Update history and version for 5.16.6 release.

Summary:
WriteBatchWithIndex's SeekForPrev() has a bug that we internally place the position just before the seek key rather than after. This makes the iterator to miss the result that is the same as the seek key. Fix it by position the iterator equal or smaller.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/4559

Differential Revision: D10468534

Pulled By: siying

fbshipit-source-id: 2fb371ae809c561b60a1c11cef71e1c66fea1f19

.gitignore

@@ -45,6 +45,8 @@ etags
 rocksdb_dump
 rocksdb_undump
 db_test2
+trace_analyzer
+trace_analyzer_test
 
 java/out
 java/target

CMakeLists.txt

@@ -48,6 +48,7 @@ option(WITH_JEMALLOC "build with JeMalloc" OFF)
 option(WITH_SNAPPY "build with SNAPPY" OFF)
 option(WITH_LZ4 "build with lz4" OFF)
 option(WITH_ZLIB "build with zlib" OFF)
+option(WITH_ZSTD "build with zstd" OFF)
 if(MSVC)
   # Defaults currently different for GFLAGS.
   #  We will address find_package work a little later
@@ -108,7 +109,6 @@ else()
     list(APPEND THIRDPARTY_LIBS ${LZ4_LIBRARIES})
   endif()
 
-  option(WITH_ZSTD "build with zstd" OFF)
   if(WITH_ZSTD)
     find_package(zstd REQUIRED)
     add_definitions(-DZSTD)
@@ -307,14 +307,6 @@ if(DISABLE_STALL_NOTIF)
   add_definitions(-DROCKSDB_DISABLE_STALL_NOTIFICATION)
 endif()
 
-# Used to run CI build and tests so we can run faster
-set(OPTIMIZE_DEBUG_DEFAULT 0)        # Debug build is unoptimized by default use -DOPTDBG=1 to optimize
-
-if(DEFINED OPTDBG)
-   set(OPTIMIZE_DEBUG ${OPTDBG})
-else()
-   set(OPTIMIZE_DEBUG ${OPTIMIZE_DEBUG_DEFAULT})
-endif()
 
 if(DEFINED USE_RTTI)
   if(USE_RTTI)
@@ -342,8 +334,10 @@ else()
   endif()
 endif()
 
+# Used to run CI build and tests so we can run faster
+option(OPTDBG "Build optimized debug build with MSVC" OFF)
 if(MSVC)
-  if((${OPTIMIZE_DEBUG} EQUAL 1))
+  if(OPTDBG)
     message(STATUS "Debug optimization is enabled")
     set(CMAKE_CXX_FLAGS_DEBUG "/Oxt /${RUNTIME_LIBRARY}d")
   else()
@@ -494,7 +488,6 @@ set(SOURCES
         db/log_reader.cc
         db/log_writer.cc
         db/malloc_stats.cc
-        db/managed_iterator.cc
         db/memtable.cc
         db/memtable_list.cc
         db/merge_helper.cc
@@ -556,6 +549,8 @@ set(SOURCES
         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/format.cc
         table/full_filter_block.cc
@@ -579,6 +574,7 @@ set(SOURCES
         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
         util/bloom.cc
@@ -603,25 +599,25 @@ set(SOURCES
         util/slice.cc
         util/sst_file_manager_impl.cc
         util/status.cc
-        util/status_message.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
         utilities/blob_db/blob_db.cc
         utilities/blob_db/blob_db_impl.cc
+        utilities/blob_db/blob_db_impl_filesnapshot.cc
         utilities/blob_db/blob_dump_tool.cc
         utilities/blob_db/blob_file.cc
         utilities/blob_db/blob_log_reader.cc
         utilities/blob_db/blob_log_writer.cc
         utilities/blob_db/blob_log_format.cc
-        utilities/blob_db/ttl_extractor.cc
         utilities/cassandra/cassandra_compaction_filter.cc
         utilities/cassandra/format.cc
         utilities/cassandra/merge_operator.cc
@@ -658,6 +654,7 @@ set(SOURCES
         utilities/simulator_cache/sim_cache.cc
         utilities/spatialdb/spatial_db.cc
         utilities/table_properties_collectors/compact_on_deletion_collector.cc
+        utilities/trace/file_trace_reader_writer.cc
         utilities/transactions/optimistic_transaction_db_impl.cc
         utilities/transactions/optimistic_transaction.cc
         utilities/transactions/pessimistic_transaction.cc
@@ -921,12 +918,14 @@ if(WITH_TESTS)
         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/merger_test.cc
         table/table_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

HISTORY.md

@@ -1,23 +1,74 @@
 # Rocksdb Change Log
-## Unreleased
+
+# 5.16.6 (11/24/2018)
+### Bug Fixes
+* Fix the bug that WriteBatchWithIndex's SeekForPrev() doesn't see the entries with the same key.
+
+## 5.16.5 (10/16/2018)
+### Bug Fixes
+* Fix slow flush/compaction when DB contains many snapshots. The problem became noticeable to us in DBs with 100,000+ snapshots, though it will affect others at different thresholds.
+* Properly set the stop key for a truncated manual CompactRange
+
+## 5.16.4 (10/10/2018)
+### Bug Fixes
+* Fix corner case where a write group leader blocked due to write stall blocks other writers in queue with WriteOptions::no_slowdown set.
+
+## 5.16.3 (10/1/2018)
+### Bug Fixes
+* Fix crash caused when `CompactFiles` run with `CompactionOptions::compression == CompressionType::kDisableCompressionOption`. Now that setting causes the compression type to be chosen according to the column family-wide compression options.
+
+## 5.16.2 (9/21/2018)
+### Bug Fixes
+* Fix bug in partition filters with format_version=4.
+
+## 5.16.1 (9/17/2018)
+### Bug Fixes
+* Remove trace_analyzer_tool from rocksdb_lib target in TARGETS file.
+* Fix RocksDB Java build and tests.
+* Remove sync point in Block destructor.
+
+## 5.16.0 (8/21/2018)
 ### Public API Change
-* For users of `Statistics` objects created via `CreateDBStatistics()`, the format of the string returned by its `ToString()` method has changed.
-* With LRUCache, when high_pri_pool_ratio > 0, midpoint insertion strategy will be enabled to put low-pri items to the tail of low-pri list (the midpoint) when they first inserted into the cache. This is to make cache entries never get hit age out faster, improving cache efficiency when large background scan presents.
+* `OnTableFileCreated` will now be called for empty files generated during compaction. In that case, `TableFileCreationInfo::file_path` will be "(nil)" and `TableFileCreationInfo::file_size` will be zero.
+* Add `FlushOptions::allow_write_stall`, which controls whether Flush calls start working immediately, even if it causes user writes to stall, or will wait until flush can be performed without causing write stall (similar to `CompactRangeOptions::allow_write_stall`). Note that the default value is false, meaning we add delay to Flush calls until stalling can be avoided when possible. This is behavior change compared to previous RocksDB versions, where Flush calls didn't check if they might cause stall or not.
+
+* The merge operands are passed to `MergeOperator::ShouldMerge` in the reversed order relative to how they were merged (passed to FullMerge or FullMergeV2) for performance reasons
+* GetAllKeyVersions() to take an extra argument of `max_num_ikeys`.
+
+### New Features
+* Changes the format of index blocks by delta encoding the index values, which are the block handles. This saves the encoding of BlockHandle::offset of the non-head index entries in each restart interval. The feature is backward compatible but not forward compatible. It is disabled by default unless format_version 4 or above is used.
+* Add a new tool: trace_analyzer. Trace_analyzer analyzes the trace file generated by using trace_replay API. It can convert the binary format trace file to a human readable txt file, output the statistics of the analyzed query types such as access statistics and size statistics, combining the dumped whole key space file to analyze, support query correlation analyzing, and etc. Current supported query types are: Get, Put, Delete, SingleDelete, DeleteRange, Merge, Iterator (Seek, SeekForPrev only).
+* Add hash index support to data blocks, which helps reducing the cpu utilization of point-lookup operations. This feature is backward compatible with the data block created without the hash index. It is disabled by default unless BlockBasedTableOptions::data_block_index_type is set to data_block_index_type = kDataBlockBinaryAndHash.
+
+### Bug Fixes
+* Fix a bug in misreporting the estimated partition index size in properties block.
+* Avoid creating empty SSTs and subsequently deleting them in certain cases during compaction.
+
+## 5.15.0 (7/17/2018)
+### Public API Change
+* Remove managed iterator. ReadOptions.managed is not effective anymore.
 * For bottommost_compression, a compatible CompressionOptions is added via `bottommost_compression_opts`. To keep backward compatible, a new boolean `enabled` is added to CompressionOptions. For compression_opts, it will be always used no matter what value of `enabled` is. For bottommost_compression_opts, it will only be used when user set `enabled=true`, otherwise, compression_opts will be used for bottommost_compression as default.
+* With LRUCache, when high_pri_pool_ratio > 0, midpoint insertion strategy will be enabled to put low-pri items to the tail of low-pri list (the midpoint) when they first inserted into the cache. This is to make cache entries never get hit age out faster, improving cache efficiency when large background scan presents.
+* For users of `Statistics` objects created via `CreateDBStatistics()`, the format of the string returned by its `ToString()` method has changed.
 * The "rocksdb.num.entries" table property no longer counts range deletion tombstones as entries.
 
 ### New Features
-* Changes the format of index blocks by storing the key in their raw form rather than converting them to InternalKey. This saves 8 bytes per index key. The feature is backward compatbile but not forward compatible. It is disabled by default unless format_version 3 or above is used.
+* Changes the format of index blocks by storing the key in their raw form rather than converting them to InternalKey. This saves 8 bytes per index key. The feature is backward compatible but not forward compatible. It is disabled by default unless format_version 3 or above is used.
+* Avoid memcpy when reading mmap files with OpenReadOnly and max_open_files==-1.
+* Support dynamically changing `ColumnFamilyOptions::ttl` via `SetOptions()`.
 * Add a new table property, "rocksdb.num.range-deletions", which counts the number of range deletion tombstones in the table.
 * Improve the performance of iterators doing long range scans by using readahead, when using direct IO.
 * pin_top_level_index_and_filter (default true) in BlockBasedTableOptions can be used in combination with cache_index_and_filter_blocks to prefetch and pin the top-level index of partitioned index and filter blocks in cache. It has no impact when cache_index_and_filter_blocks is false.
-* Avoid memcpy when reading mmap files with OpenReadOnly and max_open_files==-1
+* Write properties meta-block at the end of block-based table to save read-ahead IO.
 
 ### Bug Fixes
-* fix deadlock with enable_pipelined_write=true and max_successive_merges > 0
+* Fix deadlock with enable_pipelined_write=true and max_successive_merges > 0
+* Check conflict at output level in CompactFiles.
 * Fix corruption in non-iterator reads when mmap is used for file reads
 * Fix bug with prefix search in partition filters where a shared prefix would be ignored from the later partitions. The bug could report an eixstent key as missing. The bug could be triggered if prefix_extractor is set and partition filters is enabled.
 * Change default value of `bytes_max_delete_chunk` to 0 in NewSstFileManager() as it doesn't work well with checkpoints.
+* Fix a bug caused by not copying the block trailer with compressed SST file, direct IO, prefetcher and no compressed block cache.
+* Fix write can stuck indefinitely if enable_pipelined_write=true. The issue exists since pipelined write was introduced in 5.5.0.
 
 ## 5.14.0 (5/16/2018)
 ### Public API Change

Makefile

@@ -382,6 +382,8 @@ VALGRIND_OPTS = --error-exitcode=$(VALGRIND_ERROR) --leak-check=full
 
 BENCHTOOLOBJECTS = $(BENCH_LIB_SOURCES:.cc=.o) $(LIBOBJECTS) $(TESTUTIL)
 
+ANALYZETOOLOBJECTS = $(ANALYZER_LIB_SOURCES:.cc=.o)
+
 EXPOBJECTS = $(EXP_LIB_SOURCES:.cc=.o) $(LIBOBJECTS) $(TESTUTIL)
 
 TESTS = \
@@ -437,6 +439,7 @@ TESTS = \
 	table_properties_collector_test \
 	arena_test \
 	block_test \
+	data_block_hash_index_test \
 	cache_test \
 	corruption_test \
 	slice_transform_test \
@@ -529,6 +532,7 @@ TESTS = \
 	write_prepared_transaction_test \
 	write_unprepared_transaction_test \
 	db_universal_compaction_test \
+	trace_analyzer_test \
 
 PARALLEL_TEST = \
 	backupable_db_test \
@@ -572,6 +576,7 @@ TOOLS = \
 	rocksdb_dump \
 	rocksdb_undump \
 	blob_dump \
+	trace_analyzer \
 
 TEST_LIBS = \
 	librocksdb_env_basic_test.a
@@ -665,7 +670,7 @@ $(SHARED4): $(shared_all_libobjects)
 endif  # PLATFORM_SHARED_EXT
 
 .PHONY: blackbox_crash_test check clean coverage crash_test ldb_tests package \
-	release tags valgrind_check whitebox_crash_test format static_lib shared_lib all \
+	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
 
@@ -1018,6 +1023,13 @@ tags:
 	cscope -b `$(FIND) . -name '*.cc'` `$(FIND) . -name '*.h'` `$(FIND) . -name '*.c'`
 	ctags -e -R -o etags *
 
+tags0:
+	ctags -R .
+	cscope -b `$(FIND) . -name '*.cc' -and ! -name '*_test.cc'` \
+		  `$(FIND) . -name '*.c' -and ! -name '*_test.c'` \
+		  `$(FIND) . -name '*.h' -and ! -name '*_test.h'`
+	ctags -e -R -o etags *
+
 format:
 	build_tools/format-diff.sh
 
@@ -1031,7 +1043,7 @@ $(LIBRARY): $(LIBOBJECTS)
 	$(AM_V_AR)rm -f $@
 	$(AM_V_at)$(AR) $(ARFLAGS) $@ $(LIBOBJECTS)
 
-$(TOOLS_LIBRARY): $(BENCH_LIB_SOURCES:.cc=.o) $(TOOL_LIB_SOURCES:.cc=.o) $(LIB_SOURCES:.cc=.o) $(TESTUTIL)
+$(TOOLS_LIBRARY): $(BENCH_LIB_SOURCES:.cc=.o) $(TOOL_LIB_SOURCES:.cc=.o) $(LIB_SOURCES:.cc=.o) $(TESTUTIL) $(ANALYZER_LIB_SOURCES:.cc=.o)
 	$(AM_V_AR)rm -f $@
 	$(AM_V_at)$(AR) $(ARFLAGS) $@ $^
 
@@ -1042,6 +1054,9 @@ librocksdb_env_basic_test.a: env/env_basic_test.o $(LIBOBJECTS) $(TESTHARNESS)
 db_bench: tools/db_bench.o $(BENCHTOOLOBJECTS)
 	$(AM_LINK)
 
+trace_analyzer: tools/trace_analyzer.o $(ANALYZETOOLOBJECTS) $(LIBOBJECTS)
+	$(AM_LINK)
+
 cache_bench: cache/cache_bench.o $(LIBOBJECTS) $(TESTUTIL)
 	$(AM_LINK)
 
@@ -1350,6 +1365,9 @@ table_test: table/table_test.o $(LIBOBJECTS) $(TESTHARNESS)
 block_test: table/block_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)
 
+data_block_hash_index_test: table/data_block_hash_index_test.o $(LIBOBJECTS) $(TESTHARNESS)
+	$(AM_LINK)
+
 inlineskiplist_test: memtable/inlineskiplist_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)
 
@@ -1446,6 +1464,9 @@ options_util_test: utilities/options/options_util_test.o $(LIBOBJECTS) $(TESTHAR
 db_bench_tool_test: tools/db_bench_tool_test.o $(BENCHTOOLOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)
 
+trace_analyzer_test: tools/trace_analyzer_test.o $(LIBOBJECTS) $(ANALYZETOOLOBJECTS) $(TESTHARNESS)
+	$(AM_LINK)
+
 event_logger_test: util/event_logger_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)
 
@@ -1600,7 +1621,7 @@ ZLIB_SHA256 ?= c3e5e9fdd5004dcb542feda5ee4f0ff0744628baf8ed2dd5d66f8ca1197cb1a1
 ZLIB_DOWNLOAD_BASE ?= http://zlib.net
 BZIP2_VER ?= 1.0.6
 BZIP2_SHA256 ?= a2848f34fcd5d6cf47def00461fcb528a0484d8edef8208d6d2e2909dc61d9cd
-BZIP2_DOWNLOAD_BASE ?= http://www.bzip.org
+BZIP2_DOWNLOAD_BASE ?= https://web.archive.org/web/20180624184835/http://www.bzip.org
 SNAPPY_VER ?= 1.1.4
 SNAPPY_SHA256 ?= 134bfe122fd25599bb807bb8130e7ba6d9bdb851e0b16efcb83ac4f5d0b70057
 SNAPPY_DOWNLOAD_BASE ?= https://github.com/google/snappy/releases/download
@@ -1760,20 +1781,26 @@ rocksdbjavastaticrelease: rocksdbjavastatic
 	cd java/target;jar -uf $(ROCKSDB_JAR_ALL) librocksdbjni-*.so librocksdbjni-*.jnilib
 	cd java/target/classes;jar -uf ../$(ROCKSDB_JAR_ALL) org/rocksdb/*.class org/rocksdb/util/*.class
 
-rocksdbjavastaticreleasedocker: rocksdbjavastatic
-	DOCKER_LINUX_X64_CONTAINER=`docker ps -aqf name=rocksdb_linux_x64-be`; \
-	if [ -z "$$DOCKER_LINUX_X64_CONTAINER" ]; then \
-		docker container create --attach stdin --attach stdout --attach stderr --volume `pwd`:/rocksdb-host --name rocksdb_linux_x64-be evolvedbinary/rocksjava:centos6_x64-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh; \
-	fi
-	docker start -a rocksdb_linux_x64-be
+rocksdbjavastaticreleasedocker: rocksdbjavastatic rocksdbjavastaticdockerx86 rocksdbjavastaticdockerx86_64
+	cd java;jar -cf target/$(ROCKSDB_JAR_ALL) HISTORY*.md
+	cd java/target;jar -uf $(ROCKSDB_JAR_ALL) librocksdbjni-*.so librocksdbjni-*.jnilib
+	cd java/target/classes;jar -uf ../$(ROCKSDB_JAR_ALL) org/rocksdb/*.class org/rocksdb/util/*.class
+
+rocksdbjavastaticdockerx86:
+	mkdir -p java/target
 	DOCKER_LINUX_X86_CONTAINER=`docker ps -aqf name=rocksdb_linux_x86-be`; \
 	if [ -z "$$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
-	cd java;jar -cf target/$(ROCKSDB_JAR_ALL) HISTORY*.md
-	cd java/target;jar -uf $(ROCKSDB_JAR_ALL) librocksdbjni-*.so librocksdbjni-*.jnilib
-	cd java/target/classes;jar -uf ../$(ROCKSDB_JAR_ALL) org/rocksdb/*.class org/rocksdb/util/*.class
+
+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
 
 rocksdbjavastaticdockerppc64le:
 	mkdir -p java/target
@@ -1829,7 +1856,7 @@ $(java_libobjects): jl/%.o: %.cc
 rocksdbjava: $(java_all_libobjects)
 	$(AM_V_GEN)cd java;$(MAKE) javalib;
 	$(AM_V_at)rm -f ./java/target/$(ROCKSDBJNILIB)
-	$(AM_V_at)$(CXX) $(CXXFLAGS) -I./java/. $(JAVA_INCLUDE) -shared -fPIC -o ./java/target/$(ROCKSDBJNILIB) $(JNI_NATIVE_SOURCES) $(java_libobjects) $(JAVA_LDFLAGS) $(COVERAGEFLAGS)
+	$(AM_V_at)$(CXX) $(CXXFLAGS) -I./java/. $(JAVA_INCLUDE) -shared -fPIC -o ./java/target/$(ROCKSDBJNILIB) $(JNI_NATIVE_SOURCES) $(java_all_libobjects) $(JAVA_LDFLAGS) $(COVERAGEFLAGS)
 	$(AM_V_at)cd java;jar -cf target/$(ROCKSDB_JAR) HISTORY*.md
 	$(AM_V_at)cd java/target;jar -uf $(ROCKSDB_JAR) $(ROCKSDBJNILIB)
 	$(AM_V_at)cd java/target/classes;jar -uf ../$(ROCKSDB_JAR) org/rocksdb/*.class org/rocksdb/util/*.class
@@ -1898,7 +1925,7 @@ endif
 #  	Source files dependencies detection
 # ---------------------------------------------------------------------------
 
-all_sources = $(LIB_SOURCES) $(MAIN_SOURCES) $(MOCK_LIB_SOURCES) $(TOOL_LIB_SOURCES) $(BENCH_LIB_SOURCES) $(TEST_LIB_SOURCES) $(EXP_LIB_SOURCES)
+all_sources = $(LIB_SOURCES) $(MAIN_SOURCES) $(MOCK_LIB_SOURCES) $(TOOL_LIB_SOURCES) $(BENCH_LIB_SOURCES) $(TEST_LIB_SOURCES) $(EXP_LIB_SOURCES) $(ANALYZER_LIB_SOURCES)
 DEPFILES = $(all_sources:.cc=.cc.d)
 
 # Add proper dependency support so changing a .h file forces a .cc file to

TARGETS

@@ -1,3 +1,5 @@
+load("@fbcode_macros//build_defs:auto_headers.bzl", "AutoHeaders")
+
 REPO_PATH = package_name() + "/"
 
 BUCK_BINS = "buck-out/gen/" + REPO_PATH
@@ -106,7 +108,6 @@ cpp_library(
         "db/log_writer.cc",
         "db/logs_with_prep_tracker.cc",
         "db/malloc_stats.cc",
-        "db/managed_iterator.cc",
         "db/memtable.cc",
         "db/memtable_list.cc",
         "db/merge_helper.cc",
@@ -172,6 +173,8 @@ cpp_library(
         "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/format.cc",
         "table/full_filter_block.cc",
@@ -219,24 +222,24 @@ cpp_library(
         "util/slice.cc",
         "util/sst_file_manager_impl.cc",
         "util/status.cc",
-        "util/status_message.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",
         "utilities/blob_db/blob_db.cc",
         "utilities/blob_db/blob_db_impl.cc",
+        "utilities/blob_db/blob_db_impl_filesnapshot.cc",
         "utilities/blob_db/blob_dump_tool.cc",
         "utilities/blob_db/blob_file.cc",
         "utilities/blob_db/blob_log_format.cc",
         "utilities/blob_db/blob_log_reader.cc",
         "utilities/blob_db/blob_log_writer.cc",
-        "utilities/blob_db/ttl_extractor.cc",
         "utilities/cassandra/cassandra_compaction_filter.cc",
         "utilities/cassandra/format.cc",
         "utilities/cassandra/merge_operator.cc",
@@ -271,6 +274,7 @@ cpp_library(
         "utilities/simulator_cache/sim_cache.cc",
         "utilities/spatialdb/spatial_db.cc",
         "utilities/table_properties_collectors/compact_on_deletion_collector.cc",
+        "utilities/trace/file_trace_reader_writer.cc",
         "utilities/transactions/optimistic_transaction.cc",
         "utilities/transactions/optimistic_transaction_db_impl.cc",
         "utilities/transactions/pessimistic_transaction.cc",
@@ -301,6 +305,7 @@ cpp_library(
     srcs = [
         "db/db_test_util.cc",
         "table/mock_table.cc",
+        "tools/trace_analyzer_tool.cc",
         "util/fault_injection_test_env.cc",
         "util/testharness.cc",
         "util/testutil.cc",
@@ -499,6 +504,11 @@ ROCKS_TESTS = [
         "table/cuckoo_table_reader_test.cc",
         "serial",
     ],
+    [
+        "data_block_hash_index_test",
+        "table/data_block_hash_index_test.cc",
+        "serial",
+    ],
     [
         "date_tiered_test",
         "utilities/date_tiered/date_tiered_test.cc",
@@ -984,6 +994,11 @@ ROCKS_TESTS = [
         "util/timer_queue_test.cc",
         "serial",
     ],
+    [
+        "trace_analyzer_test",
+        "tools/trace_analyzer_test.cc",
+        "serial",
+    ],
     [
         "transaction_test",
         "utilities/transactions/transaction_test.cc",
@@ -1082,4 +1097,3 @@ if not is_opt_mode:
           deps = [":" + test_bin],
           command = [TEST_RUNNER, BUCK_BINS + test_bin]
         )
-

buckifier/targets_cfg.py

@@ -2,7 +2,10 @@ from __future__ import absolute_import
 from __future__ import division
 from __future__ import print_function
 from __future__ import unicode_literals
-rocksdb_target_header = """REPO_PATH = package_name() + "/"
+rocksdb_target_header = """
+load("@fbcode_macros//build_defs:auto_headers.bzl", "AutoHeaders")
+
+REPO_PATH = package_name() + "/"
 
 BUCK_BINS = "buck-out/gen/" + REPO_PATH
 

build_tools/dependencies.sh

@@ -1,19 +1,18 @@
-# shellcheck disable=SC2148
-GCC_BASE=/mnt/gvfs/third-party2/gcc/8219ec1bcedf8ad9da05e121e193364de2cc4f61/5.x/centos6-native/c447969
-CLANG_BASE=/mnt/gvfs/third-party2/llvm-fb/64d8d58e3d84f8bde7a029763d4f5baf39d0d5b9/stable/centos6-native/6aaf4de
-LIBGCC_BASE=/mnt/gvfs/third-party2/libgcc/ba9be983c81de7299b59fe71950c664a84dcb5f8/5.x/gcc-5-glibc-2.23/339d858
-GLIBC_BASE=/mnt/gvfs/third-party2/glibc/f20197cf3d4bd50339c9777aaa0b2ccadad9e2cb/2.23/gcc-5-glibc-2.23/ca1d1c0
-SNAPPY_BASE=/mnt/gvfs/third-party2/snappy/6427ce8c7496e4ab06c2da81543b94c0de8be3d0/1.1.3/gcc-5-glibc-2.23/9bc6787
-ZLIB_BASE=/mnt/gvfs/third-party2/zlib/8f1e8b867d26efef93eac2fabbdb2e1d512665d7/1.2.8/gcc-5-glibc-2.23/9bc6787
-BZIP2_BASE=/mnt/gvfs/third-party2/bzip2/70471c0571559fe0af7db6d7e8860b93a7eadfe1/1.0.6/gcc-5-glibc-2.23/9bc6787
-LZ4_BASE=/mnt/gvfs/third-party2/lz4/453c89d6f0e68cdf1c151c769197fabedad9cac8/r131/gcc-5-glibc-2.23/9bc6787
-ZSTD_BASE=/mnt/gvfs/third-party2/zstd/00a40fa5f8bd2cd0622f2e868552793aef37ccf4/1.3.0/gcc-5-glibc-2.23/03859b5
-GFLAGS_BASE=/mnt/gvfs/third-party2/gflags/47eef08f9acb77de982fbda6047c26d330739538/2.2.0/gcc-5-glibc-2.23/9bc6787
-JEMALLOC_BASE=/mnt/gvfs/third-party2/jemalloc/4414ddc78df8008b35cc4adac23590ad29148584/master/gcc-5-glibc-2.23/d506c82
-NUMA_BASE=/mnt/gvfs/third-party2/numa/9d7ae2693d05d62f9a579cb21e6b717cf257a75d/2.0.11/gcc-5-glibc-2.23/9bc6787
-LIBUNWIND_BASE=/mnt/gvfs/third-party2/libunwind/2b2dd58e3a52ccf2c1d827def59e5f740de0ad15/1.2/gcc-5-glibc-2.23/b443de1
-TBB_BASE=/mnt/gvfs/third-party2/tbb/379addf7ab2468a2b4293b47456cfcd1c9cb318d/4.3/gcc-5-glibc-2.23/9bc6787
-KERNEL_HEADERS_BASE=/mnt/gvfs/third-party2/kernel-headers/3f68f5fe65a85b7c2d3e66852268fbd1efdb3151/4.0.9-36_fbk5_2933_gd092e3f/gcc-5-glibc-2.23/da39a3e
-BINUTILS_BASE=/mnt/gvfs/third-party2/binutils/b9fab0aec99d9c36408e810b2677e91c12807afd/2.28/centos6-native/da39a3e
-VALGRIND_BASE=/mnt/gvfs/third-party2/valgrind/423431d61786b20bcc3bde8972901130cb29e6b3/3.11.0/gcc-5-glibc-2.23/9bc6787
-LUA_BASE=/mnt/gvfs/third-party2/lua/3b0bb3bd9a0f690a069c479fcc0f7424fc7456d2/5.2.3/gcc-5-glibc-2.23/65372bd
+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
+GLIBC_BASE=/mnt/gvfs/third-party2/glibc/3d8698d5973ba94f41620a80a67e4457fdf01e90/2.23/gcc-5-glibc-2.23/ca1d1c0
+SNAPPY_BASE=/mnt/gvfs/third-party2/snappy/7f9bdaada18f59bc27ec2b0871eb8a6144343aef/1.1.3/gcc-5-glibc-2.23/9bc6787
+ZLIB_BASE=/mnt/gvfs/third-party2/zlib/22c2d65676fb7c23cfa797c4f6937f38b026f3cf/1.2.8/gcc-5-glibc-2.23/9bc6787
+BZIP2_BASE=/mnt/gvfs/third-party2/bzip2/dc49a21c5fceec6456a7a28a94dcd16690af1337/1.0.6/gcc-5-glibc-2.23/9bc6787
+LZ4_BASE=/mnt/gvfs/third-party2/lz4/907b498203d297947f3bb70b9466f47e100f1873/r131/gcc-5-glibc-2.23/9bc6787
+ZSTD_BASE=/mnt/gvfs/third-party2/zstd/af6628a46758f1a15484a1760cd7294164bc5ba1/1.3.5/gcc-5-glibc-2.23/03859b5
+GFLAGS_BASE=/mnt/gvfs/third-party2/gflags/0b9929d2588991c65a57168bf88aff2db87c5d48/2.2.0/gcc-5-glibc-2.23/9bc6787
+JEMALLOC_BASE=/mnt/gvfs/third-party2/jemalloc/b1a0e56c1e3e6929813a4331ade3a58ff083afbb/master/gcc-5-glibc-2.23/aa64d6b
+NUMA_BASE=/mnt/gvfs/third-party2/numa/9cbf2460284c669ed19c3ccb200a71f7dd7e53c7/2.0.11/gcc-5-glibc-2.23/9bc6787
+LIBUNWIND_BASE=/mnt/gvfs/third-party2/libunwind/bf3d7497fe4e6d007354f0adffa16ce3003f8338/1.3/gcc-5-glibc-2.23/b443de1
+TBB_BASE=/mnt/gvfs/third-party2/tbb/ff4e0b093534704d8abab678a4fd7f5ea7b094c7/2018_U5/gcc-5-glibc-2.23/9bc6787
+KERNEL_HEADERS_BASE=/mnt/gvfs/third-party2/kernel-headers/b5c4a61a5c483ba24722005ae07895971a2ac707/4.0.9-36_fbk5_2933_gd092e3f/gcc-5-glibc-2.23/da39a3e
+BINUTILS_BASE=/mnt/gvfs/third-party2/binutils/55031de95a2b46c82948743419a603b3d6aefe28/2.29.1/centos7-native/da39a3e
+VALGRIND_BASE=/mnt/gvfs/third-party2/valgrind/f3f697a28122e6bcd513273dd9c1ff23852fc59f/3.13.0/gcc-5-glibc-2.23/9bc6787
+LUA_BASE=/mnt/gvfs/third-party2/lua/f0cd714433206d5139df61659eb7b28b1dea6683/5.2.3/gcc-5-glibc-2.23/65372bd

build_tools/error_filter.py

@@ -64,8 +64,12 @@ class MatchErrorParser(ErrorParserBase):
 
 class CompilerErrorParser(MatchErrorParser):
     def __init__(self):
-        # format: '<filename>:<line #>:<column #>: error: <error msg>'
-        super(CompilerErrorParser, self).__init__(r'\S+:\d+:\d+: error:')
+        # format (compile error):
+        #   '<filename>:<line #>:<column #>: error: <error msg>'
+        # format (link error):
+        #   '<filename>:<line #>: error: <error msg>'
+        # The below regex catches both
+        super(CompilerErrorParser, self).__init__(r'\S+:\d+: error:')
 
 
 class ScanBuildErrorParser(MatchErrorParser):

build_tools/fbcode_config.sh

@@ -43,12 +43,16 @@ if test -z $PIC_BUILD; then
   LZ4_INCLUDE=" -I $LZ4_BASE/include/"
   LZ4_LIBS=" $LZ4_BASE/lib/liblz4.a"
   CFLAGS+=" -DLZ4"
-
-  ZSTD_INCLUDE=" -I $ZSTD_BASE/include/"
-  ZSTD_LIBS=" $ZSTD_BASE/lib/libzstd.a"
-  CFLAGS+=" -DZSTD"
 fi
 
+ZSTD_INCLUDE=" -I $ZSTD_BASE/include/"
+if test -z $PIC_BUILD; then
+  ZSTD_LIBS=" $ZSTD_BASE/lib/libzstd.a"
+else
+  ZSTD_LIBS=" $ZSTD_BASE/lib/libzstd_pic.a"
+fi
+CFLAGS+=" -DZSTD"
+
 # location of gflags headers and libraries
 GFLAGS_INCLUDE=" -I $GFLAGS_BASE/include/"
 if test -z $PIC_BUILD; then

build_tools/rocksdb-lego-determinator

@@ -88,6 +88,8 @@ CLANG="USE_CLANG=1"
 LITE="OPT=\"-DROCKSDB_LITE -g\""
 TSAN="COMPILE_WITH_TSAN=1"
 UBSAN="COMPILE_WITH_UBSAN=1"
+TSAN_CRASH='CRASH_TEST_EXT_ARGS="--compression_type=zstd --log2_keys_per_lock=22"'
+NON_TSAN_CRASH="CRASH_TEST_EXT_ARGS=--compression_type=zstd"
 DISABLE_JEMALLOC="DISABLE_JEMALLOC=1"
 HTTP_PROXY="https_proxy=http://fwdproxy.29.prn1:8080 http_proxy=http://fwdproxy.29.prn1:8080 ftp_proxy=http://fwdproxy.29.prn1:8080"
 SETUP_JAVA_ENV="export $HTTP_PROXY; export JAVA_HOME=/usr/local/jdk-8u60-64/; export PATH=\$JAVA_HOME/bin:\$PATH"
@@ -380,14 +382,14 @@ STRESS_CRASH_TEST_COMMANDS="[
             $CLEANUP_ENV,
             {
                 'name':'Build and run RocksDB debug stress tests',
-                'shell':'$SHM $DEBUG make J=1 db_stress || $CONTRUN_NAME=db_stress $TASK_CREATION_TOOL',
+                'shell':'$SHM $DEBUG $NON_TSAN_CRASH make J=1 db_stress || $CONTRUN_NAME=db_stress $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
             {
                 'name':'Build and run RocksDB debug crash tests',
                 'timeout': 86400,
-                'shell':'$SHM $DEBUG make J=1 crash_test || $CONTRUN_NAME=crash_test $TASK_CREATION_TOOL',
+                'shell':'$SHM $DEBUG $NON_TSAN_CRASH make J=1 crash_test || $CONTRUN_NAME=crash_test $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             }
@@ -452,7 +454,7 @@ ASAN_CRASH_TEST_COMMANDS="[
             {
                 'name':'Build and run RocksDB debug asan_crash_test',
                 'timeout': 86400,
-                'shell':'$SHM $DEBUG make J=1 asan_crash_test || $CONTRUN_NAME=asan_crash_test $TASK_CREATION_TOOL',
+                'shell':'$SHM $DEBUG $NON_TSAN_CRASH make J=1 asan_crash_test || $CONTRUN_NAME=asan_crash_test $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -494,7 +496,7 @@ UBSAN_CRASH_TEST_COMMANDS="[
             {
                 'name':'Build and run RocksDB debug ubsan_crash_test',
                 'timeout': 86400,
-                'shell':'$SHM $DEBUG make J=1 ubsan_crash_test || $CONTRUN_NAME=ubsan_crash_test $TASK_CREATION_TOOL',
+                'shell':'$SHM $DEBUG $NON_TSAN_CRASH make J=1 ubsan_crash_test || $CONTRUN_NAME=ubsan_crash_test $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -560,7 +562,7 @@ TSAN_CRASH_TEST_COMMANDS="[
             {
                 'name':'Compile and run',
                 'timeout': 86400,
-                'shell':'set -o pipefail && $SHM $DEBUG $TSAN CRASH_TEST_KILL_ODD=1887 CRASH_TEST_EXT_ARGS=--log2_keys_per_lock=22  make J=1 crash_test || $CONTRUN_NAME=tsan_crash_test $TASK_CREATION_TOOL',
+                'shell':'set -o pipefail && $SHM $DEBUG $TSAN $TSAN_CRASH CRASH_TEST_KILL_ODD=1887 make J=1 crash_test || $CONTRUN_NAME=tsan_crash_test $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },

build_tools/update_dependencies.sh

@@ -65,8 +65,8 @@ touch "$OUTPUT"
 echo "Writing dependencies to $OUTPUT"
 
 # Compilers locations
-GCC_BASE=`readlink -f $TP2_LATEST/gcc/5.x/centos6-native/*/`
-CLANG_BASE=`readlink -f $TP2_LATEST/llvm-fb/stable/centos6-native/*/`
+GCC_BASE=`readlink -f $TP2_LATEST/gcc/5.x/centos7-native/*/`
+CLANG_BASE=`readlink -f $TP2_LATEST/llvm-fb/stable/centos7-native/*/`
 
 log_variable GCC_BASE
 log_variable CLANG_BASE
@@ -86,7 +86,7 @@ get_lib_base libunwind  LATEST  gcc-5-glibc-2.23
 get_lib_base tbb        LATEST  gcc-5-glibc-2.23
 
 get_lib_base kernel-headers 4.0.9-36_fbk5_2933_gd092e3f gcc-5-glibc-2.23
-get_lib_base binutils   LATEST centos6-native 
+get_lib_base binutils   LATEST centos7-native
 get_lib_base valgrind   LATEST gcc-5-glibc-2.23
 get_lib_base lua        5.2.3 gcc-5-glibc-2.23
 

db/builder.cc

@@ -230,11 +230,9 @@ Status BuildTable(
   }
 
   // Output to event logger and fire events.
-  if (!s.ok() || meta->fd.GetFileSize() > 0) {
-    EventHelpers::LogAndNotifyTableFileCreationFinished(
-        event_logger, ioptions.listeners, dbname, column_family_name, fname,
-        job_id, meta->fd, tp, reason, s);
-  }
+  EventHelpers::LogAndNotifyTableFileCreationFinished(
+      event_logger, ioptions.listeners, dbname, column_family_name, fname,
+      job_id, meta->fd, tp, reason, s);
 
   return s;
 }

db/builder.h

@@ -35,7 +35,6 @@ class VersionEdit;
 class TableBuilder;
 class WritableFileWriter;
 class InternalStats;
-class InternalIterator;
 
 // @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

db/c.cc

@@ -2433,6 +2433,18 @@ void rocksdb_options_set_max_write_buffer_number_to_maintain(
   opt->rep.max_write_buffer_number_to_maintain = n;
 }
 
+void rocksdb_options_set_enable_pipelined_write(rocksdb_options_t* opt, unsigned char v) {
+  opt->rep.enable_pipelined_write = v;
+}
+
+void rocksdb_options_set_max_subcompactions(rocksdb_options_t* opt, uint32_t n) {
+  opt->rep.max_subcompactions = n;
+}
+
+void rocksdb_options_set_max_background_jobs(rocksdb_options_t* opt, int n) {
+  opt->rep.max_background_jobs = n;
+}
+
 void rocksdb_options_set_max_background_compactions(rocksdb_options_t* opt, int n) {
   opt->rep.max_background_compactions = n;
 }

db/compact_files_test.cc

@@ -308,6 +308,47 @@ TEST_F(CompactFilesTest, CompactionFilterWithGetSv) {
   delete db;
 }
 
+TEST_F(CompactFilesTest, SentinelCompressionType) {
+  // Check that passing `CompressionType::kDisableCompressionOption` to
+  // `CompactFiles` causes it to use the column family compression options.
+  for (auto compaction_style :
+       {CompactionStyle::kCompactionStyleLevel,
+        CompactionStyle::kCompactionStyleUniversal,
+        CompactionStyle::kCompactionStyleNone}) {
+    DestroyDB(db_name_, Options());
+    Options options;
+    options.compaction_style = compaction_style;
+    // L0: Snappy, L1: ZSTD, L2: Snappy
+    options.compression_per_level = {CompressionType::kSnappyCompression,
+                                     CompressionType::kZlibCompression,
+                                     CompressionType::kSnappyCompression};
+    options.create_if_missing = true;
+    FlushedFileCollector* collector = new FlushedFileCollector();
+    options.listeners.emplace_back(collector);
+    DB* db = nullptr;
+    ASSERT_OK(DB::Open(options, db_name_, &db));
+
+    db->Put(WriteOptions(), "key", "val");
+    db->Flush(FlushOptions());
+
+    auto l0_files = collector->GetFlushedFiles();
+    ASSERT_EQ(1, l0_files.size());
+
+    // L0->L1 compaction, so output should be ZSTD-compressed
+    CompactionOptions compaction_opts;
+    compaction_opts.compression = CompressionType::kDisableCompressionOption;
+    ASSERT_OK(db->CompactFiles(compaction_opts, l0_files, 1));
+
+    rocksdb::TablePropertiesCollection all_tables_props;
+    ASSERT_OK(db->GetPropertiesOfAllTables(&all_tables_props));
+    for (const auto& name_and_table_props : all_tables_props) {
+      ASSERT_EQ(CompressionTypeToString(CompressionType::kZlibCompression),
+                name_and_table_props.second->compression_name);
+    }
+    delete db;
+  }
+}
+
 }  // namespace rocksdb
 
 int main(int argc, char** argv) {

db/compaction_iterator.cc

@@ -77,6 +77,12 @@ CompactionIterator::CompactionIterator(
     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;
@@ -603,18 +609,23 @@ void CompactionIterator::PrepareOutput() {
 inline SequenceNumber CompactionIterator::findEarliestVisibleSnapshot(
     SequenceNumber in, SequenceNumber* prev_snapshot) {
   assert(snapshots_->size());
-  SequenceNumber prev = kMaxSequenceNumber;
-  for (const auto cur : *snapshots_) {
-    assert(prev == kMaxSequenceNumber || prev <= cur);
-    if (cur >= in && (snapshot_checker_ == nullptr ||
-                      snapshot_checker_->IsInSnapshot(in, cur))) {
-      *prev_snapshot = prev == kMaxSequenceNumber ? 0 : prev;
+  auto snapshots_iter = std::lower_bound(
+      snapshots_->begin(), snapshots_->end(), in);
+  if (snapshots_iter == snapshots_->begin()) {
+    *prev_snapshot = 0;
+  } else {
+    *prev_snapshot = *std::prev(snapshots_iter);
+    assert(*prev_snapshot < in);
+  }
+  for (; snapshots_iter != snapshots_->end(); ++snapshots_iter) {
+    auto cur = *snapshots_iter;
+    assert(in <= cur);
+    if (snapshot_checker_ == nullptr ||
+        snapshot_checker_->IsInSnapshot(in, cur)) {
       return cur;
     }
-    prev = cur;
-    assert(prev < kMaxSequenceNumber);
+    *prev_snapshot = cur;
   }
-  *prev_snapshot = prev;
   return kMaxSequenceNumber;
 }
 

db/compaction_job.cc

@@ -600,6 +600,8 @@ Status CompactionJob::Run() {
   compaction_stats_.micros = env_->NowMicros() - start_micros;
   MeasureTime(stats_, COMPACTION_TIME, compaction_stats_.micros);
 
+  TEST_SYNC_POINT("CompactionJob::Run:BeforeVerify");
+
   // Check if any thread encountered an error during execution
   Status status;
   for (const auto& state : compact_->sub_compact_states) {
@@ -1075,7 +1077,8 @@ void CompactionJob::ProcessKeyValueCompaction(SubcompactionState* sub_compact) {
   }
 
   if (status.ok() && sub_compact->builder == nullptr &&
-      sub_compact->outputs.size() == 0) {
+      sub_compact->outputs.size() == 0 &&
+      !range_del_agg->IsEmpty()) {
     // handle subcompaction containing only range deletions
     status = OpenCompactionOutputFile(sub_compact);
   }

db/compaction_picker.cc

@@ -219,7 +219,8 @@ void CompactionPicker::GetRange(const std::vector<CompactionInputFiles>& inputs,
 
 bool CompactionPicker::ExpandInputsToCleanCut(const std::string& /*cf_name*/,
                                               VersionStorageInfo* vstorage,
-                                              CompactionInputFiles* inputs) {
+                                              CompactionInputFiles* inputs,
+                                              InternalKey** next_smallest) {
   // This isn't good compaction
   assert(!inputs->empty());
 
@@ -242,7 +243,8 @@ bool CompactionPicker::ExpandInputsToCleanCut(const std::string& /*cf_name*/,
     GetRange(*inputs, &smallest, &largest);
     inputs->clear();
     vstorage->GetOverlappingInputs(level, &smallest, &largest, &inputs->files,
-                                   hint_index, &hint_index);
+                                   hint_index, &hint_index, true,
+                                   next_smallest);
   } while (inputs->size() > old_size);
 
   // we started off with inputs non-empty and the previous loop only grew
@@ -315,13 +317,29 @@ Compaction* CompactionPicker::CompactFiles(
   // shouldn't have been released since.
   assert(!FilesRangeOverlapWithCompaction(input_files, output_level));
 
-  auto c =
-      new Compaction(vstorage, ioptions_, mutable_cf_options, input_files,
-                     output_level, compact_options.output_file_size_limit,
-                     mutable_cf_options.max_compaction_bytes, output_path_id,
-                     compact_options.compression, ioptions_.compression_opts,
-                     compact_options.max_subcompactions,
-                     /* grandparents */ {}, true);
+  CompressionType compression_type;
+  if (compact_options.compression == kDisableCompressionOption) {
+    int base_level;
+    if (ioptions_.compaction_style == kCompactionStyleLevel) {
+      base_level = vstorage->base_level();
+    } else {
+      base_level = 1;
+    }
+    compression_type =
+        GetCompressionType(ioptions_, vstorage, mutable_cf_options,
+                           output_level, base_level);
+  } else {
+    // TODO(ajkr): `CompactionOptions` offers configurable `CompressionType`
+    // without configurable `CompressionOptions`, which is inconsistent.
+    compression_type = compact_options.compression;
+  }
+  auto c = new Compaction(
+      vstorage, ioptions_, mutable_cf_options, input_files, output_level,
+      compact_options.output_file_size_limit,
+      mutable_cf_options.max_compaction_bytes, output_path_id, compression_type,
+      GetCompressionOptions(ioptions_, vstorage, output_level),
+      compact_options.max_subcompactions,
+      /* grandparents */ {}, true);
   RegisterCompaction(c);
   return c;
 }
@@ -633,7 +651,6 @@ Compaction* CompactionPicker::CompactRange(
       uint64_t s = inputs[i]->compensated_file_size;
       total += s;
       if (total >= limit) {
-        **compaction_end = inputs[i + 1]->smallest;
         covering_the_whole_range = false;
         inputs.files.resize(i + 1);
         break;
@@ -642,7 +659,10 @@ Compaction* CompactionPicker::CompactRange(
   }
   assert(output_path_id < static_cast<uint32_t>(ioptions_.cf_paths.size()));
 
-  if (ExpandInputsToCleanCut(cf_name, vstorage, &inputs) == false) {
+  InternalKey key_storage;
+  InternalKey* next_smallest = &key_storage;
+  if (ExpandInputsToCleanCut(cf_name, vstorage, &inputs, &next_smallest) ==
+      false) {
     // manual compaction is now multi-threaded, so it can
     // happen that ExpandWhileOverlapping fails
     // we handle it higher in RunManualCompaction
@@ -650,8 +670,10 @@ Compaction* CompactionPicker::CompactRange(
     return nullptr;
   }
 
-  if (covering_the_whole_range) {
+  if (covering_the_whole_range || !next_smallest) {
     *compaction_end = nullptr;
+  } else {
+    **compaction_end = *next_smallest;
   }
 
   CompactionInputFiles output_level_inputs;

db/compaction_picker.h

@@ -151,7 +151,8 @@ class CompactionPicker {
   // Will return false if it is impossible to apply this compaction.
   bool ExpandInputsToCleanCut(const std::string& cf_name,
                               VersionStorageInfo* vstorage,
-                              CompactionInputFiles* inputs);
+                              CompactionInputFiles* inputs,
+                              InternalKey** next_smallest = nullptr);
 
   // Returns true if any one of the parent files are being compacted
   bool IsRangeInCompaction(VersionStorageInfo* vstorage,

db/compaction_picker_test.cc

@@ -90,8 +90,8 @@ class CompactionPickerTest : public testing::Test {
     f->fd = FileDescriptor(file_number, path_id, file_size);
     f->smallest = InternalKey(smallest, smallest_seq, kTypeValue);
     f->largest = InternalKey(largest, largest_seq, kTypeValue);
-    f->smallest_seqno = smallest_seq;
-    f->largest_seqno = largest_seq;
+    f->fd.smallest_seqno = smallest_seq;
+    f->fd.largest_seqno = largest_seq;
     f->compensated_file_size = file_size;
     f->refs = 0;
     vstorage_->AddFile(level, f);

db/compaction_picker_universal.cc

@@ -97,17 +97,17 @@ void GetSmallestLargestSeqno(const std::vector<FileMetaData*>& files,
                              SequenceNumber* largest_seqno) {
   bool is_first = true;
   for (FileMetaData* f : files) {
-    assert(f->smallest_seqno <= f->largest_seqno);
+    assert(f->fd.smallest_seqno <= f->fd.largest_seqno);
     if (is_first) {
       is_first = false;
-      *smallest_seqno = f->smallest_seqno;
-      *largest_seqno = f->largest_seqno;
+      *smallest_seqno = f->fd.smallest_seqno;
+      *largest_seqno = f->fd.largest_seqno;
     } else {
-      if (f->smallest_seqno < *smallest_seqno) {
-        *smallest_seqno = f->smallest_seqno;
+      if (f->fd.smallest_seqno < *smallest_seqno) {
+        *smallest_seqno = f->fd.smallest_seqno;
       }
-      if (f->largest_seqno > *largest_seqno) {
-        *largest_seqno = f->largest_seqno;
+      if (f->fd.largest_seqno > *largest_seqno) {
+        *largest_seqno = f->fd.largest_seqno;
       }
     }
   }
@@ -365,11 +365,11 @@ Compaction* UniversalCompactionPicker::PickCompaction(
   size_t level_index = 0U;
   if (c->start_level() == 0) {
     for (auto f : *c->inputs(0)) {
-      assert(f->smallest_seqno <= f->largest_seqno);
+      assert(f->fd.smallest_seqno <= f->fd.largest_seqno);
       if (is_first) {
         is_first = false;
       }
-      prev_smallest_seqno = f->smallest_seqno;
+      prev_smallest_seqno = f->fd.smallest_seqno;
     }
     level_index = 1U;
   }

db/db_bloom_filter_test.cc

@@ -22,11 +22,12 @@ class DBBloomFilterTest : public DBTestBase {
 
 class DBBloomFilterTestWithParam
     : public DBTestBase,
-      public testing::WithParamInterface<std::tuple<bool, bool>> {
+      public testing::WithParamInterface<std::tuple<bool, bool, uint32_t>> {
   //                             public testing::WithParamInterface<bool> {
  protected:
   bool use_block_based_filter_;
   bool partition_filters_;
+  uint32_t format_version_;
 
  public:
   DBBloomFilterTestWithParam() : DBTestBase("/db_bloom_filter_tests") {}
@@ -36,9 +37,12 @@ class DBBloomFilterTestWithParam
   void SetUp() override {
     use_block_based_filter_ = std::get<0>(GetParam());
     partition_filters_ = std::get<1>(GetParam());
+    format_version_ = std::get<2>(GetParam());
   }
 };
 
+class DBBloomFilterTestDefFormatVersion : public DBBloomFilterTestWithParam {};
+
 class SliceTransformLimitedDomainGeneric : public SliceTransform {
   const char* Name() const override {
     return "SliceTransformLimitedDomainGeneric";
@@ -62,7 +66,7 @@ class SliceTransformLimitedDomainGeneric : public SliceTransform {
 // KeyMayExist can lead to a few false positives, but not false negatives.
 // To make test deterministic, use a much larger number of bits per key-20 than
 // bits in the key, so that false positives are eliminated
-TEST_P(DBBloomFilterTestWithParam, KeyMayExist) {
+TEST_P(DBBloomFilterTestDefFormatVersion, KeyMayExist) {
   do {
     ReadOptions ropts;
     std::string value;
@@ -401,6 +405,11 @@ TEST_P(DBBloomFilterTestWithParam, BloomFilter) {
       table_options.index_type =
           BlockBasedTableOptions::IndexType::kTwoLevelIndexSearch;
     }
+    table_options.format_version = format_version_;
+    if (format_version_ >= 4) {
+      // value delta encoding challenged more with index interval > 1
+      table_options.index_block_restart_interval = 8;
+    }
     table_options.metadata_block_size = 32;
     options.table_factory.reset(NewBlockBasedTableFactory(table_options));
 
@@ -456,10 +465,26 @@ TEST_P(DBBloomFilterTestWithParam, BloomFilter) {
   } while (ChangeCompactOptions());
 }
 
-INSTANTIATE_TEST_CASE_P(DBBloomFilterTestWithParam, DBBloomFilterTestWithParam,
-                        ::testing::Values(std::make_tuple(true, false),
-                                          std::make_tuple(false, true),
-                                          std::make_tuple(false, false)));
+INSTANTIATE_TEST_CASE_P(
+    FormatDef, DBBloomFilterTestDefFormatVersion,
+    ::testing::Values(std::make_tuple(true, false, test::kDefaultFormatVersion),
+                      std::make_tuple(false, true, test::kDefaultFormatVersion),
+                      std::make_tuple(false, false,
+                                      test::kDefaultFormatVersion)));
+
+INSTANTIATE_TEST_CASE_P(
+    FormatDef, DBBloomFilterTestWithParam,
+    ::testing::Values(std::make_tuple(true, false, test::kDefaultFormatVersion),
+                      std::make_tuple(false, true, test::kDefaultFormatVersion),
+                      std::make_tuple(false, false,
+                                      test::kDefaultFormatVersion)));
+
+INSTANTIATE_TEST_CASE_P(
+    FormatLatest, DBBloomFilterTestWithParam,
+    ::testing::Values(std::make_tuple(true, false, test::kLatestFormatVersion),
+                      std::make_tuple(false, true, test::kLatestFormatVersion),
+                      std::make_tuple(false, false,
+                                      test::kLatestFormatVersion)));
 
 TEST_F(DBBloomFilterTest, BloomFilterRate) {
   while (ChangeFilterOptions()) {

db/db_compaction_test.cc

@@ -116,6 +116,22 @@ private:
   std::vector<std::atomic<int>> compaction_completed_;
 };
 
+class SstStatsCollector : public EventListener {
+ public:
+  SstStatsCollector() : num_ssts_creation_started_(0) {}
+
+  void OnTableFileCreationStarted(const TableFileCreationBriefInfo& /* info */) override {
+    ++num_ssts_creation_started_;
+  }
+
+  int num_ssts_creation_started() {
+    return num_ssts_creation_started_;
+  }
+
+ private:
+  std::atomic<int> num_ssts_creation_started_;
+};
+
 static const int kCDTValueSize = 1000;
 static const int kCDTKeysPerBuffer = 4;
 static const int kCDTNumLevels = 8;
@@ -3376,14 +3392,13 @@ TEST_F(DBCompactionTest, LevelCompactExpiredTtlFiles) {
     }
     Flush();
   }
-  Flush();
   dbfull()->TEST_WaitForCompact();
   MoveFilesToLevel(3);
   ASSERT_EQ("0,0,0,2", FilesPerLevel());
 
+  // Delete previously written keys.
   for (int i = 0; i < kNumLevelFiles; ++i) {
     for (int j = 0; j < kNumKeysPerFile; ++j) {
-      // Overwrite previous keys with smaller, but predictable, values.
       ASSERT_OK(Delete(Key(i * kNumKeysPerFile + j)));
     }
     Flush();
@@ -3396,7 +3411,7 @@ TEST_F(DBCompactionTest, LevelCompactExpiredTtlFiles) {
   env_->addon_time_.fetch_add(36 * 60 * 60);  // 36 hours
   ASSERT_EQ("0,2,0,2", FilesPerLevel());
 
-  // Just do a siimple write + flush so that the Ttl expired files get
+  // Just do a simple write + flush so that the Ttl expired files get
   // compacted.
   ASSERT_OK(Put("a", "1"));
   Flush();
@@ -3410,6 +3425,57 @@ TEST_F(DBCompactionTest, LevelCompactExpiredTtlFiles) {
   // All non-L0 files are deleted, as they contained only deleted data.
   ASSERT_EQ("1", FilesPerLevel());
   rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+
+  // Test dynamically changing ttl.
+
+  env_->addon_time_.store(0);
+  DestroyAndReopen(options);
+
+  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());
+
+  // Delete previously written keys.
+  for (int i = 0; i < kNumLevelFiles; ++i) {
+    for (int j = 0; j < kNumKeysPerFile; ++j) {
+      ASSERT_OK(Delete(Key(i * kNumKeysPerFile + j)));
+    }
+    Flush();
+  }
+  dbfull()->TEST_WaitForCompact();
+  ASSERT_EQ("2,0,0,2", FilesPerLevel());
+  MoveFilesToLevel(1);
+  ASSERT_EQ("0,2,0,2", FilesPerLevel());
+
+  // Move time forward by 12 hours, and make sure that compaction still doesn't
+  // trigger as ttl is set to 24 hours.
+  env_->addon_time_.fetch_add(12 * 60 * 60);
+  ASSERT_OK(Put("a", "1"));
+  Flush();
+  dbfull()->TEST_WaitForCompact();
+  ASSERT_EQ("1,2,0,2", FilesPerLevel());
+
+  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+      "LevelCompactionPicker::PickCompaction:Return", [&](void* arg) {
+        Compaction* compaction = reinterpret_cast<Compaction*>(arg);
+        ASSERT_TRUE(compaction->compaction_reason() == CompactionReason::kTtl);
+      });
+  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+
+  // Dynamically change ttl to 10 hours.
+  // This should trigger a ttl compaction, as 12 hours have already passed.
+  ASSERT_OK(dbfull()->SetOptions({{"ttl", "36000"}}));
+  dbfull()->TEST_WaitForCompact();
+  // All non-L0 files are deleted, as they contained only deleted data.
+  ASSERT_EQ("1", FilesPerLevel());
+  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
 }
 
 TEST_F(DBCompactionTest, CompactRangeDelayedByL0FileCount) {
@@ -3435,12 +3501,13 @@ TEST_F(DBCompactionTest, CompactRangeDelayedByL0FileCount) {
       // ensure the auto compaction doesn't finish until manual compaction has
       // had a chance to be delayed.
       rocksdb::SyncPoint::GetInstance()->LoadDependency(
-          {{"DBImpl::CompactRange:StallWait", "CompactionJob::Run():End"}});
+          {{"DBImpl::WaitUntilFlushWouldNotStallWrites:StallWait",
+            "CompactionJob::Run():End"}});
     } else {
       // ensure the auto-compaction doesn't finish until manual compaction has
       // continued without delay.
       rocksdb::SyncPoint::GetInstance()->LoadDependency(
-          {{"DBImpl::CompactRange:StallWaitDone", "CompactionJob::Run():End"}});
+          {{"DBImpl::FlushMemTable:StallWaitDone", "CompactionJob::Run():End"}});
     }
     rocksdb::SyncPoint::GetInstance()->EnableProcessing();
 
@@ -3488,12 +3555,13 @@ TEST_F(DBCompactionTest, CompactRangeDelayedByImmMemTableCount) {
       // ensure the flush doesn't finish until manual compaction has had a
       // chance to be delayed.
       rocksdb::SyncPoint::GetInstance()->LoadDependency(
-          {{"DBImpl::CompactRange:StallWait", "FlushJob::WriteLevel0Table"}});
+          {{"DBImpl::WaitUntilFlushWouldNotStallWrites:StallWait",
+            "FlushJob::WriteLevel0Table"}});
     } else {
       // ensure the flush doesn't finish until manual compaction has continued
       // without delay.
       rocksdb::SyncPoint::GetInstance()->LoadDependency(
-          {{"DBImpl::CompactRange:StallWaitDone",
+          {{"DBImpl::FlushMemTable:StallWaitDone",
             "FlushJob::WriteLevel0Table"}});
     }
     rocksdb::SyncPoint::GetInstance()->EnableProcessing();
@@ -3503,6 +3571,7 @@ TEST_F(DBCompactionTest, CompactRangeDelayedByImmMemTableCount) {
       ASSERT_OK(Put(Key(0), RandomString(&rnd, 1024)));
       FlushOptions flush_opts;
       flush_opts.wait = false;
+      flush_opts.allow_write_stall = true;
       dbfull()->Flush(flush_opts);
     }
 
@@ -3538,7 +3607,7 @@ TEST_F(DBCompactionTest, CompactRangeShutdownWhileDelayed) {
     // The auto-compaction waits until the manual compaction finishes to ensure
     // the signal comes from closing CF/DB, not from compaction making progress.
     rocksdb::SyncPoint::GetInstance()->LoadDependency(
-        {{"DBImpl::CompactRange:StallWait",
+        {{"DBImpl::WaitUntilFlushWouldNotStallWrites:StallWait",
           "DBCompactionTest::CompactRangeShutdownWhileDelayed:PreShutdown"},
          {"DBCompactionTest::CompactRangeShutdownWhileDelayed:PostManual",
           "CompactionJob::Run():End"}});
@@ -3589,18 +3658,21 @@ TEST_F(DBCompactionTest, CompactRangeSkipFlushAfterDelay) {
   // began. So it unblocks CompactRange and precludes its flush. Throughout the
   // test, stall conditions are upheld via high L0 file count.
   rocksdb::SyncPoint::GetInstance()->LoadDependency(
-      {{"DBImpl::CompactRange:StallWait",
+      {{"DBImpl::WaitUntilFlushWouldNotStallWrites:StallWait",
         "DBCompactionTest::CompactRangeSkipFlushAfterDelay:PreFlush"},
        {"DBCompactionTest::CompactRangeSkipFlushAfterDelay:PostFlush",
-        "DBImpl::CompactRange:StallWaitDone"},
-       {"DBImpl::CompactRange:StallWaitDone", "CompactionJob::Run():End"}});
+        "DBImpl::FlushMemTable:StallWaitDone"},
+       {"DBImpl::FlushMemTable:StallWaitDone", "CompactionJob::Run():End"}});
   rocksdb::SyncPoint::GetInstance()->EnableProcessing();
 
+  //used for the delayable flushes
+  FlushOptions flush_opts;
+  flush_opts.allow_write_stall = true;
   for (int i = 0; i < kNumL0FilesLimit - 1; ++i) {
     for (int j = 0; j < 2; ++j) {
       ASSERT_OK(Put(Key(j), RandomString(&rnd, 1024)));
     }
-    Flush();
+    dbfull()->Flush(flush_opts);
   }
   auto manual_compaction_thread = port::Thread([this]() {
     CompactRangeOptions cro;
@@ -3610,7 +3682,7 @@ TEST_F(DBCompactionTest, CompactRangeSkipFlushAfterDelay) {
 
   TEST_SYNC_POINT("DBCompactionTest::CompactRangeSkipFlushAfterDelay:PreFlush");
   Put(ToString(0), RandomString(&rnd, 1024));
-  Flush();
+  dbfull()->Flush(flush_opts);
   Put(ToString(0), RandomString(&rnd, 1024));
   TEST_SYNC_POINT("DBCompactionTest::CompactRangeSkipFlushAfterDelay:PostFlush");
   manual_compaction_thread.join();
@@ -3766,6 +3838,30 @@ TEST_F(DBCompactionTest, CompactFilesOutputRangeConflict) {
   bg_thread.join();
 }
 
+TEST_F(DBCompactionTest, CompactionHasEmptyOutput) {
+  Options options = CurrentOptions();
+  SstStatsCollector* collector = new SstStatsCollector();
+  options.level0_file_num_compaction_trigger = 2;
+  options.listeners.emplace_back(collector);
+  Reopen(options);
+
+  // Make sure the L0 files overlap to prevent trivial move.
+  ASSERT_OK(Put("a", "val"));
+  ASSERT_OK(Put("b", "val"));
+  ASSERT_OK(Flush());
+  ASSERT_OK(Delete("a"));
+  ASSERT_OK(Delete("b"));
+  ASSERT_OK(Flush());
+
+  dbfull()->TEST_WaitForCompact();
+  ASSERT_EQ(NumTableFilesAtLevel(0), 0);
+  ASSERT_EQ(NumTableFilesAtLevel(1), 0);
+
+  // Expect one file creation to start for each flush, and zero for compaction
+  // since no keys are written.
+  ASSERT_EQ(2, collector->num_ssts_creation_started());
+}
+
 INSTANTIATE_TEST_CASE_P(DBCompactionTestWithParam, DBCompactionTestWithParam,
                         ::testing::Values(std::make_tuple(1, true),
                                           std::make_tuple(1, false),
@@ -3863,6 +3959,50 @@ INSTANTIATE_TEST_CASE_P(
                       CompactionPri::kOldestSmallestSeqFirst,
                       CompactionPri::kMinOverlappingRatio));
 
+class NoopMergeOperator : public MergeOperator {
+ public:
+  NoopMergeOperator() {}
+
+  virtual 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"; }
+};
+
+TEST_F(DBCompactionTest, PartialManualCompaction) {
+  Options opts = CurrentOptions();
+  opts.num_levels = 3;
+  opts.level0_file_num_compaction_trigger = 10;
+  opts.compression = kNoCompression;
+  opts.merge_operator.reset(new NoopMergeOperator());
+  opts.target_file_size_base = 10240;
+  DestroyAndReopen(opts);
+
+  Random rnd(301);
+  for (auto i = 0; i < 8; ++i) {
+    for (auto j = 0; j < 10; ++j) {
+      Merge("foo", RandomString(&rnd, 1024));
+    }
+    Flush();
+  }
+
+  MoveFilesToLevel(2);
+
+  std::string prop;
+  EXPECT_TRUE(dbfull()->GetProperty(DB::Properties::kLiveSstFilesSize, &prop));
+  uint64_t max_compaction_bytes = atoi(prop.c_str()) / 2;
+  ASSERT_OK(dbfull()->SetOptions(
+      {{"max_compaction_bytes", std::to_string(max_compaction_bytes)}}));
+
+  CompactRangeOptions cro;
+  cro.bottommost_level_compaction = BottommostLevelCompaction::kForce;
+  dbfull()->CompactRange(cro, nullptr, nullptr);
+}
+
 #endif // !defined(ROCKSDB_LITE)
 }  // namespace rocksdb
 

db/db_flush_test.cc

@@ -35,6 +35,7 @@ TEST_F(DBFlushTest, FlushWhileWritingManifest) {
   Reopen(options);
   FlushOptions no_wait;
   no_wait.wait = false;
+  no_wait.allow_write_stall=true;
 
   SyncPoint::GetInstance()->LoadDependency(
       {{"VersionSet::LogAndApply:WriteManifest",
@@ -55,6 +56,9 @@ TEST_F(DBFlushTest, FlushWhileWritingManifest) {
 #endif  // ROCKSDB_LITE
 }
 
+#ifndef TRAVIS
+// Disable this test temporarily on Travis as it fails intermittently.
+// Github issue: #4151
 TEST_F(DBFlushTest, SyncFail) {
   std::unique_ptr<FaultInjectionTestEnv> fault_injection_env(
       new FaultInjectionTestEnv(env_));
@@ -92,6 +96,7 @@ TEST_F(DBFlushTest, SyncFail) {
   ASSERT_EQ(refs_before, cfd->current()->TEST_refs());
   Destroy(options);
 }
+#endif  // TRAVIS
 
 TEST_F(DBFlushTest, FlushInLowPriThreadPool) {
   // Verify setting an empty high-pri (flush) thread pool causes flushes to be

db/db_impl.cc

@@ -41,7 +41,6 @@
 #include "db/log_reader.h"
 #include "db/log_writer.h"
 #include "db/malloc_stats.h"
-#include "db/managed_iterator.h"
 #include "db/memtable.h"
 #include "db/memtable_list.h"
 #include "db/merge_context.h"
@@ -1048,7 +1047,7 @@ InternalIterator* DBImpl::NewInternalIterator(
   } else {
     CleanupSuperVersion(super_version);
   }
-  return NewErrorInternalIterator(s, arena);
+  return NewErrorInternalIterator<Slice>(s, arena);
 }
 
 ColumnFamilyHandle* DBImpl::DefaultColumnFamily() const {
@@ -1072,6 +1071,15 @@ Status DBImpl::GetImpl(const ReadOptions& read_options,
   auto cfh = reinterpret_cast<ColumnFamilyHandleImpl*>(column_family);
   auto cfd = cfh->cfd();
 
+  if (tracer_) {
+    // TODO: This mutex should be removed later, to improve performance when
+    // tracing is enabled.
+    InstrumentedMutexLock lock(&trace_mutex_);
+    if (tracer_) {
+      tracer_->Get(column_family, key);
+    }
+  }
+
   // Acquire SuperVersion
   SuperVersion* sv = GetAndRefSuperVersion(cfd);
 
@@ -1578,6 +1586,10 @@ bool DBImpl::KeyMayExist(const ReadOptions& read_options,
 
 Iterator* DBImpl::NewIterator(const ReadOptions& read_options,
                               ColumnFamilyHandle* column_family) {
+  if (read_options.managed) {
+    return NewErrorIterator(
+        Status::NotSupported("Managed iterator is not supported anymore."));
+  }
   Iterator* result = nullptr;
   if (read_options.read_tier == kPersistedTier) {
     return NewErrorIterator(Status::NotSupported(
@@ -1595,22 +1607,7 @@ Iterator* DBImpl::NewIterator(const ReadOptions& read_options,
   auto cfh = reinterpret_cast<ColumnFamilyHandleImpl*>(column_family);
   auto cfd = cfh->cfd();
   ReadCallback* read_callback = nullptr;  // No read callback provided.
-  if (read_options.managed) {
-#ifdef ROCKSDB_LITE
-    // not supported in lite version
-    result = NewErrorIterator(Status::InvalidArgument(
-        "Managed Iterators not supported in RocksDBLite."));
-#else
-    if ((read_options.tailing) || (read_options.snapshot != nullptr) ||
-        (is_snapshot_supported_)) {
-      result = new ManagedIterator(this, read_options, cfd);
-    } else {
-      // Managed iter not supported
-      result = NewErrorIterator(Status::InvalidArgument(
-          "Managed Iterators not supported without snapshots."));
-    }
-#endif
-  } else if (read_options.tailing) {
+if (read_options.tailing) {
 #ifdef ROCKSDB_LITE
     // not supported in lite version
     result = nullptr;
@@ -1621,8 +1618,8 @@ Iterator* DBImpl::NewIterator(const ReadOptions& read_options,
     result = NewDBIterator(
         env_, read_options, *cfd->ioptions(), sv->mutable_cf_options,
         cfd->user_comparator(), iter, kMaxSequenceNumber,
-        sv->mutable_cf_options.max_sequential_skip_in_iterations,
-        read_callback);
+        sv->mutable_cf_options.max_sequential_skip_in_iterations, read_callback,
+        this, cfd);
 #endif
   } else {
     // Note: no need to consider the special case of
@@ -1689,9 +1686,8 @@ ArenaWrappedDBIter* DBImpl::NewIteratorImpl(const ReadOptions& read_options,
   ArenaWrappedDBIter* db_iter = NewArenaWrappedDbIterator(
       env_, read_options, *cfd->ioptions(), sv->mutable_cf_options, snapshot,
       sv->mutable_cf_options.max_sequential_skip_in_iterations,
-      sv->version_number, read_callback,
-      ((read_options.snapshot != nullptr) ? nullptr : this), cfd, allow_blob,
-      allow_refresh);
+      sv->version_number, read_callback, this, cfd, allow_blob,
+      ((read_options.snapshot != nullptr) ? false : allow_refresh));
 
   InternalIterator* internal_iter =
       NewInternalIterator(read_options, cfd, sv, db_iter->GetArena(),
@@ -1705,6 +1701,9 @@ Status DBImpl::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.");
@@ -1712,23 +1711,7 @@ Status DBImpl::NewIterators(
   ReadCallback* read_callback = nullptr;  // No read callback provided.
   iterators->clear();
   iterators->reserve(column_families.size());
-  if (read_options.managed) {
-#ifdef ROCKSDB_LITE
-    return Status::InvalidArgument(
-        "Managed interator not supported in RocksDB lite");
-#else
-    if ((!read_options.tailing) && (read_options.snapshot == nullptr) &&
-        (!is_snapshot_supported_)) {
-      return Status::InvalidArgument(
-          "Managed interator not supported without snapshots");
-    }
-    for (auto cfh : column_families) {
-      auto cfd = reinterpret_cast<ColumnFamilyHandleImpl*>(cfh)->cfd();
-      auto iter = new ManagedIterator(this, read_options, cfd);
-      iterators->push_back(iter);
-    }
-#endif
-  } else if (read_options.tailing) {
+  if (read_options.tailing) {
 #ifdef ROCKSDB_LITE
     return Status::InvalidArgument(
         "Tailing interator not supported in RocksDB lite");
@@ -1741,7 +1724,7 @@ Status DBImpl::NewIterators(
           env_, read_options, *cfd->ioptions(), sv->mutable_cf_options,
           cfd->user_comparator(), iter, kMaxSequenceNumber,
           sv->mutable_cf_options.max_sequential_skip_in_iterations,
-          read_callback));
+          read_callback, this, cfd));
     }
 #endif
   } else {
@@ -2252,9 +2235,9 @@ Status DBImpl::DeleteFile(std::string name) {
     status = versions_->LogAndApply(cfd, *cfd->GetLatestMutableCFOptions(),
                                     &edit, &mutex_, directories_.GetDbDir());
     if (status.ok()) {
-      InstallSuperVersionAndScheduleWork(cfd, &job_context.superversion_context,
-                                         *cfd->GetLatestMutableCFOptions(),
-                                         FlushReason::kDeleteFiles);
+      InstallSuperVersionAndScheduleWork(
+          cfd, &job_context.superversion_contexts[0],
+          *cfd->GetLatestMutableCFOptions(), FlushReason::kDeleteFiles);
     }
     FindObsoleteFiles(&job_context, false);
   }  // lock released here
@@ -2337,9 +2320,9 @@ Status DBImpl::DeleteFilesInRanges(ColumnFamilyHandle* column_family,
     status = versions_->LogAndApply(cfd, *cfd->GetLatestMutableCFOptions(),
                                     &edit, &mutex_, directories_.GetDbDir());
     if (status.ok()) {
-      InstallSuperVersionAndScheduleWork(cfd, &job_context.superversion_context,
-                                         *cfd->GetLatestMutableCFOptions(),
-                                         FlushReason::kDeleteFiles);
+      InstallSuperVersionAndScheduleWork(
+          cfd, &job_context.superversion_contexts[0],
+          *cfd->GetLatestMutableCFOptions(), FlushReason::kDeleteFiles);
     }
     for (auto* deleted_file : deleted_files) {
       deleted_file->being_compacted = false;
@@ -2904,6 +2887,10 @@ Status DBImpl::IngestExternalFile(
     ColumnFamilyHandle* column_family,
     const std::vector<std::string>& external_files,
     const IngestExternalFileOptions& ingestion_options) {
+  if (external_files.empty()) {
+    return Status::InvalidArgument("external_files is empty");
+  }
+
   Status status;
   auto cfh = reinterpret_cast<ColumnFamilyHandleImpl*>(column_family);
   auto cfd = cfh->cfd();
@@ -2921,6 +2908,9 @@ Status DBImpl::IngestExternalFile(
                                             immutable_db_options_, env_options_,
                                             &snapshots_, ingestion_options);
 
+  SuperVersionContext dummy_sv_ctx(/* create_superversion */ true);
+  VersionEdit dummy_edit;
+  uint64_t next_file_number = 0;
   std::list<uint64_t>::iterator pending_output_elem;
   {
     InstrumentedMutexLock l(&mutex_);
@@ -2931,12 +2921,33 @@ Status DBImpl::IngestExternalFile(
 
     // Make sure that bg cleanup wont delete the files that we are ingesting
     pending_output_elem = CaptureCurrentFileNumberInPendingOutputs();
+
+    // If crash happen after a hard link established, Recover function may
+    // reuse the file number that has already assigned to the internal file,
+    // and this will overwrite the external file. To protect the external
+    // file, we have to make sure the file number will never being reused.
+    next_file_number = versions_->FetchAddFileNumber(external_files.size());
+    auto cf_options = cfd->GetLatestMutableCFOptions();
+    status = versions_->LogAndApply(cfd, *cf_options, &dummy_edit, &mutex_,
+                                    directories_.GetDbDir());
+    if (status.ok()) {
+      InstallSuperVersionAndScheduleWork(cfd, &dummy_sv_ctx, *cf_options);
+    }
+  }
+  dummy_sv_ctx.Clean();
+  if (!status.ok()) {
+    InstrumentedMutexLock l(&mutex_);
+    ReleaseFileNumberFromPendingOutputs(pending_output_elem);
+    return status;
   }
 
   SuperVersion* super_version = cfd->GetReferencedSuperVersion(&mutex_);
-  status = ingestion_job.Prepare(external_files, super_version);
+  status =
+      ingestion_job.Prepare(external_files, next_file_number, super_version);
   CleanupSuperVersion(super_version);
   if (!status.ok()) {
+    InstrumentedMutexLock l(&mutex_);
+    ReleaseFileNumberFromPendingOutputs(pending_output_elem);
     return status;
   }
 
@@ -3030,8 +3041,6 @@ Status DBImpl::IngestExternalFile(
 
 Status DBImpl::VerifyChecksum() {
   Status s;
-  Options options;
-  EnvOptions env_options;
   std::vector<ColumnFamilyData*> cfd_list;
   {
     InstrumentedMutexLock l(&mutex_);
@@ -3049,13 +3058,19 @@ Status DBImpl::VerifyChecksum() {
   for (auto& sv : sv_list) {
     VersionStorageInfo* vstorage = sv->current->storage_info();
     ColumnFamilyData* cfd = sv->current->cfd();
+    Options opts;
+    {
+      InstrumentedMutexLock l(&mutex_);
+      opts = Options(BuildDBOptions(immutable_db_options_,
+         mutable_db_options_), cfd->GetLatestCFOptions());
+    }
     for (int i = 0; i < vstorage->num_non_empty_levels() && s.ok(); i++) {
       for (size_t j = 0; j < vstorage->LevelFilesBrief(i).num_files && s.ok();
            j++) {
         const auto& fd = vstorage->LevelFilesBrief(i).files[j].fd;
         std::string fname = TableFileName(cfd->ioptions()->cf_paths,
                                           fd.GetNumber(), fd.GetPathId());
-        s = rocksdb::VerifySstFileChecksum(options, env_options, fname);
+        s = rocksdb::VerifySstFileChecksum(opts, env_options_, fname);
       }
     }
     if (!s.ok()) {
@@ -3079,7 +3094,6 @@ Status DBImpl::VerifyChecksum() {
 
 void DBImpl::NotifyOnExternalFileIngested(
     ColumnFamilyData* cfd, const ExternalSstFileIngestionJob& ingestion_job) {
-#ifndef ROCKSDB_LITE
   if (immutable_db_options_.listeners.empty()) {
     return;
   }
@@ -3095,8 +3109,6 @@ void DBImpl::NotifyOnExternalFileIngested(
       listener->OnExternalFileIngested(this, info);
     }
   }
-
-#endif
 }
 
 void DBImpl::WaitForIngestFile() {
@@ -3106,5 +3118,42 @@ void DBImpl::WaitForIngestFile() {
   }
 }
 
+Status DBImpl::StartTrace(const TraceOptions& /* options */,
+                          std::unique_ptr<TraceWriter>&& trace_writer) {
+  InstrumentedMutexLock lock(&trace_mutex_);
+  tracer_.reset(new Tracer(env_, std::move(trace_writer)));
+  return Status::OK();
+}
+
+Status DBImpl::EndTrace() {
+  InstrumentedMutexLock lock(&trace_mutex_);
+  Status s = tracer_->Close();
+  tracer_.reset();
+  return s;
+}
+
+Status DBImpl::TraceIteratorSeek(const uint32_t& cf_id, const Slice& key) {
+  Status s;
+  if (tracer_) {
+    InstrumentedMutexLock lock(&trace_mutex_);
+    if (tracer_) {
+      s = tracer_->IteratorSeek(cf_id, key);
+    }
+  }
+  return s;
+}
+
+Status DBImpl::TraceIteratorSeekForPrev(const uint32_t& cf_id, const Slice& key) {
+  Status s;
+  if (tracer_) {
+    InstrumentedMutexLock lock(&trace_mutex_);
+    if (tracer_) {
+      s = tracer_->IteratorSeekForPrev(cf_id, key);
+    }
+  }
+  return s;
+}
+
 #endif  // ROCKSDB_LITE
+
 }  // namespace rocksdb

db/db_impl.h

@@ -46,6 +46,7 @@
 #include "rocksdb/env.h"
 #include "rocksdb/memtablerep.h"
 #include "rocksdb/status.h"
+#include "rocksdb/trace_reader_writer.h"
 #include "rocksdb/transaction_log.h"
 #include "rocksdb/write_buffer_manager.h"
 #include "table/scoped_arena_iterator.h"
@@ -54,6 +55,7 @@
 #include "util/hash.h"
 #include "util/stop_watch.h"
 #include "util/thread_local.h"
+#include "util/trace_replay.h"
 
 namespace rocksdb {
 
@@ -333,6 +335,15 @@ class DBImpl : public DB {
 
   virtual Status VerifyChecksum() override;
 
+  using DB::StartTrace;
+  virtual Status StartTrace(
+      const TraceOptions& options,
+      std::unique_ptr<TraceWriter>&& trace_writer) override;
+
+  using DB::EndTrace;
+  virtual Status EndTrace() override;
+  Status TraceIteratorSeek(const uint32_t& cf_id, const Slice& key);
+  Status TraceIteratorSeekForPrev(const uint32_t& cf_id, const Slice& key);
 #endif  // ROCKSDB_LITE
 
   // Similar to GetSnapshot(), but also lets the db know that this snapshot
@@ -385,7 +396,7 @@ class DBImpl : public DB {
   Status TEST_SwitchMemtable(ColumnFamilyData* cfd = nullptr);
 
   // Force current memtable contents to be flushed.
-  Status TEST_FlushMemTable(bool wait = true,
+  Status TEST_FlushMemTable(bool wait = true, bool allow_write_stall = false,
                             ColumnFamilyHandle* cfh = nullptr);
 
   // Wait for memtable compaction
@@ -697,6 +708,8 @@ class DBImpl : public DB {
   Statistics* stats_;
   std::unordered_map<std::string, RecoveredTransaction*>
       recovered_transactions_;
+  std::unique_ptr<Tracer> tracer_;
+  InstrumentedMutex trace_mutex_;
 
   // Except in DB::Open(), WriteOptionsFile can only be called when:
   // Persist options to options file.
@@ -790,18 +803,21 @@ class DBImpl : public DB {
   friend class WritePreparedTxnDB;
   friend class WriteBatchWithIndex;
   friend class WriteUnpreparedTxnDB;
+  friend class WriteUnpreparedTxn;
+
 #ifndef ROCKSDB_LITE
   friend class ForwardIterator;
 #endif
   friend struct SuperVersion;
   friend class CompactedDBImpl;
   friend class DBTest_ConcurrentFlushWAL_Test;
+  friend class DBTest_MixedSlowdownOptionsStop_Test;
 #ifndef NDEBUG
   friend class DBTest2_ReadCallbackTest_Test;
   friend class WriteCallbackTest_WriteWithCallbackTest_Test;
   friend class XFTransactionWriteHandler;
   friend class DBBlobIndexTest;
-  friend class WriteUnpreparedTransactionTest_RecoveryRollbackUnprepared_Test;
+  friend class WriteUnpreparedTransactionTest_RecoveryTest_Test;
 #endif
   struct CompactionState;
 
@@ -903,6 +919,10 @@ class DBImpl : public DB {
   Status FlushMemTable(ColumnFamilyData* cfd, const FlushOptions& options,
                        FlushReason flush_reason, bool writes_stopped = false);
 
+  // Wait until flushing this column family won't stall writes
+  Status WaitUntilFlushWouldNotStallWrites(ColumnFamilyData* cfd,
+                                           bool* flush_needed);
+
   // Wait for memtable flushed.
   // If flush_memtable_id is non-null, wait until the memtable with the ID
   // gets flush. Otherwise, wait until the column family don't have any

db/db_impl_compaction_flush.cc

@@ -160,8 +160,8 @@ Status DBImpl::FlushMemTableToOutputFile(
   }
 
   if (s.ok()) {
-    InstallSuperVersionAndScheduleWork(cfd, &job_context->superversion_context,
-                                       mutable_cf_options);
+    InstallSuperVersionAndScheduleWork(
+        cfd, &job_context->superversion_contexts[0], mutable_cf_options);
     if (made_progress) {
       *made_progress = 1;
     }
@@ -230,8 +230,8 @@ void DBImpl::NotifyOnFlushBegin(ColumnFamilyData* cfd, FileMetaData* file_meta,
     info.job_id = job_id;
     info.triggered_writes_slowdown = triggered_writes_slowdown;
     info.triggered_writes_stop = triggered_writes_stop;
-    info.smallest_seqno = file_meta->smallest_seqno;
-    info.largest_seqno = file_meta->largest_seqno;
+    info.smallest_seqno = file_meta->fd.smallest_seqno;
+    info.largest_seqno = file_meta->fd.largest_seqno;
     info.table_properties = prop;
     info.flush_reason = cfd->GetFlushReason();
     for (auto listener : immutable_db_options_.listeners) {
@@ -281,8 +281,8 @@ void DBImpl::NotifyOnFlushCompleted(ColumnFamilyData* cfd,
     info.job_id = job_id;
     info.triggered_writes_slowdown = triggered_writes_slowdown;
     info.triggered_writes_stop = triggered_writes_stop;
-    info.smallest_seqno = file_meta->smallest_seqno;
-    info.largest_seqno = file_meta->largest_seqno;
+    info.smallest_seqno = file_meta->fd.smallest_seqno;
+    info.largest_seqno = file_meta->fd.largest_seqno;
     info.table_properties = prop;
     info.flush_reason = cfd->GetFlushReason();
     for (auto listener : immutable_db_options_.listeners) {
@@ -324,60 +324,12 @@ Status DBImpl::CompactRange(const CompactRangeOptions& options,
     CleanupSuperVersion(super_version);
   }
 
-  if (!options.allow_write_stall && flush_needed) {
-    InstrumentedMutexLock l(&mutex_);
-    uint64_t orig_active_memtable_id = cfd->mem()->GetID();
-    WriteStallCondition write_stall_condition = WriteStallCondition::kNormal;
-    do {
-      if (write_stall_condition != WriteStallCondition::kNormal) {
-        TEST_SYNC_POINT("DBImpl::CompactRange:StallWait");
-        ROCKS_LOG_INFO(immutable_db_options_.info_log,
-                       "[%s] CompactRange waiting on stall conditions to clear",
-                       cfd->GetName().c_str());
-        bg_cv_.Wait();
-      }
-      if (cfd->IsDropped() || shutting_down_.load(std::memory_order_acquire)) {
-        return Status::ShutdownInProgress();
-      }
-
-      uint64_t earliest_memtable_id =
-          std::min(cfd->mem()->GetID(), cfd->imm()->GetEarliestMemTableID());
-      if (earliest_memtable_id > orig_active_memtable_id) {
-        // We waited so long that the memtable we were originally waiting on was
-        // flushed.
-        flush_needed = false;
-        break;
-      }
-
-      const auto& mutable_cf_options = *cfd->GetLatestMutableCFOptions();
-      const auto* vstorage = cfd->current()->storage_info();
-
-      // Skip stalling check if we're below auto-flush and auto-compaction
-      // triggers. If it stalled in these conditions, that'd mean the stall
-      // triggers are so low that stalling is needed for any background work. In
-      // that case we shouldn't wait since background work won't be scheduled.
-      if (cfd->imm()->NumNotFlushed() <
-              cfd->ioptions()->min_write_buffer_number_to_merge &&
-          vstorage->l0_delay_trigger_count() <
-              mutable_cf_options.level0_file_num_compaction_trigger) {
-        break;
-      }
-
-      // check whether one extra immutable memtable or an extra L0 file would
-      // cause write stalling mode to be entered. It could still enter stall
-      // mode due to pending compaction bytes, but that's less common
-      write_stall_condition =
-          ColumnFamilyData::GetWriteStallConditionAndCause(
-              cfd->imm()->NumNotFlushed() + 1,
-              vstorage->l0_delay_trigger_count() + 1,
-              vstorage->estimated_compaction_needed_bytes(), mutable_cf_options)
-              .first;
-    } while (write_stall_condition != WriteStallCondition::kNormal);
-  }
-  TEST_SYNC_POINT("DBImpl::CompactRange:StallWaitDone");
   Status s;
   if (flush_needed) {
-    s = FlushMemTable(cfd, FlushOptions(), FlushReason::kManualCompaction);
+    FlushOptions fo;
+    fo.allow_write_stall = options.allow_write_stall;
+    s = FlushMemTable(cfd, fo, FlushReason::kManualCompaction,
+      false /* writes_stopped*/);
     if (!s.ok()) {
       LogFlush(immutable_db_options_.info_log);
       return s;
@@ -691,7 +643,7 @@ Status DBImpl::CompactFilesImpl(
   Status status = compaction_job.Install(*c->mutable_cf_options());
   if (status.ok()) {
     InstallSuperVersionAndScheduleWork(
-        c->column_family_data(), &job_context->superversion_context,
+        c->column_family_data(), &job_context->superversion_contexts[0],
         *c->mutable_cf_options(), FlushReason::kManualCompaction);
   }
   c->ReleaseCompactionFiles(s);
@@ -885,7 +837,7 @@ Status DBImpl::ReFitLevel(ColumnFamilyData* cfd, int level, int target_level) {
       edit.DeleteFile(level, f->fd.GetNumber());
       edit.AddFile(to_level, f->fd.GetNumber(), f->fd.GetPathId(),
                    f->fd.GetFileSize(), f->smallest, f->largest,
-                   f->smallest_seqno, f->largest_seqno,
+                   f->fd.smallest_seqno, f->fd.largest_seqno,
                    f->marked_for_compaction);
     }
     ROCKS_LOG_DEBUG(immutable_db_options_.info_log,
@@ -1077,6 +1029,14 @@ Status DBImpl::FlushMemTable(ColumnFamilyData* cfd,
                              FlushReason flush_reason, bool writes_stopped) {
   Status s;
   uint64_t flush_memtable_id = 0;
+  if (!flush_options.allow_write_stall) {
+    bool flush_needed = true;
+    s = WaitUntilFlushWouldNotStallWrites(cfd, &flush_needed);
+    TEST_SYNC_POINT("DBImpl::FlushMemTable:StallWaitDone");
+    if (!s.ok() || !flush_needed) {
+      return s;
+    }
+  }
   {
     WriteContext context;
     InstrumentedMutexLock guard_lock(&mutex_);
@@ -1115,6 +1075,69 @@ Status DBImpl::FlushMemTable(ColumnFamilyData* cfd,
   return s;
 }
 
+// Calling FlushMemTable(), whether from DB::Flush() or from Backup Engine, can
+// cause write stall, for example if one memtable is being flushed already.
+// This method tries to avoid write stall (similar to CompactRange() behavior)
+// it emulates how the SuperVersion / LSM would change if flush happens, checks
+// 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) {
+  {
+    *flush_needed = true;
+    InstrumentedMutexLock l(&mutex_);
+    uint64_t orig_active_memtable_id = cfd->mem()->GetID();
+    WriteStallCondition write_stall_condition = WriteStallCondition::kNormal;
+    do {
+      if (write_stall_condition != WriteStallCondition::kNormal) {
+        TEST_SYNC_POINT("DBImpl::WaitUntilFlushWouldNotStallWrites:StallWait");
+        ROCKS_LOG_INFO(immutable_db_options_.info_log,
+                       "[%s] WaitUntilFlushWouldNotStallWrites"
+                       " waiting on stall conditions to clear",
+                       cfd->GetName().c_str());
+        bg_cv_.Wait();
+      }
+      if (cfd->IsDropped() || shutting_down_.load(std::memory_order_acquire)) {
+        return Status::ShutdownInProgress();
+      }
+
+      uint64_t earliest_memtable_id =
+          std::min(cfd->mem()->GetID(), cfd->imm()->GetEarliestMemTableID());
+      if (earliest_memtable_id > orig_active_memtable_id) {
+        // We waited so long that the memtable we were originally waiting on was
+        // flushed.
+        *flush_needed = false;
+        return Status::OK();
+      }
+
+      const auto& mutable_cf_options = *cfd->GetLatestMutableCFOptions();
+      const auto* vstorage = cfd->current()->storage_info();
+
+      // Skip stalling check if we're below auto-flush and auto-compaction
+      // triggers. If it stalled in these conditions, that'd mean the stall
+      // triggers are so low that stalling is needed for any background work. In
+      // that case we shouldn't wait since background work won't be scheduled.
+      if (cfd->imm()->NumNotFlushed() <
+              cfd->ioptions()->min_write_buffer_number_to_merge &&
+          vstorage->l0_delay_trigger_count() <
+              mutable_cf_options.level0_file_num_compaction_trigger) {
+        break;
+      }
+
+      // check whether one extra immutable memtable or an extra L0 file would
+      // cause write stalling mode to be entered. It could still enter stall
+      // mode due to pending compaction bytes, but that's less common
+      write_stall_condition =
+          ColumnFamilyData::GetWriteStallConditionAndCause(
+              cfd->imm()->NumNotFlushed() + 1,
+              vstorage->l0_delay_trigger_count() + 1,
+              vstorage->estimated_compaction_needed_bytes(), mutable_cf_options)
+              .first;
+    } while (write_stall_condition != WriteStallCondition::kNormal);
+  }
+  return Status::OK();
+}
+
 Status DBImpl::WaitForFlushMemTable(ColumnFamilyData* cfd,
                                     const uint64_t* flush_memtable_id) {
   Status s;
@@ -1775,7 +1798,7 @@ Status DBImpl::BackgroundCompaction(bool* made_progress,
                                     *c->mutable_cf_options(), c->edit(),
                                     &mutex_, directories_.GetDbDir());
     InstallSuperVersionAndScheduleWork(
-        c->column_family_data(), &job_context->superversion_context,
+        c->column_family_data(), &job_context->superversion_contexts[0],
         *c->mutable_cf_options(), FlushReason::kAutoCompaction);
     ROCKS_LOG_BUFFER(log_buffer, "[%s] Deleted %d files\n",
                      c->column_family_data()->GetName().c_str(),
@@ -1804,8 +1827,8 @@ Status DBImpl::BackgroundCompaction(bool* made_progress,
         c->edit()->DeleteFile(c->level(l), f->fd.GetNumber());
         c->edit()->AddFile(c->output_level(), f->fd.GetNumber(),
                            f->fd.GetPathId(), f->fd.GetFileSize(), f->smallest,
-                           f->largest, f->smallest_seqno, f->largest_seqno,
-                           f->marked_for_compaction);
+                           f->largest, f->fd.smallest_seqno,
+                           f->fd.largest_seqno, f->marked_for_compaction);
 
         ROCKS_LOG_BUFFER(
             log_buffer,
@@ -1822,7 +1845,7 @@ Status DBImpl::BackgroundCompaction(bool* made_progress,
                                     &mutex_, directories_.GetDbDir());
     // Use latest MutableCFOptions
     InstallSuperVersionAndScheduleWork(
-        c->column_family_data(), &job_context->superversion_context,
+        c->column_family_data(), &job_context->superversion_contexts[0],
         *c->mutable_cf_options(), FlushReason::kAutoCompaction);
 
     VersionStorageInfo::LevelSummaryStorage tmp;
@@ -1899,7 +1922,7 @@ Status DBImpl::BackgroundCompaction(bool* made_progress,
     status = compaction_job.Install(*c->mutable_cf_options());
     if (status.ok()) {
       InstallSuperVersionAndScheduleWork(
-          c->column_family_data(), &job_context->superversion_context,
+          c->column_family_data(), &job_context->superversion_contexts[0],
           *c->mutable_cf_options(), FlushReason::kAutoCompaction);
     }
     *made_progress = true;
@@ -2091,7 +2114,8 @@ void DBImpl::InstallSuperVersionAndScheduleWork(
                         old_sv->mutable_cf_options.max_write_buffer_number;
   }
 
-  if (sv_context->new_superversion == nullptr) {
+  // this branch is unlikely to step in
+  if (UNLIKELY(sv_context->new_superversion == nullptr)) {
     sv_context->NewSuperVersion();
   }
   cfd->InstallSuperVersion(sv_context, &mutex_, mutable_cf_options);

db/db_impl_debug.cc

@@ -100,9 +100,11 @@ Status DBImpl::TEST_SwitchMemtable(ColumnFamilyData* cfd) {
   return SwitchMemtable(cfd, &write_context);
 }
 
-Status DBImpl::TEST_FlushMemTable(bool wait, ColumnFamilyHandle* cfh) {
+Status DBImpl::TEST_FlushMemTable(bool wait, bool allow_write_stall,
+    ColumnFamilyHandle* cfh) {
   FlushOptions fo;
   fo.wait = wait;
+  fo.allow_write_stall = allow_write_stall;
   ColumnFamilyData* cfd;
   if (cfh == nullptr) {
     cfd = default_cf_handle_->cfd();

db/db_impl_experimental.cc

@@ -131,16 +131,16 @@ Status DBImpl::PromoteL0(ColumnFamilyHandle* column_family, int target_level) {
       edit.DeleteFile(0, f->fd.GetNumber());
       edit.AddFile(target_level, f->fd.GetNumber(), f->fd.GetPathId(),
                    f->fd.GetFileSize(), f->smallest, f->largest,
-                   f->smallest_seqno, f->largest_seqno,
+                   f->fd.smallest_seqno, f->fd.largest_seqno,
                    f->marked_for_compaction);
     }
 
     status = versions_->LogAndApply(cfd, *cfd->GetLatestMutableCFOptions(),
                                     &edit, &mutex_, directories_.GetDbDir());
     if (status.ok()) {
-      InstallSuperVersionAndScheduleWork(
-          cfd, &job_context.superversion_context,
-          *cfd->GetLatestMutableCFOptions());
+      InstallSuperVersionAndScheduleWork(cfd,
+                                         &job_context.superversion_contexts[0],
+                                         *cfd->GetLatestMutableCFOptions());
     }
   }  // lock released here
   LogFlush(immutable_db_options_.info_log);

db/db_impl_open.cc

@@ -361,7 +361,7 @@ Status DBImpl::Recover(
       s = env_->NewRandomAccessFile(IdentityFileName(dbname_), &idfile,
                                     customized_env);
       if (!s.ok()) {
-        const char* error_msg = s.ToString().c_str();
+        std::string error_str = s.ToString();
         // Check if unsupported Direct I/O is the root cause
         customized_env.use_direct_reads = false;
         s = env_->NewRandomAccessFile(IdentityFileName(dbname_), &idfile,
@@ -371,7 +371,7 @@ Status DBImpl::Recover(
               "Direct I/O is not supported by the specified DB.");
         } else {
           return Status::InvalidArgument(
-              "Found options incompatible with filesystem", error_msg);
+              "Found options incompatible with filesystem", error_str.c_str());
         }
       }
     }
@@ -969,7 +969,7 @@ Status DBImpl::WriteLevel0TableForRecovery(int job_id, ColumnFamilyData* cfd,
   if (s.ok() && meta.fd.GetFileSize() > 0) {
     edit->AddFile(level, meta.fd.GetNumber(), meta.fd.GetPathId(),
                   meta.fd.GetFileSize(), meta.smallest, meta.largest,
-                  meta.smallest_seqno, meta.largest_seqno,
+                  meta.fd.smallest_seqno, meta.fd.largest_seqno,
                   meta.marked_for_compaction);
   }
 

db/db_impl_readonly.cc

@@ -35,6 +35,12 @@ Status DBImplReadOnly::Get(const ReadOptions& read_options,
   SequenceNumber snapshot = versions_->LastSequence();
   auto cfh = reinterpret_cast<ColumnFamilyHandleImpl*>(column_family);
   auto cfd = cfh->cfd();
+  if (tracer_) {
+    InstrumentedMutexLock lock(&trace_mutex_);
+    if (tracer_) {
+      tracer_->Get(column_family, key);
+    }
+  }
   SuperVersion* super_version = cfd->GetSuperVersion();
   MergeContext merge_context;
   RangeDelAggregator range_del_agg(cfd->internal_comparator(), snapshot);

db/db_impl_write.cc

@@ -76,6 +76,12 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
   if (my_batch == nullptr) {
     return Status::Corruption("Batch is nullptr!");
   }
+  if (tracer_) {
+    InstrumentedMutexLock lock(&trace_mutex_);
+    if (tracer_) {
+      tracer_->Write(my_batch);
+    }
+  }
   if (write_options.sync && write_options.disableWAL) {
     return Status::InvalidArgument("Sync writes has to enable WAL.");
   }
@@ -311,7 +317,8 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
         w.status = WriteBatchInternal::InsertInto(
             write_group, current_sequence, column_family_memtables_.get(),
             &flush_scheduler_, write_options.ignore_missing_column_families,
-            0 /*recovery_log_number*/, this, parallel, seq_per_batch_);
+            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
@@ -346,7 +353,7 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
               &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_,
-              w.batch_cnt);
+              w.batch_cnt, batch_per_txn_);
         }
       }
       if (seq_used != nullptr) {
@@ -508,7 +515,8 @@ Status DBImpl::PipelinedWriteImpl(const WriteOptions& write_options,
       memtable_write_group.status = WriteBatchInternal::InsertInto(
           memtable_write_group, w.sequence, column_family_memtables_.get(),
           &flush_scheduler_, write_options.ignore_missing_column_families,
-          0 /*log_number*/, this, seq_per_batch_);
+          0 /*log_number*/, this, false /*concurrent_memtable_writes*/,
+          seq_per_batch_, batch_per_txn_);
       versions_->SetLastSequence(memtable_write_group.last_sequence);
       write_thread_.ExitAsMemTableWriter(&w, memtable_write_group);
     }
@@ -565,7 +573,6 @@ Status DBImpl::WriteImplWALOnly(const WriteOptions& write_options,
   }
   // else we are the leader of the write batch group
   assert(w.state == WriteThread::STATE_GROUP_LEADER);
-  WriteContext write_context;
   WriteThread::WriteGroup write_group;
   uint64_t last_sequence;
   nonmem_write_thread_.EnterAsBatchGroupLeader(&w, &write_group);
@@ -1139,10 +1146,14 @@ Status DBImpl::DelayWrite(uint64_t num_bytes,
     uint64_t delay = write_controller_.GetDelay(env_, num_bytes);
     if (delay > 0) {
       if (write_options.no_slowdown) {
-        return Status::Incomplete();
+        return Status::Incomplete("Write stall");
       }
       TEST_SYNC_POINT("DBImpl::DelayWrite:Sleep");
 
+      // Notify write_thread_ about the stall so it can setup a barrier and
+      // fail any pending writers with no_slowdown
+      write_thread_.BeginWriteStall();
+      TEST_SYNC_POINT("DBImpl::DelayWrite:BeginWriteStallDone");
       mutex_.Unlock();
       // We will delay the write until we have slept for delay ms or
       // we don't need a delay anymore
@@ -1159,15 +1170,21 @@ Status DBImpl::DelayWrite(uint64_t num_bytes,
         env_->SleepForMicroseconds(kDelayInterval);
       }
       mutex_.Lock();
+      write_thread_.EndWriteStall();
     }
 
     while (!error_handler_.IsDBStopped() && write_controller_.IsStopped()) {
       if (write_options.no_slowdown) {
-        return Status::Incomplete();
+        return Status::Incomplete("Write stall");
       }
       delayed = true;
+
+      // Notify write_thread_ about the stall so it can setup a barrier and
+      // fail any pending writers with no_slowdown
+      write_thread_.BeginWriteStall();
       TEST_SYNC_POINT("DBImpl::DelayWrite:Wait");
       bg_cv_.Wait();
+      write_thread_.EndWriteStall();
     }
   }
   assert(!delayed || !write_options.no_slowdown);

db/db_iter.cc

@@ -27,6 +27,7 @@
 #include "util/logging.h"
 #include "util/mutexlock.h"
 #include "util/string_util.h"
+#include "util/trace_replay.h"
 
 namespace rocksdb {
 
@@ -114,7 +115,8 @@ class DBIter final: public Iterator {
          const MutableCFOptions& mutable_cf_options, const Comparator* cmp,
          InternalIterator* iter, SequenceNumber s, bool arena_mode,
          uint64_t max_sequential_skip_in_iterations,
-         ReadCallback* read_callback, bool allow_blob)
+         ReadCallback* read_callback, DBImpl* db_impl, ColumnFamilyData* cfd,
+         bool allow_blob)
       : arena_mode_(arena_mode),
         env_(_env),
         logger_(cf_options.info_log),
@@ -135,6 +137,8 @@ class DBIter final: public Iterator {
         range_del_agg_(cf_options.internal_comparator, s,
                        true /* collapse_deletions */),
         read_callback_(read_callback),
+        db_impl_(db_impl),
+        cfd_(cfd),
         allow_blob_(allow_blob),
         is_blob_(false),
         start_seqnum_(read_options.iter_start_seqnum) {
@@ -344,6 +348,8 @@ class DBIter final: public Iterator {
   LocalStatistics local_stats_;
   PinnedIteratorsManager pinned_iters_mgr_;
   ReadCallback* read_callback_;
+  DBImpl* db_impl_;
+  ColumnFamilyData* cfd_;
   bool allow_blob_;
   bool is_blob_;
   // for diff snapshots we want the lower bound on the seqnum;
@@ -1267,6 +1273,12 @@ void DBIter::Seek(const Slice& target) {
   saved_key_.Clear();
   saved_key_.SetInternalKey(target, seq);
 
+#ifndef ROCKSDB_LITE
+  if (db_impl_ != nullptr && cfd_ != nullptr) {
+    db_impl_->TraceIteratorSeek(cfd_->GetID(), target);
+  }
+#endif  // ROCKSDB_LITE
+
   if (iterate_lower_bound_ != nullptr &&
       user_comparator_->Compare(saved_key_.GetUserKey(),
                                 *iterate_lower_bound_) < 0) {
@@ -1331,6 +1343,12 @@ void DBIter::SeekForPrev(const Slice& target) {
     range_del_agg_.InvalidateRangeDelMapPositions();
   }
 
+#ifndef ROCKSDB_LITE
+  if (db_impl_ != nullptr && cfd_ != nullptr) {
+    db_impl_->TraceIteratorSeekForPrev(cfd_->GetID(), target);
+  }
+#endif  // ROCKSDB_LITE
+
   RecordTick(statistics_, NUMBER_DB_SEEK);
   if (iter_->Valid()) {
     if (prefix_extractor_ && prefix_same_as_start_) {
@@ -1453,11 +1471,12 @@ Iterator* NewDBIterator(Env* env, const ReadOptions& read_options,
                         InternalIterator* internal_iter,
                         const SequenceNumber& sequence,
                         uint64_t max_sequential_skip_in_iterations,
-                        ReadCallback* read_callback, bool allow_blob) {
-  DBIter* db_iter =
-      new DBIter(env, read_options, cf_options, mutable_cf_options,
-                 user_key_comparator, internal_iter, sequence, false,
-                 max_sequential_skip_in_iterations, read_callback, allow_blob);
+                        ReadCallback* read_callback, DBImpl* db_impl,
+                        ColumnFamilyData* cfd, bool allow_blob) {
+  DBIter* db_iter = new DBIter(
+      env, read_options, cf_options, mutable_cf_options, user_key_comparator,
+      internal_iter, sequence, false, max_sequential_skip_in_iterations,
+      read_callback, db_impl, cfd, allow_blob);
   return db_iter;
 }
 
@@ -1504,13 +1523,14 @@ void ArenaWrappedDBIter::Init(Env* env, const ReadOptions& read_options,
                               const SequenceNumber& sequence,
                               uint64_t max_sequential_skip_in_iteration,
                               uint64_t version_number,
-                              ReadCallback* read_callback, bool allow_blob,
+                              ReadCallback* read_callback, DBImpl* db_impl,
+                              ColumnFamilyData* cfd, bool allow_blob,
                               bool allow_refresh) {
   auto mem = arena_.AllocateAligned(sizeof(DBIter));
-  db_iter_ = new (mem)
-      DBIter(env, read_options, cf_options, mutable_cf_options,
-             cf_options.user_comparator, nullptr, sequence, true,
-             max_sequential_skip_in_iteration, read_callback, allow_blob);
+  db_iter_ = new (mem) DBIter(env, read_options, cf_options, mutable_cf_options,
+                              cf_options.user_comparator, nullptr, sequence,
+                              true, max_sequential_skip_in_iteration,
+                              read_callback, db_impl, cfd, allow_blob);
   sv_number_ = version_number;
   allow_refresh_ = allow_refresh;
 }
@@ -1534,7 +1554,8 @@ Status ArenaWrappedDBIter::Refresh() {
     SuperVersion* sv = cfd_->GetReferencedSuperVersion(db_impl_->mutex());
     Init(env, read_options_, *(cfd_->ioptions()), sv->mutable_cf_options,
          latest_seq, sv->mutable_cf_options.max_sequential_skip_in_iterations,
-         cur_sv_number, read_callback_, allow_blob_, allow_refresh_);
+         cur_sv_number, read_callback_, db_impl_, cfd_, allow_blob_,
+         allow_refresh_);
 
     InternalIterator* internal_iter = db_impl_->NewInternalIterator(
         read_options_, cfd_, sv, &arena_, db_iter_->GetRangeDelAggregator());
@@ -1556,7 +1577,7 @@ ArenaWrappedDBIter* NewArenaWrappedDbIterator(
   ArenaWrappedDBIter* iter = new ArenaWrappedDBIter();
   iter->Init(env, read_options, cf_options, mutable_cf_options, sequence,
              max_sequential_skip_in_iterations, version_number, read_callback,
-             allow_blob, allow_refresh);
+             db_impl, cfd, allow_blob, allow_refresh);
   if (db_impl != nullptr && cfd != nullptr && allow_refresh) {
     iter->StoreRefreshInfo(read_options, db_impl, cfd, read_callback,
                            allow_blob);

db/db_iter.h

@@ -23,20 +23,18 @@ namespace rocksdb {
 
 class Arena;
 class DBIter;
-class InternalIterator;
 
 // Return a new iterator that converts internal keys (yielded by
 // "*internal_iter") that were live at the specified "sequence" number
 // into appropriate user keys.
-extern Iterator* NewDBIterator(Env* env, const ReadOptions& read_options,
-                               const ImmutableCFOptions& cf_options,
-                               const MutableCFOptions& mutable_cf_options,
-                               const Comparator* user_key_comparator,
-                               InternalIterator* internal_iter,
-                               const SequenceNumber& sequence,
-                               uint64_t max_sequential_skip_in_iterations,
-                               ReadCallback* read_callback,
-                               bool allow_blob = false);
+extern Iterator* NewDBIterator(
+    Env* env, const ReadOptions& read_options,
+    const ImmutableCFOptions& cf_options,
+    const MutableCFOptions& mutable_cf_options,
+    const Comparator* user_key_comparator, InternalIterator* internal_iter,
+    const SequenceNumber& sequence, uint64_t max_sequential_skip_in_iterations,
+    ReadCallback* read_callback, DBImpl* db_impl = nullptr,
+    ColumnFamilyData* cfd = nullptr, bool allow_blob = false);
 
 // A wrapper iterator which wraps DB Iterator and the arena, with which the DB
 // iterator is supposed be allocated. This class is used as an entry point of
@@ -75,7 +73,8 @@ class ArenaWrappedDBIter : public Iterator {
             const MutableCFOptions& mutable_cf_options,
             const SequenceNumber& sequence,
             uint64_t max_sequential_skip_in_iterations, uint64_t version_number,
-            ReadCallback* read_callback, bool allow_blob, bool allow_refresh);
+            ReadCallback* read_callback, DBImpl* db_impl, ColumnFamilyData* cfd,
+            bool allow_blob, bool allow_refresh);
 
   void StoreRefreshInfo(const ReadOptions& read_options, DBImpl* db_impl,
                         ColumnFamilyData* cfd, ReadCallback* read_callback,

db/db_iterator_test.cc

@@ -184,73 +184,6 @@ TEST_P(DBIteratorTest, NonBlockingIteration) {
                          kSkipMmapReads));
 }
 
-#ifndef ROCKSDB_LITE
-TEST_P(DBIteratorTest, ManagedNonBlockingIteration) {
-  do {
-    ReadOptions non_blocking_opts, regular_opts;
-    Options options = CurrentOptions();
-    options.statistics = rocksdb::CreateDBStatistics();
-    non_blocking_opts.read_tier = kBlockCacheTier;
-    non_blocking_opts.managed = true;
-    CreateAndReopenWithCF({"pikachu"}, options);
-    // write one kv to the database.
-    ASSERT_OK(Put(1, "a", "b"));
-
-    // scan using non-blocking iterator. We should find it because
-    // it is in memtable.
-    Iterator* iter = NewIterator(non_blocking_opts, handles_[1]);
-    int count = 0;
-    for (iter->SeekToFirst(); iter->Valid(); iter->Next()) {
-      ASSERT_OK(iter->status());
-      count++;
-    }
-    ASSERT_EQ(count, 1);
-    delete iter;
-
-    // flush memtable to storage. Now, the key should not be in the
-    // memtable neither in the block cache.
-    ASSERT_OK(Flush(1));
-
-    // verify that a non-blocking iterator does not find any
-    // kvs. Neither does it do any IOs to storage.
-    int64_t numopen = TestGetTickerCount(options, NO_FILE_OPENS);
-    int64_t cache_added = TestGetTickerCount(options, BLOCK_CACHE_ADD);
-    iter = NewIterator(non_blocking_opts, handles_[1]);
-    count = 0;
-    for (iter->SeekToFirst(); iter->Valid(); iter->Next()) {
-      count++;
-    }
-    ASSERT_EQ(count, 0);
-    ASSERT_TRUE(iter->status().IsIncomplete());
-    ASSERT_EQ(numopen, TestGetTickerCount(options, NO_FILE_OPENS));
-    ASSERT_EQ(cache_added, TestGetTickerCount(options, BLOCK_CACHE_ADD));
-    delete iter;
-
-    // read in the specified block via a regular get
-    ASSERT_EQ(Get(1, "a"), "b");
-
-    // verify that we can find it via a non-blocking scan
-    numopen = TestGetTickerCount(options, NO_FILE_OPENS);
-    cache_added = TestGetTickerCount(options, BLOCK_CACHE_ADD);
-    iter = NewIterator(non_blocking_opts, handles_[1]);
-    count = 0;
-    for (iter->SeekToFirst(); iter->Valid(); iter->Next()) {
-      ASSERT_OK(iter->status());
-      count++;
-    }
-    ASSERT_EQ(count, 1);
-    ASSERT_EQ(numopen, TestGetTickerCount(options, NO_FILE_OPENS));
-    ASSERT_EQ(cache_added, TestGetTickerCount(options, BLOCK_CACHE_ADD));
-    delete iter;
-
-    // This test verifies block cache behaviors, which is not used by plain
-    // table format.
-    // Exclude kHashCuckoo as it does not support iteration currently
-  } while (ChangeOptions(kSkipPlainTable | kSkipNoSeekToLast | kSkipHashCuckoo |
-                         kSkipMmapReads));
-}
-#endif  // ROCKSDB_LITE
-
 TEST_P(DBIteratorTest, IterSeekBeforePrev) {
   ASSERT_OK(Put("a", "b"));
   ASSERT_OK(Put("c", "d"));
@@ -2160,6 +2093,34 @@ TEST_P(DBIteratorTest, Refresh) {
   iter.reset();
 }
 
+TEST_P(DBIteratorTest, RefreshWithSnapshot) {
+  ASSERT_OK(Put("x", "y"));
+  const Snapshot* snapshot = db_->GetSnapshot();
+  ReadOptions options;
+  options.snapshot = snapshot;
+  Iterator* iter = NewIterator(options);
+
+  iter->Seek(Slice("a"));
+  ASSERT_TRUE(iter->Valid());
+  ASSERT_EQ(iter->key().compare(Slice("x")), 0);
+  iter->Next();
+  ASSERT_FALSE(iter->Valid());
+
+  ASSERT_OK(Put("c", "d"));
+
+  iter->Seek(Slice("a"));
+  ASSERT_TRUE(iter->Valid());
+  ASSERT_EQ(iter->key().compare(Slice("x")), 0);
+  iter->Next();
+  ASSERT_FALSE(iter->Valid());
+
+  Status s;
+  s = iter->Refresh();
+  ASSERT_TRUE(s.IsNotSupported());
+  db_->ReleaseSnapshot(snapshot);
+  delete iter;
+}
+
 TEST_P(DBIteratorTest, CreationFailure) {
   SyncPoint::GetInstance()->SetCallBack(
       "DBImpl::NewInternalIterator:StatusCallback", [](void* arg) {
@@ -2480,6 +2441,30 @@ TEST_P(DBIteratorTest, NonBlockingIterationBugRepro) {
   EXPECT_TRUE(iter->status().IsIncomplete());
 }
 
+TEST_P(DBIteratorTest, SeekBackwardAfterOutOfUpperBound) {
+  Put("a", "");
+  Put("b", "");
+  Flush();
+
+  ReadOptions ropt;
+  Slice ub = "b";
+  ropt.iterate_upper_bound = &ub;
+
+  std::unique_ptr<Iterator> it(dbfull()->NewIterator(ropt));
+  it->SeekForPrev("a");
+  ASSERT_TRUE(it->Valid());
+  ASSERT_OK(it->status());
+  ASSERT_EQ("a", it->key().ToString());
+  it->Next();
+  ASSERT_FALSE(it->Valid());
+  ASSERT_OK(it->status());
+  it->SeekForPrev("a");
+  ASSERT_OK(it->status());
+
+  ASSERT_TRUE(it->Valid());
+  ASSERT_EQ("a", it->key().ToString());
+}
+
 INSTANTIATE_TEST_CASE_P(DBIteratorTestInstance, DBIteratorTest,
                         testing::Values(true, false));
 

db/db_properties_test.cc

@@ -180,17 +180,16 @@ void ParseTablePropertiesString(std::string tp_string, TableProperties* tp) {
   ResetTableProperties(tp);
   sscanf(tp_string.c_str(),
          "# data blocks %" SCNu64 " # entries %" SCNu64
-         " # range deletions %" SCNu64
-         " raw key size %" SCNu64
+         " # range deletions %" SCNu64 " raw key size %" SCNu64
          " raw average key size %lf "
          " raw value size %" SCNu64
          " raw average value size %lf "
          " data block size %" SCNu64 " index block size (user-key? %" SCNu64
-         ") %" SCNu64 " filter block size %" SCNu64,
+         ", delta-value? %" SCNu64 ") %" SCNu64 " filter block size %" SCNu64,
          &tp->num_data_blocks, &tp->num_entries, &tp->num_range_deletions,
          &tp->raw_key_size, &dummy_double, &tp->raw_value_size, &dummy_double,
-         &tp->data_size, &tp->index_key_is_user_key, &tp->index_size,
-         &tp->filter_size);
+         &tp->data_size, &tp->index_key_is_user_key,
+         &tp->index_value_is_delta_encoded, &tp->index_size, &tp->filter_size);
 }
 
 void VerifySimilar(uint64_t a, uint64_t b, double bias) {
@@ -224,14 +223,11 @@ void VerifyTableProperties(const TableProperties& base_tp,
   ASSERT_EQ(base_tp.num_range_deletions, new_tp.num_range_deletions);
 }
 
-void GetExpectedTableProperties(TableProperties* expected_tp,
-                                const int kKeySize, const int kValueSize,
-                                const int kKeysPerTable,
-                                const int kRangeDeletionsPerTable,
-                                const int kTableCount,
-                                const int kBloomBitsPerKey,
-                                const size_t kBlockSize,
-                                const bool index_key_is_user_key) {
+void GetExpectedTableProperties(
+    TableProperties* expected_tp, const int kKeySize, const int kValueSize,
+    const int kKeysPerTable, const int kRangeDeletionsPerTable,
+    const int kTableCount, const int kBloomBitsPerKey, const size_t kBlockSize,
+    const bool index_key_is_user_key, const bool value_delta_encoding) {
   const int kKeyCount = kTableCount * kKeysPerTable;
   const int kRangeDeletionCount = kTableCount * kRangeDeletionsPerTable;
   const int kAvgSuccessorSize = kKeySize / 5;
@@ -248,7 +244,9 @@ void GetExpectedTableProperties(TableProperties* expected_tp,
       kTableCount * (kKeysPerTable * (kKeySize + 8 + kValueSize));
   expected_tp->index_size =
       expected_tp->num_data_blocks *
-      (kAvgSuccessorSize + (index_key_is_user_key ? 0 : 8));
+      (kAvgSuccessorSize + (index_key_is_user_key ? 0 : 8) -
+       // discount 1 byte as value size is not encoded in value delta encoding
+       (value_delta_encoding ? 1 : 0));
   expected_tp->filter_size =
       kTableCount * (kKeysPerTable * kBloomBitsPerKey / 8);
 }
@@ -342,12 +340,14 @@ TEST_F(DBPropertiesTest, AggregatedTableProperties) {
     TableProperties output_tp;
     ParseTablePropertiesString(property, &output_tp);
     bool index_key_is_user_key = output_tp.index_key_is_user_key > 0;
+    bool value_is_delta_encoded = output_tp.index_value_is_delta_encoded > 0;
 
     TableProperties expected_tp;
     GetExpectedTableProperties(&expected_tp, kKeySize, kValueSize,
                                kKeysPerTable, kRangeDeletionsPerTable,
                                kTableCount, kBloomBitsPerKey,
-                               table_options.block_size, index_key_is_user_key);
+                               table_options.block_size, index_key_is_user_key,
+                               value_is_delta_encoded);
 
     VerifyTableProperties(expected_tp, output_tp);
   }
@@ -533,6 +533,7 @@ TEST_F(DBPropertiesTest, AggregatedTablePropertiesAtLevel) {
     db_->GetProperty(DB::Properties::kAggregatedTableProperties, &tp_string);
     ParseTablePropertiesString(tp_string, &tp);
     bool index_key_is_user_key = tp.index_key_is_user_key > 0;
+    bool value_is_delta_encoded = tp.index_value_is_delta_encoded > 0;
     ASSERT_EQ(sum_tp.data_size, tp.data_size);
     ASSERT_EQ(sum_tp.index_size, tp.index_size);
     ASSERT_EQ(sum_tp.filter_size, tp.filter_size);
@@ -542,10 +543,10 @@ TEST_F(DBPropertiesTest, AggregatedTablePropertiesAtLevel) {
     ASSERT_EQ(sum_tp.num_entries, tp.num_entries);
     ASSERT_EQ(sum_tp.num_range_deletions, tp.num_range_deletions);
     if (table > 3) {
-      GetExpectedTableProperties(
-          &expected_tp, kKeySize, kValueSize, kKeysPerTable,
-          kRangeDeletionsPerTable, table, kBloomBitsPerKey,
-          table_options.block_size, index_key_is_user_key);
+      GetExpectedTableProperties(&expected_tp, kKeySize, kValueSize,
+                                 kKeysPerTable, kRangeDeletionsPerTable, table,
+                                 kBloomBitsPerKey, table_options.block_size,
+                                 index_key_is_user_key, value_is_delta_encoded);
       // Gives larger bias here as index block size, filter block size,
       // and data block size become much harder to estimate in this test.
       VerifyTableProperties(expected_tp, tp, 0.5, 0.4, 0.4, 0.25);

db/db_range_del_test.cc

@@ -191,7 +191,7 @@ TEST_F(DBRangeDelTest, SentinelsOmittedFromOutputFile) {
 
   std::vector<std::vector<FileMetaData>> files;
   dbfull()->TEST_GetFilesMetaData(db_->DefaultColumnFamily(), &files);
-  ASSERT_GT(files[0][0].smallest_seqno, 0);
+  ASSERT_GT(files[0][0].fd.smallest_seqno, 0);
 
   db_->ReleaseSnapshot(snapshot);
 }

db/db_sst_test.cc

@@ -436,7 +436,7 @@ TEST_F(DBSSTTest, OpenDBWithExistingTrash) {
 // deleted from first db_path were deleted using DeleteScheduler and
 // files in the second path were not.
 TEST_F(DBSSTTest, DeleteSchedulerMultipleDBPaths) {
-  int bg_delete_file = 0;
+  std::atomic<int> bg_delete_file(0);
   rocksdb::SyncPoint::GetInstance()->SetCallBack(
       "DeleteScheduler::DeleteTrashFile:DeleteFile",
       [&](void* /*arg*/) { bg_delete_file++; });
@@ -703,26 +703,19 @@ TEST_F(DBSSTTest, DBWithMaxSpaceAllowedRandomized) {
   // When bg_error_ is set we will verify that the DB size is greater
   // than the limit.
 
-  std::vector<int> max_space_limits_mbs = {1, 2, 4, 8, 10};
-  decltype(max_space_limits_mbs)::value_type limit_mb_cb;
-  bool bg_error_set = false;
-  uint64_t total_sst_files_size = 0;
+  std::vector<int> max_space_limits_mbs = {1, 10};
+  std::atomic<bool> bg_error_set(false);
 
-  std::atomic<int> estimate_multiplier(1);
-  int reached_max_space_on_flush = 0;
-  int reached_max_space_on_compaction = 0;
+  std::atomic<int> reached_max_space_on_flush(0);
+  std::atomic<int> reached_max_space_on_compaction(0);
   rocksdb::SyncPoint::GetInstance()->SetCallBack(
       "DBImpl::FlushMemTableToOutputFile:MaxAllowedSpaceReached",
       [&](void* arg) {
         Status* bg_error = static_cast<Status*>(arg);
         bg_error_set = true;
-        GetAllSSTFiles(&total_sst_files_size);
         reached_max_space_on_flush++;
-        // low limit for size calculated using sst files
-        ASSERT_GE(total_sst_files_size, limit_mb_cb * 1024 * 1024);
         // clear error to ensure compaction callback is called
         *bg_error = Status::OK();
-        estimate_multiplier++;  // used in the main loop assert
       });
 
   rocksdb::SyncPoint::GetInstance()->SetCallBack(
@@ -735,15 +728,11 @@ TEST_F(DBSSTTest, DBWithMaxSpaceAllowedRandomized) {
       "CompactionJob::FinishCompactionOutputFile:MaxAllowedSpaceReached",
       [&](void* /*arg*/) {
         bg_error_set = true;
-        GetAllSSTFiles(&total_sst_files_size);
         reached_max_space_on_compaction++;
       });
 
   for (auto limit_mb : max_space_limits_mbs) {
     bg_error_set = false;
-    total_sst_files_size = 0;
-    estimate_multiplier = 1;
-    limit_mb_cb = limit_mb;
     rocksdb::SyncPoint::GetInstance()->ClearTrace();
     rocksdb::SyncPoint::GetInstance()->EnableProcessing();
     std::shared_ptr<SstFileManager> sst_file_manager(NewSstFileManager(env_));
@@ -757,21 +746,17 @@ TEST_F(DBSSTTest, DBWithMaxSpaceAllowedRandomized) {
 
     sfm->SetMaxAllowedSpaceUsage(limit_mb * 1024 * 1024);
 
-    int keys_written = 0;
-    uint64_t estimated_db_size = 0;
+    // It is easy to detect if the test is stuck in a loop. No need for
+    // complex termination logic.
     while (true) {
       auto s = Put(RandomString(&rnd, 10), RandomString(&rnd, 50));
       if (!s.ok()) {
         break;
       }
-      keys_written++;
-      // Check the estimated db size vs the db limit just to make sure we
-      // dont run into an infinite loop
-      estimated_db_size = keys_written * 60;  // ~60 bytes per key
-      ASSERT_LT(estimated_db_size,
-                estimate_multiplier * limit_mb * 1024 * 1024 * 2);
     }
     ASSERT_TRUE(bg_error_set);
+    uint64_t total_sst_files_size = 0;
+    GetAllSSTFiles(&total_sst_files_size);
     ASSERT_GE(total_sst_files_size, limit_mb * 1024 * 1024);
     rocksdb::SyncPoint::GetInstance()->DisableProcessing();
   }

db/db_tailing_iter_test.cc

@@ -479,275 +479,6 @@ TEST_F(DBTestTailingIterator, TailingIteratorGap) {
   ASSERT_EQ("40", it->key().ToString());
 }
 
-TEST_F(DBTestTailingIterator, ManagedTailingIteratorSingle) {
-  ReadOptions read_options;
-  read_options.tailing = true;
-  read_options.managed = true;
-
-  std::unique_ptr<Iterator> iter(db_->NewIterator(read_options));
-  iter->SeekToFirst();
-  ASSERT_TRUE(!iter->Valid());
-
-  // add a record and check that iter can see it
-  ASSERT_OK(db_->Put(WriteOptions(), "mirko", "fodor"));
-  iter->SeekToFirst();
-  ASSERT_TRUE(iter->Valid());
-  ASSERT_EQ(iter->key().ToString(), "mirko");
-
-  iter->Next();
-  ASSERT_TRUE(!iter->Valid());
-}
-
-TEST_F(DBTestTailingIterator, ManagedTailingIteratorKeepAdding) {
-  CreateAndReopenWithCF({"pikachu"}, CurrentOptions());
-  ReadOptions read_options;
-  read_options.tailing = true;
-  read_options.managed = true;
-
-  std::unique_ptr<Iterator> iter(db_->NewIterator(read_options, handles_[1]));
-  std::string value(1024, 'a');
-
-  const int num_records = 10000;
-  for (int i = 0; i < num_records; ++i) {
-    char buf[32];
-    snprintf(buf, sizeof(buf), "%016d", i);
-
-    Slice key(buf, 16);
-    ASSERT_OK(Put(1, key, value));
-
-    iter->Seek(key);
-    ASSERT_TRUE(iter->Valid());
-    ASSERT_EQ(iter->key().compare(key), 0);
-  }
-}
-
-TEST_F(DBTestTailingIterator, ManagedTailingIteratorSeekToNext) {
-  CreateAndReopenWithCF({"pikachu"}, CurrentOptions());
-  ReadOptions read_options;
-  read_options.tailing = true;
-  read_options.managed = true;
-
-  std::unique_ptr<Iterator> iter(db_->NewIterator(read_options, handles_[1]));
-  std::string value(1024, 'a');
-
-  const int num_records = 1000;
-  for (int i = 1; i < num_records; ++i) {
-    char buf1[32];
-    char buf2[32];
-    snprintf(buf1, sizeof(buf1), "00a0%016d", i * 5);
-
-    Slice key(buf1, 20);
-    ASSERT_OK(Put(1, key, value));
-
-    if (i % 100 == 99) {
-      ASSERT_OK(Flush(1));
-    }
-
-    snprintf(buf2, sizeof(buf2), "00a0%016d", i * 5 - 2);
-    Slice target(buf2, 20);
-    iter->Seek(target);
-    ASSERT_TRUE(iter->Valid());
-    ASSERT_EQ(iter->key().compare(key), 0);
-  }
-  for (int i = 2 * num_records; i > 0; --i) {
-    char buf1[32];
-    char buf2[32];
-    snprintf(buf1, sizeof(buf1), "00a0%016d", i * 5);
-
-    Slice key(buf1, 20);
-    ASSERT_OK(Put(1, key, value));
-
-    if (i % 100 == 99) {
-      ASSERT_OK(Flush(1));
-    }
-
-    snprintf(buf2, sizeof(buf2), "00a0%016d", i * 5 - 2);
-    Slice target(buf2, 20);
-    iter->Seek(target);
-    ASSERT_TRUE(iter->Valid());
-    ASSERT_EQ(iter->key().compare(key), 0);
-  }
-}
-
-TEST_F(DBTestTailingIterator, ManagedTailingIteratorDeletes) {
-  CreateAndReopenWithCF({"pikachu"}, CurrentOptions());
-  ReadOptions read_options;
-  read_options.tailing = true;
-  read_options.managed = true;
-
-  std::unique_ptr<Iterator> iter(db_->NewIterator(read_options, handles_[1]));
-
-  // write a single record, read it using the iterator, then delete it
-  ASSERT_OK(Put(1, "0test", "test"));
-  iter->SeekToFirst();
-  ASSERT_TRUE(iter->Valid());
-  ASSERT_EQ(iter->key().ToString(), "0test");
-  ASSERT_OK(Delete(1, "0test"));
-
-  // write many more records
-  const int num_records = 10000;
-  std::string value(1024, 'A');
-
-  for (int i = 0; i < num_records; ++i) {
-    char buf[32];
-    snprintf(buf, sizeof(buf), "1%015d", i);
-
-    Slice key(buf, 16);
-    ASSERT_OK(Put(1, key, value));
-  }
-
-  // force a flush to make sure that no records are read from memtable
-  ASSERT_OK(Flush(1));
-
-  // skip "0test"
-  iter->Next();
-
-  // make sure we can read all new records using the existing iterator
-  int count = 0;
-  for (; iter->Valid(); iter->Next(), ++count) {
-  }
-
-  ASSERT_EQ(count, num_records);
-}
-
-TEST_F(DBTestTailingIterator, ManagedTailingIteratorPrefixSeek) {
-  ReadOptions read_options;
-  read_options.tailing = true;
-  read_options.managed = true;
-
-  Options options = CurrentOptions();
-  options.create_if_missing = true;
-  options.disable_auto_compactions = true;
-  options.prefix_extractor.reset(NewFixedPrefixTransform(2));
-  options.memtable_factory.reset(NewHashSkipListRepFactory(16));
-  options.allow_concurrent_memtable_write = false;
-  DestroyAndReopen(options);
-  CreateAndReopenWithCF({"pikachu"}, options);
-
-  std::unique_ptr<Iterator> iter(db_->NewIterator(read_options, handles_[1]));
-  ASSERT_OK(Put(1, "0101", "test"));
-
-  ASSERT_OK(Flush(1));
-
-  ASSERT_OK(Put(1, "0202", "test"));
-
-  // Seek(0102) shouldn't find any records since 0202 has a different prefix
-  iter->Seek("0102");
-  ASSERT_TRUE(!iter->Valid());
-
-  iter->Seek("0202");
-  ASSERT_TRUE(iter->Valid());
-  ASSERT_EQ(iter->key().ToString(), "0202");
-
-  iter->Next();
-  ASSERT_TRUE(!iter->Valid());
-}
-
-TEST_F(DBTestTailingIterator, ManagedTailingIteratorIncomplete) {
-  CreateAndReopenWithCF({"pikachu"}, CurrentOptions());
-  ReadOptions read_options;
-  read_options.tailing = true;
-  read_options.managed = true;
-  read_options.read_tier = kBlockCacheTier;
-
-  std::string key = "key";
-  std::string value = "value";
-
-  ASSERT_OK(db_->Put(WriteOptions(), key, value));
-
-  std::unique_ptr<Iterator> iter(db_->NewIterator(read_options));
-  iter->SeekToFirst();
-  // we either see the entry or it's not in cache
-  ASSERT_TRUE(iter->Valid() || iter->status().IsIncomplete());
-
-  ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
-  iter->SeekToFirst();
-  // should still be true after compaction
-  ASSERT_TRUE(iter->Valid() || iter->status().IsIncomplete());
-}
-
-TEST_F(DBTestTailingIterator, ManagedTailingIteratorSeekToSame) {
-  Options options = CurrentOptions();
-  options.compaction_style = kCompactionStyleUniversal;
-  options.write_buffer_size = 1000;
-  CreateAndReopenWithCF({"pikachu"}, options);
-
-  ReadOptions read_options;
-  read_options.tailing = true;
-  read_options.managed = true;
-
-  const int NROWS = 10000;
-  // Write rows with keys 00000, 00002, 00004 etc.
-  for (int i = 0; i < NROWS; ++i) {
-    char buf[100];
-    snprintf(buf, sizeof(buf), "%05d", 2 * i);
-    std::string key(buf);
-    std::string value("value");
-    ASSERT_OK(db_->Put(WriteOptions(), key, value));
-  }
-
-  std::unique_ptr<Iterator> iter(db_->NewIterator(read_options));
-  // Seek to 00001.  We expect to find 00002.
-  std::string start_key = "00001";
-  iter->Seek(start_key);
-  ASSERT_TRUE(iter->Valid());
-
-  std::string found = iter->key().ToString();
-  ASSERT_EQ("00002", found);
-
-  // Now seek to the same key.  The iterator should remain in the same
-  // position.
-  iter->Seek(found);
-  ASSERT_TRUE(iter->Valid());
-  ASSERT_EQ(found, iter->key().ToString());
-}
-
-TEST_F(DBTestTailingIterator, ForwardIteratorVersionProperty) {
-  Options options = CurrentOptions();
-  options.write_buffer_size = 1000;
-
-  ReadOptions read_options;
-  read_options.tailing = true;
-
-  Put("foo", "bar");
-
-  uint64_t v1, v2, v3, v4;
-  {
-    std::unique_ptr<Iterator> iter(db_->NewIterator(read_options));
-    iter->Seek("foo");
-    std::string prop_value;
-    ASSERT_OK(iter->GetProperty("rocksdb.iterator.super-version-number",
-                                &prop_value));
-    v1 = static_cast<uint64_t>(std::atoi(prop_value.c_str()));
-
-    Put("foo1", "bar1");
-    Flush();
-
-    ASSERT_OK(iter->GetProperty("rocksdb.iterator.super-version-number",
-                                &prop_value));
-    v2 = static_cast<uint64_t>(std::atoi(prop_value.c_str()));
-
-    iter->Seek("f");
-
-    ASSERT_OK(iter->GetProperty("rocksdb.iterator.super-version-number",
-                                &prop_value));
-    v3 = static_cast<uint64_t>(std::atoi(prop_value.c_str()));
-
-    ASSERT_EQ(v1, v2);
-    ASSERT_GT(v3, v2);
-  }
-
-  {
-    std::unique_ptr<Iterator> iter(db_->NewIterator(read_options));
-    iter->Seek("foo");
-    std::string prop_value;
-    ASSERT_OK(iter->GetProperty("rocksdb.iterator.super-version-number",
-                                &prop_value));
-    v4 = static_cast<uint64_t>(std::atoi(prop_value.c_str()));
-  }
-  ASSERT_EQ(v3, v4);
-}
-
 TEST_F(DBTestTailingIterator, SeekWithUpperBoundBug) {
   ReadOptions read_options;
   read_options.tailing = true;

db/db_test.cc

@@ -262,6 +262,196 @@ TEST_F(DBTest, SkipDelay) {
   }
 }
 
+TEST_F(DBTest, MixedSlowdownOptions) {
+  Options options = CurrentOptions();
+  options.env = env_;
+  options.write_buffer_size = 100000;
+  CreateAndReopenWithCF({"pikachu"}, options);
+  std::vector<port::Thread> threads;
+  std::atomic<int> thread_num(0);
+
+  std::function<void()> write_slowdown_func = [&]() {
+    int a = thread_num.fetch_add(1);
+    std::string key = "foo" + std::to_string(a);
+    WriteOptions wo;
+    wo.no_slowdown = false;
+    ASSERT_OK(dbfull()->Put(wo, key, "bar"));
+  };
+  std::function<void()> write_no_slowdown_func = [&]() {
+    int a = thread_num.fetch_add(1);
+    std::string key = "foo" + std::to_string(a);
+    WriteOptions wo;
+    wo.no_slowdown = true;
+    ASSERT_NOK(dbfull()->Put(wo, key, "bar"));
+  };
+  // Use a small number to ensure a large delay that is still effective
+  // when we do Put
+  // TODO(myabandeh): this is time dependent and could potentially make
+  // the test flaky
+  auto token = dbfull()->TEST_write_controler().GetDelayToken(1);
+  std::atomic<int> sleep_count(0);
+  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+      "DBImpl::DelayWrite:BeginWriteStallDone",
+      [&](void* /*arg*/) {
+        sleep_count.fetch_add(1);
+        if (threads.empty()) {
+          for (int i = 0; i < 2; ++i) {
+            threads.emplace_back(write_slowdown_func);
+          }
+          for (int i = 0; i < 2; ++i) {
+            threads.emplace_back(write_no_slowdown_func);
+          }
+        }
+      });
+  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+
+  WriteOptions wo;
+  wo.sync = false;
+  wo.disableWAL = false;
+  wo.no_slowdown = false;
+  dbfull()->Put(wo, "foo", "bar");
+  // We need the 2nd write to trigger delay. This is because delay is
+  // estimated based on the last write size which is 0 for the first write.
+  ASSERT_OK(dbfull()->Put(wo, "foo2", "bar2"));
+          token.reset();
+
+  for (auto& t : threads) {
+    t.join();
+  }
+  ASSERT_GE(sleep_count.load(), 1);
+
+  wo.no_slowdown = true;
+  ASSERT_OK(dbfull()->Put(wo, "foo3", "bar"));
+}
+
+TEST_F(DBTest, MixedSlowdownOptionsInQueue) {
+  Options options = CurrentOptions();
+  options.env = env_;
+  options.write_buffer_size = 100000;
+  CreateAndReopenWithCF({"pikachu"}, options);
+  std::vector<port::Thread> threads;
+  std::atomic<int> thread_num(0);
+
+  std::function<void()> write_no_slowdown_func = [&]() {
+    int a = thread_num.fetch_add(1);
+    std::string key = "foo" + std::to_string(a);
+    WriteOptions wo;
+    wo.no_slowdown = true;
+    ASSERT_NOK(dbfull()->Put(wo, key, "bar"));
+  };
+  // Use a small number to ensure a large delay that is still effective
+  // when we do Put
+  // TODO(myabandeh): this is time dependent and could potentially make
+  // the test flaky
+  auto token = dbfull()->TEST_write_controler().GetDelayToken(1);
+  std::atomic<int> sleep_count(0);
+  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+      "DBImpl::DelayWrite:Sleep",
+      [&](void* /*arg*/) {
+        sleep_count.fetch_add(1);
+        if (threads.empty()) {
+          for (int i = 0; i < 2; ++i) {
+            threads.emplace_back(write_no_slowdown_func);
+          }
+          // Sleep for 2s to allow the threads to insert themselves into the
+          // write queue
+          env_->SleepForMicroseconds(3000000ULL);
+        }
+      });
+  std::atomic<int> wait_count(0);
+  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+      "DBImpl::DelayWrite:Wait",
+      [&](void* /*arg*/) { wait_count.fetch_add(1); });
+  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+
+  WriteOptions wo;
+  wo.sync = false;
+  wo.disableWAL = false;
+  wo.no_slowdown = false;
+  dbfull()->Put(wo, "foo", "bar");
+  // We need the 2nd write to trigger delay. This is because delay is
+  // estimated based on the last write size which is 0 for the first write.
+  ASSERT_OK(dbfull()->Put(wo, "foo2", "bar2"));
+          token.reset();
+
+  for (auto& t : threads) {
+    t.join();
+  }
+  ASSERT_EQ(sleep_count.load(), 1);
+  ASSERT_GE(wait_count.load(), 0);
+}
+
+TEST_F(DBTest, MixedSlowdownOptionsStop) {
+  Options options = CurrentOptions();
+  options.env = env_;
+  options.write_buffer_size = 100000;
+  CreateAndReopenWithCF({"pikachu"}, options);
+  std::vector<port::Thread> threads;
+  std::atomic<int> thread_num(0);
+
+  std::function<void()> write_slowdown_func = [&]() {
+    int a = thread_num.fetch_add(1);
+    std::string key = "foo" + std::to_string(a);
+    WriteOptions wo;
+    wo.no_slowdown = false;
+    ASSERT_OK(dbfull()->Put(wo, key, "bar"));
+  };
+  std::function<void()> write_no_slowdown_func = [&]() {
+    int a = thread_num.fetch_add(1);
+    std::string key = "foo" + std::to_string(a);
+    WriteOptions wo;
+    wo.no_slowdown = true;
+    ASSERT_NOK(dbfull()->Put(wo, key, "bar"));
+  };
+  std::function<void()> wakeup_writer = [&]() {
+    dbfull()->mutex_.Lock();
+    dbfull()->bg_cv_.SignalAll();
+    dbfull()->mutex_.Unlock();
+  };
+  // Use a small number to ensure a large delay that is still effective
+  // when we do Put
+  // TODO(myabandeh): this is time dependent and could potentially make
+  // the test flaky
+  auto token = dbfull()->TEST_write_controler().GetStopToken();
+  std::atomic<int> wait_count(0);
+  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+      "DBImpl::DelayWrite:Wait",
+      [&](void* /*arg*/) {
+        wait_count.fetch_add(1);
+        if (threads.empty()) {
+          for (int i = 0; i < 2; ++i) {
+            threads.emplace_back(write_slowdown_func);
+          }
+          for (int i = 0; i < 2; ++i) {
+            threads.emplace_back(write_no_slowdown_func);
+          }
+          // Sleep for 2s to allow the threads to insert themselves into the
+          // write queue
+          env_->SleepForMicroseconds(3000000ULL);
+        }
+        token.reset();
+        threads.emplace_back(wakeup_writer);
+      });
+  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+
+  WriteOptions wo;
+  wo.sync = false;
+  wo.disableWAL = false;
+  wo.no_slowdown = false;
+  dbfull()->Put(wo, "foo", "bar");
+  // We need the 2nd write to trigger delay. This is because delay is
+  // estimated based on the last write size which is 0 for the first write.
+  ASSERT_OK(dbfull()->Put(wo, "foo2", "bar2"));
+          token.reset();
+
+  for (auto& t : threads) {
+    t.join();
+  }
+  ASSERT_GE(wait_count.load(), 1);
+
+  wo.no_slowdown = true;
+  ASSERT_OK(dbfull()->Put(wo, "foo3", "bar"));
+}
 #ifndef ROCKSDB_LITE
 
 TEST_F(DBTest, LevelLimitReopen) {
@@ -2149,6 +2339,9 @@ static void GCThreadBody(void* arg) {
 
 }  // namespace
 
+#ifndef TRAVIS
+// Disable this test temporarily on Travis as it fails intermittently.
+// Github issue: #4151
 TEST_F(DBTest, GroupCommitTest) {
   do {
     Options options = CurrentOptions();
@@ -2195,6 +2388,7 @@ TEST_F(DBTest, GroupCommitTest) {
     ASSERT_GT(hist_data.average, 0.0);
   } while (ChangeOptions(kSkipNoSeekToLast));
 }
+#endif  // TRAVIS
 
 namespace {
 typedef std::map<std::string, std::string> KVMap;
@@ -4327,7 +4521,7 @@ TEST_F(DBTest, DynamicCompactionOptions) {
   // Clean up memtable and L0. Block compaction threads. If continue to write
   // and flush memtables. We should see put stop after 8 memtable flushes
   // since level0_stop_writes_trigger = 8
-  dbfull()->TEST_FlushMemTable(true);
+  dbfull()->TEST_FlushMemTable(true, true);
   dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr);
   // Block compaction
   test::SleepingBackgroundTask sleeping_task_low;
@@ -4340,7 +4534,7 @@ TEST_F(DBTest, DynamicCompactionOptions) {
   WriteOptions wo;
   while (count < 64) {
     ASSERT_OK(Put(Key(count), RandomString(&rnd, 1024), wo));
-    dbfull()->TEST_FlushMemTable(true);
+    dbfull()->TEST_FlushMemTable(true, true);
     count++;
     if (dbfull()->TEST_write_controler().IsStopped()) {
       sleeping_task_low.WakeUp();
@@ -4368,7 +4562,7 @@ TEST_F(DBTest, DynamicCompactionOptions) {
   count = 0;
   while (count < 64) {
     ASSERT_OK(Put(Key(count), RandomString(&rnd, 1024), wo));
-    dbfull()->TEST_FlushMemTable(true);
+    dbfull()->TEST_FlushMemTable(true, true);
     count++;
     if (dbfull()->TEST_write_controler().IsStopped()) {
       sleeping_task_low.WakeUp();
@@ -5508,7 +5702,7 @@ TEST_F(DBTest, SoftLimit) {
   for (int i = 0; i < 72; i++) {
     Put(Key(i), std::string(5000, 'x'));
     if (i % 10 == 0) {
-      Flush();
+      dbfull()->TEST_FlushMemTable(true, true);
     }
   }
   dbfull()->TEST_WaitForCompact();
@@ -5518,7 +5712,7 @@ TEST_F(DBTest, SoftLimit) {
   for (int i = 0; i < 72; i++) {
     Put(Key(i), std::string(5000, 'x'));
     if (i % 10 == 0) {
-      Flush();
+      dbfull()->TEST_FlushMemTable(true, true);
     }
   }
   dbfull()->TEST_WaitForCompact();
@@ -5537,7 +5731,7 @@ TEST_F(DBTest, SoftLimit) {
     Put(Key(i), std::string(5000, 'x'));
     Put(Key(100 - i), std::string(5000, 'x'));
     // Flush the file. File size is around 30KB.
-    Flush();
+    dbfull()->TEST_FlushMemTable(true, true);
   }
   ASSERT_TRUE(dbfull()->TEST_write_controler().NeedsDelay());
   ASSERT_TRUE(listener->CheckCondition(WriteStallCondition::kDelayed));
@@ -5572,7 +5766,7 @@ TEST_F(DBTest, SoftLimit) {
     Put(Key(10 + i), std::string(5000, 'x'));
     Put(Key(90 - i), std::string(5000, 'x'));
     // Flush the file. File size is around 30KB.
-    Flush();
+    dbfull()->TEST_FlushMemTable(true, true);
   }
 
   // Wake up sleep task to enable compaction to run and waits
@@ -5593,7 +5787,7 @@ TEST_F(DBTest, SoftLimit) {
     Put(Key(20 + i), std::string(5000, 'x'));
     Put(Key(80 - i), std::string(5000, 'x'));
     // Flush the file. File size is around 30KB.
-    Flush();
+    dbfull()->TEST_FlushMemTable(true, true);
   }
   // Wake up sleep task to enable compaction to run and waits
   // for it to go to sleep state again to make sure one compaction

db/db_test2.cc

@@ -2321,9 +2321,9 @@ TEST_F(DBTest2, RateLimitedCompactionReads) {
         options.rate_limiter->GetTotalBytesThrough(Env::IO_LOW);
     // Include the explicit prefetch of the footer in direct I/O case.
     size_t direct_io_extra = use_direct_io ? 512 * 1024 : 0;
-    ASSERT_GE(rate_limited_bytes,
-              static_cast<size_t>(kNumKeysPerFile * kBytesPerKey * kNumL0Files +
-                                  direct_io_extra));
+    ASSERT_GE(
+        rate_limited_bytes,
+        static_cast<size_t>(kNumKeysPerFile * kBytesPerKey * kNumL0Files));
     ASSERT_LT(
         rate_limited_bytes,
         static_cast<size_t>(2 * kNumKeysPerFile * kBytesPerKey * kNumL0Files +
@@ -2500,6 +2500,108 @@ TEST_F(DBTest2, LiveFilesOmitObsoleteFiles) {
   rocksdb::SyncPoint::GetInstance()->DisableProcessing();
 }
 
+TEST_F(DBTest2, TraceAndReplay) {
+  Options options = CurrentOptions();
+  options.merge_operator = MergeOperators::CreatePutOperator();
+  ReadOptions ro;
+  WriteOptions wo;
+  TraceOptions trace_opts;
+  EnvOptions env_opts;
+  CreateAndReopenWithCF({"pikachu"}, options);
+  Random rnd(301);
+  Iterator* single_iter = nullptr;
+
+  std::string trace_filename = dbname_ + "/rocksdb.trace";
+  std::unique_ptr<TraceWriter> trace_writer;
+  ASSERT_OK(NewFileTraceWriter(env_, env_opts, trace_filename, &trace_writer));
+  ASSERT_OK(db_->StartTrace(trace_opts, std::move(trace_writer)));
+
+  ASSERT_OK(Put(0, "a", "1"));
+  ASSERT_OK(Merge(0, "b", "2"));
+  ASSERT_OK(Delete(0, "c"));
+  ASSERT_OK(SingleDelete(0, "d"));
+  ASSERT_OK(db_->DeleteRange(wo, dbfull()->DefaultColumnFamily(), "e", "f"));
+
+  WriteBatch batch;
+  ASSERT_OK(batch.Put("f", "11"));
+  ASSERT_OK(batch.Merge("g", "12"));
+  ASSERT_OK(batch.Delete("h"));
+  ASSERT_OK(batch.SingleDelete("i"));
+  ASSERT_OK(batch.DeleteRange("j", "k"));
+  ASSERT_OK(db_->Write(wo, &batch));
+
+  single_iter = db_->NewIterator(ro);
+  single_iter->Seek("f");
+  single_iter->SeekForPrev("g");
+  delete single_iter;
+
+  ASSERT_EQ("1", Get(0, "a"));
+  ASSERT_EQ("12", Get(0, "g"));
+
+  ASSERT_OK(Put(1, "foo", "bar"));
+  ASSERT_OK(Put(1, "rocksdb", "rocks"));
+  ASSERT_EQ("NOT_FOUND", Get(1, "leveldb"));
+
+  ASSERT_OK(db_->EndTrace());
+  // These should not get into the trace file as it is after EndTrace.
+  Put("hello", "world");
+  Merge("foo", "bar");
+
+  // Open another db, replay, and verify the data
+  std::string value;
+  std::string dbname2 = test::TmpDir(env_) + "/db_replay";
+  ASSERT_OK(DestroyDB(dbname2, options));
+
+  // Using a different name than db2, to pacify infer's use-after-lifetime
+  // warnings (http://fbinfer.com).
+  DB* db2_init = nullptr;
+  options.create_if_missing = true;
+  ASSERT_OK(DB::Open(options, dbname2, &db2_init));
+  ColumnFamilyHandle* cf;
+  ASSERT_OK(
+      db2_init->CreateColumnFamily(ColumnFamilyOptions(), "pikachu", &cf));
+  delete cf;
+  delete db2_init;
+
+  DB* db2 = nullptr;
+  std::vector<ColumnFamilyDescriptor> column_families;
+  ColumnFamilyOptions cf_options;
+  cf_options.merge_operator = MergeOperators::CreatePutOperator();
+  column_families.push_back(ColumnFamilyDescriptor("default", cf_options));
+  column_families.push_back(
+      ColumnFamilyDescriptor("pikachu", ColumnFamilyOptions()));
+  std::vector<ColumnFamilyHandle*> handles;
+  ASSERT_OK(DB::Open(DBOptions(), dbname2, column_families, &handles, &db2));
+
+  env_->SleepForMicroseconds(100);
+  // Verify that the keys don't already exist
+  ASSERT_TRUE(db2->Get(ro, handles[0], "a", &value).IsNotFound());
+  ASSERT_TRUE(db2->Get(ro, handles[0], "g", &value).IsNotFound());
+
+  std::unique_ptr<TraceReader> trace_reader;
+  ASSERT_OK(NewFileTraceReader(env_, env_opts, trace_filename, &trace_reader));
+  Replayer replayer(db2, handles_, std::move(trace_reader));
+  ASSERT_OK(replayer.Replay());
+
+  ASSERT_OK(db2->Get(ro, handles[0], "a", &value));
+  ASSERT_EQ("1", value);
+  ASSERT_OK(db2->Get(ro, handles[0], "g", &value));
+  ASSERT_EQ("12", value);
+  ASSERT_TRUE(db2->Get(ro, handles[0], "hello", &value).IsNotFound());
+  ASSERT_TRUE(db2->Get(ro, handles[0], "world", &value).IsNotFound());
+
+  ASSERT_OK(db2->Get(ro, handles[1], "foo", &value));
+  ASSERT_EQ("bar", value);
+  ASSERT_OK(db2->Get(ro, handles[1], "rocksdb", &value));
+  ASSERT_EQ("rocks", value);
+
+  for (auto handle : handles) {
+    delete handle;
+  }
+  delete db2;
+  ASSERT_OK(DestroyDB(dbname2, options));
+}
+
 #endif  // ROCKSDB_LITE
 
 TEST_F(DBTest2, PinnableSliceAndMmapReads) {
@@ -2547,6 +2649,192 @@ TEST_F(DBTest2, PinnableSliceAndMmapReads) {
 #endif
 }
 
+TEST_F(DBTest2, DISABLED_IteratorPinnedMemory) {
+  Options options = CurrentOptions();
+  options.create_if_missing = true;
+  options.statistics = rocksdb::CreateDBStatistics();
+  BlockBasedTableOptions bbto;
+  bbto.no_block_cache = false;
+  bbto.cache_index_and_filter_blocks = false;
+  bbto.block_cache = NewLRUCache(100000);
+  bbto.block_size = 400;  // small block size
+  options.table_factory.reset(new BlockBasedTableFactory(bbto));
+  Reopen(options);
+
+  Random rnd(301);
+  std::string v = RandomString(&rnd, 400);
+
+  // Since v is the size of a block, each key should take a block
+  // of 400+ bytes.
+  Put("1", v);
+  Put("3", v);
+  Put("5", v);
+  Put("7", v);
+  ASSERT_OK(Flush());
+
+  ASSERT_EQ(0, bbto.block_cache->GetPinnedUsage());
+
+  // Verify that iterators don't pin more than one data block in block cache
+  // at each time.
+  {
+    unique_ptr<Iterator> iter(db_->NewIterator(ReadOptions()));
+    iter->SeekToFirst();
+
+    for (int i = 0; i < 4; i++) {
+      ASSERT_TRUE(iter->Valid());
+      // Block cache should contain exactly one block.
+      ASSERT_GT(bbto.block_cache->GetPinnedUsage(), 0);
+      ASSERT_LT(bbto.block_cache->GetPinnedUsage(), 800);
+      iter->Next();
+    }
+    ASSERT_FALSE(iter->Valid());
+
+    iter->Seek("4");
+    ASSERT_TRUE(iter->Valid());
+
+    ASSERT_GT(bbto.block_cache->GetPinnedUsage(), 0);
+    ASSERT_LT(bbto.block_cache->GetPinnedUsage(), 800);
+
+    iter->Seek("3");
+    ASSERT_TRUE(iter->Valid());
+
+    ASSERT_GT(bbto.block_cache->GetPinnedUsage(), 0);
+    ASSERT_LT(bbto.block_cache->GetPinnedUsage(), 800);
+  }
+  ASSERT_EQ(0, bbto.block_cache->GetPinnedUsage());
+
+  // Test compaction case
+  Put("2", v);
+  Put("5", v);
+  Put("6", v);
+  Put("8", v);
+  ASSERT_OK(Flush());
+
+  // Clear existing data in block cache
+  bbto.block_cache->SetCapacity(0);
+  bbto.block_cache->SetCapacity(100000);
+
+  // Verify compaction input iterators don't hold more than one data blocks at
+  // one time.
+  std::atomic<bool> finished(false);
+  std::atomic<int> block_newed(0);
+  std::atomic<int> block_destroyed(0);
+  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+      "Block::Block:0", [&](void* /*arg*/) {
+        if (finished) {
+          return;
+        }
+        // Two iterators. At most 2 outstanding blocks.
+        EXPECT_GE(block_newed.load(), block_destroyed.load());
+        EXPECT_LE(block_newed.load(), block_destroyed.load() + 1);
+        block_newed.fetch_add(1);
+      });
+  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+      "Block::~Block", [&](void* /*arg*/) {
+        if (finished) {
+          return;
+        }
+        // Two iterators. At most 2 outstanding blocks.
+        EXPECT_GE(block_newed.load(), block_destroyed.load() + 1);
+        EXPECT_LE(block_newed.load(), block_destroyed.load() + 2);
+        block_destroyed.fetch_add(1);
+      });
+  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+      "CompactionJob::Run:BeforeVerify",
+      [&](void* /*arg*/) { finished = true; });
+  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+
+  ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
+
+  // Two input files. Each of them has 4 data blocks.
+  ASSERT_EQ(8, block_newed.load());
+  ASSERT_EQ(8, block_destroyed.load());
+
+  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+}
+
+TEST_F(DBTest2, TestBBTTailPrefetch) {
+  std::atomic<bool> called(false);
+  size_t expected_lower_bound = 512 * 1024;
+  size_t expected_higher_bound = 512 * 1024;
+  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+      "BlockBasedTable::Open::TailPrefetchLen", [&](void* arg) {
+        size_t* prefetch_size = static_cast<size_t*>(arg);
+        EXPECT_LE(expected_lower_bound, *prefetch_size);
+        EXPECT_GE(expected_higher_bound, *prefetch_size);
+        called = true;
+      });
+  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+
+  Put("1", "1");
+  Put("9", "1");
+  Flush();
+
+  expected_lower_bound = 0;
+  expected_higher_bound = 8 * 1024;
+
+  Put("1", "1");
+  Put("9", "1");
+  Flush();
+
+  Put("1", "1");
+  Put("9", "1");
+  Flush();
+
+  // Full compaction to make sure there is no L0 file after the open.
+  ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
+
+  ASSERT_TRUE(called.load());
+  called = false;
+
+  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+  rocksdb::SyncPoint::GetInstance()->ClearAllCallBacks();
+
+  std::atomic<bool> first_call(true);
+  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+      "BlockBasedTable::Open::TailPrefetchLen", [&](void* arg) {
+        size_t* prefetch_size = static_cast<size_t*>(arg);
+        if (first_call) {
+          EXPECT_EQ(4 * 1024, *prefetch_size);
+          first_call = false;
+        } else {
+          EXPECT_GE(4 * 1024, *prefetch_size);
+        }
+        called = true;
+      });
+  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+
+  Options options = CurrentOptions();
+  options.max_file_opening_threads = 1;  // one thread
+  BlockBasedTableOptions table_options;
+  table_options.cache_index_and_filter_blocks = true;
+  options.table_factory.reset(NewBlockBasedTableFactory(table_options));
+  options.max_open_files = -1;
+  Reopen(options);
+
+  Put("1", "1");
+  Put("9", "1");
+  Flush();
+
+  Put("1", "1");
+  Put("9", "1");
+  Flush();
+
+  ASSERT_TRUE(called.load());
+  called = false;
+
+  // Parallel loading SST files
+  options.max_file_opening_threads = 16;
+  Reopen(options);
+
+  ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
+
+  ASSERT_TRUE(called.load());
+
+  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+  rocksdb::SyncPoint::GetInstance()->ClearAllCallBacks();
+}
+
 }  // namespace rocksdb
 
 int main(int argc, char** argv) {

db/db_test_util.cc

@@ -449,15 +449,16 @@ Options DBTestBase::GetOptions(
       options.prefix_extractor.reset(NewNoopTransform());
       break;
     }
-    case kBlockBasedTableWithPartitionedIndexFormat3: {
-      table_options.format_version = 3;
-      // Format 3 changes the binary index format. Since partitioned index is a
+    case kBlockBasedTableWithPartitionedIndexFormat4: {
+      table_options.format_version = 4;
+      // Format 4 changes the binary index format. Since partitioned index is a
       // super-set of simple indexes, we are also using kTwoLevelIndexSearch to
       // test this format.
       table_options.index_type = BlockBasedTableOptions::kTwoLevelIndexSearch;
-      // The top-level index in partition filters are also affected by format 3.
+      // The top-level index in partition filters are also affected by format 4.
       table_options.filter_policy.reset(NewBloomFilterPolicy(10, false));
       table_options.partition_filters = true;
+      table_options.index_block_restart_interval = 8;
       break;
     }
     case kBlockBasedTableWithIndexRestartInterval: {

db/db_test_util.h

@@ -109,8 +109,6 @@ struct OptionsOverride {
   // These will be used only if filter_policy is set
   bool partition_filters = false;
   uint64_t metadata_block_size = 1024;
-  BlockBasedTableOptions::IndexType index_type =
-      BlockBasedTableOptions::IndexType::kBinarySearch;
 
   // Used as a bit mask of individual enums in which to skip an XF test point
   int skip_policy = 0;
@@ -698,7 +696,7 @@ class DBTestBase : public testing::Test {
     kLevelSubcompactions,
     kBlockBasedTableWithIndexRestartInterval,
     kBlockBasedTableWithPartitionedIndex,
-    kBlockBasedTableWithPartitionedIndexFormat3,
+    kBlockBasedTableWithPartitionedIndexFormat4,
     kPartitionedFilterWithNewTableReaderForCompactions,
     kUniversalSubcompactions,
     // This must be the last line

db/db_universal_compaction_test.cc

@@ -1822,9 +1822,9 @@ TEST_P(DBTestUniversalCompaction, FinalSortedRunCompactFilesConflict) {
         "CompactFilesImpl:1"}});
   rocksdb::SyncPoint::GetInstance()->EnableProcessing();
 
-  std::thread compact_files_thread([&]() {
+  port::Thread compact_files_thread([&]() {
     ASSERT_OK(dbfull()->CompactFiles(CompactionOptions(), default_cfh,
-        {first_sst_filename}, num_levels_ - 1));
+                                     {first_sst_filename}, num_levels_ - 1));
   });
 
   TEST_SYNC_POINT(

db/dbformat.cc

@@ -48,6 +48,8 @@ EntryType GetEntryType(ValueType value_type) {
       return kEntryMerge;
     case kTypeRangeDeletion:
       return kEntryRangeDeletion;
+    case kTypeBlobIndex:
+      return kEntryBlobIndex;
     default:
       return kEntryOther;
   }

db/dbformat.h

@@ -346,6 +346,12 @@ class IterKey {
 
   ~IterKey() { ResetBuffer(); }
 
+  // The bool will be picked up by the next calls to SetKey
+  void SetIsUserKey(bool is_user_key) { is_user_key_ = is_user_key; }
+
+  // Returns the key in whichever format that was provided to KeyIter
+  Slice GetKey() const { return Slice(key_, key_size_); }
+
   Slice GetInternalKey() const {
     assert(!IsUserKey());
     return Slice(key_, key_size_);
@@ -395,6 +401,11 @@ class IterKey {
     key_size_ = total_size;
   }
 
+  Slice SetKey(const Slice& key, bool copy = true) {
+    // is_user_key_ expected to be set already via SetIsUserKey
+    return SetKeyImpl(key, copy);
+  }
+
   Slice SetUserKey(const Slice& key, bool copy = true) {
     is_user_key_ = true;
     return SetKeyImpl(key, copy);

db/external_sst_file_ingestion_job.cc

@@ -29,7 +29,8 @@
 namespace rocksdb {
 
 Status ExternalSstFileIngestionJob::Prepare(
-    const std::vector<std::string>& external_files_paths, SuperVersion* sv) {
+    const std::vector<std::string>& external_files_paths,
+    uint64_t next_file_number, SuperVersion* sv) {
   Status status;
 
   // Read the information of files we are ingesting
@@ -90,7 +91,7 @@ Status ExternalSstFileIngestionJob::Prepare(
 
   // Copy/Move external files into DB
   for (IngestedFileInfo& f : files_to_ingest_) {
-    f.fd = FileDescriptor(versions_->NewFileNumber(), 0, f.file_size);
+    f.fd = FileDescriptor(next_file_number++, 0, f.file_size);
 
     const std::string path_outside_db = f.external_file_path;
     const std::string path_inside_db =
@@ -475,9 +476,9 @@ Status ExternalSstFileIngestionJob::AssignLevelAndSeqnoForIngestedFile(
         const SequenceNumber level_largest_seqno =
             (*max_element(level_files.begin(), level_files.end(),
                           [](FileMetaData* f1, FileMetaData* f2) {
-                            return f1->largest_seqno < f2->largest_seqno;
+                            return f1->fd.largest_seqno < f2->fd.largest_seqno;
                           }))
-                ->largest_seqno;
+                ->fd.largest_seqno;
         // should only assign seqno to current level's largest seqno when
         // the file fits
         if (level_largest_seqno != 0 &&
@@ -522,7 +523,7 @@ Status ExternalSstFileIngestionJob::CheckLevelForIngestedBehindFile(
   // at some upper level
   for (int lvl = 0; lvl < cfd_->NumberLevels() - 1; lvl++) {
     for (auto file : vstorage->LevelFiles(lvl)) {
-      if (file->smallest_seqno == 0) {
+      if (file->fd.smallest_seqno == 0) {
         return Status::InvalidArgument(
           "Can't ingest_behind file as despite allow_ingest_behind=true "
           "there are files with 0 seqno in database at upper levels!");
@@ -547,24 +548,27 @@ Status ExternalSstFileIngestionJob::AssignGlobalSeqnoForIngestedFile(
         "field");
   }
 
-  std::unique_ptr<RandomRWFile> rwfile;
-  Status status = env_->NewRandomRWFile(file_to_ingest->internal_file_path,
-                                        &rwfile, env_options_);
-  if (!status.ok()) {
-    return status;
+  if (ingestion_options_.write_global_seqno) {
+    // Determine if we can write global_seqno to a given offset of file.
+    // If the file system does not support random write, then we should not.
+    // Otherwise we should.
+    std::unique_ptr<RandomRWFile> rwfile;
+    Status status = env_->NewRandomRWFile(file_to_ingest->internal_file_path,
+                                          &rwfile, env_options_);
+    if (status.ok()) {
+      std::string seqno_val;
+      PutFixed64(&seqno_val, seqno);
+      status = rwfile->Write(file_to_ingest->global_seqno_offset, seqno_val);
+      if (!status.ok()) {
+        return status;
+      }
+    } else if (!status.IsNotSupported()) {
+      return status;
+    }
   }
 
-  // Write the new seqno in the global sequence number field in the file
-  std::string seqno_val;
-  PutFixed64(&seqno_val, seqno);
-  status = rwfile->Write(file_to_ingest->global_seqno_offset, seqno_val);
-  if (status.ok()) {
-    status = rwfile->Fsync();
-  }
-  if (status.ok()) {
-    file_to_ingest->assigned_seqno = seqno;
-  }
-  return status;
+  file_to_ingest->assigned_seqno = seqno;
+  return Status::OK();
 }
 
 bool ExternalSstFileIngestionJob::IngestedFileFitInLevel(

db/external_sst_file_ingestion_job.h

@@ -89,7 +89,7 @@ class ExternalSstFileIngestionJob {
 
   // Prepare the job by copying external files into the DB.
   Status Prepare(const std::vector<std::string>& external_files_paths,
-                 SuperVersion* sv);
+                 uint64_t next_file_number, SuperVersion* sv);
 
   // Check if we need to flush the memtable before running the ingestion job
   // This will be true if the files we are ingesting are overlapping with any

db/external_sst_file_test.cc

@@ -10,6 +10,7 @@
 #include "port/port.h"
 #include "port/stack_trace.h"
 #include "rocksdb/sst_file_writer.h"
+#include "util/filename.h"
 #include "util/testutil.h"
 
 namespace rocksdb {
@@ -1292,6 +1293,39 @@ TEST_F(ExternalSSTFileTest, PickedLevelBug) {
   rocksdb::SyncPoint::GetInstance()->DisableProcessing();
 }
 
+TEST_F(ExternalSSTFileTest, IngestNonExistingFile) {
+  Options options = CurrentOptions();
+  DestroyAndReopen(options);
+
+  Status s = db_->IngestExternalFile({"non_existing_file"},
+                                     IngestExternalFileOptions());
+  ASSERT_NOK(s);
+
+  // Verify file deletion is not impacted (verify a bug fix)
+  ASSERT_OK(Put(Key(1), Key(1)));
+  ASSERT_OK(Put(Key(9), Key(9)));
+  ASSERT_OK(Flush());
+
+  ASSERT_OK(Put(Key(1), Key(1)));
+  ASSERT_OK(Put(Key(9), Key(9)));
+  ASSERT_OK(Flush());
+
+  ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
+
+  // After full compaction, there should be only 1 file.
+  std::vector<std::string> files;
+  env_->GetChildren(dbname_, &files);
+  int num_sst_files = 0;
+  for (auto& f : files) {
+    uint64_t number;
+    FileType type;
+    if (ParseFileName(f, &number, &type) && type == kTableFile) {
+      num_sst_files++;
+    }
+  }
+  ASSERT_EQ(1, num_sst_files);
+}
+
 TEST_F(ExternalSSTFileTest, CompactDuringAddFileRandom) {
   Options options = CurrentOptions();
   options.disable_auto_compactions = false;

db/flush_job.cc

@@ -389,7 +389,7 @@ Status FlushJob::WriteLevel0Table() {
     // Add file to L0
     edit_->AddFile(0 /* level */, meta_.fd.GetNumber(), meta_.fd.GetPathId(),
                    meta_.fd.GetFileSize(), meta_.smallest, meta_.largest,
-                   meta_.smallest_seqno, meta_.largest_seqno,
+                   meta_.fd.smallest_seqno, meta_.fd.largest_seqno,
                    meta_.marked_for_compaction);
   }
 

db/flush_job_test.cc

@@ -147,19 +147,20 @@ TEST_F(FlushJobTest, NonEmpty) {
                      db_options_.statistics.get(), &event_logger, true);
 
   HistogramData hist;
-  FileMetaData fd;
+  FileMetaData file_meta;
   mutex_.Lock();
   flush_job.PickMemTable();
-  ASSERT_OK(flush_job.Run(nullptr, &fd));
+  ASSERT_OK(flush_job.Run(nullptr, &file_meta));
   mutex_.Unlock();
   db_options_.statistics->histogramData(FLUSH_TIME, &hist);
   ASSERT_GT(hist.average, 0.0);
 
-  ASSERT_EQ(ToString(0), fd.smallest.user_key().ToString());
-  ASSERT_EQ("9999a",
-            fd.largest.user_key().ToString());  // range tombstone end key
-  ASSERT_EQ(1, fd.smallest_seqno);
-  ASSERT_EQ(10000, fd.largest_seqno);  // range tombstone seqnum 10000
+  ASSERT_EQ(ToString(0), file_meta.smallest.user_key().ToString());
+  ASSERT_EQ(
+      "9999a",
+      file_meta.largest.user_key().ToString());  // range tombstone end key
+  ASSERT_EQ(1, file_meta.fd.smallest_seqno);
+  ASSERT_EQ(10000, file_meta.fd.largest_seqno);  // range tombstone seqnum 10000
   mock_table_factory_->AssertSingleFile(inserted_keys);
   job_context.Clean();
 }

db/job_context.h

@@ -35,6 +35,14 @@ struct SuperVersionContext {
   explicit SuperVersionContext(bool create_superversion = false)
     : new_superversion(create_superversion ? new SuperVersion() : nullptr) {}
 
+  explicit SuperVersionContext(SuperVersionContext&& other)
+      : superversions_to_free(std::move(other.superversions_to_free)),
+#ifndef ROCKSDB_DISABLE_STALL_NOTIFICATION
+        write_stall_notifications(std::move(other.write_stall_notifications)),
+#endif
+        new_superversion(std::move(other.new_superversion)) {
+  }
+
   void NewSuperVersion() {
     new_superversion = unique_ptr<SuperVersion>(new SuperVersion());
   }
@@ -98,8 +106,15 @@ struct JobContext {
   }
 
   inline bool HaveSomethingToClean() const {
+    bool sv_have_sth = false;
+    for (const auto& sv_ctx : superversion_contexts) {
+      if (sv_ctx.HaveSomethingToDelete()) {
+        sv_have_sth = true;
+        break;
+      }
+    }
     return memtables_to_free.size() > 0 || logs_to_free.size() > 0 ||
-           superversion_context.HaveSomethingToDelete();
+           sv_have_sth;
   }
 
   // Structure to store information for candidate files to delete.
@@ -142,7 +157,8 @@ struct JobContext {
   // a list of memtables to be free
   autovector<MemTable*> memtables_to_free;
 
-  SuperVersionContext superversion_context;
+  // contexts for installing superversions for multiple column families
+  std::vector<SuperVersionContext> superversion_contexts;
 
   autovector<log::Writer*> logs_to_free;
 
@@ -158,13 +174,14 @@ struct JobContext {
   size_t num_alive_log_files = 0;
   uint64_t size_log_to_delete = 0;
 
-  explicit JobContext(int _job_id, bool create_superversion = false)
-    : superversion_context(create_superversion) {
+  explicit JobContext(int _job_id, bool create_superversion = false) {
     job_id = _job_id;
     manifest_file_number = 0;
     pending_manifest_file_number = 0;
     log_number = 0;
     prev_log_number = 0;
+    superversion_contexts.emplace_back(
+        SuperVersionContext(create_superversion));
   }
 
   // For non-empty JobContext Clean() has to be called at least once before
@@ -173,7 +190,9 @@ struct JobContext {
   // doing potentially slow Clean() with locked DB mutex.
   void Clean() {
     // free superversions
-    superversion_context.Clean();
+    for (auto& sv_context : superversion_contexts) {
+      sv_context.Clean();
+    }
     // free pending memtables
     for (auto m : memtables_to_free) {
       delete m;

db/listener_test.cc

@@ -417,7 +417,9 @@ TEST_F(EventListenerTest, DisableBGCompaction) {
   for (int i = 0; static_cast<int>(cf_meta.file_count) < kSlowdownTrigger * 10;
        ++i) {
     Put(1, ToString(i), std::string(10000, 'x'), WriteOptions());
-    db_->Flush(FlushOptions(), handles_[1]);
+    FlushOptions fo;
+    fo.allow_write_stall = true;
+    db_->Flush(fo, handles_[1]);
     db_->GetColumnFamilyMetaData(handles_[1], &cf_meta);
   }
   ASSERT_GE(listener->slowdown_count, kSlowdownTrigger * 9);

db/managed_iterator.cc

@@ -1,254 +0,0 @@
-//  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).
-
-#ifndef ROCKSDB_LITE
-
-#include "db/managed_iterator.h"
-
-#include <limits>
-#include <string>
-#include <utility>
-
-#include "db/column_family.h"
-#include "db/db_impl.h"
-#include "db/db_iter.h"
-#include "db/dbformat.h"
-#include "rocksdb/env.h"
-#include "rocksdb/slice.h"
-#include "rocksdb/slice_transform.h"
-#include "table/merging_iterator.h"
-
-namespace rocksdb {
-
-namespace {
-// Helper class that locks a mutex on construction and unlocks the mutex when
-// the destructor of the MutexLock object is invoked.
-//
-// Typical usage:
-//
-//   void MyClass::MyMethod() {
-//     MILock l(&mu_);       // mu_ is an instance variable
-//     ... some complex code, possibly with multiple return paths ...
-//   }
-
-class MILock {
- public:
-  explicit MILock(std::mutex* mu, ManagedIterator* mi) : mu_(mu), mi_(mi) {
-    this->mu_->lock();
-  }
-  ~MILock() {
-    this->mu_->unlock();
-  }
-  ManagedIterator* GetManagedIterator() { return mi_; }
-
- private:
-  std::mutex* const mu_;
-  ManagedIterator* mi_;
-  // No copying allowed
-  MILock(const MILock&) = delete;
-  void operator=(const MILock&) = delete;
-};
-}  // anonymous namespace
-
-//
-// Synchronization between modifiers, releasers, creators
-// If iterator operation, wait till (!in_use), set in_use, do op, reset in_use
-//  if modifying mutable_iter, atomically exchange in_use:
-//  return if in_use set / otherwise set in use,
-//  atomically replace new iter with old , reset in use
-//  The releaser is the new operation and it holds a lock for a very short time
-//  The existing non-const iterator operations are supposed to be single
-//  threaded and hold the lock for the duration of the operation
-//  The existing const iterator operations use the cached key/values
-//  and don't do any locking.
-ManagedIterator::ManagedIterator(DBImpl* db, const ReadOptions& read_options,
-                                 ColumnFamilyData* cfd)
-    : db_(db),
-      read_options_(read_options),
-      cfd_(cfd),
-      svnum_(cfd->GetSuperVersionNumber()),
-      mutable_iter_(nullptr),
-      valid_(false),
-      snapshot_created_(false),
-      release_supported_(true) {
-  read_options_.managed = false;
-  if ((!read_options_.tailing) && (read_options_.snapshot == nullptr)) {
-    assert(nullptr != (read_options_.snapshot = db_->GetSnapshot()));
-    snapshot_created_ = true;
-  }
-  cfh_.SetCFD(cfd);
-  mutable_iter_ = unique_ptr<Iterator>(db->NewIterator(read_options_, &cfh_));
-}
-
-ManagedIterator::~ManagedIterator() {
-  Lock();
-  if (snapshot_created_) {
-    db_->ReleaseSnapshot(read_options_.snapshot);
-    snapshot_created_ = false;
-    read_options_.snapshot = nullptr;
-  }
-  UnLock();
-}
-
-bool ManagedIterator::Valid() const { return valid_; }
-
-void ManagedIterator::SeekToLast() {
-  MILock l(&in_use_, this);
-  if (NeedToRebuild()) {
-    RebuildIterator();
-  }
-  assert(mutable_iter_ != nullptr);
-  mutable_iter_->SeekToLast();
-  UpdateCurrent();
-}
-
-void ManagedIterator::SeekToFirst() {
-  MILock l(&in_use_, this);
-  SeekInternal(Slice(), true);
-}
-
-void ManagedIterator::Seek(const Slice& user_key) {
-  MILock l(&in_use_, this);
-  SeekInternal(user_key, false);
-}
-
-void ManagedIterator::SeekForPrev(const Slice& user_key) {
-  MILock l(&in_use_, this);
-  if (NeedToRebuild()) {
-    RebuildIterator();
-  }
-  assert(mutable_iter_ != nullptr);
-  mutable_iter_->SeekForPrev(user_key);
-  UpdateCurrent();
-}
-
-void ManagedIterator::SeekInternal(const Slice& user_key, bool seek_to_first) {
-  if (NeedToRebuild()) {
-    RebuildIterator();
-  }
-  assert(mutable_iter_ != nullptr);
-  if (seek_to_first) {
-    mutable_iter_->SeekToFirst();
-  } else {
-    mutable_iter_->Seek(user_key);
-  }
-  UpdateCurrent();
-}
-
-void ManagedIterator::Prev() {
-  if (!valid_) {
-    status_ = Status::InvalidArgument("Iterator value invalid");
-    return;
-  }
-  MILock l(&in_use_, this);
-  if (NeedToRebuild()) {
-    RebuildIterator(true);
-    if (!valid_) {
-      return;
-    }
-  }
-  mutable_iter_->Prev();
-  UpdateCurrent();
-}
-
-void ManagedIterator::Next() {
-  if (!valid_) {
-    status_ = Status::InvalidArgument("Iterator value invalid");
-    return;
-  }
-  MILock l(&in_use_, this);
-  if (NeedToRebuild()) {
-    RebuildIterator(true);
-    if (!valid_) {
-      return;
-    }
-  }
-  mutable_iter_->Next();
-  UpdateCurrent();
-}
-
-Slice ManagedIterator::key() const {
-  assert(valid_);
-  return cached_key_.GetUserKey();
-}
-
-Slice ManagedIterator::value() const {
-  assert(valid_);
-  return cached_value_.GetUserKey();
-}
-
-Status ManagedIterator::status() const { return status_; }
-
-void ManagedIterator::RebuildIterator(bool reseek) {
-  std::string current_key;
-  if (reseek) {
-    current_key = key().ToString();
-  }
-
-  svnum_ = cfd_->GetSuperVersionNumber();
-  mutable_iter_ = unique_ptr<Iterator>(db_->NewIterator(read_options_, &cfh_));
-
-  if (reseek) {
-    Slice old_key(current_key.data(), current_key.size());
-    SeekInternal(old_key, false);
-    UpdateCurrent();
-    if (!valid_ || key().compare(old_key) != 0) {
-      valid_ = false;
-      status_ = Status::Incomplete(
-          "Next/Prev failed because current key has "
-          "been removed");
-    }
-  }
-}
-
-void ManagedIterator::UpdateCurrent() {
-  assert(mutable_iter_ != nullptr);
-
-  valid_ = mutable_iter_->Valid();
-  status_ = mutable_iter_->status();
-
-  if (!valid_) {
-    return;
-  }
-
-  cached_key_.SetUserKey(mutable_iter_->key());
-  cached_value_.SetUserKey(mutable_iter_->value());
-}
-
-void ManagedIterator::ReleaseIter(bool only_old) {
-  if ((mutable_iter_ == nullptr) || (!release_supported_)) {
-    return;
-  }
-  if (svnum_ != cfd_->GetSuperVersionNumber() || !only_old) {
-    if (!TryLock()) {  // Don't release iter if in use
-      return;
-    }
-    mutable_iter_ = nullptr;  // in_use for a very short time
-    UnLock();
-  }
-}
-
-bool ManagedIterator::NeedToRebuild() {
-  if ((mutable_iter_ == nullptr) || (status_.IsIncomplete()) ||
-      (!only_drop_old_ && (svnum_ != cfd_->GetSuperVersionNumber()))) {
-    return true;
-  }
-  return false;
-}
-
-void ManagedIterator::Lock() {
-  in_use_.lock();
-  return;
-}
-
-bool ManagedIterator::TryLock() { return in_use_.try_lock(); }
-
-void ManagedIterator::UnLock() {
-  in_use_.unlock();
-}
-
-}  // namespace rocksdb
-
-#endif  // ROCKSDB_LITE

db/managed_iterator.h

@@ -1,85 +0,0 @@
-//  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 <mutex>
-#include <queue>
-#include <string>
-#include <vector>
-
-#include "db/column_family.h"
-#include "rocksdb/db.h"
-#include "rocksdb/iterator.h"
-#include "rocksdb/options.h"
-#include "util/arena.h"
-
-namespace rocksdb {
-
-class DBImpl;
-struct SuperVersion;
-class ColumnFamilyData;
-
-/**
- * ManagedIterator is a special type of iterator that supports freeing the
- * underlying iterator and still being able to access the current key/value
- * pair.  This is done by copying the key/value pair so that clients can
- * continue to access the data without getting a SIGSEGV.
- * The underlying iterator can be freed manually through the  call to
- * ReleaseIter or automatically (as needed on space pressure or age.)
- * The iterator is recreated using the saved original arguments.
- */
-class ManagedIterator : public Iterator {
- public:
-  ManagedIterator(DBImpl* db, const ReadOptions& read_options,
-                  ColumnFamilyData* cfd);
-  virtual ~ManagedIterator();
-
-  virtual void SeekToLast() override;
-  virtual void Prev() override;
-  virtual bool Valid() const override;
-  void SeekToFirst() override;
-  virtual void Seek(const Slice& target) override;
-  virtual void SeekForPrev(const Slice& target) override;
-  virtual void Next() override;
-  virtual Slice key() const override;
-  virtual Slice value() const override;
-  virtual Status status() const override;
-  void ReleaseIter(bool only_old);
-  void SetDropOld(bool only_old) {
-    only_drop_old_ = read_options_.tailing || only_old;
-  }
-
- private:
-  void RebuildIterator(bool reseek = false);
-  void UpdateCurrent();
-  void SeekInternal(const Slice& user_key, bool seek_to_first);
-  bool NeedToRebuild();
-  void Lock();
-  bool TryLock();
-  void UnLock();
-  DBImpl* const db_;
-  ReadOptions read_options_;
-  ColumnFamilyData* const cfd_;
-  ColumnFamilyHandleInternal cfh_;
-
-  uint64_t svnum_;
-  std::unique_ptr<Iterator> mutable_iter_;
-  // internal iterator status
-  Status status_;
-  bool valid_;
-
-  IterKey cached_key_;
-  IterKey cached_value_;
-
-  bool only_drop_old_ = true;
-  bool snapshot_created_;
-  bool release_supported_;
-  std::mutex in_use_;  // is managed iterator in use
-};
-
-}  // namespace rocksdb
-#endif  // !ROCKSDB_LITE

db/memtable.cc

@@ -697,7 +697,7 @@ static bool SaveValue(void* arg, const char* entry) {
         *(s->merge_in_progress) = true;
         merge_context->PushOperand(
             v, s->inplace_update_support == false /* operand_pinned */);
-        if (merge_operator->ShouldMerge(merge_context->GetOperands())) {
+        if (merge_operator->ShouldMerge(merge_context->GetOperandsDirectionBackward())) {
           *(s->status) = MergeHelper::TimedFullMerge(
               merge_operator, s->key->user_key(), nullptr,
               merge_context->GetOperands(), s->value, s->logger, s->statistics,

db/memtable.h

@@ -34,7 +34,6 @@ namespace rocksdb {
 class Mutex;
 class MemTableIterator;
 class MergeContext;
-class InternalIterator;
 
 struct ImmutableMemTableOptions {
   explicit ImmutableMemTableOptions(const ImmutableCFOptions& ioptions,

db/merge_context.h

@@ -74,8 +74,13 @@ class MergeContext {
     return (*operand_list_)[index];
   }
 
-  // Return all the operands.
+  // Same as GetOperandsDirectionForward
   const std::vector<Slice>& GetOperands() {
+    return GetOperandsDirectionForward();
+  }
+
+  // Return all the operands in the order as they were merged (passed to FullMerge or FullMergeV2)
+  const std::vector<Slice>& GetOperandsDirectionForward() {
     if (!operand_list_) {
       return empty_operand_list;
     }
@@ -84,6 +89,16 @@ class MergeContext {
     return *operand_list_;
   }
 
+  // Return all the operands in the reversed order relative to how they were merged (passed to FullMerge or FullMergeV2)
+  const std::vector<Slice>& GetOperandsDirectionBackward() {
+    if (!operand_list_) {
+      return empty_operand_list;
+    }
+
+    SetDirectionBackward();
+    return *operand_list_;
+  }
+
  private:
   void Initialize() {
     if (!operand_list_) {

db/merge_helper.h

@@ -26,7 +26,6 @@ class Iterator;
 class Logger;
 class MergeOperator;
 class Statistics;
-class InternalIterator;
 
 class MergeHelper {
  public:

db/obsolete_files_test.cc

@@ -227,16 +227,24 @@ TEST_F(ObsoleteFilesTest, DeleteObsoleteOptionsFile) {
   }
   ASSERT_OK(dbi->EnableFileDeletions(true /* force */));
   ASSERT_EQ(optsfiles_nums.size(), optsfiles_keep.size());
-  int size = static_cast<int>(optsfiles_nums.size());
-  int kept_opts_files_count = 0;
-  for (int i = 0; i != size; ++i) {
-    if (optsfiles_keep[i]) {
-      ++kept_opts_files_count;
-    }
-  }
-  ASSERT_EQ(2, kept_opts_files_count);
 
   CloseDB();
+
+  std::vector<std::string> files;
+  int opts_file_count = 0;
+  ASSERT_OK(env_->GetChildren(dbname_, &files));
+  for (const auto& file : files) {
+    uint64_t file_num;
+    Slice dummy_info_log_name_prefix;
+    FileType type;
+    WalFileType log_type;
+    if (ParseFileName(file, &file_num, dummy_info_log_name_prefix, &type,
+                      &log_type) &&
+        type == kOptionsFile) {
+      opts_file_count++;
+    }
+  }
+  ASSERT_EQ(2, opts_file_count);
 }
 
 } //namespace rocksdb

db/range_del_aggregator.cc

@@ -50,7 +50,7 @@ class UncollapsedRangeDelMap : public RangeDelMap {
       : rep_(TombstoneStartKeyComparator(ucmp)), ucmp_(ucmp) {}
 
   bool ShouldDelete(const ParsedInternalKey& parsed,
-                    RangeDelPositioningMode mode) {
+                    RangeDelPositioningMode mode) override {
     (void)mode;
     assert(mode == RangeDelPositioningMode::kFullScan);
     for (const auto& tombstone : rep_) {
@@ -65,7 +65,7 @@ class UncollapsedRangeDelMap : public RangeDelMap {
     return false;
   }
 
-  bool IsRangeOverlapped(const Slice& start, const Slice& end) {
+  bool IsRangeOverlapped(const Slice& start, const Slice& end) override {
     for (const auto& tombstone : rep_) {
       if (ucmp_->Compare(start, tombstone.end_key_) < 0 &&
           ucmp_->Compare(tombstone.start_key_, end) <= 0 &&
@@ -76,13 +76,13 @@ class UncollapsedRangeDelMap : public RangeDelMap {
     return false;
   }
 
-  void AddTombstone(RangeTombstone tombstone) { rep_.emplace(tombstone); }
+  void AddTombstone(RangeTombstone tombstone) override { rep_.emplace(tombstone); }
 
-  size_t Size() const { return rep_.size(); }
+  size_t Size() const override { return rep_.size(); }
 
-  void InvalidatePosition() {}  // no-op
+  void InvalidatePosition() override {}  // no-op
 
-  std::unique_ptr<RangeDelIterator> NewIterator() {
+  std::unique_ptr<RangeDelIterator> NewIterator() override {
     return std::unique_ptr<RangeDelIterator>(new Iterator(this->rep_));
   }
 };
@@ -155,6 +155,9 @@ class CollapsedRangeDelMap : public RangeDelMap {
     }
 
     RangeTombstone Tombstone() const override {
+      assert(Valid());
+      assert(std::next(iter_) != rep_.end());
+      assert(iter_->second != 0);
       RangeTombstone tombstone;
       tombstone.start_key_ = iter_->first;
       tombstone.end_key_ = std::next(iter_)->first;
@@ -173,7 +176,7 @@ class CollapsedRangeDelMap : public RangeDelMap {
   }
 
   bool ShouldDelete(const ParsedInternalKey& parsed,
-                    RangeDelPositioningMode mode) {
+                    RangeDelPositioningMode mode) override {
     if (iter_ == rep_.end() &&
         (mode == RangeDelPositioningMode::kForwardTraversal ||
          mode == RangeDelPositioningMode::kBackwardTraversal)) {
@@ -224,15 +227,15 @@ class CollapsedRangeDelMap : public RangeDelMap {
     return parsed.sequence < iter_->second;
   }
 
-  bool IsRangeOverlapped(const Slice&, const Slice&) {
+  bool IsRangeOverlapped(const Slice&, const Slice&) override {
     // Unimplemented because the only client of this method, file ingestion,
     // uses uncollapsed maps.
     fprintf(stderr, "CollapsedRangeDelMap::IsRangeOverlapped unimplemented");
     abort();
   }
 
-  void AddTombstone(RangeTombstone t) {
-    if (ucmp_->Compare(t.start_key_, t.end_key_) >= 0) {
+  void AddTombstone(RangeTombstone t) override {
+    if (ucmp_->Compare(t.start_key_, t.end_key_) >= 0 || t.seq_ == 0) {
       // The tombstone covers no keys. Nothing to do.
       return;
     }
@@ -341,11 +344,11 @@ class CollapsedRangeDelMap : public RangeDelMap {
     }
   }
 
-  size_t Size() const { return rep_.size() - 1; }
+  size_t Size() const override { return rep_.empty() ? 0 : rep_.size() - 1; }
 
-  void InvalidatePosition() { iter_ = rep_.end(); }
+  void InvalidatePosition() override { iter_ = rep_.end(); }
 
-  std::unique_ptr<RangeDelIterator> NewIterator() {
+  std::unique_ptr<RangeDelIterator> NewIterator() override {
     return std::unique_ptr<RangeDelIterator>(new Iterator(this->rep_));
   }
 };
@@ -490,7 +493,8 @@ Status RangeDelAggregator::AddTombstones(
         tombstone.end_key_ = largest->user_key();
       }
     }
-    GetRangeDelMap(tombstone.seq_).AddTombstone(std::move(tombstone));
+    auto seq = tombstone.seq_;
+    GetRangeDelMap(seq).AddTombstone(std::move(tombstone));
     input->Next();
   }
   if (!first_iter) {

db/snapshot_checker.h

@@ -20,7 +20,7 @@ class DisableGCSnapshotChecker : public SnapshotChecker {
  public:
   virtual ~DisableGCSnapshotChecker() {}
   virtual bool IsInSnapshot(SequenceNumber /*sequence*/,
-                            SequenceNumber /*snapshot_sequence*/) const {
+                            SequenceNumber /*snapshot_sequence*/) const override {
     // By returning false, we prevent all the values from being GCed
     return false;
   }

db/snapshot_impl.h

@@ -56,6 +56,9 @@ class SnapshotList {
     count_ = 0;
   }
 
+  // No copy-construct.
+  SnapshotList(const SnapshotList&) = delete;
+
   bool empty() const { return list_.next_ == &list_; }
   SnapshotImpl* oldest() const { assert(!empty()); return list_.next_; }
   SnapshotImpl* newest() const { assert(!empty()); return list_.prev_; }

db/table_cache.cc

@@ -120,7 +120,7 @@ Status TableCache::GetTableReader(
     s = ioptions_.table_factory->NewTableReader(
         TableReaderOptions(ioptions_, prefix_extractor, env_options,
                            internal_comparator, skip_filters, immortal_tables_,
-                           level),
+                           level, fd.largest_seqno),
         std::move(file_reader), fd.GetFileSize(), table_reader,
         prefetch_index_and_filter_in_cache);
     TEST_SYNC_POINT("TableCache::GetTableReader:0");
@@ -238,7 +238,7 @@ InternalIterator* TableCache::NewIterator(
   if (s.ok()) {
     if (options.table_filter &&
         !options.table_filter(*table_reader->GetTableProperties())) {
-      result = NewEmptyInternalIterator(arena);
+      result = NewEmptyInternalIterator<Slice>(arena);
     } else {
       result = table_reader->NewIterator(options, prefix_extractor, arena,
                                          skip_filters, for_compaction);
@@ -279,7 +279,7 @@ InternalIterator* TableCache::NewIterator(
   }
   if (!s.ok()) {
     assert(result == nullptr);
-    result = NewErrorInternalIterator(s, arena);
+    result = NewErrorInternalIterator<Slice>(s, arena);
   }
   return result;
 }

db/table_cache.h

@@ -31,7 +31,6 @@ class Arena;
 struct FileDescriptor;
 class GetContext;
 class HistogramImpl;
-class InternalIterator;
 
 class TableCache {
  public:

db/version_builder.cc

@@ -35,11 +35,11 @@
 namespace rocksdb {
 
 bool NewestFirstBySeqNo(FileMetaData* a, FileMetaData* b) {
-  if (a->largest_seqno != b->largest_seqno) {
-    return a->largest_seqno > b->largest_seqno;
+  if (a->fd.largest_seqno != b->fd.largest_seqno) {
+    return a->fd.largest_seqno > b->fd.largest_seqno;
   }
-  if (a->smallest_seqno != b->smallest_seqno) {
-    return a->smallest_seqno > b->smallest_seqno;
+  if (a->fd.smallest_seqno != b->fd.smallest_seqno) {
+    return a->fd.smallest_seqno > b->fd.smallest_seqno;
   }
   // Break ties by file number
   return a->fd.GetNumber() > b->fd.GetNumber();
@@ -162,22 +162,24 @@ class VersionBuilder::Rep {
             abort();
           }
 
-          if (f2->smallest_seqno == f2->largest_seqno) {
+          if (f2->fd.smallest_seqno == f2->fd.largest_seqno) {
             // This is an external file that we ingested
-            SequenceNumber external_file_seqno = f2->smallest_seqno;
-            if (!(external_file_seqno < f1->largest_seqno ||
+            SequenceNumber external_file_seqno = f2->fd.smallest_seqno;
+            if (!(external_file_seqno < f1->fd.largest_seqno ||
                   external_file_seqno == 0)) {
-              fprintf(stderr, "L0 file with seqno %" PRIu64 " %" PRIu64
-                              " vs. file with global_seqno %" PRIu64 "\n",
-                      f1->smallest_seqno, f1->largest_seqno,
+              fprintf(stderr,
+                      "L0 file with seqno %" PRIu64 " %" PRIu64
+                      " vs. file with global_seqno %" PRIu64 "\n",
+                      f1->fd.smallest_seqno, f1->fd.largest_seqno,
                       external_file_seqno);
               abort();
             }
-          } else if (f1->smallest_seqno <= f2->smallest_seqno) {
-            fprintf(stderr, "L0 files seqno %" PRIu64 " %" PRIu64
-                            " vs. %" PRIu64 " %" PRIu64 "\n",
-                    f1->smallest_seqno, f1->largest_seqno, f2->smallest_seqno,
-                    f2->largest_seqno);
+          } else if (f1->fd.smallest_seqno <= f2->fd.smallest_seqno) {
+            fprintf(stderr,
+                    "L0 files seqno %" PRIu64 " %" PRIu64 " vs. %" PRIu64
+                    " %" PRIu64 "\n",
+                    f1->fd.smallest_seqno, f1->fd.largest_seqno,
+                    f2->fd.smallest_seqno, f2->fd.largest_seqno);
             abort();
           }
         } else {
@@ -382,7 +384,7 @@ class VersionBuilder::Rep {
     }
 
     std::atomic<size_t> next_file_meta_idx(0);
-    std::function<void()> load_handlers_func = [&]() {
+    std::function<void()> load_handlers_func([&]() {
       while (true) {
         size_t file_idx = next_file_meta_idx.fetch_add(1);
         if (file_idx >= files_meta.size()) {
@@ -403,7 +405,7 @@ class VersionBuilder::Rep {
               file_meta->table_reader_handle);
         }
       }
-    };
+    });
 
     std::vector<port::Thread> threads;
     for (int i = 1; i < max_threads; i++) {

db/version_builder_test.cc

@@ -63,8 +63,8 @@ class VersionBuilderTest : public testing::Test {
     f->fd = FileDescriptor(file_number, path_id, file_size);
     f->smallest = GetInternalKey(smallest, smallest_seq);
     f->largest = GetInternalKey(largest, largest_seq);
-    f->smallest_seqno = smallest_seqno;
-    f->largest_seqno = largest_seqno;
+    f->fd.smallest_seqno = smallest_seqno;
+    f->fd.largest_seqno = largest_seqno;
     f->compensated_file_size = file_size;
     f->refs = 0;
     f->num_entries = num_entries;

db/version_edit.cc

@@ -40,13 +40,15 @@ enum Tag : uint32_t {
   kColumnFamilyAdd = 201,
   kColumnFamilyDrop = 202,
   kMaxColumnFamily = 203,
+
+  kInAtomicGroup = 300,
 };
 
 enum CustomTag : uint32_t {
   kTerminate = 1,  // The end of customized fields
   kNeedCompaction = 2,
   // Since Manifest is not entirely currently forward-compatible, and the only
-  // forward-compatbile part is the CutsomtTag of kNewFile, we currently encode
+  // forward-compatible part is the CutsomtTag of kNewFile, we currently encode
   // kMinLogNumberToKeep as part of a CustomTag as a hack. This should be
   // removed when manifest becomes forward-comptabile.
   kMinLogNumberToKeepHack = 3,
@@ -83,6 +85,8 @@ void VersionEdit::Clear() {
   is_column_family_add_ = 0;
   is_column_family_drop_ = 0;
   column_family_name_.clear();
+  is_in_atomic_group_ = false;
+  remaining_entries_ = 0;
 }
 
 bool VersionEdit::EncodeTo(std::string* dst) const {
@@ -135,7 +139,7 @@ bool VersionEdit::EncodeTo(std::string* dst) const {
     PutVarint64(dst, f.fd.GetFileSize());
     PutLengthPrefixedSlice(dst, f.smallest.Encode());
     PutLengthPrefixedSlice(dst, f.largest.Encode());
-    PutVarint64Varint64(dst, f.smallest_seqno, f.largest_seqno);
+    PutVarint64Varint64(dst, f.fd.smallest_seqno, f.fd.largest_seqno);
     if (has_customized_fields) {
       // Customized fields' format:
       // +-----------------------------+
@@ -200,6 +204,11 @@ bool VersionEdit::EncodeTo(std::string* dst) const {
   if (is_column_family_drop_) {
     PutVarint32(dst, kColumnFamilyDrop);
   }
+
+  if (is_in_atomic_group_) {
+    PutVarint32(dst, kInAtomicGroup);
+    PutVarint32(dst, remaining_entries_);
+  }
   return true;
 }
 
@@ -233,14 +242,16 @@ const char* VersionEdit::DecodeNewFile4From(Slice* input) {
   uint64_t number;
   uint32_t path_id = 0;
   uint64_t file_size;
+  SequenceNumber smallest_seqno;
+  SequenceNumber largest_seqno;
   // Since this is the only forward-compatible part of the code, we hack new
   // extension into this record. When we do, we set this boolean to distinguish
   // the record from the normal NewFile records.
   if (GetLevel(input, &level, &msg) && GetVarint64(input, &number) &&
       GetVarint64(input, &file_size) && GetInternalKey(input, &f.smallest) &&
       GetInternalKey(input, &f.largest) &&
-      GetVarint64(input, &f.smallest_seqno) &&
-      GetVarint64(input, &f.largest_seqno)) {
+      GetVarint64(input, &smallest_seqno) &&
+      GetVarint64(input, &largest_seqno)) {
     // See comments in VersionEdit::EncodeTo() for format of customized fields
     while (true) {
       uint32_t custom_tag;
@@ -272,7 +283,7 @@ const char* VersionEdit::DecodeNewFile4From(Slice* input) {
           break;
         case kMinLogNumberToKeepHack:
           // This is a hack to encode kMinLogNumberToKeep in a
-          // forward-compatbile fashion.
+          // forward-compatible fashion.
           if (!GetFixed64(&field, &min_log_number_to_keep_)) {
             return "deleted log number malformatted";
           }
@@ -289,7 +300,8 @@ const char* VersionEdit::DecodeNewFile4From(Slice* input) {
   } else {
     return "new-file4 entry";
   }
-  f.fd = FileDescriptor(number, path_id, file_size);
+  f.fd =
+      FileDescriptor(number, path_id, file_size, smallest_seqno, largest_seqno);
   new_files_.push_back(std::make_pair(level, f));
   return nullptr;
 }
@@ -409,13 +421,16 @@ Status VersionEdit::DecodeFrom(const Slice& src) {
       case kNewFile2: {
         uint64_t number;
         uint64_t file_size;
+        SequenceNumber smallest_seqno;
+        SequenceNumber largest_seqno;
         if (GetLevel(&input, &level, &msg) && GetVarint64(&input, &number) &&
             GetVarint64(&input, &file_size) &&
             GetInternalKey(&input, &f.smallest) &&
             GetInternalKey(&input, &f.largest) &&
-            GetVarint64(&input, &f.smallest_seqno) &&
-            GetVarint64(&input, &f.largest_seqno)) {
-          f.fd = FileDescriptor(number, 0, file_size);
+            GetVarint64(&input, &smallest_seqno) &&
+            GetVarint64(&input, &largest_seqno)) {
+          f.fd = FileDescriptor(number, 0, file_size, smallest_seqno,
+                                largest_seqno);
           new_files_.push_back(std::make_pair(level, f));
         } else {
           if (!msg) {
@@ -429,13 +444,16 @@ Status VersionEdit::DecodeFrom(const Slice& src) {
         uint64_t number;
         uint32_t path_id;
         uint64_t file_size;
+        SequenceNumber smallest_seqno;
+        SequenceNumber largest_seqno;
         if (GetLevel(&input, &level, &msg) && GetVarint64(&input, &number) &&
             GetVarint32(&input, &path_id) && GetVarint64(&input, &file_size) &&
             GetInternalKey(&input, &f.smallest) &&
             GetInternalKey(&input, &f.largest) &&
-            GetVarint64(&input, &f.smallest_seqno) &&
-            GetVarint64(&input, &f.largest_seqno)) {
-          f.fd = FileDescriptor(number, path_id, file_size);
+            GetVarint64(&input, &smallest_seqno) &&
+            GetVarint64(&input, &largest_seqno)) {
+          f.fd = FileDescriptor(number, path_id, file_size, smallest_seqno,
+                                largest_seqno);
           new_files_.push_back(std::make_pair(level, f));
         } else {
           if (!msg) {
@@ -473,6 +491,15 @@ Status VersionEdit::DecodeFrom(const Slice& src) {
         is_column_family_drop_ = true;
         break;
 
+      case kInAtomicGroup:
+        is_in_atomic_group_ = true;
+        if (!GetVarint32(&input, &remaining_entries_)) {
+          if (!msg) {
+            msg = "remaining entries";
+          }
+        }
+        break;
+
       default:
         msg = "unknown tag";
         break;
@@ -551,6 +578,11 @@ std::string VersionEdit::DebugString(bool hex_key) const {
     r.append("\n  MaxColumnFamily: ");
     AppendNumberTo(&r, max_column_family_);
   }
+  if (is_in_atomic_group_) {
+    r.append("\n AtomicGroup: ");
+    AppendNumberTo(&r, remaining_entries_);
+    r.append(" entries remains");
+  }
   r.append("\n}\n");
   return r;
 }
@@ -623,6 +655,9 @@ std::string VersionEdit::DebugJSON(int edit_num, bool hex_key) const {
   if (has_min_log_number_to_keep_) {
     jw << "MinLogNumberToKeep" << min_log_number_to_keep_;
   }
+  if (is_in_atomic_group_) {
+    jw << "AtomicGroup" << remaining_entries_;
+  }
 
   jw.EndObject();
 

db/version_edit.h

@@ -27,7 +27,7 @@ const uint64_t kFileNumberMask = 0x3FFFFFFFFFFFFFFF;
 extern uint64_t PackFileNumberAndPathId(uint64_t number, uint64_t path_id);
 
 // A copyable structure contains information needed to read data from an SST
-// file. It can contains a pointer to a table reader opened for the file, or
+// file. It can contain a pointer to a table reader opened for the file, or
 // file number and size, which can be used to create a new table reader for it.
 // The behavior is undefined when a copied of the structure is used when the
 // file is not in any live version any more.
@@ -36,18 +36,28 @@ struct FileDescriptor {
   TableReader* table_reader;
   uint64_t packed_number_and_path_id;
   uint64_t file_size;  // File size in bytes
+  SequenceNumber smallest_seqno;  // The smallest seqno in this file
+  SequenceNumber largest_seqno;   // The largest seqno in this file
 
   FileDescriptor() : FileDescriptor(0, 0, 0) {}
 
   FileDescriptor(uint64_t number, uint32_t path_id, uint64_t _file_size)
+      : FileDescriptor(number, path_id, _file_size, kMaxSequenceNumber, 0) {}
+
+  FileDescriptor(uint64_t number, uint32_t path_id, uint64_t _file_size,
+                 SequenceNumber _smallest_seqno, SequenceNumber _largest_seqno)
       : table_reader(nullptr),
         packed_number_and_path_id(PackFileNumberAndPathId(number, path_id)),
-        file_size(_file_size) {}
+        file_size(_file_size),
+        smallest_seqno(_smallest_seqno),
+        largest_seqno(_largest_seqno) {}
 
   FileDescriptor& operator=(const FileDescriptor& fd) {
     table_reader = fd.table_reader;
     packed_number_and_path_id = fd.packed_number_and_path_id;
     file_size = fd.file_size;
+    smallest_seqno = fd.smallest_seqno;
+    largest_seqno = fd.largest_seqno;
     return *this;
   }
 
@@ -77,8 +87,6 @@ struct FileMetaData {
   FileDescriptor fd;
   InternalKey smallest;            // Smallest internal key served by table
   InternalKey largest;             // Largest internal key served by table
-  SequenceNumber smallest_seqno;   // The smallest seqno in this file
-  SequenceNumber largest_seqno;    // The largest seqno in this file
 
   // Needs to be disposed when refs becomes 0.
   Cache::Handle* table_reader_handle;
@@ -108,9 +116,7 @@ struct FileMetaData {
                                // file.
 
   FileMetaData()
-      : smallest_seqno(kMaxSequenceNumber),
-        largest_seqno(0),
-        table_reader_handle(nullptr),
+      : table_reader_handle(nullptr),
         compensated_file_size(0),
         num_entries(0),
         num_deletions(0),
@@ -128,8 +134,8 @@ struct FileMetaData {
       smallest.DecodeFrom(key);
     }
     largest.DecodeFrom(key);
-    smallest_seqno = std::min(smallest_seqno, seqno);
-    largest_seqno = std::max(largest_seqno, seqno);
+    fd.smallest_seqno = std::min(fd.smallest_seqno, seqno);
+    fd.largest_seqno = std::max(fd.largest_seqno, seqno);
   }
 
   // Unlike UpdateBoundaries, ranges do not need to be presented in any
@@ -143,8 +149,8 @@ struct FileMetaData {
     if (largest.size() == 0 || icmp.Compare(largest, end) < 0) {
       largest = end;
     }
-    smallest_seqno = std::min(smallest_seqno, seqno);
-    largest_seqno = std::max(largest_seqno, seqno);
+    fd.smallest_seqno = std::min(fd.smallest_seqno, seqno);
+    fd.largest_seqno = std::max(fd.largest_seqno, seqno);
   }
 };
 
@@ -233,17 +239,18 @@ class VersionEdit {
                bool marked_for_compaction) {
     assert(smallest_seqno <= largest_seqno);
     FileMetaData f;
-    f.fd = FileDescriptor(file, file_path_id, file_size);
+    f.fd = FileDescriptor(file, file_path_id, file_size, smallest_seqno,
+                          largest_seqno);
     f.smallest = smallest;
     f.largest = largest;
-    f.smallest_seqno = smallest_seqno;
-    f.largest_seqno = largest_seqno;
+    f.fd.smallest_seqno = smallest_seqno;
+    f.fd.largest_seqno = largest_seqno;
     f.marked_for_compaction = marked_for_compaction;
     new_files_.emplace_back(level, std::move(f));
   }
 
   void AddFile(int level, const FileMetaData& f) {
-    assert(f.smallest_seqno <= f.largest_seqno);
+    assert(f.fd.smallest_seqno <= f.fd.largest_seqno);
     new_files_.emplace_back(level, f);
   }
 
@@ -293,6 +300,11 @@ class VersionEdit {
     return new_files_;
   }
 
+  void MarkAtomicGroup(uint32_t remaining_entries) {
+    is_in_atomic_group_ = true;
+    remaining_entries_ = remaining_entries;
+  }
+
   std::string DebugString(bool hex_key = false) const;
   std::string DebugJSON(int edit_num, bool hex_key = false) const;
 
@@ -322,7 +334,7 @@ class VersionEdit {
   DeletedFileSet deleted_files_;
   std::vector<std::pair<int, FileMetaData>> new_files_;
 
-  // Each version edit record should have column_family_id set
+  // Each version edit record should have column_family_ set
   // If it's not set, it is default (0)
   uint32_t column_family_;
   // a version edit can be either column_family add or
@@ -331,6 +343,9 @@ class VersionEdit {
   bool is_column_family_drop_;
   bool is_column_family_add_;
   std::string column_family_name_;
+
+  bool is_in_atomic_group_;
+  uint32_t remaining_entries_;
 };
 
 }  // namespace rocksdb

db/version_edit_test.cc

@@ -191,6 +191,12 @@ TEST_F(VersionEditTest, MinLogNumberToKeep) {
   TestEncodeDecode(edit);
 }
 
+TEST_F(VersionEditTest, AtomicGroupTest) {
+  VersionEdit edit;
+  edit.MarkAtomicGroup(1);
+  TestEncodeDecode(edit);
+}
+
 }  // namespace rocksdb
 
 int main(int argc, char** argv) {

db/version_set.cc

@@ -897,8 +897,8 @@ void Version::GetColumnFamilyMetaData(ColumnFamilyMetaData* cf_meta) {
       }
       files.emplace_back(
           MakeTableFileName("", file->fd.GetNumber()), file_path,
-          file->fd.GetFileSize(), file->smallest_seqno, file->largest_seqno,
-          file->smallest.user_key().ToString(),
+          file->fd.GetFileSize(), file->fd.smallest_seqno,
+          file->fd.largest_seqno, file->smallest.user_key().ToString(),
           file->largest.user_key().ToString(),
           file->stats.num_reads_sampled.load(std::memory_order_relaxed),
           file->being_compacted);
@@ -1212,12 +1212,9 @@ void Version::Get(const ReadOptions& read_options, const LookupKey& k,
 
     // report the counters before returning
     if (get_context.State() != GetContext::kNotFound &&
-        get_context.State() != GetContext::kMerge) {
-      for (uint32_t t = 0; t < Tickers::TICKER_ENUM_MAX; t++) {
-        if (get_context.tickers_value[t] > 0) {
-          RecordTick(db_statistics_, t, get_context.tickers_value[t]);
-        }
-      }
+        get_context.State() != GetContext::kMerge &&
+        db_statistics_ != nullptr) {
+      get_context.ReportCounters();
     }
     switch (get_context.State()) {
       case GetContext::kNotFound:
@@ -1251,10 +1248,8 @@ void Version::Get(const ReadOptions& read_options, const LookupKey& k,
     f = fp.GetNextFile();
   }
 
-  for (uint32_t t = 0; t < Tickers::TICKER_ENUM_MAX; t++) {
-    if (get_context.tickers_value[t] > 0) {
-      RecordTick(db_statistics_, t, get_context.tickers_value[t]);
-    }
+  if (db_statistics_ != nullptr) {
+    get_context.ReportCounters();
   }
   if (GetContext::kMerge == get_context.State()) {
     if (!merge_operator_) {
@@ -1665,8 +1660,8 @@ void VersionStorageInfo::ComputeCompactionScore(
   }
   ComputeFilesMarkedForCompaction();
   ComputeBottommostFilesMarkedForCompaction();
-  if (immutable_cf_options.ttl > 0) {
-    ComputeExpiredTtlFiles(immutable_cf_options);
+  if (mutable_cf_options.ttl > 0) {
+    ComputeExpiredTtlFiles(immutable_cf_options, mutable_cf_options.ttl);
   }
   EstimateCompactionBytesNeeded(mutable_cf_options);
 }
@@ -1695,8 +1690,8 @@ void VersionStorageInfo::ComputeFilesMarkedForCompaction() {
 }
 
 void VersionStorageInfo::ComputeExpiredTtlFiles(
-    const ImmutableCFOptions& ioptions) {
-  assert(ioptions.ttl > 0);
+    const ImmutableCFOptions& ioptions, const uint64_t ttl) {
+  assert(ttl > 0);
 
   expired_ttl_files_.clear();
 
@@ -1713,8 +1708,7 @@ void VersionStorageInfo::ComputeExpiredTtlFiles(
           f->fd.table_reader->GetTableProperties() != nullptr) {
         auto creation_time =
             f->fd.table_reader->GetTableProperties()->creation_time;
-        if (creation_time > 0 &&
-            creation_time < (current_time - ioptions.ttl)) {
+        if (creation_time > 0 && creation_time < (current_time - ttl)) {
           expired_ttl_files_.emplace_back(level, f);
         }
       }
@@ -1897,13 +1891,15 @@ void VersionStorageInfo::UpdateFilesByCompactionPri(
       case kOldestLargestSeqFirst:
         std::sort(temp.begin(), temp.end(),
                   [](const Fsize& f1, const Fsize& f2) -> bool {
-                    return f1.file->largest_seqno < f2.file->largest_seqno;
+                    return f1.file->fd.largest_seqno <
+                           f2.file->fd.largest_seqno;
                   });
         break;
       case kOldestSmallestSeqFirst:
         std::sort(temp.begin(), temp.end(),
                   [](const Fsize& f1, const Fsize& f2) -> bool {
-                    return f1.file->smallest_seqno < f2.file->smallest_seqno;
+                    return f1.file->fd.smallest_seqno <
+                           f2.file->fd.smallest_seqno;
                   });
         break;
       case kMinOverlappingRatio:
@@ -1987,17 +1983,17 @@ void VersionStorageInfo::ComputeBottommostFilesMarkedForCompaction() {
   bottommost_files_mark_threshold_ = kMaxSequenceNumber;
   for (auto& level_and_file : bottommost_files_) {
     if (!level_and_file.second->being_compacted &&
-        level_and_file.second->largest_seqno != 0 &&
+        level_and_file.second->fd.largest_seqno != 0 &&
         level_and_file.second->num_deletions > 1) {
       // largest_seqno might be nonzero due to containing the final key in an
       // earlier compaction, whose seqnum we didn't zero out. Multiple deletions
       // ensures the file really contains deleted or overwritten keys.
-      if (level_and_file.second->largest_seqno < oldest_snapshot_seqnum_) {
+      if (level_and_file.second->fd.largest_seqno < oldest_snapshot_seqnum_) {
         bottommost_files_marked_for_compaction_.push_back(level_and_file);
       } else {
         bottommost_files_mark_threshold_ =
             std::min(bottommost_files_mark_threshold_,
-                     level_and_file.second->largest_seqno);
+                     level_and_file.second->fd.largest_seqno);
       }
     }
   }
@@ -2036,7 +2032,7 @@ bool VersionStorageInfo::OverlapInLevel(int level,
 void VersionStorageInfo::GetOverlappingInputs(
     int level, const InternalKey* begin, const InternalKey* end,
     std::vector<FileMetaData*>* inputs, int hint_index, int* file_index,
-    bool expand_range) const {
+    bool expand_range, InternalKey** next_smallest) const {
   if (level >= num_non_empty_levels_) {
     // this level is empty, no overlapping inputs
     return;
@@ -2055,11 +2051,17 @@ void VersionStorageInfo::GetOverlappingInputs(
   }
   const Comparator* user_cmp = user_comparator_;
   if (level > 0) {
-    GetOverlappingInputsRangeBinarySearch(level, begin, end, inputs,
-                                          hint_index, file_index);
+    GetOverlappingInputsRangeBinarySearch(level, begin, end, inputs, hint_index,
+                                          file_index, false, next_smallest);
     return;
   }
 
+  if (next_smallest) {
+    // next_smallest key only makes sense for non-level 0, where files are
+    // non-overlapping
+    *next_smallest = nullptr;
+  }
+
   for (size_t i = 0; i < level_files_brief_[level].num_files; ) {
     FdWithKeyRange* f = &(level_files_brief_[level].files[i++]);
     const Slice file_start = ExtractUserKey(f->smallest_key);
@@ -2187,7 +2189,7 @@ int sstableKeyCompare(const Comparator* user_cmp,
 void VersionStorageInfo::GetOverlappingInputsRangeBinarySearch(
     int level, const InternalKey* begin, const InternalKey* end,
     std::vector<FileMetaData*>* inputs, int hint_index, int* file_index,
-    bool within_interval) const {
+    bool within_interval, InternalKey** next_smallest) const {
   assert(level > 0);
   int min = 0;
   int mid = 0;
@@ -2223,6 +2225,9 @@ void VersionStorageInfo::GetOverlappingInputsRangeBinarySearch(
 
   // If there were no overlapping files, return immediately.
   if (!foundOverlap) {
+    if (next_smallest) {
+      next_smallest = nullptr;
+    }
     return;
   }
   // returns the index where an overlap is found
@@ -2243,6 +2248,15 @@ void VersionStorageInfo::GetOverlappingInputsRangeBinarySearch(
   for (int i = start_index; i <= end_index; i++) {
     inputs->push_back(files_[level][i]);
   }
+
+  if (next_smallest != nullptr) {
+    // Provide the next key outside the range covered by inputs
+    if (++end_index < static_cast<int>(files_[level].size())) {
+      **next_smallest = files_[level][end_index]->smallest;
+    } else {
+      *next_smallest = nullptr;
+    }
+  }
 }
 
 // Store in *start_index and *end_index the range of all files in
@@ -2423,7 +2437,7 @@ const char* VersionStorageInfo::LevelFileSummary(FileSummaryStorage* scratch,
     AppendHumanBytes(f->fd.GetFileSize(), sztxt, sizeof(sztxt));
     int ret = snprintf(scratch->buffer + len, sz,
                        "#%" PRIu64 "(seq=%" PRIu64 ",sz=%s,%d) ",
-                       f->fd.GetNumber(), f->smallest_seqno, sztxt,
+                       f->fd.GetNumber(), f->fd.smallest_seqno, sztxt,
                        static_cast<int>(f->being_compacted));
     if (ret < 0 || ret >= sz)
       break;
@@ -2995,25 +3009,31 @@ Status VersionSet::ProcessManifestWrites(
         delete first_writer.cfd;
       }
     } else {
-      uint64_t max_log_number_in_batch  = 0;
+      // Each version in versions corresponds to a column family.
+      // For each column family, update its log number indicating that logs
+      // with number smaller than this should be ignored.
+      for (const auto version : versions) {
+        uint64_t max_log_number_in_batch = 0;
+        uint32_t cf_id = version->cfd_->GetID();
+        for (const auto& e : batch_edits) {
+          if (e->has_log_number_ && e->column_family_ == cf_id) {
+            max_log_number_in_batch =
+                std::max(max_log_number_in_batch, e->log_number_);
+          }
+        }
+        if (max_log_number_in_batch != 0) {
+          assert(version->cfd_->GetLogNumber() <= max_log_number_in_batch);
+          version->cfd_->SetLogNumber(max_log_number_in_batch);
+        }
+      }
+
       uint64_t last_min_log_number_to_keep = 0;
       for (auto& e : batch_edits) {
-        if (e->has_log_number_) {
-          max_log_number_in_batch =
-              std::max(max_log_number_in_batch, e->log_number_);
-        }
         if (e->has_min_log_number_to_keep_) {
           last_min_log_number_to_keep =
               std::max(last_min_log_number_to_keep, e->min_log_number_to_keep_);
         }
       }
-      if (max_log_number_in_batch != 0) {
-        for (int i = 0; i < static_cast<int>(versions.size()); ++i) {
-          ColumnFamilyData* cfd = versions[i]->cfd_;
-          assert(cfd->GetLogNumber() <= max_log_number_in_batch);
-          cfd->SetLogNumber(max_log_number_in_batch);
-        }
-      }
 
       if (last_min_log_number_to_keep != 0) {
         // Should only be set in 2PC mode.
@@ -3206,6 +3226,133 @@ void VersionSet::LogAndApplyHelper(ColumnFamilyData* cfd,
   builder->Apply(edit);
 }
 
+Status VersionSet::ApplyOneVersionEdit(
+    VersionEdit& edit,
+    const std::unordered_map<std::string, ColumnFamilyOptions>& name_to_options,
+    std::unordered_map<int, std::string>& column_families_not_found,
+    std::unordered_map<uint32_t, BaseReferencedVersionBuilder*>& builders,
+    bool* have_log_number, uint64_t* /* log_number */,
+    bool* have_prev_log_number, uint64_t* previous_log_number,
+    bool* have_next_file, uint64_t* next_file, bool* have_last_sequence,
+    SequenceNumber* last_sequence, uint64_t* min_log_number_to_keep,
+    uint32_t* max_column_family) {
+  // Not found means that user didn't supply that column
+  // family option AND we encountered column family add
+  // record. Once we encounter column family drop record,
+  // we will delete the column family from
+  // column_families_not_found.
+  bool cf_in_not_found = (column_families_not_found.find(edit.column_family_) !=
+                          column_families_not_found.end());
+  // in builders means that user supplied that column family
+  // option AND that we encountered column family add record
+  bool cf_in_builders = builders.find(edit.column_family_) != builders.end();
+
+  // they can't both be true
+  assert(!(cf_in_not_found && cf_in_builders));
+
+  ColumnFamilyData* cfd = nullptr;
+
+  if (edit.is_column_family_add_) {
+    if (cf_in_builders || cf_in_not_found) {
+      return Status::Corruption(
+          "Manifest adding the same column family twice: " +
+          edit.column_family_name_);
+    }
+    auto cf_options = name_to_options.find(edit.column_family_name_);
+    if (cf_options == name_to_options.end()) {
+      column_families_not_found.insert(
+          {edit.column_family_, edit.column_family_name_});
+    } else {
+      cfd = CreateColumnFamily(cf_options->second, &edit);
+      cfd->set_initialized();
+      builders.insert(
+          {edit.column_family_, new BaseReferencedVersionBuilder(cfd)});
+    }
+  } else if (edit.is_column_family_drop_) {
+    if (cf_in_builders) {
+      auto builder = builders.find(edit.column_family_);
+      assert(builder != builders.end());
+      delete builder->second;
+      builders.erase(builder);
+      cfd = column_family_set_->GetColumnFamily(edit.column_family_);
+      assert(cfd != nullptr);
+      if (cfd->Unref()) {
+        delete cfd;
+        cfd = nullptr;
+      } else {
+        // who else can have reference to cfd!?
+        assert(false);
+      }
+    } else if (cf_in_not_found) {
+      column_families_not_found.erase(edit.column_family_);
+    } else {
+      return Status::Corruption(
+          "Manifest - dropping non-existing column family");
+    }
+  } else if (!cf_in_not_found) {
+    if (!cf_in_builders) {
+      return Status::Corruption(
+          "Manifest record referencing unknown column family");
+    }
+
+    cfd = column_family_set_->GetColumnFamily(edit.column_family_);
+    // this should never happen since cf_in_builders is true
+    assert(cfd != nullptr);
+
+    // if it is not column family add or column family drop,
+    // then it's a file add/delete, which should be forwarded
+    // to builder
+    auto builder = builders.find(edit.column_family_);
+    assert(builder != builders.end());
+    builder->second->version_builder()->Apply(&edit);
+  }
+
+  if (cfd != nullptr) {
+    if (edit.has_log_number_) {
+      if (cfd->GetLogNumber() > edit.log_number_) {
+        ROCKS_LOG_WARN(
+            db_options_->info_log,
+            "MANIFEST corruption detected, but ignored - Log numbers in "
+            "records NOT monotonically increasing");
+      } else {
+        cfd->SetLogNumber(edit.log_number_);
+        *have_log_number = true;
+      }
+    }
+    if (edit.has_comparator_ &&
+        edit.comparator_ != cfd->user_comparator()->Name()) {
+      return Status::InvalidArgument(
+          cfd->user_comparator()->Name(),
+          "does not match existing comparator " + edit.comparator_);
+    }
+  }
+
+  if (edit.has_prev_log_number_) {
+    *previous_log_number = edit.prev_log_number_;
+    *have_prev_log_number = true;
+  }
+
+  if (edit.has_next_file_number_) {
+    *next_file = edit.next_file_number_;
+    *have_next_file = true;
+  }
+
+  if (edit.has_max_column_family_) {
+    *max_column_family = edit.max_column_family_;
+  }
+
+  if (edit.has_min_log_number_to_keep_) {
+    *min_log_number_to_keep =
+        std::max(*min_log_number_to_keep, edit.min_log_number_to_keep_);
+  }
+
+  if (edit.has_last_sequence_) {
+    *last_sequence = edit.last_sequence_;
+    *have_last_sequence = true;
+  }
+  return Status::OK();
+}
+
 Status VersionSet::Recover(
     const std::vector<ColumnFamilyDescriptor>& column_families,
     bool read_only) {
@@ -3294,6 +3441,8 @@ Status VersionSet::Recover(
                        true /*checksum*/, 0 /*initial_offset*/, 0);
     Slice record;
     std::string scratch;
+    std::vector<VersionEdit> replay_buffer;
+    size_t num_entries_decoded = 0;
     while (reader.ReadRecord(&record, &scratch) && s.ok()) {
       VersionEdit edit;
       s = edit.DecodeFrom(record);
@@ -3301,123 +3450,44 @@ Status VersionSet::Recover(
         break;
       }
 
-      // Not found means that user didn't supply that column
-      // family option AND we encountered column family add
-      // record. Once we encounter column family drop record,
-      // we will delete the column family from
-      // column_families_not_found.
-      bool cf_in_not_found =
-          column_families_not_found.find(edit.column_family_) !=
-          column_families_not_found.end();
-      // in builders means that user supplied that column family
-      // option AND that we encountered column family add record
-      bool cf_in_builders =
-          builders.find(edit.column_family_) != builders.end();
-
-      // they can't both be true
-      assert(!(cf_in_not_found && cf_in_builders));
-
-      ColumnFamilyData* cfd = nullptr;
-
-      if (edit.is_column_family_add_) {
-        if (cf_in_builders || cf_in_not_found) {
-          s = Status::Corruption(
-              "Manifest adding the same column family twice");
-          break;
+      if (edit.is_in_atomic_group_) {
+        if (replay_buffer.empty()) {
+          replay_buffer.resize(edit.remaining_entries_ + 1);
         }
-        auto cf_options = cf_name_to_options.find(edit.column_family_name_);
-        if (cf_options == cf_name_to_options.end()) {
-          column_families_not_found.insert(
-              {edit.column_family_, edit.column_family_name_});
-        } else {
-          cfd = CreateColumnFamily(cf_options->second, &edit);
-          cfd->set_initialized();
-          builders.insert(
-              {edit.column_family_, new BaseReferencedVersionBuilder(cfd)});
+        ++num_entries_decoded;
+        if (num_entries_decoded + edit.remaining_entries_ !=
+            static_cast<uint32_t>(replay_buffer.size())) {
+          return Status::Corruption("corrupted atomic group");
         }
-      } else if (edit.is_column_family_drop_) {
-        if (cf_in_builders) {
-          auto builder = builders.find(edit.column_family_);
-          assert(builder != builders.end());
-          delete builder->second;
-          builders.erase(builder);
-          cfd = column_family_set_->GetColumnFamily(edit.column_family_);
-          if (cfd->Unref()) {
-            delete cfd;
-            cfd = nullptr;
-          } else {
-            // who else can have reference to cfd!?
-            assert(false);
+        replay_buffer[num_entries_decoded - 1] = std::move(edit);
+        if (num_entries_decoded == replay_buffer.size()) {
+          for (auto& e : replay_buffer) {
+            s = ApplyOneVersionEdit(
+                e, cf_name_to_options, column_families_not_found, builders,
+                &have_log_number, &log_number, &have_prev_log_number,
+                &previous_log_number, &have_next_file, &next_file,
+                &have_last_sequence, &last_sequence, &min_log_number_to_keep,
+                &max_column_family);
+            if (!s.ok()) {
+              break;
+            }
           }
-        } else if (cf_in_not_found) {
-          column_families_not_found.erase(edit.column_family_);
-        } else {
-          s = Status::Corruption(
-              "Manifest - dropping non-existing column family");
-          break;
+          replay_buffer.clear();
+          num_entries_decoded = 0;
         }
-      } else if (!cf_in_not_found) {
-        if (!cf_in_builders) {
-          s = Status::Corruption(
-              "Manifest record referencing unknown column family");
-          break;
+      } else {
+        if (!replay_buffer.empty()) {
+          return Status::Corruption("corrupted atomic group");
         }
-
-        cfd = column_family_set_->GetColumnFamily(edit.column_family_);
-        // this should never happen since cf_in_builders is true
-        assert(cfd != nullptr);
-
-        // if it is not column family add or column family drop,
-        // then it's a file add/delete, which should be forwarded
-        // to builder
-        auto builder = builders.find(edit.column_family_);
-        assert(builder != builders.end());
-        builder->second->version_builder()->Apply(&edit);
+        s = ApplyOneVersionEdit(
+            edit, cf_name_to_options, column_families_not_found, builders,
+            &have_log_number, &log_number, &have_prev_log_number,
+            &previous_log_number, &have_next_file, &next_file,
+            &have_last_sequence, &last_sequence, &min_log_number_to_keep,
+            &max_column_family);
       }
-
-      if (cfd != nullptr) {
-        if (edit.has_log_number_) {
-          if (cfd->GetLogNumber() > edit.log_number_) {
-            ROCKS_LOG_WARN(
-                db_options_->info_log,
-                "MANIFEST corruption detected, but ignored - Log numbers in "
-                "records NOT monotonically increasing");
-          } else {
-            cfd->SetLogNumber(edit.log_number_);
-            have_log_number = true;
-          }
-        }
-        if (edit.has_comparator_ &&
-            edit.comparator_ != cfd->user_comparator()->Name()) {
-          s = Status::InvalidArgument(
-              cfd->user_comparator()->Name(),
-              "does not match existing comparator " + edit.comparator_);
-          break;
-        }
-      }
-
-      if (edit.has_prev_log_number_) {
-        previous_log_number = edit.prev_log_number_;
-        have_prev_log_number = true;
-      }
-
-      if (edit.has_next_file_number_) {
-        next_file = edit.next_file_number_;
-        have_next_file = true;
-      }
-
-      if (edit.has_max_column_family_) {
-        max_column_family = edit.max_column_family_;
-      }
-
-      if (edit.has_min_log_number_to_keep_) {
-        min_log_number_to_keep =
-            std::max(min_log_number_to_keep, edit.min_log_number_to_keep_);
-      }
-
-      if (edit.has_last_sequence_) {
-        last_sequence = edit.last_sequence_;
-        have_last_sequence = true;
+      if (!s.ok()) {
+        break;
       }
     }
   }
@@ -3963,7 +4033,7 @@ Status VersionSet::WriteSnapshot(log::Writer* log) {
              cfd->current()->storage_info()->LevelFiles(level)) {
           edit.AddFile(level, f->fd.GetNumber(), f->fd.GetPathId(),
                        f->fd.GetFileSize(), f->smallest, f->largest,
-                       f->smallest_seqno, f->largest_seqno,
+                       f->fd.smallest_seqno, f->fd.largest_seqno,
                        f->marked_for_compaction);
         }
       }
@@ -4288,8 +4358,8 @@ void VersionSet::GetLiveFilesMetaData(std::vector<LiveFileMetaData>* metadata) {
         filemetadata.size = file->fd.GetFileSize();
         filemetadata.smallestkey = file->smallest.user_key().ToString();
         filemetadata.largestkey = file->largest.user_key().ToString();
-        filemetadata.smallest_seqno = file->smallest_seqno;
-        filemetadata.largest_seqno = file->largest_seqno;
+        filemetadata.smallest_seqno = file->fd.smallest_seqno;
+        filemetadata.largest_seqno = file->fd.largest_seqno;
         metadata->push_back(filemetadata);
       }
     }

db/version_set.h

@@ -52,7 +52,6 @@ class Writer;
 }
 
 class Compaction;
-class InternalIterator;
 class LogBuffer;
 class LookupKey;
 class MemTable;
@@ -137,7 +136,8 @@ class VersionStorageInfo {
 
   // This computes ttl_expired_files_ and is called by
   // ComputeCompactionScore()
-  void ComputeExpiredTtlFiles(const ImmutableCFOptions& ioptions);
+  void ComputeExpiredTtlFiles(const ImmutableCFOptions& ioptions,
+                              const uint64_t ttl);
 
   // This computes bottommost_files_marked_for_compaction_ and is called by
   // ComputeCompactionScore() or UpdateOldestSnapshot().
@@ -188,9 +188,11 @@ class VersionStorageInfo {
       std::vector<FileMetaData*>* inputs,
       int hint_index = -1,        // index of overlap file
       int* file_index = nullptr,  // return index of overlap file
-      bool expand_range = true)   // if set, returns files which overlap the
-      const;                      // range and overlap each other. If false,
+      bool expand_range = true,   // if set, returns files which overlap the
+                                  // range and overlap each other. If false,
                                   // then just files intersecting the range
+      InternalKey** next_smallest = nullptr)  // if non-null, returns the
+      const;  // smallest key of next file not included
   void GetCleanInputsWithinInterval(
       int level, const InternalKey* begin,  // nullptr means before all keys
       const InternalKey* end,               // nullptr means after all keys
@@ -200,14 +202,15 @@ class VersionStorageInfo {
       const;
 
   void GetOverlappingInputsRangeBinarySearch(
-      int level,           // level > 0
+      int level,                 // level > 0
       const InternalKey* begin,  // nullptr means before all keys
       const InternalKey* end,    // nullptr means after all keys
       std::vector<FileMetaData*>* inputs,
       int hint_index,                // index of overlap file
       int* file_index,               // return index of overlap file
-      bool within_interval = false)  // if set, force the inputs within interval
-      const;
+      bool within_interval = false,  // if set, force the inputs within interval
+      InternalKey** next_smallest = nullptr)  // if non-null, returns the
+      const;  // smallest key of next file not included
 
   void ExtendFileRangeOverlappingInterval(
       int level,
@@ -728,6 +731,10 @@ struct ObsoleteFileInfo {
   }
 };
 
+namespace {
+class BaseReferencedVersionBuilder;
+}
+
 class VersionSet {
  public:
   VersionSet(const std::string& dbname, const ImmutableDBOptions* db_options,
@@ -831,6 +838,11 @@ class VersionSet {
   // Allocate and return a new file number
   uint64_t NewFileNumber() { return next_file_number_.fetch_add(1); }
 
+  // Fetch And Add n new file number
+  uint64_t FetchAddFileNumber(uint64_t n) {
+    return next_file_number_.fetch_add(n);
+  }
+
   // Return the last sequence number.
   uint64_t LastSequence() const {
     return last_sequence_.load(std::memory_order_acquire);
@@ -984,6 +996,16 @@ class VersionSet {
   ColumnFamilyData* CreateColumnFamily(const ColumnFamilyOptions& cf_options,
                                        VersionEdit* edit);
 
+  Status ApplyOneVersionEdit(
+      VersionEdit& edit,
+      const std::unordered_map<std::string, ColumnFamilyOptions>& name_to_opts,
+      std::unordered_map<int, std::string>& column_families_not_found,
+      std::unordered_map<uint32_t, BaseReferencedVersionBuilder*>& builders,
+      bool* have_log_number, uint64_t* log_number, bool* have_prev_log_number,
+      uint64_t* previous_log_number, bool* have_next_file, uint64_t* next_file,
+      bool* have_last_sequence, SequenceNumber* last_sequence,
+      uint64_t* min_log_number_to_keep, uint32_t* max_column_family);
+
   Status ProcessManifestWrites(std::deque<ManifestWriter>& writers,
                                InstrumentedMutex* mu, Directory* db_directory,
                                bool new_descriptor_log,

db/write_batch.cc

@@ -727,6 +727,11 @@ Status WriteBatchInternal::MarkEndPrepare(WriteBatch* b, const Slice& xid,
                               ContentFlags::HAS_END_PREPARE |
                               ContentFlags::HAS_BEGIN_PREPARE,
                           std::memory_order_relaxed);
+  if (unprepared_batch) {
+    b->content_flags_.store(b->content_flags_.load(std::memory_order_relaxed) |
+                                ContentFlags::HAS_BEGIN_UNPREPARE,
+                            std::memory_order_relaxed);
+  }
   return Status::OK();
 }
 

db/write_thread.cc

@@ -24,7 +24,10 @@ WriteThread::WriteThread(const ImmutableDBOptions& db_options)
       enable_pipelined_write_(db_options.enable_pipelined_write),
       newest_writer_(nullptr),
       newest_memtable_writer_(nullptr),
-      last_sequence_(0) {}
+      last_sequence_(0),
+      write_stall_dummy_(),
+      stall_mu_(),
+      stall_cv_(&stall_mu_) {}
 
 uint8_t WriteThread::BlockingAwaitState(Writer* w, uint8_t goal_mask) {
   // We're going to block.  Lazily create the mutex.  We guarantee
@@ -219,6 +222,28 @@ bool WriteThread::LinkOne(Writer* w, std::atomic<Writer*>* newest_writer) {
   assert(w->state == STATE_INIT);
   Writer* writers = newest_writer->load(std::memory_order_relaxed);
   while (true) {
+    // If write stall in effect, and w->no_slowdown is not true,
+    // block here until stall is cleared. If its true, then return
+    // immediately
+    if (writers == &write_stall_dummy_) {
+      if (w->no_slowdown) {
+        w->status = Status::Incomplete("Write stall");
+        SetState(w, STATE_COMPLETED);
+        return false;
+      }
+      // Since no_slowdown is false, wait here to be notified of the write
+      // stall clearing
+      {
+        MutexLock lock(&stall_mu_);
+        writers = newest_writer->load(std::memory_order_relaxed);
+        if (writers == &write_stall_dummy_) {
+          stall_cv_.Wait();
+          // Load newest_writers_ again since it may have changed
+          writers = newest_writer->load(std::memory_order_relaxed);
+          continue;
+        }
+      }
+    }
     w->link_older = writers;
     if (newest_writer->compare_exchange_weak(writers, w)) {
       return (writers == nullptr);
@@ -263,6 +288,17 @@ void WriteThread::CreateMissingNewerLinks(Writer* head) {
   }
 }
 
+WriteThread::Writer* WriteThread::FindNextLeader(Writer* from,
+                                                 Writer* boundary) {
+  assert(from != nullptr && from != boundary);
+  Writer* current = from;
+  while (current->link_older != boundary) {
+    current = current->link_older;
+    assert(current != nullptr);
+  }
+  return current;
+}
+
 void WriteThread::CompleteLeader(WriteGroup& write_group) {
   assert(write_group.size > 0);
   Writer* leader = write_group.leader;
@@ -292,12 +328,44 @@ void WriteThread::CompleteFollower(Writer* w, WriteGroup& write_group) {
   SetState(w, STATE_COMPLETED);
 }
 
+void WriteThread::BeginWriteStall() {
+  LinkOne(&write_stall_dummy_, &newest_writer_);
+
+  // Walk writer list until w->write_group != nullptr. The current write group
+  // will not have a mix of slowdown/no_slowdown, so its ok to stop at that
+  // point
+  Writer* w = write_stall_dummy_.link_older;
+  Writer* prev = &write_stall_dummy_;
+  while (w != nullptr && w->write_group == nullptr) {
+    if (w->no_slowdown) {
+      prev->link_older = w->link_older;
+      w->status = Status::Incomplete("Write stall");
+      SetState(w, STATE_COMPLETED);
+      w = prev->link_older;
+    } else {
+      prev = w;
+      w = w->link_older;
+    }
+  }
+}
+
+void WriteThread::EndWriteStall() {
+  MutexLock lock(&stall_mu_);
+
+  assert(newest_writer_.load(std::memory_order_relaxed) == &write_stall_dummy_);
+  newest_writer_.exchange(write_stall_dummy_.link_older);
+
+  // Wake up writers
+  stall_cv_.SignalAll();
+}
+
 static WriteThread::AdaptationContext jbg_ctx("JoinBatchGroup");
 void WriteThread::JoinBatchGroup(Writer* w) {
   TEST_SYNC_POINT_CALLBACK("WriteThread::JoinBatchGroup:Start", w);
   assert(w->batch != nullptr);
 
   bool linked_as_leader = LinkOne(w, &newest_writer_);
+
   if (linked_as_leader) {
     SetState(w, STATE_GROUP_LEADER);
   }
@@ -558,21 +626,49 @@ void WriteThread::ExitAsBatchGroupLeader(WriteGroup& write_group,
     if (!leader->ShouldWriteToMemtable()) {
       CompleteLeader(write_group);
     }
+
+    Writer* next_leader = nullptr;
+
+    // Look for next leader before we call LinkGroup. If there isn't
+    // pending writers, place a dummy writer at the tail of the queue
+    // so we know the boundary of the current write group.
+    Writer dummy;
+    Writer* expected = last_writer;
+    bool has_dummy = newest_writer_.compare_exchange_strong(expected, &dummy);
+    if (!has_dummy) {
+      // We find at least one pending writer when we insert dummy. We search
+      // for next leader from there.
+      next_leader = FindNextLeader(expected, last_writer);
+      assert(next_leader != nullptr && next_leader != last_writer);
+    }
+
     // Link the ramaining of the group to memtable writer list.
+    //
+    // We have to link our group to memtable writer queue before wake up the
+    // next leader or set newest_writer_ to null, otherwise the next leader
+    // can run ahead of us and link to memtable writer queue before we do.
     if (write_group.size > 0) {
       if (LinkGroup(write_group, &newest_memtable_writer_)) {
         // The leader can now be different from current writer.
         SetState(write_group.leader, STATE_MEMTABLE_WRITER_LEADER);
       }
     }
-    // Reset newest_writer_ and wake up the next leader.
-    Writer* newest_writer = last_writer;
-    if (!newest_writer_.compare_exchange_strong(newest_writer, nullptr)) {
-      Writer* next_leader = newest_writer;
-      while (next_leader->link_older != last_writer) {
-        next_leader = next_leader->link_older;
-        assert(next_leader != nullptr);
+
+    // If we have inserted dummy in the queue, remove it now and check if there
+    // are pending writer join the queue since we insert the dummy. If so,
+    // look for next leader again.
+    if (has_dummy) {
+      assert(next_leader == nullptr);
+      expected = &dummy;
+      bool has_pending_writer =
+          !newest_writer_.compare_exchange_strong(expected, nullptr);
+      if (has_pending_writer) {
+        next_leader = FindNextLeader(expected, &dummy);
+        assert(next_leader != nullptr && next_leader != &dummy);
       }
+    }
+
+    if (next_leader != nullptr) {
       next_leader->link_older = nullptr;
       SetState(next_leader, STATE_GROUP_LEADER);
     }

db/write_thread.h

@@ -342,6 +342,13 @@ class WriteThread {
     return last_sequence_;
   }
 
+  // Insert a dummy writer at the tail of the write queue to indicate a write
+  // stall, and fail any writers in the queue with no_slowdown set to true
+  void BeginWriteStall();
+
+  // Remove the dummy writer and wake up waiting writers
+  void EndWriteStall();
+
  private:
   // See AwaitState.
   const uint64_t max_yield_usec_;
@@ -365,6 +372,17 @@ class WriteThread {
   // is not necessary visible to reads because the writer can be ongoing.
   SequenceNumber last_sequence_;
 
+  // A dummy writer to indicate a write stall condition. This will be inserted
+  // at the tail of the writer queue by the leader, so newer writers can just
+  // check for this and bail
+  Writer write_stall_dummy_;
+
+  // Mutex and condvar for writers to block on a write stall. During a write
+  // stall, writers with no_slowdown set to false will wait on this rather
+  // on the writer queue
+  port::Mutex stall_mu_;
+  port::CondVar stall_cv_;
+
   // Waits for w->state & goal_mask using w->StateMutex().  Returns
   // the state that satisfies goal_mask.
   uint8_t BlockingAwaitState(Writer* w, uint8_t goal_mask);
@@ -392,6 +410,10 @@ class WriteThread {
   // concurrently with itself.
   void CreateMissingNewerLinks(Writer* head);
 
+  // Starting from a pending writer, follow link_older to search for next
+  // leader, until we hit boundary.
+  Writer* FindNextLeader(Writer* pending_writer, Writer* boundary);
+
   // Set the leader in write_group to completed state and remove it from the
   // write group.
   void CompleteLeader(WriteGroup& write_group);

docs/_posts/2018-08-01-rocksdb-tuning-advisor.markdown

@@ -0,0 +1,58 @@
+---
+title: Rocksdb Tuning Advisor
+layout: post
+author: poojam23
+category: blog
+---
+
+The performance of Rocksdb is contingent on its tuning. However, because
+of the complexity of its underlying technology and a large number of
+configurable parameters, a good configuration is sometimes hard to obtain. The aim of
+the python command-line tool, Rocksdb Advisor, is to automate the process of
+suggesting improvements in the configuration based on advice from Rocksdb
+experts.
+
+### Overview
+
+Experts share their wisdom as rules comprising of conditions and suggestions in the INI format (refer
+[rules.ini](https://github.com/facebook/rocksdb/blob/master/tools/advisor/advisor/rules.ini)).
+Users provide the Rocksdb configuration that they want to improve upon (as the
+familiar Rocksdb OPTIONS file —
+[example](https://github.com/facebook/rocksdb/blob/master/examples/rocksdb_option_file_example.ini))
+and the path of the file which contains Rocksdb logs and statistics.
+The [Advisor](https://github.com/facebook/rocksdb/blob/master/tools/advisor/advisor/rule_parser_example.py)
+creates appropriate DataSource objects (for Rocksdb
+[logs](https://github.com/facebook/rocksdb/blob/master/tools/advisor/advisor/db_log_parser.py),
+[options](https://github.com/facebook/rocksdb/blob/master/tools/advisor/advisor/db_options_parser.py),
+[statistics](https://github.com/facebook/rocksdb/blob/master/tools/advisor/advisor/db_stats_fetcher.py) etc.)
+and provides them to the [Rules Engine](https://github.com/facebook/rocksdb/blob/master/tools/advisor/advisor/rule_parser.py).
+The Rules uses rules from experts to parse data-sources and trigger appropriate rules.
+The Advisor's output gives information about which rules were triggered,
+why they were triggered and what each of them suggests. Each suggestion
+provided by a triggered rule advises some action on a Rocksdb
+configuration option, for example, increase CFOptions.write_buffer_size,
+set bloom_bits to 2 etc.
+
+### Usage
+
+An example command to run the tool:
+
+```shell
+cd rocksdb/tools/advisor
+python3 -m advisor.rule_parser_example --rules_spec=advisor/rules.ini --rocksdb_options=test/input_files/OPTIONS-000005 --log_files_path_prefix=test/input_files/LOG-0 --stats_dump_period_sec=20
+```
+
+Sample output where a Rocksdb log-based rule has been triggered :
+
+```shell
+Rule: stall-too-many-memtables
+LogCondition: stall-too-many-memtables regex: Stopping writes because we have \d+ immutable memtables \(waiting for flush\), max_write_buffer_number is set to \d+
+Suggestion: inc-bg-flush option : DBOptions.max_background_flushes action : increase suggested_values : ['2']
+Suggestion: inc-write-buffer option : CFOptions.max_write_buffer_number action : increase
+scope: col_fam:
+{'default'}
+```
+
+### Read more
+
+For more information, refer to [advisor](https://github.com/facebook/rocksdb/tree/master/tools/advisor/README.md).

env/env_posix.cc

@@ -20,7 +20,7 @@
 #include <sys/ioctl.h>
 #include <sys/mman.h>
 #include <sys/stat.h>
-#if defined(OS_LINUX) || defined(OS_SOLARIS)
+#if defined(OS_LINUX) || defined(OS_SOLARIS) || defined(OS_ANDROID)
 #include <sys/statfs.h>
 #include <sys/syscall.h>
 #include <sys/sysmacros.h>

include/rocksdb/advanced_options.h

@@ -596,6 +596,8 @@ struct AdvancedColumnFamilyOptions {
   // Enabled only for level compaction for now.
   //
   // Default: 0 (disabled)
+  //
+  // Dynamically changeable through SetOptions() API
   uint64_t ttl = 0;
 
   // Create ColumnFamilyOptions with default values for all fields

include/rocksdb/c.h

@@ -831,6 +831,12 @@ rocksdb_options_set_min_write_buffer_number_to_merge(rocksdb_options_t*, int);
 extern ROCKSDB_LIBRARY_API void
 rocksdb_options_set_max_write_buffer_number_to_maintain(rocksdb_options_t*,
                                                         int);
+extern ROCKSDB_LIBRARY_API void rocksdb_options_set_enable_pipelined_write(
+    rocksdb_options_t*, unsigned char);
+extern ROCKSDB_LIBRARY_API void rocksdb_options_set_max_subcompactions(
+    rocksdb_options_t*, uint32_t);
+extern ROCKSDB_LIBRARY_API void rocksdb_options_set_max_background_jobs(
+    rocksdb_options_t*, int);
 extern ROCKSDB_LIBRARY_API void rocksdb_options_set_max_background_compactions(
     rocksdb_options_t*, int);
 extern ROCKSDB_LIBRARY_API void rocksdb_options_set_base_background_compactions(
@@ -1195,6 +1201,7 @@ extern ROCKSDB_LIBRARY_API void rocksdb_readoptions_set_read_tier(
     rocksdb_readoptions_t*, int);
 extern ROCKSDB_LIBRARY_API void rocksdb_readoptions_set_tailing(
     rocksdb_readoptions_t*, unsigned char);
+// The functionality that this option controlled has been removed.
 extern ROCKSDB_LIBRARY_API void rocksdb_readoptions_set_managed(
     rocksdb_readoptions_t*, unsigned char);
 extern ROCKSDB_LIBRARY_API void rocksdb_readoptions_set_readahead_size(

include/rocksdb/comparator.h

@@ -74,6 +74,12 @@ class Comparator {
                                               const Slice& /*t*/) const {
     return false;
   }
+
+  // return true if two keys with different byte sequences can be regarded
+  // as equal by this comparator.
+  // The major use case is to determine if DataBlockHashIndex is compatible
+  // with the customized comparator.
+  virtual bool CanKeysWithDifferentByteContentsBeEqual() const { return true; }
 };
 
 // Return a builtin comparator that uses lexicographic byte-wise

include/rocksdb/db.h

@@ -53,6 +53,7 @@ struct ExternalSstFileInfo;
 class WriteBatch;
 class Env;
 class EventListener;
+class TraceWriter;
 
 using std::unique_ptr;
 
@@ -996,11 +997,6 @@ class DB {
       std::vector<LiveFileMetaData>* /*metadata*/) {}
 
   // Obtains the meta data of the specified column family of the DB.
-  // Status::NotFound() will be returned if the current DB does not have
-  // any column family match the specified name.
-  //
-  // If cf_name is not specified, then the metadata of the default
-  // column family will be returned.
   virtual void GetColumnFamilyMetaData(ColumnFamilyHandle* /*column_family*/,
                                        ColumnFamilyMetaData* /*metadata*/) {}
 
@@ -1173,6 +1169,15 @@ class DB {
     return Status::NotSupported("PromoteL0() is not implemented.");
   }
 
+  // Trace DB operations. Use EndTrace() to stop tracing.
+  virtual Status StartTrace(const TraceOptions& /*options*/,
+                            std::unique_ptr<TraceWriter>&& /*trace_writer*/) {
+    return Status::NotSupported("StartTrace() is not implemented.");
+  }
+
+  virtual Status EndTrace() {
+    return Status::NotSupported("EndTrace() is not implemented.");
+  }
 #endif  // ROCKSDB_LITE
 
   // Needed for StackableDB

include/rocksdb/merge_operator.h

@@ -195,6 +195,11 @@ class MergeOperator {
   // during a point lookup, thereby helping in limiting the number of levels to
   // read from.
   // Doesn't help with iterators.
+  //
+  // Note: the merge operands are passed to this function in the reversed order
+  // relative to how they were merged (passed to FullMerge or FullMergeV2)
+  // for performance reasons, see also:
+  // https://github.com/facebook/rocksdb/issues/3865
   virtual bool ShouldMerge(const std::vector<Slice>& /*operands*/) const {
     return false;
   }

include/rocksdb/options.h

@@ -430,6 +430,8 @@ struct DBOptions {
   // (i.e. the ones that are causing all the space amplification). If set to 0
   // (default), we will dynamically choose the WAL size limit to be
   // [sum of all write_buffer_size * max_write_buffer_number] * 4
+  // This option takes effect only when there are more than one column family as
+  // otherwise the wal size is dictated by the write_buffer_size.
   // Default: 0
   uint64_t max_total_wal_size = 0;
 
@@ -1064,11 +1066,8 @@ struct ReadOptions {
   // Not supported in ROCKSDB_LITE mode!
   bool tailing;
 
-  // Specify to create a managed iterator -- a special iterator that
-  // uses less resources by having the ability to free its underlying
-  // resources on request.
-  // Default: false
-  // Not supported in ROCKSDB_LITE mode!
+  // This options is not used anymore. It was to turn on a functionality that
+  // has been removed.
   bool managed;
 
   // Enable a total order seek regardless of index format (e.g. hash index)
@@ -1184,8 +1183,13 @@ struct FlushOptions {
   // If true, the flush will wait until the flush is done.
   // Default: true
   bool wait;
-
-  FlushOptions() : wait(true) {}
+  // If true, the flush would proceed immediately even it means writes will
+  // stall for the duration of the flush; if false the operation will wait
+  // until it's possible to do flush w/o causing stall or until required flush
+  // is performed by someone else (foreground call or background thread).
+  // Default: false
+  bool allow_write_stall;
+  FlushOptions() : wait(true), allow_write_stall(false) {}
 };
 
 // Create a Logger from provided DBOptions
@@ -1197,6 +1201,9 @@ extern Status CreateLoggerFromOptions(const std::string& dbname,
 struct CompactionOptions {
   // Compaction output compression type
   // Default: snappy
+  // If set to `kDisableCompressionOption`, RocksDB will choose compression type
+  // according to the `ColumnFamilyOptions`, taking into account the output
+  // level if `compression_per_level` is specified.
   CompressionType compression;
   // Compaction will create files of size `output_file_size_limit`.
   // Default: MAX, which means that compaction will create a single file
@@ -1268,8 +1275,22 @@ struct IngestExternalFileOptions {
   // with allow_ingest_behind=true since the dawn of time.
   // All files will be ingested at the bottommost level with seqno=0.
   bool ingest_behind = false;
+  // Set to true if you would like to write global_seqno to a given offset in
+  // the external SST file for backward compatibility. Older versions of
+  // RocksDB writes a global_seqno to a given offset within ingested SST files,
+  // and new versions of RocksDB do not. If you ingest an external SST using
+  // new version of RocksDB and would like to be able to downgrade to an
+  // older version of RocksDB, you should set 'write_global_seqno' to true. If
+  // your service is just starting to use the new RocksDB, we recommend that
+  // you set this option to false, which brings two benefits:
+  // 1. No extra random write for global_seqno during ingestion.
+  // 2. Without writing external SST file, it's possible to do checksum.
+  // We have a plan to set this option to false by default in the future.
+  bool write_global_seqno = true;
 };
 
+struct TraceOptions {};
+
 }  // namespace rocksdb
 
 #endif  // STORAGE_ROCKSDB_INCLUDE_OPTIONS_H_

include/rocksdb/slice.h

@@ -25,6 +25,10 @@
 #include <string.h>
 #include <string>
 
+#ifdef __cpp_lib_string_view
+#include <string_view>
+#endif
+
 #include "rocksdb/cleanable.h"
 
 namespace rocksdb {
@@ -41,6 +45,12 @@ class Slice {
   /* implicit */
   Slice(const std::string& s) : data_(s.data()), size_(s.size()) { }
 
+#ifdef __cpp_lib_string_view
+  // Create a slice that refers to the same contents as "sv"
+  /* implicit */
+  Slice(std::string_view sv) : data_(sv.data()), size_(sv.size()) { }
+#endif
+
   // Create a slice that refers to s[0,strlen(s)-1]
   /* implicit */
   Slice(const char* s) : data_(s) {
@@ -86,6 +96,13 @@ class Slice {
   // when hex is true, returns a string of twice the length hex encoded (0-9A-F)
   std::string ToString(bool hex = false) const;
 
+#ifdef __cpp_lib_string_view
+  // Return a string_view that references the same data as this slice.
+  std::string_view ToStringView() const {
+    return std::string_view(data_, size_);
+  }
+#endif
+
   // Decodes the current slice interpreted as an hexadecimal string into result,
   // if successful returns true, if this isn't a valid hex string
   // (e.g not coming from Slice::ToString(true)) DecodeHex returns false.

include/rocksdb/sst_file_manager.h

@@ -75,6 +75,10 @@ class SstFileManager {
   // Update trash/DB size ratio where new files will be deleted immediately
   // thread-safe
   virtual void SetMaxTrashDBRatio(double ratio) = 0;
+
+  // Return the total size of trash files
+  // thread-safe
+  virtual uint64_t GetTotalTrashSize() = 0;
 };
 
 // Create a new SstFileManager that can be shared among multiple RocksDB

include/rocksdb/status.h

@@ -282,8 +282,6 @@ class Status {
   Severity sev_;
   const char* state_;
 
-  static const char* msgs[static_cast<int>(kMaxSubCode)];
-
   explicit Status(Code _code, SubCode _subcode = kNone)
       : code_(_code), subcode_(_subcode), sev_(kNoError), state_(nullptr) {}
 

include/rocksdb/table.h

@@ -100,6 +100,18 @@ struct BlockBasedTableOptions {
 
   IndexType index_type = kBinarySearch;
 
+  // The index type that will be used for the data block.
+  enum DataBlockIndexType : char {
+    kDataBlockBinarySearch = 0,   // traditional block type
+    kDataBlockBinaryAndHash = 1,  // additional hash index
+  };
+
+  DataBlockIndexType data_block_index_type = kDataBlockBinarySearch;
+
+  // #entries/#buckets. It is valid only when data_block_hash_index_type is
+  // kDataBlockBinaryAndHash.
+  double data_block_hash_table_util_ratio = 0.75;
+
   // This option is now deprecated. No matter what value it is set to,
   // it will behave as if hash_index_allow_collision=true.
   bool hash_index_allow_collision = true;

include/rocksdb/table_properties.h

@@ -34,6 +34,7 @@ struct TablePropertiesNames {
   static const std::string kIndexPartitions;
   static const std::string kTopLevelIndexSize;
   static const std::string kIndexKeyIsUserKey;
+  static const std::string kIndexValueIsDeltaEncoded;
   static const std::string kFilterSize;
   static const std::string kRawKeySize;
   static const std::string kRawValueSize;
@@ -139,6 +140,8 @@ struct TableProperties {
   // Whether the index key is user key. Otherwise it includes 8 byte of sequence
   // number added by internal key format.
   uint64_t index_key_is_user_key = 0;
+  // Whether delta encoding is used to encode the index values.
+  uint64_t index_value_is_delta_encoded = 0;
   // the size of filter block.
   uint64_t filter_size = 0;
   // total raw key size

include/rocksdb/trace_reader_writer.h

@@ -0,0 +1,47 @@
+//  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
+
+#include "rocksdb/env.h"
+
+namespace rocksdb {
+
+// Allow custom implementations of TraceWriter and TraceReader.
+// By default, RocksDB provides a way to capture the traces to a file using the
+// factory NewFileTraceWriter(). But users could also choose to export traces to
+// any other system by providing custom implementations of TraceWriter and
+// TraceReader.
+
+// TraceWriter allows exporting RocksDB traces to any system, one operation at
+// a time.
+class TraceWriter {
+ public:
+  TraceWriter() {}
+  virtual ~TraceWriter() {}
+
+  virtual Status Write(const Slice& data) = 0;
+  virtual Status Close() = 0;
+};
+
+// TraceReader allows reading RocksDB traces from any system, one operation at
+// a time. A RocksDB Replayer could depend on this to replay opertions.
+class TraceReader {
+ public:
+  TraceReader() {}
+  virtual ~TraceReader() {}
+
+  virtual Status Read(std::string* data) = 0;
+  virtual Status Close() = 0;
+};
+
+// Factory methods to read/write traces from/to a file.
+Status NewFileTraceWriter(Env* env, const EnvOptions& env_options,
+                          const std::string& trace_filename,
+                          std::unique_ptr<TraceWriter>* trace_writer);
+Status NewFileTraceReader(Env* env, const EnvOptions& env_options,
+                          const std::string& trace_filename,
+                          std::unique_ptr<TraceReader>* trace_reader);
+}  // namespace rocksdb

include/rocksdb/types.h

@@ -23,6 +23,7 @@ enum EntryType {
   kEntrySingleDelete,
   kEntryMerge,
   kEntryRangeDeletion,
+  kEntryBlobIndex,
   kEntryOther,
 };
 

include/rocksdb/utilities/debug.h

@@ -31,9 +31,13 @@ struct KeyVersion {
 };
 
 // Returns listing of all versions of keys in the provided user key range.
-// The range is inclusive-inclusive, i.e., [`begin_key`, `end_key`].
+// The range is inclusive-inclusive, i.e., [`begin_key`, `end_key`], or
+// `max_num_ikeys` has been reached. Since all those keys returned will be
+// copied to memory, if the range covers too many keys, the memory usage
+// may be huge. `max_num_ikeys` can be used to cap the memory usage.
 // The result is inserted into the provided vector, `key_versions`.
 Status GetAllKeyVersions(DB* db, Slice begin_key, Slice end_key,
+                         size_t max_num_ikeys,
                          std::vector<KeyVersion>* key_versions);
 
 }  // namespace rocksdb

include/rocksdb/utilities/ldb_cmd.h

@@ -210,12 +210,20 @@ class LDBCommand {
   bool ParseStringOption(const std::map<std::string, std::string>& options,
                          const std::string& option, std::string* value);
 
+  /**
+   * Returns the value of the specified option as a boolean.
+   * default_val is used if the option is not found in options.
+   * Throws an exception if the value of the option is not
+   * "true" or "false" (case insensitive).
+   */
+  bool ParseBooleanOption(const std::map<std::string, std::string>& options,
+                          const std::string& option, bool default_val);
+
   Options options_;
   std::vector<ColumnFamilyDescriptor> column_families_;
   LDBOptions ldb_options_;
 
  private:
-  friend class WALDumperCommand;
   /**
    * Interpret command line options and flags to determine if the key
    * should be input/output in hex.
@@ -230,15 +238,6 @@ class LDBCommand {
   bool IsValueHex(const std::map<std::string, std::string>& options,
                   const std::vector<std::string>& flags);
 
-  /**
-   * Returns the value of the specified option as a boolean.
-   * default_val is used if the option is not found in options.
-   * Throws an exception if the value of the option is not
-   * "true" or "false" (case insensitive).
-   */
-  bool ParseBooleanOption(const std::map<std::string, std::string>& options,
-                          const std::string& option, bool default_val);
-
   /**
    * Converts val to a boolean.
    * val must be either true or false (case insensitive).

include/rocksdb/utilities/transaction.h

@@ -152,6 +152,12 @@ class Transaction {
   // If there is no previous call to SetSavePoint(), returns Status::NotFound()
   virtual Status RollbackToSavePoint() = 0;
 
+  // Pop the most recent save point.
+  // If there is no previous call to SetSavePoint(), Status::NotFound()
+  // will be returned.
+  // Otherwise returns Status::OK().
+  virtual Status PopSavePoint() = 0;
+
   // This function is similar to DB::Get() except it will also read pending
   // changes in this transaction.  Currently, this function will return
   // Status::MergeInProgress if the most recent write to the queried key in

include/rocksdb/utilities/write_batch_with_index.h

@@ -231,6 +231,7 @@ class WriteBatchWithIndex : public WriteBatchBase {
   Status PopSavePoint() override;
 
   void SetMaxBytes(size_t max_bytes) override;
+  size_t GetDataSize() const;
 
  private:
   friend class PessimisticTransactionDB;

include/rocksdb/version.h

@@ -5,8 +5,8 @@
 #pragma once
 
 #define ROCKSDB_MAJOR 5
-#define ROCKSDB_MINOR 14
-#define ROCKSDB_PATCH 0
+#define ROCKSDB_MINOR 16
+#define ROCKSDB_PATCH 6
 
 // Do not use these. We made the mistake of declaring macros starting with
 // double underscore. Now we have to live with our choice. We'll deprecate these

java/CMakeLists.txt

@@ -3,6 +3,8 @@ cmake_minimum_required(VERSION 3.4)
 set(JNI_NATIVE_SOURCES
         rocksjni/backupablejni.cc
         rocksjni/backupenginejni.cc
+        rocksjni/cassandra_compactionfilterjni.cc
+        rocksjni/cassandra_value_operator.cc
         rocksjni/checkpoint.cc
         rocksjni/clock_cache.cc
         rocksjni/columnfamilyhandle.cc
@@ -11,6 +13,7 @@ set(JNI_NATIVE_SOURCES
         rocksjni/compaction_filter_factory_jnicallback.cc
         rocksjni/compaction_options_fifo.cc
         rocksjni/compaction_options_universal.cc
+        rocksjni/compact_range_options.cc
         rocksjni/comparator.cc
         rocksjni/comparatorjnicallback.cc
         rocksjni/compression_options.cc
@@ -25,14 +28,14 @@ set(JNI_NATIVE_SOURCES
         rocksjni/memtablejni.cc
         rocksjni/merge_operator.cc
         rocksjni/native_comparator_wrapper_test.cc
+        rocksjni/optimistic_transaction_db.cc
+        rocksjni/optimistic_transaction_options.cc
         rocksjni/options.cc
         rocksjni/options_util.cc
         rocksjni/ratelimiterjni.cc
         rocksjni/remove_emptyvalue_compactionfilterjni.cc
-        rocksjni/rocks_callback_object.cc
-        rocksjni/cassandra_compactionfilterjni.cc
-        rocksjni/cassandra_value_operator.cc
         rocksjni/restorejni.cc
+        rocksjni/rocks_callback_object.cc
         rocksjni/rocksdb_exception_test.cc
         rocksjni/rocksjni.cc
         rocksjni/slice.cc
@@ -42,28 +45,33 @@ set(JNI_NATIVE_SOURCES
         rocksjni/statistics.cc
         rocksjni/statisticsjni.cc
         rocksjni/table.cc
+        rocksjni/transaction.cc
+        rocksjni/transaction_db.cc
+        rocksjni/transaction_db_options.cc
         rocksjni/transaction_log.cc
+        rocksjni/transaction_notifier.cc
+        rocksjni/transaction_notifier_jnicallback.cc
+        rocksjni/transaction_options.cc
         rocksjni/ttl.cc
         rocksjni/write_batch.cc
+        rocksjni/writebatchhandlerjnicallback.cc
         rocksjni/write_batch_test.cc
         rocksjni/write_batch_with_index.cc
-        rocksjni/writebatchhandlerjnicallback.cc
 )
 
 set(NATIVE_JAVA_CLASSES
         org.rocksdb.AbstractCompactionFilter
         org.rocksdb.AbstractCompactionFilterFactory
+        org.rocksdb.AbstractComparator
         org.rocksdb.AbstractImmutableNativeReference
         org.rocksdb.AbstractNativeReference
         org.rocksdb.AbstractRocksIterator
         org.rocksdb.AbstractSlice
-        org.rocksdb.AbstractWriteBatch
+        org.rocksdb.AbstractTransactionNotifier
         org.rocksdb.BackupableDBOptions
         org.rocksdb.BackupEngine
-        org.rocksdb.BackupEngineTest
         org.rocksdb.BlockBasedTableConfig
         org.rocksdb.BloomFilter
-        org.rocksdb.Cache
         org.rocksdb.CassandraCompactionFilter
         org.rocksdb.CassandraValueMergeOperator
         org.rocksdb.Checkpoint
@@ -72,6 +80,7 @@ set(NATIVE_JAVA_CLASSES
         org.rocksdb.ColumnFamilyOptions
         org.rocksdb.CompactionOptionsFIFO
         org.rocksdb.CompactionOptionsUniversal
+        org.rocksdb.CompactRangeOptions
         org.rocksdb.Comparator
         org.rocksdb.ComparatorOptions
         org.rocksdb.CompressionOptions
@@ -88,10 +97,10 @@ set(NATIVE_JAVA_CLASSES
         org.rocksdb.Logger
         org.rocksdb.LRUCache
         org.rocksdb.MemTableConfig
-        org.rocksdb.MergeOperator
         org.rocksdb.NativeComparatorWrapper
-        org.rocksdb.NativeComparatorWrapperTest.NativeStringComparatorWrapper
         org.rocksdb.NativeLibraryLoader
+        org.rocksdb.OptimisticTransactionDB
+        org.rocksdb.OptimisticTransactionOptions
         org.rocksdb.Options
         org.rocksdb.OptionsUtil
         org.rocksdb.PlainTableConfig
@@ -101,7 +110,6 @@ set(NATIVE_JAVA_CLASSES
         org.rocksdb.RestoreOptions
         org.rocksdb.RocksCallbackObject
         org.rocksdb.RocksDB
-        org.rocksdb.RocksDBExceptionTest
         org.rocksdb.RocksEnv
         org.rocksdb.RocksIterator
         org.rocksdb.RocksIteratorInterface
@@ -111,24 +119,29 @@ set(NATIVE_JAVA_CLASSES
         org.rocksdb.SkipListMemTableConfig
         org.rocksdb.Slice
         org.rocksdb.Snapshot
-        org.rocksdb.SnapshotTest
         org.rocksdb.SstFileManager
         org.rocksdb.SstFileWriter
         org.rocksdb.Statistics
         org.rocksdb.StringAppendOperator
         org.rocksdb.TableFormatConfig
+        org.rocksdb.Transaction
+        org.rocksdb.TransactionDB
+        org.rocksdb.TransactionDBOptions
         org.rocksdb.TransactionLogIterator
+        org.rocksdb.TransactionOptions
         org.rocksdb.TtlDB
         org.rocksdb.VectorMemTableConfig
         org.rocksdb.WBWIRocksIterator
         org.rocksdb.WriteBatch
         org.rocksdb.WriteBatch.Handler
-        org.rocksdb.WriteBatchTest
-        org.rocksdb.WriteBatchTestInternalHelper
+        org.rocksdb.WriteBatchInterface
         org.rocksdb.WriteBatchWithIndex
         org.rocksdb.WriteOptions
-        org.rocksdb.util.CapturingWriteBatchHandler
-        org.rocksdb.util.WriteBatchGetter
+        org.rocksdb.NativeComparatorWrapperTest
+        org.rocksdb.RocksDBExceptionTest
+        org.rocksdb.SnapshotTest
+        org.rocksdb.WriteBatchTest
+        org.rocksdb.WriteBatchTestInternalHelper
 )
 
 include(FindJava)
@@ -150,13 +163,14 @@ set(JAVA_TESTCLASSPATH ${JAVA_JUNIT_JAR} ${JAVA_HAMCR_JAR} ${JAVA_MOCKITO_JAR} $
 add_jar(
   rocksdbjni_classes
   SOURCES
-  src/main/java/org/rocksdb/AbstractCompactionFilter.java
   src/main/java/org/rocksdb/AbstractCompactionFilterFactory.java
+  src/main/java/org/rocksdb/AbstractCompactionFilter.java
   src/main/java/org/rocksdb/AbstractComparator.java
   src/main/java/org/rocksdb/AbstractImmutableNativeReference.java
   src/main/java/org/rocksdb/AbstractNativeReference.java
   src/main/java/org/rocksdb/AbstractRocksIterator.java
   src/main/java/org/rocksdb/AbstractSlice.java
+  src/main/java/org/rocksdb/AbstractTransactionNotifier.java
   src/main/java/org/rocksdb/AbstractWriteBatch.java
   src/main/java/org/rocksdb/AccessHint.java
   src/main/java/org/rocksdb/AdvancedColumnFamilyOptionsInterface.java
@@ -175,11 +189,12 @@ add_jar(
   src/main/java/org/rocksdb/ClockCache.java
   src/main/java/org/rocksdb/ColumnFamilyDescriptor.java
   src/main/java/org/rocksdb/ColumnFamilyHandle.java
-  src/main/java/org/rocksdb/ColumnFamilyOptions.java
   src/main/java/org/rocksdb/ColumnFamilyOptionsInterface.java
+  src/main/java/org/rocksdb/ColumnFamilyOptions.java
   src/main/java/org/rocksdb/CompactionOptionsFIFO.java
   src/main/java/org/rocksdb/CompactionOptionsUniversal.java
   src/main/java/org/rocksdb/CompactionPriority.java
+  src/main/java/org/rocksdb/CompactRangeOptions.java
   src/main/java/org/rocksdb/CompactionStopStyle.java
   src/main/java/org/rocksdb/CompactionStyle.java
   src/main/java/org/rocksdb/Comparator.java
@@ -187,8 +202,8 @@ add_jar(
   src/main/java/org/rocksdb/ComparatorType.java
   src/main/java/org/rocksdb/CompressionOptions.java
   src/main/java/org/rocksdb/CompressionType.java
-  src/main/java/org/rocksdb/DBOptions.java
   src/main/java/org/rocksdb/DBOptionsInterface.java
+  src/main/java/org/rocksdb/DBOptions.java
   src/main/java/org/rocksdb/DbPath.java
   src/main/java/org/rocksdb/DirectComparator.java
   src/main/java/org/rocksdb/DirectSlice.java
@@ -209,10 +224,12 @@ add_jar(
   src/main/java/org/rocksdb/LRUCache.java
   src/main/java/org/rocksdb/MemTableConfig.java
   src/main/java/org/rocksdb/MergeOperator.java
-  src/main/java/org/rocksdb/MutableColumnFamilyOptions.java
   src/main/java/org/rocksdb/MutableColumnFamilyOptionsInterface.java
+  src/main/java/org/rocksdb/MutableColumnFamilyOptions.java
   src/main/java/org/rocksdb/NativeComparatorWrapper.java
   src/main/java/org/rocksdb/NativeLibraryLoader.java
+  src/main/java/org/rocksdb/OptimisticTransactionDB.java
+  src/main/java/org/rocksdb/OptimisticTransactionOptions.java
   src/main/java/org/rocksdb/Options.java
   src/main/java/org/rocksdb/OptionsUtil.java
   src/main/java/org/rocksdb/PlainTableConfig.java
@@ -223,11 +240,11 @@ add_jar(
   src/main/java/org/rocksdb/RemoveEmptyValueCompactionFilter.java
   src/main/java/org/rocksdb/RestoreOptions.java
   src/main/java/org/rocksdb/RocksCallbackObject.java
-  src/main/java/org/rocksdb/RocksDB.java
   src/main/java/org/rocksdb/RocksDBException.java
+  src/main/java/org/rocksdb/RocksDB.java
   src/main/java/org/rocksdb/RocksEnv.java
-  src/main/java/org/rocksdb/RocksIterator.java
   src/main/java/org/rocksdb/RocksIteratorInterface.java
+  src/main/java/org/rocksdb/RocksIterator.java
   src/main/java/org/rocksdb/RocksMemEnv.java
   src/main/java/org/rocksdb/RocksMutableObject.java
   src/main/java/org/rocksdb/RocksObject.java
@@ -236,22 +253,36 @@ add_jar(
   src/main/java/org/rocksdb/Snapshot.java
   src/main/java/org/rocksdb/SstFileManager.java
   src/main/java/org/rocksdb/SstFileWriter.java
+  src/main/java/org/rocksdb/StatisticsCollectorCallback.java
+  src/main/java/org/rocksdb/StatisticsCollector.java
   src/main/java/org/rocksdb/Statistics.java
+  src/main/java/org/rocksdb/StatsCollectorInput.java
   src/main/java/org/rocksdb/StatsLevel.java
   src/main/java/org/rocksdb/Status.java
   src/main/java/org/rocksdb/StringAppendOperator.java
   src/main/java/org/rocksdb/TableFormatConfig.java
   src/main/java/org/rocksdb/TickerType.java
+  src/main/java/org/rocksdb/TransactionalDB.java
+  src/main/java/org/rocksdb/TransactionalOptions.java
+  src/main/java/org/rocksdb/TransactionDB.java
+  src/main/java/org/rocksdb/TransactionDBOptions.java
+  src/main/java/org/rocksdb/Transaction.java
   src/main/java/org/rocksdb/TransactionLogIterator.java
+  src/main/java/org/rocksdb/TransactionOptions.java
   src/main/java/org/rocksdb/TtlDB.java
-  src/main/java/org/rocksdb/util/Environment.java
+  src/main/java/org/rocksdb/TxnDBWritePolicy.java
   src/main/java/org/rocksdb/VectorMemTableConfig.java
   src/main/java/org/rocksdb/WALRecoveryMode.java
   src/main/java/org/rocksdb/WBWIRocksIterator.java
-  src/main/java/org/rocksdb/WriteBatch.java
   src/main/java/org/rocksdb/WriteBatchInterface.java
+  src/main/java/org/rocksdb/WriteBatch.java
   src/main/java/org/rocksdb/WriteBatchWithIndex.java
   src/main/java/org/rocksdb/WriteOptions.java
+  src/main/java/org/rocksdb/util/BytewiseComparator.java
+  src/main/java/org/rocksdb/util/DirectBytewiseComparator.java
+  src/main/java/org/rocksdb/util/Environment.java
+  src/main/java/org/rocksdb/util/ReverseBytewiseComparator.java
+  src/main/java/org/rocksdb/util/SizeUnit.java
   src/test/java/org/rocksdb/BackupEngineTest.java
   src/test/java/org/rocksdb/IngestExternalFileOptionsTest.java
   src/test/java/org/rocksdb/NativeComparatorWrapperTest.java

java/Makefile

@@ -14,6 +14,7 @@ NATIVE_JAVA_CLASSES = org.rocksdb.AbstractCompactionFilter\
 	org.rocksdb.ColumnFamilyOptions\
 	org.rocksdb.CompactionOptionsFIFO\
 	org.rocksdb.CompactionOptionsUniversal\
+	org.rocksdb.CompactRangeOptions\
 	org.rocksdb.Comparator\
 	org.rocksdb.ComparatorOptions\
 	org.rocksdb.CompressionOptions\

java/crossbuild/build-linux-centos.sh

@@ -26,6 +26,6 @@ export JAVA_HOME=/usr/lib/jvm/java-1.7.0
 # build rocksdb
 cd /rocksdb
 scl enable devtoolset-2 'make jclean clean'
-scl enable devtoolset-2 'PORTABLE=1 make rocksdbjavastatic'
+scl enable devtoolset-2 'PORTABLE=1 make -j8 rocksdbjavastatic'
 cp /rocksdb/java/target/librocksdbjni-* /rocksdb-build
 cp /rocksdb/java/target/rocksdbjni-* /rocksdb-build

java/crossbuild/docker-build-linux-centos.sh

@@ -9,10 +9,10 @@ cd /rocksdb-local
 # Use scl devtoolset if available (i.e. CentOS <7)
 if hash scl 2>/dev/null; then
 	scl enable devtoolset-2 'make jclean clean'
-	scl enable devtoolset-2 'PORTABLE=1 make rocksdbjavastatic'
+	scl enable devtoolset-2 'PORTABLE=1 make -j8 rocksdbjavastatic'
 else
 	make jclean clean
-        PORTABLE=1 make rocksdbjavastatic
+        PORTABLE=1 make -j8 rocksdbjavastatic
 fi
 
 cp java/target/librocksdbjni-linux*.so java/target/rocksdbjni-*-linux*.jar /rocksdb-host/java/target

java/rocksjni/compact_range_options.cc

@@ -0,0 +1,196 @@
+// 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).
+//
+// This file implements the "bridge" between Java and C++ for
+// rocksdb::CompactRangeOptions.
+
+#include <jni.h>
+
+#include "include/org_rocksdb_CompactRangeOptions.h"
+#include "rocksdb/options.h"
+#include "rocksjni/portal.h"
+
+/*
+ * Class:     org_rocksdb_CompactRangeOptions
+ * Method:    newCompactRangeOptions
+ * Signature: ()J
+ */
+jlong Java_org_rocksdb_CompactRangeOptions_newCompactRangeOptions(
+    JNIEnv* /*env*/, jclass /*jclazz*/) {
+  auto* options = new rocksdb::CompactRangeOptions();
+  return reinterpret_cast<jlong>(options);
+}
+
+
+/*
+ * Class:     org_rocksdb_CompactRangeOptions
+ * Method:    exclusiveManualCompaction
+ * Signature: (J)Z
+ */
+jboolean Java_org_rocksdb_CompactRangeOptions_exclusiveManualCompaction(
+    JNIEnv* /*env*/, jobject /*jobj*/, jlong jhandle) {
+  auto* options = reinterpret_cast<rocksdb::CompactRangeOptions*>(jhandle);
+  return static_cast<jboolean>(options->exclusive_manual_compaction);
+}
+
+/*
+ * Class:     org_rocksdb_CompactRangeOptions
+ * Method:    setExclusiveManualCompaction
+ * Signature: (JZ)V
+ */
+void Java_org_rocksdb_CompactRangeOptions_setExclusiveManualCompaction(
+    JNIEnv* /*env*/, jobject /*jobj*/, jlong jhandle, jboolean exclusive_manual_compaction) {
+  auto* options =
+      reinterpret_cast<rocksdb::CompactRangeOptions*>(jhandle);
+  options->exclusive_manual_compaction = static_cast<bool>(exclusive_manual_compaction);
+}
+
+
+/*
+ * Class:     org_rocksdb_CompactRangeOptions
+ * Method:    bottommostLevelCompaction
+ * Signature: (J)I
+ */
+jint Java_org_rocksdb_CompactRangeOptions_bottommostLevelCompaction(
+    JNIEnv* /*env*/, jobject /*jobj*/, jlong jhandle) {
+  auto* options = reinterpret_cast<rocksdb::CompactRangeOptions*>(jhandle);
+  return rocksdb::BottommostLevelCompactionJni::toJavaBottommostLevelCompaction(
+    options->bottommost_level_compaction);
+}
+
+/*
+ * Class:     org_rocksdb_CompactRangeOptions
+ * Method:    setBottommostLevelCompaction
+ * Signature: (JI)V
+ */
+void Java_org_rocksdb_CompactRangeOptions_setBottommostLevelCompaction(
+    JNIEnv* /*env*/, jobject /*jobj*/, jlong jhandle,
+    jint bottommost_level_compaction) {
+  auto* options = reinterpret_cast<rocksdb::CompactRangeOptions*>(jhandle);
+  options->bottommost_level_compaction =
+    rocksdb::BottommostLevelCompactionJni::toCppBottommostLevelCompaction(bottommost_level_compaction);
+}
+
+/*
+ * Class:     org_rocksdb_CompactRangeOptions
+ * Method:    changeLevel
+ * Signature: (J)Z
+ */
+jboolean Java_org_rocksdb_CompactRangeOptions_changeLevel
+  (JNIEnv* /*env*/, jobject /*jobj*/, jlong jhandle) {
+  auto* options = reinterpret_cast<rocksdb::CompactRangeOptions*>(jhandle);
+  return static_cast<jboolean>(options->change_level);
+}
+
+/*
+ * Class:     org_rocksdb_CompactRangeOptions
+ * Method:    setChangeLevel
+ * Signature: (JZ)V
+ */
+void Java_org_rocksdb_CompactRangeOptions_setChangeLevel
+  (JNIEnv* /*env*/, jobject /*jobj*/, jlong jhandle, jboolean change_level) {
+  auto* options = reinterpret_cast<rocksdb::CompactRangeOptions*>(jhandle);
+  options->change_level = static_cast<bool>(change_level);
+}
+
+/*
+ * Class:     org_rocksdb_CompactRangeOptions
+ * Method:    targetLevel
+ * Signature: (J)I
+ */
+jint Java_org_rocksdb_CompactRangeOptions_targetLevel
+  (JNIEnv* /*env*/, jobject /*jobj*/, jlong jhandle) {
+  auto* options = reinterpret_cast<rocksdb::CompactRangeOptions*>(jhandle);
+  return static_cast<jint>(options->target_level);
+}
+
+/*
+ * Class:     org_rocksdb_CompactRangeOptions
+ * Method:    setTargetLevel
+ * Signature: (JI)V
+ */
+void Java_org_rocksdb_CompactRangeOptions_setTargetLevel
+  (JNIEnv* /*env*/, jobject /*jobj*/, jlong jhandle, jint target_level) {
+  auto* options = reinterpret_cast<rocksdb::CompactRangeOptions*>(jhandle);
+  options->target_level = static_cast<int>(target_level);
+}
+
+/*
+ * Class:     org_rocksdb_CompactRangeOptions
+ * Method:    targetPathId
+ * Signature: (J)I
+ */
+jint Java_org_rocksdb_CompactRangeOptions_targetPathId
+  (JNIEnv* /*env*/, jobject /*jobj*/, jlong jhandle) {
+  auto* options = reinterpret_cast<rocksdb::CompactRangeOptions*>(jhandle);
+  return static_cast<jint>(options->target_path_id);
+}
+
+/*
+ * Class:     org_rocksdb_CompactRangeOptions
+ * Method:    setTargetPathId
+ * Signature: (JI)V
+ */
+void Java_org_rocksdb_CompactRangeOptions_setTargetPathId
+  (JNIEnv* /*env*/, jobject /*jobj*/, jlong jhandle, jint target_path_id) {
+  auto* options = reinterpret_cast<rocksdb::CompactRangeOptions*>(jhandle);
+  options->target_path_id = static_cast<uint32_t>(target_path_id);
+}
+
+/*
+ * Class:     org_rocksdb_CompactRangeOptions
+ * Method:    allowWriteStall
+ * Signature: (J)Z
+ */
+jboolean Java_org_rocksdb_CompactRangeOptions_allowWriteStall
+  (JNIEnv* /*env*/, jobject /*jobj*/, jlong jhandle) {
+  auto* options = reinterpret_cast<rocksdb::CompactRangeOptions*>(jhandle);
+  return static_cast<jboolean>(options->allow_write_stall);
+}
+
+/*
+ * Class:     org_rocksdb_CompactRangeOptions
+ * Method:    setAllowWriteStall
+ * Signature: (JZ)V
+ */
+void Java_org_rocksdb_CompactRangeOptions_setAllowWriteStall
+  (JNIEnv* /*env*/, jobject /*jobj*/, jlong jhandle, jboolean allow_write_stall) {
+  auto* options = reinterpret_cast<rocksdb::CompactRangeOptions*>(jhandle);
+  options->allow_write_stall = static_cast<bool>(allow_write_stall);
+}
+
+
+/*
+ * Class:     org_rocksdb_CompactRangeOptions
+ * Method:    maxSubcompactions
+ * Signature: (J)I
+ */
+jint Java_org_rocksdb_CompactRangeOptions_maxSubcompactions
+  (JNIEnv* /*env*/, jobject /*jobj*/, jlong jhandle) {
+  auto* options = reinterpret_cast<rocksdb::CompactRangeOptions*>(jhandle);
+  return static_cast<jint>(options->max_subcompactions);
+}
+
+/*
+ * Class:     org_rocksdb_CompactRangeOptions
+ * Method:    setMaxSubcompactions
+ * Signature: (JI)V
+ */
+void Java_org_rocksdb_CompactRangeOptions_setMaxSubcompactions
+  (JNIEnv* /*env*/, jobject /*jobj*/, jlong jhandle, jint max_subcompactions) {
+  auto* options = reinterpret_cast<rocksdb::CompactRangeOptions*>(jhandle);
+  options->max_subcompactions = static_cast<uint32_t>(max_subcompactions);
+}
+
+/*
+ * Class:     org_rocksdb_CompactRangeOptions
+ * Method:    disposeInternal
+ * Signature: (J)V
+ */
+void Java_org_rocksdb_CompactRangeOptions_disposeInternal(
+    JNIEnv* /*env*/, jobject /*jobj*/, jlong jhandle) {
+  auto* options = reinterpret_cast<rocksdb::CompactRangeOptions*>(jhandle);
+  delete options;
+}