forked from facebook/rocksdb
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathdb_compaction_test.cc
2561 lines (2190 loc) · 83.6 KB
/
db_compaction_test.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 the BSD-style license found in the
// LICENSE file in the root directory of this source tree. An additional grant
// of patent rights can be found in the PATENTS file in the same 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.
#include "db/db_test_util.h"
#include "port/stack_trace.h"
#include "rocksdb/experimental.h"
#include "rocksdb/utilities/convenience.h"
#include "util/sync_point.h"
namespace rocksdb {
// SYNC_POINT is not supported in released Windows mode.
#if !defined(ROCKSDB_LITE)
class DBCompactionTest : public DBTestBase {
public:
DBCompactionTest() : DBTestBase("/db_compaction_test") {}
};
class DBCompactionTestWithParam
: public DBTestBase,
public testing::WithParamInterface<std::tuple<uint32_t, bool>> {
public:
DBCompactionTestWithParam() : DBTestBase("/db_compaction_test") {
max_subcompactions_ = std::get<0>(GetParam());
exclusive_manual_compaction_ = std::get<1>(GetParam());
}
// Required if inheriting from testing::WithParamInterface<>
static void SetUpTestCase() {}
static void TearDownTestCase() {}
uint32_t max_subcompactions_;
bool exclusive_manual_compaction_;
};
namespace {
class FlushedFileCollector : public EventListener {
public:
FlushedFileCollector() {}
~FlushedFileCollector() {}
virtual void OnFlushCompleted(DB* db, const FlushJobInfo& info) override {
std::lock_guard<std::mutex> lock(mutex_);
flushed_files_.push_back(info.file_path);
}
std::vector<std::string> GetFlushedFiles() {
std::lock_guard<std::mutex> lock(mutex_);
std::vector<std::string> result;
for (auto fname : flushed_files_) {
result.push_back(fname);
}
return result;
}
void ClearFlushedFiles() { flushed_files_.clear(); }
private:
std::vector<std::string> flushed_files_;
std::mutex mutex_;
};
static const int kCDTValueSize = 1000;
static const int kCDTKeysPerBuffer = 4;
static const int kCDTNumLevels = 8;
Options DeletionTriggerOptions(Options options) {
options.compression = kNoCompression;
options.write_buffer_size = kCDTKeysPerBuffer * (kCDTValueSize + 24);
options.min_write_buffer_number_to_merge = 1;
options.max_write_buffer_number_to_maintain = 0;
options.num_levels = kCDTNumLevels;
options.level0_file_num_compaction_trigger = 1;
options.target_file_size_base = options.write_buffer_size * 2;
options.target_file_size_multiplier = 2;
options.max_bytes_for_level_base =
options.target_file_size_base * options.target_file_size_multiplier;
options.max_bytes_for_level_multiplier = 2;
options.disable_auto_compactions = false;
return options;
}
bool HaveOverlappingKeyRanges(
const Comparator* c,
const SstFileMetaData& a, const SstFileMetaData& b) {
if (c->Compare(a.smallestkey, b.smallestkey) >= 0) {
if (c->Compare(a.smallestkey, b.largestkey) <= 0) {
// b.smallestkey <= a.smallestkey <= b.largestkey
return true;
}
} else if (c->Compare(a.largestkey, b.smallestkey) >= 0) {
// a.smallestkey < b.smallestkey <= a.largestkey
return true;
}
if (c->Compare(a.largestkey, b.largestkey) <= 0) {
if (c->Compare(a.largestkey, b.smallestkey) >= 0) {
// b.smallestkey <= a.largestkey <= b.largestkey
return true;
}
} else if (c->Compare(a.smallestkey, b.largestkey) <= 0) {
// a.smallestkey <= b.largestkey < a.largestkey
return true;
}
return false;
}
// Identifies all files between level "min_level" and "max_level"
// which has overlapping key range with "input_file_meta".
void GetOverlappingFileNumbersForLevelCompaction(
const ColumnFamilyMetaData& cf_meta,
const Comparator* comparator,
int min_level, int max_level,
const SstFileMetaData* input_file_meta,
std::set<std::string>* overlapping_file_names) {
std::set<const SstFileMetaData*> overlapping_files;
overlapping_files.insert(input_file_meta);
for (int m = min_level; m <= max_level; ++m) {
for (auto& file : cf_meta.levels[m].files) {
for (auto* included_file : overlapping_files) {
if (HaveOverlappingKeyRanges(
comparator, *included_file, file)) {
overlapping_files.insert(&file);
overlapping_file_names->insert(file.name);
break;
}
}
}
}
}
void VerifyCompactionResult(
const ColumnFamilyMetaData& cf_meta,
const std::set<std::string>& overlapping_file_numbers) {
#ifndef NDEBUG
for (auto& level : cf_meta.levels) {
for (auto& file : level.files) {
assert(overlapping_file_numbers.find(file.name) ==
overlapping_file_numbers.end());
}
}
#endif
}
const SstFileMetaData* PickFileRandomly(
const ColumnFamilyMetaData& cf_meta,
Random* rand,
int* level = nullptr) {
auto file_id = rand->Uniform(static_cast<int>(
cf_meta.file_count)) + 1;
for (auto& level_meta : cf_meta.levels) {
if (file_id <= level_meta.files.size()) {
if (level != nullptr) {
*level = level_meta.level;
}
auto result = rand->Uniform(file_id);
return &(level_meta.files[result]);
}
file_id -= static_cast<uint32_t>(level_meta.files.size());
}
assert(false);
return nullptr;
}
} // anonymous namespace
// All the TEST_P tests run once with sub_compactions disabled (i.e.
// options.max_subcompactions = 1) and once with it enabled
TEST_P(DBCompactionTestWithParam, CompactionDeletionTrigger) {
for (int tid = 0; tid < 3; ++tid) {
uint64_t db_size[2];
Options options = DeletionTriggerOptions(CurrentOptions());
options.max_subcompactions = max_subcompactions_;
if (tid == 1) {
// the following only disable stats update in DB::Open()
// and should not affect the result of this test.
options.skip_stats_update_on_db_open = true;
} else if (tid == 2) {
// third pass with universal compaction
options.compaction_style = kCompactionStyleUniversal;
options.num_levels = 1;
}
DestroyAndReopen(options);
Random rnd(301);
const int kTestSize = kCDTKeysPerBuffer * 1024;
std::vector<std::string> values;
for (int k = 0; k < kTestSize; ++k) {
values.push_back(RandomString(&rnd, kCDTValueSize));
ASSERT_OK(Put(Key(k), values[k]));
}
dbfull()->TEST_WaitForFlushMemTable();
dbfull()->TEST_WaitForCompact();
db_size[0] = Size(Key(0), Key(kTestSize - 1));
for (int k = 0; k < kTestSize; ++k) {
ASSERT_OK(Delete(Key(k)));
}
dbfull()->TEST_WaitForFlushMemTable();
dbfull()->TEST_WaitForCompact();
db_size[1] = Size(Key(0), Key(kTestSize - 1));
// must have much smaller db size.
ASSERT_GT(db_size[0] / 3, db_size[1]);
}
}
TEST_F(DBCompactionTest, SkipStatsUpdateTest) {
// This test verify UpdateAccumulatedStats is not on
// if options.skip_stats_update_on_db_open = true
// The test will need to be updated if the internal behavior changes.
Options options = DeletionTriggerOptions(CurrentOptions());
options.env = env_;
DestroyAndReopen(options);
Random rnd(301);
const int kTestSize = kCDTKeysPerBuffer * 512;
std::vector<std::string> values;
for (int k = 0; k < kTestSize; ++k) {
values.push_back(RandomString(&rnd, kCDTValueSize));
ASSERT_OK(Put(Key(k), values[k]));
}
dbfull()->TEST_WaitForFlushMemTable();
dbfull()->TEST_WaitForCompact();
// Reopen the DB with stats-update disabled
options.skip_stats_update_on_db_open = true;
env_->random_file_open_counter_.store(0);
Reopen(options);
// As stats-update is disabled, we expect a very low number of
// random file open.
// Note that this number must be changed accordingly if we change
// the number of files needed to be opened in the DB::Open process.
const int kMaxFileOpenCount = 10;
ASSERT_LT(env_->random_file_open_counter_.load(), kMaxFileOpenCount);
// Repeat the reopen process, but this time we enable
// stats-update.
options.skip_stats_update_on_db_open = false;
env_->random_file_open_counter_.store(0);
Reopen(options);
// Since we do a normal stats update on db-open, there
// will be more random open files.
ASSERT_GT(env_->random_file_open_counter_.load(), kMaxFileOpenCount);
}
TEST_F(DBCompactionTest, TestTableReaderForCompaction) {
Options options = CurrentOptions();
options.env = env_;
options.new_table_reader_for_compaction_inputs = true;
options.max_open_files = 100;
options.level0_file_num_compaction_trigger = 3;
DestroyAndReopen(options);
Random rnd(301);
int num_table_cache_lookup = 0;
int num_new_table_reader = 0;
rocksdb::SyncPoint::GetInstance()->SetCallBack(
"TableCache::FindTable:0", [&](void* arg) {
assert(arg != nullptr);
bool no_io = *(reinterpret_cast<bool*>(arg));
if (!no_io) {
// filter out cases for table properties queries.
num_table_cache_lookup++;
}
});
rocksdb::SyncPoint::GetInstance()->SetCallBack(
"TableCache::GetTableReader:0",
[&](void* arg) { num_new_table_reader++; });
rocksdb::SyncPoint::GetInstance()->EnableProcessing();
for (int k = 0; k < options.level0_file_num_compaction_trigger; ++k) {
ASSERT_OK(Put(Key(k), Key(k)));
ASSERT_OK(Put(Key(10 - k), "bar"));
if (k < options.level0_file_num_compaction_trigger - 1) {
num_table_cache_lookup = 0;
Flush();
dbfull()->TEST_WaitForCompact();
// preloading iterator issues one table cache lookup and create
// a new table reader.
ASSERT_EQ(num_table_cache_lookup, 1);
ASSERT_EQ(num_new_table_reader, 1);
num_table_cache_lookup = 0;
num_new_table_reader = 0;
ASSERT_EQ(Key(k), Get(Key(k)));
// lookup iterator from table cache and no need to create a new one.
ASSERT_EQ(num_table_cache_lookup, 1);
ASSERT_EQ(num_new_table_reader, 0);
}
}
num_table_cache_lookup = 0;
num_new_table_reader = 0;
Flush();
dbfull()->TEST_WaitForCompact();
// Preloading iterator issues one table cache lookup and creates
// a new table reader. One file is created for flush and one for compaction.
// Compaction inputs make no table cache look-up for data/range deletion
// iterators
ASSERT_EQ(num_table_cache_lookup, 2);
// Create new iterator for:
// (1) 1 for verifying flush results
// (2) 3 for compaction input files
// (3) 1 for verifying compaction results.
ASSERT_EQ(num_new_table_reader, 5);
num_table_cache_lookup = 0;
num_new_table_reader = 0;
ASSERT_EQ(Key(1), Get(Key(1)));
ASSERT_EQ(num_table_cache_lookup, 1);
ASSERT_EQ(num_new_table_reader, 0);
num_table_cache_lookup = 0;
num_new_table_reader = 0;
CompactRangeOptions cro;
cro.change_level = true;
cro.target_level = 2;
cro.bottommost_level_compaction = BottommostLevelCompaction::kForce;
db_->CompactRange(cro, nullptr, nullptr);
// Only verifying compaction outputs issues one table cache lookup
// for both data block and range deletion block).
ASSERT_EQ(num_table_cache_lookup, 1);
// One for compaction input, one for verifying compaction results.
ASSERT_EQ(num_new_table_reader, 2);
num_table_cache_lookup = 0;
num_new_table_reader = 0;
ASSERT_EQ(Key(1), Get(Key(1)));
ASSERT_EQ(num_table_cache_lookup, 1);
ASSERT_EQ(num_new_table_reader, 0);
rocksdb::SyncPoint::GetInstance()->ClearAllCallBacks();
}
TEST_P(DBCompactionTestWithParam, CompactionDeletionTriggerReopen) {
for (int tid = 0; tid < 2; ++tid) {
uint64_t db_size[3];
Options options = DeletionTriggerOptions(CurrentOptions());
options.max_subcompactions = max_subcompactions_;
if (tid == 1) {
// second pass with universal compaction
options.compaction_style = kCompactionStyleUniversal;
options.num_levels = 1;
}
DestroyAndReopen(options);
Random rnd(301);
// round 1 --- insert key/value pairs.
const int kTestSize = kCDTKeysPerBuffer * 512;
std::vector<std::string> values;
for (int k = 0; k < kTestSize; ++k) {
values.push_back(RandomString(&rnd, kCDTValueSize));
ASSERT_OK(Put(Key(k), values[k]));
}
dbfull()->TEST_WaitForFlushMemTable();
dbfull()->TEST_WaitForCompact();
db_size[0] = Size(Key(0), Key(kTestSize - 1));
Close();
// round 2 --- disable auto-compactions and issue deletions.
options.create_if_missing = false;
options.disable_auto_compactions = true;
Reopen(options);
for (int k = 0; k < kTestSize; ++k) {
ASSERT_OK(Delete(Key(k)));
}
db_size[1] = Size(Key(0), Key(kTestSize - 1));
Close();
// as auto_compaction is off, we shouldn't see too much reduce
// in db size.
ASSERT_LT(db_size[0] / 3, db_size[1]);
// round 3 --- reopen db with auto_compaction on and see if
// deletion compensation still work.
options.disable_auto_compactions = false;
Reopen(options);
// insert relatively small amount of data to trigger auto compaction.
for (int k = 0; k < kTestSize / 10; ++k) {
ASSERT_OK(Put(Key(k), values[k]));
}
dbfull()->TEST_WaitForFlushMemTable();
dbfull()->TEST_WaitForCompact();
db_size[2] = Size(Key(0), Key(kTestSize - 1));
// this time we're expecting significant drop in size.
ASSERT_GT(db_size[0] / 3, db_size[2]);
}
}
TEST_F(DBCompactionTest, DisableStatsUpdateReopen) {
uint64_t db_size[3];
for (int test = 0; test < 2; ++test) {
Options options = DeletionTriggerOptions(CurrentOptions());
options.skip_stats_update_on_db_open = (test == 0);
env_->random_read_counter_.Reset();
DestroyAndReopen(options);
Random rnd(301);
// round 1 --- insert key/value pairs.
const int kTestSize = kCDTKeysPerBuffer * 512;
std::vector<std::string> values;
for (int k = 0; k < kTestSize; ++k) {
values.push_back(RandomString(&rnd, kCDTValueSize));
ASSERT_OK(Put(Key(k), values[k]));
}
dbfull()->TEST_WaitForFlushMemTable();
dbfull()->TEST_WaitForCompact();
db_size[0] = Size(Key(0), Key(kTestSize - 1));
Close();
// round 2 --- disable auto-compactions and issue deletions.
options.create_if_missing = false;
options.disable_auto_compactions = true;
env_->random_read_counter_.Reset();
Reopen(options);
for (int k = 0; k < kTestSize; ++k) {
ASSERT_OK(Delete(Key(k)));
}
db_size[1] = Size(Key(0), Key(kTestSize - 1));
Close();
// as auto_compaction is off, we shouldn't see too much reduce
// in db size.
ASSERT_LT(db_size[0] / 3, db_size[1]);
// round 3 --- reopen db with auto_compaction on and see if
// deletion compensation still work.
options.disable_auto_compactions = false;
Reopen(options);
dbfull()->TEST_WaitForFlushMemTable();
dbfull()->TEST_WaitForCompact();
db_size[2] = Size(Key(0), Key(kTestSize - 1));
if (options.skip_stats_update_on_db_open) {
// If update stats on DB::Open is disable, we don't expect
// deletion entries taking effect.
ASSERT_LT(db_size[0] / 3, db_size[2]);
} else {
// Otherwise, we should see a significant drop in db size.
ASSERT_GT(db_size[0] / 3, db_size[2]);
}
}
}
TEST_P(DBCompactionTestWithParam, CompactionTrigger) {
const int kNumKeysPerFile = 100;
Options options = CurrentOptions();
options.write_buffer_size = 110 << 10; // 110KB
options.arena_block_size = 4 << 10;
options.num_levels = 3;
options.level0_file_num_compaction_trigger = 3;
options.max_subcompactions = max_subcompactions_;
options.memtable_factory.reset(new SpecialSkipListFactory(kNumKeysPerFile));
CreateAndReopenWithCF({"pikachu"}, options);
Random rnd(301);
for (int num = 0; num < options.level0_file_num_compaction_trigger - 1;
num++) {
std::vector<std::string> values;
// Write 100KB (100 values, each 1K)
for (int i = 0; i < kNumKeysPerFile; i++) {
values.push_back(RandomString(&rnd, 990));
ASSERT_OK(Put(1, Key(i), values[i]));
}
// put extra key to trigger flush
ASSERT_OK(Put(1, "", ""));
dbfull()->TEST_WaitForFlushMemTable(handles_[1]);
ASSERT_EQ(NumTableFilesAtLevel(0, 1), num + 1);
}
// generate one more file in level-0, and should trigger level-0 compaction
std::vector<std::string> values;
for (int i = 0; i < kNumKeysPerFile; i++) {
values.push_back(RandomString(&rnd, 990));
ASSERT_OK(Put(1, Key(i), values[i]));
}
// put extra key to trigger flush
ASSERT_OK(Put(1, "", ""));
dbfull()->TEST_WaitForCompact();
ASSERT_EQ(NumTableFilesAtLevel(0, 1), 0);
ASSERT_EQ(NumTableFilesAtLevel(1, 1), 1);
}
TEST_F(DBCompactionTest, BGCompactionsAllowed) {
// Create several column families. Make compaction triggers in all of them
// and see number of compactions scheduled to be less than allowed.
const int kNumKeysPerFile = 100;
Options options = CurrentOptions();
options.write_buffer_size = 110 << 10; // 110KB
options.arena_block_size = 4 << 10;
options.num_levels = 3;
// Should speed up compaction when there are 4 files.
options.level0_file_num_compaction_trigger = 2;
options.level0_slowdown_writes_trigger = 20;
options.soft_pending_compaction_bytes_limit = 1 << 30; // Infinitely large
options.base_background_compactions = 1;
options.max_background_compactions = 3;
options.memtable_factory.reset(new SpecialSkipListFactory(kNumKeysPerFile));
// Block all threads in thread pool.
const size_t kTotalTasks = 4;
env_->SetBackgroundThreads(4, Env::LOW);
test::SleepingBackgroundTask sleeping_tasks[kTotalTasks];
for (size_t i = 0; i < kTotalTasks; i++) {
env_->Schedule(&test::SleepingBackgroundTask::DoSleepTask,
&sleeping_tasks[i], Env::Priority::LOW);
sleeping_tasks[i].WaitUntilSleeping();
}
CreateAndReopenWithCF({"one", "two", "three"}, options);
Random rnd(301);
for (int cf = 0; cf < 4; cf++) {
for (int num = 0; num < options.level0_file_num_compaction_trigger; num++) {
for (int i = 0; i < kNumKeysPerFile; i++) {
ASSERT_OK(Put(cf, Key(i), ""));
}
// put extra key to trigger flush
ASSERT_OK(Put(cf, "", ""));
dbfull()->TEST_WaitForFlushMemTable(handles_[cf]);
ASSERT_EQ(NumTableFilesAtLevel(0, cf), num + 1);
}
}
// Now all column families qualify compaction but only one should be
// scheduled, because no column family hits speed up condition.
ASSERT_EQ(1, env_->GetThreadPoolQueueLen(Env::Priority::LOW));
// Create two more files for one column family, which triggers speed up
// condition, three compactions will be scheduled.
for (int num = 0; num < options.level0_file_num_compaction_trigger; num++) {
for (int i = 0; i < kNumKeysPerFile; i++) {
ASSERT_OK(Put(2, Key(i), ""));
}
// put extra key to trigger flush
ASSERT_OK(Put(2, "", ""));
dbfull()->TEST_WaitForFlushMemTable(handles_[2]);
ASSERT_EQ(options.level0_file_num_compaction_trigger + num + 1,
NumTableFilesAtLevel(0, 2));
}
ASSERT_EQ(3, env_->GetThreadPoolQueueLen(Env::Priority::LOW));
// Unblock all threads to unblock all compactions.
for (size_t i = 0; i < kTotalTasks; i++) {
sleeping_tasks[i].WakeUp();
sleeping_tasks[i].WaitUntilDone();
}
dbfull()->TEST_WaitForCompact();
// Verify number of compactions allowed will come back to 1.
for (size_t i = 0; i < kTotalTasks; i++) {
sleeping_tasks[i].Reset();
env_->Schedule(&test::SleepingBackgroundTask::DoSleepTask,
&sleeping_tasks[i], Env::Priority::LOW);
sleeping_tasks[i].WaitUntilSleeping();
}
for (int cf = 0; cf < 4; cf++) {
for (int num = 0; num < options.level0_file_num_compaction_trigger; num++) {
for (int i = 0; i < kNumKeysPerFile; i++) {
ASSERT_OK(Put(cf, Key(i), ""));
}
// put extra key to trigger flush
ASSERT_OK(Put(cf, "", ""));
dbfull()->TEST_WaitForFlushMemTable(handles_[cf]);
ASSERT_EQ(NumTableFilesAtLevel(0, cf), num + 1);
}
}
// Now all column families qualify compaction but only one should be
// scheduled, because no column family hits speed up condition.
ASSERT_EQ(1, env_->GetThreadPoolQueueLen(Env::Priority::LOW));
for (size_t i = 0; i < kTotalTasks; i++) {
sleeping_tasks[i].WakeUp();
sleeping_tasks[i].WaitUntilDone();
}
}
TEST_P(DBCompactionTestWithParam, CompactionsGenerateMultipleFiles) {
Options options = CurrentOptions();
options.write_buffer_size = 100000000; // Large write buffer
options.max_subcompactions = max_subcompactions_;
CreateAndReopenWithCF({"pikachu"}, options);
Random rnd(301);
// Write 8MB (80 values, each 100K)
ASSERT_EQ(NumTableFilesAtLevel(0, 1), 0);
std::vector<std::string> values;
for (int i = 0; i < 80; i++) {
values.push_back(RandomString(&rnd, 100000));
ASSERT_OK(Put(1, Key(i), values[i]));
}
// Reopening moves updates to level-0
ReopenWithColumnFamilies({"default", "pikachu"}, options);
dbfull()->TEST_CompactRange(0, nullptr, nullptr, handles_[1],
true /* disallow trivial move */);
ASSERT_EQ(NumTableFilesAtLevel(0, 1), 0);
ASSERT_GT(NumTableFilesAtLevel(1, 1), 1);
for (int i = 0; i < 80; i++) {
ASSERT_EQ(Get(1, Key(i)), values[i]);
}
}
TEST_F(DBCompactionTest, MinorCompactionsHappen) {
do {
Options options = CurrentOptions();
options.write_buffer_size = 10000;
CreateAndReopenWithCF({"pikachu"}, options);
const int N = 500;
int starting_num_tables = TotalTableFiles(1);
for (int i = 0; i < N; i++) {
ASSERT_OK(Put(1, Key(i), Key(i) + std::string(1000, 'v')));
}
int ending_num_tables = TotalTableFiles(1);
ASSERT_GT(ending_num_tables, starting_num_tables);
for (int i = 0; i < N; i++) {
ASSERT_EQ(Key(i) + std::string(1000, 'v'), Get(1, Key(i)));
}
ReopenWithColumnFamilies({"default", "pikachu"}, options);
for (int i = 0; i < N; i++) {
ASSERT_EQ(Key(i) + std::string(1000, 'v'), Get(1, Key(i)));
}
} while (ChangeCompactOptions());
}
TEST_F(DBCompactionTest, UserKeyCrossFile1) {
Options options = CurrentOptions();
options.compaction_style = kCompactionStyleLevel;
options.level0_file_num_compaction_trigger = 3;
DestroyAndReopen(options);
// create first file and flush to l0
Put("4", "A");
Put("3", "A");
Flush();
dbfull()->TEST_WaitForFlushMemTable();
Put("2", "A");
Delete("3");
Flush();
dbfull()->TEST_WaitForFlushMemTable();
ASSERT_EQ("NOT_FOUND", Get("3"));
// move both files down to l1
dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr);
ASSERT_EQ("NOT_FOUND", Get("3"));
for (int i = 0; i < 3; i++) {
Put("2", "B");
Flush();
dbfull()->TEST_WaitForFlushMemTable();
}
dbfull()->TEST_WaitForCompact();
ASSERT_EQ("NOT_FOUND", Get("3"));
}
TEST_F(DBCompactionTest, UserKeyCrossFile2) {
Options options = CurrentOptions();
options.compaction_style = kCompactionStyleLevel;
options.level0_file_num_compaction_trigger = 3;
DestroyAndReopen(options);
// create first file and flush to l0
Put("4", "A");
Put("3", "A");
Flush();
dbfull()->TEST_WaitForFlushMemTable();
Put("2", "A");
SingleDelete("3");
Flush();
dbfull()->TEST_WaitForFlushMemTable();
ASSERT_EQ("NOT_FOUND", Get("3"));
// move both files down to l1
dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr);
ASSERT_EQ("NOT_FOUND", Get("3"));
for (int i = 0; i < 3; i++) {
Put("2", "B");
Flush();
dbfull()->TEST_WaitForFlushMemTable();
}
dbfull()->TEST_WaitForCompact();
ASSERT_EQ("NOT_FOUND", Get("3"));
}
TEST_F(DBCompactionTest, ZeroSeqIdCompaction) {
Options options = CurrentOptions();
options.compaction_style = kCompactionStyleLevel;
options.level0_file_num_compaction_trigger = 3;
FlushedFileCollector* collector = new FlushedFileCollector();
options.listeners.emplace_back(collector);
// compaction options
CompactionOptions compact_opt;
compact_opt.compression = kNoCompression;
compact_opt.output_file_size_limit = 4096;
const size_t key_len =
static_cast<size_t>(compact_opt.output_file_size_limit) / 5;
DestroyAndReopen(options);
std::vector<const Snapshot*> snaps;
// create first file and flush to l0
for (auto& key : {"1", "2", "3", "3", "3", "3"}) {
Put(key, std::string(key_len, 'A'));
snaps.push_back(dbfull()->GetSnapshot());
}
Flush();
dbfull()->TEST_WaitForFlushMemTable();
// create second file and flush to l0
for (auto& key : {"3", "4", "5", "6", "7", "8"}) {
Put(key, std::string(key_len, 'A'));
snaps.push_back(dbfull()->GetSnapshot());
}
Flush();
dbfull()->TEST_WaitForFlushMemTable();
// move both files down to l1
dbfull()->CompactFiles(compact_opt, collector->GetFlushedFiles(), 1);
// release snap so that first instance of key(3) can have seqId=0
for (auto snap : snaps) {
dbfull()->ReleaseSnapshot(snap);
}
// create 3 files in l0 so to trigger compaction
for (int i = 0; i < options.level0_file_num_compaction_trigger; i++) {
Put("2", std::string(1, 'A'));
Flush();
dbfull()->TEST_WaitForFlushMemTable();
}
dbfull()->TEST_WaitForCompact();
ASSERT_OK(Put("", ""));
}
// Check that writes done during a memtable compaction are recovered
// if the database is shutdown during the memtable compaction.
TEST_F(DBCompactionTest, RecoverDuringMemtableCompaction) {
do {
Options options = CurrentOptions();
options.env = env_;
CreateAndReopenWithCF({"pikachu"}, options);
// Trigger a long memtable compaction and reopen the database during it
ASSERT_OK(Put(1, "foo", "v1")); // Goes to 1st log file
ASSERT_OK(Put(1, "big1", std::string(10000000, 'x'))); // Fills memtable
ASSERT_OK(Put(1, "big2", std::string(1000, 'y'))); // Triggers compaction
ASSERT_OK(Put(1, "bar", "v2")); // Goes to new log file
ReopenWithColumnFamilies({"default", "pikachu"}, options);
ASSERT_EQ("v1", Get(1, "foo"));
ASSERT_EQ("v2", Get(1, "bar"));
ASSERT_EQ(std::string(10000000, 'x'), Get(1, "big1"));
ASSERT_EQ(std::string(1000, 'y'), Get(1, "big2"));
} while (ChangeOptions());
}
TEST_P(DBCompactionTestWithParam, TrivialMoveOneFile) {
int32_t trivial_move = 0;
rocksdb::SyncPoint::GetInstance()->SetCallBack(
"DBImpl::BackgroundCompaction:TrivialMove",
[&](void* arg) { trivial_move++; });
rocksdb::SyncPoint::GetInstance()->EnableProcessing();
Options options = CurrentOptions();
options.write_buffer_size = 100000000;
options.max_subcompactions = max_subcompactions_;
DestroyAndReopen(options);
int32_t num_keys = 80;
int32_t value_size = 100 * 1024; // 100 KB
Random rnd(301);
std::vector<std::string> values;
for (int i = 0; i < num_keys; i++) {
values.push_back(RandomString(&rnd, value_size));
ASSERT_OK(Put(Key(i), values[i]));
}
// Reopening moves updates to L0
Reopen(options);
ASSERT_EQ(NumTableFilesAtLevel(0, 0), 1); // 1 file in L0
ASSERT_EQ(NumTableFilesAtLevel(1, 0), 0); // 0 files in L1
std::vector<LiveFileMetaData> metadata;
db_->GetLiveFilesMetaData(&metadata);
ASSERT_EQ(metadata.size(), 1U);
LiveFileMetaData level0_file = metadata[0]; // L0 file meta
CompactRangeOptions cro;
cro.exclusive_manual_compaction = exclusive_manual_compaction_;
// Compaction will initiate a trivial move from L0 to L1
dbfull()->CompactRange(cro, nullptr, nullptr);
// File moved From L0 to L1
ASSERT_EQ(NumTableFilesAtLevel(0, 0), 0); // 0 files in L0
ASSERT_EQ(NumTableFilesAtLevel(1, 0), 1); // 1 file in L1
metadata.clear();
db_->GetLiveFilesMetaData(&metadata);
ASSERT_EQ(metadata.size(), 1U);
ASSERT_EQ(metadata[0].name /* level1_file.name */, level0_file.name);
ASSERT_EQ(metadata[0].size /* level1_file.size */, level0_file.size);
for (int i = 0; i < num_keys; i++) {
ASSERT_EQ(Get(Key(i)), values[i]);
}
ASSERT_EQ(trivial_move, 1);
rocksdb::SyncPoint::GetInstance()->DisableProcessing();
}
TEST_P(DBCompactionTestWithParam, TrivialMoveNonOverlappingFiles) {
int32_t trivial_move = 0;
int32_t non_trivial_move = 0;
rocksdb::SyncPoint::GetInstance()->SetCallBack(
"DBImpl::BackgroundCompaction:TrivialMove",
[&](void* arg) { trivial_move++; });
rocksdb::SyncPoint::GetInstance()->SetCallBack(
"DBImpl::BackgroundCompaction:NonTrivial",
[&](void* arg) { non_trivial_move++; });
rocksdb::SyncPoint::GetInstance()->EnableProcessing();
Options options = CurrentOptions();
options.disable_auto_compactions = true;
options.write_buffer_size = 10 * 1024 * 1024;
options.max_subcompactions = max_subcompactions_;
DestroyAndReopen(options);
// non overlapping ranges
std::vector<std::pair<int32_t, int32_t>> ranges = {
{100, 199},
{300, 399},
{0, 99},
{200, 299},
{600, 699},
{400, 499},
{500, 550},
{551, 599},
};
int32_t value_size = 10 * 1024; // 10 KB
Random rnd(301);
std::map<int32_t, std::string> values;
for (size_t i = 0; i < ranges.size(); i++) {
for (int32_t j = ranges[i].first; j <= ranges[i].second; j++) {
values[j] = RandomString(&rnd, value_size);
ASSERT_OK(Put(Key(j), values[j]));
}
ASSERT_OK(Flush());
}
int32_t level0_files = NumTableFilesAtLevel(0, 0);
ASSERT_EQ(level0_files, ranges.size()); // Multiple files in L0
ASSERT_EQ(NumTableFilesAtLevel(1, 0), 0); // No files in L1
CompactRangeOptions cro;
cro.exclusive_manual_compaction = exclusive_manual_compaction_;
// Since data is non-overlapping we expect compaction to initiate
// a trivial move
db_->CompactRange(cro, nullptr, nullptr);
// We expect that all the files were trivially moved from L0 to L1
ASSERT_EQ(NumTableFilesAtLevel(0, 0), 0);
ASSERT_EQ(NumTableFilesAtLevel(1, 0) /* level1_files */, level0_files);
for (size_t i = 0; i < ranges.size(); i++) {
for (int32_t j = ranges[i].first; j <= ranges[i].second; j++) {
ASSERT_EQ(Get(Key(j)), values[j]);
}
}
ASSERT_EQ(trivial_move, 1);
ASSERT_EQ(non_trivial_move, 0);
trivial_move = 0;
non_trivial_move = 0;
values.clear();
DestroyAndReopen(options);
// Same ranges as above but overlapping
ranges = {
{100, 199},
{300, 399},
{0, 99},
{200, 299},
{600, 699},
{400, 499},
{500, 560}, // this range overlap with the next one
{551, 599},
};
for (size_t i = 0; i < ranges.size(); i++) {
for (int32_t j = ranges[i].first; j <= ranges[i].second; j++) {
values[j] = RandomString(&rnd, value_size);
ASSERT_OK(Put(Key(j), values[j]));
}
ASSERT_OK(Flush());
}
db_->CompactRange(cro, nullptr, nullptr);
for (size_t i = 0; i < ranges.size(); i++) {
for (int32_t j = ranges[i].first; j <= ranges[i].second; j++) {
ASSERT_EQ(Get(Key(j)), values[j]);
}
}
ASSERT_EQ(trivial_move, 0);
ASSERT_EQ(non_trivial_move, 1);
rocksdb::SyncPoint::GetInstance()->DisableProcessing();
}
TEST_P(DBCompactionTestWithParam, TrivialMoveTargetLevel) {
int32_t trivial_move = 0;
int32_t non_trivial_move = 0;
rocksdb::SyncPoint::GetInstance()->SetCallBack(
"DBImpl::BackgroundCompaction:TrivialMove",
[&](void* arg) { trivial_move++; });
rocksdb::SyncPoint::GetInstance()->SetCallBack(
"DBImpl::BackgroundCompaction:NonTrivial",
[&](void* arg) { non_trivial_move++; });
rocksdb::SyncPoint::GetInstance()->EnableProcessing();
Options options = CurrentOptions();
options.disable_auto_compactions = true;
options.write_buffer_size = 10 * 1024 * 1024;
options.num_levels = 7;
options.max_subcompactions = max_subcompactions_;
DestroyAndReopen(options);
int32_t value_size = 10 * 1024; // 10 KB
// Add 2 non-overlapping files
Random rnd(301);
std::map<int32_t, std::string> values;
// file 1 [0 => 300]
for (int32_t i = 0; i <= 300; i++) {
values[i] = RandomString(&rnd, value_size);
ASSERT_OK(Put(Key(i), values[i]));
}
ASSERT_OK(Flush());
// file 2 [600 => 700]
for (int32_t i = 600; i <= 700; i++) {
values[i] = RandomString(&rnd, value_size);
ASSERT_OK(Put(Key(i), values[i]));
}
ASSERT_OK(Flush());
// 2 files in L0
ASSERT_EQ("2", FilesPerLevel(0));
CompactRangeOptions compact_options;
compact_options.change_level = true;
compact_options.target_level = 6;
compact_options.exclusive_manual_compaction = exclusive_manual_compaction_;
ASSERT_OK(db_->CompactRange(compact_options, nullptr, nullptr));
// 2 files in L6
ASSERT_EQ("0,0,0,0,0,0,2", FilesPerLevel(0));
ASSERT_EQ(trivial_move, 1);
ASSERT_EQ(non_trivial_move, 0);