db_impl.h 26.3 KB
Newer Older
1 2 3 4 5
//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
//  This source code is licensed under the BSD-style license found in the
//  LICENSE file in the root directory of this source tree. An additional grant
//  of patent rights can be found in the PATENTS file in the same directory.
//
J
jorlow@chromium.org 已提交
6 7 8
// 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.
9
#pragma once
K
Kai Liu 已提交
10

H
Haobo Xu 已提交
11
#include <atomic>
12
#include <deque>
13
#include <limits>
J
jorlow@chromium.org 已提交
14
#include <set>
T
Tomislav Novak 已提交
15
#include <utility>
16
#include <vector>
I
Igor Canadi 已提交
17
#include <string>
K
kailiu 已提交
18

J
jorlow@chromium.org 已提交
19 20 21
#include "db/dbformat.h"
#include "db/log_writer.h"
#include "db/snapshot.h"
22
#include "db/column_family.h"
23
#include "db/version_edit.h"
K
Kai Liu 已提交
24 25
#include "memtable_list.h"
#include "port/port.h"
26 27 28 29
#include "rocksdb/db.h"
#include "rocksdb/env.h"
#include "rocksdb/memtablerep.h"
#include "rocksdb/transaction_log.h"
30
#include "util/autovector.h"
31
#include "util/stop_watch.h"
32
#include "util/thread_local.h"
33
#include "util/scoped_arena_iterator.h"
I
Igor Canadi 已提交
34
#include "db/internal_stats.h"
35

36
namespace rocksdb {
J
jorlow@chromium.org 已提交
37 38 39 40 41 42

class MemTable;
class TableCache;
class Version;
class VersionEdit;
class VersionSet;
D
Danny Guo 已提交
43
class CompactionFilterV2;
44
class Arena;
J
jorlow@chromium.org 已提交
45 46 47

class DBImpl : public DB {
 public:
I
Igor Canadi 已提交
48
  DBImpl(const DBOptions& options, const std::string& dbname);
J
jorlow@chromium.org 已提交
49 50 51
  virtual ~DBImpl();

  // Implementations of the DB interface
52 53
  using DB::Put;
  virtual Status Put(const WriteOptions& options,
54
                     ColumnFamilyHandle* column_family, const Slice& key,
55 56 57
                     const Slice& value);
  using DB::Merge;
  virtual Status Merge(const WriteOptions& options,
58 59
                       ColumnFamilyHandle* column_family, const Slice& key,
                       const Slice& value);
60 61
  using DB::Delete;
  virtual Status Delete(const WriteOptions& options,
62
                        ColumnFamilyHandle* column_family, const Slice& key);
63
  using DB::Write;
J
jorlow@chromium.org 已提交
64
  virtual Status Write(const WriteOptions& options, WriteBatch* updates);
65
  using DB::Get;
J
jorlow@chromium.org 已提交
66
  virtual Status Get(const ReadOptions& options,
67
                     ColumnFamilyHandle* column_family, const Slice& key,
J
jorlow@chromium.org 已提交
68
                     std::string* value);
69 70 71
  using DB::MultiGet;
  virtual std::vector<Status> MultiGet(
      const ReadOptions& options,
72
      const std::vector<ColumnFamilyHandle*>& column_family,
73
      const std::vector<Slice>& keys, std::vector<std::string>* values);
74

75
  virtual Status CreateColumnFamily(const ColumnFamilyOptions& options,
76
                                    const std::string& column_family,
77 78
                                    ColumnFamilyHandle** handle);
  virtual Status DropColumnFamily(ColumnFamilyHandle* column_family);
79

80 81 82 83
  // Returns false if key doesn't exist in the database and true if it may.
  // If value_found is not passed in as null, then return the value if found in
  // memory. On return, if value was found, then value_found will be set to true
  // , otherwise false.
84
  using DB::KeyMayExist;
85
  virtual bool KeyMayExist(const ReadOptions& options,
86 87
                           ColumnFamilyHandle* column_family, const Slice& key,
                           std::string* value, bool* value_found = nullptr);
88 89
  using DB::NewIterator;
  virtual Iterator* NewIterator(const ReadOptions& options,
90
                                ColumnFamilyHandle* column_family);
91 92
  virtual Status NewIterators(
      const ReadOptions& options,
I
Igor Canadi 已提交
93
      const std::vector<ColumnFamilyHandle*>& column_families,
94
      std::vector<Iterator*>* iterators);
J
jorlow@chromium.org 已提交
95 96
  virtual const Snapshot* GetSnapshot();
  virtual void ReleaseSnapshot(const Snapshot* snapshot);
97
  using DB::GetProperty;
98
  virtual bool GetProperty(ColumnFamilyHandle* column_family,
99
                           const Slice& property, std::string* value);
100 101 102
  using DB::GetIntProperty;
  virtual bool GetIntProperty(ColumnFamilyHandle* column_family,
                              const Slice& property, uint64_t* value) override;
103
  using DB::GetApproximateSizes;
104
  virtual void GetApproximateSizes(ColumnFamilyHandle* column_family,
105 106
                                   const Range* range, int n, uint64_t* sizes);
  using DB::CompactRange;
107
  virtual Status CompactRange(ColumnFamilyHandle* column_family,
108
                              const Slice* begin, const Slice* end,
109 110
                              bool reduce_level = false, int target_level = -1,
                              uint32_t target_path_id = 0);
111 112

  using DB::NumberLevels;
113
  virtual int NumberLevels(ColumnFamilyHandle* column_family);
114
  using DB::MaxMemCompactionLevel;
115
  virtual int MaxMemCompactionLevel(ColumnFamilyHandle* column_family);
116
  using DB::Level0StopWriteTrigger;
117
  virtual int Level0StopWriteTrigger(ColumnFamilyHandle* column_family);
I
Igor Canadi 已提交
118
  virtual const std::string& GetName() const;
119
  virtual Env* GetEnv() const;
120
  using DB::GetOptions;
121
  virtual const Options& GetOptions(ColumnFamilyHandle* column_family) const;
122 123
  using DB::Flush;
  virtual Status Flush(const FlushOptions& options,
124
                       ColumnFamilyHandle* column_family);
I
Igor Canadi 已提交
125 126 127 128

  virtual SequenceNumber GetLatestSequenceNumber() const;

#ifndef ROCKSDB_LITE
129
  virtual Status DisableFileDeletions();
130
  virtual Status EnableFileDeletions(bool force);
131
  virtual int IsFileDeletionsEnabled() const;
I
Igor Canadi 已提交
132
  // All the returned filenames start with "/"
133
  virtual Status GetLiveFiles(std::vector<std::string>&,
134 135
                              uint64_t* manifest_file_size,
                              bool flush_memtable = true);
136
  virtual Status GetSortedWalFiles(VectorLogPtr& files);
I
Igor Canadi 已提交
137

138 139 140 141
  virtual Status GetUpdatesSince(
      SequenceNumber seq_number, unique_ptr<TransactionLogIterator>* iter,
      const TransactionLogIterator::ReadOptions&
          read_options = TransactionLogIterator::ReadOptions());
142 143
  virtual Status DeleteFile(std::string name);

144
  virtual void GetLiveFilesMetaData(std::vector<LiveFileMetaData>* metadata);
I
Igor Canadi 已提交
145
#endif  // ROCKSDB_LITE
146

I
Igor Canadi 已提交
147 148 149 150
  // checks if all live files exist on file system and that their file sizes
  // match to our in-memory records
  virtual Status CheckConsistency();

151 152
  virtual Status GetDbIdentity(std::string& identity);

I
Igor Canadi 已提交
153
  Status RunManualCompaction(ColumnFamilyData* cfd, int input_level,
154 155
                             int output_level, uint32_t output_path_id,
                             const Slice* begin, const Slice* end);
156

157
#ifndef ROCKSDB_LITE
J
jorlow@chromium.org 已提交
158
  // Extra methods (for testing) that are not in the public DB interface
I
Igor Canadi 已提交
159
  // Implemented in db_impl_debug.cc
J
jorlow@chromium.org 已提交
160

161
  // Compact any files in the named level that overlap [*begin, *end]
162 163
  Status TEST_CompactRange(int level, const Slice* begin, const Slice* end,
                           ColumnFamilyHandle* column_family = nullptr);
J
jorlow@chromium.org 已提交
164

165
  // Force current memtable contents to be flushed.
166
  Status TEST_FlushMemTable(bool wait = true);
J
jorlow@chromium.org 已提交
167

168
  // Wait for memtable compaction
169
  Status TEST_WaitForFlushMemTable(ColumnFamilyHandle* column_family = nullptr);
170 171 172 173

  // Wait for any compaction
  Status TEST_WaitForCompact();

J
jorlow@chromium.org 已提交
174 175 176
  // Return an internal iterator over the current state of the database.
  // The keys of this iterator are internal keys (see format.h).
  // The returned iterator should be deleted when no longer needed.
177 178
  Iterator* TEST_NewInternalIterator(
      Arena* arena, ColumnFamilyHandle* column_family = nullptr);
J
jorlow@chromium.org 已提交
179

180 181
  // Return the maximum overlapping data (in bytes) at next level for any
  // file at a level >= 1.
182 183
  int64_t TEST_MaxNextLevelOverlappingBytes(ColumnFamilyHandle* column_family =
                                                nullptr);
184

A
Abhishek Kona 已提交
185 186
  // Return the current manifest file no.
  uint64_t TEST_Current_Manifest_FileNo();
187 188 189 190

  // Trigger's a background call for testing.
  void TEST_PurgeObsoleteteWAL();

191
  // get total level0 file size. Only for testing.
192
  uint64_t TEST_GetLevel0TotalSize();
193

194 195 196 197 198
  void TEST_SetDefaultTimeToCheck(uint64_t default_interval_to_delete_obsolete_WAL)
  {
    default_interval_to_delete_obsolete_WAL_ = default_interval_to_delete_obsolete_WAL;
  }

199 200
  void TEST_GetFilesMetaData(ColumnFamilyHandle* column_family,
                             std::vector<std::vector<FileMetaData>>* metadata);
201

I
Igor Canadi 已提交
202 203 204 205
  Status TEST_ReadFirstRecord(const WalFileType type, const uint64_t number,
                              SequenceNumber* sequence);

  Status TEST_ReadFirstLine(const std::string& fname, SequenceNumber* sequence);
I
Igor Canadi 已提交
206 207
#endif  // NDEBUG

208 209 210 211 212 213 214 215 216 217 218
  // Structure to store information for candidate files to delete.
  struct CandidateFileInfo {
    std::string file_name;
    uint32_t path_id;
    CandidateFileInfo(std::string name, uint32_t path)
        : file_name(name), path_id(path) {}
    bool operator==(const CandidateFileInfo& other) const {
      return file_name == other.file_name && path_id == other.path_id;
    }
  };

I
Igor Canadi 已提交
219
  // needed for CleanupIteratorState
I
Igor Canadi 已提交
220 221
  struct DeletionState {
    inline bool HaveSomethingToDelete() const {
K
kailiu 已提交
222
      return  candidate_files.size() ||
I
Igor Canadi 已提交
223 224 225
        sst_delete_files.size() ||
        log_delete_files.size();
    }
226

I
Igor Canadi 已提交
227 228 229
    // a list of all files that we'll consider deleting
    // (every once in a while this is filled up with all files
    // in the DB directory)
230
    std::vector<CandidateFileInfo> candidate_files;
I
Igor Canadi 已提交
231 232

    // the list of all live sst files that cannot be deleted
233
    std::vector<FileDescriptor> sst_live;
I
Igor Canadi 已提交
234 235 236 237 238 239 240

    // a list of sst files that we need to delete
    std::vector<FileMetaData*> sst_delete_files;

    // a list of log files that we need to delete
    std::vector<uint64_t> log_delete_files;

241
    // a list of memtables to be free
K
Kai Liu 已提交
242
    autovector<MemTable*> memtables_to_free;
243

244
    autovector<SuperVersion*> superversions_to_free;
I
Igor Canadi 已提交
245

246
    SuperVersion* new_superversion;  // if nullptr no new superversion
I
Igor Canadi 已提交
247

I
Igor Canadi 已提交
248 249
    // the current manifest_file_number, log_number and prev_log_number
    // that corresponds to the set of files in 'live'.
250 251
    uint64_t manifest_file_number, pending_manifest_file_number, log_number,
        prev_log_number;
I
Igor Canadi 已提交
252

K
Kai Liu 已提交
253
    explicit DeletionState(bool create_superversion = false) {
I
Igor Canadi 已提交
254
      manifest_file_number = 0;
255
      pending_manifest_file_number = 0;
I
Igor Canadi 已提交
256 257
      log_number = 0;
      prev_log_number = 0;
258
      new_superversion = create_superversion ? new SuperVersion() : nullptr;
I
Igor Canadi 已提交
259 260 261 262 263 264 265
    }

    ~DeletionState() {
      // free pending memtables
      for (auto m : memtables_to_free) {
        delete m;
      }
266 267 268 269
      // free superversions
      for (auto s : superversions_to_free) {
        delete s;
      }
I
Igor Canadi 已提交
270 271 272
      // if new_superversion was not used, it will be non-nullptr and needs
      // to be freed here
      delete new_superversion;
I
Igor Canadi 已提交
273 274 275 276
    }
  };

  // Returns the list of live files in 'live' and the list
K
kailiu 已提交
277
  // of all files in the filesystem in 'candidate_files'.
I
Igor Canadi 已提交
278 279 280 281 282 283 284 285 286 287 288 289 290
  // If force == false and the last call was less than
  // options_.delete_obsolete_files_period_micros microseconds ago,
  // it will not fill up the deletion_state
  void FindObsoleteFiles(DeletionState& deletion_state,
                         bool force,
                         bool no_full_scan = false);

  // Diffs the files listed in filenames and those that do not
  // belong to live files are posibly removed. Also, removes all the
  // files in sst_delete_files and log_delete_files.
  // It is not necessary to hold the mutex when invoking this method.
  void PurgeObsoleteFiles(DeletionState& deletion_state);

291 292
  ColumnFamilyHandle* DefaultColumnFamily() const;

293
 protected:
H
heyongqiang 已提交
294 295
  Env* const env_;
  const std::string dbname_;
296
  unique_ptr<VersionSet> versions_;
I
Igor Canadi 已提交
297
  const DBOptions options_;
L
Lei Jin 已提交
298
  Statistics* stats_;
H
heyongqiang 已提交
299

300
  Iterator* NewInternalIterator(const ReadOptions&, ColumnFamilyData* cfd,
301
                                SuperVersion* super_version, Arena* arena);
302

J
jorlow@chromium.org 已提交
303 304
 private:
  friend class DB;
305
  friend class InternalStats;
I
Igor Canadi 已提交
306
#ifndef ROCKSDB_LITE
T
Tomislav Novak 已提交
307
  friend class TailingIterator;
L
Lei Jin 已提交
308
  friend class ForwardIterator;
I
Igor Canadi 已提交
309
#endif
310
  friend struct SuperVersion;
311 312
  struct CompactionState;
  struct Writer;
S
Stanislau Hlebik 已提交
313
  struct WriteContext;
J
jorlow@chromium.org 已提交
314 315 316 317 318 319

  Status NewDB();

  // Recover the descriptor from persistent storage.  May do a significant
  // amount of work to recover recently logged updates.  Any changes to
  // be made to the descriptor are added to *edit.
320 321
  Status Recover(const std::vector<ColumnFamilyDescriptor>& column_families,
                 bool read_only = false, bool error_if_log_file_exist = false);
J
jorlow@chromium.org 已提交
322 323 324

  void MaybeIgnoreError(Status* s) const;

325 326
  const Status CreateArchivalDirectory();

J
jorlow@chromium.org 已提交
327 328 329
  // Delete any unneeded files and stale in-memory entries.
  void DeleteObsoleteFiles();

330
  // Flush the in-memory write buffer to storage.  Switches to a new
J
jorlow@chromium.org 已提交
331
  // log-file/memtable and writes a new descriptor iff successful.
332
  Status FlushMemTableToOutputFile(ColumnFamilyData* cfd, bool* madeProgress,
H
Haobo Xu 已提交
333 334
                                   DeletionState& deletion_state,
                                   LogBuffer* log_buffer);
J
jorlow@chromium.org 已提交
335

I
Igor Canadi 已提交
336 337
  Status RecoverLogFile(uint64_t log_number, SequenceNumber* max_sequence,
                        bool read_only);
J
jorlow@chromium.org 已提交
338

339 340 341 342 343
  // The following two methods are used to flush a memtable to
  // storage. The first one is used atdatabase RecoveryTime (when the
  // database is opened) and is heavyweight because it holds the mutex
  // for the entire period. The second method WriteLevel0Table supports
  // concurrent flush memtables to storage.
344 345
  Status WriteLevel0TableForRecovery(ColumnFamilyData* cfd, MemTable* mem,
                                     VersionEdit* edit);
346
  Status WriteLevel0Table(ColumnFamilyData* cfd, autovector<MemTable*>& mems,
347
                          VersionEdit* edit, uint64_t* filenumber,
H
Haobo Xu 已提交
348
                          LogBuffer* log_buffer);
J
jorlow@chromium.org 已提交
349

M
Mark Callaghan 已提交
350
  uint64_t SlowdownAmount(int n, double bottom, double top);
351

S
Stanislau Hlebik 已提交
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
  // Before applying write operation (such as DBImpl::Write, DBImpl::Flush)
  // thread should grab the mutex_ and be the first on writers queue.
  // BeginWrite is used for it.
  // Be aware! Writer's job can be done by other thread (see DBImpl::Write
  // for examples), so check it via w.done before applying changes.
  //
  // Writer* w:                writer to be placed in the queue
  // uint64_t expiration_time: maximum time to be in the queue
  // See also: EndWrite
  Status BeginWrite(Writer* w, uint64_t expiration_time);

  // After doing write job, we need to remove already used writers from
  // writers_ queue and notify head of the queue about it.
  // EndWrite is used for this.
  //
  // Writer* w:           Writer, that was added by BeginWrite function
  // Writer* last_writer: Since we can join a few Writers (as DBImpl::Write
  //                      does)
  //                      we should pass last_writer as a parameter to
  //                      EndWrite
  //                      (if you don't touch other writers, just pass w)
  // Status status:       Status of write operation
  // See also: BeginWrite
  void EndWrite(Writer* w, Writer* last_writer, Status status);

377
  Status MakeRoomForWrite(ColumnFamilyData* cfd,
S
Stanislau Hlebik 已提交
378
                          WriteContext* context,
379
                          uint64_t expiration_time);
380

S
Stanislau Hlebik 已提交
381 382 383
  Status SetNewMemtableAndNewLogFile(ColumnFamilyData* cfd,
                                     WriteContext* context);

384 385
  void BuildBatchGroup(Writer** last_writer,
                       autovector<WriteBatch*>* write_batch_group);
J
jorlow@chromium.org 已提交
386

H
heyongqiang 已提交
387
  // Force current memtable contents to be flushed.
388
  Status FlushMemTable(ColumnFamilyData* cfd, const FlushOptions& options);
H
heyongqiang 已提交
389

390
  // Wait for memtable flushed
391
  Status WaitForFlushMemTable(ColumnFamilyData* cfd);
H
heyongqiang 已提交
392

393 394 395
  void RecordFlushIOStats();
  void RecordCompactionIOStats();

396
  void MaybeScheduleFlushOrCompaction();
397 398 399 400
  static void BGWorkCompaction(void* db);
  static void BGWorkFlush(void* db);
  void BackgroundCallCompaction();
  void BackgroundCallFlush();
401 402
  Status BackgroundCompaction(bool* madeProgress, DeletionState& deletion_state,
                              LogBuffer* log_buffer);
H
Haobo Xu 已提交
403 404
  Status BackgroundFlush(bool* madeProgress, DeletionState& deletion_state,
                         LogBuffer* log_buffer);
405
  void CleanupCompaction(CompactionState* compact, Status status);
I
Igor Canadi 已提交
406
  Status DoCompactionWork(CompactionState* compact,
H
Haobo Xu 已提交
407 408
                          DeletionState& deletion_state,
                          LogBuffer* log_buffer);
J
jorlow@chromium.org 已提交
409

I
Igor Canadi 已提交
410 411 412 413 414 415 416
  // This function is called as part of compaction. It enables Flush process to
  // preempt compaction, since it's higher prioirty
  // Returns: micros spent executing
  uint64_t CallFlushDuringCompaction(ColumnFamilyData* cfd,
                                     DeletionState& deletion_state,
                                     LogBuffer* log_buffer);

D
Danny Guo 已提交
417 418 419
  // Call compaction filter if is_compaction_v2 is not true. Then iterate
  // through input and compact the kv-pairs
  Status ProcessKeyValueCompaction(
420
    bool is_snapshot_supported,
D
Danny Guo 已提交
421 422 423 424 425 426 427 428 429 430 431 432 433 434 435
    SequenceNumber visible_at_tip,
    SequenceNumber earliest_snapshot,
    SequenceNumber latest_snapshot,
    DeletionState& deletion_state,
    bool bottommost_level,
    int64_t& imm_micros,
    Iterator* input,
    CompactionState* compact,
    bool is_compaction_v2,
    LogBuffer* log_buffer);

  // Call compaction_filter_v2->Filter() on kv-pairs in compact
  void CallCompactionFilterV2(CompactionState* compact,
    CompactionFilterV2* compaction_filter_v2);

J
jorlow@chromium.org 已提交
436 437
  Status OpenCompactionOutputFile(CompactionState* compact);
  Status FinishCompactionOutputFile(CompactionState* compact, Iterator* input);
438 439
  Status InstallCompactionResults(CompactionState* compact,
                                  LogBuffer* log_buffer);
440 441
  void AllocateCompactionOutputFileNumbers(CompactionState* compact);
  void ReleaseCompactionUnusedFileNumbers(CompactionState* compact);
A
Abhishek Kona 已提交
442

I
Igor Canadi 已提交
443 444 445 446 447 448
#ifdef ROCKSDB_LITE
  void PurgeObsoleteWALFiles() {
    // this function is used for archiving WAL files. we don't need this in
    // ROCKSDB_LITE
  }
#else
449
  void PurgeObsoleteWALFiles();
450

451 452 453
  Status GetSortedWalsOfType(const std::string& path,
                             VectorLogPtr& log_files,
                             WalFileType type);
454

455 456 457 458 459
  // Requires: all_logs should be sorted with earliest log file first
  // Retains all log files in all_logs which contain updates with seq no.
  // Greater Than or Equal to the requested SequenceNumber.
  Status RetainProbableWalFiles(VectorLogPtr& all_logs,
                                const SequenceNumber target);
460

461
  Status ReadFirstRecord(const WalFileType type, const uint64_t number,
I
Igor Canadi 已提交
462
                         SequenceNumber* sequence);
463

I
Igor Canadi 已提交
464
  Status ReadFirstLine(const std::string& fname, SequenceNumber* sequence);
I
Igor Canadi 已提交
465
#endif  // ROCKSDB_LITE
466

467 468
  void PrintStatistics();

469
  // dump rocksdb.stats to LOG
470 471
  void MaybeDumpStats();

472 473 474 475 476 477 478
  // Return true if the current db supports snapshot.  If the current
  // DB does not support snapshot, then calling GetSnapshot() will always
  // return nullptr.
  //
  // @see GetSnapshot()
  virtual bool IsSnapshotSupported() const;

479 480
  // Return the minimum empty level that could hold the total data in the
  // input level. Return the input level, if such level could not be found.
I
Igor Canadi 已提交
481
  int FindMinimumEmptyLevelFitting(ColumnFamilyData* cfd, int level);
482

483 484 485
  // Move the files in the input level to the target level.
  // If target_level < 0, automatically calculate the minimum level that could
  // hold the data set.
I
Igor Canadi 已提交
486
  Status ReFitLevel(ColumnFamilyData* cfd, int level, int target_level = -1);
487

J
jorlow@chromium.org 已提交
488
  // table_cache_ provides its own synchronization
I
Igor Canadi 已提交
489
  std::shared_ptr<Cache> table_cache_;
J
jorlow@chromium.org 已提交
490

491
  // Lock over the persistent DB state.  Non-nullptr iff successfully acquired.
J
jorlow@chromium.org 已提交
492 493 494 495 496
  FileLock* db_lock_;

  // State below is protected by mutex_
  port::Mutex mutex_;
  port::AtomicPointer shutting_down_;
497 498 499 500 501 502 503 504 505
  // This condition variable is signaled on these conditions:
  // * whenever bg_compaction_scheduled_ goes down to 0
  // * if bg_manual_only_ > 0, whenever a compaction finishes, even if it hasn't
  // made any progress
  // * whenever a compaction made any progress
  // * whenever bg_flush_scheduled_ value decreases (i.e. whenever a flush is
  // done, even if it didn't make any progress)
  // * whenever there is an error in background flush or compaction
  port::CondVar bg_cv_;
506
  uint64_t logfile_number_;
507
  unique_ptr<log::Writer> log_;
I
Igor Canadi 已提交
508
  bool log_empty_;
509
  ColumnFamilyHandleImpl* default_cf_handle_;
510
  InternalStats* default_cf_internal_stats_;
511
  unique_ptr<ColumnFamilyMemTablesImpl> column_family_memtables_;
I
Igor Canadi 已提交
512 513 514 515 516 517 518 519 520 521 522 523 524
  struct LogFileNumberSize {
    explicit LogFileNumberSize(uint64_t _number)
        : number(_number), size(0), getting_flushed(false) {}
    void AddSize(uint64_t new_size) { size += new_size; }
    uint64_t number;
    uint64_t size;
    bool getting_flushed;
  };
  std::deque<LogFileNumberSize> alive_log_files_;
  uint64_t total_log_size_;
  // only used for dynamically adjusting max_total_wal_size. it is a sum of
  // [write_buffer_size * max_write_buffer_number] over all column families
  uint64_t max_total_in_memory_state_;
525 526 527
  // If true, we have only one (default) column family. We use this to optimize
  // some code-paths
  bool single_column_family_mode_;
I
Igor Canadi 已提交
528

529 530
  std::unique_ptr<Directory> db_directory_;

531 532
  // Queue of writers.
  std::deque<Writer*> writers_;
533
  WriteBatch tmp_batch_;
534

J
jorlow@chromium.org 已提交
535 536
  SnapshotList snapshots_;

I
Igor Canadi 已提交
537 538 539 540
  // cache for ReadFirstRecord() calls
  std::unordered_map<uint64_t, SequenceNumber> read_first_record_cache_;
  port::Mutex read_first_record_cache_mutex_;

J
jorlow@chromium.org 已提交
541 542
  // Set of table files to protect from deletion because they are
  // part of ongoing compactions.
543 544
  // map from pending file number ID to their path IDs.
  FileNumToPathIdMap pending_outputs_;
J
jorlow@chromium.org 已提交
545

546 547 548 549
  // At least one compaction or flush job is pending but not yet scheduled
  // because of the max background thread limit.
  bool bg_schedule_needed_;

550
  // count how many background compactions are running or have been scheduled
551
  int bg_compaction_scheduled_;
J
jorlow@chromium.org 已提交
552

553 554 555 556 557
  // If non-zero, MaybeScheduleFlushOrCompaction() will only schedule manual
  // compactions (if manual_compaction_ is not null). This mechanism enables
  // manual compactions to wait until all other compactions are finished.
  int bg_manual_only_;

558 559 560
  // number of background memtable flush jobs, submitted to the HIGH pool
  int bg_flush_scheduled_;

H
hans@chromium.org 已提交
561 562
  // Information for a manual compaction
  struct ManualCompaction {
I
Igor Canadi 已提交
563
    ColumnFamilyData* cfd;
564 565
    int input_level;
    int output_level;
566
    uint32_t output_path_id;
G
Gabor Cselle 已提交
567
    bool done;
L
Lei Jin 已提交
568
    Status status;
569
    bool in_progress;           // compaction request being processed?
570 571
    const InternalKey* begin;   // nullptr means beginning of key range
    const InternalKey* end;     // nullptr means end of key range
G
Gabor Cselle 已提交
572
    InternalKey tmp_storage;    // Used to keep track of compaction progress
H
hans@chromium.org 已提交
573 574
  };
  ManualCompaction* manual_compaction_;
J
jorlow@chromium.org 已提交
575 576 577 578

  // Have we encountered a background error in paranoid mode?
  Status bg_error_;

579
  // shall we disable deletion of obsolete files
580 581 582 583 584 585
  // if 0 the deletion is enabled.
  // if non-zero, files will not be getting deleted
  // This enables two different threads to call
  // EnableFileDeletions() and DisableFileDeletions()
  // without any synchronization
  int disable_delete_obsolete_files_;
586

587 588 589
  // last time when DeleteObsoleteFiles was invoked
  uint64_t delete_obsolete_files_last_run_;

590 591 592
  // last time when PurgeObsoleteWALFiles ran.
  uint64_t purge_wal_files_last_run_;

593
  // last time stats were dumped to LOG
H
Haobo Xu 已提交
594
  std::atomic<uint64_t> last_stats_dump_time_microsec_;
595

596 597 598 599
  // obsolete files will be deleted every this seconds if ttl deletion is
  // enabled and archive size_limit is disabled.
  uint64_t default_interval_to_delete_obsolete_WAL_;

600 601
  bool flush_on_destroy_; // Used when disableWAL is true.

H
heyongqiang 已提交
602
  static const int KEEP_LOG_FILE_NUM = 1000;
603
  static const uint64_t kNoTimeOut = std::numeric_limits<uint64_t>::max();
H
heyongqiang 已提交
604
  std::string db_absolute_path_;
H
heyongqiang 已提交
605

606 607 608
  // count of the number of contiguous delaying writes
  int delayed_writes_;

609
  // The options to access storage files
L
Lei Jin 已提交
610
  const EnvOptions env_options_;
611

612 613 614 615 616 617
  // A value of true temporarily disables scheduling of background work
  bool bg_work_gate_closed_;

  // Guard against multiple concurrent refitting
  bool refitting_level_;

618 619 620
  // Indicate DB was opened successfully
  bool opened_successfully_;

J
jorlow@chromium.org 已提交
621 622 623 624
  // No copying allowed
  DBImpl(const DBImpl&);
  void operator=(const DBImpl&);

625 626
  // dump the delayed_writes_ to the log file and reset counter.
  void DelayLoggingAndReset();
627

628 629 630 631 632 633
  // Return the earliest snapshot where seqno is visible.
  // Store the snapshot right before that, if any, in prev_snapshot
  inline SequenceNumber findEarliestVisibleSnapshot(
    SequenceNumber in,
    std::vector<SequenceNumber>& snapshots,
    SequenceNumber* prev_snapshot);
634

I
Igor Canadi 已提交
635
  // Background threads call this function, which is just a wrapper around
636
  // the cfd->InstallSuperVersion() function. Background threads carry
I
Igor Canadi 已提交
637
  // deletion_state which can have new_superversion already allocated.
638 639
  void InstallSuperVersion(ColumnFamilyData* cfd,
                           DeletionState& deletion_state);
I
Igor Canadi 已提交
640

641 642 643 644 645 646 647 648 649 650
  // Find Super version and reference it. Based on options, it might return
  // the thread local cached one.
  inline SuperVersion* GetAndRefSuperVersion(ColumnFamilyData* cfd);

  // Un-reference the super version and return it to thread local cache if
  // needed. If it is the last reference of the super version. Clean it up
  // after un-referencing it.
  inline void ReturnAndCleanupSuperVersion(ColumnFamilyData* cfd,
                                           SuperVersion* sv);

I
Igor Canadi 已提交
651
#ifndef ROCKSDB_LITE
I
Igor Canadi 已提交
652 653 654
  using DB::GetPropertiesOfAllTables;
  virtual Status GetPropertiesOfAllTables(ColumnFamilyHandle* column_family,
                                          TablePropertiesCollection* props)
655
      override;
I
Igor Canadi 已提交
656
#endif  // ROCKSDB_LITE
657

658 659
  // Function that Get and KeyMayExist call with no_io true or false
  // Note: 'value_found' from KeyMayExist propagates here
660 661 662
  Status GetImpl(const ReadOptions& options, ColumnFamilyHandle* column_family,
                 const Slice& key, std::string* value,
                 bool* value_found = nullptr);
663 664 665 666

  bool GetIntPropertyInternal(ColumnFamilyHandle* column_family,
                              DBPropertyType property_type,
                              bool need_out_of_mutex, uint64_t* value);
J
jorlow@chromium.org 已提交
667 668 669 670 671 672 673
};

// Sanitize db options.  The caller should delete result.info_log if
// it is not equal to src.info_log.
extern Options SanitizeOptions(const std::string& db,
                               const InternalKeyComparator* icmp,
                               const Options& src);
674
extern DBOptions SanitizeOptions(const std::string& db, const DBOptions& src);
S
Siying Dong 已提交
675

M
miguelportilla 已提交
676 677 678 679 680 681 682
// Fix user-supplied options to be reasonable
template <class T, class V>
static void ClipToRange(T* ptr, V minvalue, V maxvalue) {
  if (static_cast<V>(*ptr) > maxvalue) *ptr = maxvalue;
  if (static_cast<V>(*ptr) < minvalue) *ptr = minvalue;
}

F
Feng Zhu 已提交
683 684 685 686
// Dump db file summary, implemented in util/
extern void DumpDBFileSummary(const DBOptions& options,
                              const std::string& dbname);

687
}  // namespace rocksdb