Coverage Report

Created: 2025-04-24 12:23

/root/doris/be/src/common/config.h
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
#pragma once
19
20
#include <cstdint>
21
#include <functional>
22
#include <map>
23
#include <mutex>
24
#include <string>
25
#include <utility>
26
#include <vector>
27
28
#define DECLARE_FIELD(FIELD_TYPE, FIELD_NAME) extern FIELD_TYPE FIELD_NAME
29
30
#define DECLARE_Bool(name) DECLARE_FIELD(bool, name)
31
#define DECLARE_Int16(name) DECLARE_FIELD(int16_t, name)
32
#define DECLARE_Int32(name) DECLARE_FIELD(int32_t, name)
33
#define DECLARE_Int64(name) DECLARE_FIELD(int64_t, name)
34
#define DECLARE_Double(name) DECLARE_FIELD(double, name)
35
#define DECLARE_String(name) DECLARE_FIELD(std::string, name)
36
#define DECLARE_Bools(name) DECLARE_FIELD(std::vector<bool>, name)
37
#define DECLARE_Int16s(name) DECLARE_FIELD(std::vector<int16_t>, name)
38
#define DECLARE_Int32s(name) DECLARE_FIELD(std::vector<int32_t>, name)
39
#define DECLARE_Int64s(name) DECLARE_FIELD(std::vector<int64_t>, name)
40
#define DECLARE_Doubles(name) DECLARE_FIELD(std::vector<double>, name)
41
#define DECLARE_Strings(name) DECLARE_FIELD(std::vector<std::string>, name)
42
#define DECLARE_mBool(name) DECLARE_FIELD(bool, name)
43
#define DECLARE_mInt16(name) DECLARE_FIELD(int16_t, name)
44
#define DECLARE_mInt32(name) DECLARE_FIELD(int32_t, name)
45
#define DECLARE_mInt64(name) DECLARE_FIELD(int64_t, name)
46
#define DECLARE_mDouble(name) DECLARE_FIELD(double, name)
47
#define DECLARE_mString(name) DECLARE_FIELD(std::string, name)
48
49
#define DEFINE_FIELD(FIELD_TYPE, FIELD_NAME, FIELD_DEFAULT, VALMUTABLE)                      \
50
21
    FIELD_TYPE FIELD_NAME;                                                                   \
51
21
    static Register reg_##FIELD_NAME(#FIELD_TYPE, #FIELD_NAME, &(FIELD_NAME), FIELD_DEFAULT, \
52
21
                                     VALMUTABLE);
53
54
#define DEFINE_VALIDATOR(FIELD_NAME, VALIDATOR)              \
55
    static auto validator_##FIELD_NAME = VALIDATOR;          \
56
    static RegisterConfValidator reg_validator_##FIELD_NAME( \
57
            #FIELD_NAME, []() -> bool { return validator_##FIELD_NAME(FIELD_NAME); });
58
59
1
#define DEFINE_Int16(name, defaultstr) DEFINE_FIELD(int16_t, name, defaultstr, false)
60
1
#define DEFINE_Bools(name, defaultstr) DEFINE_FIELD(std::vector<bool>, name, defaultstr, false)
61
1
#define DEFINE_Doubles(name, defaultstr) DEFINE_FIELD(std::vector<double>, name, defaultstr, false)
62
1
#define DEFINE_Int16s(name, defaultstr) DEFINE_FIELD(std::vector<int16_t>, name, defaultstr, false)
63
1
#define DEFINE_Int32s(name, defaultstr) DEFINE_FIELD(std::vector<int32_t>, name, defaultstr, false)
64
1
#define DEFINE_Int64s(name, defaultstr) DEFINE_FIELD(std::vector<int64_t>, name, defaultstr, false)
65
3
#define DEFINE_Bool(name, defaultstr) DEFINE_FIELD(bool, name, defaultstr, false)
66
1
#define DEFINE_Double(name, defaultstr) DEFINE_FIELD(double, name, defaultstr, false)
67
1
#define DEFINE_Int32(name, defaultstr) DEFINE_FIELD(int32_t, name, defaultstr, false)
68
1
#define DEFINE_Int64(name, defaultstr) DEFINE_FIELD(int64_t, name, defaultstr, false)
69
2
#define DEFINE_String(name, defaultstr) DEFINE_FIELD(std::string, name, defaultstr, false)
70
#define DEFINE_Strings(name, defaultstr) \
71
1
    DEFINE_FIELD(std::vector<std::string>, name, defaultstr, false)
72
1
#define DEFINE_mBool(name, defaultstr) DEFINE_FIELD(bool, name, defaultstr, true)
73
1
#define DEFINE_mInt16(name, defaultstr) DEFINE_FIELD(int16_t, name, defaultstr, true)
74
1
#define DEFINE_mInt32(name, defaultstr) DEFINE_FIELD(int32_t, name, defaultstr, true)
75
1
#define DEFINE_mInt64(name, defaultstr) DEFINE_FIELD(int64_t, name, defaultstr, true)
76
1
#define DEFINE_mDouble(name, defaultstr) DEFINE_FIELD(double, name, defaultstr, true)
77
1
#define DEFINE_mString(name, defaultstr) DEFINE_FIELD(std::string, name, defaultstr, true)
78
#define DEFINE_Validator(name, validator) DEFINE_VALIDATOR(name, validator)
79
80
namespace doris {
81
class Status;
82
83
// If you want to modify the value of config, please go to common/config.cpp to modify.
84
namespace config {
85
// Dir of custom config file
86
DECLARE_String(custom_config_dir);
87
88
// Dir of jdbc drivers
89
DECLARE_String(jdbc_drivers_dir);
90
91
// cluster id
92
DECLARE_Int32(cluster_id);
93
// port on which BackendService is exported
94
DECLARE_Int32(be_port);
95
96
// port for brpc
97
DECLARE_Int32(brpc_port);
98
99
// port for arrow flight sql
100
// Default -1, do not start arrow flight sql server.
101
DECLARE_Int32(arrow_flight_sql_port);
102
103
// If the external client cannot directly access priority_networks, set public_host to be accessible
104
// to external client.
105
// There are usually two usage scenarios:
106
// 1. in production environment, it is often inconvenient to expose Doris BE nodes to the external network.
107
// However, a reverse proxy (such as Nginx) can be added to all Doris BE nodes, and the external client will be
108
// randomly routed to a Doris BE node when connecting to Nginx. set public_host to the host of Nginx.
109
// 2. if priority_networks is an internal network IP, and BE node has its own independent external IP,
110
// but Doris currently does not support modifying priority_networks, setting public_host to the real external IP.
111
DECLARE_mString(public_host);
112
113
// If the BE node is connected to the external network through a reverse proxy like Nginx
114
// and need to use Arrow Flight SQL, should add a server in Nginx to reverse proxy
115
// `Nginx:arrow_flight_sql_proxy_port` to `BE_priority_networks:arrow_flight_sql_port`. For example:
116
// upstream arrowflight {
117
//    server 10.16.10.8:8069;
118
//    server 10.16.10.8:8068;
119
//}
120
// server {
121
//    listen 8167 http2;
122
//    listen [::]:8167 http2;
123
//    server_name doris.arrowflight.com;
124
// }
125
DECLARE_Int32(arrow_flight_sql_proxy_port);
126
127
// the number of bthreads for brpc, the default value is set to -1,
128
// which means the number of bthreads is #cpu-cores
129
DECLARE_Int32(brpc_num_threads);
130
DECLARE_Int32(brpc_idle_timeout_sec);
131
132
// Declare a selection strategy for those servers have many ips.
133
// Note that there should at most one ip match this list.
134
// This is a list in semicolon-delimited format, in CIDR notation, e.g. 10.10.10.0/24
135
// If no ip match this rule, will choose one randomly.
136
DECLARE_String(priority_networks);
137
138
// performance moderate or compact, only tcmalloc compile
139
DECLARE_String(memory_mode);
140
141
// if true, process memory limit and memory usage based on cgroup memory info.
142
DECLARE_mBool(enable_use_cgroup_memory_info);
143
144
// process memory limit specified as number of bytes
145
// ('<int>[bB]?'), megabytes ('<float>[mM]'), gigabytes ('<float>[gG]'),
146
// or percentage of the physical memory ('<int>%').
147
// defaults to bytes if no unit is given"
148
// must larger than 0. and if larger than physical memory size,
149
// it will be set to physical memory size.
150
DECLARE_String(mem_limit);
151
152
// Soft memory limit as a fraction of hard memory limit.
153
DECLARE_Double(soft_mem_limit_frac);
154
155
// Schema change memory limit as a fraction of soft memory limit.
156
DECLARE_Double(schema_change_mem_limit_frac);
157
158
// Many modern allocators (for example) do not do a mremap for
159
// realloc, even in case of large enough chunks of memory. Although this allows
160
// you to increase performance and reduce memory consumption during realloc.
161
// To fix this, we do mremap manually if the chunk of memory is large enough.
162
//
163
// The threshold (128 MB, 128 * (1ULL << 20)) is chosen quite large, since changing the address
164
// space is very slow, especially in the case of a large number of threads. We
165
// expect that the set of operations mmap/something to do/mremap can only be
166
// performed about 1000 times per second.
167
//
168
// P.S. This is also required, because tcmalloc can not allocate a chunk of
169
// memory greater than 16 GB.
170
DECLARE_mInt64(mmap_threshold); // bytes
171
172
// When hash table capacity is greater than 2^double_grow_degree(default 2G), grow when 75% of the capacity is satisfied.
173
// Increase can reduce the number of hash table resize, but may waste more memory.
174
DECLARE_mInt32(hash_table_double_grow_degree);
175
176
// The max fill rate for hash table
177
DECLARE_mInt32(max_fill_rate);
178
179
DECLARE_mInt32(double_resize_threshold);
180
181
// The maximum low water mark of the system `/proc/meminfo/MemAvailable`, Unit byte, default -1.
182
// if it is -1, then low water mark = min(MemTotal - MemLimit, MemTotal * 5%), which is 3.2G on a 64G machine.
183
// Turn up max. more memory buffers will be reserved for Memory GC.
184
// Turn down max. will use as much memory as possible.
185
// note that: `max_` prefix should be removed, but keep it for compatibility.
186
DECLARE_Int64(max_sys_mem_available_low_water_mark_bytes);
187
188
// reserve a small amount of memory so we do not trigger MinorGC
189
DECLARE_Int64(memtable_limiter_reserved_memory_bytes);
190
191
// The size of the memory that gc wants to release each time, as a percentage of the mem limit.
192
DECLARE_mString(process_minor_gc_size);
193
DECLARE_mString(process_full_gc_size);
194
195
// If true, when the process does not exceed the soft mem limit, the query memory will not be limited;
196
// when the process memory exceeds the soft mem limit, the query with the largest ratio between the currently
197
// used memory and the exec_mem_limit will be canceled.
198
// If false, cancel query when the memory used exceeds exec_mem_limit, same as before.
199
DECLARE_mBool(enable_query_memory_overcommit);
200
201
// gc will release cache, cancel task, and task will wait for gc to release memory,
202
// default gc strategy is conservative, if you want to exclude the interference of gc, let it be true
203
DECLARE_mBool(disable_memory_gc);
204
205
// when alloc memory larger than stacktrace_in_alloc_large_memory_bytes, default 2G,
206
// if alloc successful, will print a warning with stacktrace, but not prevent memory alloc.
207
// if alloc failed using Doris Allocator, will print stacktrace in error log.
208
// if is -1, disable print stacktrace when alloc large memory.
209
DECLARE_mInt64(stacktrace_in_alloc_large_memory_bytes);
210
211
// when alloc memory larger than crash_in_alloc_large_memory_bytes will crash, default -1 means disabled.
212
// if you need a core dump to analyze large memory allocation,
213
// modify this parameter to crash when large memory allocation occur will help
214
DECLARE_mInt64(crash_in_alloc_large_memory_bytes);
215
216
// If memory tracker value is inaccurate, BE will crash. usually used in test environments, default value is false.
217
DECLARE_mBool(crash_in_memory_tracker_inaccurate);
218
219
// default is true. if any memory tracking in Orphan mem tracker will report error.
220
// !! not modify the default value of this conf!! otherwise memory errors cannot be detected in time.
221
// allocator free memory not need to check, because when the thread memory tracker label is Orphan,
222
// use the tracker saved in Allocator.
223
DECLARE_mBool(enable_memory_orphan_check);
224
225
// The maximum time a thread waits for a full GC. Currently only query will wait for full gc.
226
DECLARE_mInt32(thread_wait_gc_max_milliseconds);
227
228
// reach mem limit, don't serialize in batch
229
DECLARE_mInt64(pre_serialize_keys_limit_bytes);
230
231
// the port heartbeat service used
232
DECLARE_Int32(heartbeat_service_port);
233
// the count of heart beat service
234
DECLARE_Int32(heartbeat_service_thread_count);
235
// the count of thread to create table
236
DECLARE_Int32(create_tablet_worker_count);
237
// the count of thread to drop table
238
DECLARE_Int32(drop_tablet_worker_count);
239
// the count of thread to batch load
240
DECLARE_Int32(push_worker_count_normal_priority);
241
// the count of thread to high priority batch load
242
DECLARE_Int32(push_worker_count_high_priority);
243
// the count of thread to publish version
244
DECLARE_Int32(publish_version_worker_count);
245
// the count of tablet thread to publish version
246
DECLARE_Int32(tablet_publish_txn_max_thread);
247
// the timeout of EnginPublishVersionTask
248
DECLARE_Int32(publish_version_task_timeout_s);
249
// the count of thread to calc delete bitmap
250
DECLARE_Int32(calc_delete_bitmap_max_thread);
251
// the count of thread to clear transaction task
252
DECLARE_Int32(clear_transaction_task_worker_count);
253
// the count of thread to delete
254
DECLARE_Int32(delete_worker_count);
255
// the count of thread to alter table
256
DECLARE_Int32(alter_tablet_worker_count);
257
// the count of thread to alter index
258
DECLARE_Int32(alter_index_worker_count);
259
// the count of thread to clone
260
DECLARE_Int32(clone_worker_count);
261
// the count of thread to clone
262
DECLARE_Int32(storage_medium_migrate_count);
263
// the count of thread to check consistency
264
DECLARE_Int32(check_consistency_worker_count);
265
// the count of thread to upload
266
DECLARE_Int32(upload_worker_count);
267
// the count of thread to download
268
DECLARE_Int32(download_worker_count);
269
// the count of thread to make snapshot
270
DECLARE_Int32(make_snapshot_worker_count);
271
// the count of thread to release snapshot
272
DECLARE_Int32(release_snapshot_worker_count);
273
// report random wait a little time to avoid FE receiving multiple be reports at the same time.
274
// do not set it to false for production environment
275
DECLARE_mBool(report_random_wait);
276
// the interval time(seconds) for agent report tasks signature to FE
277
DECLARE_mInt32(report_task_interval_seconds);
278
// the interval time(seconds) for refresh storage policy from FE
279
DECLARE_mInt32(storage_refresh_storage_policy_task_interval_seconds);
280
// the interval time(seconds) for agent report disk state to FE
281
DECLARE_mInt32(report_disk_state_interval_seconds);
282
// the interval time(seconds) for agent report olap table to FE
283
DECLARE_mInt32(report_tablet_interval_seconds);
284
// the max download speed(KB/s)
285
DECLARE_mInt32(max_download_speed_kbps);
286
// download low speed limit(KB/s)
287
DECLARE_mInt32(download_low_speed_limit_kbps);
288
// download low speed time(seconds)
289
DECLARE_mInt32(download_low_speed_time);
290
// whether to download small files in batch.
291
DECLARE_mBool(enable_batch_download);
292
// whether to check md5sum when download
293
DECLARE_mBool(enable_download_md5sum_check);
294
// download binlog meta timeout
295
DECLARE_mInt32(download_binlog_meta_timeout_ms);
296
297
// deprecated, use env var LOG_DIR in be.conf
298
DECLARE_String(sys_log_dir);
299
// for udf
300
DECLARE_String(user_function_dir);
301
// INFO, WARNING, ERROR, FATAL
302
DECLARE_String(sys_log_level);
303
// TIME-DAY, TIME-HOUR, SIZE-MB-nnn
304
DECLARE_String(sys_log_roll_mode);
305
// log roll num
306
DECLARE_Int32(sys_log_roll_num);
307
// verbose log
308
DECLARE_Strings(sys_log_verbose_modules);
309
// verbose log level
310
DECLARE_Int32(sys_log_verbose_level);
311
// verbose log FLAGS_v
312
DECLARE_Int32(sys_log_verbose_flags_v);
313
// log buffer level
314
DECLARE_String(log_buffer_level);
315
// log enable custom date time format
316
DECLARE_Bool(sys_log_enable_custom_date_time_format);
317
// log custom date time format (https://en.cppreference.com/w/cpp/io/manip/put_time)
318
DECLARE_String(sys_log_custom_date_time_format);
319
// log custom date time milliseconds format (fmt::format)
320
DECLARE_String(sys_log_custom_date_time_ms_format);
321
322
// number of threads available to serve backend execution requests
323
DECLARE_Int32(be_service_threads);
324
325
// interval between profile reports; in seconds
326
DECLARE_mInt32(status_report_interval);
327
DECLARE_mInt32(pipeline_status_report_interval);
328
// if true, each disk will have a separate thread pool for scanner
329
DECLARE_Bool(doris_enable_scanner_thread_pool_per_disk);
330
// the timeout of a work thread to wait the blocking priority queue to get a task
331
DECLARE_mInt64(doris_blocking_priority_queue_wait_timeout_ms);
332
// number of scanner thread pool size for olap table
333
// and the min thread num of remote scanner thread pool
334
DECLARE_mInt32(doris_scanner_thread_pool_thread_num);
335
DECLARE_mInt32(doris_scanner_min_thread_pool_thread_num);
336
// number of batch size to fetch the remote split source
337
DECLARE_mInt32(remote_split_source_batch_size);
338
// max number of remote scanner thread pool size
339
// if equal to -1, value is std::max(512, CpuInfo::num_cores() * 10)
340
DECLARE_Int32(doris_max_remote_scanner_thread_pool_thread_num);
341
// number of olap scanner thread pool queue size
342
DECLARE_Int32(doris_scanner_thread_pool_queue_size);
343
// default thrift client connect timeout(in seconds)
344
DECLARE_mInt32(thrift_connect_timeout_seconds);
345
DECLARE_mInt32(fetch_rpc_timeout_seconds);
346
// default thrift client retry interval (in milliseconds)
347
DECLARE_mInt64(thrift_client_retry_interval_ms);
348
// max message size of thrift request
349
// default: 100 * 1024 * 1024
350
DECLARE_mInt64(thrift_max_message_size);
351
// max row count number for single scan range, used in segmentv1
352
DECLARE_mInt32(doris_scan_range_row_count);
353
// max bytes number for single scan range, used in segmentv2
354
DECLARE_mInt32(doris_scan_range_max_mb);
355
// max bytes number for single scan block, used in segmentv2
356
DECLARE_mInt32(doris_scan_block_max_mb);
357
// size of scanner queue between scanner thread and compute thread
358
DECLARE_mInt32(doris_scanner_queue_size);
359
// single read execute fragment row number
360
DECLARE_mInt32(doris_scanner_row_num);
361
// single read execute fragment row bytes
362
DECLARE_mInt32(doris_scanner_row_bytes);
363
// single read execute fragment max run time millseconds
364
DECLARE_mInt32(doris_scanner_max_run_time_ms);
365
DECLARE_mInt32(min_bytes_in_scanner_queue);
366
// number of max scan keys
367
DECLARE_mInt32(doris_max_scan_key_num);
368
// the max number of push down values of a single column.
369
// if exceed, no conditions will be pushed down for that column.
370
DECLARE_mInt32(max_pushdown_conditions_per_column);
371
// (Advanced) Maximum size of per-query receive-side buffer
372
DECLARE_mInt32(exchg_node_buffer_size_bytes);
373
DECLARE_mInt32(exchg_buffer_queue_capacity_factor);
374
375
DECLARE_mInt64(column_dictionary_key_ratio_threshold);
376
DECLARE_mInt64(column_dictionary_key_size_threshold);
377
// memory_limitation_per_thread_for_schema_change_bytes unit bytes
378
DECLARE_mInt64(memory_limitation_per_thread_for_schema_change_bytes);
379
DECLARE_mInt64(memory_limitation_per_thread_for_storage_migration_bytes);
380
381
// all cache prune interval, used by GC and periodic thread.
382
DECLARE_mInt32(cache_prune_interval_sec);
383
DECLARE_mInt32(cache_periodic_prune_stale_sweep_sec);
384
// the clean interval of tablet lookup cache
385
DECLARE_mInt32(tablet_lookup_cache_stale_sweep_time_sec);
386
DECLARE_mInt32(point_query_row_cache_stale_sweep_time_sec);
387
DECLARE_mInt32(disk_stat_monitor_interval);
388
DECLARE_mInt32(unused_rowset_monitor_interval);
389
DECLARE_String(storage_root_path);
390
DECLARE_mString(broken_storage_path);
391
392
// Config is used to check incompatible old format hdr_ format
393
// whether doris uses strict way. When config is true, process will log fatal
394
// and exit. When config is false, process will only log warning.
395
DECLARE_Bool(storage_strict_check_incompatible_old_format);
396
397
// BE process will exit if the percentage of error disk reach this value.
398
DECLARE_mInt32(max_percentage_of_error_disk);
399
DECLARE_mInt32(default_num_rows_per_column_file_block);
400
// pending data policy
401
DECLARE_mInt32(pending_data_expire_time_sec);
402
// inc_rowset snapshot rs sweep time interval
403
DECLARE_mInt32(tablet_rowset_stale_sweep_time_sec);
404
// tablet stale rowset sweep by threshold size
405
DECLARE_Bool(tablet_rowset_stale_sweep_by_size);
406
DECLARE_mInt32(tablet_rowset_stale_sweep_threshold_size);
407
// garbage sweep policy
408
DECLARE_Int32(max_garbage_sweep_interval);
409
DECLARE_Int32(min_garbage_sweep_interval);
410
// garbage sweep every batch will sleep 1ms
411
DECLARE_mInt32(garbage_sweep_batch_size);
412
DECLARE_mInt32(snapshot_expire_time_sec);
413
// It is only a recommended value. When the disk space is insufficient,
414
// the file storage period under trash dose not have to comply with this parameter.
415
DECLARE_mInt32(trash_file_expire_time_sec);
416
// minimum file descriptor number
417
// modify them upon necessity
418
DECLARE_Int32(min_file_descriptor_number);
419
DECLARE_mBool(disable_segment_cache);
420
DECLARE_String(row_cache_mem_limit);
421
422
// Cache for storage page size
423
DECLARE_String(storage_page_cache_limit);
424
// Shard size for page cache, the value must be power of two.
425
// It's recommended to set it to a value close to the number of BE cores in order to reduce lock contentions.
426
DECLARE_Int32(storage_page_cache_shard_size);
427
// Percentage for index page cache
428
// all storage page cache will be divided into data_page_cache and index_page_cache
429
DECLARE_Int32(index_page_cache_percentage);
430
// whether to disable page cache feature in storage
431
// TODO delete it. Divided into Data page, Index page, pk index page
432
DECLARE_Bool(disable_storage_page_cache);
433
// whether to disable row cache feature in storage
434
DECLARE_mBool(disable_storage_row_cache);
435
// whether to disable pk page cache feature in storage
436
DECLARE_Bool(disable_pk_storage_page_cache);
437
438
// Cache for mow primary key storage page size, it's seperated from
439
// storage_page_cache_limit
440
DECLARE_String(pk_storage_page_cache_limit);
441
// data page size for primary key index
442
DECLARE_Int32(primary_key_data_page_size);
443
444
// inc_rowset snapshot rs sweep time interval
445
DECLARE_mInt32(data_page_cache_stale_sweep_time_sec);
446
DECLARE_mInt32(index_page_cache_stale_sweep_time_sec);
447
// great impact on the performance of MOW, so it can be longer.
448
DECLARE_mInt32(pk_index_page_cache_stale_sweep_time_sec);
449
450
DECLARE_Bool(enable_low_cardinality_optimize);
451
DECLARE_Bool(enable_low_cardinality_cache_code);
452
453
// be policy
454
// whether check compaction checksum
455
DECLARE_mBool(enable_compaction_checksum);
456
// whether disable automatic compaction task
457
DECLARE_mBool(disable_auto_compaction);
458
// whether enable vertical compaction
459
DECLARE_mBool(enable_vertical_compaction);
460
// whether enable ordered data compaction
461
DECLARE_mBool(enable_ordered_data_compaction);
462
// In vertical compaction, column number for every group
463
DECLARE_mInt32(vertical_compaction_num_columns_per_group);
464
// In vertical compaction, max memory usage for row_source_buffer
465
DECLARE_Int32(vertical_compaction_max_row_source_memory_mb);
466
// In vertical compaction, max dest segment file size
467
DECLARE_mInt64(vertical_compaction_max_segment_size);
468
469
// If enabled, segments will be flushed column by column
470
DECLARE_mBool(enable_vertical_segment_writer);
471
472
// In ordered data compaction, min segment size for input rowset
473
DECLARE_mInt32(ordered_data_compaction_min_segment_size);
474
475
// This config can be set to limit thread number in compaction thread pool.
476
DECLARE_mInt32(max_base_compaction_threads);
477
DECLARE_mInt32(max_cumu_compaction_threads);
478
DECLARE_mInt32(max_single_replica_compaction_threads);
479
480
DECLARE_Bool(enable_base_compaction_idle_sched);
481
DECLARE_mInt64(base_compaction_min_rowset_num);
482
DECLARE_mInt64(base_compaction_max_compaction_score);
483
DECLARE_mDouble(base_compaction_min_data_ratio);
484
DECLARE_mInt64(base_compaction_dup_key_max_file_size_mbytes);
485
486
DECLARE_Bool(enable_skip_tablet_compaction);
487
DECLARE_mInt32(skip_tablet_compaction_second);
488
// output rowset of cumulative compaction total disk size exceed this config size,
489
// this rowset will be given to base compaction, unit is m byte.
490
DECLARE_mInt64(compaction_promotion_size_mbytes);
491
492
// output rowset of cumulative compaction total disk size exceed this config ratio of
493
// base rowset's total disk size, this rowset will be given to base compaction. The value must be between
494
// 0 and 1.
495
DECLARE_mDouble(compaction_promotion_ratio);
496
497
// the smallest size of rowset promotion. When the rowset is less than this config, this
498
// rowset will be not given to base compaction. The unit is m byte.
499
DECLARE_mInt64(compaction_promotion_min_size_mbytes);
500
501
// When output rowset of cumulative compaction total version count (end_version - start_version)
502
// exceed this config count, the rowset will be moved to base compaction
503
// NOTE: this config will work for unique key merge-on-write table only, to reduce version count
504
// related cost on delete bitmap more effectively.
505
DECLARE_mInt64(compaction_promotion_version_count);
506
507
// The lower bound size to do cumulative compaction. When total disk size of candidate rowsets is less than
508
// this size, size_based policy may not do to cumulative compaction. The unit is m byte.
509
DECLARE_mInt64(compaction_min_size_mbytes);
510
511
// cumulative compaction policy: min and max delta file's number
512
DECLARE_mInt64(cumulative_compaction_min_deltas);
513
DECLARE_mInt64(cumulative_compaction_max_deltas);
514
DECLARE_mInt32(cumulative_compaction_max_deltas_factor);
515
516
// This config can be set to limit thread number in  multiget thread pool.
517
DECLARE_mInt32(multi_get_max_threads);
518
519
// The upper limit of "permits" held by all compaction tasks. This config can be set to limit memory consumption for compaction.
520
DECLARE_mInt64(total_permits_for_compaction_score);
521
522
// sleep interval in ms after generated compaction tasks
523
DECLARE_mInt32(generate_compaction_tasks_interval_ms);
524
// sleep interval in second after update replica infos
525
DECLARE_mInt32(update_replica_infos_interval_seconds);
526
527
// Compaction task number per disk.
528
// Must be greater than 2, because Base compaction and Cumulative compaction have at least one thread each.
529
DECLARE_mInt32(compaction_task_num_per_disk);
530
// compaction thread num for fast disk(typically .SSD), must be greater than 2.
531
DECLARE_mInt32(compaction_task_num_per_fast_disk);
532
533
// How many rounds of cumulative compaction for each round of base compaction when compaction tasks generation.
534
DECLARE_mInt32(cumulative_compaction_rounds_for_each_base_compaction_round);
535
536
// Not compact the invisible versions, but with some limitations:
537
// if not timeout, keep no more than compaction_keep_invisible_version_max_count versions;
538
// if timeout, keep no more than compaction_keep_invisible_version_min_count versions.
539
DECLARE_mInt32(compaction_keep_invisible_version_timeout_sec);
540
DECLARE_mInt32(compaction_keep_invisible_version_min_count);
541
DECLARE_mInt32(compaction_keep_invisible_version_max_count);
542
543
// Threshold to logging compaction trace, in seconds.
544
DECLARE_mInt32(base_compaction_trace_threshold);
545
DECLARE_mInt32(cumulative_compaction_trace_threshold);
546
DECLARE_mBool(disable_compaction_trace_log);
547
548
// Interval to picking rowset to compact, in seconds
549
DECLARE_mInt64(pick_rowset_to_compact_interval_sec);
550
551
// Compaction priority schedule
552
DECLARE_mBool(enable_compaction_priority_scheduling);
553
DECLARE_mInt32(low_priority_compaction_task_num_per_disk);
554
DECLARE_mInt32(low_priority_compaction_score_threshold);
555
556
// Thread count to do tablet meta checkpoint, -1 means use the data directories count.
557
DECLARE_Int32(max_meta_checkpoint_threads);
558
559
// Threshold to logging agent task trace, in seconds.
560
DECLARE_mInt32(agent_task_trace_threshold_sec);
561
562
// This config can be set to limit thread number in tablet migration thread pool.
563
DECLARE_Int32(min_tablet_migration_threads);
564
DECLARE_Int32(max_tablet_migration_threads);
565
566
DECLARE_mInt32(finished_migration_tasks_size);
567
// If size less than this, the remaining rowsets will be force to complete
568
DECLARE_mInt32(migration_remaining_size_threshold_mb);
569
// If the task runs longer than this time, the task will be terminated, in seconds.
570
// timeout = std::max(migration_task_timeout_secs,  tablet size / 1MB/s)
571
DECLARE_mInt32(migration_task_timeout_secs);
572
// timeout for try_lock migration lock
573
DECLARE_Int64(migration_lock_timeout_ms);
574
575
// Port to start debug webserver on
576
DECLARE_Int32(webserver_port);
577
// Https enable flag
578
DECLARE_Bool(enable_https);
579
// Path of certificate
580
DECLARE_String(ssl_certificate_path);
581
// Path of private key
582
DECLARE_String(ssl_private_key_path);
583
// Whether to check authorization
584
DECLARE_Bool(enable_all_http_auth);
585
// Number of webserver workers
586
DECLARE_Int32(webserver_num_workers);
587
588
DECLARE_Bool(enable_single_replica_load);
589
// Number of download workers for single replica load
590
DECLARE_Int32(single_replica_load_download_num_workers);
591
592
// Used for mini Load. mini load data file will be removed after this time.
593
DECLARE_Int64(load_data_reserve_hours);
594
// log error log will be removed after this time
595
DECLARE_mInt64(load_error_log_reserve_hours);
596
// error log size limit, default 200MB
597
DECLARE_mInt64(load_error_log_limit_bytes);
598
599
// be brpc interface is classified into two categories: light and heavy
600
// each category has diffrent thread number
601
// threads to handle heavy api interface, such as transmit_data/transmit_block etc
602
// Default, if less than or equal 32 core, the following are 128, 128, 10240, 10240 in turn.
603
//          if greater than 32 core, the following are core num * 4, core num * 4, core num * 320, core num * 320 in turn
604
DECLARE_Int32(brpc_heavy_work_pool_threads);
605
// threads to handle light api interface, such as exec_plan_fragment_prepare/exec_plan_fragment_start
606
DECLARE_Int32(brpc_light_work_pool_threads);
607
DECLARE_Int32(brpc_heavy_work_pool_max_queue_size);
608
DECLARE_Int32(brpc_light_work_pool_max_queue_size);
609
DECLARE_Int32(brpc_arrow_flight_work_pool_threads);
610
DECLARE_Int32(brpc_arrow_flight_work_pool_max_queue_size);
611
612
// The maximum amount of data that can be processed by a stream load
613
DECLARE_mInt64(streaming_load_max_mb);
614
// Some data formats, such as JSON, cannot be streamed.
615
// Therefore, it is necessary to limit the maximum number of
616
// such data when using stream load to prevent excessive memory consumption.
617
DECLARE_mInt64(streaming_load_json_max_mb);
618
// the alive time of a TabletsChannel.
619
// If the channel does not receive any data till this time,
620
// the channel will be removed.
621
DECLARE_mInt32(streaming_load_rpc_max_alive_time_sec);
622
// the timeout of a rpc to open the tablet writer in remote BE.
623
// short operation time, can set a short timeout
624
DECLARE_Int32(tablet_writer_open_rpc_timeout_sec);
625
// You can ignore brpc error '[E1011]The server is overcrowded' when writing data.
626
DECLARE_mBool(tablet_writer_ignore_eovercrowded);
627
DECLARE_mBool(exchange_sink_ignore_eovercrowded);
628
DECLARE_mInt32(slave_replica_writer_rpc_timeout_sec);
629
// Whether to enable stream load record function, the default is false.
630
// False: disable stream load record
631
DECLARE_mBool(enable_stream_load_record);
632
// batch size of stream load record reported to FE
633
DECLARE_mInt32(stream_load_record_batch_size);
634
// expire time of stream load record in rocksdb.
635
DECLARE_Int32(stream_load_record_expire_time_secs);
636
// time interval to clean expired stream load records
637
DECLARE_mInt64(clean_stream_load_record_interval_secs);
638
// The buffer size to store stream table function schema info
639
DECLARE_Int64(stream_tvf_buffer_size);
640
641
// OlapTableSink sender's send interval, should be less than the real response time of a tablet writer rpc.
642
// You may need to lower the speed when the sink receiver bes are too busy.
643
DECLARE_mInt32(olap_table_sink_send_interval_microseconds);
644
// For auto partition, the send interval will multiply the factor
645
DECLARE_mDouble(olap_table_sink_send_interval_auto_partition_factor);
646
647
// Fragment thread pool
648
DECLARE_Int32(fragment_pool_thread_num_min);
649
DECLARE_Int32(fragment_pool_thread_num_max);
650
DECLARE_Int32(fragment_pool_queue_size);
651
652
// Control the number of disks on the machine.  If 0, this comes from the system settings.
653
DECLARE_Int32(num_disks);
654
// The maximum number of the threads per disk is also the max queue depth per disk.
655
DECLARE_Int32(num_threads_per_disk);
656
// The read size is the size of the reads sent to os.
657
// There is a trade off of latency and throughout, trying to keep disks busy but
658
// not introduce seeks.  The literature seems to agree that with 8 MB reads, random
659
// io and sequential io perform similarly.
660
DECLARE_Int32(read_size);       // 8 * 1024 * 1024, Read Size (in bytes)
661
DECLARE_Int32(min_buffer_size); // 1024, The minimum read buffer size (in bytes)
662
663
// for pprof
664
DECLARE_String(pprof_profile_dir);
665
// for jeprofile in jemalloc
666
DECLARE_mString(jeprofile_dir);
667
// Purge all unused dirty pages for all arenas.
668
DECLARE_mBool(enable_je_purge_dirty_pages);
669
// Purge all unused Jemalloc dirty pages for all arenas when exceed je_dirty_pages_mem_limit and process exceed soft limit.
670
DECLARE_mString(je_dirty_pages_mem_limit_percent);
671
672
// to forward compatibility, will be removed later
673
DECLARE_mBool(enable_token_check);
674
675
// to open/close system metrics
676
DECLARE_Bool(enable_system_metrics);
677
678
// Number of cores Doris will used, this will effect only when it's greater than 0.
679
// Otherwise, Doris will use all cores returned from "/proc/cpuinfo".
680
DECLARE_Int32(num_cores);
681
682
// When BE start, If there is a broken disk, BE process will exit by default.
683
// Otherwise, we will ignore the broken disk,
684
DECLARE_Bool(ignore_broken_disk);
685
686
// Sleep time in milliseconds between memory maintenance iterations
687
DECLARE_mInt32(memory_maintenance_sleep_time_ms);
688
689
// After full gc, no longer full gc and minor gc during sleep.
690
// After minor gc, no minor gc during sleep, but full gc is possible.
691
DECLARE_mInt32(memory_gc_sleep_time_ms);
692
693
// Sleep time in milliseconds between memtbale flush mgr memory refresh iterations
694
DECLARE_mInt64(memtable_mem_tracker_refresh_interval_ms);
695
696
// Sleep time in milliseconds between refresh iterations of workload group weighted memory ratio
697
DECLARE_mInt64(wg_weighted_memory_ratio_refresh_interval_ms);
698
699
// percent of (active memtables size / all memtables size) when reach hard limit
700
DECLARE_mInt32(memtable_hard_limit_active_percent);
701
702
// percent of (active memtables size / all memtables size) when reach soft limit
703
DECLARE_mInt32(memtable_soft_limit_active_percent);
704
705
// memtable insert memory tracker will multiply input block size with this ratio
706
DECLARE_mDouble(memtable_insert_memory_ratio);
707
// max write buffer size before flush, default 200MB
708
DECLARE_mInt64(write_buffer_size);
709
// max buffer size used in memtable for the aggregated table, default 400MB
710
DECLARE_mInt64(write_buffer_size_for_agg);
711
// max parallel flush task per memtable writer
712
DECLARE_mInt32(memtable_flush_running_count_limit);
713
714
DECLARE_Int32(load_process_max_memory_limit_percent); // 50%
715
716
// If the memory consumption of load jobs exceed load_process_max_memory_limit,
717
// all load jobs will hang there to wait for memtable flush. We should have a
718
// soft limit which can trigger the memtable flush for the load channel who
719
// consumes lagest memory size before we reach the hard limit. The soft limit
720
// might avoid all load jobs hang at the same time.
721
DECLARE_Int32(load_process_soft_mem_limit_percent);
722
723
// If load memory consumption is within load_process_safe_mem_permit_percent,
724
// memtable memory limiter will do nothing.
725
DECLARE_Int32(load_process_safe_mem_permit_percent);
726
727
// result buffer cancelled time (unit: second)
728
DECLARE_mInt32(result_buffer_cancelled_interval_time);
729
730
// arrow flight result sink buffer rows size, default 4096 * 8
731
DECLARE_mInt32(arrow_flight_result_sink_buffer_size_rows);
732
// The timeout for ADBC Client to wait for data using arrow flight reader.
733
// If the query is very complex and no result is generated after this time, consider increasing this timeout.
734
DECLARE_mInt32(arrow_flight_reader_brpc_controller_timeout_ms);
735
736
// the increased frequency of priority for remaining tasks in BlockingPriorityQueue
737
DECLARE_mInt32(priority_queue_remaining_tasks_increased_frequency);
738
739
// sync tablet_meta when modifying meta
740
DECLARE_mBool(sync_tablet_meta);
741
742
// default thrift rpc timeout ms
743
DECLARE_mInt32(thrift_rpc_timeout_ms);
744
745
// txn commit rpc timeout
746
DECLARE_mInt32(txn_commit_rpc_timeout_ms);
747
748
// If set to true, metric calculator will run
749
DECLARE_Bool(enable_metric_calculator);
750
751
// max consumer num in one data consumer group, for routine load
752
DECLARE_mInt32(max_consumer_num_per_group);
753
754
// the max size of thread pool for routine load task.
755
// this should be larger than FE config 'max_routine_load_task_num_per_be' (default 5)
756
DECLARE_Int32(max_routine_load_thread_pool_size);
757
758
// max external scan cache batch count, means cache max_memory_cache_batch_count * batch_size row
759
// default is 20, batch_size's default value is 1024 means 20 * 1024 rows will be cached
760
DECLARE_mInt32(max_memory_sink_batch_count);
761
762
// This configuration is used for the context gc thread schedule period
763
// note: unit is minute, default is 5min
764
DECLARE_mInt32(scan_context_gc_interval_min);
765
766
// es scroll keep-alive
767
DECLARE_String(es_scroll_keepalive);
768
769
// HTTP connection timeout for es
770
DECLARE_mInt32(es_http_timeout_ms);
771
772
// the max client cache number per each host
773
// There are variety of client cache in BE, but currently we use the
774
// same cache size configuration.
775
// TODO(cmy): use different config to set different client cache if necessary.
776
DECLARE_Int32(max_client_cache_size_per_host);
777
778
// Dir to save files downloaded by SmallFileMgr
779
DECLARE_String(small_file_dir);
780
// path gc
781
DECLARE_Bool(path_gc_check);
782
DECLARE_mInt32(path_gc_check_interval_second);
783
DECLARE_mInt32(path_gc_check_step);
784
DECLARE_mInt32(path_gc_check_step_interval_ms);
785
786
// The following 2 configs limit the max usage of disk capacity of a data dir.
787
// If both of these 2 threshold reached, no more data can be writen into that data dir.
788
// The percent of max used capacity of a data dir
789
DECLARE_mInt32(storage_flood_stage_usage_percent); // 90%
790
// The min bytes that should be left of a data dir
791
DECLARE_mInt64(storage_flood_stage_left_capacity_bytes); // 1GB
792
// number of thread for flushing memtable per store
793
DECLARE_Int32(flush_thread_num_per_store);
794
// number of thread for flushing memtable per store, for high priority load task
795
DECLARE_Int32(high_priority_flush_thread_num_per_store);
796
797
// workload group's flush thread num
798
DECLARE_Int32(wg_flush_thread_num_per_store);
799
800
// config for tablet meta checkpoint
801
DECLARE_mInt32(tablet_meta_checkpoint_min_new_rowsets_num);
802
DECLARE_mInt32(tablet_meta_checkpoint_min_interval_secs);
803
DECLARE_Int32(generate_tablet_meta_checkpoint_tasks_interval_secs);
804
805
// config for default rowset type
806
// Valid configs: ALPHA, BETA
807
DECLARE_String(default_rowset_type);
808
809
// Maximum size of a single message body in all protocols
810
DECLARE_Int64(brpc_max_body_size);
811
// Max unwritten bytes in each socket, if the limit is reached, Socket.Write fails with EOVERCROWDED
812
// Default, if the physical memory is less than or equal to 64G, the value is 1G
813
//          if the physical memory is greater than 64G, the value is physical memory * mem_limit(0.8) / 1024 * 20
814
DECLARE_Int64(brpc_socket_max_unwritten_bytes);
815
// TODO(zxy): expect to be true in v1.3
816
// Whether to embed the ProtoBuf Request serialized string together with Tuple/Block data into
817
// Controller Attachment and send it through http brpc when the length of the Tuple/Block data
818
// is greater than 1.8G. This is to avoid the error of Request length overflow (2G).
819
DECLARE_mBool(transfer_large_data_by_brpc);
820
821
// max number of txns for every txn_partition_map in txn manager
822
// this is a self protection to avoid too many txns saving in manager
823
DECLARE_mInt64(max_runnings_transactions_per_txn_map);
824
825
// tablet_map_lock shard size, the value is 2^n, n=0,1,2,3,4
826
// this is a an enhancement for better performance to manage tablet
827
DECLARE_Int32(tablet_map_shard_size);
828
829
// txn_map_lock shard size, the value is 2^n, n=0,1,2,3,4
830
// this is a an enhancement for better performance to manage txn
831
DECLARE_Int32(txn_map_shard_size);
832
833
// txn_lock shard size, the value is 2^n, n=0,1,2,3,4
834
// this is a an enhancement for better performance to commit and publish txn
835
DECLARE_Int32(txn_shard_size);
836
837
// Whether to continue to start be when load tablet from header failed.
838
DECLARE_Bool(ignore_load_tablet_failure);
839
840
// Whether to continue to start be when load tablet from header failed.
841
DECLARE_mBool(ignore_rowset_stale_unconsistent_delete);
842
843
// Set max cache's size of query results, the unit is M byte
844
DECLARE_Int32(query_cache_max_size_mb);
845
846
// Cache memory is pruned when reach query_cache_max_size_mb + query_cache_elasticity_size_mb
847
DECLARE_Int32(query_cache_elasticity_size_mb);
848
849
// Maximum number of cache partitions corresponding to a SQL
850
DECLARE_Int32(query_cache_max_partition_count);
851
852
// Maximum number of version of a tablet. If the version num of a tablet exceed limit,
853
// the load process will reject new incoming load job of this tablet.
854
// This is to avoid too many version num.
855
DECLARE_mInt32(max_tablet_version_num);
856
857
// Frontend mainly use two thrift sever type: THREAD_POOL, THREADED_SELECTOR. if fe use THREADED_SELECTOR model for thrift server,
858
// the thrift_server_type_of_fe should be set THREADED_SELECTOR to make be thrift client to fe constructed with TFramedTransport
859
DECLARE_String(thrift_server_type_of_fe);
860
861
// disable zone map index when page row is too few
862
DECLARE_mInt32(zone_map_row_num_threshold);
863
864
// aws sdk log level
865
//    Off = 0,
866
//    Fatal = 1,
867
//    Error = 2,
868
//    Warn = 3,
869
//    Info = 4,
870
//    Debug = 5,
871
//    Trace = 6
872
DECLARE_Int32(aws_log_level);
873
874
// the buffer size when read data from remote storage like s3
875
DECLARE_mInt32(remote_storage_read_buffer_mb);
876
877
// The minimum length when TCMalloc Hook consumes/releases MemTracker, consume size
878
// smaller than this value will continue to accumulate. specified as number of bytes.
879
// Decreasing this value will increase the frequency of consume/release.
880
// Increasing this value will cause MemTracker statistics to be inaccurate.
881
DECLARE_mInt32(mem_tracker_consume_min_size_bytes);
882
883
// The version information of the tablet will be stored in the memory
884
// in an adjacency graph data structure.
885
// And as the new version is written and the old version is deleted,
886
// the data structure will begin to have empty vertex with no edge associations(orphan vertex).
887
// This config is used to control that when the proportion of orphan vertex is greater than the threshold,
888
// the adjacency graph will be rebuilt to ensure that the data structure will not expand indefinitely.
889
// This config usually only needs to be modified during testing.
890
// In most cases, it does not need to be modified.
891
DECLARE_mDouble(tablet_version_graph_orphan_vertex_ratio);
892
893
// share delta writers when memtable_on_sink_node = true
894
DECLARE_Bool(share_delta_writers);
895
// timeout for open load stream rpc in ms
896
DECLARE_Int64(open_load_stream_timeout_ms);
897
// enable write background when using brpc stream
898
DECLARE_mBool(enable_brpc_stream_write_background);
899
900
// brpc streaming max_buf_size in bytes
901
DECLARE_Int64(load_stream_max_buf_size);
902
// brpc streaming messages_in_batch
903
DECLARE_Int32(load_stream_messages_in_batch);
904
// brpc streaming StreamWait seconds on EAGAIN
905
DECLARE_Int32(load_stream_eagain_wait_seconds);
906
// max tasks per flush token in load stream
907
DECLARE_Int32(load_stream_flush_token_max_tasks);
908
// max wait flush token time in load stream
909
DECLARE_Int32(load_stream_max_wait_flush_token_time_ms);
910
911
// max send batch parallelism for OlapTableSink
912
// The value set by the user for send_batch_parallelism is not allowed to exceed max_send_batch_parallelism_per_job,
913
// if exceed, the value of send_batch_parallelism would be max_send_batch_parallelism_per_job
914
DECLARE_mInt32(max_send_batch_parallelism_per_job);
915
916
// number of send batch thread pool size
917
DECLARE_Int32(send_batch_thread_pool_thread_num);
918
// number of send batch thread pool queue size
919
DECLARE_Int32(send_batch_thread_pool_queue_size);
920
921
// Limit the number of segment of a newly created rowset.
922
// The newly created rowset may to be compacted after loading,
923
// so if there are too many segment in a rowset, the compaction process
924
// will run out of memory.
925
// When doing compaction, each segment may take at least 1MB buffer.
926
DECLARE_mInt32(max_segment_num_per_rowset);
927
928
// Store segment without compression if a segment is smaller than
929
// segment_compression_threshold_kb.
930
DECLARE_mInt32(segment_compression_threshold_kb);
931
932
// The connection timeout when connecting to external table such as odbc table.
933
DECLARE_mInt32(external_table_connect_timeout_sec);
934
935
// Time to clean up useless JDBC connection pool cache
936
DECLARE_mInt32(jdbc_connection_pool_cache_clear_time_sec);
937
938
// Global bitmap cache capacity for aggregation cache, size in bytes
939
DECLARE_Int64(delete_bitmap_agg_cache_capacity);
940
DECLARE_String(delete_bitmap_dynamic_agg_cache_limit);
941
DECLARE_mInt32(delete_bitmap_agg_cache_stale_sweep_time_sec);
942
943
// A common object cache depends on an Sharded LRU Cache.
944
DECLARE_mInt32(common_obj_lru_cache_stale_sweep_time_sec);
945
946
// reference https://github.com/edenhill/librdkafka/blob/master/INTRODUCTION.md#broker-version-compatibility
947
// If the dependent kafka broker version older than 0.10.0.0,
948
// the value of kafka_api_version_request should be false, and the
949
// value set by the fallback version kafka_broker_version_fallback will be used,
950
// and the valid values are: 0.9.0.x, 0.8.x.y.
951
DECLARE_String(kafka_api_version_request);
952
DECLARE_String(kafka_broker_version_fallback);
953
DECLARE_mString(kafka_debug);
954
955
// The number of pool siz of routine load consumer.
956
// If you meet the error describe in https://github.com/edenhill/librdkafka/issues/3608
957
// Change this size to 0 to fix it temporarily.
958
DECLARE_mInt32(routine_load_consumer_pool_size);
959
960
// Used in single-stream-multi-table load. When receive a batch of messages from kafka,
961
// if the size of batch is more than this threshold, we will request plans for all related tables.
962
DECLARE_Int32(multi_table_batch_plan_threshold);
963
964
// Used in single-stream-multi-table load. When receiving a batch of messages from Kafka,
965
// if the size of the table wait for plan is more than this threshold, we will request plans for all related tables.
966
// The param is aimed to avoid requesting and executing too many plans at once.
967
// Performing small batch processing on multiple tables during the loaded process can reduce the pressure of a single RPC
968
// and improve the real-time processing of data.
969
DECLARE_Int32(multi_table_max_wait_tables);
970
971
// When the timeout of a load task is less than this threshold,
972
// Doris treats it as a high priority task.
973
// high priority tasks use a separate thread pool for flush and do not block rpc by memory cleanup logic.
974
// this threshold is mainly used to identify routine load tasks and should not be modified if not necessary.
975
DECLARE_mInt32(load_task_high_priority_threshold_second);
976
977
// The min timeout of load rpc (add batch, close)
978
// Because a load rpc may be blocked for a while.
979
// Increase this config may avoid rpc timeout.
980
DECLARE_mInt32(min_load_rpc_timeout_ms);
981
982
// use which protocol to access function service, candicate is baidu_std/h2:grpc
983
DECLARE_String(function_service_protocol);
984
985
// use which load balancer to select server to connect
986
DECLARE_String(rpc_load_balancer);
987
988
// a soft limit of string type length, the hard limit is 2GB - 4, but if too long will cause very low performance,
989
// so we set a soft limit, default is 1MB
990
DECLARE_mInt32(string_type_length_soft_limit_bytes);
991
992
DECLARE_mInt32(jsonb_type_length_soft_limit_bytes);
993
994
// Threshold fo reading a small file into memory
995
DECLARE_mInt32(in_memory_file_size);
996
997
// ParquetReaderWrap prefetch buffer size
998
DECLARE_Int32(parquet_reader_max_buffer_size);
999
// Max size of parquet page header in bytes
1000
DECLARE_mInt32(parquet_header_max_size_mb);
1001
// Max buffer size for parquet row group
1002
DECLARE_mInt32(parquet_rowgroup_max_buffer_mb);
1003
// Max buffer size for parquet chunk column
1004
DECLARE_mInt32(parquet_column_max_buffer_mb);
1005
// Merge small IO, the max amplified read ratio
1006
DECLARE_mDouble(max_amplified_read_ratio);
1007
// Equivalent min size of each IO that can reach the maximum storage speed limit
1008
// 1MB for oss, 8KB for hdfs
1009
DECLARE_mInt32(merged_oss_min_io_size);
1010
DECLARE_mInt32(merged_hdfs_min_io_size);
1011
1012
// OrcReader
1013
DECLARE_mInt32(orc_natural_read_size_mb);
1014
DECLARE_mInt64(big_column_size_buffer);
1015
DECLARE_mInt64(small_column_size_buffer);
1016
1017
// When the rows number reached this limit, will check the filter rate the of bloomfilter
1018
// if it is lower than a specific threshold, the predicate will be disabled.
1019
DECLARE_mInt32(rf_predicate_check_row_num);
1020
1021
// cooldown task configs
1022
DECLARE_Int32(cooldown_thread_num);
1023
DECLARE_mInt64(generate_cooldown_task_interval_sec);
1024
DECLARE_mInt32(remove_unused_remote_files_interval_sec); // 6h
1025
DECLARE_mInt32(confirm_unused_remote_files_interval_sec);
1026
DECLARE_Int32(cold_data_compaction_thread_num);
1027
DECLARE_mInt32(cold_data_compaction_interval_sec);
1028
1029
DECLARE_Int32(s3_transfer_executor_pool_size);
1030
1031
DECLARE_Bool(enable_time_lut);
1032
DECLARE_mBool(enable_simdjson_reader);
1033
1034
DECLARE_mBool(enable_query_like_bloom_filter);
1035
// number of s3 scanner thread pool size
1036
DECLARE_Int32(doris_remote_scanner_thread_pool_thread_num);
1037
// number of s3 scanner thread pool queue size
1038
DECLARE_Int32(doris_remote_scanner_thread_pool_queue_size);
1039
1040
// limit the queue of pending batches which will be sent by a single nodechannel
1041
DECLARE_mInt64(nodechannel_pending_queue_max_bytes);
1042
1043
// The batch size for sending data by brpc streaming client
1044
DECLARE_mInt64(brpc_streaming_client_batch_bytes);
1045
1046
DECLARE_Bool(enable_brpc_builtin_services);
1047
1048
DECLARE_Bool(enable_brpc_connection_check);
1049
1050
DECLARE_mInt64(brpc_connection_check_timeout_ms);
1051
1052
// Max waiting time to wait the "plan fragment start" rpc.
1053
// If timeout, the fragment will be cancelled.
1054
// This parameter is usually only used when the FE loses connection,
1055
// and the BE can automatically cancel the relevant fragment after the timeout,
1056
// so as to avoid occupying the execution thread for a long time.
1057
DECLARE_mInt32(max_fragment_start_wait_time_seconds);
1058
1059
DECLARE_Int32(fragment_mgr_cancel_worker_interval_seconds);
1060
1061
// Node role tag for backend. Mix role is the default role, and computation role have no
1062
// any tablet.
1063
DECLARE_String(be_node_role);
1064
1065
// Hide webserver page for safety.
1066
// Hide the be config page for webserver.
1067
DECLARE_Bool(hide_webserver_config_page);
1068
1069
DECLARE_Bool(enable_segcompaction);
1070
1071
// Max number of segments allowed in a single segcompaction task.
1072
DECLARE_Int32(segcompaction_batch_size);
1073
1074
// Max row count allowed in a single source segment, bigger segments will be skipped.
1075
DECLARE_Int32(segcompaction_candidate_max_rows);
1076
1077
// Max file size allowed in a single source segment, bigger segments will be skipped.
1078
DECLARE_Int64(segcompaction_candidate_max_bytes);
1079
1080
// Max total row count allowed in a single segcompaction task.
1081
DECLARE_Int32(segcompaction_task_max_rows);
1082
1083
// Max total file size allowed in a single segcompaction task.
1084
DECLARE_Int64(segcompaction_task_max_bytes);
1085
1086
// Global segcompaction thread pool size.
1087
DECLARE_mInt32(segcompaction_num_threads);
1088
1089
// enable java udf and jdbc scannode
1090
DECLARE_Bool(enable_java_support);
1091
1092
// Set config randomly to check more issues in github workflow
1093
DECLARE_Bool(enable_fuzzy_mode);
1094
1095
DECLARE_Bool(enable_debug_points);
1096
1097
DECLARE_Int32(pipeline_executor_size);
1098
1099
// block file cache
1100
DECLARE_Bool(enable_file_cache);
1101
// format: [{"path":"/path/to/file_cache","total_size":21474836480,"query_limit":10737418240}]
1102
// format: [{"path":"/path/to/file_cache","total_size":21474836480,"query_limit":10737418240},{"path":"/path/to/file_cache2","total_size":21474836480,"query_limit":10737418240}]
1103
// format: [{"path":"/path/to/file_cache","total_size":21474836480,"query_limit":10737418240,"normal_percent":85, "disposable_percent":10, "index_percent":5}]
1104
DECLARE_String(file_cache_path);
1105
DECLARE_Int64(async_file_cache_init_file_num_interval);
1106
DECLARE_Int64(async_file_cache_init_sleep_interval_ms);
1107
DECLARE_Int64(file_cache_min_file_segment_size);
1108
DECLARE_Int64(file_cache_max_file_segment_size);
1109
DECLARE_Bool(clear_file_cache);
1110
DECLARE_Bool(enable_file_cache_query_limit);
1111
// only for debug, will be removed after finding out the root cause
1112
DECLARE_mInt32(file_cache_wait_sec_after_fail); // zero for no waiting and retrying
1113
DECLARE_mInt32(file_cache_max_evict_num_per_round);
1114
1115
// inverted index searcher cache
1116
// cache entry stay time after lookup
1117
DECLARE_mInt32(index_cache_entry_stay_time_after_lookup_s);
1118
// cache entry that have not been visited for a certain period of time can be cleaned up by GC thread
1119
DECLARE_mInt32(inverted_index_cache_stale_sweep_time_sec);
1120
// inverted index searcher cache size
1121
DECLARE_String(inverted_index_searcher_cache_limit);
1122
// set `true` to enable insert searcher into cache when write inverted index data
1123
DECLARE_Bool(enable_write_index_searcher_cache);
1124
DECLARE_Bool(enable_inverted_index_cache_check_timestamp);
1125
DECLARE_mBool(enable_inverted_index_correct_term_write);
1126
DECLARE_Int32(inverted_index_fd_number_limit_percent); // 50%
1127
DECLARE_Int32(inverted_index_query_cache_shards);
1128
1129
// inverted index match bitmap cache size
1130
DECLARE_String(inverted_index_query_cache_limit);
1131
1132
// inverted index
1133
DECLARE_mDouble(inverted_index_ram_buffer_size);
1134
DECLARE_mInt32(inverted_index_max_buffered_docs);
1135
// dict path for chinese analyzer
1136
DECLARE_String(inverted_index_dict_path);
1137
DECLARE_Int32(inverted_index_read_buffer_size);
1138
// tree depth for bkd index
1139
DECLARE_Int32(max_depth_in_bkd_tree);
1140
// index compaction
1141
DECLARE_mBool(inverted_index_compaction_enable);
1142
// Only for debug, do not use in production
1143
DECLARE_mBool(debug_inverted_index_compaction);
1144
// index by RAM directory
1145
DECLARE_mBool(inverted_index_ram_dir_enable);
1146
// use num_broadcast_buffer blocks as buffer to do broadcast
1147
DECLARE_Int32(num_broadcast_buffer);
1148
1149
// max depth of expression tree allowed.
1150
DECLARE_Int32(max_depth_of_expr_tree);
1151
1152
// Report a tablet as bad when io errors occurs more than this value.
1153
DECLARE_mInt64(max_tablet_io_errors);
1154
1155
// Report a tablet as bad when its path not found
1156
DECLARE_Int32(tablet_path_check_interval_seconds);
1157
DECLARE_mInt32(tablet_path_check_batch_size);
1158
1159
// it must be larger than or equal to 5MB
1160
DECLARE_mInt32(s3_write_buffer_size);
1161
// The timeout config for S3 buffer allocation
1162
DECLARE_mInt32(s3_writer_buffer_allocation_timeout);
1163
// the max number of cached file handle for block segemnt
1164
DECLARE_mInt64(file_cache_max_file_reader_cache_size);
1165
//enable shrink memory
1166
DECLARE_mBool(enable_shrink_memory);
1167
// enable cache for high concurrent point query work load
1168
DECLARE_mInt32(schema_cache_capacity);
1169
DECLARE_mInt32(schema_cache_sweep_time_sec);
1170
1171
// max number of segment cache
1172
DECLARE_Int32(segment_cache_capacity);
1173
DECLARE_Int32(segment_cache_fd_percentage);
1174
DECLARE_Int32(segment_cache_memory_percentage);
1175
DECLARE_mInt32(estimated_mem_per_column_reader);
1176
1177
// enable binlog
1178
DECLARE_Bool(enable_feature_binlog);
1179
1180
// enable set in BitmapValue
1181
DECLARE_Bool(enable_set_in_bitmap_value);
1182
1183
// max number of hdfs file handle in cache
1184
DECLARE_Int64(max_hdfs_file_handle_cache_num);
1185
DECLARE_Int32(max_hdfs_file_handle_cache_time_sec);
1186
1187
// max number of meta info of external files, such as parquet footer
1188
DECLARE_Int64(max_external_file_meta_cache_num);
1189
// Apply delete pred in cumu compaction
1190
DECLARE_mBool(enable_delete_when_cumu_compaction);
1191
1192
// max_write_buffer_number for rocksdb
1193
DECLARE_Int32(rocksdb_max_write_buffer_number);
1194
1195
// Convert date 0000-00-00 to 0000-01-01. It's recommended to set to false.
1196
DECLARE_mBool(allow_zero_date);
1197
// Allow invalid decimalv2 literal for compatible with old version. Recommend set it false strongly.
1198
DECLARE_mBool(allow_invalid_decimalv2_literal);
1199
// Allow to specify kerberos credentials cache path.
1200
DECLARE_mString(kerberos_ccache_path);
1201
// set krb5.conf path, use "/etc/krb5.conf" by default
1202
DECLARE_mString(kerberos_krb5_conf_path);
1203
1204
// Values include `none`, `glog`, `boost`, `glibc`, `libunwind`
1205
DECLARE_mString(get_stack_trace_tool);
1206
DECLARE_mBool(enable_address_sanitizers_with_stack_trace);
1207
1208
// DISABLED: Don't resolve location info.
1209
// FAST: Perform CU lookup using .debug_aranges (might be incomplete).
1210
// FULL: Scan all CU in .debug_info (slow!) on .debug_aranges lookup failure.
1211
// FULL_WITH_INLINE: Scan .debug_info (super slower, use with caution) for inline functions in addition to FULL.
1212
DECLARE_mString(dwarf_location_info_mode);
1213
1214
// the ratio of _prefetch_size/_batch_size in AutoIncIDBuffer
1215
DECLARE_mInt64(auto_inc_prefetch_size_ratio);
1216
1217
// the ratio of _low_level_water_level_mark/_batch_size in AutoIncIDBuffer
1218
DECLARE_mInt64(auto_inc_low_water_level_mark_size_ratio);
1219
1220
// number of threads that fetch auto-inc ranges from FE
1221
DECLARE_mInt64(auto_inc_fetch_thread_num);
1222
// Max connection cache num for point lookup queries
1223
DECLARE_mInt64(lookup_connection_cache_bytes_limit);
1224
1225
// level of compression when using LZ4_HC, whose defalut value is LZ4HC_CLEVEL_DEFAULT
1226
DECLARE_mInt64(LZ4_HC_compression_level);
1227
// Whether flatten nested arrays in variant column
1228
// Notice: TEST ONLY
1229
DECLARE_mBool(variant_enable_flatten_nested);
1230
// Threshold of a column as sparse column
1231
// Notice: TEST ONLY
1232
DECLARE_mDouble(variant_ratio_of_defaults_as_sparse_column);
1233
// Threshold to estimate a column is sparsed
1234
// Notice: TEST ONLY
1235
DECLARE_mInt64(variant_threshold_rows_to_estimate_sparse_column);
1236
// Treat invalid json format str as string, instead of throwing exception if false
1237
DECLARE_mBool(variant_throw_exeception_on_invalid_json);
1238
1239
DECLARE_mBool(enable_merge_on_write_correctness_check);
1240
// rowid conversion correctness check when compaction for mow table
1241
DECLARE_mBool(enable_rowid_conversion_correctness_check);
1242
// missing rows correctness check when compaction for mow table
1243
DECLARE_mBool(enable_missing_rows_correctness_check);
1244
// When the number of missing versions is more than this value, do not directly
1245
// retry the publish and handle it through async publish.
1246
DECLARE_mInt32(mow_publish_max_discontinuous_version_num);
1247
// When the version is not continuous for MOW table in publish phase and the gap between
1248
// current txn's publishing version and the max version of the tablet exceeds this value,
1249
// don't print warning log
1250
DECLARE_mInt32(publish_version_gap_logging_threshold);
1251
1252
// The secure path with user files, used in the `local` table function.
1253
DECLARE_mString(user_files_secure_path);
1254
1255
// If fe's frontend info has not been updated for more than fe_expire_duration_seconds, it will be regarded
1256
// as an abnormal fe, this will cause be to cancel this fe's related query.
1257
DECLARE_Int32(fe_expire_duration_seconds);
1258
1259
// If use stop_be.sh --grace, then BE has to wait all running queries to stop to avoiding running query failure
1260
// , but if the waiting time exceed the limit, then be will exit directly.
1261
// During this period, FE will not send any queries to BE and waiting for all running queries to stop.
1262
DECLARE_Int32(grace_shutdown_wait_seconds);
1263
1264
// BitmapValue serialize version.
1265
DECLARE_Int16(bitmap_serialize_version);
1266
1267
// group commit config
1268
DECLARE_String(group_commit_wal_path);
1269
DECLARE_Int32(group_commit_replay_wal_retry_num);
1270
DECLARE_Int32(group_commit_replay_wal_retry_interval_seconds);
1271
DECLARE_Int32(group_commit_replay_wal_retry_interval_max_seconds);
1272
DECLARE_mInt32(group_commit_relay_wal_threads);
1273
// This config can be set to limit thread number in group commit request fragment thread pool.
1274
DECLARE_mInt32(group_commit_insert_threads);
1275
DECLARE_mInt32(group_commit_memory_rows_for_max_filter_ratio);
1276
DECLARE_Bool(wait_internal_group_commit_finish);
1277
// Max size(bytes) of group commit queues, used for mem back pressure.
1278
DECLARE_mInt32(group_commit_queue_mem_limit);
1279
// Max size(bytes) or percentage(%) of wal disk usage, used for disk space back pressure, default 10% of the disk available space.
1280
// group_commit_wal_max_disk_limit=1024 or group_commit_wal_max_disk_limit=10% can be automatically identified.
1281
DECLARE_mString(group_commit_wal_max_disk_limit);
1282
DECLARE_Bool(group_commit_wait_replay_wal_finish);
1283
1284
// The configuration item is used to lower the priority of the scanner thread,
1285
// typically employed to ensure CPU scheduling for write operations.
1286
// Default is 0, which is default value of thread nice value, increase this value
1287
// to lower the priority of scan threads
1288
DECLARE_Int32(scan_thread_nice_value);
1289
// Used to modify the recycle interval of tablet schema cache
1290
DECLARE_mInt32(tablet_schema_cache_recycle_interval);
1291
// Granularity is at the column level
1292
DECLARE_mInt32(tablet_schema_cache_capacity);
1293
1294
// Use `LOG(FATAL)` to replace `throw` when true
1295
DECLARE_mBool(exit_on_exception);
1296
1297
// cgroup
1298
DECLARE_String(doris_cgroup_cpu_path);
1299
1300
DECLARE_Int32(workload_group_metrics_interval_ms);
1301
1302
DECLARE_mBool(enable_workload_group_memory_gc);
1303
1304
// This config controls whether the s3 file writer would flush cache asynchronously
1305
DECLARE_Bool(enable_flush_file_cache_async);
1306
1307
// Remove predicate that is always true for a segment.
1308
DECLARE_Bool(ignore_always_true_predicate_for_segment);
1309
1310
// Ingest binlog work pool size
1311
DECLARE_Int32(ingest_binlog_work_pool_size);
1312
1313
// Ingest binlog with persistent connection
1314
DECLARE_Bool(enable_ingest_binlog_with_persistent_connection);
1315
1316
// Download binlog rate limit, unit is KB/s
1317
DECLARE_Int32(download_binlog_rate_limit_kbs);
1318
1319
DECLARE_mInt32(buffered_reader_read_timeout_ms);
1320
1321
// whether to enable /api/snapshot api
1322
DECLARE_Bool(enable_snapshot_action);
1323
1324
// The max columns size for a tablet schema
1325
DECLARE_mInt32(variant_max_merged_tablet_schema_size);
1326
1327
DECLARE_mInt64(local_exchange_buffer_mem_limit);
1328
1329
DECLARE_mInt64(enable_debug_log_timeout_secs);
1330
1331
DECLARE_mBool(enable_column_type_check);
1332
1333
// Tolerance for the number of partition id 0 in rowset, default 0
1334
DECLARE_Int32(ignore_invalid_partition_id_rowset_num);
1335
1336
DECLARE_mInt32(report_query_statistics_interval_ms);
1337
DECLARE_mInt32(query_statistics_reserve_timeout_ms);
1338
1339
// consider two high usage disk at the same available level if they do not exceed this diff.
1340
DECLARE_mDouble(high_disk_avail_level_diff_usages);
1341
1342
// create tablet in partition random robin idx lru size, default 10000
1343
DECLARE_Int32(partition_disk_index_lru_size);
1344
DECLARE_String(spill_storage_root_path);
1345
// Spill storage limit specified as number of bytes
1346
// ('<int>[bB]?'), megabytes ('<float>[mM]'), gigabytes ('<float>[gG]'),
1347
// or percentage of capaity ('<int>%').
1348
// Defaults to bytes if no unit is given.
1349
// Must larger than 0.
1350
// If specified as percentage, the final limit value is:
1351
//   disk_capacity_bytes * storage_flood_stage_usage_percent * spill_storage_limit
1352
DECLARE_String(spill_storage_limit);
1353
DECLARE_mInt32(spill_gc_interval_ms);
1354
DECLARE_mInt32(spill_gc_work_time_ms);
1355
DECLARE_Int32(spill_io_thread_pool_thread_num);
1356
DECLARE_Int32(spill_io_thread_pool_queue_size);
1357
1358
DECLARE_mBool(check_segment_when_build_rowset_meta);
1359
DECLARE_Int32(num_query_ctx_map_partitions);
1360
// max s3 client retry times
1361
DECLARE_mInt32(max_s3_client_retry);
1362
// When meet s3 429 error, the "get" request will
1363
// sleep s3_read_base_wait_time_ms (*1, *2, *3, *4) ms
1364
// get try again.
1365
// The max sleep time is s3_read_max_wait_time_ms
1366
// and the max retry time is max_s3_client_retry
1367
DECLARE_mInt32(s3_read_base_wait_time_ms);
1368
DECLARE_mInt32(s3_read_max_wait_time_ms);
1369
1370
// write as inverted index tmp directory
1371
DECLARE_String(tmp_file_dir);
1372
1373
// the file paths(one or more) of CA cert, splite using ";" aws s3 lib use it to init s3client
1374
DECLARE_mString(ca_cert_file_paths);
1375
1376
/** Table sink configurations(currently contains only external table types) **/
1377
// Minimum data processed to scale writers in exchange when non partition writing
1378
DECLARE_mInt64(table_sink_non_partition_write_scaling_data_processed_threshold);
1379
// Minimum data processed to trigger skewed partition rebalancing in exchange when partition writing
1380
DECLARE_mInt64(table_sink_partition_write_min_data_processed_rebalance_threshold);
1381
// Minimum partition data processed to rebalance writers in exchange when partition writing
1382
DECLARE_mInt64(table_sink_partition_write_min_partition_data_processed_rebalance_threshold);
1383
// Maximum processed partition nums of per writer when partition writing
1384
DECLARE_mInt32(table_sink_partition_write_max_partition_nums_per_writer);
1385
1386
/** Hive sink configurations **/
1387
DECLARE_mInt64(hive_sink_max_file_size);
1388
1389
/** Iceberg sink configurations **/
1390
DECLARE_mInt64(iceberg_sink_max_file_size);
1391
1392
// Number of open tries, default 1 means only try to open once.
1393
// Retry the Open num_retries time waiting 100 milliseconds between retries.
1394
DECLARE_mInt32(thrift_client_open_num_tries);
1395
1396
DECLARE_mBool(ignore_schema_change_check);
1397
1398
//JVM monitoring enable. To prevent be from crashing due to jvm compatibility issues.
1399
DECLARE_Bool(enable_jvm_monitor);
1400
1401
// Skip loading stale rowset meta when initializing `TabletMeta` from protobuf
1402
DECLARE_mBool(skip_loading_stale_rowset_meta);
1403
// Whether to use file to record log. When starting BE with --console,
1404
// all logs will be written to both standard output and file.
1405
// Disable this option will no longer use file to record log.
1406
// Only works when starting BE with --console.
1407
DECLARE_Bool(enable_file_logger);
1408
// The time out milliseconds for remote fetch schema RPC
1409
DECLARE_mInt64(fetch_remote_schema_rpc_timeout_ms);
1410
1411
// The minimum row group size when exporting Parquet files.
1412
DECLARE_Int64(min_row_group_size);
1413
1414
DECLARE_mInt64(compaction_memory_bytes_limit);
1415
1416
DECLARE_mInt64(compaction_batch_size);
1417
1418
DECLARE_mBool(enable_parquet_page_index);
1419
1420
// Wheather to ignore not found file in external teble(eg, hive)
1421
// Default is true, if set to false, the not found file will result in query failure.
1422
DECLARE_mBool(ignore_not_found_file_in_external_table);
1423
1424
DECLARE_mInt64(tablet_meta_serialize_size_limit);
1425
1426
DECLARE_mInt64(pipeline_task_leakage_detect_period_secs);
1427
// To be compatible with hadoop's block compression
1428
DECLARE_mInt32(snappy_compression_block_size);
1429
DECLARE_mInt32(lz4_compression_block_size);
1430
1431
DECLARE_mBool(enable_pipeline_task_leakage_detect);
1432
DECLARE_Bool(force_regenerate_rowsetid_on_start_error);
1433
1434
DECLARE_mInt32(compaction_num_per_round);
1435
1436
// Enable sleep 5s between delete cumulative compaction.
1437
DECLARE_mBool(enable_sleep_between_delete_cumu_compaction);
1438
1439
// whether to prune rows with delete sign = 1 in base compaction
1440
// ATTN: this config is only for test
1441
DECLARE_mBool(enable_prune_delete_sign_when_base_compaction);
1442
1443
#ifdef BE_TEST
1444
// test s3
1445
DECLARE_String(test_s3_resource);
1446
DECLARE_String(test_s3_ak);
1447
DECLARE_String(test_s3_sk);
1448
DECLARE_String(test_s3_endpoint);
1449
DECLARE_String(test_s3_region);
1450
DECLARE_String(test_s3_bucket);
1451
DECLARE_String(test_s3_prefix);
1452
#endif
1453
1454
class Register {
1455
public:
1456
    struct Field {
1457
        const char* type = nullptr;
1458
        const char* name = nullptr;
1459
        void* storage = nullptr;
1460
        const char* defval = nullptr;
1461
        bool valmutable = false;
1462
        Field(const char* ftype, const char* fname, void* fstorage, const char* fdefval,
1463
              bool fvalmutable)
1464
                : type(ftype),
1465
                  name(fname),
1466
                  storage(fstorage),
1467
                  defval(fdefval),
1468
557
                  valmutable(fvalmutable) {}
1469
    };
1470
1471
public:
1472
    static std::map<std::string, Field>* _s_field_map;
1473
1474
public:
1475
    Register(const char* ftype, const char* fname, void* fstorage, const char* fdefval,
1476
553
             bool fvalmutable) {
1477
553
        if (_s_field_map == nullptr) {
1478
1
            _s_field_map = new std::map<std::string, Field>();
1479
1
        }
1480
553
        Field field(ftype, fname, fstorage, fdefval, fvalmutable);
1481
553
        _s_field_map->insert(std::make_pair(std::string(fname), field));
1482
553
    }
1483
};
1484
1485
// RegisterConfValidator class is used to store validator function of registered config fields in
1486
// Register::_s_field_map.
1487
// If any validator return false when BE bootstart, the bootstart will be terminated.
1488
// If validator return false when use http API to update some config, the config will not
1489
// be modified and the API will return failure.
1490
class RegisterConfValidator {
1491
public:
1492
    // Validator for each config name.
1493
    static std::map<std::string, std::function<bool()>>* _s_field_validator;
1494
1495
public:
1496
    RegisterConfValidator(const char* fname, const std::function<bool()>& validator) {
1497
        if (_s_field_validator == nullptr) {
1498
            _s_field_validator = new std::map<std::string, std::function<bool()>>();
1499
        }
1500
        // register validator to _s_field_validator
1501
        _s_field_validator->insert(std::make_pair(std::string(fname), validator));
1502
    }
1503
};
1504
1505
// configuration properties load from config file.
1506
class Properties {
1507
public:
1508
    // load conf from file, if must_exist is true and file does not exist, return false
1509
    bool load(const char* conf_file, bool must_exist = true);
1510
1511
    // Find the config value by key from `file_conf_map`.
1512
    // If found, set `retval` to the config value,
1513
    // or set `retval` to `defstr`
1514
    // if retval is not set(in case defstr is nullptr), set is_retval_set to false
1515
    template <typename T>
1516
    bool get_or_default(const char* key, const char* defstr, T& retval, bool* is_retval_set,
1517
                        std::string& rawval) const;
1518
1519
    void set(const std::string& key, const std::string& val);
1520
1521
    void set_force(const std::string& key, const std::string& val);
1522
1523
    // dump props to conf file
1524
    Status dump(const std::string& conffile);
1525
1526
private:
1527
    std::map<std::string, std::string> file_conf_map;
1528
};
1529
1530
// full configurations.
1531
extern std::map<std::string, std::string>* full_conf_map;
1532
1533
extern std::mutex custom_conf_lock;
1534
1535
// Init the config from `conf_file`.
1536
// If fill_conf_map is true, the updated config will also update the `full_conf_map`.
1537
// If must_exist is true and `conf_file` does not exist, this function will return false.
1538
// If set_to_default is true, the config value will be set to default value if not found in `conf_file`.
1539
bool init(const char* conf_file, bool fill_conf_map = false, bool must_exist = true,
1540
          bool set_to_default = true);
1541
1542
Status set_config(const std::string& field, const std::string& value, bool need_persist = false,
1543
                  bool force = false);
1544
1545
Status persist_config(const std::string& field, const std::string& value);
1546
1547
std::mutex* get_mutable_string_config_lock();
1548
1549
std::vector<std::vector<std::string>> get_config_info();
1550
1551
Status set_fuzzy_configs();
1552
1553
void update_config(const std::string& field, const std::string& value);
1554
1555
} // namespace config
1556
} // namespace doris