Fix asan_check failures

Summary: Pull Request resolved: https://github.com/facebook/rocksdb/pull/5589 Test Plan: TEST_TMPDIR=/dev/shm/rocksdb COMPILE_WITH_ASAN=1 OPT=-g make J=64 -j64 asan_check Differential Revision: D16361081 Pulled By: anand1976 fbshipit-source-id: 09474832b9cfb318a840d4b633e22dfad105d58c
HISTORY update for export and import column family APIs
2026-07-07 22:55:23 +08:00 · 2019-07-18 14:51:25 -07:00 · 2019-07-18 10:16:38 -07:00 · 2019-07-17 22:06:12 -07:00 · 2019-07-17 17:04:39 -07:00 · 2019-07-17 13:05:58 -07:00
833 changed files with 64587 additions and 27604 deletions
@@ -32,6 +32,7 @@ ldb
 manifest_dump
 sst_dump
 blob_dump
+block_cache_trace_analyzer
 column_aware_encoding_exp
 util/build_version.cc
 build_tools/VALGRIND_LOGS/
@@ -47,6 +48,7 @@ rocksdb_undump
 db_test2
 trace_analyzer
 trace_analyzer_test
+.DS_Store

 java/out
 java/target
@@ -1,5 +1,4 @@
-sudo: false
-dist: trusty
+dist: xenial
 language: cpp
 os:
  - linux
@@ -9,7 +8,7 @@ compiler:
  - gcc
 osx_image: xcode8.3
 jdk:
-  - oraclejdk7
+  - openjdk7
 cache:
  - ccache
  - apt
@@ -71,7 +70,10 @@ install:
      CC=gcc-8 && CXX=g++-8;
    fi
  - if [[ "${JOB_NAME}" == cmake* ]] && [ "${TRAVIS_OS_NAME}" == linux ]; then
-      mkdir cmake-dist && curl -sfSL https://cmake.org/files/v3.8/cmake-3.8.1-Linux-x86_64.tar.gz | tar --strip-components=1 -C cmake-dist -xz && export PATH=$PWD/cmake-dist/bin:$PATH;
+      mkdir cmake-dist && curl -sfSL https://github.com/Kitware/CMake/releases/download/v3.14.5/cmake-3.14.5-Linux-x86_64.tar.gz | tar --strip-components=1 -C cmake-dist -xz && export PATH=$PWD/cmake-dist/bin:$PATH;
+    fi
+  - if [[ "${JOB_NAME}" == java_test ]]; then
+      java -version && echo "JAVA_HOME=${JAVA_HOME}";
    fi

 before_script:
@@ -101,7 +103,7 @@ script:
    esac
  - case $JOB_NAME in
    java_test)
-      OPT=-DTRAVIS V=1 make clean jclean && make rocksdbjava jtest
+      OPT=-DTRAVIS V=1 make rocksdbjava jtest
      ;;
    lite_build)
      OPT='-DTRAVIS -DROCKSDB_LITE' V=1 make -j4 static_lib tools
@@ -110,6 +112,7 @@ script:
      OPT=-DTRAVIS V=1 make -j4 static_lib && cd examples && make -j4
      ;;
    cmake-mingw)
+      sudo update-alternatives --set x86_64-w64-mingw32-g++ /usr/bin/x86_64-w64-mingw32-g++-posix;
      mkdir build && cd build && cmake -DJNI=1 .. -DCMAKE_C_COMPILER=x86_64-w64-mingw32-gcc -DCMAKE_CXX_COMPILER=x86_64-w64-mingw32-g++ -DCMAKE_SYSTEM_NAME=Windows && make -j4 rocksdb rocksdbjni
      ;;
    cmake*)
@@ -44,6 +44,12 @@ endif()

 list(APPEND CMAKE_MODULE_PATH "${CMAKE_CURRENT_LIST_DIR}/cmake/modules/")

+find_program(CCACHE_FOUND ccache)
+if(CCACHE_FOUND)
+  set_property(GLOBAL PROPERTY RULE_LAUNCH_COMPILE ccache)
+  set_property(GLOBAL PROPERTY RULE_LAUNCH_LINK ccache)
+endif(CCACHE_FOUND)
+
 option(WITH_JEMALLOC "build with JeMalloc" OFF)
 option(WITH_SNAPPY "build with SNAPPY" OFF)
 option(WITH_LZ4 "build with lz4" OFF)
@@ -92,7 +98,7 @@ else()
  endif()

  if(WITH_ZLIB)
-    find_package(zlib REQUIRED)
+    find_package(ZLIB REQUIRED)
    add_definitions(-DZLIB)
    if(ZLIB_INCLUDE_DIRS)
      # CMake 3
@@ -177,6 +183,7 @@ else()
  set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wsign-compare -Wshadow -Wno-unused-parameter -Wno-unused-variable -Woverloaded-virtual -Wnon-virtual-dtor -Wno-missing-field-initializers -Wno-strict-aliasing")
  if(MINGW)
    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wno-format")
+    add_definitions(-D_POSIX_C_SOURCE=1)
  endif()
  set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++11")
  if(NOT CMAKE_BUILD_TYPE STREQUAL "Debug")
@@ -201,6 +208,15 @@ if(CMAKE_SYSTEM_PROCESSOR MATCHES "ppc64le")
  endif(HAS_ALTIVEC)
 endif(CMAKE_SYSTEM_PROCESSOR MATCHES "ppc64le")

+if(CMAKE_SYSTEM_PROCESSOR MATCHES "aarch64|AARCH64")
+        CHECK_C_COMPILER_FLAG("-march=armv8-a+crc" HAS_ARMV8_CRC)
+  if(HAS_ARMV8_CRC)
+    message(STATUS " HAS_ARMV8_CRC yes")
+    set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -march=armv8-a+crc -Wno-unused-function")
+    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -march=armv8-a+crc -Wno-unused-function")
+  endif(HAS_ARMV8_CRC)
+endif(CMAKE_SYSTEM_PROCESSOR MATCHES "aarch64|AARCH64")
+
 option(PORTABLE "build a portable binary" OFF)
 option(FORCE_SSE42 "force building with SSE4.2, even when PORTABLE=ON" OFF)
 if(PORTABLE)
@@ -213,7 +229,7 @@ else()
  if(MSVC)
    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /arch:AVX2")
  else()
-    if(NOT HAVE_POWER8)
+    if(NOT HAVE_POWER8 AND NOT HAS_ARMV8_CRC)
      set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -march=native")
    endif()
  endif()
@@ -317,6 +333,10 @@ if(DISABLE_STALL_NOTIF)
  add_definitions(-DROCKSDB_DISABLE_STALL_NOTIFICATION)
 endif()

+option(WITH_DYNAMIC_EXTENSION "build with dynamic extension support" OFF)
+if(NOT WITH_DYNAMIC_EXTENSION)
+  add_definitions(-DROCKSDB_NO_DYNAMIC_EXTENSION)
+endif()

 if(DEFINED USE_RTTI)
  if(USE_RTTI)
@@ -416,7 +436,7 @@ if(WITH_FALLOCATE)
 #include <linux/falloc.h>
 int main() {
 int fd = open(\"/dev/null\", 0);
- fallocate(fd, FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE, 0, 1024);
+ fallocate(fd, FALLOC_FL_KEEP_SIZE, 0, 1024);
 }
 " HAVE_FALLOCATE)
  if(HAVE_FALLOCATE)
@@ -461,8 +481,6 @@ include_directories(${PROJECT_SOURCE_DIR}/include)
 include_directories(SYSTEM ${PROJECT_SOURCE_DIR}/third-party/gtest-1.7.0/fused-src)
 find_package(Threads REQUIRED)

-add_subdirectory(third-party/gtest-1.7.0/fused-src/gtest)
-
 # Main library source code

 set(SOURCES
@@ -473,22 +491,24 @@ set(SOURCES
        db/c.cc
        db/column_family.cc
        db/compacted_db_impl.cc
-        db/compaction.cc
-        db/compaction_iterator.cc
-        db/compaction_job.cc
-        db/compaction_picker.cc
-        db/compaction_picker_fifo.cc
-        db/compaction_picker_universal.cc
+        db/compaction/compaction.cc
+        db/compaction/compaction_iterator.cc
+        db/compaction/compaction_picker.cc
+        db/compaction/compaction_job.cc
+        db/compaction/compaction_picker_fifo.cc
+        db/compaction/compaction_picker_level.cc
+        db/compaction/compaction_picker_universal.cc
        db/convenience.cc
        db/db_filesnapshot.cc
-        db/db_impl.cc
-        db/db_impl_write.cc
-        db/db_impl_compaction_flush.cc
-        db/db_impl_files.cc
-        db/db_impl_open.cc
-        db/db_impl_debug.cc
-        db/db_impl_experimental.cc
-        db/db_impl_readonly.cc
+        db/db_impl/db_impl.cc
+        db/db_impl/db_impl_write.cc
+        db/db_impl/db_impl_compaction_flush.cc
+        db/db_impl/db_impl_files.cc
+        db/db_impl/db_impl_open.cc
+        db/db_impl/db_impl_debug.cc
+        db/db_impl/db_impl_experimental.cc
+        db/db_impl/db_impl_readonly.cc
+        db/db_impl/db_impl_secondary.cc
        db/db_info_dumper.cc
        db/db_iter.cc
        db/dbformat.cc
@@ -500,6 +520,7 @@ set(SOURCES
        db/flush_job.cc
        db/flush_scheduler.cc
        db/forward_iterator.cc
+	db/import_column_family_job.cc
        db/internal_stats.cc
        db/logs_with_prep_tracker.cc
        db/log_reader.cc
@@ -529,8 +550,17 @@ set(SOURCES
        env/env_encryption.cc
        env/env_hdfs.cc
        env/mock_env.cc
+        file/delete_scheduler.cc
+        file/file_util.cc
+        file/filename.cc
+        file/sst_file_manager_impl.cc
+        logging/auto_roll_logger.cc
+        logging/event_logger.cc
+        logging/log_buffer.cc
+        memory/arena.cc
+        memory/concurrent_arena.cc
+        memory/jemalloc_nodump_allocator.cc
        memtable/alloc_tracker.cc
-        memtable/hash_cuckoo_rep.cc
        memtable/hash_linklist_rep.cc
        memtable/hash_skiplist_rep.cc
        memtable/skiplistrep.cc
@@ -538,10 +568,12 @@ set(SOURCES
        memtable/write_buffer_manager.cc
        monitoring/histogram.cc
        monitoring/histogram_windowing.cc
+        monitoring/in_memory_stats_history.cc
        monitoring/instrumented_mutex.cc
        monitoring/iostats_context.cc
        monitoring/perf_context.cc
        monitoring/perf_level.cc
+        monitoring/persistent_stats_history.cc
        monitoring/statistics.cc
        monitoring/thread_status_impl.cc
        monitoring/thread_status_updater.cc
@@ -554,80 +586,73 @@ set(SOURCES
        options/options_parser.cc
        options/options_sanity_check.cc
        port/stack_trace.cc
-        table/adaptive_table_factory.cc
-        table/block.cc
-        table/block_based_filter_block.cc
-        table/block_based_table_builder.cc
-        table/block_based_table_factory.cc
-        table/block_based_table_reader.cc
-        table/block_builder.cc
+        table/adaptive/adaptive_table_factory.cc
+        table/block_based/block.cc
+        table/block_based/block_based_filter_block.cc
+        table/block_based/block_based_table_builder.cc
+        table/block_based/block_based_table_factory.cc
+        table/block_based/block_based_table_reader.cc
+        table/block_based/block_builder.cc
+        table/block_based/block_prefix_index.cc
+        table/block_based/data_block_hash_index.cc
+        table/block_based/data_block_footer.cc
+        table/block_based/filter_block_reader_common.cc
+        table/block_based/flush_block_policy.cc
+        table/block_based/full_filter_block.cc
+        table/block_based/index_builder.cc
+        table/block_based/partitioned_filter_block.cc
        table/block_fetcher.cc
-        table/block_prefix_index.cc
        table/bloom_block.cc
-        table/cuckoo_table_builder.cc
-        table/cuckoo_table_factory.cc
-        table/cuckoo_table_reader.cc
-        table/data_block_hash_index.cc
-        table/data_block_footer.cc
-        table/flush_block_policy.cc
+        table/cuckoo/cuckoo_table_builder.cc
+        table/cuckoo/cuckoo_table_factory.cc
+        table/cuckoo/cuckoo_table_reader.cc
        table/format.cc
-        table/full_filter_block.cc
        table/get_context.cc
-        table/index_builder.cc
        table/iterator.cc
        table/merging_iterator.cc
        table/meta_blocks.cc
-        table/partitioned_filter_block.cc
        table/persistent_cache_helper.cc
-        table/plain_table_builder.cc
-        table/plain_table_factory.cc
-        table/plain_table_index.cc
-        table/plain_table_key_coding.cc
-        table/plain_table_reader.cc
+        table/plain/plain_table_builder.cc
+        table/plain/plain_table_factory.cc
+        table/plain/plain_table_index.cc
+        table/plain/plain_table_key_coding.cc
+        table/plain/plain_table_reader.cc
        table/sst_file_reader.cc
        table/sst_file_writer.cc
        table/table_properties.cc
        table/two_level_iterator.cc
+        test_util/sync_point.cc
+        test_util/sync_point_impl.cc
+        test_util/testutil.cc
+        test_util/transaction_test_util.cc
+        tools/block_cache_trace_analyzer.cc
        tools/db_bench_tool.cc
        tools/dump/db_dump_tool.cc
        tools/ldb_cmd.cc
        tools/ldb_tool.cc
        tools/sst_dump_tool.cc
        tools/trace_analyzer_tool.cc
-        util/arena.cc
-        util/auto_roll_logger.cc
+        trace_replay/trace_replay.cc
+        trace_replay/block_cache_tracer.cc
        util/bloom.cc
        util/coding.cc
        util/compaction_job_stats_impl.cc
        util/comparator.cc
        util/compression_context_cache.cc
-        util/concurrent_arena.cc
        util/concurrent_task_limiter_impl.cc
        util/crc32c.cc
-        util/delete_scheduler.cc
        util/dynamic_bloom.cc
-        util/event_logger.cc
        util/file_reader_writer.cc
-        util/file_util.cc
-        util/filename.cc
        util/filter_policy.cc
        util/hash.cc
-        util/jemalloc_nodump_allocator.cc
-        util/log_buffer.cc
        util/murmurhash.cc
        util/random.cc
        util/rate_limiter.cc
        util/slice.cc
-        util/sst_file_manager_impl.cc
        util/status.cc
        util/string_util.cc
-        util/sync_point.cc
-        util/sync_point_impl.cc
-        util/testutil.cc
        util/thread_local.cc
        util/threadpool_imp.cc
-        util/trace_replay.cc
-        util/transaction_test_util.cc
        util/xxhash.cc
        utilities/backupable/backupable_db.cc
        utilities/blob_db/blob_compaction_filter.cc
@@ -648,7 +673,6 @@ set(SOURCES
        utilities/env_mirror.cc
        utilities/env_timed.cc
        utilities/leveldb_options/leveldb_options.cc
-        utilities/lua/rocks_lua_compaction_filter.cc
        utilities/memory/memory_util.cc
        utilities/merge_operators/bytesxor.cc
        utilities/merge_operators/max.cc
@@ -663,6 +687,7 @@ set(SOURCES
        utilities/persistent_cache/block_cache_tier_metadata.cc
        utilities/persistent_cache/persistent_cache_tier.cc
        utilities/persistent_cache/volatile_tier_impl.cc
+        utilities/simulator_cache/cache_simulator.cc
        utilities/simulator_cache/sim_cache.cc
        utilities/table_properties_collectors/compact_on_deletion_collector.cc
        utilities/trace/file_trace_reader_writer.cc
@@ -696,6 +721,11 @@ if(HAVE_POWER8)
    util/crc32c_ppc_asm.S)
 endif(HAVE_POWER8)

+if(HAS_ARMV8_CRC)
+  list(APPEND SOURCES
+    util/crc32c_arm64.cc)
+endif(HAS_ARMV8_CRC)
+
 if(WIN32)
  list(APPEND SOURCES
    port/win/io_win.cc
@@ -843,15 +873,16 @@ endif()

 option(WITH_TESTS "build with tests" ON)
 if(WITH_TESTS)
+  add_subdirectory(third-party/gtest-1.7.0/fused-src/gtest)
  set(TESTS
        cache/cache_test.cc
        cache/lru_cache_test.cc
        db/column_family_test.cc
        db/compact_files_test.cc
-        db/compaction_iterator_test.cc
-        db/compaction_job_stats_test.cc
-        db/compaction_job_test.cc
-        db/compaction_picker_test.cc
+        db/compaction/compaction_job_stats_test.cc
+        db/compaction/compaction_job_test.cc
+        db/compaction/compaction_iterator_test.cc
+        db/compaction/compaction_picker_test.cc
        db/comparator_db_test.cc
        db/corruption_test.cc
        db/cuckoo_table_db_test.cc
@@ -874,6 +905,7 @@ if(WITH_TESTS)
        db/db_options_test.cc
        db/db_properties_test.cc
        db/db_range_del_test.cc
+        db/db_impl/db_secondary_test.cc
        db/db_sst_test.cc
        db/db_statistics_test.cc
        db/db_table_properties_test.cc
@@ -917,37 +949,41 @@ if(WITH_TESTS)
        env/env_basic_test.cc
        env/env_test.cc
        env/mock_env_test.cc
+        file/delete_scheduler_test.cc
+        logging/auto_roll_logger_test.cc
+        logging/env_logger_test.cc
+        logging/event_logger_test.cc
+        memory/arena_test.cc
        memtable/inlineskiplist_test.cc
        memtable/skiplist_test.cc
        memtable/write_buffer_manager_test.cc
        monitoring/histogram_test.cc
        monitoring/iostats_context_test.cc
        monitoring/statistics_test.cc
+        monitoring/stats_history_test.cc
        options/options_settable_test.cc
        options/options_test.cc
-        table/block_based_filter_block_test.cc
-        table/block_test.cc
+        table/block_based/block_based_filter_block_test.cc
+        table/block_based/block_test.cc
+        table/block_based/data_block_hash_index_test.cc
+        table/block_based/full_filter_block_test.cc
+        table/block_based/partitioned_filter_block_test.cc
        table/cleanable_test.cc
-        table/cuckoo_table_builder_test.cc
-        table/cuckoo_table_reader_test.cc
-        table/data_block_hash_index_test.cc
-        table/full_filter_block_test.cc
+        table/cuckoo/cuckoo_table_builder_test.cc
+        table/cuckoo/cuckoo_table_reader_test.cc
        table/merger_test.cc
        table/sst_file_reader_test.cc
        table/table_test.cc
+        tools/block_cache_trace_analyzer_test.cc
        tools/ldb_cmd_test.cc
        tools/reduce_levels_test.cc
        tools/sst_dump_test.cc
        tools/trace_analyzer_test.cc
-        util/arena_test.cc
-        util/auto_roll_logger_test.cc
        util/autovector_test.cc
        util/bloom_test.cc
        util/coding_test.cc
        util/crc32c_test.cc
-        util/delete_scheduler_test.cc
        util/dynamic_bloom_test.cc
-        util/event_logger_test.cc
        util/file_reader_writer_test.cc
        util/filelock_test.cc
        util/hash_test.cc
@@ -965,7 +1001,6 @@ if(WITH_TESTS)
        utilities/cassandra/cassandra_row_merge_test.cc
        utilities/cassandra/cassandra_serialize_test.cc
        utilities/checkpoint/checkpoint_test.cc
-        utilities/lua/rocks_lua_test.cc
        utilities/memory/memory_test.cc
        utilities/merge_operators/string_append/stringappend_test.cc
        utilities/object_registry_test.cc
@@ -973,6 +1008,7 @@ if(WITH_TESTS)
        utilities/options/options_util_test.cc
        utilities/persistent_cache/hash_table_test.cc
        utilities/persistent_cache/persistent_cache_test.cc
+        utilities/simulator_cache/cache_simulator_test.cc
        utilities/simulator_cache/sim_cache_test.cc
        utilities/table_properties_collectors/compact_on_deletion_collector_test.cc
        utilities/transactions/optimistic_transaction_test.cc
@@ -993,7 +1029,7 @@ if(WITH_TESTS)
    tools/db_bench.cc
    table/table_reader_bench.cc
    utilities/persistent_cache/hash_table_bench.cc)
-  add_library(testharness OBJECT util/testharness.cc)
+  add_library(testharness OBJECT test_util/testharness.cc)
  foreach(sourcefile ${BENCHMARKS})
    get_filename_component(exename ${sourcefile} NAME_WE)
    add_executable(${exename}${ARTIFACT_SUFFIX} ${sourcefile}
@@ -1007,7 +1043,7 @@ if(WITH_TESTS)
      db/db_test_util.cc
      monitoring/thread_status_updater_debug.cc
      table/mock_table.cc
-      util/fault_injection_test_env.cc
+      test_util/fault_injection_test_env.cc
      utilities/cassandra/test_utils.cc
  )
  # test utilities are only build in debug
@@ -1,5 +1,103 @@
 # Rocksdb Change Log
 ## Unreleased
+### Default Option Change
+* LRUCacheOptions.high_pri_pool_ratio is set to 0.5 (previously 0.0) by default, which means that by default midpoint insertion is enabled. The same change is made for the default value of high_pri_pool_ratio argument in NewLRUCache(). When block cache is not explictly created, the small block cache created by BlockBasedTable will still has this option to be 0.0.
+* Change BlockBasedTableOptions.cache_index_and_filter_blocks_with_high_priority's default value from false to true.
+
+### Public API Change
+* Now DB::Close() will return Aborted() error when there is unreleased snapshot. Users can retry after all snapshots are released.
+* Index and filter blocks are now handled similarly to data blocks with regards to the block cache: instead of storing reader objects in the cache, only the blocks themselves are cached. In addition, index and filter blocks (as well as filter partitions) no longer get evicted from the cache when a table is closed. Moreover, index blocks can now use the compressed block cache (if any).
+* Partitions of partitioned indexes no longer affect the read amplification statistics.
+* Due to the above refactoring, block cache eviction statistics for indexes and filters are temporarily broken. We plan to reintroduce them in a later phase.
+* options.keep_log_file_num will be enforced strictly all the time. File names of all log files will be tracked, which may take significantly amount of memory if options.keep_log_file_num is large and either of options.max_log_file_size or options.log_file_time_to_roll is set.
+* Add initial support for Get/Put with user timestamps. Users can specify timestamps via ReadOptions and WriteOptions when calling DB::Get and DB::Put.
+* Accessing a partition of a partitioned filter or index through a pinned reference is no longer considered a cache hit.
+* The semantics of the per-block-type block read counts in the performance context now match those of the generic block_read_count.
+* Add C bindings for secondary instance, i.e. DBImplSecondary.
+* db_bench adds a "benchmark" stats_history, which prints out the whole stats history.
+* Rate limited deletion of WALs is only enabled if DBOptions::wal_dir is not set, or explicitly set to db_name passed to DB::Open and DBOptions::db_paths is empty, or same as db_paths[0].path
+* Overload GetAllKeyVersions() to support non-default column family.
+* Added new APIs ExportColumnFamily() and CreateColumnFamilyWithImport() to support export and import of a Column Family. https://github.com/facebook/rocksdb/issues/3469
+
+### New Features
+* Add an option `snap_refresh_nanos` (default to 0.1s) to periodically refresh the snapshot list in compaction jobs. Assign to 0 to disable the feature.
+* Add an option `unordered_write` which trades snapshot guarantees with higher write throughput. When used with WRITE_PREPARED transactions with two_write_queues=true, it offers higher throughput with however no compromise on guarantees.
+* Allow DBImplSecondary to remove memtables with obsolete data after replaying MANIFEST and WAL.
+* Add an option `failed_move_fall_back_to_copy` (default is true) for external SST ingestion. When `move_files` is true and hard link fails, ingestion falls back to copy if `failed_move_fall_back_to_copy` is true. Otherwise, ingestion reports an error.
+* Add argument `--secondary_path` to ldb to open the database as the secondary instance. This would keep the original DB intact.
+
+### Performance Improvements
+* Reduce binary search when iterator reseek into the same data block.
+* DBIter::Next() can skip user key checking if previous entry's seqnum is 0.
+* Merging iterator to avoid child iterator reseek for some cases
+* Reduce iterator key comparision for upper/lower bound check.
+* Log Writer will flush after finishing the whole record, rather than a fragment.
+* Lower MultiGet batching API latency by reading data blocks from disk in parallel
+
+### General Improvements
+* Added new status code kColumnFamilyDropped to distinguish between Column Family Dropped and DB Shutdown in progress.
+* Improve ColumnFamilyOptions validation when creating a new column family.
+
+### Bug Fixes
+* Fix a bug in WAL replay of secondary instance by skipping write batches with older sequence numbers than the current last sequence number.
+* Fix flush's/compaction's merge processing logic which allowed `Put`s covered by range tombstones to reappear. Note `Put`s may exist even if the user only ever called `Merge()` due to an internal conversion during compaction to the bottommost level.
+* Fix/improve memtable earliest sequence assignment and WAL replay so that WAL entries of unflushed column families will not be skipped after replaying the MANIFEST and increasing db sequence due to another flushed/compacted column family.
+* Fix ingested file and directory not being fsync.
+* Return TryAgain status in place of Corruption when new tail is not visible to TransactionLogIterator.
+* Fix a bug caused by secondary not skipping the beginning of new MANIFEST.
+* On DB open, delete WAL trash files left behind in wal_dir
+
+## 6.2.0 (4/30/2019)
+### New Features
+* Add an option `strict_bytes_per_sync` that causes a file-writing thread to block rather than exceed the limit on bytes pending writeback specified by `bytes_per_sync` or `wal_bytes_per_sync`.
+* Improve range scan performance by avoiding per-key upper bound check in BlockBasedTableIterator.
+* Introduce Periodic Compaction for Level style compaction. Files are re-compacted periodically and put in the same level.
+* Block-based table index now contains exact highest key in the file, rather than an upper bound. This may improve Get() and iterator Seek() performance in some situations, especially when direct IO is enabled and block cache is disabled. A setting BlockBasedTableOptions::index_shortening is introduced to control this behavior. Set it to kShortenSeparatorsAndSuccessor to get the old behavior.
+* When reading from option file/string/map, customized envs can be filled according to object registry.
+* Improve range scan performance when using explicit user readahead by not creating new table readers for every iterator.
+* Add index type BlockBasedTableOptions::IndexType::kBinarySearchWithFirstKey. It significantly reduces read amplification in some setups, especially for iterator seeks. It's not fully implemented yet: IO errors are not handled right.
+
+### Public API Change
+* Change the behavior of OptimizeForPointLookup(): move away from hash-based block-based-table index, and use whole key memtable filtering.
+* Change the behavior of OptimizeForSmallDb(): use a 16MB block cache, put index and filter blocks into it, and cost the memtable size to it. DBOptions.OptimizeForSmallDb() and ColumnFamilyOptions.OptimizeForSmallDb() start to take an optional cache object.
+* Added BottommostLevelCompaction::kForceOptimized to avoid double compacting newly compacted files in the bottommost level compaction of manual compaction. Note this option may prohibit the manual compaction to produce a single file in the bottommost level.
+
+### Bug Fixes
+* Adjust WriteBufferManager's dummy entry size to block cache from 1MB to 256KB.
+* Fix a race condition between WritePrepared::Get and ::Put with duplicate keys.
+* Fix crash when memtable prefix bloom is enabled and read/write a key out of domain of prefix extractor.
+* Close a WAL file before another thread deletes it.
+* Fix an assertion failure `IsFlushPending() == true` caused by one bg thread releasing the db mutex in ~ColumnFamilyData and another thread clearing `flush_requested_` flag.
+
+## 6.1.1 (4/9/2019)
+### New Features
+* When reading from option file/string/map, customized comparators and/or merge operators can be filled according to object registry.
+
+### Public API Change
+
+### Bug Fixes
+* Fix a bug in 2PC where a sequence of txn prepare, memtable flush, and crash could result in losing the prepared transaction.
+* Fix a bug in Encryption Env which could cause encrypted files to be read beyond file boundaries.
+
+## 6.1.0 (3/27/2019)
+### New Features
+* Introduce two more stats levels, kExceptHistogramOrTimers and kExceptTimers.
+* Added a feature to perform data-block sampling for compressibility, and report stats to user.
+* Add support for trace filtering.
+* Add DBOptions.avoid_unnecessary_blocking_io. If true, we avoid file deletion when destorying ColumnFamilyHandle and Iterator. Instead, a job is scheduled to delete the files in background.
+
+### Public API Change
+* Remove bundled fbson library.
+* statistics.stats_level_ becomes atomic. It is preferred to use statistics.set_stats_level() and statistics.get_stats_level() to access it.
+* Introduce a new IOError subcode, PathNotFound, to indicate trying to open a nonexistent file or directory for read.
+* Add initial support for multiple db instances sharing the same data in single-writer, multi-reader mode.
+* Removed some "using std::xxx" from public headers.
+
+### Bug Fixes
+* Fix JEMALLOC_CXX_THROW macro missing from older Jemalloc versions, causing build failures on some platforms.
+* Fix SstFileReader not able to open file ingested with write_glbal_seqno=true.
+
+## 6.0.0 (2/19/2019)
 ### New Features
 * Enabled checkpoint on readonly db (DBImplReadOnly).
 * Make DB ignore dropped column families while committing results of atomic flush.
@@ -7,8 +105,16 @@
 * For users of dictionary compression with ZSTD v0.7.0+, we now reuse the same digested dictionary when compressing each of an SST file's data blocks for faster compression speeds.
 * For all users of dictionary compression who set `cache_index_and_filter_blocks == true`, we now store dictionary data used for decompression in the block cache for better control over memory usage. For users of ZSTD v1.1.4+ who compile with -DZSTD_STATIC_LINKING_ONLY, this includes a digested dictionary, which is used to increase decompression speed.
 * Add support for block checksums verification for external SST files before ingestion.
+* Introduce stats history which periodically saves Statistics snapshots and added `GetStatsHistory` API to retrieve these snapshots.
+* Add a place holder in manifest which indicate a record from future that can be safely ignored.
+* Add support for trace sampling.
+* Enable properties block checksum verification for block-based tables.
+* For all users of dictionary compression, we now generate a separate dictionary for compressing each bottom-level SST file. Previously we reused a single dictionary for a whole compaction to bottom level. The new approach achieves better compression ratios; however, it uses more memory and CPU for buffering/sampling data blocks and training dictionaries.
+* Add whole key bloom filter support in memtable.
+* Files written by `SstFileWriter` will now use dictionary compression if it is configured in the file writer's `CompressionOptions`.

 ### Public API Change
+* Disallow CompactionFilter::IgnoreSnapshots() = false, because it is not very useful and the behavior is confusing. The filter will filter everything if there is no snapshot declared by the time the compaction starts. However, users can define a snapshot after the compaction starts and before it finishes and this new snapshot won't be repeatable, because after the compaction finishes, some keys may be dropped.
 * CompactionPri = kMinOverlappingRatio also uses compensated file size, which boosts file with lots of tombstones to be compacted first.
 * Transaction::GetForUpdate is extended with a do_validate parameter with default value of true. If false it skips validating the snapshot before doing the read. Similarly ::Merge, ::Put, ::Delete, and ::SingleDelete are extended with assume_tracked with default value of false. If true it indicates that call is assumed to be after a ::GetForUpdate.
 * `TableProperties::num_entries` and `TableProperties::num_deletions` now also account for number of range tombstones.
@@ -16,7 +122,11 @@
 * With "ldb ----try_load_options", when wal_dir specified by the option file doesn't exist, ignore it.
 * Change time resolution in FileOperationInfo.
 * Deleting Blob files also go through SStFileManager.
-* Remove PlainTable's store_index_in_file feature. When opening an existing DB with index in SST files, the index and bloom filter will still be rebuild while SST files are opened, in the same way as there is no index in the file.
+* Remove CuckooHash memtable.
+* The counter stat `number.block.not_compressed` now also counts blocks not compressed due to poor compression ratio.
+* Remove ttl option from `CompactionOptionsFIFO`. The option has been deprecated and ttl in `ColumnFamilyOptions` is used instead.
+* Support SST file ingestion across multiple column families via DB::IngestExternalFiles. See the function's comment about atomicity.
+* Remove Lua compaction filter.

 ### Bug Fixes
 * Fix a deadlock caused by compaction and file ingestion waiting for each other in the event of write stalls.
@@ -1,7 +1,9 @@
 This is the list of all known third-party language bindings for RocksDB. If something is missing, please open a pull request to add it.

 * Java - https://github.com/facebook/rocksdb/tree/master/java
-* Python - http://pyrocksdb.readthedocs.org/en/latest/
+* Python
+    * http://python-rocksdb.readthedocs.io/en/latest/
+    * http://pyrocksdb.readthedocs.org/en/latest/ (unmaintained)
 * Perl - https://metacpan.org/pod/RocksDB
 * Node.js - https://npmjs.org/package/rocksdb
 * Go - https://github.com/tecbot/gorocksdb
@@ -82,17 +82,23 @@ ifeq ($(MAKECMDGOALS),rocksdbjavastatic)
 endif

 ifeq ($(MAKECMDGOALS),rocksdbjavastaticrelease)
-	DEBUG_LEVEL=0
+	ifneq ($(DEBUG_LEVEL),2)
+		DEBUG_LEVEL=0
+	endif
 endif

 ifeq ($(MAKECMDGOALS),rocksdbjavastaticreleasedocker)
-        DEBUG_LEVEL=0
+	ifneq ($(DEBUG_LEVEL),2)
+        	DEBUG_LEVEL=0
+	endif
 endif

 ifeq ($(MAKECMDGOALS),rocksdbjavastaticpublish)
 	DEBUG_LEVEL=0
 endif

+$(info $$DEBUG_LEVEL is ${DEBUG_LEVEL})
+
 # Lite build flag.
 LITE ?= 0
 ifeq ($(LITE), 0)
@@ -137,6 +143,12 @@ CFLAGS +=  -DHAVE_POWER8
 HAVE_POWER8=1
 endif

+ifeq (,$(shell $(CXX) -fsyntax-only -march=armv8-a+crc -xc /dev/null 2>&1))
+CXXFLAGS += -march=armv8-a+crc+crypto
+CFLAGS += -march=armv8-a+crc+crypto
+ARMCRC_SOURCE=1
+endif
+
 # if we're compiling for release, compile without debug code (-DNDEBUG)
 ifeq ($(DEBUG_LEVEL),0)
 OPT += -DNDEBUG
@@ -392,8 +404,8 @@ LIBOBJECTS += $(TOOL_LIB_SOURCES:.cc=.o)
 MOCKOBJECTS = $(MOCK_LIB_SOURCES:.cc=.o)

 GTEST = $(GTEST_DIR)/gtest/gtest-all.o
-TESTUTIL = ./util/testutil.o
-TESTHARNESS = ./util/testharness.o $(TESTUTIL) $(MOCKOBJECTS) $(GTEST)
+TESTUTIL = ./test_util/testutil.o
+TESTHARNESS = ./test_util/testharness.o $(TESTUTIL) $(MOCKOBJECTS) $(GTEST)
 VALGRIND_ERROR = 2
 VALGRIND_VER := $(join $(VALGRIND_VER),valgrind)

@@ -420,6 +432,7 @@ TESTS = \
 	inlineskiplist_test \
 	env_basic_test \
 	env_test \
+	env_logger_test \
 	hash_test \
 	thread_local_test \
 	rate_limiter_test \
@@ -443,6 +456,7 @@ TESTS = \
 	db_merge_operator_test \
 	db_options_test \
 	db_range_del_test \
+	db_secondary_test \
 	db_sst_test \
 	db_tailing_iter_test \
 	db_io_failure_test \
@@ -486,6 +500,7 @@ TESTS = \
 	plain_table_db_test \
 	comparator_db_test \
 	external_sst_file_test \
+	import_column_family_test \
 	prefix_test \
 	skiplist_test \
 	write_buffer_manager_test \
@@ -496,6 +511,7 @@ TESTS = \
 	cassandra_serialize_test \
 	ttl_test \
 	backupable_db_test \
+	cache_simulator_test \
 	sim_cache_test \
 	version_edit_test \
 	version_set_test \
@@ -535,7 +551,7 @@ TESTS = \
 	ldb_cmd_test \
 	persistent_cache_test \
 	statistics_test \
-	lua_test \
+	stats_history_test \
 	lru_cache_test \
 	object_registry_test \
 	repair_test \
@@ -548,6 +564,9 @@ TESTS = \
 	range_tombstone_fragmenter_test \
 	range_del_aggregator_test \
 	sst_file_reader_test \
+	db_secondary_test \
+	block_cache_tracer_test \
+	block_cache_trace_analyzer_test \

 PARALLEL_TEST = \
 	backupable_db_test \
@@ -559,6 +578,7 @@ PARALLEL_TEST = \
 	db_universal_compaction_test \
 	db_wal_test \
 	external_sst_file_test \
+	import_column_family_test \
 	fault_injection_test \
 	inlineskiplist_test \
 	manual_compaction_test \
@@ -592,6 +612,7 @@ TOOLS = \
 	rocksdb_undump \
 	blob_dump \
 	trace_analyzer \
+	block_cache_trace_analyzer \

 TEST_LIBS = \
 	librocksdb_env_basic_test.a
@@ -687,7 +708,8 @@ endif  # PLATFORM_SHARED_EXT
 .PHONY: blackbox_crash_test check clean coverage crash_test ldb_tests package \
 	release tags tags0 valgrind_check whitebox_crash_test format static_lib shared_lib all \
 	dbg rocksdbjavastatic rocksdbjava install install-static install-shared uninstall \
-	analyze tools tools_lib
+	analyze tools tools_lib \
+	blackbox_crash_test_with_atomic_flush whitebox_crash_test_with_atomic_flush


 all: $(LIBRARY) $(BENCHMARKS) tools tools_lib test_libs $(TESTS)
@@ -895,23 +917,29 @@ ldb_tests: ldb

 crash_test: whitebox_crash_test blackbox_crash_test

+crash_test_with_atomic_flush: whitebox_crash_test_with_atomic_flush blackbox_crash_test_with_atomic_flush
+
 blackbox_crash_test: db_stress
 	python -u tools/db_crashtest.py --simple blackbox $(CRASH_TEST_EXT_ARGS)
-	python -u tools/db_crashtest.py --enable_atomic_flush blackbox $(CRASH_TEST_EXT_ARGS)
 	python -u tools/db_crashtest.py blackbox $(CRASH_TEST_EXT_ARGS)

+blackbox_crash_test_with_atomic_flush: db_stress
+	python -u tools/db_crashtest.py --enable_atomic_flush blackbox $(CRASH_TEST_EXT_ARGS)
+
 ifeq ($(CRASH_TEST_KILL_ODD),)
  CRASH_TEST_KILL_ODD=888887
 endif

 whitebox_crash_test: db_stress
 	python -u tools/db_crashtest.py --simple whitebox --random_kill_odd \
-      $(CRASH_TEST_KILL_ODD) $(CRASH_TEST_EXT_ARGS)
-	python -u tools/db_crashtest.py --enable_atomic_flush whitebox  --random_kill_odd \
      $(CRASH_TEST_KILL_ODD) $(CRASH_TEST_EXT_ARGS)
 	python -u tools/db_crashtest.py whitebox  --random_kill_odd \
      $(CRASH_TEST_KILL_ODD) $(CRASH_TEST_EXT_ARGS)

+whitebox_crash_test_with_atomic_flush: db_stress
+	python -u tools/db_crashtest.py --enable_atomic_flush whitebox  --random_kill_odd \
+      $(CRASH_TEST_KILL_ODD) $(CRASH_TEST_EXT_ARGS)
+
 asan_check:
 	$(MAKE) clean
 	COMPILE_WITH_ASAN=1 $(MAKE) check -j32
@@ -922,6 +950,11 @@ asan_crash_test:
 	COMPILE_WITH_ASAN=1 $(MAKE) crash_test
 	$(MAKE) clean

+asan_crash_test_with_atomic_flush:
+	$(MAKE) clean
+	COMPILE_WITH_ASAN=1 $(MAKE) crash_test_with_atomic_flush
+	$(MAKE) clean
+
 ubsan_check:
 	$(MAKE) clean
 	COMPILE_WITH_UBSAN=1 $(MAKE) check -j32
@@ -932,6 +965,11 @@ ubsan_crash_test:
 	COMPILE_WITH_UBSAN=1 $(MAKE) crash_test
 	$(MAKE) clean

+ubsan_crash_test_with_atomic_flush:
+	$(MAKE) clean
+	COMPILE_WITH_UBSAN=1 $(MAKE) crash_test_with_atomic_flush
+	$(MAKE) clean
+
 valgrind_test:
 	ROCKSDB_VALGRIND_RUN=1 DISABLE_JEMALLOC=1 $(MAKE) valgrind_check

@@ -1076,6 +1114,9 @@ db_bench: tools/db_bench.o $(BENCHTOOLOBJECTS)
 trace_analyzer: tools/trace_analyzer.o $(ANALYZETOOLOBJECTS) $(LIBOBJECTS)
 	$(AM_LINK)

+block_cache_trace_analyzer: tools/block_cache_trace_analyzer_tool.o $(ANALYZETOOLOBJECTS) $(LIBOBJECTS)
+	$(AM_LINK)
+
 cache_bench: cache/cache_bench.o $(LIBOBJECTS) $(TESTUTIL)
 	$(AM_LINK)

@@ -1097,7 +1138,7 @@ db_sanity_test: tools/db_sanity_test.o $(LIBOBJECTS) $(TESTUTIL)
 db_repl_stress: tools/db_repl_stress.o $(LIBOBJECTS) $(TESTUTIL)
 	$(AM_LINK)

-arena_test: util/arena_test.o $(LIBOBJECTS) $(TESTHARNESS)
+arena_test: memory/arena_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

 autovector_test: util/autovector_test.o $(LIBOBJECTS) $(TESTHARNESS)
@@ -1235,6 +1276,9 @@ external_sst_file_basic_test: db/external_sst_file_basic_test.o db/db_test_util.
 external_sst_file_test: db/external_sst_file_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

+import_column_family_test: db/import_column_family_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
+	$(AM_LINK)
+
 db_tailing_iter_test: db/db_tailing_iter_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

@@ -1283,6 +1327,9 @@ backupable_db_test: utilities/backupable/backupable_db_test.o $(LIBOBJECTS) $(TE
 checkpoint_test: utilities/checkpoint/checkpoint_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

+cache_simulator_test: utilities/simulator_cache/cache_simulator_test.o $(LIBOBJECTS) $(TESTHARNESS)
+	$(AM_LINK)
+
 sim_cache_test: utilities/simulator_cache/sim_cache_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

@@ -1309,13 +1356,13 @@ write_batch_with_index_test: utilities/write_batch_with_index/write_batch_with_i
 flush_job_test: db/flush_job_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

-compaction_iterator_test: db/compaction_iterator_test.o $(LIBOBJECTS) $(TESTHARNESS)
+compaction_iterator_test: db/compaction/compaction_iterator_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

-compaction_job_test: db/compaction_job_test.o $(LIBOBJECTS) $(TESTHARNESS)
+compaction_job_test: db/compaction/compaction_job_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

-compaction_job_stats_test: db/compaction_job_stats_test.o $(LIBOBJECTS) $(TESTHARNESS)
+compaction_job_stats_test: db/compaction/compaction_job_stats_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

 compact_on_deletion_collector_test: utilities/table_properties_collectors/compact_on_deletion_collector_test.o $(LIBOBJECTS) $(TESTHARNESS)
@@ -1339,7 +1386,7 @@ fault_injection_test: db/fault_injection_test.o $(LIBOBJECTS) $(TESTHARNESS)
 rate_limiter_test: util/rate_limiter_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

-delete_scheduler_test: util/delete_scheduler_test.o $(LIBOBJECTS) $(TESTHARNESS)
+delete_scheduler_test: file/delete_scheduler_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

 filename_test: db/filename_test.o $(LIBOBJECTS) $(TESTHARNESS)
@@ -1348,13 +1395,13 @@ filename_test: db/filename_test.o $(LIBOBJECTS) $(TESTHARNESS)
 file_reader_writer_test: util/file_reader_writer_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

-block_based_filter_block_test: table/block_based_filter_block_test.o $(LIBOBJECTS) $(TESTHARNESS)
+block_based_filter_block_test: table/block_based/block_based_filter_block_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

-full_filter_block_test: table/full_filter_block_test.o $(LIBOBJECTS) $(TESTHARNESS)
+full_filter_block_test: table/block_based/full_filter_block_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

-partitioned_filter_block_test: table/partitioned_filter_block_test.o $(LIBOBJECTS) $(TESTHARNESS)
+partitioned_filter_block_test: table/block_based/partitioned_filter_block_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

 log_test: db/log_test.o $(LIBOBJECTS) $(TESTHARNESS)
@@ -1366,10 +1413,10 @@ cleanable_test: table/cleanable_test.o $(LIBOBJECTS) $(TESTHARNESS)
 table_test: table/table_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

-block_test: table/block_test.o $(LIBOBJECTS) $(TESTHARNESS)
+block_test: table/block_based/block_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

-data_block_hash_index_test: table/data_block_hash_index_test.o $(LIBOBJECTS) $(TESTHARNESS)
+data_block_hash_index_test: table/block_based/data_block_hash_index_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

 inlineskiplist_test: memtable/inlineskiplist_test.o $(LIBOBJECTS) $(TESTHARNESS)
@@ -1387,7 +1434,7 @@ version_edit_test: db/version_edit_test.o $(LIBOBJECTS) $(TESTHARNESS)
 version_set_test: db/version_set_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

-compaction_picker_test: db/compaction_picker_test.o $(LIBOBJECTS) $(TESTHARNESS)
+compaction_picker_test: db/compaction/compaction_picker_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

 version_builder_test: db/version_builder_test.o $(LIBOBJECTS) $(TESTHARNESS)
@@ -1435,10 +1482,10 @@ rocksdb_dump: tools/dump/rocksdb_dump.o $(LIBOBJECTS)
 rocksdb_undump: tools/dump/rocksdb_undump.o $(LIBOBJECTS)
 	$(AM_LINK)

-cuckoo_table_builder_test: table/cuckoo_table_builder_test.o $(LIBOBJECTS) $(TESTHARNESS)
+cuckoo_table_builder_test: table/cuckoo/cuckoo_table_builder_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

-cuckoo_table_reader_test: table/cuckoo_table_reader_test.o $(LIBOBJECTS) $(TESTHARNESS)
+cuckoo_table_reader_test: table/cuckoo/cuckoo_table_reader_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

 cuckoo_table_db_test: db/cuckoo_table_db_test.o $(LIBOBJECTS) $(TESTHARNESS)
@@ -1468,7 +1515,7 @@ db_bench_tool_test: tools/db_bench_tool_test.o $(BENCHTOOLOBJECTS) $(TESTHARNESS
 trace_analyzer_test: tools/trace_analyzer_test.o $(LIBOBJECTS) $(ANALYZETOOLOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

-event_logger_test: util/event_logger_test.o $(LIBOBJECTS) $(TESTHARNESS)
+event_logger_test: logging/event_logger_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

 timer_queue_test: util/timer_queue_test.o $(LIBOBJECTS) $(TESTHARNESS)
@@ -1489,7 +1536,10 @@ manual_compaction_test: db/manual_compaction_test.o $(LIBOBJECTS) $(TESTHARNESS)
 filelock_test: util/filelock_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

-auto_roll_logger_test: util/auto_roll_logger_test.o $(LIBOBJECTS) $(TESTHARNESS)
+auto_roll_logger_test: logging/auto_roll_logger_test.o $(LIBOBJECTS) $(TESTHARNESS)
+	$(AM_LINK)
+
+env_logger_test: logging/env_logger_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

 memtable_list_test: db/memtable_list_test.o $(LIBOBJECTS) $(TESTHARNESS)
@@ -1534,10 +1584,10 @@ persistent_cache_test: utilities/persistent_cache/persistent_cache_test.o  db/db
 statistics_test: monitoring/statistics_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

-lru_cache_test: cache/lru_cache_test.o $(LIBOBJECTS) $(TESTHARNESS)
+stats_history_test: monitoring/stats_history_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

-lua_test: utilities/lua/rocks_lua_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
+lru_cache_test: cache/lru_cache_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

 range_del_aggregator_test: db/range_del_aggregator_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
@@ -1558,6 +1608,15 @@ range_tombstone_fragmenter_test: db/range_tombstone_fragmenter_test.o db/db_test
 sst_file_reader_test: table/sst_file_reader_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

+db_secondary_test: db/db_impl/db_secondary_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
+	$(AM_LINK)
+
+block_cache_tracer_test: trace_replay/block_cache_tracer_test.o trace_replay/block_cache_tracer.o $(LIBOBJECTS) $(TESTHARNESS)
+	$(AM_LINK)
+
+block_cache_trace_analyzer_test: tools/block_cache_trace_analyzer_test.o tools/block_cache_trace_analyzer.o $(LIBOBJECTS) $(TESTHARNESS)
+	$(AM_LINK)
+
 #-------------------------------------------------
 # make install related stuff
 INSTALL_PATH ?= /usr/local
@@ -1603,7 +1662,7 @@ JAVA_INCLUDE = -I$(JAVA_HOME)/include/ -I$(JAVA_HOME)/include/linux
 ifeq ($(PLATFORM), OS_SOLARIS)
 	ARCH := $(shell isainfo -b)
 else ifeq ($(PLATFORM), OS_OPENBSD)
-	ifneq (,$(filter $(MACHINE), amd64 arm64 sparc64))
+	ifneq (,$(filter $(MACHINE), amd64 arm64 sparc64 aarch64))
 		ARCH := 64
 	else
 		ARCH := 32
@@ -1617,6 +1676,9 @@ ifeq (,$(findstring ppc,$(MACHINE)))
 else
        ROCKSDBJNILIB = librocksdbjni-linux-$(MACHINE).so
 endif
+ifneq (,$(findstring aarch64,$(MACHINE)))
+        ROCKSDBJNILIB = librocksdbjni-linux-$(MACHINE).so
+endif
 ROCKSDB_JAR = rocksdbjni-$(ROCKSDB_MAJOR).$(ROCKSDB_MINOR).$(ROCKSDB_PATCH)-linux$(ARCH).jar
 ROCKSDB_JAR_ALL = rocksdbjni-$(ROCKSDB_MAJOR).$(ROCKSDB_MINOR).$(ROCKSDB_PATCH).jar
 ROCKSDB_JAVADOCS_JAR = rocksdbjni-$(ROCKSDB_MAJOR).$(ROCKSDB_MINOR).$(ROCKSDB_PATCH)-javadoc.jar
@@ -1635,8 +1697,8 @@ SNAPPY_DOWNLOAD_BASE ?= https://github.com/google/snappy/archive
 LZ4_VER ?= 1.8.3
 LZ4_SHA256 ?= 33af5936ac06536805f9745e0b6d61da606a1f8b4cc5c04dd3cbaca3b9b4fc43
 LZ4_DOWNLOAD_BASE ?= https://github.com/lz4/lz4/archive
-ZSTD_VER ?= 1.3.7
-ZSTD_SHA256 ?= 5dd1e90eb16c25425880c8a91327f63de22891ffed082fcc17e5ae84fce0d5fb
+ZSTD_VER ?= 1.4.0
+ZSTD_SHA256 ?= 63be339137d2b683c6d19a9e34f4fb684790e864fee13c7dd40e197a64c705c1
 ZSTD_DOWNLOAD_BASE ?= https://github.com/facebook/zstd/archive
 CURL_SSL_OPTS ?= --tlsv1

@@ -1713,7 +1775,7 @@ endif
 	fi
 	tar xvzf snappy-$(SNAPPY_VER).tar.gz
 	mkdir snappy-$(SNAPPY_VER)/build
-	cd snappy-$(SNAPPY_VER)/build && CFLAGS='${EXTRA_CFLAGS}' CXXFLAGS='${EXTRA_CXXFLAGS}' LDFLAGS='${EXTRA_LDFLAGS}' cmake .. && $(MAKE) ${SNAPPY_MAKE_TARGET}
+	cd snappy-$(SNAPPY_VER)/build && CFLAGS='${EXTRA_CFLAGS}' CXXFLAGS='${EXTRA_CXXFLAGS}' LDFLAGS='${EXTRA_LDFLAGS}' cmake -DCMAKE_POSITION_INDEPENDENT_CODE=ON .. && $(MAKE) ${SNAPPY_MAKE_TARGET}
 	cp snappy-$(SNAPPY_VER)/build/libsnappy.a .

 liblz4.a:
@@ -1803,27 +1865,15 @@ rocksdbjavastaticreleasedocker: rocksdbjavastatic rocksdbjavastaticdockerx86 roc

 rocksdbjavastaticdockerx86:
 	mkdir -p java/target
-	DOCKER_LINUX_X86_CONTAINER=`docker ps -aqf name=rocksdb_linux_x86-be`; \
-	if [ -z "$$DOCKER_LINUX_X86_CONTAINER" ]; then \
-		docker container create --attach stdin --attach stdout --attach stderr --volume `pwd`:/rocksdb-host --name rocksdb_linux_x86-be evolvedbinary/rocksjava:centos6_x86-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh; \
-	fi
-	docker start -a rocksdb_linux_x86-be
+	docker run --rm --name rocksdb_linux_x86-be --attach stdin --attach stdout --attach stderr --volume `pwd`:/rocksdb-host --env DEBUG_LEVEL=$(DEBUG_LEVEL) evolvedbinary/rocksjava:centos6_x86-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh

 rocksdbjavastaticdockerx86_64:
 	mkdir -p java/target
-	DOCKER_LINUX_X64_CONTAINER=`docker ps -aqf name=rocksdb_linux_x64-be`; \
-	if [ -z "$$DOCKER_LINUX_X64_CONTAINER" ]; then \
-		docker container create --attach stdin --attach stdout --attach stderr --volume `pwd`:/rocksdb-host --name rocksdb_linux_x64-be evolvedbinary/rocksjava:centos6_x64-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh; \
-	fi
-	docker start -a rocksdb_linux_x64-be
+	docker run --rm --name rocksdb_linux_x64-be --attach stdin --attach stdout --attach stderr --volume `pwd`:/rocksdb-host --env DEBUG_LEVEL=$(DEBUG_LEVEL) evolvedbinary/rocksjava:centos6_x64-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh

 rocksdbjavastaticdockerppc64le:
 	mkdir -p java/target
-	DOCKER_LINUX_PPC64LE_CONTAINER=`docker ps -aqf name=rocksdb_linux_ppc64le-be`; \
-	if [ -z "$$DOCKER_LINUX_PPC64LE_CONTAINER" ]; then \
-		docker container create --attach stdin --attach stdout --attach stderr --volume `pwd`:/rocksdb-host --name rocksdb_linux_ppc64le-be evolvedbinary/rocksjava:centos7_ppc64le-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh; \
-	fi
-	docker start -a rocksdb_linux_ppc64le-be
+	docker run --rm --name rocksdb_linux_ppc64le-be --attach stdin --attach stdout --attach stderr --volume `pwd`:/rocksdb-host --env DEBUG_LEVEL=$(DEBUG_LEVEL) evolvedbinary/rocksjava:centos7_ppc64le-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh

 rocksdbjavastaticpublish: rocksdbjavastaticrelease rocksdbjavastaticpublishcentral

@@ -5,7 +5,7 @@
 [![PPC64le Build Status](http://140.211.168.68:8080/buildStatus/icon?job=Rocksdb)](http://140.211.168.68:8080/job/Rocksdb)

 RocksDB is developed and maintained by Facebook Database Engineering Team.
-It is built on earlier work on LevelDB by Sanjay Ghemawat (sanjay@google.com)
+It is built on earlier work on [LevelDB](https://github.com/google/leveldb) by Sanjay Ghemawat (sanjay@google.com)
 and Jeff Dean (jeff@google.com)

 This code is a library that forms the core building block for a fast
@@ -30,6 +30,7 @@ ROCKSDB_COMPILER_FLAGS = [
    "-DROCKSDB_PTHREAD_ADAPTIVE_MUTEX",
    "-DROCKSDB_BACKTRACE",
    "-Wnarrowing",
+    "-DROCKSDB_NO_DYNAMIC_EXTENSION",
 ]

 ROCKSDB_EXTERNAL_DEPS = [
@@ -83,22 +84,24 @@ cpp_library(
        "db/c.cc",
        "db/column_family.cc",
        "db/compacted_db_impl.cc",
-        "db/compaction.cc",
-        "db/compaction_iterator.cc",
-        "db/compaction_job.cc",
-        "db/compaction_picker.cc",
-        "db/compaction_picker_fifo.cc",
-        "db/compaction_picker_universal.cc",
+        "db/compaction/compaction.cc",
+        "db/compaction/compaction_iterator.cc",
+        "db/compaction/compaction_job.cc",
+        "db/compaction/compaction_picker.cc",
+        "db/compaction/compaction_picker_fifo.cc",
+        "db/compaction/compaction_picker_level.cc",
+        "db/compaction/compaction_picker_universal.cc",
        "db/convenience.cc",
        "db/db_filesnapshot.cc",
-        "db/db_impl.cc",
-        "db/db_impl_compaction_flush.cc",
-        "db/db_impl_debug.cc",
-        "db/db_impl_experimental.cc",
-        "db/db_impl_files.cc",
-        "db/db_impl_open.cc",
-        "db/db_impl_readonly.cc",
-        "db/db_impl_write.cc",
+        "db/db_impl/db_impl.cc",
+        "db/db_impl/db_impl_compaction_flush.cc",
+        "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_open.cc",
+        "db/db_impl/db_impl_readonly.cc",
+        "db/db_impl/db_impl_secondary.cc",
+        "db/db_impl/db_impl_write.cc",
        "db/db_info_dumper.cc",
        "db/db_iter.cc",
        "db/dbformat.cc",
@@ -110,6 +113,7 @@ cpp_library(
        "db/flush_job.cc",
        "db/flush_scheduler.cc",
        "db/forward_iterator.cc",
+        "db/import_column_family_job.cc",
        "db/internal_stats.cc",
        "db/log_reader.cc",
        "db/log_writer.cc",
@@ -141,8 +145,17 @@ cpp_library(
        "env/env_posix.cc",
        "env/io_posix.cc",
        "env/mock_env.cc",
+        "file/delete_scheduler.cc",
+        "file/file_util.cc",
+        "file/filename.cc",
+        "file/sst_file_manager_impl.cc",
+        "logging/auto_roll_logger.cc",
+        "logging/event_logger.cc",
+        "logging/log_buffer.cc",
+        "memory/arena.cc",
+        "memory/concurrent_arena.cc",
+        "memory/jemalloc_nodump_allocator.cc",
        "memtable/alloc_tracker.cc",
-        "memtable/hash_cuckoo_rep.cc",
        "memtable/hash_linklist_rep.cc",
        "memtable/hash_skiplist_rep.cc",
        "memtable/skiplistrep.cc",
@@ -150,10 +163,12 @@ cpp_library(
        "memtable/write_buffer_manager.cc",
        "monitoring/histogram.cc",
        "monitoring/histogram_windowing.cc",
+        "monitoring/in_memory_stats_history.cc",
        "monitoring/instrumented_mutex.cc",
        "monitoring/iostats_context.cc",
        "monitoring/perf_context.cc",
        "monitoring/perf_level.cc",
+        "monitoring/persistent_stats_history.cc",
        "monitoring/statistics.cc",
        "monitoring/thread_status_impl.cc",
        "monitoring/thread_status_updater.cc",
@@ -168,78 +183,70 @@ cpp_library(
        "options/options_sanity_check.cc",
        "port/port_posix.cc",
        "port/stack_trace.cc",
-        "table/adaptive_table_factory.cc",
-        "table/block.cc",
-        "table/block_based_filter_block.cc",
-        "table/block_based_table_builder.cc",
-        "table/block_based_table_factory.cc",
-        "table/block_based_table_reader.cc",
-        "table/block_builder.cc",
+        "table/adaptive/adaptive_table_factory.cc",
+        "table/block_based/block.cc",
+        "table/block_based/block_based_filter_block.cc",
+        "table/block_based/block_based_table_builder.cc",
+        "table/block_based/block_based_table_factory.cc",
+        "table/block_based/block_based_table_reader.cc",
+        "table/block_based/block_builder.cc",
+        "table/block_based/block_prefix_index.cc",
+        "table/block_based/data_block_footer.cc",
+        "table/block_based/data_block_hash_index.cc",
+        "table/block_based/filter_block_reader_common.cc",
+        "table/block_based/flush_block_policy.cc",
+        "table/block_based/full_filter_block.cc",
+        "table/block_based/index_builder.cc",
+        "table/block_based/partitioned_filter_block.cc",
        "table/block_fetcher.cc",
-        "table/block_prefix_index.cc",
        "table/bloom_block.cc",
-        "table/cuckoo_table_builder.cc",
-        "table/cuckoo_table_factory.cc",
-        "table/cuckoo_table_reader.cc",
-        "table/data_block_footer.cc",
-        "table/data_block_hash_index.cc",
-        "table/flush_block_policy.cc",
+        "table/cuckoo/cuckoo_table_builder.cc",
+        "table/cuckoo/cuckoo_table_factory.cc",
+        "table/cuckoo/cuckoo_table_reader.cc",
        "table/format.cc",
-        "table/full_filter_block.cc",
        "table/get_context.cc",
-        "table/index_builder.cc",
        "table/iterator.cc",
        "table/merging_iterator.cc",
        "table/meta_blocks.cc",
-        "table/partitioned_filter_block.cc",
        "table/persistent_cache_helper.cc",
-        "table/plain_table_builder.cc",
-        "table/plain_table_factory.cc",
-        "table/plain_table_index.cc",
-        "table/plain_table_key_coding.cc",
-        "table/plain_table_reader.cc",
+        "table/plain/plain_table_builder.cc",
+        "table/plain/plain_table_factory.cc",
+        "table/plain/plain_table_index.cc",
+        "table/plain/plain_table_key_coding.cc",
+        "table/plain/plain_table_reader.cc",
        "table/sst_file_reader.cc",
        "table/sst_file_writer.cc",
        "table/table_properties.cc",
        "table/two_level_iterator.cc",
+        "test_util/sync_point.cc",
+        "test_util/sync_point_impl.cc",
+        "test_util/transaction_test_util.cc",
        "tools/dump/db_dump_tool.cc",
        "tools/ldb_cmd.cc",
        "tools/ldb_tool.cc",
        "tools/sst_dump_tool.cc",
-        "util/arena.cc",
-        "util/auto_roll_logger.cc",
+        "trace_replay/block_cache_tracer.cc",
+        "trace_replay/trace_replay.cc",
        "util/bloom.cc",
        "util/build_version.cc",
        "util/coding.cc",
        "util/compaction_job_stats_impl.cc",
        "util/comparator.cc",
        "util/compression_context_cache.cc",
-        "util/concurrent_arena.cc",
        "util/concurrent_task_limiter_impl.cc",
        "util/crc32c.cc",
-        "util/delete_scheduler.cc",
        "util/dynamic_bloom.cc",
-        "util/event_logger.cc",
        "util/file_reader_writer.cc",
-        "util/file_util.cc",
-        "util/filename.cc",
        "util/filter_policy.cc",
        "util/hash.cc",
-        "util/jemalloc_nodump_allocator.cc",
-        "util/log_buffer.cc",
        "util/murmurhash.cc",
        "util/random.cc",
        "util/rate_limiter.cc",
        "util/slice.cc",
-        "util/sst_file_manager_impl.cc",
        "util/status.cc",
        "util/string_util.cc",
-        "util/sync_point.cc",
-        "util/sync_point_impl.cc",
        "util/thread_local.cc",
        "util/threadpool_imp.cc",
-        "util/trace_replay.cc",
-        "util/transaction_test_util.cc",
        "util/xxhash.cc",
        "utilities/backupable/backupable_db.cc",
        "utilities/blob_db/blob_compaction_filter.cc",
@@ -261,7 +268,6 @@ cpp_library(
        "utilities/env_mirror.cc",
        "utilities/env_timed.cc",
        "utilities/leveldb_options/leveldb_options.cc",
-        "utilities/lua/rocks_lua_compaction_filter.cc",
        "utilities/memory/memory_util.cc",
        "utilities/merge_operators/bytesxor.cc",
        "utilities/merge_operators/max.cc",
@@ -276,6 +282,7 @@ cpp_library(
        "utilities/persistent_cache/block_cache_tier_metadata.cc",
        "utilities/persistent_cache/persistent_cache_tier.cc",
        "utilities/persistent_cache/volatile_tier_impl.cc",
+        "utilities/simulator_cache/cache_simulator.cc",
        "utilities/simulator_cache/sim_cache.cc",
        "utilities/table_properties_collectors/compact_on_deletion_collector.cc",
        "utilities/trace/file_trace_reader_writer.cc",
@@ -309,10 +316,11 @@ cpp_library(
    srcs = [
        "db/db_test_util.cc",
        "table/mock_table.cc",
+        "test_util/fault_injection_test_env.cc",
+        "test_util/testharness.cc",
+        "test_util/testutil.cc",
+        "tools/block_cache_trace_analyzer.cc",
        "tools/trace_analyzer_tool.cc",
-        "util/fault_injection_test_env.cc",
-        "util/testharness.cc",
-        "util/testutil.cc",
        "utilities/cassandra/test_utils.cc",
    ],
    auto_headers = AutoHeaders.RECURSIVE_GLOB,
@@ -326,9 +334,10 @@ cpp_library(
 cpp_library(
    name = "rocksdb_tools_lib",
    srcs = [
+        "test_util/testutil.cc",
+        "tools/block_cache_trace_analyzer.cc",
        "tools/db_bench_tool.cc",
        "tools/trace_analyzer_tool.cc",
-        "util/testutil.cc",
    ],
    auto_headers = AutoHeaders.RECURSIVE_GLOB,
    arch_preprocessor_flags = ROCKSDB_ARCH_PREPROCESSOR_FLAGS,
@@ -353,12 +362,17 @@ cpp_library(
 ROCKS_TESTS = [
    [
        "arena_test",
-        "util/arena_test.cc",
+        "memory/arena_test.cc",
        "serial",
    ],
    [
        "auto_roll_logger_test",
-        "util/auto_roll_logger_test.cc",
+        "logging/auto_roll_logger_test.cc",
+        "serial",
+    ],
+    [
+        "env_logger_test",
+        "logging/env_logger_test.cc",
        "serial",
    ],
    [
@@ -378,12 +392,22 @@ ROCKS_TESTS = [
    ],
    [
        "block_based_filter_block_test",
-        "table/block_based_filter_block_test.cc",
+        "table/block_based/block_based_filter_block_test.cc",
+        "serial",
+    ],
+    [
+        "block_cache_trace_analyzer_test",
+        "tools/block_cache_trace_analyzer_test.cc",
+        "serial",
+    ],
+    [
+        "block_cache_tracer_test",
+        "trace_replay/block_cache_tracer_test.cc",
        "serial",
    ],
    [
        "block_test",
-        "table/block_test.cc",
+        "table/block_based/block_test.cc",
        "serial",
    ],
    [
@@ -401,6 +425,11 @@ ROCKS_TESTS = [
        "cache/cache_test.cc",
        "serial",
    ],
+    [
+        "cache_simulator_test",
+        "utilities/simulator_cache/cache_simulator_test.cc",
+        "serial",
+    ],
    [
        "cassandra_format_test",
        "utilities/cassandra/cassandra_format_test.cc",
@@ -453,22 +482,22 @@ ROCKS_TESTS = [
    ],
    [
        "compaction_iterator_test",
-        "db/compaction_iterator_test.cc",
+        "db/compaction/compaction_iterator_test.cc",
        "serial",
    ],
    [
        "compaction_job_stats_test",
-        "db/compaction_job_stats_test.cc",
+        "db/compaction/compaction_job_stats_test.cc",
        "serial",
    ],
    [
        "compaction_job_test",
-        "db/compaction_job_test.cc",
+        "db/compaction/compaction_job_test.cc",
        "serial",
    ],
    [
        "compaction_picker_test",
-        "db/compaction_picker_test.cc",
+        "db/compaction/compaction_picker_test.cc",
        "serial",
    ],
    [
@@ -488,7 +517,7 @@ ROCKS_TESTS = [
    ],
    [
        "cuckoo_table_builder_test",
-        "table/cuckoo_table_builder_test.cc",
+        "table/cuckoo/cuckoo_table_builder_test.cc",
        "serial",
    ],
    [
@@ -498,12 +527,12 @@ ROCKS_TESTS = [
    ],
    [
        "cuckoo_table_reader_test",
-        "table/cuckoo_table_reader_test.cc",
+        "table/cuckoo/cuckoo_table_reader_test.cc",
        "serial",
    ],
    [
        "data_block_hash_index_test",
-        "table/data_block_hash_index_test.cc",
+        "table/block_based/data_block_hash_index_test.cc",
        "serial",
    ],
    [
@@ -524,7 +553,7 @@ ROCKS_TESTS = [
    [
        "db_bloom_filter_test",
        "db/db_bloom_filter_test.cc",
-        "serial",
+        "parallel",
    ],
    [
        "db_compaction_filter_test",
@@ -606,6 +635,11 @@ ROCKS_TESTS = [
        "db/db_range_del_test.cc",
        "serial",
    ],
+    [
+        "db_secondary_test",
+        "db/db_impl/db_secondary_test.cc",
+        "serial",
+    ],
    [
        "db_sst_test",
        "db/db_sst_test.cc",
@@ -658,7 +692,7 @@ ROCKS_TESTS = [
    ],
    [
        "delete_scheduler_test",
-        "util/delete_scheduler_test.cc",
+        "file/delete_scheduler_test.cc",
        "serial",
    ],
    [
@@ -693,7 +727,7 @@ ROCKS_TESTS = [
    ],
    [
        "event_logger_test",
-        "util/event_logger_test.cc",
+        "logging/event_logger_test.cc",
        "serial",
    ],
    [
@@ -738,7 +772,7 @@ ROCKS_TESTS = [
    ],
    [
        "full_filter_block_test",
-        "table/full_filter_block_test.cc",
+        "table/block_based/full_filter_block_test.cc",
        "serial",
    ],
    [
@@ -868,7 +902,7 @@ ROCKS_TESTS = [
    ],
    [
        "partitioned_filter_block_test",
-        "table/partitioned_filter_block_test.cc",
+        "table/block_based/partitioned_filter_block_test.cc",
        "serial",
    ],
    [
@@ -951,6 +985,11 @@ ROCKS_TESTS = [
        "monitoring/statistics_test.cc",
        "serial",
    ],
+    [
+        "stats_history_test",
+        "monitoring/stats_history_test.cc",
+        "serial",
+    ],
    [
        "stringappend_test",
        "utilities/merge_operators/string_append/stringappend_test.cc",
@@ -50,6 +50,10 @@ Check out their blog post: http://blog.cloudera.com/blog/2015/08/inside-santande
 ## Airbnb
 Airbnb is using RocksDB as a storage engine for their personalized search service. You can learn more about it here: https://www.youtube.com/watch?v=ASQ6XMtogMs

+## Alluxio
+[Alluxio](https://www.alluxio.io) uses RocksDB to serve and scale file system metadata to beyond 1 Billion files. The detailed design and implementation is described in this engineering blog:
+https://www.alluxio.io/blog/scalable-metadata-service-in-alluxio-storing-billions-of-files/
+
 ## Pinterest
 Pinterest's Object Retrieval System uses RocksDB for storage: https://www.youtube.com/watch?v=MtFEVEs_2Vo

@@ -91,4 +95,4 @@ LzLabs is using RocksDB as a storage engine in their multi-database distributed
 [ProfaneDB](https://profanedb.gitlab.io/) is a database for Protocol Buffers, and uses RocksDB for storage. It is accessible via gRPC, and the schema is defined using directly `.proto` files.

 ## IOTA Foundation
- [IOTA Foundation](https://www.iota.org/) is using RocksDB in the [IOTA Reference Implementation (IRI)](https://github.com/iotaledger/iri) to store the local state of the Tangle. The Tangle is the first open-source distributed ledger powering the future of the Internet of Things.
+ [IOTA Foundation](https://www.iota.org/) is using RocksDB in the [IOTA Reference Implementation (IRI)](https://github.com/iotaledger/iri) to store the local state of the Tangle. The Tangle is the first open-source distributed ledger powering the future of the Internet of Things.
@@ -43,7 +43,7 @@ We plan to use this port for our business purposes here at Bing and this provide

 * Certain headers that are not present and not necessary on Windows were simply `#ifndef OS_WIN` in a few places (`unistd.h`)
 * All posix specific headers were replaced to port/port.h which worked well
-* Replaced `dirent.h` for `port/dirent.h` (very few places) with the implementation of the relevant interfaces within `rocksdb::port` namespace
+* Replaced `dirent.h` for `port/port_dirent.h` (very few places) with the implementation of the relevant interfaces within `rocksdb::port` namespace
 * Replaced `sys/time.h` to `port/sys_time.h` (few places) implemented equivalents within `rocksdb::port`
 * `printf %z` specification is not supported on Windows. To imitate existing standards we came up with a string macro `ROCKSDB_PRIszt` which expands to `zu` on posix systems and to `Iu` on windows.
 * in class member initialization were moved to a __ctors in some cases
@@ -1,15 +1,67 @@
 version: 1.0.{build}
+
 image: Visual Studio 2017
+
+environment:
+  JAVA_HOME: C:\Program Files\Java\jdk1.8.0
+  THIRDPARTY_HOME: $(APPVEYOR_BUILD_FOLDER)\thirdparty
+  SNAPPY_HOME: $(THIRDPARTY_HOME)\snappy-1.1.7
+  SNAPPY_INCLUDE: $(SNAPPY_HOME);$(SNAPPY_HOME)\build
+  SNAPPY_LIB_DEBUG: $(SNAPPY_HOME)\build\Debug\snappy.lib
+  SNAPPY_LIB_RELEASE: $(SNAPPY_HOME)\build\Release\snappy.lib
+  LZ4_HOME: $(THIRDPARTY_HOME)\lz4-1.8.3
+  LZ4_INCLUDE: $(LZ4_HOME)\lib
+  LZ4_LIB_DEBUG: $(LZ4_HOME)\visual\VS2010\bin\x64_Debug\liblz4_static.lib
+  LZ4_LIB_RELEASE: $(LZ4_HOME)\visual\VS2010\bin\x64_Release\liblz4_static.lib
+  ZSTD_HOME: $(THIRDPARTY_HOME)\zstd-1.4.0
+  ZSTD_INCLUDE: $(ZSTD_HOME)\lib;$(ZSTD_HOME)\lib\dictBuilder
+  ZSTD_LIB_DEBUG: $(ZSTD_HOME)\build\VS2010\bin\x64_Debug\libzstd_static.lib
+  ZSTD_LIB_RELEASE: $(ZSTD_HOME)\build\VS2010\bin\x64_Release\libzstd_static.lib
+
+install:
+  - md %THIRDPARTY_HOME%
+  - echo "Building Snappy dependency..."
+  - cd %THIRDPARTY_HOME%
+  - curl -fsSL -o snappy-1.1.7.zip https://github.com/google/snappy/archive/1.1.7.zip
+  - unzip snappy-1.1.7.zip
+  - cd snappy-1.1.7
+  - mkdir build
+  - cd build
+  - cmake -DCMAKE_GENERATOR_PLATFORM=x64 ..
+  - msbuild Snappy.sln /p:Configuration=Debug /p:Platform=x64
+  - msbuild Snappy.sln /p:Configuration=Release /p:Platform=x64
+  - echo "Building LZ4 dependency..."
+  - cd %THIRDPARTY_HOME%
+  - curl -fsSL -o lz4-1.8.3.zip https://github.com/lz4/lz4/archive/v1.8.3.zip
+  - unzip lz4-1.8.3.zip
+  - cd lz4-1.8.3\visual\VS2010
+  - ps: $CMD="C:\Program Files (x86)\Microsoft Visual Studio\2017\Community\Common7\IDE\devenv.com"; & $CMD lz4.sln /upgrade
+  - msbuild lz4.sln /p:Configuration=Debug /p:Platform=x64
+  - msbuild lz4.sln /p:Configuration=Release /p:Platform=x64
+  - echo "Building ZStd dependency..."
+  - cd %THIRDPARTY_HOME%
+  - curl -fsSL -o zstd-1.4.0.zip https://github.com/facebook/zstd/archive/v1.4.0.zip
+  - unzip zstd-1.4.0.zip
+  - cd zstd-1.4.0\build\VS2010
+  - ps: $CMD="C:\Program Files (x86)\Microsoft Visual Studio\2017\Community\Common7\IDE\devenv.com"; & $CMD zstd.sln /upgrade
+  - msbuild zstd.sln /p:Configuration=Debug /p:Platform=x64
+  - msbuild zstd.sln /p:Configuration=Release /p:Platform=x64
+
 before_build:
- md %APPVEYOR_BUILD_FOLDER%\build
- cd %APPVEYOR_BUILD_FOLDER%\build
- cmake -G "Visual Studio 15 Win64" -DOPTDBG=1 -DWITH_XPRESS=1 -DPORTABLE=1 -DJNI=1 ..
- cd ..
+  - md %APPVEYOR_BUILD_FOLDER%\build
+  - cd %APPVEYOR_BUILD_FOLDER%\build
+  - cmake -G "Visual Studio 15 Win64" -DOPTDBG=1 -DPORTABLE=1 -DSNAPPY=1 -DLZ4=1 -DZSTD=1 -DXPRESS=1 -DJNI=1 ..
+  - cd ..
 build:
  project: build\rocksdb.sln
  parallel: true
  verbosity: normal
-test:
-test_script:
- ps: build_tools\run_ci_db_test.ps1 -SuiteRun db_basic_test,db_test2,db_test,env_basic_test,env_test -Concurrency 8
+
+test:
+
+test_script:
+  - ps: build_tools\run_ci_db_test.ps1 -SuiteRun db_basic_test,db_test2,db_test,env_basic_test,env_test -Concurrency 8
+
+on_failure:
+  - cmd: 7z a build-failed.zip %APPVEYOR_BUILD_FOLDER%\build\ && appveyor PushArtifact build-failed.zip

@@ -1,3 +1,4 @@
+# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved.
 from __future__ import absolute_import
 from __future__ import division
 from __future__ import print_function
@@ -81,7 +82,7 @@ def get_tests(repo_path):
            else:
                # we consumed all the parallel tests
                break
-    
+
    return tests


@@ -117,7 +118,7 @@ def generate_targets(repo_path):
        "rocksdb_tools_lib",
        src_mk.get("BENCH_LIB_SOURCES", []) +
        src_mk.get("ANALYZER_LIB_SOURCES", []) +
-        ["util/testutil.cc"],
+        ["test_util/testutil.cc"],
        [":rocksdb_lib"])

    # test for every test we found in the Makefile
@@ -1,4 +1,5 @@
 #!/usr/bin/env bash
+# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved.
 # Create a tmp directory for the test to use
 TEST_DIR=$(mktemp -d /dev/shm/fbcode_rocksdb_XXXXXXX)
 # shellcheck disable=SC2068
@@ -1,3 +1,4 @@
+# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved.
 from __future__ import absolute_import
 from __future__ import division
 from __future__ import print_function
@@ -1,8 +1,10 @@
+# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved.
 from __future__ import absolute_import
 from __future__ import division
 from __future__ import print_function
 from __future__ import unicode_literals
 rocksdb_target_header = """load("@fbcode_macros//build_defs:auto_headers.bzl", "AutoHeaders")
+load("@fbcode_macros//build_defs:cpp_library.bzl", "cpp_library")
 load(":defs.bzl", "test_binary")

 REPO_PATH = package_name() + "/"
@@ -23,6 +25,7 @@ ROCKSDB_COMPILER_FLAGS = [
    "-DBZIP2",
    "-DLZ4",
    "-DZSTD",
+    "-DZSTD_STATIC_LINKING_ONLY",
    "-DGFLAGS=gflags",
    "-DNUMA",
    "-DTBB",
@@ -32,6 +35,7 @@ ROCKSDB_COMPILER_FLAGS = [
    "-DROCKSDB_PTHREAD_ADAPTIVE_MUTEX",
    "-DROCKSDB_BACKTRACE",
    "-Wnarrowing",
+    "-DROCKSDB_NO_DYNAMIC_EXTENSION",
 ]

 ROCKSDB_EXTERNAL_DEPS = [
@@ -1,3 +1,4 @@
+# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved.
 """
 This module keeps commonly used components.
 """
@@ -1,4 +1,5 @@
 #!/usr/bin/python
+# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved.

 # amalgamate.py creates an amalgamation from a unity build.
 # It can be run with either Python 2 or 3.
@@ -234,7 +234,7 @@ else
          #include <linux/falloc.h>
          int main() {
      int fd = open("/dev/null", 0);
-      fallocate(fd, FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE, 0, 1024);
+      fallocate(fd, FALLOC_FL_KEEP_SIZE, 0, 1024);
          }
 EOF
        if [ "$?" = 0 ]; then
@@ -397,6 +397,7 @@ EOF
          #include <malloc.h>
          int main() {
            size_t res = malloc_usable_size(0);
+            (void)res;
            return 0;
          }
 EOF
@@ -411,6 +412,7 @@ EOF
          #include <pthread.h>
          int main() {
            int x = PTHREAD_MUTEX_ADAPTIVE_NP;
+            (void)x;
            return 0;
          }
 EOF
@@ -422,7 +424,7 @@ EOF
    if ! test $ROCKSDB_DISABLE_BACKTRACE; then
        # Test whether backtrace is available
        $CXX $CFLAGS -x c++ - -o /dev/null 2>/dev/null  <<EOF
-          #include <execinfo.h>>
+          #include <execinfo.h>
          int main() {
            void* frames[1];
            backtrace_symbols(frames, backtrace(frames, 1));
@@ -480,6 +482,7 @@ EOF
          #include <sched.h>
          int main() {
            int cpuid = sched_getcpu();
+            (void)cpuid;
          }
 EOF
        if [ "$?" = 0 ]; then
@@ -515,11 +518,11 @@ fi

 if test "$USE_HDFS"; then
  if test -z "$JAVA_HOME"; then
-    echo "JAVA_HOME has to be set for HDFS usage."
+    echo "JAVA_HOME has to be set for HDFS usage." >&2
    exit 1
  fi
-  HDFS_CCFLAGS="$HDFS_CCFLAGS -I$JAVA_HOME/include -I$JAVA_HOME/include/linux -DUSE_HDFS"
-  HDFS_LDFLAGS="$HDFS_LDFLAGS -lhdfs -L$JAVA_HOME/jre/lib/amd64"
+  HDFS_CCFLAGS="$HDFS_CCFLAGS -I$JAVA_HOME/include -I$JAVA_HOME/include/linux -DUSE_HDFS -I$HADOOP_HOME/include"
+  HDFS_LDFLAGS="$HDFS_LDFLAGS -lhdfs -L$JAVA_HOME/jre/lib/amd64 -L$HADOOP_HOME/lib/native"
  HDFS_LDFLAGS="$HDFS_LDFLAGS -L$JAVA_HOME/jre/lib/amd64/server -L$GLIBC_RUNTIME_PATH/lib"
  HDFS_LDFLAGS="$HDFS_LDFLAGS -ldl -lverify -ljava -ljvm"
  COMMON_FLAGS="$COMMON_FLAGS $HDFS_CCFLAGS"
@@ -534,9 +537,11 @@ if test -z "$PORTABLE"; then
    COMMON_FLAGS="$COMMON_FLAGS -mcpu=$POWER -mtune=$POWER "
  elif test -n "`echo $TARGET_ARCHITECTURE | grep ^s390x`"; then
    COMMON_FLAGS="$COMMON_FLAGS -march=z10 "
-  elif test -n "`echo $TARGET_ARCHITECTURE | grep ^arm`"; then
+  elif test -n "`echo $TARGET_ARCHITECTURE | grep -e^arm -e^aarch64`"; then
    # TODO: Handle this with approprite options.
    COMMON_FLAGS="$COMMON_FLAGS"
+  elif test -n "`echo $TARGET_ARCHITECTURE | grep ^aarch64`"; then
+    COMMON_FLAGS="$COMMON_FLAGS"
  elif [ "$TARGET_OS" == "IOS" ]; then
    COMMON_FLAGS="$COMMON_FLAGS"
  elif [ "$TARGET_OS" != "AIX" ] && [ "$TARGET_OS" != "SunOS" ]; then
@@ -553,12 +558,13 @@ $CXX $PLATFORM_CXXFLAGS $COMMON_FLAGS -x c++ - -o /dev/null 2>/dev/null <<EOF
  #include <nmmintrin.h>
  int main() {
    volatile uint32_t x = _mm_crc32_u32(0, 0);
+    (void)x;
  }
 EOF
 if [ "$?" = 0 ]; then
  COMMON_FLAGS="$COMMON_FLAGS -DHAVE_SSE42"
 elif test "$USE_SSE"; then
-  echo "warning: USE_SSE specified but compiler could not use SSE intrinsics, disabling"
+  echo "warning: USE_SSE specified but compiler could not use SSE intrinsics, disabling" >&2
  exit 1
 fi

@@ -570,12 +576,13 @@ $CXX $PLATFORM_CXXFLAGS $COMMON_FLAGS -x c++ - -o /dev/null 2>/dev/null <<EOF
    const auto b = _mm_set_epi64x(0, 0);
    const auto c = _mm_clmulepi64_si128(a, b, 0x00);
    auto d = _mm_cvtsi128_si64(c);
+    (void)d;
  }
 EOF
 if [ "$?" = 0 ]; then
  COMMON_FLAGS="$COMMON_FLAGS -DHAVE_PCLMUL"
 elif test "$USE_SSE"; then
-  echo "warning: USE_SSE specified but compiler could not use PCLMUL intrinsics, disabling"
+  echo "warning: USE_SSE specified but compiler could not use PCLMUL intrinsics, disabling" >&2
  exit 1
 fi

@@ -589,6 +596,7 @@ if [ "$PLATFORM" != IOS ]; then
  #endif
  int main() {
    static __thread int tls;
+    (void)tls;
  }
 EOF
  if [ "$?" = 0 ]; then
@@ -596,6 +604,19 @@ EOF
  fi
 fi

+if [ "$FBCODE_BUILD" != "true" -a "$PLATFORM" = OS_LINUX ]; then
+  $CXX $COMMON_FLAGS $PLATFORM_SHARED_CFLAGS -x c++ -c - -o test_dl.o 2>/dev/null <<EOF
+  void dummy_func() {}
+EOF
+  if [ "$?" = 0 ]; then
+    $CXX $COMMON_FLAGS $PLATFORM_SHARED_LDFLAGS test_dl.o -o /dev/null 2>/dev/null
+    if [ "$?" = 0 ]; then
+      EXEC_LDFLAGS+="-ldl"
+      rm -f test_dl.o
+    fi
+  fi
+fi
+
 PLATFORM_CCFLAGS="$PLATFORM_CCFLAGS $COMMON_FLAGS"
 PLATFORM_CXXFLAGS="$PLATFORM_CXXFLAGS $COMMON_FLAGS"

@@ -1,3 +1,4 @@
+# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved.
 GCC_BASE=/mnt/gvfs/third-party2/gcc/112ec378fec7002ad3e09afde022e656049f7191/5.x/centos7-native/c447969
 CLANG_BASE=/mnt/gvfs/third-party2/llvm-fb/04999bdb3ce81a11073535dcb00b5e13dc1cbaf5/stable/centos7-native/c9f9104
 LIBGCC_BASE=/mnt/gvfs/third-party2/libgcc/92b0c8e5c8eecc71eb042594ce1ab3413799b385/5.x/gcc-5-glibc-2.23/339d858
@@ -1,4 +1,5 @@
 # shellcheck disable=SC2148
+# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved.
 GCC_BASE=/mnt/gvfs/third-party2/gcc/cf7d14c625ce30bae1a4661c2319c5a283e4dd22/4.8.1/centos6-native/cc6c9dc
 CLANG_BASE=/mnt/gvfs/third-party2/llvm-fb/8598c375b0e94e1448182eb3df034704144a838d/stable/centos6-native/3f16ddd
 LIBGCC_BASE=/mnt/gvfs/third-party2/libgcc/d6e0a7da6faba45f5e5b1638f9edd7afc2f34e7d/4.8.1/gcc-4.8.1-glibc-2.17/8aac7fc
@@ -1,3 +1,4 @@
+# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved.
 GCC_BASE=/mnt/gvfs/third-party2/gcc/6e8e715624fd15256a7970073387793dfcf79b46/7.x/centos7-native/b2ef2b6
 CLANG_BASE=/mnt/gvfs/third-party2/llvm-fb/ef37e1faa1c29782abfac1ae65a291b9b7966f6d/stable/centos7-native/c9f9104
 LIBGCC_BASE=/mnt/gvfs/third-party2/libgcc/c67031f0f739ac61575a061518d6ef5038f99f90/7.x/platform007/5620abc
@@ -1,2 +1,3 @@
 #!/usr/bin/env bash
+# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved.
 docker run -v $PWD:/rocks -w /rocks buildpack-deps make
@@ -132,11 +132,14 @@ _TEST_NAME_TO_PARSERS = {
    'lite': [CompilerErrorParser],
    'lite_test': [CompilerErrorParser, GTestErrorParser],
    'stress_crash': [CompilerErrorParser, DbCrashErrorParser],
+    'stress_crash_with_atomic_flush': [CompilerErrorParser, DbCrashErrorParser],
    'write_stress': [CompilerErrorParser, WriteStressErrorParser],
    'asan': [CompilerErrorParser, GTestErrorParser, AsanErrorParser],
    'asan_crash': [CompilerErrorParser, AsanErrorParser, DbCrashErrorParser],
+    'asan_crash_with_atomic_flush': [CompilerErrorParser, AsanErrorParser, DbCrashErrorParser],
    'ubsan': [CompilerErrorParser, GTestErrorParser, UbsanErrorParser],
    'ubsan_crash': [CompilerErrorParser, UbsanErrorParser, DbCrashErrorParser],
+    'ubsan_crash_with_atomic_flush': [CompilerErrorParser, UbsanErrorParser, DbCrashErrorParser],
    'valgrind': [CompilerErrorParser, GTestErrorParser, ValgrindErrorParser],
    'tsan': [CompilerErrorParser, GTestErrorParser, TsanErrorParser],
    'format_compatible': [CompilerErrorParser, CompatErrorParser],
@@ -1,5 +1,5 @@
 #!/bin/sh
-
+# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved.
 # fail early
 set -e

@@ -1,4 +1,5 @@
 #!/bin/sh
+# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved.
 #
 # Set environment variables so that we can compile rocksdb using
 # fbcode settings.  It uses the latest g++ and clang compilers and also
@@ -1,4 +1,5 @@
 #!/bin/sh
+# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved.
 #
 # Set environment variables so that we can compile rocksdb using
 # fbcode settings.  It uses the latest g++ compiler and also
@@ -1,4 +1,5 @@
 #!/bin/sh
+# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved.
 #
 # Set environment variables so that we can compile rocksdb using
 # fbcode settings.  It uses the latest g++ and clang compilers and also
@@ -1,4 +1,5 @@
 #!/usr/bin/env bash
+# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved.
 # If clang_format_diff.py command is not specfied, we assume we are able to
 # access directly without any path.
 if [ -z $CLANG_FORMAT_DIFF ]
@@ -1,5 +1,6 @@
 # shellcheck disable=SC1113
 #/usr/bin/env bash
+# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved.

 set -e

@@ -1,4 +1,5 @@
 #!/usr/local/fbcode/gcc-4.9-glibc-2.20-fb/bin/python2.7
+# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved.

 from __future__ import absolute_import
 from __future__ import division
@@ -1,4 +1,5 @@
 #!/usr/bin/env bash
+# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved.

 set -e

@@ -63,6 +63,21 @@ CLEANUP_ENV="
    'user':'root'
 }"

+UPLOAD_DB_DIR="
+{
+    'name':'Upload database directory',
+    'shell':'tar -cvzf rocksdb_db.tar.gz /dev/shm/rocksdb/',
+    'user':'root',
+    'cleanup':true,
+    'provide_artifacts': [
+    {
+      'name':'rocksdb_db_dir',
+      'paths': ['rocksdb_db.tar.gz'],
+      'bundle': false,
+    },
+    ],
+}"
+
 # We will eventually set the RATIO to 1, but we want do this
 # in steps. RATIO=$(nproc) will make it work as J=1
 if [ -z $RATIO ]; then
@@ -109,13 +124,6 @@ else
  TASK_CREATION_TOOL="false"
 fi

-ARTIFACTS=" 'artifacts': [
-    {
-        'name':'database',
-        'paths':[ '/dev/shm/rocksdb' ],
-    }
-]"
-
 #
 # A mechanism to disable tests temporarily
 #
@@ -140,6 +148,7 @@ UNIT_TEST_COMMANDS="[
    {
        'name':'Rocksdb Unit Test',
        'oncall':'$ONCALL',
+        'executeLocal': 'true',
        'steps': [
            $CLEANUP_ENV,
            {
@@ -160,6 +169,7 @@ UNIT_TEST_NON_SHM_COMMANDS="[
    {
        'name':'Rocksdb Unit Test',
        'oncall':'$ONCALL',
+        'executeLocal': 'true',
        'timeout': 86400,
        'steps': [
            $CLEANUP_ENV,
@@ -182,6 +192,7 @@ RELEASE_BUILD_COMMANDS="[
    {
        'name':'Rocksdb Release Build',
        'oncall':'$ONCALL',
+        'executeLocal': 'true',
        'steps': [
            $CLEANUP_ENV,
            {
@@ -202,6 +213,7 @@ UNIT_TEST_COMMANDS_481="[
    {
        'name':'Rocksdb Unit Test on GCC 4.8.1',
        'oncall':'$ONCALL',
+        'executeLocal': 'true',
        'steps': [
            $CLEANUP_ENV,
            {
@@ -222,6 +234,7 @@ RELEASE_BUILD_COMMANDS_481="[
    {
        'name':'Rocksdb Release on GCC 4.8.1',
        'oncall':'$ONCALL',
+        'executeLocal': 'true',
        'steps': [
            $CLEANUP_ENV,
            {
@@ -242,6 +255,7 @@ CLANG_UNIT_TEST_COMMANDS="[
    {
        'name':'Rocksdb Unit Test',
        'oncall':'$ONCALL',
+        'executeLocal': 'true',
        'steps': [
            $CLEANUP_ENV,
            {
@@ -262,6 +276,7 @@ CLANG_RELEASE_BUILD_COMMANDS="[
    {
        'name':'Rocksdb CLANG Release Build',
        'oncall':'$ONCALL',
+        'executeLocal': 'true',
        'steps': [
            $CLEANUP_ENV,
            {
@@ -282,6 +297,7 @@ CLANG_ANALYZE_COMMANDS="[
    {
        'name':'Rocksdb analyze',
        'oncall':'$ONCALL',
+        'executeLocal': 'true',
        'steps': [
            $CLEANUP_ENV,
            {
@@ -302,6 +318,7 @@ CODE_COV_COMMANDS="[
    {
        'name':'Rocksdb Unit Test Code Coverage',
        'oncall':'$ONCALL',
+        'executeLocal': 'true',
        'steps': [
            $CLEANUP_ENV,
            {
@@ -322,6 +339,7 @@ UNITY_COMMANDS="[
    {
        'name':'Rocksdb Unity',
        'oncall':'$ONCALL',
+        'executeLocal': 'true',
        'steps': [
            $CLEANUP_ENV,
            {
@@ -342,6 +360,7 @@ LITE_BUILD_COMMANDS="[
    {
        'name':'Rocksdb Lite build',
        'oncall':'$ONCALL',
+        'executeLocal': 'true',
        'steps': [
            $CLEANUP_ENV,
            {
@@ -361,6 +380,7 @@ REPORT_LITE_BINARY_SIZE_COMMANDS="[
    {
        'name':'Rocksdb Lite Binary Size',
        'oncall':'$ONCALL',
+        'executeLocal': 'true',
        'steps': [
            $CLEANUP_ENV,
            {
@@ -376,8 +396,9 @@ REPORT_LITE_BINARY_SIZE_COMMANDS="[
 #
 STRESS_CRASH_TEST_COMMANDS="[
    {
-        'name':'Rocksdb Stress/Crash Test',
+        'name':'Rocksdb Stress and Crash Test',
        'oncall':'$ONCALL',
+        'executeLocal': 'true',
        'timeout': 86400,
        'steps': [
            $CLEANUP_ENV,
@@ -395,7 +416,36 @@ STRESS_CRASH_TEST_COMMANDS="[
                $PARSER
            }
        ],
-        $ARTIFACTS,
+        $REPORT
+    }
+]"
+
+#
+# RocksDB stress/crash test with atomic flush
+#
+STRESS_CRASH_TEST_WITH_ATOMIC_FLUSH_COMMANDS="[
+    {
+        'name':'Rocksdb Stress and Crash Test with atomic flush',
+        'oncall':'$ONCALL',
+        'executeLocal': 'true',
+        'timeout': 86400,
+        'steps': [
+            $CLEANUP_ENV,
+            {
+                'name':'Build and run RocksDB debug stress tests',
+                'shell':'$SHM $DEBUG $NON_TSAN_CRASH make J=1 db_stress || $CONTRUN_NAME=db_stress $TASK_CREATION_TOOL',
+                'user':'root',
+                $PARSER
+            },
+            {
+                'name':'Build and run RocksDB debug crash tests with atomic flush',
+                'timeout': 86400,
+                'shell':'$SHM $DEBUG $NON_TSAN_CRASH make J=1 crash_test_with_atomic_flush || $CONTRUN_NAME=crash_test_with_atomic_flush $TASK_CREATION_TOOL',
+                'user':'root',
+                $PARSER
+            },
+            $UPLOAD_DB_DIR,
+        ],
        $REPORT
    }
 ]"
@@ -407,6 +457,7 @@ WRITE_STRESS_COMMANDS="[
    {
        'name':'Rocksdb Write Stress Test',
        'oncall':'$ONCALL',
+        'executeLocal': 'true',
        'steps': [
            $CLEANUP_ENV,
            {
@@ -429,6 +480,7 @@ ASAN_TEST_COMMANDS="[
    {
        'name':'Rocksdb Unit Test under ASAN',
        'oncall':'$ONCALL',
+        'executeLocal': 'true',
        'steps': [
            $CLEANUP_ENV,
            {
@@ -449,6 +501,7 @@ ASAN_CRASH_TEST_COMMANDS="[
    {
        'name':'Rocksdb crash test under ASAN',
        'oncall':'$ONCALL',
+        'executeLocal': 'true',
        'timeout': 86400,
        'steps': [
            $CLEANUP_ENV,
@@ -464,6 +517,30 @@ ASAN_CRASH_TEST_COMMANDS="[
    }
 ]"

+#
+# RocksDB crash testing with atomic flush under address sanitizer
+#
+ASAN_CRASH_TEST_WITH_ATOMIC_FLUSH_COMMANDS="[
+    {
+        'name':'Rocksdb crash test with atomic flush under ASAN',
+        'oncall':'$ONCALL',
+        'executeLocal': 'true',
+        'timeout': 86400,
+        'steps': [
+            $CLEANUP_ENV,
+            {
+                'name':'Build and run RocksDB debug asan_crash_test_with_atomic_flush',
+                'timeout': 86400,
+                'shell':'$SHM $DEBUG $NON_TSAN_CRASH make J=1 asan_crash_test_with_atomic_flush || $CONTRUN_NAME=asan_crash_test_with_atomic_flush $TASK_CREATION_TOOL',
+                'user':'root',
+                $PARSER
+            },
+            $UPLOAD_DB_DIR,
+        ],
+        $REPORT
+    }
+]"
+
 #
 # RocksDB test under undefined behavior sanitizer
 #
@@ -471,6 +548,7 @@ UBSAN_TEST_COMMANDS="[
    {
        'name':'Rocksdb Unit Test under UBSAN',
        'oncall':'$ONCALL',
+        'executeLocal': 'true',
        'steps': [
            $CLEANUP_ENV,
            {
@@ -491,6 +569,7 @@ UBSAN_CRASH_TEST_COMMANDS="[
    {
        'name':'Rocksdb crash test under UBSAN',
        'oncall':'$ONCALL',
+        'executeLocal': 'true',
        'timeout': 86400,
        'steps': [
            $CLEANUP_ENV,
@@ -506,6 +585,30 @@ UBSAN_CRASH_TEST_COMMANDS="[
    }
 ]"

+#
+# RocksDB crash testing with atomic flush under undefined behavior sanitizer
+#
+UBSAN_CRASH_TEST_WITH_ATOMIC_FLUSH_COMMANDS="[
+    {
+        'name':'Rocksdb crash test with atomic flush under UBSAN',
+        'oncall':'$ONCALL',
+        'executeLocal': 'true',
+        'timeout': 86400,
+        'steps': [
+            $CLEANUP_ENV,
+            {
+                'name':'Build and run RocksDB debug ubsan_crash_test_with_atomic_flush',
+                'timeout': 86400,
+                'shell':'$SHM $DEBUG $NON_TSAN_CRASH make J=1 ubsan_crash_test_with_atomic_flush || $CONTRUN_NAME=ubsan_crash_test_with_atomic_flush $TASK_CREATION_TOOL',
+                'user':'root',
+                $PARSER
+            },
+            $UPLOAD_DB_DIR,
+        ],
+        $REPORT
+    }
+]"
+
 #
 # RocksDB unit test under valgrind
 #
@@ -513,6 +616,7 @@ VALGRIND_TEST_COMMANDS="[
    {
        'name':'Rocksdb Unit Test under valgrind',
        'oncall':'$ONCALL',
+        'executeLocal': 'true',
        'timeout': 86400,
        'steps': [
            $CLEANUP_ENV,
@@ -535,6 +639,7 @@ TSAN_UNIT_TEST_COMMANDS="[
    {
        'name':'Rocksdb Unit Test under TSAN',
        'oncall':'$ONCALL',
+        'executeLocal': 'true',
        'timeout': 86400,
        'steps': [
            $CLEANUP_ENV,
@@ -557,6 +662,7 @@ TSAN_CRASH_TEST_COMMANDS="[
    {
        'name':'Rocksdb Crash Test under TSAN',
        'oncall':'$ONCALL',
+        'executeLocal': 'true',
        'timeout': 86400,
        'steps': [
            $CLEANUP_ENV,
@@ -572,6 +678,30 @@ TSAN_CRASH_TEST_COMMANDS="[
    }
 ]"

+#
+# RocksDB crash test with atomic flush under TSAN
+#
+TSAN_CRASH_TEST_WITH_ATOMIC_FLUSH_COMMANDS="[
+    {
+        'name':'Rocksdb Crash Test with atomic flush under TSAN',
+        'oncall':'$ONCALL',
+        'executeLocal': 'true',
+        'timeout': 86400,
+        'steps': [
+            $CLEANUP_ENV,
+            {
+                'name':'Compile and run',
+                'timeout': 86400,
+                'shell':'set -o pipefail && $SHM $DEBUG $TSAN $TSAN_CRASH CRASH_TEST_KILL_ODD=1887 make J=1 crash_test_with_atomic_flush || $CONTRUN_NAME=tsan_crash_test_with_atomic_flush $TASK_CREATION_TOOL',
+                'user':'root',
+                $PARSER
+            },
+            $UPLOAD_DB_DIR,
+        ],
+        $REPORT
+    }
+]"
+
 #
 # RocksDB format compatible
 #
@@ -589,6 +719,7 @@ FORMAT_COMPATIBLE_COMMANDS="[
    {
        'name':'Rocksdb Format Compatible tests',
        'oncall':'$ONCALL',
+        'executeLocal': 'true',
        'steps': [
            $CLEANUP_ENV,
            {
@@ -611,7 +742,7 @@ run_no_compression()
  rm -rf /dev/shm/rocksdb
  mkdir /dev/shm/rocksdb
  make clean
-  cat build_tools/fbcode_config.sh | grep -iv dzlib | grep -iv dlz4 | grep -iv dsnappy | grep -iv dbzip2 > .tmp.fbcode_config.sh
+  cat build_tools/fbcode_config.sh | grep -iv dzstd | grep -iv dzlib | grep -iv dlz4 | grep -iv dsnappy | grep -iv dbzip2 > .tmp.fbcode_config.sh
  mv .tmp.fbcode_config.sh build_tools/fbcode_config.sh
  cat Makefile | grep -v tools/ldb_test.py > .tmp.Makefile
  mv .tmp.Makefile Makefile
@@ -622,6 +753,7 @@ NO_COMPRESSION_COMMANDS="[
    {
        'name':'Rocksdb No Compression tests',
        'oncall':'$ONCALL',
+        'executeLocal': 'true',
        'steps': [
            $CLEANUP_ENV,
            {
@@ -699,6 +831,7 @@ JAVA_BUILD_TEST_COMMANDS="[
    {
        'name':'Rocksdb Java Build',
        'oncall':'$ONCALL',
+        'executeLocal': 'true',
        'steps': [
            $CLEANUP_ENV,
            {
@@ -753,6 +886,9 @@ case $1 in
  stress_crash)
    echo $STRESS_CRASH_TEST_COMMANDS
    ;;
+  stress_crash_with_atomic_flush)
+    echo $STRESS_CRASH_TEST_WITH_ATOMIC_FLUSH_COMMANDS
+    ;;
  write_stress)
    echo $WRITE_STRESS_COMMANDS
    ;;
@@ -762,12 +898,18 @@ case $1 in
  asan_crash)
    echo $ASAN_CRASH_TEST_COMMANDS
    ;;
+  asan_crash_with_atomic_flush)
+    echo $ASAN_CRASH_TEST_WITH_ATOMIC_FLUSH_COMMANDS
+    ;;
  ubsan)
    echo $UBSAN_TEST_COMMANDS
    ;;
  ubsan_crash)
    echo $UBSAN_CRASH_TEST_COMMANDS
    ;;
+  ubsan_crash_with_atomic_flush)
+    echo $UBSAN_CRASH_TEST_WITH_ATOMIC_FLUSH_COMMANDS
+    ;;
  valgrind)
    echo $VALGRIND_TEST_COMMANDS
    ;;
@@ -777,6 +919,9 @@ case $1 in
  tsan_crash)
    echo $TSAN_CRASH_TEST_COMMANDS
    ;;
+  tsan_crash_with_atomic_flush)
+    echo $TSAN_CRASH_TEST_WITH_ATOMIC_FLUSH_COMMANDS
+    ;;
  format_compatible)
    echo $FORMAT_COMPATIBLE_COMMANDS
    ;;
@@ -1,3 +1,4 @@
+# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved.
 # This script enables you running RocksDB tests by running
 # All the tests concurrently and utilizing all the cores
 Param(
@@ -1,4 +1,5 @@
 #!/bin/bash
+# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved.
 set -e

 ROCKSDB_VERSION="5.10.3"
@@ -1,4 +1,5 @@
 #!/bin/sh
+# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved.
 #
 # Update dependencies.sh file with the latest avaliable versions

@@ -1,4 +1,5 @@
 #!/usr/bin/env bash
+# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved.
 if [ "$#" = "0" ]; then
  echo "Usage: $0 major|minor|patch|full"
  exit 1
@@ -3,9 +3,6 @@
 //  COPYING file in the root directory) and Apache 2.0 License
 //  (found in the LICENSE.Apache file in the root directory).

-#ifndef __STDC_FORMAT_MACROS
-#define __STDC_FORMAT_MACROS
-#endif
 #ifndef GFLAGS
 #include <cstdio>
 int main() {
@@ -14,7 +11,7 @@ int main() {
 }
 #else

-#include <inttypes.h>
+#include <cinttypes>
 #include <sys/types.h>
 #include <stdio.h>

@@ -16,9 +16,9 @@
 #include <vector>
 #include "cache/clock_cache.h"
 #include "cache/lru_cache.h"
+#include "test_util/testharness.h"
 #include "util/coding.h"
 #include "util/string_util.h"
-#include "util/testharness.h"

 namespace rocksdb {

@@ -73,8 +73,7 @@ class CacheTest : public testing::TestWithParam<std::string> {
    current_ = this;
  }

-  ~CacheTest() {
-  }
+  ~CacheTest() override {}

  std::shared_ptr<Cache> NewCache(size_t capacity) {
    auto type = GetParam();
@@ -91,7 +90,7 @@ class CacheTest : public testing::TestWithParam<std::string> {
                                  bool strict_capacity_limit) {
    auto type = GetParam();
    if (type == kLRU) {
-      return NewLRUCache(capacity, num_shard_bits, strict_capacity_limit);
+      return NewLRUCache(capacity, num_shard_bits, strict_capacity_limit, 0.0);
    }
    if (type == kClock) {
      return NewClockCache(capacity, num_shard_bits, strict_capacity_limit);
@@ -99,7 +98,7 @@ class CacheTest : public testing::TestWithParam<std::string> {
    return nullptr;
  }

-  int Lookup(shared_ptr<Cache> cache, int key) {
+  int Lookup(std::shared_ptr<Cache> cache, int key) {
    Cache::Handle* handle = cache->Lookup(EncodeKey(key));
    const int r = (handle == nullptr) ? -1 : DecodeValue(cache->Value(handle));
    if (handle != nullptr) {
@@ -108,16 +107,16 @@ class CacheTest : public testing::TestWithParam<std::string> {
    return r;
  }

-  void Insert(shared_ptr<Cache> cache, int key, int value, int charge = 1) {
+  void Insert(std::shared_ptr<Cache> cache, int key, int value,
+              int charge = 1) {
    cache->Insert(EncodeKey(key), EncodeValue(value), charge,
                  &CacheTest::Deleter);
  }

-  void Erase(shared_ptr<Cache> cache, int key) {
+  void Erase(std::shared_ptr<Cache> cache, int key) {
    cache->Erase(EncodeKey(key));
  }

-
  int Lookup(int key) {
    return Lookup(cache_, key);
  }
@@ -307,7 +306,7 @@ TEST_P(CacheTest, EvictionPolicy) {
  Insert(200, 201);

  // Frequently used entry must be kept around
-  for (int i = 0; i < kCacheSize + 100; i++) {
+  for (int i = 0; i < kCacheSize + 200; i++) {
    Insert(1000+i, 2000+i);
    ASSERT_EQ(101, Lookup(100));
  }
@@ -360,7 +359,7 @@ TEST_P(CacheTest, EvictionPolicyRef) {
  Insert(303, 104);

  // Insert entries much more than Cache capacity
-  for (int i = 0; i < kCacheSize + 100; i++) {
+  for (int i = 0; i < kCacheSize + 200; i++) {
    Insert(1000 + i, 2000 + i);
  }

@@ -563,6 +562,7 @@ TEST_P(CacheTest, SetStrictCapacityLimit) {
    ASSERT_OK(s);
    ASSERT_NE(nullptr, handles[i]);
  }
+  ASSERT_EQ(10, cache->GetUsage());

  // test2: set the flag to true. Insert and check if it fails.
  std::string extra_key = "extra";
@@ -572,6 +572,7 @@ TEST_P(CacheTest, SetStrictCapacityLimit) {
  s = cache->Insert(extra_key, extra_value, 1, &deleter, &handle);
  ASSERT_TRUE(s.IsIncomplete());
  ASSERT_EQ(nullptr, handle);
+  ASSERT_EQ(10, cache->GetUsage());

  for (size_t i = 0; i < 10; i++) {
    cache->Release(handles[i]);
@@ -592,7 +593,7 @@ TEST_P(CacheTest, SetStrictCapacityLimit) {
  s = cache2->Insert(extra_key, extra_value, 1, &deleter);
  // AS if the key have been inserted into cache but get evicted immediately.
  ASSERT_OK(s);
-  ASSERT_EQ(5, cache->GetUsage());
+  ASSERT_EQ(5, cache2->GetUsage());
  ASSERT_EQ(nullptr, cache2->Lookup(extra_key));

  for (size_t i = 0; i < 5; i++) {
@@ -687,8 +688,17 @@ TEST_P(CacheTest, DefaultShardBits) {
  ASSERT_EQ(6, sc->GetNumShardBits());
 }

+TEST_P(CacheTest, GetCharge) {
+  Insert(1, 2);
+  Cache::Handle* h1 = cache_->Lookup(EncodeKey(1));
+  ASSERT_EQ(2, DecodeValue(cache_->Value(h1)));
+  ASSERT_EQ(1, cache_->GetCharge(h1));
+  cache_->Release(h1);
+}
+
 #ifdef SUPPORT_CLOCK_CACHE
-shared_ptr<Cache> (*new_clock_cache_func)(size_t, int, bool) = NewClockCache;
+std::shared_ptr<Cache> (*new_clock_cache_func)(size_t, int,
+                                               bool) = NewClockCache;
 INSTANTIATE_TEST_CASE_P(CacheTestInstance, CacheTest,
                        testing::Values(kLRU, kClock));
 #else
@@ -234,38 +234,35 @@ struct CleanupContext {
 };

 // A cache shard which maintains its own CLOCK cache.
-class ClockCacheShard : public CacheShard {
+class ClockCacheShard final : public CacheShard {
 public:
  // Hash map type.
  typedef tbb::concurrent_hash_map<CacheKey, CacheHandle*, CacheKey> HashTable;

  ClockCacheShard();
-  ~ClockCacheShard();
+  ~ClockCacheShard() override;

  // Interfaces
-  virtual void SetCapacity(size_t capacity) override;
-  virtual void SetStrictCapacityLimit(bool strict_capacity_limit) override;
-  virtual Status Insert(const Slice& key, uint32_t hash, void* value,
-                        size_t charge,
-                        void (*deleter)(const Slice& key, void* value),
-                        Cache::Handle** handle,
-                        Cache::Priority priority) override;
-  virtual Cache::Handle* Lookup(const Slice& key, uint32_t hash) override;
+  void SetCapacity(size_t capacity) override;
+  void SetStrictCapacityLimit(bool strict_capacity_limit) override;
+  Status Insert(const Slice& key, uint32_t hash, void* value, size_t charge,
+                void (*deleter)(const Slice& key, void* value),
+                Cache::Handle** handle, Cache::Priority priority) override;
+  Cache::Handle* Lookup(const Slice& key, uint32_t hash) override;
  // If the entry in in cache, increase reference count and return true.
  // Return false otherwise.
  //
  // Not necessary to hold mutex_ before being called.
-  virtual bool Ref(Cache::Handle* handle) override;
-  virtual bool Release(Cache::Handle* handle,
-                       bool force_erase = false) override;
-  virtual void Erase(const Slice& key, uint32_t hash) override;
+  bool Ref(Cache::Handle* handle) override;
+  bool Release(Cache::Handle* handle, bool force_erase = false) override;
+  void Erase(const Slice& key, uint32_t hash) override;
  bool EraseAndConfirm(const Slice& key, uint32_t hash,
                       CleanupContext* context);
-  virtual size_t GetUsage() const override;
-  virtual size_t GetPinnedUsage() const override;
-  virtual void EraseUnRefEntries() override;
-  virtual void ApplyToAllCacheEntries(void (*callback)(void*, size_t),
-                                      bool thread_safe) override;
+  size_t GetUsage() const override;
+  size_t GetPinnedUsage() const override;
+  void EraseUnRefEntries() override;
+  void ApplyToAllCacheEntries(void (*callback)(void*, size_t),
+                              bool thread_safe) override;

 private:
  static const uint32_t kInCacheBit = 1;
@@ -675,7 +672,7 @@ void ClockCacheShard::EraseUnRefEntries() {
  Cleanup(context);
 }

-class ClockCache : public ShardedCache {
+class ClockCache final : public ShardedCache {
 public:
  ClockCache(size_t capacity, int num_shard_bits, bool strict_capacity_limit)
      : ShardedCache(capacity, num_shard_bits, strict_capacity_limit) {
@@ -685,31 +682,31 @@ class ClockCache : public ShardedCache {
    SetStrictCapacityLimit(strict_capacity_limit);
  }

-  virtual ~ClockCache() { delete[] shards_; }
+  ~ClockCache() override { delete[] shards_; }

-  virtual const char* Name() const override { return "ClockCache"; }
+  const char* Name() const override { return "ClockCache"; }

-  virtual CacheShard* GetShard(int shard) override {
+  CacheShard* GetShard(int shard) override {
    return reinterpret_cast<CacheShard*>(&shards_[shard]);
  }

-  virtual const CacheShard* GetShard(int shard) const override {
+  const CacheShard* GetShard(int shard) const override {
    return reinterpret_cast<CacheShard*>(&shards_[shard]);
  }

-  virtual void* Value(Handle* handle) override {
+  void* Value(Handle* handle) override {
    return reinterpret_cast<const CacheHandle*>(handle)->value;
  }

-  virtual size_t GetCharge(Handle* handle) const override {
+  size_t GetCharge(Handle* handle) const override {
    return reinterpret_cast<const CacheHandle*>(handle)->charge;
  }

-  virtual uint32_t GetHash(Handle* handle) const override {
+  uint32_t GetHash(Handle* handle) const override {
    return reinterpret_cast<const CacheHandle*>(handle)->hash;
  }

-  virtual void DisownData() override { shards_ = nullptr; }
+  void DisownData() override { shards_ = nullptr; }

 private:
  ClockCacheShard* shards_;
@@ -7,10 +7,6 @@
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file. See the AUTHORS file for names of contributors.

-#ifndef __STDC_FORMAT_MACROS
-#define __STDC_FORMAT_MACROS
-#endif
-
 #include "cache/lru_cache.h"

 #include <assert.h>
@@ -28,7 +24,7 @@ LRUHandleTable::LRUHandleTable() : list_(nullptr), length_(0), elems_(0) {

 LRUHandleTable::~LRUHandleTable() {
  ApplyToAllCacheEntries([](LRUHandle* h) {
-    if (h->refs == 1) {
+    if (!h->HasRefs()) {
      h->Free();
    }
  });
@@ -100,14 +96,16 @@ void LRUHandleTable::Resize() {
 }

 LRUCacheShard::LRUCacheShard(size_t capacity, bool strict_capacity_limit,
-                             double high_pri_pool_ratio)
+                             double high_pri_pool_ratio,
+                             bool use_adaptive_mutex)
    : capacity_(0),
      high_pri_pool_usage_(0),
      strict_capacity_limit_(strict_capacity_limit),
      high_pri_pool_ratio_(high_pri_pool_ratio),
      high_pri_pool_capacity_(0),
      usage_(0),
-      lru_usage_(0) {
+      lru_usage_(0),
+      mutex_(use_adaptive_mutex) {
  // Make empty circular linked list
  lru_.next = &lru_;
  lru_.prev = &lru_;
@@ -115,29 +113,17 @@ LRUCacheShard::LRUCacheShard(size_t capacity, bool strict_capacity_limit,
  SetCapacity(capacity);
 }

-LRUCacheShard::~LRUCacheShard() {}
-
-bool LRUCacheShard::Unref(LRUHandle* e) {
-  assert(e->refs > 0);
-  e->refs--;
-  return e->refs == 0;
-}
-
-// Call deleter and free
-
 void LRUCacheShard::EraseUnRefEntries() {
  autovector<LRUHandle*> last_reference_list;
  {
    MutexLock l(&mutex_);
    while (lru_.next != &lru_) {
      LRUHandle* old = lru_.next;
-      assert(old->InCache());
-      assert(old->refs ==
-             1);  // LRU list contains elements which may be evicted
+      // LRU list contains only elements which can be evicted
+      assert(old->InCache() && !old->HasRefs());
      LRU_Remove(old);
      table_.Remove(old->key(), old->hash);
      old->SetInCache(false);
-      Unref(old);
      usage_ -= old->charge;
      last_reference_list.push_back(old);
    }
@@ -150,22 +136,27 @@ void LRUCacheShard::EraseUnRefEntries() {

 void LRUCacheShard::ApplyToAllCacheEntries(void (*callback)(void*, size_t),
                                           bool thread_safe) {
+  const auto applyCallback = [&]() {
+    table_.ApplyToAllCacheEntries(
+        [callback](LRUHandle* h) { callback(h->value, h->charge); });
+  };
+
  if (thread_safe) {
-    mutex_.Lock();
-  }
-  table_.ApplyToAllCacheEntries(
-      [callback](LRUHandle* h) { callback(h->value, h->charge); });
-  if (thread_safe) {
-    mutex_.Unlock();
+    MutexLock l(&mutex_);
+    applyCallback();
+  } else {
+    applyCallback();
  }
 }

 void LRUCacheShard::TEST_GetLRUList(LRUHandle** lru, LRUHandle** lru_low_pri) {
+  MutexLock l(&mutex_);
  *lru = &lru_;
  *lru_low_pri = lru_low_pri_;
 }

 size_t LRUCacheShard::TEST_GetLRUSize() {
+  MutexLock l(&mutex_);
  LRUHandle* lru_handle = lru_.next;
  size_t lru_size = 0;
  while (lru_handle != &lru_) {
@@ -233,14 +224,13 @@ void LRUCacheShard::MaintainPoolSize() {

 void LRUCacheShard::EvictFromLRU(size_t charge,
                                 autovector<LRUHandle*>* deleted) {
-  while (usage_ + charge > capacity_ && lru_.next != &lru_) {
+  while ((usage_ + charge) > capacity_ && lru_.next != &lru_) {
    LRUHandle* old = lru_.next;
-    assert(old->InCache());
-    assert(old->refs == 1);  // LRU list contains elements which may be evicted
+    // LRU list contains only elements which can be evicted
+    assert(old->InCache() && !old->HasRefs());
    LRU_Remove(old);
    table_.Remove(old->key(), old->hash);
    old->SetInCache(false);
-    Unref(old);
    usage_ -= old->charge;
    deleted->push_back(old);
  }
@@ -254,8 +244,8 @@ void LRUCacheShard::SetCapacity(size_t capacity) {
    high_pri_pool_capacity_ = capacity_ * high_pri_pool_ratio_;
    EvictFromLRU(0, &last_reference_list);
  }
-  // we free the entries here outside of mutex for
-  // performance reasons
+
+  // Free the entries outside of mutex for performance reasons
  for (auto entry : last_reference_list) {
    entry->Free();
  }
@@ -271,22 +261,22 @@ Cache::Handle* LRUCacheShard::Lookup(const Slice& key, uint32_t hash) {
  LRUHandle* e = table_.Lookup(key, hash);
  if (e != nullptr) {
    assert(e->InCache());
-    if (e->refs == 1) {
+    if (!e->HasRefs()) {
+      // The entry is in LRU since it's in hash and has no external references
      LRU_Remove(e);
    }
-    e->refs++;
+    e->Ref();
    e->SetHit();
  }
  return reinterpret_cast<Cache::Handle*>(e);
 }

 bool LRUCacheShard::Ref(Cache::Handle* h) {
-  LRUHandle* handle = reinterpret_cast<LRUHandle*>(h);
+  LRUHandle* e = reinterpret_cast<LRUHandle*>(h);
  MutexLock l(&mutex_);
-  if (handle->InCache() && handle->refs == 1) {
-    LRU_Remove(handle);
-  }
-  handle->refs++;
+  // To create another reference - entry must be already externally referenced
+  assert(e->HasRefs());
+  e->Ref();
  return true;
 }

@@ -305,30 +295,27 @@ bool LRUCacheShard::Release(Cache::Handle* handle, bool force_erase) {
  bool last_reference = false;
  {
    MutexLock l(&mutex_);
-    last_reference = Unref(e);
+    last_reference = e->Unref();
+    if (last_reference && e->InCache()) {
+      // The item is still in cache, and nobody else holds a reference to it
+      if (usage_ > capacity_ || force_erase) {
+        // The LRU list must be empty since the cache is full
+        assert(lru_.next == &lru_ || force_erase);
+        // Take this opportunity and remove the item
+        table_.Remove(e->key(), e->hash);
+        e->SetInCache(false);
+      } else {
+        // Put the item back on the LRU list, and don't free it
+        LRU_Insert(e);
+        last_reference = false;
+      }
+    }
    if (last_reference) {
      usage_ -= e->charge;
    }
-    if (e->refs == 1 && e->InCache()) {
-      // The item is still in cache, and nobody else holds a reference to it
-      if (usage_ > capacity_ || force_erase) {
-        // the cache is full
-        // The LRU list must be empty since the cache is full
-        assert(!(usage_ > capacity_) || lru_.next == &lru_);
-        // take this opportunity and remove the item
-        table_.Remove(e->key(), e->hash);
-        e->SetInCache(false);
-        Unref(e);
-        usage_ -= e->charge;
-        last_reference = true;
-      } else {
-        // put the item on the list to be potentially freed
-        LRU_Insert(e);
-      }
-    }
  }

-  // free outside of mutex
+  // Free the entry here outside of mutex for performance reasons
  if (last_reference) {
    e->Free();
  }
@@ -344,7 +331,7 @@ Status LRUCacheShard::Insert(const Slice& key, uint32_t hash, void* value,
  // It shouldn't happen very often though.
  LRUHandle* e = reinterpret_cast<LRUHandle*>(
      new char[sizeof(LRUHandle) - 1 + key.size()]);
-  Status s;
+  Status s = Status::OK();
  autovector<LRUHandle*> last_reference_list;

  e->value = value;
@@ -353,9 +340,7 @@ Status LRUCacheShard::Insert(const Slice& key, uint32_t hash, void* value,
  e->key_length = key.size();
  e->flags = 0;
  e->hash = hash;
-  e->refs = (handle == nullptr
-                 ? 1
-                 : 2);  // One from LRUCache, one for the returned handle
+  e->refs = 0;
  e->next = e->prev = nullptr;
  e->SetInCache(true);
  e->SetPriority(priority);
@@ -368,11 +353,12 @@ Status LRUCacheShard::Insert(const Slice& key, uint32_t hash, void* value,
    // is freed or the lru list is empty
    EvictFromLRU(charge, &last_reference_list);

-    if (usage_ - lru_usage_ + charge > capacity_ &&
+    if ((usage_ + charge) > capacity_ &&
        (strict_capacity_limit_ || handle == nullptr)) {
      if (handle == nullptr) {
        // Don't insert the entry but still return ok, as if the entry inserted
        // into cache and get evicted immediately.
+        e->SetInCache(false);
        last_reference_list.push_back(e);
      } else {
        delete[] reinterpret_cast<char*>(e);
@@ -380,32 +366,30 @@ Status LRUCacheShard::Insert(const Slice& key, uint32_t hash, void* value,
        s = Status::Incomplete("Insert failed due to LRU cache being full.");
      }
    } else {
-      // insert into the cache
-      // note that the cache might get larger than its capacity if not enough
-      // space was freed
+      // Insert into the cache. Note that the cache might get larger than its
+      // capacity if not enough space was freed up.
      LRUHandle* old = table_.Insert(e);
      usage_ += e->charge;
      if (old != nullptr) {
+        assert(old->InCache());
        old->SetInCache(false);
-        if (Unref(old)) {
-          usage_ -= old->charge;
-          // old is on LRU because it's in cache and its reference count
-          // was just 1 (Unref returned 0)
+        if (!old->HasRefs()) {
+          // old is on LRU because it's in cache and its reference count is 0
          LRU_Remove(old);
+          usage_ -= old->charge;
          last_reference_list.push_back(old);
        }
      }
      if (handle == nullptr) {
        LRU_Insert(e);
      } else {
+        e->Ref();
        *handle = reinterpret_cast<Cache::Handle*>(e);
      }
-      s = Status::OK();
    }
  }

-  // we free the entries here outside of mutex for
-  // performance reasons
+  // Free the entries here outside of mutex for performance reasons
  for (auto entry : last_reference_list) {
    entry->Free();
  }
@@ -420,18 +404,18 @@ void LRUCacheShard::Erase(const Slice& key, uint32_t hash) {
    MutexLock l(&mutex_);
    e = table_.Remove(key, hash);
    if (e != nullptr) {
-      last_reference = Unref(e);
-      if (last_reference) {
-        usage_ -= e->charge;
-      }
-      if (last_reference && e->InCache()) {
-        LRU_Remove(e);
-      }
+      assert(e->InCache());
      e->SetInCache(false);
+      if (!e->HasRefs()) {
+        // The entry is in LRU since it's in hash and has no external references
+        LRU_Remove(e);
+        usage_ -= e->charge;
+        last_reference = true;
+      }
    }
  }

-  // mutex not held here
+  // Free the entry here outside of mutex for performance reasons
  // last_reference will only be true if e != nullptr
  if (last_reference) {
    e->Free();
@@ -462,7 +446,8 @@ std::string LRUCacheShard::GetPrintableOptions() const {

 LRUCache::LRUCache(size_t capacity, int num_shard_bits,
                   bool strict_capacity_limit, double high_pri_pool_ratio,
-                   std::shared_ptr<MemoryAllocator> allocator)
+                   std::shared_ptr<MemoryAllocator> allocator,
+                   bool use_adaptive_mutex)
    : ShardedCache(capacity, num_shard_bits, strict_capacity_limit,
                   std::move(allocator)) {
  num_shards_ = 1 << num_shard_bits;
@@ -471,7 +456,8 @@ LRUCache::LRUCache(size_t capacity, int num_shard_bits,
  size_t per_shard = (capacity + (num_shards_ - 1)) / num_shards_;
  for (int i = 0; i < num_shards_; i++) {
    new (&shards_[i])
-        LRUCacheShard(per_shard, strict_capacity_limit, high_pri_pool_ratio);
+        LRUCacheShard(per_shard, strict_capacity_limit, high_pri_pool_ratio,
+            use_adaptive_mutex);
  }
 }

@@ -540,13 +526,15 @@ std::shared_ptr<Cache> NewLRUCache(const LRUCacheOptions& cache_opts) {
  return NewLRUCache(cache_opts.capacity, cache_opts.num_shard_bits,
                     cache_opts.strict_capacity_limit,
                     cache_opts.high_pri_pool_ratio,
-                     cache_opts.memory_allocator);
+                     cache_opts.memory_allocator,
+                     cache_opts.use_adaptive_mutex);
 }

 std::shared_ptr<Cache> NewLRUCache(
    size_t capacity, int num_shard_bits, bool strict_capacity_limit,
    double high_pri_pool_ratio,
-    std::shared_ptr<MemoryAllocator> memory_allocator) {
+    std::shared_ptr<MemoryAllocator> memory_allocator,
+    bool use_adaptive_mutex) {
  if (num_shard_bits >= 20) {
    return nullptr;  // the cache cannot be sharded into too many fine pieces
  }
@@ -559,7 +547,8 @@ std::shared_ptr<Cache> NewLRUCache(
  }
  return std::make_shared<LRUCache>(capacity, num_shard_bits,
                                    strict_capacity_limit, high_pri_pool_ratio,
-                                    std::move(memory_allocator));
+                                    std::move(memory_allocator),
+                                    use_adaptive_mutex);
 }

 }  // namespace rocksdb
@@ -17,31 +17,34 @@

 namespace rocksdb {

-// LRU cache implementation
+// LRU cache implementation. This class is not thread-safe.

 // An entry is a variable length heap-allocated structure.
 // Entries are referenced by cache and/or by any external entity.
-// The cache keeps all its entries in table. Some elements
+// The cache keeps all its entries in a hash table. Some elements
 // are also stored on LRU list.
 //
 // LRUHandle can be in these states:
 // 1. Referenced externally AND in hash table.
-//  In that case the entry is *not* in the LRU. (refs > 1 && in_cache == true)
-// 2. Not referenced externally and in hash table. In that case the entry is
-// in the LRU and can be freed. (refs == 1 && in_cache == true)
-// 3. Referenced externally and not in hash table. In that case the entry is
-// in not on LRU and not in table. (refs >= 1 && in_cache == false)
+//    In that case the entry is *not* in the LRU list
+//    (refs >= 1 && in_cache == true)
+// 2. Not referenced externally AND in hash table.
+//    In that case the entry is in the LRU list and can be freed.
+//    (refs == 0 && in_cache == true)
+// 3. Referenced externally AND not in hash table.
+//    In that case the entry is not in the LRU list and not in hash table.
+//    The entry can be freed when refs becomes 0.
+//    (refs >= 1 && in_cache == false)
 //
 // All newly created LRUHandles are in state 1. If you call
-// LRUCacheShard::Release
-// on entry in state 1, it will go into state 2. To move from state 1 to
-// state 3, either call LRUCacheShard::Erase or LRUCacheShard::Insert with the
-// same key.
+// LRUCacheShard::Release on entry in state 1, it will go into state 2.
+// To move from state 1 to state 3, either call LRUCacheShard::Erase or
+// LRUCacheShard::Insert with the same key (but possibly different value).
 // To move from state 2 to state 1, use LRUCacheShard::Lookup.
 // Before destruction, make sure that no handles are in state 1. This means
 // that any successful LRUCacheShard::Lookup/LRUCacheShard::Insert have a
-// matching
-// RUCache::Release (to move into state 2) or LRUCacheShard::Erase (for state 3)
+// matching LRUCache::Release (to move into state 2) or LRUCacheShard::Erase
+// (to move into state 3).

 struct LRUHandle {
  void* value;
@@ -51,62 +54,75 @@ struct LRUHandle {
  LRUHandle* prev;
  size_t charge;  // TODO(opt): Only allow uint32_t?
  size_t key_length;
-  uint32_t refs;     // a number of refs to this entry
-                     // cache itself is counted as 1
+  // The hash of key(). Used for fast sharding and comparisons.
+  uint32_t hash;
+  // The number of external refs to this entry. The cache itself is not counted.
+  uint32_t refs;

-  // Include the following flags:
-  //   in_cache:    whether this entry is referenced by the hash table.
-  //   is_high_pri: whether this entry is high priority entry.
-  //   in_high_pri_pool: whether this entry is in high-pri pool.
-  char flags;
+  enum Flags : uint8_t {
+    // Whether this entry is referenced by the hash table.
+    IN_CACHE = (1 << 0),
+    // Whether this entry is high priority entry.
+    IS_HIGH_PRI = (1 << 1),
+    // Whether this entry is in high-pri pool.
+    IN_HIGH_PRI_POOL = (1 << 2),
+    // Wwhether this entry has had any lookups (hits).
+    HAS_HIT = (1 << 3),
+  };

-  uint32_t hash;     // Hash of key(); used for fast sharding and comparisons
+  uint8_t flags;

-  char key_data[1];  // Beginning of key
+  // Beginning of the key (MUST BE THE LAST FIELD IN THIS STRUCT!)
+  char key_data[1];

-  Slice key() const {
-    // For cheaper lookups, we allow a temporary Handle object
-    // to store a pointer to a key in "value".
-    if (next == this) {
-      return *(reinterpret_cast<Slice*>(value));
-    } else {
-      return Slice(key_data, key_length);
-    }
+  Slice key() const { return Slice(key_data, key_length); }
+
+  // Increase the reference count by 1.
+  void Ref() { refs++; }
+
+  // Just reduce the reference count by 1. Return true if it was last reference.
+  bool Unref() {
+    assert(refs > 0);
+    refs--;
+    return refs == 0;
  }

-  bool InCache() { return flags & 1; }
-  bool IsHighPri() { return flags & 2; }
-  bool InHighPriPool() { return flags & 4; }
-  bool HasHit() { return flags & 8; }
+  // Return true if there are external refs, false otherwise.
+  bool HasRefs() const { return refs > 0; }
+
+  bool InCache() const { return flags & IN_CACHE; }
+  bool IsHighPri() const { return flags & IS_HIGH_PRI; }
+  bool InHighPriPool() const { return flags & IN_HIGH_PRI_POOL; }
+  bool HasHit() const { return flags & HAS_HIT; }

  void SetInCache(bool in_cache) {
    if (in_cache) {
-      flags |= 1;
+      flags |= IN_CACHE;
    } else {
-      flags &= ~1;
+      flags &= ~IN_CACHE;
    }
  }

  void SetPriority(Cache::Priority priority) {
    if (priority == Cache::Priority::HIGH) {
-      flags |= 2;
+      flags |= IS_HIGH_PRI;
    } else {
-      flags &= ~2;
+      flags &= ~IS_HIGH_PRI;
    }
  }

  void SetInHighPriPool(bool in_high_pri_pool) {
    if (in_high_pri_pool) {
-      flags |= 4;
+      flags |= IN_HIGH_PRI_POOL;
    } else {
-      flags &= ~4;
+      flags &= ~IN_HIGH_PRI_POOL;
    }
  }

-  void SetHit() { flags |= 8; }
+  void SetHit() { flags |= HAS_HIT; }

  void Free() {
-    assert((refs == 1 && InCache()) || (refs == 0 && !InCache()));
+    assert(refs == 0);
    if (deleter) {
      (*deleter)(key(), value);
    }
@@ -157,11 +173,11 @@ class LRUHandleTable {
 };

 // A single shard of sharded cache.
-class ALIGN_AS(CACHE_LINE_SIZE) LRUCacheShard : public CacheShard {
+class ALIGN_AS(CACHE_LINE_SIZE) LRUCacheShard final : public CacheShard {
 public:
  LRUCacheShard(size_t capacity, bool strict_capacity_limit,
-                double high_pri_pool_ratio);
-  virtual ~LRUCacheShard();
+                double high_pri_pool_ratio, bool use_adaptive_mutex);
+  virtual ~LRUCacheShard() override = default;

  // Separate from constructor so caller can easily make an array of LRUCache
  // if current usage is more than new capacity, the function will attempt to
@@ -217,10 +233,6 @@ class ALIGN_AS(CACHE_LINE_SIZE) LRUCacheShard : public CacheShard {
  // high-pri pool is no larger than the size specify by high_pri_pool_pct.
  void MaintainPoolSize();

-  // Just reduce the reference count by 1.
-  // Return true if last reference
-  bool Unref(LRUHandle* e);
-
  // Free some space following strict LRU policy until enough space
  // to hold (usage_ + charge) is freed or the lru list is empty
  // This function is not thread safe - it needs to be executed while
@@ -276,11 +288,16 @@ class ALIGN_AS(CACHE_LINE_SIZE) LRUCacheShard : public CacheShard {
  mutable port::Mutex mutex_;
 };

-class LRUCache : public ShardedCache {
+class LRUCache
+#ifdef NDEBUG
+    final
+#endif
+    : public ShardedCache {
 public:
  LRUCache(size_t capacity, int num_shard_bits, bool strict_capacity_limit,
           double high_pri_pool_ratio,
-           std::shared_ptr<MemoryAllocator> memory_allocator = nullptr);
+           std::shared_ptr<MemoryAllocator> memory_allocator = nullptr,
+           bool use_adaptive_mutex = kDefaultToAdaptiveMutex);
  virtual ~LRUCache();
  virtual const char* Name() const override { return "LRUCache"; }
  virtual CacheShard* GetShard(int shard) override;
@@ -8,14 +8,14 @@
 #include <string>
 #include <vector>
 #include "port/port.h"
-#include "util/testharness.h"
+#include "test_util/testharness.h"

 namespace rocksdb {

 class LRUCacheTest : public testing::Test {
 public:
  LRUCacheTest() {}
-  ~LRUCacheTest() { DeleteCache(); }
+  ~LRUCacheTest() override { DeleteCache(); }

  void DeleteCache() {
    if (cache_ != nullptr) {
@@ -25,12 +25,13 @@ class LRUCacheTest : public testing::Test {
    }
  }

-  void NewCache(size_t capacity, double high_pri_pool_ratio = 0.0) {
+  void NewCache(size_t capacity, double high_pri_pool_ratio = 0.0,
+                bool use_adaptive_mutex = kDefaultToAdaptiveMutex) {
    DeleteCache();
    cache_ = reinterpret_cast<LRUCacheShard*>(
        port::cacheline_aligned_alloc(sizeof(LRUCacheShard)));
    new (cache_) LRUCacheShard(capacity, false /*strict_capcity_limit*/,
-                               high_pri_pool_ratio);
+                               high_pri_pool_ratio, use_adaptive_mutex);
  }

  void Insert(const std::string& key,
@@ -7,10 +7,6 @@
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file. See the AUTHORS file for names of contributors.

-#ifndef __STDC_FORMAT_MACROS
-#define __STDC_FORMAT_MACROS
-#endif
-
 #include "cache/sharded_cache.h"

 #include <string>
@@ -54,7 +54,8 @@ class ShardedCache : public Cache {
  virtual CacheShard* GetShard(int shard) = 0;
  virtual const CacheShard* GetShard(int shard) const = 0;
  virtual void* Value(Handle* handle) override = 0;
-  virtual size_t GetCharge(Handle* handle) const = 0;
+  virtual size_t GetCharge(Handle* handle) const override = 0;
+
  virtual uint32_t GetHash(Handle* handle) const = 0;
  virtual void DisownData() override = 0;

@@ -83,7 +84,7 @@ class ShardedCache : public Cache {

 private:
  static inline uint32_t HashSlice(const Slice& s) {
-    return Hash(s.data(), s.size(), 0);
+    return static_cast<uint32_t>(GetSliceNPHash64(s));
  }

  uint32_t Shard(uint32_t hash) {
@@ -1,4 +1,5 @@
 #!/usr/bin/env bash
+# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved.

 # Exit on error.
 set -e
@@ -1,3 +1,5 @@
+#!/usr/bin/env python2
+# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved.
 import re
 import sys

@@ -13,7 +13,7 @@
 #include <deque>
 #include <vector>

-#include "db/compaction_iterator.h"
+#include "db/compaction/compaction_iterator.h"
 #include "db/dbformat.h"
 #include "db/event_helpers.h"
 #include "db/internal_stats.h"
@@ -21,6 +21,7 @@
 #include "db/range_del_aggregator.h"
 #include "db/table_cache.h"
 #include "db/version_edit.h"
+#include "file/filename.h"
 #include "monitoring/iostats_context_imp.h"
 #include "monitoring/thread_status_util.h"
 #include "rocksdb/db.h"
@@ -28,13 +29,12 @@
 #include "rocksdb/iterator.h"
 #include "rocksdb/options.h"
 #include "rocksdb/table.h"
-#include "table/block_based_table_builder.h"
+#include "table/block_based/block_based_table_builder.h"
 #include "table/format.h"
 #include "table/internal_iterator.h"
+#include "test_util/sync_point.h"
 #include "util/file_reader_writer.h"
-#include "util/filename.h"
 #include "util/stop_watch.h"
-#include "util/sync_point.h"

 namespace rocksdb {

@@ -47,18 +47,20 @@ TableBuilder* NewTableBuilder(
        int_tbl_prop_collector_factories,
    uint32_t column_family_id, const std::string& column_family_name,
    WritableFileWriter* file, const CompressionType compression_type,
-    const CompressionOptions& compression_opts, int level,
-    const std::string* compression_dict, const bool skip_filters,
-    const uint64_t creation_time, const uint64_t oldest_key_time) {
+    uint64_t sample_for_compression, const CompressionOptions& compression_opts,
+    int level, const bool skip_filters, const uint64_t creation_time,
+    const uint64_t oldest_key_time, const uint64_t target_file_size,
+    const uint64_t file_creation_time) {
  assert((column_family_id ==
          TablePropertiesCollectorFactory::Context::kUnknownColumnFamily) ==
         column_family_name.empty());
  return ioptions.table_factory->NewTableBuilder(
      TableBuilderOptions(ioptions, moptions, internal_comparator,
                          int_tbl_prop_collector_factories, compression_type,
-                          compression_opts, compression_dict, skip_filters,
-                          column_family_name, level, creation_time,
-                          oldest_key_time),
+                          sample_for_compression, compression_opts,
+                          skip_filters, column_family_name, level,
+                          creation_time, oldest_key_time, target_file_size,
+                          file_creation_time),
      column_family_id, file);
 }

@@ -75,11 +77,12 @@ Status BuildTable(
    std::vector<SequenceNumber> snapshots,
    SequenceNumber earliest_write_conflict_snapshot,
    SnapshotChecker* snapshot_checker, const CompressionType compression,
-    const CompressionOptions& compression_opts, bool paranoid_file_checks,
-    InternalStats* internal_stats, TableFileCreationReason reason,
-    EventLogger* event_logger, int job_id, const Env::IOPriority io_priority,
-    TableProperties* table_properties, int level, const uint64_t creation_time,
-    const uint64_t oldest_key_time, Env::WriteLifeTimeHint write_hint) {
+    uint64_t sample_for_compression, const CompressionOptions& compression_opts,
+    bool paranoid_file_checks, InternalStats* internal_stats,
+    TableFileCreationReason reason, EventLogger* event_logger, int job_id,
+    const Env::IOPriority io_priority, TableProperties* table_properties,
+    int level, const uint64_t creation_time, const uint64_t oldest_key_time,
+    Env::WriteLifeTimeHint write_hint, const uint64_t file_creation_time) {
  assert((column_family_id ==
          TablePropertiesCollectorFactory::Context::kUnknownColumnFamily) ==
         column_family_name.empty());
@@ -105,6 +108,11 @@ Status BuildTable(
  if (iter->Valid() || !range_del_agg->IsEmpty()) {
    TableBuilder* builder;
    std::unique_ptr<WritableFileWriter> file_writer;
+    // Currently we only enable dictionary compression during compaction to the
+    // bottommost level.
+    CompressionOptions compression_opts_for_flush(compression_opts);
+    compression_opts_for_flush.max_dict_bytes = 0;
+    compression_opts_for_flush.zstd_max_train_bytes = 0;
    {
      std::unique_ptr<WritableFile> file;
 #ifndef NDEBUG
@@ -127,9 +135,10 @@ Status BuildTable(
      builder = NewTableBuilder(
          ioptions, mutable_cf_options, internal_comparator,
          int_tbl_prop_collector_factories, column_family_id,
-          column_family_name, file_writer.get(), compression, compression_opts,
-          level, nullptr /* compression_dict */, false /* skip_filters */,
-          creation_time, oldest_key_time);
+          column_family_name, file_writer.get(), compression,
+          sample_for_compression, compression_opts_for_flush, level,
+          false /* skip_filters */, creation_time, oldest_key_time,
+          0 /*target_file_size*/, file_creation_time);
    }

    MergeHelper merge(env, internal_comparator.user_comparator(),
@@ -212,8 +221,9 @@ Status BuildTable(
          mutable_cf_options.prefix_extractor.get(), nullptr,
          (internal_stats == nullptr) ? nullptr
                                      : internal_stats->GetFileReadHist(0),
-          false /* for_compaction */, nullptr /* arena */,
-          false /* skip_filter */, level));
+          TableReaderCaller::kFlush, /*arena=*/nullptr,
+          /*skip_filter=*/false, level, /*smallest_compaction_key=*/nullptr,
+          /*largest_compaction_key*/ nullptr));
      s = it->status();
      if (s.ok() && paranoid_file_checks) {
        for (it->SeekToFirst(); it->Valid(); it->Next()) {
@@ -11,6 +11,7 @@
 #include <vector>
 #include "db/range_tombstone_fragmenter.h"
 #include "db/table_properties_collector.h"
+#include "logging/event_logger.h"
 #include "options/cf_options.h"
 #include "rocksdb/comparator.h"
 #include "rocksdb/env.h"
@@ -20,7 +21,6 @@
 #include "rocksdb/table_properties.h"
 #include "rocksdb/types.h"
 #include "table/scoped_arena_iterator.h"
-#include "util/event_logger.h"

 namespace rocksdb {

@@ -40,8 +40,6 @@ class InternalStats;
 // @param column_family_name Name of the column family that is also identified
 //    by column_family_id, or empty string if unknown. It must outlive the
 //    TableBuilder returned by this function.
-// @param compression_dict Data for presetting the compression library's
-//    dictionary, or nullptr.
 TableBuilder* NewTableBuilder(
    const ImmutableCFOptions& options, const MutableCFOptions& moptions,
    const InternalKeyComparator& internal_comparator,
@@ -49,10 +47,11 @@ TableBuilder* NewTableBuilder(
        int_tbl_prop_collector_factories,
    uint32_t column_family_id, const std::string& column_family_name,
    WritableFileWriter* file, const CompressionType compression_type,
+    const uint64_t sample_for_compression,
    const CompressionOptions& compression_opts, int level,
-    const std::string* compression_dict = nullptr,
    const bool skip_filters = false, const uint64_t creation_time = 0,
-    const uint64_t oldest_key_time = 0);
+    const uint64_t oldest_key_time = 0, const uint64_t target_file_size = 0,
+    const uint64_t file_creation_time = 0);

 // Build a Table file from the contents of *iter.  The generated file
 // will be named according to number specified in meta. On success, the rest of
@@ -75,12 +74,14 @@ extern Status BuildTable(
    std::vector<SequenceNumber> snapshots,
    SequenceNumber earliest_write_conflict_snapshot,
    SnapshotChecker* snapshot_checker, const CompressionType compression,
+    const uint64_t sample_for_compression,
    const CompressionOptions& compression_opts, bool paranoid_file_checks,
    InternalStats* internal_stats, TableFileCreationReason reason,
    EventLogger* event_logger = nullptr, int job_id = 0,
    const Env::IOPriority io_priority = Env::IO_HIGH,
    TableProperties* table_properties = nullptr, int level = -1,
    const uint64_t creation_time = 0, const uint64_t oldest_key_time = 0,
-    Env::WriteLifeTimeHint write_hint = Env::WLTH_NOT_SET);
+    Env::WriteLifeTimeHint write_hint = Env::WLTH_NOT_SET,
+    const uint64_t file_creation_time = 0);

 }  // namespace rocksdb
@@ -208,13 +208,10 @@ struct rocksdb_compactionfilter_t : public CompactionFilter {
  const char* (*name_)(void*);
  unsigned char ignore_snapshots_;

-  virtual ~rocksdb_compactionfilter_t() {
-    (*destructor_)(state_);
-  }
+  ~rocksdb_compactionfilter_t() override { (*destructor_)(state_); }

-  virtual bool Filter(int level, const Slice& key, const Slice& existing_value,
-                      std::string* new_value,
-                      bool* value_changed) const override {
+  bool Filter(int level, const Slice& key, const Slice& existing_value,
+              std::string* new_value, bool* value_changed) const override {
    char* c_new_value = nullptr;
    size_t new_value_length = 0;
    unsigned char c_value_changed = 0;
@@ -231,9 +228,9 @@ struct rocksdb_compactionfilter_t : public CompactionFilter {
    return result;
  }

-  virtual const char* Name() const override { return (*name_)(state_); }
+  const char* Name() const override { return (*name_)(state_); }

-  virtual bool IgnoreSnapshots() const override { return ignore_snapshots_; }
+  bool IgnoreSnapshots() const override { return ignore_snapshots_; }
 };

 struct rocksdb_compactionfilterfactory_t : public CompactionFilterFactory {
@@ -243,9 +240,9 @@ struct rocksdb_compactionfilterfactory_t : public CompactionFilterFactory {
      void*, rocksdb_compactionfiltercontext_t* context);
  const char* (*name_)(void*);

-  virtual ~rocksdb_compactionfilterfactory_t() { (*destructor_)(state_); }
+  ~rocksdb_compactionfilterfactory_t() override { (*destructor_)(state_); }

-  virtual std::unique_ptr<CompactionFilter> CreateCompactionFilter(
+  std::unique_ptr<CompactionFilter> CreateCompactionFilter(
      const CompactionFilter::Context& context) override {
    rocksdb_compactionfiltercontext_t ccontext;
    ccontext.rep = context;
@@ -253,7 +250,7 @@ struct rocksdb_compactionfilterfactory_t : public CompactionFilterFactory {
    return std::unique_ptr<CompactionFilter>(cf);
  }

-  virtual const char* Name() const override { return (*name_)(state_); }
+  const char* Name() const override { return (*name_)(state_); }
 };

 struct rocksdb_comparator_t : public Comparator {
@@ -265,20 +262,17 @@ struct rocksdb_comparator_t : public Comparator {
      const char* b, size_t blen);
  const char* (*name_)(void*);

-  virtual ~rocksdb_comparator_t() {
-    (*destructor_)(state_);
-  }
+  ~rocksdb_comparator_t() override { (*destructor_)(state_); }

-  virtual int Compare(const Slice& a, const Slice& b) const override {
+  int Compare(const Slice& a, const Slice& b) const override {
    return (*compare_)(state_, a.data(), a.size(), b.data(), b.size());
  }

-  virtual const char* Name() const override { return (*name_)(state_); }
+  const char* Name() const override { return (*name_)(state_); }

  // No-ops since the C binding does not support key shortening methods.
-  virtual void FindShortestSeparator(std::string*,
-                                     const Slice&) const override {}
-  virtual void FindShortSuccessor(std::string* /*key*/) const override {}
+  void FindShortestSeparator(std::string*, const Slice&) const override {}
+  void FindShortSuccessor(std::string* /*key*/) const override {}
 };

 struct rocksdb_filterpolicy_t : public FilterPolicy {
@@ -298,14 +292,11 @@ struct rocksdb_filterpolicy_t : public FilterPolicy {
      void*,
      const char* filter, size_t filter_length);

-  virtual ~rocksdb_filterpolicy_t() {
-    (*destructor_)(state_);
-  }
+  ~rocksdb_filterpolicy_t() override { (*destructor_)(state_); }

-  virtual const char* Name() const override { return (*name_)(state_); }
+  const char* Name() const override { return (*name_)(state_); }

-  virtual void CreateFilter(const Slice* keys, int n,
-                            std::string* dst) const override {
+  void CreateFilter(const Slice* keys, int n, std::string* dst) const override {
    std::vector<const char*> key_pointers(n);
    std::vector<size_t> key_sizes(n);
    for (int i = 0; i < n; i++) {
@@ -323,8 +314,7 @@ struct rocksdb_filterpolicy_t : public FilterPolicy {
    }
  }

-  virtual bool KeyMayMatch(const Slice& key,
-                           const Slice& filter) const override {
+  bool KeyMayMatch(const Slice& key, const Slice& filter) const override {
    return (*key_match_)(state_, key.data(), key.size(),
                         filter.data(), filter.size());
  }
@@ -349,14 +339,12 @@ struct rocksdb_mergeoperator_t : public MergeOperator {
      void*,
      const char* value, size_t value_length);

-  virtual ~rocksdb_mergeoperator_t() {
-    (*destructor_)(state_);
-  }
+  ~rocksdb_mergeoperator_t() override { (*destructor_)(state_); }

-  virtual const char* Name() const override { return (*name_)(state_); }
+  const char* Name() const override { return (*name_)(state_); }

-  virtual bool FullMergeV2(const MergeOperationInput& merge_in,
-                           MergeOperationOutput* merge_out) const override {
+  bool FullMergeV2(const MergeOperationInput& merge_in,
+                   MergeOperationOutput* merge_out) const override {
    size_t n = merge_in.operand_list.size();
    std::vector<const char*> operand_pointers(n);
    std::vector<size_t> operand_sizes(n);
@@ -390,10 +378,10 @@ struct rocksdb_mergeoperator_t : public MergeOperator {
    return success;
  }

-  virtual bool PartialMergeMulti(const Slice& key,
-                                 const std::deque<Slice>& operand_list,
-                                 std::string* new_value,
-                                 Logger* /*logger*/) const override {
+  bool PartialMergeMulti(const Slice& key,
+                         const std::deque<Slice>& operand_list,
+                         std::string* new_value,
+                         Logger* /*logger*/) const override {
    size_t operand_count = operand_list.size();
    std::vector<const char*> operand_pointers(operand_count);
    std::vector<size_t> operand_sizes(operand_count);
@@ -444,23 +432,21 @@ struct rocksdb_slicetransform_t : public SliceTransform {
      void*,
      const char* key, size_t length);

-  virtual ~rocksdb_slicetransform_t() {
-    (*destructor_)(state_);
-  }
+  ~rocksdb_slicetransform_t() override { (*destructor_)(state_); }

-  virtual const char* Name() const override { return (*name_)(state_); }
+  const char* Name() const override { return (*name_)(state_); }

-  virtual Slice Transform(const Slice& src) const override {
+  Slice Transform(const Slice& src) const override {
    size_t len;
    char* dst = (*transform_)(state_, src.data(), src.size(), &len);
    return Slice(dst, len);
  }

-  virtual bool InDomain(const Slice& src) const override {
+  bool InDomain(const Slice& src) const override {
    return (*in_domain_)(state_, src.data(), src.size());
  }

-  virtual bool InRange(const Slice& src) const override {
+  bool InRange(const Slice& src) const override {
    return (*in_range_)(state_, src.data(), src.size());
  }
 };
@@ -531,6 +517,21 @@ rocksdb_t* rocksdb_open_for_read_only(
  return result;
 }

+rocksdb_t* rocksdb_open_as_secondary(const rocksdb_options_t* options,
+                                     const char* name,
+                                     const char* secondary_path,
+                                     char** errptr) {
+  DB* db;
+  if (SaveError(errptr,
+                DB::OpenAsSecondary(options->rep, std::string(name),
+                                    std::string(secondary_path), &db))) {
+    return nullptr;
+  }
+  rocksdb_t* result = new rocksdb_t;
+  result->rep = db;
+  return result;
+}
+
 rocksdb_backup_engine_t* rocksdb_backup_engine_open(
    const rocksdb_options_t* options, const char* path, char** errptr) {
  BackupEngine* be;
@@ -731,6 +732,37 @@ rocksdb_t* rocksdb_open_for_read_only_column_families(
  return result;
 }

+rocksdb_t* rocksdb_open_as_secondary_column_families(
+    const rocksdb_options_t* db_options, const char* name,
+    const char* secondary_path, int num_column_families,
+    const char** column_family_names,
+    const rocksdb_options_t** column_family_options,
+    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.emplace_back(
+        std::string(column_family_names[i]),
+        ColumnFamilyOptions(column_family_options[i]->rep));
+  }
+  DB* db;
+  std::vector<ColumnFamilyHandle*> handles;
+  if (SaveError(errptr, DB::OpenAsSecondary(DBOptions(db_options->rep),
+                                            std::string(name),
+                                            std::string(secondary_path),
+                                            column_families, &handles, &db))) {
+    return nullptr;
+  }
+  for (size_t i = 0; i != handles.size(); ++i) {
+    rocksdb_column_family_handle_t* c_handle =
+        new rocksdb_column_family_handle_t;
+    c_handle->rep = handles[i];
+    column_family_handles[i] = c_handle;
+  }
+  rocksdb_t* result = new rocksdb_t;
+  result->rep = db;
+  return result;
+}
+
 char** rocksdb_list_column_families(
    const rocksdb_options_t* options,
    const char* name,
@@ -1002,7 +1034,7 @@ void rocksdb_wal_iter_destroy (const rocksdb_wal_iterator_t* iter) {
 rocksdb_writebatch_t* rocksdb_wal_iter_get_batch (const rocksdb_wal_iterator_t* iter, uint64_t* seq) {
  rocksdb_writebatch_t* result = rocksdb_writebatch_create();
  BatchResult wal_batch = iter->rep->GetBatch();
-  result->rep = * wal_batch.writeBatchPtr.release();
+  result->rep = std::move(*wal_batch.writeBatchPtr);
  if (seq != nullptr) {
    *seq = wal_batch.sequence;
  }
@@ -1084,6 +1116,18 @@ int rocksdb_property_int(
  }
 }

+int rocksdb_property_int_cf(
+    rocksdb_t* db,
+    rocksdb_column_family_handle_t* column_family,
+    const char* propname,
+    uint64_t *out_val) {
+  if (db->rep->GetIntProperty(column_family->rep, Slice(propname), out_val)) {
+    return 0;
+  } else {
+    return -1;
+  }
+}
+
 char* rocksdb_property_value_cf(
    rocksdb_t* db,
    rocksdb_column_family_handle_t* column_family,
@@ -1197,6 +1241,14 @@ void rocksdb_flush(
  SaveError(errptr, db->rep->Flush(options->rep));
 }

+void rocksdb_flush_cf(
+    rocksdb_t* db,
+    const rocksdb_flushoptions_t* options,
+    rocksdb_column_family_handle_t* column_family,
+    char** errptr) {
+  SaveError(errptr, db->rep->Flush(options->rep, column_family->rep));
+}
+
 void rocksdb_disable_file_deletions(
    rocksdb_t* db,
    char** errptr) {
@@ -1495,10 +1547,10 @@ class H : public WriteBatch::Handler {
  void* state_;
  void (*put_)(void*, const char* k, size_t klen, const char* v, size_t vlen);
  void (*deleted_)(void*, const char* k, size_t klen);
-  virtual void Put(const Slice& key, const Slice& value) override {
+  void Put(const Slice& key, const Slice& value) override {
    (*put_)(state_, key.data(), key.size(), value.data(), value.size());
  }
-  virtual void Delete(const Slice& key) override {
+  void Delete(const Slice& key) override {
    (*deleted_)(state_, key.data(), key.size());
  }
 };
@@ -2220,6 +2272,11 @@ void rocksdb_options_set_max_bytes_for_level_base(
  opt->rep.max_bytes_for_level_base = n;
 }

+void rocksdb_options_set_snap_refresh_nanos(rocksdb_options_t* opt,
+                                            uint64_t n) {
+  opt->rep.snap_refresh_nanos = n;
+}
+
 void rocksdb_options_set_level_compaction_dynamic_level_bytes(
    rocksdb_options_t* opt, unsigned char v) {
  opt->rep.level_compaction_dynamic_level_bytes = v;
@@ -2462,6 +2519,11 @@ void rocksdb_options_set_enable_pipelined_write(rocksdb_options_t* opt,
  opt->rep.enable_pipelined_write = v;
 }

+void rocksdb_options_set_unordered_write(rocksdb_options_t* opt,
+                                         unsigned char v) {
+  opt->rep.unordered_write = v;
+}
+
 void rocksdb_options_set_max_subcompactions(rocksdb_options_t* opt,
                                            uint32_t n) {
  opt->rep.max_subcompactions = n;
@@ -2866,7 +2928,7 @@ rocksdb_compactionfilter_t* rocksdb_compactionfilter_create(
  result->state_ = state;
  result->destructor_ = destructor;
  result->filter_ = filter;
-  result->ignore_snapshots_ = false;
+  result->ignore_snapshots_ = true;
  result->name_ = name;
  return result;
 }
@@ -2966,7 +3028,7 @@ rocksdb_filterpolicy_t* rocksdb_filterpolicy_create_bloom_format(int bits_per_ke
  // supplied C functions.
  struct Wrapper : public rocksdb_filterpolicy_t {
    const FilterPolicy* rep_;
-    ~Wrapper() { delete rep_; }
+    ~Wrapper() override { delete rep_; }
    const char* Name() const override { return rep_->Name(); }
    void CreateFilter(const Slice* keys, int n,
                      std::string* dst) const override {
@@ -3257,6 +3319,22 @@ void rocksdb_env_join_all_threads(rocksdb_env_t* env) {
  env->rep->WaitForJoin();
 }

+void rocksdb_env_lower_thread_pool_io_priority(rocksdb_env_t* env) {
+  env->rep->LowerThreadPoolIOPriority();
+}
+
+void rocksdb_env_lower_high_priority_thread_pool_io_priority(rocksdb_env_t* env) {
+  env->rep->LowerThreadPoolIOPriority(Env::HIGH);
+}
+
+void rocksdb_env_lower_thread_pool_cpu_priority(rocksdb_env_t* env) {
+  env->rep->LowerThreadPoolCPUPriority();
+}
+
+void rocksdb_env_lower_high_priority_thread_pool_cpu_priority(rocksdb_env_t* env) {
+  env->rep->LowerThreadPoolCPUPriority(Env::HIGH);
+}
+
 void rocksdb_env_destroy(rocksdb_env_t* env) {
  if (!env->is_default) delete env->rep;
  delete env;
@@ -3391,6 +3469,10 @@ void rocksdb_ingest_external_file_cf(
  SaveError(errptr, db->rep->IngestExternalFile(handle->rep, files, opt->rep));
 }

+void rocksdb_try_catch_up_with_primary(rocksdb_t* db, char** errptr) {
+  SaveError(errptr, db->rep->TryCatchUpWithPrimary());
+}
+
 rocksdb_slicetransform_t* rocksdb_slicetransform_create(
    void* state,
    void (*destructor)(void*),
@@ -3421,7 +3503,7 @@ void rocksdb_slicetransform_destroy(rocksdb_slicetransform_t* st) {

 struct Wrapper : public rocksdb_slicetransform_t {
  const SliceTransform* rep_;
-  ~Wrapper() { delete rep_; }
+  ~Wrapper() override { delete rep_; }
  const char* Name() const override { return rep_->Name(); }
  Slice Transform(const Slice& src) const override {
    return rep_->Transform(src);
@@ -3716,6 +3798,38 @@ rocksdb_transactiondb_t* rocksdb_transactiondb_open(
  return result;
 }

+rocksdb_transactiondb_t* rocksdb_transactiondb_open_column_families(
+    const rocksdb_options_t* options,
+    const rocksdb_transactiondb_options_t* txn_db_options, const char* name,
+    int num_column_families, const char** column_family_names,
+    const rocksdb_options_t** column_family_options,
+    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(
+        std::string(column_family_names[i]),
+        ColumnFamilyOptions(column_family_options[i]->rep)));
+  }
+
+  TransactionDB* txn_db;
+  std::vector<ColumnFamilyHandle*> handles;
+  if (SaveError(errptr, TransactionDB::Open(options->rep, txn_db_options->rep,
+                                            std::string(name), column_families,
+                                            &handles, &txn_db))) {
+    return nullptr;
+  }
+
+  for (size_t i = 0; i < handles.size(); i++) {
+    rocksdb_column_family_handle_t* c_handle =
+        new rocksdb_column_family_handle_t;
+    c_handle->rep = handles[i];
+    column_family_handles[i] = c_handle;
+  }
+  rocksdb_transactiondb_t* result = new rocksdb_transactiondb_t;
+  result->rep = txn_db;
+  return result;
+}
+
 const rocksdb_snapshot_t* rocksdb_transactiondb_create_snapshot(
    rocksdb_transactiondb_t* txn_db) {
  rocksdb_snapshot_t* result = new rocksdb_snapshot_t;
@@ -3836,6 +3950,26 @@ char* rocksdb_transaction_get_for_update(rocksdb_transaction_t* txn,
  return result;
 }

+char* rocksdb_transaction_get_for_update_cf(
+    rocksdb_transaction_t* txn, const rocksdb_readoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, const char* key, size_t klen,
+    size_t* vlen, unsigned char exclusive, char** errptr) {
+  char* result = nullptr;
+  std::string tmp;
+  Status s = txn->rep->GetForUpdate(options->rep, column_family->rep,
+                                    Slice(key, klen), &tmp, exclusive);
+  if (s.ok()) {
+    *vlen = tmp.size();
+    result = CopyString(tmp);
+  } else {
+    *vlen = 0;
+    if (!s.IsNotFound()) {
+      SaveError(errptr, s);
+    }
+  }
+  return result;
+}
+
 // Read a key outside a transaction
 char* rocksdb_transactiondb_get(
    rocksdb_transactiondb_t* txn_db,
@@ -3928,6 +4062,14 @@ void rocksdb_transaction_merge(rocksdb_transaction_t* txn, const char* key,
  SaveError(errptr, txn->rep->Merge(Slice(key, klen), Slice(val, vlen)));
 }

+void rocksdb_transaction_merge_cf(rocksdb_transaction_t* txn,
+                                  rocksdb_column_family_handle_t* column_family,
+                                  const char* key, size_t klen, const char* val,
+                                  size_t vlen, char** errptr) {
+  SaveError(errptr, txn->rep->Merge(column_family->rep, Slice(key, klen),
+                                    Slice(val, vlen)));
+}
+
 // Merge a key outside a transaction
 void rocksdb_transactiondb_merge(rocksdb_transactiondb_t* txn_db,
                                 const rocksdb_writeoptions_t* options,
@@ -3937,6 +4079,14 @@ void rocksdb_transactiondb_merge(rocksdb_transactiondb_t* txn_db,
                                       Slice(val, vlen)));
 }

+void rocksdb_transactiondb_merge_cf(
+    rocksdb_transactiondb_t* txn_db, const rocksdb_writeoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, const char* key, size_t klen,
+    const char* val, size_t vlen, char** errptr) {
+  SaveError(errptr, txn_db->rep->Merge(options->rep, column_family->rep,
+                                       Slice(key, klen), Slice(val, vlen)));
+}
+
 // Delete a key inside a transaction
 void rocksdb_transaction_delete(rocksdb_transaction_t* txn, const char* key,
                                size_t klen, char** errptr) {
@@ -3989,6 +4139,14 @@ rocksdb_iterator_t* rocksdb_transactiondb_create_iterator(
  return result;
 }

+rocksdb_iterator_t* rocksdb_transactiondb_create_iterator_cf(
+    rocksdb_transactiondb_t* txn_db, const rocksdb_readoptions_t* options,
+    rocksdb_column_family_handle_t* column_family) {
+  rocksdb_iterator_t* result = new rocksdb_iterator_t;
+  result->rep = txn_db->rep->NewIterator(options->rep, column_family->rep);
+  return result;
+}
+
 void rocksdb_transactiondb_close(rocksdb_transactiondb_t* txn_db) {
  delete txn_db->rep;
  delete txn_db;
@@ -1,6 +1,7 @@
 /* Copyright (c) 2011 The LevelDB Authors. All rights reserved.
   Use of this source code is governed by a BSD-style license that can be
   found in the LICENSE file. See the AUTHORS file for names of contributors. */
+// Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved.

 #include <stdio.h>

@@ -44,6 +45,7 @@ static char sstfilename[200];
 static char dbbackupname[200];
 static char dbcheckpointname[200];
 static char dbpathname[200];
+static char secondary_path[200];

 static void StartPhase(const char* name) {
  fprintf(stderr, "=== Test %s\n", name);
@@ -1187,6 +1189,12 @@ int main(int argc, char** argv) {
    rocksdb_put_cf(db, woptions, handles[1], "foo", 3, "hello", 5, &err);
    CheckNoError(err);

+    rocksdb_flushoptions_t *flush_options = rocksdb_flushoptions_create();
+    rocksdb_flushoptions_set_wait(flush_options, 1);
+    rocksdb_flush_cf(db, flush_options, handles[1], &err);
+    CheckNoError(err)
+    rocksdb_flushoptions_destroy(flush_options);
+
    CheckGetCF(db, roptions, handles[1], "foo", "hello");
    CheckPinGetCF(db, roptions, handles[1], "foo", "hello");

@@ -1715,6 +1723,59 @@ int main(int argc, char** argv) {
    CheckNoError(err);
  }

+  // Check that secondary instance works.
+  StartPhase("open_as_secondary");
+  {
+    rocksdb_close(db);
+    rocksdb_destroy_db(options, dbname, &err);
+
+    rocksdb_options_t* db_options = rocksdb_options_create();
+    rocksdb_options_set_create_if_missing(db_options, 1);
+    db = rocksdb_open(db_options, dbname, &err);
+    CheckNoError(err);
+    rocksdb_t* db1;
+    rocksdb_options_t* opts = rocksdb_options_create();
+    rocksdb_options_set_max_open_files(opts, -1);
+    rocksdb_options_set_create_if_missing(opts, 1);
+    snprintf(secondary_path, sizeof(secondary_path),
+             "%s/rocksdb_c_test_secondary-%d", GetTempDir(), ((int)geteuid()));
+    db1 = rocksdb_open_as_secondary(opts, dbname, secondary_path, &err);
+    CheckNoError(err);
+
+    rocksdb_writeoptions_set_sync(woptions, 0);
+    rocksdb_writeoptions_disable_WAL(woptions, 1);
+    rocksdb_put(db, woptions, "key0", 4, "value0", 6, &err);
+    CheckNoError(err);
+    rocksdb_flushoptions_t* flush_opts = rocksdb_flushoptions_create();
+    rocksdb_flushoptions_set_wait(flush_opts, 1);
+    rocksdb_flush(db, flush_opts, &err);
+    CheckNoError(err);
+    rocksdb_try_catch_up_with_primary(db1, &err);
+    CheckNoError(err);
+    rocksdb_readoptions_t* ropts = rocksdb_readoptions_create();
+    rocksdb_readoptions_set_verify_checksums(ropts, 1);
+    rocksdb_readoptions_set_snapshot(ropts, NULL);
+    CheckGet(db, ropts, "key0", "value0");
+    CheckGet(db1, ropts, "key0", "value0");
+
+    rocksdb_writeoptions_disable_WAL(woptions, 0);
+    rocksdb_put(db, woptions, "key1", 4, "value1", 6, &err);
+    CheckNoError(err);
+    rocksdb_try_catch_up_with_primary(db1, &err);
+    CheckNoError(err);
+    CheckGet(db1, ropts, "key0", "value0");
+    CheckGet(db1, ropts, "key1", "value1");
+
+    rocksdb_close(db1);
+    rocksdb_destroy_db(opts, secondary_path, &err);
+    CheckNoError(err);
+
+    rocksdb_options_destroy(db_options);
+    rocksdb_options_destroy(opts);
+    rocksdb_readoptions_destroy(ropts);
+    rocksdb_flushoptions_destroy(flush_opts);
+  }
+
  // Simple sanity check that options setting db_paths work.
  StartPhase("open_db_paths");
  {
@@ -9,34 +9,31 @@

 #include "db/column_family.h"

-#ifndef __STDC_FORMAT_MACROS
-#define __STDC_FORMAT_MACROS
-#endif
-
-#include <inttypes.h>
+#include <cinttypes>
 #include <vector>
 #include <string>
 #include <algorithm>
 #include <limits>

-#include "db/compaction_picker.h"
-#include "db/compaction_picker_fifo.h"
-#include "db/compaction_picker_universal.h"
-#include "db/db_impl.h"
+#include "db/compaction/compaction_picker.h"
+#include "db/compaction/compaction_picker_fifo.h"
+#include "db/compaction/compaction_picker_level.h"
+#include "db/compaction/compaction_picker_universal.h"
+#include "db/db_impl/db_impl.h"
 #include "db/internal_stats.h"
 #include "db/job_context.h"
 #include "db/range_del_aggregator.h"
 #include "db/table_properties_collector.h"
 #include "db/version_set.h"
 #include "db/write_controller.h"
+#include "file/sst_file_manager_impl.h"
 #include "memtable/hash_skiplist_rep.h"
 #include "monitoring/thread_status_util.h"
 #include "options/options_helper.h"
-#include "table/block_based_table_factory.h"
+#include "table/block_based/block_based_table_factory.h"
 #include "table/merging_iterator.h"
 #include "util/autovector.h"
 #include "util/compression.h"
-#include "util/sst_file_manager_impl.h"

 namespace rocksdb {

@@ -63,12 +60,24 @@ ColumnFamilyHandleImpl::~ColumnFamilyHandleImpl() {
    JobContext job_context(0);
    mutex_->Lock();
    if (cfd_->Unref()) {
+      bool dropped = cfd_->IsDropped();
+
      delete cfd_;
+
+      if (dropped) {
+        db_->FindObsoleteFiles(&job_context, false, true);
+      }
    }
-    db_->FindObsoleteFiles(&job_context, false, true);
    mutex_->Unlock();
    if (job_context.HaveSomethingToDelete()) {
-      db_->PurgeObsoleteFiles(job_context);
+      bool defer_purge =
+          db_->immutable_db_options().avoid_unnecessary_blocking_io;
+      db_->PurgeObsoleteFiles(job_context, defer_purge);
+      if (defer_purge) {
+        mutex_->Lock();
+        db_->SchedulePurge();
+        mutex_->Unlock();
+      }
    }
    job_context.Clean();
  }
@@ -396,7 +405,8 @@ ColumnFamilyData::ColumnFamilyData(
    uint32_t id, const std::string& name, Version* _dummy_versions,
    Cache* _table_cache, WriteBufferManager* write_buffer_manager,
    const ColumnFamilyOptions& cf_options, const ImmutableDBOptions& db_options,
-    const EnvOptions& env_options, ColumnFamilySet* column_family_set)
+    const EnvOptions& env_options, ColumnFamilySet* column_family_set,
+    BlockCacheTracer* const block_cache_tracer)
    : id_(id),
      name_(name),
      dummy_versions_(_dummy_versions),
@@ -436,7 +446,8 @@ ColumnFamilyData::ColumnFamilyData(
  if (_dummy_versions != nullptr) {
    internal_stats_.reset(
        new InternalStats(ioptions_.num_levels, db_options.env, this));
-    table_cache_.reset(new TableCache(ioptions_, env_options, _table_cache));
+    table_cache_.reset(new TableCache(ioptions_, env_options, _table_cache,
+                                      block_cache_tracer));
    if (ioptions_.compaction_style == kCompactionStyleLevel) {
      compaction_picker_.reset(
          new LevelCompactionPicker(ioptions_, &internal_comparator_));
@@ -982,13 +993,14 @@ const int ColumnFamilyData::kCompactToBaseLevel = -2;

 Compaction* ColumnFamilyData::CompactRange(
    const MutableCFOptions& mutable_cf_options, int input_level,
-    int output_level, uint32_t output_path_id, uint32_t max_subcompactions,
+    int output_level, const CompactRangeOptions& compact_range_options,
    const InternalKey* begin, const InternalKey* end,
-    InternalKey** compaction_end, bool* conflict) {
+    InternalKey** compaction_end, bool* conflict,
+    uint64_t max_file_num_to_ignore) {
  auto* result = compaction_picker_->CompactRange(
      GetName(), mutable_cf_options, current_->storage_info(), input_level,
-      output_level, output_path_id, max_subcompactions, begin, end,
-      compaction_end, conflict);
+      output_level, compact_range_options, begin, end, compaction_end, conflict,
+      max_file_num_to_ignore);
  if (result != nullptr) {
    result->SetInputVersion(current_);
  }
@@ -1134,13 +1146,60 @@ void ColumnFamilyData::ResetThreadLocalSuperVersions() {
  }
 }

+Status ColumnFamilyData::ValidateOptions(
+    const DBOptions& db_options, const ColumnFamilyOptions& cf_options) {
+  Status s;
+  s = CheckCompressionSupported(cf_options);
+  if (s.ok() && db_options.allow_concurrent_memtable_write) {
+    s = CheckConcurrentWritesSupported(cf_options);
+  }
+  if (s.ok()) {
+    s = CheckCFPathsSupported(db_options, cf_options);
+  }
+  if (!s.ok()) {
+    return s;
+  }
+
+  if (cf_options.ttl > 0) {
+    if (db_options.max_open_files != -1) {
+      return Status::NotSupported(
+          "TTL is only supported when files are always "
+          "kept open (set max_open_files = -1). ");
+    }
+    if (cf_options.table_factory->Name() != BlockBasedTableFactory().Name()) {
+      return Status::NotSupported(
+          "TTL is only supported in Block-Based Table format. ");
+    }
+  }
+
+  if (cf_options.periodic_compaction_seconds > 0) {
+    if (db_options.max_open_files != -1) {
+      return Status::NotSupported(
+          "Periodic Compaction is only supported when files are always "
+          "kept open (set max_open_files = -1). ");
+    }
+    if (cf_options.table_factory->Name() != BlockBasedTableFactory().Name()) {
+      return Status::NotSupported(
+          "Periodic Compaction is only supported in "
+          "Block-Based Table format. ");
+    }
+  }
+  return s;
+}
+
 #ifndef ROCKSDB_LITE
 Status ColumnFamilyData::SetOptions(
-      const std::unordered_map<std::string, std::string>& options_map) {
+    const DBOptions& db_options,
+    const std::unordered_map<std::string, std::string>& options_map) {
  MutableCFOptions new_mutable_cf_options;
  Status s =
      GetMutableOptionsFromStrings(mutable_cf_options_, options_map,
                                   ioptions_.info_log, &new_mutable_cf_options);
+  if (s.ok()) {
+    ColumnFamilyOptions cf_options =
+        BuildColumnFamilyOptions(initial_cf_options_, new_mutable_cf_options);
+    s = ValidateOptions(db_options, cf_options);
+  }
  if (s.ok()) {
    mutable_cf_options_ = new_mutable_cf_options;
    mutable_cf_options_.RefreshDerivedOptions(ioptions_);
@@ -1197,18 +1256,20 @@ ColumnFamilySet::ColumnFamilySet(const std::string& dbname,
                                 const EnvOptions& env_options,
                                 Cache* table_cache,
                                 WriteBufferManager* write_buffer_manager,
-                                 WriteController* write_controller)
+                                 WriteController* write_controller,
+                                 BlockCacheTracer* const block_cache_tracer)
    : max_column_family_(0),
-      dummy_cfd_(new ColumnFamilyData(0, "", nullptr, nullptr, nullptr,
-                                      ColumnFamilyOptions(), *db_options,
-                                      env_options, nullptr)),
+      dummy_cfd_(new ColumnFamilyData(
+          0, "", nullptr, nullptr, nullptr, ColumnFamilyOptions(), *db_options,
+          env_options, nullptr, block_cache_tracer)),
      default_cfd_cache_(nullptr),
      db_name_(dbname),
      db_options_(db_options),
      env_options_(env_options),
      table_cache_(table_cache),
      write_buffer_manager_(write_buffer_manager),
-      write_controller_(write_controller) {
+      write_controller_(write_controller),
+      block_cache_tracer_(block_cache_tracer) {
  // initialize linked list
  dummy_cfd_->prev_ = dummy_cfd_;
  dummy_cfd_->next_ = dummy_cfd_;
@@ -1276,7 +1337,7 @@ ColumnFamilyData* ColumnFamilySet::CreateColumnFamily(
  assert(column_families_.find(name) == column_families_.end());
  ColumnFamilyData* new_cfd = new ColumnFamilyData(
      id, name, dummy_versions, table_cache_, write_buffer_manager_, options,
-      *db_options_, env_options_, this);
+      *db_options_, env_options_, this, block_cache_tracer_);
  column_families_.insert({name, id});
  column_family_data_.insert({id, new_cfd});
  max_column_family_ = std::max(max_column_family_, id);
@@ -24,6 +24,7 @@
 #include "rocksdb/db.h"
 #include "rocksdb/env.h"
 #include "rocksdb/options.h"
+#include "trace_replay/block_cache_tracer.h"
 #include "util/thread_local.h"

 namespace rocksdb {
@@ -45,6 +46,112 @@ class InstrumentedMutexLock;
 struct SuperVersionContext;

 extern const double kIncSlowdownRatio;
+// This file contains a list of data structures for managing column family
+// level metadata.
+//
+// The basic relationships among classes declared here are illustrated as
+// following:
+//
+//       +----------------------+    +----------------------+   +--------+
+//   +---+ ColumnFamilyHandle 1 | +--+ ColumnFamilyHandle 2 |   | DBImpl |
+//   |   +----------------------+ |  +----------------------+   +----+---+
+//   | +--------------------------+                                  |
+//   | |                               +-----------------------------+
+//   | |                               |
+//   | | +-----------------------------v-------------------------------+
+//   | | |                                                             |
+//   | | |                      ColumnFamilySet                        |
+//   | | |                                                             |
+//   | | +-------------+--------------------------+----------------+---+
+//   | |               |                          |                |
+//   | +-------------------------------------+    |                |
+//   |                 |                     |    |                v
+//   |   +-------------v-------------+ +-----v----v---------+
+//   |   |                           | |                    |
+//   |   |     ColumnFamilyData 1    | | ColumnFamilyData 2 |    ......
+//   |   |                           | |                    |
+//   +--->                           | |                    |
+//       |                 +---------+ |                    |
+//       |                 | MemTable| |                    |
+//       |                 |  List   | |                    |
+//       +--------+---+--+-+----+----+ +--------------------++
+//                |   |  |      |
+//                |   |  |      |
+//                |   |  |      +-----------------------+
+//                |   |  +-----------+                  |
+//                v   +--------+     |                  |
+//       +--------+--------+   |     |                  |
+//       |                 |   |     |       +----------v----------+
+// +---> |SuperVersion 1.a +----------------->                     |
+//       |                 +------+  |       | MemTableListVersion |
+//       +---+-------------+   |  |  |       |                     |
+//           |                 |  |  |       +----+------------+---+
+//           |      current    |  |  |            |            |
+//           |   +-------------+  |  |mem         |            |
+//           |   |                |  |            |            |
+//         +-v---v-------+    +---v--v---+  +-----v----+  +----v-----+
+//         |             |    |          |  |          |  |          |
+//         | Version 1.a |    | memtable |  | memtable |  | memtable |
+//         |             |    |   1.a    |  |   1.b    |  |   1.c    |
+//         +-------------+    |          |  |          |  |          |
+//                            +----------+  +----------+  +----------+
+//
+// DBImpl keeps a ColumnFamilySet, which references to all column families by
+// pointing to respective ColumnFamilyData object of each column family.
+// This is how DBImpl can list and operate on all the column families.
+// ColumnFamilyHandle also points to ColumnFamilyData directly, so that
+// when a user executes a query, it can directly find memtables and Version
+// as well as SuperVersion to the column family, without going through
+// ColumnFamilySet.
+//
+// ColumnFamilySet points to the latest view of the LSM-tree (list of memtables
+// and SST files) indirectly, while ongoing operations may hold references
+// to a current or an out-of-date SuperVersion, which in turn points to a
+// point-in-time view of the LSM-tree. This guarantees the memtables and SST
+// files being operated on will not go away, until the SuperVersion is
+// unreferenced to 0 and destoryed.
+//
+// The following graph illustrates a possible referencing relationships:
+//
+// Column       +--------------+      current       +-----------+
+// Family +---->+              +------------------->+           |
+//  Data        | SuperVersion +----------+         | Version A |
+//              |      3       |   imm    |         |           |
+// Iter2 +----->+              |  +-------v------+  +-----------+
+//              +-----+--------+  | MemtableList +----------------> Empty
+//                    |           |   Version r  |  +-----------+
+//                    |           +--------------+  |           |
+//                    +------------------+   current| Version B |
+//              +--------------+         |   +----->+           |
+//              |              |         |   |      +-----+-----+
+// Compaction +>+ SuperVersion +-------------+            ^
+//    Job       |      2       +------+  |                |current
+//              |              +----+ |  |     mem        |    +------------+
+//              +--------------+    | |  +--------------------->            |
+//                                  | +------------------------> MemTable a |
+//                                  |          mem        |    |            |
+//              +--------------+    |                     |    +------------+
+//              |              +--------------------------+
+//  Iter1 +-----> SuperVersion |    |                          +------------+
+//              |      1       +------------------------------>+            |
+//              |              +-+  |        mem               | MemTable b |
+//              +--------------+ |  |                          |            |
+//                               |  |    +--------------+      +-----^------+
+//                               |  |imm | MemtableList |            |
+//                               |  +--->+   Version s  +------------+
+//                               |       +--------------+
+//                               |       +--------------+
+//                               |       | MemtableList |
+//                               +------>+   Version t  +-------->  Empty
+//                                 imm   +--------------+
+//
+// In this example, even if the current LSM-tree consists of Version A and
+// memtable a, which is also referenced by SuperVersion, two older SuperVersion
+// SuperVersion2 and Superversion1 still exist, and are referenced by a
+// compaction job and an old iterator Iter1, respectively. SuperVersion2
+// contains Version B, memtable a and memtable b; SuperVersion1 contains
+// Version B and memtable b (mutable). As a result, Version B and memtable b
+// are prevented from being destroyed or deleted.

 // ColumnFamilyHandleImpl is the class that clients use to access different
 // column families. It has non-trivial destructor, which gets called when client
@@ -232,9 +339,13 @@ class ColumnFamilyData {

  bool is_delete_range_supported() { return is_delete_range_supported_; }

+  // Validate CF options against DB options
+  static Status ValidateOptions(const DBOptions& db_options,
+                                const ColumnFamilyOptions& cf_options);
 #ifndef ROCKSDB_LITE
  // REQUIRES: DB mutex held
  Status SetOptions(
+      const DBOptions& db_options,
      const std::unordered_map<std::string, std::string>& options_map);
 #endif  // ROCKSDB_LITE

@@ -296,9 +407,10 @@ class ColumnFamilyData {
  // REQUIRES: DB mutex held
  Compaction* CompactRange(const MutableCFOptions& mutable_cf_options,
                           int input_level, int output_level,
-                           uint32_t output_path_id, uint32_t max_subcompactions,
+                           const CompactRangeOptions& compact_range_options,
                           const InternalKey* begin, const InternalKey* end,
-                           InternalKey** compaction_end, bool* manual_conflict);
+                           InternalKey** compaction_end, bool* manual_conflict,
+                           uint64_t max_file_num_to_ignore);

  CompactionPicker* compaction_picker() { return compaction_picker_.get(); }
  // thread-safe
@@ -393,7 +505,8 @@ class ColumnFamilyData {
                   const ColumnFamilyOptions& options,
                   const ImmutableDBOptions& db_options,
                   const EnvOptions& env_options,
-                   ColumnFamilySet* column_family_set);
+                   ColumnFamilySet* column_family_set,
+                   BlockCacheTracer* const block_cache_tracer);

  uint32_t id_;
  const std::string name_;
@@ -521,7 +634,8 @@ class ColumnFamilySet {
                  const ImmutableDBOptions* db_options,
                  const EnvOptions& env_options, Cache* table_cache,
                  WriteBufferManager* write_buffer_manager,
-                  WriteController* write_controller);
+                  WriteController* write_controller,
+                  BlockCacheTracer* const block_cache_tracer);
  ~ColumnFamilySet();

  ColumnFamilyData* GetDefault() const;
@@ -580,6 +694,7 @@ class ColumnFamilySet {
  Cache* table_cache_;
  WriteBufferManager* write_buffer_manager_;
  WriteController* write_controller_;
+  BlockCacheTracer* const block_cache_tracer_;
 };

 // We use ColumnFamilyMemTablesImpl to provide WriteBatch a way to access
@@ -12,19 +12,20 @@
 #include <string>
 #include <thread>

-#include "db/db_impl.h"
+#include "db/db_impl/db_impl.h"
 #include "db/db_test_util.h"
+#include "memtable/hash_skiplist_rep.h"
 #include "options/options_parser.h"
 #include "port/port.h"
 #include "rocksdb/db.h"
 #include "rocksdb/env.h"
 #include "rocksdb/iterator.h"
+#include "test_util/fault_injection_test_env.h"
+#include "test_util/sync_point.h"
+#include "test_util/testharness.h"
+#include "test_util/testutil.h"
 #include "util/coding.h"
-#include "util/fault_injection_test_env.h"
 #include "util/string_util.h"
-#include "util/sync_point.h"
-#include "util/testharness.h"
-#include "util/testutil.h"
 #include "utilities/merge_operators.h"

 namespace rocksdb {
@@ -68,7 +69,7 @@ class ColumnFamilyTestBase : public testing::Test {
    DestroyDB(dbname_, Options(db_options_, column_family_options_));
  }

-  virtual ~ColumnFamilyTestBase() {
+  ~ColumnFamilyTestBase() override {
    std::vector<ColumnFamilyDescriptor> column_families;
    for (auto h : handles_) {
      ColumnFamilyDescriptor cfdescriptor;
@@ -1206,29 +1207,32 @@ TEST_P(ColumnFamilyTest, DifferentWriteBufferSizes) {
 }
 #endif  // !ROCKSDB_LITE

-#ifndef ROCKSDB_LITE  // Cuckoo is not supported in lite
-TEST_P(ColumnFamilyTest, MemtableNotSupportSnapshot) {
-  db_options_.allow_concurrent_memtable_write = false;
-  Open();
-  auto* s1 = dbfull()->GetSnapshot();
-  ASSERT_TRUE(s1 != nullptr);
-  dbfull()->ReleaseSnapshot(s1);
+// The test is commented out because we want to test that snapshot is
+// not created for memtables not supported it, but There isn't a memtable
+// that doesn't support snapshot right now. If we have one later, we can
+// re-enable the test.
+//
+// #ifndef ROCKSDB_LITE  // Cuckoo is not supported in lite
+//   TEST_P(ColumnFamilyTest, MemtableNotSupportSnapshot) {
+//   db_options_.allow_concurrent_memtable_write = false;
+//   Open();
+//   auto* s1 = dbfull()->GetSnapshot();
+//   ASSERT_TRUE(s1 != nullptr);
+//   dbfull()->ReleaseSnapshot(s1);

-  // Add a column family that doesn't support snapshot
-  ColumnFamilyOptions first;
-  first.memtable_factory.reset(NewHashCuckooRepFactory(1024 * 1024));
-  CreateColumnFamilies({"first"}, {first});
-  auto* s2 = dbfull()->GetSnapshot();
-  ASSERT_TRUE(s2 == nullptr);
+//   // Add a column family that doesn't support snapshot
+//   ColumnFamilyOptions first;
+//   first.memtable_factory.reset(new DummyMemtableNotSupportingSnapshot());
+//   CreateColumnFamilies({"first"}, {first});
+//   auto* s2 = dbfull()->GetSnapshot();
+//   ASSERT_TRUE(s2 == nullptr);

-  // Add a column family that supports snapshot. Snapshot stays not supported.
-  ColumnFamilyOptions second;
-  CreateColumnFamilies({"second"}, {second});
-  auto* s3 = dbfull()->GetSnapshot();
-  ASSERT_TRUE(s3 == nullptr);
-  Close();
-}
-#endif  // !ROCKSDB_LITE
+//   // Add a column family that supports snapshot. Snapshot stays not
+//   supported. ColumnFamilyOptions second; CreateColumnFamilies({"second"},
+//   {second}); auto* s3 = dbfull()->GetSnapshot(); ASSERT_TRUE(s3 == nullptr);
+//   Close();
+// }
+// #endif  // !ROCKSDB_LITE

 class TestComparator : public Comparator {
  int Compare(const rocksdb::Slice& /*a*/,
@@ -10,13 +10,13 @@
 #include <thread>
 #include <vector>

-#include "db/db_impl.h"
+#include "db/db_impl/db_impl.h"
 #include "port/port.h"
 #include "rocksdb/db.h"
 #include "rocksdb/env.h"
+#include "test_util/sync_point.h"
+#include "test_util/testharness.h"
 #include "util/string_util.h"
-#include "util/sync_point.h"
-#include "util/testharness.h"

 namespace rocksdb {

@@ -35,9 +35,9 @@ class CompactFilesTest : public testing::Test {
 class FlushedFileCollector : public EventListener {
 public:
  FlushedFileCollector() {}
-  ~FlushedFileCollector() {}
+  ~FlushedFileCollector() override {}

-  virtual void OnFlushCompleted(DB* /*db*/, const FlushJobInfo& info) override {
+  void OnFlushCompleted(DB* /*db*/, const FlushJobInfo& info) override {
    std::lock_guard<std::mutex> lock(mutex_);
    flushed_files_.push_back(info.file_path);
  }
@@ -256,9 +256,9 @@ TEST_F(CompactFilesTest, CapturingPendingFiles) {
 TEST_F(CompactFilesTest, CompactionFilterWithGetSv) {
  class FilterWithGet : public CompactionFilter {
   public:
-    virtual bool Filter(int /*level*/, const Slice& /*key*/,
-                        const Slice& /*value*/, std::string* /*new_value*/,
-                        bool* /*value_changed*/) const override {
+    bool Filter(int /*level*/, const Slice& /*key*/, const Slice& /*value*/,
+                std::string* /*new_value*/,
+                bool* /*value_changed*/) const override {
      if (db_ == nullptr) {
        return true;
      }
@@ -271,7 +271,7 @@ TEST_F(CompactFilesTest, CompactionFilterWithGetSv) {
      db_ = db;
    }

-    virtual const char* Name() const override { return "FilterWithGet"; }
+    const char* Name() const override { return "FilterWithGet"; }

   private:
    DB* db_;
@@ -5,7 +5,7 @@

 #ifndef ROCKSDB_LITE
 #include "db/compacted_db_impl.h"
-#include "db/db_impl.h"
+#include "db/db_impl/db_impl.h"
 #include "db/version_set.h"
 #include "table/get_context.h"

@@ -5,9 +5,9 @@

 #pragma once
 #ifndef ROCKSDB_LITE
-#include "db/db_impl.h"
-#include <vector>
 #include <string>
+#include <vector>
+#include "db/db_impl/db_impl.h"

 namespace rocksdb {

@@ -85,6 +85,15 @@ class CompactedDBImpl : public DBImpl {
      const IngestExternalFileOptions& /*ingestion_options*/) override {
    return Status::NotSupported("Not supported in compacted db mode.");
  }
+  using DB::CreateColumnFamilyWithImport;
+  virtual Status CreateColumnFamilyWithImport(
+      const ColumnFamilyOptions& /*options*/,
+      const std::string& /*column_family_name*/,
+      const ImportColumnFamilyOptions& /*import_options*/,
+      const ExportImportFilesMetaData& /*metadata*/,
+      ColumnFamilyHandle** /*handle*/) override {
+    return Status::NotSupported("Not supported in compacted db mode.");
+  }

 private:
  friend class DB;
@@ -7,19 +7,14 @@
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file. See the AUTHORS file for names of contributors.

-#include "db/compaction.h"
-
-#ifndef __STDC_FORMAT_MACROS
-#define __STDC_FORMAT_MACROS
-#endif
-
-#include <inttypes.h>
+#include <cinttypes>
 #include <vector>

 #include "db/column_family.h"
+#include "db/compaction/compaction.h"
 #include "rocksdb/compaction_filter.h"
+#include "test_util/sync_point.h"
 #include "util/string_util.h"
-#include "util/sync_point.h"

 namespace rocksdb {

@@ -250,6 +245,12 @@ Compaction::Compaction(VersionStorageInfo* vstorage,
  if (max_subcompactions_ == 0) {
    max_subcompactions_ = immutable_cf_options_.max_subcompactions;
  }
+  if (!bottommost_level_) {
+    // Currently we only enable dictionary compression during compaction to the
+    // bottommost level.
+    output_compression_opts_.max_dict_bytes = 0;
+    output_compression_opts_.zstd_max_train_bytes = 0;
+  }

 #ifndef NDEBUG
  for (size_t i = 1; i < inputs_.size(); ++i) {
@@ -9,11 +9,13 @@

 #pragma once
 #include "db/version_set.h"
+#include "memory/arena.h"
 #include "options/cf_options.h"
-#include "util/arena.h"
 #include "util/autovector.h"

 namespace rocksdb {
+// The file contains class Compaction, as well as some helper functions
+// and data structures used by the class.

 // Utility for comparing sstable boundary keys. Returns -1 if either a or b is
 // null which provides the property that a==null indicates a key that is less
@@ -63,7 +65,7 @@ class ColumnFamilyData;
 class VersionStorageInfo;
 class CompactionFilter;

-// A Compaction encapsulates information about a compaction.
+// A Compaction encapsulates metadata about a compaction.
 class Compaction {
 public:
  Compaction(VersionStorageInfo* input_version,
@@ -376,7 +378,7 @@ class Compaction {
  CompactionReason compaction_reason_;
 };

-// Utility function
+// Return sum of sizes of all files in `files`.
 extern uint64_t TotalFileSize(const std::vector<FileMetaData*>& files);

 }  // namespace rocksdb
@@ -3,13 +3,12 @@
 //  COPYING file in the root directory) and Apache 2.0 License
 //  (found in the LICENSE.Apache file in the root directory).

-#include "db/compaction_iterator.h"
-
+#include "db/compaction/compaction_iterator.h"
 #include "db/snapshot_checker.h"
 #include "port/likely.h"
 #include "rocksdb/listener.h"
 #include "table/internal_iterator.h"
-#include "util/sync_point.h"
+#include "test_util/sync_point.h"

 #define DEFINITELY_IN_SNAPSHOT(seq, snapshot)                       \
  ((seq) <= (snapshot) &&                                           \
@@ -38,14 +37,16 @@ CompactionIterator::CompactionIterator(
    CompactionRangeDelAggregator* range_del_agg, const Compaction* compaction,
    const CompactionFilter* compaction_filter,
    const std::atomic<bool>* shutting_down,
-    const SequenceNumber preserve_deletes_seqnum)
+    const SequenceNumber preserve_deletes_seqnum,
+    SnapshotListFetchCallback* snap_list_callback)
    : CompactionIterator(
          input, cmp, merge_helper, last_sequence, snapshots,
          earliest_write_conflict_snapshot, snapshot_checker, env,
          report_detailed_time, expect_valid_internal_key, range_del_agg,
          std::unique_ptr<CompactionProxy>(
              compaction ? new CompactionProxy(compaction) : nullptr),
-          compaction_filter, shutting_down, preserve_deletes_seqnum) {}
+          compaction_filter, shutting_down, preserve_deletes_seqnum,
+          snap_list_callback) {}

 CompactionIterator::CompactionIterator(
    InternalIterator* input, const Comparator* cmp, MergeHelper* merge_helper,
@@ -57,7 +58,8 @@ CompactionIterator::CompactionIterator(
    std::unique_ptr<CompactionProxy> compaction,
    const CompactionFilter* compaction_filter,
    const std::atomic<bool>* shutting_down,
-    const SequenceNumber preserve_deletes_seqnum)
+    const SequenceNumber preserve_deletes_seqnum,
+    SnapshotListFetchCallback* snap_list_callback)
    : input_(input),
      cmp_(cmp),
      merge_helper_(merge_helper),
@@ -72,11 +74,11 @@ CompactionIterator::CompactionIterator(
      compaction_filter_(compaction_filter),
      shutting_down_(shutting_down),
      preserve_deletes_seqnum_(preserve_deletes_seqnum),
-      ignore_snapshots_(false),
      current_user_key_sequence_(0),
      current_user_key_snapshot_(0),
      merge_out_iter_(merge_helper_),
-      current_key_committed_(false) {
+      current_key_committed_(false),
+      snap_list_callback_(snap_list_callback) {
  assert(compaction_filter_ == nullptr || compaction_ != nullptr);
  assert(snapshots_ != nullptr);
  bottommost_level_ =
@@ -84,31 +86,7 @@ CompactionIterator::CompactionIterator(
  if (compaction_ != nullptr) {
    level_ptrs_ = std::vector<size_t>(compaction_->number_levels(), 0);
  }
-  if (snapshots_->size() == 0) {
-    // optimize for fast path if there are no snapshots
-    visible_at_tip_ = true;
-    earliest_snapshot_iter_ = snapshots_->end();
-    earliest_snapshot_ = kMaxSequenceNumber;
-    latest_snapshot_ = 0;
-  } else {
-    visible_at_tip_ = false;
-    earliest_snapshot_iter_ = snapshots_->begin();
-    earliest_snapshot_ = snapshots_->at(0);
-    latest_snapshot_ = snapshots_->back();
-  }
-#ifndef NDEBUG
-  // findEarliestVisibleSnapshot assumes this ordering.
-  for (size_t i = 1; i < snapshots_->size(); ++i) {
-    assert(snapshots_->at(i - 1) < snapshots_->at(i));
-  }
-#endif
-  if (compaction_filter_ != nullptr) {
-    if (compaction_filter_->IgnoreSnapshots()) {
-      ignore_snapshots_ = true;
-    }
-  } else {
-    ignore_snapshots_ = false;
-  }
+  ProcessSnapshotList();
  input_->SetPinnedItersMgr(&pinned_iters_mgr_);
  TEST_SYNC_POINT_CALLBACK("CompactionIterator:AfterInit", compaction_.get());
 }
@@ -180,9 +158,7 @@ void CompactionIterator::Next() {
 void CompactionIterator::InvokeFilterIfNeeded(bool* need_skip,
                                              Slice* skip_until) {
  if (compaction_filter_ != nullptr &&
-      (ikey_.type == kTypeValue || ikey_.type == kTypeBlobIndex) &&
-      (visible_at_tip_ || ignore_snapshots_ ||
-       DEFINITELY_NOT_IN_SNAPSHOT(ikey_.sequence, latest_snapshot_))) {
+      (ikey_.type == kTypeValue || ikey_.type == kTypeBlobIndex)) {
    // If the user has specified a compaction filter and the sequence
    // number is greater than any external snapshot, then invoke the
    // filter. If the return value of the compaction filter is true,
@@ -232,6 +208,28 @@ void CompactionIterator::InvokeFilterIfNeeded(bool* need_skip,
  }
 }

+void CompactionIterator::ProcessSnapshotList() {
+#ifndef NDEBUG
+  // findEarliestVisibleSnapshot assumes this ordering.
+  for (size_t i = 1; i < snapshots_->size(); ++i) {
+    assert(snapshots_->at(i - 1) < snapshots_->at(i));
+  }
+#endif
+  if (snapshots_->size() == 0) {
+    // optimize for fast path if there are no snapshots
+    visible_at_tip_ = true;
+    earliest_snapshot_iter_ = snapshots_->end();
+    earliest_snapshot_ = kMaxSequenceNumber;
+    latest_snapshot_ = 0;
+  } else {
+    visible_at_tip_ = false;
+    earliest_snapshot_iter_ = snapshots_->begin();
+    earliest_snapshot_ = snapshots_->at(0);
+    latest_snapshot_ = snapshots_->back();
+  }
+  released_snapshots_.clear();
+}
+
 void CompactionIterator::NextFromInput() {
  at_next_ = false;
  valid_ = false;
@@ -279,6 +277,13 @@ void CompactionIterator::NextFromInput() {
    // compaction filter). ikey_.user_key is pointing to the copy.
    if (!has_current_user_key_ ||
        !cmp_->Equal(ikey_.user_key, current_user_key_)) {
+      num_keys_++;
+      // Use num_keys_ to reduce the overhead of reading current time
+      if (snap_list_callback_ && snapshots_->size() &&
+          snap_list_callback_->TimeToRefresh(num_keys_)) {
+        snap_list_callback_->Refresh(snapshots_, latest_snapshot_);
+        ProcessSnapshotList();
+      }
      // First occurrence of this user key
      // Copy key for output
      key_ = current_key_.SetInternalKey(key_, &ikey_);
@@ -487,17 +492,6 @@ void CompactionIterator::NextFromInput() {
      // in this snapshot.
      assert(last_sequence >= current_user_key_sequence_);

-      // Note2: if last_snapshot < current_user_key_snapshot, it can only
-      // mean last_snapshot is released between we process last value and
-      // this value, and findEarliestVisibleSnapshot returns the next snapshot
-      // as current_user_key_snapshot. In this case last value and current
-      // value are both in current_user_key_snapshot currently.
-      assert(last_snapshot == current_user_key_snapshot_ ||
-             (snapshot_checker_ != nullptr &&
-              snapshot_checker_->CheckInSnapshot(current_user_key_sequence_,
-                                                 last_snapshot) ==
-                  SnapshotCheckerResult::kSnapshotReleased));
-
      ++iter_stats_.num_record_drop_hidden;  // (A)
      input_->Next();
    } else if (compaction_ != nullptr && ikey_.type == kTypeDeletion &&
@@ -634,8 +628,7 @@ void CompactionIterator::PrepareOutput() {
  // KeyNotExistsBeyondOutputLevel() return true?
  if ((compaction_ != nullptr && !compaction_->allow_ingest_behind()) &&
      ikeyNotNeededForIncrementalSnapshot() && bottommost_level_ && valid_ &&
-      IN_EARLIEST_SNAPSHOT(ikey_.sequence) && ikey_.type != kTypeMerge &&
-      !cmp_->Equal(compaction_->GetLargestUserKey(), ikey_.user_key)) {
+      IN_EARLIEST_SNAPSHOT(ikey_.sequence) && ikey_.type != kTypeMerge) {
    assert(ikey_.type != kTypeDeletion && ikey_.type != kTypeSingleDeletion);
    ikey_.sequence = 0;
    current_key_.UpdateInternalKey(0, ikey_.type);
@@ -10,8 +10,8 @@
 #include <unordered_set>
 #include <vector>

-#include "db/compaction.h"
-#include "db/compaction_iteration_stats.h"
+#include "db/compaction/compaction.h"
+#include "db/compaction/compaction_iteration_stats.h"
 #include "db/merge_helper.h"
 #include "db/pinned_iterators_manager.h"
 #include "db/range_del_aggregator.h"
@@ -21,6 +21,53 @@

 namespace rocksdb {

+// This callback can be used to refresh the snapshot list from the db. It
+// includes logics to exponentially decrease the refresh rate to limit the
+// overhead of refresh.
+class SnapshotListFetchCallback {
+ public:
+  SnapshotListFetchCallback(Env* env, uint64_t snap_refresh_nanos,
+                            size_t every_nth_key = 1024)
+      : timer_(env, /*auto restart*/ true),
+        snap_refresh_nanos_(snap_refresh_nanos),
+        every_nth_key_minus_one_(every_nth_key - 1) {
+    assert(every_nth_key > 0);
+    assert((ceil(log2(every_nth_key)) == floor(log2(every_nth_key))));
+  }
+  // Refresh the snapshot list. snapshots will bre replacted with the new list.
+  // max is the upper bound. Note: this function will acquire the db_mutex_.
+  virtual void Refresh(std::vector<SequenceNumber>* snapshots,
+                       SequenceNumber max) = 0;
+  inline bool TimeToRefresh(const size_t key_index) {
+    // skip the key if key_index % every_nth_key (which is of power 2) is not 0.
+    if ((key_index & every_nth_key_minus_one_) != 0) {
+      return false;
+    }
+    const uint64_t elapsed = timer_.ElapsedNanos();
+    auto ret = elapsed > snap_refresh_nanos_;
+    // pre-compute the next time threshold
+    if (ret) {
+      // inc next refresh period exponentially (by x4)
+      auto next_refresh_threshold = snap_refresh_nanos_ << 2;
+      // make sure the shift has not overflown the highest 1 bit
+      snap_refresh_nanos_ =
+          std::max(snap_refresh_nanos_, next_refresh_threshold);
+    }
+    return ret;
+  }
+  static constexpr SnapshotListFetchCallback* kDisabled = nullptr;
+
+  virtual ~SnapshotListFetchCallback() {}
+
+ private:
+  // Time since the callback was created
+  StopWatchNano timer_;
+  // The delay before calling ::Refresh. To be increased exponentially.
+  uint64_t snap_refresh_nanos_;
+  // Skip evey nth key. Number n if of power 2. The math will require n-1.
+  const uint64_t every_nth_key_minus_one_;
+};
+
 class CompactionIterator {
 public:
  // A wrapper around Compaction. Has a much smaller interface, only what
@@ -69,7 +116,8 @@ class CompactionIterator {
                     const Compaction* compaction = nullptr,
                     const CompactionFilter* compaction_filter = nullptr,
                     const std::atomic<bool>* shutting_down = nullptr,
-                     const SequenceNumber preserve_deletes_seqnum = 0);
+                     const SequenceNumber preserve_deletes_seqnum = 0,
+                     SnapshotListFetchCallback* snap_list_callback = nullptr);

  // Constructor with custom CompactionProxy, used for tests.
  CompactionIterator(InternalIterator* input, const Comparator* cmp,
@@ -82,7 +130,8 @@ class CompactionIterator {
                     std::unique_ptr<CompactionProxy> compaction,
                     const CompactionFilter* compaction_filter = nullptr,
                     const std::atomic<bool>* shutting_down = nullptr,
-                     const SequenceNumber preserve_deletes_seqnum = 0);
+                     const SequenceNumber preserve_deletes_seqnum = 0,
+                     SnapshotListFetchCallback* snap_list_callback = nullptr);

  ~CompactionIterator();

@@ -110,6 +159,8 @@ class CompactionIterator {
 private:
  // Processes the input stream to find the next output
  void NextFromInput();
+  // Process snapshots_ and assign related variables
+  void ProcessSnapshotList();

  // Do last preparations before presenting the output to the callee. At this
  // point this only zeroes out the sequence number if possible for better
@@ -144,7 +195,7 @@ class CompactionIterator {
  InternalIterator* input_;
  const Comparator* cmp_;
  MergeHelper* merge_helper_;
-  const std::vector<SequenceNumber>* snapshots_;
+  std::vector<SequenceNumber>* snapshots_;
  // List of snapshots released during compaction.
  // findEarliestVisibleSnapshot() find them out from return of
  // snapshot_checker, and make sure they will not be returned as
@@ -168,8 +219,6 @@ class CompactionIterator {
  SequenceNumber earliest_snapshot_;
  SequenceNumber latest_snapshot_;

-  bool ignore_snapshots_;
-
  // State
  //
  // Points to a copy of the current compaction iterator output (current_key_)
@@ -221,6 +270,9 @@ class CompactionIterator {
  // Used to avoid purging uncommitted values. The application can specify
  // uncommitted values by providing a SnapshotChecker object.
  bool current_key_committed_;
+  SnapshotListFetchCallback* snap_list_callback_;
+  // number of distinct keys processed
+  size_t num_keys_ = 0;

  bool IsShuttingDown() {
    // This is a best-effort facility, so memory_order_relaxed is sufficient.
@@ -3,15 +3,15 @@
 //  COPYING file in the root directory) and Apache 2.0 License
 //  (found in the LICENSE.Apache file in the root directory).

-#include "db/compaction_iterator.h"

 #include <string>
 #include <vector>

+#include "db/compaction/compaction_iterator.h"
 #include "port/port.h"
+#include "test_util/testharness.h"
+#include "test_util/testutil.h"
 #include "util/string_util.h"
-#include "util/testharness.h"
-#include "util/testutil.h"
 #include "utilities/merge_operators.h"

 namespace rocksdb {
@@ -112,39 +112,39 @@ class LoggingForwardVectorIterator : public InternalIterator {
    assert(keys_.size() == values_.size());
  }

-  virtual bool Valid() const override { return current_ < keys_.size(); }
+  bool Valid() const override { return current_ < keys_.size(); }

-  virtual void SeekToFirst() override {
+  void SeekToFirst() override {
    log.emplace_back(Action::Type::SEEK_TO_FIRST);
    current_ = 0;
  }
-  virtual void SeekToLast() override { assert(false); }
+  void SeekToLast() override { assert(false); }

-  virtual void Seek(const Slice& target) override {
+  void Seek(const Slice& target) override {
    log.emplace_back(Action::Type::SEEK, target.ToString());
    current_ = std::lower_bound(keys_.begin(), keys_.end(), target.ToString()) -
               keys_.begin();
  }

-  virtual void SeekForPrev(const Slice& /*target*/) override { assert(false); }
+  void SeekForPrev(const Slice& /*target*/) override { assert(false); }

-  virtual void Next() override {
+  void Next() override {
    assert(Valid());
    log.emplace_back(Action::Type::NEXT);
    current_++;
  }
-  virtual void Prev() override { assert(false); }
+  void Prev() override { assert(false); }

-  virtual Slice key() const override {
+  Slice key() const override {
    assert(Valid());
    return Slice(keys_[current_]);
  }
-  virtual Slice value() const override {
+  Slice value() const override {
    assert(Valid());
    return Slice(values_[current_]);
  }

-  virtual Status status() const override { return Status::OK(); }
+  Status status() const override { return Status::OK(); }

  std::vector<Action> log;

@@ -158,22 +158,20 @@ class FakeCompaction : public CompactionIterator::CompactionProxy {
 public:
  FakeCompaction() = default;

-  virtual int level(size_t /*compaction_input_level*/) const override {
-    return 0;
-  }
-  virtual bool KeyNotExistsBeyondOutputLevel(
+  int level(size_t /*compaction_input_level*/) const override { return 0; }
+  bool KeyNotExistsBeyondOutputLevel(
      const Slice& /*user_key*/,
      std::vector<size_t>* /*level_ptrs*/) const override {
    return is_bottommost_level || key_not_exists_beyond_output_level;
  }
-  virtual bool bottommost_level() const override { return is_bottommost_level; }
-  virtual int number_levels() const override { return 1; }
-  virtual Slice GetLargestUserKey() const override {
+  bool bottommost_level() const override { return is_bottommost_level; }
+  int number_levels() const override { return 1; }
+  Slice GetLargestUserKey() const override {
    return "\xff\xff\xff\xff\xff\xff\xff\xff\xff";
  }
-  virtual bool allow_ingest_behind() const override { return false; }
+  bool allow_ingest_behind() const override { return false; }

-  virtual bool preserve_deletes() const override { return false; }
+  bool preserve_deletes() const override { return false; }

  bool key_not_exists_beyond_output_level = false;

@@ -377,10 +375,9 @@ TEST_P(CompactionIteratorTest, RangeDeletionWithSnapshots) {

 TEST_P(CompactionIteratorTest, CompactionFilterSkipUntil) {
  class Filter : public CompactionFilter {
-    virtual Decision FilterV2(int /*level*/, const Slice& key, ValueType t,
-                              const Slice& existing_value,
-                              std::string* /*new_value*/,
-                              std::string* skip_until) const override {
+    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
@@ -560,10 +557,9 @@ TEST_P(CompactionIteratorTest, ShuttingDownInMerge) {

 TEST_P(CompactionIteratorTest, SingleMergeOperand) {
  class Filter : public CompactionFilter {
-    virtual Decision FilterV2(int /*level*/, const Slice& key, ValueType t,
-                              const Slice& existing_value,
-                              std::string* /*new_value*/,
-                              std::string* /*skip_until*/) const override {
+    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();

@@ -7,13 +7,7 @@
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file. See the AUTHORS file for names of contributors.

-#include "db/compaction_job.h"
-
-#ifndef __STDC_FORMAT_MACROS
-#define __STDC_FORMAT_MACROS
-#endif
-
-#include <inttypes.h>
+#include <cinttypes>
 #include <algorithm>
 #include <functional>
 #include <list>
@@ -25,7 +19,8 @@
 #include <vector>

 #include "db/builder.h"
-#include "db/db_impl.h"
+#include "db/compaction/compaction_job.h"
+#include "db/db_impl/db_impl.h"
 #include "db/db_iter.h"
 #include "db/dbformat.h"
 #include "db/error_handler.h"
@@ -38,6 +33,10 @@
 #include "db/merge_helper.h"
 #include "db/range_del_aggregator.h"
 #include "db/version_set.h"
+#include "file/filename.h"
+#include "file/sst_file_manager_impl.h"
+#include "logging/log_buffer.h"
+#include "logging/logging.h"
 #include "monitoring/iostats_context_imp.h"
 #include "monitoring/perf_context_imp.h"
 #include "monitoring/thread_status_util.h"
@@ -47,21 +46,17 @@
 #include "rocksdb/statistics.h"
 #include "rocksdb/status.h"
 #include "rocksdb/table.h"
-#include "table/block.h"
-#include "table/block_based_table_factory.h"
+#include "table/block_based/block.h"
+#include "table/block_based/block_based_table_factory.h"
 #include "table/merging_iterator.h"
 #include "table/table_builder.h"
+#include "test_util/sync_point.h"
 #include "util/coding.h"
 #include "util/file_reader_writer.h"
-#include "util/filename.h"
-#include "util/log_buffer.h"
-#include "util/logging.h"
 #include "util/mutexlock.h"
 #include "util/random.h"
-#include "util/sst_file_manager_impl.h"
 #include "util/stop_watch.h"
 #include "util/string_util.h"
-#include "util/sync_point.h"

 namespace rocksdb {

@@ -97,6 +92,8 @@ const char* GetCompactionReasonString(CompactionReason compaction_reason) {
      return "Flush";
    case CompactionReason::kExternalSstIngestion:
      return "ExternalSstIngestion";
+    case CompactionReason::kPeriodicCompaction:
+      return "PeriodicCompaction";
    case CompactionReason::kNumOfReasons:
      // fall through
    default:
@@ -157,7 +154,6 @@ struct CompactionJob::SubcompactionState {
  uint64_t overlapped_bytes = 0;
  // A flag determine whether the key has been seen in ShouldStopBefore()
  bool seen_key = false;
-  std::string compression_dict;

  SubcompactionState(Compaction* c, Slice* _start, Slice* _end,
                     uint64_t size = 0)
@@ -173,8 +169,7 @@ struct CompactionJob::SubcompactionState {
        approx_size(size),
        grandparent_index(0),
        overlapped_bytes(0),
-        seen_key(false),
-        compression_dict() {
+        seen_key(false) {
    assert(compaction != nullptr);
  }

@@ -197,7 +192,6 @@ struct CompactionJob::SubcompactionState {
    grandparent_index = std::move(o.grandparent_index);
    overlapped_bytes = std::move(o.overlapped_bytes);
    seen_key = std::move(o.seen_key);
-    compression_dict = std::move(o.compression_dict);
    return *this;
  }

@@ -315,7 +309,8 @@ CompactionJob::CompactionJob(
    SequenceNumber earliest_write_conflict_snapshot,
    const SnapshotChecker* snapshot_checker, std::shared_ptr<Cache> table_cache,
    EventLogger* event_logger, bool paranoid_file_checks, bool measure_io_stats,
-    const std::string& dbname, CompactionJobStats* compaction_job_stats)
+    const std::string& dbname, CompactionJobStats* compaction_job_stats,
+    Env::Priority thread_pri, SnapshotListFetchCallback* snap_list_callback)
    : job_id_(job_id),
      compact_(new CompactionState(compaction)),
      compaction_job_stats_(compaction_job_stats),
@@ -324,7 +319,7 @@ CompactionJob::CompactionJob(
      db_options_(db_options),
      env_options_(env_options),
      env_(db_options.env),
-      env_optiosn_for_read_(
+      env_options_for_read_(
          env_->OptimizeForCompactionTableRead(env_options, db_options_)),
      versions_(versions),
      shutting_down_(shutting_down),
@@ -336,6 +331,7 @@ CompactionJob::CompactionJob(
      db_mutex_(db_mutex),
      db_error_handler_(db_error_handler),
      existing_snapshots_(std::move(existing_snapshots)),
+      snap_list_callback_(snap_list_callback),
      earliest_write_conflict_snapshot_(earliest_write_conflict_snapshot),
      snapshot_checker_(snapshot_checker),
      table_cache_(std::move(table_cache)),
@@ -343,7 +339,8 @@ CompactionJob::CompactionJob(
      bottommost_level_(false),
      paranoid_file_checks_(paranoid_file_checks),
      measure_io_stats_(measure_io_stats),
-      write_hint_(Env::WLTH_NOT_SET) {
+      write_hint_(Env::WLTH_NOT_SET),
+      thread_pri_(thread_pri) {
  assert(log_buffer_ != nullptr);
  const auto* cfd = compact_->compaction->column_family_data();
  ThreadStatusUtil::SetColumnFamily(cfd, cfd->ioptions()->env,
@@ -413,15 +410,13 @@ void CompactionJob::Prepare() {

  write_hint_ =
      c->column_family_data()->CalculateSSTWriteHint(c->output_level());
-  // Is this compaction producing files at the bottommost level?
  bottommost_level_ = c->bottommost_level();

  if (c->ShouldFormSubcompactions()) {
-    const uint64_t start_micros = env_->NowMicros();
-    GenSubcompactionBoundaries();
-    MeasureTime(stats_, SUBCOMPACTION_SETUP_TIME,
-                env_->NowMicros() - start_micros);
-
+    {
+      StopWatch sw(env_, stats_, SUBCOMPACTION_SETUP_TIME);
+      GenSubcompactionBoundaries();
+    }
    assert(sizes_.size() == boundaries_.size() + 1);

    for (size_t i = 0; i <= boundaries_.size(); i++) {
@@ -429,8 +424,8 @@ void CompactionJob::Prepare() {
      Slice* end = i == boundaries_.size() ? nullptr : &boundaries_[i];
      compact_->sub_compact_states.emplace_back(c, start, end, sizes_[i]);
    }
-    MeasureTime(stats_, NUM_SUBCOMPACTIONS_SCHEDULED,
-                compact_->sub_compact_states.size());
+    RecordInHistogram(stats_, NUM_SUBCOMPACTIONS_SCHEDULED,
+                      compact_->sub_compact_states.size());
  } else {
    compact_->sub_compact_states.emplace_back(c, nullptr, nullptr);
  }
@@ -444,11 +439,6 @@ struct RangeWithSize {
      : range(a, b), size(s) {}
 };

-// Generates a histogram representing potential divisions of key ranges from
-// the input. It adds the starting and/or ending keys of certain input files
-// to the working set and then finds the approximate size of data in between
-// each consecutive pair of slices. Then it divides these ranges into
-// consecutive groups such that each group has a similar size.
 void CompactionJob::GenSubcompactionBoundaries() {
  auto* c = compact_->compaction;
  auto* cfd = c->column_family_data();
@@ -518,7 +508,7 @@ void CompactionJob::GenSubcompactionBoundaries() {
  auto* v = compact_->compaction->input_version();
  for (auto it = bounds.begin();;) {
    const Slice a = *it;
-    it++;
+    ++it;

    if (it == bounds.end()) {
      break;
@@ -530,7 +520,8 @@ void CompactionJob::GenSubcompactionBoundaries() {
    // to the index block and may incur I/O cost in the process. Unlock db
    // mutex to reduce contention
    db_mutex_->Unlock();
-    uint64_t size = versions_->ApproximateSize(v, a, b, start_lvl, out_lvl + 1);
+    uint64_t size = versions_->ApproximateSize(v, a, b, start_lvl, out_lvl + 1,
+                                               TableReaderCaller::kCompaction);
    db_mutex_->Lock();
    ranges.emplace_back(a, b, size);
    sum += size;
@@ -610,8 +601,9 @@ Status CompactionJob::Run() {
        compact_->sub_compact_states[i].compaction_job_stats.cpu_micros;
  }

-  MeasureTime(stats_, COMPACTION_TIME, compaction_stats_.micros);
-  MeasureTime(stats_, COMPACTION_CPU_TIME, compaction_stats_.cpu_micros);
+  RecordTimeToHistogram(stats_, COMPACTION_TIME, compaction_stats_.micros);
+  RecordTimeToHistogram(stats_, COMPACTION_CPU_TIME,
+                        compaction_stats_.cpu_micros);

  TEST_SYNC_POINT("CompactionJob::Run:BeforeVerify");

@@ -654,12 +646,14 @@ Status CompactionJob::Run() {
        // to cache it here for further user reads
        InternalIterator* iter = cfd->table_cache()->NewIterator(
            ReadOptions(), env_options_, cfd->internal_comparator(),
-            *files_meta[file_idx], nullptr /* range_del_agg */,
-            prefix_extractor, nullptr,
+            *files_meta[file_idx], /*range_del_agg=*/nullptr, prefix_extractor,
+            /*table_reader_ptr=*/nullptr,
            cfd->internal_stats()->GetFileReadHist(
                compact_->compaction->output_level()),
-            false, nullptr /* arena */, false /* skip_filters */,
-            compact_->compaction->output_level());
+            TableReaderCaller::kCompactionRefill, /*arena=*/nullptr,
+            /*skip_filters=*/false, compact_->compaction->output_level(),
+            /*smallest_compaction_key=*/nullptr,
+            /*largest_compaction_key=*/nullptr);
        auto s = iter->status();

        if (s.ok() && paranoid_file_checks_) {
@@ -720,7 +714,7 @@ Status CompactionJob::Install(const MutableCFOptions& mutable_cf_options) {
  Status status = compact_->status;
  ColumnFamilyData* cfd = compact_->compaction->column_family_data();
  cfd->internal_stats()->AddCompactionStats(
-      compact_->compaction->output_level(), compaction_stats_);
+      compact_->compaction->output_level(), thread_pri_, compaction_stats_);

  if (status.ok()) {
    status = InstallCompactionResults(mutable_cf_options);
@@ -816,13 +810,31 @@ void CompactionJob::ProcessKeyValueCompaction(SubcompactionState* sub_compact) {
  uint64_t prev_cpu_micros = env_->NowCPUNanos() / 1000;

  ColumnFamilyData* cfd = sub_compact->compaction->column_family_data();
+
+  // Create compaction filter and fail the compaction if
+  // IgnoreSnapshots() = false because it is not supported anymore
+  const CompactionFilter* compaction_filter =
+      cfd->ioptions()->compaction_filter;
+  std::unique_ptr<CompactionFilter> compaction_filter_from_factory = nullptr;
+  if (compaction_filter == nullptr) {
+    compaction_filter_from_factory =
+        sub_compact->compaction->CreateCompactionFilter();
+    compaction_filter = compaction_filter_from_factory.get();
+  }
+  if (compaction_filter != nullptr && !compaction_filter->IgnoreSnapshots()) {
+    sub_compact->status = Status::NotSupported(
+        "CompactionFilter::IgnoreSnapshots() = false is not supported "
+        "anymore.");
+    return;
+  }
+
  CompactionRangeDelAggregator range_del_agg(&cfd->internal_comparator(),
                                             existing_snapshots_);

  // Although the v2 aggregator is what the level iterator(s) know about,
  // the AddTombstones calls will be propagated down to the v1 aggregator.
  std::unique_ptr<InternalIterator> input(versions_->MakeInputIterator(
-      sub_compact->compaction, &range_del_agg, env_optiosn_for_read_));
+      sub_compact->compaction, &range_del_agg, env_options_for_read_));

  AutoThreadOperationStageUpdater stage_updater(
      ThreadStatus::STAGE_COMPACTION_PROCESS_KV);
@@ -847,49 +859,6 @@ void CompactionJob::ProcessKeyValueCompaction(SubcompactionState* sub_compact) {
    prev_cpu_read_nanos = IOSTATS(cpu_read_nanos);
  }

-  const MutableCFOptions* mutable_cf_options =
-      sub_compact->compaction->mutable_cf_options();
-
-  // To build compression dictionary, we sample the first output file, assuming
-  // it'll reach the maximum length. We optionally pass these samples through
-  // zstd's dictionary trainer, or just use them directly. Then, the dictionary
-  // is used for compressing subsequent output files in the same subcompaction.
-  const bool kUseZstdTrainer =
-      sub_compact->compaction->output_compression_opts().zstd_max_train_bytes >
-      0;
-  const size_t kSampleBytes =
-      kUseZstdTrainer
-          ? sub_compact->compaction->output_compression_opts()
-                .zstd_max_train_bytes
-          : sub_compact->compaction->output_compression_opts().max_dict_bytes;
-  const int kSampleLenShift = 6;  // 2^6 = 64-byte samples
-  std::set<size_t> sample_begin_offsets;
-  if (bottommost_level_ && kSampleBytes > 0) {
-    const size_t kMaxSamples = kSampleBytes >> kSampleLenShift;
-    const size_t kOutFileLen =
-        static_cast<size_t>(MaxFileSizeForLevel(*mutable_cf_options,
-            compact_->compaction->output_level(),
-            cfd->ioptions()->compaction_style,
-            compact_->compaction->GetInputBaseLevel(),
-            cfd->ioptions()->level_compaction_dynamic_level_bytes));
-    if (kOutFileLen != port::kMaxSizet) {
-      const size_t kOutFileNumSamples = kOutFileLen >> kSampleLenShift;
-      Random64 generator{versions_->NewFileNumber()};
-      for (size_t i = 0; i < kMaxSamples; ++i) {
-        sample_begin_offsets.insert(
-            static_cast<size_t>(generator.Uniform(kOutFileNumSamples))
-            << kSampleLenShift);
-      }
-    }
-  }
-
-  auto compaction_filter = cfd->ioptions()->compaction_filter;
-  std::unique_ptr<CompactionFilter> compaction_filter_from_factory = nullptr;
-  if (compaction_filter == nullptr) {
-    compaction_filter_from_factory =
-        sub_compact->compaction->CreateCompactionFilter();
-    compaction_filter = compaction_filter_from_factory.get();
-  }
  MergeHelper merge(
      env_, cfd->user_comparator(), cfd->ioptions()->merge_operator,
      compaction_filter, db_options_.info_log.get(),
@@ -916,7 +885,9 @@ void CompactionJob::ProcessKeyValueCompaction(SubcompactionState* sub_compact) {
      &existing_snapshots_, earliest_write_conflict_snapshot_,
      snapshot_checker_, env_, ShouldReportDetailedTime(env_, stats_), false,
      &range_del_agg, sub_compact->compaction, compaction_filter,
-      shutting_down_, preserve_deletes_seqnum_));
+      shutting_down_, preserve_deletes_seqnum_,
+      // Currently range_del_agg is incompatible with snapshot refresh feature.
+      range_del_agg.IsEmpty() ? snap_list_callback_ : nullptr));
  auto c_iter = sub_compact->c_iter.get();
  c_iter->SeekToFirst();
  if (c_iter->Valid() && sub_compact->compaction->output_level() != 0) {
@@ -927,12 +898,6 @@ void CompactionJob::ProcessKeyValueCompaction(SubcompactionState* sub_compact) {
                                  sub_compact->current_output_file_size);
  }
  const auto& c_iter_stats = c_iter->iter_stats();
-  auto sample_begin_offset_iter = sample_begin_offsets.cbegin();
-  // data_begin_offset and dict_sample_data are only valid while generating
-  // dictionary from the first output file.
-  size_t data_begin_offset = 0;
-  std::string dict_sample_data;
-  dict_sample_data.reserve(kSampleBytes);

  while (status.ok() && !cfd->IsDropped() && c_iter->Valid()) {
    // Invariant: c_iter.status() is guaranteed to be OK if c_iter->Valid()
@@ -968,55 +933,6 @@ void CompactionJob::ProcessKeyValueCompaction(SubcompactionState* sub_compact) {
        key, c_iter->ikey().sequence);
    sub_compact->num_output_records++;

-    if (sub_compact->outputs.size() == 1) {  // first output file
-      // Check if this key/value overlaps any sample intervals; if so, appends
-      // overlapping portions to the dictionary.
-      for (const auto& data_elmt : {key, value}) {
-        size_t data_end_offset = data_begin_offset + data_elmt.size();
-        while (sample_begin_offset_iter != sample_begin_offsets.cend() &&
-               *sample_begin_offset_iter < data_end_offset) {
-          size_t sample_end_offset =
-              *sample_begin_offset_iter + (1 << kSampleLenShift);
-          // Invariant: Because we advance sample iterator while processing the
-          // data_elmt containing the sample's last byte, the current sample
-          // cannot end before the current data_elmt.
-          assert(data_begin_offset < sample_end_offset);
-
-          size_t data_elmt_copy_offset, data_elmt_copy_len;
-          if (*sample_begin_offset_iter <= data_begin_offset) {
-            // The sample starts before data_elmt starts, so take bytes starting
-            // at the beginning of data_elmt.
-            data_elmt_copy_offset = 0;
-          } else {
-            // data_elmt starts before the sample starts, so take bytes starting
-            // at the below offset into data_elmt.
-            data_elmt_copy_offset =
-                *sample_begin_offset_iter - data_begin_offset;
-          }
-          if (sample_end_offset <= data_end_offset) {
-            // The sample ends before data_elmt ends, so take as many bytes as
-            // needed.
-            data_elmt_copy_len =
-                sample_end_offset - (data_begin_offset + data_elmt_copy_offset);
-          } else {
-            // data_elmt ends before the sample ends, so take all remaining
-            // bytes in data_elmt.
-            data_elmt_copy_len =
-                data_end_offset - (data_begin_offset + data_elmt_copy_offset);
-          }
-          dict_sample_data.append(&data_elmt.data()[data_elmt_copy_offset],
-                                  data_elmt_copy_len);
-          if (sample_end_offset > data_end_offset) {
-            // Didn't finish sample. Try to finish it with the next data_elmt.
-            break;
-          }
-          // Next sample may require bytes from same data_elmt.
-          sample_begin_offset_iter++;
-        }
-        data_begin_offset = data_end_offset;
-      }
-    }
-
    // Close output file if it is big enough. Two possibilities determine it's
    // time to close it: (1) the current key should be this file's last key, (2)
    // the next key should not be in this file.
@@ -1058,18 +974,6 @@ void CompactionJob::ProcessKeyValueCompaction(SubcompactionState* sub_compact) {
                                     &range_del_out_stats, next_key);
      RecordDroppedKeys(range_del_out_stats,
                        &sub_compact->compaction_job_stats);
-      if (sub_compact->outputs.size() == 1) {
-        // Use samples from first output file to create dictionary for
-        // compression of subsequent files.
-        if (kUseZstdTrainer) {
-          sub_compact->compression_dict = ZSTD_TrainDictionary(
-              dict_sample_data, kSampleLenShift,
-              sub_compact->compaction->output_compression_opts()
-                  .max_dict_bytes);
-        } else {
-          sub_compact->compression_dict = std::move(dict_sample_data);
-        }
-      }
    }
  }

@@ -1092,10 +996,13 @@ void CompactionJob::ProcessKeyValueCompaction(SubcompactionState* sub_compact) {
  RecordDroppedKeys(c_iter_stats, &sub_compact->compaction_job_stats);
  RecordCompactionIOStats();

-  if (status.ok() &&
-      (shutting_down_->load(std::memory_order_relaxed) || cfd->IsDropped())) {
-    status = Status::ShutdownInProgress(
-        "Database shutdown or Column family drop during compaction");
+  if (status.ok() && cfd->IsDropped()) {
+    status =
+        Status::ColumnFamilyDropped("Column family dropped during compaction");
+  }
+  if ((status.ok() || status.IsColumnFamilyDropped()) &&
+      shutting_down_->load(std::memory_order_relaxed)) {
+    status = Status::ShutdownInProgress("Database shutdown");
  }
  if (status.ok()) {
    status = input->status();
@@ -1228,10 +1135,19 @@ Status CompactionJob::FinishCompactionOutputFile(
      lower_bound = nullptr;
    }
    if (next_table_min_key != nullptr) {
-      // This isn't the last file in the subcompaction, so extend until the next
-      // file starts.
+      // This may be the last file in the subcompaction in some cases, so we
+      // need to compare the end key of subcompaction with the next file start
+      // key. When the end key is chosen by the subcompaction, we know that
+      // it must be the biggest key in output file. Therefore, it is safe to
+      // use the smaller key as the upper bound of the output file, to ensure
+      // that there is no overlapping between different output files.
      upper_bound_guard = ExtractUserKey(*next_table_min_key);
-      upper_bound = &upper_bound_guard;
+      if (sub_compact->end != nullptr &&
+          ucmp->Compare(upper_bound_guard, *sub_compact->end) >= 0) {
+        upper_bound = sub_compact->end;
+      } else {
+        upper_bound = &upper_bound_guard;
+      }
    } else {
      // This is the last file in the subcompaction, so extend until the
      // subcompaction ends.
@@ -1249,6 +1165,13 @@ Status CompactionJob::FinishCompactionOutputFile(
      has_overlapping_endpoints = false;
    }

+    // The end key of the subcompaction must be bigger or equal to the upper
+    // bound. If the end of subcompaction is null or the upper bound is null,
+    // it means that this file is the last file in the compaction. So there
+    // will be no overlapping between this file and others.
+    assert(sub_compact->end == nullptr ||
+           upper_bound == nullptr ||
+           ucmp->Compare(*upper_bound , *sub_compact->end) <= 0);
    auto it = range_del_agg->NewIterator(lower_bound, upper_bound,
                                         has_overlapping_endpoints);
    // Position the range tombstone output iterator. There may be tombstone
@@ -1557,20 +1480,20 @@ Status CompactionJob::OpenCompactionOutputFile(
  bool skip_filters =
      cfd->ioptions()->optimize_filters_for_hits && bottommost_level_;

-  uint64_t output_file_creation_time =
+  int64_t temp_current_time = 0;
+  auto get_time_status = env_->GetCurrentTime(&temp_current_time);
+  // Safe to proceed even if GetCurrentTime fails. So, log and proceed.
+  if (!get_time_status.ok()) {
+    ROCKS_LOG_WARN(db_options_.info_log,
+                   "Failed to get current time. Status: %s",
+                   get_time_status.ToString().c_str());
+  }
+  uint64_t current_time = static_cast<uint64_t>(temp_current_time);
+
+  uint64_t latest_key_time =
      sub_compact->compaction->MaxInputFileCreationTime();
-  if (output_file_creation_time == 0) {
-    int64_t _current_time = 0;
-    auto status = db_options_.env->GetCurrentTime(&_current_time);
-    // Safe to proceed even if GetCurrentTime fails. So, log and proceed.
-    if (!status.ok()) {
-      ROCKS_LOG_WARN(
-          db_options_.info_log,
-          "Failed to get current time to populate creation_time property. "
-          "Status: %s",
-          status.ToString().c_str());
-    }
-    output_file_creation_time = static_cast<uint64_t>(_current_time);
+  if (latest_key_time == 0) {
+    latest_key_time = current_time;
  }

  sub_compact->builder.reset(NewTableBuilder(
@@ -1578,9 +1501,11 @@ Status CompactionJob::OpenCompactionOutputFile(
      cfd->internal_comparator(), cfd->int_tbl_prop_collector_factories(),
      cfd->GetID(), cfd->GetName(), sub_compact->outfile.get(),
      sub_compact->compaction->output_compression(),
+      0 /*sample_for_compression */,
      sub_compact->compaction->output_compression_opts(),
-      sub_compact->compaction->output_level(), &sub_compact->compression_dict,
-      skip_filters, output_file_creation_time));
+      sub_compact->compaction->output_level(), skip_filters, latest_key_time,
+      0 /* oldest_key_time */, sub_compact->compaction->max_output_file_size(),
+      current_time));
  LogFlush(db_options_.info_log);
  return s;
 }
@@ -18,7 +18,7 @@
 #include <vector>

 #include "db/column_family.h"
-#include "db/compaction_iterator.h"
+#include "db/compaction/compaction_iterator.h"
 #include "db/dbformat.h"
 #include "db/flush_scheduler.h"
 #include "db/internal_stats.h"
@@ -29,6 +29,7 @@
 #include "db/version_edit.h"
 #include "db/write_controller.h"
 #include "db/write_thread.h"
+#include "logging/event_logger.h"
 #include "options/cf_options.h"
 #include "options/db_options.h"
 #include "port/port.h"
@@ -40,7 +41,6 @@
 #include "rocksdb/transaction_log.h"
 #include "table/scoped_arena_iterator.h"
 #include "util/autovector.h"
-#include "util/event_logger.h"
 #include "util/stop_watch.h"
 #include "util/thread_local.h"

@@ -55,24 +55,27 @@ class Version;
 class VersionEdit;
 class VersionSet;

+// CompactionJob is responsible for executing the compaction. Each (manual or
+// automated) compaction corresponds to a CompactionJob object, and usually
+// goes through the stages of `Prepare()`->`Run()`->`Install()`. CompactionJob
+// will divide the compaction into subcompactions and execute them in parallel
+// if needed.
 class CompactionJob {
 public:
-  CompactionJob(int job_id, Compaction* compaction,
-                const ImmutableDBOptions& db_options,
-                const EnvOptions env_options, VersionSet* versions,
-                const std::atomic<bool>* shutting_down,
-                const SequenceNumber preserve_deletes_seqnum,
-                LogBuffer* log_buffer,
-                Directory* db_directory, Directory* output_directory,
-                Statistics* stats, InstrumentedMutex* db_mutex,
-                ErrorHandler* db_error_handler,
-                std::vector<SequenceNumber> existing_snapshots,
-                SequenceNumber earliest_write_conflict_snapshot,
-                const SnapshotChecker* snapshot_checker,
-                std::shared_ptr<Cache> table_cache, EventLogger* event_logger,
-                bool paranoid_file_checks, bool measure_io_stats,
-                const std::string& dbname,
-                CompactionJobStats* compaction_job_stats);
+  CompactionJob(
+      int job_id, Compaction* compaction, const ImmutableDBOptions& db_options,
+      const EnvOptions env_options, VersionSet* versions,
+      const std::atomic<bool>* shutting_down,
+      const SequenceNumber preserve_deletes_seqnum, LogBuffer* log_buffer,
+      Directory* db_directory, Directory* output_directory, Statistics* stats,
+      InstrumentedMutex* db_mutex, ErrorHandler* db_error_handler,
+      std::vector<SequenceNumber> existing_snapshots,
+      SequenceNumber earliest_write_conflict_snapshot,
+      const SnapshotChecker* snapshot_checker,
+      std::shared_ptr<Cache> table_cache, EventLogger* event_logger,
+      bool paranoid_file_checks, bool measure_io_stats,
+      const std::string& dbname, CompactionJobStats* compaction_job_stats,
+      Env::Priority thread_pri, SnapshotListFetchCallback* snap_list_callback);

  ~CompactionJob();

@@ -82,17 +85,28 @@ class CompactionJob {
  CompactionJob& operator=(const CompactionJob& job) = delete;

  // REQUIRED: mutex held
+  // Prepare for the compaction by setting up boundaries for each subcompaction
  void Prepare();
  // REQUIRED mutex not held
+  // Launch threads for each subcompaction and wait for them to finish. After
+  // that, verify table is usable and finally do bookkeeping to unify
+  // subcompaction results
  Status Run();

  // REQUIRED: mutex held
+  // Add compaction input/output to the current version
  Status Install(const MutableCFOptions& mutable_cf_options);

 private:
  struct SubcompactionState;

  void AggregateStatistics();
+
+  // Generates a histogram representing potential divisions of key ranges from
+  // the input. It adds the starting and/or ending keys of certain input files
+  // to the working set and then finds the approximate size of data in between
+  // each consecutive pair of slices. Then it divides these ranges into
+  // consecutive groups such that each group has a similar size.
  void GenSubcompactionBoundaries();

  // update the thread status for starting a compaction.
@@ -137,7 +151,7 @@ class CompactionJob {

  Env* env_;
  // env_option optimized for compaction table reads
-  EnvOptions env_optiosn_for_read_;
+  EnvOptions env_options_for_read_;
  VersionSet* versions_;
  const std::atomic<bool>* shutting_down_;
  const SequenceNumber preserve_deletes_seqnum_;
@@ -152,6 +166,7 @@ class CompactionJob {
  // entirely within s1 and s2, then the earlier version of k1 can be safely
  // deleted because that version is not visible in any snapshot.
  std::vector<SequenceNumber> existing_snapshots_;
+  SnapshotListFetchCallback* snap_list_callback_;

  // This is the earliest snapshot that could be used for write-conflict
  // checking by a transaction.  For any user-key newer than this snapshot, we
@@ -164,6 +179,7 @@ class CompactionJob {

  EventLogger* event_logger_;

+  // Is this compaction creating a file in the bottom most level?
  bool bottommost_level_;
  bool paranoid_file_checks_;
  bool measure_io_stats_;
@@ -172,6 +188,7 @@ class CompactionJob {
  // Stores the approx size of keys covered in the range of each subcompaction
  std::vector<uint64_t> sizes_;
  Env::WriteLifeTimeHint write_hint_;
+  Env::Priority thread_pri_;
 };

 }  // namespace rocksdb
@@ -7,11 +7,7 @@
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file. See the AUTHORS file for names of contributors.

-#ifndef __STDC_FORMAT_MACROS
-#define __STDC_FORMAT_MACROS
-#endif
-
-#include <inttypes.h>
+#include <cinttypes>
 #include <algorithm>
 #include <iostream>
 #include <mutex>
@@ -21,12 +17,14 @@
 #include <unordered_set>
 #include <utility>

-#include "db/db_impl.h"
+#include "db/db_impl/db_impl.h"
 #include "db/dbformat.h"
 #include "db/job_context.h"
 #include "db/version_set.h"
 #include "db/write_batch_internal.h"
 #include "env/mock_env.h"
+#include "file/filename.h"
+#include "logging/logging.h"
 #include "memtable/hash_linklist_rep.h"
 #include "monitoring/statistics.h"
 #include "monitoring/thread_status_util.h"
@@ -47,20 +45,18 @@
 #include "rocksdb/thread_status.h"
 #include "rocksdb/utilities/checkpoint.h"
 #include "rocksdb/utilities/write_batch_with_index.h"
-#include "table/block_based_table_factory.h"
+#include "table/block_based/block_based_table_factory.h"
 #include "table/mock_table.h"
-#include "table/plain_table_factory.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"
 #include "util/compression.h"
-#include "util/filename.h"
 #include "util/hash.h"
-#include "util/logging.h"
 #include "util/mutexlock.h"
 #include "util/rate_limiter.h"
 #include "util/string_util.h"
-#include "util/sync_point.h"
-#include "util/testharness.h"
-#include "util/testutil.h"
 #include "utilities/merge_operators.h"

 #if !defined(IOS_CROSS_COMPILE)
@@ -113,7 +109,7 @@ class CompactionJobStatsTest : public testing::Test,
    Reopen(options);
  }

-  ~CompactionJobStatsTest() {
+  ~CompactionJobStatsTest() override {
    rocksdb::SyncPoint::GetInstance()->DisableProcessing();
    rocksdb::SyncPoint::GetInstance()->LoadDependency({});
    rocksdb::SyncPoint::GetInstance()->ClearAllCallBacks();
@@ -426,7 +422,7 @@ class CompactionJobStatsChecker : public EventListener {
  // Once a compaction completed, this function will verify the returned
  // CompactionJobInfo with the oldest CompactionJobInfo added earlier
  // in "expected_stats_" which has not yet being used for verification.
-  virtual void OnCompactionCompleted(DB* /*db*/, const CompactionJobInfo& ci) {
+  void OnCompactionCompleted(DB* /*db*/, const CompactionJobInfo& ci) override {
    if (verify_next_comp_io_stats_) {
      ASSERT_GT(ci.stats.file_write_nanos, 0);
      ASSERT_GT(ci.stats.file_range_sync_nanos, 0);
@@ -523,7 +519,7 @@ class CompactionJobDeletionStatsChecker : public CompactionJobStatsChecker {
 public:
  // Verifies whether two CompactionJobStats match.
  void Verify(const CompactionJobStats& current_stats,
-              const CompactionJobStats& stats) {
+              const CompactionJobStats& stats) override {
    ASSERT_EQ(
      current_stats.num_input_deletion_records,
      stats.num_input_deletion_records);
@@ -5,13 +5,15 @@

 #ifndef ROCKSDB_LITE

+#include <cinttypes>
 #include <algorithm>
+#include <array>
 #include <map>
 #include <string>
 #include <tuple>

 #include "db/column_family.h"
-#include "db/compaction_job.h"
+#include "db/compaction/compaction_job.h"
 #include "db/error_handler.h"
 #include "db/version_set.h"
 #include "rocksdb/cache.h"
@@ -19,10 +21,10 @@
 #include "rocksdb/options.h"
 #include "rocksdb/write_buffer_manager.h"
 #include "table/mock_table.h"
+#include "test_util/testharness.h"
+#include "test_util/testutil.h"
 #include "util/file_reader_writer.h"
 #include "util/string_util.h"
-#include "util/testharness.h"
-#include "util/testutil.h"
 #include "utilities/merge_operators.h"

 namespace rocksdb {
@@ -75,7 +77,8 @@ class CompactionJobTest : public testing::Test {
        write_buffer_manager_(db_options_.db_write_buffer_size),
        versions_(new VersionSet(dbname_, &db_options_, env_options_,
                                 table_cache_.get(), &write_buffer_manager_,
-                                 &write_controller_)),
+                                 &write_controller_,
+                                 /*block_cache_tracer=*/nullptr)),
        shutting_down_(false),
        preserve_deletes_seqnum_(0),
        mock_table_factory_(new mock::MockTableFactory()),
@@ -172,7 +175,7 @@ class CompactionJobTest : public testing::Test {
        // This is how the key will look like once it's written in bottommost
        // file
        InternalKey bottommost_internal_key(
-            key, (key == "9999") ? sequence_number : 0, kTypeValue);
+            key, 0, kTypeValue);

        if (corrupt_id(k)) {
          test::CorruptKeyType(&internal_key);
@@ -194,6 +197,14 @@ class CompactionJobTest : public testing::Test {
  }

  void NewDB() {
+    DestroyDB(dbname_, Options());
+    EXPECT_OK(env_->CreateDirIfMissing(dbname_));
+    versions_.reset(new VersionSet(dbname_, &db_options_, env_options_,
+                                   table_cache_.get(), &write_buffer_manager_,
+                                   &write_controller_,
+                                   /*block_cache_tracer=*/nullptr));
+    compaction_job_stats_.Reset();
+
    VersionEdit new_db;
    new_db.SetLogNumber(0);
    new_db.SetNextFile(2);
@@ -230,7 +241,10 @@ class CompactionJobTest : public testing::Test {
      const std::vector<std::vector<FileMetaData*>>& input_files,
      const stl_wrappers::KVMap& expected_results,
      const std::vector<SequenceNumber>& snapshots = {},
-      SequenceNumber earliest_write_conflict_snapshot = kMaxSequenceNumber) {
+      SequenceNumber earliest_write_conflict_snapshot = kMaxSequenceNumber,
+      int output_level = 1, bool verify = true,
+      SnapshotListFetchCallback* snapshot_fetcher =
+          SnapshotListFetchCallback::kDisabled) {
    auto cfd = versions_->GetColumnFamilySet()->GetDefault();

    size_t num_input_files = 0;
@@ -247,7 +261,7 @@ class CompactionJobTest : public testing::Test {

    Compaction compaction(cfd->current()->storage_info(), *cfd->ioptions(),
                          *cfd->GetLatestMutableCFOptions(),
-                          compaction_input_files, 1, 1024 * 1024,
+                          compaction_input_files, output_level, 1024 * 1024,
                          10 * 1024 * 1024, 0, kNoCompression,
                          cfd->ioptions()->compression_opts, 0, {}, true);
    compaction.SetInputVersion(cfd->current());
@@ -262,7 +276,8 @@ class CompactionJobTest : public testing::Test {
        &shutting_down_, preserve_deletes_seqnum_, &log_buffer, nullptr,
        nullptr, nullptr, &mutex_, &error_handler_, snapshots,
        earliest_write_conflict_snapshot, snapshot_checker, table_cache_,
-        &event_logger, false, false, dbname_, &compaction_job_stats_);
+        &event_logger, false, false, dbname_, &compaction_job_stats_,
+        Env::Priority::USER, snapshot_fetcher);
    VerifyInitializationOfCompactionJobStats(compaction_job_stats_);

    compaction_job.Prepare();
@@ -274,15 +289,17 @@ class CompactionJobTest : public testing::Test {
    ASSERT_OK(compaction_job.Install(*cfd->GetLatestMutableCFOptions()));
    mutex_.Unlock();

-    if (expected_results.size() == 0) {
-      ASSERT_GE(compaction_job_stats_.elapsed_micros, 0U);
-      ASSERT_EQ(compaction_job_stats_.num_input_files, num_input_files);
-      ASSERT_EQ(compaction_job_stats_.num_output_files, 0U);
-    } else {
-      ASSERT_GE(compaction_job_stats_.elapsed_micros, 0U);
-      ASSERT_EQ(compaction_job_stats_.num_input_files, num_input_files);
-      ASSERT_EQ(compaction_job_stats_.num_output_files, 1U);
-      mock_table_factory_->AssertLatestFile(expected_results);
+    if (verify) {
+      if (expected_results.size() == 0) {
+        ASSERT_GE(compaction_job_stats_.elapsed_micros, 0U);
+        ASSERT_EQ(compaction_job_stats_.num_input_files, num_input_files);
+        ASSERT_EQ(compaction_job_stats_.num_output_files, 0U);
+      } else {
+        ASSERT_GE(compaction_job_stats_.elapsed_micros, 0U);
+        ASSERT_EQ(compaction_job_stats_.num_input_files, num_input_files);
+        ASSERT_EQ(compaction_job_stats_.num_output_files, 1U);
+        mock_table_factory_->AssertLatestFile(expected_results);
+      }
    }
  }

@@ -379,7 +396,7 @@ TEST_F(CompactionJobTest, SimpleOverwrite) {

  auto expected_results =
      mock::MakeMockFile({{KeyStr("a", 0U, kTypeValue), "val2"},
-                          {KeyStr("b", 4U, kTypeValue), "val3"}});
+                          {KeyStr("b", 0U, kTypeValue), "val3"}});

  SetLastSequence(4U);
  auto files = cfd_->current()->storage_info()->LevelFiles(0);
@@ -432,7 +449,7 @@ TEST_F(CompactionJobTest, SimpleMerge) {

  auto expected_results =
      mock::MakeMockFile({{KeyStr("a", 0U, kTypeValue), "3,4,5"},
-                          {KeyStr("b", 2U, kTypeValue), "1,2"}});
+                          {KeyStr("b", 0U, kTypeValue), "1,2"}});

  SetLastSequence(5U);
  auto files = cfd_->current()->storage_info()->LevelFiles(0);
@@ -456,7 +473,7 @@ TEST_F(CompactionJobTest, NonAssocMerge) {

  auto expected_results =
      mock::MakeMockFile({{KeyStr("a", 0U, kTypeValue), "3,4,5"},
-                          {KeyStr("b", 2U, kTypeValue), "1,2"}});
+                          {KeyStr("b", 0U, kTypeValue), "1,2"}});

  SetLastSequence(5U);
  auto files = cfd_->current()->storage_info()->LevelFiles(0);
@@ -483,7 +500,7 @@ TEST_F(CompactionJobTest, MergeOperandFilter) {

  auto expected_results =
      mock::MakeMockFile({{KeyStr("a", 0U, kTypeValue), test::EncodeInt(8U)},
-                          {KeyStr("b", 2U, kTypeValue), test::EncodeInt(2U)}});
+                          {KeyStr("b", 0U, kTypeValue), test::EncodeInt(2U)}});

  SetLastSequence(5U);
  auto files = cfd_->current()->storage_info()->LevelFiles(0);
@@ -746,7 +763,7 @@ TEST_F(CompactionJobTest, SingleDeleteZeroSeq) {
  AddMockFile(file2);

  auto expected_results = mock::MakeMockFile({
-      {KeyStr("dummy", 5U, kTypeValue), "val2"},
+      {KeyStr("dummy", 0U, kTypeValue), "val2"},
  });

  SetLastSequence(22U);
@@ -930,13 +947,112 @@ TEST_F(CompactionJobTest, CorruptionAfterDeletion) {
      mock::MakeMockFile({{test::KeyStr("A", 0U, kTypeValue), "val3"},
                          {test::KeyStr("a", 0U, kTypeValue, true), "val"},
                          {test::KeyStr("b", 0U, kTypeValue, true), "val"},
-                          {test::KeyStr("c", 1U, kTypeValue), "val2"}});
+                          {test::KeyStr("c", 0U, kTypeValue), "val2"}});

  SetLastSequence(6U);
  auto files = cfd_->current()->storage_info()->LevelFiles(0);
  RunCompaction({files}, expected_results);
 }

+// Test the snapshot fetcher in compaction
+TEST_F(CompactionJobTest, SnapshotRefresh) {
+  uint64_t time_seed = env_->NowMicros();
+  printf("time_seed is %" PRIu64 "\n", time_seed);  // would help to reproduce
+  Random64 rand(time_seed);
+  std::vector<SequenceNumber> db_snapshots;
+  class SnapshotListFetchCallbackTest : public SnapshotListFetchCallback {
+   public:
+    SnapshotListFetchCallbackTest(Env* env, Random64& rand,
+                                  std::vector<SequenceNumber>* snapshots)
+        : SnapshotListFetchCallback(env, 0 /*no time delay*/,
+                                    1 /*fetch after each key*/),
+          rand_(rand),
+          snapshots_(snapshots) {}
+    virtual void Refresh(std::vector<SequenceNumber>* snapshots,
+                         SequenceNumber) override {
+      assert(snapshots->size());
+      assert(snapshots_->size());
+      assert(snapshots_->size() == snapshots->size());
+      if (rand_.OneIn(2)) {
+        uint64_t release_index = rand_.Uniform(snapshots_->size());
+        snapshots_->erase(snapshots_->begin() + release_index);
+        *snapshots = *snapshots_;
+      }
+    }
+
+   private:
+    Random64 rand_;
+    std::vector<SequenceNumber>* snapshots_;
+  } snapshot_fetcher(env_, rand, &db_snapshots);
+
+  std::vector<std::pair<const std::string, std::string>> file1_kvs, file2_kvs;
+  std::array<ValueType, 4> types = {kTypeValue, kTypeDeletion,
+                                    kTypeSingleDeletion};
+  SequenceNumber last_seq = 0;
+  for (int i = 1; i < 100; i++) {
+    SequenceNumber seq = last_seq + 1;
+    last_seq = seq;
+    if (rand.OneIn(2)) {
+      auto type = types[rand.Uniform(types.size())];
+      file1_kvs.push_back(
+          {test::KeyStr("k" + ToString(i), seq, type), "v" + ToString(i)});
+    }
+  }
+  auto file1 = mock::MakeMockFile(file1_kvs);
+  for (int i = 1; i < 100; i++) {
+    SequenceNumber seq = last_seq + 1;
+    last_seq++;
+    if (rand.OneIn(2)) {
+      auto type = types[rand.Uniform(types.size())];
+      file2_kvs.push_back(
+          {test::KeyStr("k" + ToString(i), seq, type), "v" + ToString(i)});
+    }
+  }
+  auto file2 = mock::MakeMockFile(file2_kvs);
+  for (SequenceNumber i = 1; i < last_seq + 1; i++) {
+    if (rand.OneIn(5)) {
+      db_snapshots.push_back(i);
+    }
+  }
+
+  const bool kVerify = true;
+  const int output_level_0 = 0;
+  NewDB();
+  AddMockFile(file1);
+  AddMockFile(file2);
+  SetLastSequence(last_seq);
+  auto files = cfd_->current()->storage_info()->LevelFiles(0);
+  // put the output on L0 since it is easier to feed them again to the 2nd
+  // compaction
+  RunCompaction({files}, file1, db_snapshots, kMaxSequenceNumber,
+                output_level_0, !kVerify, &snapshot_fetcher);
+
+  // Now db_snapshots are changed. Run the compaction again without snapshot
+  // fetcher but with the updated snapshot list.
+  compaction_job_stats_.Reset();
+  files = cfd_->current()->storage_info()->LevelFiles(0);
+  RunCompaction({files}, file1, db_snapshots, kMaxSequenceNumber,
+                output_level_0 + 1, !kVerify);
+  // The result should be what we get if we run compaction without snapshot
+  // fetcher on the updated list of snapshots
+  auto expected = mock_table_factory_->output();
+
+  NewDB();
+  AddMockFile(file1);
+  AddMockFile(file2);
+  SetLastSequence(last_seq);
+  files = cfd_->current()->storage_info()->LevelFiles(0);
+  RunCompaction({files}, expected, db_snapshots, kMaxSequenceNumber,
+                output_level_0, !kVerify);
+  // The 2nd compaction above would get rid of useless delete markers. To get
+  // the output here exactly as what we got above after two compactions, we also
+  // run the compaction for 2nd time.
+  compaction_job_stats_.Reset();
+  files = cfd_->current()->storage_info()->LevelFiles(0);
+  RunCompaction({files}, expected, db_snapshots, kMaxSequenceNumber,
+                output_level_0 + 1, !kVerify);
+}
+
 }  // namespace rocksdb

 int main(int argc, char** argv) {
@@ -7,25 +7,21 @@
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file. See the AUTHORS file for names of contributors.

-#include "db/compaction_picker.h"
+#include "db/compaction/compaction_picker.h"

-#ifndef __STDC_FORMAT_MACROS
-#define __STDC_FORMAT_MACROS
-#endif
-
-#include <inttypes.h>
+#include <cinttypes>
 #include <limits>
 #include <queue>
 #include <string>
 #include <utility>
 #include <vector>
 #include "db/column_family.h"
+#include "file/filename.h"
+#include "logging/log_buffer.h"
 #include "monitoring/statistics.h"
-#include "util/filename.h"
-#include "util/log_buffer.h"
+#include "test_util/sync_point.h"
 #include "util/random.h"
 #include "util/string_util.h"
-#include "util/sync_point.h"

 namespace rocksdb {

@@ -42,19 +38,23 @@ uint64_t TotalCompensatedFileSize(const std::vector<FileMetaData*>& files) {
 bool FindIntraL0Compaction(const std::vector<FileMetaData*>& level_files,
                           size_t min_files_to_compact,
                           uint64_t max_compact_bytes_per_del_file,
+                           uint64_t max_compaction_bytes,
                           CompactionInputFiles* comp_inputs) {
  size_t compact_bytes = static_cast<size_t>(level_files[0]->fd.file_size);
+  uint64_t compensated_compact_bytes = level_files[0]->compensated_file_size;
  size_t compact_bytes_per_del_file = port::kMaxSizet;
-  // compaction range will be [0, span_len).
+  // Compaction range will be [0, span_len).
  size_t span_len;
-  // pull in files until the amount of compaction work per deleted file begins
-  // increasing.
+  // Pull in files until the amount of compaction work per deleted file begins
+  // increasing or maximum total compaction size is reached.
  size_t new_compact_bytes_per_del_file = 0;
  for (span_len = 1; span_len < level_files.size(); ++span_len) {
    compact_bytes += static_cast<size_t>(level_files[span_len]->fd.file_size);
+    compensated_compact_bytes += level_files[span_len]->compensated_file_size;
    new_compact_bytes_per_del_file = compact_bytes / span_len;
    if (level_files[span_len]->being_compacted ||
-        new_compact_bytes_per_del_file > compact_bytes_per_del_file) {
+        new_compact_bytes_per_del_file > compact_bytes_per_del_file ||
+        compensated_compact_bytes > max_compaction_bytes) {
      break;
    }
    compact_bytes_per_del_file = new_compact_bytes_per_del_file;
@@ -543,9 +543,9 @@ void CompactionPicker::GetGrandparents(
 Compaction* CompactionPicker::CompactRange(
    const std::string& cf_name, const MutableCFOptions& mutable_cf_options,
    VersionStorageInfo* vstorage, int input_level, int output_level,
-    uint32_t output_path_id, uint32_t max_subcompactions,
-    const InternalKey* begin, const InternalKey* end,
-    InternalKey** compaction_end, bool* manual_conflict) {
+    const CompactRangeOptions& compact_range_options, const InternalKey* begin,
+    const InternalKey* end, InternalKey** compaction_end, bool* manual_conflict,
+    uint64_t max_file_num_to_ignore) {
  // CompactionPickerFIFO has its own implementation of compact range
  assert(ioptions_.compaction_style != kCompactionStyleFIFO);

@@ -609,11 +609,13 @@ Compaction* CompactionPicker::CompactRange(
        output_level,
        MaxFileSizeForLevel(mutable_cf_options, output_level,
                            ioptions_.compaction_style),
-        /* max_compaction_bytes */ LLONG_MAX, output_path_id,
+        /* max_compaction_bytes */ LLONG_MAX,
+        compact_range_options.target_path_id,
        GetCompressionType(ioptions_, vstorage, mutable_cf_options,
                           output_level, 1),
        GetCompressionOptions(ioptions_, vstorage, output_level),
-        max_subcompactions, /* grandparents */ {}, /* is manual */ true);
+        compact_range_options.max_subcompactions, /* grandparents */ {},
+        /* is manual */ true);
    RegisterCompaction(c);
    return c;
  }
@@ -658,7 +660,43 @@ Compaction* CompactionPicker::CompactRange(
      }
    }
  }
-  assert(output_path_id < static_cast<uint32_t>(ioptions_.cf_paths.size()));
+  assert(compact_range_options.target_path_id <
+         static_cast<uint32_t>(ioptions_.cf_paths.size()));
+
+  // for BOTTOM LEVEL compaction only, use max_file_num_to_ignore to filter out
+  // files that are created during the current compaction.
+  if (compact_range_options.bottommost_level_compaction ==
+          BottommostLevelCompaction::kForceOptimized &&
+      max_file_num_to_ignore != port::kMaxUint64) {
+    assert(input_level == output_level);
+    // inputs_shrunk holds a continuous subset of input files which were all
+    // created before the current manual compaction
+    std::vector<FileMetaData*> inputs_shrunk;
+    size_t skip_input_index = inputs.size();
+    for (size_t i = 0; i < inputs.size(); ++i) {
+      if (inputs[i]->fd.GetNumber() < max_file_num_to_ignore) {
+        inputs_shrunk.push_back(inputs[i]);
+      } else if (!inputs_shrunk.empty()) {
+        // inputs[i] was created during the current manual compaction and
+        // need to be skipped
+        skip_input_index = i;
+        break;
+      }
+    }
+    if (inputs_shrunk.empty()) {
+      return nullptr;
+    }
+    if (inputs.size() != inputs_shrunk.size()) {
+      inputs.files.swap(inputs_shrunk);
+    }
+    // set covering_the_whole_range to false if there is any file that need to
+    // be compacted in the range of inputs[skip_input_index+1, inputs.size())
+    for (size_t i = skip_input_index + 1; i < inputs.size(); ++i) {
+      if (inputs[i]->fd.GetNumber() < max_file_num_to_ignore) {
+        covering_the_whole_range = false;
+      }
+    }
+  }

  InternalKey key_storage;
  InternalKey* next_smallest = &key_storage;
@@ -724,11 +762,12 @@ Compaction* CompactionPicker::CompactRange(
      MaxFileSizeForLevel(mutable_cf_options, output_level,
                          ioptions_.compaction_style, vstorage->base_level(),
                          ioptions_.level_compaction_dynamic_level_bytes),
-      mutable_cf_options.max_compaction_bytes, output_path_id,
+      mutable_cf_options.max_compaction_bytes,
+      compact_range_options.target_path_id,
      GetCompressionType(ioptions_, vstorage, mutable_cf_options, output_level,
                         vstorage->base_level()),
      GetCompressionOptions(ioptions_, vstorage, output_level),
-      max_subcompactions, std::move(grandparents),
+      compact_range_options.max_subcompactions, std::move(grandparents),
      /* is manual compaction */ true);

  TEST_SYNC_POINT_CALLBACK("CompactionPicker::CompactRange:Return", compaction);
@@ -1068,484 +1107,4 @@ bool CompactionPicker::GetOverlappingL0Files(
  return true;
 }

-bool LevelCompactionPicker::NeedsCompaction(
-    const VersionStorageInfo* vstorage) const {
-  if (!vstorage->ExpiredTtlFiles().empty()) {
-    return true;
-  }
-  if (!vstorage->BottommostFilesMarkedForCompaction().empty()) {
-    return true;
-  }
-  if (!vstorage->FilesMarkedForCompaction().empty()) {
-    return true;
-  }
-  for (int i = 0; i <= vstorage->MaxInputLevel(); i++) {
-    if (vstorage->CompactionScore(i) >= 1) {
-      return true;
-    }
-  }
-  return false;
-}
-
-namespace {
-// A class to build a leveled compaction step-by-step.
-class LevelCompactionBuilder {
- public:
-  LevelCompactionBuilder(const std::string& cf_name,
-                         VersionStorageInfo* vstorage,
-                         CompactionPicker* compaction_picker,
-                         LogBuffer* log_buffer,
-                         const MutableCFOptions& mutable_cf_options,
-                         const ImmutableCFOptions& ioptions)
-      : cf_name_(cf_name),
-        vstorage_(vstorage),
-        compaction_picker_(compaction_picker),
-        log_buffer_(log_buffer),
-        mutable_cf_options_(mutable_cf_options),
-        ioptions_(ioptions) {}
-
-  // Pick and return a compaction.
-  Compaction* PickCompaction();
-
-  // Pick the initial files to compact to the next level. (or together
-  // in Intra-L0 compactions)
-  void SetupInitialFiles();
-
-  // If the initial files are from L0 level, pick other L0
-  // files if needed.
-  bool SetupOtherL0FilesIfNeeded();
-
-  // Based on initial files, setup other files need to be compacted
-  // in this compaction, accordingly.
-  bool SetupOtherInputsIfNeeded();
-
-  Compaction* GetCompaction();
-
-  // For the specfied level, pick a file that we want to compact.
-  // Returns false if there is no file to compact.
-  // If it returns true, inputs->files.size() will be exactly one.
-  // If level is 0 and there is already a compaction on that level, this
-  // function will return false.
-  bool PickFileToCompact();
-
-  // For L0->L0, picks the longest span of files that aren't currently
-  // undergoing compaction for which work-per-deleted-file decreases. The span
-  // always starts from the newest L0 file.
-  //
-  // Intra-L0 compaction is independent of all other files, so it can be
-  // performed even when L0->base_level compactions are blocked.
-  //
-  // Returns true if `inputs` is populated with a span of files to be compacted;
-  // otherwise, returns false.
-  bool PickIntraL0Compaction();
-
-  void PickExpiredTtlFiles();
-
-  const std::string& cf_name_;
-  VersionStorageInfo* vstorage_;
-  CompactionPicker* compaction_picker_;
-  LogBuffer* log_buffer_;
-  int start_level_ = -1;
-  int output_level_ = -1;
-  int parent_index_ = -1;
-  int base_index_ = -1;
-  double start_level_score_ = 0;
-  bool is_manual_ = false;
-  CompactionInputFiles start_level_inputs_;
-  std::vector<CompactionInputFiles> compaction_inputs_;
-  CompactionInputFiles output_level_inputs_;
-  std::vector<FileMetaData*> grandparents_;
-  CompactionReason compaction_reason_ = CompactionReason::kUnknown;
-
-  const MutableCFOptions& mutable_cf_options_;
-  const ImmutableCFOptions& ioptions_;
-  // Pick a path ID to place a newly generated file, with its level
-  static uint32_t GetPathId(const ImmutableCFOptions& ioptions,
-                            const MutableCFOptions& mutable_cf_options,
-                            int level);
-
-  static const int kMinFilesForIntraL0Compaction = 4;
-};
-
-void LevelCompactionBuilder::PickExpiredTtlFiles() {
-  if (vstorage_->ExpiredTtlFiles().empty()) {
-    return;
-  }
-
-  auto continuation = [&](std::pair<int, FileMetaData*> level_file) {
-    // If it's being compacted it has nothing to do here.
-    // If this assert() fails that means that some function marked some
-    // files as being_compacted, but didn't call ComputeCompactionScore()
-    assert(!level_file.second->being_compacted);
-    start_level_ = level_file.first;
-    output_level_ =
-        (start_level_ == 0) ? vstorage_->base_level() : start_level_ + 1;
-
-    if ((start_level_ == vstorage_->num_non_empty_levels() - 1) ||
-        (start_level_ == 0 &&
-         !compaction_picker_->level0_compactions_in_progress()->empty())) {
-      return false;
-    }
-
-    start_level_inputs_.files = {level_file.second};
-    start_level_inputs_.level = start_level_;
-    return compaction_picker_->ExpandInputsToCleanCut(cf_name_, vstorage_,
-                                                      &start_level_inputs_);
-  };
-
-  for (auto& level_file : vstorage_->ExpiredTtlFiles()) {
-    if (continuation(level_file)) {
-      // found the compaction!
-      return;
-    }
-  }
-
-  start_level_inputs_.files.clear();
-}
-
-void LevelCompactionBuilder::SetupInitialFiles() {
-  // Find the compactions by size on all levels.
-  bool skipped_l0_to_base = false;
-  for (int i = 0; i < compaction_picker_->NumberLevels() - 1; i++) {
-    start_level_score_ = vstorage_->CompactionScore(i);
-    start_level_ = vstorage_->CompactionScoreLevel(i);
-    assert(i == 0 || start_level_score_ <= vstorage_->CompactionScore(i - 1));
-    if (start_level_score_ >= 1) {
-      if (skipped_l0_to_base && start_level_ == vstorage_->base_level()) {
-        // If L0->base_level compaction is pending, don't schedule further
-        // compaction from base level. Otherwise L0->base_level compaction
-        // may starve.
-        continue;
-      }
-      output_level_ =
-          (start_level_ == 0) ? vstorage_->base_level() : start_level_ + 1;
-      if (PickFileToCompact()) {
-        // found the compaction!
-        if (start_level_ == 0) {
-          // L0 score = `num L0 files` / `level0_file_num_compaction_trigger`
-          compaction_reason_ = CompactionReason::kLevelL0FilesNum;
-        } else {
-          // L1+ score = `Level files size` / `MaxBytesForLevel`
-          compaction_reason_ = CompactionReason::kLevelMaxLevelSize;
-        }
-        break;
-      } else {
-        // didn't find the compaction, clear the inputs
-        start_level_inputs_.clear();
-        if (start_level_ == 0) {
-          skipped_l0_to_base = true;
-          // L0->base_level may be blocked due to ongoing L0->base_level
-          // compactions. It may also be blocked by an ongoing compaction from
-          // base_level downwards.
-          //
-          // In these cases, to reduce L0 file count and thus reduce likelihood
-          // of write stalls, we can attempt compacting a span of files within
-          // L0.
-          if (PickIntraL0Compaction()) {
-            output_level_ = 0;
-            compaction_reason_ = CompactionReason::kLevelL0FilesNum;
-            break;
-          }
-        }
-      }
-    }
-  }
-
-  // if we didn't find a compaction, check if there are any files marked for
-  // compaction
-  if (start_level_inputs_.empty()) {
-    parent_index_ = base_index_ = -1;
-
-    // PickFilesMarkedForCompaction();
-    compaction_picker_->PickFilesMarkedForCompaction(
-        cf_name_, vstorage_, &start_level_, &output_level_, &start_level_inputs_);
-    if (!start_level_inputs_.empty()) {
-      is_manual_ = true;
-      compaction_reason_ = CompactionReason::kFilesMarkedForCompaction;
-      return;
-    }
-
-    size_t i;
-    for (i = 0; i < vstorage_->BottommostFilesMarkedForCompaction().size();
-         ++i) {
-      auto& level_and_file = vstorage_->BottommostFilesMarkedForCompaction()[i];
-      assert(!level_and_file.second->being_compacted);
-      start_level_inputs_.level = output_level_ = start_level_ =
-          level_and_file.first;
-      start_level_inputs_.files = {level_and_file.second};
-      if (compaction_picker_->ExpandInputsToCleanCut(cf_name_, vstorage_,
-                                                     &start_level_inputs_)) {
-        break;
-      }
-    }
-    if (i == vstorage_->BottommostFilesMarkedForCompaction().size()) {
-      start_level_inputs_.clear();
-    } else {
-      assert(!start_level_inputs_.empty());
-      compaction_reason_ = CompactionReason::kBottommostFiles;
-      return;
-    }
-
-    assert(start_level_inputs_.empty());
-    PickExpiredTtlFiles();
-    if (!start_level_inputs_.empty()) {
-      compaction_reason_ = CompactionReason::kTtl;
-    }
-  }
-}
-
-bool LevelCompactionBuilder::SetupOtherL0FilesIfNeeded() {
-  if (start_level_ == 0 && output_level_ != 0) {
-    return compaction_picker_->GetOverlappingL0Files(
-        vstorage_, &start_level_inputs_, output_level_, &parent_index_);
-  }
-  return true;
-}
-
-bool LevelCompactionBuilder::SetupOtherInputsIfNeeded() {
-  // Setup input files from output level. For output to L0, we only compact
-  // spans of files that do not interact with any pending compactions, so don't
-  // need to consider other levels.
-  if (output_level_ != 0) {
-    output_level_inputs_.level = output_level_;
-    if (!compaction_picker_->SetupOtherInputs(
-            cf_name_, mutable_cf_options_, vstorage_, &start_level_inputs_,
-            &output_level_inputs_, &parent_index_, base_index_)) {
-      return false;
-    }
-
-    compaction_inputs_.push_back(start_level_inputs_);
-    if (!output_level_inputs_.empty()) {
-      compaction_inputs_.push_back(output_level_inputs_);
-    }
-
-    // In some edge cases we could pick a compaction that will be compacting
-    // a key range that overlap with another running compaction, and both
-    // of them have the same output level. This could happen if
-    // (1) we are running a non-exclusive manual compaction
-    // (2) AddFile ingest a new file into the LSM tree
-    // We need to disallow this from happening.
-    if (compaction_picker_->FilesRangeOverlapWithCompaction(compaction_inputs_,
-                                                            output_level_)) {
-      // This compaction output could potentially conflict with the output
-      // of a currently running compaction, we cannot run it.
-      return false;
-    }
-    compaction_picker_->GetGrandparents(vstorage_, start_level_inputs_,
-                                        output_level_inputs_, &grandparents_);
-  } else {
-    compaction_inputs_.push_back(start_level_inputs_);
-  }
-  return true;
-}
-
-Compaction* LevelCompactionBuilder::PickCompaction() {
-  // Pick up the first file to start compaction. It may have been extended
-  // to a clean cut.
-  SetupInitialFiles();
-  if (start_level_inputs_.empty()) {
-    return nullptr;
-  }
-  assert(start_level_ >= 0 && output_level_ >= 0);
-
-  // If it is a L0 -> base level compaction, we need to set up other L0
-  // files if needed.
-  if (!SetupOtherL0FilesIfNeeded()) {
-    return nullptr;
-  }
-
-  // Pick files in the output level and expand more files in the start level
-  // if needed.
-  if (!SetupOtherInputsIfNeeded()) {
-    return nullptr;
-  }
-
-  // Form a compaction object containing the files we picked.
-  Compaction* c = GetCompaction();
-
-  TEST_SYNC_POINT_CALLBACK("LevelCompactionPicker::PickCompaction:Return", c);
-
-  return c;
-}
-
-Compaction* LevelCompactionBuilder::GetCompaction() {
-  auto c = new Compaction(
-      vstorage_, ioptions_, mutable_cf_options_, std::move(compaction_inputs_),
-      output_level_,
-      MaxFileSizeForLevel(mutable_cf_options_, output_level_,
-                          ioptions_.compaction_style, vstorage_->base_level(),
-                          ioptions_.level_compaction_dynamic_level_bytes),
-      mutable_cf_options_.max_compaction_bytes,
-      GetPathId(ioptions_, mutable_cf_options_, output_level_),
-      GetCompressionType(ioptions_, vstorage_, mutable_cf_options_,
-                         output_level_, vstorage_->base_level()),
-      GetCompressionOptions(ioptions_, vstorage_, output_level_),
-      /* max_subcompactions */ 0, std::move(grandparents_), is_manual_,
-      start_level_score_, false /* deletion_compaction */, compaction_reason_);
-
-  // If it's level 0 compaction, make sure we don't execute any other level 0
-  // compactions in parallel
-  compaction_picker_->RegisterCompaction(c);
-
-  // Creating a compaction influences the compaction score because the score
-  // takes running compactions into account (by skipping files that are already
-  // being compacted). Since we just changed compaction score, we recalculate it
-  // here
-  vstorage_->ComputeCompactionScore(ioptions_, mutable_cf_options_);
-  return c;
-}
-
-/*
- * Find the optimal path to place a file
- * Given a level, finds the path where levels up to it will fit in levels
- * up to and including this path
- */
-uint32_t LevelCompactionBuilder::GetPathId(
-    const ImmutableCFOptions& ioptions,
-    const MutableCFOptions& mutable_cf_options, int level) {
-  uint32_t p = 0;
-  assert(!ioptions.cf_paths.empty());
-
-  // size remaining in the most recent path
-  uint64_t current_path_size = ioptions.cf_paths[0].target_size;
-
-  uint64_t level_size;
-  int cur_level = 0;
-
-  // max_bytes_for_level_base denotes L1 size.
-  // We estimate L0 size to be the same as L1.
-  level_size = mutable_cf_options.max_bytes_for_level_base;
-
-  // Last path is the fallback
-  while (p < ioptions.cf_paths.size() - 1) {
-    if (level_size <= current_path_size) {
-      if (cur_level == level) {
-        // Does desired level fit in this path?
-        return p;
-      } else {
-        current_path_size -= level_size;
-        if (cur_level > 0) {
-          if (ioptions.level_compaction_dynamic_level_bytes) {
-            // Currently, level_compaction_dynamic_level_bytes is ignored when
-            // multiple db paths are specified. https://github.com/facebook/
-            // rocksdb/blob/master/db/column_family.cc.
-            // Still, adding this check to avoid accidentally using
-            // max_bytes_for_level_multiplier_additional
-            level_size = static_cast<uint64_t>(
-                level_size * mutable_cf_options.max_bytes_for_level_multiplier);
-          } else {
-            level_size = static_cast<uint64_t>(
-                level_size * mutable_cf_options.max_bytes_for_level_multiplier *
-                mutable_cf_options.MaxBytesMultiplerAdditional(cur_level));
-          }
-        }
-        cur_level++;
-        continue;
-      }
-    }
-    p++;
-    current_path_size = ioptions.cf_paths[p].target_size;
-  }
-  return p;
-}
-
-bool LevelCompactionBuilder::PickFileToCompact() {
-  // level 0 files are overlapping. So we cannot pick more
-  // than one concurrent compactions at this level. This
-  // could be made better by looking at key-ranges that are
-  // being compacted at level 0.
-  if (start_level_ == 0 &&
-      !compaction_picker_->level0_compactions_in_progress()->empty()) {
-    TEST_SYNC_POINT("LevelCompactionPicker::PickCompactionBySize:0");
-    return false;
-  }
-
-  start_level_inputs_.clear();
-
-  assert(start_level_ >= 0);
-
-  // Pick the largest file in this level that is not already
-  // being compacted
-  const std::vector<int>& file_size =
-      vstorage_->FilesByCompactionPri(start_level_);
-  const std::vector<FileMetaData*>& level_files =
-      vstorage_->LevelFiles(start_level_);
-
-  unsigned int cmp_idx;
-  for (cmp_idx = vstorage_->NextCompactionIndex(start_level_);
-       cmp_idx < file_size.size(); cmp_idx++) {
-    int index = file_size[cmp_idx];
-    auto* f = level_files[index];
-
-    // do not pick a file to compact if it is being compacted
-    // from n-1 level.
-    if (f->being_compacted) {
-      continue;
-    }
-
-    start_level_inputs_.files.push_back(f);
-    start_level_inputs_.level = start_level_;
-    if (!compaction_picker_->ExpandInputsToCleanCut(cf_name_, vstorage_,
-                                                    &start_level_inputs_) ||
-        compaction_picker_->FilesRangeOverlapWithCompaction(
-            {start_level_inputs_}, output_level_)) {
-      // A locked (pending compaction) input-level file was pulled in due to
-      // user-key overlap.
-      start_level_inputs_.clear();
-      continue;
-    }
-
-    // Now that input level is fully expanded, we check whether any output files
-    // are locked due to pending compaction.
-    //
-    // Note we rely on ExpandInputsToCleanCut() to tell us whether any output-
-    // level files are locked, not just the extra ones pulled in for user-key
-    // overlap.
-    InternalKey smallest, largest;
-    compaction_picker_->GetRange(start_level_inputs_, &smallest, &largest);
-    CompactionInputFiles output_level_inputs;
-    output_level_inputs.level = output_level_;
-    vstorage_->GetOverlappingInputs(output_level_, &smallest, &largest,
-                                    &output_level_inputs.files);
-    if (!output_level_inputs.empty() &&
-        !compaction_picker_->ExpandInputsToCleanCut(cf_name_, vstorage_,
-                                                    &output_level_inputs)) {
-      start_level_inputs_.clear();
-      continue;
-    }
-    base_index_ = index;
-    break;
-  }
-
-  // store where to start the iteration in the next call to PickCompaction
-  vstorage_->SetNextCompactionIndex(start_level_, cmp_idx);
-
-  return start_level_inputs_.size() > 0;
-}
-
-bool LevelCompactionBuilder::PickIntraL0Compaction() {
-  start_level_inputs_.clear();
-  const std::vector<FileMetaData*>& level_files =
-      vstorage_->LevelFiles(0 /* level */);
-  if (level_files.size() <
-          static_cast<size_t>(
-              mutable_cf_options_.level0_file_num_compaction_trigger + 2) ||
-      level_files[0]->being_compacted) {
-    // If L0 isn't accumulating much files beyond the regular trigger, don't
-    // resort to L0->L0 compaction yet.
-    return false;
-  }
-  return FindIntraL0Compaction(level_files, kMinFilesForIntraL0Compaction,
-                               port::kMaxUint64, &start_level_inputs_);
-}
-}  // namespace
-
-Compaction* LevelCompactionPicker::PickCompaction(
-    const std::string& cf_name, const MutableCFOptions& mutable_cf_options,
-    VersionStorageInfo* vstorage, LogBuffer* log_buffer) {
-  LevelCompactionBuilder builder(cf_name, vstorage, this, log_buffer,
-                                 mutable_cf_options, ioptions_);
-  return builder.PickCompaction();
-}
-
 }  // namespace rocksdb
@@ -15,7 +15,7 @@
 #include <unordered_set>
 #include <vector>

-#include "db/compaction.h"
+#include "db/compaction/compaction.h"
 #include "db/version_set.h"
 #include "options/cf_options.h"
 #include "rocksdb/env.h"
@@ -24,11 +24,26 @@

 namespace rocksdb {

+// The file contains an abstract class CompactionPicker, and its two
+// sub-classes LevelCompactionPicker and NullCompactionPicker, as
+// well as some helper functions used by them.
+
 class LogBuffer;
 class Compaction;
 class VersionStorageInfo;
 struct CompactionInputFiles;

+// An abstract class to pick compactions from an existing LSM-tree.
+//
+// Each compaction style inherits the class and implement the
+// interface to form automatic compactions. If NeedCompaction() is true,
+// then call PickCompaction() to find what files need to be compacted
+// and where to put the output files.
+//
+// Non-virtual functions CompactRange() and CompactFiles() are used to
+// pick files to compact based on users' DB::CompactRange() and
+// DB::CompactFiles() requests, respectively. There is little
+// compaction style specific logic for them.
 class CompactionPicker {
 public:
  CompactionPicker(const ImmutableCFOptions& ioptions,
@@ -58,9 +73,10 @@ class CompactionPicker {
  virtual Compaction* CompactRange(
      const std::string& cf_name, const MutableCFOptions& mutable_cf_options,
      VersionStorageInfo* vstorage, int input_level, int output_level,
-      uint32_t output_path_id, uint32_t max_subcompactions,
+      const CompactRangeOptions& compact_range_options,
      const InternalKey* begin, const InternalKey* end,
-      InternalKey** compaction_end, bool* manual_conflict);
+      InternalKey** compaction_end, bool* manual_conflict,
+      uint64_t max_file_num_to_ignore);

  // The maximum allowed output level.  Default value is NumberLevels() - 1.
  virtual int MaxOutputLevel() const { return NumberLevels() - 1; }
@@ -220,21 +236,9 @@ class CompactionPicker {
  const InternalKeyComparator* const icmp_;
 };

-class LevelCompactionPicker : public CompactionPicker {
- public:
-  LevelCompactionPicker(const ImmutableCFOptions& ioptions,
-                        const InternalKeyComparator* icmp)
-      : CompactionPicker(ioptions, icmp) {}
-  virtual Compaction* PickCompaction(const std::string& cf_name,
-                                     const MutableCFOptions& mutable_cf_options,
-                                     VersionStorageInfo* vstorage,
-                                     LogBuffer* log_buffer) override;
-
-  virtual bool NeedsCompaction(
-      const VersionStorageInfo* vstorage) const override;
-};
-
 #ifndef ROCKSDB_LITE
+// A dummy compaction that never triggers any automatic
+// compaction.
 class NullCompactionPicker : public CompactionPicker {
 public:
  NullCompactionPicker(const ImmutableCFOptions& ioptions,
@@ -255,12 +259,12 @@ class NullCompactionPicker : public CompactionPicker {
                           const MutableCFOptions& /*mutable_cf_options*/,
                           VersionStorageInfo* /*vstorage*/,
                           int /*input_level*/, int /*output_level*/,
-                           uint32_t /*output_path_id*/,
-                           uint32_t /*max_subcompactions*/,
+                           const CompactRangeOptions& /*compact_range_options*/,
                           const InternalKey* /*begin*/,
                           const InternalKey* /*end*/,
                           InternalKey** /*compaction_end*/,
-                           bool* /*manual_conflict*/) override {
+                           bool* /*manual_conflict*/,
+                           uint64_t /*max_file_num_to_ignore*/) override {
    return nullptr;
  }

@@ -272,9 +276,26 @@ class NullCompactionPicker : public CompactionPicker {
 };
 #endif  // !ROCKSDB_LITE

+// Attempts to find an intra L0 compaction conforming to the given parameters.
+//
+// @param level_files                     Metadata for L0 files.
+// @param min_files_to_compact            Minimum number of files required to
+//                                        do the compaction.
+// @param max_compact_bytes_per_del_file  Maximum average size in bytes per
+//                                        file that is going to get deleted by
+//                                        the compaction.
+// @param max_compaction_bytes            Maximum total size in bytes (in terms
+//                                        of compensated file size) for files
+//                                        to be compacted.
+// @param [out] comp_inputs               If a compaction was found, will be
+//                                        initialized with corresponding input
+//                                        files. Cannot be nullptr.
+//
+// @return                                true iff compaction was found.
 bool FindIntraL0Compaction(const std::vector<FileMetaData*>& level_files,
                           size_t min_files_to_compact,
                           uint64_t max_compact_bytes_per_del_file,
+                           uint64_t max_compaction_bytes,
                           CompactionInputFiles* comp_inputs);

 CompressionType GetCompressionType(const ImmutableCFOptions& ioptions,
@@ -7,18 +7,14 @@
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file. See the AUTHORS file for names of contributors.

-#include "db/compaction_picker_fifo.h"
+#include "db/compaction/compaction_picker_fifo.h"
 #ifndef ROCKSDB_LITE

-#ifndef __STDC_FORMAT_MACROS
-#define __STDC_FORMAT_MACROS
-#endif
-
-#include <inttypes.h>
+#include <cinttypes>
 #include <string>
 #include <vector>
 #include "db/column_family.h"
-#include "util/log_buffer.h"
+#include "logging/log_buffer.h"
 #include "util/string_util.h"

 namespace rocksdb {
@@ -41,7 +37,7 @@ bool FIFOCompactionPicker::NeedsCompaction(
 Compaction* FIFOCompactionPicker::PickTTLCompaction(
    const std::string& cf_name, const MutableCFOptions& mutable_cf_options,
    VersionStorageInfo* vstorage, LogBuffer* log_buffer) {
-  assert(mutable_cf_options.compaction_options_fifo.ttl > 0);
+  assert(mutable_cf_options.ttl > 0);

  const int kLevel0 = 0;
  const std::vector<FileMetaData*>& level_files = vstorage->LevelFiles(kLevel0);
@@ -63,7 +59,7 @@ Compaction* FIFOCompactionPicker::PickTTLCompaction(
  inputs[0].level = 0;

  // avoid underflow
-  if (current_time > mutable_cf_options.compaction_options_fifo.ttl) {
+  if (current_time > mutable_cf_options.ttl) {
    for (auto ritr = level_files.rbegin(); ritr != level_files.rend(); ++ritr) {
      auto f = *ritr;
      if (f->fd.table_reader != nullptr &&
@@ -71,8 +67,7 @@ Compaction* FIFOCompactionPicker::PickTTLCompaction(
        auto creation_time =
            f->fd.table_reader->GetTableProperties()->creation_time;
        if (creation_time == 0 ||
-            creation_time >= (current_time -
-                              mutable_cf_options.compaction_options_fifo.ttl)) {
+            creation_time >= (current_time - mutable_cf_options.ttl)) {
          break;
        }
        total_size -= f->compensated_file_size;
@@ -135,7 +130,8 @@ Compaction* FIFOCompactionPicker::PickSizeCompaction(
              mutable_cf_options
                  .level0_file_num_compaction_trigger /* min_files_to_compact */
              ,
-              max_compact_bytes_per_del_file, &comp_inputs)) {
+              max_compact_bytes_per_del_file,
+              mutable_cf_options.max_compaction_bytes, &comp_inputs)) {
        Compaction* c = new Compaction(
            vstorage, ioptions_, mutable_cf_options, {comp_inputs}, 0,
            16 * 1024 * 1024 /* output file size limit */,
@@ -201,7 +197,7 @@ Compaction* FIFOCompactionPicker::PickCompaction(
  assert(vstorage->num_levels() == 1);

  Compaction* c = nullptr;
-  if (mutable_cf_options.compaction_options_fifo.ttl > 0) {
+  if (mutable_cf_options.ttl > 0) {
    c = PickTTLCompaction(cf_name, mutable_cf_options, vstorage, log_buffer);
  }
  if (c == nullptr) {
@@ -214,9 +210,10 @@ Compaction* FIFOCompactionPicker::PickCompaction(
 Compaction* FIFOCompactionPicker::CompactRange(
    const std::string& cf_name, const MutableCFOptions& mutable_cf_options,
    VersionStorageInfo* vstorage, int input_level, int output_level,
-    uint32_t /*output_path_id*/, uint32_t /*max_subcompactions*/,
+    const CompactRangeOptions& /*compact_range_options*/,
    const InternalKey* /*begin*/, const InternalKey* /*end*/,
-    InternalKey** compaction_end, bool* /*manual_conflict*/) {
+    InternalKey** compaction_end, bool* /*manual_conflict*/,
+    uint64_t /*max_file_num_to_ignore*/) {
 #ifdef NDEBUG
  (void)input_level;
  (void)output_level;
@@ -10,7 +10,7 @@
 #pragma once
 #ifndef ROCKSDB_LITE

-#include "db/compaction_picker.h"
+#include "db/compaction/compaction_picker.h"

 namespace rocksdb {
 class FIFOCompactionPicker : public CompactionPicker {
@@ -27,9 +27,10 @@ class FIFOCompactionPicker : public CompactionPicker {
  virtual Compaction* CompactRange(
      const std::string& cf_name, const MutableCFOptions& mutable_cf_options,
      VersionStorageInfo* vstorage, int input_level, int output_level,
-      uint32_t output_path_id, uint32_t max_subcompactions,
+      const CompactRangeOptions& compact_range_options,
      const InternalKey* begin, const InternalKey* end,
-      InternalKey** compaction_end, bool* manual_conflict) override;
+      InternalKey** compaction_end, bool* manual_conflict,
+      uint64_t max_file_num_to_ignore) override;

  // The maximum allowed output level.  Always returns 0.
  virtual int MaxOutputLevel() const override { return 0; }
@@ -0,0 +1,553 @@
+//  Copyright (c) 2011-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).
+//
+// Copyright (c) 2011 The LevelDB Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file. See the AUTHORS file for names of contributors.
+
+#include <string>
+#include <utility>
+#include <vector>
+
+#include "db/compaction/compaction_picker_level.h"
+#include "logging/log_buffer.h"
+#include "test_util/sync_point.h"
+
+namespace rocksdb {
+
+bool LevelCompactionPicker::NeedsCompaction(
+    const VersionStorageInfo* vstorage) const {
+  if (!vstorage->ExpiredTtlFiles().empty()) {
+    return true;
+  }
+  if (!vstorage->FilesMarkedForPeriodicCompaction().empty()) {
+    return true;
+  }
+  if (!vstorage->BottommostFilesMarkedForCompaction().empty()) {
+    return true;
+  }
+  if (!vstorage->FilesMarkedForCompaction().empty()) {
+    return true;
+  }
+  for (int i = 0; i <= vstorage->MaxInputLevel(); i++) {
+    if (vstorage->CompactionScore(i) >= 1) {
+      return true;
+    }
+  }
+  return false;
+}
+
+namespace {
+// A class to build a leveled compaction step-by-step.
+class LevelCompactionBuilder {
+ public:
+  LevelCompactionBuilder(const std::string& cf_name,
+                         VersionStorageInfo* vstorage,
+                         CompactionPicker* compaction_picker,
+                         LogBuffer* log_buffer,
+                         const MutableCFOptions& mutable_cf_options,
+                         const ImmutableCFOptions& ioptions)
+      : cf_name_(cf_name),
+        vstorage_(vstorage),
+        compaction_picker_(compaction_picker),
+        log_buffer_(log_buffer),
+        mutable_cf_options_(mutable_cf_options),
+        ioptions_(ioptions) {}
+
+  // Pick and return a compaction.
+  Compaction* PickCompaction();
+
+  // Pick the initial files to compact to the next level. (or together
+  // in Intra-L0 compactions)
+  void SetupInitialFiles();
+
+  // If the initial files are from L0 level, pick other L0
+  // files if needed.
+  bool SetupOtherL0FilesIfNeeded();
+
+  // Based on initial files, setup other files need to be compacted
+  // in this compaction, accordingly.
+  bool SetupOtherInputsIfNeeded();
+
+  Compaction* GetCompaction();
+
+  // For the specfied level, pick a file that we want to compact.
+  // Returns false if there is no file to compact.
+  // If it returns true, inputs->files.size() will be exactly one.
+  // If level is 0 and there is already a compaction on that level, this
+  // function will return false.
+  bool PickFileToCompact();
+
+  // For L0->L0, picks the longest span of files that aren't currently
+  // undergoing compaction for which work-per-deleted-file decreases. The span
+  // always starts from the newest L0 file.
+  //
+  // Intra-L0 compaction is independent of all other files, so it can be
+  // performed even when L0->base_level compactions are blocked.
+  //
+  // Returns true if `inputs` is populated with a span of files to be compacted;
+  // otherwise, returns false.
+  bool PickIntraL0Compaction();
+
+  void PickExpiredTtlFiles();
+
+  void PickFilesMarkedForPeriodicCompaction();
+
+  const std::string& cf_name_;
+  VersionStorageInfo* vstorage_;
+  CompactionPicker* compaction_picker_;
+  LogBuffer* log_buffer_;
+  int start_level_ = -1;
+  int output_level_ = -1;
+  int parent_index_ = -1;
+  int base_index_ = -1;
+  double start_level_score_ = 0;
+  bool is_manual_ = false;
+  CompactionInputFiles start_level_inputs_;
+  std::vector<CompactionInputFiles> compaction_inputs_;
+  CompactionInputFiles output_level_inputs_;
+  std::vector<FileMetaData*> grandparents_;
+  CompactionReason compaction_reason_ = CompactionReason::kUnknown;
+
+  const MutableCFOptions& mutable_cf_options_;
+  const ImmutableCFOptions& ioptions_;
+  // Pick a path ID to place a newly generated file, with its level
+  static uint32_t GetPathId(const ImmutableCFOptions& ioptions,
+                            const MutableCFOptions& mutable_cf_options,
+                            int level);
+
+  static const int kMinFilesForIntraL0Compaction = 4;
+};
+
+void LevelCompactionBuilder::PickExpiredTtlFiles() {
+  if (vstorage_->ExpiredTtlFiles().empty()) {
+    return;
+  }
+
+  auto continuation = [&](std::pair<int, FileMetaData*> level_file) {
+    // If it's being compacted it has nothing to do here.
+    // If this assert() fails that means that some function marked some
+    // files as being_compacted, but didn't call ComputeCompactionScore()
+    assert(!level_file.second->being_compacted);
+    start_level_ = level_file.first;
+    output_level_ =
+        (start_level_ == 0) ? vstorage_->base_level() : start_level_ + 1;
+
+    if ((start_level_ == vstorage_->num_non_empty_levels() - 1) ||
+        (start_level_ == 0 &&
+         !compaction_picker_->level0_compactions_in_progress()->empty())) {
+      return false;
+    }
+
+    start_level_inputs_.files = {level_file.second};
+    start_level_inputs_.level = start_level_;
+    return compaction_picker_->ExpandInputsToCleanCut(cf_name_, vstorage_,
+                                                      &start_level_inputs_);
+  };
+
+  for (auto& level_file : vstorage_->ExpiredTtlFiles()) {
+    if (continuation(level_file)) {
+      // found the compaction!
+      return;
+    }
+  }
+
+  start_level_inputs_.files.clear();
+}
+
+void LevelCompactionBuilder::PickFilesMarkedForPeriodicCompaction() {
+  if (vstorage_->FilesMarkedForPeriodicCompaction().empty()) {
+    return;
+  }
+
+  auto continuation = [&](std::pair<int, FileMetaData*> level_file) {
+    // If it's being compacted it has nothing to do here.
+    // If this assert() fails that means that some function marked some
+    // files as being_compacted, but didn't call ComputeCompactionScore()
+    assert(!level_file.second->being_compacted);
+    output_level_ = start_level_ = level_file.first;
+
+    if (start_level_ == 0 &&
+        !compaction_picker_->level0_compactions_in_progress()->empty()) {
+      return false;
+    }
+
+    start_level_inputs_.files = {level_file.second};
+    start_level_inputs_.level = start_level_;
+    return compaction_picker_->ExpandInputsToCleanCut(cf_name_, vstorage_,
+                                                      &start_level_inputs_);
+  };
+
+  for (auto& level_file : vstorage_->FilesMarkedForPeriodicCompaction()) {
+    if (continuation(level_file)) {
+      // found the compaction!
+      return;
+    }
+  }
+
+  start_level_inputs_.files.clear();
+}
+
+void LevelCompactionBuilder::SetupInitialFiles() {
+  // Find the compactions by size on all levels.
+  bool skipped_l0_to_base = false;
+  for (int i = 0; i < compaction_picker_->NumberLevels() - 1; i++) {
+    start_level_score_ = vstorage_->CompactionScore(i);
+    start_level_ = vstorage_->CompactionScoreLevel(i);
+    assert(i == 0 || start_level_score_ <= vstorage_->CompactionScore(i - 1));
+    if (start_level_score_ >= 1) {
+      if (skipped_l0_to_base && start_level_ == vstorage_->base_level()) {
+        // If L0->base_level compaction is pending, don't schedule further
+        // compaction from base level. Otherwise L0->base_level compaction
+        // may starve.
+        continue;
+      }
+      output_level_ =
+          (start_level_ == 0) ? vstorage_->base_level() : start_level_ + 1;
+      if (PickFileToCompact()) {
+        // found the compaction!
+        if (start_level_ == 0) {
+          // L0 score = `num L0 files` / `level0_file_num_compaction_trigger`
+          compaction_reason_ = CompactionReason::kLevelL0FilesNum;
+        } else {
+          // L1+ score = `Level files size` / `MaxBytesForLevel`
+          compaction_reason_ = CompactionReason::kLevelMaxLevelSize;
+        }
+        break;
+      } else {
+        // didn't find the compaction, clear the inputs
+        start_level_inputs_.clear();
+        if (start_level_ == 0) {
+          skipped_l0_to_base = true;
+          // L0->base_level may be blocked due to ongoing L0->base_level
+          // compactions. It may also be blocked by an ongoing compaction from
+          // base_level downwards.
+          //
+          // In these cases, to reduce L0 file count and thus reduce likelihood
+          // of write stalls, we can attempt compacting a span of files within
+          // L0.
+          if (PickIntraL0Compaction()) {
+            output_level_ = 0;
+            compaction_reason_ = CompactionReason::kLevelL0FilesNum;
+            break;
+          }
+        }
+      }
+    }
+  }
+
+  // if we didn't find a compaction, check if there are any files marked for
+  // compaction
+  if (start_level_inputs_.empty()) {
+    parent_index_ = base_index_ = -1;
+
+    compaction_picker_->PickFilesMarkedForCompaction(
+        cf_name_, vstorage_, &start_level_, &output_level_,
+        &start_level_inputs_);
+    if (!start_level_inputs_.empty()) {
+      is_manual_ = true;
+      compaction_reason_ = CompactionReason::kFilesMarkedForCompaction;
+      return;
+    }
+  }
+
+  // Bottommost Files Compaction on deleting tombstones
+  if (start_level_inputs_.empty()) {
+    size_t i;
+    for (i = 0; i < vstorage_->BottommostFilesMarkedForCompaction().size();
+         ++i) {
+      auto& level_and_file = vstorage_->BottommostFilesMarkedForCompaction()[i];
+      assert(!level_and_file.second->being_compacted);
+      start_level_inputs_.level = output_level_ = start_level_ =
+          level_and_file.first;
+      start_level_inputs_.files = {level_and_file.second};
+      if (compaction_picker_->ExpandInputsToCleanCut(cf_name_, vstorage_,
+                                                     &start_level_inputs_)) {
+        break;
+      }
+    }
+    if (i == vstorage_->BottommostFilesMarkedForCompaction().size()) {
+      start_level_inputs_.clear();
+    } else {
+      assert(!start_level_inputs_.empty());
+      compaction_reason_ = CompactionReason::kBottommostFiles;
+      return;
+    }
+  }
+
+  // TTL Compaction
+  if (start_level_inputs_.empty()) {
+    PickExpiredTtlFiles();
+    if (!start_level_inputs_.empty()) {
+      compaction_reason_ = CompactionReason::kTtl;
+      return;
+    }
+  }
+
+  // Periodic Compaction
+  if (start_level_inputs_.empty()) {
+    PickFilesMarkedForPeriodicCompaction();
+    if (!start_level_inputs_.empty()) {
+      compaction_reason_ = CompactionReason::kPeriodicCompaction;
+      return;
+    }
+  }
+}
+
+bool LevelCompactionBuilder::SetupOtherL0FilesIfNeeded() {
+  if (start_level_ == 0 && output_level_ != 0) {
+    return compaction_picker_->GetOverlappingL0Files(
+        vstorage_, &start_level_inputs_, output_level_, &parent_index_);
+  }
+  return true;
+}
+
+bool LevelCompactionBuilder::SetupOtherInputsIfNeeded() {
+  // Setup input files from output level. For output to L0, we only compact
+  // spans of files that do not interact with any pending compactions, so don't
+  // need to consider other levels.
+  if (output_level_ != 0) {
+    output_level_inputs_.level = output_level_;
+    if (!compaction_picker_->SetupOtherInputs(
+            cf_name_, mutable_cf_options_, vstorage_, &start_level_inputs_,
+            &output_level_inputs_, &parent_index_, base_index_)) {
+      return false;
+    }
+
+    compaction_inputs_.push_back(start_level_inputs_);
+    if (!output_level_inputs_.empty()) {
+      compaction_inputs_.push_back(output_level_inputs_);
+    }
+
+    // In some edge cases we could pick a compaction that will be compacting
+    // a key range that overlap with another running compaction, and both
+    // of them have the same output level. This could happen if
+    // (1) we are running a non-exclusive manual compaction
+    // (2) AddFile ingest a new file into the LSM tree
+    // We need to disallow this from happening.
+    if (compaction_picker_->FilesRangeOverlapWithCompaction(compaction_inputs_,
+                                                            output_level_)) {
+      // This compaction output could potentially conflict with the output
+      // of a currently running compaction, we cannot run it.
+      return false;
+    }
+    compaction_picker_->GetGrandparents(vstorage_, start_level_inputs_,
+                                        output_level_inputs_, &grandparents_);
+  } else {
+    compaction_inputs_.push_back(start_level_inputs_);
+  }
+  return true;
+}
+
+Compaction* LevelCompactionBuilder::PickCompaction() {
+  // Pick up the first file to start compaction. It may have been extended
+  // to a clean cut.
+  SetupInitialFiles();
+  if (start_level_inputs_.empty()) {
+    return nullptr;
+  }
+  assert(start_level_ >= 0 && output_level_ >= 0);
+
+  // If it is a L0 -> base level compaction, we need to set up other L0
+  // files if needed.
+  if (!SetupOtherL0FilesIfNeeded()) {
+    return nullptr;
+  }
+
+  // Pick files in the output level and expand more files in the start level
+  // if needed.
+  if (!SetupOtherInputsIfNeeded()) {
+    return nullptr;
+  }
+
+  // Form a compaction object containing the files we picked.
+  Compaction* c = GetCompaction();
+
+  TEST_SYNC_POINT_CALLBACK("LevelCompactionPicker::PickCompaction:Return", c);
+
+  return c;
+}
+
+Compaction* LevelCompactionBuilder::GetCompaction() {
+  auto c = new Compaction(
+      vstorage_, ioptions_, mutable_cf_options_, std::move(compaction_inputs_),
+      output_level_,
+      MaxFileSizeForLevel(mutable_cf_options_, output_level_,
+                          ioptions_.compaction_style, vstorage_->base_level(),
+                          ioptions_.level_compaction_dynamic_level_bytes),
+      mutable_cf_options_.max_compaction_bytes,
+      GetPathId(ioptions_, mutable_cf_options_, output_level_),
+      GetCompressionType(ioptions_, vstorage_, mutable_cf_options_,
+                         output_level_, vstorage_->base_level()),
+      GetCompressionOptions(ioptions_, vstorage_, output_level_),
+      /* max_subcompactions */ 0, std::move(grandparents_), is_manual_,
+      start_level_score_, false /* deletion_compaction */, compaction_reason_);
+
+  // If it's level 0 compaction, make sure we don't execute any other level 0
+  // compactions in parallel
+  compaction_picker_->RegisterCompaction(c);
+
+  // Creating a compaction influences the compaction score because the score
+  // takes running compactions into account (by skipping files that are already
+  // being compacted). Since we just changed compaction score, we recalculate it
+  // here
+  vstorage_->ComputeCompactionScore(ioptions_, mutable_cf_options_);
+  return c;
+}
+
+/*
+ * Find the optimal path to place a file
+ * Given a level, finds the path where levels up to it will fit in levels
+ * up to and including this path
+ */
+uint32_t LevelCompactionBuilder::GetPathId(
+    const ImmutableCFOptions& ioptions,
+    const MutableCFOptions& mutable_cf_options, int level) {
+  uint32_t p = 0;
+  assert(!ioptions.cf_paths.empty());
+
+  // size remaining in the most recent path
+  uint64_t current_path_size = ioptions.cf_paths[0].target_size;
+
+  uint64_t level_size;
+  int cur_level = 0;
+
+  // max_bytes_for_level_base denotes L1 size.
+  // We estimate L0 size to be the same as L1.
+  level_size = mutable_cf_options.max_bytes_for_level_base;
+
+  // Last path is the fallback
+  while (p < ioptions.cf_paths.size() - 1) {
+    if (level_size <= current_path_size) {
+      if (cur_level == level) {
+        // Does desired level fit in this path?
+        return p;
+      } else {
+        current_path_size -= level_size;
+        if (cur_level > 0) {
+          if (ioptions.level_compaction_dynamic_level_bytes) {
+            // Currently, level_compaction_dynamic_level_bytes is ignored when
+            // multiple db paths are specified. https://github.com/facebook/
+            // rocksdb/blob/master/db/column_family.cc.
+            // Still, adding this check to avoid accidentally using
+            // max_bytes_for_level_multiplier_additional
+            level_size = static_cast<uint64_t>(
+                level_size * mutable_cf_options.max_bytes_for_level_multiplier);
+          } else {
+            level_size = static_cast<uint64_t>(
+                level_size * mutable_cf_options.max_bytes_for_level_multiplier *
+                mutable_cf_options.MaxBytesMultiplerAdditional(cur_level));
+          }
+        }
+        cur_level++;
+        continue;
+      }
+    }
+    p++;
+    current_path_size = ioptions.cf_paths[p].target_size;
+  }
+  return p;
+}
+
+bool LevelCompactionBuilder::PickFileToCompact() {
+  // level 0 files are overlapping. So we cannot pick more
+  // than one concurrent compactions at this level. This
+  // could be made better by looking at key-ranges that are
+  // being compacted at level 0.
+  if (start_level_ == 0 &&
+      !compaction_picker_->level0_compactions_in_progress()->empty()) {
+    TEST_SYNC_POINT("LevelCompactionPicker::PickCompactionBySize:0");
+    return false;
+  }
+
+  start_level_inputs_.clear();
+
+  assert(start_level_ >= 0);
+
+  // Pick the largest file in this level that is not already
+  // being compacted
+  const std::vector<int>& file_size =
+      vstorage_->FilesByCompactionPri(start_level_);
+  const std::vector<FileMetaData*>& level_files =
+      vstorage_->LevelFiles(start_level_);
+
+  unsigned int cmp_idx;
+  for (cmp_idx = vstorage_->NextCompactionIndex(start_level_);
+       cmp_idx < file_size.size(); cmp_idx++) {
+    int index = file_size[cmp_idx];
+    auto* f = level_files[index];
+
+    // do not pick a file to compact if it is being compacted
+    // from n-1 level.
+    if (f->being_compacted) {
+      continue;
+    }
+
+    start_level_inputs_.files.push_back(f);
+    start_level_inputs_.level = start_level_;
+    if (!compaction_picker_->ExpandInputsToCleanCut(cf_name_, vstorage_,
+                                                    &start_level_inputs_) ||
+        compaction_picker_->FilesRangeOverlapWithCompaction(
+            {start_level_inputs_}, output_level_)) {
+      // A locked (pending compaction) input-level file was pulled in due to
+      // user-key overlap.
+      start_level_inputs_.clear();
+      continue;
+    }
+
+    // Now that input level is fully expanded, we check whether any output files
+    // are locked due to pending compaction.
+    //
+    // Note we rely on ExpandInputsToCleanCut() to tell us whether any output-
+    // level files are locked, not just the extra ones pulled in for user-key
+    // overlap.
+    InternalKey smallest, largest;
+    compaction_picker_->GetRange(start_level_inputs_, &smallest, &largest);
+    CompactionInputFiles output_level_inputs;
+    output_level_inputs.level = output_level_;
+    vstorage_->GetOverlappingInputs(output_level_, &smallest, &largest,
+                                    &output_level_inputs.files);
+    if (!output_level_inputs.empty() &&
+        !compaction_picker_->ExpandInputsToCleanCut(cf_name_, vstorage_,
+                                                    &output_level_inputs)) {
+      start_level_inputs_.clear();
+      continue;
+    }
+    base_index_ = index;
+    break;
+  }
+
+  // store where to start the iteration in the next call to PickCompaction
+  vstorage_->SetNextCompactionIndex(start_level_, cmp_idx);
+
+  return start_level_inputs_.size() > 0;
+}
+
+bool LevelCompactionBuilder::PickIntraL0Compaction() {
+  start_level_inputs_.clear();
+  const std::vector<FileMetaData*>& level_files =
+      vstorage_->LevelFiles(0 /* level */);
+  if (level_files.size() <
+          static_cast<size_t>(
+              mutable_cf_options_.level0_file_num_compaction_trigger + 2) ||
+      level_files[0]->being_compacted) {
+    // If L0 isn't accumulating much files beyond the regular trigger, don't
+    // resort to L0->L0 compaction yet.
+    return false;
+  }
+  return FindIntraL0Compaction(
+      level_files, kMinFilesForIntraL0Compaction, port::kMaxUint64,
+      mutable_cf_options_.max_compaction_bytes, &start_level_inputs_);
+}
+}  // namespace
+
+Compaction* LevelCompactionPicker::PickCompaction(
+    const std::string& cf_name, const MutableCFOptions& mutable_cf_options,
+    VersionStorageInfo* vstorage, LogBuffer* log_buffer) {
+  LevelCompactionBuilder builder(cf_name, vstorage, this, log_buffer,
+                                 mutable_cf_options, ioptions_);
+  return builder.PickCompaction();
+}
+}  // namespace rocksdb
@@ -0,0 +1,32 @@
+//  Copyright (c) 2011-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).
+//
+// Copyright (c) 2011 The LevelDB Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file. See the AUTHORS file for names of contributors.
+
+#pragma once
+
+#include "db/compaction/compaction_picker.h"
+
+namespace rocksdb {
+// Picking compactions for leveled compaction. See wiki page
+// https://github.com/facebook/rocksdb/wiki/Leveled-Compaction
+// for description of Leveled compaction.
+class LevelCompactionPicker : public CompactionPicker {
+ public:
+  LevelCompactionPicker(const ImmutableCFOptions& ioptions,
+                        const InternalKeyComparator* icmp)
+      : CompactionPicker(ioptions, icmp) {}
+  virtual Compaction* PickCompaction(const std::string& cf_name,
+                                     const MutableCFOptions& mutable_cf_options,
+                                     VersionStorageInfo* vstorage,
+                                     LogBuffer* log_buffer) override;
+
+  virtual bool NeedsCompaction(
+      const VersionStorageInfo* vstorage) const override;
+};
+
+}  // namespace rocksdb
@@ -3,28 +3,26 @@
 //  COPYING file in the root directory) and Apache 2.0 License
 //  (found in the LICENSE.Apache file in the root directory).

-#include "db/compaction_picker.h"

 #include <limits>
 #include <string>
 #include <utility>
-#include "db/compaction.h"
-#include "db/compaction_picker_fifo.h"
-#include "db/compaction_picker_universal.h"
+#include "db/compaction/compaction.h"
+#include "db/compaction/compaction_picker_fifo.h"
+#include "db/compaction/compaction_picker_level.h"
+#include "db/compaction/compaction_picker_universal.h"

-#include "util/logging.h"
+#include "logging/logging.h"
+#include "test_util/testharness.h"
+#include "test_util/testutil.h"
 #include "util/string_util.h"
-#include "util/testharness.h"
-#include "util/testutil.h"

 namespace rocksdb {

 class CountingLogger : public Logger {
 public:
  using Logger::Logv;
-  virtual void Logv(const char* /*format*/, va_list /*ap*/) override {
-    log_count++;
-  }
+  void Logv(const char* /*format*/, va_list /*ap*/) override { log_count++; }
  size_t log_count;
 };

@@ -68,8 +66,7 @@ class CompactionPickerTest : public testing::Test {
                                    std::numeric_limits<uint64_t>::max());
  }

-  ~CompactionPickerTest() {
-  }
+  ~CompactionPickerTest() override {}

  void NewVersionStorage(int num_levels, CompactionStyle style) {
    DeleteVersionStorage();
@@ -1481,6 +1478,65 @@ TEST_F(CompactionPickerTest, CacheNextCompactionIndex) {
  ASSERT_EQ(4, vstorage_->NextCompactionIndex(1 /* level */));
 }

+TEST_F(CompactionPickerTest, IntraL0MaxCompactionBytesNotHit) {
+  // Intra L0 compaction triggers only if there are at least
+  // level0_file_num_compaction_trigger + 2 L0 files.
+  mutable_cf_options_.level0_file_num_compaction_trigger = 3;
+  mutable_cf_options_.max_compaction_bytes = 1000000u;
+  NewVersionStorage(6, kCompactionStyleLevel);
+
+  // All 5 L0 files will be picked for intra L0 compaction. The one L1 file
+  // spans entire L0 key range and is marked as being compacted to avoid
+  // L0->L1 compaction.
+  Add(0, 1U, "100", "150", 200000U);
+  Add(0, 2U, "151", "200", 200000U);
+  Add(0, 3U, "201", "250", 200000U);
+  Add(0, 4U, "251", "300", 200000U);
+  Add(0, 5U, "301", "350", 200000U);
+  Add(1, 6U, "100", "350", 200000U);
+  vstorage_->LevelFiles(1)[0]->being_compacted = true;
+  UpdateVersionStorageInfo();
+
+  std::unique_ptr<Compaction> compaction(level_compaction_picker.PickCompaction(
+      cf_name_, mutable_cf_options_, vstorage_.get(), &log_buffer_));
+  ASSERT_TRUE(compaction.get() != nullptr);
+  ASSERT_EQ(1U, compaction->num_input_levels());
+  ASSERT_EQ(5U, compaction->num_input_files(0));
+  ASSERT_EQ(CompactionReason::kLevelL0FilesNum,
+            compaction->compaction_reason());
+  ASSERT_EQ(0U, compaction->output_level());
+}
+
+TEST_F(CompactionPickerTest, IntraL0MaxCompactionBytesHit) {
+  // Intra L0 compaction triggers only if there are at least
+  // level0_file_num_compaction_trigger + 2 L0 files.
+  mutable_cf_options_.level0_file_num_compaction_trigger = 3;
+  mutable_cf_options_.max_compaction_bytes = 999999u;
+  NewVersionStorage(6, kCompactionStyleLevel);
+
+  // 4 out of 5 L0 files will be picked for intra L0 compaction due to
+  // max_compaction_bytes limit (the minimum number of files for triggering
+  // intra L0 compaction is 4). The one L1 file spans entire L0 key range and
+  // is marked as being compacted to avoid L0->L1 compaction.
+  Add(0, 1U, "100", "150", 200000U);
+  Add(0, 2U, "151", "200", 200000U);
+  Add(0, 3U, "201", "250", 200000U);
+  Add(0, 4U, "251", "300", 200000U);
+  Add(0, 5U, "301", "350", 200000U);
+  Add(1, 6U, "100", "350", 200000U);
+  vstorage_->LevelFiles(1)[0]->being_compacted = true;
+  UpdateVersionStorageInfo();
+
+  std::unique_ptr<Compaction> compaction(level_compaction_picker.PickCompaction(
+      cf_name_, mutable_cf_options_, vstorage_.get(), &log_buffer_));
+  ASSERT_TRUE(compaction.get() != nullptr);
+  ASSERT_EQ(1U, compaction->num_input_levels());
+  ASSERT_EQ(4U, compaction->num_input_files(0));
+  ASSERT_EQ(CompactionReason::kLevelL0FilesNum,
+            compaction->compaction_reason());
+  ASSERT_EQ(0U, compaction->output_level());
+}
+
 }  // namespace rocksdb

 int main(int argc, char** argv) {
@@ -7,25 +7,21 @@
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file. See the AUTHORS file for names of contributors.

-#include "db/compaction_picker_universal.h"
+#include "db/compaction/compaction_picker_universal.h"
 #ifndef ROCKSDB_LITE

-#ifndef __STDC_FORMAT_MACROS
-#define __STDC_FORMAT_MACROS
-#endif
-
-#include <inttypes.h>
+#include <cinttypes>
 #include <limits>
 #include <queue>
 #include <string>
 #include <utility>
 #include "db/column_family.h"
+#include "file/filename.h"
+#include "logging/log_buffer.h"
 #include "monitoring/statistics.h"
-#include "util/filename.h"
-#include "util/log_buffer.h"
+#include "test_util/sync_point.h"
 #include "util/random.h"
 #include "util/string_util.h"
-#include "util/sync_point.h"

 namespace rocksdb {
 namespace {
@@ -395,8 +391,8 @@ Compaction* UniversalCompactionPicker::PickCompaction(
  }
 #endif
  // update statistics
-  MeasureTime(ioptions_.statistics, NUM_FILES_IN_SINGLE_COMPACTION,
-              c->inputs(0)->size());
+  RecordInHistogram(ioptions_.statistics, NUM_FILES_IN_SINGLE_COMPACTION,
+                    c->inputs(0)->size());

  RegisterCompaction(c);
  vstorage->ComputeCompactionScore(ioptions_, mutable_cf_options);
@@ -10,7 +10,7 @@
 #pragma once
 #ifndef ROCKSDB_LITE

-#include "db/compaction_picker.h"
+#include "db/compaction/compaction_picker.h"

 namespace rocksdb {
 class UniversalCompactionPicker : public CompactionPicker {
@@ -9,11 +9,11 @@
 #include "memtable/stl_wrappers.h"
 #include "rocksdb/db.h"
 #include "rocksdb/env.h"
+#include "test_util/testharness.h"
+#include "test_util/testutil.h"
 #include "util/hash.h"
 #include "util/kv_map.h"
 #include "util/string_util.h"
-#include "util/testharness.h"
-#include "util/testutil.h"
 #include "utilities/merge_operators.h"

 using std::unique_ptr;
@@ -27,24 +27,24 @@ class KVIter : public Iterator {
 public:
  explicit KVIter(const stl_wrappers::KVMap* map)
      : map_(map), iter_(map_->end()) {}
-  virtual bool Valid() const override { return iter_ != map_->end(); }
-  virtual void SeekToFirst() override { iter_ = map_->begin(); }
-  virtual void SeekToLast() override {
+  bool Valid() const override { return iter_ != map_->end(); }
+  void SeekToFirst() override { iter_ = map_->begin(); }
+  void SeekToLast() override {
    if (map_->empty()) {
      iter_ = map_->end();
    } else {
      iter_ = map_->find(map_->rbegin()->first);
    }
  }
-  virtual void Seek(const Slice& k) override {
+  void Seek(const Slice& k) override {
    iter_ = map_->lower_bound(k.ToString());
  }
-  virtual void SeekForPrev(const Slice& k) override {
+  void SeekForPrev(const Slice& k) override {
    iter_ = map_->upper_bound(k.ToString());
    Prev();
  }
-  virtual void Next() override { ++iter_; }
-  virtual void Prev() override {
+  void Next() override { ++iter_; }
+  void Prev() override {
    if (iter_ == map_->begin()) {
      iter_ = map_->end();
      return;
@@ -52,9 +52,9 @@ class KVIter : public Iterator {
    --iter_;
  }

-  virtual Slice key() const override { return iter_->first; }
-  virtual Slice value() const override { return iter_->second; }
-  virtual Status status() const override { return Status::OK(); }
+  Slice key() const override { return iter_->first; }
+  Slice value() const override { return iter_->second; }
+  Status status() const override { return Status::OK(); }

 private:
  const stl_wrappers::KVMap* const map_;
@@ -171,9 +171,9 @@ class DoubleComparator : public Comparator {
 public:
  DoubleComparator() {}

-  virtual const char* Name() const override { return "DoubleComparator"; }
+  const char* Name() const override { return "DoubleComparator"; }

-  virtual int Compare(const Slice& a, const Slice& b) const override {
+  int Compare(const Slice& a, const Slice& b) const override {
 #ifndef CYGWIN
    double da = std::stod(a.ToString());
    double db = std::stod(b.ToString());
@@ -189,19 +189,19 @@ class DoubleComparator : public Comparator {
      return -1;
    }
  }
-  virtual void FindShortestSeparator(std::string* /*start*/,
-                                     const Slice& /*limit*/) const override {}
+  void FindShortestSeparator(std::string* /*start*/,
+                             const Slice& /*limit*/) const override {}

-  virtual void FindShortSuccessor(std::string* /*key*/) const override {}
+  void FindShortSuccessor(std::string* /*key*/) const override {}
 };

 class HashComparator : public Comparator {
 public:
  HashComparator() {}

-  virtual const char* Name() const override { return "HashComparator"; }
+  const char* Name() const override { return "HashComparator"; }

-  virtual int Compare(const Slice& a, const Slice& b) const override {
+  int Compare(const Slice& a, const Slice& b) const override {
    uint32_t ha = Hash(a.data(), a.size(), 66);
    uint32_t hb = Hash(b.data(), b.size(), 66);
    if (ha == hb) {
@@ -212,19 +212,19 @@ class HashComparator : public Comparator {
      return -1;
    }
  }
-  virtual void FindShortestSeparator(std::string* /*start*/,
-                                     const Slice& /*limit*/) const override {}
+  void FindShortestSeparator(std::string* /*start*/,
+                             const Slice& /*limit*/) const override {}

-  virtual void FindShortSuccessor(std::string* /*key*/) const override {}
+  void FindShortSuccessor(std::string* /*key*/) const override {}
 };

 class TwoStrComparator : public Comparator {
 public:
  TwoStrComparator() {}

-  virtual const char* Name() const override { return "TwoStrComparator"; }
+  const char* Name() const override { return "TwoStrComparator"; }

-  virtual int Compare(const Slice& a, const Slice& b) const override {
+  int Compare(const Slice& a, const Slice& b) const override {
    assert(a.size() >= 2);
    assert(b.size() >= 2);
    size_t size_a1 = static_cast<size_t>(a[0]);
@@ -244,10 +244,10 @@ class TwoStrComparator : public Comparator {
    }
    return a2.compare(b2);
  }
-  virtual void FindShortestSeparator(std::string* /*start*/,
-                                     const Slice& /*limit*/) const override {}
+  void FindShortestSeparator(std::string* /*start*/,
+                             const Slice& /*limit*/) const override {}

-  virtual void FindShortSuccessor(std::string* /*key*/) const override {}
+  void FindShortSuccessor(std::string* /*key*/) const override {}
 };
 }  // namespace

@@ -272,7 +272,7 @@ class ComparatorDBTest
    EXPECT_OK(DestroyDB(dbname_, last_options_));
  }

-  ~ComparatorDBTest() {
+  ~ComparatorDBTest() override {
    delete db_;
    EXPECT_OK(DestroyDB(dbname_, last_options_));
    comparator = BytewiseComparator();
@@ -8,7 +8,7 @@

 #include "rocksdb/convenience.h"

-#include "db/db_impl.h"
+#include "db/db_impl/db_impl.h"
 #include "util/cast_util.h"

 namespace rocksdb {
@@ -59,7 +59,7 @@ Status VerifySstFileChecksum(const Options& options,
  if (!s.ok()) {
    return s;
  }
-  s = table_reader->VerifyChecksum();
+  s = table_reader->VerifyChecksum(TableReaderCaller::kUserVerifyChecksum);
  return s;
 }

@@ -13,23 +13,23 @@

 #include <errno.h>
 #include <fcntl.h>
-#include <inttypes.h>
+#include <cinttypes>
 #include <sys/stat.h>
 #include <sys/types.h>
-#include "db/db_impl.h"
+#include "db/db_impl/db_impl.h"
 #include "db/log_format.h"
 #include "db/version_set.h"
+#include "file/filename.h"
 #include "rocksdb/cache.h"
 #include "rocksdb/convenience.h"
 #include "rocksdb/env.h"
 #include "rocksdb/table.h"
 #include "rocksdb/write_batch.h"
-#include "table/block_based_table_builder.h"
+#include "table/block_based/block_based_table_builder.h"
 #include "table/meta_blocks.h"
-#include "util/filename.h"
+#include "test_util/testharness.h"
+#include "test_util/testutil.h"
 #include "util/string_util.h"
-#include "util/testharness.h"
-#include "util/testutil.h"

 namespace rocksdb {

@@ -62,9 +62,9 @@ class CorruptionTest : public testing::Test {
    options_.create_if_missing = false;
  }

-  ~CorruptionTest() {
-     delete db_;
-     DestroyDB(dbname_, Options());
+  ~CorruptionTest() override {
+    delete db_;
+    DestroyDB(dbname_, Options());
  }

  void CloseDb() {
@@ -5,15 +5,15 @@

 #ifndef ROCKSDB_LITE

-#include "db/db_impl.h"
+#include "db/db_impl/db_impl.h"
 #include "rocksdb/db.h"
 #include "rocksdb/env.h"
-#include "table/cuckoo_table_factory.h"
-#include "table/cuckoo_table_reader.h"
+#include "table/cuckoo/cuckoo_table_factory.h"
+#include "table/cuckoo/cuckoo_table_reader.h"
 #include "table/meta_blocks.h"
+#include "test_util/testharness.h"
+#include "test_util/testutil.h"
 #include "util/string_util.h"
-#include "util/testharness.h"
-#include "util/testutil.h"

 namespace rocksdb {

@@ -31,7 +31,7 @@ class CuckooTableDBTest : public testing::Test {
    Reopen();
  }

-  ~CuckooTableDBTest() {
+  ~CuckooTableDBTest() override {
    delete db_;
    EXPECT_OK(DestroyDB(dbname_, Options()));
  }
@@ -10,9 +10,11 @@
 #include "db/db_test_util.h"
 #include "port/stack_trace.h"
 #include "rocksdb/perf_context.h"
-#include "util/fault_injection_test_env.h"
+#include "rocksdb/utilities/debug.h"
+#include "table/block_based/block_builder.h"
+#include "test_util/fault_injection_test_env.h"
 #if !defined(ROCKSDB_LITE)
-#include "util/sync_point.h"
+#include "test_util/sync_point.h"
 #endif

 namespace rocksdb {
@@ -215,11 +217,11 @@ TEST_F(DBBasicTest, PutSingleDeleteGet) {
    ASSERT_EQ("v2", Get(1, "foo2"));
    ASSERT_OK(SingleDelete(1, "foo"));
    ASSERT_EQ("NOT_FOUND", Get(1, "foo"));
-    // Skip HashCuckooRep as it does not support single delete. FIFO and
-    // universal compaction do not apply to the test case. Skip MergePut
-    // because single delete does not get removed when it encounters a merge.
-  } while (ChangeOptions(kSkipHashCuckoo | kSkipFIFOCompaction |
-                         kSkipUniversalCompaction | kSkipMergePut));
+    // Ski FIFO and universal compaction because they do not apply to the test
+    // case. Skip MergePut because single delete does not get removed when it
+    // encounters a merge.
+  } while (ChangeOptions(kSkipFIFOCompaction | kSkipUniversalCompaction |
+                         kSkipMergePut));
 }

 TEST_F(DBBasicTest, EmptyFlush) {
@@ -237,11 +239,11 @@ TEST_F(DBBasicTest, EmptyFlush) {
    ASSERT_OK(Flush(1));

    ASSERT_EQ("[ ]", AllEntriesFor("a", 1));
-    // Skip HashCuckooRep as it does not support single delete. FIFO and
-    // universal compaction do not apply to the test case. Skip MergePut
-    // because merges cannot be combined with single deletions.
-  } while (ChangeOptions(kSkipHashCuckoo | kSkipFIFOCompaction |
-                         kSkipUniversalCompaction | kSkipMergePut));
+    // Skip FIFO and  universal compaction as they do not apply to the test
+    // case. Skip MergePut because merges cannot be combined with single
+    // deletions.
+  } while (ChangeOptions(kSkipFIFOCompaction | kSkipUniversalCompaction |
+                         kSkipMergePut));
 }

 TEST_F(DBBasicTest, GetFromVersions) {
@@ -265,11 +267,6 @@ TEST_F(DBBasicTest, GetSnapshot) {
      std::string key = (i == 0) ? std::string("foo") : std::string(200, 'x');
      ASSERT_OK(Put(1, key, "v1"));
      const Snapshot* s1 = db_->GetSnapshot();
-      if (option_config_ == kHashCuckoo) {
-        // Unsupported case.
-        ASSERT_TRUE(s1 == nullptr);
-        break;
-      }
      ASSERT_OK(Put(1, key, "v2"));
      ASSERT_EQ("v2", Get(1, key));
      ASSERT_EQ("v1", Get(1, key, s1));
@@ -510,7 +507,7 @@ TEST_F(DBBasicTest, Snapshot) {
    ASSERT_EQ(0U, GetNumSnapshots());
    ASSERT_EQ("0v4", Get(0, "foo"));
    ASSERT_EQ("1v4", Get(1, "foo"));
-  } while (ChangeOptions(kSkipHashCuckoo));
+  } while (ChangeOptions());
 }

 #endif  // ROCKSDB_LITE
@@ -566,8 +563,7 @@ TEST_F(DBBasicTest, CompactBetweenSnapshots) {
                           nullptr);
    ASSERT_EQ("sixth", Get(1, "foo"));
    ASSERT_EQ(AllEntriesFor("foo", 1), "[ sixth ]");
-    // skip HashCuckooRep as it does not support snapshot
-  } while (ChangeOptions(kSkipHashCuckoo | kSkipFIFOCompaction));
+  } while (ChangeOptions(kSkipFIFOCompaction));
 }

 TEST_F(DBBasicTest, DBOpen_Options) {
@@ -858,18 +854,17 @@ class TestEnv : public EnvWrapper {
      public:
        using Logger::Logv;
        TestLogger(TestEnv *env_ptr) : Logger() { env = env_ptr; }
-        ~TestLogger() {
+        ~TestLogger() override {
          if (!closed_) {
            CloseHelper();
          }
        }
-        virtual void Logv(const char* /*format*/, va_list /*ap*/) override{};
+        void Logv(const char* /*format*/, va_list /*ap*/) override{};

       protected:
-        virtual Status CloseImpl() override {
-          return CloseHelper();
-        }
-      private:
+        Status CloseImpl() override { return CloseHelper(); }
+
+       private:
        Status CloseHelper() {
          env->CloseCountInc();;
          return Status::IOError();
@@ -881,8 +876,8 @@ class TestEnv : public EnvWrapper {

    int GetCloseCount() { return close_count; }

-    virtual Status NewLogger(const std::string& /*fname*/,
-                             std::shared_ptr<Logger>* result) {
+    Status NewLogger(const std::string& /*fname*/,
+                     std::shared_ptr<Logger>* result) override {
      result->reset(new TestLogger(this));
      return Status::OK();
    }
@@ -1001,7 +996,7 @@ TEST_F(DBBasicTest, MultiGetMultiCF) {
    keys.push_back("cf" + std::to_string(i) + "_key");
  }

-  values = MultiGet(cfs, keys);
+  values = MultiGet(cfs, keys, nullptr);
  ASSERT_EQ(values.size(), 8);
  for (unsigned int j = 0; j < values.size(); ++j) {
    ASSERT_EQ(values[j], "cf" + std::to_string(j) + "_val2");
@@ -1061,7 +1056,7 @@ TEST_F(DBBasicTest, MultiGetMultiCFMutex) {
    keys.push_back("cf" + std::to_string(i) + "_key");
  }

-  values = MultiGet(cfs, keys);
+  values = MultiGet(cfs, keys, nullptr);
  ASSERT_TRUE(last_try);
  ASSERT_EQ(values.size(), 8);
  for (unsigned int j = 0; j < values.size(); ++j) {
@@ -1135,6 +1130,650 @@ TEST_F(DBBasicTest, MultiGetMultiCFSnapshot) {
  }
 }

+TEST_F(DBBasicTest, MultiGetBatchedSimpleUnsorted) {
+  do {
+    CreateAndReopenWithCF({"pikachu"}, CurrentOptions());
+    SetPerfLevel(kEnableCount);
+    ASSERT_OK(Put(1, "k1", "v1"));
+    ASSERT_OK(Put(1, "k2", "v2"));
+    ASSERT_OK(Put(1, "k3", "v3"));
+    ASSERT_OK(Put(1, "k4", "v4"));
+    ASSERT_OK(Delete(1, "k4"));
+    ASSERT_OK(Put(1, "k5", "v5"));
+    ASSERT_OK(Delete(1, "no_key"));
+
+    get_perf_context()->Reset();
+
+    std::vector<Slice> keys({"no_key", "k5", "k4", "k3", "k2", "k1"});
+    std::vector<PinnableSlice> values(keys.size());
+    std::vector<ColumnFamilyHandle*> cfs(keys.size(), handles_[1]);
+    std::vector<Status> s(keys.size());
+
+    db_->MultiGet(ReadOptions(), handles_[1], keys.size(), keys.data(),
+                  values.data(), s.data(), false);
+
+    ASSERT_EQ(values.size(), keys.size());
+    ASSERT_EQ(std::string(values[5].data(), values[5].size()), "v1");
+    ASSERT_EQ(std::string(values[4].data(), values[4].size()), "v2");
+    ASSERT_EQ(std::string(values[3].data(), values[3].size()), "v3");
+    ASSERT_EQ(std::string(values[1].data(), values[1].size()), "v5");
+    // four kv pairs * two bytes per value
+    ASSERT_EQ(8, (int)get_perf_context()->multiget_read_bytes);
+
+    ASSERT_TRUE(s[0].IsNotFound());
+    ASSERT_OK(s[1]);
+    ASSERT_TRUE(s[2].IsNotFound());
+    ASSERT_OK(s[3]);
+    ASSERT_OK(s[4]);
+    ASSERT_OK(s[5]);
+
+    SetPerfLevel(kDisable);
+  } while (ChangeCompactOptions());
+}
+
+TEST_F(DBBasicTest, MultiGetBatchedSimpleSorted) {
+  do {
+    CreateAndReopenWithCF({"pikachu"}, CurrentOptions());
+    SetPerfLevel(kEnableCount);
+    ASSERT_OK(Put(1, "k1", "v1"));
+    ASSERT_OK(Put(1, "k2", "v2"));
+    ASSERT_OK(Put(1, "k3", "v3"));
+    ASSERT_OK(Put(1, "k4", "v4"));
+    ASSERT_OK(Delete(1, "k4"));
+    ASSERT_OK(Put(1, "k5", "v5"));
+    ASSERT_OK(Delete(1, "no_key"));
+
+    get_perf_context()->Reset();
+
+    std::vector<Slice> keys({"k1", "k2", "k3", "k4", "k5", "no_key"});
+    std::vector<PinnableSlice> values(keys.size());
+    std::vector<ColumnFamilyHandle*> cfs(keys.size(), handles_[1]);
+    std::vector<Status> s(keys.size());
+
+    db_->MultiGet(ReadOptions(), handles_[1], keys.size(), keys.data(),
+                  values.data(), s.data(), true);
+
+    ASSERT_EQ(values.size(), keys.size());
+    ASSERT_EQ(std::string(values[0].data(), values[0].size()), "v1");
+    ASSERT_EQ(std::string(values[1].data(), values[1].size()), "v2");
+    ASSERT_EQ(std::string(values[2].data(), values[2].size()), "v3");
+    ASSERT_EQ(std::string(values[4].data(), values[4].size()), "v5");
+    // four kv pairs * two bytes per value
+    ASSERT_EQ(8, (int)get_perf_context()->multiget_read_bytes);
+
+    ASSERT_OK(s[0]);
+    ASSERT_OK(s[1]);
+    ASSERT_OK(s[2]);
+    ASSERT_TRUE(s[3].IsNotFound());
+    ASSERT_OK(s[4]);
+    ASSERT_TRUE(s[5].IsNotFound());
+
+    SetPerfLevel(kDisable);
+  } while (ChangeCompactOptions());
+}
+
+TEST_F(DBBasicTest, MultiGetBatchedMultiLevel) {
+  Options options = CurrentOptions();
+  options.disable_auto_compactions = true;
+  Reopen(options);
+  int num_keys = 0;
+
+  for (int i = 0; i < 128; ++i) {
+    ASSERT_OK(Put("key_" + std::to_string(i), "val_l2_" + std::to_string(i)));
+    num_keys++;
+    if (num_keys == 8) {
+      Flush();
+      num_keys = 0;
+    }
+  }
+  if (num_keys > 0) {
+    Flush();
+    num_keys = 0;
+  }
+  MoveFilesToLevel(2);
+
+  for (int i = 0; i < 128; i += 3) {
+    ASSERT_OK(Put("key_" + std::to_string(i), "val_l1_" + std::to_string(i)));
+    num_keys++;
+    if (num_keys == 8) {
+      Flush();
+      num_keys = 0;
+    }
+  }
+  if (num_keys > 0) {
+    Flush();
+    num_keys = 0;
+  }
+  MoveFilesToLevel(1);
+
+  for (int i = 0; i < 128; i += 5) {
+    ASSERT_OK(Put("key_" + std::to_string(i), "val_l0_" + std::to_string(i)));
+    num_keys++;
+    if (num_keys == 8) {
+      Flush();
+      num_keys = 0;
+    }
+  }
+  if (num_keys > 0) {
+    Flush();
+    num_keys = 0;
+  }
+  ASSERT_EQ(0, num_keys);
+
+  for (int i = 0; i < 128; i += 9) {
+    ASSERT_OK(Put("key_" + std::to_string(i), "val_mem_" + std::to_string(i)));
+  }
+
+  std::vector<std::string> keys;
+  std::vector<std::string> values;
+
+  for (int i = 64; i < 80; ++i) {
+    keys.push_back("key_" + std::to_string(i));
+  }
+
+  values = MultiGet(keys, nullptr);
+  ASSERT_EQ(values.size(), 16);
+  for (unsigned int j = 0; j < values.size(); ++j) {
+    int key = j + 64;
+    if (key % 9 == 0) {
+      ASSERT_EQ(values[j], "val_mem_" + std::to_string(key));
+    } else if (key % 5 == 0) {
+      ASSERT_EQ(values[j], "val_l0_" + std::to_string(key));
+    } else if (key % 3 == 0) {
+      ASSERT_EQ(values[j], "val_l1_" + std::to_string(key));
+    } else {
+      ASSERT_EQ(values[j], "val_l2_" + std::to_string(key));
+    }
+  }
+}
+
+#ifndef ROCKSDB_LITE
+TEST_F(DBBasicTest, GetAllKeyVersions) {
+  Options options = CurrentOptions();
+  options.env = env_;
+  options.create_if_missing = true;
+  options.disable_auto_compactions = true;
+  CreateAndReopenWithCF({"pikachu"}, options);
+  ASSERT_EQ(2, handles_.size());
+  const size_t kNumInserts = 4;
+  const size_t kNumDeletes = 4;
+  const size_t kNumUpdates = 4;
+
+  // Check default column family
+  for (size_t i = 0; i != kNumInserts; ++i) {
+    ASSERT_OK(Put(std::to_string(i), "value"));
+  }
+  for (size_t i = 0; i != kNumUpdates; ++i) {
+    ASSERT_OK(Put(std::to_string(i), "value1"));
+  }
+  for (size_t i = 0; i != kNumDeletes; ++i) {
+    ASSERT_OK(Delete(std::to_string(i)));
+  }
+  std::vector<KeyVersion> key_versions;
+  ASSERT_OK(rocksdb::GetAllKeyVersions(db_, Slice(), Slice(),
+                                       std::numeric_limits<size_t>::max(),
+                                       &key_versions));
+  ASSERT_EQ(kNumInserts + kNumDeletes + kNumUpdates, key_versions.size());
+  ASSERT_OK(rocksdb::GetAllKeyVersions(db_, handles_[0], Slice(), Slice(),
+                                       std::numeric_limits<size_t>::max(),
+                                       &key_versions));
+  ASSERT_EQ(kNumInserts + kNumDeletes + kNumUpdates, key_versions.size());
+
+  // Check non-default column family
+  for (size_t i = 0; i != kNumInserts - 1; ++i) {
+    ASSERT_OK(Put(1, std::to_string(i), "value"));
+  }
+  for (size_t i = 0; i != kNumUpdates - 1; ++i) {
+    ASSERT_OK(Put(1, std::to_string(i), "value1"));
+  }
+  for (size_t i = 0; i != kNumDeletes - 1; ++i) {
+    ASSERT_OK(Delete(1, std::to_string(i)));
+  }
+  ASSERT_OK(rocksdb::GetAllKeyVersions(db_, handles_[1], Slice(), Slice(),
+                                       std::numeric_limits<size_t>::max(),
+                                       &key_versions));
+  ASSERT_EQ(kNumInserts + kNumDeletes + kNumUpdates - 3, key_versions.size());
+}
+#endif  // !ROCKSDB_LITE
+
+class DBBasicTestWithParallelIO
+    : public DBTestBase,
+      public testing::WithParamInterface<std::tuple<bool,bool,bool,bool>> {
+ public:
+  DBBasicTestWithParallelIO()
+      : DBTestBase("/db_basic_test_with_parallel_io") {
+    bool compressed_cache = std::get<0>(GetParam());
+    bool uncompressed_cache = std::get<1>(GetParam());
+    compression_enabled_ = std::get<2>(GetParam());
+    fill_cache_ = std::get<3>(GetParam());
+
+    if (compressed_cache) {
+      std::shared_ptr<Cache> cache = NewLRUCache(1048576);
+      compressed_cache_ = std::make_shared<MyBlockCache>(cache);
+    }
+    if (uncompressed_cache) {
+      std::shared_ptr<Cache> cache = NewLRUCache(1048576);
+      uncompressed_cache_ = std::make_shared<MyBlockCache>(cache);
+    }
+
+    env_->count_random_reads_ = true;
+
+    Options options = CurrentOptions();
+    Random rnd(301);
+    BlockBasedTableOptions table_options;
+    table_options.pin_l0_filter_and_index_blocks_in_cache = true;
+    table_options.block_cache = uncompressed_cache_;
+    table_options.block_cache_compressed = compressed_cache_;
+    table_options.flush_block_policy_factory.reset(
+                      new MyFlushBlockPolicyFactory());
+    options.table_factory.reset(new BlockBasedTableFactory(table_options));
+    if (!compression_enabled_) {
+      options.compression = kNoCompression;
+    }
+    Reopen(options);
+
+    std::string zero_str(128, '\0');
+    for (int i = 0; i < 100; ++i) {
+      // Make the value compressible. A purely random string doesn't compress
+      // and the resultant data block will not be compressed
+      values_.emplace_back(RandomString(&rnd, 128) + zero_str);
+      assert(Put(Key(i), values_[i]) == Status::OK());
+    }
+    Flush();
+  }
+
+  bool CheckValue(int i, const std::string& value) {
+    if (values_[i].compare(value) == 0) {
+      return true;
+    }
+    return false;
+  }
+
+  int num_lookups() { return uncompressed_cache_->num_lookups(); }
+  int num_found() { return uncompressed_cache_->num_found(); }
+  int num_inserts() { return uncompressed_cache_->num_inserts(); }
+
+  int num_lookups_compressed() {
+    return compressed_cache_->num_lookups();
+  }
+  int num_found_compressed() {
+    return compressed_cache_->num_found();
+  }
+  int num_inserts_compressed() {
+    return compressed_cache_->num_inserts();
+  }
+
+  bool fill_cache() { return fill_cache_; }
+
+  static void SetUpTestCase() {}
+  static void TearDownTestCase() {}
+
+ private:
+  class MyFlushBlockPolicyFactory
+    : public FlushBlockPolicyFactory {
+   public:
+    MyFlushBlockPolicyFactory() {}
+
+    virtual const char* Name() const override {
+      return "MyFlushBlockPolicyFactory";
+    }
+
+    virtual FlushBlockPolicy* NewFlushBlockPolicy(
+        const BlockBasedTableOptions& /*table_options*/,
+        const BlockBuilder& data_block_builder) const override {
+      return new MyFlushBlockPolicy(data_block_builder);
+    }
+  };
+
+  class MyFlushBlockPolicy
+    : public FlushBlockPolicy {
+   public:
+    explicit MyFlushBlockPolicy(const BlockBuilder& data_block_builder)
+      : num_keys_(0), data_block_builder_(data_block_builder) {}
+
+    bool Update(const Slice& /*key*/, const Slice& /*value*/) override {
+      if (data_block_builder_.empty()) {
+        // First key in this block
+        num_keys_ = 1;
+        return false;
+      }
+      // Flush every 10 keys
+      if (num_keys_ == 10) {
+        num_keys_ = 1;
+        return true;
+      }
+      num_keys_++;
+      return false;
+    }
+
+   private:
+    int num_keys_;
+    const BlockBuilder& data_block_builder_;
+  };
+
+  class MyBlockCache
+    : public Cache {
+   public:
+    explicit MyBlockCache(std::shared_ptr<Cache>& target)
+      : target_(target), num_lookups_(0), num_found_(0), num_inserts_(0) {}
+
+    virtual const char* Name() const override { return "MyBlockCache"; }
+
+    virtual Status Insert(const Slice& key, void* value, size_t charge,
+                          void (*deleter)(const Slice& key, void* value),
+                          Handle** handle = nullptr,
+                          Priority priority = Priority::LOW) override {
+      num_inserts_++;
+      return target_->Insert(key, value, charge, deleter, handle, priority);
+    }
+
+    virtual Handle* Lookup(const Slice& key,
+                           Statistics* stats = nullptr) override {
+      num_lookups_++;
+      Handle* handle = target_->Lookup(key, stats);
+      if (handle != nullptr) {
+        num_found_++;
+      }
+      return handle;
+    }
+
+    virtual bool Ref(Handle* handle) override {
+      return target_->Ref(handle);
+    }
+
+    virtual bool Release(Handle* handle, bool force_erase = false) override {
+      return target_->Release(handle, force_erase);
+    }
+
+    virtual void* Value(Handle* handle) override {
+      return target_->Value(handle);
+    }
+
+    virtual void Erase(const Slice& key) override {
+      target_->Erase(key);
+    }
+    virtual uint64_t NewId() override {
+      return target_->NewId();
+    }
+
+    virtual void SetCapacity(size_t capacity) override {
+      target_->SetCapacity(capacity);
+    }
+
+    virtual void SetStrictCapacityLimit(bool strict_capacity_limit) override {
+      target_->SetStrictCapacityLimit(strict_capacity_limit);
+    }
+
+    virtual bool HasStrictCapacityLimit() const override {
+      return target_->HasStrictCapacityLimit();
+    }
+
+    virtual size_t GetCapacity() const override {
+      return target_->GetCapacity();
+    }
+
+    virtual size_t GetUsage() const override {
+      return target_->GetUsage();
+    }
+
+    virtual size_t GetUsage(Handle* handle) const override {
+      return target_->GetUsage(handle);
+    }
+
+    virtual size_t GetPinnedUsage() const override {
+      return target_->GetPinnedUsage();
+    }
+
+    virtual size_t GetCharge(Handle* /*handle*/) const override { return 0; }
+
+    virtual void ApplyToAllCacheEntries(void (*callback)(void*, size_t),
+                                        bool thread_safe) override {
+      return target_->ApplyToAllCacheEntries(callback, thread_safe);
+    }
+
+    virtual void EraseUnRefEntries() override {
+      return target_->EraseUnRefEntries();
+    }
+
+    int num_lookups() { return num_lookups_; }
+
+    int num_found() { return num_found_; }
+
+    int num_inserts() { return num_inserts_; }
+   private:
+    std::shared_ptr<Cache> target_;
+    int num_lookups_;
+    int num_found_;
+    int num_inserts_;
+  };
+
+  std::shared_ptr<MyBlockCache> compressed_cache_;
+  std::shared_ptr<MyBlockCache> uncompressed_cache_;
+  bool compression_enabled_;
+  std::vector<std::string> values_;
+  bool fill_cache_;
+};
+
+TEST_P(DBBasicTestWithParallelIO, MultiGet) {
+  std::vector<std::string> key_data(10);
+  std::vector<Slice> keys;
+  // We cannot resize a PinnableSlice vector, so just set initial size to
+  // largest we think we will need
+  std::vector<PinnableSlice> values(10);
+  std::vector<Status> statuses;
+  ReadOptions ro;
+  ro.fill_cache = fill_cache();
+
+  // Warm up the cache first
+  key_data.emplace_back(Key(0));
+  keys.emplace_back(Slice(key_data.back()));
+  key_data.emplace_back(Key(50));
+  keys.emplace_back(Slice(key_data.back()));
+  statuses.resize(keys.size());
+
+  dbfull()->MultiGet(ro, dbfull()->DefaultColumnFamily(), keys.size(),
+           keys.data(), values.data(), statuses.data(), true);
+  ASSERT_TRUE(CheckValue(0, values[0].ToString()));
+  ASSERT_TRUE(CheckValue(50, values[1].ToString()));
+
+  int random_reads = env_->random_read_counter_.Read();
+  key_data[0] = Key(1);
+  key_data[1] = Key(51);
+  keys[0] = Slice(key_data[0]);
+  keys[1] = Slice(key_data[1]);
+  values[0].Reset();
+  values[1].Reset();
+  dbfull()->MultiGet(ro, dbfull()->DefaultColumnFamily(), keys.size(),
+           keys.data(), values.data(), statuses.data(), true);
+  ASSERT_TRUE(CheckValue(1, values[0].ToString()));
+  ASSERT_TRUE(CheckValue(51, values[1].ToString()));
+
+  int expected_reads = random_reads + (fill_cache() ? 0 : 2);
+  ASSERT_EQ(env_->random_read_counter_.Read(), expected_reads);
+
+  keys.resize(10);
+  statuses.resize(10);
+  std::vector<int> key_ints{1,2,15,16,55,81,82,83,84,85};
+  for (size_t i = 0; i < key_ints.size(); ++i) {
+    key_data[i] = Key(key_ints[i]);
+    keys[i] = Slice(key_data[i]);
+    statuses[i] = Status::OK();
+    values[i].Reset();
+  }
+  dbfull()->MultiGet(ro, dbfull()->DefaultColumnFamily(), keys.size(),
+           keys.data(), values.data(), statuses.data(), true);
+  for (size_t i = 0; i < key_ints.size(); ++i) {
+    ASSERT_OK(statuses[i]);
+    ASSERT_TRUE(CheckValue(key_ints[i], values[i].ToString()));
+  }
+  expected_reads += (fill_cache() ? 2 : 4);
+  ASSERT_EQ(env_->random_read_counter_.Read(), expected_reads);
+}
+
+INSTANTIATE_TEST_CASE_P(
+    ParallelIO, DBBasicTestWithParallelIO,
+    // Params are as follows -
+    // Param 0 - Compressed cache enabled
+    // Param 1 - Uncompressed cache enabled
+    // Param 2 - Data compression enabled
+    // Param 3 - ReadOptions::fill_cache
+    ::testing::Values(std::make_tuple(false, true, true, true),
+                      std::make_tuple(true, true, true, true),
+                      std::make_tuple(false, true, false, true),
+                      std::make_tuple(false, true, true, false),
+                      std::make_tuple(true, true, true, false),
+                      std::make_tuple(false, true, false, false)));
+
+class DBBasicTestWithTimestampWithParam
+    : public DBTestBase,
+      public testing::WithParamInterface<bool> {
+ public:
+  DBBasicTestWithTimestampWithParam()
+      : DBTestBase("/db_basic_test_with_timestamp") {}
+
+ protected:
+  class TestComparator : public Comparator {
+   private:
+    const Comparator* cmp_without_ts_;
+
+   public:
+    explicit TestComparator(size_t ts_sz)
+        : Comparator(ts_sz), cmp_without_ts_(nullptr) {
+      cmp_without_ts_ = BytewiseComparator();
+    }
+
+    const char* Name() const override { return "TestComparator"; }
+
+    void FindShortSuccessor(std::string*) const override {}
+
+    void FindShortestSeparator(std::string*, const Slice&) const override {}
+
+    int Compare(const Slice& a, const Slice& b) const override {
+      int r = CompareWithoutTimestamp(a, b);
+      if (r != 0 || 0 == timestamp_size()) {
+        return r;
+      }
+      return CompareTimestamp(
+          Slice(a.data() + a.size() - timestamp_size(), timestamp_size()),
+          Slice(b.data() + b.size() - timestamp_size(), timestamp_size()));
+    }
+
+    int CompareWithoutTimestamp(const Slice& a, const Slice& b) const override {
+      assert(a.size() >= timestamp_size());
+      assert(b.size() >= timestamp_size());
+      Slice k1 = StripTimestampFromUserKey(a, timestamp_size());
+      Slice k2 = StripTimestampFromUserKey(b, timestamp_size());
+
+      return cmp_without_ts_->Compare(k1, k2);
+    }
+
+    int CompareTimestamp(const Slice& ts1, const Slice& ts2) const override {
+      if (!ts1.data() && !ts2.data()) {
+        return 0;
+      } else if (ts1.data() && !ts2.data()) {
+        return 1;
+      } else if (!ts1.data() && ts2.data()) {
+        return -1;
+      }
+      assert(ts1.size() == ts2.size());
+      uint64_t low1 = 0;
+      uint64_t low2 = 0;
+      uint64_t high1 = 0;
+      uint64_t high2 = 0;
+      auto* ptr1 = const_cast<Slice*>(&ts1);
+      auto* ptr2 = const_cast<Slice*>(&ts2);
+      if (!GetFixed64(ptr1, &low1) || !GetFixed64(ptr1, &high1) ||
+          !GetFixed64(ptr2, &low2) || !GetFixed64(ptr2, &high2)) {
+        assert(false);
+      }
+      if (high1 < high2) {
+        return 1;
+      } else if (high1 > high2) {
+        return -1;
+      }
+      if (low1 < low2) {
+        return 1;
+      } else if (low1 > low2) {
+        return -1;
+      }
+      return 0;
+    }
+  };
+
+  Slice EncodeTimestamp(uint64_t low, uint64_t high, std::string* ts) {
+    assert(nullptr != ts);
+    ts->clear();
+    PutFixed64(ts, low);
+    PutFixed64(ts, high);
+    assert(ts->size() == sizeof(low) + sizeof(high));
+    return Slice(*ts);
+  }
+};
+
+TEST_P(DBBasicTestWithTimestampWithParam, PutAndGet) {
+  const int kNumKeysPerFile = 8192;
+  const size_t kNumTimestamps = 6;
+  bool memtable_only = GetParam();
+  Options options = CurrentOptions();
+  options.create_if_missing = true;
+  options.env = env_;
+  options.memtable_factory.reset(new SpecialSkipListFactory(kNumKeysPerFile));
+  std::string tmp;
+  size_t ts_sz = EncodeTimestamp(0, 0, &tmp).size();
+  TestComparator test_cmp(ts_sz);
+  options.comparator = &test_cmp;
+  BlockBasedTableOptions bbto;
+  bbto.filter_policy.reset(NewBloomFilterPolicy(
+      10 /*bits_per_key*/, false /*use_block_based_builder*/));
+  bbto.whole_key_filtering = true;
+  options.table_factory.reset(NewBlockBasedTableFactory(bbto));
+  DestroyAndReopen(options);
+  CreateAndReopenWithCF({"pikachu"}, options);
+  size_t num_cfs = handles_.size();
+  ASSERT_EQ(2, num_cfs);
+  std::vector<std::string> write_ts_strs(kNumTimestamps);
+  std::vector<std::string> read_ts_strs(kNumTimestamps);
+  std::vector<Slice> write_ts_list;
+  std::vector<Slice> read_ts_list;
+
+  for (size_t i = 0; i != kNumTimestamps; ++i) {
+    write_ts_list.emplace_back(EncodeTimestamp(i * 2, 0, &write_ts_strs[i]));
+    read_ts_list.emplace_back(EncodeTimestamp(1 + i * 2, 0, &read_ts_strs[i]));
+    const Slice& write_ts = write_ts_list.back();
+    WriteOptions wopts;
+    wopts.timestamp = &write_ts;
+    for (int cf = 0; cf != static_cast<int>(num_cfs); ++cf) {
+      for (size_t j = 0; j != (kNumKeysPerFile - 1) / kNumTimestamps; ++j) {
+        ASSERT_OK(Put(cf, "key" + std::to_string(j),
+                      "value_" + std::to_string(j) + "_" + std::to_string(i),
+                      wopts));
+      }
+      if (!memtable_only) {
+        ASSERT_OK(Flush(cf));
+      }
+    }
+  }
+  const auto& verify_db_func = [&]() {
+    for (size_t i = 0; i != kNumTimestamps; ++i) {
+      ReadOptions ropts;
+      ropts.timestamp = &read_ts_list[i];
+      for (int cf = 0; cf != static_cast<int>(num_cfs); ++cf) {
+        ColumnFamilyHandle* cfh = handles_[cf];
+        for (size_t j = 0; j != (kNumKeysPerFile - 1) / kNumTimestamps; ++j) {
+          std::string value;
+          ASSERT_OK(db_->Get(ropts, cfh, "key" + std::to_string(j), &value));
+          ASSERT_EQ("value_" + std::to_string(j) + "_" + std::to_string(i),
+                    value);
+        }
+      }
+    }
+  };
+  verify_db_func();
+}
+
+INSTANTIATE_TEST_CASE_P(Timestamp, DBBasicTestWithTimestampWithParam,
+                        ::testing::Bool());
+
 }  // namespace rocksdb

 int main(int argc, char** argv) {
@@ -346,14 +346,14 @@ TEST_F(DBBlockCacheTest, IndexAndFilterBlocksStats) {
  options.statistics = rocksdb::CreateDBStatistics();
  BlockBasedTableOptions table_options;
  table_options.cache_index_and_filter_blocks = true;
-  // 200 bytes are enough to hold the first two blocks
-  std::shared_ptr<Cache> cache = NewLRUCache(200, 0, false);
+  // 500 bytes are enough to hold the first two blocks
+  std::shared_ptr<Cache> cache = NewLRUCache(500, 0, false);
  table_options.block_cache = cache;
  table_options.filter_policy.reset(NewBloomFilterPolicy(20, true));
  options.table_factory.reset(new BlockBasedTableFactory(table_options));
  CreateAndReopenWithCF({"pikachu"}, options);

-  ASSERT_OK(Put(1, "key", "val"));
+  ASSERT_OK(Put(1, "longer_key", "val"));
  // Create a new table
  ASSERT_OK(Flush(1));
  size_t index_bytes_insert =
@@ -365,9 +365,14 @@ TEST_F(DBBlockCacheTest, IndexAndFilterBlocksStats) {
  ASSERT_EQ(cache->GetUsage(), index_bytes_insert + filter_bytes_insert);
  // set the cache capacity to the current usage
  cache->SetCapacity(index_bytes_insert + filter_bytes_insert);
-  ASSERT_EQ(TestGetTickerCount(options, BLOCK_CACHE_INDEX_BYTES_EVICT), 0);
-  ASSERT_EQ(TestGetTickerCount(options, BLOCK_CACHE_FILTER_BYTES_EVICT), 0);
-  ASSERT_OK(Put(1, "key2", "val"));
+  // The index and filter eviction statistics were broken by the refactoring
+  // that moved the readers out of the block cache. Disabling these until we can
+  // bring the stats back.
+  // ASSERT_EQ(TestGetTickerCount(options, BLOCK_CACHE_INDEX_BYTES_EVICT), 0);
+  // ASSERT_EQ(TestGetTickerCount(options, BLOCK_CACHE_FILTER_BYTES_EVICT), 0);
+  // Note that the second key needs to be no longer than the first one.
+  // Otherwise the second index block may not fit in cache.
+  ASSERT_OK(Put(1, "key", "val"));
  // Create a new table
  ASSERT_OK(Flush(1));
  // cache evicted old index and block entries
@@ -375,10 +380,13 @@ TEST_F(DBBlockCacheTest, IndexAndFilterBlocksStats) {
            index_bytes_insert);
  ASSERT_GT(TestGetTickerCount(options, BLOCK_CACHE_FILTER_BYTES_INSERT),
            filter_bytes_insert);
-  ASSERT_EQ(TestGetTickerCount(options, BLOCK_CACHE_INDEX_BYTES_EVICT),
-            index_bytes_insert);
-  ASSERT_EQ(TestGetTickerCount(options, BLOCK_CACHE_FILTER_BYTES_EVICT),
-            filter_bytes_insert);
+  // The index and filter eviction statistics were broken by the refactoring
+  // that moved the readers out of the block cache. Disabling these until we can
+  // bring the stats back.
+  // ASSERT_EQ(TestGetTickerCount(options, BLOCK_CACHE_INDEX_BYTES_EVICT),
+  //           index_bytes_insert);
+  // ASSERT_EQ(TestGetTickerCount(options, BLOCK_CACHE_FILTER_BYTES_EVICT),
+  //           filter_bytes_insert);
 }

 namespace {
@@ -395,9 +403,9 @@ class MockCache : public LRUCache {
                 false /*strict_capacity_limit*/, 0.0 /*high_pri_pool_ratio*/) {
  }

-  virtual Status Insert(const Slice& key, void* value, size_t charge,
-                        void (*deleter)(const Slice& key, void* value),
-                        Handle** handle, Priority priority) override {
+  Status Insert(const Slice& key, void* value, size_t charge,
+                void (*deleter)(const Slice& key, void* value), Handle** handle,
+                Priority priority) override {
    if (priority == Priority::LOW) {
      low_pri_insert_count++;
    } else {
@@ -633,7 +641,7 @@ TEST_F(DBBlockCacheTest, CompressedCache) {

 TEST_F(DBBlockCacheTest, CacheCompressionDict) {
  const int kNumFiles = 4;
-  const int kNumEntriesPerFile = 32;
+  const int kNumEntriesPerFile = 128;
  const int kNumBytesPerEntry = 1024;

  // Try all the available libraries that support dictionary compression
@@ -32,7 +32,7 @@ class DBBloomFilterTestWithParam
 public:
  DBBloomFilterTestWithParam() : DBTestBase("/db_bloom_filter_tests") {}

-  ~DBBloomFilterTestWithParam() {}
+  ~DBBloomFilterTestWithParam() override {}

  void SetUp() override {
    use_block_based_filter_ = std::get<0>(GetParam());
@@ -642,7 +642,7 @@ class WrappedBloom : public FilterPolicy {
  explicit WrappedBloom(int bits_per_key)
      : filter_(NewBloomFilterPolicy(bits_per_key)), counter_(0) {}

-  ~WrappedBloom() { delete filter_; }
+  ~WrappedBloom() override { delete filter_; }

  const char* Name() const override { return "WrappedRocksDbFilterPolicy"; }

@@ -786,6 +786,75 @@ TEST_F(DBBloomFilterTest, PrefixExtractorBlockFilter) {
  delete iter;
 }

+TEST_F(DBBloomFilterTest, MemtableWholeKeyBloomFilter) {
+  // regression test for #2743. the range delete tombstones in memtable should
+  // be added even when Get() skips searching due to its prefix bloom filter
+  const int kMemtableSize = 1 << 20;              // 1MB
+  const int kMemtablePrefixFilterSize = 1 << 13;  // 8KB
+  const int kPrefixLen = 4;
+  Options options = CurrentOptions();
+  options.memtable_prefix_bloom_size_ratio =
+      static_cast<double>(kMemtablePrefixFilterSize) / kMemtableSize;
+  options.prefix_extractor.reset(rocksdb::NewFixedPrefixTransform(kPrefixLen));
+  options.write_buffer_size = kMemtableSize;
+  options.memtable_whole_key_filtering = false;
+  Reopen(options);
+  std::string key1("AAAABBBB");
+  std::string key2("AAAACCCC");  // not in DB
+  std::string key3("AAAADDDD");
+  std::string key4("AAAAEEEE");
+  std::string value1("Value1");
+  std::string value3("Value3");
+  std::string value4("Value4");
+
+  ASSERT_OK(Put(key1, value1, WriteOptions()));
+
+  // check memtable bloom stats
+  ASSERT_EQ("NOT_FOUND", Get(key2));
+  ASSERT_EQ(0, get_perf_context()->bloom_memtable_miss_count);
+  // same prefix, bloom filter false positive
+  ASSERT_EQ(1, get_perf_context()->bloom_memtable_hit_count);
+
+  // enable whole key bloom filter
+  options.memtable_whole_key_filtering = true;
+  Reopen(options);
+  // check memtable bloom stats
+  ASSERT_OK(Put(key3, value3, WriteOptions()));
+  ASSERT_EQ("NOT_FOUND", Get(key2));
+  // whole key bloom filter kicks in and determines it's a miss
+  ASSERT_EQ(1, get_perf_context()->bloom_memtable_miss_count);
+  ASSERT_EQ(1, get_perf_context()->bloom_memtable_hit_count);
+
+  // verify whole key filtering does not depend on prefix_extractor
+  options.prefix_extractor.reset();
+  Reopen(options);
+  // check memtable bloom stats
+  ASSERT_OK(Put(key4, value4, WriteOptions()));
+  ASSERT_EQ("NOT_FOUND", Get(key2));
+  // whole key bloom filter kicks in and determines it's a miss
+  ASSERT_EQ(2, get_perf_context()->bloom_memtable_miss_count);
+  ASSERT_EQ(1, get_perf_context()->bloom_memtable_hit_count);
+}
+
+TEST_F(DBBloomFilterTest, MemtablePrefixBloomOutOfDomain) {
+  constexpr size_t kPrefixSize = 8;
+  const std::string kKey = "key";
+  assert(kKey.size() < kPrefixSize);
+  Options options = CurrentOptions();
+  options.prefix_extractor.reset(NewFixedPrefixTransform(kPrefixSize));
+  options.memtable_prefix_bloom_size_ratio = 0.25;
+  Reopen(options);
+  ASSERT_OK(Put(kKey, "v"));
+  ASSERT_EQ("v", Get(kKey));
+  std::unique_ptr<Iterator> iter(dbfull()->NewIterator(ReadOptions()));
+  iter->Seek(kKey);
+  ASSERT_TRUE(iter->Valid());
+  ASSERT_EQ(kKey, iter->key());
+  iter->SeekForPrev(kKey);
+  ASSERT_TRUE(iter->Valid());
+  ASSERT_EQ(kKey, iter->key());
+}
+
 #ifndef ROCKSDB_LITE
 class BloomStatsTestWithParam
    : public DBBloomFilterTest,
@@ -823,7 +892,7 @@ class BloomStatsTestWithParam
    DestroyAndReopen(options_);
  }

-  ~BloomStatsTestWithParam() {
+  ~BloomStatsTestWithParam() override {
    get_perf_context()->Reset();
    Destroy(options_);
  }
@@ -935,13 +1004,16 @@ TEST_P(BloomStatsTestWithParam, BloomStatsTestWithIter) {
  ASSERT_OK(iter->status());
  ASSERT_TRUE(iter->Valid());
  ASSERT_EQ(value3, iter->value().ToString());
-  ASSERT_EQ(2, get_perf_context()->bloom_sst_hit_count);
+  // The seek doesn't check block-based bloom filter because last index key
+  // starts with the same prefix we're seeking to.
+  uint64_t expected_hits = use_block_based_builder_ ? 1 : 2;
+  ASSERT_EQ(expected_hits, get_perf_context()->bloom_sst_hit_count);

  iter->Seek(key2);
  ASSERT_OK(iter->status());
  ASSERT_TRUE(!iter->Valid());
  ASSERT_EQ(1, get_perf_context()->bloom_sst_miss_count);
-  ASSERT_EQ(2, get_perf_context()->bloom_sst_hit_count);
+  ASSERT_EQ(expected_hits, get_perf_context()->bloom_sst_hit_count);
 }

 INSTANTIATE_TEST_CASE_P(BloomStatsTestWithParam, BloomStatsTestWithParam,
@@ -1023,6 +1095,8 @@ TEST_F(DBBloomFilterTest, PrefixScan) {
    options.max_background_compactions = 2;
    options.create_if_missing = true;
    options.memtable_factory.reset(NewHashSkipListRepFactory(16));
+    assert(!options.unordered_write);
+    // It is incompatible with allow_concurrent_memtable_write=false
    options.allow_concurrent_memtable_write = false;

    BlockBasedTableOptions table_options;
@@ -1266,6 +1340,8 @@ TEST_F(DBBloomFilterTest, DynamicBloomFilterUpperBound) {
    table_options.cache_index_and_filter_blocks = true;
    table_options.filter_policy.reset(
        NewBloomFilterPolicy(10, use_block_based_builder));
+    table_options.index_shortening = BlockBasedTableOptions::
+        IndexShorteningMode::kShortenSeparatorsAndSuccessor;
    options.table_factory.reset(NewBlockBasedTableFactory(table_options));
    DestroyAndReopen(options);

@@ -63,33 +63,33 @@ INSTANTIATE_TEST_CASE_P(DBTestCompactionFilterWithCompactOption,

 class KeepFilter : public CompactionFilter {
 public:
-  virtual bool Filter(int /*level*/, const Slice& /*key*/,
-                      const Slice& /*value*/, std::string* /*new_value*/,
-                      bool* /*value_changed*/) const override {
+  bool Filter(int /*level*/, const Slice& /*key*/, const Slice& /*value*/,
+              std::string* /*new_value*/,
+              bool* /*value_changed*/) const override {
    cfilter_count++;
    return false;
  }

-  virtual const char* Name() const override { return "KeepFilter"; }
+  const char* Name() const override { return "KeepFilter"; }
 };

 class DeleteFilter : public CompactionFilter {
 public:
-  virtual bool Filter(int /*level*/, const Slice& /*key*/,
-                      const Slice& /*value*/, std::string* /*new_value*/,
-                      bool* /*value_changed*/) const override {
+  bool Filter(int /*level*/, const Slice& /*key*/, const Slice& /*value*/,
+              std::string* /*new_value*/,
+              bool* /*value_changed*/) const override {
    cfilter_count++;
    return true;
  }

-  virtual const char* Name() const override { return "DeleteFilter"; }
+  const char* Name() const override { return "DeleteFilter"; }
 };

 class DeleteISFilter : public CompactionFilter {
 public:
-  virtual bool Filter(int /*level*/, const Slice& key, const Slice& /*value*/,
-                      std::string* /*new_value*/,
-                      bool* /*value_changed*/) const override {
+  bool Filter(int /*level*/, const Slice& key, const Slice& /*value*/,
+              std::string* /*new_value*/,
+              bool* /*value_changed*/) const override {
    cfilter_count++;
    int i = std::stoi(key.ToString());
    if (i > 5 && i <= 105) {
@@ -98,20 +98,18 @@ class DeleteISFilter : public CompactionFilter {
    return false;
  }

-  virtual bool IgnoreSnapshots() const override { return true; }
+  bool IgnoreSnapshots() const override { return true; }

-  virtual const char* Name() const override { return "DeleteFilter"; }
+  const char* Name() const override { return "DeleteFilter"; }
 };

 // Skip x if floor(x/10) is even, use range skips. Requires that keys are
 // zero-padded to length 10.
 class SkipEvenFilter : public CompactionFilter {
 public:
-  virtual Decision FilterV2(int /*level*/, const Slice& key,
-                            ValueType /*value_type*/,
-                            const Slice& /*existing_value*/,
-                            std::string* /*new_value*/,
-                            std::string* skip_until) const override {
+  Decision FilterV2(int /*level*/, const Slice& key, ValueType /*value_type*/,
+                    const Slice& /*existing_value*/, std::string* /*new_value*/,
+                    std::string* skip_until) const override {
    cfilter_count++;
    int i = std::stoi(key.ToString());
    if (i / 10 % 2 == 0) {
@@ -124,22 +122,22 @@ class SkipEvenFilter : public CompactionFilter {
    return Decision::kKeep;
  }

-  virtual bool IgnoreSnapshots() const override { return true; }
+  bool IgnoreSnapshots() const override { return true; }

-  virtual const char* Name() const override { return "DeleteFilter"; }
+  const char* Name() const override { return "DeleteFilter"; }
 };

 class DelayFilter : public CompactionFilter {
 public:
  explicit DelayFilter(DBTestBase* d) : db_test(d) {}
-  virtual bool Filter(int /*level*/, const Slice& /*key*/,
-                      const Slice& /*value*/, std::string* /*new_value*/,
-                      bool* /*value_changed*/) const override {
+  bool Filter(int /*level*/, const Slice& /*key*/, const Slice& /*value*/,
+              std::string* /*new_value*/,
+              bool* /*value_changed*/) const override {
    db_test->env_->addon_time_.fetch_add(1000);
    return true;
  }

-  virtual const char* Name() const override { return "DelayFilter"; }
+  const char* Name() const override { return "DelayFilter"; }

 private:
  DBTestBase* db_test;
@@ -149,13 +147,13 @@ class ConditionalFilter : public CompactionFilter {
 public:
  explicit ConditionalFilter(const std::string* filtered_value)
      : filtered_value_(filtered_value) {}
-  virtual bool Filter(int /*level*/, const Slice& /*key*/, const Slice& value,
-                      std::string* /*new_value*/,
-                      bool* /*value_changed*/) const override {
+  bool Filter(int /*level*/, const Slice& /*key*/, const Slice& value,
+              std::string* /*new_value*/,
+              bool* /*value_changed*/) const override {
    return value.ToString() == *filtered_value_;
  }

-  virtual const char* Name() const override { return "ConditionalFilter"; }
+  const char* Name() const override { return "ConditionalFilter"; }

 private:
  const std::string* filtered_value_;
@@ -165,16 +163,15 @@ class ChangeFilter : public CompactionFilter {
 public:
  explicit ChangeFilter() {}

-  virtual bool Filter(int /*level*/, const Slice& /*key*/,
-                      const Slice& /*value*/, std::string* new_value,
-                      bool* value_changed) const override {
+  bool Filter(int /*level*/, const Slice& /*key*/, const Slice& /*value*/,
+              std::string* new_value, bool* value_changed) const override {
    assert(new_value != nullptr);
    *new_value = NEW_VALUE;
    *value_changed = true;
    return false;
  }

-  virtual const char* Name() const override { return "ChangeFilter"; }
+  const char* Name() const override { return "ChangeFilter"; }
 };

 class KeepFilterFactory : public CompactionFilterFactory {
@@ -185,7 +182,7 @@ class KeepFilterFactory : public CompactionFilterFactory {
        check_context_cf_id_(check_context_cf_id),
        compaction_filter_created_(false) {}

-  virtual std::unique_ptr<CompactionFilter> CreateCompactionFilter(
+  std::unique_ptr<CompactionFilter> CreateCompactionFilter(
      const CompactionFilter::Context& context) override {
    if (check_context_) {
      EXPECT_EQ(expect_full_compaction_.load(), context.is_full_compaction);
@@ -200,7 +197,7 @@ class KeepFilterFactory : public CompactionFilterFactory {

  bool compaction_filter_created() const { return compaction_filter_created_; }

-  virtual const char* Name() const override { return "KeepFilterFactory"; }
+  const char* Name() const override { return "KeepFilterFactory"; }
  bool check_context_;
  bool check_context_cf_id_;
  std::atomic_bool expect_full_compaction_;
@@ -211,7 +208,7 @@ class KeepFilterFactory : public CompactionFilterFactory {

 class DeleteFilterFactory : public CompactionFilterFactory {
 public:
-  virtual std::unique_ptr<CompactionFilter> CreateCompactionFilter(
+  std::unique_ptr<CompactionFilter> CreateCompactionFilter(
      const CompactionFilter::Context& context) override {
    if (context.is_manual_compaction) {
      return std::unique_ptr<CompactionFilter>(new DeleteFilter());
@@ -220,13 +217,13 @@ class DeleteFilterFactory : public CompactionFilterFactory {
    }
  }

-  virtual const char* Name() const override { return "DeleteFilterFactory"; }
+  const char* Name() const override { return "DeleteFilterFactory"; }
 };

 // Delete Filter Factory which ignores snapshots
 class DeleteISFilterFactory : public CompactionFilterFactory {
 public:
-  virtual std::unique_ptr<CompactionFilter> CreateCompactionFilter(
+  std::unique_ptr<CompactionFilter> CreateCompactionFilter(
      const CompactionFilter::Context& context) override {
    if (context.is_manual_compaction) {
      return std::unique_ptr<CompactionFilter>(new DeleteISFilter());
@@ -235,12 +232,12 @@ class DeleteISFilterFactory : public CompactionFilterFactory {
    }
  }

-  virtual const char* Name() const override { return "DeleteFilterFactory"; }
+  const char* Name() const override { return "DeleteFilterFactory"; }
 };

 class SkipEvenFilterFactory : public CompactionFilterFactory {
 public:
-  virtual std::unique_ptr<CompactionFilter> CreateCompactionFilter(
+  std::unique_ptr<CompactionFilter> CreateCompactionFilter(
      const CompactionFilter::Context& context) override {
    if (context.is_manual_compaction) {
      return std::unique_ptr<CompactionFilter>(new SkipEvenFilter());
@@ -249,18 +246,18 @@ class SkipEvenFilterFactory : public CompactionFilterFactory {
    }
  }

-  virtual const char* Name() const override { return "SkipEvenFilterFactory"; }
+  const char* Name() const override { return "SkipEvenFilterFactory"; }
 };

 class DelayFilterFactory : public CompactionFilterFactory {
 public:
  explicit DelayFilterFactory(DBTestBase* d) : db_test(d) {}
-  virtual std::unique_ptr<CompactionFilter> CreateCompactionFilter(
+  std::unique_ptr<CompactionFilter> CreateCompactionFilter(
      const CompactionFilter::Context& /*context*/) override {
    return std::unique_ptr<CompactionFilter>(new DelayFilter(db_test));
  }

-  virtual const char* Name() const override { return "DelayFilterFactory"; }
+  const char* Name() const override { return "DelayFilterFactory"; }

 private:
  DBTestBase* db_test;
@@ -271,15 +268,13 @@ class ConditionalFilterFactory : public CompactionFilterFactory {
  explicit ConditionalFilterFactory(const Slice& filtered_value)
      : filtered_value_(filtered_value.ToString()) {}

-  virtual std::unique_ptr<CompactionFilter> CreateCompactionFilter(
+  std::unique_ptr<CompactionFilter> CreateCompactionFilter(
      const CompactionFilter::Context& /*context*/) override {
    return std::unique_ptr<CompactionFilter>(
        new ConditionalFilter(&filtered_value_));
  }

-  virtual const char* Name() const override {
-    return "ConditionalFilterFactory";
-  }
+  const char* Name() const override { return "ConditionalFilterFactory"; }

 private:
  std::string filtered_value_;
@@ -289,12 +284,12 @@ class ChangeFilterFactory : public CompactionFilterFactory {
 public:
  explicit ChangeFilterFactory() {}

-  virtual std::unique_ptr<CompactionFilter> CreateCompactionFilter(
+  std::unique_ptr<CompactionFilter> CreateCompactionFilter(
      const CompactionFilter::Context& /*context*/) override {
    return std::unique_ptr<CompactionFilter>(new ChangeFilter());
  }

-  virtual const char* Name() const override { return "ChangeFilterFactory"; }
+  const char* Name() const override { return "ChangeFilterFactory"; }
 };

 #ifndef ROCKSDB_LITE
@@ -357,7 +352,7 @@ TEST_F(DBTestCompactionFilter, CompactionFilter) {
    }
  }
  ASSERT_EQ(total, 100000);
-  ASSERT_EQ(count, 1);
+  ASSERT_EQ(count, 0);

  // overwrite all the 100K keys once again.
  for (int i = 0; i < 100000; i++) {
@@ -664,7 +659,7 @@ TEST_F(DBTestCompactionFilter, CompactionFilterContextManual) {
      iter->Next();
    }
    ASSERT_EQ(total, 700);
-    ASSERT_EQ(count, 1);
+    ASSERT_EQ(count, 0);
  }
 }
 #endif  // ROCKSDB_LITE
@@ -699,44 +694,7 @@ TEST_F(DBTestCompactionFilter, CompactionFilterContextCfId) {
 }

 #ifndef ROCKSDB_LITE
-// Compaction filters should only be applied to records that are newer than the
-// latest snapshot. This test inserts records and applies a delete filter.
-TEST_F(DBTestCompactionFilter, CompactionFilterSnapshot) {
-  Options options = CurrentOptions();
-  options.compaction_filter_factory = std::make_shared<DeleteFilterFactory>();
-  options.disable_auto_compactions = true;
-  options.create_if_missing = true;
-  DestroyAndReopen(options);
-
-  // Put some data.
-  const Snapshot* snapshot = nullptr;
-  for (int table = 0; table < 4; ++table) {
-    for (int i = 0; i < 10; ++i) {
-      Put(ToString(table * 100 + i), "val");
-    }
-    Flush();
-
-    if (table == 0) {
-      snapshot = db_->GetSnapshot();
-    }
-  }
-  assert(snapshot != nullptr);
-
-  cfilter_count = 0;
-  ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
-  // The filter should delete 10 records.
-  ASSERT_EQ(30U, cfilter_count);
-
-  // Release the snapshot and compact again -> now all records should be
-  // removed.
-  db_->ReleaseSnapshot(snapshot);
-  ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
-  ASSERT_EQ(0U, CountLiveFiles());
-}
-
-// Compaction filters should only be applied to records that are newer than the
-// latest snapshot. However, if the compaction filter asks to ignore snapshots
-// records newer than the snapshot will also be processed
+// Compaction filters aplies to all records, regardless snapshots.
 TEST_F(DBTestCompactionFilter, CompactionFilterIgnoreSnapshot) {
  std::string five = ToString(5);
  Options options = CurrentOptions();
@@ -874,6 +832,38 @@ TEST_F(DBTestCompactionFilter, SkipUntilWithBloomFilter) {
  EXPECT_EQ("v50", val);
 }

+class TestNotSupportedFilter : public CompactionFilter {
+ public:
+  bool Filter(int /*level*/, const Slice& /*key*/, const Slice& /*value*/,
+              std::string* /*new_value*/,
+              bool* /*value_changed*/) const override {
+    return true;
+  }
+
+  const char* Name() const override { return "NotSupported"; }
+  bool IgnoreSnapshots() const override { return false; }
+};
+
+TEST_F(DBTestCompactionFilter, IgnoreSnapshotsFalse) {
+  Options options = CurrentOptions();
+  options.compaction_filter = new TestNotSupportedFilter();
+  DestroyAndReopen(options);
+
+  Put("a", "v10");
+  Put("z", "v20");
+  Flush();
+
+  Put("a", "v10");
+  Put("z", "v20");
+  Flush();
+
+  // Comapction should fail because IgnoreSnapshots() = false
+  EXPECT_TRUE(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr)
+                  .IsNotSupported());
+
+  delete options.compaction_filter;
+}
+
 }  // namespace rocksdb

 int main(int argc, char** argv) {
@@ -13,9 +13,9 @@
 #include "rocksdb/concurrent_task_limiter.h"
 #include "rocksdb/experimental.h"
 #include "rocksdb/utilities/convenience.h"
+#include "test_util/fault_injection_test_env.h"
+#include "test_util/sync_point.h"
 #include "util/concurrent_task_limiter_impl.h"
-#include "util/fault_injection_test_env.h"
-#include "util/sync_point.h"

 namespace rocksdb {

@@ -55,9 +55,9 @@ namespace {
 class FlushedFileCollector : public EventListener {
 public:
  FlushedFileCollector() {}
-  ~FlushedFileCollector() {}
+  ~FlushedFileCollector() override {}

-  virtual void OnFlushCompleted(DB* /*db*/, const FlushJobInfo& info) override {
+  void OnFlushCompleted(DB* /*db*/, const FlushJobInfo& info) override {
    std::lock_guard<std::mutex> lock(mutex_);
    flushed_files_.push_back(info.file_path);
  }
@@ -87,24 +87,23 @@ public:
    }
  }

-  ~CompactionStatsCollector() {}
+  ~CompactionStatsCollector() override {}

-  virtual void OnCompactionCompleted(DB* /* db */,
-      const CompactionJobInfo& info) override {
+  void OnCompactionCompleted(DB* /* db */,
+                             const CompactionJobInfo& info) override {
    int k = static_cast<int>(info.compaction_reason);
    int num_of_reasons = static_cast<int>(CompactionReason::kNumOfReasons);
    assert(k >= 0 && k < num_of_reasons);
    compaction_completed_[k]++;
  }

-  virtual void OnExternalFileIngested(DB* /* db */,
-      const ExternalFileIngestionInfo& /* info */) override {
+  void OnExternalFileIngested(
+      DB* /* db */, const ExternalFileIngestionInfo& /* info */) override {
    int k = static_cast<int>(CompactionReason::kExternalSstIngestion);
    compaction_completed_[k]++;
  }

-  virtual void OnFlushCompleted(DB* /* db */,
-      const FlushJobInfo& /* info */) override {
+  void OnFlushCompleted(DB* /* db */, const FlushJobInfo& /* info */) override {
    int k = static_cast<int>(CompactionReason::kFlush);
    compaction_completed_[k]++;
  }
@@ -354,7 +353,8 @@ TEST_P(DBCompactionTestWithParam, CompactionsPreserveDeletes) {
    CompactRangeOptions cro;
    cro.change_level = true;
    cro.target_level = 2;
-    cro.bottommost_level_compaction = BottommostLevelCompaction::kForce;
+    cro.bottommost_level_compaction =
+        BottommostLevelCompaction::kForceOptimized;

    dbfull()->TEST_WaitForFlushMemTable();
    dbfull()->CompactRange(cro, nullptr, nullptr);
@@ -497,14 +497,14 @@ TEST_F(DBCompactionTest, TestTableReaderForCompaction) {

  // Create new iterator for:
  // (1) 1 for verifying flush results
-  // (2) 3 for compaction input files
-  // (3) 1 for verifying compaction results.
-  ASSERT_EQ(num_new_table_reader, 5);
+  // (2) 1 for verifying compaction results.
+  // (3) New TableReaders will not be created for compaction inputs
+  ASSERT_EQ(num_new_table_reader, 2);

  num_table_cache_lookup = 0;
  num_new_table_reader = 0;
  ASSERT_EQ(Key(1), Get(Key(1)));
-  ASSERT_EQ(num_table_cache_lookup + old_num_table_cache_lookup2, 3);
+  ASSERT_EQ(num_table_cache_lookup + old_num_table_cache_lookup2, 5);
  ASSERT_EQ(num_new_table_reader, 0);

  num_table_cache_lookup = 0;
@@ -512,20 +512,21 @@ TEST_F(DBCompactionTest, TestTableReaderForCompaction) {
  CompactRangeOptions cro;
  cro.change_level = true;
  cro.target_level = 2;
-  cro.bottommost_level_compaction = BottommostLevelCompaction::kForce;
+  cro.bottommost_level_compaction = BottommostLevelCompaction::kForceOptimized;
  db_->CompactRange(cro, nullptr, nullptr);
  // Only verifying compaction outputs issues one table cache lookup
  // for both data block and range deletion block).
  // May preload table cache too.
  ASSERT_GE(num_table_cache_lookup, 1);
  old_num_table_cache_lookup2 = num_table_cache_lookup;
-  // One for compaction input, one for verifying compaction results.
-  ASSERT_EQ(num_new_table_reader, 2);
+  // One for verifying compaction results.
+  // No new iterator created for compaction.
+  ASSERT_EQ(num_new_table_reader, 1);

  num_table_cache_lookup = 0;
  num_new_table_reader = 0;
  ASSERT_EQ(Key(1), Get(Key(1)));
-  ASSERT_EQ(num_table_cache_lookup + old_num_table_cache_lookup2, 2);
+  ASSERT_EQ(num_table_cache_lookup + old_num_table_cache_lookup2, 3);
  ASSERT_EQ(num_new_table_reader, 0);

  rocksdb::SyncPoint::GetInstance()->ClearAllCallBacks();
@@ -2261,6 +2262,8 @@ TEST_P(DBCompactionTestWithParam, ConvertCompactionStyle) {
  CompactRangeOptions compact_options;
  compact_options.change_level = true;
  compact_options.target_level = 0;
+  // cannot use kForceOptimized here because the compaction here is expected
+  // to generate one output file
  compact_options.bottommost_level_compaction =
      BottommostLevelCompaction::kForce;
  compact_options.exclusive_manual_compaction = exclusive_manual_compaction_;
@@ -3040,7 +3043,7 @@ TEST_P(DBCompactionTestWithParam, ForceBottommostLevelCompaction) {
  // then compacte the bottommost level L3=>L3 (non trivial move)
  compact_options = CompactRangeOptions();
  compact_options.bottommost_level_compaction =
-      BottommostLevelCompaction::kForce;
+      BottommostLevelCompaction::kForceOptimized;
  ASSERT_OK(db_->CompactRange(compact_options, nullptr, nullptr));
  ASSERT_EQ("0,0,0,1", FilesPerLevel(0));
  ASSERT_EQ(trivial_move, 4);
@@ -3349,7 +3352,7 @@ TEST_F(DBCompactionTest, CompactBottomLevelFilesWithDeletions) {
  options.level0_file_num_compaction_trigger = kNumLevelFiles;
  // inflate it a bit to account for key/metadata overhead
  options.target_file_size_base = 120 * kNumKeysPerFile * kValueSize / 100;
-  Reopen(options);
+  CreateAndReopenWithCF({"one"}, options);

  Random rnd(301);
  const Snapshot* snapshot = nullptr;
@@ -3380,10 +3383,12 @@ TEST_F(DBCompactionTest, CompactBottomLevelFilesWithDeletions) {
  // just need to bump seqnum so ReleaseSnapshot knows the newest key in the SST
  // files does not need to be preserved in case of a future snapshot.
  ASSERT_OK(Put(Key(0), "val"));
+  ASSERT_NE(kMaxSequenceNumber, dbfull()->bottommost_files_mark_threshold_);
  // release snapshot and wait for compactions to finish. Single-file
  // compactions should be triggered, which reduce the size of each bottom-level
  // file without changing file count.
  db_->ReleaseSnapshot(snapshot);
+  ASSERT_EQ(kMaxSequenceNumber, dbfull()->bottommost_files_mark_threshold_);
  rocksdb::SyncPoint::GetInstance()->SetCallBack(
      "LevelCompactionPicker::PickCompaction:Return", [&](void* arg) {
        Compaction* compaction = reinterpret_cast<Compaction*>(arg);
@@ -3516,6 +3521,234 @@ TEST_F(DBCompactionTest, LevelCompactExpiredTtlFiles) {
  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
 }

+TEST_F(DBCompactionTest, LevelPeriodicCompaction) {
+  const int kNumKeysPerFile = 32;
+  const int kNumLevelFiles = 2;
+  const int kValueSize = 100;
+
+  Options options = CurrentOptions();
+  options.periodic_compaction_seconds = 48 * 60 * 60;  // 2 days
+  options.max_open_files = -1;       // needed for ttl compaction
+  env_->time_elapse_only_sleep_ = false;
+  options.env = env_;
+
+  env_->addon_time_.store(0);
+  DestroyAndReopen(options);
+
+  int periodic_compactions = 0;
+  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+      "LevelCompactionPicker::PickCompaction:Return", [&](void* arg) {
+        Compaction* compaction = reinterpret_cast<Compaction*>(arg);
+        auto compaction_reason = compaction->compaction_reason();
+        if (compaction_reason == CompactionReason::kPeriodicCompaction) {
+          periodic_compactions++;
+        }
+      });
+  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+
+  Random rnd(301);
+  for (int i = 0; i < kNumLevelFiles; ++i) {
+    for (int j = 0; j < kNumKeysPerFile; ++j) {
+      ASSERT_OK(
+          Put(Key(i * kNumKeysPerFile + j), RandomString(&rnd, kValueSize)));
+    }
+    Flush();
+  }
+  dbfull()->TEST_WaitForCompact();
+
+  ASSERT_EQ("2", FilesPerLevel());
+  ASSERT_EQ(0, periodic_compactions);
+
+  // Add 50 hours and do a write
+  env_->addon_time_.fetch_add(50 * 60 * 60);
+  ASSERT_OK(Put("a", "1"));
+  Flush();
+  dbfull()->TEST_WaitForCompact();
+  // Assert that the files stay in the same level
+  ASSERT_EQ("3", FilesPerLevel());
+  // The two old files go through the periodic compaction process
+  ASSERT_EQ(2, periodic_compactions);
+
+  MoveFilesToLevel(1);
+  ASSERT_EQ("0,3", FilesPerLevel());
+
+  // Add another 50 hours and do another write
+  env_->addon_time_.fetch_add(50 * 60 * 60);
+  ASSERT_OK(Put("b", "2"));
+  Flush();
+  dbfull()->TEST_WaitForCompact();
+  ASSERT_EQ("1,3", FilesPerLevel());
+  // The three old files now go through the periodic compaction process. 2 + 3.
+  ASSERT_EQ(5, periodic_compactions);
+
+  // Add another 50 hours and do another write
+  env_->addon_time_.fetch_add(50 * 60 * 60);
+  ASSERT_OK(Put("c", "3"));
+  Flush();
+  dbfull()->TEST_WaitForCompact();
+  ASSERT_EQ("2,3", FilesPerLevel());
+  // The four old files now go through the periodic compaction process. 5 + 4.
+  ASSERT_EQ(9, periodic_compactions);
+
+  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+}
+
+TEST_F(DBCompactionTest, LevelPeriodicCompactionWithOldDB) {
+  // This test makes sure that periodic compactions are working with a DB
+  // where file_creation_time of some files is 0.
+  // After compactions the new files are created with a valid file_creation_time
+
+  const int kNumKeysPerFile = 32;
+  const int kNumFiles = 4;
+  const int kValueSize = 100;
+
+  Options options = CurrentOptions();
+  options.max_open_files = -1;  // needed for ttl compaction
+  env_->time_elapse_only_sleep_ = false;
+  options.env = env_;
+
+  env_->addon_time_.store(0);
+  DestroyAndReopen(options);
+
+  int periodic_compactions = 0;
+  bool set_file_creation_time_to_zero = true;
+  bool set_creation_time_to_zero = true;
+  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+      "LevelCompactionPicker::PickCompaction:Return", [&](void* arg) {
+        Compaction* compaction = reinterpret_cast<Compaction*>(arg);
+        auto compaction_reason = compaction->compaction_reason();
+        if (compaction_reason == CompactionReason::kPeriodicCompaction) {
+          periodic_compactions++;
+        }
+      });
+  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+      "PropertyBlockBuilder::AddTableProperty:Start", [&](void* arg) {
+        TableProperties* props = reinterpret_cast<TableProperties*>(arg);
+        if (set_file_creation_time_to_zero) {
+          props->file_creation_time = 0;
+        }
+        if (set_creation_time_to_zero) {
+          props->creation_time = 0;
+        }
+      });
+  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+
+  Random rnd(301);
+  for (int i = 0; i < kNumFiles; ++i) {
+    for (int j = 0; j < kNumKeysPerFile; ++j) {
+      ASSERT_OK(
+          Put(Key(i * kNumKeysPerFile + j), RandomString(&rnd, kValueSize)));
+    }
+    Flush();
+    // Move the first two files to L2.
+    if (i == 1) {
+      MoveFilesToLevel(2);
+      set_creation_time_to_zero = false;
+    }
+  }
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
+
+  ASSERT_EQ("2,0,2", FilesPerLevel());
+  ASSERT_EQ(0, periodic_compactions);
+
+  Close();
+
+  set_file_creation_time_to_zero = false;
+  // Forward the clock by 2 days.
+  env_->addon_time_.fetch_add(2 * 24 * 60 * 60);
+  options.periodic_compaction_seconds = 1 * 24 * 60 * 60;  // 1 day
+
+  Reopen(options);
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
+  ASSERT_EQ("2,0,2", FilesPerLevel());
+  // Make sure that all files go through periodic compaction.
+  ASSERT_EQ(kNumFiles, periodic_compactions);
+
+  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+}
+
+TEST_F(DBCompactionTest, LevelPeriodicAndTtlCompaction) {
+  const int kNumKeysPerFile = 32;
+  const int kNumLevelFiles = 2;
+  const int kValueSize = 100;
+
+  Options options = CurrentOptions();
+  options.ttl = 10 * 60 * 60;  // 10 hours
+  options.periodic_compaction_seconds = 48 * 60 * 60;  // 2 days
+  options.max_open_files = -1;   // needed for both periodic and ttl compactions
+  env_->time_elapse_only_sleep_ = false;
+  options.env = env_;
+
+  env_->addon_time_.store(0);
+  DestroyAndReopen(options);
+
+  int periodic_compactions = 0;
+  int ttl_compactions = 0;
+  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+      "LevelCompactionPicker::PickCompaction:Return", [&](void* arg) {
+        Compaction* compaction = reinterpret_cast<Compaction*>(arg);
+        auto compaction_reason = compaction->compaction_reason();
+        if (compaction_reason == CompactionReason::kPeriodicCompaction) {
+          periodic_compactions++;
+        } else if (compaction_reason == CompactionReason::kTtl) {
+          ttl_compactions++;
+        }
+      });
+  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+
+  Random rnd(301);
+  for (int i = 0; i < kNumLevelFiles; ++i) {
+    for (int j = 0; j < kNumKeysPerFile; ++j) {
+      ASSERT_OK(
+          Put(Key(i * kNumKeysPerFile + j), RandomString(&rnd, kValueSize)));
+    }
+    Flush();
+  }
+  dbfull()->TEST_WaitForCompact();
+
+  MoveFilesToLevel(3);
+
+  ASSERT_EQ("0,0,0,2", FilesPerLevel());
+  ASSERT_EQ(0, periodic_compactions);
+  ASSERT_EQ(0, ttl_compactions);
+
+  // Add some time greater than periodic_compaction_time.
+  env_->addon_time_.fetch_add(50 * 60 * 60);
+  ASSERT_OK(Put("a", "1"));
+  Flush();
+  dbfull()->TEST_WaitForCompact();
+  // Files in the bottom level go through periodic compactions.
+  ASSERT_EQ("1,0,0,2", FilesPerLevel());
+  ASSERT_EQ(2, periodic_compactions);
+  ASSERT_EQ(0, ttl_compactions);
+
+  // Add a little more time than ttl
+  env_->addon_time_.fetch_add(11 * 60 * 60);
+  ASSERT_OK(Put("b", "1"));
+  Flush();
+  dbfull()->TEST_WaitForCompact();
+  // Notice that the previous file in level 1 falls down to the bottom level
+  // due to ttl compactions, one level at a time.
+  // And bottom level files don't get picked up for ttl compactions.
+  ASSERT_EQ("1,0,0,3", FilesPerLevel());
+  ASSERT_EQ(2, periodic_compactions);
+  ASSERT_EQ(3, ttl_compactions);
+
+  // Add some time greater than periodic_compaction_time.
+  env_->addon_time_.fetch_add(50 * 60 * 60);
+  ASSERT_OK(Put("c", "1"));
+  Flush();
+  dbfull()->TEST_WaitForCompact();
+  // Previous L0 file falls one level at a time to bottom level due to ttl.
+  // And all 4 bottom files go through periodic compactions.
+  ASSERT_EQ("1,0,0,4", FilesPerLevel());
+  ASSERT_EQ(6, periodic_compactions);
+  ASSERT_EQ(6, ttl_compactions);
+
+  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+}
+
+
 TEST_F(DBCompactionTest, CompactRangeDelayedByL0FileCount) {
  // Verify that, when `CompactRangeOptions::allow_write_stall == false`, manual
  // compaction only triggers flush after it's sure stall won't be triggered for
@@ -3658,11 +3891,17 @@ TEST_F(DBCompactionTest, CompactRangeShutdownWhileDelayed) {
      }
      Flush(1);
    }
-    auto manual_compaction_thread = port::Thread([this]() {
+    auto manual_compaction_thread = port::Thread([this, i]() {
      CompactRangeOptions cro;
      cro.allow_write_stall = false;
-      ASSERT_TRUE(db_->CompactRange(cro, handles_[1], nullptr, nullptr)
-                      .IsShutdownInProgress());
+      Status s = db_->CompactRange(cro, handles_[1], nullptr, nullptr);
+      if (i == 0) {
+        ASSERT_TRUE(db_->CompactRange(cro, handles_[1], nullptr, nullptr)
+                        .IsColumnFamilyDropped());
+      } else {
+        ASSERT_TRUE(db_->CompactRange(cro, handles_[1], nullptr, nullptr)
+                        .IsShutdownInProgress());
+      }
    });

    TEST_SYNC_POINT(
@@ -4101,12 +4340,6 @@ TEST_P(DBCompactionDirectIOTest, DirectIO) {
  options.env = new MockEnv(Env::Default());
  Reopen(options);
  bool readahead = false;
-  SyncPoint::GetInstance()->SetCallBack(
-      "TableCache::NewIterator:for_compaction", [&](void* arg) {
-        bool* use_direct_reads = static_cast<bool*>(arg);
-        ASSERT_EQ(*use_direct_reads,
-                  options.use_direct_reads);
-      });
  SyncPoint::GetInstance()->SetCallBack(
      "CompactionJob::OpenCompactionOutputFile", [&](void* arg) {
        bool* use_direct_writes = static_cast<bool*>(arg);
@@ -4187,14 +4420,14 @@ class NoopMergeOperator : public MergeOperator {
 public:
  NoopMergeOperator() {}

-  virtual bool FullMergeV2(const MergeOperationInput& /*merge_in*/,
-                           MergeOperationOutput* merge_out) const override {
+  bool FullMergeV2(const MergeOperationInput& /*merge_in*/,
+                   MergeOperationOutput* merge_out) const override {
    std::string val("bar");
    merge_out->new_value = val;
    return true;
  }

-  virtual const char* Name() const override { return "Noop"; }
+  const char* Name() const override { return "Noop"; }
 };

 TEST_F(DBCompactionTest, PartialManualCompaction) {
@@ -4223,7 +4456,7 @@ TEST_F(DBCompactionTest, PartialManualCompaction) {
      {{"max_compaction_bytes", std::to_string(max_compaction_bytes)}}));

  CompactRangeOptions cro;
-  cro.bottommost_level_compaction = BottommostLevelCompaction::kForce;
+  cro.bottommost_level_compaction = BottommostLevelCompaction::kForceOptimized;
  dbfull()->CompactRange(cro, nullptr, nullptr);
 }

@@ -4267,6 +4500,88 @@ TEST_F(DBCompactionTest, ManualCompactionFailsInReadOnlyMode) {
  Close();
 }

+// ManualCompactionBottomLevelOptimization tests the bottom level manual
+// compaction optimization to skip recompacting files created by Ln-1 to Ln
+// compaction
+TEST_F(DBCompactionTest, ManualCompactionBottomLevelOptimized) {
+  Options opts = CurrentOptions();
+  opts.num_levels = 3;
+  opts.level0_file_num_compaction_trigger = 5;
+  opts.compression = kNoCompression;
+  opts.merge_operator.reset(new NoopMergeOperator());
+  opts.target_file_size_base = 1024;
+  opts.max_bytes_for_level_multiplier = 2;
+  opts.disable_auto_compactions = true;
+  DestroyAndReopen(opts);
+  ColumnFamilyHandleImpl* cfh =
+      static_cast<ColumnFamilyHandleImpl*>(dbfull()->DefaultColumnFamily());
+  ColumnFamilyData* cfd = cfh->cfd();
+  InternalStats* internal_stats_ptr = cfd->internal_stats();
+  ASSERT_NE(internal_stats_ptr, nullptr);
+
+  Random rnd(301);
+  for (auto i = 0; i < 8; ++i) {
+    for (auto j = 0; j < 10; ++j) {
+      ASSERT_OK(
+          Put("foo" + std::to_string(i * 10 + j), RandomString(&rnd, 1024)));
+    }
+    Flush();
+  }
+
+  MoveFilesToLevel(2);
+
+  for (auto i = 0; i < 8; ++i) {
+    for (auto j = 0; j < 10; ++j) {
+      ASSERT_OK(
+          Put("bar" + std::to_string(i * 10 + j), RandomString(&rnd, 1024)));
+    }
+    Flush();
+  }
+  const std::vector<InternalStats::CompactionStats>& comp_stats =
+      internal_stats_ptr->TEST_GetCompactionStats();
+  int num = comp_stats[2].num_input_files_in_output_level;
+  ASSERT_EQ(num, 0);
+
+  CompactRangeOptions cro;
+  cro.bottommost_level_compaction = BottommostLevelCompaction::kForceOptimized;
+  dbfull()->CompactRange(cro, nullptr, nullptr);
+
+  const std::vector<InternalStats::CompactionStats>& comp_stats2 =
+      internal_stats_ptr->TEST_GetCompactionStats();
+  num = comp_stats2[2].num_input_files_in_output_level;
+  ASSERT_EQ(num, 0);
+}
+
+TEST_F(DBCompactionTest, CompactionDuringShutdown) {
+  Options opts = CurrentOptions();
+  opts.level0_file_num_compaction_trigger = 2;
+  opts.disable_auto_compactions = true;
+  DestroyAndReopen(opts);
+  ColumnFamilyHandleImpl* cfh =
+      static_cast<ColumnFamilyHandleImpl*>(dbfull()->DefaultColumnFamily());
+  ColumnFamilyData* cfd = cfh->cfd();
+  InternalStats* internal_stats_ptr = cfd->internal_stats();
+  ASSERT_NE(internal_stats_ptr, nullptr);
+
+  Random rnd(301);
+  for (auto i = 0; i < 2; ++i) {
+    for (auto j = 0; j < 10; ++j) {
+      ASSERT_OK(
+          Put("foo" + std::to_string(i * 10 + j), RandomString(&rnd, 1024)));
+    }
+    Flush();
+  }
+
+  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+      "DBImpl::BackgroundCompaction:NonTrivial:BeforeRun",
+      [&](void* /*arg*/) {
+    dbfull()->shutting_down_.store(true);
+  });
+  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+  dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr);
+  ASSERT_OK(dbfull()->error_handler_.GetBGError());
+}
+
 // FixFileIngestionCompactionDeadlock tests and verifies that compaction and
 // file ingestion do not cause deadlock in the event of write stall triggered
 // by number of L0 files reaching level0_stop_writes_trigger.
@@ -7,7 +7,7 @@
 #include "port/stack_trace.h"
 #include "rocksdb/perf_context.h"
 #if !defined(ROCKSDB_LITE)
-#include "util/sync_point.h"
+#include "test_util/sync_point.h"
 #endif
 #include <iostream>
 #include <string>
@@ -6,24 +6,20 @@

 #ifndef ROCKSDB_LITE

-#ifndef __STDC_FORMAT_MACROS
-#define __STDC_FORMAT_MACROS
-#endif
-
-#include <inttypes.h>
+#include <cinttypes>
 #include <stdint.h>
 #include <algorithm>
 #include <string>
-#include "db/db_impl.h"
+#include "db/db_impl/db_impl.h"
 #include "db/job_context.h"
 #include "db/version_set.h"
+#include "file/file_util.h"
+#include "file/filename.h"
 #include "port/port.h"
 #include "rocksdb/db.h"
 #include "rocksdb/env.h"
-#include "util/file_util.h"
-#include "util/filename.h"
+#include "test_util/sync_point.h"
 #include "util/mutexlock.h"
-#include "util/sync_point.h"

 namespace rocksdb {

@@ -61,7 +57,9 @@ Status DBImpl::EnableFileDeletions(bool force) {
  }
  if (file_deletion_enabled) {
    ROCKS_LOG_INFO(immutable_db_options_.info_log, "File Deletions Enabled");
-    PurgeObsoleteFiles(job_context);
+    if (job_context.HaveSomethingToDelete()) {
+      PurgeObsoleteFiles(job_context);
+    }
  } else {
    ROCKS_LOG_WARN(immutable_db_options_.info_log,
                   "File Deletions Enable, but not really enabled. Counter: %d",
@@ -9,8 +9,8 @@

 #include "db/db_test_util.h"
 #include "port/stack_trace.h"
-#include "util/fault_injection_test_env.h"
-#include "util/sync_point.h"
+#include "test_util/fault_injection_test_env.h"
+#include "test_util/sync_point.h"

 namespace rocksdb {

@@ -290,6 +290,39 @@ TEST_F(DBFlushTest, ManualFlushFailsInReadOnlyMode) {
  Close();
 }

+TEST_F(DBFlushTest, CFDropRaceWithWaitForFlushMemTables) {
+  Options options = CurrentOptions();
+  options.create_if_missing = true;
+  CreateAndReopenWithCF({"pikachu"}, options);
+  SyncPoint::GetInstance()->DisableProcessing();
+  SyncPoint::GetInstance()->LoadDependency(
+      {{"DBImpl::FlushMemTable:AfterScheduleFlush",
+        "DBFlushTest::CFDropRaceWithWaitForFlushMemTables:BeforeDrop"},
+       {"DBFlushTest::CFDropRaceWithWaitForFlushMemTables:AfterFree",
+        "DBImpl::BackgroundCallFlush:start"},
+       {"DBImpl::BackgroundCallFlush:start",
+        "DBImpl::FlushMemTable:BeforeWaitForBgFlush"}});
+  SyncPoint::GetInstance()->EnableProcessing();
+  ASSERT_EQ(2, handles_.size());
+  ASSERT_OK(Put(1, "key", "value"));
+  auto* cfd = static_cast<ColumnFamilyHandleImpl*>(handles_[1])->cfd();
+  port::Thread drop_cf_thr([&]() {
+    TEST_SYNC_POINT(
+        "DBFlushTest::CFDropRaceWithWaitForFlushMemTables:BeforeDrop");
+    ASSERT_OK(dbfull()->DropColumnFamily(handles_[1]));
+    ASSERT_OK(dbfull()->DestroyColumnFamilyHandle(handles_[1]));
+    handles_.resize(1);
+    TEST_SYNC_POINT(
+        "DBFlushTest::CFDropRaceWithWaitForFlushMemTables:AfterFree");
+  });
+  FlushOptions flush_opts;
+  flush_opts.allow_write_stall = true;
+  ASSERT_NOK(dbfull()->TEST_FlushMemTable(cfd, flush_opts));
+  drop_cf_thr.join();
+  Close();
+  SyncPoint::GetInstance()->DisableProcessing();
+}
+
 TEST_P(DBAtomicFlushTest, ManualAtomicFlush) {
  Options options = CurrentOptions();
  options.create_if_missing = true;
@@ -431,7 +464,7 @@ TEST_P(DBAtomicFlushTest, FlushMultipleCFs_DropSomeBeforeRequestFlush) {
    cf_ids.push_back(cf_id);
  }
  ASSERT_OK(dbfull()->DropColumnFamily(handles_[1]));
-  ASSERT_TRUE(Flush(cf_ids).IsShutdownInProgress());
+  ASSERT_TRUE(Flush(cf_ids).IsColumnFamilyDropped());
  Destroy(options);
 }

@@ -488,6 +521,106 @@ TEST_P(DBAtomicFlushTest,
  Destroy(options);
 }

+TEST_P(DBAtomicFlushTest, TriggerFlushAndClose) {
+  bool atomic_flush = GetParam();
+  if (!atomic_flush) {
+    return;
+  }
+  const int kNumKeysTriggerFlush = 4;
+  Options options = CurrentOptions();
+  options.create_if_missing = true;
+  options.atomic_flush = atomic_flush;
+  options.memtable_factory.reset(
+      new SpecialSkipListFactory(kNumKeysTriggerFlush));
+  CreateAndReopenWithCF({"pikachu"}, options);
+
+  for (int i = 0; i != kNumKeysTriggerFlush; ++i) {
+    ASSERT_OK(Put(0, "key" + std::to_string(i), "value" + std::to_string(i)));
+  }
+  SyncPoint::GetInstance()->DisableProcessing();
+  SyncPoint::GetInstance()->ClearAllCallBacks();
+  SyncPoint::GetInstance()->EnableProcessing();
+  ASSERT_OK(Put(0, "key", "value"));
+  Close();
+
+  ReopenWithColumnFamilies({kDefaultColumnFamilyName, "pikachu"}, options);
+  ASSERT_EQ("value", Get(0, "key"));
+}
+
+TEST_P(DBAtomicFlushTest, PickMemtablesRaceWithBackgroundFlush) {
+  bool atomic_flush = GetParam();
+  Options options = CurrentOptions();
+  options.create_if_missing = true;
+  options.atomic_flush = atomic_flush;
+  options.max_write_buffer_number = 4;
+  // Set min_write_buffer_number_to_merge to be greater than 1, so that
+  // a column family with one memtable in the imm will not cause IsFlushPending
+  // to return true when flush_requested_ is false.
+  options.min_write_buffer_number_to_merge = 2;
+  CreateAndReopenWithCF({"pikachu"}, options);
+  ASSERT_EQ(2, handles_.size());
+  ASSERT_OK(dbfull()->PauseBackgroundWork());
+  ASSERT_OK(Put(0, "key00", "value00"));
+  ASSERT_OK(Put(1, "key10", "value10"));
+  FlushOptions flush_opts;
+  flush_opts.wait = false;
+  ASSERT_OK(dbfull()->Flush(flush_opts, handles_));
+  ASSERT_OK(Put(0, "key01", "value01"));
+  // Since max_write_buffer_number is 4, the following flush won't cause write
+  // stall.
+  ASSERT_OK(dbfull()->Flush(flush_opts));
+  ASSERT_OK(dbfull()->DropColumnFamily(handles_[1]));
+  ASSERT_OK(dbfull()->DestroyColumnFamilyHandle(handles_[1]));
+  handles_[1] = nullptr;
+  ASSERT_OK(dbfull()->ContinueBackgroundWork());
+  ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable(handles_[0]));
+  delete handles_[0];
+  handles_.clear();
+}
+
+TEST_P(DBAtomicFlushTest, CFDropRaceWithWaitForFlushMemTables) {
+  bool atomic_flush = GetParam();
+  if (!atomic_flush) {
+    return;
+  }
+  Options options = CurrentOptions();
+  options.create_if_missing = true;
+  options.atomic_flush = atomic_flush;
+  CreateAndReopenWithCF({"pikachu"}, options);
+  SyncPoint::GetInstance()->DisableProcessing();
+  SyncPoint::GetInstance()->LoadDependency(
+      {{"DBImpl::AtomicFlushMemTables:AfterScheduleFlush",
+        "DBAtomicFlushTest::CFDropRaceWithWaitForFlushMemTables:BeforeDrop"},
+       {"DBAtomicFlushTest::CFDropRaceWithWaitForFlushMemTables:AfterFree",
+        "DBImpl::BackgroundCallFlush:start"},
+       {"DBImpl::BackgroundCallFlush:start",
+        "DBImpl::AtomicFlushMemTables:BeforeWaitForBgFlush"}});
+  SyncPoint::GetInstance()->EnableProcessing();
+  ASSERT_EQ(2, handles_.size());
+  ASSERT_OK(Put(0, "key", "value"));
+  ASSERT_OK(Put(1, "key", "value"));
+  auto* cfd_default =
+      static_cast<ColumnFamilyHandleImpl*>(dbfull()->DefaultColumnFamily())
+          ->cfd();
+  auto* cfd_pikachu = static_cast<ColumnFamilyHandleImpl*>(handles_[1])->cfd();
+  port::Thread drop_cf_thr([&]() {
+    TEST_SYNC_POINT(
+        "DBAtomicFlushTest::CFDropRaceWithWaitForFlushMemTables:BeforeDrop");
+    ASSERT_OK(dbfull()->DropColumnFamily(handles_[1]));
+    delete handles_[1];
+    handles_.resize(1);
+    TEST_SYNC_POINT(
+        "DBAtomicFlushTest::CFDropRaceWithWaitForFlushMemTables:AfterFree");
+  });
+  FlushOptions flush_opts;
+  flush_opts.allow_write_stall = true;
+  ASSERT_OK(dbfull()->TEST_AtomicFlushMemTables({cfd_default, cfd_pikachu},
+                                                flush_opts));
+  drop_cf_thr.join();
+  Close();
+  SyncPoint::GetInstance()->DisableProcessing();
+}
+
 INSTANTIATE_TEST_CASE_P(DBFlushDirectIOTest, DBFlushDirectIOTest,
                        testing::Bool());

@@ -6,23 +6,20 @@
 // Copyright (c) 2011 The LevelDB Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file. See the AUTHORS file for names of contributors.
-#include "db/db_impl.h"
+#include "db/db_impl/db_impl.h"

-#ifndef __STDC_FORMAT_MACROS
-#define __STDC_FORMAT_MACROS
-#endif
-#include <inttypes.h>
+#include <cinttypes>

 #include "db/builder.h"
 #include "db/error_handler.h"
 #include "db/event_helpers.h"
+#include "file/sst_file_manager_impl.h"
 #include "monitoring/iostats_context_imp.h"
 #include "monitoring/perf_context_imp.h"
 #include "monitoring/thread_status_updater.h"
 #include "monitoring/thread_status_util.h"
+#include "test_util/sync_point.h"
 #include "util/concurrent_task_limiter_impl.h"
-#include "util/sst_file_manager_impl.h"
-#include "util/sync_point.h"

 namespace rocksdb {

@@ -110,6 +107,13 @@ Status DBImpl::SyncClosedLogs(JobContext* job_context) {
      if (!s.ok()) {
        break;
      }
+
+      if (immutable_db_options_.recycle_log_file_num > 0) {
+        s = log->Close();
+        if (!s.ok()) {
+          break;
+        }
+      }
    }
    if (s.ok()) {
      s = directories_.GetWalDir()->Fsync();
@@ -135,12 +139,12 @@ Status DBImpl::FlushMemTableToOutputFile(
    SuperVersionContext* superversion_context,
    std::vector<SequenceNumber>& snapshot_seqs,
    SequenceNumber earliest_write_conflict_snapshot,
-    SnapshotChecker* snapshot_checker, LogBuffer* log_buffer) {
+    SnapshotChecker* snapshot_checker, LogBuffer* log_buffer,
+    Env::Priority thread_pri) {
  mutex_.AssertHeld();
  assert(cfd->imm()->NumNotFlushed() != 0);
  assert(cfd->imm()->IsFlushPending());

-
  FlushJob flush_job(
      dbname_, cfd, immutable_db_options_, mutable_cf_options,
      nullptr /* memtable_id */, env_options_for_compaction_, versions_.get(),
@@ -149,7 +153,7 @@ Status DBImpl::FlushMemTableToOutputFile(
      GetDataDir(cfd, 0U),
      GetCompressionFlush(*cfd->ioptions(), mutable_cf_options), stats_,
      &event_logger_, mutable_cf_options.report_bg_io_stats,
-      true /* sync_output_directory */, true /* write_manifest */);
+      true /* sync_output_directory */, true /* write_manifest */, thread_pri);

  FileMetaData file_meta;

@@ -201,7 +205,7 @@ Status DBImpl::FlushMemTableToOutputFile(
                     cfd->current()->storage_info()->LevelSummary(&tmp));
  }

-  if (!s.ok() && !s.IsShutdownInProgress()) {
+  if (!s.ok() && !s.IsShutdownInProgress() && !s.IsColumnFamilyDropped()) {
    Status new_bg_error = s;
    error_handler_.SetBGError(new_bg_error, BackgroundErrorReason::kFlush);
  }
@@ -218,7 +222,8 @@ Status DBImpl::FlushMemTableToOutputFile(
          cfd->ioptions()->cf_paths[0].path, file_meta.fd.GetNumber());
      sfm->OnAddFile(file_path);
      if (sfm->IsMaxAllowedSpaceReached()) {
-        Status new_bg_error = Status::SpaceLimit("Max allowed space was reached");
+        Status new_bg_error =
+            Status::SpaceLimit("Max allowed space was reached");
        TEST_SYNC_POINT_CALLBACK(
            "DBImpl::FlushMemTableToOutputFile:MaxAllowedSpaceReached",
            &new_bg_error);
@@ -232,10 +237,10 @@ Status DBImpl::FlushMemTableToOutputFile(

 Status DBImpl::FlushMemTablesToOutputFiles(
    const autovector<BGFlushArg>& bg_flush_args, bool* made_progress,
-    JobContext* job_context, LogBuffer* log_buffer) {
+    JobContext* job_context, LogBuffer* log_buffer, Env::Priority thread_pri) {
  if (immutable_db_options_.atomic_flush) {
-    return AtomicFlushMemTablesToOutputFiles(bg_flush_args, made_progress,
-                                             job_context, log_buffer);
+    return AtomicFlushMemTablesToOutputFiles(
+        bg_flush_args, made_progress, job_context, log_buffer, thread_pri);
  }
  std::vector<SequenceNumber> snapshot_seqs;
  SequenceNumber earliest_write_conflict_snapshot;
@@ -250,10 +255,10 @@ Status DBImpl::FlushMemTablesToOutputFiles(
    Status s = FlushMemTableToOutputFile(
        cfd, mutable_cf_options, made_progress, job_context,
        superversion_context, snapshot_seqs, earliest_write_conflict_snapshot,
-        snapshot_checker, log_buffer);
+        snapshot_checker, log_buffer, thread_pri);
    if (!s.ok()) {
      status = s;
-      if (!s.IsShutdownInProgress()) {
+      if (!s.IsShutdownInProgress() && !s.IsColumnFamilyDropped()) {
        // At this point, DB is not shutting down, nor is cfd dropped.
        // Something is wrong, thus we break out of the loop.
        break;
@@ -274,7 +279,7 @@ Status DBImpl::FlushMemTablesToOutputFiles(
 */
 Status DBImpl::AtomicFlushMemTablesToOutputFiles(
    const autovector<BGFlushArg>& bg_flush_args, bool* made_progress,
-    JobContext* job_context, LogBuffer* log_buffer) {
+    JobContext* job_context, LogBuffer* log_buffer, Env::Priority thread_pri) {
  mutex_.AssertHeld();

  autovector<ColumnFamilyData*> cfds;
@@ -296,6 +301,7 @@ Status DBImpl::AtomicFlushMemTablesToOutputFiles(
                     &earliest_write_conflict_snapshot, &snapshot_checker);

  autovector<Directory*> distinct_output_dirs;
+  autovector<std::string> distinct_output_dir_paths;
  std::vector<FlushJob> jobs;
  std::vector<MutableCFOptions> all_mutable_cf_options;
  int num_cfs = static_cast<int>(cfds.size());
@@ -303,18 +309,20 @@ Status DBImpl::AtomicFlushMemTablesToOutputFiles(
  for (int i = 0; i < num_cfs; ++i) {
    auto cfd = cfds[i];
    Directory* data_dir = GetDataDir(cfd, 0U);
+    const std::string& curr_path = cfd->ioptions()->cf_paths[0].path;

    // Add to distinct output directories if eligible. Use linear search. Since
    // the number of elements in the vector is not large, performance should be
    // tolerable.
    bool found = false;
-    for (const auto dir : distinct_output_dirs) {
-      if (dir == data_dir) {
+    for (const auto& path : distinct_output_dir_paths) {
+      if (path == curr_path) {
        found = true;
        break;
      }
    }
    if (!found) {
+      distinct_output_dir_paths.emplace_back(curr_path);
      distinct_output_dirs.emplace_back(data_dir);
    }

@@ -322,31 +330,29 @@ Status DBImpl::AtomicFlushMemTablesToOutputFiles(
    const MutableCFOptions& mutable_cf_options = all_mutable_cf_options.back();
    const uint64_t* max_memtable_id = &(bg_flush_args[i].max_memtable_id_);
    jobs.emplace_back(
-        dbname_, cfds[i], immutable_db_options_, mutable_cf_options,
+        dbname_, cfd, immutable_db_options_, mutable_cf_options,
        max_memtable_id, env_options_for_compaction_, versions_.get(), &mutex_,
        &shutting_down_, snapshot_seqs, earliest_write_conflict_snapshot,
        snapshot_checker, job_context, log_buffer, directories_.GetDbDir(),
        data_dir, GetCompressionFlush(*cfd->ioptions(), mutable_cf_options),
        stats_, &event_logger_, mutable_cf_options.report_bg_io_stats,
-        false /* sync_output_directory */, false /* write_manifest */);
+        false /* sync_output_directory */, false /* write_manifest */,
+        thread_pri);
    jobs.back().PickMemTable();
  }

-  autovector<FileMetaData> file_meta;
+  std::vector<FileMetaData> file_meta(num_cfs);
  Status s;
  assert(num_cfs == static_cast<int>(jobs.size()));

-  for (int i = 0; i != num_cfs; ++i) {
-    file_meta.emplace_back();
-
 #ifndef ROCKSDB_LITE
-    const MutableCFOptions& mutable_cf_options =
-        *cfds[i]->GetLatestMutableCFOptions();
+  for (int i = 0; i != num_cfs; ++i) {
+    const MutableCFOptions& mutable_cf_options = all_mutable_cf_options.at(i);
    // may temporarily unlock and lock the mutex.
    NotifyOnFlushBegin(cfds[i], &file_meta[i], mutable_cf_options,
                       job_context->job_id, jobs[i].GetTableProperties());
-#endif /* !ROCKSDB_LITE */
  }
+#endif /* !ROCKSDB_LITE */

  if (logfile_number_ > 0) {
    // TODO (yanqin) investigate whether we should sync the closed logs for
@@ -383,7 +389,8 @@ Status DBImpl::AtomicFlushMemTablesToOutputFiles(
    for (const auto& e : exec_status) {
      if (!e.second.ok()) {
        s = e.second;
-        if (!e.second.IsShutdownInProgress()) {
+        if (!e.second.IsShutdownInProgress() &&
+            !e.second.IsColumnFamilyDropped()) {
          // If a flush job did not return OK, and the CF is not dropped, and
          // the DB is not shutting down, then we have to return this result to
          // caller later.
@@ -395,12 +402,17 @@ Status DBImpl::AtomicFlushMemTablesToOutputFiles(
    s = error_status.ok() ? s : error_status;
  }

-  if (s.ok() || s.IsShutdownInProgress()) {
+  if (s.IsColumnFamilyDropped()) {
+    s = Status::OK();
+  }
+
+  if (s.ok() || s.IsShutdownInProgress() || s.IsColumnFamilyDropped()) {
    // Sync on all distinct output directories.
    for (auto dir : distinct_output_dirs) {
      if (dir != nullptr) {
-        s = dir->Fsync();
-        if (!s.ok()) {
+        Status error_status = dir->Fsync();
+        if (!error_status.ok()) {
+          s = error_status;
          break;
        }
      }
@@ -450,22 +462,24 @@ Status DBImpl::AtomicFlushMemTablesToOutputFiles(
    autovector<ColumnFamilyData*> tmp_cfds;
    autovector<const autovector<MemTable*>*> mems_list;
    autovector<const MutableCFOptions*> mutable_cf_options_list;
+    autovector<FileMetaData*> tmp_file_meta;
    for (int i = 0; i != num_cfs; ++i) {
      const auto& mems = jobs[i].GetMemTables();
      if (!cfds[i]->IsDropped() && !mems.empty()) {
        tmp_cfds.emplace_back(cfds[i]);
        mems_list.emplace_back(&mems);
        mutable_cf_options_list.emplace_back(&all_mutable_cf_options[i]);
+        tmp_file_meta.emplace_back(&file_meta[i]);
      }
    }

    s = InstallMemtableAtomicFlushResults(
        nullptr /* imm_lists */, tmp_cfds, mutable_cf_options_list, mems_list,
-        versions_.get(), &mutex_, file_meta, &job_context->memtables_to_free,
-        directories_.GetDbDir(), log_buffer);
+        versions_.get(), &mutex_, tmp_file_meta,
+        &job_context->memtables_to_free, directories_.GetDbDir(), log_buffer);
  }

-  if (s.ok() || s.IsShutdownInProgress()) {
+  if (s.ok()) {
    assert(num_cfs ==
           static_cast<int>(job_context->superversion_contexts.size()));
    for (int i = 0; i != num_cfs; ++i) {
@@ -474,7 +488,7 @@ Status DBImpl::AtomicFlushMemTablesToOutputFiles(
      }
      InstallSuperVersionAndScheduleWork(cfds[i],
                                         &job_context->superversion_contexts[i],
-                                         *cfds[i]->GetLatestMutableCFOptions());
+                                         all_mutable_cf_options[i]);
      VersionStorageInfo::LevelSummaryStorage tmp;
      ROCKS_LOG_BUFFER(log_buffer, "[%s] Level summary: %s\n",
                       cfds[i]->GetName().c_str(),
@@ -490,8 +504,7 @@ Status DBImpl::AtomicFlushMemTablesToOutputFiles(
      if (cfds[i]->IsDropped()) {
        continue;
      }
-      NotifyOnFlushCompleted(cfds[i], &file_meta[i],
-                             *cfds[i]->GetLatestMutableCFOptions(),
+      NotifyOnFlushCompleted(cfds[i], &file_meta[i], all_mutable_cf_options[i],
                             job_context->job_id, jobs[i].GetTableProperties());
      if (sfm) {
        std::string file_path = MakeTableFileName(
@@ -511,7 +524,7 @@ Status DBImpl::AtomicFlushMemTablesToOutputFiles(

  // Need to undo atomic flush if something went wrong, i.e. s is not OK and
  // it is not because of CF drop.
-  if (!s.ok() && !s.IsShutdownInProgress()) {
+  if (!s.ok() && !s.IsColumnFamilyDropped()) {
    // Have to cancel the flush jobs that have NOT executed because we need to
    // unref the versions.
    for (int i = 0; i != num_cfs; ++i) {
@@ -681,6 +694,10 @@ Status DBImpl::CompactRange(const CompactRangeOptions& options,
  }

  int max_level_with_files = 0;
+  // max_file_num_to_ignore can be used to filter out newly created SST files,
+  // useful for bottom level compaction in a manual compaction
+  uint64_t max_file_num_to_ignore = port::kMaxUint64;
+  uint64_t next_file_number = port::kMaxUint64;
  {
    InstrumentedMutexLock l(&mutex_);
    Version* base = cfd->current();
@@ -690,9 +707,11 @@ Status DBImpl::CompactRange(const CompactRangeOptions& options,
        max_level_with_files = level;
      }
    }
+    next_file_number = versions_->current_next_file_number();
  }

  int final_output_level = 0;
+
  if (cfd->ioptions()->compaction_style == kCompactionStyleUniversal &&
      cfd->NumberLevels() > 1) {
    // Always compact all files together.
@@ -702,8 +721,8 @@ Status DBImpl::CompactRange(const CompactRangeOptions& options,
      final_output_level--;
    }
    s = RunManualCompaction(cfd, ColumnFamilyData::kCompactAllLevels,
-                            final_output_level, options.target_path_id,
-                            options.max_subcompactions, begin, end, exclusive);
+                            final_output_level, options, begin, end, exclusive,
+                            false, max_file_num_to_ignore);
  } else {
    for (int level = 0; level <= max_level_with_files; level++) {
      int output_level;
@@ -728,6 +747,10 @@ Status DBImpl::CompactRange(const CompactRangeOptions& options,
          continue;
        }
        output_level = level;
+        // update max_file_num_to_ignore only for bottom level compaction
+        // because data in newly compacted files in middle levels may still need
+        // to be pushed down
+        max_file_num_to_ignore = next_file_number;
      } else {
        output_level = level + 1;
        if (cfd->ioptions()->compaction_style == kCompactionStyleLevel &&
@@ -736,8 +759,8 @@ Status DBImpl::CompactRange(const CompactRangeOptions& options,
          output_level = ColumnFamilyData::kCompactToBaseLevel;
        }
      }
-      s = RunManualCompaction(cfd, level, output_level, options.target_path_id,
-                              options.max_subcompactions, begin, end, exclusive);
+      s = RunManualCompaction(cfd, level, output_level, options, begin, end,
+                              exclusive, false, max_file_num_to_ignore);
      if (!s.ok()) {
        break;
      }
@@ -776,6 +799,31 @@ Status DBImpl::CompactRange(const CompactRangeOptions& options,
  return s;
 }

+namespace {
+class SnapshotListFetchCallbackImpl : public SnapshotListFetchCallback {
+ public:
+  SnapshotListFetchCallbackImpl(DBImpl* db_impl, Env* env,
+                                uint64_t snap_refresh_nanos, Logger* info_log)
+      : SnapshotListFetchCallback(env, snap_refresh_nanos),
+        db_impl_(db_impl),
+        info_log_(info_log) {}
+  virtual void Refresh(std::vector<SequenceNumber>* snapshots,
+                       SequenceNumber max) override {
+    size_t prev = snapshots->size();
+    snapshots->clear();
+    db_impl_->LoadSnapshots(snapshots, nullptr, max);
+    size_t now = snapshots->size();
+    ROCKS_LOG_DEBUG(info_log_,
+                    "Compaction snapshot count refreshed from %zu to %zu", prev,
+                    now);
+  }
+
+ private:
+  DBImpl* db_impl_;
+  Logger* info_log_;
+};
+}  // namespace
+
 Status DBImpl::CompactFiles(const CompactionOptions& compact_options,
                            ColumnFamilyHandle* column_family,
                            const std::vector<std::string>& input_file_names,
@@ -947,6 +995,9 @@ Status DBImpl::CompactFilesImpl(

  assert(is_snapshot_supported_ || snapshots_.empty());
  CompactionJobStats compaction_job_stats;
+  SnapshotListFetchCallbackImpl fetch_callback(
+      this, env_, c->mutable_cf_options()->snap_refresh_nanos,
+      immutable_db_options_.info_log.get());
  CompactionJob compaction_job(
      job_context->job_id, c.get(), immutable_db_options_,
      env_options_for_compaction_, versions_.get(), &shutting_down_,
@@ -956,7 +1007,9 @@ Status DBImpl::CompactFilesImpl(
      snapshot_checker, table_cache_, &event_logger_,
      c->mutable_cf_options()->paranoid_file_checks,
      c->mutable_cf_options()->report_bg_io_stats, dbname_,
-      &compaction_job_stats);
+      &compaction_job_stats, Env::Priority::USER,
+      immutable_db_options_.max_subcompactions <= 1 ? &fetch_callback
+                                                    : nullptr);

  // Creating a compaction influences the compaction score because the score
  // takes running compactions into account (by skipping files that are already
@@ -1000,7 +1053,7 @@ Status DBImpl::CompactFilesImpl(

  if (status.ok()) {
    // Done
-  } else if (status.IsShutdownInProgress()) {
+  } else if (status.IsColumnFamilyDropped() || status.IsShutdownInProgress()) {
    // Ignore compaction errors found during shutting down
  } else {
    ROCKS_LOG_WARN(immutable_db_options_.info_log,
@@ -1025,6 +1078,7 @@ Status DBImpl::CompactFilesImpl(
  if (bg_compaction_scheduled_ == 0) {
    bg_cv_.SignalAll();
  }
+  MaybeScheduleFlushOrCompaction();
  TEST_SYNC_POINT("CompactFilesImpl:End");

  return status;
@@ -1059,8 +1113,8 @@ Status DBImpl::ContinueBackgroundWork() {
  return Status::OK();
 }

-void DBImpl::NotifyOnCompactionBegin(ColumnFamilyData* cfd,
-                                     Compaction *c, const Status &st,
+void DBImpl::NotifyOnCompactionBegin(ColumnFamilyData* cfd, Compaction* c,
+                                     const Status& st,
                                     const CompactionJobStats& job_stats,
                                     int job_id) {
 #ifndef ROCKSDB_LITE
@@ -1305,8 +1359,9 @@ Status DBImpl::Flush(const FlushOptions& flush_options,
                  });
    s = AtomicFlushMemTables(cfds, flush_options, FlushReason::kManualFlush);
    ROCKS_LOG_INFO(immutable_db_options_.info_log,
-                   "Manual atomic flush finished, status: %s\n",
-                   "=====Column families:=====", s.ToString().c_str());
+                   "Manual atomic flush finished, status: %s\n"
+                   "=====Column families:=====",
+                   s.ToString().c_str());
    for (auto cfh : column_families) {
      auto cfhi = static_cast<ColumnFamilyHandleImpl*>(cfh);
      ROCKS_LOG_INFO(immutable_db_options_.info_log, "%s",
@@ -1318,11 +1373,11 @@ Status DBImpl::Flush(const FlushOptions& flush_options,
  return s;
 }

-Status DBImpl::RunManualCompaction(ColumnFamilyData* cfd, int input_level,
-                                   int output_level, uint32_t output_path_id,
-                                   uint32_t max_subcompactions,
-                                   const Slice* begin, const Slice* end,
-                                   bool exclusive, bool disallow_trivial_move) {
+Status DBImpl::RunManualCompaction(
+    ColumnFamilyData* cfd, int input_level, int output_level,
+    const CompactRangeOptions& compact_range_options, const Slice* begin,
+    const Slice* end, bool exclusive, bool disallow_trivial_move,
+    uint64_t max_file_num_to_ignore) {
  assert(input_level == ColumnFamilyData::kCompactAllLevels ||
         input_level >= 0);

@@ -1335,7 +1390,7 @@ Status DBImpl::RunManualCompaction(ColumnFamilyData* cfd, int input_level,
  manual.cfd = cfd;
  manual.input_level = input_level;
  manual.output_level = output_level;
-  manual.output_path_id = output_path_id;
+  manual.output_path_id = compact_range_options.target_path_id;
  manual.done = false;
  manual.in_progress = false;
  manual.incomplete = false;
@@ -1410,9 +1465,9 @@ Status DBImpl::RunManualCompaction(ColumnFamilyData* cfd, int input_level,
        (((manual.manual_end = &manual.tmp_storage1) != nullptr) &&
         ((compaction = manual.cfd->CompactRange(
               *manual.cfd->GetLatestMutableCFOptions(), manual.input_level,
-               manual.output_level, manual.output_path_id, max_subcompactions,
-               manual.begin, manual.end, &manual.manual_end,
-               &manual_conflict)) == nullptr &&
+               manual.output_level, compact_range_options, manual.begin,
+               manual.end, &manual.manual_end, &manual_conflict,
+               max_file_num_to_ignore)) == nullptr &&
          manual_conflict))) {
      // exclusive manual compactions should not see a conflict during
      // CompactRange
@@ -1443,7 +1498,7 @@ Status DBImpl::RunManualCompaction(ColumnFamilyData* cfd, int input_level,
      manual.incomplete = false;
      bg_compaction_scheduled_++;
      env_->Schedule(&DBImpl::BGWorkCompaction, ca, Env::Priority::LOW, this,
-                     &DBImpl::UnscheduleCallback);
+                     &DBImpl::UnscheduleCompactionCallback);
      scheduled = true;
    }
  }
@@ -1496,13 +1551,39 @@ Status DBImpl::FlushMemTable(ColumnFamilyData* cfd,
    if (!cfd->mem()->IsEmpty() || !cached_recoverable_state_empty_.load()) {
      s = SwitchMemtable(cfd, &context);
    }
-
    if (s.ok()) {
      if (cfd->imm()->NumNotFlushed() != 0 || !cfd->mem()->IsEmpty() ||
          !cached_recoverable_state_empty_.load()) {
        flush_memtable_id = cfd->imm()->GetLatestMemTableID();
        flush_req.emplace_back(cfd, flush_memtable_id);
      }
+      if (immutable_db_options_.persist_stats_to_disk) {
+        ColumnFamilyData* cfd_stats =
+            versions_->GetColumnFamilySet()->GetColumnFamily(
+                kPersistentStatsColumnFamilyName);
+        if (cfd_stats != nullptr && cfd_stats != cfd &&
+            !cfd_stats->mem()->IsEmpty()) {
+          // only force flush stats CF when it will be the only CF lagging
+          // behind after the current flush
+          bool stats_cf_flush_needed = true;
+          for (auto* loop_cfd : *versions_->GetColumnFamilySet()) {
+            if (loop_cfd == cfd_stats || loop_cfd == cfd) {
+              continue;
+            }
+            if (loop_cfd->GetLogNumber() <= cfd_stats->GetLogNumber()) {
+              stats_cf_flush_needed = false;
+            }
+          }
+          if (stats_cf_flush_needed) {
+            ROCKS_LOG_INFO(immutable_db_options_.info_log,
+                           "Force flushing stats CF with manual flush of %s "
+                           "to avoid holding old logs", cfd->GetName().c_str());
+            s = SwitchMemtable(cfd_stats, &context);
+            flush_memtable_id = cfd_stats->imm()->GetLatestMemTableID();
+            flush_req.emplace_back(cfd_stats, flush_memtable_id);
+          }
+        }
+      }
    }

    if (s.ok() && !flush_req.empty()) {
@@ -1510,6 +1591,16 @@ Status DBImpl::FlushMemTable(ColumnFamilyData* cfd,
        ColumnFamilyData* loop_cfd = elem.first;
        loop_cfd->imm()->FlushRequested();
      }
+      // If the caller wants to wait for this flush to complete, it indicates
+      // that the caller expects the ColumnFamilyData not to be free'ed by
+      // other threads which may drop the column family concurrently.
+      // Therefore, we increase the cfd's ref count.
+      if (flush_options.wait) {
+        for (auto& elem : flush_req) {
+          ColumnFamilyData* loop_cfd = elem.first;
+          loop_cfd->Ref();
+        }
+      }
      SchedulePendingFlush(flush_req, flush_reason);
      MaybeScheduleFlushOrCompaction();
    }
@@ -1518,7 +1609,8 @@ Status DBImpl::FlushMemTable(ColumnFamilyData* cfd,
      write_thread_.ExitUnbatched(&w);
    }
  }
-
+  TEST_SYNC_POINT("DBImpl::FlushMemTable:AfterScheduleFlush");
+  TEST_SYNC_POINT("DBImpl::FlushMemTable:BeforeWaitForBgFlush");
  if (s.ok() && flush_options.wait) {
    autovector<ColumnFamilyData*> cfds;
    autovector<const uint64_t*> flush_memtable_ids;
@@ -1528,6 +1620,13 @@ Status DBImpl::FlushMemTable(ColumnFamilyData* cfd,
    }
    s = WaitForFlushMemTables(cfds, flush_memtable_ids,
                              (flush_reason == FlushReason::kErrorRecovery));
+    for (auto* tmp_cfd : cfds) {
+      if (tmp_cfd->Unref()) {
+        // Only one thread can reach here.
+        InstrumentedMutexLock lock_guard(&mutex_);
+        delete tmp_cfd;
+      }
+    }
  }
  TEST_SYNC_POINT("FlushMemTableFinished");
  return s;
@@ -1591,6 +1690,15 @@ Status DBImpl::AtomicFlushMemTables(
      for (auto cfd : cfds) {
        cfd->imm()->FlushRequested();
      }
+      // If the caller wants to wait for this flush to complete, it indicates
+      // that the caller expects the ColumnFamilyData not to be free'ed by
+      // other threads which may drop the column family concurrently.
+      // Therefore, we increase the cfd's ref count.
+      if (flush_options.wait) {
+        for (auto cfd : cfds) {
+          cfd->Ref();
+        }
+      }
      GenerateFlushRequest(cfds, &flush_req);
      SchedulePendingFlush(flush_req, flush_reason);
      MaybeScheduleFlushOrCompaction();
@@ -1601,7 +1709,7 @@ Status DBImpl::AtomicFlushMemTables(
    }
  }
  TEST_SYNC_POINT("DBImpl::AtomicFlushMemTables:AfterScheduleFlush");
-
+  TEST_SYNC_POINT("DBImpl::AtomicFlushMemTables:BeforeWaitForBgFlush");
  if (s.ok() && flush_options.wait) {
    autovector<const uint64_t*> flush_memtable_ids;
    for (auto& iter : flush_req) {
@@ -1609,6 +1717,13 @@ Status DBImpl::AtomicFlushMemTables(
    }
    s = WaitForFlushMemTables(cfds, flush_memtable_ids,
                              (flush_reason == FlushReason::kErrorRecovery));
+    for (auto* cfd : cfds) {
+      if (cfd->Unref()) {
+        // Only one thread can reach here.
+        InstrumentedMutexLock lock_guard(&mutex_);
+        delete cfd;
+      }
+    }
  }
  return s;
 }
@@ -1620,7 +1735,7 @@ Status DBImpl::AtomicFlushMemTables(
 // it against various constrains and delays flush if it'd cause write stall.
 // Called should check status and flush_needed to see if flush already happened.
 Status DBImpl::WaitUntilFlushWouldNotStallWrites(ColumnFamilyData* cfd,
-    bool* flush_needed) {
+                                                 bool* flush_needed) {
  {
    *flush_needed = true;
    InstrumentedMutexLock l(&mutex_);
@@ -1643,7 +1758,10 @@ Status DBImpl::WaitUntilFlushWouldNotStallWrites(ColumnFamilyData* cfd,
                       cfd->GetName().c_str());
        bg_cv_.Wait();
      }
-      if (cfd->IsDropped() || shutting_down_.load(std::memory_order_acquire)) {
+      if (cfd->IsDropped()) {
+        return Status::ColumnFamilyDropped();
+      }
+      if (shutting_down_.load(std::memory_order_acquire)) {
        return Status::ShutdownInProgress();
      }

@@ -1768,7 +1886,7 @@ void DBImpl::MaybeScheduleFlushOrCompaction() {
    // we paused the background work
    return;
  } else if (error_handler_.IsBGWorkStopped() &&
-      !error_handler_.IsRecoveryInProgress()) {
+             !error_handler_.IsRecoveryInProgress()) {
    // There has been a hard error and this call is not part of the recovery
    // sequence. Bail out here so we don't get into an endless loop of
    // scheduling BG work which will again call this function
@@ -1783,7 +1901,11 @@ void DBImpl::MaybeScheduleFlushOrCompaction() {
  while (!is_flush_pool_empty && unscheduled_flushes_ > 0 &&
         bg_flush_scheduled_ < bg_job_limits.max_flushes) {
    bg_flush_scheduled_++;
-    env_->Schedule(&DBImpl::BGWorkFlush, this, Env::Priority::HIGH, this);
+    FlushThreadArg* fta = new FlushThreadArg;
+    fta->db_ = this;
+    fta->thread_pri_ = Env::Priority::HIGH;
+    env_->Schedule(&DBImpl::BGWorkFlush, fta, Env::Priority::HIGH, this,
+                   &DBImpl::UnscheduleFlushCallback);
  }

  // special case -- if high-pri (flush) thread pool is empty, then schedule
@@ -1793,7 +1915,11 @@ void DBImpl::MaybeScheduleFlushOrCompaction() {
           bg_flush_scheduled_ + bg_compaction_scheduled_ <
               bg_job_limits.max_flushes) {
      bg_flush_scheduled_++;
-      env_->Schedule(&DBImpl::BGWorkFlush, this, Env::Priority::LOW, this);
+      FlushThreadArg* fta = new FlushThreadArg;
+      fta->db_ = this;
+      fta->thread_pri_ = Env::Priority::LOW;
+      env_->Schedule(&DBImpl::BGWorkFlush, fta, Env::Priority::LOW, this,
+                     &DBImpl::UnscheduleFlushCallback);
    }
  }

@@ -1823,7 +1949,7 @@ void DBImpl::MaybeScheduleFlushOrCompaction() {
    bg_compaction_scheduled_++;
    unscheduled_compactions_--;
    env_->Schedule(&DBImpl::BGWorkCompaction, ca, Env::Priority::LOW, this,
-                   &DBImpl::UnscheduleCallback);
+                   &DBImpl::UnscheduleCompactionCallback);
  }
 }

@@ -1938,10 +2064,13 @@ void DBImpl::SchedulePendingPurge(std::string fname, std::string dir_to_sync,
  purge_queue_.push_back(std::move(file_info));
 }

-void DBImpl::BGWorkFlush(void* db) {
-  IOSTATS_SET_THREAD_POOL_ID(Env::Priority::HIGH);
+void DBImpl::BGWorkFlush(void* arg) {
+  FlushThreadArg fta = *(reinterpret_cast<FlushThreadArg*>(arg));
+  delete reinterpret_cast<FlushThreadArg*>(arg);
+
+  IOSTATS_SET_THREAD_POOL_ID(fta.thread_pri_);
  TEST_SYNC_POINT("DBImpl::BGWorkFlush");
-  reinterpret_cast<DBImpl*>(db)->BackgroundCallFlush();
+  reinterpret_cast<DBImpl*>(fta.db_)->BackgroundCallFlush(fta.thread_pri_);
  TEST_SYNC_POINT("DBImpl::BGWorkFlush:done");
 }

@@ -1976,7 +2105,7 @@ void DBImpl::BGWorkPurge(void* db) {
  TEST_SYNC_POINT("DBImpl::BGWorkPurge:end");
 }

-void DBImpl::UnscheduleCallback(void* arg) {
+void DBImpl::UnscheduleCompactionCallback(void* arg) {
  CompactionArg ca = *(reinterpret_cast<CompactionArg*>(arg));
  delete reinterpret_cast<CompactionArg*>(arg);
  if (ca.prepicked_compaction != nullptr) {
@@ -1985,11 +2114,17 @@ void DBImpl::UnscheduleCallback(void* arg) {
    }
    delete ca.prepicked_compaction;
  }
-  TEST_SYNC_POINT("DBImpl::UnscheduleCallback");
+  TEST_SYNC_POINT("DBImpl::UnscheduleCompactionCallback");
+}
+
+void DBImpl::UnscheduleFlushCallback(void* arg) {
+  delete reinterpret_cast<FlushThreadArg*>(arg);
+  TEST_SYNC_POINT("DBImpl::UnscheduleFlushCallback");
 }

 Status DBImpl::BackgroundFlush(bool* made_progress, JobContext* job_context,
-                               LogBuffer* log_buffer, FlushReason* reason) {
+                               LogBuffer* log_buffer, FlushReason* reason,
+                               Env::Priority thread_pri) {
  mutex_.AssertHeld();

  Status status;
@@ -2011,6 +2146,7 @@ Status DBImpl::BackgroundFlush(bool* made_progress, JobContext* job_context,
  autovector<BGFlushArg> bg_flush_args;
  std::vector<SuperVersionContext>& superversion_contexts =
      job_context->superversion_contexts;
+  autovector<ColumnFamilyData*> column_families_not_to_flush;
  while (!flush_queue_.empty()) {
    // This cfd is already referenced
    const FlushRequest& flush_req = PopFirstFromFlushQueue();
@@ -2021,9 +2157,7 @@ Status DBImpl::BackgroundFlush(bool* made_progress, JobContext* job_context,
      ColumnFamilyData* cfd = iter.first;
      if (cfd->IsDropped() || !cfd->imm()->IsFlushPending()) {
        // can't flush this CF, try next one
-        if (cfd->Unref()) {
-          delete cfd;
-        }
+        column_families_not_to_flush.push_back(cfd);
        continue;
      }
      superversion_contexts.emplace_back(SuperVersionContext(true));
@@ -2050,7 +2184,8 @@ Status DBImpl::BackgroundFlush(bool* made_progress, JobContext* job_context,
          bg_compaction_scheduled_);
    }
    status = FlushMemTablesToOutputFiles(bg_flush_args, made_progress,
-                                         job_context, log_buffer);
+                                         job_context, log_buffer, thread_pri);
+    TEST_SYNC_POINT("DBImpl::BackgroundFlush:BeforeFlush");
    // All the CFDs in the FlushReq must have the same flush reason, so just
    // grab the first one
    *reason = bg_flush_args[0].cfd_->GetFlushReason();
@@ -2062,10 +2197,15 @@ Status DBImpl::BackgroundFlush(bool* made_progress, JobContext* job_context,
      }
    }
  }
+  for (auto cfd : column_families_not_to_flush) {
+    if (cfd->Unref()) {
+      delete cfd;
+    }
+  }
  return status;
 }

-void DBImpl::BackgroundCallFlush() {
+void DBImpl::BackgroundCallFlush(Env::Priority thread_pri) {
  bool made_progress = false;
  JobContext job_context(next_job_id_.fetch_add(1), true);

@@ -2082,9 +2222,9 @@ void DBImpl::BackgroundCallFlush() {
        CaptureCurrentFileNumberInPendingOutputs();
    FlushReason reason;

-    Status s =
-        BackgroundFlush(&made_progress, &job_context, &log_buffer, &reason);
-    if (!s.ok() && !s.IsShutdownInProgress() &&
+    Status s = BackgroundFlush(&made_progress, &job_context, &log_buffer,
+                               &reason, thread_pri);
+    if (!s.ok() && !s.IsShutdownInProgress() && !s.IsColumnFamilyDropped() &&
        reason != FlushReason::kErrorRecovery) {
      // Wait a little bit before retrying background flush in
      // case this is an environmental problem and we do not want to
@@ -2109,7 +2249,8 @@ void DBImpl::BackgroundCallFlush() {

    // If flush failed, we want to delete all temporary files that we might have
    // created. Thus, we force full scan in FindObsoleteFiles()
-    FindObsoleteFiles(&job_context, !s.ok() && !s.IsShutdownInProgress());
+    FindObsoleteFiles(&job_context, !s.ok() && !s.IsShutdownInProgress() &&
+                                        !s.IsColumnFamilyDropped());
    // delete unnecessary files if any, this is done outside the mutex
    if (job_context.HaveSomethingToClean() ||
        job_context.HaveSomethingToDelete() || !log_buffer.IsEmpty()) {
@@ -2166,14 +2307,15 @@ void DBImpl::BackgroundCallCompaction(PrepickedCompaction* prepicked_compaction,
            bg_bottom_compaction_scheduled_) ||
           (bg_thread_pri == Env::Priority::LOW && bg_compaction_scheduled_));
    Status s = BackgroundCompaction(&made_progress, &job_context, &log_buffer,
-                                    prepicked_compaction);
+                                    prepicked_compaction, bg_thread_pri);
    TEST_SYNC_POINT("BackgroundCallCompaction:1");
    if (s.IsBusy()) {
      bg_cv_.SignalAll();  // In case a waiter can proceed despite the error
      mutex_.Unlock();
-      env_->SleepForMicroseconds(10000); // prevent hot loop
+      env_->SleepForMicroseconds(10000);  // prevent hot loop
      mutex_.Lock();
-    } else if (!s.ok() && !s.IsShutdownInProgress()) {
+    } else if (!s.ok() && !s.IsShutdownInProgress() &&
+               !s.IsColumnFamilyDropped()) {
      // Wait a little bit before retrying background compaction in
      // case this is an environmental problem and we do not want to
      // chew up resources for failed compactions for the duration of
@@ -2197,7 +2339,8 @@ void DBImpl::BackgroundCallCompaction(PrepickedCompaction* prepicked_compaction,
    // If compaction failed, we want to delete all temporary files that we might
    // have created (they might not be all recorded in job_context in case of a
    // failure). Thus, we force full scan in FindObsoleteFiles()
-    FindObsoleteFiles(&job_context, !s.ok() && !s.IsShutdownInProgress());
+    FindObsoleteFiles(&job_context, !s.ok() && !s.IsShutdownInProgress() &&
+                                        !s.IsColumnFamilyDropped());
    TEST_SYNC_POINT("DBImpl::BackgroundCallCompaction:FoundObsoleteFiles");

    // delete unnecessary files if any, this is done outside the mutex
@@ -2253,7 +2396,8 @@ void DBImpl::BackgroundCallCompaction(PrepickedCompaction* prepicked_compaction,
 Status DBImpl::BackgroundCompaction(bool* made_progress,
                                    JobContext* job_context,
                                    LogBuffer* log_buffer,
-                                    PrepickedCompaction* prepicked_compaction) {
+                                    PrepickedCompaction* prepicked_compaction,
+                                    Env::Priority thread_pri) {
  ManualCompactionState* manual_compaction =
      prepicked_compaction == nullptr
          ? nullptr
@@ -2416,8 +2560,8 @@ Status DBImpl::BackgroundCompaction(bool* made_progress,
          status = Status::CompactionTooLarge();
        } else {
          // update statistics
-          MeasureTime(stats_, NUM_FILES_IN_SINGLE_COMPACTION,
-                      c->inputs(0)->size());
+          RecordInHistogram(stats_, NUM_FILES_IN_SINGLE_COMPACTION,
+                            c->inputs(0)->size());
          // There are three things that can change compaction score:
          // 1) When flush or compaction finish. This case is covered by
          // InstallSuperVersionAndScheduleWork
@@ -2566,7 +2710,7 @@ Status DBImpl::BackgroundCompaction(bool* made_progress,
    ca->prepicked_compaction->task_token = std::move(task_token);
    ++bg_bottom_compaction_scheduled_;
    env_->Schedule(&DBImpl::BGWorkBottomCompaction, ca, Env::Priority::BOTTOM,
-                   this, &DBImpl::UnscheduleCallback);
+                   this, &DBImpl::UnscheduleCompactionCallback);
  } else {
    TEST_SYNC_POINT_CALLBACK("DBImpl::BackgroundCompaction:BeforeCompaction",
                             c->column_family_data());
@@ -2580,22 +2724,29 @@ Status DBImpl::BackgroundCompaction(bool* made_progress,
    GetSnapshotContext(job_context, &snapshot_seqs,
                       &earliest_write_conflict_snapshot, &snapshot_checker);
    assert(is_snapshot_supported_ || snapshots_.empty());
+    SnapshotListFetchCallbackImpl fetch_callback(
+        this, env_, c->mutable_cf_options()->snap_refresh_nanos,
+        immutable_db_options_.info_log.get());
    CompactionJob compaction_job(
        job_context->job_id, c.get(), immutable_db_options_,
        env_options_for_compaction_, versions_.get(), &shutting_down_,
        preserve_deletes_seqnum_.load(), log_buffer, directories_.GetDbDir(),
        GetDataDir(c->column_family_data(), c->output_path_id()), stats_,
-        &mutex_, &error_handler_, snapshot_seqs, earliest_write_conflict_snapshot,
-        snapshot_checker, table_cache_, &event_logger_,
-        c->mutable_cf_options()->paranoid_file_checks,
+        &mutex_, &error_handler_, snapshot_seqs,
+        earliest_write_conflict_snapshot, snapshot_checker, table_cache_,
+        &event_logger_, c->mutable_cf_options()->paranoid_file_checks,
        c->mutable_cf_options()->report_bg_io_stats, dbname_,
-        &compaction_job_stats);
+        &compaction_job_stats, thread_pri,
+        immutable_db_options_.max_subcompactions <= 1 ? &fetch_callback
+                                                      : nullptr);
    compaction_job.Prepare();

    NotifyOnCompactionBegin(c->column_family_data(), c.get(), status,
                            compaction_job_stats, job_context->job_id);

    mutex_.Unlock();
+    TEST_SYNC_POINT_CALLBACK(
+        "DBImpl::BackgroundCompaction:NonTrivial:BeforeRun", nullptr);
    compaction_job.Run();
    TEST_SYNC_POINT("DBImpl::BackgroundCompaction:NonTrivial:AfterRun");
    mutex_.Lock();
@@ -2629,7 +2780,7 @@ Status DBImpl::BackgroundCompaction(bool* made_progress,

  if (status.ok() || status.IsCompactionTooLarge()) {
    // Done
-  } else if (status.IsShutdownInProgress()) {
+  } else if (status.IsColumnFamilyDropped() || status.IsShutdownInProgress()) {
    // Ignore compaction errors found during shutting down
  } else {
    ROCKS_LOG_WARN(immutable_db_options_.info_log, "Compaction error: %s",
@@ -2713,7 +2864,7 @@ void DBImpl::RemoveManualCompaction(DBImpl::ManualCompactionState* m) {
      it = manual_compaction_dequeue_.erase(it);
      return;
    }
-    it++;
+    ++it;
  }
  assert(false);
  return;
@@ -2734,7 +2885,7 @@ bool DBImpl::ShouldntRunManualCompaction(ManualCompactionState* m) {
  bool seen = false;
  while (it != manual_compaction_dequeue_.end()) {
    if (m == (*it)) {
-      it++;
+      ++it;
      seen = true;
      continue;
    } else if (MCOverlap(m, (*it)) && (!seen && !(*it)->in_progress)) {
@@ -2743,7 +2894,7 @@ bool DBImpl::ShouldntRunManualCompaction(ManualCompactionState* m) {
      // and (*it) is ahead in the queue and is not yet in progress
      return true;
    }
-    it++;
+    ++it;
  }
  return false;
 }
@@ -2761,7 +2912,7 @@ bool DBImpl::HaveManualCompaction(ColumnFamilyData* cfd) {
      // in progress
      return true;
    }
-    it++;
+    ++it;
  }
  return false;
 }
@@ -2774,7 +2925,7 @@ bool DBImpl::HasExclusiveManualCompaction() {
    if ((*it)->exclusive) {
      return true;
    }
-    it++;
+    ++it;
  }
  return false;
 }
@@ -2812,7 +2963,7 @@ void DBImpl::BuildCompactionJobInfo(
                              fmd->fd.GetNumber(), fmd->fd.GetPathId());
      compaction_job_info->input_files.push_back(fn);
      if (compaction_job_info->table_properties.count(fn) == 0) {
-        shared_ptr<const TableProperties> tp;
+        std::shared_ptr<const TableProperties> tp;
        auto s = current->GetTableProperties(&tp, fmd, &fn);
        if (s.ok()) {
          compaction_job_info->table_properties[fn] = tp;
@@ -2859,6 +3010,17 @@ void DBImpl::InstallSuperVersionAndScheduleWork(
  }
  cfd->InstallSuperVersion(sv_context, &mutex_, mutable_cf_options);

+  // There may be a small data race here. The snapshot tricking bottommost
+  // compaction may already be released here. But assuming there will always be
+  // newer snapshot created and released frequently, the compaction will be
+  // triggered soon anyway.
+  bottommost_files_mark_threshold_ = kMaxSequenceNumber;
+  for (auto* my_cfd : *versions_->GetColumnFamilySet()) {
+    bottommost_files_mark_threshold_ = std::min(
+        bottommost_files_mark_threshold_,
+        my_cfd->current()->storage_info()->bottommost_files_mark_threshold());
+  }
+
  // Whenever we install new SuperVersion, we might need to issue new flushes or
  // compactions.
  SchedulePendingCompaction(cfd);
@@ -9,7 +9,7 @@

 #ifndef NDEBUG

-#include "db/db_impl.h"
+#include "db/db_impl/db_impl.h"
 #include "db/error_handler.h"
 #include "monitoring/thread_status_updater.h"

@@ -26,10 +26,16 @@ void DBImpl::TEST_SwitchWAL() {
  SwitchWAL(&write_context);
 }

-bool DBImpl::TEST_WALBufferIsEmpty() {
-  InstrumentedMutexLock wl(&log_write_mutex_);
+bool DBImpl::TEST_WALBufferIsEmpty(bool lock) {
+  if (lock) {
+    log_write_mutex_.Lock();
+  }
  log::Writer* cur_log_writer = logs_.back().writer;
-  return cur_log_writer->TEST_BufferIsEmpty();
+  auto res = cur_log_writer->TEST_BufferIsEmpty();
+  if (lock) {
+    log_write_mutex_.Unlock();
+  }
+  return res;
 }

 int64_t DBImpl::TEST_MaxNextLevelOverlappingBytes(
@@ -87,8 +93,9 @@ Status DBImpl::TEST_CompactRange(int level, const Slice* begin,
       cfd->ioptions()->compaction_style == kCompactionStyleFIFO)
          ? level
          : level + 1;
-  return RunManualCompaction(cfd, level, output_level, 0, 0, begin, end, true,
-                             disallow_trivial_move);
+  return RunManualCompaction(cfd, level, output_level, CompactRangeOptions(),
+                             begin, end, true, disallow_trivial_move,
+                             port::kMaxUint64 /*max_file_num_to_ignore*/);
 }

 Status DBImpl::TEST_SwitchMemtable(ColumnFamilyData* cfd) {
@@ -101,7 +108,7 @@ Status DBImpl::TEST_SwitchMemtable(ColumnFamilyData* cfd) {
 }

 Status DBImpl::TEST_FlushMemTable(bool wait, bool allow_write_stall,
-    ColumnFamilyHandle* cfh) {
+                                  ColumnFamilyHandle* cfh) {
  FlushOptions fo;
  fo.wait = wait;
  fo.allow_write_stall = allow_write_stall;
@@ -115,6 +122,16 @@ Status DBImpl::TEST_FlushMemTable(bool wait, bool allow_write_stall,
  return FlushMemTable(cfd, fo, FlushReason::kTest);
 }

+Status DBImpl::TEST_FlushMemTable(ColumnFamilyData* cfd,
+                                  const FlushOptions& flush_opts) {
+  return FlushMemTable(cfd, flush_opts, FlushReason::kTest);
+}
+
+Status DBImpl::TEST_AtomicFlushMemTables(
+    const autovector<ColumnFamilyData*>& cfds, const FlushOptions& flush_opts) {
+  return AtomicFlushMemTables(cfds, flush_opts, FlushReason::kTest);
+}
+
 Status DBImpl::TEST_WaitForFlushMemTable(ColumnFamilyHandle* column_family) {
  ColumnFamilyData* cfd;
  if (column_family == nullptr) {
@@ -143,13 +160,9 @@ Status DBImpl::TEST_WaitForCompact(bool wait_unscheduled) {
  return error_handler_.GetBGError();
 }

-void DBImpl::TEST_LockMutex() {
-  mutex_.Lock();
-}
+void DBImpl::TEST_LockMutex() { mutex_.Lock(); }

-void DBImpl::TEST_UnlockMutex() {
-  mutex_.Unlock();
-}
+void DBImpl::TEST_UnlockMutex() { mutex_.Unlock(); }

 void* DBImpl::TEST_BeginWrite() {
  auto w = new WriteThread::Writer();
@@ -243,10 +256,24 @@ size_t DBImpl::TEST_GetWalPreallocateBlockSize(
  return GetWalPreallocateBlockSize(write_buffer_size);
 }

-void DBImpl::TEST_WaitForTimedTaskRun(std::function<void()> callback) const {
+void DBImpl::TEST_WaitForDumpStatsRun(std::function<void()> callback) const {
  if (thread_dump_stats_ != nullptr) {
    thread_dump_stats_->TEST_WaitForRun(callback);
  }
 }
+
+void DBImpl::TEST_WaitForPersistStatsRun(std::function<void()> callback) const {
+  if (thread_persist_stats_ != nullptr) {
+    thread_persist_stats_->TEST_WaitForRun(callback);
+  }
+}
+
+bool DBImpl::TEST_IsPersistentStatsEnabled() const {
+  return thread_persist_stats_ && thread_persist_stats_->IsRunning();
+}
+
+size_t DBImpl::TEST_EstimateInMemoryStatsHistorySize() const {
+  return EstimateInMemoryStatsHistorySize();
+}
 }  // namespace rocksdb
 #endif  // NDEBUG
@@ -7,13 +7,9 @@
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file. See the AUTHORS file for names of contributors.

-#include "db/db_impl.h"
+#include "db/db_impl/db_impl.h"

-#ifndef __STDC_FORMAT_MACROS
-#define __STDC_FORMAT_MACROS
-#endif
-
-#include <inttypes.h>
+#include <cinttypes>
 #include <vector>

 #include "db/column_family.h"
@@ -6,18 +6,15 @@
 // Copyright (c) 2011 The LevelDB Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file. See the AUTHORS file for names of contributors.
-#include "db/db_impl.h"
+#include "db/db_impl/db_impl.h"

-#ifndef __STDC_FORMAT_MACROS
-#define __STDC_FORMAT_MACROS
-#endif
-#include <inttypes.h>
+#include <cinttypes>
 #include <set>
 #include <unordered_set>
 #include "db/event_helpers.h"
 #include "db/memtable_list.h"
-#include "util/file_util.h"
-#include "util/sst_file_manager_impl.h"
+#include "file/file_util.h"
+#include "file/sst_file_manager_impl.h"

 namespace rocksdb {

@@ -108,7 +105,7 @@ void DBImpl::FindObsoleteFiles(JobContext* job_context, bool force,
  versions_->AddLiveFiles(&job_context->sst_live);
  if (doing_the_full_scan) {
    InfoLogPrefix info_log_prefix(!immutable_db_options_.db_log_dir.empty(),
-        dbname_);
+                                  dbname_);
    std::set<std::string> paths;
    for (size_t path_id = 0; path_id < immutable_db_options_.db_paths.size();
         path_id++) {
@@ -152,8 +149,7 @@ void DBImpl::FindObsoleteFiles(JobContext* job_context, bool force,
        }

        // TODO(icanadi) clean up this mess to avoid having one-off "/" prefixes
-        job_context->full_scan_candidate_files.emplace_back(
-            "/" + file, path);
+        job_context->full_scan_candidate_files.emplace_back("/" + file, path);
      }
    }

@@ -163,8 +159,8 @@ void DBImpl::FindObsoleteFiles(JobContext* job_context, bool force,
      env_->GetChildren(immutable_db_options_.wal_dir,
                        &log_files);  // Ignore errors
      for (const std::string& log_file : log_files) {
-        job_context->full_scan_candidate_files.emplace_back(log_file,
-            immutable_db_options_.wal_dir);
+        job_context->full_scan_candidate_files.emplace_back(
+            log_file, immutable_db_options_.wal_dir);
      }
    }
    // Add info log files in db_log_dir
@@ -174,8 +170,8 @@ void DBImpl::FindObsoleteFiles(JobContext* job_context, bool force,
      // Ignore errors
      env_->GetChildren(immutable_db_options_.db_log_dir, &info_log_files);
      for (std::string& log_file : info_log_files) {
-        job_context->full_scan_candidate_files.emplace_back(log_file,
-            immutable_db_options_.db_log_dir);
+        job_context->full_scan_candidate_files.emplace_back(
+            log_file, immutable_db_options_.db_log_dir);
      }
    }
  }
@@ -260,14 +256,15 @@ void DBImpl::DeleteObsoleteFileImpl(int job_id, const std::string& fname,
                                    const std::string& path_to_sync,
                                    FileType type, uint64_t number) {
  Status file_deletion_status;
-  if (type == kTableFile) {
+  if (type == kTableFile || type == kLogFile) {
    file_deletion_status =
-        DeleteSSTFile(&immutable_db_options_, fname, path_to_sync);
+        DeleteDBFile(&immutable_db_options_, fname, path_to_sync,
+                     /*force_bg=*/false, /*force_fg=*/!wal_in_db_path_);
  } else {
    file_deletion_status = env_->DeleteFile(fname);
  }
  TEST_SYNC_POINT_CALLBACK("DBImpl::DeleteObsoleteFileImpl:AfterDeletion",
-      &file_deletion_status);
+                           &file_deletion_status);
  if (file_deletion_status.ok()) {
    ROCKS_LOG_DEBUG(immutable_db_options_.info_log,
                    "[JOB %d] Delete %s type=%d #%" PRIu64 " -- %s\n", job_id,
@@ -322,7 +319,8 @@ void DBImpl::PurgeObsoleteFiles(JobContext& state, bool schedule_only) {
  const char* kDumbDbName = "";
  for (auto& file : state.sst_delete_files) {
    candidate_files.emplace_back(
-        MakeTableFileName(kDumbDbName, file.metadata->fd.GetNumber()), file.path);
+        MakeTableFileName(kDumbDbName, file.metadata->fd.GetNumber()),
+        file.path);
    if (file.metadata->table_reader_handle) {
      table_cache_->Release(file.metadata->table_reader_handle);
    }
@@ -332,7 +330,7 @@ void DBImpl::PurgeObsoleteFiles(JobContext& state, bool schedule_only) {
  for (auto file_num : state.log_delete_files) {
    if (file_num > 0) {
      candidate_files.emplace_back(LogFileName(kDumbDbName, file_num),
-          immutable_db_options_.wal_dir);
+                                   immutable_db_options_.wal_dir);
    }
  }
  for (const auto& filename : state.manifest_delete_files) {
@@ -465,13 +463,12 @@ void DBImpl::PurgeObsoleteFiles(JobContext& state, bool schedule_only) {
    } else {
      dir_to_sync =
          (type == kLogFile) ? immutable_db_options_.wal_dir : dbname_;
-      fname = dir_to_sync
-            + (
-                (!dir_to_sync.empty() && dir_to_sync.back() == '/') ||
-                (!to_delete.empty() && to_delete.front() == '/')
-                ? "" : "/"
-              )
-            + to_delete;
+      fname = dir_to_sync +
+              ((!dir_to_sync.empty() && dir_to_sync.back() == '/') ||
+                       (!to_delete.empty() && to_delete.front() == '/')
+                   ? ""
+                   : "/") +
+              to_delete;
    }

 #ifndef ROCKSDB_LITE
@@ -482,6 +479,11 @@ void DBImpl::PurgeObsoleteFiles(JobContext& state, bool schedule_only) {
    }
 #endif  // !ROCKSDB_LITE

+    for (const auto w : state.logs_to_free) {
+      // TODO: maybe check the return value of Close.
+      w->Close();
+    }
+
    Status file_deletion_status;
    if (schedule_only) {
      InstrumentedMutexLock guard_lock(&mutex_);
@@ -6,21 +6,19 @@
 // Copyright (c) 2011 The LevelDB Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file. See the AUTHORS file for names of contributors.
-#include "db/db_impl.h"
+#include "db/db_impl/db_impl.h"

-#ifndef __STDC_FORMAT_MACROS
-#define __STDC_FORMAT_MACROS
-#endif
-#include <inttypes.h>
+#include <cinttypes>

 #include "db/builder.h"
 #include "db/error_handler.h"
+#include "file/sst_file_manager_impl.h"
+#include "monitoring/persistent_stats_history.h"
 #include "options/options_helper.h"
 #include "rocksdb/wal_filter.h"
-#include "table/block_based_table_factory.h"
+#include "table/block_based/block_based_table_factory.h"
+#include "test_util/sync_point.h"
 #include "util/rate_limiter.h"
-#include "util/sst_file_manager_impl.h"
-#include "util/sync_point.h"

 namespace rocksdb {
 Options SanitizeOptions(const std::string& dbname, const Options& src) {
@@ -124,6 +122,25 @@ DBOptions SanitizeOptions(const std::string& dbname, const DBOptions& src) {
  }

 #ifndef ROCKSDB_LITE
+  ImmutableDBOptions immutable_db_options(result);
+  if (!IsWalDirSameAsDBPath(&immutable_db_options)) {
+    // Either the WAL dir and db_paths[0]/db_name are not the same, or we
+    // cannot tell for sure. In either case, assume they're different and
+    // explicitly cleanup the trash log files (bypass DeleteScheduler)
+    // Do this first so even if we end up calling
+    // DeleteScheduler::CleanupDirectory on the same dir later, it will be
+    // safe
+    std::vector<std::string> filenames;
+    result.env->GetChildren(result.wal_dir, &filenames);
+    for (std::string& filename : filenames) {
+      if (filename.find(".log.trash",
+                  filename.length() - std::string(".log.trash").length()) !=
+                  std::string::npos) {
+        std::string trash_file = result.wal_dir + "/" + filename;
+        result.env->DeleteFile(trash_file);
+      }
+    }
+  }
  // 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
@@ -145,7 +162,6 @@ DBOptions SanitizeOptions(const std::string& dbname, const DBOptions& src) {
 }

 namespace {
-
 Status SanitizeOptionsByTable(
    const DBOptions& db_opts,
    const std::vector<ColumnFamilyDescriptor>& column_families) {
@@ -158,38 +174,23 @@ Status SanitizeOptionsByTable(
  }
  return Status::OK();
 }
+}  // namespace

-static Status ValidateOptions(
+Status DBImpl::ValidateOptions(
    const DBOptions& db_options,
    const std::vector<ColumnFamilyDescriptor>& column_families) {
  Status s;
-
  for (auto& cfd : column_families) {
-    s = CheckCompressionSupported(cfd.options);
-    if (s.ok() && db_options.allow_concurrent_memtable_write) {
-      s = CheckConcurrentWritesSupported(cfd.options);
-    }
-    if (s.ok()) {
-      s = CheckCFPathsSupported(db_options, cfd.options);
-    }
+    s = ColumnFamilyData::ValidateOptions(db_options, cfd.options);
    if (!s.ok()) {
      return s;
    }
-
-    if (cfd.options.ttl > 0 || cfd.options.compaction_options_fifo.ttl > 0) {
-      if (db_options.max_open_files != -1) {
-        return Status::NotSupported(
-            "TTL is only supported when files are always "
-            "kept open (set max_open_files = -1). ");
-      }
-      if (cfd.options.table_factory->Name() !=
-          BlockBasedTableFactory().Name()) {
-        return Status::NotSupported(
-            "TTL is only supported in Block-Based Table format. ");
-      }
-    }
  }
+  s = ValidateOptions(db_options);
+  return s;
+}

+Status DBImpl::ValidateOptions(const DBOptions& db_options) {
  if (db_options.db_paths.size() > 4) {
    return Status::NotSupported(
        "More than four DB paths are not supported yet. ");
@@ -214,9 +215,20 @@ static Status ValidateOptions(
    return Status::InvalidArgument("keep_log_file_num must be greater than 0");
  }

+  if (db_options.unordered_write &&
+      !db_options.allow_concurrent_memtable_write) {
+    return Status::InvalidArgument(
+        "unordered_write is incompatible with !allow_concurrent_memtable_write");
+  }
+
+  if (db_options.unordered_write && db_options.enable_pipelined_write) {
+    return Status::InvalidArgument(
+        "unordered_write is incompatible with enable_pipelined_write");
+  }
+
  return Status::OK();
 }
-}  // namespace
+
 Status DBImpl::NewDB() {
  VersionEdit new_db;
  new_db.SetLogNumber(0);
@@ -272,9 +284,9 @@ Status DBImpl::CreateAndNewDirectory(Env* env, const std::string& dirname,
  return env->NewDirectory(dirname, directory);
 }

-Status DBImpl::Directories::SetDirectories(
-    Env* env, const std::string& dbname, const std::string& wal_dir,
-    const std::vector<DbPath>& data_paths) {
+Status Directories::SetDirectories(Env* env, const std::string& dbname,
+                                   const std::string& wal_dir,
+                                   const std::vector<DbPath>& data_paths) {
  Status s = DBImpl::CreateAndNewDirectory(env, dbname, &db_dir_);
  if (!s.ok()) {
    return s;
@@ -383,6 +395,7 @@ Status DBImpl::Recover(
  }

  Status s = versions_->Recover(column_families, read_only);
+
  if (immutable_db_options_.paranoid_checks && s.ok()) {
    s = CheckConsistency();
  }
@@ -394,6 +407,10 @@ Status DBImpl::Recover(
      }
    }
  }
+  // DB mutex is already held
+  if (s.ok() && immutable_db_options_.persist_stats_to_disk) {
+    s = InitPersistStatsColumnFamily();
+  }

  // Initial max_total_in_memory_state_ before recovery logs. Log recovery
  // may check this value to decide whether to flush.
@@ -409,6 +426,8 @@ Status DBImpl::Recover(
    default_cf_handle_ = new ColumnFamilyHandleImpl(
        versions_->GetColumnFamilySet()->GetDefault(), this, &mutex_);
    default_cf_internal_stats_ = default_cf_handle_->cfd()->internal_stats();
+    // TODO(Zhongyi): handle single_column_family_mode_ when
+    // persistent_stats is enabled
    single_column_family_mode_ =
        versions_->GetColumnFamilySet()->NumberOfColumnFamilies() == 1;

@@ -504,6 +523,98 @@ Status DBImpl::Recover(
  return s;
 }

+Status DBImpl::PersistentStatsProcessFormatVersion() {
+  mutex_.AssertHeld();
+  Status s;
+  // persist version when stats CF doesn't exist
+  bool should_persist_format_version = !persistent_stats_cfd_exists_;
+  mutex_.Unlock();
+  if (persistent_stats_cfd_exists_) {
+    // Check persistent stats format version compatibility. Drop and recreate
+    // persistent stats CF if format version is incompatible
+    uint64_t format_version_recovered = 0;
+    Status s_format = DecodePersistentStatsVersionNumber(
+        this, StatsVersionKeyType::kFormatVersion, &format_version_recovered);
+    uint64_t compatible_version_recovered = 0;
+    Status s_compatible = DecodePersistentStatsVersionNumber(
+        this, StatsVersionKeyType::kCompatibleVersion,
+        &compatible_version_recovered);
+    // abort reading from existing stats CF if any of following is true:
+    // 1. failed to read format version or compatible version from disk
+    // 2. sst's format version is greater than current format version, meaning
+    // this sst is encoded with a newer RocksDB release, and current compatible
+    // version is below the sst's compatible version
+    if (!s_format.ok() || !s_compatible.ok() ||
+        (kStatsCFCurrentFormatVersion < format_version_recovered &&
+         kStatsCFCompatibleFormatVersion < compatible_version_recovered)) {
+      if (!s_format.ok() || !s_compatible.ok()) {
+        ROCKS_LOG_INFO(
+            immutable_db_options_.info_log,
+            "Reading persistent stats version key failed. Format key: %s, "
+            "compatible key: %s",
+            s_format.ToString().c_str(), s_compatible.ToString().c_str());
+      } else {
+        ROCKS_LOG_INFO(
+            immutable_db_options_.info_log,
+            "Disable persistent stats due to corrupted or incompatible format "
+            "version\n");
+      }
+      DropColumnFamily(persist_stats_cf_handle_);
+      DestroyColumnFamilyHandle(persist_stats_cf_handle_);
+      ColumnFamilyHandle* handle = nullptr;
+      ColumnFamilyOptions cfo;
+      OptimizeForPersistentStats(&cfo);
+      s = CreateColumnFamily(cfo, kPersistentStatsColumnFamilyName, &handle);
+      persist_stats_cf_handle_ = static_cast<ColumnFamilyHandleImpl*>(handle);
+      // should also persist version here because old stats CF is discarded
+      should_persist_format_version = true;
+    }
+  }
+  if (s.ok() && should_persist_format_version) {
+    // Persistent stats CF being created for the first time, need to write
+    // format version key
+    WriteBatch batch;
+    batch.Put(persist_stats_cf_handle_, kFormatVersionKeyString,
+              ToString(kStatsCFCurrentFormatVersion));
+    batch.Put(persist_stats_cf_handle_, kCompatibleVersionKeyString,
+              ToString(kStatsCFCompatibleFormatVersion));
+    WriteOptions wo;
+    wo.low_pri = true;
+    wo.no_slowdown = true;
+    wo.sync = false;
+    s = Write(wo, &batch);
+  }
+  mutex_.Lock();
+  return s;
+}
+
+Status DBImpl::InitPersistStatsColumnFamily() {
+  mutex_.AssertHeld();
+  assert(!persist_stats_cf_handle_);
+  ColumnFamilyData* persistent_stats_cfd =
+      versions_->GetColumnFamilySet()->GetColumnFamily(
+          kPersistentStatsColumnFamilyName);
+  persistent_stats_cfd_exists_ = persistent_stats_cfd != nullptr;
+
+  Status s;
+  if (persistent_stats_cfd != nullptr) {
+    // We are recovering from a DB which already contains persistent stats CF,
+    // the CF is already created in VersionSet::ApplyOneVersionEdit, but
+    // column family handle was not. Need to explicitly create handle here.
+    persist_stats_cf_handle_ =
+        new ColumnFamilyHandleImpl(persistent_stats_cfd, this, &mutex_);
+  } else {
+    mutex_.Unlock();
+    ColumnFamilyHandle* handle = nullptr;
+    ColumnFamilyOptions cfo;
+    OptimizeForPersistentStats(&cfo);
+    s = CreateColumnFamily(cfo, kPersistentStatsColumnFamilyName, &handle);
+    persist_stats_cf_handle_ = static_cast<ColumnFamilyHandleImpl*>(handle);
+    mutex_.Lock();
+  }
+  return s;
+}
+
 // REQUIRES: log_numbers are sorted in ascending order
 Status DBImpl::RecoverLogFiles(const std::vector<uint64_t>& log_numbers,
                               SequenceNumber* next_sequence, bool read_only) {
@@ -512,7 +623,7 @@ Status DBImpl::RecoverLogFiles(const std::vector<uint64_t>& log_numbers,
    Logger* info_log;
    const char* fname;
    Status* status;  // nullptr if immutable_db_options_.paranoid_checks==false
-    virtual void Corruption(size_t bytes, const Status& s) override {
+    void Corruption(size_t bytes, const Status& s) override {
      ROCKS_LOG_WARN(info_log, "%s%s: dropping %d bytes; %s",
                     (this->status == nullptr ? "(ignoring error) " : ""),
                     fname, static_cast<int>(bytes), s.ToString().c_str());
@@ -563,12 +674,13 @@ Status DBImpl::RecoverLogFiles(const std::vector<uint64_t>& log_numbers,
  bool stop_replay_for_corruption = false;
  bool flushed = false;
  uint64_t corrupted_log_number = kMaxSequenceNumber;
+  uint64_t min_log_number = MinLogNumberToKeep();
  for (auto log_number : log_numbers) {
-    if (log_number < versions_->min_log_number_to_keep_2pc()) {
+    if (log_number < min_log_number) {
      ROCKS_LOG_INFO(immutable_db_options_.info_log,
                     "Skipping log #%" PRIu64
                     " since it is older than min log to keep #%" PRIu64,
-                     log_number, versions_->min_log_number_to_keep_2pc());
+                     log_number, min_log_number);
      continue;
    }
    // The previous incarnation may not have written any MANIFEST
@@ -580,7 +692,7 @@ Status DBImpl::RecoverLogFiles(const std::vector<uint64_t>& log_numbers,

    ROCKS_LOG_INFO(immutable_db_options_.info_log,
                   "Recovering log #%" PRIu64 " mode %d", log_number,
-                   immutable_db_options_.wal_recovery_mode);
+                   static_cast<int>(immutable_db_options_.wal_recovery_mode));
    auto logFileDropped = [this, &fname]() {
      uint64_t bytes;
      if (env_->GetFileSize(fname, &bytes).ok()) {
@@ -629,8 +741,7 @@ Status DBImpl::RecoverLogFiles(const std::vector<uint64_t>& log_numbers,
    // to be skipped instead of propagating bad information (like overly
    // large sequence numbers).
    log::Reader reader(immutable_db_options_.info_log, std::move(file_reader),
-                       &reporter, true /*checksum*/, log_number,
-                       false /* retry_after_eof */);
+                       &reporter, true /*checksum*/, log_number);

    // Determine if we should tolerate incomplete records at the tail end of the
    // Read all the records and add to a memtable
@@ -724,7 +835,8 @@ Status DBImpl::RecoverLogFiles(const std::vector<uint64_t>& log_numbers,
                " mode %d log filter %s returned "
                "more records (%d) than original (%d) which is not allowed. "
                "Aborting recovery.",
-                log_number, immutable_db_options_.wal_recovery_mode,
+                log_number,
+                static_cast<int>(immutable_db_options_.wal_recovery_mode),
                immutable_db_options_.wal_filter->Name(), new_count,
                original_count);
            status = Status::NotSupported(
@@ -1029,6 +1141,7 @@ Status DBImpl::WriteLevel0TableForRecovery(int job_id, ColumnFamilyData* cfd,
          cfd->int_tbl_prop_collector_factories(), cfd->GetID(), cfd->GetName(),
          snapshot_seqs, earliest_write_conflict_snapshot, snapshot_checker,
          GetCompressionFlush(*cfd->ioptions(), mutable_cf_options),
+          mutable_cf_options.sample_for_compression,
          cfd->ioptions()->compression_opts, paranoid_file_checks,
          cfd->internal_stats(), TableFileCreationReason::kRecovery,
          &event_logger_, job_id, Env::IO_HIGH, nullptr /* table_properties */,
@@ -1058,7 +1171,7 @@ Status DBImpl::WriteLevel0TableForRecovery(int job_id, ColumnFamilyData* cfd,
  stats.micros = env_->NowMicros() - start_micros;
  stats.bytes_written = meta.fd.GetFileSize();
  stats.num_output_files = 1;
-  cfd->internal_stats()->AddCompactionStats(level, stats);
+  cfd->internal_stats()->AddCompactionStats(level, Env::Priority::USER, stats);
  cfd->internal_stats()->AddCFStats(InternalStats::BYTES_FLUSHED,
                                    meta.fd.GetFileSize());
  RecordTick(stats_, COMPACT_WRITE_BYTES, meta.fd.GetFileSize());
@@ -1071,12 +1184,23 @@ Status DB::Open(const Options& options, const std::string& dbname, DB** dbptr) {
  std::vector<ColumnFamilyDescriptor> column_families;
  column_families.push_back(
      ColumnFamilyDescriptor(kDefaultColumnFamilyName, cf_options));
+  if (db_options.persist_stats_to_disk) {
+    column_families.push_back(
+        ColumnFamilyDescriptor(kPersistentStatsColumnFamilyName, cf_options));
+  }
  std::vector<ColumnFamilyHandle*> handles;
  Status s = DB::Open(db_options, dbname, column_families, &handles, dbptr);
  if (s.ok()) {
-    assert(handles.size() == 1);
+    if (db_options.persist_stats_to_disk) {
+      assert(handles.size() == 2);
+    } else {
+      assert(handles.size() == 1);
+    }
    // i can delete the handle since DBImpl is always holding a reference to
    // default column family
+    if (db_options.persist_stats_to_disk && handles[1] != nullptr) {
+      delete handles[1];
+    }
    delete handles[0];
  }
  return s;
@@ -1091,6 +1215,45 @@ Status DB::Open(const DBOptions& db_options, const std::string& dbname,
                      !kSeqPerBatch, kBatchPerTxn);
 }

+Status DBImpl::CreateWAL(uint64_t log_file_num, uint64_t recycle_log_number,
+                         size_t preallocate_block_size, log::Writer** new_log) {
+  Status s;
+  std::unique_ptr<WritableFile> lfile;
+
+  DBOptions db_options =
+      BuildDBOptions(immutable_db_options_, mutable_db_options_);
+  EnvOptions opt_env_options =
+      env_->OptimizeForLogWrite(env_options_, db_options);
+  std::string log_fname =
+      LogFileName(immutable_db_options_.wal_dir, log_file_num);
+
+  if (recycle_log_number) {
+    ROCKS_LOG_INFO(immutable_db_options_.info_log,
+                   "reusing log %" PRIu64 " from recycle list\n",
+                   recycle_log_number);
+    std::string old_log_fname =
+        LogFileName(immutable_db_options_.wal_dir, recycle_log_number);
+    s = env_->ReuseWritableFile(log_fname, old_log_fname, &lfile,
+                                opt_env_options);
+  } else {
+    s = NewWritableFile(env_, log_fname, &lfile, opt_env_options);
+  }
+
+  if (s.ok()) {
+    lfile->SetWriteLifeTimeHint(CalculateWALWriteHint());
+    lfile->SetPreallocationBlockSize(preallocate_block_size);
+
+    const auto& listeners = immutable_db_options_.listeners;
+    std::unique_ptr<WritableFileWriter> file_writer(
+        new WritableFileWriter(std::move(lfile), log_fname, opt_env_options,
+                               env_, nullptr /* stats */, listeners));
+    *new_log = new log::Writer(std::move(file_writer), log_file_num,
+                               immutable_db_options_.recycle_log_file_num > 0,
+                               immutable_db_options_.manual_wal_flush);
+  }
+  return s;
+}
+
 Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
                    const std::vector<ColumnFamilyDescriptor>& column_families,
                    std::vector<ColumnFamilyHandle*>* handles, DB** dbptr,
@@ -1150,41 +1313,28 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
    delete impl;
    return s;
  }
+
+  impl->wal_in_db_path_ =
+      IsWalDirSameAsDBPath(&impl->immutable_db_options_);
+
  impl->mutex_.Lock();
-  auto write_hint = impl->CalculateWALWriteHint();
  // Handles create_if_missing, error_if_exists
  s = impl->Recover(column_families);
  if (s.ok()) {
    uint64_t new_log_number = impl->versions_->NewFileNumber();
-    std::unique_ptr<WritableFile> lfile;
-    EnvOptions soptions(db_options);
-    EnvOptions opt_env_options =
-        impl->immutable_db_options_.env->OptimizeForLogWrite(
-            soptions, BuildDBOptions(impl->immutable_db_options_,
-                                     impl->mutable_db_options_));
-    std::string log_fname =
-        LogFileName(impl->immutable_db_options_.wal_dir, new_log_number);
-    s = NewWritableFile(impl->immutable_db_options_.env, log_fname, &lfile,
-                        opt_env_options);
+    log::Writer* new_log = nullptr;
+    const size_t preallocate_block_size =
+        impl->GetWalPreallocateBlockSize(max_write_buffer_size);
+    s = impl->CreateWAL(new_log_number, 0 /*recycle_log_number*/,
+                        preallocate_block_size, &new_log);
    if (s.ok()) {
-      lfile->SetWriteLifeTimeHint(write_hint);
-      lfile->SetPreallocationBlockSize(
-          impl->GetWalPreallocateBlockSize(max_write_buffer_size));
-      {
-        InstrumentedMutexLock wl(&impl->log_write_mutex_);
-        impl->logfile_number_ = new_log_number;
-        const auto& listeners = impl->immutable_db_options_.listeners;
-        std::unique_ptr<WritableFileWriter> file_writer(
-            new WritableFileWriter(std::move(lfile), log_fname, opt_env_options,
-                                   impl->env_, nullptr /* stats */, listeners));
-        impl->logs_.emplace_back(
-            new_log_number,
-            new log::Writer(
-                std::move(file_writer), new_log_number,
-                impl->immutable_db_options_.recycle_log_file_num > 0,
-                impl->immutable_db_options_.manual_wal_flush));
-      }
+      InstrumentedMutexLock wl(&impl->log_write_mutex_);
+      impl->logfile_number_ = new_log_number;
+      assert(new_log != nullptr);
+      impl->logs_.emplace_back(new_log_number, new_log);
+    }

+    if (s.ok()) {
      // set column family handles
      for (auto cf : column_families) {
        auto cfd =
@@ -1231,6 +1381,10 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
      s = impl->directories_.GetDbDir()->Fsync();
    }
  }
+  if (s.ok() && impl->immutable_db_options_.persist_stats_to_disk) {
+    // try to read format version but no need to fail Open() even if it fails
+    s = impl->PersistentStatsProcessFormatVersion();
+  }

  if (s.ok()) {
    for (auto cfd : *impl->versions_->GetColumnFamilySet()) {
@@ -1316,7 +1470,8 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
 #endif  // !ROCKSDB_LITE

  if (s.ok()) {
-    ROCKS_LOG_INFO(impl->immutable_db_options_.info_log, "DB pointer %p", impl);
+    ROCKS_LOG_HEADER(impl->immutable_db_options_.info_log, "DB pointer %p",
+                     impl);
    LogFlush(impl->immutable_db_options_.info_log);
    assert(impl->TEST_WALBufferIsEmpty());
    // If the assert above fails then we need to FlushWAL before returning
@@ -3,10 +3,10 @@
 //  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_readonly.h"
+#include "db/db_impl/db_impl_readonly.h"

 #include "db/compacted_db_impl.h"
-#include "db/db_impl.h"
+#include "db/db_impl/db_impl.h"
 #include "db/db_iter.h"
 #include "db/merge_context.h"
 #include "monitoring/perf_context_imp.h"
@@ -60,7 +60,7 @@ Status DBImplReadOnly::Get(const ReadOptions& read_options,
  RecordTick(stats_, NUMBER_KEYS_READ);
  size_t size = pinnable_val->size();
  RecordTick(stats_, BYTES_READ, size);
-  MeasureTime(stats_, BYTES_PER_READ, size);
+  RecordInHistogram(stats_, BYTES_PER_READ, size);
  PERF_COUNTER_ADD(get_read_bytes, size);
  return s;
 }
@@ -7,9 +7,9 @@

 #ifndef ROCKSDB_LITE

-#include "db/db_impl.h"
-#include <vector>
 #include <string>
+#include <vector>
+#include "db/db_impl/db_impl.h"

 namespace rocksdb {

@@ -115,6 +115,16 @@ class DBImplReadOnly : public DBImpl {
    return Status::NotSupported("Not supported operation in read only mode.");
  }

+  using DB::CreateColumnFamilyWithImport;
+  virtual Status CreateColumnFamilyWithImport(
+      const ColumnFamilyOptions& /*options*/,
+      const std::string& /*column_family_name*/,
+      const ImportColumnFamilyOptions& /*import_options*/,
+      const ExportImportFilesMetaData& /*metadata*/,
+      ColumnFamilyHandle** /*handle*/) override {
+    return Status::NotSupported("Not supported operation in read only mode.");
+  }
+
 private:
  friend class DB;

@@ -122,6 +132,6 @@ class DBImplReadOnly : public DBImpl {
  DBImplReadOnly(const DBImplReadOnly&);
  void operator=(const DBImplReadOnly&);
 };
-}
+}  // namespace rocksdb

 #endif  // !ROCKSDB_LITE
@@ -0,0 +1,648 @@
+//  Copyright (c) 2011-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_secondary.h"
+
+#include <cinttypes>
+
+#include "db/db_iter.h"
+#include "db/merge_context.h"
+#include "logging/auto_roll_logger.h"
+#include "monitoring/perf_context_imp.h"
+
+namespace rocksdb {
+
+#ifndef ROCKSDB_LITE
+DBImplSecondary::DBImplSecondary(const DBOptions& db_options,
+                                 const std::string& dbname)
+    : DBImpl(db_options, dbname) {
+  ROCKS_LOG_INFO(immutable_db_options_.info_log,
+                 "Opening the db in secondary mode");
+  LogFlush(immutable_db_options_.info_log);
+}
+
+DBImplSecondary::~DBImplSecondary() {}
+
+Status DBImplSecondary::Recover(
+    const std::vector<ColumnFamilyDescriptor>& column_families,
+    bool /*readonly*/, bool /*error_if_log_file_exist*/,
+    bool /*error_if_data_exists_in_logs*/) {
+  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()) {
+    return s;
+  }
+  if (immutable_db_options_.paranoid_checks && s.ok()) {
+    s = CheckConsistency();
+  }
+  // Initial max_total_in_memory_state_ before recovery logs.
+  max_total_in_memory_state_ = 0;
+  for (auto cfd : *versions_->GetColumnFamilySet()) {
+    auto* mutable_cf_options = cfd->GetLatestMutableCFOptions();
+    max_total_in_memory_state_ += mutable_cf_options->write_buffer_size *
+                                  mutable_cf_options->max_write_buffer_number;
+  }
+  if (s.ok()) {
+    default_cf_handle_ = new ColumnFamilyHandleImpl(
+        versions_->GetColumnFamilySet()->GetDefault(), this, &mutex_);
+    default_cf_internal_stats_ = default_cf_handle_->cfd()->internal_stats();
+    single_column_family_mode_ =
+        versions_->GetColumnFamilySet()->NumberOfColumnFamilies() == 1;
+
+    std::unordered_set<ColumnFamilyData*> cfds_changed;
+    s = FindAndRecoverLogFiles(&cfds_changed, &job_context);
+  }
+
+  if (s.IsPathNotFound()) {
+    ROCKS_LOG_INFO(immutable_db_options_.info_log,
+                   "Secondary tries to read WAL, but WAL file(s) have already "
+                   "been purged by primary.");
+    s = Status::OK();
+  }
+  // TODO: update options_file_number_ needed?
+
+  job_context.Clean();
+  return s;
+}
+
+// find new WAL and apply them in order to the secondary instance
+Status DBImplSecondary::FindAndRecoverLogFiles(
+    std::unordered_set<ColumnFamilyData*>* cfds_changed,
+    JobContext* job_context) {
+  assert(nullptr != cfds_changed);
+  assert(nullptr != job_context);
+  Status s;
+  std::vector<uint64_t> logs;
+  s = FindNewLogNumbers(&logs);
+  if (s.ok() && !logs.empty()) {
+    SequenceNumber next_sequence(kMaxSequenceNumber);
+    s = RecoverLogFiles(logs, &next_sequence, cfds_changed, job_context);
+  }
+  return s;
+}
+
+// List wal_dir and find all new WALs, return these log numbers
+Status DBImplSecondary::FindNewLogNumbers(std::vector<uint64_t>* logs) {
+  assert(logs != nullptr);
+  std::vector<std::string> filenames;
+  Status s;
+  s = env_->GetChildren(immutable_db_options_.wal_dir, &filenames);
+  if (s.IsNotFound()) {
+    return Status::InvalidArgument("Failed to open wal_dir",
+                                   immutable_db_options_.wal_dir);
+  } else if (!s.ok()) {
+    return s;
+  }
+
+  // if log_readers_ is non-empty, it means we have applied all logs with log
+  // numbers smaller than the smallest log in log_readers_, so there is no
+  // need to pass these logs to RecoverLogFiles
+  uint64_t log_number_min = 0;
+  if (!log_readers_.empty()) {
+    log_number_min = log_readers_.begin()->first;
+  }
+  for (size_t i = 0; i < filenames.size(); i++) {
+    uint64_t number;
+    FileType type;
+    if (ParseFileName(filenames[i], &number, &type) && type == kLogFile &&
+        number >= log_number_min) {
+      logs->push_back(number);
+    }
+  }
+  // Recover logs in the order that they were generated
+  if (!logs->empty()) {
+    std::sort(logs->begin(), logs->end());
+  }
+  return s;
+}
+
+Status DBImplSecondary::MaybeInitLogReader(
+    uint64_t log_number, log::FragmentBufferedReader** log_reader) {
+  auto iter = log_readers_.find(log_number);
+  // make sure the log file is still present
+  if (iter == log_readers_.end() ||
+      iter->second->reader_->GetLogNumber() != log_number) {
+    // delete the obsolete log reader if log number mismatch
+    if (iter != log_readers_.end()) {
+      log_readers_.erase(iter);
+    }
+    // initialize log reader from log_number
+    // TODO: min_log_number_to_keep_2pc check needed?
+    // Open the log file
+    std::string fname = LogFileName(immutable_db_options_.wal_dir, log_number);
+    ROCKS_LOG_INFO(immutable_db_options_.info_log,
+                   "Recovering log #%" PRIu64 " mode %d", log_number,
+                   static_cast<int>(immutable_db_options_.wal_recovery_mode));
+
+    std::unique_ptr<SequentialFileReader> file_reader;
+    {
+      std::unique_ptr<SequentialFile> file;
+      Status status = env_->NewSequentialFile(
+          fname, &file, env_->OptimizeForLogRead(env_options_));
+      if (!status.ok()) {
+        *log_reader = nullptr;
+        return status;
+      }
+      file_reader.reset(new SequentialFileReader(std::move(file), fname));
+    }
+
+    // Create the log reader.
+    LogReaderContainer* log_reader_container = new LogReaderContainer(
+        env_, immutable_db_options_.info_log, std::move(fname),
+        std::move(file_reader), log_number);
+    log_readers_.insert(std::make_pair(
+        log_number, std::unique_ptr<LogReaderContainer>(log_reader_container)));
+  }
+  iter = log_readers_.find(log_number);
+  assert(iter != log_readers_.end());
+  *log_reader = iter->second->reader_;
+  return Status::OK();
+}
+
+// After manifest recovery, replay WALs and refresh log_readers_ if necessary
+// REQUIRES: log_numbers are sorted in ascending order
+Status DBImplSecondary::RecoverLogFiles(
+    const std::vector<uint64_t>& log_numbers, SequenceNumber* next_sequence,
+    std::unordered_set<ColumnFamilyData*>* cfds_changed,
+    JobContext* job_context) {
+  assert(nullptr != cfds_changed);
+  assert(nullptr != job_context);
+  mutex_.AssertHeld();
+  Status status;
+  for (auto log_number : log_numbers) {
+    log::FragmentBufferedReader* reader = nullptr;
+    status = MaybeInitLogReader(log_number, &reader);
+    if (!status.ok()) {
+      return status;
+    }
+    assert(reader != nullptr);
+  }
+  for (auto log_number : log_numbers) {
+    auto it  = log_readers_.find(log_number);
+    assert(it != log_readers_.end());
+    log::FragmentBufferedReader* reader = it->second->reader_;
+    // Manually update the file number allocation counter in VersionSet.
+    versions_->MarkFileNumberUsed(log_number);
+
+    // Determine if we should tolerate incomplete records at the tail end of the
+    // Read all the records and add to a memtable
+    std::string scratch;
+    Slice record;
+    WriteBatch batch;
+
+    while (reader->ReadRecord(&record, &scratch,
+                              immutable_db_options_.wal_recovery_mode) &&
+           status.ok()) {
+      if (record.size() < WriteBatchInternal::kHeader) {
+        reader->GetReporter()->Corruption(
+            record.size(), Status::Corruption("log record too small"));
+        continue;
+      }
+      WriteBatchInternal::SetContents(&batch, record);
+      SequenceNumber seq_of_batch = WriteBatchInternal::Sequence(&batch);
+      std::vector<uint32_t> column_family_ids;
+      status = CollectColumnFamilyIdsFromWriteBatch(batch, &column_family_ids);
+      if (status.ok()) {
+        for (const auto id : column_family_ids) {
+          ColumnFamilyData* cfd =
+              versions_->GetColumnFamilySet()->GetColumnFamily(id);
+          if (cfd == nullptr) {
+            continue;
+          }
+          if (cfds_changed->count(cfd) == 0) {
+            cfds_changed->insert(cfd);
+          }
+          const std::vector<FileMetaData*>& l0_files =
+              cfd->current()->storage_info()->LevelFiles(0);
+          SequenceNumber seq =
+              l0_files.empty() ? 0 : l0_files.back()->fd.largest_seqno;
+          // If the write batch's sequence number is smaller than the last
+          // sequence number of the largest sequence persisted for this column
+          // family, then its data must reside in an SST that has already been
+          // added in the prior MANIFEST replay.
+          if (seq_of_batch <= seq) {
+            continue;
+          }
+          auto curr_log_num = port::kMaxUint64;
+          if (cfd_to_current_log_.count(cfd) > 0) {
+            curr_log_num = cfd_to_current_log_[cfd];
+          }
+          // If the active memtable contains records added by replaying an
+          // earlier WAL, then we need to seal the memtable, add it to the
+          // immutable memtable list and create a new active memtable.
+          if (!cfd->mem()->IsEmpty() && (curr_log_num == port::kMaxUint64 ||
+                                         curr_log_num != log_number)) {
+            const MutableCFOptions mutable_cf_options =
+                *cfd->GetLatestMutableCFOptions();
+            MemTable* new_mem =
+                cfd->ConstructNewMemtable(mutable_cf_options, seq_of_batch);
+            cfd->mem()->SetNextLogNumber(log_number);
+            cfd->imm()->Add(cfd->mem(), &job_context->memtables_to_free);
+            new_mem->Ref();
+            cfd->SetMemtable(new_mem);
+          }
+        }
+        bool has_valid_writes = false;
+        status = WriteBatchInternal::InsertInto(
+            &batch, column_family_memtables_.get(),
+            nullptr /* flush_scheduler */, true, log_number, this,
+            false /* concurrent_memtable_writes */, next_sequence,
+            &has_valid_writes, seq_per_batch_, batch_per_txn_);
+      }
+      // If column family was not found, it might mean that the WAL write
+      // batch references to the column family that was dropped after the
+      // insert. We don't want to fail the whole write batch in that case --
+      // we just ignore the update.
+      // That's why we set ignore missing column families to true
+      // passing null flush_scheduler will disable memtable flushing which is
+      // needed for secondary instances
+      if (status.ok()) {
+        for (const auto id : column_family_ids) {
+          ColumnFamilyData* cfd =
+              versions_->GetColumnFamilySet()->GetColumnFamily(id);
+          if (cfd == nullptr) {
+            continue;
+          }
+          std::unordered_map<ColumnFamilyData*, uint64_t>::iterator iter =
+              cfd_to_current_log_.find(cfd);
+          if (iter == cfd_to_current_log_.end()) {
+            cfd_to_current_log_.insert({cfd, log_number});
+          } else if (log_number > iter->second) {
+            iter->second = log_number;
+          }
+        }
+        auto last_sequence = *next_sequence - 1;
+        if ((*next_sequence != kMaxSequenceNumber) &&
+            (versions_->LastSequence() <= last_sequence)) {
+          versions_->SetLastAllocatedSequence(last_sequence);
+          versions_->SetLastPublishedSequence(last_sequence);
+          versions_->SetLastSequence(last_sequence);
+        }
+      } else {
+        // We are treating this as a failure while reading since we read valid
+        // blocks that do not form coherent data
+        reader->GetReporter()->Corruption(record.size(), status);
+      }
+    }
+    if (!status.ok()) {
+      return status;
+    }
+  }
+  // remove logreaders from map after successfully recovering the WAL
+  if (log_readers_.size() > 1) {
+    auto erase_iter = log_readers_.begin();
+    std::advance(erase_iter, log_readers_.size() - 1);
+    log_readers_.erase(log_readers_.begin(), erase_iter);
+  }
+  return status;
+}
+
+// Implementation of the DB interface
+Status DBImplSecondary::Get(const ReadOptions& read_options,
+                            ColumnFamilyHandle* column_family, const Slice& key,
+                            PinnableSlice* value) {
+  return GetImpl(read_options, column_family, key, value);
+}
+
+Status DBImplSecondary::GetImpl(const ReadOptions& read_options,
+                                ColumnFamilyHandle* column_family,
+                                const Slice& key, PinnableSlice* pinnable_val) {
+  assert(pinnable_val != nullptr);
+  PERF_CPU_TIMER_GUARD(get_cpu_nanos, env_);
+  StopWatch sw(env_, stats_, DB_GET);
+  PERF_TIMER_GUARD(get_snapshot_time);
+
+  auto cfh = static_cast<ColumnFamilyHandleImpl*>(column_family);
+  ColumnFamilyData* cfd = cfh->cfd();
+  if (tracer_) {
+    InstrumentedMutexLock lock(&trace_mutex_);
+    if (tracer_) {
+      tracer_->Get(column_family, key);
+    }
+  }
+  // Acquire SuperVersion
+  SuperVersion* super_version = GetAndRefSuperVersion(cfd);
+  SequenceNumber snapshot = versions_->LastSequence();
+  MergeContext merge_context;
+  SequenceNumber max_covering_tombstone_seq = 0;
+  Status s;
+  LookupKey lkey(key, snapshot);
+  PERF_TIMER_STOP(get_snapshot_time);
+
+  bool done = false;
+  if (super_version->mem->Get(lkey, pinnable_val->GetSelf(), &s, &merge_context,
+                              &max_covering_tombstone_seq, read_options)) {
+    done = true;
+    pinnable_val->PinSelf();
+    RecordTick(stats_, MEMTABLE_HIT);
+  } else if ((s.ok() || s.IsMergeInProgress()) &&
+             super_version->imm->Get(
+                 lkey, pinnable_val->GetSelf(), &s, &merge_context,
+                 &max_covering_tombstone_seq, read_options)) {
+    done = true;
+    pinnable_val->PinSelf();
+    RecordTick(stats_, MEMTABLE_HIT);
+  }
+  if (!done && !s.ok() && !s.IsMergeInProgress()) {
+    ReturnAndCleanupSuperVersion(cfd, super_version);
+    return s;
+  }
+  if (!done) {
+    PERF_TIMER_GUARD(get_from_output_files_time);
+    super_version->current->Get(read_options, lkey, pinnable_val, &s,
+                                &merge_context, &max_covering_tombstone_seq);
+    RecordTick(stats_, MEMTABLE_MISS);
+  }
+  {
+    PERF_TIMER_GUARD(get_post_process_time);
+    ReturnAndCleanupSuperVersion(cfd, super_version);
+    RecordTick(stats_, NUMBER_KEYS_READ);
+    size_t size = pinnable_val->size();
+    RecordTick(stats_, BYTES_READ, size);
+    RecordTimeToHistogram(stats_, BYTES_PER_READ, size);
+    PERF_COUNTER_ADD(get_read_bytes, size);
+  }
+  return s;
+}
+
+Iterator* DBImplSecondary::NewIterator(const ReadOptions& read_options,
+                                       ColumnFamilyHandle* column_family) {
+  if (read_options.managed) {
+    return NewErrorIterator(
+        Status::NotSupported("Managed iterator is not supported anymore."));
+  }
+  if (read_options.read_tier == kPersistedTier) {
+    return NewErrorIterator(Status::NotSupported(
+        "ReadTier::kPersistedData is not yet supported in iterators."));
+  }
+  Iterator* result = nullptr;
+  auto cfh = reinterpret_cast<ColumnFamilyHandleImpl*>(column_family);
+  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"));
+  } else if (read_options.snapshot != nullptr) {
+    // TODO (yanqin) support snapshot.
+    return NewErrorIterator(
+        Status::NotSupported("snapshot not supported in secondary mode"));
+  } else {
+    auto snapshot = versions_->LastSequence();
+    result = NewIteratorImpl(read_options, cfd, snapshot, read_callback);
+  }
+  return result;
+}
+
+ArenaWrappedDBIter* DBImplSecondary::NewIteratorImpl(
+    const ReadOptions& read_options, ColumnFamilyData* cfd,
+    SequenceNumber snapshot, ReadCallback* read_callback) {
+  assert(nullptr != cfd);
+  SuperVersion* super_version = cfd->GetReferencedSuperVersion(&mutex_);
+  auto db_iter = NewArenaWrappedDbIterator(
+      env_, read_options, *cfd->ioptions(), super_version->mutable_cf_options,
+      snapshot,
+      super_version->mutable_cf_options.max_sequential_skip_in_iterations,
+      super_version->version_number, read_callback);
+  auto internal_iter =
+      NewInternalIterator(read_options, cfd, super_version, db_iter->GetArena(),
+                          db_iter->GetRangeDelAggregator(), snapshot);
+  db_iter->SetIterUnderDBIter(internal_iter);
+  return db_iter;
+}
+
+Status DBImplSecondary::NewIterators(
+    const ReadOptions& read_options,
+    const std::vector<ColumnFamilyHandle*>& column_families,
+    std::vector<Iterator*>* iterators) {
+  if (read_options.managed) {
+    return Status::NotSupported("Managed iterator is not supported anymore.");
+  }
+  if (read_options.read_tier == kPersistedTier) {
+    return Status::NotSupported(
+        "ReadTier::kPersistedData is not yet supported in iterators.");
+  }
+  ReadCallback* read_callback = nullptr;  // No read callback provided.
+  if (iterators == nullptr) {
+    return Status::InvalidArgument("iterators not allowed to be nullptr");
+  }
+  iterators->clear();
+  iterators->reserve(column_families.size());
+  if (read_options.tailing) {
+    return Status::NotSupported(
+        "tailing iterator not supported in secondary mode");
+  } else if (read_options.snapshot != nullptr) {
+    // TODO (yanqin) support snapshot.
+    return Status::NotSupported("snapshot not supported in secondary mode");
+  } else {
+    SequenceNumber read_seq = versions_->LastSequence();
+    for (auto cfh : column_families) {
+      ColumnFamilyData* cfd = static_cast<ColumnFamilyHandleImpl*>(cfh)->cfd();
+      iterators->push_back(
+          NewIteratorImpl(read_options, cfd, read_seq, read_callback));
+    }
+  }
+  return Status::OK();
+}
+
+Status DBImplSecondary::CheckConsistency() {
+  mutex_.AssertHeld();
+  Status s = DBImpl::CheckConsistency();
+  // If DBImpl::CheckConsistency() which is stricter returns success, then we
+  // do not need to give a second chance.
+  if (s.ok()) {
+    return s;
+  }
+  // It's possible that DBImpl::CheckConssitency() can fail because the primary
+  // may have removed certain files, causing the GetFileSize(name) call to
+  // fail and returning a PathNotFound. In this case, we take a best-effort
+  // approach and just proceed.
+  TEST_SYNC_POINT_CALLBACK(
+      "DBImplSecondary::CheckConsistency:AfterFirstAttempt", &s);
+  std::vector<LiveFileMetaData> metadata;
+  versions_->GetLiveFilesMetaData(&metadata);
+
+  std::string corruption_messages;
+  for (const auto& md : metadata) {
+    // md.name has a leading "/".
+    std::string file_path = md.db_path + md.name;
+
+    uint64_t fsize = 0;
+    s = env_->GetFileSize(file_path, &fsize);
+    if (!s.ok() &&
+        (env_->GetFileSize(Rocks2LevelTableFileName(file_path), &fsize).ok() ||
+         s.IsPathNotFound())) {
+      s = Status::OK();
+    }
+    if (!s.ok()) {
+      corruption_messages +=
+          "Can't access " + md.name + ": " + s.ToString() + "\n";
+    }
+  }
+  return corruption_messages.empty() ? Status::OK()
+                                     : Status::Corruption(corruption_messages);
+}
+
+Status DBImplSecondary::TryCatchUpWithPrimary() {
+  assert(versions_.get() != nullptr);
+  assert(manifest_reader_.get() != nullptr);
+  Status s;
+  // read the manifest and apply new changes to the secondary instance
+  std::unordered_set<ColumnFamilyData*> cfds_changed;
+  JobContext job_context(0, true /*create_superversion*/);
+  InstrumentedMutexLock lock_guard(&mutex_);
+  s = static_cast<ReactiveVersionSet*>(versions_.get())
+          ->ReadAndApply(&mutex_, &manifest_reader_, &cfds_changed);
+
+  ROCKS_LOG_INFO(immutable_db_options_.info_log, "Last sequence is %" PRIu64,
+                 static_cast<uint64_t>(versions_->LastSequence()));
+  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));
+  }
+
+  // list wal_dir to discover new WALs and apply new changes to the secondary
+  // instance
+  if (s.ok()) {
+    s = FindAndRecoverLogFiles(&cfds_changed, &job_context);
+  }
+  if (s.IsPathNotFound()) {
+    ROCKS_LOG_INFO(immutable_db_options_.info_log,
+                   "Secondary tries to read WAL, but WAL file(s) have already "
+                   "been purged by primary.");
+    s = Status::OK();
+  }
+  if (s.ok()) {
+    for (auto cfd : cfds_changed) {
+      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();
+    }
+    job_context.Clean();
+  }
+  return s;
+}
+
+Status DB::OpenAsSecondary(const Options& options, const std::string& dbname,
+                           const std::string& secondary_path, DB** dbptr) {
+  *dbptr = nullptr;
+
+  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::OpenAsSecondary(db_options, dbname, secondary_path,
+                                 column_families, &handles, dbptr);
+  if (s.ok()) {
+    assert(handles.size() == 1);
+    delete handles[0];
+  }
+  return s;
+}
+
+Status DB::OpenAsSecondary(
+    const DBOptions& db_options, const std::string& dbname,
+    const std::string& secondary_path,
+    const std::vector<ColumnFamilyDescriptor>& column_families,
+    std::vector<ColumnFamilyHandle*>* handles, DB** dbptr) {
+  *dbptr = nullptr;
+  if (db_options.max_open_files != -1) {
+    // TODO (yanqin) maybe support max_open_files != -1 by creating hard links
+    // on SST files so that db secondary can still have access to old SSTs
+    // while primary instance may delete original.
+    return Status::InvalidArgument("require max_open_files to be -1");
+  }
+
+  DBOptions tmp_opts(db_options);
+  Status s;
+  if (nullptr == tmp_opts.info_log) {
+    s = CreateLoggerFromOptions(secondary_path, tmp_opts, &tmp_opts.info_log);
+    if (!s.ok()) {
+      tmp_opts.info_log = nullptr;
+    }
+  }
+
+  handles->clear();
+  DBImplSecondary* impl = new DBImplSecondary(tmp_opts, dbname);
+  impl->versions_.reset(new ReactiveVersionSet(
+      dbname, &impl->immutable_db_options_, impl->env_options_,
+      impl->table_cache_.get(), impl->write_buffer_manager_,
+      &impl->write_controller_));
+  impl->column_family_memtables_.reset(
+      new ColumnFamilyMemTablesImpl(impl->versions_->GetColumnFamilySet()));
+  impl->wal_in_db_path_ =
+      IsWalDirSameAsDBPath(&impl->immutable_db_options_);
+
+  impl->mutex_.Lock();
+  s = impl->Recover(column_families, true, false, false);
+  if (s.ok()) {
+    for (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(true /* 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 = impl;
+    for (auto h : *handles) {
+      impl->NewThreadStatusCfInfo(
+          reinterpret_cast<ColumnFamilyHandleImpl*>(h)->cfd());
+    }
+  } else {
+    for (auto h : *handles) {
+      delete h;
+    }
+    handles->clear();
+    delete impl;
+  }
+  return s;
+}
+#else   // !ROCKSDB_LITE
+
+Status DB::OpenAsSecondary(const Options& /*options*/,
+                           const std::string& /*name*/,
+                           const std::string& /*secondary_path*/,
+                           DB** /*dbptr*/) {
+  return Status::NotSupported("Not supported in ROCKSDB_LITE.");
+}
+
+Status DB::OpenAsSecondary(
+    const DBOptions& /*db_options*/, const std::string& /*dbname*/,
+    const std::string& /*secondary_path*/,
+    const std::vector<ColumnFamilyDescriptor>& /*column_families*/,
+    std::vector<ColumnFamilyHandle*>* /*handles*/, DB** /*dbptr*/) {
+  return Status::NotSupported("Not supported in ROCKSDB_LITE.");
+}
+#endif  // !ROCKSDB_LITE
+
+}  // namespace rocksdb
@@ -0,0 +1,306 @@
+//  Copyright (c) 2011-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
+
+#ifndef ROCKSDB_LITE
+
+#include <string>
+#include <vector>
+#include "db/db_impl/db_impl.h"
+
+namespace rocksdb {
+
+// A wrapper class to hold log reader, log reporter, log status.
+class LogReaderContainer {
+ public:
+  LogReaderContainer()
+      : reader_(nullptr), reporter_(nullptr), status_(nullptr) {}
+  LogReaderContainer(Env* env, std::shared_ptr<Logger> info_log,
+                     std::string fname,
+                     std::unique_ptr<SequentialFileReader>&& file_reader,
+                     uint64_t log_number) {
+    LogReporter* reporter = new LogReporter();
+    status_ = new Status();
+    reporter->env = env;
+    reporter->info_log = info_log.get();
+    reporter->fname = std::move(fname);
+    reporter->status = status_;
+    reporter_ = reporter;
+    // We intentially make log::Reader do checksumming even if
+    // paranoid_checks==false so that corruptions cause entire commits
+    // to be skipped instead of propagating bad information (like overly
+    // large sequence numbers).
+    reader_ = new log::FragmentBufferedReader(info_log, std::move(file_reader),
+                                              reporter, true /*checksum*/,
+                                              log_number);
+  }
+  log::FragmentBufferedReader* reader_;
+  log::Reader::Reporter* reporter_;
+  Status* status_;
+  ~LogReaderContainer() {
+    delete reader_;
+    delete reporter_;
+    delete status_;
+  }
+ private:
+  struct LogReporter : public log::Reader::Reporter {
+    Env* env;
+    Logger* info_log;
+    std::string fname;
+    Status* status;  // nullptr if immutable_db_options_.paranoid_checks==false
+    void Corruption(size_t bytes, const Status& s) override {
+      ROCKS_LOG_WARN(info_log, "%s%s: dropping %d bytes; %s",
+                     (this->status == nullptr ? "(ignoring error) " : ""),
+                     fname.c_str(), static_cast<int>(bytes),
+                     s.ToString().c_str());
+      if (this->status != nullptr && this->status->ok()) {
+        *this->status = s;
+      }
+    }
+  };
+};
+
+// The secondary instance shares access to the storage as the primary.
+// The secondary is able to read and replay changes described in both the
+// MANIFEST and the WAL files without coordination with the primary.
+// The secondary instance can be opened using `DB::OpenAsSecondary`. After
+// that, it can call `DBImplSecondary::TryCatchUpWithPrimary` to make best
+// effort attempts to catch up with the primary.
+class DBImplSecondary : public DBImpl {
+ public:
+  DBImplSecondary(const DBOptions& options, const std::string& dbname);
+  ~DBImplSecondary() override;
+
+  // Recover by replaying MANIFEST and WAL. Also initialize manifest_reader_
+  // and log_readers_ to facilitate future operations.
+  Status Recover(const std::vector<ColumnFamilyDescriptor>& column_families,
+                 bool read_only, bool error_if_log_file_exist,
+                 bool error_if_data_exists_in_logs) override;
+
+  // Implementations of the DB interface
+  using DB::Get;
+  Status Get(const ReadOptions& options, ColumnFamilyHandle* column_family,
+             const Slice& key, PinnableSlice* value) override;
+
+  Status GetImpl(const ReadOptions& options, ColumnFamilyHandle* column_family,
+                 const Slice& key, PinnableSlice* value);
+
+  using DBImpl::NewIterator;
+  Iterator* NewIterator(const ReadOptions&,
+                        ColumnFamilyHandle* column_family) override;
+
+  ArenaWrappedDBIter* NewIteratorImpl(const ReadOptions& read_options,
+                                      ColumnFamilyData* cfd,
+                                      SequenceNumber snapshot,
+                                      ReadCallback* read_callback);
+
+  Status NewIterators(const ReadOptions& options,
+                      const std::vector<ColumnFamilyHandle*>& column_families,
+                      std::vector<Iterator*>* iterators) override;
+
+  using DBImpl::Put;
+  Status Put(const WriteOptions& /*options*/,
+             ColumnFamilyHandle* /*column_family*/, const Slice& /*key*/,
+             const Slice& /*value*/) override {
+    return Status::NotSupported("Not supported operation in secondary mode.");
+  }
+
+  using DBImpl::Merge;
+  Status Merge(const WriteOptions& /*options*/,
+               ColumnFamilyHandle* /*column_family*/, const Slice& /*key*/,
+               const Slice& /*value*/) override {
+    return Status::NotSupported("Not supported operation in secondary mode.");
+  }
+
+  using DBImpl::Delete;
+  Status Delete(const WriteOptions& /*options*/,
+                ColumnFamilyHandle* /*column_family*/,
+                const Slice& /*key*/) override {
+    return Status::NotSupported("Not supported operation in secondary mode.");
+  }
+
+  using DBImpl::SingleDelete;
+  Status SingleDelete(const WriteOptions& /*options*/,
+                      ColumnFamilyHandle* /*column_family*/,
+                      const Slice& /*key*/) override {
+    return Status::NotSupported("Not supported operation in secondary mode.");
+  }
+
+  Status Write(const WriteOptions& /*options*/,
+               WriteBatch* /*updates*/) override {
+    return Status::NotSupported("Not supported operation in secondary mode.");
+  }
+
+  using DBImpl::CompactRange;
+  Status CompactRange(const CompactRangeOptions& /*options*/,
+                      ColumnFamilyHandle* /*column_family*/,
+                      const Slice* /*begin*/, const Slice* /*end*/) override {
+    return Status::NotSupported("Not supported operation in secondary mode.");
+  }
+
+  using DBImpl::CompactFiles;
+  Status CompactFiles(
+      const CompactionOptions& /*compact_options*/,
+      ColumnFamilyHandle* /*column_family*/,
+      const std::vector<std::string>& /*input_file_names*/,
+      const int /*output_level*/, const int /*output_path_id*/ = -1,
+      std::vector<std::string>* const /*output_file_names*/ = nullptr,
+      CompactionJobInfo* /*compaction_job_info*/ = nullptr) override {
+    return Status::NotSupported("Not supported operation in secondary mode.");
+  }
+
+  Status DisableFileDeletions() override {
+    return Status::NotSupported("Not supported operation in secondary mode.");
+  }
+
+  Status EnableFileDeletions(bool /*force*/) override {
+    return Status::NotSupported("Not supported operation in secondary mode.");
+  }
+
+  Status GetLiveFiles(std::vector<std::string>&,
+                      uint64_t* /*manifest_file_size*/,
+                      bool /*flush_memtable*/ = true) override {
+    return Status::NotSupported("Not supported operation in secondary mode.");
+  }
+
+  using DBImpl::Flush;
+  Status Flush(const FlushOptions& /*options*/,
+               ColumnFamilyHandle* /*column_family*/) override {
+    return Status::NotSupported("Not supported operation in secondary mode.");
+  }
+
+  using DBImpl::SyncWAL;
+  Status SyncWAL() override {
+    return Status::NotSupported("Not supported operation in secondary mode.");
+  }
+
+  using DB::IngestExternalFile;
+  Status IngestExternalFile(
+      ColumnFamilyHandle* /*column_family*/,
+      const std::vector<std::string>& /*external_files*/,
+      const IngestExternalFileOptions& /*ingestion_options*/) override {
+    return Status::NotSupported("Not supported operation in secondary mode.");
+  }
+
+  // Try to catch up with the primary by reading as much as possible from the
+  // log files until there is nothing more to read or encounters an error. If
+  // the amount of information in the log files to process is huge, this
+  // method can take long time due to all the I/O and CPU costs.
+  Status TryCatchUpWithPrimary() override;
+
+
+  // Try to find log reader using log_number from log_readers_ map, initialize
+  // if it doesn't exist
+  Status MaybeInitLogReader(uint64_t log_number,
+                            log::FragmentBufferedReader** log_reader);
+
+  // Check if all live files exist on file system and that their file sizes
+  // matche to the in-memory records. It is possible that some live files may
+  // have been deleted by the primary. In this case, CheckConsistency() does
+  // not flag the missing file as inconsistency.
+  Status CheckConsistency() override;
+
+ protected:
+  // ColumnFamilyCollector is a write batch handler which does nothing
+  // except recording unique column family IDs
+  class ColumnFamilyCollector : public WriteBatch::Handler {
+    std::unordered_set<uint32_t> column_family_ids_;
+
+    Status AddColumnFamilyId(uint32_t column_family_id) {
+      if (column_family_ids_.find(column_family_id) ==
+          column_family_ids_.end()) {
+        column_family_ids_.insert(column_family_id);
+      }
+      return Status::OK();
+    }
+
+   public:
+    explicit ColumnFamilyCollector() {}
+
+    ~ColumnFamilyCollector() override {}
+
+    Status PutCF(uint32_t column_family_id, const Slice&,
+                 const Slice&) override {
+      return AddColumnFamilyId(column_family_id);
+    }
+
+    Status DeleteCF(uint32_t column_family_id, const Slice&) override {
+      return AddColumnFamilyId(column_family_id);
+    }
+
+    Status SingleDeleteCF(uint32_t column_family_id, const Slice&) override {
+      return AddColumnFamilyId(column_family_id);
+    }
+
+    Status DeleteRangeCF(uint32_t column_family_id, const Slice&,
+                         const Slice&) override {
+      return AddColumnFamilyId(column_family_id);
+    }
+
+    Status MergeCF(uint32_t column_family_id, const Slice&,
+                   const Slice&) override {
+      return AddColumnFamilyId(column_family_id);
+    }
+
+    Status PutBlobIndexCF(uint32_t column_family_id, const Slice&,
+                          const Slice&) override {
+      return AddColumnFamilyId(column_family_id);
+    }
+
+    const std::unordered_set<uint32_t>& column_families() const {
+      return column_family_ids_;
+    }
+  };
+
+  Status CollectColumnFamilyIdsFromWriteBatch(
+      const WriteBatch& batch, std::vector<uint32_t>* column_family_ids) {
+    assert(column_family_ids != nullptr);
+    column_family_ids->clear();
+    ColumnFamilyCollector handler;
+    Status s = batch.Iterate(&handler);
+    if (s.ok()) {
+      for (const auto& cf : handler.column_families()) {
+        column_family_ids->push_back(cf);
+      }
+    }
+    return s;
+  }
+
+ private:
+  friend class DB;
+
+  // No copying allowed
+  DBImplSecondary(const DBImplSecondary&);
+  void operator=(const DBImplSecondary&);
+
+  using DBImpl::Recover;
+
+  Status FindAndRecoverLogFiles(
+      std::unordered_set<ColumnFamilyData*>* cfds_changed,
+      JobContext* job_context);
+  Status FindNewLogNumbers(std::vector<uint64_t>* logs);
+  // After manifest recovery, replay WALs and refresh log_readers_ if necessary
+  // REQUIRES: log_numbers are sorted in ascending order
+  Status RecoverLogFiles(const std::vector<uint64_t>& log_numbers,
+                         SequenceNumber* next_sequence,
+                         std::unordered_set<ColumnFamilyData*>* cfds_changed,
+                         JobContext* job_context);
+
+  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_;
+
+  // Current WAL number replayed for each column family.
+  std::unordered_map<ColumnFamilyData*, uint64_t> cfd_to_current_log_;
+};
+
+}  // namespace rocksdb
+
+#endif  // !ROCKSDB_LITE
@@ -6,17 +6,14 @@
 // Copyright (c) 2011 The LevelDB Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file. See the AUTHORS file for names of contributors.
-#include "db/db_impl.h"
+#include "db/db_impl/db_impl.h"

-#ifndef __STDC_FORMAT_MACROS
-#define __STDC_FORMAT_MACROS
-#endif
-#include <inttypes.h>
+#include <cinttypes>
 #include "db/error_handler.h"
 #include "db/event_helpers.h"
 #include "monitoring/perf_context_imp.h"
 #include "options/options_helper.h"
-#include "util/sync_point.h"
+#include "test_util/sync_point.h"

 namespace rocksdb {
 // Convenience methods
@@ -94,6 +91,11 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
    return Status::NotSupported(
        "pipelined_writes is not compatible with seq_per_batch");
  }
+  if (immutable_db_options_.unordered_write &&
+      immutable_db_options_.enable_pipelined_write) {
+    return Status::NotSupported(
+        "pipelined_writes is not compatible with unordered_write");
+  }
  // Otherwise IsLatestPersistentState optimization does not make sense
  assert(!WriteBatchInternal::IsLatestPersistentState(my_batch) ||
         disable_memtable);
@@ -107,8 +109,39 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
  }

  if (two_write_queues_ && disable_memtable) {
-    return WriteImplWALOnly(write_options, my_batch, callback, log_used,
-                            log_ref, seq_used, batch_cnt, pre_release_callback);
+    AssignOrder assign_order =
+        seq_per_batch_ ? kDoAssignOrder : kDontAssignOrder;
+    // Otherwise it is WAL-only Prepare batches in WriteCommitted policy and
+    // they don't consume sequence.
+    return WriteImplWALOnly(&nonmem_write_thread_, write_options, my_batch,
+                            callback, log_used, log_ref, seq_used, batch_cnt,
+                            pre_release_callback, assign_order,
+                            kDontPublishLastSeq, disable_memtable);
+  }
+
+  if (immutable_db_options_.unordered_write) {
+    const size_t sub_batch_cnt = batch_cnt != 0
+                                     ? batch_cnt
+                                     // every key is a sub-batch consuming a seq
+                                     : WriteBatchInternal::Count(my_batch);
+    uint64_t seq;
+    // Use a write thread to i) optimize for WAL write, ii) publish last
+    // sequence in in increasing order, iii) call pre_release_callback serially
+    status = WriteImplWALOnly(&write_thread_, write_options, my_batch, callback,
+                              log_used, log_ref, &seq, sub_batch_cnt,
+                              pre_release_callback, kDoAssignOrder,
+                              kDoPublishLastSeq, disable_memtable);
+    if (!status.ok()) {
+      return status;
+    }
+    if (seq_used) {
+      *seq_used = seq;
+    }
+    if (!disable_memtable) {
+      status = UnorderedWriteMemtable(write_options, my_batch, callback,
+                                      log_ref, seq, sub_batch_cnt);
+    }
+    return status;
  }

  if (immutable_db_options_.enable_pipelined_write) {
@@ -146,17 +179,6 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,

    if (write_thread_.CompleteParallelMemTableWriter(&w)) {
      // we're responsible for exit batch group
-      for (auto* writer : *(w.write_group)) {
-        if (!writer->CallbackFailed() && writer->pre_release_callback) {
-          assert(writer->sequence != kMaxSequenceNumber);
-          Status ws = writer->pre_release_callback->Callback(writer->sequence,
-                                                             disable_memtable);
-          if (!ws.ok()) {
-            status = ws;
-            break;
-          }
-        }
-      }
      // TODO(myabandeh): propagate status to write_group
      auto last_sequence = w.write_group->last_sequence;
      versions_->SetLastSequence(last_sequence);
@@ -189,9 +211,6 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
  WriteThread::WriteGroup write_group;
  bool in_parallel_group = false;
  uint64_t last_sequence = kMaxSequenceNumber;
-  if (!two_write_queues_) {
-    last_sequence = versions_->LastSequence();
-  }

  mutex_.Lock();

@@ -206,6 +225,11 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
    PERF_TIMER_STOP(write_pre_and_post_process_time);

    status = PreprocessWrite(write_options, &need_log_sync, &write_context);
+    if (!two_write_queues_) {
+      // Assign it after ::PreprocessWrite since the sequence might advance
+      // inside it by WriteRecoverableState
+      last_sequence = versions_->LastSequence();
+    }

    PERF_TIMER_START(write_pre_and_post_process_time);
  }
@@ -239,6 +263,7 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
    size_t total_count = 0;
    size_t valid_batches = 0;
    size_t total_byte_size = 0;
+    size_t pre_release_callback_cnt = 0;
    for (auto* writer : write_group) {
      if (writer->CheckCallback(this)) {
        valid_batches += writer->batch_cnt;
@@ -246,9 +271,11 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
          total_count += WriteBatchInternal::Count(writer->batch);
          parallel = parallel && !writer->batch->HasMerge();
        }
-
        total_byte_size = WriteBatchInternal::AppendedByteSize(
            total_byte_size, WriteBatchInternal::ByteSize(writer->batch));
+        if (writer->pre_release_callback) {
+          pre_release_callback_cnt++;
+        }
      }
    }
    // Note about seq_per_batch_: either disableWAL is set for the entire write
@@ -279,7 +306,7 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
                        concurrent_update);
      RecordTick(stats_, WRITE_DONE_BY_OTHER, write_done_by_other);
    }
-    MeasureTime(stats_, BYTES_PER_WRITE, total_byte_size);
+    RecordInHistogram(stats_, BYTES_PER_WRITE, total_byte_size);

    if (write_options.disableWAL) {
      has_unpersisted_data_.store(true, std::memory_order_relaxed);
@@ -309,6 +336,37 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
    const SequenceNumber current_sequence = last_sequence + 1;
    last_sequence += seq_inc;

+    // PreReleaseCallback is called after WAL write and before memtable write
+    if (status.ok()) {
+      SequenceNumber next_sequence = current_sequence;
+      size_t index = 0;
+      // Note: the logic for advancing seq here must be consistent with the
+      // logic in WriteBatchInternal::InsertInto(write_group...) as well as
+      // with WriteBatchInternal::InsertInto(write_batch...) that is called on
+      // the merged batch during recovery from the WAL.
+      for (auto* writer : write_group) {
+        if (writer->CallbackFailed()) {
+          continue;
+        }
+        writer->sequence = next_sequence;
+        if (writer->pre_release_callback) {
+          Status ws = writer->pre_release_callback->Callback(
+              writer->sequence, disable_memtable, writer->log_used, index++,
+              pre_release_callback_cnt);
+          if (!ws.ok()) {
+            status = ws;
+            break;
+          }
+        }
+        if (seq_per_batch_) {
+          assert(writer->batch_cnt);
+          next_sequence += writer->batch_cnt;
+        } else if (writer->ShouldWriteToMemtable()) {
+          next_sequence += WriteBatchInternal::Count(writer->batch);
+        }
+      }
+    }
+
    if (status.ok()) {
      PERF_TIMER_GUARD(write_memtable_time);

@@ -320,23 +378,6 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
            0 /*recovery_log_number*/, this, parallel, seq_per_batch_,
            batch_per_txn_);
      } else {
-        SequenceNumber next_sequence = current_sequence;
-        // Note: the logic for advancing seq here must be consistent with the
-        // logic in WriteBatchInternal::InsertInto(write_group...) as well as
-        // with WriteBatchInternal::InsertInto(write_batch...) that is called on
-        // the merged batch during recovery from the WAL.
-        for (auto* writer : write_group) {
-          if (writer->CallbackFailed()) {
-            continue;
-          }
-          writer->sequence = next_sequence;
-          if (seq_per_batch_) {
-            assert(writer->batch_cnt);
-            next_sequence += writer->batch_cnt;
-          } else if (writer->ShouldWriteToMemtable()) {
-            next_sequence += WriteBatchInternal::Count(writer->batch);
-          }
-        }
        write_group.last_sequence = last_sequence;
        write_thread_.LaunchParallelMemTableWriters(&write_group);
        in_parallel_group = true;
@@ -388,17 +429,8 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
  }
  if (should_exit_batch_group) {
    if (status.ok()) {
-      for (auto* writer : write_group) {
-        if (!writer->CallbackFailed() && writer->pre_release_callback) {
-          assert(writer->sequence != kMaxSequenceNumber);
-          Status ws = writer->pre_release_callback->Callback(writer->sequence,
-                                                             disable_memtable);
-          if (!ws.ok()) {
-            status = ws;
-            break;
-          }
-        }
-      }
+      // Note: if we are to resume after non-OK statuses we need to revisit how
+      // we reacts to non-OK statuses here.
      versions_->SetLastSequence(last_sequence);
    }
    MemTableInsertStatusCheck(w.status);
@@ -471,7 +503,7 @@ Status DBImpl::PipelinedWriteImpl(const WriteOptions& write_options,
    RecordTick(stats_, NUMBER_KEYS_WRITTEN, total_count);
    stats->AddDBStats(InternalStats::BYTES_WRITTEN, total_byte_size);
    RecordTick(stats_, BYTES_WRITTEN, total_byte_size);
-    MeasureTime(stats_, BYTES_PER_WRITE, total_byte_size);
+    RecordInHistogram(stats_, BYTES_PER_WRITE, total_byte_size);

    PERF_TIMER_STOP(write_pre_and_post_process_time);

@@ -542,23 +574,70 @@ Status DBImpl::PipelinedWriteImpl(const WriteOptions& write_options,
  return w.FinalStatus();
 }

+Status DBImpl::UnorderedWriteMemtable(const WriteOptions& write_options,
+                                      WriteBatch* my_batch,
+                                      WriteCallback* callback, uint64_t log_ref,
+                                      SequenceNumber seq,
+                                      const size_t sub_batch_cnt) {
+  PERF_TIMER_GUARD(write_pre_and_post_process_time);
+  StopWatch write_sw(env_, immutable_db_options_.statistics.get(), DB_WRITE);
+
+  WriteThread::Writer w(write_options, my_batch, callback, log_ref,
+                        false /*disable_memtable*/);
+
+  if (w.CheckCallback(this) && w.ShouldWriteToMemtable()) {
+    w.sequence = seq;
+    size_t total_count = WriteBatchInternal::Count(my_batch);
+    InternalStats* stats = default_cf_internal_stats_;
+    stats->AddDBStats(InternalStats::NUMBER_KEYS_WRITTEN, total_count);
+    RecordTick(stats_, NUMBER_KEYS_WRITTEN, total_count);
+
+    ColumnFamilyMemTablesImpl column_family_memtables(
+        versions_->GetColumnFamilySet());
+    w.status = WriteBatchInternal::InsertInto(
+        &w, w.sequence, &column_family_memtables, &flush_scheduler_,
+        write_options.ignore_missing_column_families, 0 /*log_number*/, this,
+        true /*concurrent_memtable_writes*/, seq_per_batch_, sub_batch_cnt);
+
+    WriteStatusCheck(w.status);
+    if (write_options.disableWAL) {
+      has_unpersisted_data_.store(true, std::memory_order_relaxed);
+    }
+  }
+
+  size_t pending_cnt = pending_memtable_writes_.fetch_sub(1) - 1;
+  if (pending_cnt == 0) {
+    // switch_cv_ waits until pending_memtable_writes_ = 0. Locking its mutex
+    // before notify ensures that cv is in waiting state when it is notified
+    // thus not missing the update to pending_memtable_writes_ even though it is
+    // not modified under the mutex.
+    std::lock_guard<std::mutex> lck(switch_mutex_);
+    switch_cv_.notify_all();
+  }
+
+  if (!w.FinalStatus().ok()) {
+    return w.FinalStatus();
+  }
+  return Status::OK();
+}
+
 // The 2nd write queue. If enabled it will be used only for WAL-only writes.
 // This is the only queue that updates LastPublishedSequence which is only
 // applicable in a two-queue setting.
-Status DBImpl::WriteImplWALOnly(const WriteOptions& write_options,
-                                WriteBatch* my_batch, WriteCallback* callback,
-                                uint64_t* log_used, uint64_t log_ref,
-                                uint64_t* seq_used, size_t batch_cnt,
-                                PreReleaseCallback* pre_release_callback) {
+Status DBImpl::WriteImplWALOnly(
+    WriteThread* write_thread, const WriteOptions& write_options,
+    WriteBatch* my_batch, WriteCallback* callback, uint64_t* log_used,
+    const uint64_t log_ref, uint64_t* seq_used, const size_t sub_batch_cnt,
+    PreReleaseCallback* pre_release_callback, const AssignOrder assign_order,
+    const PublishLastSeq publish_last_seq, const bool disable_memtable) {
  Status status;
  PERF_TIMER_GUARD(write_pre_and_post_process_time);
  WriteThread::Writer w(write_options, my_batch, callback, log_ref,
-                        true /* disable_memtable */, batch_cnt,
-                        pre_release_callback);
+                        disable_memtable, sub_batch_cnt, pre_release_callback);
  RecordTick(stats_, WRITE_WITH_WAL);
  StopWatch write_sw(env_, immutable_db_options_.statistics.get(), DB_WRITE);

-  nonmem_write_thread_.JoinBatchGroup(&w);
+  write_thread->JoinBatchGroup(&w);
  assert(w.state != WriteThread::STATE_PARALLEL_MEMTABLE_WRITER);
  if (w.state == WriteThread::STATE_COMPLETED) {
    if (log_used != nullptr) {
@@ -571,17 +650,45 @@ Status DBImpl::WriteImplWALOnly(const WriteOptions& write_options,
  }
  // else we are the leader of the write batch group
  assert(w.state == WriteThread::STATE_GROUP_LEADER);
+
+  if (publish_last_seq == kDoPublishLastSeq) {
+    // Currently we only use kDoPublishLastSeq in unordered_write
+    assert(immutable_db_options_.unordered_write);
+    WriteContext write_context;
+    if (error_handler_.IsDBStopped()) {
+      status = error_handler_.GetBGError();
+    }
+    // TODO(myabandeh): Make preliminary checks thread-safe so we could do them
+    // without paying the cost of obtaining the mutex.
+    if (status.ok()) {
+      InstrumentedMutexLock l(&mutex_);
+      bool need_log_sync = false;
+      status = PreprocessWrite(write_options, &need_log_sync, &write_context);
+      WriteStatusCheck(status);
+    }
+    if (!status.ok()) {
+      WriteThread::WriteGroup write_group;
+      write_thread->EnterAsBatchGroupLeader(&w, &write_group);
+      write_thread->ExitAsBatchGroupLeader(write_group, status);
+      return status;
+    }
+  }
+
  WriteThread::WriteGroup write_group;
  uint64_t last_sequence;
-  nonmem_write_thread_.EnterAsBatchGroupLeader(&w, &write_group);
+  write_thread->EnterAsBatchGroupLeader(&w, &write_group);
  // Note: no need to update last_batch_group_size_ here since the batch writes
  // to WAL only

+  size_t pre_release_callback_cnt = 0;
  size_t total_byte_size = 0;
  for (auto* writer : write_group) {
    if (writer->CheckCallback(this)) {
      total_byte_size = WriteBatchInternal::AppendedByteSize(
          total_byte_size, WriteBatchInternal::ByteSize(writer->batch));
+      if (writer->pre_release_callback) {
+        pre_release_callback_cnt++;
+      }
    }
  }

@@ -602,7 +709,7 @@ Status DBImpl::WriteImplWALOnly(const WriteOptions& write_options,
                      concurrent_update);
    RecordTick(stats_, WRITE_DONE_BY_OTHER, write_done_by_other);
  }
-  MeasureTime(stats_, BYTES_PER_WRITE, total_byte_size);
+  RecordInHistogram(stats_, BYTES_PER_WRITE, total_byte_size);

  PERF_TIMER_STOP(write_pre_and_post_process_time);

@@ -610,11 +717,13 @@ Status DBImpl::WriteImplWALOnly(const WriteOptions& write_options,
  // LastAllocatedSequence is increased inside WriteToWAL under
  // wal_write_mutex_ to ensure ordered events in WAL
  size_t seq_inc = 0 /* total_count */;
-  if (seq_per_batch_) {
+  if (assign_order == kDoAssignOrder) {
    size_t total_batch_cnt = 0;
    for (auto* writer : write_group) {
-      assert(writer->batch_cnt);
-      total_batch_cnt += writer->batch_cnt;
+      assert(writer->batch_cnt || !seq_per_batch_);
+      if (!writer->CallbackFailed()) {
+        total_batch_cnt += writer->batch_cnt;
+      }
    }
    seq_inc = total_batch_cnt;
  }
@@ -625,16 +734,21 @@ Status DBImpl::WriteImplWALOnly(const WriteOptions& write_options,
    // Otherwise we inc seq number to do solely the seq allocation
    last_sequence = versions_->FetchAddLastAllocatedSequence(seq_inc);
  }
+
+  size_t memtable_write_cnt = 0;
  auto curr_seq = last_sequence + 1;
  for (auto* writer : write_group) {
    if (writer->CallbackFailed()) {
      continue;
    }
    writer->sequence = curr_seq;
-    if (seq_per_batch_) {
-      assert(writer->batch_cnt);
+    if (assign_order == kDoAssignOrder) {
+      assert(writer->batch_cnt || !seq_per_batch_);
      curr_seq += writer->batch_cnt;
    }
+    if (!writer->disable_memtable) {
+      memtable_write_cnt++;
+    }
    // else seq advances only by memtable writes
  }
  if (status.ok() && write_options.sync) {
@@ -653,12 +767,13 @@ Status DBImpl::WriteImplWALOnly(const WriteOptions& write_options,
    WriteStatusCheck(status);
  }
  if (status.ok()) {
+    size_t index = 0;
    for (auto* writer : write_group) {
      if (!writer->CallbackFailed() && writer->pre_release_callback) {
        assert(writer->sequence != kMaxSequenceNumber);
-        const bool DISABLE_MEMTABLE = true;
-        Status ws = writer->pre_release_callback->Callback(writer->sequence,
-                                                           DISABLE_MEMTABLE);
+        Status ws = writer->pre_release_callback->Callback(
+            writer->sequence, disable_memtable, writer->log_used, index++,
+            pre_release_callback_cnt);
        if (!ws.ok()) {
          status = ws;
          break;
@@ -666,7 +781,15 @@ Status DBImpl::WriteImplWALOnly(const WriteOptions& write_options,
      }
    }
  }
-  nonmem_write_thread_.ExitAsBatchGroupLeader(write_group, status);
+  if (publish_last_seq == kDoPublishLastSeq) {
+    versions_->SetLastSequence(last_sequence + seq_inc);
+    // Currently we only use kDoPublishLastSeq in unordered_write
+    assert(immutable_db_options_.unordered_write);
+  }
+  if (immutable_db_options_.unordered_write && status.ok()) {
+    pending_memtable_writes_ += memtable_write_cnt;
+  }
+  write_thread->ExitAsBatchGroupLeader(write_group, status);
  if (status.ok()) {
    status = w.FinalStatus();
  }
@@ -718,6 +841,7 @@ Status DBImpl::PreprocessWrite(const WriteOptions& write_options,
         versions_->GetColumnFamilySet()->NumberOfColumnFamilies() == 1);
  if (UNLIKELY(status.ok() && !single_column_family_mode_ &&
               total_log_size_ > GetMaxTotalWalSize())) {
+    WaitForPendingWrites();
    status = SwitchWAL(write_context);
  }

@@ -727,10 +851,12 @@ Status DBImpl::PreprocessWrite(const WriteOptions& write_options,
    // thread is writing to another DB with the same write buffer, they may also
    // be flushed. We may end up with flushing much more DBs than needed. It's
    // suboptimal but still correct.
+    WaitForPendingWrites();
    status = HandleWriteBufferFull(write_context);
  }

  if (UNLIKELY(status.ok() && !flush_scheduler_.Empty())) {
+    WaitForPendingWrites();
    status = ScheduleFlushes(write_context);
  }

@@ -876,7 +1002,7 @@ Status DBImpl::WriteToWAL(const WriteThread::WriteGroup& write_group,
  status = WriteToWAL(*merged_batch, log_writer, log_used, &log_size);
  if (to_be_cached_state) {
    cached_recoverable_state_ = *to_be_cached_state;
-      cached_recoverable_state_empty_ = false;
+    cached_recoverable_state_empty_ = false;
  }

  if (status.ok() && need_log_sync) {
@@ -952,7 +1078,7 @@ Status DBImpl::ConcurrentWriteToWAL(const WriteThread::WriteGroup& write_group,
  status = WriteToWAL(*merged_batch, log_writer, log_used, &log_size);
  if (to_be_cached_state) {
    cached_recoverable_state_ = *to_be_cached_state;
-      cached_recoverable_state_empty_ = false;
+    cached_recoverable_state_empty_ = false;
  }
  log_write_mutex_.Unlock();

@@ -1001,8 +1127,13 @@ Status DBImpl::WriteRecoverableState() {
      const bool DISABLE_MEMTABLE = true;
      for (uint64_t sub_batch_seq = seq + 1;
           sub_batch_seq < next_seq && status.ok(); sub_batch_seq++) {
+        uint64_t const no_log_num = 0;
+        // Unlock it since the callback might end up locking mutex. e.g.,
+        // AddCommitted -> AdvanceMaxEvictedSeq -> GetSnapshotListFromDB
+        mutex_.Unlock();
        status = recoverable_state_pre_release_callback_->Callback(
-            sub_batch_seq, !DISABLE_MEMTABLE);
+            sub_batch_seq, !DISABLE_MEMTABLE, no_log_num, 0, 1);
+        mutex_.Lock();
      }
    }
    if (status.ok()) {
@@ -1072,16 +1203,17 @@ Status DBImpl::SwitchWAL(WriteContext* write_context) {
  if (!flush_wont_release_oldest_log) {
    // we only mark this log as getting flushed if we have successfully
    // flushed all data in this log. If this log contains outstanding prepared
-    // transactions then we cannot flush this log until those transactions are commited.
+    // transactions then we cannot flush this log until those transactions are
+    // commited.
    unable_to_release_oldest_log_ = false;
    alive_log_files_.begin()->getting_flushed = true;
  }

-  ROCKS_LOG_INFO(immutable_db_options_.info_log,
-                 "Flushing all column families with data in WAL number %" PRIu64
-                 ". Total log size is %" PRIu64
-                 " while max_total_wal_size is %" PRIu64,
-                 oldest_alive_log, total_log_size_.load(), GetMaxTotalWalSize());
+  ROCKS_LOG_INFO(
+      immutable_db_options_.info_log,
+      "Flushing all column families with data in WAL number %" PRIu64
+      ". Total log size is %" PRIu64 " while max_total_wal_size is %" PRIu64,
+      oldest_alive_log, total_log_size_.load(), GetMaxTotalWalSize());
  // no need to refcount because drop is happening in write thread, so can't
  // happen while we're in the write thread
  autovector<ColumnFamilyData*> cfds;
@@ -1096,6 +1228,7 @@ Status DBImpl::SwitchWAL(WriteContext* write_context) {
        cfds.push_back(cfd);
      }
    }
+    MaybeFlushStatsCF(&cfds);
  }
  for (const auto cfd : cfds) {
    cfd->Ref();
@@ -1133,7 +1266,7 @@ Status DBImpl::HandleWriteBufferFull(WriteContext* write_context) {
  ROCKS_LOG_INFO(
      immutable_db_options_.info_log,
      "Flushing column family with largest mem table size. Write buffer is "
-      "using %" PRIu64 " bytes out of a total of %" PRIu64 ".",
+      "using %" ROCKSDB_PRIszt " bytes out of a total of %" ROCKSDB_PRIszt ".",
      write_buffer_manager_->memory_usage(),
      write_buffer_manager_->buffer_size());
  // no need to refcount because drop is happening in write thread, so can't
@@ -1162,6 +1295,7 @@ Status DBImpl::HandleWriteBufferFull(WriteContext* write_context) {
    if (cfd_picked != nullptr) {
      cfds.push_back(cfd_picked);
    }
+    MaybeFlushStatsCF(&cfds);
  }

  for (const auto cfd : cfds) {
@@ -1305,6 +1439,40 @@ Status DBImpl::ThrottleLowPriWritesIfNeeded(const WriteOptions& write_options,
  return Status::OK();
 }

+void DBImpl::MaybeFlushStatsCF(autovector<ColumnFamilyData*>* cfds) {
+  assert(cfds != nullptr);
+  if (!cfds->empty() && immutable_db_options_.persist_stats_to_disk) {
+    ColumnFamilyData* cfd_stats =
+        versions_->GetColumnFamilySet()->GetColumnFamily(
+            kPersistentStatsColumnFamilyName);
+    if (cfd_stats != nullptr && !cfd_stats->mem()->IsEmpty()) {
+      for (ColumnFamilyData* cfd : *cfds) {
+        if (cfd == cfd_stats) {
+          // stats CF already included in cfds
+          return;
+        }
+      }
+      // force flush stats CF when its log number is less than all other CF's
+      // log numbers
+      bool force_flush_stats_cf = true;
+      for (auto* loop_cfd : *versions_->GetColumnFamilySet()) {
+        if (loop_cfd == cfd_stats) {
+          continue;
+        }
+        if (loop_cfd->GetLogNumber() <= cfd_stats->GetLogNumber()) {
+          force_flush_stats_cf = false;
+        }
+      }
+      if (force_flush_stats_cf) {
+        cfds->push_back(cfd_stats);
+        ROCKS_LOG_INFO(immutable_db_options_.info_log,
+                       "Force flushing stats CF with automated flush "
+                       "to avoid holding old logs");
+      }
+    }
+  }
+}
+
 Status DBImpl::ScheduleFlushes(WriteContext* context) {
  autovector<ColumnFamilyData*> cfds;
  if (immutable_db_options_.atomic_flush) {
@@ -1318,6 +1486,7 @@ Status DBImpl::ScheduleFlushes(WriteContext* context) {
    while ((tmp_cfd = flush_scheduler_.TakeNextColumnFamily()) != nullptr) {
      cfds.push_back(tmp_cfd);
    }
+    MaybeFlushStatsCF(&cfds);
  }
  Status status;
  for (auto& cfd : cfds) {
@@ -1424,58 +1593,19 @@ Status DBImpl::SwitchMemtable(ColumnFamilyData* cfd, WriteContext* context) {
  // Log this later after lock release. It may be outdated, e.g., if background
  // flush happens before logging, but that should be ok.
  int num_imm_unflushed = cfd->imm()->NumNotFlushed();
-  DBOptions db_options =
-      BuildDBOptions(immutable_db_options_, mutable_db_options_);
  const auto preallocate_block_size =
-    GetWalPreallocateBlockSize(mutable_cf_options.write_buffer_size);
-  auto write_hint = CalculateWALWriteHint();
+      GetWalPreallocateBlockSize(mutable_cf_options.write_buffer_size);
  mutex_.Unlock();
-  {
-    std::string log_fname =
-        LogFileName(immutable_db_options_.wal_dir, new_log_number);
-    if (creating_new_log) {
-      EnvOptions opt_env_opt =
-          env_->OptimizeForLogWrite(env_options_, db_options);
-      if (recycle_log_number) {
-        ROCKS_LOG_INFO(immutable_db_options_.info_log,
-                       "reusing log %" PRIu64 " from recycle list\n",
-                       recycle_log_number);
-        std::string old_log_fname =
-            LogFileName(immutable_db_options_.wal_dir, recycle_log_number);
-        s = env_->ReuseWritableFile(log_fname, old_log_fname, &lfile,
-                                    opt_env_opt);
-      } else {
-        s = NewWritableFile(env_, log_fname, &lfile, opt_env_opt);
-      }
-      if (s.ok()) {
-        // Our final size should be less than write_buffer_size
-        // (compression, etc) but err on the side of caution.
-
-        // use preallocate_block_size instead
-        // of calling GetWalPreallocateBlockSize()
-        lfile->SetPreallocationBlockSize(preallocate_block_size);
-        lfile->SetWriteLifeTimeHint(write_hint);
-        std::unique_ptr<WritableFileWriter> file_writer(new WritableFileWriter(
-            std::move(lfile), log_fname, opt_env_opt, env_, nullptr /* stats */,
-            immutable_db_options_.listeners));
-        new_log = new log::Writer(
-            std::move(file_writer), new_log_number,
-            immutable_db_options_.recycle_log_file_num > 0, manual_wal_flush_);
-      }
-    }
-
-    if (s.ok()) {
-      SequenceNumber seq = versions_->LastSequence();
-      new_mem = cfd->ConstructNewMemtable(mutable_cf_options, seq);
-      context->superversion_context.NewSuperVersion();
-    }
-
-#ifndef ROCKSDB_LITE
-    // PLEASE NOTE: We assume that there are no failable operations
-    // after lock is acquired below since we are already notifying
-    // client about mem table becoming immutable.
-    NotifyOnMemTableSealed(cfd, memtable_info);
-#endif //ROCKSDB_LITE
+  if (creating_new_log) {
+    // TODO: Write buffer size passed in should be max of all CF's instead
+    // of mutable_cf_options.write_buffer_size.
+    s = CreateWAL(new_log_number, recycle_log_number, preallocate_block_size,
+                  &new_log);
+  }
+  if (s.ok()) {
+    SequenceNumber seq = versions_->LastSequence();
+    new_mem = cfd->ConstructNewMemtable(mutable_cf_options, seq);
+    context->superversion_context.NewSuperVersion();
  }
  ROCKS_LOG_INFO(immutable_db_options_.info_log,
                 "[%s] New memtable created with log file: #%" PRIu64
@@ -1484,10 +1614,7 @@ Status DBImpl::SwitchMemtable(ColumnFamilyData* cfd, WriteContext* context) {
  mutex_.Lock();
  if (s.ok() && creating_new_log) {
    log_write_mutex_.Lock();
-    logfile_number_ = new_log_number;
    assert(new_log != nullptr);
-    log_empty_ = true;
-    log_dir_synced_ = false;
    if (!logs_.empty()) {
      // Alway flush the buffer of the last log before switching to a new one
      log::Writer* cur_log_writer = logs_.back().writer;
@@ -1495,21 +1622,41 @@ Status DBImpl::SwitchMemtable(ColumnFamilyData* cfd, WriteContext* context) {
      if (!s.ok()) {
        ROCKS_LOG_WARN(immutable_db_options_.info_log,
                       "[%s] Failed to switch from #%" PRIu64 " to #%" PRIu64
-                       "  WAL file -- %s\n",
+                       "  WAL file\n",
                       cfd->GetName().c_str(), cur_log_writer->get_log_number(),
                       new_log_number);
      }
    }
-    logs_.emplace_back(logfile_number_, new_log);
-    alive_log_files_.push_back(LogFileNumberSize(logfile_number_));
+    if (s.ok()) {
+      logfile_number_ = new_log_number;
+      log_empty_ = true;
+      log_dir_synced_ = false;
+      logs_.emplace_back(logfile_number_, new_log);
+      alive_log_files_.push_back(LogFileNumberSize(logfile_number_));
+    }
    log_write_mutex_.Unlock();
  }

  if (!s.ok()) {
    // how do we fail if we're not creating new log?
    assert(creating_new_log);
-    assert(!new_mem);
-    assert(!new_log);
+    if (new_mem) {
+      delete new_mem;
+    }
+    if (new_log) {
+      delete new_log;
+    }
+    SuperVersion* new_superversion =
+        context->superversion_context.new_superversion.release();
+    if (new_superversion != nullptr) {
+      delete new_superversion;
+    }
+    // We may have lost data from the WritableFileBuffer in-memory buffer for
+    // the current log, so treat it as a fatal error and set bg_error
+    error_handler_.SetBGError(s, BackgroundErrorReason::kMemTable);
+    // Read back bg_error in order to get the right severity
+    s = error_handler_.GetBGError();
+
    if (two_write_queues_) {
      nonmem_write_thread_.ExitUnbatched(&nonmem_w);
    }
@@ -1536,6 +1683,13 @@ Status DBImpl::SwitchMemtable(ColumnFamilyData* cfd, WriteContext* context) {
  cfd->SetMemtable(new_mem);
  InstallSuperVersionAndScheduleWork(cfd, &context->superversion_context,
                                     mutable_cf_options);
+#ifndef ROCKSDB_LITE
+  mutex_.Unlock();
+  // Notify client that memtable is sealed, now that we have successfully
+  // installed a new memtable
+  NotifyOnMemTableSealed(cfd, memtable_info);
+  mutex_.Lock();
+#endif  // ROCKSDB_LITE
  if (two_write_queues_) {
    nonmem_write_thread_.ExitUnbatched(&nonmem_w);
  }
@@ -1544,13 +1698,13 @@ Status DBImpl::SwitchMemtable(ColumnFamilyData* cfd, WriteContext* context) {

 size_t DBImpl::GetWalPreallocateBlockSize(uint64_t write_buffer_size) const {
  mutex_.AssertHeld();
-  size_t bsize = static_cast<size_t>(
-    write_buffer_size / 10 + write_buffer_size);
+  size_t bsize =
+      static_cast<size_t>(write_buffer_size / 10 + write_buffer_size);
  // Some users might set very high write_buffer_size and rely on
  // max_total_wal_size or other parameters to control the WAL size.
  if (mutable_db_options_.max_total_wal_size > 0) {
-    bsize = std::min<size_t>(bsize, static_cast<size_t>(
-      mutable_db_options_.max_total_wal_size));
+    bsize = std::min<size_t>(
+        bsize, static_cast<size_t>(mutable_db_options_.max_total_wal_size));
  }
  if (immutable_db_options_.db_write_buffer_size > 0) {
    bsize = std::min<size_t>(bsize, immutable_db_options_.db_write_buffer_size);
@@ -1568,11 +1722,25 @@ size_t DBImpl::GetWalPreallocateBlockSize(uint64_t write_buffer_size) const {
 // can call if they wish
 Status DB::Put(const WriteOptions& opt, ColumnFamilyHandle* column_family,
               const Slice& key, const Slice& value) {
-  // Pre-allocate size of write batch conservatively.
-  // 8 bytes are taken by header, 4 bytes for count, 1 byte for type,
-  // and we allocate 11 extra bytes for key length, as well as value length.
-  WriteBatch batch(key.size() + value.size() + 24);
-  Status s = batch.Put(column_family, key, value);
+  if (nullptr == opt.timestamp) {
+    // Pre-allocate size of write batch conservatively.
+    // 8 bytes are taken by header, 4 bytes for count, 1 byte for type,
+    // and we allocate 11 extra bytes for key length, as well as value length.
+    WriteBatch batch(key.size() + value.size() + 24);
+    Status s = batch.Put(column_family, key, value);
+    if (!s.ok()) {
+      return s;
+    }
+    return Write(opt, &batch);
+  }
+  Slice akey;
+  std::string buf;
+  Status s = AppendTimestamp(key, *(opt.timestamp), &akey, &buf);
+  if (!s.ok()) {
+    return s;
+  }
+  WriteBatch batch(akey.size() + value.size() + 24);
+  s = batch.Put(column_family, akey, value);
  if (!s.ok()) {
    return s;
  }
@@ -0,0 +1,784 @@
+//  Copyright (c) 2011-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).
+//
+// Copyright (c) 2011 The LevelDB Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file. See the AUTHORS file for names of contributors.
+
+#include "db/db_impl/db_impl_secondary.h"
+#include "db/db_test_util.h"
+#include "port/stack_trace.h"
+#include "test_util/fault_injection_test_env.h"
+#include "test_util/sync_point.h"
+
+namespace rocksdb {
+
+#ifndef ROCKSDB_LITE
+class DBSecondaryTest : public DBTestBase {
+ public:
+  DBSecondaryTest()
+      : DBTestBase("/db_secondary_test"),
+        secondary_path_(),
+        handles_secondary_(),
+        db_secondary_(nullptr) {
+    secondary_path_ =
+        test::PerThreadDBPath(env_, "/db_secondary_test_secondary");
+  }
+
+  ~DBSecondaryTest() override {
+    CloseSecondary();
+    if (getenv("KEEP_DB") != nullptr) {
+      fprintf(stdout, "Secondary DB is still at %s\n", secondary_path_.c_str());
+    } else {
+      Options options;
+      options.env = env_;
+      EXPECT_OK(DestroyDB(secondary_path_, options));
+    }
+  }
+
+ protected:
+  Status ReopenAsSecondary(const Options& options) {
+    return DB::OpenAsSecondary(options, dbname_, secondary_path_, &db_);
+  }
+
+  void OpenSecondary(const Options& options);
+
+  void OpenSecondaryWithColumnFamilies(
+      const std::vector<std::string>& column_families, const Options& options);
+
+  void CloseSecondary() {
+    for (auto h : handles_secondary_) {
+      db_secondary_->DestroyColumnFamilyHandle(h);
+    }
+    handles_secondary_.clear();
+    delete db_secondary_;
+    db_secondary_ = nullptr;
+  }
+
+  DBImplSecondary* db_secondary_full() {
+    return static_cast<DBImplSecondary*>(db_secondary_);
+  }
+
+  void CheckFileTypeCounts(const std::string& dir, int expected_log,
+                           int expected_sst, int expected_manifest) const;
+
+  std::string secondary_path_;
+  std::vector<ColumnFamilyHandle*> handles_secondary_;
+  DB* db_secondary_;
+};
+
+void DBSecondaryTest::OpenSecondary(const Options& options) {
+  Status s =
+      DB::OpenAsSecondary(options, dbname_, secondary_path_, &db_secondary_);
+  ASSERT_OK(s);
+}
+
+void DBSecondaryTest::OpenSecondaryWithColumnFamilies(
+    const std::vector<std::string>& column_families, const Options& options) {
+  std::vector<ColumnFamilyDescriptor> cf_descs;
+  cf_descs.emplace_back(kDefaultColumnFamilyName, options);
+  for (const auto& cf_name : column_families) {
+    cf_descs.emplace_back(cf_name, options);
+  }
+  Status s = DB::OpenAsSecondary(options, dbname_, secondary_path_, cf_descs,
+                                 &handles_secondary_, &db_secondary_);
+  ASSERT_OK(s);
+}
+
+void DBSecondaryTest::CheckFileTypeCounts(const std::string& dir,
+                                          int expected_log, int expected_sst,
+                                          int expected_manifest) const {
+  std::vector<std::string> filenames;
+  env_->GetChildren(dir, &filenames);
+
+  int log_cnt = 0, sst_cnt = 0, manifest_cnt = 0;
+  for (auto file : filenames) {
+    uint64_t number;
+    FileType type;
+    if (ParseFileName(file, &number, &type)) {
+      log_cnt += (type == kLogFile);
+      sst_cnt += (type == kTableFile);
+      manifest_cnt += (type == kDescriptorFile);
+    }
+  }
+  ASSERT_EQ(expected_log, log_cnt);
+  ASSERT_EQ(expected_sst, sst_cnt);
+  ASSERT_EQ(expected_manifest, manifest_cnt);
+}
+
+TEST_F(DBSecondaryTest, ReopenAsSecondary) {
+  Options options;
+  options.env = env_;
+  Reopen(options);
+  ASSERT_OK(Put("foo", "foo_value"));
+  ASSERT_OK(Put("bar", "bar_value"));
+  ASSERT_OK(dbfull()->Flush(FlushOptions()));
+  Close();
+
+  ASSERT_OK(ReopenAsSecondary(options));
+  ASSERT_EQ("foo_value", Get("foo"));
+  ASSERT_EQ("bar_value", Get("bar"));
+  ReadOptions ropts;
+  ropts.verify_checksums = true;
+  auto db1 = static_cast<DBImplSecondary*>(db_);
+  ASSERT_NE(nullptr, db1);
+  Iterator* iter = db1->NewIterator(ropts);
+  ASSERT_NE(nullptr, iter);
+  size_t count = 0;
+  for (iter->SeekToFirst(); iter->Valid(); iter->Next()) {
+    if (0 == count) {
+      ASSERT_EQ("bar", iter->key().ToString());
+      ASSERT_EQ("bar_value", iter->value().ToString());
+    } else if (1 == count) {
+      ASSERT_EQ("foo", iter->key().ToString());
+      ASSERT_EQ("foo_value", iter->value().ToString());
+    }
+    ++count;
+  }
+  delete iter;
+  ASSERT_EQ(2, count);
+}
+
+TEST_F(DBSecondaryTest, OpenAsSecondary) {
+  Options options;
+  options.env = env_;
+  options.level0_file_num_compaction_trigger = 4;
+  Reopen(options);
+  for (int i = 0; i < 3; ++i) {
+    ASSERT_OK(Put("foo", "foo_value" + std::to_string(i)));
+    ASSERT_OK(Put("bar", "bar_value" + std::to_string(i)));
+    ASSERT_OK(Flush());
+  }
+  Options options1;
+  options1.env = env_;
+  options1.max_open_files = -1;
+  OpenSecondary(options1);
+  ASSERT_OK(dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr));
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
+
+  ReadOptions ropts;
+  ropts.verify_checksums = true;
+  const auto verify_db_func = [&](const std::string& foo_val,
+                                  const std::string& bar_val) {
+    std::string value;
+    ASSERT_OK(db_secondary_->Get(ropts, "foo", &value));
+    ASSERT_EQ(foo_val, value);
+    ASSERT_OK(db_secondary_->Get(ropts, "bar", &value));
+    ASSERT_EQ(bar_val, value);
+    Iterator* iter = db_secondary_->NewIterator(ropts);
+    ASSERT_NE(nullptr, iter);
+    iter->Seek("foo");
+    ASSERT_TRUE(iter->Valid());
+    ASSERT_EQ("foo", iter->key().ToString());
+    ASSERT_EQ(foo_val, iter->value().ToString());
+    iter->Seek("bar");
+    ASSERT_TRUE(iter->Valid());
+    ASSERT_EQ("bar", iter->key().ToString());
+    ASSERT_EQ(bar_val, iter->value().ToString());
+    size_t count = 0;
+    for (iter->SeekToFirst(); iter->Valid(); iter->Next()) {
+      ++count;
+    }
+    ASSERT_EQ(2, count);
+    delete iter;
+  };
+
+  verify_db_func("foo_value2", "bar_value2");
+
+  ASSERT_OK(Put("foo", "new_foo_value"));
+  ASSERT_OK(Put("bar", "new_bar_value"));
+  ASSERT_OK(Flush());
+
+  ASSERT_OK(db_secondary_->TryCatchUpWithPrimary());
+  verify_db_func("new_foo_value", "new_bar_value");
+}
+
+TEST_F(DBSecondaryTest, OpenAsSecondaryWALTailing) {
+  Options options;
+  options.env = env_;
+  options.level0_file_num_compaction_trigger = 4;
+  Reopen(options);
+  for (int i = 0; i < 3; ++i) {
+    ASSERT_OK(Put("foo", "foo_value" + std::to_string(i)));
+    ASSERT_OK(Put("bar", "bar_value" + std::to_string(i)));
+  }
+  Options options1;
+  options1.env = env_;
+  options1.max_open_files = -1;
+  OpenSecondary(options1);
+
+  ReadOptions ropts;
+  ropts.verify_checksums = true;
+  const auto verify_db_func = [&](const std::string& foo_val,
+                                  const std::string& bar_val) {
+    std::string value;
+    ASSERT_OK(db_secondary_->Get(ropts, "foo", &value));
+    ASSERT_EQ(foo_val, value);
+    ASSERT_OK(db_secondary_->Get(ropts, "bar", &value));
+    ASSERT_EQ(bar_val, value);
+    Iterator* iter = db_secondary_->NewIterator(ropts);
+    ASSERT_NE(nullptr, iter);
+    iter->Seek("foo");
+    ASSERT_TRUE(iter->Valid());
+    ASSERT_EQ("foo", iter->key().ToString());
+    ASSERT_EQ(foo_val, iter->value().ToString());
+    iter->Seek("bar");
+    ASSERT_TRUE(iter->Valid());
+    ASSERT_EQ("bar", iter->key().ToString());
+    ASSERT_EQ(bar_val, iter->value().ToString());
+    size_t count = 0;
+    for (iter->SeekToFirst(); iter->Valid(); iter->Next()) {
+      ++count;
+    }
+    ASSERT_EQ(2, count);
+    delete iter;
+  };
+
+  verify_db_func("foo_value2", "bar_value2");
+
+  ASSERT_OK(Put("foo", "new_foo_value"));
+  ASSERT_OK(Put("bar", "new_bar_value"));
+
+  ASSERT_OK(db_secondary_->TryCatchUpWithPrimary());
+  verify_db_func("new_foo_value", "new_bar_value");
+
+  ASSERT_OK(Flush());
+  ASSERT_OK(Put("foo", "new_foo_value_1"));
+  ASSERT_OK(db_secondary_->TryCatchUpWithPrimary());
+  verify_db_func("new_foo_value_1", "new_bar_value");
+}
+
+TEST_F(DBSecondaryTest, OpenWithNonExistColumnFamily) {
+  Options options;
+  options.env = env_;
+  CreateAndReopenWithCF({"pikachu"}, options);
+
+  Options options1;
+  options1.env = env_;
+  options1.max_open_files = -1;
+  std::vector<ColumnFamilyDescriptor> cf_descs;
+  cf_descs.emplace_back(kDefaultColumnFamilyName, options1);
+  cf_descs.emplace_back("pikachu", options1);
+  cf_descs.emplace_back("eevee", options1);
+  Status s = DB::OpenAsSecondary(options1, dbname_, secondary_path_, cf_descs,
+                                 &handles_secondary_, &db_secondary_);
+  ASSERT_NOK(s);
+}
+
+TEST_F(DBSecondaryTest, OpenWithSubsetOfColumnFamilies) {
+  Options options;
+  options.env = env_;
+  CreateAndReopenWithCF({"pikachu"}, options);
+  Options options1;
+  options1.env = env_;
+  options1.max_open_files = -1;
+  OpenSecondary(options1);
+  ASSERT_EQ(0, handles_secondary_.size());
+  ASSERT_NE(nullptr, db_secondary_);
+
+  ASSERT_OK(Put(0 /*cf*/, "foo", "foo_value"));
+  ASSERT_OK(Put(1 /*cf*/, "foo", "foo_value"));
+  ASSERT_OK(Flush(0 /*cf*/));
+  ASSERT_OK(Flush(1 /*cf*/));
+  ASSERT_OK(db_secondary_->TryCatchUpWithPrimary());
+  ReadOptions ropts;
+  ropts.verify_checksums = true;
+  std::string value;
+  ASSERT_OK(db_secondary_->Get(ropts, "foo", &value));
+  ASSERT_EQ("foo_value", value);
+}
+
+TEST_F(DBSecondaryTest, SwitchToNewManifestDuringOpen) {
+  Options options;
+  options.env = env_;
+  Reopen(options);
+  Close();
+
+  SyncPoint::GetInstance()->DisableProcessing();
+  SyncPoint::GetInstance()->ClearAllCallBacks();
+  SyncPoint::GetInstance()->LoadDependency(
+      {{"ReactiveVersionSet::MaybeSwitchManifest:AfterGetCurrentManifestPath:0",
+        "VersionSet::ProcessManifestWrites:BeforeNewManifest"},
+       {"VersionSet::ProcessManifestWrites:AfterNewManifest",
+        "ReactiveVersionSet::MaybeSwitchManifest:AfterGetCurrentManifestPath:"
+        "1"}});
+  SyncPoint::GetInstance()->EnableProcessing();
+
+  // Make sure db calls RecoverLogFiles so as to trigger a manifest write,
+  // which causes the db to switch to a new MANIFEST upon start.
+  port::Thread ro_db_thread([&]() {
+    Options options1;
+    options1.env = env_;
+    options1.max_open_files = -1;
+    OpenSecondary(options1);
+    CloseSecondary();
+  });
+  Reopen(options);
+  ro_db_thread.join();
+}
+
+TEST_F(DBSecondaryTest, MissingTableFileDuringOpen) {
+  Options options;
+  options.env = env_;
+  options.level0_file_num_compaction_trigger = 4;
+  Reopen(options);
+  for (int i = 0; i != options.level0_file_num_compaction_trigger; ++i) {
+    ASSERT_OK(Put("foo", "foo_value" + std::to_string(i)));
+    ASSERT_OK(Put("bar", "bar_value" + std::to_string(i)));
+    ASSERT_OK(dbfull()->Flush(FlushOptions()));
+  }
+  ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable());
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
+  Options options1;
+  options1.env = env_;
+  options1.max_open_files = -1;
+  OpenSecondary(options1);
+  ReadOptions ropts;
+  ropts.verify_checksums = true;
+  std::string value;
+  ASSERT_OK(db_secondary_->Get(ropts, "foo", &value));
+  ASSERT_EQ("foo_value" +
+                std::to_string(options.level0_file_num_compaction_trigger - 1),
+            value);
+  ASSERT_OK(db_secondary_->Get(ropts, "bar", &value));
+  ASSERT_EQ("bar_value" +
+                std::to_string(options.level0_file_num_compaction_trigger - 1),
+            value);
+  Iterator* iter = db_secondary_->NewIterator(ropts);
+  ASSERT_NE(nullptr, iter);
+  iter->Seek("bar");
+  ASSERT_TRUE(iter->Valid());
+  ASSERT_EQ("bar", iter->key().ToString());
+  ASSERT_EQ("bar_value" +
+                std::to_string(options.level0_file_num_compaction_trigger - 1),
+            iter->value().ToString());
+  iter->Seek("foo");
+  ASSERT_TRUE(iter->Valid());
+  ASSERT_EQ("foo", iter->key().ToString());
+  ASSERT_EQ("foo_value" +
+                std::to_string(options.level0_file_num_compaction_trigger - 1),
+            iter->value().ToString());
+  size_t count = 0;
+  for (iter->SeekToFirst(); iter->Valid(); iter->Next()) {
+    ++count;
+  }
+  ASSERT_EQ(2, count);
+  delete iter;
+}
+
+TEST_F(DBSecondaryTest, MissingTableFile) {
+  int table_files_not_exist = 0;
+  SyncPoint::GetInstance()->DisableProcessing();
+  SyncPoint::GetInstance()->ClearAllCallBacks();
+  SyncPoint::GetInstance()->SetCallBack(
+      "ReactiveVersionSet::ApplyOneVersionEditToBuilder:AfterLoadTableHandlers",
+      [&](void* arg) {
+        Status s = *reinterpret_cast<Status*>(arg);
+        if (s.IsPathNotFound()) {
+          ++table_files_not_exist;
+        } else if (!s.ok()) {
+          assert(false);  // Should not reach here
+        }
+      });
+  SyncPoint::GetInstance()->EnableProcessing();
+  Options options;
+  options.env = env_;
+  options.level0_file_num_compaction_trigger = 4;
+  Reopen(options);
+
+  Options options1;
+  options1.env = env_;
+  options1.max_open_files = -1;
+  OpenSecondary(options1);
+
+  for (int i = 0; i != options.level0_file_num_compaction_trigger; ++i) {
+    ASSERT_OK(Put("foo", "foo_value" + std::to_string(i)));
+    ASSERT_OK(Put("bar", "bar_value" + std::to_string(i)));
+    ASSERT_OK(dbfull()->Flush(FlushOptions()));
+  }
+  ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable());
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
+
+  ASSERT_NE(nullptr, db_secondary_full());
+  ReadOptions ropts;
+  ropts.verify_checksums = true;
+  std::string value;
+  ASSERT_NOK(db_secondary_->Get(ropts, "foo", &value));
+  ASSERT_NOK(db_secondary_->Get(ropts, "bar", &value));
+
+  ASSERT_OK(db_secondary_->TryCatchUpWithPrimary());
+  ASSERT_EQ(options.level0_file_num_compaction_trigger, table_files_not_exist);
+  ASSERT_OK(db_secondary_->Get(ropts, "foo", &value));
+  ASSERT_EQ("foo_value" +
+                std::to_string(options.level0_file_num_compaction_trigger - 1),
+            value);
+  ASSERT_OK(db_secondary_->Get(ropts, "bar", &value));
+  ASSERT_EQ("bar_value" +
+                std::to_string(options.level0_file_num_compaction_trigger - 1),
+            value);
+  Iterator* iter = db_secondary_->NewIterator(ropts);
+  ASSERT_NE(nullptr, iter);
+  iter->Seek("bar");
+  ASSERT_TRUE(iter->Valid());
+  ASSERT_EQ("bar", iter->key().ToString());
+  ASSERT_EQ("bar_value" +
+                std::to_string(options.level0_file_num_compaction_trigger - 1),
+            iter->value().ToString());
+  iter->Seek("foo");
+  ASSERT_TRUE(iter->Valid());
+  ASSERT_EQ("foo", iter->key().ToString());
+  ASSERT_EQ("foo_value" +
+                std::to_string(options.level0_file_num_compaction_trigger - 1),
+            iter->value().ToString());
+  size_t count = 0;
+  for (iter->SeekToFirst(); iter->Valid(); iter->Next()) {
+    ++count;
+  }
+  ASSERT_EQ(2, count);
+  delete iter;
+}
+
+TEST_F(DBSecondaryTest, PrimaryDropColumnFamily) {
+  Options options;
+  options.env = env_;
+  const std::string kCfName1 = "pikachu";
+  CreateAndReopenWithCF({kCfName1}, options);
+
+  Options options1;
+  options1.env = env_;
+  options1.max_open_files = -1;
+  OpenSecondaryWithColumnFamilies({kCfName1}, options1);
+  ASSERT_EQ(2, handles_secondary_.size());
+
+  ASSERT_OK(Put(1 /*cf*/, "foo", "foo_val_1"));
+  ASSERT_OK(Flush(1 /*cf*/));
+
+  ASSERT_OK(db_secondary_->TryCatchUpWithPrimary());
+  ReadOptions ropts;
+  ropts.verify_checksums = true;
+  std::string value;
+  ASSERT_OK(db_secondary_->Get(ropts, handles_secondary_[1], "foo", &value));
+  ASSERT_EQ("foo_val_1", value);
+
+  ASSERT_OK(dbfull()->DropColumnFamily(handles_[1]));
+  Close();
+  CheckFileTypeCounts(dbname_, 1, 0, 1);
+  ASSERT_OK(db_secondary_->TryCatchUpWithPrimary());
+  value.clear();
+  ASSERT_OK(db_secondary_->Get(ropts, handles_secondary_[1], "foo", &value));
+  ASSERT_EQ("foo_val_1", value);
+}
+
+TEST_F(DBSecondaryTest, SwitchManifest) {
+  Options options;
+  options.env = env_;
+  options.level0_file_num_compaction_trigger = 4;
+  Reopen(options);
+
+  Options options1;
+  options1.env = env_;
+  options1.max_open_files = -1;
+  OpenSecondary(options1);
+
+  const int kNumFiles = options.level0_file_num_compaction_trigger - 1;
+  // Keep it smaller than 10 so that key0, key1, ..., key9 are sorted as 0, 1,
+  // ..., 9.
+  const int kNumKeys = 10;
+  // Create two sst
+  for (int i = 0; i != kNumFiles; ++i) {
+    for (int j = 0; j != kNumKeys; ++j) {
+      ASSERT_OK(Put("key" + std::to_string(j), "value_" + std::to_string(i)));
+    }
+    ASSERT_OK(Flush());
+  }
+
+  ASSERT_OK(db_secondary_->TryCatchUpWithPrimary());
+  const auto& range_scan_db = [&]() {
+    ReadOptions tmp_ropts;
+    tmp_ropts.total_order_seek = true;
+    tmp_ropts.verify_checksums = true;
+    std::unique_ptr<Iterator> iter(db_secondary_->NewIterator(tmp_ropts));
+    int cnt = 0;
+    for (iter->SeekToFirst(); iter->Valid(); iter->Next(), ++cnt) {
+      ASSERT_EQ("key" + std::to_string(cnt), iter->key().ToString());
+      ASSERT_EQ("value_" + std::to_string(kNumFiles - 1),
+                iter->value().ToString());
+    }
+  };
+
+  range_scan_db();
+
+  // While secondary instance still keeps old MANIFEST open, we close primary,
+  // restart primary, performs full compaction, close again, restart again so
+  // that next time secondary tries to catch up with primary, the secondary
+  // will skip the MANIFEST in middle.
+  Reopen(options);
+  ASSERT_OK(dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr));
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
+
+  Reopen(options);
+  ASSERT_OK(dbfull()->SetOptions({{"disable_auto_compactions", "false"}}));
+
+  ASSERT_OK(db_secondary_->TryCatchUpWithPrimary());
+  range_scan_db();
+}
+
+// Here, "Snapshot" refers to the version edits written by
+// VersionSet::WriteSnapshot() at the beginning of the new MANIFEST after
+// switching from the old one.
+TEST_F(DBSecondaryTest, SkipSnapshotAfterManifestSwitch) {
+  Options options;
+  options.env = env_;
+  options.disable_auto_compactions = true;
+  Reopen(options);
+
+  Options options1;
+  options1.env = env_;
+  options1.max_open_files = -1;
+  OpenSecondary(options1);
+
+  ASSERT_OK(Put("0", "value0"));
+  ASSERT_OK(Flush());
+  ASSERT_OK(db_secondary_->TryCatchUpWithPrimary());
+  std::string value;
+  ReadOptions ropts;
+  ropts.verify_checksums = true;
+  ASSERT_OK(db_secondary_->Get(ropts, "0", &value));
+  ASSERT_EQ("value0", value);
+
+  Reopen(options);
+  ASSERT_OK(dbfull()->SetOptions({{"disable_auto_compactions", "false"}}));
+  ASSERT_OK(db_secondary_->TryCatchUpWithPrimary());
+}
+
+TEST_F(DBSecondaryTest, SwitchWAL) {
+  const int kNumKeysPerMemtable = 1;
+  Options options;
+  options.env = env_;
+  options.max_write_buffer_number = 4;
+  options.min_write_buffer_number_to_merge = 2;
+  options.memtable_factory.reset(
+      new SpecialSkipListFactory(kNumKeysPerMemtable));
+  Reopen(options);
+
+  Options options1;
+  options1.env = env_;
+  options1.max_open_files = -1;
+  OpenSecondary(options1);
+
+  const auto& verify_db = [](DB* db1, DB* db2) {
+    ASSERT_NE(nullptr, db1);
+    ASSERT_NE(nullptr, db2);
+    ReadOptions read_opts;
+    read_opts.verify_checksums = true;
+    std::unique_ptr<Iterator> it1(db1->NewIterator(read_opts));
+    std::unique_ptr<Iterator> it2(db2->NewIterator(read_opts));
+    it1->SeekToFirst();
+    it2->SeekToFirst();
+    for (; it1->Valid() && it2->Valid(); it1->Next(), it2->Next()) {
+      ASSERT_EQ(it1->key(), it2->key());
+      ASSERT_EQ(it1->value(), it2->value());
+    }
+    ASSERT_FALSE(it1->Valid());
+    ASSERT_FALSE(it2->Valid());
+
+    for (it1->SeekToFirst(); it1->Valid(); it1->Next()) {
+      std::string value;
+      ASSERT_OK(db2->Get(read_opts, it1->key(), &value));
+      ASSERT_EQ(it1->value(), value);
+    }
+    for (it2->SeekToFirst(); it2->Valid(); it2->Next()) {
+      std::string value;
+      ASSERT_OK(db1->Get(read_opts, it2->key(), &value));
+      ASSERT_EQ(it2->value(), value);
+    }
+  };
+  for (int k = 0; k != 16; ++k) {
+    ASSERT_OK(Put("key" + std::to_string(k), "value" + std::to_string(k)));
+    ASSERT_OK(db_secondary_->TryCatchUpWithPrimary());
+    verify_db(dbfull(), db_secondary_);
+  }
+}
+
+TEST_F(DBSecondaryTest, SwitchWALMultiColumnFamilies) {
+  const int kNumKeysPerMemtable = 1;
+  SyncPoint::GetInstance()->DisableProcessing();
+  SyncPoint::GetInstance()->LoadDependency({
+      {"DBImpl::BackgroundCallFlush:ContextCleanedUp",
+       "DBSecondaryTest::SwitchWALMultipleColumnFamilies:BeforeCatchUp"}});
+  SyncPoint::GetInstance()->EnableProcessing();
+  const std::string kCFName1 = "pikachu";
+  Options options;
+  options.env = env_;
+  options.max_write_buffer_number = 4;
+  options.min_write_buffer_number_to_merge = 2;
+  options.memtable_factory.reset(
+      new SpecialSkipListFactory(kNumKeysPerMemtable));
+  CreateAndReopenWithCF({kCFName1}, options);
+
+  Options options1;
+  options1.env = env_;
+  options1.max_open_files = -1;
+  OpenSecondaryWithColumnFamilies({kCFName1}, options1);
+  ASSERT_EQ(2, handles_secondary_.size());
+
+  const auto& verify_db = [](DB* db1,
+                             const std::vector<ColumnFamilyHandle*>& handles1,
+                             DB* db2,
+                             const std::vector<ColumnFamilyHandle*>& handles2) {
+    ASSERT_NE(nullptr, db1);
+    ASSERT_NE(nullptr, db2);
+    ReadOptions read_opts;
+    read_opts.verify_checksums = true;
+    ASSERT_EQ(handles1.size(), handles2.size());
+    for (size_t i = 0; i != handles1.size(); ++i) {
+      std::unique_ptr<Iterator> it1(db1->NewIterator(read_opts, handles1[i]));
+      std::unique_ptr<Iterator> it2(db2->NewIterator(read_opts, handles2[i]));
+      it1->SeekToFirst();
+      it2->SeekToFirst();
+      for (; it1->Valid() && it2->Valid(); it1->Next(), it2->Next()) {
+        ASSERT_EQ(it1->key(), it2->key());
+        ASSERT_EQ(it1->value(), it2->value());
+      }
+      ASSERT_FALSE(it1->Valid());
+      ASSERT_FALSE(it2->Valid());
+
+      for (it1->SeekToFirst(); it1->Valid(); it1->Next()) {
+        std::string value;
+        ASSERT_OK(db2->Get(read_opts, handles2[i], it1->key(), &value));
+        ASSERT_EQ(it1->value(), value);
+      }
+      for (it2->SeekToFirst(); it2->Valid(); it2->Next()) {
+        std::string value;
+        ASSERT_OK(db1->Get(read_opts, handles1[i], it2->key(), &value));
+        ASSERT_EQ(it2->value(), value);
+      }
+    }
+  };
+  for (int k = 0; k != 8; ++k) {
+    ASSERT_OK(
+        Put(0 /*cf*/, "key" + std::to_string(k), "value" + std::to_string(k)));
+    ASSERT_OK(
+        Put(1 /*cf*/, "key" + std::to_string(k), "value" + std::to_string(k)));
+    TEST_SYNC_POINT("DBSecondaryTest::SwitchWALMultipleColumnFamilies:BeforeCatchUp");
+    ASSERT_OK(db_secondary_->TryCatchUpWithPrimary());
+    verify_db(dbfull(), handles_, db_secondary_, handles_secondary_);
+    SyncPoint::GetInstance()->ClearTrace();
+  }
+}
+
+TEST_F(DBSecondaryTest, CatchUpAfterFlush) {
+  const int kNumKeysPerMemtable = 16;
+  Options options;
+  options.env = env_;
+  options.max_write_buffer_number = 4;
+  options.min_write_buffer_number_to_merge = 2;
+  options.memtable_factory.reset(
+      new SpecialSkipListFactory(kNumKeysPerMemtable));
+  Reopen(options);
+
+  Options options1;
+  options1.env = env_;
+  options1.max_open_files = -1;
+  OpenSecondary(options1);
+
+  WriteOptions write_opts;
+  WriteBatch wb;
+  wb.Put("key0", "value0");
+  wb.Put("key1", "value1");
+  ASSERT_OK(dbfull()->Write(write_opts, &wb));
+  ReadOptions read_opts;
+  std::unique_ptr<Iterator> iter1(db_secondary_->NewIterator(read_opts));
+  iter1->Seek("key0");
+  ASSERT_FALSE(iter1->Valid());
+  iter1->Seek("key1");
+  ASSERT_FALSE(iter1->Valid());
+  ASSERT_OK(db_secondary_->TryCatchUpWithPrimary());
+  iter1->Seek("key0");
+  ASSERT_FALSE(iter1->Valid());
+  iter1->Seek("key1");
+  ASSERT_FALSE(iter1->Valid());
+  std::unique_ptr<Iterator> iter2(db_secondary_->NewIterator(read_opts));
+  iter2->Seek("key0");
+  ASSERT_TRUE(iter2->Valid());
+  ASSERT_EQ("value0", iter2->value());
+  iter2->Seek("key1");
+  ASSERT_TRUE(iter2->Valid());
+  ASSERT_EQ("value1", iter2->value());
+
+  {
+    WriteBatch wb1;
+    wb1.Put("key0", "value01");
+    wb1.Put("key1", "value11");
+    ASSERT_OK(dbfull()->Write(write_opts, &wb1));
+  }
+
+  {
+    WriteBatch wb2;
+    wb2.Put("key0", "new_value0");
+    wb2.Delete("key1");
+    ASSERT_OK(dbfull()->Write(write_opts, &wb2));
+  }
+
+  ASSERT_OK(Flush());
+
+  ASSERT_OK(db_secondary_->TryCatchUpWithPrimary());
+  std::unique_ptr<Iterator> iter3(db_secondary_->NewIterator(read_opts));
+  // iter3 should not see value01 and value11 at all.
+  iter3->Seek("key0");
+  ASSERT_TRUE(iter3->Valid());
+  ASSERT_EQ("new_value0", iter3->value());
+  iter3->Seek("key1");
+  ASSERT_FALSE(iter3->Valid());
+}
+
+TEST_F(DBSecondaryTest, CheckConsistencyWhenOpen) {
+  bool called = false;
+  Options options;
+  options.env = env_;
+  options.disable_auto_compactions = true;
+  Reopen(options);
+  SyncPoint::GetInstance()->DisableProcessing();
+  SyncPoint::GetInstance()->ClearAllCallBacks();
+  SyncPoint::GetInstance()->SetCallBack(
+      "DBImplSecondary::CheckConsistency:AfterFirstAttempt", [&](void* arg) {
+        ASSERT_NE(nullptr, arg);
+        called = true;
+        auto* s = reinterpret_cast<Status*>(arg);
+        ASSERT_NOK(*s);
+      });
+  SyncPoint::GetInstance()->LoadDependency(
+      {{"DBImpl::CheckConsistency:AfterGetLiveFilesMetaData",
+        "BackgroundCallCompaction:0"},
+       {"DBImpl::BackgroundCallCompaction:PurgedObsoleteFiles",
+        "DBImpl::CheckConsistency:BeforeGetFileSize"}});
+  SyncPoint::GetInstance()->EnableProcessing();
+
+  ASSERT_OK(Put("a", "value0"));
+  ASSERT_OK(Put("c", "value0"));
+  ASSERT_OK(Flush());
+  ASSERT_OK(Put("b", "value1"));
+  ASSERT_OK(Put("d", "value1"));
+  ASSERT_OK(Flush());
+  port::Thread thread([this]() {
+    Options opts;
+    opts.env = env_;
+    opts.max_open_files = -1;
+    OpenSecondary(opts);
+  });
+  ASSERT_OK(dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr));
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
+  thread.join();
+  ASSERT_TRUE(called);
+}
+#endif  //! ROCKSDB_LITE
+
+}  // namespace rocksdb
+
+int main(int argc, char** argv) {
+  rocksdb::port::InstallStackTraceHandler();
+  ::testing::InitGoogleTest(&argc, argv);
+  return RUN_ALL_TESTS();
+}
--- a/Show More
+++ b/Show More