@nxtedition/rocksdb 7.0.4 → 7.0.7

package/deps/rocksdb/rocksdb/db/blob/db_blob_basic_test.cc

@@ -5,6 +5,7 @@
 
 #include <array>
 #include <sstream>
+#include <string>
 
 #include "db/blob/blob_index.h"
 #include "db/blob/blob_log_format.h"
@@ -48,6 +49,177 @@ TEST_F(DBBlobBasicTest, GetBlob) {
                   .IsIncomplete());
 }
 
+TEST_F(DBBlobBasicTest, GetBlobFromCache) {
+  Options options = GetDefaultOptions();
+
+  LRUCacheOptions co;
+  co.capacity = 2048;
+  co.num_shard_bits = 2;
+  co.metadata_charge_policy = kDontChargeCacheMetadata;
+  auto backing_cache = NewLRUCache(co);
+
+  options.enable_blob_files = true;
+  options.blob_cache = backing_cache;
+
+  BlockBasedTableOptions block_based_options;
+  block_based_options.no_block_cache = false;
+  block_based_options.block_cache = backing_cache;
+  block_based_options.cache_index_and_filter_blocks = true;
+  options.table_factory.reset(NewBlockBasedTableFactory(block_based_options));
+
+  Reopen(options);
+
+  constexpr char key[] = "key";
+  constexpr char blob_value[] = "blob_value";
+
+  ASSERT_OK(Put(key, blob_value));
+
+  ASSERT_OK(Flush());
+
+  ReadOptions read_options;
+
+  read_options.fill_cache = false;
+
+  {
+    PinnableSlice result;
+
+    read_options.read_tier = kReadAllTier;
+    ASSERT_OK(db_->Get(read_options, db_->DefaultColumnFamily(), key, &result));
+    ASSERT_EQ(result, blob_value);
+
+    result.Reset();
+    read_options.read_tier = kBlockCacheTier;
+
+    // Try again with no I/O allowed. Since we didn't re-fill the cache, the
+    // blob itself can only be read from the blob file, so the read should
+    // return Incomplete.
+    ASSERT_TRUE(db_->Get(read_options, db_->DefaultColumnFamily(), key, &result)
+                    .IsIncomplete());
+    ASSERT_TRUE(result.empty());
+  }
+
+  read_options.fill_cache = true;
+
+  {
+    PinnableSlice result;
+
+    read_options.read_tier = kReadAllTier;
+    ASSERT_OK(db_->Get(read_options, db_->DefaultColumnFamily(), key, &result));
+    ASSERT_EQ(result, blob_value);
+
+    result.Reset();
+    read_options.read_tier = kBlockCacheTier;
+
+    // Try again with no I/O allowed. The table and the necessary blocks/blobs
+    // should already be in their respective caches.
+    ASSERT_OK(db_->Get(read_options, db_->DefaultColumnFamily(), key, &result));
+    ASSERT_EQ(result, blob_value);
+  }
+}
+
+TEST_F(DBBlobBasicTest, IterateBlobsFromCache) {
+  Options options = GetDefaultOptions();
+
+  LRUCacheOptions co;
+  co.capacity = 2048;
+  co.num_shard_bits = 2;
+  co.metadata_charge_policy = kDontChargeCacheMetadata;
+  auto backing_cache = NewLRUCache(co);
+
+  options.enable_blob_files = true;
+  options.blob_cache = backing_cache;
+
+  BlockBasedTableOptions block_based_options;
+  block_based_options.no_block_cache = false;
+  block_based_options.block_cache = backing_cache;
+  block_based_options.cache_index_and_filter_blocks = true;
+  options.table_factory.reset(NewBlockBasedTableFactory(block_based_options));
+
+  Reopen(options);
+
+  int num_blobs = 5;
+  std::vector<std::string> keys;
+  std::vector<std::string> blobs;
+
+  for (int i = 0; i < num_blobs; ++i) {
+    keys.push_back("key" + std::to_string(i));
+    blobs.push_back("blob" + std::to_string(i));
+    ASSERT_OK(Put(keys[i], blobs[i]));
+  }
+  ASSERT_OK(Flush());
+
+  ReadOptions read_options;
+
+  {
+    read_options.fill_cache = false;
+    read_options.read_tier = kReadAllTier;
+
+    std::unique_ptr<Iterator> iter(db_->NewIterator(read_options));
+    ASSERT_OK(iter->status());
+
+    int i = 0;
+    for (iter->SeekToFirst(); iter->Valid(); iter->Next()) {
+      ASSERT_OK(iter->status());
+      ASSERT_EQ(iter->key().ToString(), keys[i]);
+      ASSERT_EQ(iter->value().ToString(), blobs[i]);
+      ++i;
+    }
+    ASSERT_EQ(i, num_blobs);
+  }
+
+  {
+    read_options.fill_cache = false;
+    read_options.read_tier = kBlockCacheTier;
+
+    std::unique_ptr<Iterator> iter(db_->NewIterator(read_options));
+    ASSERT_OK(iter->status());
+
+    // Try again with no I/O allowed. Since we didn't re-fill the cache,
+    // the blob itself can only be read from the blob file, so iter->Valid()
+    // should be false.
+    iter->SeekToFirst();
+    ASSERT_NOK(iter->status());
+    ASSERT_FALSE(iter->Valid());
+  }
+
+  {
+    read_options.fill_cache = true;
+    read_options.read_tier = kReadAllTier;
+
+    std::unique_ptr<Iterator> iter(db_->NewIterator(read_options));
+    ASSERT_OK(iter->status());
+
+    // Read blobs from the file and refill the cache.
+    int i = 0;
+    for (iter->SeekToFirst(); iter->Valid(); iter->Next()) {
+      ASSERT_OK(iter->status());
+      ASSERT_EQ(iter->key().ToString(), keys[i]);
+      ASSERT_EQ(iter->value().ToString(), blobs[i]);
+      ++i;
+    }
+    ASSERT_EQ(i, num_blobs);
+  }
+
+  {
+    read_options.fill_cache = false;
+    read_options.read_tier = kBlockCacheTier;
+
+    std::unique_ptr<Iterator> iter(db_->NewIterator(read_options));
+    ASSERT_OK(iter->status());
+
+    // Try again with no I/O allowed. The table and the necessary blocks/blobs
+    // should already be in their respective caches.
+    int i = 0;
+    for (iter->SeekToFirst(); iter->Valid(); iter->Next()) {
+      ASSERT_OK(iter->status());
+      ASSERT_EQ(iter->key().ToString(), keys[i]);
+      ASSERT_EQ(iter->value().ToString(), blobs[i]);
+      ++i;
+    }
+    ASSERT_EQ(i, num_blobs);
+  }
+}
+
 TEST_F(DBBlobBasicTest, MultiGetBlobs) {
   constexpr size_t min_blob_size = 6;
 

package/deps/rocksdb/rocksdb/db/column_family.cc

@@ -17,6 +17,7 @@
 #include <vector>
 
 #include "db/blob/blob_file_cache.h"
+#include "db/blob/blob_source.h"
 #include "db/compaction/compaction_picker.h"
 #include "db/compaction/compaction_picker_fifo.h"
 #include "db/compaction/compaction_picker_level.h"
@@ -516,7 +517,7 @@ ColumnFamilyData::ColumnFamilyData(
     const ColumnFamilyOptions& cf_options, const ImmutableDBOptions& db_options,
     const FileOptions* file_options, ColumnFamilySet* column_family_set,
     BlockCacheTracer* const block_cache_tracer,
-    const std::shared_ptr<IOTracer>& io_tracer,
+    const std::shared_ptr<IOTracer>& io_tracer, const std::string& db_id,
     const std::string& db_session_id)
     : id_(id),
       name_(name),
@@ -580,6 +581,8 @@ ColumnFamilyData::ColumnFamilyData(
     blob_file_cache_.reset(
         new BlobFileCache(_table_cache, ioptions(), soptions(), id_,
                           internal_stats_->GetBlobFileReadHist(), io_tracer));
+    blob_source_.reset(new BlobSource(ioptions(), db_id, db_session_id,
+                                      blob_file_cache_.get()));
 
     if (ioptions_.compaction_style == kCompactionStyleLevel) {
       compaction_picker_.reset(
@@ -1504,13 +1507,14 @@ ColumnFamilySet::ColumnFamilySet(const std::string& dbname,
                                  WriteController* _write_controller,
                                  BlockCacheTracer* const block_cache_tracer,
                                  const std::shared_ptr<IOTracer>& io_tracer,
+                                 const std::string& db_id,
                                  const std::string& db_session_id)
     : max_column_family_(0),
       file_options_(file_options),
       dummy_cfd_(new ColumnFamilyData(
           ColumnFamilyData::kDummyColumnFamilyDataId, "", nullptr, nullptr,
           nullptr, ColumnFamilyOptions(), *db_options, &file_options_, nullptr,
-          block_cache_tracer, io_tracer, db_session_id)),
+          block_cache_tracer, io_tracer, db_id, db_session_id)),
       default_cfd_cache_(nullptr),
       db_name_(dbname),
       db_options_(db_options),
@@ -1519,6 +1523,7 @@ ColumnFamilySet::ColumnFamilySet(const std::string& dbname,
       write_controller_(_write_controller),
       block_cache_tracer_(block_cache_tracer),
       io_tracer_(io_tracer),
+      db_id_(db_id),
       db_session_id_(db_session_id) {
   // initialize linked list
   dummy_cfd_->prev_ = dummy_cfd_;
@@ -1586,7 +1591,7 @@ ColumnFamilyData* ColumnFamilySet::CreateColumnFamily(
   ColumnFamilyData* new_cfd = new ColumnFamilyData(
       id, name, dummy_versions, table_cache_, write_buffer_manager_, options,
       *db_options_, &file_options_, this, block_cache_tracer_, io_tracer_,
-      db_session_id_);
+      db_id_, db_session_id_);
   column_families_.insert({name, id});
   column_family_data_.insert({id, new_cfd});
   max_column_family_ = std::max(max_column_family_, id);

package/deps/rocksdb/rocksdb/db/column_family.h

@@ -47,6 +47,7 @@ class InstrumentedMutex;
 class InstrumentedMutexLock;
 struct SuperVersionContext;
 class BlobFileCache;
+class BlobSource;
 
 extern const double kIncSlowdownRatio;
 // This file contains a list of data structures for managing column family
@@ -376,7 +377,7 @@ class ColumnFamilyData {
                          SequenceNumber earliest_seq);
 
   TableCache* table_cache() const { return table_cache_.get(); }
-  BlobFileCache* blob_file_cache() const { return blob_file_cache_.get(); }
+  BlobSource* blob_source() const { return blob_source_.get(); }
 
   // See documentation in compaction_picker.h
   // REQUIRES: DB mutex held
@@ -539,7 +540,7 @@ class ColumnFamilyData {
                    ColumnFamilySet* column_family_set,
                    BlockCacheTracer* const block_cache_tracer,
                    const std::shared_ptr<IOTracer>& io_tracer,
-                   const std::string& db_session_id);
+                   const std::string& db_id, const std::string& db_session_id);
 
   std::vector<std::string> GetDbPaths() const;
 
@@ -563,6 +564,7 @@ class ColumnFamilyData {
 
   std::unique_ptr<TableCache> table_cache_;
   std::unique_ptr<BlobFileCache> blob_file_cache_;
+  std::unique_ptr<BlobSource> blob_source_;
 
   std::unique_ptr<InternalStats> internal_stats_;
 
@@ -673,7 +675,7 @@ class ColumnFamilySet {
                   WriteController* _write_controller,
                   BlockCacheTracer* const block_cache_tracer,
                   const std::shared_ptr<IOTracer>& io_tracer,
-                  const std::string& db_session_id);
+                  const std::string& db_id, const std::string& db_session_id);
   ~ColumnFamilySet();
 
   ColumnFamilyData* GetDefault() const;
@@ -735,6 +737,7 @@ class ColumnFamilySet {
   WriteController* write_controller_;
   BlockCacheTracer* const block_cache_tracer_;
   std::shared_ptr<IOTracer> io_tracer_;
+  const std::string& db_id_;
   std::string db_session_id_;
 };
 

package/deps/rocksdb/rocksdb/db/compaction/compaction_job.cc

@@ -2160,6 +2160,16 @@ Status CompactionJob::InstallCompactionResults(
                              stats.GetBytes());
   }
 
+  if (compaction->compaction_reason() == CompactionReason::kLevelMaxLevelSize &&
+      compaction->immutable_options()->compaction_pri == kRoundRobin) {
+    int start_level = compaction->start_level();
+    if (start_level > 0) {
+      auto vstorage = compaction->input_version()->storage_info();
+      edit->AddCompactCursor(start_level,
+                             vstorage->GetNextCompactCursor(start_level));
+    }
+  }
+
   return versions_->LogAndApply(compaction->column_family_data(),
                                 mutable_cf_options, edit, db_mutex_,
                                 db_directory_);

package/deps/rocksdb/rocksdb/db/compaction/compaction_job_test.cc

@@ -208,11 +208,11 @@ class CompactionJobTestBase : public testing::Test {
         mutable_db_options_(),
         table_cache_(NewLRUCache(50000, 16)),
         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_,
-                                 /*block_cache_tracer=*/nullptr,
-                                 /*io_tracer=*/nullptr, /*db_session_id*/ "")),
+        versions_(new VersionSet(
+            dbname_, &db_options_, env_options_, table_cache_.get(),
+            &write_buffer_manager_, &write_controller_,
+            /*block_cache_tracer=*/nullptr,
+            /*io_tracer=*/nullptr, /*db_id*/ "", /*db_session_id*/ "")),
         shutting_down_(false),
         mock_table_factory_(new mock::MockTableFactory()),
         error_handler_(nullptr, db_options_, &mutex_),
@@ -444,7 +444,7 @@ class CompactionJobTestBase : public testing::Test {
         new VersionSet(dbname_, &db_options_, env_options_, table_cache_.get(),
                        &write_buffer_manager_, &write_controller_,
                        /*block_cache_tracer=*/nullptr, /*io_tracer=*/nullptr,
-                       /*db_session_id*/ ""));
+                       /*db_id*/ "", /*db_session_id*/ ""));
     compaction_job_stats_.Reset();
     ASSERT_OK(SetIdentityFile(env_, dbname_));
 

package/deps/rocksdb/rocksdb/db/compaction/compaction_picker_level.cc

@@ -448,6 +448,14 @@ bool LevelCompactionBuilder::PickFileToCompact() {
     // do not pick a file to compact if it is being compacted
     // from n-1 level.
     if (f->being_compacted) {
+      if (ioptions_.compaction_pri == kRoundRobin) {
+        // TODO(zichen): this file may be involved in one compaction from
+        // an upper level, cannot advance the cursor for round-robin policy.
+        // Currently, we do not pick any file to compact in this case. We
+        // should fix this later to ensure a compaction is picked but the
+        // cursor shall not be advanced.
+        return false;
+      }
       continue;
     }
 
@@ -460,6 +468,13 @@ bool LevelCompactionBuilder::PickFileToCompact() {
       // A locked (pending compaction) input-level file was pulled in due to
       // user-key overlap.
       start_level_inputs_.clear();
+
+      // To ensure every file is selcted in a round-robin manner, we cannot
+      // skip the current file. So we return false and wait for the next time
+      // we can pick this file to compact
+      if (ioptions_.compaction_pri == kRoundRobin) {
+        return false;
+      }
       continue;
     }
 
@@ -479,6 +494,10 @@ bool LevelCompactionBuilder::PickFileToCompact() {
         !compaction_picker_->ExpandInputsToCleanCut(cf_name_, vstorage_,
                                                     &output_level_inputs)) {
       start_level_inputs_.clear();
+      // The same reason as above to ensure the round-robin compaction
+      if (ioptions_.compaction_pri == kRoundRobin) {
+        return false;
+      }
       continue;
     }
     base_index_ = index;
@@ -486,8 +505,9 @@ bool LevelCompactionBuilder::PickFileToCompact() {
   }
 
   // store where to start the iteration in the next call to PickCompaction
-  vstorage_->SetNextCompactionIndex(start_level_, cmp_idx);
-
+  if (ioptions_.compaction_pri != kRoundRobin) {
+    vstorage_->SetNextCompactionIndex(start_level_, cmp_idx);
+  }
   return start_level_inputs_.size() > 0;
 }
 

package/deps/rocksdb/rocksdb/db/compaction/compaction_picker_test.cc

@@ -1311,6 +1311,44 @@ TEST_F(CompactionPickerTest, CompactionPriMinOverlapping4) {
   ASSERT_EQ(6U, compaction->input(0, 0)->fd.GetNumber());
 }
 
+TEST_F(CompactionPickerTest, CompactionPriRoundRobin) {
+  std::vector<InternalKey> test_cursors = {InternalKey("249", 100, kTypeValue),
+                                           InternalKey("600", 100, kTypeValue),
+                                           InternalKey()};
+  std::vector<uint32_t> selected_files = {8U, 6U, 6U};
+
+  ioptions_.compaction_pri = kRoundRobin;
+  mutable_cf_options_.max_bytes_for_level_base = 10000000;
+  mutable_cf_options_.max_bytes_for_level_multiplier = 10;
+  for (size_t i = 0; i < test_cursors.size(); i++) {
+    // start a brand new version in each test.
+    NewVersionStorage(6, kCompactionStyleLevel);
+    vstorage_->ResizeCompactCursors(6);
+    // Set the cursor
+    vstorage_->AddCursorForOneLevel(2, test_cursors[i]);
+    Add(2, 6U, "150", "199", 50000000U);  // Overlap with 26U, 27U
+    Add(2, 7U, "200", "249", 50000000U);  // File not overlapping
+    Add(2, 8U, "300", "600", 50000000U);  // Overlap with 28U, 29U
+
+    Add(3, 26U, "130", "165", 60000000U);
+    Add(3, 27U, "166", "170", 60000000U);
+    Add(3, 28U, "270", "340", 60000000U);
+    Add(3, 29U, "401", "500", 60000000U);
+    UpdateVersionStorageInfo();
+    LevelCompactionPicker local_level_compaction_picker =
+        LevelCompactionPicker(ioptions_, &icmp_);
+    std::unique_ptr<Compaction> compaction(
+        local_level_compaction_picker.PickCompaction(
+            cf_name_, mutable_cf_options_, mutable_db_options_, vstorage_.get(),
+            &log_buffer_));
+    ASSERT_TRUE(compaction.get() != nullptr);
+    ASSERT_EQ(1U, compaction->num_input_files(0));
+    ASSERT_EQ(selected_files[i], compaction->input(0, 0)->fd.GetNumber());
+    // release the version storage
+    DeleteVersionStorage();
+  }
+}
+
 // This test exhibits the bug where we don't properly reset parent_index in
 // PickCompaction()
 TEST_F(CompactionPickerTest, ParentIndexResetBug) {

package/deps/rocksdb/rocksdb/db/db_basic_test.cc

@@ -2221,6 +2221,7 @@ TEST_F(DBMultiGetAsyncIOTest, GetFromL0) {
   // No async IO in this case since we don't do parallel lookup in L0
   ASSERT_EQ(multiget_io_batch_size.count, 0);
   ASSERT_EQ(multiget_io_batch_size.max, 0);
+  ASSERT_EQ(statistics()->getTickerCount(MULTIGET_COROUTINE_COUNT), 0);
 }
 
 TEST_F(DBMultiGetAsyncIOTest, GetFromL1) {
@@ -2257,6 +2258,7 @@ TEST_F(DBMultiGetAsyncIOTest, GetFromL1) {
   // A batch of 3 async IOs is expected, one for each overlapping file in L1
   ASSERT_EQ(multiget_io_batch_size.count, 1);
   ASSERT_EQ(multiget_io_batch_size.max, 3);
+  ASSERT_EQ(statistics()->getTickerCount(MULTIGET_COROUTINE_COUNT), 3);
 }
 
 TEST_F(DBMultiGetAsyncIOTest, LastKeyInFile) {
@@ -2407,27 +2409,37 @@ TEST_F(DBBasicTest, MultiGetStats) {
                 values.data(), s.data(), false);
 
   ASSERT_EQ(values.size(), kMultiGetBatchSize);
-  HistogramData hist_data_blocks;
+  HistogramData hist_level;
   HistogramData hist_index_and_filter_blocks;
   HistogramData hist_sst;
 
-  options.statistics->histogramData(NUM_DATA_BLOCKS_READ_PER_LEVEL,
-                                    &hist_data_blocks);
+  options.statistics->histogramData(NUM_LEVEL_READ_PER_MULTIGET, &hist_level);
   options.statistics->histogramData(NUM_INDEX_AND_FILTER_BLOCKS_READ_PER_LEVEL,
                                     &hist_index_and_filter_blocks);
   options.statistics->histogramData(NUM_SST_READ_PER_LEVEL, &hist_sst);
 
   // Maximum number of blocks read from a file system in a level.
-  ASSERT_EQ(hist_data_blocks.max, 32);
+  ASSERT_EQ(hist_level.max, 1);
   ASSERT_GT(hist_index_and_filter_blocks.max, 0);
   // Maximum number of sst files read from file system in a level.
   ASSERT_EQ(hist_sst.max, 2);
 
   // Minimun number of blocks read in a level.
-  ASSERT_EQ(hist_data_blocks.min, 4);
+  ASSERT_EQ(hist_level.min, 1);
   ASSERT_GT(hist_index_and_filter_blocks.min, 0);
   // Minimun number of sst files read in a level.
   ASSERT_EQ(hist_sst.min, 1);
+
+  for (PinnableSlice& value : values) {
+    value.Reset();
+  }
+  for (Status& status : s) {
+    status = Status::OK();
+  }
+  db_->MultiGet(read_opts, handles_[1], kMultiGetBatchSize, &keys[950],
+                values.data(), s.data(), false);
+  options.statistics->histogramData(NUM_LEVEL_READ_PER_MULTIGET, &hist_level);
+  ASSERT_EQ(hist_level.max, 2);
 }
 
 // Test class for batched MultiGet with prefix extractor

package/deps/rocksdb/rocksdb/db/db_block_cache_test.cc

@@ -670,7 +670,7 @@ class DBBlockCacheTest1 : public DBTestBase,
 };
 
 INSTANTIATE_TEST_CASE_P(DBBlockCacheTest1, DBBlockCacheTest1,
-                        ::testing::Values(1, 2, 3));
+                        ::testing::Values(1, 2));
 
 TEST_P(DBBlockCacheTest1, WarmCacheWithBlocksDuringFlush) {
   Options options = CurrentOptions();
@@ -686,13 +686,10 @@ TEST_P(DBBlockCacheTest1, WarmCacheWithBlocksDuringFlush) {
       table_options.partition_filters = true;
       table_options.index_type =
           BlockBasedTableOptions::IndexType::kTwoLevelIndexSearch;
-      table_options.filter_policy.reset(NewBloomFilterPolicy(10, false));
+      table_options.filter_policy.reset(NewBloomFilterPolicy(10));
       break;
-    case 2:  // block-based filter
-      table_options.filter_policy.reset(NewBloomFilterPolicy(10, true));
-      break;
-    case 3:  // full filter
-      table_options.filter_policy.reset(NewBloomFilterPolicy(10, false));
+    case 2:  // full filter
+      table_options.filter_policy.reset(NewBloomFilterPolicy(10));
       break;
     default:
       assert(false);