Line data Source code
1 : // Copyright (c) 2011 The LevelDB Authors. All rights reserved.
2 : // Use of this source code is governed by a BSD-style license that can be
3 : // found in the LICENSE file. See the AUTHORS file for names of contributors.
4 :
5 : #include "db/db_impl.h"
6 :
7 : #include <algorithm>
8 : #include <set>
9 : #include <string>
10 : #include <stdint.h>
11 : #include <stdio.h>
12 : #include <vector>
13 : #include "db/builder.h"
14 : #include "db/db_iter.h"
15 : #include "db/dbformat.h"
16 : #include "db/filename.h"
17 : #include "db/log_reader.h"
18 : #include "db/log_writer.h"
19 : #include "db/memtable.h"
20 : #include "db/table_cache.h"
21 : #include "db/version_set.h"
22 : #include "db/write_batch_internal.h"
23 : #include "leveldb/db.h"
24 : #include "leveldb/env.h"
25 : #include "leveldb/status.h"
26 : #include "leveldb/table.h"
27 : #include "leveldb/table_builder.h"
28 : #include "port/port.h"
29 : #include "table/block.h"
30 : #include "table/merger.h"
31 : #include "table/two_level_iterator.h"
32 : #include "util/coding.h"
33 : #include "util/logging.h"
34 : #include "util/mutexlock.h"
35 :
36 : namespace leveldb {
37 :
38 : const int kNumNonTableCacheFiles = 10;
39 :
40 : // Information kept for every waiting writer
41 1102 : struct DBImpl::Writer {
42 : Status status;
43 : WriteBatch* batch;
44 : bool sync;
45 : bool done;
46 : port::CondVar cv;
47 :
48 1102 : explicit Writer(port::Mutex* mu) : cv(mu) { }
49 : };
50 :
51 7 : struct DBImpl::CompactionState {
52 : Compaction* const compaction;
53 :
54 : // Sequence numbers < smallest_snapshot are not significant since we
55 : // will never have to service a snapshot below smallest_snapshot.
56 : // Therefore if we have seen a sequence number S <= smallest_snapshot,
57 : // we can drop all entries for the same key with sequence numbers < S.
58 : SequenceNumber smallest_snapshot;
59 :
60 : // Files produced by compaction
61 77 : struct Output {
62 : uint64_t number;
63 : uint64_t file_size;
64 : InternalKey smallest, largest;
65 : };
66 : std::vector<Output> outputs;
67 :
68 : // State kept for output being generated
69 : WritableFile* outfile;
70 : TableBuilder* builder;
71 :
72 : uint64_t total_bytes;
73 :
74 3080 : Output* current_output() { return &outputs[outputs.size()-1]; }
75 :
76 0 : explicit CompactionState(Compaction* c)
77 : : compaction(c),
78 : outfile(NULL),
79 : builder(NULL),
80 14 : total_bytes(0) {
81 0 : }
82 : };
83 :
84 : // Fix user-supplied options to be reasonable
85 : template <class T,class V>
86 : static void ClipToRange(T* ptr, V minvalue, V maxvalue) {
87 732 : if (static_cast<V>(*ptr) > maxvalue) *ptr = maxvalue;
88 732 : if (static_cast<V>(*ptr) < minvalue) *ptr = minvalue;
89 : }
90 244 : Options SanitizeOptions(const std::string& dbname,
91 : const InternalKeyComparator* icmp,
92 : const InternalFilterPolicy* ipolicy,
93 : const Options& src) {
94 244 : Options result = src;
95 244 : result.comparator = icmp;
96 244 : result.filter_policy = (src.filter_policy != NULL) ? ipolicy : NULL;
97 : ClipToRange(&result.max_open_files, 64 + kNumNonTableCacheFiles, 50000);
98 : ClipToRange(&result.write_buffer_size, 64<<10, 1<<30);
99 : ClipToRange(&result.block_size, 1<<10, 4<<20);
100 244 : if (result.info_log == NULL) {
101 : // Open a log file in the same directory as the db
102 488 : src.env->CreateDir(dbname); // In case it does not exist
103 976 : src.env->RenameFile(InfoLogFileName(dbname), OldInfoLogFileName(dbname));
104 488 : Status s = src.env->NewLogger(InfoLogFileName(dbname), &result.info_log);
105 244 : if (!s.ok()) {
106 : // No place suitable for logging
107 0 : result.info_log = NULL;
108 : }
109 : }
110 244 : if (result.block_cache == NULL) {
111 0 : result.block_cache = NewLRUCache(8 << 20);
112 : }
113 244 : return result;
114 : }
115 :
116 244 : DBImpl::DBImpl(const Options& raw_options, const std::string& dbname)
117 : : env_(raw_options.env),
118 : internal_comparator_(raw_options.comparator),
119 : internal_filter_policy_(raw_options.filter_policy),
120 : options_(SanitizeOptions(dbname, &internal_comparator_,
121 244 : &internal_filter_policy_, raw_options)),
122 244 : owns_info_log_(options_.info_log != raw_options.info_log),
123 244 : owns_cache_(options_.block_cache != raw_options.block_cache),
124 : dbname_(dbname),
125 : db_lock_(NULL),
126 : shutting_down_(NULL),
127 : bg_cv_(&mutex_),
128 244 : mem_(new MemTable(internal_comparator_)),
129 : imm_(NULL),
130 : logfile_(NULL),
131 : logfile_number_(0),
132 : log_(NULL),
133 : seed_(0),
134 244 : tmp_batch_(new WriteBatch),
135 : bg_compaction_scheduled_(false),
136 5124 : manual_compaction_(NULL) {
137 244 : mem_->Ref();
138 244 : has_imm_.Release_Store(NULL);
139 :
140 : // Reserve ten files or so for other uses and give the rest to TableCache.
141 244 : const int table_cache_size = options_.max_open_files - kNumNonTableCacheFiles;
142 244 : table_cache_ = new TableCache(dbname_, &options_, table_cache_size);
143 :
144 : versions_ = new VersionSet(dbname_, &options_, table_cache_,
145 244 : &internal_comparator_);
146 244 : }
147 :
148 2440 : DBImpl::~DBImpl() {
149 : // Wait for background work to finish
150 244 : mutex_.Lock();
151 244 : shutting_down_.Release_Store(this); // Any non-NULL value is ok
152 244 : while (bg_compaction_scheduled_) {
153 0 : bg_cv_.Wait();
154 : }
155 244 : mutex_.Unlock();
156 :
157 244 : if (db_lock_ != NULL) {
158 488 : env_->UnlockFile(db_lock_);
159 : }
160 :
161 244 : delete versions_;
162 244 : if (mem_ != NULL) mem_->Unref();
163 244 : if (imm_ != NULL) imm_->Unref();
164 244 : delete tmp_batch_;
165 244 : delete log_;
166 244 : delete logfile_;
167 244 : delete table_cache_;
168 :
169 244 : if (owns_info_log_) {
170 244 : delete options_.info_log;
171 : }
172 244 : if (owns_cache_) {
173 0 : delete options_.block_cache;
174 : }
175 488 : }
176 :
177 129 : Status DBImpl::NewDB() {
178 129 : VersionEdit new_db;
179 258 : new_db.SetComparatorName(user_comparator()->Name());
180 129 : new_db.SetLogNumber(0);
181 129 : new_db.SetNextFile(2);
182 129 : new_db.SetLastSequence(0);
183 :
184 129 : const std::string manifest = DescriptorFileName(dbname_, 1);
185 : WritableFile* file;
186 129 : Status s = env_->NewWritableFile(manifest, &file);
187 129 : if (!s.ok()) {
188 : return s;
189 : }
190 : {
191 129 : log::Writer log(file);
192 : std::string record;
193 129 : new_db.EncodeTo(&record);
194 258 : s = log.AddRecord(record);
195 129 : if (s.ok()) {
196 258 : s = file->Close();
197 129 : }
198 : }
199 129 : delete file;
200 129 : if (s.ok()) {
201 : // Make "CURRENT" file that points to the new manifest file.
202 258 : s = SetCurrentFile(env_, dbname_, 1);
203 : } else {
204 0 : env_->DeleteFile(manifest);
205 : }
206 129 : return s;
207 : }
208 :
209 283 : void DBImpl::MaybeIgnoreError(Status* s) const {
210 283 : if (s->ok() || options_.paranoid_checks) {
211 : // No change needed
212 : } else {
213 0 : Log(options_.info_log, "Ignoring error %s", s->ToString().c_str());
214 0 : *s = Status::OK();
215 : }
216 283 : }
217 :
218 251 : void DBImpl::DeleteObsoleteFiles() {
219 502 : if (!bg_error_.ok()) {
220 : // After a background error, we don't know whether a new version may
221 : // or may not have been committed, so we cannot safely garbage collect.
222 0 : return;
223 : }
224 :
225 : // Make a set of all of the live files
226 251 : std::set<uint64_t> live = pending_outputs_;
227 251 : versions_->AddLiveFiles(&live);
228 :
229 251 : std::vector<std::string> filenames;
230 502 : env_->GetChildren(dbname_, &filenames); // Ignoring errors on purpose
231 : uint64_t number;
232 : FileType type;
233 4918 : for (size_t i = 0; i < filenames.size(); i++) {
234 4416 : if (ParseFileName(filenames[i], &number, &type)) {
235 1810 : bool keep = true;
236 1810 : switch (type) {
237 : case kLogFile:
238 481 : keep = ((number >= versions_->LogNumber()) ||
239 481 : (number == versions_->PrevLogNumber()));
240 366 : break;
241 : case kDescriptorFile:
242 : // Keep my manifest file, and any newer incarnations'
243 : // (in case there is a race that allows other incarnations)
244 495 : keep = (number >= versions_->ManifestFileNumber());
245 495 : break;
246 : case kTableFile:
247 534 : keep = (live.find(number) != live.end());
248 178 : break;
249 : case kTempFile:
250 : // Any temp files that are currently being written to must
251 : // be recorded in pending_outputs_, which is inserted into "live"
252 0 : keep = (live.find(number) != live.end());
253 0 : break;
254 : case kCurrentFile:
255 : case kDBLockFile:
256 : case kInfoLogFile:
257 : keep = true;
258 : break;
259 : }
260 :
261 1810 : if (!keep) {
262 374 : if (type == kTableFile) {
263 15 : table_cache_->Evict(number);
264 : }
265 : Log(options_.info_log, "Delete type=%d #%lld\n",
266 : int(type),
267 374 : static_cast<unsigned long long>(number));
268 2244 : env_->DeleteFile(dbname_ + "/" + filenames[i]);
269 : }
270 : }
271 : }
272 : }
273 :
274 244 : Status DBImpl::Recover(VersionEdit* edit) {
275 244 : mutex_.AssertHeld();
276 :
277 : // Ignore error from CreateDir since the creation of the DB is
278 : // committed only when the descriptor is created, and this directory
279 : // may already exist from a previous failed creation attempt.
280 488 : env_->CreateDir(dbname_);
281 244 : assert(db_lock_ == NULL);
282 488 : Status s = env_->LockFile(LockFileName(dbname_), &db_lock_);
283 244 : if (!s.ok()) {
284 : return s;
285 : }
286 :
287 488 : if (!env_->FileExists(CurrentFileName(dbname_))) {
288 129 : if (options_.create_if_missing) {
289 258 : s = NewDB();
290 129 : if (!s.ok()) {
291 : return s;
292 : }
293 : } else {
294 : return Status::InvalidArgument(
295 0 : dbname_, "does not exist (create_if_missing is false)");
296 : }
297 : } else {
298 115 : if (options_.error_if_exists) {
299 : return Status::InvalidArgument(
300 0 : dbname_, "exists (error_if_exists is true)");
301 : }
302 : }
303 :
304 488 : s = versions_->Recover();
305 244 : if (s.ok()) {
306 244 : SequenceNumber max_sequence(0);
307 :
308 : // Recover from all newer log files than the ones named in the
309 : // descriptor (new log files may have been added by the previous
310 : // incarnation without registering them in the descriptor).
311 : //
312 : // Note that PrevLogNumber() is no longer used, but we pay
313 : // attention to it in case we are recovering a database
314 : // produced by an older version of leveldb.
315 244 : const uint64_t min_log = versions_->LogNumber();
316 244 : const uint64_t prev_log = versions_->PrevLogNumber();
317 : std::vector<std::string> filenames;
318 488 : s = env_->GetChildren(dbname_, &filenames);
319 244 : if (!s.ok()) {
320 0 : return s;
321 : }
322 : std::set<uint64_t> expected;
323 244 : versions_->AddLiveFiles(&expected);
324 : uint64_t number;
325 : FileType type;
326 : std::vector<uint64_t> logs;
327 3536 : for (size_t i = 0; i < filenames.size(); i++) {
328 3048 : if (ParseFileName(filenames[i], &number, &type)) {
329 : expected.erase(number);
330 1140 : if (type == kLogFile && ((number >= min_log) || (number == prev_log)))
331 115 : logs.push_back(number);
332 : }
333 : }
334 244 : if (!expected.empty()) {
335 : char buf[50];
336 : snprintf(buf, sizeof(buf), "%d missing files; e.g.",
337 0 : static_cast<int>(expected.size()));
338 0 : return Status::Corruption(buf, TableFileName(dbname_, *(expected.begin())));
339 : }
340 :
341 : // Recover in the order in which the logs were generated
342 : std::sort(logs.begin(), logs.end());
343 718 : for (size_t i = 0; i < logs.size(); i++) {
344 345 : s = RecoverLogFile(logs[i], edit, &max_sequence);
345 :
346 : // The previous incarnation may not have written any MANIFEST
347 : // records after allocating this log number. So we manually
348 : // update the file number allocation counter in VersionSet.
349 230 : versions_->MarkFileNumberUsed(logs[i]);
350 : }
351 :
352 244 : if (s.ok()) {
353 244 : if (versions_->LastSequence() < max_sequence) {
354 115 : versions_->SetLastSequence(max_sequence);
355 : }
356 244 : }
357 : }
358 :
359 : return s;
360 : }
361 :
362 115 : Status DBImpl::RecoverLogFile(uint64_t log_number,
363 : VersionEdit* edit,
364 : SequenceNumber* max_sequence) {
365 345 : struct LogReporter : public log::Reader::Reporter {
366 : Env* env;
367 : Logger* info_log;
368 : const char* fname;
369 : Status* status; // NULL if options_.paranoid_checks==false
370 0 : virtual void Corruption(size_t bytes, const Status& s) {
371 : Log(info_log, "%s%s: dropping %d bytes; %s",
372 0 : (this->status == NULL ? "(ignoring error) " : ""),
373 0 : fname, static_cast<int>(bytes), s.ToString().c_str());
374 0 : if (this->status != NULL && this->status->ok()) *this->status = s;
375 0 : }
376 : };
377 :
378 115 : mutex_.AssertHeld();
379 :
380 : // Open the log file
381 115 : std::string fname = LogFileName(dbname_, log_number);
382 : SequentialFile* file;
383 115 : Status status = env_->NewSequentialFile(fname, &file);
384 115 : if (!status.ok()) {
385 0 : MaybeIgnoreError(&status);
386 : return status;
387 : }
388 :
389 : // Create the log reader.
390 : LogReporter reporter;
391 115 : reporter.env = env_;
392 115 : reporter.info_log = options_.info_log;
393 115 : reporter.fname = fname.c_str();
394 115 : reporter.status = (options_.paranoid_checks ? &status : NULL);
395 : // We intentionally make log::Reader do checksumming even if
396 : // paranoid_checks==false so that corruptions cause entire commits
397 : // to be skipped instead of propagating bad information (like overly
398 : // large sequence numbers).
399 : log::Reader reader(file, &reporter, true/*checksum*/,
400 230 : 0/*initial_offset*/);
401 : Log(options_.info_log, "Recovering log #%llu",
402 115 : (unsigned long long) log_number);
403 :
404 : // Read all the records and add to a memtable
405 : std::string scratch;
406 : Slice record;
407 230 : WriteBatch batch;
408 : MemTable* mem = NULL;
409 681 : while (reader.ReadRecord(&record, &scratch) &&
410 : status.ok()) {
411 283 : if (record.size() < 12) {
412 : reporter.Corruption(
413 0 : record.size(), Status::Corruption("log record too small"));
414 0 : continue;
415 : }
416 283 : WriteBatchInternal::SetContents(&batch, record);
417 :
418 283 : if (mem == NULL) {
419 115 : mem = new MemTable(internal_comparator_);
420 115 : mem->Ref();
421 : }
422 566 : status = WriteBatchInternal::InsertInto(&batch, mem);
423 283 : MaybeIgnoreError(&status);
424 283 : if (!status.ok()) {
425 : break;
426 : }
427 : const SequenceNumber last_seq =
428 566 : WriteBatchInternal::Sequence(&batch) +
429 566 : WriteBatchInternal::Count(&batch) - 1;
430 283 : if (last_seq > *max_sequence) {
431 283 : *max_sequence = last_seq;
432 : }
433 :
434 283 : if (mem->ApproximateMemoryUsage() > options_.write_buffer_size) {
435 0 : status = WriteLevel0Table(mem, edit, NULL);
436 0 : if (!status.ok()) {
437 : // Reflect errors immediately so that conditions like full
438 : // file-systems cause the DB::Open() to fail.
439 : break;
440 : }
441 0 : mem->Unref();
442 0 : mem = NULL;
443 : }
444 : }
445 :
446 115 : if (status.ok() && mem != NULL) {
447 230 : status = WriteLevel0Table(mem, edit, NULL);
448 : // Reflect errors immediately so that conditions like full
449 : // file-systems cause the DB::Open() to fail.
450 : }
451 :
452 115 : if (mem != NULL) mem->Unref();
453 115 : delete file;
454 : return status;
455 : }
456 :
457 115 : Status DBImpl::WriteLevel0Table(MemTable* mem, VersionEdit* edit,
458 : Version* base) {
459 115 : mutex_.AssertHeld();
460 115 : const uint64_t start_micros = env_->NowMicros();
461 : FileMetaData meta;
462 230 : meta.number = versions_->NewFileNumber();
463 115 : pending_outputs_.insert(meta.number);
464 115 : Iterator* iter = mem->NewIterator();
465 : Log(options_.info_log, "Level-0 table #%llu: started",
466 115 : (unsigned long long) meta.number);
467 :
468 : Status s;
469 : {
470 115 : mutex_.Unlock();
471 230 : s = BuildTable(dbname_, env_, options_, table_cache_, iter, &meta);
472 115 : mutex_.Lock();
473 : }
474 :
475 : Log(options_.info_log, "Level-0 table #%llu: %lld bytes %s",
476 : (unsigned long long) meta.number,
477 : (unsigned long long) meta.file_size,
478 345 : s.ToString().c_str());
479 115 : delete iter;
480 115 : pending_outputs_.erase(meta.number);
481 :
482 :
483 : // Note that if file_size is zero, the file has been deleted and
484 : // should not be added to the manifest.
485 115 : int level = 0;
486 115 : if (s.ok() && meta.file_size > 0) {
487 115 : const Slice min_user_key = meta.smallest.user_key();
488 115 : const Slice max_user_key = meta.largest.user_key();
489 115 : if (base != NULL) {
490 0 : level = base->PickLevelForMemTableOutput(min_user_key, max_user_key);
491 : }
492 : edit->AddFile(level, meta.number, meta.file_size,
493 115 : meta.smallest, meta.largest);
494 : }
495 :
496 : CompactionStats stats;
497 115 : stats.micros = env_->NowMicros() - start_micros;
498 115 : stats.bytes_written = meta.file_size;
499 115 : stats_[level].Add(stats);
500 115 : return s;
501 : }
502 :
503 0 : void DBImpl::CompactMemTable() {
504 0 : mutex_.AssertHeld();
505 0 : assert(imm_ != NULL);
506 :
507 : // Save the contents of the memtable as a new Table
508 0 : VersionEdit edit;
509 0 : Version* base = versions_->current();
510 0 : base->Ref();
511 0 : Status s = WriteLevel0Table(imm_, &edit, base);
512 0 : base->Unref();
513 :
514 0 : if (s.ok() && shutting_down_.Acquire_Load()) {
515 0 : s = Status::IOError("Deleting DB during memtable compaction");
516 : }
517 :
518 : // Replace immutable memtable with the generated Table
519 0 : if (s.ok()) {
520 0 : edit.SetPrevLogNumber(0);
521 0 : edit.SetLogNumber(logfile_number_); // Earlier logs no longer needed
522 0 : s = versions_->LogAndApply(&edit, &mutex_);
523 : }
524 :
525 0 : if (s.ok()) {
526 : // Commit to the new state
527 0 : imm_->Unref();
528 0 : imm_ = NULL;
529 0 : has_imm_.Release_Store(NULL);
530 0 : DeleteObsoleteFiles();
531 : } else {
532 0 : RecordBackgroundError(s);
533 0 : }
534 0 : }
535 :
536 0 : void DBImpl::CompactRange(const Slice* begin, const Slice* end) {
537 0 : int max_level_with_files = 1;
538 : {
539 0 : MutexLock l(&mutex_);
540 0 : Version* base = versions_->current();
541 0 : for (int level = 1; level < config::kNumLevels; level++) {
542 0 : if (base->OverlapInLevel(level, begin, end)) {
543 0 : max_level_with_files = level;
544 : }
545 : }
546 : }
547 0 : TEST_CompactMemTable(); // TODO(sanjay): Skip if memtable does not overlap
548 0 : for (int level = 0; level < max_level_with_files; level++) {
549 0 : TEST_CompactRange(level, begin, end);
550 : }
551 0 : }
552 :
553 0 : void DBImpl::TEST_CompactRange(int level, const Slice* begin,const Slice* end) {
554 0 : assert(level >= 0);
555 0 : assert(level + 1 < config::kNumLevels);
556 :
557 : InternalKey begin_storage, end_storage;
558 :
559 : ManualCompaction manual;
560 0 : manual.level = level;
561 0 : manual.done = false;
562 0 : if (begin == NULL) {
563 0 : manual.begin = NULL;
564 : } else {
565 0 : begin_storage = InternalKey(*begin, kMaxSequenceNumber, kValueTypeForSeek);
566 0 : manual.begin = &begin_storage;
567 : }
568 0 : if (end == NULL) {
569 0 : manual.end = NULL;
570 : } else {
571 0 : end_storage = InternalKey(*end, 0, static_cast<ValueType>(0));
572 0 : manual.end = &end_storage;
573 : }
574 :
575 0 : MutexLock l(&mutex_);
576 0 : while (!manual.done && !shutting_down_.Acquire_Load() && bg_error_.ok()) {
577 0 : if (manual_compaction_ == NULL) { // Idle
578 0 : manual_compaction_ = &manual;
579 0 : MaybeScheduleCompaction();
580 : } else { // Running either my compaction or another compaction.
581 0 : bg_cv_.Wait();
582 : }
583 : }
584 0 : if (manual_compaction_ == &manual) {
585 : // Cancel my manual compaction since we aborted early for some reason.
586 0 : manual_compaction_ = NULL;
587 : }
588 0 : }
589 :
590 0 : Status DBImpl::TEST_CompactMemTable() {
591 : // NULL batch means just wait for earlier writes to be done
592 0 : Status s = Write(WriteOptions(), NULL);
593 0 : if (s.ok()) {
594 : // Wait until the compaction completes
595 0 : MutexLock l(&mutex_);
596 0 : while (imm_ != NULL && bg_error_.ok()) {
597 0 : bg_cv_.Wait();
598 : }
599 0 : if (imm_ != NULL) {
600 0 : s = bg_error_;
601 : }
602 : }
603 0 : return s;
604 : }
605 :
606 0 : void DBImpl::RecordBackgroundError(const Status& s) {
607 0 : mutex_.AssertHeld();
608 0 : if (bg_error_.ok()) {
609 0 : bg_error_ = s;
610 0 : bg_cv_.SignalAll();
611 : }
612 0 : }
613 :
614 258 : void DBImpl::MaybeScheduleCompaction() {
615 258 : mutex_.AssertHeld();
616 258 : if (bg_compaction_scheduled_) {
617 : // Already scheduled
618 516 : } else if (shutting_down_.Acquire_Load()) {
619 : // DB is being deleted; no more background compactions
620 516 : } else if (!bg_error_.ok()) {
621 : // Already got an error; no more changes
622 774 : } else if (imm_ == NULL &&
623 516 : manual_compaction_ == NULL &&
624 258 : !versions_->NeedsCompaction()) {
625 : // No work to be done
626 : } else {
627 7 : bg_compaction_scheduled_ = true;
628 7 : env_->Schedule(&DBImpl::BGWork, this);
629 : }
630 258 : }
631 :
632 7 : void DBImpl::BGWork(void* db) {
633 7 : reinterpret_cast<DBImpl*>(db)->BackgroundCall();
634 7 : }
635 :
636 7 : void DBImpl::BackgroundCall() {
637 7 : MutexLock l(&mutex_);
638 7 : assert(bg_compaction_scheduled_);
639 14 : if (shutting_down_.Acquire_Load()) {
640 : // No more background work when shutting down.
641 14 : } else if (!bg_error_.ok()) {
642 : // No more background work after a background error.
643 : } else {
644 7 : BackgroundCompaction();
645 : }
646 :
647 7 : bg_compaction_scheduled_ = false;
648 :
649 : // Previous compaction may have produced too many files in a level,
650 : // so reschedule another compaction if needed.
651 7 : MaybeScheduleCompaction();
652 7 : bg_cv_.SignalAll();
653 7 : }
654 :
655 7 : void DBImpl::BackgroundCompaction() {
656 7 : mutex_.AssertHeld();
657 :
658 7 : if (imm_ != NULL) {
659 0 : CompactMemTable();
660 7 : return;
661 : }
662 :
663 : Compaction* c;
664 7 : bool is_manual = (manual_compaction_ != NULL);
665 : InternalKey manual_end;
666 7 : if (is_manual) {
667 0 : ManualCompaction* m = manual_compaction_;
668 0 : c = versions_->CompactRange(m->level, m->begin, m->end);
669 0 : m->done = (c == NULL);
670 0 : if (c != NULL) {
671 0 : manual_end = c->input(0, c->num_input_files(0) - 1)->largest;
672 : }
673 : Log(options_.info_log,
674 : "Manual compaction at level-%d from %s .. %s; will stop at %s\n",
675 : m->level,
676 0 : (m->begin ? m->begin->DebugString().c_str() : "(begin)"),
677 0 : (m->end ? m->end->DebugString().c_str() : "(end)"),
678 0 : (m->done ? "(end)" : manual_end.DebugString().c_str()));
679 : } else {
680 7 : c = versions_->PickCompaction();
681 : }
682 :
683 : Status status;
684 7 : if (c == NULL) {
685 : // Nothing to do
686 7 : } else if (!is_manual && c->IsTrivialMove()) {
687 : // Move file to next level
688 0 : assert(c->num_input_files(0) == 1);
689 0 : FileMetaData* f = c->input(0, 0);
690 0 : c->edit()->DeleteFile(c->level(), f->number);
691 0 : c->edit()->AddFile(c->level() + 1, f->number, f->file_size,
692 0 : f->smallest, f->largest);
693 0 : status = versions_->LogAndApply(c->edit(), &mutex_);
694 0 : if (!status.ok()) {
695 0 : RecordBackgroundError(status);
696 : }
697 : VersionSet::LevelSummaryStorage tmp;
698 : Log(options_.info_log, "Moved #%lld to level-%d %lld bytes %s: %s\n",
699 : static_cast<unsigned long long>(f->number),
700 0 : c->level() + 1,
701 : static_cast<unsigned long long>(f->file_size),
702 : status.ToString().c_str(),
703 0 : versions_->LevelSummary(&tmp));
704 : } else {
705 14 : CompactionState* compact = new CompactionState(c);
706 14 : status = DoCompactionWork(compact);
707 7 : if (!status.ok()) {
708 0 : RecordBackgroundError(status);
709 : }
710 7 : CleanupCompaction(compact);
711 7 : c->ReleaseInputs();
712 7 : DeleteObsoleteFiles();
713 : }
714 7 : delete c;
715 :
716 7 : if (status.ok()) {
717 : // Done
718 0 : } else if (shutting_down_.Acquire_Load()) {
719 : // Ignore compaction errors found during shutting down
720 : } else {
721 : Log(options_.info_log,
722 0 : "Compaction error: %s", status.ToString().c_str());
723 : }
724 :
725 7 : if (is_manual) {
726 0 : ManualCompaction* m = manual_compaction_;
727 0 : if (!status.ok()) {
728 0 : m->done = true;
729 : }
730 0 : if (!m->done) {
731 : // We only compacted part of the requested range. Update *m
732 : // to the range that is left to be compacted.
733 0 : m->tmp_storage = manual_end;
734 0 : m->begin = &m->tmp_storage;
735 : }
736 0 : manual_compaction_ = NULL;
737 : }
738 : }
739 :
740 7 : void DBImpl::CleanupCompaction(CompactionState* compact) {
741 7 : mutex_.AssertHeld();
742 7 : if (compact->builder != NULL) {
743 : // May happen if we get a shutdown call in the middle of compaction
744 0 : compact->builder->Abandon();
745 0 : delete compact->builder;
746 : } else {
747 7 : assert(compact->outfile == NULL);
748 : }
749 7 : delete compact->outfile;
750 21 : for (size_t i = 0; i < compact->outputs.size(); i++) {
751 14 : const CompactionState::Output& out = compact->outputs[i];
752 7 : pending_outputs_.erase(out.number);
753 : }
754 14 : delete compact;
755 7 : }
756 :
757 7 : Status DBImpl::OpenCompactionOutputFile(CompactionState* compact) {
758 7 : assert(compact != NULL);
759 7 : assert(compact->builder == NULL);
760 : uint64_t file_number;
761 : {
762 7 : mutex_.Lock();
763 14 : file_number = versions_->NewFileNumber();
764 7 : pending_outputs_.insert(file_number);
765 : CompactionState::Output out;
766 7 : out.number = file_number;
767 : out.smallest.Clear();
768 : out.largest.Clear();
769 7 : compact->outputs.push_back(out);
770 7 : mutex_.Unlock();
771 : }
772 :
773 : // Make the output file
774 7 : std::string fname = TableFileName(dbname_, file_number);
775 7 : Status s = env_->NewWritableFile(fname, &compact->outfile);
776 7 : if (s.ok()) {
777 7 : compact->builder = new TableBuilder(options_, compact->outfile);
778 : }
779 7 : return s;
780 : }
781 :
782 7 : Status DBImpl::FinishCompactionOutputFile(CompactionState* compact,
783 : Iterator* input) {
784 7 : assert(compact != NULL);
785 7 : assert(compact->outfile != NULL);
786 7 : assert(compact->builder != NULL);
787 :
788 7 : const uint64_t output_number = compact->current_output()->number;
789 7 : assert(output_number != 0);
790 :
791 : // Check for iterator errors
792 7 : Status s = input->status();
793 7 : const uint64_t current_entries = compact->builder->NumEntries();
794 7 : if (s.ok()) {
795 14 : s = compact->builder->Finish();
796 : } else {
797 0 : compact->builder->Abandon();
798 : }
799 7 : const uint64_t current_bytes = compact->builder->FileSize();
800 7 : compact->current_output()->file_size = current_bytes;
801 7 : compact->total_bytes += current_bytes;
802 7 : delete compact->builder;
803 7 : compact->builder = NULL;
804 :
805 : // Finish and check for file errors
806 7 : if (s.ok()) {
807 14 : s = compact->outfile->Sync();
808 : }
809 7 : if (s.ok()) {
810 14 : s = compact->outfile->Close();
811 : }
812 7 : delete compact->outfile;
813 7 : compact->outfile = NULL;
814 :
815 7 : if (s.ok() && current_entries > 0) {
816 : // Verify that the table is usable
817 : Iterator* iter = table_cache_->NewIterator(ReadOptions(),
818 : output_number,
819 7 : current_bytes);
820 14 : s = iter->status();
821 7 : delete iter;
822 7 : if (s.ok()) {
823 : Log(options_.info_log,
824 : "Generated table #%llu: %lld keys, %lld bytes",
825 : (unsigned long long) output_number,
826 : (unsigned long long) current_entries,
827 7 : (unsigned long long) current_bytes);
828 : }
829 : }
830 7 : return s;
831 : }
832 :
833 :
834 7 : Status DBImpl::InstallCompactionResults(CompactionState* compact) {
835 7 : mutex_.AssertHeld();
836 : Log(options_.info_log, "Compacted %d@%d + %d@%d files => %lld bytes",
837 : compact->compaction->num_input_files(0),
838 : compact->compaction->level(),
839 : compact->compaction->num_input_files(1),
840 7 : compact->compaction->level() + 1,
841 28 : static_cast<long long>(compact->total_bytes));
842 :
843 : // Add compaction outputs
844 14 : compact->compaction->AddInputDeletions(compact->compaction->edit());
845 7 : const int level = compact->compaction->level();
846 28 : for (size_t i = 0; i < compact->outputs.size(); i++) {
847 14 : const CompactionState::Output& out = compact->outputs[i];
848 : compact->compaction->edit()->AddFile(
849 : level + 1,
850 14 : out.number, out.file_size, out.smallest, out.largest);
851 : }
852 14 : return versions_->LogAndApply(compact->compaction->edit(), &mutex_);
853 : }
854 :
855 7 : Status DBImpl::DoCompactionWork(CompactionState* compact) {
856 7 : const uint64_t start_micros = env_->NowMicros();
857 7 : int64_t imm_micros = 0; // Micros spent doing imm_ compactions
858 :
859 : Log(options_.info_log, "Compacting %d@%d + %d@%d files",
860 : compact->compaction->num_input_files(0),
861 : compact->compaction->level(),
862 : compact->compaction->num_input_files(1),
863 21 : compact->compaction->level() + 1);
864 :
865 7 : assert(versions_->NumLevelFiles(compact->compaction->level()) > 0);
866 7 : assert(compact->builder == NULL);
867 7 : assert(compact->outfile == NULL);
868 14 : if (snapshots_.empty()) {
869 7 : compact->smallest_snapshot = versions_->LastSequence();
870 : } else {
871 0 : compact->smallest_snapshot = snapshots_.oldest()->number_;
872 : }
873 :
874 : // Release mutex while we're actually doing the compaction work
875 7 : mutex_.Unlock();
876 :
877 7 : Iterator* input = versions_->MakeInputIterator(compact->compaction);
878 7 : input->SeekToFirst();
879 : Status status;
880 : ParsedInternalKey ikey;
881 : std::string current_user_key;
882 7 : bool has_current_user_key = false;
883 7 : SequenceNumber last_sequence_for_key = kMaxSequenceNumber;
884 3094 : for (; input->Valid() && !shutting_down_.Acquire_Load(); ) {
885 : // Prioritize immutable compaction work
886 3080 : if (has_imm_.NoBarrier_Load() != NULL) {
887 0 : const uint64_t imm_start = env_->NowMicros();
888 0 : mutex_.Lock();
889 0 : if (imm_ != NULL) {
890 0 : CompactMemTable();
891 0 : bg_cv_.SignalAll(); // Wakeup MakeRoomForWrite() if necessary
892 : }
893 0 : mutex_.Unlock();
894 0 : imm_micros += (env_->NowMicros() - imm_start);
895 : }
896 :
897 1540 : Slice key = input->key();
898 1540 : if (compact->compaction->ShouldStopBefore(key) &&
899 0 : compact->builder != NULL) {
900 0 : status = FinishCompactionOutputFile(compact, input);
901 0 : if (!status.ok()) {
902 : break;
903 : }
904 : }
905 :
906 : // Handle key/value, add to state, etc.
907 1540 : bool drop = false;
908 1540 : if (!ParseInternalKey(key, &ikey)) {
909 : // Do not hide error keys
910 : current_user_key.clear();
911 : has_current_user_key = false;
912 : last_sequence_for_key = kMaxSequenceNumber;
913 : } else {
914 4620 : if (!has_current_user_key ||
915 : user_comparator()->Compare(ikey.user_key,
916 6139 : Slice(current_user_key)) != 0) {
917 : // First occurrence of this user key
918 1519 : current_user_key.assign(ikey.user_key.data(), ikey.user_key.size());
919 : has_current_user_key = true;
920 : last_sequence_for_key = kMaxSequenceNumber;
921 : }
922 :
923 1540 : if (last_sequence_for_key <= compact->smallest_snapshot) {
924 : // Hidden by an newer entry for same user key
925 : drop = true; // (A)
926 3038 : } else if (ikey.type == kTypeDeletion &&
927 1519 : ikey.sequence <= compact->smallest_snapshot &&
928 0 : compact->compaction->IsBaseLevelForKey(ikey.user_key)) {
929 : // For this user key:
930 : // (1) there is no data in higher levels
931 : // (2) data in lower levels will have larger sequence numbers
932 : // (3) data in layers that are being compacted here and have
933 : // smaller sequence numbers will be dropped in the next
934 : // few iterations of this loop (by rule (A) above).
935 : // Therefore this deletion marker is obsolete and can be dropped.
936 0 : drop = true;
937 : }
938 :
939 1540 : last_sequence_for_key = ikey.sequence;
940 : }
941 : #if 0
942 : Log(options_.info_log,
943 : " Compact: %s, seq %d, type: %d %d, drop: %d, is_base: %d, "
944 : "%d smallest_snapshot: %d",
945 : ikey.user_key.ToString().c_str(),
946 : (int)ikey.sequence, ikey.type, kTypeValue, drop,
947 : compact->compaction->IsBaseLevelForKey(ikey.user_key),
948 : (int)last_sequence_for_key, (int)compact->smallest_snapshot);
949 : #endif
950 :
951 1540 : if (!drop) {
952 : // Open output file if necessary
953 1519 : if (compact->builder == NULL) {
954 14 : status = OpenCompactionOutputFile(compact);
955 7 : if (!status.ok()) {
956 : break;
957 : }
958 : }
959 1519 : if (compact->builder->NumEntries() == 0) {
960 7 : compact->current_output()->smallest.DecodeFrom(key);
961 : }
962 1519 : compact->current_output()->largest.DecodeFrom(key);
963 1519 : compact->builder->Add(key, input->value());
964 :
965 : // Close output file if it is big enough
966 3038 : if (compact->builder->FileSize() >=
967 1519 : compact->compaction->MaxOutputFileSize()) {
968 0 : status = FinishCompactionOutputFile(compact, input);
969 0 : if (!status.ok()) {
970 : break;
971 : }
972 : }
973 : }
974 :
975 1540 : input->Next();
976 : }
977 :
978 14 : if (status.ok() && shutting_down_.Acquire_Load()) {
979 0 : status = Status::IOError("Deleting DB during compaction");
980 : }
981 7 : if (status.ok() && compact->builder != NULL) {
982 14 : status = FinishCompactionOutputFile(compact, input);
983 : }
984 7 : if (status.ok()) {
985 14 : status = input->status();
986 : }
987 7 : delete input;
988 7 : input = NULL;
989 :
990 : CompactionStats stats;
991 7 : stats.micros = env_->NowMicros() - start_micros - imm_micros;
992 21 : for (int which = 0; which < 2; which++) {
993 50 : for (int i = 0; i < compact->compaction->num_input_files(which); i++) {
994 36 : stats.bytes_read += compact->compaction->input(which, i)->file_size;
995 : }
996 : }
997 21 : for (size_t i = 0; i < compact->outputs.size(); i++) {
998 14 : stats.bytes_written += compact->outputs[i].file_size;
999 : }
1000 :
1001 7 : mutex_.Lock();
1002 7 : stats_[compact->compaction->level() + 1].Add(stats);
1003 :
1004 7 : if (status.ok()) {
1005 14 : status = InstallCompactionResults(compact);
1006 : }
1007 7 : if (!status.ok()) {
1008 0 : RecordBackgroundError(status);
1009 : }
1010 : VersionSet::LevelSummaryStorage tmp;
1011 : Log(options_.info_log,
1012 7 : "compacted to: %s", versions_->LevelSummary(&tmp));
1013 7 : return status;
1014 : }
1015 :
1016 : namespace {
1017 : struct IterState {
1018 : port::Mutex* mu;
1019 : Version* version;
1020 : MemTable* mem;
1021 : MemTable* imm;
1022 : };
1023 :
1024 159 : static void CleanupIteratorState(void* arg1, void* arg2) {
1025 159 : IterState* state = reinterpret_cast<IterState*>(arg1);
1026 159 : state->mu->Lock();
1027 159 : state->mem->Unref();
1028 159 : if (state->imm != NULL) state->imm->Unref();
1029 159 : state->version->Unref();
1030 159 : state->mu->Unlock();
1031 159 : delete state;
1032 159 : }
1033 : } // namespace
1034 :
1035 159 : Iterator* DBImpl::NewInternalIterator(const ReadOptions& options,
1036 : SequenceNumber* latest_snapshot,
1037 : uint32_t* seed) {
1038 159 : IterState* cleanup = new IterState;
1039 159 : mutex_.Lock();
1040 159 : *latest_snapshot = versions_->LastSequence();
1041 :
1042 : // Collect together all needed child iterators
1043 : std::vector<Iterator*> list;
1044 318 : list.push_back(mem_->NewIterator());
1045 159 : mem_->Ref();
1046 159 : if (imm_ != NULL) {
1047 0 : list.push_back(imm_->NewIterator());
1048 0 : imm_->Ref();
1049 : }
1050 159 : versions_->current()->AddIterators(options, &list);
1051 : Iterator* internal_iter =
1052 318 : NewMergingIterator(&internal_comparator_, &list[0], list.size());
1053 159 : versions_->current()->Ref();
1054 :
1055 159 : cleanup->mu = &mutex_;
1056 159 : cleanup->mem = mem_;
1057 159 : cleanup->imm = imm_;
1058 159 : cleanup->version = versions_->current();
1059 159 : internal_iter->RegisterCleanup(CleanupIteratorState, cleanup, NULL);
1060 :
1061 159 : *seed = ++seed_;
1062 159 : mutex_.Unlock();
1063 318 : return internal_iter;
1064 : }
1065 :
1066 0 : Iterator* DBImpl::TEST_NewInternalIterator() {
1067 : SequenceNumber ignored;
1068 : uint32_t ignored_seed;
1069 0 : return NewInternalIterator(ReadOptions(), &ignored, &ignored_seed);
1070 : }
1071 :
1072 0 : int64_t DBImpl::TEST_MaxNextLevelOverlappingBytes() {
1073 0 : MutexLock l(&mutex_);
1074 0 : return versions_->MaxNextLevelOverlappingBytes();
1075 : }
1076 :
1077 36658 : Status DBImpl::Get(const ReadOptions& options,
1078 : const Slice& key,
1079 : std::string* value) {
1080 : Status s;
1081 36658 : MutexLock l(&mutex_);
1082 : SequenceNumber snapshot;
1083 36658 : if (options.snapshot != NULL) {
1084 0 : snapshot = reinterpret_cast<const SnapshotImpl*>(options.snapshot)->number_;
1085 : } else {
1086 36658 : snapshot = versions_->LastSequence();
1087 : }
1088 :
1089 36658 : MemTable* mem = mem_;
1090 36658 : MemTable* imm = imm_;
1091 36658 : Version* current = versions_->current();
1092 36658 : mem->Ref();
1093 36658 : if (imm != NULL) imm->Ref();
1094 36658 : current->Ref();
1095 :
1096 36658 : bool have_stat_update = false;
1097 : Version::GetStats stats;
1098 :
1099 : // Unlock while reading from files and memtables
1100 : {
1101 36658 : mutex_.Unlock();
1102 : // First look in the memtable, then in the immutable memtable (if any).
1103 36658 : LookupKey lkey(key, snapshot);
1104 36658 : if (mem->Get(lkey, value, &s)) {
1105 : // Done
1106 36651 : } else if (imm != NULL && imm->Get(lkey, value, &s)) {
1107 : // Done
1108 : } else {
1109 73302 : s = current->Get(options, lkey, value, &stats);
1110 36651 : have_stat_update = true;
1111 : }
1112 36658 : mutex_.Lock();
1113 : }
1114 :
1115 36658 : if (have_stat_update && current->UpdateStats(stats)) {
1116 7 : MaybeScheduleCompaction();
1117 : }
1118 36658 : mem->Unref();
1119 36658 : if (imm != NULL) imm->Unref();
1120 36658 : current->Unref();
1121 36658 : return s;
1122 : }
1123 :
1124 159 : Iterator* DBImpl::NewIterator(const ReadOptions& options) {
1125 : SequenceNumber latest_snapshot;
1126 : uint32_t seed;
1127 159 : Iterator* iter = NewInternalIterator(options, &latest_snapshot, &seed);
1128 : return NewDBIterator(
1129 : this, user_comparator(), iter,
1130 159 : (options.snapshot != NULL
1131 : ? reinterpret_cast<const SnapshotImpl*>(options.snapshot)->number_
1132 : : latest_snapshot),
1133 477 : seed);
1134 : }
1135 :
1136 57 : void DBImpl::RecordReadSample(Slice key) {
1137 57 : MutexLock l(&mutex_);
1138 57 : if (versions_->current()->RecordReadSample(key)) {
1139 0 : MaybeScheduleCompaction();
1140 : }
1141 57 : }
1142 :
1143 0 : const Snapshot* DBImpl::GetSnapshot() {
1144 0 : MutexLock l(&mutex_);
1145 0 : return snapshots_.New(versions_->LastSequence());
1146 : }
1147 :
1148 0 : void DBImpl::ReleaseSnapshot(const Snapshot* s) {
1149 0 : MutexLock l(&mutex_);
1150 0 : snapshots_.Delete(reinterpret_cast<const SnapshotImpl*>(s));
1151 0 : }
1152 :
1153 : // Convenience methods
1154 0 : Status DBImpl::Put(const WriteOptions& o, const Slice& key, const Slice& val) {
1155 0 : return DB::Put(o, key, val);
1156 : }
1157 :
1158 0 : Status DBImpl::Delete(const WriteOptions& options, const Slice& key) {
1159 0 : return DB::Delete(options, key);
1160 : }
1161 :
1162 551 : Status DBImpl::Write(const WriteOptions& options, WriteBatch* my_batch) {
1163 551 : Writer w(&mutex_);
1164 551 : w.batch = my_batch;
1165 551 : w.sync = options.sync;
1166 551 : w.done = false;
1167 :
1168 551 : MutexLock l(&mutex_);
1169 1102 : writers_.push_back(&w);
1170 1653 : while (!w.done && &w != writers_.front()) {
1171 0 : w.cv.Wait();
1172 : }
1173 551 : if (w.done) {
1174 : return w.status;
1175 : }
1176 :
1177 : // May temporarily unlock and wait.
1178 551 : Status status = MakeRoomForWrite(my_batch == NULL);
1179 551 : uint64_t last_sequence = versions_->LastSequence();
1180 551 : Writer* last_writer = &w;
1181 551 : if (status.ok() && my_batch != NULL) { // NULL batch is for compactions
1182 551 : WriteBatch* updates = BuildBatchGroup(&last_writer);
1183 551 : WriteBatchInternal::SetSequence(updates, last_sequence + 1);
1184 551 : last_sequence += WriteBatchInternal::Count(updates);
1185 :
1186 : // Add to log and apply to memtable. We can release the lock
1187 : // during this phase since &w is currently responsible for logging
1188 : // and protects against concurrent loggers and concurrent writes
1189 : // into mem_.
1190 : {
1191 551 : mutex_.Unlock();
1192 1102 : status = log_->AddRecord(WriteBatchInternal::Contents(updates));
1193 551 : bool sync_error = false;
1194 551 : if (status.ok() && options.sync) {
1195 310 : status = logfile_->Sync();
1196 155 : if (!status.ok()) {
1197 0 : sync_error = true;
1198 : }
1199 : }
1200 551 : if (status.ok()) {
1201 1102 : status = WriteBatchInternal::InsertInto(updates, mem_);
1202 : }
1203 551 : mutex_.Lock();
1204 551 : if (sync_error) {
1205 : // The state of the log file is indeterminate: the log record we
1206 : // just added may or may not show up when the DB is re-opened.
1207 : // So we force the DB into a mode where all future writes fail.
1208 0 : RecordBackgroundError(status);
1209 : }
1210 : }
1211 551 : if (updates == tmp_batch_) tmp_batch_->Clear();
1212 :
1213 551 : versions_->SetLastSequence(last_sequence);
1214 : }
1215 :
1216 : while (true) {
1217 1102 : Writer* ready = writers_.front();
1218 551 : writers_.pop_front();
1219 551 : if (ready != &w) {
1220 0 : ready->status = status;
1221 0 : ready->done = true;
1222 0 : ready->cv.Signal();
1223 : }
1224 551 : if (ready == last_writer) break;
1225 : }
1226 :
1227 : // Notify new head of write queue
1228 1102 : if (!writers_.empty()) {
1229 0 : writers_.front()->cv.Signal();
1230 : }
1231 :
1232 551 : return status;
1233 : }
1234 :
1235 : // REQUIRES: Writer list must be non-empty
1236 : // REQUIRES: First writer must have a non-NULL batch
1237 551 : WriteBatch* DBImpl::BuildBatchGroup(Writer** last_writer) {
1238 1102 : assert(!writers_.empty());
1239 1102 : Writer* first = writers_.front();
1240 551 : WriteBatch* result = first->batch;
1241 551 : assert(result != NULL);
1242 :
1243 1102 : size_t size = WriteBatchInternal::ByteSize(first->batch);
1244 :
1245 : // Allow the group to grow up to a maximum size, but if the
1246 : // original write is small, limit the growth so we do not slow
1247 : // down the small write too much.
1248 551 : size_t max_size = 1 << 20;
1249 551 : if (size <= (128<<10)) {
1250 550 : max_size = size + (128<<10);
1251 : }
1252 :
1253 551 : *last_writer = first;
1254 551 : std::deque<Writer*>::iterator iter = writers_.begin();
1255 : ++iter; // Advance past "first"
1256 1102 : for (; iter != writers_.end(); ++iter) {
1257 0 : Writer* w = *iter;
1258 0 : if (w->sync && !first->sync) {
1259 : // Do not include a sync write into a batch handled by a non-sync write.
1260 : break;
1261 : }
1262 :
1263 0 : if (w->batch != NULL) {
1264 0 : size += WriteBatchInternal::ByteSize(w->batch);
1265 0 : if (size > max_size) {
1266 : // Do not make batch too big
1267 : break;
1268 : }
1269 :
1270 : // Append to *result
1271 0 : if (result == first->batch) {
1272 : // Switch to temporary batch instead of disturbing caller's batch
1273 0 : result = tmp_batch_;
1274 0 : assert(WriteBatchInternal::Count(result) == 0);
1275 0 : WriteBatchInternal::Append(result, first->batch);
1276 : }
1277 0 : WriteBatchInternal::Append(result, w->batch);
1278 : }
1279 0 : *last_writer = w;
1280 : }
1281 551 : return result;
1282 : }
1283 :
1284 : // REQUIRES: mutex_ is held
1285 : // REQUIRES: this thread is currently at the front of the writer queue
1286 551 : Status DBImpl::MakeRoomForWrite(bool force) {
1287 551 : mutex_.AssertHeld();
1288 1102 : assert(!writers_.empty());
1289 551 : bool allow_delay = !force;
1290 : Status s;
1291 : while (true) {
1292 1102 : if (!bg_error_.ok()) {
1293 : // Yield previous error
1294 0 : s = bg_error_;
1295 : break;
1296 551 : } else if (
1297 1102 : allow_delay &&
1298 551 : versions_->NumLevelFiles(0) >= config::kL0_SlowdownWritesTrigger) {
1299 : // We are getting close to hitting a hard limit on the number of
1300 : // L0 files. Rather than delaying a single write by several
1301 : // seconds when we hit the hard limit, start delaying each
1302 : // individual write by 1ms to reduce latency variance. Also,
1303 : // this delay hands over some CPU to the compaction thread in
1304 : // case it is sharing the same core as the writer.
1305 0 : mutex_.Unlock();
1306 0 : env_->SleepForMicroseconds(1000);
1307 0 : allow_delay = false; // Do not delay a single write more than once
1308 0 : mutex_.Lock();
1309 1102 : } else if (!force &&
1310 551 : (mem_->ApproximateMemoryUsage() <= options_.write_buffer_size)) {
1311 : // There is room in current memtable
1312 : break;
1313 0 : } else if (imm_ != NULL) {
1314 : // We have filled up the current memtable, but the previous
1315 : // one is still being compacted, so we wait.
1316 0 : Log(options_.info_log, "Current memtable full; waiting...\n");
1317 0 : bg_cv_.Wait();
1318 0 : } else if (versions_->NumLevelFiles(0) >= config::kL0_StopWritesTrigger) {
1319 : // There are too many level-0 files.
1320 0 : Log(options_.info_log, "Too many L0 files; waiting...\n");
1321 0 : bg_cv_.Wait();
1322 : } else {
1323 : // Attempt to switch to a new memtable and trigger compaction of old
1324 0 : assert(versions_->PrevLogNumber() == 0);
1325 0 : uint64_t new_log_number = versions_->NewFileNumber();
1326 0 : WritableFile* lfile = NULL;
1327 0 : s = env_->NewWritableFile(LogFileName(dbname_, new_log_number), &lfile);
1328 0 : if (!s.ok()) {
1329 : // Avoid chewing through file number space in a tight loop.
1330 0 : versions_->ReuseFileNumber(new_log_number);
1331 0 : break;
1332 : }
1333 0 : delete log_;
1334 0 : delete logfile_;
1335 0 : logfile_ = lfile;
1336 0 : logfile_number_ = new_log_number;
1337 0 : log_ = new log::Writer(lfile);
1338 0 : imm_ = mem_;
1339 0 : has_imm_.Release_Store(imm_);
1340 0 : mem_ = new MemTable(internal_comparator_);
1341 0 : mem_->Ref();
1342 0 : force = false; // Do not force another compaction if have room
1343 0 : MaybeScheduleCompaction();
1344 : }
1345 : }
1346 551 : return s;
1347 : }
1348 :
1349 0 : bool DBImpl::GetProperty(const Slice& property, std::string* value) {
1350 : value->clear();
1351 :
1352 0 : MutexLock l(&mutex_);
1353 0 : Slice in = property;
1354 : Slice prefix("leveldb.");
1355 0 : if (!in.starts_with(prefix)) return false;
1356 0 : in.remove_prefix(prefix.size());
1357 :
1358 0 : if (in.starts_with("num-files-at-level")) {
1359 0 : in.remove_prefix(strlen("num-files-at-level"));
1360 : uint64_t level;
1361 0 : bool ok = ConsumeDecimalNumber(&in, &level) && in.empty();
1362 0 : if (!ok || level >= config::kNumLevels) {
1363 : return false;
1364 : } else {
1365 : char buf[100];
1366 : snprintf(buf, sizeof(buf), "%d",
1367 0 : versions_->NumLevelFiles(static_cast<int>(level)));
1368 : *value = buf;
1369 : return true;
1370 : }
1371 0 : } else if (in == "stats") {
1372 : char buf[200];
1373 : snprintf(buf, sizeof(buf),
1374 : " Compactions\n"
1375 : "Level Files Size(MB) Time(sec) Read(MB) Write(MB)\n"
1376 : "--------------------------------------------------\n"
1377 : );
1378 0 : value->append(buf);
1379 0 : for (int level = 0; level < config::kNumLevels; level++) {
1380 0 : int files = versions_->NumLevelFiles(level);
1381 0 : if (stats_[level].micros > 0 || files > 0) {
1382 : snprintf(
1383 : buf, sizeof(buf),
1384 : "%3d %8d %8.0f %9.0f %8.0f %9.0f\n",
1385 : level,
1386 : files,
1387 0 : versions_->NumLevelBytes(level) / 1048576.0,
1388 : stats_[level].micros / 1e6,
1389 : stats_[level].bytes_read / 1048576.0,
1390 0 : stats_[level].bytes_written / 1048576.0);
1391 0 : value->append(buf);
1392 : }
1393 : }
1394 : return true;
1395 0 : } else if (in == "sstables") {
1396 0 : *value = versions_->current()->DebugString();
1397 0 : return true;
1398 : }
1399 :
1400 : return false;
1401 : }
1402 :
1403 0 : void DBImpl::GetApproximateSizes(
1404 : const Range* range, int n,
1405 : uint64_t* sizes) {
1406 : // TODO(opt): better implementation
1407 : Version* v;
1408 : {
1409 0 : MutexLock l(&mutex_);
1410 0 : versions_->current()->Ref();
1411 0 : v = versions_->current();
1412 : }
1413 :
1414 0 : for (int i = 0; i < n; i++) {
1415 : // Convert user_key into a corresponding internal key.
1416 0 : InternalKey k1(range[i].start, kMaxSequenceNumber, kValueTypeForSeek);
1417 0 : InternalKey k2(range[i].limit, kMaxSequenceNumber, kValueTypeForSeek);
1418 0 : uint64_t start = versions_->ApproximateOffsetOf(v, k1);
1419 0 : uint64_t limit = versions_->ApproximateOffsetOf(v, k2);
1420 0 : sizes[i] = (limit >= start ? limit - start : 0);
1421 : }
1422 :
1423 : {
1424 0 : MutexLock l(&mutex_);
1425 0 : v->Unref();
1426 : }
1427 0 : }
1428 :
1429 : // Default implementations of convenience methods that subclasses of DB
1430 : // can call if they wish
1431 0 : Status DB::Put(const WriteOptions& opt, const Slice& key, const Slice& value) {
1432 0 : WriteBatch batch;
1433 0 : batch.Put(key, value);
1434 0 : return Write(opt, &batch);
1435 : }
1436 :
1437 0 : Status DB::Delete(const WriteOptions& opt, const Slice& key) {
1438 0 : WriteBatch batch;
1439 0 : batch.Delete(key);
1440 0 : return Write(opt, &batch);
1441 : }
1442 :
1443 244 : DB::~DB() { }
1444 :
1445 244 : Status DB::Open(const Options& options, const std::string& dbname,
1446 : DB** dbptr) {
1447 244 : *dbptr = NULL;
1448 :
1449 244 : DBImpl* impl = new DBImpl(options, dbname);
1450 244 : impl->mutex_.Lock();
1451 244 : VersionEdit edit;
1452 244 : Status s = impl->Recover(&edit); // Handles create_if_missing, error_if_exists
1453 244 : if (s.ok()) {
1454 488 : uint64_t new_log_number = impl->versions_->NewFileNumber();
1455 : WritableFile* lfile;
1456 732 : s = options.env->NewWritableFile(LogFileName(dbname, new_log_number),
1457 488 : &lfile);
1458 244 : if (s.ok()) {
1459 244 : edit.SetLogNumber(new_log_number);
1460 244 : impl->logfile_ = lfile;
1461 244 : impl->logfile_number_ = new_log_number;
1462 244 : impl->log_ = new log::Writer(lfile);
1463 488 : s = impl->versions_->LogAndApply(&edit, &impl->mutex_);
1464 : }
1465 244 : if (s.ok()) {
1466 244 : impl->DeleteObsoleteFiles();
1467 244 : impl->MaybeScheduleCompaction();
1468 : }
1469 : }
1470 244 : impl->mutex_.Unlock();
1471 244 : if (s.ok()) {
1472 244 : *dbptr = impl;
1473 : } else {
1474 0 : delete impl;
1475 : }
1476 244 : return s;
1477 : }
1478 :
1479 244 : Snapshot::~Snapshot() {
1480 244 : }
1481 :
1482 3 : Status DestroyDB(const std::string& dbname, const Options& options) {
1483 3 : Env* env = options.env;
1484 : std::vector<std::string> filenames;
1485 : // Ignore error in case directory does not exist
1486 6 : env->GetChildren(dbname, &filenames);
1487 3 : if (filenames.empty()) {
1488 : return Status::OK();
1489 : }
1490 :
1491 : FileLock* lock;
1492 3 : const std::string lockname = LockFileName(dbname);
1493 3 : Status result = env->LockFile(lockname, &lock);
1494 3 : if (result.ok()) {
1495 : uint64_t number;
1496 : FileType type;
1497 45 : for (size_t i = 0; i < filenames.size(); i++) {
1498 57 : if (ParseFileName(filenames[i], &number, &type) &&
1499 15 : type != kDBLockFile) { // Lock file will be deleted at end
1500 60 : Status del = env->DeleteFile(dbname + "/" + filenames[i]);
1501 24 : if (result.ok() && !del.ok()) {
1502 0 : result = del;
1503 : }
1504 : }
1505 : }
1506 6 : env->UnlockFile(lock); // Ignore error since state is already gone
1507 6 : env->DeleteFile(lockname);
1508 6 : env->DeleteDir(dbname); // Ignore error in case dir contains other files
1509 : }
1510 3 : return result;
1511 : }
1512 :
1513 : } // namespace leveldb
|
