db_test_util.h 37.9 KB
Newer Older
1
// Copyright (c) 2011-present, Facebook, Inc.  All rights reserved.
S
Siying Dong 已提交
2 3 4
//  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).
5 6 7 8 9 10
//
// 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.

#pragma once
11

12 13 14
#include <fcntl.h>

#include <algorithm>
15
#include <cinttypes>
16
#include <map>
17 18 19 20 21 22 23
#include <set>
#include <string>
#include <thread>
#include <unordered_set>
#include <utility>
#include <vector>

24
#include "db/db_impl/db_impl.h"
25
#include "db/dbformat.h"
26
#include "file/filename.h"
27
#include "memtable/hash_linklist_rep.h"
28 29
#include "rocksdb/cache.h"
#include "rocksdb/compaction_filter.h"
A
agiardullo 已提交
30
#include "rocksdb/convenience.h"
31 32 33 34 35
#include "rocksdb/db.h"
#include "rocksdb/env.h"
#include "rocksdb/filter_policy.h"
#include "rocksdb/options.h"
#include "rocksdb/slice.h"
36
#include "rocksdb/sst_file_writer.h"
Y
Yi Wu 已提交
37
#include "rocksdb/statistics.h"
38 39 40
#include "rocksdb/table.h"
#include "rocksdb/utilities/checkpoint.h"
#include "table/mock_table.h"
S
sdong 已提交
41
#include "table/scoped_arena_iterator.h"
42 43
#include "test_util/sync_point.h"
#include "test_util/testharness.h"
44 45 46
#include "util/cast_util.h"
#include "util/compression.h"
#include "util/mutexlock.h"
47
#include "util/string_util.h"
48 49
#include "utilities/merge_operators.h"

50
namespace ROCKSDB_NAMESPACE {
51
class MockEnv;
52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75

namespace anon {
class AtomicCounter {
 public:
  explicit AtomicCounter(Env* env = NULL)
      : env_(env), cond_count_(&mu_), count_(0) {}

  void Increment() {
    MutexLock l(&mu_);
    count_++;
    cond_count_.SignalAll();
  }

  int Read() {
    MutexLock l(&mu_);
    return count_;
  }

  bool WaitFor(int count) {
    MutexLock l(&mu_);

    uint64_t start = env_->NowMicros();
    while (count_ < count) {
      uint64_t now = env_->NowMicros();
76 77
      cond_count_.TimedWait(now + /*1s*/ 1 * 1000 * 1000);
      if (env_->NowMicros() - start > /*10s*/ 10 * 1000 * 1000) {
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
        return false;
      }
      if (count_ < count) {
        GTEST_LOG_(WARNING) << "WaitFor is taking more time than usual";
      }
    }

    return true;
  }

  void Reset() {
    MutexLock l(&mu_);
    count_ = 0;
    cond_count_.SignalAll();
  }

 private:
  Env* env_;
  port::Mutex mu_;
  port::CondVar cond_count_;
  int count_;
};

struct OptionsOverride {
  std::shared_ptr<const FilterPolicy> filter_policy = nullptr;
M
Maysam Yabandeh 已提交
103 104
  // These will be used only if filter_policy is set
  bool partition_filters = false;
M
Maysam Yabandeh 已提交
105
  uint64_t metadata_block_size = 1024;
106 107 108 109 110 111 112

  // Used as a bit mask of individual enums in which to skip an XF test point
  int skip_policy = 0;
};

}  // namespace anon

S
Siying Dong 已提交
113 114
enum SkipPolicy { kSkipNone = 0, kSkipNoSnapshot = 1, kSkipNoPrefix = 2 };

115 116 117
// A hacky skip list mem table that triggers flush after number of entries.
class SpecialMemTableRep : public MemTableRep {
 public:
118 119
  explicit SpecialMemTableRep(Allocator* allocator, MemTableRep* memtable,
                              int num_entries_flush)
120 121 122 123 124 125 126 127 128 129 130
      : MemTableRep(allocator),
        memtable_(memtable),
        num_entries_flush_(num_entries_flush),
        num_entries_(0) {}

  virtual KeyHandle Allocate(const size_t len, char** buf) override {
    return memtable_->Allocate(len, buf);
  }

  // Insert key into the list.
  // REQUIRES: nothing that compares equal to key is currently in the list.
M
Maysam Yabandeh 已提交
131
  virtual void Insert(KeyHandle handle) override {
132
    num_entries_++;
M
Maysam Yabandeh 已提交
133
    memtable_->Insert(handle);
134 135
  }

M
Maysam Yabandeh 已提交
136 137 138 139 140
  void InsertConcurrently(KeyHandle handle) override {
    num_entries_++;
    memtable_->Insert(handle);
  }

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
  // Returns true iff an entry that compares equal to key is in the list.
  virtual bool Contains(const char* key) const override {
    return memtable_->Contains(key);
  }

  virtual size_t ApproximateMemoryUsage() override {
    // Return a high memory usage when number of entries exceeds the threshold
    // to trigger a flush.
    return (num_entries_ < num_entries_flush_) ? 0 : 1024 * 1024 * 1024;
  }

  virtual void Get(const LookupKey& k, void* callback_args,
                   bool (*callback_func)(void* arg,
                                         const char* entry)) override {
    memtable_->Get(k, callback_args, callback_func);
  }

  uint64_t ApproximateNumEntries(const Slice& start_ikey,
                                 const Slice& end_ikey) override {
    return memtable_->ApproximateNumEntries(start_ikey, end_ikey);
  }

  virtual MemTableRep::Iterator* GetIterator(Arena* arena = nullptr) override {
    return memtable_->GetIterator(arena);
  }

  virtual ~SpecialMemTableRep() override {}

 private:
170
  std::unique_ptr<MemTableRep> memtable_;
171 172 173 174 175 176 177 178 179 180 181 182 183
  int num_entries_flush_;
  int num_entries_;
};

// The factory for the hacky skip list mem table that triggers flush after
// number of entries exceeds a threshold.
class SpecialSkipListFactory : public MemTableRepFactory {
 public:
  // After number of inserts exceeds `num_entries_flush` in a mem table, trigger
  // flush.
  explicit SpecialSkipListFactory(int num_entries_flush)
      : num_entries_flush_(num_entries_flush) {}

184
  using MemTableRepFactory::CreateMemTableRep;
185
  virtual MemTableRep* CreateMemTableRep(
186
      const MemTableRep::KeyComparator& compare, Allocator* allocator,
A
Andrew Kryczka 已提交
187
      const SliceTransform* transform, Logger* /*logger*/) override {
188 189 190 191 192 193
    return new SpecialMemTableRep(
        allocator, factory_.CreateMemTableRep(compare, allocator, transform, 0),
        num_entries_flush_);
  }
  virtual const char* Name() const override { return "SkipListFactory"; }

194 195 196 197
  bool IsInsertConcurrentlySupported() const override {
    return factory_.IsInsertConcurrentlySupported();
  }

198 199 200 201 202
 private:
  SkipListFactory factory_;
  int num_entries_flush_;
};

203 204 205
// Special Env used to delay background operations
class SpecialEnv : public EnvWrapper {
 public:
206
  explicit SpecialEnv(Env* base, bool time_elapse_only_sleep = false);
207

208
  Status NewWritableFile(const std::string& f, std::unique_ptr<WritableFile>* r,
209 210 211 212
                         const EnvOptions& soptions) override {
    class SSTableFile : public WritableFile {
     private:
      SpecialEnv* env_;
213
      std::unique_ptr<WritableFile> base_;
214 215

     public:
216
      SSTableFile(SpecialEnv* env, std::unique_ptr<WritableFile>&& base)
S
sdong 已提交
217
          : env_(env), base_(std::move(base)) {}
218 219 220 221 222 223 224 225
      Status Append(const Slice& data) override {
        if (env_->table_write_callback_) {
          (*env_->table_write_callback_)();
        }
        if (env_->drop_writes_.load(std::memory_order_acquire)) {
          // Drop writes on the floor
          return Status::OK();
        } else if (env_->no_space_.load(std::memory_order_acquire)) {
226
          return Status::NoSpace("No space left on device");
227 228 229 230 231
        } else {
          env_->bytes_written_ += data.size();
          return base_->Append(data);
        }
      }
A
Aaron Gao 已提交
232
      Status PositionedAppend(const Slice& data, uint64_t offset) override {
A
Aaron Gao 已提交
233 234 235 236 237 238 239 240 241 242 243 244 245
        if (env_->table_write_callback_) {
          (*env_->table_write_callback_)();
        }
        if (env_->drop_writes_.load(std::memory_order_acquire)) {
          // Drop writes on the floor
          return Status::OK();
        } else if (env_->no_space_.load(std::memory_order_acquire)) {
          return Status::NoSpace("No space left on device");
        } else {
          env_->bytes_written_ += data.size();
          return base_->PositionedAppend(data, offset);
        }
      }
S
sdong 已提交
246
      Status Truncate(uint64_t size) override { return base_->Truncate(size); }
247 248 249 250 251 252 253
      Status RangeSync(uint64_t offset, uint64_t nbytes) override {
        Status s = base_->RangeSync(offset, nbytes);
#if !(defined NDEBUG) || !defined(OS_WIN)
        TEST_SYNC_POINT_CALLBACK("SpecialEnv::SStableFile::RangeSync", &s);
#endif  // !(defined NDEBUG) || !defined(OS_WIN)
        return s;
      }
254
      Status Close() override {
255 256
// SyncPoint is not supported in Released Windows Mode.
#if !(defined NDEBUG) || !defined(OS_WIN)
257 258 259 260 261
        // Check preallocation size
        // preallocation size is never passed to base file.
        size_t preallocation_size = preallocation_block_size();
        TEST_SYNC_POINT_CALLBACK("DBTestWritableFile.GetPreallocationStatus",
                                 &preallocation_size);
262
#endif  // !(defined NDEBUG) || !defined(OS_WIN)
263 264 265 266 267
        Status s = base_->Close();
#if !(defined NDEBUG) || !defined(OS_WIN)
        TEST_SYNC_POINT_CALLBACK("SpecialEnv::SStableFile::Close", &s);
#endif  // !(defined NDEBUG) || !defined(OS_WIN)
        return s;
268 269 270 271 272 273 274
      }
      Status Flush() override { return base_->Flush(); }
      Status Sync() override {
        ++env_->sync_counter_;
        while (env_->delay_sstable_sync_.load(std::memory_order_acquire)) {
          env_->SleepForMicroseconds(100000);
        }
275 276 277 278
        Status s;
        if (!env_->skip_fsync_) {
          s = base_->Sync();
        }
279 280 281 282
#if !(defined NDEBUG) || !defined(OS_WIN)
        TEST_SYNC_POINT_CALLBACK("SpecialEnv::SStableFile::Sync", &s);
#endif  // !(defined NDEBUG) || !defined(OS_WIN)
        return s;
283 284 285 286
      }
      void SetIOPriority(Env::IOPriority pri) override {
        base_->SetIOPriority(pri);
      }
287 288 289
      Env::IOPriority GetIOPriority() override {
        return base_->GetIOPriority();
      }
A
Aaron Gao 已提交
290 291 292
      bool use_direct_io() const override {
        return base_->use_direct_io();
      }
293 294 295
      Status Allocate(uint64_t offset, uint64_t len) override {
        return base_->Allocate(offset, len);
      }
296 297 298
    };
    class ManifestFile : public WritableFile {
     public:
299
      ManifestFile(SpecialEnv* env, std::unique_ptr<WritableFile>&& b)
S
sdong 已提交
300
          : env_(env), base_(std::move(b)) {}
301 302 303 304 305 306 307
      Status Append(const Slice& data) override {
        if (env_->manifest_write_error_.load(std::memory_order_acquire)) {
          return Status::IOError("simulated writer error");
        } else {
          return base_->Append(data);
        }
      }
308
      Status Truncate(uint64_t size) override { return base_->Truncate(size); }
309 310 311 312 313 314 315
      Status Close() override { return base_->Close(); }
      Status Flush() override { return base_->Flush(); }
      Status Sync() override {
        ++env_->sync_counter_;
        if (env_->manifest_sync_error_.load(std::memory_order_acquire)) {
          return Status::IOError("simulated sync error");
        } else {
316 317 318 319 320
          if (env_->skip_fsync_) {
            return Status::OK();
          } else {
            return base_->Sync();
          }
321 322 323
        }
      }
      uint64_t GetFileSize() override { return base_->GetFileSize(); }
324 325 326
      Status Allocate(uint64_t offset, uint64_t len) override {
        return base_->Allocate(offset, len);
      }
327 328 329

     private:
      SpecialEnv* env_;
330
      std::unique_ptr<WritableFile> base_;
331 332 333
    };
    class WalFile : public WritableFile {
     public:
334
      WalFile(SpecialEnv* env, std::unique_ptr<WritableFile>&& b)
335 336 337 338
          : env_(env), base_(std::move(b)) {
        env_->num_open_wal_file_.fetch_add(1);
      }
      virtual ~WalFile() { env_->num_open_wal_file_.fetch_add(-1); }
339
      Status Append(const Slice& data) override {
S
sdong 已提交
340 341 342 343
#if !(defined NDEBUG) || !defined(OS_WIN)
        TEST_SYNC_POINT("SpecialEnv::WalFile::Append:1");
#endif
        Status s;
344
        if (env_->log_write_error_.load(std::memory_order_acquire)) {
S
sdong 已提交
345
          s = Status::IOError("simulated writer error");
346 347 348 349 350 351
        } else {
          int slowdown =
              env_->log_write_slowdown_.load(std::memory_order_acquire);
          if (slowdown > 0) {
            env_->SleepForMicroseconds(slowdown);
          }
S
sdong 已提交
352
          s = base_->Append(data);
353
        }
S
sdong 已提交
354 355 356 357
#if !(defined NDEBUG) || !defined(OS_WIN)
        TEST_SYNC_POINT("SpecialEnv::WalFile::Append:2");
#endif
        return s;
358
      }
359
      Status Truncate(uint64_t size) override { return base_->Truncate(size); }
360 361 362 363 364 365 366 367 368 369 370 371
      Status Close() override {
// SyncPoint is not supported in Released Windows Mode.
#if !(defined NDEBUG) || !defined(OS_WIN)
        // Check preallocation size
        // preallocation size is never passed to base file.
        size_t preallocation_size = preallocation_block_size();
        TEST_SYNC_POINT_CALLBACK("DBTestWalFile.GetPreallocationStatus",
                                 &preallocation_size);
#endif  // !(defined NDEBUG) || !defined(OS_WIN)

        return base_->Close();
      }
372 373 374
      Status Flush() override { return base_->Flush(); }
      Status Sync() override {
        ++env_->sync_counter_;
375 376 377 378 379
        if (env_->skip_fsync_) {
          return Status::OK();
        } else {
          return base_->Sync();
        }
380
      }
381 382 383
      bool IsSyncThreadSafe() const override {
        return env_->is_wal_sync_thread_safe_.load();
      }
384 385 386
      Status Allocate(uint64_t offset, uint64_t len) override {
        return base_->Allocate(offset, len);
      }
387 388 389

     private:
      SpecialEnv* env_;
390
      std::unique_ptr<WritableFile> base_;
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
    class OtherFile : public WritableFile {
     public:
      OtherFile(SpecialEnv* env, std::unique_ptr<WritableFile>&& b)
          : env_(env), base_(std::move(b)) {}
      Status Append(const Slice& data) override { return base_->Append(data); }
      Status Truncate(uint64_t size) override { return base_->Truncate(size); }
      Status Close() override { return base_->Close(); }
      Status Flush() override { return base_->Flush(); }
      Status Sync() override {
        if (env_->skip_fsync_) {
          return Status::OK();
        } else {
          return base_->Sync();
        }
      }
      uint64_t GetFileSize() override { return base_->GetFileSize(); }
      Status Allocate(uint64_t offset, uint64_t len) override {
        return base_->Allocate(offset, len);
      }

     private:
      SpecialEnv* env_;
      std::unique_ptr<WritableFile> base_;
    };
416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434

    if (non_writeable_rate_.load(std::memory_order_acquire) > 0) {
      uint32_t random_number;
      {
        MutexLock l(&rnd_mutex_);
        random_number = rnd_.Uniform(100);
      }
      if (random_number < non_writeable_rate_.load()) {
        return Status::IOError("simulated random write error");
      }
    }

    new_writable_count_++;

    if (non_writable_count_.load() > 0) {
      non_writable_count_--;
      return Status::IOError("simulated write error");
    }

A
Aaron Gao 已提交
435 436 437 438 439 440 441 442
    EnvOptions optimized = soptions;
    if (strstr(f.c_str(), "MANIFEST") != nullptr ||
        strstr(f.c_str(), "log") != nullptr) {
      optimized.use_mmap_writes = false;
      optimized.use_direct_writes = false;
    }

    Status s = target()->NewWritableFile(f, r, optimized);
443 444 445 446 447 448 449
    if (s.ok()) {
      if (strstr(f.c_str(), ".sst") != nullptr) {
        r->reset(new SSTableFile(this, std::move(*r)));
      } else if (strstr(f.c_str(), "MANIFEST") != nullptr) {
        r->reset(new ManifestFile(this, std::move(*r)));
      } else if (strstr(f.c_str(), "log") != nullptr) {
        r->reset(new WalFile(this, std::move(*r)));
450 451
      } else {
        r->reset(new OtherFile(this, std::move(*r)));
452 453 454 455 456 457
      }
    }
    return s;
  }

  Status NewRandomAccessFile(const std::string& f,
458
                             std::unique_ptr<RandomAccessFile>* r,
459 460 461
                             const EnvOptions& soptions) override {
    class CountingFile : public RandomAccessFile {
     public:
462
      CountingFile(std::unique_ptr<RandomAccessFile>&& target,
463
                   anon::AtomicCounter* counter,
Y
Yi Wu 已提交
464
                   std::atomic<size_t>* bytes_read)
465 466 467
          : target_(std::move(target)),
            counter_(counter),
            bytes_read_(bytes_read) {}
468 469 470
      virtual Status Read(uint64_t offset, size_t n, Slice* result,
                          char* scratch) const override {
        counter_->Increment();
471 472 473
        Status s = target_->Read(offset, n, result, scratch);
        *bytes_read_ += result->size();
        return s;
474 475
      }

476 477 478 479 480 481
      virtual Status Prefetch(uint64_t offset, size_t n) override {
        Status s = target_->Prefetch(offset, n);
        *bytes_read_ += n;
        return s;
      }

482
     private:
483
      std::unique_ptr<RandomAccessFile> target_;
484
      anon::AtomicCounter* counter_;
Y
Yi Wu 已提交
485
      std::atomic<size_t>* bytes_read_;
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
    class RandomFailureFile : public RandomAccessFile {
     public:
      RandomFailureFile(std::unique_ptr<RandomAccessFile>&& target,
                        std::atomic<uint64_t>* failure_cnt, uint32_t fail_odd)
          : target_(std::move(target)),
            fail_cnt_(failure_cnt),
            fail_odd_(fail_odd) {}
      virtual Status Read(uint64_t offset, size_t n, Slice* result,
                          char* scratch) const override {
        if (Random::GetTLSInstance()->OneIn(fail_odd_)) {
          fail_cnt_->fetch_add(1);
          return Status::IOError("random error");
        }
        return target_->Read(offset, n, result, scratch);
      }

      virtual Status Prefetch(uint64_t offset, size_t n) override {
        return target_->Prefetch(offset, n);
      }

     private:
      std::unique_ptr<RandomAccessFile> target_;
      std::atomic<uint64_t>* fail_cnt_;
      uint32_t fail_odd_;
    };

514
    Status s = target()->NewRandomAccessFile(f, r, soptions);
515
    random_file_open_counter_++;
516 517 518 519 520 521 522 523
    if (s.ok()) {
      if (count_random_reads_) {
        r->reset(new CountingFile(std::move(*r), &random_read_counter_,
                                  &random_read_bytes_counter_));
      } else if (rand_reads_fail_odd_ > 0) {
        r->reset(new RandomFailureFile(std::move(*r), &num_reads_fails_,
                                       rand_reads_fail_odd_));
      }
524
    }
525

526 527 528
    if (s.ok() && soptions.compaction_readahead_size > 0) {
      compaction_readahead_size_ = soptions.compaction_readahead_size;
    }
529 530 531
    return s;
  }

S
Siying Dong 已提交
532
  virtual Status NewSequentialFile(const std::string& f,
533
                                   std::unique_ptr<SequentialFile>* r,
S
Siying Dong 已提交
534
                                   const EnvOptions& soptions) override {
535 536
    class CountingFile : public SequentialFile {
     public:
537
      CountingFile(std::unique_ptr<SequentialFile>&& target,
538 539 540 541 542 543 544 545 546
                   anon::AtomicCounter* counter)
          : target_(std::move(target)), counter_(counter) {}
      virtual Status Read(size_t n, Slice* result, char* scratch) override {
        counter_->Increment();
        return target_->Read(n, result, scratch);
      }
      virtual Status Skip(uint64_t n) override { return target_->Skip(n); }

     private:
547
      std::unique_ptr<SequentialFile> target_;
548 549 550 551 552 553 554 555 556 557 558 559
      anon::AtomicCounter* counter_;
    };

    Status s = target()->NewSequentialFile(f, r, soptions);
    if (s.ok() && count_sequential_reads_) {
      r->reset(new CountingFile(std::move(*r), &sequential_read_counter_));
    }
    return s;
  }

  virtual void SleepForMicroseconds(int micros) override {
    sleep_counter_.Increment();
M
Maysam Yabandeh 已提交
560
    if (no_slowdown_ || time_elapse_only_sleep_) {
561
      addon_microseconds_.fetch_add(micros);
562
    }
M
Maysam Yabandeh 已提交
563
    if (!no_slowdown_) {
564 565 566 567
      target()->SleepForMicroseconds(micros);
    }
  }

568 569 570 571 572 573 574 575 576 577 578 579
  void MockSleepForMicroseconds(int64_t micros) {
    sleep_counter_.Increment();
    assert(no_slowdown_);
    addon_microseconds_.fetch_add(micros);
  }

  void MockSleepForSeconds(int64_t seconds) {
    sleep_counter_.Increment();
    assert(no_slowdown_);
    addon_microseconds_.fetch_add(seconds * 1000000);
  }

580
  virtual Status GetCurrentTime(int64_t* unix_time) override {
581
    Status s;
582 583 584
    if (time_elapse_only_sleep_) {
      *unix_time = maybe_starting_time_;
    } else {
585 586
      s = target()->GetCurrentTime(unix_time);
    }
587
    if (s.ok()) {
588 589
      // mock microseconds elapsed to seconds of time
      *unix_time += addon_microseconds_.load() / 1000000;
590 591 592 593
    }
    return s;
  }

594 595 596 597 598
  virtual uint64_t NowCPUNanos() override {
    now_cpu_count_.fetch_add(1);
    return target()->NowCPUNanos();
  }

599
  virtual uint64_t NowNanos() override {
600
    return (time_elapse_only_sleep_ ? 0 : target()->NowNanos()) +
601
           addon_microseconds_.load() * 1000;
602 603 604
  }

  virtual uint64_t NowMicros() override {
605
    return (time_elapse_only_sleep_ ? 0 : target()->NowMicros()) +
606
           addon_microseconds_.load();
607 608
  }

609 610 611 612 613
  virtual Status DeleteFile(const std::string& fname) override {
    delete_count_.fetch_add(1);
    return target()->DeleteFile(fname);
  }

614
  void SetMockSleep(bool enabled = true) { no_slowdown_ = enabled; }
615

616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633
  Status NewDirectory(const std::string& name,
                      std::unique_ptr<Directory>* result) override {
    if (!skip_fsync_) {
      return target()->NewDirectory(name, result);
    } else {
      class NoopDirectory : public Directory {
       public:
        NoopDirectory() {}
        ~NoopDirectory() {}

        Status Fsync() override { return Status::OK(); }
      };

      result->reset(new NoopDirectory());
      return Status::OK();
    }
  }

634 635 636
  // Something to return when mocking current time
  const int64_t maybe_starting_time_;

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
  Random rnd_;
  port::Mutex rnd_mutex_;  // Lock to pretect rnd_

  // sstable Sync() calls are blocked while this pointer is non-nullptr.
  std::atomic<bool> delay_sstable_sync_;

  // Drop writes on the floor while this pointer is non-nullptr.
  std::atomic<bool> drop_writes_;

  // Simulate no-space errors while this pointer is non-nullptr.
  std::atomic<bool> no_space_;

  // Simulate non-writable file system while this pointer is non-nullptr
  std::atomic<bool> non_writable_;

  // Force sync of manifest files to fail while this pointer is non-nullptr
  std::atomic<bool> manifest_sync_error_;

  // Force write to manifest files to fail while this pointer is non-nullptr
  std::atomic<bool> manifest_write_error_;

  // Force write to log files to fail while this pointer is non-nullptr
  std::atomic<bool> log_write_error_;

  // Slow down every log write, in micro-seconds.
  std::atomic<int> log_write_slowdown_;

664 665 666
  // Number of WAL files that are still open for write.
  std::atomic<int> num_open_wal_file_;

667
  bool count_random_reads_;
668 669
  uint32_t rand_reads_fail_odd_ = 0;
  std::atomic<uint64_t> num_reads_fails_;
670
  anon::AtomicCounter random_read_counter_;
Y
Yi Wu 已提交
671
  std::atomic<size_t> random_read_bytes_counter_;
672
  std::atomic<int> random_file_open_counter_;
673 674 675 676 677 678 679 680 681 682

  bool count_sequential_reads_;
  anon::AtomicCounter sequential_read_counter_;

  anon::AtomicCounter sleep_counter_;

  std::atomic<int64_t> bytes_written_;

  std::atomic<int> sync_counter_;

683 684 685
  // If true, all fsync to files and directories are skipped.
  bool skip_fsync_ = false;

686 687 688 689 690 691 692 693
  std::atomic<uint32_t> non_writeable_rate_;

  std::atomic<uint32_t> new_writable_count_;

  std::atomic<uint32_t> non_writable_count_;

  std::function<void()>* table_write_callback_;

694 695
  std::atomic<int> now_cpu_count_;

696 697
  std::atomic<int> delete_count_;

S
sdong 已提交
698
  std::atomic<bool> is_wal_sync_thread_safe_{true};
699

700
  std::atomic<size_t> compaction_readahead_size_{};
701 702 703 704 705 706 707 708 709 710

 private:  // accessing these directly is prone to error
  friend class DBTestBase;

  std::atomic<int64_t> addon_microseconds_{0};

  // Do not modify in the env of a running DB (could cause deadlock)
  std::atomic<bool> time_elapse_only_sleep_;

  bool no_slowdown_;
711 712
};

Y
Yi Wu 已提交
713
#ifndef ROCKSDB_LITE
Y
Yi Wu 已提交
714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737
class OnFileDeletionListener : public EventListener {
 public:
  OnFileDeletionListener() : matched_count_(0), expected_file_name_("") {}

  void SetExpectedFileName(const std::string file_name) {
    expected_file_name_ = file_name;
  }

  void VerifyMatchedCount(size_t expected_value) {
    ASSERT_EQ(matched_count_, expected_value);
  }

  void OnTableFileDeleted(const TableFileDeletionInfo& info) override {
    if (expected_file_name_ != "") {
      ASSERT_EQ(expected_file_name_, info.file_path);
      expected_file_name_ = "";
      matched_count_++;
    }
  }

 private:
  size_t matched_count_;
  std::string expected_file_name_;
};
Y
Yi Wu 已提交
738
#endif
Y
Yi Wu 已提交
739

740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761
// A test merge operator mimics put but also fails if one of merge operands is
// "corrupted".
class TestPutOperator : public MergeOperator {
 public:
  virtual bool FullMergeV2(const MergeOperationInput& merge_in,
                           MergeOperationOutput* merge_out) const override {
    if (merge_in.existing_value != nullptr &&
        *(merge_in.existing_value) == "corrupted") {
      return false;
    }
    for (auto value : merge_in.operand_list) {
      if (value == "corrupted") {
        return false;
      }
    }
    merge_out->existing_operand = merge_in.operand_list.back();
    return true;
  }

  virtual const char* Name() const override { return "TestPutOperator"; }
};

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
// A wrapper around Cache that can easily be extended with instrumentation,
// etc.
class CacheWrapper : public Cache {
 public:
  explicit CacheWrapper(std::shared_ptr<Cache> target)
      : target_(std::move(target)) {}

  const char* Name() const override { return target_->Name(); }

  Status Insert(const Slice& key, void* value, size_t charge,
                void (*deleter)(const Slice& key, void* value),
                Handle** handle = nullptr,
                Priority priority = Priority::LOW) override {
    return target_->Insert(key, value, charge, deleter, handle, priority);
  }

  Handle* Lookup(const Slice& key, Statistics* stats = nullptr) override {
    return target_->Lookup(key, stats);
  }

  bool Ref(Handle* handle) override { return target_->Ref(handle); }

  bool Release(Handle* handle, bool force_erase = false) override {
    return target_->Release(handle, force_erase);
  }

  void* Value(Handle* handle) override { return target_->Value(handle); }

  void Erase(const Slice& key) override { target_->Erase(key); }
  uint64_t NewId() override { return target_->NewId(); }

  void SetCapacity(size_t capacity) override { target_->SetCapacity(capacity); }

  void SetStrictCapacityLimit(bool strict_capacity_limit) override {
    target_->SetStrictCapacityLimit(strict_capacity_limit);
  }

  bool HasStrictCapacityLimit() const override {
    return target_->HasStrictCapacityLimit();
  }

  size_t GetCapacity() const override { return target_->GetCapacity(); }

  size_t GetUsage() const override { return target_->GetUsage(); }

  size_t GetUsage(Handle* handle) const override {
    return target_->GetUsage(handle);
  }

  size_t GetPinnedUsage() const override { return target_->GetPinnedUsage(); }

  size_t GetCharge(Handle* handle) const override {
    return target_->GetCharge(handle);
  }

  void ApplyToAllCacheEntries(void (*callback)(void*, size_t),
                              bool thread_safe) override {
    target_->ApplyToAllCacheEntries(callback, thread_safe);
  }

  void EraseUnRefEntries() override { target_->EraseUnRefEntries(); }

 protected:
  std::shared_ptr<Cache> target_;
};

828
class DBTestBase : public testing::Test {
Y
Yi Wu 已提交
829
 public:
830
  // Sequence of option configurations to try
Y
Yi Wu 已提交
831
  enum OptionConfig : int {
832 833 834 835 836
    kDefault = 0,
    kBlockBasedTableWithPrefixHashIndex = 1,
    kBlockBasedTableWithWholeKeyHashIndex = 2,
    kPlainTableFirstBytePrefix = 3,
    kPlainTableCappedPrefix = 4,
837 838 839 840
    kPlainTableCappedPrefixNonMmap = 5,
    kPlainTableAllBytesPrefix = 6,
    kVectorRep = 7,
    kHashLinkList = 8,
841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862
    kMergePut = 9,
    kFilter = 10,
    kFullFilterWithNewTableReaderForCompactions = 11,
    kUncompressed = 12,
    kNumLevel_3 = 13,
    kDBLogDir = 14,
    kWalDirAndMmapReads = 15,
    kManifestFileSize = 16,
    kPerfOptions = 17,
    kHashSkipList = 18,
    kUniversalCompaction = 19,
    kUniversalCompactionMultiLevel = 20,
    kCompressedBlockCache = 21,
    kInfiniteMaxOpenFiles = 22,
    kxxHashChecksum = 23,
    kFIFOCompaction = 24,
    kOptimizeFiltersForHits = 25,
    kRowCache = 26,
    kRecycleLogFiles = 27,
    kConcurrentSkipList = 28,
    kPipelinedWrite = 29,
    kConcurrentWALWrites = 30,
863 864 865 866
    kDirectIO,
    kLevelSubcompactions,
    kBlockBasedTableWithIndexRestartInterval,
    kBlockBasedTableWithPartitionedIndex,
867
    kBlockBasedTableWithPartitionedIndexFormat4,
868
    kPartitionedFilterWithNewTableReaderForCompactions,
869
    kUniversalSubcompactions,
B
Bo Hou 已提交
870
    kxxHash64Checksum,
M
Maysam Yabandeh 已提交
871
    kUnorderedWrite,
872
    // This must be the last line
873
    kEnd,
874 875 876 877 878
  };

 public:
  std::string dbname_;
  std::string alternative_wal_dir_;
S
sdong 已提交
879
  std::string alternative_db_log_dir_;
880
  MockEnv* mem_env_;
E
Ewout Prangsma 已提交
881
  Env* encrypted_env_;
882
  SpecialEnv* env_;
883
  std::shared_ptr<Env> env_guard_;
884 885 886
  DB* db_;
  std::vector<ColumnFamilyHandle*> handles_;

Y
Yi Wu 已提交
887
  int option_config_;
888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903
  Options last_options_;

  // Skip some options, as they may not be applicable to a specific test.
  // To add more skip constants, use values 4, 8, 16, etc.
  enum OptionSkip {
    kNoSkip = 0,
    kSkipDeletesFilterFirst = 1,
    kSkipUniversalCompaction = 2,
    kSkipMergePut = 4,
    kSkipPlainTable = 8,
    kSkipHashIndex = 16,
    kSkipNoSeekToLast = 32,
    kSkipFIFOCompaction = 128,
    kSkipMmapReads = 256,
  };

904 905 906 907 908 909 910
  const int kRangeDelSkipConfigs =
      // Plain tables do not support range deletions.
      kSkipPlainTable |
      // MmapReads disables the iterator pinning that RangeDelAggregator
      // requires.
      kSkipMmapReads;

S
sdong 已提交
911 912 913 914
  // `env_do_fsync` decides whether the special Env would do real
  // fsync for files and directories. Skipping fsync can speed up
  // tests, but won't cover the exact fsync logic.
  DBTestBase(const std::string path, bool env_do_fsync);
915 916 917

  ~DBTestBase();

918 919 920 921 922 923
  static std::string Key(int i) {
    char buf[100];
    snprintf(buf, sizeof(buf), "key%06d", i);
    return std::string(buf);
  }

S
sdong 已提交
924 925
  static bool ShouldSkipOptions(int option_config, int skip_mask = kNoSkip);

926 927 928 929
  // Switch to a fresh database with the next option configuration to
  // test.  Return false if there are no more configurations to test.
  bool ChangeOptions(int skip_mask = kNoSkip);

S
Siying Dong 已提交
930
  // Switch between different compaction styles.
931 932
  bool ChangeCompactOptions();

S
Siying Dong 已提交
933 934 935
  // Switch between different WAL-realted options.
  bool ChangeWalOptions();

936 937 938 939
  // Switch between different filter policy
  // Jump from kDefault to kFilter to kFullFilter
  bool ChangeFilterOptions();

940 941 942
  // Switch between different DB options for file ingestion tests.
  bool ChangeOptionsForFileIngestionTest();

943
  // Return the current option configuration.
Y
Yi Wu 已提交
944 945 946 947 948 949 950
  Options CurrentOptions(const anon::OptionsOverride& options_override =
                             anon::OptionsOverride()) const;

  Options CurrentOptions(const Options& default_options,
                         const anon::OptionsOverride& options_override =
                             anon::OptionsOverride()) const;

951
  Options GetDefaultOptions() const;
952

953 954 955 956 957
  Options GetOptions(int option_config) const {
    return GetOptions(option_config, GetDefaultOptions());
  }

  Options GetOptions(int option_config, const Options& default_options,
Y
Yi Wu 已提交
958 959
                     const anon::OptionsOverride& options_override =
                         anon::OptionsOverride()) const;
960

961
  DBImpl* dbfull() { return static_cast_with_check<DBImpl>(db_); }
962 963 964 965 966 967 968 969 970 971 972 973 974

  void CreateColumnFamilies(const std::vector<std::string>& cfs,
                            const Options& options);

  void CreateAndReopenWithCF(const std::vector<std::string>& cfs,
                             const Options& options);

  void ReopenWithColumnFamilies(const std::vector<std::string>& cfs,
                                const std::vector<Options>& options);

  void ReopenWithColumnFamilies(const std::vector<std::string>& cfs,
                                const Options& options);

S
sdong 已提交
975 976
  Status TryReopenWithColumnFamilies(const std::vector<std::string>& cfs,
                                     const std::vector<Options>& options);
977 978 979 980 981 982 983 984 985 986

  Status TryReopenWithColumnFamilies(const std::vector<std::string>& cfs,
                                     const Options& options);

  void Reopen(const Options& options);

  void Close();

  void DestroyAndReopen(const Options& options);

987
  void Destroy(const Options& options, bool delete_cf_paths = false);
988 989 990 991 992

  Status ReadOnlyReopen(const Options& options);

  Status TryReopen(const Options& options);

A
Aaron Gao 已提交
993 994
  bool IsDirectIOSupported();

E
Ewout Prangsma 已提交
995 996
  bool IsMemoryMappedAccessSupported() const;

997 998
  Status Flush(int cf = 0);

Y
Yanqin Jin 已提交
999 1000
  Status Flush(const std::vector<int>& cf_ids);

1001 1002 1003 1004 1005
  Status Put(const Slice& k, const Slice& v, WriteOptions wo = WriteOptions());

  Status Put(int cf, const Slice& k, const Slice& v,
             WriteOptions wo = WriteOptions());

1006 1007 1008 1009 1010 1011
  Status Merge(const Slice& k, const Slice& v,
               WriteOptions wo = WriteOptions());

  Status Merge(int cf, const Slice& k, const Slice& v,
               WriteOptions wo = WriteOptions());

1012 1013 1014 1015
  Status Delete(const std::string& k);

  Status Delete(int cf, const std::string& k);

A
Andres Noetzli 已提交
1016 1017 1018 1019
  Status SingleDelete(const std::string& k);

  Status SingleDelete(int cf, const std::string& k);

1020 1021
  bool SetPreserveDeletesSequenceNumber(SequenceNumber sn);

1022 1023 1024 1025 1026
  std::string Get(const std::string& k, const Snapshot* snapshot = nullptr);

  std::string Get(int cf, const std::string& k,
                  const Snapshot* snapshot = nullptr);

1027 1028
  Status Get(const std::string& k, PinnableSlice* v);

A
Anand Ananthabhotla 已提交
1029 1030
  std::vector<std::string> MultiGet(std::vector<int> cfs,
                                    const std::vector<std::string>& k,
1031 1032
                                    const Snapshot* snapshot,
                                    const bool batched);
A
Anand Ananthabhotla 已提交
1033

1034 1035 1036
  std::vector<std::string> MultiGet(const std::vector<std::string>& k,
                                    const Snapshot* snapshot = nullptr);

1037 1038 1039 1040
  uint64_t GetNumSnapshots();

  uint64_t GetTimeOldestSnapshots();

1041 1042
  uint64_t GetSequenceOldestSnapshots();

1043 1044 1045 1046 1047 1048
  // Return a string that contains all key,value pairs in order,
  // formatted like "(k1->v1)(k2->v2)".
  std::string Contents(int cf = 0);

  std::string AllEntriesFor(const Slice& user_key, int cf = 0);

1049
#ifndef ROCKSDB_LITE
1050 1051 1052 1053 1054 1055
  int NumSortedRuns(int cf = 0);

  uint64_t TotalSize(int cf = 0);

  uint64_t SizeAtLevel(int level);

V
Vasili Svirski 已提交
1056
  size_t TotalLiveFiles(int cf = 0);
1057

1058 1059 1060 1061
  size_t CountLiveFiles();

  int NumTableFilesAtLevel(int level, int cf = 0);

1062 1063
  double CompressionRatioAtLevel(int level, int cf = 0);

1064
  int TotalTableFiles(int cf = 0, int levels = -1);
Y
Yi Wu 已提交
1065
#endif  // ROCKSDB_LITE
1066

1067 1068
  std::vector<uint64_t> GetBlobFileNumbers();

1069 1070 1071 1072 1073
  // Return spread of files per level
  std::string FilesPerLevel(int cf = 0);

  size_t CountFiles();

1074 1075 1076 1077 1078 1079 1080
  Status CountFiles(size_t* count);

  Status Size(const Slice& start, const Slice& limit, uint64_t* size) {
    return Size(start, limit, 0, size);
  }

  Status Size(const Slice& start, const Slice& limit, int cf, uint64_t* size);
1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098

  void Compact(int cf, const Slice& start, const Slice& limit,
               uint32_t target_path_id);

  void Compact(int cf, const Slice& start, const Slice& limit);

  void Compact(const Slice& start, const Slice& limit);

  // Do n memtable compactions, each of which produces an sstable
  // covering the range [small,large].
  void MakeTables(int n, const std::string& small, const std::string& large,
                  int cf = 0);

  // Prevent pushing of new sstables into deeper levels by adding
  // tables that cover a specified range to all levels.
  void FillLevels(const std::string& smallest, const std::string& largest,
                  int cf);

1099 1100
  void MoveFilesToLevel(int level, int cf = 0);

Y
Yi Wu 已提交
1101
#ifndef ROCKSDB_LITE
1102
  void DumpFileCounts(const char* label);
Y
Yi Wu 已提交
1103
#endif  // ROCKSDB_LITE
1104 1105 1106

  std::string DumpSSTableList();

1107 1108
  static void GetSstFiles(Env* env, std::string path,
                          std::vector<std::string>* files);
D
dyniusz 已提交
1109

1110 1111 1112 1113 1114
  int GetSstFileCount(std::string path);

  // this will generate non-overlapping files since it keeps increasing key_idx
  void GenerateNewFile(Random* rnd, int* key_idx, bool nowait = false);

1115 1116
  void GenerateNewFile(int fd, Random* rnd, int* key_idx, bool nowait = false);

1117
  static const int kNumKeysByGenerateNewRandomFile;
K
krad 已提交
1118
  static const int KNumKeysByGenerateNewFile = 100;
1119

1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135
  void GenerateNewRandomFile(Random* rnd, bool nowait = false);

  std::string IterStatus(Iterator* iter);

  Options OptionsForLogIterTest();

  std::string DummyString(size_t len, char c = 'a');

  void VerifyIterLast(std::string expected_key, int cf = 0);

  // Used to test InplaceUpdate

  // If previous value is nullptr or delta is > than previous value,
  //   sets newValue with delta
  // If previous value is not empty,
  //   updates previous value with 'b' string of previous value size - 1.
S
sdong 已提交
1136 1137 1138
  static UpdateStatus updateInPlaceSmallerSize(char* prevValue,
                                               uint32_t* prevSize, Slice delta,
                                               std::string* newValue);
1139

S
sdong 已提交
1140 1141 1142 1143
  static UpdateStatus updateInPlaceSmallerVarintSize(char* prevValue,
                                                     uint32_t* prevSize,
                                                     Slice delta,
                                                     std::string* newValue);
1144

S
sdong 已提交
1145 1146 1147
  static UpdateStatus updateInPlaceLargerSize(char* prevValue,
                                              uint32_t* prevSize, Slice delta,
                                              std::string* newValue);
1148

S
sdong 已提交
1149 1150
  static UpdateStatus updateInPlaceNoAction(char* prevValue, uint32_t* prevSize,
                                            Slice delta, std::string* newValue);
1151 1152 1153 1154 1155 1156

  // Utility method to test InplaceUpdate
  void validateNumberOfEntries(int numValues, int cf = 0);

  void CopyFile(const std::string& source, const std::string& destination,
                uint64_t size = 0);
1157

1158 1159 1160
  Status GetAllDataFiles(const FileType file_type,
                         std::unordered_map<std::string, uint64_t>* sst_files,
                         uint64_t* total_size = nullptr);
Y
Yi Wu 已提交
1161 1162 1163

  std::vector<std::uint64_t> ListTableFiles(Env* env, const std::string& path);

I
Islam AbdelRahman 已提交
1164 1165 1166 1167
  void VerifyDBFromMap(
      std::map<std::string, std::string> true_data,
      size_t* total_reads_res = nullptr, bool tailing_iter = false,
      std::map<std::string, Status> status = std::map<std::string, Status>());
1168 1169 1170

  void VerifyDBInternal(
      std::vector<std::pair<std::string, std::string>> true_data);
1171

Y
Yi Wu 已提交
1172 1173 1174 1175 1176 1177 1178 1179
#ifndef ROCKSDB_LITE
  uint64_t GetNumberOfSstFilesForColumnFamily(DB* db,
                                              std::string column_family_name);
#endif  // ROCKSDB_LITE

  uint64_t TestGetTickerCount(const Options& options, Tickers ticker_type) {
    return options.statistics->getTickerCount(ticker_type);
  }
1180 1181 1182 1183 1184

  uint64_t TestGetAndResetTickerCount(const Options& options,
                                      Tickers ticker_type) {
    return options.statistics->getAndResetTickerCount(ticker_type);
  }
1185 1186 1187 1188 1189 1190 1191 1192 1193

  // Note: reverting this setting within the same test run is not yet
  // supported
  void SetTimeElapseOnlySleepOnReopen(DBOptions* options);

 private:  // Prone to error on direct use
  void MaybeInstallTimeElapseOnlySleep(const DBOptions& options);

  bool time_elapse_only_sleep_on_reopen_ = false;
1194 1195
};

1196
}  // namespace ROCKSDB_NAMESPACE