Coverage Report

Created: 2025-04-12 08:06

/root/doris/be/src/runtime/runtime_state.h
Line
Count
Source (jump to first uncovered line)
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
// This file is copied from
18
// https://github.com/apache/impala/blob/branch-2.9.0/be/src/runtime/runtime-state.h
19
// and modified by Doris
20
21
#pragma once
22
23
#include <gen_cpp/PaloInternalService_types.h>
24
#include <gen_cpp/Types_types.h>
25
#include <gen_cpp/segment_v2.pb.h>
26
#include <stdint.h>
27
28
#include <atomic>
29
#include <cstdint>
30
#include <fstream>
31
#include <functional>
32
#include <memory>
33
#include <mutex>
34
#include <shared_mutex>
35
#include <string>
36
#include <utility>
37
#include <vector>
38
39
#include "agent/be_exec_version_manager.h"
40
#include "cctz/time_zone.h"
41
#include "common/be_mock_util.h"
42
#include "common/compiler_util.h" // IWYU pragma: keep
43
#include "common/config.h"
44
#include "common/factory_creator.h"
45
#include "common/status.h"
46
#include "io/fs/file_system.h"
47
#include "io/fs/s3_file_system.h"
48
#include "runtime/task_execution_context.h"
49
#include "runtime/workload_group/workload_group.h"
50
#include "util/debug_util.h"
51
#include "util/runtime_profile.h"
52
#include "vec/columns/columns_number.h"
53
54
namespace doris {
55
class RuntimeFilter;
56
57
23
inline int32_t get_execution_rpc_timeout_ms(int32_t execution_timeout_sec) {
58
23
    return std::min(config::execution_max_rpc_timeout_sec, execution_timeout_sec) * 1000;
59
23
}
60
61
namespace pipeline {
62
class PipelineXLocalStateBase;
63
class PipelineXSinkLocalStateBase;
64
class PipelineFragmentContext;
65
class PipelineTask;
66
class Dependency;
67
} // namespace pipeline
68
69
class DescriptorTbl;
70
class ObjectPool;
71
class ExecEnv;
72
class RuntimeFilterMgr;
73
class MemTrackerLimiter;
74
class QueryContext;
75
class RuntimeFilterConsumer;
76
class RuntimeFilterProducer;
77
78
// A collection of items that are part of the global state of a
79
// query and shared across all execution nodes of that query.
80
class RuntimeState {
81
    ENABLE_FACTORY_CREATOR(RuntimeState);
82
83
public:
84
    RuntimeState(const TPlanFragmentExecParams& fragment_exec_params,
85
                 const TQueryOptions& query_options, const TQueryGlobals& query_globals,
86
                 ExecEnv* exec_env, QueryContext* ctx,
87
                 const std::shared_ptr<MemTrackerLimiter>& query_mem_tracker = nullptr);
88
89
    RuntimeState(const TUniqueId& instance_id, const TUniqueId& query_id, int32 fragment_id,
90
                 const TQueryOptions& query_options, const TQueryGlobals& query_globals,
91
                 ExecEnv* exec_env, QueryContext* ctx);
92
93
    // Used by pipeline. This runtime state is only used for setup.
94
    RuntimeState(const TUniqueId& query_id, int32 fragment_id, const TQueryOptions& query_options,
95
                 const TQueryGlobals& query_globals, ExecEnv* exec_env, QueryContext* ctx);
96
97
    // RuntimeState for executing expr in fe-support.
98
    RuntimeState(const TQueryGlobals& query_globals);
99
100
    // for job task only
101
    RuntimeState();
102
103
    // Empty d'tor to avoid issues with unique_ptr.
104
    MOCK_DEFINE(virtual) ~RuntimeState();
105
106
    // Set per-query state.
107
    Status init(const TUniqueId& fragment_instance_id, const TQueryOptions& query_options,
108
                const TQueryGlobals& query_globals, ExecEnv* exec_env);
109
110
    // for ut and non-query.
111
0
    void set_exec_env(ExecEnv* exec_env) { _exec_env = exec_env; }
112
113
    // for ut and non-query.
114
    void init_mem_trackers(const std::string& name = "ut", const TUniqueId& id = TUniqueId());
115
116
307
    const TQueryOptions& query_options() const { return _query_options; }
117
8
    int64_t scan_queue_mem_limit() const {
118
8
        return _query_options.__isset.scan_queue_mem_limit ? _query_options.scan_queue_mem_limit
119
8
                                                           : _query_options.mem_limit / 20;
120
8
    }
121
122
6
    int32_t max_column_reader_num() const {
123
6
        return _query_options.__isset.max_column_reader_num ? _query_options.max_column_reader_num
124
6
                                                            : 20000;
125
6
    }
126
127
215
    ObjectPool* obj_pool() const { return _obj_pool.get(); }
128
129
392
    const DescriptorTbl& desc_tbl() const { return *_desc_tbl; }
130
207
    void set_desc_tbl(const DescriptorTbl* desc_tbl) { _desc_tbl = desc_tbl; }
131
93
    MOCK_FUNCTION int batch_size() const { return _query_options.batch_size; }
132
8
    int query_parallel_instance_num() const { return _query_options.parallel_instance; }
133
0
    int max_errors() const { return _query_options.max_errors; }
134
28
    int execution_timeout() const {
135
28
        return _query_options.__isset.execution_timeout ? _query_options.execution_timeout
136
28
                                                        : _query_options.query_timeout;
137
28
    }
138
6
    int num_scanner_threads() const {
139
6
        return _query_options.__isset.num_scanner_threads ? _query_options.num_scanner_threads : 0;
140
6
    }
141
14
    int min_scan_concurrency_of_scan_scheduler() const {
142
14
        return _query_options.__isset.min_scan_scheduler_concurrency
143
14
                       ? _query_options.min_scan_scheduler_concurrency
144
14
                       : 0;
145
14
    }
146
147
14
    int min_scan_concurrency_of_scanner() const {
148
14
        return _query_options.__isset.min_scanner_concurrency
149
14
                       ? _query_options.min_scanner_concurrency
150
14
                       : 1;
151
14
    }
152
153
0
    TQueryType::type query_type() const { return _query_options.query_type; }
154
6
    int64_t timestamp_ms() const { return _timestamp_ms; }
155
0
    int32_t nano_seconds() const { return _nano_seconds; }
156
    // if possible, use timezone_obj() rather than timezone()
157
13
    const std::string& timezone() const { return _timezone; }
158
33
    const cctz::time_zone& timezone_obj() const { return _timezone_obj; }
159
0
    const std::string& user() const { return _user; }
160
473
    const TUniqueId& query_id() const { return _query_id; }
161
22.4k
    const TUniqueId& fragment_instance_id() const { return _fragment_instance_id; }
162
    // should only be called in pipeline engine
163
42
    int32_t fragment_id() const { return _fragment_id; }
164
67
    ExecEnv* exec_env() { return _exec_env; }
165
    std::shared_ptr<MemTrackerLimiter> query_mem_tracker() const;
166
167
    // Returns runtime state profile
168
4
    RuntimeProfile* runtime_profile() { return &_profile; }
169
0
    RuntimeProfile* load_channel_profile() { return &_load_channel_profile; }
170
171
0
    bool enable_function_pushdown() const {
172
0
        return _query_options.__isset.enable_function_pushdown &&
173
0
               _query_options.enable_function_pushdown;
174
0
    }
175
176
53
    bool check_overflow_for_decimal() const {
177
53
        return _query_options.__isset.check_overflow_for_decimal &&
178
53
               _query_options.check_overflow_for_decimal;
179
53
    }
180
181
75
    bool enable_decimal256() const {
182
75
        return _query_options.__isset.enable_decimal256 && _query_options.enable_decimal256;
183
75
    }
184
185
0
    bool enable_common_expr_pushdown() const {
186
0
        return _query_options.__isset.enable_common_expr_pushdown &&
187
0
               _query_options.enable_common_expr_pushdown;
188
0
    }
189
190
0
    bool enable_common_expr_pushdown_for_inverted_index() const {
191
0
        return enable_common_expr_pushdown() &&
192
0
               _query_options.__isset.enable_common_expr_pushdown_for_inverted_index &&
193
0
               _query_options.enable_common_expr_pushdown_for_inverted_index;
194
0
    };
195
196
0
    bool mysql_row_binary_format() const {
197
0
        return _query_options.__isset.mysql_row_binary_format &&
198
0
               _query_options.mysql_row_binary_format;
199
0
    }
200
201
0
    bool enable_short_circuit_query_access_column_store() const {
202
0
        return _query_options.__isset.enable_short_circuit_query_access_column_store &&
203
0
               _query_options.enable_short_circuit_query_access_column_store;
204
0
    }
205
206
    // Appends error to the _error_log if there is space
207
    bool log_error(const std::string& error);
208
209
    // Returns true if the error log has not reached _max_errors.
210
0
    bool log_has_space() {
211
0
        std::lock_guard<std::mutex> l(_error_log_lock);
212
0
        return _error_log.size() < _query_options.max_errors;
213
0
    }
214
215
    // Append all _error_log[_unreported_error_idx+] to new_errors and set
216
    // _unreported_error_idx to _errors_log.size()
217
    void get_unreported_errors(std::vector<std::string>* new_errors);
218
219
    [[nodiscard]] MOCK_FUNCTION bool is_cancelled() const;
220
    MOCK_FUNCTION Status cancel_reason() const;
221
0
    void cancel(const Status& reason) {
222
0
        if (_exec_status.update(reason)) {
223
            // Create a error status, so that we could print error stack, and
224
            // we could know which path call cancel.
225
0
            LOG(WARNING) << "Task is cancelled, instance: "
226
0
                         << PrintInstanceStandardInfo(_query_id, _fragment_instance_id)
227
0
                         << ", st = " << reason;
228
0
        } else {
229
0
            LOG(WARNING) << "Task is already cancelled, instance: "
230
0
                         << PrintInstanceStandardInfo(_query_id, _fragment_instance_id)
231
0
                         << ", original cancel msg: " << _exec_status.status()
232
0
                         << ", new cancel msg: " << reason;
233
0
        }
234
0
    }
235
236
0
    void set_backend_id(int64_t backend_id) { _backend_id = backend_id; }
237
0
    int64_t backend_id() const { return _backend_id; }
238
239
46
    void set_be_number(int be_number) { _be_number = be_number; }
240
93
    int be_number(void) const { return _be_number; }
241
242
0
    std::vector<std::string>& output_files() { return _output_files; }
243
244
0
    void set_import_label(const std::string& import_label) { _import_label = import_label; }
245
246
0
    const std::vector<std::string>& export_output_files() const { return _export_output_files; }
247
248
0
    void add_export_output_file(const std::string& file) { _export_output_files.push_back(file); }
249
250
0
    void set_db_name(const std::string& db_name) { _db_name = db_name; }
251
252
0
    const std::string& db_name() { return _db_name; }
253
254
0
    void set_wal_id(int64_t wal_id) { _wal_id = wal_id; }
255
256
3
    int64_t wal_id() const { return _wal_id; }
257
258
0
    void set_content_length(size_t content_length) { _content_length = content_length; }
259
260
0
    size_t content_length() const { return _content_length; }
261
262
0
    const std::string& import_label() { return _import_label; }
263
264
0
    const std::string& load_dir() const { return _load_dir; }
265
266
0
    void set_load_job_id(int64_t job_id) { _load_job_id = job_id; }
267
268
0
    int64_t load_job_id() const { return _load_job_id; }
269
270
    std::string get_error_log_file_path();
271
272
    // append error msg and error line to file when loading data.
273
    // is_summary is true, means we are going to write the summary line
274
    // If we need to stop the processing, set stop_processing to true
275
    Status append_error_msg_to_file(std::function<std::string()> line,
276
                                    std::function<std::string()> error_msg,
277
                                    bool is_summary = false);
278
279
0
    int64_t num_bytes_load_total() { return _num_bytes_load_total.load(); }
280
281
0
    int64_t num_finished_range() { return _num_finished_scan_range.load(); }
282
283
0
    int64_t num_rows_load_total() { return _num_rows_load_total.load(); }
284
285
0
    int64_t num_rows_load_filtered() { return _num_rows_load_filtered.load(); }
286
287
0
    int64_t num_rows_load_unselected() { return _num_rows_load_unselected.load(); }
288
289
0
    int64_t num_rows_filtered_in_strict_mode_partial_update() {
290
0
        return _num_rows_filtered_in_strict_mode_partial_update;
291
0
    }
292
293
0
    int64_t num_rows_load_success() {
294
0
        return num_rows_load_total() - num_rows_load_filtered() - num_rows_load_unselected();
295
0
    }
296
297
0
    void update_num_rows_load_total(int64_t num_rows) { _num_rows_load_total.fetch_add(num_rows); }
298
299
0
    void set_num_rows_load_total(int64_t num_rows) { _num_rows_load_total.store(num_rows); }
300
301
0
    void update_num_bytes_load_total(int64_t bytes_load) {
302
0
        _num_bytes_load_total.fetch_add(bytes_load);
303
0
    }
304
305
2
    void update_num_finished_scan_range(int64_t finished_range) {
306
2
        _num_finished_scan_range.fetch_add(finished_range);
307
2
    }
308
309
0
    void update_num_rows_load_filtered(int64_t num_rows) {
310
0
        _num_rows_load_filtered.fetch_add(num_rows);
311
0
    }
312
313
0
    void update_num_rows_load_unselected(int64_t num_rows) {
314
0
        _num_rows_load_unselected.fetch_add(num_rows);
315
0
    }
316
317
0
    void set_num_rows_filtered_in_strict_mode_partial_update(int64_t num_rows) {
318
0
        _num_rows_filtered_in_strict_mode_partial_update = num_rows;
319
0
    }
320
321
5
    void set_per_fragment_instance_idx(int idx) { _per_fragment_instance_idx = idx; }
322
323
15
    int per_fragment_instance_idx() const { return _per_fragment_instance_idx; }
324
325
5
    void set_num_per_fragment_instances(int num_instances) {
326
5
        _num_per_fragment_instances = num_instances;
327
5
    }
328
329
0
    int num_per_fragment_instances() const { return _num_per_fragment_instances; }
330
331
5
    void set_load_stream_per_node(int load_stream_per_node) {
332
5
        _load_stream_per_node = load_stream_per_node;
333
5
    }
334
335
0
    int load_stream_per_node() const { return _load_stream_per_node; }
336
337
5
    void set_total_load_streams(int total_load_streams) {
338
5
        _total_load_streams = total_load_streams;
339
5
    }
340
341
0
    int total_load_streams() const { return _total_load_streams; }
342
343
5
    void set_num_local_sink(int num_local_sink) { _num_local_sink = num_local_sink; }
344
345
0
    int num_local_sink() const { return _num_local_sink; }
346
347
0
    bool disable_stream_preaggregations() const {
348
0
        return _query_options.disable_stream_preaggregations;
349
0
    }
350
351
0
    int32_t runtime_filter_wait_time_ms() const {
352
0
        return _query_options.runtime_filter_wait_time_ms;
353
0
    }
354
355
0
    int32_t runtime_filter_max_in_num() const { return _query_options.runtime_filter_max_in_num; }
356
357
96
    int be_exec_version() const {
358
96
        DCHECK(_query_options.__isset.be_exec_version &&
359
96
               BeExecVersionManager::check_be_exec_version(_query_options.be_exec_version));
360
96
        return _query_options.be_exec_version;
361
96
    }
362
2
    bool enable_local_shuffle() const {
363
2
        return _query_options.__isset.enable_local_shuffle && _query_options.enable_local_shuffle;
364
2
    }
365
366
3
    MOCK_FUNCTION bool enable_local_exchange() const {
367
3
        return _query_options.__isset.enable_local_exchange && _query_options.enable_local_exchange;
368
3
    }
369
370
0
    bool trim_tailing_spaces_for_external_table_query() const {
371
0
        return _query_options.trim_tailing_spaces_for_external_table_query;
372
0
    }
373
374
0
    bool return_object_data_as_binary() const {
375
0
        return _query_options.return_object_data_as_binary;
376
0
    }
377
378
19
    segment_v2::CompressionTypePB fragement_transmission_compression_type() const {
379
19
        if (_query_options.__isset.fragment_transmission_compression_codec) {
380
0
            if (_query_options.fragment_transmission_compression_codec == "lz4") {
381
0
                return segment_v2::CompressionTypePB::LZ4;
382
0
            } else if (_query_options.fragment_transmission_compression_codec == "snappy") {
383
0
                return segment_v2::CompressionTypePB::SNAPPY;
384
0
            } else {
385
0
                return segment_v2::CompressionTypePB::NO_COMPRESSION;
386
0
            }
387
0
        }
388
19
        return segment_v2::CompressionTypePB::NO_COMPRESSION;
389
19
    }
390
391
0
    bool skip_storage_engine_merge() const {
392
0
        return _query_options.__isset.skip_storage_engine_merge &&
393
0
               _query_options.skip_storage_engine_merge;
394
0
    }
395
396
0
    bool skip_delete_predicate() const {
397
0
        return _query_options.__isset.skip_delete_predicate && _query_options.skip_delete_predicate;
398
0
    }
399
400
0
    bool skip_delete_bitmap() const {
401
0
        return _query_options.__isset.skip_delete_bitmap && _query_options.skip_delete_bitmap;
402
0
    }
403
404
0
    bool skip_missing_version() const {
405
0
        return _query_options.__isset.skip_missing_version && _query_options.skip_missing_version;
406
0
    }
407
408
3
    int64_t data_queue_max_blocks() const {
409
3
        return _query_options.__isset.data_queue_max_blocks ? _query_options.data_queue_max_blocks
410
3
                                                            : 1;
411
3
    }
412
413
    bool enable_page_cache() const;
414
415
0
    std::vector<TTabletCommitInfo> tablet_commit_infos() const {
416
0
        std::lock_guard<std::mutex> lock(_tablet_infos_mutex);
417
0
        return _tablet_commit_infos;
418
0
    }
419
420
0
    void add_tablet_commit_infos(std::vector<TTabletCommitInfo>& commit_infos) {
421
0
        std::lock_guard<std::mutex> lock(_tablet_infos_mutex);
422
0
        _tablet_commit_infos.insert(_tablet_commit_infos.end(),
423
0
                                    std::make_move_iterator(commit_infos.begin()),
424
0
                                    std::make_move_iterator(commit_infos.end()));
425
0
    }
426
427
0
    std::vector<TErrorTabletInfo> error_tablet_infos() const {
428
0
        std::lock_guard<std::mutex> lock(_tablet_infos_mutex);
429
0
        return _error_tablet_infos;
430
0
    }
431
432
0
    void add_error_tablet_infos(std::vector<TErrorTabletInfo>& tablet_infos) {
433
0
        std::lock_guard<std::mutex> lock(_tablet_infos_mutex);
434
0
        _error_tablet_infos.insert(_error_tablet_infos.end(),
435
0
                                   std::make_move_iterator(tablet_infos.begin()),
436
0
                                   std::make_move_iterator(tablet_infos.end()));
437
0
    }
438
439
0
    std::vector<THivePartitionUpdate> hive_partition_updates() const {
440
0
        std::lock_guard<std::mutex> lock(_hive_partition_updates_mutex);
441
0
        return _hive_partition_updates;
442
0
    }
443
444
0
    void add_hive_partition_updates(const THivePartitionUpdate& hive_partition_update) {
445
0
        std::lock_guard<std::mutex> lock(_hive_partition_updates_mutex);
446
0
        _hive_partition_updates.emplace_back(hive_partition_update);
447
0
    }
448
449
0
    std::vector<TIcebergCommitData> iceberg_commit_datas() const {
450
0
        std::lock_guard<std::mutex> lock(_iceberg_commit_datas_mutex);
451
0
        return _iceberg_commit_datas;
452
0
    }
453
454
0
    void add_iceberg_commit_datas(const TIcebergCommitData& iceberg_commit_data) {
455
0
        std::lock_guard<std::mutex> lock(_iceberg_commit_datas_mutex);
456
0
        _iceberg_commit_datas.emplace_back(iceberg_commit_data);
457
0
    }
458
459
    // local runtime filter mgr, the runtime filter do not have remote target or
460
    // not need local merge should regist here. the instance exec finish, the local
461
    // runtime filter mgr can release the memory of local runtime filter
462
79
    RuntimeFilterMgr* local_runtime_filter_mgr() { return _runtime_filter_mgr; }
463
464
    RuntimeFilterMgr* global_runtime_filter_mgr();
465
466
118
    void set_runtime_filter_mgr(RuntimeFilterMgr* runtime_filter_mgr) {
467
118
        _runtime_filter_mgr = runtime_filter_mgr;
468
118
    }
469
470
672k
    QueryContext* get_query_ctx() { return _query_ctx; }
471
472
    [[nodiscard]] bool low_memory_mode() const;
473
474
    std::weak_ptr<QueryContext> get_query_ctx_weak();
475
    WorkloadGroupPtr workload_group();
476
477
0
    void set_query_mem_tracker(const std::shared_ptr<MemTrackerLimiter>& tracker) {
478
0
        _query_mem_tracker = tracker;
479
0
    }
480
481
14
    void set_query_options(const TQueryOptions& query_options) { _query_options = query_options; }
482
483
0
    bool enable_profile() const {
484
0
        return _query_options.__isset.enable_profile && _query_options.enable_profile;
485
0
    }
486
487
0
    int rpc_verbose_profile_max_instance_count() const {
488
0
        return _query_options.__isset.rpc_verbose_profile_max_instance_count
489
0
                       ? _query_options.rpc_verbose_profile_max_instance_count
490
0
                       : 0;
491
0
    }
492
493
0
    bool enable_share_hash_table_for_broadcast_join() const {
494
0
        return _query_options.__isset.enable_share_hash_table_for_broadcast_join &&
495
0
               _query_options.enable_share_hash_table_for_broadcast_join;
496
0
    }
497
498
0
    bool enable_parallel_scan() const {
499
0
        return _query_options.__isset.enable_parallel_scan && _query_options.enable_parallel_scan;
500
0
    }
501
502
0
    bool is_read_csv_empty_line_as_null() const {
503
0
        return _query_options.__isset.read_csv_empty_line_as_null &&
504
0
               _query_options.read_csv_empty_line_as_null;
505
0
    }
506
507
4
    int parallel_scan_max_scanners_count() const {
508
4
        return _query_options.__isset.parallel_scan_max_scanners_count
509
4
                       ? _query_options.parallel_scan_max_scanners_count
510
4
                       : 0;
511
4
    }
512
513
400
    int partition_topn_max_partitions() const {
514
400
        return _query_options.__isset.partition_topn_max_partitions
515
400
                       ? _query_options.partition_topn_max_partitions
516
400
                       : 1024;
517
400
    }
518
519
0
    int partition_topn_per_partition_rows() const {
520
0
        return _query_options.__isset.partition_topn_pre_partition_rows
521
0
                       ? _query_options.partition_topn_pre_partition_rows
522
0
                       : 1000;
523
0
    }
524
525
0
    int64_t parallel_scan_min_rows_per_scanner() const {
526
0
        return _query_options.__isset.parallel_scan_min_rows_per_scanner
527
0
                       ? _query_options.parallel_scan_min_rows_per_scanner
528
0
                       : 0;
529
0
    }
530
531
0
    void set_be_exec_version(int32_t version) noexcept { _query_options.be_exec_version = version; }
532
533
    using LocalState = doris::pipeline::PipelineXLocalStateBase;
534
    using SinkLocalState = doris::pipeline::PipelineXSinkLocalStateBase;
535
    // get result can return an error message, and we will only call it during the prepare.
536
    void emplace_local_state(int id, std::unique_ptr<LocalState> state);
537
538
    LocalState* get_local_state(int id);
539
    Result<LocalState*> get_local_state_result(int id);
540
541
    void emplace_sink_local_state(int id, std::unique_ptr<SinkLocalState> state);
542
543
    SinkLocalState* get_sink_local_state();
544
545
    Result<SinkLocalState*> get_sink_local_state_result();
546
547
    void resize_op_id_to_local_state(int operator_size);
548
549
    std::vector<std::shared_ptr<RuntimeProfile>> pipeline_id_to_profile();
550
551
    std::vector<std::shared_ptr<RuntimeProfile>> build_pipeline_profile(std::size_t pipeline_size);
552
553
374
    void set_task_execution_context(std::shared_ptr<TaskExecutionContext> context) {
554
374
        _task_execution_context_inited = true;
555
374
        _task_execution_context = context;
556
374
    }
557
558
128
    std::weak_ptr<TaskExecutionContext> get_task_execution_context() {
559
128
        CHECK(_task_execution_context_inited)
560
0
                << "_task_execution_context_inited == false, the ctx is not inited";
561
128
        return _task_execution_context;
562
128
    }
563
564
    Status register_producer_runtime_filter(const doris::TRuntimeFilterDesc& desc,
565
                                            std::shared_ptr<RuntimeFilterProducer>* producer_filter,
566
                                            RuntimeProfile* parent_profile);
567
568
    Status register_consumer_runtime_filter(
569
            const doris::TRuntimeFilterDesc& desc, bool need_local_merge, int node_id,
570
            std::shared_ptr<RuntimeFilterConsumer>* consumer_filter);
571
572
    bool is_nereids() const;
573
574
88
    bool enable_spill() const {
575
88
        return (_query_options.__isset.enable_force_spill && _query_options.enable_force_spill) ||
576
88
               (_query_options.__isset.enable_spill && _query_options.enable_spill);
577
88
    }
578
579
0
    bool enable_force_spill() const {
580
0
        return _query_options.__isset.enable_force_spill && _query_options.enable_force_spill;
581
0
    }
582
583
0
    int64_t spill_min_revocable_mem() const {
584
0
        if (_query_options.__isset.min_revocable_mem) {
585
0
            return std::max(_query_options.min_revocable_mem, (int64_t)1);
586
0
        }
587
0
        return 1;
588
0
    }
589
590
5
    int64_t spill_sort_mem_limit() const {
591
5
        if (_query_options.__isset.spill_sort_mem_limit) {
592
5
            return std::max(_query_options.spill_sort_mem_limit, (int64_t)16777216);
593
5
        }
594
0
        return 134217728;
595
5
    }
596
597
15
    int64_t spill_sort_batch_bytes() const {
598
15
        if (_query_options.__isset.spill_sort_batch_bytes) {
599
15
            return std::max(_query_options.spill_sort_batch_bytes, (int64_t)8388608);
600
15
        }
601
0
        return 8388608;
602
15
    }
603
604
11
    int spill_aggregation_partition_count() const {
605
11
        if (_query_options.__isset.spill_aggregation_partition_count) {
606
11
            return std::min(std::max(_query_options.spill_aggregation_partition_count, 16), 8192);
607
11
        }
608
0
        return 32;
609
11
    }
610
611
0
    int spill_hash_join_partition_count() const {
612
0
        if (_query_options.__isset.spill_hash_join_partition_count) {
613
0
            return std::min(std::max(_query_options.spill_hash_join_partition_count, 16), 8192);
614
0
        }
615
0
        return 32;
616
0
    }
617
618
0
    int64_t low_memory_mode_buffer_limit() const {
619
0
        if (_query_options.__isset.low_memory_mode_buffer_limit) {
620
0
            return std::max(_query_options.low_memory_mode_buffer_limit, (int64_t)1);
621
0
        }
622
0
        return 32L * 1024 * 1024;
623
0
    }
624
625
1
    int spill_revocable_memory_high_watermark_percent() const {
626
1
        if (_query_options.__isset.revocable_memory_high_watermark_percent) {
627
1
            return _query_options.revocable_memory_high_watermark_percent;
628
1
        }
629
0
        return -1;
630
1
    }
631
632
0
    MOCK_FUNCTION bool enable_shared_exchange_sink_buffer() const {
633
0
        return _query_options.__isset.enable_shared_exchange_sink_buffer &&
634
0
               _query_options.enable_shared_exchange_sink_buffer;
635
0
    }
636
637
1.34M
    size_t minimum_operator_memory_required_bytes() const {
638
1.34M
        if (_query_options.__isset.minimum_operator_memory_required_kb) {
639
1.34M
            return _query_options.minimum_operator_memory_required_kb * 1024;
640
1.34M
        } else {
641
            // refer other database
642
0
            return 100 * 1024;
643
0
        }
644
1.34M
    }
645
646
60
    void set_max_operator_id(int max_operator_id) { _max_operator_id = max_operator_id; }
647
648
10
    int max_operator_id() const { return _max_operator_id; }
649
650
16
    void set_task_id(int id) { _task_id = id; }
651
652
177
    int task_id() const { return _task_id; }
653
654
5
    void set_task_num(int task_num) { _task_num = task_num; }
655
656
0
    int task_num() const { return _task_num; }
657
658
0
    int profile_level() const { return _profile_level; }
659
660
private:
661
    Status create_error_log_file();
662
663
    static const int DEFAULT_BATCH_SIZE = 4062;
664
665
    std::shared_ptr<MemTrackerLimiter> _query_mem_tracker = nullptr;
666
667
    // Could not find a better way to record if the weak ptr is inited, use a bool to record
668
    // it. In some unit test cases, the runtime state's task ctx is not inited, then the test
669
    // hang, it is very hard to debug.
670
    bool _task_execution_context_inited = false;
671
    // Hold execution context for other threads
672
    std::weak_ptr<TaskExecutionContext> _task_execution_context;
673
674
    // put runtime state before _obj_pool, so that it will be deconstructed after
675
    // _obj_pool. Because some of object in _obj_pool will use profile when deconstructing.
676
    RuntimeProfile _profile;
677
    RuntimeProfile _load_channel_profile;
678
    // Why 2?
679
    // During cluster upgrade, fe will not pass profile_level to be, so we need to set it to 2
680
    // to make sure user can see all profile counters like before.
681
    int _profile_level = 2;
682
683
    const DescriptorTbl* _desc_tbl = nullptr;
684
    std::shared_ptr<ObjectPool> _obj_pool;
685
686
    // owned by PipelineFragmentContext
687
    RuntimeFilterMgr* _runtime_filter_mgr = nullptr;
688
689
    // Lock protecting _error_log and _unreported_error_idx
690
    std::mutex _error_log_lock;
691
692
    // Logs error messages.
693
    std::vector<std::string> _error_log;
694
695
    // _error_log[_unreported_error_idx+] has been not reported to the coordinator.
696
    int _unreported_error_idx;
697
698
    // Username of user that is executing the query to which this RuntimeState belongs.
699
    std::string _user;
700
701
    //Query-global timestamp_ms
702
    int64_t _timestamp_ms;
703
    int32_t _nano_seconds;
704
    std::string _timezone;
705
    cctz::time_zone _timezone_obj;
706
707
    TUniqueId _query_id;
708
    // fragment id for each TPipelineFragmentParams
709
    int32_t _fragment_id;
710
    TUniqueId _fragment_instance_id;
711
    TQueryOptions _query_options;
712
    ExecEnv* _exec_env = nullptr;
713
714
    AtomicStatus _exec_status;
715
716
    int _per_fragment_instance_idx;
717
    int _num_per_fragment_instances = 0;
718
    int _load_stream_per_node = 0;
719
    int _total_load_streams = 0;
720
    int _num_local_sink = 0;
721
722
    // The backend id on which this fragment instance runs
723
    int64_t _backend_id = -1;
724
725
    // used as send id
726
    int _be_number;
727
728
    // put here to collect files??
729
    std::vector<std::string> _output_files;
730
    std::atomic<int64_t> _num_rows_load_total;      // total rows read from source
731
    std::atomic<int64_t> _num_rows_load_filtered;   // unqualified rows
732
    std::atomic<int64_t> _num_rows_load_unselected; // rows filtered by predicates
733
    std::atomic<int64_t> _num_rows_filtered_in_strict_mode_partial_update;
734
    std::atomic<int64_t> _num_print_error_rows;
735
736
    std::atomic<int64_t> _num_bytes_load_total; // total bytes read from source
737
    std::atomic<int64_t> _num_finished_scan_range;
738
739
    std::vector<std::string> _export_output_files;
740
    std::string _import_label;
741
    std::string _db_name;
742
    std::string _load_dir;
743
    int64_t _load_job_id;
744
    int64_t _wal_id = -1;
745
    size_t _content_length = 0;
746
747
    // mini load
748
    int64_t _error_row_number;
749
    std::string _error_log_file_path;
750
    std::unique_ptr<std::ofstream> _error_log_file; // error file path, absolute path
751
    mutable std::mutex _tablet_infos_mutex;
752
    std::vector<TTabletCommitInfo> _tablet_commit_infos;
753
    std::vector<TErrorTabletInfo> _error_tablet_infos;
754
    int _max_operator_id = 0;
755
    pipeline::PipelineTask* _task = nullptr;
756
    int _task_id = -1;
757
    int _task_num = 0;
758
759
    mutable std::mutex _hive_partition_updates_mutex;
760
    std::vector<THivePartitionUpdate> _hive_partition_updates;
761
762
    mutable std::mutex _iceberg_commit_datas_mutex;
763
    std::vector<TIcebergCommitData> _iceberg_commit_datas;
764
765
    std::vector<std::unique_ptr<doris::pipeline::PipelineXLocalStateBase>> _op_id_to_local_state;
766
767
    std::unique_ptr<doris::pipeline::PipelineXSinkLocalStateBase> _sink_local_state;
768
769
    QueryContext* _query_ctx = nullptr;
770
771
    // true if max_filter_ratio is 0
772
    bool _load_zero_tolerance = false;
773
774
    // only to lock _pipeline_id_to_profile
775
    std::shared_mutex _pipeline_profile_lock;
776
    std::vector<std::shared_ptr<RuntimeProfile>> _pipeline_id_to_profile;
777
778
    // prohibit copies
779
    RuntimeState(const RuntimeState&);
780
781
    // save error log to s3
782
    std::shared_ptr<io::S3FileSystem> _s3_error_fs;
783
    // error file path on s3, ${bucket}/${prefix}/error_log/${label}_${fragment_instance_id}
784
    std::string _s3_error_log_file_path;
785
    std::mutex _s3_error_log_file_lock;
786
};
787
788
#define RETURN_IF_CANCELLED(state)               \
789
595
    do {                                         \
790
595
        if (UNLIKELY((state)->is_cancelled())) { \
791
0
            return (state)->cancel_reason();     \
792
0
        }                                        \
793
595
    } while (false)
794
795
} // namespace doris