update point_lock_manager.cc

Summary: Pull Request resolved: https://github.com/facebook/rocksdb/pull/13485

Reviewed By: jaykorean, pdillinger

Differential Revision: D71787995

Pulled By: hx235

fbshipit-source-id: 59b6ff7c824adbdef34b6ae12d7dbcc3e0852961

BUCK

@@ -214,7 +214,7 @@ cpp_library_wrapper(name="rocksdb_lib", srcs=[
         "table/cuckoo/cuckoo_table_builder.cc",
         "table/cuckoo/cuckoo_table_factory.cc",
         "table/cuckoo/cuckoo_table_reader.cc",
-        "table/external_table_reader.cc",
+        "table/external_table.cc",
         "table/format.cc",
         "table/get_context.cc",
         "table/iterator.cc",

CMakeLists.txt

@@ -835,7 +835,7 @@ set(SOURCES
         table/cuckoo/cuckoo_table_builder.cc
         table/cuckoo/cuckoo_table_factory.cc
         table/cuckoo/cuckoo_table_reader.cc
-        table/external_table_reader.cc
+        table/external_table.cc
         table/format.cc
         table/get_context.cc
         table/iterator.cc

HISTORY.md

@@ -1,6 +1,51 @@
 # Rocksdb Change Log
 > NOTE: Entries for next release do not go here. Follow instructions in `unreleased_history/README.txt`
 
+## 10.1.0 (03/24/2025)
+### New Features
+* Added a new `DBOptions.calculate_sst_write_lifetime_hint_set` setting that allows to customize which compaction styles SST write lifetime hint calculation is allowed on. Today RocksDB supports only two modes `kCompactionStyleLevel` and `kCompactionStyleUniversal`.
+* Add a new field `num_l0_files` in `CompactionJobInfo` about the number of L0 files in the CF right before and after the compaction
+* Added per-key-placement feature in Remote Compaction
+* Implemented API DB::GetPropertiesOfTablesByLevel that retrieves table properties for files in each LSM tree level
+
+### Public API Changes
+* `GetAllKeyVersions()` now interprets empty slices literally, as valid keys, and uses new `OptSlice` type default value for extreme upper and lower range limits.
+* `DeleteFilesInRanges()` now takes `RangeOpt` which is based on `OptSlice`. The overload taking `RangePtr` is deprecated.
+* Add an unordered map of name/value pairs, ReadOptions::property_bag, to pass opaque options through to an external table when creating an Iterator.
+* Introduced CompactionServiceJobStatus::kAborted to allow handling aborted scenario in Schedule(), Wait() or OnInstallation() APIs in Remote Compactions.
+* format\_version < 2 in BlockBasedTableOptions is no longer supported for writing new files. Support for reading such files is deprecated and might be removed in the future. `CompressedSecondaryCacheOptions::compress_format_version == 1` is also deprecated.
+
+### Behavior Changes
+* `ldb` now returns an error if the specified `--compression_type` is not supported in the build.
+* MultiGet with snapshot and ReadOptions::read_tier = kPersistedTier will now read a consistent view across CFs (instead of potentially reading some CF before and some CF after a flush).
+* CreateColumnFamily() is no longer allowed on a read-only DB (OpenForReadOnly())
+
+### Bug Fixes
+* Fixed stats for Tiered Storage with preclude_last_level feature
+
+## 10.0.0 (02/21/2025)
+### New Features
+* Introduced new `auto_refresh_iterator_with_snapshot` opt-in knob that (when enabled) will periodically release obsolete memory and storage resources for as long as the iterator is making progress and its supplied `read_options.snapshot` was initialized with non-nullptr value.
+* Added the ability to plug-in a custom table reader implementation. See include/rocksdb/external_table_reader.h for more details.
+* Experimental feature: RocksDB now supports FAISS inverted file based indices via the secondary indexing framework. Applications can use FAISS secondary indices to automatically quantize embeddings and perform K-nearest-neighbors similarity searches. See `FaissIVFIndex` and `SecondaryIndex` for more details. Note: the FAISS integration currently requires using the BUCK build.
+* Add new DB property `num_running_compaction_sorted_runs` that tracks the number of sorted runs being processed by currently running compactions
+* Experimental feature: added support for simple secondary indices that index the specified column as-is. See `SimpleSecondaryIndex` and `SecondaryIndex` for more details.
+* Added new `TransactionDBOptions::txn_commit_bypass_memtable_threshold`, which enables optimized transaction commit (see `TransactionOptions::commit_bypass_memtable`) when the transaction size exceeds a configured threshold.
+
+### Public API Changes
+* Updated the query API of the experimental secondary indexing feature by removing the earlier `SecondaryIndex::NewIterator` virtual and adding a `SecondaryIndexIterator` class that can be utilized by applications to find the primary keys for a given search target.
+* Added back the ability to leverage the primary key when building secondary index entries. This involved changes to the signatures of `SecondaryIndex::GetSecondary{KeyPrefix,Value}` as well as the addition of a new method `SecondaryIndex::FinalizeSecondaryKeyPrefix`. See the API comments for more details.
+* Minimum supported version of ZSTD is now 1.4.0, for code simplification. Obsolete `CompressionType` `kZSTDNotFinalCompression` is also removed.
+
+### Behavior Changes
+* `VerifyBackup` in `verify_with_checksum`=`true` mode will now evaluate checksums in parallel. As a result, unlike in case of original implementation, the API won't bail out on a very first corruption / mismatch and instead will iterate over all the backup files logging success / _degree_of_failure_ for each.
+* Reversed the order of updates to the same key in WriteBatchWithIndex. This means if there are multiple updates to the same key, the most recent update is ordered first. This affects the output of WBWIIterator. When WriteBatchWithIndex is created with `overwrite_key=true`, this affects the output only if Merge is used (#13387).
+* Added support for Merge operations in transactions using option `TransactionOptions::commit_bypass_memtable`.
+
+### Bug Fixes
+* Fixed GetMergeOperands() API in ReadOnlyDB and SecondaryDB
+* Fix a bug in `GetMergeOperands()` that can return incorrect status (MergeInProgress) and incorrect number of merge operands. This can happen when `GetMergeOperandsOptions::continue_cb` is set, both active and immutable memtables have merge operands and the callback stops the look up at the immutable memtable.
+
 ## 9.11.0 (01/17/2025)
 ### New Features
 * Introduce CancelAwaitingJobs() in CompactionService interface which will allow users to implement cancellation of running remote compactions from the primary instance

Makefile

@@ -2489,7 +2489,7 @@ checkout_folly:
 	fi
 	@# Pin to a particular version for public CI, so that PR authors don't
 	@# need to worry about folly breaking our integration. Update periodically
-	cd third-party/folly && git reset --hard f1c1542fb071e9d88181279ec24043400222aef1
+	cd third-party/folly && git reset --hard 78286282478e1ae05b2e8cbcf0e2139eab283bea
 	@# NOTE: this hack is required for clang in some cases
 	perl -pi -e 's/int rv = syscall/int rv = (int)syscall/' third-party/folly/folly/detail/Futex.cpp
 	@# NOTE: this hack is required for gcc in some cases

db/blob/blob_file_builder.cc

@@ -188,10 +188,12 @@ Status BlobFileBuilder::OpenBlobFileIfNeeded() {
   }
 
   std::unique_ptr<FSWritableFile> file;
-
+  FileOptions fo_copy;
   {
     assert(file_options_);
-    Status s = NewWritableFile(fs_, blob_file_path, &file, *file_options_);
+    fo_copy = *file_options_;
+    fo_copy.write_hint = write_hint_;
+    Status s = NewWritableFile(fs_, blob_file_path, &file, fo_copy);
 
     TEST_SYNC_POINT_CALLBACK(
         "BlobFileBuilder::OpenBlobFileIfNeeded:NewWritableFile", &s);
@@ -209,7 +211,9 @@ Status BlobFileBuilder::OpenBlobFileIfNeeded() {
 
   assert(file);
   file->SetIOPriority(write_options_->rate_limiter_priority);
-  file->SetWriteLifeTimeHint(write_hint_);
+  // Subsequent attempts to override the hint via SetWriteLifeTimeHint
+  // with the very same value will be ignored by the fs.
+  file->SetWriteLifeTimeHint(fo_copy.write_hint);
   FileTypeSet tmp_set = immutable_options_->checksum_handoff_file_types;
   Statistics* const statistics = immutable_options_->stats;
   std::unique_ptr<WritableFileWriter> file_writer(new WritableFileWriter(

db/builder.cc

@@ -145,7 +145,9 @@ Status BuildTable(
       bool use_direct_writes = file_options.use_direct_writes;
       TEST_SYNC_POINT_CALLBACK("BuildTable:create_file", &use_direct_writes);
 #endif  // !NDEBUG
-      IOStatus io_s = NewWritableFile(fs, fname, &file, file_options);
+      FileOptions fo_copy = file_options;
+      fo_copy.write_hint = write_hint;
+      IOStatus io_s = NewWritableFile(fs, fname, &file, fo_copy);
       assert(s.ok());
       s = io_s;
       if (io_status->ok()) {
@@ -163,7 +165,9 @@ Status BuildTable(
       table_file_created = true;
       FileTypeSet tmp_set = ioptions.checksum_handoff_file_types;
       file->SetIOPriority(tboptions.write_options.rate_limiter_priority);
-      file->SetWriteLifeTimeHint(write_hint);
+      // Subsequent attempts to override the hint via SetWriteLifeTimeHint
+      // with the very same value will be ignored by the fs.
+      file->SetWriteLifeTimeHint(fo_copy.write_hint);
       file_writer.reset(new WritableFileWriter(
           std::move(file), fname, file_options, ioptions.clock, io_tracer,
           ioptions.stats, Histograms::SST_WRITE_MICROS, ioptions.listeners,

db/column_family.cc

@@ -168,7 +168,8 @@ Status CheckConcurrentWritesSupported(const ColumnFamilyOptions& cf_options) {
   }
   if (!cf_options.memtable_factory->IsInsertConcurrentlySupported()) {
     return Status::InvalidArgument(
-        "Memtable doesn't allow concurrent writes (allow_concurrent_memtable_write)");
+        "Memtable doesn't allow concurrent writes "
+        "(allow_concurrent_memtable_write)");
   }
   return Status::OK();
 }
@@ -199,8 +200,9 @@ const uint64_t kDefaultTtl = 0xfffffffffffffffe;
 const uint64_t kDefaultPeriodicCompSecs = 0xfffffffffffffffe;
 }  // anonymous namespace
 
-ColumnFamilyOptions SanitizeOptions(const ImmutableDBOptions& db_options,
-                                    const ColumnFamilyOptions& src) {
+ColumnFamilyOptions SanitizeCfOptions(const ImmutableDBOptions& db_options,
+                                      bool read_only,
+                                      const ColumnFamilyOptions& src) {
   ColumnFamilyOptions result = src;
   size_t clamp_max = std::conditional<
       sizeof(size_t) == 4, std::integral_constant<size_t, 0xffffffff>,
@@ -239,6 +241,10 @@ ColumnFamilyOptions SanitizeOptions(const ImmutableDBOptions& db_options,
 
     result.min_write_buffer_number_to_merge = 1;
   }
+  if (result.disallow_memtable_writes) {
+    // A simple memtable that enforces MarkReadOnly (unlike skip list)
+    result.memtable_factory = std::make_shared<VectorRepFactory>();
+  }
 
   if (result.num_levels < 1) {
     result.num_levels = 1;
@@ -435,6 +441,18 @@ ColumnFamilyOptions SanitizeOptions(const ImmutableDBOptions& db_options,
     result.periodic_compaction_seconds = 0;
   }
 
+  if (read_only && (result.preserve_internal_time_seconds > 0 ||
+                    result.preclude_last_level_data_seconds > 0)) {
+    // With no writes coming in, we don't need periodic SeqnoToTime entries.
+    // Existing SST files may or may not have that info associated with them.
+    ROCKS_LOG_WARN(
+        db_options.info_log.get(),
+        "preserve_internal_time_seconds and preclude_last_level_data_seconds "
+        "are ignored in read-only DB");
+    result.preserve_internal_time_seconds = 0;
+    result.preclude_last_level_data_seconds = 0;
+  }
+
   return result;
 }
 
@@ -492,6 +510,17 @@ void SuperVersion::Init(
   imm->Ref();
   current->Ref();
   refs.store(1, std::memory_order_relaxed);
+
+  // There should be at least one mapping entry iff time tracking is enabled.
+#ifndef NDEBUG
+  MinAndMaxPreserveSeconds preserve_info{mutable_cf_options};
+  if (preserve_info.IsEnabled()) {
+    assert(seqno_to_time_mapping);
+    assert(!seqno_to_time_mapping->Empty());
+  } else {
+    assert(seqno_to_time_mapping == nullptr);
+  }
+#endif  // NDEBUG
 }
 
 namespace {
@@ -530,7 +559,7 @@ ColumnFamilyData::ColumnFamilyData(
     const FileOptions* file_options, ColumnFamilySet* column_family_set,
     BlockCacheTracer* const block_cache_tracer,
     const std::shared_ptr<IOTracer>& io_tracer, const std::string& db_id,
-    const std::string& db_session_id)
+    const std::string& db_session_id, bool read_only)
     : id_(id),
       name_(name),
       dummy_versions_(_dummy_versions),
@@ -540,7 +569,7 @@ ColumnFamilyData::ColumnFamilyData(
       dropped_(false),
       flush_skip_reschedule_(false),
       internal_comparator_(cf_options.comparator),
-      initial_cf_options_(SanitizeOptions(db_options, cf_options)),
+      initial_cf_options_(SanitizeCfOptions(db_options, read_only, cf_options)),
       ioptions_(db_options, initial_cf_options_),
       mutable_cf_options_(initial_cf_options_),
       is_delete_range_supported_(
@@ -1339,20 +1368,17 @@ bool ColumnFamilyData::ReturnThreadLocalSuperVersion(SuperVersion* sv) {
   return false;
 }
 
-void ColumnFamilyData::InstallSuperVersion(SuperVersionContext* sv_context,
-                                           InstrumentedMutex* db_mutex) {
-  db_mutex->AssertHeld();
-  return InstallSuperVersion(sv_context, mutable_cf_options_);
-}
-
 void ColumnFamilyData::InstallSuperVersion(
-    SuperVersionContext* sv_context,
-    const MutableCFOptions& mutable_cf_options) {
+    SuperVersionContext* sv_context, InstrumentedMutex* db_mutex,
+    std::optional<std::shared_ptr<SeqnoToTimeMapping>>
+        new_seqno_to_time_mapping) {
+  db_mutex->AssertHeld();
+
   SuperVersion* new_superversion = sv_context->new_superversion.release();
-  new_superversion->mutable_cf_options = mutable_cf_options;
+  new_superversion->mutable_cf_options = GetLatestMutableCFOptions();
   new_superversion->Init(this, mem_, imm_.current(), current_,
-                         sv_context->new_seqno_to_time_mapping
-                             ? std::move(sv_context->new_seqno_to_time_mapping)
+                         new_seqno_to_time_mapping.has_value()
+                             ? std::move(new_seqno_to_time_mapping.value())
                          : super_version_
                              ? super_version_->ShareSeqnoToTimeMapping()
                              : nullptr);
@@ -1365,7 +1391,7 @@ void ColumnFamilyData::InstallSuperVersion(
     // currently RecalculateWriteStallConditions() treats it as further slowing
     // down is needed.
     super_version_->write_stall_condition =
-        RecalculateWriteStallConditions(mutable_cf_options);
+        RecalculateWriteStallConditions(new_superversion->mutable_cf_options);
   } else {
     super_version_->write_stall_condition =
         old_superversion->write_stall_condition;
@@ -1378,8 +1404,9 @@ void ColumnFamilyData::InstallSuperVersion(
     ResetThreadLocalSuperVersions();
 
     if (old_superversion->mutable_cf_options.write_buffer_size !=
-        mutable_cf_options.write_buffer_size) {
-      mem_->UpdateWriteBufferSize(mutable_cf_options.write_buffer_size);
+        new_superversion->mutable_cf_options.write_buffer_size) {
+      mem_->UpdateWriteBufferSize(
+          new_superversion->mutable_cf_options.write_buffer_size);
     }
     if (old_superversion->write_stall_condition !=
         new_superversion->write_stall_condition) {
@@ -1680,7 +1707,8 @@ ColumnFamilySet::ColumnFamilySet(const std::string& dbname,
       dummy_cfd_(new ColumnFamilyData(
           ColumnFamilyData::kDummyColumnFamilyDataId, "", nullptr, nullptr,
           nullptr, ColumnFamilyOptions(), *db_options, &file_options_, nullptr,
-          block_cache_tracer, io_tracer, db_id, db_session_id)),
+          block_cache_tracer, io_tracer, db_id, db_session_id,
+          /*read_only*/ true)),
       default_cfd_cache_(nullptr),
       db_name_(dbname),
       db_options_(db_options),
@@ -1752,12 +1780,12 @@ size_t ColumnFamilySet::NumberOfColumnFamilies() const {
 // under a DB mutex AND write thread
 ColumnFamilyData* ColumnFamilySet::CreateColumnFamily(
     const std::string& name, uint32_t id, Version* dummy_versions,
-    const ColumnFamilyOptions& options) {
+    const ColumnFamilyOptions& options, bool read_only) {
   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_, &file_options_, this, block_cache_tracer_, io_tracer_,
-      db_id_, db_session_id_);
+      db_id_, db_session_id_, read_only);
   column_families_.insert({name, id});
   column_family_data_.insert({id, new_cfd});
   auto ucmp = new_cfd->user_comparator();

db/column_family.h

@@ -281,8 +281,9 @@ Status CheckConcurrentWritesSupported(const ColumnFamilyOptions& cf_options);
 Status CheckCFPathsSupported(const DBOptions& db_options,
                              const ColumnFamilyOptions& cf_options);
 
-ColumnFamilyOptions SanitizeOptions(const ImmutableDBOptions& db_options,
-                                    const ColumnFamilyOptions& src);
+ColumnFamilyOptions SanitizeCfOptions(const ImmutableDBOptions& db_options,
+                                      bool read_only,
+                                      const ColumnFamilyOptions& src);
 // Wrap user defined table properties collector factories `from cf_options`
 // into internal ones in internal_tbl_prop_coll_factories. Add a system internal
 // one too.
@@ -384,14 +385,17 @@ class ColumnFamilyData {
   Version* dummy_versions() { return dummy_versions_; }
   Version* current() { return current_; }  // REQUIRE: DB mutex held
   void SetCurrent(Version* _current);      // REQUIRE: DB mutex held
-  uint64_t GetNumLiveVersions() const;    // REQUIRE: DB mutex held
-  uint64_t GetTotalSstFilesSize() const;  // REQUIRE: DB mutex held
-  uint64_t GetLiveSstFilesSize() const;   // REQUIRE: DB mutex held
-  uint64_t GetTotalBlobFileSize() const;  // REQUIRE: DB mutex held
+  uint64_t GetNumLiveVersions() const;     // REQUIRE: DB mutex held
+  uint64_t GetTotalSstFilesSize() const;   // REQUIRE: DB mutex held
+  uint64_t GetLiveSstFilesSize() const;    // REQUIRE: DB mutex held
+  uint64_t GetTotalBlobFileSize() const;   // REQUIRE: DB mutex held
   // REQUIRE: DB mutex held
   void SetMemtable(MemTable* new_mem) {
     AssignMemtableID(new_mem);
     mem_ = new_mem;
+    if (ioptions_.disallow_memtable_writes) {
+      mem_->MarkImmutable();
+    }
   }
 
   void AssignMemtableID(ReadOnlyMemTable* new_imm) {
@@ -487,15 +491,11 @@ class ColumnFamilyData {
   uint64_t GetSuperVersionNumberRelaxed() const {
     return super_version_number_.load(std::memory_order_relaxed);
   }
-  // will return a pointer to SuperVersion* if previous SuperVersion
-  // if its reference count is zero and needs deletion or nullptr if not
-  // As argument takes a pointer to allocated SuperVersion to enable
-  // the clients to allocate SuperVersion outside of mutex.
-  // IMPORTANT: Only call this from DBImpl::InstallSuperVersion()
+  // Only intended for use by DBImpl::InstallSuperVersion() and variants
   void InstallSuperVersion(SuperVersionContext* sv_context,
-                           const MutableCFOptions& mutable_cf_options);
-  void InstallSuperVersion(SuperVersionContext* sv_context,
-                           InstrumentedMutex* db_mutex);
+                           InstrumentedMutex* db_mutex,
+                           std::optional<std::shared_ptr<SeqnoToTimeMapping>>
+                               new_seqno_to_time_mapping = {});
 
   void ResetThreadLocalSuperVersions();
 
@@ -590,16 +590,14 @@ class ColumnFamilyData {
 
  private:
   friend class ColumnFamilySet;
-  ColumnFamilyData(uint32_t id, const std::string& name,
-                   Version* dummy_versions, Cache* table_cache,
-                   WriteBufferManager* write_buffer_manager,
-                   const ColumnFamilyOptions& 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::string& db_id, const std::string& db_session_id);
+  ColumnFamilyData(
+      uint32_t id, const std::string& name, Version* dummy_versions,
+      Cache* table_cache, WriteBufferManager* write_buffer_manager,
+      const ColumnFamilyOptions& 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::string& db_id,
+      const std::string& db_session_id, bool read_only);
 
   std::vector<std::string> GetDbPaths() const;
 
@@ -761,7 +759,8 @@ class ColumnFamilySet {
 
   ColumnFamilyData* CreateColumnFamily(const std::string& name, uint32_t id,
                                        Version* dummy_version,
-                                       const ColumnFamilyOptions& options);
+                                       const ColumnFamilyOptions& options,
+                                       bool read_only);
 
   const UnorderedMap<uint32_t, size_t>& GetRunningColumnFamiliesTimestampSize()
       const {

db/column_family_test.cc

@@ -271,7 +271,8 @@ class ColumnFamilyTestBase : public testing::Test {
       // them.
       ASSERT_OK(RocksDBOptionsParser::VerifyCFOptions(
           ConfigOptions(), desc.options,
-          SanitizeOptions(dbfull()->immutable_db_options(), current_cf_opt)));
+          SanitizeCfOptions(dbfull()->immutable_db_options(),
+                            /*read_only*/ false, current_cf_opt)));
       cfi++;
     }
   }
@@ -2232,7 +2233,7 @@ TEST_P(ColumnFamilyTest, CreateMissingColumnFamilies) {
   ASSERT_EQ(my_fs->options_files_created.load(), 2);
 }
 
-TEST_P(ColumnFamilyTest, SanitizeOptions) {
+TEST_P(ColumnFamilyTest, SanitizeCfOptions) {
   DBOptions db_options;
   for (int s = kCompactionStyleLevel; s <= kCompactionStyleUniversal; ++s) {
     for (int l = 0; l <= 2; l++) {
@@ -2248,8 +2249,8 @@ TEST_P(ColumnFamilyTest, SanitizeOptions) {
             original.write_buffer_size =
                 l * 4 * 1024 * 1024 + i * 1024 * 1024 + j * 1024 + k;
 
-            ColumnFamilyOptions result =
-                SanitizeOptions(ImmutableDBOptions(db_options), original);
+            ColumnFamilyOptions result = SanitizeCfOptions(
+                ImmutableDBOptions(db_options), /*read_only*/ false, original);
             ASSERT_TRUE(result.level0_stop_writes_trigger >=
                         result.level0_slowdown_writes_trigger);
             ASSERT_TRUE(result.level0_slowdown_writes_trigger >=

db/compaction/compaction.cc

@@ -338,16 +338,16 @@ Compaction::Compaction(
                   _blob_garbage_collection_age_cutoff > 1
               ? mutable_cf_options().blob_garbage_collection_age_cutoff
               : _blob_garbage_collection_age_cutoff),
-      penultimate_level_(
-          // For simplicity, we don't support the concept of "penultimate level"
+      proximal_level_(
+          // For simplicity, we don't support the concept of "proximal level"
           // with `CompactionReason::kExternalSstIngestion` and
           // `CompactionReason::kRefitLevel`
           _compaction_reason == CompactionReason::kExternalSstIngestion ||
                   _compaction_reason == CompactionReason::kRefitLevel
               ? Compaction::kInvalidLevel
-              : EvaluatePenultimateLevel(vstorage, mutable_cf_options_,
-                                         immutable_options_, start_level_,
-                                         output_level_)) {
+              : EvaluateProximalLevel(vstorage, mutable_cf_options_,
+                                      immutable_options_, start_level_,
+                                      output_level_)) {
   MarkFilesBeingCompacted(true);
   if (is_manual_compaction_) {
     compaction_reason_ = CompactionReason::kManualCompaction;
@@ -405,10 +405,10 @@ Compaction::Compaction(
     }
   }
 
-  PopulatePenultimateLevelOutputRange();
+  PopulateProximalLevelOutputRange();
 }
 
-void Compaction::PopulatePenultimateLevelOutputRange() {
+void Compaction::PopulateProximalLevelOutputRange() {
   if (!SupportsPerKeyPlacement()) {
     assert(keep_in_last_level_through_seqno_ == kMaxSequenceNumber);
     return;
@@ -417,46 +417,42 @@ void Compaction::PopulatePenultimateLevelOutputRange() {
   // exclude the last level, the range of all input levels is the safe range
   // of keys that can be moved up.
   int exclude_level = number_levels_ - 1;
-  penultimate_output_range_type_ = PenultimateOutputRangeType::kNonLastRange;
+  proximal_output_range_type_ = ProximalOutputRangeType::kNonLastRange;
 
-  // For universal compaction, the penultimate_output_range could be extended if
-  // all penultimate level files are included in the compaction (which includes
-  // the case that the penultimate level is empty).
+  // For universal compaction, the proximal_output_range could be extended if
+  // all proximal level files are included in the compaction (which includes
+  // the case that the proximal level is empty).
   if (immutable_options_.compaction_style == kCompactionStyleUniversal) {
     exclude_level = kInvalidLevel;
-    penultimate_output_range_type_ = PenultimateOutputRangeType::kFullRange;
-    std::set<uint64_t> penultimate_inputs;
+    proximal_output_range_type_ = ProximalOutputRangeType::kFullRange;
+    std::set<uint64_t> proximal_inputs;
     for (const auto& input_lvl : inputs_) {
-      if (input_lvl.level == penultimate_level_) {
+      if (input_lvl.level == proximal_level_) {
         for (const auto& file : input_lvl.files) {
-          penultimate_inputs.emplace(file->fd.GetNumber());
+          proximal_inputs.emplace(file->fd.GetNumber());
         }
       }
     }
-    auto penultimate_files = input_vstorage_->LevelFiles(penultimate_level_);
-    for (const auto& file : penultimate_files) {
-      if (penultimate_inputs.find(file->fd.GetNumber()) ==
-          penultimate_inputs.end()) {
+    auto proximal_files = input_vstorage_->LevelFiles(proximal_level_);
+    for (const auto& file : proximal_files) {
+      if (proximal_inputs.find(file->fd.GetNumber()) == proximal_inputs.end()) {
         exclude_level = number_levels_ - 1;
-        penultimate_output_range_type_ =
-            PenultimateOutputRangeType::kNonLastRange;
+        proximal_output_range_type_ = ProximalOutputRangeType::kNonLastRange;
         break;
       }
     }
   }
 
-  // FIXME: should make use of `penultimate_output_range_type_`.
+  // FIXME: should make use of `proximal_output_range_type_`.
   // FIXME: when last level's input range does not overlap with
-  //  penultimate level, and penultimate level input is empty,
-  //  this call will not set penultimate_level_smallest_ or
-  //  penultimate_level_largest_. No keys will be compacted up.
-  GetBoundaryInternalKeys(input_vstorage_, inputs_,
-                          &penultimate_level_smallest_,
-                          &penultimate_level_largest_, exclude_level);
+  //  proximal level, and proximal level input is empty,
+  //  this call will not set proximal_level_smallest_ or
+  //  proximal_level_largest_. No keys will be compacted up.
+  GetBoundaryInternalKeys(input_vstorage_, inputs_, &proximal_level_smallest_,
+                          &proximal_level_largest_, exclude_level);
 
-  if (penultimate_output_range_type_ !=
-      PenultimateOutputRangeType::kFullRange) {
-    // If not full range in penultimate level, must keep everything already
+  if (proximal_output_range_type_ != ProximalOutputRangeType::kFullRange) {
+    // If not full range in proximal level, must keep everything already
     // in the last level there, because moving it back up might cause
     // overlap/placement issues that are difficult to resolve properly in the
     // presence of range deletes
@@ -486,23 +482,23 @@ Compaction::~Compaction() {
 }
 
 bool Compaction::SupportsPerKeyPlacement() const {
-  return penultimate_level_ != kInvalidLevel;
+  return proximal_level_ != kInvalidLevel;
 }
 
-int Compaction::GetPenultimateLevel() const { return penultimate_level_; }
+int Compaction::GetProximalLevel() const { return proximal_level_; }
 
 // smallest_key and largest_key include timestamps if user-defined timestamp is
 // enabled.
-bool Compaction::OverlapPenultimateLevelOutputRange(
+bool Compaction::OverlapProximalLevelOutputRange(
     const Slice& smallest_key, const Slice& largest_key) const {
   if (!SupportsPerKeyPlacement()) {
     return false;
   }
 
-  // See FIXME in Compaction::PopulatePenultimateLevelOutputRange().
+  // See FIXME in Compaction::PopulateProximalLevelOutputRange().
   // We do not compact any key up in this case.
-  if (penultimate_level_smallest_.size() == 0 ||
-      penultimate_level_largest_.size() == 0) {
+  if (proximal_level_smallest_.size() == 0 ||
+      proximal_level_largest_.size() == 0) {
     return false;
   }
 
@@ -510,13 +506,13 @@ bool Compaction::OverlapPenultimateLevelOutputRange(
       input_vstorage_->InternalComparator()->user_comparator();
 
   return ucmp->CompareWithoutTimestamp(
-             smallest_key, penultimate_level_largest_.user_key()) <= 0 &&
+             smallest_key, proximal_level_largest_.user_key()) <= 0 &&
          ucmp->CompareWithoutTimestamp(
-             largest_key, penultimate_level_smallest_.user_key()) >= 0;
+             largest_key, proximal_level_smallest_.user_key()) >= 0;
 }
 
 // key includes timestamp if user-defined timestamp is enabled.
-void Compaction::TEST_AssertWithinPenultimateLevelOutputRange(
+void Compaction::TEST_AssertWithinProximalLevelOutputRange(
     const Slice& user_key, bool expect_failure) const {
 #ifdef NDEBUG
   (void)user_key;
@@ -524,15 +520,15 @@ void Compaction::TEST_AssertWithinPenultimateLevelOutputRange(
 #else
   assert(SupportsPerKeyPlacement());
 
-  assert(penultimate_level_smallest_.size() > 0);
-  assert(penultimate_level_largest_.size() > 0);
+  assert(proximal_level_smallest_.size() > 0);
+  assert(proximal_level_largest_.size() > 0);
 
   auto* cmp = input_vstorage_->user_comparator();
 
   // op_type of a key can change during compaction, e.g. Merge -> Put.
-  if (!(cmp->Compare(user_key, penultimate_level_smallest_.user_key()) >= 0)) {
+  if (!(cmp->Compare(user_key, proximal_level_smallest_.user_key()) >= 0)) {
     assert(expect_failure);
-  } else if (!(cmp->Compare(user_key, penultimate_level_largest_.user_key()) <=
+  } else if (!(cmp->Compare(user_key, proximal_level_largest_.user_key()) <=
                0)) {
     assert(expect_failure);
   } else {
@@ -1018,7 +1014,7 @@ uint64_t Compaction::MinInputFileEpochNumber() const {
   return min_epoch_number;
 }
 
-int Compaction::EvaluatePenultimateLevel(
+int Compaction::EvaluateProximalLevel(
     const VersionStorageInfo* vstorage,
     const MutableCFOptions& mutable_cf_options,
     const ImmutableOptions& immutable_options, const int start_level,
@@ -1033,21 +1029,21 @@ int Compaction::EvaluatePenultimateLevel(
     return kInvalidLevel;
   }
 
-  int penultimate_level = output_level - 1;
-  assert(penultimate_level < immutable_options.num_levels);
-  if (penultimate_level <= 0) {
+  int proximal_level = output_level - 1;
+  assert(proximal_level < immutable_options.num_levels);
+  if (proximal_level <= 0) {
     return kInvalidLevel;
   }
 
-  // If the penultimate level is not within input level -> output level range
-  // check if the penultimate output level is empty, if it's empty, it could
-  // also be locked for the penultimate output.
+  // If the proximal level is not within input level -> output level range
+  // check if the proximal output level is empty, if it's empty, it could
+  // also be locked for the proximal output.
   // TODO: ideally, it only needs to check if there's a file within the
   //  compaction output key range. For simplicity, it just check if there's any
-  //  file on the penultimate level.
+  //  file on the proximal level.
   if (start_level == immutable_options.num_levels - 1 &&
       (immutable_options.compaction_style != kCompactionStyleUniversal ||
-       !vstorage->LevelFiles(penultimate_level).empty())) {
+       !vstorage->LevelFiles(proximal_level).empty())) {
     return kInvalidLevel;
   }
 
@@ -1061,7 +1057,7 @@ int Compaction::EvaluatePenultimateLevel(
     return kInvalidLevel;
   }
 
-  return penultimate_level;
+  return proximal_level;
 }
 
 void Compaction::FilterInputsForCompactionIterator() {

db/compaction/compaction.h

@@ -102,13 +102,13 @@ class Compaction {
                  BlobGarbageCollectionPolicy::kUseDefault,
              double blob_garbage_collection_age_cutoff = -1);
 
-  // The type of the penultimate level output range
-  enum class PenultimateOutputRangeType : int {
-    kNotSupported,  // it cannot output to the penultimate level
-    kFullRange,     // any data could be output to the penultimate level
+  // The type of the proximal level output range
+  enum class ProximalOutputRangeType : int {
+    kNotSupported,  // it cannot output to the proximal level
+    kFullRange,     // any data could be output to the proximal level
     kNonLastRange,  // only the keys within non_last_level compaction inputs can
-                    // be outputted to the penultimate level
-    kDisabled,      // no data can be outputted to the penultimate level
+                    // be outputted to the proximal level
+    kDisabled,      // no data can be outputted to the proximal level
   };
 
   // No copying allowed
@@ -370,29 +370,29 @@ class Compaction {
 
   Slice GetLargestUserKey() const { return largest_user_key_; }
 
-  PenultimateOutputRangeType GetPenultimateOutputRangeType() const {
-    return penultimate_output_range_type_;
+  ProximalOutputRangeType GetProximalOutputRangeType() const {
+    return proximal_output_range_type_;
   }
 
   // Return true if the compaction supports per_key_placement
   bool SupportsPerKeyPlacement() const;
 
-  // Get per_key_placement penultimate output level, which is `last_level - 1`
+  // Get per_key_placement proximal output level, which is `last_level - 1`
   // if per_key_placement feature is supported. Otherwise, return -1.
-  int GetPenultimateLevel() const;
+  int GetProximalLevel() const;
 
-  // Return true if the given range is overlap with penultimate level output
+  // Return true if the given range is overlap with proximal level output
   // range.
   // Both smallest_key and largest_key include timestamps if user-defined
   // timestamp is enabled.
-  bool OverlapPenultimateLevelOutputRange(const Slice& smallest_key,
-                                          const Slice& largest_key) const;
+  bool OverlapProximalLevelOutputRange(const Slice& smallest_key,
+                                       const Slice& largest_key) const;
 
-  // For testing purposes, check that a key is within penultimate level
+  // For testing purposes, check that a key is within proximal level
   // output range for per_key_placement feature, which is safe to place the key
-  // to the penultimate level. Different compaction strategies have different
+  // to the proximal level. Different compaction strategies have different
   // rules. `user_key` includes timestamp if user-defined timestamp is enabled.
-  void TEST_AssertWithinPenultimateLevelOutputRange(
+  void TEST_AssertWithinProximalLevelOutputRange(
       const Slice& user_key, bool expect_failure = false) const;
 
   CompactionReason compaction_reason() const { return compaction_reason_; }
@@ -441,20 +441,20 @@ class Compaction {
 
   static constexpr int kInvalidLevel = -1;
 
-  // Evaluate penultimate output level. If the compaction supports
-  // per_key_placement feature, it returns the penultimate level number.
+  // Evaluate proximal output level. If the compaction supports
+  // per_key_placement feature, it returns the proximal level number.
   // Otherwise, it's set to kInvalidLevel (-1), which means
-  // output_to_penultimate_level is not supported.
-  // Note: even the penultimate level output is supported (PenultimateLevel !=
+  // output_to_proximal_level is not supported.
+  // Note: even the proximal level output is supported (ProximalLevel !=
   // kInvalidLevel), some key range maybe unsafe to be outputted to the
-  // penultimate level. The safe key range is populated by
-  // `PopulatePenultimateLevelOutputRange()`.
-  // Which could potentially disable all penultimate level output.
-  static int EvaluatePenultimateLevel(
-      const VersionStorageInfo* vstorage,
-      const MutableCFOptions& mutable_cf_options,
-      const ImmutableOptions& immutable_options, const int start_level,
-      const int output_level);
+  // proximal level. The safe key range is populated by
+  // `PopulateProximalLevelOutputRange()`.
+  // Which could potentially disable all proximal level output.
+  static int EvaluateProximalLevel(const VersionStorageInfo* vstorage,
+                                   const MutableCFOptions& mutable_cf_options,
+                                   const ImmutableOptions& immutable_options,
+                                   const int start_level,
+                                   const int output_level);
 
   // If some data cannot be safely migrated "up" the LSM tree due to a change
   // in the preclude_last_level_data_seconds setting, this indicates a sequence
@@ -482,10 +482,10 @@ class Compaction {
       InternalKey* smallest_key, InternalKey* largest_key,
       int exclude_level = -1);
 
-  // populate penultimate level output range, which will be used to determine if
-  // a key is safe to output to the penultimate level (details see
-  // `Compaction::WithinPenultimateLevelOutputRange()`.
-  void PopulatePenultimateLevelOutputRange();
+  // populate proximal level output range, which will be used to determine if
+  // a key is safe to output to the proximal level (details see
+  // `Compaction::WithinProximalLevelOutputRange()`.
+  void PopulateProximalLevelOutputRange();
 
   // If oldest snapshot is specified at Compaction construction time, we have
   // an opportunity to optimize inputs for compaction iterator for this case:
@@ -616,20 +616,20 @@ class Compaction {
 
   // only set when per_key_placement feature is enabled, -1 (kInvalidLevel)
   // means not supported.
-  const int penultimate_level_;
+  const int proximal_level_;
 
-  // Key range for penultimate level output
+  // Key range for proximal level output
   // includes timestamp if user-defined timestamp is enabled.
-  // penultimate_output_range_type_ shows the range type
-  InternalKey penultimate_level_smallest_;
-  InternalKey penultimate_level_largest_;
-  PenultimateOutputRangeType penultimate_output_range_type_ =
-      PenultimateOutputRangeType::kNotSupported;
+  // proximal_output_range_type_ shows the range type
+  InternalKey proximal_level_smallest_;
+  InternalKey proximal_level_largest_;
+  ProximalOutputRangeType proximal_output_range_type_ =
+      ProximalOutputRangeType::kNotSupported;
 };
 
 #ifndef NDEBUG
 // Helper struct only for tests, which contains the data to decide if a key
-// should be output to the penultimate level.
+// should be output to the proximal level.
 // TODO: remove this when the public feature knob is available
 struct PerKeyPlacementContext {
   const int level;
@@ -637,16 +637,16 @@ struct PerKeyPlacementContext {
   const Slice value;
   const SequenceNumber seq_num;
 
-  bool& output_to_penultimate_level;
+  bool& output_to_proximal_level;
 
   PerKeyPlacementContext(int _level, Slice _key, Slice _value,
                          SequenceNumber _seq_num,
-                         bool& _output_to_penultimate_level)
+                         bool& _output_to_proximal_level)
       : level(_level),
         key(_key),
         value(_value),
         seq_num(_seq_num),
-        output_to_penultimate_level(_output_to_penultimate_level) {}
+        output_to_proximal_level(_output_to_proximal_level) {}
 };
 #endif /* !NDEBUG */
 

db/compaction/compaction_job.cc

@@ -109,16 +109,16 @@ const char* GetCompactionReasonString(CompactionReason compaction_reason) {
   }
 }
 
-const char* GetCompactionPenultimateOutputRangeTypeString(
-    Compaction::PenultimateOutputRangeType range_type) {
+const char* GetCompactionProximalOutputRangeTypeString(
+    Compaction::ProximalOutputRangeType range_type) {
   switch (range_type) {
-    case Compaction::PenultimateOutputRangeType::kNotSupported:
+    case Compaction::ProximalOutputRangeType::kNotSupported:
       return "NotSupported";
-    case Compaction::PenultimateOutputRangeType::kFullRange:
+    case Compaction::ProximalOutputRangeType::kFullRange:
       return "FullRange";
-    case Compaction::PenultimateOutputRangeType::kNonLastRange:
+    case Compaction::ProximalOutputRangeType::kNonLastRange:
       return "NonLastRange";
-    case Compaction::PenultimateOutputRangeType::kDisabled:
+    case Compaction::ProximalOutputRangeType::kDisabled:
       return "Disabled";
     default:
       assert(false);
@@ -147,7 +147,7 @@ CompactionJob::CompactionJob(
     BlobFileCompletionCallback* blob_callback, int* bg_compaction_scheduled,
     int* bg_bottom_compaction_scheduled)
     : compact_(new CompactionState(compaction)),
-      compaction_stats_(compaction->compaction_reason(), 1),
+      internal_stats_(compaction->compaction_reason(), 1),
       db_options_(db_options),
       mutable_db_options_copy_(mutable_db_options),
       log_buffer_(log_buffer),
@@ -155,7 +155,7 @@ CompactionJob::CompactionJob(
       stats_(stats),
       bottommost_level_(false),
       write_hint_(Env::WLTH_NOT_SET),
-      compaction_job_stats_(compaction_job_stats),
+      job_stats_(compaction_job_stats),
       job_id_(job_id),
       dbname_(dbname),
       db_id_(db_id),
@@ -191,7 +191,7 @@ CompactionJob::CompactionJob(
       extra_num_subcompaction_threads_reserved_(0),
       bg_compaction_scheduled_(bg_compaction_scheduled),
       bg_bottom_compaction_scheduled_(bg_bottom_compaction_scheduled) {
-  assert(compaction_job_stats_ != nullptr);
+  assert(job_stats_ != nullptr);
   assert(log_buffer_ != nullptr);
 
   const auto* cfd = compact_->compaction->column_family_data();
@@ -240,9 +240,8 @@ void CompactionJob::ReportStartedCompaction(Compaction* compaction) {
   // to ensure GetThreadList() can always show them all together.
   ThreadStatusUtil::SetThreadOperation(ThreadStatus::OP_COMPACTION);
 
-  compaction_job_stats_->is_manual_compaction =
-      compaction->is_manual_compaction();
-  compaction_job_stats_->is_full_compaction = compaction->is_full_compaction();
+  job_stats_->is_manual_compaction = compaction->is_manual_compaction();
+  job_stats_->is_full_compaction = compaction->is_full_compaction();
 }
 
 void CompactionJob::Prepare(
@@ -260,7 +259,8 @@ void CompactionJob::Prepare(
   assert(storage_info);
   assert(storage_info->NumLevelFiles(compact_->compaction->level()) > 0);
 
-  write_hint_ = storage_info->CalculateSSTWriteHint(c->output_level());
+  write_hint_ = storage_info->CalculateSSTWriteHint(
+      c->output_level(), db_options_.calculate_sst_write_lifetime_hint_set);
   bottommost_level_ = c->bottommost_level();
 
   if (!known_single_subcompact.has_value() && c->ShouldFormSubcompactions()) {
@@ -301,8 +301,7 @@ void CompactionJob::Prepare(
   SequenceNumber preserve_time_min_seqno = kMaxSequenceNumber;
   SequenceNumber preclude_last_level_min_seqno = kMaxSequenceNumber;
   uint64_t preserve_time_duration =
-      std::max(c->mutable_cf_options().preserve_internal_time_seconds,
-               c->mutable_cf_options().preclude_last_level_data_seconds);
+      MinAndMaxPreserveSeconds(c->mutable_cf_options()).max_preserve_seconds;
 
   if (preserve_time_duration > 0) {
     const ReadOptions read_options(Env::IOActivity::kCompaction);
@@ -379,8 +378,8 @@ void CompactionJob::Prepare(
   }
   // Now combine what we would like to preclude from last level with what we
   // can safely support without dangerously moving data back up the LSM tree,
-  // to get the final seqno threshold for penultimate vs. last. In particular,
-  // when the reserved output key range for the penultimate level does not
+  // to get the final seqno threshold for proximal vs. last. In particular,
+  // when the reserved output key range for the proximal level does not
   // include the entire last level input key range, we need to keep entries
   // already in the last level there. (Even allowing within-range entries to
   // move back up could cause problems with range tombstones. Perhaps it
@@ -389,8 +388,8 @@ void CompactionJob::Prepare(
   // tracking and complexity to CompactionIterator that is probably not
   // worthwhile overall. Correctness is also more clear when splitting by
   // seqno threshold.)
-  penultimate_after_seqno_ = std::max(preclude_last_level_min_seqno,
-                                      c->GetKeepInLastLevelThroughSeqno());
+  proximal_after_seqno_ = std::max(preclude_last_level_min_seqno,
+                                   c->GetKeepInLastLevelThroughSeqno());
 
   options_file_number_ = versions_->options_file_number();
 }
@@ -696,17 +695,17 @@ Status CompactionJob::Run() {
     thread.join();
   }
 
-  compaction_stats_.SetMicros(db_options_.clock->NowMicros() - start_micros);
+  internal_stats_.SetMicros(db_options_.clock->NowMicros() - start_micros);
 
   for (auto& state : compact_->sub_compact_states) {
-    compaction_stats_.AddCpuMicros(state.compaction_job_stats.cpu_micros);
+    internal_stats_.AddCpuMicros(state.compaction_job_stats.cpu_micros);
     state.RemoveLastEmptyOutput();
   }
 
   RecordTimeToHistogram(stats_, COMPACTION_TIME,
-                        compaction_stats_.stats.micros);
+                        internal_stats_.output_level_stats.micros);
   RecordTimeToHistogram(stats_, COMPACTION_CPU_TIME,
-                        compaction_stats_.stats.cpu_micros);
+                        internal_stats_.output_level_stats.cpu_micros);
 
   TEST_SYNC_POINT("CompactionJob::Run:BeforeVerify");
 
@@ -841,7 +840,6 @@ Status CompactionJob::Run() {
   TEST_SYNC_POINT("CompactionJob::ReleaseSubcompactionResources:0");
   TEST_SYNC_POINT("CompactionJob::ReleaseSubcompactionResources:1");
 
-  TablePropertiesCollection tp;
   for (const auto& state : compact_->sub_compact_states) {
     for (const auto& output : state.GetOutputs()) {
       auto fn =
@@ -856,46 +854,60 @@ Status CompactionJob::Run() {
   // compaction_service is set. We now know whether each sub_compaction was
   // done remotely or not. Reset is_remote_compaction back to false and allow
   // AggregateCompactionStats() to set the right value.
-  compaction_job_stats_->is_remote_compaction = false;
+  job_stats_->is_remote_compaction = false;
 
   // Finish up all bookkeeping to unify the subcompaction results.
-  compact_->AggregateCompactionStats(compaction_stats_, *compaction_job_stats_);
+  compact_->AggregateCompactionStats(internal_stats_, *job_stats_);
+
   uint64_t num_input_range_del = 0;
-  bool ok = UpdateCompactionStats(&num_input_range_del);
-  // (Sub)compactions returned ok, do sanity check on the number of input keys.
-  if (status.ok() && ok && compaction_job_stats_->has_num_input_records) {
-    size_t ts_sz = compact_->compaction->column_family_data()
-                       ->user_comparator()
-                       ->timestamp_size();
-    // When trim_ts_ is non-empty, CompactionIterator takes
-    // HistoryTrimmingIterator as input iterator and sees a trimmed view of
-    // input keys. So the number of keys it processed is not suitable for
-    // verification here.
-    // TODO: support verification when trim_ts_ is non-empty.
-    if (!(ts_sz > 0 && !trim_ts_.empty())) {
-      assert(compaction_stats_.stats.num_input_records > 0);
-      // TODO: verify the number of range deletion entries.
-      uint64_t expected =
-          compaction_stats_.stats.num_input_records - num_input_range_del;
-      uint64_t actual = compaction_job_stats_->num_input_records;
-      if (expected != actual) {
-        char scratch[2345];
-        compact_->compaction->Summary(scratch, sizeof(scratch));
-        std::string msg =
-            "Compaction number of input keys does not match "
-            "number of keys processed. Expected " +
-            std::to_string(expected) + " but processed " +
-            std::to_string(actual) + ". Compaction summary: " + scratch;
+  bool ok = BuildStatsFromInputTableProperties(&num_input_range_del);
+  // (Sub)compactions returned ok, do sanity check on the number of input
+  // keys.
+  if (status.ok() && ok) {
+    if (job_stats_->has_num_input_records) {
+      status = VerifyInputRecordCount(num_input_range_del);
+      if (!status.ok()) {
         ROCKS_LOG_WARN(
             db_options_.info_log, "[%s] [JOB %d] Compaction with status: %s",
             compact_->compaction->column_family_data()->GetName().c_str(),
-            job_context_->job_id, msg.c_str());
-        if (db_options_.compaction_verify_record_count) {
-          status = Status::Corruption(msg);
-        }
+            job_context_->job_id, status.ToString().c_str());
       }
     }
+    UpdateCompactionJobInputStats(internal_stats_, num_input_range_del);
   }
+  UpdateCompactionJobOutputStats(internal_stats_);
+
+  // Verify number of output records
+  if (status.ok() && db_options_.compaction_verify_record_count) {
+    uint64_t total_output_num = 0;
+    for (const auto& state : compact_->sub_compact_states) {
+      for (const auto& output : state.GetOutputs()) {
+        total_output_num += output.table_properties->num_entries -
+                            output.table_properties->num_range_deletions;
+      }
+    }
+
+    uint64_t expected = internal_stats_.output_level_stats.num_output_records;
+    if (internal_stats_.has_proximal_level_output) {
+      expected += internal_stats_.proximal_level_stats.num_output_records;
+    }
+    if (expected != total_output_num) {
+      char scratch[2345];
+      compact_->compaction->Summary(scratch, sizeof(scratch));
+      std::string msg =
+          "Number of keys in compaction output SST files does not match "
+          "number of keys added. Expected " +
+          std::to_string(expected) + " but there are " +
+          std::to_string(total_output_num) +
+          " in output SST files. Compaction summary: " + scratch;
+      ROCKS_LOG_WARN(
+          db_options_.info_log, "[%s] [JOB %d] Compaction with status: %s",
+          compact_->compaction->column_family_data()->GetName().c_str(),
+          job_context_->job_id, msg.c_str());
+      status = Status::Corruption(msg);
+    }
+  }
+
   RecordCompactionIOStats();
   LogFlush(db_options_.info_log);
   TEST_SYNC_POINT("CompactionJob::Run():End");
@@ -917,7 +929,7 @@ Status CompactionJob::Install(bool* compaction_released) {
 
   int output_level = compact_->compaction->output_level();
   cfd->internal_stats()->AddCompactionStats(output_level, thread_pri_,
-                                            compaction_stats_);
+                                            internal_stats_);
 
   if (status.ok()) {
     status = InstallCompactionResults(compaction_released);
@@ -928,7 +940,7 @@ Status CompactionJob::Install(bool* compaction_released) {
 
   VersionStorageInfo::LevelSummaryStorage tmp;
   auto vstorage = cfd->current()->storage_info();
-  const auto& stats = compaction_stats_.stats;
+  const auto& stats = internal_stats_.output_level_stats;
 
   double read_write_amp = 0.0;
   double write_amp = 0.0;
@@ -994,19 +1006,20 @@ Status CompactionJob::Install(bool* compaction_released) {
         blob_files.back()->GetBlobFileNumber());
   }
 
-  if (compaction_stats_.has_penultimate_level_output) {
-    ROCKS_LOG_BUFFER(
-        log_buffer_,
-        "[%s] has Penultimate Level output: %" PRIu64
-        ", level %d, number of files: %" PRIu64 ", number of records: %" PRIu64,
-        column_family_name.c_str(),
-        compaction_stats_.penultimate_level_stats.bytes_written,
-        compact_->compaction->GetPenultimateLevel(),
-        compaction_stats_.penultimate_level_stats.num_output_files,
-        compaction_stats_.penultimate_level_stats.num_output_records);
+  if (internal_stats_.has_proximal_level_output) {
+    ROCKS_LOG_BUFFER(log_buffer_,
+                     "[%s] has Proximal Level output: %" PRIu64
+                     ", level %d, number of files: %" PRIu64
+                     ", number of records: %" PRIu64,
+                     column_family_name.c_str(),
+                     internal_stats_.proximal_level_stats.bytes_written,
+                     compact_->compaction->GetProximalLevel(),
+                     internal_stats_.proximal_level_stats.num_output_files,
+                     internal_stats_.proximal_level_stats.num_output_records);
   }
 
-  UpdateCompactionJobStats(stats);
+  TEST_SYNC_POINT_CALLBACK(
+      "CompactionJob::Install:AfterUpdateCompactionJobStats", job_stats_);
 
   auto stream = event_logger_->LogToBuffer(log_buffer_, 8192);
   stream << "job" << job_id_ << "event" << "compaction_finished"
@@ -1028,17 +1041,16 @@ Status CompactionJob::Install(bool* compaction_released) {
          << CompressionTypeToString(compact_->compaction->output_compression());
 
   stream << "num_single_delete_mismatches"
-         << compaction_job_stats_->num_single_del_mismatch;
+         << job_stats_->num_single_del_mismatch;
   stream << "num_single_delete_fallthrough"
-         << compaction_job_stats_->num_single_del_fallthru;
+         << job_stats_->num_single_del_fallthru;
 
   if (measure_io_stats_) {
-    stream << "file_write_nanos" << compaction_job_stats_->file_write_nanos;
-    stream << "file_range_sync_nanos"
-           << compaction_job_stats_->file_range_sync_nanos;
-    stream << "file_fsync_nanos" << compaction_job_stats_->file_fsync_nanos;
+    stream << "file_write_nanos" << job_stats_->file_write_nanos;
+    stream << "file_range_sync_nanos" << job_stats_->file_range_sync_nanos;
+    stream << "file_fsync_nanos" << job_stats_->file_fsync_nanos;
     stream << "file_prepare_write_nanos"
-           << compaction_job_stats_->file_prepare_write_nanos;
+           << job_stats_->file_prepare_write_nanos;
   }
 
   stream << "lsm_state";
@@ -1056,16 +1068,16 @@ Status CompactionJob::Install(bool* compaction_released) {
     stream << "blob_file_tail" << blob_files.back()->GetBlobFileNumber();
   }
 
-  if (compaction_stats_.has_penultimate_level_output) {
+  if (internal_stats_.has_proximal_level_output) {
     InternalStats::CompactionStats& pl_stats =
-        compaction_stats_.penultimate_level_stats;
-    stream << "penultimate_level_num_output_files" << pl_stats.num_output_files;
-    stream << "penultimate_level_bytes_written" << pl_stats.bytes_written;
-    stream << "penultimate_level_num_output_records"
+        internal_stats_.proximal_level_stats;
+    stream << "proximal_level_num_output_files" << pl_stats.num_output_files;
+    stream << "proximal_level_bytes_written" << pl_stats.bytes_written;
+    stream << "proximal_level_num_output_records"
            << pl_stats.num_output_records;
-    stream << "penultimate_level_num_output_files_blob"
+    stream << "proximal_level_num_output_files_blob"
            << pl_stats.num_output_files_blob;
-    stream << "penultimate_level_bytes_written_blob"
+    stream << "proximal_level_bytes_written_blob"
            << pl_stats.bytes_written_blob;
   }
 
@@ -1130,8 +1142,7 @@ void CompactionJob::ProcessKeyValueCompaction(SubcompactionState* sub_compact) {
   if (db_options_.compaction_service) {
     CompactionServiceJobStatus comp_status =
         ProcessKeyValueCompactionWithCompactionService(sub_compact);
-    if (comp_status == CompactionServiceJobStatus::kSuccess ||
-        comp_status == CompactionServiceJobStatus::kFailure) {
+    if (comp_status != CompactionServiceJobStatus::kUseLocal) {
       return;
     }
     // fallback to local compaction
@@ -1314,7 +1325,7 @@ void CompactionJob::ProcessKeyValueCompaction(SubcompactionState* sub_compact) {
 
   std::vector<std::string> blob_file_paths;
 
-  // TODO: BlobDB to support output_to_penultimate_level compaction, which needs
+  // TODO: BlobDB to support output_to_proximal_level compaction, which needs
   //  2 builders, so may need to move to `CompactionOutputs`
   std::unique_ptr<BlobFileBuilder> blob_file_builder(
       (mutable_cf_options.enable_blob_files &&
@@ -1399,30 +1410,30 @@ void CompactionJob::ProcessKeyValueCompaction(SubcompactionState* sub_compact) {
     }
 
     const auto& ikey = c_iter->ikey();
-    bool use_penultimate_output = ikey.sequence > penultimate_after_seqno_;
+    bool use_proximal_output = ikey.sequence > proximal_after_seqno_;
 #ifndef NDEBUG
     if (sub_compact->compaction->SupportsPerKeyPlacement()) {
       // Could be overridden by unittest
       PerKeyPlacementContext context(sub_compact->compaction->output_level(),
                                      ikey.user_key, c_iter->value(),
-                                     ikey.sequence, use_penultimate_output);
+                                     ikey.sequence, use_proximal_output);
       TEST_SYNC_POINT_CALLBACK("CompactionIterator::PrepareOutput.context",
                                &context);
-      if (use_penultimate_output) {
-        // Verify that entries sent to the penultimate level are within the
+      if (use_proximal_output) {
+        // Verify that entries sent to the proximal level are within the
         // allowed range (because the input key range of the last level could
-        // be larger than the allowed output key range of the penultimate
+        // be larger than the allowed output key range of the proximal
         // level). This check uses user keys (ignores sequence numbers) because
         // compaction boundaries are a "clean cut" between user keys (see
         // CompactionPicker::ExpandInputsToCleanCut()), which is especially
         // important when preferred sequence numbers has been swapped in for
         // kTypeValuePreferredSeqno / TimedPut.
-        sub_compact->compaction->TEST_AssertWithinPenultimateLevelOutputRange(
+        sub_compact->compaction->TEST_AssertWithinProximalLevelOutputRange(
             c_iter->user_key());
       }
     } else {
-      assert(penultimate_after_seqno_ == kMaxSequenceNumber);
-      assert(!use_penultimate_output);
+      assert(proximal_after_seqno_ == kMaxSequenceNumber);
+      assert(!use_proximal_output);
     }
 #endif  // NDEBUG
 
@@ -1431,7 +1442,7 @@ void CompactionJob::ProcessKeyValueCompaction(SubcompactionState* sub_compact) {
     // and `close_file_func`.
     // TODO: it would be better to have the compaction file open/close moved
     // into `CompactionOutputs` which has the output file information.
-    status = sub_compact->AddToOutput(*c_iter, use_penultimate_output,
+    status = sub_compact->AddToOutput(*c_iter, use_proximal_output,
                                       open_file_func, close_file_func);
     if (!status.ok()) {
       break;
@@ -1643,15 +1654,15 @@ Status CompactionJob::FinishCompactionOutputFile(
     std::pair<SequenceNumber, SequenceNumber> keep_seqno_range{
         0, kMaxSequenceNumber};
     if (sub_compact->compaction->SupportsPerKeyPlacement()) {
-      if (outputs.IsPenultimateLevel()) {
-        keep_seqno_range.first = penultimate_after_seqno_;
+      if (outputs.IsProximalLevel()) {
+        keep_seqno_range.first = proximal_after_seqno_;
       } else {
-        keep_seqno_range.second = penultimate_after_seqno_;
+        keep_seqno_range.second = proximal_after_seqno_;
       }
     }
     CompactionIterationStats range_del_out_stats;
     // NOTE1: Use `bottommost_level_ = true` for both bottommost and
-    // output_to_penultimate_level compaction here, as it's only used to decide
+    // output_to_proximal_level compaction here, as it's only used to decide
     // if range dels could be dropped. (Logically, we are taking a single sorted
     // run returned from CompactionIterator and physically splitting it between
     // two output levels.)
@@ -1814,22 +1825,22 @@ Status CompactionJob::InstallCompactionResults(bool* compaction_released) {
 
   {
     Compaction::InputLevelSummaryBuffer inputs_summary;
-    if (compaction_stats_.has_penultimate_level_output) {
+    if (internal_stats_.has_proximal_level_output) {
       ROCKS_LOG_BUFFER(
           log_buffer_,
-          "[%s] [JOB %d] Compacted %s => output_to_penultimate_level: %" PRIu64
+          "[%s] [JOB %d] Compacted %s => output_to_proximal_level: %" PRIu64
           " bytes + last: %" PRIu64 " bytes. Total: %" PRIu64 " bytes",
           compaction->column_family_data()->GetName().c_str(), job_id_,
           compaction->InputLevelSummary(&inputs_summary),
-          compaction_stats_.penultimate_level_stats.bytes_written,
-          compaction_stats_.stats.bytes_written,
-          compaction_stats_.TotalBytesWritten());
+          internal_stats_.proximal_level_stats.bytes_written,
+          internal_stats_.output_level_stats.bytes_written,
+          internal_stats_.TotalBytesWritten());
     } else {
       ROCKS_LOG_BUFFER(log_buffer_,
                        "[%s] [JOB %d] Compacted %s => %" PRIu64 " bytes",
                        compaction->column_family_data()->GetName().c_str(),
                        job_id_, compaction->InputLevelSummary(&inputs_summary),
-                       compaction_stats_.TotalBytesWritten());
+                       internal_stats_.TotalBytesWritten());
     }
   }
 
@@ -1948,11 +1959,11 @@ Status CompactionJob::OpenCompactionOutputFile(SubcompactionState* sub_compact,
   // Here last_level_temperature supersedes default_write_temperature, when
   // enabled and applicable
   if (last_level_temp != Temperature::kUnknown &&
-      sub_compact->compaction->is_last_level() &&
-      !outputs.IsPenultimateLevel()) {
+      sub_compact->compaction->is_last_level() && !outputs.IsProximalLevel()) {
     temperature = last_level_temp;
   }
   fo_copy.temperature = temperature;
+  fo_copy.write_hint = write_hint_;
 
   Status s;
   IOStatus io_s = NewWritableFile(fs_.get(), fname, &writable_file, fo_copy);
@@ -2038,7 +2049,9 @@ Status CompactionJob::OpenCompactionOutputFile(SubcompactionState* sub_compact,
   }
 
   writable_file->SetIOPriority(GetRateLimiterPriority());
-  writable_file->SetWriteLifeTimeHint(write_hint_);
+  // Subsequent attempts to override the hint via SetWriteLifeTimeHint
+  // with the very same value will be ignored by the fs.
+  writable_file->SetWriteLifeTimeHint(fo_copy.write_hint);
   FileTypeSet tmp_set = db_options_.checksum_handoff_file_types;
   writable_file->SetPreallocationBlockSize(static_cast<size_t>(
       sub_compact->compaction->OutputFilePreallocationSize()));
@@ -2063,7 +2076,7 @@ Status CompactionJob::OpenCompactionOutputFile(SubcompactionState* sub_compact,
       bottommost_level_, TableFileCreationReason::kCompaction,
       0 /* oldest_key_time */, current_time, db_id_, db_session_id_,
       sub_compact->compaction->max_output_file_size(), file_number,
-      penultimate_after_seqno_ /*last_level_inclusive_max_seqno_threshold*/);
+      proximal_after_seqno_ /*last_level_inclusive_max_seqno_threshold*/);
 
   outputs.NewBuilder(tboptions);
 
@@ -2087,12 +2100,13 @@ void CopyPrefix(const Slice& src, size_t prefix_length, std::string* dst) {
 }
 }  // namespace
 
-bool CompactionJob::UpdateCompactionStats(uint64_t* num_input_range_del) {
+bool CompactionJob::BuildStatsFromInputTableProperties(
+    uint64_t* num_input_range_del) {
   assert(compact_);
 
   Compaction* compaction = compact_->compaction;
-  compaction_stats_.stats.num_input_files_in_non_output_levels = 0;
-  compaction_stats_.stats.num_input_files_in_output_level = 0;
+  internal_stats_.output_level_stats.num_input_files_in_non_output_levels = 0;
+  internal_stats_.output_level_stats.num_input_files_in_output_level = 0;
 
   bool has_error = false;
   const ReadOptions read_options(Env::IOActivity::kCompaction);
@@ -2104,13 +2118,14 @@ bool CompactionJob::UpdateCompactionStats(uint64_t* num_input_range_del) {
     size_t num_input_files = flevel->num_files;
     uint64_t* bytes_read;
     if (compaction->level(input_level) != compaction->output_level()) {
-      compaction_stats_.stats.num_input_files_in_non_output_levels +=
+      internal_stats_.output_level_stats.num_input_files_in_non_output_levels +=
           static_cast<int>(num_input_files);
-      bytes_read = &compaction_stats_.stats.bytes_read_non_output_levels;
+      bytes_read =
+          &internal_stats_.output_level_stats.bytes_read_non_output_levels;
     } else {
-      compaction_stats_.stats.num_input_files_in_output_level +=
+      internal_stats_.output_level_stats.num_input_files_in_output_level +=
           static_cast<int>(num_input_files);
-      bytes_read = &compaction_stats_.stats.bytes_read_output_level;
+      bytes_read = &internal_stats_.output_level_stats.bytes_read_output_level;
     }
     for (size_t i = 0; i < num_input_files; ++i) {
       const FileMetaData* file_meta = flevel->files[i].file_metadata;
@@ -2130,7 +2145,8 @@ bool CompactionJob::UpdateCompactionStats(uint64_t* num_input_range_del) {
           has_error = true;
         }
       }
-      compaction_stats_.stats.num_input_records += file_input_entries;
+      internal_stats_.output_level_stats.num_input_records +=
+          file_input_entries;
       if (num_input_range_del) {
         *num_input_range_del += file_num_range_del;
       }
@@ -2141,62 +2157,122 @@ bool CompactionJob::UpdateCompactionStats(uint64_t* num_input_range_del) {
     size_t num_filtered_input_files = filtered_flevel.size();
     uint64_t* bytes_skipped;
     if (compaction->level(input_level) != compaction->output_level()) {
-      compaction_stats_.stats.num_filtered_input_files_in_non_output_levels +=
+      internal_stats_.output_level_stats
+          .num_filtered_input_files_in_non_output_levels +=
           static_cast<int>(num_filtered_input_files);
-      bytes_skipped = &compaction_stats_.stats.bytes_skipped_non_output_levels;
+      bytes_skipped =
+          &internal_stats_.output_level_stats.bytes_skipped_non_output_levels;
     } else {
-      compaction_stats_.stats.num_filtered_input_files_in_output_level +=
+      internal_stats_.output_level_stats
+          .num_filtered_input_files_in_output_level +=
           static_cast<int>(num_filtered_input_files);
-      bytes_skipped = &compaction_stats_.stats.bytes_skipped_output_level;
+      bytes_skipped =
+          &internal_stats_.output_level_stats.bytes_skipped_output_level;
     }
     for (const FileMetaData* filtered_file_meta : filtered_flevel) {
       *bytes_skipped += filtered_file_meta->fd.GetFileSize();
     }
   }
 
-  assert(compaction_job_stats_);
-  compaction_stats_.stats.bytes_read_blob =
-      compaction_job_stats_->total_blob_bytes_read;
-
-  compaction_stats_.stats.num_dropped_records =
-      compaction_stats_.DroppedRecords();
+  // TODO - find a better place to set these two
+  assert(job_stats_);
+  internal_stats_.output_level_stats.bytes_read_blob =
+      job_stats_->total_blob_bytes_read;
+  internal_stats_.output_level_stats.num_dropped_records =
+      internal_stats_.DroppedRecords();
   return !has_error;
 }
 
-void CompactionJob::UpdateCompactionJobStats(
-    const InternalStats::CompactionStats& stats) const {
-  compaction_job_stats_->elapsed_micros = stats.micros;
-
+void CompactionJob::UpdateCompactionJobInputStats(
+    const InternalStats::CompactionStatsFull& internal_stats,
+    uint64_t num_input_range_del) const {
+  assert(job_stats_);
   // input information
-  compaction_job_stats_->total_input_bytes =
-      stats.bytes_read_non_output_levels + stats.bytes_read_output_level;
-  compaction_job_stats_->num_input_records = stats.num_input_records;
-  compaction_job_stats_->num_input_files =
-      stats.num_input_files_in_non_output_levels +
-      stats.num_input_files_in_output_level;
-  compaction_job_stats_->num_input_files_at_output_level =
-      stats.num_input_files_in_output_level;
-  compaction_job_stats_->num_filtered_input_files =
-      stats.num_filtered_input_files_in_non_output_levels +
-      stats.num_filtered_input_files_in_output_level;
-  compaction_job_stats_->num_filtered_input_files_at_output_level =
-      stats.num_filtered_input_files_in_output_level;
-  compaction_job_stats_->total_skipped_input_bytes =
-      stats.bytes_skipped_non_output_levels + stats.bytes_skipped_output_level;
+  job_stats_->total_input_bytes =
+      internal_stats.output_level_stats.bytes_read_non_output_levels +
+      internal_stats.output_level_stats.bytes_read_output_level;
+  job_stats_->num_input_records =
+      internal_stats.output_level_stats.num_input_records - num_input_range_del;
+  job_stats_->num_input_files =
+      internal_stats.output_level_stats.num_input_files_in_non_output_levels +
+      internal_stats.output_level_stats.num_input_files_in_output_level;
+  job_stats_->num_input_files_at_output_level =
+      internal_stats.output_level_stats.num_input_files_in_output_level;
+  job_stats_->num_filtered_input_files =
+      internal_stats.output_level_stats
+          .num_filtered_input_files_in_non_output_levels +
+      internal_stats.output_level_stats
+          .num_filtered_input_files_in_output_level;
+  job_stats_->num_filtered_input_files_at_output_level =
+      internal_stats.output_level_stats
+          .num_filtered_input_files_in_output_level;
+  job_stats_->total_skipped_input_bytes =
+      internal_stats.output_level_stats.bytes_skipped_non_output_levels +
+      internal_stats.output_level_stats.bytes_skipped_output_level;
+
+  if (internal_stats.has_proximal_level_output) {
+    job_stats_->total_input_bytes +=
+        internal_stats.proximal_level_stats.bytes_read_non_output_levels +
+        internal_stats.proximal_level_stats.bytes_read_output_level;
+    job_stats_->num_input_records +=
+        internal_stats.proximal_level_stats.num_input_records;
+    job_stats_->num_input_files +=
+        internal_stats.proximal_level_stats
+            .num_input_files_in_non_output_levels +
+        internal_stats.proximal_level_stats.num_input_files_in_output_level;
+    job_stats_->num_input_files_at_output_level +=
+        internal_stats.proximal_level_stats.num_input_files_in_output_level;
+    job_stats_->num_filtered_input_files +=
+        internal_stats.proximal_level_stats
+            .num_filtered_input_files_in_non_output_levels +
+        internal_stats.proximal_level_stats
+            .num_filtered_input_files_in_output_level;
+    job_stats_->num_filtered_input_files_at_output_level +=
+        internal_stats.proximal_level_stats
+            .num_filtered_input_files_in_output_level;
+    job_stats_->total_skipped_input_bytes +=
+        internal_stats.proximal_level_stats.bytes_skipped_non_output_levels +
+        internal_stats.proximal_level_stats.bytes_skipped_output_level;
+  }
+}
+
+void CompactionJob::UpdateCompactionJobOutputStats(
+    const InternalStats::CompactionStatsFull& internal_stats) const {
+  assert(job_stats_);
+  job_stats_->elapsed_micros = internal_stats.output_level_stats.micros;
+  job_stats_->cpu_micros = internal_stats.output_level_stats.cpu_micros;
 
   // output information
-  compaction_job_stats_->total_output_bytes = stats.bytes_written;
-  compaction_job_stats_->total_output_bytes_blob = stats.bytes_written_blob;
-  compaction_job_stats_->num_output_records = stats.num_output_records;
-  compaction_job_stats_->num_output_files = stats.num_output_files;
-  compaction_job_stats_->num_output_files_blob = stats.num_output_files_blob;
+  job_stats_->total_output_bytes =
+      internal_stats.output_level_stats.bytes_written;
+  job_stats_->total_output_bytes_blob =
+      internal_stats.output_level_stats.bytes_written_blob;
+  job_stats_->num_output_records =
+      internal_stats.output_level_stats.num_output_records;
+  job_stats_->num_output_files =
+      internal_stats.output_level_stats.num_output_files;
+  job_stats_->num_output_files_blob =
+      internal_stats.output_level_stats.num_output_files_blob;
 
-  if (stats.num_output_files > 0) {
+  if (internal_stats.has_proximal_level_output) {
+    job_stats_->total_output_bytes +=
+        internal_stats.proximal_level_stats.bytes_written;
+    job_stats_->total_output_bytes_blob +=
+        internal_stats.proximal_level_stats.bytes_written_blob;
+    job_stats_->num_output_records +=
+        internal_stats.proximal_level_stats.num_output_records;
+    job_stats_->num_output_files +=
+        internal_stats.proximal_level_stats.num_output_files;
+    job_stats_->num_output_files_blob +=
+        internal_stats.proximal_level_stats.num_output_files_blob;
+  }
+
+  if (job_stats_->num_output_files > 0) {
     CopyPrefix(compact_->SmallestUserKey(),
                CompactionJobStats::kMaxPrefixLength,
-               &compaction_job_stats_->smallest_output_key_prefix);
+               &job_stats_->smallest_output_key_prefix);
     CopyPrefix(compact_->LargestUserKey(), CompactionJobStats::kMaxPrefixLength,
-               &compaction_job_stats_->largest_output_key_prefix);
+               &job_stats_->largest_output_key_prefix);
   }
 }
 
@@ -2234,19 +2310,19 @@ void CompactionJob::LogCompaction() {
                    ? int64_t{-1}  // Use -1 for "none"
                    : static_cast<int64_t>(existing_snapshots_[0]));
     if (compaction->SupportsPerKeyPlacement()) {
-      stream << "prenultimate_after_seqno" << penultimate_after_seqno_;
+      stream << "proximal_after_seqno" << proximal_after_seqno_;
       stream << "preserve_seqno_after" << preserve_seqno_after_;
-      stream << "penultimate_output_level" << compaction->GetPenultimateLevel();
-      stream << "penultimate_output_range"
-             << GetCompactionPenultimateOutputRangeTypeString(
-                    compaction->GetPenultimateOutputRangeType());
+      stream << "proximal_output_level" << compaction->GetProximalLevel();
+      stream << "proximal_output_range"
+             << GetCompactionProximalOutputRangeTypeString(
+                    compaction->GetProximalOutputRangeType());
 
-      if (compaction->GetPenultimateOutputRangeType() ==
-          Compaction::PenultimateOutputRangeType::kDisabled) {
+      if (compaction->GetProximalOutputRangeType() ==
+          Compaction::ProximalOutputRangeType::kDisabled) {
         ROCKS_LOG_WARN(
             db_options_.info_log,
-            "[%s] [JOB %d] Penultimate level output is disabled, likely "
-            "because of the range conflict in the penultimate level",
+            "[%s] [JOB %d] Proximal level output is disabled, likely "
+            "because of the range conflict in the proximal level",
             cfd->GetName().c_str(), job_id_);
       }
     }
@@ -2271,4 +2347,36 @@ Env::IOPriority CompactionJob::GetRateLimiterPriority() {
   return Env::IO_LOW;
 }
 
+Status CompactionJob::VerifyInputRecordCount(
+    uint64_t num_input_range_del) const {
+  size_t ts_sz = compact_->compaction->column_family_data()
+                     ->user_comparator()
+                     ->timestamp_size();
+  // When trim_ts_ is non-empty, CompactionIterator takes
+  // HistoryTrimmingIterator as input iterator and sees a trimmed view of
+  // input keys. So the number of keys it processed is not suitable for
+  // verification here.
+  // TODO: support verification when trim_ts_ is non-empty.
+  if (!(ts_sz > 0 && !trim_ts_.empty())) {
+    assert(internal_stats_.output_level_stats.num_input_records > 0);
+    // TODO: verify the number of range deletion entries.
+    uint64_t expected = internal_stats_.output_level_stats.num_input_records -
+                        num_input_range_del;
+    uint64_t actual = job_stats_->num_input_records;
+    if (expected != actual) {
+      char scratch[2345];
+      compact_->compaction->Summary(scratch, sizeof(scratch));
+      std::string msg =
+          "Compaction number of input keys does not match "
+          "number of keys processed. Expected " +
+          std::to_string(expected) + " but processed " +
+          std::to_string(actual) + ". Compaction summary: " + scratch;
+      if (db_options_.compaction_verify_record_count) {
+        return Status::Corruption(msg);
+      }
+    }
+  }
+  return Status::OK();
+}
+
 }  // namespace ROCKSDB_NAMESPACE

db/compaction/compaction_job.h

@@ -67,7 +67,7 @@ class SubcompactionState;
 // if needed.
 //
 // CompactionJob has 2 main stats:
-// 1. CompactionJobStats compaction_job_stats_
+// 1. CompactionJobStats job_stats_
 //    CompactionJobStats is a public data structure which is part of Compaction
 //    event listener that rocksdb share the job stats with the user.
 //    Internally it's an aggregation of all the compaction_job_stats from each
@@ -81,7 +81,7 @@ class SubcompactionState;
 // +------------------------+     |
 // | CompactionJob          |     |          +------------------------+
 // |                        |     |          | SubcompactionState     |
-// |   compaction_job_stats +-----+          |                        |
+// |   job_stats            +-----+          |                        |
 // |                        |     +--------->|   compaction_job_stats |
 // |                        |     |          |                        |
 // +------------------------+     |          +------------------------+
@@ -98,16 +98,13 @@ class SubcompactionState;
 //                                +--------->+                        |
 //                                           +------------------------+
 //
-// 2. CompactionStatsFull compaction_stats_
+// 2. CompactionStatsFull internal_stats_
 //    `CompactionStatsFull` is an internal stats about the compaction, which
 //    is eventually sent to `ColumnFamilyData::internal_stats_` and used for
 //    logging and public metrics.
 //    Internally, it's an aggregation of stats_ from each `SubcompactionState`.
-//    It has 2 parts, normal stats about the main compaction information and
-//    the penultimate level output stats.
-//    `SubcompactionState` maintains the CompactionOutputs for normal output and
-//    the penultimate level output if exists, the per_level stats is
-//    stored with the outputs.
+//    It has 2 parts, ordinary output level stats and the proximal level output
+//    stats.
 //                                                +---------------------------+
 //                                                | SubcompactionState        |
 //                                                |                           |
@@ -119,15 +116,15 @@ class SubcompactionState;
 //                                            |   |                           |
 //                                            |   | +----------------------+  |
 // +--------------------------------+         |   | | CompactionOutputs    |  |
-// | CompactionJob                  |         |   | | (penultimate_level)  |  |
+// | CompactionJob                  |         |   | | (proximal_level)     |  |
 // |                                |    +--------->|   stats_             |  |
-// |   compaction_stats_            |    |    |   | +----------------------+  |
+// |   internal_stats_              |    |    |   | +----------------------+  |
 // |    +-------------------------+ |    |    |   |                           |
-// |    |stats (normal)           |------|----+   +---------------------------+
+// |    |output_level_stats       |------|----+   +---------------------------+
 // |    +-------------------------+ |    |    |
 // |                                |    |    |
 // |    +-------------------------+ |    |    |   +---------------------------+
-// |    |penultimate_level_stats  +------+    |   | SubcompactionState        |
+// |    |proximal_level_stats     |------+    |   | SubcompactionState        |
 // |    +-------------------------+ |    |    |   |                           |
 // |                                |    |    |   | +----------------------+  |
 // |                                |    |    |   | | CompactionOutputs    |  |
@@ -137,7 +134,7 @@ class SubcompactionState;
 //                                       |        |                           |
 //                                       |        | +----------------------+  |
 //                                       |        | | CompactionOutputs    |  |
-//                                       |        | | (penultimate_level)  |  |
+//                                       |        | | (proximal_level)     |  |
 //                                       +--------->|   stats_             |  |
 //                                                | +----------------------+  |
 //                                                |                           |
@@ -199,23 +196,9 @@ class CompactionJob {
   IOStatus io_status() const { return io_status_; }
 
  protected:
-  // Update the following stats in compaction_stats_.stats
-  // - num_input_files_in_non_output_levels
-  // - num_input_files_in_output_level
-  // - bytes_read_non_output_levels
-  // - bytes_read_output_level
-  // - num_input_records
-  // - bytes_read_blob
-  // - num_dropped_records
-  //
-  // @param num_input_range_del if non-null, will be set to the number of range
-  // deletion entries in this compaction input.
-  //
-  // Returns true iff compaction_stats_.stats.num_input_records and
-  // num_input_range_del are calculated successfully.
-  bool UpdateCompactionStats(uint64_t* num_input_range_del = nullptr);
-  virtual void UpdateCompactionJobStats(
-      const InternalStats::CompactionStats& stats) const;
+  void UpdateCompactionJobOutputStats(
+      const InternalStats::CompactionStatsFull& internal_stats) const;
+
   void LogCompaction();
   virtual void RecordCompactionIOStats();
   void CleanupCompaction();
@@ -224,7 +207,7 @@ class CompactionJob {
   void ProcessKeyValueCompaction(SubcompactionState* sub_compact);
 
   CompactionState* compact_;
-  InternalStats::CompactionStatsFull compaction_stats_;
+  InternalStats::CompactionStatsFull internal_stats_;
   const ImmutableDBOptions& db_options_;
   const MutableDBOptions mutable_db_options_copy_;
   LogBuffer* log_buffer_;
@@ -237,11 +220,37 @@ class CompactionJob {
 
   IOStatus io_status_;
 
-  CompactionJobStats* compaction_job_stats_;
+  CompactionJobStats* job_stats_;
 
  private:
   friend class CompactionJobTestBase;
 
+  // Collect the following stats from Input Table Properties
+  // - num_input_files_in_non_output_levels
+  // - num_input_files_in_output_level
+  // - bytes_read_non_output_levels
+  // - bytes_read_output_level
+  // - num_input_records
+  // - bytes_read_blob
+  // - num_dropped_records
+  // and set them in internal_stats_.output_level_stats
+  //
+  // @param num_input_range_del if non-null, will be set to the number of range
+  // deletion entries in this compaction input.
+  //
+  // Returns true iff internal_stats_.output_level_stats.num_input_records and
+  // num_input_range_del are calculated successfully.
+  //
+  // This should be called only once for compactions (not per subcompaction)
+  bool BuildStatsFromInputTableProperties(
+      uint64_t* num_input_range_del = nullptr);
+
+  void UpdateCompactionJobInputStats(
+      const InternalStats::CompactionStatsFull& internal_stats,
+      uint64_t num_input_range_del) const;
+
+  Status VerifyInputRecordCount(uint64_t num_input_range_del) const;
+
   // Generates a histogram representing potential divisions of key ranges from
   // the input. It adds the starting and/or ending keys of certain input files
   // to the working set and then finds the approximate size of data in between
@@ -363,8 +372,8 @@ class CompactionJob {
 
   // Minimal sequence number to preclude the data from the last level. If the
   // key has bigger (newer) sequence number than this, it will be precluded from
-  // the last level (output to penultimate level).
-  SequenceNumber penultimate_after_seqno_ = kMaxSequenceNumber;
+  // the last level (output to proximal level).
+  SequenceNumber proximal_after_seqno_ = kMaxSequenceNumber;
 
   // Options File Number used for Remote Compaction
   // Setting this requires DBMutex.
@@ -431,6 +440,8 @@ struct CompactionServiceOutputFile {
   bool marked_for_compaction;
   UniqueId64x2 unique_id{};
   TableProperties table_properties;
+  bool is_proximal_level_output;
+  Temperature file_temperature;
 
   CompactionServiceOutputFile() = default;
   CompactionServiceOutputFile(
@@ -440,7 +451,8 @@ struct CompactionServiceOutputFile {
       uint64_t _epoch_number, const std::string& _file_checksum,
       const std::string& _file_checksum_func_name, uint64_t _paranoid_hash,
       bool _marked_for_compaction, UniqueId64x2 _unique_id,
-      const TableProperties& _table_properties)
+      const TableProperties& _table_properties, bool _is_proximal_level_output,
+      Temperature _file_temperature)
       : file_name(name),
         smallest_seqno(smallest),
         largest_seqno(largest),
@@ -454,7 +466,9 @@ struct CompactionServiceOutputFile {
         paranoid_hash(_paranoid_hash),
         marked_for_compaction(_marked_for_compaction),
         unique_id(std::move(_unique_id)),
-        table_properties(_table_properties) {}
+        table_properties(_table_properties),
+        is_proximal_level_output(_is_proximal_level_output),
+        file_temperature(_file_temperature) {}
 };
 
 // CompactionServiceResult contains the compaction result from a different db
@@ -470,8 +484,21 @@ struct CompactionServiceResult {
 
   uint64_t bytes_read = 0;
   uint64_t bytes_written = 0;
+
+  // Job-level Compaction Stats.
+  //
+  // NOTE: Job level stats cannot be rebuilt from scratch by simply aggregating
+  // per-level stats due to some fields populated directly during compaction
+  // (e.g. RecordDroppedKeys()). This is why we need both job-level stats and
+  // per-level in the serialized result. If rebuilding job-level stats from
+  // per-level stats become possible in the future, consider deprecating this
+  // field.
   CompactionJobStats stats;
 
+  // Per-level Compaction Stats for both output_level_stats and
+  // proximal_level_stats
+  InternalStats::CompactionStatsFull internal_stats;
+
   // serialization interface to read and write the object
   static Status Read(const std::string& data_str, CompactionServiceResult* obj);
   Status Write(std::string* output);
@@ -517,9 +544,6 @@ class CompactionServiceCompactionJob : private CompactionJob {
  protected:
   void RecordCompactionIOStats() override;
 
-  void UpdateCompactionJobStats(
-      const InternalStats::CompactionStats& stats) const override;
-
  private:
   // Get table file name in output_path
   std::string GetTableFileName(uint64_t file_number) override;

db/compaction/compaction_job_test.cc

@@ -232,6 +232,11 @@ class CompactionJobTestBase : public testing::Test {
     // set default for the tests
     mutable_cf_options_.target_file_size_base = 1024 * 1024;
     mutable_cf_options_.max_compaction_bytes = 10 * 1024 * 1024;
+
+    // Turn off compaction_verify_record_count MockTables
+    if (table_type == TableTypeForTest::kMockTable) {
+      db_options_.compaction_verify_record_count = false;
+    }
   }
 
   void SetUp() override {
@@ -1474,7 +1479,7 @@ TEST_F(CompactionJobTest, OldestBlobFileNumber) {
                 /* expected_oldest_blob_file_numbers */ {19});
 }
 
-TEST_F(CompactionJobTest, VerifyPenultimateLevelOutput) {
+TEST_F(CompactionJobTest, VerifyProximalLevelOutput) {
   cf_options_.last_level_temperature = Temperature::kCold;
   SyncPoint::GetInstance()->SetCallBack(
       "Compaction::SupportsPerKeyPlacement:Enabled", [&](void* arg) {
@@ -1487,8 +1492,7 @@ TEST_F(CompactionJobTest, VerifyPenultimateLevelOutput) {
   SyncPoint::GetInstance()->SetCallBack(
       "CompactionIterator::PrepareOutput.context", [&](void* arg) {
         auto context = static_cast<PerKeyPlacementContext*>(arg);
-        context->output_to_penultimate_level =
-            context->seq_num > latest_cold_seq;
+        context->output_to_proximal_level = context->seq_num > latest_cold_seq;
       });
   SyncPoint::GetInstance()->EnableProcessing();
 
@@ -1534,11 +1538,11 @@ TEST_F(CompactionJobTest, VerifyPenultimateLevelOutput) {
       /*verify_func=*/[&](Compaction& comp) {
         for (char c = 'a'; c <= 'z'; c++) {
           if (c == 'a') {
-            comp.TEST_AssertWithinPenultimateLevelOutputRange(
+            comp.TEST_AssertWithinProximalLevelOutputRange(
                 "a", true /*expect_failure*/);
           } else {
             std::string c_str{c};
-            comp.TEST_AssertWithinPenultimateLevelOutputRange(c_str);
+            comp.TEST_AssertWithinProximalLevelOutputRange(c_str);
           }
         }
       });
@@ -1682,7 +1686,8 @@ TEST_F(CompactionJobTest, ResultSerialization) {
         file_checksum /* file_checksum */,
         file_checksum_func_name /* file_checksum_func_name */,
         rnd64.Uniform(UINT64_MAX) /* paranoid_hash */,
-        rnd.OneIn(2) /* marked_for_compaction */, id /* unique_id */, tp);
+        rnd.OneIn(2) /* marked_for_compaction */, id /* unique_id */, tp,
+        false /* is_proximal_level_output */, Temperature::kHot);
   }
   result.output_level = rnd.Uniform(10);
   result.output_path = rnd.RandomString(rnd.Uniform(kStrMaxLen));
@@ -1736,6 +1741,8 @@ TEST_F(CompactionJobTest, ResultSerialization) {
     ASSERT_EQ(deserialized_tmp.output_files[0].file_checksum, file_checksum);
     ASSERT_EQ(deserialized_tmp.output_files[0].file_checksum_func_name,
               file_checksum_func_name);
+    ASSERT_EQ(deserialized_tmp.output_files[0].file_temperature,
+              Temperature::kHot);
   }
 
   // Test unknown field

db/compaction/compaction_outputs.cc

@@ -54,7 +54,7 @@ Status CompactionOutputs::Finish(
   }
   current_output().finished = true;
   stats_.bytes_written += current_bytes;
-  stats_.num_output_files = outputs_.size();
+  stats_.num_output_files = static_cast<int>(outputs_.size());
 
   return s;
 }
@@ -792,8 +792,8 @@ void CompactionOutputs::FillFilesToCutForTtl() {
 }
 
 CompactionOutputs::CompactionOutputs(const Compaction* compaction,
-                                     const bool is_penultimate_level)
-    : compaction_(compaction), is_penultimate_level_(is_penultimate_level) {
+                                     const bool is_proximal_level)
+    : compaction_(compaction), is_proximal_level_(is_proximal_level) {
   partitioner_ = compaction->output_level() == 0
                      ? nullptr
                      : compaction->CreateSstPartitioner();

db/compaction/compaction_outputs.h

@@ -30,29 +30,32 @@ class CompactionOutputs {
   // compaction output file
   struct Output {
     Output(FileMetaData&& _meta, const InternalKeyComparator& _icmp,
-           bool _enable_hash, bool _finished, uint64_t precalculated_hash)
+           bool _enable_hash, bool _finished, uint64_t precalculated_hash,
+           bool _is_proximal_level)
         : meta(std::move(_meta)),
           validator(_icmp, _enable_hash, precalculated_hash),
-          finished(_finished) {}
+          finished(_finished),
+          is_proximal_level(_is_proximal_level) {}
     FileMetaData meta;
     OutputValidator validator;
     bool finished;
+    bool is_proximal_level;
     std::shared_ptr<const TableProperties> table_properties;
   };
 
   CompactionOutputs() = delete;
 
   explicit CompactionOutputs(const Compaction* compaction,
-                             const bool is_penultimate_level);
+                             const bool is_proximal_level);
 
-  bool IsPenultimateLevel() const { return is_penultimate_level_; }
+  bool IsProximalLevel() const { return is_proximal_level_; }
 
   // Add generated output to the list
   void AddOutput(FileMetaData&& meta, const InternalKeyComparator& icmp,
                  bool enable_hash, bool finished = false,
                  uint64_t precalculated_hash = 0) {
     outputs_.emplace_back(std::move(meta), icmp, enable_hash, finished,
-                          precalculated_hash);
+                          precalculated_hash, is_proximal_level_);
   }
 
   // Set new table builder for the current output
@@ -63,34 +66,29 @@ class CompactionOutputs {
     file_writer_.reset(writer);
   }
 
-  // TODO: Remove it when remote compaction support tiered compaction
-  void AddBytesWritten(uint64_t bytes) { stats_.bytes_written += bytes; }
-  void SetNumOutputRecords(uint64_t num) { stats_.num_output_records = num; }
-  void SetNumOutputFiles(uint64_t num) { stats_.num_output_files = num; }
-
   // TODO: Move the BlobDB builder into CompactionOutputs
   const std::vector<BlobFileAddition>& GetBlobFileAdditions() const {
-    if (is_penultimate_level_) {
+    if (is_proximal_level_) {
       assert(blob_file_additions_.empty());
     }
     return blob_file_additions_;
   }
 
   std::vector<BlobFileAddition>* GetBlobFileAdditionsPtr() {
-    assert(!is_penultimate_level_);
+    assert(!is_proximal_level_);
     return &blob_file_additions_;
   }
 
   bool HasBlobFileAdditions() const { return !blob_file_additions_.empty(); }
 
   BlobGarbageMeter* CreateBlobGarbageMeter() {
-    assert(!is_penultimate_level_);
+    assert(!is_proximal_level_);
     blob_garbage_meter_ = std::make_unique<BlobGarbageMeter>();
     return blob_garbage_meter_.get();
   }
 
   BlobGarbageMeter* GetBlobGarbageMeter() const {
-    if (is_penultimate_level_) {
+    if (is_proximal_level_) {
       // blobdb doesn't support per_key_placement yet
       assert(blob_garbage_meter_ == nullptr);
       return nullptr;
@@ -99,8 +97,9 @@ class CompactionOutputs {
   }
 
   void UpdateBlobStats() {
-    assert(!is_penultimate_level_);
-    stats_.num_output_files_blob = blob_file_additions_.size();
+    assert(!is_proximal_level_);
+    stats_.num_output_files_blob =
+        static_cast<int>(blob_file_additions_.size());
     for (const auto& blob : blob_file_additions_) {
       stats_.bytes_written_blob += blob.GetTotalBlobBytes();
     }
@@ -304,12 +303,12 @@ class CompactionOutputs {
   std::vector<BlobFileAddition> blob_file_additions_;
   std::unique_ptr<BlobGarbageMeter> blob_garbage_meter_;
 
-  // Basic compaction output stats for this level's outputs
-  InternalStats::CompactionOutputsStats stats_;
+  // Per level's output stat
+  InternalStats::CompactionStats stats_;
 
-  // indicate if this CompactionOutputs obj for penultimate_level, should always
+  // indicate if this CompactionOutputs obj for proximal_level, should always
   // be false if per_key_placement feature is not enabled.
-  const bool is_penultimate_level_;
+  const bool is_proximal_level_;
 
   // partitioner information
   std::string last_key_for_partitioner_;
@@ -363,7 +362,7 @@ class CompactionOutputs {
   std::vector<size_t> level_ptrs_;
 };
 
-// helper struct to concatenate the last level and penultimate level outputs
+// helper struct to concatenate the last level and proximal level outputs
 // which could be replaced by std::ranges::join_view() in c++20
 struct OutputIterator {
  public:

db/compaction/compaction_picker.cc

@@ -272,8 +272,8 @@ bool CompactionPicker::RangeOverlapWithCompaction(
       return true;
     }
     if (c->SupportsPerKeyPlacement()) {
-      if (c->OverlapPenultimateLevelOutputRange(smallest_user_key,
-                                                largest_user_key)) {
+      if (c->OverlapProximalLevelOutputRange(smallest_user_key,
+                                             largest_user_key)) {
         return true;
       }
     }
@@ -284,7 +284,7 @@ bool CompactionPicker::RangeOverlapWithCompaction(
 
 bool CompactionPicker::FilesRangeOverlapWithCompaction(
     const std::vector<CompactionInputFiles>& inputs, int level,
-    int penultimate_level) const {
+    int proximal_level) const {
   bool is_empty = true;
   for (auto& in : inputs) {
     if (!in.empty()) {
@@ -301,18 +301,18 @@ bool CompactionPicker::FilesRangeOverlapWithCompaction(
   //  files cannot be overlapped in the order of L0 files.
   InternalKey smallest, largest;
   GetRange(inputs, &smallest, &largest, Compaction::kInvalidLevel);
-  if (penultimate_level != Compaction::kInvalidLevel) {
+  if (proximal_level != Compaction::kInvalidLevel) {
     if (ioptions_.compaction_style == kCompactionStyleUniversal) {
       if (RangeOverlapWithCompaction(smallest.user_key(), largest.user_key(),
-                                     penultimate_level)) {
+                                     proximal_level)) {
         return true;
       }
     } else {
-      InternalKey penultimate_smallest, penultimate_largest;
-      GetRange(inputs, &penultimate_smallest, &penultimate_largest, level);
-      if (RangeOverlapWithCompaction(penultimate_smallest.user_key(),
-                                     penultimate_largest.user_key(),
-                                     penultimate_level)) {
+      InternalKey proximal_smallest, proximal_largest;
+      GetRange(inputs, &proximal_smallest, &proximal_largest, level);
+      if (RangeOverlapWithCompaction(proximal_smallest.user_key(),
+                                     proximal_largest.user_key(),
+                                     proximal_level)) {
         return true;
       }
     }
@@ -353,7 +353,7 @@ Compaction* CompactionPicker::CompactFiles(
   }
   assert(output_level == 0 || !FilesRangeOverlapWithCompaction(
                                   input_files, output_level,
-                                  Compaction::EvaluatePenultimateLevel(
+                                  Compaction::EvaluateProximalLevel(
                                       vstorage, mutable_cf_options, ioptions_,
                                       start_level, output_level)));
 #endif /* !NDEBUG */
@@ -659,9 +659,9 @@ Compaction* CompactionPicker::CompactRange(
     // overlaping outputs in the same level.
     if (FilesRangeOverlapWithCompaction(
             inputs, output_level,
-            Compaction::EvaluatePenultimateLevel(vstorage, mutable_cf_options,
-                                                 ioptions_, start_level,
-                                                 output_level))) {
+            Compaction::EvaluateProximalLevel(vstorage, mutable_cf_options,
+                                              ioptions_, start_level,
+                                              output_level))) {
       // This compaction output could potentially conflict with the output
       // of a currently running compaction, we cannot run it.
       *manual_conflict = true;
@@ -848,9 +848,9 @@ Compaction* CompactionPicker::CompactRange(
   // overlaping outputs in the same level.
   if (FilesRangeOverlapWithCompaction(
           compaction_inputs, output_level,
-          Compaction::EvaluatePenultimateLevel(vstorage, mutable_cf_options,
-                                               ioptions_, input_level,
-                                               output_level))) {
+          Compaction::EvaluateProximalLevel(vstorage, mutable_cf_options,
+                                            ioptions_, input_level,
+                                            output_level))) {
     // This compaction output could potentially conflict with the output
     // of a currently running compaction, we cannot run it.
     *manual_conflict = true;
@@ -1137,7 +1137,7 @@ Status CompactionPicker::SanitizeAndConvertCompactionInputFiles(
   if (output_level != 0 &&
       FilesRangeOverlapWithCompaction(
           *converted_input_files, output_level,
-          Compaction::EvaluatePenultimateLevel(
+          Compaction::EvaluateProximalLevel(
               version->storage_info(), version->GetMutableCFOptions(),
               ioptions_, (*converted_input_files)[0].level, output_level))) {
     return Status::Aborted(
@@ -1154,7 +1154,7 @@ void CompactionPicker::RegisterCompaction(Compaction* c) {
   assert(ioptions_.compaction_style != kCompactionStyleLevel ||
          c->output_level() == 0 ||
          !FilesRangeOverlapWithCompaction(*c->inputs(), c->output_level(),
-                                          c->GetPenultimateLevel()));
+                                          c->GetProximalLevel()));
   // CompactionReason::kExternalSstIngestion's start level is just a placeholder
   // number without actual meaning as file ingestion technically does not have
   // an input level like other compactions

db/compaction/compaction_picker.h

@@ -138,6 +138,12 @@ class CompactionPicker {
     return !level0_compactions_in_progress_.empty();
   }
 
+  // Is any compaction in progress
+  bool IsCompactionInProgress() const {
+    return !(level0_compactions_in_progress_.empty() &&
+             compactions_in_progress_.empty());
+  }
+
   // Return true if the passed key range overlap with a compaction output
   // that is currently running.
   bool RangeOverlapWithCompaction(const Slice& smallest_user_key,
@@ -190,7 +196,7 @@ class CompactionPicker {
   // key range of a currently running compaction.
   bool FilesRangeOverlapWithCompaction(
       const std::vector<CompactionInputFiles>& inputs, int level,
-      int penultimate_level) const;
+      int proximal_level) const;
 
   bool SetupOtherInputs(const std::string& cf_name,
                         const MutableCFOptions& mutable_cf_options,

db/compaction/compaction_picker_level.cc

@@ -414,9 +414,9 @@ void LevelCompactionBuilder::SetupOtherFilesWithRoundRobinExpansion() {
                                                     &tmp_start_level_inputs) ||
         compaction_picker_->FilesRangeOverlapWithCompaction(
             {tmp_start_level_inputs}, output_level_,
-            Compaction::EvaluatePenultimateLevel(vstorage_, mutable_cf_options_,
-                                                 ioptions_, start_level_,
-                                                 output_level_))) {
+            Compaction::EvaluateProximalLevel(vstorage_, mutable_cf_options_,
+                                              ioptions_, start_level_,
+                                              output_level_))) {
       // Constraint 1a
       tmp_start_level_inputs.clear();
       return;
@@ -490,9 +490,9 @@ bool LevelCompactionBuilder::SetupOtherInputsIfNeeded() {
     // We need to disallow this from happening.
     if (compaction_picker_->FilesRangeOverlapWithCompaction(
             compaction_inputs_, output_level_,
-            Compaction::EvaluatePenultimateLevel(vstorage_, mutable_cf_options_,
-                                                 ioptions_, start_level_,
-                                                 output_level_))) {
+            Compaction::EvaluateProximalLevel(vstorage_, mutable_cf_options_,
+                                              ioptions_, start_level_,
+                                              output_level_))) {
       // This compaction output could potentially conflict with the output
       // of a currently running compaction, we cannot run it.
       return false;
@@ -846,9 +846,9 @@ bool LevelCompactionBuilder::PickFileToCompact() {
                                                     &start_level_inputs_) ||
         compaction_picker_->FilesRangeOverlapWithCompaction(
             {start_level_inputs_}, output_level_,
-            Compaction::EvaluatePenultimateLevel(vstorage_, mutable_cf_options_,
-                                                 ioptions_, start_level_,
-                                                 output_level_))) {
+            Compaction::EvaluateProximalLevel(vstorage_, mutable_cf_options_,
+                                              ioptions_, start_level_,
+                                              output_level_))) {
       // A locked (pending compaction) input-level file was pulled in due to
       // user-key overlap.
       start_level_inputs_.clear();

db/compaction/compaction_picker_test.cc

@@ -3677,7 +3677,7 @@ TEST_F(CompactionPickerTest, UniversalSizeRatioTierCompactionLastLevel) {
   const uint64_t kFileSize = 100000;
   const int kNumLevels = 7;
   const int kLastLevel = kNumLevels - 1;
-  const int kPenultimateLevel = kLastLevel - 1;
+  const int kProximalLevel = kLastLevel - 1;
 
   ioptions_.compaction_style = kCompactionStyleUniversal;
   mutable_cf_options_.preclude_last_level_data_seconds = 1000;
@@ -3702,14 +3702,14 @@ TEST_F(CompactionPickerTest, UniversalSizeRatioTierCompactionLastLevel) {
   // Here to make sure it's size ratio compaction instead of size amp
   ASSERT_EQ(compaction->compaction_reason(),
             CompactionReason::kUniversalSizeRatio);
-  ASSERT_EQ(compaction->output_level(), kPenultimateLevel - 1);
+  ASSERT_EQ(compaction->output_level(), kProximalLevel - 1);
   ASSERT_EQ(compaction->input_levels(0)->num_files, 2);
   ASSERT_EQ(compaction->input_levels(5)->num_files, 0);
   ASSERT_EQ(compaction->input_levels(6)->num_files, 0);
 }
 
 TEST_F(CompactionPickerTest, UniversalSizeAmpTierCompactionNotSuport) {
-  // Tiered compaction only support level_num > 2 (otherwise the penultimate
+  // Tiered compaction only support level_num > 2 (otherwise the proximal
   // level is going to be level 0, which may make thing more complicated), so
   // when there's only 2 level, still treating level 1 as the last level for
   // size amp compaction
@@ -3753,7 +3753,7 @@ TEST_F(CompactionPickerTest, UniversalSizeAmpTierCompactionLastLevel) {
   const uint64_t kFileSize = 100000;
   const int kNumLevels = 7;
   const int kLastLevel = kNumLevels - 1;
-  const int kPenultimateLevel = kLastLevel - 1;
+  const int kProximalLevel = kLastLevel - 1;
 
   ioptions_.compaction_style = kCompactionStyleUniversal;
   mutable_cf_options_.preclude_last_level_data_seconds = 1000;
@@ -3775,10 +3775,10 @@ TEST_F(CompactionPickerTest, UniversalSizeAmpTierCompactionLastLevel) {
           vstorage_.get(), &log_buffer_));
 
   // It's a Size Amp compaction, but doesn't include the last level file and
-  // output to the penultimate level.
+  // output to the proximal level.
   ASSERT_EQ(compaction->compaction_reason(),
             CompactionReason::kUniversalSizeAmplification);
-  ASSERT_EQ(compaction->output_level(), kPenultimateLevel);
+  ASSERT_EQ(compaction->output_level(), kProximalLevel);
   ASSERT_EQ(compaction->input_levels(0)->num_files, 2);
   ASSERT_EQ(compaction->input_levels(5)->num_files, 1);
   ASSERT_EQ(compaction->input_levels(6)->num_files, 0);
@@ -3940,7 +3940,7 @@ TEST_P(PerKeyPlacementCompactionPickerTest, OverlapWithNormalCompaction) {
   ASSERT_EQ(enable_per_key_placement_,
             level_compaction_picker.FilesRangeOverlapWithCompaction(
                 input_files, 6,
-                Compaction::EvaluatePenultimateLevel(
+                Compaction::EvaluateProximalLevel(
                     vstorage_.get(), mutable_cf_options_, ioptions_, 0, 6)));
 }
 
@@ -4028,7 +4028,7 @@ TEST_P(PerKeyPlacementCompactionPickerTest,
   ASSERT_EQ(enable_per_key_placement_,
             universal_compaction_picker.FilesRangeOverlapWithCompaction(
                 input_files, 6,
-                Compaction::EvaluatePenultimateLevel(
+                Compaction::EvaluateProximalLevel(
                     vstorage_.get(), mutable_cf_options_, ioptions_, 0, 6)));
 }
 
@@ -4076,9 +4076,9 @@ TEST_P(PerKeyPlacementCompactionPickerTest, NormalCompactionOverlapUniversal) {
                 input_files, 5, Compaction::kInvalidLevel));
 }
 
-TEST_P(PerKeyPlacementCompactionPickerTest, PenultimateOverlapUniversal) {
+TEST_P(PerKeyPlacementCompactionPickerTest, ProximalOverlapUniversal) {
   // This test is make sure the Tiered compaction would lock whole range of
-  // both output level and penultimate level
+  // both output level and proximal level
   if (enable_per_key_placement_) {
     mutable_cf_options_.preclude_last_level_data_seconds = 10000;
   }
@@ -4098,7 +4098,7 @@ TEST_P(PerKeyPlacementCompactionPickerTest, PenultimateOverlapUniversal) {
   UpdateVersionStorageInfo();
 
   // the existing compaction is the 1st L4 file + L6 file
-  // then compaction of the 2nd L4 file to L5 (penultimate level) is overlapped
+  // then compaction of the 2nd L4 file to L5 (proximal level) is overlapped
   // when the tiered compaction feature is on.
   CompactionOptions comp_options;
   std::unordered_set<uint64_t> input_set;
@@ -4187,9 +4187,9 @@ TEST_P(PerKeyPlacementCompactionPickerTest, LastLevelOnlyOverlapUniversal) {
 }
 
 TEST_P(PerKeyPlacementCompactionPickerTest,
-       LastLevelOnlyFailPenultimateUniversal) {
+       LastLevelOnlyFailProximalUniversal) {
   // This is to test last_level only compaction still unable to do the
-  // penultimate level compaction if there's already a file in the penultimate
+  // proximal level compaction if there's already a file in the proximal
   // level.
   // This should rarely happen in universal compaction, as the non-empty L5
   // should be included in the compaction.
@@ -4222,9 +4222,9 @@ TEST_P(PerKeyPlacementCompactionPickerTest,
       mutable_db_options_, 0));
 
   ASSERT_TRUE(comp1);
-  ASSERT_EQ(comp1->GetPenultimateLevel(), Compaction::kInvalidLevel);
+  ASSERT_EQ(comp1->GetProximalLevel(), Compaction::kInvalidLevel);
 
-  // As comp1 cannot be output to the penultimate level, compacting file 40 to
+  // As comp1 cannot be output to the proximal level, compacting file 40 to
   // L5 is always safe.
   input_set.clear();
   input_files.clear();
@@ -4239,14 +4239,14 @@ TEST_P(PerKeyPlacementCompactionPickerTest,
       comp_options, input_files, 5, vstorage_.get(), mutable_cf_options_,
       mutable_db_options_, 0));
   ASSERT_TRUE(comp2);
-  ASSERT_EQ(Compaction::kInvalidLevel, comp2->GetPenultimateLevel());
+  ASSERT_EQ(Compaction::kInvalidLevel, comp2->GetProximalLevel());
 }
 
 TEST_P(PerKeyPlacementCompactionPickerTest,
        LastLevelOnlyConflictWithOngoingUniversal) {
   // This is to test last_level only compaction still unable to do the
-  // penultimate level compaction if there's already an ongoing compaction to
-  // the penultimate level
+  // proximal level compaction if there's already an ongoing compaction to
+  // the proximal level
   if (enable_per_key_placement_) {
     mutable_cf_options_.preclude_last_level_data_seconds = 10000;
   }
@@ -4265,7 +4265,7 @@ TEST_P(PerKeyPlacementCompactionPickerTest,
   Add(6, 60U, "101", "351", 60000000U);
   UpdateVersionStorageInfo();
 
-  // create an ongoing compaction to L5 (penultimate level)
+  // create an ongoing compaction to L5 (proximal level)
   CompactionOptions comp_options;
   std::unordered_set<uint64_t> input_set;
   input_set.insert(40);
@@ -4278,7 +4278,7 @@ TEST_P(PerKeyPlacementCompactionPickerTest,
       mutable_db_options_, 0));
 
   ASSERT_TRUE(comp1);
-  ASSERT_EQ(comp1->GetPenultimateLevel(), Compaction::kInvalidLevel);
+  ASSERT_EQ(comp1->GetProximalLevel(), Compaction::kInvalidLevel);
 
   input_set.clear();
   input_files.clear();
@@ -4289,7 +4289,7 @@ TEST_P(PerKeyPlacementCompactionPickerTest,
   ASSERT_EQ(enable_per_key_placement_,
             universal_compaction_picker.FilesRangeOverlapWithCompaction(
                 input_files, 6,
-                Compaction::EvaluatePenultimateLevel(
+                Compaction::EvaluateProximalLevel(
                     vstorage_.get(), mutable_cf_options_, ioptions_, 6, 6)));
 
   if (!enable_per_key_placement_) {
@@ -4297,7 +4297,7 @@ TEST_P(PerKeyPlacementCompactionPickerTest,
         comp_options, input_files, 6, vstorage_.get(), mutable_cf_options_,
         mutable_db_options_, 0));
     ASSERT_TRUE(comp2);
-    ASSERT_EQ(Compaction::kInvalidLevel, comp2->GetPenultimateLevel());
+    ASSERT_EQ(Compaction::kInvalidLevel, comp2->GetProximalLevel());
   }
 }
 
@@ -4306,7 +4306,7 @@ TEST_P(PerKeyPlacementCompactionPickerTest,
   // This is similar to `LastLevelOnlyConflictWithOngoingUniversal`, the only
   // change is the ongoing compaction to L5 has no overlap with the last level
   // compaction, so it's safe to move data from the last level to the
-  // penultimate level.
+  // proximal level.
   if (enable_per_key_placement_) {
     mutable_cf_options_.preclude_last_level_data_seconds = 10000;
   }
@@ -4325,7 +4325,7 @@ TEST_P(PerKeyPlacementCompactionPickerTest,
   Add(6, 60U, "101", "351", 60000000U);
   UpdateVersionStorageInfo();
 
-  // create an ongoing compaction to L5 (penultimate level)
+  // create an ongoing compaction to L5 (proximal level)
   CompactionOptions comp_options;
   std::unordered_set<uint64_t> input_set;
   input_set.insert(42);
@@ -4338,7 +4338,7 @@ TEST_P(PerKeyPlacementCompactionPickerTest,
       mutable_db_options_, 0));
 
   ASSERT_TRUE(comp1);
-  ASSERT_EQ(comp1->GetPenultimateLevel(), Compaction::kInvalidLevel);
+  ASSERT_EQ(comp1->GetProximalLevel(), Compaction::kInvalidLevel);
 
   input_set.clear();
   input_files.clear();
@@ -4349,8 +4349,8 @@ TEST_P(PerKeyPlacementCompactionPickerTest,
   // always safe to move data up
   ASSERT_FALSE(universal_compaction_picker.FilesRangeOverlapWithCompaction(
       input_files, 6,
-      Compaction::EvaluatePenultimateLevel(vstorage_.get(), mutable_cf_options_,
-                                           ioptions_, 6, 6)));
+      Compaction::EvaluateProximalLevel(vstorage_.get(), mutable_cf_options_,
+                                        ioptions_, 6, 6)));
 
   // 2 compactions can be run in parallel
   std::unique_ptr<Compaction> comp2(universal_compaction_picker.CompactFiles(
@@ -4358,9 +4358,9 @@ TEST_P(PerKeyPlacementCompactionPickerTest,
       mutable_db_options_, 0));
   ASSERT_TRUE(comp2);
   if (enable_per_key_placement_) {
-    ASSERT_NE(Compaction::kInvalidLevel, comp2->GetPenultimateLevel());
+    ASSERT_NE(Compaction::kInvalidLevel, comp2->GetProximalLevel());
   } else {
-    ASSERT_EQ(Compaction::kInvalidLevel, comp2->GetPenultimateLevel());
+    ASSERT_EQ(Compaction::kInvalidLevel, comp2->GetProximalLevel());
   }
 }
 

db/compaction/compaction_picker_universal.cc

@@ -288,7 +288,9 @@ class UniversalCompactionBuilder {
 // and the index of the file in that level
 
 struct InputFileInfo {
-  InputFileInfo() : f(nullptr), level(0), index(0) {}
+  InputFileInfo() : InputFileInfo(nullptr, 0, 0) {}
+  InputFileInfo(FileMetaData* file_meta, size_t l, size_t i)
+      : f(file_meta), level(l), index(i) {}
 
   FileMetaData* f;
   size_t level;
@@ -321,22 +323,14 @@ SmallestKeyHeap create_level_heap(Compaction* c, const Comparator* ucmp) {
   SmallestKeyHeap smallest_key_priority_q =
       SmallestKeyHeap(SmallestKeyHeapComparator(ucmp));
 
-  InputFileInfo input_file;
-
   for (size_t l = 0; l < c->num_input_levels(); l++) {
     if (c->num_input_files(l) != 0) {
       if (l == 0 && c->start_level() == 0) {
         for (size_t i = 0; i < c->num_input_files(0); i++) {
-          input_file.f = c->input(0, i);
-          input_file.level = 0;
-          input_file.index = i;
-          smallest_key_priority_q.push(std::move(input_file));
+          smallest_key_priority_q.emplace(c->input(0, i), 0, i);
         }
       } else {
-        input_file.f = c->input(l, 0);
-        input_file.level = l;
-        input_file.index = 0;
-        smallest_key_priority_q.push(std::move(input_file));
+        smallest_key_priority_q.emplace(c->input(l, 0), l, 0);
       }
     }
   }
@@ -374,7 +368,7 @@ bool UniversalCompactionBuilder::IsInputFilesNonOverlapping(Compaction* c) {
   auto comparator = icmp_->user_comparator();
   int first_iter = 1;
 
-  InputFileInfo prev, curr, next;
+  InputFileInfo prev, curr;
 
   SmallestKeyHeap smallest_key_priority_q =
       create_level_heap(c, icmp_->user_comparator());
@@ -397,17 +391,10 @@ bool UniversalCompactionBuilder::IsInputFilesNonOverlapping(Compaction* c) {
       prev = curr;
     }
 
-    next.f = nullptr;
-
     if (c->level(curr.level) != 0 &&
         curr.index < c->num_input_files(curr.level) - 1) {
-      next.f = c->input(curr.level, curr.index + 1);
-      next.level = curr.level;
-      next.index = curr.index + 1;
-    }
-
-    if (next.f) {
-      smallest_key_priority_q.push(std::move(next));
+      smallest_key_priority_q.emplace(c->input(curr.level, curr.index + 1),
+                                      curr.level, curr.index + 1);
     }
   }
   return true;
@@ -996,7 +983,7 @@ Compaction* UniversalCompactionBuilder::PickCompactionToReduceSortedRuns(
 
   if (output_level != 0 && picker_->FilesRangeOverlapWithCompaction(
                                inputs, output_level,
-                               Compaction::EvaluatePenultimateLevel(
+                               Compaction::EvaluateProximalLevel(
                                    vstorage_, mutable_cf_options_, ioptions_,
                                    start_level, output_level))) {
     return nullptr;
@@ -1345,7 +1332,7 @@ Compaction* UniversalCompactionBuilder::PickIncrementalForReduceSizeAmp(
   // intra L0 compactions outputs could have overlap
   if (output_level != 0 && picker_->FilesRangeOverlapWithCompaction(
                                inputs, output_level,
-                               Compaction::EvaluatePenultimateLevel(
+                               Compaction::EvaluateProximalLevel(
                                    vstorage_, mutable_cf_options_, ioptions_,
                                    start_level, output_level))) {
     return nullptr;
@@ -1486,9 +1473,9 @@ Compaction* UniversalCompactionBuilder::PickDeleteTriggeredCompaction() {
       }
       if (picker_->FilesRangeOverlapWithCompaction(
               inputs, output_level,
-              Compaction::EvaluatePenultimateLevel(
-                  vstorage_, mutable_cf_options_, ioptions_, start_level,
-                  output_level))) {
+              Compaction::EvaluateProximalLevel(vstorage_, mutable_cf_options_,
+                                                ioptions_, start_level,
+                                                output_level))) {
         return nullptr;
       }
 
@@ -1590,7 +1577,7 @@ Compaction* UniversalCompactionBuilder::PickCompactionWithSortedRunRange(
   // intra L0 compactions outputs could have overlap
   if (output_level != 0 && picker_->FilesRangeOverlapWithCompaction(
                                inputs, output_level,
-                               Compaction::EvaluatePenultimateLevel(
+                               Compaction::EvaluateProximalLevel(
                                    vstorage_, mutable_cf_options_, ioptions_,
                                    start_level, output_level))) {
     return nullptr;

db/compaction/compaction_service_job.cc

@@ -83,6 +83,14 @@ CompactionJob::ProcessKeyValueCompactionWithCompactionService(
   switch (response.status) {
     case CompactionServiceJobStatus::kSuccess:
       break;
+    case CompactionServiceJobStatus::kAborted:
+      sub_compact->status =
+          Status::Aborted("Scheduling a remote compaction job was aborted");
+      ROCKS_LOG_WARN(
+          db_options_.info_log,
+          "[%s] [JOB %d] Remote compaction was aborted at Schedule()",
+          compaction->column_family_data()->GetName().c_str(), job_id_);
+      return response.status;
     case CompactionServiceJobStatus::kFailure:
       sub_compact->status = Status::Incomplete(
           "CompactionService failed to schedule a remote compaction job.");
@@ -102,6 +110,9 @@ CompactionJob::ProcessKeyValueCompactionWithCompactionService(
       break;
   }
 
+  std::string debug_str_before_wait =
+      compaction->input_version()->DebugString();
+
   ROCKS_LOG_INFO(db_options_.info_log,
                  "[%s] [JOB %d] Waiting for remote compaction...",
                  compaction->column_family_data()->GetName().c_str(), job_id_);
@@ -110,6 +121,16 @@ CompactionJob::ProcessKeyValueCompactionWithCompactionService(
       db_options_.compaction_service->Wait(response.scheduled_job_id,
                                            &compaction_result_binary);
 
+  if (compaction_status != CompactionServiceJobStatus::kSuccess) {
+    ROCKS_LOG_ERROR(db_options_.info_log,
+                    "[%s] [JOB %d] Wait() status is not kSuccess. "
+                    "\nDebugString Before Wait():\n%s"
+                    "\nDebugString After Wait():\n%s",
+                    compaction->column_family_data()->GetName().c_str(),
+                    job_id_, debug_str_before_wait.c_str(),
+                    compaction->input_version()->DebugString().c_str());
+  }
+
   if (compaction_status == CompactionServiceJobStatus::kUseLocal) {
     ROCKS_LOG_INFO(
         db_options_.info_log,
@@ -118,6 +139,16 @@ CompactionJob::ProcessKeyValueCompactionWithCompactionService(
     return compaction_status;
   }
 
+  if (compaction_status == CompactionServiceJobStatus::kAborted) {
+    sub_compact->status =
+        Status::Aborted("Waiting a remote compaction job was aborted");
+    ROCKS_LOG_INFO(db_options_.info_log,
+                   "[%s] [JOB %d] Remote compaction was aborted during Wait()",
+                   compaction->column_family_data()->GetName().c_str(),
+                   job_id_);
+    return compaction_status;
+  }
+
   CompactionServiceResult compaction_result;
   s = CompactionServiceResult::Read(compaction_result_binary,
                                     &compaction_result);
@@ -208,19 +239,34 @@ CompactionJob::ProcessKeyValueCompactionWithCompactionService(
     meta.file_checksum_func_name = file.file_checksum_func_name;
     meta.marked_for_compaction = file.marked_for_compaction;
     meta.unique_id = file.unique_id;
+    meta.temperature = file.file_temperature;
 
     auto cfd = compaction->column_family_data();
-    sub_compact->Current().AddOutput(std::move(meta),
-                                     cfd->internal_comparator(), false, true,
-                                     file.paranoid_hash);
-    sub_compact->Current().UpdateTableProperties(file.table_properties);
+    CompactionOutputs* compaction_outputs =
+        sub_compact->Outputs(file.is_proximal_level_output);
+    assert(compaction_outputs);
+    compaction_outputs->AddOutput(std::move(meta), cfd->internal_comparator(),
+                                  false, true, file.paranoid_hash);
+    compaction_outputs->UpdateTableProperties(file.table_properties);
   }
+
+  // Set per-level stats
+  auto compaction_output_stats =
+      sub_compact->OutputStats(false /* is_proximal_level */);
+  assert(compaction_output_stats);
+  compaction_output_stats->Add(
+      compaction_result.internal_stats.output_level_stats);
+  if (compaction->SupportsPerKeyPlacement()) {
+    compaction_output_stats =
+        sub_compact->OutputStats(true /* is_proximal_level */);
+    assert(compaction_output_stats);
+    compaction_output_stats->Add(
+        compaction_result.internal_stats.proximal_level_stats);
+  }
+
+  // Set job stats
   sub_compact->compaction_job_stats = compaction_result.stats;
-  sub_compact->Current().SetNumOutputRecords(
-      compaction_result.stats.num_output_records);
-  sub_compact->Current().SetNumOutputFiles(
-      compaction_result.stats.num_output_files);
-  sub_compact->Current().AddBytesWritten(compaction_result.bytes_written);
+
   RecordTick(stats_, REMOTE_COMPACT_READ_BYTES, compaction_result.bytes_read);
   RecordTick(stats_, REMOTE_COMPACT_WRITE_BYTES,
              compaction_result.bytes_written);
@@ -240,18 +286,6 @@ void CompactionServiceCompactionJob::RecordCompactionIOStats() {
   CompactionJob::RecordCompactionIOStats();
 }
 
-void CompactionServiceCompactionJob::UpdateCompactionJobStats(
-    const InternalStats::CompactionStats& stats) const {
-  compaction_job_stats_->elapsed_micros = stats.micros;
-
-  // output information only in remote compaction
-  compaction_job_stats_->total_output_bytes = stats.bytes_written;
-  compaction_job_stats_->total_output_bytes_blob = stats.bytes_written_blob;
-  compaction_job_stats_->num_output_records = stats.num_output_records;
-  compaction_job_stats_->num_output_files = stats.num_output_files;
-  compaction_job_stats_->num_output_files_blob = stats.num_output_files_blob;
-}
-
 CompactionServiceCompactionJob::CompactionServiceCompactionJob(
     int job_id, Compaction* compaction, const ImmutableDBOptions& db_options,
     const MutableDBOptions& mutable_db_options, const FileOptions& file_options,
@@ -313,15 +347,14 @@ Status CompactionServiceCompactionJob::Run() {
 
   ProcessKeyValueCompaction(sub_compact);
 
-  compaction_stats_.stats.micros =
-      db_options_.clock->NowMicros() - start_micros;
-  compaction_stats_.stats.cpu_micros =
-      sub_compact->compaction_job_stats.cpu_micros;
+  uint64_t elapsed_micros = db_options_.clock->NowMicros() - start_micros;
+  internal_stats_.SetMicros(elapsed_micros);
+  internal_stats_.AddCpuMicros(elapsed_micros);
 
   RecordTimeToHistogram(stats_, COMPACTION_TIME,
-                        compaction_stats_.stats.micros);
+                        internal_stats_.output_level_stats.micros);
   RecordTimeToHistogram(stats_, COMPACTION_CPU_TIME,
-                        compaction_stats_.stats.cpu_micros);
+                        internal_stats_.output_level_stats.cpu_micros);
 
   Status status = sub_compact->status;
   IOStatus io_s = sub_compact->io_status;
@@ -351,25 +384,31 @@ Status CompactionServiceCompactionJob::Run() {
 
   // Build Compaction Job Stats
 
-  // 1. Aggregate CompactionOutputStats into Internal Compaction Stats
-  // (compaction_stats_) and aggregate Compaction Job Stats
-  // (compaction_job_stats_) from the sub compactions
-  compact_->AggregateCompactionStats(compaction_stats_, *compaction_job_stats_);
+  // 1. Aggregate internal stats and job stats for all subcompactions
+  // internal stats: sub_compact.proximal_level_outputs_.stats and
+  //                 sub_compact.compaction_outputs_.stats into
+  //                 internal_stats_.output_level_stats and
+  //                 internal_stats_.proximal_level_stats
+  // job-level stats: sub_compact.compaction_job_stats into compact.job_stats_
+  //
+  // For remote compaction, there's only one subcompaction.
+  compact_->AggregateCompactionStats(internal_stats_, *job_stats_);
 
-  // 2. Update the Output information in the Compaction Job Stats with
-  // aggregated Internal Compaction Stats.
-  UpdateCompactionJobStats(compaction_stats_.stats);
-
-  // 3. Set fields that are not propagated as part of aggregations above
+  // 2. Update job-level output stats with the aggregated internal_stats_
+  // Please note that input stats will be updated by primary host when all
+  // subcompactions are finished
+  UpdateCompactionJobOutputStats(internal_stats_);
+  // and set fields that are not propagated as part of the update
   compaction_result_->stats.is_manual_compaction = c->is_manual_compaction();
   compaction_result_->stats.is_full_compaction = c->is_full_compaction();
   compaction_result_->stats.is_remote_compaction = true;
 
-  // 4. Update IO Stats that are not part of the aggregations above (bytes_read,
-  // bytes_written)
+  // 3. Update IO Stats that are not part of the the update above
+  // (bytes_read, bytes_written)
   RecordCompactionIOStats();
 
   // Build Output
+  compaction_result_->internal_stats = internal_stats_;
   compaction_result_->output_level = compact_->compaction->output_level();
   compaction_result_->output_path = output_path_;
   if (status.ok()) {
@@ -382,7 +421,8 @@ Status CompactionServiceCompactionJob::Run() {
           meta.file_creation_time, meta.epoch_number, meta.file_checksum,
           meta.file_checksum_func_name, output_file.validator.GetHash(),
           meta.marked_for_compaction, meta.unique_id,
-          *output_file.table_properties);
+          *output_file.table_properties, output_file.is_proximal_level,
+          meta.temperature);
     }
   }
 
@@ -554,7 +594,16 @@ static std::unordered_map<std::string, OptionTypeInfo>
             const auto this_one = static_cast<const TableProperties*>(addr1);
             const auto that_one = static_cast<const TableProperties*>(addr2);
             return this_one->AreEqual(opts, that_one, mismatch);
-          }}}};
+          }}},
+        {"is_proximal_level_output",
+         {offsetof(struct CompactionServiceOutputFile,
+                   is_proximal_level_output),
+          OptionType::kBoolean, OptionVerificationType::kNormal,
+          OptionTypeFlags::kNone}},
+        {"file_temperature",
+         {offsetof(struct CompactionServiceOutputFile, file_temperature),
+          OptionType::kTemperature, OptionVerificationType::kNormal,
+          OptionTypeFlags::kNone}}};
 
 static std::unordered_map<std::string, OptionTypeInfo>
     compaction_job_stats_type_info = {
@@ -679,6 +728,125 @@ static std::unordered_map<std::string, OptionTypeInfo>
           OptionTypeFlags::kNone}},
 };
 
+static std::unordered_map<std::string, OptionTypeInfo>
+    compaction_stats_type_info = {
+        {"micros",
+         {offsetof(struct InternalStats::CompactionStats, micros),
+          OptionType::kUInt64T, OptionVerificationType::kNormal,
+          OptionTypeFlags::kNone}},
+        {"cpu_micros",
+         {offsetof(struct InternalStats::CompactionStats, cpu_micros),
+          OptionType::kUInt64T, OptionVerificationType::kNormal,
+          OptionTypeFlags::kNone}},
+        {"bytes_read_non_output_levels",
+         {offsetof(struct InternalStats::CompactionStats,
+                   bytes_read_non_output_levels),
+          OptionType::kUInt64T, OptionVerificationType::kNormal,
+          OptionTypeFlags::kNone}},
+        {"bytes_read_output_level",
+         {offsetof(struct InternalStats::CompactionStats,
+                   bytes_read_output_level),
+          OptionType::kUInt64T, OptionVerificationType::kNormal,
+          OptionTypeFlags::kNone}},
+        {"bytes_skipped_non_output_levels",
+         {offsetof(struct InternalStats::CompactionStats,
+                   bytes_skipped_non_output_levels),
+          OptionType::kUInt64T, OptionVerificationType::kNormal,
+          OptionTypeFlags::kNone}},
+        {"bytes_skipped_output_level",
+         {offsetof(struct InternalStats::CompactionStats,
+                   bytes_skipped_output_level),
+          OptionType::kUInt64T, OptionVerificationType::kNormal,
+          OptionTypeFlags::kNone}},
+        {"bytes_read_blob",
+         {offsetof(struct InternalStats::CompactionStats, bytes_read_blob),
+          OptionType::kUInt64T, OptionVerificationType::kNormal,
+          OptionTypeFlags::kNone}},
+        {"bytes_written",
+         {offsetof(struct InternalStats::CompactionStats, bytes_written),
+          OptionType::kUInt64T, OptionVerificationType::kNormal,
+          OptionTypeFlags::kNone}},
+        {"bytes_written_blob",
+         {offsetof(struct InternalStats::CompactionStats, bytes_written_blob),
+          OptionType::kUInt64T, OptionVerificationType::kNormal,
+          OptionTypeFlags::kNone}},
+        {"bytes_moved",
+         {offsetof(struct InternalStats::CompactionStats, bytes_moved),
+          OptionType::kUInt64T, OptionVerificationType::kNormal,
+          OptionTypeFlags::kNone}},
+        {"num_input_files_in_non_output_levels",
+         {offsetof(struct InternalStats::CompactionStats,
+                   num_input_files_in_non_output_levels),
+          OptionType::kInt, OptionVerificationType::kNormal,
+          OptionTypeFlags::kNone}},
+        {"num_input_files_in_output_level",
+         {offsetof(struct InternalStats::CompactionStats,
+                   num_input_files_in_output_level),
+          OptionType::kInt, OptionVerificationType::kNormal,
+          OptionTypeFlags::kNone}},
+        {"num_filtered_input_files_in_non_output_levels",
+         {offsetof(struct InternalStats::CompactionStats,
+                   num_filtered_input_files_in_non_output_levels),
+          OptionType::kInt, OptionVerificationType::kNormal,
+          OptionTypeFlags::kNone}},
+        {"num_filtered_input_files_in_output_level",
+         {offsetof(struct InternalStats::CompactionStats,
+                   num_filtered_input_files_in_output_level),
+          OptionType::kInt, OptionVerificationType::kNormal,
+          OptionTypeFlags::kNone}},
+        {"num_output_files",
+         {offsetof(struct InternalStats::CompactionStats, num_output_files),
+          OptionType::kInt, OptionVerificationType::kNormal,
+          OptionTypeFlags::kNone}},
+        {"num_output_files_blob",
+         {offsetof(struct InternalStats::CompactionStats,
+                   num_output_files_blob),
+          OptionType::kInt, OptionVerificationType::kNormal,
+          OptionTypeFlags::kNone}},
+        {"num_input_records",
+         {offsetof(struct InternalStats::CompactionStats, num_input_records),
+          OptionType::kUInt64T, OptionVerificationType::kNormal,
+          OptionTypeFlags::kNone}},
+        {"num_dropped_records",
+         {offsetof(struct InternalStats::CompactionStats, num_dropped_records),
+          OptionType::kUInt64T, OptionVerificationType::kNormal,
+          OptionTypeFlags::kNone}},
+        {"num_output_records",
+         {offsetof(struct InternalStats::CompactionStats, num_output_records),
+          OptionType::kUInt64T, OptionVerificationType::kNormal,
+          OptionTypeFlags::kNone}},
+        {"count",
+         {offsetof(struct InternalStats::CompactionStats, count),
+          OptionType::kUInt64T, OptionVerificationType::kNormal,
+          OptionTypeFlags::kNone}},
+        {"counts", OptionTypeInfo::Array<
+                       int, static_cast<int>(CompactionReason::kNumOfReasons)>(
+                       offsetof(struct InternalStats::CompactionStats, counts),
+                       OptionVerificationType::kNormal, OptionTypeFlags::kNone,
+                       {0, OptionType::kInt})},
+};
+
+static std::unordered_map<std::string, OptionTypeInfo>
+    compaction_internal_stats_type_info = {
+        {"output_level_stats",
+         OptionTypeInfo::Struct(
+             "output_level_stats", &compaction_stats_type_info,
+             offsetof(struct InternalStats::CompactionStatsFull,
+                      output_level_stats),
+             OptionVerificationType::kNormal, OptionTypeFlags::kNone)},
+        {"has_proximal_level_output",
+         {offsetof(struct InternalStats::CompactionStatsFull,
+                   has_proximal_level_output),
+          OptionType::kBoolean, OptionVerificationType::kNormal,
+          OptionTypeFlags::kNone}},
+        {"proximal_level_stats",
+         OptionTypeInfo::Struct(
+             "proximal_level_stats", &compaction_stats_type_info,
+             offsetof(struct InternalStats::CompactionStatsFull,
+                      proximal_level_stats),
+             OptionVerificationType::kNormal, OptionTypeFlags::kNone)},
+};
+
 namespace {
 // this is a helper struct to serialize and deserialize class Status, because
 // Status's members are not public.
@@ -785,6 +953,11 @@ static std::unordered_map<std::string, OptionTypeInfo> cs_result_type_info = {
                   "stats", &compaction_job_stats_type_info,
                   offsetof(struct CompactionServiceResult, stats),
                   OptionVerificationType::kNormal, OptionTypeFlags::kNone)},
+    {"internal_stats",
+     OptionTypeInfo::Struct(
+         "internal_stats", &compaction_internal_stats_type_info,
+         offsetof(struct CompactionServiceResult, internal_stats),
+         OptionVerificationType::kNormal, OptionTypeFlags::kNone)},
 };
 
 Status CompactionServiceInput::Read(const std::string& data_str,

db/compaction/compaction_service_test.cc

@@ -357,11 +357,12 @@ TEST_F(CompactionServiceTest, BasicCompactions) {
   } else {
     ASSERT_OK(result.status);
   }
-  ASSERT_GE(result.stats.elapsed_micros, 1);
-  ASSERT_GE(result.stats.cpu_micros, 1);
+  ASSERT_GE(result.internal_stats.output_level_stats.micros, 1);
+  ASSERT_GE(result.internal_stats.output_level_stats.cpu_micros, 1);
 
-  ASSERT_EQ(20, result.stats.num_output_records);
-  ASSERT_EQ(result.output_files.size(), result.stats.num_output_files);
+  ASSERT_EQ(20, result.internal_stats.output_level_stats.num_output_records);
+  ASSERT_EQ(result.output_files.size(),
+            result.internal_stats.output_level_stats.num_output_files);
 
   uint64_t total_size = 0;
   for (auto output_file : result.output_files) {
@@ -372,7 +373,7 @@ TEST_F(CompactionServiceTest, BasicCompactions) {
     ASSERT_GT(file_size, 0);
     total_size += file_size;
   }
-  ASSERT_EQ(total_size, result.stats.total_output_bytes);
+  ASSERT_EQ(total_size, result.internal_stats.TotalBytesWritten());
 
   ASSERT_TRUE(result.stats.is_remote_compaction);
   ASSERT_TRUE(result.stats.is_manual_compaction);
@@ -716,6 +717,46 @@ TEST_F(CompactionServiceTest, VerifyStatsLocalFallback) {
   VerifyTestData();
 }
 
+TEST_F(CompactionServiceTest, VerifyInputRecordCount) {
+  Options options = CurrentOptions();
+  options.disable_auto_compactions = true;
+  ReopenWithCompactionService(&options);
+  GenerateTestData();
+
+  auto my_cs = GetCompactionService();
+
+  std::string start_str = Key(15);
+  std::string end_str = Key(45);
+  Slice start(start_str);
+  Slice end(end_str);
+  uint64_t comp_num = my_cs->GetCompactionNum();
+
+  // Only iterator through 10 keys and force compaction to finish.
+  int num_iter = 0;
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
+      "CompactionJob::ProcessKeyValueCompaction()::stop", [&](void* stop_ptr) {
+        num_iter++;
+        if (num_iter == 10) {
+          *(bool*)stop_ptr = true;
+        }
+      });
+  SyncPoint::GetInstance()->EnableProcessing();
+
+  // CompactRange() should fail
+  Status s = db_->CompactRange(CompactRangeOptions(), &start, &end);
+  ASSERT_NOK(s);
+  ASSERT_TRUE(s.IsCorruption());
+  const char* expected_message =
+      "Compaction number of input keys does not match number of keys "
+      "processed.";
+  ASSERT_TRUE(std::strstr(s.getState(), expected_message));
+
+  ASSERT_GE(my_cs->GetCompactionNum(), comp_num + 1);
+
+  SyncPoint::GetInstance()->DisableProcessing();
+  SyncPoint::GetInstance()->ClearAllCallBacks();
+}
+
 TEST_F(CompactionServiceTest, CorruptedOutput) {
   Options options = CurrentOptions();
   options.disable_auto_compactions = true;
@@ -1188,34 +1229,38 @@ TEST_F(CompactionServiceTest, PrecludeLastLevel) {
 
   for (int i = 0; i < kNumTrigger; i++) {
     for (int j = 0; j < kNumKeys; j++) {
-      // FIXME: need to assign outputs to levels to allow overlapping ranges:
-      // ASSERT_OK(Put(Key(j * kNumTrigger + i), "v" + std::to_string(i)));
-      // instead of this (too easy):
-      ASSERT_OK(Put(Key(i * kNumKeys + j), "v" + std::to_string(i)));
+      ASSERT_OK(Put(Key(j * kNumTrigger + i), "v" + std::to_string(i)));
     }
     ASSERT_OK(Flush());
   }
   ASSERT_OK(dbfull()->TEST_WaitForCompact());
 
-  // Data split between penultimate (kUnknown) and last (kCold) levels
-  // FIXME: need to assign outputs to levels to get this:
-  // ASSERT_EQ("0,0,0,0,0,1,1", FilesPerLevel());
-  // ASSERT_GT(GetSstSizeHelper(Temperature::kUnknown), 0);
-  // ASSERT_GT(GetSstSizeHelper(Temperature::kCold), 0);
-  // instead of this (WRONG but currently expected):
-  ASSERT_EQ("0,0,0,0,0,0,2", FilesPerLevel());
-  // Check manifest temperatures
+  // Data split between proximal (kUnknown) and last (kCold) levels
+  ASSERT_EQ("0,0,0,0,0,1,1", FilesPerLevel());
   ASSERT_GT(GetSstSizeHelper(Temperature::kUnknown), 0);
-  ASSERT_EQ(GetSstSizeHelper(Temperature::kCold), 0);
+  ASSERT_GT(GetSstSizeHelper(Temperature::kCold), 0);
+
   // TODO: Check FileSystem temperatures with FileTemperatureTestFS
 
   for (int i = 0; i < kNumTrigger; i++) {
     for (int j = 0; j < kNumKeys; j++) {
-      // FIXME
-      // ASSERT_EQ(Get(Key(j * kNumTrigger + i)), "v" + std::to_string(i));
-      ASSERT_EQ(Get(Key(i * kNumKeys + j)), "v" + std::to_string(i));
+      ASSERT_EQ(Get(Key(j * kNumTrigger + i)), "v" + std::to_string(i));
     }
   }
+
+  // Verify Output Stats
+  auto my_cs = GetCompactionService();
+  CompactionServiceResult result;
+  my_cs->GetResult(&result);
+  ASSERT_OK(result.status);
+  ASSERT_GT(result.internal_stats.output_level_stats.cpu_micros, 0);
+  ASSERT_GT(result.internal_stats.output_level_stats.micros, 0);
+  ASSERT_EQ(result.internal_stats.output_level_stats.num_output_records +
+                result.internal_stats.proximal_level_stats.num_output_records,
+            kNumTrigger * kNumKeys);
+  ASSERT_EQ(result.internal_stats.output_level_stats.num_output_files +
+                result.internal_stats.proximal_level_stats.num_output_files,
+            2);
 }
 
 TEST_F(CompactionServiceTest, ConcurrentCompaction) {
@@ -1471,6 +1516,40 @@ TEST_F(CompactionServiceTest, FallbackLocalManual) {
   VerifyTestData();
 }
 
+TEST_F(CompactionServiceTest, AbortedWhileWait) {
+  Options options = CurrentOptions();
+  options.disable_auto_compactions = true;
+  ReopenWithCompactionService(&options);
+
+  GenerateTestData();
+  VerifyTestData();
+
+  auto my_cs = GetCompactionService();
+  Statistics* compactor_statistics = GetCompactorStatistics();
+  Statistics* primary_statistics = GetPrimaryStatistics();
+
+  my_cs->ResetOverride();
+  std::string start_str = Key(15);
+  std::string end_str = Key(45);
+  Slice start(start_str);
+  Slice end(end_str);
+
+  // Override Wait() result with kAborted
+  my_cs->OverrideWaitStatus(CompactionServiceJobStatus::kAborted);
+  start_str = Key(120);
+  start = start_str;
+
+  Status s = db_->CompactRange(CompactRangeOptions(), &start, nullptr);
+  ASSERT_NOK(s);
+  ASSERT_TRUE(s.IsAborted());
+  // no remote compaction is run
+  ASSERT_EQ(my_cs->GetCompactionNum(), 0);
+  // make sure the compaction statistics is not recorded any side
+  ASSERT_EQ(primary_statistics->getTickerCount(COMPACT_WRITE_BYTES), 0);
+  ASSERT_EQ(primary_statistics->getTickerCount(REMOTE_COMPACT_WRITE_BYTES), 0);
+  ASSERT_EQ(compactor_statistics->getTickerCount(COMPACT_WRITE_BYTES), 0);
+}
+
 TEST_F(CompactionServiceTest, RemoteEventListener) {
   class RemoteEventListenerTest : public EventListener {
    public:

db/compaction/compaction_state.cc

@@ -36,11 +36,11 @@ Slice CompactionState::LargestUserKey() {
 }
 
 void CompactionState::AggregateCompactionStats(
-    InternalStats::CompactionStatsFull& compaction_stats,
-    CompactionJobStats& compaction_job_stats) {
+    InternalStats::CompactionStatsFull& internal_stats,
+    CompactionJobStats& job_stats) {
   for (const auto& sc : sub_compact_states) {
-    sc.AggregateCompactionOutputStats(compaction_stats);
-    compaction_job_stats.Add(sc.compaction_job_stats);
+    sc.AggregateCompactionOutputStats(internal_stats);
+    job_stats.Add(sc.compaction_job_stats);
   }
 }
 }  // namespace ROCKSDB_NAMESPACE

db/compaction/compaction_state.h

@@ -29,8 +29,8 @@ class CompactionState {
   Status status;
 
   void AggregateCompactionStats(
-      InternalStats::CompactionStatsFull& compaction_stats,
-      CompactionJobStats& compaction_job_stats);
+      InternalStats::CompactionStatsFull& internal_stats,
+      CompactionJobStats& job_stats);
 
   explicit CompactionState(Compaction* c) : compaction(c) {}
 

db/compaction/subcompaction_state.cc

@@ -14,33 +14,32 @@
 
 namespace ROCKSDB_NAMESPACE {
 void SubcompactionState::AggregateCompactionOutputStats(
-    InternalStats::CompactionStatsFull& compaction_stats) const {
+    InternalStats::CompactionStatsFull& internal_stats) const {
   // Outputs should be closed. By extension, any files created just for
   // range deletes have already been written also.
   assert(compaction_outputs_.HasBuilder() == false);
-  assert(penultimate_level_outputs_.HasBuilder() == false);
+  assert(proximal_level_outputs_.HasBuilder() == false);
 
   // FIXME: These stats currently include abandonned output files
   // assert(compaction_outputs_.stats_.num_output_files ==
   //        compaction_outputs_.outputs_.size());
-  // assert(penultimate_level_outputs_.stats_.num_output_files ==
-  //        penultimate_level_outputs_.outputs_.size());
+  // assert(proximal_level_outputs_.stats_.num_output_files ==
+  //        proximal_level_outputs_.outputs_.size());
 
-  compaction_stats.stats.Add(compaction_outputs_.stats_);
-  if (penultimate_level_outputs_.HasOutput()) {
-    compaction_stats.has_penultimate_level_output = true;
-    compaction_stats.penultimate_level_stats.Add(
-        penultimate_level_outputs_.stats_);
+  internal_stats.output_level_stats.Add(compaction_outputs_.stats_);
+  if (proximal_level_outputs_.HasOutput()) {
+    internal_stats.has_proximal_level_output = true;
+    internal_stats.proximal_level_stats.Add(proximal_level_outputs_.stats_);
   }
 }
 
 OutputIterator SubcompactionState::GetOutputs() const {
-  return OutputIterator(penultimate_level_outputs_.outputs_,
+  return OutputIterator(proximal_level_outputs_.outputs_,
                         compaction_outputs_.outputs_);
 }
 
 void SubcompactionState::Cleanup(Cache* cache) {
-  penultimate_level_outputs_.Cleanup();
+  proximal_level_outputs_.Cleanup();
   compaction_outputs_.Cleanup();
 
   if (!status.ok()) {
@@ -63,9 +62,9 @@ void SubcompactionState::Cleanup(Cache* cache) {
 }
 
 Slice SubcompactionState::SmallestUserKey() const {
-  if (penultimate_level_outputs_.HasOutput()) {
+  if (proximal_level_outputs_.HasOutput()) {
     Slice a = compaction_outputs_.SmallestUserKey();
-    Slice b = penultimate_level_outputs_.SmallestUserKey();
+    Slice b = proximal_level_outputs_.SmallestUserKey();
     if (a.empty()) {
       return b;
     }
@@ -85,9 +84,9 @@ Slice SubcompactionState::SmallestUserKey() const {
 }
 
 Slice SubcompactionState::LargestUserKey() const {
-  if (penultimate_level_outputs_.HasOutput()) {
+  if (proximal_level_outputs_.HasOutput()) {
     Slice a = compaction_outputs_.LargestUserKey();
-    Slice b = penultimate_level_outputs_.LargestUserKey();
+    Slice b = proximal_level_outputs_.LargestUserKey();
     if (a.empty()) {
       return b;
     }
@@ -107,12 +106,12 @@ Slice SubcompactionState::LargestUserKey() const {
 }
 
 Status SubcompactionState::AddToOutput(
-    const CompactionIterator& iter, bool use_penultimate_output,
+    const CompactionIterator& iter, bool use_proximal_output,
     const CompactionFileOpenFunc& open_file_func,
     const CompactionFileCloseFunc& close_file_func) {
   // update target output
-  current_outputs_ = use_penultimate_output ? &penultimate_level_outputs_
-                                            : &compaction_outputs_;
+  current_outputs_ =
+      use_proximal_output ? &proximal_level_outputs_ : &compaction_outputs_;
   return current_outputs_->AddToOutput(iter, open_file_func, close_file_func);
 }
 

db/compaction/subcompaction_state.h

@@ -26,13 +26,13 @@ namespace ROCKSDB_NAMESPACE {
 // Maintains state and outputs for each sub-compaction
 // It contains 2 `CompactionOutputs`:
 //  1. one for the normal output files
-//  2. another for the penultimate level outputs
+//  2. another for the proximal level outputs
 // a `current` pointer maintains the current output group, when calling
 // `AddToOutput()`, it checks the output of the current compaction_iterator key
 // and point `current` to the target output group. By default, it just points to
 // normal compaction_outputs, if the compaction_iterator key should be placed on
-// the penultimate level, `current` is changed to point to
-// `penultimate_level_outputs`.
+// the proximal level, `current` is changed to point to
+// `proximal_level_outputs`.
 // The later operations uses `Current()` to get the target group.
 //
 // +----------+          +-----------------------------+      +---------+
@@ -43,7 +43,7 @@ namespace ROCKSDB_NAMESPACE {
 //       |                                                    |  ...    |
 //       |
 //       |               +-----------------------------+      +---------+
-//       +-------------> | penultimate_level_outputs   |----->| output  |
+//       +-------------> | proximal_level_outputs      |----->| output  |
 //                       +-----------------------------+      +---------+
 //                                                            |  ...    |
 
@@ -78,7 +78,7 @@ class SubcompactionState {
   Slice LargestUserKey() const;
 
   // Get all outputs from the subcompaction. For per_key_placement compaction,
-  // it returns both the last level outputs and penultimate level outputs.
+  // it returns both the last level outputs and proximal level outputs.
   OutputIterator GetOutputs() const;
 
   // Assign range dels aggregator. The various tombstones will potentially
@@ -92,7 +92,7 @@ class SubcompactionState {
 
   void RemoveLastEmptyOutput() {
     compaction_outputs_.RemoveLastEmptyOutput();
-    penultimate_level_outputs_.RemoveLastEmptyOutput();
+    proximal_level_outputs_.RemoveLastEmptyOutput();
   }
 
   void BuildSubcompactionJobInfo(
@@ -119,14 +119,14 @@ class SubcompactionState {
         start(_start),
         end(_end),
         sub_job_id(_sub_job_id),
-        compaction_outputs_(c, /*is_penultimate_level=*/false),
-        penultimate_level_outputs_(c, /*is_penultimate_level=*/true) {
+        compaction_outputs_(c, /*is_proximal_level=*/false),
+        proximal_level_outputs_(c, /*is_proximal_level=*/true) {
     assert(compaction != nullptr);
     // Set output split key (used for RoundRobin feature) only for normal
-    // compaction_outputs, output to penultimate_level feature doesn't support
+    // compaction_outputs, output to proximal_level feature doesn't support
     // RoundRobin feature (and may never going to be supported, because for
     // RoundRobin, the data time is mostly naturally sorted, no need to have
-    // per-key placement with output_to_penultimate_level).
+    // per-key placement with output_to_proximal_level).
     compaction_outputs_.SetOutputSlitKey(start, end);
   }
 
@@ -141,18 +141,17 @@ class SubcompactionState {
         compaction_job_stats(std::move(state.compaction_job_stats)),
         sub_job_id(state.sub_job_id),
         compaction_outputs_(std::move(state.compaction_outputs_)),
-        penultimate_level_outputs_(std::move(state.penultimate_level_outputs_)),
+        proximal_level_outputs_(std::move(state.proximal_level_outputs_)),
         range_del_agg_(std::move(state.range_del_agg_)) {
-    current_outputs_ =
-        state.current_outputs_ == &state.penultimate_level_outputs_
-            ? &penultimate_level_outputs_
-            : &compaction_outputs_;
+    current_outputs_ = state.current_outputs_ == &state.proximal_level_outputs_
+                           ? &proximal_level_outputs_
+                           : &compaction_outputs_;
   }
 
   // Add all the new files from this compaction to version_edit
   void AddOutputsEdit(VersionEdit* out_edit) const {
-    for (const auto& file : penultimate_level_outputs_.outputs_) {
-      out_edit->AddFile(compaction->GetPenultimateLevel(), file.meta);
+    for (const auto& file : proximal_level_outputs_.outputs_) {
+      out_edit->AddFile(compaction->GetProximalLevel(), file.meta);
     }
     for (const auto& file : compaction_outputs_.outputs_) {
       out_edit->AddFile(compaction->output_level(), file.meta);
@@ -162,13 +161,32 @@ class SubcompactionState {
   void Cleanup(Cache* cache);
 
   void AggregateCompactionOutputStats(
-      InternalStats::CompactionStatsFull& compaction_stats) const;
+      InternalStats::CompactionStatsFull& internal_stats) const;
 
   CompactionOutputs& Current() const {
     assert(current_outputs_);
     return *current_outputs_;
   }
 
+  CompactionOutputs* Outputs(bool is_proximal_level) {
+    assert(compaction);
+    if (is_proximal_level) {
+      assert(compaction->SupportsPerKeyPlacement());
+      return &proximal_level_outputs_;
+    }
+    return &compaction_outputs_;
+  }
+
+  // Per-level stats for the output
+  InternalStats::CompactionStats* OutputStats(bool is_proximal_level) {
+    assert(compaction);
+    if (is_proximal_level) {
+      assert(compaction->SupportsPerKeyPlacement());
+      return &proximal_level_outputs_.stats_;
+    }
+    return &compaction_outputs_.stats_;
+  }
+
   CompactionRangeDelAggregator* RangeDelAgg() const {
     return range_del_agg_.get();
   }
@@ -179,12 +197,11 @@ class SubcompactionState {
   }
 
   // Add compaction_iterator key/value to the `Current` output group.
-  Status AddToOutput(const CompactionIterator& iter,
-                     bool use_penultimate_output,
+  Status AddToOutput(const CompactionIterator& iter, bool use_proximal_output,
                      const CompactionFileOpenFunc& open_file_func,
                      const CompactionFileCloseFunc& close_file_func);
 
-  // Close all compaction output files, both output_to_penultimate_level outputs
+  // Close all compaction output files, both output_to_proximal_level outputs
   // and normal outputs.
   Status CloseCompactionFiles(const Status& curr_status,
                               const CompactionFileOpenFunc& open_file_func,
@@ -195,11 +212,11 @@ class SubcompactionState {
     // CloseOutput() may open new compaction output files.
     Status s = curr_status;
     if (per_key) {
-      s = penultimate_level_outputs_.CloseOutput(
-          s, range_del_agg_.get(), open_file_func, close_file_func);
+      s = proximal_level_outputs_.CloseOutput(s, range_del_agg_.get(),
+                                              open_file_func, close_file_func);
     } else {
-      assert(penultimate_level_outputs_.HasBuilder() == false);
-      assert(penultimate_level_outputs_.HasOutput() == false);
+      assert(proximal_level_outputs_.HasBuilder() == false);
+      assert(proximal_level_outputs_.HasOutput() == false);
     }
     s = compaction_outputs_.CloseOutput(s, range_del_agg_.get(), open_file_func,
                                         close_file_func);
@@ -209,7 +226,7 @@ class SubcompactionState {
  private:
   // State kept for output being generated
   CompactionOutputs compaction_outputs_;
-  CompactionOutputs penultimate_level_outputs_;
+  CompactionOutputs proximal_level_outputs_;
   CompactionOutputs* current_outputs_ = &compaction_outputs_;
   std::unique_ptr<CompactionRangeDelAggregator> range_del_agg_;
 };

db/compaction/tiered_compaction_test.cc

@@ -33,42 +33,13 @@ ConfigOptions GetStrictConfigOptions() {
 class TieredCompactionTest : public DBTestBase {
  public:
   TieredCompactionTest()
-      : DBTestBase("tiered_compaction_test", /*env_do_fsync=*/true),
-        kBasicCompStats(CompactionReason::kUniversalSizeAmplification, 1),
-        kBasicPerKeyPlacementCompStats(
-            CompactionReason::kUniversalSizeAmplification, 1),
-        kBasicFlushStats(CompactionReason::kFlush, 1) {
-    kBasicCompStats.micros = kHasValue;
-    kBasicCompStats.cpu_micros = kHasValue;
-    kBasicCompStats.bytes_read_non_output_levels = kHasValue;
-    kBasicCompStats.num_input_files_in_non_output_levels = kHasValue;
-    kBasicCompStats.num_input_records = kHasValue;
-    kBasicCompStats.num_dropped_records = kHasValue;
-
-    kBasicPerLevelStats.num_output_records = kHasValue;
-    kBasicPerLevelStats.bytes_written = kHasValue;
-    kBasicPerLevelStats.num_output_files = kHasValue;
-
-    kBasicPerKeyPlacementCompStats.micros = kHasValue;
-    kBasicPerKeyPlacementCompStats.cpu_micros = kHasValue;
-    kBasicPerKeyPlacementCompStats.Add(kBasicPerLevelStats);
-
-    kBasicFlushStats.micros = kHasValue;
-    kBasicFlushStats.cpu_micros = kHasValue;
-    kBasicFlushStats.bytes_written = kHasValue;
-    kBasicFlushStats.num_output_files = kHasValue;
-  }
+      : DBTestBase("tiered_compaction_test", /*env_do_fsync=*/true) {}
 
  protected:
-  static constexpr uint8_t kHasValue = 1;
-
-  InternalStats::CompactionStats kBasicCompStats;
-  InternalStats::CompactionStats kBasicPerKeyPlacementCompStats;
-  InternalStats::CompactionOutputsStats kBasicPerLevelStats;
-  InternalStats::CompactionStats kBasicFlushStats;
-
   std::atomic_bool enable_per_key_placement = true;
 
+  CompactionJobStats job_stats;
+
   void SetUp() override {
     SyncPoint::GetInstance()->SetCallBack(
         "Compaction::SupportsPerKeyPlacement:Enabled", [&](void* arg) {
@@ -108,21 +79,36 @@ class TieredCompactionTest : public DBTestBase {
 
   // Verify the compaction stats, the stats are roughly compared
   void VerifyCompactionStats(
-      const std::vector<InternalStats::CompactionStats>& expect_stats,
-      const InternalStats::CompactionStats& expect_pl_stats) {
+      const std::vector<InternalStats::CompactionStats>& expected_stats,
+      const InternalStats::CompactionStats& expected_pl_stats,
+      size_t output_level, uint64_t num_input_range_del = 0) {
     const std::vector<InternalStats::CompactionStats>& stats =
         GetCompactionStats();
-    const size_t kLevels = expect_stats.size();
+    const size_t kLevels = expected_stats.size();
     ASSERT_EQ(kLevels, stats.size());
+    ASSERT_TRUE(output_level < kLevels);
 
-    for (auto it = stats.begin(), expect = expect_stats.begin();
-         it != stats.end(); it++, expect++) {
-      VerifyCompactionStats(*it, *expect);
+    for (size_t level = 0; level < kLevels; level++) {
+      VerifyCompactionStats(stats[level], expected_stats[level]);
     }
 
     const InternalStats::CompactionStats& pl_stats =
         GetPerKeyPlacementCompactionStats();
-    VerifyCompactionStats(pl_stats, expect_pl_stats);
+    VerifyCompactionStats(pl_stats, expected_pl_stats);
+
+    const auto& output_level_stats = stats[output_level];
+    CompactionJobStats expected_job_stats;
+    expected_job_stats.cpu_micros = output_level_stats.cpu_micros;
+    expected_job_stats.num_input_files =
+        output_level_stats.num_input_files_in_output_level +
+        output_level_stats.num_input_files_in_non_output_levels;
+    expected_job_stats.num_input_records =
+        output_level_stats.num_input_records - num_input_range_del;
+    expected_job_stats.num_output_files =
+        output_level_stats.num_output_files + pl_stats.num_output_files;
+    expected_job_stats.num_output_records =
+        output_level_stats.num_output_records + pl_stats.num_output_records;
+    VerifyCompactionJobStats(job_stats, expected_job_stats);
   }
 
   void ResetAllStats(std::vector<InternalStats::CompactionStats>& stats,
@@ -139,42 +125,52 @@ class TieredCompactionTest : public DBTestBase {
   }
 
  private:
-  void CompareStats(uint64_t val, uint64_t expect) {
-    if (expect > 0) {
-      ASSERT_TRUE(val > 0);
-    } else {
-      ASSERT_EQ(val, 0);
-    }
-  }
-
   void VerifyCompactionStats(
       const InternalStats::CompactionStats& stats,
       const InternalStats::CompactionStats& expect_stats) {
-    CompareStats(stats.micros, expect_stats.micros);
-    CompareStats(stats.cpu_micros, expect_stats.cpu_micros);
-    CompareStats(stats.bytes_read_non_output_levels,
-                 expect_stats.bytes_read_non_output_levels);
-    CompareStats(stats.bytes_read_output_level,
-                 expect_stats.bytes_read_output_level);
-    CompareStats(stats.bytes_read_blob, expect_stats.bytes_read_blob);
-    CompareStats(stats.bytes_written, expect_stats.bytes_written);
-    CompareStats(stats.bytes_moved, expect_stats.bytes_moved);
-    CompareStats(stats.num_input_files_in_non_output_levels,
-                 expect_stats.num_input_files_in_non_output_levels);
-    CompareStats(stats.num_input_files_in_output_level,
-                 expect_stats.num_input_files_in_output_level);
-    CompareStats(stats.num_output_files, expect_stats.num_output_files);
-    CompareStats(stats.num_output_files_blob,
-                 expect_stats.num_output_files_blob);
-    CompareStats(stats.num_input_records, expect_stats.num_input_records);
-    CompareStats(stats.num_dropped_records, expect_stats.num_dropped_records);
-    CompareStats(stats.num_output_records, expect_stats.num_output_records);
+    ASSERT_EQ(stats.micros > 0, expect_stats.micros > 0);
+    ASSERT_EQ(stats.cpu_micros > 0, expect_stats.cpu_micros > 0);
+
+    // Hard to get consistent byte sizes of SST files.
+    // Use ASSERT_NEAR for comparison
+    ASSERT_NEAR(stats.bytes_read_non_output_levels * 1.0f,
+                expect_stats.bytes_read_non_output_levels * 1.0f,
+                stats.bytes_read_non_output_levels * 0.5f);
+    ASSERT_NEAR(stats.bytes_read_output_level * 1.0f,
+                expect_stats.bytes_read_output_level * 1.0f,
+                stats.bytes_read_output_level * 0.5f);
+    ASSERT_NEAR(stats.bytes_read_blob * 1.0f,
+                expect_stats.bytes_read_blob * 1.0f,
+                stats.bytes_read_blob * 0.5f);
+    ASSERT_NEAR(stats.bytes_written * 1.0f, expect_stats.bytes_written * 1.0f,
+                stats.bytes_written * 0.5f);
+
+    ASSERT_EQ(stats.bytes_moved, expect_stats.bytes_moved);
+    ASSERT_EQ(stats.num_input_files_in_non_output_levels,
+              expect_stats.num_input_files_in_non_output_levels);
+    ASSERT_EQ(stats.num_input_files_in_output_level,
+              expect_stats.num_input_files_in_output_level);
+    ASSERT_EQ(stats.num_output_files, expect_stats.num_output_files);
+    ASSERT_EQ(stats.num_output_files_blob, expect_stats.num_output_files_blob);
+    ASSERT_EQ(stats.num_input_records, expect_stats.num_input_records);
+    ASSERT_EQ(stats.num_dropped_records, expect_stats.num_dropped_records);
+    ASSERT_EQ(stats.num_output_records, expect_stats.num_output_records);
+
     ASSERT_EQ(stats.count, expect_stats.count);
     for (int i = 0; i < static_cast<int>(CompactionReason::kNumOfReasons);
          i++) {
       ASSERT_EQ(stats.counts[i], expect_stats.counts[i]);
     }
   }
+
+  void VerifyCompactionJobStats(const CompactionJobStats& stats,
+                                const CompactionJobStats& expected_stats) {
+    ASSERT_EQ(stats.cpu_micros, expected_stats.cpu_micros);
+    ASSERT_EQ(stats.num_input_files, expected_stats.num_input_files);
+    ASSERT_EQ(stats.num_input_records, expected_stats.num_input_records);
+    ASSERT_EQ(job_stats.num_output_files, expected_stats.num_output_files);
+    ASSERT_EQ(job_stats.num_output_records, expected_stats.num_output_records);
+  }
 };
 
 TEST_F(TieredCompactionTest, SequenceBasedTieredStorageUniversal) {
@@ -199,19 +195,37 @@ TEST_F(TieredCompactionTest, SequenceBasedTieredStorageUniversal) {
       [&](void* arg) {
         *static_cast<SequenceNumber*>(arg) = latest_cold_seq.load();
       });
+  SyncPoint::GetInstance()->SetCallBack(
+      "CompactionJob::Install:AfterUpdateCompactionJobStats", [&](void* arg) {
+        job_stats.Reset();
+        job_stats.Add(*(static_cast<CompactionJobStats*>(arg)));
+      });
   SyncPoint::GetInstance()->EnableProcessing();
 
   std::vector<InternalStats::CompactionStats> expect_stats(kNumLevels);
-  InternalStats::CompactionStats& last_stats = expect_stats[kLastLevel];
   InternalStats::CompactionStats expect_pl_stats;
 
+  // Put keys in the following way to create overlaps
+  // First file from 0 ~ 99
+  // Second file from 10 ~ 109
+  // ...
+  size_t bytes_per_file = 1952;
+  uint64_t total_input_key_count = kNumTrigger * kNumKeys;
+  uint64_t total_output_key_count = 130;  // 0 ~ 129
+
   for (int i = 0; i < kNumTrigger; i++) {
     for (int j = 0; j < kNumKeys; j++) {
       ASSERT_OK(Put(Key(i * 10 + j), "value" + std::to_string(i)));
     }
     ASSERT_OK(Flush());
+
     seq_history.emplace_back(dbfull()->GetLatestSequenceNumber());
-    expect_stats[0].Add(kBasicFlushStats);
+    InternalStats::CompactionStats flush_stats(CompactionReason::kFlush, 1);
+    flush_stats.cpu_micros = 1;
+    flush_stats.micros = 1;
+    flush_stats.bytes_written = bytes_per_file;
+    flush_stats.num_output_files = 1;
+    expect_stats[0].Add(flush_stats);
   }
   ASSERT_OK(dbfull()->TEST_WaitForCompact());
 
@@ -221,32 +235,97 @@ TEST_F(TieredCompactionTest, SequenceBasedTieredStorageUniversal) {
   ASSERT_GT(GetSstSizeHelper(Temperature::kUnknown), 0);
   ASSERT_EQ(GetSstSizeHelper(Temperature::kCold), 0);
 
-  // basic compaction stats are still counted to the last level
-  expect_stats[kLastLevel].Add(kBasicCompStats);
-  expect_pl_stats.Add(kBasicPerKeyPlacementCompStats);
+  uint64_t bytes_written_penultimate_level =
+      GetPerKeyPlacementCompactionStats().bytes_written;
 
-  VerifyCompactionStats(expect_stats, expect_pl_stats);
+  // TODO - Use designated initializer when c++20 support is required
+  {
+    InternalStats::CompactionStats last_level_compaction_stats(
+        CompactionReason::kUniversalSizeAmplification, 1);
+    last_level_compaction_stats.cpu_micros = 1;
+    last_level_compaction_stats.micros = 1;
+    last_level_compaction_stats.bytes_written = 0;
+    last_level_compaction_stats.bytes_read_non_output_levels =
+        bytes_per_file * kNumTrigger;
+    last_level_compaction_stats.num_input_files_in_non_output_levels =
+        kNumTrigger;
+    last_level_compaction_stats.num_input_records = total_input_key_count;
+    last_level_compaction_stats.num_dropped_records =
+        total_input_key_count - total_output_key_count;
+    last_level_compaction_stats.num_output_records = 0;
+    last_level_compaction_stats.num_output_files = 0;
+    expect_stats[kLastLevel].Add(last_level_compaction_stats);
+  }
+  {
+    InternalStats::CompactionStats penultimate_level_compaction_stats(
+        CompactionReason::kUniversalSizeAmplification, 1);
+    penultimate_level_compaction_stats.cpu_micros = 1;
+    penultimate_level_compaction_stats.micros = 1;
+    penultimate_level_compaction_stats.bytes_written =
+        bytes_written_penultimate_level;
+    penultimate_level_compaction_stats.num_output_files = 1;
+    penultimate_level_compaction_stats.num_output_records =
+        total_output_key_count;
+    expect_pl_stats.Add(penultimate_level_compaction_stats);
+  }
+  VerifyCompactionStats(expect_stats, expect_pl_stats, kLastLevel);
 
   ResetAllStats(expect_stats, expect_pl_stats);
 
   // move forward the cold_seq to split the file into 2 levels, so should have
-  // both the last level stats and the output_to_penultimate_level stats
+  // both the last level stats and the penultimate level stats
   latest_cold_seq = seq_history[0];
   ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
   ASSERT_EQ("0,0,0,0,0,1,1", FilesPerLevel());
 
   ASSERT_GT(GetSstSizeHelper(Temperature::kUnknown), 0);
   ASSERT_GT(GetSstSizeHelper(Temperature::kCold), 0);
 
-  last_stats.Add(kBasicCompStats);
-  last_stats.ResetCompactionReason(CompactionReason::kManualCompaction);
-  last_stats.Add(kBasicPerLevelStats);
-  last_stats.num_dropped_records = 0;
-  expect_pl_stats.Add(kBasicPerKeyPlacementCompStats);
-  expect_pl_stats.ResetCompactionReason(CompactionReason::kManualCompaction);
-  VerifyCompactionStats(expect_stats, expect_pl_stats);
+  // Now update the input count to be the total count from the previous
+  total_input_key_count = total_output_key_count;
+  uint64_t moved_to_last_level_key_count = 10;
 
-  // delete all cold data, so all data will be on penultimate level
+  // bytes read in non output = bytes written in penultimate level from previous
+  uint64_t bytes_read_in_non_output_level = bytes_written_penultimate_level;
+  uint64_t bytes_written_output_level =
+      GetCompactionStats()[kLastLevel].bytes_written;
+
+  // Now get the new bytes written in penultimate level
+  bytes_written_penultimate_level =
+      GetPerKeyPlacementCompactionStats().bytes_written;
+  {
+    InternalStats::CompactionStats last_level_compaction_stats(
+        CompactionReason::kManualCompaction, 1);
+    last_level_compaction_stats.cpu_micros = 1;
+    last_level_compaction_stats.micros = 1;
+    last_level_compaction_stats.bytes_written = bytes_written_output_level;
+    last_level_compaction_stats.bytes_read_non_output_levels =
+        bytes_read_in_non_output_level;
+    last_level_compaction_stats.num_input_files_in_non_output_levels = 1;
+    last_level_compaction_stats.num_input_records = total_input_key_count;
+    last_level_compaction_stats.num_dropped_records =
+        total_input_key_count - total_output_key_count;
+    last_level_compaction_stats.num_output_records =
+        moved_to_last_level_key_count;
+    last_level_compaction_stats.num_output_files = 1;
+    expect_stats[kLastLevel].Add(last_level_compaction_stats);
+  }
+  {
+    InternalStats::CompactionStats penultimate_level_compaction_stats(
+        CompactionReason::kManualCompaction, 1);
+    penultimate_level_compaction_stats.cpu_micros = 1;
+    penultimate_level_compaction_stats.micros = 1;
+    penultimate_level_compaction_stats.bytes_written =
+        bytes_written_penultimate_level;
+    penultimate_level_compaction_stats.num_output_files = 1;
+    penultimate_level_compaction_stats.num_output_records =
+        total_output_key_count - moved_to_last_level_key_count;
+    expect_pl_stats.Add(penultimate_level_compaction_stats);
+  }
+  VerifyCompactionStats(expect_stats, expect_pl_stats, kLastLevel);
+
+  // delete all cold data, so all data will be on proximal level
   for (int i = 0; i < 10; i++) {
     ASSERT_OK(Delete(Key(i)));
   }
@@ -255,17 +334,54 @@ TEST_F(TieredCompactionTest, SequenceBasedTieredStorageUniversal) {
   ResetAllStats(expect_stats, expect_pl_stats);
 
   ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
   ASSERT_EQ("0,0,0,0,0,1", FilesPerLevel());
   ASSERT_GT(GetSstSizeHelper(Temperature::kUnknown), 0);
   ASSERT_EQ(GetSstSizeHelper(Temperature::kCold), 0);
 
-  last_stats.Add(kBasicCompStats);
-  last_stats.ResetCompactionReason(CompactionReason::kManualCompaction);
-  last_stats.bytes_read_output_level = kHasValue;
-  last_stats.num_input_files_in_output_level = kHasValue;
-  expect_pl_stats.Add(kBasicPerKeyPlacementCompStats);
-  expect_pl_stats.ResetCompactionReason(CompactionReason::kManualCompaction);
-  VerifyCompactionStats(expect_stats, expect_pl_stats);
+  // 10 tombstones added
+  total_input_key_count = total_input_key_count + 10;
+  total_output_key_count = total_output_key_count - 10;
+
+  auto last_level_stats = GetCompactionStats()[kLastLevel];
+  bytes_written_penultimate_level =
+      GetPerKeyPlacementCompactionStats().bytes_written;
+
+  ASSERT_LT(bytes_written_penultimate_level,
+            last_level_stats.bytes_read_non_output_levels +
+                last_level_stats.bytes_read_output_level);
+  {
+    InternalStats::CompactionStats last_level_compaction_stats(
+        CompactionReason::kManualCompaction, 1);
+    last_level_compaction_stats.cpu_micros = 1;
+    last_level_compaction_stats.micros = 1;
+    last_level_compaction_stats.bytes_written = 0;
+    last_level_compaction_stats.bytes_read_non_output_levels =
+        last_level_stats.bytes_read_non_output_levels;
+    last_level_compaction_stats.bytes_read_output_level =
+        last_level_stats.bytes_read_output_level;
+    last_level_compaction_stats.num_input_files_in_non_output_levels = 2;
+    last_level_compaction_stats.num_input_files_in_output_level = 1;
+    last_level_compaction_stats.num_input_records = total_input_key_count;
+    last_level_compaction_stats.num_dropped_records =
+        total_input_key_count - total_output_key_count;
+    last_level_compaction_stats.num_output_records = 0;
+    last_level_compaction_stats.num_output_files = 0;
+    expect_stats[kLastLevel].Add(last_level_compaction_stats);
+  }
+  {
+    InternalStats::CompactionStats penultimate_level_compaction_stats(
+        CompactionReason::kManualCompaction, 1);
+    penultimate_level_compaction_stats.cpu_micros = 1;
+    penultimate_level_compaction_stats.micros = 1;
+    penultimate_level_compaction_stats.bytes_written =
+        bytes_written_penultimate_level;
+    penultimate_level_compaction_stats.num_output_files = 1;
+    penultimate_level_compaction_stats.num_output_records =
+        total_output_key_count;
+    expect_pl_stats.Add(penultimate_level_compaction_stats);
+  }
+  VerifyCompactionStats(expect_stats, expect_pl_stats, kLastLevel);
 
   // move forward the cold_seq again with range delete, take a snapshot to keep
   // the range dels in both cold and hot SSTs
@@ -275,6 +391,7 @@ TEST_F(TieredCompactionTest, SequenceBasedTieredStorageUniversal) {
   ASSERT_OK(
       db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(), start, end));
   ASSERT_OK(Flush());
+  uint64_t num_input_range_del = 1;
 
   ResetAllStats(expect_stats, expect_pl_stats);
 
@@ -283,12 +400,49 @@ TEST_F(TieredCompactionTest, SequenceBasedTieredStorageUniversal) {
   ASSERT_GT(GetSstSizeHelper(Temperature::kUnknown), 0);
   ASSERT_GT(GetSstSizeHelper(Temperature::kCold), 0);
 
-  last_stats.Add(kBasicCompStats);
-  last_stats.Add(kBasicPerLevelStats);
-  last_stats.ResetCompactionReason(CompactionReason::kManualCompaction);
-  expect_pl_stats.Add(kBasicPerKeyPlacementCompStats);
-  expect_pl_stats.ResetCompactionReason(CompactionReason::kManualCompaction);
-  VerifyCompactionStats(expect_stats, expect_pl_stats);
+  // Previous output + one delete range
+  total_input_key_count = total_output_key_count + num_input_range_del;
+  moved_to_last_level_key_count = 20;
+
+  last_level_stats = GetCompactionStats()[kLastLevel];
+  bytes_written_penultimate_level =
+      GetPerKeyPlacementCompactionStats().bytes_written;
+  // Expected to write more in last level
+  ASSERT_GT(bytes_written_penultimate_level, last_level_stats.bytes_written);
+  {
+    InternalStats::CompactionStats last_level_compaction_stats(
+        CompactionReason::kManualCompaction, 1);
+    last_level_compaction_stats.cpu_micros = 1;
+    last_level_compaction_stats.micros = 1;
+    last_level_compaction_stats.bytes_written = last_level_stats.bytes_written;
+    last_level_compaction_stats.bytes_read_non_output_levels =
+        last_level_stats.bytes_read_non_output_levels;
+    last_level_compaction_stats.bytes_read_output_level = 0;
+    last_level_compaction_stats.num_input_files_in_non_output_levels = 2;
+    last_level_compaction_stats.num_input_files_in_output_level = 0;
+    last_level_compaction_stats.num_input_records = total_input_key_count;
+    last_level_compaction_stats.num_dropped_records =
+        num_input_range_del;  // delete range tombstone
+    last_level_compaction_stats.num_output_records =
+        moved_to_last_level_key_count;
+    last_level_compaction_stats.num_output_files = 1;
+    expect_stats[kLastLevel].Add(last_level_compaction_stats);
+  }
+  {
+    InternalStats::CompactionStats penultimate_level_compaction_stats(
+        CompactionReason::kManualCompaction, 1);
+    penultimate_level_compaction_stats.cpu_micros = 1;
+    penultimate_level_compaction_stats.micros = 1;
+    penultimate_level_compaction_stats.bytes_written =
+        bytes_written_penultimate_level;
+    penultimate_level_compaction_stats.num_output_files = 1;
+    penultimate_level_compaction_stats.num_output_records =
+        total_input_key_count - moved_to_last_level_key_count -
+        num_input_range_del;
+    expect_pl_stats.Add(penultimate_level_compaction_stats);
+  }
+  VerifyCompactionStats(expect_stats, expect_pl_stats, kLastLevel,
+                        num_input_range_del);
 
   // verify data
   std::string value;
@@ -341,11 +495,11 @@ TEST_F(TieredCompactionTest, SequenceBasedTieredStorageUniversal) {
 // This test was essentially for a hacked-up version on future functionality.
 // It can be resurrected if/when a form of range-based tiering is properly
 // implemented.
+// TODO - Add stats verification when adding this test back
 TEST_F(TieredCompactionTest, DISABLED_RangeBasedTieredStorageUniversal) {
   const int kNumTrigger = 4;
   const int kNumLevels = 7;
   const int kNumKeys = 100;
-  const int kLastLevel = kNumLevels - 1;
 
   auto options = CurrentOptions();
   options.compaction_style = kCompactionStyleUniversal;
@@ -364,14 +518,13 @@ TEST_F(TieredCompactionTest, DISABLED_RangeBasedTieredStorageUniversal) {
       "CompactionIterator::PrepareOutput.context", [&](void* arg) {
         auto context = static_cast<PerKeyPlacementContext*>(arg);
         MutexLock l(&mutex);
-        context->output_to_penultimate_level =
+        context->output_to_proximal_level =
             cmp->Compare(context->key, hot_start) >= 0 &&
             cmp->Compare(context->key, hot_end) < 0;
       });
   SyncPoint::GetInstance()->EnableProcessing();
 
   std::vector<InternalStats::CompactionStats> expect_stats(kNumLevels);
-  InternalStats::CompactionStats& last_stats = expect_stats[kLastLevel];
   InternalStats::CompactionStats expect_pl_stats;
 
   for (int i = 0; i < kNumTrigger; i++) {
@@ -379,21 +532,15 @@ TEST_F(TieredCompactionTest, DISABLED_RangeBasedTieredStorageUniversal) {
       ASSERT_OK(Put(Key(j), "value" + std::to_string(j)));
     }
     ASSERT_OK(Flush());
-    expect_stats[0].Add(kBasicFlushStats);
   }
   ASSERT_OK(dbfull()->TEST_WaitForCompact());
   ASSERT_EQ("0,0,0,0,0,1,1", FilesPerLevel());
   ASSERT_GT(GetSstSizeHelper(Temperature::kUnknown), 0);
   ASSERT_GT(GetSstSizeHelper(Temperature::kCold), 0);
 
-  last_stats.Add(kBasicCompStats);
-  last_stats.Add(kBasicPerLevelStats);
-  expect_pl_stats.Add(kBasicPerKeyPlacementCompStats);
-  VerifyCompactionStats(expect_stats, expect_pl_stats);
-
   ResetAllStats(expect_stats, expect_pl_stats);
 
-  // change to all cold, no output_to_penultimate_level output
+  // change to all cold, no output_to_proximal_level output
   {
     MutexLock l(&mutex);
     hot_start = Key(100);
@@ -404,14 +551,6 @@ TEST_F(TieredCompactionTest, DISABLED_RangeBasedTieredStorageUniversal) {
   ASSERT_EQ(GetSstSizeHelper(Temperature::kUnknown), 0);
   ASSERT_GT(GetSstSizeHelper(Temperature::kCold), 0);
 
-  last_stats.Add(kBasicCompStats);
-  last_stats.ResetCompactionReason(CompactionReason::kManualCompaction);
-  last_stats.Add(kBasicPerLevelStats);
-  last_stats.num_dropped_records = 0;
-  last_stats.bytes_read_output_level = kHasValue;
-  last_stats.num_input_files_in_output_level = kHasValue;
-  VerifyCompactionStats(expect_stats, expect_pl_stats);
-
   // change to all hot, universal compaction support moving data to up level if
   // it's within compaction level range.
   {
@@ -421,7 +560,7 @@ TEST_F(TieredCompactionTest, DISABLED_RangeBasedTieredStorageUniversal) {
   }
 
   // No data is moved from cold tier to hot tier because no input files from L5
-  // or higher, it's not safe to move data to output_to_penultimate_level level.
+  // or higher, it's not safe to move data to output_to_proximal_level level.
   ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
   ASSERT_EQ("0,0,0,0,0,1", FilesPerLevel());
 
@@ -567,7 +706,7 @@ TEST_F(TieredCompactionTest, LevelColdRangeDelete) {
 
   // 20->30 will be marked as cold data, but it cannot be placed to cold tier
   // (bottommost) otherwise, it will be "deleted" by the range del in
-  // output_to_penultimate_level level verify that these data will be able to
+  // output_to_proximal_level level verify that these data will be able to
   // queried
   for (int i = 20; i < 30; i++) {
     ASSERT_OK(Put(Key(i), "value" + std::to_string(i)));
@@ -677,17 +816,17 @@ TEST_F(TieredCompactionTest, LevelOutofBoundaryRangeDelete) {
   std::vector<std::vector<FileMetaData>> level_to_files;
   dbfull()->TEST_GetFilesMetaData(dbfull()->DefaultColumnFamily(),
                                   &level_to_files);
-  // range tombstone is in the penultimate level
-  const int penultimate_level = kNumLevels - 2;
-  ASSERT_EQ(level_to_files[penultimate_level].size(), 1);
-  ASSERT_EQ(level_to_files[penultimate_level][0].num_entries, 1);
-  ASSERT_EQ(level_to_files[penultimate_level][0].num_deletions, 1);
-  ASSERT_EQ(level_to_files[penultimate_level][0].temperature,
+  // range tombstone is in the proximal level
+  const int proximal_level = kNumLevels - 2;
+  ASSERT_EQ(level_to_files[proximal_level].size(), 1);
+  ASSERT_EQ(level_to_files[proximal_level][0].num_entries, 1);
+  ASSERT_EQ(level_to_files[proximal_level][0].num_deletions, 1);
+  ASSERT_EQ(level_to_files[proximal_level][0].temperature,
             Temperature::kUnknown);
 
   ASSERT_GT(GetSstSizeHelper(Temperature::kCold), 0);
   ASSERT_EQ("0,1,10",
-            FilesPerLevel());  // one file is at the penultimate level which
+            FilesPerLevel());  // one file is at the proximal level which
                                // only contains a range delete
 
   // Add 2 hot keys, each is a new SST, they will be placed in the same level as
@@ -701,7 +840,7 @@ TEST_F(TieredCompactionTest, LevelOutofBoundaryRangeDelete) {
 
   ASSERT_OK(db_->CompactRange(cro, nullptr, nullptr));
   ASSERT_EQ("0,2,10",
-            FilesPerLevel());  // one file is at the penultimate level
+            FilesPerLevel());  // one file is at the proximal level
                                // which only contains a range delete
   std::vector<LiveFileMetaData> live_file_meta;
   db_->GetLiveFilesMetaData(&live_file_meta);
@@ -711,7 +850,7 @@ TEST_F(TieredCompactionTest, LevelOutofBoundaryRangeDelete) {
     if (meta.num_deletions > 0) {
       // found SST with del, which has 2 entries, one for data one for range del
       ASSERT_EQ(meta.level,
-                kNumLevels - 2);  // output to penultimate level
+                kNumLevels - 2);  // output to proximal level
       ASSERT_EQ(meta.num_entries, 2);
       ASSERT_EQ(meta.num_deletions, 1);
       found_sst_with_del = true;
@@ -722,7 +861,7 @@ TEST_F(TieredCompactionTest, LevelOutofBoundaryRangeDelete) {
 
   // release the first snapshot and compact, which should compact the range del
   // but new inserted key `0` and `6` are still hot data which will be placed on
-  // the penultimate level
+  // the proximal level
   db_->ReleaseSnapshot(snap);
   ASSERT_OK(db_->CompactRange(cro, nullptr, nullptr));
   ASSERT_EQ("0,2,7", FilesPerLevel());
@@ -738,7 +877,7 @@ TEST_F(TieredCompactionTest, LevelOutofBoundaryRangeDelete) {
   ASSERT_FALSE(found_sst_with_del);
 
   // Now make all data cold, key 0 will be moved to the last level, but key 6 is
-  // still in snap2, so it will be kept at the penultimate level
+  // still in snap2, so it will be kept at the proximal level
   latest_cold_seq = dbfull()->GetLatestSequenceNumber();
   ASSERT_OK(db_->CompactRange(cro, nullptr, nullptr));
   ASSERT_EQ("0,1,8", FilesPerLevel());
@@ -783,7 +922,7 @@ TEST_F(TieredCompactionTest, UniversalRangeDelete) {
   }
   ASSERT_OK(Flush());
 
-  // compact to the penultimate level with 10 files
+  // compact to the proximal level with 10 files
   CompactRangeOptions cro;
   cro.bottommost_level_compaction = BottommostLevelCompaction::kForce;
   ASSERT_OK(db_->CompactRange(cro, nullptr, nullptr));
@@ -810,7 +949,7 @@ TEST_F(TieredCompactionTest, UniversalRangeDelete) {
 
   ASSERT_EQ("0,0,0,0,0,0,8", FilesPerLevel());
 
-  // range del with snapshot should be preserved in the penultimate level
+  // range del with snapshot should be preserved in the proximal level
   auto snap = db_->GetSnapshot();
 
   start = Key(6);
@@ -841,7 +980,7 @@ TEST_F(TieredCompactionTest, UniversalRangeDelete) {
     if (meta.num_deletions > 0) {
       // found SST with del, which has 2 entries, one for data one for range del
       ASSERT_EQ(meta.level,
-                kNumLevels - 2);  // output_to_penultimate_level level
+                kNumLevels - 2);  // output_to_proximal_level level
       ASSERT_EQ(meta.num_entries, 2);
       ASSERT_EQ(meta.num_deletions, 1);
       found_sst_with_del = true;
@@ -890,6 +1029,8 @@ TEST_F(TieredCompactionTest, SequenceBasedTieredStorageLevel) {
   const int kNumKeys = 100;
   const int kLastLevel = kNumLevels - 1;
 
+  int output_level = 0;
+
   auto options = CurrentOptions();
   SetColdTemperature(options);
   options.level0_file_num_compaction_trigger = kNumTrigger;
@@ -906,18 +1047,40 @@ TEST_F(TieredCompactionTest, SequenceBasedTieredStorageLevel) {
       [&](void* arg) {
         *static_cast<SequenceNumber*>(arg) = latest_cold_seq.load();
       });
+  SyncPoint::GetInstance()->SetCallBack(
+      "CompactionJob::Install:AfterUpdateCompactionJobStats", [&](void* arg) {
+        job_stats.Reset();
+        job_stats.Add(*(static_cast<CompactionJobStats*>(arg)));
+      });
+  SyncPoint::GetInstance()->SetCallBack(
+      "CompactionJob::ProcessKeyValueCompaction()::Processing", [&](void* arg) {
+        auto compaction = static_cast<Compaction*>(arg);
+        output_level = compaction->output_level();
+      });
   SyncPoint::GetInstance()->EnableProcessing();
 
   std::vector<InternalStats::CompactionStats> expect_stats(kNumLevels);
-  InternalStats::CompactionStats& last_stats = expect_stats[kLastLevel];
   InternalStats::CompactionStats expect_pl_stats;
 
+  // Put keys in the following way to create overlaps
+  // First file from 0 ~ 99
+  // Second file from 10 ~ 109
+  // ...
+  size_t bytes_per_file = 1952;
+  uint64_t total_input_key_count = kNumTrigger * kNumKeys;
+  uint64_t total_output_key_count = 130;  // 0 ~ 129
+
   for (int i = 0; i < kNumTrigger; i++) {
     for (int j = 0; j < kNumKeys; j++) {
       ASSERT_OK(Put(Key(i * 10 + j), "value" + std::to_string(i)));
     }
     ASSERT_OK(Flush());
-    expect_stats[0].Add(kBasicFlushStats);
+    InternalStats::CompactionStats flush_stats(CompactionReason::kFlush, 1);
+    flush_stats.cpu_micros = 1;
+    flush_stats.micros = 1;
+    flush_stats.bytes_written = bytes_per_file;
+    flush_stats.num_output_files = 1;
+    expect_stats[0].Add(flush_stats);
   }
   ASSERT_OK(dbfull()->TEST_WaitForCompact());
 
@@ -926,10 +1089,30 @@ TEST_F(TieredCompactionTest, SequenceBasedTieredStorageLevel) {
   ASSERT_GT(GetSstSizeHelper(Temperature::kUnknown), 0);
   ASSERT_EQ(GetSstSizeHelper(Temperature::kCold), 0);
 
-  expect_stats[1].Add(kBasicCompStats);
-  expect_stats[1].Add(kBasicPerLevelStats);
-  expect_stats[1].ResetCompactionReason(CompactionReason::kLevelL0FilesNum);
-  VerifyCompactionStats(expect_stats, expect_pl_stats);
+  uint64_t bytes_written_output_level =
+      GetCompactionStats()[output_level].bytes_written;
+  ASSERT_GT(bytes_written_output_level, 0);
+
+  {
+    InternalStats::CompactionStats output_level_compaction_stats(
+        CompactionReason::kLevelL0FilesNum, 1);
+    output_level_compaction_stats.cpu_micros = 1;
+    output_level_compaction_stats.micros = 1;
+    output_level_compaction_stats.bytes_written = bytes_written_output_level;
+    output_level_compaction_stats.bytes_read_non_output_levels =
+        bytes_per_file * kNumTrigger;
+    output_level_compaction_stats.bytes_read_output_level = 0;
+    output_level_compaction_stats.num_input_files_in_non_output_levels =
+        kNumTrigger;
+    output_level_compaction_stats.num_input_files_in_output_level = 0;
+    output_level_compaction_stats.num_input_records = total_input_key_count;
+    output_level_compaction_stats.num_dropped_records =
+        total_input_key_count - total_output_key_count;
+    output_level_compaction_stats.num_output_records = total_output_key_count;
+    output_level_compaction_stats.num_output_files = 1;
+    expect_stats[output_level].Add(output_level_compaction_stats);
+  }
+  VerifyCompactionStats(expect_stats, expect_pl_stats, output_level);
 
   // move all data to the last level
   MoveFilesToLevel(kLastLevel);
@@ -944,15 +1127,26 @@ TEST_F(TieredCompactionTest, SequenceBasedTieredStorageLevel) {
   ASSERT_EQ(GetSstSizeHelper(Temperature::kUnknown), 0);
   ASSERT_GT(GetSstSizeHelper(Temperature::kCold), 0);
 
-  last_stats.Add(kBasicCompStats);
-  last_stats.Add(kBasicPerLevelStats);
-  last_stats.num_dropped_records = 0;
-  last_stats.bytes_read_non_output_levels = 0;
-  last_stats.num_input_files_in_non_output_levels = 0;
-  last_stats.bytes_read_output_level = kHasValue;
-  last_stats.num_input_files_in_output_level = kHasValue;
-  last_stats.ResetCompactionReason(CompactionReason::kManualCompaction);
-  VerifyCompactionStats(expect_stats, expect_pl_stats);
+  total_input_key_count = total_output_key_count;
+  {
+    InternalStats::CompactionStats output_level_compaction_stats(
+        CompactionReason::kManualCompaction, 1);
+    output_level_compaction_stats.cpu_micros = 1;
+    output_level_compaction_stats.micros = 1;
+    output_level_compaction_stats.bytes_written = bytes_written_output_level;
+    output_level_compaction_stats.bytes_read_non_output_levels = 0;
+    output_level_compaction_stats.bytes_read_output_level =
+        bytes_written_output_level;
+    output_level_compaction_stats.num_input_files_in_non_output_levels = 0;
+    output_level_compaction_stats.num_input_files_in_output_level = 1;
+    output_level_compaction_stats.num_input_records = total_input_key_count;
+    output_level_compaction_stats.num_dropped_records =
+        total_input_key_count - total_output_key_count;
+    output_level_compaction_stats.num_output_records = total_output_key_count;
+    output_level_compaction_stats.num_output_files = 1;
+    expect_stats[output_level].Add(output_level_compaction_stats);
+  }
+  VerifyCompactionStats(expect_stats, expect_pl_stats, output_level);
 
   // Add new data, which is all hot and overriding all existing data
   latest_cold_seq = dbfull()->GetLatestSequenceNumber();
@@ -976,17 +1170,47 @@ TEST_F(TieredCompactionTest, SequenceBasedTieredStorageLevel) {
   ASSERT_GT(GetSstSizeHelper(Temperature::kUnknown), 0);
   ASSERT_EQ(GetSstSizeHelper(Temperature::kCold), 0);
 
+  uint64_t bytes_written_in_proximal_level =
+      GetPerKeyPlacementCompactionStats().bytes_written;
   for (int level = 2; level < kNumLevels - 1; level++) {
-    expect_stats[level].bytes_moved = kHasValue;
+    expect_stats[level].bytes_moved = bytes_written_in_proximal_level;
   }
 
-  last_stats.Add(kBasicCompStats);
-  last_stats.bytes_read_output_level = kHasValue;
-  last_stats.num_input_files_in_output_level = kHasValue;
-  last_stats.ResetCompactionReason(CompactionReason::kManualCompaction);
-  expect_pl_stats.Add(kBasicPerKeyPlacementCompStats);
-  expect_pl_stats.ResetCompactionReason(CompactionReason::kManualCompaction);
-  VerifyCompactionStats(expect_stats, expect_pl_stats);
+  // Another set of 130 keys + from the previous
+  total_input_key_count = total_output_key_count + 130;
+  // Merged into 130
+  total_output_key_count = 130;
+
+  {
+    InternalStats::CompactionStats output_level_compaction_stats(
+        CompactionReason::kManualCompaction, 1);
+    output_level_compaction_stats.cpu_micros = 1;
+    output_level_compaction_stats.micros = 1;
+    output_level_compaction_stats.bytes_written = 0;
+    output_level_compaction_stats.bytes_read_non_output_levels =
+        bytes_written_in_proximal_level;
+    output_level_compaction_stats.bytes_read_output_level =
+        bytes_written_output_level;
+    output_level_compaction_stats.num_input_files_in_non_output_levels = 1;
+    output_level_compaction_stats.num_input_files_in_output_level = 1;
+    output_level_compaction_stats.num_input_records = total_input_key_count;
+    output_level_compaction_stats.num_dropped_records =
+        total_input_key_count - total_output_key_count;
+    output_level_compaction_stats.num_output_records = 0;
+    output_level_compaction_stats.num_output_files = 0;
+    expect_stats[output_level].Add(output_level_compaction_stats);
+  }
+  {
+    InternalStats::CompactionStats proximal_level_compaction_stats(
+        CompactionReason::kManualCompaction, 1);
+    expect_pl_stats.cpu_micros = 1;
+    expect_pl_stats.micros = 1;
+    expect_pl_stats.bytes_written = bytes_written_in_proximal_level;
+    expect_pl_stats.num_output_files = 1;
+    expect_pl_stats.num_output_records = total_output_key_count;
+    expect_pl_stats.Add(proximal_level_compaction_stats);
+  }
+  VerifyCompactionStats(expect_stats, expect_pl_stats, output_level);
 
   // move forward the cold_seq, try to split the data into cold and hot, but in
   // this case it's unsafe to split the data
@@ -1138,7 +1362,7 @@ TEST_F(TieredCompactionTest, DISABLED_RangeBasedTieredStorageLevel) {
       "CompactionIterator::PrepareOutput.context", [&](void* arg) {
         auto context = static_cast<PerKeyPlacementContext*>(arg);
         MutexLock l(&mutex);
-        context->output_to_penultimate_level =
+        context->output_to_proximal_level =
             cmp->Compare(context->key, hot_start) >= 0 &&
             cmp->Compare(context->key, hot_end) < 0;
       });
@@ -1221,10 +1445,10 @@ TEST_F(TieredCompactionTest, DISABLED_RangeBasedTieredStorageLevel) {
       options.statistics->getTickerCount(COMPACTION_RANGE_DEL_DROP_OBSOLETE),
       1);
 
-  // Tests that we only compact keys up to penultimate level
-  // that are within penultimate level input's internal key range.
-  // UPDATE: this functionality has changed. With penultimate-enabled
-  // compaction, the expanded potential output range in the penultimate
+  // Tests that we only compact keys up to proximal level
+  // that are within proximal level input's internal key range.
+  // UPDATE: this functionality has changed. With proximal-enabled
+  // compaction, the expanded potential output range in the proximal
   // level is reserved so should be safe to use.
   {
     MutexLock l(&mutex);
@@ -1376,7 +1600,7 @@ TEST_P(PrecludeLastLevelTest, MigrationFromPreserveTimeManualCompaction) {
   cro.bottommost_level_compaction = BottommostLevelCompaction::kForce;
   ASSERT_OK(db_->CompactRange(cro, nullptr, nullptr));
 
-  // all data is moved up to the penultimate level
+  // all data is moved up to the proximal level
   ASSERT_EQ("0,0,0,0,0,1", FilesPerLevel());
   ASSERT_EQ(GetSstSizeHelper(Temperature::kCold), 0);
   ASSERT_GT(GetSstSizeHelper(Temperature::kUnknown), 0);
@@ -1448,7 +1672,7 @@ TEST_P(PrecludeLastLevelTest, MigrationFromPreserveTimeAutoCompaction) {
     ASSERT_OK(dbfull()->TEST_WaitForCompact());
   }
 
-  // all data is moved up to the penultimate level
+  // all data is moved up to the proximal level
   ASSERT_EQ("0,0,0,0,0,1", FilesPerLevel());
   ASSERT_EQ(GetSstSizeHelper(Temperature::kCold), 0);
   ASSERT_GT(GetSstSizeHelper(Temperature::kUnknown), 0);
@@ -1489,8 +1713,7 @@ TEST_P(PrecludeLastLevelTest, MigrationFromPreserveTimePartial) {
   ASSERT_EQ("0,0,0,0,0,0,1", FilesPerLevel());
 
   std::vector<KeyVersion> key_versions;
-  ASSERT_OK(GetAllKeyVersions(db_, Slice(), Slice(),
-                              std::numeric_limits<size_t>::max(),
+  ASSERT_OK(GetAllKeyVersions(db_, {}, {}, std::numeric_limits<size_t>::max(),
                               &key_versions));
 
   // make sure there're more than 300 keys and first 100 keys are having seqno
@@ -1579,9 +1802,9 @@ TEST_P(PrecludeLastLevelTest, SmallPrecludeTime) {
 }
 
 TEST_P(PrecludeLastLevelTest, CheckInternalKeyRange) {
-  // When compacting keys from the last level to penultimate level,
-  // output to penultimate level should be within internal key range
-  // of input files from penultimate level.
+  // When compacting keys from the last level to proximal level,
+  // output to proximal level should be within internal key range
+  // of input files from proximal level.
   // Set up:
   // L5:
   //  File 1: DeleteRange[1, 3)@4, File 2: [3@5, 100@6]
@@ -1719,8 +1942,8 @@ TEST_P(PrecludeWithCompactStyleTest, RangeTombstoneSnapshotMigrateFromLast) {
 
   ApplyConfigChange(&options, {{"preclude_last_level_data_seconds", "10000"}});
 
-  // To exercise the WithinPenultimateLevelOutputRange feature, we want files
-  // around the middle file to be compacted on the penultimate level
+  // To exercise the WithinProximalLevelOutputRange feature, we want files
+  // around the middle file to be compacted on the proximal level
   ASSERT_OK(Put(Key(0), "val0"));
   ASSERT_OK(Flush());
   ASSERT_OK(Put(Key(3), "val3"));
@@ -1777,9 +2000,9 @@ TEST_P(PrecludeWithCompactStyleTest, RangeTombstoneSnapshotMigrateFromLast) {
   EXPECT_EQ("0,0,0,0,0,3,1", FilesPerLevel());
   VerifyLogicalState(__LINE__);
 
-  // Compact everything, but some data still goes to both penultimate and last
+  // Compact everything, but some data still goes to both proximal and last
   // levels. A full-range compaction should be safe to "migrate" data from the
-  // last level to penultimate (because of preclude setting change).
+  // last level to proximal (because of preclude setting change).
   ASSERT_OK(CompactRange({}, {}, {}));
   EXPECT_EQ("0,0,0,0,0,1,1", FilesPerLevel());
   VerifyLogicalState(__LINE__);
@@ -1898,7 +2121,7 @@ TEST_P(TimedPutPrecludeLastLevelTest, InterleavedTimedPutAndPut) {
   Close();
 }
 
-TEST_P(TimedPutPrecludeLastLevelTest, PreserveTimedPutOnPenultimateLevel) {
+TEST_P(TimedPutPrecludeLastLevelTest, PreserveTimedPutOnProximalLevel) {
   Options options = CurrentOptions();
   options.compaction_style = kCompactionStyleUniversal;
   options.disable_auto_compactions = true;
@@ -1924,14 +2147,14 @@ TEST_P(TimedPutPrecludeLastLevelTest, PreserveTimedPutOnPenultimateLevel) {
   ASSERT_OK(TimedPut(0, Key(2), "v2", kMockStartTime - 1 * 24 * 60 * 60, wo));
   ASSERT_OK(Flush());
 
-  // Should still be in penultimate level.
+  // Should still be in proximal level.
   ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
   ASSERT_EQ("0,0,0,0,0,1", FilesPerLevel());
   ASSERT_GT(GetSstSizeHelper(Temperature::kHot), 0);
   ASSERT_EQ(GetSstSizeHelper(Temperature::kCold), 0);
 
   // Wait one more day and release snapshot. Data's preferred seqno should be
-  // swapped in, but data should still stay in penultimate level. SST file's
+  // swapped in, but data should still stay in proximal level. SST file's
   // seqno to time mapping should continue to cover preferred seqno after
   // compaction.
   db_->ReleaseSnapshot(snap1);
@@ -2079,8 +2302,7 @@ TEST_P(PrecludeLastLevelTest, LastLevelOnlyCompactionPartial) {
   ASSERT_GT(GetSstSizeHelper(Temperature::kUnknown), 0);
 
   std::vector<KeyVersion> key_versions;
-  ASSERT_OK(GetAllKeyVersions(db_, Slice(), Slice(),
-                              std::numeric_limits<size_t>::max(),
+  ASSERT_OK(GetAllKeyVersions(db_, {}, {}, std::numeric_limits<size_t>::max(),
                               &key_versions));
 
   // make sure there're more than 300 keys and first 100 keys are having seqno
@@ -2253,13 +2475,13 @@ TEST_P(PrecludeLastLevelOptionalTest, LastLevelOnlyCompactionNoPreclude) {
   Close();
 }
 
-TEST_P(PrecludeLastLevelOptionalTest, PeriodicCompactionToPenultimateLevel) {
+TEST_P(PrecludeLastLevelOptionalTest, PeriodicCompactionToProximalLevel) {
   // Test the last level only periodic compaction should also be blocked by an
-  // ongoing compaction in penultimate level if tiered compaction is enabled
+  // ongoing compaction in proximal level if tiered compaction is enabled
   // otherwise, the periodic compaction should just run for the last level.
   const int kNumTrigger = 4;
   const int kNumLevels = 7;
-  const int kPenultimateLevel = kNumLevels - 2;
+  const int kProximalLevel = kNumLevels - 2;
   const int kKeyPerSec = 1;
   const int kNumKeys = 100;
 
@@ -2301,13 +2523,13 @@ TEST_P(PrecludeLastLevelOptionalTest, PeriodicCompactionToPenultimateLevel) {
   SyncPoint::GetInstance()->SetCallBack(
       "CompactionJob::ProcessKeyValueCompaction()::Processing", [&](void* arg) {
         auto compaction = static_cast<Compaction*>(arg);
-        if (compaction->output_level() == kPenultimateLevel) {
+        if (compaction->output_level() == kProximalLevel) {
           is_size_ratio_compaction_running = true;
           TEST_SYNC_POINT(
-              "PrecludeLastLevelTest::PeriodicCompactionToPenultimateLevel:"
+              "PrecludeLastLevelTest::PeriodicCompactionToProximalLevel:"
               "SizeRatioCompaction1");
           TEST_SYNC_POINT(
-              "PrecludeLastLevelTest::PeriodicCompactionToPenultimateLevel:"
+              "PrecludeLastLevelTest::PeriodicCompactionToProximalLevel:"
               "SizeRatioCompaction2");
           is_size_ratio_compaction_running = false;
         }
@@ -2329,17 +2551,17 @@ TEST_P(PrecludeLastLevelOptionalTest, PeriodicCompactionToPenultimateLevel) {
           verified_last_level_compaction = true;
         }
         TEST_SYNC_POINT(
-            "PrecludeLastLevelTest::PeriodicCompactionToPenultimateLevel:"
+            "PrecludeLastLevelTest::PeriodicCompactionToProximalLevel:"
             "AutoCompactionPicked");
       });
 
   SyncPoint::GetInstance()->LoadDependency({
-      {"PrecludeLastLevelTest::PeriodicCompactionToPenultimateLevel:"
+      {"PrecludeLastLevelTest::PeriodicCompactionToProximalLevel:"
        "SizeRatioCompaction1",
-       "PrecludeLastLevelTest::PeriodicCompactionToPenultimateLevel:DoneWrite"},
-      {"PrecludeLastLevelTest::PeriodicCompactionToPenultimateLevel:"
+       "PrecludeLastLevelTest::PeriodicCompactionToProximalLevel:DoneWrite"},
+      {"PrecludeLastLevelTest::PeriodicCompactionToProximalLevel:"
        "AutoCompactionPicked",
-       "PrecludeLastLevelTest::PeriodicCompactionToPenultimateLevel:"
+       "PrecludeLastLevelTest::PeriodicCompactionToProximalLevel:"
        "SizeRatioCompaction2"},
   });
 
@@ -2356,11 +2578,11 @@ TEST_P(PrecludeLastLevelOptionalTest, PeriodicCompactionToPenultimateLevel) {
   }
 
   TEST_SYNC_POINT(
-      "PrecludeLastLevelTest::PeriodicCompactionToPenultimateLevel:DoneWrite");
+      "PrecludeLastLevelTest::PeriodicCompactionToProximalLevel:DoneWrite");
 
   // wait for periodic compaction time and flush to trigger the periodic
   // compaction, which should be blocked by ongoing compaction in the
-  // penultimate level
+  // proximal level
   mock_clock_->MockSleepForSeconds(10000);
   for (int i = 0; i < 3 * kNumKeys; i++) {
     ASSERT_OK(Put(Key(i), rnd.RandomString(10)));
@@ -2423,7 +2645,7 @@ class ThreeRangesPartitionerFactory : public SstPartitionerFactory {
   }
 };
 
-TEST_P(PrecludeLastLevelTest, PartialPenultimateLevelCompaction) {
+TEST_P(PrecludeLastLevelTest, PartialProximalLevelCompaction) {
   const int kNumTrigger = 4;
   const int kNumLevels = 7;
   const int kKeyPerSec = 10;
@@ -2593,8 +2815,8 @@ TEST_P(PrecludeLastLevelTest, RangeDelsCauseFileEndpointsToOverlap) {
       "UniversalCompactionBuilder::PickCompaction:Return", [&](void* arg) {
         auto compaction = static_cast<Compaction*>(arg);
         if (compaction->SupportsPerKeyPlacement()) {
-          ASSERT_EQ(compaction->GetPenultimateOutputRangeType(),
-                    Compaction::PenultimateOutputRangeType::kNonLastRange);
+          ASSERT_EQ(compaction->GetProximalOutputRangeType(),
+                    Compaction::ProximalOutputRangeType::kNonLastRange);
           per_key_comp_num++;
         }
       });
@@ -2650,7 +2872,7 @@ TEST_P(PrecludeLastLevelTest, RangeDelsCauseFileEndpointsToOverlap) {
   ASSERT_EQ(3, per_key_comp_num);
   verify_db();
 
-  // Finish off the penultimate level.
+  // Finish off the proximal level.
   ASSERT_OK(db_->CompactRange(cro, nullptr, nullptr));
   ASSERT_EQ("0,0,0,0,0,0,3", FilesPerLevel());
   verify_db();

db/convenience.cc

@@ -26,6 +26,17 @@ Status DeleteFilesInRange(DB* db, ColumnFamilyHandle* column_family,
 
 Status DeleteFilesInRanges(DB* db, ColumnFamilyHandle* column_family,
                            const RangePtr* ranges, size_t n, bool include_end) {
+  std::vector<RangeOpt> range_opts(n);
+  for (size_t i = 0; i < n; ++i) {
+    range_opts[i] = {OptSlice::CopyFromPtr(ranges[i].start),
+                     OptSlice::CopyFromPtr(ranges[i].limit)};
+  }
+  return DeleteFilesInRanges(db, column_family, range_opts.data(), n,
+                             include_end);
+}
+
+Status DeleteFilesInRanges(DB* db, ColumnFamilyHandle* column_family,
+                           const RangeOpt* ranges, size_t n, bool include_end) {
   return (static_cast_with_check<DBImpl>(db->GetRootDB()))
       ->DeleteFilesInRanges(column_family, ranges, n, include_end);
 }

db/corruption_test.cc

@@ -851,6 +851,9 @@ TEST_F(CorruptionTest, ParanoidFileChecksOnCompact) {
   options.env = env_.get();
   options.paranoid_file_checks = true;
   options.create_if_missing = true;
+  // Skip verifying record count against TableProperties for
+  // MockTables
+  options.compaction_verify_record_count = false;
   Status s;
   for (const auto& mode : corruption_modes) {
     delete db_;
@@ -863,7 +866,6 @@ TEST_F(CorruptionTest, ParanoidFileChecksOnCompact) {
     ASSERT_OK(DB::Open(options, dbname_, &db_));
     assert(db_ != nullptr);  // suppress false clang-analyze report
     Build(100, 2);
-    // ASSERT_OK(db_->Flush(FlushOptions()));
     DBImpl* dbi = static_cast_with_check<DBImpl>(db_);
     ASSERT_OK(dbi->TEST_FlushMemTable());
     mock->SetCorruptionMode(mode);

db/db_basic_test.cc

@@ -161,6 +161,7 @@ TEST_F(DBBasicTest, UniqueSession) {
 
   ASSERT_EQ(sid2, sid3);
 
+  DestroyAndReopen(options);
   CreateAndReopenWithCF({"goku"}, options);
   ASSERT_OK(db_->GetDbSessionId(sid1));
   ASSERT_OK(Put("bar", "e1"));
@@ -179,6 +180,7 @@ TEST_F(DBBasicTest, UniqueSession) {
 TEST_F(DBBasicTest, ReadOnlyDB) {
   ASSERT_OK(Put("foo", "v1"));
   ASSERT_OK(Put("bar", "v2"));
+  ASSERT_OK(Flush());
   ASSERT_OK(Put("foo", "v3"));
   Close();
 
@@ -208,10 +210,11 @@ TEST_F(DBBasicTest, ReadOnlyDB) {
 
   auto options = CurrentOptions();
   assert(options.env == env_);
-  ASSERT_OK(ReadOnlyReopen(options));
+  ASSERT_OK(EnforcedReadOnlyReopen(options));
   ASSERT_EQ("v3", Get("foo"));
   ASSERT_EQ("v2", Get("bar"));
   verify_all_iters();
+  ASSERT_EQ(Flush().code(), Status::Code::kNotSupported);
   Close();
 
   // Reopen and flush memtable.
@@ -219,26 +222,38 @@ TEST_F(DBBasicTest, ReadOnlyDB) {
   ASSERT_OK(Flush());
   Close();
   // Now check keys in read only mode.
-  ASSERT_OK(ReadOnlyReopen(options));
+  ASSERT_OK(EnforcedReadOnlyReopen(options));
   ASSERT_EQ("v3", Get("foo"));
   ASSERT_EQ("v2", Get("bar"));
   verify_all_iters();
-  ASSERT_TRUE(db_->SyncWAL().IsNotSupported());
+  ASSERT_EQ(db_->SyncWAL().code(), Status::Code::kNotSupported);
+
+  // More ops that should fail
+  std::vector<ColumnFamilyHandle*> cfhs{{}};
+  ASSERT_EQ(db_->CreateColumnFamily(options, "blah", &cfhs[0]).code(),
+            Status::Code::kNotSupported);
+
+  ASSERT_EQ(db_->CreateColumnFamilies(options, {"blah"}, &cfhs).code(),
+            Status::Code::kNotSupported);
+
+  std::vector<ColumnFamilyDescriptor> cfds;
+  cfds.push_back({"blah", options});
+  ASSERT_EQ(db_->CreateColumnFamilies(cfds, &cfhs).code(),
+            Status::Code::kNotSupported);
 }
 
-// TODO akanksha: Update the test to check that combination
-// does not actually write to FS (use open read-only with
-// CompositeEnvWrapper+ReadOnlyFileSystem).
-TEST_F(DBBasicTest, DISABLED_ReadOnlyDBWithWriteDBIdToManifestSet) {
+TEST_F(DBBasicTest, ReadOnlyDBWithWriteDBIdToManifestSet) {
+  auto options = CurrentOptions();
+  options.write_dbid_to_manifest = false;
+  DestroyAndReopen(options);
   ASSERT_OK(Put("foo", "v1"));
   ASSERT_OK(Put("bar", "v2"));
   ASSERT_OK(Put("foo", "v3"));
   Close();
 
-  auto options = CurrentOptions();
   options.write_dbid_to_manifest = true;
   assert(options.env == env_);
-  ASSERT_OK(ReadOnlyReopen(options));
+  ASSERT_OK(EnforcedReadOnlyReopen(options));
   std::string db_id1;
   ASSERT_OK(db_->GetDbIdentity(db_id1));
   ASSERT_EQ("v3", Get("foo"));
@@ -258,7 +273,7 @@ TEST_F(DBBasicTest, DISABLED_ReadOnlyDBWithWriteDBIdToManifestSet) {
   ASSERT_OK(Flush());
   Close();
   // Now check keys in read only mode.
-  ASSERT_OK(ReadOnlyReopen(options));
+  ASSERT_OK(EnforcedReadOnlyReopen(options));
   ASSERT_EQ("v3", Get("foo"));
   ASSERT_EQ("v2", Get("bar"));
   ASSERT_TRUE(db_->SyncWAL().IsNotSupported());
@@ -3273,8 +3288,7 @@ TEST_F(DBBasicTest, GetAllKeyVersions) {
     ASSERT_OK(Delete(std::to_string(i)));
   }
   std::vector<KeyVersion> key_versions;
-  ASSERT_OK(GetAllKeyVersions(db_, Slice(), Slice(),
-                              std::numeric_limits<size_t>::max(),
+  ASSERT_OK(GetAllKeyVersions(db_, {}, {}, std::numeric_limits<size_t>::max(),
                               &key_versions));
   ASSERT_EQ(kNumInserts + kNumDeletes + kNumUpdates, key_versions.size());
   for (size_t i = 0; i < kNumInserts + kNumDeletes + kNumUpdates; i++) {
@@ -3284,7 +3298,7 @@ TEST_F(DBBasicTest, GetAllKeyVersions) {
       ASSERT_EQ(key_versions[i].GetTypeName(), "TypeValue");
     }
   }
-  ASSERT_OK(GetAllKeyVersions(db_, handles_[0], Slice(), Slice(),
+  ASSERT_OK(GetAllKeyVersions(db_, handles_[0], {}, {},
                               std::numeric_limits<size_t>::max(),
                               &key_versions));
   ASSERT_EQ(kNumInserts + kNumDeletes + kNumUpdates, key_versions.size());
@@ -3299,10 +3313,17 @@ TEST_F(DBBasicTest, GetAllKeyVersions) {
   for (size_t i = 0; i + 1 != kNumDeletes; ++i) {
     ASSERT_OK(Delete(1, std::to_string(i)));
   }
-  ASSERT_OK(GetAllKeyVersions(db_, handles_[1], Slice(), Slice(),
+  ASSERT_OK(GetAllKeyVersions(db_, handles_[1], {}, {},
                               std::numeric_limits<size_t>::max(),
                               &key_versions));
   ASSERT_EQ(kNumInserts + kNumDeletes + kNumUpdates - 3, key_versions.size());
+
+  // Change from historical behavior: empty key is now interpreted literally as
+  // a legal key (rather than as a "not present" key)
+  ASSERT_OK(GetAllKeyVersions(db_, handles_[1], Slice(), Slice(),
+                              std::numeric_limits<size_t>::max(),
+                              &key_versions));
+  ASSERT_EQ(key_versions.size(), 0);
 }
 
 TEST_F(DBBasicTest, ValueTypeString) {
@@ -3354,6 +3375,69 @@ TEST_F(DBBasicTest, MultiGetIOBufferOverrun) {
                      keys.data(), values.data(), statuses.data(), true);
 }
 
+TEST_F(DBBasicTest, MultiGetWithSnapshotsAndPersistedTier) {
+  Options options = CurrentOptions();
+  options.create_if_missing = true;
+  options.atomic_flush = true;
+  DestroyAndReopen(options);
+  CreateAndReopenWithCF({"cf1", "cf2"}, options);
+
+  // Insert initial data
+  ASSERT_OK(Put(0, "key1", "value1_cf0"));
+  ASSERT_OK(Put(1, "key1", "value1_cf1"));
+  ASSERT_OK(Put(2, "key1", "value1_cf2"));
+  ASSERT_OK(Flush({0, 1, 2}));
+  for (auto cf : {0, 1, 2}) {
+    ASSERT_EQ(1, NumTableFilesAtLevel(0, cf));
+  }
+
+  ASSERT_OK(Put(0, "key1", "value2_cf0"));
+  ASSERT_OK(Put(1, "key1", "value2_cf1"));
+  ASSERT_OK(Put(2, "key1", "value2_cf2"));
+
+  // Prepare for concurrent atomic flush
+  std::atomic<bool> flush_done(false);
+  std::thread flush_thread([&]() {
+    ASSERT_OK(Flush({0, 1, 2}));
+    flush_done.store(true);
+  });
+
+  // Perform MultiGet with snapshot and read_tier = kPersistentTier
+  ReadOptions ro;
+  const Snapshot* snapshot = db_->GetSnapshot();
+  ro.snapshot = snapshot;
+  ro.read_tier = kPersistedTier;
+
+  std::string k = "key1";
+  std::vector<Slice> keys(3, Slice(k));
+  std::vector<Status> statuses(keys.size());
+  std::vector<ColumnFamilyHandle*> cfs(keys.size());
+  std::vector<Slice> new_keys(keys.size());
+  std::vector<PinnableSlice> pin_values(keys.size());
+  for (size_t i = 0; i < keys.size(); ++i) {
+    cfs[i] = handles_[i];
+  }
+  db_->MultiGet(ro, cfs.size(), cfs.data(), keys.data(), pin_values.data(),
+                statuses.data());
+  for (const auto& s : statuses) {
+    ASSERT_OK(s);
+  }
+
+  if (pin_values[0] == "value1_cf0") {
+    // Check if the first value matches expected value
+    ASSERT_EQ(pin_values[1], "value1_cf1");
+    ASSERT_EQ(pin_values[2], "value1_cf2");
+  } else {
+    // If first value doesn't match, check if we got the updated values
+    ASSERT_EQ(pin_values[0], "value2_cf0");
+    ASSERT_EQ(pin_values[1], "value2_cf1");
+    ASSERT_EQ(pin_values[2], "value2_cf2");
+  }
+
+  flush_thread.join();
+  db_->ReleaseSnapshot(snapshot);
+}
+
 TEST_F(DBBasicTest, IncrementalRecoveryNoCorrupt) {
   Options options = CurrentOptions();
   DestroyAndReopen(options);
@@ -4994,6 +5078,97 @@ TEST_F(DBBasicTest, VerifyFileChecksumsReadahead) {
             (sst_size + alignment - 1) / (alignment));
 }
 
+TEST_F(DBBasicTest, DisallowMemtableWrite) {
+  // This test is mostly about what you can't do with memtable writes
+  // disallowed. For what you can do, see
+  // ExternalSSTFileBasicTest.FailIfNotBottommostLevelAndDisallowMemtable
+  Options options_allow = GetDefaultOptions();
+  options_allow.create_if_missing = true;
+  Options options_disallow = options_allow;
+  options_disallow.disallow_memtable_writes = true;
+
+  DestroyAndReopen(options_allow);
+  // CFs allowing and disallowing memtable write
+  CreateColumnFamilies({"cf1", "cf2"}, options_allow);
+  CreateColumnFamilies({"cf3"}, options_disallow);
+  // XXX: needed to get consistent handles_ mappings
+  ReopenWithColumnFamilies(
+      {"default", "cf1", "cf2", "cf3"},
+      {options_allow, options_allow, options_allow, options_disallow});
+
+  EXPECT_EQ(Put(0, "a0", "1").code(), Status::Code::kOk);
+  EXPECT_EQ(Put(1, "a1", "1").code(), Status::Code::kOk);
+  EXPECT_EQ(Put(2, "a2", "1").code(), Status::Code::kOk);
+  EXPECT_EQ(Put(3, "a3", "1").code(), Status::Code::kInvalidArgument);
+
+  EXPECT_EQ(Get(0, "a0"), "1");
+  EXPECT_EQ(Get(1, "a1"), "1");
+  EXPECT_EQ(Get(2, "a2"), "1");
+  EXPECT_EQ(Get(3, "a3"), "NOT_FOUND");
+
+  EXPECT_EQ(Delete(0, "z0").code(), Status::Code::kOk);
+  EXPECT_EQ(Delete(1, "z1").code(), Status::Code::kOk);
+  EXPECT_EQ(Delete(2, "z2").code(), Status::Code::kOk);
+  EXPECT_EQ(Delete(3, "z3").code(), Status::Code::kInvalidArgument);
+
+  WriteBatch wb;
+  EXPECT_EQ(wb.Put(handles_[0], "b0", "2").code(), Status::Code::kOk);
+  EXPECT_EQ(wb.Put(handles_[1], "b1", "2").code(), Status::Code::kOk);
+  EXPECT_EQ(wb.Put(handles_[2], "b2", "2").code(), Status::Code::kOk);
+  EXPECT_EQ(wb.Put(handles_[3], "b3", "2").code(),
+            Status::Code::kInvalidArgument);
+  ASSERT_OK(db_->Write({}, &wb));
+  wb.Clear();
+
+  EXPECT_EQ(Get(0, "b0"), "2");
+  EXPECT_EQ(Get(1, "b1"), "2");
+  EXPECT_EQ(Get(2, "b2"), "2");
+  EXPECT_EQ(Get(3, "b3"), "NOT_FOUND");
+
+  // When the DB is re-opened with WAL entries for a CF that is newly setting
+  // disallow_memtable_writes, we detect that and fail the open gracefully.
+  ASSERT_EQ(TryReopenWithColumnFamilies(
+                {"default", "cf1", "cf2", "cf3"},
+                {options_allow, options_allow, options_disallow, options_allow})
+                .code(),
+            Status::Code::kInvalidArgument);
+
+  // Successfully opening with allow creates L0 files from the WAL
+  ReopenWithColumnFamilies({"default", "cf1", "cf2", "cf3"}, options_allow);
+
+  EXPECT_EQ(Get(0, "a0"), "1");
+  EXPECT_EQ(Get(1, "a1"), "1");
+  EXPECT_EQ(Get(2, "a2"), "1");
+  EXPECT_EQ(Get(3, "a3"), "NOT_FOUND");
+
+  // Now able to disallow on CF2 because no relevant WAL entries
+  ReopenWithColumnFamilies(
+      {"default", "cf1", "cf2", "cf3"},
+      {options_allow, options_allow, options_disallow, options_allow});
+
+  EXPECT_EQ(Get(0, "a0"), "1");
+  EXPECT_EQ(Get(1, "a1"), "1");
+  EXPECT_EQ(Get(2, "a2"), "1");
+  EXPECT_EQ(Get(3, "a3"), "NOT_FOUND");
+
+  // Now able to write to CF 3 but not CF 2
+  EXPECT_EQ(Put(0, "c0", "3").code(), Status::Code::kOk);
+  EXPECT_EQ(Put(1, "c1", "3").code(), Status::Code::kOk);
+  EXPECT_EQ(Put(2, "c2", "3").code(), Status::Code::kInvalidArgument);
+  EXPECT_EQ(Put(3, "c3", "3").code(), Status::Code::kOk);
+
+  EXPECT_EQ(Get(0, "c0"), "3");
+  EXPECT_EQ(Get(1, "c1"), "3");
+  EXPECT_EQ(Get(2, "c2"), "NOT_FOUND");
+  EXPECT_EQ(Get(3, "c3"), "3");
+
+  // disallow_memtable_writes not supported on default column family.
+  // (Would be complicated to make a WriteBatch aware of the setting in order
+  // to reject the write before entering the write path.)
+  Destroy(options_allow);
+  EXPECT_EQ(TryReopen(options_disallow).code(), Status::Code::kInvalidArgument);
+}
+
 // TODO: re-enable after we provide finer-grained control for WAL tracking to
 // meet the needs of different use cases, durability levels and recovery modes.
 TEST_F(DBBasicTest, DISABLED_ManualWalSync) {

db/db_compaction_test.cc

@@ -127,6 +127,19 @@ class DBCompactionTestWithParam
     exclusive_manual_compaction_ = std::get<1>(GetParam());
   }
 
+  class TrivialMoveEventListener : public EventListener {
+   public:
+    explicit TrivialMoveEventListener(size_t expected_trivially_moved_files)
+        : expected_trivially_moved_files_(expected_trivially_moved_files) {}
+    void OnCompactionBegin(DB* /*db*/, const CompactionJobInfo& ci) override {
+      ASSERT_EQ(ci.stats.num_input_files_trivially_moved,
+                expected_trivially_moved_files_);
+    }
+
+   private:
+    size_t expected_trivially_moved_files_ = 0;
+  };
+
   // Required if inheriting from testing::WithParamInterface<>
   static void SetUpTestCase() {}
   static void TearDownTestCase() {}
@@ -1301,6 +1314,9 @@ TEST_P(DBCompactionTestWithParam, TrivialMoveOneFile) {
 
   Options options = CurrentOptions();
   options.write_buffer_size = 100000000;
+  TrivialMoveEventListener* trivial_move_listener =
+      new TrivialMoveEventListener(1 /*expected_trivially_moved_files*/);
+  options.listeners.emplace_back(trivial_move_listener);
   options.max_subcompactions = max_subcompactions_;
   DestroyAndReopen(options);
 
@@ -1361,6 +1377,10 @@ TEST_P(DBCompactionTestWithParam, TrivialMoveNonOverlappingFiles) {
 
   Options options = CurrentOptions();
   options.disable_auto_compactions = true;
+  // 8 is number of `ranges` that each is a non overlapping file.
+  TrivialMoveEventListener* trivial_move_listener =
+      new TrivialMoveEventListener(8 /*expected_trivially_moved_files*/);
+  options.listeners.emplace_back(trivial_move_listener);
   options.write_buffer_size = 10 * 1024 * 1024;
   options.max_subcompactions = max_subcompactions_;
 
@@ -1408,6 +1428,11 @@ TEST_P(DBCompactionTestWithParam, TrivialMoveNonOverlappingFiles) {
   trivial_move = 0;
   non_trivial_move = 0;
   values.clear();
+  options.listeners.clear();
+  // Same ranges of files, but now overlapping, trivial move not applicable.
+  TrivialMoveEventListener* trivial_move_listener2 =
+      new TrivialMoveEventListener(0 /*expected_trivially_moved_files*/);
+  options.listeners.emplace_back(trivial_move_listener2);
   DestroyAndReopen(options);
   // Same ranges as above but overlapping
   ranges = {
@@ -1455,6 +1480,11 @@ TEST_P(DBCompactionTestWithParam, TrivialMoveTargetLevel) {
 
   Options options = CurrentOptions();
   options.disable_auto_compactions = true;
+  // Two non overlapping files in L0 trivialy moved:
+  // file 1 [0 => 300], file 2 [600 => 700]
+  TrivialMoveEventListener* trivial_move_listener1 =
+      new TrivialMoveEventListener(2 /*expected_trivially_moved_files*/);
+  options.listeners.emplace_back(trivial_move_listener1);
   options.write_buffer_size = 10 * 1024 * 1024;
   options.num_levels = 7;
   options.max_subcompactions = max_subcompactions_;
@@ -2087,13 +2117,10 @@ TEST_P(DBDeleteFileRangeTest, DeleteFilesInRanges) {
     auto begin_str1 = Key(0), end_str1 = Key(100);
     auto begin_str2 = Key(100), end_str2 = Key(200);
     auto begin_str3 = Key(200), end_str3 = Key(299);
-    Slice begin1(begin_str1), end1(end_str1);
-    Slice begin2(begin_str2), end2(end_str2);
-    Slice begin3(begin_str3), end3(end_str3);
-    std::vector<RangePtr> ranges;
-    ranges.emplace_back(&begin1, &end1);
-    ranges.emplace_back(&begin2, &end2);
-    ranges.emplace_back(&begin3, &end3);
+    std::vector<RangeOpt> ranges;
+    ranges.emplace_back(begin_str1, end_str1);
+    ranges.emplace_back(begin_str2, end_str2);
+    ranges.emplace_back(begin_str3, end_str3);
     ASSERT_OK(DeleteFilesInRanges(db_, db_->DefaultColumnFamily(),
                                   ranges.data(), ranges.size()));
     ASSERT_EQ("0,3,7", FilesPerLevel(0));
@@ -2141,7 +2168,7 @@ TEST_P(DBDeleteFileRangeTest, DeleteFilesInRanges) {
 
   // Delete all files.
   {
-    RangePtr range;
+    RangeOpt range;
     ASSERT_OK(DeleteFilesInRanges(db_, db_->DefaultColumnFamily(), &range, 1));
     ASSERT_EQ("", FilesPerLevel(0));
 
@@ -5912,6 +5939,9 @@ TEST_F(DBCompactionTest, SubcompactionEvent) {
       ASSERT_EQ(running_compactions_.find(ci.job_id),
                 running_compactions_.end());
       running_compactions_.emplace(ci.job_id, std::unordered_set<int>());
+      if (expected_num_l0_files_pre_compaction_ != -1) {
+        ASSERT_EQ(expected_num_l0_files_pre_compaction_, ci.num_l0_files);
+      }
     }
 
     void OnCompactionCompleted(DB* /*db*/,
@@ -5921,6 +5951,9 @@ TEST_F(DBCompactionTest, SubcompactionEvent) {
       ASSERT_NE(it, running_compactions_.end());
       ASSERT_EQ(it->second.size(), 0);
       running_compactions_.erase(it);
+      if (expected_num_l0_files_post_compaction_ != -1) {
+        ASSERT_EQ(expected_num_l0_files_post_compaction_, ci.num_l0_files);
+      }
     }
 
     void OnSubcompactionBegin(const SubcompactionJobInfo& si) override {
@@ -5950,10 +5983,25 @@ TEST_F(DBCompactionTest, SubcompactionEvent) {
       return total_subcompaction_cnt_;
     }
 
+    void SetExpectedNumL0FilesPreCompaction(int num) {
+      expected_num_l0_files_pre_compaction_ = num;
+    }
+
+    void SetExpectedNumL0FilesPostCompaction(int num) {
+      expected_num_l0_files_post_compaction_ = num;
+    }
+
+    void ResetExpectedNumL0Files() {
+      SetExpectedNumL0FilesPreCompaction(-1);
+      SetExpectedNumL0FilesPostCompaction(-1);
+    }
+
    private:
     InstrumentedMutex mutex_;
     std::unordered_map<int, std::unordered_set<int>> running_compactions_;
     size_t total_subcompaction_cnt_ = 0;
+    int expected_num_l0_files_pre_compaction_ = -1;
+    int expected_num_l0_files_post_compaction_ = -1;
   };
 
   Options options = CurrentOptions();
@@ -5973,6 +6021,7 @@ TEST_F(DBCompactionTest, SubcompactionEvent) {
     ASSERT_OK(Flush());
   }
   MoveFilesToLevel(2);
+  ASSERT_EQ(FilesPerLevel(), "0,0,4");
 
   // generate 2 files @ L1 which overlaps with L2 files
   for (int i = 0; i < 2; i++) {
@@ -5982,11 +6031,18 @@ TEST_F(DBCompactionTest, SubcompactionEvent) {
     }
     ASSERT_OK(Flush());
   }
+  listener->SetExpectedNumL0FilesPreCompaction(2 /* num */);
+  listener->SetExpectedNumL0FilesPostCompaction(0 /* num */);
+
   MoveFilesToLevel(1);
   ASSERT_EQ(FilesPerLevel(), "0,2,4");
 
+  listener->ResetExpectedNumL0Files();
+
   CompactRangeOptions comp_opts;
   comp_opts.max_subcompactions = 4;
+
+  listener->SetExpectedNumL0FilesPreCompaction(0 /* num */);
   Status s = dbfull()->CompactRange(comp_opts, nullptr, nullptr);
   ASSERT_OK(s);
   ASSERT_OK(dbfull()->TEST_WaitForCompact());
@@ -5994,6 +6050,8 @@ TEST_F(DBCompactionTest, SubcompactionEvent) {
   ASSERT_EQ(listener->GetRunningCompactionCount(), 0);
   // and sub compaction is triggered
   ASSERT_GT(listener->GetTotalSubcompactionCount(), 0);
+
+  listener->ResetExpectedNumL0Files();
 }
 
 TEST_F(DBCompactionTest, CompactFilesOutputRangeConflict) {
@@ -10472,7 +10530,7 @@ TEST_F(DBCompactionTest, NumberOfSubcompactions) {
   }
 }
 
-TEST_F(DBCompactionTest, VerifyRecordCount) {
+TEST_F(DBCompactionTest, VerifyInputRecordCount) {
   Options options = CurrentOptions();
   options.compaction_style = kCompactionStyleLevel;
   options.level0_file_num_compaction_trigger = 3;
@@ -10510,6 +10568,103 @@ TEST_F(DBCompactionTest, VerifyRecordCount) {
   ASSERT_TRUE(std::strstr(s.getState(), expect));
 }
 
+TEST_F(DBCompactionTest, VerifyOutputRecordCountBlockBasedTable) {
+  Options options = CurrentOptions();
+  options.compaction_style = kCompactionStyleLevel;
+  options.level0_file_num_compaction_trigger = 3;
+  options.compaction_verify_record_count = true;
+  DestroyAndReopen(options);
+  Random rnd(301);
+
+  // Create 2 overlapping L0 files
+  for (int i = 1; i < 20; i += 2) {
+    ASSERT_OK(Put(Key(i), rnd.RandomString(100)));
+  }
+  ASSERT_OK(Flush());
+
+  ASSERT_OK(db_->DeleteRange(WriteOptions(), Key(10), Key(15)));
+
+  for (int i = 0; i < 20; i += 2) {
+    ASSERT_OK(Put(Key(i), rnd.RandomString(100)));
+  }
+  ASSERT_OK(Flush());
+
+  // Skip adding every 7th key in the output table
+  int num_iter = 0;
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
+      "BlockBasedTableBuilder::Add::skip", [&](void* skip) {
+        num_iter++;
+        if (num_iter % 7 == 0) {
+          *(bool*)skip = true;
+        }
+      });
+  SyncPoint::GetInstance()->EnableProcessing();
+
+  Status s = db_->CompactRange(CompactRangeOptions(), nullptr, nullptr);
+  ASSERT_TRUE(s.IsCorruption());
+  const char* expect =
+      "Number of keys in compaction output SST files does not match number of "
+      "keys added.";
+  ASSERT_TRUE(std::strstr(s.getState(), expect));
+}
+
+TEST_F(DBCompactionTest, VerifyOutputRecordCountPlainTable) {
+  Options options = CurrentOptions();
+  options.compaction_style = kCompactionStyleLevel;
+  options.level0_file_num_compaction_trigger = 3;
+  options.compaction_verify_record_count = true;
+
+  PlainTableOptions plain_table_options;
+  plain_table_options.user_key_len = 0;
+  plain_table_options.bloom_bits_per_key = 2;
+  plain_table_options.hash_table_ratio = 0.8;
+  plain_table_options.index_sparseness = 3;
+  plain_table_options.huge_page_tlb_size = 0;
+  plain_table_options.encoding_type = kPrefix;
+  plain_table_options.full_scan_mode = false;
+  plain_table_options.store_index_in_file = false;
+
+  options.table_factory.reset(NewPlainTableFactory(plain_table_options));
+  options.memtable_factory.reset(NewHashLinkListRepFactory(4, 0, 3, true));
+
+  options.prefix_extractor.reset(NewFixedPrefixTransform(8));
+  options.allow_mmap_reads = false;
+  options.allow_concurrent_memtable_write = false;
+  options.unordered_write = false;
+
+  DestroyAndReopen(options);
+  Random rnd(301);
+
+  // Create 2 overlapping L0 files
+  for (int i = 1; i < 20; i += 2) {
+    ASSERT_OK(Put(Key(i), rnd.RandomString(100)));
+  }
+  ASSERT_OK(Flush());
+
+  for (int i = 0; i < 20; i += 2) {
+    ASSERT_OK(Put(Key(i), rnd.RandomString(100)));
+  }
+  ASSERT_OK(Flush());
+
+  // Skip adding every 7th key in the output table
+  int num_iter = 0;
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
+      "PlainTableBuilder::Add::skip", [&](void* skip) {
+        num_iter++;
+        if (num_iter % 7 == 0) {
+          *(bool*)skip = true;
+        }
+      });
+  SyncPoint::GetInstance()->EnableProcessing();
+
+  Status s = db_->CompactRange(CompactRangeOptions(), nullptr, nullptr);
+  ASSERT_TRUE(s.IsCorruption());
+  const char* expect =
+      "Number of keys in compaction output SST files does not match number of "
+      "keys added.";
+  ASSERT_TRUE(std::strstr(s.getState(), expect));
+}
+
 TEST_F(DBCompactionTest, ErrorWhenReadFileHead) {
   // This is to test a bug that is fixed in
   // https://github.com/facebook/rocksdb/pull/11782.

db/db_impl/db_impl.cc

@@ -80,6 +80,7 @@
 #include "port/port.h"
 #include "rocksdb/cache.h"
 #include "rocksdb/compaction_filter.h"
+#include "rocksdb/convenience.h"
 #include "rocksdb/db.h"
 #include "rocksdb/env.h"
 #include "rocksdb/merge_operator.h"
@@ -284,9 +285,8 @@ DBImpl::DBImpl(const DBOptions& options, const std::string& dbname,
   periodic_task_functions_.emplace(PeriodicTaskType::kFlushInfoLog,
                                    [this]() { this->FlushInfoLog(); });
   periodic_task_functions_.emplace(
-      PeriodicTaskType::kRecordSeqnoTime, [this]() {
-        this->RecordSeqnoToTimeMapping(/*populate_historical_seconds=*/0);
-      });
+      PeriodicTaskType::kRecordSeqnoTime,
+      [this]() { this->RecordSeqnoToTimeMapping(); });
 
   versions_.reset(new VersionSet(
       dbname_, &immutable_db_options_, file_options_, table_cache_.get(),
@@ -843,57 +843,40 @@ Status DBImpl::StartPeriodicTaskScheduler() {
   return s;
 }
 
-Status DBImpl::RegisterRecordSeqnoTimeWorker(const ReadOptions& read_options,
-                                             const WriteOptions& write_options,
-                                             bool is_new_db) {
+Status DBImpl::RegisterRecordSeqnoTimeWorker() {
   options_mutex_.AssertHeld();
 
-  uint64_t min_preserve_seconds = std::numeric_limits<uint64_t>::max();
-  uint64_t max_preserve_seconds = std::numeric_limits<uint64_t>::min();
-  std::vector<SuperVersionContext> sv_contexts;
+  // We assume InstallSuperVersionForConfigChange has already ensured suitable
+  // mappings are present for each relevant CF. We just need to be sure the DB's
+  // seqno_to_time_mapping_ and worker scheduler are appropriate for the
+  // combination of CF settings.
+
+  MinAndMaxPreserveSeconds preserve_info;
+  uint64_t seqno_time_cadence;
   {
     InstrumentedMutexLock l(&mutex_);
 
     for (auto cfd : *versions_->GetColumnFamilySet()) {
       auto& mopts = cfd->GetLatestMutableCFOptions();
-      // preserve time is the max of 2 options.
-      uint64_t preserve_seconds =
-          std::max(mopts.preserve_internal_time_seconds,
-                   mopts.preclude_last_level_data_seconds);
-      if (!cfd->IsDropped() && preserve_seconds > 0) {
-        min_preserve_seconds = std::min(preserve_seconds, min_preserve_seconds);
-        max_preserve_seconds = std::max(preserve_seconds, max_preserve_seconds);
+      if (!cfd->IsDropped()) {
+        preserve_info.Combine(mopts);
       }
     }
-    size_t old_mapping_size = seqno_to_time_mapping_.Size();
-    if (min_preserve_seconds == std::numeric_limits<uint64_t>::max()) {
-      // Don't track
+    seqno_time_cadence = preserve_info.GetRecodingCadence();
+    if (seqno_time_cadence == 0) {
+      // To return as much as possible to the feature being disabled,
+      // clear the existing mapping
       seqno_to_time_mapping_.SetCapacity(0);
       seqno_to_time_mapping_.SetMaxTimeSpan(UINT64_MAX);
+      assert(seqno_to_time_mapping_.Empty());
     } else {
       uint64_t cap = std::min(kMaxSeqnoToTimeEntries,
-                              max_preserve_seconds * kMaxSeqnoTimePairsPerCF /
-                                  min_preserve_seconds);
+                              preserve_info.max_preserve_seconds *
+                                  kMaxSeqnoTimePairsPerCF /
+                                  preserve_info.min_preserve_seconds);
       seqno_to_time_mapping_.SetCapacity(cap);
-      seqno_to_time_mapping_.SetMaxTimeSpan(max_preserve_seconds);
+      seqno_to_time_mapping_.SetMaxTimeSpan(preserve_info.max_preserve_seconds);
     }
-    if (old_mapping_size != seqno_to_time_mapping_.Size()) {
-      InstallSeqnoToTimeMappingInSV(&sv_contexts);
-    }
-  }
-
-  // clean up outside db mutex
-  for (SuperVersionContext& sv_context : sv_contexts) {
-    sv_context.Clean();
-  }
-  sv_contexts.clear();
-
-  uint64_t seqno_time_cadence = 0;
-  if (min_preserve_seconds != std::numeric_limits<uint64_t>::max()) {
-    // round up to 1 when the time_duration is smaller than
-    // kMaxSeqnoTimePairsPerCF
-    seqno_time_cadence = (min_preserve_seconds + kMaxSeqnoTimePairsPerCF - 1) /
-                         kMaxSeqnoTimePairsPerCF;
   }
 
   TEST_SYNC_POINT_CALLBACK(
@@ -903,64 +886,6 @@ Status DBImpl::RegisterRecordSeqnoTimeWorker(const ReadOptions& read_options,
   if (seqno_time_cadence == 0) {
     s = periodic_task_scheduler_.Unregister(PeriodicTaskType::kRecordSeqnoTime);
   } else {
-    // Before registering the periodic task, we need to be sure to fulfill two
-    // promises:
-    // 1) Any DB created with preserve/preclude options set from the beginning
-    // will get pre-allocated seqnos with pre-populated time mappings back to
-    // the times we are interested in. (This will enable future import of data
-    // while preserving rough write time. We can only do this reliably from
-    // DB::Open, as otherwise there could be a race between CreateColumnFamily
-    // and the first Write to the DB, and seqno-to-time mappings need to be
-    // monotonic.
-    // 2) In any DB, any data written after setting preserve/preclude options
-    // must have a reasonable time estimate (so that we can accurately place
-    // the data), which means at least one entry in seqno_to_time_mapping_.
-    //
-    // FIXME: We don't currently guarantee that if the first column family with
-    // that setting is added or configured after initial DB::Open but before
-    // the first user Write. Fixing this causes complications with the crash
-    // test because if DB starts without preserve/preclude option, does some
-    // user writes but all those writes are lost in crash, then re-opens with
-    // preserve/preclude option, it sees seqno==1 which looks like one of the
-    // user writes was recovered, when actually it was not.
-    bool last_seqno_zero = GetLatestSequenceNumber() == 0;
-    assert(!is_new_db || last_seqno_zero);
-    if (is_new_db && last_seqno_zero) {
-      // Pre-allocate seqnos and pre-populate historical mapping
-      // We can simply modify these, before writes are allowed
-      constexpr uint64_t kMax = kMaxSeqnoTimePairsPerSST;
-      versions_->SetLastAllocatedSequence(kMax);
-      versions_->SetLastPublishedSequence(kMax);
-      versions_->SetLastSequence(kMax);
-
-      // And record in manifest, to avoid going backwards in seqno on re-open
-      // (potentially with different options). Concurrency is simple because we
-      // are in DB::Open
-      {
-        InstrumentedMutexLock l(&mutex_);
-        VersionEdit edit;
-        edit.SetLastSequence(kMax);
-        s = versions_->LogAndApplyToDefaultColumnFamily(
-            read_options, write_options, &edit, &mutex_,
-            directories_.GetDbDir());
-        if (!s.ok() && versions_->io_status().IsIOError()) {
-          error_handler_.SetBGError(versions_->io_status(),
-                                    BackgroundErrorReason::kManifestWrite);
-        }
-      }
-
-      // Pre-populate mappings for reserved sequence numbers.
-      RecordSeqnoToTimeMapping(max_preserve_seconds);
-    } else {
-      if (!last_seqno_zero) {
-        // Ensure at least one mapping (or log a warning), and
-        // an updated entry whenever relevant SetOptions is called
-        RecordSeqnoToTimeMapping(/*populate_historical_seconds=*/0);
-      } else {
-        // FIXME (see limitation described above)
-      }
-    }
-
     s = periodic_task_scheduler_.Register(
         PeriodicTaskType::kRecordSeqnoTime,
         periodic_task_functions_.at(PeriodicTaskType::kRecordSeqnoTime),
@@ -1309,8 +1234,9 @@ Status DBImpl::SetOptions(
     //
     // (b) Append a new Version without manifest write nor DB mutex release
     //
-    // Thus aren't releasing the DB mutex again until the end of this block,
-    // after installing the new SuperVersion.
+    // Thus aren't releasing the DB mutex from LogAndApply calling pre_cb,
+    // through installing the new Version until the end of this block, after
+    // installing the new SuperVersion.
     auto pre_cb = [&]() -> Status {
       Status cb_s = cfd->SetOptions(db_options, options_map);
       if (cb_s.ok()) {
@@ -1335,7 +1261,7 @@ Status DBImpl::SetOptions(
       // Trigger possible flush/compactions. This has to be before we persist
       // options to file, otherwise there will be a deadlock with writer
       // thread.
-      InstallSuperVersionAndScheduleWork(cfd, &sv_context);
+      InstallSuperVersionForConfigChange(cfd, &sv_context);
       persist_options_status =
           WriteOptionsFile(write_options, true /*db_mutex_already_held*/);
       bg_cv_.SignalAll();
@@ -1353,8 +1279,7 @@ Status DBImpl::SetOptions(
 
   if (s.ok() && (options_map.count("preserve_internal_time_seconds") > 0 ||
                  options_map.count("preclude_last_level_data_seconds") > 0)) {
-    s = RegisterRecordSeqnoTimeWorker(read_options, write_options,
-                                      false /* is_new_db*/);
+    s = RegisterRecordSeqnoTimeWorker();
   }
 
   ROCKS_LOG_INFO(
@@ -2704,7 +2629,7 @@ Status DBImpl::MultiCFSnapshot(const ReadOptions& read_options,
     }
   };
 
-  bool last_try = false;
+  bool acquire_mutex = false;
   if (cf_list->size() == 1) {
     // Fast path for a single column family. We can simply get the thread local
     // super version
@@ -2753,29 +2678,32 @@ Status DBImpl::MultiCFSnapshot(const ReadOptions& read_options,
     // sure.
     constexpr int num_retries = 3;
     for (int i = 0; i < num_retries; ++i) {
-      last_try = (i == num_retries - 1);
+      // When reading from kPersistedTier, we want a consistent view into CFs.
+      // So we take mutex to prevent any SV change in any CF.
+      acquire_mutex = ((i == num_retries - 1) && !read_options.snapshot) ||
+                      read_options.read_tier == kPersistedTier;
       bool retry = false;
 
       if (i > 0) {
         sv_cleanup_func();
       }
       if (read_options.snapshot == nullptr) {
-        if (last_try) {
-          TEST_SYNC_POINT("DBImpl::MultiCFSnapshot::LastTry");
-          // We're close to max number of retries. For the last retry,
-          // acquire the lock so we're sure to succeed
-          mutex_.Lock();
-        }
         *snapshot = GetLastPublishedSequence();
       } else {
         *snapshot =
             static_cast_with_check<const SnapshotImpl>(read_options.snapshot)
                 ->number_;
       }
+      if (acquire_mutex) {
+        TEST_SYNC_POINT("DBImpl::MultiCFSnapshot::LastTry");
+        // We're close to max number of retries. For the last retry,
+        // acquire the lock so we're sure to succeed
+        mutex_.Lock();
+      }
       for (auto cf_iter = cf_list->begin(); cf_iter != cf_list->end();
            ++cf_iter) {
         auto node = iter_deref_func(cf_iter);
-        if (!last_try) {
+        if (!acquire_mutex) {
           if (extra_sv_ref) {
             node->super_version = node->cfd->GetReferencedSuperVersion(this);
           } else {
@@ -2799,7 +2727,7 @@ Status DBImpl::MultiCFSnapshot(const ReadOptions& read_options,
           }
         }
         TEST_SYNC_POINT("DBImpl::MultiCFSnapshot::BeforeCheckingSnapshot");
-        if (read_options.snapshot != nullptr || last_try) {
+        if (read_options.snapshot != nullptr || acquire_mutex) {
           // If user passed a snapshot, then we don't care if a memtable is
           // sealed or compaction happens because the snapshot would ensure
           // that older key versions are kept around. If this is the last
@@ -2810,7 +2738,7 @@ Status DBImpl::MultiCFSnapshot(const ReadOptions& read_options,
         // memtables, which will include immutable memtables as well, but that
         // might be tricky to maintain in case we decide, in future, to do
         // memtable compaction.
-        if (!last_try) {
+        if (!acquire_mutex) {
           SequenceNumber seq =
               node->super_version->mem->GetEarliestSequenceNumber();
           if (seq > *snapshot) {
@@ -2820,19 +2748,20 @@ Status DBImpl::MultiCFSnapshot(const ReadOptions& read_options,
         }
       }
       if (!retry) {
-        if (last_try) {
+        if (acquire_mutex) {
           mutex_.Unlock();
           TEST_SYNC_POINT("DBImpl::MultiCFSnapshot::AfterLastTryRefSV");
         }
         break;
       }
+      assert(!acquire_mutex);
     }
   }
 
   TEST_SYNC_POINT("DBImpl::MultiCFSnapshot:AfterGetSeqNum1");
   TEST_SYNC_POINT("DBImpl::MultiCFSnapshot:AfterGetSeqNum2");
   PERF_TIMER_STOP(get_snapshot_time);
-  *sv_from_thread_local = !last_try;
+  *sv_from_thread_local = !acquire_mutex;
   if (!s.ok()) {
     sv_cleanup_func();
   }
@@ -3538,7 +3467,7 @@ void DBImpl::MultiGetEntityWithCallback(
 }
 
 Status DBImpl::WrapUpCreateColumnFamilies(
-    const ReadOptions& read_options, const WriteOptions& write_options,
+    const WriteOptions& write_options,
     const std::vector<const ColumnFamilyOptions*>& cf_options) {
   options_mutex_.AssertHeld();
 
@@ -3555,8 +3484,7 @@ Status DBImpl::WrapUpCreateColumnFamilies(
   // Attempt both follow-up actions even if one fails
   Status s = WriteOptionsFile(write_options, false /*db_mutex_already_held*/);
   if (register_worker) {
-    s.UpdateIfOk(RegisterRecordSeqnoTimeWorker(read_options, write_options,
-                                               /* is_new_db */ false));
+    s.UpdateIfOk(RegisterRecordSeqnoTimeWorker());
   }
   return s;
 }
@@ -3571,8 +3499,7 @@ Status DBImpl::CreateColumnFamily(const ReadOptions& read_options,
   Status s = CreateColumnFamilyImpl(read_options, write_options, cf_options,
                                     column_family, handle);
   if (s.ok()) {
-    s.UpdateIfOk(
-        WrapUpCreateColumnFamilies(read_options, write_options, {&cf_options}));
+    s.UpdateIfOk(WrapUpCreateColumnFamilies(write_options, {&cf_options}));
   }
   return s;
 }
@@ -3599,8 +3526,7 @@ Status DBImpl::CreateColumnFamilies(
     success_once = true;
   }
   if (success_once) {
-    s.UpdateIfOk(
-        WrapUpCreateColumnFamilies(read_options, write_options, {&cf_options}));
+    s.UpdateIfOk(WrapUpCreateColumnFamilies(write_options, {&cf_options}));
   }
   return s;
 }
@@ -3630,8 +3556,7 @@ Status DBImpl::CreateColumnFamilies(
     cf_opts.push_back(&column_families[i].options);
   }
   if (success_once) {
-    s.UpdateIfOk(
-        WrapUpCreateColumnFamilies(read_options, write_options, cf_opts));
+    s.UpdateIfOk(WrapUpCreateColumnFamilies(write_options, cf_opts));
   }
   return s;
 }
@@ -3700,7 +3625,7 @@ Status DBImpl::CreateColumnFamilyImpl(const ReadOptions& read_options,
       auto* cfd =
           versions_->GetColumnFamilySet()->GetColumnFamily(column_family_name);
       assert(cfd != nullptr);
-      InstallSuperVersionAndScheduleWork(cfd, &sv_context);
+      InstallSuperVersionForConfigChange(cfd, &sv_context);
 
       if (!cfd->mem()->IsSnapshotSupported()) {
         is_snapshot_supported_ = false;
@@ -3784,7 +3709,7 @@ Status DBImpl::DropColumnFamilyImpl(ColumnFamilyHandle* column_family) {
   Status s;
   // Save re-aquiring lock for RegisterRecordSeqnoTimeWorker when not
   // applicable
-  bool used_preserve_preclude = false;
+  MinAndMaxPreserveSeconds preserve_info;
   {
     InstrumentedMutexLock l(&mutex_);
     if (cfd->IsDropped()) {
@@ -3802,8 +3727,7 @@ Status DBImpl::DropColumnFamilyImpl(ColumnFamilyHandle* column_family) {
       auto& moptions = cfd->GetLatestMutableCFOptions();
       max_total_in_memory_state_ -=
           moptions.write_buffer_size * moptions.max_write_buffer_number;
-      used_preserve_preclude = moptions.preserve_internal_time_seconds > 0 ||
-                               moptions.preclude_last_level_data_seconds > 0;
+      preserve_info.Combine(moptions);
     }
 
     if (!cf_support_snapshot) {
@@ -3821,9 +3745,8 @@ Status DBImpl::DropColumnFamilyImpl(ColumnFamilyHandle* column_family) {
     bg_cv_.SignalAll();
   }
 
-  if (used_preserve_preclude) {
-    s = RegisterRecordSeqnoTimeWorker(read_options, write_options,
-                                      /* is_new_db */ false);
+  if (preserve_info.IsEnabled()) {
+    s = RegisterRecordSeqnoTimeWorker();
   }
 
   if (s.ok()) {
@@ -4485,7 +4408,7 @@ Status DBImpl::GetPropertiesOfTablesInRange(ColumnFamilyHandle* column_family,
   // Add timestamp if needed
   for (size_t i = 0; i < n; i++) {
     auto [start, limit] = MaybeAddTimestampsToRange(
-        &range[i].start, &range[i].limit, ts_sz, &keys.emplace_back(),
+        range[i].start, range[i].limit, ts_sz, &keys.emplace_back(),
         &keys.emplace_back(), /*exclusive_end=*/false);
     assert(start.has_value());
     assert(limit.has_value());
@@ -4502,6 +4425,29 @@ Status DBImpl::GetPropertiesOfTablesInRange(ColumnFamilyHandle* column_family,
   return s;
 }
 
+Status DBImpl::GetPropertiesOfTablesByLevel(
+    ColumnFamilyHandle* column_family,
+    std::vector<std::unique_ptr<TablePropertiesCollection>>* props_by_level) {
+  auto cfh = static_cast_with_check<ColumnFamilyHandleImpl>(column_family);
+  auto cfd = cfh->cfd();
+
+  // Increment the ref count
+  mutex_.Lock();
+  auto version = cfd->current();
+  version->Ref();
+  mutex_.Unlock();
+
+  const ReadOptions read_options;
+  auto s = version->GetPropertiesOfTablesByLevel(read_options, props_by_level);
+
+  // Decrement the ref count
+  mutex_.Lock();
+  version->Unref();
+  mutex_.Unlock();
+
+  return s;
+}
+
 const std::string& DBImpl::GetName() const { return dbname_; }
 
 Env* DBImpl::GetEnv() const { return env_; }
@@ -4802,7 +4748,7 @@ void DBImpl::GetApproximateMemTableStats(ColumnFamilyHandle* column_family,
   // Add timestamp if needed
   std::string start_with_ts, limit_with_ts;
   auto [start, limit] = MaybeAddTimestampsToRange(
-      &range.start, &range.limit, ts_sz, &start_with_ts, &limit_with_ts);
+      range.start, range.limit, ts_sz, &start_with_ts, &limit_with_ts);
   assert(start.has_value());
   assert(limit.has_value());
   // Convert user_key into a corresponding internal key.
@@ -4840,9 +4786,8 @@ Status DBImpl::GetApproximateSizes(const SizeApproximationOptions& options,
   for (int i = 0; i < n; i++) {
     // Add timestamp if needed
     std::string start_with_ts, limit_with_ts;
-    auto [start, limit] =
-        MaybeAddTimestampsToRange(&range[i].start, &range[i].limit, ts_sz,
-                                  &start_with_ts, &limit_with_ts);
+    auto [start, limit] = MaybeAddTimestampsToRange(
+        range[i].start, range[i].limit, ts_sz, &start_with_ts, &limit_with_ts);
     assert(start.has_value());
     assert(limit.has_value());
     // Convert user_key into a corresponding internal key.
@@ -4918,7 +4863,7 @@ Status DBImpl::GetUpdatesSince(
 }
 
 Status DBImpl::DeleteFilesInRanges(ColumnFamilyHandle* column_family,
-                                   const RangePtr* ranges, size_t n,
+                                   const RangeOpt* ranges, size_t n,
                                    bool include_end) {
   // TODO: plumb Env::IOActivity, Env::IOPriority
   const ReadOptions read_options;
@@ -4930,7 +4875,7 @@ Status DBImpl::DeleteFilesInRanges(ColumnFamilyHandle* column_family,
   const Comparator* ucmp = cfd->user_comparator();
   assert(ucmp);
   const size_t ts_sz = ucmp->timestamp_size();
-  autovector<UserKeyRangePtr> ukey_ranges;
+  autovector<UserKeyRangeOpt> ukey_ranges;
   std::vector<std::string> keys;
   std::vector<Slice> key_slices;
   ukey_ranges.reserve(n);
@@ -4940,8 +4885,8 @@ Status DBImpl::DeleteFilesInRanges(ColumnFamilyHandle* column_family,
     auto [start, limit] = MaybeAddTimestampsToRange(
         ranges[i].start, ranges[i].limit, ts_sz, &keys.emplace_back(),
         &keys.emplace_back(), !include_end);
-    assert((ranges[i].start != nullptr) == start.has_value());
-    assert((ranges[i].limit != nullptr) == limit.has_value());
+    assert(ranges[i].start.has_value() == start.has_value());
+    assert(ranges[i].limit.has_value() == limit.has_value());
     ukey_ranges.emplace_back(start, limit);
   }
 
@@ -5877,6 +5822,27 @@ Status DBImpl::IngestExternalFiles(
             "timestamps enabled doesn't support ingest behind.");
       }
     }
+    if (arg.atomic_replace_range.has_value()) {
+      if (ingest_opts.ingest_behind) {
+        return Status::InvalidArgument(
+            "Can't combine atomic_replace_range with ingest_behind.");
+      }
+      if (ingest_opts.snapshot_consistency) {
+        // TODO: support generating and ingesting a big tombstone file, which
+        // might depend on non-nullptr start and limit
+        return Status::NotSupported(
+            "atomic_replace_range not yet supported with "
+            "snapshot_consistency.");
+      } else {
+        if (arg.atomic_replace_range->start.has_value() ^
+            arg.atomic_replace_range->limit.has_value()) {
+          return Status::NotSupported(
+              "Only one of atomic_replace_range.{start,limit}.has_value() is "
+              "not supported.");
+        }
+      }
+    }
+
     if (ingest_opts.allow_db_generated_files) {
       if (ingest_opts.write_global_seqno) {
         return Status::NotSupported(
@@ -5925,8 +5891,8 @@ Status DBImpl::IngestExternalFiles(
             this);
     Status es = ingestion_jobs[i].Prepare(
         args[i].external_files, args[i].files_checksums,
-        args[i].files_checksum_func_names, args[i].file_temperature,
-        start_file_number, super_version);
+        args[i].files_checksum_func_names, args[i].atomic_replace_range,
+        args[i].file_temperature, start_file_number, super_version);
     // capture first error only
     if (!es.ok() && status.ok()) {
       status = es;
@@ -5941,8 +5907,8 @@ Status DBImpl::IngestExternalFiles(
             this);
     Status es = ingestion_jobs[0].Prepare(
         args[0].external_files, args[0].files_checksums,
-        args[0].files_checksum_func_names, args[0].file_temperature,
-        next_file_number, super_version);
+        args[0].files_checksum_func_names, args[0].atomic_replace_range,
+        args[0].file_temperature, next_file_number, super_version);
     if (!es.ok()) {
       status = es;
     }
@@ -6235,7 +6201,7 @@ Status DBImpl::CreateColumnFamilyWithImport(
           versions_->LogAndApply(cfd, read_options, write_options, &dummy_edit,
                                  &mutex_, directories_.GetDbDir());
       if (status.ok()) {
-        InstallSuperVersionAndScheduleWork(cfd, &dummy_sv_ctx);
+        InstallSuperVersionForConfigChange(cfd, &dummy_sv_ctx);
       }
     }
   }
@@ -6272,7 +6238,7 @@ Status DBImpl::CreateColumnFamilyWithImport(
                                         import_job.edit(), &mutex_,
                                         directories_.GetDbDir());
         if (status.ok()) {
-          InstallSuperVersionAndScheduleWork(cfd, &sv_context);
+          InstallSuperVersionForConfigChange(cfd, &sv_context);
         }
       }
 
@@ -6333,9 +6299,9 @@ Status DBImpl::ClipColumnFamily(ColumnFamilyHandle* column_family,
 
   if (status.ok()) {
     // DeleteFilesInRanges non-overlap files except L0
-    std::vector<RangePtr> ranges;
-    ranges.emplace_back(nullptr, &begin_key);
-    ranges.emplace_back(&end_key, nullptr);
+    std::vector<RangeOpt> ranges;
+    ranges.emplace_back(OptSlice{}, begin_key);
+    ranges.emplace_back(end_key, OptSlice{});
     status = DeleteFilesInRanges(column_family, ranges.data(), ranges.size());
   }
 
@@ -6732,7 +6698,7 @@ Status DBImpl::GetCreationTimeOfOldestFile(uint64_t* creation_time) {
   }
 }
 
-void DBImpl::RecordSeqnoToTimeMapping(uint64_t populate_historical_seconds) {
+std::pair<SequenceNumber, uint64_t> DBImpl::GetSeqnoToTimeSample() const {
   // TECHNICALITY: Sample last sequence number *before* time, as prescribed
   // for SeqnoToTimeMapping. We don't know how long it has been since the last
   // sequence number was written, so we at least have a one-sided bound by
@@ -6741,63 +6707,162 @@ void DBImpl::RecordSeqnoToTimeMapping(uint64_t populate_historical_seconds) {
   // while holding the DB mutex. (This is really to make testing happy because
   // it's fine to throw out extra close-but-not-quite-consistent mappings in
   // production.)
-  std::vector<SuperVersionContext> sv_contexts;
-  bool success = true;
-  SequenceNumber seqno;
-  uint64_t unix_time;
+  mutex_.AssertHeld();
+  SequenceNumber seqno = GetLatestSequenceNumber();
+  // HACK/TODO: seqno might be zero but we can't record a mapping for that.
+  // Start with 1, which should be close enough.
+  seqno = std::max(seqno, SequenceNumber{1});
+  int64_t unix_time_signed = 0;
+  immutable_db_options_.clock->GetCurrentTime(&unix_time_signed)
+      .PermitUncheckedError();  // Ignore error
+  return {seqno, static_cast<uint64_t>(unix_time_signed)};
+}
+
+void DBImpl::EnsureSeqnoToTimeMapping(
+    const MinAndMaxPreserveSeconds& preserve_info) {
+  mutex_.AssertHeld();
+  assert(preserve_info.IsEnabled());
+
+  // Atomically with CF creation or mutable option change (see
+  // InstallSuperVersionForConfigChange()), we need to be sure any data written
+  // after setting preserve/preclude options must have a reasonable time
+  // estimate (so that we can accurately place the data), which means at least
+  // one entry in seqno_to_time_mapping_. It's not critical that `preserve_info`
+  // take into account all CFs, as that's mostly relevant to how we add
+  // recurring entries and purge old ones.
+
+  auto [seqno, unix_time_now] = GetSeqnoToTimeSample();
+  // Ensure at least one sample that is sufficiently recent
+  uint64_t unix_time_last_sample = 0;
+  if (seqno_to_time_mapping_.Empty()) {
+    // The exact best settings will be found and applied in
+    // RegisterRecordSeqnoTimeWorker()
+    seqno_to_time_mapping_.SetCapacity(kMaxSeqnoToTimeEntries);
+  } else {
+    unix_time_last_sample =
+        seqno_to_time_mapping_.GetProximalTimeBeforeSeqno(kMaxSequenceNumber);
+  }
+  uint64_t cadence = preserve_info.GetRecodingCadence();
+  // Extend cadence so as to avoid stepping on toes of recorder job, which
+  // could lag a bit.
+  cadence += 3 + cadence / 100;
+  if (unix_time_now >= cadence &&
+      unix_time_last_sample <= unix_time_now - cadence) {
+    assert(seqno > 0);  // See GetSeqnoToTimeSample()
+    // Always successful assuming seqno never go backwards
+    seqno_to_time_mapping_.Append(seqno, unix_time_now);
+  }
+}
+
+void DBImpl::PrepopulateSeqnoToTimeMapping(
+    const MinAndMaxPreserveSeconds& preserve_info) {
+  // Only for opening a new DB, with preserve/preclude options set
+  if (!preserve_info.IsEnabled()) {
+    assert(false);
+    return;
+  }
+  if (GetLatestSequenceNumber() != 0) {
+    assert(false);
+    return;
+  }
+
+  // Here we fulfill the following promise:
+  //
+  // Any DB/CF created with preserve/preclude options set from the beginning
+  // will get pre-allocated seqnos with pre-populated time mappings back to
+  // the times we are interested in. (This will enable future import of data
+  // while preserving rough write time. We can only do this reliably from
+  // DB::Open, as otherwise there could be a race between CreateColumnFamily
+  // and the first Write to the DB, and seqno-to-time mappings need to be
+  // monotonic.
+  //
+  // FIXME: We don't currently guarantee that if the first column family with
+  // that setting is added or configured after initial DB::Open but before
+  // the first user Write. Fixing this causes complications with the crash
+  // test because if DB starts without preserve/preclude option, does some
+  // user writes but all those writes are lost in crash, then re-opens with
+  // preserve/preclude option, it sees seqno==1 which looks like one of the
+  // user writes was recovered, when actually it was not.
+
+  // Pre-allocate seqnos and pre-populate historical mapping
+  // We can simply modify these, before writes are allowed
+  constexpr uint64_t kMax = kMaxSeqnoTimePairsPerSST;
+  versions_->SetLastAllocatedSequence(kMax);
+  versions_->SetLastPublishedSequence(kMax);
+  versions_->SetLastSequence(kMax);
+
+  // And record in manifest, to avoid going backwards in seqno on re-open
+  // (potentially with different options). Concurrency is simple because we
+  // are in DB::Open
+  const WriteOptions write_options(Env::IOActivity::kDBOpen);
+  const ReadOptions read_options(Env::IOActivity::kDBOpen);
+  VersionEdit edit;
+  edit.SetLastSequence(kMax);
+  Status s = versions_->LogAndApplyToDefaultColumnFamily(
+      read_options, write_options, &edit, &mutex_, directories_.GetDbDir());
+  if (!s.ok() && versions_->io_status().IsIOError()) {
+    error_handler_.SetBGError(versions_->io_status(),
+                              BackgroundErrorReason::kManifestWrite);
+  }
+
+  auto [seqno, unix_time_now] = GetSeqnoToTimeSample();
+  uint64_t populate_historical_seconds = preserve_info.max_preserve_seconds;
+  if (seqno > 1 && unix_time_now > populate_historical_seconds) {
+    // seqno=0 is reserved
+    SequenceNumber from_seqno = 1;
+    seqno_to_time_mapping_.PrePopulate(
+        from_seqno, seqno, unix_time_now - populate_historical_seconds,
+        unix_time_now);
+  } else {
+    // One of these will fail
+    assert(seqno > 1);
+    assert(unix_time_now > populate_historical_seconds);
+  }
+}
+
+void DBImpl::InstallSuperVersionForConfigChange(
+    ColumnFamilyData* cfd, SuperVersionContext* sv_context) {
+  MinAndMaxPreserveSeconds preserve_info{cfd->GetLatestCFOptions()};
+  std::shared_ptr<SeqnoToTimeMapping> new_seqno_to_time_mapping;
+  if (preserve_info.IsEnabled()) {
+    // TODO: detect & optimize if mapping hasn't changed from previous
+    // SuperVersion
+    EnsureSeqnoToTimeMapping(preserve_info);
+    new_seqno_to_time_mapping = std::make_shared<SeqnoToTimeMapping>();
+    new_seqno_to_time_mapping->CopyFrom(seqno_to_time_mapping_);
+  }
+  InstallSuperVersionAndScheduleWork(cfd, sv_context,
+                                     std::move(new_seqno_to_time_mapping));
+}
+
+void DBImpl::RecordSeqnoToTimeMapping() {
+  SuperVersionContext sv_context;
   {
     InstrumentedMutexLock l(&mutex_);
+    // Record next sample
+    seqno_to_time_mapping_.Append(GetSeqnoToTimeSample());
+    // Create an immutable snapshot for sharing across CFs
+    std::shared_ptr<SeqnoToTimeMapping> new_seqno_to_time_mapping =
+        std::make_shared<SeqnoToTimeMapping>();
+    new_seqno_to_time_mapping->CopyFrom(seqno_to_time_mapping_);
 
-    seqno = GetLatestSequenceNumber();
-    int64_t unix_time_signed = 0;
-    immutable_db_options_.clock->GetCurrentTime(&unix_time_signed)
-        .PermitUncheckedError();  // Ignore error
-    unix_time = static_cast<uint64_t>(unix_time_signed);
-
-    if (populate_historical_seconds > 0) {
-      if (seqno > 1 && unix_time > populate_historical_seconds) {
-        // seqno=0 is reserved
-        SequenceNumber from_seqno = 1;
-        success = seqno_to_time_mapping_.PrePopulate(
-            from_seqno, seqno, unix_time - populate_historical_seconds,
-            unix_time);
-        InstallSeqnoToTimeMappingInSV(&sv_contexts);
-      } else {
-        // One of these will fail
-        assert(seqno > 1);
-        assert(unix_time > populate_historical_seconds);
-        success = false;
+    // Update in SV of all applicable CFs
+    for (ColumnFamilyData* cfd : *versions_->GetColumnFamilySet()) {
+      if (cfd->IsDropped()) {
+        continue;
+      }
+      MinAndMaxPreserveSeconds preserve_info{cfd->GetLatestCFOptions()};
+      if (preserve_info.IsEnabled()) {
+        sv_context.NewSuperVersion();
+        cfd->InstallSuperVersion(&sv_context, &mutex_,
+                                 new_seqno_to_time_mapping);
       }
-    } else {
-      // FIXME: assert(seqno > 0);
-      // Always successful assuming seqno never go backwards
-      seqno_to_time_mapping_.Append(seqno, unix_time);
-      InstallSeqnoToTimeMappingInSV(&sv_contexts);
     }
+    bg_cv_.SignalAll();
   }
 
   // clean up & report outside db mutex
-  for (SuperVersionContext& sv_context : sv_contexts) {
-    sv_context.Clean();
-  }
-
-  if (populate_historical_seconds > 0) {
-    if (success) {
-      ROCKS_LOG_INFO(
-          immutable_db_options_.info_log,
-          "Pre-populated sequence number to time entries: [1,%" PRIu64
-          "] -> [%" PRIu64 ",%" PRIu64 "]",
-          seqno, unix_time - populate_historical_seconds, unix_time);
-    } else {
-      ROCKS_LOG_WARN(
-          immutable_db_options_.info_log,
-          "Failed to pre-populate sequence number to time entries: [1,%" PRIu64
-          "] -> [%" PRIu64 ",%" PRIu64 "]",
-          seqno, unix_time - populate_historical_seconds, unix_time);
-    }
-  } else {
-    assert(success);
-  }
+  sv_context.Clean();
 }
 
 void DBImpl::TrackOrUntrackFiles(
@@ -6856,22 +6921,4 @@ void DBImpl::TrackOrUntrackFiles(
   }
 }
 
-void DBImpl::InstallSeqnoToTimeMappingInSV(
-    std::vector<SuperVersionContext>* sv_contexts) {
-  mutex_.AssertHeld();
-  std::shared_ptr<SeqnoToTimeMapping> new_seqno_to_time_mapping =
-      std::make_shared<SeqnoToTimeMapping>();
-  new_seqno_to_time_mapping->CopyFrom(seqno_to_time_mapping_);
-  for (ColumnFamilyData* cfd : *versions_->GetColumnFamilySet()) {
-    if (cfd->IsDropped()) {
-      continue;
-    }
-    sv_contexts->emplace_back(/*create_superversion=*/true);
-    sv_contexts->back().new_seqno_to_time_mapping = new_seqno_to_time_mapping;
-    cfd->InstallSuperVersion(&sv_contexts->back(),
-                             cfd->GetLatestMutableCFOptions());
-  }
-  bg_cv_.SignalAll();
-}
-
 }  // namespace ROCKSDB_NAMESPACE

db/db_impl/db_impl.h

@@ -542,7 +542,7 @@ class DBImpl : public DB {
       const TransactionLogIterator::ReadOptions& read_options =
           TransactionLogIterator::ReadOptions()) override;
   Status DeleteFilesInRanges(ColumnFamilyHandle* column_family,
-                             const RangePtr* ranges, size_t n,
+                             const RangeOpt* ranges, size_t n,
                              bool include_end = true);
 
   void GetLiveFilesMetaData(std::vector<LiveFileMetaData>* metadata) override;
@@ -651,6 +651,11 @@ class DBImpl : public DB {
       ColumnFamilyHandle* column_family, const Range* range, std::size_t n,
       TablePropertiesCollection* props) override;
 
+  Status GetPropertiesOfTablesByLevel(
+      ColumnFamilyHandle* column_family,
+      std::vector<std::unique_ptr<TablePropertiesCollection>>* props_by_level)
+      override;
+
   // ---- End of implementations of the DB interface ----
   SystemClock* GetSystemClock() const;
 
@@ -1267,27 +1272,18 @@ class DBImpl : public DB {
   // flush LOG out of application buffer
   void FlushInfoLog();
 
-  // record current sequence number to time mapping. If
-  // populate_historical_seconds > 0 then pre-populate all the
-  // sequence numbers from [1, last] to map to [now minus
-  // populate_historical_seconds, now].
-  void RecordSeqnoToTimeMapping(uint64_t populate_historical_seconds);
+  // For the background timer job
+  void RecordSeqnoToTimeMapping();
 
-  // Everytime DB's seqno to time mapping changed (which already hold the db
-  // mutex), we install a new SuperVersion in each column family with a shared
-  // copy of the new mapping while holding the db mutex.
-  // This is done for all column families even though the column family does not
-  // explicitly enabled the
-  // `preclude_last_level_data_seconds` or `preserve_internal_time_seconds`
-  // features.
-  // This mapping supports iterators to fulfill the
-  // "rocksdb.iterator.write-time" iterator property for entries in memtables.
-  //
-  // Since this new SuperVersion doesn't involve an LSM tree shape change, we
-  // don't schedule work after installing this SuperVersion. It returns the used
-  // `SuperVersionContext` for clean up after release mutex.
-  void InstallSeqnoToTimeMappingInSV(
-      std::vector<SuperVersionContext>* sv_contexts);
+  // REQUIRES: DB mutex held
+  std::pair<SequenceNumber, uint64_t> GetSeqnoToTimeSample() const;
+
+  // REQUIRES: DB mutex held or during open
+  void EnsureSeqnoToTimeMapping(const MinAndMaxPreserveSeconds& preserve_secs);
+
+  // Only called during open
+  void PrepopulateSeqnoToTimeMapping(
+      const MinAndMaxPreserveSeconds& preserve_secs);
 
   // Interface to block and signal the DB in case of stalling writes by
   // WriteBufferManager. Each DBImpl object contains ptr to WBMStallInterface.
@@ -1979,7 +1975,7 @@ class DBImpl : public DB {
 
   // Follow-up work to user creating a column family or (families)
   Status WrapUpCreateColumnFamilies(
-      const ReadOptions& read_options, const WriteOptions& write_options,
+      const WriteOptions& write_options,
       const std::vector<const ColumnFamilyOptions*>& cf_options);
 
   Status DropColumnFamilyImpl(ColumnFamilyHandle* column_family);
@@ -2138,6 +2134,11 @@ class DBImpl : public DB {
       bool flushed, std::unordered_map<int, VersionEdit>* version_edits,
       RecoveryContext* recovery_ctx);
 
+  // Check that DB sequence number is not set back during recovery between
+  // replaying of WAL files and between replaying of WriteBatches.
+  Status CheckSeqnoNotSetBackDuringRecovery(SequenceNumber prev_next_seqno,
+                                            SequenceNumber current_next_seqno);
+
   void FinishLogFilesRecovery(int job_id, const Status& status);
   // The following two methods are used to flush a memtable to
   // storage. The first one is used at database RecoveryTime (when the
@@ -2450,9 +2451,7 @@ class DBImpl : public DB {
   // Cancel scheduled periodic tasks
   Status CancelPeriodicTaskScheduler();
 
-  Status RegisterRecordSeqnoTimeWorker(const ReadOptions& read_options,
-                                       const WriteOptions& write_options,
-                                       bool is_new_db);
+  Status RegisterRecordSeqnoTimeWorker();
 
   void PrintStatistics();
 
@@ -2518,12 +2517,21 @@ class DBImpl : public DB {
 
   // Background threads call this function, which is just a wrapper around
   // the InstallSuperVersion() function. Background threads carry
-  // sv_context which can have new_superversion already
-  // allocated.
+  // sv_context to allow allocation of SuperVersion object outside of holding
+  // the DB mutex.
   // All ColumnFamily state changes go through this function. Here we analyze
   // the new state and we schedule background work if we detect that the new
   // state needs flush or compaction.
-  void InstallSuperVersionAndScheduleWork(ColumnFamilyData* cfd,
+  // See also InstallSuperVersionForConfigChange().
+  void InstallSuperVersionAndScheduleWork(
+      ColumnFamilyData* cfd, SuperVersionContext* sv_context,
+      std::optional<std::shared_ptr<SeqnoToTimeMapping>>
+          new_seqno_to_time_mapping = {});
+
+  // A variant of InstallSuperVersionAndScheduleWork() that must be used for
+  // new CFs or for changes to mutable_cf_options. This is so that it can
+  // update seqno_to_time_mapping cached for the new SuperVersion as relevant.
+  void InstallSuperVersionForConfigChange(ColumnFamilyData* cfd,
                                           SuperVersionContext* sv_context);
 
   bool GetIntPropertyInternal(ColumnFamilyData* cfd,

db/db_impl/db_impl_compaction_flush.cc

@@ -981,7 +981,8 @@ Status DBImpl::CompactRange(const CompactRangeOptions& options,
 
   std::string begin_str, end_str;
   auto [begin, end] =
-      MaybeAddTimestampsToRange(begin_without_ts, end_without_ts, ts_sz,
+      MaybeAddTimestampsToRange(OptSlice::CopyFromPtr(begin_without_ts),
+                                OptSlice::CopyFromPtr(end_without_ts), ts_sz,
                                 &begin_str, &end_str, false /*exclusive_end*/);
 
   return CompactRangeInternal(
@@ -1695,11 +1696,13 @@ void DBImpl::NotifyOnCompactionBegin(ColumnFamilyData* cfd, Compaction* c,
   }
 
   c->SetNotifyOnCompactionCompleted();
+  int num_l0_files = c->input_version()->storage_info()->NumLevelFiles(0);
   // release lock while notifying events
   mutex_.Unlock();
   TEST_SYNC_POINT("DBImpl::NotifyOnCompactionBegin::UnlockMutex");
   {
     CompactionJobInfo info{};
+    info.num_l0_files = num_l0_files;
     BuildCompactionJobInfo(cfd, c, st, job_stats, job_id, &info);
     for (const auto& listener : immutable_db_options_.listeners) {
       listener->OnCompactionBegin(this, info);
@@ -1724,11 +1727,13 @@ void DBImpl::NotifyOnCompactionCompleted(
     return;
   }
 
+  int num_l0_files = cfd->current()->storage_info()->NumLevelFiles(0);
   // release lock while notifying events
   mutex_.Unlock();
   TEST_SYNC_POINT("DBImpl::NotifyOnCompactionCompleted::UnlockMutex");
   {
     CompactionJobInfo info{};
+    info.num_l0_files = num_l0_files;
     BuildCompactionJobInfo(cfd, c, st, compaction_job_stats, job_id, &info);
     for (const auto& listener : immutable_db_options_.listeners) {
       listener->OnCompactionCompleted(this, info);
@@ -3798,6 +3803,8 @@ Status DBImpl::BackgroundCompaction(bool* made_progress,
     compaction_job_stats.num_input_files = c->num_input_files(0);
     // Trivial moves do not get compacted remotely
     compaction_job_stats.is_remote_compaction = false;
+    compaction_job_stats.num_input_files_trivially_moved =
+        compaction_job_stats.num_input_files;
 
     NotifyOnCompactionBegin(c->column_family_data(), c.get(), status,
                             compaction_job_stats, job_context->job_id);
@@ -4273,9 +4280,10 @@ void DBImpl::BuildCompactionJobInfo(
 // for superversion_to_free
 
 void DBImpl::InstallSuperVersionAndScheduleWork(
-    ColumnFamilyData* cfd, SuperVersionContext* sv_context) {
+    ColumnFamilyData* cfd, SuperVersionContext* sv_context,
+    std::optional<std::shared_ptr<SeqnoToTimeMapping>>
+        new_seqno_to_time_mapping) {
   mutex_.AssertHeld();
-  const auto& mutable_cf_options = cfd->GetLatestMutableCFOptions();
 
   // Update max_total_in_memory_state_
   size_t old_memtable_size = 0;
@@ -4289,7 +4297,8 @@ void DBImpl::InstallSuperVersionAndScheduleWork(
   if (UNLIKELY(sv_context->new_superversion == nullptr)) {
     sv_context->NewSuperVersion();
   }
-  cfd->InstallSuperVersion(sv_context, mutable_cf_options);
+  cfd->InstallSuperVersion(sv_context, &mutex_,
+                           std::move(new_seqno_to_time_mapping));
 
   // There may be a small data race here. The snapshot tricking bottommost
   // compaction may already be released here. But assuming there will always be
@@ -4316,9 +4325,10 @@ void DBImpl::InstallSuperVersionAndScheduleWork(
   MaybeScheduleFlushOrCompaction();
 
   // Update max_total_in_memory_state_
-  max_total_in_memory_state_ = max_total_in_memory_state_ - old_memtable_size +
-                               mutable_cf_options.write_buffer_size *
-                                   mutable_cf_options.max_write_buffer_number;
+  max_total_in_memory_state_ =
+      max_total_in_memory_state_ - old_memtable_size +
+      cfd->GetLatestMutableCFOptions().write_buffer_size *
+          cfd->GetLatestMutableCFOptions().max_write_buffer_number;
 }
 
 // ShouldPurge is called by FindObsoleteFiles when doing a full scan,

db/db_impl/db_impl_debug.cc

@@ -379,10 +379,13 @@ void DBImpl::TEST_VerifyNoObsoleteFilesCached(
     uint64_t file_number;
     GetUnaligned(reinterpret_cast<const uint64_t*>(key.data()), &file_number);
     // Assert file is in live/quarantined set
-    if (live_and_quar_files.find(file_number) == live_and_quar_files.end()) {
+    bool cached_file_is_live_or_quar =
+        live_and_quar_files.find(file_number) != live_and_quar_files.end();
+    if (!cached_file_is_live_or_quar) {
+      // Fail with useful info
       std::cerr << "File " << file_number << " is not live nor quarantined"
                 << std::endl;
-      assert(false);
+      assert(cached_file_is_live_or_quar);
     }
   };
   table_cache_->ApplyToAllEntries(fn, {});

db/db_impl/db_impl_files.cc

@@ -368,6 +368,7 @@ void DBImpl::DeleteObsoleteFileImpl(int job_id, const std::string& fname,
                                     FileType type, uint64_t number) {
   TEST_SYNC_POINT_CALLBACK("DBImpl::DeleteObsoleteFileImpl::BeforeDeletion",
                            const_cast<std::string*>(&fname));
+  IGNORE_STATUS_IF_ERROR(Status::IOError());
 
   Status file_deletion_status;
   if (type == kTableFile || type == kBlobFile || type == kWalFile) {
@@ -423,6 +424,8 @@ void DBImpl::PurgeObsoleteFiles(JobContext& state, bool schedule_only) {
   // FindObsoleteFiles() should've populated this so nonzero
   assert(state.manifest_file_number != 0);
 
+  IGNORE_STATUS_IF_ERROR(Status::IOError());
+
   // Now, convert lists to unordered sets, WITHOUT mutex held; set is slow.
   std::unordered_set<uint64_t> sst_live_set(state.sst_live.begin(),
                                             state.sst_live.end());

db/db_impl/db_impl_open.cc

@@ -35,8 +35,8 @@ Options SanitizeOptions(const std::string& dbname, const Options& src,
   auto db_options =
       SanitizeOptions(dbname, DBOptions(src), read_only, logger_creation_s);
   ImmutableDBOptions immutable_db_options(db_options);
-  auto cf_options =
-      SanitizeOptions(immutable_db_options, ColumnFamilyOptions(src));
+  auto cf_options = SanitizeCfOptions(immutable_db_options, read_only,
+                                      ColumnFamilyOptions(src));
   return Options(db_options, cf_options);
 }
 
@@ -224,6 +224,12 @@ Status DBImpl::ValidateOptions(
     if (!s.ok()) {
       return s;
     }
+    if (cfd.name == kDefaultColumnFamilyName) {
+      if (cfd.options.disallow_memtable_writes) {
+        return Status::InvalidArgument(
+            "Default column family cannot use disallow_memtable_writes=true");
+      }
+    }
   }
   s = ValidateOptions(db_options);
   return s;
@@ -1197,6 +1203,13 @@ Status DBImpl::ProcessLogFiles(
   PredecessorWALInfo predecessor_wal_info;
 
   for (auto wal_number : wal_numbers) {
+    // Detecting early break on the next iteration after `wal_number` has been
+    // advanced since this `wal_number` doesn't affect follow-up handling after
+    // breaking out of the for loop.
+    if (!status.ok()) {
+      break;
+    }
+    SequenceNumber prev_next_sequence = *next_sequence;
     if (status.ok()) {
       status = ProcessLogFile(
           wal_number, min_wal_number, is_retry, read_only, job_id,
@@ -1204,6 +1217,10 @@ Status DBImpl::ProcessLogFiles(
           &stop_replay_by_wal_filter, &corrupted_wal_number,
           corrupted_wal_found, version_edits, &flushed, predecessor_wal_info);
     }
+    if (status.ok()) {
+      status = CheckSeqnoNotSetBackDuringRecovery(prev_next_sequence,
+                                                  *next_sequence);
+    }
   }
 
   if (status.ok()) {
@@ -1311,6 +1328,7 @@ Status DBImpl::ProcessLogFile(
     }
 
     // FIXME(hx235): consolidate `process_status` and `status`
+    SequenceNumber prev_next_sequence = *next_sequence;
     Status process_status = ProcessLogRecord(
         record, reader, running_ts_sz, wal_number, fname, read_only, job_id,
         logFileDropped, &reporter, &record_checksum, &last_seqno_observed,
@@ -1319,6 +1337,12 @@ Status DBImpl::ProcessLogFile(
 
     if (!process_status.ok()) {
       return process_status;
+    } else if (Status seqno_check_status = CheckSeqnoNotSetBackDuringRecovery(
+                   prev_next_sequence, *next_sequence);
+               !seqno_check_status.ok()) {
+      // Sequence number being set back indicates a serious software bug, the DB
+      // should not be opened in this case.
+      return seqno_check_status;
     } else if (*stop_replay_for_corruption) {
       break;
     }
@@ -1857,6 +1881,20 @@ Status DBImpl::MaybeFlushFinalMemtableOrRestoreActiveLogFiles(
   return status;
 }
 
+Status DBImpl::CheckSeqnoNotSetBackDuringRecovery(
+    SequenceNumber prev_next_seqno, SequenceNumber current_next_seqno) {
+  if (prev_next_seqno == kMaxSequenceNumber ||
+      prev_next_seqno <= current_next_seqno) {
+    return Status::OK();
+  }
+  std::string msg =
+      "Sequence number is being set backwards during recovery, this is likely "
+      "a software bug or a data corruption. Prev next seqno: " +
+      std::to_string(prev_next_seqno) +
+      " , current next seqno: " + std::to_string(current_next_seqno);
+  return Status::Corruption(msg);
+}
+
 void DBImpl::FinishLogFilesRecovery(int job_id, const Status& status) {
   event_logger_.Log() << "job" << job_id << "event"
                       << (status.ok() ? "recovery_finished" : "recovery_failed")
@@ -1989,8 +2027,9 @@ Status DBImpl::WriteLevel0TableForRecovery(int job_id, ColumnFamilyData* cfd,
     meta.oldest_ancester_time = current_time;
     meta.epoch_number = cfd->NewEpochNumber();
     {
-      auto write_hint =
-          cfd->current()->storage_info()->CalculateSSTWriteHint(/*level=*/0);
+      auto write_hint = cfd->current()->storage_info()->CalculateSSTWriteHint(
+          /*level=*/0,
+          immutable_db_options_.calculate_sst_write_lifetime_hint_set);
       mutex_.Unlock();
 
       SequenceNumber earliest_write_conflict_snapshot;
@@ -2258,6 +2297,7 @@ IOStatus DBImpl::CreateWAL(const WriteOptions& write_options,
       BuildDBOptions(immutable_db_options_, mutable_db_options_);
   FileOptions opt_file_options =
       fs_->OptimizeForLogWrite(file_options_, db_options);
+  opt_file_options.write_hint = CalculateWALWriteHint();
   // DB option takes precedence when not kUnknown
   if (immutable_db_options_.wal_write_temperature != Temperature::kUnknown) {
     opt_file_options.temperature = immutable_db_options_.wal_write_temperature;
@@ -2279,7 +2319,9 @@ IOStatus DBImpl::CreateWAL(const WriteOptions& write_options,
   }
 
   if (io_s.ok()) {
-    lfile->SetWriteLifeTimeHint(CalculateWALWriteHint());
+    // Subsequent attempts to override the hint via SetWriteLifeTimeHint
+    // with the very same value will be ignored by the fs.
+    lfile->SetWriteLifeTimeHint(opt_file_options.write_hint);
     lfile->SetPreallocationBlockSize(preallocate_block_size);
 
     const auto& listeners = immutable_db_options_.listeners;
@@ -2334,9 +2376,11 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
   handles->clear();
 
   size_t max_write_buffer_size = 0;
+  MinAndMaxPreserveSeconds preserve_info;
   for (const auto& cf : column_families) {
     max_write_buffer_size =
         std::max(max_write_buffer_size, cf.options.write_buffer_size);
+    preserve_info.Combine(cf.options);
   }
 
   auto impl = std::make_unique<DBImpl>(db_options, dbname, seq_per_batch,
@@ -2469,6 +2513,12 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
     s = impl->InitPersistStatsColumnFamily();
   }
 
+  // After reaching the post-recovery seqno but before creating SuperVersions
+  // ensure seqno to time mapping is pre-populated as needed.
+  if (s.ok() && recovery_ctx.is_new_db_ && preserve_info.IsEnabled()) {
+    impl->PrepopulateSeqnoToTimeMapping(preserve_info);
+  }
+
   if (s.ok()) {
     // set column family handles
     for (const auto& cf : column_families) {
@@ -2478,6 +2528,9 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
         handles->push_back(
             new ColumnFamilyHandleImpl(cfd, impl.get(), &impl->mutex_));
         impl->NewThreadStatusCfInfo(cfd);
+        SuperVersionContext sv_context(/* create_superversion */ true);
+        impl->InstallSuperVersionForConfigChange(cfd, &sv_context);
+        sv_context.Clean();
       } else {
         if (db_options.create_missing_column_families) {
           // missing column family, create it
@@ -2485,6 +2538,7 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
           impl->mutex_.Unlock();
           // NOTE: the work normally done in WrapUpCreateColumnFamilies will
           // be done separately below.
+          // This includes InstallSuperVersionForConfigChange.
           s = impl->CreateColumnFamilyImpl(read_options, write_options,
                                            cf.options, cf.name, &handle);
           impl->mutex_.Lock();
@@ -2501,15 +2555,14 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
     }
   }
 
-  if (s.ok()) {
-    SuperVersionContext sv_context(/* create_superversion */ true);
-    for (auto cfd : *impl->versions_->GetColumnFamilySet()) {
-      impl->InstallSuperVersionAndScheduleWork(cfd, &sv_context);
-    }
-    sv_context.Clean();
-  }
-
   if (s.ok() && impl->immutable_db_options_.persist_stats_to_disk) {
+    // Install SuperVersion for hidden column family
+    assert(impl->persist_stats_cf_handle_);
+    assert(impl->persist_stats_cf_handle_->cfd());
+    SuperVersionContext sv_context(/* create_superversion */ true);
+    impl->InstallSuperVersionForConfigChange(
+        impl->persist_stats_cf_handle_->cfd(), &sv_context);
+    sv_context.Clean();
     // try to read format version
     s = impl->PersistentStatsProcessFormatVersion();
   }
@@ -2618,8 +2671,7 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
     s = impl->StartPeriodicTaskScheduler();
   }
   if (s.ok()) {
-    s = impl->RegisterRecordSeqnoTimeWorker(read_options, write_options,
-                                            recovery_ctx.is_new_db_);
+    s = impl->RegisterRecordSeqnoTimeWorker();
   }
   impl->options_mutex_.Unlock();
   if (s.ok()) {

db/db_impl/db_impl_readonly.h

@@ -155,6 +155,29 @@ class DBImplReadOnly : public DBImpl {
     return Status::NotSupported("Not supported operation in read only mode.");
   }
 
+  using DB::CreateColumnFamily;
+  using DBImpl::CreateColumnFamily;
+  Status CreateColumnFamily(const ColumnFamilyOptions& /*cf_options*/,
+                            const std::string& /*column_family*/,
+                            ColumnFamilyHandle** /*handle*/) override {
+    return Status::NotSupported("Not supported operation in read only mode.");
+  }
+
+  using DB::CreateColumnFamilies;
+  using DBImpl::CreateColumnFamilies;
+  Status CreateColumnFamilies(
+      const ColumnFamilyOptions& /*cf_options*/,
+      const std::vector<std::string>& /*column_family_names*/,
+      std::vector<ColumnFamilyHandle*>* /*handles*/) override {
+    return Status::NotSupported("Not supported operation in read only mode.");
+  }
+
+  Status CreateColumnFamilies(
+      const std::vector<ColumnFamilyDescriptor>& /*column_families*/,
+      std::vector<ColumnFamilyHandle*>* /*handles*/) override {
+    return Status::NotSupported("Not supported operation in read only mode.");
+  }
+
   // FIXME: some missing overrides for more "write" functions
 
  protected:

db/db_impl/db_impl_secondary.cc

@@ -913,6 +913,10 @@ Status DBImplSecondary::CompactWithoutInstallation(
   Status s = cfd->compaction_picker()->GetCompactionInputsFromFileNumbers(
       &input_files, &input_set, vstorage, comp_options);
   if (!s.ok()) {
+    ROCKS_LOG_ERROR(
+        immutable_db_options_.info_log,
+        "GetCompactionInputsFromFileNumbers() failed - %s.\n DebugString: %s",
+        s.ToString().c_str(), version->DebugString().c_str());
     return s;
   }
 
@@ -990,6 +994,7 @@ Status DB::OpenAndCompact(
   DBOptions db_options;
   ConfigOptions config_options;
   config_options.env = override_options.env;
+  config_options.ignore_unknown_options = true;
   std::vector<ColumnFamilyDescriptor> all_column_families;
 
   TEST_SYNC_POINT_CALLBACK(

db/db_merge_operator_test.cc

@@ -971,7 +971,7 @@ TEST_F(DBMergeOperatorTest, MaxSuccessiveMergesBaseValues) {
 
   // No base value
   {
-    constexpr char key[] = "key1";
+    const std::string key = "key1";
 
     ASSERT_OK(db_->Merge(WriteOptions(), db_->DefaultColumnFamily(), key, foo));
     ASSERT_OK(db_->Merge(WriteOptions(), db_->DefaultColumnFamily(), key, bar));
@@ -994,7 +994,7 @@ TEST_F(DBMergeOperatorTest, MaxSuccessiveMergesBaseValues) {
 
   // Plain base value
   {
-    constexpr char key[] = "key2";
+    const std::string key = "key2";
 
     ASSERT_OK(db_->Put(WriteOptions(), db_->DefaultColumnFamily(), key, foo));
     ASSERT_OK(db_->Merge(WriteOptions(), db_->DefaultColumnFamily(), key, bar));
@@ -1019,7 +1019,7 @@ TEST_F(DBMergeOperatorTest, MaxSuccessiveMergesBaseValues) {
 
   // Wide-column base value
   {
-    constexpr char key[] = "key3";
+    const std::string key = "key3";
     const WideColumns columns{{kDefaultWideColumnName, foo}, {bar, baz}};
 
     ASSERT_OK(db_->PutEntity(WriteOptions(), db_->DefaultColumnFamily(), key,

db/db_options_test.cc

@@ -70,7 +70,8 @@ class DBOptionsTest : public DBTestBase {
     options.env = env_;
     ImmutableDBOptions db_options(options);
     test::RandomInitCFOptions(&options, options, rnd);
-    auto sanitized_options = SanitizeOptions(db_options, options);
+    auto sanitized_options =
+        SanitizeCfOptions(db_options, /*read_only*/ false, options);
     auto opt_map = GetMutableCFOptionsMap(sanitized_options);
     delete options.compaction_filter;
     return opt_map;

db/db_secondary_test.cc

@@ -160,6 +160,7 @@ TEST_F(DBSecondaryTest, NonExistingDb) {
 TEST_F(DBSecondaryTest, ReopenAsSecondary) {
   Options options;
   options.env = env_;
+  options.preserve_internal_time_seconds = 300;
   Reopen(options);
   ASSERT_OK(Put("foo", "foo_value"));
   ASSERT_OK(Put("bar", "bar_value"));

db/db_table_properties_test.cc

@@ -229,6 +229,56 @@ TEST_F(DBTablePropertiesTest, CreateOnDeletionCollectorFactory) {
   ASSERT_EQ(0.5, del_factory->GetDeletionRatio());
 }
 
+TEST_F(DBTablePropertiesTest, GetPropertiesOfTablesByLevelTest) {
+  Random rnd(202);
+  Options options;
+  options.level_compaction_dynamic_level_bytes = false;
+  options.create_if_missing = true;
+  options.write_buffer_size = 4096;
+  options.max_write_buffer_number = 2;
+  options.level0_file_num_compaction_trigger = 2;
+  options.level0_slowdown_writes_trigger = 2;
+  options.level0_stop_writes_trigger = 2;
+  options.target_file_size_base = 2048;
+  options.max_bytes_for_level_base = 40960;
+  options.max_bytes_for_level_multiplier = 4;
+  options.hard_pending_compaction_bytes_limit = 16 * 1024;
+  options.num_levels = 8;
+  options.env = env_;
+
+  DestroyAndReopen(options);
+
+  // build a decent LSM
+  for (int i = 0; i < 10000; i++) {
+    EXPECT_OK(Put(test::RandomKey(&rnd, 5), rnd.RandomString(102)));
+  }
+  ASSERT_OK(Flush());
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
+  if (NumTableFilesAtLevel(0) == 0) {
+    EXPECT_OK(Put(test::RandomKey(&rnd, 5), rnd.RandomString(102)));
+    ASSERT_OK(Flush());
+  }
+
+  ASSERT_OK(db_->PauseBackgroundWork());
+
+  // Ensure that we have at least L0, L1 and L2
+  ASSERT_GT(NumTableFilesAtLevel(0), 0);
+  ASSERT_GT(NumTableFilesAtLevel(1), 0);
+  ASSERT_GT(NumTableFilesAtLevel(2), 0);
+  ColumnFamilyMetaData cf_meta;
+  db_->GetColumnFamilyMetaData(&cf_meta);
+  std::vector<std::unique_ptr<TablePropertiesCollection>> levels_props;
+  ASSERT_OK(db_->GetPropertiesOfTablesByLevel(db_->DefaultColumnFamily(),
+                                              &levels_props));
+  for (int i = 0; i < 8; i++) {
+    const std::unique_ptr<TablePropertiesCollection>& level_props =
+        levels_props[i];
+    ASSERT_EQ(level_props->size(), cf_meta.levels[i].files.size());
+  }
+
+  Close();
+}
+
 // Test params:
 // 1) whether to enable user-defined timestamps
 class DBTablePropertiesInRangeTest : public DBTestBase,
@@ -292,7 +342,7 @@ class DBTablePropertiesInRangeTest : public DBTestBase,
     keys.reserve(range_size * 2);
     for (auto& r : ranges) {
       auto [start, limit] = MaybeAddTimestampsToRange(
-          &r.start, &r.limit, ts_sz, &keys.emplace_back(), &keys.emplace_back(),
+          r.start, r.limit, ts_sz, &keys.emplace_back(), &keys.emplace_back(),
           /*exclusive_end=*/false);
       EXPECT_TRUE(start.has_value());
       EXPECT_TRUE(limit.has_value());

db/db_test.cc

@@ -59,11 +59,13 @@
 #include "rocksdb/utilities/checkpoint.h"
 #include "rocksdb/utilities/optimistic_transaction_db.h"
 #include "rocksdb/utilities/write_batch_with_index.h"
+#include "table/block_based/block_based_table_factory.h"
 #include "table/mock_table.h"
 #include "test_util/sync_point.h"
 #include "test_util/testharness.h"
 #include "test_util/testutil.h"
 #include "util/compression.h"
+#include "util/defer.h"
 #include "util/mutexlock.h"
 #include "util/random.h"
 #include "util/rate_limiter_impl.h"
@@ -3180,6 +3182,15 @@ class ModelDB : public DB {
     return Status();
   }
 
+  using DB::GetPropertiesOfTablesByLevel;
+  Status GetPropertiesOfTablesByLevel(
+      ColumnFamilyHandle* /* column_family */,
+      std::vector<
+          std::unique_ptr<TablePropertiesCollection>>* /* props_by_level */)
+      override {
+    return Status();
+  }
+
   using DB::KeyMayExist;
   bool KeyMayExist(const ReadOptions& /*options*/,
                    ColumnFamilyHandle* /*column_family*/, const Slice& /*key*/,
@@ -5400,8 +5411,7 @@ TEST_F(DBTest, DynamicLevelCompressionPerLevel) {
 
   for (const auto& file : cf_meta.levels[4].files) {
     listener->SetExpectedFileName(dbname_ + file.name);
-    Slice start(file.smallestkey), limit(file.largestkey);
-    const RangePtr ranges(&start, &limit);
+    const RangeOpt ranges(file.smallestkey, file.largestkey);
     // Given verification from above, we're guaranteed that by deleting all the
     // files in [<smallestkey>, <largestkey>] range, we're effectively deleting
     // that very single file and nothing more.
@@ -5453,6 +5463,7 @@ TEST_F(DBTest, DynamicLevelCompressionPerLevel2) {
   options.max_bytes_for_level_multiplier = 8;
   options.max_background_compactions = 1;
   options.num_levels = 5;
+  options.compaction_verify_record_count = false;
   std::shared_ptr<mock::MockTableFactory> mtf(new mock::MockTableFactory);
   options.table_factory = mtf;
 
@@ -6084,6 +6095,11 @@ TEST_F(DBTest, L0L1L2AndUpHitCounter) {
 }
 
 TEST_F(DBTest, EncodeDecompressedBlockSizeTest) {
+  bool& allow_unsupported_fv =
+      BlockBasedTableFactory::AllowUnsupportedFormatVersion();
+  SaveAndRestore guard(&allow_unsupported_fv);
+  ASSERT_FALSE(allow_unsupported_fv);
+
   // iter 0 -- zlib
   // iter 1 -- bzip2
   // iter 2 -- lz4
@@ -6106,7 +6122,16 @@ TEST_F(DBTest, EncodeDecompressedBlockSizeTest) {
       table_options.format_version = first_table_version;
       table_options.filter_policy.reset(NewBloomFilterPolicy(10));
       Options options = CurrentOptions();
+
+      // Hack to generate old files (checked in factory construction)
+      allow_unsupported_fv = true;
       options.table_factory.reset(NewBlockBasedTableFactory(table_options));
+      ASSERT_EQ(options.table_factory->GetOptions<BlockBasedTableOptions>()
+                    ->format_version,
+                first_table_version);
+      // Able to read old files without the hack
+      allow_unsupported_fv = false;
+
       options.create_if_missing = true;
       options.compression = comp;
       DestroyAndReopen(options);
@@ -6118,9 +6143,14 @@ TEST_F(DBTest, EncodeDecompressedBlockSizeTest) {
         // compressible string
         ASSERT_OK(Put(Key(i), rnd.RandomString(128) + std::string(128, 'a')));
       }
+      ASSERT_OK(Flush());
 
       table_options.format_version = first_table_version == 1 ? 2 : 1;
       options.table_factory.reset(NewBlockBasedTableFactory(table_options));
+      // format_version (for writing) is sanitized to minimum supported
+      ASSERT_EQ(options.table_factory->GetOptions<BlockBasedTableOptions>()
+                    ->format_version,
+                BlockBasedTableFactory::kMinSupportedFormatVersion);
       Reopen(options);
       for (int i = 0; i < kNumKeysWritten; ++i) {
         auto r = Get(Key(i));

db/db_test_util.cc

@@ -11,6 +11,7 @@
 
 #include "cache/cache_reservation_manager.h"
 #include "db/forward_iterator.h"
+#include "env/fs_readonly.h"
 #include "env/mock_env.h"
 #include "port/lang.h"
 #include "rocksdb/cache.h"
@@ -716,6 +717,17 @@ Status DBTestBase::ReadOnlyReopen(const Options& options) {
   return DB::OpenForReadOnly(options, dbname_, &db_);
 }
 
+Status DBTestBase::EnforcedReadOnlyReopen(const Options& options) {
+  Close();
+  Options options_copy = options;
+  MaybeInstallTimeElapseOnlySleep(options_copy);
+  auto fs_read_only =
+      std::make_shared<ReadOnlyFileSystem>(env_->GetFileSystem());
+  env_read_only_ = std::make_shared<CompositeEnvWrapper>(env_, fs_read_only);
+  options_copy.env = env_read_only_.get();
+  return DB::OpenForReadOnly(options_copy, dbname_, &db_);
+}
+
 Status DBTestBase::TryReopen(const Options& options) {
   Close();
   last_options_.table_factory.reset();

db/db_test_util.h

@@ -1062,6 +1062,7 @@ class DBTestBase : public testing::Test {
   MockEnv* mem_env_;
   Env* encrypted_env_;
   SpecialEnv* env_;
+  std::shared_ptr<Env> env_read_only_;
   std::shared_ptr<Env> env_guard_;
   DB* db_;
   std::vector<ColumnFamilyHandle*> handles_;
@@ -1178,6 +1179,9 @@ class DBTestBase : public testing::Test {
 
   Status ReadOnlyReopen(const Options& options);
 
+  // With a filesystem wrapper that fails on attempted write
+  Status EnforcedReadOnlyReopen(const Options& options);
+
   Status TryReopen(const Options& options);
 
   bool IsDirectIOSupported();

db/dbformat.h

@@ -83,6 +83,8 @@ extern const ValueType kValueTypeForSeekForPrev;
 
 // A range of user keys used internally by RocksDB. Also see `Range` used by
 // public APIs.
+// TODO: merge with Range in pubic API, but this is generally inclusive limit
+// and it is maybe exclusive limit
 struct UserKeyRange {
   // In case of user_defined timestamp, if enabled, `start` and `limit` should
   // include user_defined timestamps.
@@ -93,18 +95,17 @@ struct UserKeyRange {
   UserKeyRange(const Slice& s, const Slice& l) : start(s), limit(l) {}
 };
 
-// A range of user keys used internally by RocksDB. Also see `RangePtr` used by
+// A range of user keys used internally by RocksDB. Also see `RangeOpt` used by
 // public APIs.
-struct UserKeyRangePtr {
+struct UserKeyRangeOpt {
   // In case of user_defined timestamp, if enabled, `start` and `limit` should
   // point to key with timestamp part.
   // An optional range start, if missing, indicating a start before all keys.
-  std::optional<Slice> start;
+  OptSlice start;
   // An optional range end, if missing, indicating an end after all keys.
-  std::optional<Slice> limit;
+  OptSlice limit;
 
-  UserKeyRangePtr(const std::optional<Slice>& s, const std::optional<Slice>& l)
-      : start(s), limit(l) {}
+  UserKeyRangeOpt(const OptSlice& s, const OptSlice& l) : start(s), limit(l) {}
 };
 
 // Checks whether a type is an inline value type
@@ -469,6 +470,7 @@ class InternalKey {
 
   Slice user_key() const { return ExtractUserKey(rep_); }
   size_t size() const { return rep_.size(); }
+  bool unset() const { return rep_.empty(); }
 
   void Set(const Slice& _user_key, SequenceNumber s, ValueType t) {
     SetFrom(ParsedInternalKey(_user_key, s, t));

db/experimental.cc

@@ -57,7 +57,8 @@ Status GetFileChecksumsFromCurrentManifest(FileSystem* fs,
   }
   assert(checksum_list);
 
-  const ReadOptions read_options(Env::IOActivity::kReadManifest);
+  const ReadOptions read_options(
+      Env::IOActivity::kGetFileChecksumsFromCurrentManifest);
   checksum_list->reset();
 
   std::unique_ptr<SequentialFileReader> file_reader;

db/external_sst_file_basic_test.cc

@@ -1954,21 +1954,44 @@ TEST_F(ExternalSSTFileBasicTest, OverlappingFiles) {
     SstFileWriter sst_file_writer(EnvOptions(), options);
     std::string file3 = sst_files_dir_ + "file3.sst";
     ASSERT_OK(sst_file_writer.Open(file3));
-    ASSERT_OK(sst_file_writer.Put("j", "j1"));
+    ASSERT_OK(sst_file_writer.Put("k", "k1"));
     ASSERT_OK(sst_file_writer.Put("m", "m1"));
     ExternalSstFileInfo file3_info;
     ASSERT_OK(sst_file_writer.Finish(&file3_info));
     files.push_back(std::move(file3));
   }
 
+  // This could be ingested to the same level as file3 and file4, but the
+  // greedy/simple overlap check relegates it to a later level
+  {
+    SstFileWriter sst_file_writer(EnvOptions(), options);
+    std::string file4 = sst_files_dir_ + "file4.sst";
+    ASSERT_OK(sst_file_writer.Open(file4));
+    ASSERT_OK(sst_file_writer.Put("j", "j1"));
+    ExternalSstFileInfo file4_info;
+    ASSERT_OK(sst_file_writer.Finish(&file4_info));
+    files.push_back(std::move(file4));
+  }
+
+  {
+    SstFileWriter sst_file_writer(EnvOptions(), options);
+    std::string file5 = sst_files_dir_ + "file5.sst";
+    ASSERT_OK(sst_file_writer.Open(file5));
+    ASSERT_OK(sst_file_writer.Put("i", "i3"));
+    ExternalSstFileInfo file5_info;
+    ASSERT_OK(sst_file_writer.Finish(&file5_info));
+    files.push_back(std::move(file5));
+  }
+
   IngestExternalFileOptions ifo;
   ifo.allow_global_seqno = false;
   ASSERT_NOK(db_->IngestExternalFile(files, ifo));
   ifo.allow_global_seqno = true;
   ASSERT_OK(db_->IngestExternalFile(files, ifo));
   ASSERT_EQ(Get("a"), "a1");
-  ASSERT_EQ(Get("i"), "i2");
+  ASSERT_EQ(Get("i"), "i3");
   ASSERT_EQ(Get("j"), "j1");
+  ASSERT_EQ(Get("k"), "k1");
   ASSERT_EQ(Get("m"), "m1");
 
   int total_keys = 0;
@@ -1979,10 +2002,11 @@ TEST_F(ExternalSSTFileBasicTest, OverlappingFiles) {
   }
   ASSERT_OK(iter->status());
   delete iter;
-  ASSERT_EQ(total_keys, 4);
+  ASSERT_EQ(total_keys, 5);
 
   ASSERT_EQ(1, NumTableFilesAtLevel(6));
   ASSERT_EQ(2, NumTableFilesAtLevel(5));
+  ASSERT_EQ(2, NumTableFilesAtLevel(4));
 }
 
 class CompactionJobStatsCheckerForFilteredFiles : public EventListener {
@@ -2669,51 +2693,283 @@ TEST_F(ExternalSSTFileBasicTest, IngestWithTemperature) {
   }
 }
 
-TEST_F(ExternalSSTFileBasicTest, FailIfNotBottommostLevel) {
-  Options options = GetDefaultOptions();
+TEST_F(ExternalSSTFileBasicTest, FailIfNotBottommostLevelAndDisallowMemtable) {
+  for (bool disallow_memtable : {false, true}) {
+    Options options = GetDefaultOptions();
 
-  std::string file_path = sst_files_dir_ + std::to_string(1);
-  SstFileWriter sfw(EnvOptions(), options);
+    // First test with universal compaction
+    options.create_if_missing = true;
+    options.compaction_style = CompactionStyle::kCompactionStyleUniversal;
+    DestroyAndReopen(options);
 
-  ASSERT_OK(sfw.Open(file_path));
-  ASSERT_OK(sfw.Put("b", "dontcare"));
-  ASSERT_OK(sfw.Finish());
+    // And a CF potentially disallowing memtable write
+    options.disallow_memtable_writes = disallow_memtable;
+    CreateColumnFamilies({"cf0"}, options);
+    ASSERT_EQ(db_->GetOptions(handles_[0]).disallow_memtable_writes,
+              disallow_memtable);
 
-  // Test universal compaction + ingest with snapshot consistency
-  options.create_if_missing = true;
-  options.compaction_style = CompactionStyle::kCompactionStyleUniversal;
-  DestroyAndReopen(options);
-  {
-    const Snapshot* snapshot = db_->GetSnapshot();
-    ManagedSnapshot snapshot_guard(db_, snapshot);
-    IngestExternalFileOptions ifo;
-    ifo.fail_if_not_bottommost_level = true;
-    ifo.snapshot_consistency = true;
-    const Status s = db_->IngestExternalFile({file_path}, ifo);
-    ASSERT_TRUE(s.ok());
-  }
+    // Ingest with snapshot consistency
+    std::string file_path = sst_files_dir_ + std::to_string(1);
+    std::string file_path2 = sst_files_dir_ + std::to_string(2);
+    SstFileWriter sfw(EnvOptions(), options);
 
-  // Test level compaction
-  options.compaction_style = CompactionStyle::kCompactionStyleLevel;
-  options.num_levels = 2;
-  DestroyAndReopen(options);
-  ASSERT_OK(db_->Put(WriteOptions(), "a", "dontcare"));
-  ASSERT_OK(db_->Put(WriteOptions(), "c", "dontcare"));
-  ASSERT_OK(db_->Flush(FlushOptions()));
+    ASSERT_OK(sfw.Open(file_path));
+    ASSERT_OK(sfw.Put("b", "0"));
+    ASSERT_OK(sfw.Finish());
 
-  ASSERT_OK(db_->Put(WriteOptions(), "b", "dontcare"));
-  ASSERT_OK(db_->Put(WriteOptions(), "d", "dontcare"));
-  ASSERT_OK(db_->Flush(FlushOptions()));
+    {
+      const Snapshot* snapshot = db_->GetSnapshot();
+      ManagedSnapshot snapshot_guard(db_, snapshot);
+      IngestExternalFileOptions ifo;
+      ifo.fail_if_not_bottommost_level = true;
+      ifo.snapshot_consistency = true;
+      ASSERT_OK(db_->IngestExternalFile(handles_[0], {file_path}, ifo));
+    }
+    ASSERT_EQ(Get(0, "b"), "0");
 
-  {
-    CompactRangeOptions cro;
-    cro.bottommost_level_compaction = BottommostLevelCompaction::kForce;
-    ASSERT_OK(db_->CompactRange(cro, nullptr, nullptr));
+    // Test level compaction
+    options.compaction_style = CompactionStyle::kCompactionStyleLevel;
+    options.num_levels = 2;
+    CreateColumnFamilies({"cf1"}, options);
+    ASSERT_EQ(db_->GetOptions(handles_[1]).disallow_memtable_writes,
+              disallow_memtable);
 
-    IngestExternalFileOptions ifo;
-    ifo.fail_if_not_bottommost_level = true;
-    const Status s = db_->IngestExternalFile({file_path}, ifo);
-    ASSERT_TRUE(s.IsTryAgain());
+    if (!disallow_memtable) {
+      ASSERT_OK(Put(1, "a", "1"));
+      ASSERT_OK(Put(1, "c", "3"));
+      ASSERT_OK(Flush(1));
+
+      ASSERT_OK(Put(1, "b", "2"));
+      ASSERT_OK(Put(1, "d", "4"));
+      ASSERT_OK(Flush(1));
+    } else {
+      // Memtable write disallowed
+      EXPECT_EQ(Put(1, "a", "1").code(), Status::Code::kInvalidArgument);
+
+      // Use ingestion to get to the same state as above
+      ASSERT_OK(sfw.Open(file_path2));
+      ASSERT_OK(sfw.Put("a", "1"));
+      ASSERT_OK(sfw.Put("c", "3"));
+      ASSERT_OK(sfw.Finish());
+      ASSERT_OK(db_->IngestExternalFile(handles_[1], {file_path2}, {}));
+
+      ASSERT_OK(sfw.Open(file_path2));
+      ASSERT_OK(sfw.Put("b", "2"));
+      ASSERT_OK(sfw.Put("d", "4"));
+      ASSERT_OK(sfw.Finish());
+      ASSERT_OK(db_->IngestExternalFile(handles_[1], {file_path2}, {}));
+    }
+    ASSERT_EQ(Get(1, "a"), "1");
+    ASSERT_EQ(Get(1, "b"), "2");
+    ASSERT_EQ(Get(1, "c"), "3");
+    ASSERT_EQ(Get(1, "d"), "4");
+
+    {
+      // Test fail_if_not_bottommost_level, which fails if there's any overlap
+      // anywhere, even with snapshot_consistency=false
+      IngestExternalFileOptions ifo;
+      ASSERT_FALSE(ifo.fail_if_not_bottommost_level);
+      ifo.fail_if_not_bottommost_level = true;
+      ifo.snapshot_consistency = false;
+      // Fails with overlap on earlier level
+      Status s = db_->IngestExternalFile(handles_[1], {file_path}, ifo);
+      ASSERT_EQ(s.code(), Status::Code::kTryAgain);
+
+      CompactRangeOptions cro;
+      cro.bottommost_level_compaction = BottommostLevelCompaction::kForce;
+      ASSERT_OK(db_->CompactRange(cro, handles_[1], nullptr, nullptr));
+
+      // Fails with overlap on last level
+      s = db_->IngestExternalFile(handles_[1], {file_path}, ifo);
+      ASSERT_EQ(s.code(), Status::Code::kTryAgain);
+
+      // No change to data
+      ASSERT_EQ(Get(1, "a"), "1");
+      ASSERT_EQ(Get(1, "b"), "2");
+      ASSERT_EQ(Get(1, "c"), "3");
+      ASSERT_EQ(Get(1, "d"), "4");
+    }
+
+    if (!disallow_memtable) {
+      // Test allow_blocking_flush=false (fail because of memtable overlap)
+      IngestExternalFileOptions ifo;
+      ASSERT_TRUE(ifo.allow_blocking_flush);
+      ifo.allow_blocking_flush = false;
+      ASSERT_OK(Put(1, "b", "42"));
+      Status s = db_->IngestExternalFile(handles_[1], {file_path}, ifo);
+      ASSERT_EQ(s.code(), Status::Code::kInvalidArgument);
+
+      ASSERT_EQ(Get(1, "a"), "1");
+      ASSERT_EQ(Get(1, "b"), "42");
+      ASSERT_EQ(Get(1, "c"), "3");
+      ASSERT_EQ(Get(1, "d"), "4");
+
+      // Revert state
+      ASSERT_OK(Put(1, "b", "2"));
+      ASSERT_OK(Flush(1));
+    }
+
+    {
+      // Test atomic_replace_range
+      IngestExternalFileArg arg;
+      arg.column_family = handles_[1];
+      arg.external_files = {file_path};
+      arg.atomic_replace_range = {{"a", "zzz"}};
+
+      // start with some failure cases
+      // TODO: support snapshot consistency with tombstone file
+      ASSERT_TRUE(arg.options.snapshot_consistency);
+      Status s = db_->IngestExternalFiles({arg});
+      ASSERT_EQ(s.code(), Status::Code::kNotSupported);
+
+      ASSERT_EQ(Get(1, "a"), "1");
+      ASSERT_EQ(Get(1, "b"), "2");
+      ASSERT_EQ(Get(1, "c"), "3");
+      ASSERT_EQ(Get(1, "d"), "4");
+
+      arg.options.snapshot_consistency = false;
+      // Can usually be used with atomic_replace_range and
+      // snapshot_consistency=false, except it requires no input overlap
+      arg.options.fail_if_not_bottommost_level = true;
+
+      // one-sided ranges not yet supported
+      arg.atomic_replace_range = {{{}, "zzz"}};
+      s = db_->IngestExternalFiles({arg});
+      ASSERT_EQ(s.code(), Status::Code::kNotSupported);
+
+      arg.atomic_replace_range = {{"a", {}}};
+      s = db_->IngestExternalFiles({arg});
+      ASSERT_EQ(s.code(), Status::Code::kNotSupported);
+
+      // rejected because doesn't cover ingested file
+      arg.atomic_replace_range = {{"x", "z"}};
+      s = db_->IngestExternalFiles({arg});
+      ASSERT_EQ(s.code(), Status::Code::kInvalidArgument);
+
+      // rejected because of partial file overlap
+      arg.atomic_replace_range = {{"a", "c"}};
+      s = db_->IngestExternalFiles({arg});
+      ASSERT_EQ(s.code(), Status::Code::kInvalidArgument);
+
+      if (!disallow_memtable) {
+        // memtable overlap with replace range
+        ASSERT_OK(Put(1, "e", "5"));
+        arg.options.allow_blocking_flush = false;
+
+        // rejected because of memtable overlap
+        arg.atomic_replace_range = {{"a", "z"}};
+        s = db_->IngestExternalFiles({arg});
+        ASSERT_EQ(s.code(), Status::Code::kInvalidArgument);
+
+        // rejected because of memtable overlap
+        arg.atomic_replace_range = {{nullptr, nullptr}};
+        s = db_->IngestExternalFiles({arg});
+        ASSERT_EQ(s.code(), Status::Code::kInvalidArgument);
+
+        // FIXME: upper bound should be exclusive (DeleteRange semantics).
+        // currently rejected because of documented bug
+        arg.atomic_replace_range = {{"a", "e"}};
+        s = db_->IngestExternalFiles({arg});
+        ASSERT_EQ(s.code(), Status::Code::kInvalidArgument);
+
+        // work-around ensuring no memtable overlap
+        arg.atomic_replace_range = {{"a", "d2"}};
+        ASSERT_OK(db_->IngestExternalFiles({arg}));
+
+        ASSERT_EQ(Get(1, "e"), "5");
+      } else {
+        // rejected because of partial file overlap
+        arg.atomic_replace_range = {{"b", "z"}};
+        s = db_->IngestExternalFiles({arg});
+        ASSERT_EQ(s.code(), Status::Code::kInvalidArgument);
+
+        // no memtable complications
+        arg.atomic_replace_range = {{"a", "z"}};
+        ASSERT_OK(db_->IngestExternalFiles({arg}));
+
+        ASSERT_EQ(Get(1, "e"), "NOT_FOUND");
+      }
+      ASSERT_EQ(Get(1, "a"), "NOT_FOUND");
+      ASSERT_EQ(Get(1, "b"), "0");
+      ASSERT_EQ(Get(1, "c"), "NOT_FOUND");
+      ASSERT_EQ(Get(1, "d"), "NOT_FOUND");
+
+      // The single ingested file replaced everything (except perhaps memtable)
+      std::vector<LiveFileMetaData> live_files;
+      db_->GetLiveFilesMetaData(&live_files);
+      // One file in each CF
+      ASSERT_EQ(live_files.size(), 2);
+
+      ASSERT_OK(sfw.Open(file_path));
+      ASSERT_OK(sfw.Put("f", "6"));
+      ASSERT_OK(sfw.Finish());
+
+      // Another file
+      ASSERT_OK(sfw.Open(file_path2));
+      ASSERT_OK(sfw.Put("f", "7"));
+      ASSERT_OK(sfw.Put("g", "8"));
+      ASSERT_OK(sfw.Finish());
+
+      if (!disallow_memtable) {
+        // rejected because of memtable overlap with range
+        arg.atomic_replace_range = {{"e", "z"}};
+        s = db_->IngestExternalFiles({arg});
+        ASSERT_EQ(s.code(), Status::Code::kInvalidArgument);
+
+        // allow blocking flush of "e" (which is then replaced), and the file
+        // with just "b" is not replaced
+        arg.options.allow_blocking_flush = true;
+        ASSERT_OK(db_->IngestExternalFiles({arg}));
+
+        ASSERT_EQ(Get(1, "b"), "0");
+        ASSERT_EQ(Get(1, "e"), "NOT_FOUND");
+        ASSERT_EQ(Get(1, "f"), "6");
+        ASSERT_EQ(Get(1, "g"), "NOT_FOUND");
+
+        // memtable overlap with replace range
+        ASSERT_OK(Put(1, "e", "5"));
+        arg.options.allow_blocking_flush = false;
+        arg.external_files = {file_path2};
+
+        // rejected because of memtable overlap
+        arg.atomic_replace_range = {{nullptr, nullptr}};
+        s = db_->IngestExternalFiles({arg});
+        ASSERT_EQ(s.code(), Status::Code::kInvalidArgument);
+
+        // Replace everything, including with memtable flush
+        arg.options.allow_blocking_flush = true;
+        ASSERT_OK(db_->IngestExternalFiles({arg}));
+
+        ASSERT_EQ(Get(1, "b"), "NOT_FOUND");
+        ASSERT_EQ(Get(1, "e"), "NOT_FOUND");
+        ASSERT_EQ(Get(1, "f"), "7");
+        ASSERT_EQ(Get(1, "g"), "8");
+      } else {
+        arg.external_files = {file_path2, file_path};
+
+        // rejected because of overlap in files to ingest with fail_if_ = true
+        arg.atomic_replace_range = {{"e", "z"}};
+        s = db_->IngestExternalFiles({arg});
+        ASSERT_EQ(s.code(), Status::Code::kTryAgain);
+
+        arg.options.fail_if_not_bottommost_level = false;
+
+        // rejected because range doesn't cover ingested files
+        // FIXME: upper bound should be exclusive "g" instead
+        arg.atomic_replace_range = {{"e", "f2"}};
+        s = db_->IngestExternalFiles({arg});
+        ASSERT_EQ(s.code(), Status::Code::kInvalidArgument);
+
+        // Loaded into different levels, and the file with just "b" is not
+        // replaced
+        arg.atomic_replace_range = {{"e", "z"}};
+        ASSERT_OK(db_->IngestExternalFiles({arg}));
+
+        ASSERT_EQ(Get(1, "b"), "0");
+        ASSERT_EQ(Get(1, "f"), "6");  // earlier file listed later to ingest
+        ASSERT_EQ(Get(1, "g"), "8");
+      }
+    }
   }
 }
 

db/external_sst_file_ingestion_job.cc

@@ -29,6 +29,7 @@ Status ExternalSstFileIngestionJob::Prepare(
     const std::vector<std::string>& external_files_paths,
     const std::vector<std::string>& files_checksums,
     const std::vector<std::string>& files_checksum_func_names,
+    const std::optional<RangeOpt>& atomic_replace_range,
     const Temperature& file_temperature, uint64_t next_file_number,
     SuperVersion* sv) {
   Status status;
@@ -80,15 +81,47 @@ Status ExternalSstFileIngestionJob::Prepare(
     std::sort(sorted_files.begin(), sorted_files.end(), file_range_checker_);
 
     for (size_t i = 0; i + 1 < num_files; i++) {
-      if (file_range_checker_.OverlapsWithPrev(sorted_files[i],
-                                               sorted_files[i + 1],
-                                               /* ranges_sorted= */ true)) {
+      if (file_range_checker_.Overlaps(*sorted_files[i], *sorted_files[i + 1],
+                                       /* known_sorted= */ true)) {
         files_overlap_ = true;
         break;
       }
     }
   }
 
+  if (atomic_replace_range.has_value()) {
+    atomic_replace_range_.emplace();
+
+    if (atomic_replace_range->start && atomic_replace_range->limit) {
+      // User keys to internal keys (with timestamps)
+      const size_t ts_sz = ucmp_->timestamp_size();
+      std::string start_with_ts, limit_with_ts;
+      auto [start, limit] = MaybeAddTimestampsToRange(
+          atomic_replace_range->start, atomic_replace_range->limit, ts_sz,
+          &start_with_ts, &limit_with_ts);
+      assert(start.has_value());
+      assert(limit.has_value());
+      atomic_replace_range_->smallest_internal_key.Set(
+          *start, kMaxSequenceNumber, kValueTypeForSeek);
+      atomic_replace_range_->largest_internal_key.Set(
+          *limit, kMaxSequenceNumber, kValueTypeForSeek);
+      // Check files to ingest against replace range
+      for (size_t i = 0; i < num_files; i++) {
+        if (!file_range_checker_.Contains(*atomic_replace_range_,
+                                          files_to_ingest_[i])) {
+          return Status::InvalidArgument(
+              "Atomic replace range does not contain all files");
+        }
+      }
+    } else {
+      // Currently if either bound is not present, both must be
+      assert(atomic_replace_range->start.has_value() == false);
+      assert(atomic_replace_range->limit.has_value() == false);
+      assert(atomic_replace_range_->smallest_internal_key.unset());
+      assert(atomic_replace_range_->largest_internal_key.unset());
+    }
+  }
+
   if (ingestion_options_.ingest_behind && files_overlap_) {
     return Status::NotSupported(
         "Files with overlapping ranges cannot be ingested with ingestion "
@@ -359,9 +392,9 @@ void ExternalSstFileIngestionJob::DivideInputFilesIntoBatches() {
 
   file_batches_to_ingest_.emplace_back(/* _track_batch_range= */ true);
   for (auto& file : files_to_ingest_) {
-    if (file_range_checker_.OverlapsWithPrev(&file_batches_to_ingest_.back(),
-                                             &file,
-                                             /* ranges_sorted= */ false)) {
+    if (!file_batches_to_ingest_.back().unset() &&
+        file_range_checker_.Overlaps(file_batches_to_ingest_.back(), file,
+                                     /* known_sorted= */ false)) {
       file_batches_to_ingest_.emplace_back(/* _track_batch_range= */ true);
     }
     file_batches_to_ingest_.back().AddFile(&file, file_range_checker_);
@@ -370,14 +403,32 @@ void ExternalSstFileIngestionJob::DivideInputFilesIntoBatches() {
 
 Status ExternalSstFileIngestionJob::NeedsFlush(bool* flush_needed,
                                                SuperVersion* super_version) {
-  size_t n = files_to_ingest_.size();
-  autovector<UserKeyRange> ranges;
-  ranges.reserve(n);
-  for (const IngestedFileInfo& file_to_ingest : files_to_ingest_) {
-    ranges.emplace_back(file_to_ingest.start_ukey, file_to_ingest.limit_ukey);
+  Status status;
+  if (atomic_replace_range_.has_value() && atomic_replace_range_->unset()) {
+    // For replacing whole CF, we can simply check whether memtable is empty
+    *flush_needed = !super_version->mem->IsEmpty();
+  } else {
+    autovector<UserKeyRange> ranges;
+    if (atomic_replace_range_.has_value()) {
+      assert(!atomic_replace_range_->smallest_internal_key.unset());
+      assert(!atomic_replace_range_->largest_internal_key.unset());
+      // NOTE: we already checked in Prepare() that the atomic_replace_range
+      // covers all the files_to_ingest
+      // FIXME: need to make upper bound key exclusive (not easy here because
+      // the existing internal APIs deal in inclusive upper bound user keys)
+      ranges.emplace_back(
+          atomic_replace_range_->smallest_internal_key.user_key(),
+          atomic_replace_range_->largest_internal_key.user_key());
+    } else {
+      ranges.reserve(files_to_ingest_.size());
+      for (const IngestedFileInfo& file_to_ingest : files_to_ingest_) {
+        ranges.emplace_back(file_to_ingest.start_ukey,
+                            file_to_ingest.limit_ukey);
+      }
+    }
+    status = cfd_->RangesOverlapWithMemtables(
+        ranges, super_version, db_options_.allow_data_in_errors, flush_needed);
   }
-  Status status = cfd_->RangesOverlapWithMemtables(
-      ranges, super_version, db_options_.allow_data_in_errors, flush_needed);
   if (status.ok() && *flush_needed) {
     if (!ingestion_options_.allow_blocking_flush) {
       status = Status::InvalidArgument("External file requires flush");
@@ -430,8 +481,49 @@ Status ExternalSstFileIngestionJob::Run() {
   // the only active writer, and hence they are equal
   SequenceNumber last_seqno = versions_->LastSequence();
   edit_.SetColumnFamily(cfd_->GetID());
-  // The levels that the files will be ingested into
 
+  if (atomic_replace_range_.has_value()) {
+    auto* vstorage = super_version->current->storage_info();
+    if (atomic_replace_range_->unset()) {
+      if (cfd_->compaction_picker()->IsCompactionInProgress()) {
+        return Status::InvalidArgument(
+            "Atomic replace range (full) overlaps with pending compaction");
+      }
+      for (int lvl = 0; lvl < cfd_->NumberLevels(); lvl++) {
+        for (auto file : vstorage->LevelFiles(lvl)) {
+          // Set up to delete file to be replaced
+          edit_.DeleteFile(lvl, file->fd.GetNumber());
+        }
+      }
+    } else {
+      assert(!atomic_replace_range_->smallest_internal_key.unset());
+      assert(!atomic_replace_range_->largest_internal_key.unset());
+      for (int lvl = 0; lvl < cfd_->NumberLevels(); lvl++) {
+        if (cfd_->RangeOverlapWithCompaction(
+                atomic_replace_range_->smallest_internal_key.user_key(),
+                atomic_replace_range_->largest_internal_key.user_key(), lvl)) {
+          return Status::InvalidArgument(
+              "Atomic replace range overlaps with pending compaction");
+        }
+        for (auto file : vstorage->LevelFiles(lvl)) {
+          if (file_range_checker_.Overlaps(*atomic_replace_range_,
+                                           file->smallest, file->largest)) {
+            if (file_range_checker_.Contains(*atomic_replace_range_,
+                                             file->smallest, file->largest)) {
+              // Set up to delete file to be replaced
+              edit_.DeleteFile(lvl, file->fd.GetNumber());
+            } else {
+              // TODO: generate and ingest a tombstone file also
+              return Status::InvalidArgument(
+                  "Atomic replace range partially overlaps with existing file");
+            }
+          }
+        }
+      }
+    }
+  }
+
+  // Find levels to ingest into
   std::optional<int> prev_batch_uppermost_level;
   for (auto& batch : file_batches_to_ingest_) {
     int batch_uppermost_level = 0;
@@ -1104,6 +1196,10 @@ Status ExternalSstFileIngestionJob::AssignLevelAndSeqnoForIngestedFile(
     if (lvl > 0 && lvl < vstorage->base_level()) {
       continue;
     }
+    if (atomic_replace_range_.has_value()) {
+      target_level = lvl;
+      continue;
+    }
     if (cfd_->RangeOverlapWithCompaction(file_to_ingest->start_ukey,
                                          file_to_ingest->limit_ukey, lvl)) {
       // We must use L0 or any level higher than `lvl` to be able to overwrite
@@ -1172,6 +1268,8 @@ Status ExternalSstFileIngestionJob::AssignLevelAndSeqnoForIngestedFile(
 
 Status ExternalSstFileIngestionJob::CheckLevelForIngestedBehindFile(
     IngestedFileInfo* file_to_ingest) {
+  assert(!atomic_replace_range_.has_value());
+
   auto* vstorage = cfd_->current()->storage_info();
   // First, check if new files fit in the last level
   int last_lvl = cfd_->NumberLevels() - 1;

db/external_sst_file_ingestion_job.h

@@ -27,50 +27,77 @@ class SystemClock;
 
 struct KeyRangeInfo {
   // Smallest internal key in an external file or for a batch of external files.
+  // unset() could be either invalid or "before all keys"
   InternalKey smallest_internal_key;
   // Largest internal key in an external file or for a batch of external files.
+  // unset() could be either invalid or "after all keys"
   InternalKey largest_internal_key;
 
-  bool empty() const {
-    return smallest_internal_key.size() == 0 &&
-           largest_internal_key.size() == 0;
+  bool unset() const {
+    // Legal internal keys are at least 8 bytes.
+    return smallest_internal_key.unset() || largest_internal_key.unset();
   }
 };
 
 // Helper class to apply SST file key range checks to the external files.
+// XXX: using sstableKeyCompare with user comparator on internal keys is
+// very broken
 class ExternalFileRangeChecker {
  public:
   explicit ExternalFileRangeChecker(const Comparator* ucmp) : ucmp_(ucmp) {}
 
   // Operator used for sorting ranges.
-  bool operator()(const KeyRangeInfo* prev_range,
-                  const KeyRangeInfo* range) const {
-    assert(prev_range);
-    assert(range);
-    return sstableKeyCompare(ucmp_, prev_range->smallest_internal_key,
-                             range->smallest_internal_key) < 0;
+  bool operator()(const KeyRangeInfo* range1,
+                  const KeyRangeInfo* range2) const {
+    assert(range1);
+    assert(range2);
+    assert(!range1->unset());
+    assert(!range2->unset());
+    return sstableKeyCompare(ucmp_, range1->smallest_internal_key,
+                             range2->smallest_internal_key) < 0;
   }
 
-  // Check whether `range` overlaps with `prev_range`. `ranges_sorted` can be
-  // set to true when the inputs are already sorted based on the sorting logic
-  // provided by this checker's operator(), which can help simplify the check.
-  bool OverlapsWithPrev(const KeyRangeInfo* prev_range,
-                        const KeyRangeInfo* range,
-                        bool ranges_sorted = false) const {
-    assert(prev_range);
-    assert(range);
-    if (prev_range->empty() || range->empty()) {
+  bool Overlaps(const KeyRangeInfo& range1, const KeyRangeInfo& range2,
+                bool known_sorted = false) const {
+    return Overlaps(range1, range2.smallest_internal_key,
+                    range2.largest_internal_key, known_sorted);
+  }
+  bool Overlaps(const KeyRangeInfo& range1, const InternalKey& range2_smallest,
+                const InternalKey& range2_largest,
+                bool known_sorted = false) const {
+    bool any_unset =
+        range1.unset() || range2_smallest.unset() || range2_largest.unset();
+    if (any_unset) {
+      assert(!any_unset);
       return false;
     }
-    if (ranges_sorted) {
-      return sstableKeyCompare(ucmp_, prev_range->largest_internal_key,
-                               range->smallest_internal_key) >= 0;
+    if (known_sorted) {
+      return sstableKeyCompare(ucmp_, range1.largest_internal_key,
+                               range2_smallest) >= 0;
     }
 
-    return sstableKeyCompare(ucmp_, prev_range->largest_internal_key,
-                             range->smallest_internal_key) >= 0 &&
-           sstableKeyCompare(ucmp_, prev_range->smallest_internal_key,
-                             range->largest_internal_key) <= 0;
+    return sstableKeyCompare(ucmp_, range1.largest_internal_key,
+                             range2_smallest) >= 0 &&
+           sstableKeyCompare(ucmp_, range1.smallest_internal_key,
+                             range2_largest) <= 0;
+  }
+
+  bool Contains(const KeyRangeInfo& range1, const KeyRangeInfo& range2) {
+    return Contains(range1, range2.smallest_internal_key,
+                    range2.largest_internal_key);
+  }
+  bool Contains(const KeyRangeInfo& range1, const InternalKey& range2_smallest,
+                const InternalKey& range2_largest) {
+    bool any_unset =
+        range1.unset() || range2_smallest.unset() || range2_largest.unset();
+    if (any_unset) {
+      assert(!any_unset);
+      return false;
+    }
+    return sstableKeyCompare(ucmp_, range1.smallest_internal_key,
+                             range2_smallest) <= 0 &&
+           sstableKeyCompare(ucmp_, range1.largest_internal_key,
+                             range2_largest) >= 0;
   }
 
   void MaybeUpdateRange(const InternalKey& start_key,
@@ -218,6 +245,7 @@ class ExternalSstFileIngestionJob {
   Status Prepare(const std::vector<std::string>& external_files_paths,
                  const std::vector<std::string>& files_checksums,
                  const std::vector<std::string>& files_checksum_func_names,
+                 const std::optional<RangeOpt>& atomic_replace_range,
                  const Temperature& file_temperature, uint64_t next_file_number,
                  SuperVersion* sv);
 
@@ -362,6 +390,7 @@ class ExternalSstFileIngestionJob {
   autovector<IngestedFileInfo> files_to_ingest_;
   std::vector<FileBatchInfo> file_batches_to_ingest_;
   const IngestExternalFileOptions& ingestion_options_;
+  std::optional<KeyRangeInfo> atomic_replace_range_;
   Directories* directories_;
   EventLogger* event_logger_;
   VersionEdit edit_;

db/external_sst_file_test.cc

@@ -3915,9 +3915,7 @@ TEST_P(IngestDBGeneratedFileTest, FailureCase) {
     s = db_->IngestExternalFile(to_ingest_files, ingest_opts);
     ASSERT_TRUE(s.ToString().find(err) != std::string::npos);
     ASSERT_NOK(s);
-    if (options.compaction_style != kCompactionStyleUniversal) {
-      // FIXME: after fixing ingestion with universal compaction, currently
-      //  will always ingest into L0.
+    if (options.num_levels > 1) {
       ingest_opts.fail_if_not_bottommost_level = true;
       s = db_->IngestExternalFile(to_ingest_files, ingest_opts);
       ASSERT_NOK(s);

db/flush_job.cc

@@ -870,7 +870,8 @@ Status FlushJob::WriteLevel0Table() {
   std::vector<BlobFileAddition> blob_file_additions;
 
   {
-    auto write_hint = base_->storage_info()->CalculateSSTWriteHint(/*level=*/0);
+    auto write_hint = base_->storage_info()->CalculateSSTWriteHint(
+        /*level=*/0, db_options_.calculate_sst_write_lifetime_hint_set);
     Env::IOPriority io_priority = GetRateLimiterPriority();
     db_mutex_->Unlock();
     if (log_buffer_) {
@@ -1193,13 +1194,12 @@ void FlushJob::GetEffectiveCutoffUDTForPickedMemTables() {
 }
 
 void FlushJob::GetPrecludeLastLevelMinSeqno() {
-  if (mutable_cf_options_.preclude_last_level_data_seconds == 0 ||
-      // FIXME: create FlushJob and build SuperVersions such that
-      // preclude_last_level_data_seconds > 0 implies
-      // seqno_to_time_mapping_ != nullptr
-      seqno_to_time_mapping_ == nullptr) {
+  if (mutable_cf_options_.preclude_last_level_data_seconds == 0) {
     return;
   }
+  // SuperVersion should guarantee this
+  assert(seqno_to_time_mapping_);
+  assert(!seqno_to_time_mapping_->Empty());
   int64_t current_time = 0;
   Status s = db_options_.clock->GetCurrentTime(&current_time);
   if (!s.ok()) {

db/flush_job.h

@@ -234,7 +234,7 @@ class FlushJob {
 
   // The current minimum seqno that compaction jobs will preclude the data from
   // the last level. Data with seqnos larger than this or larger than
-  // `earliest_snapshot_` will be output to the penultimate level had it gone
+  // `earliest_snapshot_` will be output to the proximal level had it gone
   // through a compaction to the last level.
   SequenceNumber preclude_last_level_min_seqno_ = kMaxSequenceNumber;
 };

db/internal_stats.h

@@ -153,23 +153,6 @@ class InternalStats {
 
   InternalStats(int num_levels, SystemClock* clock, ColumnFamilyData* cfd);
 
-  // Per level compaction stats
-  struct CompactionOutputsStats {
-    uint64_t num_output_records = 0;
-    uint64_t bytes_written = 0;
-    uint64_t bytes_written_blob = 0;
-    uint64_t num_output_files = 0;
-    uint64_t num_output_files_blob = 0;
-
-    void Add(const CompactionOutputsStats& stats) {
-      this->num_output_records += stats.num_output_records;
-      this->bytes_written += stats.bytes_written;
-      this->bytes_written_blob += stats.bytes_written_blob;
-      this->num_output_files += stats.num_output_files;
-      this->num_output_files_blob += stats.num_output_files_blob;
-    }
-  };
-
   // Per level compaction stats.  comp_stats_[level] stores the stats for
   // compactions that produced data for the specified "level".
   struct CompactionStats {
@@ -420,15 +403,6 @@ class InternalStats {
       }
     }
 
-    void Add(const CompactionOutputsStats& stats) {
-      this->num_output_files += static_cast<int>(stats.num_output_files);
-      this->num_output_records += stats.num_output_records;
-      this->bytes_written += stats.bytes_written;
-      this->bytes_written_blob += stats.bytes_written_blob;
-      this->num_output_files_blob +=
-          static_cast<int>(stats.num_output_files_blob);
-    }
-
     void Subtract(const CompactionStats& c) {
       this->micros -= c.micros;
       this->cpu_micros -= c.cpu_micros;
@@ -473,49 +447,51 @@ class InternalStats {
     }
   };
 
-  // Compaction stats, for per_key_placement compaction, it includes 2 levels
-  // stats: the last level and the penultimate level.
+  // Compaction internal stats, for per_key_placement compaction, it includes 2
+  // output level stats: the last level and the proximal level.
   struct CompactionStatsFull {
     // the stats for the target primary output level
-    CompactionStats stats;
+    CompactionStats output_level_stats;
 
-    // stats for penultimate level output if exist
-    bool has_penultimate_level_output = false;
-    CompactionStats penultimate_level_stats;
+    // stats for proximal level output if exist
+    bool has_proximal_level_output = false;
+    CompactionStats proximal_level_stats;
 
-    explicit CompactionStatsFull() : stats(), penultimate_level_stats() {}
+    explicit CompactionStatsFull()
+        : output_level_stats(), proximal_level_stats() {}
 
     explicit CompactionStatsFull(CompactionReason reason, int c)
-        : stats(reason, c), penultimate_level_stats(reason, c) {}
+        : output_level_stats(reason, c), proximal_level_stats(reason, c) {}
 
     uint64_t TotalBytesWritten() const {
-      uint64_t bytes_written = stats.bytes_written + stats.bytes_written_blob;
-      if (has_penultimate_level_output) {
-        bytes_written += penultimate_level_stats.bytes_written +
-                         penultimate_level_stats.bytes_written_blob;
+      uint64_t bytes_written = output_level_stats.bytes_written +
+                               output_level_stats.bytes_written_blob;
+      if (has_proximal_level_output) {
+        bytes_written += proximal_level_stats.bytes_written +
+                         proximal_level_stats.bytes_written_blob;
       }
       return bytes_written;
     }
 
     uint64_t DroppedRecords() {
-      uint64_t output_records = stats.num_output_records;
-      if (has_penultimate_level_output) {
-        output_records += penultimate_level_stats.num_output_records;
+      uint64_t output_records = output_level_stats.num_output_records;
+      if (has_proximal_level_output) {
+        output_records += proximal_level_stats.num_output_records;
       }
-      if (stats.num_input_records > output_records) {
-        return stats.num_input_records - output_records;
+      if (output_level_stats.num_input_records > output_records) {
+        return output_level_stats.num_input_records - output_records;
       }
       return 0;
     }
 
     void SetMicros(uint64_t val) {
-      stats.micros = val;
-      penultimate_level_stats.micros = val;
+      output_level_stats.micros = val;
+      proximal_level_stats.micros = val;
     }
 
     void AddCpuMicros(uint64_t val) {
-      stats.cpu_micros += val;
-      penultimate_level_stats.cpu_micros += val;
+      output_level_stats.cpu_micros += val;
+      proximal_level_stats.cpu_micros += val;
     }
   };
 
@@ -587,10 +563,9 @@ class InternalStats {
 
   void AddCompactionStats(int level, Env::Priority thread_pri,
                           const CompactionStatsFull& comp_stats_full) {
-    AddCompactionStats(level, thread_pri, comp_stats_full.stats);
-    if (comp_stats_full.has_penultimate_level_output) {
-      per_key_placement_comp_stats_.Add(
-          comp_stats_full.penultimate_level_stats);
+    AddCompactionStats(level, thread_pri, comp_stats_full.output_level_stats);
+    if (comp_stats_full.has_proximal_level_output) {
+      per_key_placement_comp_stats_.Add(comp_stats_full.proximal_level_stats);
     }
   }
 

db/job_context.h

@@ -22,6 +22,9 @@ namespace ROCKSDB_NAMESPACE {
 class MemTable;
 struct SuperVersion;
 
+// The purpose of this struct is to simplify pushing work such as
+// allocation/construction, de-allocation/destruction, and notifications to
+// outside of holding the DB mutex.
 struct SuperVersionContext {
   struct WriteStallNotification {
     WriteStallInfo write_stall_info;
@@ -35,12 +38,6 @@ struct SuperVersionContext {
   std::unique_ptr<SuperVersion>
       new_superversion;  // if nullptr no new superversion
 
-  // If not nullptr, a new seqno to time mapping is available to be installed.
-  // Otherwise, make a shared copy of the one in the existing SuperVersion and
-  // carry it over to the new SuperVersion. This is moved to the SuperVersion
-  // during installation.
-  std::shared_ptr<const SeqnoToTimeMapping> new_seqno_to_time_mapping{nullptr};
-
   explicit SuperVersionContext(bool create_superversion = false)
       : new_superversion(create_superversion ? new SuperVersion() : nullptr) {}
 

db/plain_table_db_test.cc

@@ -1338,6 +1338,50 @@ TEST_P(PlainTableDBTest, AdaptiveTable) {
 
 INSTANTIATE_TEST_CASE_P(PlainTableDBTest, PlainTableDBTest, ::testing::Bool());
 
+TEST_P(PlainTableDBTest, DeleteRangeNotSupported) {
+  // XXX: After attempting DeleteRange with PlainTable, Writes will permanently
+  // fail. Even if re-opening the DB, if WAL is used, the WAL is not recoverable
+  // (without manual intervention). Furthermore, a partial write batch can
+  // be exposed to readers, breaking WriteBatch atomicity.
+  for (bool use_write_batch : {/*false, */ true}) {
+    DestroyAndReopen();
+
+    ASSERT_OK(Put("a0001111", "1"));
+    ASSERT_OK(Put("b0001111", "2"));
+    ASSERT_OK(Put("c0001111", "3"));
+    if (use_write_batch) {
+      WriteBatch wb;
+      ASSERT_OK(wb.Put("d0001111", "4"));
+      ASSERT_OK(wb.DeleteRange("a", "b"));
+      ASSERT_OK(wb.Put("e0001111", "5"));
+      ASSERT_EQ(dbfull()->Write({}, &wb).code(), Status::Code::kNotSupported);
+    } else {
+      ASSERT_EQ(dbfull()->DeleteRange({}, "az", "bz").code(),
+                Status::Code::kNotSupported);
+    }
+    ASSERT_EQ(Get("a0001111"), "1");
+    ASSERT_EQ(Get("b0001111"), "2");
+    ASSERT_EQ(Get("c0001111"), "3");
+    if (use_write_batch) {
+      // XXX: broken WriteBatch atomicity
+      ASSERT_EQ(Get("d0001111"), "4");
+    } else {
+      ASSERT_EQ(Get("d0001111"), "NOT_FOUND");
+    }
+    ASSERT_EQ(Get("e0001111"), "NOT_FOUND");
+
+    ASSERT_EQ(Put("e0001111", "5").code(), Status::Code::kNotSupported);
+    ASSERT_EQ(Get("e0001111"), "NOT_FOUND");
+
+    // Even trying to flush
+    ASSERT_EQ(dbfull()->TEST_FlushMemTable().code(),
+              Status::Code::kNotSupported);
+
+    // XXX: WAL is not recoverable
+    ASSERT_EQ(TryReopen().code(), Status::Code::kNotSupported);
+  }
+}
+
 }  // namespace ROCKSDB_NAMESPACE
 
 int main(int argc, char** argv) {

db/repair.cc

@@ -100,13 +100,15 @@ class Repairer {
         db_options_(SanitizeOptions(dbname_, db_options)),
         immutable_db_options_(ImmutableDBOptions(db_options_)),
         icmp_(default_cf_opts.comparator),
-        default_cf_opts_(
-            SanitizeOptions(immutable_db_options_, default_cf_opts)),
+        default_cf_opts_(SanitizeCfOptions(immutable_db_options_,
+                                           /*read_only*/ false,
+                                           default_cf_opts)),
         default_iopts_(
             ImmutableOptions(immutable_db_options_, default_cf_opts_)),
         default_mopts_(MutableCFOptions(default_cf_opts_)),
-        unknown_cf_opts_(
-            SanitizeOptions(immutable_db_options_, unknown_cf_opts)),
+        unknown_cf_opts_(SanitizeCfOptions(immutable_db_options_,
+                                           /*read_only*/ false,
+                                           unknown_cf_opts)),
         create_unknown_cfs_(create_unknown_cfs),
         raw_table_cache_(
             // TableCache can be small since we expect each table to be opened
@@ -456,8 +458,9 @@ class Repairer {
       meta.file_creation_time = current_time;
       SnapshotChecker* snapshot_checker = DisableGCSnapshotChecker::Instance();
 
-      auto write_hint =
-          cfd->current()->storage_info()->CalculateSSTWriteHint(/*level=*/0);
+      auto write_hint = cfd->current()->storage_info()->CalculateSSTWriteHint(
+          /*level=*/0, db_options_.calculate_sst_write_lifetime_hint_set);
+
       std::vector<std::unique_ptr<FragmentedRangeTombstoneIterator>>
           range_del_iters;
       auto range_del_iter = mem->NewRangeTombstoneIterator(

db/seqno_time_test.cc

@@ -96,7 +96,7 @@ TEST_F(SeqnoTimeTest, TemperatureBasicUniversal) {
   }
   ASSERT_OK(dbfull()->TEST_WaitForCompact());
 
-  // All data is hot, only output to penultimate level
+  // All data is hot, only output to proximal level
   ASSERT_EQ("0,0,0,0,0,1", FilesPerLevel());
   ASSERT_GT(GetSstSizeHelper(Temperature::kUnknown), 0);
   ASSERT_EQ(GetSstSizeHelper(Temperature::kCold), 0);
@@ -185,7 +185,7 @@ TEST_F(SeqnoTimeTest, TemperatureBasicLevel) {
   options.num_levels = kNumLevels;
   options.level_compaction_dynamic_level_bytes = true;
   // TODO(zjay): for level compaction, auto-compaction may stuck in deadloop, if
-  //  the penultimate level score > 1, but the hot is not cold enough to compact
+  //  the proximal level score > 1, but the hot is not cold enough to compact
   //  to last level, which will keep triggering compaction.
   options.disable_auto_compactions = true;
   DestroyAndReopen(options);
@@ -205,7 +205,7 @@ TEST_F(SeqnoTimeTest, TemperatureBasicLevel) {
   cro.bottommost_level_compaction = BottommostLevelCompaction::kForce;
   ASSERT_OK(db_->CompactRange(cro, nullptr, nullptr));
 
-  // All data is hot, only output to penultimate level
+  // All data is hot, only output to proximal level
   ASSERT_EQ("0,0,0,0,0,1", FilesPerLevel());
   ASSERT_GT(GetSstSizeHelper(Temperature::kUnknown), 0);
   ASSERT_EQ(GetSstSizeHelper(Temperature::kCold), 0);
@@ -753,7 +753,7 @@ TEST_P(SeqnoTimeTablePropTest, SeqnoToTimeMappingUniversal) {
   CompactRangeOptions cro;
   cro.bottommost_level_compaction = BottommostLevelCompaction::kForce;
   ASSERT_OK(db_->CompactRange(cro, nullptr, nullptr));
-  // make sure the data is all compacted to penultimate level if the feature is
+  // make sure the data is all compacted to proximal level if the feature is
   // on, otherwise, compacted to the last level.
   if (options.preclude_last_level_data_seconds > 0) {
     ASSERT_GT(NumTableFilesAtLevel(5), 0);
@@ -792,8 +792,7 @@ TEST_P(SeqnoTimeTablePropTest, SeqnoToTimeMappingUniversal) {
   }
   ASSERT_GT(num_seqno_zeroing, 0);
   std::vector<KeyVersion> key_versions;
-  ASSERT_OK(GetAllKeyVersions(db_, Slice(), Slice(),
-                              std::numeric_limits<size_t>::max(),
+  ASSERT_OK(GetAllKeyVersions(db_, {}, {}, std::numeric_limits<size_t>::max(),
                               &key_versions));
   // make sure there're more than 300 keys and first 100 keys are having seqno
   // zeroed out, the last 100 key seqno not zeroed out
@@ -919,10 +918,11 @@ TEST_P(SeqnoTimeTablePropTest, PrePopulateInDB) {
       ASSERT_EQ(db_->GetLatestSequenceNumber(), 0);
 
       // And even if we re-open read-write, we do not get pre-population,
-      // because that's only for new DBs.
+      // because that's only for new DBs. We just get a single bootstrap
+      // entry as a lower bound on write times of future writes.
       Reopen(track_options);
       sttm = dbfull()->TEST_GetSeqnoToTimeMapping();
-      ASSERT_EQ(sttm.Size(), 0);
+      ASSERT_EQ(sttm.Size(), 1);
       ASSERT_EQ(db_->GetLatestSequenceNumber(), 0);
     }
   }

db/seqno_to_time_mapping.cc

@@ -490,7 +490,7 @@ bool SeqnoToTimeMapping::Append(SequenceNumber seqno, uint64_t time) {
   return added;
 }
 
-bool SeqnoToTimeMapping::PrePopulate(SequenceNumber from_seqno,
+void SeqnoToTimeMapping::PrePopulate(SequenceNumber from_seqno,
                                      SequenceNumber to_seqno,
                                      uint64_t from_time, uint64_t to_time) {
   assert(Empty());
@@ -505,8 +505,6 @@ bool SeqnoToTimeMapping::PrePopulate(SequenceNumber from_seqno,
                                  (to_seqno - from_seqno);
     pairs_.emplace_back(i, t);
   }
-
-  return /*success*/ true;
 }
 
 std::string SeqnoToTimeMapping::ToHumanString() const {

db/seqno_to_time_mapping.h

@@ -138,7 +138,7 @@ class SeqnoToTimeMapping {
   // Adds a series of mappings interpolating from from_seqno->from_time to
   // to_seqno->to_time. This can only be called on an empty object and both
   // seqno range and time range are inclusive.
-  bool PrePopulate(SequenceNumber from_seqno, SequenceNumber to_seqno,
+  void PrePopulate(SequenceNumber from_seqno, SequenceNumber to_seqno,
                    uint64_t from_time, uint64_t to_time);
 
   // Append a new entry to the list. The `seqno` should be >= all previous
@@ -148,6 +148,10 @@ class SeqnoToTimeMapping {
   // rather than creating a new entry.
   bool Append(SequenceNumber seqno, uint64_t time);
 
+  bool Append(std::pair<SequenceNumber, uint64_t> seqno_time_pair) {
+    return Append(seqno_time_pair.first, seqno_time_pair.second);
+  }
+
   // Clear all entries and (re-)enter enforced mode if not already in that
   // state. Enforced limits are unchanged.
   void Clear() {
@@ -274,6 +278,48 @@ class SeqnoToTimeMapping {
   pair_const_iterator FindGreaterEqSeqno(SequenceNumber seqno) const;
 };
 
+// A struct to help combining settings across column families
+struct MinAndMaxPreserveSeconds {
+  uint64_t min_preserve_seconds = std::numeric_limits<uint64_t>::max();
+  uint64_t max_preserve_seconds = std::numeric_limits<uint64_t>::min();
+
+  MinAndMaxPreserveSeconds() = default;
+
+  template <class CFOpts>
+  explicit MinAndMaxPreserveSeconds(const CFOpts& opts) {
+    Combine(opts);
+  }
+
+  bool IsEnabled() const {
+    return min_preserve_seconds != std::numeric_limits<uint64_t>::max();
+  }
+
+  // Incorporate another CF's settings into the result. If preserve/preclude are
+  // disabled for this CF, they are excluded from the result.
+  template <class CFOpts>
+  void Combine(const CFOpts& opts) {
+    uint64_t preserve_seconds = std::max(opts.preserve_internal_time_seconds,
+                                         opts.preclude_last_level_data_seconds);
+    if (preserve_seconds > 0) {
+      min_preserve_seconds = std::min(preserve_seconds, min_preserve_seconds);
+      max_preserve_seconds = std::max(preserve_seconds, max_preserve_seconds);
+    }
+  }
+
+  // Choose how many seconds between mapping samples
+  uint64_t GetRecodingCadence() const {
+    if (IsEnabled()) {
+      // round up to 1 when the time_duration is smaller than
+      // kMaxSeqnoTimePairsPerCF
+      return (min_preserve_seconds + kMaxSeqnoTimePairsPerCF - 1) /
+             kMaxSeqnoTimePairsPerCF;
+    } else {
+      // disabled
+      return 0;
+    }
+  }
+};
+
 // === Utility methods used for TimedPut === //
 
 // Pack a value Slice and a unix write time into buffer `buf` and return a Slice

db/table_cache.cc

@@ -205,6 +205,7 @@ Status TableCache::FindTable(
       RecordTick(ioptions_.stats, NO_FILE_ERRORS);
       // We do not cache error results so that if the error is transient,
       // or somebody repairs the file, we recover automatically.
+      IGNORE_STATUS_IF_ERROR(s);
     } else {
       s = cache_.Insert(key, table_reader.get(), 1, handle);
       if (s.ok()) {

db/version_edit_handler.cc

@@ -408,7 +408,7 @@ void VersionEditHandler::CheckIterationResult(const log::Reader& reader,
       if (cfd->IsDropped()) {
         continue;
       }
-      if (read_only_) {
+      if (version_set_->unchanging()) {
         cfd->table_cache()->SetTablesAreImmortal();
       }
       *s = LoadTables(cfd, /*prefetch_index_and_filter_in_cache=*/false,
@@ -471,8 +471,8 @@ void VersionEditHandler::CheckIterationResult(const log::Reader& reader,
 ColumnFamilyData* VersionEditHandler::CreateCfAndInit(
     const ColumnFamilyOptions& cf_options, const VersionEdit& edit) {
   uint32_t cf_id = edit.GetColumnFamily();
-  ColumnFamilyData* cfd =
-      version_set_->CreateColumnFamily(cf_options, read_options_, &edit);
+  ColumnFamilyData* cfd = version_set_->CreateColumnFamily(
+      cf_options, read_options_, &edit, read_only_);
   assert(cfd != nullptr);
   cfd->set_initialized();
   assert(builders_.find(cf_id) == builders_.end());

db/version_edit_handler.h

@@ -198,7 +198,9 @@ class VersionEditHandler : public VersionEditHandlerBase {
                             bool prefetch_index_and_filter_in_cache,
                             bool is_initial_load);
 
-  virtual bool MustOpenAllColumnFamilies() const { return !read_only_; }
+  virtual bool MustOpenAllColumnFamilies() const {
+    return !version_set_->unchanging();
+  }
 
   const bool read_only_;
   std::vector<ColumnFamilyDescriptor> column_families_;
@@ -334,10 +336,10 @@ class ManifestTailer : public VersionEditHandlerPointInTime {
                           const ReadOptions& read_options,
                           EpochNumberRequirement epoch_number_requirement =
                               EpochNumberRequirement::kMustPresent)
-      : VersionEditHandlerPointInTime(/*read_only=*/false, column_families,
-                                      version_set, io_tracer, read_options,
-                                      /*allow_incomplete_valid_version=*/false,
-                                      epoch_number_requirement),
+      : VersionEditHandlerPointInTime(
+            /*read_only=*/true, column_families, version_set, io_tracer,
+            read_options,
+            /*allow_incomplete_valid_version=*/false, epoch_number_requirement),
         mode_(Mode::kRecovery) {}
 
   Status VerifyFile(ColumnFamilyData* cfd, const std::string& fpath, int level,

db/version_set.cc

@@ -1627,8 +1627,8 @@ Status Version::GetTableProperties(const ReadOptions& read_options,
   return s;
 }
 
-Status Version::GetPropertiesOfAllTables(const ReadOptions& read_options,
-                                         TablePropertiesCollection* props) {
+Status Version::GetPropertiesOfAllTables(
+    const ReadOptions& read_options, TablePropertiesCollection* props) const {
   Status s;
   for (int level = 0; level < storage_info_.num_levels_; level++) {
     s = GetPropertiesOfAllTables(read_options, props, level);
@@ -1699,7 +1699,7 @@ Status Version::TablesRangeTombstoneSummary(int max_entries_to_print,
 
 Status Version::GetPropertiesOfAllTables(const ReadOptions& read_options,
                                          TablePropertiesCollection* props,
-                                         int level) {
+                                         int level) const {
   for (const auto& file_meta : storage_info_.files_[level]) {
     auto fname =
         TableFileName(cfd_->ioptions().cf_paths, file_meta->fd.GetNumber(),
@@ -1753,6 +1753,24 @@ Status Version::GetPropertiesOfTablesInRange(
   return Status::OK();
 }
 
+Status Version::GetPropertiesOfTablesByLevel(
+    const ReadOptions& read_options,
+    std::vector<std::unique_ptr<TablePropertiesCollection>>* props_by_level)
+    const {
+  Status s;
+
+  props_by_level->reserve(storage_info_.num_levels_);
+  for (int level = 0; level < storage_info_.num_levels_; level++) {
+    props_by_level->push_back(std::make_unique<TablePropertiesCollection>());
+    s = GetPropertiesOfAllTables(read_options, props_by_level->back().get(),
+                                 level);
+    if (!s.ok()) {
+      return s;
+    }
+  }
+  return Status::OK();
+}
+
 Status Version::GetAggregatedTableProperties(
     const ReadOptions& read_options, std::shared_ptr<const TableProperties>* tp,
     int level) {
@@ -4761,7 +4779,7 @@ void VersionStorageInfo::CalculateBaseBytes(const ImmutableOptions& ioptions,
             cur_level_size <= base_bytes_min &&
             (options.preclude_last_level_data_seconds == 0 ||
              i < num_levels_ - 2)) {
-          // When per_key_placement is enabled, the penultimate level is
+          // When per_key_placement is enabled, the proximal level is
           // necessary.
           lowest_unnecessary_level_ = i;
         }
@@ -4903,24 +4921,38 @@ bool VersionStorageInfo::RangeMightExistAfterSortedRun(
 }
 
 Env::WriteLifeTimeHint VersionStorageInfo::CalculateSSTWriteHint(
-    int level) const {
-  if (compaction_style_ != kCompactionStyleLevel) {
+    int level, CompactionStyleSet compaction_style_set) const {
+  if (!compaction_style_set.Contains(compaction_style_)) {
     return Env::WLTH_NOT_SET;
   }
-  if (level == 0) {
-    return Env::WLTH_MEDIUM;
-  }
 
-  // L1: medium, L2: long, ...
-  if (level - base_level_ >= 2) {
-    return Env::WLTH_EXTREME;
-  } else if (level < base_level_) {
-    // There is no restriction which prevents level passed in to be smaller
-    // than base_level.
-    return Env::WLTH_MEDIUM;
+  switch (compaction_style_) {
+    case kCompactionStyleLevel:
+      if (level == 0) {
+        return Env::WLTH_MEDIUM;
+      }
+
+      // L1: medium, L2: long, ...
+      if (level - base_level_ >= 2) {
+        return Env::WLTH_EXTREME;
+      } else if (level < base_level_) {
+        // There is no restriction which prevents level passed in to be smaller
+        // than base_level.
+        return Env::WLTH_MEDIUM;
+      }
+      return static_cast<Env::WriteLifeTimeHint>(
+          level - base_level_ + static_cast<int>(Env::WLTH_MEDIUM));
+    case kCompactionStyleUniversal:
+      if (level == 0) {
+        return Env::WLTH_SHORT;
+      }
+      if (level == 1) {
+        return Env::WLTH_MEDIUM;
+      }
+      return Env::WLTH_LONG;
+    default:
+      return Env::WLTH_NOT_SET;
   }
-  return static_cast<Env::WriteLifeTimeHint>(
-      level - base_level_ + static_cast<int>(Env::WLTH_MEDIUM));
 }
 
 void Version::AddLiveFiles(std::vector<uint64_t>* live_table_files,
@@ -5114,7 +5146,7 @@ VersionSet::VersionSet(
     BlockCacheTracer* const block_cache_tracer,
     const std::shared_ptr<IOTracer>& io_tracer, const std::string& db_id,
     const std::string& db_session_id, const std::string& daily_offpeak_time_utc,
-    ErrorHandler* const error_handler, const bool read_only)
+    ErrorHandler* error_handler, bool unchanging)
     : column_family_set_(new ColumnFamilySet(
           dbname, _db_options, storage_options, table_cache,
           write_buffer_manager, write_controller, block_cache_tracer, io_tracer,
@@ -5143,12 +5175,12 @@ VersionSet::VersionSet(
       db_session_id_(db_session_id),
       offpeak_time_option_(OffpeakTimeOption(daily_offpeak_time_utc)),
       error_handler_(error_handler),
-      read_only_(read_only),
+      unchanging_(unchanging),
       closed_(false) {}
 
 Status VersionSet::Close(FSDirectory* db_dir, InstrumentedMutex* mu) {
   Status s;
-  if (closed_ || read_only_ || !manifest_file_number_ || !descriptor_log_) {
+  if (closed_ || unchanging_ || !manifest_file_number_ || !descriptor_log_) {
     return s;
   }
 
@@ -5741,7 +5773,8 @@ Status VersionSet::ProcessManifestWrites(
       assert(new_cf_options != nullptr);
       assert(max_last_sequence == descriptor_last_sequence_);
       CreateColumnFamily(*new_cf_options, read_options,
-                         first_writer.edit_list.front());
+                         first_writer.edit_list.front(),
+                         /*read_only*/ false);
     } else if (first_writer.edit_list.front()->IsColumnFamilyDrop()) {
       assert(batch_edits.size() == 1);
       assert(max_last_sequence == descriptor_last_sequence_);
@@ -7294,8 +7327,10 @@ uint64_t VersionSet::GetObsoleteSstFilesSize() const {
 
 ColumnFamilyData* VersionSet::CreateColumnFamily(
     const ColumnFamilyOptions& cf_options, const ReadOptions& read_options,
-    const VersionEdit* edit) {
+    const VersionEdit* edit, bool read_only) {
   assert(edit->IsColumnFamilyAdd());
+  // Unchanging LSM tree implies no writes to the CF
+  assert(!unchanging_ || read_only);
 
   MutableCFOptions dummy_cf_options;
   Version* dummy_versions =
@@ -7305,7 +7340,7 @@ ColumnFamilyData* VersionSet::CreateColumnFamily(
   dummy_versions->Ref();
   auto new_cfd = column_family_set_->CreateColumnFamily(
       edit->GetColumnFamilyName(), edit->GetColumnFamily(), dummy_versions,
-      cf_options);
+      cf_options, read_only);
 
   Version* v = new Version(new_cfd, this, file_options_,
                            new_cfd->GetLatestMutableCFOptions(), io_tracer_,
@@ -7429,7 +7464,7 @@ ReactiveVersionSet::ReactiveVersionSet(
                  write_buffer_manager, write_controller,
                  /*block_cache_tracer=*/nullptr, io_tracer, /*db_id*/ "",
                  /*db_session_id*/ "", /*daily_offpeak_time_utc*/ "",
-                 /*error_handler=*/nullptr, /*read_only=*/true) {}
+                 /*error_handler=*/nullptr, /*unchanging=*/false) {}
 
 ReactiveVersionSet::~ReactiveVersionSet() = default;
 

db/version_set.h

@@ -630,7 +630,8 @@ class VersionStorageInfo {
                                      const Slice& largest_user_key,
                                      int last_level, int last_l0_idx);
 
-  Env::WriteLifeTimeHint CalculateSSTWriteHint(int level) const;
+  Env::WriteLifeTimeHint CalculateSSTWriteHint(
+      int level, CompactionStyleSet compaction_style_set) const;
 
   const Comparator* user_comparator() const { return user_comparator_; }
 
@@ -993,17 +994,21 @@ class Version {
                             const FileMetaData* file_meta,
                             const std::string* fname = nullptr) const;
 
-  // REQUIRES: lock is held
   // On success, *props will be populated with all SSTables' table properties.
   // The keys of `props` are the sst file name, the values of `props` are the
   // tables' properties, represented as std::shared_ptr.
   Status GetPropertiesOfAllTables(const ReadOptions& read_options,
-                                  TablePropertiesCollection* props);
+                                  TablePropertiesCollection* props) const;
   Status GetPropertiesOfAllTables(const ReadOptions& read_options,
-                                  TablePropertiesCollection* props, int level);
+                                  TablePropertiesCollection* props,
+                                  int level) const;
   Status GetPropertiesOfTablesInRange(const ReadOptions& read_options,
                                       const autovector<UserKeyRange>& ranges,
                                       TablePropertiesCollection* props) const;
+  Status GetPropertiesOfTablesByLevel(
+      const ReadOptions& read_options,
+      std::vector<std::unique_ptr<TablePropertiesCollection>>* props_by_level)
+      const;
 
   // Print summary of range delete tombstones in SST files into out_str,
   // with maximum max_entries_to_print entries printed out.
@@ -1174,6 +1179,9 @@ class AtomicGroupReadBuffer {
 // VersionSet is the collection of versions of all the column families of the
 // database. Each database owns one VersionSet. A VersionSet has access to all
 // column families via ColumnFamilySet, i.e. set of the column families.
+// `unchanging` means the LSM tree structure of the column families will not
+// change during the lifetime of this VersionSet (true for read-only instance,
+// but false for secondary instance or writable DB).
 class VersionSet {
  public:
   VersionSet(const std::string& dbname, const ImmutableDBOptions* db_options,
@@ -1184,7 +1192,7 @@ class VersionSet {
              const std::shared_ptr<IOTracer>& io_tracer,
              const std::string& db_id, const std::string& db_session_id,
              const std::string& daily_offpeak_time_utc,
-             ErrorHandler* const error_handler, const bool read_only);
+             ErrorHandler* error_handler, bool unchanging);
   // No copying allowed
   VersionSet(const VersionSet&) = delete;
   void operator=(const VersionSet&) = delete;
@@ -1263,8 +1271,11 @@ class VersionSet {
   void WakeUpWaitingManifestWriters();
 
   // Recover the last saved descriptor (MANIFEST) from persistent storage.
-  // If read_only == true, Recover() will not complain if some column families
-  // are not opened
+  // Unlike `unchanging` on the VersionSet, `read_only` here and in other
+  // functions below refers to the CF receiving no writes or modifications
+  // through this VersionSet, but could through external manifest updates
+  // etc. Thus, `read_only=true` for secondary instances as well as read-only
+  // instances.
   Status Recover(const std::vector<ColumnFamilyDescriptor>& column_families,
                  bool read_only = false, std::string* db_id = nullptr,
                  bool no_error_if_files_missing = false, bool is_retry = false,
@@ -1342,6 +1353,8 @@ class VersionSet {
     return min_log_number_to_keep_.load();
   }
 
+  bool unchanging() const { return unchanging_; }
+
   // Allocate and return a new file number
   uint64_t NewFileNumber() { return next_file_number_.fetch_add(1); }
 
@@ -1573,6 +1586,8 @@ class VersionSet {
     AppendVersion(cfd, version);
   }
 
+  bool& TEST_unchanging() { return const_cast<bool&>(unchanging_); }
+
  protected:
   struct ManifestWriter;
 
@@ -1625,7 +1640,7 @@ class VersionSet {
 
   ColumnFamilyData* CreateColumnFamily(const ColumnFamilyOptions& cf_options,
                                        const ReadOptions& read_options,
-                                       const VersionEdit* edit);
+                                       const VersionEdit* edit, bool read_only);
 
   Status VerifyFileMetadata(const ReadOptions& read_options,
                             ColumnFamilyData* cfd, const std::string& fpath,
@@ -1722,7 +1737,7 @@ class VersionSet {
                            VersionEdit* edit, SequenceNumber* max_last_sequence,
                            InstrumentedMutex* mu);
 
-  const bool read_only_;
+  const bool unchanging_;
   bool closed_;
 };
 

db/version_set_test.cc

@@ -26,6 +26,7 @@
 #include "test_util/mock_time_env.h"
 #include "test_util/testharness.h"
 #include "test_util/testutil.h"
+#include "util/defer.h"
 #include "util/string_util.h"
 
 namespace ROCKSDB_NAMESPACE {
@@ -1905,7 +1906,7 @@ TEST_F(VersionSetTest, WalAddition) {
         &write_buffer_manager_, &write_controller_,
         /*block_cache_tracer=*/nullptr, /*io_tracer=*/nullptr,
         /*db_id=*/"", /*db_session_id=*/"", /*daily_offpeak_time_utc=*/"",
-        /*error_handler=*/nullptr, /*read_only=*/false));
+        /*error_handler=*/nullptr, /*unchanging=*/false));
     ASSERT_OK(new_versions->Recover(column_families_, /*read_only=*/false));
     const auto& wals = new_versions->GetWalSet().GetWals();
     ASSERT_EQ(wals.size(), 1);
@@ -1973,7 +1974,7 @@ TEST_F(VersionSetTest, WalCloseWithoutSync) {
         &write_buffer_manager_, &write_controller_,
         /*block_cache_tracer=*/nullptr, /*io_tracer=*/nullptr,
         /*db_id=*/"", /*db_session_id=*/"", /*daily_offpeak_time_utc=*/"",
-        /*error_handler=*/nullptr, /*read_only=*/false));
+        /*error_handler=*/nullptr, /*unchanging=*/false));
     ASSERT_OK(new_versions->Recover(column_families_, false));
     const auto& wals = new_versions->GetWalSet().GetWals();
     ASSERT_EQ(wals.size(), 2);
@@ -2027,7 +2028,7 @@ TEST_F(VersionSetTest, WalDeletion) {
         &write_buffer_manager_, &write_controller_,
         /*block_cache_tracer=*/nullptr, /*io_tracer=*/nullptr,
         /*db_id=*/"", /*db_session_id=*/"", /*daily_offpeak_time_utc=*/"",
-        /*error_handler=*/nullptr, /*read_only=*/false));
+        /*error_handler=*/nullptr, /*unchanging=*/false));
     ASSERT_OK(new_versions->Recover(column_families_, false));
     const auto& wals = new_versions->GetWalSet().GetWals();
     ASSERT_EQ(wals.size(), 1);
@@ -2066,7 +2067,7 @@ TEST_F(VersionSetTest, WalDeletion) {
         &write_buffer_manager_, &write_controller_,
         /*block_cache_tracer=*/nullptr, /*io_tracer=*/nullptr,
         /*db_id=*/"", /*db_session_id=*/"", /*daily_offpeak_time_utc=*/"",
-        /*error_handler=*/nullptr, /*read_only=*/false));
+        /*error_handler=*/nullptr, /*unchanging=*/false));
     ASSERT_OK(new_versions->Recover(column_families_, false));
     const auto& wals = new_versions->GetWalSet().GetWals();
     ASSERT_EQ(wals.size(), 1);
@@ -2187,7 +2188,7 @@ TEST_F(VersionSetTest, DeleteWalsBeforeNonExistingWalNumber) {
         &write_buffer_manager_, &write_controller_,
         /*block_cache_tracer=*/nullptr, /*io_tracer=*/nullptr,
         /*db_id=*/"", /*db_session_id=*/"", /*daily_offpeak_time_utc=*/"",
-        /*error_handler=*/nullptr, /*read_only=*/false));
+        /*error_handler=*/nullptr, /*unchanging=*/false));
     ASSERT_OK(new_versions->Recover(column_families_, false));
     const auto& wals = new_versions->GetWalSet().GetWals();
     ASSERT_EQ(wals.size(), 1);
@@ -2224,7 +2225,7 @@ TEST_F(VersionSetTest, DeleteAllWals) {
         &write_buffer_manager_, &write_controller_,
         /*block_cache_tracer=*/nullptr, /*io_tracer=*/nullptr,
         /*db_id=*/"", /*db_session_id=*/"", /*daily_offpeak_time_utc=*/"",
-        /*error_handler=*/nullptr, /*read_only=*/false));
+        /*error_handler=*/nullptr, /*unchanging=*/false));
     ASSERT_OK(new_versions->Recover(column_families_, false));
     const auto& wals = new_versions->GetWalSet().GetWals();
     ASSERT_EQ(wals.size(), 0);
@@ -2267,7 +2268,7 @@ TEST_F(VersionSetTest, AtomicGroupWithWalEdits) {
         &write_buffer_manager_, &write_controller_,
         /*block_cache_tracer=*/nullptr, /*io_tracer=*/nullptr,
         /*db_id=*/"", /*db_session_id=*/"", /*daily_offpeak_time_utc=*/"",
-        /*error_handler=*/nullptr, /*read_only=*/false));
+        /*error_handler=*/nullptr, /*unchanging=*/false));
     std::string db_id;
     ASSERT_OK(
         new_versions->Recover(column_families_, /*read_only=*/false, &db_id));
@@ -2447,7 +2448,7 @@ class VersionSetWithTimestampTest : public VersionSetTest {
         &write_buffer_manager_, &write_controller_,
         /*block_cache_tracer=*/nullptr, /*io_tracer=*/nullptr,
         /*db_id=*/"", /*db_session_id=*/"", /*daily_offpeak_time_utc=*/"",
-        /*error_handler=*/nullptr, /*read_only=*/false));
+        /*error_handler=*/nullptr, /*unchanging=*/false));
     ASSERT_OK(vset->Recover(column_families_, /*read_only=*/false,
                             /*db_id=*/nullptr));
     for (auto* cfd : *(vset->GetColumnFamilySet())) {
@@ -3749,6 +3750,8 @@ TEST_P(VersionSetTestEmptyDb, OpenCompleteManifest) {
   }
   std::string db_id;
   bool has_missing_table_file = false;
+  SaveAndRestore<bool> override_unchanging(&versions_->TEST_unchanging(),
+                                           read_only);
   s = versions_->TryRecoverFromOneManifest(manifest_path, column_families,
                                            read_only, &db_id,
                                            &has_missing_table_file);

db/wal_manager.cc

@@ -283,6 +283,7 @@ void WalManager::ArchiveWALFile(const std::string& fname, uint64_t number) {
   // The sync point below is used in (DBTest,TransactionLogIteratorRace)
   TEST_SYNC_POINT("WalManager::PurgeObsoleteFiles:1");
   Status s = env_->RenameFile(fname, archived_log_name);
+  IGNORE_STATUS_IF_ERROR(s);
   // The sync point below is used in (DBTest,TransactionLogIteratorRace)
   TEST_SYNC_POINT("WalManager::PurgeObsoleteFiles:2");
   // The sync point below is used in

db/write_batch.cc

@@ -815,6 +815,12 @@ WriteBatchInternal::GetColumnFamilyIdAndTimestampSize(
         s = Status::InvalidArgument("Default cf timestamp size mismatch");
       }
     }
+    auto* cfd =
+        static_cast_with_check<ColumnFamilyHandleImpl>(column_family)->cfd();
+    if (cfd && cfd->ioptions().disallow_memtable_writes) {
+      s = Status::InvalidArgument(
+          "This column family has disallow_memtable_writes=true");
+    }
   } else if (b->default_cf_ts_sz_ > 0) {
     ts_sz = b->default_cf_ts_sz_;
   }
@@ -836,6 +842,12 @@ Status CheckColumnFamilyTimestampSize(ColumnFamilyHandle* column_family,
   if (cf_ts_sz != ts.size()) {
     return Status::InvalidArgument("timestamp size mismatch");
   }
+  auto* cfd =
+      static_cast_with_check<ColumnFamilyHandleImpl>(column_family)->cfd();
+  if (cfd && cfd->ioptions().disallow_memtable_writes) {
+    return Status::InvalidArgument(
+        "This column family has disallow_memtable_writes=true");
+  }
   return Status::OK();
 }
 }  // anonymous namespace
@@ -2185,6 +2197,13 @@ class MemTableInserter : public WriteBatch::Handler {
       }
       return false;
     }
+    auto* current = cf_mems_->current();
+    if (current && current->ioptions().disallow_memtable_writes) {
+      *s = Status::InvalidArgument(
+          "This column family has disallow_memtable_writes=true");
+      return false;
+    }
+
     if (recovering_log_number_ != 0 &&
         recovering_log_number_ < cf_mems_->GetLogNumber()) {
       // This is true only in recovery environment (recovering_log_number_ is

db_stress_tool/db_stress_shared_state.h

@@ -137,7 +137,7 @@ class SharedState {
     for (int i = 0; i < FLAGS_column_families; ++i) {
       key_locks_[i].reset(new port::Mutex[num_locks]);
     }
-    if (FLAGS_read_fault_one_in) {
+    if (FLAGS_read_fault_one_in || FLAGS_metadata_read_fault_one_in) {
 #ifdef NDEBUG
       // Unsupported in release mode because it relies on
       // `IGNORE_STATUS_IF_ERROR` to distinguish faults not expected to lead to

db_stress_tool/no_batched_ops_stress.cc

@@ -1600,12 +1600,6 @@ class NonBatchedOpsStressTest : public StressTest {
     Slice ub_slice;
     ReadOptions ro_copy = read_opts;
 
-    // There is a narrow window in iterator auto refresh run where injected read
-    // errors are simply untraceable, ex. failure to delete file as a part of
-    // superversion cleanup callback invoked by the DBIter destructor.
-    bool ignore_injected_read_error_in_iter =
-        ro_copy.auto_refresh_iterator_with_snapshot;
-
     // Randomly test with `iterate_upper_bound` and `prefix_same_as_start`
     //
     // Get the next prefix first and then see if we want to set it to be the
@@ -1698,8 +1692,7 @@ class NonBatchedOpsStressTest : public StressTest {
               FaultInjectionIOType::kRead),
           fault_fs_guard->GetAndResetInjectedThreadLocalErrorCount(
               FaultInjectionIOType::kMetadataRead));
-      if (!ignore_injected_read_error_in_iter &&
-          !SharedState::ignore_read_error && injected_error_count > 0 &&
+      if (!SharedState::ignore_read_error && injected_error_count > 0 &&
           s.ok()) {
         // Grab mutex so multiple thread don't try to print the
         // stack trace at the same time

env/env.cc

@@ -732,6 +732,37 @@ std::string Env::PriorityToString(Env::Priority priority) {
   return "Invalid";
 }
 
+std::string Env::IOActivityToString(IOActivity activity) {
+  switch (activity) {
+    case Env::IOActivity::kFlush:
+      return "Flush";
+    case Env::IOActivity::kCompaction:
+      return "Compaction";
+    case Env::IOActivity::kDBOpen:
+      return "DBOpen";
+    case Env::IOActivity::kGet:
+      return "Get";
+    case Env::IOActivity::kMultiGet:
+      return "MultiGet";
+    case Env::IOActivity::kDBIterator:
+      return "DBIterator";
+    case Env::IOActivity::kVerifyDBChecksum:
+      return "VerifyDBChecksum";
+    case Env::IOActivity::kVerifyFileChecksums:
+      return "VerifyFileChecksums";
+    case Env::IOActivity::kGetEntity:
+      return "GetEntity";
+    case Env::IOActivity::kMultiGetEntity:
+      return "MultiGetEntity";
+    case Env::IOActivity::kGetFileChecksumsFromCurrentManifest:
+      return "GetFileChecksumsFromCurrentManifest";
+    case Env::IOActivity::kUnknown:
+      return "Unknown";
+  };
+  assert(false);
+  return "Invalid";
+}
+
 uint64_t Env::GetThreadID() const {
   std::hash<std::thread::id> hasher;
   return hasher(std::this_thread::get_id());

file/delete_scheduler.cc

@@ -130,6 +130,7 @@ Status DeleteScheduler::AddFileToDeletionQueue(const std::string& file_path,
                  s.ToString().c_str());
 
   if (!s.ok()) {
+    IGNORE_STATUS_IF_ERROR(s);
     ROCKS_LOG_ERROR(info_log_, "Failed to mark %s as trash -- %s",
                     file_path.c_str(), s.ToString().c_str());
     s = fs_->DeleteFile(file_path, IOOptions(), nullptr);
@@ -151,6 +152,7 @@ Status DeleteScheduler::AddFileToDeletionQueue(const std::string& file_path,
     if (io_s.ok()) {
       total_trash_size_.fetch_add(trash_file_size);
     }
+    IGNORE_STATUS_IF_ERROR(s);
   }
   //**TODO: What should we do if we failed to
   // get the file size?

include/rocksdb/advanced_options.h

@@ -719,6 +719,17 @@ struct AdvancedColumnFamilyOptions {
   // Dynamically changeable through SetOptions() API
   bool report_bg_io_stats = false;
 
+  // Setting this option to true disallows ordinary writes to the column family
+  // and it can only be populated through import and ingestion. It is intended
+  // to protect "ingestion only" column families. This option is not currently
+  // supported on the default column family because of error handling challenges
+  // analogous to https://github.com/facebook/rocksdb/issues/13429
+  //
+  // This option is not mutable with SetOptions(). It can be changed between
+  // DB::Open() calls, but open will fail if recovering WAL writes to a CF with
+  // this option set.
+  bool disallow_memtable_writes = false;
+
   // This option has different meanings for different compaction styles:
   //
   // Leveled: Non-bottom-level files with all keys older than TTL will go

include/rocksdb/cache.h

@@ -300,7 +300,7 @@ struct CompressedSecondaryCacheOptions : LRUCacheOptions {
 
   // compress_format_version can have two values:
   // compress_format_version == 1 -- decompressed size is not included in the
-  // block header.
+  // block header. DEPRECATED
   // compress_format_version == 2 -- decompressed size is included in the block
   // header in varint32 format.
   uint32_t compress_format_version = 2;

include/rocksdb/compaction_job_stats.h

@@ -33,6 +33,8 @@ struct CompactionJobStats {
   uint64_t num_blobs_read = 0;
   // the number of compaction input files (table files)
   size_t num_input_files = 0;
+  // The number of input files that get trivially moved.
+  size_t num_input_files_trivially_moved = 0;
   // the number of compaction input files at the output level (table files)
   size_t num_input_files_at_output_level = 0;
   // the number of compaction input files that are filtered out by compaction
@@ -118,6 +120,6 @@ struct CompactionJobStats {
   // number of single-deletes which meet something other than a put
   uint64_t num_single_del_mismatch = 0;
 
-  // TODO: Add output_to_penultimate_level output information
+  // TODO: Add output_to_proximal_level output information
 };
 }  // namespace ROCKSDB_NAMESPACE

include/rocksdb/convenience.h

@@ -410,7 +410,7 @@ Status GetStringFromColumnFamilyOptions(std::string* opts_str,
 Status GetStringFromCompressionType(std::string* compression_str,
                                     CompressionType compression_type);
 
-std::vector<CompressionType> GetSupportedCompressions();
+const std::vector<CompressionType>& GetSupportedCompressions();
 
 Status GetBlockBasedTableOptionsFromString(
     const ConfigOptions& config_options,
@@ -450,6 +450,22 @@ Status DeleteFilesInRange(DB* db, ColumnFamilyHandle* column_family,
 // Delete files in multiple ranges at once
 // Delete files in a lot of ranges one at a time can be slow, use this API for
 // better performance in that case.
+Status DeleteFilesInRanges(DB* db, ColumnFamilyHandle* column_family,
+                           const RangeOpt* ranges, size_t n,
+                           bool include_end = true);
+
+// DEPRECATED
+struct RangePtr {
+  // In case of user_defined timestamp, if enabled, `start` and `limit` should
+  // point to key without timestamp part.
+  const Slice* start;
+  const Slice* limit;
+
+  RangePtr() : start(nullptr), limit(nullptr) {}
+  RangePtr(const Slice* s, const Slice* l) : start(s), limit(l) {}
+};
+
+// DEPRECATED
 Status DeleteFilesInRanges(DB* db, ColumnFamilyHandle* column_family,
                            const RangePtr* ranges, size_t n,
                            bool include_end = true);

include/rocksdb/db.h

@@ -96,43 +96,6 @@ class ColumnFamilyHandle {
 static const int kMajorVersion = __ROCKSDB_MAJOR__;
 static const int kMinorVersion = __ROCKSDB_MINOR__;
 
-// A range of keys
-struct Range {
-  // In case of user_defined timestamp, if enabled, `start` and `limit` should
-  // point to key without timestamp part.
-  Slice start;
-  Slice limit;
-
-  Range() {}
-  Range(const Slice& s, const Slice& l) : start(s), limit(l) {}
-};
-
-struct RangePtr {
-  // In case of user_defined timestamp, if enabled, `start` and `limit` should
-  // point to key without timestamp part.
-  const Slice* start;
-  const Slice* limit;
-
-  RangePtr() : start(nullptr), limit(nullptr) {}
-  RangePtr(const Slice* s, const Slice* l) : start(s), limit(l) {}
-};
-
-// It is valid that files_checksums and files_checksum_func_names are both
-// empty (no checksum information is provided for ingestion). Otherwise,
-// their sizes should be the same as external_files. The file order should
-// be the same in three vectors and guaranteed by the caller.
-// Note that, we assume the temperatures of this batch of files to be
-// ingested are the same.
-struct IngestExternalFileArg {
-  ColumnFamilyHandle* column_family = nullptr;
-  std::vector<std::string> external_files;
-  IngestExternalFileOptions options;
-  std::vector<std::string> files_checksums;
-  std::vector<std::string> files_checksum_func_names;
-  // A hint as to the temperature for *reading* the files to be ingested.
-  Temperature file_temperature = Temperature::kUnknown;
-};
-
 struct GetMergeOperandsOptions {
   using ContinueCallback = std::function<bool(Slice)>;
 
@@ -2081,14 +2044,11 @@ class DB {
       ColumnFamilyHandle* column_family, const Range* range, std::size_t n,
       TablePropertiesCollection* props) = 0;
 
-  // Get the table properties of files per level.
-  virtual Status GetPropertiesOfTablesForLevels(
-      ColumnFamilyHandle* /* column_family */,
-      std::vector<
-          std::unique_ptr<TablePropertiesCollection>>* /* levels_props */) {
-    return Status::NotSupported(
-        "GetPropertiesOfTablesForLevels() is not implemented.");
-  }
+  // Get the table properties of files by level.
+  virtual Status GetPropertiesOfTablesByLevel(
+      ColumnFamilyHandle* column_family,
+      std::vector<std::unique_ptr<TablePropertiesCollection>>*
+          props_by_level) = 0;
 
   virtual Status SuggestCompactRange(ColumnFamilyHandle* /*column_family*/,
                                      const Slice* /*begin*/,

include/rocksdb/env.h

@@ -455,10 +455,12 @@ class Env : public Customizable {
     kVerifyFileChecksums = 7,
     kGetEntity = 8,
     kMultiGetEntity = 9,
-    kReadManifest = 10,
+    kGetFileChecksumsFromCurrentManifest = 10,
     kUnknown,  // Keep last for easy array of non-unknowns
   };
 
+  static std::string IOActivityToString(IOActivity activity);
+
   // Arrange to run "(*function)(arg)" once in a background thread, in
   // the thread pool specified by pri. By default, jobs go to the 'LOW'
   // priority thread pool.

include/rocksdb/external_table.h

@@ -6,6 +6,7 @@
 #pragma once
 
 #include "rocksdb/customizable.h"
+#include "rocksdb/file_checksum.h"
 #include "rocksdb/iterator.h"
 #include "rocksdb/options.h"
 #include "rocksdb/status.h"
@@ -18,22 +19,21 @@ class ExternalTableFactory;
 // The interface defined in this file is subject to change at any time without
 // warning!!
 
-// This file defines an interface for plugging in an external table reader
+// This file defines an interface for plugging in an external table
 // into RocksDB. The external table reader will be used instead of the
-// BlockBasedTable to load and query sst files. As of now, creating the
-// external table files using RocksDB is not supported, but will be added in
-// the near future. The external table files can be created outside and
-// RocksDB and ingested into a RocksDB instance using the IngestExternalFIle()
-// API.
+// BlockBasedTable to load and query sst files.
+// The external table files can be created using an SstFileWriter. Eventually
+// external tables will be allowed to be ingested into a RocksDB instance
+// using the IngestExternalFIle() API.
 //
-// Initial support is for loading and querying the files using an
-// SstFileReader. We will add support for ingestion of an external table
-// into a limited RocksDB instance that only supports ingestion and not live
-// writes in the near future. It'll be followed by support for replacing the
-// column family by ingesting a new set of files. In all cases, the external
-// table files will only be allowed in the bottommost level.
+// Initial support is for writing and querying the files using an
+// SstFileWriter and SstFileReader. We will add support for ingestion of an
+// external table into a limited RocksDB instance that only supports ingestion
+// and not live writes in the near future. It'll be followed by support for
+// replacing the column family by ingesting a new set of files. In all cases,
+// the external table files will only be allowed in the bottommost level.
 //
-// The external table reader can support one or both of the following layouts -
+// The external table can support one or both of the following layouts -
 // 1. Total order seek - All the keys in the files are in sorted order, and a
 //    user can seek to the first, last, or any key in between and iterate
 //    forwards or backwards till the end of the range. To support this mode,
@@ -54,8 +54,8 @@ class ExternalTableFactory;
 //    true to seek to the first non-empty prefix (as determined by the key
 //    order) if the seek prefix is empty.
 //
-// Many of the options in ReadOptions may not be relevant to the external
-// table implementation.
+// Many of the options in ReadOptions and WriteOptions may not be relevant to
+// the external table implementation.
 // TODO: Specify which options are relevant
 
 class ExternalTableReader {
@@ -94,6 +94,62 @@ class ExternalTableReader {
   }
 };
 
+// A table builder interface that can be used by SstFileWriter to allow
+// RocksDB users to write external table files. The sequence of operations
+// to write an external table is as follows -
+// 1. Add() is called one or more times to write all key-values to the table.
+//    Its called in increasing key order, as determined by the comparator.
+//    The input key is a user key, i.e sequence number and value type are
+//    stripped out.
+// 2. After every Add() operation, status() is called to check the current
+//    status.
+// 3. After the last key is added, Finish() is called to do whatever is
+//    necessary to ensure the data is persisted in the table file.
+// 4. If there is a failure midway for some reason, Abandon() is called
+//    instead of Finish().
+// 5. At the end, FileSize(), GetTableProperties(), and status() are called to
+//    get the final size of the file, the table properties, and the final
+//    status. GetFileChecksum() and GetFileChecksumFuncName() may also be
+//    called to get checksum information about the whole file, but their
+//    implementation is optional.
+class ExternalTableBuilder {
+ public:
+  virtual ~ExternalTableBuilder() {}
+
+  // Write a single KV to the table file. This is guaranteed to be called
+  // in key order, and the write may be buffered and flushed at a later time.
+  virtual void Add(const Slice& key, const Slice& value) = 0;
+
+  // Return the current Status. This could return non-ok, for example, if
+  // Add() fails for some reason.
+  virtual Status status() const = 0;
+
+  // Flush and close the table file
+  virtual Status Finish() = 0;
+
+  // Delete the partial file and release any allocated resources. Either this
+  // or Finish() will be called, but not both.
+  virtual void Abandon() = 0;
+
+  // Return the size of the table file. Will be called at the end, after
+  // Finish().
+  virtual uint64_t FileSize() const = 0;
+
+  //  As mentioned in earlier comments, the following table properties must be
+  //  returned at a minimum -
+  //  comparator_name
+  //  num_entries
+  //  raw_key_size
+  //  raw_value_size
+  virtual TableProperties GetTableProperties() const = 0;
+
+  virtual std::string GetFileChecksum() const { return kUnknownFileChecksum; }
+
+  virtual const char* GetFileChecksumFuncName() const {
+    return kUnknownFileChecksumFuncName;
+  }
+};
+
 struct ExternalTableOptions {
   const std::shared_ptr<const SliceTransform>& prefix_extractor;
   const Comparator* comparator;
@@ -104,6 +160,29 @@ struct ExternalTableOptions {
       : prefix_extractor(_prefix_extractor), comparator(_comparator) {}
 };
 
+struct ExternalTableBuilderOptions {
+  const ReadOptions& read_options;
+  const WriteOptions& write_options;
+  const std::shared_ptr<const SliceTransform>& prefix_extractor;
+  const Comparator* comparator;
+  const std::string& column_family_name;
+  const std::string db_id;
+  const std::string db_session_id;
+  const TableFileCreationReason reason;
+
+  ExternalTableBuilderOptions(
+      const ReadOptions& _read_options, const WriteOptions& _write_options,
+      const std::shared_ptr<const SliceTransform>& _prefix_extractor,
+      const Comparator* _comparator, const std::string& _column_family_name,
+      const TableFileCreationReason _reason)
+      : read_options(_read_options),
+        write_options(_write_options),
+        prefix_extractor(_prefix_extractor),
+        comparator(_comparator),
+        column_family_name(_column_family_name),
+        reason(_reason) {}
+};
+
 class ExternalTableFactory : public Customizable {
  public:
   ~ExternalTableFactory() override {}
@@ -113,7 +192,11 @@ class ExternalTableFactory : public Customizable {
   virtual Status NewTableReader(
       const ReadOptions& read_options, const std::string& file_path,
       const ExternalTableOptions& table_options,
-      std::unique_ptr<ExternalTableReader>* table_reader) = 0;
+      std::unique_ptr<ExternalTableReader>* table_reader) const = 0;
+
+  virtual ExternalTableBuilder* NewTableBuilder(
+      const ExternalTableBuilderOptions& builder_options,
+      const std::string& file_path) const = 0;
 };
 
 // Allocate a TableFactory that wraps around an ExternalTableFactory. Use this

include/rocksdb/file_system.h

@@ -192,6 +192,12 @@ struct FileOptions : EnvOptions {
   // handoff during file writes.
   ChecksumType handoff_checksum_type;
 
+  // Expose write lifetime hint on the FileOptions level to provide more
+  // flexibility in setting the hint in downstream, custom implementations
+  // that might be able to process the hint only at the time of the actual
+  // FSWritableFile object creation.
+  Env::WriteLifeTimeHint write_hint = Env::WLTH_NOT_SET;
+
   FileOptions() : EnvOptions(), handoff_checksum_type(ChecksumType::kCRC32c) {}
 
   FileOptions(const DBOptions& opts)
@@ -206,7 +212,8 @@ struct FileOptions : EnvOptions {
       : EnvOptions(opts),
         io_options(opts.io_options),
         temperature(opts.temperature),
-        handoff_checksum_type(opts.handoff_checksum_type) {}
+        handoff_checksum_type(opts.handoff_checksum_type),
+        write_hint(opts.write_hint) {}
 
   FileOptions& operator=(const FileOptions&) = default;
 };

include/rocksdb/listener.h

@@ -439,6 +439,9 @@ struct CompactionJobInfo {
   // the job id, which is unique in the same thread.
   int job_id;
 
+  // the number of L0 files in the CF right before and after the compaction
+  int num_l0_files;
+
   // the smallest input level of the compaction.
   int base_input_level;
   // the output level of the compaction.

include/rocksdb/options.h

@@ -62,6 +62,8 @@ struct Options;
 struct DbPath;
 
 using FileTypeSet = SmallEnumSet<FileType, FileType::kBlobFile>;
+using CompactionStyleSet =
+    SmallEnumSet<CompactionStyle, CompactionStyle::kCompactionStyleNone>;
 
 struct ColumnFamilyOptions : public AdvancedColumnFamilyOptions {
   // The function recovers options to a previous version. Only 4.6 or later
@@ -454,6 +456,7 @@ extern const char* kHostnameForDbHostId;
 enum class CompactionServiceJobStatus : char {
   kSuccess,
   kFailure,
+  kAborted,
   kUseLocal,
 };
 
@@ -1619,6 +1622,24 @@ struct DBOptions {
   // `kUnknown`, this overrides any temperature set by OptimizeForLogWrite
   // functions.
   Temperature wal_write_temperature = Temperature::kUnknown;
+
+  // Enum set indicative of which compaction styles SST write lifetime hint
+  // calculation is allowed on. Today, RocksDB provides native support for
+  // kCompactionStyleLevel and kCompactionStyleUniversal (experimental version).
+  // Other compaction styles, even when enabled in the set, won't have any
+  // effect in the default PosixWritableFile file implementation. There are
+  // numerous benefits coming from employing the hints including reduction in
+  // write amplification caused by OS file movement during garbage collection,
+  // and reduction in wear-leveling (SSDs). However, as currently implemented,
+  // SST write lifetime hints are calculated in a static way and solely based on
+  // the level, which might not be suitable for non-uniform workloads with
+  // dynamic / high-variance lifespan of data within the same level. In those
+  // cases (or when the performance is not satisfactory), it's recommended to
+  // disable the hints by assigning the setting to the empty set (= {});
+  //
+  // Default: Enabled in kCompactionStyleLevel mode.
+  CompactionStyleSet calculate_sst_write_lifetime_hint_set = {
+      CompactionStyle::kCompactionStyleLevel};
   // End EXPERIMENTAL
 };
 
@@ -1985,6 +2006,12 @@ struct ReadOptions {
   // reader implementation.
   uint64_t weight = 0;
 
+  // A map of name,value pairs that can be passed by the user to an
+  // external table reader. This is completely opaque to RocksDB and is
+  // ignored by the natively supported table readers like block based and plain
+  // table. This is only useful for Iterator.
+  std::optional<std::unordered_map<std::string, std::string>> property_bag;
+
   // *** END options for RocksDB internal use only ***
 
   ReadOptions() {}
@@ -2218,6 +2245,29 @@ struct CompactRangeOptions {
   double blob_garbage_collection_age_cutoff = -1;
 };
 
+// A range of keys. In case of user_defined timestamp, if enabled, `start` and
+// `limit` should point to key without timestamp part.
+struct Range {
+  Slice start;
+  Slice limit;
+
+  Range() {}
+  Range(const Slice& s, const Slice& l) : start(s), limit(l) {}
+};
+
+// A key range with optional endpoints. In case of user_defined timestamp, if
+// enabled, `start` and `limit` should point to key without timestamp part.
+struct RangeOpt {
+  // When start.has_value() == false, refers to starting before every key
+  OptSlice start;
+  // When limit.has_value() == false, refers to ending after every key
+  OptSlice limit;
+
+  RangeOpt() {}
+  RangeOpt(const OptSlice& s, const OptSlice& l) : start(s), limit(l) {}
+  // RangeOpt(const Slice& s, const Slice& l) : start(s), limit(l) {}
+};
+
 // IngestExternalFileOptions is used by IngestExternalFile()
 struct IngestExternalFileOptions {
   // Can be set to true to move the files instead of copying them.
@@ -2240,8 +2290,11 @@ struct IngestExternalFileOptions {
   // during file ingestion in the DB (the conditions under which a global_seqno
   // must be assigned to the ingested file).
   bool allow_global_seqno = true;
-  // If set to false and the file key range overlaps with the memtable key range
-  // (memtable flush required), IngestExternalFile will fail.
+  // Normally (true), IngestExternalFile() will trigger and block for flushing
+  // memtable(s) if there is overlap between ingested files and memtable(s). If
+  // allow_blocking_flush is set to false, IngestExternalFile() will fail if the
+  // file key range overlaps with the memtable key range (memtable flush
+  // required).
   bool allow_blocking_flush = true;
   // Set to true if you would like duplicate keys in the file being ingested
   // to be skipped rather than overwriting existing data under that key.
@@ -2324,6 +2377,44 @@ struct IngestExternalFileOptions {
   bool fill_cache = true;
 };
 
+// It is valid that files_checksums and files_checksum_func_names are both
+// empty (no checksum information is provided for ingestion). Otherwise,
+// their sizes should be the same as external_files. The file order should
+// be the same in three vectors and guaranteed by the caller.
+// Note that, we assume the temperatures of this batch of files to be
+// ingested are the same.
+struct IngestExternalFileArg {
+  ColumnFamilyHandle* column_family = nullptr;
+  std::vector<std::string> external_files;
+  IngestExternalFileOptions options;
+  std::vector<std::string> files_checksums;
+  std::vector<std::string> files_checksum_func_names;
+  // A hint as to the temperature for *reading* the files to be ingested.
+  Temperature file_temperature = Temperature::kUnknown;
+  // EXPERIMENTAL: When specified, existing keys in the given range will be
+  // cleared atomically as part of the ingestion, where the ingested files are
+  // logically applied on top of the cleared key range.
+  // * If both `start` and `limit` are nullptr, the entire column family is
+  // cleared; however, setting just one bound to nullptr is not yet supported.
+  // * When a range is specified, all the external files in this batch must
+  //   be contained in that key range.
+  // * Checks for memtable overlap and possible blocking flush will apply
+  //   to this range (not just the file ranges).
+  // * Not compatible with ingest_behind=true.
+  // * When options.snapshot_consistency = false, the range is cleared
+  // similarly to DeleteFilesInRange, but fails if any files overlap the range
+  // only partially.
+  //   * It is recommended to use fail_if_not_bottommost_level=true to ensure
+  //     data in the key range is ingested to a single compacted level (the
+  //     last level). (fail_if_not_bottommost_level=false allows overlap between
+  //     the ingested files.)
+  // * options.snapshot_consistency = true is not yet supported.
+  // BUG: the upper bound of the range may be interpreted as inclusive or
+  // exclusive, so it is best not to depend on one or the other until it is
+  // sorted out.
+  std::optional<RangeOpt> atomic_replace_range;
+};
+
 enum TraceFilterType : uint64_t {
   // Trace all the operations
   kTraceFilterNone = 0x0,

include/rocksdb/slice.h

@@ -20,6 +20,7 @@
 
 #include <cassert>
 #include <cstddef>
+#include <cstdint>
 #include <cstdio>
 #include <cstring>
 #include <string>
@@ -129,6 +130,46 @@ class Slice {
   // Intentionally copyable
 };
 
+// A likely more efficient alternative to std::optional<Slice>. For example,
+// an empty key might be distinct from "not specified" (and Slice* as an
+// optional is more troublesome to deal with).
+class OptSlice {
+ public:
+  OptSlice() : slice_(nullptr, SIZE_MAX) {}
+  /*implicit*/ OptSlice(const Slice& s) : slice_(s) {}
+  /*implicit*/ OptSlice(const std::string& s) : slice_(s) {}
+  /*implicit*/ OptSlice(const std::string_view& sv) : slice_(sv) {}
+  /*implicit*/ OptSlice(const char* c_str) : slice_(c_str) {}
+  // For easier migrating from APIs uing Slice* as an optional type.
+  // CAUTION: OptSlice{nullptr} is "no value" while Slice{nullptr} is "empty"
+  /*implicit*/ OptSlice(std::nullptr_t) : OptSlice() {}
+
+  bool has_value() const noexcept { return slice_.size() != SIZE_MAX; }
+  explicit operator bool() const noexcept { return has_value(); }
+
+  const Slice& value() const noexcept {
+    assert(has_value());
+    return slice_;
+  }
+  const Slice& operator*() const noexcept { return value(); }
+  const Slice* operator->() const noexcept { return &value(); }
+
+  const Slice* AsPtr() const noexcept {
+    return has_value() ? &slice_ : nullptr;
+  }
+  // Populate from an optional pointer. This is a very explicit conversion
+  // to minimize risk of bugs as in
+  //   Slice start, limit;
+  //   RangeOpt rng = {&start, &limit};
+  //   start = ...;  // BUG: would not affect rng
+  static OptSlice CopyFromPtr(const Slice* ptr) {
+    return ptr ? OptSlice{*ptr} : OptSlice{};
+  }
+
+ protected:
+  Slice slice_;
+};
+
 /**
  * A Slice that can be pinned with some cleanup tasks, which will be run upon
  * ::Reset() or object destruction, whichever is invoked first. This can be used

include/rocksdb/table.h

@@ -126,7 +126,15 @@ struct CacheUsageOptions {
 };
 
 // Configures how SST files using the block-based table format (standard)
-// are written and read.
+// are written and read. With few exceptions, each option only affects either
+// (a) how new SST files are written, or (b) how SST files are read. If an
+// option seems to affect how the SST file is constructed, e.g. format_version,
+// that option *ONLY* has an effect at construction time. Contrast this with
+// options like the various `cache` and `pin` options, that only affect
+// in-memory and IO behavior at read time. In general, any version of RocksDB
+// able to read the full key-value and indexing data in the SST file will read
+// it as written regardless of current options for writing new files. See
+// filter_policy regarding filters.
 //
 // Except as specifically noted, all options here are "mutable" using
 // SetOptions(), with the caveat that only new table builders and new table
@@ -480,6 +488,10 @@ struct BlockBasedTableOptions {
   // If non-nullptr, use the specified filter policy to reduce disk reads.
   // Many applications will benefit from passing the result of
   // NewBloomFilterPolicy() here.
+  //
+  // Because filters only impact performance and are not data-critical, an
+  // SST file can be opened and used without filters if (a) the filter
+  // policy name or schema is unrecognized, or (b) filter_policy is nullptr.
   std::shared_ptr<const FilterPolicy> filter_policy = nullptr;
 
   // If true, place whole keys in the filter (not just prefixes).
@@ -524,13 +536,9 @@ struct BlockBasedTableOptions {
   // Default: 0 (disabled)
   uint32_t read_amp_bytes_per_bit = 0;
 
-  // We currently have these versions:
-  // 0 -- This version can be read by really old RocksDB's. Doesn't support
-  // changing checksum type (default is CRC32).
-  // 1 -- Can be read by RocksDB's versions since 3.0. Supports non-default
-  // checksum, like xxHash. It is written by RocksDB when
-  // BlockBasedTableOptions::checksum is something other than kCRC32c. (version
-  // 0 is silently upconverted)
+  // We currently have these format versions:
+  // 0 - 1 -- Unsupported for writing new files and quietly sanitized to 2.
+  // Read support is deprecated and could be removed in the future.
   // 2 -- Can be read by RocksDB's versions since 3.10. Changes the way we
   // encode compressed blocks with LZ4, BZip2 and Zlib compression. If you
   // don't plan to run RocksDB before version 3.10, you should probably use

include/rocksdb/thread_status.h

@@ -64,7 +64,7 @@ struct ThreadStatus {
     OP_VERIFY_FILE_CHECKSUMS,
     OP_GETENTITY,
     OP_MULTIGETENTITY,
-    OP_READ_MANIFEST,
+    OP_GET_FILE_CHECKSUMS_FROM_CURRENT_MANIFEST,
     NUM_OP_TYPES
   };
 

include/rocksdb/utilities/debug.h

@@ -33,12 +33,12 @@ struct KeyVersion {
 // copied to memory, if the range covers too many keys, the memory usage
 // may be huge. `max_num_ikeys` can be used to cap the memory usage.
 // The result is inserted into the provided vector, `key_versions`.
-Status GetAllKeyVersions(DB* db, Slice begin_key, Slice end_key,
+Status GetAllKeyVersions(DB* db, OptSlice begin_key, OptSlice end_key,
                          size_t max_num_ikeys,
                          std::vector<KeyVersion>* key_versions);
 
-Status GetAllKeyVersions(DB* db, ColumnFamilyHandle* cfh, Slice begin_key,
-                         Slice end_key, size_t max_num_ikeys,
+Status GetAllKeyVersions(DB* db, ColumnFamilyHandle* cfh, OptSlice begin_key,
+                         OptSlice end_key, size_t max_num_ikeys,
                          std::vector<KeyVersion>* key_versions);
 
 }  // namespace ROCKSDB_NAMESPACE

include/rocksdb/utilities/stackable_db.h

@@ -554,6 +554,14 @@ class StackableDB : public DB {
     return db_->GetPropertiesOfTablesInRange(column_family, range, n, props);
   }
 
+  using DB::GetPropertiesOfTablesByLevel;
+  Status GetPropertiesOfTablesByLevel(
+      ColumnFamilyHandle* column_family,
+      std::vector<std::unique_ptr<TablePropertiesCollection>>* props_by_level)
+      override {
+    return db_->GetPropertiesOfTablesByLevel(column_family, props_by_level);
+  }
+
   Status GetUpdatesSince(
       SequenceNumber seq_number, std::unique_ptr<TransactionLogIterator>* iter,
       const TransactionLogIterator::ReadOptions& read_options) override {

include/rocksdb/version.h

@@ -12,7 +12,7 @@
 // NOTE: in 'main' development branch, this should be the *next*
 // minor or major version number planned for release.
 #define ROCKSDB_MAJOR 10
-#define ROCKSDB_MINOR 0
+#define ROCKSDB_MINOR 2
 #define ROCKSDB_PATCH 0
 
 // Do not use these. We made the mistake of declaring macros starting with

java/rocksjni/rocksjni.cc

@@ -3637,7 +3637,7 @@ void Java_org_rocksdb_RocksDB_destroyDB(JNIEnv* env, jclass, jstring jdb_path,
 }
 
 bool get_slice_helper(JNIEnv* env, jobjectArray ranges, jsize index,
-                      std::unique_ptr<ROCKSDB_NAMESPACE::Slice>& slice,
+                      ROCKSDB_NAMESPACE::OptSlice& opt_slice,
                       std::vector<std::unique_ptr<jbyte[]>>& ranges_to_free) {
   jobject jArray = env->GetObjectArrayElement(ranges, index);
   if (env->ExceptionCheck()) {
@@ -3659,8 +3659,8 @@ bool get_slice_helper(JNIEnv* env, jobjectArray ranges, jsize index,
     return false;
   }
   env->DeleteLocalRef(jArray);
-  slice.reset(new ROCKSDB_NAMESPACE::Slice(
-      reinterpret_cast<char*>(ranges_to_free.back().get()), len_ba));
+  opt_slice = ROCKSDB_NAMESPACE::Slice(
+      reinterpret_cast<char*>(ranges_to_free.back().get()), len_ba);
   return true;
 }
 /*
@@ -3675,24 +3675,24 @@ void Java_org_rocksdb_RocksDB_deleteFilesInRanges(JNIEnv* env, jclass /*jdb*/,
                                                   jboolean include_end) {
   jsize length = env->GetArrayLength(ranges);
 
-  std::vector<ROCKSDB_NAMESPACE::RangePtr> rangesVector;
-  std::vector<std::unique_ptr<ROCKSDB_NAMESPACE::Slice>> slices;
+  std::vector<ROCKSDB_NAMESPACE::RangeOpt> rangesVector;
+  std::vector<ROCKSDB_NAMESPACE::OptSlice> slices;
   std::vector<std::unique_ptr<jbyte[]>> ranges_to_free;
   for (jsize i = 0; (i + 1) < length; i += 2) {
-    slices.push_back(std::unique_ptr<ROCKSDB_NAMESPACE::Slice>());
+    slices.emplace_back();
     if (!get_slice_helper(env, ranges, i, slices.back(), ranges_to_free)) {
       // exception thrown
       return;
     }
 
-    slices.push_back(std::unique_ptr<ROCKSDB_NAMESPACE::Slice>());
+    slices.emplace_back();
     if (!get_slice_helper(env, ranges, i + 1, slices.back(), ranges_to_free)) {
       // exception thrown
       return;
     }
 
-    rangesVector.push_back(ROCKSDB_NAMESPACE::RangePtr(
-        slices[slices.size() - 2].get(), slices[slices.size() - 1].get()));
+    rangesVector.push_back(ROCKSDB_NAMESPACE::RangeOpt(
+        slices[slices.size() - 2], slices[slices.size() - 1]));
   }
 
   auto* db = reinterpret_cast<ROCKSDB_NAMESPACE::DB*>(jdb_handle);

memtable/skiplist.h

@@ -134,8 +134,8 @@ class SkipList {
   // i up to max_height_ is the predecessor of prev_[0] and prev_height_
   // is the height of prev_[0].  prev_[0] can only be equal to head before
   // insertion, in which case max_height_ and prev_height_ are 1.
-  Node** prev_;
   int32_t prev_height_;
+  Node** prev_;
 
   inline int GetMaxHeight() const {
     return max_height_.load(std::memory_order_relaxed);
@@ -438,7 +438,7 @@ SkipList<Key, Comparator>::SkipList(const Comparator cmp, Allocator* allocator,
       kScaledInverseBranching_((Random::kMaxNext + 1) / kBranching_),
       compare_(cmp),
       allocator_(allocator),
-      head_(NewNode(0 /* any key will do */, max_height)),
+      head_(NewNode({} /* any key will do */, max_height)),
       max_height_(1),
       prev_height_(1) {
   assert(max_height > 0 && kMaxHeight_ == static_cast<uint32_t>(max_height));

memtable/wbwi_memtable.cc

@@ -61,6 +61,7 @@ bool WBWIMemTable::Get(const LookupKey& key, std::string* value,
   assert(!wbwi_->GetWriteBatch()->HasDeleteRange());
   assert(merge_context);
 
+  *out_seq = kMaxSequenceNumber;
   [[maybe_unused]] SequenceNumber read_seq =
       GetInternalKeySeqno(key.internal_key());
   // This is memtable is a single write batch, no snapshot can be taken within

monitoring/thread_status_util_debug.cc

@@ -50,8 +50,9 @@ Env::IOActivity ThreadStatusUtil::TEST_GetExpectedIOActivity(
       return Env::IOActivity::kGetEntity;
     case ThreadStatus::OperationType::OP_MULTIGETENTITY:
       return Env::IOActivity::kMultiGetEntity;
-    case ThreadStatus::OperationType::OP_READ_MANIFEST:
-      return Env::IOActivity::kReadManifest;
+    case ThreadStatus::OperationType::
+        OP_GET_FILE_CHECKSUMS_FROM_CURRENT_MANIFEST:
+      return Env::IOActivity::kGetFileChecksumsFromCurrentManifest;
     default:
       return Env::IOActivity::kUnknown;
   }

options/cf_options.cc

@@ -736,6 +736,10 @@ static std::unordered_map<std::string, OptionTypeInfo>
          {offsetof(struct ImmutableCFOptions, force_consistency_checks),
           OptionType::kBoolean, OptionVerificationType::kNormal,
           OptionTypeFlags::kNone}},
+        {"disallow_memtable_writes",
+         {offsetof(struct ImmutableCFOptions, disallow_memtable_writes),
+          OptionType::kBoolean, OptionVerificationType::kNormal,
+          OptionTypeFlags::kNone}},
         {"default_temperature",
          {offsetof(struct ImmutableCFOptions, default_temperature),
           OptionType::kTemperature, OptionVerificationType::kNormal,
@@ -998,6 +1002,7 @@ ImmutableCFOptions::ImmutableCFOptions(const ColumnFamilyOptions& cf_options)
       num_levels(cf_options.num_levels),
       optimize_filters_for_hits(cf_options.optimize_filters_for_hits),
       force_consistency_checks(cf_options.force_consistency_checks),
+      disallow_memtable_writes(cf_options.disallow_memtable_writes),
       default_temperature(cf_options.default_temperature),
       memtable_insert_with_hint_prefix_extractor(
           cf_options.memtable_insert_with_hint_prefix_extractor),

options/cf_options.h

@@ -68,6 +68,8 @@ struct ImmutableCFOptions {
 
   bool force_consistency_checks;
 
+  bool disallow_memtable_writes;
+
   Temperature default_temperature;
 
   std::shared_ptr<const SliceTransform>

options/db_options.cc

@@ -141,6 +141,7 @@ static std::unordered_map<std::string, OptionTypeInfo>
           std::shared_ptr<Statistics> statistics;
           std::vector<DbPath> db_paths;
           FileTypeSet checksum_handoff_file_types;
+          CompactionStyleSet calculate_sst_write_lifetime_hint_set;
          */
         {"advise_random_on_open",
          {offsetof(struct ImmutableDBOptions, advise_random_on_open),
@@ -801,7 +802,9 @@ ImmutableDBOptions::ImmutableDBOptions(const DBOptions& options)
       follower_catchup_retry_count(options.follower_catchup_retry_count),
       follower_catchup_retry_wait_ms(options.follower_catchup_retry_wait_ms),
       metadata_write_temperature(options.metadata_write_temperature),
-      wal_write_temperature(options.wal_write_temperature) {
+      wal_write_temperature(options.wal_write_temperature),
+      calculate_sst_write_lifetime_hint_set(
+          options.calculate_sst_write_lifetime_hint_set) {
   fs = env->GetFileSystem();
   clock = env->GetSystemClock().get();
   logger = info_log.get();

options/db_options.h

@@ -107,6 +107,7 @@ struct ImmutableDBOptions {
   uint64_t follower_catchup_retry_wait_ms;
   Temperature metadata_write_temperature;
   Temperature wal_write_temperature;
+  CompactionStyleSet calculate_sst_write_lifetime_hint_set;
 
   // Beginning convenience/helper objects that are not part of the base
   // DBOptions

options/options.cc

@@ -90,6 +90,7 @@ AdvancedColumnFamilyOptions::AdvancedColumnFamilyOptions(const Options& options)
       paranoid_file_checks(options.paranoid_file_checks),
       force_consistency_checks(options.force_consistency_checks),
       report_bg_io_stats(options.report_bg_io_stats),
+      disallow_memtable_writes(options.disallow_memtable_writes),
       ttl(options.ttl),
       periodic_compaction_seconds(options.periodic_compaction_seconds),
       sample_for_compression(options.sample_for_compression),
@@ -395,6 +396,8 @@ void ColumnFamilyOptions::Dump(Logger* log) const {
                    force_consistency_checks);
   ROCKS_LOG_HEADER(log, "               Options.report_bg_io_stats: %d",
                    report_bg_io_stats);
+  ROCKS_LOG_HEADER(log, "               Options.disallow_memtable_writes: %d",
+                   disallow_memtable_writes);
   ROCKS_LOG_HEADER(log, "                              Options.ttl: %" PRIu64,
                    ttl);
   ROCKS_LOG_HEADER(log,

options/options_helper.cc

@@ -199,6 +199,8 @@ void BuildDBOptions(const ImmutableDBOptions& immutable_db_options,
       immutable_db_options.metadata_write_temperature;
   options.wal_write_temperature = immutable_db_options.wal_write_temperature;
   options.compaction_service = immutable_db_options.compaction_service;
+  options.calculate_sst_write_lifetime_hint_set =
+      immutable_db_options.calculate_sst_write_lifetime_hint_set;
 }
 
 ColumnFamilyOptions BuildColumnFamilyOptions(
@@ -326,6 +328,7 @@ void UpdateColumnFamilyOptions(const ImmutableCFOptions& ioptions,
   cf_opts->num_levels = ioptions.num_levels;
   cf_opts->optimize_filters_for_hits = ioptions.optimize_filters_for_hits;
   cf_opts->force_consistency_checks = ioptions.force_consistency_checks;
+  cf_opts->disallow_memtable_writes = ioptions.disallow_memtable_writes;
   cf_opts->memtable_insert_with_hint_prefix_extractor =
       ioptions.memtable_insert_with_hint_prefix_extractor;
   cf_opts->cf_paths = ioptions.cf_paths;
@@ -384,38 +387,47 @@ std::unordered_map<std::string, CompressionType>
         {"kZSTD", kZSTD},
         {"kDisableCompressionOption", kDisableCompressionOption}};
 
-std::vector<CompressionType> GetSupportedCompressions() {
-  // std::set internally to deduplicate potential name aliases
-  std::set<CompressionType> supported_compressions;
-  for (const auto& comp_to_name : OptionsHelper::compression_type_string_map) {
-    CompressionType t = comp_to_name.second;
-    if (t != kDisableCompressionOption && CompressionTypeSupported(t)) {
-      supported_compressions.insert(t);
+const std::vector<CompressionType>& GetSupportedCompressions() {
+  static std::vector<CompressionType> supported_compressions = []() {
+    // std::set internally to deduplicate potential name aliases
+    std::set<CompressionType> comp_set;
+    for (const auto& comp_to_name :
+         OptionsHelper::compression_type_string_map) {
+      CompressionType t = comp_to_name.second;
+      if (t != kDisableCompressionOption && CompressionTypeSupported(t)) {
+        comp_set.insert(t);
+      }
     }
-  }
-  return std::vector<CompressionType>(supported_compressions.begin(),
-                                      supported_compressions.end());
+    return std::vector<CompressionType>(comp_set.begin(), comp_set.end());
+  }();
+  return supported_compressions;
 }
 
-std::vector<CompressionType> GetSupportedDictCompressions() {
-  std::set<CompressionType> dict_compression_types;
-  for (const auto& comp_to_name : OptionsHelper::compression_type_string_map) {
-    CompressionType t = comp_to_name.second;
-    if (t != kDisableCompressionOption && DictCompressionTypeSupported(t)) {
-      dict_compression_types.insert(t);
+const std::vector<CompressionType>& GetSupportedDictCompressions() {
+  static std::vector<CompressionType> supported_dict_compressions = []() {
+    std::set<CompressionType> comp_set;
+    for (const auto& comp_to_name :
+         OptionsHelper::compression_type_string_map) {
+      CompressionType t = comp_to_name.second;
+      if (t != kDisableCompressionOption && DictCompressionTypeSupported(t)) {
+        comp_set.insert(t);
+      }
     }
-  }
-  return std::vector<CompressionType>(dict_compression_types.begin(),
-                                      dict_compression_types.end());
+    return std::vector<CompressionType>(comp_set.begin(), comp_set.end());
+  }();
+  return supported_dict_compressions;
 }
 
-std::vector<ChecksumType> GetSupportedChecksums() {
-  std::set<ChecksumType> checksum_types;
-  for (const auto& e : OptionsHelper::checksum_type_string_map) {
-    checksum_types.insert(e.second);
-  }
-  return std::vector<ChecksumType>(checksum_types.begin(),
-                                   checksum_types.end());
+const std::vector<ChecksumType>& GetSupportedChecksums() {
+  static std::vector<ChecksumType> supported_checksums = []() {
+    std::set<ChecksumType> checksum_types;
+    for (const auto& e : OptionsHelper::checksum_type_string_map) {
+      checksum_types.insert(e.second);
+    }
+    return std::vector<ChecksumType>(checksum_types.begin(),
+                                     checksum_types.end());
+  }();
+  return supported_checksums;
 }
 
 static bool ParseOptionHelper(void* opt_address, const OptionType& opt_type,