db_test_util.h 38.2 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 "file/filename.h"
26 27
#include "rocksdb/cache.h"
#include "rocksdb/compaction_filter.h"
A
agiardullo 已提交
28
#include "rocksdb/convenience.h"
29 30 31 32 33
#include "rocksdb/db.h"
#include "rocksdb/env.h"
#include "rocksdb/filter_policy.h"
#include "rocksdb/options.h"
#include "rocksdb/slice.h"
34
#include "rocksdb/sst_file_writer.h"
Y
Yi Wu 已提交
35
#include "rocksdb/statistics.h"
36 37 38
#include "rocksdb/table.h"
#include "rocksdb/utilities/checkpoint.h"
#include "table/mock_table.h"
S
sdong 已提交
39
#include "table/scoped_arena_iterator.h"
40 41
#include "test_util/sync_point.h"
#include "test_util/testharness.h"
42 43 44
#include "util/cast_util.h"
#include "util/compression.h"
#include "util/mutexlock.h"
45
#include "util/string_util.h"
46 47
#include "utilities/merge_operators.h"

48
namespace ROCKSDB_NAMESPACE {
49
class MockEnv;
50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73

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();
74 75
      cond_count_.TimedWait(now + /*1s*/ 1 * 1000 * 1000);
      if (env_->NowMicros() - start > /*10s*/ 10 * 1000 * 1000) {
76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100
        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
101 102
  // These will be used only if filter_policy is set
  bool partition_filters = false;
M
Maysam Yabandeh 已提交
103
  uint64_t metadata_block_size = 1024;
104 105 106 107 108 109 110

  // 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
111 112
enum SkipPolicy { kSkipNone = 0, kSkipNoSnapshot = 1, kSkipNoPrefix = 2 };

113 114 115
// Special Env used to delay background operations
class SpecialEnv : public EnvWrapper {
 public:
116
  explicit SpecialEnv(Env* base, bool time_elapse_only_sleep = false);
117

118
  Status NewWritableFile(const std::string& f, std::unique_ptr<WritableFile>* r,
119 120 121 122
                         const EnvOptions& soptions) override {
    class SSTableFile : public WritableFile {
     private:
      SpecialEnv* env_;
123
      std::unique_ptr<WritableFile> base_;
124 125

     public:
126
      SSTableFile(SpecialEnv* env, std::unique_ptr<WritableFile>&& base)
S
sdong 已提交
127
          : env_(env), base_(std::move(base)) {}
128 129 130 131 132 133 134 135
      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)) {
136
          return Status::NoSpace("No space left on device");
137 138 139 140 141
        } else {
          env_->bytes_written_ += data.size();
          return base_->Append(data);
        }
      }
142 143 144 145 146
      Status Append(
          const Slice& data,
          const DataVerificationInfo& /* verification_info */) override {
        return Append(data);
      }
A
Aaron Gao 已提交
147
      Status PositionedAppend(const Slice& data, uint64_t offset) override {
A
Aaron Gao 已提交
148 149 150 151 152 153 154 155 156 157 158 159 160
        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);
        }
      }
161 162 163 164 165
      Status PositionedAppend(
          const Slice& data, uint64_t offset,
          const DataVerificationInfo& /* verification_info */) override {
        return PositionedAppend(data, offset);
      }
S
sdong 已提交
166
      Status Truncate(uint64_t size) override { return base_->Truncate(size); }
167 168 169 170 171 172 173
      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;
      }
174
      Status Close() override {
175 176
// SyncPoint is not supported in Released Windows Mode.
#if !(defined NDEBUG) || !defined(OS_WIN)
177 178 179 180 181
        // 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);
182
#endif  // !(defined NDEBUG) || !defined(OS_WIN)
183 184 185 186 187
        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;
188 189 190 191 192 193 194
      }
      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);
        }
195 196 197 198
        Status s;
        if (!env_->skip_fsync_) {
          s = base_->Sync();
        }
199 200 201 202
#if !(defined NDEBUG) || !defined(OS_WIN)
        TEST_SYNC_POINT_CALLBACK("SpecialEnv::SStableFile::Sync", &s);
#endif  // !(defined NDEBUG) || !defined(OS_WIN)
        return s;
203 204 205 206
      }
      void SetIOPriority(Env::IOPriority pri) override {
        base_->SetIOPriority(pri);
      }
207 208 209
      Env::IOPriority GetIOPriority() override {
        return base_->GetIOPriority();
      }
A
Aaron Gao 已提交
210 211 212
      bool use_direct_io() const override {
        return base_->use_direct_io();
      }
213 214 215
      Status Allocate(uint64_t offset, uint64_t len) override {
        return base_->Allocate(offset, len);
      }
216 217 218
      size_t GetUniqueId(char* id, size_t max_size) const override {
        return base_->GetUniqueId(id, max_size);
      }
219 220 221
    };
    class ManifestFile : public WritableFile {
     public:
222
      ManifestFile(SpecialEnv* env, std::unique_ptr<WritableFile>&& b)
S
sdong 已提交
223
          : env_(env), base_(std::move(b)) {}
224 225 226 227 228 229 230
      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);
        }
      }
231 232 233 234 235 236
      Status Append(
          const Slice& data,
          const DataVerificationInfo& /*verification_info*/) override {
        return Append(data);
      }

237
      Status Truncate(uint64_t size) override { return base_->Truncate(size); }
238 239 240 241 242 243 244
      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 {
245 246 247 248 249
          if (env_->skip_fsync_) {
            return Status::OK();
          } else {
            return base_->Sync();
          }
250 251 252
        }
      }
      uint64_t GetFileSize() override { return base_->GetFileSize(); }
253 254 255
      Status Allocate(uint64_t offset, uint64_t len) override {
        return base_->Allocate(offset, len);
      }
256 257 258

     private:
      SpecialEnv* env_;
259
      std::unique_ptr<WritableFile> base_;
260 261 262
    };
    class WalFile : public WritableFile {
     public:
263
      WalFile(SpecialEnv* env, std::unique_ptr<WritableFile>&& b)
264 265 266 267
          : env_(env), base_(std::move(b)) {
        env_->num_open_wal_file_.fetch_add(1);
      }
      virtual ~WalFile() { env_->num_open_wal_file_.fetch_add(-1); }
268
      Status Append(const Slice& data) override {
S
sdong 已提交
269 270 271 272
#if !(defined NDEBUG) || !defined(OS_WIN)
        TEST_SYNC_POINT("SpecialEnv::WalFile::Append:1");
#endif
        Status s;
273
        if (env_->log_write_error_.load(std::memory_order_acquire)) {
S
sdong 已提交
274
          s = Status::IOError("simulated writer error");
275 276 277 278 279 280
        } else {
          int slowdown =
              env_->log_write_slowdown_.load(std::memory_order_acquire);
          if (slowdown > 0) {
            env_->SleepForMicroseconds(slowdown);
          }
S
sdong 已提交
281
          s = base_->Append(data);
282
        }
S
sdong 已提交
283 284 285 286
#if !(defined NDEBUG) || !defined(OS_WIN)
        TEST_SYNC_POINT("SpecialEnv::WalFile::Append:2");
#endif
        return s;
287
      }
288 289 290 291 292
      Status Append(
          const Slice& data,
          const DataVerificationInfo& /* verification_info */) override {
        return Append(data);
      }
293
      Status Truncate(uint64_t size) override { return base_->Truncate(size); }
J
Jay Zhuang 已提交
294 295 296 297 298 299
      void PrepareWrite(size_t offset, size_t len) override {
        base_->PrepareWrite(offset, len);
      }
      void SetPreallocationBlockSize(size_t size) override {
        base_->SetPreallocationBlockSize(size);
      }
300 301 302 303
      Status Close() override {
// SyncPoint is not supported in Released Windows Mode.
#if !(defined NDEBUG) || !defined(OS_WIN)
        // Check preallocation size
J
Jay Zhuang 已提交
304 305
        size_t block_size, last_allocated_block;
        base_->GetPreallocationStatus(&block_size, &last_allocated_block);
306
        TEST_SYNC_POINT_CALLBACK("DBTestWalFile.GetPreallocationStatus",
J
Jay Zhuang 已提交
307
                                 &block_size);
308 309 310 311
#endif  // !(defined NDEBUG) || !defined(OS_WIN)

        return base_->Close();
      }
312 313 314
      Status Flush() override { return base_->Flush(); }
      Status Sync() override {
        ++env_->sync_counter_;
315
        if (env_->corrupt_in_sync_) {
316
          EXPECT_OK(Append(std::string(33000, ' ')));
317 318
          return Status::IOError("Ingested Sync Failure");
        }
319 320 321 322 323
        if (env_->skip_fsync_) {
          return Status::OK();
        } else {
          return base_->Sync();
        }
324
      }
325 326 327
      bool IsSyncThreadSafe() const override {
        return env_->is_wal_sync_thread_safe_.load();
      }
328 329 330
      Status Allocate(uint64_t offset, uint64_t len) override {
        return base_->Allocate(offset, len);
      }
331 332 333

     private:
      SpecialEnv* env_;
334
      std::unique_ptr<WritableFile> base_;
335
    };
336 337 338 339 340
    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); }
341 342 343 344 345
      Status Append(
          const Slice& data,
          const DataVerificationInfo& /*verification_info*/) override {
        return Append(data);
      }
346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364
      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_;
    };
365

366 367 368 369 370
    if (no_file_overwrite_.load(std::memory_order_acquire) &&
        target()->FileExists(f).ok()) {
      return Status::NotSupported("SpecialEnv::no_file_overwrite_ is true.");
    }

371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388
    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
389 390 391 392 393 394 395 396
    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);
397 398 399 400 401 402 403
    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)));
404 405
      } else {
        r->reset(new OtherFile(this, std::move(*r)));
406 407 408 409 410 411
      }
    }
    return s;
  }

  Status NewRandomAccessFile(const std::string& f,
412
                             std::unique_ptr<RandomAccessFile>* r,
413 414 415
                             const EnvOptions& soptions) override {
    class CountingFile : public RandomAccessFile {
     public:
416
      CountingFile(std::unique_ptr<RandomAccessFile>&& target,
417
                   anon::AtomicCounter* counter,
Y
Yi Wu 已提交
418
                   std::atomic<size_t>* bytes_read)
419 420 421
          : target_(std::move(target)),
            counter_(counter),
            bytes_read_(bytes_read) {}
422 423 424
      virtual Status Read(uint64_t offset, size_t n, Slice* result,
                          char* scratch) const override {
        counter_->Increment();
425 426 427
        Status s = target_->Read(offset, n, result, scratch);
        *bytes_read_ += result->size();
        return s;
428 429
      }

430 431 432 433 434 435
      virtual Status Prefetch(uint64_t offset, size_t n) override {
        Status s = target_->Prefetch(offset, n);
        *bytes_read_ += n;
        return s;
      }

436
     private:
437
      std::unique_ptr<RandomAccessFile> target_;
438
      anon::AtomicCounter* counter_;
Y
Yi Wu 已提交
439
      std::atomic<size_t>* bytes_read_;
440 441
    };

442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467
    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_;
    };

468
    Status s = target()->NewRandomAccessFile(f, r, soptions);
469
    random_file_open_counter_++;
470 471 472 473 474 475 476 477
    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_));
      }
478
    }
479

480 481 482
    if (s.ok() && soptions.compaction_readahead_size > 0) {
      compaction_readahead_size_ = soptions.compaction_readahead_size;
    }
483 484 485
    return s;
  }

S
Siying Dong 已提交
486
  virtual Status NewSequentialFile(const std::string& f,
487
                                   std::unique_ptr<SequentialFile>* r,
S
Siying Dong 已提交
488
                                   const EnvOptions& soptions) override {
489 490
    class CountingFile : public SequentialFile {
     public:
491
      CountingFile(std::unique_ptr<SequentialFile>&& target,
492 493 494 495 496 497 498 499 500
                   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:
501
      std::unique_ptr<SequentialFile> target_;
502 503 504 505 506 507 508 509 510 511 512 513
      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
514
    if (no_slowdown_ || time_elapse_only_sleep_) {
515
      addon_microseconds_.fetch_add(micros);
516
    }
M
Maysam Yabandeh 已提交
517
    if (!no_slowdown_) {
518 519 520 521
      target()->SleepForMicroseconds(micros);
    }
  }

522 523 524 525 526 527 528 529 530 531 532 533
  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);
  }

534
  virtual Status GetCurrentTime(int64_t* unix_time) override {
535
    Status s;
536 537 538
    if (time_elapse_only_sleep_) {
      *unix_time = maybe_starting_time_;
    } else {
539 540
      s = target()->GetCurrentTime(unix_time);
    }
541
    if (s.ok()) {
542 543
      // mock microseconds elapsed to seconds of time
      *unix_time += addon_microseconds_.load() / 1000000;
544 545 546 547
    }
    return s;
  }

548 549 550 551 552
  virtual uint64_t NowCPUNanos() override {
    now_cpu_count_.fetch_add(1);
    return target()->NowCPUNanos();
  }

553
  virtual uint64_t NowNanos() override {
554
    return (time_elapse_only_sleep_ ? 0 : target()->NowNanos()) +
555
           addon_microseconds_.load() * 1000;
556 557 558
  }

  virtual uint64_t NowMicros() override {
559
    return (time_elapse_only_sleep_ ? 0 : target()->NowMicros()) +
560
           addon_microseconds_.load();
561 562
  }

563 564 565 566 567
  virtual Status DeleteFile(const std::string& fname) override {
    delete_count_.fetch_add(1);
    return target()->DeleteFile(fname);
  }

568
  void SetMockSleep(bool enabled = true) { no_slowdown_ = enabled; }
569

570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587
  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();
    }
  }

588 589 590 591 592 593 594 595
  Status RenameFile(const std::string& src, const std::string& dest) override {
    rename_count_.fetch_add(1);
    if (rename_error_.load(std::memory_order_acquire)) {
      return Status::NotSupported("Simulated `RenameFile()` error.");
    }
    return target()->RenameFile(src, dest);
  }

596 597 598
  // Something to return when mocking current time
  const int64_t maybe_starting_time_;

599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622
  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_;

623 624 625
  // Force `RenameFile()` to fail while this pointer is non-nullptr
  std::atomic<bool> rename_error_{false};

626 627 628
  // Slow down every log write, in micro-seconds.
  std::atomic<int> log_write_slowdown_;

629 630 631
  // If true, returns Status::NotSupported for file overwrite.
  std::atomic<bool> no_file_overwrite_;

632 633 634
  // Number of WAL files that are still open for write.
  std::atomic<int> num_open_wal_file_;

635
  bool count_random_reads_;
636 637
  uint32_t rand_reads_fail_odd_ = 0;
  std::atomic<uint64_t> num_reads_fails_;
638
  anon::AtomicCounter random_read_counter_;
Y
Yi Wu 已提交
639
  std::atomic<size_t> random_read_bytes_counter_;
640
  std::atomic<int> random_file_open_counter_;
641 642 643 644 645 646 647 648 649 650

  bool count_sequential_reads_;
  anon::AtomicCounter sequential_read_counter_;

  anon::AtomicCounter sleep_counter_;

  std::atomic<int64_t> bytes_written_;

  std::atomic<int> sync_counter_;

651 652 653
  // If true, all fsync to files and directories are skipped.
  bool skip_fsync_ = false;

654 655 656
  // If true, ingest the corruption to file during sync.
  bool corrupt_in_sync_ = false;

657 658 659 660 661 662 663 664
  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_;

665 666
  std::atomic<int> now_cpu_count_;

667 668
  std::atomic<int> delete_count_;

669 670
  std::atomic<int> rename_count_{0};

S
sdong 已提交
671
  std::atomic<bool> is_wal_sync_thread_safe_{true};
672

673
  std::atomic<size_t> compaction_readahead_size_{};
674 675 676 677 678 679 680 681 682 683

 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_;
684 685
};

Y
Yi Wu 已提交
686
#ifndef ROCKSDB_LITE
Y
Yi Wu 已提交
687 688 689
class OnFileDeletionListener : public EventListener {
 public:
  OnFileDeletionListener() : matched_count_(0), expected_file_name_("") {}
690 691
  const char* Name() const override { return kClassName(); }
  static const char* kClassName() { return "OnFileDeletionListener"; }
Y
Yi Wu 已提交
692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712

  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_;
};
713 714 715

class FlushCounterListener : public EventListener {
 public:
716 717
  const char* Name() const override { return kClassName(); }
  static const char* kClassName() { return "FlushCounterListener"; }
718 719 720 721 722 723 724 725
  std::atomic<int> count{0};
  std::atomic<FlushReason> expected_flush_reason{FlushReason::kOthers};

  void OnFlushBegin(DB* /*db*/, const FlushJobInfo& flush_job_info) override {
    count++;
    ASSERT_EQ(expected_flush_reason.load(), flush_job_info.flush_reason);
  }
};
Y
Yi Wu 已提交
726
#endif
Y
Yi Wu 已提交
727

728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749
// 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"; }
};

750 751 752 753 754 755 756 757 758
// 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(); }

759
  using Cache::Insert;
760 761 762 763 764 765 766
  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);
  }

767
  using Cache::Lookup;
768 769 770 771 772 773
  Handle* Lookup(const Slice& key, Statistics* stats = nullptr) override {
    return target_->Lookup(key, stats);
  }

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

774
  using Cache::Release;
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
  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);
  }

808 809 810 811
  DeleterFn GetDeleter(Handle* handle) const override {
    return target_->GetDeleter(handle);
  }

812 813 814 815 816
  void ApplyToAllCacheEntries(void (*callback)(void*, size_t),
                              bool thread_safe) override {
    target_->ApplyToAllCacheEntries(callback, thread_safe);
  }

817 818 819 820 821 822 823
  void ApplyToAllEntries(
      const std::function<void(const Slice& key, void* value, size_t charge,
                               DeleterFn deleter)>& callback,
      const ApplyToAllEntriesOptions& opts) override {
    target_->ApplyToAllEntries(callback, opts);
  }

824 825 826 827 828 829
  void EraseUnRefEntries() override { target_->EraseUnRefEntries(); }

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

830
class DBTestBase : public testing::Test {
Y
Yi Wu 已提交
831
 public:
832
  // Sequence of option configurations to try
Y
Yi Wu 已提交
833
  enum OptionConfig : int {
834 835 836 837 838
    kDefault = 0,
    kBlockBasedTableWithPrefixHashIndex = 1,
    kBlockBasedTableWithWholeKeyHashIndex = 2,
    kPlainTableFirstBytePrefix = 3,
    kPlainTableCappedPrefix = 4,
839 840 841 842
    kPlainTableCappedPrefixNonMmap = 5,
    kPlainTableAllBytesPrefix = 6,
    kVectorRep = 7,
    kHashLinkList = 8,
843 844 845 846 847 848 849 850 851 852 853 854 855 856
    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,
857
    kXXH3Checksum = 23,
858 859 860 861 862 863 864
    kFIFOCompaction = 24,
    kOptimizeFiltersForHits = 25,
    kRowCache = 26,
    kRecycleLogFiles = 27,
    kConcurrentSkipList = 28,
    kPipelinedWrite = 29,
    kConcurrentWALWrites = 30,
865 866 867 868
    kDirectIO,
    kLevelSubcompactions,
    kBlockBasedTableWithIndexRestartInterval,
    kBlockBasedTableWithPartitionedIndex,
869
    kBlockBasedTableWithPartitionedIndexFormat4,
870
    kPartitionedFilterWithNewTableReaderForCompactions,
871
    kUniversalSubcompactions,
M
Maysam Yabandeh 已提交
872
    kUnorderedWrite,
873
    // This must be the last line
874
    kEnd,
875 876 877 878 879
  };

 public:
  std::string dbname_;
  std::string alternative_wal_dir_;
S
sdong 已提交
880
  std::string alternative_db_log_dir_;
881
  MockEnv* mem_env_;
E
Ewout Prangsma 已提交
882
  Env* encrypted_env_;
883
  SpecialEnv* env_;
884
  std::shared_ptr<Env> env_guard_;
885 886 887
  DB* db_;
  std::vector<ColumnFamilyHandle*> handles_;

Y
Yi Wu 已提交
888
  int option_config_;
889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904
  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,
  };

905 906 907 908 909 910 911
  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
912 913 914 915
  // `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);
916 917 918

  ~DBTestBase();

919 920 921 922 923 924
  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
925 926
  static bool ShouldSkipOptions(int option_config, int skip_mask = kNoSkip);

927 928 929 930
  // 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
931
  // Switch between different compaction styles.
932 933
  bool ChangeCompactOptions();

S
Siying Dong 已提交
934 935 936
  // Switch between different WAL-realted options.
  bool ChangeWalOptions();

937 938 939 940
  // Switch between different filter policy
  // Jump from kDefault to kFilter to kFullFilter
  bool ChangeFilterOptions();

941 942 943
  // Switch between different DB options for file ingestion tests.
  bool ChangeOptionsForFileIngestionTest();

944
  // Return the current option configuration.
Y
Yi Wu 已提交
945 946 947 948 949 950 951
  Options CurrentOptions(const anon::OptionsOverride& options_override =
                             anon::OptionsOverride()) const;

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

952
  Options GetDefaultOptions() const;
953

954 955 956 957 958
  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
959 960
                     const anon::OptionsOverride& options_override =
                         anon::OptionsOverride()) const;
961

962
  DBImpl* dbfull() { return static_cast_with_check<DBImpl>(db_); }
963 964 965 966 967 968 969 970 971 972 973 974 975

  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
976 977
  Status TryReopenWithColumnFamilies(const std::vector<std::string>& cfs,
                                     const std::vector<Options>& options);
978 979 980 981 982 983 984 985 986 987

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

  void Reopen(const Options& options);

  void Close();

  void DestroyAndReopen(const Options& options);

988
  void Destroy(const Options& options, bool delete_cf_paths = false);
989 990 991 992 993

  Status ReadOnlyReopen(const Options& options);

  Status TryReopen(const Options& options);

A
Aaron Gao 已提交
994 995
  bool IsDirectIOSupported();

E
Ewout Prangsma 已提交
996 997
  bool IsMemoryMappedAccessSupported() const;

998 999
  Status Flush(int cf = 0);

Y
Yanqin Jin 已提交
1000 1001
  Status Flush(const std::vector<int>& cf_ids);

1002 1003 1004 1005 1006
  Status Put(const Slice& k, const Slice& v, WriteOptions wo = WriteOptions());

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

1007 1008 1009 1010 1011 1012
  Status Merge(const Slice& k, const Slice& v,
               WriteOptions wo = WriteOptions());

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

1013 1014 1015 1016
  Status Delete(const std::string& k);

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

A
Andres Noetzli 已提交
1017 1018 1019 1020
  Status SingleDelete(const std::string& k);

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

1021 1022
  bool SetPreserveDeletesSequenceNumber(SequenceNumber sn);

1023 1024 1025 1026 1027
  std::string Get(const std::string& k, const Snapshot* snapshot = nullptr);

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

1028 1029
  Status Get(const std::string& k, PinnableSlice* v);

A
Anand Ananthabhotla 已提交
1030 1031
  std::vector<std::string> MultiGet(std::vector<int> cfs,
                                    const std::vector<std::string>& k,
1032 1033
                                    const Snapshot* snapshot,
                                    const bool batched);
A
Anand Ananthabhotla 已提交
1034

1035 1036 1037
  std::vector<std::string> MultiGet(const std::vector<std::string>& k,
                                    const Snapshot* snapshot = nullptr);

1038 1039 1040 1041
  uint64_t GetNumSnapshots();

  uint64_t GetTimeOldestSnapshots();

1042 1043
  uint64_t GetSequenceOldestSnapshots();

1044 1045 1046 1047 1048 1049
  // 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);

1050
#ifndef ROCKSDB_LITE
1051 1052 1053 1054 1055 1056
  int NumSortedRuns(int cf = 0);

  uint64_t TotalSize(int cf = 0);

  uint64_t SizeAtLevel(int level);

V
Vasili Svirski 已提交
1057
  size_t TotalLiveFiles(int cf = 0);
1058

1059 1060 1061 1062
  size_t CountLiveFiles();

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

1063 1064
  double CompressionRatioAtLevel(int level, int cf = 0);

1065
  int TotalTableFiles(int cf = 0, int levels = -1);
Y
Yi Wu 已提交
1066
#endif  // ROCKSDB_LITE
1067

1068 1069
  std::vector<uint64_t> GetBlobFileNumbers();

1070 1071 1072 1073 1074
  // Return spread of files per level
  std::string FilesPerLevel(int cf = 0);

  size_t CountFiles();

1075 1076 1077 1078 1079 1080 1081
  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);
1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099

  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);

1100 1101
  void MoveFilesToLevel(int level, int cf = 0);

Y
Yi Wu 已提交
1102
#ifndef ROCKSDB_LITE
1103
  void DumpFileCounts(const char* label);
Y
Yi Wu 已提交
1104
#endif  // ROCKSDB_LITE
1105 1106 1107

  std::string DumpSSTableList();

1108 1109
  static void GetSstFiles(Env* env, std::string path,
                          std::vector<std::string>* files);
D
dyniusz 已提交
1110

1111 1112 1113 1114 1115
  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);

1116 1117
  void GenerateNewFile(int fd, Random* rnd, int* key_idx, bool nowait = false);

1118
  static const int kNumKeysByGenerateNewRandomFile;
K
krad 已提交
1119
  static const int KNumKeysByGenerateNewFile = 100;
1120

1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136
  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
1137 1138 1139
  static UpdateStatus updateInPlaceSmallerSize(char* prevValue,
                                               uint32_t* prevSize, Slice delta,
                                               std::string* newValue);
1140

S
sdong 已提交
1141 1142 1143 1144
  static UpdateStatus updateInPlaceSmallerVarintSize(char* prevValue,
                                                     uint32_t* prevSize,
                                                     Slice delta,
                                                     std::string* newValue);
1145

S
sdong 已提交
1146 1147 1148
  static UpdateStatus updateInPlaceLargerSize(char* prevValue,
                                              uint32_t* prevSize, Slice delta,
                                              std::string* newValue);
1149

S
sdong 已提交
1150 1151
  static UpdateStatus updateInPlaceNoAction(char* prevValue, uint32_t* prevSize,
                                            Slice delta, std::string* newValue);
1152 1153 1154 1155 1156 1157

  // 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);
1158

1159 1160 1161
  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
1162 1163 1164

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

I
Islam AbdelRahman 已提交
1165 1166 1167 1168
  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>());
1169 1170 1171

  void VerifyDBInternal(
      std::vector<std::pair<std::string, std::string>> true_data);
1172

Y
Yi Wu 已提交
1173 1174 1175 1176 1177 1178 1179 1180
#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);
  }
1181 1182 1183 1184 1185

  uint64_t TestGetAndResetTickerCount(const Options& options,
                                      Tickers ticker_type) {
    return options.statistics->getAndResetTickerCount(ticker_type);
  }
1186 1187 1188 1189 1190 1191 1192 1193 1194

  // 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;
1195 1196
};

1197 1198 1199 1200
// For verifying that all files generated by current version have SST
// unique ids.
void VerifySstUniqueIds(const TablePropertiesCollection& props);

1201
}  // namespace ROCKSDB_NAMESPACE