operator/hashjoin_probe_operator.h"
72
#include "exec/operator/hive_table_sink_operator.h"
73
#include "exec/operator/iceberg_delete_sink_operator.h"
74
#include "exec/operator/iceberg_merge_sink_operator.h"
75
#include "exec/operator/iceberg_table_sink_operator.h"
76
#include "exec/operator/jdbc_scan_operator.h"
77
#include "exec/operator/jdbc_table_sink_operator.h"
78
#include "exec/operator/local_merge_sort_source_operator.h"
79
#include "exec/operator/materialization_opertor.h"
80
#include "exec/operator/maxcompute_table_sink_operator.h"
81
#include "exec/operator/memory_scratch_sink_operator.h"
82
#include "exec/operator/meta_scan_operator.h"
83
#include "exec/operator/multi_cast_data_stream_sink.h"
84
#include "exec/operator/multi_cast_data_stream_source.h"
85
#include "exec/operator/nested_loop_join_build_operator.h"
86
#include "exec/operator/nested_loop_join_probe_operator.h"
87
#include "exec/operator/olap_scan_operator.h"
88
#include "exec/operator/olap_table_sink_operator.h"
89
#include "exec/operator/olap_table_sink_v2_operator.h"
90
#include "exec/operator/partition_sort_sink_operator.h"
91
#include "exec/operator/partition_sort_source_operator.h"
92
#include "exec/operator/partitioned_aggregation_sink_operator.h"
93
#include "exec/operator/partitioned_aggregation_source_operator.h"
94
#include "exec/operator/partitioned_hash_join_probe_operator.h"
95
#include "exec/operator/partitioned_hash_join_sink_operator.h"
96
#include "exec/operator/rec_cte_anchor_sink_operator.h"
97
#include "exec/operator/rec_cte_scan_operator.h"
98
#include "exec/operator/rec_cte_sink_operator.h"
99
#include "exec/operator/rec_cte_source_operator.h"
100
#include "exec/operator/repeat_operator.h"
101
#include "exec/operator/result_file_sink_operator.h"
102
#include "exec/operator/result_sink_operator.h"
103
#include "exec/operator/schema_scan_operator.h"
104
#include "exec/operator/select_operator.h"
105
#include "exec/operator/set_probe_sink_operator.h"
106
#include "exec/operator/set_sink_operator.h"
107
#include "exec/operator/set_source_operator.h"
108
#include "exec/operator/sort_sink_operator.h"
109
#include "exec/operator/sort_source_operator.h"
110
#include "exec/operator/spill_iceberg_table_sink_operator.h"
111
#include "exec/operator/spill_sort_sink_operator.h"
112
#include "exec/operator/spill_sort_source_operator.h"
113
#include "exec/operator/streaming_aggregation_operator.h"
114
#include "exec/operator/table_function_operator.h"
115
#include "exec/operator/tvf_table_sink_operator.h"
116
#include "exec/operator/union_sink_operator.h"
117
#include "exec/operator/union_source_operator.h"
118
#include "exec/pipeline/dependency.h"
119
#include "exec/pipeline/pipeline_task.h"
120
#include "exec/pipeline/task_scheduler.h"
121
#include "exec/runtime_filter/runtime_filter_mgr.h"
122
#include "exec/sort/topn_sorter.h"
123
#include "exec/spill/spill_file.h"
124
#include "io/fs/stream_load_pipe.h"
125
#include "load/stream_load/new_load_stream_mgr.h"
126
#include "runtime/exec_env.h"
127
#include "runtime/fragment_mgr.h"
128
#include "runtime/result_buffer_mgr.h"
129
#include "runtime/runtime_state.h"
130
#include "runtime/thread_context.h"
131
#include "service/backend_options.h"
132
#include "util/client_cache.h"
133
#include "util/countdown_latch.h"
134
#include "util/debug_util.h"
135
#include "util/network_util.h"
136
#include "util/uid_util.h"
137
138
namespace doris {
139
PipelineFragmentContext::PipelineFragmentContext(
140
        TUniqueId query_id, const TPipelineFragmentParams& request,
141
        std::shared_ptr<QueryContext> query_ctx, ExecEnv* exec_env,
142
        const std::function<void(RuntimeState*, Status*)>& call_back)
143
447k
        : _query_id(std::move(query_id)),
144
447k
          _fragment_id(request.fragment_id),
145
447k
          _exec_env(exec_env),
146
447k
          _query_ctx(std::move(query_ctx)),
147
447k
          _call_back(call_back),
148
447k
          _is_report_on_cancel(true),
149
447k
          _params(request),
150
447k
          _parallel_instances(_params.__isset.parallel_instances ? _params.parallel_instances : 0),
151
447k
          _need_notify_close(request.__isset.need_notify_close ? request.need_notify_close
152
447k
                                                               : false) {
153
447k
    _fragment_watcher.start();
154
447k
}
155
156
447k
PipelineFragmentContext::~PipelineFragmentContext() {
157
447k
    LOG_INFO("PipelineFragmentContext::~PipelineFragmentContext")
158
447k
            .tag("query_id", print_id(_query_id))
159
447k
            .tag("fragment_id", _fragment_id);
160
447k
    _release_resource();
161
447k
    {
162
        // The memory released by the query end is recorded in the query mem tracker.
163
447k
        SCOPED_SWITCH_THREAD_MEM_TRACKER_LIMITER(_query_ctx->query_mem_tracker());
164
447k
        _runtime_state.reset();
165
447k
        _query_ctx.reset();
166
447k
    }
167
447k
}
168
169
149
bool PipelineFragmentContext::is_timeout(timespec now) const {
170
149
    if (_timeout <= 0) {
171
0
        return false;
172
0
    }
173
149
    return _fragment_watcher.elapsed_time_seconds(now) > _timeout;
174
149
}
175
176
// notify_close() transitions the PFC from "waiting for external close notification" to
177
// "self-managed close". For recursive CTE fragments, the old PFC is kept alive until
178
// the rerun_fragment(wait_for_destroy) RPC calls this to trigger shutdown.
179
// Returns true if all tasks have already closed (i.e., the PFC can be safely destroyed).
180
9.58k
bool PipelineFragmentContext::notify_close() {
181
9.58k
    bool all_closed = false;
182
9.58k
    bool need_remove = false;
183
9.58k
    {
184
9.58k
        std::lock_guard<std::mutex> l(_task_mutex);
185
9.58k
        if (_closed_tasks >= _total_tasks) {
186
3.45k
            if (_need_notify_close) {
187
                // Fragment was cancelled and waiting for notify to close.
188
                // Record that we need to remove from fragment mgr, but do it
189
                // after releasing _task_mutex to avoid ABBA deadlock with
190
                // dump_pipeline_tasks() (which acquires _pipeline_map lock
191
                // first, then _task_mutex via debug_string()).
192
3.38k
                need_remove = true;
193
3.38k
            }
194
3.45k
            all_closed = true;
195
3.45k
        }
196
        // make fragment release by self after cancel
197
9.58k
        _need_notify_close = false;
198
9.58k
    }
199
9.58k
    if (need_remove) {
200
3.38k
        _exec_env->fragment_mgr()->remove_pipeline_context({_query_id, _fragment_id});
201
3.38k
    }
202
9.58k
    return all_closed;
203
9.58k
}
204
205
// Must not add lock in this method. Because it will call query ctx cancel. And
206
// QueryCtx cancel will call fragment ctx cancel. And Also Fragment ctx's running
207
// Method like exchange sink buffer will call query ctx cancel. If we add lock here
208
// There maybe dead lock.
209
6.10k
void PipelineFragmentContext::cancel(const Status reason) {
210
6.10k
    LOG_INFO("PipelineFragmentContext::cancel")
211
6.10k
            .tag("query_id", print_id(_query_id))
212
6.10k
            .tag("fragment_id", _fragment_id)
213
6.10k
            .tag("reason", reason.to_string());
214
6.10k
    if (notify_close()) {
215
80
        return;
216
80
    }
217
    // Timeout is a special error code, we need print current stack to debug timeout issue.
218
6.02k
    if (reason.is<ErrorCode::TIMEOUT>()) {
219
1
        auto dbg_str = fmt::format("PipelineFragmentContext is cancelled due to timeout:\n{}",
220
1
                                   debug_string());
221
1
        LOG_LONG_STRING(WARNING, dbg_str);
222
1
    }
223
224
    // `ILLEGAL_STATE` means queries this fragment belongs to was not found in FE (maybe finished)
225
6.02k
    if (reason.is<ErrorCode::ILLEGAL_STATE>()) {
226
0
        LOG_WARNING("PipelineFragmentContext is cancelled due to illegal state : {}",
227
0
                    debug_string());
228
0
    }
229
230
6.02k
    if (reason.is<ErrorCode::MEM_LIMIT_EXCEEDED>() || reason.is<ErrorCode::MEM_ALLOC_FAILED>()) {
231
0
        print_profile("cancel pipeline, reason: " + reason.to_string());
232
0
    }
233
234
6.02k
    if (auto error_url = get_load_error_url(); !error_url.empty()) {
235
23
        _query_ctx->set_load_error_url(error_url);
236
23
    }
237
238
6.02k
    if (auto first_error_msg = get_first_error_msg(); !first_error_msg.empty()) {
239
23
        _query_ctx->set_first_error_msg(first_error_msg);
240
23
    }
241
242
6.02k
    _query_ctx->cancel(reason, _fragment_id);
243
6.02k
    if (reason.is<ErrorCode::LIMIT_REACH>()) {
244
458
        _is_report_on_cancel = false;
245
5.56k
    } else {
246
24.3k
        for (auto& id : _fragment_instance_ids) {
247
24.3k
            LOG(WARNING) << "PipelineFragmentContext cancel instance: " << print_id(id);
248
24.3k
        }
249
5.56k
    }
250
    // Get pipe from new load stream manager and send cancel to it or the fragment may hang to wait read from pipe
251
    // For stream load the fragment's query_id == load id, it is set in FE.
252
6.02k
    auto stream_load_ctx = _exec_env->new_load_stream_mgr()->get(_query_id);
253
6.02k
    if (stream_load_ctx != nullptr) {
254
31
        stream_load_ctx->pipe->cancel(reason.to_string());
255
        // Set error URL here because after pipe is cancelled, stream load execution may return early.
256
        // We need to set the error URL at this point to ensure error information is properly
257
        // propagated to the client.
258
31
        stream_load_ctx->error_url = get_load_error_url();
259
31
        stream_load_ctx->first_error_msg = get_first_error_msg();
260
31
    }
261
262
25.6k
    for (auto& tasks : _tasks) {
263
57.2k
        for (auto& task : tasks) {
264
57.2k
            task.first->unblock_all_dependencies();
265
57.2k
        }
266
25.6k
    }
267
6.02k
}
268
269
708k
PipelinePtr PipelineFragmentContext::add_pipeline(PipelinePtr parent, int idx) {
270
708k
    PipelineId id = _next_pipeline_id++;
271
708k
    auto pipeline = std::make_shared<Pipeline>(
272
708k
            id, parent ? std::min(parent->num_tasks(), _num_instances) : _num_instances,
273
708k
            parent ? parent->num_tasks() : _num_instances);
274
708k
    if (idx >= 0) {
275
122k
        _pipelines.insert(_pipelines.begin() + idx, pipeline);
276
586k
    } else {
277
586k
        _pipelines.emplace_back(pipeline);
278
586k
    }
279
708k
    if (parent) {
280
255k
        parent->set_children(pipeline);
281
255k
    }
282
708k
    return pipeline;
283
708k
}
284
285
447k
Status PipelineFragmentContext::_build_and_prepare_full_pipeline(ThreadPool* thread_pool) {
286
447k
    {
287
447k
        SCOPED_TIMER(_build_pipelines_timer);
288
        // 2. Build pipelines with operators in this fragment.
289
447k
        auto root_pipeline = add_pipeline();
290
447k
        RETURN_IF_ERROR(_build_pipelines(_runtime_state->obj_pool(), *_query_ctx->desc_tbl,
291
447k
                                         &_root_op, root_pipeline));
292
293
        // 3. Create sink operator
294
447k
        if (!_params.fragment.__isset.output_sink) {
295
0
            return Status::InternalError("No output sink in this fragment!");
296
0
        }
297
447k
        RETURN_IF_ERROR(_create_data_sink(_runtime_state->obj_pool(), _params.fragment.output_sink,
298
447k
                                          _params.fragment.output_exprs, _params,
299
447k
                                          root_pipeline->output_row_desc(), _runtime_state.get(),
300
447k
                                          *_desc_tbl, root_pipeline->id()));
301
447k
        RETURN_IF_ERROR(_sink->init(_params.fragment.output_sink));
302
447k
        RETURN_IF_ERROR(root_pipeline->set_sink(_sink));
303
304
585k
        for (PipelinePtr& pipeline : _pipelines) {
305
585k
            DCHECK(pipeline->sink() != nullptr) << pipeline->operators().size();
306
585k
            RETURN_IF_ERROR(pipeline->sink()->set_child(pipeline->operators().back()));
307
585k
        }
308
447k
    }
309
    // 4. Build local exchanger
310
447k
    if (_runtime_state->enable_local_shuffle()) {
311
443k
        SCOPED_TIMER(_plan_local_exchanger_timer);
312
443k
        RETURN_IF_ERROR(_plan_local_exchange(_params.num_buckets,
313
443k
                                             _params.bucket_seq_to_instance_idx,
314
443k
                                             _params.shuffle_idx_to_instance_idx));
315
443k
    }
316
317
    // 5. Initialize global states in pipelines.
318
709k
    for (PipelinePtr& pipeline : _pipelines) {
319
709k
        SCOPED_TIMER(_prepare_all_pipelines_timer);
320
709k
        pipeline->children().clear();
321
709k
        RETURN_IF_ERROR(pipeline->prepare(_runtime_state.get()));
322
709k
    }
323
324
445k
    {
325
445k
        SCOPED_TIMER(_build_tasks_timer);
326
        // 6. Build pipeline tasks and initialize local state.
327
445k
        RETURN_IF_ERROR(_build_pipeline_tasks(thread_pool));
328
445k
    }
329
330
445k
    return Status::OK();
331
445k
}
332
333
447k
Status PipelineFragmentContext::prepare(ThreadPool* thread_pool) {
334
447k
    if (_prepared) {
335
0
        return Status::InternalError("Already prepared");
336
0
    }
337
447k
    if (_params.__isset.query_options && _params.query_options.__isset.execution_timeout) {
338
447k
        _timeout = _params.query_options.execution_timeout;
339
447k
    }
340
341
447k
    _fragment_level_profile = std::make_unique<RuntimeProfile>("PipelineContext");
342
447k
    _prepare_timer = ADD_TIMER(_fragment_level_profile, "PrepareTime");
343
447k
    SCOPED_TIMER(_prepare_timer);
344
447k
    _build_pipelines_timer = ADD_TIMER(_fragment_level_profile, "BuildPipelinesTime");
345
447k
    _init_context_timer = ADD_TIMER(_fragment_level_profile, "InitContextTime");
346
447k
    _plan_local_exchanger_timer = ADD_TIMER(_fragment_level_profile, "PlanLocalLocalExchangerTime");
347
447k
    _build_tasks_timer = ADD_TIMER(_fragment_level_profile, "BuildTasksTime");
348
447k
    _prepare_all_pipelines_timer = ADD_TIMER(_fragment_level_profile, "PrepareAllPipelinesTime");
349
447k
    {
350
447k
        SCOPED_TIMER(_init_context_timer);
351
447k
        cast_set(_num_instances, _params.local_params.size());
352
447k
        _total_instances =
353
447k
                _params.__isset.total_instances ? _params.total_instances : _num_instances;
354
355
447k
        auto* fragment_context = this;
356
357
447k
        if (_params.query_options.__isset.is_report_success) {
358
446k
            fragment_context->set_is_report_success(_params.query_options.is_report_success);
359
446k
        }
360
361
        // 1. Set up the global runtime state.
362
447k
        _runtime_state = RuntimeState::create_unique(
363
447k
                _params.query_id, _params.fragment_id, _params.query_options,
364
447k
                _query_ctx->query_globals, _exec_env, _query_ctx.get());
365
447k
        _runtime_state->set_task_execution_context(shared_from_this());
366
447k
        SCOPED_SWITCH_THREAD_MEM_TRACKER_LIMITER(_runtime_state->query_mem_tracker());
367
447k
        if (_params.__isset.backend_id) {
368
442k
            _runtime_state->set_backend_id(_params.backend_id);
369
442k
        }
370
447k
        if (_params.__isset.import_label) {
371
241
            _runtime_state->set_import_label(_params.import_label);
372
241
        }
373
447k
        if (_params.__isset.db_name) {
374
193
            _runtime_state->set_db_name(_params.db_name);
375
193
        }
376
447k
        if (_params.__isset.load_job_id) {
377
0
            _runtime_state->set_load_job_id(_params.load_job_id);
378
0
        }
379
380
447k
        if (_params.is_simplified_param) {
381
152k
            _desc_tbl = _query_ctx->desc_tbl;
382
295k
        } else {
383
295k
            DCHECK(_params.__isset.desc_tbl);
384
295k
            RETURN_IF_ERROR(DescriptorTbl::create(_runtime_state->obj_pool(), _params.desc_tbl,
385
295k
                                                  &_desc_tbl));
386
295k
        }
387
447k
        _runtime_state->set_desc_tbl(_desc_tbl);
388
447k
        _runtime_state->set_num_per_fragment_instances(_params.num_senders);
389
447k
        _runtime_state->set_load_stream_per_node(_params.load_stream_per_node);
390
447k
        _runtime_state->set_total_load_streams(_params.total_load_streams);
391
447k
        _runtime_state->set_num_local_sink(_params.num_local_sink);
392
393
        // init fragment_instance_ids
394
447k
        const auto target_size = _params.local_params.size();
395
447k
        _fragment_instance_ids.resize(target_size);
396
1.70M
        for (size_t i = 0; i < _params.local_params.size(); i++) {
397
1.26M
            auto fragment_instance_id = _params.local_params[i].fragment_instance_id;
398
1.26M
            _fragment_instance_ids[i] = fragment_instance_id;
399
1.26M
        }
400
447k
    }
401
402
447k
    RETURN_IF_ERROR(_build_and_prepare_full_pipeline(thread_pool));
403
404
446k
    _init_next_report_time();
405
406
446k
    _prepared = true;
407
446k
    return Status::OK();
408
447k
}
409
410
Status PipelineFragmentContext::_build_pipeline_tasks_for_instance(
411
        int instance_idx,
412
1.26M
        const std::vector<std::shared_ptr<RuntimeProfile>>& pipeline_id_to_profile) {
413
1.26M
    const auto& local_params = _params.local_params[instance_idx];
414
1.26M
    auto fragment_instance_id = local_params.fragment_instance_id;
415
1.26M
    auto runtime_filter_mgr = std::make_unique<RuntimeFilterMgr>(false);
416
1.26M
    std::map<PipelineId, PipelineTask*> pipeline_id_to_task;
417
1.26M
    auto get_shared_state = [&](PipelinePtr pipeline)
418
1.26M
            -> std::map<int, std::pair<std::shared_ptr<BasicSharedState>,
419
2.07M
                                       std::vector<std::shared_ptr<Dependency>>>> {
420
2.07M
        std::map<int, std::pair<std::shared_ptr<BasicSharedState>,
421
2.07M
                                std::vector<std::shared_ptr<Dependency>>>>
422
2.07M
                shared_state_map;
423
2.66M
        for (auto& op : pipeline->operators()) {
424
2.66M
            auto source_id = op->operator_id();
425
2.66M
            if (auto iter = _op_id_to_shared_state.find(source_id);
426
2.66M
                iter != _op_id_to_shared_state.end()) {
427
808k
                shared_state_map.insert({source_id, iter->second});
428
808k
            }
429
2.66M
        }
430
2.07M
        for (auto sink_to_source_id : pipeline->sink()->dests_id()) {
431
2.07M
            if (auto iter = _op_id_to_shared_state.find(sink_to_source_id);
432
2.07M
                iter != _op_id_to_shared_state.end()) {
433
302k
                shared_state_map.insert({sink_to_source_id, iter->second});
434
302k
            }
435
2.07M
        }
436
2.07M
        return shared_state_map;
437
2.07M
    };
438
439
3.83M
    for (size_t pip_idx = 0; pip_idx < _pipelines.size(); pip_idx++) {
440
2.57M
        auto& pipeline = _pipelines[pip_idx];
441
2.57M
        if (pipeline->num_tasks() > 1 || instance_idx == 0) {
442
2.06M
            auto task_runtime_state = RuntimeState::create_unique(
443
2.06M
                    local_params.fragment_instance_id, _params.query_id, _params.fragment_id,
444
2.06M
                    _params.query_options, _query_ctx->query_globals, _exec_env, _query_ctx.get());
445
2.06M
            {
446
                // Initialize runtime state for this task
447
2.06M
                task_runtime_state->set_query_mem_tracker(_query_ctx->query_mem_tracker());
448
449
2.06M
                task_runtime_state->set_task_execution_context(shared_from_this());
450
2.06M
                task_runtime_state->set_be_number(local_params.backend_num);
451
452
2.06M
                if (_params.__isset.backend_id) {
453
2.06M
                    task_runtime_state->set_backend_id(_params.backend_id);
454
2.06M
                }
455
2.06M
                if (_params.__isset.import_label) {
456
242
                    task_runtime_state->set_import_label(_params.import_label);
457
242
                }
458
2.06M
                if (_params.__isset.db_name) {
459
194
                    task_runtime_state->set_db_name(_params.db_name);
460
194
                }
461
2.06M
                if (_params.__isset.load_job_id) {
462
0
                    task_runtime_state->set_load_job_id(_params.load_job_id);
463
0
                }
464
2.06M
                if (_params.__isset.wal_id) {
465
114
                    task_runtime_state->set_wal_id(_params.wal_id);
466
114
                }
467
2.06M
                if (_params.__isset.content_length) {
468
32
                    task_runtime_state->set_content_length(_params.content_length);
469
32
                }
470
471
2.06M
                task_runtime_state->set_desc_tbl(_desc_tbl);
472
2.06M
                task_runtime_state->set_per_fragment_instance_idx(local_params.sender_id);
473
2.06M
                task_runtime_state->set_num_per_fragment_instances(_params.num_senders);
474
2.06M
                task_runtime_state->resize_op_id_to_local_state(max_operator_id());
475
2.06M
                task_runtime_state->set_max_operator_id(max_operator_id());
476
2.06M
                task_runtime_state->set_load_stream_per_node(_params.load_stream_per_node);
477
2.06M
                task_runtime_state->set_total_load_streams(_params.total_load_streams);
478
2.06M
                task_runtime_state->set_num_local_sink(_params.num_local_sink);
479
480
2.06M
                task_runtime_state->set_runtime_filter_mgr(runtime_filter_mgr.get());
481
2.06M
            }
482
2.06M
            auto cur_task_id = _total_tasks++;
483
2.06M
            task_runtime_state->set_task_id(cur_task_id);
484
2.06M
            task_runtime_state->set_task_num(pipeline->num_tasks());
485
2.06M
            auto task = std::make_shared<PipelineTask>(
486
2.06M
                    pipeline, cur_task_id, task_runtime_state.get(),
487
2.06M
                    std::dynamic_pointer_cast<PipelineFragmentContext>(shared_from_this()),
488
2.06M
                    pipeline_id_to_profile[pip_idx].get(), get_shared_state(pipeline),
489
2.06M
                    instance_idx);
490
2.06M
            pipeline->incr_created_tasks(instance_idx, task.get());
491
2.06M
            pipeline_id_to_task.insert({pipeline->id(), task.get()});
492
2.06M
            _tasks[instance_idx].emplace_back(
493
2.06M
                    std::pair<std::shared_ptr<PipelineTask>, std::unique_ptr<RuntimeState>> {
494
2.06M
                            std::move(task), std::move(task_runtime_state)});
495
2.06M
        }
496
2.57M
    }
497
498
    /**
499
         * Build DAG for pipeline tasks.
500
         * For example, we have
501
         *
502
         *   ExchangeSink (Pipeline1)     JoinBuildSink (Pipeline2)
503
         *            \                      /
504
         *          JoinProbeOperator1 (Pipeline1)    JoinBuildSink (Pipeline3)
505
         *                 \                          /
506
         *               JoinProbeOperator2 (Pipeline1)
507
         *
508
         * In this fragment, we have three pipelines and pipeline 1 depends on pipeline 2 and pipeline 3.
509
         * To build this DAG, `_dag` manage dependencies between pipelines by pipeline ID and
510
         * `pipeline_id_to_task` is used to find the task by a unique pipeline ID.
511
         *
512
         * Finally, we have two upstream dependencies in Pipeline1 corresponding to JoinProbeOperator1
513
         * and JoinProbeOperator2.
514
         */
515
2.57M
    for (auto& _pipeline : _pipelines) {
516
2.57M
        if (pipeline_id_to_task.contains(_pipeline->id())) {
517
2.06M
            auto* task = pipeline_id_to_task[_pipeline->id()];
518
2.06M
            DCHECK(task != nullptr);
519
520
            // If this task has upstream dependency, then inject it into this task.
521
2.06M
            if (_dag.contains(_pipeline->id())) {
522
1.32M
                auto& deps = _dag[_pipeline->id()];
523
2.10M
                for (auto& dep : deps) {
524
2.10M
                    if (pipeline_id_to_task.contains(dep)) {
525
1.09M
                        auto ss = pipeline_id_to_task[dep]->get_sink_shared_state();
526
1.09M
                        if (ss) {
527
495k
                            task->inject_shared_state(ss);
528
598k
                        } else {
529
598k
                            pipeline_id_to_task[dep]->inject_shared_state(
530
598k
                                    task->get_source_shared_state());
531
598k
                        }
532
1.09M
                    }
533
2.10M
                }
534
1.32M
            }
535
2.06M
        }
536
2.57M
    }
537
3.83M
    for (size_t pip_idx = 0; pip_idx < _pipelines.size(); pip_idx++) {
538
2.56M
        if (pipeline_id_to_task.contains(_pipelines[pip_idx]->id())) {
539
2.06M
            auto* task = pipeline_id_to_task[_pipelines[pip_idx]->id()];
540
2.06M
            DCHECK(pipeline_id_to_profile[pip_idx]);
541
2.06M
            std::vector<TScanRangeParams> scan_ranges;
542
2.06M
            auto node_id = _pipelines[pip_idx]->operators().front()->node_id();
543
2.06M
            if (local_params.per_node_scan_ranges.contains(node_id)) {
544
331k
                scan_ranges = local_params.per_node_scan_ranges.find(node_id)->second;
545
331k
            }
546
2.06M
            RETURN_IF_ERROR_OR_CATCH_EXCEPTION(task->prepare(scan_ranges, local_params.sender_id,
547
2.06M
                                                             _params.fragment.output_sink));
548
2.06M
        }
549
2.56M
    }
550
1.26M
    {
551
1.26M
        std::lock_guard<std::mutex> l(_state_map_lock);
552
1.26M
        _runtime_filter_mgr_map[instance_idx] = std::move(runtime_filter_mgr);
553
1.26M
    }
554
1.26M
    return Status::OK();
555
1.26M
}
556
557
446k
Status PipelineFragmentContext::_build_pipeline_tasks(ThreadPool* thread_pool) {
558
446k
    _total_tasks = 0;
559
446k
    _closed_tasks = 0;
560
446k
    const auto target_size = _params.local_params.size();
561
446k
    _tasks.resize(target_size);
562
446k
    _runtime_filter_mgr_map.resize(target_size);
563
1.15M
    for (size_t pip_idx = 0; pip_idx < _pipelines.size(); pip_idx++) {
564
707k
        _pip_id_to_pipeline[_pipelines[pip_idx]->id()] = _pipelines[pip_idx].get();
565
707k
    }
566
446k
    auto pipeline_id_to_profile = _runtime_state->build_pipeline_profile(_pipelines.size());
567
568
446k
    if (target_size > 1 &&
569
446k
        (_runtime_state->query_options().__isset.parallel_prepare_threshold &&
570
150k
         target_size > _runtime_state->query_options().parallel_prepare_threshold)) {
571
        // If instances parallelism is big enough ( > parallel_prepare_threshold), we will prepare all tasks by multi-threads
572
39.4k
        std::vector<Status> prepare_status(target_size);
573
39.4k
        int submitted_tasks = 0;
574
39.4k
        Status submit_status;
575
39.4k
        CountDownLatch latch((int)target_size);
576
377k
        for (int i = 0; i < target_size; i++) {
577
337k
            submit_status = thread_pool->submit_func([&, i]() {
578
337k
                SCOPED_ATTACH_TASK(_query_ctx.get());
579
337k
                prepare_status[i] = _build_pipeline_tasks_for_instance(i, pipeline_id_to_profile);
580
337k
                latch.count_down();
581
337k
            });
582
337k
            if (LIKELY(submit_status.ok())) {
583
337k
                submitted_tasks++;
584
18.4E
            } else {
585
18.4E
                break;
586
18.4E
            }
587
337k
        }
588
39.4k
        latch.arrive_and_wait(target_size - submitted_tasks);
589
39.4k
        if (UNLIKELY(!submit_status.ok())) {
590
0
            return submit_status;
591
0
        }
592
377k
        for (int i = 0; i < submitted_tasks; i++) {
593
337k
            if (!prepare_status[i].ok()) {
594
0
                return prepare_status[i];
595
0
            }
596
337k
        }
597
406k
    } else {
598
1.33M
        for (int i = 0; i < target_size; i++) {
599
925k
            RETURN_IF_ERROR(_build_pipeline_tasks_for_instance(i, pipeline_id_to_profile));
600
925k
        }
601
406k
    }
602
446k
    _pipeline_parent_map.clear();
603
446k
    _op_id_to_shared_state.clear();
604
    // Record task cardinality once when this fragment context finishes task initialization.
605
446k
    _query_ctx->add_total_task_num(_total_tasks.load(std::memory_order_relaxed));
606
607
446k
    return Status::OK();
608
446k
}
609
610
445k
void PipelineFragmentContext::_init_next_report_time() {
611
445k
    auto interval_s = config::pipeline_status_report_interval;
612
445k
    if (_is_report_success && interval_s > 0 && _timeout > interval_s) {
613
42.0k
        VLOG_FILE << "enable period report: fragment id=" << _fragment_id;
614
42.0k
        uint64_t report_fragment_offset = (uint64_t)(rand() % interval_s) * NANOS_PER_SEC;
615
        // We don't want to wait longer than it takes to run the entire fragment.
616
42.0k
        _previous_report_time =
617
42.0k
                MonotonicNanos() + report_fragment_offset - (uint64_t)(interval_s)*NANOS_PER_SEC;
618
42.0k
        _disable_period_report = false;
619
42.0k
    }
620
445k
}
621
622
4.92k
void PipelineFragmentContext::refresh_next_report_time() {
623
4.92k
    auto disable = _disable_period_report.load(std::memory_order_acquire);
624
4.92k
    DCHECK(disable == true);
625
4.92k
    _previous_report_time.store(MonotonicNanos(), std::memory_order_release);
626
4.92k
    _disable_period_report.compare_exchange_strong(disable, false);
627
4.92k
}
628
629
7.54M
void PipelineFragmentContext::trigger_report_if_necessary() {
630
7.54M
    if (!_is_report_success) {
631
7.10M
        return;
632
7.10M
    }
633
439k
    auto disable = _disable_period_report.load(std::memory_order_acquire);
634
439k
    if (disable) {
635
7.79k
        return;
636
7.79k
    }
637
431k
    int32_t interval_s = config::pipeline_status_report_interval;
638
431k
    if (interval_s <= 0) {
639
0
        LOG(WARNING) << "config::status_report_interval is equal to or less than zero, do not "
640
0
                        "trigger "
641
0
                        "report.";
642
0
    }
643
431k
    uint64_t next_report_time = _previous_report_time.load(std::memory_order_acquire) +
644
431k
                                (uint64_t)(interval_s)*NANOS_PER_SEC;
645
431k
    if (MonotonicNanos() > next_report_time) {
646
4.93k
        if (!_disable_period_report.compare_exchange_strong(disable, true,
647
4.93k
                                                            std::memory_order_acq_rel)) {
648
10
            return;
649
10
        }
650
4.92k
        if (VLOG_FILE_IS_ON) {
651
0
            VLOG_FILE << "Reporting "
652
0
                      << "profile for query_id " << print_id(_query_id)
653
0
                      << ", fragment id: " << _fragment_id;
654
655
0
            std::stringstream ss;
656
0
            _runtime_state->runtime_profile()->compute_time_in_profile();
657
0
            _runtime_state->runtime_profile()->pretty_print(&ss);
658
0
            if (_runtime_state->load_channel_profile()) {
659
0
                _runtime_state->load_channel_profile()->pretty_print(&ss);
660
0
            }
661
662
0
            VLOG_FILE << "Query " << print_id(get_query_id()) << " fragment " << get_fragment_id()
663
0
                      << " profile:\n"
664
0
                      << ss.str();
665
0
        }
666
4.92k
        auto st = send_report(false);
667
4.92k
        if (!st.ok()) {
668
0
            disable = true;
669
0
            _disable_period_report.compare_exchange_strong(disable, false,
670
0
                                                           std::memory_order_acq_rel);
671
0
        }
672
4.92k
    }
673
431k
}
674
675
Status PipelineFragmentContext::_build_pipelines(ObjectPool* pool, const DescriptorTbl& descs,
676
444k
                                                 OperatorPtr* root, PipelinePtr cur_pipe) {
677
444k
    if (_params.fragment.plan.nodes.empty()) {
678
0
        throw Exception(ErrorCode::INTERNAL_ERROR, "Invalid plan which has no plan node!");
679
0
    }
680
681
444k
    int node_idx = 0;
682
683
444k
    RETURN_IF_ERROR(_create_tree_helper(pool, _params.fragment.plan.nodes, descs, nullptr,
684
444k
                                        &node_idx, root, cur_pipe, 0, false, false));
685
686
444k
    if (node_idx + 1 != _params.fragment.plan.nodes.size()) {
687
0
        return Status::InternalError(
688
0
                "Plan tree only partially reconstructed. Not all thrift nodes were used.");
689
0
    }
690
444k
    return Status::OK();
691
444k
}
692
693
Status PipelineFragmentContext::_create_tree_helper(
694
        ObjectPool* pool, const std::vector<TPlanNode>& tnodes, const DescriptorTbl& descs,
695
        OperatorPtr parent, int* node_idx, OperatorPtr* root, PipelinePtr& cur_pipe, int child_idx,
696
694k
        const bool followed_by_shuffled_operator, const bool require_bucket_distribution) {
697
    // propagate error case
698
694k
    if (*node_idx >= tnodes.size()) {
699
0
        return Status::InternalError(
700
0
                "Failed to reconstruct plan tree from thrift. Node id: {}, number of nodes: {}",
701
0
                *node_idx, tnodes.size());
702
0
    }
703
694k
    const TPlanNode& tnode = tnodes[*node_idx];
704
705
694k
    int num_children = tnodes[*node_idx].num_children;
706
694k
    bool current_followed_by_shuffled_operator = followed_by_shuffled_operator;
707
694k
    bool current_require_bucket_distribution = require_bucket_distribution;
708
    // TODO: Create CacheOperator is confused now
709
694k
    OperatorPtr op = nullptr;
710
694k
    OperatorPtr cache_op = nullptr;
711
694k
    RETURN_IF_ERROR(_create_operator(pool, tnodes[*node_idx], descs, op, cur_pipe,
712
694k
                                     parent == nullptr ? -1 : parent->node_id(), child_idx,
713
694k
                                     followed_by_shuffled_operator,
714
694k
                                     current_require_bucket_distribution, cache_op));
715
    // Initialization must be done here. For example, group by expressions in agg will be used to
716
    // decide if a local shuffle should be planed, so it must be initialized here.
717
694k
    RETURN_IF_ERROR(op->init(tnode, _runtime_state.get()));
718
    // assert(parent != nullptr || (node_idx == 0 && root_expr != nullptr));
719
694k
    if (parent != nullptr) {
720
        // add to parent's child(s)
721
248k
        RETURN_IF_ERROR(parent->set_child(cache_op ? cache_op : op));
722
445k
    } else {
723
445k
        *root = op;
724
445k
    }
725
    /**
726
     * `ExchangeType::HASH_SHUFFLE` should be used if an operator is followed by a shuffled operator (shuffled hash join, union operator followed by co-located operators).
727
     *
728
     * For plan:
729
     * LocalExchange(id=0) -> Aggregation(id=1) -> ShuffledHashJoin(id=2)
730
     *                           Exchange(id=3) -> ShuffledHashJoinBuild(id=2)
731
     * We must ensure data distribution of `LocalExchange(id=0)` is same as Exchange(id=3).
732
     *
733
     * If an operator's is followed by a local exchange without shuffle (e.g. passthrough), a
734
     * shuffled local exchanger will be used before join so it is not followed by shuffle join.
735
     */
736
694k
    auto required_data_distribution =
737
694k
            cur_pipe->operators().empty()
738
694k
                    ? cur_pipe->sink()->required_data_distribution(_runtime_state.get())
739
694k
                    : op->required_data_distribution(_runtime_state.get());
740
694k
    current_followed_by_shuffled_operator =
741
694k
            ((followed_by_shuffled_operator ||
742
694k
              (cur_pipe->operators().empty() ? cur_pipe->sink()->is_shuffled_operator()
743
634k
                                             : op->is_shuffled_operator())) &&
744
694k
             Pipeline::is_hash_exchange(required_data_distribution.distribution_type)) ||
745
694k
            (followed_by_shuffled_operator &&
746
581k
             required_data_distribution.distribution_type == ExchangeType::NOOP);
747
748
694k
    current_require_bucket_distribution =
749
694k
            ((require_bucket_distribution ||
750
694k
              (cur_pipe->operators().empty() ? cur_pipe->sink()->is_colocated_operator()
751
639k
                                             : op->is_colocated_operator())) &&
752
694k
             Pipeline::is_hash_exchange(required_data_distribution.distribution_type)) ||
753
694k
            (require_bucket_distribution &&
754
587k
             required_data_distribution.distribution_type == ExchangeType::NOOP);
755
756
694k
    if (num_children == 0) {
757
462k
        _use_serial_source = op->is_serial_operator();
758
462k
    }
759
    // rely on that tnodes is preorder of the plan
760
943k
    for (int i = 0; i < num_children; i++) {
761
249k
        ++*node_idx;
762
249k
        RETURN_IF_ERROR(_create_tree_helper(pool, tnodes, descs, op, node_idx, nullptr, cur_pipe, i,
763
249k
                                            current_followed_by_shuffled_operator,
764
249k
                                            current_require_bucket_distribution));
765
766
        // we are expecting a child, but have used all nodes
767
        // this means we have been given a bad tree and must fail
768
249k
        if (*node_idx >= tnodes.size()) {
769
0
            return Status::InternalError(
770
0
                    "Failed to reconstruct plan tree from thrift. Node id: {}, number of "
771
0
                    "nodes: {}",
772
0
                    *node_idx, tnodes.size());
773
0
        }
774
249k
    }
775
776
694k
    return Status::OK();
777
694k
}
778
779
void PipelineFragmentContext::_inherit_pipeline_properties(
780
        const DataDistribution& data_distribution, PipelinePtr pipe_with_source,
781
122k
        PipelinePtr pipe_with_sink) {
782
122k
    pipe_with_sink->set_num_tasks(pipe_with_source->num_tasks());
783
122k
    pipe_with_source->set_num_tasks(_num_instances);
784
122k
    pipe_with_source->set_data_distribution(data_distribution);
785
122k
}
786
787
Status PipelineFragmentContext::_add_local_exchange_impl(
788
        int idx, ObjectPool* pool, PipelinePtr cur_pipe, PipelinePtr new_pip,
789
        DataDistribution data_distribution, bool* do_local_exchange, int num_buckets,
790
        const std::map<int, int>& bucket_seq_to_instance_idx,
791
122k
        const std::map<int, int>& shuffle_idx_to_instance_idx) {
792
122k
    auto& operators = cur_pipe->operators();
793
122k
    const auto downstream_pipeline_id = cur_pipe->id();
794
122k
    auto local_exchange_id = next_operator_id();
795
    // 1. Create a new pipeline with local exchange sink.
796
122k
    DataSinkOperatorPtr sink;
797
122k
    auto sink_id = next_sink_operator_id();
798
799
    /**
800
     * `bucket_seq_to_instance_idx` is empty if no scan operator is contained in this fragment.
801
     * So co-located operators(e.g. Agg, Analytic) should use `HASH_SHUFFLE` instead of `BUCKET_HASH_SHUFFLE`.
802
     */
803
122k
    const bool followed_by_shuffled_operator =
804
122k
            operators.size() > idx ? operators[idx]->followed_by_shuffled_operator()
805
122k
                                   : cur_pipe->sink()->followed_by_shuffled_operator();
806
122k
    const bool use_global_hash_shuffle = bucket_seq_to_instance_idx.empty() &&
807
122k
                                         !shuffle_idx_to_instance_idx.contains(-1) &&
808
122k
                                         followed_by_shuffled_operator && !_use_serial_source;
809
122k
    sink = std::make_shared<LocalExchangeSinkOperatorX>(
810
122k
            sink_id, local_exchange_id, use_global_hash_shuffle ? _total_instances : _num_instances,
811
122k
            data_distribution.partition_exprs, bucket_seq_to_instance_idx);
812
122k
    if (bucket_seq_to_instance_idx.empty() &&
813
122k
        data_distribution.distribution_type == ExchangeType::BUCKET_HASH_SHUFFLE) {
814
7
        data_distribution.distribution_type = ExchangeType::HASH_SHUFFLE;
815
7
    }
816
122k
    RETURN_IF_ERROR(new_pip->set_sink(sink));
817
122k
    RETURN_IF_ERROR(new_pip->sink()->init(_runtime_state.get(), data_distribution.distribution_type,
818
122k
                                          num_buckets, use_global_hash_shuffle,
819
122k
                                          shuffle_idx_to_instance_idx));
820
821
    // 2. Create and initialize LocalExchangeSharedState.
822
122k
    std::shared_ptr<LocalExchangeSharedState> shared_state =
823
122k
            LocalExchangeSharedState::create_shared(_num_instances);
824
122k
    switch (data_distribution.distribution_type) {
825
17.5k
    case ExchangeType::HASH_SHUFFLE:
826
17.5k
        shared_state->exchanger = ShuffleExchanger::create_unique(
827
17.5k
                std::max(cur_pipe->num_tasks(), _num_instances), _num_instances,
828
17.5k
                use_global_hash_shuffle ? _total_instances : _num_instances,
829
17.5k
                _runtime_state->query_options().__isset.local_exchange_free_blocks_limit
830
17.5k
                        ? cast_set<int>(
831
17.5k
                                  _runtime_state->query_options().local_exchange_free_blocks_limit)
832
17.5k
                        : 0);
833
17.5k
        break;
834
526
    case ExchangeType::BUCKET_HASH_SHUFFLE:
835
526
        shared_state->exchanger = BucketShuffleExchanger::create_unique(
836
526
                std::max(cur_pipe->num_tasks(), _num_instances), _num_instances, num_buckets,
837
526
                _runtime_state->query_options().__isset.local_exchange_free_blocks_limit
838
526
                        ? cast_set<int>(
839
526
                                  _runtime_state->query_options().local_exchange_free_blocks_limit)
840
526
                        : 0);
841
526
        break;
842
100k
    case ExchangeType::PASSTHROUGH:
843
100k
        shared_state->exchanger = PassthroughExchanger::create_unique(
844
100k
                cur_pipe->num_tasks(), _num_instances,
845
100k
                _runtime_state->query_options().__isset.local_exchange_free_blocks_limit
846
100k
                        ? cast_set<int>(
847
100k
                                  _runtime_state->query_options().local_exchange_free_blocks_limit)
848
100k
                        : 0);
849
100k
        break;
850
313
    case ExchangeType::BROADCAST:
851
313
        shared_state->exchanger = BroadcastExchanger::create_unique(
852
313
                cur_pipe->num_tasks(), _num_instances,
853
313
                _runtime_state->query_options().__isset.local_exchange_free_blocks_limit
854
313
                        ? cast_set<int>(
855
313
                                  _runtime_state->query_options().local_exchange_free_blocks_limit)
856
313
                        : 0);
857
313
        break;
858
2.39k
    case ExchangeType::PASS_TO_ONE:
859
2.39k
        if (_runtime_state->enable_share_hash_table_for_broadcast_join()) {
860
            // If shared hash table is enabled for BJ, hash table will be built by only one task
861
1.47k
            shared_state->exchanger = PassToOneExchanger::create_unique(
862
1.47k
                    cur_pipe->num_tasks(), _num_instances,
863
1.47k
                    _runtime_state->query_options().__isset.local_exchange_free_blocks_limit
864
1.47k
                            ? cast_set<int>(_runtime_state->query_options()
865
1.47k
                                                    .local_exchange_free_blocks_limit)
866
1.47k
                            : 0);
867
1.47k
        } else {
868
914
            shared_state->exchanger = BroadcastExchanger::create_unique(
869
914
                    cur_pipe->num_tasks(), _num_instances,
870
914
                    _runtime_state->query_options().__isset.local_exchange_free_blocks_limit
871
914
                            ? cast_set<int>(_runtime_state->query_options()
872
914
                                                    .local_exchange_free_blocks_limit)
873
914
                            : 0);
874
914
        }
875
2.39k
        break;
876
898
    case ExchangeType::ADAPTIVE_PASSTHROUGH:
877
898
        shared_state->exchanger = AdaptivePassthroughExchanger::create_unique(
878
898
                std::max(cur_pipe->num_tasks(), _num_instances), _num_instances,
879
898
                _runtime_state->query_options().__isset.local_exchange_free_blocks_limit
880
898
                        ? cast_set<int>(
881
898
                                  _runtime_state->query_options().local_exchange_free_blocks_limit)
882
898
                        : 0);
883
898
        break;
884
0
    default:
885
0
        return Status::InternalError("Unsupported local exchange type : " +
886
0
                                     std::to_string((int)data_distribution.distribution_type));
887
122k
    }
888
122k
    shared_state->create_source_dependencies(_num_instances, local_exchange_id, local_exchange_id,
889
122k
                                             "LOCAL_EXCHANGE_OPERATOR");
890
122k
    shared_state->create_sink_dependency(sink_id, local_exchange_id, "LOCAL_EXCHANGE_SINK");
891
122k
    _op_id_to_shared_state.insert({local_exchange_id, {shared_state, shared_state->sink_deps}});
892
893
    // 3. Set two pipelines' operator list. For example, split pipeline [Scan - AggSink] to
894
    // pipeline1 [Scan - LocalExchangeSink] and pipeline2 [LocalExchangeSource - AggSink].
895
896
    // 3.1 Initialize new pipeline's operator list.
897
122k
    std::copy(operators.begin(), operators.begin() + idx,
898
122k
              std::inserter(new_pip->operators(), new_pip->operators().end()));
899
900
    // 3.2 Erase unused operators in previous pipeline.
901
122k
    operators.erase(operators.begin(), operators.begin() + idx);
902
903
    // 4. Initialize LocalExchangeSource and insert it into this pipeline.
904
122k
    OperatorPtr source_op;
905
122k
    source_op = std::make_shared<LocalExchangeSourceOperatorX>(pool, local_exchange_id);
906
122k
    RETURN_IF_ERROR(source_op->set_child(new_pip->operators().back()));
907
122k
    RETURN_IF_ERROR(source_op->init(data_distribution.distribution_type));
908
122k
    if (!operators.empty()) {
909
48.2k
        RETURN_IF_ERROR(operators.front()->set_child(nullptr));
910
48.2k
        RETURN_IF_ERROR(operators.front()->set_child(source_op));
911
48.2k
    }
912
122k
    operators.insert(operators.begin(), source_op);
913
914
    // 5. Set children for two pipelines separately.
915
122k
    std::vector<std::shared_ptr<Pipeline>> new_children;
916
122k
    std::vector<PipelineId> edges_with_source;
917
140k
    for (auto child : cur_pipe->children()) {
918
140k
        bool found = false;
919
154k
        for (auto op : new_pip->operators()) {
920
154k
            if (child->sink()->node_id() == op->node_id()) {
921
12.6k
                new_pip->set_children(child);
922
12.6k
                found = true;
923
12.6k
            };
924
154k
        }
925
140k
        if (!found) {
926
127k
            new_children.push_back(child);
927
127k
            edges_with_source.push_back(child->id());
928
127k
        }
929
140k
    }
930
122k
    new_children.push_back(new_pip);
931
122k
    edges_with_source.push_back(new_pip->id());
932
933
    // 6. Set DAG for new pipelines.
934
122k
    if (!new_pip->children().empty()) {
935
7.33k
        std::vector<PipelineId> edges_with_sink;
936
12.6k
        for (auto child : new_pip->children()) {
937
12.6k
            edges_with_sink.push_back(child->id());
938
12.6k
        }
939
7.33k
        _dag.insert({new_pip->id(), edges_with_sink});
940
7.33k
    }
941
122k
    cur_pipe->set_children(new_children);
942
122k
    _dag[downstream_pipeline_id] = edges_with_source;
943
122k
    RETURN_IF_ERROR(new_pip->sink()->set_child(new_pip->operators().back()));
944
122k
    RETURN_IF_ERROR(cur_pipe->sink()->set_child(nullptr));
945
122k
    RETURN_IF_ERROR(cur_pipe->sink()->set_child(cur_pipe->operators().back()));
946
947
    // 7. Inherit properties from current pipeline.
948
122k
    _inherit_pipeline_properties(data_distribution, cur_pipe, new_pip);
949
122k
    return Status::OK();
950
122k
}
951
952
Status PipelineFragmentContext::_add_local_exchange(
953
        int pip_idx, int idx, int node_id, ObjectPool* pool, PipelinePtr cur_pipe,
954
        DataDistribution data_distribution, bool* do_local_exchange, int num_buckets,
955
        const std::map<int, int>& bucket_seq_to_instance_idx,
956
196k
        const std::map<int, int>& shuffle_idx_to_instance_idx) {
957
196k
    if (_num_instances <= 1 || cur_pipe->num_tasks_of_parent() <= 1) {
958
41.9k
        return Status::OK();
959
41.9k
    }
960
961
154k
    if (!cur_pipe->need_to_local_exchange(data_distribution, idx)) {
962
50.2k
        return Status::OK();
963
50.2k
    }
964
104k
    *do_local_exchange = true;
965
966
104k
    auto& operators = cur_pipe->operators();
967
104k
    auto total_op_num = operators.size();
968
104k
    auto new_pip = add_pipeline(cur_pipe, pip_idx + 1);
969
104k
    RETURN_IF_ERROR(_add_local_exchange_impl(
970
104k
            idx, pool, cur_pipe, new_pip, data_distribution, do_local_exchange, num_buckets,
971
104k
            bucket_seq_to_instance_idx, shuffle_idx_to_instance_idx));
972
973
104k
    CHECK(total_op_num + 1 == cur_pipe->operators().size() + new_pip->operators().size())
974
409
            << "total_op_num: " << total_op_num
975
409
            << " cur_pipe->operators().size(): " << cur_pipe->operators().size()
976
409
            << " new_pip->operators().size(): " << new_pip->operators().size();
977
978
    // There are some local shuffles with relatively heavy operations on the sink.
979
    // If the local sink concurrency is 1 and the local source concurrency is n, the sink becomes a bottleneck.
980
    // Therefore, local passthrough is used to increase the concurrency of the sink.
981
    // op -> local sink(1) -> local source (n)
982
    // op -> local passthrough(1) -> local passthrough(n) ->  local sink(n) -> local source (n)
983
104k
    if (cur_pipe->num_tasks() > 1 && new_pip->num_tasks() == 1 &&
984
104k
        Pipeline::heavy_operations_on_the_sink(data_distribution.distribution_type)) {
985
17.8k
        RETURN_IF_ERROR(_add_local_exchange_impl(
986
17.8k
                cast_set<int>(new_pip->operators().size()), pool, new_pip,
987
17.8k
                add_pipeline(new_pip, pip_idx + 2), DataDistribution(ExchangeType::PASSTHROUGH),
988
17.8k
                do_local_exchange, num_buckets, bucket_seq_to_instance_idx,
989
17.8k
                shuffle_idx_to_instance_idx));
990
17.8k
    }
991
104k
    return Status::OK();
992
104k
}
993
994
Status PipelineFragmentContext::_plan_local_exchange(
995
        int num_buckets, const std::map<int, int>& bucket_seq_to_instance_idx,
996
442k
        const std::map<int, int>& shuffle_idx_to_instance_idx) {
997
1.02M
    for (int pip_idx = cast_set<int>(_pipelines.size()) - 1; pip_idx >= 0; pip_idx--) {
998
582k
        _pipelines[pip_idx]->init_data_distribution(_runtime_state.get());
999
        // Set property if child pipeline is not join operator's child.
1000
582k
        if (!_pipelines[pip_idx]->children().empty()) {
1001
133k
            for (auto& child : _pipelines[pip_idx]->children()) {
1002
133k
                if (child->sink()->node_id() ==
1003
133k
                    _pipelines[pip_idx]->operators().front()->node_id()) {
1004
118k
                    _pipelines[pip_idx]->set_data_distribution(child->data_distribution());
1005
118k
                }
1006
133k
            }
1007
127k
        }
1008
1009
        // if 'num_buckets == 0' means the fragment is colocated by exchange node not the
1010
        // scan node. so here use `_num_instance` to replace the `num_buckets` to prevent dividing 0
1011
        // still keep colocate plan after local shuffle
1012
582k
        RETURN_IF_ERROR(_plan_local_exchange(num_buckets, pip_idx, _pipelines[pip_idx],
1013
582k
                                             bucket_seq_to_instance_idx,
1014
582k
                                             shuffle_idx_to_instance_idx));
1015
582k
    }
1016
442k
    return Status::OK();
1017
442k
}
1018
1019
Status PipelineFragmentContext::_plan_local_exchange(
1020
        int num_buckets, int pip_idx, PipelinePtr pip,
1021
        const std::map<int, int>& bucket_seq_to_instance_idx,
1022
581k
        const std::map<int, int>& shuffle_idx_to_instance_idx) {
1023
581k
    int idx = 1;
1024
581k
    bool do_local_exchange = false;
1025
629k
    do {
1026
629k
        auto& ops = pip->operators();
1027
629k
        do_local_exchange = false;
1028
        // Plan local exchange for each operator.
1029
697k
        for (; idx < ops.size();) {
1030
116k
            if (ops[idx]->required_data_distribution(_runtime_state.get()).need_local_exchange()) {
1031
107k
                RETURN_IF_ERROR(_add_local_exchange(
1032
107k
                        pip_idx, idx, ops[idx]->node_id(), _runtime_state->obj_pool(), pip,
1033
107k
                        ops[idx]->required_data_distribution(_runtime_state.get()),
1034
107k
                        &do_local_exchange, num_buckets, bucket_seq_to_instance_idx,
1035
107k
                        shuffle_idx_to_instance_idx));
1036
107k
            }
1037
116k
            if (do_local_exchange) {
1038
                // If local exchange is needed for current operator, we will split this pipeline to
1039
                // two pipelines by local exchange sink/source. And then we need to process remaining
1040
                // operators in this pipeline so we set idx to 2 (0 is local exchange source and 1
1041
                // is current operator was already processed) and continue to plan local exchange.
1042
48.4k
                idx = 2;
1043
48.4k
                break;
1044
48.4k
            }
1045
67.8k
            idx++;
1046
67.8k
        }
1047
629k
    } while (do_local_exchange);
1048
581k
    if (pip->sink()->required_data_distribution(_runtime_state.get()).need_local_exchange()) {
1049
89.5k
        RETURN_IF_ERROR(_add_local_exchange(
1050
89.5k
                pip_idx, idx, pip->sink()->node_id(), _runtime_state->obj_pool(), pip,
1051
89.5k
                pip->sink()->required_data_distribution(_runtime_state.get()), &do_local_exchange,
1052
89.5k
                num_buckets, bucket_seq_to_instance_idx, shuffle_idx_to_instance_idx));
1053
89.5k
    }
1054
581k
    return Status::OK();
1055
581k
}
1056
1057
Status PipelineFragmentContext::_create_data_sink(ObjectPool* pool, const TDataSink& thrift_sink,
1058
                                                  const std::vector<TExpr>& output_exprs,
1059
                                                  const TPipelineFragmentParams& params,
1060
                                                  const RowDescriptor& row_desc,
1061
                                                  RuntimeState* state, DescriptorTbl& desc_tbl,
1062
447k
                                                  PipelineId cur_pipeline_id) {
1063
447k
    switch (thrift_sink.type) {
1064
150k
    case TDataSinkType::DATA_STREAM_SINK: {
1065
150k
        if (!thrift_sink.__isset.stream_sink) {
1066
0
            return Status::InternalError("Missing data stream sink.");
1067
0
        }
1068
150k
        _sink = std::make_shared<ExchangeSinkOperatorX>(
1069
150k
                state, row_desc, next_sink_operator_id(), thrift_sink.stream_sink,
1070
150k
                params.destinations, _fragment_instance_ids);
1071
150k
        break;
1072
150k
    }
1073
256k
    case TDataSinkType::RESULT_SINK: {
1074
256k
        if (!thrift_sink.__isset.result_sink) {
1075
0
            return Status::InternalError("Missing data buffer sink.");
1076
0
        }
1077
1078
256k
        auto& pipeline = _pipelines[cur_pipeline_id];
1079
256k
        int child_node_id = pipeline->operators().back()->node_id();
1080
256k
        _sink = std::make_shared<ResultSinkOperatorX>(next_sink_operator_id(), child_node_id + 1,
1081
256k
                                                      row_desc, output_exprs,
1082
256k
                                                      thrift_sink.result_sink);
1083
256k
        break;
1084
256k
    }
1085
105
    case TDataSinkType::DICTIONARY_SINK: {
1086
105
        if (!thrift_sink.__isset.dictionary_sink) {
1087
0
            return Status::InternalError("Missing dict sink.");
1088
0
        }
1089
1090
105
        _sink = std::make_shared<DictSinkOperatorX>(next_sink_operator_id(), row_desc, output_exprs,
1091
105
                                                    thrift_sink.dictionary_sink);
1092
105
        break;
1093
105
    }
1094
0
    case TDataSinkType::GROUP_COMMIT_OLAP_TABLE_SINK:
1095
33.5k
    case TDataSinkType::OLAP_TABLE_SINK: {
1096
33.5k
        auto& pipeline = _pipelines[cur_pipeline_id];
1097
33.5k
        int child_node_id = pipeline->operators().back()->node_id();
1098
33.5k
        if (state->query_options().enable_memtable_on_sink_node &&
1099
33.5k
            !_has_inverted_index_v1_or_partial_update(thrift_sink.olap_table_sink) &&
1100
33.5k
            !_has_row_binlog(thrift_sink.olap_table_sink) && !config::is_cloud_mode()) {
1101
2.73k
            _sink = std::make_shared<OlapTableSinkV2OperatorX>(
1102
2.73k
                    pool, next_sink_operator_id(), child_node_id + 1, row_desc, output_exprs);
1103
30.8k
        } else {
1104
30.8k
            _sink = std::make_shared<OlapTableSinkOperatorX>(
1105
30.8k
                    pool, next_sink_operator_id(), child_node_id + 1, row_desc, output_exprs);
1106
30.8k
        }
1107
33.5k
        break;
1108
0
    }
1109
166
    case TDataSinkType::GROUP_COMMIT_BLOCK_SINK: {
1110
166
        DCHECK(thrift_sink.__isset.olap_table_sink);
1111
166
        DCHECK(state->get_query_ctx() != nullptr);
1112
166
        state->get_query_ctx()->query_mem_tracker()->is_group_commit_load = true;
1113
166
        _sink = std::make_shared<GroupCommitBlockSinkOperatorX>(next_sink_operator_id(), row_desc,
1114
166
                                                                output_exprs);
1115
166
        break;
1116
0
    }
1117
1.46k
    case TDataSinkType::HIVE_TABLE_SINK: {
1118
1.46k
        if (!thrift_sink.__isset.hive_table_sink) {
1119
0
            return Status::InternalError("Missing hive table sink.");
1120
0
        }
1121
1.46k
        _sink = std::make_shared<HiveTableSinkOperatorX>(pool, next_sink_operator_id(), row_desc,
1122
1.46k
                                                         output_exprs);
1123
1.46k
        break;
1124
1.46k
    }
1125
1.73k
    case TDataSinkType::ICEBERG_TABLE_SINK: {
1126
1.73k
        if (!thrift_sink.__isset.iceberg_table_sink) {
1127
0
            return Status::InternalError("Missing iceberg table sink.");
1128
0
        }
1129
1.73k
        if (thrift_sink.iceberg_table_sink.__isset.sort_info) {
1130
0
            _sink = std::make_shared<SpillIcebergTableSinkOperatorX>(pool, next_sink_operator_id(),
1131
0
                                                                     row_desc, output_exprs);
1132
1.73k
        } else {
1133
1.73k
            _sink = std::make_shared<IcebergTableSinkOperatorX>(pool, next_sink_operator_id(),
1134
1.73k
                                                                row_desc, output_exprs);
1135
1.73k
        }
1136
1.73k
        break;
1137
1.73k
    }
1138
20
    case TDataSinkType::ICEBERG_DELETE_SINK: {
1139
20
        if (!thrift_sink.__isset.iceberg_delete_sink) {
1140
0
            return Status::InternalError("Missing iceberg delete sink.");
1141
0
        }
1142
20
        _sink = std::make_shared<IcebergDeleteSinkOperatorX>(pool, next_sink_operator_id(),
1143
20
                                                             row_desc, output_exprs);
1144
20
        break;
1145
20
    }
1146
80
    case TDataSinkType::ICEBERG_MERGE_SINK: {
1147
80
        if (!thrift_sink.__isset.iceberg_merge_sink) {
1148
0
            return Status::InternalError("Missing iceberg merge sink.");
1149
0
        }
1150
80
        _sink = std::make_shared<IcebergMergeSinkOperatorX>(pool, next_sink_operator_id(), row_desc,
1151
80
                                                            output_exprs);
1152
80
        break;
1153
80
    }
1154
0
    case TDataSinkType::MAXCOMPUTE_TABLE_SINK: {
1155
0
        if (!thrift_sink.__isset.max_compute_table_sink) {
1156
0
            return Status::InternalError("Missing max compute table sink.");
1157
0
        }
1158
0
        _sink = std::make_shared<MCTableSinkOperatorX>(pool, next_sink_operator_id(), row_desc,
1159
0
                                                       output_exprs);
1160
0
        break;
1161
0
    }
1162
88
    case TDataSinkType::JDBC_TABLE_SINK: {
1163
88
        if (!thrift_sink.__isset.jdbc_table_sink) {
1164
0
            return Status::InternalError("Missing data jdbc sink.");
1165
0
        }
1166
88
        if (config::enable_java_support) {
1167
88
            _sink = std::make_shared<JdbcTableSinkOperatorX>(row_desc, next_sink_operator_id(),
1168
88
                                                             output_exprs);
1169
88
        } else {
1170
0
            return Status::InternalError(
1171
0
                    "Jdbc table sink is not enabled, you can change be config "
1172
0
                    "enable_java_support to true and restart be.");
1173
0
        }
1174
88
        break;
1175
88
    }
1176
88
    case TDataSinkType::MEMORY_SCRATCH_SINK: {
1177
3
        if (!thrift_sink.__isset.memory_scratch_sink) {
1178
0
            return Status::InternalError("Missing data buffer sink.");
1179
0
        }
1180
1181
3
        _sink = std::make_shared<MemoryScratchSinkOperatorX>(row_desc, next_sink_operator_id(),
1182
3
                                                             output_exprs);
1183
3
        break;
1184
3
    }
1185
502
    case TDataSinkType::RESULT_FILE_SINK: {
1186
502
        if (!thrift_sink.__isset.result_file_sink) {
1187
0
            return Status::InternalError("Missing result file sink.");
1188
0
        }
1189
1190
        // Result file sink is not the top sink
1191
502
        if (params.__isset.destinations && !params.destinations.empty()) {
1192
0
            _sink = std::make_shared<ResultFileSinkOperatorX>(
1193
0
                    next_sink_operator_id(), row_desc, thrift_sink.result_file_sink,
1194
0
                    params.destinations, output_exprs, desc_tbl);
1195
502
        } else {
1196
502
            _sink = std::make_shared<ResultFileSinkOperatorX>(next_sink_operator_id(), row_desc,
1197
502
                                                              output_exprs);
1198
502
        }
1199
502
        break;
1200
502
    }
1201
2.19k
    case TDataSinkType::MULTI_CAST_DATA_STREAM_SINK: {
1202
2.19k
        DCHECK(thrift_sink.__isset.multi_cast_stream_sink);
1203
2.19k
        DCHECK_GT(thrift_sink.multi_cast_stream_sink.sinks.size(), 0);
1204
2.19k
        auto sink_id = next_sink_operator_id();
1205
2.19k
        const int multi_cast_node_id = sink_id;
1206
2.19k
        auto sender_size = thrift_sink.multi_cast_stream_sink.sinks.size();
1207
        // one sink has multiple sources.
1208
2.19k
        std::vector<int> sources;
1209
8.57k
        for (int i = 0; i < sender_size; ++i) {
1210
6.38k
            auto source_id = next_operator_id();
1211
6.38k
            sources.push_back(source_id);
1212
6.38k
        }
1213
1214
2.19k
        _sink = std::make_shared<MultiCastDataStreamSinkOperatorX>(
1215
2.19k
                sink_id, multi_cast_node_id, sources, pool, thrift_sink.multi_cast_stream_sink);
1216
8.57k
        for (int i = 0; i < sender_size; ++i) {
1217
6.38k
            auto new_pipeline = add_pipeline();
1218
            // use to exchange sink
1219
6.38k
            RowDescriptor* exchange_row_desc = nullptr;
1220
6.38k
            {
1221
6.38k
                const auto& tmp_row_desc =
1222
6.38k
                        !thrift_sink.multi_cast_stream_sink.sinks[i].output_exprs.empty()
1223
6.38k
                                ? RowDescriptor(state->desc_tbl(),
1224
6.38k
                                                {thrift_sink.multi_cast_stream_sink.sinks[i]
1225
6.38k
                                                         .output_tuple_id})
1226
6.38k
                                : row_desc;
1227
6.38k
                exchange_row_desc = pool->add(new RowDescriptor(tmp_row_desc));
1228
6.38k
            }
1229
6.38k
            auto source_id = sources[i];
1230
6.38k
            OperatorPtr source_op;
1231
            // 1. create and set the source operator of multi_cast_data_stream_source for new pipeline
1232
6.38k
            source_op = std::make_shared<MultiCastDataStreamerSourceOperatorX>(
1233
6.38k
                    /*node_id*/ source_id, /*consumer_id*/ i, pool,
1234
6.38k
                    thrift_sink.multi_cast_stream_sink.sinks[i], row_desc,
1235
6.38k
                    /*operator_id=*/source_id);
1236
6.38k
            RETURN_IF_ERROR(new_pipeline->add_operator(
1237
6.38k
                    source_op, params.__isset.parallel_instances ? params.parallel_instances : 0));
1238
            // 2. create and set sink operator of data stream sender for new pipeline
1239
1240
6.38k
            DataSinkOperatorPtr sink_op;
1241
6.38k
            sink_op = std::make_shared<ExchangeSinkOperatorX>(
1242
6.38k
                    state, *exchange_row_desc, next_sink_operator_id(),
1243
6.38k
                    thrift_sink.multi_cast_stream_sink.sinks[i],
1244
6.38k
                    thrift_sink.multi_cast_stream_sink.destinations[i], _fragment_instance_ids);
1245
1246
6.38k
            RETURN_IF_ERROR(new_pipeline->set_sink(sink_op));
1247
6.38k
            {
1248
6.38k
                TDataSink* t = pool->add(new TDataSink());
1249
6.38k
                t->stream_sink = thrift_sink.multi_cast_stream_sink.sinks[i];
1250
6.38k
                RETURN_IF_ERROR(sink_op->init(*t));
1251
6.38k
            }
1252
1253
            // 3. set dependency dag
1254
6.38k
            _dag[new_pipeline->id()].push_back(cur_pipeline_id);
1255
6.38k
        }
1256
2.19k
        if (sources.empty()) {
1257
0
            return Status::InternalError("size of sources must be greater than 0");
1258
0
        }
1259
2.19k
        break;
1260
2.19k
    }
1261
2.19k
    case TDataSinkType::BLACKHOLE_SINK: {
1262
13
        if (!thrift_sink.__isset.blackhole_sink) {
1263
0
            return Status::InternalError("Missing blackhole sink.");
1264
0
        }
1265
1266
13
        _sink.reset(new BlackholeSinkOperatorX(next_sink_operator_id()));
1267
13
        break;
1268
13
    }
1269
156
    case TDataSinkType::TVF_TABLE_SINK: {
1270
156
        if (!thrift_sink.__isset.tvf_table_sink) {
1271
0
            return Status::InternalError("Missing TVF table sink.");
1272
0
        }
1273
156
        _sink = std::make_shared<TVFTableSinkOperatorX>(pool, next_sink_operator_id(), row_desc,
1274
156
                                                        output_exprs);
1275
156
        break;
1276
156
    }
1277
0
    default:
1278
0
        return Status::InternalError("Unsuported sink type in pipeline: {}", thrift_sink.type);
1279
447k
    }
1280
446k
    return Status::OK();
1281
447k
}
1282
1283
// NOLINTBEGIN(readability-function-size)
1284
// NOLINTBEGIN(readability-function-cognitive-complexity)
1285
Status PipelineFragmentContext::_create_operator(ObjectPool* pool, const TPlanNode& tnode,
1286
                                                 const DescriptorTbl& descs, OperatorPtr& op,
1287
                                                 PipelinePtr& cur_pipe, int parent_idx,
1288
                                                 int child_idx,
1289
                                                 const bool followed_by_shuffled_operator,
1290
                                                 const bool require_bucket_distribution,
1291
696k
                                                 OperatorPtr& cache_op) {
1292
696k
    std::vector<DataSinkOperatorPtr> sink_ops;
1293
696k
    Defer defer = Defer([&]() {
1294
695k
        if (op) {
1295
695k
            op->update_operator(tnode, followed_by_shuffled_operator, require_bucket_distribution);
1296
695k
        }
1297
695k
        for (auto& s : sink_ops) {
1298
132k
            s->update_operator(tnode, followed_by_shuffled_operator, require_bucket_distribution);
1299
132k
        }
1300
695k
    });
1301
    // We directly construct the operator from Thrift because the given array is in the order of preorder traversal.
1302
    // Therefore, here we need to use a stack-like structure.
1303
696k
    _pipeline_parent_map.pop(cur_pipe, parent_idx, child_idx);
1304
696k
    std::stringstream error_msg;
1305
696k
    bool enable_query_cache = _params.fragment.__isset.query_cache_param;
1306
1307
696k
    bool fe_with_old_version = false;
1308
696k
    switch (tnode.node_type) {
1309
216k
    case TPlanNodeType::OLAP_SCAN_NODE: {
1310
216k
        op = std::make_shared<OlapScanOperatorX>(
1311
216k
                pool, tnode, next_operator_id(), descs, _num_instances,
1312
216k
                enable_query_cache ? _params.fragment.query_cache_param : TQueryCacheParam {});
1313
216k
        RETURN_IF_ERROR(cur_pipe->add_operator(op, _parallel_instances));
1314
216k
        fe_with_old_version = !tnode.__isset.is_serial_operator;
1315
216k
        break;
1316
216k
    }
1317
80
    case TPlanNodeType::GROUP_COMMIT_SCAN_NODE: {
1318
80
        DCHECK(_query_ctx != nullptr);
1319
80
        _query_ctx->query_mem_tracker()->is_group_commit_load = true;
1320
80
        op = std::make_shared<GroupCommitOperatorX>(pool, tnode, next_operator_id(), descs,
1321
80
                                                    _num_instances);
1322
80
        RETURN_IF_ERROR(cur_pipe->add_operator(op, _parallel_instances));
1323
80
        fe_with_old_version = !tnode.__isset.is_serial_operator;
1324
80
        break;
1325
80
    }
1326
0
    case TPlanNodeType::JDBC_SCAN_NODE: {
1327
0
        if (config::enable_java_support) {
1328
0
            op = std::make_shared<JDBCScanOperatorX>(pool, tnode, next_operator_id(), descs,
1329
0
                                                     _num_instances);
1330
0
            RETURN_IF_ERROR(cur_pipe->add_operator(op, _parallel_instances));
1331
0
        } else {
1332
0
            return Status::InternalError(
1333
0
                    "Jdbc scan node is disabled, you can change be config enable_java_support "
1334
0
                    "to true and restart be.");
1335
0
        }
1336
0
        fe_with_old_version = !tnode.__isset.is_serial_operator;
1337
0
        break;
1338
0
    }
1339
25.2k
    case TPlanNodeType::FILE_SCAN_NODE: {
1340
25.2k
        op = std::make_shared<FileScanOperatorX>(pool, tnode, next_operator_id(), descs,
1341
25.2k
                                                 _num_instances);
1342
25.2k
        RETURN_IF_ERROR(cur_pipe->add_operator(op, _parallel_instances));
1343
25.2k
        fe_with_old_version = !tnode.__isset.is_serial_operator;
1344
25.2k
        break;
1345
25.2k
    }
1346
154k
    case TPlanNodeType::EXCHANGE_NODE: {
1347
154k
        int num_senders = _params.per_exch_num_senders.contains(tnode.node_id)
1348
154k
                                  ? _params.per_exch_num_senders.find(tnode.node_id)->second
1349
18.4E
                                  : 0;
1350
154k
        DCHECK_GT(num_senders, 0);
1351
154k
        op = std::make_shared<ExchangeSourceOperatorX>(pool, tnode, next_operator_id(), descs,
1352
154k
                                                       num_senders);
1353
154k
        RETURN_IF_ERROR(cur_pipe->add_operator(op, _parallel_instances));
1354
154k
        fe_with_old_version = !tnode.__isset.is_serial_operator;
1355
154k
        break;
1356
154k
    }
1357
165k
    case TPlanNodeType::AGGREGATION_NODE: {
1358
165k
        if (tnode.agg_node.grouping_exprs.empty() &&
1359
165k
            descs.get_tuple_descriptor(tnode.agg_node.output_tuple_id)->slots().empty()) {
1360
0
            return Status::InternalError("Illegal aggregate node " + std::to_string(tnode.node_id) +
1361
0
                                         ": group by and output is empty");
1362
0
        }
1363
165k
        bool need_create_cache_op =
1364
165k
                enable_query_cache && tnode.node_id == _params.fragment.query_cache_param.node_id;
1365
165k
        auto create_query_cache_operator = [&](PipelinePtr& new_pipe) {
1366
10
            auto cache_node_id = _params.local_params[0].per_node_scan_ranges.begin()->first;
1367
10
            auto cache_source_id = next_operator_id();
1368
10
            op = std::make_shared<CacheSourceOperatorX>(pool, cache_node_id, cache_source_id,
1369
10
                                                        _params.fragment.query_cache_param);
1370
10
            RETURN_IF_ERROR(cur_pipe->add_operator(op, _parallel_instances));
1371
1372
10
            const auto downstream_pipeline_id = cur_pipe->id();
1373
10
            if (!_dag.contains(downstream_pipeline_id)) {
1374
10
                _dag.insert({downstream_pipeline_id, {}});
1375
10
            }
1376
10
            new_pipe = add_pipeline(cur_pipe);
1377
10
            _dag[downstream_pipeline_id].push_back(new_pipe->id());
1378
1379
10
            DataSinkOperatorPtr cache_sink(new CacheSinkOperatorX(
1380
10
                    next_sink_operator_id(), op->node_id(), op->operator_id()));
1381
10
            RETURN_IF_ERROR(new_pipe->set_sink(cache_sink));
1382
10
            return Status::OK();
1383
10
        };
1384
165k
        const bool group_by_limit_opt =
1385
165k
                tnode.agg_node.__isset.agg_sort_info_by_group_key && tnode.limit > 0;
1386
1387
        /// PartitionedAggSourceOperatorX does not support "group by limit opt(#29641)" yet.
1388
        /// If `group_by_limit_opt` is true, then it might not need to spill at all.
1389
165k
        const bool enable_spill = _runtime_state->enable_spill() &&
1390
165k
                                  !tnode.agg_node.grouping_exprs.empty() && !group_by_limit_opt;
1391
165k
        const bool is_streaming_agg = tnode.agg_node.__isset.use_streaming_preaggregation &&
1392
165k
                                      tnode.agg_node.use_streaming_preaggregation &&
1393
165k
                                      !tnode.agg_node.grouping_exprs.empty();
1394
        // TODO: distinct streaming agg does not support spill.
1395
165k
        const bool can_use_distinct_streaming_agg =
1396
165k
                (!enable_spill || is_streaming_agg) && tnode.agg_node.aggregate_functions.empty() &&
1397
165k
                !tnode.agg_node.__isset.agg_sort_info_by_group_key &&
1398
165k
                _params.query_options.__isset.enable_distinct_streaming_aggregation &&
1399
165k
                _params.query_options.enable_distinct_streaming_aggregation;
1400
1401
165k
        if (can_use_distinct_streaming_agg) {
1402
92.4k
            if (need_create_cache_op) {
1403
8
                PipelinePtr new_pipe;
1404
8
                RETURN_IF_ERROR(create_query_cache_operator(new_pipe));
1405
1406
8
                cache_op = op;
1407
8
                op = std::make_shared<DistinctStreamingAggOperatorX>(pool, next_operator_id(),
1408
8
                                                                     tnode, descs);
1409
8
                RETURN_IF_ERROR(new_pipe->add_operator(op, _parallel_instances));
1410
8
                RETURN_IF_ERROR(cur_pipe->operators().front()->set_child(op));
1411
8
                cur_pipe = new_pipe;
1412
92.4k
            } else {
1413
92.4k
                op = std::make_shared<DistinctStreamingAggOperatorX>(pool, next_operator_id(),
1414
92.4k
                                                                     tnode, descs);
1415
92.4k
                RETURN_IF_ERROR(cur_pipe->add_operator(op, _parallel_instances));
1416
92.4k
            }
1417
92.4k
        } else if (is_streaming_agg) {
1418
3.66k
            if (need_create_cache_op) {
1419
0
                PipelinePtr new_pipe;
1420
0
                RETURN_IF_ERROR(create_query_cache_operator(new_pipe));
1421
0
                cache_op = op;
1422
0
                op = std::make_shared<StreamingAggOperatorX>(pool, next_operator_id(), tnode,
1423
0
                                                             descs);
1424
0
                RETURN_IF_ERROR(cur_pipe->operators().front()->set_child(op));
1425
0
                RETURN_IF_ERROR(new_pipe->add_operator(op, _parallel_instances));
1426
0
                cur_pipe = new_pipe;
1427
3.66k
            } else {
1428
3.66k
                op = std::make_shared<StreamingAggOperatorX>(pool, next_operator_id(), tnode,
1429
3.66k
                                                             descs);
1430
3.66k
                RETURN_IF_ERROR(cur_pipe->add_operator(op, _parallel_instances));
1431
3.66k
            }
1432
69.5k
        } else {
1433
            // create new pipeline to add query cache operator
1434
69.5k
            PipelinePtr new_pipe;
1435
69.5k
            if (need_create_cache_op) {
1436
2
                RETURN_IF_ERROR(create_query_cache_operator(new_pipe));
1437
2
                cache_op = op;
1438
2
            }
1439
1440
69.5k
            if (enable_spill) {
1441
1
                op = std::make_shared<PartitionedAggSourceOperatorX>(pool, tnode,
1442
1
                                                                     next_operator_id(), descs);
1443
69.5k
            } else {
1444
69.5k
                op = std::make_shared<AggSourceOperatorX>(pool, tnode, next_operator_id(), descs);
1445
69.5k
            }
1446
69.5k
            if (need_create_cache_op) {
1447
2
                RETURN_IF_ERROR(cur_pipe->operators().front()->set_child(op));
1448
2
                RETURN_IF_ERROR(new_pipe->add_operator(op, _parallel_instances));
1449
2
                cur_pipe = new_pipe;
1450
69.5k
            } else {
1451
69.5k
                RETURN_IF_ERROR(cur_pipe->add_operator(op, _parallel_instances));
1452
69.5k
            }
1453
1454
69.5k
            const auto downstream_pipeline_id = cur_pipe->id();
1455
69.5k
            if (!_dag.contains(downstream_pipeline_id)) {
1456
67.1k
                _dag.insert({downstream_pipeline_id, {}});
1457
67.1k
            }
1458
69.5k
            cur_pipe = add_pipeline(cur_pipe);
1459
69.5k
            _dag[downstream_pipeline_id].push_back(cur_pipe->id());
1460
1461
69.5k
            if (enable_spill) {
1462
1
                sink_ops.push_back(std::make_shared<PartitionedAggSinkOperatorX>(
1463
1
                        pool, next_sink_operator_id(), op->operator_id(), tnode, descs));
1464
69.5k
            } else {
1465
69.5k
                sink_ops.push_back(std::make_shared<AggSinkOperatorX>(
1466
69.5k
                        pool, next_sink_operator_id(), op->operator_id(), tnode, descs));
1467
69.5k
            }
1468
69.5k
            RETURN_IF_ERROR(cur_pipe->set_sink(sink_ops.back()));
1469
69.5k
            RETURN_IF_ERROR(cur_pipe->sink()->init(tnode, _runtime_state.get()));
1470
69.5k
        }
1471
165k
        break;
1472
165k
    }
1473
165k
    case TPlanNodeType::BUCKETED_AGGREGATION_NODE: {
1474
80
        if (tnode.bucketed_agg_node.grouping_exprs.empty()) {
1475
0
            return Status::InternalError(
1476
0
                    "Bucketed aggregation node {} should not be used without group by keys",
1477
0
                    tnode.node_id);
1478
0
        }
1479
1480
        // Create source operator (goes on the current / downstream pipeline).
1481
80
        op = std::make_shared<BucketedAggSourceOperatorX>(pool, tnode, next_operator_id(), descs);
1482
80
        RETURN_IF_ERROR(cur_pipe->add_operator(op, _parallel_instances));
1483
1484
        // Create a new pipeline for the sink side.
1485
80
        const auto downstream_pipeline_id = cur_pipe->id();
1486
80
        if (!_dag.contains(downstream_pipeline_id)) {
1487
80
            _dag.insert({downstream_pipeline_id, {}});
1488
80
        }
1489
80
        cur_pipe = add_pipeline(cur_pipe);
1490
80
        _dag[downstream_pipeline_id].push_back(cur_pipe->id());
1491
1492
        // Create sink operator.
1493
80
        sink_ops.push_back(std::make_shared<BucketedAggSinkOperatorX>(
1494
80
                pool, next_sink_operator_id(), op->operator_id(), tnode, descs));
1495
80
        RETURN_IF_ERROR(cur_pipe->set_sink(sink_ops.back()));
1496
80
        RETURN_IF_ERROR(cur_pipe->sink()->init(tnode, _runtime_state.get()));
1497
1498
        // Pre-register a single shared state for ALL instances so that every
1499
        // sink instance writes its per-instance hash table into the same
1500
        // BucketedAggSharedState and every source instance can merge across
1501
        // all of them.
1502
80
        {
1503
80
            auto shared_state = BucketedAggSharedState::create_shared();
1504
80
            shared_state->id = op->operator_id();
1505
80
            shared_state->related_op_ids.insert(op->operator_id());
1506
1507
512
            for (int i = 0; i < _num_instances; i++) {
1508
432
                auto sink_dep = std::make_shared<Dependency>(op->operator_id(), op->node_id(),
1509
432
                                                             "BUCKETED_AGG_SINK_DEPENDENCY");
1510
432
                sink_dep->set_shared_state(shared_state.get());
1511
432
                shared_state->sink_deps.push_back(sink_dep);
1512
432
            }
1513
80
            shared_state->create_source_dependencies(_num_instances, op->operator_id(),
1514
80
                                                     op->node_id(), "BUCKETED_AGG_SOURCE");
1515
80
            _op_id_to_shared_state.insert(
1516
80
                    {op->operator_id(), {shared_state, shared_state->sink_deps}});
1517
80
        }
1518
80
        break;
1519
80
    }
1520
9.62k
    case TPlanNodeType::HASH_JOIN_NODE: {
1521
9.62k
        const auto is_broadcast_join = tnode.hash_join_node.__isset.is_broadcast_join &&
1522
9.62k
                                       tnode.hash_join_node.is_broadcast_join;
1523
9.62k
        const auto enable_spill = _runtime_state->enable_spill();
1524
9.62k
        if (enable_spill && !is_broadcast_join) {
1525
0
            auto tnode_ = tnode;
1526
0
            tnode_.runtime_filters.clear();
1527
0
            auto inner_probe_operator =
1528
0
                    std::make_shared<HashJoinProbeOperatorX>(pool, tnode_, 0, descs);
1529
1530
            // probe side inner sink operator is used to build hash table on probe side when data is spilled.
1531
            // So here use `tnode_` which has no runtime filters.
1532
0
            auto probe_side_inner_sink_operator =
1533
0
                    std::make_shared<HashJoinBuildSinkOperatorX>(pool, 0, 0, tnode_, descs);
1534
1535
0
            RETURN_IF_ERROR(inner_probe_operator->init(tnode_, _runtime_state.get()));
1536
0
            RETURN_IF_ERROR(probe_side_inner_sink_operator->init(tnode_, _runtime_state.get()));
1537
1538
0
            auto probe_operator = std::make_shared<PartitionedHashJoinProbeOperatorX>(
1539
0
                    pool, tnode_, next_operator_id(), descs);
1540
0
            probe_operator->set_inner_operators(probe_side_inner_sink_operator,
1541
0
                                                inner_probe_operator);
1542
0
            op = std::move(probe_operator);
1543
0
            RETURN_IF_ERROR(cur_pipe->add_operator(op, _parallel_instances));
1544
1545
0
            const auto downstream_pipeline_id = cur_pipe->id();
1546
0
            if (!_dag.contains(downstream_pipeline_id)) {
1547
0
                _dag.insert({downstream_pipeline_id, {}});
1548
0
            }
1549
0
            PipelinePtr build_side_pipe = add_pipeline(cur_pipe);
1550
0
            _dag[downstream_pipeline_id].push_back(build_side_pipe->id());
1551
1552
0
            auto inner_sink_operator =
1553
0
                    std::make_shared<HashJoinBuildSinkOperatorX>(pool, 0, 0, tnode, descs);
1554
0
            auto sink_operator = std::make_shared<PartitionedHashJoinSinkOperatorX>(
1555
0
                    pool, next_sink_operator_id(), op->operator_id(), tnode_, descs);
1556
0
            RETURN_IF_ERROR(inner_sink_operator->init(tnode, _runtime_state.get()));
1557
1558
0
            sink_operator->set_inner_operators(inner_sink_operator, inner_probe_operator);
1559
0
            sink_ops.push_back(std::move(sink_operator));
1560
0
            RETURN_IF_ERROR(build_side_pipe->set_sink(sink_ops.back()));
1561
0
            RETURN_IF_ERROR(build_side_pipe->sink()->init(tnode_, _runtime_state.get()));
1562
1563
0
            _pipeline_parent_map.push(op->node_id(), cur_pipe);
1564
0
            _pipeline_parent_map.push(op->node_id(), build_side_pipe);
1565
9.62k
        } else {
1566
9.62k
            op = std::make_shared<HashJoinProbeOperatorX>(pool, tnode, next_operator_id(), descs);
1567
9.62k
            RETURN_IF_ERROR(cur_pipe->add_operator(op, _parallel_instances));
1568
1569
9.62k
            const auto downstream_pipeline_id = cur_pipe->id();
1570
9.62k
            if (!_dag.contains(downstream_pipeline_id)) {
1571
7.91k
                _dag.insert({downstream_pipeline_id, {}});
1572
7.91k
            }
1573
9.62k
            PipelinePtr build_side_pipe = add_pipeline(cur_pipe);
1574
9.62k
            _dag[downstream_pipeline_id].push_back(build_side_pipe->id());
1575
1576
9.62k
            sink_ops.push_back(std::make_shared<HashJoinBuildSinkOperatorX>(
1577
9.62k
                    pool, next_sink_operator_id(), op->operator_id(), tnode, descs));
1578
9.62k
            RETURN_IF_ERROR(build_side_pipe->set_sink(sink_ops.back()));
1579
9.62k
            RETURN_IF_ERROR(build_side_pipe->sink()->init(tnode, _runtime_state.get()));
1580
1581
9.62k
            _pipeline_parent_map.push(op->node_id(), cur_pipe);
1582
9.62k
            _pipeline_parent_map.push(op->node_id(), build_side_pipe);
1583
9.62k
        }
1584
9.62k
        if (is_broadcast_join && _runtime_state->enable_share_hash_table_for_broadcast_join()) {
1585
2.87k
            std::shared_ptr<HashJoinSharedState> shared_state =
1586
2.87k
                    HashJoinSharedState::create_shared(_num_instances);
1587
18.4k
            for (int i = 0; i < _num_instances; i++) {
1588
15.5k
                auto sink_dep = std::make_shared<Dependency>(op->operator_id(), op->node_id(),
1589
15.5k
                                                             "HASH_JOIN_BUILD_DEPENDENCY");
1590
15.5k
                sink_dep->set_shared_state(shared_state.get());
1591
15.5k
                shared_state->sink_deps.push_back(sink_dep);
1592
15.5k
            }
1593
2.87k
            shared_state->create_source_dependencies(_num_instances, op->operator_id(),
1594
2.87k
                                                     op->node_id(), "HASH_JOIN_PROBE");
1595
2.87k
            _op_id_to_shared_state.insert(
1596
2.87k
                    {op->operator_id(), {shared_state, shared_state->sink_deps}});
1597
2.87k
        }
1598
9.62k
        break;
1599
9.62k
    }
1600
5.34k
    case TPlanNodeType::CROSS_JOIN_NODE: {
1601
5.34k
        op = std::make_shared<NestedLoopJoinProbeOperatorX>(pool, tnode, next_operator_id(), descs);
1602
5.34k
        RETURN_IF_ERROR(cur_pipe->add_operator(op, _parallel_instances));
1603
1604
5.34k
        const auto downstream_pipeline_id = cur_pipe->id();
1605
5.34k
        if (!_dag.contains(downstream_pipeline_id)) {
1606
5.11k
            _dag.insert({downstream_pipeline_id, {}});
1607
5.11k
        }
1608
5.34k
        PipelinePtr build_side_pipe = add_pipeline(cur_pipe);
1609
5.34k
        _dag[downstream_pipeline_id].push_back(build_side_pipe->id());
1610
1611
5.34k
        sink_ops.push_back(std::make_shared<NestedLoopJoinBuildSinkOperatorX>(
1612
5.34k
                pool, next_sink_operator_id(), op->operator_id(), tnode, descs));
1613
5.34k
        RETURN_IF_ERROR(build_side_pipe->set_sink(sink_ops.back()));
1614
5.34k
        RETURN_IF_ERROR(build_side_pipe->sink()->init(tnode, _runtime_state.get()));
1615
5.34k
        _pipeline_parent_map.push(op->node_id(), cur_pipe);
1616
5.34k
        _pipeline_parent_map.push(op->node_id(), build_side_pipe);
1617
5.34k
        break;
1618
5.34k
    }
1619
53.8k
    case TPlanNodeType::UNION_NODE: {
1620
53.8k
        int child_count = tnode.num_children;
1621
53.8k
        op = std::make_shared<UnionSourceOperatorX>(pool, tnode, next_operator_id(), descs);
1622
53.8k
        RETURN_IF_ERROR(cur_pipe->add_operator(op, _parallel_instances));
1623
1624
53.8k
        const auto downstream_pipeline_id = cur_pipe->id();
1625
53.8k
        if (!_dag.contains(downstream_pipeline_id)) {
1626
53.5k
            _dag.insert({downstream_pipeline_id, {}});
1627
53.5k
        }
1628
55.2k
        for (int i = 0; i < child_count; i++) {
1629
1.41k
            PipelinePtr build_side_pipe = add_pipeline(cur_pipe);
1630
1.41k
            _dag[downstream_pipeline_id].push_back(build_side_pipe->id());
1631
1.41k
            sink_ops.push_back(std::make_shared<UnionSinkOperatorX>(
1632
1.41k
                    i, next_sink_operator_id(), op->operator_id(), pool, tnode, descs));
1633
1.41k
            RETURN_IF_ERROR(build_side_pipe->set_sink(sink_ops.back()));
1634
1.41k
            RETURN_IF_ERROR(build_side_pipe->sink()->init(tnode, _runtime_state.get()));
1635
            // preset children pipelines. if any pipeline found this as its father, will use the prepared pipeline to build.
1636
1.41k
            _pipeline_parent_map.push(op->node_id(), build_side_pipe);
1637
1.41k
        }
1638
53.8k
        break;
1639
53.8k
    }
1640
53.8k
    case TPlanNodeType::SORT_NODE: {
1641
44.6k
        const auto should_spill = _runtime_state->enable_spill() &&
1642
44.6k
                                  tnode.sort_node.algorithm == TSortAlgorithm::FULL_SORT;
1643
44.6k
        const bool use_local_merge =
1644
44.6k
                tnode.sort_node.__isset.use_local_merge && tnode.sort_node.use_local_merge;
1645
44.6k
        if (should_spill) {
1646
9
            op = std::make_shared<SpillSortSourceOperatorX>(pool, tnode, next_operator_id(), descs);
1647
44.6k
        } else if (use_local_merge) {
1648
42.4k
            op = std::make_shared<LocalMergeSortSourceOperatorX>(pool, tnode, next_operator_id(),
1649
42.4k
                                                                 descs);
1650
42.4k
        } else {
1651
2.23k
            op = std::make_shared<SortSourceOperatorX>(pool, tnode, next_operator_id(), descs);
1652
2.23k
        }
1653
44.6k
        RETURN_IF_ERROR(cur_pipe->add_operator(op, _parallel_instances));
1654
1655
44.6k
        const auto downstream_pipeline_id = cur_pipe->id();
1656
44.6k
        if (!_dag.contains(downstream_pipeline_id)) {
1657
44.6k
            _dag.insert({downstream_pipeline_id, {}});
1658
44.6k
        }
1659
44.6k
        cur_pipe = add_pipeline(cur_pipe);
1660
44.6k
        _dag[downstream_pipeline_id].push_back(cur_pipe->id());
1661
1662
44.6k
        if (should_spill) {
1663
9
            sink_ops.push_back(std::make_shared<SpillSortSinkOperatorX>(
1664
9
                    pool, next_sink_operator_id(), op->operator_id(), tnode, descs));
1665
44.6k
        } else {
1666
44.6k
            sink_ops.push_back(std::make_shared<SortSinkOperatorX>(
1667
44.6k
                    pool, next_sink_operator_id(), op->operator_id(), tnode, descs));
1668
44.6k
        }
1669
44.6k
        RETURN_IF_ERROR(cur_pipe->set_sink(sink_ops.back()));
1670
44.6k
        RETURN_IF_ERROR(cur_pipe->sink()->init(tnode, _runtime_state.get()));
1671
44.6k
        break;
1672
44.6k
    }
1673
44.6k
    case TPlanNodeType::PARTITION_SORT_NODE: {
1674
62
        op = std::make_shared<PartitionSortSourceOperatorX>(pool, tnode, next_operator_id(), descs);
1675
62
        RETURN_IF_ERROR(cur_pipe->add_operator(op, _parallel_instances));
1676
1677
62
        const auto downstream_pipeline_id = cur_pipe->id();
1678
62
        if (!_dag.contains(downstream_pipeline_id)) {
1679
62
            _dag.insert({downstream_pipeline_id, {}});
1680
62
        }
1681
62
        cur_pipe = add_pipeline(cur_pipe);
1682
62
        _dag[downstream_pipeline_id].push_back(cur_pipe->id());
1683
1684
62
        sink_ops.push_back(std::make_shared<PartitionSortSinkOperatorX>(
1685
62
                pool, next_sink_operator_id(), op->operator_id(), tnode, descs));
1686
62
        RETURN_IF_ERROR(cur_pipe->set_sink(sink_ops.back()));
1687
62
        RETURN_IF_ERROR(cur_pipe->sink()->init(tnode, _runtime_state.get()));
1688
62
        break;
1689
62
    }
1690
1.53k
    case TPlanNodeType::ANALYTIC_EVAL_NODE: {
1691
1.53k
        op = std::make_shared<AnalyticSourceOperatorX>(pool, tnode, next_operator_id(), descs);
1692
1.53k
        RETURN_IF_ERROR(cur_pipe->add_operator(op, _parallel_instances));
1693
1694
1.53k
        const auto downstream_pipeline_id = cur_pipe->id();
1695
1.53k
        if (!_dag.contains(downstream_pipeline_id)) {
1696
1.51k
            _dag.insert({downstream_pipeline_id, {}});
1697
1.51k
        }
1698
1.53k
        cur_pipe = add_pipeline(cur_pipe);
1699
1.53k
        _dag[downstream_pipeline_id].push_back(cur_pipe->id());
1700
1701
1.53k
        sink_ops.push_back(std::make_shared<AnalyticSinkOperatorX>(
1702
1.53k
                pool, next_sink_operator_id(), op->operator_id(), tnode, descs));
1703
1.53k
        RETURN_IF_ERROR(cur_pipe->set_sink(sink_ops.back()));
1704
1.53k
        RETURN_IF_ERROR(cur_pipe->sink()->init(tnode, _runtime_state.get()));
1705
1.53k
        break;
1706
1.53k
    }
1707
1.60k
    case TPlanNodeType::MATERIALIZATION_NODE: {
1708
1.60k
        op = std::make_shared<MaterializationOperator>(pool, tnode, next_operator_id(), descs);
1709
1.60k
        RETURN_IF_ERROR(cur_pipe->add_operator(op, _parallel_instances));
1710
1.60k
        break;
1711
1.60k
    }
1712
1.60k
    case TPlanNodeType::INTERSECT_NODE: {
1713
134
        RETURN_IF_ERROR(_build_operators_for_set_operation_node<true>(pool, tnode, descs, op,
1714
134
                                                                      cur_pipe, sink_ops));
1715
134
        break;
1716
134
    }
1717
134
    case TPlanNodeType::EXCEPT_NODE: {
1718
133
        RETURN_IF_ERROR(_build_operators_for_set_operation_node<false>(pool, tnode, descs, op,
1719
133
                                                                       cur_pipe, sink_ops));
1720
133
        break;
1721
133
    }
1722
296
    case TPlanNodeType::REPEAT_NODE: {
1723
296
        op = std::make_shared<RepeatOperatorX>(pool, tnode, next_operator_id(), descs);
1724
296
        RETURN_IF_ERROR(cur_pipe->add_operator(op, _parallel_instances));
1725
296
        break;
1726
296
    }
1727
914
    case TPlanNodeType::TABLE_FUNCTION_NODE: {
1728
914
        op = std::make_shared<TableFunctionOperatorX>(pool, tnode, next_operator_id(), descs);
1729
914
        RETURN_IF_ERROR(cur_pipe->add_operator(op, _parallel_instances));
1730
914
        break;
1731
914
    }
1732
914
    case TPlanNodeType::ASSERT_NUM_ROWS_NODE: {
1733
218
        op = std::make_shared<AssertNumRowsOperatorX>(pool, tnode, next_operator_id(), descs);
1734
218
        RETURN_IF_ERROR(cur_pipe->add_operator(op, _parallel_instances));
1735
218
        break;
1736
218
    }
1737
1.67k
    case TPlanNodeType::EMPTY_SET_NODE: {
1738
1.67k
        op = std::make_shared<EmptySetSourceOperatorX>(pool, tnode, next_operator_id(), descs);
1739
1.67k
        RETURN_IF_ERROR(cur_pipe->add_operator(op, _parallel_instances));
1740
1.67k
        break;
1741
1.67k
    }
1742
1.67k
    case TPlanNodeType::DATA_GEN_SCAN_NODE: {
1743
464
        op = std::make_shared<DataGenSourceOperatorX>(pool, tnode, next_operator_id(), descs);
1744
464
        RETURN_IF_ERROR(cur_pipe->add_operator(op, _parallel_instances));
1745
464
        fe_with_old_version = !tnode.__isset.is_serial_operator;
1746
464
        break;
1747
464
    }
1748
2.06k
    case TPlanNodeType::SCHEMA_SCAN_NODE: {
1749
2.06k
        op = std::make_shared<SchemaScanOperatorX>(pool, tnode, next_operator_id(), descs);
1750
2.06k
        RETURN_IF_ERROR(cur_pipe->add_operator(op, _parallel_instances));
1751
2.06k
        break;
1752
2.06k
    }
1753
7.29k
    case TPlanNodeType::META_SCAN_NODE: {
1754
7.29k
        op = std::make_shared<MetaScanOperatorX>(pool, tnode, next_operator_id(), descs);
1755
7.29k
        RETURN_IF_ERROR(cur_pipe->add_operator(op, _parallel_instances));
1756
7.29k
        break;
1757
7.29k
    }
1758
7.29k
    case TPlanNodeType::SELECT_NODE: {
1759
2.21k
        op = std::make_shared<SelectOperatorX>(pool, tnode, next_operator_id(), descs);
1760
2.21k
        RETURN_IF_ERROR(cur_pipe->add_operator(op, _parallel_instances));
1761
2.21k
        break;
1762
2.21k
    }
1763
2.21k
    case TPlanNodeType::REC_CTE_NODE: {
1764
151
        op = std::make_shared<RecCTESourceOperatorX>(pool, tnode, next_operator_id(), descs);
1765
151
        RETURN_IF_ERROR(cur_pipe->add_operator(op, _parallel_instances));
1766
1767
151
        const auto downstream_pipeline_id = cur_pipe->id();
1768
151
        if (!_dag.contains(downstream_pipeline_id)) {
1769
148
            _dag.insert({downstream_pipeline_id, {}});
1770
148
        }
1771
1772
151
        PipelinePtr anchor_side_pipe = add_pipeline(cur_pipe);
1773
151
        _dag[downstream_pipeline_id].push_back(anchor_side_pipe->id());
1774
1775
151
        DataSinkOperatorPtr anchor_sink;
1776
151
        anchor_sink = std::make_shared<RecCTEAnchorSinkOperatorX>(next_sink_operator_id(),
1777
151
                                                                  op->operator_id(), tnode, descs);
1778
151
        RETURN_IF_ERROR(anchor_side_pipe->set_sink(anchor_sink));
1779
151
        RETURN_IF_ERROR(anchor_side_pipe->sink()->init(tnode, _runtime_state.get()));
1780
151
        _pipeline_parent_map.push(op->node_id(), anchor_side_pipe);
1781
1782
151
        PipelinePtr rec_side_pipe = add_pipeline(cur_pipe);
1783
151
        _dag[downstream_pipeline_id].push_back(rec_side_pipe->id());
1784
1785
151
        DataSinkOperatorPtr rec_sink;
1786
151
        rec_sink = std::make_shared<RecCTESinkOperatorX>(next_sink_operator_id(), op->operator_id(),
1787
151
                                                         tnode, descs);
1788
151
        RETURN_IF_ERROR(rec_side_pipe->set_sink(rec_sink));
1789
151
        RETURN_IF_ERROR(rec_side_pipe->sink()->init(tnode, _runtime_state.get()));
1790
151
        _pipeline_parent_map.push(op->node_id(), rec_side_pipe);
1791
1792
151
        break;
1793
151
    }
1794
1.95k
    case TPlanNodeType::REC_CTE_SCAN_NODE: {
1795
1.95k
        op = std::make_shared<RecCTEScanOperatorX>(pool, tnode, next_operator_id(), descs);
1796
1.95k
        RETURN_IF_ERROR(cur_pipe->add_operator(op, _parallel_instances));
1797
1.95k
        break;
1798
1.95k
    }
1799
1.95k
    default:
1800
0
        return Status::InternalError("Unsupported exec type in pipeline: {}",
1801
0
                                     print_plan_node_type(tnode.node_type));
1802
696k
    }
1803
694k
    if (_params.__isset.parallel_instances && fe_with_old_version) {
1804
0
        cur_pipe->set_num_tasks(_params.parallel_instances);
1805
0
        op->set_serial_operator();
1806
0
    }
1807
1808
694k
    return Status::OK();
1809
696k
}
1810
// NOLINTEND(readability-function-cognitive-complexity)
1811
// NOLINTEND(readability-function-size)
1812
1813
template <bool is_intersect>
1814
Status PipelineFragmentContext::_build_operators_for_set_operation_node(
1815
        ObjectPool* pool, const TPlanNode& tnode, const DescriptorTbl& descs, OperatorPtr& op,
1816
267
        PipelinePtr& cur_pipe, std::vector<DataSinkOperatorPtr>& sink_ops) {
1817
267
    op.reset(new SetSourceOperatorX<is_intersect>(pool, tnode, next_operator_id(), descs));
1818
267
    RETURN_IF_ERROR(cur_pipe->add_operator(op, _parallel_instances));
1819
1820
267
    const auto downstream_pipeline_id = cur_pipe->id();
1821
267
    if (!_dag.contains(downstream_pipeline_id)) {
1822
242
        _dag.insert({downstream_pipeline_id, {}});
1823
242
    }
1824
1825
895
    for (int child_id = 0; child_id < tnode.num_children; child_id++) {
1826
628
        PipelinePtr probe_side_pipe = add_pipeline(cur_pipe);
1827
628
        _dag[downstream_pipeline_id].push_back(probe_side_pipe->id());
1828
1829
628
        if (child_id == 0) {
1830
267
            sink_ops.push_back(std::make_shared<SetSinkOperatorX<is_intersect>>(
1831
267
                    child_id, next_sink_operator_id(), op->operator_id(), pool, tnode, descs));
1832
361
        } else {
1833
361
            sink_ops.push_back(std::make_shared<SetProbeSinkOperatorX<is_intersect>>(
1834
361
                    child_id, next_sink_operator_id(), op->operator_id(), pool, tnode, descs));
1835
361
        }
1836
628
        RETURN_IF_ERROR(probe_side_pipe->set_sink(sink_ops.back()));
1837
628
        RETURN_IF_ERROR(probe_side_pipe->sink()->init(tnode, _runtime_state.get()));
1838
        // prepare children pipelines. if any pipeline found this as its father, will use the prepared pipeline to build.
1839
628
        _pipeline_parent_map.push(op->node_id(), probe_side_pipe);
1840
628
    }
1841
1842
267
    return Status::OK();
1843
267
}
_ZN5doris23PipelineFragmentContext39_build_operators_for_set_operation_nodeILb1EEENS_6StatusEPNS_10ObjectPoolERKNS_9TPlanNodeERKNS_13DescriptorTblERSt10shared_ptrINS_13OperatorXBaseEERSB_INS_8PipelineEERSt6vectorISB_INS_21DataSinkOperatorXBaseEESaISK_EE
Line
Count
Source
1816
134
        PipelinePtr& cur_pipe, std::vector<DataSinkOperatorPtr>& sink_ops) {
1817
134
    op.reset(new SetSourceOperatorX<is_intersect>(pool, tnode, next_operator_id(), descs));
1818
134
    RETURN_IF_ERROR(cur_pipe->add_operator(op, _parallel_instances));
1819
1820
134
    const auto downstream_pipeline_id = cur_pipe->id();
1821
134
    if (!_dag.contains(downstream_pipeline_id)) {
1822
118
        _dag.insert({downstream_pipeline_id, {}});
1823
118
    }
1824
1825
481
    for (int child_id = 0; child_id < tnode.num_children; child_id++) {
1826
347
        PipelinePtr probe_side_pipe = add_pipeline(cur_pipe);
1827
347
        _dag[downstream_pipeline_id].push_back(probe_side_pipe->id());
1828
1829
347
        if (child_id == 0) {
1830
134
            sink_ops.push_back(std::make_shared<SetSinkOperatorX<is_intersect>>(
1831
134
                    child_id, next_sink_operator_id(), op->operator_id(), pool, tnode, descs));
1832
213
        } else {
1833
213
            sink_ops.push_back(std::make_shared<SetProbeSinkOperatorX<is_intersect>>(
1834
213
                    child_id, next_sink_operator_id(), op->operator_id(), pool, tnode, descs));
1835
213
        }
1836
347
        RETURN_IF_ERROR(probe_side_pipe->set_sink(sink_ops.back()));
1837
347
        RETURN_IF_ERROR(probe_side_pipe->sink()->init(tnode, _runtime_state.get()));
1838
        // prepare children pipelines. if any pipeline found this as its father, will use the prepared pipeline to build.
1839
347
        _pipeline_parent_map.push(op->node_id(), probe_side_pipe);
1840
347
    }
1841
1842
134
    return Status::OK();
1843
134
}
_ZN5doris23PipelineFragmentContext39_build_operators_for_set_operation_nodeILb0EEENS_6StatusEPNS_10ObjectPoolERKNS_9TPlanNodeERKNS_13DescriptorTblERSt10shared_ptrINS_13OperatorXBaseEERSB_INS_8PipelineEERSt6vectorISB_INS_21DataSinkOperatorXBaseEESaISK_EE
Line
Count
Source
1816
133
        PipelinePtr& cur_pipe, std::vector<DataSinkOperatorPtr>& sink_ops) {
1817
133
    op.reset(new SetSourceOperatorX<is_intersect>(pool, tnode, next_operator_id(), descs));
1818
133
    RETURN_IF_ERROR(cur_pipe->add_operator(op, _parallel_instances));
1819
1820
133
    const auto downstream_pipeline_id = cur_pipe->id();
1821
133
    if (!_dag.contains(downstream_pipeline_id)) {
1822
124
        _dag.insert({downstream_pipeline_id, {}});
1823
124
    }
1824
1825
414
    for (int child_id = 0; child_id < tnode.num_children; child_id++) {
1826
281
        PipelinePtr probe_side_pipe = add_pipeline(cur_pipe);
1827
281
        _dag[downstream_pipeline_id].push_back(probe_side_pipe->id());
1828
1829
281
        if (child_id == 0) {
1830
133
            sink_ops.push_back(std::make_shared<SetSinkOperatorX<is_intersect>>(
1831
133
                    child_id, next_sink_operator_id(), op->operator_id(), pool, tnode, descs));
1832
148
        } else {
1833
148
            sink_ops.push_back(std::make_shared<SetProbeSinkOperatorX<is_intersect>>(
1834
148
                    child_id, next_sink_operator_id(), op->operator_id(), pool, tnode, descs));
1835
148
        }
1836
281
        RETURN_IF_ERROR(probe_side_pipe->set_sink(sink_ops.back()));
1837
281
        RETURN_IF_ERROR(probe_side_pipe->sink()->init(tnode, _runtime_state.get()));
1838
        // prepare children pipelines. if any pipeline found this as its father, will use the prepared pipeline to build.
1839
281
        _pipeline_parent_map.push(op->node_id(), probe_side_pipe);
1840
281
    }
1841
1842
133
    return Status::OK();
1843
133
}
1844
1845
445k
Status PipelineFragmentContext::submit() {
1846
445k
    if (_submitted) {
1847
0
        return Status::InternalError("submitted");
1848
0
    }
1849
445k
    _submitted = true;
1850
1851
445k
    int submit_tasks = 0;
1852
445k
    Status st;
1853
445k
    auto* scheduler = _query_ctx->get_pipe_exec_scheduler();
1854
1.26M
    for (auto& task : _tasks) {
1855
2.07M
        for (auto& t : task) {
1856
2.07M
            st = scheduler->submit(t.first);
1857
2.07M
            DBUG_EXECUTE_IF("PipelineFragmentContext.submit.failed",
1858
2.07M
                            { st = Status::Aborted("PipelineFragmentContext.submit.failed"); });
1859
2.07M
            if (!st) {
1860
0
                cancel(Status::InternalError("submit context to executor fail"));
1861
0
                std::lock_guard<std::mutex> l(_task_mutex);
1862
0
                _total_tasks = submit_tasks;
1863
0
                break;
1864
0
            }
1865
2.07M
            submit_tasks++;
1866
2.07M
        }
1867
1.26M
    }
1868
445k
    if (!st.ok()) {
1869
0
        bool need_remove = false;
1870
0
        {
1871
0
            std::lock_guard<std::mutex> l(_task_mutex);
1872
0
            if (_closed_tasks >= _total_tasks) {
1873
0
                need_remove = _close_fragment_instance();
1874
0
            }
1875
0
        }
1876
        // Call remove_pipeline_context() outside _task_mutex to avoid ABBA deadlock.
1877
0
        if (need_remove) {
1878
0
            _exec_env->fragment_mgr()->remove_pipeline_context({_query_id, _fragment_id});
1879
0
        }
1880
0
        return Status::InternalError("Submit pipeline failed. err = {}, BE: {}", st.to_string(),
1881
0
                                     BackendOptions::get_localhost());
1882
445k
    } else {
1883
445k
        return st;
1884
445k
    }
1885
445k
}
1886
1887
0
void PipelineFragmentContext::print_profile(const std::string& extra_info) {
1888
0
    if (_runtime_state->enable_profile()) {
1889
0
        std::stringstream ss;
1890
0
        for (auto runtime_profile_ptr : _runtime_state->pipeline_id_to_profile()) {
1891
0
            runtime_profile_ptr->pretty_print(&ss);
1892
0
        }
1893
1894
0
        if (_runtime_state->load_channel_profile()) {
1895
0
            _runtime_state->load_channel_profile()->pretty_print(&ss);
1896
0
        }
1897
1898
0
        auto profile_str =
1899
0
                fmt::format("Query {} fragment {} {}, profile, {}", print_id(this->_query_id),
1900
0
                            this->_fragment_id, extra_info, ss.str());
1901
0
        LOG_LONG_STRING(INFO, profile_str);
1902
0
    }
1903
0
}
1904
// If all pipeline tasks binded to the fragment instance are finished, then we could
1905
// close the fragment instance.
1906
// Returns true if the caller should call remove_pipeline_context() **after** releasing
1907
// _task_mutex. We must not call remove_pipeline_context() here because it acquires
1908
// _pipeline_map's shard lock, and this function is called while _task_mutex is held.
1909
// Acquiring _pipeline_map while holding _task_mutex creates an ABBA deadlock with
1910
// dump_pipeline_tasks(), which acquires _pipeline_map first and then _task_mutex
1911
// (via debug_string()).
1912
446k
bool PipelineFragmentContext::_close_fragment_instance() {
1913
446k
    if (_is_fragment_instance_closed) {
1914
0
        return false;
1915
0
    }
1916
446k
    Defer defer_op {[&]() { _is_fragment_instance_closed = true; }};
1917
446k
    _fragment_level_profile->total_time_counter()->update(_fragment_watcher.elapsed_time());
1918
446k
    if (!_need_notify_close) {
1919
443k
        auto st = send_report(true);
1920
443k
        if (!st) {
1921
0
            LOG(WARNING) << fmt::format("Failed to send report for query {}, fragment {}: {}",
1922
0
                                        print_id(_query_id), _fragment_id, st.to_string());
1923
0
        }
1924
443k
    }
1925
    // Print profile content in info log is a tempoeray solution for stream load and external_connector.
1926
    // Since stream load does not have someting like coordinator on FE, so
1927
    // backend can not report profile to FE, ant its profile can not be shown
1928
    // in the same way with other query. So we print the profile content to info log.
1929
1930
446k
    if (_runtime_state->enable_profile() &&
1931
446k
        (_query_ctx->get_query_source() == QuerySource::STREAM_LOAD ||
1932
2.24k
         _query_ctx->get_query_source() == QuerySource::EXTERNAL_CONNECTOR ||
1933
2.24k
         _query_ctx->get_query_source() == QuerySource::GROUP_COMMIT_LOAD)) {
1934
0
        std::stringstream ss;
1935
        // Compute the _local_time_percent before pretty_print the runtime_profile
1936
        // Before add this operation, the print out like that:
1937
        // UNION_NODE (id=0):(Active: 56.720us, non-child: 00.00%)
1938
        // After add the operation, the print out like that:
1939
        // UNION_NODE (id=0):(Active: 56.720us, non-child: 82.53%)
1940
        // We can easily know the exec node execute time without child time consumed.
1941
0
        for (auto runtime_profile_ptr : _runtime_state->pipeline_id_to_profile()) {
1942
0
            runtime_profile_ptr->pretty_print(&ss);
1943
0
        }
1944
1945
0
        if (_runtime_state->load_channel_profile()) {
1946
0
            _runtime_state->load_channel_profile()->pretty_print(&ss);
1947
0
        }
1948
1949
0
        LOG_INFO("Query {} fragment {} profile:\n {}", print_id(_query_id), _fragment_id, ss.str());
1950
0
    }
1951
1952
446k
    if (_query_ctx->enable_profile()) {
1953
2.24k
        _query_ctx->add_fragment_profile(_fragment_id, collect_realtime_profile(),
1954
2.24k
                                         collect_realtime_load_channel_profile());
1955
2.24k
    }
1956
1957
    // Return whether the caller needs to remove from the pipeline map.
1958
    // The caller must do this after releasing _task_mutex.
1959
446k
    return !_need_notify_close;
1960
446k
}
1961
1962
2.06M
void PipelineFragmentContext::decrement_running_task(PipelineId pipeline_id) {
1963
    // If all tasks of this pipeline has been closed, upstream tasks is never needed, and we just make those runnable here
1964
2.06M
    DCHECK(_pip_id_to_pipeline.contains(pipeline_id));
1965
2.06M
    if (_pip_id_to_pipeline[pipeline_id]->close_task()) {
1966
708k
        if (_dag.contains(pipeline_id)) {
1967
384k
            for (auto dep : _dag[pipeline_id]) {
1968
384k
                _pip_id_to_pipeline[dep]->make_all_runnable(pipeline_id);
1969
384k
            }
1970
304k
        }
1971
708k
    }
1972
2.06M
    bool need_remove = false;
1973
2.06M
    {
1974
2.06M
        std::lock_guard<std::mutex> l(_task_mutex);
1975
2.06M
        ++_closed_tasks;
1976
        // Update query-level finished task progress in real time.
1977
2.06M
        _query_ctx->inc_finished_task_num();
1978
2.06M
        if (_closed_tasks >= _total_tasks) {
1979
446k
            need_remove = _close_fragment_instance();
1980
446k
        }
1981
2.06M
    }
1982
    // Call remove_pipeline_context() outside _task_mutex to avoid ABBA deadlock.
1983
2.06M
    if (need_remove) {
1984
443k
        _exec_env->fragment_mgr()->remove_pipeline_context({_query_id, _fragment_id});
1985
443k
    }
1986
2.06M
}
1987
1988
54.5k
std::string PipelineFragmentContext::get_load_error_url() {
1989
54.5k
    if (const auto& str = _runtime_state->get_error_log_file_path(); !str.empty()) {
1990
0
        return to_load_error_http_path(str);
1991
0
    }
1992
132k
    for (auto& tasks : _tasks) {
1993
200k
        for (auto& task : tasks) {
1994
200k
            if (const auto& str = task.second->get_error_log_file_path(); !str.empty()) {
1995
175
                return to_load_error_http_path(str);
1996
175
            }
1997
200k
        }
1998
132k
    }
1999
54.3k
    return "";
2000
54.5k
}
2001
2002
54.5k
std::string PipelineFragmentContext::get_first_error_msg() {
2003
54.5k
    if (const auto& str = _runtime_state->get_first_error_msg(); !str.empty()) {
2004
0
        return str;
2005
0
    }
2006
132k
    for (auto& tasks : _tasks) {
2007
200k
        for (auto& task : tasks) {
2008
200k
            if (const auto& str = task.second->get_first_error_msg(); !str.empty()) {
2009
175
                return str;
2010
175
            }
2011
200k
        }
2012
132k
    }
2013
54.3k
    return "";
2014
54.5k
}
2015
2016
0
std::string PipelineFragmentContext::_to_http_path(const std::string& file_name) const {
2017
0
    std::stringstream url;
2018
0
    url << "http://" << BackendOptions::get_localhost() << ":" << config::webserver_port
2019
0
        << "/api/_download_load?"
2020
0
        << "token=" << _exec_env->token() << "&file=" << file_name;
2021
0
    return url.str();
2022
0
}
2023
2024
48.4k
void PipelineFragmentContext::_coordinator_callback(const ReportStatusRequest& req) {
2025
48.4k
    DBUG_EXECUTE_IF("FragmentMgr::coordinator_callback.report_delay", {
2026
48.4k
        int random_seconds = req.status.is<ErrorCode::DATA_QUALITY_ERROR>() ? 8 : 2;
2027
48.4k
        LOG_INFO("sleep : ").tag("time", random_seconds).tag("query_id", print_id(req.query_id));
2028
48.4k
        std::this_thread::sleep_for(std::chrono::seconds(random_seconds));
2029
48.4k
        LOG_INFO("sleep done").tag("query_id", print_id(req.query_id));
2030
48.4k
    });
2031
2032
48.4k
    DCHECK(req.status.ok() || req.done); // if !status.ok() => done
2033
48.4k
    if (req.coord_addr.hostname == "external") {
2034
        // External query (flink/spark read tablets) not need to report to FE.
2035
0
        return;
2036
0
    }
2037
48.4k
    int callback_retries = 10;
2038
48.4k
    const int sleep_ms = 1000;
2039
48.4k
    Status exec_status = req.status;
2040
48.4k
    Status coord_status;
2041
48.4k
    std::unique_ptr<FrontendServiceConnection> coord = nullptr;
2042
48.4k
    do {
2043
48.4k
        coord = std::make_unique<FrontendServiceConnection>(_exec_env->frontend_client_cache(),
2044
48.4k
                                                            req.coord_addr, &coord_status);
2045
48.4k
        if (!coord_status.ok()) {
2046
0
            std::this_thread::sleep_for(std::chrono::milliseconds(sleep_ms));
2047
0
        }
2048
48.4k
    } while (!coord_status.ok() && callback_retries-- > 0);
2049
2050
48.4k
    if (!coord_status.ok()) {
2051
0
        UniqueId uid(req.query_id.hi, req.query_id.lo);
2052
0
        static_cast<void>(req.cancel_fn(Status::InternalError(
2053
0
                "query_id: {}, couldn't get a client for {}, reason is {}", uid.to_string(),
2054
0
                PrintThriftNetworkAddress(req.coord_addr), coord_status.to_string())));
2055
0
        return;
2056
0
    }
2057
2058
48.4k
    TReportExecStatusParams params;
2059
48.4k
    params.protocol_version = FrontendServiceVersion::V1;
2060
48.4k
    params.__set_query_id(req.query_id);
2061
48.4k
    params.__set_backend_num(req.backend_num);
2062
48.4k
    params.__set_fragment_instance_id(req.fragment_instance_id);
2063
48.4k
    params.__set_fragment_id(req.fragment_id);
2064
48.4k
    params.__set_status(exec_status.to_thrift());
2065
48.4k
    params.__set_done(req.done);
2066
48.4k
    params.__set_query_type(req.runtime_state->query_type());
2067
48.4k
    params.__isset.profile = false;
2068
2069
48.4k
    DCHECK(req.runtime_state != nullptr);
2070
2071
48.4k
    if (req.runtime_state->query_type() == TQueryType::LOAD) {
2072
44.1k
        params.__set_loaded_rows(req.runtime_state->num_rows_load_total());
2073
44.1k
        params.__set_loaded_bytes(req.runtime_state->num_bytes_load_total());
2074
44.1k
    } else {
2075
4.37k
        DCHECK(!req.runtime_states.empty());
2076
4.37k
        if (!req.runtime_state->output_files().empty()) {
2077
0
            params.__isset.delta_urls = true;
2078
0
            for (auto& it : req.runtime_state->output_files()) {
2079
0
                params.delta_urls.push_back(_to_http_path(it));
2080
0
            }
2081
0
        }
2082
4.37k
        if (!params.delta_urls.empty()) {
2083
0
            params.__isset.delta_urls = true;
2084
0
        }
2085
4.37k
    }
2086
2087
48.4k
    static std::string s_dpp_normal_all = "dpp.norm.ALL";
2088
48.4k
    static std::string s_dpp_abnormal_all = "dpp.abnorm.ALL";
2089
48.4k
    static std::string s_unselected_rows = "unselected.rows";
2090
48.4k
    int64_t num_rows_load_success = 0;
2091
48.4k
    int64_t num_rows_load_filtered = 0;
2092
48.4k
    int64_t num_rows_load_unselected = 0;
2093
48.4k
    if (req.runtime_state->num_rows_load_total() > 0 ||
2094
48.4k
        req.runtime_state->num_rows_load_filtered() > 0 ||
2095
48.4k
        req.runtime_state->num_finished_range() > 0) {
2096
0
        params.__isset.load_counters = true;
2097
2098
0
        num_rows_load_success = req.runtime_state->num_rows_load_success();
2099
0
        num_rows_load_filtered = req.runtime_state->num_rows_load_filtered();
2100
0
        num_rows_load_unselected = req.runtime_state->num_rows_load_unselected();
2101
0
        params.__isset.fragment_instance_reports = true;
2102
0
        TFragmentInstanceReport t;
2103
0
        t.__set_fragment_instance_id(req.runtime_state->fragment_instance_id());
2104
0
        t.__set_num_finished_range(cast_set<int>(req.runtime_state->num_finished_range()));
2105
0
        t.__set_loaded_rows(req.runtime_state->num_rows_load_total());
2106
0
        t.__set_loaded_bytes(req.runtime_state->num_bytes_load_total());
2107
0
        params.fragment_instance_reports.push_back(t);
2108
48.4k
    } else if (!req.runtime_states.empty()) {
2109
143k
        for (auto* rs : req.runtime_states) {
2110
143k
            if (rs->num_rows_load_total() > 0 || rs->num_rows_load_filtered() > 0 ||
2111
143k
                rs->num_finished_range() > 0) {
2112
37.5k
                params.__isset.load_counters = true;
2113
37.5k
                num_rows_load_success += rs->num_rows_load_success();
2114
37.5k
                num_rows_load_filtered += rs->num_rows_load_filtered();
2115
37.5k
                num_rows_load_unselected += rs->num_rows_load_unselected();
2116
37.5k
                params.__isset.fragment_instance_reports = true;
2117
37.5k
                TFragmentInstanceReport t;
2118
37.5k
                t.__set_fragment_instance_id(rs->fragment_instance_id());
2119
37.5k
                t.__set_num_finished_range(cast_set<int>(rs->num_finished_range()));
2120
37.5k
                t.__set_loaded_rows(rs->num_rows_load_total());
2121
37.5k
                t.__set_loaded_bytes(rs->num_bytes_load_total());
2122
37.5k
                params.fragment_instance_reports.push_back(t);
2123
37.5k
            }
2124
143k
        }
2125
48.4k
    }
2126
48.4k
    params.load_counters.emplace(s_dpp_normal_all, std::to_string(num_rows_load_success));
2127
48.4k
    params.load_counters.emplace(s_dpp_abnormal_all, std::to_string(num_rows_load_filtered));
2128
48.4k
    params.load_counters.emplace(s_unselected_rows, std::to_string(num_rows_load_unselected));
2129
2130
48.4k
    if (!req.load_error_url.empty()) {
2131
160
        params.__set_tracking_url(req.load_error_url);
2132
160
    }
2133
48.4k
    if (!req.first_error_msg.empty()) {
2134
160
        params.__set_first_error_msg(req.first_error_msg);
2135
160
    }
2136
143k
    for (auto* rs : req.runtime_states) {
2137
143k
        if (rs->wal_id() > 0) {
2138
106
            params.__set_txn_id(rs->wal_id());
2139
106
            params.__set_label(rs->import_label());
2140
106
        }
2141
143k
    }
2142
48.4k
    if (!req.runtime_state->export_output_files().empty()) {
2143
0
        params.__isset.export_files = true;
2144
0
        params.export_files = req.runtime_state->export_output_files();
2145
48.4k
    } else if (!req.runtime_states.empty()) {
2146
143k
        for (auto* rs : req.runtime_states) {
2147
143k
            if (!rs->export_output_files().empty()) {
2148
0
                params.__isset.export_files = true;
2149
0
                params.export_files.insert(params.export_files.end(),
2150
0
                                           rs->export_output_files().begin(),
2151
0
                                           rs->export_output_files().end());
2152
0
            }
2153
143k
        }
2154
48.4k
    }
2155
48.4k
    if (auto tci = req.runtime_state->tablet_commit_infos(); !tci.empty()) {
2156
0
        params.__isset.commitInfos = true;
2157
0
        params.commitInfos.insert(params.commitInfos.end(), tci.begin(), tci.end());
2158
48.4k
    } else if (!req.runtime_states.empty()) {
2159
143k
        for (auto* rs : req.runtime_states) {
2160
143k
            if (auto rs_tci = rs->tablet_commit_infos(); !rs_tci.empty()) {
2161
27.7k
                params.__isset.commitInfos = true;
2162
27.7k
                params.commitInfos.insert(params.commitInfos.end(), rs_tci.begin(), rs_tci.end());
2163
27.7k
            }
2164
143k
        }
2165
48.4k
    }
2166
48.4k
    if (auto eti = req.runtime_state->error_tablet_infos(); !eti.empty()) {
2167
0
        params.__isset.errorTabletInfos = true;
2168
0
        params.errorTabletInfos.insert(params.errorTabletInfos.end(), eti.begin(), eti.end());
2169
48.4k
    } else if (!req.runtime_states.empty()) {
2170
143k
        for (auto* rs : req.runtime_states) {
2171
143k
            if (auto rs_eti = rs->error_tablet_infos(); !rs_eti.empty()) {
2172
0
                params.__isset.errorTabletInfos = true;
2173
0
                params.errorTabletInfos.insert(params.errorTabletInfos.end(), rs_eti.begin(),
2174
0
                                               rs_eti.end());
2175
0
            }
2176
143k
        }
2177
48.4k
    }
2178
48.4k
    if (auto hpu = req.runtime_state->hive_partition_updates(); !hpu.empty()) {
2179
0
        params.__isset.hive_partition_updates = true;
2180
0
        params.hive_partition_updates.insert(params.hive_partition_updates.end(), hpu.begin(),
2181
0
                                             hpu.end());
2182
48.4k
    } else if (!req.runtime_states.empty()) {
2183
143k
        for (auto* rs : req.runtime_states) {
2184
143k
            if (auto rs_hpu = rs->hive_partition_updates(); !rs_hpu.empty()) {
2185
2.13k
                params.__isset.hive_partition_updates = true;
2186
2.13k
                params.hive_partition_updates.insert(params.hive_partition_updates.end(),
2187
2.13k
                                                     rs_hpu.begin(), rs_hpu.end());
2188
2.13k
            }
2189
143k
        }
2190
48.4k
    }
2191
48.4k
    if (auto icd = req.runtime_state->iceberg_commit_datas(); !icd.empty()) {
2192
0
        params.__isset.iceberg_commit_datas = true;
2193
0
        params.iceberg_commit_datas.insert(params.iceberg_commit_datas.end(), icd.begin(),
2194
0
                                           icd.end());
2195
48.4k
    } else if (!req.runtime_states.empty()) {
2196
143k
        for (auto* rs : req.runtime_states) {
2197
143k
            if (auto rs_icd = rs->iceberg_commit_datas(); !rs_icd.empty()) {
2198
2.08k
                params.__isset.iceberg_commit_datas = true;
2199
2.08k
                params.iceberg_commit_datas.insert(params.iceberg_commit_datas.end(),
2200
2.08k
                                                   rs_icd.begin(), rs_icd.end());
2201
2.08k
            }
2202
143k
        }
2203
48.4k
    }
2204
2205
48.4k
    if (auto mcd = req.runtime_state->mc_commit_datas(); !mcd.empty()) {
2206
0
        params.__isset.mc_commit_datas = true;
2207
0
        params.mc_commit_datas.insert(params.mc_commit_datas.end(), mcd.begin(), mcd.end());
2208
48.4k
    } else if (!req.runtime_states.empty()) {
2209
143k
        for (auto* rs : req.runtime_states) {
2210
143k
            if (auto rs_mcd = rs->mc_commit_datas(); !rs_mcd.empty()) {
2211
0
                params.__isset.mc_commit_datas = true;
2212
0
                params.mc_commit_datas.insert(params.mc_commit_datas.end(), rs_mcd.begin(),
2213
0
                                              rs_mcd.end());
2214
0
            }
2215
143k
        }
2216
48.4k
    }
2217
2218
48.4k
    req.runtime_state->get_unreported_errors(&(params.error_log));
2219
48.4k
    params.__isset.error_log = (!params.error_log.empty());
2220
2221
48.4k
    if (_exec_env->cluster_info()->backend_id != 0) {
2222
48.4k
        params.__set_backend_id(_exec_env->cluster_info()->backend_id);
2223
48.4k
    }
2224
2225
48.4k
    TReportExecStatusResult res;
2226
48.4k
    Status rpc_status;
2227
2228
48.4k
    VLOG_DEBUG << "reportExecStatus params is "
2229
6
               << apache::thrift::ThriftDebugString(params).c_str();
2230
48.4k
    if (!exec_status.ok()) {
2231
1.70k
        LOG(WARNING) << "report error status: " << exec_status.msg()
2232
1.70k
                     << " to coordinator: " << req.coord_addr
2233
1.70k
                     << ", query id: " << print_id(req.query_id);
2234
1.70k
    }
2235
48.4k
    try {
2236
48.4k
        try {
2237
48.4k
            (*coord)->reportExecStatus(res, params);
2238
48.4k
        } catch ([[maybe_unused]] apache::thrift::transport::TTransportException& e) {
2239
#ifndef ADDRESS_SANITIZER
2240
            LOG(WARNING) << "Retrying ReportExecStatus. query id: " << print_id(req.query_id)
2241
                         << ", instance id: " << print_id(req.fragment_instance_id) << " to "
2242
                         << req.coord_addr << ", err: " << e.what();
2243
#endif
2244
0
            rpc_status = coord->reopen();
2245
2246
0
            if (!rpc_status.ok()) {
2247
0
                req.cancel_fn(rpc_status);
2248
0
                return;
2249
0
            }
2250
0
            (*coord)->reportExecStatus(res, params);
2251
0
        }
2252
2253
48.4k
        rpc_status = Status::create<false>(res.status);
2254
48.4k
    } catch (apache::thrift::TException& e) {
2255
0
        rpc_status = Status::InternalError("ReportExecStatus() to {} failed: {}",
2256
0
                                           PrintThriftNetworkAddress(req.coord_addr), e.what());
2257
0
    }
2258
2259
48.4k
    if (!rpc_status.ok()) {
2260
0
        LOG_INFO("Going to cancel query {} since report exec status got rpc failed: {}",
2261
0
                 print_id(req.query_id), rpc_status.to_string());
2262
0
        req.cancel_fn(rpc_status);
2263
0
    }
2264
48.4k
}
2265
2266
448k
Status PipelineFragmentContext::send_report(bool done) {
2267
448k
    Status exec_status = _query_ctx->exec_status();
2268
    // If plan is done successfully, but _is_report_success is false,
2269
    // no need to send report.
2270
    // Load will set _is_report_success to true because load wants to know
2271
    // the process.
2272
448k
    if (!_is_report_success && done && exec_status.ok()) {
2273
399k
        return Status::OK();
2274
399k
    }
2275
2276
    // If both _is_report_success and _is_report_on_cancel are false,
2277
    // which means no matter query is success or failed, no report is needed.
2278
    // This may happen when the query limit reached and
2279
    // a internal cancellation being processed
2280
    // When limit is reached the fragment is also cancelled, but _is_report_on_cancel will
2281
    // be set to false, to avoid sending fault report to FE.
2282
48.8k
    if (!_is_report_success && !_is_report_on_cancel) {
2283
356
        if (done) {
2284
            // if done is true, which means the query is finished successfully, we can safely close the fragment instance without sending report to FE, and just return OK status here.
2285
356
            return Status::OK();
2286
356
        }
2287
0
        return Status::NeedSendAgain("");
2288
356
    }
2289
2290
48.4k
    std::vector<RuntimeState*> runtime_states;
2291
2292
107k
    for (auto& tasks : _tasks) {
2293
143k
        for (auto& task : tasks) {
2294
143k
            runtime_states.push_back(task.second.get());
2295
143k
        }
2296
107k
    }
2297
2298
48.4k
    std::string load_eror_url = _query_ctx->get_load_error_url().empty()
2299
48.4k
                                        ? get_load_error_url()
2300
48.4k
                                        : _query_ctx->get_load_error_url();
2301
48.4k
    std::string first_error_msg = _query_ctx->get_first_error_msg().empty()
2302
48.4k
                                          ? get_first_error_msg()
2303
48.4k
                                          : _query_ctx->get_first_error_msg();
2304
2305
48.4k
    ReportStatusRequest req {.status = exec_status,
2306
48.4k
                             .runtime_states = runtime_states,
2307
48.4k
                             .done = done || !exec_status.ok(),
2308
48.4k
                             .coord_addr = _query_ctx->coord_addr,
2309
48.4k
                             .query_id = _query_id,
2310
48.4k
                             .fragment_id = _fragment_id,
2311
48.4k
                             .fragment_instance_id = TUniqueId(),
2312
48.4k
                             .backend_num = -1,
2313
48.4k
                             .runtime_state = _runtime_state.get(),
2314
48.4k
                             .load_error_url = load_eror_url,
2315
48.4k
                             .first_error_msg = first_error_msg,
2316
48.4k
                             .cancel_fn = [this](const Status& reason) { cancel(reason); }};
2317
48.4k
    auto ctx = std::dynamic_pointer_cast<PipelineFragmentContext>(shared_from_this());
2318
48.4k
    return _exec_env->fragment_mgr()->get_thread_pool()->submit_func([this, req, ctx]() {
2319
48.4k
        SCOPED_ATTACH_TASK(ctx->get_query_ctx()->query_mem_tracker());
2320
48.4k
        _coordinator_callback(req);
2321
48.4k
        if (!req.done) {
2322
4.92k
            ctx->refresh_next_report_time();
2323
4.92k
        }
2324
48.4k
    });
2325
48.8k
}
2326
2327
0
size_t PipelineFragmentContext::get_revocable_size(bool* has_running_task) const {
2328
0
    size_t res = 0;
2329
    // _tasks will be cleared during ~PipelineFragmentContext, so that it's safe
2330
    // here to traverse the vector.
2331
0
    for (const auto& task_instances : _tasks) {
2332
0
        for (const auto& task : task_instances) {
2333
0
            if (task.first->is_running()) {
2334
0
                LOG_EVERY_N(INFO, 50) << "Query: " << print_id(_query_id)
2335
0
                                      << " is running, task: " << (void*)task.first.get()
2336
0
                                      << ", is_running: " << task.first->is_running();
2337
0
                *has_running_task = true;
2338
0
                return 0;
2339
0
            }
2340
2341
0
            size_t revocable_size = task.first->get_revocable_size();
2342
0
            if (revocable_size >= SpillFile::MIN_SPILL_WRITE_BATCH_MEM) {
2343
0
                res += revocable_size;
2344
0
            }
2345
0
        }
2346
0
    }
2347
0
    return res;
2348
0
}
2349
2350
0
std::vector<PipelineTask*> PipelineFragmentContext::get_revocable_tasks() const {
2351
0
    std::vector<PipelineTask*> revocable_tasks;
2352
0
    for (const auto& task_instances : _tasks) {
2353
0
        for (const auto& task : task_instances) {
2354
0
            size_t revocable_size_ = task.first->get_revocable_size();
2355
2356
0
            if (revocable_size_ >= SpillFile::MIN_SPILL_WRITE_BATCH_MEM) {
2357
0
                revocable_tasks.emplace_back(task.first.get());
2358
0
            }
2359
0
        }
2360
0
    }
2361
0
    return revocable_tasks;
2362
0
}
2363
2364
150
std::string PipelineFragmentContext::debug_string() {
2365
150
    std::lock_guard<std::mutex> l(_task_mutex);
2366
150
    fmt::memory_buffer debug_string_buffer;
2367
150
    fmt::format_to(debug_string_buffer,
2368
150
                   "PipelineFragmentContext Info: _closed_tasks={}, _total_tasks={}, "
2369
150
                   "need_notify_close={}, fragment_id={}, _rec_cte_stage={}\n",
2370
150
                   _closed_tasks, _total_tasks, _need_notify_close, _fragment_id, _rec_cte_stage);
2371
730
    for (size_t j = 0; j < _tasks.size(); j++) {
2372
580
        fmt::format_to(debug_string_buffer, "Tasks in instance {}:\n", j);
2373
1.39k
        for (size_t i = 0; i < _tasks[j].size(); i++) {
2374
814
            fmt::format_to(debug_string_buffer, "Task {}: {}\n", i,
2375
814
                           _tasks[j][i].first->debug_string());
2376
814
        }
2377
580
    }
2378
2379
150
    return fmt::to_string(debug_string_buffer);
2380
150
}
2381
2382
std::vector<std::shared_ptr<TRuntimeProfileTree>>
2383
2.24k
PipelineFragmentContext::collect_realtime_profile() const {
2384
2.24k
    std::vector<std::shared_ptr<TRuntimeProfileTree>> res;
2385
2386
    // we do not have mutex to protect pipeline_id_to_profile
2387
    // so we need to make sure this funciton is invoked after fragment context
2388
    // has already been prepared.
2389
2.24k
    if (!_prepared) {
2390
0
        std::string msg =
2391
0
                "Query " + print_id(_query_id) + " collecting profile, but its not prepared";
2392
0
        DCHECK(false) << msg;
2393
0
        LOG_ERROR(msg);
2394
0
        return res;
2395
0
    }
2396
2397
    // Make sure first profile is fragment level profile
2398
2.24k
    auto fragment_profile = std::make_shared<TRuntimeProfileTree>();
2399
2.24k
    _fragment_level_profile->to_thrift(fragment_profile.get(), _runtime_state->profile_level());
2400
2.24k
    res.push_back(fragment_profile);
2401
2402
    // pipeline_id_to_profile is initialized in prepare stage
2403
4.09k
    for (auto pipeline_profile : _runtime_state->pipeline_id_to_profile()) {
2404
4.09k
        auto profile_ptr = std::make_shared<TRuntimeProfileTree>();
2405
4.09k
        pipeline_profile->to_thrift(profile_ptr.get(), _runtime_state->profile_level());
2406
4.09k
        res.push_back(profile_ptr);
2407
4.09k
    }
2408
2409
2.24k
    return res;
2410
2.24k
}
2411
2412
std::shared_ptr<TRuntimeProfileTree>
2413
2.24k
PipelineFragmentContext::collect_realtime_load_channel_profile() const {
2414
    // we do not have mutex to protect pipeline_id_to_profile
2415
    // so we need to make sure this funciton is invoked after fragment context
2416
    // has already been prepared.
2417
2.24k
    if (!_prepared) {
2418
0
        std::string msg =
2419
0
                "Query " + print_id(_query_id) + " collecting profile, but its not prepared";
2420
0
        DCHECK(false) << msg;
2421
0
        LOG_ERROR(msg);
2422
0
        return nullptr;
2423
0
    }
2424
2425
5.41k
    for (const auto& tasks : _tasks) {
2426
10.6k
        for (const auto& task : tasks) {
2427
10.6k
            if (task.second->load_channel_profile() == nullptr) {
2428
0
                continue;
2429
0
            }
2430
2431
10.6k
            auto tmp_load_channel_profile = std::make_shared<TRuntimeProfileTree>();
2432
2433
10.6k
            task.second->load_channel_profile()->to_thrift(tmp_load_channel_profile.get(),
2434
10.6k
                                                           _runtime_state->profile_level());
2435
10.6k
            _runtime_state->load_channel_profile()->update(*tmp_load_channel_profile);
2436
10.6k
        }
2437
5.41k
    }
2438
2439
2.24k
    auto load_channel_profile = std::make_shared<TRuntimeProfileTree>();
2440
2.24k
    _runtime_state->load_channel_profile()->to_thrift(load_channel_profile.get(),
2441
2.24k
                                                      _runtime_state->profile_level());
2442
2.24k
    return load_channel_profile;
2443
2.24k
}
2444
2445
// Collect runtime filter IDs registered by all tasks in this PFC.
2446
// Used during recursive CTE stage transitions to know which filters to deregister
2447
// before creating the new PFC for the next recursion round.
2448
// Called from rerun_fragment(wait_for_destroy) while tasks are still closing.
2449
// Thread safety: safe because _tasks is structurally immutable after prepare() —
2450
// the vector sizes do not change, and individual RuntimeState filter sets are
2451
// written only during open() which has completed by the time we reach rerun.
2452
3.28k
std::set<int> PipelineFragmentContext::get_deregister_runtime_filter() const {
2453
3.28k
    std::set<int> result;
2454
5.30k
    for (const auto& _task : _tasks) {
2455
9.28k
        for (const auto& task : _task) {
2456
9.28k
            auto set = task.first->runtime_state()->get_deregister_runtime_filter();
2457
9.28k
            result.merge(set);
2458
9.28k
        }
2459
5.30k
    }
2460
3.28k
    if (_runtime_state) {
2461
3.28k
        auto set = _runtime_state->get_deregister_runtime_filter();
2462
3.28k
        result.merge(set);
2463
3.28k
    }
2464
3.28k
    return result;
2465
3.28k
}
2466
2467
447k
void PipelineFragmentContext::_release_resource() {
2468
447k
    std::lock_guard<std::mutex> l(_task_mutex);
2469
    // The memory released by the query end is recorded in the query mem tracker.
2470
447k
    SCOPED_SWITCH_THREAD_MEM_TRACKER_LIMITER(_query_ctx->query_mem_tracker());
2471
447k
    auto st = _query_ctx->exec_status();
2472
1.26M
    for (auto& _task : _tasks) {
2473
1.26M
        if (!_task.empty()) {
2474
1.26M
            _call_back(_task.front().first->runtime_state(), &st);
2475
1.26M
        }
2476
1.26M
    }
2477
447k
    _tasks.clear();
2478
447k
    _dag.clear();
2479
447k
    _pip_id_to_pipeline.clear();
2480
447k
    _pipelines.clear();
2481
447k
    _sink.reset();
2482
447k
    _root_op.reset();
2483
447k
    _runtime_filter_mgr_map.clear();
2484
447k
    _op_id_to_shared_state.clear();
2485
447k
}
2486
2487
} // namespace doris