Update HISTORY for 9.3.2

Summary:
The `PessimisticTransaction::SetName()` code checks for an existing txn of the given name before registering the new txn. However, this is not atomic, which could result in a race condition if two txns try to register with the same name. Both might succeed and lead to unpredictable behavior. This PR makes the test and set atomic.

Pull Request resolved: https://github.com/facebook/rocksdb/pull/12895

Reviewed By: pdillinger

Differential Revision: D60460482

Pulled By: anand1976

fbshipit-source-id: e8afeb2356e1b8f4e8df785cb73532739f82579d

.github/actions/windows-build-steps/action.yml

@@ -15,30 +15,40 @@ runs:
       SNAPPY_INCLUDE: ${{ github.workspace }}/thirdparty/snappy-1.1.8;${{ github.workspace }}/thirdparty/snappy-1.1.8/build
       SNAPPY_LIB_DEBUG: ${{ github.workspace }}/thirdparty/snappy-1.1.8/build/Debug/snappy.lib
     run: |-
+      # NOTE: if ... Exit $LASTEXITCODE lines needed to exit and report failure
       echo ===================== Install Dependencies =====================
       choco install liberica8jdk -y
+      if(!$?) { Exit $LASTEXITCODE }
       mkdir $Env:THIRDPARTY_HOME
       cd $Env:THIRDPARTY_HOME
       echo "Building Snappy dependency..."
       curl -Lo snappy-1.1.8.zip https://github.com/google/snappy/archive/refs/tags/1.1.8.zip
+      if(!$?) { Exit $LASTEXITCODE }
       unzip -q snappy-1.1.8.zip
+      if(!$?) { Exit $LASTEXITCODE }
       cd snappy-1.1.8
       mkdir build
       cd build
       & cmake -G "$Env:CMAKE_GENERATOR" ..
+      if(!$?) { Exit $LASTEXITCODE }
       msbuild Snappy.sln -maxCpuCount -property:Configuration=Debug -property:Platform=x64
+      if(!$?) { Exit $LASTEXITCODE }
       echo ======================== Build RocksDB =========================
       cd ${{ github.workspace }}
       $env:Path = $env:JAVA_HOME + ";" + $env:Path
       mkdir build
       cd build
       & cmake -G "$Env:CMAKE_GENERATOR" -DCMAKE_BUILD_TYPE=Debug -DOPTDBG=1 -DPORTABLE="$Env:CMAKE_PORTABLE" -DSNAPPY=1 -DJNI=1 ..
+      if(!$?) { Exit $LASTEXITCODE }
       cd ..
       echo "Building with VS version: $Env:CMAKE_GENERATOR"
       msbuild build/rocksdb.sln -maxCpuCount -property:Configuration=Debug -property:Platform=x64
+      if(!$?) { Exit $LASTEXITCODE }
       echo ========================= Test RocksDB =========================
       build_tools\run_ci_db_test.ps1 -SuiteRun arena_test,db_basic_test,db_test,db_test2,db_merge_operand_test,bloom_test,c_test,coding_test,crc32c_test,dynamic_bloom_test,env_basic_test,env_test,hash_test,random_test -Concurrency 16
+      if(!$?) { Exit $LASTEXITCODE }
       echo ======================== Test RocksJava ========================
       cd build\java
       & ctest -C Debug -j 16
+      if(!$?) { Exit $LASTEXITCODE }
     shell: pwsh

.github/workflows/nightly.yml

@@ -96,3 +96,13 @@ jobs:
     steps:
     - uses: actions/checkout@v4.1.0
     - uses: "./.github/actions/windows-build-steps"
+  build-linux-arm-test-full:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 4-core-ubuntu-arm
+    steps:
+      - uses: actions/checkout@v4.1.0
+      - uses: "./.github/actions/pre-steps"
+      - run: sudo apt-get update && sudo apt-get install -y build-essential libgflags-dev
+      - run: make V=1 J=4 -j4 check
+      - uses: "./.github/actions/post-steps"

.github/workflows/pr-jobs.yml

@@ -607,3 +607,13 @@ jobs:
       with:
         name: maven-site
         path: "${{ github.workspace }}/java/target/site"
+  build-linux-arm:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 4-core-ubuntu-arm
+    steps:
+      - uses: actions/checkout@v4.1.0
+      - uses: "./.github/actions/pre-steps"
+      - run: sudo apt-get update && sudo apt-get install -y build-essential
+      - run: ROCKSDBTESTS_PLATFORM_DEPENDENT=only make V=1 J=4 -j4 all_but_some_tests check_some
+      - uses: "./.github/actions/post-steps"

CMakeLists.txt

@@ -44,6 +44,29 @@ project(rocksdb
   HOMEPAGE_URL https://rocksdb.org/
   LANGUAGES CXX C ASM)
 
+if(APPLE)
+  # On macOS Cmake, when cross-compiling, sometimes CMAKE_SYSTEM_PROCESSOR wrongfully stays
+  # the same as CMAKE_HOST_SYSTEM_PROCESSOR regardless the target CPU.
+  # The manual call to  set(CMAKE_SYSTEM_PROCESSOR) has to be set after the project() call.
+  # because project() might reset CMAKE_SYSTEM_PROCESSOR back to the value of CMAKE_HOST_SYSTEM_PROCESSOR.
+  # Check if CMAKE_SYSTEM_PROCESSOR is not equal to CMAKE_OSX_ARCHITECTURES
+  if(NOT CMAKE_OSX_ARCHITECTURES STREQUAL "")
+    if(NOT CMAKE_SYSTEM_PROCESSOR STREQUAL CMAKE_OSX_ARCHITECTURES)
+      # Split CMAKE_OSX_ARCHITECTURES into a list
+      string(REPLACE ";" " " ARCH_LIST ${CMAKE_OSX_ARCHITECTURES})
+      separate_arguments(ARCH_LIST UNIX_COMMAND ${ARCH_LIST})
+      # Count the number of architectures
+      list(LENGTH ARCH_LIST ARCH_COUNT)
+      # Ensure that exactly one architecture is specified
+      if(NOT ARCH_COUNT EQUAL 1)
+          message(FATAL_ERROR "CMAKE_OSX_ARCHITECTURES must have exactly one value. Current value: ${CMAKE_OSX_ARCHITECTURES}")
+      endif()
+      set(CMAKE_SYSTEM_PROCESSOR ${CMAKE_OSX_ARCHITECTURES})
+      message(STATUS "CMAKE_SYSTEM_PROCESSOR is manually set to ${CMAKE_SYSTEM_PROCESSOR}")
+    endif()
+  endif()
+endif()
+
 if(POLICY CMP0042)
   cmake_policy(SET CMP0042 NEW)
 endif()
@@ -182,7 +205,7 @@ endif()
 
 if(MSVC)
   set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /Zi /nologo /EHsc /GS /Gd /GR /GF /fp:precise /Zc:wchar_t /Zc:forScope /errorReport:queue")
-  set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /FC /d2Zi+ /W4 /wd4127 /wd4800 /wd4996 /wd4351 /wd4100 /wd4204 /wd4324")
+  set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /FC /d2Zi+ /W4 /wd4127 /wd4996 /wd4100 /wd4324")
 else()
   set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -W -Wextra -Wall -pthread")
   set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wsign-compare -Wshadow -Wno-unused-parameter -Wno-unused-variable -Woverloaded-virtual -Wnon-virtual-dtor -Wno-missing-field-initializers -Wno-strict-aliasing -Wno-invalid-offsetof")
@@ -191,6 +214,7 @@ else()
   endif()
   if(MINGW)
     set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wno-format")
+    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wa,-mbig-obj")
     add_definitions(-D_POSIX_C_SOURCE=1)
   endif()
   if(NOT CMAKE_BUILD_TYPE STREQUAL "Debug")
@@ -629,6 +653,7 @@ set(SOURCES
         cache/sharded_cache.cc
         cache/tiered_secondary_cache.cc
         db/arena_wrapped_db_iter.cc
+        db/attribute_group_iterator_impl.cc
         db/blob/blob_contents.cc
         db/blob/blob_fetcher.cc
         db/blob/blob_file_addition.cc
@@ -645,6 +670,7 @@ set(SOURCES
         db/blob/prefetch_buffer_collection.cc
         db/builder.cc
         db/c.cc
+        db/coalescing_iterator.cc
         db/column_family.cc
         db/compaction/compaction.cc
         db/compaction/compaction_iterator.cc
@@ -929,6 +955,7 @@ set(SOURCES
         utilities/transactions/write_prepared_txn_db.cc
         utilities/transactions/write_unprepared_txn.cc
         utilities/transactions/write_unprepared_txn_db.cc
+        utilities/types_util.cc
         utilities/ttl/db_ttl_impl.cc
         utilities/wal_filter.cc
         utilities/write_batch_with_index/write_batch_with_index.cc
@@ -1010,8 +1037,10 @@ endif()
 
 else()
   list(APPEND SOURCES
+    db/db_impl/db_impl_follower.cc
     port/port_posix.cc
     env/env_posix.cc
+    env/fs_on_demand.cc
     env/fs_posix.cc
     env/io_posix.cc)
 endif()
@@ -1336,6 +1365,7 @@ if(WITH_TESTS)
         db/file_indexer_test.cc
         db/filename_test.cc
         db/flush_job_test.cc
+        db/db_follower_test.cc
         db/import_column_family_test.cc
         db/listener_test.cc
         db/log_test.cc
@@ -1343,6 +1373,7 @@ if(WITH_TESTS)
         db/memtable_list_test.cc
         db/merge_helper_test.cc
         db/merge_test.cc
+        db/multi_cf_iterator_test.cc
         db/options_file_test.cc
         db/perf_context_test.cc
         db/periodic_task_scheduler_test.cc
@@ -1458,6 +1489,7 @@ if(WITH_TESTS)
         utilities/transactions/lock/range/range_locking_test.cc
         utilities/transactions/timestamped_snapshot_test.cc
         utilities/ttl/ttl_test.cc
+        utilities/types_util_test.cc
         utilities/util_merge_operators_test.cc
         utilities/write_batch_with_index/write_batch_with_index_test.cc
 	${PLUGIN_TESTS}

HISTORY.md

@@ -1,6 +1,153 @@
 # Rocksdb Change Log
 > NOTE: Entries for next release do not go here. Follow instructions in `unreleased_history/README.txt`
 
+## 9.3.2 (08/02/2024)
+### Bug Fixes
+* *Make DestroyDB supports slow deletion when it's configured in `SstFileManager`. The slow deletion is subject to the configured `rate_bytes_per_sec`, but not subject to the `max_trash_db_ratio`.
+* Fix a race condition in pessimistic transactions that could allow multiple transactions with the same name to be registered simultaneously, resulting in a crash or other unpredictable behavior.
+
+## 9.3.1 (05/25/2024)
+### Bug Fixes
+* [internal only] Build script improvement
+
+## 9.3.0 (05/17/2024)
+### New Features
+* Optimistic transactions and pessimistic transactions with the WriteCommitted policy now support the `GetEntity` API.
+* Added new `Iterator` property, "rocksdb.iterator.is-value-pinned", for checking whether the `Slice` returned by `Iterator::value()` can be used until the `Iterator` is destroyed.
+* Optimistic transactions and WriteCommitted pessimistic transactions now support the `MultiGetEntity` API.
+* Optimistic transactions and pessimistic transactions with the WriteCommitted policy now support the `PutEntity` API. Support for read APIs and other write policies (WritePrepared, WriteUnprepared) will be added later.
+
+### Public API Changes
+* Exposed block based metadata cache options via C API
+* Exposed compaction pri via c api.
+* Add a kAdmPolicyAllowAll option to TieredAdmissionPolicy that admits all blocks evicted from the primary block cache into the compressed secondary cache.
+
+### Behavior Changes
+* CompactRange() with change_level=true on a CF with FIFO compaction will return Status::NotSupported().
+* External file ingestion with FIFO compaction will always ingest to L0.
+
+### Bug Fixes
+* Fixed a bug for databases using `DBOptions::allow_2pc == true` (all `TransactionDB`s except `OptimisticTransactionDB`) that have exactly one column family. Due to a missing WAL sync, attempting to open the DB could have returned a `Status::Corruption` with a message like "SST file is ahead of WALs".
+* Fix a bug in CreateColumnFamilyWithImport() where if multiple CFs are imported, we were not resetting files' epoch number and L0 files can have overlapping key range but the same epoch number.
+* Fixed race conditions when `ColumnFamilyOptions::inplace_update_support == true` between user overwrites and reads on the same key.
+* Fix a bug where `CompactFiles()` can compact files of range conflict with other ongoing compactions' when `preclude_last_level_data_seconds > 0` is used
+* Fixed a false positive `Status::Corruption` reported when reopening a DB that used `DBOptions::recycle_log_file_num > 0` and `DBOptions::wal_compression != kNoCompression`.
+* While WAL is locked with LockWAL(), some operations like Flush() and IngestExternalFile() are now blocked as they should have been.
+* Fixed a bug causing stale memory access when using the TieredSecondaryCache with an NVM secondary cache, and a file system that supports return an FS allocated buffer for MultiRead (FSSupportedOps::kFSBuffer is set).
+
+## 9.2.0 (05/01/2024)
+### New Features
+* Added two options `deadline` and `max_size_bytes` for CacheDumper to exit early
+* Added a new API `GetEntityFromBatchAndDB` to `WriteBatchWithIndex` that can be used for wide-column point lookups with read-your-own-writes consistency. Similarly to `GetFromBatchAndDB`, the API can combine data from the write batch with data from the underlying database if needed. See the API comments for more details.
+* [Experimental] Introduce two new cross-column-family iterators - CoalescingIterator and AttributeGroupIterator. The CoalescingIterator enables users to iterate over multiple column families and access their values and columns. During this iteration, if the same key exists in more than one column family, the keys in the later column family will overshadow the previous ones. The AttributeGroupIterator allows users to gather wide columns per Column Family and create attribute groups while iterating over keys across all CFs.
+* Added a new API `MultiGetEntityFromBatchAndDB` to `WriteBatchWithIndex` that can be used for batched wide-column point lookups with read-your-own-writes consistency. Similarly to `MultiGetFromBatchAndDB`, the API can combine data from the write batch with data from the underlying database if needed. See the API comments for more details.
+* *Adds a `SstFileReader::NewTableIterator` API to support programmatically read a SST file as a raw table file.
+* Add an option to `WaitForCompactOptions` - `wait_for_purge` to make `WaitForCompact()` API wait for background purge to complete
+
+### Public API Changes
+* DeleteRange() will return NotSupported() if row_cache is configured since they don't work together in some cases.
+* Deprecated `CompactionOptions::compression` since `CompactionOptions`'s API for configuring compression was incomplete, unsafe, and likely unnecessary
+* Using `OptionChangeMigration()` to migrate from non-FIFO to FIFO compaction
+with `Options::compaction_options_fifo.max_table_files_size` > 0 can cause
+the whole DB to be dropped right after migration if the migrated data is larger than
+`max_table_files_size`
+
+### Behavior Changes
+* Enabling `BlockBasedTableOptions::block_align` is now incompatible (i.e., APIs will return `Status::InvalidArgument`) with more ways of enabling compression: `CompactionOptions::compression`, `ColumnFamilyOptions::compression_per_level`, and `ColumnFamilyOptions::bottommost_compression`.
+* Changed the default value of `CompactionOptions::compression` to `kDisableCompressionOption`, which means the compression type is determined by the `ColumnFamilyOptions`.
+* `BlockBasedTableOptions::optimize_filters_for_memory` is now set to true by default. When `partition_filters=false`, this could lead to somewhat increased average RSS memory usage by the block cache, but this "extra" usage is within the allowed memory budget and should make memory usage more consistent (by minimizing internal fragmentation for more kinds of blocks).
+* Dump all keys for cache dumper impl if `SetDumpFilter()` is not called
+* `CompactRange()` with `CompactRangeOptions::change_level = true` and `CompactRangeOptions::target_level = 0` that ends up moving more than 1 file from non-L0 to L0 will return `Status::Aborted()`.
+* On distributed file systems that support file system level checksum verification and reconstruction reads, RocksDB will now retry a file read if the initial read fails RocksDB block level or record level checksum verification. This applies to MANIFEST file reads when the DB is opened, and to SST file reads at all times.
+
+### Bug Fixes
+* Fix a bug causing `VerifyFileChecksums()` to return false-positive corruption under `BlockBasedTableOptions::block_align=true`
+* Provide consistent view of the database across the column families for `NewIterators()` API.
+* Fixed feature interaction bug for `DeleteRange()` together with `ColumnFamilyOptions::memtable_insert_with_hint_prefix_extractor`. The impact of this bug would likely be corruption or crashing.
+* Fixed hang in `DisableManualCompactions()` where compactions waiting to be scheduled due to conflicts would not be canceled promptly
+* Fixed a regression when `ColumnFamilyOptions::max_successive_merges > 0` where the CPU overhead for deciding whether to merge could have increased unless the user had set the option `ColumnFamilyOptions::strict_max_successive_merges`
+* Fixed a bug in `MultiGet()` and `MultiGetEntity()` together with blob files (`ColumnFamilyOptions::enable_blob_files == true`). An error looking up one of the keys could cause the results to be wrong for other keys for which the statuses were `Status::OK`.
+* Fixed a bug where wrong padded bytes are used to generate file checksum and `DataVerificationInfo::checksum` upon file creation
+* Correctly implemented the move semantics of `PinnableWideColumns`.
+* Fixed a bug when the recycle_log_file_num in DBOptions is changed from 0 to non-zero when a DB is reopened. On a subsequent reopen, if a log file created when recycle_log_file_num==0 was reused previously, is alive and is empty, we could end up inserting stale WAL records into the memtable.
+* *Fix a bug where obsolete files' deletion during DB::Open are not rate limited with `SstFilemManager`'s slow deletion feature even if it's configured.
+
+## 9.1.0 (03/22/2024)
+### New Features
+* Added an option, `GetMergeOperandsOptions::continue_cb`, to give users the ability to end `GetMergeOperands()`'s lookup process before all merge operands were found.
+* *Add sanity checks for ingesting external files that currently checks if the user key comparator used to create the file is compatible with the column family's user key comparator.
+*Support ingesting external files for column family that has user-defined timestamps in memtable only enabled.
+* On file systems that support storage level data checksum and reconstruction, retry SST block reads for point lookups, scans, and flush and compaction if there's a checksum mismatch on the initial read.
+* Some enhancements and fixes to experimental Temperature handling features, including new `default_write_temperature` CF option and opening an `SstFileWriter` with a temperature.
+* `WriteBatchWithIndex` now supports wide-column point lookups via the `GetEntityFromBatch` API. See the API comments for more details.
+* *Implement experimental features: API `Iterator::GetProperty("rocksdb.iterator.write-time")` to allow users to get data's approximate write unix time and write data with a specific write time via `WriteBatch::TimedPut` API.
+
+### Public API Changes
+* Best-effort recovery (`best_efforts_recovery == true`) may now be used together with atomic flush (`atomic_flush == true`). The all-or-nothing recovery guarantee for atomically flushed data will be upheld.
+* Remove deprecated option `bottommost_temperature`, already replaced by `last_level_temperature`
+* Added new PerfContext counters for block cache bytes read - block_cache_index_read_byte, block_cache_filter_read_byte, block_cache_compression_dict_read_byte, and block_cache_read_byte.
+* Deprecate experimental Remote Compaction APIs - StartV2() and WaitForCompleteV2() and introduce Schedule() and Wait(). The new APIs essentially does the same thing as the old APIs. They allow taking externally generated unique id to wait for remote compaction to complete.
+* *For API `WriteCommittedTransaction::GetForUpdate`, if the column family enables user-defined timestamp, it was mandated that argument `do_validate` cannot be false, and UDT based validation has to be done with a user set read timestamp. It's updated to make the UDT based validation optional if user sets `do_validate` to false and does not set a read timestamp. With this, `GetForUpdate` skips UDT based validation and it's users' responsibility to enforce the UDT invariant. SO DO NOT skip this UDT-based validation if users do not have ways to enforce the UDT invariant. Ways to enforce the invariant on the users side include manage a monotonically increasing timestamp, commit transactions in a single thread etc.
+* Defined a new PerfLevel `kEnableWait` to measure time spent by user threads blocked in RocksDB other than mutex, such as a write thread waiting to be added to a write group, a write thread delayed or stalled etc.
+* `RateLimiter`'s API no longer requires the burst size to be the refill size. Users of `NewGenericRateLimiter()` can now provide burst size in `single_burst_bytes`. Implementors of `RateLimiter::SetSingleBurstBytes()` need to adapt their implementations to match the changed API doc.
+* Add `write_memtable_time` to the newly introduced PerfLevel `kEnableWait`.
+
+### Behavior Changes
+* `RateLimiter`s created by `NewGenericRateLimiter()` no longer modify the refill period when `SetSingleBurstBytes()` is called.
+* Merge writes will only keep merge operand count within `ColumnFamilyOptions::max_successive_merges` when the key's merge operands are all found in memory, unless `strict_max_successive_merges` is explicitly set.
+
+### Bug Fixes
+* Fixed `kBlockCacheTier` reads to return `Status::Incomplete` when I/O is needed to fetch a merge chain's base value from a blob file.
+* Fixed `kBlockCacheTier` reads to return `Status::Incomplete` on table cache miss rather than incorrectly returning an empty value.
+* Fixed a data race in WalManager that may affect how frequent PurgeObsoleteWALFiles() runs.
+* Re-enable the recycle_log_file_num option in DBOptions for kPointInTimeRecovery WAL recovery mode, which was previously disabled due to a bug in the recovery logic. This option is incompatible with WriteOptions::disableWAL. A Status::InvalidArgument() will be returned if disableWAL is specified.
+
+### Performance Improvements
+* Java API `multiGet()` variants now take advantage of the underlying batched `multiGet()` performance improvements.
+Before
+```
+Benchmark (columnFamilyTestType) (keyCount) (keySize) (multiGetSize) (valueSize) Mode Cnt Score Error Units
+MultiGetBenchmarks.multiGetList10 no_column_family 10000 16 100 64 thrpt 25 6315.541 ± 8.106 ops/s
+MultiGetBenchmarks.multiGetList10 no_column_family 10000 16 100 1024 thrpt 25 6975.468 ± 68.964 ops/s
+```
+After
+```
+Benchmark (columnFamilyTestType) (keyCount) (keySize) (multiGetSize) (valueSize) Mode Cnt Score Error Units
+MultiGetBenchmarks.multiGetList10 no_column_family 10000 16 100 64 thrpt 25 7046.739 ± 13.299 ops/s
+MultiGetBenchmarks.multiGetList10 no_column_family 10000 16 100 1024 thrpt 25 7654.521 ± 60.121 ops/s
+```
+
+## 9.0.0 (02/16/2024)
+### New Features
+* Provide support for FSBuffer for point lookups. Also added support for scans and compactions that don't go through prefetching.
+* *Make `SstFileWriter` create SST files without persisting user defined timestamps when the `Option.persist_user_defined_timestamps` flag is set to false.
+* Add support for user-defined timestamps in APIs `DeleteFilesInRanges` and `GetPropertiesOfTablesInRange`.
+* Mark wal\_compression feature as production-ready. Currently only compatible with ZSTD compression.
+
+### Public API Changes
+* Allow setting Stderr logger via C API
+* Declare one Get and one MultiGet variant as pure virtual, and make all the other variants non-overridable. The methods required to be implemented by derived classes of DB allow returning timestamps. It is up to the implementation to check and return an error if timestamps are not supported. The non-batched MultiGet APIs are reimplemented in terms of batched MultiGet, so callers might see a performance improvement.
+* Exposed mode option to Rate Limiter via c api.
+* Removed deprecated option `access_hint_on_compaction_start`
+* Removed deprecated option `ColumnFamilyOptions::check_flush_compaction_key_order`
+* *Remove the default `WritableFile::GetFileSize` and `FSWritableFile::GetFileSize` implementation that returns 0 and make it pure virtual, so that subclasses are enforced to explicitly provide an implementation.
+* Removed deprecated option `ColumnFamilyOptions::level_compaction_dynamic_file_size`
+* *Removed tickers with typos "rocksdb.error.handler.bg.errro.count", "rocksdb.error.handler.bg.io.errro.count", "rocksdb.error.handler.bg.retryable.io.errro.count".
+* Remove the force mode for `EnableFileDeletions` API because it is unsafe with no known legitimate use.
+* Removed deprecated option `ColumnFamilyOptions::ignore_max_compaction_bytes_for_input`
+* `sst_dump --command=check` now compares the number of records in a table with `num_entries` in table property, and reports corruption if there is a mismatch. API `SstFileDumper::ReadSequential()` is updated to optionally do this verification. (#12322)
+
+### Behavior Changes
+* format\_version=6 is the new default setting in BlockBasedTableOptions, for more robust data integrity checking. DBs and SST files written with this setting cannot be read by RocksDB versions before 8.6.0.
+* Compactions can be scheduled in parallel in an additional scenario: multiple files are marked for compaction within a single column family
+* For leveled compaction, RocksDB will try to do intra-L0 compaction if the total L0 size is small compared to Lbase (#12214). Users with atomic_flush=true are more likely to see the impact of this change.
+
+### Bug Fixes
+* Fixed a data race in `DBImpl::RenameTempFileToOptionsFile`.
+* Fix some perf context statistics error in write steps. which include missing write_memtable_time in unordered_write. missing write_memtable_time in PipelineWrite when Writer stat is STATE_PARALLEL_MEMTABLE_WRITER. missing write_delay_time when calling DelayWrite in WriteImplWALOnly function.
+* Fixed a bug that can, under rare circumstances, cause MultiGet to return an incorrect result for a duplicate key in a MultiGet batch.
+* Fix a bug where older data of an ingested key can be returned for read when universal compaction is used
+
 ## 8.11.0 (01/19/2024)
 ### New Features
 * Add new statistics: `rocksdb.sst.write.micros` measures time of each write to SST file; `rocksdb.file.write.{flush|compaction|db.open}.micros` measure time of each write to SST table (currently only block-based table format) and blob file for flush, compaction and db open.

INSTALL.md

@@ -126,7 +126,6 @@ most processors made since roughly 2013.
         * Update XCode:  run `xcode-select --install` (or install it from XCode App's settting).
         * Install via [homebrew](http://brew.sh/).
             * If you're first time developer in MacOS, you still need to run: `xcode-select --install` in your command line.
-            * run `brew tap homebrew/versions; brew install gcc7 --use-llvm` to install gcc 7 (or higher).
     * run `brew install rocksdb`
 
 * **FreeBSD** (11.01):

Makefile

@@ -1610,6 +1610,9 @@ object_registry_test: $(OBJ_DIR)/utilities/object_registry_test.o $(TEST_LIBRARY
 ttl_test: $(OBJ_DIR)/utilities/ttl/ttl_test.o $(TEST_LIBRARY) $(LIBRARY)
 	$(AM_LINK)
 
+types_util_test: $(OBJ_DIR)/utilities/types_util_test.o $(TEST_LIBRARY) $(LIBRARY)
+	$(AM_LINK)
+
 write_batch_with_index_test: $(OBJ_DIR)/utilities/write_batch_with_index/write_batch_with_index_test.o $(TEST_LIBRARY) $(LIBRARY)
 	$(AM_LINK)
 
@@ -1640,6 +1643,9 @@ wal_edit_test: $(OBJ_DIR)/db/wal_edit_test.o $(TEST_LIBRARY) $(LIBRARY)
 dbformat_test: $(OBJ_DIR)/db/dbformat_test.o $(TEST_LIBRARY) $(LIBRARY)
 	$(AM_LINK)
 
+multi_cf_iterator_test: $(OBJ_DIR)/db/multi_cf_iterator_test.o $(TEST_LIBRARY) $(LIBRARY)
+	$(AM_LINK)
+
 env_basic_test: $(OBJ_DIR)/env/env_basic_test.o $(TEST_LIBRARY) $(LIBRARY)
 	$(AM_LINK)
 
@@ -1916,6 +1922,9 @@ sst_file_reader_test: $(OBJ_DIR)/table/sst_file_reader_test.o $(TEST_LIBRARY) $(
 db_secondary_test: $(OBJ_DIR)/db/db_secondary_test.o $(TEST_LIBRARY) $(LIBRARY)
 	$(AM_LINK)
 
+db_follower_test: $(OBJ_DIR)/db/db_follower_test.o $(TEST_LIBRARY) $(LIBRARY)
+	$(AM_LINK)
+
 block_cache_tracer_test: $(OBJ_DIR)/trace_replay/block_cache_tracer_test.o $(TEST_LIBRARY) $(LIBRARY)
 	$(AM_LINK)
 
@@ -2332,47 +2341,47 @@ rocksdbjavastaticreleasedocker: rocksdbjavastaticosx rocksdbjavastaticdockerx86
 
 rocksdbjavastaticdockerx86:
 	mkdir -p java/target
-	docker run --rm --name rocksdb_linux_x86-be --platform linux/386 --attach stdin --attach stdout --attach stderr --volume $(HOME)/.m2:/root/.m2:ro --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) evolvedbinary/rocksjava:centos6_x86-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh
+	docker run --rm --name rocksdb_linux_x86-be --platform linux/386 --attach stdin --attach stdout --attach stderr --volume $(HOME)/.m2:/root/.m2:ro --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) --env J=$(J) evolvedbinary/rocksjava:centos6_x86-be /rocksdb-host/java/crossbuild/docker-build-linux.sh
 
 rocksdbjavastaticdockerx86_64:
 	mkdir -p java/target
-	docker run --rm --name rocksdb_linux_x64-be --attach stdin --attach stdout --attach stderr --volume $(HOME)/.m2:/root/.m2:ro --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) evolvedbinary/rocksjava:centos6_x64-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh
+	docker run --rm --name rocksdb_linux_x64-be --attach stdin --attach stdout --attach stderr --volume $(HOME)/.m2:/root/.m2:ro --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) --env J=$(J) evolvedbinary/rocksjava:centos6_x64-be /rocksdb-host/java/crossbuild/docker-build-linux.sh
 
 rocksdbjavastaticdockerppc64le:
 	mkdir -p java/target
-	docker run --rm --name rocksdb_linux_ppc64le-be --attach stdin --attach stdout --attach stderr --volume $(HOME)/.m2:/root/.m2:ro --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) evolvedbinary/rocksjava:centos7_ppc64le-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh
+	docker run --rm --name rocksdb_linux_ppc64le-be --attach stdin --attach stdout --attach stderr --volume $(HOME)/.m2:/root/.m2:ro --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) --env J=$(J) evolvedbinary/rocksjava:centos7_ppc64le-be /rocksdb-host/java/crossbuild/docker-build-linux.sh
 
 rocksdbjavastaticdockerarm64v8:
 	mkdir -p java/target
-	docker run --rm --name rocksdb_linux_arm64v8-be --attach stdin --attach stdout --attach stderr --volume $(HOME)/.m2:/root/.m2:ro --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) evolvedbinary/rocksjava:centos7_arm64v8-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh
+	docker run --rm --name rocksdb_linux_arm64v8-be --attach stdin --attach stdout --attach stderr --volume $(HOME)/.m2:/root/.m2:ro --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) --env J=$(J) evolvedbinary/rocksjava:centos7_arm64v8-be /rocksdb-host/java/crossbuild/docker-build-linux.sh
 
 rocksdbjavastaticdockers390x:
 	mkdir -p java/target
-	docker run --rm --name rocksdb_linux_s390x-be --attach stdin --attach stdout --attach stderr --volume $(HOME)/.m2:/root/.m2:ro --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) evolvedbinary/rocksjava:ubuntu18_s390x-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh
+	docker run --rm --name rocksdb_linux_s390x-be --attach stdin --attach stdout --attach stderr --volume $(HOME)/.m2:/root/.m2:ro --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) --env J=$(J) evolvedbinary/rocksjava:ubuntu18_s390x-be /rocksdb-host/java/crossbuild/docker-build-linux.sh
 
 rocksdbjavastaticdockerriscv64:
 	mkdir -p java/target
-	docker run --rm --name rocksdb_linux_riscv64-be --attach stdin --attach stdout --attach stderr --volume $(HOME)/.m2:/root/.m2:ro --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) evolvedbinary/rocksjava:ubuntu20_riscv64-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh
+	docker run --rm --name rocksdb_linux_riscv64-be --attach stdin --attach stdout --attach stderr --volume $(HOME)/.m2:/root/.m2:ro --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) --env J=$(J) evolvedbinary/rocksjava:ubuntu20_riscv64-be /rocksdb-host/java/crossbuild/docker-build-linux.sh
 
 rocksdbjavastaticdockerx86musl:
 	mkdir -p java/target
-	docker run --rm --name rocksdb_linux_x86-musl-be --platform linux/386 --attach stdin --attach stdout --attach stderr --volume $(HOME)/.m2:/root/.m2:ro --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) evolvedbinary/rocksjava:alpine3_x86-be /rocksdb-host/java/crossbuild/docker-build-linux-alpine.sh
+	docker run --rm --name rocksdb_linux_x86-musl-be --platform linux/386 --attach stdin --attach stdout --attach stderr --volume $(HOME)/.m2:/root/.m2:ro --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) --env J=$(J) evolvedbinary/rocksjava:alpine3_x86-be /rocksdb-host/java/crossbuild/docker-build-linux.sh
 
 rocksdbjavastaticdockerx86_64musl:
 	mkdir -p java/target
-	docker run --rm --name rocksdb_linux_x64-musl-be --attach stdin --attach stdout --attach stderr --volume $(HOME)/.m2:/root/.m2:ro --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) evolvedbinary/rocksjava:alpine3_x64-be /rocksdb-host/java/crossbuild/docker-build-linux-alpine.sh
+	docker run --rm --name rocksdb_linux_x64-musl-be --attach stdin --attach stdout --attach stderr --volume $(HOME)/.m2:/root/.m2:ro --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) --env J=$(J) evolvedbinary/rocksjava:alpine3_x64-be /rocksdb-host/java/crossbuild/docker-build-linux.sh
 
 rocksdbjavastaticdockerppc64lemusl:
 	mkdir -p java/target
-	docker run --rm --name rocksdb_linux_ppc64le-musl-be --attach stdin --attach stdout --attach stderr --volume $(HOME)/.m2:/root/.m2:ro --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) evolvedbinary/rocksjava:alpine3_ppc64le-be /rocksdb-host/java/crossbuild/docker-build-linux-alpine.sh
+	docker run --rm --name rocksdb_linux_ppc64le-musl-be --attach stdin --attach stdout --attach stderr --volume $(HOME)/.m2:/root/.m2:ro --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) --env J=$(J) evolvedbinary/rocksjava:alpine3_ppc64le-be /rocksdb-host/java/crossbuild/docker-build-linux.sh
 
 rocksdbjavastaticdockerarm64v8musl:
 	mkdir -p java/target
-	docker run --rm --name rocksdb_linux_arm64v8-musl-be --attach stdin --attach stdout --attach stderr --volume $(HOME)/.m2:/root/.m2:ro --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) evolvedbinary/rocksjava:alpine3_arm64v8-be /rocksdb-host/java/crossbuild/docker-build-linux-alpine.sh
+	docker run --rm --name rocksdb_linux_arm64v8-musl-be --attach stdin --attach stdout --attach stderr --volume $(HOME)/.m2:/root/.m2:ro --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) --env J=$(J) evolvedbinary/rocksjava:alpine3_arm64v8-be /rocksdb-host/java/crossbuild/docker-build-linux.sh
 
 rocksdbjavastaticdockers390xmusl:
 	mkdir -p java/target
-	docker run --rm --name rocksdb_linux_s390x-musl-be --attach stdin --attach stdout --attach stderr --volume $(HOME)/.m2:/root/.m2:ro --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) evolvedbinary/rocksjava:alpine3_s390x-be /rocksdb-host/java/crossbuild/docker-build-linux-alpine.sh
+	docker run --rm --name rocksdb_linux_s390x-musl-be --attach stdin --attach stdout --attach stderr --volume $(HOME)/.m2:/root/.m2:ro --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) --env J=$(J) evolvedbinary/rocksjava:alpine3_s390x-be /rocksdb-host/java/crossbuild/docker-build-linux.sh
 
 rocksdbjavastaticpublish: rocksdbjavastaticrelease rocksdbjavastaticpublishcentral
 

TARGETS

@@ -4,6 +4,7 @@
 # This file is a Facebook-specific integration for buck builds, so can
 # only be validated by Facebook employees.
 load("//rocks/buckifier:defs.bzl", "cpp_library_wrapper","rocks_cpp_library_wrapper","cpp_binary_wrapper","cpp_unittest_wrapper","fancy_bench_wrapper","add_c_test_wrapper")
+load("@fbcode_macros//build_defs:export_files.bzl", "export_file")
 
 
 cpp_library_wrapper(name="rocksdb_lib", srcs=[
@@ -21,6 +22,7 @@ cpp_library_wrapper(name="rocksdb_lib", srcs=[
         "cache/sharded_cache.cc",
         "cache/tiered_secondary_cache.cc",
         "db/arena_wrapped_db_iter.cc",
+        "db/attribute_group_iterator_impl.cc",
         "db/blob/blob_contents.cc",
         "db/blob/blob_fetcher.cc",
         "db/blob/blob_file_addition.cc",
@@ -37,6 +39,7 @@ cpp_library_wrapper(name="rocksdb_lib", srcs=[
         "db/blob/prefetch_buffer_collection.cc",
         "db/builder.cc",
         "db/c.cc",
+        "db/coalescing_iterator.cc",
         "db/column_family.cc",
         "db/compaction/compaction.cc",
         "db/compaction/compaction_iterator.cc",
@@ -58,6 +61,7 @@ cpp_library_wrapper(name="rocksdb_lib", srcs=[
         "db/db_impl/db_impl_debug.cc",
         "db/db_impl/db_impl_experimental.cc",
         "db/db_impl/db_impl_files.cc",
+        "db/db_impl/db_impl_follower.cc",
         "db/db_impl/db_impl_open.cc",
         "db/db_impl/db_impl_readonly.cc",
         "db/db_impl/db_impl_secondary.cc",
@@ -115,6 +119,7 @@ cpp_library_wrapper(name="rocksdb_lib", srcs=[
         "env/env_posix.cc",
         "env/file_system.cc",
         "env/file_system_tracer.cc",
+        "env/fs_on_demand.cc",
         "env/fs_posix.cc",
         "env/fs_remap.cc",
         "env/io_posix.cc",
@@ -345,6 +350,7 @@ cpp_library_wrapper(name="rocksdb_lib", srcs=[
         "utilities/transactions/write_unprepared_txn.cc",
         "utilities/transactions/write_unprepared_txn_db.cc",
         "utilities/ttl/db_ttl_impl.cc",
+        "utilities/types_util.cc",
         "utilities/wal_filter.cc",
         "utilities/write_batch_with_index/write_batch_with_index.cc",
         "utilities/write_batch_with_index/write_batch_with_index_internal.cc",
@@ -4792,6 +4798,12 @@ cpp_unittest_wrapper(name="db_flush_test",
             extra_compiler_flags=[])
 
 
+cpp_unittest_wrapper(name="db_follower_test",
+            srcs=["db/db_follower_test.cc"],
+            deps=[":rocksdb_test_lib"],
+            extra_compiler_flags=[])
+
+
 cpp_unittest_wrapper(name="db_inplace_update_test",
             srcs=["db/db_inplace_update_test.cc"],
             deps=[":rocksdb_test_lib"],
@@ -5226,6 +5238,12 @@ cpp_unittest_wrapper(name="mock_env_test",
             extra_compiler_flags=[])
 
 
+cpp_unittest_wrapper(name="multi_cf_iterator_test",
+            srcs=["db/multi_cf_iterator_test.cc"],
+            deps=[":rocksdb_test_lib"],
+            extra_compiler_flags=[])
+
+
 cpp_unittest_wrapper(name="object_registry_test",
             srcs=["utilities/object_registry_test.cc"],
             deps=[":rocksdb_test_lib"],
@@ -5520,6 +5538,12 @@ cpp_unittest_wrapper(name="ttl_test",
             extra_compiler_flags=[])
 
 
+cpp_unittest_wrapper(name="types_util_test",
+            srcs=["utilities/types_util_test.cc"],
+            deps=[":rocksdb_test_lib"],
+            extra_compiler_flags=[])
+
+
 cpp_unittest_wrapper(name="udt_util_test",
             srcs=["util/udt_util_test.cc"],
             deps=[":rocksdb_test_lib"],

buckifier/targets_cfg.py

@@ -7,6 +7,7 @@ rocksdb_target_header_template = """# This file \100generated by:
 # This file is a Facebook-specific integration for buck builds, so can
 # only be validated by Facebook employees.
 load("//rocks/buckifier:defs.bzl", "cpp_library_wrapper","rocks_cpp_library_wrapper","cpp_binary_wrapper","cpp_unittest_wrapper","fancy_bench_wrapper","add_c_test_wrapper")
+load("@fbcode_macros//build_defs:export_files.bzl", "export_file")
 
 """
 

build_tools/build_detect_platform

@@ -163,24 +163,6 @@ case "$TARGET_OS" in
             PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -latomic"
         fi
         PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lpthread -lrt -ldl"
-        if test -z "$ROCKSDB_USE_IO_URING"; then
-            ROCKSDB_USE_IO_URING=1
-        fi
-        if test "$ROCKSDB_USE_IO_URING" -ne 0; then
-            # check for liburing
-            $CXX $PLATFORM_CXXFLAGS -x c++ - -luring -o test.o 2>/dev/null  <<EOF
-              #include <liburing.h>
-              int main() {
-                struct io_uring ring;
-                io_uring_queue_init(1, &ring, 0);
-                return 0;
-              }
-EOF
-            if [ "$?" = 0 ]; then
-                PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -luring"
-                COMMON_FLAGS="$COMMON_FLAGS -DROCKSDB_IOURING_PRESENT"
-            fi
-        fi
         # PORT_FILES=port/linux/linux_specific.cc
         ;;
     SunOS)
@@ -614,6 +596,24 @@ EOF
         fi
     fi
 
+    if test -z "$ROCKSDB_USE_IO_URING"; then
+        ROCKSDB_USE_IO_URING=1
+    fi
+    if [ "$ROCKSDB_USE_IO_URING" -ne 0 -a "$PLATFORM" = OS_LINUX ]; then
+        # check for liburing
+        $CXX $PLATFORM_CXXFLAGS -x c++ - -luring -o test.o 2>/dev/null  <<EOF
+          #include <liburing.h>
+          int main() {
+            struct io_uring ring;
+            io_uring_queue_init(1, &ring, 0);
+            return 0;
+          }
+EOF
+        if [ "$?" = 0 ]; then
+            PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -luring"
+            COMMON_FLAGS="$COMMON_FLAGS -DROCKSDB_IOURING_PRESENT"
+        fi
+    fi
 fi
 
 # TODO(tec): Fix -Wshorten-64-to-32 errors on FreeBSD and enable the warning.

build_tools/dependencies_platform010.sh

@@ -1,6 +1,6 @@
 # Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved.
 # The file is generated using update_dependencies.sh.
-GCC_BASE=/mnt/gvfs/third-party2/gcc/b30c2e3be652b0135fb3df27fdef5eb4ffc2dc6d/11.x/centos9-native/886b5eb
+GCC_BASE=/mnt/gvfs/third-party2/gcc/62de5a92e5f23c661c3d4b9f322e04eb14e7a5bd/11.x/centos8-native/886b5eb
 CLANG_BASE=/mnt/gvfs/third-party2/llvm-fb/1f6edd1ff15c99c861afc8f3cd69054cd974dd64/15/platform010/72a2ff8
 LIBGCC_BASE=/mnt/gvfs/third-party2/libgcc/d1129753c8361ac8e9453c0f4291337a4507ebe6/11.x/platform010/5684a5a
 GLIBC_BASE=/mnt/gvfs/third-party2/glibc/fed6e93d87571fb162734c86636119d45a398963/2.34/platform010/f259413
@@ -17,6 +17,6 @@ TBB_BASE=/mnt/gvfs/third-party2/tbb/53953ebc4e3eda85ad6fc3e429ba146035e97b90/201
 LIBURING_BASE=/mnt/gvfs/third-party2/liburing/a98e2d137007e3ebf7f33bd6f99c2c56bdaf8488/20210212/platform010/76ebdda
 BENCHMARK_BASE=/mnt/gvfs/third-party2/benchmark/780c7a0f9cf0967961e69ad08e61cddd85d61821/trunk/platform010/76ebdda
 KERNEL_HEADERS_BASE=/mnt/gvfs/third-party2/kernel-headers/624a2f8f6c93c3c1df8aa4a6255d8202631a6c80/fb/platform010/da39a3e
-BINUTILS_BASE=/mnt/gvfs/third-party2/binutils/8dec50d6a10944a2e851f090a3872f9ba0b0b402/2.37/centos9-native/da39a3e
+BINUTILS_BASE=/mnt/gvfs/third-party2/binutils/39579e8603b48b3540f8b0633f43adf29acccb8b/2.37/centos8-native/da39a3e
 VALGRIND_BASE=/mnt/gvfs/third-party2/valgrind/cd9cc656d49ecb53797ce4d055e49fde29fd57ff/3.19.0/platform010/76ebdda
 LUA_BASE=/mnt/gvfs/third-party2/lua/363787fa5cac2a8aa20638909210443278fa138e/5.3.4/platform010/9079c97

build_tools/fbcode_config.sh

@@ -113,7 +113,7 @@ CLANG_LIB="$CLANG_BASE/lib"
 CLANG_SRC="$CLANG_BASE/../../src"
 
 CLANG_ANALYZER="$CLANG_BIN/clang++"
-CLANG_SCAN_BUILD="$CLANG_SRC/llvm/tools/clang/tools/scan-build/bin/scan-build"
+CLANG_SCAN_BUILD="$CLANG_BIN/scan-build
 
 if [ -z "$USE_CLANG" ]; then
   # gcc

build_tools/fbcode_config_platform010.sh

@@ -110,7 +110,7 @@ CLANG_LIB="$CLANG_BASE/lib"
 CLANG_SRC="$CLANG_BASE/../../src"
 
 CLANG_ANALYZER="$CLANG_BIN/clang++"
-CLANG_SCAN_BUILD="$CLANG_SRC/llvm/clang/tools/scan-build/bin/scan-build"
+CLANG_SCAN_BUILD="$CLANG_BIN/scan-build"
 
 if [ -z "$USE_CLANG" ]; then
   # gcc

build_tools/update_dependencies.sh

@@ -75,7 +75,7 @@ touch "$OUTPUT"
 echo "Writing dependencies to $OUTPUT"
 
 # Compilers locations
-GCC_BASE=`readlink -f $TP2_LATEST/gcc/11.x/centos9-native/*/`
+GCC_BASE=`readlink -f $TP2_LATEST/gcc/11.x/centos8-native/*/`
 CLANG_BASE=`readlink -f $TP2_LATEST/llvm-fb/15/platform010/*/`
 
 log_header
@@ -99,7 +99,7 @@ get_lib_base liburing   LATEST  platform010
 get_lib_base benchmark  LATEST  platform010
 
 get_lib_base kernel-headers fb platform010
-get_lib_base binutils   LATEST centos9-native
+get_lib_base binutils   LATEST centos8-native
 get_lib_base valgrind   LATEST platform010
 get_lib_base lua        5.3.4  platform010
 

cache/cache_bench_tool.cc

@@ -194,7 +194,7 @@ class SharedState {
       : cv_(&mu_),
         cache_bench_(cache_bench) {}
 
-  ~SharedState() {}
+  ~SharedState() = default;
 
   port::Mutex* GetMutex() { return &mu_; }
 
@@ -425,7 +425,7 @@ class CacheBench {
     }
   }
 
-  ~CacheBench() {}
+  ~CacheBench() = default;
 
   void PopulateCache() {
     Random64 rnd(FLAGS_seed);

cache/cache_reservation_manager_test.cc

@@ -129,7 +129,6 @@ TEST_F(CacheReservationManagerTest,
 
 TEST(CacheReservationManagerIncreaseReservcationOnFullCacheTest,
      IncreaseCacheReservationOnFullCache) {
-  ;
   constexpr std::size_t kSizeDummyEntry =
       CacheReservationManagerImpl<CacheEntryRole::kMisc>::GetDummyEntrySize();
   constexpr std::size_t kSmallCacheCapacity = 4 * kSizeDummyEntry;

cache/cache_test.cc

@@ -106,7 +106,7 @@ class CacheTest : public testing::Test,
     type_ = GetParam();
   }
 
-  ~CacheTest() override {}
+  ~CacheTest() override = default;
 
   // These functions encode/decode keys in tests cases that use
   // int keys.
@@ -766,7 +766,9 @@ TEST_P(CacheTest, OverCapacity) {
     std::string key = EncodeKey(i + 1);
     auto h = cache.Lookup(key);
     ASSERT_TRUE(h != nullptr);
-    if (h) cache.Release(h);
+    if (h) {
+      cache.Release(h);
+    }
   }
 
   // the cache is over capacity since nothing could be evicted
@@ -777,7 +779,7 @@ TEST_P(CacheTest, OverCapacity) {
 
   if (IsHyperClock()) {
     // Make sure eviction is triggered.
-    ASSERT_OK(cache.Insert(EncodeKey(-1), nullptr, 1, &handles[0]));
+    ASSERT_OK(cache.Insert(EncodeKey(-1), nullptr, 1, handles.data()));
 
     // cache is under capacity now since elements were released
     ASSERT_GE(n, cache.get()->GetUsage());

cache/compressed_secondary_cache.cc

@@ -26,7 +26,7 @@ CompressedSecondaryCache::CompressedSecondaryCache(
               cache_))),
       disable_cache_(opts.capacity == 0) {}
 
-CompressedSecondaryCache::~CompressedSecondaryCache() {}
+CompressedSecondaryCache::~CompressedSecondaryCache() = default;
 
 std::unique_ptr<SecondaryCacheResultHandle> CompressedSecondaryCache::Lookup(
     const Slice& key, const Cache::CacheItemHelper* helper,
@@ -384,7 +384,7 @@ const Cache::CacheItemHelper* CompressedSecondaryCache::GetHelper(
             chunks_head = chunks_head->next;
             tmp_chunk->Free();
             obj = nullptr;
-          };
+          }
         }};
     return &kHelper;
   } else {

cache/compressed_secondary_cache_test.cc

@@ -33,7 +33,7 @@ const std::string key3 = "____    ____key3";
 class CompressedSecondaryCacheTestBase : public testing::Test,
                                          public WithCacheType {
  public:
-  CompressedSecondaryCacheTestBase() {}
+  CompressedSecondaryCacheTestBase() = default;
   ~CompressedSecondaryCacheTestBase() override = default;
 
  protected:

cache/lru_cache.cc

@@ -277,8 +277,8 @@ void LRUCacheShard::LRU_Insert(LRUHandle* e) {
     e->SetInHighPriPool(false);
     e->SetInLowPriPool(true);
     low_pri_pool_usage_ += e->total_charge;
-    MaintainPoolSize();
     lru_low_pri_ = e;
+    MaintainPoolSize();
   } else {
     // Insert "e" to the head of bottom-pri pool.
     e->next = lru_bottom_pri_->next;
@@ -301,6 +301,7 @@ void LRUCacheShard::MaintainPoolSize() {
     // Overflow last entry in high-pri pool to low-pri pool.
     lru_low_pri_ = lru_low_pri_->next;
     assert(lru_low_pri_ != &lru_);
+    assert(lru_low_pri_->InHighPriPool());
     lru_low_pri_->SetInHighPriPool(false);
     lru_low_pri_->SetInLowPriPool(true);
     assert(high_pri_pool_usage_ >= lru_low_pri_->total_charge);
@@ -312,6 +313,7 @@ void LRUCacheShard::MaintainPoolSize() {
     // Overflow last entry in low-pri pool to bottom-pri pool.
     lru_bottom_pri_ = lru_bottom_pri_->next;
     assert(lru_bottom_pri_ != &lru_);
+    assert(lru_bottom_pri_->InLowPriPool());
     lru_bottom_pri_->SetInHighPriPool(false);
     lru_bottom_pri_->SetInLowPriPool(false);
     assert(low_pri_pool_usage_ >= lru_bottom_pri_->total_charge);

cache/lru_cache_test.cc

@@ -32,7 +32,7 @@ namespace ROCKSDB_NAMESPACE {
 
 class LRUCacheTest : public testing::Test {
  public:
-  LRUCacheTest() {}
+  LRUCacheTest() = default;
   ~LRUCacheTest() override { DeleteCache(); }
 
   void DeleteCache() {
@@ -57,10 +57,11 @@ class LRUCacheTest : public testing::Test {
   }
 
   void Insert(const std::string& key,
-              Cache::Priority priority = Cache::Priority::LOW) {
+              Cache::Priority priority = Cache::Priority::LOW,
+              size_t charge = 1) {
     EXPECT_OK(cache_->Insert(key, 0 /*hash*/, nullptr /*value*/,
-                             &kNoopCacheItemHelper, 1 /*charge*/,
-                             nullptr /*handle*/, priority));
+                             &kNoopCacheItemHelper, charge, nullptr /*handle*/,
+                             priority));
   }
 
   void Insert(char key, Cache::Priority priority = Cache::Priority::LOW) {
@@ -144,8 +145,10 @@ class LRUCacheTest : public testing::Test {
     ASSERT_EQ(num_bottom_pri_pool_keys, bottom_pri_pool_keys);
   }
 
- private:
+ protected:
   LRUCacheShard* cache_ = nullptr;
+
+ private:
   Cache::EvictionCallback eviction_callback_;
 };
 
@@ -378,7 +381,7 @@ class ClockCacheTest : public testing::Test {
   using Table = typename Shard::Table;
   using TableOpts = typename Table::Opts;
 
-  ClockCacheTest() {}
+  ClockCacheTest() = default;
   ~ClockCacheTest() override { DeleteShard(); }
 
   void DeleteShard() {
@@ -1976,7 +1979,7 @@ TEST_P(BasicSecondaryCacheTest, BasicWaitAllTest) {
     ah.priority = Cache::Priority::LOW;
     cache->StartAsyncLookup(ah);
   }
-  cache->WaitAll(&async_handles[0], async_handles.size());
+  cache->WaitAll(async_handles.data(), async_handles.size());
   for (size_t i = 0; i < async_handles.size(); ++i) {
     SCOPED_TRACE("i = " + std::to_string(i));
     Cache::Handle* result = async_handles[i].Result();
@@ -2703,6 +2706,23 @@ TEST_P(DBSecondaryCacheTest, TestSecondaryCacheOptionTwoDB) {
   ASSERT_OK(DestroyDB(dbname2, options));
 }
 
+TEST_F(LRUCacheTest, InsertAfterReducingCapacity) {
+  // Fix a bug in LRU cache where it may try to remove a low pri entry's
+  // charge from high pri pool. It causes
+  // Assertion failed: (high_pri_pool_usage_ >= lru_low_pri_->total_charge),
+  // function MaintainPoolSize, file lru_cache.cc
+  NewCache(/*capacity=*/10, /*high_pri_pool_ratio=*/0.2,
+           /*low_pri_pool_ratio=*/0.8);
+  // high pri pool size and usage are both 2
+  Insert("x", Cache::Priority::HIGH);
+  Insert("y", Cache::Priority::HIGH);
+  cache_->SetCapacity(5);
+  // high_pri_pool_size is 1, the next time we try to maintain pool size,
+  // we will move entries from high pri pool to low pri pool
+  // The bug was deducting this entry's charge from high pri pool usage.
+  Insert("aaa", Cache::Priority::LOW, /*charge=*/3);
+}
+
 }  // namespace ROCKSDB_NAMESPACE
 
 int main(int argc, char** argv) {

cache/secondary_cache_adapter.cc

@@ -134,12 +134,14 @@ bool CacheWithSecondaryAdapter::EvictionHandler(const Slice& key,
     auto obj = target_->Value(handle);
     // Ignore dummy entry
     if (obj != kDummyObj) {
-      bool hit = false;
+      bool force = false;
       if (adm_policy_ == TieredAdmissionPolicy::kAdmPolicyAllowCacheHits) {
-        hit = was_hit;
+        force = was_hit;
+      } else if (adm_policy_ == TieredAdmissionPolicy::kAdmPolicyAllowAll) {
+        force = true;
       }
       // Spill into secondary cache.
-      secondary_cache_->Insert(key, obj, helper, hit).PermitUncheckedError();
+      secondary_cache_->Insert(key, obj, helper, force).PermitUncheckedError();
     }
   }
   // Never takes ownership of obj
@@ -661,6 +663,7 @@ std::shared_ptr<Cache> NewTieredCache(const TieredCacheOptions& _opts) {
         break;
       case TieredAdmissionPolicy::kAdmPolicyPlaceholder:
       case TieredAdmissionPolicy::kAdmPolicyAllowCacheHits:
+      case TieredAdmissionPolicy::kAdmPolicyAllowAll:
         if (opts.nvm_sec_cache) {
           valid_adm_policy = false;
         }

cache/tiered_secondary_cache_test.cc

@@ -386,7 +386,7 @@ TEST_F(DBTieredSecondaryCacheTest, BasicMultiGetTest) {
   keys.push_back(Key(8));
   values = MultiGet(keys, /*snapshot=*/nullptr, /*async=*/true);
   ASSERT_EQ(values.size(), keys.size());
-  for (auto value : values) {
+  for (const auto& value : values) {
     ASSERT_EQ(1007, value.size());
   }
   ASSERT_EQ(nvm_sec_cache()->num_insert_saved(), 3u);
@@ -400,7 +400,7 @@ TEST_F(DBTieredSecondaryCacheTest, BasicMultiGetTest) {
   keys.push_back(Key(20));
   values = MultiGet(keys, /*snapshot=*/nullptr, /*async=*/true);
   ASSERT_EQ(values.size(), keys.size());
-  for (auto value : values) {
+  for (const auto& value : values) {
     ASSERT_EQ(1007, value.size());
   }
   ASSERT_EQ(nvm_sec_cache()->num_insert_saved(), 6u);
@@ -414,7 +414,7 @@ TEST_F(DBTieredSecondaryCacheTest, BasicMultiGetTest) {
   keys.push_back(Key(8));
   values = MultiGet(keys, /*snapshot=*/nullptr, /*async=*/true);
   ASSERT_EQ(values.size(), keys.size());
-  for (auto value : values) {
+  for (const auto& value : values) {
     ASSERT_EQ(1007, value.size());
   }
   ASSERT_EQ(nvm_sec_cache()->num_insert_saved(), 6u);
@@ -428,7 +428,7 @@ TEST_F(DBTieredSecondaryCacheTest, BasicMultiGetTest) {
   keys.push_back(Key(8));
   values = MultiGet(keys, /*snapshot=*/nullptr, /*async=*/true);
   ASSERT_EQ(values.size(), keys.size());
-  for (auto value : values) {
+  for (const auto& value : values) {
     ASSERT_EQ(1007, value.size());
   }
   ASSERT_EQ(nvm_sec_cache()->num_insert_saved(), 6u);
@@ -442,7 +442,7 @@ TEST_F(DBTieredSecondaryCacheTest, BasicMultiGetTest) {
   keys.push_back(Key(8));
   values = MultiGet(keys, /*snapshot=*/nullptr, /*async=*/true);
   ASSERT_EQ(values.size(), keys.size());
-  for (auto value : values) {
+  for (const auto& value : values) {
     ASSERT_EQ(1007, value.size());
   }
   ASSERT_EQ(nvm_sec_cache()->num_insert_saved(), 6u);
@@ -456,7 +456,7 @@ TEST_F(DBTieredSecondaryCacheTest, BasicMultiGetTest) {
   keys.push_back(Key(20));
   values = MultiGet(keys, /*snapshot=*/nullptr, /*async=*/true);
   ASSERT_EQ(values.size(), keys.size());
-  for (auto value : values) {
+  for (const auto& value : values) {
     ASSERT_EQ(1007, value.size());
   }
   ASSERT_EQ(nvm_sec_cache()->num_insert_saved(), 6u);
@@ -470,7 +470,7 @@ TEST_F(DBTieredSecondaryCacheTest, BasicMultiGetTest) {
   keys.push_back(Key(20));
   values = MultiGet(keys, /*snapshot=*/nullptr, /*async=*/true);
   ASSERT_EQ(values.size(), keys.size());
-  for (auto value : values) {
+  for (const auto& value : values) {
     ASSERT_EQ(1007, value.size());
   }
   ASSERT_EQ(nvm_sec_cache()->num_insert_saved(), 6u);
@@ -484,7 +484,7 @@ TEST_F(DBTieredSecondaryCacheTest, BasicMultiGetTest) {
   keys.push_back(Key(20));
   values = MultiGet(keys, /*snapshot=*/nullptr, /*async=*/true);
   ASSERT_EQ(values.size(), keys.size());
-  for (auto value : values) {
+  for (const auto& value : values) {
     ASSERT_EQ(1007, value.size());
   }
   ASSERT_EQ(nvm_sec_cache()->num_insert_saved(), 6u);
@@ -528,7 +528,7 @@ TEST_F(DBTieredSecondaryCacheTest, WaitAllTest) {
   keys.push_back(Key(8));
   values = MultiGet(keys, /*snapshot=*/nullptr, /*async=*/true);
   ASSERT_EQ(values.size(), keys.size());
-  for (auto value : values) {
+  for (const auto& value : values) {
     ASSERT_EQ(1007, value.size());
   }
   ASSERT_EQ(nvm_sec_cache()->num_insert_saved(), 3u);
@@ -542,7 +542,7 @@ TEST_F(DBTieredSecondaryCacheTest, WaitAllTest) {
   keys.push_back(Key(20));
   values = MultiGet(keys, /*snapshot=*/nullptr, /*async=*/true);
   ASSERT_EQ(values.size(), keys.size());
-  for (auto value : values) {
+  for (const auto& value : values) {
     ASSERT_EQ(1007, value.size());
   }
   ASSERT_EQ(nvm_sec_cache()->num_insert_saved(), 6u);
@@ -561,7 +561,7 @@ TEST_F(DBTieredSecondaryCacheTest, WaitAllTest) {
   keys.push_back(Key(36));
   values = MultiGet(keys, /*snapshot=*/nullptr, /*async=*/true);
   ASSERT_EQ(values.size(), keys.size());
-  for (auto value : values) {
+  for (const auto& value : values) {
     ASSERT_EQ(1007, value.size());
   }
   ASSERT_EQ(nvm_sec_cache()->num_insert_saved(), 10u);
@@ -582,7 +582,7 @@ TEST_F(DBTieredSecondaryCacheTest, WaitAllTest) {
   keys.push_back(Key(8));
   values = MultiGet(keys, /*snapshot=*/nullptr, /*async=*/true);
   ASSERT_EQ(values.size(), keys.size());
-  for (auto value : values) {
+  for (const auto& value : values) {
     ASSERT_EQ(1007, value.size());
   }
   ASSERT_EQ(nvm_sec_cache()->num_insert_saved(), 10u);
@@ -629,7 +629,7 @@ TEST_F(DBTieredSecondaryCacheTest, ReadyBeforeWaitAllTest) {
   keys.push_back(Key(8));
   values = MultiGet(keys, /*snapshot=*/nullptr, /*async=*/true);
   ASSERT_EQ(values.size(), keys.size());
-  for (auto value : values) {
+  for (const auto& value : values) {
     ASSERT_EQ(1007, value.size());
   }
   ASSERT_EQ(nvm_sec_cache()->num_insert_saved(), 3u);
@@ -644,7 +644,7 @@ TEST_F(DBTieredSecondaryCacheTest, ReadyBeforeWaitAllTest) {
   keys.push_back(Key(20));
   values = MultiGet(keys, /*snapshot=*/nullptr, /*async=*/true);
   ASSERT_EQ(values.size(), keys.size());
-  for (auto value : values) {
+  for (const auto& value : values) {
     ASSERT_EQ(1007, value.size());
   }
   ASSERT_EQ(nvm_sec_cache()->num_insert_saved(), 6u);
@@ -659,7 +659,7 @@ TEST_F(DBTieredSecondaryCacheTest, ReadyBeforeWaitAllTest) {
   keys.push_back(Key(8));
   values = MultiGet(keys, /*snapshot=*/nullptr, /*async=*/true);
   ASSERT_EQ(values.size(), keys.size());
-  for (auto value : values) {
+  for (const auto& value : values) {
     ASSERT_EQ(1007, value.size());
   }
   ASSERT_EQ(nvm_sec_cache()->num_insert_saved(), 6u);
@@ -676,7 +676,7 @@ TEST_F(DBTieredSecondaryCacheTest, ReadyBeforeWaitAllTest) {
   keys.push_back(Key(36));
   values = MultiGet(keys, /*snapshot=*/nullptr, /*async=*/true);
   ASSERT_EQ(values.size(), keys.size());
-  for (auto value : values) {
+  for (const auto& value : values) {
     ASSERT_EQ(1007, value.size());
   }
   ASSERT_EQ(nvm_sec_cache()->num_insert_saved(), 8u);
@@ -691,7 +691,7 @@ TEST_F(DBTieredSecondaryCacheTest, ReadyBeforeWaitAllTest) {
   keys.push_back(Key(36));
   values = MultiGet(keys, /*snapshot=*/nullptr, /*async=*/true);
   ASSERT_EQ(values.size(), keys.size());
-  for (auto value : values) {
+  for (const auto& value : values) {
     ASSERT_EQ(1007, value.size());
   }
   ASSERT_EQ(nvm_sec_cache()->num_insert_saved(), 8u);
@@ -816,11 +816,174 @@ TEST_P(DBTieredAdmPolicyTest, CompressedOnlyTest) {
   Destroy(options);
 }
 
+TEST_P(DBTieredAdmPolicyTest, CompressedCacheAdmission) {
+  if (!LZ4_Supported()) {
+    ROCKSDB_GTEST_SKIP("This test requires LZ4 support.");
+    return;
+  }
+
+  BlockBasedTableOptions table_options;
+  // We want a block cache of size 5KB, and a compressed secondary cache of
+  // size 5KB. However, we specify a block cache size of 256KB here in order
+  // to take into account the cache reservation in the block cache on
+  // behalf of the compressed cache. The unit of cache reservation is 256KB.
+  // The effective block cache capacity will be calculated as 256 + 5 = 261KB,
+  // and 256KB will be reserved for the compressed cache, leaving 10KB for
+  // the primary block cache. We only have to worry about this here because
+  // the cache size is so small.
+  table_options.block_cache = NewCache(256 * 1024, 5 * 1024, 0, GetParam());
+  table_options.block_size = 4 * 1024;
+  table_options.cache_index_and_filter_blocks = false;
+  Options options = GetDefaultOptions();
+  options.create_if_missing = true;
+  options.table_factory.reset(NewBlockBasedTableFactory(table_options));
+
+  size_t comp_cache_usage = compressed_secondary_cache()->TEST_GetUsage();
+  // Disable paranoid_file_checks so that flush will not read back the newly
+  // written file
+  options.paranoid_file_checks = false;
+  DestroyAndReopen(options);
+  Random rnd(301);
+  const int N = 256;
+  for (int i = 0; i < N; i++) {
+    std::string p_v;
+    test::CompressibleString(&rnd, 0.5, 1007, &p_v);
+    ASSERT_OK(Put(Key(i), p_v));
+  }
+
+  ASSERT_OK(Flush());
+
+  // The second Get (for 5) will evict the data block loaded by the first
+  // Get, which will be admitted into the compressed secondary cache only
+  // for the kAdmPolicyAllowAll policy
+  std::string v = Get(Key(0));
+  ASSERT_EQ(1007, v.size());
+
+  v = Get(Key(5));
+  ASSERT_EQ(1007, v.size());
+
+  if (GetParam() == TieredAdmissionPolicy::kAdmPolicyAllowAll) {
+    ASSERT_GT(compressed_secondary_cache()->TEST_GetUsage(),
+              comp_cache_usage + 128);
+  } else {
+    ASSERT_LT(compressed_secondary_cache()->TEST_GetUsage(),
+              comp_cache_usage + 128);
+  }
+
+  Destroy(options);
+}
+
+TEST_F(DBTieredSecondaryCacheTest, FSBufferTest) {
+  class WrapFS : public FileSystemWrapper {
+   public:
+    explicit WrapFS(const std::shared_ptr<FileSystem>& _target)
+        : FileSystemWrapper(_target) {}
+    ~WrapFS() override {}
+    const char* Name() const override { return "WrapFS"; }
+
+    IOStatus NewRandomAccessFile(const std::string& fname,
+                                 const FileOptions& opts,
+                                 std::unique_ptr<FSRandomAccessFile>* result,
+                                 IODebugContext* dbg) override {
+      class WrappedRandomAccessFile : public FSRandomAccessFileOwnerWrapper {
+       public:
+        explicit WrappedRandomAccessFile(
+            std::unique_ptr<FSRandomAccessFile>& file)
+            : FSRandomAccessFileOwnerWrapper(std::move(file)) {}
+
+        IOStatus MultiRead(FSReadRequest* reqs, size_t num_reqs,
+                           const IOOptions& options,
+                           IODebugContext* dbg) override {
+          for (size_t i = 0; i < num_reqs; ++i) {
+            FSReadRequest& req = reqs[i];
+            FSAllocationPtr buffer(new char[req.len], [](void* ptr) {
+              delete[] static_cast<char*>(ptr);
+            });
+            req.fs_scratch = std::move(buffer);
+            req.status = Read(req.offset, req.len, options, &req.result,
+                              static_cast<char*>(req.fs_scratch.get()), dbg);
+          }
+          return IOStatus::OK();
+        }
+      };
+
+      std::unique_ptr<FSRandomAccessFile> file;
+      IOStatus s = target()->NewRandomAccessFile(fname, opts, &file, dbg);
+      EXPECT_OK(s);
+      result->reset(new WrappedRandomAccessFile(file));
+
+      return s;
+    }
+
+    void SupportedOps(int64_t& supported_ops) override {
+      supported_ops = 1 << FSSupportedOps::kAsyncIO;
+      supported_ops |= 1 << FSSupportedOps::kFSBuffer;
+    }
+  };
+
+  if (!LZ4_Supported()) {
+    ROCKSDB_GTEST_SKIP("This test requires LZ4 support.");
+    return;
+  }
+
+  std::shared_ptr<WrapFS> wrap_fs =
+      std::make_shared<WrapFS>(env_->GetFileSystem());
+  std::unique_ptr<Env> wrap_env(new CompositeEnvWrapper(env_, wrap_fs));
+  BlockBasedTableOptions table_options;
+  table_options.block_cache = NewCache(250 * 1024, 20 * 1024, 256 * 1024,
+                                       TieredAdmissionPolicy::kAdmPolicyAuto,
+                                       /*ready_before_wait=*/true);
+  table_options.block_size = 4 * 1024;
+  table_options.cache_index_and_filter_blocks = false;
+  Options options = GetDefaultOptions();
+  options.create_if_missing = true;
+  options.table_factory.reset(NewBlockBasedTableFactory(table_options));
+  options.statistics = CreateDBStatistics();
+  options.env = wrap_env.get();
+
+  options.paranoid_file_checks = false;
+  DestroyAndReopen(options);
+  Random rnd(301);
+  const int N = 256;
+  for (int i = 0; i < N; i++) {
+    std::string p_v;
+    test::CompressibleString(&rnd, 0.5, 1007, &p_v);
+    ASSERT_OK(Put(Key(i), p_v));
+  }
+
+  ASSERT_OK(Flush());
+
+  std::vector<std::string> keys;
+  std::vector<std::string> values;
+
+  keys.push_back(Key(0));
+  keys.push_back(Key(4));
+  keys.push_back(Key(8));
+  values = MultiGet(keys, /*snapshot=*/nullptr, /*async=*/true);
+  ASSERT_EQ(values.size(), keys.size());
+  for (const auto& value : values) {
+    ASSERT_EQ(1007, value.size());
+  }
+  ASSERT_EQ(nvm_sec_cache()->num_insert_saved(), 3u);
+  ASSERT_EQ(nvm_sec_cache()->num_misses(), 3u);
+  ASSERT_EQ(nvm_sec_cache()->num_hits(), 0u);
+
+  std::string v = Get(Key(12));
+  ASSERT_EQ(1007, v.size());
+  ASSERT_EQ(nvm_sec_cache()->num_insert_saved(), 4u);
+  ASSERT_EQ(nvm_sec_cache()->num_misses(), 4u);
+  ASSERT_EQ(options.statistics->getTickerCount(BLOCK_CACHE_MISS), 4u);
+
+  Close();
+  Destroy(options);
+}
+
 INSTANTIATE_TEST_CASE_P(
     DBTieredAdmPolicyTest, DBTieredAdmPolicyTest,
     ::testing::Values(TieredAdmissionPolicy::kAdmPolicyAuto,
                       TieredAdmissionPolicy::kAdmPolicyPlaceholder,
-                      TieredAdmissionPolicy::kAdmPolicyAllowCacheHits));
+                      TieredAdmissionPolicy::kAdmPolicyAllowCacheHits,
+                      TieredAdmissionPolicy::kAdmPolicyAllowAll));
 
 }  // namespace ROCKSDB_NAMESPACE
 

cmake/modules/Findzstd.cmake

@@ -1,29 +1,29 @@
 # - Find zstd
 # Find the zstd compression library and includes
 #
-# zstd_INCLUDE_DIRS - where to find zstd.h, etc.
-# zstd_LIBRARIES - List of libraries when using zstd.
-# zstd_FOUND - True if zstd found.
+# ZSTD_INCLUDE_DIRS - where to find zstd.h, etc.
+# ZSTD_LIBRARIES - List of libraries when using zstd.
+# ZSTD_FOUND - True if zstd found.
 
-find_path(zstd_INCLUDE_DIRS
+find_path(ZSTD_INCLUDE_DIRS
   NAMES zstd.h
   HINTS ${zstd_ROOT_DIR}/include)
 
-find_library(zstd_LIBRARIES
+find_library(ZSTD_LIBRARIES
   NAMES zstd
   HINTS ${zstd_ROOT_DIR}/lib)
 
 include(FindPackageHandleStandardArgs)
-find_package_handle_standard_args(zstd DEFAULT_MSG zstd_LIBRARIES ZSTD_INCLUDE_DIRS)
+find_package_handle_standard_args(zstd DEFAULT_MSG ZSTD_LIBRARIES ZSTD_INCLUDE_DIRS)
 
 mark_as_advanced(
-  zstd_LIBRARIES
-  zstd_INCLUDE_DIRS)
+  ZSTD_LIBRARIES
+  ZSTD_INCLUDE_DIRS)
 
-if(zstd_FOUND AND NOT (TARGET zstd::zstd))
+if(ZSTD_FOUND AND NOT (TARGET zstd::zstd))
   add_library (zstd::zstd UNKNOWN IMPORTED)
   set_target_properties(zstd::zstd
     PROPERTIES
-      IMPORTED_LOCATION ${zstd_LIBRARIES}
-      INTERFACE_INCLUDE_DIRECTORIES ${zstd_INCLUDE_DIRS})
+      IMPORTED_LOCATION ${ZSTD_LIBRARIES}
+      INTERFACE_INCLUDE_DIRECTORIES ${ZSTD_INCLUDE_DIRS})
 endif()

db/arena_wrapped_db_iter.cc

@@ -43,14 +43,13 @@ void ArenaWrappedDBIter::Init(
     Env* env, const ReadOptions& read_options, const ImmutableOptions& ioptions,
     const MutableCFOptions& mutable_cf_options, const Version* version,
     const SequenceNumber& sequence, uint64_t max_sequential_skip_in_iteration,
-    uint64_t version_number, ReadCallback* read_callback, DBImpl* db_impl,
-    ColumnFamilyData* cfd, bool expose_blob_index, bool allow_refresh) {
+    uint64_t version_number, ReadCallback* read_callback,
+    ColumnFamilyHandleImpl* cfh, bool expose_blob_index, bool allow_refresh) {
   auto mem = arena_.AllocateAligned(sizeof(DBIter));
-  db_iter_ =
-      new (mem) DBIter(env, read_options, ioptions, mutable_cf_options,
-                       ioptions.user_comparator, /* iter */ nullptr, version,
-                       sequence, true, max_sequential_skip_in_iteration,
-                       read_callback, db_impl, cfd, expose_blob_index);
+  db_iter_ = new (mem) DBIter(
+      env, read_options, ioptions, mutable_cf_options, ioptions.user_comparator,
+      /* iter */ nullptr, version, sequence, true,
+      max_sequential_skip_in_iteration, read_callback, cfh, expose_blob_index);
   sv_number_ = version_number;
   read_options_ = read_options;
   allow_refresh_ = allow_refresh;
@@ -65,40 +64,44 @@ void ArenaWrappedDBIter::Init(
 Status ArenaWrappedDBIter::Refresh() { return Refresh(nullptr); }
 
 Status ArenaWrappedDBIter::Refresh(const Snapshot* snapshot) {
-  if (cfd_ == nullptr || db_impl_ == nullptr || !allow_refresh_) {
+  if (cfh_ == nullptr || !allow_refresh_) {
     return Status::NotSupported("Creating renew iterator is not allowed.");
   }
   assert(db_iter_ != nullptr);
+  auto cfd = cfh_->cfd();
+  auto db_impl = cfh_->db();
+
   // TODO(yiwu): For last_seq_same_as_publish_seq_==false, this is not the
   // correct behavior. Will be corrected automatically when we take a snapshot
   // here for the case of WritePreparedTxnDB.
-  uint64_t cur_sv_number = cfd_->GetSuperVersionNumber();
+  uint64_t cur_sv_number = cfd->GetSuperVersionNumber();
   // If we recreate a new internal iterator below (NewInternalIterator()),
   // we will pass in read_options_. We need to make sure it
   // has the right snapshot.
   read_options_.snapshot = snapshot;
   TEST_SYNC_POINT("ArenaWrappedDBIter::Refresh:1");
   TEST_SYNC_POINT("ArenaWrappedDBIter::Refresh:2");
+
   auto reinit_internal_iter = [&]() {
     Env* env = db_iter_->env();
     db_iter_->~DBIter();
     arena_.~Arena();
     new (&arena_) Arena();
 
-    SuperVersion* sv = cfd_->GetReferencedSuperVersion(db_impl_);
+    SuperVersion* sv = cfd->GetReferencedSuperVersion(db_impl);
     assert(sv->version_number >= cur_sv_number);
-    SequenceNumber read_seq = GetSeqNum(db_impl_, snapshot);
+    SequenceNumber read_seq = GetSeqNum(db_impl, snapshot);
     if (read_callback_) {
       read_callback_->Refresh(read_seq);
     }
-    Init(env, read_options_, *(cfd_->ioptions()), sv->mutable_cf_options,
+    Init(env, read_options_, *(cfd->ioptions()), sv->mutable_cf_options,
          sv->current, read_seq,
          sv->mutable_cf_options.max_sequential_skip_in_iterations,
-         sv->version_number, read_callback_, db_impl_, cfd_, expose_blob_index_,
+         sv->version_number, read_callback_, cfh_, expose_blob_index_,
          allow_refresh_);
 
-    InternalIterator* internal_iter = db_impl_->NewInternalIterator(
-        read_options_, cfd_, sv, &arena_, read_seq,
+    InternalIterator* internal_iter = db_impl->NewInternalIterator(
+        read_options_, cfd, sv, &arena_, read_seq,
         /* allow_unprepared_value */ true, /* db_iter */ this);
     SetIterUnderDBIter(internal_iter);
   };
@@ -107,10 +110,10 @@ Status ArenaWrappedDBIter::Refresh(const Snapshot* snapshot) {
       reinit_internal_iter();
       break;
     } else {
-      SequenceNumber read_seq = GetSeqNum(db_impl_, snapshot);
+      SequenceNumber read_seq = GetSeqNum(db_impl, snapshot);
       // Refresh range-tombstones in MemTable
       if (!read_options_.ignore_range_deletions) {
-        SuperVersion* sv = cfd_->GetThreadLocalSuperVersion(db_impl_);
+        SuperVersion* sv = cfd->GetThreadLocalSuperVersion(db_impl);
         TEST_SYNC_POINT_CALLBACK("ArenaWrappedDBIter::Refresh:SV", nullptr);
         auto t = sv->mem->NewRangeTombstoneIterator(
             read_options_, read_seq, false /* immutable_memtable */);
@@ -123,13 +126,13 @@ Status ArenaWrappedDBIter::Refresh(const Snapshot* snapshot) {
           // will be freed during db_iter destruction there.
           if (memtable_range_tombstone_iter_) {
             assert(!*memtable_range_tombstone_iter_ ||
-                   sv_number_ != cfd_->GetSuperVersionNumber());
+                   sv_number_ != cfd->GetSuperVersionNumber());
           }
           delete t;
         } else {  // current mutable memtable has range tombstones
           if (!memtable_range_tombstone_iter_) {
             delete t;
-            db_impl_->ReturnAndCleanupSuperVersion(cfd_, sv);
+            db_impl->ReturnAndCleanupSuperVersion(cfd, sv);
             // The memtable under DBIter did not have range tombstone before
             // refresh.
             reinit_internal_iter();
@@ -138,13 +141,13 @@ Status ArenaWrappedDBIter::Refresh(const Snapshot* snapshot) {
             delete *memtable_range_tombstone_iter_;
             *memtable_range_tombstone_iter_ = new TruncatedRangeDelIterator(
                 std::unique_ptr<FragmentedRangeTombstoneIterator>(t),
-                &cfd_->internal_comparator(), nullptr, nullptr);
+                &cfd->internal_comparator(), nullptr, nullptr);
           }
         }
-        db_impl_->ReturnAndCleanupSuperVersion(cfd_, sv);
+        db_impl->ReturnAndCleanupSuperVersion(cfd, sv);
       }
       // Check again if the latest super version number is changed
-      uint64_t latest_sv_number = cfd_->GetSuperVersionNumber();
+      uint64_t latest_sv_number = cfd->GetSuperVersionNumber();
       if (latest_sv_number != cur_sv_number) {
         // If the super version number is changed after refreshing,
         // fallback to Re-Init the InternalIterator
@@ -163,14 +166,14 @@ ArenaWrappedDBIter* NewArenaWrappedDbIterator(
     Env* env, const ReadOptions& read_options, const ImmutableOptions& ioptions,
     const MutableCFOptions& mutable_cf_options, const Version* version,
     const SequenceNumber& sequence, uint64_t max_sequential_skip_in_iterations,
-    uint64_t version_number, ReadCallback* read_callback, DBImpl* db_impl,
-    ColumnFamilyData* cfd, bool expose_blob_index, bool allow_refresh) {
+    uint64_t version_number, ReadCallback* read_callback,
+    ColumnFamilyHandleImpl* cfh, bool expose_blob_index, bool allow_refresh) {
   ArenaWrappedDBIter* iter = new ArenaWrappedDBIter();
   iter->Init(env, read_options, ioptions, mutable_cf_options, version, sequence,
              max_sequential_skip_in_iterations, version_number, read_callback,
-             db_impl, cfd, expose_blob_index, allow_refresh);
-  if (db_impl != nullptr && cfd != nullptr && allow_refresh) {
-    iter->StoreRefreshInfo(db_impl, cfd, read_callback, expose_blob_index);
+             cfh, expose_blob_index, allow_refresh);
+  if (cfh != nullptr && allow_refresh) {
+    iter->StoreRefreshInfo(cfh, read_callback, expose_blob_index);
   }
 
   return iter;

db/arena_wrapped_db_iter.h

@@ -87,15 +87,14 @@ class ArenaWrappedDBIter : public Iterator {
             const MutableCFOptions& mutable_cf_options, const Version* version,
             const SequenceNumber& sequence,
             uint64_t max_sequential_skip_in_iterations, uint64_t version_number,
-            ReadCallback* read_callback, DBImpl* db_impl, ColumnFamilyData* cfd,
+            ReadCallback* read_callback, ColumnFamilyHandleImpl* cfh,
             bool expose_blob_index, bool allow_refresh);
 
   // Store some parameters so we can refresh the iterator at a later point
   // with these same params
-  void StoreRefreshInfo(DBImpl* db_impl, ColumnFamilyData* cfd,
+  void StoreRefreshInfo(ColumnFamilyHandleImpl* cfh,
                         ReadCallback* read_callback, bool expose_blob_index) {
-    db_impl_ = db_impl;
-    cfd_ = cfd;
+    cfh_ = cfh;
     read_callback_ = read_callback;
     expose_blob_index_ = expose_blob_index;
   }
@@ -104,8 +103,7 @@ class ArenaWrappedDBIter : public Iterator {
   DBIter* db_iter_ = nullptr;
   Arena arena_;
   uint64_t sv_number_;
-  ColumnFamilyData* cfd_ = nullptr;
-  DBImpl* db_impl_ = nullptr;
+  ColumnFamilyHandleImpl* cfh_ = nullptr;
   ReadOptions read_options_;
   ReadCallback* read_callback_;
   bool expose_blob_index_ = false;
@@ -116,13 +114,13 @@ class ArenaWrappedDBIter : public Iterator {
 };
 
 // Generate the arena wrapped iterator class.
-// `db_impl` and `cfd` are used for reneweal. If left null, renewal will not
+// `cfh` is used for reneweal. If left null, renewal will not
 // be supported.
 ArenaWrappedDBIter* NewArenaWrappedDbIterator(
     Env* env, const ReadOptions& read_options, const ImmutableOptions& ioptions,
     const MutableCFOptions& mutable_cf_options, const Version* version,
     const SequenceNumber& sequence, uint64_t max_sequential_skip_in_iterations,
     uint64_t version_number, ReadCallback* read_callback,
-    DBImpl* db_impl = nullptr, ColumnFamilyData* cfd = nullptr,
-    bool expose_blob_index = false, bool allow_refresh = true);
+    ColumnFamilyHandleImpl* cfh = nullptr, bool expose_blob_index = false,
+    bool allow_refresh = true);
 }  // namespace ROCKSDB_NAMESPACE

db/attribute_group_iterator_impl.cc

@@ -0,0 +1,20 @@
+//  Copyright (c) Meta Platforms, Inc. and affiliates.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+
+#include "db/attribute_group_iterator_impl.h"
+
+namespace ROCKSDB_NAMESPACE {
+
+const AttributeGroups kNoAttributeGroups;
+const IteratorAttributeGroups kNoIteratorAttributeGroups;
+
+void AttributeGroupIteratorImpl::AddToAttributeGroups(
+    const autovector<MultiCfIteratorInfo>& items) {
+  for (const auto& item : items) {
+    attribute_groups_.emplace_back(item.cfh, &item.iterator->columns());
+  }
+}
+
+}  // namespace ROCKSDB_NAMESPACE

db/attribute_group_iterator_impl.h

@@ -0,0 +1,83 @@
+//  Copyright (c) Meta Platforms, Inc. and affiliates.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+
+#pragma once
+
+#include "db/multi_cf_iterator_impl.h"
+#include "rocksdb/attribute_groups.h"
+
+namespace ROCKSDB_NAMESPACE {
+
+class AttributeGroupIteratorImpl : public AttributeGroupIterator {
+ public:
+  AttributeGroupIteratorImpl(
+      const Comparator* comparator,
+      const std::vector<ColumnFamilyHandle*>& column_families,
+      const std::vector<Iterator*>& child_iterators)
+      : impl_(
+            comparator, column_families, child_iterators, [this]() { Reset(); },
+            [this](const autovector<MultiCfIteratorInfo>& items) {
+              AddToAttributeGroups(items);
+            }) {}
+  ~AttributeGroupIteratorImpl() override {}
+
+  // No copy allowed
+  AttributeGroupIteratorImpl(const AttributeGroupIteratorImpl&) = delete;
+  AttributeGroupIteratorImpl& operator=(const AttributeGroupIteratorImpl&) =
+      delete;
+
+  bool Valid() const override { return impl_.Valid(); }
+  void SeekToFirst() override { impl_.SeekToFirst(); }
+  void SeekToLast() override { impl_.SeekToLast(); }
+  void Seek(const Slice& target) override { impl_.Seek(target); }
+  void SeekForPrev(const Slice& target) override { impl_.SeekForPrev(target); }
+  void Next() override { impl_.Next(); }
+  void Prev() override { impl_.Prev(); }
+  Slice key() const override { return impl_.key(); }
+  Status status() const override { return impl_.status(); }
+
+  const IteratorAttributeGroups& attribute_groups() const override {
+    assert(Valid());
+    return attribute_groups_;
+  }
+
+  void Reset() { attribute_groups_.clear(); }
+
+ private:
+  MultiCfIteratorImpl impl_;
+  IteratorAttributeGroups attribute_groups_;
+  void AddToAttributeGroups(const autovector<MultiCfIteratorInfo>& items);
+};
+
+class EmptyAttributeGroupIterator : public AttributeGroupIterator {
+ public:
+  explicit EmptyAttributeGroupIterator(const Status& s) : status_(s) {}
+  bool Valid() const override { return false; }
+  void Seek(const Slice& /*target*/) override {}
+  void SeekForPrev(const Slice& /*target*/) override {}
+  void SeekToFirst() override {}
+  void SeekToLast() override {}
+  void Next() override { assert(false); }
+  void Prev() override { assert(false); }
+  Slice key() const override {
+    assert(false);
+    return Slice();
+  }
+  Status status() const override { return status_; }
+
+  const IteratorAttributeGroups& attribute_groups() const override {
+    return kNoIteratorAttributeGroups;
+  }
+
+ private:
+  Status status_;
+};
+
+inline std::unique_ptr<AttributeGroupIterator> NewAttributeGroupErrorIterator(
+    const Status& status) {
+  return std::make_unique<EmptyAttributeGroupIterator>(status);
+}
+
+}  // namespace ROCKSDB_NAMESPACE

db/blob/blob_file_reader_test.cc

@@ -405,7 +405,7 @@ TEST_F(BlobFileReaderTest, CreateReaderAndGetBlob) {
 
     requests_buf[0] =
         BlobReadRequest(key_refs[0], blob_offsets[0], blob_sizes[0],
-                        kNoCompression, nullptr, &statuses_buf[0]);
+                        kNoCompression, nullptr, statuses_buf.data());
     requests_buf[1] =
         BlobReadRequest(key_refs[1], blob_offsets[1], blob_sizes[1] + 1,
                         kNoCompression, nullptr, &statuses_buf[1]);

db/blob/blob_source_test.cc

@@ -168,8 +168,8 @@ TEST_F(BlobSourceTest, GetBlobsFromCache) {
 
   uint64_t file_size = BlobLogHeader::kSize;
   for (size_t i = 0; i < num_blobs; ++i) {
-    keys.push_back({key_strs[i]});
-    blobs.push_back({blob_strs[i]});
+    keys.emplace_back(key_strs[i]);
+    blobs.emplace_back(blob_strs[i]);
     file_size += BlobLogRecord::kHeaderSize + keys[i].size() + blobs[i].size();
   }
   file_size += BlobLogFooter::kSize;
@@ -482,8 +482,8 @@ TEST_F(BlobSourceTest, GetCompressedBlobs) {
   std::vector<Slice> blobs;
 
   for (size_t i = 0; i < num_blobs; ++i) {
-    keys.push_back({key_strs[i]});
-    blobs.push_back({blob_strs[i]});
+    keys.emplace_back(key_strs[i]);
+    blobs.emplace_back(blob_strs[i]);
   }
 
   std::vector<uint64_t> blob_offsets(keys.size());
@@ -610,8 +610,8 @@ TEST_F(BlobSourceTest, MultiGetBlobsFromMultiFiles) {
   uint64_t file_size = BlobLogHeader::kSize;
   uint64_t blob_value_bytes = 0;
   for (size_t i = 0; i < num_blobs; ++i) {
-    keys.push_back({key_strs[i]});
-    blobs.push_back({blob_strs[i]});
+    keys.emplace_back(key_strs[i]);
+    blobs.emplace_back(blob_strs[i]);
     blob_value_bytes += blobs[i].size();
     file_size += BlobLogRecord::kHeaderSize + keys[i].size() + blobs[i].size();
   }
@@ -802,8 +802,8 @@ TEST_F(BlobSourceTest, MultiGetBlobsFromCache) {
 
   uint64_t file_size = BlobLogHeader::kSize;
   for (size_t i = 0; i < num_blobs; ++i) {
-    keys.push_back({key_strs[i]});
-    blobs.push_back({blob_strs[i]});
+    keys.emplace_back(key_strs[i]);
+    blobs.emplace_back(blob_strs[i]);
     file_size += BlobLogRecord::kHeaderSize + keys[i].size() + blobs[i].size();
   }
   file_size += BlobLogFooter::kSize;
@@ -1164,7 +1164,7 @@ TEST_F(BlobSecondaryCacheTest, GetBlobsFromSecondaryCache) {
     ASSERT_OK(blob_source.GetBlob(read_options, keys[0], file_number,
                                   blob_offsets[0], file_size, blob_sizes[0],
                                   kNoCompression, nullptr /* prefetch_buffer */,
-                                  &values[0], nullptr /* bytes_read */));
+                                  values.data(), nullptr /* bytes_read */));
     // Release cache handle
     values[0].Reset();
 
@@ -1183,7 +1183,7 @@ TEST_F(BlobSecondaryCacheTest, GetBlobsFromSecondaryCache) {
     ASSERT_OK(blob_source.GetBlob(read_options, keys[0], file_number,
                                   blob_offsets[0], file_size, blob_sizes[0],
                                   kNoCompression, nullptr /* prefetch_buffer */,
-                                  &values[0], nullptr /* bytes_read */));
+                                  values.data(), nullptr /* bytes_read */));
     ASSERT_EQ(values[0], blobs[0]);
     ASSERT_TRUE(
         blob_source.TEST_BlobInCache(file_number, file_size, blob_offsets[0]));
@@ -1263,7 +1263,7 @@ TEST_F(BlobSecondaryCacheTest, GetBlobsFromSecondaryCache) {
       ASSERT_OK(blob_source.GetBlob(
           read_options, keys[0], file_number, blob_offsets[0], file_size,
           blob_sizes[0], kNoCompression, nullptr /* prefetch_buffer */,
-          &values[0], nullptr /* bytes_read */));
+          values.data(), nullptr /* bytes_read */));
       ASSERT_EQ(values[0], blobs[0]);
 
       // Release cache handle
@@ -1365,8 +1365,8 @@ class BlobSourceCacheReservationTest : public DBTestBase {
 
     blob_file_size_ = BlobLogHeader::kSize;
     for (size_t i = 0; i < kNumBlobs; ++i) {
-      keys_.push_back({key_strs_[i]});
-      blobs_.push_back({blob_strs_[i]});
+      keys_.emplace_back(key_strs_[i]);
+      blobs_.emplace_back(blob_strs_[i]);
       blob_file_size_ +=
           BlobLogRecord::kHeaderSize + keys_[i].size() + blobs_[i].size();
     }

db/blob/db_blob_basic_test.cc

@@ -418,8 +418,8 @@ TEST_F(DBBlobBasicTest, MultiGetBlobs) {
     std::array<PinnableSlice, num_keys> values;
     std::array<Status, num_keys> statuses;
 
-    db_->MultiGet(read_options, db_->DefaultColumnFamily(), num_keys, &keys[0],
-                  &values[0], &statuses[0]);
+    db_->MultiGet(read_options, db_->DefaultColumnFamily(), num_keys,
+                  keys.data(), values.data(), statuses.data());
 
     ASSERT_OK(statuses[0]);
     ASSERT_EQ(values[0], first_value);
@@ -441,8 +441,8 @@ TEST_F(DBBlobBasicTest, MultiGetBlobs) {
     std::array<PinnableSlice, num_keys> values;
     std::array<Status, num_keys> statuses;
 
-    db_->MultiGet(read_options, db_->DefaultColumnFamily(), num_keys, &keys[0],
-                  &values[0], &statuses[0]);
+    db_->MultiGet(read_options, db_->DefaultColumnFamily(), num_keys,
+                  keys.data(), values.data(), statuses.data());
 
     ASSERT_OK(statuses[0]);
     ASSERT_EQ(values[0], first_value);
@@ -512,8 +512,8 @@ TEST_F(DBBlobBasicTest, MultiGetBlobsFromCache) {
     std::array<PinnableSlice, num_keys> values;
     std::array<Status, num_keys> statuses;
 
-    db_->MultiGet(read_options, db_->DefaultColumnFamily(), num_keys, &keys[0],
-                  &values[0], &statuses[0]);
+    db_->MultiGet(read_options, db_->DefaultColumnFamily(), num_keys,
+                  keys.data(), values.data(), statuses.data());
 
     ASSERT_OK(statuses[0]);
     ASSERT_EQ(values[0], first_value);
@@ -534,8 +534,8 @@ TEST_F(DBBlobBasicTest, MultiGetBlobsFromCache) {
     std::array<PinnableSlice, num_keys> values;
     std::array<Status, num_keys> statuses;
 
-    db_->MultiGet(read_options, db_->DefaultColumnFamily(), num_keys, &keys[0],
-                  &values[0], &statuses[0]);
+    db_->MultiGet(read_options, db_->DefaultColumnFamily(), num_keys,
+                  keys.data(), values.data(), statuses.data());
 
     ASSERT_OK(statuses[0]);
     ASSERT_EQ(values[0], first_value);
@@ -553,8 +553,8 @@ TEST_F(DBBlobBasicTest, MultiGetBlobsFromCache) {
     std::array<PinnableSlice, num_keys> values;
     std::array<Status, num_keys> statuses;
 
-    db_->MultiGet(read_options, db_->DefaultColumnFamily(), num_keys, &keys[0],
-                  &values[0], &statuses[0]);
+    db_->MultiGet(read_options, db_->DefaultColumnFamily(), num_keys,
+                  keys.data(), values.data(), statuses.data());
 
     ASSERT_OK(statuses[0]);
     ASSERT_EQ(values[0], first_value);
@@ -574,8 +574,8 @@ TEST_F(DBBlobBasicTest, MultiGetBlobsFromCache) {
     std::array<PinnableSlice, num_keys> values;
     std::array<Status, num_keys> statuses;
 
-    db_->MultiGet(read_options, db_->DefaultColumnFamily(), num_keys, &keys[0],
-                  &values[0], &statuses[0]);
+    db_->MultiGet(read_options, db_->DefaultColumnFamily(), num_keys,
+                  keys.data(), values.data(), statuses.data());
 
     ASSERT_OK(statuses[0]);
     ASSERT_EQ(values[0], first_value);
@@ -758,8 +758,8 @@ TEST_F(DBBlobBasicTest, MultiGetWithDirectIO) {
     //
     // [offset=0, len=12288]
 
-    db_->MultiGet(ReadOptions(), db_->DefaultColumnFamily(), num_keys, &keys[0],
-                  &values[0], &statuses[0]);
+    db_->MultiGet(ReadOptions(), db_->DefaultColumnFamily(), num_keys,
+                  keys.data(), values.data(), statuses.data());
 
     SyncPoint::GetInstance()->DisableProcessing();
     SyncPoint::GetInstance()->ClearAllCallBacks();
@@ -829,8 +829,8 @@ TEST_F(DBBlobBasicTest, MultiGetBlobsFromMultipleFiles) {
   {
     std::array<PinnableSlice, kNumKeys> values;
     std::array<Status, kNumKeys> statuses;
-    db_->MultiGet(read_options, db_->DefaultColumnFamily(), kNumKeys, &keys[0],
-                  &values[0], &statuses[0]);
+    db_->MultiGet(read_options, db_->DefaultColumnFamily(), kNumKeys,
+                  keys.data(), values.data(), statuses.data());
 
     for (size_t i = 0; i < kNumKeys; ++i) {
       ASSERT_OK(statuses[i]);
@@ -843,8 +843,8 @@ TEST_F(DBBlobBasicTest, MultiGetBlobsFromMultipleFiles) {
   {
     std::array<PinnableSlice, kNumKeys> values;
     std::array<Status, kNumKeys> statuses;
-    db_->MultiGet(read_options, db_->DefaultColumnFamily(), kNumKeys, &keys[0],
-                  &values[0], &statuses[0]);
+    db_->MultiGet(read_options, db_->DefaultColumnFamily(), kNumKeys,
+                  keys.data(), values.data(), statuses.data());
 
     for (size_t i = 0; i < kNumKeys; ++i) {
       ASSERT_TRUE(statuses[i].IsIncomplete());
@@ -858,8 +858,8 @@ TEST_F(DBBlobBasicTest, MultiGetBlobsFromMultipleFiles) {
   {
     std::array<PinnableSlice, kNumKeys> values;
     std::array<Status, kNumKeys> statuses;
-    db_->MultiGet(read_options, db_->DefaultColumnFamily(), kNumKeys, &keys[0],
-                  &values[0], &statuses[0]);
+    db_->MultiGet(read_options, db_->DefaultColumnFamily(), kNumKeys,
+                  keys.data(), values.data(), statuses.data());
 
     for (size_t i = 0; i < kNumKeys; ++i) {
       ASSERT_OK(statuses[i]);
@@ -872,8 +872,8 @@ TEST_F(DBBlobBasicTest, MultiGetBlobsFromMultipleFiles) {
   {
     std::array<PinnableSlice, kNumKeys> values;
     std::array<Status, kNumKeys> statuses;
-    db_->MultiGet(read_options, db_->DefaultColumnFamily(), kNumKeys, &keys[0],
-                  &values[0], &statuses[0]);
+    db_->MultiGet(read_options, db_->DefaultColumnFamily(), kNumKeys,
+                  keys.data(), values.data(), statuses.data());
 
     for (size_t i = 0; i < kNumKeys; ++i) {
       ASSERT_OK(statuses[i]);
@@ -1182,6 +1182,30 @@ TEST_F(DBBlobBasicTest, GetMergeBlobWithPut) {
   ASSERT_EQ(Get("Key1"), "v1,v2,v3");
 }
 
+TEST_F(DBBlobBasicTest, GetMergeBlobFromMemoryTier) {
+  Options options = GetDefaultOptions();
+  options.merge_operator = MergeOperators::CreateStringAppendOperator();
+  options.enable_blob_files = true;
+  options.min_blob_size = 0;
+
+  Reopen(options);
+
+  ASSERT_OK(Put(Key(0), "v1"));
+  ASSERT_OK(Flush());
+  ASSERT_OK(Merge(Key(0), "v2"));
+  ASSERT_OK(Flush());
+
+  // Regular `Get()` loads data block to cache.
+  std::string value;
+  ASSERT_OK(db_->Get(ReadOptions(), Key(0), &value));
+  ASSERT_EQ("v1,v2", value);
+
+  // Base value blob is still uncached, so an in-memory read will fail.
+  ReadOptions read_options;
+  read_options.read_tier = kBlockCacheTier;
+  ASSERT_TRUE(db_->Get(read_options, Key(0), &value).IsIncomplete());
+}
+
 TEST_F(DBBlobBasicTest, MultiGetMergeBlobWithPut) {
   constexpr size_t num_keys = 3;
 
@@ -1206,8 +1230,8 @@ TEST_F(DBBlobBasicTest, MultiGetMergeBlobWithPut) {
   std::array<PinnableSlice, num_keys> values;
   std::array<Status, num_keys> statuses;
 
-  db_->MultiGet(ReadOptions(), db_->DefaultColumnFamily(), num_keys, &keys[0],
-                &values[0], &statuses[0]);
+  db_->MultiGet(ReadOptions(), db_->DefaultColumnFamily(), num_keys,
+                keys.data(), values.data(), statuses.data());
 
   ASSERT_OK(statuses[0]);
   ASSERT_EQ(values[0], "v0_0,v0_1,v0_2");
@@ -1470,8 +1494,8 @@ TEST_P(DBBlobBasicIOErrorMultiGetTest, MultiGetBlobs_IOError) {
   });
   SyncPoint::GetInstance()->EnableProcessing();
 
-  db_->MultiGet(ReadOptions(), db_->DefaultColumnFamily(), num_keys, &keys[0],
-                &values[0], &statuses[0]);
+  db_->MultiGet(ReadOptions(), db_->DefaultColumnFamily(), num_keys,
+                keys.data(), values.data(), statuses.data());
 
   SyncPoint::GetInstance()->DisableProcessing();
   SyncPoint::GetInstance()->ClearAllCallBacks();
@@ -1527,6 +1551,57 @@ TEST_P(DBBlobBasicIOErrorMultiGetTest, MultipleBlobFiles) {
   ASSERT_TRUE(statuses[1].IsIOError());
 }
 
+TEST_F(DBBlobBasicTest, MultiGetFindTable_IOError) {
+  // Repro test for a specific bug where `MultiGet()` would fail to open a table
+  // in `FindTable()` and then proceed to return raw blob handles for the other
+  // keys.
+  Options options = GetDefaultOptions();
+  options.enable_blob_files = true;
+  options.min_blob_size = 0;
+
+  Reopen(options);
+
+  // Force no table cache so every read will preload the SST file.
+  dbfull()->TEST_table_cache()->SetCapacity(0);
+
+  constexpr size_t num_keys = 2;
+
+  constexpr char key1[] = "key1";
+  constexpr char value1[] = "blob1";
+
+  ASSERT_OK(Put(key1, value1));
+  ASSERT_OK(Flush());
+
+  constexpr char key2[] = "key2";
+  constexpr char value2[] = "blob2";
+
+  ASSERT_OK(Put(key2, value2));
+  ASSERT_OK(Flush());
+
+  std::atomic<int> num_files_opened = 0;
+  // This test would be more realistic if we injected an `IOError` from the
+  // `FileSystem`
+  SyncPoint::GetInstance()->SetCallBack(
+      "TableCache::MultiGet:FindTable", [&](void* status) {
+        num_files_opened++;
+        if (num_files_opened == 2) {
+          Status* s = static_cast<Status*>(status);
+          *s = Status::IOError();
+        }
+      });
+  SyncPoint::GetInstance()->EnableProcessing();
+
+  std::array<Slice, num_keys> keys{{key1, key2}};
+  std::array<PinnableSlice, num_keys> values;
+  std::array<Status, num_keys> statuses;
+  db_->MultiGet(ReadOptions(), db_->DefaultColumnFamily(), num_keys,
+                keys.data(), values.data(), statuses.data());
+
+  ASSERT_TRUE(statuses[0].IsIOError());
+  ASSERT_OK(statuses[1]);
+  ASSERT_EQ(value2, values[1]);
+}
+
 namespace {
 
 class ReadBlobCompactionFilter : public CompactionFilter {
@@ -1820,7 +1895,7 @@ TEST_F(DBBlobBasicTest, GetEntityBlob) {
     std::array<Status, num_keys> statuses;
 
     db_->MultiGetEntity(ReadOptions(), db_->DefaultColumnFamily(), num_keys,
-                        &keys[0], &results[0], &statuses[0]);
+                        keys.data(), results.data(), statuses.data());
 
     ASSERT_OK(statuses[0]);
     ASSERT_EQ(results[0].columns(), expected_columns);
@@ -1917,8 +1992,8 @@ TEST_F(DBBlobWithTimestampTest, MultiGetBlobs) {
     std::array<PinnableSlice, num_keys> values;
     std::array<Status, num_keys> statuses;
 
-    db_->MultiGet(read_options, db_->DefaultColumnFamily(), num_keys, &keys[0],
-                  &values[0], &statuses[0]);
+    db_->MultiGet(read_options, db_->DefaultColumnFamily(), num_keys,
+                  keys.data(), values.data(), statuses.data());
 
     ASSERT_OK(statuses[0]);
     ASSERT_EQ(values[0], first_value);
@@ -2001,8 +2076,8 @@ TEST_F(DBBlobWithTimestampTest, MultiGetMergeBlobWithPut) {
   std::array<PinnableSlice, num_keys> values;
   std::array<Status, num_keys> statuses;
 
-  db_->MultiGet(read_opts, db_->DefaultColumnFamily(), num_keys, &keys[0],
-                &values[0], &statuses[0]);
+  db_->MultiGet(read_opts, db_->DefaultColumnFamily(), num_keys, keys.data(),
+                values.data(), statuses.data());
 
   ASSERT_OK(statuses[0]);
   ASSERT_EQ(values[0], "v0_0,v0_1,v0_2");

db/blob/db_blob_index_test.cc

@@ -45,10 +45,10 @@ class DBBlobIndexTest : public DBTestBase {
   DBBlobIndexTest() : DBTestBase("db_blob_index_test", /*env_do_fsync=*/true) {}
 
   ColumnFamilyHandle* cfh() { return dbfull()->DefaultColumnFamily(); }
-
-  ColumnFamilyData* cfd() {
-    return static_cast_with_check<ColumnFamilyHandleImpl>(cfh())->cfd();
+  ColumnFamilyHandleImpl* cfh_impl() {
+    return static_cast_with_check<ColumnFamilyHandleImpl>(cfh());
   }
+  ColumnFamilyData* cfd() { return cfh_impl()->cfd(); }
 
   Status PutBlobIndex(WriteBatch* batch, const Slice& key,
                       const Slice& blob_index) {
@@ -96,11 +96,9 @@ class DBBlobIndexTest : public DBTestBase {
   }
 
   ArenaWrappedDBIter* GetBlobIterator() {
-    ColumnFamilyData* column_family = cfd();
     DBImpl* db_impl = dbfull();
     return db_impl->NewIteratorImpl(
-        ReadOptions(), column_family,
-        column_family->GetReferencedSuperVersion(db_impl),
+        ReadOptions(), cfh_impl(), cfd()->GetReferencedSuperVersion(db_impl),
         db_impl->GetLatestSequenceNumber(), nullptr /*read_callback*/,
         true /*expose_blob_index*/);
   }

db/builder.cc

@@ -70,8 +70,8 @@ Status BuildTable(
     bool paranoid_file_checks, InternalStats* internal_stats,
     IOStatus* io_status, const std::shared_ptr<IOTracer>& io_tracer,
     BlobFileCreationReason blob_creation_reason,
-    const SeqnoToTimeMapping& seqno_to_time_mapping, EventLogger* event_logger,
-    int job_id, TableProperties* table_properties,
+    UnownedPtr<const SeqnoToTimeMapping> seqno_to_time_mapping,
+    EventLogger* event_logger, int job_id, TableProperties* table_properties,
     Env::WriteLifeTimeHint write_hint, const std::string* full_history_ts_low,
     BlobFileCompletionCallback* blob_callback, Version* version,
     uint64_t* num_input_entries, uint64_t* memtable_payload_bytes,
@@ -210,13 +210,18 @@ Status BuildTable(
     const bool logical_strip_timestamp =
         ts_sz > 0 && !ioptions.persist_user_defined_timestamps;
 
+    SequenceNumber smallest_preferred_seqno = kMaxSequenceNumber;
     std::string key_after_flush_buf;
+    std::string value_buf;
     c_iter.SeekToFirst();
     for (; c_iter.Valid(); c_iter.Next()) {
       const Slice& key = c_iter.key();
       const Slice& value = c_iter.value();
-      const ParsedInternalKey& ikey = c_iter.ikey();
-      Slice key_after_flush = key;
+      ParsedInternalKey ikey = c_iter.ikey();
+      key_after_flush_buf.assign(key.data(), key.size());
+      Slice key_after_flush = key_after_flush_buf;
+      Slice value_after_flush = value;
+
       // If user defined timestamps will be stripped from user key after flush,
       // the in memory version of the key act logically the same as one with a
       // minimum timestamp. We update the timestamp here so file boundary and
@@ -227,17 +232,39 @@ Status BuildTable(
         key_after_flush = key_after_flush_buf;
       }
 
+      if (ikey.type == kTypeValuePreferredSeqno) {
+        auto [unpacked_value, unix_write_time] =
+            ParsePackedValueWithWriteTime(value);
+        SequenceNumber preferred_seqno =
+            seqno_to_time_mapping
+                ? seqno_to_time_mapping->GetProximalSeqnoBeforeTime(
+                      unix_write_time)
+                : kMaxSequenceNumber;
+        if (preferred_seqno < ikey.sequence) {
+          value_after_flush =
+              PackValueAndSeqno(unpacked_value, preferred_seqno, &value_buf);
+          smallest_preferred_seqno =
+              std::min(smallest_preferred_seqno, preferred_seqno);
+        } else {
+          // Cannot get a useful preferred seqno, convert it to a kTypeValue.
+          UpdateInternalKey(&key_after_flush_buf, ikey.sequence, kTypeValue);
+          ikey = ParsedInternalKey(ikey.user_key, ikey.sequence, kTypeValue);
+          key_after_flush = key_after_flush_buf;
+          value_after_flush = ParsePackedValueForValue(value);
+        }
+      }
+
       //  Generate a rolling 64-bit hash of the key and values
       //  Note :
       //  Here "key" integrates 'sequence_number'+'kType'+'user key'.
-      s = output_validator.Add(key_after_flush, value);
+      s = output_validator.Add(key_after_flush, value_after_flush);
       if (!s.ok()) {
         break;
       }
-      builder->Add(key_after_flush, value);
+      builder->Add(key_after_flush, value_after_flush);
 
-      s = meta->UpdateBoundaries(key_after_flush, value, ikey.sequence,
-                                 ikey.type);
+      s = meta->UpdateBoundaries(key_after_flush, value_after_flush,
+                                 ikey.sequence, ikey.type);
       if (!s.ok()) {
         break;
       }
@@ -301,11 +328,14 @@ Status BuildTable(
       builder->Abandon();
     } else {
       SeqnoToTimeMapping relevant_mapping;
-      relevant_mapping.CopyFromSeqnoRange(seqno_to_time_mapping,
-                                          meta->fd.smallest_seqno,
-                                          meta->fd.largest_seqno);
-      relevant_mapping.SetCapacity(kMaxSeqnoTimePairsPerSST);
-      relevant_mapping.Enforce(tboptions.file_creation_time);
+      if (seqno_to_time_mapping) {
+        relevant_mapping.CopyFromSeqnoRange(
+            *seqno_to_time_mapping,
+            std::min(meta->fd.smallest_seqno, smallest_preferred_seqno),
+            meta->fd.largest_seqno);
+        relevant_mapping.SetCapacity(kMaxSeqnoTimePairsPerSST);
+        relevant_mapping.Enforce(tboptions.file_creation_time);
+      }
       builder->SetSeqnoTimeTableProperties(
           relevant_mapping,
           ioptions.compaction_style == CompactionStyle::kCompactionStyleFIFO

db/builder.h

@@ -23,7 +23,6 @@
 #include "rocksdb/status.h"
 #include "rocksdb/table_properties.h"
 #include "rocksdb/types.h"
-#include "table/scoped_arena_iterator.h"
 
 namespace ROCKSDB_NAMESPACE {
 
@@ -64,7 +63,7 @@ Status BuildTable(
     bool paranoid_file_checks, InternalStats* internal_stats,
     IOStatus* io_status, const std::shared_ptr<IOTracer>& io_tracer,
     BlobFileCreationReason blob_creation_reason,
-    const SeqnoToTimeMapping& seqno_to_time_mapping,
+    UnownedPtr<const SeqnoToTimeMapping> seqno_to_time_mapping,
     EventLogger* event_logger = nullptr, int job_id = 0,
     TableProperties* table_properties = nullptr,
     Env::WriteLifeTimeHint write_hint = Env::WLTH_NOT_SET,

db/c.cc

@@ -446,7 +446,7 @@ struct rocksdb_mergeoperator_t : public MergeOperator {
     size_t new_value_len;
     char* tmp_new_value = (*full_merge_)(
         state_, merge_in.key.data(), merge_in.key.size(), existing_value_data,
-        existing_value_len, &operand_pointers[0], &operand_sizes[0],
+        existing_value_len, operand_pointers.data(), operand_sizes.data(),
         static_cast<int>(n), &success, &new_value_len);
     merge_out->new_value.assign(tmp_new_value, new_value_len);
 
@@ -475,8 +475,9 @@ struct rocksdb_mergeoperator_t : public MergeOperator {
     unsigned char success;
     size_t new_value_len;
     char* tmp_new_value = (*partial_merge_)(
-        state_, key.data(), key.size(), &operand_pointers[0], &operand_sizes[0],
-        static_cast<int>(operand_count), &success, &new_value_len);
+        state_, key.data(), key.size(), operand_pointers.data(),
+        operand_sizes.data(), static_cast<int>(operand_count), &success,
+        &new_value_len);
     new_value->assign(tmp_new_value, new_value_len);
 
     if (delete_value_ != nullptr) {
@@ -886,9 +887,9 @@ rocksdb_t* rocksdb_open_and_trim_history(
     size_t trim_tslen, char** errptr) {
   std::vector<ColumnFamilyDescriptor> column_families;
   for (int i = 0; i < num_column_families; i++) {
-    column_families.push_back(ColumnFamilyDescriptor(
+    column_families.emplace_back(
         std::string(column_family_names[i]),
-        ColumnFamilyOptions(column_family_options[i]->rep)));
+        ColumnFamilyOptions(column_family_options[i]->rep));
   }
 
   std::string trim_ts_(trim_ts, trim_tslen);
@@ -919,9 +920,9 @@ rocksdb_t* rocksdb_open_column_families(
     rocksdb_column_family_handle_t** column_family_handles, char** errptr) {
   std::vector<ColumnFamilyDescriptor> column_families;
   for (int i = 0; i < num_column_families; i++) {
-    column_families.push_back(ColumnFamilyDescriptor(
+    column_families.emplace_back(
         std::string(column_family_names[i]),
-        ColumnFamilyOptions(column_family_options[i]->rep)));
+        ColumnFamilyOptions(column_family_options[i]->rep));
   }
 
   DB* db;
@@ -953,9 +954,9 @@ rocksdb_t* rocksdb_open_column_families_with_ttl(
   for (int i = 0; i < num_column_families; i++) {
     ttls_vec.push_back(ttls[i]);
 
-    column_families.push_back(ColumnFamilyDescriptor(
+    column_families.emplace_back(
         std::string(column_family_names[i]),
-        ColumnFamilyOptions(column_family_options[i]->rep)));
+        ColumnFamilyOptions(column_family_options[i]->rep));
   }
 
   ROCKSDB_NAMESPACE::DBWithTTL* db;
@@ -985,9 +986,9 @@ rocksdb_t* rocksdb_open_for_read_only_column_families(
     unsigned char error_if_wal_file_exists, char** errptr) {
   std::vector<ColumnFamilyDescriptor> column_families;
   for (int i = 0; i < num_column_families; i++) {
-    column_families.push_back(ColumnFamilyDescriptor(
+    column_families.emplace_back(
         std::string(column_family_names[i]),
-        ColumnFamilyOptions(column_family_options[i]->rep)));
+        ColumnFamilyOptions(column_family_options[i]->rep));
   }
 
   DB* db;
@@ -1081,7 +1082,7 @@ rocksdb_column_family_handle_t** rocksdb_create_column_families(
   std::vector<ColumnFamilyHandle*> handles;
   std::vector<std::string> names;
   for (int i = 0; i != num_column_families; ++i) {
-    names.push_back(std::string(column_family_names[i]));
+    names.emplace_back(column_family_names[i]);
   }
   SaveError(errptr, db->rep->CreateColumnFamilies(
                         ColumnFamilyOptions(column_family_options->rep), names,
@@ -2744,6 +2745,24 @@ void rocksdb_options_set_block_based_table_factory(
   }
 }
 
+void rocksdb_block_based_options_set_top_level_index_pinning_tier(
+    rocksdb_block_based_table_options_t* options, int v) {
+  options->rep.metadata_cache_options.top_level_index_pinning =
+      static_cast<ROCKSDB_NAMESPACE::PinningTier>(v);
+}
+
+void rocksdb_block_based_options_set_partition_pinning_tier(
+    rocksdb_block_based_table_options_t* options, int v) {
+  options->rep.metadata_cache_options.partition_pinning =
+      static_cast<ROCKSDB_NAMESPACE::PinningTier>(v);
+}
+
+void rocksdb_block_based_options_set_unpartitioned_pinning_tier(
+    rocksdb_block_based_table_options_t* options, int v) {
+  options->rep.metadata_cache_options.unpartitioned_pinning =
+      static_cast<ROCKSDB_NAMESPACE::PinningTier>(v);
+}
+
 rocksdb_cuckoo_table_options_t* rocksdb_cuckoo_options_create() {
   return new rocksdb_cuckoo_table_options_t;
 }
@@ -2788,7 +2807,9 @@ void rocksdb_options_set_cuckoo_table_factory(
 void rocksdb_set_options(rocksdb_t* db, int count, const char* const keys[],
                          const char* const values[], char** errptr) {
   std::unordered_map<std::string, std::string> options_map;
-  for (int i = 0; i < count; i++) options_map[keys[i]] = values[i];
+  for (int i = 0; i < count; i++) {
+    options_map[keys[i]] = values[i];
+  }
   SaveError(errptr, db->rep->SetOptions(options_map));
 }
 
@@ -2797,7 +2818,9 @@ void rocksdb_set_options_cf(rocksdb_t* db,
                             const char* const keys[],
                             const char* const values[], char** errptr) {
   std::unordered_map<std::string, std::string> options_map;
-  for (int i = 0; i < count; i++) options_map[keys[i]] = values[i];
+  for (int i = 0; i < count; i++) {
+    options_map[keys[i]] = values[i];
+  }
   SaveError(errptr, db->rep->SetOptions(handle->rep, options_map));
 }
 
@@ -3925,6 +3948,14 @@ void rocksdb_options_set_fifo_compaction_options(
   opt->rep.compaction_options_fifo = fifo->rep;
 }
 
+void rocksdb_options_set_compaction_pri(rocksdb_options_t* opt, int pri) {
+  opt->rep.compaction_pri = static_cast<ROCKSDB_NAMESPACE::CompactionPri>(pri);
+}
+
+int rocksdb_options_get_compaction_pri(rocksdb_options_t* opt) {
+  return opt->rep.compaction_pri;
+}
+
 char* rocksdb_options_statistics_get_string(rocksdb_options_t* opt) {
   ROCKSDB_NAMESPACE::Statistics* statistics = opt->rep.statistics.get();
   if (statistics) {
@@ -5060,7 +5091,9 @@ void rocksdb_env_lower_high_priority_thread_pool_cpu_priority(
 }
 
 void rocksdb_env_destroy(rocksdb_env_t* env) {
-  if (!env->is_default) delete env->rep;
+  if (!env->is_default) {
+    delete env->rep;
+  }
   delete env;
 }
 
@@ -5524,7 +5557,7 @@ size_t rocksdb_column_family_metadata_get_level_count(
 rocksdb_level_metadata_t* rocksdb_column_family_metadata_get_level_metadata(
     rocksdb_column_family_metadata_t* cf_meta, size_t i) {
   if (i >= cf_meta->rep.levels.size()) {
-    return NULL;
+    return nullptr;
   }
   rocksdb_level_metadata_t* level_meta =
       (rocksdb_level_metadata_t*)malloc(sizeof(rocksdb_level_metadata_t));
@@ -5739,9 +5772,9 @@ rocksdb_transactiondb_t* rocksdb_transactiondb_open_column_families(
     rocksdb_column_family_handle_t** column_family_handles, char** errptr) {
   std::vector<ColumnFamilyDescriptor> column_families;
   for (int i = 0; i < num_column_families; i++) {
-    column_families.push_back(ColumnFamilyDescriptor(
+    column_families.emplace_back(
         std::string(column_family_names[i]),
-        ColumnFamilyOptions(column_family_options[i]->rep)));
+        ColumnFamilyOptions(column_family_options[i]->rep));
   }
 
   TransactionDB* txn_db;
@@ -6533,9 +6566,9 @@ rocksdb_optimistictransactiondb_open_column_families(
     rocksdb_column_family_handle_t** column_family_handles, char** errptr) {
   std::vector<ColumnFamilyDescriptor> column_families;
   for (int i = 0; i < num_column_families; i++) {
-    column_families.push_back(ColumnFamilyDescriptor(
+    column_families.emplace_back(
         std::string(column_family_names[i]),
-        ColumnFamilyOptions(column_family_options[i]->rep)));
+        ColumnFamilyOptions(column_family_options[i]->rep));
   }
 
   OptimisticTransactionDB* otxn_db;

db/c_test.c

@@ -50,12 +50,13 @@ static void StartPhase(const char* name) {
 #endif
 static const char* GetTempDir(void) {
   const char* ret = getenv("TEST_TMPDIR");
-  if (ret == NULL || ret[0] == '\0')
+  if (ret == NULL || ret[0] == '\0') {
 #ifdef OS_WIN
     ret = getenv("TEMP");
 #else
     ret = "/tmp";
 #endif
+  }
   return ret;
 }
 #ifdef _MSC_VER
@@ -206,10 +207,11 @@ static int CmpCompare(void* arg, const char* a, size_t alen, const char* b,
   size_t n = (alen < blen) ? alen : blen;
   int r = memcmp(a, b, n);
   if (r == 0) {
-    if (alen < blen)
+    if (alen < blen) {
       r = -1;
-    else if (alen > blen)
+    } else if (alen > blen) {
       r = +1;
+    }
   }
   return r;
 }
@@ -702,6 +704,10 @@ int main(int argc, char** argv) {
   rocksdb_block_based_options_set_block_cache(table_options, cache);
   rocksdb_block_based_options_set_data_block_index_type(table_options, 1);
   rocksdb_block_based_options_set_data_block_hash_ratio(table_options, 0.75);
+  rocksdb_block_based_options_set_top_level_index_pinning_tier(table_options,
+                                                               1);
+  rocksdb_block_based_options_set_partition_pinning_tier(table_options, 2);
+  rocksdb_block_based_options_set_unpartitioned_pinning_tier(table_options, 3);
   rocksdb_options_set_block_based_table_factory(options, table_options);
 
   rocksdb_options_set_compression(options, rocksdb_no_compression);
@@ -883,9 +889,8 @@ int main(int argc, char** argv) {
   StartPhase("addfile");
   {
     rocksdb_envoptions_t* env_opt = rocksdb_envoptions_create();
-    rocksdb_options_t* io_options = rocksdb_options_create();
     rocksdb_sstfilewriter_t* writer =
-        rocksdb_sstfilewriter_create(env_opt, io_options);
+        rocksdb_sstfilewriter_create(env_opt, options);
 
     remove(sstfilename);
     rocksdb_sstfilewriter_open(writer, sstfilename, &err);
@@ -944,7 +949,6 @@ int main(int argc, char** argv) {
 
     rocksdb_ingestexternalfileoptions_destroy(ing_opt);
     rocksdb_sstfilewriter_destroy(writer);
-    rocksdb_options_destroy(io_options);
     rocksdb_envoptions_destroy(env_opt);
 
     // Delete all keys we just ingested
@@ -1318,6 +1322,8 @@ int main(int argc, char** argv) {
       policy = rocksdb_filterpolicy_create_ribbon_hybrid(8.0, 1);
     }
     rocksdb_block_based_options_set_filter_policy(table_options, policy);
+    rocksdb_block_based_options_set_optimize_filters_for_memory(table_options,
+                                                                0);
 
     // Create new database
     rocksdb_close(db);
@@ -2039,7 +2045,6 @@ int main(int argc, char** argv) {
     rocksdb_options_set_advise_random_on_open(o, 1);
     CheckCondition(1 == rocksdb_options_get_advise_random_on_open(o));
 
-
     rocksdb_options_set_use_adaptive_mutex(o, 1);
     CheckCondition(1 == rocksdb_options_get_use_adaptive_mutex(o));
 
@@ -2110,6 +2115,9 @@ int main(int argc, char** argv) {
     rocksdb_options_set_compaction_style(o, 2);
     CheckCondition(2 == rocksdb_options_get_compaction_style(o));
 
+    rocksdb_options_set_compaction_pri(o, 4);
+    CheckCondition(4 == rocksdb_options_get_compaction_pri(o));
+
     rocksdb_options_set_atomic_flush(o, 1);
     CheckCondition(1 == rocksdb_options_get_atomic_flush(o));
 
@@ -2614,6 +2622,10 @@ int main(int argc, char** argv) {
     CheckCondition(1 == rocksdb_options_get_compaction_style(copy));
     CheckCondition(2 == rocksdb_options_get_compaction_style(o));
 
+    rocksdb_options_set_compaction_pri(copy, 1);
+    CheckCondition(1 == rocksdb_options_get_compaction_pri(copy));
+    CheckCondition(4 == rocksdb_options_get_compaction_pri(o));
+
     rocksdb_options_set_atomic_flush(copy, 0);
     CheckCondition(0 == rocksdb_options_get_atomic_flush(copy));
     CheckCondition(1 == rocksdb_options_get_atomic_flush(o));

db/coalescing_iterator.cc

@@ -0,0 +1,47 @@
+//  Copyright (c) Meta Platforms, Inc. and affiliates.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+
+#include "db/coalescing_iterator.h"
+
+#include "db/wide/wide_columns_helper.h"
+
+namespace ROCKSDB_NAMESPACE {
+
+void CoalescingIterator::Coalesce(
+    const autovector<MultiCfIteratorInfo>& items) {
+  assert(wide_columns_.empty());
+  MinHeap heap;
+  for (const auto& item : items) {
+    assert(item.iterator);
+    for (auto& column : item.iterator->columns()) {
+      heap.push(WideColumnWithOrder{&column, item.order});
+    }
+  }
+  if (heap.empty()) {
+    return;
+  }
+  wide_columns_.reserve(heap.size());
+  auto current = heap.top();
+  heap.pop();
+  while (!heap.empty()) {
+    int comparison = current.column->name().compare(heap.top().column->name());
+    if (comparison < 0) {
+      wide_columns_.push_back(*current.column);
+    } else if (comparison > 0) {
+      // Shouldn't reach here.
+      // Current item in the heap is greater than the top item in the min heap
+      assert(false);
+    }
+    current = heap.top();
+    heap.pop();
+  }
+  wide_columns_.push_back(*current.column);
+
+  if (WideColumnsHelper::HasDefaultColumn(wide_columns_)) {
+    value_ = WideColumnsHelper::GetDefaultColumn(wide_columns_);
+  }
+}
+
+}  // namespace ROCKSDB_NAMESPACE

db/coalescing_iterator.h

@@ -0,0 +1,79 @@
+//  Copyright (c) Meta Platforms, Inc. and affiliates.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+
+#pragma once
+
+#include "db/multi_cf_iterator_impl.h"
+
+namespace ROCKSDB_NAMESPACE {
+
+// EXPERIMENTAL
+class CoalescingIterator : public Iterator {
+ public:
+  CoalescingIterator(const Comparator* comparator,
+                     const std::vector<ColumnFamilyHandle*>& column_families,
+                     const std::vector<Iterator*>& child_iterators)
+      : impl_(
+            comparator, column_families, child_iterators, [this]() { Reset(); },
+            [this](const autovector<MultiCfIteratorInfo>& items) {
+              Coalesce(items);
+            }) {}
+  ~CoalescingIterator() override {}
+
+  // No copy allowed
+  CoalescingIterator(const CoalescingIterator&) = delete;
+  CoalescingIterator& operator=(const CoalescingIterator&) = delete;
+
+  bool Valid() const override { return impl_.Valid(); }
+  void SeekToFirst() override { impl_.SeekToFirst(); }
+  void SeekToLast() override { impl_.SeekToLast(); }
+  void Seek(const Slice& target) override { impl_.Seek(target); }
+  void SeekForPrev(const Slice& target) override { impl_.SeekForPrev(target); }
+  void Next() override { impl_.Next(); }
+  void Prev() override { impl_.Prev(); }
+  Slice key() const override { return impl_.key(); }
+  Status status() const override { return impl_.status(); }
+
+  Slice value() const override {
+    assert(Valid());
+    return value_;
+  }
+  const WideColumns& columns() const override {
+    assert(Valid());
+    return wide_columns_;
+  }
+
+  void Reset() {
+    value_.clear();
+    wide_columns_.clear();
+  }
+
+ private:
+  MultiCfIteratorImpl impl_;
+  Slice value_;
+  WideColumns wide_columns_;
+
+  struct WideColumnWithOrder {
+    const WideColumn* column;
+    int order;
+  };
+
+  class WideColumnWithOrderComparator {
+   public:
+    explicit WideColumnWithOrderComparator() {}
+    bool operator()(const WideColumnWithOrder& a,
+                    const WideColumnWithOrder& b) const {
+      int c = a.column->name().compare(b.column->name());
+      return c == 0 ? a.order - b.order > 0 : c > 0;
+    }
+  };
+
+  using MinHeap =
+      BinaryHeap<WideColumnWithOrder, WideColumnWithOrderComparator>;
+
+  void Coalesce(const autovector<MultiCfIteratorInfo>& items);
+};
+
+}  // namespace ROCKSDB_NAMESPACE

db/column_family.cc

@@ -411,6 +411,13 @@ ColumnFamilyOptions SanitizeOptions(const ImmutableDBOptions& db_options,
           "periodic_compaction_seconds does not support FIFO compaction. You"
           "may want to set option TTL instead.");
     }
+    if (result.last_level_temperature != Temperature::kUnknown) {
+      ROCKS_LOG_WARN(
+          db_options.info_log.get(),
+          "last_level_temperature is ignored with FIFO compaction. Consider "
+          "CompactionOptionsFIFO::file_temperature_age_thresholds.");
+      result.last_level_temperature = Temperature::kUnknown;
+    }
   }
 
   // For universal compaction, `ttl` and `periodic_compaction_seconds` mean the
@@ -470,13 +477,16 @@ void SuperVersion::Cleanup() {
   cfd->UnrefAndTryDelete();
 }
 
-void SuperVersion::Init(ColumnFamilyData* new_cfd, MemTable* new_mem,
-                        MemTableListVersion* new_imm, Version* new_current) {
+void SuperVersion::Init(
+    ColumnFamilyData* new_cfd, MemTable* new_mem, MemTableListVersion* new_imm,
+    Version* new_current,
+    std::shared_ptr<const SeqnoToTimeMapping> new_seqno_to_time_mapping) {
   cfd = new_cfd;
   mem = new_mem;
   imm = new_imm;
   current = new_current;
   full_history_ts_low = cfd->GetFullHistoryTsLow();
+  seqno_to_time_mapping = std::move(new_seqno_to_time_mapping);
   cfd->Ref();
   mem->Ref();
   imm->Ref();
@@ -1189,11 +1199,12 @@ Status ColumnFamilyData::RangesOverlapWithMemtables(
   ReadOptions read_opts;
   read_opts.total_order_seek = true;
   MergeIteratorBuilder merge_iter_builder(&internal_comparator_, &arena);
-  merge_iter_builder.AddIterator(
-      super_version->mem->NewIterator(read_opts, &arena));
-  super_version->imm->AddIterators(read_opts, &merge_iter_builder,
+  merge_iter_builder.AddIterator(super_version->mem->NewIterator(
+      read_opts, /*seqno_to_time_mapping=*/nullptr, &arena));
+  super_version->imm->AddIterators(read_opts, /*seqno_to_time_mapping=*/nullptr,
+                                   &merge_iter_builder,
                                    false /* add_range_tombstone_iter */);
-  ScopedArenaIterator memtable_iter(merge_iter_builder.Finish());
+  ScopedArenaPtr<InternalIterator> memtable_iter(merge_iter_builder.Finish());
 
   auto read_seq = super_version->current->version_set()->LastSequence();
   ReadRangeDelAggregator range_del_agg(&internal_comparator_, read_seq);
@@ -1329,7 +1340,12 @@ void ColumnFamilyData::InstallSuperVersion(
     const MutableCFOptions& mutable_cf_options) {
   SuperVersion* new_superversion = sv_context->new_superversion.release();
   new_superversion->mutable_cf_options = mutable_cf_options;
-  new_superversion->Init(this, mem_, imm_.current(), current_);
+  new_superversion->Init(this, mem_, imm_.current(), current_,
+                         sv_context->new_seqno_to_time_mapping
+                             ? std::move(sv_context->new_seqno_to_time_mapping)
+                         : super_version_
+                             ? super_version_->ShareSeqnoToTimeMapping()
+                             : nullptr);
   SuperVersion* old_superversion = super_version_;
   super_version_ = new_superversion;
   if (old_superversion == nullptr || old_superversion->current != current() ||

db/column_family.h

@@ -26,6 +26,7 @@
 #include "rocksdb/env.h"
 #include "rocksdb/options.h"
 #include "trace_replay/block_cache_tracer.h"
+#include "util/cast_util.h"
 #include "util/hash_containers.h"
 #include "util/thread_local.h"
 
@@ -168,6 +169,7 @@ class ColumnFamilyHandleImpl : public ColumnFamilyHandle {
   // destroy without mutex
   virtual ~ColumnFamilyHandleImpl();
   virtual ColumnFamilyData* cfd() const { return cfd_; }
+  virtual DBImpl* db() const { return db_; }
 
   uint32_t GetID() const override;
   const std::string& GetName() const override;
@@ -218,6 +220,9 @@ struct SuperVersion {
   // enable UDT feature, this is an empty string.
   std::string full_history_ts_low;
 
+  // A shared copy of the DB's seqno to time mapping.
+  std::shared_ptr<const SeqnoToTimeMapping> seqno_to_time_mapping{nullptr};
+
   // should be called outside the mutex
   SuperVersion() = default;
   ~SuperVersion();
@@ -231,8 +236,23 @@ struct SuperVersion {
   // that needs to be deleted in to_delete vector. Unrefing those
   // objects needs to be done in the mutex
   void Cleanup();
-  void Init(ColumnFamilyData* new_cfd, MemTable* new_mem,
-            MemTableListVersion* new_imm, Version* new_current);
+  void Init(
+      ColumnFamilyData* new_cfd, MemTable* new_mem,
+      MemTableListVersion* new_imm, Version* new_current,
+      std::shared_ptr<const SeqnoToTimeMapping> new_seqno_to_time_mapping);
+
+  // Share the ownership of the seqno to time mapping object referred to in this
+  // SuperVersion. To be used by the new SuperVersion to be installed after this
+  // one if seqno to time mapping does not change in between these two
+  // SuperVersions. Or to share the ownership of the mapping with a FlushJob.
+  std::shared_ptr<const SeqnoToTimeMapping> ShareSeqnoToTimeMapping() {
+    return seqno_to_time_mapping;
+  }
+
+  // Access the seqno to time mapping object in this SuperVersion.
+  UnownedPtr<const SeqnoToTimeMapping> GetSeqnoToTimeMapping() const {
+    return seqno_to_time_mapping.get();
+  }
 
   // The value of dummy is not actually used. kSVInUse takes its address as a
   // mark in the thread local storage to indicate the SuperVersion is in use

db/column_family_test.cc

@@ -270,7 +270,7 @@ class ColumnFamilyTestBase : public testing::Test {
 
   void Reopen(const std::vector<ColumnFamilyOptions> options = {}) {
     std::vector<std::string> names;
-    for (auto name : names_) {
+    for (const auto& name : names_) {
       if (name != "") {
         names.push_back(name);
       }
@@ -607,7 +607,7 @@ TEST_P(FlushEmptyCFTestWithParam, FlushEmptyCFTest) {
   // Preserve file system state up to here to simulate a crash condition.
   fault_env->SetFilesystemActive(false);
   std::vector<std::string> names;
-  for (auto name : names_) {
+  for (const auto& name : names_) {
     if (name != "") {
       names.push_back(name);
     }
@@ -669,7 +669,7 @@ TEST_P(FlushEmptyCFTestWithParam, FlushEmptyCFTest2) {
   // Preserve file system state up to here to simulate a crash condition.
   fault_env->SetFilesystemActive(false);
   std::vector<std::string> names;
-  for (auto name : names_) {
+  for (const auto& name : names_) {
     if (name != "") {
       names.push_back(name);
     }
@@ -1034,7 +1034,7 @@ TEST_P(ColumnFamilyTest, CrashAfterFlush) {
   fault_env->SetFilesystemActive(false);
 
   std::vector<std::string> names;
-  for (auto name : names_) {
+  for (const auto& name : names_) {
     if (name != "") {
       names.push_back(name);
     }
@@ -3407,9 +3407,13 @@ TEST_P(ColumnFamilyTest, DISABLED_LogTruncationTest) {
   for (size_t i = 0; i < filenames.size(); i++) {
     uint64_t number;
     FileType type;
-    if (!(ParseFileName(filenames[i], &number, &type))) continue;
+    if (!(ParseFileName(filenames[i], &number, &type))) {
+      continue;
+    }
 
-    if (type != kWalFile) continue;
+    if (type != kWalFile) {
+      continue;
+    }
 
     logfs.push_back(filenames[i]);
   }

db/compact_files_test.cc

@@ -34,8 +34,8 @@ class CompactFilesTest : public testing::Test {
 // A class which remembers the name of each flushed file.
 class FlushedFileCollector : public EventListener {
  public:
-  FlushedFileCollector() {}
-  ~FlushedFileCollector() override {}
+  FlushedFileCollector() = default;
+  ~FlushedFileCollector() override = default;
 
   void OnFlushCompleted(DB* /*db*/, const FlushJobInfo& info) override {
     std::lock_guard<std::mutex> lock(mutex_);
@@ -45,7 +45,7 @@ class FlushedFileCollector : public EventListener {
   std::vector<std::string> GetFlushedFiles() {
     std::lock_guard<std::mutex> lock(mutex_);
     std::vector<std::string> result;
-    for (auto fname : flushed_files_) {
+    for (const auto& fname : flushed_files_) {
       result.push_back(fname);
     }
     return result;
@@ -159,7 +159,9 @@ TEST_F(CompactFilesTest, MultipleLevel) {
   // Compact files except the file in L3
   std::vector<std::string> files;
   for (int i = 0; i < 6; ++i) {
-    if (i == 3) continue;
+    if (i == 3) {
+      continue;
+    }
     for (auto& file : meta.levels[i].files) {
       files.push_back(file.db_path + "/" + file.name);
     }
@@ -228,7 +230,7 @@ TEST_F(CompactFilesTest, ObsoleteFiles) {
   ASSERT_OK(static_cast_with_check<DBImpl>(db)->TEST_WaitForCompact());
 
   // verify all compaction input files are deleted
-  for (auto fname : l0_files) {
+  for (const auto& fname : l0_files) {
     ASSERT_EQ(Status::NotFound(), env_->FileExists(fname));
   }
   delete db;
@@ -439,6 +441,50 @@ TEST_F(CompactFilesTest, SentinelCompressionType) {
   }
 }
 
+TEST_F(CompactFilesTest, CompressionWithBlockAlign) {
+  Options options;
+  options.compression = CompressionType::kNoCompression;
+  options.create_if_missing = true;
+  options.disable_auto_compactions = true;
+
+  std::shared_ptr<FlushedFileCollector> collector =
+      std::make_shared<FlushedFileCollector>();
+  options.listeners.push_back(collector);
+
+  {
+    BlockBasedTableOptions bbto;
+    bbto.block_align = true;
+    options.table_factory.reset(NewBlockBasedTableFactory(bbto));
+  }
+
+  std::unique_ptr<DB> db;
+  {
+    DB* _db = nullptr;
+    ASSERT_OK(DB::Open(options, db_name_, &_db));
+    db.reset(_db);
+  }
+
+  ASSERT_OK(db->Put(WriteOptions(), "key", "val"));
+  ASSERT_OK(db->Flush(FlushOptions()));
+
+  // Ensure background work is fully finished including listener callbacks
+  // before accessing listener state.
+  ASSERT_OK(
+      static_cast_with_check<DBImpl>(db.get())->TEST_WaitForBackgroundWork());
+  auto l0_files = collector->GetFlushedFiles();
+  ASSERT_EQ(1, l0_files.size());
+
+  // We can run this test even without Snappy support because we expect the
+  // `CompactFiles()` to fail before actually invoking Snappy compression.
+  CompactionOptions compaction_opts;
+  compaction_opts.compression = CompressionType::kSnappyCompression;
+  ASSERT_TRUE(db->CompactFiles(compaction_opts, l0_files, 1 /* output_level */)
+                  .IsInvalidArgument());
+
+  compaction_opts.compression = CompressionType::kDisableCompressionOption;
+  ASSERT_OK(db->CompactFiles(compaction_opts, l0_files, 1 /* output_level */));
+}
+
 TEST_F(CompactFilesTest, GetCompactionJobInfo) {
   Options options;
   options.create_if_missing = true;
@@ -492,4 +538,3 @@ int main(int argc, char** argv) {
   ::testing::InitGoogleTest(&argc, argv);
   return RUN_ALL_TESTS();
 }
-

db/compaction/compaction.cc

@@ -160,7 +160,9 @@ std::vector<CompactionInputFiles> Compaction::PopulateWithAtomicBoundaries(
     AtomicCompactionUnitBoundary cur_boundary;
     size_t first_atomic_idx = 0;
     auto add_unit_boundary = [&](size_t to) {
-      if (first_atomic_idx == to) return;
+      if (first_atomic_idx == to) {
+        return;
+      }
       for (size_t k = first_atomic_idx; k < to; k++) {
         inputs[i].atomic_compaction_unit_boundaries.push_back(cur_boundary);
       }
@@ -753,7 +755,9 @@ int InputSummary(const std::vector<FileMetaData*>& files, char* output,
     AppendHumanBytes(files.at(i)->fd.GetFileSize(), sztxt, 16);
     ret = snprintf(output + write, sz, "%" PRIu64 "(%s) ",
                    files.at(i)->fd.GetNumber(), sztxt);
-    if (ret < 0 || ret >= sz) break;
+    if (ret < 0 || ret >= sz) {
+      break;
+    }
     write += ret;
   }
   // if files.size() is non-zero, overwrite the last space

db/compaction/compaction_iteration_stats.h

@@ -44,6 +44,13 @@ struct CompactionIterationStats {
   uint64_t total_blob_bytes_read = 0;
   uint64_t num_blobs_relocated = 0;
   uint64_t total_blob_bytes_relocated = 0;
+
+  // TimedPut diagnostics
+  // Total number of kTypeValuePreferredSeqno records encountered.
+  uint64_t num_input_timed_put_records = 0;
+  // Number of kTypeValuePreferredSeqno records we ended up swapping in
+  // preferred seqno.
+  uint64_t num_timed_put_swap_preferred_seqno = 0;
 };
 
 }  // namespace ROCKSDB_NAMESPACE

db/compaction/compaction_iterator.cc

@@ -491,6 +491,8 @@ void CompactionIterator::NextFromInput() {
     if (ikey_.type == kTypeDeletion || ikey_.type == kTypeSingleDeletion ||
         ikey_.type == kTypeDeletionWithTimestamp) {
       iter_stats_.num_input_deletion_records++;
+    } else if (ikey_.type == kTypeValuePreferredSeqno) {
+      iter_stats_.num_input_timed_put_records++;
     }
     iter_stats_.total_input_raw_key_bytes += key_.size();
     iter_stats_.total_input_raw_value_bytes += value_.size();
@@ -618,7 +620,8 @@ void CompactionIterator::NextFromInput() {
       // not compact out.  We will keep this Put, but can drop it's data.
       // (See Optimization 3, below.)
       if (ikey_.type != kTypeValue && ikey_.type != kTypeBlobIndex &&
-          ikey_.type != kTypeWideColumnEntity) {
+          ikey_.type != kTypeWideColumnEntity &&
+          ikey_.type != kTypeValuePreferredSeqno) {
         ROCKS_LOG_FATAL(info_log_, "Unexpected key %s for compaction output",
                         ikey_.DebugString(allow_data_in_errors_, true).c_str());
         assert(false);
@@ -632,7 +635,8 @@ void CompactionIterator::NextFromInput() {
         assert(false);
       }
 
-      if (ikey_.type == kTypeBlobIndex || ikey_.type == kTypeWideColumnEntity) {
+      if (ikey_.type == kTypeBlobIndex || ikey_.type == kTypeWideColumnEntity ||
+          ikey_.type == kTypeValuePreferredSeqno) {
         ikey_.type = kTypeValue;
         current_key_.UpdateInternalKey(ikey_.sequence, ikey_.type);
       }
@@ -798,7 +802,8 @@ void CompactionIterator::NextFromInput() {
             // happened
             if (next_ikey.type != kTypeValue &&
                 next_ikey.type != kTypeBlobIndex &&
-                next_ikey.type != kTypeWideColumnEntity) {
+                next_ikey.type != kTypeWideColumnEntity &&
+                next_ikey.type != kTypeValuePreferredSeqno) {
               ++iter_stats_.num_single_del_mismatch;
             }
 
@@ -968,6 +973,62 @@ void CompactionIterator::NextFromInput() {
         validity_info_.SetValid(ValidContext::kKeepDel);
         at_next_ = true;
       }
+    } else if (ikey_.type == kTypeValuePreferredSeqno &&
+               DefinitelyInSnapshot(ikey_.sequence, earliest_snapshot_) &&
+               (bottommost_level_ ||
+                (compaction_ != nullptr &&
+                 compaction_->KeyNotExistsBeyondOutputLevel(ikey_.user_key,
+                                                            &level_ptrs_)))) {
+      // This section that attempts to swap preferred sequence number will not
+      // be invoked if this is a CompactionIterator created for flush, since
+      // `compaction_` will be nullptr and it's not bottommost either.
+      //
+      // The entries with the same user key and smaller sequence numbers are
+      // all in this earliest snapshot range to be iterated. Since those entries
+      // will be hidden by this entry [rule A], it's safe to swap in the
+      // preferred seqno now.
+      //
+      // It's otherwise not safe to swap in the preferred seqno since it's
+      // possible for entries in earlier snapshots to have sequence number that
+      // is smaller than this entry's sequence number but bigger than this
+      // entry's preferred sequence number. Swapping in the preferred sequence
+      // number will break the internal key ordering invariant for this key.
+      //
+      // A special case involving range deletion is handled separately below.
+      auto [unpacked_value, preferred_seqno] =
+          ParsePackedValueWithSeqno(value_);
+      assert(preferred_seqno < ikey_.sequence || ikey_.sequence == 0);
+      if (range_del_agg_->ShouldDelete(
+              key_, RangeDelPositioningMode::kForwardTraversal)) {
+        ++iter_stats_.num_record_drop_hidden;
+        ++iter_stats_.num_record_drop_range_del;
+        AdvanceInputIter();
+      } else {
+        InternalKey ikey_after_swap(ikey_.user_key,
+                                    std::min(preferred_seqno, ikey_.sequence),
+                                    kTypeValue);
+        Slice ikey_after_swap_slice(*ikey_after_swap.rep());
+        if (range_del_agg_->ShouldDelete(
+                ikey_after_swap_slice,
+                RangeDelPositioningMode::kForwardTraversal)) {
+          // A range tombstone that doesn't cover this kTypeValuePreferredSeqno
+          // entry will end up covering the entry, so it's not safe to swap
+          // preferred sequence number. In this case, we output the entry as is.
+          validity_info_.SetValid(ValidContext::kNewUserKey);
+        } else {
+          if (ikey_.sequence != 0) {
+            iter_stats_.num_timed_put_swap_preferred_seqno++;
+            saved_seq_for_penul_check_ = ikey_.sequence;
+            ikey_.sequence = preferred_seqno;
+          }
+          ikey_.type = kTypeValue;
+          current_key_.UpdateInternalKey(ikey_.sequence, ikey_.type);
+          key_ = current_key_.GetInternalKey();
+          ikey_.user_key = current_key_.GetUserKey();
+          value_ = unpacked_value;
+          validity_info_.SetValid(ValidContext::kSwapPreferredSeqno);
+        }
+      }
     } else if (ikey_.type == kTypeMerge) {
       if (!merge_helper_->HasOperator()) {
         status_ = Status::InvalidArgument(
@@ -1219,6 +1280,21 @@ void CompactionIterator::DecideOutputLevel() {
   }
 #endif  // NDEBUG
 
+  // saved_seq_for_penul_check_ is populated in `NextFromInput` when the
+  // entry's sequence number is non zero and validity context for output this
+  // entry is kSwapPreferredSeqno for use in `DecideOutputLevel`. It should be
+  // cleared out here unconditionally. Otherwise, it may end up getting consumed
+  // incorrectly by a different entry.
+  SequenceNumber seq_for_range_check =
+      (saved_seq_for_penul_check_.has_value() &&
+       saved_seq_for_penul_check_.value() != kMaxSequenceNumber)
+          ? saved_seq_for_penul_check_.value()
+          : ikey_.sequence;
+  saved_seq_for_penul_check_ = std::nullopt;
+  ParsedInternalKey ikey_for_range_check = ikey_;
+  if (seq_for_range_check != ikey_.sequence) {
+    ikey_for_range_check.sequence = seq_for_range_check;
+  }
   if (output_to_penultimate_level_) {
     // If it's decided to output to the penultimate level, but unsafe to do so,
     // still output to the last level. For example, moving the data from a lower
@@ -1227,7 +1303,7 @@ void CompactionIterator::DecideOutputLevel() {
     // not from this compaction.
     // TODO: add statistic for declined output_to_penultimate_level
     bool safe_to_penultimate_level =
-        compaction_->WithinPenultimateLevelOutputRange(ikey_);
+        compaction_->WithinPenultimateLevelOutputRange(ikey_for_range_check);
     if (!safe_to_penultimate_level) {
       output_to_penultimate_level_ = false;
       // It could happen when disable/enable `last_level_temperature` while
@@ -1239,7 +1315,7 @@ void CompactionIterator::DecideOutputLevel() {
       // snapshot is released before enabling `last_level_temperature` feature
       // We will migrate the feature to `last_level_temperature` and maybe make
       // it not dynamically changeable.
-      if (ikey_.sequence > earliest_snapshot_) {
+      if (seq_for_range_check > earliest_snapshot_) {
         status_ = Status::Corruption(
             "Unsafe to store Seq later than snapshot in the last level if "
             "per_key_placement is enabled");

db/compaction/compaction_iterator.h

@@ -410,6 +410,7 @@ class CompactionIterator {
     kKeepDel = 9,
     kNewUserKey = 10,
     kRangeDeletion = 11,
+    kSwapPreferredSeqno = 12,
   };
 
   struct ValidityInfo {
@@ -436,6 +437,17 @@ class CompactionIterator {
   // iterator output (or current key in the underlying iterator during
   // NextFromInput()).
   ParsedInternalKey ikey_;
+
+  // When a kTypeValuePreferredSeqno entry's preferred seqno is safely swapped
+  // in in this compaction, this field saves its original sequence number for
+  // range checking whether it's safe to be placed on the penultimate level.
+  // This is to ensure when such an entry happens to be the right boundary of
+  // penultimate safe range, it won't get excluded because with the preferred
+  // seqno swapped in, it's now larger than the right boundary (itself before
+  // the swap). This is safe to do, because preferred seqno is swapped in only
+  // when no entries with the same user key exist on lower levels and this entry
+  // is already visible in the earliest snapshot.
+  std::optional<SequenceNumber> saved_seq_for_penul_check_ = kMaxSequenceNumber;
   // Stores whether ikey_.user_key is valid. If set to false, the user key is
   // not compared against the current key in the underlying iterator.
   bool has_current_user_key_ = false;

db/compaction/compaction_iterator_test.cc

@@ -17,6 +17,14 @@
 #include "utilities/merge_operators.h"
 
 namespace ROCKSDB_NAMESPACE {
+namespace {
+std::string ValueWithPreferredSeqno(std::string val,
+                                    SequenceNumber preferred_seqno = 0) {
+  std::string result = val;
+  PutFixed64(&result, preferred_seqno);
+  return result;
+}
+}  // namespace
 
 // Expects no merging attempts.
 class NoMergingMergeOp : public MergeOperator {
@@ -392,6 +400,17 @@ TEST_P(CompactionIteratorTest, CorruptionAfterSingleDeletion) {
   ASSERT_FALSE(c_iter_->Valid());
 }
 
+// Tests compatibility of TimedPut and SingleDelete. TimedPut should act as if
+// it's a Put.
+TEST_P(CompactionIteratorTest, TimedPutAndSingleDelete) {
+  InitIterators({test::KeyStr("a", 5, kTypeSingleDeletion),
+                 test::KeyStr("a", 3, kTypeValuePreferredSeqno)},
+                {"", "val"}, {}, {}, 5);
+  c_iter_->SeekToFirst();
+  ASSERT_OK(c_iter_->status());
+  ASSERT_FALSE(c_iter_->Valid());
+}
+
 TEST_P(CompactionIteratorTest, SimpleRangeDeletion) {
   InitIterators({test::KeyStr("morning", 5, kTypeValue),
                  test::KeyStr("morning", 2, kTypeValue),
@@ -431,6 +450,31 @@ TEST_P(CompactionIteratorTest, RangeDeletionWithSnapshots) {
   ASSERT_FALSE(c_iter_->Valid());
 }
 
+// Tests compatibility of TimedPut and Range delete. TimedPut should act as if
+// it's a Put.
+TEST_P(CompactionIteratorTest, TimedPutAndRangeDeletion) {
+  InitIterators(
+      {test::KeyStr("morning", 5, kTypeValuePreferredSeqno),
+       test::KeyStr("morning", 2, kTypeValuePreferredSeqno),
+       test::KeyStr("night", 3, kTypeValuePreferredSeqno)},
+      {ValueWithPreferredSeqno("zao5"), ValueWithPreferredSeqno("zao2"),
+       ValueWithPreferredSeqno("wan")},
+      {test::KeyStr("ma", 4, kTypeRangeDeletion)}, {"mz"}, 5);
+  c_iter_->SeekToFirst();
+  ASSERT_TRUE(c_iter_->Valid());
+  ASSERT_EQ(test::KeyStr("morning", 5, kTypeValuePreferredSeqno),
+            c_iter_->key().ToString());
+  ASSERT_EQ(ValueWithPreferredSeqno("zao5"), c_iter_->value().ToString());
+  c_iter_->Next();
+  ASSERT_TRUE(c_iter_->Valid());
+  ASSERT_EQ(test::KeyStr("night", 3, kTypeValuePreferredSeqno),
+            c_iter_->key().ToString());
+  ASSERT_EQ(ValueWithPreferredSeqno("wan"), c_iter_->value().ToString());
+  c_iter_->Next();
+  ASSERT_OK(c_iter_->status());
+  ASSERT_FALSE(c_iter_->Valid());
+}
+
 TEST_P(CompactionIteratorTest, CompactionFilterSkipUntil) {
   class Filter : public CompactionFilter {
     Decision FilterV2(int /*level*/, const Slice& key, ValueType t,
@@ -502,9 +546,11 @@ TEST_P(CompactionIteratorTest, CompactionFilterSkipUntil) {
        test::KeyStr("f", 25, kTypeValue), test::KeyStr("g", 90, kTypeValue),
        test::KeyStr("h", 91, kTypeValue),  // keep
        test::KeyStr("i", 95, kTypeMerge),  // skip to "z"
-       test::KeyStr("j", 99, kTypeValue)},
+       test::KeyStr("j", 99, kTypeValue),
+       test::KeyStr("k", 100, kTypeValuePreferredSeqno)},
       {"av50", "am45", "bv60", "bv40", "cv35", "dm70", "em71", "fm65", "fm30",
-       "fv25", "gv90", "hv91", "im95", "jv99"},
+       "fv25", "gv90", "hv91", "im95", "jv99",
+       ValueWithPreferredSeqno("kv100")},
       {}, {}, kMaxSequenceNumber, kMaxSequenceNumber, &merge_op, &filter);
 
   // Compaction should output just "a", "e" and "h" keys.
@@ -614,87 +660,87 @@ TEST_P(CompactionIteratorTest, ShuttingDownInMerge) {
   EXPECT_EQ(2, filter.last_seen.load());
 }
 
-TEST_P(CompactionIteratorTest, SingleMergeOperand) {
-  class Filter : public CompactionFilter {
-    Decision FilterV2(int /*level*/, const Slice& key, ValueType t,
-                      const Slice& existing_value, std::string* /*new_value*/,
-                      std::string* /*skip_until*/) const override {
-      std::string k = key.ToString();
-      std::string v = existing_value.ToString();
+class Filter : public CompactionFilter {
+  Decision FilterV2(int /*level*/, const Slice& key, ValueType t,
+                    const Slice& existing_value, std::string* /*new_value*/,
+                    std::string* /*skip_until*/) const override {
+    std::string k = key.ToString();
+    std::string v = existing_value.ToString();
 
-      // See InitIterators() call below for the sequence of keys and their
-      // filtering decisions. Here we closely assert that compaction filter is
-      // called with the expected keys and only them, and with the right values.
-      if (k == "a") {
-        EXPECT_EQ(ValueType::kMergeOperand, t);
-        EXPECT_EQ("av1", v);
-        return Decision::kKeep;
-      } else if (k == "b") {
-        EXPECT_EQ(ValueType::kMergeOperand, t);
-        return Decision::kKeep;
-      } else if (k == "c") {
-        return Decision::kKeep;
-      }
-
-      ADD_FAILURE();
+    // See InitIterators() call below for the sequence of keys and their
+    // filtering decisions. Here we closely assert that compaction filter is
+    // called with the expected keys and only them, and with the right values.
+    if (k == "a") {
+      EXPECT_EQ(ValueType::kMergeOperand, t);
+      EXPECT_EQ("av1", v);
+      return Decision::kKeep;
+    } else if (k == "b") {
+      EXPECT_EQ(ValueType::kMergeOperand, t);
+      return Decision::kKeep;
+    } else if (k == "c") {
       return Decision::kKeep;
     }
 
-    const char* Name() const override {
-      return "CompactionIteratorTest.SingleMergeOperand::Filter";
-    }
-  };
+    ADD_FAILURE();
+    return Decision::kKeep;
+  }
 
-  class SingleMergeOp : public MergeOperator {
-   public:
-    bool FullMergeV2(const MergeOperationInput& merge_in,
-                     MergeOperationOutput* merge_out) const override {
-      // See InitIterators() call below for why "c" is the only key for which
-      // FullMergeV2 should be called.
-      EXPECT_EQ("c", merge_in.key.ToString());
+  const char* Name() const override {
+    return "CompactionIteratorTest.SingleMergeOperand::Filter";
+  }
+};
 
-      std::string temp_value;
-      if (merge_in.existing_value != nullptr) {
-        temp_value = merge_in.existing_value->ToString();
-      }
+class SingleMergeOp : public MergeOperator {
+ public:
+  bool FullMergeV2(const MergeOperationInput& merge_in,
+                   MergeOperationOutput* merge_out) const override {
+    // See InitIterators() call below for why "c" is the only key for which
+    // FullMergeV2 should be called.
+    EXPECT_EQ("c", merge_in.key.ToString());
 
-      for (auto& operand : merge_in.operand_list) {
-        temp_value.append(operand.ToString());
-      }
-      merge_out->new_value = temp_value;
-
-      return true;
+    std::string temp_value;
+    if (merge_in.existing_value != nullptr) {
+      temp_value = merge_in.existing_value->ToString();
     }
 
-    bool PartialMergeMulti(const Slice& key,
-                           const std::deque<Slice>& operand_list,
-                           std::string* new_value,
-                           Logger* /*logger*/) const override {
-      std::string string_key = key.ToString();
-      EXPECT_TRUE(string_key == "a" || string_key == "b");
+    for (auto& operand : merge_in.operand_list) {
+      temp_value.append(operand.ToString());
+    }
+    merge_out->new_value = temp_value;
 
-      if (string_key == "a") {
-        EXPECT_EQ(1, operand_list.size());
-      } else if (string_key == "b") {
-        EXPECT_EQ(2, operand_list.size());
-      }
+    return true;
+  }
 
-      std::string temp_value;
-      for (auto& operand : operand_list) {
-        temp_value.append(operand.ToString());
-      }
-      swap(temp_value, *new_value);
+  bool PartialMergeMulti(const Slice& key,
+                         const std::deque<Slice>& operand_list,
+                         std::string* new_value,
+                         Logger* /*logger*/) const override {
+    std::string string_key = key.ToString();
+    EXPECT_TRUE(string_key == "a" || string_key == "b");
 
-      return true;
+    if (string_key == "a") {
+      EXPECT_EQ(1, operand_list.size());
+    } else if (string_key == "b") {
+      EXPECT_EQ(2, operand_list.size());
     }
 
-    const char* Name() const override {
-      return "CompactionIteratorTest SingleMergeOp";
+    std::string temp_value;
+    for (auto& operand : operand_list) {
+      temp_value.append(operand.ToString());
     }
+    swap(temp_value, *new_value);
 
-    bool AllowSingleOperand() const override { return true; }
-  };
+    return true;
+  }
 
+  const char* Name() const override {
+    return "CompactionIteratorTest SingleMergeOp";
+  }
+
+  bool AllowSingleOperand() const override { return true; }
+};
+
+TEST_P(CompactionIteratorTest, SingleMergeOperand) {
   SingleMergeOp merge_op;
   Filter filter;
   InitIterators(
@@ -719,6 +765,24 @@ TEST_P(CompactionIteratorTest, SingleMergeOperand) {
   ASSERT_EQ("cv1cv2", c_iter_->value().ToString());
 }
 
+// Tests compatibility of TimedPut and Merge operation. When a TimedPut is
+// merged with some merge operand in compaction, it will become a regular Put
+// and lose its preferred sequence number.
+TEST_P(CompactionIteratorTest, TimedPutAndMerge) {
+  SingleMergeOp merge_op;
+  Filter filter;
+  InitIterators({test::KeyStr("c", 90, kTypeMerge),
+                 test::KeyStr("c", 80, kTypeValuePreferredSeqno)},
+                {"cv2", ValueWithPreferredSeqno("cv1")}, {}, {},
+                kMaxSequenceNumber, kMaxSequenceNumber, &merge_op, &filter);
+
+  c_iter_->SeekToFirst();
+  ASSERT_TRUE(c_iter_->Valid());
+  ASSERT_EQ(test::KeyStr("c", 90, kTypeValue), c_iter_->key().ToString());
+  ASSERT_OK(c_iter_->status());
+  ASSERT_EQ("cv1cv2", c_iter_->value().ToString());
+}
+
 // In bottommost level, values earlier than earliest snapshot can be output
 // with sequence = 0.
 TEST_P(CompactionIteratorTest, ZeroOutSequenceAtBottomLevel) {
@@ -963,6 +1027,22 @@ TEST_F(CompactionIteratorWithSnapshotCheckerTest, DedupSameSnapshot_Value) {
       {"v4", "v3", "v1"}, 3 /*last_committed_seq*/);
 }
 
+TEST_F(CompactionIteratorWithSnapshotCheckerTest, DedupSameSnapshot_TimedPut) {
+  AddSnapshot(2, 1);
+  RunTest({test::KeyStr("foo", 4, kTypeValuePreferredSeqno),
+           test::KeyStr("foo", 3, kTypeValuePreferredSeqno),
+           test::KeyStr("foo", 2, kTypeValuePreferredSeqno),
+           test::KeyStr("foo", 1, kTypeValuePreferredSeqno)},
+          {ValueWithPreferredSeqno("v4"), ValueWithPreferredSeqno("v3"),
+           ValueWithPreferredSeqno("v2"), ValueWithPreferredSeqno("v1")},
+          {test::KeyStr("foo", 4, kTypeValuePreferredSeqno),
+           test::KeyStr("foo", 3, kTypeValuePreferredSeqno),
+           test::KeyStr("foo", 1, kTypeValuePreferredSeqno)},
+          {ValueWithPreferredSeqno("v4"), ValueWithPreferredSeqno("v3"),
+           ValueWithPreferredSeqno("v1")},
+          3 /*last_committed_seq*/);
+}
+
 TEST_F(CompactionIteratorWithSnapshotCheckerTest, DedupSameSnapshot_Deletion) {
   AddSnapshot(2, 1);
   RunTest(
@@ -1128,6 +1208,149 @@ TEST_F(CompactionIteratorWithSnapshotCheckerTest,
           2 /* earliest_write_conflict_snapshot */);
 }
 
+// Same as above but with a value with preferred seqno entry. In addition to the
+// value getting trimmed, the type of the KV is changed to kTypeValue.
+TEST_F(CompactionIteratorWithSnapshotCheckerTest,
+       KeepSingleDeletionForWriteConflictChecking_TimedPut) {
+  AddSnapshot(2, 0);
+  RunTest({test::KeyStr("a", 2, kTypeSingleDeletion),
+           test::KeyStr("a", 1, kTypeValuePreferredSeqno)},
+          {"", ValueWithPreferredSeqno("v1")},
+          {test::KeyStr("a", 2, kTypeSingleDeletion),
+           test::KeyStr("a", 1, kTypeValue)},
+          {"", ""}, 2 /* last_committed_seq */, nullptr /* merge_operator */,
+          nullptr /* compaction_filter */, false /* bottommost_level */,
+          2 /* earliest_write_conflict_snapshot */);
+}
+
+// Tests when a kTypeValuePreferredSeqno entry can have its preferred sequence
+// number swapped in. The required and sufficient conditions for an entry's
+// preferred sequence number to get swapped in are:
+// 1) The entry is visible to the earliest snapshot, AND
+// 2) No more entries with the same user key on lower levels, AND
+//    This is either because:
+//    2a) This is a compaction to the bottommost level, OR
+//    2b) Keys do not exist beyond output level
+// 3) The entry will not resurface a range deletion entry after swapping in the
+//    preferred sequence number.
+TEST_F(CompactionIteratorWithSnapshotCheckerTest,
+       TimedPut_NotVisibleToEarliestSnapshot_NoSwapPreferredSeqno) {
+  AddSnapshot(3);
+  RunTest({test::KeyStr("bar", 5, kTypeValuePreferredSeqno)},
+          {ValueWithPreferredSeqno("bv2", 2)},
+          {test::KeyStr("bar", 5, kTypeValuePreferredSeqno)},
+          {ValueWithPreferredSeqno("bv2", 2), "bv1"}, 5 /*last_committed_seq*/,
+          nullptr /*merge_operator*/, nullptr /*compaction_filter*/,
+          true /*bottommost_level*/,
+          kMaxSequenceNumber /*earliest_write_conflict_snapshot*/,
+          true /*key_not_exists_beyond_output_level*/);
+}
+
+TEST_F(CompactionIteratorWithSnapshotCheckerTest,
+       TimedPut_MoreEntriesInLowerLevels_NoSwapPreferredSeqno) {
+  // This tests mimics more entries in lower levels with `bottommost_level` and
+  // `key_not_exists_beyond_output_level` set to false.
+  RunTest({test::KeyStr("bar", 5, kTypeValuePreferredSeqno)},
+          {ValueWithPreferredSeqno("bv2", 2)},
+          {test::KeyStr("bar", 5, kTypeValuePreferredSeqno)},
+          {ValueWithPreferredSeqno("bv2", 2)}, 5 /*last_committed_seq*/,
+          nullptr /*merge_operator*/, nullptr /*compaction_filter*/,
+          false /*bottommost_level*/,
+          kMaxSequenceNumber /*earliest_write_conflict_snapshot*/,
+          false /*key_not_exists_beyond_output_level*/);
+}
+
+TEST_F(CompactionIteratorWithSnapshotCheckerTest,
+       TimedPut_ShouldBeCoverredByRangeDeletionBeforeSwap_NoOutput) {
+  InitIterators({test::KeyStr("morning", 5, kTypeValuePreferredSeqno),
+                 test::KeyStr("morning", 2, kTypeValuePreferredSeqno),
+                 test::KeyStr("night", 6, kTypeValue)},
+                {ValueWithPreferredSeqno("zao", 3),
+                 ValueWithPreferredSeqno("zao", 1), "wan"},
+                {test::KeyStr("ma", 6, kTypeRangeDeletion)}, {"mz"}, 6,
+                kMaxSequenceNumber /*last_committed_sequence*/,
+                nullptr /*merge_op*/, nullptr /*filter*/,
+                false /*bottommost_level*/,
+                kMaxSequenceNumber /*earliest_write_conflict_snapshot*/,
+                true /*key_not_exists_beyond_output_level*/);
+  c_iter_->SeekToFirst();
+  ASSERT_TRUE(c_iter_->Valid());
+  ASSERT_EQ(test::KeyStr("night", 6, kTypeValue), c_iter_->key().ToString());
+  ASSERT_EQ("wan", c_iter_->value().ToString());
+  c_iter_->Next();
+  ASSERT_FALSE(c_iter_->Valid());
+  ASSERT_OK(c_iter_->status());
+}
+
+TEST_F(CompactionIteratorWithSnapshotCheckerTest,
+       TimedPut_WillBeHiddenByRangeDeletionAfterSwap_NoSwap) {
+  InitIterators({test::KeyStr("morning", 5, kTypeValuePreferredSeqno),
+                 test::KeyStr("night", 6, kTypeValue)},
+                {ValueWithPreferredSeqno("zao", 3), "wan"},
+                {test::KeyStr("ma", 4, kTypeRangeDeletion)}, {"mz"}, 6,
+                kMaxSequenceNumber /*last_committed_sequence*/,
+                nullptr /*merge_op*/, nullptr /*filter*/,
+                false /*bottommost_level*/,
+                kMaxSequenceNumber /*earliest_write_conflict_snapshot*/,
+                true /*key_not_exists_beyond_output_level*/);
+  c_iter_->SeekToFirst();
+  ASSERT_TRUE(c_iter_->Valid());
+  ASSERT_EQ(test::KeyStr("morning", 5, kTypeValuePreferredSeqno),
+            c_iter_->key().ToString());
+  ASSERT_EQ(ValueWithPreferredSeqno("zao", 3), c_iter_->value().ToString());
+  c_iter_->Next();
+  ASSERT_TRUE(c_iter_->Valid());
+  ASSERT_EQ(test::KeyStr("night", 6, kTypeValue), c_iter_->key().ToString());
+  ASSERT_EQ("wan", c_iter_->value().ToString());
+  c_iter_->Next();
+  ASSERT_FALSE(c_iter_->Valid());
+  ASSERT_OK(c_iter_->status());
+}
+
+TEST_F(CompactionIteratorWithSnapshotCheckerTest,
+       TimedPut_BottomMostLevelVisibleToEarliestSnapshot_SwapPreferredSeqno) {
+  // Preferred seqno got swapped in and also zeroed out as a bottommost level
+  // optimization.
+  RunTest(
+      {test::KeyStr("bar", 5, kTypeValuePreferredSeqno),
+       test::KeyStr("bar", 4, kTypeValuePreferredSeqno),
+       test::KeyStr("foo", 6, kTypeValue)},
+      {ValueWithPreferredSeqno("bv2", 2), ValueWithPreferredSeqno("bv1", 1),
+       "fv1"},
+      {test::KeyStr("bar", 0, kTypeValue), test::KeyStr("foo", 0, kTypeValue)},
+      {"bv2", "fv1"}, 6 /*last_committed_seq*/, nullptr /*merge_operator*/,
+      nullptr /*compaction_filter*/, true /*bottommost_level*/);
+}
+
+TEST_F(
+    CompactionIteratorWithSnapshotCheckerTest,
+    TimedPut_NonBottomMostLevelVisibleToEarliestSnapshot_SwapPreferredSeqno) {
+  RunTest(
+      {test::KeyStr("bar", 5, kTypeValuePreferredSeqno),
+       test::KeyStr("bar", 4, kTypeValuePreferredSeqno),
+       test::KeyStr("foo", 6, kTypeValue)},
+      {ValueWithPreferredSeqno("bv2", 2), ValueWithPreferredSeqno("bv1", 1),
+       "fv1"},
+      {test::KeyStr("bar", 2, kTypeValue), test::KeyStr("foo", 6, kTypeValue)},
+      {"bv2", "fv1"}, 6 /*last_committed_seq*/, nullptr /*merge_operator*/,
+      nullptr /*compaction_filter*/, false /*bottommost_level*/,
+      kMaxSequenceNumber /*earliest_write_conflict_snapshot*/,
+      true /*key_not_exists_beyond_output_level*/);
+}
+
+TEST_F(CompactionIteratorWithSnapshotCheckerTest,
+       TimedPut_SequenceNumberAlreadyZeroedOut_ChangeType) {
+  RunTest(
+      {test::KeyStr("bar", 0, kTypeValuePreferredSeqno),
+       test::KeyStr("bar", 0, kTypeValuePreferredSeqno),
+       test::KeyStr("foo", 0, kTypeValue)},
+      {ValueWithPreferredSeqno("bv2", 2), ValueWithPreferredSeqno("bv1", 1),
+       "fv1"},
+      {test::KeyStr("bar", 0, kTypeValue), test::KeyStr("foo", 0, kTypeValue)},
+      {"bv2", "fv1"}, 6 /*last_committed_seq*/, nullptr /*merge_operator*/,
+      nullptr /*compaction_filter*/, true /*bottommost_level*/);
+}
+
 // Compaction filter should keep uncommitted key as-is, and
 //   * Convert the latest value to deletion, and/or
 //   * if latest value is a merge, apply filter to all subsequent merges.
@@ -1145,6 +1368,22 @@ TEST_F(CompactionIteratorWithSnapshotCheckerTest, CompactionFilter_Value) {
       nullptr /*merge_operator*/, compaction_filter.get());
 }
 
+TEST_F(CompactionIteratorWithSnapshotCheckerTest, CompactionFilter_TimedPut) {
+  // TODO(yuzhangyu): Add support for this type in compaction filter.
+  // Type kTypeValuePreferredSeqno is not explicitly exposed in the compaction
+  // filter API, so users can not operate on it through compaction filter API
+  // to remove/purge/change value etc. But this type of entry can be impacted by
+  // other entries' filter result, currently only kRemoveAndSkip type of result
+  // can affect it.
+  std::unique_ptr<CompactionFilter> compaction_filter(
+      new FilterAllKeysCompactionFilter());
+  RunTest({test::KeyStr("a", 2, kTypeValuePreferredSeqno)},
+          {ValueWithPreferredSeqno("v1")},
+          {test::KeyStr("a", 2, kTypeValuePreferredSeqno)},
+          {ValueWithPreferredSeqno("v1")}, 2 /*last_committed_seq*/,
+          nullptr /*merge_operator*/, compaction_filter.get());
+}
+
 TEST_F(CompactionIteratorWithSnapshotCheckerTest, CompactionFilter_Deletion) {
   std::unique_ptr<CompactionFilter> compaction_filter(
       new FilterAllKeysCompactionFilter());

db/compaction/compaction_job.cc

@@ -404,7 +404,9 @@ void CompactionJob::AcquireSubcompactionResources(
 
 void CompactionJob::ShrinkSubcompactionResources(uint64_t num_extra_resources) {
   // Do nothing when we have zero resources to shrink
-  if (num_extra_resources == 0) return;
+  if (num_extra_resources == 0) {
+    return;
+  }
   db_mutex_->Lock();
   // We cannot release threads more than what we reserved before
   int extra_num_subcompaction_threads_released = env_->ReleaseThreads(
@@ -584,7 +586,9 @@ void CompactionJob::GenSubcompactionBoundaries() {
 
   TEST_SYNC_POINT_CALLBACK("CompactionJob::GenSubcompactionBoundaries:0",
                            &num_planned_subcompactions);
-  if (num_planned_subcompactions == 1) return;
+  if (num_planned_subcompactions == 1) {
+    return;
+  }
 
   // Group the ranges into subcompactions
   uint64_t target_range_size = std::max(
@@ -641,7 +645,7 @@ Status CompactionJob::Run() {
 
   // Always schedule the first subcompaction (whether or not there are also
   // others) in the current thread to be efficient with resources
-  ProcessKeyValueCompaction(&compact_->sub_compact_states[0]);
+  ProcessKeyValueCompaction(compact_->sub_compact_states.data());
 
   // Wait for all other threads (if there are any) to finish execution
   for (auto& thread : thread_pool) {
@@ -1850,13 +1854,14 @@ Status CompactionJob::OpenCompactionOutputFile(SubcompactionState* sub_compact,
   // Pass temperature of the last level files to FileSystem.
   FileOptions fo_copy = file_options_;
   Temperature temperature = sub_compact->compaction->output_temperature();
-  // only set for the last level compaction and also it's not output to
-  // penultimate level (when preclude_last_level feature is enabled)
-  if (temperature == Temperature::kUnknown &&
+  Temperature last_level_temp =
+      sub_compact->compaction->mutable_cf_options()->last_level_temperature;
+  // Here last_level_temperature supersedes default_write_temperature, when
+  // enabled and applicable
+  if (last_level_temp != Temperature::kUnknown &&
       sub_compact->compaction->is_last_level() &&
       !sub_compact->IsCurrentPenultimateLevel()) {
-    temperature =
-        sub_compact->compaction->mutable_cf_options()->last_level_temperature;
+    temperature = last_level_temp;
   }
   fo_copy.temperature = temperature;
 

db/compaction/compaction_job.h

@@ -41,7 +41,6 @@
 #include "rocksdb/env.h"
 #include "rocksdb/memtablerep.h"
 #include "rocksdb/transaction_log.h"
-#include "table/scoped_arena_iterator.h"
 #include "util/autovector.h"
 #include "util/stop_watch.h"
 #include "util/thread_local.h"

db/compaction/compaction_job_stats_test.cc

@@ -46,7 +46,6 @@
 #include "table/block_based/block_based_table_factory.h"
 #include "table/mock_table.h"
 #include "table/plain/plain_table_factory.h"
-#include "table/scoped_arena_iterator.h"
 #include "test_util/sync_point.h"
 #include "test_util/testharness.h"
 #include "test_util/testutil.h"
@@ -131,7 +130,7 @@ class CompactionJobStatsTest : public testing::Test,
     ColumnFamilyOptions cf_opts(options);
     size_t cfi = handles_.size();
     handles_.resize(cfi + cfs.size());
-    for (auto cf : cfs) {
+    for (const auto& cf : cfs) {
       ASSERT_OK(db_->CreateColumnFamily(cf_opts, cf, &handles_[cfi++]));
     }
   }
@@ -160,7 +159,7 @@ class CompactionJobStatsTest : public testing::Test,
     EXPECT_EQ(cfs.size(), options.size());
     std::vector<ColumnFamilyDescriptor> column_families;
     for (size_t i = 0; i < cfs.size(); ++i) {
-      column_families.push_back(ColumnFamilyDescriptor(cfs[i], options[i]));
+      column_families.emplace_back(cfs[i], options[i]);
     }
     DBOptions db_opts = DBOptions(options[0]);
     return DB::Open(db_opts, dbname_, column_families, &handles_, &db_);

db/compaction/compaction_job_test.cc

@@ -308,7 +308,7 @@ class CompactionJobTestBase : public testing::Test {
                                 kDefaultColumnFamilyName, -1 /* level */),
             file_writer.get()));
     // Build table.
-    for (auto kv : contents) {
+    for (const auto& kv : contents) {
       std::string key;
       std::string value;
       std::tie(key, value) = kv;
@@ -327,7 +327,7 @@ class CompactionJobTestBase : public testing::Test {
     SequenceNumber smallest_seqno = kMaxSequenceNumber;
     SequenceNumber largest_seqno = 0;
     uint64_t oldest_blob_file_number = kInvalidBlobFileNumber;
-    for (auto kv : contents) {
+    for (const auto& kv : contents) {
       ParsedInternalKey key;
       std::string skey;
       std::string value;
@@ -1473,7 +1473,7 @@ TEST_F(CompactionJobTest, OldestBlobFileNumber) {
 }
 
 TEST_F(CompactionJobTest, VerifyPenultimateLevelOutput) {
-  cf_options_.bottommost_temperature = Temperature::kCold;
+  cf_options_.last_level_temperature = Temperature::kCold;
   SyncPoint::GetInstance()->SetCallBack(
       "Compaction::SupportsPerKeyPlacement:Enabled", [&](void* arg) {
         auto supports_per_key_placement = static_cast<bool*>(arg);

db/compaction/compaction_outputs.cc

@@ -27,7 +27,9 @@ Status CompactionOutputs::Finish(
   if (s.ok()) {
     SeqnoToTimeMapping relevant_mapping;
     relevant_mapping.CopyFromSeqnoRange(
-        seqno_to_time_mapping, meta->fd.smallest_seqno, meta->fd.largest_seqno);
+        seqno_to_time_mapping,
+        std::min(smallest_preferred_seqno_, meta->fd.smallest_seqno),
+        meta->fd.largest_seqno);
     relevant_mapping.SetCapacity(kMaxSeqnoTimePairsPerSST);
     builder_->SetSeqnoTimeTableProperties(relevant_mapping,
                                           meta->oldest_ancester_time);
@@ -422,6 +424,11 @@ Status CompactionOutputs::AddToOutput(
   }
 
   const ParsedInternalKey& ikey = c_iter.ikey();
+  if (ikey.type == kTypeValuePreferredSeqno) {
+    SequenceNumber preferred_seqno = ParsePackedValueForSeqno(value);
+    smallest_preferred_seqno_ =
+        std::min(smallest_preferred_seqno_, preferred_seqno);
+  }
   s = current_output().meta.UpdateBoundaries(key, value, ikey.sequence,
                                              ikey.type);
 

db/compaction/compaction_outputs.h

@@ -297,6 +297,7 @@ class CompactionOutputs {
   std::unique_ptr<TableBuilder> builder_;
   std::unique_ptr<WritableFileWriter> file_writer_;
   uint64_t current_output_file_size_ = 0;
+  SequenceNumber smallest_preferred_seqno_ = kMaxSequenceNumber;
 
   // all the compaction outputs so far
   std::vector<Output> outputs_;

db/compaction/compaction_picker.cc

@@ -130,7 +130,7 @@ CompactionPicker::CompactionPicker(const ImmutableOptions& ioptions,
                                    const InternalKeyComparator* icmp)
     : ioptions_(ioptions), icmp_(icmp) {}
 
-CompactionPicker::~CompactionPicker() {}
+CompactionPicker::~CompactionPicker() = default;
 
 // Delete this compaction from the list of running compactions.
 void CompactionPicker::ReleaseCompactionFiles(Compaction* c, Status status) {
@@ -340,8 +340,8 @@ Compaction* CompactionPicker::CompactFiles(
 #ifndef NDEBUG
   assert(input_files.size());
   // This compaction output should not overlap with a running compaction as
-  // `SanitizeCompactionInputFiles` should've checked earlier and db mutex
-  // shouldn't have been released since.
+  // `SanitizeAndConvertCompactionInputFiles` should've checked earlier and db
+  // mutex shouldn't have been released since.
   int start_level = Compaction::kInvalidLevel;
   for (const auto& in : input_files) {
     // input_files should already be sorted by level
@@ -377,7 +377,8 @@ Compaction* CompactionPicker::CompactFiles(
       output_level, compact_options.output_file_size_limit,
       mutable_cf_options.max_compaction_bytes, output_path_id, compression_type,
       GetCompressionOptions(mutable_cf_options, vstorage, output_level),
-      Temperature::kUnknown, compact_options.max_subcompactions,
+      mutable_cf_options.default_write_temperature,
+      compact_options.max_subcompactions,
       /* grandparents */ {}, true);
   RegisterCompaction(c);
   return c;
@@ -456,9 +457,10 @@ bool CompactionPicker::IsRangeInCompaction(VersionStorageInfo* vstorage,
 // Returns false if files on parent level are currently in compaction, which
 // means that we can't compact them
 bool CompactionPicker::SetupOtherInputs(
-    const std::string& cf_name, VersionStorageInfo* vstorage,
-    CompactionInputFiles* inputs, CompactionInputFiles* output_level_inputs,
-    int* parent_index, int base_index, bool only_expand_towards_right) {
+    const std::string& cf_name, const MutableCFOptions& mutable_cf_options,
+    VersionStorageInfo* vstorage, CompactionInputFiles* inputs,
+    CompactionInputFiles* output_level_inputs, int* parent_index,
+    int base_index, bool only_expand_towards_right) {
   assert(!inputs->empty());
   assert(output_level_inputs->empty());
   const int input_level = inputs->level;
@@ -524,8 +526,15 @@ bool CompactionPicker::SetupOtherInputs(
     if (!ExpandInputsToCleanCut(cf_name, vstorage, &expanded_inputs)) {
       try_overlapping_inputs = false;
     }
+    // It helps to reduce write amp and avoid a further separate compaction
+    // to include more input level files without expanding output level files.
+    // So we apply a softer limit. We still need a limit to avoid overly large
+    // compactions and potential high space amp spikes.
+    const uint64_t limit =
+        MultiplyCheckOverflow(mutable_cf_options.max_compaction_bytes, 2.0);
     if (try_overlapping_inputs && expanded_inputs.size() > inputs->size() &&
-        !AreFilesInCompaction(expanded_inputs.files)) {
+        !AreFilesInCompaction(expanded_inputs.files) &&
+        output_level_inputs_size + expanded_inputs_size < limit) {
       InternalKey new_start, new_limit;
       GetRange(expanded_inputs, &new_start, &new_limit);
       CompactionInputFiles expanded_output_level_inputs;
@@ -547,7 +556,8 @@ bool CompactionPicker::SetupOtherInputs(
                                              base_index, nullptr);
       expanded_inputs_size = TotalFileSize(expanded_inputs.files);
       if (expanded_inputs.size() > inputs->size() &&
-          !AreFilesInCompaction(expanded_inputs.files)) {
+          !AreFilesInCompaction(expanded_inputs.files) &&
+          (output_level_inputs_size + expanded_inputs_size) < limit) {
         expand_inputs = true;
       }
     }
@@ -664,7 +674,8 @@ Compaction* CompactionPicker::CompactRange(
         compact_range_options.target_path_id,
         GetCompressionType(vstorage, mutable_cf_options, output_level, 1),
         GetCompressionOptions(mutable_cf_options, vstorage, output_level),
-        Temperature::kUnknown, compact_range_options.max_subcompactions,
+        mutable_cf_options.default_write_temperature,
+        compact_range_options.max_subcompactions,
         /* grandparents */ {}, /* is manual */ true, trim_ts, /* score */ -1,
         /* deletion_compaction */ false, /* l0_files_might_overlap */ true,
         CompactionReason::kUnknown,
@@ -806,8 +817,8 @@ Compaction* CompactionPicker::CompactRange(
   output_level_inputs.level = output_level;
   if (input_level != output_level) {
     int parent_index = -1;
-    if (!SetupOtherInputs(cf_name, vstorage, &inputs, &output_level_inputs,
-                          &parent_index, -1)) {
+    if (!SetupOtherInputs(cf_name, mutable_cf_options, vstorage, &inputs,
+                          &output_level_inputs, &parent_index, -1)) {
       // manual compaction is now multi-threaded, so it can
       // happen that SetupOtherInputs fails
       // we handle it higher in RunManualCompaction
@@ -852,8 +863,9 @@ Compaction* CompactionPicker::CompactRange(
       GetCompressionType(vstorage, mutable_cf_options, output_level,
                          vstorage->base_level()),
       GetCompressionOptions(mutable_cf_options, vstorage, output_level),
-      Temperature::kUnknown, compact_range_options.max_subcompactions,
-      std::move(grandparents), /* is manual */ true, trim_ts, /* score */ -1,
+      mutable_cf_options.default_write_temperature,
+      compact_range_options.max_subcompactions, std::move(grandparents),
+      /* is manual */ true, trim_ts, /* score */ -1,
       /* deletion_compaction */ false, /* l0_files_might_overlap */ true,
       CompactionReason::kUnknown,
       compact_range_options.blob_garbage_collection_policy,
@@ -1028,19 +1040,18 @@ Status CompactionPicker::SanitizeCompactionInputFilesForAllLevels(
       }
     }
   }
-  if (RangeOverlapWithCompaction(smallestkey, largestkey, output_level)) {
-    return Status::Aborted(
-        "A running compaction is writing to the same output level in an "
-        "overlapping key range");
-  }
   return Status::OK();
 }
 
-Status CompactionPicker::SanitizeCompactionInputFiles(
+Status CompactionPicker::SanitizeAndConvertCompactionInputFiles(
     std::unordered_set<uint64_t>* input_files,
-    const ColumnFamilyMetaData& cf_meta, const int output_level) const {
+    const ColumnFamilyMetaData& cf_meta, const int output_level,
+    const VersionStorageInfo* vstorage,
+    std::vector<CompactionInputFiles>* converted_input_files) const {
   assert(static_cast<int>(cf_meta.levels.size()) - 1 ==
          cf_meta.levels[cf_meta.levels.size() - 1].level);
+  assert(converted_input_files);
+
   if (output_level >= static_cast<int>(cf_meta.levels.size())) {
     return Status::InvalidArgument(
         "Output level for column family " + cf_meta.name +
@@ -1066,7 +1077,6 @@ Status CompactionPicker::SanitizeCompactionInputFiles(
 
   Status s = SanitizeCompactionInputFilesForAllLevels(input_files, cf_meta,
                                                       output_level);
-
   if (!s.ok()) {
     return s;
   }
@@ -1107,6 +1117,22 @@ Status CompactionPicker::SanitizeCompactionInputFiles(
     }
   }
 
+  s = GetCompactionInputsFromFileNumbers(converted_input_files, input_files,
+                                         vstorage, CompactionOptions());
+  if (!s.ok()) {
+    return s;
+  }
+  assert(converted_input_files->size() > 0);
+  if (output_level != 0 &&
+      FilesRangeOverlapWithCompaction(
+          *converted_input_files, output_level,
+          Compaction::EvaluatePenultimateLevel(
+              vstorage, ioptions_, (*converted_input_files)[0].level,
+              output_level))) {
+    return Status::Aborted(
+        "A running compaction is writing to the same output level(s) in an "
+        "overlapping key range");
+  }
   return Status::OK();
 }
 

db/compaction/compaction_picker.h

@@ -86,16 +86,20 @@ class CompactionPicker {
 
   virtual bool NeedsCompaction(const VersionStorageInfo* vstorage) const = 0;
 
-// Sanitize the input set of compaction input files.
-// When the input parameters do not describe a valid compaction, the
-// function will try to fix the input_files by adding necessary
-// files.  If it's not possible to conver an invalid input_files
-// into a valid one by adding more files, the function will return a
-// non-ok status with specific reason.
-//
-  Status SanitizeCompactionInputFiles(std::unordered_set<uint64_t>* input_files,
-                                      const ColumnFamilyMetaData& cf_meta,
-                                      const int output_level) const;
+  // Sanitize the input set of compaction input files and convert it to
+  // `std::vector<CompactionInputFiles>` in the output parameter
+  // `converted_input_files`.
+  // When the input parameters do not describe a valid
+  // compaction, the function will try to fix the input_files by adding
+  // necessary files.  If it's not possible to convert an invalid input_files
+  // into a valid one by adding more files, the function will return a
+  // non-ok status with specific reason.
+  //
+  Status SanitizeAndConvertCompactionInputFiles(
+      std::unordered_set<uint64_t>* input_files,
+      const ColumnFamilyMetaData& cf_meta, const int output_level,
+      const VersionStorageInfo* vstorage,
+      std::vector<CompactionInputFiles>* converted_input_files) const;
 
   // Free up the files that participated in a compaction
   //
@@ -109,8 +113,8 @@ class CompactionPicker {
   // object.
   //
   // Caller must provide a set of input files that has been passed through
-  // `SanitizeCompactionInputFiles` earlier. The lock should not be released
-  // between that call and this one.
+  // `SanitizeAndConvertCompactionInputFiles` earlier. The lock should not be
+  // released between that call and this one.
   Compaction* CompactFiles(const CompactionOptions& compact_options,
                            const std::vector<CompactionInputFiles>& input_files,
                            int output_level, VersionStorageInfo* vstorage,
@@ -120,6 +124,7 @@ class CompactionPicker {
 
   // Converts a set of compaction input file numbers into
   // a list of CompactionInputFiles.
+  // TODO(hx235): remove the unused paramter `compact_options`
   Status GetCompactionInputsFromFileNumbers(
       std::vector<CompactionInputFiles>* input_files,
       std::unordered_set<uint64_t>* input_set,
@@ -186,6 +191,7 @@ class CompactionPicker {
       int penultimate_level) const;
 
   bool SetupOtherInputs(const std::string& cf_name,
+                        const MutableCFOptions& mutable_cf_options,
                         VersionStorageInfo* vstorage,
                         CompactionInputFiles* inputs,
                         CompactionInputFiles* output_level_inputs,
@@ -224,8 +230,8 @@ class CompactionPicker {
  protected:
   const ImmutableOptions& ioptions_;
 
-// A helper function to SanitizeCompactionInputFiles() that
-// sanitizes "input_files" by adding necessary files.
+  // A helper function to SanitizeAndConvertCompactionInputFiles() that
+  // sanitizes "input_files" by adding necessary files.
   virtual Status SanitizeCompactionInputFilesForAllLevels(
       std::unordered_set<uint64_t>* input_files,
       const ColumnFamilyMetaData& cf_meta, const int output_level) const;

db/compaction/compaction_picker_fifo.cc

@@ -116,7 +116,8 @@ Compaction* FIFOCompactionPicker::PickTTLCompaction(
   Compaction* c = new Compaction(
       vstorage, ioptions_, mutable_cf_options, mutable_db_options,
       std::move(inputs), 0, 0, 0, 0, kNoCompression,
-      mutable_cf_options.compression_opts, Temperature::kUnknown,
+      mutable_cf_options.compression_opts,
+      mutable_cf_options.default_write_temperature,
       /* max_subcompactions */ 0, {}, /* is manual */ false,
       /* trim_ts */ "", vstorage->CompactionScore(0),
       /* is deletion compaction */ true, /* l0_files_might_overlap */ true,
@@ -185,7 +186,8 @@ Compaction* FIFOCompactionPicker::PickSizeCompaction(
             {comp_inputs}, 0, 16 * 1024 * 1024 /* output file size limit */,
             0 /* max compaction bytes, not applicable */,
             0 /* output path ID */, mutable_cf_options.compression,
-            mutable_cf_options.compression_opts, Temperature::kUnknown,
+            mutable_cf_options.compression_opts,
+            mutable_cf_options.default_write_temperature,
             0 /* max_subcompactions */, {}, /* is manual */ false,
             /* trim_ts */ "", vstorage->CompactionScore(0),
             /* is deletion compaction */ false,
@@ -280,7 +282,8 @@ Compaction* FIFOCompactionPicker::PickSizeCompaction(
       /* target_file_size */ 0,
       /* max_compaction_bytes */ 0,
       /* output_path_id */ 0, kNoCompression,
-      mutable_cf_options.compression_opts, Temperature::kUnknown,
+      mutable_cf_options.compression_opts,
+      mutable_cf_options.default_write_temperature,
       /* max_subcompactions */ 0, {}, /* is manual */ false,
       /* trim_ts */ "", vstorage->CompactionScore(0),
       /* is deletion compaction */ true,
@@ -414,6 +417,7 @@ Compaction* FIFOCompactionPicker::PickTemperatureChangeCompaction(
   if (inputs[0].files.empty()) {
     return nullptr;
   }
+  assert(compaction_target_temp != Temperature::kLastTemperature);
 
   Compaction* c = new Compaction(
       vstorage, ioptions_, mutable_cf_options, mutable_db_options,

db/compaction/compaction_picker_level.cc

@@ -355,7 +355,9 @@ void LevelCompactionBuilder::SetupOtherFilesWithRoundRobinExpansion() {
   TEST_SYNC_POINT("LevelCompactionPicker::RoundRobin");
 
   // Only expand the inputs when we have selected a file in start_level_inputs_
-  if (start_level_inputs_.size() == 0) return;
+  if (start_level_inputs_.size() == 0) {
+    return;
+  }
 
   uint64_t start_lvl_bytes_no_compacting = 0;
   uint64_t curr_bytes_to_compact = 0;
@@ -465,8 +467,9 @@ bool LevelCompactionBuilder::SetupOtherInputsIfNeeded() {
     }
     if (!is_l0_trivial_move_ &&
         !compaction_picker_->SetupOtherInputs(
-            cf_name_, vstorage_, &start_level_inputs_, &output_level_inputs_,
-            &parent_index_, base_index_, round_robin_expanding)) {
+            cf_name_, mutable_cf_options_, vstorage_, &start_level_inputs_,
+            &output_level_inputs_, &parent_index_, base_index_,
+            round_robin_expanding)) {
       return false;
     }
 
@@ -550,7 +553,7 @@ Compaction* LevelCompactionBuilder::GetCompaction() {
       GetCompressionType(vstorage_, mutable_cf_options_, output_level_,
                          vstorage_->base_level()),
       GetCompressionOptions(mutable_cf_options_, vstorage_, output_level_),
-      Temperature::kUnknown,
+      mutable_cf_options_.default_write_temperature,
       /* max_subcompactions */ 0, std::move(grandparents_), is_manual_,
       /* trim_ts */ "", start_level_score_, false /* deletion_compaction */,
       l0_files_might_overlap, compaction_reason_);

db/compaction/compaction_picker_test.cc

@@ -77,7 +77,7 @@ class CompactionPickerTestBase : public testing::Test {
     ioptions_.level_compaction_dynamic_level_bytes = false;
   }
 
-  ~CompactionPickerTestBase() override {}
+  ~CompactionPickerTestBase() override = default;
 
   void NewVersionStorage(int num_levels, CompactionStyle style) {
     DeleteVersionStorage();
@@ -214,7 +214,7 @@ class CompactionPickerTest : public CompactionPickerTestBase {
   explicit CompactionPickerTest()
       : CompactionPickerTestBase(BytewiseComparator()) {}
 
-  ~CompactionPickerTest() override {}
+  ~CompactionPickerTest() override = default;
 };
 
 class CompactionPickerU64TsTest : public CompactionPickerTestBase {
@@ -222,7 +222,7 @@ class CompactionPickerU64TsTest : public CompactionPickerTestBase {
   explicit CompactionPickerU64TsTest()
       : CompactionPickerTestBase(test::BytewiseComparatorWithU64TsWrapper()) {}
 
-  ~CompactionPickerU64TsTest() override {}
+  ~CompactionPickerU64TsTest() override = default;
 };
 
 TEST_F(CompactionPickerTest, Empty) {
@@ -2465,7 +2465,7 @@ TEST_F(CompactionPickerTest, IsBottommostLevelTest) {
   DeleteVersionStorage();
 }
 
-TEST_F(CompactionPickerTest, IgnoreCompactionLimitWhenAddFileFromInputLevel) {
+TEST_F(CompactionPickerTest, CompactionLimitWhenAddFileFromInputLevel) {
   mutable_cf_options_.max_bytes_for_level_base = 1000000u;
   mutable_cf_options_.max_compaction_bytes = 800000u;
   ioptions_.level_compaction_dynamic_level_bytes = false;
@@ -2473,8 +2473,10 @@ TEST_F(CompactionPickerTest, IgnoreCompactionLimitWhenAddFileFromInputLevel) {
   // A compaction should be triggered and pick file 2 and 5.
   // It pulls in other compaction input file from the input level L1
   // without pulling in more output level files.
-  // Files 1, 3, 4 will be included in the compaction.
+  // Files 1, 3, 4 are eligible.
   // File 6 is excluded since it overlaps with file 7.
+  // It can expand input level since in this case, the limit on compaction size
+  // is 2 * max_compaction_bytes.
   Add(1, 1U, "100", "150", 300000U);
   Add(1, 2U, "151", "200", 300001U, 0, 0);
   Add(1, 3U, "201", "250", 300000U, 0, 0);
@@ -2498,6 +2500,38 @@ TEST_F(CompactionPickerTest, IgnoreCompactionLimitWhenAddFileFromInputLevel) {
   ASSERT_EQ(5U, compaction->input(1, 0)->fd.GetNumber());
 }
 
+TEST_F(CompactionPickerTest, HitCompactionLimitWhenAddFileFromInputLevel) {
+  mutable_cf_options_.max_bytes_for_level_base = 1000000u;
+  mutable_cf_options_.max_compaction_bytes = 800000u;
+  ioptions_.level_compaction_dynamic_level_bytes = false;
+  NewVersionStorage(6, kCompactionStyleLevel);
+  // A compaction should be triggered and pick file 2 and 5.
+  // It pulls in other compaction input file from the input level L1
+  // without pulling in more output level files.
+  // Files 1, 3, 4 are eligible.
+  // File 6 is excluded since it overlaps with file 7.
+  // It can not expand input level since total compaction size hit the limit
+  // 2 * max_compaction_bytes.
+  Add(1, 1U, "100", "150", 400000U);
+  Add(1, 2U, "151", "200", 400001U, 0, 0);
+  Add(1, 3U, "201", "250", 400000U, 0, 0);
+  Add(1, 4U, "251", "300", 400000U, 0, 0);
+  Add(1, 6U, "325", "400", 400000U, 0, 0);
+  Add(2, 5U, "100", "350", 1U);
+  Add(2, 7U, "375", "425", 1U);
+  UpdateVersionStorageInfo();
+
+  std::unique_ptr<Compaction> compaction(level_compaction_picker.PickCompaction(
+      cf_name_, mutable_cf_options_, mutable_db_options_, vstorage_.get(),
+      &log_buffer_));
+  ASSERT_TRUE(compaction.get() != nullptr);
+  ASSERT_EQ(2U, compaction->num_input_levels());
+  ASSERT_EQ(1U, compaction->num_input_files(0));
+  ASSERT_EQ(1U, compaction->num_input_files(1));
+  ASSERT_EQ(2U, compaction->input(0, 0)->fd.GetNumber());
+  ASSERT_EQ(5U, compaction->input(1, 0)->fd.GetNumber());
+}
+
 TEST_F(CompactionPickerTest, IsTrivialMoveOn) {
   mutable_cf_options_.max_bytes_for_level_base = 10000u;
   mutable_cf_options_.max_compaction_bytes = 10001u;

db/compaction/compaction_picker_universal.cc

@@ -870,7 +870,7 @@ Compaction* UniversalCompactionBuilder::PickCompactionToReduceSortedRuns(
                                            output_level, 1, enable_compression),
                         GetCompressionOptions(mutable_cf_options_, vstorage_,
                                               output_level, enable_compression),
-                        Temperature::kUnknown,
+                        mutable_cf_options_.default_write_temperature,
                         /* max_subcompactions */ 0, grandparents,
                         /* is manual */ false, /* trim_ts */ "", score_,
                         false /* deletion_compaction */,
@@ -1139,7 +1139,8 @@ Compaction* UniversalCompactionBuilder::PickIncrementalForReduceSizeAmp(
   // from bottom_start_idx and bottom_end_idx, but for now, we use
   // SetupOtherInputs() for simplicity.
   int parent_index = -1;  // Create and use bottom_start_idx?
-  if (!picker_->SetupOtherInputs(cf_name_, vstorage_, &second_last_level_inputs,
+  if (!picker_->SetupOtherInputs(cf_name_, mutable_cf_options_, vstorage_,
+                                 &second_last_level_inputs,
                                  &bottom_level_inputs, &parent_index,
                                  /*base_index=*/-1)) {
     return nullptr;
@@ -1204,7 +1205,7 @@ Compaction* UniversalCompactionBuilder::PickIncrementalForReduceSizeAmp(
                          true /* enable_compression */),
       GetCompressionOptions(mutable_cf_options_, vstorage_, output_level,
                             true /* enable_compression */),
-      Temperature::kUnknown,
+      mutable_cf_options_.default_write_temperature,
       /* max_subcompactions */ 0, /* grandparents */ {}, /* is manual */ false,
       /* trim_ts */ "", score_, false /* deletion_compaction */,
       /* l0_files_might_overlap */ true,
@@ -1310,8 +1311,9 @@ Compaction* UniversalCompactionBuilder::PickDeleteTriggeredCompaction() {
       int parent_index = -1;
 
       output_level_inputs.level = output_level;
-      if (!picker_->SetupOtherInputs(cf_name_, vstorage_, &start_level_inputs,
-                                     &output_level_inputs, &parent_index, -1)) {
+      if (!picker_->SetupOtherInputs(cf_name_, mutable_cf_options_, vstorage_,
+                                     &start_level_inputs, &output_level_inputs,
+                                     &parent_index, -1)) {
         return nullptr;
       }
       inputs.push_back(start_level_inputs);
@@ -1347,7 +1349,7 @@ Compaction* UniversalCompactionBuilder::PickDeleteTriggeredCompaction() {
       /* max_grandparent_overlap_bytes */ GetMaxOverlappingBytes(), path_id,
       GetCompressionType(vstorage_, mutable_cf_options_, output_level, 1),
       GetCompressionOptions(mutable_cf_options_, vstorage_, output_level),
-      Temperature::kUnknown,
+      mutable_cf_options_.default_write_temperature,
       /* max_subcompactions */ 0, grandparents, /* is manual */ false,
       /* trim_ts */ "", score_, false /* deletion_compaction */,
       /* l0_files_might_overlap */ true,
@@ -1440,7 +1442,7 @@ Compaction* UniversalCompactionBuilder::PickCompactionWithSortedRunRange(
                          true /* enable_compression */),
       GetCompressionOptions(mutable_cf_options_, vstorage_, output_level,
                             true /* enable_compression */),
-      Temperature::kUnknown,
+      mutable_cf_options_.default_write_temperature,
       /* max_subcompactions */ 0, /* grandparents */ {}, /* is manual */ false,
       /* trim_ts */ "", score_, false /* deletion_compaction */,
       /* l0_files_might_overlap */ true, compaction_reason);

db/compaction/compaction_service_job.cc

@@ -74,24 +74,24 @@ CompactionJob::ProcessKeyValueCompactionWithCompactionService(
       compaction_input.output_level, input_files_oss.str().c_str());
   CompactionServiceJobInfo info(dbname_, db_id_, db_session_id_,
                                 GetCompactionId(sub_compact), thread_pri_);
-  CompactionServiceJobStatus compaction_status =
-      db_options_.compaction_service->StartV2(info, compaction_input_binary);
-  switch (compaction_status) {
+  CompactionServiceScheduleResponse response =
+      db_options_.compaction_service->Schedule(info, compaction_input_binary);
+  switch (response.status) {
     case CompactionServiceJobStatus::kSuccess:
       break;
     case CompactionServiceJobStatus::kFailure:
       sub_compact->status = Status::Incomplete(
-          "CompactionService failed to start compaction job.");
+          "CompactionService failed to schedule a remote compaction job.");
       ROCKS_LOG_WARN(db_options_.info_log,
                      "[%s] [JOB %d] Remote compaction failed to start.",
                      compaction_input.column_family.name.c_str(), job_id_);
-      return compaction_status;
+      return response.status;
     case CompactionServiceJobStatus::kUseLocal:
       ROCKS_LOG_INFO(
           db_options_.info_log,
-          "[%s] [JOB %d] Remote compaction fallback to local by API Start.",
+          "[%s] [JOB %d] Remote compaction fallback to local by API (Schedule)",
           compaction_input.column_family.name.c_str(), job_id_);
-      return compaction_status;
+      return response.status;
     default:
       assert(false);  // unknown status
       break;
@@ -101,14 +101,15 @@ CompactionJob::ProcessKeyValueCompactionWithCompactionService(
                  "[%s] [JOB %d] Waiting for remote compaction...",
                  compaction_input.column_family.name.c_str(), job_id_);
   std::string compaction_result_binary;
-  compaction_status = db_options_.compaction_service->WaitForCompleteV2(
-      info, &compaction_result_binary);
+  CompactionServiceJobStatus compaction_status =
+      db_options_.compaction_service->Wait(response.scheduled_job_id,
+                                           &compaction_result_binary);
 
   if (compaction_status == CompactionServiceJobStatus::kUseLocal) {
-    ROCKS_LOG_INFO(db_options_.info_log,
-                   "[%s] [JOB %d] Remote compaction fallback to local by API "
-                   "WaitForComplete.",
-                   compaction_input.column_family.name.c_str(), job_id_);
+    ROCKS_LOG_INFO(
+        db_options_.info_log,
+        "[%s] [JOB %d] Remote compaction fallback to local by API (Wait)",
+        compaction_input.column_family.name.c_str(), job_id_);
     return compaction_status;
   }
 
@@ -830,4 +831,3 @@ bool CompactionServiceInput::TEST_Equals(CompactionServiceInput* other,
 }
 #endif  // NDEBUG
 }  // namespace ROCKSDB_NAMESPACE
-

db/compaction/compaction_service_test.cc

@@ -31,40 +31,44 @@ class MyTestCompactionService : public CompactionService {
 
   const char* Name() const override { return kClassName(); }
 
-  CompactionServiceJobStatus StartV2(
+  CompactionServiceScheduleResponse Schedule(
       const CompactionServiceJobInfo& info,
       const std::string& compaction_service_input) override {
     InstrumentedMutexLock l(&mutex_);
     start_info_ = info;
     assert(info.db_name == db_path_);
-    jobs_.emplace(info.job_id, compaction_service_input);
-    CompactionServiceJobStatus s = CompactionServiceJobStatus::kSuccess;
-    if (is_override_start_status_) {
-      return override_start_status_;
-    }
-    return s;
+    std::string unique_id = Env::Default()->GenerateUniqueId();
+    jobs_.emplace(unique_id, compaction_service_input);
+    infos_.emplace(unique_id, info);
+    CompactionServiceScheduleResponse response(
+        unique_id, is_override_start_status_
+                       ? override_start_status_
+                       : CompactionServiceJobStatus::kSuccess);
+    return response;
   }
 
-  CompactionServiceJobStatus WaitForCompleteV2(
-      const CompactionServiceJobInfo& info,
-      std::string* compaction_service_result) override {
+  CompactionServiceJobStatus Wait(const std::string& scheduled_job_id,
+                                  std::string* result) override {
     std::string compaction_input;
-    assert(info.db_name == db_path_);
     {
       InstrumentedMutexLock l(&mutex_);
-      wait_info_ = info;
-      auto i = jobs_.find(info.job_id);
-      if (i == jobs_.end()) {
+      auto job_index = jobs_.find(scheduled_job_id);
+      if (job_index == jobs_.end()) {
         return CompactionServiceJobStatus::kFailure;
       }
-      compaction_input = std::move(i->second);
-      jobs_.erase(i);
-    }
+      compaction_input = std::move(job_index->second);
+      jobs_.erase(job_index);
 
+      auto info_index = infos_.find(scheduled_job_id);
+      if (info_index == infos_.end()) {
+        return CompactionServiceJobStatus::kFailure;
+      }
+      wait_info_ = std::move(info_index->second);
+      infos_.erase(info_index);
+    }
     if (is_override_wait_status_) {
       return override_wait_status_;
     }
-
     CompactionServiceOptionsOverride options_override;
     options_override.env = options_.env;
     options_override.file_checksum_gen_factory =
@@ -90,11 +94,11 @@ class MyTestCompactionService : public CompactionService {
     OpenAndCompactOptions options;
     options.canceled = &canceled_;
 
-    Status s = DB::OpenAndCompact(
-        options, db_path_, db_path_ + "/" + std::to_string(info.job_id),
-        compaction_input, compaction_service_result, options_override);
+    Status s =
+        DB::OpenAndCompact(options, db_path_, db_path_ + "/" + scheduled_job_id,
+                           compaction_input, result, options_override);
     if (is_override_wait_result_) {
-      *compaction_service_result = override_wait_result_;
+      *result = override_wait_result_;
     }
     compaction_num_.fetch_add(1);
     if (s.ok()) {
@@ -135,7 +139,8 @@ class MyTestCompactionService : public CompactionService {
  private:
   InstrumentedMutex mutex_;
   std::atomic_int compaction_num_{0};
-  std::map<uint64_t, std::string> jobs_;
+  std::map<std::string, std::string> jobs_;
+  std::map<std::string, CompactionServiceJobInfo> infos_;
   const std::string db_path_;
   Options options_;
   std::shared_ptr<Statistics> statistics_;
@@ -172,6 +177,7 @@ class CompactionServiceTest : public DBTestBase {
         remote_table_properties_collector_factories);
     options->compaction_service = compaction_service_;
     DestroyAndReopen(*options);
+    CreateAndReopenWithCF({"cf_1", "cf_2", "cf_3"}, *options);
   }
 
   Statistics* GetCompactorStatistics() { return compactor_statistics_.get(); }
@@ -183,36 +189,45 @@ class CompactionServiceTest : public DBTestBase {
     return static_cast_with_check<MyTestCompactionService>(cs);
   }
 
-  void GenerateTestData() {
-    // Generate 20 files @ L2
-    for (int i = 0; i < 20; i++) {
-      for (int j = 0; j < 10; j++) {
-        int key_id = i * 10 + j;
-        ASSERT_OK(Put(Key(key_id), "value" + std::to_string(key_id)));
+  void GenerateTestData(bool move_files_manually = false) {
+    // Generate 20 files @ L2 Per CF
+    for (int cf_id = 0; cf_id < static_cast<int>(handles_.size()); cf_id++) {
+      for (int i = 0; i < 20; i++) {
+        for (int j = 0; j < 10; j++) {
+          int key_id = i * 10 + j;
+          ASSERT_OK(Put(cf_id, Key(key_id), "value" + std::to_string(key_id)));
+        }
+        ASSERT_OK(Flush(cf_id));
+      }
+      if (move_files_manually) {
+        MoveFilesToLevel(2, cf_id);
       }
-      ASSERT_OK(Flush());
-    }
-    MoveFilesToLevel(2);
 
-    // Generate 10 files @ L1 overlap with all 20 files @ L2
-    for (int i = 0; i < 10; i++) {
-      for (int j = 0; j < 10; j++) {
-        int key_id = i * 20 + j * 2;
-        ASSERT_OK(Put(Key(key_id), "value_new" + std::to_string(key_id)));
+      // Generate 10 files @ L1 overlap with all 20 files @ L2
+      for (int i = 0; i < 10; i++) {
+        for (int j = 0; j < 10; j++) {
+          int key_id = i * 20 + j * 2;
+          ASSERT_OK(
+              Put(cf_id, Key(key_id), "value_new" + std::to_string(key_id)));
+        }
+        ASSERT_OK(Flush(cf_id));
+      }
+      if (move_files_manually) {
+        MoveFilesToLevel(1, cf_id);
+        ASSERT_EQ(FilesPerLevel(cf_id), "0,10,20");
       }
-      ASSERT_OK(Flush());
     }
-    MoveFilesToLevel(1);
-    ASSERT_EQ(FilesPerLevel(), "0,10,20");
   }
 
   void VerifyTestData() {
-    for (int i = 0; i < 200; i++) {
-      auto result = Get(Key(i));
-      if (i % 2) {
-        ASSERT_EQ(result, "value" + std::to_string(i));
-      } else {
-        ASSERT_EQ(result, "value_new" + std::to_string(i));
+    for (int cf_id = 0; cf_id < static_cast<int>(handles_.size()); cf_id++) {
+      for (int i = 0; i < 200; i++) {
+        auto result = Get(cf_id, Key(i));
+        if (i % 2) {
+          ASSERT_EQ(result, "value" + std::to_string(i));
+        } else {
+          ASSERT_EQ(result, "value_new" + std::to_string(i));
+        }
       }
     }
   }
@@ -234,32 +249,10 @@ TEST_F(CompactionServiceTest, BasicCompactions) {
   Statistics* primary_statistics = GetPrimaryStatistics();
   Statistics* compactor_statistics = GetCompactorStatistics();
 
-  for (int i = 0; i < 20; i++) {
-    for (int j = 0; j < 10; j++) {
-      int key_id = i * 10 + j;
-      ASSERT_OK(Put(Key(key_id), "value" + std::to_string(key_id)));
-    }
-    ASSERT_OK(Flush());
-  }
-
-  for (int i = 0; i < 10; i++) {
-    for (int j = 0; j < 10; j++) {
-      int key_id = i * 20 + j * 2;
-      ASSERT_OK(Put(Key(key_id), "value_new" + std::to_string(key_id)));
-    }
-    ASSERT_OK(Flush());
-  }
+  GenerateTestData();
   ASSERT_OK(dbfull()->TEST_WaitForCompact());
+  VerifyTestData();
 
-  // verify result
-  for (int i = 0; i < 200; i++) {
-    auto result = Get(Key(i));
-    if (i % 2) {
-      ASSERT_EQ(result, "value" + std::to_string(i));
-    } else {
-      ASSERT_EQ(result, "value_new" + std::to_string(i));
-    }
-  }
   auto my_cs = GetCompactionService();
   ASSERT_GE(my_cs->GetCompactionNum(), 1);
 
@@ -322,7 +315,8 @@ TEST_F(CompactionServiceTest, BasicCompactions) {
         assert(*id != kNullUniqueId64x2);
         verify_passed++;
       });
-  Reopen(options);
+  ReopenWithColumnFamilies({kDefaultColumnFamilyName, "cf_1", "cf_2", "cf_3"},
+                           options);
   ASSERT_GT(verify_passed, 0);
   Close();
 }
@@ -490,26 +484,9 @@ TEST_F(CompactionServiceTest, CompactionFilter) {
       new PartialDeleteCompactionFilter());
   options.compaction_filter = delete_comp_filter.get();
   ReopenWithCompactionService(&options);
-
-  for (int i = 0; i < 20; i++) {
-    for (int j = 0; j < 10; j++) {
-      int key_id = i * 10 + j;
-      ASSERT_OK(Put(Key(key_id), "value" + std::to_string(key_id)));
-    }
-    ASSERT_OK(Flush());
-  }
-
-  for (int i = 0; i < 10; i++) {
-    for (int j = 0; j < 10; j++) {
-      int key_id = i * 20 + j * 2;
-      ASSERT_OK(Put(Key(key_id), "value_new" + std::to_string(key_id)));
-    }
-    ASSERT_OK(Flush());
-  }
+  GenerateTestData();
   ASSERT_OK(dbfull()->TEST_WaitForCompact());
-
   ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
-
   // verify result
   for (int i = 0; i < 200; i++) {
     auto result = Get(Key(i));
@@ -551,17 +528,17 @@ TEST_F(CompactionServiceTest, ConcurrentCompaction) {
   options.level0_file_num_compaction_trigger = 100;
   options.max_background_jobs = 20;
   ReopenWithCompactionService(&options);
-  GenerateTestData();
+  GenerateTestData(true);
 
   ColumnFamilyMetaData meta;
   db_->GetColumnFamilyMetaData(&meta);
 
   std::vector<std::thread> threads;
   for (const auto& file : meta.levels[1].files) {
-    threads.emplace_back(std::thread([&]() {
+    threads.emplace_back([&]() {
       std::string fname = file.db_path + "/" + file.name;
       ASSERT_OK(db_->CompactFiles(CompactionOptions(), {fname}, 2));
-    }));
+    });
   }
 
   for (auto& thread : threads) {
@@ -570,14 +547,7 @@ TEST_F(CompactionServiceTest, ConcurrentCompaction) {
   ASSERT_OK(dbfull()->TEST_WaitForCompact());
 
   // verify result
-  for (int i = 0; i < 200; i++) {
-    auto result = Get(Key(i));
-    if (i % 2) {
-      ASSERT_EQ(result, "value" + std::to_string(i));
-    } else {
-      ASSERT_EQ(result, "value_new" + std::to_string(i));
-    }
-  }
+  VerifyTestData();
   auto my_cs = GetCompactionService();
   ASSERT_EQ(my_cs->GetCompactionNum(), 10);
   ASSERT_EQ(FilesPerLevel(), "0,0,10");
@@ -587,21 +557,7 @@ TEST_F(CompactionServiceTest, CompactionInfo) {
   Options options = CurrentOptions();
   ReopenWithCompactionService(&options);
 
-  for (int i = 0; i < 20; i++) {
-    for (int j = 0; j < 10; j++) {
-      int key_id = i * 10 + j;
-      ASSERT_OK(Put(Key(key_id), "value" + std::to_string(key_id)));
-    }
-    ASSERT_OK(Flush());
-  }
-
-  for (int i = 0; i < 10; i++) {
-    for (int j = 0; j < 10; j++) {
-      int key_id = i * 20 + j * 2;
-      ASSERT_OK(Put(Key(key_id), "value_new" + std::to_string(key_id)));
-    }
-    ASSERT_OK(Flush());
-  }
+  GenerateTestData();
   ASSERT_OK(dbfull()->TEST_WaitForCompact());
   auto my_cs =
       static_cast_with_check<MyTestCompactionService>(GetCompactionService());
@@ -676,32 +632,9 @@ TEST_F(CompactionServiceTest, FallbackLocalAuto) {
 
   my_cs->OverrideStartStatus(CompactionServiceJobStatus::kUseLocal);
 
-  for (int i = 0; i < 20; i++) {
-    for (int j = 0; j < 10; j++) {
-      int key_id = i * 10 + j;
-      ASSERT_OK(Put(Key(key_id), "value" + std::to_string(key_id)));
-    }
-    ASSERT_OK(Flush());
-  }
-
-  for (int i = 0; i < 10; i++) {
-    for (int j = 0; j < 10; j++) {
-      int key_id = i * 20 + j * 2;
-      ASSERT_OK(Put(Key(key_id), "value_new" + std::to_string(key_id)));
-    }
-    ASSERT_OK(Flush());
-  }
+  GenerateTestData();
   ASSERT_OK(dbfull()->TEST_WaitForCompact());
-
-  // verify result
-  for (int i = 0; i < 200; i++) {
-    auto result = Get(Key(i));
-    if (i % 2) {
-      ASSERT_EQ(result, "value" + std::to_string(i));
-    } else {
-      ASSERT_EQ(result, "value_new" + std::to_string(i));
-    }
-  }
+  VerifyTestData();
 
   ASSERT_EQ(my_cs->GetCompactionNum(), 0);
 

db/compaction/tiered_compaction_test.cc

@@ -14,12 +14,11 @@
 #include "rocksdb/listener.h"
 #include "rocksdb/utilities/debug.h"
 #include "test_util/mock_time_env.h"
+#include "utilities/merge_operators.h"
 
 namespace ROCKSDB_NAMESPACE {
 
-
-class TieredCompactionTest : public DBTestBase,
-                             public testing::WithParamInterface<bool> {
+class TieredCompactionTest : public DBTestBase {
  public:
   TieredCompactionTest()
       : DBTestBase("tiered_compaction_test", /*env_do_fsync=*/true),
@@ -123,14 +122,8 @@ class TieredCompactionTest : public DBTestBase,
     pl_stats.Clear();
   }
 
-  // bottommost_temperature is renaming to last_level_temperature, set either
-  // of them should have the same effect.
   void SetColdTemperature(Options& options) {
-    if (GetParam()) {
-      options.bottommost_temperature = Temperature::kCold;
-    } else {
-      options.last_level_temperature = Temperature::kCold;
-    }
+    options.last_level_temperature = Temperature::kCold;
   }
 
  private:
@@ -172,7 +165,7 @@ class TieredCompactionTest : public DBTestBase,
   }
 };
 
-TEST_P(TieredCompactionTest, SequenceBasedTieredStorageUniversal) {
+TEST_F(TieredCompactionTest, SequenceBasedTieredStorageUniversal) {
   const int kNumTrigger = 4;
   const int kNumLevels = 7;
   const int kNumKeys = 100;
@@ -334,7 +327,7 @@ TEST_P(TieredCompactionTest, SequenceBasedTieredStorageUniversal) {
   ASSERT_GT(GetSstSizeHelper(Temperature::kCold), 0);
 }
 
-TEST_P(TieredCompactionTest, RangeBasedTieredStorageUniversal) {
+TEST_F(TieredCompactionTest, RangeBasedTieredStorageUniversal) {
   const int kNumTrigger = 4;
   const int kNumLevels = 7;
   const int kNumKeys = 100;
@@ -508,7 +501,7 @@ TEST_P(TieredCompactionTest, RangeBasedTieredStorageUniversal) {
       1);
 }
 
-TEST_P(TieredCompactionTest, LevelColdRangeDelete) {
+TEST_F(TieredCompactionTest, LevelColdRangeDelete) {
   const int kNumTrigger = 4;
   const int kNumLevels = 7;
   const int kNumKeys = 100;
@@ -614,7 +607,7 @@ class SingleKeySstPartitionerFactory : public SstPartitionerFactory {
   }
 };
 
-TEST_P(TieredCompactionTest, LevelOutofBoundaryRangeDelete) {
+TEST_F(TieredCompactionTest, LevelOutofBoundaryRangeDelete) {
   const int kNumTrigger = 4;
   const int kNumLevels = 3;
   const int kNumKeys = 10;
@@ -743,7 +736,7 @@ TEST_P(TieredCompactionTest, LevelOutofBoundaryRangeDelete) {
   ASSERT_GT(GetSstSizeHelper(Temperature::kCold), 0);
 }
 
-TEST_P(TieredCompactionTest, UniversalRangeDelete) {
+TEST_F(TieredCompactionTest, UniversalRangeDelete) {
   const int kNumTrigger = 4;
   const int kNumLevels = 7;
   const int kNumKeys = 10;
@@ -875,7 +868,7 @@ TEST_P(TieredCompactionTest, UniversalRangeDelete) {
   ASSERT_GT(GetSstSizeHelper(Temperature::kCold), 0);
 }
 
-TEST_P(TieredCompactionTest, SequenceBasedTieredStorageLevel) {
+TEST_F(TieredCompactionTest, SequenceBasedTieredStorageLevel) {
   const int kNumTrigger = 4;
   const int kNumLevels = 7;
   const int kNumKeys = 100;
@@ -1099,7 +1092,7 @@ TEST_P(TieredCompactionTest, SequenceBasedTieredStorageLevel) {
   ASSERT_GT(GetSstSizeHelper(Temperature::kCold), 0);
 }
 
-TEST_P(TieredCompactionTest, RangeBasedTieredStorageLevel) {
+TEST_F(TieredCompactionTest, RangeBasedTieredStorageLevel) {
   const int kNumTrigger = 4;
   const int kNumLevels = 7;
   const int kNumKeys = 100;
@@ -1240,10 +1233,7 @@ TEST_P(TieredCompactionTest, RangeBasedTieredStorageLevel) {
   db_->ReleaseSnapshot(temp_snap);
 }
 
-INSTANTIATE_TEST_CASE_P(TieredCompactionTest, TieredCompactionTest,
-                        testing::Bool());
-
-TEST_P(TieredCompactionTest, CheckInternalKeyRange) {
+TEST_F(TieredCompactionTest, 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.
@@ -1318,8 +1308,8 @@ TEST_P(TieredCompactionTest, CheckInternalKeyRange) {
 
 class PrecludeLastLevelTest : public DBTestBase {
  public:
-  PrecludeLastLevelTest()
-      : DBTestBase("preclude_last_level_test", /*env_do_fsync=*/false) {
+  PrecludeLastLevelTest(std::string test_name = "preclude_last_level_test")
+      : DBTestBase(test_name, /*env_do_fsync=*/false) {
     mock_clock_ = std::make_shared<MockSystemClock>(env_->GetSystemClock());
     mock_clock_->SetCurrentTime(kMockStartTime);
     mock_env_ = std::make_unique<CompositeEnvWrapper>(env_, mock_clock_);
@@ -1594,6 +1584,159 @@ TEST_F(PrecludeLastLevelTest, SmallPrecludeTime) {
   Close();
 }
 
+// Test Param: protection_bytes_per_key for WriteBatch
+class TimedPutPrecludeLastLevelTest
+    : public PrecludeLastLevelTest,
+      public testing::WithParamInterface<size_t> {
+ public:
+  TimedPutPrecludeLastLevelTest()
+      : PrecludeLastLevelTest("timed_put_preclude_last_level_test") {}
+};
+
+TEST_P(TimedPutPrecludeLastLevelTest, FastTrackTimedPutToLastLevel) {
+  const int kNumTrigger = 4;
+  const int kNumLevels = 7;
+  const int kNumKeys = 100;
+
+  Options options = CurrentOptions();
+  options.compaction_style = kCompactionStyleUniversal;
+  options.preclude_last_level_data_seconds = 60;
+  options.preserve_internal_time_seconds = 0;
+  options.env = mock_env_.get();
+  options.level0_file_num_compaction_trigger = kNumTrigger;
+  options.num_levels = kNumLevels;
+  options.last_level_temperature = Temperature::kCold;
+  DestroyAndReopen(options);
+  WriteOptions wo;
+  wo.protection_bytes_per_key = GetParam();
+
+  Random rnd(301);
+
+  dbfull()->TEST_WaitForPeriodicTaskRun([&] {
+    mock_clock_->MockSleepForSeconds(static_cast<int>(rnd.Uniform(10) + 1));
+  });
+
+  for (int i = 0; i < kNumKeys / 2; i++) {
+    ASSERT_OK(Put(Key(i), rnd.RandomString(100), wo));
+    dbfull()->TEST_WaitForPeriodicTaskRun([&] {
+      mock_clock_->MockSleepForSeconds(static_cast<int>(rnd.Uniform(2)));
+    });
+  }
+  // Create one file with regular Put.
+  ASSERT_OK(Flush());
+
+  // Create one file with TimedPut.
+  // With above mock clock operations, write_unix_time 50 should be before
+  // current_time - preclude_last_level_seconds.
+  // These data are eligible to be put on the last level once written to db
+  // and compaction will fast track them to the last level.
+  for (int i = kNumKeys / 2; i < kNumKeys; i++) {
+    ASSERT_OK(TimedPut(0, Key(i), rnd.RandomString(100), 50, wo));
+  }
+  ASSERT_OK(Flush());
+
+  // TimedPut file moved to the last level immediately.
+  ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
+  ASSERT_EQ("0,0,0,0,0,1,1", FilesPerLevel());
+
+  // Wait more than preclude_last_level time, Put file eventually moved to the
+  // last level.
+  mock_clock_->MockSleepForSeconds(100);
+
+  ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
+  ASSERT_EQ("0,0,0,0,0,0,1", FilesPerLevel());
+  ASSERT_EQ(GetSstSizeHelper(Temperature::kUnknown), 0);
+  ASSERT_GT(GetSstSizeHelper(Temperature::kCold), 0);
+
+  Close();
+}
+
+TEST_P(TimedPutPrecludeLastLevelTest, InterleavedTimedPutAndPut) {
+  Options options = CurrentOptions();
+  options.compaction_style = kCompactionStyleUniversal;
+  options.disable_auto_compactions = true;
+  options.preclude_last_level_data_seconds = 1 * 24 * 60 * 60;
+  options.env = mock_env_.get();
+  options.num_levels = 7;
+  options.last_level_temperature = Temperature::kCold;
+  options.default_write_temperature = Temperature::kHot;
+  DestroyAndReopen(options);
+  WriteOptions wo;
+  wo.protection_bytes_per_key = GetParam();
+
+  // Start time: kMockStartTime = 10000000;
+  ASSERT_OK(TimedPut(0, Key(0), "v0", kMockStartTime - 1 * 24 * 60 * 60, wo));
+  ASSERT_OK(Put(Key(1), "v1", wo));
+  ASSERT_OK(Flush());
+
+  ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
+  ASSERT_EQ("0,0,0,0,0,1,1", FilesPerLevel());
+  ASSERT_GT(GetSstSizeHelper(Temperature::kHot), 0);
+  ASSERT_GT(GetSstSizeHelper(Temperature::kCold), 0);
+  Close();
+}
+
+TEST_P(TimedPutPrecludeLastLevelTest, PreserveTimedPutOnPenultimateLevel) {
+  Options options = CurrentOptions();
+  options.compaction_style = kCompactionStyleUniversal;
+  options.disable_auto_compactions = true;
+  options.preclude_last_level_data_seconds = 3 * 24 * 60 * 60;
+  int seconds_between_recording = (3 * 24 * 60 * 60) / kMaxSeqnoTimePairsPerCF;
+  options.env = mock_env_.get();
+  options.num_levels = 7;
+  options.last_level_temperature = Temperature::kCold;
+  options.default_write_temperature = Temperature::kHot;
+  DestroyAndReopen(options);
+  WriteOptions wo;
+  wo.protection_bytes_per_key = GetParam();
+
+  // Creating a snapshot to manually control when preferred sequence number is
+  // swapped in. An entry's preferred seqno won't get swapped in until it's
+  // visible to the earliest snapshot. With this, we can test relevant seqno to
+  // time mapping recorded in SST file also covers preferred seqno, not just
+  // the seqno in the internal keys.
+  auto* snap1 = db_->GetSnapshot();
+  // Start time: kMockStartTime = 10000000;
+  ASSERT_OK(TimedPut(0, Key(0), "v0", kMockStartTime - 1 * 24 * 60 * 60, wo));
+  ASSERT_OK(TimedPut(0, Key(1), "v1", kMockStartTime - 1 * 24 * 60 * 60, wo));
+  ASSERT_OK(TimedPut(0, Key(2), "v2", kMockStartTime - 1 * 24 * 60 * 60, wo));
+  ASSERT_OK(Flush());
+
+  // Should still be in penultimate 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
+  // seqno to time mapping should continue to cover preferred seqno after
+  // compaction.
+  db_->ReleaseSnapshot(snap1);
+  mock_clock_->MockSleepForSeconds(1 * 24 * 60 * 60);
+  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 data are eligible to be placed on last level.
+  // Instead of waiting exactly one more day, here we waited
+  // `seconds_between_recording` less seconds to show that it's not precise.
+  // Data could start to be placed on cold tier one recording interval before
+  // they exactly become cold based on the setting. For this one column family
+  // setting preserving 3 days of recording, it's about 43 minutes.
+  mock_clock_->MockSleepForSeconds(1 * 24 * 60 * 60 -
+                                   seconds_between_recording);
+  ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
+  ASSERT_EQ("0,0,0,0,0,0,1", FilesPerLevel());
+  ASSERT_EQ(GetSstSizeHelper(Temperature::kHot), 0);
+  ASSERT_GT(GetSstSizeHelper(Temperature::kCold), 0);
+  Close();
+}
+
+INSTANTIATE_TEST_CASE_P(TimedPutPrecludeLastLevelTest,
+                        TimedPutPrecludeLastLevelTest, ::testing::Values(0, 8));
+
 TEST_F(PrecludeLastLevelTest, LastLevelOnlyCompactionPartial) {
   const int kNumTrigger = 4;
   const int kNumLevels = 7;
@@ -2211,6 +2354,309 @@ TEST_F(PrecludeLastLevelTest, RangeDelsCauseFileEndpointsToOverlap) {
   Close();
 }
 
+// Tests DBIter::GetProperty("rocksdb.iterator.write-time") return a data's
+// approximate write unix time.
+// Test Param:
+// 1) use tailing iterator or regular iterator (when it applies)
+class IteratorWriteTimeTest : public PrecludeLastLevelTest,
+                              public testing::WithParamInterface<bool> {
+ public:
+  IteratorWriteTimeTest() : PrecludeLastLevelTest("iterator_write_time_test") {}
+
+  uint64_t VerifyKeyAndGetWriteTime(Iterator* iter,
+                                    const std::string& expected_key) {
+    std::string prop;
+    uint64_t write_time = 0;
+    EXPECT_TRUE(iter->Valid());
+    EXPECT_EQ(expected_key, iter->key());
+    EXPECT_OK(iter->GetProperty("rocksdb.iterator.write-time", &prop));
+    Slice prop_slice = prop;
+    EXPECT_TRUE(GetFixed64(&prop_slice, &write_time));
+    return write_time;
+  }
+
+  void VerifyKeyAndWriteTime(Iterator* iter, const std::string& expected_key,
+                             uint64_t expected_write_time) {
+    std::string prop;
+    uint64_t write_time = 0;
+    EXPECT_TRUE(iter->Valid());
+    EXPECT_EQ(expected_key, iter->key());
+    EXPECT_OK(iter->GetProperty("rocksdb.iterator.write-time", &prop));
+    Slice prop_slice = prop;
+    EXPECT_TRUE(GetFixed64(&prop_slice, &write_time));
+    EXPECT_EQ(expected_write_time, write_time);
+  }
+};
+
+TEST_P(IteratorWriteTimeTest, ReadFromMemtables) {
+  const int kNumTrigger = 4;
+  const int kNumLevels = 7;
+  const int kNumKeys = 100;
+  const int kSecondsPerRecording = 101;
+  const int kKeyWithWriteTime = 25;
+  const uint64_t kUserSpecifiedWriteTime =
+      kMockStartTime + kSecondsPerRecording * 15;
+
+  Options options = CurrentOptions();
+  options.compaction_style = kCompactionStyleUniversal;
+  options.env = mock_env_.get();
+  options.level0_file_num_compaction_trigger = kNumTrigger;
+  options.preserve_internal_time_seconds = 10000;
+  options.num_levels = kNumLevels;
+  DestroyAndReopen(options);
+
+  Random rnd(301);
+  for (int i = 0; i < kNumKeys; i++) {
+    dbfull()->TEST_WaitForPeriodicTaskRun(
+        [&] { mock_clock_->MockSleepForSeconds(kSecondsPerRecording); });
+    if (i == kKeyWithWriteTime) {
+      ASSERT_OK(
+          TimedPut(Key(i), rnd.RandomString(100), kUserSpecifiedWriteTime));
+    } else {
+      ASSERT_OK(Put(Key(i), rnd.RandomString(100)));
+    }
+  }
+
+  ReadOptions ropts;
+  ropts.tailing = GetParam();
+  int i;
+
+  // Forward iteration
+  uint64_t start_time = 0;
+  {
+    std::unique_ptr<Iterator> iter(dbfull()->NewIterator(ropts));
+    for (iter->SeekToFirst(), i = 0; iter->Valid(); iter->Next(), i++) {
+      if (start_time == 0) {
+        start_time = VerifyKeyAndGetWriteTime(iter.get(), Key(i));
+      } else if (i == kKeyWithWriteTime) {
+        VerifyKeyAndWriteTime(iter.get(), Key(i), kUserSpecifiedWriteTime);
+      } else {
+        VerifyKeyAndWriteTime(iter.get(), Key(i),
+                              start_time + kSecondsPerRecording * (i + 1));
+      }
+    }
+    ASSERT_OK(iter->status());
+  }
+
+  // Backward iteration
+  {
+    ropts.tailing = false;
+    std::unique_ptr<Iterator> iter(dbfull()->NewIterator(ropts));
+    for (iter->SeekToLast(), i = kNumKeys - 1; iter->Valid();
+         iter->Prev(), i--) {
+      if (i == 0) {
+        VerifyKeyAndWriteTime(iter.get(), Key(i), start_time);
+      } else if (i == kKeyWithWriteTime) {
+        VerifyKeyAndWriteTime(iter.get(), Key(i), kUserSpecifiedWriteTime);
+      } else {
+        VerifyKeyAndWriteTime(iter.get(), Key(i),
+                              start_time + kSecondsPerRecording * (i + 1));
+      }
+    }
+    ASSERT_OK(iter->status());
+  }
+
+  // Reopen the DB and disable the seqno to time recording, data with user
+  // specified write time can still get a write time before it's flushed.
+  options.preserve_internal_time_seconds = 0;
+  DestroyAndReopen(options);
+  ASSERT_OK(TimedPut(Key(kKeyWithWriteTime), rnd.RandomString(100),
+                     kUserSpecifiedWriteTime));
+  {
+    std::unique_ptr<Iterator> iter(dbfull()->NewIterator(ropts));
+    iter->Seek(Key(kKeyWithWriteTime));
+    VerifyKeyAndWriteTime(iter.get(), Key(kKeyWithWriteTime),
+                          kUserSpecifiedWriteTime);
+    ASSERT_OK(iter->status());
+  }
+
+  ASSERT_OK(Flush());
+  {
+    std::unique_ptr<Iterator> iter(dbfull()->NewIterator(ropts));
+    iter->Seek(Key(kKeyWithWriteTime));
+    VerifyKeyAndWriteTime(iter.get(), Key(kKeyWithWriteTime),
+                          std::numeric_limits<uint64_t>::max());
+    ASSERT_OK(iter->status());
+  }
+
+  Close();
+}
+
+TEST_P(IteratorWriteTimeTest, ReadFromSstFile) {
+  const int kNumTrigger = 4;
+  const int kNumLevels = 7;
+  const int kNumKeys = 100;
+  const int kSecondsPerRecording = 101;
+  const int kKeyWithWriteTime = 25;
+  const uint64_t kUserSpecifiedWriteTime =
+      kMockStartTime + kSecondsPerRecording * 15;
+
+  Options options = CurrentOptions();
+  options.compaction_style = kCompactionStyleUniversal;
+  options.env = mock_env_.get();
+  options.level0_file_num_compaction_trigger = kNumTrigger;
+  options.preserve_internal_time_seconds = 10000;
+  options.num_levels = kNumLevels;
+  DestroyAndReopen(options);
+
+  Random rnd(301);
+  for (int i = 0; i < kNumKeys; i++) {
+    dbfull()->TEST_WaitForPeriodicTaskRun(
+        [&] { mock_clock_->MockSleepForSeconds(kSecondsPerRecording); });
+    if (i == kKeyWithWriteTime) {
+      ASSERT_OK(
+          TimedPut(Key(i), rnd.RandomString(100), kUserSpecifiedWriteTime));
+    } else {
+      ASSERT_OK(Put(Key(i), rnd.RandomString(100)));
+    }
+  }
+
+  ASSERT_OK(Flush());
+  ReadOptions ropts;
+  ropts.tailing = GetParam();
+  std::string prop;
+  int i;
+
+  // Forward iteration
+  uint64_t start_time = 0;
+  {
+    std::unique_ptr<Iterator> iter(dbfull()->NewIterator(ropts));
+    for (iter->SeekToFirst(), i = 0; iter->Valid(); iter->Next(), i++) {
+      if (start_time == 0) {
+        start_time = VerifyKeyAndGetWriteTime(iter.get(), Key(i));
+      } else if (i == kKeyWithWriteTime) {
+        // It's not precisely kUserSpecifiedWriteTime, instead it has a margin
+        // of error that is one recording apart while we convert write time to
+        // sequence number, and then back to write time.
+        VerifyKeyAndWriteTime(iter.get(), Key(i),
+                              kUserSpecifiedWriteTime - kSecondsPerRecording);
+      } else {
+        VerifyKeyAndWriteTime(iter.get(), Key(i),
+                              start_time + kSecondsPerRecording * (i + 1));
+      }
+    }
+    ASSERT_OK(iter->status());
+  }
+
+  // Backward iteration
+  {
+    ropts.tailing = false;
+    std::unique_ptr<Iterator> iter(dbfull()->NewIterator(ropts));
+    for (iter->SeekToLast(), i = kNumKeys - 1; iter->Valid();
+         iter->Prev(), i--) {
+      if (i == 0) {
+        VerifyKeyAndWriteTime(iter.get(), Key(i), start_time);
+      } else if (i == kKeyWithWriteTime) {
+        VerifyKeyAndWriteTime(iter.get(), Key(i),
+                              kUserSpecifiedWriteTime - kSecondsPerRecording);
+      } else {
+        VerifyKeyAndWriteTime(iter.get(), Key(i),
+                              start_time + kSecondsPerRecording * (i + 1));
+      }
+    }
+    ASSERT_OK(iter->status());
+  }
+
+  // Reopen the DB and disable the seqno to time recording. Data retrieved from
+  // SST files still have write time available.
+  options.preserve_internal_time_seconds = 0;
+  DestroyAndReopen(options);
+
+  dbfull()->TEST_WaitForPeriodicTaskRun(
+      [&] { mock_clock_->MockSleepForSeconds(kSecondsPerRecording); });
+  ASSERT_OK(Put("a", "val"));
+  ASSERT_TRUE(dbfull()->TEST_GetSeqnoToTimeMapping().Empty());
+
+  {
+    std::unique_ptr<Iterator> iter(dbfull()->NewIterator(ropts));
+    iter->SeekToFirst();
+    ASSERT_TRUE(iter->Valid());
+    // "a" is retrieved from memtable, its write time is unknown because the
+    // seqno to time mapping recording is not available.
+    VerifyKeyAndWriteTime(iter.get(), "a",
+                          std::numeric_limits<uint64_t>::max());
+    for (iter->Next(), i = 0; iter->Valid(); iter->Next(), i++) {
+      if (i == 0) {
+        VerifyKeyAndWriteTime(iter.get(), Key(i), start_time);
+      } else if (i == kKeyWithWriteTime) {
+        VerifyKeyAndWriteTime(iter.get(), Key(i),
+                              kUserSpecifiedWriteTime - kSecondsPerRecording);
+      } else {
+        VerifyKeyAndWriteTime(iter.get(), Key(i),
+                              start_time + kSecondsPerRecording * (i + 1));
+      }
+    }
+    ASSERT_OK(iter->status());
+  }
+
+  // There is no write time info for "a" after it's flushed to SST file either.
+  ASSERT_OK(Flush());
+  {
+    std::unique_ptr<Iterator> iter(dbfull()->NewIterator(ropts));
+    iter->SeekToFirst();
+    ASSERT_TRUE(iter->Valid());
+    VerifyKeyAndWriteTime(iter.get(), "a",
+                          std::numeric_limits<uint64_t>::max());
+  }
+
+  // Sequence number zeroed out after compacted to the last level, write time
+  // all becomes zero.
+  ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
+  {
+    std::unique_ptr<Iterator> iter(dbfull()->NewIterator(ropts));
+    iter->SeekToFirst();
+    for (iter->Next(), i = 0; iter->Valid(); iter->Next(), i++) {
+      VerifyKeyAndWriteTime(iter.get(), Key(i), 0);
+    }
+    ASSERT_OK(iter->status());
+  }
+  Close();
+}
+
+TEST_P(IteratorWriteTimeTest, MergeReturnsBaseValueWriteTime) {
+  const int kNumTrigger = 4;
+  const int kNumLevels = 7;
+  const int kSecondsPerRecording = 101;
+
+  Options options = CurrentOptions();
+  options.compaction_style = kCompactionStyleUniversal;
+  options.env = mock_env_.get();
+  options.level0_file_num_compaction_trigger = kNumTrigger;
+  options.preserve_internal_time_seconds = 10000;
+  options.num_levels = kNumLevels;
+  options.merge_operator = MergeOperators::CreateStringAppendOperator();
+  DestroyAndReopen(options);
+
+  dbfull()->TEST_WaitForPeriodicTaskRun(
+      [&] { mock_clock_->MockSleepForSeconds(kSecondsPerRecording); });
+  ASSERT_OK(Put("foo", "fv1"));
+
+  dbfull()->TEST_WaitForPeriodicTaskRun(
+      [&] { mock_clock_->MockSleepForSeconds(kSecondsPerRecording); });
+  ASSERT_OK(Put("bar", "bv1"));
+  ASSERT_OK(Merge("foo", "bv1"));
+
+  ReadOptions ropts;
+  ropts.tailing = GetParam();
+  {
+    std::unique_ptr<Iterator> iter(dbfull()->NewIterator(ropts));
+    iter->SeekToFirst();
+    uint64_t bar_time = VerifyKeyAndGetWriteTime(iter.get(), "bar");
+    iter->Next();
+    uint64_t foo_time = VerifyKeyAndGetWriteTime(iter.get(), "foo");
+    // "foo" has an older write time because its base value's write time is used
+    ASSERT_GT(bar_time, foo_time);
+    iter->Next();
+    ASSERT_FALSE(iter->Valid());
+    ASSERT_OK(iter->status());
+  }
+
+  Close();
+}
+
+INSTANTIATE_TEST_CASE_P(IteratorWriteTimeTest, IteratorWriteTimeTest,
+                        testing::Bool());
+
 }  // namespace ROCKSDB_NAMESPACE
 
 int main(int argc, char** argv) {

db/comparator_db_test.cc

@@ -170,7 +170,7 @@ void DoRandomIteraratorTest(DB* db, std::vector<std::string> source_strings,
 
 class DoubleComparator : public Comparator {
  public:
-  DoubleComparator() {}
+  DoubleComparator() = default;
 
   const char* Name() const override { return "DoubleComparator"; }
 
@@ -198,7 +198,7 @@ class DoubleComparator : public Comparator {
 
 class HashComparator : public Comparator {
  public:
-  HashComparator() {}
+  HashComparator() = default;
 
   const char* Name() const override { return "HashComparator"; }
 
@@ -221,7 +221,7 @@ class HashComparator : public Comparator {
 
 class TwoStrComparator : public Comparator {
  public:
-  TwoStrComparator() {}
+  TwoStrComparator() = default;
 
   const char* Name() const override { return "TwoStrComparator"; }
 
@@ -372,7 +372,7 @@ TEST_P(ComparatorDBTest, Uint64Comparator) {
       uint64_t r = rnd64.Next();
       std::string str;
       str.resize(8);
-      memcpy(&str[0], static_cast<void*>(&r), 8);
+      memcpy(str.data(), static_cast<void*>(&r), 8);
       source_strings.push_back(str);
     }
 

db/cuckoo_table_db_test.cc

@@ -209,7 +209,7 @@ static std::string Key(int i) {
 static std::string Uint64Key(uint64_t i) {
   std::string str;
   str.resize(8);
-  memcpy(&str[0], static_cast<void*>(&i), 8);
+  memcpy(str.data(), static_cast<void*>(&i), 8);
   return str;
 }
 }  // namespace.

db/db_basic_test.cc

@@ -442,6 +442,43 @@ TEST_F(DBBasicTest, PutSingleDeleteGet) {
                          kSkipMergePut));
 }
 
+TEST_F(DBBasicTest, TimedPutBasic) {
+  do {
+    Options options = CurrentOptions();
+    options.merge_operator = MergeOperators::CreateStringAppendOperator();
+    CreateAndReopenWithCF({"pikachu"}, options);
+    ASSERT_OK(TimedPut(1, "foo", "v1", /*write_unix_time=*/0));
+    // Read from memtable
+    ASSERT_EQ("v1", Get(1, "foo"));
+    ASSERT_OK(TimedPut(1, "foo", "v2.1", /*write_unix_time=*/3));
+    ASSERT_EQ("v2.1", Get(1, "foo"));
+
+    // Read from sst file
+    ASSERT_OK(db_->Flush(FlushOptions(), handles_[1]));
+    ASSERT_OK(Merge(1, "foo", "v2.2"));
+    ASSERT_EQ("v2.1,v2.2", Get(1, "foo"));
+    ASSERT_OK(Delete(1, "foo"));
+    ASSERT_EQ("NOT_FOUND", Get(1, "foo"));
+
+    ASSERT_OK(TimedPut(1, "bar", "bv1", /*write_unix_time=*/0));
+    ASSERT_EQ("bv1", Get(1, "bar"));
+    ASSERT_OK(TimedPut(1, "baz", "bzv1", /*write_unix_time=*/0));
+    ASSERT_EQ("bzv1", Get(1, "baz"));
+    if (option_config_ != kRowCache) {
+      std::string range_del_begin = "b";
+      std::string range_del_end = "baz";
+      Slice begin_rdel = range_del_begin, end_rdel = range_del_end;
+      ASSERT_OK(
+          db_->DeleteRange(WriteOptions(), handles_[1], begin_rdel, end_rdel));
+      ASSERT_EQ("NOT_FOUND", Get(1, "bar"));
+    }
+
+    ASSERT_EQ("bzv1", Get(1, "baz"));
+    ASSERT_OK(SingleDelete(1, "baz"));
+    ASSERT_EQ("NOT_FOUND", Get(1, "baz"));
+  } while (ChangeOptions(kSkipPlainTable));
+}
+
 TEST_F(DBBasicTest, EmptyFlush) {
   // It is possible to produce empty flushes when using single deletes. Tests
   // whether empty flushes cause issues.
@@ -1368,9 +1405,9 @@ TEST_P(DBMultiGetTestWithParam, MultiGetMultiCF) {
   for (int i = 0; i < num_keys; ++i) {
     int cf = i / 3;
     int cf_key = 1 % 3;
-    cf_kv_vec.emplace_back(std::make_tuple(
+    cf_kv_vec.emplace_back(
         cf, "cf" + std::to_string(cf) + "_key_" + std::to_string(cf_key),
-        "cf" + std::to_string(cf) + "_val_" + std::to_string(cf_key)));
+        "cf" + std::to_string(cf) + "_val_" + std::to_string(cf_key));
     ASSERT_OK(Put(std::get<0>(cf_kv_vec[i]), std::get<1>(cf_kv_vec[i]),
                   std::get<2>(cf_kv_vec[i])));
   }
@@ -1378,7 +1415,7 @@ TEST_P(DBMultiGetTestWithParam, MultiGetMultiCF) {
   int get_sv_count = 0;
   ROCKSDB_NAMESPACE::DBImpl* db = static_cast_with_check<DBImpl>(db_);
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
-      "DBImpl::MultiGet::AfterRefSV", [&](void* /*arg*/) {
+      "DBImpl::MultiCFSnapshot::AfterRefSV", [&](void* /*arg*/) {
         if (++get_sv_count == 2) {
           // After MultiGet refs a couple of CFs, flush all CFs so MultiGet
           // is forced to repeat the process
@@ -1478,9 +1515,10 @@ TEST_P(DBMultiGetTestWithParam, MultiGetMultiCFMutex) {
   int retries = 0;
   bool last_try = false;
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
-      "DBImpl::MultiGet::LastTry", [&](void* /*arg*/) { last_try = true; });
+      "DBImpl::MultiCFSnapshot::LastTry",
+      [&](void* /*arg*/) { last_try = true; });
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
-      "DBImpl::MultiGet::AfterRefSV", [&](void* /*arg*/) {
+      "DBImpl::MultiCFSnapshot::AfterRefSV", [&](void* /*arg*/) {
         if (last_try) {
           return;
         }
@@ -1496,10 +1534,10 @@ TEST_P(DBMultiGetTestWithParam, MultiGetMultiCFMutex) {
         }
       });
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->LoadDependency({
-      {"DBImpl::MultiGet::AfterLastTryRefSV",
+      {"DBImpl::MultiCFSnapshot::AfterLastTryRefSV",
        "DBMultiGetTestWithParam::MultiGetMultiCFMutex:BeforeCreateSV"},
       {"DBMultiGetTestWithParam::MultiGetMultiCFMutex:AfterCreateSV",
-       "DBImpl::MultiGet::BeforeLastTryUnRefSV"},
+       "DBImpl::MultiCFSnapshot::BeforeLastTryUnRefSV"},
   });
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
 
@@ -1565,7 +1603,7 @@ TEST_P(DBMultiGetTestWithParam, MultiGetMultiCFSnapshot) {
   int get_sv_count = 0;
   ROCKSDB_NAMESPACE::DBImpl* db = static_cast_with_check<DBImpl>(db_);
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
-      "DBImpl::MultiGet::AfterRefSV", [&](void* /*arg*/) {
+      "DBImpl::MultiCFSnapshot::AfterRefSV", [&](void* /*arg*/) {
         if (++get_sv_count == 2) {
           for (int i = 0; i < 8; ++i) {
             ASSERT_OK(Flush(i));
@@ -2607,9 +2645,9 @@ TEST_P(DBMultiGetAsyncIOTest, GetFromL1) {
   key_strs.push_back(Key(33));
   key_strs.push_back(Key(54));
   key_strs.push_back(Key(102));
-  keys.push_back(key_strs[0]);
-  keys.push_back(key_strs[1]);
-  keys.push_back(key_strs[2]);
+  keys.emplace_back(key_strs[0]);
+  keys.emplace_back(key_strs[1]);
+  keys.emplace_back(key_strs[2]);
   values.resize(keys.size());
   statuses.resize(keys.size());
 
@@ -2652,9 +2690,9 @@ TEST_P(DBMultiGetAsyncIOTest, GetFromL1Error) {
   key_strs.push_back(Key(33));
   key_strs.push_back(Key(54));
   key_strs.push_back(Key(102));
-  keys.push_back(key_strs[0]);
-  keys.push_back(key_strs[1]);
-  keys.push_back(key_strs[2]);
+  keys.emplace_back(key_strs[0]);
+  keys.emplace_back(key_strs[1]);
+  keys.emplace_back(key_strs[2]);
   values.resize(keys.size());
   statuses.resize(keys.size());
 
@@ -2717,9 +2755,9 @@ TEST_P(DBMultiGetAsyncIOTest, LastKeyInFile) {
   key_strs.push_back(Key(21));
   key_strs.push_back(Key(54));
   key_strs.push_back(Key(102));
-  keys.push_back(key_strs[0]);
-  keys.push_back(key_strs[1]);
-  keys.push_back(key_strs[2]);
+  keys.emplace_back(key_strs[0]);
+  keys.emplace_back(key_strs[1]);
+  keys.emplace_back(key_strs[2]);
   values.resize(keys.size());
   statuses.resize(keys.size());
 
@@ -2762,9 +2800,9 @@ TEST_P(DBMultiGetAsyncIOTest, GetFromL1AndL2) {
   key_strs.push_back(Key(33));
   key_strs.push_back(Key(56));
   key_strs.push_back(Key(102));
-  keys.push_back(key_strs[0]);
-  keys.push_back(key_strs[1]);
-  keys.push_back(key_strs[2]);
+  keys.emplace_back(key_strs[0]);
+  keys.emplace_back(key_strs[1]);
+  keys.emplace_back(key_strs[2]);
   values.resize(keys.size());
   statuses.resize(keys.size());
 
@@ -2805,8 +2843,8 @@ TEST_P(DBMultiGetAsyncIOTest, GetFromL2WithRangeOverlapL0L1) {
   // 19 and 26 are in L2, but overlap with L0 and L1 file ranges
   key_strs.push_back(Key(19));
   key_strs.push_back(Key(26));
-  keys.push_back(key_strs[0]);
-  keys.push_back(key_strs[1]);
+  keys.emplace_back(key_strs[0]);
+  keys.emplace_back(key_strs[1]);
   values.resize(keys.size());
   statuses.resize(keys.size());
 
@@ -2841,8 +2879,8 @@ TEST_P(DBMultiGetAsyncIOTest, GetFromL2WithRangeDelInL1) {
   // 139 and 163 are in L2, but overlap with a range deletes in L1
   key_strs.push_back(Key(139));
   key_strs.push_back(Key(163));
-  keys.push_back(key_strs[0]);
-  keys.push_back(key_strs[1]);
+  keys.emplace_back(key_strs[0]);
+  keys.emplace_back(key_strs[1]);
   values.resize(keys.size());
   statuses.resize(keys.size());
 
@@ -2871,9 +2909,9 @@ TEST_P(DBMultiGetAsyncIOTest, GetFromL1AndL2WithRangeDelInL1) {
   key_strs.push_back(Key(139));
   key_strs.push_back(Key(144));
   key_strs.push_back(Key(163));
-  keys.push_back(key_strs[0]);
-  keys.push_back(key_strs[1]);
-  keys.push_back(key_strs[2]);
+  keys.emplace_back(key_strs[0]);
+  keys.emplace_back(key_strs[1]);
+  keys.emplace_back(key_strs[2]);
   values.resize(keys.size());
   statuses.resize(keys.size());
 
@@ -2904,9 +2942,9 @@ TEST_P(DBMultiGetAsyncIOTest, GetNoIOUring) {
   key_strs.push_back(Key(33));
   key_strs.push_back(Key(54));
   key_strs.push_back(Key(102));
-  keys.push_back(key_strs[0]);
-  keys.push_back(key_strs[1]);
-  keys.push_back(key_strs[2]);
+  keys.emplace_back(key_strs[0]);
+  keys.emplace_back(key_strs[1]);
+  keys.emplace_back(key_strs[2]);
   values.resize(keys.size());
   statuses.resize(keys.size());
 
@@ -3285,9 +3323,9 @@ TEST_F(DBBasicTest, MultiGetIOBufferOverrun) {
 
   // Warm up the cache first
   key_data.emplace_back(Key(0));
-  keys.emplace_back(Slice(key_data.back()));
+  keys.emplace_back(key_data.back());
   key_data.emplace_back(Key(50));
-  keys.emplace_back(Slice(key_data.back()));
+  keys.emplace_back(key_data.back());
   statuses.resize(keys.size());
 
   dbfull()->MultiGet(ro, dbfull()->DefaultColumnFamily(), keys.size(),
@@ -3661,10 +3699,10 @@ TEST_F(DBBasicTest, ConcurrentlyCloseDB) {
   DestroyAndReopen(options);
   std::vector<std::thread> workers;
   for (int i = 0; i < 10; i++) {
-    workers.push_back(std::thread([&]() {
+    workers.emplace_back([&]() {
       auto s = db_->Close();
       ASSERT_OK(s);
-    }));
+    });
   }
   for (auto& w : workers) {
     w.join();
@@ -3938,9 +3976,9 @@ TEST_P(DBBasicTestWithParallelIO, MultiGet) {
 
   // Warm up the cache first
   key_data.emplace_back(Key(0));
-  keys.emplace_back(Slice(key_data.back()));
+  keys.emplace_back(key_data.back());
   key_data.emplace_back(Key(50));
-  keys.emplace_back(Slice(key_data.back()));
+  keys.emplace_back(key_data.back());
   statuses.resize(keys.size());
 
   dbfull()->MultiGet(ro, dbfull()->DefaultColumnFamily(), keys.size(),
@@ -4119,9 +4157,9 @@ TEST_P(DBBasicTestWithParallelIO, MultiGetDirectIO) {
 
   // Warm up the cache first
   key_data.emplace_back(Key(0));
-  keys.emplace_back(Slice(key_data.back()));
+  keys.emplace_back(key_data.back());
   key_data.emplace_back(Key(50));
-  keys.emplace_back(Slice(key_data.back()));
+  keys.emplace_back(key_data.back());
   statuses.resize(keys.size());
 
   dbfull()->MultiGet(ro, dbfull()->DefaultColumnFamily(), keys.size(),
@@ -4189,9 +4227,9 @@ TEST_P(DBBasicTestWithParallelIO, MultiGetWithChecksumMismatch) {
 
   // Warm up the cache first
   key_data.emplace_back(Key(0));
-  keys.emplace_back(Slice(key_data.back()));
+  keys.emplace_back(key_data.back());
   key_data.emplace_back(Key(50));
-  keys.emplace_back(Slice(key_data.back()));
+  keys.emplace_back(key_data.back());
   statuses.resize(keys.size());
 
   dbfull()->MultiGet(ro, dbfull()->DefaultColumnFamily(), keys.size(),
@@ -4237,9 +4275,9 @@ TEST_P(DBBasicTestWithParallelIO, MultiGetWithMissingFile) {
 
   // Warm up the cache first
   key_data.emplace_back(Key(0));
-  keys.emplace_back(Slice(key_data.back()));
+  keys.emplace_back(key_data.back());
   key_data.emplace_back(Key(50));
-  keys.emplace_back(Slice(key_data.back()));
+  keys.emplace_back(key_data.back());
   statuses.resize(keys.size());
 
   dbfull()->MultiGet(ro, dbfull()->DefaultColumnFamily(), keys.size(),
@@ -4743,7 +4781,7 @@ TEST_F(DBBasicTest, VerifyFileChecksumsReadahead) {
   uint64_t number;
   FileType type;
   ASSERT_OK(env_->GetChildren(dbname_, &filenames));
-  for (auto name : filenames) {
+  for (const auto& name : filenames) {
     if (ParseFileName(name, &number, &type)) {
       if (type == kTableFile) {
         sst_cnt++;

db/db_block_cache_test.cc

@@ -744,7 +744,7 @@ TEST_F(DBBlockCacheTest, AddRedundantStats) {
   const size_t capacity = size_t{1} << 25;
   const int num_shard_bits = 0;  // 1 shard
   int iterations_tested = 0;
-  for (std::shared_ptr<Cache> base_cache :
+  for (const std::shared_ptr<Cache>& base_cache :
        {NewLRUCache(capacity, num_shard_bits),
         // FixedHyperClockCache
         HyperClockCacheOptions(
@@ -990,7 +990,7 @@ TEST_F(DBBlockCacheTest, CacheEntryRoleStats) {
   int iterations_tested = 0;
   for (bool partition : {false, true}) {
     SCOPED_TRACE("Partition? " + std::to_string(partition));
-    for (std::shared_ptr<Cache> cache :
+    for (const std::shared_ptr<Cache>& cache :
          {NewLRUCache(capacity),
           HyperClockCacheOptions(
               capacity,
@@ -1251,7 +1251,7 @@ void DummyFillCache(Cache& cache, size_t entry_size,
 
 class CountingLogger : public Logger {
  public:
-  ~CountingLogger() override {}
+  ~CountingLogger() override = default;
   using Logger::Logv;
   void Logv(const InfoLogLevel log_level, const char* format,
             va_list /*ap*/) override {
@@ -1373,7 +1373,7 @@ class StableCacheKeyTestFS : public FaultInjectionTestFS {
     SetFailGetUniqueId(true);
   }
 
-  ~StableCacheKeyTestFS() override {}
+  ~StableCacheKeyTestFS() override = default;
 
   IOStatus LinkFile(const std::string&, const std::string&, const IOOptions&,
                     IODebugContext*) override {
@@ -1566,7 +1566,7 @@ class CacheKeyTest : public testing::Test {
     tp_.db_id = std::to_string(db_id_);
     tp_.orig_file_number = file_number;
     bool is_stable;
-    std::string cur_session_id = "";  // ignored
+    std::string cur_session_id;       // ignored
     uint64_t cur_file_number = 42;    // ignored
     OffsetableCacheKey rv;
     BlockBasedTable::SetupBaseCacheKey(&tp_, cur_session_id, cur_file_number,
@@ -1834,6 +1834,7 @@ class DBBlockCachePinningTest
   PinningTier unpartitioned_pinning_;
 };
 
+#ifdef LZ4
 TEST_P(DBBlockCachePinningTest, TwoLevelDB) {
   // Creates one file in L0 and one file in L1. Both files have enough data that
   // their index and filter blocks are partitioned. The L1 file will also have
@@ -1845,10 +1846,7 @@ TEST_P(DBBlockCachePinningTest, TwoLevelDB) {
   const int kNumKeysPerFile = kBlockSize * kNumBlocksPerFile / kKeySize;
 
   Options options = CurrentOptions();
-  // `kNoCompression` makes the unit test more portable. But it relies on the
-  // current behavior of persisting/accessing dictionary even when there's no
-  // (de)compression happening, which seems fairly likely to change over time.
-  options.compression = kNoCompression;
+  options.compression = kLZ4Compression;
   options.compression_opts.max_dict_bytes = 4 << 10;
   options.statistics = ROCKSDB_NAMESPACE::CreateDBStatistics();
   BlockBasedTableOptions table_options;
@@ -1961,6 +1959,7 @@ TEST_P(DBBlockCachePinningTest, TwoLevelDB) {
   ASSERT_EQ(expected_compression_dict_misses,
             TestGetTickerCount(options, BLOCK_CACHE_COMPRESSION_DICT_MISS));
 }
+#endif
 
 INSTANTIATE_TEST_CASE_P(
     DBBlockCachePinningTest, DBBlockCachePinningTest,

db/db_bloom_filter_test.cc

@@ -78,7 +78,7 @@ class DBBloomFilterTestWithParam
   DBBloomFilterTestWithParam()
       : DBTestBase("db_bloom_filter_tests", /*env_do_fsync=*/true) {}
 
-  ~DBBloomFilterTestWithParam() override {}
+  ~DBBloomFilterTestWithParam() override = default;
 
   void SetUp() override {
     bfp_impl_ = std::get<0>(GetParam());
@@ -529,6 +529,7 @@ TEST_P(DBBloomFilterTestWithParam, BloomFilter) {
     BlockBasedTableOptions table_options;
     table_options.no_block_cache = true;
     table_options.filter_policy = Create(10, bfp_impl_);
+    table_options.optimize_filters_for_memory = false;
     table_options.partition_filters = partition_filters_;
     if (partition_filters_) {
       table_options.index_type =
@@ -1695,6 +1696,7 @@ TEST_F(DBBloomFilterTest, ContextCustomFilterPolicy) {
     BlockBasedTableOptions table_options;
     table_options.filter_policy = policy;
     table_options.format_version = 5;
+    table_options.optimize_filters_for_memory = false;
     options.table_factory.reset(NewBlockBasedTableFactory(table_options));
 
     ASSERT_OK(TryReopen(options));
@@ -2051,7 +2053,7 @@ class DBBloomFilterTestVaryPrefixAndFormatVer
   DBBloomFilterTestVaryPrefixAndFormatVer()
       : DBTestBase("db_bloom_filter_tests", /*env_do_fsync=*/true) {}
 
-  ~DBBloomFilterTestVaryPrefixAndFormatVer() override {}
+  ~DBBloomFilterTestVaryPrefixAndFormatVer() override = default;
 
   void SetUp() override {
     use_prefix_ = std::get<0>(GetParam());
@@ -2126,8 +2128,9 @@ TEST_P(DBBloomFilterTestVaryPrefixAndFormatVer, PartitionedMultiGet) {
       values[i] = PinnableSlice();
     }
 
-    db_->MultiGet(ropts, Q, &column_families[0], &key_slices[0], &values[0],
-                  /*timestamps=*/nullptr, &statuses[0], true);
+    db_->MultiGet(ropts, Q, column_families.data(), key_slices.data(),
+                  values.data(),
+                  /*timestamps=*/nullptr, statuses.data(), true);
 
     // Confirm correct status results
     uint32_t number_not_found = 0;
@@ -2177,8 +2180,9 @@ TEST_P(DBBloomFilterTestVaryPrefixAndFormatVer, PartitionedMultiGet) {
       values[i] = PinnableSlice();
     }
 
-    db_->MultiGet(ropts, Q, &column_families[0], &key_slices[0], &values[0],
-                  /*timestamps=*/nullptr, &statuses[0], true);
+    db_->MultiGet(ropts, Q, column_families.data(), key_slices.data(),
+                  values.data(),
+                  /*timestamps=*/nullptr, statuses.data(), true);
 
     // Confirm correct status results
     uint32_t number_not_found = 0;

db/db_compaction_filter_test.cc

@@ -150,7 +150,7 @@ class ConditionalFilter : public CompactionFilter {
 
 class ChangeFilter : public CompactionFilter {
  public:
-  explicit ChangeFilter() {}
+  explicit ChangeFilter() = default;
 
   bool Filter(int /*level*/, const Slice& /*key*/, const Slice& /*value*/,
               std::string* new_value, bool* value_changed) const override {
@@ -289,7 +289,7 @@ class ConditionalFilterFactory : public CompactionFilterFactory {
 
 class ChangeFilterFactory : public CompactionFilterFactory {
  public:
-  explicit ChangeFilterFactory() {}
+  explicit ChangeFilterFactory() = default;
 
   std::unique_ptr<CompactionFilter> CreateCompactionFilter(
       const CompactionFilter::Context& /*context*/) override {
@@ -342,7 +342,7 @@ TEST_F(DBTestCompactionFilter, CompactionFilter) {
   {
     InternalKeyComparator icmp(options.comparator);
     ReadOptions read_options;
-    ScopedArenaIterator iter(dbfull()->NewInternalIterator(
+    ScopedArenaPtr<InternalIterator> iter(dbfull()->NewInternalIterator(
         read_options, &arena, kMaxSequenceNumber, handles_[1]));
     iter->SeekToFirst();
     ASSERT_OK(iter->status());
@@ -434,7 +434,7 @@ TEST_F(DBTestCompactionFilter, CompactionFilter) {
   {
     InternalKeyComparator icmp(options.comparator);
     ReadOptions read_options;
-    ScopedArenaIterator iter(dbfull()->NewInternalIterator(
+    ScopedArenaPtr<InternalIterator> iter(dbfull()->NewInternalIterator(
         read_options, &arena, kMaxSequenceNumber, handles_[1]));
     iter->SeekToFirst();
     ASSERT_OK(iter->status());
@@ -717,8 +717,8 @@ TEST_F(DBTestCompactionFilter, CompactionFilterContextManual) {
     Arena arena;
     InternalKeyComparator icmp(options.comparator);
     ReadOptions read_options;
-    ScopedArenaIterator iter(dbfull()->NewInternalIterator(read_options, &arena,
-                                                           kMaxSequenceNumber));
+    ScopedArenaPtr<InternalIterator> iter(dbfull()->NewInternalIterator(
+        read_options, &arena, kMaxSequenceNumber));
     iter->SeekToFirst();
     ASSERT_OK(iter->status());
     while (iter->Valid()) {

db/db_compaction_test.cc

@@ -16,6 +16,7 @@
 #include "env/mock_env.h"
 #include "port/port.h"
 #include "port/stack_trace.h"
+#include "rocksdb/advanced_options.h"
 #include "rocksdb/concurrent_task_limiter.h"
 #include "rocksdb/experimental.h"
 #include "rocksdb/sst_file_writer.h"
@@ -41,7 +42,7 @@ class CompactionStatsCollector : public EventListener {
     }
   }
 
-  ~CompactionStatsCollector() override {}
+  ~CompactionStatsCollector() override = default;
 
   void OnCompactionCompleted(DB* /* db */,
                              const CompactionJobInfo& info) override {
@@ -241,8 +242,8 @@ class RoundRobinSubcompactionsAgainstResources
 namespace {
 class FlushedFileCollector : public EventListener {
  public:
-  FlushedFileCollector() {}
-  ~FlushedFileCollector() override {}
+  FlushedFileCollector() = default;
+  ~FlushedFileCollector() override = default;
 
   void OnFlushCompleted(DB* /*db*/, const FlushJobInfo& info) override {
     std::lock_guard<std::mutex> lock(mutex_);
@@ -252,7 +253,7 @@ class FlushedFileCollector : public EventListener {
   std::vector<std::string> GetFlushedFiles() {
     std::lock_guard<std::mutex> lock(mutex_);
     std::vector<std::string> result;
-    for (auto fname : flushed_files_) {
+    for (const auto& fname : flushed_files_) {
       result.push_back(fname);
     }
     return result;
@@ -2090,9 +2091,9 @@ TEST_P(DBDeleteFileRangeTest, DeleteFilesInRanges) {
     Slice begin2(begin_str2), end2(end_str2);
     Slice begin3(begin_str3), end3(end_str3);
     std::vector<RangePtr> ranges;
-    ranges.push_back(RangePtr(&begin1, &end1));
-    ranges.push_back(RangePtr(&begin2, &end2));
-    ranges.push_back(RangePtr(&begin3, &end3));
+    ranges.emplace_back(&begin1, &end1);
+    ranges.emplace_back(&begin2, &end2);
+    ranges.emplace_back(&begin3, &end3);
     ASSERT_OK(DeleteFilesInRanges(db_, db_->DefaultColumnFamily(),
                                   ranges.data(), ranges.size()));
     ASSERT_EQ("0,3,7", FilesPerLevel(0));
@@ -2117,9 +2118,9 @@ TEST_P(DBDeleteFileRangeTest, DeleteFilesInRanges) {
     Slice begin2(begin_str2), end2(end_str2);
     Slice begin3(begin_str3), end3(end_str3);
     std::vector<RangePtr> ranges;
-    ranges.push_back(RangePtr(&begin1, &end1));
-    ranges.push_back(RangePtr(&begin2, &end2));
-    ranges.push_back(RangePtr(&begin3, &end3));
+    ranges.emplace_back(&begin1, &end1);
+    ranges.emplace_back(&begin2, &end2);
+    ranges.emplace_back(&begin3, &end3);
     ASSERT_OK(DeleteFilesInRanges(db_, db_->DefaultColumnFamily(),
                                   ranges.data(), ranges.size(), false));
     ASSERT_EQ("0,1,4", FilesPerLevel(0));
@@ -3997,7 +3998,7 @@ TEST_P(DBCompactionTestWithParam, FullCompactionInBottomPriThreadPool) {
   Env::Default()->SetBackgroundThreads(0, Env::Priority::BOTTOM);
 }
 
-TEST_F(DBCompactionTest, CancelCompactionWaitingOnConflict) {
+TEST_F(DBCompactionTest, CancelCompactionWaitingOnRunningConflict) {
   // This test verifies cancellation of a compaction waiting to be scheduled due
   // to conflict with a running compaction.
   //
@@ -4036,7 +4037,7 @@ TEST_F(DBCompactionTest, CancelCompactionWaitingOnConflict) {
   // Make sure the manual compaction has seen the conflict before being canceled
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->LoadDependency(
       {{"ColumnFamilyData::CompactRange:Return",
-        "DBCompactionTest::CancelCompactionWaitingOnConflict:"
+        "DBCompactionTest::CancelCompactionWaitingOnRunningConflict:"
         "PreDisableManualCompaction"}});
   auto manual_compaction_thread = port::Thread([this]() {
     ASSERT_TRUE(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr)
@@ -4047,12 +4048,73 @@ TEST_F(DBCompactionTest, CancelCompactionWaitingOnConflict) {
   // despite finding a conflict with an automatic compaction that is still
   // running
   TEST_SYNC_POINT(
-      "DBCompactionTest::CancelCompactionWaitingOnConflict:"
+      "DBCompactionTest::CancelCompactionWaitingOnRunningConflict:"
       "PreDisableManualCompaction");
   db_->DisableManualCompaction();
   manual_compaction_thread.join();
 }
 
+TEST_F(DBCompactionTest, CancelCompactionWaitingOnScheduledConflict) {
+  // This test verifies cancellation of a compaction waiting to be scheduled due
+  // to conflict with a scheduled (but not running) compaction.
+  //
+  // A `CompactRange()` in universal compacts all files, waiting for files to
+  // become available if they are locked for another compaction. This test
+  // blocks the compaction thread pool and then calls `CompactRange()` twice.
+  // The first call to `CompactRange()` schedules a compaction that is queued
+  // in the thread pool. The second call to `CompactRange()` blocks on the first
+  // call due to the conflict in file picking. The test verifies that
+  // `DisableManualCompaction()` can cancel both while the thread pool remains
+  // blocked.
+  const int kNumSortedRuns = 4;
+
+  Options options = CurrentOptions();
+  options.compaction_style = kCompactionStyleUniversal;
+  options.disable_auto_compactions = true;
+  options.memtable_factory.reset(
+      test::NewSpecialSkipListFactory(KNumKeysByGenerateNewFile - 1));
+  Reopen(options);
+
+  test::SleepingBackgroundTask sleeping_task_low;
+  env_->Schedule(&test::SleepingBackgroundTask::DoSleepTask, &sleeping_task_low,
+                 Env::Priority::LOW);
+
+  // Fill overlapping files in L0
+  Random rnd(301);
+  for (int i = 0; i < kNumSortedRuns; ++i) {
+    int key_idx = 0;
+    GenerateNewFile(&rnd, &key_idx, false /* nowait */);
+  }
+
+  std::atomic<int> num_compact_range_calls{0};
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
+      "ColumnFamilyData::CompactRange:Return",
+      [&](void* /* arg */) { num_compact_range_calls++; });
+
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
+
+  const int kNumManualCompactions = 2;
+  port::Thread manual_compaction_threads[kNumManualCompactions];
+  for (int i = 0; i < kNumManualCompactions; i++) {
+    manual_compaction_threads[i] = port::Thread([this]() {
+      ASSERT_TRUE(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr)
+                      .IsIncomplete());
+    });
+  }
+  while (num_compact_range_calls < kNumManualCompactions) {
+  }
+
+  // Cancel it. Threads should be joinable, i.e., both the scheduled and blocked
+  // manual compactions were canceled despite no compaction could have ever run.
+  db_->DisableManualCompaction();
+  for (int i = 0; i < kNumManualCompactions; i++) {
+    manual_compaction_threads[i].join();
+  }
+
+  sleeping_task_low.WakeUp();
+  sleeping_task_low.WaitUntilDone();
+}
+
 TEST_F(DBCompactionTest, OptimizedDeletionObsoleting) {
   // Deletions can be dropped when compacted to non-last level if they fall
   // outside the lower-level files' key-ranges.
@@ -6082,6 +6144,14 @@ TEST_F(DBCompactionTest, CompactionLimiter) {
         }
       });
 
+  std::vector<std::string> pending_compaction_cfs;
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
+      "SchedulePendingCompaction::cfd", [&](void* arg) {
+        const std::string& cf_name =
+            static_cast<ColumnFamilyData*>(arg)->GetName();
+        pending_compaction_cfs.emplace_back(cf_name);
+      });
+
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
 
   // Block all compact threads in thread pool.
@@ -6135,8 +6205,15 @@ TEST_F(DBCompactionTest, CompactionLimiter) {
               NumTableFilesAtLevel(0, 0));
   }
 
-  // All CFs are pending compaction
-  ASSERT_EQ(cf_count, env_->GetThreadPoolQueueLen(Env::LOW));
+  // Wait until all CFs are pending compaction. WaitForFlushMemtable() can
+  // return before the next compaction is scheduled, so we need to do some
+  // waiting here.
+  unsigned int tp_len = env_->GetThreadPoolQueueLen(Env::LOW);
+  for (int i = 0; i < 10000 && tp_len < cf_count; i++) {
+    env_->SleepForMicroseconds(1000);
+    tp_len = env_->GetThreadPoolQueueLen(Env::LOW);
+  }
+  ASSERT_EQ(cf_count, tp_len);
 
   // Unblock all compaction threads
   for (size_t i = 0; i < kTotalCompactTasks; i++) {
@@ -6641,7 +6718,7 @@ TEST_F(DBCompactionTest, RoundRobinCutOutputAtCompactCursor) {
 
 class NoopMergeOperator : public MergeOperator {
  public:
-  NoopMergeOperator() {}
+  NoopMergeOperator() = default;
 
   bool FullMergeV2(const MergeOperationInput& /*merge_in*/,
                    MergeOperationOutput* merge_out) const override {
@@ -7044,7 +7121,8 @@ class DBCompactionTestWithOngoingFileIngestionParam
       SyncPoint::GetInstance()->SetCallBack(
           "ExternalSstFileIngestionJob::Run", [&](void*) {
             SyncPoint::GetInstance()->LoadDependency(
-                {{"DBImpl::CompactFilesImpl::PostSanitizeCompactionInputFiles",
+                {{"DBImpl::CompactFilesImpl::"
+                  "PostSanitizeAndConvertCompactionInputFiles",
                   "VersionSet::LogAndApply:WriteManifest"}});
           });
     } else {
@@ -7526,6 +7604,72 @@ class DBCompactionTestL0FilesMisorderCorruption : public DBCompactionTest {
   std::shared_ptr<test::SleepingBackgroundTask> sleeping_task_;
 };
 
+TEST_F(DBCompactionTest, CompactFilesSupportKeyPlacementRangeConflict) {
+  Options options;
+  options.create_if_missing = true;
+  options.compaction_style = kCompactionStyleUniversal;
+  options.num_levels = 3;
+  DestroyAndReopen(options);
+
+  // To create LSM of below shape:
+  // L0: [k2]
+  // L1: [k3],[k4]
+  // L2: [k1,  k5]
+  ASSERT_OK(Put("k1", "v"));
+  ASSERT_OK(Put("k5", "v"));
+  ASSERT_OK(Flush());
+  CompactRangeOptions cro;
+  cro.bottommost_level_compaction = BottommostLevelCompaction::kForceOptimized;
+  ASSERT_OK(db_->CompactRange(cro, nullptr, nullptr));
+  ASSERT_EQ("0,0,1", FilesPerLevel());
+
+  ASSERT_OK(Put("k3", "v"));
+  ASSERT_OK(Flush());
+  ASSERT_OK(Put("k4", "v"));
+  ASSERT_OK(Flush());
+  ASSERT_OK(experimental::PromoteL0(db_, db_->DefaultColumnFamily(), 1));
+  ASSERT_EQ("0,2,1", FilesPerLevel());
+
+  ASSERT_OK(Put("k2", "v"));
+  ASSERT_OK(Flush());
+  ASSERT_EQ("1,2,1", FilesPerLevel());
+
+  Close();
+
+  // To force below two CompactFiles() in order to coerce range conflict on L1
+  // upon (2)
+  // (1): Compact [k2] at L0 and [k3] at L1 with output to L1
+  // (2): Compact [k4] at L1 and [k1, k5] at L2 and output to L1 and L2
+  options.preclude_last_level_data_seconds = 1;
+  Reopen(options);
+
+  ColumnFamilyMetaData cf_meta_data;
+  db_->GetColumnFamilyMetaData(&cf_meta_data);
+
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
+      "CompactFilesImpl:0", [&](void* /*arg*/) {
+        std::vector<std::string> c2_input_files;
+        c2_input_files.push_back(cf_meta_data.levels[1].files[1].name);
+        c2_input_files.push_back(cf_meta_data.levels[2].files[0].name);
+        // To verify CompactFiles() is aborted upon range conflict instead
+        // of crashing upon internal assertion
+        Status s = db_->CompactFiles(CompactionOptions(), c2_input_files,
+                                     2 /* output_level */);
+        ASSERT_TRUE(s.IsAborted());
+      });
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
+
+  std::vector<std::string> c1_input_files;
+  c1_input_files.push_back(cf_meta_data.levels[0].files[0].name);
+  c1_input_files.push_back(cf_meta_data.levels[1].files[0].name);
+  Status s = db_->CompactFiles(CompactionOptions(), c1_input_files,
+                               1 /* output_level */);
+  ASSERT_OK(s);
+
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->ClearAllCallBacks();
+}
+
 TEST_F(DBCompactionTestL0FilesMisorderCorruption,
        FlushAfterIntraL0LevelCompactionWithIngestedFile) {
   SetupOptions(CompactionStyle::kCompactionStyleLevel, "");
@@ -9164,66 +9308,104 @@ TEST_F(DBCompactionTest, CompactionWithChecksumHandoffManifest2) {
 }
 
 TEST_F(DBCompactionTest, FIFOChangeTemperature) {
-  Options options = CurrentOptions();
-  options.compaction_style = kCompactionStyleFIFO;
-  options.num_levels = 1;
-  options.max_open_files = -1;
-  options.level0_file_num_compaction_trigger = 2;
-  options.create_if_missing = true;
-  CompactionOptionsFIFO fifo_options;
-  fifo_options.file_temperature_age_thresholds = {{Temperature::kCold, 1000}};
-  fifo_options.max_table_files_size = 100000000;
-  options.compaction_options_fifo = fifo_options;
-  env_->SetMockSleep();
-  Reopen(options);
+  for (bool write_time_default : {false, true}) {
+    SCOPED_TRACE("write time default? " + std::to_string(write_time_default));
 
-  int total_cold = 0;
-  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
-      "NewWritableFile::FileOptions.temperature", [&](void* arg) {
-        Temperature temperature = *(static_cast<Temperature*>(arg));
-        if (temperature == Temperature::kCold) {
-          total_cold++;
-        }
-      });
-  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
+    Options options = CurrentOptions();
+    options.compaction_style = kCompactionStyleFIFO;
+    options.num_levels = 1;
+    options.max_open_files = -1;
+    options.level0_file_num_compaction_trigger = 2;
+    options.create_if_missing = true;
+    CompactionOptionsFIFO fifo_options;
+    fifo_options.file_temperature_age_thresholds = {{Temperature::kCold, 1000}};
+    fifo_options.max_table_files_size = 100000000;
+    options.compaction_options_fifo = fifo_options;
+    env_->SetMockSleep();
+    if (write_time_default) {
+      options.default_write_temperature = Temperature::kWarm;
+    }
+    // Should be ignored (TODO: fail?)
+    options.last_level_temperature = Temperature::kHot;
+    Reopen(options);
 
-  // The file system does not support checksum handoff. The check
-  // will be ignored.
-  ASSERT_OK(Put(Key(0), "value1"));
-  env_->MockSleepForSeconds(800);
-  ASSERT_OK(Put(Key(2), "value2"));
-  ASSERT_OK(Flush());
+    int total_cold = 0;
+    int total_warm = 0;
+    int total_hot = 0;
+    int total_unknown = 0;
+    ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
+        "NewWritableFile::FileOptions.temperature", [&](void* arg) {
+          Temperature temperature = *(static_cast<Temperature*>(arg));
+          if (temperature == Temperature::kCold) {
+            total_cold++;
+          } else if (temperature == Temperature::kWarm) {
+            total_warm++;
+          } else if (temperature == Temperature::kHot) {
+            total_hot++;
+          } else {
+            assert(temperature == Temperature::kUnknown);
+            total_unknown++;
+          }
+        });
+    ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
 
-  ASSERT_OK(Put(Key(0), "value1"));
-  env_->MockSleepForSeconds(800);
-  ASSERT_OK(Put(Key(2), "value2"));
-  ASSERT_OK(Flush());
+    // The file system does not support checksum handoff. The check
+    // will be ignored.
+    ASSERT_OK(Put(Key(0), "value1"));
+    env_->MockSleepForSeconds(800);
+    ASSERT_OK(Put(Key(2), "value2"));
+    ASSERT_OK(Flush());
 
-  ASSERT_OK(Put(Key(0), "value1"));
-  env_->MockSleepForSeconds(800);
-  ASSERT_OK(Put(Key(2), "value2"));
-  ASSERT_OK(Flush());
-  ASSERT_OK(dbfull()->TEST_WaitForCompact());
+    ASSERT_OK(Put(Key(0), "value1"));
+    env_->MockSleepForSeconds(800);
+    ASSERT_OK(Put(Key(2), "value2"));
+    ASSERT_OK(Flush());
 
-  ASSERT_OK(Put(Key(0), "value1"));
-  env_->MockSleepForSeconds(800);
-  ASSERT_OK(Put(Key(2), "value2"));
-  ASSERT_OK(Flush());
+    ASSERT_OK(Put(Key(0), "value1"));
+    env_->MockSleepForSeconds(800);
+    ASSERT_OK(Put(Key(2), "value2"));
+    ASSERT_OK(Flush());
+    ASSERT_OK(dbfull()->TEST_WaitForCompact());
 
-  ASSERT_OK(dbfull()->TEST_WaitForCompact());
+    if (write_time_default) {
+      // Also test dynamic option change
+      ASSERT_OK(db_->SetOptions({{"default_write_temperature", "kHot"}}));
+    }
 
-  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
+    ASSERT_OK(Put(Key(0), "value1"));
+    env_->MockSleepForSeconds(800);
+    ASSERT_OK(Put(Key(2), "value2"));
+    ASSERT_OK(Flush());
 
-  ColumnFamilyMetaData metadata;
-  db_->GetColumnFamilyMetaData(&metadata);
-  ASSERT_EQ(4, metadata.file_count);
-  ASSERT_EQ(Temperature::kUnknown, metadata.levels[0].files[0].temperature);
-  ASSERT_EQ(Temperature::kUnknown, metadata.levels[0].files[1].temperature);
-  ASSERT_EQ(Temperature::kCold, metadata.levels[0].files[2].temperature);
-  ASSERT_EQ(Temperature::kCold, metadata.levels[0].files[3].temperature);
-  ASSERT_EQ(2, total_cold);
+    ASSERT_OK(dbfull()->TEST_WaitForCompact());
 
-  Destroy(options);
+    ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
+
+    ColumnFamilyMetaData metadata;
+    db_->GetColumnFamilyMetaData(&metadata);
+    ASSERT_EQ(4, metadata.file_count);
+    if (write_time_default) {
+      ASSERT_EQ(Temperature::kHot, metadata.levels[0].files[0].temperature);
+      ASSERT_EQ(Temperature::kWarm, metadata.levels[0].files[1].temperature);
+      // Includes obsolete/deleted files moved to cold
+      ASSERT_EQ(total_warm, 3);
+      ASSERT_EQ(total_hot, 1);
+      // Includes non-SST DB files
+      ASSERT_GT(total_unknown, 0);
+    } else {
+      ASSERT_EQ(Temperature::kUnknown, metadata.levels[0].files[0].temperature);
+      ASSERT_EQ(Temperature::kUnknown, metadata.levels[0].files[1].temperature);
+      ASSERT_EQ(total_warm, 0);
+      ASSERT_EQ(total_hot, 0);
+      // Includes non-SST DB files
+      ASSERT_GT(total_unknown, 4);
+    }
+    ASSERT_EQ(Temperature::kCold, metadata.levels[0].files[2].temperature);
+    ASSERT_EQ(Temperature::kCold, metadata.levels[0].files[3].temperature);
+    ASSERT_EQ(2, total_cold);
+
+    Destroy(options);
+  }
 }
 
 TEST_F(DBCompactionTest, DisableMultiManualCompaction) {
@@ -9840,7 +10022,7 @@ TEST_F(DBCompactionTest, TurnOnLevelCompactionDynamicLevelBytesUCToLC) {
   options.compaction_style = CompactionStyle::kCompactionStyleLevel;
   options.level_compaction_dynamic_level_bytes = true;
   ReopenWithColumnFamilies({"default", "pikachu"}, options);
-  std::string expected_lsm = "";
+  std::string expected_lsm;
   for (int i = 0; i < 49; ++i) {
     expected_lsm += "0,";
   }
@@ -9852,6 +10034,62 @@ TEST_F(DBCompactionTest, TurnOnLevelCompactionDynamicLevelBytesUCToLC) {
   ASSERT_EQ(expected_lsm, FilesPerLevel(1));
 }
 
+TEST_F(DBCompactionTest, DisallowRefitFilesFromNonL0ToL02) {
+  Options options = CurrentOptions();
+  options.compaction_style = CompactionStyle::kCompactionStyleLevel;
+  options.num_levels = 3;
+  DestroyAndReopen(options);
+
+  // To set up LSM shape:
+  // L0
+  // L1
+  // L2:[a@1, k@3], [k@2, z@4] (sorted by ascending smallest key)
+  // Both of these 2 files have epoch number = 1
+  const Snapshot* s1 = db_->GetSnapshot();
+  ASSERT_OK(Put("a", "@1"));
+  ASSERT_OK(Put("k", "@2"));
+  const Snapshot* s2 = db_->GetSnapshot();
+  ASSERT_OK(Put("k", "@3"));
+  ASSERT_OK(Put("z", "v3"));
+  ASSERT_OK(Flush());
+  // Cut file between k@3 and k@2
+  SyncPoint::GetInstance()->SetCallBack(
+      "CompactionOutputs::ShouldStopBefore::manual_decision",
+      [options](void* p) {
+        auto* pair = (std::pair<bool*, const Slice>*)p;
+        if ((options.comparator->Compare(ExtractUserKey(pair->second), "k") ==
+             0) &&
+            (GetInternalKeySeqno(pair->second) == 2)) {
+          *(pair->first) = true;
+        }
+      });
+  SyncPoint::GetInstance()->EnableProcessing();
+  CompactRangeOptions cro;
+  cro.bottommost_level_compaction = BottommostLevelCompaction::kForceOptimized;
+  cro.change_level = true;
+  cro.target_level = 2;
+  Status s = dbfull()->CompactRange(cro, nullptr, nullptr);
+  ASSERT_OK(s);
+  ASSERT_EQ("0,0,2", FilesPerLevel());
+  std::vector<LiveFileMetaData> files;
+  dbfull()->GetLiveFilesMetaData(&files);
+  ASSERT_EQ(files.size(), 2);
+  ASSERT_EQ(files[0].smallestkey, "a");
+  ASSERT_EQ(files[0].largestkey, "k");
+  ASSERT_EQ(files[1].smallestkey, "k");
+  ASSERT_EQ(files[1].largestkey, "z");
+
+  // Disallow moving 2 non-L0 files to L0
+  CompactRangeOptions cro2;
+  cro2.change_level = true;
+  cro2.target_level = 0;
+  s = dbfull()->CompactRange(cro2, nullptr, nullptr);
+  ASSERT_TRUE(s.IsAborted());
+
+  db_->ReleaseSnapshot(s1);
+  db_->ReleaseSnapshot(s2);
+}
+
 TEST_F(DBCompactionTest, DrainUnnecessaryLevelsAfterMultiplierChanged) {
   // When the level size multiplier increases such that fewer levels become
   // necessary, unnecessary levels should to be drained.
@@ -10356,20 +10594,20 @@ TEST_F(DBCompactionTest, ReleaseCompactionDuringManifestWrite) {
   SyncPoint::GetInstance()->EnableProcessing();
 
   std::vector<std::thread> threads;
-  threads.emplace_back(std::thread([&]() {
+  threads.emplace_back([&]() {
     std::string k1_str = Key(1);
     std::string k2_str = Key(2);
     Slice k1 = k1_str;
     Slice k2 = k2_str;
     ASSERT_OK(db_->CompactRange(CompactRangeOptions(), &k1, &k2));
-  }));
-  threads.emplace_back(std::thread([&]() {
+  });
+  threads.emplace_back([&]() {
     std::string k10_str = Key(10);
     std::string k11_str = Key(11);
     Slice k10 = k10_str;
     Slice k11 = k11_str;
     ASSERT_OK(db_->CompactRange(CompactRangeOptions(), &k10, &k11));
-  }));
+  });
   std::string k100_str = Key(100);
   std::string k101_str = Key(101);
   Slice k100 = k100_str;

db/db_filesnapshot.cc

@@ -151,6 +151,29 @@ Status DBImpl::GetSortedWalFiles(VectorLogPtr& files) {
     }
   }
 
+  if (s.ok()) {
+    size_t wal_size = files.size();
+    ROCKS_LOG_INFO(immutable_db_options_.info_log,
+                   "Number of log files %" ROCKSDB_PRIszt " (%" ROCKSDB_PRIszt
+                   " required by manifest)",
+                   wal_size, required_by_manifest.size());
+#ifndef NDEBUG
+    std::ostringstream wal_names;
+    for (const auto& wal : files) {
+      wal_names << wal->PathName() << " ";
+    }
+
+    std::ostringstream wal_required_by_manifest_names;
+    for (const auto& wal : required_by_manifest) {
+      wal_required_by_manifest_names << wal << ".log ";
+    }
+
+    ROCKS_LOG_INFO(immutable_db_options_.info_log,
+                   "Log files : %s .Log files required by manifest: %s.",
+                   wal_names.str().c_str(),
+                   wal_required_by_manifest_names.str().c_str());
+#endif  // NDEBUG
+  }
   return s;
 }
 
@@ -206,12 +229,18 @@ Status DBImpl::GetLiveFilesStorageInfo(
   // metadata.
   mutex_.Lock();
   if (flush_memtable) {
-    Status status = FlushForGetLiveFiles();
-    if (!status.ok()) {
-      mutex_.Unlock();
-      ROCKS_LOG_ERROR(immutable_db_options_.info_log, "Cannot Flush data %s\n",
-                      status.ToString().c_str());
-      return status;
+    bool wal_locked = lock_wal_count_ > 0;
+    if (wal_locked) {
+      ROCKS_LOG_INFO(immutable_db_options_.info_log,
+                     "Can't FlushForGetLiveFiles while WAL is locked");
+    } else {
+      Status status = FlushForGetLiveFiles();
+      if (!status.ok()) {
+        mutex_.Unlock();
+        ROCKS_LOG_ERROR(immutable_db_options_.info_log,
+                        "Cannot Flush data %s\n", status.ToString().c_str());
+        return status;
+      }
     }
   }
 
@@ -371,10 +400,6 @@ Status DBImpl::GetLiveFilesStorageInfo(
   }
 
   size_t wal_size = live_wal_files.size();
-
-  ROCKS_LOG_INFO(immutable_db_options_.info_log,
-                 "Number of log files %" ROCKSDB_PRIszt, live_wal_files.size());
-
   // Link WAL files. Copy exact size of last one because it is the only one
   // that has changes after the last flush.
   auto wal_dir = immutable_db_options_.GetWalDir();
@@ -390,8 +415,11 @@ Status DBImpl::GetLiveFilesStorageInfo(
       info.file_number = live_wal_files[i]->LogNumber();
       info.file_type = kWalFile;
       info.size = live_wal_files[i]->SizeFileBytes();
-      // Only last should need to be trimmed
-      info.trim_to_size = (i + 1 == wal_size);
+      // Trim the log either if its the last one, or log file recycling is
+      // enabled. In the latter case, a hard link doesn't prevent the file
+      // from being renamed and recycled. So we need to copy it instead.
+      info.trim_to_size = (i + 1 == wal_size) ||
+                          (immutable_db_options_.recycle_log_file_num > 0);
       if (opts.include_checksum_info) {
         info.file_checksum_func_name = kUnknownFileChecksumFuncName;
         info.file_checksum = kUnknownFileChecksum;

db/db_flush_test.cc

@@ -1367,14 +1367,15 @@ TEST_F(DBFlushTest, MemPurgeDeleteAndDeleteRange) {
       ASSERT_OK(iter->status());
       key = (iter->key()).ToString(false);
       value = (iter->value()).ToString(false);
-      if (key.compare(KEY3) == 0)
+      if (key.compare(KEY3) == 0) {
         ASSERT_EQ(value, p_v3b);
-      else if (key.compare(KEY4) == 0)
+      } else if (key.compare(KEY4) == 0) {
         ASSERT_EQ(value, p_v4);
-      else if (key.compare(KEY5) == 0)
+      } else if (key.compare(KEY5) == 0) {
         ASSERT_EQ(value, p_v5);
-      else
+      } else {
         ASSERT_EQ(value, NOT_FOUND);
+      }
       count++;
     }
     ASSERT_OK(iter->status());
@@ -1404,22 +1405,25 @@ TEST_F(DBFlushTest, MemPurgeDeleteAndDeleteRange) {
     ASSERT_OK(iter->status());
     key = (iter->key()).ToString(false);
     value = (iter->value()).ToString(false);
-    if (key.compare(KEY2) == 0)
+    if (key.compare(KEY2) == 0) {
       ASSERT_EQ(value, p_v2);
-    else if (key.compare(KEY3) == 0)
+    } else if (key.compare(KEY3) == 0) {
       ASSERT_EQ(value, p_v3b);
-    else if (key.compare(KEY4) == 0)
+    } else if (key.compare(KEY4) == 0) {
       ASSERT_EQ(value, p_v4);
-    else if (key.compare(KEY5) == 0)
+    } else if (key.compare(KEY5) == 0) {
       ASSERT_EQ(value, p_v5);
-    else
+    } else {
       ASSERT_EQ(value, NOT_FOUND);
+    }
     count++;
   }
 
   // Expected count here is 4: KEY2, KEY3, KEY4, KEY5.
   ASSERT_EQ(count, EXPECTED_COUNT_END);
-  if (iter) delete iter;
+  if (iter) {
+    delete iter;
+  }
 
   Close();
 }
@@ -2499,7 +2503,7 @@ TEST_F(DBFlushTest, TombstoneVisibleInSnapshot) {
   class SimpleTestFlushListener : public EventListener {
    public:
     explicit SimpleTestFlushListener(DBFlushTest* _test) : test_(_test) {}
-    ~SimpleTestFlushListener() override {}
+    ~SimpleTestFlushListener() override = default;
 
     void OnFlushBegin(DB* db, const FlushJobInfo& info) override {
       ASSERT_EQ(static_cast<uint32_t>(0), info.cf_id);

db/db_follower_test.cc

@@ -0,0 +1,527 @@
+//  Copyright (c) 2024-present, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+
+#include "db/db_test_util.h"
+#include "port/stack_trace.h"
+#include "test_util/sync_point.h"
+
+namespace ROCKSDB_NAMESPACE {
+
+#ifdef OS_LINUX
+
+class DBFollowerTest : public DBTestBase {
+ public:
+  // Create directories for leader and follower
+  // Create the leader DB object
+  DBFollowerTest() : DBTestBase("/db_follower_test", /*env_do_fsync*/ false) {
+    follower_name_ = dbname_ + "/follower";
+    db_parent_ = dbname_;
+    Close();
+    Destroy(CurrentOptions());
+    EXPECT_EQ(env_->CreateDirIfMissing(dbname_), Status::OK());
+    dbname_ = dbname_ + "/leader";
+    Reopen(CurrentOptions());
+  }
+
+  ~DBFollowerTest() {
+    follower_.reset();
+    EXPECT_EQ(DestroyDB(follower_name_, CurrentOptions()), Status::OK());
+    Destroy(CurrentOptions());
+    dbname_ = db_parent_;
+  }
+
+ protected:
+  class DBFollowerTestFS : public FileSystemWrapper {
+   public:
+    explicit DBFollowerTestFS(const std::shared_ptr<FileSystem>& target)
+        : FileSystemWrapper(target),
+          cv_(&mutex_),
+          barrier_(false),
+          count_(0),
+          reinit_count_(0) {}
+
+    const char* Name() const override { return "DBFollowerTestFS"; }
+
+    IOStatus NewSequentialFile(const std::string& fname,
+                               const FileOptions& file_opts,
+                               std::unique_ptr<FSSequentialFile>* result,
+                               IODebugContext* dbg = nullptr) override {
+      class DBFollowerTestSeqFile : public FSSequentialFileWrapper {
+       public:
+        DBFollowerTestSeqFile(DBFollowerTestFS* fs,
+                              std::unique_ptr<FSSequentialFile>&& file,
+                              uint64_t /*size*/)
+            : FSSequentialFileWrapper(file.get()),
+              fs_(fs),
+              file_(std::move(file)) {}
+
+        IOStatus Read(size_t n, const IOOptions& options, Slice* result,
+                      char* scratch, IODebugContext* dbg) override {
+          fs_->BarrierWait();
+          return target()->Read(n, options, result, scratch, dbg);
+        }
+
+       private:
+        DBFollowerTestFS* fs_;
+        std::unique_ptr<FSSequentialFile> file_;
+      };
+
+      std::unique_ptr<FSSequentialFile> file;
+      IOStatus s = target()->NewSequentialFile(fname, file_opts, &file, dbg);
+
+      if (s.ok() && test::GetFileType(fname) == kDescriptorFile) {
+        uint64_t size = 0;
+        EXPECT_EQ(target()->GetFileSize(fname, IOOptions(), &size, nullptr),
+                  IOStatus::OK());
+        result->reset(new DBFollowerTestSeqFile(this, std::move(file), size));
+      } else {
+        *result = std::move(file);
+      }
+      return s;
+    }
+
+    void BarrierInit(int count) {
+      MutexLock l(&mutex_);
+      barrier_ = true;
+      count_ = count;
+    }
+
+    void BarrierWait() {
+      MutexLock l(&mutex_);
+      if (!barrier_) {
+        return;
+      }
+      if (--count_ == 0) {
+        if (reinit_count_ > 0) {
+          count_ = reinit_count_;
+          reinit_count_ = 0;
+        } else {
+          barrier_ = false;
+        }
+        cv_.SignalAll();
+      } else {
+        cv_.Wait();
+      }
+    }
+
+    void BarrierWaitAndReinit(int count) {
+      MutexLock l(&mutex_);
+      if (!barrier_) {
+        return;
+      }
+      reinit_count_ = count;
+      if (--count_ == 0) {
+        if (reinit_count_ > 0) {
+          count_ = reinit_count_;
+          reinit_count_ = 0;
+        } else {
+          barrier_ = false;
+        }
+        cv_.SignalAll();
+      } else {
+        cv_.Wait();
+      }
+    }
+
+   private:
+    port::Mutex mutex_;
+    port::CondVar cv_;
+    bool barrier_;
+    int count_;
+    int reinit_count_;
+  };
+
+  class DBFollowerTestSstPartitioner : public SstPartitioner {
+   public:
+    explicit DBFollowerTestSstPartitioner(uint64_t max_keys)
+        : max_keys_(max_keys), num_keys_(0) {}
+
+    const char* Name() const override { return "DBFollowerTestSstPartitioner"; }
+
+    PartitionerResult ShouldPartition(
+        const PartitionerRequest& /*request*/) override {
+      if (++num_keys_ > max_keys_) {
+        num_keys_ = 0;
+        return PartitionerResult::kRequired;
+      } else {
+        return PartitionerResult::kNotRequired;
+      }
+    }
+
+    bool CanDoTrivialMove(const Slice& /*smallest_user_key*/,
+                          const Slice& /*largest_user_key*/) override {
+      return true;
+    }
+
+   private:
+    uint64_t max_keys_;
+    uint64_t num_keys_;
+  };
+
+  class DBFollowerTestSstPartitionerFactory : public SstPartitionerFactory {
+   public:
+    explicit DBFollowerTestSstPartitionerFactory(uint64_t max_keys)
+        : max_keys_(max_keys) {}
+
+    std::unique_ptr<SstPartitioner> CreatePartitioner(
+        const SstPartitioner::Context& /*context*/) const override {
+      std::unique_ptr<SstPartitioner> partitioner;
+      partitioner.reset(new DBFollowerTestSstPartitioner(max_keys_));
+      return partitioner;
+    }
+
+    const char* Name() const override {
+      return "DBFollowerTestSstPartitionerFactory";
+    }
+
+   private:
+    uint64_t max_keys_;
+  };
+
+  Status OpenAsFollower() {
+    Options opts = CurrentOptions();
+    if (!follower_env_) {
+      follower_env_ = NewCompositeEnv(
+          std::make_shared<DBFollowerTestFS>(env_->GetFileSystem()));
+    }
+    opts.env = follower_env_.get();
+    opts.follower_refresh_catchup_period_ms = 100;
+    return DB::OpenAsFollower(opts, follower_name_, dbname_, &follower_);
+  }
+
+  std::string FollowerGet(const std::string& k) {
+    ReadOptions options;
+    options.verify_checksums = true;
+    std::string result;
+    Status s = follower()->Get(options, k, &result);
+    if (s.IsNotFound()) {
+      result = "NOT_FOUND";
+    } else if (!s.ok()) {
+      result = s.ToString();
+    }
+    return result;
+  }
+
+  DB* follower() { return follower_.get(); }
+  DBFollowerTestFS* follower_fs() {
+    return static_cast<DBFollowerTestFS*>(follower_env_->GetFileSystem().get());
+  }
+
+  void CheckDirs() {
+    std::vector<std::string> db_children;
+    std::vector<std::string> follower_children;
+    EXPECT_OK(env_->GetChildren(dbname_, &db_children));
+    EXPECT_OK(env_->GetChildren(follower_name_, &follower_children));
+
+    std::set<uint64_t> db_filenums;
+    std::set<uint64_t> follower_filenums;
+    for (auto& name : db_children) {
+      if (test::GetFileType(name) != kTableFile) {
+        continue;
+      }
+      db_filenums.insert(test::GetFileNumber(name));
+    }
+    for (auto& name : follower_children) {
+      if (test::GetFileType(name) != kTableFile) {
+        continue;
+      }
+      follower_filenums.insert(test::GetFileNumber(name));
+    }
+    db_filenums.merge(follower_filenums);
+    EXPECT_EQ(follower_filenums.size(), db_filenums.size());
+  }
+
+ private:
+  std::string follower_name_;
+  std::string db_parent_;
+  std::unique_ptr<Env> follower_env_;
+  std::unique_ptr<DB> follower_;
+};
+
+TEST_F(DBFollowerTest, Basic) {
+  ASSERT_OK(Put("k1", "v1"));
+  ASSERT_OK(Flush());
+  ASSERT_OK(Put("k2", "v2"));
+  ASSERT_OK(Flush());
+
+  ASSERT_OK(OpenAsFollower());
+  std::string val;
+  ASSERT_OK(follower()->Get(ReadOptions(), "k1", &val));
+  ASSERT_EQ(val, "v1");
+  CheckDirs();
+}
+
+TEST_F(DBFollowerTest, Flush) {
+  SyncPoint::GetInstance()->LoadDependency({
+      {"DBImplFollower::TryCatchupWithLeader:Begin1", "Leader::Start"},
+      {"Leader::Done", "DBImplFollower::TryCatchupWithLeader:Begin2"},
+      {"DBImplFollower::TryCatchupWithLeader:End", "Follower::WaitForCatchup"},
+  });
+  SyncPoint::GetInstance()->EnableProcessing();
+
+  ASSERT_OK(OpenAsFollower());
+  TEST_SYNC_POINT("Leader::Start");
+  ASSERT_OK(Put("k1", "v1"));
+  ASSERT_OK(Flush());
+  TEST_SYNC_POINT("Leader::Done");
+
+  TEST_SYNC_POINT("Follower::WaitForCatchup");
+  std::string val;
+  ASSERT_OK(follower()->Get(ReadOptions(), "k1", &val));
+  ASSERT_EQ(val, "v1");
+  CheckDirs();
+
+  SyncPoint::GetInstance()->DisableProcessing();
+}
+
+// This test creates 4 L0 files, immediately followed by a compaction to L1.
+// The follower replays the 4 flush records from the MANIFEST unsuccessfully,
+// and then successfully recovers a Version from the compaction record
+TEST_F(DBFollowerTest, RetryCatchup) {
+  Options opts = CurrentOptions();
+  opts.disable_auto_compactions = true;
+  Reopen(opts);
+
+  ASSERT_OK(OpenAsFollower());
+  SyncPoint::GetInstance()->LoadDependency({
+      {"DBImplFollower::TryCatchupWithLeader:Begin1", "Leader::Start"},
+      {"DBImpl::BackgroundCompaction:Start",
+       "DBImplFollower::TryCatchupWithLeader:Begin2"},
+      {"VersionEditHandlerPointInTime::MaybeCreateVersion:Begin1",
+       "DBImpl::BackgroundCompaction:BeforeCompaction"},
+      {"DBImpl::BackgroundCallCompaction:PurgedObsoleteFiles",
+       "VersionEditHandlerPointInTime::MaybeCreateVersion:Begin2"},
+      {"DBImplFollower::TryCatchupWithLeader:End", "Follower::WaitForCatchup"},
+  });
+  SyncPoint::GetInstance()->EnableProcessing();
+  TEST_SYNC_POINT("Leader::Start");
+  ASSERT_OK(Put("k1", "v1"));
+  ASSERT_OK(Flush());
+  ASSERT_OK(Put("k1", "v2"));
+  ASSERT_OK(Flush());
+  ASSERT_OK(Put("k1", "v3"));
+  ASSERT_OK(Flush());
+  ASSERT_OK(Put("k1", "v4"));
+  ASSERT_OK(Flush());
+
+  ASSERT_OK(dbfull()->TEST_CompactRange(0, nullptr, nullptr, nullptr, true));
+
+  TEST_SYNC_POINT("Follower::WaitForCatchup");
+  ASSERT_EQ(FollowerGet("k1"), "v4");
+  CheckDirs();
+
+  SyncPoint::GetInstance()->ClearAllCallBacks();
+  SyncPoint::GetInstance()->DisableProcessing();
+}
+
+// This test validates the same as the previous test, except there is a
+// MANIFEST rollover between the flushes and compaction. The follower
+// does not switch to a new MANIFEST in ReadAndApply. So it would require
+// another round of refresh before catching up.
+TEST_F(DBFollowerTest, RetryCatchupManifestRollover) {
+  Options opts = CurrentOptions();
+  opts.disable_auto_compactions = true;
+  Reopen(opts);
+
+  ASSERT_OK(Put("k1", "v1"));
+  ASSERT_OK(Flush());
+  ASSERT_OK(Put("k1", "v2"));
+  ASSERT_OK(Flush());
+  ASSERT_OK(Put("k1", "v3"));
+  ASSERT_OK(Flush());
+  ASSERT_OK(OpenAsFollower());
+  SyncPoint::GetInstance()->LoadDependency({
+      {"DBImplFollower::TryCatchupWithLeader:Begin1", "Leader::Start"},
+      {"Leader::Flushed", "DBImplFollower::TryCatchupWithLeader:Begin2"},
+      {"VersionEditHandlerPointInTime::MaybeCreateVersion:Begin1",
+       "Leader::Done"},
+      {"DBImpl::BackgroundCallCompaction:PurgedObsoleteFiles",
+       "VersionEditHandlerPointInTime::MaybeCreateVersion:Begin2"},
+      {"DBImplFollower::TryCatchupWithLeader:End",
+       "Follower::WaitForCatchup:1"},
+  });
+  SyncPoint::GetInstance()->EnableProcessing();
+  TEST_SYNC_POINT("Leader::Start");
+  ASSERT_OK(Put("k1", "v4"));
+  ASSERT_OK(Flush());
+
+  TEST_SYNC_POINT("Leader::Flushed");
+  TEST_SYNC_POINT("Leader::Done");
+  Reopen(opts);
+  ASSERT_OK(dbfull()->TEST_CompactRange(0, nullptr, nullptr, nullptr, true));
+
+  TEST_SYNC_POINT("Follower::WaitForCatchup:1");
+  SyncPoint::GetInstance()->LoadDependency({
+      {"DBImplFollower::TryCatchupWithLeader:End",
+       "Follower::WaitForCatchup:2"},
+  });
+  TEST_SYNC_POINT("Follower::WaitForCatchup:2");
+  ASSERT_EQ(FollowerGet("k1"), "v4");
+
+  SyncPoint::GetInstance()->ClearAllCallBacks();
+  SyncPoint::GetInstance()->DisableProcessing();
+}
+
+// This test creates 4 L0 files and compacts them. The follower, during catchup,
+// successfully instantiates 4 Versions corresponding to the 4 files (but
+// donesn't install them yet), followed by deleting those 4 and adding a new
+// file from compaction. The test verifies that the 4 L0 files are deleted
+// correctly by the follower.
+// We use teh Barrier* functions to ensure that the follower first sees the 4
+// L0 files and is able to link them, and then sees the compaction that
+// obsoletes those L0 files (so those L0 files are intermediates that it has
+// to explicitly delete). Suppose we don't have any barriers, its possible
+// the follower reads the L0 records and compaction records from the MANIFEST
+// in one read, which means those L0 files would have already been deleted
+// by the leader and the follower cannot link to them.
+TEST_F(DBFollowerTest, IntermediateObsoleteFiles) {
+  Options opts = CurrentOptions();
+  opts.disable_auto_compactions = true;
+  Reopen(opts);
+  ASSERT_OK(OpenAsFollower());
+
+  follower_fs()->BarrierInit(2);
+  ASSERT_OK(Put("k1", "v1"));
+  ASSERT_OK(Flush());
+  ASSERT_OK(Put("k1", "v2"));
+  ASSERT_OK(Flush());
+  ASSERT_OK(Put("k1", "v3"));
+  ASSERT_OK(Flush());
+  ASSERT_OK(Put("k1", "v4"));
+  ASSERT_OK(Flush());
+  follower_fs()->BarrierWaitAndReinit(2);
+
+  ASSERT_OK(dbfull()->TEST_CompactRange(0, nullptr, nullptr, nullptr, true));
+  follower_fs()->BarrierWait();
+
+  SyncPoint::GetInstance()->LoadDependency({
+      {"DBImplFollower::TryCatchupWithLeader:End",
+       "Follower::WaitForCatchup:1"},
+  });
+  SyncPoint::GetInstance()->EnableProcessing();
+  TEST_SYNC_POINT("Follower::WaitForCatchup:1");
+  CheckDirs();
+  ASSERT_EQ(FollowerGet("k1"), "v4");
+}
+
+// This test verifies a scenario where the follower can recover a Version
+// partially (i.e some of the additions cannot be found), and the files
+// that are found are obsoleted by a subsequent VersionEdit.
+TEST_F(DBFollowerTest, PartialVersionRecovery) {
+  Options opts = CurrentOptions();
+  opts.disable_auto_compactions = true;
+  opts.sst_partitioner_factory =
+      std::make_shared<DBFollowerTestSstPartitionerFactory>(1);
+  Reopen(opts);
+
+  ASSERT_OK(Put("k1", "v1"));
+  ASSERT_OK(Put("k2", "v1"));
+  ASSERT_OK(Put("k3", "v1"));
+  ASSERT_OK(Flush());
+  MoveFilesToLevel(2);
+
+  ASSERT_OK(Put("k1", "v2"));
+  ASSERT_OK(Flush());
+  ASSERT_OK(Put("k3", "v2"));
+  ASSERT_OK(Flush());
+  MoveFilesToLevel(1);
+
+  ASSERT_OK(OpenAsFollower());
+  ASSERT_OK(dbfull()->SetOptions(dbfull()->DefaultColumnFamily(),
+                                 {{"max_compaction_bytes", "1"}}));
+
+  follower_fs()->BarrierInit(2);
+  Slice key("k1");
+  ASSERT_OK(dbfull()->TEST_CompactRange(1, &key, &key, nullptr, true));
+
+  follower_fs()->BarrierWaitAndReinit(2);
+
+  // The second compaction input overlaps the previous compaction outputs
+  // by one file. This file is never added to  VersionStorageInfo since it
+  // was added and deleted before the catch up completes. We later verify that
+  // the follower correctly deleted this file.
+  key = Slice("k3");
+  ASSERT_OK(dbfull()->TEST_CompactRange(1, &key, &key, nullptr, true));
+  follower_fs()->BarrierWait();
+
+  SyncPoint::GetInstance()->LoadDependency({
+      {"DBImplFollower::TryCatchupWithLeader:End",
+       "Follower::WaitForCatchup:1"},
+  });
+  SyncPoint::GetInstance()->EnableProcessing();
+  TEST_SYNC_POINT("Follower::WaitForCatchup:1");
+  CheckDirs();
+  ASSERT_EQ(FollowerGet("k1"), "v2");
+  ASSERT_EQ(FollowerGet("k2"), "v1");
+  ASSERT_EQ(FollowerGet("k3"), "v2");
+  SyncPoint::GetInstance()->DisableProcessing();
+}
+
+// This test verifies a scenario similar to the PartialVersionRecovery, except
+// with a MANIFEST rollover in between. When there is a rollover, the
+// follower's attempt ends without installing a new Version. The next catch up
+// attempt will recover a full Version.
+TEST_F(DBFollowerTest, PartialVersionRecoveryWithRollover) {
+  Options opts = CurrentOptions();
+  opts.disable_auto_compactions = true;
+  opts.sst_partitioner_factory =
+      std::make_shared<DBFollowerTestSstPartitionerFactory>(1);
+  Reopen(opts);
+
+  ASSERT_OK(Put("k1", "v1"));
+  ASSERT_OK(Put("k2", "v1"));
+  ASSERT_OK(Put("k3", "v1"));
+  ASSERT_OK(Flush());
+  MoveFilesToLevel(2);
+
+  ASSERT_OK(Put("k1", "v2"));
+  ASSERT_OK(Flush());
+  ASSERT_OK(Put("k3", "v2"));
+  ASSERT_OK(Flush());
+  MoveFilesToLevel(1);
+
+  opts.max_compaction_bytes = 1;
+  Reopen(opts);
+
+  ASSERT_OK(OpenAsFollower());
+
+  follower_fs()->BarrierInit(2);
+  Slice key("k1");
+  ASSERT_OK(dbfull()->TEST_CompactRange(1, &key, &key, nullptr, true));
+
+  follower_fs()->BarrierWaitAndReinit(2);
+  Reopen(opts);
+  key = Slice("k3");
+  ASSERT_OK(dbfull()->TEST_CompactRange(1, &key, &key, nullptr, true));
+  follower_fs()->BarrierWait();
+
+  SyncPoint::GetInstance()->LoadDependency({
+      {"DBImplFollower::TryCatchupWithLeader:Begin1",
+       "Follower::WaitForCatchup:1"},
+      {"Follower::WaitForCatchup:2",
+       "DBImplFollower::TryCatchupWithLeader:Begin2"},
+  });
+  SyncPoint::GetInstance()->EnableProcessing();
+  TEST_SYNC_POINT("Follower::WaitForCatchup:1");
+  TEST_SYNC_POINT("Follower::WaitForCatchup:2");
+  SyncPoint::GetInstance()->LoadDependency({
+      {"DBImplFollower::TryCatchupWithLeader:End",
+       "Follower::WaitForCatchup:3"},
+  });
+  TEST_SYNC_POINT("Follower::WaitForCatchup:3");
+  CheckDirs();
+  ASSERT_EQ(FollowerGet("k1"), "v2");
+  ASSERT_EQ(FollowerGet("k2"), "v1");
+  ASSERT_EQ(FollowerGet("k3"), "v2");
+  SyncPoint::GetInstance()->DisableProcessing();
+}
+#endif
+}  // namespace ROCKSDB_NAMESPACE
+
+int main(int argc, char** argv) {
+  ROCKSDB_NAMESPACE::port::InstallStackTraceHandler();
+  ::testing::InitGoogleTest(&argc, argv);
+  return RUN_ALL_TESTS();
+}

db/db_impl/compacted_db_impl.cc

@@ -21,7 +21,7 @@ CompactedDBImpl::CompactedDBImpl(const DBOptions& options,
       version_(nullptr),
       user_comparator_(nullptr) {}
 
-CompactedDBImpl::~CompactedDBImpl() {}
+CompactedDBImpl::~CompactedDBImpl() = default;
 
 size_t CompactedDBImpl::FindFile(const Slice& key) {
   size_t right = files_.num_files - 1;

db/db_impl/db_impl.h

@@ -50,6 +50,7 @@
 #include "monitoring/instrumented_mutex.h"
 #include "options/db_options.h"
 #include "port/port.h"
+#include "rocksdb/attribute_groups.h"
 #include "rocksdb/db.h"
 #include "rocksdb/env.h"
 #include "rocksdb/memtablerep.h"
@@ -59,7 +60,6 @@
 #include "rocksdb/utilities/replayer.h"
 #include "rocksdb/write_buffer_manager.h"
 #include "table/merging_iterator.h"
-#include "table/scoped_arena_iterator.h"
 #include "util/autovector.h"
 #include "util/hash.h"
 #include "util/repeatable_thread.h"
@@ -291,6 +291,11 @@ class DBImpl : public DB {
                       const Slice* keys,
                       PinnableAttributeGroups* results) override;
 
+  void MultiGetEntityWithCallback(
+      const ReadOptions& read_options, ColumnFamilyHandle* column_family,
+      ReadCallback* callback,
+      autovector<KeyContext*, MultiGetContext::MAX_BATCH_SIZE>* sorted_keys);
+
   Status CreateColumnFamily(const ColumnFamilyOptions& cf_options,
                             const std::string& column_family,
                             ColumnFamilyHandle** handle) override {
@@ -351,6 +356,17 @@ class DBImpl : public DB {
 
   const Snapshot* GetSnapshot() override;
   void ReleaseSnapshot(const Snapshot* snapshot) override;
+
+  // EXPERIMENTAL
+  std::unique_ptr<Iterator> NewCoalescingIterator(
+      const ReadOptions& options,
+      const std::vector<ColumnFamilyHandle*>& column_families) override;
+
+  // EXPERIMENTAL
+  std::unique_ptr<AttributeGroupIterator> NewAttributeGroupIterator(
+      const ReadOptions& options,
+      const std::vector<ColumnFamilyHandle*>& column_families) override;
+
   // Create a timestamped snapshot. This snapshot can be shared by multiple
   // readers. If any of them uses it for write conflict checking, then
   // is_write_conflict_boundary is true. For simplicity, set it to true by
@@ -646,8 +662,8 @@ class DBImpl : public DB {
 
   // If `snapshot` == kMaxSequenceNumber, set a recent one inside the file.
   ArenaWrappedDBIter* NewIteratorImpl(const ReadOptions& options,
-                                      ColumnFamilyData* cfd, SuperVersion* sv,
-                                      SequenceNumber snapshot,
+                                      ColumnFamilyHandleImpl* cfh,
+                                      SuperVersion* sv, SequenceNumber snapshot,
                                       ReadCallback* read_callback,
                                       bool expose_blob_index = false,
                                       bool allow_refresh = true);
@@ -810,6 +826,8 @@ class DBImpl : public DB {
 
   uint64_t MinLogNumberToKeep();
 
+  uint64_t MinLogNumberToRecycle();
+
   // Returns the lower bound file number for SSTs that won't be deleted, even if
   // they're obsolete. This lower bound is used internally to prevent newly
   // created flush/compaction output files from being deleted before they're
@@ -1046,7 +1064,8 @@ class DBImpl : public DB {
   static Status Open(const DBOptions& db_options, const std::string& name,
                      const std::vector<ColumnFamilyDescriptor>& column_families,
                      std::vector<ColumnFamilyHandle*>* handles, DB** dbptr,
-                     const bool seq_per_batch, const bool batch_per_txn);
+                     const bool seq_per_batch, const bool batch_per_txn,
+                     const bool is_retry, bool* can_retry);
 
   static IOStatus CreateAndNewDirectory(
       FileSystem* fs, const std::string& dirname,
@@ -1192,6 +1211,8 @@ class DBImpl : public DB {
     return logs_.back().number;
   }
 
+  void TEST_DeleteObsoleteFiles();
+
   const std::unordered_set<uint64_t>& TEST_GetFilesGrabbedForPurge() const {
     return files_grabbed_for_purge_;
   }
@@ -1219,6 +1240,22 @@ class DBImpl : public DB {
   // populate_historical_seconds, now].
   void RecordSeqnoToTimeMapping(uint64_t populate_historical_seconds);
 
+  // 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);
+
   // Interface to block and signal the DB in case of stalling writes by
   // WriteBufferManager. Each DBImpl object contains ptr to WBMStallInterface.
   // When DB needs to be blocked or signalled by WriteBufferManager,
@@ -1391,9 +1428,8 @@ class DBImpl : public DB {
     autovector<ColumnFamilyData*> cfds_;
     autovector<const MutableCFOptions*> mutable_cf_opts_;
     autovector<autovector<VersionEdit*>> edit_lists_;
-    // Stale SST files to delete found upon recovery. This stores a mapping from
-    // such a file's absolute path to its parent directory.
-    std::unordered_map<std::string, std::string> files_to_delete_;
+    // All existing data files (SST files and Blob files) found during DB::Open.
+    std::vector<std::string> existing_data_files_;
     bool is_new_db_ = false;
   };
 
@@ -1507,7 +1543,7 @@ class DBImpl : public DB {
   Status WriteRecoverableState();
 
   // Actual implementation of Close()
-  Status CloseImpl();
+  virtual Status CloseImpl();
 
   // Recover the descriptor from persistent storage.  May do a significant
   // amount of work to recover recently logged updates.  Any changes to
@@ -1519,9 +1555,9 @@ class DBImpl : public DB {
   virtual Status Recover(
       const std::vector<ColumnFamilyDescriptor>& column_families,
       bool read_only = false, bool error_if_wal_file_exists = false,
-      bool error_if_data_exists_in_wals = false,
+      bool error_if_data_exists_in_wals = false, bool is_retry = false,
       uint64_t* recovered_seq = nullptr,
-      RecoveryContext* recovery_ctx = nullptr);
+      RecoveryContext* recovery_ctx = nullptr, bool* can_retry = nullptr);
 
   virtual bool OwnTablesAndLogs() const { return true; }
 
@@ -1531,22 +1567,36 @@ class DBImpl : public DB {
   // Assign db_id_ and write DB ID to manifest if necessary.
   void SetDBId(std::string&& id, bool read_only, RecoveryContext* recovery_ctx);
 
+  // Collect a deduplicated collection of paths used by this DB, including
+  // dbname_, DBOptions.db_paths, ColumnFamilyOptions.cf_paths.
+  std::set<std::string> CollectAllDBPaths();
+
   // REQUIRES: db mutex held when calling this function, but the db mutex can
   // be released and re-acquired. Db mutex will be held when the function
   // returns.
-  // After recovery, there may be SST files in db/cf paths that are
-  // not referenced in the MANIFEST (e.g.
+  // It stores all existing data files (SST and Blob) in RecoveryContext. In
+  // the meantime, we find out the largest file number present in the paths, and
+  // bump up the version set's next_file_number_ to be 1 + largest_file_number.
+  // recovery_ctx stores the context about version edits. All those edits are
+  // persisted to new Manifest after successfully syncing the new WAL.
+  Status MaybeUpdateNextFileNumber(RecoveryContext* recovery_ctx);
+
+  // Track existing data files, including both referenced and unreferenced SST
+  // and Blob files in SstFileManager. This is only called during DB::Open and
+  // it's called before any file deletion start so that their deletion can be
+  // properly rate limited.
+  // Files may not be referenced in the MANIFEST because (e.g.
   // 1. It's best effort recovery;
   // 2. The VersionEdits referencing the SST files are appended to
   // RecoveryContext, DB crashes when syncing the MANIFEST, the VersionEdits are
   // still not synced to MANIFEST during recovery.)
-  // It stores the SST files to be deleted in RecoveryContext. In the
-  // meantime, we find out the largest file number present in the paths, and
-  // bump up the version set's next_file_number_ to be 1 + largest_file_number.
-  // recovery_ctx stores the context about version edits and files to be
-  // deleted. All those edits are persisted to new Manifest after successfully
-  // syncing the new WAL.
-  Status DeleteUnreferencedSstFiles(RecoveryContext* recovery_ctx);
+  //
+  // If the file is referenced in Manifest (typically that's the
+  // vast majority of all files), since it already has the file size
+  // on record, we don't need to query the file system. Otherwise, we query the
+  // file system for the size of an unreferenced file.
+  void TrackExistingDataFiles(
+      const std::vector<std::string>& existing_data_files);
 
   // SetDbSessionId() should be called in the constuctor DBImpl()
   // to ensure that db_session_id_ gets updated every time the DB is opened
@@ -1600,6 +1650,7 @@ class DBImpl : public DB {
   friend class ForwardIterator;
   friend struct SuperVersion;
   friend class CompactedDBImpl;
+  friend class DBImplFollower;
 #ifndef NDEBUG
   friend class DBTest_ConcurrentFlushWAL_Test;
   friend class DBTest_MixedSlowdownOptionsStop_Test;
@@ -1901,7 +1952,7 @@ class DBImpl : public DB {
   // corrupted_log_found is set to true if we recover from a corrupted log file.
   Status RecoverLogFiles(const std::vector<uint64_t>& log_numbers,
                          SequenceNumber* next_sequence, bool read_only,
-                         bool* corrupted_log_found,
+                         bool is_retry, bool* corrupted_log_found,
                          RecoveryContext* recovery_ctx);
 
   // The following two methods are used to flush a memtable to
@@ -2005,17 +2056,22 @@ class DBImpl : public DB {
       mutex_.Lock();
     }
 
-    if (!immutable_db_options_.unordered_write) {
-      // Then the writes are finished before the next write group starts
-      return;
+    if (immutable_db_options_.unordered_write) {
+      // Wait for the ones who already wrote to the WAL to finish their
+      // memtable write.
+      if (pending_memtable_writes_.load() != 0) {
+        // XXX: suspicious wait while holding DB mutex?
+        std::unique_lock<std::mutex> guard(switch_mutex_);
+        switch_cv_.wait(guard,
+                        [&] { return pending_memtable_writes_.load() == 0; });
+      }
+    } else {
+      // (Writes are finished before the next write group starts.)
     }
 
-    // Wait for the ones who already wrote to the WAL to finish their
-    // memtable write.
-    if (pending_memtable_writes_.load() != 0) {
-      std::unique_lock<std::mutex> guard(switch_mutex_);
-      switch_cv_.wait(guard,
-                      [&] { return pending_memtable_writes_.load() == 0; });
+    // Wait for any LockWAL to clear
+    while (lock_wal_count_ > 0) {
+      bg_cv_.Wait();
     }
   }
 
@@ -2309,10 +2365,7 @@ class DBImpl : public DB {
   // A structure to hold the information required to process MultiGet of keys
   // belonging to one column family. For a multi column family MultiGet, there
   // will be a container of these objects.
-  struct MultiGetColumnFamilyData {
-    ColumnFamilyHandle* cf;
-    ColumnFamilyData* cfd;
-
+  struct MultiGetKeyRangePerCf {
     // For the batched MultiGet which relies on sorted keys, start specifies
     // the index of first key belonging to this column family in the sorted
     // list.
@@ -2322,31 +2375,33 @@ class DBImpl : public DB {
     // belonging to this column family in the sorted list
     size_t num_keys;
 
+    MultiGetKeyRangePerCf() : start(0), num_keys(0) {}
+
+    MultiGetKeyRangePerCf(size_t first, size_t count)
+        : start(first), num_keys(count) {}
+  };
+
+  // A structure to contain ColumnFamilyData and the SuperVersion obtained for
+  // the consistent view of DB
+  struct ColumnFamilySuperVersionPair {
+    ColumnFamilyHandleImpl* cfh;
+    ColumnFamilyData* cfd;
+
     // SuperVersion for the column family obtained in a manner that ensures a
     // consistent view across all column families in the DB
     SuperVersion* super_version;
-    MultiGetColumnFamilyData(ColumnFamilyHandle* column_family,
-                             SuperVersion* sv)
-        : cf(column_family),
-          cfd(static_cast<ColumnFamilyHandleImpl*>(cf)->cfd()),
-          start(0),
-          num_keys(0),
+    ColumnFamilySuperVersionPair(ColumnFamilyHandle* column_family,
+                                 SuperVersion* sv)
+        : cfh(static_cast<ColumnFamilyHandleImpl*>(column_family)),
+          cfd(cfh->cfd()),
           super_version(sv) {}
 
-    MultiGetColumnFamilyData(ColumnFamilyHandle* column_family, size_t first,
-                             size_t count, SuperVersion* sv)
-        : cf(column_family),
-          cfd(static_cast<ColumnFamilyHandleImpl*>(cf)->cfd()),
-          start(first),
-          num_keys(count),
-          super_version(sv) {}
-
-    MultiGetColumnFamilyData() = default;
+    ColumnFamilySuperVersionPair() = default;
   };
 
   // A common function to obtain a consistent snapshot, which can be implicit
   // if the user doesn't specify a snapshot in read_options, across
-  // multiple column families for MultiGet. It will attempt to get an implicit
+  // multiple column families. It will attempt to get an implicit
   // snapshot without acquiring the db_mutes, but will give up after a few
   // tries and acquire the mutex if a memtable flush happens. The template
   // allows both the batched and non-batched MultiGet to call this with
@@ -2355,18 +2410,26 @@ class DBImpl : public DB {
   // If callback is non-null, the callback is refreshed with the snapshot
   // sequence number
   //
+  // `extra_sv_ref` is used to indicate whether thread-local SuperVersion
+  // should be obtained with an extra ref (by GetReferencedSuperVersion()) or
+  // not (by GetAndRefSuperVersion()). For instance, point lookup like MultiGet
+  // does not require SuperVersion to be re-acquired throughout the entire
+  // invocation (no need extra ref), while MultiCfIterators may need the
+  // SuperVersion to be updated during Refresh() (requires extra ref).
+  //
   // `sv_from_thread_local` being set to false indicates that the SuperVersion
   // obtained from the ColumnFamilyData, whereas true indicates they are thread
   // local.
+  //
   // A non-OK status will be returned if for a column family that enables
   // user-defined timestamp feature, the specified `ReadOptions.timestamp`
   // attemps to read collapsed history.
-  template <class T>
-  Status MultiCFSnapshot(
-      const ReadOptions& read_options, ReadCallback* callback,
-      std::function<MultiGetColumnFamilyData*(typename T::iterator&)>&
-          iter_deref_func,
-      T* cf_list, SequenceNumber* snapshot, bool* sv_from_thread_local);
+  template <class T, typename IterDerefFuncType>
+  Status MultiCFSnapshot(const ReadOptions& read_options,
+                         ReadCallback* callback,
+                         IterDerefFuncType iter_deref_func, T* cf_list,
+                         bool extra_sv_ref, SequenceNumber* snapshot,
+                         bool* sv_from_thread_local);
 
   // The actual implementation of the batching MultiGet. The caller is expected
   // to have acquired the SuperVersion and pass in a snapshot sequence number
@@ -2389,6 +2452,13 @@ class DBImpl : public DB {
 
   bool ShouldReferenceSuperVersion(const MergeContext& merge_context);
 
+  template <typename IterType, typename ImplType,
+            typename ErrorIteratorFuncType>
+  std::unique_ptr<IterType> NewMultiCfIterator(
+      const ReadOptions& _read_options,
+      const std::vector<ColumnFamilyHandle*>& column_families,
+      ErrorIteratorFuncType error_iterator_func);
+
   // Lock over the persistent DB state.  Non-nullptr iff successfully acquired.
   FileLock* db_lock_;
 
@@ -2440,6 +2510,11 @@ class DBImpl : public DB {
   uint64_t logfile_number_;
   // Log files that we can recycle. Must be protected by db mutex_.
   std::deque<uint64_t> log_recycle_files_;
+  // The minimum log file number taht can be recycled, if log recycling is
+  // enabled. This is used to ensure that log files created by previous
+  // instances of the database are not recycled, as we cannot be sure they
+  // were created in the recyclable format.
+  uint64_t min_log_number_to_recycle_;
   // Protected by log_write_mutex_.
   bool log_dir_synced_;
   // Without two_write_queues, read and writes to log_empty_ are protected by
@@ -2925,9 +3000,9 @@ inline Status DBImpl::FailIfReadCollapsedHistory(const ColumnFamilyData* cfd,
   if (!full_history_ts_low.empty() &&
       ucmp->CompareTimestamp(ts, full_history_ts_low) < 0) {
     std::stringstream oss;
-    oss << "Read timestamp: " << ts.ToString(true)
+    oss << "Read timestamp: " << ucmp->TimestampToString(ts)
         << " is smaller than full_history_ts_low: "
-        << Slice(full_history_ts_low).ToString(true) << std::endl;
+        << ucmp->TimestampToString(full_history_ts_low) << std::endl;
     return Status::InvalidArgument(oss.str());
   }
   return Status::OK();

db/db_impl/db_impl_compaction_flush.cc

@@ -163,8 +163,10 @@ IOStatus DBImpl::SyncClosedLogs(const WriteOptions& write_options,
         if (error_recovery_in_prog) {
           log->file()->reset_seen_error();
         }
-        // TODO: plumb Env::IOActivity, Env::IOPriority
-        io_s = log->Close(WriteOptions());
+        // Normally the log file is closed when purging obsolete file, but if
+        // log recycling is enabled, the log file is closed here so that it
+        // can be reused.
+        io_s = log->Close(write_options);
         if (!io_s.ok()) {
           break;
         }
@@ -222,9 +224,15 @@ Status DBImpl::FlushMemTableToOutputFile(
   // the host crashes after flushing and before WAL is persistent, the
   // flushed SST may contain data from write batches whose updates to
   // other (unflushed) column families are missing.
+  //
+  // When 2PC is enabled, non-recent WAL(s) may be needed for crash-recovery,
+  // even when there is only one CF in the DB, for prepared transactions that
+  // had not been committed yet. Make sure we sync them to keep the persisted
+  // WAL state at least as new as the persisted SST state.
   const bool needs_to_sync_closed_wals =
       logfile_number_ > 0 &&
-      versions_->GetColumnFamilySet()->NumberOfColumnFamilies() > 1;
+      (versions_->GetColumnFamilySet()->NumberOfColumnFamilies() > 1 ||
+       allow_2pc());
 
   // If needs_to_sync_closed_wals is true, we need to record the current
   // maximum memtable ID of this column family so that a later PickMemtables()
@@ -270,8 +278,8 @@ Status DBImpl::FlushMemTableToOutputFile(
       GetCompressionFlush(*cfd->ioptions(), mutable_cf_options), stats_,
       &event_logger_, mutable_cf_options.report_bg_io_stats,
       true /* sync_output_directory */, true /* write_manifest */, thread_pri,
-      io_tracer_, seqno_to_time_mapping_, db_id_, db_session_id_,
-      cfd->GetFullHistoryTsLow(), &blob_callback_);
+      io_tracer_, cfd->GetSuperVersion()->ShareSeqnoToTimeMapping(), db_id_,
+      db_session_id_, cfd->GetFullHistoryTsLow(), &blob_callback_);
   FileMetaData file_meta;
 
   Status s;
@@ -545,8 +553,9 @@ Status DBImpl::AtomicFlushMemTablesToOutputFiles(
         GetCompressionFlush(*cfd->ioptions(), mutable_cf_options), stats_,
         &event_logger_, mutable_cf_options.report_bg_io_stats,
         false /* sync_output_directory */, false /* write_manifest */,
-        thread_pri, io_tracer_, seqno_to_time_mapping_, db_id_, db_session_id_,
-        cfd->GetFullHistoryTsLow(), &blob_callback_));
+        thread_pri, io_tracer_,
+        cfd->GetSuperVersion()->ShareSeqnoToTimeMapping(), db_id_,
+        db_session_id_, cfd->GetFullHistoryTsLow(), &blob_callback_));
   }
 
   std::vector<FileMetaData> file_meta(num_cfs);
@@ -970,7 +979,7 @@ void DBImpl::NotifyOnFlushBegin(ColumnFamilyData* cfd, FileMetaData* file_meta,
     info.smallest_seqno = file_meta->fd.smallest_seqno;
     info.largest_seqno = file_meta->fd.largest_seqno;
     info.flush_reason = flush_reason;
-    for (auto listener : immutable_db_options_.listeners) {
+    for (const auto& listener : immutable_db_options_.listeners) {
       listener->OnFlushBegin(this, info);
     }
   }
@@ -1002,7 +1011,7 @@ void DBImpl::NotifyOnFlushCompleted(
     for (auto& info : *flush_jobs_info) {
       info->triggered_writes_slowdown = triggered_writes_slowdown;
       info->triggered_writes_stop = triggered_writes_stop;
-      for (auto listener : immutable_db_options_.listeners) {
+      for (const auto& listener : immutable_db_options_.listeners) {
         listener->OnFlushCompleted(this, *info);
       }
       TEST_SYNC_POINT(
@@ -1085,7 +1094,12 @@ Status DBImpl::IncreaseFullHistoryTsLowImpl(ColumnFamilyData* cfd,
   assert(ucmp->timestamp_size() == ts_low.size() && !ts_low.empty());
   if (!current_ts_low.empty() &&
       ucmp->CompareTimestamp(ts_low, current_ts_low) < 0) {
-    return Status::InvalidArgument("Cannot decrease full_history_ts_low");
+    std::stringstream oss;
+    oss << "Current full_history_ts_low: "
+        << ucmp->TimestampToString(current_ts_low)
+        << " is higher than provided ts: " << ucmp->TimestampToString(ts_low)
+        << std::endl;
+    return Status::InvalidArgument(oss.str());
   }
 
   Status s = versions_->LogAndApply(cfd, *cfd->GetLatestMutableCFOptions(),
@@ -1117,6 +1131,11 @@ Status DBImpl::CompactRangeInternal(const CompactRangeOptions& options,
   if (options.target_path_id >= cfd->ioptions()->cf_paths.size()) {
     return Status::InvalidArgument("Invalid target path ID");
   }
+  if (options.change_level &&
+      cfd->ioptions()->compaction_style == kCompactionStyleFIFO) {
+    return Status::NotSupported(
+        "FIFO compaction does not support change_level.");
+  }
 
   bool flush_needed = true;
 
@@ -1177,6 +1196,16 @@ Status DBImpl::CompactRangeInternal(const CompactRangeOptions& options,
                             final_output_level, options, begin, end, exclusive,
                             false /* disable_trivial_move */,
                             std::numeric_limits<uint64_t>::max(), trim_ts);
+  } else if (cfd->ioptions()->compaction_style == kCompactionStyleFIFO) {
+    // FIFOCompactionPicker::CompactRange() will ignore the input key range
+    // [begin, end] and just try to pick compaction based on the configured
+    // option `compaction_options_fifo`. So we skip checking if [begin, end]
+    // overlaps with the DB here.
+    final_output_level = 0;
+    s = RunManualCompaction(cfd, /*input_level=*/0, final_output_level, options,
+                            begin, end, exclusive,
+                            false /* disable_trivial_move */,
+                            std::numeric_limits<uint64_t>::max(), trim_ts);
   } else {
     int first_overlapped_level = kInvalidLevel;
     {
@@ -1261,8 +1290,7 @@ Status DBImpl::CompactRangeInternal(const CompactRangeOptions& options,
       CleanupSuperVersion(super_version);
     }
     if (s.ok() && first_overlapped_level != kInvalidLevel) {
-      if (cfd->ioptions()->compaction_style == kCompactionStyleUniversal ||
-          cfd->ioptions()->compaction_style == kCompactionStyleFIFO) {
+      if (cfd->ioptions()->compaction_style == kCompactionStyleUniversal) {
         assert(first_overlapped_level == 0);
         s = RunManualCompaction(
             cfd, first_overlapped_level, first_overlapped_level, options, begin,
@@ -1417,6 +1445,14 @@ Status DBImpl::CompactFiles(const CompactionOptions& compact_options,
   LogBuffer log_buffer(InfoLogLevel::INFO_LEVEL,
                        immutable_db_options_.info_log.get());
 
+  if (compact_options.compression !=
+      CompressionType::kDisableCompressionOption) {
+    ROCKS_LOG_WARN(immutable_db_options_.info_log,
+                   "[%s] [JOB %d] Found use of deprecated option "
+                   "`CompactionOptions::compression`",
+                   cfd->GetName().c_str(), job_context.job_id);
+  }
+
   // Perform CompactFiles
   TEST_SYNC_POINT("TestCompactFiles::IngestExternalFile2");
   TEST_SYNC_POINT_CALLBACK("TestCompactFiles:PausingManualCompaction:3",
@@ -1506,16 +1542,11 @@ Status DBImpl::CompactFilesImpl(
         std::to_string(cfd->ioptions()->num_levels - 1));
   }
 
-  Status s = cfd->compaction_picker()->SanitizeCompactionInputFiles(
-      &input_set, cf_meta, output_level);
-  TEST_SYNC_POINT("DBImpl::CompactFilesImpl::PostSanitizeCompactionInputFiles");
-  if (!s.ok()) {
-    return s;
-  }
-
   std::vector<CompactionInputFiles> input_files;
-  s = cfd->compaction_picker()->GetCompactionInputsFromFileNumbers(
-      &input_files, &input_set, version->storage_info(), compact_options);
+  Status s = cfd->compaction_picker()->SanitizeAndConvertCompactionInputFiles(
+      &input_set, cf_meta, output_level, version->storage_info(), &input_files);
+  TEST_SYNC_POINT(
+      "DBImpl::CompactFilesImpl::PostSanitizeAndConvertCompactionInputFiles");
   if (!s.ok()) {
     return s;
   }
@@ -1609,9 +1640,9 @@ Status DBImpl::CompactFilesImpl(
   }
   if (status.ok()) {
     assert(compaction_job.io_status().ok());
-    InstallSuperVersionAndScheduleWork(c->column_family_data(),
-                                       &job_context->superversion_contexts[0],
-                                       *c->mutable_cf_options());
+    InstallSuperVersionAndScheduleWork(
+        c->column_family_data(), job_context->superversion_contexts.data(),
+        *c->mutable_cf_options());
   }
   // status above captures any error during compaction_job.Install, so its ok
   // not check compaction_job.io_status() explicitly if we're not calling
@@ -1731,7 +1762,7 @@ void DBImpl::NotifyOnCompactionBegin(ColumnFamilyData* cfd, Compaction* c,
   {
     CompactionJobInfo info{};
     BuildCompactionJobInfo(cfd, c, st, job_stats, job_id, &info);
-    for (auto listener : immutable_db_options_.listeners) {
+    for (const auto& listener : immutable_db_options_.listeners) {
       listener->OnCompactionBegin(this, info);
     }
     info.status.PermitUncheckedError();
@@ -1760,7 +1791,7 @@ void DBImpl::NotifyOnCompactionCompleted(
   {
     CompactionJobInfo info{};
     BuildCompactionJobInfo(cfd, c, st, compaction_job_stats, job_id, &info);
-    for (auto listener : immutable_db_options_.listeners) {
+    for (const auto& listener : immutable_db_options_.listeners) {
       listener->OnCompactionCompleted(this, info);
     }
   }
@@ -1771,6 +1802,8 @@ void DBImpl::NotifyOnCompactionCompleted(
 
 // REQUIREMENT: block all background work by calling PauseBackgroundWork()
 // before calling this function
+// TODO (hx235): Replace Status::NotSupported() with Status::Aborted() for
+// better semantics like CompactFiles()
 Status DBImpl::ReFitLevel(ColumnFamilyData* cfd, int level, int target_level) {
   assert(level < cfd->NumberLevels());
   if (target_level >= cfd->NumberLevels()) {
@@ -1806,6 +1839,8 @@ Status DBImpl::ReFitLevel(ColumnFamilyData* cfd, int level, int target_level) {
   if (to_level != level) {
     std::vector<CompactionInputFiles> input(1);
     input[0].level = level;
+    // TODO (hx235): Only refit the output files in the current manual
+    // compaction instead of all the files in the output level
     for (auto& f : vstorage->LevelFiles(level)) {
       input[0].files.push_back(f);
     }
@@ -1837,6 +1872,12 @@ Status DBImpl::ReFitLevel(ColumnFamilyData* cfd, int level, int target_level) {
       }
     } else {
       // to_level < level
+      if (to_level == 0 && input[0].files.size() > 1) {
+        refitting_level_ = false;
+        return Status::Aborted(
+            "Moving more than 1 file from non-L0 to L0 is not allowed as it "
+            "does not bring any benefit to read nor write throughput.");
+      }
       // Check levels are empty for a trivial move
       for (int l = to_level; l < level; l++) {
         if (vstorage->NumLevelFiles(l) > 0) {
@@ -1868,7 +1909,8 @@ Status DBImpl::ReFitLevel(ColumnFamilyData* cfd, int level, int target_level) {
         ,
         LLONG_MAX /* max compaction bytes, not applicable */,
         0 /* output path ID, not applicable */, mutable_cf_options.compression,
-        mutable_cf_options.compression_opts, Temperature::kUnknown,
+        mutable_cf_options.compression_opts,
+        mutable_cf_options.default_write_temperature,
         0 /* max_subcompactions, not applicable */,
         {} /* grandparents, not applicable */, false /* is manual */,
         "" /* trim_ts */, -1 /* score, not applicable */,
@@ -2150,16 +2192,6 @@ Status DBImpl::RunManualCompaction(
                manual.begin, manual.end, &manual.manual_end, &manual_conflict,
                max_file_num_to_ignore, trim_ts)) == nullptr &&
           manual_conflict))) {
-      if (!scheduled) {
-        // There is a conflicting compaction
-        if (manual_compaction_paused_ > 0 || manual.canceled == true) {
-          // Stop waiting since it was canceled. Pretend the error came from
-          // compaction so the below cleanup/error handling code can process it.
-          manual.done = true;
-          manual.status =
-              Status::Incomplete(Status::SubCode::kManualCompactionPaused);
-        }
-      }
       if (!manual.done) {
         bg_cv_.Wait();
       }
@@ -2234,6 +2266,17 @@ Status DBImpl::RunManualCompaction(
         *final_output_level = compaction->output_level();
       }
     }
+    if (!scheduled) {
+      // There is nothing scheduled to wait on, so any cancellation can end the
+      // manual now.
+      if (manual_compaction_paused_ > 0 || manual.canceled == true) {
+        // Stop waiting since it was canceled. Pretend the error came from
+        // compaction so the below cleanup/error handling code can process it.
+        manual.done = true;
+        manual.status =
+            Status::Incomplete(Status::SubCode::kManualCompactionPaused);
+      }
+    }
   }
 
   log_buffer.FlushBufferToLog();
@@ -3028,6 +3071,8 @@ void DBImpl::SchedulePendingCompaction(ColumnFamilyData* cfd) {
     return;
   }
   if (!cfd->queued_for_compaction() && cfd->NeedsCompaction()) {
+    TEST_SYNC_POINT_CALLBACK("SchedulePendingCompaction::cfd",
+                             static_cast<void*>(cfd));
     AddToCompactionQueue(cfd);
     ++unscheduled_compactions_;
   }
@@ -3220,7 +3265,7 @@ Status DBImpl::BackgroundFlush(bool* made_progress, JobContext* job_context,
         column_families_not_to_flush.push_back(cfd);
         continue;
       }
-      superversion_contexts.emplace_back(SuperVersionContext(true));
+      superversion_contexts.emplace_back(true);
       bg_flush_args.emplace_back(cfd, max_memtable_id,
                                  &(superversion_contexts.back()), flush_reason);
     }
@@ -3315,7 +3360,7 @@ void DBImpl::BackgroundCallFlush(Env::Priority thread_pri) {
       bg_cv_.SignalAll();  // In case a waiter can proceed despite the error
       mutex_.Unlock();
       ROCKS_LOG_ERROR(immutable_db_options_.info_log,
-                      "[JOB %d] Waiting after background flush error: %s"
+                      "[JOB %d] Waiting after background flush error: %s, "
                       "Accumulated background error counts: %" PRIu64,
                       job_context.job_id, s.ToString().c_str(), error_cnt);
       log_buffer.FlushBufferToLog();
@@ -3725,9 +3770,9 @@ Status DBImpl::BackgroundCompaction(bool* made_progress,
           compaction_released = true;
         });
     io_s = versions_->io_status();
-    InstallSuperVersionAndScheduleWork(c->column_family_data(),
-                                       &job_context->superversion_contexts[0],
-                                       *c->mutable_cf_options());
+    InstallSuperVersionAndScheduleWork(
+        c->column_family_data(), job_context->superversion_contexts.data(),
+        *c->mutable_cf_options());
     ROCKS_LOG_BUFFER(log_buffer, "[%s] Deleted %d files\n",
                      c->column_family_data()->GetName().c_str(),
                      c->num_input_files(0));
@@ -3800,9 +3845,9 @@ Status DBImpl::BackgroundCompaction(bool* made_progress,
         });
     io_s = versions_->io_status();
     // Use latest MutableCFOptions
-    InstallSuperVersionAndScheduleWork(c->column_family_data(),
-                                       &job_context->superversion_contexts[0],
-                                       *c->mutable_cf_options());
+    InstallSuperVersionAndScheduleWork(
+        c->column_family_data(), job_context->superversion_contexts.data(),
+        *c->mutable_cf_options());
 
     VersionStorageInfo::LevelSummaryStorage tmp;
     c->column_family_data()->internal_stats()->IncBytesMoved(c->output_level(),
@@ -3895,9 +3940,9 @@ Status DBImpl::BackgroundCompaction(bool* made_progress,
         compaction_job.Install(*c->mutable_cf_options(), &compaction_released);
     io_s = compaction_job.io_status();
     if (status.ok()) {
-      InstallSuperVersionAndScheduleWork(c->column_family_data(),
-                                         &job_context->superversion_contexts[0],
-                                         *c->mutable_cf_options());
+      InstallSuperVersionAndScheduleWork(
+          c->column_family_data(), job_context->superversion_contexts.data(),
+          *c->mutable_cf_options());
     }
     *made_progress = true;
     TEST_SYNC_POINT_CALLBACK("DBImpl::BackgroundCompaction:AfterCompaction",
@@ -4044,7 +4089,6 @@ void DBImpl::RemoveManualCompaction(DBImpl::ManualCompactionState* m) {
     ++it;
   }
   assert(false);
-  return;
 }
 
 bool DBImpl::ShouldntRunManualCompaction(ManualCompactionState* m) {
@@ -4341,6 +4385,7 @@ Status DBImpl::WaitForCompact(
     }
     if ((bg_bottom_compaction_scheduled_ || bg_compaction_scheduled_ ||
          bg_flush_scheduled_ || unscheduled_compactions_ ||
+         (wait_for_compact_options.wait_for_purge && bg_purge_scheduled_) ||
          unscheduled_flushes_ || error_handler_.IsRecoveryInProgress()) &&
         (error_handler_.GetBGError().ok())) {
       if (wait_for_compact_options.timeout.count()) {
@@ -4348,6 +4393,7 @@ Status DBImpl::WaitForCompact(
           return Status::TimedOut();
         }
       } else {
+        TEST_SYNC_POINT("DBImpl::WaitForCompact:InsideLoop");
         bg_cv_.Wait();
       }
     } else if (wait_for_compact_options.close_db) {

db/db_impl/db_impl_debug.cc

@@ -314,6 +314,11 @@ const autovector<uint64_t>& DBImpl::TEST_GetFilesToQuarantine() const {
   return error_handler_.GetFilesToQuarantine();
 }
 
+void DBImpl::TEST_DeleteObsoleteFiles() {
+  InstrumentedMutexLock l(&mutex_);
+  DeleteObsoleteFiles();
+}
+
 size_t DBImpl::TEST_EstimateInMemoryStatsHistorySize() const {
   InstrumentedMutexLock l(&const_cast<DBImpl*>(this)->stats_history_mutex_);
   return EstimateInMemoryStatsHistorySize();

db/db_impl/db_impl_experimental.cc

@@ -104,7 +104,9 @@ Status DBImpl::PromoteL0(ColumnFamilyHandle* column_family, int target_level) {
         return status;
       }
 
-      if (i == 0) continue;
+      if (i == 0) {
+        continue;
+      }
       auto prev_f = l0_files[i - 1];
       if (icmp->Compare(prev_f->largest, f->smallest) >= 0) {
         ROCKS_LOG_INFO(immutable_db_options_.info_log,
@@ -148,9 +150,9 @@ Status DBImpl::PromoteL0(ColumnFamilyHandle* column_family, int target_level) {
                                     read_options, write_options, &edit, &mutex_,
                                     directories_.GetDbDir());
     if (status.ok()) {
-      InstallSuperVersionAndScheduleWork(cfd,
-                                         &job_context.superversion_contexts[0],
-                                         *cfd->GetLatestMutableCFOptions());
+      InstallSuperVersionAndScheduleWork(
+          cfd, job_context.superversion_contexts.data(),
+          *cfd->GetLatestMutableCFOptions());
     }
   }  // lock released here
   LogFlush(immutable_db_options_.info_log);

db/db_impl/db_impl_files.cc

@@ -28,6 +28,8 @@ uint64_t DBImpl::MinLogNumberToKeep() {
   return versions_->min_log_number_to_keep();
 }
 
+uint64_t DBImpl::MinLogNumberToRecycle() { return min_log_number_to_recycle_; }
+
 uint64_t DBImpl::MinObsoleteSstNumberToKeep() {
   mutex_.AssertHeld();
   if (!pending_outputs_.empty()) {
@@ -175,31 +177,10 @@ void DBImpl::FindObsoleteFiles(JobContext* job_context, bool force,
     versions_->AddLiveFiles(&job_context->sst_live, &job_context->blob_live);
     InfoLogPrefix info_log_prefix(!immutable_db_options_.db_log_dir.empty(),
                                   dbname_);
-    std::set<std::string> paths;
-    for (size_t path_id = 0; path_id < immutable_db_options_.db_paths.size();
-         path_id++) {
-      paths.insert(immutable_db_options_.db_paths[path_id].path);
-    }
-
-    // Note that if cf_paths is not specified in the ColumnFamilyOptions
-    // of a particular column family, we use db_paths as the cf_paths
-    // setting. Hence, there can be multiple duplicates of files from db_paths
-    // in the following code. The duplicate are removed while identifying
-    // unique files in PurgeObsoleteFiles.
-    for (auto cfd : *versions_->GetColumnFamilySet()) {
-      for (size_t path_id = 0; path_id < cfd->ioptions()->cf_paths.size();
-           path_id++) {
-        auto& path = cfd->ioptions()->cf_paths[path_id].path;
-
-        if (paths.find(path) == paths.end()) {
-          paths.insert(path);
-        }
-      }
-    }
-
+    // PurgeObsoleteFiles will dedupe duplicate files.
     IOOptions io_opts;
     io_opts.do_not_recurse = true;
-    for (auto& path : paths) {
+    for (auto& path : CollectAllDBPaths()) {
       // set of all files in the directory. We'll exclude files that are still
       // alive in the subsequent processings.
       std::vector<std::string> files;
@@ -298,7 +279,8 @@ void DBImpl::FindObsoleteFiles(JobContext* job_context, bool force,
     while (alive_log_files_.begin()->number < min_log_number) {
       auto& earliest = *alive_log_files_.begin();
       if (immutable_db_options_.recycle_log_file_num >
-          log_recycle_files_.size()) {
+              log_recycle_files_.size() &&
+          earliest.number >= MinLogNumberToRecycle()) {
         ROCKS_LOG_INFO(immutable_db_options_.info_log,
                        "adding log %" PRIu64 " to recycle list\n",
                        earliest.number);
@@ -963,28 +945,26 @@ Status DBImpl::SetupDBId(const WriteOptions& write_options, bool read_only,
   return s;
 }
 
-Status DBImpl::DeleteUnreferencedSstFiles(RecoveryContext* recovery_ctx) {
-  mutex_.AssertHeld();
-  std::vector<std::string> paths;
-  paths.push_back(NormalizePath(dbname_ + std::string(1, kFilePathSeparator)));
+std::set<std::string> DBImpl::CollectAllDBPaths() {
+  std::set<std::string> all_db_paths;
+  all_db_paths.insert(NormalizePath(dbname_));
   for (const auto& db_path : immutable_db_options_.db_paths) {
-    paths.push_back(
-        NormalizePath(db_path.path + std::string(1, kFilePathSeparator)));
+    all_db_paths.insert(NormalizePath(db_path.path));
   }
   for (const auto* cfd : *versions_->GetColumnFamilySet()) {
     for (const auto& cf_path : cfd->ioptions()->cf_paths) {
-      paths.push_back(
-          NormalizePath(cf_path.path + std::string(1, kFilePathSeparator)));
+      all_db_paths.insert(NormalizePath(cf_path.path));
     }
   }
-  // Dedup paths
-  std::sort(paths.begin(), paths.end());
-  paths.erase(std::unique(paths.begin(), paths.end()), paths.end());
+  return all_db_paths;
+}
 
+Status DBImpl::MaybeUpdateNextFileNumber(RecoveryContext* recovery_ctx) {
+  mutex_.AssertHeld();
   uint64_t next_file_number = versions_->current_next_file_number();
   uint64_t largest_file_number = next_file_number;
   Status s;
-  for (const auto& path : paths) {
+  for (const auto& path : CollectAllDBPaths()) {
     std::vector<std::string> files;
     s = env_->GetChildren(path, &files);
     if (!s.ok()) {
@@ -996,13 +976,10 @@ Status DBImpl::DeleteUnreferencedSstFiles(RecoveryContext* recovery_ctx) {
       if (!ParseFileName(fname, &number, &type)) {
         continue;
       }
-      // path ends with '/' or '\\'
-      const std::string normalized_fpath = path + fname;
+      const std::string normalized_fpath = path + kFilePathSeparator + fname;
       largest_file_number = std::max(largest_file_number, number);
-      if (type == kTableFile && number >= next_file_number &&
-          recovery_ctx->files_to_delete_.find(normalized_fpath) ==
-              recovery_ctx->files_to_delete_.end()) {
-        recovery_ctx->files_to_delete_.emplace(normalized_fpath, path);
+      if ((type == kTableFile || type == kBlobFile)) {
+        recovery_ctx->existing_data_files_.push_back(normalized_fpath);
       }
     }
   }
@@ -1022,5 +999,4 @@ Status DBImpl::DeleteUnreferencedSstFiles(RecoveryContext* recovery_ctx) {
   recovery_ctx->UpdateVersionEdits(default_cfd, edit);
   return s;
 }
-
 }  // namespace ROCKSDB_NAMESPACE

db/db_impl/db_impl_follower.cc

@@ -0,0 +1,347 @@
+//  Copyright (c) 2024-present, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+
+#include "db/db_impl/db_impl_follower.h"
+
+#include <cinttypes>
+
+#include "db/arena_wrapped_db_iter.h"
+#include "db/merge_context.h"
+#include "env/composite_env_wrapper.h"
+#include "env/fs_on_demand.h"
+#include "logging/auto_roll_logger.h"
+#include "logging/logging.h"
+#include "monitoring/perf_context_imp.h"
+#include "rocksdb/configurable.h"
+#include "rocksdb/db.h"
+#include "util/cast_util.h"
+#include "util/write_batch_util.h"
+
+namespace ROCKSDB_NAMESPACE {
+
+DBImplFollower::DBImplFollower(const DBOptions& db_options,
+                               std::unique_ptr<Env>&& env,
+                               const std::string& dbname, std::string src_path)
+    : DBImplSecondary(db_options, dbname, ""),
+      env_guard_(std::move(env)),
+      stop_requested_(false),
+      src_path_(std::move(src_path)),
+      cv_(&mu_) {
+  ROCKS_LOG_INFO(immutable_db_options_.info_log,
+                 "Opening the db in follower mode");
+  LogFlush(immutable_db_options_.info_log);
+}
+
+DBImplFollower::~DBImplFollower() {
+  Status s = Close();
+  if (!s.ok()) {
+    ROCKS_LOG_INFO(immutable_db_options_.info_log, "Error closing DB : %s",
+                   s.ToString().c_str());
+  }
+}
+
+// Recover a follower DB instance by reading the MANIFEST. The verification
+// as part of the MANIFEST replay will ensure that local links to the
+// leader's files are created, thus ensuring we can continue reading them
+// even if the leader deletes those files due to compaction.
+// TODO:
+//  1. Devise a mechanism to prevent misconfiguration by, for example,
+//     keeping a local copy of the IDENTITY file and cross checking
+//  2. Make the recovery more robust by retrying if the first attempt
+//     fails.
+Status DBImplFollower::Recover(
+    const std::vector<ColumnFamilyDescriptor>& column_families,
+    bool /*readonly*/, bool /*error_if_wal_file_exists*/,
+    bool /*error_if_data_exists_in_wals*/, bool /*is_retry*/, uint64_t*,
+    RecoveryContext* /*recovery_ctx*/, bool* /*can_retry*/) {
+  mutex_.AssertHeld();
+
+  JobContext job_context(0);
+  Status s;
+  s = static_cast<ReactiveVersionSet*>(versions_.get())
+          ->Recover(column_families, &manifest_reader_, &manifest_reporter_,
+                    &manifest_reader_status_);
+  if (!s.ok()) {
+    if (manifest_reader_status_) {
+      manifest_reader_status_->PermitUncheckedError();
+    }
+    return s;
+  }
+  if (immutable_db_options_.paranoid_checks && s.ok()) {
+    s = CheckConsistency();
+  }
+  if (s.ok()) {
+    default_cf_handle_ = new ColumnFamilyHandleImpl(
+        versions_->GetColumnFamilySet()->GetDefault(), this, &mutex_);
+    default_cf_internal_stats_ = default_cf_handle_->cfd()->internal_stats();
+
+    // Start the periodic catch-up thread
+    // TODO: See if it makes sense to have a threadpool, rather than a thread
+    // per follower DB instance
+    catch_up_thread_.reset(
+        new port::Thread(&DBImplFollower::PeriodicRefresh, this));
+  }
+
+  return s;
+}
+
+// Try to catch up by tailing the MANIFEST.
+// TODO:
+//   1. Cleanup obsolete files afterward
+//   2. Add some error notifications and statistics
+Status DBImplFollower::TryCatchUpWithLeader() {
+  assert(versions_.get() != nullptr);
+  assert(manifest_reader_.get() != nullptr);
+  Status s;
+
+  TEST_SYNC_POINT("DBImplFollower::TryCatchupWithLeader:Begin1");
+  TEST_SYNC_POINT("DBImplFollower::TryCatchupWithLeader:Begin2");
+  // read the manifest and apply new changes to the follower instance
+  std::unordered_set<ColumnFamilyData*> cfds_changed;
+  JobContext job_context(0, true /*create_superversion*/);
+  {
+    InstrumentedMutexLock lock_guard(&mutex_);
+    std::vector<std::string> files_to_delete;
+    s = static_cast_with_check<ReactiveVersionSet>(versions_.get())
+            ->ReadAndApply(&mutex_, &manifest_reader_,
+                           manifest_reader_status_.get(), &cfds_changed,
+                           &files_to_delete);
+    ReleaseFileNumberFromPendingOutputs(pending_outputs_inserted_elem_);
+    pending_outputs_inserted_elem_.reset(new std::list<uint64_t>::iterator(
+        CaptureCurrentFileNumberInPendingOutputs()));
+
+    ROCKS_LOG_INFO(immutable_db_options_.info_log, "Last sequence is %" PRIu64,
+                   static_cast<uint64_t>(versions_->LastSequence()));
+    ROCKS_LOG_INFO(
+        immutable_db_options_.info_log, "Next file number is %" PRIu64,
+        static_cast<uint64_t>(versions_->current_next_file_number()));
+    for (ColumnFamilyData* cfd : cfds_changed) {
+      if (cfd->IsDropped()) {
+        ROCKS_LOG_DEBUG(immutable_db_options_.info_log, "[%s] is dropped\n",
+                        cfd->GetName().c_str());
+        continue;
+      }
+      VersionStorageInfo::LevelSummaryStorage tmp;
+      ROCKS_LOG_DEBUG(immutable_db_options_.info_log,
+                      "[%s] Level summary: %s\n", cfd->GetName().c_str(),
+                      cfd->current()->storage_info()->LevelSummary(&tmp));
+    }
+
+    if (s.ok()) {
+      for (auto cfd : cfds_changed) {
+        if (cfd->mem()->GetEarliestSequenceNumber() <
+            versions_->LastSequence()) {
+          // Construct a new memtable with earliest sequence number set to the
+          // last sequence number in the VersionSet. This matters when
+          // DBImpl::MultiCFSnapshot tries to get consistent references
+          // to super versions in a lock free manner, it checks the earliest
+          // sequence number to detect if there was a change in version in
+          // the meantime.
+          const MutableCFOptions mutable_cf_options =
+              *cfd->GetLatestMutableCFOptions();
+          MemTable* new_mem = cfd->ConstructNewMemtable(
+              mutable_cf_options, versions_->LastSequence());
+          cfd->mem()->SetNextLogNumber(cfd->GetLogNumber());
+          cfd->mem()->ConstructFragmentedRangeTombstones();
+          cfd->imm()->Add(cfd->mem(), &job_context.memtables_to_free);
+          new_mem->Ref();
+          cfd->SetMemtable(new_mem);
+        }
+
+        // This will check if the old memtable is still referenced
+        cfd->imm()->RemoveOldMemTables(cfd->GetLogNumber(),
+                                       &job_context.memtables_to_free);
+        auto& sv_context = job_context.superversion_contexts.back();
+        cfd->InstallSuperVersion(&sv_context, &mutex_);
+        sv_context.NewSuperVersion();
+      }
+    }
+
+    for (auto& file : files_to_delete) {
+      IOStatus io_s = fs_->DeleteFile(file, IOOptions(), nullptr);
+      if (!io_s.ok()) {
+        ROCKS_LOG_INFO(immutable_db_options_.info_log,
+                       "Cannot delete file %s: %s", file.c_str(),
+                       io_s.ToString().c_str());
+      }
+    }
+  }
+  job_context.Clean();
+
+  // Cleanup unused, obsolete files.
+  JobContext purge_files_job_context(0);
+  {
+    InstrumentedMutexLock lock_guard(&mutex_);
+    // Currently, follower instance does not create any database files, thus
+    // is unnecessary for the follower to force full scan.
+    FindObsoleteFiles(&purge_files_job_context, /*force=*/false);
+  }
+  if (purge_files_job_context.HaveSomethingToDelete()) {
+    PurgeObsoleteFiles(purge_files_job_context);
+  }
+  purge_files_job_context.Clean();
+
+  TEST_SYNC_POINT("DBImplFollower::TryCatchupWithLeader:End");
+
+  return s;
+}
+
+void DBImplFollower::PeriodicRefresh() {
+  while (!stop_requested_.load()) {
+    MutexLock l(&mu_);
+    int64_t wait_until =
+        immutable_db_options_.clock->NowMicros() +
+        immutable_db_options_.follower_refresh_catchup_period_ms * 1000;
+    immutable_db_options_.clock->TimedWait(
+        &cv_, std::chrono::microseconds(wait_until));
+    if (stop_requested_.load()) {
+      break;
+    }
+    Status s;
+    for (uint64_t i = 0;
+         i < immutable_db_options_.follower_catchup_retry_count &&
+         !stop_requested_.load();
+         ++i) {
+      s = TryCatchUpWithLeader();
+
+      if (s.ok()) {
+        ROCKS_LOG_INFO(immutable_db_options_.info_log,
+                       "Successful catch up on attempt %llu",
+                       static_cast<unsigned long long>(i));
+        break;
+      }
+      wait_until = immutable_db_options_.clock->NowMicros() +
+                   immutable_db_options_.follower_catchup_retry_wait_ms * 1000;
+      immutable_db_options_.clock->TimedWait(
+          &cv_, std::chrono::microseconds(wait_until));
+    }
+    if (!s.ok()) {
+      ROCKS_LOG_INFO(immutable_db_options_.info_log, "Catch up unsuccessful");
+    }
+  }
+}
+
+Status DBImplFollower::Close() {
+  if (catch_up_thread_) {
+    stop_requested_.store(true);
+    {
+      MutexLock l(&mu_);
+      cv_.SignalAll();
+    }
+    catch_up_thread_->join();
+    catch_up_thread_.reset();
+  }
+
+  ReleaseFileNumberFromPendingOutputs(pending_outputs_inserted_elem_);
+
+  return DBImpl::Close();
+}
+
+Status DB::OpenAsFollower(const Options& options, const std::string& dbname,
+                          const std::string& leader_path,
+                          std::unique_ptr<DB>* dbptr) {
+  dbptr->reset();
+
+  DBOptions db_options(options);
+  ColumnFamilyOptions cf_options(options);
+  std::vector<ColumnFamilyDescriptor> column_families;
+  column_families.emplace_back(kDefaultColumnFamilyName, cf_options);
+  std::vector<ColumnFamilyHandle*> handles;
+
+  Status s = DB::OpenAsFollower(db_options, dbname, leader_path,
+                                column_families, &handles, dbptr);
+  if (s.ok()) {
+    assert(handles.size() == 1);
+    delete handles[0];
+  }
+  return s;
+}
+
+Status DB::OpenAsFollower(
+    const DBOptions& db_options, const std::string& dbname,
+    const std::string& src_path,
+    const std::vector<ColumnFamilyDescriptor>& column_families,
+    std::vector<ColumnFamilyHandle*>* handles, std::unique_ptr<DB>* dbptr) {
+  dbptr->reset();
+
+  FileSystem* fs = db_options.env->GetFileSystem().get();
+  {
+    IOStatus io_s;
+    if (db_options.create_if_missing) {
+      io_s = fs->CreateDirIfMissing(dbname, IOOptions(), nullptr);
+    } else {
+      io_s = fs->FileExists(dbname, IOOptions(), nullptr);
+    }
+    if (!io_s.ok()) {
+      return static_cast<Status>(io_s);
+    }
+  }
+  std::unique_ptr<Env> new_env(new CompositeEnvWrapper(
+      db_options.env, NewOnDemandFileSystem(db_options.env->GetFileSystem(),
+                                            src_path, dbname)));
+
+  DBOptions tmp_opts(db_options);
+  Status s;
+  tmp_opts.env = new_env.get();
+  if (nullptr == tmp_opts.info_log) {
+    s = CreateLoggerFromOptions(dbname, tmp_opts, &tmp_opts.info_log);
+    if (!s.ok()) {
+      tmp_opts.info_log = nullptr;
+      return s;
+    }
+  }
+
+  handles->clear();
+  DBImplFollower* impl =
+      new DBImplFollower(tmp_opts, std::move(new_env), dbname, src_path);
+  impl->versions_.reset(new ReactiveVersionSet(
+      dbname, &impl->immutable_db_options_, impl->file_options_,
+      impl->table_cache_.get(), impl->write_buffer_manager_,
+      &impl->write_controller_, impl->io_tracer_));
+  impl->column_family_memtables_.reset(
+      new ColumnFamilyMemTablesImpl(impl->versions_->GetColumnFamilySet()));
+  impl->wal_in_db_path_ = impl->immutable_db_options_.IsWalDirSameAsDBPath();
+
+  impl->mutex_.Lock();
+  s = impl->Recover(column_families, /*read_only=*/true,
+                    /*error_if_wal_file_exists=*/false,
+                    /*error_if_data_exists_in_wals=*/false);
+  if (s.ok()) {
+    for (const auto& cf : column_families) {
+      auto cfd =
+          impl->versions_->GetColumnFamilySet()->GetColumnFamily(cf.name);
+      if (nullptr == cfd) {
+        s = Status::InvalidArgument("Column family not found", cf.name);
+        break;
+      }
+      handles->push_back(new ColumnFamilyHandleImpl(cfd, impl, &impl->mutex_));
+    }
+  }
+  SuperVersionContext sv_context(false /* create_superversion */);
+  if (s.ok()) {
+    for (auto cfd : *impl->versions_->GetColumnFamilySet()) {
+      sv_context.NewSuperVersion();
+      cfd->InstallSuperVersion(&sv_context, &impl->mutex_);
+    }
+  }
+  impl->mutex_.Unlock();
+  sv_context.Clean();
+  if (s.ok()) {
+    dbptr->reset(impl);
+    for (auto h : *handles) {
+      impl->NewThreadStatusCfInfo(
+          static_cast_with_check<ColumnFamilyHandleImpl>(h)->cfd());
+    }
+  } else {
+    for (auto h : *handles) {
+      delete h;
+    }
+    handles->clear();
+    delete impl;
+  }
+  return s;
+}
+
+}  // namespace ROCKSDB_NAMESPACE

db/db_impl/db_impl_follower.h

@@ -0,0 +1,54 @@
+//  Copyright (c) 2024-present, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+
+#pragma once
+
+#include <string>
+#include <vector>
+
+#include "db/db_impl/db_impl.h"
+#include "db/db_impl/db_impl_secondary.h"
+#include "logging/logging.h"
+#include "port/port.h"
+
+namespace ROCKSDB_NAMESPACE {
+
+class DBImplFollower : public DBImplSecondary {
+ public:
+  DBImplFollower(const DBOptions& db_options, std::unique_ptr<Env>&& env,
+                 const std::string& dbname, std::string src_path);
+  ~DBImplFollower();
+
+  Status Close() override;
+
+ protected:
+  bool OwnTablesAndLogs() const override {
+    // TODO: Change this to true once we've properly implemented file
+    // deletion for the read scaling case
+    return true;
+  }
+
+  Status Recover(const std::vector<ColumnFamilyDescriptor>& column_families,
+                 bool /*readonly*/, bool /*error_if_wal_file_exists*/,
+                 bool /*error_if_data_exists_in_wals*/,
+                 bool /*is_retry*/ = false, uint64_t* = nullptr,
+                 RecoveryContext* /*recovery_ctx*/ = nullptr,
+                 bool* /*can_retry*/ = nullptr) override;
+
+ private:
+  friend class DB;
+
+  Status TryCatchUpWithLeader();
+  void PeriodicRefresh();
+
+  std::unique_ptr<Env> env_guard_;
+  std::unique_ptr<port::Thread> catch_up_thread_;
+  std::atomic<bool> stop_requested_;
+  std::string src_path_;
+  port::Mutex mu_;
+  port::CondVar cv_;
+  std::unique_ptr<std::list<uint64_t>::iterator> pending_outputs_inserted_elem_;
+};
+}  // namespace ROCKSDB_NAMESPACE

db/db_impl/db_impl_open.cc

@@ -104,7 +104,6 @@ DBOptions SanitizeOptions(const std::string& dbname, const DBOptions& src,
   if (result.recycle_log_file_num &&
       (result.wal_recovery_mode ==
            WALRecoveryMode::kTolerateCorruptedTailRecords ||
-       result.wal_recovery_mode == WALRecoveryMode::kPointInTimeRecovery ||
        result.wal_recovery_mode == WALRecoveryMode::kAbsoluteConsistency)) {
     // - kTolerateCorruptedTailRecords is inconsistent with recycle log file
     //   feature. WAL recycling expects recovery success upon encountering a
@@ -176,15 +175,6 @@ DBOptions SanitizeOptions(const std::string& dbname, const DBOptions& src,
       }
     }
   }
-  // When the DB is stopped, it's possible that there are some .trash files that
-  // were not deleted yet, when we open the DB we will find these .trash files
-  // and schedule them to be deleted (or delete immediately if SstFileManager
-  // was not used)
-  auto sfm = static_cast<SstFileManagerImpl*>(result.sst_file_manager.get());
-  for (size_t i = 0; i < result.db_paths.size(); i++) {
-    DeleteScheduler::CleanupDirectory(result.env, sfm, result.db_paths[i].path)
-        .PermitUncheckedError();
-  }
 
   // Create a default SstFileManager for purposes of tracking compaction size
   // and facilitating recovery from out of space errors.
@@ -281,12 +271,6 @@ Status DBImpl::ValidateOptions(const DBOptions& db_options) {
         "atomic_flush is incompatible with enable_pipelined_write");
   }
 
-  // TODO remove this restriction
-  if (db_options.atomic_flush && db_options.best_efforts_recovery) {
-    return Status::InvalidArgument(
-        "atomic_flush is currently incompatible with best-efforts recovery");
-  }
-
   if (db_options.use_direct_io_for_flush_and_compaction &&
       0 == db_options.writable_file_max_buffer_size) {
     return Status::InvalidArgument(
@@ -420,7 +404,8 @@ IOStatus Directories::SetDirectories(FileSystem* fs, const std::string& dbname,
 Status DBImpl::Recover(
     const std::vector<ColumnFamilyDescriptor>& column_families, bool read_only,
     bool error_if_wal_file_exists, bool error_if_data_exists_in_wals,
-    uint64_t* recovered_seq, RecoveryContext* recovery_ctx) {
+    bool is_retry, uint64_t* recovered_seq, RecoveryContext* recovery_ctx,
+    bool* can_retry) {
   mutex_.AssertHeld();
 
   const WriteOptions write_options(Env::IOActivity::kDBOpen);
@@ -536,7 +521,31 @@ Status DBImpl::Recover(
   Status s;
   bool missing_table_file = false;
   if (!immutable_db_options_.best_efforts_recovery) {
-    s = versions_->Recover(column_families, read_only, &db_id_);
+    // Status of reading the descriptor file
+    Status desc_status;
+    s = versions_->Recover(column_families, read_only, &db_id_,
+                           /*no_error_if_files_missing=*/false, is_retry,
+                           &desc_status);
+    desc_status.PermitUncheckedError();
+    if (can_retry) {
+      // If we're opening for the first time and the failure is likely due to
+      // a corrupt MANIFEST file (could result in either the log::Reader
+      // detecting a corrupt record, or SST files not found error due to
+      // discarding badly formed tail records)
+      if (!is_retry &&
+          (desc_status.IsCorruption() || s.IsNotFound() || s.IsCorruption()) &&
+          CheckFSFeatureSupport(fs_.get(),
+                                FSSupportedOps::kVerifyAndReconstructRead)) {
+        *can_retry = true;
+        ROCKS_LOG_ERROR(
+            immutable_db_options_.info_log,
+            "Possible corruption detected while replaying MANIFEST %s, %s. "
+            "Will be retried.",
+            desc_status.ToString().c_str(), s.ToString().c_str());
+      } else {
+        *can_retry = false;
+      }
+    }
   } else {
     assert(!files_in_dbname.empty());
     s = versions_->TryRecover(column_families, read_only, files_in_dbname,
@@ -651,7 +660,7 @@ Status DBImpl::Recover(
   s = SetupDBId(write_options, read_only, recovery_ctx);
   ROCKS_LOG_INFO(immutable_db_options_.info_log, "DB ID: %s\n", db_id_.c_str());
   if (s.ok() && !read_only) {
-    s = DeleteUnreferencedSstFiles(recovery_ctx);
+    s = MaybeUpdateNextFileNumber(recovery_ctx);
   }
 
   if (immutable_db_options_.paranoid_checks && s.ok()) {
@@ -774,8 +783,8 @@ Status DBImpl::Recover(
       std::sort(wals.begin(), wals.end());
 
       bool corrupted_wal_found = false;
-      s = RecoverLogFiles(wals, &next_sequence, read_only, &corrupted_wal_found,
-                          recovery_ctx);
+      s = RecoverLogFiles(wals, &next_sequence, read_only, is_retry,
+                          &corrupted_wal_found, recovery_ctx);
       if (corrupted_wal_found && recovered_seq != nullptr) {
         *recovered_seq = next_sequence;
       }
@@ -953,19 +962,6 @@ Status DBImpl::LogAndApplyForRecovery(const RecoveryContext& recovery_ctx) {
                                     recovery_ctx.mutable_cf_opts_, read_options,
                                     write_options, recovery_ctx.edit_lists_,
                                     &mutex_, directories_.GetDbDir());
-  if (s.ok() && !(recovery_ctx.files_to_delete_.empty())) {
-    mutex_.Unlock();
-    for (const auto& stale_sst_file : recovery_ctx.files_to_delete_) {
-      s = DeleteDBFile(&immutable_db_options_, stale_sst_file.first,
-                       stale_sst_file.second,
-                       /*force_bg=*/false,
-                       /*force_fg=*/false);
-      if (!s.ok()) {
-        break;
-      }
-    }
-    mutex_.Lock();
-  }
   return s;
 }
 
@@ -1085,13 +1081,14 @@ bool DBImpl::InvokeWalFilterIfNeededOnWalRecord(uint64_t wal_number,
 // REQUIRES: wal_numbers are sorted in ascending order
 Status DBImpl::RecoverLogFiles(const std::vector<uint64_t>& wal_numbers,
                                SequenceNumber* next_sequence, bool read_only,
-                               bool* corrupted_wal_found,
+                               bool is_retry, bool* corrupted_wal_found,
                                RecoveryContext* recovery_ctx) {
   struct LogReporter : public log::Reader::Reporter {
     Env* env;
     Logger* info_log;
     const char* fname;
     Status* status;  // nullptr if immutable_db_options_.paranoid_checks==false
+    bool* old_log_record;
     void Corruption(size_t bytes, const Status& s) override {
       ROCKS_LOG_WARN(info_log, "%s%s: dropping %d bytes; %s",
                      (status == nullptr ? "(ignoring error) " : ""), fname,
@@ -1100,10 +1097,19 @@ Status DBImpl::RecoverLogFiles(const std::vector<uint64_t>& wal_numbers,
         *status = s;
       }
     }
+
+    void OldLogRecord(size_t bytes) override {
+      if (old_log_record != nullptr) {
+        *old_log_record = true;
+      }
+      ROCKS_LOG_WARN(info_log, "%s: dropping %d bytes; possibly recycled",
+                     fname, static_cast<int>(bytes));
+    }
   };
 
   mutex_.AssertHeld();
   Status status;
+  bool old_log_record = false;
   std::unordered_map<int, VersionEdit> version_edits;
   // no need to refcount because iteration is under mutex
   for (auto cfd : *versions_->GetColumnFamilySet()) {
@@ -1186,7 +1192,7 @@ Status DBImpl::RecoverLogFiles(const std::vector<uint64_t>& wal_numbers,
       }
       file_reader.reset(new SequentialFileReader(
           std::move(file), fname, immutable_db_options_.log_readahead_size,
-          io_tracer_));
+          io_tracer_, /*listeners=*/{}, /*rate_limiter=*/nullptr, is_retry));
     }
 
     // Create the log reader.
@@ -1194,6 +1200,7 @@ Status DBImpl::RecoverLogFiles(const std::vector<uint64_t>& wal_numbers,
     reporter.env = env_;
     reporter.info_log = immutable_db_options_.info_log.get();
     reporter.fname = fname.c_str();
+    reporter.old_log_record = &old_log_record;
     if (!immutable_db_options_.paranoid_checks ||
         immutable_db_options_.wal_recovery_mode ==
             WALRecoveryMode::kSkipAnyCorruptedRecords) {
@@ -1341,7 +1348,7 @@ Status DBImpl::RecoverLogFiles(const std::vector<uint64_t>& wal_numbers,
       }
     }
 
-    if (!status.ok()) {
+    if (!status.ok() || old_log_record) {
       if (status.IsNotSupported()) {
         // We should not treat NotSupported as corruption. It is rather a clear
         // sign that we are processing a WAL that is produced by an incompatible
@@ -1366,6 +1373,7 @@ Status DBImpl::RecoverLogFiles(const std::vector<uint64_t>& wal_numbers,
         }
         // We should ignore the error but not continue replaying
         status = Status::OK();
+        old_log_record = false;
         stop_replay_for_corruption = true;
         corrupted_wal_number = wal_number;
         if (corrupted_wal_found != nullptr) {
@@ -1636,7 +1644,8 @@ Status DBImpl::WriteLevel0TableForRecovery(int job_id, ColumnFamilyData* cfd,
   Status s;
   TableProperties table_properties;
   {
-    ScopedArenaIterator iter(mem->NewIterator(ro, &arena));
+    ScopedArenaPtr<InternalIterator> iter(
+        mem->NewIterator(ro, /*seqno_to_time_mapping=*/nullptr, &arena));
     ROCKS_LOG_DEBUG(immutable_db_options_.info_log,
                     "[%s] [WriteLevel0TableForRecovery]"
                     " Level-0 table #%" PRIu64 ": started",
@@ -1689,7 +1698,6 @@ Status DBImpl::WriteLevel0TableForRecovery(int job_id, ColumnFamilyData* cfd,
           TableFileCreationReason::kRecovery, 0 /* oldest_key_time */,
           0 /* file_creation_time */, db_id_, db_session_id_,
           0 /* target_file_size */, meta.fd.GetNumber());
-      SeqnoToTimeMapping empty_seqno_to_time_mapping;
       Version* version = cfd->current();
       version->Ref();
       uint64_t num_input_entries = 0;
@@ -1700,7 +1708,7 @@ Status DBImpl::WriteLevel0TableForRecovery(int job_id, ColumnFamilyData* cfd,
           snapshot_seqs, earliest_write_conflict_snapshot, kMaxSequenceNumber,
           snapshot_checker, paranoid_file_checks, cfd->internal_stats(), &io_s,
           io_tracer_, BlobFileCreationReason::kRecovery,
-          empty_seqno_to_time_mapping, &event_logger_, job_id,
+          nullptr /* seqno_to_time_mapping */, &event_logger_, job_id,
           nullptr /* table_properties */, write_hint,
           nullptr /*full_history_ts_low*/, &blob_callback_, version,
           &num_input_entries);
@@ -1799,11 +1807,9 @@ Status DB::Open(const Options& options, const std::string& dbname, DB** dbptr) {
   DBOptions db_options(options);
   ColumnFamilyOptions cf_options(options);
   std::vector<ColumnFamilyDescriptor> column_families;
-  column_families.push_back(
-      ColumnFamilyDescriptor(kDefaultColumnFamilyName, cf_options));
+  column_families.emplace_back(kDefaultColumnFamilyName, cf_options);
   if (db_options.persist_stats_to_disk) {
-    column_families.push_back(
-        ColumnFamilyDescriptor(kPersistentStatsColumnFamilyName, cf_options));
+    column_families.emplace_back(kPersistentStatsColumnFamilyName, cf_options);
   }
   std::vector<ColumnFamilyHandle*> handles;
   Status s = DB::Open(db_options, dbname, column_families, &handles, dbptr);
@@ -1830,8 +1836,12 @@ Status DB::Open(const DBOptions& db_options, const std::string& dbname,
   const bool kBatchPerTxn = true;
   ThreadStatusUtil::SetEnableTracking(db_options.enable_thread_tracking);
   ThreadStatusUtil::SetThreadOperation(ThreadStatus::OperationType::OP_DBOPEN);
-  Status s = DBImpl::Open(db_options, dbname, column_families, handles, dbptr,
-                          !kSeqPerBatch, kBatchPerTxn);
+  bool can_retry = false;
+  Status s;
+  do {
+    s = DBImpl::Open(db_options, dbname, column_families, handles, dbptr,
+                     !kSeqPerBatch, kBatchPerTxn, can_retry, &can_retry);
+  } while (!s.ok() && can_retry);
   ThreadStatusUtil::ResetThreadStatus();
   return s;
 }
@@ -1950,10 +1960,55 @@ IOStatus DBImpl::CreateWAL(const WriteOptions& write_options,
   return io_s;
 }
 
+void DBImpl::TrackExistingDataFiles(
+    const std::vector<std::string>& existing_data_files) {
+  auto sfm = static_cast<SstFileManagerImpl*>(
+      immutable_db_options_.sst_file_manager.get());
+  assert(sfm);
+  std::vector<ColumnFamilyMetaData> metadata;
+  GetAllColumnFamilyMetaData(&metadata);
+
+  std::unordered_set<std::string> referenced_files;
+  for (const auto& md : metadata) {
+    for (const auto& lmd : md.levels) {
+      for (const auto& fmd : lmd.files) {
+        // We're assuming that each sst file name exists in at most one of
+        // the paths.
+        std::string file_path =
+            fmd.directory + kFilePathSeparator + fmd.relative_filename;
+        sfm->OnAddFile(file_path, fmd.size).PermitUncheckedError();
+        referenced_files.insert(file_path);
+      }
+    }
+    for (const auto& bmd : md.blob_files) {
+      std::string name = bmd.blob_file_name;
+      // The BlobMetaData.blob_file_name may start with "/".
+      if (!name.empty() && name[0] == kFilePathSeparator) {
+        name = name.substr(1);
+      }
+      // We're assuming that each blob file name exists in at most one of
+      // the paths.
+      std::string file_path = bmd.blob_file_path + kFilePathSeparator + name;
+      sfm->OnAddFile(file_path, bmd.blob_file_size).PermitUncheckedError();
+      referenced_files.insert(file_path);
+    }
+  }
+
+  for (const auto& file_path : existing_data_files) {
+    if (referenced_files.find(file_path) != referenced_files.end()) {
+      continue;
+    }
+    // There shouldn't be any duplicated files. In case there is, SstFileManager
+    // will take care of deduping it.
+    sfm->OnAddFile(file_path).PermitUncheckedError();
+  }
+}
+
 Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
                     const std::vector<ColumnFamilyDescriptor>& column_families,
                     std::vector<ColumnFamilyHandle*>* handles, DB** dbptr,
-                    const bool seq_per_batch, const bool batch_per_txn) {
+                    const bool seq_per_batch, const bool batch_per_txn,
+                    const bool is_retry, bool* can_retry) {
   const WriteOptions write_options(Env::IOActivity::kDBOpen);
   const ReadOptions read_options(Env::IOActivity::kDBOpen);
 
@@ -1972,7 +2027,7 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
   handles->clear();
 
   size_t max_write_buffer_size = 0;
-  for (auto cf : column_families) {
+  for (const auto& cf : column_families) {
     max_write_buffer_size =
         std::max(max_write_buffer_size, cf.options.write_buffer_size);
   }
@@ -1996,7 +2051,7 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
         paths.emplace_back(cf_path.path);
       }
     }
-    for (auto& path : paths) {
+    for (const auto& path : paths) {
       s = impl->env_->CreateDirIfMissing(path);
       if (!s.ok()) {
         break;
@@ -2026,8 +2081,8 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
   uint64_t recovered_seq(kMaxSequenceNumber);
   s = impl->Recover(column_families, false /* read_only */,
                     false /* error_if_wal_file_exists */,
-                    false /* error_if_data_exists_in_wals */, &recovered_seq,
-                    &recovery_ctx);
+                    false /* error_if_data_exists_in_wals */, is_retry,
+                    &recovered_seq, &recovery_ctx, can_retry);
   if (s.ok()) {
     uint64_t new_log_number = impl->versions_->NewFileNumber();
     log::Writer* new_log = nullptr;
@@ -2035,6 +2090,11 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
         impl->GetWalPreallocateBlockSize(max_write_buffer_size);
     s = impl->CreateWAL(write_options, new_log_number, 0 /*recycle_log_number*/,
                         preallocate_block_size, &new_log);
+    if (s.ok()) {
+      // Prevent log files created by previous instance from being recycled.
+      // They might be in alive_log_file_, and might get recycled otherwise.
+      impl->min_log_number_to_recycle_ = new_log_number;
+    }
     if (s.ok()) {
       InstrumentedMutexLock wl(&impl->log_write_mutex_);
       impl->logfile_number_ = new_log_number;
@@ -2044,8 +2104,7 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
     }
 
     if (s.ok()) {
-      impl->alive_log_files_.push_back(
-          DBImpl::LogFileNumberSize(impl->logfile_number_));
+      impl->alive_log_files_.emplace_back(impl->logfile_number_);
       // In WritePrepared there could be gap in sequence numbers. This breaks
       // the trick we use in kPointInTimeRecovery which assumes the first seq in
       // the log right after the corrupted log is one larger than the last seq
@@ -2093,7 +2152,7 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
 
   if (s.ok()) {
     // set column family handles
-    for (auto cf : column_families) {
+    for (const auto& cf : column_families) {
       auto cfd =
           impl->versions_->GetColumnFamilySet()->GetColumnFamily(cf.name);
       if (cfd != nullptr) {
@@ -2163,9 +2222,6 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
         impl->WriteOptionsFile(write_options, true /*db_mutex_already_held*/);
     *dbptr = impl;
     impl->opened_successfully_ = true;
-    impl->DeleteObsoleteFiles();
-    TEST_SYNC_POINT("DBImpl::Open:AfterDeleteFiles");
-    impl->MaybeScheduleFlushOrCompaction();
   } else {
     persist_options_status.PermitUncheckedError();
   }
@@ -2180,73 +2236,10 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
     ROCKS_LOG_INFO(impl->immutable_db_options_.info_log,
                    "SstFileManager instance %p", sfm);
 
-    // Notify SstFileManager about all sst files that already exist in
-    // db_paths[0] and cf_paths[0] when the DB is opened.
-
-    // SstFileManagerImpl needs to know sizes of the files. For files whose size
-    // we already know (sst files that appear in manifest - typically that's the
-    // vast majority of all files), we'll pass the size to SstFileManager.
-    // For all other files SstFileManager will query the size from filesystem.
-
-    std::vector<ColumnFamilyMetaData> metadata;
-    impl->GetAllColumnFamilyMetaData(&metadata);
-
-    std::unordered_map<std::string, uint64_t> known_file_sizes;
-    for (const auto& md : metadata) {
-      for (const auto& lmd : md.levels) {
-        for (const auto& fmd : lmd.files) {
-          known_file_sizes[fmd.relative_filename] = fmd.size;
-        }
-      }
-      for (const auto& bmd : md.blob_files) {
-        std::string name = bmd.blob_file_name;
-        // The BlobMetaData.blob_file_name may start with "/".
-        if (!name.empty() && name[0] == '/') {
-          name = name.substr(1);
-        }
-        known_file_sizes[name] = bmd.blob_file_size;
-      }
-    }
-
-    std::vector<std::string> paths;
-    paths.emplace_back(impl->immutable_db_options_.db_paths[0].path);
-    for (auto& cf : column_families) {
-      if (!cf.options.cf_paths.empty()) {
-        paths.emplace_back(cf.options.cf_paths[0].path);
-      }
-    }
-    // Remove duplicate paths.
-    std::sort(paths.begin(), paths.end());
-    paths.erase(std::unique(paths.begin(), paths.end()), paths.end());
-    IOOptions io_opts;
-    io_opts.do_not_recurse = true;
-    for (auto& path : paths) {
-      std::vector<std::string> existing_files;
-      impl->immutable_db_options_.fs
-          ->GetChildren(path, io_opts, &existing_files,
-                        /*IODebugContext*=*/nullptr)
-          .PermitUncheckedError();  //**TODO: What do to on error?
-      for (auto& file_name : existing_files) {
-        uint64_t file_number;
-        FileType file_type;
-        std::string file_path = path + "/" + file_name;
-        if (ParseFileName(file_name, &file_number, &file_type) &&
-            (file_type == kTableFile || file_type == kBlobFile)) {
-          // TODO: Check for errors from OnAddFile?
-          if (known_file_sizes.count(file_name)) {
-            // We're assuming that each sst file name exists in at most one of
-            // the paths.
-            sfm->OnAddFile(file_path, known_file_sizes.at(file_name))
-                .PermitUncheckedError();
-          } else {
-            sfm->OnAddFile(file_path).PermitUncheckedError();
-          }
-        }
-      }
-    }
+    impl->TrackExistingDataFiles(recovery_ctx.existing_data_files_);
 
     // Reserve some disk buffer space. This is a heuristic - when we run out
-    // of disk space, this ensures that there is atleast write_buffer_size
+    // of disk space, this ensures that there is at least write_buffer_size
     // amount of free space before we resume DB writes. In low disk space
     // conditions, we want to avoid a lot of small L0 files due to frequent
     // WAL write failures and resultant forced flushes
@@ -2254,6 +2247,27 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
                            impl->immutable_db_options_.db_paths[0].path);
   }
 
+  if (s.ok()) {
+    // When the DB is stopped, it's possible that there are some .trash files
+    // that were not deleted yet, when we open the DB we will find these .trash
+    // files and schedule them to be deleted (or delete immediately if
+    // SstFileManager was not used).
+    // Note that we only start doing this and below delete obsolete file after
+    // `TrackExistingDataFiles` are called, the `max_trash_db_ratio` is
+    // ineffective otherwise and these files' deletion won't be rate limited
+    // which can cause discard stall.
+    for (const auto& path : impl->CollectAllDBPaths()) {
+      DeleteScheduler::CleanupDirectory(impl->immutable_db_options_.env, sfm,
+                                        path)
+          .PermitUncheckedError();
+    }
+    impl->mutex_.Lock();
+    // This will do a full scan.
+    impl->DeleteObsoleteFiles();
+    TEST_SYNC_POINT("DBImpl::Open:AfterDeleteFiles");
+    impl->MaybeScheduleFlushOrCompaction();
+    impl->mutex_.Unlock();
+  }
 
   if (s.ok()) {
     ROCKS_LOG_HEADER(impl->immutable_db_options_.info_log, "DB pointer %p",

db/db_impl/db_impl_readonly.cc

@@ -26,7 +26,7 @@ DBImplReadOnly::DBImplReadOnly(const DBOptions& db_options,
   LogFlush(immutable_db_options_.info_log);
 }
 
-DBImplReadOnly::~DBImplReadOnly() {}
+DBImplReadOnly::~DBImplReadOnly() = default;
 
 // Implementations of the DB interface
 Status DBImplReadOnly::GetImpl(const ReadOptions& read_options,
@@ -293,8 +293,7 @@ Status DB::OpenForReadOnly(const Options& options, const std::string& dbname,
   DBOptions db_options(options);
   ColumnFamilyOptions cf_options(options);
   std::vector<ColumnFamilyDescriptor> column_families;
-  column_families.push_back(
-      ColumnFamilyDescriptor(kDefaultColumnFamilyName, cf_options));
+  column_families.emplace_back(kDefaultColumnFamilyName, cf_options);
   std::vector<ColumnFamilyHandle*> handles;
 
   s = DBImplReadOnly::OpenForReadOnlyWithoutCheck(
@@ -339,7 +338,7 @@ Status DBImplReadOnly::OpenForReadOnlyWithoutCheck(
                            error_if_wal_file_exists);
   if (s.ok()) {
     // set column family handles
-    for (auto cf : column_families) {
+    for (const auto& cf : column_families) {
       auto cfd =
           impl->versions_->GetColumnFamilySet()->GetColumnFamily(cf.name);
       if (cfd == nullptr) {

db/db_impl/db_impl_secondary.cc

@@ -28,13 +28,13 @@ DBImplSecondary::DBImplSecondary(const DBOptions& db_options,
   LogFlush(immutable_db_options_.info_log);
 }
 
-DBImplSecondary::~DBImplSecondary() {}
+DBImplSecondary::~DBImplSecondary() = default;
 
 Status DBImplSecondary::Recover(
     const std::vector<ColumnFamilyDescriptor>& column_families,
     bool /*readonly*/, bool /*error_if_wal_file_exists*/,
-    bool /*error_if_data_exists_in_wals*/, uint64_t*,
-    RecoveryContext* /*recovery_ctx*/) {
+    bool /*error_if_data_exists_in_wals*/, bool /*is_retry*/, uint64_t*,
+    RecoveryContext* /*recovery_ctx*/, bool* /*can_retry*/) {
   mutex_.AssertHeld();
 
   JobContext job_context(0);
@@ -327,6 +327,7 @@ Status DBImplSecondary::RecoverLogFiles(
       status = *wal_read_status;
     }
     if (!status.ok()) {
+      wal_read_status->PermitUncheckedError();
       return status;
     }
   }
@@ -497,8 +498,8 @@ Iterator* DBImplSecondary::NewIterator(const ReadOptions& _read_options,
 
   Iterator* result = nullptr;
   auto cfh = static_cast_with_check<ColumnFamilyHandleImpl>(column_family);
+  assert(cfh != nullptr);
   auto cfd = cfh->cfd();
-  ReadCallback* read_callback = nullptr;  // No read callback provided.
   if (read_options.tailing) {
     return NewErrorIterator(Status::NotSupported(
         "tailing iterator not supported in secondary mode"));
@@ -517,27 +518,28 @@ Iterator* DBImplSecondary::NewIterator(const ReadOptions& _read_options,
         return NewErrorIterator(s);
       }
     }
-    result = NewIteratorImpl(read_options, cfd, sv, snapshot, read_callback);
+    result = NewIteratorImpl(read_options, cfh, sv, snapshot,
+                             nullptr /*read_callback*/);
   }
   return result;
 }
 
 ArenaWrappedDBIter* DBImplSecondary::NewIteratorImpl(
-    const ReadOptions& read_options, ColumnFamilyData* cfd,
+    const ReadOptions& read_options, ColumnFamilyHandleImpl* cfh,
     SuperVersion* super_version, SequenceNumber snapshot,
     ReadCallback* read_callback, bool expose_blob_index, bool allow_refresh) {
-  assert(nullptr != cfd);
+  assert(nullptr != cfh);
   assert(snapshot == kMaxSequenceNumber);
   snapshot = versions_->LastSequence();
   assert(snapshot != kMaxSequenceNumber);
   auto db_iter = NewArenaWrappedDbIterator(
-      env_, read_options, *cfd->ioptions(), super_version->mutable_cf_options,
-      super_version->current, snapshot,
+      env_, read_options, *cfh->cfd()->ioptions(),
+      super_version->mutable_cf_options, super_version->current, snapshot,
       super_version->mutable_cf_options.max_sequential_skip_in_iterations,
-      super_version->version_number, read_callback, this, cfd,
-      expose_blob_index, allow_refresh);
+      super_version->version_number, read_callback, cfh, expose_blob_index,
+      allow_refresh);
   auto internal_iter = NewInternalIterator(
-      db_iter->GetReadOptions(), cfd, super_version, db_iter->GetArena(),
+      db_iter->GetReadOptions(), cfh->cfd(), super_version, db_iter->GetArena(),
       snapshot, /* allow_unprepared_value */ true, db_iter);
   db_iter->SetIterUnderDBIter(internal_iter);
   return db_iter;
@@ -596,28 +598,29 @@ Status DBImplSecondary::NewIterators(
     return Status::NotSupported("snapshot not supported in secondary mode");
   } else {
     SequenceNumber read_seq(kMaxSequenceNumber);
-    autovector<std::tuple<ColumnFamilyData*, SuperVersion*>> cfd_to_sv;
+    autovector<std::tuple<ColumnFamilyHandleImpl*, SuperVersion*>> cfh_to_sv;
     const bool check_read_ts =
         read_options.timestamp && read_options.timestamp->size() > 0;
-    for (auto cfh : column_families) {
-      ColumnFamilyData* cfd = static_cast<ColumnFamilyHandleImpl*>(cfh)->cfd();
+    for (auto cf : column_families) {
+      auto cfh = static_cast_with_check<ColumnFamilyHandleImpl>(cf);
+      auto cfd = cfh->cfd();
       SuperVersion* sv = cfd->GetReferencedSuperVersion(this);
-      cfd_to_sv.emplace_back(cfd, sv);
+      cfh_to_sv.emplace_back(cfh, sv);
       if (check_read_ts) {
         const Status s =
             FailIfReadCollapsedHistory(cfd, sv, *(read_options.timestamp));
         if (!s.ok()) {
-          for (auto prev_entry : cfd_to_sv) {
+          for (auto prev_entry : cfh_to_sv) {
             CleanupSuperVersion(std::get<1>(prev_entry));
           }
           return s;
         }
       }
     }
-    assert(cfd_to_sv.size() == column_families.size());
-    for (auto [cfd, sv] : cfd_to_sv) {
+    assert(cfh_to_sv.size() == column_families.size());
+    for (auto [cfh, sv] : cfh_to_sv) {
       iterators->push_back(
-          NewIteratorImpl(read_options, cfd, sv, read_seq, read_callback));
+          NewIteratorImpl(read_options, cfh, sv, read_seq, read_callback));
     }
   }
   return Status::OK();
@@ -677,7 +680,8 @@ Status DBImplSecondary::TryCatchUpWithPrimary() {
     InstrumentedMutexLock lock_guard(&mutex_);
     s = static_cast_with_check<ReactiveVersionSet>(versions_.get())
             ->ReadAndApply(&mutex_, &manifest_reader_,
-                           manifest_reader_status_.get(), &cfds_changed);
+                           manifest_reader_status_.get(), &cfds_changed,
+                           /*files_to_delete=*/nullptr);
 
     ROCKS_LOG_INFO(immutable_db_options_.info_log, "Last sequence is %" PRIu64,
                    static_cast<uint64_t>(versions_->LastSequence()));
@@ -802,7 +806,7 @@ Status DB::OpenAsSecondary(
   impl->mutex_.Lock();
   s = impl->Recover(column_families, true, false, false);
   if (s.ok()) {
-    for (auto cf : column_families) {
+    for (const auto& cf : column_families) {
       auto cfd =
           impl->versions_->GetColumnFamilySet()->GetColumnFamily(cf.name);
       if (nullptr == cfd) {

db/db_impl/db_impl_secondary.h

@@ -82,8 +82,9 @@ class DBImplSecondary : public DBImpl {
   // and log_readers_ to facilitate future operations.
   Status Recover(const std::vector<ColumnFamilyDescriptor>& column_families,
                  bool read_only, bool error_if_wal_file_exists,
-                 bool error_if_data_exists_in_wals, uint64_t* = nullptr,
-                 RecoveryContext* recovery_ctx = nullptr) override;
+                 bool error_if_data_exists_in_wals, bool is_retry = false,
+                 uint64_t* = nullptr, RecoveryContext* recovery_ctx = nullptr,
+                 bool* can_retry = nullptr) override;
 
   // Can return IOError due to files being deleted by the primary. To avoid
   // IOError in this case, application can coordinate between primary and
@@ -112,8 +113,8 @@ class DBImplSecondary : public DBImpl {
                         ColumnFamilyHandle* column_family) override;
 
   ArenaWrappedDBIter* NewIteratorImpl(const ReadOptions& read_options,
-                                      ColumnFamilyData* cfd, SuperVersion* sv,
-                                      SequenceNumber snapshot,
+                                      ColumnFamilyHandleImpl* cfh,
+                                      SuperVersion* sv, SequenceNumber snapshot,
                                       ReadCallback* read_callback,
                                       bool expose_blob_index = false,
                                       bool allow_refresh = true);
@@ -276,6 +277,10 @@ class DBImplSecondary : public DBImpl {
     return false;
   }
 
+  std::unique_ptr<log::FragmentBufferedReader> manifest_reader_;
+  std::unique_ptr<log::Reader::Reporter> manifest_reporter_;
+  std::unique_ptr<Status> manifest_reader_status_;
+
  private:
   friend class DB;
 
@@ -304,10 +309,6 @@ class DBImplSecondary : public DBImpl {
                                     const CompactionServiceInput& input,
                                     CompactionServiceResult* result);
 
-  std::unique_ptr<log::FragmentBufferedReader> manifest_reader_;
-  std::unique_ptr<log::Reader::Reporter> manifest_reporter_;
-  std::unique_ptr<Status> manifest_reader_status_;
-
   // Cache log readers for each log number, used for continue WAL replay
   // after recovery
   std::map<uint64_t, std::unique_ptr<LogReaderContainer>> log_readers_;

db/db_impl/db_impl_write.cc

@@ -220,6 +220,15 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
              write_options.protection_bytes_per_key != 8) {
     return Status::InvalidArgument(
         "`WriteOptions::protection_bytes_per_key` must be zero or eight");
+  } else if (write_options.disableWAL &&
+             immutable_db_options_.recycle_log_file_num > 0 &&
+             !(two_write_queues_ && disable_memtable)) {
+    // Corruption detection in recycled WALs relies on sequential sequence
+    // numbers, but WritePreparedTxnDB uses disableWAL internally for split
+    // writes
+    return Status::InvalidArgument(
+        "WriteOptions::disableWAL option is not supported if "
+        "DBOptions::recycle_log_file_num > 0");
   }
   // TODO: this use of operator bool on `tracer_` can avoid unnecessary lock
   // grabs but does not seem thread-safe.
@@ -325,7 +334,7 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
 
     if (w.ShouldWriteToMemtable()) {
       PERF_TIMER_STOP(write_pre_and_post_process_time);
-      PERF_TIMER_GUARD(write_memtable_time);
+      PERF_TIMER_FOR_WAIT_GUARD(write_memtable_time);
 
       ColumnFamilyMemTablesImpl column_family_memtables(
           versions_->GetColumnFamilySet());
@@ -557,7 +566,7 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
     }
 
     if (status.ok()) {
-      PERF_TIMER_GUARD(write_memtable_time);
+      PERF_TIMER_FOR_WAIT_GUARD(write_memtable_time);
 
       if (!parallel) {
         // w.sequence will be set inside InsertInto
@@ -797,7 +806,7 @@ Status DBImpl::PipelinedWriteImpl(const WriteOptions& write_options,
   WriteThread::WriteGroup memtable_write_group;
 
   if (w.state == WriteThread::STATE_MEMTABLE_WRITER_LEADER) {
-    PERF_TIMER_GUARD(write_memtable_time);
+    PERF_TIMER_FOR_WAIT_GUARD(write_memtable_time);
     assert(w.ShouldWriteToMemtable());
     write_thread_.EnterAsMemTableWriter(&w, &memtable_write_group);
     if (memtable_write_group.size > 1 &&
@@ -820,7 +829,7 @@ Status DBImpl::PipelinedWriteImpl(const WriteOptions& write_options,
 
   if (w.state == WriteThread::STATE_PARALLEL_MEMTABLE_WRITER) {
     PERF_TIMER_STOP(write_pre_and_post_process_time);
-    PERF_TIMER_GUARD(write_memtable_time);
+    PERF_TIMER_FOR_WAIT_GUARD(write_memtable_time);
 
     assert(w.ShouldWriteToMemtable());
     ColumnFamilyMemTablesImpl column_family_memtables(
@@ -868,7 +877,7 @@ Status DBImpl::UnorderedWriteMemtable(const WriteOptions& write_options,
     RecordTick(stats_, NUMBER_KEYS_WRITTEN, total_count);
 
     PERF_TIMER_STOP(write_pre_and_post_process_time);
-    PERF_TIMER_GUARD(write_memtable_time);
+    PERF_TIMER_FOR_WAIT_GUARD(write_memtable_time);
 
     ColumnFamilyMemTablesImpl column_family_memtables(
         versions_->GetColumnFamilySet());
@@ -954,7 +963,7 @@ Status DBImpl::WriteImplWALOnly(
     }
   } else {
     PERF_TIMER_STOP(write_pre_and_post_process_time);
-    PERF_TIMER_GUARD(write_delay_time);
+    PERF_TIMER_FOR_WAIT_GUARD(write_delay_time);
     InstrumentedMutexLock lock(&mutex_);
     Status status =
         DelayWrite(/*num_bytes=*/0ull, *write_thread, write_options);
@@ -1228,7 +1237,7 @@ Status DBImpl::PreprocessWrite(const WriteOptions& write_options,
   if (UNLIKELY(status.ok() && (write_controller_.IsStopped() ||
                                write_controller_.NeedsDelay()))) {
     PERF_TIMER_STOP(write_pre_and_post_process_time);
-    PERF_TIMER_GUARD(write_delay_time);
+    PERF_TIMER_FOR_WAIT_GUARD(write_delay_time);
     // We don't know size of curent batch so that we always use the size
     // for previous one. It might create a fairness issue that expiration
     // might happen for smaller writes but larger writes can go through.
@@ -2000,7 +2009,7 @@ Status DBImpl::ThrottleLowPriWritesIfNeeded(const WriteOptions& write_options,
       // is that in case the write is heavy, low pri writes may never have
       // a chance to run. Now we guarantee we are still slowly making
       // progress.
-      PERF_TIMER_GUARD(write_delay_time);
+      PERF_TIMER_FOR_WAIT_GUARD(write_delay_time);
       auto data_size = my_batch->GetDataSize();
       while (data_size > 0) {
         size_t allowed = write_controller_.low_pri_rate_limiter()->RequestToken(
@@ -2135,7 +2144,7 @@ void DBImpl::NotifyOnMemTableSealed(ColumnFamilyData* /*cfd*/,
   }
 
   mutex_.Unlock();
-  for (auto listener : immutable_db_options_.listeners) {
+  for (const auto& listener : immutable_db_options_.listeners) {
     listener->OnMemTableSealed(mem_table_info);
   }
   mutex_.Lock();
@@ -2147,6 +2156,8 @@ void DBImpl::NotifyOnMemTableSealed(ColumnFamilyData* /*cfd*/,
 // two_write_queues_ is true (This is to simplify the reasoning.)
 Status DBImpl::SwitchMemtable(ColumnFamilyData* cfd, WriteContext* context) {
   mutex_.AssertHeld();
+  assert(lock_wal_count_ == 0);
+
   // TODO: plumb Env::IOActivity, Env::IOPriority
   const ReadOptions read_options;
   const WriteOptions write_options;
@@ -2173,8 +2184,10 @@ Status DBImpl::SwitchMemtable(ColumnFamilyData* cfd, WriteContext* context) {
     log_write_mutex_.Unlock();
   }
   uint64_t recycle_log_number = 0;
+  // If file deletion is disabled, don't recycle logs since it'll result in
+  // the file getting renamed
   if (creating_new_log && immutable_db_options_.recycle_log_file_num &&
-      !log_recycle_files_.empty()) {
+      !log_recycle_files_.empty() && IsFileDeletionsEnabled()) {
     recycle_log_number = log_recycle_files_.front();
   }
   uint64_t new_log_number =
@@ -2252,7 +2265,7 @@ Status DBImpl::SwitchMemtable(ColumnFamilyData* cfd, WriteContext* context) {
       log_empty_ = true;
       log_dir_synced_ = false;
       logs_.emplace_back(logfile_number_, new_log);
-      alive_log_files_.push_back(LogFileNumberSize(logfile_number_));
+      alive_log_files_.emplace_back(logfile_number_);
     }
   }
 

db/db_io_failure_test.cc

@@ -13,6 +13,124 @@
 #include "util/random.h"
 
 namespace ROCKSDB_NAMESPACE {
+namespace {
+// A wrapper that allows injection of errors.
+class CorruptionFS : public FileSystemWrapper {
+ public:
+  bool writable_file_error_;
+  int num_writable_file_errors_;
+
+  explicit CorruptionFS(const std::shared_ptr<FileSystem>& _target,
+                        bool fs_buffer, bool verify_read)
+      : FileSystemWrapper(_target),
+        writable_file_error_(false),
+        num_writable_file_errors_(0),
+        corruption_trigger_(INT_MAX),
+        read_count_(0),
+        rnd_(300),
+        fs_buffer_(fs_buffer),
+        verify_read_(verify_read) {}
+  ~CorruptionFS() override {
+    // Assert that the corruption was reset, which means it got triggered
+    assert(corruption_trigger_ == INT_MAX);
+  }
+  const char* Name() const override { return "ErrorEnv"; }
+
+  IOStatus NewWritableFile(const std::string& fname, const FileOptions& opts,
+                           std::unique_ptr<FSWritableFile>* result,
+                           IODebugContext* dbg) override {
+    result->reset();
+    if (writable_file_error_) {
+      ++num_writable_file_errors_;
+      return IOStatus::IOError(fname, "fake error");
+    }
+    return target()->NewWritableFile(fname, opts, result, dbg);
+  }
+
+  void SetCorruptionTrigger(const int trigger) {
+    corruption_trigger_ = trigger;
+    read_count_ = 0;
+  }
+
+  IOStatus NewRandomAccessFile(const std::string& fname,
+                               const FileOptions& opts,
+                               std::unique_ptr<FSRandomAccessFile>* result,
+                               IODebugContext* dbg) override {
+    class CorruptionRandomAccessFile : public FSRandomAccessFileOwnerWrapper {
+     public:
+      CorruptionRandomAccessFile(CorruptionFS& fs,
+                                 std::unique_ptr<FSRandomAccessFile>& file)
+          : FSRandomAccessFileOwnerWrapper(std::move(file)), fs_(fs) {}
+
+      IOStatus Read(uint64_t offset, size_t len, const IOOptions& opts,
+                    Slice* result, char* scratch,
+                    IODebugContext* dbg) const override {
+        IOStatus s = target()->Read(offset, len, opts, result, scratch, dbg);
+        if (opts.verify_and_reconstruct_read) {
+          return s;
+        }
+        if (s.ok() && ++fs_.read_count_ >= fs_.corruption_trigger_) {
+          fs_.read_count_ = 0;
+          fs_.corruption_trigger_ = INT_MAX;
+          char* data = const_cast<char*>(result->data());
+          std::memcpy(
+              data,
+              fs_.rnd_.RandomString(static_cast<int>(result->size())).c_str(),
+              result->size());
+        }
+        return s;
+      }
+
+      IOStatus MultiRead(FSReadRequest* reqs, size_t num_reqs,
+                         const IOOptions& options,
+                         IODebugContext* dbg) override {
+        for (size_t i = 0; i < num_reqs; ++i) {
+          FSReadRequest& req = reqs[i];
+          if (fs_.fs_buffer_) {
+            FSAllocationPtr buffer(new char[req.len], [](void* ptr) {
+              delete[] static_cast<char*>(ptr);
+            });
+            req.fs_scratch = std::move(buffer);
+            req.status = Read(req.offset, req.len, options, &req.result,
+                              static_cast<char*>(req.fs_scratch.get()), dbg);
+          } else {
+            req.status = Read(req.offset, req.len, options, &req.result,
+                              req.scratch, dbg);
+          }
+        }
+        return IOStatus::OK();
+      }
+
+     private:
+      CorruptionFS& fs_;
+    };
+
+    std::unique_ptr<FSRandomAccessFile> file;
+    IOStatus s = target()->NewRandomAccessFile(fname, opts, &file, dbg);
+    EXPECT_OK(s);
+    result->reset(new CorruptionRandomAccessFile(*this, file));
+
+    return s;
+  }
+
+  void SupportedOps(int64_t& supported_ops) override {
+    supported_ops = 1 << FSSupportedOps::kAsyncIO;
+    if (fs_buffer_) {
+      supported_ops |= 1 << FSSupportedOps::kFSBuffer;
+    }
+    if (verify_read_) {
+      supported_ops |= 1 << FSSupportedOps::kVerifyAndReconstructRead;
+    }
+  }
+
+ private:
+  int corruption_trigger_;
+  int read_count_;
+  Random rnd_;
+  bool fs_buffer_;
+  bool verify_read_;
+};
+}  // anonymous namespace
 
 class DBIOFailureTest : public DBTestBase {
  public:
@@ -579,6 +697,182 @@ TEST_F(DBIOFailureTest, CompactionSstSyncError) {
   ASSERT_EQ("bar3", Get(1, "foo"));
 }
 #endif  // !(defined NDEBUG) || !defined(OS_WIN)
+
+class DBIOCorruptionTest
+    : public DBIOFailureTest,
+      public testing::WithParamInterface<std::tuple<bool, bool, bool>> {
+ public:
+  DBIOCorruptionTest() : DBIOFailureTest() {
+    BlockBasedTableOptions bbto;
+    options_ = CurrentOptions();
+
+    base_env_ = env_;
+    EXPECT_NE(base_env_, nullptr);
+    fs_.reset(new CorruptionFS(base_env_->GetFileSystem(),
+                               std::get<0>(GetParam()),
+                               std::get<2>(GetParam())));
+    env_guard_ = NewCompositeEnv(fs_);
+    options_.env = env_guard_.get();
+    bbto.num_file_reads_for_auto_readahead = 0;
+    options_.table_factory.reset(NewBlockBasedTableFactory(bbto));
+    options_.disable_auto_compactions = true;
+
+    Reopen(options_);
+  }
+
+  ~DBIOCorruptionTest() {
+    Close();
+    db_ = nullptr;
+  }
+
+  Status ReopenDB() { return TryReopen(options_); }
+
+ protected:
+  std::unique_ptr<Env> env_guard_;
+  std::shared_ptr<CorruptionFS> fs_;
+  Env* base_env_;
+  Options options_;
+};
+
+TEST_P(DBIOCorruptionTest, GetReadCorruptionRetry) {
+  CorruptionFS* fs =
+      static_cast<CorruptionFS*>(env_guard_->GetFileSystem().get());
+
+  ASSERT_OK(Put("key1", "val1"));
+  ASSERT_OK(Flush());
+  fs->SetCorruptionTrigger(1);
+
+  std::string val;
+  ReadOptions ro;
+  ro.async_io = std::get<1>(GetParam());
+  Status s = dbfull()->Get(ReadOptions(), "key1", &val);
+  if (std::get<2>(GetParam())) {
+    ASSERT_OK(s);
+    ASSERT_EQ(val, "val1");
+  } else {
+    ASSERT_TRUE(s.IsCorruption());
+  }
+}
+
+TEST_P(DBIOCorruptionTest, IterReadCorruptionRetry) {
+  CorruptionFS* fs =
+      static_cast<CorruptionFS*>(env_guard_->GetFileSystem().get());
+
+  ASSERT_OK(Put("key1", "val1"));
+  ASSERT_OK(Flush());
+  fs->SetCorruptionTrigger(1);
+
+  ReadOptions ro;
+  ro.readahead_size = 65536;
+  ro.async_io = std::get<1>(GetParam());
+
+  Iterator* iter = dbfull()->NewIterator(ro);
+  iter->SeekToFirst();
+  while (iter->status().ok() && iter->Valid()) {
+    iter->Next();
+  }
+  if (std::get<2>(GetParam())) {
+    ASSERT_OK(iter->status());
+  } else {
+    ASSERT_TRUE(iter->status().IsCorruption());
+  }
+  delete iter;
+}
+
+TEST_P(DBIOCorruptionTest, MultiGetReadCorruptionRetry) {
+  CorruptionFS* fs =
+      static_cast<CorruptionFS*>(env_guard_->GetFileSystem().get());
+
+  ASSERT_OK(Put("key1", "val1"));
+  ASSERT_OK(Put("key2", "val2"));
+  ASSERT_OK(Flush());
+  fs->SetCorruptionTrigger(1);
+
+  std::vector<std::string> keystr{"key1", "key2"};
+  std::vector<Slice> keys{Slice(keystr[0]), Slice(keystr[1])};
+  std::vector<PinnableSlice> values(keys.size());
+  std::vector<Status> statuses(keys.size());
+  ReadOptions ro;
+  ro.async_io = std::get<1>(GetParam());
+  dbfull()->MultiGet(ro, dbfull()->DefaultColumnFamily(), keys.size(),
+                     keys.data(), values.data(), statuses.data());
+  if (std::get<2>(GetParam())) {
+    ASSERT_EQ(values[0].ToString(), "val1");
+    ASSERT_EQ(values[1].ToString(), "val2");
+  } else {
+    ASSERT_TRUE(statuses[0].IsCorruption());
+    ASSERT_TRUE(statuses[1].IsCorruption());
+  }
+}
+
+TEST_P(DBIOCorruptionTest, CompactionReadCorruptionRetry) {
+  CorruptionFS* fs =
+      static_cast<CorruptionFS*>(env_guard_->GetFileSystem().get());
+
+  ASSERT_OK(Put("key1", "val1"));
+  ASSERT_OK(Put("key3", "val3"));
+  ASSERT_OK(Flush());
+  ASSERT_OK(Put("key2", "val2"));
+  ASSERT_OK(Flush());
+  fs->SetCorruptionTrigger(1);
+  Status s = dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr);
+  if (std::get<2>(GetParam())) {
+    ASSERT_OK(s);
+
+    std::string val;
+    ReadOptions ro;
+    ro.async_io = std::get<1>(GetParam());
+    ASSERT_OK(dbfull()->Get(ro, "key1", &val));
+    ASSERT_EQ(val, "val1");
+  } else {
+    ASSERT_TRUE(s.IsCorruption());
+  }
+}
+
+TEST_P(DBIOCorruptionTest, FlushReadCorruptionRetry) {
+  CorruptionFS* fs =
+      static_cast<CorruptionFS*>(env_guard_->GetFileSystem().get());
+
+  ASSERT_OK(Put("key1", "val1"));
+  fs->SetCorruptionTrigger(1);
+  Status s = Flush();
+  if (std::get<2>(GetParam())) {
+    ASSERT_OK(s);
+
+    std::string val;
+    ReadOptions ro;
+    ro.async_io = std::get<1>(GetParam());
+    ASSERT_OK(dbfull()->Get(ro, "key1", &val));
+    ASSERT_EQ(val, "val1");
+  } else {
+    ASSERT_NOK(s);
+  }
+}
+
+TEST_P(DBIOCorruptionTest, ManifestCorruptionRetry) {
+  CorruptionFS* fs =
+      static_cast<CorruptionFS*>(env_guard_->GetFileSystem().get());
+
+  ASSERT_OK(Put("key1", "val1"));
+  ASSERT_OK(Flush());
+  SyncPoint::GetInstance()->SetCallBack(
+      "VersionSet::Recover:StartManifestRead",
+      [&](void* /*arg*/) { fs->SetCorruptionTrigger(0); });
+  SyncPoint::GetInstance()->EnableProcessing();
+
+  if (std::get<2>(GetParam())) {
+    ASSERT_OK(ReopenDB());
+  } else {
+    ASSERT_EQ(ReopenDB(), Status::Corruption());
+  }
+  SyncPoint::GetInstance()->DisableProcessing();
+}
+
+// The parameters are - 1. Use FS provided buffer, 2. Use async IO ReadOption,
+// 3. Retry with verify_and_reconstruct_read IOOption
+INSTANTIATE_TEST_CASE_P(DBIOCorruptionTest, DBIOCorruptionTest,
+                        testing::Combine(testing::Bool(), testing::Bool(),
+                                         testing::Bool()));
 }  // namespace ROCKSDB_NAMESPACE
 
 int main(int argc, char** argv) {

db/db_iter.cc

@@ -43,8 +43,8 @@ DBIter::DBIter(Env* _env, const ReadOptions& read_options,
                const Comparator* cmp, InternalIterator* iter,
                const Version* version, SequenceNumber s, bool arena_mode,
                uint64_t max_sequential_skip_in_iterations,
-               ReadCallback* read_callback, DBImpl* db_impl,
-               ColumnFamilyData* cfd, bool expose_blob_index)
+               ReadCallback* read_callback, ColumnFamilyHandleImpl* cfh,
+               bool expose_blob_index)
     : prefix_extractor_(mutable_cf_options.prefix_extractor.get()),
       env_(_env),
       clock_(ioptions.clock),
@@ -79,8 +79,7 @@ DBIter::DBIter(Env* _env, const ReadOptions& read_options,
       is_blob_(false),
       arena_mode_(arena_mode),
       io_activity_(read_options.io_activity),
-      db_impl_(db_impl),
-      cfd_(cfd),
+      cfh_(cfh),
       timestamp_ub_(read_options.timestamp),
       timestamp_lb_(read_options.iter_start_ts),
       timestamp_size_(timestamp_ub_ ? timestamp_ub_->size() : 0) {
@@ -110,12 +109,22 @@ Status DBIter::GetProperty(std::string prop_name, std::string* prop) {
       *prop = "Iterator is not valid.";
     }
     return Status::OK();
+  } else if (prop_name == "rocksdb.iterator.is-value-pinned") {
+    if (valid_) {
+      *prop = (pin_thru_lifetime_ && iter_.Valid() &&
+               iter_.value().data() == value_.data())
+                  ? "1"
+                  : "0";
+    } else {
+      *prop = "Iterator is not valid.";
+    }
+    return Status::OK();
   } else if (prop_name == "rocksdb.iterator.internal-key") {
     *prop = saved_key_.GetUserKey().ToString();
     return Status::OK();
   } else if (prop_name == "rocksdb.iterator.write-time") {
-    // TODO(yuzhangyu): implement return the actual write time.
-    return Status::NotSupported("write time property is under construction");
+    PutFixed64(prop, saved_write_unix_time_);
+    return Status::OK();
   }
   return Status::InvalidArgument("Unidentified property.");
 }
@@ -232,6 +241,7 @@ bool DBIter::SetValueAndColumnsFromEntity(Slice slice) {
   if (!s.ok()) {
     status_ = s;
     valid_ = false;
+    wide_columns_.clear();
     return false;
   }
 
@@ -393,6 +403,7 @@ bool DBIter::FindNextUserEntryInternal(bool skipping_saved_key,
             }
             break;
           case kTypeValue:
+          case kTypeValuePreferredSeqno:
           case kTypeBlobIndex:
           case kTypeWideColumnEntity:
             if (!PrepareValue()) {
@@ -418,8 +429,14 @@ bool DBIter::FindNextUserEntryInternal(bool skipping_saved_key,
                 return false;
               }
             } else {
-              assert(ikey_.type == kTypeValue);
-              SetValueAndColumnsFromPlain(iter_.value());
+              assert(ikey_.type == kTypeValue ||
+                     ikey_.type == kTypeValuePreferredSeqno);
+              Slice value = iter_.value();
+              saved_write_unix_time_ = iter_.write_unix_time();
+              if (ikey_.type == kTypeValuePreferredSeqno) {
+                value = ParsePackedValueForValue(value);
+              }
+              SetValueAndColumnsFromPlain(value);
             }
 
             valid_ = true;
@@ -575,10 +592,15 @@ bool DBIter::MergeValuesNewToOld() {
       return false;
     }
 
-    if (kTypeValue == ikey.type) {
-      // hit a put, merge the put value with operands and store the
-      // final result in saved_value_. We are done!
-      if (!MergeWithPlainBaseValue(iter_.value(), ikey.user_key)) {
+    if (kTypeValue == ikey.type || kTypeValuePreferredSeqno == ikey.type) {
+      Slice value = iter_.value();
+      saved_write_unix_time_ = iter_.write_unix_time();
+      if (kTypeValuePreferredSeqno == ikey.type) {
+        value = ParsePackedValueForValue(value);
+      }
+      // hit a put or put equivalent, merge the put value with operands and
+      // store the final result in saved_value_. We are done!
+      if (!MergeWithPlainBaseValue(value, ikey.user_key)) {
         return false;
       }
       // iter_ is positioned after put
@@ -701,16 +723,21 @@ bool DBIter::ReverseToForward() {
   // not exist or may have different prefix than the current key().
   // If that's the case, seek iter_ to current key.
   if (!expect_total_order_inner_iter() || !iter_.Valid()) {
-    IterKey last_key;
-    ParsedInternalKey pikey(saved_key_.GetUserKey(), kMaxSequenceNumber,
-                            kValueTypeForSeek);
-    if (timestamp_size_ > 0) {
+    std::string last_key;
+    if (timestamp_size_ == 0) {
+      AppendInternalKey(
+          &last_key, ParsedInternalKey(saved_key_.GetUserKey(),
+                                       kMaxSequenceNumber, kValueTypeForSeek));
+    } else {
       // TODO: pre-create kTsMax.
       const std::string kTsMax(timestamp_size_, '\xff');
-      pikey.SetTimestamp(kTsMax);
+      AppendInternalKeyWithDifferentTimestamp(
+          &last_key,
+          ParsedInternalKey(saved_key_.GetUserKey(), kMaxSequenceNumber,
+                            kValueTypeForSeek),
+          kTsMax);
     }
-    last_key.SetInternalKey(pikey);
-    iter_.Seek(last_key.GetInternalKey());
+    iter_.Seek(last_key);
     RecordTick(statistics_, NUMBER_OF_RESEEKS_IN_ITERATION);
   }
 
@@ -840,8 +867,8 @@ bool DBIter::FindValueForCurrentKey() {
   merge_context_.Clear();
   current_entry_is_merged_ = false;
   // last entry before merge (could be kTypeDeletion,
-  // kTypeDeletionWithTimestamp, kTypeSingleDeletion, kTypeValue,
-  // kTypeBlobIndex, or kTypeWideColumnEntity)
+  // kTypeDeletionWithTimestamp, kTypeSingleDeletion, kTypeValue
+  // kTypeBlobIndex, kTypeWideColumnEntity or kTypeValuePreferredSeqno)
   ValueType last_not_merge_type = kTypeDeletion;
   ValueType last_key_entry_type = kTypeDeletion;
 
@@ -918,10 +945,16 @@ bool DBIter::FindValueForCurrentKey() {
     last_key_entry_type = ikey.type;
     switch (last_key_entry_type) {
       case kTypeValue:
+      case kTypeValuePreferredSeqno:
       case kTypeBlobIndex:
       case kTypeWideColumnEntity:
         if (iter_.iter()->IsValuePinned()) {
-          pinned_value_ = iter_.value();
+          saved_write_unix_time_ = iter_.write_unix_time();
+          if (last_key_entry_type == kTypeValuePreferredSeqno) {
+            pinned_value_ = ParsePackedValueForValue(iter_.value());
+          } else {
+            pinned_value_ = iter_.value();
+          }
         } else {
           valid_ = false;
           status_ = Status::NotSupported(
@@ -1031,7 +1064,8 @@ bool DBIter::FindValueForCurrentKey() {
 
         return true;
       } else {
-        assert(last_not_merge_type == kTypeValue);
+        assert(last_not_merge_type == kTypeValue ||
+               last_not_merge_type == kTypeValuePreferredSeqno);
         if (!MergeWithPlainBaseValue(pinned_value_, saved_key_.GetUserKey())) {
           return false;
         }
@@ -1039,6 +1073,7 @@ bool DBIter::FindValueForCurrentKey() {
       }
       break;
     case kTypeValue:
+    case kTypeValuePreferredSeqno:
       SetValueAndColumnsFromPlain(pinned_value_);
 
       break;
@@ -1143,10 +1178,15 @@ bool DBIter::FindValueForCurrentKeyUsingSeek() {
     Slice ts = ExtractTimestampFromUserKey(ikey.user_key, timestamp_size_);
     saved_timestamp_.assign(ts.data(), ts.size());
   }
-  if (ikey.type == kTypeValue || ikey.type == kTypeBlobIndex ||
-      ikey.type == kTypeWideColumnEntity) {
+  if (ikey.type == kTypeValue || ikey.type == kTypeValuePreferredSeqno ||
+      ikey.type == kTypeBlobIndex || ikey.type == kTypeWideColumnEntity) {
     assert(iter_.iter()->IsValuePinned());
-    pinned_value_ = iter_.value();
+    saved_write_unix_time_ = iter_.write_unix_time();
+    if (ikey.type == kTypeValuePreferredSeqno) {
+      pinned_value_ = ParsePackedValueForValue(iter_.value());
+    } else {
+      pinned_value_ = iter_.value();
+    }
     if (ikey.type == kTypeBlobIndex) {
       if (!SetBlobValueIfNeeded(ikey.user_key, pinned_value_)) {
         return false;
@@ -1159,7 +1199,7 @@ bool DBIter::FindValueForCurrentKeyUsingSeek() {
         return false;
       }
     } else {
-      assert(ikey.type == kTypeValue);
+      assert(ikey.type == kTypeValue || ikey.type == kTypeValuePreferredSeqno);
       SetValueAndColumnsFromPlain(pinned_value_);
     }
 
@@ -1205,8 +1245,12 @@ bool DBIter::FindValueForCurrentKeyUsingSeek() {
       return false;
     }
 
-    if (ikey.type == kTypeValue) {
-      if (!MergeWithPlainBaseValue(iter_.value(), saved_key_.GetUserKey())) {
+    if (ikey.type == kTypeValue || ikey.type == kTypeValuePreferredSeqno) {
+      Slice value = iter_.value();
+      if (ikey.type == kTypeValuePreferredSeqno) {
+        value = ParsePackedValueForValue(value);
+      }
+      if (!MergeWithPlainBaseValue(value, saved_key_.GetUserKey())) {
         return false;
       }
       return true;
@@ -1343,18 +1387,23 @@ bool DBIter::FindUserKeyBeforeSavedKey() {
 
     if (num_skipped >= max_skip_) {
       num_skipped = 0;
-      IterKey last_key;
-      ParsedInternalKey pikey(saved_key_.GetUserKey(), kMaxSequenceNumber,
-                              kValueTypeForSeek);
-      if (timestamp_size_ > 0) {
+      std::string last_key;
+      if (timestamp_size_ == 0) {
+        AppendInternalKey(&last_key, ParsedInternalKey(saved_key_.GetUserKey(),
+                                                       kMaxSequenceNumber,
+                                                       kValueTypeForSeek));
+      } else {
         // TODO: pre-create kTsMax.
         const std::string kTsMax(timestamp_size_, '\xff');
-        pikey.SetTimestamp(kTsMax);
+        AppendInternalKeyWithDifferentTimestamp(
+            &last_key,
+            ParsedInternalKey(saved_key_.GetUserKey(), kMaxSequenceNumber,
+                              kValueTypeForSeek),
+            kTsMax);
       }
-      last_key.SetInternalKey(pikey);
       // It would be more efficient to use SeekForPrev() here, but some
       // iterators may not support it.
-      iter_.Seek(last_key.GetInternalKey());
+      iter_.Seek(last_key);
       RecordTick(statistics_, NUMBER_OF_RESEEKS_IN_ITERATION);
       if (!iter_.Valid()) {
         break;
@@ -1459,7 +1508,7 @@ void DBIter::Seek(const Slice& target) {
   PERF_CPU_TIMER_GUARD(iter_seek_cpu_nanos, clock_);
   StopWatch sw(clock_, statistics_, DB_SEEK);
 
-  if (db_impl_ != nullptr && cfd_ != nullptr) {
+  if (cfh_ != nullptr) {
     // TODO: What do we do if this returns an error?
     Slice lower_bound, upper_bound;
     if (iterate_lower_bound_ != nullptr) {
@@ -1472,7 +1521,9 @@ void DBIter::Seek(const Slice& target) {
     } else {
       upper_bound = Slice("");
     }
-    db_impl_->TraceIteratorSeek(cfd_->GetID(), target, lower_bound, upper_bound)
+    cfh_->db()
+        ->TraceIteratorSeek(cfh_->cfd()->GetID(), target, lower_bound,
+                            upper_bound)
         .PermitUncheckedError();
   }
 
@@ -1533,7 +1584,7 @@ void DBIter::SeekForPrev(const Slice& target) {
   PERF_CPU_TIMER_GUARD(iter_seek_cpu_nanos, clock_);
   StopWatch sw(clock_, statistics_, DB_SEEK);
 
-  if (db_impl_ != nullptr && cfd_ != nullptr) {
+  if (cfh_ != nullptr) {
     // TODO: What do we do if this returns an error?
     Slice lower_bound, upper_bound;
     if (iterate_lower_bound_ != nullptr) {
@@ -1546,8 +1597,8 @@ void DBIter::SeekForPrev(const Slice& target) {
     } else {
       upper_bound = Slice("");
     }
-    db_impl_
-        ->TraceIteratorSeekForPrev(cfd_->GetID(), target, lower_bound,
+    cfh_->db()
+        ->TraceIteratorSeekForPrev(cfh_->cfd()->GetID(), target, lower_bound,
                                    upper_bound)
         .PermitUncheckedError();
   }
@@ -1711,13 +1762,12 @@ Iterator* NewDBIterator(Env* env, const ReadOptions& read_options,
                         InternalIterator* internal_iter, const Version* version,
                         const SequenceNumber& sequence,
                         uint64_t max_sequential_skip_in_iterations,
-                        ReadCallback* read_callback, DBImpl* db_impl,
-                        ColumnFamilyData* cfd, bool expose_blob_index) {
-  DBIter* db_iter =
-      new DBIter(env, read_options, ioptions, mutable_cf_options,
-                 user_key_comparator, internal_iter, version, sequence, false,
-                 max_sequential_skip_in_iterations, read_callback, db_impl, cfd,
-                 expose_blob_index);
+                        ReadCallback* read_callback,
+                        ColumnFamilyHandleImpl* cfh, bool expose_blob_index) {
+  DBIter* db_iter = new DBIter(
+      env, read_options, ioptions, mutable_cf_options, user_key_comparator,
+      internal_iter, version, sequence, false,
+      max_sequential_skip_in_iterations, read_callback, cfh, expose_blob_index);
   return db_iter;
 }
 

db/db_iter.h

@@ -118,7 +118,7 @@ class DBIter final : public Iterator {
          const MutableCFOptions& mutable_cf_options, const Comparator* cmp,
          InternalIterator* iter, const Version* version, SequenceNumber s,
          bool arena_mode, uint64_t max_sequential_skip_in_iterations,
-         ReadCallback* read_callback, DBImpl* db_impl, ColumnFamilyData* cfd,
+         ReadCallback* read_callback, ColumnFamilyHandleImpl* cfh,
          bool expose_blob_index);
 
   // No copying allowed
@@ -367,6 +367,12 @@ class DBIter final : public Iterator {
   // and should not be used across functions. Reusing this object can reduce
   // overhead of calling construction of the function if creating it each time.
   ParsedInternalKey ikey_;
+
+  // The approximate write time for the entry. It is deduced from the entry's
+  // sequence number if the seqno to time mapping is available. For a
+  // kTypeValuePreferredSeqno entry, this is the write time specified by the
+  // user.
+  uint64_t saved_write_unix_time_;
   std::string saved_value_;
   Slice pinned_value_;
   // for prefix seek mode to support prev()
@@ -417,8 +423,7 @@ class DBIter final : public Iterator {
   MergeContext merge_context_;
   LocalStatistics local_stats_;
   PinnedIteratorsManager pinned_iters_mgr_;
-  DBImpl* db_impl_;
-  ColumnFamilyData* cfd_;
+  ColumnFamilyHandleImpl* cfh_;
   const Slice* const timestamp_ub_;
   const Slice* const timestamp_lb_;
   const size_t timestamp_size_;
@@ -428,15 +433,12 @@ class DBIter final : public Iterator {
 // Return a new iterator that converts internal keys (yielded by
 // "*internal_iter") that were live at the specified `sequence` number
 // into appropriate user keys.
-Iterator* NewDBIterator(Env* env, const ReadOptions& read_options,
-                        const ImmutableOptions& ioptions,
-                        const MutableCFOptions& mutable_cf_options,
-                        const Comparator* user_key_comparator,
-                        InternalIterator* internal_iter, const Version* version,
-                        const SequenceNumber& sequence,
-                        uint64_t max_sequential_skip_in_iterations,
-                        ReadCallback* read_callback, DBImpl* db_impl = nullptr,
-                        ColumnFamilyData* cfd = nullptr,
-                        bool expose_blob_index = false);
+Iterator* NewDBIterator(
+    Env* env, const ReadOptions& read_options, const ImmutableOptions& ioptions,
+    const MutableCFOptions& mutable_cf_options,
+    const Comparator* user_key_comparator, InternalIterator* internal_iter,
+    const Version* version, const SequenceNumber& sequence,
+    uint64_t max_sequential_skip_in_iterations, ReadCallback* read_callback,
+    ColumnFamilyHandleImpl* cfh = nullptr, bool expose_blob_index = false);
 
 }  // namespace ROCKSDB_NAMESPACE

db/db_iter_test.cc

@@ -45,6 +45,14 @@ class TestIterator : public InternalIterator {
     Add(argkey, kTypeValue, argvalue);
   }
 
+  void AddTimedPut(std::string argkey, std::string argvalue,
+                   uint64_t write_unix_time) {
+    std::string packed_value;
+    [[maybe_unused]] auto packed_value_slice =
+        PackValueAndWriteTime(argvalue, write_unix_time, &packed_value);
+    Add(argkey, kTypeValuePreferredSeqno, packed_value);
+  }
+
   void AddDeletion(std::string argkey) {
     Add(argkey, kTypeDeletion, std::string());
   }
@@ -1388,6 +1396,60 @@ TEST_F(DBIteratorTest, DBIteratorSkipInternalKeys) {
   }
 }
 
+TEST_F(DBIteratorTest, DBIteratorTimedPutBasic) {
+  ReadOptions ro;
+  Options options;
+  options.merge_operator = MergeOperators::CreateFromStringId("stringappend");
+
+  TestIterator* internal_iter = new TestIterator(BytewiseComparator());
+  internal_iter->AddTimedPut("a", "0", /*write_unix_time=*/0);
+  internal_iter->AddMerge("a", "1");
+  internal_iter->AddTimedPut("b", "0", /*write_unix_time=*/0);
+  internal_iter->AddDeletion("b");
+  internal_iter->AddTimedPut("c", "01", /*write_unix_time=*/0);
+  internal_iter->AddTimedPut("c", "02", /*write_unix_time=*/0);
+  internal_iter->AddTimedPut("c", "2", /*write_unix_time=*/0);
+  internal_iter->AddTimedPut("d", "3", /*write_unix_time=*/0);
+  internal_iter->Finish();
+
+  std::unique_ptr<Iterator> db_iter(NewDBIterator(
+      env_, ro, ImmutableOptions(options), MutableCFOptions(options),
+      BytewiseComparator(), internal_iter, nullptr /* version */,
+      7 /* sequence */, /*max_sequential_skip_in_iterations*/ 1,
+      nullptr /* read_callback */));
+  db_iter->SeekToFirst();
+  ASSERT_TRUE(db_iter->Valid());
+  ASSERT_EQ(db_iter->key().ToString(), "a");
+  ASSERT_EQ(db_iter->value().ToString(), "0,1");
+  db_iter->Next();
+  ASSERT_TRUE(db_iter->Valid());
+  ASSERT_EQ(db_iter->key().ToString(), "c");
+  ASSERT_EQ(db_iter->value().ToString(), "2");
+  db_iter->Next();
+  ASSERT_TRUE(db_iter->Valid());
+  ASSERT_EQ(db_iter->key().ToString(), "d");
+  ASSERT_EQ(db_iter->value().ToString(), "3");
+  db_iter->Next();
+  ASSERT_FALSE(db_iter->Valid());
+  ASSERT_OK(db_iter->status());
+
+  db_iter->SeekToLast();
+  ASSERT_TRUE(db_iter->Valid());
+  ASSERT_EQ(db_iter->key().ToString(), "d");
+  ASSERT_EQ(db_iter->value().ToString(), "3");
+  db_iter->Prev();
+  ASSERT_TRUE(db_iter->Valid());
+  ASSERT_EQ(db_iter->key().ToString(), "c");
+  ASSERT_EQ(db_iter->value().ToString(), "2");
+  db_iter->Prev();
+  ASSERT_TRUE(db_iter->Valid());
+  ASSERT_EQ(db_iter->key().ToString(), "a");
+  ASSERT_EQ(db_iter->value().ToString(), "0,1");
+  db_iter->Prev();
+  ASSERT_FALSE(db_iter->Valid());
+  ASSERT_OK(db_iter->status());
+}
+
 TEST_F(DBIteratorTest, DBIterator1) {
   ReadOptions ro;
   Options options;

db/db_iterator_test.cc

@@ -93,8 +93,8 @@ class DBIteratorTest : public DBIteratorBaseTest,
     if (column_family == nullptr) {
       column_family = db_->DefaultColumnFamily();
     }
-    auto* cfd =
-        static_cast_with_check<ColumnFamilyHandleImpl>(column_family)->cfd();
+    auto* cfh = static_cast_with_check<ColumnFamilyHandleImpl>(column_family);
+    auto* cfd = cfh->cfd();
     SequenceNumber seq = read_options.snapshot != nullptr
                              ? read_options.snapshot->GetSequenceNumber()
                              : db_->GetLatestSequenceNumber();
@@ -109,7 +109,7 @@ class DBIteratorTest : public DBIteratorBaseTest,
     }
     DBImpl* db_impl = dbfull();
     SuperVersion* super_version = cfd->GetReferencedSuperVersion(db_impl);
-    return db_impl->NewIteratorImpl(read_options, cfd, super_version, seq,
+    return db_impl->NewIteratorImpl(read_options, cfh, super_version, seq,
                                     read_callback);
   }
 
@@ -133,15 +133,28 @@ TEST_P(DBIteratorTest, IteratorProperty) {
     ASSERT_NOK(iter->GetProperty("non_existing.value", &prop_value));
     ASSERT_OK(iter->GetProperty("rocksdb.iterator.is-key-pinned", &prop_value));
     ASSERT_EQ("0", prop_value);
+    ASSERT_OK(
+        iter->GetProperty("rocksdb.iterator.is-value-pinned", &prop_value));
+    ASSERT_EQ("0", prop_value);
     ASSERT_OK(iter->GetProperty("rocksdb.iterator.internal-key", &prop_value));
     ASSERT_EQ("1", prop_value);
     iter->Next();
     ASSERT_OK(iter->GetProperty("rocksdb.iterator.is-key-pinned", &prop_value));
     ASSERT_EQ("Iterator is not valid.", prop_value);
+    ASSERT_OK(
+        iter->GetProperty("rocksdb.iterator.is-value-pinned", &prop_value));
+    ASSERT_EQ("Iterator is not valid.", prop_value);
 
     // Get internal key at which the iteration stopped (tombstone in this case).
     ASSERT_OK(iter->GetProperty("rocksdb.iterator.internal-key", &prop_value));
     ASSERT_EQ("2", prop_value);
+
+    prop_value.clear();
+    ASSERT_OK(iter->GetProperty("rocksdb.iterator.write-time", &prop_value));
+    uint64_t write_time;
+    Slice prop_slice = prop_value;
+    ASSERT_TRUE(GetFixed64(&prop_slice, &write_time));
+    ASSERT_EQ(std::numeric_limits<uint64_t>::max(), write_time);
   }
   Close();
 }
@@ -1673,12 +1686,15 @@ TEST_P(DBIteratorTest, PinnedDataIteratorMultipleFiles) {
   ro.pin_data = true;
   auto iter = NewIterator(ro);
 
-  std::vector<std::pair<Slice, std::string>> results;
+  std::vector<std::pair<Slice, Slice>> results;
   for (iter->SeekToFirst(); iter->Valid(); iter->Next()) {
     std::string prop_value;
     ASSERT_OK(iter->GetProperty("rocksdb.iterator.is-key-pinned", &prop_value));
     ASSERT_EQ("1", prop_value);
-    results.emplace_back(iter->key(), iter->value().ToString());
+    ASSERT_OK(
+        iter->GetProperty("rocksdb.iterator.is-value-pinned", &prop_value));
+    ASSERT_EQ("1", prop_value);
+    results.emplace_back(iter->key(), iter->value());
   }
 
   ASSERT_EQ(results.size(), true_data.size());
@@ -1732,6 +1748,9 @@ TEST_P(DBIteratorTest, PinnedDataIteratorMergeOperator) {
     std::string prop_value;
     ASSERT_OK(iter->GetProperty("rocksdb.iterator.is-key-pinned", &prop_value));
     ASSERT_EQ("1", prop_value);
+    ASSERT_OK(
+        iter->GetProperty("rocksdb.iterator.is-value-pinned", &prop_value));
+    ASSERT_EQ("0", prop_value);
     results.emplace_back(iter->key(), iter->value().ToString());
   }
   ASSERT_OK(iter->status());
@@ -1785,12 +1804,15 @@ TEST_P(DBIteratorTest, PinnedDataIteratorReadAfterUpdate) {
     }
   }
 
-  std::vector<std::pair<Slice, std::string>> results;
+  std::vector<std::pair<Slice, Slice>> results;
   for (iter->SeekToFirst(); iter->Valid(); iter->Next()) {
     std::string prop_value;
     ASSERT_OK(iter->GetProperty("rocksdb.iterator.is-key-pinned", &prop_value));
     ASSERT_EQ("1", prop_value);
-    results.emplace_back(iter->key(), iter->value().ToString());
+    ASSERT_OK(
+        iter->GetProperty("rocksdb.iterator.is-value-pinned", &prop_value));
+    ASSERT_EQ("1", prop_value);
+    results.emplace_back(iter->key(), iter->value());
   }
   ASSERT_OK(iter->status());
 
@@ -3206,13 +3228,13 @@ TEST_F(DBIteratorWithReadCallbackTest, ReadCallback) {
   ASSERT_OK(Put("bar", "v7"));
 
   SequenceNumber seq2 = db_->GetLatestSequenceNumber();
-  auto* cfd =
-      static_cast_with_check<ColumnFamilyHandleImpl>(db_->DefaultColumnFamily())
-          ->cfd();
+  auto* cfh = static_cast_with_check<ColumnFamilyHandleImpl>(
+      db_->DefaultColumnFamily());
+  auto* cfd = cfh->cfd();
   // The iterator are suppose to see data before seq1.
   DBImpl* db_impl = dbfull();
   SuperVersion* super_version = cfd->GetReferencedSuperVersion(db_impl);
-  Iterator* iter = db_impl->NewIteratorImpl(ReadOptions(), cfd, super_version,
+  Iterator* iter = db_impl->NewIteratorImpl(ReadOptions(), cfh, super_version,
                                             seq2, &callback1);
 
   // Seek
@@ -3292,7 +3314,7 @@ TEST_F(DBIteratorWithReadCallbackTest, ReadCallback) {
 
   // The iterator is suppose to see data before seq3.
   super_version = cfd->GetReferencedSuperVersion(db_impl);
-  iter = db_impl->NewIteratorImpl(ReadOptions(), cfd, super_version, seq4,
+  iter = db_impl->NewIteratorImpl(ReadOptions(), cfh, super_version, seq4,
                                   &callback2);
   // Seek to "z", which is visible.
   iter->Seek("z");
@@ -3548,6 +3570,121 @@ TEST_F(DBIteratorTest, ErrorWhenReadFile) {
   iter->Reset();
 }
 
+TEST_F(DBIteratorTest, IteratorsConsistentViewImplicitSnapshot) {
+  Options options = GetDefaultOptions();
+  CreateAndReopenWithCF({"cf_1", "cf_2"}, options);
+
+  for (int i = 0; i < 3; ++i) {
+    ASSERT_OK(Put(i, "cf" + std::to_string(i) + "_key",
+                  "cf" + std::to_string(i) + "_val"));
+  }
+
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->LoadDependency(
+      {{"DBImpl::BGWorkFlush:done",
+        "DBImpl::MultiCFSnapshot::BeforeCheckingSnapshot"}});
+
+  bool flushed = false;
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
+      "DBImpl::MultiCFSnapshot::AfterRefSV", [&](void* /*arg*/) {
+        if (!flushed) {
+          for (int i = 0; i < 3; ++i) {
+            ASSERT_OK(Put(i, "cf" + std::to_string(i) + "_key",
+                          "cf" + std::to_string(i) + "_val_new"));
+          }
+          // After SV is obtained for the first CF, flush for the second CF
+          ASSERT_OK(Flush(1));
+          flushed = true;
+        }
+      });
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
+  ReadOptions read_options;
+  std::vector<Iterator*> iters;
+  ASSERT_OK(db_->NewIterators(read_options, handles_, &iters));
+
+  for (int i = 0; i < 3; ++i) {
+    auto iter = iters[i];
+    ASSERT_OK(iter->status());
+    iter->SeekToFirst();
+    ASSERT_EQ(IterStatus(iter), "cf" + std::to_string(i) + "_key->cf" +
+                                    std::to_string(i) + "_val_new");
+  }
+  for (auto* iter : iters) {
+    delete iter;
+  }
+
+  // Thread-local SVs are no longer obsolete nor in use
+  for (int i = 0; i < 3; ++i) {
+    auto* cfd =
+        static_cast_with_check<ColumnFamilyHandleImpl>(handles_[i])->cfd();
+    ASSERT_NE(cfd->TEST_GetLocalSV()->Get(), SuperVersion::kSVObsolete);
+    ASSERT_NE(cfd->TEST_GetLocalSV()->Get(), SuperVersion::kSVInUse);
+  }
+}
+
+TEST_F(DBIteratorTest, IteratorsConsistentViewExplicitSnapshot) {
+  Options options = GetDefaultOptions();
+  options.atomic_flush = true;
+  CreateAndReopenWithCF({"cf_1", "cf_2"}, options);
+
+  for (int i = 0; i < 3; ++i) {
+    ASSERT_OK(Put(i, "cf" + std::to_string(i) + "_key",
+                  "cf" + std::to_string(i) + "_val"));
+  }
+
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->LoadDependency(
+      {{"DBImpl::BGWorkFlush:done",
+        "DBImpl::MultiCFSnapshot::BeforeCheckingSnapshot"}});
+
+  bool flushed = false;
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
+      "DBImpl::MultiCFSnapshot::AfterRefSV", [&](void* /*arg*/) {
+        if (!flushed) {
+          for (int i = 0; i < 3; ++i) {
+            ASSERT_OK(Put(i, "cf" + std::to_string(i) + "_key",
+                          "cf" + std::to_string(i) + "_val_new"));
+          }
+          // After SV is obtained for the first CF, do the atomic flush()
+          ASSERT_OK(Flush());
+          flushed = true;
+        }
+      });
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
+  // Explicit snapshot wouldn't force reloading all svs. We should expect old
+  // values
+  const Snapshot* snapshot = db_->GetSnapshot();
+  ReadOptions read_options;
+  read_options.snapshot = snapshot;
+  std::vector<Iterator*> iters;
+  ASSERT_OK(db_->NewIterators(read_options, handles_, &iters));
+
+  for (int i = 0; i < 3; ++i) {
+    auto iter = iters[i];
+    ASSERT_OK(iter->status());
+    iter->SeekToFirst();
+    ASSERT_EQ(IterStatus(iter), "cf" + std::to_string(i) + "_key->cf" +
+                                    std::to_string(i) + "_val");
+  }
+  db_->ReleaseSnapshot(snapshot);
+  for (auto* iter : iters) {
+    delete iter;
+  }
+
+  // Thread-local SV for cf_0 is obsolete (atomic flush happened after the first
+  // SV Ref)
+  auto* cfd0 =
+      static_cast_with_check<ColumnFamilyHandleImpl>(handles_[0])->cfd();
+  ASSERT_EQ(cfd0->TEST_GetLocalSV()->Get(), SuperVersion::kSVObsolete);
+  ASSERT_NE(cfd0->TEST_GetLocalSV()->Get(), SuperVersion::kSVInUse);
+
+  // Rest are not InUse nor Obsolete
+  for (int i = 1; i < 3; ++i) {
+    auto* cfd =
+        static_cast_with_check<ColumnFamilyHandleImpl>(handles_[i])->cfd();
+    ASSERT_NE(cfd->TEST_GetLocalSV()->Get(), SuperVersion::kSVObsolete);
+    ASSERT_NE(cfd->TEST_GetLocalSV()->Get(), SuperVersion::kSVInUse);
+  }
+}
+
 }  // namespace ROCKSDB_NAMESPACE
 
 int main(int argc, char** argv) {

db/db_kv_checksum_test.cc

@@ -732,7 +732,7 @@ TEST_P(DbMemtableKVChecksumTest, GetWithCorruptAfterMemtableInsert) {
       });
 
   SyncPoint::GetInstance()->SetCallBack(
-      "Memtable::SaveValue:Begin:entry", [&](void* entry) {
+      "Memtable::SaveValue:Found:entry", [&](void* entry) {
         char* buf = *static_cast<char**>(entry);
         buf[corrupt_byte_offset_] += corrupt_byte_addend_;
         ++corrupt_byte_offset_;
@@ -769,7 +769,7 @@ TEST_P(DbMemtableKVChecksumTest,
       });
 
   SyncPoint::GetInstance()->SetCallBack(
-      "Memtable::SaveValue:Begin:entry", [&](void* entry) {
+      "Memtable::SaveValue:Found:entry", [&](void* entry) {
         char* buf = *static_cast<char**>(entry);
         buf[corrupt_byte_offset_] += corrupt_byte_addend_;
         ++corrupt_byte_offset_;

db/db_log_iter_test.cc

@@ -14,6 +14,7 @@
 #include "db/db_test_util.h"
 #include "env/mock_env.h"
 #include "port/stack_trace.h"
+#include "util/atomic.h"
 
 namespace ROCKSDB_NAMESPACE {
 
@@ -147,6 +148,7 @@ TEST_F(DBTestXactLogIterator, TransactionLogIteratorRace) {
 }
 
 TEST_F(DBTestXactLogIterator, TransactionLogIteratorCheckWhenArchive) {
+  RelaxedAtomic<bool> callback_hit{};
   do {
     ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->ClearTrace();
     Options options = OptionsForLogIterTest();
@@ -168,17 +170,23 @@ TEST_F(DBTestXactLogIterator, TransactionLogIteratorCheckWhenArchive) {
     ASSERT_OK(dbfull()->Put(WriteOptions(), "key4", DummyString(1024)));
     ASSERT_OK(dbfull()->Flush(FlushOptions()));
 
+    callback_hit.StoreRelaxed(false);
     ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
         "WalManager::PurgeObsoleteFiles:1", [&](void*) {
           auto iter = OpenTransactionLogIter(0);
           ExpectRecords(4, iter);
+          callback_hit.StoreRelaxed(true);
         });
 
     ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
     ASSERT_OK(dbfull()->Flush(FlushOptions(), cf));
-
     delete cf;
+    // Normally hit several times; WART: perhaps more in parallel after flush
+    // FIXME: this test is flaky
+    // ASSERT_TRUE(callback_hit.LoadRelaxed());
   } while (ChangeCompactOptions());
+  Close();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
 }
 #endif
 

db/db_memtable_test.cc

@@ -313,6 +313,10 @@ TEST_F(DBMemTableTest, InsertWithHint) {
   ASSERT_EQ("foo_v3", Get("foo_k3"));
   ASSERT_EQ("bar_v1", Get("bar_k1"));
   ASSERT_EQ("bar_v2", Get("bar_k2"));
+  ASSERT_OK(db_->DeleteRange(WriteOptions(), "foo_k1", "foo_k4"));
+  ASSERT_EQ(hint_bar, rep->last_hint_in());
+  ASSERT_EQ(hint_bar, rep->last_hint_out());
+  ASSERT_EQ(5, rep->num_insert_with_hint());
   ASSERT_EQ("vvv", Get("NotInPrefixDomain"));
 }
 

db/db_merge_operand_test.cc

@@ -430,6 +430,111 @@ TEST_F(DBMergeOperandTest, GetMergeOperandsLargeResultOptimization) {
   }
 }
 
+TEST_F(DBMergeOperandTest, GetMergeOperandsShortCircuitInMemtable) {
+  const int kNumOperands = 10;
+  const int kNumOperandsToFetch = 5;
+
+  Options options = CurrentOptions();
+  options.merge_operator = MergeOperators::CreateStringAppendOperator();
+  DestroyAndReopen(options);
+
+  Random rnd(301);
+  std::vector<std::string> expected_merge_operands;
+  expected_merge_operands.reserve(kNumOperands);
+  for (int i = 0; i < kNumOperands; ++i) {
+    expected_merge_operands.emplace_back(rnd.RandomString(7 /* len */));
+    ASSERT_OK(Merge("key", expected_merge_operands.back()));
+  }
+
+  std::vector<PinnableSlice> merge_operands(kNumOperands);
+  GetMergeOperandsOptions merge_operands_info;
+  merge_operands_info.expected_max_number_of_operands = kNumOperands;
+  int num_fetched = 0;
+  merge_operands_info.continue_cb = [&](Slice /* value */) {
+    num_fetched++;
+    return num_fetched != kNumOperandsToFetch;
+  };
+  int num_merge_operands = 0;
+  ASSERT_OK(db_->GetMergeOperands(ReadOptions(), db_->DefaultColumnFamily(),
+                                  "key", merge_operands.data(),
+                                  &merge_operands_info, &num_merge_operands));
+  ASSERT_EQ(kNumOperandsToFetch, num_merge_operands);
+  ASSERT_EQ(kNumOperandsToFetch, num_fetched);
+
+  for (int i = 0; i < kNumOperandsToFetch; ++i) {
+    ASSERT_EQ(expected_merge_operands[kNumOperands - kNumOperandsToFetch + i],
+              merge_operands[i]);
+  }
+}
+
+TEST_F(DBMergeOperandTest, GetMergeOperandsShortCircuitBaseValue) {
+  // The continuation callback doesn't need to be called on a base value because
+  // there's no remaining work to be saved.
+  Options options = CurrentOptions();
+  options.merge_operator = MergeOperators::CreateStringAppendOperator();
+  DestroyAndReopen(options);
+
+  Random rnd(301);
+  std::string expected_value = rnd.RandomString(7 /* len */);
+  ASSERT_OK(Put("key", expected_value));
+
+  std::vector<PinnableSlice> merge_operands(1);
+  GetMergeOperandsOptions merge_operands_info;
+  merge_operands_info.expected_max_number_of_operands = 1;
+  int num_fetched = 0;
+  merge_operands_info.continue_cb = [&num_fetched](Slice /* value */) {
+    num_fetched++;
+    return true;
+  };
+  int num_merge_operands = 0;
+  ASSERT_OK(db_->GetMergeOperands(ReadOptions(), db_->DefaultColumnFamily(),
+                                  "key", merge_operands.data(),
+                                  &merge_operands_info, &num_merge_operands));
+  ASSERT_EQ(1, num_merge_operands);
+  ASSERT_EQ(0, num_fetched);
+
+  ASSERT_EQ(expected_value, merge_operands[0]);
+}
+
+TEST_F(DBMergeOperandTest, GetMergeOperandsShortCircuitInSst) {
+  const int kNumOperands = 10;
+  const int kNumOperandsToFetch = 5;
+
+  Options options = CurrentOptions();
+  options.disable_auto_compactions = true;
+  options.merge_operator = MergeOperators::CreateStringAppendOperator();
+  DestroyAndReopen(options);
+
+  Random rnd(301);
+  std::vector<std::string> expected_merge_operands;
+  expected_merge_operands.reserve(kNumOperands);
+  for (int i = 0; i < kNumOperands; ++i) {
+    expected_merge_operands.emplace_back(rnd.RandomString(7 /* len */));
+    ASSERT_OK(Merge("key", expected_merge_operands.back()));
+    ASSERT_OK(Flush());
+  }
+
+  std::vector<PinnableSlice> merge_operands(kNumOperands);
+  GetMergeOperandsOptions merge_operands_info;
+  merge_operands_info.expected_max_number_of_operands = kNumOperands;
+  int num_fetched = 0;
+  merge_operands_info.continue_cb = [&](Slice /* value */) {
+    num_fetched++;
+    return num_fetched != kNumOperandsToFetch;
+  };
+  int num_merge_operands = 0;
+  ASSERT_OK(db_->GetMergeOperands(ReadOptions(), db_->DefaultColumnFamily(),
+                                  "key", merge_operands.data(),
+                                  &merge_operands_info, &num_merge_operands));
+  ASSERT_EQ(kNumOperandsToFetch, num_merge_operands);
+  ASSERT_EQ(kNumOperandsToFetch, num_fetched);
+
+  for (int i = 0; i < kNumOperandsToFetch; ++i) {
+    ASSERT_EQ(expected_merge_operands[kNumOperands - kNumOperandsToFetch + i],
+              merge_operands[i]);
+  }
+}
+
 TEST_F(DBMergeOperandTest, GetMergeOperandsBaseDeletionInImmMem) {
   // In this test, "k1" has a MERGE in a mutable memtable on top of a base
   // DELETE in an immutable memtable.

db/db_range_del_test.cc

@@ -2833,8 +2833,8 @@ TEST_F(DBRangeDelTest, LeftSentinelKeyTestWithNewerKey) {
   Arena arena;
   InternalKeyComparator icmp(options.comparator);
   ReadOptions read_options;
-  ScopedArenaIterator iter;
-  iter.set(
+  ScopedArenaPtr<InternalIterator> iter;
+  iter.reset(
       dbfull()->NewInternalIterator(read_options, &arena, kMaxSequenceNumber));
 
   auto k = Key(4);
@@ -3806,6 +3806,21 @@ TEST_F(DBRangeDelTest, RefreshWithSnapshot) {
   iter.reset();
   db_->ReleaseSnapshot(snapshot);
 }
+
+TEST_F(DBRangeDelTest, RowCache) {
+  Options options = CurrentOptions();
+  options.row_cache = NewLRUCache(8 << 10);
+  DestroyAndReopen(options);
+  ASSERT_OK(Put(Key(3), "val"));
+  ASSERT_TRUE(db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(),
+                               Key(3), Key(5))
+                  .IsNotSupported());
+  WriteBatch wb;
+  ASSERT_OK(wb.Put(Key(6), "abc"));
+  ASSERT_OK(wb.DeleteRange(Key(1), Key(5)));
+  ASSERT_TRUE(db_->Write(WriteOptions(), &wb).IsNotSupported());
+  ASSERT_EQ(Get(Key(3)), "val");
+}
 }  // namespace ROCKSDB_NAMESPACE
 
 int main(int argc, char** argv) {

db/db_sst_test.cc

@@ -21,7 +21,8 @@ namespace ROCKSDB_NAMESPACE {
 
 class DBSSTTest : public DBTestBase {
  public:
-  DBSSTTest() : DBTestBase("db_sst_test", /*env_do_fsync=*/true) {}
+  DBSSTTest(const std::string& test_name = "db_sst_test")
+      : DBTestBase(test_name, /*env_do_fsync=*/true) {}
 };
 
 // A class which remembers the name of each flushed file.
@@ -506,6 +507,23 @@ TEST_F(DBSSTTest, DBWithSstFileManagerForBlobFiles) {
   ASSERT_EQ(files_deleted, 0);
   ASSERT_EQ(files_scheduled_to_delete, 0);
   Close();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
+      "SstFileManagerImpl::ScheduleUnaccountedFileDeletion", [&](void* arg) {
+        assert(arg);
+        const std::string* const file_path =
+            static_cast<const std::string*>(arg);
+        if (EndsWith(*file_path, ".blob")) {
+          ++files_scheduled_to_delete;
+        }
+      });
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
+      "DeleteScheduler::OnDeleteFile", [&](void* arg) {
+        const std::string* const file_path =
+            static_cast<const std::string*>(arg);
+        if (EndsWith(*file_path, ".blob")) {
+          files_deleted++;
+        }
+      });
   ASSERT_OK(DestroyDB(dbname_, options));
   ASSERT_EQ(files_deleted, blob_files.size());
   ASSERT_EQ(files_scheduled_to_delete, blob_files.size());
@@ -648,6 +666,23 @@ TEST_F(DBSSTTest, DBWithSstFileManagerForBlobFilesWithGC) {
   }
 
   Close();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
+      "SstFileManagerImpl::ScheduleUnaccountedFileDeletion", [&](void* arg) {
+        assert(arg);
+        const std::string* const file_path =
+            static_cast<const std::string*>(arg);
+        if (EndsWith(*file_path, ".blob")) {
+          ++files_scheduled_to_delete;
+        }
+      });
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
+      "DeleteScheduler::OnDeleteFile", [&](void* arg) {
+        const std::string* const file_path =
+            static_cast<const std::string*>(arg);
+        if (EndsWith(*file_path, ".blob")) {
+          files_deleted++;
+        }
+      });
   ASSERT_OK(DestroyDB(dbname_, options));
   sfm->WaitForEmptyTrash();
   ASSERT_EQ(files_deleted, 5);
@@ -886,7 +921,7 @@ TEST_P(DBWALTestWithParam, WALTrashCleanupOnOpen) {
       // before restarting the DB.
       // We have to set this on the 2nd to last file for it to delay deletion
       // on the last file. (Quirk of DeleteScheduler::BackgroundEmptyTrash())
-      options.sst_file_manager->SetDeleteRateBytesPerSecond(1);
+      options.sst_file_manager->SetDeleteRateBytesPerSecond(1024 * 1024);
     }
     ASSERT_OK(Put("Key2", DummyString(1024, v)));
     ASSERT_OK(Put("Key3", DummyString(1024, v)));
@@ -937,15 +972,23 @@ INSTANTIATE_TEST_CASE_P(DBWALTestWithParam, DBWALTestWithParam,
                         ::testing::Values(std::make_tuple("", true),
                                           std::make_tuple("_wal_dir", false)));
 
-TEST_F(DBSSTTest, OpenDBWithExistingTrashAndObsoleteSstFile) {
+// Test param: max_trash_db_ratio for DeleteScheduler
+class DBObsoleteFileDeletionOnOpenTest
+    : public DBSSTTest,
+      public ::testing::WithParamInterface<double> {
+ public:
+  explicit DBObsoleteFileDeletionOnOpenTest()
+      : DBSSTTest("db_sst_deletion_on_open_test") {}
+};
+
+TEST_P(DBObsoleteFileDeletionOnOpenTest, Basic) {
   Options options = CurrentOptions();
   options.sst_file_manager.reset(
       NewSstFileManager(env_, nullptr, "", 1024 * 1024 /* 1 MB/sec */));
   auto sfm = static_cast<SstFileManagerImpl*>(options.sst_file_manager.get());
-  // Set an extra high trash ratio to prevent immediate/non-rate limited
-  // deletions
   sfm->SetDeleteRateBytesPerSecond(1024 * 1024);
-  sfm->delete_scheduler()->SetMaxTrashDBRatio(1000.0);
+  double max_trash_db_ratio = GetParam();
+  sfm->delete_scheduler()->SetMaxTrashDBRatio(max_trash_db_ratio);
 
   int bg_delete_file = 0;
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
@@ -965,10 +1008,19 @@ TEST_F(DBSSTTest, OpenDBWithExistingTrashAndObsoleteSstFile) {
       WriteStringToFile(env_, "abc", dbname_ + "/" + "003.sst.trash", false));
   // Manually add an obsolete sst file. Obsolete SST files are discovered and
   // deleted upon recovery.
-  constexpr uint64_t kSstFileNumber = 100;
-  const std::string kObsoleteSstFile =
-      MakeTableFileName(dbname_, kSstFileNumber);
-  ASSERT_OK(WriteStringToFile(env_, "abc", kObsoleteSstFile, false));
+  uint64_t sst_file_number = 100;
+  const std::string kObsoleteSstFileOne =
+      MakeTableFileName(dbname_, sst_file_number);
+  ASSERT_OK(WriteStringToFile(env_, "abc", kObsoleteSstFileOne, false));
+  // The slow deletion on recovery had a bug before where a file's size is not
+  // first tracked in `total_size_` in SstFileManager before passed to
+  // DeleteScheduler. The first obsolete file is still slow deleted because
+  // 0 (total_trash_size_) > 0 (total_size_) * 1000 (max_trash_db_ratio)
+  // is always false.
+  // Here we explicitly create a second obsolete file to verify this bug's fix
+  const std::string kObsoleteSstFileTwo =
+      MakeTableFileName(dbname_, sst_file_number - 1);
+  ASSERT_OK(WriteStringToFile(env_, "abc", kObsoleteSstFileTwo, false));
 
   // Reopen the DB and verify that it deletes existing trash files and obsolete
   // SST files with rate limiting.
@@ -977,10 +1029,26 @@ TEST_F(DBSSTTest, OpenDBWithExistingTrashAndObsoleteSstFile) {
   ASSERT_NOK(env_->FileExists(dbname_ + "/" + "001.sst.trash"));
   ASSERT_NOK(env_->FileExists(dbname_ + "/" + "002.sst.trash"));
   ASSERT_NOK(env_->FileExists(dbname_ + "/" + "003.sst.trash"));
-  ASSERT_NOK(env_->FileExists(kObsoleteSstFile));
-  ASSERT_EQ(bg_delete_file, 4);
+  ASSERT_NOK(env_->FileExists(kObsoleteSstFileOne));
+  ASSERT_NOK(env_->FileExists(kObsoleteSstFileTwo));
+  // The files in the DB's directory are all either trash or obsolete sst files.
+  // So the trash/db ratio is 1. A ratio equal to or higher than 1 should
+  // schedule all files' deletion in background. A ratio lower than 1 may
+  // send some files to be deleted immediately.
+  if (max_trash_db_ratio < 1) {
+    ASSERT_LE(bg_delete_file, 5);
+  } else {
+    ASSERT_EQ(bg_delete_file, 5);
+  }
+
+  ASSERT_EQ(sfm->GetTotalSize(), 0);
+  ASSERT_EQ(sfm->delete_scheduler()->GetTotalTrashSize(), 0);
 }
 
+INSTANTIATE_TEST_CASE_P(DBObsoleteFileDeletionOnOpenTest,
+                        DBObsoleteFileDeletionOnOpenTest,
+                        ::testing::Values(0, 0.5, 1, 1.2));
+
 // Create a DB with 2 db_paths, and generate multiple files in the 2
 // db_paths using CompactRangeOptions, make sure that files that were
 // deleted from first db_path were deleted using DeleteScheduler and
@@ -1868,6 +1936,24 @@ TEST_F(DBSSTTest, DBWithSFMForBlobFilesAtomicFlush) {
   ASSERT_EQ(files_deleted, 1);
 
   Close();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
+      "SstFileManagerImpl::ScheduleUnaccountedFileDeletion", [&](void* arg) {
+        assert(arg);
+        const std::string* const file_path =
+            static_cast<const std::string*>(arg);
+        if (EndsWith(*file_path, ".blob")) {
+          ++files_scheduled_to_delete;
+        }
+      });
+
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
+      "DeleteScheduler::OnDeleteFile", [&](void* arg) {
+        const std::string* const file_path =
+            static_cast<const std::string*>(arg);
+        if (EndsWith(*file_path, ".blob")) {
+          files_deleted++;
+        }
+      });
   ASSERT_OK(DestroyDB(dbname_, options));
 
   ASSERT_EQ(files_scheduled_to_delete, 4);

db/db_table_properties_test.cc

@@ -280,7 +280,7 @@ class DBTablePropertiesInRangeTest : public DBTestBase,
     // run the query
     TablePropertiesCollection props;
     ColumnFamilyHandle* default_cf = db_->DefaultColumnFamily();
-    EXPECT_OK(db_->GetPropertiesOfTablesInRange(default_cf, &ranges[0],
+    EXPECT_OK(db_->GetPropertiesOfTablesInRange(default_cf, ranges.data(),
                                                 ranges.size(), &props));
 
     const Comparator* ucmp = default_cf->GetComparator();

db/db_test.cc

@@ -26,6 +26,7 @@
 #endif
 
 #include "cache/lru_cache.h"
+#include "db/attribute_group_iterator_impl.h"
 #include "db/blob/blob_index.h"
 #include "db/blob/blob_log_format.h"
 #include "db/db_impl/db_impl.h"
@@ -59,7 +60,6 @@
 #include "rocksdb/utilities/optimistic_transaction_db.h"
 #include "rocksdb/utilities/write_batch_with_index.h"
 #include "table/mock_table.h"
-#include "table/scoped_arena_iterator.h"
 #include "test_util/sync_point.h"
 #include "test_util/testharness.h"
 #include "test_util/testutil.h"
@@ -3199,6 +3199,21 @@ class ModelDB : public DB {
                       std::vector<Iterator*>* /*iterators*/) override {
     return Status::NotSupported("Not supported yet");
   }
+
+  std::unique_ptr<Iterator> NewCoalescingIterator(
+      const ReadOptions& /*options*/,
+      const std::vector<ColumnFamilyHandle*>& /*column_families*/) override {
+    return std::unique_ptr<Iterator>(
+        NewErrorIterator(Status::NotSupported("Not supported yet")));
+  }
+
+  std::unique_ptr<AttributeGroupIterator> NewAttributeGroupIterator(
+      const ReadOptions& /*options*/,
+      const std::vector<ColumnFamilyHandle*>& /*column_families*/) override {
+    return NewAttributeGroupErrorIterator(
+        Status::NotSupported("Not supported yet"));
+  }
+
   const Snapshot* GetSnapshot() override {
     ModelSnapshot* snapshot = new ModelSnapshot;
     snapshot->map_ = map_;
@@ -3829,6 +3844,9 @@ TEST_P(DBTestWithParam, FIFOCompactionTest) {
     } else {
       CompactRangeOptions cro;
       cro.exclusive_manual_compaction = exclusive_manual_compaction_;
+      cro.change_level = true;
+      ASSERT_TRUE(db_->CompactRange(cro, nullptr, nullptr).IsNotSupported());
+      cro.change_level = false;
       ASSERT_OK(db_->CompactRange(cro, nullptr, nullptr));
     }
     // only 5 files should survive
@@ -4659,24 +4677,27 @@ TEST_F(DBTest, DynamicMemtableOptions) {
 
 #ifdef ROCKSDB_USING_THREAD_STATUS
 namespace {
-void VerifyOperationCount(Env* env, ThreadStatus::OperationType op_type,
+bool VerifyOperationCount(Env* env, ThreadStatus::OperationType op_type,
                           int expected_count) {
   int op_count = 0;
   std::vector<ThreadStatus> thread_list;
-  ASSERT_OK(env->GetThreadList(&thread_list));
+  EXPECT_OK(env->GetThreadList(&thread_list));
   for (const auto& thread : thread_list) {
     if (thread.operation_type == op_type) {
       op_count++;
     }
   }
   if (op_count != expected_count) {
+    fprintf(stderr, "op_count: %d, expected_count %d\n", op_count,
+            expected_count);
     for (const auto& thread : thread_list) {
-      fprintf(stderr, "thread id: %" PRIu64 ", thread status: %s\n",
+      fprintf(stderr, "thread id: %" PRIu64 ", thread status: %s, cf_name %s\n",
               thread.thread_id,
-              thread.GetOperationName(thread.operation_type).c_str());
+              thread.GetOperationName(thread.operation_type).c_str(),
+              thread.cf_name.c_str());
     }
   }
-  ASSERT_EQ(op_count, expected_count);
+  return op_count == expected_count;
 }
 }  // anonymous namespace
 
@@ -4776,11 +4797,11 @@ TEST_F(DBTest, ThreadStatusFlush) {
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
 
   CreateAndReopenWithCF({"pikachu"}, options);
-  VerifyOperationCount(env_, ThreadStatus::OP_FLUSH, 0);
+  ASSERT_TRUE(VerifyOperationCount(env_, ThreadStatus::OP_FLUSH, 0));
 
   ASSERT_OK(Put(1, "foo", "v1"));
   ASSERT_EQ("v1", Get(1, "foo"));
-  VerifyOperationCount(env_, ThreadStatus::OP_FLUSH, 0);
+  ASSERT_TRUE(VerifyOperationCount(env_, ThreadStatus::OP_FLUSH, 0));
 
   uint64_t num_running_flushes = 0;
   ASSERT_TRUE(db_->GetIntProperty(DB::Properties::kNumRunningFlushes,
@@ -4793,7 +4814,7 @@ TEST_F(DBTest, ThreadStatusFlush) {
   // The first sync point is to make sure there's one flush job
   // running when we perform VerifyOperationCount().
   TEST_SYNC_POINT("DBTest::ThreadStatusFlush:1");
-  VerifyOperationCount(env_, ThreadStatus::OP_FLUSH, 1);
+  ASSERT_TRUE(VerifyOperationCount(env_, ThreadStatus::OP_FLUSH, 1));
   ASSERT_TRUE(db_->GetIntProperty(DB::Properties::kNumRunningFlushes,
                                   &num_running_flushes));
   ASSERT_EQ(num_running_flushes, 1);
@@ -4804,17 +4825,11 @@ TEST_F(DBTest, ThreadStatusFlush) {
 }
 
 TEST_P(DBTestWithParam, ThreadStatusSingleCompaction) {
-  const int kTestKeySize = 16;
   const int kTestValueSize = 984;
-  const int kEntrySize = kTestKeySize + kTestValueSize;
   const int kEntriesPerBuffer = 100;
   Options options;
   options.create_if_missing = true;
-  options.write_buffer_size = kEntrySize * kEntriesPerBuffer;
   options.compaction_style = kCompactionStyleLevel;
-  options.target_file_size_base = options.write_buffer_size;
-  options.max_bytes_for_level_base = options.target_file_size_base * 2;
-  options.max_bytes_for_level_multiplier = 2;
   options.compression = kNoCompression;
   options = CurrentOptions(options);
   options.env = env_;
@@ -4849,7 +4864,7 @@ TEST_P(DBTestWithParam, ThreadStatusSingleCompaction) {
                                     &num_running_compactions));
     ASSERT_EQ(num_running_compactions, 0);
     TEST_SYNC_POINT("DBTest::ThreadStatusSingleCompaction:0");
-    ASSERT_GE(NumTableFilesAtLevel(0),
+    ASSERT_EQ(NumTableFilesAtLevel(0),
               options.level0_file_num_compaction_trigger);
 
     // This makes sure at least one compaction is running.
@@ -4857,10 +4872,18 @@ TEST_P(DBTestWithParam, ThreadStatusSingleCompaction) {
 
     if (options.enable_thread_tracking) {
       // expecting one single L0 to L1 compaction
-      VerifyOperationCount(env_, ThreadStatus::OP_COMPACTION, 1);
+      // This test is flaky and fails here.
+      bool match = VerifyOperationCount(env_, ThreadStatus::OP_COMPACTION, 1);
+      if (!match) {
+        ASSERT_TRUE(db_->GetIntProperty(DB::Properties::kNumRunningCompactions,
+                                        &num_running_compactions));
+        fprintf(stderr, "running compaction: %" PRIu64 " lsm state: %s\n",
+                num_running_compactions, FilesPerLevel().c_str());
+      }
+      ASSERT_TRUE(match);
     } else {
       // If thread tracking is not enabled, compaction count should be 0.
-      VerifyOperationCount(env_, ThreadStatus::OP_COMPACTION, 0);
+      ASSERT_TRUE(VerifyOperationCount(env_, ThreadStatus::OP_COMPACTION, 0));
     }
     ASSERT_TRUE(db_->GetIntProperty(DB::Properties::kNumRunningCompactions,
                                     &num_running_compactions));

db/db_test2.cc

@@ -6604,7 +6604,7 @@ TEST_F(DBTest2, LastLevelTemperature) {
   auto* listener = new TestListener();
 
   Options options = CurrentOptions();
-  options.bottommost_temperature = Temperature::kWarm;
+  options.last_level_temperature = Temperature::kWarm;
   options.level0_file_num_compaction_trigger = 2;
   options.level_compaction_dynamic_level_bytes = true;
   options.num_levels = kNumLevels;
@@ -6816,8 +6816,8 @@ TEST_F(DBTest2, LastLevelTemperatureUniversal) {
   size = GetSstSizeHelper(Temperature::kWarm);
   ASSERT_EQ(size, 0);
 
-  // Update bottommost temperature
-  options.bottommost_temperature = Temperature::kWarm;
+  // Update last level temperature
+  options.last_level_temperature = Temperature::kWarm;
   Reopen(options);
   db_->GetColumnFamilyMetaData(&metadata);
   // Should not impact existing ones
@@ -6869,10 +6869,10 @@ TEST_F(DBTest2, LastLevelTemperatureUniversal) {
       &prop));
   ASSERT_EQ(std::atoi(prop.c_str()), 0);
 
-  // Update bottommost temperature dynamically with SetOptions
+  // Update last level temperature dynamically with SetOptions
   auto s = db_->SetOptions({{"last_level_temperature", "kCold"}});
   ASSERT_OK(s);
-  ASSERT_EQ(db_->GetOptions().bottommost_temperature, Temperature::kCold);
+  ASSERT_EQ(db_->GetOptions().last_level_temperature, Temperature::kCold);
   db_->GetColumnFamilyMetaData(&metadata);
   // Should not impact the existing files
   ASSERT_EQ(Temperature::kWarm,
@@ -6898,94 +6898,133 @@ TEST_F(DBTest2, LastLevelTemperatureUniversal) {
   ASSERT_GT(size, 0);
 
   // kLastTemperature is an invalid temperature
-  options.bottommost_temperature = Temperature::kLastTemperature;
+  options.last_level_temperature = Temperature::kLastTemperature;
   s = TryReopen(options);
   ASSERT_TRUE(s.IsIOError());
 }
 
 TEST_F(DBTest2, LastLevelStatistics) {
-  Options options = CurrentOptions();
-  options.bottommost_temperature = Temperature::kWarm;
-  options.default_temperature = Temperature::kHot;
-  options.level0_file_num_compaction_trigger = 2;
-  options.level_compaction_dynamic_level_bytes = true;
-  options.statistics = CreateDBStatistics();
-  Reopen(options);
+  for (bool write_time_default : {false, true}) {
+    SCOPED_TRACE("write time default? " + std::to_string(write_time_default));
+    Options options = CurrentOptions();
+    options.last_level_temperature = Temperature::kWarm;
+    if (write_time_default) {
+      options.default_write_temperature = Temperature::kHot;
+      ASSERT_EQ(options.default_temperature, Temperature::kUnknown);
+    } else {
+      options.default_temperature = Temperature::kHot;
+      ASSERT_EQ(options.default_write_temperature, Temperature::kUnknown);
+    }
+    options.level0_file_num_compaction_trigger = 2;
+    options.level_compaction_dynamic_level_bytes = true;
+    options.statistics = CreateDBStatistics();
+    BlockBasedTableOptions bbto;
+    bbto.no_block_cache = true;
+    options.table_factory.reset(NewBlockBasedTableFactory(bbto));
 
-  // generate 1 sst on level 0
-  ASSERT_OK(Put("foo", "bar"));
-  ASSERT_OK(Put("bar", "bar"));
-  ASSERT_OK(Flush());
-  ASSERT_EQ("bar", Get("bar"));
+    DestroyAndReopen(options);
 
-  ASSERT_GT(options.statistics->getTickerCount(NON_LAST_LEVEL_READ_BYTES), 0);
-  ASSERT_GT(options.statistics->getTickerCount(NON_LAST_LEVEL_READ_COUNT), 0);
-  ASSERT_EQ(options.statistics->getTickerCount(NON_LAST_LEVEL_READ_BYTES),
-            options.statistics->getTickerCount(HOT_FILE_READ_BYTES));
-  ASSERT_EQ(options.statistics->getTickerCount(NON_LAST_LEVEL_READ_COUNT),
-            options.statistics->getTickerCount(HOT_FILE_READ_COUNT));
-  ASSERT_EQ(options.statistics->getTickerCount(LAST_LEVEL_READ_BYTES), 0);
-  ASSERT_EQ(options.statistics->getTickerCount(LAST_LEVEL_READ_COUNT), 0);
+    // generate 1 sst on level 0
+    ASSERT_OK(Put("foo1", "bar"));
+    ASSERT_OK(Put("bar", "bar"));
+    ASSERT_OK(Flush());
+    ASSERT_EQ("bar", Get("bar"));
 
-  // 2nd flush to trigger compaction
-  ASSERT_OK(Put("foo", "bar"));
-  ASSERT_OK(Put("bar", "bar"));
-  ASSERT_OK(Flush());
-  ASSERT_OK(dbfull()->TEST_WaitForCompact());
-  ASSERT_EQ("bar", Get("bar"));
+    ASSERT_GT(options.statistics->getTickerCount(NON_LAST_LEVEL_READ_BYTES), 0);
+    ASSERT_GT(options.statistics->getTickerCount(NON_LAST_LEVEL_READ_COUNT), 0);
+    ASSERT_EQ(options.statistics->getTickerCount(NON_LAST_LEVEL_READ_BYTES),
+              options.statistics->getTickerCount(HOT_FILE_READ_BYTES));
+    ASSERT_EQ(options.statistics->getTickerCount(NON_LAST_LEVEL_READ_COUNT),
+              options.statistics->getTickerCount(HOT_FILE_READ_COUNT));
+    ASSERT_EQ(options.statistics->getTickerCount(LAST_LEVEL_READ_BYTES), 0);
+    ASSERT_EQ(options.statistics->getTickerCount(LAST_LEVEL_READ_COUNT), 0);
 
-  ASSERT_EQ(options.statistics->getTickerCount(NON_LAST_LEVEL_READ_BYTES),
-            options.statistics->getTickerCount(HOT_FILE_READ_BYTES));
-  ASSERT_EQ(options.statistics->getTickerCount(NON_LAST_LEVEL_READ_COUNT),
-            options.statistics->getTickerCount(HOT_FILE_READ_COUNT));
-  ASSERT_EQ(options.statistics->getTickerCount(LAST_LEVEL_READ_BYTES),
-            options.statistics->getTickerCount(WARM_FILE_READ_BYTES));
-  ASSERT_EQ(options.statistics->getTickerCount(LAST_LEVEL_READ_COUNT),
-            options.statistics->getTickerCount(WARM_FILE_READ_COUNT));
+    // 2nd flush to trigger compaction
+    ASSERT_OK(Put("foo2", "bar"));
+    ASSERT_OK(Put("bar", "bar"));
+    ASSERT_OK(Flush());
+    ASSERT_OK(dbfull()->TEST_WaitForCompact());
+    ASSERT_EQ("bar", Get("bar"));
 
-  auto pre_bytes =
-      options.statistics->getTickerCount(NON_LAST_LEVEL_READ_BYTES);
-  auto pre_count =
-      options.statistics->getTickerCount(NON_LAST_LEVEL_READ_COUNT);
+    ASSERT_EQ(options.statistics->getTickerCount(NON_LAST_LEVEL_READ_BYTES),
+              options.statistics->getTickerCount(HOT_FILE_READ_BYTES));
+    ASSERT_EQ(options.statistics->getTickerCount(NON_LAST_LEVEL_READ_COUNT),
+              options.statistics->getTickerCount(HOT_FILE_READ_COUNT));
+    ASSERT_EQ(options.statistics->getTickerCount(LAST_LEVEL_READ_BYTES),
+              options.statistics->getTickerCount(WARM_FILE_READ_BYTES));
+    ASSERT_EQ(options.statistics->getTickerCount(LAST_LEVEL_READ_COUNT),
+              options.statistics->getTickerCount(WARM_FILE_READ_COUNT));
 
-  // 3rd flush to generate 1 sst on level 0
-  ASSERT_OK(Put("foo", "bar"));
-  ASSERT_OK(Put("bar", "bar"));
-  ASSERT_OK(Flush());
-  ASSERT_EQ("bar", Get("bar"));
+    auto pre_bytes =
+        options.statistics->getTickerCount(NON_LAST_LEVEL_READ_BYTES);
+    auto pre_count =
+        options.statistics->getTickerCount(NON_LAST_LEVEL_READ_COUNT);
 
-  ASSERT_GT(options.statistics->getTickerCount(NON_LAST_LEVEL_READ_BYTES),
-            pre_bytes);
-  ASSERT_GT(options.statistics->getTickerCount(NON_LAST_LEVEL_READ_COUNT),
-            pre_count);
-  ASSERT_EQ(options.statistics->getTickerCount(NON_LAST_LEVEL_READ_BYTES),
-            options.statistics->getTickerCount(HOT_FILE_READ_BYTES));
-  ASSERT_EQ(options.statistics->getTickerCount(NON_LAST_LEVEL_READ_COUNT),
-            options.statistics->getTickerCount(HOT_FILE_READ_COUNT));
-  ASSERT_EQ(options.statistics->getTickerCount(LAST_LEVEL_READ_BYTES),
-            options.statistics->getTickerCount(WARM_FILE_READ_BYTES));
-  ASSERT_EQ(options.statistics->getTickerCount(LAST_LEVEL_READ_COUNT),
-            options.statistics->getTickerCount(WARM_FILE_READ_COUNT));
+    // 3rd flush to generate 1 sst on level 0
+    ASSERT_OK(Put("foo3", "bar"));
+    ASSERT_OK(Put("bar", "bar"));
+    ASSERT_OK(Flush());
+    ASSERT_EQ("bar", Get("foo1"));
+    ASSERT_EQ("bar", Get("foo2"));
+    ASSERT_EQ("bar", Get("foo3"));
+    ASSERT_EQ("bar", Get("bar"));
 
-  // Not a realistic setting to make last level kWarm and default temp kCold.
-  // This is just for testing default temp can be reset on reopen while the
-  // last level temp is consistent across DB reopen because those file's temp
-  // are persisted in manifest.
-  options.default_temperature = Temperature::kCold;
-  ASSERT_OK(options.statistics->Reset());
-  Reopen(options);
-  ASSERT_EQ("bar", Get("bar"));
+    ASSERT_GT(options.statistics->getTickerCount(NON_LAST_LEVEL_READ_BYTES),
+              pre_bytes);
+    ASSERT_GT(options.statistics->getTickerCount(NON_LAST_LEVEL_READ_COUNT),
+              pre_count);
+    ASSERT_EQ(options.statistics->getTickerCount(NON_LAST_LEVEL_READ_BYTES),
+              options.statistics->getTickerCount(HOT_FILE_READ_BYTES));
+    ASSERT_EQ(options.statistics->getTickerCount(NON_LAST_LEVEL_READ_COUNT),
+              options.statistics->getTickerCount(HOT_FILE_READ_COUNT));
+    ASSERT_EQ(options.statistics->getTickerCount(LAST_LEVEL_READ_BYTES),
+              options.statistics->getTickerCount(WARM_FILE_READ_BYTES));
+    ASSERT_EQ(options.statistics->getTickerCount(LAST_LEVEL_READ_COUNT),
+              options.statistics->getTickerCount(WARM_FILE_READ_COUNT));
+    // Control
+    ASSERT_NE(options.statistics->getTickerCount(LAST_LEVEL_READ_COUNT),
+              options.statistics->getTickerCount(NON_LAST_LEVEL_READ_COUNT));
 
-  ASSERT_EQ(0, options.statistics->getTickerCount(HOT_FILE_READ_BYTES));
+    // Not a realistic setting to make last level kWarm and default temp kCold.
+    // This is just for testing default temp can be reset on reopen while the
+    // last level temp is consistent across DB reopen because those file's temp
+    // are persisted in manifest.
+    options.default_temperature = Temperature::kCold;
+    ASSERT_OK(options.statistics->Reset());
+    Reopen(options);
+    ASSERT_EQ("bar", Get("foo1"));
+    ASSERT_EQ("bar", Get("foo2"));
+    ASSERT_EQ("bar", Get("foo3"));
+    ASSERT_EQ("bar", Get("bar"));
 
-  ASSERT_EQ(options.statistics->getTickerCount(NON_LAST_LEVEL_READ_BYTES),
-            options.statistics->getTickerCount(COLD_FILE_READ_BYTES));
-  ASSERT_EQ(options.statistics->getTickerCount(NON_LAST_LEVEL_READ_COUNT),
-            options.statistics->getTickerCount(COLD_FILE_READ_COUNT));
-  ASSERT_EQ(options.statistics->getTickerCount(LAST_LEVEL_READ_BYTES),
-            options.statistics->getTickerCount(WARM_FILE_READ_BYTES));
-  ASSERT_EQ(options.statistics->getTickerCount(LAST_LEVEL_READ_COUNT),
-            options.statistics->getTickerCount(WARM_FILE_READ_COUNT));
+    if (write_time_default) {
+      // Unchanged
+      ASSERT_EQ(options.statistics->getTickerCount(NON_LAST_LEVEL_READ_BYTES),
+                options.statistics->getTickerCount(HOT_FILE_READ_BYTES));
+      ASSERT_EQ(options.statistics->getTickerCount(NON_LAST_LEVEL_READ_COUNT),
+                options.statistics->getTickerCount(HOT_FILE_READ_COUNT));
+
+      ASSERT_LT(0, options.statistics->getTickerCount(HOT_FILE_READ_BYTES));
+      ASSERT_EQ(0, options.statistics->getTickerCount(COLD_FILE_READ_BYTES));
+    } else {
+      // Changed (in how we map kUnknown)
+      ASSERT_EQ(options.statistics->getTickerCount(NON_LAST_LEVEL_READ_BYTES),
+                options.statistics->getTickerCount(COLD_FILE_READ_BYTES));
+      ASSERT_EQ(options.statistics->getTickerCount(NON_LAST_LEVEL_READ_COUNT),
+                options.statistics->getTickerCount(COLD_FILE_READ_COUNT));
+
+      ASSERT_EQ(0, options.statistics->getTickerCount(HOT_FILE_READ_BYTES));
+      ASSERT_LT(0, options.statistics->getTickerCount(COLD_FILE_READ_BYTES));
+    }
+
+    ASSERT_EQ(options.statistics->getTickerCount(LAST_LEVEL_READ_BYTES),
+              options.statistics->getTickerCount(WARM_FILE_READ_BYTES));
+    ASSERT_EQ(options.statistics->getTickerCount(LAST_LEVEL_READ_COUNT),
+              options.statistics->getTickerCount(WARM_FILE_READ_COUNT));
+    // Control
+    ASSERT_NE(options.statistics->getTickerCount(LAST_LEVEL_READ_COUNT),
+              options.statistics->getTickerCount(NON_LAST_LEVEL_READ_COUNT));
+  }
 }
 
 TEST_F(DBTest2, CheckpointFileTemperature) {
@@ -7002,7 +7041,7 @@ TEST_F(DBTest2, CheckpointFileTemperature) {
   auto test_fs = std::make_shared<NoLinkTestFS>(env_->GetFileSystem());
   std::unique_ptr<Env> env(new CompositeEnvWrapper(env_, test_fs));
   Options options = CurrentOptions();
-  options.bottommost_temperature = Temperature::kWarm;
+  options.last_level_temperature = Temperature::kWarm;
   // set dynamic_level to true so the compaction would compact the data to the
   // last level directly which will have the last_level_temperature
   options.level_compaction_dynamic_level_bytes = true;
@@ -7061,7 +7100,7 @@ TEST_F(DBTest2, FileTemperatureManifestFixup) {
   auto test_fs = std::make_shared<FileTemperatureTestFS>(env_->GetFileSystem());
   std::unique_ptr<Env> env(new CompositeEnvWrapper(env_, test_fs));
   Options options = CurrentOptions();
-  options.bottommost_temperature = Temperature::kWarm;
+  options.last_level_temperature = Temperature::kWarm;
   // set dynamic_level to true so the compaction would compact the data to the
   // last level directly which will have the last_level_temperature
   options.level_compaction_dynamic_level_bytes = true;
@@ -7769,6 +7808,27 @@ TEST_F(DBTest2, ZSTDChecksum) {
 }
 #endif
 
+TEST_F(DBTest2, TableCacheMissDuringReadFromBlockCacheTier) {
+  Options options = CurrentOptions();
+  options.statistics = ROCKSDB_NAMESPACE::CreateDBStatistics();
+  Reopen(options);
+
+  // Give table cache zero capacity to prevent preloading tables. That way,
+  // `kBlockCacheTier` reads will fail due to table cache misses.
+  dbfull()->TEST_table_cache()->SetCapacity(0);
+  ASSERT_OK(Put("foo", "bar"));
+  ASSERT_OK(Flush());
+
+  uint64_t orig_num_file_opens = TestGetTickerCount(options, NO_FILE_OPENS);
+
+  ReadOptions non_blocking_opts;
+  non_blocking_opts.read_tier = kBlockCacheTier;
+  std::string value;
+  ASSERT_TRUE(db_->Get(non_blocking_opts, "foo", &value).IsIncomplete());
+
+  ASSERT_EQ(orig_num_file_opens, TestGetTickerCount(options, NO_FILE_OPENS));
+}
+
 }  // namespace ROCKSDB_NAMESPACE
 
 int main(int argc, char** argv) {