forked from facebook/rocksdb
-
Notifications
You must be signed in to change notification settings - Fork 0
/
db_sst_test.cc
1891 lines (1597 loc) · 65.3 KB
/
db_sst_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 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.
#include "db/db_test_util.h"
#include "env/mock_env.h"
#include "file/sst_file_manager_impl.h"
#include "port/port.h"
#include "port/stack_trace.h"
#include "rocksdb/cache.h"
#include "rocksdb/sst_file_manager.h"
#include "rocksdb/table.h"
#include "util/random.h"
namespace ROCKSDB_NAMESPACE {
class DBSSTTest : public DBTestBase {
public:
DBSSTTest() : DBTestBase("db_sst_test", /*env_do_fsync=*/true) {}
};
// A class which remembers the name of each flushed file.
class FlushedFileCollector : public EventListener {
public:
FlushedFileCollector() = default;
~FlushedFileCollector() override = default;
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 (const auto& fname : flushed_files_) {
result.push_back(fname);
}
return result;
}
void ClearFlushedFiles() {
std::lock_guard<std::mutex> lock(mutex_);
flushed_files_.clear();
}
private:
std::vector<std::string> flushed_files_;
std::mutex mutex_;
};
TEST_F(DBSSTTest, DontDeletePendingOutputs) {
Options options;
options.env = env_;
options.create_if_missing = true;
DestroyAndReopen(options);
// Every time we write to a table file, call FOF/POF with full DB scan. This
// will make sure our pending_outputs_ protection work correctly
std::function<void()> purge_obsolete_files_function = [&]() {
JobContext job_context(0);
dbfull()->TEST_LockMutex();
dbfull()->FindObsoleteFiles(&job_context, true /*force*/);
dbfull()->TEST_UnlockMutex();
dbfull()->PurgeObsoleteFiles(job_context);
job_context.Clean();
};
env_->table_write_callback_ = &purge_obsolete_files_function;
for (int i = 0; i < 2; ++i) {
ASSERT_OK(Put("a", "begin"));
ASSERT_OK(Put("z", "end"));
ASSERT_OK(Flush());
}
// If pending output guard does not work correctly, PurgeObsoleteFiles() will
// delete the file that Compaction is trying to create, causing this: error
// db/db_test.cc:975: IO error:
// /tmp/rocksdbtest-1552237650/db_test/000009.sst: No such file or directory
Compact("a", "b");
}
// 1 Create some SST files by inserting K-V pairs into DB
// 2 Close DB and change suffix from ".sst" to ".ldb" for every other SST file
// 3 Open DB and check if all key can be read
TEST_F(DBSSTTest, SSTsWithLdbSuffixHandling) {
Options options = CurrentOptions();
options.write_buffer_size = 110 << 10; // 110KB
options.num_levels = 4;
DestroyAndReopen(options);
Random rnd(301);
int key_id = 0;
for (int i = 0; i < 10; ++i) {
GenerateNewFile(&rnd, &key_id, false);
}
ASSERT_OK(Flush());
Close();
int const num_files = GetSstFileCount(dbname_);
ASSERT_GT(num_files, 0);
Reopen(options);
std::vector<std::string> values;
values.reserve(key_id);
for (int k = 0; k < key_id; ++k) {
values.push_back(Get(Key(k)));
}
Close();
std::vector<std::string> filenames;
GetSstFiles(env_, dbname_, &filenames);
int num_ldb_files = 0;
for (size_t i = 0; i < filenames.size(); ++i) {
if (i & 1) {
continue;
}
std::string const rdb_name = dbname_ + "/" + filenames[i];
std::string const ldb_name = Rocks2LevelTableFileName(rdb_name);
ASSERT_TRUE(env_->RenameFile(rdb_name, ldb_name).ok());
++num_ldb_files;
}
ASSERT_GT(num_ldb_files, 0);
ASSERT_EQ(num_files, GetSstFileCount(dbname_));
Reopen(options);
for (int k = 0; k < key_id; ++k) {
ASSERT_EQ(values[k], Get(Key(k)));
}
Destroy(options);
}
// Check that we don't crash when opening DB with
// DBOptions::skip_checking_sst_file_sizes_on_db_open = true.
TEST_F(DBSSTTest, SkipCheckingSSTFileSizesOnDBOpen) {
ASSERT_OK(Put("pika", "choo"));
ASSERT_OK(Flush());
// Just open the DB with the option set to true and check that we don't crash.
Options options;
options.env = env_;
options.skip_checking_sst_file_sizes_on_db_open = true;
Reopen(options);
ASSERT_EQ("choo", Get("pika"));
}
TEST_F(DBSSTTest, DontDeleteMovedFile) {
// This test triggers move compaction and verifies that the file is not
// deleted when it's part of move compaction
Options options = CurrentOptions();
options.env = env_;
options.create_if_missing = true;
options.max_bytes_for_level_base = 1024 * 1024; // 1 MB
options.level0_file_num_compaction_trigger =
2; // trigger compaction when we have 2 files
DestroyAndReopen(options);
Random rnd(301);
// Create two 1MB sst files
for (int i = 0; i < 2; ++i) {
// Create 1MB sst file
for (int j = 0; j < 100; ++j) {
ASSERT_OK(Put(Key(i * 50 + j), rnd.RandomString(10 * 1024)));
}
ASSERT_OK(Flush());
}
// this should execute both L0->L1 and L1->(move)->L2 compactions
ASSERT_OK(dbfull()->TEST_WaitForCompact());
ASSERT_EQ("0,0,1", FilesPerLevel(0));
// If the moved file is actually deleted (the move-safeguard in
// ~Version::Version() is not there), we get this failure:
// Corruption: Can't access /000009.sst
Reopen(options);
}
// This reproduces a bug where we don't delete a file because when it was
// supposed to be deleted, it was blocked by pending_outputs
// Consider:
// 1. current file_number is 13
// 2. compaction (1) starts, blocks deletion of all files starting with 13
// (pending outputs)
// 3. file 13 is created by compaction (2)
// 4. file 13 is consumed by compaction (3) and file 15 was created. Since file
// 13 has no references, it is put into VersionSet::obsolete_files_
// 5. FindObsoleteFiles() gets file 13 from VersionSet::obsolete_files_. File 13
// is deleted from obsolete_files_ set.
// 6. PurgeObsoleteFiles() tries to delete file 13, but this file is blocked by
// pending outputs since compaction (1) is still running. It is not deleted and
// it is not present in obsolete_files_ anymore. Therefore, we never delete it.
TEST_F(DBSSTTest, DeleteObsoleteFilesPendingOutputs) {
Options options = CurrentOptions();
options.env = env_;
options.write_buffer_size = 2 * 1024 * 1024; // 2 MB
options.max_bytes_for_level_base = 1024 * 1024; // 1 MB
options.level0_file_num_compaction_trigger =
2; // trigger compaction when we have 2 files
options.max_background_flushes = 2;
options.max_background_compactions = 2;
OnFileDeletionListener* listener = new OnFileDeletionListener();
options.listeners.emplace_back(listener);
Reopen(options);
Random rnd(301);
// Create two 1MB sst files
for (int i = 0; i < 2; ++i) {
// Create 1MB sst file
for (int j = 0; j < 100; ++j) {
ASSERT_OK(Put(Key(i * 50 + j), rnd.RandomString(10 * 1024)));
}
ASSERT_OK(Flush());
}
// this should execute both L0->L1 and L1->(move)->L2 compactions
ASSERT_OK(dbfull()->TEST_WaitForCompact());
ASSERT_EQ("0,0,1", FilesPerLevel(0));
test::SleepingBackgroundTask blocking_thread;
port::Mutex mutex_;
bool already_blocked(false);
// block the flush
std::function<void()> block_first_time = [&]() {
bool blocking = false;
{
MutexLock l(&mutex_);
if (!already_blocked) {
blocking = true;
already_blocked = true;
}
}
if (blocking) {
blocking_thread.DoSleep();
}
};
env_->table_write_callback_ = &block_first_time;
// Insert 2.5MB data, which should trigger a flush because we exceed
// write_buffer_size. The flush will be blocked with block_first_time
// pending_file is protecting all the files created after
for (int j = 0; j < 256; ++j) {
ASSERT_OK(Put(Key(j), rnd.RandomString(10 * 1024)));
}
blocking_thread.WaitUntilSleeping();
ASSERT_OK(dbfull()->TEST_CompactRange(2, nullptr, nullptr));
ASSERT_EQ("0,0,0,1", FilesPerLevel(0));
std::vector<LiveFileMetaData> metadata;
db_->GetLiveFilesMetaData(&metadata);
ASSERT_EQ(metadata.size(), 1U);
auto file_on_L2 = metadata[0].name;
listener->SetExpectedFileName(dbname_ + file_on_L2);
ASSERT_OK(dbfull()->TEST_CompactRange(3, nullptr, nullptr, nullptr,
true /* disallow trivial move */));
ASSERT_EQ("0,0,0,0,1", FilesPerLevel(0));
// finish the flush!
blocking_thread.WakeUp();
blocking_thread.WaitUntilDone();
ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable());
// File just flushed is too big for L0 and L1 so gets moved to L2.
ASSERT_OK(dbfull()->TEST_WaitForCompact());
ASSERT_EQ("0,0,1,0,1", FilesPerLevel(0));
metadata.clear();
db_->GetLiveFilesMetaData(&metadata);
ASSERT_EQ(metadata.size(), 2U);
// This file should have been deleted during last compaction
ASSERT_EQ(Status::NotFound(), env_->FileExists(dbname_ + file_on_L2));
listener->VerifyMatchedCount(1);
}
// Test that producing an empty .sst file does not write it out to
// disk, and that the DeleteFile() env method is not called for
// removing the non-existing file later.
TEST_F(DBSSTTest, DeleteFileNotCalledForNotCreatedSSTFile) {
Options options = CurrentOptions();
options.env = env_;
OnFileDeletionListener* listener = new OnFileDeletionListener();
options.listeners.emplace_back(listener);
Reopen(options);
// Flush the empty database.
ASSERT_OK(Flush());
ASSERT_EQ("", FilesPerLevel(0));
// We expect no .sst files.
std::vector<LiveFileMetaData> metadata;
db_->GetLiveFilesMetaData(&metadata);
ASSERT_EQ(metadata.size(), 0U);
// We expect no file deletions.
listener->VerifyMatchedCount(0);
}
// Test that producing a non-empty .sst file does write it out to
// disk, and that the DeleteFile() env method is not called for removing
// the file later.
TEST_F(DBSSTTest, DeleteFileNotCalledForCreatedSSTFile) {
Options options = CurrentOptions();
options.env = env_;
OnFileDeletionListener* listener = new OnFileDeletionListener();
options.listeners.emplace_back(listener);
Reopen(options);
ASSERT_OK(Put("pika", "choo"));
// Flush the non-empty database.
ASSERT_OK(Flush());
ASSERT_EQ("1", FilesPerLevel(0));
// We expect 1 .sst files.
std::vector<LiveFileMetaData> metadata;
db_->GetLiveFilesMetaData(&metadata);
ASSERT_EQ(metadata.size(), 1U);
// We expect no file deletions.
listener->VerifyMatchedCount(0);
}
TEST_F(DBSSTTest, DBWithSstFileManager) {
std::shared_ptr<SstFileManager> sst_file_manager(NewSstFileManager(env_));
auto sfm = static_cast<SstFileManagerImpl*>(sst_file_manager.get());
int files_added = 0;
int files_deleted = 0;
int files_moved = 0;
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
"SstFileManagerImpl::OnAddFile", [&](void* /*arg*/) { files_added++; });
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
"SstFileManagerImpl::OnDeleteFile",
[&](void* /*arg*/) { files_deleted++; });
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
"SstFileManagerImpl::OnMoveFile", [&](void* /*arg*/) { files_moved++; });
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
Options options = CurrentOptions();
options.sst_file_manager = sst_file_manager;
DestroyAndReopen(options);
Random rnd(301);
for (int i = 0; i < 25; i++) {
GenerateNewRandomFile(&rnd);
ASSERT_OK(Flush());
ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable());
ASSERT_OK(dbfull()->TEST_WaitForCompact());
// Verify that we are tracking all sst files in dbname_
std::unordered_map<std::string, uint64_t> files_in_db;
ASSERT_OK(GetAllDataFiles(kTableFile, &files_in_db));
ASSERT_EQ(sfm->GetTrackedFiles(), files_in_db);
}
ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
ASSERT_OK(dbfull()->TEST_WaitForCompact());
std::unordered_map<std::string, uint64_t> files_in_db;
ASSERT_OK(GetAllDataFiles(kTableFile, &files_in_db));
// Verify that we are tracking all sst files in dbname_
ASSERT_EQ(sfm->GetTrackedFiles(), files_in_db);
// Verify the total files size
uint64_t total_files_size = 0;
for (auto& file_to_size : files_in_db) {
total_files_size += file_to_size.second;
}
ASSERT_EQ(sfm->GetTotalSize(), total_files_size);
// We flushed at least 25 files
ASSERT_GE(files_added, 25);
// Compaction must have deleted some files
ASSERT_GT(files_deleted, 0);
// No files were moved
ASSERT_EQ(files_moved, 0);
Close();
Reopen(options);
ASSERT_EQ(sfm->GetTrackedFiles(), files_in_db);
ASSERT_EQ(sfm->GetTotalSize(), total_files_size);
// Verify that we track all the files again after the DB is closed and opened
Close();
sst_file_manager.reset(NewSstFileManager(env_));
options.sst_file_manager = sst_file_manager;
sfm = static_cast<SstFileManagerImpl*>(sst_file_manager.get());
Reopen(options);
ASSERT_EQ(sfm->GetTrackedFiles(), files_in_db);
ASSERT_EQ(sfm->GetTotalSize(), total_files_size);
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
}
TEST_F(DBSSTTest, DBWithSstFileManagerForBlobFiles) {
std::shared_ptr<SstFileManager> sst_file_manager(NewSstFileManager(env_));
auto sfm = static_cast<SstFileManagerImpl*>(sst_file_manager.get());
int files_added = 0;
int files_deleted = 0;
int files_moved = 0;
int files_scheduled_to_delete = 0;
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
"SstFileManagerImpl::OnAddFile", [&](void* arg) {
const std::string* const file_path =
static_cast<const std::string*>(arg);
if (file_path->find(".blob") != std::string::npos) {
files_added++;
}
});
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
"SstFileManagerImpl::OnDeleteFile", [&](void* arg) {
const std::string* const file_path =
static_cast<const std::string*>(arg);
if (file_path->find(".blob") != std::string::npos) {
files_deleted++;
}
});
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
"SstFileManagerImpl::ScheduleFileDeletion", [&](void* arg) {
assert(arg);
const std::string* const file_path =
static_cast<const std::string*>(arg);
if (file_path->find(".blob") != std::string::npos) {
++files_scheduled_to_delete;
}
});
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
"SstFileManagerImpl::OnMoveFile", [&](void* /*arg*/) { files_moved++; });
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
Options options = CurrentOptions();
options.sst_file_manager = sst_file_manager;
options.enable_blob_files = true;
options.blob_file_size = 32; // create one blob per file
DestroyAndReopen(options);
Random rnd(301);
for (int i = 0; i < 10; i++) {
ASSERT_OK(Put("Key_" + std::to_string(i), "Value_" + std::to_string(i)));
ASSERT_OK(Flush());
ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable());
ASSERT_OK(dbfull()->TEST_WaitForCompact());
// Verify that we are tracking all sst and blob files in dbname_
std::unordered_map<std::string, uint64_t> files_in_db;
ASSERT_OK(GetAllDataFiles(kTableFile, &files_in_db));
ASSERT_OK(GetAllDataFiles(kBlobFile, &files_in_db));
ASSERT_EQ(sfm->GetTrackedFiles(), files_in_db);
}
std::vector<uint64_t> blob_files = GetBlobFileNumbers();
ASSERT_EQ(files_added, blob_files.size());
// No blob file is obsoleted.
ASSERT_EQ(files_deleted, 0);
ASSERT_EQ(files_scheduled_to_delete, 0);
// No files were moved.
ASSERT_EQ(files_moved, 0);
ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
ASSERT_OK(dbfull()->TEST_WaitForCompact());
std::unordered_map<std::string, uint64_t> files_in_db;
ASSERT_OK(GetAllDataFiles(kTableFile, &files_in_db));
ASSERT_OK(GetAllDataFiles(kBlobFile, &files_in_db));
// Verify that we are tracking all sst and blob files in dbname_
ASSERT_EQ(sfm->GetTrackedFiles(), files_in_db);
// Verify the total files size
uint64_t total_files_size = 0;
for (auto& file_to_size : files_in_db) {
total_files_size += file_to_size.second;
}
ASSERT_EQ(sfm->GetTotalSize(), total_files_size);
Close();
Reopen(options);
ASSERT_EQ(sfm->GetTrackedFiles(), files_in_db);
ASSERT_EQ(sfm->GetTotalSize(), total_files_size);
// Verify that we track all the files again after the DB is closed and opened.
Close();
sst_file_manager.reset(NewSstFileManager(env_));
options.sst_file_manager = sst_file_manager;
sfm = static_cast<SstFileManagerImpl*>(sst_file_manager.get());
Reopen(options);
ASSERT_EQ(sfm->GetTrackedFiles(), files_in_db);
ASSERT_EQ(sfm->GetTotalSize(), total_files_size);
// Destroy DB and it will remove all the blob files from sst file manager and
// blob files deletion will go through ScheduleFileDeletion.
ASSERT_EQ(files_deleted, 0);
ASSERT_EQ(files_scheduled_to_delete, 0);
Close();
ASSERT_OK(DestroyDB(dbname_, options));
ASSERT_EQ(files_deleted, blob_files.size());
ASSERT_EQ(files_scheduled_to_delete, blob_files.size());
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->ClearAllCallBacks();
}
TEST_F(DBSSTTest, DBWithSstFileManagerForBlobFilesWithGC) {
std::shared_ptr<SstFileManager> sst_file_manager(NewSstFileManager(env_));
auto sfm = static_cast<SstFileManagerImpl*>(sst_file_manager.get());
Options options = CurrentOptions();
options.sst_file_manager = sst_file_manager;
options.enable_blob_files = true;
options.blob_file_size = 32; // create one blob per file
options.disable_auto_compactions = true;
options.enable_blob_garbage_collection = true;
options.blob_garbage_collection_age_cutoff = 0.5;
int files_added = 0;
int files_deleted = 0;
int files_moved = 0;
int files_scheduled_to_delete = 0;
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
"SstFileManagerImpl::OnAddFile", [&](void* arg) {
const std::string* const file_path =
static_cast<const std::string*>(arg);
if (file_path->find(".blob") != std::string::npos) {
files_added++;
}
});
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
"SstFileManagerImpl::OnDeleteFile", [&](void* arg) {
const std::string* const file_path =
static_cast<const std::string*>(arg);
if (file_path->find(".blob") != std::string::npos) {
files_deleted++;
}
});
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
"SstFileManagerImpl::ScheduleFileDeletion", [&](void* arg) {
assert(arg);
const std::string* const file_path =
static_cast<const std::string*>(arg);
if (file_path->find(".blob") != std::string::npos) {
++files_scheduled_to_delete;
}
});
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
"SstFileManagerImpl::OnMoveFile", [&](void* /*arg*/) { files_moved++; });
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
DestroyAndReopen(options);
Random rnd(301);
constexpr char first_key[] = "first_key";
constexpr char first_value[] = "first_value";
constexpr char second_key[] = "second_key";
constexpr char second_value[] = "second_value";
ASSERT_OK(Put(first_key, first_value));
ASSERT_OK(Put(second_key, second_value));
ASSERT_OK(Flush());
constexpr char third_key[] = "third_key";
constexpr char third_value[] = "third_value";
constexpr char fourth_key[] = "fourth_key";
constexpr char fourth_value[] = "fourth_value";
constexpr char fifth_key[] = "fifth_key";
constexpr char fifth_value[] = "fifth_value";
ASSERT_OK(Put(third_key, third_value));
ASSERT_OK(Put(fourth_key, fourth_value));
ASSERT_OK(Put(fifth_key, fifth_value));
ASSERT_OK(Flush());
const std::vector<uint64_t> original_blob_files = GetBlobFileNumbers();
ASSERT_EQ(original_blob_files.size(), 5);
ASSERT_EQ(files_added, 5);
ASSERT_EQ(files_deleted, 0);
ASSERT_EQ(files_scheduled_to_delete, 0);
ASSERT_EQ(files_moved, 0);
{
// Verify that we are tracking all sst and blob files in dbname_
std::unordered_map<std::string, uint64_t> files_in_db;
ASSERT_OK(GetAllDataFiles(kTableFile, &files_in_db));
ASSERT_OK(GetAllDataFiles(kBlobFile, &files_in_db));
ASSERT_EQ(sfm->GetTrackedFiles(), files_in_db);
}
const size_t cutoff_index = static_cast<size_t>(
options.blob_garbage_collection_age_cutoff * original_blob_files.size());
size_t expected_number_of_files = original_blob_files.size();
// Note: turning off enable_blob_files before the compaction results in
// garbage collected values getting inlined.
ASSERT_OK(db_->SetOptions({{"enable_blob_files", "false"}}));
expected_number_of_files -= cutoff_index;
files_added = 0;
constexpr Slice* begin = nullptr;
constexpr Slice* end = nullptr;
ASSERT_OK(db_->CompactRange(CompactRangeOptions(), begin, end));
ASSERT_OK(dbfull()->TEST_WaitForCompact());
sfm->WaitForEmptyTrash();
ASSERT_EQ(Get(first_key), first_value);
ASSERT_EQ(Get(second_key), second_value);
ASSERT_EQ(Get(third_key), third_value);
ASSERT_EQ(Get(fourth_key), fourth_value);
ASSERT_EQ(Get(fifth_key), fifth_value);
const std::vector<uint64_t> new_blob_files = GetBlobFileNumbers();
ASSERT_EQ(new_blob_files.size(), expected_number_of_files);
// No new file is added.
ASSERT_EQ(files_added, 0);
ASSERT_EQ(files_deleted, cutoff_index);
ASSERT_EQ(files_scheduled_to_delete, cutoff_index);
ASSERT_EQ(files_moved, 0);
// Original blob files below the cutoff should be gone, original blob files at
// or above the cutoff should be still there
for (size_t i = cutoff_index; i < original_blob_files.size(); ++i) {
ASSERT_EQ(new_blob_files[i - cutoff_index], original_blob_files[i]);
}
{
// Verify that we are tracking all sst and blob files in dbname_
std::unordered_map<std::string, uint64_t> files_in_db;
ASSERT_OK(GetAllDataFiles(kTableFile, &files_in_db));
ASSERT_OK(GetAllDataFiles(kBlobFile, &files_in_db));
ASSERT_EQ(sfm->GetTrackedFiles(), files_in_db);
}
Close();
ASSERT_OK(DestroyDB(dbname_, options));
sfm->WaitForEmptyTrash();
ASSERT_EQ(files_deleted, 5);
ASSERT_EQ(files_scheduled_to_delete, 5);
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->ClearAllCallBacks();
}
class DBSSTTestRateLimit : public DBSSTTest,
public ::testing::WithParamInterface<bool> {
public:
DBSSTTestRateLimit() : DBSSTTest() {}
~DBSSTTestRateLimit() override = default;
};
TEST_P(DBSSTTestRateLimit, RateLimitedDelete) {
Destroy(last_options_);
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->LoadDependency({
{"DBSSTTest::RateLimitedDelete:1",
"DeleteScheduler::BackgroundEmptyTrash"},
});
std::vector<uint64_t> penalties;
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
"DeleteScheduler::BackgroundEmptyTrash:Wait",
[&](void* arg) { penalties.push_back(*(static_cast<uint64_t*>(arg))); });
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
"InstrumentedCondVar::TimedWaitInternal", [&](void* arg) {
// Turn timed wait into a simulated sleep
uint64_t* abs_time_us = static_cast<uint64_t*>(arg);
uint64_t cur_time = env_->NowMicros();
if (*abs_time_us > cur_time) {
env_->MockSleepForMicroseconds(*abs_time_us - cur_time);
}
// Plus an additional short, random amount
env_->MockSleepForMicroseconds(Random::GetTLSInstance()->Uniform(10));
// Set wait until time to before (actual) current time to force not
// to sleep
*abs_time_us = Env::Default()->NowMicros();
});
// Disable PeriodicTaskScheduler as it also has TimedWait, which could update
// the simulated sleep time
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
"DBImpl::StartPeriodicTaskScheduler:DisableScheduler", [&](void* arg) {
bool* disable_scheduler = static_cast<bool*>(arg);
*disable_scheduler = true;
});
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
bool different_wal_dir = GetParam();
Options options = CurrentOptions();
SetTimeElapseOnlySleepOnReopen(&options);
options.disable_auto_compactions = true;
options.env = env_;
options.statistics = CreateDBStatistics();
if (different_wal_dir) {
options.wal_dir = alternative_wal_dir_;
}
int64_t rate_bytes_per_sec = 1024 * 10; // 10 Kbs / Sec
Status s;
options.sst_file_manager.reset(
NewSstFileManager(env_, nullptr, "", 0, false, &s, 0));
ASSERT_OK(s);
options.sst_file_manager->SetDeleteRateBytesPerSecond(rate_bytes_per_sec);
auto sfm = static_cast<SstFileManagerImpl*>(options.sst_file_manager.get());
sfm->delete_scheduler()->SetMaxTrashDBRatio(1.1);
WriteOptions wo;
if (!different_wal_dir) {
wo.disableWAL = true;
}
Reopen(options);
// Create 4 files in L0
for (char v = 'a'; v <= 'd'; v++) {
ASSERT_OK(Put("Key2", DummyString(1024, v), wo));
ASSERT_OK(Put("Key3", DummyString(1024, v), wo));
ASSERT_OK(Put("Key4", DummyString(1024, v), wo));
ASSERT_OK(Put("Key1", DummyString(1024, v), wo));
ASSERT_OK(Put("Key4", DummyString(1024, v), wo));
ASSERT_OK(Flush());
}
// We created 4 sst files in L0
ASSERT_EQ("4", FilesPerLevel(0));
std::vector<LiveFileMetaData> metadata;
db_->GetLiveFilesMetaData(&metadata);
// Compaction will move the 4 files in L0 to trash and create 1 L1 file
ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
ASSERT_OK(dbfull()->TEST_WaitForCompact());
ASSERT_EQ("0,1", FilesPerLevel(0));
uint64_t delete_start_time = env_->NowMicros();
// Hold BackgroundEmptyTrash
TEST_SYNC_POINT("DBSSTTest::RateLimitedDelete:1");
sfm->WaitForEmptyTrash();
uint64_t time_spent_deleting = env_->NowMicros() - delete_start_time;
uint64_t total_files_size = 0;
uint64_t expected_penlty = 0;
ASSERT_EQ(penalties.size(), metadata.size());
for (size_t i = 0; i < metadata.size(); i++) {
total_files_size += metadata[i].size;
expected_penlty = ((total_files_size * 1000000) / rate_bytes_per_sec);
ASSERT_EQ(expected_penlty, penalties[i]);
}
ASSERT_GT(time_spent_deleting, expected_penlty * 0.9);
ASSERT_LT(time_spent_deleting, expected_penlty * 1.1);
ASSERT_EQ(4, options.statistics->getAndResetTickerCount(FILES_MARKED_TRASH));
ASSERT_EQ(
0, options.statistics->getAndResetTickerCount(FILES_DELETED_IMMEDIATELY));
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
}
INSTANTIATE_TEST_CASE_P(RateLimitedDelete, DBSSTTestRateLimit,
::testing::Bool());
TEST_F(DBSSTTest, RateLimitedWALDelete) {
Destroy(last_options_);
std::vector<uint64_t> penalties;
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
"DeleteScheduler::BackgroundEmptyTrash:Wait",
[&](void* arg) { penalties.push_back(*(static_cast<uint64_t*>(arg))); });
Options options = CurrentOptions();
options.disable_auto_compactions = true;
options.compression = kNoCompression;
options.env = env_;
int64_t rate_bytes_per_sec = 1024 * 10; // 10 Kbs / Sec
Status s;
options.sst_file_manager.reset(
NewSstFileManager(env_, nullptr, "", 0, false, &s, 0));
ASSERT_OK(s);
options.sst_file_manager->SetDeleteRateBytesPerSecond(rate_bytes_per_sec);
auto sfm = static_cast<SstFileManagerImpl*>(options.sst_file_manager.get());
sfm->delete_scheduler()->SetMaxTrashDBRatio(3.1);
SetTimeElapseOnlySleepOnReopen(&options);
ASSERT_OK(TryReopen(options));
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
// Create 4 files in L0
for (char v = 'a'; v <= 'd'; v++) {
ASSERT_OK(Put("Key2", DummyString(1024, v)));
ASSERT_OK(Put("Key3", DummyString(1024, v)));
ASSERT_OK(Put("Key4", DummyString(1024, v)));
ASSERT_OK(Put("Key1", DummyString(1024, v)));
ASSERT_OK(Put("Key4", DummyString(1024, v)));
ASSERT_OK(Flush());
}
// We created 4 sst files in L0
ASSERT_EQ("4", FilesPerLevel(0));
// Compaction will move the 4 files in L0 to trash and create 1 L1 file.
// Use kForceOptimized to not rewrite the new L1 file.
CompactRangeOptions cro;
cro.bottommost_level_compaction = BottommostLevelCompaction::kForceOptimized;
ASSERT_OK(db_->CompactRange(cro, nullptr, nullptr));
ASSERT_OK(dbfull()->TEST_WaitForCompact());
ASSERT_EQ("0,1", FilesPerLevel(0));
sfm->WaitForEmptyTrash();
ASSERT_EQ(penalties.size(), 8);
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
}
class DBWALTestWithParam
: public DBTestBase,
public testing::WithParamInterface<std::tuple<std::string, bool>> {
public:
explicit DBWALTestWithParam()
: DBTestBase("db_wal_test_with_params", /*env_do_fsync=*/true) {
wal_dir_ = std::get<0>(GetParam());
wal_dir_same_as_dbname_ = std::get<1>(GetParam());
}
std::string wal_dir_;
bool wal_dir_same_as_dbname_;
};
TEST_P(DBWALTestWithParam, WALTrashCleanupOnOpen) {
class MyEnv : public EnvWrapper {
public:
MyEnv(Env* t) : EnvWrapper(t), fake_log_delete(false) {}
const char* Name() const override { return "MyEnv"; }
Status DeleteFile(const std::string& fname) override {
if (fname.find(".log.trash") != std::string::npos && fake_log_delete) {
return Status::OK();
}
return target()->DeleteFile(fname);
}
void set_fake_log_delete(bool fake) { fake_log_delete = fake; }
private:
bool fake_log_delete;
};
std::unique_ptr<MyEnv> env(new MyEnv(env_));
Destroy(last_options_);
env->set_fake_log_delete(true);
Options options = CurrentOptions();
options.disable_auto_compactions = true;
options.compression = kNoCompression;
options.env = env.get();
options.wal_dir = dbname_ + wal_dir_;
int64_t rate_bytes_per_sec = 1024 * 10; // 10 Kbs / Sec
Status s;
options.sst_file_manager.reset(
NewSstFileManager(env_, nullptr, "", 0, false, &s, 0));
ASSERT_OK(s);
options.sst_file_manager->SetDeleteRateBytesPerSecond(rate_bytes_per_sec);
auto sfm = static_cast<SstFileManagerImpl*>(options.sst_file_manager.get());
sfm->delete_scheduler()->SetMaxTrashDBRatio(3.1);
Reopen(options);
// Create 4 files in L0
for (char v = 'a'; v <= 'd'; v++) {
if (v == 'c') {
// Maximize the change that the last log file will be preserved in trash
// before restarting the DB.
// We have to set this on the 2nd to last file for it to delay deletion
// on the last file. (Quirk of DeleteScheduler::BackgroundEmptyTrash())
options.sst_file_manager->SetDeleteRateBytesPerSecond(1);
}
ASSERT_OK(Put("Key2", DummyString(1024, v)));
ASSERT_OK(Put("Key3", DummyString(1024, v)));
ASSERT_OK(Put("Key4", DummyString(1024, v)));
ASSERT_OK(Put("Key1", DummyString(1024, v)));
ASSERT_OK(Put("Key4", DummyString(1024, v)));
ASSERT_OK(Flush());
}
// We created 4 sst files in L0
ASSERT_EQ("4", FilesPerLevel(0));
Close();
options.sst_file_manager.reset();
std::vector<std::string> filenames;
int trash_log_count = 0;
if (!wal_dir_same_as_dbname_) {
// Forcibly create some trash log files
std::unique_ptr<WritableFile> result;
ASSERT_OK(env->NewWritableFile(options.wal_dir + "/1000.log.trash", &result,
EnvOptions()));
result.reset();
}
ASSERT_OK(env->GetChildren(options.wal_dir, &filenames));
for (const std::string& fname : filenames) {
if (fname.find(".log.trash") != std::string::npos) {
trash_log_count++;
}
}
ASSERT_GE(trash_log_count, 1);
env->set_fake_log_delete(false);
Reopen(options);
filenames.clear();
trash_log_count = 0;
ASSERT_OK(env->GetChildren(options.wal_dir, &filenames));
for (const std::string& fname : filenames) {
if (fname.find(".log.trash") != std::string::npos) {
trash_log_count++;
}
}
ASSERT_EQ(trash_log_count, 0);
Close();
}
INSTANTIATE_TEST_CASE_P(DBWALTestWithParam, DBWALTestWithParam,
::testing::Values(std::make_tuple("", true),
std::make_tuple("_wal_dir", false)));
TEST_F(DBSSTTest, OpenDBWithExistingTrashAndObsoleteSstFile) {
Options options = CurrentOptions();
options.sst_file_manager.reset(
NewSstFileManager(env_, nullptr, "", 1024 * 1024 /* 1 MB/sec */));
auto sfm = static_cast<SstFileManagerImpl*>(options.sst_file_manager.get());
// Set an extra high trash ratio to prevent immediate/non-rate limited
// deletions
sfm->SetDeleteRateBytesPerSecond(1024 * 1024);
sfm->delete_scheduler()->SetMaxTrashDBRatio(1000.0);
int bg_delete_file = 0;
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
"DeleteScheduler::DeleteTrashFile:DeleteFile",
[&](void* /*arg*/) { bg_delete_file++; });
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
Destroy(last_options_);
// Add some trash files to the db directory so the DB can clean them up
ASSERT_OK(env_->CreateDirIfMissing(dbname_));
ASSERT_OK(
WriteStringToFile(env_, "abc", dbname_ + "/" + "001.sst.trash", false));
ASSERT_OK(
WriteStringToFile(env_, "abc", dbname_ + "/" + "002.sst.trash", false));
ASSERT_OK(
WriteStringToFile(env_, "abc", dbname_ + "/" + "003.sst.trash", false));
// Manually add an obsolete sst file. Obsolete SST files are discovered and
// deleted upon recovery.
constexpr uint64_t kSstFileNumber = 100;
const std::string kObsoleteSstFile =
MakeTableFileName(dbname_, kSstFileNumber);
ASSERT_OK(WriteStringToFile(env_, "abc", kObsoleteSstFile, false));
// Reopen the DB and verify that it deletes existing trash files and obsolete
// SST files with rate limiting.
Reopen(options);
sfm->WaitForEmptyTrash();
ASSERT_NOK(env_->FileExists(dbname_ + "/" + "001.sst.trash"));
ASSERT_NOK(env_->FileExists(dbname_ + "/" + "002.sst.trash"));
ASSERT_NOK(env_->FileExists(dbname_ + "/" + "003.sst.trash"));
ASSERT_NOK(env_->FileExists(kObsoleteSstFile));
ASSERT_EQ(bg_delete_file, 4);
}
// Create a DB with 2 db_paths, and generate multiple files in the 2
// db_paths using CompactRangeOptions, make sure that files that were
// deleted from first db_path were deleted using DeleteScheduler and
// files in the second path were not.
TEST_F(DBSSTTest, DeleteSchedulerMultipleDBPaths) {
std::atomic<int> bg_delete_file(0);
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
"DeleteScheduler::DeleteTrashFile:DeleteFile",
[&](void* /*arg*/) { bg_delete_file++; });
// The deletion scheduler sometimes skips marking file as trash according to
// a heuristic. In that case the deletion will go through the below SyncPoint.
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
"DeleteScheduler::DeleteFile", [&](void* /*arg*/) { bg_delete_file++; });
Options options = CurrentOptions();
options.disable_auto_compactions = true;
options.db_paths.emplace_back(dbname_, 1024 * 100);