Integrate block cache tracer into db_impl (#5433)

Summary:
This PR integrates the block cache tracer class into db_impl.cc.
db_impl.cc contains a member variable of AtomicBlockCacheTraceWriter class and passes its reference to the block_based_table_reader.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/5433

Differential Revision: D15728016

Pulled By: HaoyuHuang

fbshipit-source-id: 23d5659e8c82d556833dcc1a5558aac8c1f7db71

TARGETS

@@ -222,6 +222,7 @@ cpp_library(
         "tools/ldb_cmd.cc",
         "tools/ldb_tool.cc",
         "tools/sst_dump_tool.cc",
+        "trace_replay/block_cache_tracer.cc",
         "trace_replay/trace_replay.cc",
         "util/bloom.cc",
         "util/build_version.cc",
@@ -314,6 +315,7 @@ cpp_library(
         "test_util/fault_injection_test_env.cc",
         "test_util/testharness.cc",
         "test_util/testutil.cc",
+        "tools/block_cache_trace_analyzer.cc",
         "tools/trace_analyzer_tool.cc",
         "utilities/cassandra/test_utils.cc",
     ],
@@ -329,6 +331,7 @@ cpp_library(
     name = "rocksdb_tools_lib",
     srcs = [
         "test_util/testutil.cc",
+        "tools/block_cache_trace_analyzer.cc",
         "tools/db_bench_tool.cc",
         "tools/trace_analyzer_tool.cc",
     ],
@@ -383,6 +386,16 @@ ROCKS_TESTS = [
         "table/block_based/block_based_filter_block_test.cc",
         "serial",
     ],
+    [
+        "block_cache_trace_analyzer_test",
+        "tools/block_cache_trace_analyzer_test.cc",
+        "serial",
+    ],
+    [
+        "block_cache_tracer_test",
+        "trace_replay/block_cache_tracer_test.cc",
+        "serial",
+    ],
     [
         "block_test",
         "table/block_based/block_test.cc",

db/column_family.cc

@@ -405,7 +405,8 @@ ColumnFamilyData::ColumnFamilyData(
     uint32_t id, const std::string& name, Version* _dummy_versions,
     Cache* _table_cache, WriteBufferManager* write_buffer_manager,
     const ColumnFamilyOptions& cf_options, const ImmutableDBOptions& db_options,
-    const EnvOptions& env_options, ColumnFamilySet* column_family_set)
+    const EnvOptions& env_options, ColumnFamilySet* column_family_set,
+    BlockCacheTracer* const block_cache_tracer)
     : id_(id),
       name_(name),
       dummy_versions_(_dummy_versions),
@@ -445,7 +446,8 @@ ColumnFamilyData::ColumnFamilyData(
   if (_dummy_versions != nullptr) {
     internal_stats_.reset(
         new InternalStats(ioptions_.num_levels, db_options.env, this));
-    table_cache_.reset(new TableCache(ioptions_, env_options, _table_cache));
+    table_cache_.reset(new TableCache(ioptions_, env_options, _table_cache,
+                                      block_cache_tracer));
     if (ioptions_.compaction_style == kCompactionStyleLevel) {
       compaction_picker_.reset(
           new LevelCompactionPicker(ioptions_, &internal_comparator_));
@@ -1254,18 +1256,20 @@ ColumnFamilySet::ColumnFamilySet(const std::string& dbname,
                                  const EnvOptions& env_options,
                                  Cache* table_cache,
                                  WriteBufferManager* write_buffer_manager,
-                                 WriteController* write_controller)
+                                 WriteController* write_controller,
+                                 BlockCacheTracer* const block_cache_tracer)
     : max_column_family_(0),
-      dummy_cfd_(new ColumnFamilyData(0, "", nullptr, nullptr, nullptr,
-                                      ColumnFamilyOptions(), *db_options,
-                                      env_options, nullptr)),
+      dummy_cfd_(new ColumnFamilyData(
+          0, "", nullptr, nullptr, nullptr, ColumnFamilyOptions(), *db_options,
+          env_options, nullptr, block_cache_tracer)),
       default_cfd_cache_(nullptr),
       db_name_(dbname),
       db_options_(db_options),
       env_options_(env_options),
       table_cache_(table_cache),
       write_buffer_manager_(write_buffer_manager),
-      write_controller_(write_controller) {
+      write_controller_(write_controller),
+      block_cache_tracer_(block_cache_tracer) {
   // initialize linked list
   dummy_cfd_->prev_ = dummy_cfd_;
   dummy_cfd_->next_ = dummy_cfd_;
@@ -1333,7 +1337,7 @@ ColumnFamilyData* ColumnFamilySet::CreateColumnFamily(
   assert(column_families_.find(name) == column_families_.end());
   ColumnFamilyData* new_cfd = new ColumnFamilyData(
       id, name, dummy_versions, table_cache_, write_buffer_manager_, options,
-      *db_options_, env_options_, this);
+      *db_options_, env_options_, this, block_cache_tracer_);
   column_families_.insert({name, id});
   column_family_data_.insert({id, new_cfd});
   max_column_family_ = std::max(max_column_family_, id);

db/column_family.h

@@ -24,6 +24,7 @@
 #include "rocksdb/db.h"
 #include "rocksdb/env.h"
 #include "rocksdb/options.h"
+#include "trace_replay/block_cache_tracer.h"
 #include "util/thread_local.h"
 
 namespace rocksdb {
@@ -46,7 +47,7 @@ struct SuperVersionContext;
 
 extern const double kIncSlowdownRatio;
 // This file contains a list of data structures for managing column family
-// level metadata. 
+// level metadata.
 //
 // The basic relationships among classes declared here are illustrated as
 // following:
@@ -94,7 +95,7 @@ extern const double kIncSlowdownRatio;
 //         |             |    |   1.a    |  |   1.b    |  |   1.c    |
 //         +-------------+    |          |  |          |  |          |
 //                            +----------+  +----------+  +----------+
-// 
+//
 // DBImpl keeps a ColumnFamilySet, which references to all column families by
 // pointing to respective ColumnFamilyData object of each column family.
 // This is how DBImpl can list and operate on all the column families.
@@ -151,7 +152,7 @@ extern const double kIncSlowdownRatio;
 // contains Version B, memtable a and memtable b; SuperVersion1 contains
 // Version B and memtable b (mutable). As a result, Version B and memtable b
 // are prevented from being destroyed or deleted.
-  
+
 // ColumnFamilyHandleImpl is the class that clients use to access different
 // column families. It has non-trivial destructor, which gets called when client
 // is done using the column family
@@ -504,7 +505,8 @@ class ColumnFamilyData {
                    const ColumnFamilyOptions& options,
                    const ImmutableDBOptions& db_options,
                    const EnvOptions& env_options,
-                   ColumnFamilySet* column_family_set);
+                   ColumnFamilySet* column_family_set,
+                   BlockCacheTracer* const block_cache_tracer);
 
   uint32_t id_;
   const std::string name_;
@@ -632,7 +634,8 @@ class ColumnFamilySet {
                   const ImmutableDBOptions* db_options,
                   const EnvOptions& env_options, Cache* table_cache,
                   WriteBufferManager* write_buffer_manager,
-                  WriteController* write_controller);
+                  WriteController* write_controller,
+                  BlockCacheTracer* const block_cache_tracer);
   ~ColumnFamilySet();
 
   ColumnFamilyData* GetDefault() const;
@@ -691,6 +694,7 @@ class ColumnFamilySet {
   Cache* table_cache_;
   WriteBufferManager* write_buffer_manager_;
   WriteController* write_controller_;
+  BlockCacheTracer* const block_cache_tracer_;
 };
 
 // We use ColumnFamilyMemTablesImpl to provide WriteBatch a way to access

db/compaction/compaction_job_test.cc

@@ -77,7 +77,8 @@ class CompactionJobTest : public testing::Test {
         write_buffer_manager_(db_options_.db_write_buffer_size),
         versions_(new VersionSet(dbname_, &db_options_, env_options_,
                                  table_cache_.get(), &write_buffer_manager_,
-                                 &write_controller_)),
+                                 &write_controller_,
+                                 /*block_cache_tracer=*/nullptr)),
         shutting_down_(false),
         preserve_deletes_seqnum_(0),
         mock_table_factory_(new mock::MockTableFactory()),
@@ -200,7 +201,8 @@ class CompactionJobTest : public testing::Test {
     EXPECT_OK(env_->CreateDirIfMissing(dbname_));
     versions_.reset(new VersionSet(dbname_, &db_options_, env_options_,
                                    table_cache_.get(), &write_buffer_manager_,
-                                   &write_controller_));
+                                   &write_controller_,
+                                   /*block_cache_tracer=*/nullptr));
     compaction_job_stats_.Reset();
 
     VersionEdit new_db;

db/db_impl/db_impl.cc

@@ -237,7 +237,7 @@ DBImpl::DBImpl(const DBOptions& options, const std::string& dbname,
 
   versions_.reset(new VersionSet(dbname_, &immutable_db_options_, env_options_,
                                  table_cache_.get(), write_buffer_manager_,
-                                 &write_controller_));
+                                 &write_controller_, &block_cache_tracer_));
   column_family_memtables_.reset(
       new ColumnFamilyMemTablesImpl(versions_->GetColumnFamilySet()));
 
@@ -3924,6 +3924,18 @@ Status DBImpl::EndTrace() {
   return s;
 }
 
+Status DBImpl::StartBlockCacheTrace(
+    const TraceOptions& trace_options,
+    std::unique_ptr<TraceWriter>&& trace_writer) {
+  return block_cache_tracer_.StartTrace(env_, trace_options,
+                                        std::move(trace_writer));
+}
+
+Status DBImpl::EndBlockCacheTrace() {
+  block_cache_tracer_.EndTrace();
+  return Status::OK();
+}
+
 Status DBImpl::TraceIteratorSeek(const uint32_t& cf_id, const Slice& key) {
   Status s;
   if (tracer_) {

db/db_impl/db_impl.h

@@ -40,7 +40,6 @@
 #include "db/wal_manager.h"
 #include "db/write_controller.h"
 #include "db/write_thread.h"
-#include "db/memtable_list.h"
 #include "logging/event_logger.h"
 #include "monitoring/instrumented_mutex.h"
 #include "options/db_options.h"
@@ -53,6 +52,7 @@
 #include "rocksdb/transaction_log.h"
 #include "rocksdb/write_buffer_manager.h"
 #include "table/scoped_arena_iterator.h"
+#include "trace_replay/block_cache_tracer.h"
 #include "trace_replay/trace_replay.h"
 #include "util/autovector.h"
 #include "util/hash.h"
@@ -331,6 +331,14 @@ class DBImpl : public DB {
   using DB::EndTrace;
   virtual Status EndTrace() override;
 
+  using DB::StartBlockCacheTrace;
+  Status StartBlockCacheTrace(
+      const TraceOptions& options,
+      std::unique_ptr<TraceWriter>&& trace_writer) override;
+
+  using DB::EndBlockCacheTrace;
+  Status EndBlockCacheTrace() override;
+
   using DB::GetPropertiesOfAllTables;
   virtual Status GetPropertiesOfAllTables(
       ColumnFamilyHandle* column_family,
@@ -832,6 +840,7 @@ class DBImpl : public DB {
       recovered_transactions_;
   std::unique_ptr<Tracer> tracer_;
   InstrumentedMutex trace_mutex_;
+  BlockCacheTracer block_cache_tracer_;
 
   // State below is protected by mutex_
   // With two_write_queues enabled, some of the variables that accessed during

db/db_wal_test.cc

@@ -838,7 +838,8 @@ class RecoveryTestHelper {
 
     versions.reset(new VersionSet(test->dbname_, &db_options, env_options,
                                   table_cache.get(), &write_buffer_manager,
-                                  &write_controller));
+                                  &write_controller,
+                                  /*block_cache_tracer=*/nullptr));
 
     wal_manager.reset(new WalManager(db_options, env_options));
 

db/flush_job_test.cc

@@ -35,7 +35,8 @@ class FlushJobTest : public testing::Test {
         write_buffer_manager_(db_options_.db_write_buffer_size),
         versions_(new VersionSet(dbname_, &db_options_, env_options_,
                                  table_cache_.get(), &write_buffer_manager_,
-                                 &write_controller_)),
+                                 &write_controller_,
+                                 /*block_cache_tracer=*/nullptr)),
         shutting_down_(false),
         mock_table_factory_(new mock::MockTableFactory()) {
     EXPECT_OK(env_->CreateDirIfMissing(dbname_));

db/memtable_list_test.cc

@@ -100,7 +100,7 @@ class MemTableListTest : public testing::Test {
 
     VersionSet versions(dbname, &immutable_db_options, env_options,
                         table_cache.get(), &write_buffer_manager,
-                        &write_controller);
+                        &write_controller, /*block_cache_tracer=*/nullptr);
     std::vector<ColumnFamilyDescriptor> cf_descs;
     cf_descs.emplace_back(kDefaultColumnFamilyName, ColumnFamilyOptions());
     cf_descs.emplace_back("one", ColumnFamilyOptions());
@@ -144,7 +144,7 @@ class MemTableListTest : public testing::Test {
 
     VersionSet versions(dbname, &immutable_db_options, env_options,
                         table_cache.get(), &write_buffer_manager,
-                        &write_controller);
+                        &write_controller, /*block_cache_tracer=*/nullptr);
     std::vector<ColumnFamilyDescriptor> cf_descs;
     cf_descs.emplace_back(kDefaultColumnFamilyName, ColumnFamilyOptions());
     cf_descs.emplace_back("one", ColumnFamilyOptions());

db/repair.cc

@@ -109,11 +109,13 @@ class Repairer {
             // once.
             NewLRUCache(10, db_options_.table_cache_numshardbits)),
         table_cache_(new TableCache(default_cf_iopts_, env_options_,
-                                    raw_table_cache_.get())),
+                                    raw_table_cache_.get(),
+                                    /*block_cache_tracer=*/nullptr)),
         wb_(db_options_.db_write_buffer_size),
         wc_(db_options_.delayed_write_rate),
         vset_(dbname_, &immutable_db_options_, env_options_,
-              raw_table_cache_.get(), &wb_, &wc_),
+              raw_table_cache_.get(), &wb_, &wc_,
+              /*block_cache_tracer=*/nullptr),
         next_file_number_(1),
         db_lock_(nullptr) {
     for (const auto& cfd : column_families) {

db/table_cache.cc

@@ -68,11 +68,13 @@ void AppendVarint64(IterKey* key, uint64_t v) {
 }  // namespace
 
 TableCache::TableCache(const ImmutableCFOptions& ioptions,
-                       const EnvOptions& env_options, Cache* const cache)
+                       const EnvOptions& env_options, Cache* const cache,
+                       BlockCacheTracer* const block_cache_tracer)
     : ioptions_(ioptions),
       env_options_(env_options),
       cache_(cache),
-      immortal_tables_(false) {
+      immortal_tables_(false),
+      block_cache_tracer_(block_cache_tracer) {
   if (ioptions_.row_cache) {
     // If the same cache is shared by multiple instances, we need to
     // disambiguate its entries.
@@ -125,7 +127,7 @@ Status TableCache::GetTableReader(
     s = ioptions_.table_factory->NewTableReader(
         TableReaderOptions(ioptions_, prefix_extractor, env_options,
                            internal_comparator, skip_filters, immortal_tables_,
-                           level, fd.largest_seqno),
+                           level, fd.largest_seqno, block_cache_tracer_),
         std::move(file_reader), fd.GetFileSize(), table_reader,
         prefetch_index_and_filter_in_cache);
     TEST_SYNC_POINT("TableCache::GetTableReader:0");

db/table_cache.h

@@ -23,6 +23,7 @@
 #include "rocksdb/options.h"
 #include "rocksdb/table.h"
 #include "table/table_reader.h"
+#include "trace_replay/block_cache_tracer.h"
 
 namespace rocksdb {
 
@@ -48,7 +49,8 @@ class HistogramImpl;
 class TableCache {
  public:
   TableCache(const ImmutableCFOptions& ioptions,
-             const EnvOptions& storage_options, Cache* cache);
+             const EnvOptions& storage_options, Cache* cache,
+             BlockCacheTracer* const block_cache_tracer);
   ~TableCache();
 
   // Return an iterator for the specified file number (the corresponding
@@ -188,6 +190,7 @@ class TableCache {
   Cache* const cache_;
   std::string row_cache_id_;
   bool immortal_tables_;
+  BlockCacheTracer* const block_cache_tracer_;
 };
 
 }  // namespace rocksdb

db/version_set.cc

@@ -3342,10 +3342,11 @@ VersionSet::VersionSet(const std::string& dbname,
                        const ImmutableDBOptions* _db_options,
                        const EnvOptions& storage_options, Cache* table_cache,
                        WriteBufferManager* write_buffer_manager,
-                       WriteController* write_controller)
-    : column_family_set_(
-          new ColumnFamilySet(dbname, _db_options, storage_options, table_cache,
-                              write_buffer_manager, write_controller)),
+                       WriteController* write_controller,
+                       BlockCacheTracer* const block_cache_tracer)
+    : column_family_set_(new ColumnFamilySet(
+          dbname, _db_options, storage_options, table_cache,
+          write_buffer_manager, write_controller, block_cache_tracer)),
       env_(_db_options->env),
       dbname_(dbname),
       db_options_(_db_options),
@@ -3359,7 +3360,8 @@ VersionSet::VersionSet(const std::string& dbname,
       prev_log_number_(0),
       current_version_number_(0),
       manifest_file_size_(0),
-      env_options_(storage_options) {}
+      env_options_(storage_options),
+      block_cache_tracer_(block_cache_tracer) {}
 
 void CloseTables(void* ptr, size_t) {
   TableReader* table_reader = reinterpret_cast<TableReader*>(ptr);
@@ -4445,7 +4447,8 @@ Status VersionSet::ReduceNumberOfLevels(const std::string& dbname,
                                         options->table_cache_numshardbits));
   WriteController wc(options->delayed_write_rate);
   WriteBufferManager wb(options->db_write_buffer_size);
-  VersionSet versions(dbname, &db_options, env_options, tc.get(), &wb, &wc);
+  VersionSet versions(dbname, &db_options, env_options, tc.get(), &wb, &wc,
+                      /*block_cache_tracer=*/nullptr);
   Status status;
 
   std::vector<ColumnFamilyDescriptor> dummy;
@@ -5200,7 +5203,8 @@ ReactiveVersionSet::ReactiveVersionSet(const std::string& dbname,
                                        WriteBufferManager* write_buffer_manager,
                                        WriteController* write_controller)
     : VersionSet(dbname, _db_options, _env_options, table_cache,
-                 write_buffer_manager, write_controller) {}
+                 write_buffer_manager, write_controller,
+                 /*block_cache_tracer=*/nullptr) {}
 
 ReactiveVersionSet::~ReactiveVersionSet() {}
 

db/version_set.h

@@ -46,6 +46,7 @@
 #include "rocksdb/env.h"
 #include "table/get_context.h"
 #include "table/multiget_context.h"
+#include "trace_replay/block_cache_tracer.h"
 
 namespace rocksdb {
 
@@ -777,7 +778,8 @@ class VersionSet {
   VersionSet(const std::string& dbname, const ImmutableDBOptions* db_options,
              const EnvOptions& env_options, Cache* table_cache,
              WriteBufferManager* write_buffer_manager,
-             WriteController* write_controller);
+             WriteController* write_controller,
+             BlockCacheTracer* const block_cache_tracer);
   virtual ~VersionSet();
 
   // Apply *edit to the current version to form a new descriptor that
@@ -1125,6 +1127,8 @@ class VersionSet {
   // env options for all reads and writes except compactions
   EnvOptions env_options_;
 
+  BlockCacheTracer* const block_cache_tracer_;
+
  private:
   // No copying allowed
   VersionSet(const VersionSet&);

db/version_set_test.cc

@@ -618,7 +618,8 @@ class VersionSetTestBase {
         write_buffer_manager_(db_options_.db_write_buffer_size),
         versions_(new VersionSet(dbname_, &db_options_, env_options_,
                                  table_cache_.get(), &write_buffer_manager_,
-                                 &write_controller_)),
+                                 &write_controller_,
+                                 /*block_cache_tracer=*/nullptr)),
         reactive_versions_(std::make_shared<ReactiveVersionSet>(
             dbname_, &db_options_, env_options_, table_cache_.get(),
             &write_buffer_manager_, &write_controller_)),

db/wal_manager_test.cc

@@ -50,7 +50,8 @@ class WalManagerTest : public testing::Test {
 
     versions_.reset(new VersionSet(dbname_, &db_options_, env_options_,
                                    table_cache_.get(), &write_buffer_manager_,
-                                   &write_controller_));
+                                   &write_controller_,
+                                   /*block_cache_tracer=*/nullptr));
 
     wal_manager_.reset(new WalManager(db_options_, env_options_));
   }

include/rocksdb/db.h

@@ -1317,6 +1317,17 @@ class DB {
   virtual Status EndTrace() {
     return Status::NotSupported("EndTrace() is not implemented.");
   }
+
+  // Trace block cache accesses. Use EndBlockCacheTrace() to stop tracing.
+  virtual Status StartBlockCacheTrace(
+      const TraceOptions& /*options*/,
+      std::unique_ptr<TraceWriter>&& /*trace_writer*/) {
+    return Status::NotSupported("StartBlockCacheTrace() is not implemented.");
+  }
+
+  virtual Status EndBlockCacheTrace() {
+    return Status::NotSupported("EndBlockCacheTrace() is not implemented.");
+  }
 #endif  // ROCKSDB_LITE
 
   // Needed for StackableDB

include/rocksdb/utilities/stackable_db.h

@@ -315,6 +315,16 @@ class StackableDB : public DB {
     db_->GetColumnFamilyMetaData(column_family, cf_meta);
   }
 
+  using DB::StartBlockCacheTrace;
+  Status StartBlockCacheTrace(
+      const TraceOptions& options,
+      std::unique_ptr<TraceWriter>&& trace_writer) override {
+    return db_->StartBlockCacheTrace(options, std::move(trace_writer));
+  }
+
+  using DB::EndBlockCacheTrace;
+  Status EndBlockCacheTrace() override { return db_->EndBlockCacheTrace(); }
+
 #endif  // ROCKSDB_LITE
 
   virtual Status GetLiveFiles(std::vector<std::string>& vec, uint64_t* mfs,

table/block_based/block_based_table_factory.cc

@@ -198,7 +198,8 @@ Status BlockBasedTableFactory::NewTableReader(
       file_size, table_reader, table_reader_options.prefix_extractor,
       prefetch_index_and_filter_in_cache, table_reader_options.skip_filters,
       table_reader_options.level, table_reader_options.immortal,
-      table_reader_options.largest_seqno, &tail_prefetch_stats_);
+      table_reader_options.largest_seqno, &tail_prefetch_stats_,
+      table_reader_options.block_cache_tracer);
 }
 
 TableBuilder* BlockBasedTableFactory::NewTableBuilder(

table/block_based/block_based_table_reader.cc

@@ -1020,19 +1020,17 @@ Slice BlockBasedTable::GetCacheKey(const char* cache_key_prefix,
   return Slice(cache_key, static_cast<size_t>(end - cache_key));
 }
 
-Status BlockBasedTable::Open(const ImmutableCFOptions& ioptions,
-                             const EnvOptions& env_options,
-                             const BlockBasedTableOptions& table_options,
-                             const InternalKeyComparator& internal_comparator,
-                             std::unique_ptr<RandomAccessFileReader>&& file,
-                             uint64_t file_size,
-                             std::unique_ptr<TableReader>* table_reader,
-                             const SliceTransform* prefix_extractor,
-                             const bool prefetch_index_and_filter_in_cache,
-                             const bool skip_filters, const int level,
-                             const bool immortal_table,
-                             const SequenceNumber largest_seqno,
-                             TailPrefetchStats* tail_prefetch_stats) {
+Status BlockBasedTable::Open(
+    const ImmutableCFOptions& ioptions, const EnvOptions& env_options,
+    const BlockBasedTableOptions& table_options,
+    const InternalKeyComparator& internal_comparator,
+    std::unique_ptr<RandomAccessFileReader>&& file, uint64_t file_size,
+    std::unique_ptr<TableReader>* table_reader,
+    const SliceTransform* prefix_extractor,
+    const bool prefetch_index_and_filter_in_cache, const bool skip_filters,
+    const int level, const bool immortal_table,
+    const SequenceNumber largest_seqno, TailPrefetchStats* tail_prefetch_stats,
+    BlockCacheTracer* const block_cache_tracer) {
   table_reader->reset();
 
   Status s;
@@ -1082,7 +1080,8 @@ Status BlockBasedTable::Open(const ImmutableCFOptions& ioptions,
   rep->internal_prefix_transform.reset(
       new InternalKeySliceTransform(prefix_extractor));
   SetupCacheKeyPrefix(rep);
-  std::unique_ptr<BlockBasedTable> new_table(new BlockBasedTable(rep));
+  std::unique_ptr<BlockBasedTable> new_table(
+      new BlockBasedTable(rep, block_cache_tracer));
 
   // page cache options
   rep->persistent_cache_options =

table/block_based/block_based_table_reader.h

@@ -35,6 +35,7 @@
 #include "table/table_properties_internal.h"
 #include "table/table_reader.h"
 #include "table/two_level_iterator.h"
+#include "trace_replay/block_cache_tracer.h"
 #include "util/coding.h"
 #include "util/file_reader_writer.h"
 #include "util/user_comparator_wrapper.h"
@@ -108,7 +109,8 @@ class BlockBasedTable : public TableReader {
                      bool skip_filters = false, int level = -1,
                      const bool immortal_table = false,
                      const SequenceNumber largest_seqno = 0,
-                     TailPrefetchStats* tail_prefetch_stats = nullptr);
+                     TailPrefetchStats* tail_prefetch_stats = nullptr,
+                     BlockCacheTracer* const block_cache_tracer = nullptr);
 
   bool PrefixMayMatch(const Slice& internal_key,
                       const ReadOptions& read_options,
@@ -239,11 +241,13 @@ class BlockBasedTable : public TableReader {
 
  protected:
   Rep* rep_;
-  explicit BlockBasedTable(Rep* rep) : rep_(rep) {}
+  explicit BlockBasedTable(Rep* rep, BlockCacheTracer* const block_cache_tracer)
+      : rep_(rep), block_cache_tracer_(block_cache_tracer) {}
 
  private:
   friend class MockedBlockBasedTable;
   static std::atomic<uint64_t> next_cache_key_id_;
+  BlockCacheTracer* const block_cache_tracer_;
 
   void UpdateCacheHitMetrics(BlockType block_type, GetContext* get_context,
                              size_t usage) const;

table/block_based/partitioned_filter_block_test.cc

@@ -23,7 +23,8 @@ std::map<uint64_t, Slice> slices;
 
 class MockedBlockBasedTable : public BlockBasedTable {
  public:
-  explicit MockedBlockBasedTable(Rep* rep) : BlockBasedTable(rep) {
+  explicit MockedBlockBasedTable(Rep* rep)
+      : BlockBasedTable(rep, /*block_cache_tracer=*/nullptr) {
     // Initialize what Open normally does as much as necessary for the test
     rep->cache_key_prefix_size = 10;
   }

table/table_builder.h

@@ -18,6 +18,7 @@
 #include "options/cf_options.h"
 #include "rocksdb/options.h"
 #include "rocksdb/table_properties.h"
+#include "trace_replay/block_cache_tracer.h"
 #include "util/file_reader_writer.h"
 
 namespace rocksdb {
@@ -32,10 +33,12 @@ struct TableReaderOptions {
                      const EnvOptions& _env_options,
                      const InternalKeyComparator& _internal_comparator,
                      bool _skip_filters = false, bool _immortal = false,
-                     int _level = -1)
+                     int _level = -1,
+                     BlockCacheTracer* const _block_cache_tracer = nullptr)
       : TableReaderOptions(_ioptions, _prefix_extractor, _env_options,
                            _internal_comparator, _skip_filters, _immortal,
-                           _level, 0 /* _largest_seqno */) {}
+                           _level, 0 /* _largest_seqno */,
+                           _block_cache_tracer) {}
 
   // @param skip_filters Disables loading/accessing the filter block
   TableReaderOptions(const ImmutableCFOptions& _ioptions,
@@ -43,7 +46,8 @@ struct TableReaderOptions {
                      const EnvOptions& _env_options,
                      const InternalKeyComparator& _internal_comparator,
                      bool _skip_filters, bool _immortal, int _level,
-                     SequenceNumber _largest_seqno)
+                     SequenceNumber _largest_seqno,
+                     BlockCacheTracer* const _block_cache_tracer)
       : ioptions(_ioptions),
         prefix_extractor(_prefix_extractor),
         env_options(_env_options),
@@ -51,7 +55,8 @@ struct TableReaderOptions {
         skip_filters(_skip_filters),
         immortal(_immortal),
         level(_level),
-        largest_seqno(_largest_seqno) {}
+        largest_seqno(_largest_seqno),
+        block_cache_tracer(_block_cache_tracer) {}
 
   const ImmutableCFOptions& ioptions;
   const SliceTransform* prefix_extractor;
@@ -65,6 +70,7 @@ struct TableReaderOptions {
   int level;
   // largest seqno in the table
   SequenceNumber largest_seqno;
+  BlockCacheTracer* const block_cache_tracer;
 };
 
 struct TableBuilderOptions {

tools/ldb_cmd.cc

@@ -954,7 +954,8 @@ void DumpManifestFile(Options options, std::string file, bool verbose, bool hex,
   WriteController wc(options.delayed_write_rate);
   WriteBufferManager wb(options.db_write_buffer_size);
   ImmutableDBOptions immutable_db_options(options);
-  VersionSet versions(dbname, &immutable_db_options, sopt, tc.get(), &wb, &wc);
+  VersionSet versions(dbname, &immutable_db_options, sopt, tc.get(), &wb, &wc,
+                      /*block_cache_tracer=*/nullptr);
   Status s = versions.DumpManifest(options, file, verbose, hex, json);
   if (!s.ok()) {
     printf("Error in processing file %s %s\n", file.c_str(),
@@ -1664,7 +1665,8 @@ Status ReduceDBLevelsCommand::GetOldNumOfLevels(Options& opt,
   const InternalKeyComparator cmp(opt.comparator);
   WriteController wc(opt.delayed_write_rate);
   WriteBufferManager wb(opt.db_write_buffer_size);
-  VersionSet versions(db_path_, &db_options, soptions, tc.get(), &wb, &wc);
+  VersionSet versions(db_path_, &db_options, soptions, tc.get(), &wb, &wc,
+                      /*block_cache_tracer=*/nullptr);
   std::vector<ColumnFamilyDescriptor> dummy;
   ColumnFamilyDescriptor dummy_descriptor(kDefaultColumnFamilyName,
                                           ColumnFamilyOptions(opt));

trace_replay/block_cache_tracer.cc

@@ -23,30 +23,29 @@ bool ShouldTraceReferencedKey(const BlockCacheTraceRecord& record) {
           record.caller == BlockCacheLookupCaller::kUserMGet);
 }
 
-BlockCacheTraceWriter::BlockCacheTraceWriter(
-    Env* env, const TraceOptions& trace_options,
-    std::unique_ptr<TraceWriter>&& trace_writer)
-    : env_(env),
-      trace_options_(trace_options),
-      trace_writer_(std::move(trace_writer)) {}
-
-bool BlockCacheTraceWriter::ShouldTrace(
-    const BlockCacheTraceRecord& record) const {
-  if (trace_options_.sampling_frequency == 0 ||
-      trace_options_.sampling_frequency == 1) {
+bool ShouldTrace(const BlockCacheTraceRecord& record,
+                 const TraceOptions& trace_options) {
+  if (trace_options.sampling_frequency == 0 ||
+      trace_options.sampling_frequency == 1) {
     return true;
   }
   // We use spatial downsampling so that we have a complete access history for a
   // block.
   const uint64_t hash = GetSliceNPHash64(Slice(record.block_key));
-  return hash % trace_options_.sampling_frequency == 0;
+  return hash % trace_options.sampling_frequency == 0;
 }
 
+BlockCacheTraceWriter::BlockCacheTraceWriter(
+    Env* env, const TraceOptions& trace_options,
+    std::unique_ptr<TraceWriter>&& trace_writer)
+    : env_(env),
+      trace_options_(trace_options),
+      trace_writer_(std::move(trace_writer)) {}
+
 Status BlockCacheTraceWriter::WriteBlockAccess(
     const BlockCacheTraceRecord& record) {
   uint64_t trace_file_size = trace_writer_->GetFileSize();
-  if (trace_file_size > trace_options_.max_trace_file_size ||
-      !ShouldTrace(record)) {
+  if (trace_file_size > trace_options_.max_trace_file_size) {
     return Status::OK();
   }
   Trace trace;
@@ -68,7 +67,6 @@ Status BlockCacheTraceWriter::WriteBlockAccess(
   }
   std::string encoded_trace;
   TracerHelper::EncodeTrace(trace, &encoded_trace);
-  InstrumentedMutexLock lock_guard(&trace_writer_mutex_);
   return trace_writer_->Write(encoded_trace);
 }
 
@@ -81,7 +79,6 @@ Status BlockCacheTraceWriter::WriteHeader() {
   PutFixed32(&trace.payload, kMinorVersion);
   std::string encoded_trace;
   TracerHelper::EncodeTrace(trace, &encoded_trace);
-  InstrumentedMutexLock lock_guard(&trace_writer_mutex_);
   return trace_writer_->Write(encoded_trace);
 }
 
@@ -216,4 +213,41 @@ Status BlockCacheTraceReader::ReadAccess(BlockCacheTraceRecord* record) {
   return Status::OK();
 }
 
+BlockCacheTracer::BlockCacheTracer() { writer_.store(nullptr); }
+
+BlockCacheTracer::~BlockCacheTracer() { EndTrace(); }
+
+Status BlockCacheTracer::StartTrace(
+    Env* env, const TraceOptions& trace_options,
+    std::unique_ptr<TraceWriter>&& trace_writer) {
+  InstrumentedMutexLock lock_guard(&trace_writer_mutex_);
+  if (writer_.load()) {
+    return Status::OK();
+  }
+  trace_options_ = trace_options;
+  writer_.store(
+      new BlockCacheTraceWriter(env, trace_options, std::move(trace_writer)));
+  return writer_.load()->WriteHeader();
+}
+
+void BlockCacheTracer::EndTrace() {
+  InstrumentedMutexLock lock_guard(&trace_writer_mutex_);
+  if (!writer_.load()) {
+    return;
+  }
+  delete writer_.load();
+  writer_.store(nullptr);
+}
+
+Status BlockCacheTracer::WriteBlockAccess(const BlockCacheTraceRecord& record) {
+  if (!writer_.load() || !ShouldTrace(record, trace_options_)) {
+    return Status::OK();
+  }
+  InstrumentedMutexLock lock_guard(&trace_writer_mutex_);
+  if (!writer_.load()) {
+    return Status::OK();
+  }
+  return writer_.load()->WriteBlockAccess(record);
+}
+
 }  // namespace rocksdb

trace_replay/block_cache_tracer.h

@@ -5,6 +5,8 @@
 
 #pragma once
 
+#include <atomic>
+
 #include "monitoring/instrumented_mutex.h"
 #include "rocksdb/env.h"
 #include "rocksdb/options.h"
@@ -101,13 +103,9 @@ class BlockCacheTraceWriter {
   Status WriteHeader();
 
  private:
-  bool ShouldTrace(const BlockCacheTraceRecord& record) const;
-
   Env* env_;
   TraceOptions trace_options_;
   std::unique_ptr<TraceWriter> trace_writer_;
-  /*Mutex to protect trace_writer_ */
-  InstrumentedMutex trace_writer_mutex_;
 };
 
 // BlockCacheTraceReader helps read the trace file generated by
@@ -130,4 +128,33 @@ class BlockCacheTraceReader {
   std::unique_ptr<TraceReader> trace_reader_;
 };
 
+// A block cache tracer. It downsamples the accesses according to
+// trace_options and uses BlockCacheTraceWriter to write the access record to
+// the trace file.
+class BlockCacheTracer {
+ public:
+  BlockCacheTracer();
+  ~BlockCacheTracer();
+  // No copy and move.
+  BlockCacheTracer(const BlockCacheTracer&) = delete;
+  BlockCacheTracer& operator=(const BlockCacheTracer&) = delete;
+  BlockCacheTracer(BlockCacheTracer&&) = delete;
+  BlockCacheTracer& operator=(BlockCacheTracer&&) = delete;
+
+  // Start writing block cache accesses to the trace_writer.
+  Status StartTrace(Env* env, const TraceOptions& trace_options,
+                    std::unique_ptr<TraceWriter>&& trace_writer);
+
+  // Stop writing block cache accesses to the trace_writer.
+  void EndTrace();
+
+  Status WriteBlockAccess(const BlockCacheTraceRecord& record);
+
+ private:
+  TraceOptions trace_options_;
+  // A mutex protects the writer_.
+  InstrumentedMutex trace_writer_mutex_;
+  std::atomic<BlockCacheTraceWriter*> writer_;
+};
+
 }  // namespace rocksdb

trace_replay/block_cache_tracer_test.cc

@@ -80,6 +80,26 @@ class BlockCacheTracerTest : public testing::Test {
     }
   }
 
+  BlockCacheTraceRecord GenerateAccessRecord() {
+    uint32_t key_id = 0;
+    BlockCacheTraceRecord record;
+    record.block_type = TraceType::kBlockTraceDataBlock;
+    record.block_size = kBlockSize;
+    record.block_key = kBlockKeyPrefix + std::to_string(key_id);
+    record.access_timestamp = env_->NowMicros();
+    record.cf_id = kCFId;
+    record.cf_name = kDefaultColumnFamilyName;
+    record.caller = GetCaller(key_id);
+    record.level = kLevel;
+    record.sst_fd_number = kSSTFDNumber + key_id;
+    record.is_cache_hit = Boolean::kFalse;
+    record.no_insert = Boolean::kFalse;
+    record.referenced_key = kRefKeyPrefix + std::to_string(key_id);
+    record.is_referenced_key_exist_in_block = Boolean::kTrue;
+    record.num_keys_in_block = kNumKeysInBlock;
+    return record;
+  }
+
   void VerifyAccess(BlockCacheTraceReader* reader, uint32_t from_key_id,
                     TraceType block_type, uint32_t nblocks) {
     assert(reader);
@@ -118,6 +138,88 @@ class BlockCacheTracerTest : public testing::Test {
   std::string test_path_;
 };
 
+TEST_F(BlockCacheTracerTest, AtomicWriteBeforeStartTrace) {
+  BlockCacheTraceRecord record = GenerateAccessRecord();
+  {
+    TraceOptions trace_opt;
+    std::unique_ptr<TraceWriter> trace_writer;
+    ASSERT_OK(NewFileTraceWriter(env_, env_options_, trace_file_path_,
+                                 &trace_writer));
+    BlockCacheTracer writer;
+    // The record should be written to the trace_file since StartTrace is not
+    // called.
+    ASSERT_OK(writer.WriteBlockAccess(record));
+    ASSERT_OK(env_->FileExists(trace_file_path_));
+  }
+  {
+    // Verify trace file contains nothing.
+    std::unique_ptr<TraceReader> trace_reader;
+    ASSERT_OK(NewFileTraceReader(env_, env_options_, trace_file_path_,
+                                 &trace_reader));
+    BlockCacheTraceReader reader(std::move(trace_reader));
+    BlockCacheTraceHeader header;
+    ASSERT_NOK(reader.ReadHeader(&header));
+  }
+}
+
+TEST_F(BlockCacheTracerTest, AtomicWrite) {
+  BlockCacheTraceRecord record = GenerateAccessRecord();
+  {
+    TraceOptions trace_opt;
+    std::unique_ptr<TraceWriter> trace_writer;
+    ASSERT_OK(NewFileTraceWriter(env_, env_options_, trace_file_path_,
+                                 &trace_writer));
+    BlockCacheTracer writer;
+    ASSERT_OK(writer.StartTrace(env_, trace_opt, std::move(trace_writer)));
+    ASSERT_OK(writer.WriteBlockAccess(record));
+    ASSERT_OK(env_->FileExists(trace_file_path_));
+  }
+  {
+    // Verify trace file contains one record.
+    std::unique_ptr<TraceReader> trace_reader;
+    ASSERT_OK(NewFileTraceReader(env_, env_options_, trace_file_path_,
+                                 &trace_reader));
+    BlockCacheTraceReader reader(std::move(trace_reader));
+    BlockCacheTraceHeader header;
+    ASSERT_OK(reader.ReadHeader(&header));
+    ASSERT_EQ(kMajorVersion, header.rocksdb_major_version);
+    ASSERT_EQ(kMinorVersion, header.rocksdb_minor_version);
+    VerifyAccess(&reader, 0, TraceType::kBlockTraceDataBlock, 1);
+    ASSERT_NOK(reader.ReadAccess(&record));
+  }
+}
+
+TEST_F(BlockCacheTracerTest, AtomicNoWriteAfterEndTrace) {
+  BlockCacheTraceRecord record = GenerateAccessRecord();
+  {
+    TraceOptions trace_opt;
+    std::unique_ptr<TraceWriter> trace_writer;
+    ASSERT_OK(NewFileTraceWriter(env_, env_options_, trace_file_path_,
+                                 &trace_writer));
+    BlockCacheTracer writer;
+    ASSERT_OK(writer.StartTrace(env_, trace_opt, std::move(trace_writer)));
+    ASSERT_OK(writer.WriteBlockAccess(record));
+    writer.EndTrace();
+    // Write the record again. This time the record should not be written since
+    // EndTrace is called.
+    ASSERT_OK(writer.WriteBlockAccess(record));
+    ASSERT_OK(env_->FileExists(trace_file_path_));
+  }
+  {
+    // Verify trace file contains one record.
+    std::unique_ptr<TraceReader> trace_reader;
+    ASSERT_OK(NewFileTraceReader(env_, env_options_, trace_file_path_,
+                                 &trace_reader));
+    BlockCacheTraceReader reader(std::move(trace_reader));
+    BlockCacheTraceHeader header;
+    ASSERT_OK(reader.ReadHeader(&header));
+    ASSERT_EQ(kMajorVersion, header.rocksdb_major_version);
+    ASSERT_EQ(kMinorVersion, header.rocksdb_minor_version);
+    VerifyAccess(&reader, 0, TraceType::kBlockTraceDataBlock, 1);
+    ASSERT_NOK(reader.ReadAccess(&record));
+  }
+}
+
 TEST_F(BlockCacheTracerTest, MixedBlocks) {
   {
     // Generate a trace file containing a mix of blocks.