forked from facebook/rocksdb
-
Notifications
You must be signed in to change notification settings - Fork 0
/
column_family.cc
1831 lines (1666 loc) · 70.6 KB
/
column_family.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/column_family.h"
#include <algorithm>
#include <cinttypes>
#include <limits>
#include <sstream>
#include <string>
#include <vector>
#include "db/blob/blob_file_cache.h"
#include "db/blob/blob_source.h"
#include "db/compaction/compaction_picker.h"
#include "db/compaction/compaction_picker_fifo.h"
#include "db/compaction/compaction_picker_level.h"
#include "db/compaction/compaction_picker_universal.h"
#include "db/db_impl/db_impl.h"
#include "db/internal_stats.h"
#include "db/job_context.h"
#include "db/range_del_aggregator.h"
#include "db/table_properties_collector.h"
#include "db/version_set.h"
#include "db/write_controller.h"
#include "file/sst_file_manager_impl.h"
#include "logging/logging.h"
#include "monitoring/thread_status_util.h"
#include "options/options_helper.h"
#include "port/port.h"
#include "rocksdb/convenience.h"
#include "rocksdb/table.h"
#include "table/merging_iterator.h"
#include "util/autovector.h"
#include "util/cast_util.h"
#include "util/compression.h"
namespace ROCKSDB_NAMESPACE {
ColumnFamilyHandleImpl::ColumnFamilyHandleImpl(
ColumnFamilyData* column_family_data, DBImpl* db, InstrumentedMutex* mutex)
: cfd_(column_family_data), db_(db), mutex_(mutex) {
if (cfd_ != nullptr) {
cfd_->Ref();
}
}
ColumnFamilyHandleImpl::~ColumnFamilyHandleImpl() {
if (cfd_ != nullptr) {
for (auto& listener : cfd_->ioptions()->listeners) {
listener->OnColumnFamilyHandleDeletionStarted(this);
}
// Job id == 0 means that this is not our background process, but rather
// user thread
// Need to hold some shared pointers owned by the initial_cf_options
// before final cleaning up finishes.
ColumnFamilyOptions initial_cf_options_copy = cfd_->initial_cf_options();
JobContext job_context(0);
mutex_->Lock();
bool dropped = cfd_->IsDropped();
if (cfd_->UnrefAndTryDelete()) {
if (dropped) {
db_->FindObsoleteFiles(&job_context, false, true);
}
}
mutex_->Unlock();
if (job_context.HaveSomethingToDelete()) {
bool defer_purge =
db_->immutable_db_options().avoid_unnecessary_blocking_io;
db_->PurgeObsoleteFiles(job_context, defer_purge);
}
job_context.Clean();
}
}
uint32_t ColumnFamilyHandleImpl::GetID() const { return cfd()->GetID(); }
const std::string& ColumnFamilyHandleImpl::GetName() const {
return cfd()->GetName();
}
Status ColumnFamilyHandleImpl::GetDescriptor(ColumnFamilyDescriptor* desc) {
// accessing mutable cf-options requires db mutex.
InstrumentedMutexLock l(mutex_);
*desc = ColumnFamilyDescriptor(cfd()->GetName(), cfd()->GetLatestCFOptions());
return Status::OK();
}
const Comparator* ColumnFamilyHandleImpl::GetComparator() const {
return cfd()->user_comparator();
}
void GetInternalTblPropCollFactory(
const ImmutableCFOptions& ioptions,
InternalTblPropCollFactories* internal_tbl_prop_coll_factories) {
assert(internal_tbl_prop_coll_factories);
auto& collector_factories = ioptions.table_properties_collector_factories;
for (size_t i = 0; i < ioptions.table_properties_collector_factories.size();
++i) {
assert(collector_factories[i]);
internal_tbl_prop_coll_factories->emplace_back(
new UserKeyTablePropertiesCollectorFactory(collector_factories[i]));
}
}
Status CheckCompressionSupported(const ColumnFamilyOptions& cf_options) {
if (!cf_options.compression_per_level.empty()) {
for (size_t level = 0; level < cf_options.compression_per_level.size();
++level) {
if (!CompressionTypeSupported(cf_options.compression_per_level[level])) {
return Status::InvalidArgument(
"Compression type " +
CompressionTypeToString(cf_options.compression_per_level[level]) +
" is not linked with the binary.");
}
}
} else {
if (!CompressionTypeSupported(cf_options.compression)) {
return Status::InvalidArgument(
"Compression type " +
CompressionTypeToString(cf_options.compression) +
" is not linked with the binary.");
}
}
if (cf_options.compression_opts.zstd_max_train_bytes > 0) {
if (cf_options.compression_opts.use_zstd_dict_trainer) {
if (!ZSTD_TrainDictionarySupported()) {
return Status::InvalidArgument(
"zstd dictionary trainer cannot be used because ZSTD 1.1.3+ "
"is not linked with the binary.");
}
} else if (!ZSTD_FinalizeDictionarySupported()) {
return Status::InvalidArgument(
"zstd finalizeDictionary cannot be used because ZSTD 1.4.5+ "
"is not linked with the binary.");
}
if (cf_options.compression_opts.max_dict_bytes == 0) {
return Status::InvalidArgument(
"The dictionary size limit (`CompressionOptions::max_dict_bytes`) "
"should be nonzero if we're using zstd's dictionary generator.");
}
}
if (!CompressionTypeSupported(cf_options.blob_compression_type)) {
std::ostringstream oss;
oss << "The specified blob compression type "
<< CompressionTypeToString(cf_options.blob_compression_type)
<< " is not available.";
return Status::InvalidArgument(oss.str());
}
return Status::OK();
}
Status CheckConcurrentWritesSupported(const ColumnFamilyOptions& cf_options) {
if (cf_options.inplace_update_support) {
return Status::InvalidArgument(
"In-place memtable updates (inplace_update_support) is not compatible "
"with concurrent writes (allow_concurrent_memtable_write)");
}
if (!cf_options.memtable_factory->IsInsertConcurrentlySupported()) {
return Status::InvalidArgument(
"Memtable doesn't concurrent writes (allow_concurrent_memtable_write)");
}
return Status::OK();
}
Status CheckCFPathsSupported(const DBOptions& db_options,
const ColumnFamilyOptions& cf_options) {
// More than one cf_paths are supported only in universal
// and level compaction styles. This function also checks the case
// in which cf_paths is not specified, which results in db_paths
// being used.
if ((cf_options.compaction_style != kCompactionStyleUniversal) &&
(cf_options.compaction_style != kCompactionStyleLevel)) {
if (cf_options.cf_paths.size() > 1) {
return Status::NotSupported(
"More than one CF paths are only supported in "
"universal and level compaction styles. ");
} else if (cf_options.cf_paths.empty() && db_options.db_paths.size() > 1) {
return Status::NotSupported(
"More than one DB paths are only supported in "
"universal and level compaction styles. ");
}
}
return Status::OK();
}
namespace {
const uint64_t kDefaultTtl = 0xfffffffffffffffe;
const uint64_t kDefaultPeriodicCompSecs = 0xfffffffffffffffe;
} // anonymous namespace
ColumnFamilyOptions SanitizeOptions(const ImmutableDBOptions& db_options,
const ColumnFamilyOptions& src) {
ColumnFamilyOptions result = src;
size_t clamp_max = std::conditional<
sizeof(size_t) == 4, std::integral_constant<size_t, 0xffffffff>,
std::integral_constant<uint64_t, 64ull << 30>>::type::value;
ClipToRange(&result.write_buffer_size, (static_cast<size_t>(64)) << 10,
clamp_max);
// if user sets arena_block_size, we trust user to use this value. Otherwise,
// calculate a proper value from writer_buffer_size;
if (result.arena_block_size <= 0) {
result.arena_block_size =
std::min(size_t{1024 * 1024}, result.write_buffer_size / 8);
// Align up to 4k
const size_t align = 4 * 1024;
result.arena_block_size =
((result.arena_block_size + align - 1) / align) * align;
}
result.min_write_buffer_number_to_merge =
std::min(result.min_write_buffer_number_to_merge,
result.max_write_buffer_number - 1);
if (result.min_write_buffer_number_to_merge < 1) {
result.min_write_buffer_number_to_merge = 1;
}
if (db_options.atomic_flush && result.min_write_buffer_number_to_merge > 1) {
ROCKS_LOG_WARN(
db_options.logger,
"Currently, if atomic_flush is true, then triggering flush for any "
"column family internally (non-manual flush) will trigger flushing "
"all column families even if the number of memtables is smaller "
"min_write_buffer_number_to_merge. Therefore, configuring "
"min_write_buffer_number_to_merge > 1 is not compatible and should "
"be satinized to 1. Not doing so will lead to data loss and "
"inconsistent state across multiple column families when WAL is "
"disabled, which is a common setting for atomic flush");
result.min_write_buffer_number_to_merge = 1;
}
if (result.num_levels < 1) {
result.num_levels = 1;
}
if (result.compaction_style == kCompactionStyleLevel &&
result.num_levels < 2) {
result.num_levels = 2;
}
if (result.compaction_style == kCompactionStyleUniversal &&
db_options.allow_ingest_behind && result.num_levels < 3) {
result.num_levels = 3;
}
if (result.max_write_buffer_number < 2) {
result.max_write_buffer_number = 2;
}
// fall back max_write_buffer_number_to_maintain if
// max_write_buffer_size_to_maintain is not set
if (result.max_write_buffer_size_to_maintain < 0) {
result.max_write_buffer_size_to_maintain =
result.max_write_buffer_number *
static_cast<int64_t>(result.write_buffer_size);
} else if (result.max_write_buffer_size_to_maintain == 0 &&
result.max_write_buffer_number_to_maintain < 0) {
result.max_write_buffer_number_to_maintain = result.max_write_buffer_number;
}
// bloom filter size shouldn't exceed 1/4 of memtable size.
if (result.memtable_prefix_bloom_size_ratio > 0.25) {
result.memtable_prefix_bloom_size_ratio = 0.25;
} else if (result.memtable_prefix_bloom_size_ratio < 0) {
result.memtable_prefix_bloom_size_ratio = 0;
}
if (!result.prefix_extractor) {
assert(result.memtable_factory);
Slice name = result.memtable_factory->Name();
if (name.compare("HashSkipListRepFactory") == 0 ||
name.compare("HashLinkListRepFactory") == 0) {
result.memtable_factory = std::make_shared<SkipListFactory>();
}
}
if (result.compaction_style == kCompactionStyleFIFO) {
// since we delete level0 files in FIFO compaction when there are too many
// of them, these options don't really mean anything
result.level0_slowdown_writes_trigger = std::numeric_limits<int>::max();
result.level0_stop_writes_trigger = std::numeric_limits<int>::max();
}
if (result.max_bytes_for_level_multiplier <= 0) {
result.max_bytes_for_level_multiplier = 1;
}
if (result.level0_file_num_compaction_trigger == 0) {
ROCKS_LOG_WARN(db_options.logger,
"level0_file_num_compaction_trigger cannot be 0");
result.level0_file_num_compaction_trigger = 1;
}
if (result.level0_stop_writes_trigger <
result.level0_slowdown_writes_trigger ||
result.level0_slowdown_writes_trigger <
result.level0_file_num_compaction_trigger) {
ROCKS_LOG_WARN(db_options.logger,
"This condition must be satisfied: "
"level0_stop_writes_trigger(%d) >= "
"level0_slowdown_writes_trigger(%d) >= "
"level0_file_num_compaction_trigger(%d)",
result.level0_stop_writes_trigger,
result.level0_slowdown_writes_trigger,
result.level0_file_num_compaction_trigger);
if (result.level0_slowdown_writes_trigger <
result.level0_file_num_compaction_trigger) {
result.level0_slowdown_writes_trigger =
result.level0_file_num_compaction_trigger;
}
if (result.level0_stop_writes_trigger <
result.level0_slowdown_writes_trigger) {
result.level0_stop_writes_trigger = result.level0_slowdown_writes_trigger;
}
ROCKS_LOG_WARN(db_options.logger,
"Adjust the value to "
"level0_stop_writes_trigger(%d) "
"level0_slowdown_writes_trigger(%d) "
"level0_file_num_compaction_trigger(%d)",
result.level0_stop_writes_trigger,
result.level0_slowdown_writes_trigger,
result.level0_file_num_compaction_trigger);
}
if (result.soft_pending_compaction_bytes_limit == 0) {
result.soft_pending_compaction_bytes_limit =
result.hard_pending_compaction_bytes_limit;
} else if (result.hard_pending_compaction_bytes_limit > 0 &&
result.soft_pending_compaction_bytes_limit >
result.hard_pending_compaction_bytes_limit) {
result.soft_pending_compaction_bytes_limit =
result.hard_pending_compaction_bytes_limit;
}
// When the DB is stopped, it's possible that there are some .trash files that
// were not deleted yet, when we open the DB we will find these .trash files
// and schedule them to be deleted (or delete immediately if SstFileManager
// was not used)
auto sfm =
static_cast<SstFileManagerImpl*>(db_options.sst_file_manager.get());
for (size_t i = 0; i < result.cf_paths.size(); i++) {
DeleteScheduler::CleanupDirectory(db_options.env, sfm,
result.cf_paths[i].path)
.PermitUncheckedError();
}
if (result.cf_paths.empty()) {
result.cf_paths = db_options.db_paths;
}
if (result.level_compaction_dynamic_level_bytes) {
if (result.compaction_style != kCompactionStyleLevel) {
ROCKS_LOG_INFO(db_options.info_log.get(),
"level_compaction_dynamic_level_bytes only makes sense "
"for level-based compaction");
result.level_compaction_dynamic_level_bytes = false;
} else if (result.cf_paths.size() > 1U) {
// we don't yet know how to make both of this feature and multiple
// DB path work.
ROCKS_LOG_WARN(db_options.info_log.get(),
"multiple cf_paths/db_paths and "
"level_compaction_dynamic_level_bytes "
"can't be used together");
result.level_compaction_dynamic_level_bytes = false;
}
}
if (result.max_compaction_bytes == 0) {
result.max_compaction_bytes = result.target_file_size_base * 25;
}
bool is_block_based_table = (result.table_factory->IsInstanceOf(
TableFactory::kBlockBasedTableName()));
const uint64_t kAdjustedTtl = 30 * 24 * 60 * 60;
if (result.ttl == kDefaultTtl) {
if (is_block_based_table) {
// FIFO also requires max_open_files=-1, which is checked in
// ValidateOptions().
result.ttl = kAdjustedTtl;
} else {
result.ttl = 0;
}
}
const uint64_t kAdjustedPeriodicCompSecs = 30 * 24 * 60 * 60;
if (result.compaction_style == kCompactionStyleLevel) {
if ((result.compaction_filter != nullptr ||
result.compaction_filter_factory != nullptr) &&
result.periodic_compaction_seconds == kDefaultPeriodicCompSecs &&
is_block_based_table) {
result.periodic_compaction_seconds = kAdjustedPeriodicCompSecs;
}
} else if (result.compaction_style == kCompactionStyleUniversal) {
if (result.periodic_compaction_seconds == kDefaultPeriodicCompSecs &&
is_block_based_table) {
result.periodic_compaction_seconds = kAdjustedPeriodicCompSecs;
}
} else if (result.compaction_style == kCompactionStyleFIFO) {
if (result.periodic_compaction_seconds != kDefaultPeriodicCompSecs) {
ROCKS_LOG_WARN(
db_options.info_log.get(),
"periodic_compaction_seconds does not support FIFO compaction. You"
"may want to set option TTL instead.");
}
if (result.last_level_temperature != Temperature::kUnknown) {
ROCKS_LOG_WARN(
db_options.info_log.get(),
"last_level_temperature is ignored with FIFO compaction. Consider "
"CompactionOptionsFIFO::file_temperature_age_thresholds.");
result.last_level_temperature = Temperature::kUnknown;
}
}
// For universal compaction, `ttl` and `periodic_compaction_seconds` mean the
// same thing, take the stricter value.
if (result.compaction_style == kCompactionStyleUniversal) {
if (result.periodic_compaction_seconds == 0) {
result.periodic_compaction_seconds = result.ttl;
} else if (result.ttl != 0) {
result.periodic_compaction_seconds =
std::min(result.ttl, result.periodic_compaction_seconds);
}
}
if (result.periodic_compaction_seconds == kDefaultPeriodicCompSecs) {
result.periodic_compaction_seconds = 0;
}
return result;
}
int SuperVersion::dummy = 0;
void* const SuperVersion::kSVInUse = &SuperVersion::dummy;
void* const SuperVersion::kSVObsolete = nullptr;
SuperVersion::~SuperVersion() {
for (auto td : to_delete) {
delete td;
}
}
SuperVersion* SuperVersion::Ref() {
refs.fetch_add(1, std::memory_order_relaxed);
return this;
}
bool SuperVersion::Unref() {
// fetch_sub returns the previous value of ref
uint32_t previous_refs = refs.fetch_sub(1);
assert(previous_refs > 0);
return previous_refs == 1;
}
void SuperVersion::Cleanup() {
assert(refs.load(std::memory_order_relaxed) == 0);
// Since this SuperVersion object is being deleted,
// decrement reference to the immutable MemtableList
// this SV object was pointing to.
imm->Unref(&to_delete);
MemTable* m = mem->Unref();
if (m != nullptr) {
auto* memory_usage = current->cfd()->imm()->current_memory_usage();
assert(*memory_usage >= m->ApproximateMemoryUsage());
*memory_usage -= m->ApproximateMemoryUsage();
to_delete.push_back(m);
}
current->Unref();
cfd->UnrefAndTryDelete();
}
void SuperVersion::Init(
ColumnFamilyData* new_cfd, MemTable* new_mem, MemTableListVersion* new_imm,
Version* new_current,
std::shared_ptr<const SeqnoToTimeMapping> new_seqno_to_time_mapping) {
cfd = new_cfd;
mem = new_mem;
imm = new_imm;
current = new_current;
full_history_ts_low = cfd->GetFullHistoryTsLow();
seqno_to_time_mapping = std::move(new_seqno_to_time_mapping);
cfd->Ref();
mem->Ref();
imm->Ref();
current->Ref();
refs.store(1, std::memory_order_relaxed);
}
namespace {
void SuperVersionUnrefHandle(void* ptr) {
// UnrefHandle is called when a thread exits or a ThreadLocalPtr gets
// destroyed. When the former happens, the thread shouldn't see kSVInUse.
// When the latter happens, only super_version_ holds a reference
// to ColumnFamilyData, so no further queries are possible.
SuperVersion* sv = static_cast<SuperVersion*>(ptr);
bool was_last_ref __attribute__((__unused__));
was_last_ref = sv->Unref();
// Thread-local SuperVersions can't outlive ColumnFamilyData::super_version_.
// This is important because we can't do SuperVersion cleanup here.
// That would require locking DB mutex, which would deadlock because
// SuperVersionUnrefHandle is called with locked ThreadLocalPtr mutex.
assert(!was_last_ref);
}
} // anonymous namespace
std::vector<std::string> ColumnFamilyData::GetDbPaths() const {
std::vector<std::string> paths;
paths.reserve(ioptions_.cf_paths.size());
for (const DbPath& db_path : ioptions_.cf_paths) {
paths.emplace_back(db_path.path);
}
return paths;
}
const uint32_t ColumnFamilyData::kDummyColumnFamilyDataId =
std::numeric_limits<uint32_t>::max();
ColumnFamilyData::ColumnFamilyData(
uint32_t id, const std::string& name, Version* _dummy_versions,
Cache* _table_cache, WriteBufferManager* write_buffer_manager,
const ColumnFamilyOptions& cf_options, const ImmutableDBOptions& db_options,
const FileOptions* file_options, ColumnFamilySet* column_family_set,
BlockCacheTracer* const block_cache_tracer,
const std::shared_ptr<IOTracer>& io_tracer, const std::string& db_id,
const std::string& db_session_id)
: id_(id),
name_(name),
dummy_versions_(_dummy_versions),
current_(nullptr),
refs_(0),
initialized_(false),
dropped_(false),
internal_comparator_(cf_options.comparator),
initial_cf_options_(SanitizeOptions(db_options, cf_options)),
ioptions_(db_options, initial_cf_options_),
mutable_cf_options_(initial_cf_options_),
is_delete_range_supported_(
cf_options.table_factory->IsDeleteRangeSupported()),
write_buffer_manager_(write_buffer_manager),
mem_(nullptr),
imm_(ioptions_.min_write_buffer_number_to_merge,
ioptions_.max_write_buffer_number_to_maintain,
ioptions_.max_write_buffer_size_to_maintain),
super_version_(nullptr),
super_version_number_(0),
local_sv_(new ThreadLocalPtr(&SuperVersionUnrefHandle)),
next_(nullptr),
prev_(nullptr),
log_number_(0),
column_family_set_(column_family_set),
queued_for_flush_(false),
queued_for_compaction_(false),
prev_compaction_needed_bytes_(0),
allow_2pc_(db_options.allow_2pc),
last_memtable_id_(0),
db_paths_registered_(false),
mempurge_used_(false),
next_epoch_number_(1) {
if (id_ != kDummyColumnFamilyDataId) {
// TODO(cc): RegisterDbPaths can be expensive, considering moving it
// outside of this constructor which might be called with db mutex held.
// TODO(cc): considering using ioptions_.fs, currently some tests rely on
// EnvWrapper, that's the main reason why we use env here.
Status s = ioptions_.env->RegisterDbPaths(GetDbPaths());
if (s.ok()) {
db_paths_registered_ = true;
} else {
ROCKS_LOG_ERROR(
ioptions_.logger,
"Failed to register data paths of column family (id: %d, name: %s)",
id_, name_.c_str());
}
}
Ref();
// Convert user defined table properties collector factories to internal ones.
GetInternalTblPropCollFactory(ioptions_, &internal_tbl_prop_coll_factories_);
// if _dummy_versions is nullptr, then this is a dummy column family.
if (_dummy_versions != nullptr) {
internal_stats_.reset(
new InternalStats(ioptions_.num_levels, ioptions_.clock, this));
table_cache_.reset(new TableCache(ioptions_, file_options, _table_cache,
block_cache_tracer, io_tracer,
db_session_id));
blob_file_cache_.reset(
new BlobFileCache(_table_cache, ioptions(), soptions(), id_,
internal_stats_->GetBlobFileReadHist(), io_tracer));
blob_source_.reset(new BlobSource(ioptions(), db_id, db_session_id,
blob_file_cache_.get()));
if (ioptions_.compaction_style == kCompactionStyleLevel) {
compaction_picker_.reset(
new LevelCompactionPicker(ioptions_, &internal_comparator_));
} else if (ioptions_.compaction_style == kCompactionStyleUniversal) {
compaction_picker_.reset(
new UniversalCompactionPicker(ioptions_, &internal_comparator_));
} else if (ioptions_.compaction_style == kCompactionStyleFIFO) {
compaction_picker_.reset(
new FIFOCompactionPicker(ioptions_, &internal_comparator_));
} else if (ioptions_.compaction_style == kCompactionStyleNone) {
compaction_picker_.reset(
new NullCompactionPicker(ioptions_, &internal_comparator_));
ROCKS_LOG_WARN(ioptions_.logger,
"Column family %s does not use any background compaction. "
"Compactions can only be done via CompactFiles\n",
GetName().c_str());
} else {
ROCKS_LOG_ERROR(ioptions_.logger,
"Unable to recognize the specified compaction style %d. "
"Column family %s will use kCompactionStyleLevel.\n",
ioptions_.compaction_style, GetName().c_str());
compaction_picker_.reset(
new LevelCompactionPicker(ioptions_, &internal_comparator_));
}
if (column_family_set_->NumberOfColumnFamilies() < 10) {
ROCKS_LOG_INFO(ioptions_.logger,
"--------------- Options for column family [%s]:\n",
name.c_str());
initial_cf_options_.Dump(ioptions_.logger);
} else {
ROCKS_LOG_INFO(ioptions_.logger, "\t(skipping printing options)\n");
}
}
RecalculateWriteStallConditions(mutable_cf_options_);
if (cf_options.table_factory->IsInstanceOf(
TableFactory::kBlockBasedTableName()) &&
cf_options.table_factory->GetOptions<BlockBasedTableOptions>()) {
const BlockBasedTableOptions* bbto =
cf_options.table_factory->GetOptions<BlockBasedTableOptions>();
const auto& options_overrides = bbto->cache_usage_options.options_overrides;
const auto file_metadata_charged =
options_overrides.at(CacheEntryRole::kFileMetadata).charged;
if (bbto->block_cache &&
file_metadata_charged == CacheEntryRoleOptions::Decision::kEnabled) {
// TODO(hx235): Add a `ConcurrentCacheReservationManager` at DB scope
// responsible for reservation of `ObsoleteFileInfo` so that we can keep
// this `file_metadata_cache_res_mgr_` nonconcurrent
file_metadata_cache_res_mgr_.reset(new ConcurrentCacheReservationManager(
std::make_shared<
CacheReservationManagerImpl<CacheEntryRole::kFileMetadata>>(
bbto->block_cache)));
}
}
}
// DB mutex held
ColumnFamilyData::~ColumnFamilyData() {
assert(refs_.load(std::memory_order_relaxed) == 0);
// remove from linked list
auto prev = prev_;
auto next = next_;
prev->next_ = next;
next->prev_ = prev;
if (!dropped_ && column_family_set_ != nullptr) {
// If it's dropped, it's already removed from column family set
// If column_family_set_ == nullptr, this is dummy CFD and not in
// ColumnFamilySet
column_family_set_->RemoveColumnFamily(this);
}
if (current_ != nullptr) {
current_->Unref();
}
// It would be wrong if this ColumnFamilyData is in flush_queue_ or
// compaction_queue_ and we destroyed it
assert(!queued_for_flush_);
assert(!queued_for_compaction_);
assert(super_version_ == nullptr);
if (dummy_versions_ != nullptr) {
// List must be empty
assert(dummy_versions_->Next() == dummy_versions_);
bool deleted __attribute__((__unused__));
deleted = dummy_versions_->Unref();
assert(deleted);
}
if (mem_ != nullptr) {
delete mem_->Unref();
}
autovector<MemTable*> to_delete;
imm_.current()->Unref(&to_delete);
for (MemTable* m : to_delete) {
delete m;
}
if (db_paths_registered_) {
// TODO(cc): considering using ioptions_.fs, currently some tests rely on
// EnvWrapper, that's the main reason why we use env here.
Status s = ioptions_.env->UnregisterDbPaths(GetDbPaths());
if (!s.ok()) {
ROCKS_LOG_ERROR(
ioptions_.logger,
"Failed to unregister data paths of column family (id: %d, name: %s)",
id_, name_.c_str());
}
}
}
bool ColumnFamilyData::UnrefAndTryDelete() {
int old_refs = refs_.fetch_sub(1);
assert(old_refs > 0);
if (old_refs == 1) {
assert(super_version_ == nullptr);
delete this;
return true;
}
if (old_refs == 2 && super_version_ != nullptr) {
// Only the super_version_ holds me
SuperVersion* sv = super_version_;
super_version_ = nullptr;
// Release SuperVersion references kept in ThreadLocalPtr.
local_sv_.reset();
if (sv->Unref()) {
// Note: sv will delete this ColumnFamilyData during Cleanup()
assert(sv->cfd == this);
sv->Cleanup();
delete sv;
return true;
}
}
return false;
}
void ColumnFamilyData::SetDropped() {
// can't drop default CF
assert(id_ != 0);
dropped_ = true;
write_controller_token_.reset();
// remove from column_family_set
column_family_set_->RemoveColumnFamily(this);
}
ColumnFamilyOptions ColumnFamilyData::GetLatestCFOptions() const {
return BuildColumnFamilyOptions(initial_cf_options_, mutable_cf_options_);
}
uint64_t ColumnFamilyData::OldestLogToKeep() {
auto current_log = GetLogNumber();
if (allow_2pc_) {
auto imm_prep_log = imm()->PrecomputeMinLogContainingPrepSection();
auto mem_prep_log = mem()->GetMinLogContainingPrepSection();
if (imm_prep_log > 0 && imm_prep_log < current_log) {
current_log = imm_prep_log;
}
if (mem_prep_log > 0 && mem_prep_log < current_log) {
current_log = mem_prep_log;
}
}
return current_log;
}
const double kIncSlowdownRatio = 0.8;
const double kDecSlowdownRatio = 1 / kIncSlowdownRatio;
const double kNearStopSlowdownRatio = 0.6;
const double kDelayRecoverSlowdownRatio = 1.4;
namespace {
// If penalize_stop is true, we further reduce slowdown rate.
std::unique_ptr<WriteControllerToken> SetupDelay(
WriteController* write_controller, uint64_t compaction_needed_bytes,
uint64_t prev_compaction_need_bytes, bool penalize_stop,
bool auto_compactions_disabled) {
const uint64_t kMinWriteRate = 16 * 1024u; // Minimum write rate 16KB/s.
uint64_t max_write_rate = write_controller->max_delayed_write_rate();
uint64_t write_rate = write_controller->delayed_write_rate();
if (auto_compactions_disabled) {
// When auto compaction is disabled, always use the value user gave.
write_rate = max_write_rate;
} else if (write_controller->NeedsDelay() && max_write_rate > kMinWriteRate) {
// If user gives rate less than kMinWriteRate, don't adjust it.
//
// If already delayed, need to adjust based on previous compaction debt.
// When there are two or more column families require delay, we always
// increase or reduce write rate based on information for one single
// column family. It is likely to be OK but we can improve if there is a
// problem.
// Ignore compaction_needed_bytes = 0 case because compaction_needed_bytes
// is only available in level-based compaction
//
// If the compaction debt stays the same as previously, we also further slow
// down. It usually means a mem table is full. It's mainly for the case
// where both of flush and compaction are much slower than the speed we
// insert to mem tables, so we need to actively slow down before we get
// feedback signal from compaction and flushes to avoid the full stop
// because of hitting the max write buffer number.
//
// If DB just falled into the stop condition, we need to further reduce
// the write rate to avoid the stop condition.
if (penalize_stop) {
// Penalize the near stop or stop condition by more aggressive slowdown.
// This is to provide the long term slowdown increase signal.
// The penalty is more than the reward of recovering to the normal
// condition.
write_rate = static_cast<uint64_t>(static_cast<double>(write_rate) *
kNearStopSlowdownRatio);
if (write_rate < kMinWriteRate) {
write_rate = kMinWriteRate;
}
} else if (prev_compaction_need_bytes > 0 &&
prev_compaction_need_bytes <= compaction_needed_bytes) {
write_rate = static_cast<uint64_t>(static_cast<double>(write_rate) *
kIncSlowdownRatio);
if (write_rate < kMinWriteRate) {
write_rate = kMinWriteRate;
}
} else if (prev_compaction_need_bytes > compaction_needed_bytes) {
// We are speeding up by ratio of kSlowdownRatio when we have paid
// compaction debt. But we'll never speed up to faster than the write rate
// given by users.
write_rate = static_cast<uint64_t>(static_cast<double>(write_rate) *
kDecSlowdownRatio);
if (write_rate > max_write_rate) {
write_rate = max_write_rate;
}
}
}
return write_controller->GetDelayToken(write_rate);
}
int GetL0FileCountForCompactionSpeedup(int level0_file_num_compaction_trigger,
int level0_slowdown_writes_trigger) {
// SanitizeOptions() ensures it.
assert(level0_file_num_compaction_trigger <= level0_slowdown_writes_trigger);
if (level0_file_num_compaction_trigger < 0) {
return std::numeric_limits<int>::max();
}
const int64_t twice_level0_trigger =
static_cast<int64_t>(level0_file_num_compaction_trigger) * 2;
const int64_t one_fourth_trigger_slowdown =
static_cast<int64_t>(level0_file_num_compaction_trigger) +
((level0_slowdown_writes_trigger - level0_file_num_compaction_trigger) /
4);
assert(twice_level0_trigger >= 0);
assert(one_fourth_trigger_slowdown >= 0);
// 1/4 of the way between L0 compaction trigger threshold and slowdown
// condition.
// Or twice as compaction trigger, if it is smaller.
int64_t res = std::min(twice_level0_trigger, one_fourth_trigger_slowdown);
if (res >= std::numeric_limits<int32_t>::max()) {
return std::numeric_limits<int32_t>::max();
} else {
// res fits in int
return static_cast<int>(res);
}
}
uint64_t GetPendingCompactionBytesForCompactionSpeedup(
const MutableCFOptions& mutable_cf_options,
const VersionStorageInfo* vstorage) {
// Compaction debt relatively large compared to the stable (bottommost) data
// size indicates compaction fell behind.
const uint64_t kBottommostSizeDivisor = 8;
// Meaningful progress toward the slowdown trigger is another good indication.
const uint64_t kSlowdownTriggerDivisor = 4;
uint64_t bottommost_files_size = 0;
for (const auto& level_and_file : vstorage->BottommostFiles()) {
bottommost_files_size += level_and_file.second->fd.GetFileSize();
}
// Slowdown trigger might be zero but that means compaction speedup should
// always happen (undocumented/historical), so no special treatment is needed.
uint64_t slowdown_threshold =
mutable_cf_options.soft_pending_compaction_bytes_limit /
kSlowdownTriggerDivisor;
// Size of zero, however, should not be used to decide to speedup compaction.
if (bottommost_files_size == 0) {
return slowdown_threshold;
}
uint64_t size_threshold = bottommost_files_size / kBottommostSizeDivisor;
return std::min(size_threshold, slowdown_threshold);
}
uint64_t GetMarkedFileCountForCompactionSpeedup() {
// When just one file is marked, it is not clear that parallel compaction will
// help the compaction that the user nicely requested to happen sooner. When
// multiple files are marked, however, it is pretty clearly helpful, except
// for the rare case in which a single compaction grabs all the marked files.
return 2;
}
} // anonymous namespace
std::pair<WriteStallCondition, WriteStallCause>
ColumnFamilyData::GetWriteStallConditionAndCause(
int num_unflushed_memtables, int num_l0_files,
uint64_t num_compaction_needed_bytes,
const MutableCFOptions& mutable_cf_options,
const ImmutableCFOptions& immutable_cf_options) {
if (num_unflushed_memtables >= mutable_cf_options.max_write_buffer_number) {
return {WriteStallCondition::kStopped, WriteStallCause::kMemtableLimit};
} else if (!mutable_cf_options.disable_auto_compactions &&
num_l0_files >= mutable_cf_options.level0_stop_writes_trigger) {
return {WriteStallCondition::kStopped, WriteStallCause::kL0FileCountLimit};
} else if (!mutable_cf_options.disable_auto_compactions &&
mutable_cf_options.hard_pending_compaction_bytes_limit > 0 &&
num_compaction_needed_bytes >=
mutable_cf_options.hard_pending_compaction_bytes_limit) {
return {WriteStallCondition::kStopped,
WriteStallCause::kPendingCompactionBytes};
} else if (mutable_cf_options.max_write_buffer_number > 3 &&
num_unflushed_memtables >=
mutable_cf_options.max_write_buffer_number - 1 &&
num_unflushed_memtables - 1 >=
immutable_cf_options.min_write_buffer_number_to_merge) {
return {WriteStallCondition::kDelayed, WriteStallCause::kMemtableLimit};
} else if (!mutable_cf_options.disable_auto_compactions &&
mutable_cf_options.level0_slowdown_writes_trigger >= 0 &&
num_l0_files >=
mutable_cf_options.level0_slowdown_writes_trigger) {
return {WriteStallCondition::kDelayed, WriteStallCause::kL0FileCountLimit};
} else if (!mutable_cf_options.disable_auto_compactions &&
mutable_cf_options.soft_pending_compaction_bytes_limit > 0 &&
num_compaction_needed_bytes >=
mutable_cf_options.soft_pending_compaction_bytes_limit) {
return {WriteStallCondition::kDelayed,
WriteStallCause::kPendingCompactionBytes};
}
return {WriteStallCondition::kNormal, WriteStallCause::kNone};
}
WriteStallCondition ColumnFamilyData::RecalculateWriteStallConditions(
const MutableCFOptions& mutable_cf_options) {
auto write_stall_condition = WriteStallCondition::kNormal;
if (current_ != nullptr) {
auto* vstorage = current_->storage_info();
auto write_controller = column_family_set_->write_controller_;
uint64_t compaction_needed_bytes =
vstorage->estimated_compaction_needed_bytes();
auto write_stall_condition_and_cause = GetWriteStallConditionAndCause(
imm()->NumNotFlushed(), vstorage->l0_delay_trigger_count(),
vstorage->estimated_compaction_needed_bytes(), mutable_cf_options,
*ioptions());
write_stall_condition = write_stall_condition_and_cause.first;
auto write_stall_cause = write_stall_condition_and_cause.second;
bool was_stopped = write_controller->IsStopped();
bool needed_delay = write_controller->NeedsDelay();
if (write_stall_condition == WriteStallCondition::kStopped &&
write_stall_cause == WriteStallCause::kMemtableLimit) {
write_controller_token_ = write_controller->GetStopToken();
internal_stats_->AddCFStats(InternalStats::MEMTABLE_LIMIT_STOPS, 1);
ROCKS_LOG_WARN(
ioptions_.logger,
"[%s] Stopping writes because we have %d immutable memtables "
"(waiting for flush), max_write_buffer_number is set to %d",
name_.c_str(), imm()->NumNotFlushed(),
mutable_cf_options.max_write_buffer_number);
} else if (write_stall_condition == WriteStallCondition::kStopped &&
write_stall_cause == WriteStallCause::kL0FileCountLimit) {
write_controller_token_ = write_controller->GetStopToken();
internal_stats_->AddCFStats(InternalStats::L0_FILE_COUNT_LIMIT_STOPS, 1);
if (compaction_picker_->IsLevel0CompactionInProgress()) {
internal_stats_->AddCFStats(
InternalStats::L0_FILE_COUNT_LIMIT_STOPS_WITH_ONGOING_COMPACTION,
1);
}
ROCKS_LOG_WARN(ioptions_.logger,
"[%s] Stopping writes because we have %d level-0 files",
name_.c_str(), vstorage->l0_delay_trigger_count());
} else if (write_stall_condition == WriteStallCondition::kStopped &&
write_stall_cause == WriteStallCause::kPendingCompactionBytes) {
write_controller_token_ = write_controller->GetStopToken();
internal_stats_->AddCFStats(
InternalStats::PENDING_COMPACTION_BYTES_LIMIT_STOPS, 1);
ROCKS_LOG_WARN(