forked from facebook/rocksdb
-
Notifications
You must be signed in to change notification settings - Fork 0
/
db_stress.cc
4725 lines (4291 loc) · 170 KB
/
db_stress.cc
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
// Copyright (c) 2011-present, Facebook, Inc. All rights reserved.
// This source code is licensed under both the GPLv2 (found in the
// COPYING file in the root directory) and Apache 2.0 License
// (found in the LICENSE.Apache file in the root directory).
//
// Copyright (c) 2011 The LevelDB Authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file. See the AUTHORS file for names of contributors.
//
// The test uses an array to compare against values written to the database.
// Keys written to the array are in 1:1 correspondence to the actual values in
// the database according to the formula in the function GenerateValue.
// Space is reserved in the array from 0 to FLAGS_max_key and values are
// randomly written/deleted/read from those positions. During verification we
// compare all the positions in the array. To shorten/elongate the running
// time, you could change the settings: FLAGS_max_key, FLAGS_ops_per_thread,
// (sometimes also FLAGS_threads).
//
// NOTE that if FLAGS_test_batches_snapshots is set, the test will have
// different behavior. See comment of the flag for details.
#ifndef GFLAGS
#include <cstdio>
int main() {
fprintf(stderr, "Please install gflags to run rocksdb tools\n");
return 1;
}
#else
#include <fcntl.h>
#include <stdio.h>
#include <stdlib.h>
#include <sys/types.h>
#include <algorithm>
#include <array>
#include <chrono>
#include <cinttypes>
#include <exception>
#include <queue>
#include <thread>
#include "db/db_impl/db_impl.h"
#include "db/version_set.h"
#include "hdfs/env_hdfs.h"
#include "logging/logging.h"
#include "monitoring/histogram.h"
#include "options/options_helper.h"
#include "port/port.h"
#include "rocksdb/cache.h"
#include "rocksdb/env.h"
#include "rocksdb/slice.h"
#include "rocksdb/slice_transform.h"
#include "rocksdb/statistics.h"
#include "rocksdb/utilities/backupable_db.h"
#include "rocksdb/utilities/checkpoint.h"
#include "rocksdb/utilities/db_ttl.h"
#include "rocksdb/utilities/debug.h"
#include "rocksdb/utilities/options_util.h"
#include "rocksdb/utilities/transaction.h"
#include "rocksdb/utilities/transaction_db.h"
#include "rocksdb/write_batch.h"
#include "util/coding.h"
#include "util/compression.h"
#include "util/crc32c.h"
#include "util/gflags_compat.h"
#include "util/mutexlock.h"
#include "util/random.h"
#include "util/string_util.h"
// SyncPoint is not supported in Released Windows Mode.
#if !(defined NDEBUG) || !defined(OS_WIN)
#include "test_util/sync_point.h"
#endif // !(defined NDEBUG) || !defined(OS_WIN)
#include "test_util/testutil.h"
#include "utilities/merge_operators.h"
using GFLAGS_NAMESPACE::ParseCommandLineFlags;
using GFLAGS_NAMESPACE::RegisterFlagValidator;
using GFLAGS_NAMESPACE::SetUsageMessage;
static const long KB = 1024;
static const int kRandomValueMaxFactor = 3;
static const int kValueMaxLen = 100;
static bool ValidateUint32Range(const char* flagname, uint64_t value) {
if (value > std::numeric_limits<uint32_t>::max()) {
fprintf(stderr,
"Invalid value for --%s: %lu, overflow\n",
flagname,
(unsigned long)value);
return false;
}
return true;
}
DEFINE_uint64(seed, 2341234, "Seed for PRNG");
static const bool FLAGS_seed_dummy __attribute__((__unused__)) =
RegisterFlagValidator(&FLAGS_seed, &ValidateUint32Range);
DEFINE_bool(read_only, false, "True if open DB in read-only mode during tests");
DEFINE_int64(max_key, 1 * KB* KB,
"Max number of key/values to place in database");
DEFINE_int32(column_families, 10, "Number of column families");
DEFINE_string(
options_file, "",
"The path to a RocksDB options file. If specified, then db_stress will "
"run with the RocksDB options in the default column family of the "
"specified options file. Note that, when an options file is provided, "
"db_stress will ignore the flag values for all options that may be passed "
"via options file.");
DEFINE_int64(
active_width, 0,
"Number of keys in active span of the key-range at any given time. The "
"span begins with its left endpoint at key 0, gradually moves rightwards, "
"and ends with its right endpoint at max_key. If set to 0, active_width "
"will be sanitized to be equal to max_key.");
// TODO(noetzli) Add support for single deletes
DEFINE_bool(test_batches_snapshots, false,
"If set, the test uses MultiGet(), MultiPut() and MultiDelete()"
" which read/write/delete multiple keys in a batch. In this mode,"
" we do not verify db content by comparing the content with the "
"pre-allocated array. Instead, we do partial verification inside"
" MultiGet() by checking various values in a batch. Benefit of"
" this mode:\n"
"\t(a) No need to acquire mutexes during writes (less cache "
"flushes in multi-core leading to speed up)\n"
"\t(b) No long validation at the end (more speed up)\n"
"\t(c) Test snapshot and atomicity of batch writes");
DEFINE_bool(atomic_flush, false,
"If set, enables atomic flush in the options.\n");
DEFINE_bool(test_cf_consistency, false,
"If set, runs the stress test dedicated to verifying writes to "
"multiple column families are consistent. Setting this implies "
"`atomic_flush=true` is set true if `disable_wal=false`.\n");
DEFINE_int32(threads, 32, "Number of concurrent threads to run.");
DEFINE_int32(ttl, -1,
"Opens the db with this ttl value if this is not -1. "
"Carefully specify a large value such that verifications on "
"deleted values don't fail");
DEFINE_int32(value_size_mult, 8,
"Size of value will be this number times rand_int(1,3) bytes");
DEFINE_int32(compaction_readahead_size, 0, "Compaction readahead size");
DEFINE_bool(enable_pipelined_write, false, "Pipeline WAL/memtable writes");
DEFINE_bool(verify_before_write, false, "Verify before write");
DEFINE_bool(histogram, false, "Print histogram of operation timings");
DEFINE_bool(destroy_db_initially, true,
"Destroys the database dir before start if this is true");
DEFINE_bool(verbose, false, "Verbose");
DEFINE_bool(progress_reports, true,
"If true, db_stress will report number of finished operations");
DEFINE_uint64(db_write_buffer_size, rocksdb::Options().db_write_buffer_size,
"Number of bytes to buffer in all memtables before compacting");
DEFINE_int32(write_buffer_size,
static_cast<int32_t>(rocksdb::Options().write_buffer_size),
"Number of bytes to buffer in memtable before compacting");
DEFINE_int32(max_write_buffer_number,
rocksdb::Options().max_write_buffer_number,
"The number of in-memory memtables. "
"Each memtable is of size FLAGS_write_buffer_size.");
DEFINE_int32(min_write_buffer_number_to_merge,
rocksdb::Options().min_write_buffer_number_to_merge,
"The minimum number of write buffers that will be merged together "
"before writing to storage. This is cheap because it is an "
"in-memory merge. If this feature is not enabled, then all these "
"write buffers are flushed to L0 as separate files and this "
"increases read amplification because a get request has to check "
"in all of these files. Also, an in-memory merge may result in "
"writing less data to storage if there are duplicate records in"
" each of these individual write buffers.");
DEFINE_int32(max_write_buffer_number_to_maintain,
rocksdb::Options().max_write_buffer_number_to_maintain,
"The total maximum number of write buffers to maintain in memory "
"including copies of buffers that have already been flushed. "
"Unlike max_write_buffer_number, this parameter does not affect "
"flushing. This controls the minimum amount of write history "
"that will be available in memory for conflict checking when "
"Transactions are used. If this value is too low, some "
"transactions may fail at commit time due to not being able to "
"determine whether there were any write conflicts. Setting this "
"value to 0 will cause write buffers to be freed immediately "
"after they are flushed. If this value is set to -1, "
"'max_write_buffer_number' will be used.");
DEFINE_int64(max_write_buffer_size_to_maintain,
rocksdb::Options().max_write_buffer_size_to_maintain,
"The total maximum size of write buffers to maintain in memory "
"including copies of buffers that have already been flushed. "
"Unlike max_write_buffer_number, this parameter does not affect "
"flushing. This controls the minimum amount of write history "
"that will be available in memory for conflict checking when "
"Transactions are used. If this value is too low, some "
"transactions may fail at commit time due to not being able to "
"determine whether there were any write conflicts. Setting this "
"value to 0 will cause write buffers to be freed immediately "
"after they are flushed. If this value is set to -1, "
"'max_write_buffer_number' will be used.");
DEFINE_double(memtable_prefix_bloom_size_ratio,
rocksdb::Options().memtable_prefix_bloom_size_ratio,
"creates prefix blooms for memtables, each with size "
"`write_buffer_size * memtable_prefix_bloom_size_ratio`.");
DEFINE_bool(memtable_whole_key_filtering,
rocksdb::Options().memtable_whole_key_filtering,
"Enable whole key filtering in memtables.");
DEFINE_int32(open_files, rocksdb::Options().max_open_files,
"Maximum number of files to keep open at the same time "
"(use default if == 0)");
DEFINE_int64(compressed_cache_size, -1,
"Number of bytes to use as a cache of compressed data."
" Negative means use default settings.");
DEFINE_int32(compaction_style, rocksdb::Options().compaction_style, "");
DEFINE_int32(level0_file_num_compaction_trigger,
rocksdb::Options().level0_file_num_compaction_trigger,
"Level0 compaction start trigger");
DEFINE_int32(level0_slowdown_writes_trigger,
rocksdb::Options().level0_slowdown_writes_trigger,
"Number of files in level-0 that will slow down writes");
DEFINE_int32(level0_stop_writes_trigger,
rocksdb::Options().level0_stop_writes_trigger,
"Number of files in level-0 that will trigger put stop.");
DEFINE_int32(block_size,
static_cast<int32_t>(rocksdb::BlockBasedTableOptions().block_size),
"Number of bytes in a block.");
DEFINE_int32(
format_version,
static_cast<int32_t>(rocksdb::BlockBasedTableOptions().format_version),
"Format version of SST files.");
DEFINE_int32(index_block_restart_interval,
rocksdb::BlockBasedTableOptions().index_block_restart_interval,
"Number of keys between restart points "
"for delta encoding of keys in index block.");
DEFINE_int32(max_background_compactions,
rocksdb::Options().max_background_compactions,
"The maximum number of concurrent background compactions "
"that can occur in parallel.");
DEFINE_int32(num_bottom_pri_threads, 0,
"The number of threads in the bottom-priority thread pool (used "
"by universal compaction only).");
DEFINE_int32(compaction_thread_pool_adjust_interval, 0,
"The interval (in milliseconds) to adjust compaction thread pool "
"size. Don't change it periodically if the value is 0.");
DEFINE_int32(compaction_thread_pool_variations, 2,
"Range of background thread pool size variations when adjusted "
"periodically.");
DEFINE_int32(max_background_flushes, rocksdb::Options().max_background_flushes,
"The maximum number of concurrent background flushes "
"that can occur in parallel.");
DEFINE_int32(universal_size_ratio, 0, "The ratio of file sizes that trigger"
" compaction in universal style");
DEFINE_int32(universal_min_merge_width, 0, "The minimum number of files to "
"compact in universal style compaction");
DEFINE_int32(universal_max_merge_width, 0, "The max number of files to compact"
" in universal style compaction");
DEFINE_int32(universal_max_size_amplification_percent, 0,
"The max size amplification for universal style compaction");
DEFINE_int32(clear_column_family_one_in, 1000000,
"With a chance of 1/N, delete a column family and then recreate "
"it again. If N == 0, never drop/create column families. "
"When test_batches_snapshots is true, this flag has no effect");
DEFINE_int32(set_options_one_in, 0,
"With a chance of 1/N, change some random options");
DEFINE_int32(set_in_place_one_in, 0,
"With a chance of 1/N, toggle in place support option");
DEFINE_int64(cache_size, 2LL * KB * KB * KB,
"Number of bytes to use as a cache of uncompressed data.");
DEFINE_bool(cache_index_and_filter_blocks, false,
"True if indexes/filters should be cached in block cache.");
DEFINE_bool(use_clock_cache, false,
"Replace default LRU block cache with clock cache.");
DEFINE_uint64(subcompactions, 1,
"Maximum number of subcompactions to divide L0-L1 compactions "
"into.");
DEFINE_uint64(periodic_compaction_seconds, 1000,
"Files older than this value will be picked up for compaction.");
DEFINE_uint64(compaction_ttl, 1000,
"Files older than TTL will be compacted to the next level.");
DEFINE_bool(allow_concurrent_memtable_write, false,
"Allow multi-writers to update mem tables in parallel.");
DEFINE_bool(enable_write_thread_adaptive_yield, true,
"Use a yielding spin loop for brief writer thread waits.");
static const bool FLAGS_subcompactions_dummy __attribute__((__unused__)) =
RegisterFlagValidator(&FLAGS_subcompactions, &ValidateUint32Range);
static bool ValidateInt32Positive(const char* flagname, int32_t value) {
if (value < 0) {
fprintf(stderr, "Invalid value for --%s: %d, must be >=0\n",
flagname, value);
return false;
}
return true;
}
DEFINE_int32(reopen, 10, "Number of times database reopens");
static const bool FLAGS_reopen_dummy __attribute__((__unused__)) =
RegisterFlagValidator(&FLAGS_reopen, &ValidateInt32Positive);
DEFINE_int32(bloom_bits, 10, "Bloom filter bits per key. "
"Negative means use default settings.");
DEFINE_bool(use_block_based_filter, false, "use block based filter"
"instead of full filter for block based table");
DEFINE_bool(partition_filters, false,
"use partitioned filters "
"for block-based table");
DEFINE_int32(
index_type,
static_cast<int32_t>(rocksdb::BlockBasedTableOptions::kBinarySearch),
"Type of block-based table index (see `enum IndexType` in table.h)");
DEFINE_string(db, "", "Use the db with the following name.");
DEFINE_string(secondaries_base, "",
"Use this path as the base path for secondary instances.");
DEFINE_bool(enable_secondary, false, "Enable secondary instance.");
DEFINE_string(
expected_values_path, "",
"File where the array of expected uint32_t values will be stored. If "
"provided and non-empty, the DB state will be verified against these "
"values after recovery. --max_key and --column_family must be kept the "
"same across invocations of this program that use the same "
"--expected_values_path.");
DEFINE_bool(verify_checksum, false,
"Verify checksum for every block read from storage");
DEFINE_bool(mmap_read, rocksdb::Options().allow_mmap_reads,
"Allow reads to occur via mmap-ing files");
DEFINE_bool(mmap_write, rocksdb::Options().allow_mmap_writes,
"Allow writes to occur via mmap-ing files");
DEFINE_bool(use_direct_reads, rocksdb::Options().use_direct_reads,
"Use O_DIRECT for reading data");
DEFINE_bool(use_direct_io_for_flush_and_compaction,
rocksdb::Options().use_direct_io_for_flush_and_compaction,
"Use O_DIRECT for writing data");
// Database statistics
static std::shared_ptr<rocksdb::Statistics> dbstats;
DEFINE_bool(statistics, false, "Create database statistics");
DEFINE_bool(sync, false, "Sync all writes to disk");
DEFINE_bool(use_fsync, false, "If true, issue fsync instead of fdatasync");
DEFINE_int32(kill_random_test, 0,
"If non-zero, kill at various points in source code with "
"probability 1/this");
static const bool FLAGS_kill_random_test_dummy __attribute__((__unused__)) =
RegisterFlagValidator(&FLAGS_kill_random_test, &ValidateInt32Positive);
extern int rocksdb_kill_odds;
DEFINE_string(kill_prefix_blacklist, "",
"If non-empty, kill points with prefix in the list given will be"
" skipped. Items are comma-separated.");
extern std::vector<std::string> rocksdb_kill_prefix_blacklist;
DEFINE_bool(disable_wal, false, "If true, do not write WAL for write.");
DEFINE_uint64(recycle_log_file_num, rocksdb::Options().recycle_log_file_num,
"Number of old WAL files to keep around for later recycling");
DEFINE_int64(target_file_size_base, rocksdb::Options().target_file_size_base,
"Target level-1 file size for compaction");
DEFINE_int32(target_file_size_multiplier, 1,
"A multiplier to compute target level-N file size (N >= 2)");
DEFINE_uint64(max_bytes_for_level_base,
rocksdb::Options().max_bytes_for_level_base,
"Max bytes for level-1");
DEFINE_double(max_bytes_for_level_multiplier, 2,
"A multiplier to compute max bytes for level-N (N >= 2)");
DEFINE_int32(range_deletion_width, 10,
"The width of the range deletion intervals.");
DEFINE_uint64(rate_limiter_bytes_per_sec, 0, "Set options.rate_limiter value.");
DEFINE_bool(rate_limit_bg_reads, false,
"Use options.rate_limiter on compaction reads");
DEFINE_bool(use_txn, false,
"Use TransactionDB. Currently the default write policy is "
"TxnDBWritePolicy::WRITE_PREPARED");
DEFINE_int32(backup_one_in, 0,
"If non-zero, then CreateNewBackup() will be called once for "
"every N operations on average. 0 indicates CreateNewBackup() "
"is disabled.");
DEFINE_int32(checkpoint_one_in, 0,
"If non-zero, then CreateCheckpoint() will be called once for "
"every N operations on average. 0 indicates CreateCheckpoint() "
"is disabled.");
DEFINE_int32(ingest_external_file_one_in, 0,
"If non-zero, then IngestExternalFile() will be called once for "
"every N operations on average. 0 indicates IngestExternalFile() "
"is disabled.");
DEFINE_int32(ingest_external_file_width, 1000,
"The width of the ingested external files.");
DEFINE_int32(compact_files_one_in, 0,
"If non-zero, then CompactFiles() will be called once for every N "
"operations on average. 0 indicates CompactFiles() is disabled.");
DEFINE_int32(compact_range_one_in, 0,
"If non-zero, then CompactRange() will be called once for every N "
"operations on average. 0 indicates CompactRange() is disabled.");
DEFINE_int32(flush_one_in, 0,
"If non-zero, then Flush() will be called once for every N ops "
"on average. 0 indicates calls to Flush() are disabled.");
DEFINE_int32(compact_range_width, 10000,
"The width of the ranges passed to CompactRange().");
DEFINE_int32(acquire_snapshot_one_in, 0,
"If non-zero, then acquires a snapshot once every N operations on "
"average.");
DEFINE_bool(compare_full_db_state_snapshot, false,
"If set we compare state of entire db (in one of the threads) with"
"each snapshot.");
DEFINE_uint64(snapshot_hold_ops, 0,
"If non-zero, then releases snapshots N operations after they're "
"acquired.");
DEFINE_bool(use_multiget, false,
"If set, use the batched MultiGet API for reads");
static bool ValidateInt32Percent(const char* flagname, int32_t value) {
if (value < 0 || value>100) {
fprintf(stderr, "Invalid value for --%s: %d, 0<= pct <=100 \n",
flagname, value);
return false;
}
return true;
}
DEFINE_int32(readpercent, 10,
"Ratio of reads to total workload (expressed as a percentage)");
static const bool FLAGS_readpercent_dummy __attribute__((__unused__)) =
RegisterFlagValidator(&FLAGS_readpercent, &ValidateInt32Percent);
DEFINE_int32(prefixpercent, 20,
"Ratio of prefix iterators to total workload (expressed as a"
" percentage)");
static const bool FLAGS_prefixpercent_dummy __attribute__((__unused__)) =
RegisterFlagValidator(&FLAGS_prefixpercent, &ValidateInt32Percent);
DEFINE_int32(writepercent, 45,
"Ratio of writes to total workload (expressed as a percentage)");
static const bool FLAGS_writepercent_dummy __attribute__((__unused__)) =
RegisterFlagValidator(&FLAGS_writepercent, &ValidateInt32Percent);
DEFINE_int32(delpercent, 15,
"Ratio of deletes to total workload (expressed as a percentage)");
static const bool FLAGS_delpercent_dummy __attribute__((__unused__)) =
RegisterFlagValidator(&FLAGS_delpercent, &ValidateInt32Percent);
DEFINE_int32(delrangepercent, 0,
"Ratio of range deletions to total workload (expressed as a "
"percentage). Cannot be used with test_batches_snapshots");
static const bool FLAGS_delrangepercent_dummy __attribute__((__unused__)) =
RegisterFlagValidator(&FLAGS_delrangepercent, &ValidateInt32Percent);
DEFINE_int32(nooverwritepercent, 60,
"Ratio of keys without overwrite to total workload (expressed as "
" a percentage)");
static const bool FLAGS_nooverwritepercent_dummy __attribute__((__unused__)) =
RegisterFlagValidator(&FLAGS_nooverwritepercent, &ValidateInt32Percent);
DEFINE_int32(iterpercent, 10, "Ratio of iterations to total workload"
" (expressed as a percentage)");
static const bool FLAGS_iterpercent_dummy __attribute__((__unused__)) =
RegisterFlagValidator(&FLAGS_iterpercent, &ValidateInt32Percent);
DEFINE_uint64(num_iterations, 10, "Number of iterations per MultiIterate run");
static const bool FLAGS_num_iterations_dummy __attribute__((__unused__)) =
RegisterFlagValidator(&FLAGS_num_iterations, &ValidateUint32Range);
namespace {
enum rocksdb::CompressionType StringToCompressionType(const char* ctype) {
assert(ctype);
if (!strcasecmp(ctype, "none"))
return rocksdb::kNoCompression;
else if (!strcasecmp(ctype, "snappy"))
return rocksdb::kSnappyCompression;
else if (!strcasecmp(ctype, "zlib"))
return rocksdb::kZlibCompression;
else if (!strcasecmp(ctype, "bzip2"))
return rocksdb::kBZip2Compression;
else if (!strcasecmp(ctype, "lz4"))
return rocksdb::kLZ4Compression;
else if (!strcasecmp(ctype, "lz4hc"))
return rocksdb::kLZ4HCCompression;
else if (!strcasecmp(ctype, "xpress"))
return rocksdb::kXpressCompression;
else if (!strcasecmp(ctype, "zstd"))
return rocksdb::kZSTD;
fprintf(stderr, "Cannot parse compression type '%s'\n", ctype);
return rocksdb::kSnappyCompression; //default value
}
enum rocksdb::ChecksumType StringToChecksumType(const char* ctype) {
assert(ctype);
auto iter = rocksdb::checksum_type_string_map.find(ctype);
if (iter != rocksdb::checksum_type_string_map.end()) {
return iter->second;
}
fprintf(stderr, "Cannot parse checksum type '%s'\n", ctype);
return rocksdb::kCRC32c;
}
std::string ChecksumTypeToString(rocksdb::ChecksumType ctype) {
auto iter = std::find_if(
rocksdb::checksum_type_string_map.begin(),
rocksdb::checksum_type_string_map.end(),
[&](const std::pair<std::string, rocksdb::ChecksumType>&
name_and_enum_val) { return name_and_enum_val.second == ctype; });
assert(iter != rocksdb::checksum_type_string_map.end());
return iter->first;
}
std::vector<std::string> SplitString(std::string src) {
std::vector<std::string> ret;
if (src.empty()) {
return ret;
}
size_t pos = 0;
size_t pos_comma;
while ((pos_comma = src.find(',', pos)) != std::string::npos) {
ret.push_back(src.substr(pos, pos_comma - pos));
pos = pos_comma + 1;
}
ret.push_back(src.substr(pos, src.length()));
return ret;
}
} // namespace
DEFINE_string(compression_type, "snappy",
"Algorithm to use to compress the database");
static enum rocksdb::CompressionType FLAGS_compression_type_e =
rocksdb::kSnappyCompression;
DEFINE_int32(compression_max_dict_bytes, 0,
"Maximum size of dictionary used to prime the compression "
"library.");
DEFINE_int32(compression_zstd_max_train_bytes, 0,
"Maximum size of training data passed to zstd's dictionary "
"trainer.");
DEFINE_string(checksum_type, "kCRC32c", "Algorithm to use to checksum blocks");
static enum rocksdb::ChecksumType FLAGS_checksum_type_e = rocksdb::kCRC32c;
DEFINE_string(hdfs, "", "Name of hdfs environment");
// posix or hdfs environment
static rocksdb::Env* FLAGS_env = rocksdb::Env::Default();
DEFINE_uint64(ops_per_thread, 1200000, "Number of operations per thread.");
static const bool FLAGS_ops_per_thread_dummy __attribute__((__unused__)) =
RegisterFlagValidator(&FLAGS_ops_per_thread, &ValidateUint32Range);
DEFINE_uint64(log2_keys_per_lock, 2, "Log2 of number of keys per lock");
static const bool FLAGS_log2_keys_per_lock_dummy __attribute__((__unused__)) =
RegisterFlagValidator(&FLAGS_log2_keys_per_lock, &ValidateUint32Range);
DEFINE_uint64(max_manifest_file_size, 16384, "Maximum size of a MANIFEST file");
DEFINE_bool(in_place_update, false, "On true, does inplace update in memtable");
DEFINE_int32(secondary_catch_up_one_in, 0,
"If non-zero, the secondaries attemp to catch up with the primary "
"once for every N operations on average. 0 indicates the "
"secondaries do not try to catch up after open.");
static std::shared_ptr<rocksdb::Statistics> dbstats_secondaries;
enum RepFactory {
kSkipList,
kHashSkipList,
kVectorRep
};
namespace {
enum RepFactory StringToRepFactory(const char* ctype) {
assert(ctype);
if (!strcasecmp(ctype, "skip_list"))
return kSkipList;
else if (!strcasecmp(ctype, "prefix_hash"))
return kHashSkipList;
else if (!strcasecmp(ctype, "vector"))
return kVectorRep;
fprintf(stdout, "Cannot parse memreptable %s\n", ctype);
return kSkipList;
}
#ifdef _MSC_VER
#pragma warning(push)
// truncation of constant value on static_cast
#pragma warning(disable : 4309)
#endif
bool GetNextPrefix(const rocksdb::Slice& src, std::string* v) {
std::string ret = src.ToString();
for (int i = static_cast<int>(ret.size()) - 1; i >= 0; i--) {
if (ret[i] != static_cast<char>(255)) {
ret[i] = ret[i] + 1;
break;
} else if (i != 0) {
ret[i] = 0;
} else {
// all FF. No next prefix
return false;
}
}
*v = ret;
return true;
}
#ifdef _MSC_VER
#pragma warning(pop)
#endif
} // namespace
static enum RepFactory FLAGS_rep_factory;
DEFINE_string(memtablerep, "prefix_hash", "");
static bool ValidatePrefixSize(const char* flagname, int32_t value) {
if (value < -1 || value > 8) {
fprintf(stderr, "Invalid value for --%s: %d. -1 <= PrefixSize <= 8\n",
flagname, value);
return false;
}
return true;
}
DEFINE_int32(prefix_size, 7,
"Control the prefix size for HashSkipListRep. "
"-1 is disabled.");
static const bool FLAGS_prefix_size_dummy __attribute__((__unused__)) =
RegisterFlagValidator(&FLAGS_prefix_size, &ValidatePrefixSize);
DEFINE_bool(use_merge, false, "On true, replaces all writes with a Merge "
"that behaves like a Put");
DEFINE_bool(use_full_merge_v1, false,
"On true, use a merge operator that implement the deprecated "
"version of FullMerge");
namespace rocksdb {
// convert long to a big-endian slice key
static std::string Key(int64_t val) {
std::string little_endian_key;
std::string big_endian_key;
PutFixed64(&little_endian_key, val);
assert(little_endian_key.size() == sizeof(val));
big_endian_key.resize(sizeof(val));
for (size_t i = 0 ; i < sizeof(val); ++i) {
big_endian_key[i] = little_endian_key[sizeof(val) - 1 - i];
}
return big_endian_key;
}
static bool GetIntVal(std::string big_endian_key, uint64_t *key_p) {
unsigned int size_key = sizeof(*key_p);
assert(big_endian_key.size() == size_key);
std::string little_endian_key;
little_endian_key.resize(size_key);
for (size_t i = 0 ; i < size_key; ++i) {
little_endian_key[i] = big_endian_key[size_key - 1 - i];
}
Slice little_endian_slice = Slice(little_endian_key);
return GetFixed64(&little_endian_slice, key_p);
}
static std::string StringToHex(const std::string& str) {
std::string result = "0x";
result.append(Slice(str).ToString(true));
return result;
}
class StressTest;
namespace {
class Stats {
private:
uint64_t start_;
uint64_t finish_;
double seconds_;
long done_;
long gets_;
long prefixes_;
long writes_;
long deletes_;
size_t single_deletes_;
long iterator_size_sums_;
long founds_;
long iterations_;
long range_deletions_;
long covered_by_range_deletions_;
long errors_;
long num_compact_files_succeed_;
long num_compact_files_failed_;
int next_report_;
size_t bytes_;
uint64_t last_op_finish_;
HistogramImpl hist_;
public:
Stats() { }
void Start() {
next_report_ = 100;
hist_.Clear();
done_ = 0;
gets_ = 0;
prefixes_ = 0;
writes_ = 0;
deletes_ = 0;
single_deletes_ = 0;
iterator_size_sums_ = 0;
founds_ = 0;
iterations_ = 0;
range_deletions_ = 0;
covered_by_range_deletions_ = 0;
errors_ = 0;
bytes_ = 0;
seconds_ = 0;
num_compact_files_succeed_ = 0;
num_compact_files_failed_ = 0;
start_ = FLAGS_env->NowMicros();
last_op_finish_ = start_;
finish_ = start_;
}
void Merge(const Stats& other) {
hist_.Merge(other.hist_);
done_ += other.done_;
gets_ += other.gets_;
prefixes_ += other.prefixes_;
writes_ += other.writes_;
deletes_ += other.deletes_;
single_deletes_ += other.single_deletes_;
iterator_size_sums_ += other.iterator_size_sums_;
founds_ += other.founds_;
iterations_ += other.iterations_;
range_deletions_ += other.range_deletions_;
covered_by_range_deletions_ = other.covered_by_range_deletions_;
errors_ += other.errors_;
bytes_ += other.bytes_;
seconds_ += other.seconds_;
num_compact_files_succeed_ += other.num_compact_files_succeed_;
num_compact_files_failed_ += other.num_compact_files_failed_;
if (other.start_ < start_) start_ = other.start_;
if (other.finish_ > finish_) finish_ = other.finish_;
}
void Stop() {
finish_ = FLAGS_env->NowMicros();
seconds_ = (finish_ - start_) * 1e-6;
}
void FinishedSingleOp() {
if (FLAGS_histogram) {
auto now = FLAGS_env->NowMicros();
auto micros = now - last_op_finish_;
hist_.Add(micros);
if (micros > 20000) {
fprintf(stdout, "long op: %" PRIu64 " micros%30s\r", micros, "");
}
last_op_finish_ = now;
}
done_++;
if (FLAGS_progress_reports) {
if (done_ >= next_report_) {
if (next_report_ < 1000) next_report_ += 100;
else if (next_report_ < 5000) next_report_ += 500;
else if (next_report_ < 10000) next_report_ += 1000;
else if (next_report_ < 50000) next_report_ += 5000;
else if (next_report_ < 100000) next_report_ += 10000;
else if (next_report_ < 500000) next_report_ += 50000;
else next_report_ += 100000;
fprintf(stdout, "... finished %ld ops%30s\r", done_, "");
}
}
}
void AddBytesForWrites(long nwrites, size_t nbytes) {
writes_ += nwrites;
bytes_ += nbytes;
}
void AddGets(long ngets, long nfounds) {
founds_ += nfounds;
gets_ += ngets;
}
void AddPrefixes(long nprefixes, long count) {
prefixes_ += nprefixes;
iterator_size_sums_ += count;
}
void AddIterations(long n) { iterations_ += n; }
void AddDeletes(long n) { deletes_ += n; }
void AddSingleDeletes(size_t n) { single_deletes_ += n; }
void AddRangeDeletions(long n) { range_deletions_ += n; }
void AddCoveredByRangeDeletions(long n) { covered_by_range_deletions_ += n; }
void AddErrors(long n) { errors_ += n; }
void AddNumCompactFilesSucceed(long n) { num_compact_files_succeed_ += n; }
void AddNumCompactFilesFailed(long n) { num_compact_files_failed_ += n; }
void Report(const char* name) {
std::string extra;
if (bytes_ < 1 || done_ < 1) {
fprintf(stderr, "No writes or ops?\n");
return;
}
double elapsed = (finish_ - start_) * 1e-6;
double bytes_mb = bytes_ / 1048576.0;
double rate = bytes_mb / elapsed;
double throughput = (double)done_/elapsed;
fprintf(stdout, "%-12s: ", name);
fprintf(stdout, "%.3f micros/op %ld ops/sec\n",
seconds_ * 1e6 / done_, (long)throughput);
fprintf(stdout, "%-12s: Wrote %.2f MB (%.2f MB/sec) (%ld%% of %ld ops)\n",
"", bytes_mb, rate, (100*writes_)/done_, done_);
fprintf(stdout, "%-12s: Wrote %ld times\n", "", writes_);
fprintf(stdout, "%-12s: Deleted %ld times\n", "", deletes_);
fprintf(stdout, "%-12s: Single deleted %" ROCKSDB_PRIszt " times\n", "",
single_deletes_);
fprintf(stdout, "%-12s: %ld read and %ld found the key\n", "",
gets_, founds_);
fprintf(stdout, "%-12s: Prefix scanned %ld times\n", "", prefixes_);
fprintf(stdout, "%-12s: Iterator size sum is %ld\n", "",
iterator_size_sums_);
fprintf(stdout, "%-12s: Iterated %ld times\n", "", iterations_);
fprintf(stdout, "%-12s: Deleted %ld key-ranges\n", "", range_deletions_);
fprintf(stdout, "%-12s: Range deletions covered %ld keys\n", "",
covered_by_range_deletions_);
fprintf(stdout, "%-12s: Got errors %ld times\n", "", errors_);
fprintf(stdout, "%-12s: %ld CompactFiles() succeed\n", "",
num_compact_files_succeed_);
fprintf(stdout, "%-12s: %ld CompactFiles() did not succeed\n", "",
num_compact_files_failed_);
if (FLAGS_histogram) {
fprintf(stdout, "Microseconds per op:\n%s\n", hist_.ToString().c_str());
}
fflush(stdout);
}
};
// State shared by all concurrent executions of the same benchmark.
class SharedState {
public:
// indicates a key may have any value (or not be present) as an operation on
// it is incomplete.
static const uint32_t UNKNOWN_SENTINEL;
// indicates a key should definitely be deleted
static const uint32_t DELETION_SENTINEL;
explicit SharedState(StressTest* stress_test)
: cv_(&mu_),
seed_(static_cast<uint32_t>(FLAGS_seed)),
max_key_(FLAGS_max_key),
log2_keys_per_lock_(static_cast<uint32_t>(FLAGS_log2_keys_per_lock)),
num_threads_(FLAGS_threads),
num_initialized_(0),
num_populated_(0),
vote_reopen_(0),
num_done_(0),
start_(false),
start_verify_(false),
should_stop_bg_thread_(false),
bg_thread_finished_(false),
stress_test_(stress_test),
verification_failure_(false),
no_overwrite_ids_(FLAGS_column_families),
values_(nullptr),
printing_verification_results_(false) {
// Pick random keys in each column family that will not experience
// overwrite
printf("Choosing random keys with no overwrite\n");
Random64 rnd(seed_);
// Start with the identity permutation. Subsequent iterations of
// for loop below will start with perm of previous for loop
int64_t *permutation = new int64_t[max_key_];
for (int64_t i = 0; i < max_key_; i++) {
permutation[i] = i;
}
// Now do the Knuth shuffle
int64_t num_no_overwrite_keys = (max_key_ * FLAGS_nooverwritepercent) / 100;
// Only need to figure out first num_no_overwrite_keys of permutation
no_overwrite_ids_.reserve(num_no_overwrite_keys);
for (int64_t i = 0; i < num_no_overwrite_keys; i++) {
int64_t rand_index = i + rnd.Next() % (max_key_ - i);
// Swap i and rand_index;
int64_t temp = permutation[i];
permutation[i] = permutation[rand_index];
permutation[rand_index] = temp;
// Fill no_overwrite_ids_ with the first num_no_overwrite_keys of
// permutation
no_overwrite_ids_.insert(permutation[i]);
}
delete[] permutation;
size_t expected_values_size =
sizeof(std::atomic<uint32_t>) * FLAGS_column_families * max_key_;
bool values_init_needed = false;
Status status;
if (!FLAGS_expected_values_path.empty()) {
if (!std::atomic<uint32_t>{}.is_lock_free()) {
status = Status::InvalidArgument(
"Cannot use --expected_values_path on platforms without lock-free "
"std::atomic<uint32_t>");
}
if (status.ok() && FLAGS_clear_column_family_one_in > 0) {
status = Status::InvalidArgument(
"Cannot use --expected_values_path on when "
"--clear_column_family_one_in is greater than zero.");