Fix Windows strcmp for Unicode (#8190)

Summary:
The code for strcmp that was present does work when compiled for Windows unicode file paths.

Needs backporting to:
* 6.17.fb
* 6.18.fb
* 6.19.fb

Pull Request resolved: https://github.com/facebook/rocksdb/pull/8190

Reviewed By: akankshamahajan15

Differential Revision: D27765588

Pulled By: jay-zhuang

fbshipit-source-id: 89f8a5ac61fd7edc758340dfd335b0a5f96dae6e

.circleci/config.yml

@@ -62,6 +62,14 @@ commands:
           path: /tmp/test-results
       - store_artifacts: # store LOG for debugging if there's any
           path: LOG
+      - run: # on fail, compress Test Logs for diagnosing the issue
+         name: Compress Test Logs
+         command: tar -cvzf t.tar.gz t
+         when: on_fail
+      - store_artifacts: # on fail, store Test Logs for diagnosing the issue
+          path: t.tar.gz
+          destination: test_logs
+          when: on_fail
 
   install-clang-10:
     steps:
@@ -79,7 +87,7 @@ commands:
           name: Install gflags
           command: |
             sudo apt-get update -y && sudo apt-get install -y libgflags-dev
-  
+
   install-gflags-on-macos:
     steps:
       - run:
@@ -93,7 +101,14 @@ commands:
           name: Install gtest-parallel
           command: |
             git clone --single-branch --branch master --depth 1 https://github.com/google/gtest-parallel.git ~/gtest-parallel
-            echo "export PATH=$HOME/gtest-parallel:$PATH" >> $BASH_ENV
+            echo 'export PATH=$HOME/gtest-parallel:$PATH' >> $BASH_ENV
+
+  install-compression-libs:
+    steps:
+      - run:
+          name: Install compression libs
+          command: |
+            sudo apt-get update -y && sudo apt-get install -y libsnappy-dev zlib1g-dev libbz2-dev liblz4-dev libzstd-dev
 
 executors:
   windows-2xlarge:
@@ -169,13 +184,13 @@ jobs:
   build-linux-release:
     machine:
       image: ubuntu-1604:202007-01
-    resource_class: 2xlarge
+    resource_class: large
     steps:
       - checkout # check out the code in the project directory
-      - run: make V=1 -j32 release | .circleci/cat_ignore_eagain
+      - run: make V=1 -j8 release | .circleci/cat_ignore_eagain
       - run: if ./db_stress --version; then false; else true; fi # ensure without gflags
       - install-gflags
-      - run: make V=1 -j32 release | .circleci/cat_ignore_eagain
+      - run: make V=1 -j8 release | .circleci/cat_ignore_eagain
       - run: ./db_stress --version # ensure with gflags
       - post-steps
 
@@ -186,11 +201,11 @@ jobs:
     steps:
       - checkout # check out the code in the project directory
       - run: make clean
-      - run: USE_RTTI=1 DEBUG_LEVEL=0 make V=1 -j16 static_lib tools db_bench | .circleci/cat_ignore_eagain
+      - run: USE_RTTI=1 DEBUG_LEVEL=0 make V=1 -j8 static_lib tools db_bench | .circleci/cat_ignore_eagain
       - run: if ./db_stress --version; then false; else true; fi # ensure without gflags
       - run: sudo apt-get update -y && sudo apt-get install -y libgflags-dev
       - run: make clean
-      - run: USE_RTTI=1 DEBUG_LEVEL=0 make V=1 -j16 static_lib tools db_bench | .circleci/cat_ignore_eagain
+      - run: USE_RTTI=1 DEBUG_LEVEL=0 make V=1 -j8 static_lib tools db_bench | .circleci/cat_ignore_eagain
       - run: ./db_stress --version # ensure with gflags
 
   build-linux-lite:
@@ -209,21 +224,21 @@ jobs:
     resource_class: large
     steps:
       - checkout # check out the code in the project directory
-      - run: LITE=1 make V=1 -j32 release | .circleci/cat_ignore_eagain
+      - run: LITE=1 make V=1 -j8 release | .circleci/cat_ignore_eagain
       - run: if ./db_stress --version; then false; else true; fi # ensure without gflags
       - install-gflags
-      - run: LITE=1 make V=1 -j32 release | .circleci/cat_ignore_eagain
+      - run: LITE=1 make V=1 -j8 release | .circleci/cat_ignore_eagain
       - run: ./db_stress --version # ensure with gflags
       - post-steps
 
-  build-linux-clang-no-test:
+  build-linux-clang-no_test_run:
     machine:
       image: ubuntu-1604:202007-01
-    resource_class: 2xlarge
+    resource_class: xlarge
     steps:
       - checkout # check out the code in the project directory
       - run: sudo apt-get update -y && sudo apt-get install -y clang libgflags-dev
-      - run: CC=clang CXX=clang++ USE_CLANG=1 PORTABLE=1 make V=1 -j32 all | .circleci/cat_ignore_eagain
+      - run: CC=clang CXX=clang++ USE_CLANG=1 PORTABLE=1 make V=1 -j16 all | .circleci/cat_ignore_eagain
       - post-steps
 
   build-linux-clang10-asan:
@@ -271,33 +286,63 @@ jobs:
       - run: CC=clang-10 CXX=clang++-10 ROCKSDB_DISABLE_ALIGNED_NEW=1 CLANG_ANALYZER="/usr/bin/clang++-10" CLANG_SCAN_BUILD=scan-build-10 USE_CLANG=1 make V=1 -j32 analyze | .circleci/cat_ignore_eagain # aligned new doesn't work for reason we haven't figured out. For unknown, reason passing "clang++-10" as CLANG_ANALYZER doesn't work, and we need a full path.
       - post-steps
 
-  build-linux-cmake:
+  build-linux-cmake-no_test_run:
     machine:
       image: ubuntu-1604:202007-01
-    resource_class: 2xlarge
+    resource_class: large
     steps:
       - checkout # check out the code in the project directory
-      - run: (mkdir build && cd build && cmake -DWITH_GFLAGS=0 .. && make V=1 -j32) | .circleci/cat_ignore_eagain
+      - run: (mkdir build && cd build && cmake -DWITH_GFLAGS=0 .. && make V=1 -j8) | .circleci/cat_ignore_eagain
       - post-steps
 
   build-linux-unity:
     docker: # executor type
       - image: gcc:latest
-    resource_class: xlarge
+    resource_class: large
     steps:
       - checkout # check out the code in the project directory
       - run: apt-get update -y && apt-get install -y libgflags-dev
-      - run: TEST_TMPDIR=/dev/shm && make V=1 -j16 unity_test | .circleci/cat_ignore_eagain
+      - run: TEST_TMPDIR=/dev/shm && make V=1 -j8 unity_test | .circleci/cat_ignore_eagain
       - post-steps
 
-  build-linux-gcc-4-8:
+  build-linux-gcc-4_8-no_test_run:
     machine:
       image: ubuntu-1604:202007-01
     resource_class: large
     steps:
       - pre-steps
       - run: sudo apt-get update -y && sudo apt-get install gcc-4.8 g++-4.8 libgflags-dev
-      - run: CC=gcc-4.8 CXX=g++-4.8 V=1 SKIP_LINK=1 make -j4 all | .circleci/cat_ignore_eagain # Linking broken because libgflags compiled with newer ABI
+      - run: CC=gcc-4.8 CXX=g++-4.8 V=1 SKIP_LINK=1 make -j8 all | .circleci/cat_ignore_eagain # Linking broken because libgflags compiled with newer ABI
+      - post-steps
+
+  build-linux-gcc-8-no_test_run:
+    machine:
+      image: ubuntu-2004:202010-01
+    resource_class: large
+    steps:
+      - pre-steps
+      - run: sudo apt-get update -y && sudo apt-get install gcc-8 g++-8 libgflags-dev
+      - run: CC=gcc-8 CXX=g++-8 V=1 SKIP_LINK=1 make -j8 all | .circleci/cat_ignore_eagain # Linking broken because libgflags compiled with newer ABI
+      - post-steps
+
+  build-linux-gcc-9-no_test_run:
+    machine:
+      image: ubuntu-2004:202010-01
+    resource_class: large
+    steps:
+      - pre-steps
+      - run: sudo apt-get update -y && sudo apt-get install gcc-9 g++-9 libgflags-dev
+      - run: CC=gcc-9 CXX=g++-9 V=1 SKIP_LINK=1 make -j8 all | .circleci/cat_ignore_eagain # Linking broken because libgflags compiled with newer ABI
+      - post-steps
+
+  build-linux-gcc-10-cxx20-no_test_run:
+    machine:
+      image: ubuntu-2004:202010-01
+    resource_class: xlarge
+    steps:
+      - pre-steps
+      - run: sudo apt-get update -y && sudo apt-get install gcc-10 g++-10 libgflags-dev
+      - run: CC=gcc-10 CXX=g++-10 V=1 SKIP_LINK=1 ROCKSDB_CXX_STANDARD=c++20 make -j16 all | .circleci/cat_ignore_eagain # Linking broken because libgflags compiled with newer ABI
       - post-steps
 
   build-windows:
@@ -326,9 +371,13 @@ jobs:
       - run:
           name: "Setup VS"
           command: |
-            if [[ "${VS_YEAR}" == "2017" ]]; then
+            if [[ "${VS_YEAR}" == "2019" ]]; then
+              echo "VS2019 already present."
+            elif [[ "${VS_YEAR}" == "2017" ]]; then
+              echo "Installing VS2017..."
               powershell .circleci/vs2017_install.ps1
             elif [[ "${VS_YEAR}" == "2015" ]]; then
+              echo "Installing VS2015..."
               powershell .circleci/vs2015_install.ps1
             fi
       - run:
@@ -354,6 +403,7 @@ jobs:
             cd build
             ${CMAKE_BIN} -G "${CMAKE_GENERATOR}" -DCMAKE_BUILD_TYPE=Debug -DOPTDBG=1 -DPORTABLE=1 -DSNAPPY=1 -DJNI=1 << parameters.extra_cmake_opt >> ..
             cd ..
+            echo "Building with VS version: ${CMAKE_GENERATOR}"
             msbuild.exe build/rocksdb.sln -maxCpuCount -property:Configuration=Debug -property:Platform=x64
       - run:
           name: "Test RocksDB"
@@ -459,7 +509,7 @@ jobs:
   build-examples:
     machine:
       image: ubuntu-1604:202007-01
-    resource_class: medium
+    resource_class: large
     steps:
       - pre-steps
       - install-gflags
@@ -469,6 +519,29 @@ jobs:
             OPT=-DTRAVIS V=1 make -j4 static_lib && cd examples && make -j4 | ../.circleci/cat_ignore_eagain
       - post-steps
 
+  build-cmake-mingw:
+    machine:
+      image: ubuntu-1604:202007-01
+    resource_class: 2xlarge
+    steps:
+      - pre-steps
+      - install-gflags
+      - run: sudo apt-get update -y && sudo apt-get install -y mingw-w64
+      - run: sudo update-alternatives --set x86_64-w64-mingw32-g++ /usr/bin/x86_64-w64-mingw32-g++-posix
+      - run:
+          name: "Build cmake-mingw"
+          command: |
+            sudo apt-get install snapd && sudo snap install cmake --beta --classic
+            export PATH=/snap/bin:$PATH
+            sudo apt-get install -y openjdk-8-jdk
+            export JAVA_HOME=/usr/lib/jvm/java-8-openjdk-amd64
+            export PATH=$JAVA_HOME/bin:$PATH
+            echo "JAVA_HOME=${JAVA_HOME}"
+            which java && java -version
+            which javac && javac -version
+            mkdir build && cd build && cmake -DJNI=1 -DWITH_GFLAGS=OFF .. -DCMAKE_C_COMPILER=x86_64-w64-mingw32-gcc -DCMAKE_CXX_COMPILER=x86_64-w64-mingw32-g++ -DCMAKE_SYSTEM_NAME=Windows && make -j4 rocksdb rocksdbjni
+      - post-steps
+
   build-linux-non-shm:
     machine:
       image: ubuntu-1604:202007-01
@@ -496,10 +569,28 @@ jobs:
             sed -i 's/[[:space:]]*$//; s/ / \.\//g; s/.*/.\/&/' /tmp/test_list
             cat /tmp/test_list
             export TEST_TMPDIR=/tmp/rocksdb_test_tmp
-            /usr/bin/python ../gtest-parallel/gtest-parallel $(</tmp/test_list) --output_dir=/tmp | cat  # pipe to cat to continuously output status on circleci UI. Otherwise, no status will be printed while the job is running.
+            gtest-parallel $(</tmp/test_list) --output_dir=/tmp | cat  # pipe to cat to continuously output status on circleci UI. Otherwise, no status will be printed while the job is running.
+      - post-steps
+
+  build-format-compatible:
+    machine:
+      image: ubuntu-1604:202007-01
+    resource_class: 2xlarge
+    steps:
+      - pre-steps
+      - install-gflags
+      - install-compression-libs
+      - run:
+          name: "test"
+          command: |
+            export TEST_TMPDIR=/dev/shm/rocksdb
+            rm -rf /dev/shm/rocksdb
+            mkdir /dev/shm/rocksdb
+            tools/check_format_compatible.sh
       - post-steps
 
 workflows:
+  version: 2
   build-linux:
     jobs:
       - build-linux
@@ -524,9 +615,6 @@ workflows:
   build-linux-lite-release:
     jobs:
       - build-linux-lite-release
-  build-linux-clang-no-test:
-    jobs:
-      - build-linux-clang-no-test
   build-linux-clang10-asan:
     jobs:
       - build-linux-clang10-asan
@@ -539,29 +627,30 @@ workflows:
   build-linux-clang10-clang-analyze:
     jobs:
       - build-linux-clang10-clang-analyze
-  build-linux-cmake:
-    jobs:
-      - build-linux-cmake
   build-linux-unity:
     jobs:
       - build-linux-unity
-  build-windows:
+  build-windows-vs2019:
     jobs:
-      - build-windows
+      - build-windows:
+          name: "build-windows-vs2019"
+  build-windows-vs2019-cxx20:
+    jobs:
+      - build-windows:
+          name: "build-windows-vs2019-cxx20"
+          extra_cmake_opt: -DCMAKE_CXX_STANDARD=20
   build-windows-vs2017:
     jobs:
       - build-windows:
+          name: "build-windows-vs2017"
           vs_year: "2017"
           cmake_generator: "Visual Studio 15 Win64"
   build-windows-vs2015:
     jobs:
       - build-windows:
+          name: "build-windows-vs2015"
           vs_year: "2015"
           cmake_generator: "Visual Studio 14 Win64"
-  build-windows-cxx20:
-    jobs:
-      - build-windows:
-          extra_cmake_opt: -DCMAKE_CXX_STANDARD=20
   build-java:
     jobs:
       - build-linux-java
@@ -585,9 +674,30 @@ workflows:
       - build-linux-non-shm:
           start_test: "statistics_test" # make sure unique in src.mk
           end_test: ""
-  build-linux-gcc-4-8:
+  build-linux-compilers-no_test_run:
     jobs:
-      - build-linux-gcc-4-8
+      - build-linux-clang-no_test_run
+      - build-linux-cmake-no_test_run
+      - build-linux-gcc-4_8-no_test_run
+      - build-linux-gcc-8-no_test_run
+      - build-linux-gcc-9-no_test_run
+      - build-linux-gcc-10-cxx20-no_test_run
   build-macos:
     jobs:
       - build-macos
+  build-macos-cmake:
+    jobs:
+      - build-macos-cmake
+  build-cmake-mingw:
+    jobs:
+      - build-cmake-mingw
+  nightly:
+    triggers:
+      - schedule:
+          cron: "0 0 * * *"
+          filters:
+            branches:
+              only:
+                - master
+    jobs:
+      - build-format-compatible

.circleci/vs2015_install.ps1

@@ -21,3 +21,4 @@ if (($exitCode -ne 0) -and ($exitCode -ne 3010)) {
     Copy-Item -Path "C:\Users\circleci\AppData\Local\Temp\vslogs.zip" -Destination "C:\w\build-results\"
     exit 1
 }
+echo "VS 2015 installed."

.circleci/vs2017_install.ps1

@@ -32,3 +32,4 @@ if (($exitCode -ne 0) -and ($exitCode -ne 3010)) {
     Copy-Item -Path "C:\Users\circleci\AppData\Local\Temp\vslogs.zip" -Destination "C:\w\build-results\"
     exit 1
 }
+echo "VS 2017 installed."

.github/workflows/sanity_check.yml

@@ -32,7 +32,7 @@ jobs:
     - name: Download clang-format-diff.py
       uses: wei/wget@v1
       with:
-        args: https://raw.githubusercontent.com/llvm/llvm-project/master/clang/tools/clang-format/clang-format-diff.py
+        args: https://raw.githubusercontent.com/llvm/llvm-project/main/clang/tools/clang-format/clang-format-diff.py
 
     - name: Check format
       run: VERBOSE_CHECK=1 make check-format

.gitignore

@@ -90,3 +90,8 @@ buckifier/__pycache__
 compile_commands.json
 clang-format-diff.py
 .py3/
+
+fuzz/proto/gen/
+fuzz/crash-*
+
+cmake-build-*

.travis.yml

@@ -2,20 +2,18 @@ dist: xenial
 language: cpp
 os:
   - linux
-  - osx
 arch:
-  - amd64
   - arm64
   - ppc64le
 compiler:
   - clang
   - gcc
-osx_image: xcode9.4
 cache:
   - ccache
 
 addons:
   apt:
+    update: true
     sources:
       - ubuntu-toolchain-r-test
     packages:
@@ -26,15 +24,6 @@ addons:
       - liblzma-dev  # xv
       - libzstd-dev
       - zlib1g-dev
-  homebrew:
-    update: true
-    packages:
-      - ccache
-      - gflags
-      - lz4
-      - snappy
-      - xz
-      - zstd
 
 env:
   - TEST_GROUP=platform_dependent # 16-18 minutes
@@ -58,20 +47,6 @@ env:
 
 matrix:
   exclude:
-  - os: osx
-    env: JOB_NAME=cmake-gcc8
-  - os: osx
-    env: JOB_NAME=cmake-gcc9
-  - os: osx
-    env: JOB_NAME=cmake-gcc9-c++20
-  - os: osx
-    env: JOB_NAME=cmake-mingw
-  - os: osx
-    env: JOB_NAME=make-gcc4.8
-  - os: osx
-    arch: ppc64le
-  - os: osx
-    compiler: gcc
   - os : linux
     arch: arm64
     env: JOB_NAME=cmake-mingw
@@ -86,29 +61,6 @@ matrix:
     env: JOB_NAME=make-gcc4.8
   - os: linux
     compiler: clang
-  # Exclude all but most unique cmake variants for pull requests, but build all in branches
-  - if: type = pull_request AND commit_message !~ /FULL_CI/
-    os : linux
-    arch: amd64
-    env: JOB_NAME=cmake
-  - if: type = pull_request AND commit_message !~ /FULL_CI/
-    os : linux
-    arch: amd64
-    env: JOB_NAME=cmake-gcc8
-  - if: type = pull_request AND commit_message !~ /FULL_CI/
-    os : linux
-    arch: amd64
-    env: JOB_NAME=cmake-gcc9
-  # Exclude most osx, arm64 and ppc64le tests for pull requests, but build in branches
-  # Temporarily disable ppc64le cmake test while snapd is broken
-  - if: type = pull_request AND commit_message !~ /FULL_CI/
-    os: linux
-    arch: ppc64le
-    env: JOB_NAME=cmake
-  # NB: the cmake build is a partial java test
-  - if: type = pull_request AND commit_message !~ /FULL_CI/
-    os: osx
-    env: TEST_GROUP=1
   - if: type = pull_request AND commit_message !~ /FULL_CI/
     os : linux
     arch: arm64
@@ -117,9 +69,6 @@ matrix:
     os: linux
     arch: ppc64le
     env: TEST_GROUP=1
-  - if: type = pull_request AND commit_message !~ /FULL_CI/
-    os: osx
-    env: TEST_GROUP=2
   - if: type = pull_request AND commit_message !~ /FULL_CI/
     os : linux
     arch: arm64
@@ -128,9 +77,6 @@ matrix:
     os: linux
     arch: ppc64le
     env: TEST_GROUP=2
-  - if: type = pull_request AND commit_message !~ /FULL_CI/
-    os: osx
-    env: TEST_GROUP=3
   - if: type = pull_request AND commit_message !~ /FULL_CI/
     os : linux
     arch: arm64
@@ -139,9 +85,6 @@ matrix:
     os: linux
     arch: ppc64le
     env: TEST_GROUP=3
-  - if: type = pull_request AND commit_message !~ /FULL_CI/
-    os: osx
-    env: TEST_GROUP=4
   - if: type = pull_request AND commit_message !~ /FULL_CI/
     os : linux
     arch: arm64
@@ -150,9 +93,6 @@ matrix:
     os: linux
     arch: ppc64le
     env: TEST_GROUP=4
-  - if: type = pull_request AND commit_message !~ /FULL_CI/ AND commit_message !~ /java/
-    os : osx
-    env: JOB_NAME=java_test
   - if: type = pull_request AND commit_message !~ /FULL_CI/ AND commit_message !~ /java/
     os : linux
     arch: arm64
@@ -161,9 +101,6 @@ matrix:
     os: linux
     arch: ppc64le
     env: JOB_NAME=java_test
-  - if: type = pull_request AND commit_message !~ /FULL_CI/
-    os : osx
-    env: JOB_NAME=lite_build
   - if: type = pull_request AND commit_message !~ /FULL_CI/
     os : linux
     arch: arm64
@@ -172,9 +109,6 @@ matrix:
     os: linux
     arch: ppc64le
     env: JOB_NAME=lite_build
-  - if: type = pull_request AND commit_message !~ /FULL_CI/
-    os : osx
-    env: JOB_NAME=examples
   - if: type = pull_request AND commit_message !~ /FULL_CI/
     os : linux
     arch: arm64
@@ -207,9 +141,6 @@ matrix:
     os: linux
     arch: ppc64le
     env: JOB_NAME=cmake-gcc9-c++20
-  - if: type = pull_request AND commit_message !~ /FULL_CI/
-    os : osx
-    env: JOB_NAME=status_checked
   - if: type = pull_request AND commit_message !~ /FULL_CI/
     os : linux
     arch: arm64
@@ -220,9 +151,6 @@ matrix:
     env: JOB_NAME=status_checked
 
 install:
-  - if [ "${TRAVIS_OS_NAME}" == osx ]; then
-      PATH=$PATH:/usr/local/opt/ccache/libexec;
-    fi
   - if [ "${JOB_NAME}" == cmake-gcc8 ]; then
       sudo apt-get install -y g++-8 || exit $?;
       CC=gcc-8 && CXX=g++-8;
@@ -238,22 +166,21 @@ install:
       sudo apt-get install -y g++-4.8 || exit $?;
       CC=gcc-4.8 && CXX=g++-4.8;
     fi
-  - if [[ "${JOB_NAME}" == cmake* ]] && [ "${TRAVIS_OS_NAME}" == linux ]; then
-      sudo apt-get install snapd && sudo snap install cmake --beta --classic || exit $?;
-      export PATH=/snap/bin:$PATH;
+  - |
+    if [[ "${JOB_NAME}" == cmake* ]]; then
+      sudo apt-get remove -y cmake cmake-data
+      export CMAKE_DEB="cmake-3.14.5-Linux-$(uname -m).deb"
+      export CMAKE_DEB_URL="https://rocksdb-deps.s3-us-west-2.amazonaws.com/cmake/${CMAKE_DEB}"
+      curl --silent --fail --show-error --location --output "${CMAKE_DEB}" "${CMAKE_DEB_URL}" || exit $?
+      sudo dpkg -i "${CMAKE_DEB}" || exit $?
+      which cmake && cmake --version
     fi
   - |
     if [[ "${JOB_NAME}" == java_test || "${JOB_NAME}" == cmake* ]]; then
       # Ensure JDK 8
-      if [ "${TRAVIS_OS_NAME}" == osx ]; then
-        brew tap AdoptOpenJDK/openjdk || exit $?
-        brew cask install adoptopenjdk8 || exit $?
-        export JAVA_HOME=$(/usr/libexec/java_home)
-      else
-        sudo apt-get install -y openjdk-8-jdk || exit $?
-        export PATH=/usr/lib/jvm/java-8-openjdk-$(dpkg --print-architecture)/bin:$PATH
-        export JAVA_HOME=/usr/lib/jvm/java-8-openjdk-$(dpkg --print-architecture)
-      fi
+      sudo apt-get install -y openjdk-8-jdk || exit $?
+      export PATH=/usr/lib/jvm/java-8-openjdk-$(dpkg --print-architecture)/bin:$PATH
+      export JAVA_HOME=/usr/lib/jvm/java-8-openjdk-$(dpkg --print-architecture)
       echo "JAVA_HOME=${JAVA_HOME}"
       which java && java -version
       which javac && javac -version

CMakeLists.txt

@@ -170,23 +170,25 @@ else()
   endif()
 endif()
 
-string(TIMESTAMP TS "%Y/%m/%d %H:%M:%S" UTC)
-set(GIT_DATE_TIME "${TS}" CACHE STRING "the time we first built rocksdb")
+string(TIMESTAMP TS "%Y-%m-%d %H:%M:%S" UTC)
+set(BUILD_DATE "${TS}" CACHE STRING "the time we first built rocksdb")
 
 find_package(Git)
 
 if(GIT_FOUND AND EXISTS "${CMAKE_CURRENT_SOURCE_DIR}/.git")
-  if(WIN32)
-    execute_process(COMMAND $ENV{COMSPEC} /C ${GIT_EXECUTABLE} -C ${CMAKE_CURRENT_SOURCE_DIR} rev-parse HEAD OUTPUT_VARIABLE GIT_SHA)
-  else()
-    execute_process(COMMAND ${GIT_EXECUTABLE} -C ${CMAKE_CURRENT_SOURCE_DIR} rev-parse HEAD OUTPUT_VARIABLE GIT_SHA)
+  execute_process(WORKING_DIRECTORY "${CMAKE_CURRENT_SOURCE_DIR}" OUTPUT_VARIABLE GIT_SHA COMMAND "${GIT_EXECUTABLE}" rev-parse HEAD )
+  execute_process(WORKING_DIRECTORY "${CMAKE_CURRENT_SOURCE_DIR}" RESULT_VARIABLE GIT_MOD COMMAND "${GIT_EXECUTABLE}" diff-index HEAD --quiet)
+  execute_process(WORKING_DIRECTORY "${CMAKE_CURRENT_SOURCE_DIR}" OUTPUT_VARIABLE GIT_DATE COMMAND "${GIT_EXECUTABLE}" log -1 --date=format:"%Y-%m-%d %T" --format="%ad")
+  execute_process(WORKING_DIRECTORY "${CMAKE_CURRENT_SOURCE_DIR}" OUTPUT_VARIABLE GIT_TAG RESULT_VARIABLE rv COMMAND "${GIT_EXECUTABLE}" symbolic-ref -q --short HEAD OUTPUT_STRIP_TRAILING_WHITESPACE)
+  if (rv AND NOT rv EQUAL 0)
+    execute_process(WORKING_DIRECTORY "${CMAKE_CURRENT_SOURCE_DIR}" OUTPUT_VARIABLE GIT_TAG COMMAND "${GIT_EXECUTABLE}" describe --tags --exact-match OUTPUT_STRIP_TRAILING_WHITESPACE)
   endif()
 else()
   set(GIT_SHA 0)
+  set(GIT_MOD 1)
 endif()
-
-string(REGEX REPLACE "[^0-9a-f]+" "" GIT_SHA "${GIT_SHA}")
-
+string(REGEX REPLACE "[^0-9a-fA-F]+" "" GIT_SHA "${GIT_SHA}")
+string(REGEX REPLACE "[^0-9: /-]+" "" GIT_DATE "${GIT_DATE}")
 
 option(WITH_MD_LIBRARY "build with MD" ON)
 if(WIN32 AND MSVC)
@@ -199,9 +201,7 @@ endif()
 
 set(BUILD_VERSION_CC ${CMAKE_BINARY_DIR}/build_version.cc)
 configure_file(util/build_version.cc.in ${BUILD_VERSION_CC} @ONLY)
-add_library(build_version OBJECT ${BUILD_VERSION_CC})
-target_include_directories(build_version PRIVATE
-  ${CMAKE_CURRENT_SOURCE_DIR}/util)
+
 if(MSVC)
   set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /Zi /nologo /EHsc /GS /Gd /GR /GF /fp:precise /Zc:wchar_t /Zc:forScope /errorReport:queue")
   set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /FC /d2Zi+ /W4 /wd4127 /wd4800 /wd4996 /wd4351 /wd4100 /wd4204 /wd4324")
@@ -243,14 +243,14 @@ if(CMAKE_SYSTEM_PROCESSOR MATCHES "^(powerpc|ppc)64")
   endif(HAS_ALTIVEC)
 endif(CMAKE_SYSTEM_PROCESSOR MATCHES "^(powerpc|ppc)64")
 
-if(CMAKE_SYSTEM_PROCESSOR MATCHES "aarch64|AARCH64")
+if(CMAKE_SYSTEM_PROCESSOR MATCHES "arm64|aarch64|AARCH64")
         CHECK_C_COMPILER_FLAG("-march=armv8-a+crc+crypto" HAS_ARMV8_CRC)
   if(HAS_ARMV8_CRC)
     message(STATUS " HAS_ARMV8_CRC yes")
     set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -march=armv8-a+crc+crypto -Wno-unused-function")
     set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -march=armv8-a+crc+crypto -Wno-unused-function")
   endif(HAS_ARMV8_CRC)
-endif(CMAKE_SYSTEM_PROCESSOR MATCHES "aarch64|AARCH64")
+endif(CMAKE_SYSTEM_PROCESSOR MATCHES "arm64|aarch64|AARCH64")
 
 option(PORTABLE "build a portable binary" OFF)
 option(FORCE_SSE42 "force building with SSE4.2, even when PORTABLE=ON" OFF)
@@ -466,11 +466,6 @@ if(CMAKE_SYSTEM_NAME MATCHES "Cygwin")
   add_definitions(-fno-builtin-memcmp -DCYGWIN)
 elseif(CMAKE_SYSTEM_NAME MATCHES "Darwin")
   add_definitions(-DOS_MACOSX)
-  if(CMAKE_SYSTEM_PROCESSOR MATCHES arm)
-    add_definitions(-DIOS_CROSS_COMPILE -DROCKSDB_LITE)
-    # no debug info for IOS, that will make our library big
-    add_definitions(-DNDEBUG)
-  endif()
 elseif(CMAKE_SYSTEM_NAME MATCHES "Linux")
   add_definitions(-DOS_LINUX)
 elseif(CMAKE_SYSTEM_NAME MATCHES "SunOS")
@@ -814,6 +809,8 @@ set(SOURCES
         utilities/transactions/lock/lock_manager.cc
         utilities/transactions/lock/point/point_lock_tracker.cc
         utilities/transactions/lock/point/point_lock_manager.cc
+        utilities/transactions/lock/range/range_tree/range_tree_lock_manager.cc
+        utilities/transactions/lock/range/range_tree/range_tree_lock_tracker.cc
         utilities/transactions/optimistic_transaction_db_impl.cc
         utilities/transactions/optimistic_transaction.cc
         utilities/transactions/pessimistic_transaction.cc
@@ -828,8 +825,21 @@ set(SOURCES
         utilities/transactions/write_unprepared_txn_db.cc
         utilities/ttl/db_ttl_impl.cc
         utilities/write_batch_with_index/write_batch_with_index.cc
-        utilities/write_batch_with_index/write_batch_with_index_internal.cc
-        $<TARGET_OBJECTS:build_version>)
+        utilities/write_batch_with_index/write_batch_with_index_internal.cc)
+
+list(APPEND SOURCES
+  utilities/transactions/lock/range/range_tree/lib/locktree/concurrent_tree.cc
+  utilities/transactions/lock/range/range_tree/lib/locktree/keyrange.cc
+  utilities/transactions/lock/range/range_tree/lib/locktree/lock_request.cc
+  utilities/transactions/lock/range/range_tree/lib/locktree/locktree.cc
+  utilities/transactions/lock/range/range_tree/lib/locktree/manager.cc
+  utilities/transactions/lock/range/range_tree/lib/locktree/range_buffer.cc
+  utilities/transactions/lock/range/range_tree/lib/locktree/treenode.cc
+  utilities/transactions/lock/range/range_tree/lib/locktree/txnid_set.cc
+  utilities/transactions/lock/range/range_tree/lib/locktree/wfg.cc
+  utilities/transactions/lock/range/range_tree/lib/standalone_port.cc
+  utilities/transactions/lock/range/range_tree/lib/util/dbt.cc
+  utilities/transactions/lock/range/range_tree/lib/util/memarena.cc)
 
 if(HAVE_SSE42 AND NOT MSVC)
   set_source_files_properties(
@@ -906,12 +916,12 @@ else()
   set(SYSTEM_LIBS ${CMAKE_THREAD_LIBS_INIT})
 endif()
 
-add_library(${ROCKSDB_STATIC_LIB} STATIC ${SOURCES})
+add_library(${ROCKSDB_STATIC_LIB} STATIC ${SOURCES} ${BUILD_VERSION_CC})
 target_link_libraries(${ROCKSDB_STATIC_LIB} PRIVATE
   ${THIRDPARTY_LIBS} ${SYSTEM_LIBS})
 
 if(ROCKSDB_BUILD_SHARED)
-  add_library(${ROCKSDB_SHARED_LIB} SHARED ${SOURCES})
+  add_library(${ROCKSDB_SHARED_LIB} SHARED ${SOURCES} ${BUILD_VERSION_CC})
   target_link_libraries(${ROCKSDB_SHARED_LIB} PRIVATE
     ${THIRDPARTY_LIBS} ${SYSTEM_LIBS})
 
@@ -929,7 +939,7 @@ if(ROCKSDB_BUILD_SHARED)
                           LINKER_LANGUAGE CXX
                           VERSION ${rocksdb_VERSION}
                           SOVERSION ${rocksdb_VERSION_MAJOR}
-                          OUTPUT_NAME "rocksdb")
+                          OUTPUT_NAME "rocksdb${ARTIFACT_SUFFIX}")
   endif()
 endif()
 
@@ -1072,6 +1082,7 @@ if(WITH_TESTS)
         db/db_iter_test.cc
         db/db_iter_stress_test.cc
         db/db_iterator_test.cc
+        db/db_kv_checksum_test.cc
         db/db_log_iter_test.cc
         db/db_memtable_test.cc
         db/db_merge_operator_test.cc
@@ -1160,6 +1171,7 @@ if(WITH_TESTS)
         table/table_test.cc
         table/block_fetcher_test.cc
         test_util/testutil_test.cc
+        trace_replay/block_cache_tracer_test.cc
         trace_replay/io_tracer_test.cc
         tools/block_cache_analyzer/block_cache_trace_analyzer_test.cc
         tools/io_tracer_parser_test.cc
@@ -1210,6 +1222,7 @@ if(WITH_TESTS)
         utilities/transactions/lock/point/point_lock_manager_test.cc
         utilities/transactions/write_prepared_transaction_test.cc
         utilities/transactions/write_unprepared_transaction_test.cc
+        utilities/transactions/lock/range/range_locking_test.cc
         utilities/ttl/ttl_test.cc
         utilities/write_batch_with_index/write_batch_with_index_test.cc
     )

HISTORY.md

@@ -1,27 +1,67 @@
 # Rocksdb Change Log
-## 6.15.5 (02/05/2021)
+## Unreleased
+
+## 6.18.0 (02/19/2021)
+### Behavior Changes
+* When retryable IO error occurs during compaction, it is mapped to soft error and set the BG error. However, auto resume is not called to clean the soft error since compaction will reschedule by itself. In this change, When retryable IO error occurs during compaction, BG error is not set. User will be informed the error via EventHelper.
+* Introduce a new trace file format for query tracing and replay and trace file version is bump up to 0.2. A payload map is added as the first portion of the payload. We will not have backward compatible issues when adding new entries to trace records. Added the iterator_upper_bound and iterator_lower_bound in Seek and SeekForPrev tracing function. Added them as the new payload member for iterator tracing.
+
+### New Features
+* Add support for key-value integrity protection in live updates from the user buffers provided to `WriteBatch` through the write to RocksDB's in-memory update buffer (memtable). This is intended to detect some cases of in-memory data corruption, due to either software or hardware errors. Users can enable protection by constructing their `WriteBatch` with `protection_bytes_per_key == 8`.
+* Add support for updating `full_history_ts_low` option in manual compaction, which is for old timestamp data GC.
+* Add a mechanism for using Makefile to build external plugin code into the RocksDB libraries/binaries. This intends to simplify compatibility and distribution for plugins (e.g., special-purpose `FileSystem`s) whose source code resides outside the RocksDB repo. See "plugin/README.md" for developer details, and "PLUGINS.md" for a listing of available plugins.
+* Added memory pre-fetching for experimental Ribbon filter, which especially optimizes performance with batched MultiGet.
+* A new, experimental version of BlobDB (key-value separation) is now available. The new implementation is integrated into the RocksDB core, i.e. it is accessible via the usual `rocksdb::DB` API, as opposed to the separate `rocksdb::blob_db::BlobDB` interface used by the earlier version, and can be configured on a per-column family basis using the configuration options `enable_blob_files`, `min_blob_size`, `blob_file_size`, `blob_compression_type`, `enable_blob_garbage_collection`, and `blob_garbage_collection_age_cutoff`. It extends RocksDB's consistency guarantees to blobs, and offers more features and better performance. Note that some features, most notably `Merge`, compaction filters, and backup/restore are not yet supported, and there is no support for migrating a database created by the old implementation.
+
 ### Bug Fixes
 * Since 6.15.0, `TransactionDB` returns error `Status`es from calls to `DeleteRange()` and calls to `Write()` where the `WriteBatch` contains a range deletion. Previously such operations may have succeeded while not providing the expected transactional guarantees. There are certain cases where range deletion can still be used on such DBs; see the API doc on `TransactionDB::DeleteRange()` for details.
 * `OptimisticTransactionDB` now returns error `Status`es from calls to `DeleteRange()` and calls to `Write()` where the `WriteBatch` contains a range deletion. Previously such operations may have succeeded while not providing the expected transactional guarantees.
+* Fix `WRITE_PREPARED`, `WRITE_UNPREPARED` TransactionDB `MultiGet()` may return uncommitted data with snapshot.
+* In DB::OpenForReadOnly, if any error happens while checking Manifest file path, it was overridden by Status::NotFound. It has been fixed and now actual error is returned.
+
+### Public API Change
+* Added a "only_mutable_options" flag to the ConfigOptions.  When this flag is "true", the Configurable functions and convenience methods (such as GetDBOptionsFromString) will only deal with options that are marked as mutable.  When this flag is true, only options marked as mutable can be configured (a Status::InvalidArgument will be returned) and options not marked as mutable will not be returned or compared.  The default is "false", meaning to compare all options.
+* Add new Append and PositionedAppend APIs to FileSystem to bring the data verification information (data checksum information) from upper layer (e.g., WritableFileWriter) to the storage layer. In this way, the customized FileSystem is able to verify the correctness of data being written to the storage on time. Add checksum_handoff_file_types to DBOptions. User can use this option to control which file types (Currently supported file tyes: kWALFile, kTableFile, kDescriptorFile.) should use the new Append and PositionedAppend APIs to handoff the verification information. Currently, RocksDB only use crc32c to calculate the checksum for write handoff.
+* Add an option, `CompressionOptions::max_dict_buffer_bytes`, to limit the in-memory buffering for selecting samples for generating/training a dictionary. The limit is currently loosely adhered to.
+
+## 6.17.0 (01/15/2021)
+### Behavior Changes
+* When verifying full file checksum with `DB::VerifyFileChecksums()`, we now fail with `Status::InvalidArgument` if the name of the checksum generator used for verification does not match the name of the checksum generator used for protecting the file when it was created.
+* Since RocksDB does not continue write the same file if a file write fails for any reason, the file scope write IO error is treated the same as retryable IO error. More information about error handling of file scope IO error is included in `ErrorHandler::SetBGError`.
 
-## 6.15.4 (01/21/2021)
 ### Bug Fixes
+* Version older than 6.15 cannot decode VersionEdits `WalAddition` and `WalDeletion`, fixed this by changing the encoded format of them to be ignorable by older versions.
 * Fix a race condition between DB startups and shutdowns in managing the periodic background worker threads. One effect of this race condition could be the process being terminated.
 
-## 6.15.3 (01/07/2021)
-### Bug Fixes
-* For Java builds, fix errors due to missing compression library includes.
+### Public API Change
+* Add a public API WriteBufferManager::dummy_entries_in_cache_usage() which reports the size of dummy entries stored in cache (passed to WriteBufferManager). Dummy entries are used to account for DataBlocks.
+* Add a public API GetRocksBuildProperties and GetRocksBuildInfoAsString to get properties about the current build.  These properties may include settings related to the GIT settings (branch, timestamp).  This change also sets the "build date" based on the GIT properties, rather than the actual build time, thereby enabling more reproducible builds.
 
-## 6.15.2 (12/22/2020)
-### Bug Fixes
-* Fix failing RocksJava test compilation and add CI jobs
-* Fix jemalloc compilation issue on macOS
-* Fix build issues - compatibility with older gcc, older jemalloc libraries, docker warning when building i686 binaries
+## 6.16.0 (12/18/2020)
+### Behavior Changes
+* Attempting to write a merge operand without explicitly configuring `merge_operator` now fails immediately, causing the DB to enter read-only mode. Previously, failure was deferred until the `merge_operator` was needed by a user read or a background operation.
 
-## 6.15.1 (12/01/2020)
 ### Bug Fixes
-* Truncated WALs ending in incomplete records can no longer produce gaps in the recovered data when `WALRecoveryMode::kPointInTimeRecovery` is used. Gaps are still possible when WALs are truncated exactly on record boundaries.
+* Truncated WALs ending in incomplete records can no longer produce gaps in the recovered data when `WALRecoveryMode::kPointInTimeRecovery` is used. Gaps are still possible when WALs are truncated exactly on record boundaries; for complete protection, users should enable `track_and_verify_wals_in_manifest`.
 * Fix a bug where compressed blocks read by MultiGet are not inserted into the compressed block cache when use_direct_reads = true.
+* Fixed the issue of full scanning on obsolete files when there are too many outstanding compactions with ConcurrentTaskLimiter enabled.
+* Fixed the logic of populating native data structure for `read_amp_bytes_per_bit` during OPTIONS file parsing on big-endian architecture. Without this fix, original code introduced in PR7659, when running on big-endian machine, can mistakenly store read_amp_bytes_per_bit (an uint32) in little endian format. Future access to `read_amp_bytes_per_bit` will give wrong values. Little endian architecture is not affected.
+* Fixed prefix extractor with timestamp issues.
+* Fixed a bug in atomic flush: in two-phase commit mode, the minimum WAL log number to keep is incorrect.
+* Fixed a bug related to checkpoint in PR7789: if there are multiple column families, and the checkpoint is not opened as read only, then in rare cases, data loss may happen in the checkpoint. Since backup engine relies on checkpoint, it may also be affected.
+* When ldb --try_load_options is used with the --column_family option, the ColumnFamilyOptions for the specified column family was not loaded from the OPTIONS file. Fix it so its loaded from OPTIONS and then overridden with command line overrides.
+
+### New Features
+* User defined timestamp feature supports `CompactRange` and `GetApproximateSizes`.
+* Support getting aggregated table properties (kAggregatedTableProperties and kAggregatedTablePropertiesAtLevel) with DB::GetMapProperty, for easier access to the data in a structured format.
+* Experimental option BlockBasedTableOptions::optimize_filters_for_memory now works with experimental Ribbon filter (as well as Bloom filter).
+
+### Public API Change
+* Deprecated public but rarely-used FilterBitsBuilder::CalculateNumEntry, which is replaced with ApproximateNumEntries taking a size_t parameter and returning size_t.
+* To improve portability the functions `Env::GetChildren` and `Env::GetChildrenFileAttributes` will no longer return entries for the special directories `.` or `..`.
+* Added a new option `track_and_verify_wals_in_manifest`. If `true`, the log numbers and sizes of the synced WALs are tracked in MANIFEST, then during DB recovery, if a synced WAL is missing from disk, or the WAL's size does not match the recorded size in MANIFEST, an error will be reported and the recovery will be aborted. Note that this option does not work with secondary instance.
+* `rocksdb_approximate_sizes` and `rocksdb_approximate_sizes_cf` in the C API now requires an error pointer (`char** errptr`) for receiving any error.
+* All overloads of DB::GetApproximateSizes now return Status, so that any failure to obtain the sizes is indicated to the caller.
 
 ## 6.15.0 (11/13/2020)
 ### Bug Fixes
@@ -38,8 +78,7 @@
 * Fixed a potential bug caused by evaluating `TableBuilder::NeedCompact()` before `TableBuilder::Finish()` in compaction job. For example, the `NeedCompact()` method of `CompactOnDeletionCollector` returned by built-in `CompactOnDeletionCollectorFactory` requires `BlockBasedTable::Finish()` to return the correct result. The bug can cause a compaction-generated file not to be marked for future compaction based on deletion ratio.
 * Fixed a seek issue with prefix extractor and timestamp.
 * Fixed a bug of encoding and parsing BlockBasedTableOptions::read_amp_bytes_per_bit as a 64-bit integer.
-* Fixed the logic of populating native data structure for `read_amp_bytes_per_bit` during OPTIONS file parsing on big-endian architecture. Without this fix, original code introduced in PR7659, when running on big-endian machine, can mistakenly store read_amp_bytes_per_bit (an uint32) in little endian format. Future access to `read_amp_bytes_per_bit` will give wrong values. Little endian architecture is not affected.
-
+* Fixed a bug of a recovery corner case, details in PR7621.
 
 ### Public API Change
 * Deprecate `BlockBasedTableOptions::pin_l0_filter_and_index_blocks_in_cache` and `BlockBasedTableOptions::pin_top_level_index_and_filter`. These options still take effect until users migrate to the replacement APIs in `BlockBasedTableOptions::metadata_cache_options`. Migration guidance can be found in the API comments on the deprecated options.

Makefile

@@ -190,6 +190,11 @@ else
 endif
 
 ifdef ASSERT_STATUS_CHECKED
+# For ASC, turn off constructor elision, preventing the case where a constructor returned
+# by a method may pass the ASC check if the status is checked in the inner method.  Forcing
+# the copy constructor to be invoked disables the optimization and will cause the calling method
+# to check the status in order to prevent an error from being raised.
+PLATFORM_CXXFLAGS += -fno-elide-constructors
 ifeq ($(filter -DROCKSDB_ASSERT_STATUS_CHECKED,$(OPT)),)
 	OPT += -DROCKSDB_ASSERT_STATUS_CHECKED
 endif
@@ -264,6 +269,12 @@ dummy := $(shell (export ROCKSDB_ROOT="$(CURDIR)"; \
 # this file is generated by the previous line to set build flags and sources
 include make_config.mk
 
+ROCKSDB_PLUGIN_MKS = $(foreach plugin, $(ROCKSDB_PLUGINS), plugin/$(plugin)/*.mk)
+include $(ROCKSDB_PLUGIN_MKS)
+ROCKSDB_PLUGIN_SOURCES = $(foreach plugin, $(ROCKSDB_PLUGINS), $(foreach source, $($(plugin)_SOURCES), plugin/$(plugin)/$(source)))
+ROCKSDB_PLUGIN_HEADERS = $(foreach plugin, $(ROCKSDB_PLUGINS), $(foreach header, $($(plugin)_HEADERS), plugin/$(plugin)/$(header)))
+PLATFORM_LDFLAGS += $(foreach plugin, $(ROCKSDB_PLUGINS), $($(plugin)_LDFLAGS))
+
 export JAVAC_ARGS
 CLEAN_FILES += make_config.mk rocksdb.pc
 
@@ -474,33 +485,9 @@ CXXFLAGS += $(WARNING_FLAGS) -I. -I./include $(PLATFORM_CXXFLAGS) $(OPT) -Woverl
 
 LDFLAGS += $(PLATFORM_LDFLAGS)
 
-# If NO_UPDATE_BUILD_VERSION is set we don't update util/build_version.cc, but
-# the file needs to already exist or else the build will fail
-ifndef NO_UPDATE_BUILD_VERSION
-date := $(shell date +%F)
-ifdef FORCE_GIT_SHA
-	git_sha := $(FORCE_GIT_SHA)
-else
-	git_sha := $(shell git rev-parse HEAD 2>/dev/null)
-endif
-gen_build_version = sed -e s/@@GIT_SHA@@/$(git_sha)/ -e s/@@GIT_DATE_TIME@@/$(date)/ util/build_version.cc.in
-
-# Record the version of the source that we are compiling.
-# We keep a record of the git revision in this file.  It is then built
-# as a regular source file as part of the compilation process.
-# One can run "strings executable_filename | grep _build_" to find
-# the version of the source that we used to build the executable file.
-FORCE:
-util/build_version.cc: FORCE
-	$(AM_V_GEN)rm -f $@-t
-	$(AM_V_at)$(gen_build_version) > $@-t
-	$(AM_V_at)if test -f $@; then					\
-	  cmp -s $@-t $@ && rm -f $@-t || mv -f $@-t $@;		\
-	else mv -f $@-t $@; fi
-endif
-
 OBJ_DIR?=.
 LIB_OBJECTS = $(patsubst %.cc, $(OBJ_DIR)/%.o, $(LIB_SOURCES))
+LIB_OBJECTS += $(patsubst %.cc, $(OBJ_DIR)/%.o, $(ROCKSDB_PLUGIN_SOURCES))
 ifeq ($(HAVE_POWER8),1)
 LIB_OBJECTS += $(patsubst %.c, $(OBJ_DIR)/%.o, $(LIB_SOURCES_C))
 LIB_OBJECTS += $(patsubst %.S, $(OBJ_DIR)/%.o, $(LIB_SOURCES_ASM))
@@ -510,6 +497,12 @@ ifeq ($(USE_FOLLY_DISTRIBUTED_MUTEX),1)
   LIB_OBJECTS += $(patsubst %.cpp, $(OBJ_DIR)/%.o, $(FOLLY_SOURCES))
 endif
 
+# range_tree is not compatible with non GNU libc on ppc64
+# see https://jira.percona.com/browse/PS-7559
+ifneq ($(PPC_LIBC_IS_GNU),0)
+  LIB_OBJECTS += $(patsubst %.cc, $(OBJ_DIR)/%.o, $(RANGE_TREE_SOURCES))
+endif
+
 GTEST = $(OBJ_DIR)/$(GTEST_DIR)/gtest/gtest-all.o
 TESTUTIL = $(OBJ_DIR)/test_util/testutil.o
 TESTHARNESS = $(OBJ_DIR)/test_util/testharness.o $(TESTUTIL) $(GTEST)
@@ -558,6 +551,7 @@ PARALLEL_TEST = \
 	table_test \
 	transaction_test \
 	point_lock_manager_test \
+	range_locking_test \
 	write_prepared_transaction_test \
 	write_unprepared_transaction_test \
 
@@ -586,21 +580,63 @@ ifdef ASSERT_STATUS_CHECKED
 		blob_file_reader_test \
 		bloom_test \
 		cassandra_format_test \
+		cassandra_functional_test \
 		cassandra_row_merge_test \
 		cassandra_serialize_test \
 		cleanable_test \
+		checkpoint_test \
 		coding_test \
 		crc32c_test \
 		dbformat_test \
 		db_basic_test \
+		compact_files_test \
+		compaction_picker_test \
+		comparator_db_test \
+		db_encryption_test \
+		db_iter_test \
+		db_iter_stress_test \
+		db_log_iter_test \
+		db_bloom_filter_test \
 		db_blob_basic_test \
+		db_blob_index_test \
+		db_block_cache_test \
+		db_compaction_test \
+		db_compaction_filter_test \
+		db_dynamic_level_test \
 		db_flush_test \
+		db_inplace_update_test \
+		db_io_failure_test \
+		db_iterator_test \
+		db_kv_checksum_test \
+		db_logical_block_size_cache_test \
+		db_memtable_test \
+		db_merge_operand_test \
+		db_merge_operator_test \
+		db_wal_test \
 		db_with_timestamp_basic_test \
 		db_with_timestamp_compaction_test \
+		db_write_test \
 		db_options_test \
 		db_properties_test \
+		db_range_del_test \
 		db_secondary_test \
+		deletefile_test \
+		external_sst_file_test \
 		options_file_test \
+		db_sst_test \
+		db_statistics_test \
+		db_table_properties_test \
+		db_tailing_iter_test \
+		fault_injection_test \
+		listener_test \
+		log_test \
+		manual_compaction_test \
+		obsolete_files_test \
+		perf_context_test \
+		periodic_work_scheduler_test \
+		perf_context_test \
+		version_set_test \
+		wal_manager_test \
 		defer_test \
 		filename_test \
 		dynamic_bloom_test \
@@ -609,8 +645,10 @@ ifdef ASSERT_STATUS_CHECKED
 		env_logger_test \
 		event_logger_test \
 		error_handler_fs_test \
+		external_sst_file_basic_test \
 		auto_roll_logger_test \
 		file_indexer_test \
+		delete_scheduler_test \
 		flush_job_test \
 		hash_table_test \
 		hash_test \
@@ -621,9 +659,11 @@ ifdef ASSERT_STATUS_CHECKED
 		iostats_context_test \
 		ldb_cmd_test \
 		memkind_kmem_allocator_test \
+		merge_test \
 		merger_test \
 		mock_env_test \
 		object_registry_test \
+		optimistic_transaction_test \
 		prefix_test \
 		plain_table_db_test \
 		repair_test \
@@ -631,6 +671,8 @@ ifdef ASSERT_STATUS_CHECKED
 		customizable_test \
 		options_settable_test \
 		options_test \
+		point_lock_manager_test \
+		random_access_file_reader_test \
 		random_test \
 		range_del_aggregator_test \
 		sst_file_reader_test \
@@ -639,11 +681,14 @@ ifdef ASSERT_STATUS_CHECKED
 		ribbon_test \
 		skiplist_test \
 		slice_test \
+		slice_transform_test \
 		sst_dump_test \
 		statistics_test \
 		stats_history_test \
+		stringappend_test \
 		thread_local_test \
 		trace_analyzer_test \
+		transaction_test \
 		env_timed_test \
 		filelock_test \
 		timer_queue_test \
@@ -658,25 +703,45 @@ ifdef ASSERT_STATUS_CHECKED
 		version_builder_test \
 		version_edit_test \
 		work_queue_test \
+		write_buffer_manager_test \
 		write_controller_test \
+		write_prepared_transaction_test \
+		write_unprepared_transaction_test \
 		compaction_iterator_test \
 		compaction_job_test \
 		compaction_job_stats_test \
-	        io_tracer_test \
+		io_tracer_test \
+		io_tracer_parser_test \
+		prefetch_test \
 		merge_helper_test \
 		memtable_list_test \
 		flush_job_test \
 		block_based_filter_block_test \
 		block_fetcher_test \
+		block_test \
+		data_block_hash_index_test \
 		full_filter_block_test \
 		partitioned_filter_block_test \
 		column_family_test \
 		file_reader_writer_test \
+		rate_limiter_test \
 		corruption_test \
+		reduce_levels_test \
+		thread_list_test \
+		compact_on_deletion_collector_test \
 		db_universal_compaction_test \
 		import_column_family_test \
+		option_change_migration_test \
+		cuckoo_table_builder_test \
+		cuckoo_table_db_test \
+		cuckoo_table_reader_test \
 		memory_test \
 		table_test \
+		backupable_db_test \
+		blob_db_test \
+		ttl_test \
+		write_batch_test \
+		write_batch_with_index_test \
 
 ifeq ($(USE_FOLLY_DISTRIBUTED_MUTEX),1)
 TESTS_PASSING_ASC += folly_synchronization_distributed_mutex_test
@@ -787,6 +852,37 @@ ROCKSDB_MAJOR = $(shell egrep "ROCKSDB_MAJOR.[0-9]" include/rocksdb/version.h |
 ROCKSDB_MINOR = $(shell egrep "ROCKSDB_MINOR.[0-9]" include/rocksdb/version.h | cut -d ' ' -f 3)
 ROCKSDB_PATCH = $(shell egrep "ROCKSDB_PATCH.[0-9]" include/rocksdb/version.h | cut -d ' ' -f 3)
 
+# If NO_UPDATE_BUILD_VERSION is set we don't update util/build_version.cc, but
+# the file needs to already exist or else the build will fail
+ifndef NO_UPDATE_BUILD_VERSION
+
+# By default, use the current date-time as the date.  If there are no changes,
+# we will use the last commit date instead.
+build_date := $(shell date "+%Y-%m-%d %T")
+
+ifdef FORCE_GIT_SHA
+	git_sha := $(FORCE_GIT_SHA)
+	git_mod := 1
+	git_date := $(build_date)
+else
+	git_sha := $(shell git rev-parse HEAD 2>/dev/null)
+	git_tag  := $(shell git symbolic-ref -q --short HEAD 2> /dev/null || git describe --tags --exact-match 2>/dev/null)
+	git_mod  := $(shell git diff-index HEAD --quiet 2>/dev/null; echo $$?)
+	git_date := $(shell git log -1 --date=format:"%Y-%m-%d %T" --format="%ad" 2>/dev/null)
+endif
+gen_build_version = sed -e s/@GIT_SHA@/$(git_sha)/ -e s:@GIT_TAG@:"$(git_tag)": -e s/@GIT_MOD@/"$(git_mod)"/ -e s/@BUILD_DATE@/"$(build_date)"/ -e s/@GIT_DATE@/"$(git_date)"/ util/build_version.cc.in
+
+# Record the version of the source that we are compiling.
+# We keep a record of the git revision in this file.  It is then built
+# as a regular source file as part of the compilation process.
+# One can run "strings executable_filename | grep _build_" to find
+# the version of the source that we used to build the executable file.
+util/build_version.cc: $(filter-out $(OBJ_DIR)/util/build_version.o, $(LIB_OBJECTS)) util/build_version.cc.in
+	$(AM_V_GEN)rm -f $@-t
+	$(AM_V_at)$(gen_build_version) > $@
+endif
+CLEAN_FILES += util/build_version.cc
+
 default: all
 
 #-----------------------------------------------
@@ -1329,7 +1425,7 @@ $(STATIC_TEST_LIBRARY): $(TEST_OBJECTS)
 	$(AM_V_AR)rm -f $@ $(SHARED_TEST_LIBRARY)
 	$(AM_V_at)$(AR) $(ARFLAGS) $@ $^
 
-$(STATIC_TOOLS_LIBRARY): $(BENCH_OBJECTS) $(TOOL_OBJECTS)
+$(STATIC_TOOLS_LIBRARY): $(TOOL_OBJECTS)
 	$(AM_V_AR)rm -f $@ $(SHARED_TOOLS_LIBRARY)
 	$(AM_V_at)$(AR) $(ARFLAGS) $@ $^
 
@@ -1526,6 +1622,9 @@ db_inplace_update_test: $(OBJ_DIR)/db/db_inplace_update_test.o $(TEST_LIBRARY) $
 db_iterator_test: $(OBJ_DIR)/db/db_iterator_test.o $(TEST_LIBRARY) $(LIBRARY)
 	$(AM_LINK)
 
+db_kv_checksum_test: $(OBJ_DIR)/db/db_kv_checksum_test.o $(TEST_LIBRARY) $(LIBRARY)
+	$(AM_LINK)
+
 db_memtable_test: $(OBJ_DIR)/db/db_memtable_test.o $(TEST_LIBRARY) $(LIBRARY)
 	$(AM_LINK)
 
@@ -1909,6 +2008,9 @@ blob_db_test: $(OBJ_DIR)/utilities/blob_db/blob_db_test.o $(TEST_LIBRARY) $(LIBR
 repeatable_thread_test: $(OBJ_DIR)/util/repeatable_thread_test.o $(TEST_LIBRARY) $(LIBRARY)
 	$(AM_LINK)
 
+range_locking_test: utilities/transactions/lock/range/range_locking_test.o $(TEST_LIBRARY) $(LIBRARY)
+	$(AM_LINK)
+
 range_tombstone_fragmenter_test: $(OBJ_DIR)/db/range_tombstone_fragmenter_test.o $(TEST_LIBRARY) $(LIBRARY)
 	$(AM_LINK)
 
@@ -1987,6 +2089,10 @@ install-headers: gen-pc
 	for header in `$(FIND) "include/rocksdb" -type f -name *.h`; do \
 		install -C -m 644 $$header $(DESTDIR)/$(PREFIX)/$$header; \
 	done
+	for header in $(ROCKSDB_PLUGIN_HEADERS); do \
+		install -d $(DESTDIR)/$(PREFIX)/include/rocksdb/`dirname $$header`; \
+		install -C -m 644 $$header $(DESTDIR)/$(PREFIX)/include/rocksdb/$$header; \
+	done
 	install -C -m 644 rocksdb.pc $(INSTALL_LIBDIR)/pkgconfig/rocksdb.pc
 
 install-static: install-headers $(LIBRARY)
@@ -2070,11 +2176,11 @@ BZIP2_DOWNLOAD_BASE ?= https://sourceware.org/pub/bzip2
 SNAPPY_VER ?= 1.1.8
 SNAPPY_SHA256 ?= 16b677f07832a612b0836178db7f374e414f94657c138e6993cbfc5dcc58651f
 SNAPPY_DOWNLOAD_BASE ?= https://github.com/google/snappy/archive
-LZ4_VER ?= 1.9.2
-LZ4_SHA256 ?= 658ba6191fa44c92280d4aa2c271b0f4fbc0e34d249578dd05e50e76d0e5efcc
+LZ4_VER ?= 1.9.3
+LZ4_SHA256 ?= 030644df4611007ff7dc962d981f390361e6c97a34e5cbc393ddfbe019ffe2c1
 LZ4_DOWNLOAD_BASE ?= https://github.com/lz4/lz4/archive
-ZSTD_VER ?= 1.4.5
-ZSTD_SHA256 ?= 734d1f565c42f691f8420c8d06783ad818060fc390dee43ae0a89f86d0a4f8c2
+ZSTD_VER ?= 1.4.9
+ZSTD_SHA256 ?= acf714d98e3db7b876e5b540cbf6dee298f60eb3c0723104f6d3f065cd60d6a8
 ZSTD_DOWNLOAD_BASE ?= https://github.com/facebook/zstd/archive
 CURL_SSL_OPTS ?= --tlsv1
 
@@ -2122,7 +2228,7 @@ zlib-$(ZLIB_VER).tar.gz:
 libz.a: zlib-$(ZLIB_VER).tar.gz
 	-rm -rf zlib-$(ZLIB_VER)
 	tar xvzf zlib-$(ZLIB_VER).tar.gz
-	cd zlib-$(ZLIB_VER) && CFLAGS='-fPIC ${EXTRA_CFLAGS}' LDFLAGS='${EXTRA_LDFLAGS}' ./configure --static && $(MAKE)
+	cd zlib-$(ZLIB_VER) && CFLAGS='-fPIC ${JAVA_STATIC_DEPS_CCFLAGS} ${EXTRA_CFLAGS}' LDFLAGS='${JAVA_STATIC_DEPS_LDFLAGS} ${EXTRA_LDFLAGS}' ./configure --static && $(MAKE)
 	cp zlib-$(ZLIB_VER)/libz.a .
 
 bzip2-$(BZIP2_VER).tar.gz:
@@ -2136,7 +2242,7 @@ bzip2-$(BZIP2_VER).tar.gz:
 libbz2.a: bzip2-$(BZIP2_VER).tar.gz
 	-rm -rf bzip2-$(BZIP2_VER)
 	tar xvzf bzip2-$(BZIP2_VER).tar.gz
-	cd bzip2-$(BZIP2_VER) && $(MAKE) CFLAGS='-fPIC -O2 -g -D_FILE_OFFSET_BITS=64 ${EXTRA_CFLAGS}' AR='ar ${EXTRA_ARFLAGS}'
+	cd bzip2-$(BZIP2_VER) && $(MAKE) CFLAGS='-fPIC -O2 -g -D_FILE_OFFSET_BITS=64 ${JAVA_STATIC_DEPS_CCFLAGS} ${EXTRA_CFLAGS}' LDFLAGS='${JAVA_STATIC_DEPS_LDFLAGS} ${EXTRA_LDFLAGS}' AR='ar ${EXTRA_ARFLAGS}'
 	cp bzip2-$(BZIP2_VER)/libbz2.a .
 
 snappy-$(SNAPPY_VER).tar.gz:
@@ -2151,7 +2257,7 @@ libsnappy.a: snappy-$(SNAPPY_VER).tar.gz
 	-rm -rf snappy-$(SNAPPY_VER)
 	tar xvzf snappy-$(SNAPPY_VER).tar.gz
 	mkdir snappy-$(SNAPPY_VER)/build
-	cd snappy-$(SNAPPY_VER)/build && CFLAGS='${EXTRA_CFLAGS}' CXXFLAGS='${EXTRA_CXXFLAGS}' LDFLAGS='${EXTRA_LDFLAGS}' cmake -DCMAKE_POSITION_INDEPENDENT_CODE=ON .. && $(MAKE) ${SNAPPY_MAKE_TARGET}
+	cd snappy-$(SNAPPY_VER)/build && CFLAGS='${JAVA_STATIC_DEPS_CCFLAGS} ${EXTRA_CFLAGS}' CXXFLAGS='${JAVA_STATIC_DEPS_CXXFLAGS} ${EXTRA_CXXFLAGS}' LDFLAGS='${JAVA_STATIC_DEPS_LDFLAGS} ${EXTRA_LDFLAGS}' cmake -DCMAKE_POSITION_INDEPENDENT_CODE=ON ${PLATFORM_CMAKE_FLAGS} .. && $(MAKE) ${SNAPPY_MAKE_TARGET}
 	cp snappy-$(SNAPPY_VER)/build/libsnappy.a .
 
 lz4-$(LZ4_VER).tar.gz:
@@ -2165,7 +2271,7 @@ lz4-$(LZ4_VER).tar.gz:
 liblz4.a: lz4-$(LZ4_VER).tar.gz
 	-rm -rf lz4-$(LZ4_VER)
 	tar xvzf lz4-$(LZ4_VER).tar.gz
-	cd lz4-$(LZ4_VER)/lib && $(MAKE) CFLAGS='-fPIC -O2 ${EXTRA_CFLAGS}' all
+	cd lz4-$(LZ4_VER)/lib && $(MAKE) CFLAGS='-fPIC -O2 ${JAVA_STATIC_DEPS_CCFLAGS} ${EXTRA_CFLAGS}' LDFLAGS='${JAVA_STATIC_DEPS_LDFLAGS} ${EXTRA_LDFLAGS}' all
 	cp lz4-$(LZ4_VER)/lib/liblz4.a .
 
 zstd-$(ZSTD_VER).tar.gz:
@@ -2179,7 +2285,7 @@ zstd-$(ZSTD_VER).tar.gz:
 libzstd.a: zstd-$(ZSTD_VER).tar.gz
 	-rm -rf zstd-$(ZSTD_VER)
 	tar xvzf zstd-$(ZSTD_VER).tar.gz
-	cd zstd-$(ZSTD_VER)/lib && DESTDIR=. PREFIX= $(MAKE) CFLAGS='-fPIC -O2 ${EXTRA_CFLAGS}' install
+	cd zstd-$(ZSTD_VER)/lib && DESTDIR=. PREFIX= $(MAKE) CFLAGS='-fPIC -O2 ${JAVA_STATIC_DEPS_CCFLAGS} ${EXTRA_CFLAGS}' LDFLAGS='${JAVA_STATIC_DEPS_LDFLAGS} ${EXTRA_LDFLAGS}' libzstd.a
 	cp zstd-$(ZSTD_VER)/lib/libzstd.a .
 
 # A version of each $(LIB_OBJECTS) compiled with -fPIC and a fixed set of static compression libraries
@@ -2203,7 +2309,7 @@ endif
 	$(MAKE) rocksdbjavastatic_javalib
 
 rocksdbjavastatic_javalib:
-	cd java;$(MAKE) javalib
+	cd java; SHA256_CMD='$(SHA256_CMD)' $(MAKE) javalib
 	rm -f java/target/$(ROCKSDBJNILIB)
 	$(CXX) $(CXXFLAGS) -I./java/. $(JAVA_INCLUDE) -shared -fPIC \
 	  -o ./java/target/$(ROCKSDBJNILIB) $(JNI_NATIVE_SOURCES) \
@@ -2297,7 +2403,7 @@ rocksdbjava: $(LIB_OBJECTS)
 ifeq ($(JAVA_HOME),)
 	$(error JAVA_HOME is not set)
 endif
-	$(AM_V_GEN)cd java;$(MAKE) javalib;
+	$(AM_V_GEN)cd java; SHA256_CMD='$(SHA256_CMD)' $(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) $(LIB_OBJECTS) $(JAVA_LDFLAGS) $(COVERAGEFLAGS)
 	$(AM_V_at)cd java;jar -cf target/$(ROCKSDB_JAR) HISTORY*.md
@@ -2309,13 +2415,13 @@ jclean:
 	cd java;$(MAKE) clean;
 
 jtest_compile: rocksdbjava
-	cd java;$(MAKE) java_test
+	cd java; SHA256_CMD='$(SHA256_CMD)' $(MAKE) java_test
 
 jtest_run:
 	cd java;$(MAKE) run_test
 
 jtest: rocksdbjava
-	cd java;$(MAKE) sample;$(MAKE) test;
+	cd java;$(MAKE) sample; SHA256_CMD='$(SHA256_CMD)' $(MAKE) test;
 	$(PYTHON) tools/check_all_python.py # TODO peterd: find a better place for this check in CI targets
 
 jdb_bench:
@@ -2425,6 +2531,8 @@ build_subset_tests: $(ROCKSDBTESTS_SUBSET)
 # working solution.
 ifneq ($(MAKECMDGOALS),clean)
 ifneq ($(MAKECMDGOALS),format)
+ifneq ($(MAKECMDGOALS),check-format)
+ifneq ($(MAKECMDGOALS),check-buck-targets)
 ifneq ($(MAKECMDGOALS),jclean)
 ifneq ($(MAKECMDGOALS),jtest)
 ifneq ($(MAKECMDGOALS),rocksdbjavastatic)
@@ -2440,3 +2548,5 @@ endif
 endif
 endif
 endif
+endif
+endif

PLUGINS.md

@@ -0,0 +1,3 @@
+This is the list of all known third-party plugins for RocksDB. If something is missing, please open a pull request to add it.
+
+* [Dedupfs](https://github.com/ajkr/dedupfs): an example for plugin developers to reference

TARGETS

@@ -1,4 +1,5 @@
-# This file @generated by `python3 buckifier/buckify_rocksdb.py`
+# This file @generated by:
+#$ python3 buckifier/buckify_rocksdb.py
 # --> DO NOT EDIT MANUALLY <--
 # This file is a Facebook-specific integration for buck builds, so can
 # only be validated by Facebook employees.
@@ -24,7 +25,7 @@ ROCKSDB_EXTERNAL_DEPS = [
     ("zlib", None, "z"),
     ("gflags", None, "gflags"),
     ("lz4", None, "lz4"),
-    ("zstd", None),
+    ("zstd", None, "zstd"),
 ]
 
 ROCKSDB_OS_DEPS = [
@@ -388,6 +389,20 @@ cpp_library(
         "utilities/transactions/lock/lock_manager.cc",
         "utilities/transactions/lock/point/point_lock_manager.cc",
         "utilities/transactions/lock/point/point_lock_tracker.cc",
+        "utilities/transactions/lock/range/range_tree/lib/locktree/concurrent_tree.cc",
+        "utilities/transactions/lock/range/range_tree/lib/locktree/keyrange.cc",
+        "utilities/transactions/lock/range/range_tree/lib/locktree/lock_request.cc",
+        "utilities/transactions/lock/range/range_tree/lib/locktree/locktree.cc",
+        "utilities/transactions/lock/range/range_tree/lib/locktree/manager.cc",
+        "utilities/transactions/lock/range/range_tree/lib/locktree/range_buffer.cc",
+        "utilities/transactions/lock/range/range_tree/lib/locktree/treenode.cc",
+        "utilities/transactions/lock/range/range_tree/lib/locktree/txnid_set.cc",
+        "utilities/transactions/lock/range/range_tree/lib/locktree/wfg.cc",
+        "utilities/transactions/lock/range/range_tree/lib/standalone_port.cc",
+        "utilities/transactions/lock/range/range_tree/lib/util/dbt.cc",
+        "utilities/transactions/lock/range/range_tree/lib/util/memarena.cc",
+        "utilities/transactions/lock/range/range_tree/range_tree_lock_manager.cc",
+        "utilities/transactions/lock/range/range_tree/range_tree_lock_tracker.cc",
         "utilities/transactions/optimistic_transaction.cc",
         "utilities/transactions/optimistic_transaction_db_impl.cc",
         "utilities/transactions/pessimistic_transaction.cc",
@@ -678,6 +693,20 @@ cpp_library(
         "utilities/transactions/lock/lock_manager.cc",
         "utilities/transactions/lock/point/point_lock_manager.cc",
         "utilities/transactions/lock/point/point_lock_tracker.cc",
+        "utilities/transactions/lock/range/range_tree/lib/locktree/concurrent_tree.cc",
+        "utilities/transactions/lock/range/range_tree/lib/locktree/keyrange.cc",
+        "utilities/transactions/lock/range/range_tree/lib/locktree/lock_request.cc",
+        "utilities/transactions/lock/range/range_tree/lib/locktree/locktree.cc",
+        "utilities/transactions/lock/range/range_tree/lib/locktree/manager.cc",
+        "utilities/transactions/lock/range/range_tree/lib/locktree/range_buffer.cc",
+        "utilities/transactions/lock/range/range_tree/lib/locktree/treenode.cc",
+        "utilities/transactions/lock/range/range_tree/lib/locktree/txnid_set.cc",
+        "utilities/transactions/lock/range/range_tree/lib/locktree/wfg.cc",
+        "utilities/transactions/lock/range/range_tree/lib/standalone_port.cc",
+        "utilities/transactions/lock/range/range_tree/lib/util/dbt.cc",
+        "utilities/transactions/lock/range/range_tree/lib/util/memarena.cc",
+        "utilities/transactions/lock/range/range_tree/range_tree_lock_manager.cc",
+        "utilities/transactions/lock/range/range_tree/range_tree_lock_tracker.cc",
         "utilities/transactions/optimistic_transaction.cc",
         "utilities/transactions/optimistic_transaction_db_impl.cc",
         "utilities/transactions/pessimistic_transaction.cc",
@@ -786,7 +815,7 @@ cpp_binary(
 ) if not is_opt_mode else None
 
 custom_unittest(
-    "c_test",
+    name = "c_test",
     command = [
         native.package_name() + "/buckifier/rocks_test_runner.sh",
         "$(location :{})".format("c_test_bin"),
@@ -1209,6 +1238,13 @@ ROCKS_TESTS = [
         [],
         [],
     ],
+    [
+        "db_kv_checksum_test",
+        "db/db_kv_checksum_test.cc",
+        "serial",
+        [],
+        [],
+    ],
     [
         "db_log_iter_test",
         "db/db_log_iter_test.cc",
@@ -1776,6 +1812,13 @@ ROCKS_TESTS = [
         [],
         [],
     ],
+    [
+        "range_locking_test",
+        "utilities/transactions/lock/range/range_locking_test.cc",
+        "parallel",
+        [],
+        [],
+    ],
     [
         "range_tombstone_fragmenter_test",
         "db/range_tombstone_fragmenter_test.cc",

USERS.md

@@ -26,6 +26,9 @@ Learn more about those use cases in a Tech Talk by Ankit Gupta and Naveen Somasu
 ## Yahoo
 Yahoo is using RocksDB as a storage engine for their biggest distributed data store Sherpa. Learn more about it here: http://yahooeng.tumblr.com/post/120730204806/sherpa-scales-new-heights
 
+## Baidu
+[Apache Doris](http://doris.apache.org/master/en/) is a MPP analytical database engine released by Baidu. It [uses RocksDB](http://doris.apache.org/master/en/administrator-guide/operation/tablet-meta-tool.html) to manage its tablet's metadata.
+
 ## CockroachDB
 CockroachDB is an open-source geo-replicated transactional database. They are using RocksDB as their storage engine. Check out their github: https://github.com/cockroachdb/cockroach
 
@@ -44,7 +47,7 @@ Tango is using RocksDB as a graph storage to store all users' connection data an
 Turn is using RocksDB as a storage layer for their key/value store, serving at peak 2.4MM QPS out of different datacenters.
 Check out our RocksDB Protobuf merge operator at: https://github.com/vladb38/rocksdb_protobuf
 
-## Santanader UK/Cloudera Profession Services
+## Santander UK/Cloudera Profession Services
 Check out their blog post: http://blog.cloudera.com/blog/2015/08/inside-santanders-near-real-time-data-ingest-architecture/
 
 ## Airbnb
@@ -67,7 +70,7 @@ Pinterest's Object Retrieval System uses RocksDB for storage: https://www.youtub
 [VWO's](https://vwo.com/) Smart Code checker and URL helper uses RocksDB to store all the URLs where VWO's Smart Code is installed.
 
 ## quasardb
-[quasardb](https://www.quasardb.net) is a high-performance, distributed, transactional key-value database that integrates well with in-memory analytics engines such as Apache Spark. 
+[quasardb](https://www.quasardb.net) is a high-performance, distributed, transactional key-value database that integrates well with in-memory analytics engines such as Apache Spark.
 quasardb uses a heavily tuned RocksDB as its persistence layer.
 
 ## Netflix
@@ -86,7 +89,7 @@ quasardb uses a heavily tuned RocksDB as its persistence layer.
 [Uber](http://eng.uber.com/cherami/) uses RocksDB as a durable and scalable task queue.
 
 ## 360 Pika
-[360](http://www.360.cn/) [Pika](https://github.com/Qihoo360/pika) is a nosql compatible with redis. With the huge amount of data stored, redis may suffer for a capacity bottleneck, and pika was born for solving it. It has widely been widely used in many company
+[360](http://www.360.cn/) [Pika](https://github.com/Qihoo360/pika) is a nosql compatible with redis. With the huge amount of data stored, redis may suffer for a capacity bottleneck, and pika was born for solving it. It has widely been used in many companies.
 
 ## LzLabs
 LzLabs is using RocksDB as a storage engine in their multi-database distributed framework to store application configuration and user data.
@@ -96,16 +99,19 @@ LzLabs is using RocksDB as a storage engine in their multi-database distributed
 
 ## IOTA Foundation
  [IOTA Foundation](https://www.iota.org/) is using RocksDB in the [IOTA Reference Implementation (IRI)](https://github.com/iotaledger/iri) to store the local state of the Tangle. The Tangle is the first open-source distributed ledger powering the future of the Internet of Things.
- 
+
 ## Avrio Project
  [Avrio Project](http://avrio-project.github.io/avrio.network/) is using RocksDB in [Avrio ](https://github.com/avrio-project/avrio) to store blocks, account balances and data and other blockchain-releated data. Avrio is a multiblockchain decentralized cryptocurrency empowering monetary transactions.
- 
+
 ## Crux
 [Crux](https://github.com/juxt/crux) is a document database that uses RocksDB for local [EAV](https://en.wikipedia.org/wiki/Entity%E2%80%93attribute%E2%80%93value_model) index storage to enable point-in-time bitemporal Datalog queries. The "unbundled" architecture uses Kafka to provide horizontal scalability.
 
 ## Nebula Graph
-
 [Nebula Graph](https://github.com/vesoft-inc/nebula) is a distributed, scalable, lightning-fast, open source graph database capable of hosting super large scale graphs with dozens of billions of vertices (nodes) and trillions of edges, with milliseconds of latency.
 
 ## YugabyteDB
 [YugabyteDB](https://www.yugabyte.com/) is an open source, high performance, distributed SQL database that uses RocksDB as its storage layer. For more information, please see https://github.com/yugabyte/yugabyte-db/.
+
+## ArangoDB
+[ArangoDB](https://www.arangodb.com/) is a native multi-model database with flexible data models for documents, graphs, and key-values, for building high performance applications using a convenient SQL-like query language or JavaScript extensions. It uses RocksDB as its sotrage engine.
+

buckifier/buckify_rocksdb.py

@@ -24,10 +24,10 @@ from util import ColorString
 # (This generates a TARGET file without user-specified dependency for unit
 # tests.)
 # $python3 buckifier/buckify_rocksdb.py \
-#        '{"fake": { \
-#                      "extra_deps": [":test_dep", "//fakes/module:mock1"],  \
-#                      "extra_compiler_flags": ["-DROCKSDB_LITE", "-Os"], \
-#                  } \
+#        '{"fake": {
+#                      "extra_deps": [":test_dep", "//fakes/module:mock1"],
+#                      "extra_compiler_flags": ["-DROCKSDB_LITE", "-Os"]
+#                  }
 #         }'
 # (Generated TARGETS file has test_dep and mock1 as dependencies for RocksDB
 # unit tests, and will use the extra_compiler_flags to compile the unit test
@@ -129,17 +129,27 @@ def generate_targets(repo_path, deps_map):
     if src_mk is None or cc_files is None or parallel_tests is None:
         return False
 
-    TARGETS = TARGETSBuilder("%s/TARGETS" % repo_path)
+    extra_argv = ""
+    if len(sys.argv) >= 2:
+        # Heuristically quote and canonicalize whitespace for inclusion
+        # in how the file was generated.
+        extra_argv = " '{0}'".format(" ".join(sys.argv[1].split()))
+
+    TARGETS = TARGETSBuilder("%s/TARGETS" % repo_path, extra_argv)
 
     # rocksdb_lib
     TARGETS.add_library(
         "rocksdb_lib",
         src_mk["LIB_SOURCES"] +
+        # always add range_tree, it's only excluded on ppc64, which we don't use internally
+        src_mk["RANGE_TREE_SOURCES"] +
         src_mk["TOOL_LIB_SOURCES"])
     # rocksdb_whole_archive_lib
     TARGETS.add_library(
         "rocksdb_whole_archive_lib",
         src_mk["LIB_SOURCES"] +
+        # always add range_tree, it's only excluded on ppc64, which we don't use internally
+        src_mk["RANGE_TREE_SOURCES"] +
         src_mk["TOOL_LIB_SOURCES"],
         deps=None,
         headers=None,

buckifier/targets_builder.py

@@ -25,10 +25,11 @@ def pretty_list(lst, indent=8):
 
 
 class TARGETSBuilder(object):
-    def __init__(self, path):
+    def __init__(self, path, extra_argv):
         self.path = path
         self.targets_file = open(path, 'wb')
-        header = targets_cfg.rocksdb_target_header_template
+        header = targets_cfg.rocksdb_target_header_template.format(
+            extra_argv=extra_argv)
         self.targets_file.write(header.encode("utf-8"))
         self.total_lib = 0
         self.total_bin = 0
@@ -90,7 +91,7 @@ cpp_binary(
 ) if not is_opt_mode else None
 
 custom_unittest(
-    "c_test",
+    name = "c_test",
     command = [
         native.package_name() + "/buckifier/rocks_test_runner.sh",
         "$(location :{})".format("c_test_bin"),

buckifier/targets_cfg.py

@@ -5,7 +5,8 @@ from __future__ import print_function
 from __future__ import unicode_literals
 
 rocksdb_target_header_template = \
-    """# This file \100generated by `python3 buckifier/buckify_rocksdb.py`
+    """# This file \100generated by:
+#$ python3 buckifier/buckify_rocksdb.py{extra_argv}
 # --> DO NOT EDIT MANUALLY <--
 # This file is a Facebook-specific integration for buck builds, so can
 # only be validated by Facebook employees.
@@ -31,7 +32,7 @@ ROCKSDB_EXTERNAL_DEPS = [
     ("zlib", None, "z"),
     ("gflags", None, "gflags"),
     ("lz4", None, "lz4"),
-    ("zstd", None),
+    ("zstd", None, "zstd"),
 ]
 
 ROCKSDB_OS_DEPS = [
@@ -99,11 +100,11 @@ ROCKSDB_PREPROCESSOR_FLAGS = [
     "-I" + REPO_PATH,
 ]
 
-ROCKSDB_ARCH_PREPROCESSOR_FLAGS = {
+ROCKSDB_ARCH_PREPROCESSOR_FLAGS = {{
     "x86_64": [
         "-DHAVE_PCLMUL",
     ],
-}
+}}
 
 build_mode = read_config("fbcode", "build_mode")
 

build_tools/build_detect_platform

@@ -45,8 +45,13 @@ if test -z "$OUTPUT"; then
   exit 1
 fi
 
-# we depend on C++11
-PLATFORM_CXXFLAGS="-std=c++11"
+# we depend on C++11, but should be compatible with newer standards
+if [ "$ROCKSDB_CXX_STANDARD" ]; then
+  PLATFORM_CXXFLAGS="-std=$ROCKSDB_CXX_STANDARD"
+else
+  PLATFORM_CXXFLAGS="-std=c++11"
+fi
+
 # we currently depend on POSIX platform
 COMMON_FLAGS="-DROCKSDB_PLATFORM_POSIX -DROCKSDB_LIB_IO_POSIX"
 
@@ -655,6 +660,28 @@ else
     COMMON_FLAGS="$COMMON_FLAGS -mmacosx-version-min=10.12"
     PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -mmacosx-version-min=10.12"
     PLATFORM_SHARED_LDFLAGS="$PLATFORM_SHARED_LDFLAGS -mmacosx-version-min=10.12"
+    PLATFORM_CMAKE_FLAGS="-DCMAKE_OSX_DEPLOYMENT_TARGET=10.12"
+    JAVA_STATIC_DEPS_COMMON_FLAGS="-mmacosx-version-min=10.12"
+    JAVA_STATIC_DEPS_LDFLAGS="$JAVA_STATIC_DEPS_COMMON_FLAGS"
+    JAVA_STATIC_DEPS_CCFLAGS="$JAVA_STATIC_DEPS_COMMON_FLAGS"
+    JAVA_STATIC_DEPS_CXXFLAGS="$JAVA_STATIC_DEPS_COMMON_FLAGS"
+  fi
+fi
+
+if test -n "`echo $TARGET_ARCHITECTURE | grep ^ppc64`"; then
+  # check for GNU libc on ppc64
+  $CXX -x c++ - -o /dev/null 2>/dev/null <<EOF
+    #include <stdio.h>
+    #include <stdlib.h>
+    #include <gnu/libc-version.h>
+
+    int main(int argc, char *argv[]) {
+      printf("GNU libc version: %s\n", gnu_get_libc_version());
+      return 0;
+    }
+EOF
+  if [ "$?" != 0 ]; then
+      PPC_LIBC_IS_GNU=0
   fi
 fi
 
@@ -812,8 +839,12 @@ echo "CXX=$CXX" >> "$OUTPUT"
 echo "AR=$AR" >> "$OUTPUT"
 echo "PLATFORM=$PLATFORM" >> "$OUTPUT"
 echo "PLATFORM_LDFLAGS=$PLATFORM_LDFLAGS" >> "$OUTPUT"
+echo "PLATFORM_CMAKE_FLAGS=$PLATFORM_CMAKE_FLAGS" >> "$OUTPUT"
 echo "JAVA_LDFLAGS=$JAVA_LDFLAGS" >> "$OUTPUT"
 echo "JAVA_STATIC_LDFLAGS=$JAVA_STATIC_LDFLAGS" >> "$OUTPUT"
+echo "JAVA_STATIC_DEPS_CCFLAGS=$JAVA_STATIC_DEPS_CCFLAGS" >> "$OUTPUT"
+echo "JAVA_STATIC_DEPS_CXXFLAGS=$JAVA_STATIC_DEPS_CXXFLAGS" >> "$OUTPUT"
+echo "JAVA_STATIC_DEPS_LDFLAGS=$JAVA_STATIC_DEPS_LDFLAGS" >> "$OUTPUT"
 echo "JAVAC_ARGS=$JAVAC_ARGS" >> "$OUTPUT"
 echo "VALGRIND_VER=$VALGRIND_VER" >> "$OUTPUT"
 echo "PLATFORM_CCFLAGS=$PLATFORM_CCFLAGS" >> "$OUTPUT"
@@ -847,3 +878,6 @@ echo "LUA_PATH=$LUA_PATH" >> "$OUTPUT"
 if test -n "$USE_FOLLY_DISTRIBUTED_MUTEX"; then
   echo "USE_FOLLY_DISTRIBUTED_MUTEX=$USE_FOLLY_DISTRIBUTED_MUTEX" >> "$OUTPUT"
 fi
+if test -n "$PPC_LIBC_IS_GNU"; then
+  echo "PPC_LIBC_IS_GNU=$PPC_LIBC_IS_GNU" >> "$OUTPUT"
+fi

build_tools/format-diff.sh

@@ -52,7 +52,8 @@ else
     else
       echo "You didn't have clang-format-diff.py and/or clang-format available in your computer!"
       echo "You can download clang-format-diff.py by running: "
-      echo "    curl --location https://tinyurl.com/y2kvokof -o ${REPO_ROOT}/clang-format-diff.py"
+      echo "    curl --location https://raw.githubusercontent.com/llvm/llvm-project/main/clang/tools/clang-format/clang-format-diff.py -o ${REPO_ROOT}/clang-format-diff.py"
+      echo "You should make sure the downloaded script is not compromised."
       echo "You can download clang-format by running:"
       echo "    brew install clang-format"
       echo "  Or"
@@ -82,7 +83,8 @@ else
       echo "You have clang-format-diff.py for Python 2 but are using a Python 3"
       echo "interpreter (${PYTHON:-python3})."
       echo "You can download clang-format-diff.py for Python 3 by running: "
-      echo "    curl --location https://tinyurl.com/y2kvokof -o ${REPO_ROOT}/clang-format-diff.py"
+      echo "    curl --location https://raw.githubusercontent.com/llvm/llvm-project/main/clang/tools/clang-format/clang-format-diff.py -o ${REPO_ROOT}/clang-format-diff.py"
+      echo "You should make sure the downloaded script is not compromised."
       exit 130
     fi
     CLANG_FORMAT_DIFF="${PYTHON:-python3} $CFD_PATH"

build_tools/rocksdb-lego-determinator

@@ -3,7 +3,7 @@
 # to determine next steps to run
 
 # Usage:
-# EMAIL=<email> ONCALL=<email> TRIGGER=<trigger> SUBSCRIBER=<email> rocks_ci.py <test-name>
+# EMAIL=<email> ONCALL=<email> TRIGGER=<trigger> SUBSCRIBER=<email> WORKINGDIR=<working_dir> rocksdb-lego-determinator <test-name>
 #
 # Input         Value
 # -------------------------------------------------------------------------
@@ -11,7 +11,7 @@
 # ONCALL        Email address to raise a task on failure
 # TRIGGER       Trigger conditions for email. Valid values are fail, warn, all
 # SUBSCRIBER    Email addresss to add as subscriber for task
-#
+# WORKINGDIR    Working directory
 
 #
 # Report configuration
@@ -53,13 +53,19 @@ if [[ ! -z $REPORT_EMAIL || ! -z $CREATE_TASK ]]; then
   ]"
 fi
 
+# Working directory for the following command, default to current directory
+WORKING_DIR=.
+if [ ! -z $WORKINGDIR ]; then
+  WORKING_DIR=$WORKINGDIR
+fi
+
 #
 # Helper variables
 #
 CLEANUP_ENV="
 {
     'name':'Cleanup environment',
-    'shell':'rm -rf /dev/shm/rocksdb && mkdir /dev/shm/rocksdb && (chmod +t /dev/shm || true)  && make clean',
+    'shell':'cd $WORKING_DIR; rm -rf /dev/shm/rocksdb && mkdir /dev/shm/rocksdb && (chmod +t /dev/shm || true) && make clean',
     'user':'root'
 }"
 
@@ -112,6 +118,7 @@ SETUP_JAVA_ENV="export $HTTP_PROXY; export JAVA_HOME=/usr/local/jdk-8u60-64/; ex
 PARSER="'parser':'python build_tools/error_filter.py $1'"
 
 CONTRUN_NAME="ROCKSDB_CONTRUN_NAME"
+SKIP_FORMAT_CHECKS="SKIP_FORMAT_BUCK_CHECKS=1"
 
 # This code is getting called under various scenarios. What we care about is to
 # understand when it's called from nightly contruns because in that case we'll
@@ -153,7 +160,7 @@ UNIT_TEST_COMMANDS="[
             $CLEANUP_ENV,
             {
                 'name':'Build and test RocksDB debug version',
-                'shell':'$SHM $DEBUG make $PARALLELISM check || $CONTRUN_NAME=check $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; $SHM $DEBUG $SKIP_FORMAT_CHECKS make $PARALLELISM check || $CONTRUN_NAME=check $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -176,7 +183,7 @@ UNIT_TEST_NON_SHM_COMMANDS="[
             {
                 'name':'Build and test RocksDB debug version',
                 'timeout': 86400,
-                'shell':'$NON_SHM $DEBUG make $PARALLELISM check || $CONTRUN_NAME=non_shm_check $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; $NON_SHM $DEBUG $SKIP_FORMAT_CHECKS make $PARALLELISM check || $CONTRUN_NAME=non_shm_check $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -197,7 +204,7 @@ RELEASE_BUILD_COMMANDS="[
             $CLEANUP_ENV,
             {
                 'name':'Build RocksDB release',
-                'shell':'make $PARALLEL_j release || $CONTRUN_NAME=release $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; make $PARALLEL_j release || $CONTRUN_NAME=release $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -218,7 +225,7 @@ UNIT_TEST_COMMANDS_481="[
             $CLEANUP_ENV,
             {
                 'name':'Build and test RocksDB debug version',
-                'shell':'$SHM $GCC_481 $DEBUG make $PARALLELISM check || $CONTRUN_NAME=unit_gcc_481_check $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; $SHM $GCC_481 $DEBUG $SKIP_FORMAT_CHECKS make $PARALLELISM check || $CONTRUN_NAME=unit_gcc_481_check $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -239,7 +246,7 @@ RELEASE_BUILD_COMMANDS_481="[
             $CLEANUP_ENV,
             {
                 'name':'Build RocksDB release on GCC 4.8.1',
-                'shell':'$GCC_481 make $PARALLEL_j release || $CONTRUN_NAME=release_gcc481 $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; $GCC_481 make $PARALLEL_j release || $CONTRUN_NAME=release_gcc481 $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -260,7 +267,7 @@ CLANG_UNIT_TEST_COMMANDS="[
             $CLEANUP_ENV,
             {
                 'name':'Build and test RocksDB debug',
-                'shell':'$CLANG $SHM $DEBUG make $PARALLELISM check || $CONTRUN_NAME=clang_check $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; $CLANG $SHM $DEBUG $SKIP_FORMAT_CHECKS make $PARALLELISM check || $CONTRUN_NAME=clang_check $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -281,7 +288,7 @@ CLANG_RELEASE_BUILD_COMMANDS="[
             $CLEANUP_ENV,
             {
                 'name':'Build RocksDB release',
-                'shell':'$CLANG make $PARALLEL_j release|| $CONTRUN_NAME=clang_release $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; $CLANG make $PARALLEL_j release|| $CONTRUN_NAME=clang_release $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -302,7 +309,7 @@ CLANG_ANALYZE_COMMANDS="[
             $CLEANUP_ENV,
             {
                 'name':'RocksDB build and analyze',
-                'shell':'$CLANG $SHM $DEBUG make $PARALLEL_j analyze || $CONTRUN_NAME=clang_analyze $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; $CLANG $SHM $DEBUG make $PARALLEL_j analyze || $CONTRUN_NAME=clang_analyze $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -323,7 +330,7 @@ CODE_COV_COMMANDS="[
             $CLEANUP_ENV,
             {
                 'name':'Build, test and collect code coverage info',
-                'shell':'$SHM $DEBUG make $PARALLELISM coverage || $CONTRUN_NAME=coverage $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; $SHM $DEBUG $SKIP_FORMAT_CHECKS make $PARALLELISM coverage || $CONTRUN_NAME=coverage $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -344,7 +351,7 @@ UNITY_COMMANDS="[
             $CLEANUP_ENV,
             {
                 'name':'Build, test unity test',
-                'shell':'$SHM $DEBUG V=1 make J=1 unity_test || $CONTRUN_NAME=unity_test $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; $SHM $DEBUG V=1 make J=1 unity_test || $CONTRUN_NAME=unity_test $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -365,7 +372,7 @@ LITE_BUILD_COMMANDS="[
             $CLEANUP_ENV,
             {
                 'name':'Build RocksDB debug version',
-                'shell':'make J=1 LITE=1 all check || $CONTRUN_NAME=lite $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; $SKIP_FORMAT_CHECKS make J=1 LITE=1 all check || $CONTRUN_NAME=lite $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -385,7 +392,7 @@ REPORT_LITE_BINARY_SIZE_COMMANDS="[
             $CLEANUP_ENV,
             {
                 'name':'Report RocksDB Lite binary size to scuba',
-                'shell':'tools/report_lite_binary_size.sh',
+                'shell':'cd $WORKING_DIR; tools/report_lite_binary_size.sh',
                 'user':'root',
             },
         ],
@@ -404,14 +411,14 @@ STRESS_CRASH_TEST_COMMANDS="[
             $CLEANUP_ENV,
             {
                 'name':'Build and run RocksDB debug stress tests',
-                'shell':'$SHM $DEBUG $NON_TSAN_CRASH make J=1 db_stress || $CONTRUN_NAME=db_stress $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; $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 $NON_TSAN_CRASH make J=1 crash_test || $CONTRUN_NAME=crash_test $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; $SHM $DEBUG $NON_TSAN_CRASH make J=1 crash_test || $CONTRUN_NAME=crash_test $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -434,14 +441,14 @@ BLACKBOX_STRESS_CRASH_TEST_COMMANDS="[
             $CLEANUP_ENV,
             {
                 'name':'Build and run RocksDB debug stress tests',
-                'shell':'$SHM $DEBUG $NON_TSAN_CRASH make J=1 db_stress || $CONTRUN_NAME=db_stress $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; $SHM $DEBUG $NON_TSAN_CRASH make J=1 db_stress || $CONTRUN_NAME=db_stress $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
             {
                 'name':'Build and run RocksDB debug blackbox crash tests',
                 'timeout': 86400,
-                'shell':'$SHM $DEBUG $NON_TSAN_CRASH make J=1 blackbox_crash_test || $CONTRUN_NAME=blackbox_crash_test $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; $SHM $DEBUG $NON_TSAN_CRASH make J=1 blackbox_crash_test || $CONTRUN_NAME=blackbox_crash_test $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -464,14 +471,14 @@ WHITEBOX_STRESS_CRASH_TEST_COMMANDS="[
             $CLEANUP_ENV,
             {
                 'name':'Build and run RocksDB debug stress tests',
-                'shell':'$SHM $DEBUG $NON_TSAN_CRASH make J=1 db_stress || $CONTRUN_NAME=db_stress $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; $SHM $DEBUG $NON_TSAN_CRASH make J=1 db_stress || $CONTRUN_NAME=db_stress $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
             {
                 'name':'Build and run RocksDB debug whitebox crash tests',
                 'timeout': 86400,
-                'shell':'$SHM $DEBUG $NON_TSAN_CRASH make J=1 whitebox_crash_test || $CONTRUN_NAME=whitebox_crash_test $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; $SHM $DEBUG $NON_TSAN_CRASH make J=1 whitebox_crash_test || $CONTRUN_NAME=whitebox_crash_test $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -494,14 +501,14 @@ STRESS_CRASH_TEST_WITH_ATOMIC_FLUSH_COMMANDS="[
             $CLEANUP_ENV,
             {
                 'name':'Build and run RocksDB debug stress tests',
-                'shell':'$SHM $DEBUG $NON_TSAN_CRASH make J=1 db_stress || $CONTRUN_NAME=db_stress $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; $SHM $DEBUG $NON_TSAN_CRASH make J=1 db_stress || $CONTRUN_NAME=db_stress $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
             {
                 'name':'Build and run RocksDB debug crash tests with atomic flush',
                 'timeout': 86400,
-                'shell':'$SHM $DEBUG $NON_TSAN_CRASH make J=1 crash_test_with_atomic_flush || $CONTRUN_NAME=crash_test_with_atomic_flush $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; $SHM $DEBUG $NON_TSAN_CRASH make J=1 crash_test_with_atomic_flush || $CONTRUN_NAME=crash_test_with_atomic_flush $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -524,14 +531,14 @@ STRESS_CRASH_TEST_WITH_TXN_COMMANDS="[
             $CLEANUP_ENV,
             {
                 'name':'Build and run RocksDB debug stress tests',
-                'shell':'$SHM $DEBUG $NON_TSAN_CRASH make J=1 db_stress || $CONTRUN_NAME=db_stress $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; $SHM $DEBUG $NON_TSAN_CRASH make J=1 db_stress || $CONTRUN_NAME=db_stress $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
             {
                 'name':'Build and run RocksDB debug crash tests with txn',
                 'timeout': 86400,
-                'shell':'$SHM $DEBUG $NON_TSAN_CRASH make J=1 crash_test_with_txn || $CONTRUN_NAME=crash_test_with_txn $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; $SHM $DEBUG $NON_TSAN_CRASH make J=1 crash_test_with_txn || $CONTRUN_NAME=crash_test_with_txn $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -553,7 +560,7 @@ WRITE_STRESS_COMMANDS="[
             $CLEANUP_ENV,
             {
                 'name':'Build and run RocksDB write stress tests',
-                'shell':'make write_stress && python tools/write_stress_runner.py --runtime_sec=3600 --db=/tmp/rocksdb_write_stress || $CONTRUN_NAME=write_stress $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; make write_stress && python tools/write_stress_runner.py --runtime_sec=3600 --db=/tmp/rocksdb_write_stress || $CONTRUN_NAME=write_stress $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             }
@@ -576,7 +583,7 @@ ASAN_TEST_COMMANDS="[
             $CLEANUP_ENV,
             {
                 'name':'Test RocksDB debug under ASAN',
-'shell':'set -o pipefail && ($SHM $ASAN $DEBUG make $PARALLELISM asan_check || $CONTRUN_NAME=asan_check $TASK_CREATION_TOOL) |& /usr/facebook/ops/scripts/asan_symbolize.py -d',
+'shell':'cd $WORKING_DIR; set -o pipefail && ($SHM $ASAN $DEBUG $SKIP_FORMAT_CHECKS make $PARALLELISM asan_check || $CONTRUN_NAME=asan_check $TASK_CREATION_TOOL) |& /usr/facebook/ops/scripts/asan_symbolize.py -d',
                 'user':'root',
                 $PARSER
             }
@@ -599,7 +606,7 @@ ASAN_CRASH_TEST_COMMANDS="[
             {
                 'name':'Build and run RocksDB debug asan_crash_test',
                 'timeout': 86400,
-                'shell':'$SHM $DEBUG $NON_TSAN_CRASH make J=1 asan_crash_test || $CONTRUN_NAME=asan_crash_test $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; $SHM $DEBUG $NON_TSAN_CRASH $SKIP_FORMAT_CHECKS make J=1 asan_crash_test || $CONTRUN_NAME=asan_crash_test $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -623,7 +630,7 @@ ASAN_BLACKBOX_CRASH_TEST_COMMANDS="[
             {
                 'name':'Build and run RocksDB debug blackbox asan_crash_test',
                 'timeout': 86400,
-                'shell':'$SHM $DEBUG $NON_TSAN_CRASH make J=1 blackbox_asan_crash_test || $CONTRUN_NAME=blackbox_asan_crash_test $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; $SHM $DEBUG $NON_TSAN_CRASH $SKIP_FORMAT_CHECKS make J=1 blackbox_asan_crash_test || $CONTRUN_NAME=blackbox_asan_crash_test $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -647,7 +654,7 @@ ASAN_WHITEBOX_CRASH_TEST_COMMANDS="[
             {
                 'name':'Build and run RocksDB debug whitebox asan_crash_test',
                 'timeout': 86400,
-                'shell':'$SHM $DEBUG $NON_TSAN_CRASH make J=1 whitebox_asan_crash_test || $CONTRUN_NAME=whitebox_asan_crash_test $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; $SHM $DEBUG $NON_TSAN_CRASH $SKIP_FORMAT_CHECKS make J=1 whitebox_asan_crash_test || $CONTRUN_NAME=whitebox_asan_crash_test $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -671,7 +678,7 @@ ASAN_CRASH_TEST_WITH_ATOMIC_FLUSH_COMMANDS="[
             {
                 'name':'Build and run RocksDB debug asan_crash_test_with_atomic_flush',
                 'timeout': 86400,
-                'shell':'$SHM $DEBUG $NON_TSAN_CRASH make J=1 asan_crash_test_with_atomic_flush || $CONTRUN_NAME=asan_crash_test_with_atomic_flush $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; $SHM $DEBUG $NON_TSAN_CRASH $SKIP_FORMAT_CHECKS make J=1 asan_crash_test_with_atomic_flush || $CONTRUN_NAME=asan_crash_test_with_atomic_flush $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -695,7 +702,7 @@ ASAN_CRASH_TEST_WITH_TXN_COMMANDS="[
             {
                 'name':'Build and run RocksDB debug asan_crash_test_with_txn',
                 'timeout': 86400,
-                'shell':'$SHM $DEBUG $NON_TSAN_CRASH make J=1 asan_crash_test_with_txn || $CONTRUN_NAME=asan_crash_test_with_txn $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; $SHM $DEBUG $NON_TSAN_CRASH $SKIP_FORMAT_CHECKS make J=1 asan_crash_test_with_txn || $CONTRUN_NAME=asan_crash_test_with_txn $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -717,7 +724,7 @@ UBSAN_TEST_COMMANDS="[
             $CLEANUP_ENV,
             {
                 'name':'Test RocksDB debug under UBSAN',
-                'shell':'set -o pipefail && $SHM $UBSAN $CLANG $DEBUG make $PARALLELISM ubsan_check || $CONTRUN_NAME=ubsan_check $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; set -o pipefail && $SHM $UBSAN $CLANG $DEBUG $SKIP_FORMAT_CHECKS make $PARALLELISM ubsan_check || $CONTRUN_NAME=ubsan_check $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             }
@@ -740,7 +747,7 @@ UBSAN_CRASH_TEST_COMMANDS="[
             {
                 'name':'Build and run RocksDB debug ubsan_crash_test',
                 'timeout': 86400,
-                'shell':'$SHM $DEBUG $NON_TSAN_CRASH $CLANG make J=1 ubsan_crash_test || $CONTRUN_NAME=ubsan_crash_test $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; $SHM $DEBUG $NON_TSAN_CRASH $CLANG $SKIP_FORMAT_CHECKS make J=1 ubsan_crash_test || $CONTRUN_NAME=ubsan_crash_test $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -764,7 +771,7 @@ UBSAN_BLACKBOX_CRASH_TEST_COMMANDS="[
             {
                 'name':'Build and run RocksDB debug blackbox ubsan_crash_test',
                 'timeout': 86400,
-                'shell':'$SHM $DEBUG $NON_TSAN_CRASH $CLANG make J=1 blackbox_ubsan_crash_test || $CONTRUN_NAME=blackbox_ubsan_crash_test $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; $SHM $DEBUG $NON_TSAN_CRASH $CLANG $SKIP_FORMAT_CHECKS make J=1 blackbox_ubsan_crash_test || $CONTRUN_NAME=blackbox_ubsan_crash_test $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -788,7 +795,7 @@ UBSAN_WHITEBOX_CRASH_TEST_COMMANDS="[
             {
                 'name':'Build and run RocksDB debug whitebox ubsan_crash_test',
                 'timeout': 86400,
-                'shell':'$SHM $DEBUG $NON_TSAN_CRASH $CLANG make J=1 whitebox_ubsan_crash_test || $CONTRUN_NAME=whitebox_ubsan_crash_test $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; $SHM $DEBUG $NON_TSAN_CRASH $CLANG $SKIP_FORMAT_CHECKS make J=1 whitebox_ubsan_crash_test || $CONTRUN_NAME=whitebox_ubsan_crash_test $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -812,7 +819,7 @@ UBSAN_CRASH_TEST_WITH_ATOMIC_FLUSH_COMMANDS="[
             {
                 'name':'Build and run RocksDB debug ubsan_crash_test_with_atomic_flush',
                 'timeout': 86400,
-                'shell':'$SHM $DEBUG $NON_TSAN_CRASH $CLANG make J=1 ubsan_crash_test_with_atomic_flush || $CONTRUN_NAME=ubsan_crash_test_with_atomic_flush $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; $SHM $DEBUG $NON_TSAN_CRASH $CLANG $SKIP_FORMAT_CHECKS make J=1 ubsan_crash_test_with_atomic_flush || $CONTRUN_NAME=ubsan_crash_test_with_atomic_flush $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -836,7 +843,7 @@ UBSAN_CRASH_TEST_WITH_TXN_COMMANDS="[
             {
                 'name':'Build and run RocksDB debug ubsan_crash_test_with_txn',
                 'timeout': 86400,
-                'shell':'$SHM $DEBUG $NON_TSAN_CRASH $CLANG make J=1 ubsan_crash_test_with_txn || $CONTRUN_NAME=ubsan_crash_test_with_txn $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; $SHM $DEBUG $NON_TSAN_CRASH $CLANG $SKIP_FORMAT_CHECKS make J=1 ubsan_crash_test_with_txn || $CONTRUN_NAME=ubsan_crash_test_with_txn $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -860,7 +867,7 @@ VALGRIND_TEST_COMMANDS="[
             {
                 'name':'Run RocksDB debug unit tests',
                 'timeout': 86400,
-                'shell':'$SHM $DEBUG make $PARALLELISM valgrind_test || $CONTRUN_NAME=valgrind_check $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; $SHM $DEBUG make $PARALLELISM valgrind_test || $CONTRUN_NAME=valgrind_check $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -883,7 +890,7 @@ TSAN_UNIT_TEST_COMMANDS="[
             {
                 'name':'Run RocksDB debug unit test',
                 'timeout': 86400,
-                'shell':'set -o pipefail && $SHM $DEBUG $TSAN make $PARALLELISM check || $CONTRUN_NAME=tsan_check $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; set -o pipefail && $SHM $DEBUG $TSAN $SKIP_FORMAT_CHECKS make $PARALLELISM check || $CONTRUN_NAME=tsan_check $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -906,7 +913,7 @@ TSAN_CRASH_TEST_COMMANDS="[
             {
                 'name':'Compile and run',
                 'timeout': 86400,
-                'shell':'set -o pipefail && $SHM $DEBUG $TSAN $TSAN_CRASH CRASH_TEST_KILL_ODD=1887 make J=1 crash_test || $CONTRUN_NAME=tsan_crash_test $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; 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
             },
@@ -930,7 +937,7 @@ TSAN_BLACKBOX_CRASH_TEST_COMMANDS="[
             {
                 'name':'Compile and run',
                 'timeout': 86400,
-                'shell':'set -o pipefail && $SHM $DEBUG $TSAN $TSAN_CRASH CRASH_TEST_KILL_ODD=1887 make J=1 blackbox_crash_test || $CONTRUN_NAME=tsan_blackbox_crash_test $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; set -o pipefail && $SHM $DEBUG $TSAN $TSAN_CRASH CRASH_TEST_KILL_ODD=1887 make J=1 blackbox_crash_test || $CONTRUN_NAME=tsan_blackbox_crash_test $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -954,7 +961,7 @@ TSAN_WHITEBOX_CRASH_TEST_COMMANDS="[
             {
                 'name':'Compile and run',
                 'timeout': 86400,
-                'shell':'set -o pipefail && $SHM $DEBUG $TSAN $TSAN_CRASH CRASH_TEST_KILL_ODD=1887 make J=1 whitebox_crash_test || $CONTRUN_NAME=tsan_whitebox_crash_test $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; set -o pipefail && $SHM $DEBUG $TSAN $TSAN_CRASH CRASH_TEST_KILL_ODD=1887 make J=1 whitebox_crash_test || $CONTRUN_NAME=tsan_whitebox_crash_test $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -978,7 +985,7 @@ TSAN_CRASH_TEST_WITH_ATOMIC_FLUSH_COMMANDS="[
             {
                 'name':'Compile and run',
                 'timeout': 86400,
-                'shell':'set -o pipefail && $SHM $DEBUG $TSAN $TSAN_CRASH CRASH_TEST_KILL_ODD=1887 make J=1 crash_test_with_atomic_flush || $CONTRUN_NAME=tsan_crash_test_with_atomic_flush $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; set -o pipefail && $SHM $DEBUG $TSAN $TSAN_CRASH CRASH_TEST_KILL_ODD=1887 make J=1 crash_test_with_atomic_flush || $CONTRUN_NAME=tsan_crash_test_with_atomic_flush $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -1002,7 +1009,7 @@ TSAN_CRASH_TEST_WITH_TXN_COMMANDS="[
             {
                 'name':'Compile and run',
                 'timeout': 86400,
-                'shell':'set -o pipefail && $SHM $DEBUG $TSAN $TSAN_CRASH CRASH_TEST_KILL_ODD=1887 make J=1 crash_test_with_txn || $CONTRUN_NAME=tsan_crash_test_with_txn $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; set -o pipefail && $SHM $DEBUG $TSAN $TSAN_CRASH CRASH_TEST_KILL_ODD=1887 make J=1 crash_test_with_txn || $CONTRUN_NAME=tsan_crash_test_with_txn $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -1036,7 +1043,7 @@ FORMAT_COMPATIBLE_COMMANDS="[
             $CLEANUP_ENV,
             {
                 'name':'Run RocksDB debug unit test',
-                'shell':'build_tools/rocksdb-lego-determinator run_format_compatible || $CONTRUN_NAME=run_format_compatible $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; build_tools/rocksdb-lego-determinator run_format_compatible || $CONTRUN_NAME=run_format_compatible $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -1058,6 +1065,7 @@ run_no_compression()
   mv .tmp.fbcode_config.sh build_tools/fbcode_config.sh
   cat Makefile | grep -v tools/ldb_test.py > .tmp.Makefile
   mv .tmp.Makefile Makefile
+  export $SKIP_FORMAT_CHECKS
   make $DEBUG J=1 check
 }
 
@@ -1070,7 +1078,7 @@ NO_COMPRESSION_COMMANDS="[
             $CLEANUP_ENV,
             {
                 'name':'Run RocksDB debug unit test',
-                'shell':'build_tools/rocksdb-lego-determinator run_no_compression || $CONTRUN_NAME=run_no_compression $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; build_tools/rocksdb-lego-determinator run_no_compression || $CONTRUN_NAME=run_no_compression $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -1129,7 +1137,7 @@ REGRESSION_COMMANDS="[
             $CLEANUP_ENV,
             {
                 'name':'Make and run script',
-                'shell':'build_tools/rocksdb-lego-determinator run_regression || $CONTRUN_NAME=run_regression $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; build_tools/rocksdb-lego-determinator run_regression || $CONTRUN_NAME=run_regression $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -1150,7 +1158,7 @@ JAVA_BUILD_TEST_COMMANDS="[
             $CLEANUP_ENV,
             {
                 'name':'Build RocksDB for Java',
-                'shell':'$SETUP_JAVA_ENV; $SHM make rocksdbjava || $CONTRUN_NAME=rocksdbjava $TASK_CREATION_TOOL',
+                'shell':'cd $WORKING_DIR; $SETUP_JAVA_ENV; $SHM make rocksdbjava || $CONTRUN_NAME=rocksdbjava $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },

cache/lru_cache.h

@@ -67,7 +67,7 @@ struct LRUHandle {
     IS_HIGH_PRI = (1 << 1),
     // Whether this entry is in high-pri pool.
     IN_HIGH_PRI_POOL = (1 << 2),
-    // Wwhether this entry has had any lookups (hits).
+    // Whether this entry has had any lookups (hits).
     HAS_HIT = (1 << 3),
   };
 
@@ -130,7 +130,7 @@ struct LRUHandle {
     delete[] reinterpret_cast<char*>(this);
   }
 
-  // Caclculate the memory usage by metadata
+  // Calculate the memory usage by metadata
   inline size_t CalcTotalCharge(
       CacheMetadataChargePolicy metadata_charge_policy) {
     size_t meta_charge = 0;

db/arena_wrapped_db_iter.cc

@@ -30,20 +30,19 @@ Status ArenaWrappedDBIter::GetProperty(std::string prop_name,
   return db_iter_->GetProperty(prop_name, prop);
 }
 
-void ArenaWrappedDBIter::Init(Env* env, const ReadOptions& read_options,
-                              const ImmutableCFOptions& cf_options,
-                              const MutableCFOptions& mutable_cf_options,
-                              const SequenceNumber& sequence,
-                              uint64_t max_sequential_skip_in_iteration,
-                              uint64_t version_number,
-                              ReadCallback* read_callback, DBImpl* db_impl,
-                              ColumnFamilyData* cfd, bool allow_blob,
-                              bool allow_refresh) {
+void ArenaWrappedDBIter::Init(
+    Env* env, const ReadOptions& read_options,
+    const ImmutableCFOptions& cf_options,
+    const MutableCFOptions& mutable_cf_options, const Version* version,
+    const SequenceNumber& sequence, uint64_t max_sequential_skip_in_iteration,
+    uint64_t version_number, ReadCallback* read_callback, DBImpl* db_impl,
+    ColumnFamilyData* cfd, bool expose_blob_index, bool allow_refresh) {
   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, db_impl, cfd, allow_blob);
+  db_iter_ =
+      new (mem) DBIter(env, read_options, cf_options, mutable_cf_options,
+                       cf_options.user_comparator, /* iter */ nullptr, version,
+                       sequence, true, max_sequential_skip_in_iteration,
+                       read_callback, db_impl, cfd, expose_blob_index);
   sv_number_ = version_number;
   read_options_ = read_options;
   allow_refresh_ = allow_refresh;
@@ -72,8 +71,9 @@ Status ArenaWrappedDBIter::Refresh() {
       read_callback_->Refresh(latest_seq);
     }
     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_, db_impl_, cfd_, allow_blob_,
+         sv->current, latest_seq,
+         sv->mutable_cf_options.max_sequential_skip_in_iterations,
+         cur_sv_number, read_callback_, db_impl_, cfd_, expose_blob_index_,
          allow_refresh_);
 
     InternalIterator* internal_iter = db_impl_->NewInternalIterator(
@@ -90,16 +90,16 @@ Status ArenaWrappedDBIter::Refresh() {
 ArenaWrappedDBIter* NewArenaWrappedDbIterator(
     Env* env, const ReadOptions& read_options,
     const ImmutableCFOptions& cf_options,
-    const MutableCFOptions& mutable_cf_options, const SequenceNumber& sequence,
-    uint64_t max_sequential_skip_in_iterations, uint64_t version_number,
-    ReadCallback* read_callback, DBImpl* db_impl, ColumnFamilyData* cfd,
-    bool allow_blob, bool allow_refresh) {
+    const MutableCFOptions& mutable_cf_options, const Version* version,
+    const SequenceNumber& sequence, uint64_t max_sequential_skip_in_iterations,
+    uint64_t version_number, ReadCallback* read_callback, DBImpl* db_impl,
+    ColumnFamilyData* cfd, bool expose_blob_index, bool allow_refresh) {
   ArenaWrappedDBIter* iter = new ArenaWrappedDBIter();
-  iter->Init(env, read_options, cf_options, mutable_cf_options, sequence,
-             max_sequential_skip_in_iterations, version_number, read_callback,
-             db_impl, cfd, allow_blob, allow_refresh);
+  iter->Init(env, read_options, cf_options, mutable_cf_options, version,
+             sequence, max_sequential_skip_in_iterations, version_number,
+             read_callback, db_impl, cfd, expose_blob_index, allow_refresh);
   if (db_impl != nullptr && cfd != nullptr && allow_refresh) {
-    iter->StoreRefreshInfo(db_impl, cfd, read_callback, allow_blob);
+    iter->StoreRefreshInfo(db_impl, cfd, read_callback, expose_blob_index);
   }
 
   return iter;

db/arena_wrapped_db_iter.h

@@ -23,6 +23,7 @@
 namespace ROCKSDB_NAMESPACE {
 
 class Arena;
+class Version;
 
 // A wrapper iterator which wraps DB Iterator and the arena, with which the DB
 // iterator is supposed to be allocated. This class is used as an entry point of
@@ -72,20 +73,20 @@ class ArenaWrappedDBIter : public Iterator {
 
   void Init(Env* env, const ReadOptions& read_options,
             const ImmutableCFOptions& cf_options,
-            const MutableCFOptions& mutable_cf_options,
+            const MutableCFOptions& mutable_cf_options, const Version* version,
             const SequenceNumber& sequence,
             uint64_t max_sequential_skip_in_iterations, uint64_t version_number,
             ReadCallback* read_callback, DBImpl* db_impl, ColumnFamilyData* cfd,
-            bool allow_blob, bool allow_refresh);
+            bool expose_blob_index, bool allow_refresh);
 
   // Store some parameters so we can refresh the iterator at a later point
   // with these same params
   void StoreRefreshInfo(DBImpl* db_impl, ColumnFamilyData* cfd,
-                        ReadCallback* read_callback, bool allow_blob) {
+                        ReadCallback* read_callback, bool expose_blob_index) {
     db_impl_ = db_impl;
     cfd_ = cfd;
     read_callback_ = read_callback;
-    allow_blob_ = allow_blob;
+    expose_blob_index_ = expose_blob_index;
   }
 
  private:
@@ -96,7 +97,7 @@ class ArenaWrappedDBIter : public Iterator {
   DBImpl* db_impl_ = nullptr;
   ReadOptions read_options_;
   ReadCallback* read_callback_;
-  bool allow_blob_ = false;
+  bool expose_blob_index_ = false;
   bool allow_refresh_ = true;
 };
 
@@ -106,9 +107,9 @@ class ArenaWrappedDBIter : public Iterator {
 extern ArenaWrappedDBIter* NewArenaWrappedDbIterator(
     Env* env, const ReadOptions& read_options,
     const ImmutableCFOptions& cf_options,
-    const MutableCFOptions& mutable_cf_options, const SequenceNumber& sequence,
-    uint64_t max_sequential_skip_in_iterations, uint64_t version_number,
-    ReadCallback* read_callback, DBImpl* db_impl = nullptr,
-    ColumnFamilyData* cfd = nullptr, bool allow_blob = false,
-    bool allow_refresh = true);
+    const MutableCFOptions& mutable_cf_options, const Version* version,
+    const SequenceNumber& sequence, uint64_t max_sequential_skip_in_iterations,
+    uint64_t version_number, ReadCallback* read_callback,
+    DBImpl* db_impl = nullptr, ColumnFamilyData* cfd = nullptr,
+    bool expose_blob_index = false, bool allow_refresh = true);
 }  // namespace ROCKSDB_NAMESPACE

db/blob/blob_file_builder.cc

@@ -17,9 +17,11 @@
 #include "file/writable_file_writer.h"
 #include "logging/logging.h"
 #include "options/cf_options.h"
+#include "options/options_helper.h"
 #include "rocksdb/slice.h"
 #include "rocksdb/status.h"
 #include "test_util/sync_point.h"
+#include "trace_replay/io_tracer.h"
 #include "util/compression.h"
 
 namespace ROCKSDB_NAMESPACE {
@@ -31,12 +33,13 @@ BlobFileBuilder::BlobFileBuilder(
     int job_id, uint32_t column_family_id,
     const std::string& column_family_name, Env::IOPriority io_priority,
     Env::WriteLifeTimeHint write_hint,
+    const std::shared_ptr<IOTracer>& io_tracer,
     std::vector<std::string>* blob_file_paths,
     std::vector<BlobFileAddition>* blob_file_additions)
     : BlobFileBuilder([versions]() { return versions->NewFileNumber(); }, env,
                       fs, immutable_cf_options, mutable_cf_options,
                       file_options, job_id, column_family_id,
-                      column_family_name, io_priority, write_hint,
+                      column_family_name, io_priority, write_hint, io_tracer,
                       blob_file_paths, blob_file_additions) {}
 
 BlobFileBuilder::BlobFileBuilder(
@@ -46,6 +49,7 @@ BlobFileBuilder::BlobFileBuilder(
     int job_id, uint32_t column_family_id,
     const std::string& column_family_name, Env::IOPriority io_priority,
     Env::WriteLifeTimeHint write_hint,
+    const std::shared_ptr<IOTracer>& io_tracer,
     std::vector<std::string>* blob_file_paths,
     std::vector<BlobFileAddition>* blob_file_additions)
     : file_number_generator_(std::move(file_number_generator)),
@@ -61,6 +65,7 @@ BlobFileBuilder::BlobFileBuilder(
       column_family_name_(column_family_name),
       io_priority_(io_priority),
       write_hint_(write_hint),
+      io_tracer_(io_tracer),
       blob_file_paths_(blob_file_paths),
       blob_file_additions_(blob_file_additions),
       blob_count_(0),
@@ -157,11 +162,12 @@ Status BlobFileBuilder::OpenBlobFileIfNeeded() {
   std::unique_ptr<FSWritableFile> file;
 
   {
-    TEST_SYNC_POINT("BlobFileBuilder::OpenBlobFileIfNeeded:NewWritableFile");
-
     assert(file_options_);
-    const Status s =
-        NewWritableFile(fs_, blob_file_path, &file, *file_options_);
+    Status s = NewWritableFile(fs_, blob_file_path, &file, *file_options_);
+
+    TEST_SYNC_POINT_CALLBACK(
+        "BlobFileBuilder::OpenBlobFileIfNeeded:NewWritableFile", &s);
+
     if (!s.ok()) {
       return s;
     }
@@ -176,17 +182,19 @@ Status BlobFileBuilder::OpenBlobFileIfNeeded() {
   assert(file);
   file->SetIOPriority(io_priority_);
   file->SetWriteLifeTimeHint(write_hint_);
-
+  FileTypeSet tmp_set = immutable_cf_options_->checksum_handoff_file_types;
   Statistics* const statistics = immutable_cf_options_->statistics;
-
   std::unique_ptr<WritableFileWriter> file_writer(new WritableFileWriter(
       std::move(file), blob_file_paths_->back(), *file_options_, env_,
-      nullptr /*IOTracer*/, statistics, immutable_cf_options_->listeners,
-      immutable_cf_options_->file_checksum_gen_factory));
+      io_tracer_, statistics, immutable_cf_options_->listeners,
+      immutable_cf_options_->file_checksum_gen_factory,
+      tmp_set.Contains(FileType::kBlobFile)));
 
-  std::unique_ptr<BlobLogWriter> blob_log_writer(
-      new BlobLogWriter(std::move(file_writer), env_, statistics,
-                        blob_file_number, immutable_cf_options_->use_fsync));
+  constexpr bool do_flush = false;
+
+  std::unique_ptr<BlobLogWriter> blob_log_writer(new BlobLogWriter(
+      std::move(file_writer), env_, statistics, blob_file_number,
+      immutable_cf_options_->use_fsync, do_flush));
 
   constexpr bool has_ttl = false;
   constexpr ExpirationRange expiration_range;
@@ -195,9 +203,11 @@ Status BlobFileBuilder::OpenBlobFileIfNeeded() {
                        expiration_range);
 
   {
-    TEST_SYNC_POINT("BlobFileBuilder::OpenBlobFileIfNeeded:WriteHeader");
+    Status s = blob_log_writer->WriteHeader(header);
+
+    TEST_SYNC_POINT_CALLBACK(
+        "BlobFileBuilder::OpenBlobFileIfNeeded:WriteHeader", &s);
 
-    const Status s = blob_log_writer->WriteHeader(header);
     if (!s.ok()) {
       return s;
     }
@@ -247,9 +257,10 @@ Status BlobFileBuilder::WriteBlobToFile(const Slice& key, const Slice& blob,
 
   uint64_t key_offset = 0;
 
-  TEST_SYNC_POINT("BlobFileBuilder::WriteBlobToFile:AddRecord");
+  Status s = writer_->AddRecord(key, blob, &key_offset, blob_offset);
+
+  TEST_SYNC_POINT_CALLBACK("BlobFileBuilder::WriteBlobToFile:AddRecord", &s);
 
-  const Status s = writer_->AddRecord(key, blob, &key_offset, blob_offset);
   if (!s.ok()) {
     return s;
   }
@@ -271,10 +282,10 @@ Status BlobFileBuilder::CloseBlobFile() {
   std::string checksum_method;
   std::string checksum_value;
 
-  TEST_SYNC_POINT("BlobFileBuilder::WriteBlobToFile:AppendFooter");
+  Status s = writer_->AppendFooter(footer, &checksum_method, &checksum_value);
+
+  TEST_SYNC_POINT_CALLBACK("BlobFileBuilder::WriteBlobToFile:AppendFooter", &s);
 
-  const Status s =
-      writer_->AppendFooter(footer, &checksum_method, &checksum_value);
   if (!s.ok()) {
     return s;
   }

db/blob/blob_file_builder.h

@@ -25,6 +25,7 @@ class BlobFileAddition;
 class Status;
 class Slice;
 class BlobLogWriter;
+class IOTracer;
 
 class BlobFileBuilder {
  public:
@@ -36,6 +37,7 @@ class BlobFileBuilder {
                   const std::string& column_family_name,
                   Env::IOPriority io_priority,
                   Env::WriteLifeTimeHint write_hint,
+                  const std::shared_ptr<IOTracer>& io_tracer,
                   std::vector<std::string>* blob_file_paths,
                   std::vector<BlobFileAddition>* blob_file_additions);
 
@@ -48,6 +50,7 @@ class BlobFileBuilder {
                   const std::string& column_family_name,
                   Env::IOPriority io_priority,
                   Env::WriteLifeTimeHint write_hint,
+                  const std::shared_ptr<IOTracer>& io_tracer,
                   std::vector<std::string>* blob_file_paths,
                   std::vector<BlobFileAddition>* blob_file_additions);
 
@@ -81,6 +84,7 @@ class BlobFileBuilder {
   std::string column_family_name_;
   Env::IOPriority io_priority_;
   Env::WriteLifeTimeHint write_hint_;
+  std::shared_ptr<IOTracer> io_tracer_;
   std::vector<std::string>* blob_file_paths_;
   std::vector<BlobFileAddition>* blob_file_additions_;
   std::unique_ptr<BlobLogWriter> writer_;

db/blob/blob_file_builder_test.cc

@@ -15,7 +15,6 @@
 #include "db/blob/blob_index.h"
 #include "db/blob/blob_log_format.h"
 #include "db/blob/blob_log_sequential_reader.h"
-#include "env/composite_env_wrapper.h"
 #include "env/mock_env.h"
 #include "file/filename.h"
 #include "file/random_access_file_reader.h"
@@ -40,7 +39,8 @@ class TestFileNumberGenerator {
 
 class BlobFileBuilderTest : public testing::Test {
  protected:
-  BlobFileBuilderTest() : mock_env_(Env::Default()), fs_(&mock_env_) {}
+  BlobFileBuilderTest()
+      : mock_env_(Env::Default()), fs_(mock_env_.GetFileSystem().get()) {}
 
   void VerifyBlobFile(uint64_t blob_file_number,
                       const std::string& blob_file_path,
@@ -54,7 +54,7 @@ class BlobFileBuilderTest : public testing::Test {
     std::unique_ptr<FSRandomAccessFile> file;
     constexpr IODebugContext* dbg = nullptr;
     ASSERT_OK(
-        fs_.NewRandomAccessFile(blob_file_path, file_options_, &file, dbg));
+        fs_->NewRandomAccessFile(blob_file_path, file_options_, &file, dbg));
 
     std::unique_ptr<RandomAccessFileReader> file_reader(
         new RandomAccessFileReader(std::move(file), blob_file_path,
@@ -108,7 +108,7 @@ class BlobFileBuilderTest : public testing::Test {
   }
 
   MockEnv mock_env_;
-  LegacyFileSystemWrapper fs_;
+  FileSystem* fs_;
   FileOptions file_options_;
 };
 
@@ -138,11 +138,11 @@ TEST_F(BlobFileBuilderTest, BuildAndCheckOneFile) {
   std::vector<std::string> blob_file_paths;
   std::vector<BlobFileAddition> blob_file_additions;
 
-  BlobFileBuilder builder(TestFileNumberGenerator(), &mock_env_, &fs_,
-                          &immutable_cf_options, &mutable_cf_options,
-                          &file_options_, job_id, column_family_id,
-                          column_family_name, io_priority, write_hint,
-                          &blob_file_paths, &blob_file_additions);
+  BlobFileBuilder builder(
+      TestFileNumberGenerator(), &mock_env_, fs_, &immutable_cf_options,
+      &mutable_cf_options, &file_options_, job_id, column_family_id,
+      column_family_name, io_priority, write_hint, nullptr /*IOTracer*/,
+      &blob_file_paths, &blob_file_additions);
 
   std::vector<std::pair<std::string, std::string>> expected_key_value_pairs(
       number_of_blobs);
@@ -221,11 +221,11 @@ TEST_F(BlobFileBuilderTest, BuildAndCheckMultipleFiles) {
   std::vector<std::string> blob_file_paths;
   std::vector<BlobFileAddition> blob_file_additions;
 
-  BlobFileBuilder builder(TestFileNumberGenerator(), &mock_env_, &fs_,
-                          &immutable_cf_options, &mutable_cf_options,
-                          &file_options_, job_id, column_family_id,
-                          column_family_name, io_priority, write_hint,
-                          &blob_file_paths, &blob_file_additions);
+  BlobFileBuilder builder(
+      TestFileNumberGenerator(), &mock_env_, fs_, &immutable_cf_options,
+      &mutable_cf_options, &file_options_, job_id, column_family_id,
+      column_family_name, io_priority, write_hint, nullptr /*IOTracer*/,
+      &blob_file_paths, &blob_file_additions);
 
   std::vector<std::pair<std::string, std::string>> expected_key_value_pairs(
       number_of_blobs);
@@ -306,11 +306,11 @@ TEST_F(BlobFileBuilderTest, InlinedValues) {
   std::vector<std::string> blob_file_paths;
   std::vector<BlobFileAddition> blob_file_additions;
 
-  BlobFileBuilder builder(TestFileNumberGenerator(), &mock_env_, &fs_,
-                          &immutable_cf_options, &mutable_cf_options,
-                          &file_options_, job_id, column_family_id,
-                          column_family_name, io_priority, write_hint,
-                          &blob_file_paths, &blob_file_additions);
+  BlobFileBuilder builder(
+      TestFileNumberGenerator(), &mock_env_, fs_, &immutable_cf_options,
+      &mutable_cf_options, &file_options_, job_id, column_family_id,
+      column_family_name, io_priority, write_hint, nullptr /*IOTracer*/,
+      &blob_file_paths, &blob_file_additions);
 
   for (size_t i = 0; i < number_of_blobs; ++i) {
     const std::string key = std::to_string(i);
@@ -358,11 +358,11 @@ TEST_F(BlobFileBuilderTest, Compression) {
   std::vector<std::string> blob_file_paths;
   std::vector<BlobFileAddition> blob_file_additions;
 
-  BlobFileBuilder builder(TestFileNumberGenerator(), &mock_env_, &fs_,
-                          &immutable_cf_options, &mutable_cf_options,
-                          &file_options_, job_id, column_family_id,
-                          column_family_name, io_priority, write_hint,
-                          &blob_file_paths, &blob_file_additions);
+  BlobFileBuilder builder(
+      TestFileNumberGenerator(), &mock_env_, fs_, &immutable_cf_options,
+      &mutable_cf_options, &file_options_, job_id, column_family_id,
+      column_family_name, io_priority, write_hint, nullptr /*IOTracer*/,
+      &blob_file_paths, &blob_file_additions);
 
   const std::string key("1");
   const std::string uncompressed_value(value_size, 'x');
@@ -440,11 +440,11 @@ TEST_F(BlobFileBuilderTest, CompressionError) {
   std::vector<std::string> blob_file_paths;
   std::vector<BlobFileAddition> blob_file_additions;
 
-  BlobFileBuilder builder(TestFileNumberGenerator(), &mock_env_, &fs_,
-                          &immutable_cf_options, &mutable_cf_options,
-                          &file_options_, job_id, column_family_id,
-                          column_family_name, io_priority, write_hint,
-                          &blob_file_paths, &blob_file_additions);
+  BlobFileBuilder builder(
+      TestFileNumberGenerator(), &mock_env_, fs_, &immutable_cf_options,
+      &mutable_cf_options, &file_options_, job_id, column_family_id,
+      column_family_name, io_priority, write_hint, nullptr /*IOTracer*/,
+      &blob_file_paths, &blob_file_additions);
 
   SyncPoint::GetInstance()->SetCallBack("CompressData:TamperWithReturnValue",
                                         [](void* arg) {
@@ -517,11 +517,11 @@ TEST_F(BlobFileBuilderTest, Checksum) {
   std::vector<std::string> blob_file_paths;
   std::vector<BlobFileAddition> blob_file_additions;
 
-  BlobFileBuilder builder(TestFileNumberGenerator(), &mock_env_, &fs_,
-                          &immutable_cf_options, &mutable_cf_options,
-                          &file_options_, job_id, column_family_id,
-                          column_family_name, io_priority, write_hint,
-                          &blob_file_paths, &blob_file_additions);
+  BlobFileBuilder builder(
+      TestFileNumberGenerator(), &mock_env_, fs_, &immutable_cf_options,
+      &mutable_cf_options, &file_options_, job_id, column_family_id,
+      column_family_name, io_priority, write_hint, nullptr /*IOTracer*/,
+      &blob_file_paths, &blob_file_additions);
 
   const std::string key("1");
   const std::string value("deadbeef");
@@ -570,13 +570,11 @@ class BlobFileBuilderIOErrorTest
  protected:
   BlobFileBuilderIOErrorTest()
       : mock_env_(Env::Default()),
-        fault_injection_env_(&mock_env_),
-        fs_(&fault_injection_env_),
+        fs_(mock_env_.GetFileSystem().get()),
         sync_point_(GetParam()) {}
 
   MockEnv mock_env_;
-  FaultInjectionTestEnv fault_injection_env_;
-  LegacyFileSystemWrapper fs_;
+  FileSystem* fs_;
   FileOptions file_options_;
   std::string sync_point_;
 };
@@ -597,11 +595,11 @@ TEST_P(BlobFileBuilderIOErrorTest, IOError) {
 
   Options options;
   options.cf_paths.emplace_back(
-      test::PerThreadDBPath(&fault_injection_env_,
-                            "BlobFileBuilderIOErrorTest_IOError"),
+      test::PerThreadDBPath(&mock_env_, "BlobFileBuilderIOErrorTest_IOError"),
       0);
   options.enable_blob_files = true;
   options.blob_file_size = value_size;
+  options.env = &mock_env_;
 
   ImmutableCFOptions immutable_cf_options(options);
   MutableCFOptions mutable_cf_options(options);
@@ -615,15 +613,17 @@ TEST_P(BlobFileBuilderIOErrorTest, IOError) {
   std::vector<std::string> blob_file_paths;
   std::vector<BlobFileAddition> blob_file_additions;
 
-  BlobFileBuilder builder(TestFileNumberGenerator(), &fault_injection_env_,
-                          &fs_, &immutable_cf_options, &mutable_cf_options,
-                          &file_options_, job_id, column_family_id,
-                          column_family_name, io_priority, write_hint,
-                          &blob_file_paths, &blob_file_additions);
+  BlobFileBuilder builder(
+      TestFileNumberGenerator(), &mock_env_, fs_, &immutable_cf_options,
+      &mutable_cf_options, &file_options_, job_id, column_family_id,
+      column_family_name, io_priority, write_hint, nullptr /*IOTracer*/,
+      &blob_file_paths, &blob_file_additions);
 
-  SyncPoint::GetInstance()->SetCallBack(sync_point_, [this](void* /* arg */) {
-    fault_injection_env_.SetFilesystemActive(false,
-                                             Status::IOError(sync_point_));
+  SyncPoint::GetInstance()->SetCallBack(sync_point_, [this](void* arg) {
+    Status* const s = static_cast<Status*>(arg);
+    assert(s);
+
+    (*s) = Status::IOError(sync_point_);
   });
   SyncPoint::GetInstance()->EnableProcessing();
 

db/blob/blob_file_cache.cc

@@ -13,6 +13,7 @@
 #include "rocksdb/cache.h"
 #include "rocksdb/slice.h"
 #include "test_util/sync_point.h"
+#include "trace_replay/io_tracer.h"
 #include "util/hash.h"
 
 namespace ROCKSDB_NAMESPACE {
@@ -21,13 +22,15 @@ BlobFileCache::BlobFileCache(Cache* cache,
                              const ImmutableCFOptions* immutable_cf_options,
                              const FileOptions* file_options,
                              uint32_t column_family_id,
-                             HistogramImpl* blob_file_read_hist)
+                             HistogramImpl* blob_file_read_hist,
+                             const std::shared_ptr<IOTracer>& io_tracer)
     : cache_(cache),
-      mutex_(kNumberOfMutexStripes, GetSliceNPHash64),
+      mutex_(kNumberOfMutexStripes, kGetSliceNPHash64UnseededFnPtr),
       immutable_cf_options_(immutable_cf_options),
       file_options_(file_options),
       column_family_id_(column_family_id),
-      blob_file_read_hist_(blob_file_read_hist) {
+      blob_file_read_hist_(blob_file_read_hist),
+      io_tracer_(io_tracer) {
   assert(cache_);
   assert(immutable_cf_options_);
   assert(file_options_);
@@ -71,7 +74,7 @@ Status BlobFileCache::GetBlobFileReader(
     assert(file_options_);
     const Status s = BlobFileReader::Create(
         *immutable_cf_options_, *file_options_, column_family_id_,
-        blob_file_read_hist_, blob_file_number, &reader);
+        blob_file_read_hist_, blob_file_number, io_tracer_, &reader);
     if (!s.ok()) {
       RecordTick(statistics, NO_FILE_ERRORS);
       return s;

db/blob/blob_file_cache.h

@@ -20,12 +20,14 @@ class HistogramImpl;
 class Status;
 class BlobFileReader;
 class Slice;
+class IOTracer;
 
 class BlobFileCache {
  public:
   BlobFileCache(Cache* cache, const ImmutableCFOptions* immutable_cf_options,
                 const FileOptions* file_options, uint32_t column_family_id,
-                HistogramImpl* blob_file_read_hist);
+                HistogramImpl* blob_file_read_hist,
+                const std::shared_ptr<IOTracer>& io_tracer);
 
   BlobFileCache(const BlobFileCache&) = delete;
   BlobFileCache& operator=(const BlobFileCache&) = delete;
@@ -42,6 +44,7 @@ class BlobFileCache {
   const FileOptions* file_options_;
   uint32_t column_family_id_;
   HistogramImpl* blob_file_read_hist_;
+  std::shared_ptr<IOTracer> io_tracer_;
 
   static constexpr size_t kNumberOfMutexStripes = 1 << 7;
 };

db/blob/blob_file_cache_test.cc

@@ -46,10 +46,11 @@ void WriteBlobFile(uint32_t column_family_id,
 
   constexpr Statistics* statistics = nullptr;
   constexpr bool use_fsync = false;
+  constexpr bool do_flush = false;
 
   BlobLogWriter blob_log_writer(std::move(file_writer),
                                 immutable_cf_options.env, statistics,
-                                blob_file_number, use_fsync);
+                                blob_file_number, use_fsync, do_flush);
 
   constexpr bool has_ttl = false;
   constexpr ExpirationRange expiration_range;
@@ -113,7 +114,7 @@ TEST_F(BlobFileCacheTest, GetBlobFileReader) {
 
   BlobFileCache blob_file_cache(backing_cache.get(), &immutable_cf_options,
                                 &file_options, column_family_id,
-                                blob_file_read_hist);
+                                blob_file_read_hist, nullptr /*IOTracer*/);
 
   // First try: reader should be opened and put in cache
   CacheHandleGuard<BlobFileReader> first;
@@ -158,7 +159,7 @@ TEST_F(BlobFileCacheTest, GetBlobFileReader_Race) {
 
   BlobFileCache blob_file_cache(backing_cache.get(), &immutable_cf_options,
                                 &file_options, column_family_id,
-                                blob_file_read_hist);
+                                blob_file_read_hist, nullptr /*IOTracer*/);
 
   CacheHandleGuard<BlobFileReader> first;
   CacheHandleGuard<BlobFileReader> second;
@@ -206,7 +207,7 @@ TEST_F(BlobFileCacheTest, GetBlobFileReader_IOError) {
 
   BlobFileCache blob_file_cache(backing_cache.get(), &immutable_cf_options,
                                 &file_options, column_family_id,
-                                blob_file_read_hist);
+                                blob_file_read_hist, nullptr /*IOTracer*/);
 
   // Note: there is no blob file with the below number
   constexpr uint64_t blob_file_number = 123;
@@ -247,7 +248,7 @@ TEST_F(BlobFileCacheTest, GetBlobFileReader_CacheFull) {
 
   BlobFileCache blob_file_cache(backing_cache.get(), &immutable_cf_options,
                                 &file_options, column_family_id,
-                                blob_file_read_hist);
+                                blob_file_read_hist, nullptr /*IOTracer*/);
 
   // Insert into cache should fail since it has zero capacity and
   // strict_capacity_limit is set

db/blob/blob_file_reader.cc

@@ -24,6 +24,7 @@ Status BlobFileReader::Create(
     const ImmutableCFOptions& immutable_cf_options,
     const FileOptions& file_options, uint32_t column_family_id,
     HistogramImpl* blob_file_read_hist, uint64_t blob_file_number,
+    const std::shared_ptr<IOTracer>& io_tracer,
     std::unique_ptr<BlobFileReader>* blob_file_reader) {
   assert(blob_file_reader);
   assert(!*blob_file_reader);
@@ -34,7 +35,7 @@ Status BlobFileReader::Create(
   {
     const Status s =
         OpenFile(immutable_cf_options, file_options, blob_file_read_hist,
-                 blob_file_number, &file_size, &file_reader);
+                 blob_file_number, io_tracer, &file_size, &file_reader);
     if (!s.ok()) {
       return s;
     }
@@ -68,8 +69,8 @@ Status BlobFileReader::Create(
 Status BlobFileReader::OpenFile(
     const ImmutableCFOptions& immutable_cf_options,
     const FileOptions& file_opts, HistogramImpl* blob_file_read_hist,
-    uint64_t blob_file_number, uint64_t* file_size,
-    std::unique_ptr<RandomAccessFileReader>* file_reader) {
+    uint64_t blob_file_number, const std::shared_ptr<IOTracer>& io_tracer,
+    uint64_t* file_size, std::unique_ptr<RandomAccessFileReader>* file_reader) {
   assert(file_size);
   assert(file_reader);
 
@@ -117,10 +118,11 @@ Status BlobFileReader::OpenFile(
   }
 
   file_reader->reset(new RandomAccessFileReader(
-      std::move(file), blob_file_path, immutable_cf_options.env,
-      std::shared_ptr<IOTracer>(), immutable_cf_options.statistics,
-      BLOB_DB_BLOB_FILE_READ_MICROS, blob_file_read_hist,
-      immutable_cf_options.rate_limiter, immutable_cf_options.listeners));
+      std::move(file), blob_file_path,
+      immutable_cf_options.env, io_tracer,
+      immutable_cf_options.statistics, BLOB_DB_BLOB_FILE_READ_MICROS,
+      blob_file_read_hist, immutable_cf_options.rate_limiter,
+      immutable_cf_options.listeners));
 
   return Status::OK();
 }

db/blob/blob_file_reader.h

@@ -29,6 +29,7 @@ class BlobFileReader {
                        uint32_t column_family_id,
                        HistogramImpl* blob_file_read_hist,
                        uint64_t blob_file_number,
+                       const std::shared_ptr<IOTracer>& io_tracer,
                        std::unique_ptr<BlobFileReader>* reader);
 
   BlobFileReader(const BlobFileReader&) = delete;
@@ -47,7 +48,9 @@ class BlobFileReader {
   static Status OpenFile(const ImmutableCFOptions& immutable_cf_options,
                          const FileOptions& file_opts,
                          HistogramImpl* blob_file_read_hist,
-                         uint64_t blob_file_number, uint64_t* file_size,
+                         uint64_t blob_file_number,
+                         const std::shared_ptr<IOTracer>& io_tracer,
+                         uint64_t* file_size,
                          std::unique_ptr<RandomAccessFileReader>* file_reader);
 
   static Status ReadHeader(const RandomAccessFileReader* file_reader,

db/blob/blob_file_reader_test.cc

@@ -54,10 +54,11 @@ void WriteBlobFile(const ImmutableCFOptions& immutable_cf_options,
 
   constexpr Statistics* statistics = nullptr;
   constexpr bool use_fsync = false;
+  constexpr bool do_flush = false;
 
   BlobLogWriter blob_log_writer(std::move(file_writer),
                                 immutable_cf_options.env, statistics,
-                                blob_file_number, use_fsync);
+                                blob_file_number, use_fsync, do_flush);
 
   BlobLogHeader header(column_family_id, compression_type, has_ttl,
                        expiration_range_header);
@@ -141,9 +142,9 @@ TEST_F(BlobFileReaderTest, CreateReaderAndGetBlob) {
 
   std::unique_ptr<BlobFileReader> reader;
 
-  ASSERT_OK(BlobFileReader::Create(immutable_cf_options, FileOptions(),
-                                   column_family_id, blob_file_read_hist,
-                                   blob_file_number, &reader));
+  ASSERT_OK(BlobFileReader::Create(
+      immutable_cf_options, FileOptions(), column_family_id,
+      blob_file_read_hist, blob_file_number, nullptr /*IOTracer*/, &reader));
 
   // Make sure the blob can be retrieved with and without checksum verification
   ReadOptions read_options;
@@ -263,10 +264,11 @@ TEST_F(BlobFileReaderTest, Malformed) {
 
     constexpr Statistics* statistics = nullptr;
     constexpr bool use_fsync = false;
+    constexpr bool do_flush = false;
 
     BlobLogWriter blob_log_writer(std::move(file_writer),
                                   immutable_cf_options.env, statistics,
-                                  blob_file_number, use_fsync);
+                                  blob_file_number, use_fsync, do_flush);
 
     BlobLogHeader header(column_family_id, kNoCompression, has_ttl,
                          expiration_range);
@@ -280,7 +282,8 @@ TEST_F(BlobFileReaderTest, Malformed) {
 
   ASSERT_TRUE(BlobFileReader::Create(immutable_cf_options, FileOptions(),
                                      column_family_id, blob_file_read_hist,
-                                     blob_file_number, &reader)
+                                     blob_file_number, nullptr /*IOTracer*/,
+                                     &reader)
                   .IsCorruption());
 }
 
@@ -313,7 +316,8 @@ TEST_F(BlobFileReaderTest, TTL) {
 
   ASSERT_TRUE(BlobFileReader::Create(immutable_cf_options, FileOptions(),
                                      column_family_id, blob_file_read_hist,
-                                     blob_file_number, &reader)
+                                     blob_file_number, nullptr /*IOTracer*/,
+                                     &reader)
                   .IsCorruption());
 }
 
@@ -351,7 +355,8 @@ TEST_F(BlobFileReaderTest, ExpirationRangeInHeader) {
 
   ASSERT_TRUE(BlobFileReader::Create(immutable_cf_options, FileOptions(),
                                      column_family_id, blob_file_read_hist,
-                                     blob_file_number, &reader)
+                                     blob_file_number, nullptr /*IOTracer*/,
+                                     &reader)
                   .IsCorruption());
 }
 
@@ -389,7 +394,8 @@ TEST_F(BlobFileReaderTest, ExpirationRangeInFooter) {
 
   ASSERT_TRUE(BlobFileReader::Create(immutable_cf_options, FileOptions(),
                                      column_family_id, blob_file_read_hist,
-                                     blob_file_number, &reader)
+                                     blob_file_number, nullptr /*IOTracer*/,
+                                     &reader)
                   .IsCorruption());
 }
 
@@ -427,7 +433,7 @@ TEST_F(BlobFileReaderTest, IncorrectColumnFamily) {
   ASSERT_TRUE(BlobFileReader::Create(immutable_cf_options, FileOptions(),
                                      incorrect_column_family_id,
                                      blob_file_read_hist, blob_file_number,
-                                     &reader)
+                                     nullptr /*IOTracer*/, &reader)
                   .IsCorruption());
 }
 
@@ -458,9 +464,9 @@ TEST_F(BlobFileReaderTest, BlobCRCError) {
 
   std::unique_ptr<BlobFileReader> reader;
 
-  ASSERT_OK(BlobFileReader::Create(immutable_cf_options, FileOptions(),
-                                   column_family_id, blob_file_read_hist,
-                                   blob_file_number, &reader));
+  ASSERT_OK(BlobFileReader::Create(
+      immutable_cf_options, FileOptions(), column_family_id,
+      blob_file_read_hist, blob_file_number, nullptr /*IOTracer*/, &reader));
 
   SyncPoint::GetInstance()->SetCallBack(
       "BlobFileReader::VerifyBlob:CheckBlobCRC", [](void* arg) {
@@ -514,9 +520,9 @@ TEST_F(BlobFileReaderTest, Compression) {
 
   std::unique_ptr<BlobFileReader> reader;
 
-  ASSERT_OK(BlobFileReader::Create(immutable_cf_options, FileOptions(),
-                                   column_family_id, blob_file_read_hist,
-                                   blob_file_number, &reader));
+  ASSERT_OK(BlobFileReader::Create(
+      immutable_cf_options, FileOptions(), column_family_id,
+      blob_file_read_hist, blob_file_number, nullptr /*IOTracer*/, &reader));
 
   // Make sure the blob can be retrieved with and without checksum verification
   ReadOptions read_options;
@@ -574,9 +580,9 @@ TEST_F(BlobFileReaderTest, UncompressionError) {
 
   std::unique_ptr<BlobFileReader> reader;
 
-  ASSERT_OK(BlobFileReader::Create(immutable_cf_options, FileOptions(),
-                                   column_family_id, blob_file_read_hist,
-                                   blob_file_number, &reader));
+  ASSERT_OK(BlobFileReader::Create(
+      immutable_cf_options, FileOptions(), column_family_id,
+      blob_file_read_hist, blob_file_number, nullptr /*IOTracer*/, &reader));
 
   SyncPoint::GetInstance()->SetCallBack(
       "BlobFileReader::UncompressBlobIfNeeded:TamperWithResult", [](void* arg) {
@@ -659,9 +665,9 @@ TEST_P(BlobFileReaderIOErrorTest, IOError) {
 
   std::unique_ptr<BlobFileReader> reader;
 
-  const Status s = BlobFileReader::Create(immutable_cf_options, FileOptions(),
-                                          column_family_id, blob_file_read_hist,
-                                          blob_file_number, &reader);
+  const Status s = BlobFileReader::Create(
+      immutable_cf_options, FileOptions(), column_family_id,
+      blob_file_read_hist, blob_file_number, nullptr /*IOTracer*/, &reader);
 
   const bool fail_during_create =
       (sync_point_ != "BlobFileReader::GetBlob:ReadFromFile");
@@ -739,9 +745,9 @@ TEST_P(BlobFileReaderDecodingErrorTest, DecodingError) {
 
   std::unique_ptr<BlobFileReader> reader;
 
-  const Status s = BlobFileReader::Create(immutable_cf_options, FileOptions(),
-                                          column_family_id, blob_file_read_hist,
-                                          blob_file_number, &reader);
+  const Status s = BlobFileReader::Create(
+      immutable_cf_options, FileOptions(), column_family_id,
+      blob_file_read_hist, blob_file_number, nullptr /*IOTracer*/, &reader);
 
   const bool fail_during_create =
       sync_point_ != "BlobFileReader::GetBlob:TamperWithResult";

db/blob/blob_log_writer.cc

@@ -20,13 +20,15 @@ namespace ROCKSDB_NAMESPACE {
 
 BlobLogWriter::BlobLogWriter(std::unique_ptr<WritableFileWriter>&& dest,
                              Env* env, Statistics* statistics,
-                             uint64_t log_number, bool use_fs, uint64_t boffset)
+                             uint64_t log_number, bool use_fs, bool do_flush,
+                             uint64_t boffset)
     : dest_(std::move(dest)),
       env_(env),
       statistics_(statistics),
       log_number_(log_number),
       block_offset_(boffset),
       use_fsync_(use_fs),
+      do_flush_(do_flush),
       last_elem_type_(kEtNone) {}
 
 BlobLogWriter::~BlobLogWriter() = default;
@@ -49,7 +51,9 @@ Status BlobLogWriter::WriteHeader(BlobLogHeader& header) {
   Status s = dest_->Append(Slice(str));
   if (s.ok()) {
     block_offset_ += str.size();
-    s = dest_->Flush();
+    if (do_flush_) {
+      s = dest_->Flush();
+    }
   }
   last_elem_type_ = kEtFileHdr;
   RecordTick(statistics_, BLOB_DB_BLOB_FILE_BYTES_WRITTEN,
@@ -152,7 +156,7 @@ Status BlobLogWriter::EmitPhysicalRecord(const std::string& headerbuf,
   if (s.ok()) {
     s = dest_->Append(val);
   }
-  if (s.ok()) {
+  if (do_flush_ && s.ok()) {
     s = dest_->Flush();
   }
 

db/blob/blob_log_writer.h

@@ -34,7 +34,7 @@ class BlobLogWriter {
   // "*dest" must remain live while this BlobLogWriter is in use.
   BlobLogWriter(std::unique_ptr<WritableFileWriter>&& dest, Env* env,
                 Statistics* statistics, uint64_t log_number, bool use_fsync,
-                uint64_t boffset = 0);
+                bool do_flush, uint64_t boffset = 0);
   // No copying allowed
   BlobLogWriter(const BlobLogWriter&) = delete;
   BlobLogWriter& operator=(const BlobLogWriter&) = delete;
@@ -74,6 +74,7 @@ class BlobLogWriter {
   uint64_t log_number_;
   uint64_t block_offset_;  // Current offset in block
   bool use_fsync_;
+  bool do_flush_;
 
  public:
   enum ElemType { kEtNone, kEtFileHdr, kEtRecord, kEtFileFooter };

db/blob/db_blob_basic_test.cc

@@ -3,6 +3,8 @@
 //  COPYING file in the root directory) and Apache 2.0 License
 //  (found in the LICENSE.Apache file in the root directory).
 
+#include <array>
+
 #include "db/blob/blob_index.h"
 #include "db/db_test_util.h"
 #include "port/stack_trace.h"
@@ -44,6 +46,85 @@ TEST_F(DBBlobBasicTest, GetBlob) {
                   .IsIncomplete());
 }
 
+TEST_F(DBBlobBasicTest, MultiGetBlobs) {
+  constexpr size_t min_blob_size = 6;
+
+  Options options = GetDefaultOptions();
+  options.enable_blob_files = true;
+  options.min_blob_size = min_blob_size;
+
+  Reopen(options);
+
+  // Put then retrieve three key-values. The first value is below the size limit
+  // and is thus stored inline; the other two are stored separately as blobs.
+  constexpr size_t num_keys = 3;
+
+  constexpr char first_key[] = "first_key";
+  constexpr char first_value[] = "short";
+  static_assert(sizeof(first_value) - 1 < min_blob_size,
+                "first_value too long to be inlined");
+
+  ASSERT_OK(Put(first_key, first_value));
+
+  constexpr char second_key[] = "second_key";
+  constexpr char second_value[] = "long_value";
+  static_assert(sizeof(second_value) - 1 >= min_blob_size,
+                "second_value too short to be stored as blob");
+
+  ASSERT_OK(Put(second_key, second_value));
+
+  constexpr char third_key[] = "third_key";
+  constexpr char third_value[] = "other_long_value";
+  static_assert(sizeof(third_value) - 1 >= min_blob_size,
+                "third_value too short to be stored as blob");
+
+  ASSERT_OK(Put(third_key, third_value));
+
+  ASSERT_OK(Flush());
+
+  ReadOptions read_options;
+
+  std::array<Slice, num_keys> keys{{first_key, second_key, third_key}};
+
+  {
+    std::array<PinnableSlice, num_keys> values;
+    std::array<Status, num_keys> statuses;
+
+    db_->MultiGet(read_options, db_->DefaultColumnFamily(), num_keys, &keys[0],
+                  &values[0], &statuses[0]);
+
+    ASSERT_OK(statuses[0]);
+    ASSERT_EQ(values[0], first_value);
+
+    ASSERT_OK(statuses[1]);
+    ASSERT_EQ(values[1], second_value);
+
+    ASSERT_OK(statuses[2]);
+    ASSERT_EQ(values[2], third_value);
+  }
+
+  // Try again with no I/O allowed. The table and the necessary blocks should
+  // already be in their respective caches. The first (inlined) value should be
+  // successfully read; however, the two blob values could only be read from the
+  // blob file, so for those the read should return Incomplete.
+  read_options.read_tier = kBlockCacheTier;
+
+  {
+    std::array<PinnableSlice, num_keys> values;
+    std::array<Status, num_keys> statuses;
+
+    db_->MultiGet(read_options, db_->DefaultColumnFamily(), num_keys, &keys[0],
+                  &values[0], &statuses[0]);
+
+    ASSERT_OK(statuses[0]);
+    ASSERT_EQ(values[0], first_value);
+
+    ASSERT_TRUE(statuses[1].IsIncomplete());
+
+    ASSERT_TRUE(statuses[2].IsIncomplete());
+  }
+}
+
 TEST_F(DBBlobBasicTest, GetBlob_CorruptIndex) {
   Options options = GetDefaultOptions();
   options.enable_blob_files = true;
@@ -129,6 +210,63 @@ TEST_F(DBBlobBasicTest, GetBlob_IndexWithInvalidFileNumber) {
                   .IsCorruption());
 }
 
+#ifndef ROCKSDB_LITE
+TEST_F(DBBlobBasicTest, GenerateIOTracing) {
+  Options options = GetDefaultOptions();
+  options.enable_blob_files = true;
+  options.min_blob_size = 0;
+  std::string trace_file = dbname_ + "/io_trace_file";
+
+  Reopen(options);
+  {
+    // Create IO trace file
+    std::unique_ptr<TraceWriter> trace_writer;
+    ASSERT_OK(
+        NewFileTraceWriter(env_, EnvOptions(), trace_file, &trace_writer));
+    ASSERT_OK(db_->StartIOTrace(env_, TraceOptions(), std::move(trace_writer)));
+
+    constexpr char key[] = "key";
+    constexpr char blob_value[] = "blob_value";
+
+    ASSERT_OK(Put(key, blob_value));
+    ASSERT_OK(Flush());
+    ASSERT_EQ(Get(key), blob_value);
+
+    ASSERT_OK(db_->EndIOTrace());
+    ASSERT_OK(env_->FileExists(trace_file));
+  }
+  {
+    // Parse trace file to check file opertions related to blob files are
+    // recorded.
+    std::unique_ptr<TraceReader> trace_reader;
+    ASSERT_OK(
+        NewFileTraceReader(env_, EnvOptions(), trace_file, &trace_reader));
+    IOTraceReader reader(std::move(trace_reader));
+
+    IOTraceHeader header;
+    ASSERT_OK(reader.ReadHeader(&header));
+    ASSERT_EQ(kMajorVersion, static_cast<int>(header.rocksdb_major_version));
+    ASSERT_EQ(kMinorVersion, static_cast<int>(header.rocksdb_minor_version));
+
+    // Read records.
+    int blob_files_op_count = 0;
+    Status status;
+    while (true) {
+      IOTraceRecord record;
+      status = reader.ReadIOOp(&record);
+      if (!status.ok()) {
+        break;
+      }
+      if (record.file_name.find("blob") != std::string::npos) {
+        blob_files_op_count++;
+      }
+    }
+    // Assuming blob files will have Append, Close and then Read operations.
+    ASSERT_GT(blob_files_op_count, 2);
+  }
+}
+#endif  // !ROCKSDB_LITE
+
 class DBBlobBasicIOErrorTest : public DBBlobBasicTest,
                                public testing::WithParamInterface<std::string> {
  protected:
@@ -175,6 +313,48 @@ TEST_P(DBBlobBasicIOErrorTest, GetBlob_IOError) {
   SyncPoint::GetInstance()->ClearAllCallBacks();
 }
 
+TEST_P(DBBlobBasicIOErrorTest, MultiGetBlobs_IOError) {
+  Options options = GetDefaultOptions();
+  options.env = fault_injection_env_.get();
+  options.enable_blob_files = true;
+  options.min_blob_size = 0;
+
+  Reopen(options);
+
+  constexpr size_t num_keys = 2;
+
+  constexpr char first_key[] = "first_key";
+  constexpr char first_value[] = "first_value";
+
+  ASSERT_OK(Put(first_key, first_value));
+
+  constexpr char second_key[] = "second_key";
+  constexpr char second_value[] = "second_value";
+
+  ASSERT_OK(Put(second_key, second_value));
+
+  ASSERT_OK(Flush());
+
+  std::array<Slice, num_keys> keys{{first_key, second_key}};
+  std::array<PinnableSlice, num_keys> values;
+  std::array<Status, num_keys> statuses;
+
+  SyncPoint::GetInstance()->SetCallBack(sync_point_, [this](void* /* arg */) {
+    fault_injection_env_->SetFilesystemActive(false,
+                                              Status::IOError(sync_point_));
+  });
+  SyncPoint::GetInstance()->EnableProcessing();
+
+  db_->MultiGet(ReadOptions(), db_->DefaultColumnFamily(), num_keys, &keys[0],
+                &values[0], &statuses[0]);
+
+  SyncPoint::GetInstance()->DisableProcessing();
+  SyncPoint::GetInstance()->ClearAllCallBacks();
+
+  ASSERT_TRUE(statuses[0].IsIOError());
+  ASSERT_TRUE(statuses[1].IsIOError());
+}
+
 }  // namespace ROCKSDB_NAMESPACE
 
 int main(int argc, char** argv) {

db/blob/db_blob_index_test.cc

@@ -98,7 +98,7 @@ class DBBlobIndexTest : public DBTestBase {
   ArenaWrappedDBIter* GetBlobIterator() {
     return dbfull()->NewIteratorImpl(
         ReadOptions(), cfd(), dbfull()->GetLatestSequenceNumber(),
-        nullptr /*read_callback*/, true /*allow_blob*/);
+        nullptr /*read_callback*/, true /*expose_blob_index*/);
   }
 
   Options GetTestOptions() {
@@ -238,8 +238,11 @@ TEST_F(DBBlobIndexTest, Updated) {
   }
 }
 
-// Iterator should get blob value if allow_blob flag is set,
-// otherwise return Status::NotSupported status.
+// Note: the following test case pertains to the StackableDB-based BlobDB
+// implementation. When a blob iterator is used, it should set the
+// expose_blob_index flag for the underlying DBIter, and retrieve/return the
+// corresponding blob value. If a regular DBIter is created (i.e.
+// expose_blob_index is not set), it should return Status::Corruption.
 TEST_F(DBBlobIndexTest, Iterate) {
   const std::vector<std::vector<ValueType>> data = {
       /*00*/ {kTypeValue},
@@ -302,6 +305,7 @@ TEST_F(DBBlobIndexTest, Iterate) {
                     std::function<void(Iterator*)> extra_check = nullptr) {
     // Seek
     auto* iterator = create_iterator();
+    ASSERT_OK(iterator->status());
     ASSERT_OK(iterator->Refresh());
     iterator->Seek(get_key(index));
     check_iterator(iterator, expected_status, forward_value);
@@ -315,6 +319,7 @@ TEST_F(DBBlobIndexTest, Iterate) {
     ASSERT_OK(iterator->Refresh());
     iterator->Seek(get_key(index - 1));
     ASSERT_TRUE(iterator->Valid());
+    ASSERT_OK(iterator->status());
     iterator->Next();
     check_iterator(iterator, expected_status, forward_value);
     if (extra_check) {
@@ -324,6 +329,7 @@ TEST_F(DBBlobIndexTest, Iterate) {
 
     // SeekForPrev
     iterator = create_iterator();
+    ASSERT_OK(iterator->status());
     ASSERT_OK(iterator->Refresh());
     iterator->SeekForPrev(get_key(index));
     check_iterator(iterator, expected_status, backward_value);
@@ -336,6 +342,7 @@ TEST_F(DBBlobIndexTest, Iterate) {
     iterator = create_iterator();
     iterator->Seek(get_key(index + 1));
     ASSERT_TRUE(iterator->Valid());
+    ASSERT_OK(iterator->status());
     iterator->Prev();
     check_iterator(iterator, expected_status, backward_value);
     if (extra_check) {
@@ -373,7 +380,7 @@ TEST_F(DBBlobIndexTest, Iterate) {
             ASSERT_OK(Write(&batch));
             break;
           default:
-            assert(false);
+            FAIL();
         };
       }
       snapshots.push_back(dbfull()->GetSnapshot());
@@ -384,8 +391,8 @@ TEST_F(DBBlobIndexTest, Iterate) {
     MoveDataTo(tier);
 
     // Normal iterator
-    verify(1, Status::kNotSupported, "", "", create_normal_iterator);
-    verify(3, Status::kNotSupported, "", "", create_normal_iterator);
+    verify(1, Status::kCorruption, "", "", create_normal_iterator);
+    verify(3, Status::kCorruption, "", "", create_normal_iterator);
     verify(5, Status::kOk, get_value(5, 0), get_value(5, 0),
            create_normal_iterator);
     verify(7, Status::kOk, get_value(8, 0), get_value(6, 0),

db/builder.cc

@@ -28,6 +28,7 @@
 #include "file/writable_file_writer.h"
 #include "monitoring/iostats_context_imp.h"
 #include "monitoring/thread_status_util.h"
+#include "options/options_helper.h"
 #include "rocksdb/db.h"
 #include "rocksdb/env.h"
 #include "rocksdb/iterator.h"
@@ -147,13 +148,14 @@ Status BuildTable(
             file_checksum, file_checksum_func_name);
         return s;
       }
+      FileTypeSet tmp_set = ioptions.checksum_handoff_file_types;
       file->SetIOPriority(io_priority);
       file->SetWriteLifeTimeHint(write_hint);
-
       file_writer.reset(new WritableFileWriter(
           std::move(file), fname, file_options, env, io_tracer,
           ioptions.statistics, ioptions.listeners,
-          ioptions.file_checksum_gen_factory));
+          ioptions.file_checksum_gen_factory,
+          tmp_set.Contains(FileType::kTableFile)));
 
       builder = NewTableBuilder(
           ioptions, mutable_cf_options, internal_comparator,
@@ -175,8 +177,8 @@ Status BuildTable(
             ? new BlobFileBuilder(versions, env, fs, &ioptions,
                                   &mutable_cf_options, &file_options, job_id,
                                   column_family_id, column_family_name,
-                                  io_priority, write_hint, &blob_file_paths,
-                                  blob_file_additions)
+                                  io_priority, write_hint, io_tracer,
+                                  &blob_file_paths, blob_file_additions)
             : nullptr);
 
     CompactionIterator c_iter(

db/c.cc

@@ -1388,34 +1388,39 @@ char* rocksdb_property_value_cf(
   }
 }
 
-void rocksdb_approximate_sizes(
-    rocksdb_t* db,
-    int num_ranges,
-    const char* const* range_start_key, const size_t* range_start_key_len,
-    const char* const* range_limit_key, const size_t* range_limit_key_len,
-    uint64_t* sizes) {
+void rocksdb_approximate_sizes(rocksdb_t* db, int num_ranges,
+                               const char* const* range_start_key,
+                               const size_t* range_start_key_len,
+                               const char* const* range_limit_key,
+                               const size_t* range_limit_key_len,
+                               uint64_t* sizes, char** errptr) {
   Range* ranges = new Range[num_ranges];
   for (int i = 0; i < num_ranges; i++) {
     ranges[i].start = Slice(range_start_key[i], range_start_key_len[i]);
     ranges[i].limit = Slice(range_limit_key[i], range_limit_key_len[i]);
   }
-  db->rep->GetApproximateSizes(ranges, num_ranges, sizes);
+  Status s = db->rep->GetApproximateSizes(ranges, num_ranges, sizes);
+  if (!s.ok()) {
+    SaveError(errptr, s);
+  }
   delete[] ranges;
 }
 
 void rocksdb_approximate_sizes_cf(
-    rocksdb_t* db,
-    rocksdb_column_family_handle_t* column_family,
-    int num_ranges,
-    const char* const* range_start_key, const size_t* range_start_key_len,
-    const char* const* range_limit_key, const size_t* range_limit_key_len,
-    uint64_t* sizes) {
+    rocksdb_t* db, rocksdb_column_family_handle_t* column_family,
+    int num_ranges, const char* const* range_start_key,
+    const size_t* range_start_key_len, const char* const* range_limit_key,
+    const size_t* range_limit_key_len, uint64_t* sizes, char** errptr) {
   Range* ranges = new Range[num_ranges];
   for (int i = 0; i < num_ranges; i++) {
     ranges[i].start = Slice(range_start_key[i], range_start_key_len[i]);
     ranges[i].limit = Slice(range_limit_key[i], range_limit_key_len[i]);
   }
-  db->rep->GetApproximateSizes(column_family->rep, ranges, num_ranges, sizes);
+  Status s = db->rep->GetApproximateSizes(column_family->rep, ranges,
+                                          num_ranges, sizes);
+  if (!s.ok()) {
+    SaveError(errptr, s);
+  }
   delete[] ranges;
 }
 
@@ -2769,6 +2774,14 @@ void rocksdb_options_set_bottommost_compression_options_zstd_max_train_bytes(
   opt->rep.bottommost_compression_opts.enabled = enabled;
 }
 
+void rocksdb_options_set_bottommost_compression_options_max_dict_buffer_bytes(
+    rocksdb_options_t* opt, uint64_t max_dict_buffer_bytes,
+    unsigned char enabled) {
+  opt->rep.bottommost_compression_opts.max_dict_buffer_bytes =
+      max_dict_buffer_bytes;
+  opt->rep.bottommost_compression_opts.enabled = enabled;
+}
+
 void rocksdb_options_set_compression_options(rocksdb_options_t* opt, int w_bits,
                                              int level, int strategy,
                                              int max_dict_bytes) {
@@ -2783,6 +2796,11 @@ void rocksdb_options_set_compression_options_zstd_max_train_bytes(
   opt->rep.compression_opts.zstd_max_train_bytes = zstd_max_train_bytes;
 }
 
+void rocksdb_options_set_compression_options_max_dict_buffer_bytes(
+    rocksdb_options_t* opt, uint64_t max_dict_buffer_bytes) {
+  opt->rep.compression_opts.max_dict_buffer_bytes = max_dict_buffer_bytes;
+}
+
 void rocksdb_options_set_prefix_extractor(
     rocksdb_options_t* opt, rocksdb_slicetransform_t* prefix_extractor) {
   opt->rep.prefix_extractor.reset(prefix_extractor);
@@ -3931,6 +3949,25 @@ unsigned char rocksdb_readoptions_get_ignore_range_deletions(
   return opt->rep.ignore_range_deletions;
 }
 
+void rocksdb_readoptions_set_deadline(rocksdb_readoptions_t* opt,
+                                      uint64_t microseconds) {
+  opt->rep.deadline = std::chrono::microseconds(microseconds);
+}
+
+uint64_t rocksdb_readoptions_get_deadline(rocksdb_readoptions_t* opt) {
+  return opt->rep.deadline.count();
+}
+
+void rocksdb_readoptions_set_io_timeout(rocksdb_readoptions_t* opt,
+                                        uint64_t microseconds) {
+  opt->rep.io_timeout = std::chrono::microseconds(microseconds);
+}
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_readoptions_get_io_timeout(rocksdb_readoptions_t* opt) {
+  return opt->rep.io_timeout.count();
+}
+
 rocksdb_writeoptions_t* rocksdb_writeoptions_create() {
   return new rocksdb_writeoptions_t;
 }

db/c_test.c

@@ -988,7 +988,9 @@ int main(int argc, char** argv) {
                   &err);
       CheckNoError(err);
     }
-    rocksdb_approximate_sizes(db, 2, start, start_len, limit, limit_len, sizes);
+    rocksdb_approximate_sizes(db, 2, start, start_len, limit, limit_len, sizes,
+                              &err);
+    CheckNoError(err);
     CheckCondition(sizes[0] > 0);
     CheckCondition(sizes[1] > 0);
   }
@@ -2279,6 +2281,12 @@ int main(int argc, char** argv) {
     rocksdb_readoptions_set_ignore_range_deletions(ro, 1);
     CheckCondition(1 == rocksdb_readoptions_get_ignore_range_deletions(ro));
 
+    rocksdb_readoptions_set_deadline(ro, 300);
+    CheckCondition(300 == rocksdb_readoptions_get_deadline(ro));
+
+    rocksdb_readoptions_set_io_timeout(ro, 400);
+    CheckCondition(400 == rocksdb_readoptions_get_io_timeout(ro));
+
     rocksdb_readoptions_destroy(ro);
   }
 

db/column_family.cc

@@ -33,6 +33,7 @@
 #include "monitoring/thread_status_util.h"
 #include "options/options_helper.h"
 #include "port/port.h"
+#include "rocksdb/convenience.h"
 #include "rocksdb/table.h"
 #include "table/merging_iterator.h"
 #include "util/autovector.h"
@@ -335,7 +336,9 @@ ColumnFamilyOptions SanitizeOptions(const ImmutableDBOptions& db_options,
   // was not used)
   auto sfm = static_cast<SstFileManagerImpl*>(db_options.sst_file_manager.get());
   for (size_t i = 0; i < result.cf_paths.size(); i++) {
-    DeleteScheduler::CleanupDirectory(db_options.env, sfm, result.cf_paths[i].path);
+    DeleteScheduler::CleanupDirectory(db_options.env, sfm,
+                                      result.cf_paths[i].path)
+        .PermitUncheckedError();
   }
 #endif
 
@@ -449,9 +452,7 @@ void SuperVersion::Cleanup() {
     to_delete.push_back(m);
   }
   current->Unref();
-  if (cfd->Unref()) {
-    delete cfd;
-  }
+  cfd->UnrefAndTryDelete(this);
 }
 
 void SuperVersion::Init(ColumnFamilyData* new_cfd, MemTable* new_mem,
@@ -562,7 +563,7 @@ ColumnFamilyData::ColumnFamilyData(
                                       block_cache_tracer, io_tracer));
     blob_file_cache_.reset(
         new BlobFileCache(_table_cache, ioptions(), soptions(), id_,
-                          internal_stats_->GetBlobFileReadHist()));
+                          internal_stats_->GetBlobFileReadHist(), io_tracer));
 
     if (ioptions_.compaction_style == kCompactionStyleLevel) {
       compaction_picker_.reset(
@@ -660,7 +661,7 @@ ColumnFamilyData::~ColumnFamilyData() {
   }
 }
 
-bool ColumnFamilyData::UnrefAndTryDelete() {
+bool ColumnFamilyData::UnrefAndTryDelete(SuperVersion* sv_under_cleanup) {
   int old_refs = refs_.fetch_sub(1);
   assert(old_refs > 0);
 
@@ -670,7 +671,11 @@ bool ColumnFamilyData::UnrefAndTryDelete() {
     return true;
   }
 
-  if (old_refs == 2 && super_version_ != nullptr) {
+  // If called under SuperVersion::Cleanup, we should not re-enter Cleanup on
+  // the same SuperVersion. (But while installing a new SuperVersion, this
+  // cfd could be referenced only by two SuperVersions.)
+  if (old_refs == 2 && super_version_ != nullptr &&
+      super_version_ != sv_under_cleanup) {
     // Only the super_version_ holds me
     SuperVersion* sv = super_version_;
     super_version_ = nullptr;
@@ -708,9 +713,7 @@ uint64_t ColumnFamilyData::OldestLogToKeep() {
   auto current_log = GetLogNumber();
 
   if (allow_2pc_) {
-    autovector<MemTable*> empty_list;
-    auto imm_prep_log =
-        imm()->PrecomputeMinLogContainingPrepSection(empty_list);
+    auto imm_prep_log = imm()->PrecomputeMinLogContainingPrepSection();
     auto mem_prep_log = mem()->GetMinLogContainingPrepSection();
 
     if (imm_prep_log > 0 && imm_prep_log < current_log) {
@@ -1339,24 +1342,33 @@ Status ColumnFamilyData::ValidateOptions(
           "Block-Based Table format. ");
     }
   }
+
+  if (cf_options.enable_blob_garbage_collection &&
+      (cf_options.blob_garbage_collection_age_cutoff < 0.0 ||
+       cf_options.blob_garbage_collection_age_cutoff > 1.0)) {
+    return Status::InvalidArgument(
+        "The age cutoff for blob garbage collection should be in the range "
+        "[0.0, 1.0].");
+  }
+
   return s;
 }
 
 #ifndef ROCKSDB_LITE
 Status ColumnFamilyData::SetOptions(
-    const DBOptions& db_options,
+    const DBOptions& db_opts,
     const std::unordered_map<std::string, std::string>& options_map) {
-  MutableCFOptions new_mutable_cf_options;
-  Status s =
-      GetMutableOptionsFromStrings(mutable_cf_options_, options_map,
-                                   ioptions_.info_log, &new_mutable_cf_options);
+  ColumnFamilyOptions cf_opts =
+      BuildColumnFamilyOptions(initial_cf_options_, mutable_cf_options_);
+  ConfigOptions config_opts;
+  config_opts.mutable_options_only = true;
+  Status s = GetColumnFamilyOptionsFromMap(config_opts, cf_opts, options_map,
+                                           &cf_opts);
   if (s.ok()) {
-    ColumnFamilyOptions cf_options =
-        BuildColumnFamilyOptions(initial_cf_options_, new_mutable_cf_options);
-    s = ValidateOptions(db_options, cf_options);
+    s = ValidateOptions(db_opts, cf_opts);
   }
   if (s.ok()) {
-    mutable_cf_options_ = new_mutable_cf_options;
+    mutable_cf_options_ = MutableCFOptions(cf_opts);
     mutable_cf_options_.RefreshDerivedOptions(ioptions_);
   }
   return s;

db/column_family.h

@@ -279,21 +279,11 @@ class ColumnFamilyData {
   // holding a DB mutex, or as the leader in a write batch group).
   void Ref() { refs_.fetch_add(1); }
 
-  // Unref decreases the reference count, but does not handle deletion
-  // when the count goes to 0.  If this method returns true then the
-  // caller should delete the instance immediately, or later, by calling
-  // FreeDeadColumnFamilies().  Unref() can only be called while holding
-  // a DB mutex, or during single-threaded recovery.
-  bool Unref() {
-    int old_refs = refs_.fetch_sub(1);
-    assert(old_refs > 0);
-    return old_refs == 1;
-  }
-
   // UnrefAndTryDelete() decreases the reference count and do free if needed,
   // return true if this is freed else false, UnrefAndTryDelete() can only
   // be called while holding a DB mutex, or during single-threaded recovery.
-  bool UnrefAndTryDelete();
+  // sv_under_cleanup is only provided when called from SuperVersion::Cleanup.
+  bool UnrefAndTryDelete(SuperVersion* sv_under_cleanup = nullptr);
 
   // SetDropped() can only be called under following conditions:
   // 1) Holding a DB mutex,
@@ -360,6 +350,11 @@ class ColumnFamilyData {
 
   MemTableList* imm() { return &imm_; }
   MemTable* mem() { return mem_; }
+
+  bool IsEmpty() {
+    return mem()->GetFirstSequenceNumber() == 0 && imm()->NumNotFlushed() == 0;
+  }
+
   Version* current() { return current_; }
   Version* dummy_versions() { return dummy_versions_; }
   void SetCurrent(Version* _current);
@@ -509,6 +504,21 @@ class ColumnFamilyData {
 
   FSDirectory* GetDataDir(size_t path_id) const;
 
+  // full_history_ts_low_ can only increase.
+  void SetFullHistoryTsLow(std::string ts_low) {
+    assert(!ts_low.empty());
+    const Comparator* ucmp = user_comparator();
+    assert(ucmp);
+    if (full_history_ts_low_.empty() ||
+        ucmp->CompareTimestamp(ts_low, full_history_ts_low_) > 0) {
+      full_history_ts_low_ = std::move(ts_low);
+    }
+  }
+
+  const std::string& GetFullHistoryTsLow() const {
+    return full_history_ts_low_;
+  }
+
   ThreadLocalPtr* TEST_GetLocalSV() { return local_sv_.get(); }
 
  private:
@@ -605,6 +615,8 @@ class ColumnFamilyData {
   std::vector<std::shared_ptr<FSDirectory>> data_dirs_;
 
   bool db_paths_registered_;
+
+  std::string full_history_ts_low_;
 };
 
 // ColumnFamilySet has interesting thread-safety requirements

db/column_family_test.cc

@@ -73,7 +73,7 @@ class ColumnFamilyTestBase : public testing::Test {
     db_options_.create_if_missing = true;
     db_options_.fail_if_options_file_error = true;
     db_options_.env = env_;
-    DestroyDB(dbname_, Options(db_options_, column_family_options_));
+    EXPECT_OK(DestroyDB(dbname_, Options(db_options_, column_family_options_)));
   }
 
   ~ColumnFamilyTestBase() override {
@@ -187,8 +187,8 @@ class ColumnFamilyTestBase : public testing::Test {
     std::vector<ColumnFamilyDescriptor> column_families;
     names_.clear();
     for (size_t i = 0; i < cf.size(); ++i) {
-      column_families.push_back(ColumnFamilyDescriptor(
-          cf[i], options.size() == 0 ? column_family_options_ : options[i]));
+      column_families.emplace_back(
+          cf[i], options.size() == 0 ? column_family_options_ : options[i]);
       names_.push_back(cf[i]);
     }
     return DB::Open(db_options_, dbname_, column_families, &handles_, &db_);
@@ -199,8 +199,8 @@ class ColumnFamilyTestBase : public testing::Test {
     std::vector<ColumnFamilyDescriptor> column_families;
     names_.clear();
     for (size_t i = 0; i < cf.size(); ++i) {
-      column_families.push_back(ColumnFamilyDescriptor(
-          cf[i], options.size() == 0 ? column_family_options_ : options[i]));
+      column_families.emplace_back(
+          cf[i], options.size() == 0 ? column_family_options_ : options[i]);
       names_.push_back(cf[i]);
     }
     return DB::OpenForReadOnly(db_options_, dbname_, column_families, &handles_,
@@ -653,8 +653,8 @@ TEST_P(FlushEmptyCFTestWithParam, FlushEmptyCFTest) {
   // after flushing file B is deleted. At the same time, the min log number of
   // default CF is not written to manifest. Log file A still remains.
   // Flushed to SST file Y.
-  Flush(1);
-  Flush(0);
+  ASSERT_OK(Flush(1));
+  ASSERT_OK(Flush(0));
   ASSERT_OK(Put(1, "bar", "v3"));  // seqID 4
   ASSERT_OK(Put(1, "foo", "v4"));  // seqID 5
   ASSERT_OK(db_->FlushWAL(/*sync=*/false));
@@ -708,15 +708,15 @@ TEST_P(FlushEmptyCFTestWithParam, FlushEmptyCFTest2) {
   // and is set to current. Both CFs' min log number is set to file C so after
   // flushing file B is deleted. Log file A still remains.
   // Flushed to SST file Y.
-  Flush(1);
+  ASSERT_OK(Flush(1));
   ASSERT_OK(Put(0, "bar", "v2"));  // seqID 4
   ASSERT_OK(Put(2, "bar", "v2"));  // seqID 5
   ASSERT_OK(Put(1, "bar", "v3"));  // seqID 6
   // Flushing all column families. This forces all CFs' min log to current. This
   // is written to the manifest file. Log file C is cleared.
-  Flush(0);
-  Flush(1);
-  Flush(2);
+  ASSERT_OK(Flush(0));
+  ASSERT_OK(Flush(1));
+  ASSERT_OK(Flush(2));
   // Write to log file D
   ASSERT_OK(Put(1, "bar", "v4"));  // seqID 7
   ASSERT_OK(Put(1, "bar", "v5"));  // seqID 8
@@ -898,9 +898,7 @@ TEST_P(ColumnFamilyTest, IgnoreRecoveredLog) {
   std::vector<std::string> old_files;
   ASSERT_OK(env_->GetChildren(backup_logs, &old_files));
   for (auto& file : old_files) {
-    if (file != "." && file != "..") {
-      ASSERT_OK(env_->DeleteFile(backup_logs + "/" + file));
-    }
+    ASSERT_OK(env_->DeleteFile(backup_logs + "/" + file));
   }
 
   column_family_options_.merge_operator =
@@ -929,9 +927,7 @@ TEST_P(ColumnFamilyTest, IgnoreRecoveredLog) {
   std::vector<std::string> logs;
   ASSERT_OK(env_->GetChildren(db_options_.wal_dir, &logs));
   for (auto& log : logs) {
-    if (log != ".." && log != ".") {
-      CopyFile(db_options_.wal_dir + "/" + log, backup_logs + "/" + log);
-    }
+    CopyFile(db_options_.wal_dir + "/" + log, backup_logs + "/" + log);
   }
 
   // recover the DB
@@ -956,9 +952,7 @@ TEST_P(ColumnFamilyTest, IgnoreRecoveredLog) {
     if (iter == 0) {
       // copy the logs from backup back to wal dir
       for (auto& log : logs) {
-        if (log != ".." && log != ".") {
-          CopyFile(backup_logs + "/" + log, db_options_.wal_dir + "/" + log);
-        }
+        CopyFile(backup_logs + "/" + log, db_options_.wal_dir + "/" + log);
       }
     }
   }
@@ -985,7 +979,7 @@ TEST_P(ColumnFamilyTest, FlushTest) {
     for (int i = 0; i < 3; ++i) {
       uint64_t max_total_in_memory_state =
           MaxTotalInMemoryState();
-      Flush(i);
+      ASSERT_OK(Flush(i));
       AssertMaxTotalInMemoryState(max_total_in_memory_state);
     }
     ASSERT_OK(Put(1, "foofoo", "bar"));
@@ -1093,7 +1087,7 @@ TEST_P(ColumnFamilyTest, CrashAfterFlush) {
   ASSERT_OK(batch.Put(handles_[0], Slice("foo"), Slice("bar")));
   ASSERT_OK(batch.Put(handles_[1], Slice("foo"), Slice("bar")));
   ASSERT_OK(db_->Write(WriteOptions(), &batch));
-  Flush(0);
+  ASSERT_OK(Flush(0));
   fault_env->SetFilesystemActive(false);
 
   std::vector<std::string> names;
@@ -1103,7 +1097,7 @@ TEST_P(ColumnFamilyTest, CrashAfterFlush) {
     }
   }
   Close();
-  fault_env->DropUnsyncedFileData();
+  ASSERT_OK(fault_env->DropUnsyncedFileData());
   fault_env->ResetState();
   Open(names, {});
 
@@ -2236,7 +2230,7 @@ TEST_P(ColumnFamilyTest, FlushStaleColumnFamilies) {
   // files for column family [one], because it's empty
   AssertCountLiveFiles(4);
 
-  Flush(0);
+  ASSERT_OK(Flush(0));
   ASSERT_EQ(0, dbfull()->TEST_total_log_size());
   Close();
 }
@@ -3040,7 +3034,7 @@ TEST_P(ColumnFamilyTest, IteratorCloseWALFile1) {
   Iterator* it = db_->NewIterator(ReadOptions(), handles_[1]);
   ASSERT_OK(it->status());
   // A flush will make `it` hold the last reference of its super version.
-  Flush(1);
+  ASSERT_OK(Flush(1));
 
   ASSERT_OK(Put(1, "fodor", "mirko"));
   ASSERT_OK(Put(0, "fodor", "mirko"));
@@ -3093,7 +3087,7 @@ TEST_P(ColumnFamilyTest, IteratorCloseWALFile2) {
   Iterator* it = db_->NewIterator(ro, handles_[1]);
   ASSERT_OK(it->status());
   // A flush will make `it` hold the last reference of its super version.
-  Flush(1);
+  ASSERT_OK(Flush(1));
 
   ASSERT_OK(Put(1, "fodor", "mirko"));
   ASSERT_OK(Put(0, "fodor", "mirko"));
@@ -3147,7 +3141,7 @@ TEST_P(ColumnFamilyTest, ForwardIteratorCloseWALFile) {
   CreateColumnFamilies({"one"});
   ASSERT_OK(Put(1, "fodor", "mirko"));
   ASSERT_OK(Put(1, "fodar2", "mirko"));
-  Flush(1);
+  ASSERT_OK(Flush(1));
 
   // Create an iterator holding the current super version, as well as
   // the SST file just flushed.
@@ -3159,7 +3153,7 @@ TEST_P(ColumnFamilyTest, ForwardIteratorCloseWALFile) {
 
   ASSERT_OK(Put(1, "fodor", "mirko"));
   ASSERT_OK(Put(1, "fodar2", "mirko"));
-  Flush(1);
+  ASSERT_OK(Flush(1));
 
   WaitForCompaction();
 
@@ -3232,9 +3226,9 @@ TEST_P(ColumnFamilyTest, LogSyncConflictFlush) {
   ROCKSDB_NAMESPACE::port::Thread thread([&] { ASSERT_OK(db_->SyncWAL()); });
 
   TEST_SYNC_POINT("ColumnFamilyTest::LogSyncConflictFlush:1");
-  Flush(1);
+  ASSERT_OK(Flush(1));
   ASSERT_OK(Put(1, "foo", "bar"));
-  Flush(1);
+  ASSERT_OK(Flush(1));
 
   TEST_SYNC_POINT("ColumnFamilyTest::LogSyncConflictFlush:2");
 
@@ -3256,7 +3250,7 @@ TEST_P(ColumnFamilyTest, DISABLED_LogTruncationTest) {
   Build(0, 100);
 
   // Flush the 0th column family to force a roll of the wal log
-  Flush(0);
+  ASSERT_OK(Flush(0));
 
   // Add some more entries
   Build(100, 100);
@@ -3332,14 +3326,14 @@ TEST_P(ColumnFamilyTest, DefaultCfPathsTest) {
 
   // Fill Column family 1.
   PutRandomData(1, 100, 100);
-  Flush(1);
+  ASSERT_OK(Flush(1));
 
   ASSERT_EQ(1, GetSstFileCount(cf_opt1.cf_paths[0].path));
   ASSERT_EQ(0, GetSstFileCount(dbname_));
 
   // Fill column family 2
   PutRandomData(2, 100, 100);
-  Flush(2);
+  ASSERT_OK(Flush(2));
 
   // SST from Column family 2 should be generated in
   // db_paths which is dbname_ in this case.
@@ -3358,14 +3352,14 @@ TEST_P(ColumnFamilyTest, MultipleCFPathsTest) {
   Reopen({ColumnFamilyOptions(), cf_opt1, cf_opt2});
 
   PutRandomData(1, 100, 100, true /* save */);
-  Flush(1);
+  ASSERT_OK(Flush(1));
 
   // Check that files are generated in appropriate paths.
   ASSERT_EQ(1, GetSstFileCount(cf_opt1.cf_paths[0].path));
   ASSERT_EQ(0, GetSstFileCount(dbname_));
 
   PutRandomData(2, 100, 100, true /* save */);
-  Flush(2);
+  ASSERT_OK(Flush(2));
 
   ASSERT_EQ(1, GetSstFileCount(cf_opt2.cf_paths[0].path));
   ASSERT_EQ(0, GetSstFileCount(dbname_));
@@ -3391,6 +3385,30 @@ TEST_P(ColumnFamilyTest, MultipleCFPathsTest) {
   }
 }
 
+TEST(ColumnFamilyTest, ValidateBlobGCCutoff) {
+  DBOptions db_options;
+
+  ColumnFamilyOptions cf_options;
+  cf_options.enable_blob_garbage_collection = true;
+
+  cf_options.blob_garbage_collection_age_cutoff = -0.5;
+  ASSERT_TRUE(ColumnFamilyData::ValidateOptions(db_options, cf_options)
+                  .IsInvalidArgument());
+
+  cf_options.blob_garbage_collection_age_cutoff = 0.0;
+  ASSERT_OK(ColumnFamilyData::ValidateOptions(db_options, cf_options));
+
+  cf_options.blob_garbage_collection_age_cutoff = 0.5;
+  ASSERT_OK(ColumnFamilyData::ValidateOptions(db_options, cf_options));
+
+  cf_options.blob_garbage_collection_age_cutoff = 1.0;
+  ASSERT_OK(ColumnFamilyData::ValidateOptions(db_options, cf_options));
+
+  cf_options.blob_garbage_collection_age_cutoff = 1.5;
+  ASSERT_TRUE(ColumnFamilyData::ValidateOptions(db_options, cf_options)
+                  .IsInvalidArgument());
+}
+
 }  // namespace ROCKSDB_NAMESPACE
 
 #ifdef ROCKSDB_UNITTESTS_WITH_CUSTOM_OBJECTS_FROM_STATIC_LIBS

db/compact_files_test.cc

@@ -91,9 +91,9 @@ TEST_F(CompactFilesTest, L0ConflictsFiles) {
   // create couple files
   // Background compaction starts and waits in BackgroundCallCompaction:0
   for (int i = 0; i < kLevel0Trigger * 4; ++i) {
-    db->Put(WriteOptions(), ToString(i), "");
-    db->Put(WriteOptions(), ToString(100 - i), "");
-    db->Flush(FlushOptions());
+    ASSERT_OK(db->Put(WriteOptions(), ToString(i), ""));
+    ASSERT_OK(db->Put(WriteOptions(), ToString(100 - i), ""));
+    ASSERT_OK(db->Flush(FlushOptions()));
   }
 
   ROCKSDB_NAMESPACE::ColumnFamilyMetaData meta;
@@ -138,18 +138,18 @@ TEST_F(CompactFilesTest, ObsoleteFiles) {
   DB* db = nullptr;
   DestroyDB(db_name_, options);
   Status s = DB::Open(options, db_name_, &db);
-  assert(s.ok());
-  assert(db);
+  ASSERT_OK(s);
+  ASSERT_NE(db, nullptr);
 
   // create couple files
   for (int i = 1000; i < 2000; ++i) {
-    db->Put(WriteOptions(), ToString(i),
-            std::string(kWriteBufferSize / 10, 'a' + (i % 26)));
+    ASSERT_OK(db->Put(WriteOptions(), ToString(i),
+                      std::string(kWriteBufferSize / 10, 'a' + (i % 26))));
   }
 
   auto l0_files = collector->GetFlushedFiles();
   ASSERT_OK(db->CompactFiles(CompactionOptions(), l0_files, 1));
-  static_cast_with_check<DBImpl>(db)->TEST_WaitForCompact();
+  ASSERT_OK(static_cast_with_check<DBImpl>(db)->TEST_WaitForCompact());
 
   // verify all compaction input files are deleted
   for (auto fname : l0_files) {
@@ -182,15 +182,17 @@ TEST_F(CompactFilesTest, NotCutOutputOnLevel0) {
 
   // create couple files
   for (int i = 0; i < 500; ++i) {
-    db->Put(WriteOptions(), ToString(i), std::string(1000, 'a' + (i % 26)));
+    ASSERT_OK(db->Put(WriteOptions(), ToString(i),
+                      std::string(1000, 'a' + (i % 26))));
   }
-  static_cast_with_check<DBImpl>(db)->TEST_WaitForFlushMemTable();
+  ASSERT_OK(static_cast_with_check<DBImpl>(db)->TEST_WaitForFlushMemTable());
   auto l0_files_1 = collector->GetFlushedFiles();
   collector->ClearFlushedFiles();
   for (int i = 0; i < 500; ++i) {
-    db->Put(WriteOptions(), ToString(i), std::string(1000, 'a' + (i % 26)));
+    ASSERT_OK(db->Put(WriteOptions(), ToString(i),
+                      std::string(1000, 'a' + (i % 26))));
   }
-  static_cast_with_check<DBImpl>(db)->TEST_WaitForFlushMemTable();
+  ASSERT_OK(static_cast_with_check<DBImpl>(db)->TEST_WaitForFlushMemTable());
   auto l0_files_2 = collector->GetFlushedFiles();
   ASSERT_OK(db->CompactFiles(CompactionOptions(), l0_files_1, 0));
   ASSERT_OK(db->CompactFiles(CompactionOptions(), l0_files_2, 0));
@@ -213,13 +215,13 @@ TEST_F(CompactFilesTest, CapturingPendingFiles) {
   DB* db = nullptr;
   DestroyDB(db_name_, options);
   Status s = DB::Open(options, db_name_, &db);
-  assert(s.ok());
+  ASSERT_OK(s);
   assert(db);
 
   // Create 5 files.
   for (int i = 0; i < 5; ++i) {
-    db->Put(WriteOptions(), "key" + ToString(i), "value");
-    db->Flush(FlushOptions());
+    ASSERT_OK(db->Put(WriteOptions(), "key" + ToString(i), "value"));
+    ASSERT_OK(db->Flush(FlushOptions()));
   }
 
   auto l0_files = collector->GetFlushedFiles();
@@ -237,8 +239,8 @@ TEST_F(CompactFilesTest, CapturingPendingFiles) {
 
   // In the meantime flush another file.
   TEST_SYNC_POINT("CompactFilesTest.CapturingPendingFiles:0");
-  db->Put(WriteOptions(), "key5", "value");
-  db->Flush(FlushOptions());
+  ASSERT_OK(db->Put(WriteOptions(), "key5", "value"));
+  ASSERT_OK(db->Flush(FlushOptions()));
   TEST_SYNC_POINT("CompactFilesTest.CapturingPendingFiles:1");
 
   compaction_thread.join();
@@ -249,7 +251,7 @@ TEST_F(CompactFilesTest, CapturingPendingFiles) {
 
   // Make sure we can reopen the DB.
   s = DB::Open(options, db_name_, &db);
-  ASSERT_TRUE(s.ok());
+  ASSERT_OK(s);
   assert(db);
   delete db;
 }
@@ -293,8 +295,8 @@ TEST_F(CompactFilesTest, CompactionFilterWithGetSv) {
   cf->SetDB(db);
 
   // Write one L0 file
-  db->Put(WriteOptions(), "K1", "V1");
-  db->Flush(FlushOptions());
+  ASSERT_OK(db->Put(WriteOptions(), "K1", "V1"));
+  ASSERT_OK(db->Flush(FlushOptions()));
 
   // Compact all L0 files using CompactFiles
   ROCKSDB_NAMESPACE::ColumnFamilyMetaData meta;
@@ -337,8 +339,8 @@ TEST_F(CompactFilesTest, SentinelCompressionType) {
     DB* db = nullptr;
     ASSERT_OK(DB::Open(options, db_name_, &db));
 
-    db->Put(WriteOptions(), "key", "val");
-    db->Flush(FlushOptions());
+    ASSERT_OK(db->Put(WriteOptions(), "key", "val"));
+    ASSERT_OK(db->Flush(FlushOptions()));
 
     auto l0_files = collector->GetFlushedFiles();
     ASSERT_EQ(1, l0_files.size());
@@ -377,14 +379,15 @@ TEST_F(CompactFilesTest, GetCompactionJobInfo) {
   DB* db = nullptr;
   DestroyDB(db_name_, options);
   Status s = DB::Open(options, db_name_, &db);
-  assert(s.ok());
+  ASSERT_OK(s);
   assert(db);
 
   // create couple files
   for (int i = 0; i < 500; ++i) {
-    db->Put(WriteOptions(), ToString(i), std::string(1000, 'a' + (i % 26)));
+    ASSERT_OK(db->Put(WriteOptions(), ToString(i),
+                      std::string(1000, 'a' + (i % 26))));
   }
-  static_cast_with_check<DBImpl>(db)->TEST_WaitForFlushMemTable();
+  ASSERT_OK(static_cast_with_check<DBImpl>(db)->TEST_WaitForFlushMemTable());
   auto l0_files_1 = collector->GetFlushedFiles();
   CompactionOptions co;
   co.compression = CompressionType::kLZ4Compression;

db/compacted_db_impl.h

@@ -82,6 +82,11 @@ class CompactedDBImpl : public DBImpl {
                        ColumnFamilyHandle* /*column_family*/) override {
     return Status::NotSupported("Not supported in compacted db mode.");
   }
+
+  virtual Status SyncWAL() override {
+    return Status::NotSupported("Not supported in compacted db mode.");
+  }
+
   using DB::IngestExternalFile;
   virtual Status IngestExternalFile(
       ColumnFamilyHandle* /*column_family*/,

db/compaction/compaction_iterator.cc

@@ -5,9 +5,11 @@
 
 #include "db/compaction/compaction_iterator.h"
 
-#include <cinttypes>
+#include <iterator>
+#include <limits>
 
 #include "db/blob/blob_file_builder.h"
+#include "db/blob/blob_index.h"
 #include "db/snapshot_checker.h"
 #include "port/likely.h"
 #include "rocksdb/listener.h"
@@ -94,6 +96,8 @@ CompactionIterator::CompactionIterator(
       current_user_key_sequence_(0),
       current_user_key_snapshot_(0),
       merge_out_iter_(merge_helper_),
+      blob_garbage_collection_cutoff_file_number_(
+          ComputeBlobGarbageCollectionCutoffFileNumber(compaction_.get())),
       current_key_committed_(false),
       cmp_with_history_ts_low_(0) {
   assert(compaction_filter_ == nullptr || compaction_ != nullptr);
@@ -418,8 +422,8 @@ void CompactionIterator::NextFromInput() {
       // In the previous iteration we encountered a single delete that we could
       // not compact out.  We will keep this Put, but can drop it's data.
       // (See Optimization 3, below.)
-      assert(ikey_.type == kTypeValue);
-      if (ikey_.type != kTypeValue) {
+      assert(ikey_.type == kTypeValue || ikey_.type == kTypeBlobIndex);
+      if (ikey_.type != kTypeValue && ikey_.type != kTypeBlobIndex) {
         ROCKS_LOG_FATAL(info_log_,
                         "Unexpected key type %d for compaction output",
                         ikey_.type);
@@ -432,6 +436,11 @@ void CompactionIterator::NextFromInput() {
                         current_user_key_snapshot_, last_snapshot);
       }
 
+      if (ikey_.type == kTypeBlobIndex) {
+        ikey_.type = kTypeValue;
+        current_key_.UpdateInternalKey(ikey_.sequence, ikey_.type);
+      }
+
       value_.clear();
       valid_ = true;
       clear_and_output_next_key_ = false;
@@ -733,40 +742,136 @@ void CompactionIterator::NextFromInput() {
   }
 }
 
+bool CompactionIterator::ExtractLargeValueIfNeededImpl() {
+  if (!blob_file_builder_) {
+    return false;
+  }
+
+  blob_index_.clear();
+  const Status s = blob_file_builder_->Add(user_key(), value_, &blob_index_);
+
+  if (!s.ok()) {
+    status_ = s;
+    valid_ = false;
+
+    return false;
+  }
+
+  if (blob_index_.empty()) {
+    return false;
+  }
+
+  value_ = blob_index_;
+
+  return true;
+}
+
+void CompactionIterator::ExtractLargeValueIfNeeded() {
+  assert(ikey_.type == kTypeValue);
+
+  if (!ExtractLargeValueIfNeededImpl()) {
+    return;
+  }
+
+  ikey_.type = kTypeBlobIndex;
+  current_key_.UpdateInternalKey(ikey_.sequence, ikey_.type);
+}
+
+void CompactionIterator::GarbageCollectBlobIfNeeded() {
+  assert(ikey_.type == kTypeBlobIndex);
+
+  if (!compaction_) {
+    return;
+  }
+
+  // GC for integrated BlobDB
+  if (compaction_->enable_blob_garbage_collection()) {
+    BlobIndex blob_index;
+
+    {
+      const Status s = blob_index.DecodeFrom(value_);
+
+      if (!s.ok()) {
+        status_ = s;
+        valid_ = false;
+
+        return;
+      }
+    }
+
+    if (blob_index.IsInlined() || blob_index.HasTTL()) {
+      status_ = Status::Corruption("Unexpected TTL/inlined blob index");
+      valid_ = false;
+
+      return;
+    }
+
+    if (blob_index.file_number() >=
+        blob_garbage_collection_cutoff_file_number_) {
+      return;
+    }
+
+    const Version* const version = compaction_->input_version();
+    assert(version);
+
+    {
+      const Status s =
+          version->GetBlob(ReadOptions(), user_key(), blob_index, &blob_value_);
+
+      if (!s.ok()) {
+        status_ = s;
+        valid_ = false;
+
+        return;
+      }
+    }
+
+    value_ = blob_value_;
+
+    if (ExtractLargeValueIfNeededImpl()) {
+      return;
+    }
+
+    ikey_.type = kTypeValue;
+    current_key_.UpdateInternalKey(ikey_.sequence, ikey_.type);
+
+    return;
+  }
+
+  // GC for stacked BlobDB
+  if (compaction_filter_) {
+    const auto blob_decision = compaction_filter_->PrepareBlobOutput(
+        user_key(), value_, &compaction_filter_value_);
+
+    if (blob_decision == CompactionFilter::BlobDecision::kCorruption) {
+      status_ =
+          Status::Corruption("Corrupted blob reference encountered during GC");
+      valid_ = false;
+
+      return;
+    }
+
+    if (blob_decision == CompactionFilter::BlobDecision::kIOError) {
+      status_ = Status::IOError("Could not relocate blob during GC");
+      valid_ = false;
+
+      return;
+    }
+
+    if (blob_decision == CompactionFilter::BlobDecision::kChangeValue) {
+      value_ = compaction_filter_value_;
+
+      return;
+    }
+  }
+}
+
 void CompactionIterator::PrepareOutput() {
   if (valid_) {
     if (ikey_.type == kTypeValue) {
-      if (blob_file_builder_) {
-        blob_index_.clear();
-        const Status s =
-            blob_file_builder_->Add(user_key(), value_, &blob_index_);
-
-        if (!s.ok()) {
-          status_ = s;
-          valid_ = false;
-        } else if (!blob_index_.empty()) {
-          value_ = blob_index_;
-          ikey_.type = kTypeBlobIndex;
-          current_key_.UpdateInternalKey(ikey_.sequence, ikey_.type);
-        }
-      }
+      ExtractLargeValueIfNeeded();
     } else if (ikey_.type == kTypeBlobIndex) {
-      if (compaction_filter_) {
-        const auto blob_decision = compaction_filter_->PrepareBlobOutput(
-            user_key(), value_, &compaction_filter_value_);
-
-        if (blob_decision == CompactionFilter::BlobDecision::kCorruption) {
-          status_ = Status::Corruption(
-              "Corrupted blob reference encountered during GC");
-          valid_ = false;
-        } else if (blob_decision == CompactionFilter::BlobDecision::kIOError) {
-          status_ = Status::IOError("Could not relocate blob during GC");
-          valid_ = false;
-        } else if (blob_decision ==
-                   CompactionFilter::BlobDecision::kChangeValue) {
-          value_ = compaction_filter_value_;
-        }
-      }
+      GarbageCollectBlobIfNeeded();
     }
 
     // Zeroing out the sequence number leads to better compression.
@@ -894,4 +999,30 @@ bool CompactionIterator::IsInEarliestSnapshot(SequenceNumber sequence) {
   return in_snapshot == SnapshotCheckerResult::kInSnapshot;
 }
 
+uint64_t CompactionIterator::ComputeBlobGarbageCollectionCutoffFileNumber(
+    const CompactionProxy* compaction) {
+  if (!compaction) {
+    return 0;
+  }
+
+  if (!compaction->enable_blob_garbage_collection()) {
+    return 0;
+  }
+
+  Version* const version = compaction->input_version();
+  assert(version);
+
+  const VersionStorageInfo* const storage_info = version->storage_info();
+  assert(storage_info);
+
+  const auto& blob_files = storage_info->GetBlobFiles();
+
+  auto it = blob_files.begin();
+  std::advance(
+      it, compaction->blob_garbage_collection_age_cutoff() * blob_files.size());
+
+  return it != blob_files.end() ? it->first
+                                : std::numeric_limits<uint64_t>::max();
+}
+
 }  // namespace ROCKSDB_NAMESPACE

db/compaction/compaction_iterator.h

@@ -5,6 +5,7 @@
 #pragma once
 
 #include <algorithm>
+#include <cinttypes>
 #include <deque>
 #include <string>
 #include <unordered_set>
@@ -45,6 +46,12 @@ class CompactionIterator {
     virtual bool allow_ingest_behind() const = 0;
 
     virtual bool preserve_deletes() const = 0;
+
+    virtual bool enable_blob_garbage_collection() const = 0;
+
+    virtual double blob_garbage_collection_age_cutoff() const = 0;
+
+    virtual Version* input_version() const = 0;
   };
 
   class RealCompaction : public CompactionProxy {
@@ -53,6 +60,7 @@ class CompactionIterator {
         : compaction_(compaction) {
       assert(compaction_);
       assert(compaction_->immutable_cf_options());
+      assert(compaction_->mutable_cf_options());
     }
 
     int level() const override { return compaction_->level(); }
@@ -80,6 +88,19 @@ class CompactionIterator {
       return compaction_->immutable_cf_options()->preserve_deletes;
     }
 
+    bool enable_blob_garbage_collection() const override {
+      return compaction_->mutable_cf_options()->enable_blob_garbage_collection;
+    }
+
+    double blob_garbage_collection_age_cutoff() const override {
+      return compaction_->mutable_cf_options()
+          ->blob_garbage_collection_age_cutoff;
+    }
+
+    Version* input_version() const override {
+      return compaction_->input_version();
+    }
+
    private:
     const Compaction* compaction_;
   };
@@ -146,11 +167,30 @@ class CompactionIterator {
   // Processes the input stream to find the next output
   void NextFromInput();
 
-  // Do last preparations before presenting the output to the callee. At this
-  // point this only zeroes out the sequence number if possible for better
-  // compression.
+  // Do final preparations before presenting the output to the callee.
   void PrepareOutput();
 
+  // Passes the output value to the blob file builder (if any), and replaces it
+  // with the corresponding blob reference if it has been actually written to a
+  // blob file (i.e. if it passed the value size check). Returns true if the
+  // value got extracted to a blob file, false otherwise.
+  bool ExtractLargeValueIfNeededImpl();
+
+  // Extracts large values as described above, and updates the internal key's
+  // type to kTypeBlobIndex if the value got extracted. Should only be called
+  // for regular values (kTypeValue).
+  void ExtractLargeValueIfNeeded();
+
+  // Relocates valid blobs residing in the oldest blob files if garbage
+  // collection is enabled. Relocated blobs are written to new blob files or
+  // inlined in the LSM tree depending on the current settings (i.e.
+  // enable_blob_files and min_blob_size). Should only be called for blob
+  // references (kTypeBlobIndex).
+  //
+  // Note: the stacked BlobDB implementation's compaction filter based GC
+  // algorithm is also called from here.
+  void GarbageCollectBlobIfNeeded();
+
   // Invoke compaction filter if needed.
   // Return true on success, false on failures (e.g.: kIOError).
   bool InvokeFilterIfNeeded(bool* need_skip, Slice* skip_until);
@@ -191,6 +231,9 @@ class CompactionIterator {
     }
   }
 
+  static uint64_t ComputeBlobGarbageCollectionCutoffFileNumber(
+      const CompactionProxy* compaction);
+
   InternalIterator* input_;
   const Comparator* cmp_;
   MergeHelper* merge_helper_;
@@ -273,7 +316,11 @@ class CompactionIterator {
   // PinnedIteratorsManager used to pin input_ Iterator blocks while reading
   // merge operands and then releasing them after consuming them.
   PinnedIteratorsManager pinned_iters_mgr_;
+
+  uint64_t blob_garbage_collection_cutoff_file_number_;
+
   std::string blob_index_;
+  PinnableSlice blob_value_;
   std::string compaction_filter_value_;
   InternalKey compaction_filter_skip_until_;
   // "level_ptrs" holds indices that remember which file of an associated

db/compaction/compaction_iterator_test.cc

@@ -176,6 +176,12 @@ class FakeCompaction : public CompactionIterator::CompactionProxy {
 
   bool preserve_deletes() const override { return false; }
 
+  bool enable_blob_garbage_collection() const override { return false; }
+
+  double blob_garbage_collection_age_cutoff() const override { return 0.0; }
+
+  Version* input_version() const override { return nullptr; }
+
   bool key_not_exists_beyond_output_level = false;
 
   bool is_bottommost_level = false;
@@ -964,6 +970,21 @@ TEST_F(CompactionIteratorWithSnapshotCheckerTest,
           2 /*earliest_write_conflict_snapshot*/);
 }
 
+// Same as above but with a blob index. In addition to the value getting
+// trimmed, the type of the KV is changed to kTypeValue.
+TEST_F(CompactionIteratorWithSnapshotCheckerTest,
+       KeepSingleDeletionForWriteConflictChecking_BlobIndex) {
+  AddSnapshot(2, 0);
+  RunTest({test::KeyStr("a", 2, kTypeSingleDeletion),
+           test::KeyStr("a", 1, kTypeBlobIndex)},
+          {"", "fake_blob_index"},
+          {test::KeyStr("a", 2, kTypeSingleDeletion),
+           test::KeyStr("a", 1, kTypeValue)},
+          {"", ""}, 2 /*last_committed_seq*/, nullptr /*merge_operator*/,
+          nullptr /*compaction_filter*/, false /*bottommost_level*/,
+          2 /*earliest_write_conflict_snapshot*/);
+}
+
 // Compaction filter should keep uncommitted key as-is, and
 //   * Convert the latest velue to deletion, and/or
 //   * if latest value is a merge, apply filter to all suequent merges.

db/compaction/compaction_job.cc

@@ -46,6 +46,7 @@
 #include "monitoring/iostats_context_imp.h"
 #include "monitoring/perf_context_imp.h"
 #include "monitoring/thread_status_util.h"
+#include "options/options_helper.h"
 #include "port/port.h"
 #include "rocksdb/db.h"
 #include "rocksdb/env.h"
@@ -976,7 +977,7 @@ void CompactionJob::ProcessKeyValueCompaction(SubcompactionState* sub_compact) {
                 sub_compact->compaction->immutable_cf_options(),
                 mutable_cf_options, &file_options_, job_id_, cfd->GetID(),
                 cfd->GetName(), Env::IOPriority::IO_LOW, write_hint_,
-                &blob_file_paths, &sub_compact->blob_file_additions)
+                io_tracer_, &blob_file_paths, &sub_compact->blob_file_additions)
           : nullptr);
 
   TEST_SYNC_POINT("CompactionJob::Run():Inprogress");
@@ -1500,7 +1501,18 @@ Status CompactionJob::FinishCompactionOutputFile(
     std::string fname =
         TableFileName(sub_compact->compaction->immutable_cf_options()->cf_paths,
                       meta->fd.GetNumber(), meta->fd.GetPathId());
-    env_->DeleteFile(fname);
+
+    // TODO(AR) it is not clear if there are any larger implications if
+    // DeleteFile fails here
+    Status ds = env_->DeleteFile(fname);
+    if (!ds.ok()) {
+      ROCKS_LOG_WARN(
+          db_options_.info_log,
+          "[%s] [JOB %d] Unable to remove SST file for table #%" PRIu64
+          " at bottom level%s",
+          cfd->GetName().c_str(), job_id_, output_number,
+          meta->marked_for_compaction ? " (need compaction)" : "");
+    }
 
     // Also need to remove the file from outputs, or it will be added to the
     // VersionEdit.
@@ -1555,9 +1567,7 @@ Status CompactionJob::FinishCompactionOutputFile(
           "CompactionJob::FinishCompactionOutputFile:"
           "MaxAllowedSpaceReached");
       InstrumentedMutexLock l(db_mutex_);
-      // Should handle return error?
-      db_error_handler_->SetBGError(s, BackgroundErrorReason::kCompaction)
-          .PermitUncheckedError();
+      db_error_handler_->SetBGError(s, BackgroundErrorReason::kCompaction);
     }
   }
 #endif
@@ -1724,6 +1734,7 @@ Status CompactionJob::OpenCompactionOutputFile(
 
   writable_file->SetIOPriority(Env::IOPriority::IO_LOW);
   writable_file->SetWriteLifeTimeHint(write_hint_);
+  FileTypeSet tmp_set = db_options_.checksum_handoff_file_types;
   writable_file->SetPreallocationBlockSize(static_cast<size_t>(
       sub_compact->compaction->OutputFilePreallocationSize()));
   const auto& listeners =
@@ -1731,7 +1742,8 @@ Status CompactionJob::OpenCompactionOutputFile(
   sub_compact->outfile.reset(new WritableFileWriter(
       std::move(writable_file), fname, file_options_, env_, io_tracer_,
       db_options_.statistics.get(), listeners,
-      db_options_.file_checksum_gen_factory.get()));
+      db_options_.file_checksum_gen_factory.get(),
+      tmp_set.Contains(FileType::kTableFile)));
 
   // If the Column family flag is to only optimize filters for hits,
   // we can skip creating filters if this is the bottommost_level where

db/compaction/compaction_job_stats_test.cc

@@ -297,15 +297,14 @@ class CompactionJobStatsTest : public testing::Test,
     return result;
   }
 
-  uint64_t Size(const Slice& start, const Slice& limit, int cf = 0) {
+  Status Size(uint64_t* size, const Slice& start, const Slice& limit,
+              int cf = 0) {
     Range r(start, limit);
-    uint64_t size;
     if (cf == 0) {
-      db_->GetApproximateSizes(&r, 1, &size);
+      return db_->GetApproximateSizes(&r, 1, size);
     } else {
-      db_->GetApproximateSizes(handles_[1], &r, 1, &size);
+      return db_->GetApproximateSizes(handles_[1], &r, 1, size);
     }
-    return size;
   }
 
   void Compact(int cf, const Slice& start, const Slice& limit,

db/compaction/compaction_job_test.cc

@@ -72,7 +72,7 @@ class CompactionJobTestBase : public testing::Test {
   CompactionJobTestBase(std::string dbname, const Comparator* ucmp,
                         std::function<std::string(uint64_t)> encode_u64_ts)
       : env_(Env::Default()),
-        fs_(std::make_shared<LegacyFileSystemWrapper>(env_)),
+        fs_(env_->GetFileSystem()),
         dbname_(std::move(dbname)),
         ucmp_(ucmp),
         db_options_(),
@@ -269,7 +269,7 @@ class CompactionJobTestBase : public testing::Test {
                        &write_buffer_manager_, &write_controller_,
                        /*block_cache_tracer=*/nullptr, /*io_tracer=*/nullptr));
     compaction_job_stats_.Reset();
-    SetIdentityFile(env_, dbname_);
+    ASSERT_OK(SetIdentityFile(env_, dbname_));
 
     VersionEdit new_db;
     new_db.SetLogNumber(0);

db/compaction/compaction_picker_test.cc

@@ -141,7 +141,7 @@ class CompactionPickerTest : public testing::Test {
     if (temp_vstorage_) {
       VersionBuilder builder(FileOptions(), &ioptions_, nullptr,
                              vstorage_.get(), nullptr);
-      builder.SaveTo(temp_vstorage_.get());
+      ASSERT_OK(builder.SaveTo(temp_vstorage_.get()));
       vstorage_ = std::move(temp_vstorage_);
     }
     vstorage_->CalculateBaseBytes(ioptions_, mutable_cf_options_);

db/corruption_test.cc

@@ -21,6 +21,7 @@
 #include "db/version_set.h"
 #include "env/composite_env_wrapper.h"
 #include "file/filename.h"
+#include "port/stack_trace.h"
 #include "rocksdb/cache.h"
 #include "rocksdb/convenience.h"
 #include "rocksdb/db.h"
@@ -42,7 +43,8 @@ static constexpr int kValueSize = 1000;
 
 class CorruptionTest : public testing::Test {
  public:
-  test::ErrorEnv env_;
+  std::shared_ptr<Env> env_guard_;
+  test::ErrorEnv* env_;
   std::string dbname_;
   std::shared_ptr<Cache> tiny_cache_;
   Options options_;
@@ -53,9 +55,19 @@ class CorruptionTest : public testing::Test {
     // set it to 0), test SequenceNumberRecovery will fail, likely because of a
     // bug in recovery code. Keep it 4 for now to make the test passes.
     tiny_cache_ = NewLRUCache(100, 4);
+    Env* base_env = Env::Default();
+#ifndef ROCKSDB_LITE
+    const char* test_env_uri = getenv("TEST_ENV_URI");
+    if (test_env_uri) {
+      Status s = Env::LoadEnv(test_env_uri, &base_env, &env_guard_);
+      EXPECT_OK(s);
+      EXPECT_NE(Env::Default(), base_env);
+    }
+#endif  //! ROCKSDB_LITE
+    env_ = new test::ErrorEnv(base_env);
     options_.wal_recovery_mode = WALRecoveryMode::kTolerateCorruptedTailRecords;
-    options_.env = &env_;
-    dbname_ = test::PerThreadDBPath("corruption_test");
+    options_.env = env_;
+    dbname_ = test::PerThreadDBPath(env_, "corruption_test");
     Status s = DestroyDB(dbname_, options_);
     EXPECT_OK(s);
 
@@ -77,8 +89,11 @@ class CorruptionTest : public testing::Test {
     if (getenv("KEEP_DB")) {
       fprintf(stdout, "db is still at %s\n", dbname_.c_str());
     } else {
-      EXPECT_OK(DestroyDB(dbname_, Options()));
+      Options opts;
+      opts.env = env_->target();
+      EXPECT_OK(DestroyDB(dbname_, opts));
     }
+    delete env_;
   }
 
   void CloseDb() {
@@ -93,7 +108,7 @@ class CorruptionTest : public testing::Test {
     if (opt.env == Options().env) {
       // If env is not overridden, replace it with ErrorEnv.
       // Otherwise, the test already uses a non-default Env.
-      opt.env = &env_;
+      opt.env = env_;
     }
     opt.arena_block_size = 4096;
     BlockBasedTableOptions table_options;
@@ -176,7 +191,7 @@ class CorruptionTest : public testing::Test {
   void Corrupt(FileType filetype, int offset, int bytes_to_corrupt) {
     // Pick file to corrupt
     std::vector<std::string> filenames;
-    ASSERT_OK(env_.GetChildren(dbname_, &filenames));
+    ASSERT_OK(env_->GetChildren(dbname_, &filenames));
     uint64_t number;
     FileType type;
     std::string fname;
@@ -191,7 +206,7 @@ class CorruptionTest : public testing::Test {
     }
     ASSERT_TRUE(!fname.empty()) << filetype;
 
-    ASSERT_OK(test::CorruptFile(&env_, fname, offset, bytes_to_corrupt));
+    ASSERT_OK(test::CorruptFile(env_, fname, offset, bytes_to_corrupt));
   }
 
   // corrupts exactly one file at level `level`. if no file found at level,
@@ -201,7 +216,7 @@ class CorruptionTest : public testing::Test {
     db_->GetLiveFilesMetaData(&metadata);
     for (const auto& m : metadata) {
       if (m.level == level) {
-        ASSERT_OK(test::CorruptFile(&env_, dbname_ + "/" + m.name, offset,
+        ASSERT_OK(test::CorruptFile(env_, dbname_ + "/" + m.name, offset,
                                     bytes_to_corrupt));
         return;
       }
@@ -268,14 +283,14 @@ TEST_F(CorruptionTest, Recovery) {
 }
 
 TEST_F(CorruptionTest, RecoverWriteError) {
-  env_.writable_file_error_ = true;
+  env_->writable_file_error_ = true;
   Status s = TryReopen();
   ASSERT_TRUE(!s.ok());
 }
 
 TEST_F(CorruptionTest, NewFileErrorDuringWrite) {
   // Do enough writing to force minor compaction
-  env_.writable_file_error_ = true;
+  env_->writable_file_error_ = true;
   const int num =
       static_cast<int>(3 + (Options().write_buffer_size / kValueSize));
   std::string value_storage;
@@ -291,8 +306,8 @@ TEST_F(CorruptionTest, NewFileErrorDuringWrite) {
     ASSERT_TRUE(!failed || !s.ok());
   }
   ASSERT_TRUE(!s.ok());
-  ASSERT_GE(env_.num_writable_file_errors_, 1);
-  env_.writable_file_error_ = false;
+  ASSERT_GE(env_->num_writable_file_errors_, 1);
+  env_->writable_file_error_ = false;
   Reopen();
 }
 
@@ -310,7 +325,7 @@ TEST_F(CorruptionTest, TableFile) {
 
 TEST_F(CorruptionTest, VerifyChecksumReadahead) {
   Options options;
-  SpecialEnv senv(Env::Default());
+  SpecialEnv senv(env_->target());
   options.env = &senv;
   // Disable block cache as we are going to check checksum for
   // the same file twice and measure number of reads.
@@ -432,6 +447,7 @@ TEST_F(CorruptionTest, CorruptedDescriptor) {
 
 TEST_F(CorruptionTest, CompactionInputError) {
   Options options;
+  options.env = env_;
   Reopen(&options);
   Build(10);
   DBImpl* dbi = static_cast_with_check<DBImpl>(db_);
@@ -452,6 +468,7 @@ TEST_F(CorruptionTest, CompactionInputError) {
 
 TEST_F(CorruptionTest, CompactionInputErrorParanoid) {
   Options options;
+  options.env = env_;
   options.paranoid_checks = true;
   options.write_buffer_size = 131072;
   options.max_write_buffer_number = 2;
@@ -537,7 +554,7 @@ TEST_F(CorruptionTest, RangeDeletionCorrupted) {
       ImmutableCFOptions(options_), kRangeDelBlock, &range_del_handle));
 
   ASSERT_OK(TryReopen());
-  ASSERT_OK(test::CorruptFile(&env_, filename,
+  ASSERT_OK(test::CorruptFile(env_, filename,
                               static_cast<int>(range_del_handle.offset()), 1));
   ASSERT_TRUE(TryReopen().IsCorruption());
 }
@@ -545,6 +562,7 @@ TEST_F(CorruptionTest, RangeDeletionCorrupted) {
 TEST_F(CorruptionTest, FileSystemStateCorrupted) {
   for (int iter = 0; iter < 2; ++iter) {
     Options options;
+    options.env = env_;
     options.paranoid_checks = true;
     options.create_if_missing = true;
     Reopen(&options);
@@ -561,13 +579,13 @@ TEST_F(CorruptionTest, FileSystemStateCorrupted) {
 
     if (iter == 0) {  // corrupt file size
       std::unique_ptr<WritableFile> file;
-      env_.NewWritableFile(filename, &file, EnvOptions());
+      ASSERT_OK(env_->NewWritableFile(filename, &file, EnvOptions()));
       ASSERT_OK(file->Append(Slice("corrupted sst")));
       file.reset();
       Status x = TryReopen(&options);
       ASSERT_TRUE(x.IsCorruption());
     } else {  // delete the file
-      ASSERT_OK(env_.DeleteFile(filename));
+      ASSERT_OK(env_->DeleteFile(filename));
       Status x = TryReopen(&options);
       ASSERT_TRUE(x.IsCorruption());
     }
@@ -583,6 +601,7 @@ static const auto& corruption_modes = {
 
 TEST_F(CorruptionTest, ParanoidFileChecksOnFlush) {
   Options options;
+  options.env = env_;
   options.check_flush_compaction_key_order = false;
   options.paranoid_file_checks = true;
   options.create_if_missing = true;
@@ -597,7 +616,7 @@ TEST_F(CorruptionTest, ParanoidFileChecksOnFlush) {
     options.table_factory = mock;
     mock->SetCorruptionMode(mode);
     ASSERT_OK(DB::Open(options, dbname_, &db_));
-    assert(db_ != nullptr);
+    assert(db_ != nullptr);  // suppress false clang-analyze report
     Build(10);
     s = db_->Flush(FlushOptions());
     if (mode == mock::MockTableFactory::kCorruptNone) {
@@ -610,6 +629,7 @@ TEST_F(CorruptionTest, ParanoidFileChecksOnFlush) {
 
 TEST_F(CorruptionTest, ParanoidFileChecksOnCompact) {
   Options options;
+  options.env = env_;
   options.paranoid_file_checks = true;
   options.create_if_missing = true;
   options.check_flush_compaction_key_order = false;
@@ -622,7 +642,7 @@ TEST_F(CorruptionTest, ParanoidFileChecksOnCompact) {
         std::make_shared<mock::MockTableFactory>();
     options.table_factory = mock;
     ASSERT_OK(DB::Open(options, dbname_, &db_));
-    assert(db_ != nullptr);
+    assert(db_ != nullptr);  // suppress false clang-analyze report
     Build(100, 2);
     // ASSERT_OK(db_->Flush(FlushOptions()));
     DBImpl* dbi = static_cast_with_check<DBImpl>(db_);
@@ -639,6 +659,7 @@ TEST_F(CorruptionTest, ParanoidFileChecksOnCompact) {
 
 TEST_F(CorruptionTest, ParanoidFileChecksWithDeleteRangeFirst) {
   Options options;
+  options.env = env_;
   options.check_flush_compaction_key_order = false;
   options.paranoid_file_checks = true;
   options.create_if_missing = true;
@@ -648,7 +669,7 @@ TEST_F(CorruptionTest, ParanoidFileChecksWithDeleteRangeFirst) {
     ASSERT_OK(DestroyDB(dbname_, options));
     ASSERT_OK(DB::Open(options, dbname_, &db_));
     std::string start, end;
-    assert(db_ != nullptr);
+    assert(db_ != nullptr);  // suppress false clang-analyze report
     ASSERT_OK(db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(),
                                Key(3, &start), Key(7, &end)));
     auto snap = db_->GetSnapshot();
@@ -671,6 +692,7 @@ TEST_F(CorruptionTest, ParanoidFileChecksWithDeleteRangeFirst) {
 
 TEST_F(CorruptionTest, ParanoidFileChecksWithDeleteRange) {
   Options options;
+  options.env = env_;
   options.check_flush_compaction_key_order = false;
   options.paranoid_file_checks = true;
   options.create_if_missing = true;
@@ -679,7 +701,7 @@ TEST_F(CorruptionTest, ParanoidFileChecksWithDeleteRange) {
     db_ = nullptr;
     ASSERT_OK(DestroyDB(dbname_, options));
     ASSERT_OK(DB::Open(options, dbname_, &db_));
-    assert(db_ != nullptr);
+    assert(db_ != nullptr);  // suppress false clang-analyze report
     Build(10, 0, 0);
     std::string start, end;
     ASSERT_OK(db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(),
@@ -706,6 +728,7 @@ TEST_F(CorruptionTest, ParanoidFileChecksWithDeleteRange) {
 
 TEST_F(CorruptionTest, ParanoidFileChecksWithDeleteRangeLast) {
   Options options;
+  options.env = env_;
   options.check_flush_compaction_key_order = false;
   options.paranoid_file_checks = true;
   options.create_if_missing = true;
@@ -714,7 +737,7 @@ TEST_F(CorruptionTest, ParanoidFileChecksWithDeleteRangeLast) {
     db_ = nullptr;
     ASSERT_OK(DestroyDB(dbname_, options));
     ASSERT_OK(DB::Open(options, dbname_, &db_));
-    assert(db_ != nullptr);
+    assert(db_ != nullptr);  // suppress false clang-analyze report
     std::string start, end;
     Build(10);
     ASSERT_OK(db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(),
@@ -738,6 +761,7 @@ TEST_F(CorruptionTest, ParanoidFileChecksWithDeleteRangeLast) {
 
 TEST_F(CorruptionTest, LogCorruptionErrorsInCompactionIterator) {
   Options options;
+  options.env = env_;
   options.create_if_missing = true;
   options.allow_data_in_errors = true;
   auto mode = mock::MockTableFactory::kCorruptKey;
@@ -751,7 +775,7 @@ TEST_F(CorruptionTest, LogCorruptionErrorsInCompactionIterator) {
   options.table_factory = mock;
 
   ASSERT_OK(DB::Open(options, dbname_, &db_));
-  assert(db_ != nullptr);
+  assert(db_ != nullptr);  // suppress false clang-analyze report
   Build(100, 2);
 
   DBImpl* dbi = static_cast_with_check<DBImpl>(db_);
@@ -763,6 +787,7 @@ TEST_F(CorruptionTest, LogCorruptionErrorsInCompactionIterator) {
 
 TEST_F(CorruptionTest, CompactionKeyOrderCheck) {
   Options options;
+  options.env = env_;
   options.paranoid_file_checks = false;
   options.create_if_missing = true;
   options.check_flush_compaction_key_order = false;
@@ -773,7 +798,7 @@ TEST_F(CorruptionTest, CompactionKeyOrderCheck) {
       std::make_shared<mock::MockTableFactory>();
   options.table_factory = mock;
   ASSERT_OK(DB::Open(options, dbname_, &db_));
-  assert(db_ != nullptr);
+  assert(db_ != nullptr);  // suppress false clang-analyze report
   mock->SetCorruptionMode(mock::MockTableFactory::kCorruptReorderKey);
   Build(100, 2);
   DBImpl* dbi = static_cast_with_check<DBImpl>(db_);
@@ -786,6 +811,7 @@ TEST_F(CorruptionTest, CompactionKeyOrderCheck) {
 
 TEST_F(CorruptionTest, FlushKeyOrderCheck) {
   Options options;
+  options.env = env_;
   options.paranoid_file_checks = false;
   options.create_if_missing = true;
   ASSERT_OK(db_->SetOptions({{"check_flush_compaction_key_order", "true"}}));
@@ -814,7 +840,6 @@ TEST_F(CorruptionTest, FlushKeyOrderCheck) {
 }
 
 TEST_F(CorruptionTest, DisableKeyOrderCheck) {
-  Options options;
   ASSERT_OK(db_->SetOptions({{"check_flush_compaction_key_order", "false"}}));
   DBImpl* dbi = static_cast_with_check<DBImpl>(db_);
 
@@ -836,7 +861,7 @@ TEST_F(CorruptionTest, DisableKeyOrderCheck) {
 TEST_F(CorruptionTest, VerifyWholeTableChecksum) {
   CloseDb();
   Options options;
-  options.env = &env_;
+  options.env = env_;
   ASSERT_OK(DestroyDB(dbname_, options));
   options.create_if_missing = true;
   options.file_checksum_gen_factory =
@@ -859,7 +884,7 @@ TEST_F(CorruptionTest, VerifyWholeTableChecksum) {
   SyncPoint::GetInstance()->SetCallBack(
       "DBImpl::VerifySstFileChecksum:mismatch", [&](void* arg) {
         auto* s = reinterpret_cast<Status*>(arg);
-        assert(s);
+        ASSERT_NE(s, nullptr);
         ++count;
         ASSERT_NOK(*s);
       });
@@ -870,8 +895,18 @@ TEST_F(CorruptionTest, VerifyWholeTableChecksum) {
 
 }  // namespace ROCKSDB_NAMESPACE
 
+#ifdef ROCKSDB_UNITTESTS_WITH_CUSTOM_OBJECTS_FROM_STATIC_LIBS
+extern "C" {
+void RegisterCustomObjects(int argc, char** argv);
+}
+#else
+void RegisterCustomObjects(int /*argc*/, char** /*argv*/) {}
+#endif  // !ROCKSDB_UNITTESTS_WITH_CUSTOM_OBJECTS_FROM_STATIC_LIBS
+
 int main(int argc, char** argv) {
+  ROCKSDB_NAMESPACE::port::InstallStackTraceHandler();
   ::testing::InitGoogleTest(&argc, argv);
+  RegisterCustomObjects(argc, argv);
   return RUN_ALL_TESTS();
 }
 

db/cuckoo_table_db_test.cc

@@ -129,10 +129,10 @@ TEST_F(CuckooTableDBTest, Flush) {
   ASSERT_OK(Put("key1", "v1"));
   ASSERT_OK(Put("key2", "v2"));
   ASSERT_OK(Put("key3", "v3"));
-  dbfull()->TEST_FlushMemTable();
+  ASSERT_OK(dbfull()->TEST_FlushMemTable());
 
   TablePropertiesCollection ptc;
-  reinterpret_cast<DB*>(dbfull())->GetPropertiesOfAllTables(&ptc);
+  ASSERT_OK(reinterpret_cast<DB*>(dbfull())->GetPropertiesOfAllTables(&ptc));
   ASSERT_EQ(1U, ptc.size());
   ASSERT_EQ(3U, ptc.begin()->second->num_entries);
   ASSERT_EQ("1", FilesPerLevel());
@@ -146,9 +146,9 @@ TEST_F(CuckooTableDBTest, Flush) {
   ASSERT_OK(Put("key4", "v4"));
   ASSERT_OK(Put("key5", "v5"));
   ASSERT_OK(Put("key6", "v6"));
-  dbfull()->TEST_FlushMemTable();
+  ASSERT_OK(dbfull()->TEST_FlushMemTable());
 
-  reinterpret_cast<DB*>(dbfull())->GetPropertiesOfAllTables(&ptc);
+  ASSERT_OK(reinterpret_cast<DB*>(dbfull())->GetPropertiesOfAllTables(&ptc));
   ASSERT_EQ(2U, ptc.size());
   auto row = ptc.begin();
   ASSERT_EQ(3U, row->second->num_entries);
@@ -164,8 +164,8 @@ TEST_F(CuckooTableDBTest, Flush) {
   ASSERT_OK(Delete("key6"));
   ASSERT_OK(Delete("key5"));
   ASSERT_OK(Delete("key4"));
-  dbfull()->TEST_FlushMemTable();
-  reinterpret_cast<DB*>(dbfull())->GetPropertiesOfAllTables(&ptc);
+  ASSERT_OK(dbfull()->TEST_FlushMemTable());
+  ASSERT_OK(reinterpret_cast<DB*>(dbfull())->GetPropertiesOfAllTables(&ptc));
   ASSERT_EQ(3U, ptc.size());
   row = ptc.begin();
   ASSERT_EQ(3U, row->second->num_entries);
@@ -186,10 +186,10 @@ TEST_F(CuckooTableDBTest, FlushWithDuplicateKeys) {
   ASSERT_OK(Put("key1", "v1"));
   ASSERT_OK(Put("key2", "v2"));
   ASSERT_OK(Put("key1", "v3"));  // Duplicate
-  dbfull()->TEST_FlushMemTable();
+  ASSERT_OK(dbfull()->TEST_FlushMemTable());
 
   TablePropertiesCollection ptc;
-  reinterpret_cast<DB*>(dbfull())->GetPropertiesOfAllTables(&ptc);
+  ASSERT_OK(reinterpret_cast<DB*>(dbfull())->GetPropertiesOfAllTables(&ptc));
   ASSERT_EQ(1U, ptc.size());
   ASSERT_EQ(2U, ptc.begin()->second->num_entries);
   ASSERT_EQ("1", FilesPerLevel());
@@ -219,7 +219,7 @@ TEST_F(CuckooTableDBTest, Uint64Comparator) {
   ASSERT_OK(Put(Uint64Key(1), "v1"));
   ASSERT_OK(Put(Uint64Key(2), "v2"));
   ASSERT_OK(Put(Uint64Key(3), "v3"));
-  dbfull()->TEST_FlushMemTable();
+  ASSERT_OK(dbfull()->TEST_FlushMemTable());
 
   ASSERT_EQ("v1", Get(Uint64Key(1)));
   ASSERT_EQ("v2", Get(Uint64Key(2)));
@@ -228,10 +228,10 @@ TEST_F(CuckooTableDBTest, Uint64Comparator) {
 
   // Add more keys.
   ASSERT_OK(Delete(Uint64Key(2)));  // Delete.
-  dbfull()->TEST_FlushMemTable();
+  ASSERT_OK(dbfull()->TEST_FlushMemTable());
   ASSERT_OK(Put(Uint64Key(3), "v0"));  // Update.
   ASSERT_OK(Put(Uint64Key(4), "v4"));
-  dbfull()->TEST_FlushMemTable();
+  ASSERT_OK(dbfull()->TEST_FlushMemTable());
   ASSERT_EQ("v1", Get(Uint64Key(1)));
   ASSERT_EQ("NOT_FOUND", Get(Uint64Key(2)));
   ASSERT_EQ("v0", Get(Uint64Key(3)));
@@ -251,11 +251,11 @@ TEST_F(CuckooTableDBTest, CompactionIntoMultipleFiles) {
   for (int idx = 0; idx < 28; ++idx) {
     ASSERT_OK(Put(Key(idx), std::string(10000, 'a' + char(idx))));
   }
-  dbfull()->TEST_WaitForFlushMemTable();
+  ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable());
   ASSERT_EQ("1", FilesPerLevel());
 
-  dbfull()->TEST_CompactRange(0, nullptr, nullptr, nullptr,
-                              true /* disallow trivial move */);
+  ASSERT_OK(dbfull()->TEST_CompactRange(0, nullptr, nullptr, nullptr,
+                                        true /* disallow trivial move */));
   ASSERT_EQ("0,2", FilesPerLevel());
   for (int idx = 0; idx < 28; ++idx) {
     ASSERT_EQ(std::string(10000, 'a' + char(idx)), Get(Key(idx)));
@@ -274,15 +274,15 @@ TEST_F(CuckooTableDBTest, SameKeyInsertedInTwoDifferentFilesAndCompacted) {
   for (int idx = 0; idx < 11; ++idx) {
     ASSERT_OK(Put(Key(idx), std::string(10000, 'a')));
   }
-  dbfull()->TEST_WaitForFlushMemTable();
+  ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable());
   ASSERT_EQ("1", FilesPerLevel());
 
   // Generate one more file in level-0, and should trigger level-0 compaction
   for (int idx = 0; idx < 11; ++idx) {
     ASSERT_OK(Put(Key(idx), std::string(10000, 'a' + char(idx))));
   }
-  dbfull()->TEST_WaitForFlushMemTable();
-  dbfull()->TEST_CompactRange(0, nullptr, nullptr);
+  ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable());
+  ASSERT_OK(dbfull()->TEST_CompactRange(0, nullptr, nullptr));
 
   ASSERT_EQ("0,1", FilesPerLevel());
   for (int idx = 0; idx < 11; ++idx) {
@@ -303,7 +303,7 @@ TEST_F(CuckooTableDBTest, AdaptiveTable) {
   ASSERT_OK(Put("key1", "v1"));
   ASSERT_OK(Put("key2", "v2"));
   ASSERT_OK(Put("key3", "v3"));
-  dbfull()->TEST_FlushMemTable();
+  ASSERT_OK(dbfull()->TEST_FlushMemTable());
 
   // Write some keys using plain table.
   std::shared_ptr<TableFactory> block_based_factory(
@@ -319,7 +319,7 @@ TEST_F(CuckooTableDBTest, AdaptiveTable) {
   Reopen(&options);
   ASSERT_OK(Put("key4", "v4"));
   ASSERT_OK(Put("key1", "v5"));
-  dbfull()->TEST_FlushMemTable();
+  ASSERT_OK(dbfull()->TEST_FlushMemTable());
 
   // Write some keys using block based table.
   options.table_factory.reset(NewAdaptiveTableFactory(
@@ -328,7 +328,7 @@ TEST_F(CuckooTableDBTest, AdaptiveTable) {
   Reopen(&options);
   ASSERT_OK(Put("key5", "v6"));
   ASSERT_OK(Put("key2", "v7"));
-  dbfull()->TEST_FlushMemTable();
+  ASSERT_OK(dbfull()->TEST_FlushMemTable());
 
   ASSERT_EQ("v5", Get("key1"));
   ASSERT_EQ("v7", Get("key2"));

db/db_basic_test.cc

@@ -21,6 +21,7 @@
 #if !defined(ROCKSDB_LITE)
 #include "test_util/sync_point.h"
 #endif
+#include "util/file_checksum_helper.h"
 #include "util/random.h"
 #include "utilities/fault_injection_env.h"
 #include "utilities/merge_operators.h"
@@ -38,7 +39,10 @@ TEST_F(DBBasicTest, OpenWhenOpen) {
   options.env = env_;
   DB* db2 = nullptr;
   Status s = DB::Open(options, dbname_, &db2);
-  ASSERT_NOK(s);
+  ASSERT_NOK(s) << [db2]() {
+    delete db2;
+    return "db2 open: ok";
+  }();
   ASSERT_EQ(Status::Code::kIOError, s.code());
   ASSERT_EQ(Status::SubCode::kNone, s.subcode());
   ASSERT_TRUE(strstr(s.getState(), "lock ") != nullptr);
@@ -143,7 +147,7 @@ TEST_F(DBBasicTest, ReadOnlyDB) {
 
   // Reopen and flush memtable.
   Reopen(options);
-  Flush();
+  ASSERT_OK(Flush());
   Close();
   // Now check keys in read only mode.
   ASSERT_OK(ReadOnlyReopen(options));
@@ -179,7 +183,7 @@ TEST_F(DBBasicTest, ReadOnlyDBWithWriteDBIdToManifestSet) {
 
   // Reopen and flush memtable.
   Reopen(options);
-  Flush();
+  ASSERT_OK(Flush());
   Close();
   // Now check keys in read only mode.
   ASSERT_OK(ReadOnlyReopen(options));
@@ -202,7 +206,7 @@ TEST_F(DBBasicTest, CompactedDB) {
   Reopen(options);
   // 1 L0 file, use CompactedDB if max_open_files = -1
   ASSERT_OK(Put("aaa", DummyString(kFileSize / 2, '1')));
-  Flush();
+  ASSERT_OK(Flush());
   Close();
   ASSERT_OK(ReadOnlyReopen(options));
   Status s = Put("new", "value");
@@ -220,12 +224,12 @@ TEST_F(DBBasicTest, CompactedDB) {
   Reopen(options);
   // Add more L0 files
   ASSERT_OK(Put("bbb", DummyString(kFileSize / 2, '2')));
-  Flush();
+  ASSERT_OK(Flush());
   ASSERT_OK(Put("aaa", DummyString(kFileSize / 2, 'a')));
-  Flush();
+  ASSERT_OK(Flush());
   ASSERT_OK(Put("bbb", DummyString(kFileSize / 2, 'b')));
   ASSERT_OK(Put("eee", DummyString(kFileSize / 2, 'e')));
-  Flush();
+  ASSERT_OK(Flush());
   Close();
 
   ASSERT_OK(ReadOnlyReopen(options));
@@ -402,13 +406,16 @@ TEST_F(DBBasicTest, GetSnapshot) {
 
 TEST_F(DBBasicTest, CheckLock) {
   do {
-    DB* localdb;
+    DB* localdb = nullptr;
     Options options = CurrentOptions();
     ASSERT_OK(TryReopen(options));
 
     // second open should fail
     Status s = DB::Open(options, dbname_, &localdb);
-    ASSERT_NOK(s);
+    ASSERT_NOK(s) << [localdb]() {
+      delete localdb;
+      return "localdb open: ok";
+    }();
 #ifdef OS_LINUX
     ASSERT_TRUE(s.ToString().find("lock ") != std::string::npos);
 #endif  // OS_LINUX
@@ -1407,10 +1414,10 @@ TEST_F(DBBasicTest, MultiGetBatchedSortedMultiFile) {
     // mix with memtable
     ASSERT_OK(Put(1, "k1", "v1"));
     ASSERT_OK(Put(1, "k2", "v2"));
-    Flush(1);
+    ASSERT_OK(Flush(1));
     ASSERT_OK(Put(1, "k3", "v3"));
     ASSERT_OK(Put(1, "k4", "v4"));
-    Flush(1);
+    ASSERT_OK(Flush(1));
     ASSERT_OK(Delete(1, "k4"));
     ASSERT_OK(Put(1, "k5", "v5"));
     ASSERT_OK(Delete(1, "no_key"));
@@ -1453,19 +1460,19 @@ TEST_F(DBBasicTest, MultiGetBatchedDuplicateKeys) {
   // mix with memtable
   ASSERT_OK(Merge(1, "k1", "v1"));
   ASSERT_OK(Merge(1, "k2", "v2"));
-  Flush(1);
+  ASSERT_OK(Flush(1));
   MoveFilesToLevel(2, 1);
   ASSERT_OK(Merge(1, "k3", "v3"));
   ASSERT_OK(Merge(1, "k4", "v4"));
-  Flush(1);
+  ASSERT_OK(Flush(1));
   MoveFilesToLevel(2, 1);
   ASSERT_OK(Merge(1, "k4", "v4_2"));
   ASSERT_OK(Merge(1, "k6", "v6"));
-  Flush(1);
+  ASSERT_OK(Flush(1));
   MoveFilesToLevel(2, 1);
   ASSERT_OK(Merge(1, "k7", "v7"));
   ASSERT_OK(Merge(1, "k8", "v8"));
-  Flush(1);
+  ASSERT_OK(Flush(1));
   MoveFilesToLevel(2, 1);
 
   get_perf_context()->Reset();
@@ -1505,12 +1512,12 @@ TEST_F(DBBasicTest, MultiGetBatchedMultiLevel) {
     ASSERT_OK(Put("key_" + std::to_string(i), "val_l2_" + std::to_string(i)));
     num_keys++;
     if (num_keys == 8) {
-      Flush();
+      ASSERT_OK(Flush());
       num_keys = 0;
     }
   }
   if (num_keys > 0) {
-    Flush();
+    ASSERT_OK(Flush());
     num_keys = 0;
   }
   MoveFilesToLevel(2);
@@ -1519,12 +1526,12 @@ TEST_F(DBBasicTest, MultiGetBatchedMultiLevel) {
     ASSERT_OK(Put("key_" + std::to_string(i), "val_l1_" + std::to_string(i)));
     num_keys++;
     if (num_keys == 8) {
-      Flush();
+      ASSERT_OK(Flush());
       num_keys = 0;
     }
   }
   if (num_keys > 0) {
-    Flush();
+    ASSERT_OK(Flush());
     num_keys = 0;
   }
   MoveFilesToLevel(1);
@@ -1533,12 +1540,12 @@ TEST_F(DBBasicTest, MultiGetBatchedMultiLevel) {
     ASSERT_OK(Put("key_" + std::to_string(i), "val_l0_" + std::to_string(i)));
     num_keys++;
     if (num_keys == 8) {
-      Flush();
+      ASSERT_OK(Flush());
       num_keys = 0;
     }
   }
   if (num_keys > 0) {
-    Flush();
+    ASSERT_OK(Flush());
     num_keys = 0;
   }
   ASSERT_EQ(0, num_keys);
@@ -1584,12 +1591,12 @@ TEST_F(DBBasicTest, MultiGetBatchedMultiLevelMerge) {
     ASSERT_OK(Put("key_" + std::to_string(i), "val_l2_" + std::to_string(i)));
     num_keys++;
     if (num_keys == 8) {
-      Flush();
+      ASSERT_OK(Flush());
       num_keys = 0;
     }
   }
   if (num_keys > 0) {
-    Flush();
+    ASSERT_OK(Flush());
     num_keys = 0;
   }
   MoveFilesToLevel(2);
@@ -1598,12 +1605,12 @@ TEST_F(DBBasicTest, MultiGetBatchedMultiLevelMerge) {
     ASSERT_OK(Merge("key_" + std::to_string(i), "val_l1_" + std::to_string(i)));
     num_keys++;
     if (num_keys == 8) {
-      Flush();
+      ASSERT_OK(Flush());
       num_keys = 0;
     }
   }
   if (num_keys > 0) {
-    Flush();
+    ASSERT_OK(Flush());
     num_keys = 0;
   }
   MoveFilesToLevel(1);
@@ -1612,12 +1619,12 @@ TEST_F(DBBasicTest, MultiGetBatchedMultiLevelMerge) {
     ASSERT_OK(Merge("key_" + std::to_string(i), "val_l0_" + std::to_string(i)));
     num_keys++;
     if (num_keys == 8) {
-      Flush();
+      ASSERT_OK(Flush());
       num_keys = 0;
     }
   }
   if (num_keys > 0) {
-    Flush();
+    ASSERT_OK(Flush());
     num_keys = 0;
   }
   ASSERT_EQ(0, num_keys);
@@ -1699,7 +1706,7 @@ TEST_F(DBBasicTest, MultiGetBatchedValueSize) {
     ASSERT_OK(Put(1, "k7", "v7_"));
     ASSERT_OK(Put(1, "k3", "v3_"));
     ASSERT_OK(Put(1, "k4", "v4"));
-    Flush(1);
+    ASSERT_OK(Flush(1));
     ASSERT_OK(Delete(1, "k4"));
     ASSERT_OK(Put(1, "k11", "v11"));
     ASSERT_OK(Delete(1, "no_key"));
@@ -1709,7 +1716,7 @@ TEST_F(DBBasicTest, MultiGetBatchedValueSize) {
     ASSERT_OK(Put(1, "k15", "v15"));
     ASSERT_OK(Put(1, "k16", "v16"));
     ASSERT_OK(Put(1, "k17", "v17"));
-    Flush(1);
+    ASSERT_OK(Flush(1));
 
     ASSERT_OK(Put(1, "k1", "v1_"));
     ASSERT_OK(Put(1, "k2", "v2_"));
@@ -1779,12 +1786,12 @@ TEST_F(DBBasicTest, MultiGetBatchedValueSizeMultiLevelMerge) {
     ASSERT_OK(Put("key_" + std::to_string(i), "val_l2_" + std::to_string(i)));
     num_keys++;
     if (num_keys == 8) {
-      Flush();
+      ASSERT_OK(Flush());
       num_keys = 0;
     }
   }
   if (num_keys > 0) {
-    Flush();
+    ASSERT_OK(Flush());
     num_keys = 0;
   }
   MoveFilesToLevel(2);
@@ -1793,12 +1800,12 @@ TEST_F(DBBasicTest, MultiGetBatchedValueSizeMultiLevelMerge) {
     ASSERT_OK(Merge("key_" + std::to_string(i), "val_l1_" + std::to_string(i)));
     num_keys++;
     if (num_keys == 8) {
-      Flush();
+      ASSERT_OK(Flush());
       num_keys = 0;
     }
   }
   if (num_keys > 0) {
-    Flush();
+    ASSERT_OK(Flush());
     num_keys = 0;
   }
   MoveFilesToLevel(1);
@@ -1807,12 +1814,12 @@ TEST_F(DBBasicTest, MultiGetBatchedValueSizeMultiLevelMerge) {
     ASSERT_OK(Merge("key_" + std::to_string(i), "val_l0_" + std::to_string(i)));
     num_keys++;
     if (num_keys == 8) {
-      Flush();
+      ASSERT_OK(Flush());
       num_keys = 0;
     }
   }
   if (num_keys > 0) {
-    Flush();
+    ASSERT_OK(Flush());
     num_keys = 0;
   }
   ASSERT_EQ(0, num_keys);
@@ -1902,10 +1909,10 @@ TEST_F(DBBasicTest, MultiGetStats) {
     keys[i] = Slice(keys_str[i]);
     ASSERT_OK(Put(1, "k" + std::to_string(i), rnd.RandomString(1000)));
     if (i % 100 == 0) {
-      Flush(1);
+      ASSERT_OK(Flush(1));
     }
   }
-  Flush(1);
+  ASSERT_OK(Flush(1));
   MoveFilesToLevel(2, 1);
 
   for (int i = 501; i < 1000; ++i) {
@@ -1913,11 +1920,11 @@ TEST_F(DBBasicTest, MultiGetStats) {
     keys[i] = Slice(keys_str[i]);
     ASSERT_OK(Put(1, "k" + std::to_string(i), rnd.RandomString(1000)));
     if (i % 100 == 0) {
-      Flush(1);
+      ASSERT_OK(Flush(1));
     }
   }
 
-  Flush(1);
+  ASSERT_OK(Flush(1));
   MoveFilesToLevel(2, 1);
 
   for (int i = 1001; i < total_keys; ++i) {
@@ -1925,10 +1932,10 @@ TEST_F(DBBasicTest, MultiGetStats) {
     keys[i] = Slice(keys_str[i]);
     ASSERT_OK(Put(1, "k" + std::to_string(i), rnd.RandomString(1000)));
     if (i % 100 == 0) {
-      Flush(1);
+      ASSERT_OK(Flush(1));
     }
   }
-  Flush(1);
+  ASSERT_OK(Flush(1));
   Close();
 
   ReopenWithColumnFamilies({"default", "pikachu"}, options);
@@ -1955,7 +1962,7 @@ TEST_F(DBBasicTest, MultiGetStats) {
   ASSERT_GT(hist_sst.max, 0);
 
   // Minimun number of blocks read in a level.
-  ASSERT_EQ(hist_data_blocks.min, 0);
+  ASSERT_EQ(hist_data_blocks.min, 3);
   ASSERT_GT(hist_index_and_filter_blocks.min, 0);
   // Minimun number of sst files read in a level.
   ASSERT_GT(hist_sst.max, 0);
@@ -2036,11 +2043,11 @@ TEST_P(DBMultiGetRowCacheTest, MultiGetBatched) {
     ASSERT_OK(Put(1, "k2", "v2"));
     ASSERT_OK(Put(1, "k3", "v3"));
     ASSERT_OK(Put(1, "k4", "v4"));
-    Flush(1);
+    ASSERT_OK(Flush(1));
     ASSERT_OK(Put(1, "k5", "v5"));
     const Snapshot* snap1 = dbfull()->GetSnapshot();
     ASSERT_OK(Delete(1, "k4"));
-    Flush(1);
+    ASSERT_OK(Flush(1));
     const Snapshot* snap2 = dbfull()->GetSnapshot();
 
     get_perf_context()->Reset();
@@ -2179,7 +2186,7 @@ TEST_F(DBBasicTest, MultiGetIOBufferOverrun) {
     std::string value(rnd.RandomString(128) + zero_str);
     assert(Put(Key(i), value) == Status::OK());
   }
-  Flush();
+  ASSERT_OK(Flush());
 
   std::vector<std::string> key_data(10);
   std::vector<Slice> keys;
@@ -2499,6 +2506,42 @@ TEST_F(DBBasicTest, SkipWALIfMissingTableFiles) {
   ASSERT_FALSE(iter->Valid());
   ASSERT_OK(iter->status());
 }
+
+TEST_F(DBBasicTest, DisableTrackWal) {
+  // If WAL tracking was enabled, and then disabled during reopen,
+  // the previously tracked WALs should be removed from MANIFEST.
+
+  Options options = CurrentOptions();
+  options.track_and_verify_wals_in_manifest = true;
+  // extremely small write buffer size,
+  // so that new WALs are created more frequently.
+  options.write_buffer_size = 100;
+  options.env = env_;
+  DestroyAndReopen(options);
+  for (int i = 0; i < 100; i++) {
+    ASSERT_OK(Put("foo" + std::to_string(i), "value" + std::to_string(i)));
+  }
+  ASSERT_OK(dbfull()->TEST_SwitchMemtable());
+  ASSERT_OK(db_->SyncWAL());
+  // Some WALs are tracked.
+  ASSERT_FALSE(dbfull()->TEST_GetVersionSet()->GetWalSet().GetWals().empty());
+  Close();
+
+  // Disable WAL tracking.
+  options.track_and_verify_wals_in_manifest = false;
+  options.create_if_missing = false;
+  ASSERT_OK(TryReopen(options));
+  // Previously tracked WALs are cleared.
+  ASSERT_TRUE(dbfull()->TEST_GetVersionSet()->GetWalSet().GetWals().empty());
+  Close();
+
+  // Re-enable WAL tracking again.
+  options.track_and_verify_wals_in_manifest = true;
+  options.create_if_missing = false;
+  ASSERT_OK(TryReopen(options));
+  ASSERT_TRUE(dbfull()->TEST_GetVersionSet()->GetWalSet().GetWals().empty());
+  Close();
+}
 #endif  // !ROCKSDB_LITE
 
 TEST_F(DBBasicTest, ManifestChecksumMismatch) {
@@ -2528,6 +2571,64 @@ TEST_F(DBBasicTest, ManifestChecksumMismatch) {
   ASSERT_TRUE(s.IsCorruption());
 }
 
+#ifndef ROCKSDB_LITE
+class DBBasicTestTrackWal : public DBTestBase,
+                            public testing::WithParamInterface<bool> {
+ public:
+  DBBasicTestTrackWal()
+      : DBTestBase("/db_basic_test_track_wal", /*env_do_fsync=*/false) {}
+
+  int CountWalFiles() {
+    VectorLogPtr log_files;
+    EXPECT_OK(dbfull()->GetSortedWalFiles(log_files));
+    return static_cast<int>(log_files.size());
+  };
+};
+
+TEST_P(DBBasicTestTrackWal, DoNotTrackObsoleteWal) {
+  // If a WAL becomes obsolete after flushing, but is not deleted from disk yet,
+  // then if SyncWAL is called afterwards, the obsolete WAL should not be
+  // tracked in MANIFEST.
+
+  Options options = CurrentOptions();
+  options.create_if_missing = true;
+  options.track_and_verify_wals_in_manifest = true;
+  options.atomic_flush = GetParam();
+
+  DestroyAndReopen(options);
+  CreateAndReopenWithCF({"cf"}, options);
+  ASSERT_EQ(handles_.size(), 2);  // default, cf
+  // Do not delete WALs.
+  ASSERT_OK(db_->DisableFileDeletions());
+  constexpr int n = 10;
+  std::vector<std::unique_ptr<LogFile>> wals(n);
+  for (size_t i = 0; i < n; i++) {
+    // Generate a new WAL for each key-value.
+    const int cf = i % 2;
+    ASSERT_OK(db_->GetCurrentWalFile(&wals[i]));
+    ASSERT_OK(Put(cf, "k" + std::to_string(i), "v" + std::to_string(i)));
+    ASSERT_OK(Flush({0, 1}));
+  }
+  ASSERT_EQ(CountWalFiles(), n);
+  // Since all WALs are obsolete, no WAL should be tracked in MANIFEST.
+  ASSERT_OK(db_->SyncWAL());
+
+  // Manually delete all WALs.
+  Close();
+  for (const auto& wal : wals) {
+    ASSERT_OK(env_->DeleteFile(LogFileName(dbname_, wal->LogNumber())));
+  }
+
+  // If SyncWAL tracks the obsolete WALs in MANIFEST,
+  // reopen will fail because the WALs are missing from disk.
+  ASSERT_OK(TryReopenWithColumnFamilies({"default", "cf"}, options));
+  Destroy(options);
+}
+
+INSTANTIATE_TEST_CASE_P(DBBasicTestTrackWal, DBBasicTestTrackWal,
+                        testing::Bool());
+#endif  // ROCKSDB_LITE
+
 class DBBasicTestMultiGet : public DBTestBase {
  public:
   DBBasicTestMultiGet(std::string test_dir, int num_cfs, bool compressed_cache,
@@ -2614,9 +2715,9 @@ class DBBasicTestMultiGet : public DBTestBase {
                                : Put(cf, Key(i), values_[i])) == Status::OK());
       }
       if (num_cfs == 1) {
-        Flush();
+        EXPECT_OK(Flush());
       } else {
-        dbfull()->Flush(FlushOptions(), handles_[cf]);
+        EXPECT_OK(dbfull()->Flush(FlushOptions(), handles_[cf]));
       }
 
       for (int i = 0; i < 100; ++i) {
@@ -2628,9 +2729,9 @@ class DBBasicTestMultiGet : public DBTestBase {
                Status::OK());
       }
       if (num_cfs == 1) {
-        Flush();
+        EXPECT_OK(Flush());
       } else {
-        dbfull()->Flush(FlushOptions(), handles_[cf]);
+        EXPECT_OK(dbfull()->Flush(FlushOptions(), handles_[cf]));
       }
     }
   }
@@ -3173,7 +3274,9 @@ class DeadlineFS : public FileSystemWrapper {
 
   // Increment the IO counter and return a delay in microseconds
   IOStatus ShouldDelay(const IOOptions& opts) {
-    if (!deadline_.count() && !io_timeout_.count()) {
+    if (timedout_) {
+      return IOStatus::TimedOut();
+    } else if (!deadline_.count() && !io_timeout_.count()) {
       return IOStatus::OK();
     }
     if (!ignore_deadline_ && delay_trigger_ == io_count_++) {
@@ -3480,6 +3583,29 @@ TEST_F(DBBasicTest, VerifyFileChecksums) {
   ASSERT_OK(Flush());
 
   ASSERT_OK(db_->VerifyFileChecksums(ReadOptions()));
+
+  // Does the right thing but with the wrong name -- using it should lead to an
+  // error.
+  class MisnamedFileChecksumGenerator : public FileChecksumGenCrc32c {
+   public:
+    MisnamedFileChecksumGenerator(const FileChecksumGenContext& context)
+        : FileChecksumGenCrc32c(context) {}
+
+    const char* Name() const override { return "sha1"; }
+  };
+
+  class MisnamedFileChecksumGenFactory : public FileChecksumGenCrc32cFactory {
+   public:
+    std::unique_ptr<FileChecksumGenerator> CreateFileChecksumGenerator(
+        const FileChecksumGenContext& context) override {
+      return std::unique_ptr<FileChecksumGenerator>(
+          new MisnamedFileChecksumGenerator(context));
+    }
+  };
+
+  options.file_checksum_gen_factory.reset(new MisnamedFileChecksumGenFactory());
+  Reopen(options);
+  ASSERT_TRUE(db_->VerifyFileChecksums(ReadOptions()).IsInvalidArgument());
 }
 #endif  // !ROCKSDB_LITE
 
@@ -3543,7 +3669,7 @@ TEST_P(DBBasicTestDeadline, PointLookupDeadline) {
       std::string key = "k" + ToString(i);
       ASSERT_OK(Put(key, rnd.RandomString(100)));
     }
-    Flush();
+    ASSERT_OK(Flush());
 
     bool timedout = true;
     // A timeout will be forced when the IO counter reaches this value

db/db_block_cache_test.cc

@@ -677,7 +677,7 @@ TEST_F(DBBlockCacheTest, ParanoidFileChecks) {
   // Create a new SST file. This will further trigger a compaction
   // and generate another file.
   ASSERT_OK(Flush(1));
-  dbfull()->TEST_WaitForCompact();
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
   ASSERT_EQ(3, /* Totally 3 files created up to now */
             TestGetTickerCount(options, BLOCK_CACHE_ADD));
 
@@ -692,7 +692,7 @@ TEST_F(DBBlockCacheTest, ParanoidFileChecks) {
   ASSERT_OK(Put(1, "1_key4", "val4"));
   ASSERT_OK(Put(1, "9_key4", "val4"));
   ASSERT_OK(Flush(1));
-  dbfull()->TEST_WaitForCompact();
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
   ASSERT_EQ(3, /* Totally 3 files created up to now */
             TestGetTickerCount(options, BLOCK_CACHE_ADD));
 }
@@ -860,7 +860,7 @@ TEST_F(DBBlockCacheTest, CacheCompressionDict) {
       }
       ASSERT_OK(Flush());
     }
-    dbfull()->TEST_WaitForCompact();
+    ASSERT_OK(dbfull()->TEST_WaitForCompact());
     ASSERT_EQ(0, NumTableFilesAtLevel(0));
     ASSERT_EQ(kNumFiles, NumTableFilesAtLevel(1));
 

db/db_bloom_filter_test.cc

@@ -128,8 +128,8 @@ TEST_P(DBBloomFilterTestDefFormatVersion, KeyMayExist) {
     ASSERT_EQ(cache_added, TestGetTickerCount(options, BLOCK_CACHE_ADD));
 
     ASSERT_OK(Flush(1));
-    dbfull()->TEST_CompactRange(0, nullptr, nullptr, handles_[1],
-                                true /* disallow trivial move */);
+    ASSERT_OK(dbfull()->TEST_CompactRange(0, nullptr, nullptr, handles_[1],
+                                          true /* disallow trivial move */));
 
     numopen = TestGetTickerCount(options, NO_FILE_OPENS);
     cache_added = TestGetTickerCount(options, BLOCK_CACHE_ADD);
@@ -178,7 +178,7 @@ TEST_F(DBBloomFilterTest, GetFilterByPrefixBloomCustomPrefixExtractor) {
     ASSERT_OK(dbfull()->Put(wo, "barbarbar2", "foo2"));
     ASSERT_OK(dbfull()->Put(wo, "foofoofoo", "bar"));
 
-    dbfull()->Flush(fo);
+    ASSERT_OK(dbfull()->Flush(fo));
 
     ASSERT_EQ("foo", Get("barbarbar"));
     ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 0);
@@ -244,7 +244,7 @@ TEST_F(DBBloomFilterTest, GetFilterByPrefixBloom) {
     ASSERT_OK(dbfull()->Put(wo, "barbarbar2", "foo2"));
     ASSERT_OK(dbfull()->Put(wo, "foofoofoo", "bar"));
 
-    dbfull()->Flush(fo);
+    ASSERT_OK(dbfull()->Flush(fo));
 
     ASSERT_EQ("foo", Get("barbarbar"));
     ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 0);
@@ -297,7 +297,7 @@ TEST_F(DBBloomFilterTest, WholeKeyFilterProp) {
     // ranges.
     ASSERT_OK(dbfull()->Put(wo, "aaa", ""));
     ASSERT_OK(dbfull()->Put(wo, "zzz", ""));
-    dbfull()->Flush(fo);
+    ASSERT_OK(dbfull()->Flush(fo));
 
     Reopen(options);
     ASSERT_EQ("NOT_FOUND", Get("foo"));
@@ -328,7 +328,7 @@ TEST_F(DBBloomFilterTest, WholeKeyFilterProp) {
     // ranges.
     ASSERT_OK(dbfull()->Put(wo, "aaa", ""));
     ASSERT_OK(dbfull()->Put(wo, "zzz", ""));
-    db_->CompactRange(CompactRangeOptions(), nullptr, nullptr);
+    ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
 
     // Reopen with both of whole key off and prefix extractor enabled.
     // Still no bloom filter should be used.
@@ -351,7 +351,7 @@ TEST_F(DBBloomFilterTest, WholeKeyFilterProp) {
     // ranges.
     ASSERT_OK(dbfull()->Put(wo, "aaa", ""));
     ASSERT_OK(dbfull()->Put(wo, "zzz", ""));
-    db_->CompactRange(CompactRangeOptions(), nullptr, nullptr);
+    ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
 
     options.prefix_extractor.reset();
     bbto.whole_key_filtering = true;
@@ -364,7 +364,7 @@ TEST_F(DBBloomFilterTest, WholeKeyFilterProp) {
     // not filtered out by key ranges.
     ASSERT_OK(dbfull()->Put(wo, "aaa", ""));
     ASSERT_OK(dbfull()->Put(wo, "zzz", ""));
-    Flush();
+    ASSERT_OK(Flush());
 
     // Now we have two files:
     // File 1: An older file with prefix bloom.
@@ -467,7 +467,7 @@ TEST_P(DBBloomFilterTestWithParam, BloomFilter) {
     for (int i = 0; i < N; i += 100) {
       ASSERT_OK(Put(1, Key(i), Key(i)));
     }
-    Flush(1);
+    ASSERT_OK(Flush(1));
 
     // Prevent auto compactions triggered by seeks
     env_->delay_sstable_sync_.store(true, std::memory_order_release);
@@ -880,7 +880,7 @@ TEST_F(DBBloomFilterTest, ContextCustomFilterPolicy) {
 
     // Destroy
     ASSERT_OK(dbfull()->DropColumnFamily(handles_[1]));
-    dbfull()->DestroyColumnFamilyHandle(handles_[1]);
+    ASSERT_OK(dbfull()->DestroyColumnFamilyHandle(handles_[1]));
     handles_[1] = nullptr;
   }
 }
@@ -1444,9 +1444,9 @@ void PrefixScanInit(DBBloomFilterTest* dbtest) {
   snprintf(buf, sizeof(buf), "%02d______:end", 10);
   keystr = std::string(buf);
   ASSERT_OK(dbtest->Put(keystr, keystr));
-  dbtest->Flush();
-  dbtest->dbfull()->CompactRange(CompactRangeOptions(), nullptr,
-                                 nullptr);  // move to level 1
+  ASSERT_OK(dbtest->Flush());
+  ASSERT_OK(dbtest->dbfull()->CompactRange(CompactRangeOptions(), nullptr,
+                                           nullptr));  // move to level 1
 
   // GROUP 1
   for (int i = 1; i <= small_range_sstfiles; i++) {
@@ -1563,21 +1563,21 @@ TEST_F(DBBloomFilterTest, OptimizeFiltersForHits) {
   for (int key : keys) {
     ASSERT_OK(Put(1, Key(key), "val"));
     if (++num_inserted % 1000 == 0) {
-      dbfull()->TEST_WaitForFlushMemTable();
-      dbfull()->TEST_WaitForCompact();
+      ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable());
+      ASSERT_OK(dbfull()->TEST_WaitForCompact());
     }
   }
   ASSERT_OK(Put(1, Key(0), "val"));
   ASSERT_OK(Put(1, Key(numkeys), "val"));
   ASSERT_OK(Flush(1));
-  dbfull()->TEST_WaitForCompact();
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
 
   if (NumTableFilesAtLevel(0, 1) == 0) {
     // No Level 0 file. Create one.
     ASSERT_OK(Put(1, Key(0), "val"));
     ASSERT_OK(Put(1, Key(numkeys), "val"));
     ASSERT_OK(Flush(1));
-    dbfull()->TEST_WaitForCompact();
+    ASSERT_OK(dbfull()->TEST_WaitForCompact());
   }
 
   for (int i = 1; i < numkeys; i += 2) {
@@ -1682,7 +1682,8 @@ TEST_F(DBBloomFilterTest, OptimizeFiltersForHits) {
       BottommostLevelCompaction::kSkip;
   compact_options.change_level = true;
   compact_options.target_level = 7;
-  db_->CompactRange(compact_options, handles_[1], nullptr, nullptr);
+  ASSERT_TRUE(db_->CompactRange(compact_options, handles_[1], nullptr, nullptr)
+                  .IsNotSupported());
 
   ASSERT_EQ(trivial_move, 1);
   ASSERT_EQ(non_trivial_move, 0);
@@ -1714,10 +1715,10 @@ TEST_F(DBBloomFilterTest, OptimizeFiltersForHits) {
 
 int CountIter(std::unique_ptr<Iterator>& iter, const Slice& key) {
   int count = 0;
-  for (iter->Seek(key); iter->Valid() && iter->status() == Status::OK();
-       iter->Next()) {
+  for (iter->Seek(key); iter->Valid(); iter->Next()) {
     count++;
   }
+  EXPECT_OK(iter->status());
   return count;
 }
 
@@ -1747,7 +1748,7 @@ TEST_F(DBBloomFilterTest, DynamicBloomFilterUpperBound) {
     ASSERT_OK(Put("abcdxxx1", "val2"));
     ASSERT_OK(Put("abcdxxx2", "val3"));
     ASSERT_OK(Put("abcdxxx3", "val4"));
-    dbfull()->Flush(FlushOptions());
+    ASSERT_OK(dbfull()->Flush(FlushOptions()));
     {
       // prefix_extractor has not changed, BF will always be read
       Slice upper_bound("abce");
@@ -1905,7 +1906,7 @@ TEST_F(DBBloomFilterTest, DynamicBloomFilterMultipleSST) {
     ASSERT_OK(Put("foo4", "bar4"));
     ASSERT_OK(Put("foq5", "bar5"));
     ASSERT_OK(Put("fpb", "1"));
-    dbfull()->Flush(FlushOptions());
+    ASSERT_OK(dbfull()->Flush(FlushOptions()));
     {
       // BF is cappped:3 now
       std::unique_ptr<Iterator> iter_tmp(db_->NewIterator(read_options));
@@ -1929,7 +1930,7 @@ TEST_F(DBBloomFilterTest, DynamicBloomFilterMultipleSST) {
     ASSERT_OK(Put("foo7", "bar7"));
     ASSERT_OK(Put("foq8", "bar8"));
     ASSERT_OK(Put("fpc", "2"));
-    dbfull()->Flush(FlushOptions());
+    ASSERT_OK(dbfull()->Flush(FlushOptions()));
     {
       // BF is fixed:2 now
       std::unique_ptr<Iterator> iter_tmp(db_->NewIterator(read_options));
@@ -2040,10 +2041,10 @@ TEST_F(DBBloomFilterTest, DynamicBloomFilterNewColumnFamily) {
       ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_PREFIX_USEFUL), 0);
     }
     ASSERT_OK(dbfull()->DropColumnFamily(handles_[2]));
-    dbfull()->DestroyColumnFamilyHandle(handles_[2]);
+    ASSERT_OK(dbfull()->DestroyColumnFamilyHandle(handles_[2]));
     handles_[2] = nullptr;
     ASSERT_OK(dbfull()->DropColumnFamily(handles_[1]));
-    dbfull()->DestroyColumnFamilyHandle(handles_[1]);
+    ASSERT_OK(dbfull()->DestroyColumnFamilyHandle(handles_[1]));
     handles_[1] = nullptr;
     iteration++;
   }

db/db_compaction_filter_test.cc

@@ -42,7 +42,7 @@ class DBTestCompactionFilterWithCompactParam
         option_config_ == kUniversalSubcompactions) {
       assert(options.max_subcompactions > 1);
     }
-    TryReopen(options);
+    Reopen(options);
   }
 };
 
@@ -276,7 +276,7 @@ TEST_F(DBTestCompactionFilter, CompactionFilter) {
   for (int i = 0; i < 100000; i++) {
     char key[100];
     snprintf(key, sizeof(key), "B%010d", i);
-    Put(1, key, value);
+    ASSERT_OK(Put(1, key, value));
   }
   ASSERT_OK(Flush(1));
 
@@ -284,10 +284,10 @@ TEST_F(DBTestCompactionFilter, CompactionFilter) {
   // the compaction is each level invokes the filter for
   // all the keys in that level.
   cfilter_count = 0;
-  dbfull()->TEST_CompactRange(0, nullptr, nullptr, handles_[1]);
+  ASSERT_OK(dbfull()->TEST_CompactRange(0, nullptr, nullptr, handles_[1]));
   ASSERT_EQ(cfilter_count, 100000);
   cfilter_count = 0;
-  dbfull()->TEST_CompactRange(1, nullptr, nullptr, handles_[1]);
+  ASSERT_OK(dbfull()->TEST_CompactRange(1, nullptr, nullptr, handles_[1]));
   ASSERT_EQ(cfilter_count, 100000);
 
   ASSERT_EQ(NumTableFilesAtLevel(0, 1), 0);
@@ -321,6 +321,7 @@ TEST_F(DBTestCompactionFilter, CompactionFilter) {
       }
       iter->Next();
     }
+    ASSERT_OK(iter->status());
   }
   ASSERT_EQ(total, 100000);
   ASSERT_EQ(count, 0);
@@ -337,10 +338,10 @@ TEST_F(DBTestCompactionFilter, CompactionFilter) {
   // means that all keys should pass at least once
   // via the compaction filter
   cfilter_count = 0;
-  dbfull()->TEST_CompactRange(0, nullptr, nullptr, handles_[1]);
+  ASSERT_OK(dbfull()->TEST_CompactRange(0, nullptr, nullptr, handles_[1]));
   ASSERT_EQ(cfilter_count, 100000);
   cfilter_count = 0;
-  dbfull()->TEST_CompactRange(1, nullptr, nullptr, handles_[1]);
+  ASSERT_OK(dbfull()->TEST_CompactRange(1, nullptr, nullptr, handles_[1]));
   ASSERT_EQ(cfilter_count, 100000);
   ASSERT_EQ(NumTableFilesAtLevel(0, 1), 0);
   ASSERT_EQ(NumTableFilesAtLevel(1, 1), 0);
@@ -369,10 +370,10 @@ TEST_F(DBTestCompactionFilter, CompactionFilter) {
   // verify that at the end of the compaction process,
   // nothing is left.
   cfilter_count = 0;
-  dbfull()->TEST_CompactRange(0, nullptr, nullptr, handles_[1]);
+  ASSERT_OK(dbfull()->TEST_CompactRange(0, nullptr, nullptr, handles_[1]));
   ASSERT_EQ(cfilter_count, 100000);
   cfilter_count = 0;
-  dbfull()->TEST_CompactRange(1, nullptr, nullptr, handles_[1]);
+  ASSERT_OK(dbfull()->TEST_CompactRange(1, nullptr, nullptr, handles_[1]));
   ASSERT_EQ(cfilter_count, 0);
   ASSERT_EQ(NumTableFilesAtLevel(0, 1), 0);
   ASSERT_EQ(NumTableFilesAtLevel(1, 1), 0);
@@ -387,6 +388,7 @@ TEST_F(DBTestCompactionFilter, CompactionFilter) {
       count++;
       iter->Next();
     }
+    ASSERT_OK(iter->status());
     ASSERT_EQ(count, 0);
   }
 
@@ -427,9 +429,9 @@ TEST_F(DBTestCompactionFilter, CompactionFilterDeletesAll) {
   // put some data
   for (int table = 0; table < 4; ++table) {
     for (int i = 0; i < 10 + table; ++i) {
-      Put(ToString(table * 100 + i), "val");
+      ASSERT_OK(Put(ToString(table * 100 + i), "val"));
     }
-    Flush();
+    ASSERT_OK(Flush());
   }
 
   // this will produce empty file (delete compaction filter)
@@ -440,6 +442,7 @@ TEST_F(DBTestCompactionFilter, CompactionFilterDeletesAll) {
 
   Iterator* itr = db_->NewIterator(ReadOptions());
   itr->SeekToFirst();
+  ASSERT_OK(itr->status());
   // empty db
   ASSERT_TRUE(!itr->Valid());
 
@@ -463,25 +466,25 @@ TEST_P(DBTestCompactionFilterWithCompactParam,
   for (int i = 0; i < 100001; i++) {
     char key[100];
     snprintf(key, sizeof(key), "B%010d", i);
-    Put(1, key, value);
+    ASSERT_OK(Put(1, key, value));
   }
 
   // push all files to  lower levels
   ASSERT_OK(Flush(1));
   if (option_config_ != kUniversalCompactionMultiLevel &&
       option_config_ != kUniversalSubcompactions) {
-    dbfull()->TEST_CompactRange(0, nullptr, nullptr, handles_[1]);
-    dbfull()->TEST_CompactRange(1, nullptr, nullptr, handles_[1]);
+    ASSERT_OK(dbfull()->TEST_CompactRange(0, nullptr, nullptr, handles_[1]));
+    ASSERT_OK(dbfull()->TEST_CompactRange(1, nullptr, nullptr, handles_[1]));
   } else {
-    dbfull()->CompactRange(CompactRangeOptions(), handles_[1], nullptr,
-                           nullptr);
+    ASSERT_OK(dbfull()->CompactRange(CompactRangeOptions(), handles_[1],
+                                     nullptr, nullptr));
   }
 
   // re-write all data again
   for (int i = 0; i < 100001; i++) {
     char key[100];
     snprintf(key, sizeof(key), "B%010d", i);
-    Put(1, key, value);
+    ASSERT_OK(Put(1, key, value));
   }
 
   // push all files to  lower levels. This should
@@ -489,11 +492,11 @@ TEST_P(DBTestCompactionFilterWithCompactParam,
   ASSERT_OK(Flush(1));
   if (option_config_ != kUniversalCompactionMultiLevel &&
       option_config_ != kUniversalSubcompactions) {
-    dbfull()->TEST_CompactRange(0, nullptr, nullptr, handles_[1]);
-    dbfull()->TEST_CompactRange(1, nullptr, nullptr, handles_[1]);
+    ASSERT_OK(dbfull()->TEST_CompactRange(0, nullptr, nullptr, handles_[1]));
+    ASSERT_OK(dbfull()->TEST_CompactRange(1, nullptr, nullptr, handles_[1]));
   } else {
-    dbfull()->CompactRange(CompactRangeOptions(), handles_[1], nullptr,
-                           nullptr);
+    ASSERT_OK(dbfull()->CompactRange(CompactRangeOptions(), handles_[1],
+                                     nullptr, nullptr));
   }
 
   // verify that all keys now have the new value that
@@ -531,7 +534,7 @@ TEST_F(DBTestCompactionFilter, CompactionFilterWithMergeOperator) {
   ASSERT_OK(Flush());
   std::string newvalue = Get("foo");
   ASSERT_EQ(newvalue, three);
-  dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr);
+  ASSERT_OK(dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr));
   newvalue = Get("foo");
   ASSERT_EQ(newvalue, three);
 
@@ -539,12 +542,12 @@ TEST_F(DBTestCompactionFilter, CompactionFilterWithMergeOperator) {
   // merge keys.
   ASSERT_OK(db_->Put(WriteOptions(), "bar", two));
   ASSERT_OK(Flush());
-  dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr);
+  ASSERT_OK(dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr));
   newvalue = Get("bar");
   ASSERT_EQ("NOT_FOUND", newvalue);
   ASSERT_OK(db_->Merge(WriteOptions(), "bar", two));
   ASSERT_OK(Flush());
-  dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr);
+  ASSERT_OK(dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr));
   newvalue = Get("bar");
   ASSERT_EQ(two, two);
 
@@ -555,7 +558,7 @@ TEST_F(DBTestCompactionFilter, CompactionFilterWithMergeOperator) {
   ASSERT_OK(Flush());
   newvalue = Get("foobar");
   ASSERT_EQ(newvalue, three);
-  dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr);
+  ASSERT_OK(dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr));
   newvalue = Get("foobar");
   ASSERT_EQ(newvalue, three);
 
@@ -568,7 +571,7 @@ TEST_F(DBTestCompactionFilter, CompactionFilterWithMergeOperator) {
   ASSERT_OK(Flush());
   newvalue = Get("barfoo");
   ASSERT_EQ(newvalue, four);
-  dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr);
+  ASSERT_OK(dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr));
   newvalue = Get("barfoo");
   ASSERT_EQ(newvalue, four);
 }
@@ -590,21 +593,21 @@ TEST_F(DBTestCompactionFilter, CompactionFilterContextManual) {
     for (int i = 0; i < num_keys_per_file; i++) {
       char key[100];
       snprintf(key, sizeof(key), "B%08d%02d", i, j);
-      Put(key, value);
+      ASSERT_OK(Put(key, value));
     }
-    dbfull()->TEST_FlushMemTable();
+    ASSERT_OK(dbfull()->TEST_FlushMemTable());
     // Make sure next file is much smaller so automatic compaction will not
     // be triggered.
     num_keys_per_file /= 2;
   }
-  dbfull()->TEST_WaitForCompact();
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
 
   // Force a manual compaction
   cfilter_count = 0;
   filter->expect_manual_compaction_.store(true);
   filter->expect_full_compaction_.store(true);
   filter->expect_cf_id_.store(0);
-  dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr);
+  ASSERT_OK(dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr));
   ASSERT_EQ(cfilter_count, 700);
   ASSERT_EQ(NumSortedRuns(0), 1);
   ASSERT_TRUE(filter->compaction_filter_created());
@@ -654,14 +657,14 @@ TEST_F(DBTestCompactionFilter, CompactionFilterContextCfId) {
     for (int i = 0; i < num_keys_per_file; i++) {
       char key[100];
       snprintf(key, sizeof(key), "B%08d%02d", i, j);
-      Put(1, key, value);
+      ASSERT_OK(Put(1, key, value));
     }
-    Flush(1);
+    ASSERT_OK(Flush(1));
     // Make sure next file is much smaller so automatic compaction will not
     // be triggered.
     num_keys_per_file /= 2;
   }
-  dbfull()->TEST_WaitForCompact();
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
 
   ASSERT_TRUE(filter->compaction_filter_created());
 }
@@ -680,9 +683,9 @@ TEST_F(DBTestCompactionFilter, CompactionFilterIgnoreSnapshot) {
   const Snapshot* snapshot = nullptr;
   for (int table = 0; table < 4; ++table) {
     for (int i = 0; i < 10; ++i) {
-      Put(ToString(table * 100 + i), "val");
+      ASSERT_OK(Put(ToString(table * 100 + i), "val"));
     }
-    Flush();
+    ASSERT_OK(Flush());
 
     if (table == 0) {
       snapshot = db_->GetSnapshot();
@@ -702,6 +705,7 @@ TEST_F(DBTestCompactionFilter, CompactionFilterIgnoreSnapshot) {
     read_options.snapshot = snapshot;
     std::unique_ptr<Iterator> iter(db_->NewIterator(read_options));
     iter->SeekToFirst();
+    ASSERT_OK(iter->status());
     int count = 0;
     while (iter->Valid()) {
       count++;
@@ -710,6 +714,7 @@ TEST_F(DBTestCompactionFilter, CompactionFilterIgnoreSnapshot) {
     ASSERT_EQ(count, 6);
     read_options.snapshot = nullptr;
     std::unique_ptr<Iterator> iter1(db_->NewIterator(read_options));
+    ASSERT_OK(iter1->status());
     iter1->SeekToFirst();
     count = 0;
     while (iter1->Valid()) {
@@ -740,9 +745,9 @@ TEST_F(DBTestCompactionFilter, SkipUntil) {
     for (int i = table * 6; i < 39 + table * 11; ++i) {
       char key[100];
       snprintf(key, sizeof(key), "%010d", table * 100 + i);
-      Put(key, std::to_string(table * 1000 + i));
+      ASSERT_OK(Put(key, std::to_string(table * 1000 + i)));
     }
-    Flush();
+    ASSERT_OK(Flush());
   }
 
   cfilter_skips = 0;
@@ -781,10 +786,10 @@ TEST_F(DBTestCompactionFilter, SkipUntilWithBloomFilter) {
   options.create_if_missing = true;
   DestroyAndReopen(options);
 
-  Put("0000000010", "v10");
-  Put("0000000020", "v20");  // skipped
-  Put("0000000050", "v50");
-  Flush();
+  ASSERT_OK(Put("0000000010", "v10"));
+  ASSERT_OK(Put("0000000020", "v20"));  // skipped
+  ASSERT_OK(Put("0000000050", "v50"));
+  ASSERT_OK(Flush());
 
   cfilter_skips = 0;
   EXPECT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
@@ -822,13 +827,13 @@ TEST_F(DBTestCompactionFilter, IgnoreSnapshotsFalse) {
   options.compaction_filter = new TestNotSupportedFilter();
   DestroyAndReopen(options);
 
-  Put("a", "v10");
-  Put("z", "v20");
-  Flush();
+  ASSERT_OK(Put("a", "v10"));
+  ASSERT_OK(Put("z", "v20"));
+  ASSERT_OK(Flush());
 
-  Put("a", "v10");
-  Put("z", "v20");
-  Flush();
+  ASSERT_OK(Put("a", "v10"));
+  ASSERT_OK(Put("z", "v20"));
+  ASSERT_OK(Flush());
 
   // Comapction should fail because IgnoreSnapshots() = false
   EXPECT_TRUE(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr)

db/db_dynamic_level_test.cc

@@ -102,7 +102,8 @@ TEST_F(DBTestDynamicLevel, DynamicLevelMaxBytesBase) {
       }
 
       // Test compact range works
-      dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr);
+      ASSERT_OK(
+          dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr));
       // All data should be in the last level.
       ColumnFamilyMetaData cf_meta;
       db_->GetColumnFamilyMetaData(&cf_meta);
@@ -166,8 +167,8 @@ TEST_F(DBTestDynamicLevel, DynamicLevelMaxBytesBase2) {
   ASSERT_OK(dbfull()->SetOptions({
       {"disable_auto_compactions", "false"},
   }));
-  Flush();
-  dbfull()->TEST_WaitForCompact();
+  ASSERT_OK(Flush());
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
   ASSERT_TRUE(db_->GetIntProperty("rocksdb.base-level", &int_prop));
   ASSERT_EQ(4U, int_prop);
 
@@ -184,8 +185,8 @@ TEST_F(DBTestDynamicLevel, DynamicLevelMaxBytesBase2) {
   ASSERT_OK(dbfull()->SetOptions({
       {"disable_auto_compactions", "false"},
   }));
-  Flush();
-  dbfull()->TEST_WaitForCompact();
+  ASSERT_OK(Flush());
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
   ASSERT_TRUE(db_->GetIntProperty("rocksdb.base-level", &int_prop));
   ASSERT_EQ(3U, int_prop);
   ASSERT_TRUE(db_->GetProperty("rocksdb.num-files-at-level1", &str_prop));
@@ -205,8 +206,8 @@ TEST_F(DBTestDynamicLevel, DynamicLevelMaxBytesBase2) {
   ASSERT_OK(dbfull()->SetOptions({
       {"disable_auto_compactions", "false"},
   }));
-  Flush();
-  dbfull()->TEST_WaitForCompact();
+  ASSERT_OK(Flush());
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
   ASSERT_TRUE(db_->GetIntProperty("rocksdb.base-level", &int_prop));
   ASSERT_EQ(3U, int_prop);
 
@@ -234,8 +235,8 @@ TEST_F(DBTestDynamicLevel, DynamicLevelMaxBytesBase2) {
   }));
 
   TEST_SYNC_POINT("DynamicLevelMaxBytesBase2:0");
-  Flush();
-  dbfull()->TEST_WaitForCompact();
+  ASSERT_OK(Flush());
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
   ASSERT_TRUE(db_->GetIntProperty("rocksdb.base-level", &int_prop));
   ASSERT_EQ(2U, int_prop);
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
@@ -264,7 +265,7 @@ TEST_F(DBTestDynamicLevel, DynamicLevelMaxBytesBase2) {
   }
   TEST_SYNC_POINT("DynamicLevelMaxBytesBase2:2");
 
-  Flush();
+  ASSERT_OK(Flush());
 
   thread.join();
 
@@ -302,7 +303,7 @@ TEST_F(DBTestDynamicLevel, DynamicLevelMaxBytesCompactRange) {
   DestroyAndReopen(options);
 
   // Compact against empty DB
-  dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr);
+  ASSERT_OK(dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr));
 
   uint64_t int_prop;
   std::string str_prop;
@@ -316,13 +317,13 @@ TEST_F(DBTestDynamicLevel, DynamicLevelMaxBytesCompactRange) {
     ASSERT_OK(
         Put(Key(static_cast<int>(rnd.Uniform(kMaxKey))), rnd.RandomString(80)));
   }
-  Flush();
-  dbfull()->TEST_WaitForCompact();
+  ASSERT_OK(Flush());
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
   if (NumTableFilesAtLevel(0) == 0) {
     // Make sure level 0 is not empty
     ASSERT_OK(
         Put(Key(static_cast<int>(rnd.Uniform(kMaxKey))), rnd.RandomString(80)));
-    Flush();
+    ASSERT_OK(Flush());
   }
 
   ASSERT_TRUE(db_->GetIntProperty("rocksdb.base-level", &int_prop));
@@ -343,7 +344,7 @@ TEST_F(DBTestDynamicLevel, DynamicLevelMaxBytesCompactRange) {
       });
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
 
-  dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr);
+  ASSERT_OK(dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr));
   ASSERT_EQ(output_levels.size(), 2);
   ASSERT_TRUE(output_levels.find(3) != output_levels.end());
   ASSERT_TRUE(output_levels.find(4) != output_levels.end());
@@ -389,8 +390,8 @@ TEST_F(DBTestDynamicLevel, DynamicLevelMaxBytesBaseInc) {
     PutFixed32(&value, static_cast<uint32_t>(i));
     ASSERT_OK(Put(Key(i), value));
   }
-  Flush();
-  dbfull()->TEST_WaitForCompact();
+  ASSERT_OK(Flush());
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
 
   ASSERT_EQ(non_trivial, 0);
@@ -449,7 +450,7 @@ TEST_F(DBTestDynamicLevel, DISABLED_MigrateToDynamicLevelMaxBytesBase) {
     ASSERT_OK(Delete(Key(i / 10)));
   }
   verify_func(total_keys, false);
-  dbfull()->TEST_WaitForCompact();
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
 
   options.level_compaction_dynamic_level_bytes = true;
   options.disable_auto_compactions = true;
@@ -464,7 +465,7 @@ TEST_F(DBTestDynamicLevel, DISABLED_MigrateToDynamicLevelMaxBytesBase) {
     CompactRangeOptions compact_options;
     compact_options.change_level = true;
     compact_options.target_level = options.num_levels - 1;
-    dbfull()->CompactRange(compact_options, nullptr, nullptr);
+    ASSERT_OK(dbfull()->CompactRange(compact_options, nullptr, nullptr));
     compaction_finished.store(true);
   });
   do {
@@ -484,7 +485,7 @@ TEST_F(DBTestDynamicLevel, DISABLED_MigrateToDynamicLevelMaxBytesBase) {
   }
 
   verify_func(total_keys2, false);
-  dbfull()->TEST_WaitForCompact();
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
   verify_func(total_keys2, false);
 
   // Base level is not level 1

db/db_encryption_test.cc

@@ -44,7 +44,7 @@ TEST_F(DBEncryptionTest, CheckEncrypted) {
   Env* target = GetTargetEnv();
   int hits = 0;
   for (auto it = fileNames.begin() ; it != fileNames.end(); ++it) {
-    if ((*it == "..") || (*it == ".") || (*it == "LOCK")) {
+    if (*it == "LOCK") {
       continue;
     }
     auto filePath = dbname_ + "/" + *it;

db/db_flush_test.cc

@@ -14,10 +14,12 @@
 #include "file/filename.h"
 #include "port/port.h"
 #include "port/stack_trace.h"
+#include "rocksdb/utilities/transaction_db.h"
 #include "test_util/sync_point.h"
 #include "util/cast_util.h"
 #include "util/mutexlock.h"
 #include "utilities/fault_injection_env.h"
+#include "utilities/fault_injection_fs.h"
 
 namespace ROCKSDB_NAMESPACE {
 
@@ -450,7 +452,6 @@ TEST_F(DBFlushTest, FlushWithBlob) {
   constexpr uint64_t min_blob_size = 10;
 
   Options options;
-  options.env = CurrentOptions().env;
   options.enable_blob_files = true;
   options.min_blob_size = min_blob_size;
   options.disable_auto_compactions = true;
@@ -524,14 +525,209 @@ TEST_F(DBFlushTest, FlushWithBlob) {
 #endif  // ROCKSDB_LITE
 }
 
+TEST_F(DBFlushTest, FlushWithChecksumHandoff1) {
+  if (mem_env_ || encrypted_env_) {
+    ROCKSDB_GTEST_SKIP("Test requires non-mem or non-encrypted environment");
+    return;
+  }
+  std::shared_ptr<FaultInjectionTestFS> fault_fs(
+      new FaultInjectionTestFS(FileSystem::Default()));
+  std::unique_ptr<Env> fault_fs_env(NewCompositeEnv(fault_fs));
+  Options options = CurrentOptions();
+  options.write_buffer_size = 100;
+  options.max_write_buffer_number = 4;
+  options.min_write_buffer_number_to_merge = 3;
+  options.disable_auto_compactions = true;
+  options.env = fault_fs_env.get();
+  options.checksum_handoff_file_types.Add(FileType::kTableFile);
+  Reopen(options);
+
+  fault_fs->SetChecksumHandoffFuncType(ChecksumType::kCRC32c);
+  ASSERT_OK(Put("key1", "value1"));
+  ASSERT_OK(Put("key2", "value2"));
+  ASSERT_OK(dbfull()->TEST_SwitchMemtable());
+
+  // The hash does not match, write fails
+  // fault_fs->SetChecksumHandoffFuncType(ChecksumType::kxxHash);
+  // Since the file system returns IOStatus::Corruption, it is an
+  // unrecoverable error.
+  SyncPoint::GetInstance()->SetCallBack("FlushJob::Start", [&](void*) {
+    fault_fs->SetChecksumHandoffFuncType(ChecksumType::kxxHash);
+  });
+  ASSERT_OK(Put("key3", "value3"));
+  ASSERT_OK(Put("key4", "value4"));
+  SyncPoint::GetInstance()->EnableProcessing();
+  Status s = Flush();
+  ASSERT_EQ(s.severity(),
+            ROCKSDB_NAMESPACE::Status::Severity::kUnrecoverableError);
+  SyncPoint::GetInstance()->DisableProcessing();
+  Destroy(options);
+  Reopen(options);
+
+  // The file system does not support checksum handoff. The check
+  // will be ignored.
+  fault_fs->SetChecksumHandoffFuncType(ChecksumType::kNoChecksum);
+  ASSERT_OK(Put("key5", "value5"));
+  ASSERT_OK(Put("key6", "value6"));
+  ASSERT_OK(dbfull()->TEST_SwitchMemtable());
+
+  // Each write will be similated as corrupted.
+  // Since the file system returns IOStatus::Corruption, it is an
+  // unrecoverable error.
+  fault_fs->SetChecksumHandoffFuncType(ChecksumType::kCRC32c);
+  SyncPoint::GetInstance()->SetCallBack("FlushJob::Start", [&](void*) {
+    fault_fs->IngestDataCorruptionBeforeWrite();
+  });
+  ASSERT_OK(Put("key7", "value7"));
+  ASSERT_OK(Put("key8", "value8"));
+  SyncPoint::GetInstance()->EnableProcessing();
+  s = Flush();
+  ASSERT_EQ(s.severity(),
+            ROCKSDB_NAMESPACE::Status::Severity::kUnrecoverableError);
+  SyncPoint::GetInstance()->DisableProcessing();
+
+  Destroy(options);
+}
+
+TEST_F(DBFlushTest, FlushWithChecksumHandoff2) {
+  if (mem_env_ || encrypted_env_) {
+    ROCKSDB_GTEST_SKIP("Test requires non-mem or non-encrypted environment");
+    return;
+  }
+  std::shared_ptr<FaultInjectionTestFS> fault_fs(
+      new FaultInjectionTestFS(FileSystem::Default()));
+  std::unique_ptr<Env> fault_fs_env(NewCompositeEnv(fault_fs));
+  Options options = CurrentOptions();
+  options.write_buffer_size = 100;
+  options.max_write_buffer_number = 4;
+  options.min_write_buffer_number_to_merge = 3;
+  options.disable_auto_compactions = true;
+  options.env = fault_fs_env.get();
+  Reopen(options);
+
+  fault_fs->SetChecksumHandoffFuncType(ChecksumType::kCRC32c);
+  ASSERT_OK(Put("key1", "value1"));
+  ASSERT_OK(Put("key2", "value2"));
+  ASSERT_OK(Flush());
+
+  // options is not set, the checksum handoff will not be triggered
+  SyncPoint::GetInstance()->SetCallBack("FlushJob::Start", [&](void*) {
+    fault_fs->SetChecksumHandoffFuncType(ChecksumType::kxxHash);
+  });
+  ASSERT_OK(Put("key3", "value3"));
+  ASSERT_OK(Put("key4", "value4"));
+  SyncPoint::GetInstance()->EnableProcessing();
+  ASSERT_OK(Flush());
+  SyncPoint::GetInstance()->DisableProcessing();
+  Destroy(options);
+  Reopen(options);
+
+  // The file system does not support checksum handoff. The check
+  // will be ignored.
+  fault_fs->SetChecksumHandoffFuncType(ChecksumType::kNoChecksum);
+  ASSERT_OK(Put("key5", "value5"));
+  ASSERT_OK(Put("key6", "value6"));
+  ASSERT_OK(Flush());
+
+  // options is not set, the checksum handoff will not be triggered
+  fault_fs->SetChecksumHandoffFuncType(ChecksumType::kCRC32c);
+  SyncPoint::GetInstance()->SetCallBack("FlushJob::Start", [&](void*) {
+    fault_fs->IngestDataCorruptionBeforeWrite();
+  });
+  ASSERT_OK(Put("key7", "value7"));
+  ASSERT_OK(Put("key8", "value8"));
+  SyncPoint::GetInstance()->EnableProcessing();
+  ASSERT_OK(Flush());
+  SyncPoint::GetInstance()->DisableProcessing();
+
+  Destroy(options);
+}
+
+TEST_F(DBFlushTest, FlushWithChecksumHandoffManifest1) {
+  if (mem_env_ || encrypted_env_) {
+    ROCKSDB_GTEST_SKIP("Test requires non-mem or non-encrypted environment");
+    return;
+  }
+  std::shared_ptr<FaultInjectionTestFS> fault_fs(
+      new FaultInjectionTestFS(FileSystem::Default()));
+  std::unique_ptr<Env> fault_fs_env(NewCompositeEnv(fault_fs));
+  Options options = CurrentOptions();
+  options.write_buffer_size = 100;
+  options.max_write_buffer_number = 4;
+  options.min_write_buffer_number_to_merge = 3;
+  options.disable_auto_compactions = true;
+  options.env = fault_fs_env.get();
+  options.checksum_handoff_file_types.Add(FileType::kDescriptorFile);
+  fault_fs->SetChecksumHandoffFuncType(ChecksumType::kCRC32c);
+  Reopen(options);
+
+  ASSERT_OK(Put("key1", "value1"));
+  ASSERT_OK(Put("key2", "value2"));
+  ASSERT_OK(Flush());
+
+  // The hash does not match, write fails
+  // fault_fs->SetChecksumHandoffFuncType(ChecksumType::kxxHash);
+  // Since the file system returns IOStatus::Corruption, it is mapped to
+  // kFatalError error.
+  ASSERT_OK(Put("key3", "value3"));
+  SyncPoint::GetInstance()->SetCallBack(
+      "VersionSet::LogAndApply:WriteManifest", [&](void*) {
+        fault_fs->SetChecksumHandoffFuncType(ChecksumType::kxxHash);
+      });
+  ASSERT_OK(Put("key3", "value3"));
+  ASSERT_OK(Put("key4", "value4"));
+  SyncPoint::GetInstance()->EnableProcessing();
+  Status s = Flush();
+  ASSERT_EQ(s.severity(), ROCKSDB_NAMESPACE::Status::Severity::kFatalError);
+  SyncPoint::GetInstance()->DisableProcessing();
+  Destroy(options);
+}
+
+TEST_F(DBFlushTest, FlushWithChecksumHandoffManifest2) {
+  if (mem_env_ || encrypted_env_) {
+    ROCKSDB_GTEST_SKIP("Test requires non-mem or non-encrypted environment");
+    return;
+  }
+  std::shared_ptr<FaultInjectionTestFS> fault_fs(
+      new FaultInjectionTestFS(FileSystem::Default()));
+  std::unique_ptr<Env> fault_fs_env(NewCompositeEnv(fault_fs));
+  Options options = CurrentOptions();
+  options.write_buffer_size = 100;
+  options.max_write_buffer_number = 4;
+  options.min_write_buffer_number_to_merge = 3;
+  options.disable_auto_compactions = true;
+  options.env = fault_fs_env.get();
+  options.checksum_handoff_file_types.Add(FileType::kDescriptorFile);
+  fault_fs->SetChecksumHandoffFuncType(ChecksumType::kNoChecksum);
+  Reopen(options);
+  // The file system does not support checksum handoff. The check
+  // will be ignored.
+  ASSERT_OK(Put("key5", "value5"));
+  ASSERT_OK(Put("key6", "value6"));
+  ASSERT_OK(Flush());
+
+  // Each write will be similated as corrupted.
+  // Since the file system returns IOStatus::Corruption, it is mapped to
+  // kFatalError error.
+  fault_fs->SetChecksumHandoffFuncType(ChecksumType::kCRC32c);
+  SyncPoint::GetInstance()->SetCallBack(
+      "VersionSet::LogAndApply:WriteManifest",
+      [&](void*) { fault_fs->IngestDataCorruptionBeforeWrite(); });
+  ASSERT_OK(Put("key7", "value7"));
+  ASSERT_OK(Put("key8", "value8"));
+  SyncPoint::GetInstance()->EnableProcessing();
+  Status s = Flush();
+  ASSERT_EQ(s.severity(), ROCKSDB_NAMESPACE::Status::Severity::kFatalError);
+  SyncPoint::GetInstance()->DisableProcessing();
+
+  Destroy(options);
+}
+
 class DBFlushTestBlobError : public DBFlushTest,
                              public testing::WithParamInterface<std::string> {
  public:
-  DBFlushTestBlobError()
-      : fault_injection_env_(env_), sync_point_(GetParam()) {}
-  ~DBFlushTestBlobError() { Close(); }
+  DBFlushTestBlobError() : sync_point_(GetParam()) {}
 
-  FaultInjectionTestEnv fault_injection_env_;
   std::string sync_point_;
 };
 
@@ -544,20 +740,18 @@ TEST_P(DBFlushTestBlobError, FlushError) {
   Options options;
   options.enable_blob_files = true;
   options.disable_auto_compactions = true;
-  options.env = &fault_injection_env_;
+  options.env = env_;
 
   Reopen(options);
 
   ASSERT_OK(Put("key", "blob"));
 
-  SyncPoint::GetInstance()->SetCallBack(sync_point_, [this](void* /* arg */) {
-    fault_injection_env_.SetFilesystemActive(false,
-                                             Status::IOError(sync_point_));
+  SyncPoint::GetInstance()->SetCallBack(sync_point_, [this](void* arg) {
+    Status* const s = static_cast<Status*>(arg);
+    assert(s);
+
+    (*s) = Status::IOError(sync_point_);
   });
-  SyncPoint::GetInstance()->SetCallBack(
-      "BuildTable:BeforeDeleteFile", [this](void* /* arg */) {
-        fault_injection_env_.SetFilesystemActive(true);
-      });
   SyncPoint::GetInstance()->EnableProcessing();
 
   ASSERT_NOK(Flush());
@@ -620,6 +814,96 @@ TEST_P(DBFlushTestBlobError, FlushError) {
 #endif  // ROCKSDB_LITE
 }
 
+#ifndef ROCKSDB_LITE
+TEST_P(DBAtomicFlushTest, ManualFlushUnder2PC) {
+  Options options = CurrentOptions();
+  options.create_if_missing = true;
+  options.allow_2pc = true;
+  options.atomic_flush = GetParam();
+  // 64MB so that memtable flush won't be trigger by the small writes.
+  options.write_buffer_size = (static_cast<size_t>(64) << 20);
+
+  // Destroy the DB to recreate as a TransactionDB.
+  Close();
+  Destroy(options, true);
+
+  // Create a TransactionDB.
+  TransactionDB* txn_db = nullptr;
+  TransactionDBOptions txn_db_opts;
+  txn_db_opts.write_policy = TxnDBWritePolicy::WRITE_COMMITTED;
+  ASSERT_OK(TransactionDB::Open(options, txn_db_opts, dbname_, &txn_db));
+  ASSERT_NE(txn_db, nullptr);
+  db_ = txn_db;
+
+  // Create two more columns other than default CF.
+  std::vector<std::string> cfs = {"puppy", "kitty"};
+  CreateColumnFamilies(cfs, options);
+  ASSERT_EQ(handles_.size(), 2);
+  ASSERT_EQ(handles_[0]->GetName(), cfs[0]);
+  ASSERT_EQ(handles_[1]->GetName(), cfs[1]);
+  const size_t kNumCfToFlush = options.atomic_flush ? 2 : 1;
+
+  WriteOptions wopts;
+  TransactionOptions txn_opts;
+  // txn1 only prepare, but does not commit.
+  // The WAL containing the prepared but uncommitted data must be kept.
+  Transaction* txn1 = txn_db->BeginTransaction(wopts, txn_opts, nullptr);
+  // txn2 not only prepare, but also commit.
+  Transaction* txn2 = txn_db->BeginTransaction(wopts, txn_opts, nullptr);
+  ASSERT_NE(txn1, nullptr);
+  ASSERT_NE(txn2, nullptr);
+  for (size_t i = 0; i < kNumCfToFlush; i++) {
+    ASSERT_OK(txn1->Put(handles_[i], "k1", "v1"));
+    ASSERT_OK(txn2->Put(handles_[i], "k2", "v2"));
+  }
+  // A txn must be named before prepare.
+  ASSERT_OK(txn1->SetName("txn1"));
+  ASSERT_OK(txn2->SetName("txn2"));
+  // Prepare writes to WAL, but not to memtable. (WriteCommitted)
+  ASSERT_OK(txn1->Prepare());
+  ASSERT_OK(txn2->Prepare());
+  // Commit writes to memtable.
+  ASSERT_OK(txn2->Commit());
+  delete txn1;
+  delete txn2;
+
+  // There are still data in memtable not flushed.
+  // But since data is small enough to reside in the active memtable,
+  // there are no immutable memtable.
+  for (size_t i = 0; i < kNumCfToFlush; i++) {
+    auto cfh = static_cast<ColumnFamilyHandleImpl*>(handles_[i]);
+    ASSERT_EQ(0, cfh->cfd()->imm()->NumNotFlushed());
+    ASSERT_FALSE(cfh->cfd()->mem()->IsEmpty());
+  }
+
+  // Atomic flush memtables,
+  // the min log with prepared data should be written to MANIFEST.
+  std::vector<ColumnFamilyHandle*> cfs_to_flush(kNumCfToFlush);
+  for (size_t i = 0; i < kNumCfToFlush; i++) {
+    cfs_to_flush[i] = handles_[i];
+  }
+  ASSERT_OK(txn_db->Flush(FlushOptions(), cfs_to_flush));
+
+  // There are no remaining data in memtable after flush.
+  for (size_t i = 0; i < kNumCfToFlush; i++) {
+    auto cfh = static_cast<ColumnFamilyHandleImpl*>(handles_[i]);
+    ASSERT_EQ(0, cfh->cfd()->imm()->NumNotFlushed());
+    ASSERT_TRUE(cfh->cfd()->mem()->IsEmpty());
+    ASSERT_EQ(cfh->cfd()->GetFlushReason(), FlushReason::kManualFlush);
+  }
+
+  // The recovered min log number with prepared data should be non-zero.
+  // In 2pc mode, MinLogNumberToKeep returns the
+  // VersionSet::min_log_number_to_keep_2pc recovered from MANIFEST, if it's 0,
+  // it means atomic flush didn't write the min_log_number_to_keep to MANIFEST.
+  cfs.push_back(kDefaultColumnFamilyName);
+  ASSERT_OK(TryReopenWithColumnFamilies(cfs, options));
+  DBImpl* db_impl = reinterpret_cast<DBImpl*>(db_);
+  ASSERT_TRUE(db_impl->allow_2pc());
+  ASSERT_NE(db_impl->MinLogNumberToKeep(), 0);
+}
+#endif  // ROCKSDB_LITE
+
 TEST_P(DBAtomicFlushTest, ManualAtomicFlush) {
   Options options = CurrentOptions();
   options.create_if_missing = true;
@@ -634,18 +918,84 @@ TEST_P(DBAtomicFlushTest, ManualAtomicFlush) {
   for (size_t i = 0; i != num_cfs; ++i) {
     ASSERT_OK(Put(static_cast<int>(i) /*cf*/, "key", "value", wopts));
   }
+
+  for (size_t i = 0; i != num_cfs; ++i) {
+    auto cfh = static_cast<ColumnFamilyHandleImpl*>(handles_[i]);
+    ASSERT_EQ(0, cfh->cfd()->imm()->NumNotFlushed());
+    ASSERT_FALSE(cfh->cfd()->mem()->IsEmpty());
+  }
+
   std::vector<int> cf_ids;
   for (size_t i = 0; i != num_cfs; ++i) {
     cf_ids.emplace_back(static_cast<int>(i));
   }
   ASSERT_OK(Flush(cf_ids));
+
   for (size_t i = 0; i != num_cfs; ++i) {
     auto cfh = static_cast<ColumnFamilyHandleImpl*>(handles_[i]);
+    ASSERT_EQ(cfh->cfd()->GetFlushReason(), FlushReason::kManualFlush);
     ASSERT_EQ(0, cfh->cfd()->imm()->NumNotFlushed());
     ASSERT_TRUE(cfh->cfd()->mem()->IsEmpty());
   }
 }
 
+TEST_P(DBAtomicFlushTest, PrecomputeMinLogNumberToKeepNon2PC) {
+  Options options = CurrentOptions();
+  options.create_if_missing = true;
+  options.atomic_flush = GetParam();
+  options.write_buffer_size = (static_cast<size_t>(64) << 20);
+  CreateAndReopenWithCF({"pikachu"}, options);
+
+  const size_t num_cfs = handles_.size();
+  ASSERT_EQ(num_cfs, 2);
+  WriteOptions wopts;
+  for (size_t i = 0; i != num_cfs; ++i) {
+    ASSERT_OK(Put(static_cast<int>(i) /*cf*/, "key", "value", wopts));
+  }
+
+  {
+    // Flush the default CF only.
+    std::vector<int> cf_ids{0};
+    ASSERT_OK(Flush(cf_ids));
+
+    autovector<ColumnFamilyData*> flushed_cfds;
+    autovector<autovector<VersionEdit*>> flush_edits;
+    auto flushed_cfh = static_cast<ColumnFamilyHandleImpl*>(handles_[0]);
+    flushed_cfds.push_back(flushed_cfh->cfd());
+    flush_edits.push_back({});
+    auto unflushed_cfh = static_cast<ColumnFamilyHandleImpl*>(handles_[1]);
+
+    ASSERT_EQ(PrecomputeMinLogNumberToKeepNon2PC(dbfull()->TEST_GetVersionSet(),
+                                                 flushed_cfds, flush_edits),
+              unflushed_cfh->cfd()->GetLogNumber());
+  }
+
+  {
+    // Flush all CFs.
+    std::vector<int> cf_ids;
+    for (size_t i = 0; i != num_cfs; ++i) {
+      cf_ids.emplace_back(static_cast<int>(i));
+    }
+    ASSERT_OK(Flush(cf_ids));
+    uint64_t log_num_after_flush = dbfull()->TEST_GetCurrentLogNumber();
+
+    uint64_t min_log_number_to_keep = port::kMaxUint64;
+    autovector<ColumnFamilyData*> flushed_cfds;
+    autovector<autovector<VersionEdit*>> flush_edits;
+    for (size_t i = 0; i != num_cfs; ++i) {
+      auto cfh = static_cast<ColumnFamilyHandleImpl*>(handles_[i]);
+      flushed_cfds.push_back(cfh->cfd());
+      flush_edits.push_back({});
+      min_log_number_to_keep =
+          std::min(min_log_number_to_keep, cfh->cfd()->GetLogNumber());
+    }
+    ASSERT_EQ(min_log_number_to_keep, log_num_after_flush);
+    ASSERT_EQ(PrecomputeMinLogNumberToKeepNon2PC(dbfull()->TEST_GetVersionSet(),
+                                                 flushed_cfds, flush_edits),
+              min_log_number_to_keep);
+  }
+}
+
 TEST_P(DBAtomicFlushTest, AtomicFlushTriggeredByMemTableFull) {
   Options options = CurrentOptions();
   options.create_if_missing = true;

db/db_impl/db_impl.cc

@@ -53,7 +53,6 @@
 #include "db/version_set.h"
 #include "db/write_batch_internal.h"
 #include "db/write_callback.h"
-#include "env/composite_env_wrapper.h"
 #include "file/file_util.h"
 #include "file/filename.h"
 #include "file/random_access_file_reader.h"
@@ -83,6 +82,7 @@
 #include "rocksdb/stats_history.h"
 #include "rocksdb/status.h"
 #include "rocksdb/table.h"
+#include "rocksdb/version.h"
 #include "rocksdb/write_buffer_manager.h"
 #include "table/block_based/block.h"
 #include "table/block_based/block_based_table_factory.h"
@@ -94,7 +94,6 @@
 #include "table/two_level_iterator.h"
 #include "test_util/sync_point.h"
 #include "util/autovector.h"
-#include "util/build_version.h"
 #include "util/cast_util.h"
 #include "util/coding.h"
 #include "util/compression.h"
@@ -307,18 +306,21 @@ Status DBImpl::ResumeImpl(DBRecoverContext context) {
   mutex_.AssertHeld();
   WaitForBackgroundWork();
 
-  Status bg_error = error_handler_.GetBGError();
   Status s;
   if (shutdown_initiated_) {
     // Returning shutdown status to SFM during auto recovery will cause it
     // to abort the recovery and allow the shutdown to progress
     s = Status::ShutdownInProgress();
   }
-  if (s.ok() && bg_error.severity() > Status::Severity::kHardError) {
-    ROCKS_LOG_INFO(
-        immutable_db_options_.info_log,
-        "DB resume requested but failed due to Fatal/Unrecoverable error");
-    s = bg_error;
+
+  if (s.ok()) {
+    Status bg_error = error_handler_.GetBGError();
+    if (bg_error.severity() > Status::Severity::kHardError) {
+      ROCKS_LOG_INFO(
+          immutable_db_options_.info_log,
+          "DB resume requested but failed due to Fatal/Unrecoverable error");
+      s = bg_error;
+    }
   }
 
   // Make sure the IO Status stored in version set is set to OK.
@@ -393,6 +395,11 @@ Status DBImpl::ResumeImpl(DBRecoverContext context) {
   FindObsoleteFiles(&job_context, true);
   if (s.ok()) {
     s = error_handler_.ClearBGError();
+  } else {
+    // NOTE: this is needed to pass ASSERT_STATUS_CHECKED
+    // in the DBSSTTest.DBWithMaxSpaceAllowedRandomized test.
+    // See https://github.com/facebook/rocksdb/pull/7715#issuecomment-754947952
+    error_handler_.GetRecoveryError().PermitUncheckedError();
   }
   mutex_.Unlock();
 
@@ -409,6 +416,12 @@ Status DBImpl::ResumeImpl(DBRecoverContext context) {
     if (file_deletion_disabled) {
       // Always return ok
       s = EnableFileDeletions(/*force=*/true);
+      if (!s.ok()) {
+        ROCKS_LOG_INFO(
+            immutable_db_options_.info_log,
+            "DB resume requested but could not enable file deletions [%s]",
+            s.ToString().c_str());
+      }
     }
     ROCKS_LOG_INFO(immutable_db_options_.info_log, "Successfully resumed DB");
   }
@@ -690,6 +703,18 @@ void DBImpl::PrintStatistics() {
 
 void DBImpl::StartPeriodicWorkScheduler() {
 #ifndef ROCKSDB_LITE
+
+#ifndef NDEBUG
+  // It only used by test to disable scheduler
+  bool disable_scheduler = false;
+  TEST_SYNC_POINT_CALLBACK(
+      "DBImpl::StartPeriodicWorkScheduler:DisableScheduler",
+      &disable_scheduler);
+  if (disable_scheduler) {
+    return;
+  }
+#endif  // !NDEBUG
+
   {
     InstrumentedMutexLock l(&mutex_);
     periodic_work_scheduler_ = PeriodicWorkScheduler::Default();
@@ -1323,7 +1348,8 @@ Status DBImpl::MarkLogsSynced(uint64_t up_to, bool synced_dir) {
     auto& wal = *it;
     assert(wal.getting_synced);
     if (logs_.size() > 1) {
-      if (immutable_db_options_.track_and_verify_wals_in_manifest) {
+      if (immutable_db_options_.track_and_verify_wals_in_manifest &&
+          wal.writer->file()->GetFileSize() > 0) {
         synced_wals.AddWal(wal.number,
                            WalMetadata(wal.writer->file()->GetFileSize()));
       }
@@ -1692,7 +1718,9 @@ Status DBImpl::GetImpl(const ReadOptions& read_options, const Slice& key,
   }
   // If timestamp is used, we use read callback to ensure <key,t,s> is returned
   // only if t <= read_opts.timestamp and s <= snapshot.
-  if (ts_sz > 0 && !get_impl_options.callback) {
+  if (ts_sz > 0) {
+    assert(!get_impl_options
+                .callback);  // timestamp with callback is not supported
     read_cb.Refresh(snapshot);
     get_impl_options.callback = &read_cb;
   }
@@ -2204,7 +2232,7 @@ void DBImpl::MultiGet(const ReadOptions& read_options, const size_t num_keys,
   for (; cf_iter != multiget_cf_data.end(); ++cf_iter) {
     s = MultiGetImpl(read_options, cf_iter->start, cf_iter->num_keys,
                      &sorted_keys, cf_iter->super_version, consistent_seqnum,
-                     read_callback, nullptr);
+                     read_callback);
     if (!s.ok()) {
       break;
     }
@@ -2368,15 +2396,16 @@ void DBImpl::MultiGetWithCallback(
   }
 
   GetWithTimestampReadCallback timestamp_read_callback(0);
-  ReadCallback* read_callback = nullptr;
+  ReadCallback* read_callback = callback;
   if (read_options.timestamp && read_options.timestamp->size() > 0) {
+    assert(!read_callback);  // timestamp with callback is not supported
     timestamp_read_callback.Refresh(consistent_seqnum);
     read_callback = &timestamp_read_callback;
   }
 
   Status s = MultiGetImpl(read_options, 0, num_keys, sorted_keys,
                           multiget_cf_data[0].super_version, consistent_seqnum,
-                          read_callback, nullptr);
+                          read_callback);
   assert(s.ok() || s.IsTimedOut() || s.IsAborted());
   ReturnAndCleanupSuperVersion(multiget_cf_data[0].cfd,
                                multiget_cf_data[0].super_version);
@@ -2395,7 +2424,7 @@ Status DBImpl::MultiGetImpl(
     const ReadOptions& read_options, size_t start_key, size_t num_keys,
     autovector<KeyContext*, MultiGetContext::MAX_BATCH_SIZE>* sorted_keys,
     SuperVersion* super_version, SequenceNumber snapshot,
-    ReadCallback* callback, bool* is_blob_index) {
+    ReadCallback* callback) {
   PERF_CPU_TIMER_GUARD(get_cpu_nanos, env_);
   StopWatch sw(env_, stats_, DB_MULTIGET);
 
@@ -2434,11 +2463,9 @@ Status DBImpl::MultiGetImpl(
         (read_options.read_tier == kPersistedTier &&
          has_unpersisted_data_.load(std::memory_order_relaxed));
     if (!skip_memtable) {
-      super_version->mem->MultiGet(read_options, &range, callback,
-                                   is_blob_index);
+      super_version->mem->MultiGet(read_options, &range, callback);
       if (!range.empty()) {
-        super_version->imm->MultiGet(read_options, &range, callback,
-                                     is_blob_index);
+        super_version->imm->MultiGet(read_options, &range, callback);
       }
       if (!range.empty()) {
         lookup_current = true;
@@ -2448,8 +2475,7 @@ Status DBImpl::MultiGetImpl(
     }
     if (lookup_current) {
       PERF_TIMER_GUARD(get_from_output_files_time);
-      super_version->current->MultiGet(read_options, &range, callback,
-                                       is_blob_index);
+      super_version->current->MultiGet(read_options, &range, callback);
     }
     curr_value_size = range.GetValueSize();
     if (curr_value_size > read_options.value_size_soft_limit) {
@@ -2803,7 +2829,7 @@ Iterator* DBImpl::NewIterator(const ReadOptions& read_options,
                                     /* allow_unprepared_value */ true);
     result = NewDBIterator(
         env_, read_options, *cfd->ioptions(), sv->mutable_cf_options,
-        cfd->user_comparator(), iter, kMaxSequenceNumber,
+        cfd->user_comparator(), iter, sv->current, kMaxSequenceNumber,
         sv->mutable_cf_options.max_sequential_skip_in_iterations, read_callback,
         this, cfd);
 #endif
@@ -2824,7 +2850,7 @@ ArenaWrappedDBIter* DBImpl::NewIteratorImpl(const ReadOptions& read_options,
                                             ColumnFamilyData* cfd,
                                             SequenceNumber snapshot,
                                             ReadCallback* read_callback,
-                                            bool allow_blob,
+                                            bool expose_blob_index,
                                             bool allow_refresh) {
   SuperVersion* sv = cfd->GetReferencedSuperVersion(this);
 
@@ -2889,9 +2915,9 @@ ArenaWrappedDBIter* DBImpl::NewIteratorImpl(const ReadOptions& read_options,
   // likely that any iterator pointer is close to the iterator it points to so
   // that they are likely to be in the same cache line and/or page.
   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, this, cfd, allow_blob,
+      env_, read_options, *cfd->ioptions(), sv->mutable_cf_options, sv->current,
+      snapshot, sv->mutable_cf_options.max_sequential_skip_in_iterations,
+      sv->version_number, read_callback, this, cfd, expose_blob_index,
       read_options.snapshot != nullptr ? false : allow_refresh);
 
   InternalIterator* internal_iter = NewInternalIterator(
@@ -2929,7 +2955,7 @@ Status DBImpl::NewIterators(
                                       /* allow_unprepared_value */ true);
       iterators->push_back(NewDBIterator(
           env_, read_options, *cfd->ioptions(), sv->mutable_cf_options,
-          cfd->user_comparator(), iter, kMaxSequenceNumber,
+          cfd->user_comparator(), iter, sv->current, kMaxSequenceNumber,
           sv->mutable_cf_options.max_sequential_skip_in_iterations,
           read_callback, this, cfd));
     }
@@ -3101,8 +3127,8 @@ const std::string& DBImpl::GetName() const { return dbname_; }
 Env* DBImpl::GetEnv() const { return env_; }
 
 FileSystem* DB::GetFileSystem() const {
-  static LegacyFileSystemWrapper fs_wrap(GetEnv());
-  return &fs_wrap;
+  const auto& fs = GetEnv()->GetFileSystem();
+  return fs.get();
 }
 
 FileSystem* DBImpl::GetFileSystem() const {
@@ -3400,6 +3426,10 @@ Status DBImpl::GetApproximateSizes(const SizeApproximationOptions& options,
     return Status::InvalidArgument("Invalid options");
   }
 
+  const Comparator* const ucmp = column_family->GetComparator();
+  assert(ucmp);
+  size_t ts_sz = ucmp->timestamp_size();
+
   Version* v;
   auto cfh = static_cast_with_check<ColumnFamilyHandleImpl>(column_family);
   auto cfd = cfh->cfd();
@@ -3407,9 +3437,23 @@ Status DBImpl::GetApproximateSizes(const SizeApproximationOptions& options,
   v = sv->current;
 
   for (int i = 0; i < n; i++) {
+    Slice start = range[i].start;
+    Slice limit = range[i].limit;
+
+    // Add timestamp if needed
+    std::string start_with_ts, limit_with_ts;
+    if (ts_sz > 0) {
+      // Maximum timestamp means including all key with any timestamp
+      AppendKeyWithMaxTimestamp(&start_with_ts, start, ts_sz);
+      // Append a maximum timestamp as the range limit is exclusive:
+      // [start, limit)
+      AppendKeyWithMaxTimestamp(&limit_with_ts, limit, ts_sz);
+      start = start_with_ts;
+      limit = limit_with_ts;
+    }
     // Convert user_key into a corresponding internal key.
-    InternalKey k1(range[i].start, kMaxSequenceNumber, kValueTypeForSeek);
-    InternalKey k2(range[i].limit, kMaxSequenceNumber, kValueTypeForSeek);
+    InternalKey k1(start, kMaxSequenceNumber, kValueTypeForSeek);
+    InternalKey k2(limit, kMaxSequenceNumber, kValueTypeForSeek);
     sizes[i] = 0;
     if (options.include_files) {
       sizes[i] += versions_->ApproximateSize(
@@ -3467,7 +3511,6 @@ Status DBImpl::DeleteFile(std::string name) {
     return Status::InvalidArgument("Invalid file name");
   }
 
-  Status status;
   if (type == kWalFile) {
     // Only allow deleting archived log files
     if (log_type != kArchivedLogFile) {
@@ -3476,7 +3519,7 @@ Status DBImpl::DeleteFile(std::string name) {
                       name.c_str());
       return Status::NotSupported("Delete only supported for archived logs");
     }
-    status = wal_manager_.DeleteFile(name, number);
+    Status status = wal_manager_.DeleteFile(name, number);
     if (!status.ok()) {
       ROCKS_LOG_ERROR(immutable_db_options_.info_log,
                       "DeleteFile %s failed -- %s.\n", name.c_str(),
@@ -3485,6 +3528,7 @@ Status DBImpl::DeleteFile(std::string name) {
     return status;
   }
 
+  Status status;
   int level;
   FileMetaData* metadata;
   ColumnFamilyData* cfd;
@@ -3558,7 +3602,7 @@ Status DBImpl::DeleteFile(std::string name) {
 Status DBImpl::DeleteFilesInRanges(ColumnFamilyHandle* column_family,
                                    const RangePtr* ranges, size_t n,
                                    bool include_end) {
-  Status status;
+  Status status = Status::OK();
   auto cfh = static_cast_with_check<ColumnFamilyHandleImpl>(column_family);
   ColumnFamilyData* cfd = cfh->cfd();
   VersionEdit edit;
@@ -3617,7 +3661,7 @@ Status DBImpl::DeleteFilesInRanges(ColumnFamilyHandle* column_family,
     }
     if (edit.GetDeletedFiles().empty()) {
       job_context.Clean();
-      return Status::OK();
+      return status;
     }
     input_version->Ref();
     status = versions_->LogAndApply(cfd, *cfd->GetLatestMutableCFOptions(),
@@ -4164,16 +4208,17 @@ void DBImpl::EraseThreadStatusDbInfo() const {}
 //
 // A global method that can dump out the build version
 void DumpRocksDBBuildVersion(Logger* log) {
-#if !defined(IOS_CROSS_COMPILE)
-  // if we compile with Xcode, we don't run build_detect_version, so we don't
-  // generate util/build_version.cc
-  ROCKS_LOG_HEADER(log, "RocksDB version: %d.%d.%d\n", ROCKSDB_MAJOR,
-                   ROCKSDB_MINOR, ROCKSDB_PATCH);
-  ROCKS_LOG_HEADER(log, "Git sha %s", rocksdb_build_git_sha);
-  ROCKS_LOG_HEADER(log, "Compile date %s", rocksdb_build_compile_date);
-#else
-  (void)log;  // ignore "-Wunused-parameter"
-#endif
+  ROCKS_LOG_HEADER(log, "RocksDB version: %s\n",
+                   GetRocksVersionAsString().c_str());
+  const auto& props = GetRocksBuildProperties();
+  const auto& sha = props.find("rocksdb_build_git_sha");
+  if (sha != props.end()) {
+    ROCKS_LOG_HEADER(log, "Git sha %s", sha->second.c_str());
+  }
+  const auto date = props.find("rocksdb_build_date");
+  if (date != props.end()) {
+    ROCKS_LOG_HEADER(log, "Compile date %s", date->second.c_str());
+  }
 }
 
 #ifndef ROCKSDB_LITE
@@ -4338,7 +4383,7 @@ Status DBImpl::IngestExternalFiles(
     }
   }
   // Ingest multiple external SST files atomically.
-  size_t num_cfs = args.size();
+  const size_t num_cfs = args.size();
   for (size_t i = 0; i != num_cfs; ++i) {
     if (args[i].external_files.empty()) {
       char err_msg[128] = {0};
@@ -4379,10 +4424,7 @@ Status DBImpl::IngestExternalFiles(
         env_, versions_.get(), cfd, immutable_db_options_, file_options_,
         &snapshots_, arg.options, &directories_, &event_logger_, io_tracer_);
   }
-  std::vector<std::pair<bool, Status>> exec_results;
-  for (size_t i = 0; i != num_cfs; ++i) {
-    exec_results.emplace_back(false, Status::OK());
-  }
+
   // TODO(yanqin) maybe make jobs run in parallel
   uint64_t start_file_number = next_file_number;
   for (size_t i = 1; i != num_cfs; ++i) {
@@ -4390,10 +4432,13 @@ Status DBImpl::IngestExternalFiles(
     auto* cfd =
         static_cast<ColumnFamilyHandleImpl*>(args[i].column_family)->cfd();
     SuperVersion* super_version = cfd->GetReferencedSuperVersion(this);
-    exec_results[i].second = ingestion_jobs[i].Prepare(
+    Status es = ingestion_jobs[i].Prepare(
         args[i].external_files, args[i].files_checksums,
         args[i].files_checksum_func_names, start_file_number, super_version);
-    exec_results[i].first = true;
+    // capture first error only
+    if (!es.ok() && status.ok()) {
+      status = es;
+    }
     CleanupSuperVersion(super_version);
   }
   TEST_SYNC_POINT("DBImpl::IngestExternalFiles:BeforeLastJobPrepare:0");
@@ -4402,23 +4447,17 @@ Status DBImpl::IngestExternalFiles(
     auto* cfd =
         static_cast<ColumnFamilyHandleImpl*>(args[0].column_family)->cfd();
     SuperVersion* super_version = cfd->GetReferencedSuperVersion(this);
-    exec_results[0].second = ingestion_jobs[0].Prepare(
+    Status es = ingestion_jobs[0].Prepare(
         args[0].external_files, args[0].files_checksums,
         args[0].files_checksum_func_names, next_file_number, super_version);
-    exec_results[0].first = true;
-    CleanupSuperVersion(super_version);
-  }
-  for (const auto& exec_result : exec_results) {
-    if (!exec_result.second.ok()) {
-      status = exec_result.second;
-      break;
+    if (!es.ok()) {
+      status = es;
     }
+    CleanupSuperVersion(super_version);
   }
   if (!status.ok()) {
     for (size_t i = 0; i != num_cfs; ++i) {
-      if (exec_results[i].first) {
-        ingestion_jobs[i].Cleanup(status);
-      }
+      ingestion_jobs[i].Cleanup(status);
     }
     InstrumentedMutexLock l(&mutex_);
     ReleaseFileNumberFromPendingOutputs(pending_output_elem);
@@ -4588,8 +4627,7 @@ Status DBImpl::IngestExternalFiles(
       // TODO: distinguish between MANIFEST write and CURRENT renaming
       const IOStatus& io_s = versions_->io_status();
       // Should handle return error?
-      error_handler_.SetBGError(io_s, BackgroundErrorReason::kManifestWrite)
-          .PermitUncheckedError();
+      error_handler_.SetBGError(io_s, BackgroundErrorReason::kManifestWrite);
     }
 
     // Resume writes to the DB
@@ -4863,7 +4901,7 @@ Status DBImpl::VerifySstFileChecksum(const FileMetaData& fmeta,
       fs_.get(), fname, immutable_db_options_.file_checksum_gen_factory.get(),
       fmeta.file_checksum_func_name, &file_checksum, &func_name,
       read_options.readahead_size, immutable_db_options_.allow_mmap_reads,
-      io_tracer_);
+      io_tracer_, immutable_db_options_.rate_limiter.get());
   if (s.ok()) {
     assert(fmeta.file_checksum_func_name == func_name);
     if (file_checksum != fmeta.file_checksum) {
@@ -4935,24 +4973,27 @@ Status DBImpl::EndBlockCacheTrace() {
   return Status::OK();
 }
 
-Status DBImpl::TraceIteratorSeek(const uint32_t& cf_id, const Slice& key) {
+Status DBImpl::TraceIteratorSeek(const uint32_t& cf_id, const Slice& key,
+                                 const Slice& lower_bound,
+                                 const Slice upper_bound) {
   Status s;
   if (tracer_) {
     InstrumentedMutexLock lock(&trace_mutex_);
     if (tracer_) {
-      s = tracer_->IteratorSeek(cf_id, key);
+      s = tracer_->IteratorSeek(cf_id, key, lower_bound, upper_bound);
     }
   }
   return s;
 }
 
-Status DBImpl::TraceIteratorSeekForPrev(const uint32_t& cf_id,
-                                        const Slice& key) {
+Status DBImpl::TraceIteratorSeekForPrev(const uint32_t& cf_id, const Slice& key,
+                                        const Slice& lower_bound,
+                                        const Slice upper_bound) {
   Status s;
   if (tracer_) {
     InstrumentedMutexLock lock(&trace_mutex_);
     if (tracer_) {
-      s = tracer_->IteratorSeekForPrev(cf_id, key);
+      s = tracer_->IteratorSeekForPrev(cf_id, key, lower_bound, upper_bound);
     }
   }
   return s;

db/db_impl/db_impl.h

@@ -522,7 +522,7 @@ class DBImpl : public DB {
                                       ColumnFamilyData* cfd,
                                       SequenceNumber snapshot,
                                       ReadCallback* read_callback,
-                                      bool allow_blob = false,
+                                      bool expose_blob_index = false,
                                       bool allow_refresh = true);
 
   virtual SequenceNumber GetLastPublishedSequence() const {
@@ -595,8 +595,11 @@ class DBImpl : public DB {
                                  bool* found_record_for_key,
                                  bool* is_blob_index = nullptr);
 
-  Status TraceIteratorSeek(const uint32_t& cf_id, const Slice& key);
-  Status TraceIteratorSeekForPrev(const uint32_t& cf_id, const Slice& key);
+  Status TraceIteratorSeek(const uint32_t& cf_id, const Slice& key,
+                           const Slice& lower_bound, const Slice upper_bound);
+  Status TraceIteratorSeekForPrev(const uint32_t& cf_id, const Slice& key,
+                                  const Slice& lower_bound,
+                                  const Slice upper_bound);
 #endif  // ROCKSDB_LITE
 
   // Similar to GetSnapshot(), but also lets the db know that this snapshot
@@ -922,7 +925,7 @@ class DBImpl : public DB {
                            ColumnFamilyHandle* column_family = nullptr,
                            bool disallow_trivial_move = false);
 
-  void TEST_SwitchWAL();
+  Status TEST_SwitchWAL();
 
   bool TEST_UnableToReleaseOldestLog() { return unable_to_release_oldest_log_; }
 
@@ -954,6 +957,9 @@ class DBImpl : public DB {
   // is only for the special test of CancelledCompactions
   Status TEST_WaitForCompact(bool waitUnscheduled = false);
 
+  // Get the background error status
+  Status TEST_GetBGError();
+
   // Return the maximum overlapping data (in bytes) at next level for any
   // file at a level >= 1.
   int64_t TEST_MaxNextLevelOverlappingBytes(
@@ -1017,6 +1023,12 @@ class DBImpl : public DB {
 
   VersionSet* TEST_GetVersionSet() const { return versions_.get(); }
 
+  uint64_t TEST_GetCurrentLogNumber() const {
+    InstrumentedMutexLock l(mutex());
+    assert(!logs_.empty());
+    return logs_.back().number;
+  }
+
   const std::unordered_set<uint64_t>& TEST_GetFilesGrabbedForPurge() const {
     return files_grabbed_for_purge_;
   }
@@ -1104,6 +1116,15 @@ class DBImpl : public DB {
   // If need_enter_write_thread = false, the method will enter write thread.
   Status WriteOptionsFile(bool need_mutex_lock, bool need_enter_write_thread);
 
+  Status CompactRangeInternal(const CompactRangeOptions& options,
+                              ColumnFamilyHandle* column_family,
+                              const Slice* begin, const Slice* end);
+
+  Status GetApproximateSizesInternal(const SizeApproximationOptions& options,
+                                     ColumnFamilyHandle* column_family,
+                                     const Range* range, int n,
+                                     uint64_t* sizes);
+
   // The following two functions can only be called when:
   // 1. WriteThread::Writer::EnterUnbatched() is used.
   // 2. db_mutex is NOT held
@@ -1851,11 +1872,12 @@ class DBImpl : public DB {
   Status MultiGetImpl(
       const ReadOptions& read_options, size_t start_key, size_t num_keys,
       autovector<KeyContext*, MultiGetContext::MAX_BATCH_SIZE>* sorted_keys,
-      SuperVersion* sv, SequenceNumber snap_seqnum, ReadCallback* callback,
-      bool* is_blob_index);
+      SuperVersion* sv, SequenceNumber snap_seqnum, ReadCallback* callback);
 
   Status DisableFileDeletionsWithLock();
 
+  Status IncreaseFullHistoryTsLow(ColumnFamilyData* cfd, std::string ts_low);
+
   // table_cache_ provides its own synchronization
   std::shared_ptr<Cache> table_cache_;
 
@@ -2219,12 +2241,22 @@ extern uint64_t PrecomputeMinLogNumberToKeep2PC(
     const autovector<VersionEdit*>& edit_list,
     const autovector<MemTable*>& memtables_to_flush,
     LogsWithPrepTracker* prep_tracker);
+// For atomic flush.
+extern uint64_t PrecomputeMinLogNumberToKeep2PC(
+    VersionSet* vset, const autovector<ColumnFamilyData*>& cfds_to_flush,
+    const autovector<autovector<VersionEdit*>>& edit_lists,
+    const autovector<const autovector<MemTable*>*>& memtables_to_flush,
+    LogsWithPrepTracker* prep_tracker);
 
 // In non-2PC mode, WALs with log number < the returned number can be
 // deleted after the cfd_to_flush column family is flushed successfully.
 extern uint64_t PrecomputeMinLogNumberToKeepNon2PC(
     VersionSet* vset, const ColumnFamilyData& cfd_to_flush,
     const autovector<VersionEdit*>& edit_list);
+// For atomic flush.
+extern uint64_t PrecomputeMinLogNumberToKeepNon2PC(
+    VersionSet* vset, const autovector<ColumnFamilyData*>& cfds_to_flush,
+    const autovector<autovector<VersionEdit*>>& edit_lists);
 
 // `cfd_to_flush` is the column family whose memtable will be flushed and thus
 // will not depend on any WAL file. nullptr means no memtable is being flushed.
@@ -2232,6 +2264,10 @@ extern uint64_t PrecomputeMinLogNumberToKeepNon2PC(
 extern uint64_t FindMinPrepLogReferencedByMemTable(
     VersionSet* vset, const ColumnFamilyData* cfd_to_flush,
     const autovector<MemTable*>& memtables_to_flush);
+// For atomic flush.
+extern uint64_t FindMinPrepLogReferencedByMemTable(
+    VersionSet* vset, const autovector<ColumnFamilyData*>& cfds_to_flush,
+    const autovector<const autovector<MemTable*>*>& memtables_to_flush);
 
 // Fix user-supplied options to be reasonable
 template <class T, class V>

db/db_impl/db_impl_compaction_flush.cc

@@ -35,8 +35,10 @@ bool DBImpl::EnoughRoomForCompaction(
     // Pass the current bg_error_ to SFM so it can decide what checks to
     // perform. If this DB instance hasn't seen any error yet, the SFM can be
     // optimistic and not do disk space checks
-    enough_room =
-        sfm->EnoughRoomForCompaction(cfd, inputs, error_handler_.GetBGError());
+    Status bg_error = error_handler_.GetBGError();
+    enough_room = sfm->EnoughRoomForCompaction(cfd, inputs, bg_error);
+    bg_error.PermitUncheckedError();  // bg_error is just a copy of the Status
+                                      // from the error_handler_
     if (enough_room) {
       *sfm_reserved_compact_space = true;
     }
@@ -130,12 +132,10 @@ IOStatus DBImpl::SyncClosedLogs(JobContext* job_context) {
     }
     if (!io_s.ok()) {
       if (total_log_size_ > 0) {
-        error_handler_.SetBGError(io_s, BackgroundErrorReason::kFlush)
-            .PermitUncheckedError();
+        error_handler_.SetBGError(io_s, BackgroundErrorReason::kFlush);
       } else {
         // If the WAL is empty, we use different error reason
-        error_handler_.SetBGError(io_s, BackgroundErrorReason::kFlushNoWAL)
-            .PermitUncheckedError();
+        error_handler_.SetBGError(io_s, BackgroundErrorReason::kFlushNoWAL);
       }
       TEST_SYNC_POINT("DBImpl::SyncClosedLogs:Failed");
       return io_s;
@@ -159,14 +159,14 @@ Status DBImpl::FlushMemTableToOutputFile(
 
   FlushJob flush_job(
       dbname_, cfd, immutable_db_options_, mutable_cf_options,
-      nullptr /* memtable_id */, file_options_for_compaction_, versions_.get(),
-      &mutex_, &shutting_down_, snapshot_seqs, earliest_write_conflict_snapshot,
-      snapshot_checker, job_context, log_buffer, directories_.GetDbDir(),
-      GetDataDir(cfd, 0U),
+      port::kMaxUint64 /* memtable_id */, file_options_for_compaction_,
+      versions_.get(), &mutex_, &shutting_down_, snapshot_seqs,
+      earliest_write_conflict_snapshot, snapshot_checker, job_context,
+      log_buffer, directories_.GetDbDir(), GetDataDir(cfd, 0U),
       GetCompressionFlush(*cfd->ioptions(), mutable_cf_options), stats_,
       &event_logger_, mutable_cf_options.report_bg_io_stats,
       true /* sync_output_directory */, true /* write_manifest */, thread_pri,
-      io_tracer_, db_id_, db_session_id_);
+      io_tracer_, db_id_, db_session_id_, cfd->GetFullHistoryTsLow());
   FileMetaData file_meta;
 
   TEST_SYNC_POINT("DBImpl::FlushMemTableToOutputFile:BeforePickMemtables");
@@ -192,8 +192,7 @@ Status DBImpl::FlushMemTableToOutputFile(
     s = io_s;
     if (!io_s.ok() && !io_s.IsShutdownInProgress() &&
         !io_s.IsColumnFamilyDropped()) {
-      error_handler_.SetBGError(io_s, BackgroundErrorReason::kFlush)
-          .PermitUncheckedError();
+      error_handler_.SetBGError(io_s, BackgroundErrorReason::kFlush);
     }
   } else {
     TEST_SYNC_POINT("DBImpl::SyncClosedLogs:Skip");
@@ -253,24 +252,23 @@ Status DBImpl::FlushMemTableToOutputFile(
       // be pessimistic and try write to a new MANIFEST.
       // TODO: distinguish between MANIFEST write and CURRENT renaming
       if (!versions_->io_status().ok()) {
-        // Should handle return error?
-        error_handler_.SetBGError(io_s, BackgroundErrorReason::kManifestWrite)
-            .PermitUncheckedError();
+        if (total_log_size_ > 0) {
+          // If the WAL is empty, we use different error reason
+          error_handler_.SetBGError(io_s,
+                                    BackgroundErrorReason::kManifestWrite);
+        } else {
+          error_handler_.SetBGError(io_s,
+                                    BackgroundErrorReason::kManifestWriteNoWAL);
+        }
       } else if (total_log_size_ > 0) {
-        // Should handle return error?
-        error_handler_.SetBGError(io_s, BackgroundErrorReason::kFlush)
-            .PermitUncheckedError();
+        error_handler_.SetBGError(io_s, BackgroundErrorReason::kFlush);
       } else {
         // If the WAL is empty, we use different error reason
-        // Should handle return error?
-        error_handler_.SetBGError(io_s, BackgroundErrorReason::kFlushNoWAL)
-            .PermitUncheckedError();
+        error_handler_.SetBGError(io_s, BackgroundErrorReason::kFlushNoWAL);
       }
     } else {
       Status new_bg_error = s;
-      // Should handle return error?
-      error_handler_.SetBGError(new_bg_error, BackgroundErrorReason::kFlush)
-          .PermitUncheckedError();
+      error_handler_.SetBGError(new_bg_error, BackgroundErrorReason::kFlush);
     }
   } else {
     // If we got here, then we decided not to care about the i_os status (either
@@ -288,16 +286,17 @@ Status DBImpl::FlushMemTableToOutputFile(
       // Notify sst_file_manager that a new file was added
       std::string file_path = MakeTableFileName(
           cfd->ioptions()->cf_paths[0].path, file_meta.fd.GetNumber());
-      sfm->OnAddFile(file_path);
+      // TODO (PR7798).  We should only add the file to the FileManager if it
+      // exists. Otherwise, some tests may fail.  Ignore the error in the
+      // interim.
+      sfm->OnAddFile(file_path).PermitUncheckedError();
       if (sfm->IsMaxAllowedSpaceReached()) {
         Status new_bg_error =
             Status::SpaceLimit("Max allowed space was reached");
         TEST_SYNC_POINT_CALLBACK(
             "DBImpl::FlushMemTableToOutputFile:MaxAllowedSpaceReached",
             &new_bg_error);
-        // Should handle this error?
-        error_handler_.SetBGError(new_bg_error, BackgroundErrorReason::kFlush)
-            .PermitUncheckedError();
+        error_handler_.SetBGError(new_bg_error, BackgroundErrorReason::kFlush);
       }
     }
 #endif  // ROCKSDB_LITE
@@ -313,30 +312,22 @@ Status DBImpl::FlushMemTablesToOutputFiles(
     return AtomicFlushMemTablesToOutputFiles(
         bg_flush_args, made_progress, job_context, log_buffer, thread_pri);
   }
+  assert(bg_flush_args.size() == 1);
   std::vector<SequenceNumber> snapshot_seqs;
   SequenceNumber earliest_write_conflict_snapshot;
   SnapshotChecker* snapshot_checker;
   GetSnapshotContext(job_context, &snapshot_seqs,
                      &earliest_write_conflict_snapshot, &snapshot_checker);
-  Status status;
-  for (auto& arg : bg_flush_args) {
-    ColumnFamilyData* cfd = arg.cfd_;
-    MutableCFOptions mutable_cf_options = *cfd->GetLatestMutableCFOptions();
-    SuperVersionContext* superversion_context = arg.superversion_context_;
-    Status s = FlushMemTableToOutputFile(
-        cfd, mutable_cf_options, made_progress, job_context,
-        superversion_context, snapshot_seqs, earliest_write_conflict_snapshot,
-        snapshot_checker, log_buffer, thread_pri);
-    if (!s.ok()) {
-      status = s;
-      if (!s.IsShutdownInProgress() && !s.IsColumnFamilyDropped()) {
-        // At this point, DB is not shutting down, nor is cfd dropped.
-        // Something is wrong, thus we break out of the loop.
-        break;
-      }
-    }
-  }
-  return status;
+  const auto& bg_flush_arg = bg_flush_args[0];
+  ColumnFamilyData* cfd = bg_flush_arg.cfd_;
+  MutableCFOptions mutable_cf_options = *cfd->GetLatestMutableCFOptions();
+  SuperVersionContext* superversion_context =
+      bg_flush_arg.superversion_context_;
+  Status s = FlushMemTableToOutputFile(
+      cfd, mutable_cf_options, made_progress, job_context, superversion_context,
+      snapshot_seqs, earliest_write_conflict_snapshot, snapshot_checker,
+      log_buffer, thread_pri);
+  return s;
 }
 
 /*
@@ -399,7 +390,7 @@ Status DBImpl::AtomicFlushMemTablesToOutputFiles(
 
     all_mutable_cf_options.emplace_back(*cfd->GetLatestMutableCFOptions());
     const MutableCFOptions& mutable_cf_options = all_mutable_cf_options.back();
-    const uint64_t* max_memtable_id = &(bg_flush_args[i].max_memtable_id_);
+    uint64_t max_memtable_id = bg_flush_args[i].max_memtable_id_;
     jobs.emplace_back(new FlushJob(
         dbname_, cfd, immutable_db_options_, mutable_cf_options,
         max_memtable_id, file_options_for_compaction_, versions_.get(), &mutex_,
@@ -408,7 +399,8 @@ Status DBImpl::AtomicFlushMemTablesToOutputFiles(
         data_dir, GetCompressionFlush(*cfd->ioptions(), mutable_cf_options),
         stats_, &event_logger_, mutable_cf_options.report_bg_io_stats,
         false /* sync_output_directory */, false /* write_manifest */,
-        thread_pri, io_tracer_, db_id_, db_session_id_));
+        thread_pri, io_tracer_, db_id_, db_session_id_,
+        cfd->GetFullHistoryTsLow()));
     jobs.back()->PickMemTable();
   }
 
@@ -585,7 +577,7 @@ Status DBImpl::AtomicFlushMemTablesToOutputFiles(
 
     s = InstallMemtableAtomicFlushResults(
         nullptr /* imm_lists */, tmp_cfds, mutable_cf_options_list, mems_list,
-        versions_.get(), &mutex_, tmp_file_meta,
+        versions_.get(), &logs_with_prep_tracker_, &mutex_, tmp_file_meta,
         &job_context->memtables_to_free, directories_.GetDbDir(), log_buffer);
   }
 
@@ -631,7 +623,7 @@ Status DBImpl::AtomicFlushMemTablesToOutputFiles(
     auto sfm = static_cast<SstFileManagerImpl*>(
         immutable_db_options_.sst_file_manager.get());
     assert(all_mutable_cf_options.size() == static_cast<size_t>(num_cfs));
-    for (int i = 0; i != num_cfs; ++i) {
+    for (int i = 0; s.ok() && i != num_cfs; ++i) {
       if (cfds[i]->IsDropped()) {
         continue;
       }
@@ -640,14 +632,16 @@ Status DBImpl::AtomicFlushMemTablesToOutputFiles(
       if (sfm) {
         std::string file_path = MakeTableFileName(
             cfds[i]->ioptions()->cf_paths[0].path, file_meta[i].fd.GetNumber());
-        sfm->OnAddFile(file_path);
+        // TODO (PR7798).  We should only add the file to the FileManager if it
+        // exists. Otherwise, some tests may fail.  Ignore the error in the
+        // interim.
+        sfm->OnAddFile(file_path).PermitUncheckedError();
         if (sfm->IsMaxAllowedSpaceReached() &&
             error_handler_.GetBGError().ok()) {
           Status new_bg_error =
               Status::SpaceLimit("Max allowed space was reached");
-          // Should Handle this error?
-          error_handler_.SetBGError(new_bg_error, BackgroundErrorReason::kFlush)
-              .PermitUncheckedError();
+          error_handler_.SetBGError(new_bg_error,
+                                    BackgroundErrorReason::kFlush);
         }
       }
     }
@@ -664,24 +658,23 @@ Status DBImpl::AtomicFlushMemTablesToOutputFiles(
       // be pessimistic and try write to a new MANIFEST.
       // TODO: distinguish between MANIFEST write and CURRENT renaming
       if (!versions_->io_status().ok()) {
-        // Should Handle this error?
-        error_handler_.SetBGError(io_s, BackgroundErrorReason::kManifestWrite)
-            .PermitUncheckedError();
+        if (total_log_size_ > 0) {
+          // If the WAL is empty, we use different error reason
+          error_handler_.SetBGError(io_s,
+                                    BackgroundErrorReason::kManifestWrite);
+        } else {
+          error_handler_.SetBGError(io_s,
+                                    BackgroundErrorReason::kManifestWriteNoWAL);
+        }
       } else if (total_log_size_ > 0) {
-        // Should Handle this error?
-        error_handler_.SetBGError(io_s, BackgroundErrorReason::kFlush)
-            .PermitUncheckedError();
+        error_handler_.SetBGError(io_s, BackgroundErrorReason::kFlush);
       } else {
         // If the WAL is empty, we use different error reason
-        // Should Handle this error?
-        error_handler_.SetBGError(io_s, BackgroundErrorReason::kFlushNoWAL)
-            .PermitUncheckedError();
+        error_handler_.SetBGError(io_s, BackgroundErrorReason::kFlushNoWAL);
       }
     } else {
       Status new_bg_error = s;
-      // Should Handle this error?
-      error_handler_.SetBGError(new_bg_error, BackgroundErrorReason::kFlush)
-          .PermitUncheckedError();
+      error_handler_.SetBGError(new_bg_error, BackgroundErrorReason::kFlush);
     }
   }
 
@@ -781,7 +774,60 @@ void DBImpl::NotifyOnFlushCompleted(
 
 Status DBImpl::CompactRange(const CompactRangeOptions& options,
                             ColumnFamilyHandle* column_family,
-                            const Slice* begin, const Slice* end) {
+                            const Slice* begin_without_ts,
+                            const Slice* end_without_ts) {
+  const Comparator* const ucmp = column_family->GetComparator();
+  assert(ucmp);
+  size_t ts_sz = ucmp->timestamp_size();
+  if (ts_sz == 0) {
+    return CompactRangeInternal(options, column_family, begin_without_ts,
+                                end_without_ts);
+  }
+
+  std::string begin_str;
+  std::string end_str;
+
+  // CompactRange compact all keys: [begin, end] inclusively. Add maximum
+  // timestamp to include all `begin` keys, and add minimal timestamp to include
+  // all `end` keys.
+  if (begin_without_ts != nullptr) {
+    AppendKeyWithMaxTimestamp(&begin_str, *begin_without_ts, ts_sz);
+  }
+  if (end_without_ts != nullptr) {
+    AppendKeyWithMinTimestamp(&end_str, *end_without_ts, ts_sz);
+  }
+  Slice begin(begin_str);
+  Slice end(end_str);
+
+  Slice* begin_with_ts = begin_without_ts ? &begin : nullptr;
+  Slice* end_with_ts = end_without_ts ? &end : nullptr;
+
+  return CompactRangeInternal(options, column_family, begin_with_ts,
+                              end_with_ts);
+}
+
+Status DBImpl::IncreaseFullHistoryTsLow(ColumnFamilyData* cfd,
+                                        std::string ts_low) {
+  VersionEdit edit;
+  edit.SetColumnFamily(cfd->GetID());
+  edit.SetFullHistoryTsLow(ts_low);
+
+  InstrumentedMutexLock l(&mutex_);
+  std::string current_ts_low = cfd->GetFullHistoryTsLow();
+  const Comparator* ucmp = cfd->user_comparator();
+  if (!current_ts_low.empty() &&
+      ucmp->CompareTimestamp(ts_low, current_ts_low) < 0) {
+    return Status::InvalidArgument(
+        "Cannot decrease full_history_timestamp_low");
+  }
+
+  return versions_->LogAndApply(cfd, *cfd->GetLatestMutableCFOptions(), &edit,
+                                &mutex_);
+}
+
+Status DBImpl::CompactRangeInternal(const CompactRangeOptions& options,
+                                    ColumnFamilyHandle* column_family,
+                                    const Slice* begin, const Slice* end) {
   auto cfh = static_cast_with_check<ColumnFamilyHandleImpl>(column_family);
   auto cfd = cfh->cfd();
 
@@ -790,20 +836,36 @@ Status DBImpl::CompactRange(const CompactRangeOptions& options,
   }
 
   bool flush_needed = true;
+
+  // Update full_history_ts_low if it's set
+  if (options.full_history_ts_low != nullptr &&
+      !options.full_history_ts_low->empty()) {
+    std::string ts_low = options.full_history_ts_low->ToString();
+    if (begin != nullptr || end != nullptr) {
+      return Status::InvalidArgument(
+          "Cannot specify compaction range with full_history_ts_low");
+    }
+    Status s = IncreaseFullHistoryTsLow(cfd, ts_low);
+    if (!s.ok()) {
+      LogFlush(immutable_db_options_.info_log);
+      return s;
+    }
+  }
+
+  Status s;
   if (begin != nullptr && end != nullptr) {
     // TODO(ajkr): We could also optimize away the flush in certain cases where
     // one/both sides of the interval are unbounded. But it requires more
     // changes to RangesOverlapWithMemtables.
     Range range(*begin, *end);
     SuperVersion* super_version = cfd->GetReferencedSuperVersion(this);
-    cfd->RangesOverlapWithMemtables({range}, super_version,
-                                    immutable_db_options_.allow_data_in_errors,
-                                    &flush_needed);
+    s = cfd->RangesOverlapWithMemtables(
+        {range}, super_version, immutable_db_options_.allow_data_in_errors,
+        &flush_needed);
     CleanupSuperVersion(super_version);
   }
 
-  Status s;
-  if (flush_needed) {
+  if (s.ok() && flush_needed) {
     FlushOptions fo;
     fo.allow_write_stall = options.allow_write_stall;
     if (immutable_db_options_.atomic_flush) {
@@ -1160,7 +1222,8 @@ Status DBImpl::CompactFilesImpl(
       c->mutable_cf_options()->paranoid_file_checks,
       c->mutable_cf_options()->report_bg_io_stats, dbname_,
       &compaction_job_stats, Env::Priority::USER, io_tracer_,
-      &manual_compaction_paused_, db_id_, db_session_id_);
+      &manual_compaction_paused_, db_id_, db_session_id_,
+      c->column_family_data()->GetFullHistoryTsLow());
 
   // Creating a compaction influences the compaction score because the score
   // takes running compactions into account (by skipping files that are already
@@ -1174,7 +1237,8 @@ Status DBImpl::CompactFilesImpl(
   mutex_.Unlock();
   TEST_SYNC_POINT("CompactFilesImpl:0");
   TEST_SYNC_POINT("CompactFilesImpl:1");
-  compaction_job.Run();
+  // Ignore the status here, as it will be checked in the Install down below...
+  compaction_job.Run().PermitUncheckedError();
   TEST_SYNC_POINT("CompactFilesImpl:2");
   TEST_SYNC_POINT("CompactFilesImpl:3");
   mutex_.Lock();
@@ -1224,11 +1288,9 @@ Status DBImpl::CompactFilesImpl(
                    job_context->job_id, status.ToString().c_str());
     IOStatus io_s = compaction_job.io_status();
     if (!io_s.ok()) {
-      error_handler_.SetBGError(io_s, BackgroundErrorReason::kCompaction)
-          .PermitUncheckedError();
+      error_handler_.SetBGError(io_s, BackgroundErrorReason::kCompaction);
     } else {
-      error_handler_.SetBGError(status, BackgroundErrorReason::kCompaction)
-          .PermitUncheckedError();
+      error_handler_.SetBGError(status, BackgroundErrorReason::kCompaction);
     }
   }
 
@@ -1373,8 +1435,6 @@ Status DBImpl::ReFitLevel(ColumnFamilyData* cfd, int level, int target_level) {
 
   SuperVersionContext sv_context(/* create_superversion */ true);
 
-  Status status;
-
   InstrumentedMutexLock guard_lock(&mutex_);
 
   // only allow one thread refitting
@@ -1438,8 +1498,9 @@ Status DBImpl::ReFitLevel(ColumnFamilyData* cfd, int level, int target_level) {
                     "[%s] Apply version edit:\n%s", cfd->GetName().c_str(),
                     edit.DebugString().data());
 
-    status = versions_->LogAndApply(cfd, mutable_cf_options, &edit, &mutex_,
-                                    directories_.GetDbDir());
+    Status status = versions_->LogAndApply(cfd, mutable_cf_options, &edit,
+                                           &mutex_, directories_.GetDbDir());
+
     InstallSuperVersionAndScheduleWork(cfd, &sv_context, mutable_cf_options);
 
     ROCKS_LOG_DEBUG(immutable_db_options_.info_log, "[%s] LogAndApply: %s\n",
@@ -1450,12 +1511,14 @@ Status DBImpl::ReFitLevel(ColumnFamilyData* cfd, int level, int target_level) {
                       "[%s] After refitting:\n%s", cfd->GetName().c_str(),
                       cfd->current()->DebugString().data());
     }
+    sv_context.Clean();
+    refitting_level_ = false;
+
+    return status;
   }
 
-  sv_context.Clean();
   refitting_level_ = false;
-
-  return status;
+  return Status::OK();
 }
 
 int DBImpl::NumberLevels(ColumnFamilyHandle* column_family) {
@@ -1697,8 +1760,9 @@ void DBImpl::GenerateFlushRequest(const autovector<ColumnFamilyData*>& cfds,
 Status DBImpl::FlushMemTable(ColumnFamilyData* cfd,
                              const FlushOptions& flush_options,
                              FlushReason flush_reason, bool writes_stopped) {
+  // This method should not be called if atomic_flush is true.
+  assert(!immutable_db_options_.atomic_flush);
   Status s;
-  uint64_t flush_memtable_id = 0;
   if (!flush_options.allow_write_stall) {
     bool flush_needed = true;
     s = WaitUntilFlushWouldNotStallWrites(cfd, &flush_needed);
@@ -1708,7 +1772,8 @@ Status DBImpl::FlushMemTable(ColumnFamilyData* cfd,
     }
   }
 
-  FlushRequest flush_req;
+  autovector<FlushRequest> flush_reqs;
+  autovector<uint64_t> memtable_ids_to_wait;
   {
     WriteContext context;
     InstrumentedMutexLock guard_lock(&mutex_);
@@ -1723,18 +1788,24 @@ Status DBImpl::FlushMemTable(ColumnFamilyData* cfd,
     }
     WaitForPendingWrites();
 
-    if (!cfd->mem()->IsEmpty() || !cached_recoverable_state_empty_.load()) {
-      if (flush_reason != FlushReason::kErrorRecoveryRetryFlush) {
-        s = SwitchMemtable(cfd, &context);
-      } else {
-        assert(cfd->imm()->NumNotFlushed() > 0);
-      }
+    if (flush_reason != FlushReason::kErrorRecoveryRetryFlush &&
+        (!cfd->mem()->IsEmpty() || !cached_recoverable_state_empty_.load())) {
+      // Note that, when flush reason is kErrorRecoveryRetryFlush, during the
+      // auto retry resume, we want to avoid creating new small memtables.
+      // Therefore, SwitchMemtable will not be called. Also, since ResumeImpl
+      // will iterate through all the CFs and call FlushMemtable during auto
+      // retry resume, it is possible that in some CFs,
+      // cfd->imm()->NumNotFlushed() = 0. In this case, so no flush request will
+      // be created and scheduled, status::OK() will be returned.
+      s = SwitchMemtable(cfd, &context);
     }
+    const uint64_t flush_memtable_id = port::kMaxUint64;
     if (s.ok()) {
       if (cfd->imm()->NumNotFlushed() != 0 || !cfd->mem()->IsEmpty() ||
           !cached_recoverable_state_empty_.load()) {
-        flush_memtable_id = cfd->imm()->GetLatestMemTableID();
-        flush_req.emplace_back(cfd, flush_memtable_id);
+        FlushRequest req{{cfd, flush_memtable_id}};
+        flush_reqs.emplace_back(std::move(req));
+        memtable_ids_to_wait.emplace_back(cfd->imm()->GetLatestMemTableID());
       }
       if (immutable_db_options_.persist_stats_to_disk &&
           flush_reason != FlushReason::kErrorRecoveryRetryFlush) {
@@ -1760,15 +1831,19 @@ Status DBImpl::FlushMemTable(ColumnFamilyData* cfd,
                            "to avoid holding old logs",
                            cfd->GetName().c_str());
             s = SwitchMemtable(cfd_stats, &context);
-            flush_memtable_id = cfd_stats->imm()->GetLatestMemTableID();
-            flush_req.emplace_back(cfd_stats, flush_memtable_id);
+            FlushRequest req{{cfd_stats, flush_memtable_id}};
+            flush_reqs.emplace_back(std::move(req));
+            memtable_ids_to_wait.emplace_back(
+                cfd->imm()->GetLatestMemTableID());
           }
         }
       }
     }
-    if (s.ok() && !flush_req.empty()) {
-      for (auto& elem : flush_req) {
-        ColumnFamilyData* loop_cfd = elem.first;
+
+    if (s.ok() && !flush_reqs.empty()) {
+      for (const auto& req : flush_reqs) {
+        assert(req.size() == 1);
+        ColumnFamilyData* loop_cfd = req[0].first;
         loop_cfd->imm()->FlushRequested();
       }
       // If the caller wants to wait for this flush to complete, it indicates
@@ -1776,12 +1851,15 @@ Status DBImpl::FlushMemTable(ColumnFamilyData* cfd,
       // other threads which may drop the column family concurrently.
       // Therefore, we increase the cfd's ref count.
       if (flush_options.wait) {
-        for (auto& elem : flush_req) {
-          ColumnFamilyData* loop_cfd = elem.first;
+        for (const auto& req : flush_reqs) {
+          assert(req.size() == 1);
+          ColumnFamilyData* loop_cfd = req[0].first;
           loop_cfd->Ref();
         }
       }
-      SchedulePendingFlush(flush_req, flush_reason);
+      for (const auto& req : flush_reqs) {
+        SchedulePendingFlush(req, flush_reason);
+      }
       MaybeScheduleFlushOrCompaction();
     }
 
@@ -1797,9 +1875,11 @@ Status DBImpl::FlushMemTable(ColumnFamilyData* cfd,
   if (s.ok() && flush_options.wait) {
     autovector<ColumnFamilyData*> cfds;
     autovector<const uint64_t*> flush_memtable_ids;
-    for (auto& iter : flush_req) {
-      cfds.push_back(iter.first);
-      flush_memtable_ids.push_back(&(iter.second));
+    assert(flush_reqs.size() == memtable_ids_to_wait.size());
+    for (size_t i = 0; i < flush_reqs.size(); ++i) {
+      assert(flush_reqs[i].size() == 1);
+      cfds.push_back(flush_reqs[i][0].first);
+      flush_memtable_ids.push_back(&(memtable_ids_to_wait[i]));
     }
     s = WaitForFlushMemTables(
         cfds, flush_memtable_ids,
@@ -2224,6 +2304,17 @@ DBImpl::FlushRequest DBImpl::PopFirstFromFlushQueue() {
   assert(!flush_queue_.empty());
   FlushRequest flush_req = flush_queue_.front();
   flush_queue_.pop_front();
+  if (!immutable_db_options_.atomic_flush) {
+    assert(flush_req.size() == 1);
+  }
+  for (const auto& elem : flush_req) {
+    if (!immutable_db_options_.atomic_flush) {
+      ColumnFamilyData* cfd = elem.first;
+      assert(cfd);
+      assert(cfd->queued_for_flush());
+      cfd->set_queued_for_flush(false);
+    }
+  }
   // TODO: need to unset flush reason?
   return flush_req;
 }
@@ -2256,19 +2347,36 @@ ColumnFamilyData* DBImpl::PickCompactionFromQueue(
 
 void DBImpl::SchedulePendingFlush(const FlushRequest& flush_req,
                                   FlushReason flush_reason) {
+  mutex_.AssertHeld();
   if (flush_req.empty()) {
     return;
   }
-  for (auto& iter : flush_req) {
-    ColumnFamilyData* cfd = iter.first;
-    cfd->Ref();
-    cfd->SetFlushReason(flush_reason);
+  if (!immutable_db_options_.atomic_flush) {
+    // For the non-atomic flush case, we never schedule multiple column
+    // families in the same flush request.
+    assert(flush_req.size() == 1);
+    ColumnFamilyData* cfd = flush_req[0].first;
+    assert(cfd);
+    if (!cfd->queued_for_flush() && cfd->imm()->IsFlushPending()) {
+      cfd->Ref();
+      cfd->set_queued_for_flush(true);
+      cfd->SetFlushReason(flush_reason);
+      ++unscheduled_flushes_;
+      flush_queue_.push_back(flush_req);
+    }
+  } else {
+    for (auto& iter : flush_req) {
+      ColumnFamilyData* cfd = iter.first;
+      cfd->Ref();
+      cfd->SetFlushReason(flush_reason);
+    }
+    ++unscheduled_flushes_;
+    flush_queue_.push_back(flush_req);
   }
-  ++unscheduled_flushes_;
-  flush_queue_.push_back(flush_req);
 }
 
 void DBImpl::SchedulePendingCompaction(ColumnFamilyData* cfd) {
+  mutex_.AssertHeld();
   if (!cfd->queued_for_compaction() && cfd->NeedsCompaction()) {
     AddToCompactionQueue(cfd);
     ++unscheduled_compactions_;
@@ -2563,7 +2671,8 @@ void DBImpl::BackgroundCallCompaction(PrepickedCompaction* prepicked_compaction,
     // failure). Thus, we force full scan in FindObsoleteFiles()
     FindObsoleteFiles(&job_context, !s.ok() && !s.IsShutdownInProgress() &&
                                         !s.IsManualCompactionPaused() &&
-                                        !s.IsColumnFamilyDropped());
+                                        !s.IsColumnFamilyDropped() &&
+                                        !s.IsBusy());
     TEST_SYNC_POINT("DBImpl::BackgroundCallCompaction:FoundObsoleteFiles");
 
     // delete unnecessary files if any, this is done outside the mutex
@@ -2969,7 +3078,7 @@ Status DBImpl::BackgroundCompaction(bool* made_progress,
         c->mutable_cf_options()->report_bg_io_stats, dbname_,
         &compaction_job_stats, thread_pri, io_tracer_,
         is_manual ? &manual_compaction_paused_ : nullptr, db_id_,
-        db_session_id_);
+        db_session_id_, c->column_family_data()->GetFullHistoryTsLow());
     compaction_job.Prepare();
 
     NotifyOnCompactionBegin(c->column_family_data(), c.get(), status,
@@ -3034,10 +3143,9 @@ Status DBImpl::BackgroundCompaction(bool* made_progress,
       auto err_reason = versions_->io_status().ok()
                             ? BackgroundErrorReason::kCompaction
                             : BackgroundErrorReason::kManifestWrite;
-      error_handler_.SetBGError(io_s, err_reason).PermitUncheckedError();
+      error_handler_.SetBGError(io_s, err_reason);
     } else {
-      error_handler_.SetBGError(status, BackgroundErrorReason::kCompaction)
-          .PermitUncheckedError();
+      error_handler_.SetBGError(status, BackgroundErrorReason::kCompaction);
     }
     if (c != nullptr && !is_manual && !error_handler_.IsBGWorkStopped()) {
       // Put this cfd back in the compaction queue so we can retry after some

db/db_impl/db_impl_debug.cc

@@ -22,12 +22,13 @@ uint64_t DBImpl::TEST_GetLevel0TotalSize() {
   return default_cf_handle_->cfd()->current()->storage_info()->NumLevelBytes(0);
 }
 
-void DBImpl::TEST_SwitchWAL() {
+Status DBImpl::TEST_SwitchWAL() {
   WriteContext write_context;
   InstrumentedMutexLock l(&mutex_);
   void* writer = TEST_BeginWrite();
-  SwitchWAL(&write_context);
+  auto s = SwitchWAL(&write_context);
   TEST_EndWrite(writer);
+  return s;
 }
 
 bool DBImpl::TEST_WALBufferIsEmpty(bool lock) {
@@ -170,12 +171,17 @@ Status DBImpl::TEST_WaitForCompact(bool wait_unscheduled) {
   while ((bg_bottom_compaction_scheduled_ || bg_compaction_scheduled_ ||
           bg_flush_scheduled_ ||
           (wait_unscheduled && unscheduled_compactions_)) &&
-         (error_handler_.GetBGError() == Status::OK())) {
+         (error_handler_.GetBGError().ok())) {
     bg_cv_.Wait();
   }
   return error_handler_.GetBGError();
 }
 
+Status DBImpl::TEST_GetBGError() {
+  InstrumentedMutexLock l(&mutex_);
+  return error_handler_.GetBGError();
+}
+
 void DBImpl::TEST_LockMutex() { mutex_.Lock(); }
 
 void DBImpl::TEST_UnlockMutex() { mutex_.Unlock(); }

db/db_impl/db_impl_files.cc

@@ -6,16 +6,17 @@
 // Copyright (c) 2011 The LevelDB Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file. See the AUTHORS file for names of contributors.
-#include "db/db_impl/db_impl.h"
-
 #include <cinttypes>
 #include <set>
 #include <unordered_set>
+
+#include "db/db_impl/db_impl.h"
 #include "db/event_helpers.h"
 #include "db/memtable_list.h"
 #include "file/file_util.h"
 #include "file/filename.h"
 #include "file/sst_file_manager_impl.h"
+#include "port/port.h"
 #include "util/autovector.h"
 
 namespace ROCKSDB_NAMESPACE {
@@ -190,7 +191,8 @@ void DBImpl::FindObsoleteFiles(JobContext* job_context, bool force,
       // set of all files in the directory. We'll exclude files that are still
       // alive in the subsequent processings.
       std::vector<std::string> files;
-      env_->GetChildren(path, &files).PermitUncheckedError();  // Ignore errors
+      Status s = env_->GetChildren(path, &files);
+      s.PermitUncheckedError();  // TODO: What should we do on error?
       for (const std::string& file : files) {
         uint64_t number;
         FileType type;
@@ -206,7 +208,8 @@ void DBImpl::FindObsoleteFiles(JobContext* job_context, bool force,
           continue;
         }
 
-        // TODO(icanadi) clean up this mess to avoid having one-off "/" prefixes
+        // TODO(icanadi) clean up this mess to avoid having one-off "/"
+        // prefixes
         job_context->full_scan_candidate_files.emplace_back("/" + file, path);
       }
     }
@@ -214,9 +217,8 @@ void DBImpl::FindObsoleteFiles(JobContext* job_context, bool force,
     // Add log files in wal_dir
     if (immutable_db_options_.wal_dir != dbname_) {
       std::vector<std::string> log_files;
-      env_->GetChildren(immutable_db_options_.wal_dir,
-                        &log_files)
-          .PermitUncheckedError();  // Ignore errors
+      Status s = env_->GetChildren(immutable_db_options_.wal_dir, &log_files);
+      s.PermitUncheckedError();  // TODO: What should we do on error?
       for (const std::string& log_file : log_files) {
         job_context->full_scan_candidate_files.emplace_back(
             log_file, immutable_db_options_.wal_dir);
@@ -226,9 +228,9 @@ void DBImpl::FindObsoleteFiles(JobContext* job_context, bool force,
     if (!immutable_db_options_.db_log_dir.empty() &&
         immutable_db_options_.db_log_dir != dbname_) {
       std::vector<std::string> info_log_files;
-      // Ignore errors
-      env_->GetChildren(immutable_db_options_.db_log_dir, &info_log_files)
-          .PermitUncheckedError();
+      Status s =
+          env_->GetChildren(immutable_db_options_.db_log_dir, &info_log_files);
+      s.PermitUncheckedError();  // TODO: What should we do on error?
       for (std::string& log_file : info_log_files) {
         job_context->full_scan_candidate_files.emplace_back(
             log_file, immutable_db_options_.db_log_dir);
@@ -658,13 +660,15 @@ uint64_t FindMinPrepLogReferencedByMemTable(
 
   // we must look through the memtables for two phase transactions
   // that have been committed but not yet flushed
+  std::unordered_set<MemTable*> memtables_to_flush_set(
+      memtables_to_flush.begin(), memtables_to_flush.end());
   for (auto loop_cfd : *vset->GetColumnFamilySet()) {
     if (loop_cfd->IsDropped() || loop_cfd == cfd_to_flush) {
       continue;
     }
 
     auto log = loop_cfd->imm()->PrecomputeMinLogContainingPrepSection(
-        memtables_to_flush);
+        &memtables_to_flush_set);
 
     if (log > 0 && (min_log == 0 || log < min_log)) {
       min_log = log;
@@ -680,6 +684,37 @@ uint64_t FindMinPrepLogReferencedByMemTable(
   return min_log;
 }
 
+uint64_t FindMinPrepLogReferencedByMemTable(
+    VersionSet* vset, const autovector<ColumnFamilyData*>& cfds_to_flush,
+    const autovector<const autovector<MemTable*>*>& memtables_to_flush) {
+  uint64_t min_log = 0;
+
+  std::unordered_set<ColumnFamilyData*> cfds_to_flush_set(cfds_to_flush.begin(),
+                                                          cfds_to_flush.end());
+  std::unordered_set<MemTable*> memtables_to_flush_set;
+  for (const autovector<MemTable*>* memtables : memtables_to_flush) {
+    memtables_to_flush_set.insert(memtables->begin(), memtables->end());
+  }
+  for (auto loop_cfd : *vset->GetColumnFamilySet()) {
+    if (loop_cfd->IsDropped() || cfds_to_flush_set.count(loop_cfd)) {
+      continue;
+    }
+
+    auto log = loop_cfd->imm()->PrecomputeMinLogContainingPrepSection(
+        &memtables_to_flush_set);
+    if (log > 0 && (min_log == 0 || log < min_log)) {
+      min_log = log;
+    }
+
+    log = loop_cfd->mem()->GetMinLogContainingPrepSection();
+    if (log > 0 && (min_log == 0 || log < min_log)) {
+      min_log = log;
+    }
+  }
+
+  return min_log;
+}
+
 uint64_t PrecomputeMinLogNumberToKeepNon2PC(
     VersionSet* vset, const ColumnFamilyData& cfd_to_flush,
     const autovector<VersionEdit*>& edit_list) {
@@ -710,6 +745,42 @@ uint64_t PrecomputeMinLogNumberToKeepNon2PC(
   return min_log_number_to_keep;
 }
 
+uint64_t PrecomputeMinLogNumberToKeepNon2PC(
+    VersionSet* vset, const autovector<ColumnFamilyData*>& cfds_to_flush,
+    const autovector<autovector<VersionEdit*>>& edit_lists) {
+  assert(vset != nullptr);
+  assert(!cfds_to_flush.empty());
+  assert(cfds_to_flush.size() == edit_lists.size());
+
+  uint64_t min_log_number_to_keep = port::kMaxUint64;
+  for (const auto& edit_list : edit_lists) {
+    uint64_t log = 0;
+    for (const auto& e : edit_list) {
+      if (e->HasLogNumber()) {
+        log = std::max(log, e->GetLogNumber());
+      }
+    }
+    if (log != 0) {
+      min_log_number_to_keep = std::min(min_log_number_to_keep, log);
+    }
+  }
+  if (min_log_number_to_keep == port::kMaxUint64) {
+    min_log_number_to_keep = cfds_to_flush[0]->GetLogNumber();
+    for (size_t i = 1; i < cfds_to_flush.size(); i++) {
+      min_log_number_to_keep =
+          std::min(min_log_number_to_keep, cfds_to_flush[i]->GetLogNumber());
+    }
+  }
+
+  std::unordered_set<const ColumnFamilyData*> flushed_cfds(
+      cfds_to_flush.begin(), cfds_to_flush.end());
+  min_log_number_to_keep =
+      std::min(min_log_number_to_keep,
+               vset->PreComputeMinLogNumberWithUnflushedData(flushed_cfds));
+
+  return min_log_number_to_keep;
+}
+
 uint64_t PrecomputeMinLogNumberToKeep2PC(
     VersionSet* vset, const ColumnFamilyData& cfd_to_flush,
     const autovector<VersionEdit*>& edit_list,
@@ -751,6 +822,38 @@ uint64_t PrecomputeMinLogNumberToKeep2PC(
   return min_log_number_to_keep;
 }
 
+uint64_t PrecomputeMinLogNumberToKeep2PC(
+    VersionSet* vset, const autovector<ColumnFamilyData*>& cfds_to_flush,
+    const autovector<autovector<VersionEdit*>>& edit_lists,
+    const autovector<const autovector<MemTable*>*>& memtables_to_flush,
+    LogsWithPrepTracker* prep_tracker) {
+  assert(vset != nullptr);
+  assert(prep_tracker != nullptr);
+  assert(cfds_to_flush.size() == edit_lists.size());
+  assert(cfds_to_flush.size() == memtables_to_flush.size());
+
+  uint64_t min_log_number_to_keep =
+      PrecomputeMinLogNumberToKeepNon2PC(vset, cfds_to_flush, edit_lists);
+
+  uint64_t min_log_in_prep_heap =
+      prep_tracker->FindMinLogContainingOutstandingPrep();
+
+  if (min_log_in_prep_heap != 0 &&
+      min_log_in_prep_heap < min_log_number_to_keep) {
+    min_log_number_to_keep = min_log_in_prep_heap;
+  }
+
+  uint64_t min_log_refed_by_mem = FindMinPrepLogReferencedByMemTable(
+      vset, cfds_to_flush, memtables_to_flush);
+
+  if (min_log_refed_by_mem != 0 &&
+      min_log_refed_by_mem < min_log_number_to_keep) {
+    min_log_number_to_keep = min_log_refed_by_mem;
+  }
+
+  return min_log_number_to_keep;
+}
+
 Status DBImpl::SetDBId() {
   Status s;
   // Happens when immutable_db_options_.write_dbid_to_manifest is set to true

db/db_impl/db_impl_open.cc

@@ -147,7 +147,8 @@ DBOptions SanitizeOptions(const std::string& dbname, const DBOptions& src) {
     // DeleteScheduler::CleanupDirectory on the same dir later, it will be
     // safe
     std::vector<std::string> filenames;
-    result.env->GetChildren(result.wal_dir, &filenames).PermitUncheckedError();
+    Status s = result.env->GetChildren(result.wal_dir, &filenames);
+    s.PermitUncheckedError();  //**TODO: What to do on error?
     for (std::string& filename : filenames) {
       if (filename.find(".log.trash", filename.length() -
                                           std::string(".log.trash").length()) !=
@@ -163,7 +164,8 @@ DBOptions SanitizeOptions(const std::string& dbname, const DBOptions& src) {
   // was not used)
   auto sfm = static_cast<SstFileManagerImpl*>(result.sst_file_manager.get());
   for (size_t i = 0; i < result.db_paths.size(); i++) {
-    DeleteScheduler::CleanupDirectory(result.env, sfm, result.db_paths[i].path);
+    DeleteScheduler::CleanupDirectory(result.env, sfm, result.db_paths[i].path)
+        .PermitUncheckedError();
   }
 
   // Create a default SstFileManager for purposes of tracking compaction size
@@ -287,11 +289,13 @@ Status DBImpl::NewDB(std::vector<std::string>* new_filenames) {
     if (!s.ok()) {
       return s;
     }
+    FileTypeSet tmp_set = immutable_db_options_.checksum_handoff_file_types;
     file->SetPreallocationBlockSize(
         immutable_db_options_.manifest_preallocation_size);
     std::unique_ptr<WritableFileWriter> file_writer(new WritableFileWriter(
         std::move(file), manifest, file_options, env_, io_tracer_,
-        nullptr /* stats */, immutable_db_options_.listeners));
+        nullptr /* stats */, immutable_db_options_.listeners, nullptr,
+        tmp_set.Contains(FileType::kDescriptorFile)));
     log::Writer log(std::move(file_writer), 0, false);
     std::string record;
     new_db.EncodeTo(&record);
@@ -1368,6 +1372,9 @@ Status DBImpl::WriteLevel0TableForRecovery(int job_id, ColumnFamilyData* cfd,
                       cfd->GetName().c_str(), meta.fd.GetNumber(),
                       meta.fd.GetFileSize(), s.ToString().c_str());
       mutex_.Lock();
+
+      io_s.PermitUncheckedError();  // TODO(AR) is this correct, or should we
+                                    // return io_s if not ok()?
     }
   }
   ReleaseFileNumberFromPendingOutputs(pending_outputs_inserted_elem);
@@ -1386,7 +1393,9 @@ Status DBImpl::WriteLevel0TableForRecovery(int job_id, ColumnFamilyData* cfd,
                   meta.oldest_ancester_time, meta.file_creation_time,
                   meta.file_checksum, meta.file_checksum_func_name);
 
-    edit->SetBlobFileAdditions(std::move(blob_file_additions));
+    for (const auto& blob : blob_file_additions) {
+      edit->AddBlobFile(blob);
+    }
   }
 
   InternalStats::CompactionStats stats(CompactionReason::kFlush, 1);
@@ -1480,9 +1489,11 @@ IOStatus DBImpl::CreateWAL(uint64_t log_file_num, uint64_t recycle_log_number,
     lfile->SetPreallocationBlockSize(preallocate_block_size);
 
     const auto& listeners = immutable_db_options_.listeners;
+    FileTypeSet tmp_set = immutable_db_options_.checksum_handoff_file_types;
     std::unique_ptr<WritableFileWriter> file_writer(new WritableFileWriter(
         std::move(lfile), log_fname, opt_file_options, env_, io_tracer_,
-        nullptr /* stats */, listeners));
+        nullptr /* stats */, listeners, nullptr,
+        tmp_set.Contains(FileType::kWalFile)));
     *new_log = new log::Writer(std::move(file_writer), log_file_num,
                                immutable_db_options_.recycle_log_file_num > 0,
                                immutable_db_options_.manual_wal_flush);
@@ -1563,6 +1574,7 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
       InstrumentedMutexLock wl(&impl->log_write_mutex_);
       impl->logfile_number_ = new_log_number;
       assert(new_log != nullptr);
+      assert(impl->logs_.empty());
       impl->logs_.emplace_back(new_log_number, new_log);
     }
 
@@ -1735,9 +1747,8 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
     paths.erase(std::unique(paths.begin(), paths.end()), paths.end());
     for (auto& path : paths) {
       std::vector<std::string> existing_files;
-      // TODO: Check for errors here?
       impl->immutable_db_options_.env->GetChildren(path, &existing_files)
-          .PermitUncheckedError();
+          .PermitUncheckedError();  //**TODO: What do to on error?
       for (auto& file_name : existing_files) {
         uint64_t file_number;
         FileType file_type;

db/db_impl/db_impl_readonly.cc

@@ -83,7 +83,7 @@ Iterator* DBImplReadOnly::NewIterator(const ReadOptions& read_options,
   ReadCallback* read_callback = nullptr;  // No read callback provided.
   auto db_iter = NewArenaWrappedDbIterator(
       env_, read_options, *cfd->ioptions(), super_version->mutable_cf_options,
-      read_seq,
+      super_version->current, read_seq,
       super_version->mutable_cf_options.max_sequential_skip_in_iterations,
       super_version->version_number, read_callback);
   auto internal_iter = NewInternalIterator(
@@ -115,7 +115,8 @@ Status DBImplReadOnly::NewIterators(
     auto* cfd = static_cast_with_check<ColumnFamilyHandleImpl>(cfh)->cfd();
     auto* sv = cfd->GetSuperVersion()->Ref();
     auto* db_iter = NewArenaWrappedDbIterator(
-        env_, read_options, *cfd->ioptions(), sv->mutable_cf_options, read_seq,
+        env_, read_options, *cfd->ioptions(), sv->mutable_cf_options,
+        sv->current, read_seq,
         sv->mutable_cf_options.max_sequential_skip_in_iterations,
         sv->version_number, read_callback);
     auto* internal_iter = NewInternalIterator(
@@ -142,9 +143,6 @@ Status OpenForReadOnlyCheckExistence(const DBOptions& db_options,
     uint64_t manifest_file_number;
     s = VersionSet::GetCurrentManifestPath(dbname, fs.get(), &manifest_path,
                                            &manifest_file_number);
-    if (!s.ok()) {
-      return Status::NotFound(CurrentFileName(dbname), "does not exist");
-    }
   }
   return s;
 }

db/db_impl/db_impl_secondary.cc

@@ -421,7 +421,7 @@ ArenaWrappedDBIter* DBImplSecondary::NewIteratorImpl(
   SuperVersion* super_version = cfd->GetReferencedSuperVersion(this);
   auto db_iter = NewArenaWrappedDbIterator(
       env_, read_options, *cfd->ioptions(), super_version->mutable_cf_options,
-      snapshot,
+      super_version->current, snapshot,
       super_version->mutable_cf_options.max_sequential_skip_in_iterations,
       super_version->version_number, read_callback);
   auto internal_iter = NewInternalIterator(

db/db_impl/db_impl_write.cc

@@ -163,7 +163,6 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
   StopWatch write_sw(env_, immutable_db_options_.statistics.get(), DB_WRITE);
 
   write_thread_.JoinBatchGroup(&w);
-  Status status;
   if (w.state == WriteThread::STATE_PARALLEL_MEMTABLE_WRITER) {
     // we are a non-leader in a parallel group
 
@@ -193,8 +192,6 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
     }
     assert(w.state == WriteThread::STATE_COMPLETED);
     // STATE_COMPLETED conditional below handles exit
-
-    status = w.FinalStatus();
   }
   if (w.state == WriteThread::STATE_COMPLETED) {
     if (log_used != nullptr) {
@@ -204,13 +201,11 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
       *seq_used = w.sequence;
     }
     // write is complete and leader has updated sequence
-    // Should we handle it?
-    status.PermitUncheckedError();
     return w.FinalStatus();
   }
   // else we are the leader of the write batch group
   assert(w.state == WriteThread::STATE_GROUP_LEADER);
-
+  Status status;
   // Once reaches this point, the current writer "w" will try to do its write
   // job.  It may also pick up some of the remaining writers in the "writers_"
   // when it finds suitable, and finish them in the same write batch.
@@ -531,6 +526,8 @@ Status DBImpl::PipelinedWriteImpl(const WriteOptions& write_options,
     PERF_TIMER_STOP(write_pre_and_post_process_time);
 
     IOStatus io_s;
+    io_s.PermitUncheckedError();  // Allow io_s to be uninitialized
+
     if (w.status.ok() && !write_options.disableWAL) {
       PERF_TIMER_GUARD(write_wal_time);
       stats->AddDBStats(InternalStats::kIntStatsWriteDoneBySelf, 1);
@@ -566,7 +563,12 @@ Status DBImpl::PipelinedWriteImpl(const WriteOptions& write_options,
     write_thread_.ExitAsBatchGroupLeader(wal_write_group, w.status);
   }
 
+  // NOTE: the memtable_write_group is declared before the following
+  // `if` statement because its lifetime needs to be longer
+  // that the inner context  of the `if` as a reference to it
+  // may be used further below within the outer _write_thread
   WriteThread::WriteGroup memtable_write_group;
+
   if (w.state == WriteThread::STATE_MEMTABLE_WRITER_LEADER) {
     PERF_TIMER_GUARD(write_memtable_time);
     assert(w.ShouldWriteToMemtable());
@@ -583,6 +585,10 @@ Status DBImpl::PipelinedWriteImpl(const WriteOptions& write_options,
       versions_->SetLastSequence(memtable_write_group.last_sequence);
       write_thread_.ExitAsMemTableWriter(&w, memtable_write_group);
     }
+  } else {
+    // NOTE: the memtable_write_group is never really used,
+    // so we need to set its status to pass ASSERT_STATUS_CHECKED
+    memtable_write_group.status.PermitUncheckedError();
   }
 
   if (w.state == WriteThread::STATE_PARALLEL_MEMTABLE_WRITER) {
@@ -666,7 +672,6 @@ Status DBImpl::WriteImplWALOnly(
     const uint64_t log_ref, uint64_t* seq_used, const size_t sub_batch_cnt,
     PreReleaseCallback* pre_release_callback, const AssignOrder assign_order,
     const PublishLastSeq publish_last_seq, const bool disable_memtable) {
-  Status status;
   PERF_TIMER_GUARD(write_pre_and_post_process_time);
   WriteThread::Writer w(write_options, my_batch, callback, log_ref,
                         disable_memtable, sub_batch_cnt, pre_release_callback);
@@ -688,6 +693,8 @@ Status DBImpl::WriteImplWALOnly(
   assert(w.state == WriteThread::STATE_GROUP_LEADER);
 
   if (publish_last_seq == kDoPublishLastSeq) {
+    Status status;
+
     // Currently we only use kDoPublishLastSeq in unordered_write
     assert(immutable_db_options_.unordered_write);
     WriteContext write_context;
@@ -764,7 +771,9 @@ Status DBImpl::WriteImplWALOnly(
     }
     seq_inc = total_batch_cnt;
   }
+  Status status;
   IOStatus io_s;
+  io_s.PermitUncheckedError();  // Allow io_s to be uninitialized
   if (!write_options.disableWAL) {
     io_s = ConcurrentWriteToWAL(write_group, log_used, &last_sequence, seq_inc);
     status = io_s;
@@ -850,8 +859,7 @@ void DBImpl::WriteStatusCheckOnLocked(const Status& status) {
   if (immutable_db_options_.paranoid_checks && !status.ok() &&
       !status.IsBusy() && !status.IsIncomplete()) {
     // Maybe change the return status to void?
-    error_handler_.SetBGError(status, BackgroundErrorReason::kWriteCallback)
-        .PermitUncheckedError();
+    error_handler_.SetBGError(status, BackgroundErrorReason::kWriteCallback);
   }
 }
 
@@ -863,8 +871,7 @@ void DBImpl::WriteStatusCheck(const Status& status) {
       !status.IsBusy() && !status.IsIncomplete()) {
     mutex_.Lock();
     // Maybe change the return status to void?
-    error_handler_.SetBGError(status, BackgroundErrorReason::kWriteCallback)
-        .PermitUncheckedError();
+    error_handler_.SetBGError(status, BackgroundErrorReason::kWriteCallback);
     mutex_.Unlock();
   }
 }
@@ -877,8 +884,7 @@ void DBImpl::IOStatusCheck(const IOStatus& io_status) {
       io_status.IsIOFenced()) {
     mutex_.Lock();
     // Maybe change the return status to void?
-    error_handler_.SetBGError(io_status, BackgroundErrorReason::kWriteCallback)
-        .PermitUncheckedError();
+    error_handler_.SetBGError(io_status, BackgroundErrorReason::kWriteCallback);
     mutex_.Unlock();
   }
 }
@@ -1335,10 +1341,17 @@ Status DBImpl::SwitchWAL(WriteContext* write_context) {
     }
     for (auto cfd : cfds) {
       cfd->imm()->FlushRequested();
+      if (!immutable_db_options_.atomic_flush) {
+        FlushRequest flush_req;
+        GenerateFlushRequest({cfd}, &flush_req);
+        SchedulePendingFlush(flush_req, FlushReason::kWriteBufferManager);
+      }
+    }
+    if (immutable_db_options_.atomic_flush) {
+      FlushRequest flush_req;
+      GenerateFlushRequest(cfds, &flush_req);
+      SchedulePendingFlush(flush_req, FlushReason::kWriteBufferManager);
     }
-    FlushRequest flush_req;
-    GenerateFlushRequest(cfds, &flush_req);
-    SchedulePendingFlush(flush_req, FlushReason::kWriteBufferManager);
     MaybeScheduleFlushOrCompaction();
   }
   return status;
@@ -1414,10 +1427,17 @@ Status DBImpl::HandleWriteBufferFull(WriteContext* write_context) {
     }
     for (const auto cfd : cfds) {
       cfd->imm()->FlushRequested();
+      if (!immutable_db_options_.atomic_flush) {
+        FlushRequest flush_req;
+        GenerateFlushRequest({cfd}, &flush_req);
+        SchedulePendingFlush(flush_req, FlushReason::kWriteBufferFull);
+      }
+    }
+    if (immutable_db_options_.atomic_flush) {
+      FlushRequest flush_req;
+      GenerateFlushRequest(cfds, &flush_req);
+      SchedulePendingFlush(flush_req, FlushReason::kWriteBufferFull);
     }
-    FlushRequest flush_req;
-    GenerateFlushRequest(cfds, &flush_req);
-    SchedulePendingFlush(flush_req, FlushReason::kWriteBufferFull);
     MaybeScheduleFlushOrCompaction();
   }
   return status;
@@ -1641,10 +1661,16 @@ Status DBImpl::ScheduleFlushes(WriteContext* context) {
   if (status.ok()) {
     if (immutable_db_options_.atomic_flush) {
       AssignAtomicFlushSeq(cfds);
+      FlushRequest flush_req;
+      GenerateFlushRequest(cfds, &flush_req);
+      SchedulePendingFlush(flush_req, FlushReason::kWriteBufferFull);
+    } else {
+      for (auto* cfd : cfds) {
+        FlushRequest flush_req;
+        GenerateFlushRequest({cfd}, &flush_req);
+        SchedulePendingFlush(flush_req, FlushReason::kWriteBufferFull);
+      }
     }
-    FlushRequest flush_req;
-    GenerateFlushRequest(cfds, &flush_req);
-    SchedulePendingFlush(flush_req, FlushReason::kWriteBufferFull);
     MaybeScheduleFlushOrCompaction();
   }
   return status;
@@ -1790,32 +1816,74 @@ Status DBImpl::SwitchMemtable(ColumnFamilyData* cfd, WriteContext* context) {
     }
     // We may have lost data from the WritableFileBuffer in-memory buffer for
     // the current log, so treat it as a fatal error and set bg_error
-    // Should handle return error?
     if (!io_s.ok()) {
-      // Should handle return error?
-      error_handler_.SetBGError(io_s, BackgroundErrorReason::kMemTable)
-          .PermitUncheckedError();
+      error_handler_.SetBGError(io_s, BackgroundErrorReason::kMemTable);
     } else {
-      // Should handle return error?
-      error_handler_.SetBGError(s, BackgroundErrorReason::kMemTable)
-          .PermitUncheckedError();
+      error_handler_.SetBGError(s, BackgroundErrorReason::kMemTable);
     }
     // Read back bg_error in order to get the right severity
     s = error_handler_.GetBGError();
     return s;
   }
 
-  for (auto loop_cfd : *versions_->GetColumnFamilySet()) {
-    // all this is just optimization to delete logs that
-    // are no longer needed -- if CF is empty, that means it
-    // doesn't need that particular log to stay alive, so we just
-    // advance the log number. no need to persist this in the manifest
-    if (loop_cfd->mem()->GetFirstSequenceNumber() == 0 &&
-        loop_cfd->imm()->NumNotFlushed() == 0) {
-      if (creating_new_log) {
-        loop_cfd->SetLogNumber(logfile_number_);
+  bool empty_cf_updated = false;
+  if (immutable_db_options_.track_and_verify_wals_in_manifest &&
+      !immutable_db_options_.allow_2pc && creating_new_log) {
+    // In non-2pc mode, WALs become obsolete if they do not contain unflushed
+    // data. Updating the empty CF's log number might cause some WALs to become
+    // obsolete. So we should track the WAL obsoletion event before actually
+    // updating the empty CF's log number.
+    uint64_t min_wal_number_to_keep =
+        versions_->PreComputeMinLogNumberWithUnflushedData(logfile_number_);
+    if (min_wal_number_to_keep >
+        versions_->GetWalSet().GetMinWalNumberToKeep()) {
+      // Get a snapshot of the empty column families.
+      // LogAndApply may release and reacquire db
+      // mutex, during that period, column family may become empty (e.g. its
+      // flush succeeds), then it affects the computed min_log_number_to_keep,
+      // so we take a snapshot for consistency of column family data
+      // status. If a column family becomes non-empty afterwards, its active log
+      // should still be the created new log, so the min_log_number_to_keep is
+      // not affected.
+      autovector<ColumnFamilyData*> empty_cfs;
+      for (auto cf : *versions_->GetColumnFamilySet()) {
+        if (cf->IsEmpty()) {
+          empty_cfs.push_back(cf);
+        }
+      }
+
+      VersionEdit wal_deletion;
+      wal_deletion.DeleteWalsBefore(min_wal_number_to_keep);
+      s = versions_->LogAndApplyToDefaultColumnFamily(&wal_deletion, &mutex_);
+      if (!s.ok() && versions_->io_status().IsIOError()) {
+        s = error_handler_.SetBGError(versions_->io_status(),
+                                      BackgroundErrorReason::kManifestWrite);
+      }
+      if (!s.ok()) {
+        return s;
+      }
+
+      for (auto cf : empty_cfs) {
+        if (cf->IsEmpty()) {
+          cf->SetLogNumber(logfile_number_);
+          cf->mem()->SetCreationSeq(versions_->LastSequence());
+        }  // cf may become non-empty.
+      }
+      empty_cf_updated = true;
+    }
+  }
+  if (!empty_cf_updated) {
+    for (auto cf : *versions_->GetColumnFamilySet()) {
+      // all this is just optimization to delete logs that
+      // are no longer needed -- if CF is empty, that means it
+      // doesn't need that particular log to stay alive, so we just
+      // advance the log number. no need to persist this in the manifest
+      if (cf->IsEmpty()) {
+        if (creating_new_log) {
+          cf->SetLogNumber(logfile_number_);
+        }
+        cf->mem()->SetCreationSeq(versions_->LastSequence());
       }
-      loop_cfd->mem()->SetCreationSeq(versions_->LastSequence());
     }
   }
 

db/db_impl/db_secondary_test.cc

@@ -748,10 +748,12 @@ TEST_F(DBSecondaryTest, SwitchWALMultiColumnFamilies) {
     }
   };
   for (int k = 0; k != 8; ++k) {
-    ASSERT_OK(
-        Put(0 /*cf*/, "key" + std::to_string(k), "value" + std::to_string(k)));
-    ASSERT_OK(
-        Put(1 /*cf*/, "key" + std::to_string(k), "value" + std::to_string(k)));
+    for (int j = 0; j < 2; ++j) {
+      ASSERT_OK(Put(0 /*cf*/, "key" + std::to_string(k),
+                    "value" + std::to_string(k)));
+      ASSERT_OK(Put(1 /*cf*/, "key" + std::to_string(k),
+                    "value" + std::to_string(k)));
+    }
     TEST_SYNC_POINT(
         "DBSecondaryTest::SwitchWALMultipleColumnFamilies:BeforeCatchUp");
     ASSERT_OK(db_secondary_->TryCatchUpWithPrimary());

db/db_info_dumper.cc

@@ -64,9 +64,10 @@ void DumpDBFileSummary(const ImmutableDBOptions& options,
         break;
       case kWalFile:
         if (env->GetFileSize(dbname + "/" + file, &file_size).ok()) {
-          char str[16];
-          snprintf(str, sizeof(str), "%" PRIu64, file_size);
-          wal_info.append(file).append(" size: ").append(str).append(" ; ");
+          wal_info.append(file)
+              .append(" size: ")
+              .append(std::to_string(file_size))
+              .append(" ; ");
         } else {
           Error(options.info_log, "Error when reading LOG file: %s/%s\n",
                 dbname.c_str(), file.c_str());
@@ -120,9 +121,10 @@ void DumpDBFileSummary(const ImmutableDBOptions& options,
       if (ParseFileName(file, &number, &type)) {
         if (type == kWalFile) {
           if (env->GetFileSize(options.wal_dir + "/" + file, &file_size).ok()) {
-            char str[16];
-            snprintf(str, sizeof(str), "%" PRIu64, file_size);
-            wal_info.append(file).append(" size: ").append(str).append(" ; ");
+            wal_info.append(file)
+                .append(" size: ")
+                .append(std::to_string(file_size))
+                .append(" ; ");
           } else {
             Error(options.info_log, "Error when reading LOG file %s/%s\n",
                   options.wal_dir.c_str(), file.c_str());

db/db_io_failure_test.cc

@@ -43,11 +43,15 @@ TEST_F(DBIOFailureTest, DropWrites) {
           if (level > 0 && level == dbfull()->NumberLevels() - 1) {
             break;
           }
-          dbfull()->TEST_CompactRange(level, nullptr, nullptr, nullptr,
-                                      true /* disallow trivial move */);
+          Status s =
+              dbfull()->TEST_CompactRange(level, nullptr, nullptr, nullptr,
+                                          true /* disallow trivial move */);
+          ASSERT_TRUE(s.ok() || s.IsCorruption());
         }
       } else {
-        dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr);
+        Status s =
+            dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr);
+        ASSERT_TRUE(s.ok() || s.IsCorruption());
       }
     }
 
@@ -56,7 +60,8 @@ TEST_F(DBIOFailureTest, DropWrites) {
     ASSERT_EQ("5", property_value);
 
     env_->drop_writes_.store(false, std::memory_order_release);
-    ASSERT_LT(CountFiles(), num_files + 3);
+    const size_t count = CountFiles();
+    ASSERT_LT(count, num_files + 3);
 
     // Check that compaction attempts slept after errors
     // TODO @krad: Figure out why ASSERT_EQ 5 keeps failing in certain compiler
@@ -82,7 +87,8 @@ TEST_F(DBIOFailureTest, DropWritesFlush) {
     ASSERT_TRUE(db_->GetProperty("rocksdb.background-errors", &property_value));
     ASSERT_EQ("0", property_value);
 
-    dbfull()->TEST_FlushMemTable(true);
+    // ASSERT file is too short
+    ASSERT_TRUE(dbfull()->TEST_FlushMemTable(true).IsCorruption());
 
     ASSERT_TRUE(db_->GetProperty("rocksdb.background-errors", &property_value));
     ASSERT_EQ("1", property_value);
@@ -166,7 +172,7 @@ TEST_F(DBIOFailureTest, ManifestWriteError) {
     ASSERT_EQ("bar", Get("foo"));
 
     // Memtable compaction (will succeed)
-    Flush();
+    ASSERT_OK(Flush());
     ASSERT_EQ("bar", Get("foo"));
     const int last = 2;
     MoveFilesToLevel(2);
@@ -174,7 +180,8 @@ TEST_F(DBIOFailureTest, ManifestWriteError) {
 
     // Merging compaction (will fail)
     error_type->store(true, std::memory_order_release);
-    dbfull()->TEST_CompactRange(last, nullptr, nullptr);  // Should fail
+    ASSERT_NOK(
+        dbfull()->TEST_CompactRange(last, nullptr, nullptr));  // Should fail
     ASSERT_EQ("bar", Get("foo"));
 
     error_type->store(false, std::memory_order_release);
@@ -192,7 +199,13 @@ TEST_F(DBIOFailureTest, ManifestWriteError) {
 
     // Merging compaction (will fail)
     error_type->store(true, std::memory_order_release);
-    dbfull()->TEST_CompactRange(last, nullptr, nullptr);  // Should fail
+    Status s =
+        dbfull()->TEST_CompactRange(last, nullptr, nullptr);  // Should fail
+    if (iter == 0) {
+      ASSERT_OK(s);
+    } else {
+      ASSERT_TRUE(s.IsIOError());
+    }
     ASSERT_EQ("bar", Get("foo"));
 
     // Recovery: should not lose data
@@ -220,18 +233,15 @@ TEST_F(DBIOFailureTest, PutFailsParanoid) {
   options.paranoid_checks = true;
   DestroyAndReopen(options);
   CreateAndReopenWithCF({"pikachu"}, options);
-  Status s;
 
   ASSERT_OK(Put(1, "foo", "bar"));
   ASSERT_OK(Put(1, "foo1", "bar1"));
   // simulate error
   env_->log_write_error_.store(true, std::memory_order_release);
-  s = Put(1, "foo2", "bar2");
-  ASSERT_TRUE(!s.ok());
+  ASSERT_NOK(Put(1, "foo2", "bar2"));
   env_->log_write_error_.store(false, std::memory_order_release);
-  s = Put(1, "foo3", "bar3");
   // the next put should fail, too
-  ASSERT_TRUE(!s.ok());
+  ASSERT_NOK(Put(1, "foo3", "bar3"));
   // but we're still able to read
   ASSERT_EQ("bar", Get(1, "foo"));
 
@@ -244,12 +254,10 @@ TEST_F(DBIOFailureTest, PutFailsParanoid) {
   ASSERT_OK(Put(1, "foo1", "bar1"));
   // simulate error
   env_->log_write_error_.store(true, std::memory_order_release);
-  s = Put(1, "foo2", "bar2");
-  ASSERT_TRUE(!s.ok());
+  ASSERT_NOK(Put(1, "foo2", "bar2"));
   env_->log_write_error_.store(false, std::memory_order_release);
-  s = Put(1, "foo3", "bar3");
   // the next put should NOT fail
-  ASSERT_TRUE(s.ok());
+  ASSERT_OK(Put(1, "foo3", "bar3"));
 }
 #if !(defined NDEBUG) || !defined(OS_WIN)
 TEST_F(DBIOFailureTest, FlushSstRangeSyncError) {
@@ -269,14 +277,14 @@ TEST_F(DBIOFailureTest, FlushSstRangeSyncError) {
 
   DestroyAndReopen(options);
   CreateAndReopenWithCF({"pikachu"}, options);
-  Status s;
 
+  const char* io_error_msg = "range sync dummy error";
   std::atomic<int> range_sync_called(0);
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
       "SpecialEnv::SStableFile::RangeSync", [&](void* arg) {
         if (range_sync_called.fetch_add(1) == 0) {
           Status* st = static_cast<Status*>(arg);
-          *st = Status::IOError("range sync dummy error");
+          *st = Status::IOError(io_error_msg);
         }
       });
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
@@ -298,7 +306,9 @@ TEST_F(DBIOFailureTest, FlushSstRangeSyncError) {
   ASSERT_OK(Put(1, "foo3_2", rnd_str));
   ASSERT_OK(Put(1, "foo3_3", rnd_str));
   ASSERT_OK(Put(1, "foo4", "bar"));
-  dbfull()->TEST_WaitForFlushMemTable(handles_[1]);
+  Status s = dbfull()->TEST_WaitForFlushMemTable(handles_[1]);
+  ASSERT_TRUE(s.IsIOError());
+  ASSERT_STREQ(s.getState(), io_error_msg);
 
   // Following writes should fail as flush failed.
   ASSERT_NOK(Put(1, "foo2", "bar3"));
@@ -328,7 +338,6 @@ TEST_F(DBIOFailureTest, CompactSstRangeSyncError) {
   options.table_factory.reset(NewBlockBasedTableFactory(table_options));
   DestroyAndReopen(options);
   CreateAndReopenWithCF({"pikachu"}, options);
-  Status s;
 
   Random rnd(301);
   std::string rnd_str =
@@ -342,21 +351,22 @@ TEST_F(DBIOFailureTest, CompactSstRangeSyncError) {
   ASSERT_OK(Put(1, "foo1_1", rnd_str));
   ASSERT_OK(Put(1, "foo1_2", rnd_str));
   ASSERT_OK(Put(1, "foo1_3", rnd_str));
-  Flush(1);
+  ASSERT_OK(Flush(1));
   ASSERT_OK(Put(1, "foo", "bar"));
   ASSERT_OK(Put(1, "foo3_1", rnd_str));
   ASSERT_OK(Put(1, "foo3_2", rnd_str));
   ASSERT_OK(Put(1, "foo3_3", rnd_str));
   ASSERT_OK(Put(1, "foo4", "bar"));
-  Flush(1);
-  dbfull()->TEST_WaitForFlushMemTable(handles_[1]);
+  ASSERT_OK(Flush(1));
+  ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable(handles_[1]));
 
+  const char* io_error_msg = "range sync dummy error";
   std::atomic<int> range_sync_called(0);
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
       "SpecialEnv::SStableFile::RangeSync", [&](void* arg) {
         if (range_sync_called.fetch_add(1) == 0) {
           Status* st = static_cast<Status*>(arg);
-          *st = Status::IOError("range sync dummy error");
+          *st = Status::IOError(io_error_msg);
         }
       });
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
@@ -365,7 +375,9 @@ TEST_F(DBIOFailureTest, CompactSstRangeSyncError) {
                                  {
                                      {"disable_auto_compactions", "false"},
                                  }));
-  dbfull()->TEST_WaitForCompact();
+  Status s = dbfull()->TEST_WaitForCompact();
+  ASSERT_TRUE(s.IsIOError());
+  ASSERT_STREQ(s.getState(), io_error_msg);
 
   // Following writes should fail as flush failed.
   ASSERT_NOK(Put(1, "foo2", "bar3"));
@@ -389,13 +401,14 @@ TEST_F(DBIOFailureTest, FlushSstCloseError) {
 
   DestroyAndReopen(options);
   CreateAndReopenWithCF({"pikachu"}, options);
-  Status s;
+
+  const char* io_error_msg = "close dummy error";
   std::atomic<int> close_called(0);
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
       "SpecialEnv::SStableFile::Close", [&](void* arg) {
         if (close_called.fetch_add(1) == 0) {
           Status* st = static_cast<Status*>(arg);
-          *st = Status::IOError("close dummy error");
+          *st = Status::IOError(io_error_msg);
         }
       });
 
@@ -404,7 +417,9 @@ TEST_F(DBIOFailureTest, FlushSstCloseError) {
   ASSERT_OK(Put(1, "foo", "bar"));
   ASSERT_OK(Put(1, "foo1", "bar1"));
   ASSERT_OK(Put(1, "foo", "bar2"));
-  dbfull()->TEST_WaitForFlushMemTable(handles_[1]);
+  Status s = dbfull()->TEST_WaitForFlushMemTable(handles_[1]);
+  ASSERT_TRUE(s.IsIOError());
+  ASSERT_STREQ(s.getState(), io_error_msg);
 
   // Following writes should fail as flush failed.
   ASSERT_NOK(Put(1, "foo2", "bar3"));
@@ -429,25 +444,25 @@ TEST_F(DBIOFailureTest, CompactionSstCloseError) {
 
   DestroyAndReopen(options);
   CreateAndReopenWithCF({"pikachu"}, options);
-  Status s;
 
   ASSERT_OK(Put(1, "foo", "bar"));
   ASSERT_OK(Put(1, "foo2", "bar"));
-  Flush(1);
+  ASSERT_OK(Flush(1));
   ASSERT_OK(Put(1, "foo", "bar2"));
   ASSERT_OK(Put(1, "foo2", "bar"));
-  Flush(1);
+  ASSERT_OK(Flush(1));
   ASSERT_OK(Put(1, "foo", "bar3"));
   ASSERT_OK(Put(1, "foo2", "bar"));
-  Flush(1);
-  dbfull()->TEST_WaitForCompact();
+  ASSERT_OK(Flush(1));
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
 
+  const char* io_error_msg = "close dummy error";
   std::atomic<int> close_called(0);
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
       "SpecialEnv::SStableFile::Close", [&](void* arg) {
         if (close_called.fetch_add(1) == 0) {
           Status* st = static_cast<Status*>(arg);
-          *st = Status::IOError("close dummy error");
+          *st = Status::IOError(io_error_msg);
         }
       });
 
@@ -456,7 +471,9 @@ TEST_F(DBIOFailureTest, CompactionSstCloseError) {
                                  {
                                      {"disable_auto_compactions", "false"},
                                  }));
-  dbfull()->TEST_WaitForCompact();
+  Status s = dbfull()->TEST_WaitForCompact();
+  ASSERT_TRUE(s.IsIOError());
+  ASSERT_STREQ(s.getState(), io_error_msg);
 
   // Following writes should fail as compaction failed.
   ASSERT_NOK(Put(1, "foo2", "bar3"));
@@ -480,13 +497,14 @@ TEST_F(DBIOFailureTest, FlushSstSyncError) {
 
   DestroyAndReopen(options);
   CreateAndReopenWithCF({"pikachu"}, options);
-  Status s;
+
+  const char* io_error_msg = "sync dummy error";
   std::atomic<int> sync_called(0);
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
       "SpecialEnv::SStableFile::Sync", [&](void* arg) {
         if (sync_called.fetch_add(1) == 0) {
           Status* st = static_cast<Status*>(arg);
-          *st = Status::IOError("sync dummy error");
+          *st = Status::IOError(io_error_msg);
         }
       });
 
@@ -495,7 +513,9 @@ TEST_F(DBIOFailureTest, FlushSstSyncError) {
   ASSERT_OK(Put(1, "foo", "bar"));
   ASSERT_OK(Put(1, "foo1", "bar1"));
   ASSERT_OK(Put(1, "foo", "bar2"));
-  dbfull()->TEST_WaitForFlushMemTable(handles_[1]);
+  Status s = dbfull()->TEST_WaitForFlushMemTable(handles_[1]);
+  ASSERT_TRUE(s.IsIOError());
+  ASSERT_STREQ(s.getState(), io_error_msg);
 
   // Following writes should fail as flush failed.
   ASSERT_NOK(Put(1, "foo2", "bar3"));
@@ -521,25 +541,25 @@ TEST_F(DBIOFailureTest, CompactionSstSyncError) {
 
   DestroyAndReopen(options);
   CreateAndReopenWithCF({"pikachu"}, options);
-  Status s;
 
   ASSERT_OK(Put(1, "foo", "bar"));
   ASSERT_OK(Put(1, "foo2", "bar"));
-  Flush(1);
+  ASSERT_OK(Flush(1));
   ASSERT_OK(Put(1, "foo", "bar2"));
   ASSERT_OK(Put(1, "foo2", "bar"));
-  Flush(1);
+  ASSERT_OK(Flush(1));
   ASSERT_OK(Put(1, "foo", "bar3"));
   ASSERT_OK(Put(1, "foo2", "bar"));
-  Flush(1);
-  dbfull()->TEST_WaitForCompact();
+  ASSERT_OK(Flush(1));
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
 
+  const char* io_error_msg = "sync dummy error";
   std::atomic<int> sync_called(0);
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
       "SpecialEnv::SStableFile::Sync", [&](void* arg) {
         if (sync_called.fetch_add(1) == 0) {
           Status* st = static_cast<Status*>(arg);
-          *st = Status::IOError("close dummy error");
+          *st = Status::IOError(io_error_msg);
         }
       });
 
@@ -548,7 +568,9 @@ TEST_F(DBIOFailureTest, CompactionSstSyncError) {
                                  {
                                      {"disable_auto_compactions", "false"},
                                  }));
-  dbfull()->TEST_WaitForCompact();
+  Status s = dbfull()->TEST_WaitForCompact();
+  ASSERT_TRUE(s.IsIOError());
+  ASSERT_STREQ(s.getState(), io_error_msg);
 
   // Following writes should fail as compaction failed.
   ASSERT_NOK(Put(1, "foo2", "bar3"));

db/db_iter.cc

@@ -36,16 +36,18 @@ namespace ROCKSDB_NAMESPACE {
 DBIter::DBIter(Env* _env, const ReadOptions& read_options,
                const ImmutableCFOptions& cf_options,
                const MutableCFOptions& mutable_cf_options,
-               const Comparator* cmp, InternalIterator* iter, SequenceNumber s,
-               bool arena_mode, uint64_t max_sequential_skip_in_iterations,
+               const Comparator* cmp, InternalIterator* iter,
+               const Version* version, SequenceNumber s, bool arena_mode,
+               uint64_t max_sequential_skip_in_iterations,
                ReadCallback* read_callback, DBImpl* db_impl,
-               ColumnFamilyData* cfd, bool allow_blob)
+               ColumnFamilyData* cfd, bool expose_blob_index)
     : prefix_extractor_(mutable_cf_options.prefix_extractor.get()),
       env_(_env),
       logger_(cf_options.info_log),
       user_comparator_(cmp),
       merge_operator_(cf_options.merge_operator),
       iter_(iter),
+      version_(version),
       read_callback_(read_callback),
       sequence_(s),
       statistics_(cf_options.statistics),
@@ -65,7 +67,9 @@ DBIter::DBIter(Env* _env, const ReadOptions& read_options,
       expect_total_order_inner_iter_(prefix_extractor_ == nullptr ||
                                      read_options.total_order_seek ||
                                      read_options.auto_prefix_mode),
-      allow_blob_(allow_blob),
+      read_tier_(read_options.read_tier),
+      verify_checksums_(read_options.verify_checksums),
+      expose_blob_index_(expose_blob_index),
       is_blob_(false),
       arena_mode_(arena_mode),
       range_del_agg_(&cf_options.internal_comparator, s),
@@ -165,6 +169,40 @@ void DBIter::Next() {
   }
 }
 
+bool DBIter::SetBlobValueIfNeeded(const Slice& user_key,
+                                  const Slice& blob_index) {
+  assert(!is_blob_);
+
+  if (expose_blob_index_) {  // Stacked BlobDB implementation
+    is_blob_ = true;
+    return true;
+  }
+
+  if (!version_) {
+    status_ = Status::Corruption("Encountered unexpected blob index.");
+    valid_ = false;
+    return false;
+  }
+
+  // TODO: consider moving ReadOptions from ArenaWrappedDBIter to DBIter to
+  // avoid having to copy options back and forth.
+  ReadOptions read_options;
+  read_options.read_tier = read_tier_;
+  read_options.verify_checksums = verify_checksums_;
+
+  const Status s =
+      version_->GetBlob(read_options, user_key, blob_index, &blob_value_);
+
+  if (!s.ok()) {
+    status_ = s;
+    valid_ = false;
+    return false;
+  }
+
+  is_blob_ = true;
+  return true;
+}
+
 // PRE: saved_key_ has the current user key if skipping_saved_key
 // POST: saved_key_ should have the next user key if valid_,
 //       if the current entry is a result of merge
@@ -221,25 +259,28 @@ bool DBIter::FindNextUserEntryInternal(bool skipping_saved_key,
       is_key_seqnum_zero_ = false;
       return false;
     }
+    Slice user_key_without_ts =
+        StripTimestampFromUserKey(ikey_.user_key, timestamp_size_);
 
     is_key_seqnum_zero_ = (ikey_.sequence == 0);
 
     assert(iterate_upper_bound_ == nullptr ||
            iter_.UpperBoundCheckResult() != IterBoundCheck::kInbound ||
            user_comparator_.CompareWithoutTimestamp(
-               ikey_.user_key, /*a_has_ts=*/true, *iterate_upper_bound_,
+               user_key_without_ts, /*a_has_ts=*/false, *iterate_upper_bound_,
                /*b_has_ts=*/false) < 0);
     if (iterate_upper_bound_ != nullptr &&
         iter_.UpperBoundCheckResult() != IterBoundCheck::kInbound &&
         user_comparator_.CompareWithoutTimestamp(
-            ikey_.user_key, /*a_has_ts=*/true, *iterate_upper_bound_,
+            user_key_without_ts, /*a_has_ts=*/false, *iterate_upper_bound_,
             /*b_has_ts=*/false) >= 0) {
       break;
     }
 
     assert(prefix == nullptr || prefix_extractor_ != nullptr);
     if (prefix != nullptr &&
-        prefix_extractor_->Transform(ikey_.user_key).compare(*prefix) != 0) {
+        prefix_extractor_->Transform(user_key_without_ts).compare(*prefix) !=
+            0) {
       assert(prefix_same_as_start_);
       break;
     }
@@ -316,8 +357,14 @@ bool DBIter::FindNextUserEntryInternal(bool skipping_saved_key,
           case kTypeBlobIndex:
             if (start_seqnum_ > 0) {
               if (ikey_.sequence >= start_seqnum_) {
-                assert(ikey_.type != kTypeBlobIndex);
                 saved_key_.SetInternalKey(ikey_);
+
+                if (ikey_.type == kTypeBlobIndex) {
+                  if (!SetBlobValueIfNeeded(ikey_.user_key, iter_.value())) {
+                    return false;
+                  }
+                }
+
                 valid_ = true;
                 return true;
               } else {
@@ -331,6 +378,13 @@ bool DBIter::FindNextUserEntryInternal(bool skipping_saved_key,
               }
             } else if (timestamp_lb_) {
               saved_key_.SetInternalKey(ikey_);
+
+              if (ikey_.type == kTypeBlobIndex) {
+                if (!SetBlobValueIfNeeded(ikey_.user_key, iter_.value())) {
+                  return false;
+                }
+              }
+
               valid_ = true;
               return true;
             } else {
@@ -345,20 +399,13 @@ bool DBIter::FindNextUserEntryInternal(bool skipping_saved_key,
                 num_skipped = 0;
                 reseek_done = false;
                 PERF_COUNTER_ADD(internal_delete_skipped_count, 1);
-              } else if (ikey_.type == kTypeBlobIndex) {
-                if (!allow_blob_) {
-                  ROCKS_LOG_ERROR(logger_, "Encounter unexpected blob index.");
-                  status_ = Status::NotSupported(
-                      "Encounter unexpected blob index. Please open DB with "
-                      "ROCKSDB_NAMESPACE::blob_db::BlobDB instead.");
-                  valid_ = false;
-                  return false;
+              } else {
+                if (ikey_.type == kTypeBlobIndex) {
+                  if (!SetBlobValueIfNeeded(ikey_.user_key, iter_.value())) {
+                    return false;
+                  }
                 }
 
-                is_blob_ = true;
-                valid_ = true;
-                return true;
-              } else {
                 valid_ = true;
                 return true;
               }
@@ -498,7 +545,6 @@ bool DBIter::MergeValuesNewToOld() {
   TEST_SYNC_POINT("DBIter::MergeValuesNewToOld:PushedFirstOperand");
 
   ParsedInternalKey ikey;
-  Status s;
   for (iter_.Next(); iter_.Valid(); iter_.Next()) {
     TEST_SYNC_POINT("DBIter::MergeValuesNewToOld:SteppedToNextOperand");
     if (!ParseKey(&ikey)) {
@@ -526,7 +572,7 @@ bool DBIter::MergeValuesNewToOld() {
       // hit a put, merge the put value with operands and store the
       // final result in saved_value_. We are done!
       const Slice val = iter_.value();
-      s = MergeHelper::TimedFullMerge(
+      Status s = MergeHelper::TimedFullMerge(
           merge_operator_, ikey.user_key, &val, merge_context_.GetOperands(),
           &saved_value_, logger_, statistics_, env_, &pinned_value_, true);
       if (!s.ok()) {
@@ -548,15 +594,7 @@ bool DBIter::MergeValuesNewToOld() {
           iter_.value(), iter_.iter()->IsValuePinned() /* operand_pinned */);
       PERF_COUNTER_ADD(internal_merge_count, 1);
     } else if (kTypeBlobIndex == ikey.type) {
-      if (!allow_blob_) {
-        ROCKS_LOG_ERROR(logger_, "Encounter unexpected blob index.");
-        status_ = Status::NotSupported(
-            "Encounter unexpected blob index. Please open DB with "
-            "ROCKSDB_NAMESPACE::blob_db::BlobDB instead.");
-      } else {
-        status_ =
-            Status::NotSupported("Blob DB does not support merge operator.");
-      }
+      status_ = Status::NotSupported("BlobDB does not support merge operator.");
       valid_ = false;
       return false;
     } else {
@@ -577,10 +615,10 @@ bool DBIter::MergeValuesNewToOld() {
   // a deletion marker.
   // feed null as the existing value to the merge operator, such that
   // client can differentiate this scenario and do things accordingly.
-  s = MergeHelper::TimedFullMerge(merge_operator_, saved_key_.GetUserKey(),
-                                  nullptr, merge_context_.GetOperands(),
-                                  &saved_value_, logger_, statistics_, env_,
-                                  &pinned_value_, true);
+  Status s = MergeHelper::TimedFullMerge(
+      merge_operator_, saved_key_.GetUserKey(), nullptr,
+      merge_context_.GetOperands(), &saved_value_, logger_, statistics_, env_,
+      &pinned_value_, true);
   if (!s.ok()) {
     valid_ = false;
     status_ = s;
@@ -704,7 +742,9 @@ void DBIter::PrevInternal(const Slice* prefix) {
 
     assert(prefix == nullptr || prefix_extractor_ != nullptr);
     if (prefix != nullptr &&
-        prefix_extractor_->Transform(saved_key_.GetUserKey())
+        prefix_extractor_
+                ->Transform(StripTimestampFromUserKey(saved_key_.GetUserKey(),
+                                                      timestamp_size_))
                 .compare(*prefix) != 0) {
       assert(prefix_same_as_start_);
       // Current key does not have the same prefix as start
@@ -883,15 +923,8 @@ bool DBIter::FindValueForCurrentKey() {
             merge_context_.GetOperands(), &saved_value_, logger_, statistics_,
             env_, &pinned_value_, true);
       } else if (last_not_merge_type == kTypeBlobIndex) {
-        if (!allow_blob_) {
-          ROCKS_LOG_ERROR(logger_, "Encounter unexpected blob index.");
-          status_ = Status::NotSupported(
-              "Encounter unexpected blob index. Please open DB with "
-              "ROCKSDB_NAMESPACE::blob_db::BlobDB instead.");
-        } else {
-          status_ =
-              Status::NotSupported("Blob DB does not support merge operator.");
-        }
+        status_ =
+            Status::NotSupported("BlobDB does not support merge operator.");
         valid_ = false;
         return false;
       } else {
@@ -906,15 +939,10 @@ bool DBIter::FindValueForCurrentKey() {
       // do nothing - we've already has value in pinned_value_
       break;
     case kTypeBlobIndex:
-      if (!allow_blob_) {
-        ROCKS_LOG_ERROR(logger_, "Encounter unexpected blob index.");
-        status_ = Status::NotSupported(
-            "Encounter unexpected blob index. Please open DB with "
-            "ROCKSDB_NAMESPACE::blob_db::BlobDB instead.");
-        valid_ = false;
+      if (!SetBlobValueIfNeeded(saved_key_.GetUserKey(), pinned_value_)) {
         return false;
       }
-      is_blob_ = true;
+
       break;
     default:
       valid_ = false;
@@ -987,14 +1015,6 @@ bool DBIter::FindValueForCurrentKeyUsingSeek() {
     valid_ = false;
     return true;
   }
-  if (ikey.type == kTypeBlobIndex && !allow_blob_) {
-    ROCKS_LOG_ERROR(logger_, "Encounter unexpected blob index.");
-    status_ = Status::NotSupported(
-        "Encounter unexpected blob index. Please open DB with "
-        "ROCKSDB_NAMESPACE::blob_db::BlobDB instead.");
-    valid_ = false;
-    return false;
-  }
   if (!iter_.PrepareValue()) {
     valid_ = false;
     return false;
@@ -1002,7 +1022,12 @@ bool DBIter::FindValueForCurrentKeyUsingSeek() {
   if (ikey.type == kTypeValue || ikey.type == kTypeBlobIndex) {
     assert(iter_.iter()->IsValuePinned());
     pinned_value_ = iter_.value();
-    is_blob_ = (ikey.type == kTypeBlobIndex);
+    if (ikey.type == kTypeBlobIndex) {
+      if (!SetBlobValueIfNeeded(ikey.user_key, pinned_value_)) {
+        return false;
+      }
+    }
+
     valid_ = true;
     return true;
   }
@@ -1058,15 +1083,7 @@ bool DBIter::FindValueForCurrentKeyUsingSeek() {
           iter_.value(), iter_.iter()->IsValuePinned() /* operand_pinned */);
       PERF_COUNTER_ADD(internal_merge_count, 1);
     } else if (ikey.type == kTypeBlobIndex) {
-      if (!allow_blob_) {
-        ROCKS_LOG_ERROR(logger_, "Encounter unexpected blob index.");
-        status_ = Status::NotSupported(
-            "Encounter unexpected blob index. Please open DB with "
-            "ROCKSDB_NAMESPACE::blob_db::BlobDB instead.");
-      } else {
-        status_ =
-            Status::NotSupported("Blob DB does not support merge operator.");
-      }
+      status_ = Status::NotSupported("BlobDB does not support merge operator.");
       valid_ = false;
       return false;
     } else {
@@ -1237,7 +1254,19 @@ void DBIter::Seek(const Slice& target) {
 #ifndef ROCKSDB_LITE
   if (db_impl_ != nullptr && cfd_ != nullptr) {
     // TODO: What do we do if this returns an error?
-    db_impl_->TraceIteratorSeek(cfd_->GetID(), target).PermitUncheckedError();
+    Slice lower_bound, upper_bound;
+    if (iterate_lower_bound_ != nullptr) {
+      lower_bound = *iterate_lower_bound_;
+    } else {
+      lower_bound = Slice("");
+    }
+    if (iterate_upper_bound_ != nullptr) {
+      upper_bound = *iterate_upper_bound_;
+    } else {
+      upper_bound = Slice("");
+    }
+    db_impl_->TraceIteratorSeek(cfd_->GetID(), target, lower_bound, upper_bound)
+        .PermitUncheckedError();
   }
 #endif  // ROCKSDB_LITE
 
@@ -1299,7 +1328,20 @@ void DBIter::SeekForPrev(const Slice& target) {
 #ifndef ROCKSDB_LITE
   if (db_impl_ != nullptr && cfd_ != nullptr) {
     // TODO: What do we do if this returns an error?
-    db_impl_->TraceIteratorSeekForPrev(cfd_->GetID(), target)
+    Slice lower_bound, upper_bound;
+    if (iterate_lower_bound_ != nullptr) {
+      lower_bound = *iterate_lower_bound_;
+    } else {
+      lower_bound = Slice("");
+    }
+    if (iterate_upper_bound_ != nullptr) {
+      upper_bound = *iterate_upper_bound_;
+    } else {
+      upper_bound = Slice("");
+    }
+    db_impl_
+        ->TraceIteratorSeekForPrev(cfd_->GetID(), target, lower_bound,
+                                   upper_bound)
         .PermitUncheckedError();
   }
 #endif  // ROCKSDB_LITE
@@ -1399,7 +1441,8 @@ void DBIter::SeekToFirst() {
   }
   if (valid_ && prefix_same_as_start_) {
     assert(prefix_extractor_ != nullptr);
-    prefix_.SetUserKey(prefix_extractor_->Transform(saved_key_.GetUserKey()));
+    prefix_.SetUserKey(prefix_extractor_->Transform(
+        StripTimestampFromUserKey(saved_key_.GetUserKey(), timestamp_size_)));
   }
 }
 
@@ -1450,7 +1493,8 @@ void DBIter::SeekToLast() {
   }
   if (valid_ && prefix_same_as_start_) {
     assert(prefix_extractor_ != nullptr);
-    prefix_.SetUserKey(prefix_extractor_->Transform(saved_key_.GetUserKey()));
+    prefix_.SetUserKey(prefix_extractor_->Transform(
+        StripTimestampFromUserKey(saved_key_.GetUserKey(), timestamp_size_)));
   }
 }
 
@@ -1458,15 +1502,16 @@ 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,
+                        InternalIterator* internal_iter, const Version* version,
                         const SequenceNumber& sequence,
                         uint64_t max_sequential_skip_in_iterations,
                         ReadCallback* read_callback, DBImpl* db_impl,
-                        ColumnFamilyData* cfd, bool 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);
+                        ColumnFamilyData* cfd, bool expose_blob_index) {
+  DBIter* db_iter =
+      new DBIter(env, read_options, cf_options, mutable_cf_options,
+                 user_key_comparator, internal_iter, version, sequence, false,
+                 max_sequential_skip_in_iterations, read_callback, db_impl, cfd,
+                 expose_blob_index);
   return db_iter;
 }
 

db/db_iter.h

@@ -22,6 +22,8 @@
 
 namespace ROCKSDB_NAMESPACE {
 
+class Version;
+
 // This file declares the factory functions of DBIter, in its original form
 // or a wrapped form with class ArenaWrappedDBIter, which is defined here.
 // Class DBIter, which is declared and implemented inside db_iter.cc, is
@@ -114,10 +116,10 @@ class DBIter final : public Iterator {
   DBIter(Env* _env, const ReadOptions& read_options,
          const ImmutableCFOptions& cf_options,
          const MutableCFOptions& mutable_cf_options, const Comparator* cmp,
-         InternalIterator* iter, SequenceNumber s, bool arena_mode,
-         uint64_t max_sequential_skip_in_iterations,
+         InternalIterator* iter, const Version* version, SequenceNumber s,
+         bool arena_mode, uint64_t max_sequential_skip_in_iterations,
          ReadCallback* read_callback, DBImpl* db_impl, ColumnFamilyData* cfd,
-         bool allow_blob);
+         bool expose_blob_index);
 
   // No copying allowed
   DBIter(const DBIter&) = delete;
@@ -159,7 +161,10 @@ class DBIter final : public Iterator {
   }
   Slice value() const override {
     assert(valid_);
-    if (current_entry_is_merged_) {
+
+    if (!expose_blob_index_ && is_blob_) {
+      return blob_value_;
+    } else if (current_entry_is_merged_) {
       // If pinned_value_ is set then the result of merge operator is one of
       // the merge operands and we should return it.
       return pinned_value_.data() ? pinned_value_ : saved_value_;
@@ -185,7 +190,7 @@ class DBIter final : public Iterator {
     return ExtractTimestampFromUserKey(ukey_and_ts, timestamp_size_);
   }
   bool IsBlob() const {
-    assert(valid_ && (allow_blob_ || !is_blob_));
+    assert(valid_);
     return is_blob_;
   }
 
@@ -287,12 +292,17 @@ class DBIter final : public Iterator {
                : user_comparator_.CompareWithoutTimestamp(a, b);
   }
 
+  // Retrieves the blob value for the specified user key using the given blob
+  // index when using the integrated BlobDB implementation.
+  bool SetBlobValueIfNeeded(const Slice& user_key, const Slice& blob_index);
+
   const SliceTransform* prefix_extractor_;
   Env* const env_;
   Logger* logger_;
   UserComparatorWrapper user_comparator_;
   const MergeOperator* const merge_operator_;
   IteratorWrapper iter_;
+  const Version* version_;
   ReadCallback* read_callback_;
   // Max visible sequence number. It is normally the snapshot seq unless we have
   // uncommitted data in db as in WriteUnCommitted.
@@ -306,6 +316,7 @@ class DBIter final : public Iterator {
   std::string saved_value_;
   Slice pinned_value_;
   // for prefix seek mode to support prev()
+  PinnableSlice blob_value_;
   Statistics* statistics_;
   uint64_t max_skip_;
   uint64_t max_skippable_internal_keys_;
@@ -335,7 +346,11 @@ class DBIter final : public Iterator {
   // Expect the inner iterator to maintain a total order.
   // prefix_extractor_ must be non-NULL if the value is false.
   const bool expect_total_order_inner_iter_;
-  bool allow_blob_;
+  ReadTier read_tier_;
+  bool verify_checksums_;
+  // Whether the iterator is allowed to expose blob references. Set to true when
+  // the stacked BlobDB implementation is used, false otherwise.
+  bool expose_blob_index_;
   bool is_blob_;
   bool arena_mode_;
   // List of operands for merge operator.
@@ -367,8 +382,9 @@ extern Iterator* NewDBIterator(
     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);
+    const Version* version, const SequenceNumber& sequence,
+    uint64_t max_sequential_skip_in_iterations, ReadCallback* read_callback,
+    DBImpl* db_impl = nullptr, ColumnFamilyData* cfd = nullptr,
+    bool expose_blob_index = false);
 
 }  // namespace ROCKSDB_NAMESPACE

db/db_iter_stress_test.cc

@@ -513,7 +513,7 @@ TEST_F(DBIteratorStressTest, StressTest) {
                   db_iter.reset(NewDBIterator(
                       env_, ropt, ImmutableCFOptions(options),
                       MutableCFOptions(options), BytewiseComparator(),
-                      internal_iter, sequence,
+                      internal_iter, nullptr /* version */, sequence,
                       options.max_sequential_skip_in_iterations,
                       nullptr /*read_callback*/));
                 }

db/db_iter_test.cc

@@ -253,8 +253,9 @@ TEST_F(DBIteratorTest, DBIteratorPrevNext) {
     ReadOptions ro;
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 10, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 10 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
 
     db_iter->SeekToLast();
     ASSERT_TRUE(db_iter->Valid());
@@ -286,8 +287,9 @@ TEST_F(DBIteratorTest, DBIteratorPrevNext) {
     ReadOptions ro;
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 10, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 10 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
 
     db_iter->SeekToLast();
     ASSERT_TRUE(db_iter->Valid());
@@ -313,8 +315,9 @@ TEST_F(DBIteratorTest, DBIteratorPrevNext) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 10, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 10 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
 
     db_iter->SeekToLast();
     ASSERT_TRUE(db_iter->Valid());
@@ -346,8 +349,9 @@ TEST_F(DBIteratorTest, DBIteratorPrevNext) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 10, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 10 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
 
     db_iter->SeekToLast();
     ASSERT_TRUE(db_iter->Valid());
@@ -382,11 +386,13 @@ TEST_F(DBIteratorTest, DBIteratorPrevNext) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 10, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 10 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
 
     db_iter->SeekToLast();
     ASSERT_TRUE(!db_iter->Valid());
+    ASSERT_OK(db_iter->status());
   }
   // Test case to check SeekToLast with iterate_upper_bound set
   // (same key put may times - SeekToLast should start with the
@@ -412,8 +418,9 @@ TEST_F(DBIteratorTest, DBIteratorPrevNext) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 7, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 7 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
 
     SetPerfLevel(kEnableCount);
     ASSERT_TRUE(GetPerfLevel() == kEnableCount);
@@ -450,8 +457,9 @@ TEST_F(DBIteratorTest, DBIteratorPrevNext) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 4, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 4 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
 
     db_iter->SeekToLast();
     ASSERT_TRUE(db_iter->Valid());
@@ -476,11 +484,13 @@ TEST_F(DBIteratorTest, DBIteratorPrevNext) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 10, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 10 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
 
     db_iter->SeekToLast();
     ASSERT_TRUE(!db_iter->Valid());
+    ASSERT_OK(db_iter->status());
   }
   // Test to check the SeekToLast() with the iterate_upper_bound set
   // (Deletion cases)
@@ -499,8 +509,9 @@ TEST_F(DBIteratorTest, DBIteratorPrevNext) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 10, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 10 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
 
     db_iter->SeekToLast();
     ASSERT_TRUE(db_iter->Valid());
@@ -535,8 +546,9 @@ TEST_F(DBIteratorTest, DBIteratorPrevNext) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 7, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 7 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
 
     SetPerfLevel(kEnableCount);
     ASSERT_TRUE(GetPerfLevel() == kEnableCount);
@@ -565,8 +577,9 @@ TEST_F(DBIteratorTest, DBIteratorPrevNext) {
     ReadOptions ro;
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 10, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 10 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
 
     db_iter->SeekToFirst();
     ASSERT_TRUE(db_iter->Valid());
@@ -585,6 +598,7 @@ TEST_F(DBIteratorTest, DBIteratorPrevNext) {
 
     db_iter->Prev();
     ASSERT_TRUE(!db_iter->Valid());
+    ASSERT_OK(db_iter->status());
   }
 
   {
@@ -608,8 +622,9 @@ TEST_F(DBIteratorTest, DBIteratorPrevNext) {
     ReadOptions ro;
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 2, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 2 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
     db_iter->SeekToLast();
     ASSERT_TRUE(db_iter->Valid());
     ASSERT_EQ(db_iter->key().ToString(), "b");
@@ -640,8 +655,9 @@ TEST_F(DBIteratorTest, DBIteratorPrevNext) {
     ReadOptions ro;
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 10, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 10 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
     db_iter->SeekToLast();
     ASSERT_TRUE(db_iter->Valid());
     ASSERT_EQ(db_iter->key().ToString(), "c");
@@ -671,10 +687,12 @@ TEST_F(DBIteratorTest, DBIteratorEmpty) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 0, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 0 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
     db_iter->SeekToLast();
     ASSERT_TRUE(!db_iter->Valid());
+    ASSERT_OK(db_iter->status());
   }
 
   {
@@ -683,10 +701,12 @@ TEST_F(DBIteratorTest, DBIteratorEmpty) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 0, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 0 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
     db_iter->SeekToFirst();
     ASSERT_TRUE(!db_iter->Valid());
+    ASSERT_OK(db_iter->status());
   }
 }
 
@@ -706,8 +726,9 @@ TEST_F(DBIteratorTest, DBIteratorUseSkipCountSkips) {
 
   std::unique_ptr<Iterator> db_iter(NewDBIterator(
       env_, ro, ImmutableCFOptions(options), MutableCFOptions(options),
-      BytewiseComparator(), internal_iter, 2,
-      options.max_sequential_skip_in_iterations, nullptr /*read_callback*/));
+      BytewiseComparator(), internal_iter, nullptr /* version */,
+      2 /* sequence */, options.max_sequential_skip_in_iterations,
+      nullptr /* read_callback */));
   db_iter->SeekToLast();
   ASSERT_TRUE(db_iter->Valid());
   ASSERT_EQ(db_iter->key().ToString(), "c");
@@ -728,6 +749,7 @@ TEST_F(DBIteratorTest, DBIteratorUseSkipCountSkips) {
 
   db_iter->Prev();
   ASSERT_TRUE(!db_iter->Valid());
+  ASSERT_OK(db_iter->status());
   ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 3u);
 }
 
@@ -751,8 +773,9 @@ TEST_F(DBIteratorTest, DBIteratorUseSkip) {
       options.statistics = ROCKSDB_NAMESPACE::CreateDBStatistics();
       std::unique_ptr<Iterator> db_iter(NewDBIterator(
           env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-          internal_iter, i + 2, options.max_sequential_skip_in_iterations,
-          nullptr /*read_callback*/));
+          internal_iter, nullptr /* version */, i + 2 /* sequence */,
+          options.max_sequential_skip_in_iterations,
+          nullptr /* read_callback */));
       db_iter->SeekToLast();
       ASSERT_TRUE(db_iter->Valid());
 
@@ -771,6 +794,7 @@ TEST_F(DBIteratorTest, DBIteratorUseSkip) {
       db_iter->Prev();
 
       ASSERT_TRUE(!db_iter->Valid());
+      ASSERT_OK(db_iter->status());
     }
   }
 
@@ -787,8 +811,9 @@ TEST_F(DBIteratorTest, DBIteratorUseSkip) {
 
       std::unique_ptr<Iterator> db_iter(NewDBIterator(
           env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-          internal_iter, i + 2, options.max_sequential_skip_in_iterations,
-          nullptr /*read_callback*/));
+          internal_iter, nullptr /* version */, i + 2 /* sequence */,
+          options.max_sequential_skip_in_iterations,
+          nullptr /* read_callback */));
       db_iter->SeekToLast();
       ASSERT_TRUE(db_iter->Valid());
 
@@ -802,6 +827,7 @@ TEST_F(DBIteratorTest, DBIteratorUseSkip) {
       db_iter->Prev();
 
       ASSERT_TRUE(!db_iter->Valid());
+      ASSERT_OK(db_iter->status());
     }
 
     {
@@ -816,8 +842,9 @@ TEST_F(DBIteratorTest, DBIteratorUseSkip) {
 
       std::unique_ptr<Iterator> db_iter(NewDBIterator(
           env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-          internal_iter, 202, options.max_sequential_skip_in_iterations,
-          nullptr /*read_callback*/));
+          internal_iter, nullptr /* version */, 202 /* sequence */,
+          options.max_sequential_skip_in_iterations,
+          nullptr /* read_callback */));
       db_iter->SeekToLast();
       ASSERT_TRUE(db_iter->Valid());
 
@@ -836,6 +863,7 @@ TEST_F(DBIteratorTest, DBIteratorUseSkip) {
       db_iter->Prev();
 
       ASSERT_TRUE(!db_iter->Valid());
+      ASSERT_OK(db_iter->status());
     }
   }
 
@@ -849,13 +877,16 @@ TEST_F(DBIteratorTest, DBIteratorUseSkip) {
       internal_iter->Finish();
       std::unique_ptr<Iterator> db_iter(NewDBIterator(
           env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-          internal_iter, i, options.max_sequential_skip_in_iterations,
-          nullptr /*read_callback*/));
+          internal_iter, nullptr /* version */, i /* sequence */,
+          options.max_sequential_skip_in_iterations,
+          nullptr /* read_callback */));
       db_iter->SeekToLast();
       ASSERT_TRUE(!db_iter->Valid());
+      ASSERT_OK(db_iter->status());
 
       db_iter->SeekToFirst();
       ASSERT_TRUE(!db_iter->Valid());
+      ASSERT_OK(db_iter->status());
     }
 
     TestIterator* internal_iter = new TestIterator(BytewiseComparator());
@@ -866,8 +897,9 @@ TEST_F(DBIteratorTest, DBIteratorUseSkip) {
     internal_iter->Finish();
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 200, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 200 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
     db_iter->SeekToLast();
     ASSERT_TRUE(db_iter->Valid());
     ASSERT_EQ(db_iter->key().ToString(), "c");
@@ -875,6 +907,7 @@ TEST_F(DBIteratorTest, DBIteratorUseSkip) {
 
     db_iter->Prev();
     ASSERT_TRUE(!db_iter->Valid());
+    ASSERT_OK(db_iter->status());
 
     db_iter->SeekToFirst();
     ASSERT_TRUE(db_iter->Valid());
@@ -883,6 +916,7 @@ TEST_F(DBIteratorTest, DBIteratorUseSkip) {
 
     db_iter->Next();
     ASSERT_TRUE(!db_iter->Valid());
+    ASSERT_OK(db_iter->status());
   }
 
   {
@@ -901,8 +935,9 @@ TEST_F(DBIteratorTest, DBIteratorUseSkip) {
 
       std::unique_ptr<Iterator> db_iter(NewDBIterator(
           env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-          internal_iter, i + 2, options.max_sequential_skip_in_iterations,
-          nullptr /*read_callback*/));
+          internal_iter, nullptr /* version */, i + 2 /* sequence */,
+          options.max_sequential_skip_in_iterations,
+          nullptr /* read_callback */));
       db_iter->SeekToLast();
       ASSERT_TRUE(db_iter->Valid());
 
@@ -921,6 +956,7 @@ TEST_F(DBIteratorTest, DBIteratorUseSkip) {
       db_iter->Prev();
 
       ASSERT_TRUE(!db_iter->Valid());
+      ASSERT_OK(db_iter->status());
     }
   }
 
@@ -936,8 +972,9 @@ TEST_F(DBIteratorTest, DBIteratorUseSkip) {
 
       std::unique_ptr<Iterator> db_iter(NewDBIterator(
           env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-          internal_iter, i + 2, options.max_sequential_skip_in_iterations,
-          nullptr /*read_callback*/));
+          internal_iter, nullptr /* version */, i + 2 /* sequence */,
+          options.max_sequential_skip_in_iterations,
+          nullptr /* read_callback */));
       db_iter->SeekToLast();
       ASSERT_TRUE(db_iter->Valid());
 
@@ -960,6 +997,7 @@ TEST_F(DBIteratorTest, DBIteratorUseSkip) {
 
       db_iter->Prev();
       ASSERT_TRUE(!db_iter->Valid());
+      ASSERT_OK(db_iter->status());
     }
   }
 }
@@ -986,8 +1024,9 @@ TEST_F(DBIteratorTest, DBIteratorSkipInternalKeys) {
     ro.max_skippable_internal_keys = 0;
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 10, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 10 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
 
     db_iter->SeekToFirst();
     ASSERT_TRUE(db_iter->Valid());
@@ -1015,7 +1054,7 @@ TEST_F(DBIteratorTest, DBIteratorSkipInternalKeys) {
 
     db_iter->Prev();
     ASSERT_TRUE(!db_iter->Valid());
-    ASSERT_TRUE(db_iter->status().ok());
+    ASSERT_OK(db_iter->status());
   }
 
   // Test to make sure that the request will *not* fail as incomplete if
@@ -1033,8 +1072,9 @@ TEST_F(DBIteratorTest, DBIteratorSkipInternalKeys) {
     ro.max_skippable_internal_keys = 2;
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 10, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 10 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
 
     db_iter->SeekToFirst();
     ASSERT_TRUE(db_iter->Valid());
@@ -1078,8 +1118,9 @@ TEST_F(DBIteratorTest, DBIteratorSkipInternalKeys) {
     ro.max_skippable_internal_keys = 2;
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 10, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 10 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
 
     db_iter->SeekToFirst();
     ASSERT_TRUE(db_iter->Valid());
@@ -1117,8 +1158,9 @@ TEST_F(DBIteratorTest, DBIteratorSkipInternalKeys) {
     ro.max_skippable_internal_keys = 2;
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 10, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 10 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
 
     db_iter->SeekToFirst();
     ASSERT_TRUE(db_iter->Valid());
@@ -1153,8 +1195,9 @@ TEST_F(DBIteratorTest, DBIteratorSkipInternalKeys) {
     ro.max_skippable_internal_keys = 2;
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 10, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 10 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
 
     db_iter->SeekToLast();
     ASSERT_TRUE(db_iter->Valid());
@@ -1184,8 +1227,9 @@ TEST_F(DBIteratorTest, DBIteratorSkipInternalKeys) {
     ro.max_skippable_internal_keys = 2;
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 10, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 10 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
 
     db_iter->SeekToFirst();
     ASSERT_TRUE(db_iter->Valid());
@@ -1222,8 +1266,9 @@ TEST_F(DBIteratorTest, DBIteratorSkipInternalKeys) {
     ro.max_skippable_internal_keys = 2;
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 10, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 10 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
 
     db_iter->SeekToFirst();
     ASSERT_TRUE(db_iter->Valid());
@@ -1260,8 +1305,9 @@ TEST_F(DBIteratorTest, DBIteratorSkipInternalKeys) {
       ro.max_skippable_internal_keys = i;
       std::unique_ptr<Iterator> db_iter(NewDBIterator(
           env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-          internal_iter, 2 * i + 1, options.max_sequential_skip_in_iterations,
-          nullptr /*read_callback*/));
+          internal_iter, nullptr /* version */, 2 * i + 1 /* sequence */,
+          options.max_sequential_skip_in_iterations,
+          nullptr /* read_callback */));
 
       db_iter->SeekToFirst();
       ASSERT_TRUE(db_iter->Valid());
@@ -1314,8 +1360,9 @@ TEST_F(DBIteratorTest, DBIteratorSkipInternalKeys) {
       ro.max_skippable_internal_keys = i;
       std::unique_ptr<Iterator> db_iter(NewDBIterator(
           env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-          internal_iter, 2 * i + 1, options.max_sequential_skip_in_iterations,
-          nullptr /*read_callback*/));
+          internal_iter, nullptr /* version */, 2 * i + 1 /* sequence */,
+          options.max_sequential_skip_in_iterations,
+          nullptr /* read_callback */));
 
       db_iter->SeekToFirst();
       ASSERT_TRUE(db_iter->Valid());
@@ -1353,8 +1400,9 @@ TEST_F(DBIteratorTest, DBIterator1) {
 
   std::unique_ptr<Iterator> db_iter(NewDBIterator(
       env_, ro, ImmutableCFOptions(options), MutableCFOptions(options),
-      BytewiseComparator(), internal_iter, 1,
-      options.max_sequential_skip_in_iterations, nullptr /*read_callback*/));
+      BytewiseComparator(), internal_iter, nullptr /* version */,
+      1 /* sequence */, options.max_sequential_skip_in_iterations,
+      nullptr /* read_callback */));
   db_iter->SeekToFirst();
   ASSERT_TRUE(db_iter->Valid());
   ASSERT_EQ(db_iter->key().ToString(), "a");
@@ -1381,8 +1429,9 @@ TEST_F(DBIteratorTest, DBIterator2) {
 
   std::unique_ptr<Iterator> db_iter(NewDBIterator(
       env_, ro, ImmutableCFOptions(options), MutableCFOptions(options),
-      BytewiseComparator(), internal_iter, 0,
-      options.max_sequential_skip_in_iterations, nullptr /*read_callback*/));
+      BytewiseComparator(), internal_iter, nullptr /* version */,
+      0 /* sequence */, options.max_sequential_skip_in_iterations,
+      nullptr /* read_callback */));
   db_iter->SeekToFirst();
   ASSERT_TRUE(db_iter->Valid());
   ASSERT_EQ(db_iter->key().ToString(), "a");
@@ -1406,8 +1455,9 @@ TEST_F(DBIteratorTest, DBIterator3) {
 
   std::unique_ptr<Iterator> db_iter(NewDBIterator(
       env_, ro, ImmutableCFOptions(options), MutableCFOptions(options),
-      BytewiseComparator(), internal_iter, 2,
-      options.max_sequential_skip_in_iterations, nullptr /*read_callback*/));
+      BytewiseComparator(), internal_iter, nullptr /* version */,
+      2 /* sequence */, options.max_sequential_skip_in_iterations,
+      nullptr /* read_callback */));
   db_iter->SeekToFirst();
   ASSERT_TRUE(db_iter->Valid());
   ASSERT_EQ(db_iter->key().ToString(), "a");
@@ -1431,8 +1481,9 @@ TEST_F(DBIteratorTest, DBIterator4) {
 
   std::unique_ptr<Iterator> db_iter(NewDBIterator(
       env_, ro, ImmutableCFOptions(options), MutableCFOptions(options),
-      BytewiseComparator(), internal_iter, 4,
-      options.max_sequential_skip_in_iterations, nullptr /*read_callback*/));
+      BytewiseComparator(), internal_iter, nullptr /* version */,
+      4 /* sequence */, options.max_sequential_skip_in_iterations,
+      nullptr /* read_callback */));
   db_iter->SeekToFirst();
   ASSERT_TRUE(db_iter->Valid());
   ASSERT_EQ(db_iter->key().ToString(), "a");
@@ -1465,8 +1516,9 @@ TEST_F(DBIteratorTest, DBIterator5) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 0, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 0 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
     db_iter->SeekToLast();
     ASSERT_TRUE(db_iter->Valid());
     ASSERT_EQ(db_iter->key().ToString(), "a");
@@ -1488,8 +1540,9 @@ TEST_F(DBIteratorTest, DBIterator5) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 1, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 1 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
     db_iter->SeekToLast();
     ASSERT_TRUE(db_iter->Valid());
     ASSERT_EQ(db_iter->key().ToString(), "a");
@@ -1511,8 +1564,9 @@ TEST_F(DBIteratorTest, DBIterator5) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 2, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 2 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
     db_iter->SeekToLast();
     ASSERT_TRUE(db_iter->Valid());
     ASSERT_EQ(db_iter->key().ToString(), "a");
@@ -1534,8 +1588,9 @@ TEST_F(DBIteratorTest, DBIterator5) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 3, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 3 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
     db_iter->SeekToLast();
     ASSERT_TRUE(db_iter->Valid());
     ASSERT_EQ(db_iter->key().ToString(), "a");
@@ -1557,8 +1612,9 @@ TEST_F(DBIteratorTest, DBIterator5) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 4, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 4 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
     db_iter->SeekToLast();
     ASSERT_TRUE(db_iter->Valid());
     ASSERT_EQ(db_iter->key().ToString(), "a");
@@ -1580,8 +1636,9 @@ TEST_F(DBIteratorTest, DBIterator5) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 5, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 5 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
     db_iter->SeekToLast();
     ASSERT_TRUE(db_iter->Valid());
     ASSERT_EQ(db_iter->key().ToString(), "a");
@@ -1603,8 +1660,9 @@ TEST_F(DBIteratorTest, DBIterator5) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 6, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 6 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
     db_iter->SeekToLast();
     ASSERT_TRUE(db_iter->Valid());
     ASSERT_EQ(db_iter->key().ToString(), "a");
@@ -1624,8 +1682,9 @@ TEST_F(DBIteratorTest, DBIterator5) {
     internal_iter->Finish();
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 10, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 10 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
     db_iter->Seek("b");
     ASSERT_TRUE(db_iter->Valid());
     ASSERT_EQ(db_iter->key().ToString(), "b");
@@ -1655,8 +1714,9 @@ TEST_F(DBIteratorTest, DBIterator6) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 0, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 0 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
     db_iter->SeekToLast();
     ASSERT_TRUE(db_iter->Valid());
     ASSERT_EQ(db_iter->key().ToString(), "a");
@@ -1678,8 +1738,9 @@ TEST_F(DBIteratorTest, DBIterator6) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 1, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 1 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
     db_iter->SeekToLast();
     ASSERT_TRUE(db_iter->Valid());
     ASSERT_EQ(db_iter->key().ToString(), "a");
@@ -1701,8 +1762,9 @@ TEST_F(DBIteratorTest, DBIterator6) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 2, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 2 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
     db_iter->SeekToLast();
     ASSERT_TRUE(db_iter->Valid());
     ASSERT_EQ(db_iter->key().ToString(), "a");
@@ -1724,8 +1786,9 @@ TEST_F(DBIteratorTest, DBIterator6) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 3, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 3 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
     db_iter->SeekToLast();
     ASSERT_TRUE(!db_iter->Valid());
   }
@@ -1743,8 +1806,9 @@ TEST_F(DBIteratorTest, DBIterator6) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 4, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 4 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
     db_iter->SeekToLast();
     ASSERT_TRUE(db_iter->Valid());
     ASSERT_EQ(db_iter->key().ToString(), "a");
@@ -1766,8 +1830,9 @@ TEST_F(DBIteratorTest, DBIterator6) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 5, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 5 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
     db_iter->SeekToLast();
     ASSERT_TRUE(db_iter->Valid());
     ASSERT_EQ(db_iter->key().ToString(), "a");
@@ -1789,8 +1854,9 @@ TEST_F(DBIteratorTest, DBIterator6) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 6, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 6 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
     db_iter->SeekToLast();
     ASSERT_TRUE(db_iter->Valid());
     ASSERT_EQ(db_iter->key().ToString(), "a");
@@ -1832,8 +1898,9 @@ TEST_F(DBIteratorTest, DBIterator7) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 0, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 0 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
     db_iter->SeekToLast();
     ASSERT_TRUE(db_iter->Valid());
     ASSERT_EQ(db_iter->key().ToString(), "a");
@@ -1867,8 +1934,9 @@ TEST_F(DBIteratorTest, DBIterator7) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 2, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 2 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
     db_iter->SeekToLast();
     ASSERT_TRUE(db_iter->Valid());
 
@@ -1908,8 +1976,9 @@ TEST_F(DBIteratorTest, DBIterator7) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 4, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 4 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
     db_iter->SeekToLast();
     ASSERT_TRUE(db_iter->Valid());
 
@@ -1949,8 +2018,9 @@ TEST_F(DBIteratorTest, DBIterator7) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 5, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 5 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
     db_iter->SeekToLast();
     ASSERT_TRUE(db_iter->Valid());
 
@@ -1995,8 +2065,9 @@ TEST_F(DBIteratorTest, DBIterator7) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 6, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 6 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
     db_iter->SeekToLast();
     ASSERT_TRUE(db_iter->Valid());
 
@@ -2042,8 +2113,9 @@ TEST_F(DBIteratorTest, DBIterator7) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 7, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 7 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
     db_iter->SeekToLast();
     ASSERT_TRUE(db_iter->Valid());
 
@@ -2083,8 +2155,9 @@ TEST_F(DBIteratorTest, DBIterator7) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 9, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 9 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
     db_iter->SeekToLast();
     ASSERT_TRUE(db_iter->Valid());
 
@@ -2130,8 +2203,9 @@ TEST_F(DBIteratorTest, DBIterator7) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 13, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 13 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
     db_iter->SeekToLast();
     ASSERT_TRUE(db_iter->Valid());
 
@@ -2178,8 +2252,9 @@ TEST_F(DBIteratorTest, DBIterator7) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, cf_options, mutable_cf_options, BytewiseComparator(),
-        internal_iter, 14, options.max_sequential_skip_in_iterations,
-        nullptr /*read_callback*/));
+        internal_iter, nullptr /* version */, 14 /* sequence */,
+        options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
     db_iter->SeekToLast();
     ASSERT_TRUE(db_iter->Valid());
 
@@ -2209,8 +2284,9 @@ TEST_F(DBIteratorTest, DBIterator8) {
 
   std::unique_ptr<Iterator> db_iter(NewDBIterator(
       env_, ro, ImmutableCFOptions(options), MutableCFOptions(options),
-      BytewiseComparator(), internal_iter, 10,
-      options.max_sequential_skip_in_iterations, nullptr /*read_callback*/));
+      BytewiseComparator(), internal_iter, nullptr /* version */,
+      10 /* sequence */, options.max_sequential_skip_in_iterations,
+      nullptr /* read_callback */));
   db_iter->SeekToLast();
   ASSERT_TRUE(db_iter->Valid());
   ASSERT_EQ(db_iter->key().ToString(), "b");
@@ -2240,8 +2316,9 @@ TEST_F(DBIteratorTest, DBIterator9) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, ImmutableCFOptions(options), MutableCFOptions(options),
-        BytewiseComparator(), internal_iter, 10,
-        options.max_sequential_skip_in_iterations, nullptr /*read_callback*/));
+        BytewiseComparator(), internal_iter, nullptr /* version */,
+        10 /* sequence */, options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
 
     db_iter->SeekToLast();
     ASSERT_TRUE(db_iter->Valid());
@@ -2307,8 +2384,9 @@ TEST_F(DBIteratorTest, DBIterator10) {
 
   std::unique_ptr<Iterator> db_iter(NewDBIterator(
       env_, ro, ImmutableCFOptions(options), MutableCFOptions(options),
-      BytewiseComparator(), internal_iter, 10,
-      options.max_sequential_skip_in_iterations, nullptr /*read_callback*/));
+      BytewiseComparator(), internal_iter, nullptr /* version */,
+      10 /* sequence */, options.max_sequential_skip_in_iterations,
+      nullptr /* read_callback */));
 
   db_iter->Seek("c");
   ASSERT_TRUE(db_iter->Valid());
@@ -2347,8 +2425,8 @@ TEST_F(DBIteratorTest, SeekToLastOccurrenceSeq0) {
 
   std::unique_ptr<Iterator> db_iter(NewDBIterator(
       env_, ro, ImmutableCFOptions(options), MutableCFOptions(options),
-      BytewiseComparator(), internal_iter, 10, 0 /* force seek */,
-      nullptr /*read_callback*/));
+      BytewiseComparator(), internal_iter, nullptr /* version */,
+      10 /* sequence */, 0 /* force seek */, nullptr /* read_callback */));
   db_iter->SeekToFirst();
   ASSERT_TRUE(db_iter->Valid());
   ASSERT_EQ(db_iter->key().ToString(), "a");
@@ -2376,8 +2454,9 @@ TEST_F(DBIteratorTest, DBIterator11) {
 
   std::unique_ptr<Iterator> db_iter(NewDBIterator(
       env_, ro, ImmutableCFOptions(options), MutableCFOptions(options),
-      BytewiseComparator(), internal_iter, 1,
-      options.max_sequential_skip_in_iterations, nullptr /*read_callback*/));
+      BytewiseComparator(), internal_iter, nullptr /* version */,
+      1 /* sequence */, options.max_sequential_skip_in_iterations,
+      nullptr /* read_callback */));
   db_iter->SeekToFirst();
   ASSERT_TRUE(db_iter->Valid());
   ASSERT_EQ(db_iter->key().ToString(), "a");
@@ -2403,7 +2482,8 @@ TEST_F(DBIteratorTest, DBIterator12) {
 
   std::unique_ptr<Iterator> db_iter(NewDBIterator(
       env_, ro, ImmutableCFOptions(options), MutableCFOptions(options),
-      BytewiseComparator(), internal_iter, 10, 0, nullptr /*read_callback*/));
+      BytewiseComparator(), internal_iter, nullptr /* version */,
+      10 /* sequence */, 0 /* force seek */, nullptr /* read_callback */));
   db_iter->SeekToLast();
   ASSERT_TRUE(db_iter->Valid());
   ASSERT_EQ(db_iter->key().ToString(), "c");
@@ -2440,7 +2520,9 @@ TEST_F(DBIteratorTest, DBIterator13) {
 
   std::unique_ptr<Iterator> db_iter(NewDBIterator(
       env_, ro, ImmutableCFOptions(options), MutableCFOptions(options),
-      BytewiseComparator(), internal_iter, 2, 3, nullptr /*read_callback*/));
+      BytewiseComparator(), internal_iter, nullptr /* version */,
+      2 /* sequence */, 3 /* max_sequential_skip_in_iterations */,
+      nullptr /* read_callback */));
   db_iter->Seek("b");
   ASSERT_TRUE(db_iter->Valid());
   ASSERT_EQ(db_iter->key().ToString(), key);
@@ -2468,7 +2550,9 @@ TEST_F(DBIteratorTest, DBIterator14) {
 
   std::unique_ptr<Iterator> db_iter(NewDBIterator(
       env_, ro, ImmutableCFOptions(options), MutableCFOptions(options),
-      BytewiseComparator(), internal_iter, 4, 1, nullptr /*read_callback*/));
+      BytewiseComparator(), internal_iter, nullptr /* version */,
+      4 /* sequence */, 1 /* max_sequential_skip_in_iterations */,
+      nullptr /* read_callback */));
   db_iter->Seek("b");
   ASSERT_TRUE(db_iter->Valid());
   ASSERT_EQ(db_iter->key().ToString(), "b");
@@ -2495,8 +2579,9 @@ TEST_F(DBIteratorTest, DBIteratorTestDifferentialSnapshots) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, ImmutableCFOptions(options), MutableCFOptions(options),
-        BytewiseComparator(), internal_iter, 13,
-        options.max_sequential_skip_in_iterations, nullptr));
+        BytewiseComparator(), internal_iter, nullptr /* version */,
+        13 /* sequence */, options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
     // Expecting InternalKeys in [5,8] range with correct type
     int seqnums[4] = {5,8,11,13};
     std::string user_keys[4] = {"1","2","3","4"};
@@ -2530,8 +2615,9 @@ TEST_F(DBIteratorTest, DBIteratorTestDifferentialSnapshots) {
 
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, ImmutableCFOptions(options), MutableCFOptions(options),
-        BytewiseComparator(), internal_iter, 13,
-        options.max_sequential_skip_in_iterations, nullptr));
+        BytewiseComparator(), internal_iter, nullptr /* version */,
+        13 /* sequence */, options.max_sequential_skip_in_iterations,
+        nullptr /* read_callback */));
     // Expecting InternalKeys in [5,8] range with correct type
     int seqnums[4] = {5,8,11,13};
     EntryType key_types[4] = {EntryType::kEntryDelete,EntryType::kEntryDelete,
@@ -2580,9 +2666,9 @@ class DBIterWithMergeIterTest : public testing::Test {
 
     db_iter_.reset(NewDBIterator(
         env_, ro_, ImmutableCFOptions(options_), MutableCFOptions(options_),
-        BytewiseComparator(), merge_iter,
+        BytewiseComparator(), merge_iter, nullptr /* version */,
         8 /* read data earlier than seqId 8 */,
-        3 /* max iterators before reseek */, nullptr /*read_callback*/));
+        3 /* max iterators before reseek */, nullptr /* read_callback */));
   }
 
   Env* env_;
@@ -3020,8 +3106,9 @@ TEST_F(DBIteratorTest, SeekPrefixTombstones) {
   ro.prefix_same_as_start = true;
   std::unique_ptr<Iterator> db_iter(NewDBIterator(
       env_, ro, ImmutableCFOptions(options), MutableCFOptions(options),
-      BytewiseComparator(), internal_iter, 10,
-      options.max_sequential_skip_in_iterations, nullptr /*read_callback*/));
+      BytewiseComparator(), internal_iter, nullptr /* version */,
+      10 /* sequence */, options.max_sequential_skip_in_iterations,
+      nullptr /* read_callback */));
 
   int skipped_keys = 0;
 
@@ -3056,14 +3143,15 @@ TEST_F(DBIteratorTest, SeekToFirstLowerBound) {
     Options options;
     std::unique_ptr<Iterator> db_iter(NewDBIterator(
         env_, ro, ImmutableCFOptions(options), MutableCFOptions(options),
-        BytewiseComparator(), internal_iter, 10 /* sequence */,
-        options.max_sequential_skip_in_iterations,
+        BytewiseComparator(), internal_iter, nullptr /* version */,
+        10 /* sequence */, options.max_sequential_skip_in_iterations,
         nullptr /* read_callback */));
 
     db_iter->SeekToFirst();
     if (i == kNumKeys + 1) {
       // lower bound was beyond the last key
       ASSERT_FALSE(db_iter->Valid());
+      ASSERT_OK(db_iter->status());
     } else {
       ASSERT_TRUE(db_iter->Valid());
       int expected;
@@ -3095,8 +3183,9 @@ TEST_F(DBIteratorTest, PrevLowerBound) {
   Options options;
   std::unique_ptr<Iterator> db_iter(NewDBIterator(
       env_, ro, ImmutableCFOptions(options), MutableCFOptions(options),
-      BytewiseComparator(), internal_iter, 10 /* sequence */,
-      options.max_sequential_skip_in_iterations, nullptr /* read_callback */));
+      BytewiseComparator(), internal_iter, nullptr /* version */,
+      10 /* sequence */, options.max_sequential_skip_in_iterations,
+      nullptr /* read_callback */));
 
   db_iter->SeekToLast();
   for (int i = kNumKeys; i >= kLowerBound; --i) {
@@ -3123,8 +3212,9 @@ TEST_F(DBIteratorTest, SeekLessLowerBound) {
   Options options;
   std::unique_ptr<Iterator> db_iter(NewDBIterator(
       env_, ro, ImmutableCFOptions(options), MutableCFOptions(options),
-      BytewiseComparator(), internal_iter, 10 /* sequence */,
-      options.max_sequential_skip_in_iterations, nullptr /* read_callback */));
+      BytewiseComparator(), internal_iter, nullptr /* version */,
+      10 /* sequence */, options.max_sequential_skip_in_iterations,
+      nullptr /* read_callback */));
 
   auto before_lower_bound_str = std::to_string(kLowerBound - 1);
   Slice before_lower_bound(lower_bound_str);
@@ -3148,8 +3238,9 @@ TEST_F(DBIteratorTest, ReverseToForwardWithDisappearingKeys) {
 
   std::unique_ptr<Iterator> db_iter(NewDBIterator(
       env_, ReadOptions(), ImmutableCFOptions(options),
-      MutableCFOptions(options), BytewiseComparator(), internal_iter, 10,
-      options.max_sequential_skip_in_iterations, nullptr /*read_callback*/));
+      MutableCFOptions(options), BytewiseComparator(), internal_iter,
+      nullptr /* version */, 10 /* sequence */,
+      options.max_sequential_skip_in_iterations, nullptr /* read_callback */));
 
   db_iter->SeekForPrev("a");
   ASSERT_TRUE(db_iter->Valid());

db/db_iterator_test.cc

@@ -67,8 +67,8 @@ TEST_P(DBIteratorTest, IteratorProperty) {
   // The test needs to be changed if kPersistedTier is supported in iterator.
   Options options = CurrentOptions();
   CreateAndReopenWithCF({"pikachu"}, options);
-  Put(1, "1", "2");
-  Delete(1, "2");
+  ASSERT_OK(Put(1, "1", "2"));
+  ASSERT_OK(Delete(1, "2"));
   ReadOptions ropt;
   ropt.pin_data = false;
   {
@@ -172,10 +172,10 @@ TEST_P(DBIteratorTest, NonBlockingIteration) {
 TEST_P(DBIteratorTest, IterSeekBeforePrev) {
   ASSERT_OK(Put("a", "b"));
   ASSERT_OK(Put("c", "d"));
-  dbfull()->Flush(FlushOptions());
+  EXPECT_OK(dbfull()->Flush(FlushOptions()));
   ASSERT_OK(Put("0", "f"));
   ASSERT_OK(Put("1", "h"));
-  dbfull()->Flush(FlushOptions());
+  EXPECT_OK(dbfull()->Flush(FlushOptions()));
   ASSERT_OK(Put("2", "j"));
   auto iter = NewIterator(ReadOptions());
   iter->Seek(Slice("c"));
@@ -199,7 +199,7 @@ TEST_P(DBIteratorTest, IterReseekNewUpperBound) {
   ASSERT_OK(Put("aabb", rnd.RandomString(400)));
   ASSERT_OK(Put("aaef", rnd.RandomString(400)));
   ASSERT_OK(Put("b", rnd.RandomString(400)));
-  dbfull()->Flush(FlushOptions());
+  EXPECT_OK(dbfull()->Flush(FlushOptions()));
   ReadOptions opts;
   Slice ub = Slice("aa");
   opts.iterate_upper_bound = &ub;
@@ -215,10 +215,10 @@ TEST_P(DBIteratorTest, IterReseekNewUpperBound) {
 TEST_P(DBIteratorTest, IterSeekForPrevBeforeNext) {
   ASSERT_OK(Put("a", "b"));
   ASSERT_OK(Put("c", "d"));
-  dbfull()->Flush(FlushOptions());
+  EXPECT_OK(dbfull()->Flush(FlushOptions()));
   ASSERT_OK(Put("0", "f"));
   ASSERT_OK(Put("1", "h"));
-  dbfull()->Flush(FlushOptions());
+  EXPECT_OK(dbfull()->Flush(FlushOptions()));
   ASSERT_OK(Put("2", "j"));
   auto iter = NewIterator(ReadOptions());
   iter->SeekForPrev(Slice("0"));
@@ -238,7 +238,7 @@ TEST_P(DBIteratorTest, IterLongKeys) {
   ASSERT_OK(Put(MakeLongKey(20, 0), "0"));
   ASSERT_OK(Put(MakeLongKey(32, 2), "2"));
   ASSERT_OK(Put("a", "b"));
-  dbfull()->Flush(FlushOptions());
+  EXPECT_OK(dbfull()->Flush(FlushOptions()));
   ASSERT_OK(Put(MakeLongKey(50, 1), "1"));
   ASSERT_OK(Put(MakeLongKey(127, 3), "3"));
   ASSERT_OK(Put(MakeLongKey(64, 4), "4"));
@@ -276,7 +276,7 @@ TEST_P(DBIteratorTest, IterLongKeys) {
 
 TEST_P(DBIteratorTest, IterNextWithNewerSeq) {
   ASSERT_OK(Put("0", "0"));
-  dbfull()->Flush(FlushOptions());
+  EXPECT_OK(dbfull()->Flush(FlushOptions()));
   ASSERT_OK(Put("a", "b"));
   ASSERT_OK(Put("c", "d"));
   ASSERT_OK(Put("d", "e"));
@@ -302,7 +302,7 @@ TEST_P(DBIteratorTest, IterNextWithNewerSeq) {
 
 TEST_P(DBIteratorTest, IterPrevWithNewerSeq) {
   ASSERT_OK(Put("0", "0"));
-  dbfull()->Flush(FlushOptions());
+  EXPECT_OK(dbfull()->Flush(FlushOptions()));
   ASSERT_OK(Put("a", "b"));
   ASSERT_OK(Put("c", "d"));
   ASSERT_OK(Put("d", "e"));
@@ -333,7 +333,7 @@ TEST_P(DBIteratorTest, IterPrevWithNewerSeq) {
 
 TEST_P(DBIteratorTest, IterPrevWithNewerSeq2) {
   ASSERT_OK(Put("0", "0"));
-  dbfull()->Flush(FlushOptions());
+  EXPECT_OK(dbfull()->Flush(FlushOptions()));
   ASSERT_OK(Put("a", "b"));
   ASSERT_OK(Put("c", "d"));
   ASSERT_OK(Put("e", "f"));
@@ -377,6 +377,8 @@ TEST_P(DBIteratorTest, IterEmpty) {
     iter->SeekForPrev("foo");
     ASSERT_EQ(IterStatus(iter), "(invalid)");
 
+    ASSERT_OK(iter->status());
+
     delete iter;
   } while (ChangeCompactOptions());
 }
@@ -783,18 +785,18 @@ TEST_P(DBIteratorTest, IterWithSnapshot) {
 TEST_P(DBIteratorTest, IteratorPinsRef) {
   do {
     CreateAndReopenWithCF({"pikachu"}, CurrentOptions());
-    Put(1, "foo", "hello");
+    ASSERT_OK(Put(1, "foo", "hello"));
 
     // Get iterator that will yield the current contents of the DB.
     Iterator* iter = NewIterator(ReadOptions(), handles_[1]);
 
     // Write to force compactions
-    Put(1, "foo", "newvalue1");
+    ASSERT_OK(Put(1, "foo", "newvalue1"));
     for (int i = 0; i < 100; i++) {
       // 100K values
       ASSERT_OK(Put(1, Key(i), Key(i) + std::string(100000, 'v')));
     }
-    Put(1, "foo", "newvalue2");
+    ASSERT_OK(Put(1, "foo", "newvalue2"));
 
     iter->SeekToFirst();
     ASSERT_TRUE(iter->Valid());
@@ -809,8 +811,8 @@ TEST_P(DBIteratorTest, IteratorPinsRef) {
 TEST_P(DBIteratorTest, IteratorDeleteAfterCfDelete) {
   CreateAndReopenWithCF({"pikachu"}, CurrentOptions());
 
-  Put(1, "foo", "delete-cf-then-delete-iter");
-  Put(1, "hello", "value2");
+  ASSERT_OK(Put(1, "foo", "delete-cf-then-delete-iter"));
+  ASSERT_OK(Put(1, "hello", "value2"));
 
   ColumnFamilyHandle* cf = handles_[1];
   ReadOptions ro;
@@ -820,7 +822,7 @@ TEST_P(DBIteratorTest, IteratorDeleteAfterCfDelete) {
   ASSERT_EQ(IterStatus(iter), "foo->delete-cf-then-delete-iter");
 
   // delete CF handle
-  db_->DestroyColumnFamilyHandle(cf);
+  EXPECT_OK(db_->DestroyColumnFamilyHandle(cf));
   handles_.erase(std::begin(handles_) + 1);
 
   // delete Iterator after CF handle is deleted
@@ -832,7 +834,7 @@ TEST_P(DBIteratorTest, IteratorDeleteAfterCfDelete) {
 TEST_P(DBIteratorTest, IteratorDeleteAfterCfDrop) {
   CreateAndReopenWithCF({"pikachu"}, CurrentOptions());
 
-  Put(1, "foo", "drop-cf-then-delete-iter");
+  ASSERT_OK(Put(1, "foo", "drop-cf-then-delete-iter"));
 
   ReadOptions ro;
   ColumnFamilyHandle* cf = handles_[1];
@@ -842,8 +844,8 @@ TEST_P(DBIteratorTest, IteratorDeleteAfterCfDrop) {
   ASSERT_EQ(IterStatus(iter), "foo->drop-cf-then-delete-iter");
 
   // drop and delete CF
-  db_->DropColumnFamily(cf);
-  db_->DestroyColumnFamilyHandle(cf);
+  EXPECT_OK(db_->DropColumnFamily(cf));
+  EXPECT_OK(db_->DestroyColumnFamilyHandle(cf));
   handles_.erase(std::begin(handles_) + 1);
 
   // delete Iterator after CF handle is dropped
@@ -1307,9 +1309,9 @@ TEST_P(DBIteratorTest, PrevAfterAndNextAfterMerge) {
 
   // write three entries with different keys using Merge()
   WriteOptions wopts;
-  db_->Merge(wopts, "1", "data1");
-  db_->Merge(wopts, "2", "data2");
-  db_->Merge(wopts, "3", "data3");
+  ASSERT_OK(db_->Merge(wopts, "1", "data1"));
+  ASSERT_OK(db_->Merge(wopts, "2", "data2"));
+  ASSERT_OK(db_->Merge(wopts, "3", "data3"));
 
   std::unique_ptr<Iterator> it(NewIterator(ReadOptions()));
 
@@ -1393,7 +1395,7 @@ class DBIteratorTestForPinnedData : public DBIteratorTest {
 
       if (run_config == TestConfig::FLUSH_EVERY_1000) {
         if (i && i % 1000 == 0) {
-          Flush();
+          ASSERT_OK(Flush());
         }
       }
     }
@@ -1402,7 +1404,7 @@ class DBIteratorTestForPinnedData : public DBIteratorTest {
       Close();
       Reopen(options);
     } else if (run_config == TestConfig::COMPACT_BEFORE_READ) {
-      db_->CompactRange(CompactRangeOptions(), nullptr, nullptr);
+      ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
     }
 
     ReadOptions ro;
@@ -1768,6 +1770,7 @@ TEST_P(DBIteratorTest, IterSeekForPrevCrossingFiles) {
     Iterator* iter = NewIterator(ro);
     iter->SeekForPrev("c2");
     ASSERT_TRUE(!iter->Valid());
+    ASSERT_OK(iter->status());
     delete iter;
   }
 }
@@ -1823,6 +1826,7 @@ TEST_P(DBIteratorTest, IterSeekForPrevCrossingFilesCustomPrefixExtractor) {
     Iterator* iter = NewIterator(ro);
     iter->SeekForPrev("c2");
     ASSERT_TRUE(!iter->Valid());
+    ASSERT_OK(iter->status());
     delete iter;
   }
 }
@@ -2151,19 +2155,19 @@ TEST_P(DBIteratorTest, ReadAhead) {
 
   std::string value(1024, 'a');
   for (int i = 0; i < 100; i++) {
-    Put(Key(i), value);
+    ASSERT_OK(Put(Key(i), value));
   }
   ASSERT_OK(Flush());
   MoveFilesToLevel(2);
 
   for (int i = 0; i < 100; i++) {
-    Put(Key(i), value);
+    ASSERT_OK(Put(Key(i), value));
   }
   ASSERT_OK(Flush());
   MoveFilesToLevel(1);
 
   for (int i = 0; i < 100; i++) {
-    Put(Key(i), value);
+    ASSERT_OK(Put(Key(i), value));
   }
   ASSERT_OK(Flush());
 #ifndef ROCKSDB_LITE
@@ -2270,6 +2274,7 @@ TEST_P(DBIteratorTest, Refresh) {
   ASSERT_OK(Put("x", "y"));
 
   std::unique_ptr<Iterator> iter(NewIterator(ReadOptions()));
+  ASSERT_OK(iter->status());
   iter->Seek(Slice("a"));
   ASSERT_TRUE(iter->Valid());
   ASSERT_EQ(iter->key().compare(Slice("x")), 0);
@@ -2284,7 +2289,8 @@ TEST_P(DBIteratorTest, Refresh) {
   iter->Next();
   ASSERT_FALSE(iter->Valid());
 
-  iter->Refresh();
+  ASSERT_OK(iter->status());
+  ASSERT_OK(iter->Refresh());
 
   iter->Seek(Slice("a"));
   ASSERT_TRUE(iter->Valid());
@@ -2295,7 +2301,7 @@ TEST_P(DBIteratorTest, Refresh) {
   iter->Next();
   ASSERT_FALSE(iter->Valid());
 
-  dbfull()->Flush(FlushOptions());
+  EXPECT_OK(dbfull()->Flush(FlushOptions()));
 
   ASSERT_OK(Put("m", "n"));
 
@@ -2308,7 +2314,8 @@ TEST_P(DBIteratorTest, Refresh) {
   iter->Next();
   ASSERT_FALSE(iter->Valid());
 
-  iter->Refresh();
+  ASSERT_OK(iter->status());
+  ASSERT_OK(iter->Refresh());
 
   iter->Seek(Slice("a"));
   ASSERT_TRUE(iter->Valid());
@@ -2331,6 +2338,7 @@ TEST_P(DBIteratorTest, RefreshWithSnapshot) {
   ReadOptions options;
   options.snapshot = snapshot;
   Iterator* iter = NewIterator(options);
+  ASSERT_OK(iter->status());
 
   iter->Seek(Slice("a"));
   ASSERT_TRUE(iter->Valid());
@@ -2346,8 +2354,8 @@ TEST_P(DBIteratorTest, RefreshWithSnapshot) {
   iter->Next();
   ASSERT_FALSE(iter->Valid());
 
-  Status s;
-  s = iter->Refresh();
+  ASSERT_OK(iter->status());
+  Status s = iter->Refresh();
   ASSERT_TRUE(s.IsNotSupported());
   db_->ReleaseSnapshot(snapshot);
   delete iter;
@@ -2405,14 +2413,14 @@ TEST_P(DBIteratorTest, UpperBoundWithChangeDirection) {
 
 TEST_P(DBIteratorTest, TableFilter) {
   ASSERT_OK(Put("a", "1"));
-  dbfull()->Flush(FlushOptions());
+  EXPECT_OK(dbfull()->Flush(FlushOptions()));
   ASSERT_OK(Put("b", "2"));
   ASSERT_OK(Put("c", "3"));
-  dbfull()->Flush(FlushOptions());
+  EXPECT_OK(dbfull()->Flush(FlushOptions()));
   ASSERT_OK(Put("d", "4"));
   ASSERT_OK(Put("e", "5"));
   ASSERT_OK(Put("f", "6"));
-  dbfull()->Flush(FlushOptions());
+  EXPECT_OK(dbfull()->Flush(FlushOptions()));
 
   // Ensure the table_filter callback is called once for each table.
   {
@@ -2597,13 +2605,13 @@ TEST_P(DBIteratorTest, SeekAfterHittingManyInternalKeys) {
   ReadOptions ropts;
   ropts.max_skippable_internal_keys = 2;
 
-  Put("1", "val_1");
+  ASSERT_OK(Put("1", "val_1"));
   // Add more tombstones than max_skippable_internal_keys so that Next() fails.
-  Delete("2");
-  Delete("3");
-  Delete("4");
-  Delete("5");
-  Put("6", "val_6");
+  ASSERT_OK(Delete("2"));
+  ASSERT_OK(Delete("3"));
+  ASSERT_OK(Delete("4"));
+  ASSERT_OK(Delete("5"));
+  ASSERT_OK(Put("6", "val_6"));
 
   std::unique_ptr<Iterator> iter(NewIterator(ropts));
   iter->SeekToFirst();
@@ -2645,9 +2653,9 @@ TEST_P(DBIteratorTest, NonBlockingIterationBugRepro) {
   DestroyAndReopen(options);
 
   // Two records in sst file, each in its own block.
-  Put("b", "");
-  Put("d", "");
-  Flush();
+  ASSERT_OK(Put("b", ""));
+  ASSERT_OK(Put("d", ""));
+  ASSERT_OK(Flush());
 
   // Create a nonblocking iterator before writing to memtable.
   ReadOptions ropt;
@@ -2657,7 +2665,7 @@ TEST_P(DBIteratorTest, NonBlockingIterationBugRepro) {
   // Overwrite a key in memtable many times to hit
   // max_sequential_skip_in_iterations (which is 8 by default).
   for (int i = 0; i < 20; ++i) {
-    Put("c", "");
+    ASSERT_OK(Put("c", ""));
   }
 
   // Load the second block in sst file into the block cache.
@@ -2674,9 +2682,9 @@ TEST_P(DBIteratorTest, NonBlockingIterationBugRepro) {
 }
 
 TEST_P(DBIteratorTest, SeekBackwardAfterOutOfUpperBound) {
-  Put("a", "");
-  Put("b", "");
-  Flush();
+  ASSERT_OK(Put("a", ""));
+  ASSERT_OK(Put("b", ""));
+  ASSERT_OK(Flush());
 
   ReadOptions ropt;
   Slice ub = "b";
@@ -2883,6 +2891,127 @@ TEST_P(DBIteratorTest, IterateWithLowerBoundAcrossFileBoundary) {
   ASSERT_OK(iter->status());
 }
 
+TEST_P(DBIteratorTest, Blob) {
+  Options options = CurrentOptions();
+  options.enable_blob_files = true;
+  options.max_sequential_skip_in_iterations = 2;
+  options.statistics = CreateDBStatistics();
+
+  Reopen(options);
+
+  // Note: we have 4 KVs (3 of which are hidden) for key "b" and
+  // max_sequential_skip_in_iterations is set to 2. Thus, we need to do a reseek
+  // anytime we move from "b" to "c" or vice versa.
+  ASSERT_OK(Put("a", "va"));
+  ASSERT_OK(Flush());
+  ASSERT_OK(Put("b", "vb0"));
+  ASSERT_OK(Flush());
+  ASSERT_OK(Put("b", "vb1"));
+  ASSERT_OK(Flush());
+  ASSERT_OK(Put("b", "vb2"));
+  ASSERT_OK(Flush());
+  ASSERT_OK(Put("b", "vb3"));
+  ASSERT_OK(Flush());
+  ASSERT_OK(Put("c", "vc"));
+  ASSERT_OK(Flush());
+
+  std::unique_ptr<Iterator> iter_guard(NewIterator(ReadOptions()));
+  Iterator* const iter = iter_guard.get();
+
+  iter->SeekToFirst();
+  ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 0);
+  ASSERT_EQ(IterStatus(iter), "a->va");
+  iter->Next();
+  ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 0);
+  ASSERT_EQ(IterStatus(iter), "b->vb3");
+  iter->Next();
+  ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 1);
+  ASSERT_EQ(IterStatus(iter), "c->vc");
+  iter->Next();
+  ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 1);
+  ASSERT_EQ(IterStatus(iter), "(invalid)");
+  iter->SeekToFirst();
+  ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 1);
+  ASSERT_EQ(IterStatus(iter), "a->va");
+  iter->Prev();
+  ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 1);
+  ASSERT_EQ(IterStatus(iter), "(invalid)");
+
+  iter->SeekToLast();
+  ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 1);
+  ASSERT_EQ(IterStatus(iter), "c->vc");
+  iter->Prev();
+  ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 2);
+  ASSERT_EQ(IterStatus(iter), "b->vb3");
+  iter->Prev();
+  ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 2);
+  ASSERT_EQ(IterStatus(iter), "a->va");
+  iter->Prev();
+  ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 2);
+  ASSERT_EQ(IterStatus(iter), "(invalid)");
+  iter->SeekToLast();
+  ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 2);
+  ASSERT_EQ(IterStatus(iter), "c->vc");
+  iter->Next();
+  ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 2);
+  ASSERT_EQ(IterStatus(iter), "(invalid)");
+
+  iter->Seek("");
+  ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 2);
+  ASSERT_EQ(IterStatus(iter), "a->va");
+  iter->Seek("a");
+  ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 2);
+  ASSERT_EQ(IterStatus(iter), "a->va");
+  iter->Seek("ax");
+  ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 2);
+  ASSERT_EQ(IterStatus(iter), "b->vb3");
+
+  iter->SeekForPrev("d");
+  ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 2);
+  ASSERT_EQ(IterStatus(iter), "c->vc");
+  iter->SeekForPrev("c");
+  ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 2);
+  ASSERT_EQ(IterStatus(iter), "c->vc");
+  iter->SeekForPrev("bx");
+  ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 3);
+  ASSERT_EQ(IterStatus(iter), "b->vb3");
+
+  iter->Seek("b");
+  ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 3);
+  ASSERT_EQ(IterStatus(iter), "b->vb3");
+  iter->Seek("z");
+  ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 3);
+  ASSERT_EQ(IterStatus(iter), "(invalid)");
+  iter->SeekForPrev("b");
+  ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 4);
+  ASSERT_EQ(IterStatus(iter), "b->vb3");
+  iter->SeekForPrev("");
+  ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 4);
+  ASSERT_EQ(IterStatus(iter), "(invalid)");
+
+  // Switch from reverse to forward
+  iter->SeekToLast();
+  ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 4);
+  iter->Prev();
+  ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 5);
+  iter->Prev();
+  ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 5);
+  iter->Next();
+  ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 5);
+  ASSERT_EQ(IterStatus(iter), "b->vb3");
+
+  // Switch from forward to reverse
+  iter->SeekToFirst();
+  ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 5);
+  iter->Next();
+  ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 5);
+  iter->Next();
+  ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 6);
+  iter->Prev();
+  ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 7);
+  ASSERT_EQ(IterStatus(iter), "b->vb3");
+}
+
 INSTANTIATE_TEST_CASE_P(DBIteratorTestInstance, DBIteratorTest,
                         testing::Values(true, false));
 

db/db_kv_checksum_test.cc

@@ -0,0 +1,199 @@
+//  Copyright (c) 2020-present, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+
+#include "db/db_test_util.h"
+#include "rocksdb/rocksdb_namespace.h"
+
+namespace ROCKSDB_NAMESPACE {
+
+enum class WriteBatchOpType {
+  kPut = 0,
+  kDelete,
+  kSingleDelete,
+  kDeleteRange,
+  kMerge,
+  kBlobIndex,
+  kNum,
+};
+
+// Integer addition is needed for `::testing::Range()` to take the enum type.
+WriteBatchOpType operator+(WriteBatchOpType lhs, const int rhs) {
+  using T = std::underlying_type<WriteBatchOpType>::type;
+  return static_cast<WriteBatchOpType>(static_cast<T>(lhs) + rhs);
+}
+
+class DbKvChecksumTest
+    : public DBTestBase,
+      public ::testing::WithParamInterface<std::tuple<WriteBatchOpType, char>> {
+ public:
+  DbKvChecksumTest()
+      : DBTestBase("/db_kv_checksum_test", /*env_do_fsync=*/false) {
+    op_type_ = std::get<0>(GetParam());
+    corrupt_byte_addend_ = std::get<1>(GetParam());
+  }
+
+  std::pair<WriteBatch, Status> GetWriteBatch(size_t ts_sz,
+                                              ColumnFamilyHandle* cf_handle) {
+    Status s;
+    WriteBatch wb(0 /* reserved_bytes */, 0 /* max_bytes */, ts_sz,
+                  8 /* protection_bytes_per_entry */);
+    switch (op_type_) {
+      case WriteBatchOpType::kPut:
+        s = wb.Put(cf_handle, "key", "val");
+        break;
+      case WriteBatchOpType::kDelete:
+        s = wb.Delete(cf_handle, "key");
+        break;
+      case WriteBatchOpType::kSingleDelete:
+        s = wb.SingleDelete(cf_handle, "key");
+        break;
+      case WriteBatchOpType::kDeleteRange:
+        s = wb.DeleteRange(cf_handle, "begin", "end");
+        break;
+      case WriteBatchOpType::kMerge:
+        s = wb.Merge(cf_handle, "key", "val");
+        break;
+      case WriteBatchOpType::kBlobIndex:
+        // TODO(ajkr): use public API once available.
+        uint32_t cf_id;
+        if (cf_handle == nullptr) {
+          cf_id = 0;
+        } else {
+          cf_id = cf_handle->GetID();
+        }
+        s = WriteBatchInternal::PutBlobIndex(&wb, cf_id, "key", "val");
+        break;
+      case WriteBatchOpType::kNum:
+        assert(false);
+    }
+    return {std::move(wb), std::move(s)};
+  }
+
+  void CorruptNextByteCallBack(void* arg) {
+    Slice encoded = *static_cast<Slice*>(arg);
+    if (entry_len_ == port::kMaxSizet) {
+      // We learn the entry size on the first attempt
+      entry_len_ = encoded.size();
+    }
+    // All entries should be the same size
+    assert(entry_len_ == encoded.size());
+    char* buf = const_cast<char*>(encoded.data());
+    buf[corrupt_byte_offset_] += corrupt_byte_addend_;
+    ++corrupt_byte_offset_;
+  }
+
+  bool MoreBytesToCorrupt() { return corrupt_byte_offset_ < entry_len_; }
+
+ protected:
+  WriteBatchOpType op_type_;
+  char corrupt_byte_addend_;
+  size_t corrupt_byte_offset_ = 0;
+  size_t entry_len_ = port::kMaxSizet;
+};
+
+std::string GetTestNameSuffix(
+    ::testing::TestParamInfo<std::tuple<WriteBatchOpType, char>> info) {
+  std::ostringstream oss;
+  switch (std::get<0>(info.param)) {
+    case WriteBatchOpType::kPut:
+      oss << "Put";
+      break;
+    case WriteBatchOpType::kDelete:
+      oss << "Delete";
+      break;
+    case WriteBatchOpType::kSingleDelete:
+      oss << "SingleDelete";
+      break;
+    case WriteBatchOpType::kDeleteRange:
+      oss << "DeleteRange";
+      break;
+    case WriteBatchOpType::kMerge:
+      oss << "Merge";
+      break;
+    case WriteBatchOpType::kBlobIndex:
+      oss << "BlobIndex";
+      break;
+    case WriteBatchOpType::kNum:
+      assert(false);
+  }
+  oss << "Add"
+      << static_cast<int>(static_cast<unsigned char>(std::get<1>(info.param)));
+  return oss.str();
+}
+
+INSTANTIATE_TEST_CASE_P(
+    DbKvChecksumTest, DbKvChecksumTest,
+    ::testing::Combine(::testing::Range(static_cast<WriteBatchOpType>(0),
+                                        WriteBatchOpType::kNum),
+                       ::testing::Values(2, 103, 251)),
+    GetTestNameSuffix);
+
+TEST_P(DbKvChecksumTest, MemTableAddCorrupted) {
+  // This test repeatedly attempts to write `WriteBatch`es containing a single
+  // entry of type `op_type_`. Each attempt has one byte corrupted in its
+  // memtable entry by adding `corrupt_byte_addend_` to its original value. The
+  // test repeats until an attempt has been made on each byte in the encoded
+  // memtable entry. All attempts are expected to fail with `Status::Corruption`
+  SyncPoint::GetInstance()->SetCallBack(
+      "MemTable::Add:Encoded",
+      std::bind(&DbKvChecksumTest::CorruptNextByteCallBack, this,
+                std::placeholders::_1));
+
+  while (MoreBytesToCorrupt()) {
+    // Failed memtable insert always leads to read-only mode, so we have to
+    // reopen for every attempt.
+    Options options = CurrentOptions();
+    if (op_type_ == WriteBatchOpType::kMerge) {
+      options.merge_operator = MergeOperators::CreateStringAppendOperator();
+    }
+    Reopen(options);
+
+    SyncPoint::GetInstance()->EnableProcessing();
+    auto batch_and_status =
+        GetWriteBatch(0 /* ts_sz */, nullptr /* cf_handle */);
+    ASSERT_OK(batch_and_status.second);
+    ASSERT_TRUE(
+        db_->Write(WriteOptions(), &batch_and_status.first).IsCorruption());
+    SyncPoint::GetInstance()->DisableProcessing();
+  }
+}
+
+TEST_P(DbKvChecksumTest, MemTableAddWithColumnFamilyCorrupted) {
+  // This test repeatedly attempts to write `WriteBatch`es containing a single
+  // entry of type `op_type_` to a non-default column family. Each attempt has
+  // one byte corrupted in its memtable entry by adding `corrupt_byte_addend_`
+  // to its original value. The test repeats until an attempt has been made on
+  // each byte in the encoded memtable entry. All attempts are expected to fail
+  // with `Status::Corruption`.
+  Options options = CurrentOptions();
+  if (op_type_ == WriteBatchOpType::kMerge) {
+    options.merge_operator = MergeOperators::CreateStringAppendOperator();
+  }
+  CreateAndReopenWithCF({"pikachu"}, options);
+  SyncPoint::GetInstance()->SetCallBack(
+      "MemTable::Add:Encoded",
+      std::bind(&DbKvChecksumTest::CorruptNextByteCallBack, this,
+                std::placeholders::_1));
+
+  while (MoreBytesToCorrupt()) {
+    // Failed memtable insert always leads to read-only mode, so we have to
+    // reopen for every attempt.
+    ReopenWithColumnFamilies({kDefaultColumnFamilyName, "pikachu"}, options);
+
+    SyncPoint::GetInstance()->EnableProcessing();
+    auto batch_and_status = GetWriteBatch(0 /* ts_sz */, handles_[1]);
+    ASSERT_OK(batch_and_status.second);
+    ASSERT_TRUE(
+        db_->Write(WriteOptions(), &batch_and_status.first).IsCorruption());
+    SyncPoint::GetInstance()->DisableProcessing();
+  }
+}
+
+}  // namespace ROCKSDB_NAMESPACE
+
+int main(int argc, char** argv) {
+  ::testing::InitGoogleTest(&argc, argv);
+  return RUN_ALL_TESTS();
+}

db/db_log_iter_test.cc

@@ -33,9 +33,8 @@ class DBTestXactLogIterator : public DBTestBase {
 };
 
 namespace {
-SequenceNumber ReadRecords(
-    std::unique_ptr<TransactionLogIterator>& iter,
-    int& count) {
+SequenceNumber ReadRecords(std::unique_ptr<TransactionLogIterator>& iter,
+                           int& count, bool expect_ok = true) {
   count = 0;
   SequenceNumber lastSequence = 0;
   BatchResult res;
@@ -47,6 +46,11 @@ SequenceNumber ReadRecords(
     EXPECT_OK(iter->status());
     iter->Next();
   }
+  if (expect_ok) {
+    EXPECT_OK(iter->status());
+  } else {
+    EXPECT_NOK(iter->status());
+  }
   return res.sequence;
 }
 
@@ -64,9 +68,9 @@ TEST_F(DBTestXactLogIterator, TransactionLogIterator) {
     Options options = OptionsForLogIterTest();
     DestroyAndReopen(options);
     CreateAndReopenWithCF({"pikachu"}, options);
-    Put(0, "key1", DummyString(1024));
-    Put(1, "key2", DummyString(1024));
-    Put(1, "key2", DummyString(1024));
+    ASSERT_OK(Put(0, "key1", DummyString(1024)));
+    ASSERT_OK(Put(1, "key2", DummyString(1024)));
+    ASSERT_OK(Put(1, "key2", DummyString(1024)));
     ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), 3U);
     {
       auto iter = OpenTransactionLogIter(0);
@@ -75,9 +79,9 @@ TEST_F(DBTestXactLogIterator, TransactionLogIterator) {
     ReopenWithColumnFamilies({"default", "pikachu"}, options);
     env_->SleepForMicroseconds(2 * 1000 * 1000);
     {
-      Put(0, "key4", DummyString(1024));
-      Put(1, "key5", DummyString(1024));
-      Put(0, "key6", DummyString(1024));
+      ASSERT_OK(Put(0, "key4", DummyString(1024)));
+      ASSERT_OK(Put(1, "key5", DummyString(1024)));
+      ASSERT_OK(Put(0, "key6", DummyString(1024)));
     }
     {
       auto iter = OpenTransactionLogIter(0);
@@ -109,15 +113,15 @@ TEST_F(DBTestXactLogIterator, TransactionLogIteratorRace) {
       ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
       Options options = OptionsForLogIterTest();
       DestroyAndReopen(options);
-      Put("key1", DummyString(1024));
-      dbfull()->Flush(FlushOptions());
-      Put("key2", DummyString(1024));
-      dbfull()->Flush(FlushOptions());
-      Put("key3", DummyString(1024));
-      dbfull()->Flush(FlushOptions());
-      Put("key4", DummyString(1024));
+      ASSERT_OK(Put("key1", DummyString(1024)));
+      ASSERT_OK(dbfull()->Flush(FlushOptions()));
+      ASSERT_OK(Put("key2", DummyString(1024)));
+      ASSERT_OK(dbfull()->Flush(FlushOptions()));
+      ASSERT_OK(Put("key3", DummyString(1024)));
+      ASSERT_OK(dbfull()->Flush(FlushOptions()));
+      ASSERT_OK(Put("key4", DummyString(1024)));
       ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), 4U);
-      dbfull()->FlushWAL(false);
+      ASSERT_OK(dbfull()->FlushWAL(false));
 
       {
         auto iter = OpenTransactionLogIter(0);
@@ -130,11 +134,11 @@ TEST_F(DBTestXactLogIterator, TransactionLogIteratorRace) {
       // condition
       FlushOptions flush_options;
       flush_options.wait = false;
-      dbfull()->Flush(flush_options);
+      ASSERT_OK(dbfull()->Flush(flush_options));
 
       // "key5" would be written in a new memtable and log
-      Put("key5", DummyString(1024));
-      dbfull()->FlushWAL(false);
+      ASSERT_OK(Put("key5", DummyString(1024)));
+      ASSERT_OK(dbfull()->FlushWAL(false));
       {
         // this iter would miss "key4" if not fixed
         auto iter = OpenTransactionLogIter(0);
@@ -149,14 +153,14 @@ TEST_F(DBTestXactLogIterator, TransactionLogIteratorStallAtLastRecord) {
   do {
     Options options = OptionsForLogIterTest();
     DestroyAndReopen(options);
-    Put("key1", DummyString(1024));
+    ASSERT_OK(Put("key1", DummyString(1024)));
     auto iter = OpenTransactionLogIter(0);
     ASSERT_OK(iter->status());
     ASSERT_TRUE(iter->Valid());
     iter->Next();
     ASSERT_TRUE(!iter->Valid());
     ASSERT_OK(iter->status());
-    Put("key2", DummyString(1024));
+    ASSERT_OK(Put("key2", DummyString(1024)));
     iter->Next();
     ASSERT_OK(iter->status());
     ASSERT_TRUE(iter->Valid());
@@ -167,9 +171,9 @@ TEST_F(DBTestXactLogIterator, TransactionLogIteratorCheckAfterRestart) {
   do {
     Options options = OptionsForLogIterTest();
     DestroyAndReopen(options);
-    Put("key1", DummyString(1024));
-    Put("key2", DummyString(1023));
-    dbfull()->Flush(FlushOptions());
+    ASSERT_OK(Put("key1", DummyString(1024)));
+    ASSERT_OK(Put("key2", DummyString(1023)));
+    ASSERT_OK(dbfull()->Flush(FlushOptions()));
     Reopen(options);
     auto iter = OpenTransactionLogIter(0);
     ExpectRecords(2, iter);
@@ -181,10 +185,10 @@ TEST_F(DBTestXactLogIterator, TransactionLogIteratorCorruptedLog) {
     Options options = OptionsForLogIterTest();
     DestroyAndReopen(options);
     for (int i = 0; i < 1024; i++) {
-      Put("key"+ToString(i), DummyString(10));
+      ASSERT_OK(Put("key" + ToString(i), DummyString(10)));
     }
-    dbfull()->Flush(FlushOptions());
-    dbfull()->FlushWAL(false);
+    ASSERT_OK(dbfull()->Flush(FlushOptions()));
+    ASSERT_OK(dbfull()->FlushWAL(false));
     // Corrupt this log to create a gap
     ROCKSDB_NAMESPACE::VectorLogPtr wal_files;
     ASSERT_OK(dbfull()->GetSortedWalFiles(wal_files));
@@ -197,13 +201,13 @@ TEST_F(DBTestXactLogIterator, TransactionLogIteratorCorruptedLog) {
     }
 
     // Insert a new entry to a new log file
-    Put("key1025", DummyString(10));
-    dbfull()->FlushWAL(false);
+    ASSERT_OK(Put("key1025", DummyString(10)));
+    ASSERT_OK(dbfull()->FlushWAL(false));
     // Try to read from the beginning. Should stop before the gap and read less
     // than 1025 entries
     auto iter = OpenTransactionLogIter(0);
     int count;
-    SequenceNumber last_sequence_read = ReadRecords(iter, count);
+    SequenceNumber last_sequence_read = ReadRecords(iter, count, false);
     ASSERT_LT(last_sequence_read, 1025U);
     // Try to read past the gap, should be able to seek to key1025
     auto iter2 = OpenTransactionLogIter(last_sequence_read + 1);
@@ -217,15 +221,15 @@ TEST_F(DBTestXactLogIterator, TransactionLogIteratorBatchOperations) {
     DestroyAndReopen(options);
     CreateAndReopenWithCF({"pikachu"}, options);
     WriteBatch batch;
-    batch.Put(handles_[1], "key1", DummyString(1024));
-    batch.Put(handles_[0], "key2", DummyString(1024));
-    batch.Put(handles_[1], "key3", DummyString(1024));
-    batch.Delete(handles_[0], "key2");
-    dbfull()->Write(WriteOptions(), &batch);
-    Flush(1);
-    Flush(0);
+    ASSERT_OK(batch.Put(handles_[1], "key1", DummyString(1024)));
+    ASSERT_OK(batch.Put(handles_[0], "key2", DummyString(1024)));
+    ASSERT_OK(batch.Put(handles_[1], "key3", DummyString(1024)));
+    ASSERT_OK(batch.Delete(handles_[0], "key2"));
+    ASSERT_OK(dbfull()->Write(WriteOptions(), &batch));
+    ASSERT_OK(Flush(1));
+    ASSERT_OK(Flush(0));
     ReopenWithColumnFamilies({"default", "pikachu"}, options);
-    Put(1, "key4", DummyString(1024));
+    ASSERT_OK(Put(1, "key4", DummyString(1024)));
     auto iter = OpenTransactionLogIter(3);
     ExpectRecords(2, iter);
   } while (ChangeCompactOptions());
@@ -237,13 +241,13 @@ TEST_F(DBTestXactLogIterator, TransactionLogIteratorBlobs) {
   CreateAndReopenWithCF({"pikachu"}, options);
   {
     WriteBatch batch;
-    batch.Put(handles_[1], "key1", DummyString(1024));
-    batch.Put(handles_[0], "key2", DummyString(1024));
-    batch.PutLogData(Slice("blob1"));
-    batch.Put(handles_[1], "key3", DummyString(1024));
-    batch.PutLogData(Slice("blob2"));
-    batch.Delete(handles_[0], "key2");
-    dbfull()->Write(WriteOptions(), &batch);
+    ASSERT_OK(batch.Put(handles_[1], "key1", DummyString(1024)));
+    ASSERT_OK(batch.Put(handles_[0], "key2", DummyString(1024)));
+    ASSERT_OK(batch.PutLogData(Slice("blob1")));
+    ASSERT_OK(batch.Put(handles_[1], "key3", DummyString(1024)));
+    ASSERT_OK(batch.PutLogData(Slice("blob2")));
+    ASSERT_OK(batch.Delete(handles_[0], "key2"));
+    ASSERT_OK(dbfull()->Write(WriteOptions(), &batch));
     ReopenWithColumnFamilies({"default", "pikachu"}, options);
   }
 
@@ -268,7 +272,7 @@ TEST_F(DBTestXactLogIterator, TransactionLogIteratorBlobs) {
       return Status::OK();
     }
   } handler;
-  res.writeBatchPtr->Iterate(&handler);
+  ASSERT_OK(res.writeBatchPtr->Iterate(&handler));
   ASSERT_EQ(
       "Put(1, key1, 1024)"
       "Put(0, key2, 1024)"

db/db_logical_block_size_cache_test.cc

@@ -401,7 +401,7 @@ TEST_F(DBLogicalBlockSizeCacheTest, MultiDBWithDifferentPaths) {
   ColumnFamilyOptions cf_options0;
   cf_options0.cf_paths = {{cf_path_0_, 1024}};
   ColumnFamilyHandle* cf0;
-  db0->CreateColumnFamily(cf_options0, "cf", &cf0);
+  ASSERT_OK(db0->CreateColumnFamily(cf_options0, "cf", &cf0));
   ASSERT_EQ(2, cache_->Size());
   ASSERT_TRUE(cache_->Contains(data_path_0_));
   ASSERT_EQ(1, cache_->GetRefCount(data_path_0_));
@@ -421,7 +421,7 @@ TEST_F(DBLogicalBlockSizeCacheTest, MultiDBWithDifferentPaths) {
   ColumnFamilyOptions cf_options1;
   cf_options1.cf_paths = {{cf_path_1_, 1024}};
   ColumnFamilyHandle* cf1;
-  db1->CreateColumnFamily(cf_options1, "cf", &cf1);
+  ASSERT_OK(db1->CreateColumnFamily(cf_options1, "cf", &cf1));
   ASSERT_EQ(4, cache_->Size());
   ASSERT_TRUE(cache_->Contains(data_path_0_));
   ASSERT_EQ(1, cache_->GetRefCount(data_path_0_));
@@ -432,7 +432,7 @@ TEST_F(DBLogicalBlockSizeCacheTest, MultiDBWithDifferentPaths) {
   ASSERT_TRUE(cache_->Contains(cf_path_1_));
   ASSERT_EQ(1, cache_->GetRefCount(cf_path_1_));
 
-  db0->DestroyColumnFamilyHandle(cf0);
+  ASSERT_OK(db0->DestroyColumnFamilyHandle(cf0));
   delete db0;
   ASSERT_EQ(2, cache_->Size());
   ASSERT_TRUE(cache_->Contains(data_path_1_));
@@ -441,7 +441,7 @@ TEST_F(DBLogicalBlockSizeCacheTest, MultiDBWithDifferentPaths) {
   ASSERT_EQ(1, cache_->GetRefCount(cf_path_1_));
   ASSERT_OK(DestroyDB(data_path_0_, options, {{"cf", cf_options0}}));
 
-  db1->DestroyColumnFamilyHandle(cf1);
+  ASSERT_OK(db1->DestroyColumnFamilyHandle(cf1));
   delete db1;
   ASSERT_EQ(0, cache_->Size());
   ASSERT_OK(DestroyDB(data_path_1_, options, {{"cf", cf_options1}}));
@@ -466,7 +466,7 @@ TEST_F(DBLogicalBlockSizeCacheTest, MultiDBWithSamePaths) {
   ASSERT_EQ(1, cache_->GetRefCount(data_path_0_));
 
   ColumnFamilyHandle* cf0;
-  db0->CreateColumnFamily(cf_options, "cf", &cf0);
+  ASSERT_OK(db0->CreateColumnFamily(cf_options, "cf", &cf0));
   ASSERT_EQ(2, cache_->Size());
   ASSERT_TRUE(cache_->Contains(data_path_0_));
   ASSERT_EQ(1, cache_->GetRefCount(data_path_0_));
@@ -482,14 +482,14 @@ TEST_F(DBLogicalBlockSizeCacheTest, MultiDBWithSamePaths) {
   ASSERT_EQ(1, cache_->GetRefCount(cf_path_0_));
 
   ColumnFamilyHandle* cf1;
-  db1->CreateColumnFamily(cf_options, "cf", &cf1);
+  ASSERT_OK(db1->CreateColumnFamily(cf_options, "cf", &cf1));
   ASSERT_EQ(2, cache_->Size());
   ASSERT_TRUE(cache_->Contains(data_path_0_));
   ASSERT_EQ(2, cache_->GetRefCount(data_path_0_));
   ASSERT_TRUE(cache_->Contains(cf_path_0_));
   ASSERT_EQ(2, cache_->GetRefCount(cf_path_0_));
 
-  db0->DestroyColumnFamilyHandle(cf0);
+  ASSERT_OK(db0->DestroyColumnFamilyHandle(cf0));
   delete db0;
   ASSERT_EQ(2, cache_->Size());
   ASSERT_TRUE(cache_->Contains(data_path_0_));
@@ -498,7 +498,7 @@ TEST_F(DBLogicalBlockSizeCacheTest, MultiDBWithSamePaths) {
   ASSERT_EQ(1, cache_->GetRefCount(cf_path_0_));
   ASSERT_OK(DestroyDB(dbname_ + "/db0", options, {{"cf", cf_options}}));
 
-  db1->DestroyColumnFamilyHandle(cf1);
+  ASSERT_OK(db1->DestroyColumnFamilyHandle(cf1));
   delete db1;
   ASSERT_EQ(0, cache_->Size());
   ASSERT_OK(DestroyDB(dbname_ + "/db1", options, {{"cf", cf_options}}));

db/db_memtable_test.cc

@@ -129,7 +129,6 @@ class TestPrefixExtractor : public SliceTransform {
 TEST_F(DBMemTableTest, DuplicateSeq) {
   SequenceNumber seq = 123;
   std::string value;
-  Status s;
   MergeContext merge_context;
   Options options;
   InternalKeyComparator ikey_cmp(options.comparator);
@@ -146,22 +145,25 @@ TEST_F(DBMemTableTest, DuplicateSeq) {
                                kMaxSequenceNumber, 0 /* column_family_id */);
 
   // Write some keys and make sure it returns false on duplicates
-  bool res;
-  res = mem->Add(seq, kTypeValue, "key", "value2");
-  ASSERT_TRUE(res);
-  res = mem->Add(seq, kTypeValue, "key", "value2");
-  ASSERT_FALSE(res);
+  ASSERT_OK(
+      mem->Add(seq, kTypeValue, "key", "value2", nullptr /* kv_prot_info */));
+  ASSERT_TRUE(
+      mem->Add(seq, kTypeValue, "key", "value2", nullptr /* kv_prot_info */)
+          .IsTryAgain());
   // Changing the type should still cause the duplicatae key
-  res = mem->Add(seq, kTypeMerge, "key", "value2");
-  ASSERT_FALSE(res);
+  ASSERT_TRUE(
+      mem->Add(seq, kTypeMerge, "key", "value2", nullptr /* kv_prot_info */)
+          .IsTryAgain());
   // Changing the seq number will make the key fresh
-  res = mem->Add(seq + 1, kTypeMerge, "key", "value2");
-  ASSERT_TRUE(res);
+  ASSERT_OK(mem->Add(seq + 1, kTypeMerge, "key", "value2",
+                     nullptr /* kv_prot_info */));
   // Test with different types for duplicate keys
-  res = mem->Add(seq, kTypeDeletion, "key", "");
-  ASSERT_FALSE(res);
-  res = mem->Add(seq, kTypeSingleDeletion, "key", "");
-  ASSERT_FALSE(res);
+  ASSERT_TRUE(
+      mem->Add(seq, kTypeDeletion, "key", "", nullptr /* kv_prot_info */)
+          .IsTryAgain());
+  ASSERT_TRUE(
+      mem->Add(seq, kTypeSingleDeletion, "key", "", nullptr /* kv_prot_info */)
+          .IsTryAgain());
 
   // Test the duplicate keys under stress
   for (int i = 0; i < 10000; i++) {
@@ -169,11 +171,12 @@ TEST_F(DBMemTableTest, DuplicateSeq) {
     if (!insert_dup) {
       seq++;
     }
-    res = mem->Add(seq, kTypeValue, "foo", "value" + ToString(seq));
+    Status s = mem->Add(seq, kTypeValue, "foo", "value" + ToString(seq),
+                        nullptr /* kv_prot_info */);
     if (insert_dup) {
-      ASSERT_FALSE(res);
+      ASSERT_TRUE(s.IsTryAgain());
     } else {
-      ASSERT_TRUE(res);
+      ASSERT_OK(s);
     }
   }
   delete mem;
@@ -185,10 +188,11 @@ TEST_F(DBMemTableTest, DuplicateSeq) {
   mem = new MemTable(cmp, ioptions, MutableCFOptions(options), &wb,
                      kMaxSequenceNumber, 0 /* column_family_id */);
   // Insert a duplicate key with _ in it
-  res = mem->Add(seq, kTypeValue, "key_1", "value");
-  ASSERT_TRUE(res);
-  res = mem->Add(seq, kTypeValue, "key_1", "value");
-  ASSERT_FALSE(res);
+  ASSERT_OK(
+      mem->Add(seq, kTypeValue, "key_1", "value", nullptr /* kv_prot_info */));
+  ASSERT_TRUE(
+      mem->Add(seq, kTypeValue, "key_1", "value", nullptr /* kv_prot_info */)
+          .IsTryAgain());
   delete mem;
 
   // Test when InsertConcurrently will be invoked
@@ -197,10 +201,11 @@ TEST_F(DBMemTableTest, DuplicateSeq) {
   mem = new MemTable(cmp, ioptions, MutableCFOptions(options), &wb,
                      kMaxSequenceNumber, 0 /* column_family_id */);
   MemTablePostProcessInfo post_process_info;
-  res = mem->Add(seq, kTypeValue, "key", "value", true, &post_process_info);
-  ASSERT_TRUE(res);
-  res = mem->Add(seq, kTypeValue, "key", "value", true, &post_process_info);
-  ASSERT_FALSE(res);
+  ASSERT_OK(mem->Add(seq, kTypeValue, "key", "value",
+                     nullptr /* kv_prot_info */, true, &post_process_info));
+  ASSERT_TRUE(mem->Add(seq, kTypeValue, "key", "value",
+                       nullptr /* kv_prot_info */, true, &post_process_info)
+                  .IsTryAgain());
   delete mem;
 }
 
@@ -208,7 +213,6 @@ TEST_F(DBMemTableTest, DuplicateSeq) {
 TEST_F(DBMemTableTest, ConcurrentMergeWrite) {
   int num_ops = 1000;
   std::string value;
-  Status s;
   MergeContext merge_context;
   Options options;
   // A merge operator that is not sensitive to concurrent writes since in this
@@ -227,8 +231,7 @@ TEST_F(DBMemTableTest, ConcurrentMergeWrite) {
 
   // Put 0 as the base
   PutFixed64(&value, static_cast<uint64_t>(0));
-  bool res = mem->Add(0, kTypeValue, "key", value);
-  ASSERT_TRUE(res);
+  ASSERT_OK(mem->Add(0, kTypeValue, "key", value, nullptr /* kv_prot_info */));
   value.clear();
 
   // Write Merge concurrently
@@ -237,9 +240,8 @@ TEST_F(DBMemTableTest, ConcurrentMergeWrite) {
     std::string v1;
     for (int seq = 1; seq < num_ops / 2; seq++) {
       PutFixed64(&v1, seq);
-      bool res1 =
-          mem->Add(seq, kTypeMerge, "key", v1, true, &post_process_info1);
-      ASSERT_TRUE(res1);
+      ASSERT_OK(mem->Add(seq, kTypeMerge, "key", v1, nullptr /* kv_prot_info */,
+                         true, &post_process_info1));
       v1.clear();
     }
   });
@@ -248,9 +250,8 @@ TEST_F(DBMemTableTest, ConcurrentMergeWrite) {
     std::string v2;
     for (int seq = num_ops / 2; seq < num_ops; seq++) {
       PutFixed64(&v2, seq);
-      bool res2 =
-          mem->Add(seq, kTypeMerge, "key", v2, true, &post_process_info2);
-      ASSERT_TRUE(res2);
+      ASSERT_OK(mem->Add(seq, kTypeMerge, "key", v2, nullptr /* kv_prot_info */,
+                         true, &post_process_info2));
       v2.clear();
     }
   });
@@ -261,8 +262,9 @@ TEST_F(DBMemTableTest, ConcurrentMergeWrite) {
   ReadOptions roptions;
   SequenceNumber max_covering_tombstone_seq = 0;
   LookupKey lkey("key", kMaxSequenceNumber);
-  res = mem->Get(lkey, &value, /*timestamp=*/nullptr, &status, &merge_context,
-                 &max_covering_tombstone_seq, roptions);
+  bool res = mem->Get(lkey, &value, /*timestamp=*/nullptr, &status,
+                      &merge_context, &max_covering_tombstone_seq, roptions);
+  ASSERT_OK(status);
   ASSERT_TRUE(res);
   uint64_t ivalue = DecodeFixed64(Slice(value).data());
   uint64_t sum = 0;

db/db_merge_operand_test.cc

@@ -59,29 +59,29 @@ TEST_F(DBMergeOperandTest, GetMergeOperandsBasic) {
   merge_operands_info.expected_max_number_of_operands = num_records;
 
   // k0 value in memtable
-  Put("k0", "PutARock");
-  db_->GetMergeOperands(ReadOptions(), db_->DefaultColumnFamily(), "k0",
-                        values.data(), &merge_operands_info,
-                        &number_of_operands);
+  ASSERT_OK(Put("k0", "PutARock"));
+  ASSERT_OK(db_->GetMergeOperands(ReadOptions(), db_->DefaultColumnFamily(),
+                                  "k0", values.data(), &merge_operands_info,
+                                  &number_of_operands));
   ASSERT_EQ(values[0], "PutARock");
 
   // k0.1 value in SST
-  Put("k0.1", "RockInSST");
+  ASSERT_OK(Put("k0.1", "RockInSST"));
   ASSERT_OK(Flush());
-  db_->GetMergeOperands(ReadOptions(), db_->DefaultColumnFamily(), "k0.1",
-                        values.data(), &merge_operands_info,
-                        &number_of_operands);
+  ASSERT_OK(db_->GetMergeOperands(ReadOptions(), db_->DefaultColumnFamily(),
+                                  "k0.1", values.data(), &merge_operands_info,
+                                  &number_of_operands));
   ASSERT_EQ(values[0], "RockInSST");
 
   // All k1 values are in memtable.
   ASSERT_OK(Merge("k1", "a"));
-  Put("k1", "x");
+  ASSERT_OK(Put("k1", "x"));
   ASSERT_OK(Merge("k1", "b"));
   ASSERT_OK(Merge("k1", "c"));
   ASSERT_OK(Merge("k1", "d"));
-  db_->GetMergeOperands(ReadOptions(), db_->DefaultColumnFamily(), "k1",
-                        values.data(), &merge_operands_info,
-                        &number_of_operands);
+  ASSERT_OK(db_->GetMergeOperands(ReadOptions(), db_->DefaultColumnFamily(),
+                                  "k1", values.data(), &merge_operands_info,
+                                  &number_of_operands));
   ASSERT_EQ(values[0], "x");
   ASSERT_EQ(values[1], "b");
   ASSERT_EQ(values[2], "c");
@@ -98,13 +98,13 @@ TEST_F(DBMergeOperandTest, GetMergeOperandsBasic) {
 
   // All k1.1 values are in memtable.
   ASSERT_OK(Merge("k1.1", "r"));
-  Delete("k1.1");
+  ASSERT_OK(Delete("k1.1"));
   ASSERT_OK(Merge("k1.1", "c"));
   ASSERT_OK(Merge("k1.1", "k"));
   ASSERT_OK(Merge("k1.1", "s"));
-  db_->GetMergeOperands(ReadOptions(), db_->DefaultColumnFamily(), "k1.1",
-                        values.data(), &merge_operands_info,
-                        &number_of_operands);
+  ASSERT_OK(db_->GetMergeOperands(ReadOptions(), db_->DefaultColumnFamily(),
+                                  "k1.1", values.data(), &merge_operands_info,
+                                  &number_of_operands));
   ASSERT_EQ(values[0], "c");
   ASSERT_EQ(values[1], "k");
   ASSERT_EQ(values[2], "s");
@@ -115,9 +115,9 @@ TEST_F(DBMergeOperandTest, GetMergeOperandsBasic) {
   ASSERT_OK(Merge("k2", "e"));
   ASSERT_OK(Merge("k2", "r"));
   ASSERT_OK(Flush());
-  db_->GetMergeOperands(ReadOptions(), db_->DefaultColumnFamily(), "k2",
-                        values.data(), &merge_operands_info,
-                        &number_of_operands);
+  ASSERT_OK(db_->GetMergeOperands(ReadOptions(), db_->DefaultColumnFamily(),
+                                  "k2", values.data(), &merge_operands_info,
+                                  &number_of_operands));
   ASSERT_EQ(values[0], "q");
   ASSERT_EQ(values[1], "w");
   ASSERT_EQ(values[2], "e");
@@ -125,30 +125,30 @@ TEST_F(DBMergeOperandTest, GetMergeOperandsBasic) {
 
   // All k2.1 values are flushed to L0 into a single file.
   ASSERT_OK(Merge("k2.1", "m"));
-  Put("k2.1", "l");
+  ASSERT_OK(Put("k2.1", "l"));
   ASSERT_OK(Merge("k2.1", "n"));
   ASSERT_OK(Merge("k2.1", "o"));
   ASSERT_OK(Flush());
-  db_->GetMergeOperands(ReadOptions(), db_->DefaultColumnFamily(), "k2.1",
-                        values.data(), &merge_operands_info,
-                        &number_of_operands);
+  ASSERT_OK(db_->GetMergeOperands(ReadOptions(), db_->DefaultColumnFamily(),
+                                  "k2.1", values.data(), &merge_operands_info,
+                                  &number_of_operands));
   ASSERT_EQ(values[0], "l,n,o");
 
   // All k2.2 values are flushed to L0 into a single file.
   ASSERT_OK(Merge("k2.2", "g"));
-  Delete("k2.2");
+  ASSERT_OK(Delete("k2.2"));
   ASSERT_OK(Merge("k2.2", "o"));
   ASSERT_OK(Merge("k2.2", "t"));
   ASSERT_OK(Flush());
-  db_->GetMergeOperands(ReadOptions(), db_->DefaultColumnFamily(), "k2.2",
-                        values.data(), &merge_operands_info,
-                        &number_of_operands);
+  ASSERT_OK(db_->GetMergeOperands(ReadOptions(), db_->DefaultColumnFamily(),
+                                  "k2.2", values.data(), &merge_operands_info,
+                                  &number_of_operands));
   ASSERT_EQ(values[0], "o,t");
 
   // Do some compaction that will make the following tests more predictable
   //  Slice start("PutARock");
   //  Slice end("t");
-  db_->CompactRange(CompactRangeOptions(), nullptr, nullptr);
+  ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
 
   // All k3 values are flushed and are in different files.
   ASSERT_OK(Merge("k3", "ab"));
@@ -158,9 +158,9 @@ TEST_F(DBMergeOperandTest, GetMergeOperandsBasic) {
   ASSERT_OK(Merge("k3", "cd"));
   ASSERT_OK(Flush());
   ASSERT_OK(Merge("k3", "de"));
-  db_->GetMergeOperands(ReadOptions(), db_->DefaultColumnFamily(), "k3",
-                        values.data(), &merge_operands_info,
-                        &number_of_operands);
+  ASSERT_OK(db_->GetMergeOperands(ReadOptions(), db_->DefaultColumnFamily(),
+                                  "k3", values.data(), &merge_operands_info,
+                                  &number_of_operands));
   ASSERT_EQ(values[0], "ab");
   ASSERT_EQ(values[1], "bc");
   ASSERT_EQ(values[2], "cd");
@@ -169,14 +169,14 @@ TEST_F(DBMergeOperandTest, GetMergeOperandsBasic) {
   // All k3.1 values are flushed and are in different files.
   ASSERT_OK(Merge("k3.1", "ab"));
   ASSERT_OK(Flush());
-  Put("k3.1", "bc");
+  ASSERT_OK(Put("k3.1", "bc"));
   ASSERT_OK(Flush());
   ASSERT_OK(Merge("k3.1", "cd"));
   ASSERT_OK(Flush());
   ASSERT_OK(Merge("k3.1", "de"));
-  db_->GetMergeOperands(ReadOptions(), db_->DefaultColumnFamily(), "k3.1",
-                        values.data(), &merge_operands_info,
-                        &number_of_operands);
+  ASSERT_OK(db_->GetMergeOperands(ReadOptions(), db_->DefaultColumnFamily(),
+                                  "k3.1", values.data(), &merge_operands_info,
+                                  &number_of_operands));
   ASSERT_EQ(values[0], "bc");
   ASSERT_EQ(values[1], "cd");
   ASSERT_EQ(values[2], "de");
@@ -184,14 +184,14 @@ TEST_F(DBMergeOperandTest, GetMergeOperandsBasic) {
   // All k3.2 values are flushed and are in different files.
   ASSERT_OK(Merge("k3.2", "ab"));
   ASSERT_OK(Flush());
-  Delete("k3.2");
+  ASSERT_OK(Delete("k3.2"));
   ASSERT_OK(Flush());
   ASSERT_OK(Merge("k3.2", "cd"));
   ASSERT_OK(Flush());
   ASSERT_OK(Merge("k3.2", "de"));
-  db_->GetMergeOperands(ReadOptions(), db_->DefaultColumnFamily(), "k3.2",
-                        values.data(), &merge_operands_info,
-                        &number_of_operands);
+  ASSERT_OK(db_->GetMergeOperands(ReadOptions(), db_->DefaultColumnFamily(),
+                                  "k3.2", values.data(), &merge_operands_info,
+                                  &number_of_operands));
   ASSERT_EQ(values[0], "cd");
   ASSERT_EQ(values[1], "de");
 
@@ -206,9 +206,9 @@ TEST_F(DBMergeOperandTest, GetMergeOperandsBasic) {
   ASSERT_OK(Flush());
   MoveFilesToLevel(1);
   ASSERT_OK(Merge("k4", "ed"));
-  db_->GetMergeOperands(ReadOptions(), db_->DefaultColumnFamily(), "k4",
-                        values.data(), &merge_operands_info,
-                        &number_of_operands);
+  ASSERT_OK(db_->GetMergeOperands(ReadOptions(), db_->DefaultColumnFamily(),
+                                  "k4", values.data(), &merge_operands_info,
+                                  &number_of_operands));
   ASSERT_EQ(values[0], "ba");
   ASSERT_EQ(values[1], "cb");
   ASSERT_EQ(values[2], "dc");
@@ -219,14 +219,14 @@ TEST_F(DBMergeOperandTest, GetMergeOperandsBasic) {
   ASSERT_OK(Merge("k5", "am"));
   ASSERT_OK(Merge("k5", "i"));
   ASSERT_OK(Flush());
-  Put("k5", "remember");
+  ASSERT_OK(Put("k5", "remember"));
   ASSERT_OK(Merge("k5", "i"));
   ASSERT_OK(Merge("k5", "am"));
   ASSERT_OK(Merge("k5", "rocks"));
-  dbfull()->TEST_SwitchMemtable();
-  db_->GetMergeOperands(ReadOptions(), db_->DefaultColumnFamily(), "k5",
-                        values.data(), &merge_operands_info,
-                        &number_of_operands);
+  ASSERT_OK(dbfull()->TEST_SwitchMemtable());
+  ASSERT_OK(db_->GetMergeOperands(ReadOptions(), db_->DefaultColumnFamily(),
+                                  "k5", values.data(), &merge_operands_info,
+                                  &number_of_operands));
   ASSERT_EQ(values[0], "remember");
   ASSERT_EQ(values[1], "i");
   ASSERT_EQ(values[2], "am");

db/db_merge_operator_test.cc

@@ -94,7 +94,7 @@ TEST_F(DBMergeOperatorTest, LimitMergeOperands) {
   ASSERT_OK(Merge("k1", "c"));
   ASSERT_OK(Merge("k1", "d"));
   std::string value;
-  ASSERT_TRUE(db_->Get(ReadOptions(), "k1", &value).ok());
+  ASSERT_OK(db_->Get(ReadOptions(), "k1", &value));
   // Make sure that only the latest two merge operands are used. If this was
   // not the case the value would be "a,b,c,d".
   ASSERT_EQ(value, "c,d");
@@ -105,7 +105,7 @@ TEST_F(DBMergeOperatorTest, LimitMergeOperands) {
   ASSERT_OK(Merge("k2", "c"));
   ASSERT_OK(Merge("k2", "d"));
   ASSERT_OK(Flush());
-  ASSERT_TRUE(db_->Get(ReadOptions(), "k2", &value).ok());
+  ASSERT_OK(db_->Get(ReadOptions(), "k2", &value));
   ASSERT_EQ(value, "c,d");
 
   // All K3 values are flushed and are in different files.
@@ -116,7 +116,7 @@ TEST_F(DBMergeOperatorTest, LimitMergeOperands) {
   ASSERT_OK(Merge("k3", "cd"));
   ASSERT_OK(Flush());
   ASSERT_OK(Merge("k3", "de"));
-  ASSERT_TRUE(db_->Get(ReadOptions(), "k3", &value).ok());
+  ASSERT_OK(db_->Get(ReadOptions(), "k3", &value));
   ASSERT_EQ(value, "cd,de");
 
   // All K4 values are in different levels
@@ -130,7 +130,7 @@ TEST_F(DBMergeOperatorTest, LimitMergeOperands) {
   ASSERT_OK(Flush());
   MoveFilesToLevel(1);
   ASSERT_OK(Merge("k4", "de"));
-  ASSERT_TRUE(db_->Get(ReadOptions(), "k4", &value).ok());
+  ASSERT_OK(db_->Get(ReadOptions(), "k4", &value));
   ASSERT_EQ(value, "cd,de");
 }
 
@@ -344,8 +344,9 @@ TEST_P(MergeOperatorPinningTest, EvictCacheBeforeMerge) {
   // Code executed before merge operation
   merge_hook->before_merge_ = [&]() {
     // Evict all tables from cache before every merge operation
+    auto* table_cache = dbfull()->TEST_table_cache();
     for (uint64_t num : file_numbers) {
-      TableCache::Evict(dbfull()->TEST_table_cache(), num);
+      TableCache::Evict(table_cache, num);
     }
     // Decrease cache capacity to force all unrefed blocks to be evicted
     if (bbto.block_cache) {
@@ -366,7 +367,7 @@ TEST_P(MergeOperatorPinningTest, EvictCacheBeforeMerge) {
   VerifyDBFromMap(true_data, &total_reads);
   ASSERT_EQ(merge_cnt, total_reads);
 
-  db_->CompactRange(CompactRangeOptions(), nullptr, nullptr);
+  ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
 
   VerifyDBFromMap(true_data, &total_reads);
 }
@@ -385,7 +386,7 @@ TEST_P(MergeOperatorPinningTest, TailingIterator) {
   std::function<void()> writer_func = [&]() {
     int k = 0;
     for (int i = 0; i < kNumWrites; i++) {
-      db_->Merge(WriteOptions(), Key(k), Key(k));
+      ASSERT_OK(db_->Merge(WriteOptions(), Key(k), Key(k)));
 
       if (i && i % kNumOperands == 0) {
         k++;
@@ -403,7 +404,7 @@ TEST_P(MergeOperatorPinningTest, TailingIterator) {
     ReadOptions ro;
     ro.tailing = true;
     Iterator* iter = db_->NewIterator(ro);
-
+    ASSERT_OK(iter->status());
     iter->SeekToFirst();
     for (int i = 0; i < (kNumWrites / kNumOperands); i++) {
       while (!iter->Valid()) {
@@ -416,6 +417,7 @@ TEST_P(MergeOperatorPinningTest, TailingIterator) {
 
       iter->Next();
     }
+    ASSERT_OK(iter->status());
 
     delete iter;
   };
@@ -449,12 +451,13 @@ TEST_F(DBMergeOperatorTest, TailingIteratorMemtableUnrefedBySomeoneElse) {
   //    ForwardIterator to not pin it in some circumstances. This test
   //    reproduces it.
 
-  db_->Merge(WriteOptions(), "key", "sst");
-  db_->Flush(FlushOptions()); // Switch to SuperVersion A
-  db_->Merge(WriteOptions(), "key", "memtable");
+  ASSERT_OK(db_->Merge(WriteOptions(), "key", "sst"));
+  ASSERT_OK(db_->Flush(FlushOptions()));  // Switch to SuperVersion A
+  ASSERT_OK(db_->Merge(WriteOptions(), "key", "memtable"));
 
   // Pin SuperVersion A
   std::unique_ptr<Iterator> someone_else(db_->NewIterator(ReadOptions()));
+  ASSERT_OK(someone_else->status());
 
   bool pushed_first_operand = false;
   bool stepped_to_next_operand = false;
@@ -462,7 +465,7 @@ TEST_F(DBMergeOperatorTest, TailingIteratorMemtableUnrefedBySomeoneElse) {
       "DBIter::MergeValuesNewToOld:PushedFirstOperand", [&](void*) {
         EXPECT_FALSE(pushed_first_operand);
         pushed_first_operand = true;
-        db_->Flush(FlushOptions()); // Switch to SuperVersion B
+        EXPECT_OK(db_->Flush(FlushOptions()));  // Switch to SuperVersion B
       });
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
       "DBIter::MergeValuesNewToOld:SteppedToNextOperand", [&](void*) {
@@ -477,7 +480,7 @@ TEST_F(DBMergeOperatorTest, TailingIteratorMemtableUnrefedBySomeoneElse) {
   std::unique_ptr<Iterator> iter(db_->NewIterator(ro));
   iter->Seek("key");
 
-  ASSERT_TRUE(iter->status().ok());
+  ASSERT_OK(iter->status());
   ASSERT_TRUE(iter->Valid());
   EXPECT_EQ(std::string("sst,memtable"), iter->value().ToString());
   EXPECT_TRUE(pushed_first_operand);

db/db_options_test.cc

@@ -129,6 +129,83 @@ TEST_F(DBOptionsTest, GetLatestCFOptions) {
             GetMutableCFOptionsMap(dbfull()->GetOptions(handles_[1])));
 }
 
+TEST_F(DBOptionsTest, SetMutableTableOptions) {
+  Options options;
+  options.create_if_missing = true;
+  options.env = env_;
+  options.blob_file_size = 16384;
+  BlockBasedTableOptions bbto;
+  bbto.no_block_cache = true;
+  bbto.block_size = 8192;
+  bbto.block_restart_interval = 7;
+
+  options.table_factory.reset(NewBlockBasedTableFactory(bbto));
+  Reopen(options);
+
+  ColumnFamilyHandle* cfh = dbfull()->DefaultColumnFamily();
+  Options c_opts = dbfull()->GetOptions(cfh);
+  const auto* c_bbto =
+      c_opts.table_factory->GetOptions<BlockBasedTableOptions>();
+  ASSERT_NE(c_bbto, nullptr);
+  ASSERT_EQ(c_opts.blob_file_size, 16384);
+  ASSERT_EQ(c_bbto->no_block_cache, true);
+  ASSERT_EQ(c_bbto->block_size, 8192);
+  ASSERT_EQ(c_bbto->block_restart_interval, 7);
+  ASSERT_OK(dbfull()->SetOptions(
+      cfh, {{"table_factory.block_size", "16384"},
+            {"table_factory.block_restart_interval", "11"}}));
+  ASSERT_EQ(c_bbto->block_size, 16384);
+  ASSERT_EQ(c_bbto->block_restart_interval, 11);
+
+  // Now set an option that is not mutable - options should not change
+  ASSERT_NOK(
+      dbfull()->SetOptions(cfh, {{"table_factory.no_block_cache", "false"}}));
+  ASSERT_EQ(c_bbto->no_block_cache, true);
+  ASSERT_EQ(c_bbto->block_size, 16384);
+  ASSERT_EQ(c_bbto->block_restart_interval, 11);
+
+  // Set some that are mutable and some that are not - options should not change
+  ASSERT_NOK(dbfull()->SetOptions(
+      cfh, {{"table_factory.no_block_cache", "false"},
+            {"table_factory.block_size", "8192"},
+            {"table_factory.block_restart_interval", "7"}}));
+  ASSERT_EQ(c_bbto->no_block_cache, true);
+  ASSERT_EQ(c_bbto->block_size, 16384);
+  ASSERT_EQ(c_bbto->block_restart_interval, 11);
+
+  // Set some that are mutable and some that do not exist - options should not
+  // change
+  ASSERT_NOK(dbfull()->SetOptions(
+      cfh, {{"table_factory.block_size", "8192"},
+            {"table_factory.does_not_exist", "true"},
+            {"table_factory.block_restart_interval", "7"}}));
+  ASSERT_EQ(c_bbto->no_block_cache, true);
+  ASSERT_EQ(c_bbto->block_size, 16384);
+  ASSERT_EQ(c_bbto->block_restart_interval, 11);
+
+  // Trying to change the table factory fails
+  ASSERT_NOK(dbfull()->SetOptions(
+      cfh, {{"table_factory", TableFactory::kPlainTableName()}}));
+
+  // Set some on the table and some on the Column Family
+  ASSERT_OK(dbfull()->SetOptions(
+      cfh, {{"table_factory.block_size", "16384"},
+            {"blob_file_size", "32768"},
+            {"table_factory.block_restart_interval", "13"}}));
+  c_opts = dbfull()->GetOptions(cfh);
+  ASSERT_EQ(c_opts.blob_file_size, 32768);
+  ASSERT_EQ(c_bbto->block_size, 16384);
+  ASSERT_EQ(c_bbto->block_restart_interval, 13);
+  // Set some on the table and a bad one on the ColumnFamily - options should
+  // not change
+  ASSERT_NOK(dbfull()->SetOptions(
+      cfh, {{"table_factory.block_size", "1024"},
+            {"no_such_option", "32768"},
+            {"table_factory.block_restart_interval", "7"}}));
+  ASSERT_EQ(c_bbto->block_size, 16384);
+  ASSERT_EQ(c_bbto->block_restart_interval, 13);
+}
+
 TEST_F(DBOptionsTest, SetBytesPerSync) {
   const size_t kValueSize = 1024 * 1024;  // 1MB
   Options options;
@@ -190,10 +267,11 @@ TEST_F(DBOptionsTest, SetWalBytesPerSync) {
   options.env = env_;
   Reopen(options);
   ASSERT_EQ(512, dbfull()->GetDBOptions().wal_bytes_per_sync);
-  int counter = 0;
+  std::atomic_int counter{0};
   int low_bytes_per_sync = 0;
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
-      "WritableFileWriter::RangeSync:0", [&](void* /*arg*/) { counter++; });
+      "WritableFileWriter::RangeSync:0",
+      [&](void* /*arg*/) { counter.fetch_add(1); });
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
   const std::string kValue(kValueSize, 'v');
   int i = 0;
@@ -729,7 +807,7 @@ TEST_F(DBOptionsTest, SetFIFOCompactionOptions) {
     for (int j = 0; j < 10; j++) {
       ASSERT_OK(Put(ToString(i * 20 + j), rnd.RandomString(980)));
     }
-    Flush();
+    ASSERT_OK(Flush());
   }
   ASSERT_OK(dbfull()->TEST_WaitForCompact());
   ASSERT_EQ(NumTableFilesAtLevel(0), 10);

db/db_range_del_test.cc

@@ -56,7 +56,7 @@ TEST_F(DBRangeDelTest, EndSameAsStartCoversNothing) {
 }
 
 TEST_F(DBRangeDelTest, EndComesBeforeStartInvalidArgument) {
-  db_->Put(WriteOptions(), "b", "val");
+  ASSERT_OK(db_->Put(WriteOptions(), "b", "val"));
   ASSERT_TRUE(
       db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(), "b", "a")
           .IsInvalidArgument());
@@ -82,13 +82,14 @@ TEST_F(DBRangeDelTest, CompactionOutputHasOnlyRangeTombstone) {
 
     // snapshot protects range tombstone from dropping due to becoming obsolete.
     const Snapshot* snapshot = db_->GetSnapshot();
-    db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(), "a", "z");
-    db_->Flush(FlushOptions());
+    ASSERT_OK(
+        db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(), "a", "z"));
+    ASSERT_OK(db_->Flush(FlushOptions()));
 
     ASSERT_EQ(1, NumTableFilesAtLevel(0));
     ASSERT_EQ(0, NumTableFilesAtLevel(1));
-    dbfull()->TEST_CompactRange(0, nullptr, nullptr, nullptr,
-                                true /* disallow_trivial_move */);
+    ASSERT_OK(dbfull()->TEST_CompactRange(0, nullptr, nullptr, nullptr,
+                                          true /* disallow_trivial_move */));
     ASSERT_EQ(0, NumTableFilesAtLevel(0));
     ASSERT_EQ(1, NumTableFilesAtLevel(1));
     ASSERT_EQ(0, TestGetTickerCount(opts, COMPACTION_RANGE_DEL_DROP_OBSOLETE));
@@ -118,7 +119,8 @@ TEST_F(DBRangeDelTest, CompactionOutputFilesExactlyFilled) {
 
   // snapshot protects range tombstone from dropping due to becoming obsolete.
   const Snapshot* snapshot = db_->GetSnapshot();
-  db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(), Key(0), Key(1));
+  ASSERT_OK(db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(), Key(0),
+                             Key(1)));
 
   Random rnd(301);
   for (int i = 0; i < kNumFiles; ++i) {
@@ -128,18 +130,18 @@ TEST_F(DBRangeDelTest, CompactionOutputFilesExactlyFilled) {
       values.push_back(rnd.RandomString(3 << 10));
       ASSERT_OK(Put(Key(i * kNumPerFile + j), values[j]));
       if (j == 0 && i > 0) {
-        dbfull()->TEST_WaitForFlushMemTable();
+        ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable());
       }
     }
   }
   // put extra key to trigger final flush
   ASSERT_OK(Put("", ""));
-  dbfull()->TEST_WaitForFlushMemTable();
+  ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable());
   ASSERT_EQ(kNumFiles, NumTableFilesAtLevel(0));
   ASSERT_EQ(0, NumTableFilesAtLevel(1));
 
-  dbfull()->TEST_CompactRange(0, nullptr, nullptr, nullptr,
-                              true /* disallow_trivial_move */);
+  ASSERT_OK(dbfull()->TEST_CompactRange(0, nullptr, nullptr, nullptr,
+                                        true /* disallow_trivial_move */));
   ASSERT_EQ(0, NumTableFilesAtLevel(0));
   ASSERT_EQ(2, NumTableFilesAtLevel(1));
   db_->ReleaseSnapshot(snapshot);
@@ -178,12 +180,12 @@ TEST_F(DBRangeDelTest, MaxCompactionBytesCutsOutputFiles) {
     }
     // extra entry to trigger SpecialSkipListFactory's flush
     ASSERT_OK(Put(GetNumericStr(kNumPerFile), ""));
-    dbfull()->TEST_WaitForFlushMemTable();
+    ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable());
     ASSERT_EQ(i + 1, NumTableFilesAtLevel(0));
   }
 
-  dbfull()->TEST_CompactRange(0, nullptr, nullptr, nullptr,
-                              true /* disallow_trivial_move */);
+  ASSERT_OK(dbfull()->TEST_CompactRange(0, nullptr, nullptr, nullptr,
+                                        true /* disallow_trivial_move */));
   ASSERT_EQ(0, NumTableFilesAtLevel(0));
   ASSERT_GE(NumTableFilesAtLevel(1), 2);
 
@@ -221,10 +223,10 @@ TEST_F(DBRangeDelTest, SentinelsOmittedFromOutputFile) {
 }
 
 TEST_F(DBRangeDelTest, FlushRangeDelsSameStartKey) {
-  db_->Put(WriteOptions(), "b1", "val");
+  ASSERT_OK(db_->Put(WriteOptions(), "b1", "val"));
   ASSERT_OK(
       db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(), "a", "c"));
-  db_->Put(WriteOptions(), "b2", "val");
+  ASSERT_OK(db_->Put(WriteOptions(), "b2", "val"));
   ASSERT_OK(
       db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(), "a", "b"));
   // first iteration verifies query correctness in memtable, second verifies
@@ -241,8 +243,9 @@ TEST_F(DBRangeDelTest, FlushRangeDelsSameStartKey) {
 }
 
 TEST_F(DBRangeDelTest, CompactRangeDelsSameStartKey) {
-  db_->Put(WriteOptions(), "unused", "val");  // prevents empty after compaction
-  db_->Put(WriteOptions(), "b1", "val");
+  ASSERT_OK(db_->Put(WriteOptions(), "unused",
+                     "val"));  // prevents empty after compaction
+  ASSERT_OK(db_->Put(WriteOptions(), "b1", "val"));
   ASSERT_OK(db_->Flush(FlushOptions()));
   ASSERT_OK(
       db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(), "a", "c"));
@@ -254,8 +257,8 @@ TEST_F(DBRangeDelTest, CompactRangeDelsSameStartKey) {
 
   for (int i = 0; i < 2; ++i) {
     if (i > 0) {
-      dbfull()->TEST_CompactRange(0, nullptr, nullptr, nullptr,
-                                  true /* disallow_trivial_move */);
+      ASSERT_OK(dbfull()->TEST_CompactRange(0, nullptr, nullptr, nullptr,
+                                            true /* disallow_trivial_move */));
       ASSERT_EQ(0, NumTableFilesAtLevel(0));
       ASSERT_EQ(1, NumTableFilesAtLevel(1));
     }
@@ -279,12 +282,13 @@ TEST_F(DBRangeDelTest, FlushRemovesCoveredKeys) {
     if (i == kNum / 3) {
       snapshot = db_->GetSnapshot();
     } else if (i == 2 * kNum / 3) {
-      db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(),
-                       GetNumericStr(kRangeBegin), GetNumericStr(kRangeEnd));
+      ASSERT_OK(db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(),
+                                 GetNumericStr(kRangeBegin),
+                                 GetNumericStr(kRangeEnd)));
     }
-    db_->Put(WriteOptions(), GetNumericStr(i), "val");
+    ASSERT_OK(db_->Put(WriteOptions(), GetNumericStr(i), "val"));
   }
-  db_->Flush(FlushOptions());
+  ASSERT_OK(db_->Flush(FlushOptions()));
 
   for (int i = 0; i < kNum; ++i) {
     ReadOptions read_opts;
@@ -314,24 +318,27 @@ TEST_F(DBRangeDelTest, CompactionRemovesCoveredKeys) {
   for (int i = 0; i < kNumFiles; ++i) {
     if (i > 0) {
       // range tombstone covers first half of the previous file
-      db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(),
-                       GetNumericStr((i - 1) * kNumPerFile),
-                       GetNumericStr((i - 1) * kNumPerFile + kNumPerFile / 2));
+      ASSERT_OK(db_->DeleteRange(
+          WriteOptions(), db_->DefaultColumnFamily(),
+          GetNumericStr((i - 1) * kNumPerFile),
+          GetNumericStr((i - 1) * kNumPerFile + kNumPerFile / 2)));
     }
     // Make sure a given key appears in each file so compaction won't be able to
     // use trivial move, which would happen if the ranges were non-overlapping.
     // Also, we need an extra element since flush is only triggered when the
     // number of keys is one greater than SpecialSkipListFactory's limit.
     // We choose a key outside the key-range used by the test to avoid conflict.
-    db_->Put(WriteOptions(), GetNumericStr(kNumPerFile * kNumFiles), "val");
+    ASSERT_OK(db_->Put(WriteOptions(), GetNumericStr(kNumPerFile * kNumFiles),
+                       "val"));
 
     for (int j = 0; j < kNumPerFile; ++j) {
-      db_->Put(WriteOptions(), GetNumericStr(i * kNumPerFile + j), "val");
+      ASSERT_OK(
+          db_->Put(WriteOptions(), GetNumericStr(i * kNumPerFile + j), "val"));
     }
-    dbfull()->TEST_WaitForFlushMemTable();
+    ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable());
     ASSERT_EQ(i + 1, NumTableFilesAtLevel(0));
   }
-  db_->CompactRange(CompactRangeOptions(), nullptr, nullptr);
+  ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
   ASSERT_EQ(0, NumTableFilesAtLevel(0));
   ASSERT_GT(NumTableFilesAtLevel(1), 0);
   ASSERT_EQ((kNumFiles - 1) * kNumPerFile / 2,
@@ -373,8 +380,8 @@ TEST_F(DBRangeDelTest, ValidLevelSubcompactionBoundaries) {
       if (i > 0) {
         // delete [95,105) in two files, [295,305) in next two
         int mid = (j + (1 - j % 2)) * kNumPerFile;
-        db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(),
-                         Key(mid - 5), Key(mid + 5));
+        ASSERT_OK(db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(),
+                                   Key(mid - 5), Key(mid + 5)));
       }
       std::vector<std::string> values;
       // Write 100KB (100 values, each 1K)
@@ -384,7 +391,7 @@ TEST_F(DBRangeDelTest, ValidLevelSubcompactionBoundaries) {
       }
       // put extra key to trigger flush
       ASSERT_OK(Put("", ""));
-      dbfull()->TEST_WaitForFlushMemTable();
+      ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable());
       if (j < kNumFiles - 1) {
         // background compaction may happen early for kNumFiles'th file
         ASSERT_EQ(NumTableFilesAtLevel(0), j + 1);
@@ -400,7 +407,7 @@ TEST_F(DBRangeDelTest, ValidLevelSubcompactionBoundaries) {
         // oversized L0 (relative to base_level) causes the compaction to run
         // earlier.
         ASSERT_OK(db_->EnableAutoCompaction({db_->DefaultColumnFamily()}));
-        dbfull()->TEST_WaitForCompact();
+        ASSERT_OK(dbfull()->TEST_WaitForCompact());
         ASSERT_OK(db_->SetOptions(db_->DefaultColumnFamily(),
                                   {{"disable_auto_compactions", "true"}}));
         ASSERT_EQ(NumTableFilesAtLevel(0), 0);
@@ -433,8 +440,8 @@ TEST_F(DBRangeDelTest, ValidUniversalSubcompactionBoundaries) {
         // insert range deletions [95,105) in two files, [295,305) in next two
         // to prepare L1 for later manual compaction.
         int mid = (j + (1 - j % 2)) * kNumPerFile;
-        db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(),
-                         Key(mid - 5), Key(mid + 5));
+        ASSERT_OK(db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(),
+                                   Key(mid - 5), Key(mid + 5)));
       }
       std::vector<std::string> values;
       // Write 100KB (100 values, each 1K)
@@ -444,13 +451,13 @@ TEST_F(DBRangeDelTest, ValidUniversalSubcompactionBoundaries) {
       }
       // put extra key to trigger flush
       ASSERT_OK(Put("", ""));
-      dbfull()->TEST_WaitForFlushMemTable();
+      ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable());
       if (j < kFilesPerLevel - 1) {
         // background compaction may happen early for kFilesPerLevel'th file
         ASSERT_EQ(NumTableFilesAtLevel(0), j + 1);
       }
     }
-    dbfull()->TEST_WaitForCompact();
+    ASSERT_OK(dbfull()->TEST_WaitForCompact());
     ASSERT_EQ(NumTableFilesAtLevel(0), 0);
     ASSERT_GT(NumTableFilesAtLevel(kNumLevels - 1 - i), kFilesPerLevel - 1);
   }
@@ -483,17 +490,17 @@ TEST_F(DBRangeDelTest, CompactionRemovesCoveredMergeOperands) {
   for (int i = 0; i <= kNumFiles * kNumPerFile; ++i) {
     if (i % kNumPerFile == 0 && i / kNumPerFile == kNumFiles - 1) {
       // Delete merge operands from all but the last file
-      db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(), "key",
-                       "key_");
+      ASSERT_OK(db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(),
+                                 "key", "key_"));
     }
     std::string val;
     PutFixed64(&val, i);
-    db_->Merge(WriteOptions(), "key", val);
+    ASSERT_OK(db_->Merge(WriteOptions(), "key", val));
     // we need to prevent trivial move using Puts so compaction will actually
     // process the merge operands.
-    db_->Put(WriteOptions(), "prevent_trivial_move", "");
+    ASSERT_OK(db_->Put(WriteOptions(), "prevent_trivial_move", ""));
     if (i > 0 && i % kNumPerFile == 0) {
-      dbfull()->TEST_WaitForFlushMemTable();
+      ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable());
     }
   }
 
@@ -504,7 +511,7 @@ TEST_F(DBRangeDelTest, CompactionRemovesCoveredMergeOperands) {
   PutFixed64(&expected, 45);  // 1+2+...+9
   ASSERT_EQ(expected, actual);
 
-  db_->CompactRange(CompactRangeOptions(), nullptr, nullptr);
+  ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
 
   expected.clear();
   ASSERT_OK(db_->Get(read_opts, "key", &actual));
@@ -550,19 +557,19 @@ TEST_F(DBRangeDelTest, ObsoleteTombstoneCleanup) {
   opts.statistics = CreateDBStatistics();
   Reopen(opts);
 
-  db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(), "dr1",
-                   "dr10");  // obsolete after compaction
-  db_->Put(WriteOptions(), "key", "val");
-  db_->Flush(FlushOptions());
+  ASSERT_OK(db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(), "dr1",
+                             "dr10"));  // obsolete after compaction
+  ASSERT_OK(db_->Put(WriteOptions(), "key", "val"));
+  ASSERT_OK(db_->Flush(FlushOptions()));
   const Snapshot* snapshot = db_->GetSnapshot();
-  db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(), "dr2",
-                   "dr20");  // protected by snapshot
-  db_->Put(WriteOptions(), "key", "val");
-  db_->Flush(FlushOptions());
+  ASSERT_OK(db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(), "dr2",
+                             "dr20"));  // protected by snapshot
+  ASSERT_OK(db_->Put(WriteOptions(), "key", "val"));
+  ASSERT_OK(db_->Flush(FlushOptions()));
 
   ASSERT_EQ(2, NumTableFilesAtLevel(0));
   ASSERT_EQ(0, NumTableFilesAtLevel(1));
-  db_->CompactRange(CompactRangeOptions(), nullptr, nullptr);
+  ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
   ASSERT_EQ(0, NumTableFilesAtLevel(0));
   ASSERT_EQ(1, NumTableFilesAtLevel(1));
   ASSERT_EQ(1, TestGetTickerCount(opts, COMPACTION_RANGE_DEL_DROP_OBSOLETE));
@@ -609,22 +616,24 @@ TEST_F(DBRangeDelTest, TableEvictedDuringScan) {
   // to bottommost level (i.e., L1).
   const Snapshot* snapshot = db_->GetSnapshot();
   for (int i = 0; i < kNum; ++i) {
-    db_->Put(WriteOptions(), GetNumericStr(i), "val");
+    ASSERT_OK(db_->Put(WriteOptions(), GetNumericStr(i), "val"));
     if (i > 0) {
-      dbfull()->TEST_WaitForFlushMemTable();
+      ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable());
     }
     if (i >= kNum / 2 && i < kNum / 2 + kNumRanges) {
-      db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(),
-                       GetNumericStr(kRangeBegin), GetNumericStr(kRangeEnd));
+      ASSERT_OK(db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(),
+                                 GetNumericStr(kRangeBegin),
+                                 GetNumericStr(kRangeEnd)));
     }
   }
   // Must be > 1 so the first L1 file can be closed before scan finishes
-  dbfull()->TEST_WaitForCompact();
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
   ASSERT_GT(NumTableFilesAtLevel(1), 1);
   std::vector<uint64_t> file_numbers = ListTableFiles(env_, dbname_);
 
   ReadOptions read_opts;
   auto* iter = db_->NewIterator(read_opts);
+  ASSERT_OK(iter->status());
   int expected = kRangeEnd;
   iter->SeekToFirst();
   for (auto file_number : file_numbers) {
@@ -647,7 +656,7 @@ TEST_F(DBRangeDelTest, TableEvictedDuringScan) {
 TEST_F(DBRangeDelTest, GetCoveredKeyFromMutableMemtable) {
   do {
     DestroyAndReopen(CurrentOptions());
-    db_->Put(WriteOptions(), "key", "val");
+    ASSERT_OK(db_->Put(WriteOptions(), "key", "val"));
     ASSERT_OK(
         db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(), "a", "z"));
 
@@ -669,10 +678,10 @@ TEST_F(DBRangeDelTest, GetCoveredKeyFromImmutableMemtable) {
     opts.memtable_factory.reset(new SpecialSkipListFactory(1));
     DestroyAndReopen(opts);
 
-    db_->Put(WriteOptions(), "key", "val");
+    ASSERT_OK(db_->Put(WriteOptions(), "key", "val"));
     ASSERT_OK(
         db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(), "a", "z"));
-    db_->Put(WriteOptions(), "blah", "val");
+    ASSERT_OK(db_->Put(WriteOptions(), "blah", "val"));
 
     ReadOptions read_opts;
     std::string value;
@@ -683,7 +692,7 @@ TEST_F(DBRangeDelTest, GetCoveredKeyFromImmutableMemtable) {
 TEST_F(DBRangeDelTest, GetCoveredKeyFromSst) {
   do {
     DestroyAndReopen(CurrentOptions());
-    db_->Put(WriteOptions(), "key", "val");
+    ASSERT_OK(db_->Put(WriteOptions(), "key", "val"));
     // snapshot prevents key from being deleted during flush
     const Snapshot* snapshot = db_->GetSnapshot();
     ASSERT_OK(
@@ -706,11 +715,11 @@ TEST_F(DBRangeDelTest, GetCoveredMergeOperandFromMemtable) {
   for (int i = 0; i < kNumMergeOps; ++i) {
     std::string val;
     PutFixed64(&val, i);
-    db_->Merge(WriteOptions(), "key", val);
+    ASSERT_OK(db_->Merge(WriteOptions(), "key", val));
     if (i == kNumMergeOps / 2) {
       // deletes [0, 5]
-      db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(), "key",
-                       "key_");
+      ASSERT_OK(db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(),
+                                 "key", "key_"));
     }
   }
 
@@ -734,16 +743,16 @@ TEST_F(DBRangeDelTest, GetIgnoresRangeDeletions) {
   opts.memtable_factory.reset(new SpecialSkipListFactory(1));
   Reopen(opts);
 
-  db_->Put(WriteOptions(), "sst_key", "val");
+  ASSERT_OK(db_->Put(WriteOptions(), "sst_key", "val"));
   // snapshot prevents key from being deleted during flush
   const Snapshot* snapshot = db_->GetSnapshot();
   ASSERT_OK(
       db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(), "a", "z"));
   ASSERT_OK(db_->Flush(FlushOptions()));
-  db_->Put(WriteOptions(), "imm_key", "val");
+  ASSERT_OK(db_->Put(WriteOptions(), "imm_key", "val"));
   ASSERT_OK(
       db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(), "a", "z"));
-  db_->Put(WriteOptions(), "mem_key", "val");
+  ASSERT_OK(db_->Put(WriteOptions(), "mem_key", "val"));
   ASSERT_OK(
       db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(), "a", "z"));
 
@@ -768,13 +777,15 @@ TEST_F(DBRangeDelTest, IteratorRemovesCoveredKeys) {
   // should be deleted.
   for (int i = 0; i < kNum; ++i) {
     if (i == kNum / 2) {
-      db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(),
-                       GetNumericStr(kRangeBegin), GetNumericStr(kRangeEnd));
+      ASSERT_OK(db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(),
+                                 GetNumericStr(kRangeBegin),
+                                 GetNumericStr(kRangeEnd)));
     }
-    db_->Put(WriteOptions(), GetNumericStr(i), "val");
+    ASSERT_OK(db_->Put(WriteOptions(), GetNumericStr(i), "val"));
   }
   ReadOptions read_opts;
   auto* iter = db_->NewIterator(read_opts);
+  ASSERT_OK(iter->status());
 
   int expected = 0;
   for (iter->SeekToFirst(); iter->Valid(); iter->Next()) {
@@ -802,14 +813,16 @@ TEST_F(DBRangeDelTest, IteratorOverUserSnapshot) {
   for (int i = 0; i < kNum; ++i) {
     if (i == kNum / 2) {
       snapshot = db_->GetSnapshot();
-      db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(),
-                       GetNumericStr(kRangeBegin), GetNumericStr(kRangeEnd));
+      ASSERT_OK(db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(),
+                                 GetNumericStr(kRangeBegin),
+                                 GetNumericStr(kRangeEnd)));
     }
-    db_->Put(WriteOptions(), GetNumericStr(i), "val");
+    ASSERT_OK(db_->Put(WriteOptions(), GetNumericStr(i), "val"));
   }
   ReadOptions read_opts;
   read_opts.snapshot = snapshot;
   auto* iter = db_->NewIterator(read_opts);
+  ASSERT_OK(iter->status());
 
   int expected = 0;
   for (iter->SeekToFirst(); iter->Valid(); iter->Next()) {
@@ -828,22 +841,23 @@ TEST_F(DBRangeDelTest, IteratorIgnoresRangeDeletions) {
   opts.memtable_factory.reset(new SpecialSkipListFactory(1));
   Reopen(opts);
 
-  db_->Put(WriteOptions(), "sst_key", "val");
+  ASSERT_OK(db_->Put(WriteOptions(), "sst_key", "val"));
   // snapshot prevents key from being deleted during flush
   const Snapshot* snapshot = db_->GetSnapshot();
   ASSERT_OK(
       db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(), "a", "z"));
   ASSERT_OK(db_->Flush(FlushOptions()));
-  db_->Put(WriteOptions(), "imm_key", "val");
+  ASSERT_OK(db_->Put(WriteOptions(), "imm_key", "val"));
   ASSERT_OK(
       db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(), "a", "z"));
-  db_->Put(WriteOptions(), "mem_key", "val");
+  ASSERT_OK(db_->Put(WriteOptions(), "mem_key", "val"));
   ASSERT_OK(
       db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(), "a", "z"));
 
   ReadOptions read_opts;
   read_opts.ignore_range_deletions = true;
   auto* iter = db_->NewIterator(read_opts);
+  ASSERT_OK(iter->status());
   int i = 0;
   std::string expected[] = {"imm_key", "mem_key", "sst_key"};
   for (iter->SeekToFirst(); iter->Valid(); iter->Next(), ++i) {
@@ -857,7 +871,7 @@ TEST_F(DBRangeDelTest, IteratorIgnoresRangeDeletions) {
 
 #ifndef ROCKSDB_UBSAN_RUN
 TEST_F(DBRangeDelTest, TailingIteratorRangeTombstoneUnsupported) {
-  db_->Put(WriteOptions(), "key", "val");
+  ASSERT_OK(db_->Put(WriteOptions(), "key", "val"));
   // snapshot prevents key from being deleted during flush
   const Snapshot* snapshot = db_->GetSnapshot();
   ASSERT_OK(
@@ -873,6 +887,7 @@ TEST_F(DBRangeDelTest, TailingIteratorRangeTombstoneUnsupported) {
       iter->SeekToFirst();
     }
     ASSERT_TRUE(iter->status().IsNotSupported());
+
     delete iter;
     if (i == 0) {
       ASSERT_OK(db_->Flush(FlushOptions()));
@@ -882,7 +897,6 @@ TEST_F(DBRangeDelTest, TailingIteratorRangeTombstoneUnsupported) {
   }
   db_->ReleaseSnapshot(snapshot);
 }
-
 #endif  // !ROCKSDB_UBSAN_RUN
 
 TEST_F(DBRangeDelTest, SubcompactionHasEmptyDedicatedRangeDelFile) {
@@ -926,8 +940,8 @@ TEST_F(DBRangeDelTest, SubcompactionHasEmptyDedicatedRangeDelFile) {
   ASSERT_EQ(kNumFiles, NumTableFilesAtLevel(0));
   ASSERT_EQ(1, NumTableFilesAtLevel(1));
 
-  db_->EnableAutoCompaction({db_->DefaultColumnFamily()});
-  dbfull()->TEST_WaitForCompact();
+  ASSERT_OK(db_->EnableAutoCompaction({db_->DefaultColumnFamily()}));
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
   db_->ReleaseSnapshot(snapshot);
 }
 
@@ -949,7 +963,7 @@ TEST_F(DBRangeDelTest, MemtableBloomFilter) {
   for (int i = 0; i < kNumKeys; ++i) {
     ASSERT_OK(Put(Key(i), "val"));
   }
-  Flush();
+  ASSERT_OK(Flush());
   ASSERT_OK(db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(), Key(0),
                              Key(kNumKeys)));
   for (int i = 0; i < kNumKeys; ++i) {
@@ -987,8 +1001,8 @@ TEST_F(DBRangeDelTest, CompactionTreatsSplitInputLevelDeletionAtomically) {
 
     // snapshot protects range tombstone from dropping due to becoming obsolete.
     const Snapshot* snapshot = db_->GetSnapshot();
-    db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(), Key(0),
-                     Key(2 * kNumFilesPerLevel));
+    ASSERT_OK(db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(),
+                               Key(0), Key(2 * kNumFilesPerLevel)));
 
     Random rnd(301);
     std::string value = rnd.RandomString(kValueBytes);
@@ -997,14 +1011,14 @@ TEST_F(DBRangeDelTest, CompactionTreatsSplitInputLevelDeletionAtomically) {
       ASSERT_OK(Put(Key(j), value));
       ASSERT_OK(Put(Key(2 * kNumFilesPerLevel - 1 - j), value));
       if (j > 0) {
-        dbfull()->TEST_WaitForFlushMemTable();
+        ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable());
         ASSERT_EQ(j, NumTableFilesAtLevel(0));
       }
     }
     // put extra key to trigger final flush
     ASSERT_OK(Put("", ""));
-    dbfull()->TEST_WaitForFlushMemTable();
-    dbfull()->TEST_WaitForCompact();
+    ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable());
+    ASSERT_OK(dbfull()->TEST_WaitForCompact());
     ASSERT_EQ(0, NumTableFilesAtLevel(0));
     ASSERT_EQ(kNumFilesPerLevel, NumTableFilesAtLevel(1));
 
@@ -1022,7 +1036,7 @@ TEST_F(DBRangeDelTest, CompactionTreatsSplitInputLevelDeletionAtomically) {
     } else if (i == 2) {
       ASSERT_OK(db_->SetOptions(db_->DefaultColumnFamily(),
                                 {{"max_bytes_for_level_base", "10000"}}));
-      dbfull()->TEST_WaitForCompact();
+      ASSERT_OK(dbfull()->TEST_WaitForCompact());
       ASSERT_EQ(1, NumTableFilesAtLevel(1));
     }
     ASSERT_GT(NumTableFilesAtLevel(2), 0);
@@ -1056,8 +1070,8 @@ TEST_F(DBRangeDelTest, RangeTombstoneEndKeyAsSstableUpperBound) {
   // A snapshot protects the range tombstone from dropping due to
   // becoming obsolete.
   const Snapshot* snapshot = db_->GetSnapshot();
-  db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(),
-                   Key(0), Key(2 * kNumFilesPerLevel));
+  ASSERT_OK(db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(), Key(0),
+                             Key(2 * kNumFilesPerLevel)));
 
   // Create 2 additional sstables in L0. Note that the first sstable
   // contains the range tombstone.
@@ -1096,7 +1110,7 @@ TEST_F(DBRangeDelTest, RangeTombstoneEndKeyAsSstableUpperBound) {
     ASSERT_EQ(value, Get(Key(2)));
     auto begin_str = Key(3);
     const ROCKSDB_NAMESPACE::Slice begin = begin_str;
-    dbfull()->TEST_CompactRange(1, &begin, nullptr);
+    ASSERT_OK(dbfull()->TEST_CompactRange(1, &begin, nullptr));
     ASSERT_EQ(1, NumTableFilesAtLevel(1));
     ASSERT_EQ(2, NumTableFilesAtLevel(2));
     ASSERT_EQ(value, Get(Key(2)));
@@ -1115,7 +1129,7 @@ TEST_F(DBRangeDelTest, RangeTombstoneEndKeyAsSstableUpperBound) {
     //     [key000002#6,1, key000004#72057594037927935,15]
     auto begin_str = Key(0);
     const ROCKSDB_NAMESPACE::Slice begin = begin_str;
-    dbfull()->TEST_CompactRange(1, &begin, &begin);
+    ASSERT_OK(dbfull()->TEST_CompactRange(1, &begin, &begin));
     ASSERT_EQ(0, NumTableFilesAtLevel(1));
     ASSERT_EQ(3, NumTableFilesAtLevel(2));
   }
@@ -1216,9 +1230,9 @@ TEST_F(DBRangeDelTest, KeyAtOverlappingEndpointReappears) {
   std::string value;
   ASSERT_TRUE(db_->Get(ReadOptions(), "key", &value).IsNotFound());
 
-  dbfull()->TEST_CompactRange(0 /* level */, nullptr /* begin */,
-                              nullptr /* end */, nullptr /* column_family */,
-                              true /* disallow_trivial_move */);
+  ASSERT_OK(dbfull()->TEST_CompactRange(
+      0 /* level */, nullptr /* begin */, nullptr /* end */,
+      nullptr /* column_family */, true /* disallow_trivial_move */));
   ASSERT_EQ(0, NumTableFilesAtLevel(0));
   // Now we have multiple files at L1 all containing a single user key, thus
   // guaranteeing overlap in the file endpoints.
@@ -1229,9 +1243,9 @@ TEST_F(DBRangeDelTest, KeyAtOverlappingEndpointReappears) {
 
   // Compact and verify again. It's worthwhile because now the files have
   // tighter endpoints, so we can verify that doesn't mess anything up.
-  dbfull()->TEST_CompactRange(1 /* level */, nullptr /* begin */,
-                              nullptr /* end */, nullptr /* column_family */,
-                              true /* disallow_trivial_move */);
+  ASSERT_OK(dbfull()->TEST_CompactRange(
+      1 /* level */, nullptr /* begin */, nullptr /* end */,
+      nullptr /* column_family */, true /* disallow_trivial_move */));
   ASSERT_GT(NumTableFilesAtLevel(2), 1);
   ASSERT_TRUE(db_->Get(ReadOptions(), "key", &value).IsNotFound());
 
@@ -1307,6 +1321,7 @@ TEST_F(DBRangeDelTest, UntruncatedTombstoneDoesNotDeleteNewerKey) {
 
   auto get_key_count = [this]() -> int {
     auto* iter = db_->NewIterator(ReadOptions());
+    assert(iter->status().ok());
     iter->SeekToFirst();
     int keys_found = 0;
     for (; iter->Valid(); iter->Next()) {
@@ -1409,6 +1424,7 @@ TEST_F(DBRangeDelTest, DeletedMergeOperandReappearsIterPrev) {
   ASSERT_GT(NumTableFilesAtLevel(1), 1);
 
   auto* iter = db_->NewIterator(ReadOptions());
+  ASSERT_OK(iter->status());
   iter->SeekToLast();
   int keys_found = 0;
   for (; iter->Valid(); iter->Prev()) {
@@ -1435,11 +1451,12 @@ TEST_F(DBRangeDelTest, SnapshotPreventsDroppedKeys) {
   ASSERT_OK(db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(), Key(0),
                              Key(10)));
 
-  db_->Flush(FlushOptions());
+  ASSERT_OK(db_->Flush(FlushOptions()));
 
   ReadOptions read_opts;
   read_opts.snapshot = snapshot;
   auto* iter = db_->NewIterator(read_opts);
+  ASSERT_OK(iter->status());
 
   iter->SeekToFirst();
   ASSERT_TRUE(iter->Valid());
@@ -1482,6 +1499,7 @@ TEST_F(DBRangeDelTest, SnapshotPreventsDroppedKeysInImmMemTables) {
   ReadOptions read_opts;
   read_opts.snapshot = snapshot.get();
   std::unique_ptr<Iterator> iter(db_->NewIterator(read_opts));
+  ASSERT_OK(iter->status());
 
   TEST_SYNC_POINT("SnapshotPreventsDroppedKeysInImmMemTables:AfterNewIterator");
 
@@ -1519,7 +1537,7 @@ TEST_F(DBRangeDelTest, RangeTombstoneWrittenToMinimalSsts) {
       std::string value = rnd.RandomString(kValueBytes);
       ASSERT_OK(Put(key, value));
     }
-    db_->Flush(FlushOptions());
+    ASSERT_OK(db_->Flush(FlushOptions()));
     MoveFilesToLevel(2);
   }
   ASSERT_EQ(0, NumTableFilesAtLevel(0));
@@ -1538,7 +1556,7 @@ TEST_F(DBRangeDelTest, RangeTombstoneWrittenToMinimalSsts) {
   // TODO(ajkr): remove this `Put` after file cutting accounts for range
   // tombstones (#3977).
   ASSERT_OK(Put("c" + Key(1), "value"));
-  db_->Flush(FlushOptions());
+  ASSERT_OK(db_->Flush(FlushOptions()));
 
   // Ensure manual L0->L1 compaction cuts the outputs before the range tombstone
   // and the range tombstone is only placed in the second SST.
@@ -1546,9 +1564,9 @@ TEST_F(DBRangeDelTest, RangeTombstoneWrittenToMinimalSsts) {
   Slice begin_key(begin_key_storage);
   std::string end_key_storage("d");
   Slice end_key(end_key_storage);
-  dbfull()->TEST_CompactRange(0 /* level */, &begin_key /* begin */,
-                              &end_key /* end */, nullptr /* column_family */,
-                              true /* disallow_trivial_move */);
+  ASSERT_OK(dbfull()->TEST_CompactRange(
+      0 /* level */, &begin_key /* begin */, &end_key /* end */,
+      nullptr /* column_family */, true /* disallow_trivial_move */));
   ASSERT_EQ(2, NumTableFilesAtLevel(1));
 
   std::vector<LiveFileMetaData> all_metadata;
@@ -1613,15 +1631,15 @@ TEST_F(DBRangeDelTest, OverlappedTombstones) {
 
   ASSERT_EQ(1, NumTableFilesAtLevel(0));
 
-  dbfull()->TEST_CompactRange(0, nullptr, nullptr, nullptr,
-                              true /* disallow_trivial_move */);
+  ASSERT_OK(dbfull()->TEST_CompactRange(0, nullptr, nullptr, nullptr,
+                                        true /* disallow_trivial_move */));
 
   // The tombstone range is not broken up into multiple SSTs which may incur a
   // large compaction with L2.
   ASSERT_EQ(1, NumTableFilesAtLevel(1));
   std::vector<std::vector<FileMetaData>> files;
-  dbfull()->TEST_CompactRange(1, nullptr, nullptr, nullptr,
-                              true /* disallow_trivial_move */);
+  ASSERT_OK(dbfull()->TEST_CompactRange(1, nullptr, nullptr, nullptr,
+                                        true /* disallow_trivial_move */));
   ASSERT_EQ(1, NumTableFilesAtLevel(2));
   ASSERT_EQ(0, NumTableFilesAtLevel(1));
 }
@@ -1654,13 +1672,13 @@ TEST_F(DBRangeDelTest, OverlappedKeys) {
 
   // The key range is broken up into three SSTs to avoid a future big compaction
   // with the grandparent
-  dbfull()->TEST_CompactRange(0, nullptr, nullptr, nullptr,
-                              true /* disallow_trivial_move */);
+  ASSERT_OK(dbfull()->TEST_CompactRange(0, nullptr, nullptr, nullptr,
+                                        true /* disallow_trivial_move */));
   ASSERT_EQ(3, NumTableFilesAtLevel(1));
 
   std::vector<std::vector<FileMetaData>> files;
-  dbfull()->TEST_CompactRange(1, nullptr, nullptr, nullptr,
-                              true /* disallow_trivial_move */);
+  ASSERT_OK(dbfull()->TEST_CompactRange(1, nullptr, nullptr, nullptr,
+                                        true /* disallow_trivial_move */));
   ASSERT_EQ(1, NumTableFilesAtLevel(2));
   ASSERT_EQ(0, NumTableFilesAtLevel(1));
 }

db/db_sst_test.cc

@@ -98,7 +98,7 @@ TEST_F(DBSSTTest, SSTsWithLdbSuffixHandling) {
   for (int i = 0; i < 10; ++i) {
     GenerateNewFile(&rnd, &key_id, false);
   }
-  Flush();
+  ASSERT_OK(Flush());
   Close();
   int const num_files = GetSstFileCount(dbname_);
   ASSERT_GT(num_files, 0);
@@ -170,7 +170,7 @@ TEST_F(DBSSTTest, DontDeleteMovedFile) {
     ASSERT_OK(Flush());
   }
   // this should execute both L0->L1 and L1->(move)->L2 compactions
-  dbfull()->TEST_WaitForCompact();
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
   ASSERT_EQ("0,0,1", FilesPerLevel(0));
 
   // If the moved file is actually deleted (the move-safeguard in
@@ -218,7 +218,7 @@ TEST_F(DBSSTTest, DeleteObsoleteFilesPendingOutputs) {
     ASSERT_OK(Flush());
   }
   // this should execute both L0->L1 and L1->(move)->L2 compactions
-  dbfull()->TEST_WaitForCompact();
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
   ASSERT_EQ("0,0,1", FilesPerLevel(0));
 
   test::SleepingBackgroundTask blocking_thread;
@@ -264,9 +264,9 @@ TEST_F(DBSSTTest, DeleteObsoleteFilesPendingOutputs) {
   // finish the flush!
   blocking_thread.WakeUp();
   blocking_thread.WaitUntilDone();
-  dbfull()->TEST_WaitForFlushMemTable();
+  ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable());
   // File just flushed is too big for L0 and L1 so gets moved to L2.
-  dbfull()->TEST_WaitForCompact();
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
   ASSERT_EQ("0,0,1,0,1", FilesPerLevel(0));
 
   metadata.clear();
@@ -302,8 +302,8 @@ TEST_F(DBSSTTest, DBWithSstFileManager) {
   for (int i = 0; i < 25; i++) {
     GenerateNewRandomFile(&rnd);
     ASSERT_OK(Flush());
-    dbfull()->TEST_WaitForFlushMemTable();
-    dbfull()->TEST_WaitForCompact();
+    ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable());
+    ASSERT_OK(dbfull()->TEST_WaitForCompact());
     // Verify that we are tracking all sst files in dbname_
     std::unordered_map<std::string, uint64_t> files_in_db;
     ASSERT_OK(GetAllSSTFiles(&files_in_db));
@@ -374,6 +374,14 @@ TEST_F(DBSSTTest, RateLimitedDelete) {
         *abs_time_us = Env::Default()->NowMicros();
       });
 
+  // Disable PeriodicWorkScheduler as it also has TimedWait, which could update
+  // the simulated sleep time
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
+      "DBImpl::StartPeriodicWorkScheduler:DisableScheduler", [&](void* arg) {
+        bool* disable_scheduler = static_cast<bool*>(arg);
+        *disable_scheduler = true;
+      });
+
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
 
   Options options = CurrentOptions();
@@ -393,7 +401,7 @@ TEST_F(DBSSTTest, RateLimitedDelete) {
 
   WriteOptions wo;
   wo.disableWAL = true;
-  ASSERT_OK(TryReopen(options));
+  Reopen(options);
   // Create 4 files in L0
   for (char v = 'a'; v <= 'd'; v++) {
     ASSERT_OK(Put("Key2", DummyString(1024, v), wo));
@@ -540,10 +548,15 @@ TEST_P(DBWALTestWithParam, WALTrashCleanupOnOpen) {
   auto sfm = static_cast<SstFileManagerImpl*>(options.sst_file_manager.get());
   sfm->delete_scheduler()->SetMaxTrashDBRatio(3.1);
 
-  ASSERT_OK(TryReopen(options));
+  Reopen(options);
 
   // Create 4 files in L0
   for (char v = 'a'; v <= 'd'; v++) {
+    if (v == 'd') {
+      // Maximize the change that the last log file will be preserved in trash
+      // before restarting the DB.
+      options.sst_file_manager->SetDeleteRateBytesPerSecond(1);
+    }
     ASSERT_OK(Put("Key2", DummyString(1024, v)));
     ASSERT_OK(Put("Key3", DummyString(1024, v)));
     ASSERT_OK(Put("Key4", DummyString(1024, v)));
@@ -562,11 +575,11 @@ TEST_P(DBWALTestWithParam, WALTrashCleanupOnOpen) {
   if (!wal_dir_same_as_dbname_) {
     // Forcibly create some trash log files
     std::unique_ptr<WritableFile> result;
-    env->NewWritableFile(options.wal_dir + "/1000.log.trash", &result,
-                         EnvOptions());
+    ASSERT_OK(env->NewWritableFile(options.wal_dir + "/1000.log.trash", &result,
+                                   EnvOptions()));
     result.reset();
   }
-  env->GetChildren(options.wal_dir, &filenames);
+  ASSERT_OK(env->GetChildren(options.wal_dir, &filenames));
   for (const std::string& fname : filenames) {
     if (fname.find(".log.trash") != std::string::npos) {
       trash_log_count++;
@@ -575,11 +588,11 @@ TEST_P(DBWALTestWithParam, WALTrashCleanupOnOpen) {
   ASSERT_GE(trash_log_count, 1);
 
   env->set_fake_log_delete(false);
-  ASSERT_OK(TryReopen(options));
+  Reopen(options);
 
   filenames.clear();
   trash_log_count = 0;
-  env->GetChildren(options.wal_dir, &filenames);
+  ASSERT_OK(env->GetChildren(options.wal_dir, &filenames));
   for (const std::string& fname : filenames) {
     if (fname.find(".log.trash") != std::string::npos) {
       trash_log_count++;
@@ -603,13 +616,13 @@ TEST_F(DBSSTTest, OpenDBWithExistingTrash) {
   Destroy(last_options_);
 
   // Add some trash files to the db directory so the DB can clean them up
-  env_->CreateDirIfMissing(dbname_);
+  ASSERT_OK(env_->CreateDirIfMissing(dbname_));
   ASSERT_OK(WriteStringToFile(env_, "abc", dbname_ + "/" + "001.sst.trash"));
   ASSERT_OK(WriteStringToFile(env_, "abc", dbname_ + "/" + "002.sst.trash"));
   ASSERT_OK(WriteStringToFile(env_, "abc", dbname_ + "/" + "003.sst.trash"));
 
   // Reopen the DB and verify that it deletes existing trash files
-  ASSERT_OK(TryReopen(options));
+  Reopen(options);
   sfm->WaitForEmptyTrash();
   ASSERT_NOK(env_->FileExists(dbname_ + "/" + "001.sst.trash"));
   ASSERT_NOK(env_->FileExists(dbname_ + "/" + "002.sst.trash"));
@@ -728,7 +741,7 @@ TEST_F(DBSSTTest, DestroyDBWithRateLimitedDelete) {
   int num_sst_files = 0;
   int num_wal_files = 0;
   std::vector<std::string> db_files;
-  env_->GetChildren(dbname_, &db_files);
+  ASSERT_OK(env_->GetChildren(dbname_, &db_files));
   for (std::string f : db_files) {
     if (f.substr(f.find_last_of(".") + 1) == "sst") {
       num_sst_files++;
@@ -817,7 +830,7 @@ TEST_F(DBSSTTest, CancellingCompactionsWorks) {
     ASSERT_OK(Put(Key(i), rnd.RandomString(50)));
   }
   ASSERT_OK(Flush());
-  dbfull()->TEST_WaitForCompact(true);
+  ASSERT_OK(dbfull()->TEST_WaitForCompact(true));
 
   // Because we set a callback in CancelledCompaction, we actually
   // let the compaction run
@@ -867,10 +880,12 @@ TEST_F(DBSSTTest, CancellingManualCompactionsWorks) {
   ASSERT_OK(Flush());
 
   // OK, now trigger a manual compaction
-  dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr);
+  ASSERT_TRUE(dbfull()
+                  ->CompactRange(CompactRangeOptions(), nullptr, nullptr)
+                  .IsCompactionTooLarge());
 
   // Wait for manual compaction to get scheduled and finish
-  dbfull()->TEST_WaitForCompact(true);
+  ASSERT_OK(dbfull()->TEST_WaitForCompact(true));
 
   ASSERT_EQ(sfm->GetCompactionsReservedSize(), 0);
   // Make sure the stat is bumped
@@ -880,10 +895,13 @@ TEST_F(DBSSTTest, CancellingManualCompactionsWorks) {
 
   // Now make sure CompactFiles also gets cancelled
   auto l0_files = collector->GetFlushedFiles();
-  dbfull()->CompactFiles(ROCKSDB_NAMESPACE::CompactionOptions(), l0_files, 0);
+  ASSERT_TRUE(
+      dbfull()
+          ->CompactFiles(ROCKSDB_NAMESPACE::CompactionOptions(), l0_files, 0)
+          .IsCompactionTooLarge());
 
   // Wait for manual compaction to get scheduled and finish
-  dbfull()->TEST_WaitForCompact(true);
+  ASSERT_OK(dbfull()->TEST_WaitForCompact(true));
 
   ASSERT_EQ(dbfull()->immutable_db_options().statistics.get()->getTickerCount(
                 COMPACTION_CANCELLED),
@@ -898,8 +916,9 @@ TEST_F(DBSSTTest, CancellingManualCompactionsWorks) {
       "CompactFilesImpl:End", [&](void* /*arg*/) { completed_compactions++; });
 
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
-  dbfull()->CompactFiles(ROCKSDB_NAMESPACE::CompactionOptions(), l0_files, 0);
-  dbfull()->TEST_WaitForCompact(true);
+  ASSERT_OK(dbfull()->CompactFiles(ROCKSDB_NAMESPACE::CompactionOptions(),
+                                   l0_files, 0));
+  ASSERT_OK(dbfull()->TEST_WaitForCompact(true));
 
   ASSERT_EQ(sfm->GetCompactionsReservedSize(), 0);
   ASSERT_GT(completed_compactions, 0);
@@ -1003,7 +1022,7 @@ TEST_F(DBSSTTest, OpenDBWithInfiniteMaxOpenFiles) {
     CompactRangeOptions compact_options;
     compact_options.change_level = true;
     compact_options.target_level = 2;
-    db_->CompactRange(compact_options, nullptr, nullptr);
+    ASSERT_OK(db_->CompactRange(compact_options, nullptr, nullptr));
 
     // Create 12 Files in L0
     for (int i = 0; i < 12; i++) {
@@ -1055,7 +1074,7 @@ TEST_F(DBSSTTest, GetTotalSstFilesSize) {
       std::string val = "val_file_" + ToString(i);
       ASSERT_OK(Put(Key(j), val));
     }
-    Flush();
+    ASSERT_OK(Flush());
   }
   ASSERT_EQ("5", FilesPerLevel(0));
 
@@ -1079,6 +1098,7 @@ TEST_F(DBSSTTest, GetTotalSstFilesSize) {
 
   // hold current version
   std::unique_ptr<Iterator> iter1(dbfull()->NewIterator(ReadOptions()));
+  ASSERT_OK(iter1->status());
 
   // Compact 5 files into 1 file in L0
   ASSERT_OK(dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr));
@@ -1102,12 +1122,13 @@ TEST_F(DBSSTTest, GetTotalSstFilesSize) {
 
   // hold current version
   std::unique_ptr<Iterator> iter2(dbfull()->NewIterator(ReadOptions()));
+  ASSERT_OK(iter2->status());
 
   // Delete all keys and compact, this will delete all live files
   for (int i = 0; i < 10; i++) {
     ASSERT_OK(Delete(Key(i)));
   }
-  Flush();
+  ASSERT_OK(Flush());
   ASSERT_OK(dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr));
   ASSERT_EQ("", FilesPerLevel(0));
 
@@ -1121,6 +1142,7 @@ TEST_F(DBSSTTest, GetTotalSstFilesSize) {
   // Total SST files = 6 (5 original files + compacted file)
   ASSERT_EQ(total_sst_files_size, 6 * single_file_size);
 
+  ASSERT_OK(iter1->status());
   iter1.reset();
   ASSERT_TRUE(dbfull()->GetIntProperty("rocksdb.total-sst-files-size",
                                        &total_sst_files_size));
@@ -1128,6 +1150,7 @@ TEST_F(DBSSTTest, GetTotalSstFilesSize) {
   // Total SST files = 1 (compacted file)
   ASSERT_EQ(total_sst_files_size, 1 * single_file_size);
 
+  ASSERT_OK(iter2->status());
   iter2.reset();
   ASSERT_TRUE(dbfull()->GetIntProperty("rocksdb.total-sst-files-size",
                                        &total_sst_files_size));
@@ -1146,7 +1169,7 @@ TEST_F(DBSSTTest, GetTotalSstFilesSizeVersionsFilesShared) {
   // Generate 5 files in L0
   for (int i = 0; i < 5; i++) {
     ASSERT_OK(Put(Key(i), "val"));
-    Flush();
+    ASSERT_OK(Flush());
   }
   ASSERT_EQ("5", FilesPerLevel(0));
 
@@ -1171,6 +1194,7 @@ TEST_F(DBSSTTest, GetTotalSstFilesSizeVersionsFilesShared) {
 
   // hold current version
   std::unique_ptr<Iterator> iter1(dbfull()->NewIterator(ReadOptions()));
+  ASSERT_OK(iter1->status());
 
   // Compaction will do trivial move from L0 to L1
   ASSERT_OK(dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr));
@@ -1194,12 +1218,13 @@ TEST_F(DBSSTTest, GetTotalSstFilesSizeVersionsFilesShared) {
 
   // hold current version
   std::unique_ptr<Iterator> iter2(dbfull()->NewIterator(ReadOptions()));
+  ASSERT_OK(iter2->status());
 
   // Delete all keys and compact, this will delete all live files
   for (int i = 0; i < 5; i++) {
     ASSERT_OK(Delete(Key(i)));
   }
-  Flush();
+  ASSERT_OK(Flush());
   ASSERT_OK(dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr));
   ASSERT_EQ("", FilesPerLevel(0));
 
@@ -1213,7 +1238,9 @@ TEST_F(DBSSTTest, GetTotalSstFilesSizeVersionsFilesShared) {
   // Total SST files = 5 (used in 2 version)
   ASSERT_EQ(total_sst_files_size, 5 * single_file_size);
 
+  ASSERT_OK(iter1->status());
   iter1.reset();
+  ASSERT_OK(iter2->status());
   iter2.reset();
 
   ASSERT_TRUE(dbfull()->GetIntProperty("rocksdb.total-sst-files-size",

db/db_statistics_test.cc

@@ -137,7 +137,7 @@ TEST_F(DBStatisticsTest, ResetStats) {
       ASSERT_EQ(1, TestGetTickerCount(options, NUMBER_KEYS_WRITTEN));
       options.statistics->histogramData(DB_WRITE, &histogram_data);
       ASSERT_GT(histogram_data.max, 0.0);
-      options.statistics->Reset();
+      ASSERT_OK(options.statistics->Reset());
     }
   }
 }

db/db_table_properties_test.cc

@@ -65,9 +65,9 @@ TEST_F(DBTablePropertiesTest, GetPropertiesOfAllTablesTest) {
   // Create 4 tables
   for (int table = 0; table < 4; ++table) {
     for (int i = 0; i < 10 + table; ++i) {
-      db_->Put(WriteOptions(), ToString(table * 100 + i), "val");
+      ASSERT_OK(db_->Put(WriteOptions(), ToString(table * 100 + i), "val"));
     }
-    db_->Flush(FlushOptions());
+    ASSERT_OK(db_->Flush(FlushOptions()));
   }
 
   // 1. Read table properties directly from file
@@ -161,14 +161,14 @@ TEST_F(DBTablePropertiesTest, GetPropertiesOfTablesInRange) {
   for (int i = 0; i < 10000; i++) {
     ASSERT_OK(Put(test::RandomKey(&rnd, 5), rnd.RandomString(102)));
   }
-  Flush();
-  dbfull()->TEST_WaitForCompact();
+  ASSERT_OK(Flush());
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
   if (NumTableFilesAtLevel(0) == 0) {
     ASSERT_OK(Put(test::RandomKey(&rnd, 5), rnd.RandomString(102)));
-    Flush();
+    ASSERT_OK(Flush());
   }
 
-  db_->PauseBackgroundWork();
+  ASSERT_OK(db_->PauseBackgroundWork());
 
   // Ensure that we have at least L0, L1 and L2
   ASSERT_GT(NumTableFilesAtLevel(0), 0);
@@ -236,8 +236,8 @@ TEST_F(DBTablePropertiesTest, GetColumnFamilyNameProperty) {
   // Create one table per CF, then verify it was created with the column family
   // name property.
   for (uint32_t cf = 0; cf < 2; ++cf) {
-    Put(cf, "key", "val");
-    Flush(cf);
+    ASSERT_OK(Put(cf, "key", "val"));
+    ASSERT_OK(Flush(cf));
 
     TablePropertiesCollection fname_to_props;
     ASSERT_OK(db_->GetPropertiesOfAllTables(handles_[cf], &fname_to_props));
@@ -260,17 +260,17 @@ TEST_F(DBTablePropertiesTest, GetDbIdentifiersProperty) {
   CreateAndReopenWithCF({"goku"}, CurrentOptions());
 
   for (uint32_t cf = 0; cf < 2; ++cf) {
-    Put(cf, "key", "val");
-    Put(cf, "foo", "bar");
-    Flush(cf);
+    ASSERT_OK(Put(cf, "key", "val"));
+    ASSERT_OK(Put(cf, "foo", "bar"));
+    ASSERT_OK(Flush(cf));
 
     TablePropertiesCollection fname_to_props;
     ASSERT_OK(db_->GetPropertiesOfAllTables(handles_[cf], &fname_to_props));
     ASSERT_EQ(1U, fname_to_props.size());
 
     std::string id, sid;
-    db_->GetDbIdentity(id);
-    db_->GetDbSessionId(sid);
+    ASSERT_OK(db_->GetDbIdentity(id));
+    ASSERT_OK(db_->GetDbSessionId(sid));
     ASSERT_EQ(id, fname_to_props.begin()->second->db_id);
     ASSERT_EQ(sid, fname_to_props.begin()->second->db_session_id);
   }
@@ -298,9 +298,9 @@ TEST_P(DBTableHostnamePropertyTest, DbHostLocationProperty) {
   CreateAndReopenWithCF({"goku"}, opts);
 
   for (uint32_t cf = 0; cf < 2; ++cf) {
-    Put(cf, "key", "val");
-    Put(cf, "foo", "bar");
-    Flush(cf);
+    ASSERT_OK(Put(cf, "key", "val"));
+    ASSERT_OK(Put(cf, "foo", "bar"));
+    ASSERT_OK(Flush(cf));
 
     TablePropertiesCollection fname_to_props;
     ASSERT_OK(db_->GetPropertiesOfAllTables(handles_[cf], &fname_to_props));
@@ -356,8 +356,8 @@ TEST_P(DBTablePropertiesTest, DeletionTriggeredCompactionMarking) {
 
   // add an L1 file to prevent tombstones from dropping due to obsolescence
   // during flush
-  Put(Key(0), "val");
-  Flush();
+  ASSERT_OK(Put(Key(0), "val"));
+  ASSERT_OK(Flush());
   MoveFilesToLevel(1);
 
   DeletionTriggeredCompactionTestListener *listener =
@@ -368,14 +368,14 @@ TEST_P(DBTablePropertiesTest, DeletionTriggeredCompactionMarking) {
   for (int i = 0; i < kNumKeys; ++i) {
     if (i >= kNumKeys - kWindowSize &&
         i < kNumKeys - kWindowSize + kNumDelsTrigger) {
-      Delete(Key(i));
+      ASSERT_OK(Delete(Key(i)));
     } else {
-      Put(Key(i), "val");
+      ASSERT_OK(Put(Key(i), "val"));
     }
   }
-  Flush();
+  ASSERT_OK(Flush());
 
-  dbfull()->TEST_WaitForCompact();
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
   ASSERT_EQ(0, NumTableFilesAtLevel(0));
 
   // Change the window size and deletion trigger and ensure new values take
@@ -389,14 +389,14 @@ TEST_P(DBTablePropertiesTest, DeletionTriggeredCompactionMarking) {
   for (int i = 0; i < kNumKeys; ++i) {
     if (i >= kNumKeys - kWindowSize &&
         i < kNumKeys - kWindowSize + kNumDelsTrigger) {
-      Delete(Key(i));
+      ASSERT_OK(Delete(Key(i)));
     } else {
-      Put(Key(i), "val");
+      ASSERT_OK(Put(Key(i), "val"));
     }
   }
-  Flush();
+  ASSERT_OK(Flush());
 
-  dbfull()->TEST_WaitForCompact();
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
   ASSERT_EQ(0, NumTableFilesAtLevel(0));
 
   // Change the window size to disable delete triggered compaction
@@ -408,14 +408,14 @@ TEST_P(DBTablePropertiesTest, DeletionTriggeredCompactionMarking) {
   for (int i = 0; i < kNumKeys; ++i) {
     if (i >= kNumKeys - kWindowSize &&
         i < kNumKeys - kWindowSize + kNumDelsTrigger) {
-      Delete(Key(i));
+      ASSERT_OK(Delete(Key(i)));
     } else {
-      Put(Key(i), "val");
+      ASSERT_OK(Put(Key(i), "val"));
     }
   }
-  Flush();
+  ASSERT_OK(Flush());
 
-  dbfull()->TEST_WaitForCompact();
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
   ASSERT_EQ(1, NumTableFilesAtLevel(0));
   ASSERT_LT(0, opts.statistics->getTickerCount(COMPACT_WRITE_BYTES_MARKED));
   ASSERT_LT(0, opts.statistics->getTickerCount(COMPACT_READ_BYTES_MARKED));
@@ -438,8 +438,8 @@ TEST_P(DBTablePropertiesTest, RatioBasedDeletionTriggeredCompactionMarking) {
 
   // Add an L2 file to prevent tombstones from dropping due to obsolescence
   // during flush
-  Put(Key(0), "val");
-  Flush();
+  ASSERT_OK(Put(Key(0), "val"));
+  ASSERT_OK(Flush());
   MoveFilesToLevel(2);
 
   auto* listener = new DeletionTriggeredCompactionTestListener();

db/db_tailing_iter_test.cc

@@ -31,6 +31,7 @@ TEST_F(DBTestTailingIterator, TailingIteratorSingle) {
   std::unique_ptr<Iterator> iter(db_->NewIterator(read_options));
   iter->SeekToFirst();
   ASSERT_TRUE(!iter->Valid());
+  ASSERT_OK(iter->status());
 
   // add a record and check that iter can see it
   ASSERT_OK(db_->Put(WriteOptions(), "mirko", "fodor"));
@@ -48,6 +49,7 @@ TEST_F(DBTestTailingIterator, TailingIteratorKeepAdding) {
   read_options.tailing = true;
 
   std::unique_ptr<Iterator> iter(db_->NewIterator(read_options, handles_[1]));
+  ASSERT_OK(iter->status());
   std::string value(1024, 'a');
 
   const int num_records = 10000;
@@ -70,7 +72,9 @@ TEST_F(DBTestTailingIterator, TailingIteratorSeekToNext) {
   read_options.tailing = true;
 
   std::unique_ptr<Iterator> iter(db_->NewIterator(read_options, handles_[1]));
+  ASSERT_OK(iter->status());
   std::unique_ptr<Iterator> itern(db_->NewIterator(read_options, handles_[1]));
+  ASSERT_OK(itern->status());
   std::string value(1024, 'a');
 
   const int num_records = 1000;
@@ -138,8 +142,11 @@ TEST_F(DBTestTailingIterator, TailingIteratorTrimSeekToNext) {
   Slice keyu(bufe, 20);
   read_options.iterate_upper_bound = &keyu;
   std::unique_ptr<Iterator> iter(db_->NewIterator(read_options, handles_[1]));
+  ASSERT_OK(iter->status());
   std::unique_ptr<Iterator> itern(db_->NewIterator(read_options, handles_[1]));
+  ASSERT_OK(itern->status());
   std::unique_ptr<Iterator> iterh(db_->NewIterator(read_options, handles_[1]));
+  ASSERT_OK(iterh->status());
   std::string value(1024, 'a');
   bool file_iters_deleted = false;
   bool file_iters_renewed_null = false;
@@ -179,7 +186,7 @@ TEST_F(DBTestTailingIterator, TailingIteratorTrimSeekToNext) {
 
     if (i % 100 == 99) {
       ASSERT_OK(Flush(1));
-      dbfull()->TEST_WaitForCompact();
+      ASSERT_OK(dbfull()->TEST_WaitForCompact());
       if (i == 299) {
         file_iters_deleted = true;
       }
@@ -225,6 +232,7 @@ TEST_F(DBTestTailingIterator, TailingIteratorTrimSeekToNext) {
   ReopenWithColumnFamilies({"default", "pikachu"}, options);
   read_options.read_tier = kBlockCacheTier;
   std::unique_ptr<Iterator> iteri(db_->NewIterator(read_options, handles_[1]));
+  ASSERT_OK(iteri->status());
   char buf5[32];
   snprintf(buf5, sizeof(buf5), "00a0%016d", (num_records / 2) * 5 - 2);
   Slice target1(buf5, 20);
@@ -236,6 +244,7 @@ TEST_F(DBTestTailingIterator, TailingIteratorTrimSeekToNext) {
   options.table_factory.reset(NewBlockBasedTableFactory());
   ReopenWithColumnFamilies({"default", "pikachu"}, options);
   iter.reset(db_->NewIterator(read_options, handles_[1]));
+  ASSERT_OK(iter->status());
   for (int i = 2 * num_records; i > 0; --i) {
     char buf1[32];
     char buf2[32];
@@ -262,6 +271,7 @@ TEST_F(DBTestTailingIterator, TailingIteratorDeletes) {
   read_options.tailing = true;
 
   std::unique_ptr<Iterator> iter(db_->NewIterator(read_options, handles_[1]));
+  ASSERT_OK(iter->status());
 
   // write a single record, read it using the iterator, then delete it
   ASSERT_OK(Put(1, "0test", "test"));
@@ -309,6 +319,7 @@ TEST_F(DBTestTailingIterator, TailingIteratorPrefixSeek) {
   CreateAndReopenWithCF({"pikachu"}, options);
 
   std::unique_ptr<Iterator> iter(db_->NewIterator(read_options, handles_[1]));
+  ASSERT_OK(iter->status());
   ASSERT_OK(Put(1, "0101", "test"));
 
   ASSERT_OK(Flush(1));
@@ -339,6 +350,7 @@ TEST_F(DBTestTailingIterator, TailingIteratorIncomplete) {
   ASSERT_OK(db_->Put(WriteOptions(), key, value));
 
   std::unique_ptr<Iterator> iter(db_->NewIterator(read_options));
+  ASSERT_OK(iter->status());
   iter->SeekToFirst();
   // we either see the entry or it's not in cache
   ASSERT_TRUE(iter->Valid() || iter->status().IsIncomplete());
@@ -369,6 +381,7 @@ TEST_F(DBTestTailingIterator, TailingIteratorSeekToSame) {
   }
 
   std::unique_ptr<Iterator> iter(db_->NewIterator(read_options));
+  ASSERT_OK(iter->status());
   // Seek to 00001.  We expect to find 00002.
   std::string start_key = "00001";
   iter->Seek(start_key);
@@ -404,6 +417,7 @@ TEST_F(DBTestTailingIterator, TailingIteratorUpperBound) {
   ASSERT_OK(Put(1, "21", "21"));
 
   std::unique_ptr<Iterator> it(db_->NewIterator(read_options, handles_[1]));
+  ASSERT_OK(it->status());
   it->Seek("12");
   ASSERT_TRUE(it->Valid());
   ASSERT_EQ("12", it->key().ToString());
@@ -411,7 +425,7 @@ TEST_F(DBTestTailingIterator, TailingIteratorUpperBound) {
   it->Next();
   // Not valid since "21" is over the upper bound.
   ASSERT_FALSE(it->Valid());
-
+  ASSERT_OK(it->status());
   // This keeps track of the number of times NeedToSeekImmutable() was true.
   int immutable_seeks = 0;
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
@@ -424,6 +438,7 @@ TEST_F(DBTestTailingIterator, TailingIteratorUpperBound) {
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
 
   ASSERT_FALSE(it->Valid());
+  ASSERT_OK(it->status());
   ASSERT_EQ(0, immutable_seeks);
 }
 
@@ -478,6 +493,8 @@ TEST_F(DBTestTailingIterator, TailingIteratorGap) {
   it->Next();
   ASSERT_TRUE(it->Valid());
   ASSERT_EQ("40", it->key().ToString());
+
+  ASSERT_OK(it->status());
 }
 
 TEST_F(DBTestTailingIterator, SeekWithUpperBoundBug) {
@@ -496,6 +513,7 @@ TEST_F(DBTestTailingIterator, SeekWithUpperBoundBug) {
   ASSERT_OK(Flush());
 
   std::unique_ptr<Iterator> iter(db_->NewIterator(read_options));
+  ASSERT_OK(iter->status());
 
   iter->Seek("aa");
   ASSERT_TRUE(iter->Valid());
@@ -518,6 +536,7 @@ TEST_F(DBTestTailingIterator, SeekToFirstWithUpperBoundBug) {
   ASSERT_OK(Flush());
 
   std::unique_ptr<Iterator> iter(db_->NewIterator(read_options));
+  ASSERT_OK(iter->status());
 
   iter->SeekToFirst();
   ASSERT_TRUE(iter->Valid());

db/db_test.cc

@@ -1305,51 +1305,6 @@ TEST_F(DBTest, DISABLED_RepeatedWritesToSameKey) {
 }
 #endif  // ROCKSDB_LITE
 
-TEST_F(DBTest, SparseMerge) {
-  do {
-    Options options = CurrentOptions();
-    options.compression = kNoCompression;
-    CreateAndReopenWithCF({"pikachu"}, options);
-
-    FillLevels("A", "Z", 1);
-
-    // Suppose there is:
-    //    small amount of data with prefix A
-    //    large amount of data with prefix B
-    //    small amount of data with prefix C
-    // and that recent updates have made small changes to all three prefixes.
-    // Check that we do not do a compaction that merges all of B in one shot.
-    const std::string value(1000, 'x');
-    Put(1, "A", "va");
-    // Write approximately 100MB of "B" values
-    for (int i = 0; i < 100000; i++) {
-      char key[100];
-      snprintf(key, sizeof(key), "B%010d", i);
-      Put(1, key, value);
-    }
-    Put(1, "C", "vc");
-    ASSERT_OK(Flush(1));
-    dbfull()->TEST_CompactRange(0, nullptr, nullptr, handles_[1]);
-
-    // Make sparse update
-    Put(1, "A", "va2");
-    Put(1, "B100", "bvalue2");
-    Put(1, "C", "vc2");
-    ASSERT_OK(Flush(1));
-
-    // Compactions should not cause us to create a situation where
-    // a file overlaps too much data at the next level.
-    ASSERT_LE(dbfull()->TEST_MaxNextLevelOverlappingBytes(handles_[1]),
-              20 * 1048576);
-    dbfull()->TEST_CompactRange(0, nullptr, nullptr);
-    ASSERT_LE(dbfull()->TEST_MaxNextLevelOverlappingBytes(handles_[1]),
-              20 * 1048576);
-    dbfull()->TEST_CompactRange(1, nullptr, nullptr);
-    ASSERT_LE(dbfull()->TEST_MaxNextLevelOverlappingBytes(handles_[1]),
-              20 * 1048576);
-  } while (ChangeCompactOptions());
-}
-
 #ifndef ROCKSDB_LITE
 static bool Between(uint64_t val, uint64_t low, uint64_t high) {
   bool result = (val >= low) && (val <= high);
@@ -1382,17 +1337,19 @@ TEST_F(DBTest, ApproximateSizesMemTable) {
   SizeApproximationOptions size_approx_options;
   size_approx_options.include_memtabtles = true;
   size_approx_options.include_files = true;
-  db_->GetApproximateSizes(size_approx_options, default_cf, &r, 1, &size);
+  ASSERT_OK(
+      db_->GetApproximateSizes(size_approx_options, default_cf, &r, 1, &size));
   ASSERT_GT(size, 6000);
   ASSERT_LT(size, 204800);
   // Zero if not including mem table
-  db_->GetApproximateSizes(&r, 1, &size);
+  ASSERT_OK(db_->GetApproximateSizes(&r, 1, &size));
   ASSERT_EQ(size, 0);
 
   start = Key(500);
   end = Key(600);
   r = Range(start, end);
-  db_->GetApproximateSizes(size_approx_options, default_cf, &r, 1, &size);
+  ASSERT_OK(
+      db_->GetApproximateSizes(size_approx_options, default_cf, &r, 1, &size));
   ASSERT_EQ(size, 0);
 
   for (int i = 0; i < N; i++) {
@@ -1402,13 +1359,15 @@ TEST_F(DBTest, ApproximateSizesMemTable) {
   start = Key(500);
   end = Key(600);
   r = Range(start, end);
-  db_->GetApproximateSizes(size_approx_options, default_cf, &r, 1, &size);
+  ASSERT_OK(
+      db_->GetApproximateSizes(size_approx_options, default_cf, &r, 1, &size));
   ASSERT_EQ(size, 0);
 
   start = Key(100);
   end = Key(1020);
   r = Range(start, end);
-  db_->GetApproximateSizes(size_approx_options, default_cf, &r, 1, &size);
+  ASSERT_OK(
+      db_->GetApproximateSizes(size_approx_options, default_cf, &r, 1, &size));
   ASSERT_GT(size, 6000);
 
   options.max_write_buffer_number = 8;
@@ -1434,29 +1393,32 @@ TEST_F(DBTest, ApproximateSizesMemTable) {
   start = Key(100);
   end = Key(300);
   r = Range(start, end);
-  db_->GetApproximateSizes(size_approx_options, default_cf, &r, 1, &size);
+  ASSERT_OK(
+      db_->GetApproximateSizes(size_approx_options, default_cf, &r, 1, &size));
   ASSERT_EQ(size, 0);
 
   start = Key(1050);
   end = Key(1080);
   r = Range(start, end);
-  db_->GetApproximateSizes(size_approx_options, default_cf, &r, 1, &size);
+  ASSERT_OK(
+      db_->GetApproximateSizes(size_approx_options, default_cf, &r, 1, &size));
   ASSERT_GT(size, 6000);
 
   start = Key(2100);
   end = Key(2300);
   r = Range(start, end);
-  db_->GetApproximateSizes(size_approx_options, default_cf, &r, 1, &size);
+  ASSERT_OK(
+      db_->GetApproximateSizes(size_approx_options, default_cf, &r, 1, &size));
   ASSERT_EQ(size, 0);
 
   start = Key(1050);
   end = Key(1080);
   r = Range(start, end);
   uint64_t size_with_mt, size_without_mt;
-  db_->GetApproximateSizes(size_approx_options, default_cf, &r, 1,
-                           &size_with_mt);
+  ASSERT_OK(db_->GetApproximateSizes(size_approx_options, default_cf, &r, 1,
+                                     &size_with_mt));
   ASSERT_GT(size_with_mt, 6000);
-  db_->GetApproximateSizes(&r, 1, &size_without_mt);
+  ASSERT_OK(db_->GetApproximateSizes(&r, 1, &size_without_mt));
   ASSERT_EQ(size_without_mt, 0);
 
   Flush();
@@ -1468,15 +1430,16 @@ TEST_F(DBTest, ApproximateSizesMemTable) {
   start = Key(1050);
   end = Key(1080);
   r = Range(start, end);
-  db_->GetApproximateSizes(size_approx_options, default_cf, &r, 1,
-                           &size_with_mt);
-  db_->GetApproximateSizes(&r, 1, &size_without_mt);
+  ASSERT_OK(db_->GetApproximateSizes(size_approx_options, default_cf, &r, 1,
+                                     &size_with_mt));
+  ASSERT_OK(db_->GetApproximateSizes(&r, 1, &size_without_mt));
   ASSERT_GT(size_with_mt, size_without_mt);
   ASSERT_GT(size_without_mt, 6000);
 
   // Check that include_memtabtles flag works as expected
   size_approx_options.include_memtabtles = false;
-  db_->GetApproximateSizes(size_approx_options, default_cf, &r, 1, &size);
+  ASSERT_OK(
+      db_->GetApproximateSizes(size_approx_options, default_cf, &r, 1, &size));
   ASSERT_EQ(size, size_without_mt);
 
   // Check that files_size_error_margin works as expected, when the heuristic
@@ -1485,10 +1448,12 @@ TEST_F(DBTest, ApproximateSizesMemTable) {
   end = Key(1000 + N - 2);
   r = Range(start, end);
   size_approx_options.files_size_error_margin = -1.0;  // disabled
-  db_->GetApproximateSizes(size_approx_options, default_cf, &r, 1, &size);
+  ASSERT_OK(
+      db_->GetApproximateSizes(size_approx_options, default_cf, &r, 1, &size));
   uint64_t size2;
   size_approx_options.files_size_error_margin = 0.5;  // enabled, but not used
-  db_->GetApproximateSizes(size_approx_options, default_cf, &r, 1, &size2);
+  ASSERT_OK(
+      db_->GetApproximateSizes(size_approx_options, default_cf, &r, 1, &size2));
   ASSERT_EQ(size, size2);
 }
 
@@ -1539,14 +1504,16 @@ TEST_F(DBTest, ApproximateSizesFilesWithErrorMargin) {
 
     // Get the precise size without any approximation heuristic
     uint64_t size;
-    db_->GetApproximateSizes(size_approx_options, default_cf, &r, 1, &size);
+    ASSERT_OK(db_->GetApproximateSizes(size_approx_options, default_cf, &r, 1,
+                                       &size));
     ASSERT_NE(size, 0);
 
     // Get the size with an approximation heuristic
     uint64_t size2;
     const double error_margin = 0.2;
     size_approx_options.files_size_error_margin = error_margin;
-    db_->GetApproximateSizes(size_approx_options, default_cf, &r, 1, &size2);
+    ASSERT_OK(db_->GetApproximateSizes(size_approx_options, default_cf, &r, 1,
+                                       &size2));
     ASSERT_LT(size2, size * (1 + error_margin));
     ASSERT_GT(size2, size * (1 - error_margin));
   }
@@ -1562,7 +1529,7 @@ TEST_F(DBTest, ApproximateSizesFilesWithErrorMargin) {
       const std::string end = Key(i + 11);  // overlap by 1 key
       const Range r(start, end);
       uint64_t size;
-      db_->GetApproximateSizes(&r, 1, &size);
+      ASSERT_OK(db_->GetApproximateSizes(&r, 1, &size));
       ASSERT_LE(size, 11 * 100);
     }
   }
@@ -1630,9 +1597,12 @@ TEST_F(DBTest, ApproximateSizes) {
     DestroyAndReopen(options);
     CreateAndReopenWithCF({"pikachu"}, options);
 
-    ASSERT_TRUE(Between(Size("", "xyz", 1), 0, 0));
+    uint64_t size;
+    ASSERT_OK(Size("", "xyz", 1, &size));
+    ASSERT_TRUE(Between(size, 0, 0));
     ReopenWithColumnFamilies({"default", "pikachu"}, options);
-    ASSERT_TRUE(Between(Size("", "xyz", 1), 0, 0));
+    ASSERT_OK(Size("", "xyz", 1, &size));
+    ASSERT_TRUE(Between(size, 0, 0));
 
     // Write 8MB (80 values, each 100K)
     ASSERT_EQ(NumTableFilesAtLevel(0, 1), 0);
@@ -1645,7 +1615,8 @@ TEST_F(DBTest, ApproximateSizes) {
     }
 
     // 0 because GetApproximateSizes() does not account for memtable space
-    ASSERT_TRUE(Between(Size("", Key(50), 1), 0, 0));
+    ASSERT_OK(Size("", Key(50), 1, &size));
+    ASSERT_TRUE(Between(size, 0, 0));
 
     // Check sizes across recovery by reopening a few times
     for (int run = 0; run < 3; run++) {
@@ -1653,14 +1624,17 @@ TEST_F(DBTest, ApproximateSizes) {
 
       for (int compact_start = 0; compact_start < N; compact_start += 10) {
         for (int i = 0; i < N; i += 10) {
-          ASSERT_TRUE(Between(Size("", Key(i), 1), S1 * i, S2 * i));
-          ASSERT_TRUE(Between(Size("", Key(i) + ".suffix", 1), S1 * (i + 1),
-                              S2 * (i + 1)));
-          ASSERT_TRUE(Between(Size(Key(i), Key(i + 10), 1), S1 * 10, S2 * 10));
+          ASSERT_OK(Size("", Key(i), 1, &size));
+          ASSERT_TRUE(Between(size, S1 * i, S2 * i));
+          ASSERT_OK(Size("", Key(i) + ".suffix", 1, &size));
+          ASSERT_TRUE(Between(size, S1 * (i + 1), S2 * (i + 1)));
+          ASSERT_OK(Size(Key(i), Key(i + 10), 1, &size));
+          ASSERT_TRUE(Between(size, S1 * 10, S2 * 10));
         }
-        ASSERT_TRUE(Between(Size("", Key(50), 1), S1 * 50, S2 * 50));
-        ASSERT_TRUE(
-            Between(Size("", Key(50) + ".suffix", 1), S1 * 50, S2 * 50));
+        ASSERT_OK(Size("", Key(50), 1, &size));
+        ASSERT_TRUE(Between(size, S1 * 50, S2 * 50));
+        ASSERT_OK(Size("", Key(50) + ".suffix", 1, &size));
+        ASSERT_TRUE(Between(size, S1 * 50, S2 * 50));
 
         std::string cstart_str = Key(compact_start);
         std::string cend_str = Key(compact_start + 9);
@@ -1695,21 +1669,32 @@ TEST_F(DBTest, ApproximateSizes_MixOfSmallAndLarge) {
     ASSERT_OK(Put(1, Key(7), rnd.RandomString(10000)));
 
     // Check sizes across recovery by reopening a few times
+    uint64_t size;
     for (int run = 0; run < 3; run++) {
       ReopenWithColumnFamilies({"default", "pikachu"}, options);
 
-      ASSERT_TRUE(Between(Size("", Key(0), 1), 0, 0));
-      ASSERT_TRUE(Between(Size("", Key(1), 1), 10000, 11000));
-      ASSERT_TRUE(Between(Size("", Key(2), 1), 20000, 21000));
-      ASSERT_TRUE(Between(Size("", Key(3), 1), 120000, 121000));
-      ASSERT_TRUE(Between(Size("", Key(4), 1), 130000, 131000));
-      ASSERT_TRUE(Between(Size("", Key(5), 1), 230000, 232000));
-      ASSERT_TRUE(Between(Size("", Key(6), 1), 240000, 242000));
+      ASSERT_OK(Size("", Key(0), 1, &size));
+      ASSERT_TRUE(Between(size, 0, 0));
+      ASSERT_OK(Size("", Key(1), 1, &size));
+      ASSERT_TRUE(Between(size, 10000, 11000));
+      ASSERT_OK(Size("", Key(2), 1, &size));
+      ASSERT_TRUE(Between(size, 20000, 21000));
+      ASSERT_OK(Size("", Key(3), 1, &size));
+      ASSERT_TRUE(Between(size, 120000, 121000));
+      ASSERT_OK(Size("", Key(4), 1, &size));
+      ASSERT_TRUE(Between(size, 130000, 131000));
+      ASSERT_OK(Size("", Key(5), 1, &size));
+      ASSERT_TRUE(Between(size, 230000, 232000));
+      ASSERT_OK(Size("", Key(6), 1, &size));
+      ASSERT_TRUE(Between(size, 240000, 242000));
       // Ensure some overhead is accounted for, even without including all
-      ASSERT_TRUE(Between(Size("", Key(7), 1), 540500, 545000));
-      ASSERT_TRUE(Between(Size("", Key(8), 1), 550500, 555000));
+      ASSERT_OK(Size("", Key(7), 1, &size));
+      ASSERT_TRUE(Between(size, 540500, 545000));
+      ASSERT_OK(Size("", Key(8), 1, &size));
+      ASSERT_TRUE(Between(size, 550500, 555000));
 
-      ASSERT_TRUE(Between(Size(Key(3), Key(5), 1), 110100, 111000));
+      ASSERT_OK(Size(Key(3), Key(5), 1, &size));
+      ASSERT_TRUE(Between(size, 110100, 111000));
 
       dbfull()->TEST_CompactRange(0, nullptr, nullptr, handles_[1]);
     }
@@ -1793,6 +1778,7 @@ TEST_F(DBTest, Snapshot) {
 TEST_F(DBTest, HiddenValuesAreRemoved) {
   anon::OptionsOverride options_override;
   options_override.skip_policy = kSkipNoSnapshot;
+  uint64_t size;
   do {
     Options options = CurrentOptions(options_override);
     CreateAndReopenWithCF({"pikachu"}, options);
@@ -1810,7 +1796,8 @@ TEST_F(DBTest, HiddenValuesAreRemoved) {
     ASSERT_GT(NumTableFilesAtLevel(0, 1), 0);
 
     ASSERT_EQ(big, Get(1, "foo", snapshot));
-    ASSERT_TRUE(Between(Size("", "pastfoo", 1), 50000, 60000));
+    ASSERT_OK(Size("", "pastfoo", 1, &size));
+    ASSERT_TRUE(Between(size, 50000, 60000));
     db_->ReleaseSnapshot(snapshot);
     ASSERT_EQ(AllEntriesFor("foo", 1), "[ tiny, " + big + " ]");
     Slice x("x");
@@ -1821,7 +1808,8 @@ TEST_F(DBTest, HiddenValuesAreRemoved) {
     dbfull()->TEST_CompactRange(1, nullptr, &x, handles_[1]);
     ASSERT_EQ(AllEntriesFor("foo", 1), "[ tiny ]");
 
-    ASSERT_TRUE(Between(Size("", "pastfoo", 1), 0, 1000));
+    ASSERT_OK(Size("", "pastfoo", 1, &size));
+    ASSERT_TRUE(Between(size, 0, 1000));
     // ApproximateOffsetOf() is not yet implemented in plain table format,
     // which is used by Size().
   } while (ChangeOptions(kSkipUniversalCompaction | kSkipFIFOCompaction |

db/db_test2.cc

@@ -41,9 +41,7 @@ TEST_F(DBTest2, OpenForReadOnly) {
   std::vector<std::string> files;
   ASSERT_OK(env_->GetChildren(dbname, &files));
   for (auto& f : files) {
-    if (f != "." && f != "..") {
-      ASSERT_OK(env_->DeleteFile(dbname + "/" + f));
-    }
+    ASSERT_OK(env_->DeleteFile(dbname + "/" + f));
   }
   // <dbname> should be empty now and we should be able to delete it
   ASSERT_OK(env_->DeleteDir(dbname));
@@ -75,9 +73,7 @@ TEST_F(DBTest2, OpenForReadOnlyWithColumnFamilies) {
   std::vector<std::string> files;
   ASSERT_OK(env_->GetChildren(dbname, &files));
   for (auto& f : files) {
-    if (f != "." && f != "..") {
-      ASSERT_OK(env_->DeleteFile(dbname + "/" + f));
-    }
+    ASSERT_OK(env_->DeleteFile(dbname + "/" + f));
   }
   // <dbname> should be empty now and we should be able to delete it
   ASSERT_OK(env_->DeleteDir(dbname));
@@ -174,7 +170,9 @@ TEST_F(DBTest2, PartitionedIndexUserToInternalKey) {
     ASSERT_OK(Put("keykey_" + std::to_string(j), value));
     snapshots.push_back(db_->GetSnapshot());
   }
-  Flush();
+  ASSERT_OK(Flush());
+  dbfull()->TEST_WaitForFlushMemTable();
+
   for (auto s : snapshots) {
     db_->ReleaseSnapshot(s);
   }
@@ -1412,67 +1410,94 @@ INSTANTIATE_TEST_CASE_P(
 
 TEST_P(PresetCompressionDictTest, Flush) {
   // Verifies that dictionary is generated and written during flush only when
-  // `ColumnFamilyOptions::compression` enables dictionary.
+  // `ColumnFamilyOptions::compression` enables dictionary. Also verifies the
+  // size of the dictionary is within expectations according to the limit on
+  // buffering set by `CompressionOptions::max_dict_buffer_bytes`.
   const size_t kValueLen = 256;
   const size_t kKeysPerFile = 1 << 10;
-  const size_t kDictLen = 4 << 10;
+  const size_t kDictLen = 16 << 10;
+  const size_t kBlockLen = 4 << 10;
 
   Options options = CurrentOptions();
   if (bottommost_) {
     options.bottommost_compression = compression_type_;
     options.bottommost_compression_opts.enabled = true;
     options.bottommost_compression_opts.max_dict_bytes = kDictLen;
+    options.bottommost_compression_opts.max_dict_buffer_bytes = kBlockLen;
   } else {
     options.compression = compression_type_;
     options.compression_opts.max_dict_bytes = kDictLen;
+    options.compression_opts.max_dict_buffer_bytes = kBlockLen;
   }
   options.memtable_factory.reset(new SpecialSkipListFactory(kKeysPerFile));
   options.statistics = CreateDBStatistics();
   BlockBasedTableOptions bbto;
+  bbto.block_size = kBlockLen;
   bbto.cache_index_and_filter_blocks = true;
   options.table_factory.reset(NewBlockBasedTableFactory(bbto));
   Reopen(options);
 
-  uint64_t prev_compression_dict_misses =
-      TestGetTickerCount(options, BLOCK_CACHE_COMPRESSION_DICT_MISS);
   Random rnd(301);
   for (size_t i = 0; i <= kKeysPerFile; ++i) {
     ASSERT_OK(Put(Key(static_cast<int>(i)), rnd.RandomString(kValueLen)));
   }
   ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable());
 
-  // If there's a compression dictionary, it should have been loaded when the
-  // flush finished, incurring a cache miss.
-  uint64_t expected_compression_dict_misses;
+  // We can use `BLOCK_CACHE_COMPRESSION_DICT_BYTES_INSERT` to detect whether a
+  // compression dictionary exists since dictionaries would be preloaded when
+  // the flush finishes.
   if (bottommost_) {
-    expected_compression_dict_misses = prev_compression_dict_misses;
+    // Flush is never considered bottommost. This should change in the future
+    // since flushed files may have nothing underneath them, like the one in
+    // this test case.
+    ASSERT_EQ(
+        TestGetTickerCount(options, BLOCK_CACHE_COMPRESSION_DICT_BYTES_INSERT),
+        0);
   } else {
-    expected_compression_dict_misses = prev_compression_dict_misses + 1;
+    ASSERT_GT(
+        TestGetTickerCount(options, BLOCK_CACHE_COMPRESSION_DICT_BYTES_INSERT),
+        0);
+    // TODO(ajkr): fix the below assertion to work with ZSTD. The expectation on
+    // number of bytes needs to be adjusted in case the cached block is in
+    // ZSTD's digested dictionary format.
+    if (compression_type_ != kZSTD &&
+        compression_type_ != kZSTDNotFinalCompression) {
+      // Although we limited buffering to `kBlockLen`, there may be up to two
+      // blocks of data included in the dictionary since we only check limit
+      // after each block is built.
+      ASSERT_LE(TestGetTickerCount(options,
+                                   BLOCK_CACHE_COMPRESSION_DICT_BYTES_INSERT),
+                2 * kBlockLen);
+    }
   }
-  ASSERT_EQ(expected_compression_dict_misses,
-            TestGetTickerCount(options, BLOCK_CACHE_COMPRESSION_DICT_MISS));
 }
 
 TEST_P(PresetCompressionDictTest, CompactNonBottommost) {
   // Verifies that dictionary is generated and written during compaction to
   // non-bottommost level only when `ColumnFamilyOptions::compression` enables
-  // dictionary.
+  // dictionary. Also verifies the size of the dictionary is within expectations
+  // according to the limit on buffering set by
+  // `CompressionOptions::max_dict_buffer_bytes`.
   const size_t kValueLen = 256;
   const size_t kKeysPerFile = 1 << 10;
-  const size_t kDictLen = 4 << 10;
+  const size_t kDictLen = 16 << 10;
+  const size_t kBlockLen = 4 << 10;
 
   Options options = CurrentOptions();
   if (bottommost_) {
     options.bottommost_compression = compression_type_;
     options.bottommost_compression_opts.enabled = true;
     options.bottommost_compression_opts.max_dict_bytes = kDictLen;
+    options.bottommost_compression_opts.max_dict_buffer_bytes = kBlockLen;
   } else {
     options.compression = compression_type_;
     options.compression_opts.max_dict_bytes = kDictLen;
+    options.compression_opts.max_dict_buffer_bytes = kBlockLen;
   }
   options.disable_auto_compactions = true;
   options.statistics = CreateDBStatistics();
   BlockBasedTableOptions bbto;
+  bbto.block_size = kBlockLen;
   bbto.cache_index_and_filter_blocks = true;
   options.table_factory.reset(NewBlockBasedTableFactory(bbto));
   Reopen(options);
@@ -1494,8 +1519,8 @@ TEST_P(PresetCompressionDictTest, CompactNonBottommost) {
   ASSERT_EQ("2,0,1", FilesPerLevel(0));
 #endif  // ROCKSDB_LITE
 
-  uint64_t prev_compression_dict_misses =
-      TestGetTickerCount(options, BLOCK_CACHE_COMPRESSION_DICT_MISS);
+  uint64_t prev_compression_dict_bytes_inserted =
+      TestGetTickerCount(options, BLOCK_CACHE_COMPRESSION_DICT_BYTES_INSERT);
   // This L0->L1 compaction merges the two L0 files into L1. The produced L1
   // file is not bottommost due to the existing L2 file covering the same key-
   // range.
@@ -1503,38 +1528,58 @@ TEST_P(PresetCompressionDictTest, CompactNonBottommost) {
 #ifndef ROCKSDB_LITE
   ASSERT_EQ("0,1,1", FilesPerLevel(0));
 #endif  // ROCKSDB_LITE
-  // If there's a compression dictionary, it should have been loaded when the
-  // compaction finished, incurring a cache miss.
-  uint64_t expected_compression_dict_misses;
+  // We can use `BLOCK_CACHE_COMPRESSION_DICT_BYTES_INSERT` to detect whether a
+  // compression dictionary exists since dictionaries would be preloaded when
+  // the compaction finishes.
   if (bottommost_) {
-    expected_compression_dict_misses = prev_compression_dict_misses;
+    ASSERT_EQ(
+        TestGetTickerCount(options, BLOCK_CACHE_COMPRESSION_DICT_BYTES_INSERT),
+        prev_compression_dict_bytes_inserted);
   } else {
-    expected_compression_dict_misses = prev_compression_dict_misses + 1;
+    ASSERT_GT(
+        TestGetTickerCount(options, BLOCK_CACHE_COMPRESSION_DICT_BYTES_INSERT),
+        prev_compression_dict_bytes_inserted);
+    // TODO(ajkr): fix the below assertion to work with ZSTD. The expectation on
+    // number of bytes needs to be adjusted in case the cached block is in
+    // ZSTD's digested dictionary format.
+    if (compression_type_ != kZSTD &&
+        compression_type_ != kZSTDNotFinalCompression) {
+      // Although we limited buffering to `kBlockLen`, there may be up to two
+      // blocks of data included in the dictionary since we only check limit
+      // after each block is built.
+      ASSERT_LE(TestGetTickerCount(options,
+                                   BLOCK_CACHE_COMPRESSION_DICT_BYTES_INSERT),
+                prev_compression_dict_bytes_inserted + 2 * kBlockLen);
+    }
   }
-  ASSERT_EQ(expected_compression_dict_misses,
-            TestGetTickerCount(options, BLOCK_CACHE_COMPRESSION_DICT_MISS));
 }
 
 TEST_P(PresetCompressionDictTest, CompactBottommost) {
   // Verifies that dictionary is generated and written during compaction to
   // non-bottommost level only when either `ColumnFamilyOptions::compression` or
-  // `ColumnFamilyOptions::bottommost_compression` enables dictionary.
+  // `ColumnFamilyOptions::bottommost_compression` enables dictionary. Also
+  // verifies the size of the dictionary is within expectations according to the
+  // limit on buffering set by `CompressionOptions::max_dict_buffer_bytes`.
   const size_t kValueLen = 256;
   const size_t kKeysPerFile = 1 << 10;
-  const size_t kDictLen = 4 << 10;
+  const size_t kDictLen = 16 << 10;
+  const size_t kBlockLen = 4 << 10;
 
   Options options = CurrentOptions();
   if (bottommost_) {
     options.bottommost_compression = compression_type_;
     options.bottommost_compression_opts.enabled = true;
     options.bottommost_compression_opts.max_dict_bytes = kDictLen;
+    options.bottommost_compression_opts.max_dict_buffer_bytes = kBlockLen;
   } else {
     options.compression = compression_type_;
     options.compression_opts.max_dict_bytes = kDictLen;
+    options.compression_opts.max_dict_buffer_bytes = kBlockLen;
   }
   options.disable_auto_compactions = true;
   options.statistics = CreateDBStatistics();
   BlockBasedTableOptions bbto;
+  bbto.block_size = kBlockLen;
   bbto.cache_index_and_filter_blocks = true;
   options.table_factory.reset(NewBlockBasedTableFactory(bbto));
   Reopen(options);
@@ -1550,17 +1595,28 @@ TEST_P(PresetCompressionDictTest, CompactBottommost) {
   ASSERT_EQ("2", FilesPerLevel(0));
 #endif  // ROCKSDB_LITE
 
-  uint64_t prev_compression_dict_misses =
-      TestGetTickerCount(options, BLOCK_CACHE_COMPRESSION_DICT_MISS);
+  uint64_t prev_compression_dict_bytes_inserted =
+      TestGetTickerCount(options, BLOCK_CACHE_COMPRESSION_DICT_BYTES_INSERT);
   CompactRangeOptions cro;
   ASSERT_OK(db_->CompactRange(cro, nullptr, nullptr));
 #ifndef ROCKSDB_LITE
   ASSERT_EQ("0,1", FilesPerLevel(0));
 #endif  // ROCKSDB_LITE
-  // If there's a compression dictionary, it should have been loaded when the
-  // compaction finished, incurring a cache miss.
-  ASSERT_EQ(prev_compression_dict_misses + 1,
-            TestGetTickerCount(options, BLOCK_CACHE_COMPRESSION_DICT_MISS));
+  ASSERT_GT(
+      TestGetTickerCount(options, BLOCK_CACHE_COMPRESSION_DICT_BYTES_INSERT),
+      prev_compression_dict_bytes_inserted);
+  // TODO(ajkr): fix the below assertion to work with ZSTD. The expectation on
+  // number of bytes needs to be adjusted in case the cached block is in ZSTD's
+  // digested dictionary format.
+  if (compression_type_ != kZSTD &&
+      compression_type_ != kZSTDNotFinalCompression) {
+    // Although we limited buffering to `kBlockLen`, there may be up to two
+    // blocks of data included in the dictionary since we only check limit after
+    // each block is built.
+    ASSERT_LE(
+        TestGetTickerCount(options, BLOCK_CACHE_COMPRESSION_DICT_BYTES_INSERT),
+        prev_compression_dict_bytes_inserted + 2 * kBlockLen);
+  }
 }
 
 class CompactionCompressionListener : public EventListener {
@@ -3683,20 +3739,26 @@ TEST_F(DBTest2, LiveFilesOmitObsoleteFiles) {
 
 TEST_F(DBTest2, TestNumPread) {
   Options options = CurrentOptions();
+  bool prefetch_supported =
+      test::IsPrefetchSupported(env_->GetFileSystem(), dbname_);
   // disable block cache
   BlockBasedTableOptions table_options;
   table_options.no_block_cache = true;
   options.table_factory.reset(NewBlockBasedTableFactory(table_options));
   Reopen(options);
   env_->count_random_reads_ = true;
-
   env_->random_file_open_counter_.store(0);
   ASSERT_OK(Put("bar", "foo"));
   ASSERT_OK(Put("foo", "bar"));
   ASSERT_OK(Flush());
-  // After flush, we'll open the file and read footer, meta block,
-  // property block and index block.
-  ASSERT_EQ(4, env_->random_read_counter_.Read());
+  if (prefetch_supported) {
+    // After flush, we'll open the file and read footer, meta block,
+    // property block and index block.
+    ASSERT_EQ(4, env_->random_read_counter_.Read());
+  } else {
+    // With prefetch not supported, we will do a single read into a buffer
+    ASSERT_EQ(1, env_->random_read_counter_.Read());
+  }
   ASSERT_EQ(1, env_->random_file_open_counter_.load());
 
   // One pread per a normal data block read
@@ -3712,19 +3774,30 @@ TEST_F(DBTest2, TestNumPread) {
   ASSERT_OK(Put("bar2", "foo2"));
   ASSERT_OK(Put("foo2", "bar2"));
   ASSERT_OK(Flush());
-  // After flush, we'll open the file and read footer, meta block,
-  // property block and index block.
-  ASSERT_EQ(4, env_->random_read_counter_.Read());
+  if (prefetch_supported) {
+    // After flush, we'll open the file and read footer, meta block,
+    // property block and index block.
+    ASSERT_EQ(4, env_->random_read_counter_.Read());
+  } else {
+    // With prefetch not supported, we will do a single read into a buffer
+    ASSERT_EQ(1, env_->random_read_counter_.Read());
+  }
   ASSERT_EQ(1, env_->random_file_open_counter_.load());
 
-  // Compaction needs two input blocks, which requires 2 preads, and
-  // generate a new SST file which needs 4 preads (footer, meta block,
-  // property block and index block). In total 6.
   env_->random_file_open_counter_.store(0);
   env_->random_read_counter_.Reset();
   ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
-  ASSERT_EQ(6, env_->random_read_counter_.Read());
-  // All compactin input files should have already been opened.
+  if (prefetch_supported) {
+    // Compaction needs two input blocks, which requires 2 preads, and
+    // generate a new SST file which needs 4 preads (footer, meta block,
+    // property block and index block). In total 6.
+    ASSERT_EQ(6, env_->random_read_counter_.Read());
+  } else {
+    // With prefetch off, compaction needs two input blocks,
+    // followed by a single buffered read.  In total 3.
+    ASSERT_EQ(3, env_->random_read_counter_.Read());
+  }
+  // All compaction input files should have already been opened.
   ASSERT_EQ(1, env_->random_file_open_counter_.load());
 
   // One pread per a normal data block read

db/db_test_util.cc

@@ -200,28 +200,28 @@ bool DBTestBase::ChangeCompactOptions() {
     Destroy(last_options_);
     auto options = CurrentOptions();
     options.create_if_missing = true;
-    TryReopen(options);
+    Reopen(options);
     return true;
   } else if (option_config_ == kUniversalCompaction) {
     option_config_ = kUniversalCompactionMultiLevel;
     Destroy(last_options_);
     auto options = CurrentOptions();
     options.create_if_missing = true;
-    TryReopen(options);
+    Reopen(options);
     return true;
   } else if (option_config_ == kUniversalCompactionMultiLevel) {
     option_config_ = kLevelSubcompactions;
     Destroy(last_options_);
     auto options = CurrentOptions();
     assert(options.max_subcompactions > 1);
-    TryReopen(options);
+    Reopen(options);
     return true;
   } else if (option_config_ == kLevelSubcompactions) {
     option_config_ = kUniversalSubcompactions;
     Destroy(last_options_);
     auto options = CurrentOptions();
     assert(options.max_subcompactions > 1);
-    TryReopen(options);
+    Reopen(options);
     return true;
   } else {
     return false;
@@ -236,7 +236,7 @@ bool DBTestBase::ChangeWalOptions() {
     auto options = CurrentOptions();
     Destroy(options);
     options.create_if_missing = true;
-    TryReopen(options);
+    Reopen(options);
     return true;
   } else if (option_config_ == kDBLogDir) {
     option_config_ = kWalDirAndMmapReads;
@@ -244,14 +244,14 @@ bool DBTestBase::ChangeWalOptions() {
     auto options = CurrentOptions();
     Destroy(options);
     options.create_if_missing = true;
-    TryReopen(options);
+    Reopen(options);
     return true;
   } else if (option_config_ == kWalDirAndMmapReads) {
     option_config_ = kRecycleLogFiles;
     Destroy(last_options_);
     auto options = CurrentOptions();
     Destroy(options);
-    TryReopen(options);
+    Reopen(options);
     return true;
   } else {
     return false;
@@ -680,7 +680,7 @@ void DBTestBase::Close() {
 void DBTestBase::DestroyAndReopen(const Options& options) {
   // Destroy using last options
   Destroy(last_options_);
-  ASSERT_OK(TryReopen(options));
+  Reopen(options);
 }
 
 void DBTestBase::Destroy(const Options& options, bool delete_cf_paths) {
@@ -1128,26 +1128,48 @@ std::string DBTestBase::FilesPerLevel(int cf) {
 #endif  // !ROCKSDB_LITE
 
 size_t DBTestBase::CountFiles() {
+  size_t count = 0;
   std::vector<std::string> files;
-  env_->GetChildren(dbname_, &files);
-
-  std::vector<std::string> logfiles;
-  if (dbname_ != last_options_.wal_dir) {
-    env_->GetChildren(last_options_.wal_dir, &logfiles);
+  if (env_->GetChildren(dbname_, &files).ok()) {
+    count += files.size();
   }
 
-  return files.size() + logfiles.size();
+  if (dbname_ != last_options_.wal_dir) {
+    if (env_->GetChildren(last_options_.wal_dir, &files).ok()) {
+      count += files.size();
+    }
+  }
+
+  return count;
+};
+
+Status DBTestBase::CountFiles(size_t* count) {
+  std::vector<std::string> files;
+  Status s = env_->GetChildren(dbname_, &files);
+  if (!s.ok()) {
+    return s;
+  }
+  size_t files_count = files.size();
+
+  if (dbname_ != last_options_.wal_dir) {
+    s = env_->GetChildren(last_options_.wal_dir, &files);
+    if (!s.ok()) {
+      return s;
+    }
+    *count = files_count + files.size();
+  }
+
+  return Status::OK();
 }
 
-uint64_t DBTestBase::Size(const Slice& start, const Slice& limit, int cf) {
+Status DBTestBase::Size(const Slice& start, const Slice& limit, int cf,
+                        uint64_t* size) {
   Range r(start, limit);
-  uint64_t size;
   if (cf == 0) {
-    db_->GetApproximateSizes(&r, 1, &size);
+    return db_->GetApproximateSizes(&r, 1, size);
   } else {
-    db_->GetApproximateSizes(handles_[1], &r, 1, &size);
+    return db_->GetApproximateSizes(handles_[1], &r, 1, size);
   }
-  return size;
 }
 
 void DBTestBase::Compact(int cf, const Slice& start, const Slice& limit,
@@ -1266,8 +1288,8 @@ void DBTestBase::GenerateNewRandomFile(Random* rnd, bool nowait) {
   }
   ASSERT_OK(Put("key" + rnd->RandomString(7), rnd->RandomString(200)));
   if (!nowait) {
-    dbfull()->TEST_WaitForFlushMemTable();
-    dbfull()->TEST_WaitForCompact();
+    ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable());
+    ASSERT_OK(dbfull()->TEST_WaitForCompact());
   }
 }
 

db/db_test_util.h

@@ -1070,7 +1070,13 @@ class DBTestBase : public testing::Test {
 
   size_t CountFiles();
 
-  uint64_t Size(const Slice& start, const Slice& limit, int cf = 0);
+  Status CountFiles(size_t* count);
+
+  Status Size(const Slice& start, const Slice& limit, uint64_t* size) {
+    return Size(start, limit, 0, size);
+  }
+
+  Status Size(const Slice& start, const Slice& limit, int cf, uint64_t* size);
 
   void Compact(int cf, const Slice& start, const Slice& limit,
                uint32_t target_path_id);

db/db_universal_compaction_test.cc

@@ -996,7 +996,7 @@ TEST_P(DBTestUniversalCompaction, UniversalCompactionStopStyleSimilarSize) {
   //   First, clean up memtable before inserting new data. This will generate
   //   a level-0 file, with size around 0.4 (according to previously written
   //   data amount).
-  dbfull()->Flush(FlushOptions());
+  ASSERT_OK(dbfull()->Flush(FlushOptions()));
   for (int num = 0; num < options.level0_file_num_compaction_trigger - 3;
        num++) {
     // Write 110KB (11 values, each 10K)
@@ -1781,7 +1781,7 @@ TEST_P(DBTestUniversalCompaction, FinalSortedRunCompactFilesConflict) {
       dbfull()->TEST_write_controler().GetCompactionPressureToken();
 
   ASSERT_OK(Put("key", "val"));
-  Flush();
+  ASSERT_OK(Flush());
   ASSERT_OK(dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr));
   ASSERT_EQ(NumTableFilesAtLevel(num_levels_ - 1), 1);
   ColumnFamilyMetaData cf_meta;
@@ -1807,7 +1807,7 @@ TEST_P(DBTestUniversalCompaction, FinalSortedRunCompactFilesConflict) {
       "DBTestUniversalCompaction:FinalSortedRunCompactFilesConflict:0");
   for (int i = 0; i < 2; ++i) {
     ASSERT_OK(Put("key", "val"));
-    Flush();
+    ASSERT_OK(Flush());
   }
   ASSERT_OK(dbfull()->TEST_WaitForCompact());
 
@@ -1911,7 +1911,7 @@ TEST_F(DBTestUniversalCompaction2, BasicL0toL1) {
   for (i = 0; i < 2000; ++i) {
     ASSERT_OK(Put(Key(i), "val"));
   }
-  Flush();
+  ASSERT_OK(Flush());
   //  MoveFilesToLevel(6);
   ASSERT_OK(dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr));
 
@@ -1923,7 +1923,7 @@ TEST_F(DBTestUniversalCompaction2, BasicL0toL1) {
       ASSERT_OK(Put(Key(i), "val"));
     }
   }
-  Flush();
+  ASSERT_OK(Flush());
 
   ASSERT_OK(dbfull()->TEST_WaitForCompact());
   ASSERT_EQ(0, NumTableFilesAtLevel(0));
@@ -1954,7 +1954,7 @@ TEST_F(DBTestUniversalCompaction2, SingleLevel) {
   for (i = 0; i < 2000; ++i) {
     ASSERT_OK(Put(Key(i), "val"));
   }
-  Flush();
+  ASSERT_OK(Flush());
 
   for (i = 1999; i < kNumKeys; ++i) {
     if (i >= kNumKeys - kWindowSize &&
@@ -1964,7 +1964,7 @@ TEST_F(DBTestUniversalCompaction2, SingleLevel) {
       ASSERT_OK(Put(Key(i), "val"));
     }
   }
-  Flush();
+  ASSERT_OK(Flush()(;
 
   ASSERT_OK(dbfull()->TEST_WaitForCompact());
   ASSERT_EQ(1, NumTableFilesAtLevel(0));
@@ -1992,19 +1992,19 @@ TEST_F(DBTestUniversalCompaction2, MultipleLevels) {
   for (i = 0; i < 500; ++i) {
     ASSERT_OK(Put(Key(i), "val"));
   }
-  Flush();
+  ASSERT_OK(Flush());
   for (i = 500; i < 1000; ++i) {
     ASSERT_OK(Put(Key(i), "val"));
   }
-  Flush();
+  ASSERT_OK(Flush());
   for (i = 1000; i < 1500; ++i) {
     ASSERT_OK(Put(Key(i), "val"));
   }
-  Flush();
+  ASSERT_OK(Flush());
   for (i = 1500; i < 2000; ++i) {
     ASSERT_OK(Put(Key(i), "val"));
   }
-  Flush();
+  ASSERT_OK(Flush());
 
   ASSERT_OK(dbfull()->TEST_WaitForCompact());
   ASSERT_EQ(0, NumTableFilesAtLevel(0));
@@ -2013,15 +2013,15 @@ TEST_F(DBTestUniversalCompaction2, MultipleLevels) {
   for (i = 1999; i < 2333; ++i) {
     ASSERT_OK(Put(Key(i), "val"));
   }
-  Flush();
+  ASSERT_OK(Flush());
   for (i = 2333; i < 2666; ++i) {
     ASSERT_OK(Put(Key(i), "val"));
   }
-  Flush();
+  ASSERT_OK(Flush());
   for (i = 2666; i < 2999; ++i) {
     ASSERT_OK(Put(Key(i), "val"));
   }
-  Flush();
+  ASSERT_OK(Flush());
 
   ASSERT_OK(dbfull()->TEST_WaitForCompact());
   ASSERT_EQ(0, NumTableFilesAtLevel(0));
@@ -2031,7 +2031,7 @@ TEST_F(DBTestUniversalCompaction2, MultipleLevels) {
   for (i = 1900; i < 2100; ++i) {
     ASSERT_OK(Delete(Key(i)));
   }
-  Flush();
+  ASSERT_OK(Flush());
 
   ASSERT_OK(dbfull()->TEST_WaitForCompact());
   ASSERT_EQ(0, NumTableFilesAtLevel(0));
@@ -2064,19 +2064,19 @@ TEST_F(DBTestUniversalCompaction2, OverlappingL0) {
   for (i = 0; i < 2000; ++i) {
     ASSERT_OK(Put(Key(i), "val"));
   }
-  Flush();
+  ASSERT_OK(Flush());
   for (i = 2000; i < 3000; ++i) {
     ASSERT_OK(Put(Key(i), "val"));
   }
-  Flush();
+  ASSERT_OK(Flush());
   for (i = 3500; i < 4000; ++i) {
     ASSERT_OK(Put(Key(i), "val"));
   }
-  Flush();
+  ASSERT_OK(Flush());
   for (i = 2900; i < 3100; ++i) {
     ASSERT_OK(Delete(Key(i)));
   }
-  Flush();
+  ASSERT_OK(Flush());
 
   ASSERT_OK(dbfull()->TEST_WaitForCompact());
   ASSERT_EQ(2, NumTableFilesAtLevel(0));
@@ -2106,7 +2106,7 @@ TEST_F(DBTestUniversalCompaction2, IngestBehind) {
   for (i = 0; i < 2000; ++i) {
     ASSERT_OK(Put(Key(i), "val"));
   }
-  Flush();
+  ASSERT_OK(Flush());
   //  MoveFilesToLevel(6);
   ASSERT_OK(dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr));
 
@@ -2118,7 +2118,7 @@ TEST_F(DBTestUniversalCompaction2, IngestBehind) {
       ASSERT_OK(Put(Key(i), "val"));
     }
   }
-  Flush();
+  ASSERT_OK(Flush());
 
   ASSERT_OK(dbfull()->TEST_WaitForCompact());
   ASSERT_EQ(0, NumTableFilesAtLevel(0));
@@ -2184,7 +2184,7 @@ TEST_F(DBTestUniversalCompaction2, PeriodicCompaction) {
 
   // Case 1: Oldest flushed file excceeds periodic compaction threshold.
   ASSERT_OK(Put("foo", "bar"));
-  Flush();
+  ASSERT_OK(Flush());
   ASSERT_EQ(0, periodic_compactions);
   // Move clock forward so that the flushed file would qualify periodic
   // compaction.
@@ -2192,7 +2192,7 @@ TEST_F(DBTestUniversalCompaction2, PeriodicCompaction) {
 
   // Another flush would trigger compaction the oldest file.
   ASSERT_OK(Put("foo", "bar2"));
-  Flush();
+  ASSERT_OK(Flush());
   ASSERT_OK(dbfull()->TEST_WaitForCompact());
 
   ASSERT_EQ(1, periodic_compactions);
@@ -2203,7 +2203,7 @@ TEST_F(DBTestUniversalCompaction2, PeriodicCompaction) {
   periodic_compactions = 0;
   // A flush doesn't trigger a periodic compaction when threshold not hit
   ASSERT_OK(Put("foo", "bar2"));
-  Flush();
+  ASSERT_OK(Flush());
   ASSERT_OK(dbfull()->TEST_WaitForCompact());
   ASSERT_EQ(0, periodic_compactions);
 
@@ -2211,7 +2211,7 @@ TEST_F(DBTestUniversalCompaction2, PeriodicCompaction) {
   // a compaction
   ASSERT_OK(Put("foo", "bar2"));
   env_->MockSleepForSeconds(48 * 60 * 60 + 100);
-  Flush();
+  ASSERT_OK(Flush());
   ASSERT_OK(dbfull()->TEST_WaitForCompact());
   ASSERT_EQ(1, periodic_compactions);
   ASSERT_EQ(0, start_level);

db/db_wal_test.cc

@@ -14,6 +14,7 @@
 #include "port/stack_trace.h"
 #include "test_util/sync_point.h"
 #include "utilities/fault_injection_env.h"
+#include "utilities/fault_injection_fs.h"
 
 namespace ROCKSDB_NAMESPACE {
 class DBWALTestBase : public DBTestBase {
@@ -358,16 +359,16 @@ TEST_F(DBWALTest, RecoverWithBlob) {
   // There should be no files just yet since we haven't flushed.
   {
     VersionSet* const versions = dbfull()->TEST_GetVersionSet();
-    assert(versions);
+    ASSERT_NE(versions, nullptr);
 
     ColumnFamilyData* const cfd = versions->GetColumnFamilySet()->GetDefault();
-    assert(cfd);
+    ASSERT_NE(cfd, nullptr);
 
     Version* const current = cfd->current();
-    assert(current);
+    ASSERT_NE(current, nullptr);
 
     const VersionStorageInfo* const storage_info = current->storage_info();
-    assert(storage_info);
+    ASSERT_NE(storage_info, nullptr);
 
     ASSERT_EQ(storage_info->num_non_empty_levels(), 0);
     ASSERT_TRUE(storage_info->GetBlobFiles().empty());
@@ -388,28 +389,28 @@ TEST_F(DBWALTest, RecoverWithBlob) {
   ASSERT_EQ(Get("key2"), long_value);
 
   VersionSet* const versions = dbfull()->TEST_GetVersionSet();
-  assert(versions);
+  ASSERT_NE(versions, nullptr);
 
   ColumnFamilyData* const cfd = versions->GetColumnFamilySet()->GetDefault();
-  assert(cfd);
+  ASSERT_NE(cfd, nullptr);
 
   Version* const current = cfd->current();
-  assert(current);
+  ASSERT_NE(current, nullptr);
 
   const VersionStorageInfo* const storage_info = current->storage_info();
-  assert(storage_info);
+  ASSERT_NE(storage_info, nullptr);
 
   const auto& l0_files = storage_info->LevelFiles(0);
   ASSERT_EQ(l0_files.size(), 1);
 
   const FileMetaData* const table_file = l0_files[0];
-  assert(table_file);
+  ASSERT_NE(table_file, nullptr);
 
   const auto& blob_files = storage_info->GetBlobFiles();
   ASSERT_EQ(blob_files.size(), 1);
 
   const auto& blob_file = blob_files.begin()->second;
-  assert(blob_file);
+  ASSERT_NE(blob_file, nullptr);
 
   ASSERT_EQ(table_file->smallest.user_key(), "key1");
   ASSERT_EQ(table_file->largest.user_key(), "key2");
@@ -422,7 +423,7 @@ TEST_F(DBWALTest, RecoverWithBlob) {
 
 #ifndef ROCKSDB_LITE
   const InternalStats* const internal_stats = cfd->internal_stats();
-  assert(internal_stats);
+  ASSERT_NE(internal_stats, nullptr);
 
   const uint64_t expected_bytes =
       table_file->fd.GetFileSize() + blob_file->GetTotalBlobBytes();
@@ -437,15 +438,159 @@ TEST_F(DBWALTest, RecoverWithBlob) {
 #endif  // ROCKSDB_LITE
 }
 
+TEST_F(DBWALTest, RecoverWithBlobMultiSST) {
+  // Write several large (4 KB) values without flushing. Note that blob files
+  // are not actually enabled at this point.
+  std::string large_value(1 << 12, 'a');
+
+  constexpr int num_keys = 64;
+
+  for (int i = 0; i < num_keys; ++i) {
+    ASSERT_OK(Put(Key(i), large_value));
+  }
+
+  // There should be no files just yet since we haven't flushed.
+  {
+    VersionSet* const versions = dbfull()->TEST_GetVersionSet();
+    ASSERT_NE(versions, nullptr);
+
+    ColumnFamilyData* const cfd = versions->GetColumnFamilySet()->GetDefault();
+    ASSERT_NE(cfd, nullptr);
+
+    Version* const current = cfd->current();
+    ASSERT_NE(current, nullptr);
+
+    const VersionStorageInfo* const storage_info = current->storage_info();
+    ASSERT_NE(storage_info, nullptr);
+
+    ASSERT_EQ(storage_info->num_non_empty_levels(), 0);
+    ASSERT_TRUE(storage_info->GetBlobFiles().empty());
+  }
+
+  // Reopen the database with blob files enabled and write buffer size set to a
+  // smaller value. Multiple table files+blob files should be written and added
+  // to the Version during recovery.
+  Options options;
+  options.write_buffer_size = 1 << 16;  // 64 KB
+  options.enable_blob_files = true;
+  options.avoid_flush_during_recovery = false;
+  options.disable_auto_compactions = true;
+  options.env = env_;
+
+  Reopen(options);
+
+  for (int i = 0; i < num_keys; ++i) {
+    ASSERT_EQ(Get(Key(i)), large_value);
+  }
+
+  VersionSet* const versions = dbfull()->TEST_GetVersionSet();
+  ASSERT_NE(versions, nullptr);
+
+  ColumnFamilyData* const cfd = versions->GetColumnFamilySet()->GetDefault();
+  ASSERT_NE(cfd, nullptr);
+
+  Version* const current = cfd->current();
+  ASSERT_NE(current, nullptr);
+
+  const VersionStorageInfo* const storage_info = current->storage_info();
+  ASSERT_NE(storage_info, nullptr);
+
+  const auto& l0_files = storage_info->LevelFiles(0);
+  ASSERT_GT(l0_files.size(), 1);
+
+  const auto& blob_files = storage_info->GetBlobFiles();
+  ASSERT_GT(blob_files.size(), 1);
+
+  ASSERT_EQ(l0_files.size(), blob_files.size());
+}
+
+TEST_F(DBWALTest, WALWithChecksumHandoff) {
+#ifndef ROCKSDB_ASSERT_STATUS_CHECKED
+  if (mem_env_ || encrypted_env_) {
+    ROCKSDB_GTEST_SKIP("Test requires non-mem or non-encrypted environment");
+    return;
+  }
+  std::shared_ptr<FaultInjectionTestFS> fault_fs(
+      new FaultInjectionTestFS(FileSystem::Default()));
+  std::unique_ptr<Env> fault_fs_env(NewCompositeEnv(fault_fs));
+  do {
+    Options options = CurrentOptions();
+
+    options.checksum_handoff_file_types.Add(FileType::kWalFile);
+    options.env = fault_fs_env.get();
+    fault_fs->SetChecksumHandoffFuncType(ChecksumType::kCRC32c);
+
+    CreateAndReopenWithCF({"pikachu"}, options);
+    WriteOptions writeOpt = WriteOptions();
+    writeOpt.disableWAL = true;
+    ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "foo", "v1"));
+    ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "bar", "v1"));
+
+    ReopenWithColumnFamilies({"default", "pikachu"}, options);
+    ASSERT_EQ("v1", Get(1, "foo"));
+    ASSERT_EQ("v1", Get(1, "bar"));
+
+    writeOpt.disableWAL = false;
+    ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "bar", "v2"));
+    writeOpt.disableWAL = true;
+    ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "foo", "v2"));
+
+    ReopenWithColumnFamilies({"default", "pikachu"}, options);
+    // Both value's should be present.
+    ASSERT_EQ("v2", Get(1, "bar"));
+    ASSERT_EQ("v2", Get(1, "foo"));
+
+    writeOpt.disableWAL = true;
+    // This put, data is persisted by Flush
+    ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "bar", "v3"));
+    ReopenWithColumnFamilies({"default", "pikachu"}, options);
+    writeOpt.disableWAL = false;
+    // Data is persisted in the WAL
+    ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "zoo", "v3"));
+    // The hash does not match, write fails
+    fault_fs->SetChecksumHandoffFuncType(ChecksumType::kxxHash);
+    writeOpt.disableWAL = false;
+    ASSERT_NOK(dbfull()->Put(writeOpt, handles_[1], "foo", "v3"));
+
+    ReopenWithColumnFamilies({"default", "pikachu"}, options);
+    // Due to the write failure, Get should not find
+    ASSERT_NE("v3", Get(1, "foo"));
+    ASSERT_EQ("v3", Get(1, "zoo"));
+    ASSERT_EQ("v3", Get(1, "bar"));
+
+    fault_fs->SetChecksumHandoffFuncType(ChecksumType::kCRC32c);
+    // Each write will be similated as corrupted.
+    fault_fs->IngestDataCorruptionBeforeWrite();
+    writeOpt.disableWAL = true;
+    ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "bar", "v4"));
+    writeOpt.disableWAL = false;
+    ASSERT_NOK(dbfull()->Put(writeOpt, handles_[1], "foo", "v4"));
+    ReopenWithColumnFamilies({"default", "pikachu"}, options);
+    ASSERT_NE("v4", Get(1, "foo"));
+    ASSERT_NE("v4", Get(1, "bar"));
+    fault_fs->NoDataCorruptionBeforeWrite();
+
+    fault_fs->SetChecksumHandoffFuncType(ChecksumType::kNoChecksum);
+    // The file system does not provide checksum method and verification.
+    writeOpt.disableWAL = true;
+    ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "bar", "v5"));
+    writeOpt.disableWAL = false;
+    ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "foo", "v5"));
+    ReopenWithColumnFamilies({"default", "pikachu"}, options);
+    ASSERT_EQ("v5", Get(1, "foo"));
+    ASSERT_EQ("v5", Get(1, "bar"));
+
+    Destroy(options);
+  } while (ChangeWalOptions());
+#endif  // ROCKSDB_ASSERT_STATUS_CHECKED
+}
+
 class DBRecoveryTestBlobError
     : public DBWALTest,
       public testing::WithParamInterface<std::string> {
  public:
-  DBRecoveryTestBlobError()
-      : fault_injection_env_(env_), sync_point_(GetParam()) {}
-  ~DBRecoveryTestBlobError() { Close(); }
+  DBRecoveryTestBlobError() : sync_point_(GetParam()) {}
 
-  FaultInjectionTestEnv fault_injection_env_;
   std::string sync_point_;
 };
 
@@ -460,21 +605,19 @@ TEST_P(DBRecoveryTestBlobError, RecoverWithBlobError) {
 
   // Reopen with blob files enabled but make blob file writing fail during
   // recovery.
-  SyncPoint::GetInstance()->SetCallBack(sync_point_, [this](void* /* arg */) {
-    fault_injection_env_.SetFilesystemActive(false,
-                                             Status::IOError(sync_point_));
+  SyncPoint::GetInstance()->SetCallBack(sync_point_, [this](void* arg) {
+    Status* const s = static_cast<Status*>(arg);
+    assert(s);
+
+    (*s) = Status::IOError(sync_point_);
   });
-  SyncPoint::GetInstance()->SetCallBack(
-      "BuildTable:BeforeDeleteFile", [this](void* /* arg */) {
-        fault_injection_env_.SetFilesystemActive(true);
-      });
   SyncPoint::GetInstance()->EnableProcessing();
 
   Options options;
   options.enable_blob_files = true;
   options.avoid_flush_during_recovery = false;
   options.disable_auto_compactions = true;
-  options.env = &fault_injection_env_;
+  options.env = env_;
 
   ASSERT_NOK(TryReopen(options));
 
@@ -502,13 +645,11 @@ TEST_F(DBWALTest, IgnoreRecoveredLog) {
 
   do {
     // delete old files in backup_logs directory
-    env_->CreateDirIfMissing(backup_logs);
+    ASSERT_OK(env_->CreateDirIfMissing(backup_logs));
     std::vector<std::string> old_files;
-    env_->GetChildren(backup_logs, &old_files);
+    ASSERT_OK(env_->GetChildren(backup_logs, &old_files));
     for (auto& file : old_files) {
-      if (file != "." && file != "..") {
-        env_->DeleteFile(backup_logs + "/" + file);
-      }
+      ASSERT_OK(env_->DeleteFile(backup_logs + "/" + file));
     }
     Options options = CurrentOptions();
     options.create_if_missing = true;
@@ -526,11 +667,9 @@ TEST_F(DBWALTest, IgnoreRecoveredLog) {
 
     // copy the logs to backup
     std::vector<std::string> logs;
-    env_->GetChildren(options.wal_dir, &logs);
+    ASSERT_OK(env_->GetChildren(options.wal_dir, &logs));
     for (auto& log : logs) {
-      if (log != ".." && log != ".") {
-        CopyFile(options.wal_dir + "/" + log, backup_logs + "/" + log);
-      }
+      CopyFile(options.wal_dir + "/" + log, backup_logs + "/" + log);
     }
 
     // recover the DB
@@ -541,9 +680,7 @@ TEST_F(DBWALTest, IgnoreRecoveredLog) {
 
     // copy the logs from backup back to wal dir
     for (auto& log : logs) {
-      if (log != ".." && log != ".") {
-        CopyFile(backup_logs + "/" + log, options.wal_dir + "/" + log);
-      }
+      CopyFile(backup_logs + "/" + log, options.wal_dir + "/" + log);
     }
     // this should ignore the log files, recovery should not happen again
     // if the recovery happens, the same merge operator would be called twice,
@@ -557,11 +694,9 @@ TEST_F(DBWALTest, IgnoreRecoveredLog) {
     Close();
 
     // copy the logs from backup back to wal dir
-    env_->CreateDirIfMissing(options.wal_dir);
+    ASSERT_OK(env_->CreateDirIfMissing(options.wal_dir));
     for (auto& log : logs) {
-      if (log != ".." && log != ".") {
-        CopyFile(backup_logs + "/" + log, options.wal_dir + "/" + log);
-      }
+      CopyFile(backup_logs + "/" + log, options.wal_dir + "/" + log);
     }
     // assert that we successfully recovered only from logs, even though we
     // destroyed the DB
@@ -572,16 +707,14 @@ TEST_F(DBWALTest, IgnoreRecoveredLog) {
     // Recovery will fail if DB directory doesn't exist.
     Destroy(options);
     // copy the logs from backup back to wal dir
-    env_->CreateDirIfMissing(options.wal_dir);
+    ASSERT_OK(env_->CreateDirIfMissing(options.wal_dir));
     for (auto& log : logs) {
-      if (log != ".." && log != ".") {
-        CopyFile(backup_logs + "/" + log, options.wal_dir + "/" + log);
-        // we won't be needing this file no more
-        env_->DeleteFile(backup_logs + "/" + log);
-      }
+      CopyFile(backup_logs + "/" + log, options.wal_dir + "/" + log);
+      // we won't be needing this file no more
+      ASSERT_OK(env_->DeleteFile(backup_logs + "/" + log));
     }
     Status s = TryReopen(options);
-    ASSERT_TRUE(!s.ok());
+    ASSERT_NOK(s);
     Destroy(options);
   } while (ChangeWalOptions());
 }
@@ -619,9 +752,9 @@ TEST_F(DBWALTest, PreallocateBlock) {
         called.fetch_add(1);
       });
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
-  Put("", "");
-  Flush();
-  Put("", "");
+  ASSERT_OK(Put("", ""));
+  ASSERT_OK(Flush());
+  ASSERT_OK(Put("", ""));
   Close();
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
   ASSERT_EQ(2, called.load());
@@ -638,9 +771,9 @@ TEST_F(DBWALTest, PreallocateBlock) {
         called.fetch_add(1);
       });
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
-  Put("", "");
-  Flush();
-  Put("", "");
+  ASSERT_OK(Put("", ""));
+  ASSERT_OK(Flush());
+  ASSERT_OK(Put("", ""));
   Close();
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
   ASSERT_EQ(2, called.load());
@@ -658,9 +791,9 @@ TEST_F(DBWALTest, PreallocateBlock) {
         called.fetch_add(1);
       });
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
-  Put("", "");
-  Flush();
-  Put("", "");
+  ASSERT_OK(Put("", ""));
+  ASSERT_OK(Flush());
+  ASSERT_OK(Put("", ""));
   Close();
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
   ASSERT_EQ(2, called.load());
@@ -679,9 +812,9 @@ TEST_F(DBWALTest, PreallocateBlock) {
         called.fetch_add(1);
       });
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
-  Put("", "");
-  Flush();
-  Put("", "");
+  ASSERT_OK(Put("", ""));
+  ASSERT_OK(Flush());
+  ASSERT_OK(Put("", ""));
   Close();
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
   ASSERT_EQ(2, called.load());
@@ -907,7 +1040,7 @@ TEST_F(DBWALTest, RecoverCheckFileAmountWithSmallWriteBuffer) {
   // Make 'dobrynia' to be flushed and new WAL file to be created
   ASSERT_OK(Put(2, Key(10), DummyString(7500000)));
   ASSERT_OK(Put(2, Key(1), DummyString(1)));
-  dbfull()->TEST_WaitForFlushMemTable(handles_[2]);
+  ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable(handles_[2]));
   {
     auto tables = ListTableFiles(env_, dbname_);
     ASSERT_EQ(tables.size(), static_cast<size_t>(1));
@@ -961,7 +1094,7 @@ TEST_F(DBWALTest, RecoverCheckFileAmount) {
   // Make 'nikitich' memtable to be flushed
   ASSERT_OK(Put(3, Key(10), DummyString(1002400)));
   ASSERT_OK(Put(3, Key(1), DummyString(1)));
-  dbfull()->TEST_WaitForFlushMemTable(handles_[3]);
+  ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable(handles_[3]));
   // 4 memtable are not flushed, 1 sst file
   {
     auto tables = ListTableFiles(env_, dbname_);
@@ -981,7 +1114,7 @@ TEST_F(DBWALTest, RecoverCheckFileAmount) {
   ASSERT_OK(Put(3, Key(10), DummyString(1002400)));
   // make it flush
   ASSERT_OK(Put(3, Key(1), DummyString(1)));
-  dbfull()->TEST_WaitForFlushMemTable(handles_[3]);
+  ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable(handles_[3]));
   // There are still 4 memtable not flushed, and 2 sst tables
   ASSERT_OK(Put(0, Key(1), DummyString(1)));
   ASSERT_OK(Put(1, Key(1), DummyString(1)));
@@ -1029,10 +1162,10 @@ TEST_F(DBWALTest, SyncMultipleLogs) {
   for (uint64_t b = 0; b < kNumBatches; b++) {
     batch.Clear();
     for (int i = 0; i < kBatchSize; i++) {
-      batch.Put(Key(i), DummyString(128));
+      ASSERT_OK(batch.Put(Key(i), DummyString(128)));
     }
 
-    dbfull()->Write(wo, &batch);
+    ASSERT_OK(dbfull()->Write(wo, &batch));
   }
 
   ASSERT_OK(dbfull()->SyncWAL());
@@ -1060,7 +1193,7 @@ TEST_F(DBWALTest, PartOfWritesWithWALDisabled) {
   ASSERT_OK(Flush(0));
   ASSERT_OK(Put(0, "key", "v5", wal_on));  // seq id 5
   ASSERT_EQ("v5", Get(0, "key"));
-  dbfull()->FlushWAL(false);
+  ASSERT_OK(dbfull()->FlushWAL(false));
   // Simulate a crash.
   fault_env->SetFilesystemActive(false);
   Close();
@@ -1128,12 +1261,13 @@ class RecoveryTestHelper {
       for (int i = 0; i < kKeysPerWALFile; i++) {
         std::string key = "key" + ToString((*count)++);
         std::string value = test->DummyString(kValueSize);
-        assert(current_log_writer.get() != nullptr);
+        ASSERT_NE(current_log_writer.get(), nullptr);
         uint64_t seq = versions->LastSequence() + 1;
         batch.Clear();
-        batch.Put(key, value);
+        ASSERT_OK(batch.Put(key, value));
         WriteBatchInternal::SetSequence(&batch, seq);
-        current_log_writer->AddRecord(WriteBatchInternal::Contents(&batch));
+        ASSERT_OK(current_log_writer->AddRecord(
+            WriteBatchInternal::Contents(&batch)));
         versions->SetLastAllocatedSequence(seq);
         versions->SetLastPublishedSequence(seq);
         versions->SetLastSequence(seq);
@@ -1309,10 +1443,11 @@ TEST_F(DBWALTest, kPointInTimeRecoveryCFConsistency) {
 
   ASSERT_OK(Put(1, "key3", "val3"));
   // Corrupt WAL at location of key3
-  test::CorruptFile(env, fname, static_cast<int>(offset_to_corrupt), 4, false);
+  ASSERT_OK(test::CorruptFile(env, fname, static_cast<int>(offset_to_corrupt),
+                              4, false));
   ASSERT_OK(Put(2, "key4", "val4"));
   ASSERT_OK(Put(1, "key5", "val5"));
-  Flush(2);
+  ASSERT_OK(Flush(2));
 
   // PIT recovery & verify
   options.wal_recovery_mode = WALRecoveryMode::kPointInTimeRecovery;
@@ -1466,7 +1601,7 @@ TEST_F(DBWALTest, WalCleanupAfterAvoidFlushDuringRecovery) {
   for (int i = 0; i < 2; ++i) {
     if (i > 0) {
       // Flush() triggers deletion of obsolete tracked files
-      Flush();
+      ASSERT_OK(Flush());
     }
     VectorLogPtr log_files;
     ASSERT_OK(dbfull()->GetSortedWalFiles(log_files));
@@ -1508,7 +1643,7 @@ TEST_F(DBWALTest, RecoverWithoutFlush) {
   ASSERT_EQ(Get("foo"), "foo_v2");
   ASSERT_EQ(Get("bar"), "bar_v2");
   // manual flush and insert again
-  Flush();
+  ASSERT_OK(Flush());
   ASSERT_EQ(Get("foo"), "foo_v2");
   ASSERT_EQ(Get("bar"), "bar_v2");
   ASSERT_OK(Put("foo", "foo_v3"));
@@ -1529,7 +1664,9 @@ TEST_F(DBWALTest, RecoverWithoutFlushMultipleCF) {
 
   auto countWalFiles = [this]() {
     VectorLogPtr log_files;
-    dbfull()->GetSortedWalFiles(log_files);
+    if (!dbfull()->GetSortedWalFiles(log_files).ok()) {
+      return size_t{0};
+    }
     return log_files.size();
   };
 
@@ -1537,11 +1674,11 @@ TEST_F(DBWALTest, RecoverWithoutFlushMultipleCF) {
   CreateAndReopenWithCF({"one", "two"}, options);
   ASSERT_OK(Put(0, "key1", kSmallValue));
   ASSERT_OK(Put(1, "key2", kLargeValue));
-  Flush(1);
+  ASSERT_OK(Flush(1));
   ASSERT_EQ(1, countWalFiles());
   ASSERT_OK(Put(0, "key3", kSmallValue));
   ASSERT_OK(Put(2, "key4", kLargeValue));
-  Flush(2);
+  ASSERT_OK(Flush(2));
   ASSERT_EQ(2, countWalFiles());
 
   // Reopen, insert and flush.
@@ -1555,9 +1692,9 @@ TEST_F(DBWALTest, RecoverWithoutFlushMultipleCF) {
   ASSERT_OK(Put(0, "key5", kLargeValue));
   ASSERT_OK(Put(1, "key6", kLargeValue));
   ASSERT_EQ(3, countWalFiles());
-  Flush(1);
+  ASSERT_OK(Flush(1));
   ASSERT_OK(Put(2, "key7", kLargeValue));
-  dbfull()->FlushWAL(false);
+  ASSERT_OK(dbfull()->FlushWAL(false));
   ASSERT_EQ(4, countWalFiles());
 
   // Reopen twice and validate.
@@ -1646,7 +1783,7 @@ TEST_F(DBWALTest, RestoreTotalLogSizeAfterRecoverWithoutFlush) {
 
     void OnFlushBegin(DB* /*db*/, const FlushJobInfo& flush_job_info) override {
       count++;
-      assert(FlushReason::kWriteBufferManager == flush_job_info.flush_reason);
+      ASSERT_EQ(FlushReason::kWriteBufferManager, flush_job_info.flush_reason);
     }
   };
   std::shared_ptr<TestFlushListener> test_listener =
@@ -1690,7 +1827,9 @@ TEST_F(DBWALTest, RestoreTotalLogSizeAfterRecoverWithoutFlush) {
             1 * kMB);
   // Write one more key to trigger flush.
   ASSERT_OK(Put(0, "foo", "v2"));
-  dbfull()->TEST_WaitForFlushMemTable();
+  for (auto* h : handles_) {
+    ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable(h));
+  }
   // Flushed two column families.
   ASSERT_EQ(2, test_listener->count.load());
 }
@@ -1764,9 +1903,9 @@ TEST_F(DBWALTest, WalTermTest) {
   wo.disableWAL = false;
 
   WriteBatch batch;
-  batch.Put("foo", "bar");
+  ASSERT_OK(batch.Put("foo", "bar"));
   batch.MarkWalTerminationPoint();
-  batch.Put("foo2", "bar2");
+  ASSERT_OK(batch.Put("foo2", "bar2"));
 
   ASSERT_OK(dbfull()->Write(wo, &batch));
 

db/db_with_timestamp_basic_test.cc

@@ -32,6 +32,22 @@ class DBBasicTestWithTimestampBase : public DBTestBase {
     return ret;
   }
 
+  static std::string KeyWithPrefix(std::string prefix, uint64_t k) {
+    std::string ret;
+    PutFixed64(&ret, k);
+    std::reverse(ret.begin(), ret.end());
+    return prefix + ret;
+  }
+
+  static std::vector<Slice> ConvertStrToSlice(
+      std::vector<std::string>& strings) {
+    std::vector<Slice> ret;
+    for (const auto& s : strings) {
+      ret.emplace_back(s);
+    }
+    return ret;
+  }
+
   class TestComparator : public Comparator {
    private:
     const Comparator* cmp_without_ts_;
@@ -179,6 +195,218 @@ class DBBasicTestWithTimestamp : public DBBasicTestWithTimestampBase {
       : DBBasicTestWithTimestampBase("db_basic_test_with_timestamp") {}
 };
 
+TEST_F(DBBasicTestWithTimestamp, CompactRangeWithSpecifiedRange) {
+  Options options = CurrentOptions();
+  options.env = env_;
+  options.create_if_missing = true;
+  const size_t kTimestampSize = Timestamp(0, 0).size();
+  TestComparator test_cmp(kTimestampSize);
+  options.comparator = &test_cmp;
+  DestroyAndReopen(options);
+
+  WriteOptions write_opts;
+  std::string ts_str = Timestamp(1, 0);
+  Slice ts = ts_str;
+  write_opts.timestamp = &ts;
+
+  ASSERT_OK(db_->Put(write_opts, "foo1", "bar"));
+  ASSERT_OK(Flush());
+
+  ASSERT_OK(db_->Put(write_opts, "foo2", "bar"));
+  ASSERT_OK(Flush());
+
+  std::string start_str = "foo";
+  std::string end_str = "foo2";
+  Slice start(start_str), end(end_str);
+  ASSERT_OK(db_->CompactRange(CompactRangeOptions(), &start, &end));
+
+  Close();
+}
+
+TEST_F(DBBasicTestWithTimestamp, UpdateFullHistoryTsLow) {
+  Options options = CurrentOptions();
+  options.env = env_;
+  options.create_if_missing = true;
+  const size_t kTimestampSize = Timestamp(0, 0).size();
+  TestComparator test_cmp(kTimestampSize);
+  options.comparator = &test_cmp;
+  DestroyAndReopen(options);
+
+  const std::string kKey = "test kKey";
+
+  // Test set ts_low first and flush()
+  int current_ts_low = 5;
+  std::string ts_low_str = Timestamp(current_ts_low, 0);
+  Slice ts_low = ts_low_str;
+  CompactRangeOptions comp_opts;
+  comp_opts.full_history_ts_low = &ts_low;
+  comp_opts.bottommost_level_compaction = BottommostLevelCompaction::kForce;
+
+  ASSERT_OK(db_->CompactRange(comp_opts, nullptr, nullptr));
+
+  auto* cfd =
+      static_cast_with_check<ColumnFamilyHandleImpl>(db_->DefaultColumnFamily())
+          ->cfd();
+  auto result_ts_low = cfd->GetFullHistoryTsLow();
+
+  ASSERT_TRUE(test_cmp.CompareTimestamp(ts_low, result_ts_low) == 0);
+
+  for (int i = 0; i < 10; i++) {
+    WriteOptions write_opts;
+    std::string ts_str = Timestamp(i, 0);
+    Slice ts = ts_str;
+    write_opts.timestamp = &ts;
+    ASSERT_OK(db_->Put(write_opts, kKey, Key(i)));
+  }
+  ASSERT_OK(Flush());
+
+  for (int i = 0; i < 10; i++) {
+    ReadOptions read_opts;
+    std::string ts_str = Timestamp(i, 0);
+    Slice ts = ts_str;
+    read_opts.timestamp = &ts;
+    std::string value;
+    Status status = db_->Get(read_opts, kKey, &value);
+    if (i < current_ts_low) {
+      ASSERT_TRUE(status.IsNotFound());
+    } else {
+      ASSERT_OK(status);
+      ASSERT_TRUE(value.compare(Key(i)) == 0);
+    }
+  }
+
+  // Test set ts_low and then trigger compaction
+  for (int i = 10; i < 20; i++) {
+    WriteOptions write_opts;
+    std::string ts_str = Timestamp(i, 0);
+    Slice ts = ts_str;
+    write_opts.timestamp = &ts;
+    ASSERT_OK(db_->Put(write_opts, kKey, Key(i)));
+  }
+
+  ASSERT_OK(Flush());
+
+  current_ts_low = 15;
+  ts_low_str = Timestamp(current_ts_low, 0);
+  ts_low = ts_low_str;
+  comp_opts.full_history_ts_low = &ts_low;
+  ASSERT_OK(db_->CompactRange(comp_opts, nullptr, nullptr));
+  result_ts_low = cfd->GetFullHistoryTsLow();
+  ASSERT_TRUE(test_cmp.CompareTimestamp(ts_low, result_ts_low) == 0);
+
+  for (int i = 0; i < 20; i++) {
+    ReadOptions read_opts;
+    std::string ts_str = Timestamp(i, 0);
+    Slice ts = ts_str;
+    read_opts.timestamp = &ts;
+    std::string value;
+    Status status = db_->Get(read_opts, kKey, &value);
+    if (i < current_ts_low) {
+      ASSERT_TRUE(status.IsNotFound());
+    } else {
+      ASSERT_OK(status);
+      ASSERT_TRUE(value.compare(Key(i)) == 0);
+    }
+  }
+
+  // Test invalid compaction with range
+  Slice start(kKey), end(kKey);
+  Status s = db_->CompactRange(comp_opts, &start, &end);
+  ASSERT_TRUE(s.IsInvalidArgument());
+  s = db_->CompactRange(comp_opts, &start, nullptr);
+  ASSERT_TRUE(s.IsInvalidArgument());
+  s = db_->CompactRange(comp_opts, nullptr, &end);
+  ASSERT_TRUE(s.IsInvalidArgument());
+
+  // Test invalid compaction with the decreasing ts_low
+  ts_low_str = Timestamp(current_ts_low - 1, 0);
+  ts_low = ts_low_str;
+  comp_opts.full_history_ts_low = &ts_low;
+  s = db_->CompactRange(comp_opts, nullptr, nullptr);
+  ASSERT_TRUE(s.IsInvalidArgument());
+
+  Close();
+}
+
+TEST_F(DBBasicTestWithTimestamp, GetApproximateSizes) {
+  Options options = CurrentOptions();
+  options.write_buffer_size = 100000000;  // Large write buffer
+  options.compression = kNoCompression;
+  options.create_if_missing = true;
+  const size_t kTimestampSize = Timestamp(0, 0).size();
+  TestComparator test_cmp(kTimestampSize);
+  options.comparator = &test_cmp;
+  DestroyAndReopen(options);
+  auto default_cf = db_->DefaultColumnFamily();
+
+  WriteOptions write_opts;
+  std::string ts_str = Timestamp(1, 0);
+  Slice ts = ts_str;
+  write_opts.timestamp = &ts;
+
+  const int N = 128;
+  Random rnd(301);
+  for (int i = 0; i < N; i++) {
+    ASSERT_OK(db_->Put(write_opts, Key(i), rnd.RandomString(1024)));
+  }
+
+  uint64_t size;
+  std::string start = Key(50);
+  std::string end = Key(60);
+  Range r(start, end);
+  SizeApproximationOptions size_approx_options;
+  size_approx_options.include_memtabtles = true;
+  size_approx_options.include_files = true;
+  ASSERT_OK(
+      db_->GetApproximateSizes(size_approx_options, default_cf, &r, 1, &size));
+  ASSERT_GT(size, 6000);
+  ASSERT_LT(size, 204800);
+
+  // test multiple ranges
+  std::vector<Range> ranges;
+  std::string start_tmp = Key(10);
+  std::string end_tmp = Key(20);
+  ranges.emplace_back(Range(start_tmp, end_tmp));
+  ranges.emplace_back(Range(start, end));
+  uint64_t range_sizes[2];
+  ASSERT_OK(db_->GetApproximateSizes(size_approx_options, default_cf,
+                                     ranges.data(), 2, range_sizes));
+
+  ASSERT_EQ(range_sizes[1], size);
+
+  // Zero if not including mem table
+  ASSERT_OK(db_->GetApproximateSizes(&r, 1, &size));
+  ASSERT_EQ(size, 0);
+
+  start = Key(500);
+  end = Key(600);
+  r = Range(start, end);
+  ASSERT_OK(
+      db_->GetApproximateSizes(size_approx_options, default_cf, &r, 1, &size));
+  ASSERT_EQ(size, 0);
+
+  // Test range boundaries
+  ASSERT_OK(db_->Put(write_opts, Key(1000), rnd.RandomString(1024)));
+  // Should include start key
+  start = Key(1000);
+  end = Key(1100);
+  r = Range(start, end);
+  ASSERT_OK(
+      db_->GetApproximateSizes(size_approx_options, default_cf, &r, 1, &size));
+  ASSERT_GT(size, 0);
+
+  // Should exclude end key
+  start = Key(900);
+  end = Key(1000);
+  r = Range(start, end);
+  ASSERT_OK(
+      db_->GetApproximateSizes(size_approx_options, default_cf, &r, 1, &size));
+  ASSERT_EQ(size, 0);
+  std::cout << size << std::endl;
+
+  Close();
+}
+
 TEST_F(DBBasicTestWithTimestamp, SimpleForwardIterate) {
   const int kNumKeysPerFile = 128;
   const uint64_t kMaxKey = 1024;
@@ -249,6 +477,8 @@ TEST_F(DBBasicTestWithTimestamp, SeekWithPrefixLessThanKey) {
   options.env = env_;
   options.create_if_missing = true;
   options.prefix_extractor.reset(NewFixedPrefixTransform(3));
+  options.memtable_whole_key_filtering = true;
+  options.memtable_prefix_bloom_size_ratio = 0.1;
   BlockBasedTableOptions bbto;
   bbto.filter_policy.reset(NewBloomFilterPolicy(10, false));
   bbto.cache_index_and_filter_blocks = true;
@@ -265,16 +495,16 @@ TEST_F(DBBasicTestWithTimestamp, SeekWithPrefixLessThanKey) {
   write_opts.timestamp = &ts;
 
   ASSERT_OK(db_->Put(write_opts, "foo1", "bar"));
-  Flush();
+  ASSERT_OK(Flush());
 
   ASSERT_OK(db_->Put(write_opts, "foo2", "bar"));
-  Flush();
+  ASSERT_OK(Flush());
 
   // Move sst file to next level
   ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
 
   ASSERT_OK(db_->Put(write_opts, "foo3", "bar"));
-  Flush();
+  ASSERT_OK(Flush());
 
   ReadOptions read_opts;
   std::string read_ts = Timestamp(1, 0);
@@ -285,6 +515,10 @@ TEST_F(DBBasicTestWithTimestamp, SeekWithPrefixLessThanKey) {
     iter->Seek("foo");
     ASSERT_TRUE(iter->Valid());
     ASSERT_OK(iter->status());
+    iter->Next();
+    ASSERT_TRUE(iter->Valid());
+    ASSERT_OK(iter->status());
+
     iter->Seek("bbb");
     ASSERT_FALSE(iter->Valid());
     ASSERT_OK(iter->status());
@@ -298,6 +532,8 @@ TEST_F(DBBasicTestWithTimestamp, SeekWithPrefixLargerThanKey) {
   options.env = env_;
   options.create_if_missing = true;
   options.prefix_extractor.reset(NewFixedPrefixTransform(20));
+  options.memtable_whole_key_filtering = true;
+  options.memtable_prefix_bloom_size_ratio = 0.1;
   BlockBasedTableOptions bbto;
   bbto.filter_policy.reset(NewBloomFilterPolicy(10, false));
   bbto.cache_index_and_filter_blocks = true;
@@ -314,21 +550,76 @@ TEST_F(DBBasicTestWithTimestamp, SeekWithPrefixLargerThanKey) {
   write_opts.timestamp = &ts;
 
   ASSERT_OK(db_->Put(write_opts, "foo1", "bar"));
-  Flush();
+  ASSERT_OK(Flush());
 
   ASSERT_OK(db_->Put(write_opts, "foo2", "bar"));
-  Flush();
+  ASSERT_OK(Flush());
 
   // Move sst file to next level
   ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
 
   ASSERT_OK(db_->Put(write_opts, "foo3", "bar"));
-  Flush();
+  ASSERT_OK(Flush());
 
   ReadOptions read_opts;
   std::string read_ts = Timestamp(2, 0);
   ts = read_ts;
   read_opts.timestamp = &ts;
+  {
+    std::unique_ptr<Iterator> iter(db_->NewIterator(read_opts));
+    // Make sure the prefix extractor doesn't include timestamp, otherwise it
+    // may return invalid result.
+    iter->Seek("foo");
+    ASSERT_TRUE(iter->Valid());
+    ASSERT_OK(iter->status());
+    iter->Next();
+    ASSERT_TRUE(iter->Valid());
+    ASSERT_OK(iter->status());
+  }
+
+  Close();
+}
+
+TEST_F(DBBasicTestWithTimestamp, SeekWithBound) {
+  Options options = CurrentOptions();
+  options.env = env_;
+  options.create_if_missing = true;
+  options.prefix_extractor.reset(NewFixedPrefixTransform(2));
+  BlockBasedTableOptions bbto;
+  bbto.filter_policy.reset(NewBloomFilterPolicy(10, false));
+  bbto.cache_index_and_filter_blocks = true;
+  bbto.whole_key_filtering = true;
+  options.table_factory.reset(NewBlockBasedTableFactory(bbto));
+  const size_t kTimestampSize = Timestamp(0, 0).size();
+  TestComparator test_cmp(kTimestampSize);
+  options.comparator = &test_cmp;
+  DestroyAndReopen(options);
+
+  WriteOptions write_opts;
+  std::string ts_str = Timestamp(1, 0);
+  Slice ts = ts_str;
+  write_opts.timestamp = &ts;
+
+  ASSERT_OK(db_->Put(write_opts, "foo1", "bar"));
+  ASSERT_OK(Flush());
+
+  ASSERT_OK(db_->Put(write_opts, "foo2", "bar"));
+  ASSERT_OK(Flush());
+
+  // Move sst file to next level
+  ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
+
+  ASSERT_OK(db_->Put(write_opts, "foo3", "bar"));
+  ASSERT_OK(Flush());
+
+  ReadOptions read_opts;
+  std::string read_ts = Timestamp(2, 0);
+  ts = read_ts;
+  read_opts.timestamp = &ts;
+  std::string up_bound = "foo5";
+  Slice up_bound_slice = up_bound;
+  read_opts.iterate_upper_bound = &up_bound_slice;
+  read_opts.auto_prefix_mode = true;
   {
     std::unique_ptr<Iterator> iter(db_->NewIterator(read_opts));
     // Make sure the prefix extractor doesn't include timestamp, otherwise it
@@ -611,7 +902,7 @@ TEST_F(DBBasicTestWithTimestamp, MultiGetWithFastLocalBloom) {
 
   ASSERT_OK(db_->Put(write_opts, "foo", "bar"));
 
-  Flush();
+  ASSERT_OK(Flush());
 
   // Read with MultiGet
   ReadOptions read_opts;
@@ -652,7 +943,7 @@ TEST_F(DBBasicTestWithTimestamp, MultiGetWithPrefix) {
 
   ASSERT_OK(db_->Put(write_opts, "foo", "bar"));
 
-  Flush();
+  ASSERT_OK(Flush());
 
   // Read with MultiGet
   ReadOptions read_opts;
@@ -740,7 +1031,7 @@ TEST_F(DBBasicTestWithTimestamp, MultiGetRangeFiltering) {
     Slice key_slice = key;
     Slice value_slice = value;
     ASSERT_OK(db_->Put(write_opts, key_slice, value_slice));
-    Flush();
+    ASSERT_OK(Flush());
   }
 
   // Make num_levels to 2 to do key range filtering of sst files
@@ -748,7 +1039,7 @@ TEST_F(DBBasicTestWithTimestamp, MultiGetRangeFiltering) {
 
   ASSERT_OK(db_->Put(write_opts, "foo", "bar"));
 
-  Flush();
+  ASSERT_OK(Flush());
 
   // Read with MultiGet
   ts_str = Timestamp(2, 0);
@@ -791,7 +1082,7 @@ TEST_F(DBBasicTestWithTimestamp, MultiGetPrefixFilter) {
 
   ASSERT_OK(db_->Put(write_opts, "foo", "bar"));
 
-  Flush();
+  ASSERT_OK(Flush());
   // Read with MultiGet
   ts_str = Timestamp(2, 0);
   ts = ts_str;
@@ -909,6 +1200,131 @@ TEST_F(DBBasicTestWithTimestamp, CompactDeletionWithTimestampMarkerToBottom) {
   Close();
 }
 
+class DBBasicTestWithTimestampFilterPrefixSettings
+    : public DBBasicTestWithTimestampBase,
+      public testing::WithParamInterface<
+          std::tuple<std::shared_ptr<const FilterPolicy>, bool, bool,
+                     std::shared_ptr<const SliceTransform>, bool, double>> {
+ public:
+  DBBasicTestWithTimestampFilterPrefixSettings()
+      : DBBasicTestWithTimestampBase(
+            "db_basic_test_with_timestamp_filter_prefix") {}
+};
+
+TEST_P(DBBasicTestWithTimestampFilterPrefixSettings, GetAndMultiGet) {
+  Options options = CurrentOptions();
+  options.env = env_;
+  options.create_if_missing = true;
+  BlockBasedTableOptions bbto;
+  bbto.filter_policy = std::get<0>(GetParam());
+  bbto.whole_key_filtering = std::get<1>(GetParam());
+  bbto.cache_index_and_filter_blocks = std::get<2>(GetParam());
+  options.table_factory.reset(NewBlockBasedTableFactory(bbto));
+  options.prefix_extractor = std::get<3>(GetParam());
+  options.memtable_whole_key_filtering = std::get<4>(GetParam());
+  options.memtable_prefix_bloom_size_ratio = std::get<5>(GetParam());
+
+  const size_t kTimestampSize = Timestamp(0, 0).size();
+  TestComparator test_cmp(kTimestampSize);
+  options.comparator = &test_cmp;
+  DestroyAndReopen(options);
+  const int kMaxKey = 1000;
+
+  // Write any value
+  WriteOptions write_opts;
+  std::string ts_str = Timestamp(1, 0);
+  Slice ts = ts_str;
+  write_opts.timestamp = &ts;
+
+  int idx = 0;
+  for (; idx < kMaxKey / 4; idx++) {
+    ASSERT_OK(db_->Put(write_opts, Key1(idx), "bar"));
+    ASSERT_OK(db_->Put(write_opts, KeyWithPrefix("foo", idx), "bar"));
+  }
+
+  ASSERT_OK(Flush());
+  ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
+
+  for (; idx < kMaxKey / 2; idx++) {
+    ASSERT_OK(db_->Put(write_opts, Key1(idx), "bar"));
+    ASSERT_OK(db_->Put(write_opts, KeyWithPrefix("foo", idx), "bar"));
+  }
+
+  ASSERT_OK(Flush());
+
+  for (; idx < kMaxKey; idx++) {
+    ASSERT_OK(db_->Put(write_opts, Key1(idx), "bar"));
+    ASSERT_OK(db_->Put(write_opts, KeyWithPrefix("foo", idx), "bar"));
+  }
+
+  // Read with MultiGet
+  ReadOptions read_opts;
+  read_opts.timestamp = &ts;
+
+  ReadOptions read_opts_total_order;
+  read_opts_total_order.timestamp = &ts;
+  read_opts_total_order.total_order_seek = true;
+
+  for (idx = 0; idx < kMaxKey; idx++) {
+    size_t batch_size = 4;
+    std::vector<std::string> keys_str(batch_size);
+    std::vector<PinnableSlice> values(batch_size);
+    std::vector<Status> statuses(batch_size);
+    ColumnFamilyHandle* cfh = db_->DefaultColumnFamily();
+
+    keys_str[0] = Key1(idx);
+    keys_str[1] = KeyWithPrefix("foo", idx);
+    keys_str[2] = Key1(kMaxKey + idx);
+    keys_str[3] = KeyWithPrefix("foo", kMaxKey + idx);
+
+    auto keys = ConvertStrToSlice(keys_str);
+
+    db_->MultiGet(read_opts, cfh, batch_size, keys.data(), values.data(),
+                  statuses.data());
+
+    for (int i = 0; i < 2; i++) {
+      ASSERT_OK(statuses[i]);
+    }
+    for (int i = 2; i < 4; i++) {
+      ASSERT_TRUE(statuses[i].IsNotFound());
+    }
+
+    for (int i = 0; i < 2; i++) {
+      std::string value;
+      ASSERT_OK(db_->Get(read_opts, keys[i], &value));
+      std::unique_ptr<Iterator> it1(db_->NewIterator(read_opts));
+      ASSERT_NE(nullptr, it1);
+      ASSERT_OK(it1->status());
+      // TODO(zjay) Fix seek with prefix
+      // it1->Seek(keys[i]);
+      // ASSERT_TRUE(it1->Valid());
+    }
+
+    for (int i = 2; i < 4; i++) {
+      std::string value;
+      Status s = db_->Get(read_opts, keys[i], &value);
+      ASSERT_TRUE(s.IsNotFound());
+    }
+  }
+  Close();
+}
+
+INSTANTIATE_TEST_CASE_P(
+    Timestamp, DBBasicTestWithTimestampFilterPrefixSettings,
+    ::testing::Combine(
+        ::testing::Values(
+            std::shared_ptr<const FilterPolicy>(nullptr),
+            std::shared_ptr<const FilterPolicy>(NewBloomFilterPolicy(10, true)),
+            std::shared_ptr<const FilterPolicy>(NewBloomFilterPolicy(10,
+                                                                     false))),
+        ::testing::Bool(), ::testing::Bool(),
+        ::testing::Values(
+            std::shared_ptr<const SliceTransform>(NewFixedPrefixTransform(1)),
+            std::shared_ptr<const SliceTransform>(NewFixedPrefixTransform(4)),
+            std::shared_ptr<const SliceTransform>(NewFixedPrefixTransform(7)),
+            std::shared_ptr<const SliceTransform>(NewFixedPrefixTransform(8))),
+        ::testing::Bool(), ::testing::Values(0, 0.1)));
+
 class DataVisibilityTest : public DBBasicTestWithTimestampBase {
  public:
   DataVisibilityTest() : DBBasicTestWithTimestampBase("data_visibility_test") {
@@ -1372,7 +1788,7 @@ TEST_F(DataVisibilityTest, MultiGetWithTimestamp) {
   VerifyDefaultCF(snap0);
   VerifyDefaultCF(snap1);
 
-  Flush();
+  ASSERT_OK(Flush());
 
   const Snapshot* snap2 = db_->GetSnapshot();
   PutTestData(2);
@@ -1458,7 +1874,7 @@ TEST_F(DataVisibilityTest, MultiGetCrossCF) {
   VerifyDefaultCF(snap0);
   VerifyDefaultCF(snap1);
 
-  Flush();
+  ASSERT_OK(Flush());
 
   const Snapshot* snap2 = db_->GetSnapshot();
   PutTestData(2);
@@ -1839,6 +2255,8 @@ TEST_F(DBBasicTestWithTimestamp, BatchWriteAndMultiGet) {
   options.create_if_missing = true;
   options.env = env_;
   options.memtable_factory.reset(new SpecialSkipListFactory(kNumKeysPerFile));
+  options.memtable_prefix_bloom_size_ratio = 0.1;
+  options.memtable_whole_key_filtering = true;
 
   size_t ts_sz = Timestamp(0, 0).size();
   TestComparator test_cmp(ts_sz);
@@ -2093,6 +2511,7 @@ INSTANTIATE_TEST_CASE_P(
     Timestamp, DBBasicTestWithTimestampPrefixSeek,
     ::testing::Combine(
         ::testing::Values(
+            std::shared_ptr<const SliceTransform>(NewFixedPrefixTransform(1)),
             std::shared_ptr<const SliceTransform>(NewFixedPrefixTransform(4)),
             std::shared_ptr<const SliceTransform>(NewFixedPrefixTransform(7)),
             std::shared_ptr<const SliceTransform>(NewFixedPrefixTransform(8))),

db/db_write_test.cc

@@ -60,14 +60,15 @@ TEST_P(DBWriteTest, WriteStallRemoveNoSlowdownWrite) {
     std::string key = "foo" + std::to_string(a);
     WriteOptions wo;
     wo.no_slowdown = false;
-    dbfull()->Put(wo, key, "bar");
+    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;
-    dbfull()->Put(wo, key, "bar");
+    Status s = dbfull()->Put(wo, key, "bar");
+    ASSERT_TRUE(s.ok() || s.IsIncomplete());
   };
   std::function<void(void*)> unblock_main_thread_func = [&](void*) {
     mutex.Lock();
@@ -77,13 +78,13 @@ TEST_P(DBWriteTest, WriteStallRemoveNoSlowdownWrite) {
   };
 
   // Create 3 L0 files and schedule 4th without waiting
-  Put("foo" + std::to_string(thread_num.fetch_add(1)), "bar");
-  Flush();
-  Put("foo" + std::to_string(thread_num.fetch_add(1)), "bar");
-  Flush();
-  Put("foo" + std::to_string(thread_num.fetch_add(1)), "bar");
-  Flush();
-  Put("foo" + std::to_string(thread_num.fetch_add(1)), "bar");
+  ASSERT_OK(Put("foo" + std::to_string(thread_num.fetch_add(1)), "bar"));
+  ASSERT_OK(Flush());
+  ASSERT_OK(Put("foo" + std::to_string(thread_num.fetch_add(1)), "bar"));
+  ASSERT_OK(Flush());
+  ASSERT_OK(Put("foo" + std::to_string(thread_num.fetch_add(1)), "bar"));
+  ASSERT_OK(Flush());
+  ASSERT_OK(Put("foo" + std::to_string(thread_num.fetch_add(1)), "bar"));
 
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
       "WriteThread::JoinBatchGroup:Start", unblock_main_thread_func);
@@ -104,7 +105,7 @@ TEST_P(DBWriteTest, WriteStallRemoveNoSlowdownWrite) {
   // write_thread
   FlushOptions fopt;
   fopt.wait = false;
-  dbfull()->Flush(fopt);
+  ASSERT_OK(dbfull()->Flush(fopt));
 
   // Create a mix of slowdown/no_slowdown write threads
   mutex.Lock();
@@ -145,7 +146,7 @@ TEST_P(DBWriteTest, WriteStallRemoveNoSlowdownWrite) {
   mutex.Unlock();
 
   TEST_SYNC_POINT("DBWriteTest::WriteStallRemoveNoSlowdownWrite:1");
-  dbfull()->TEST_WaitForFlushMemTable(nullptr);
+  ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable(nullptr));
   // This would have triggered a write stall. Unblock the write group leader
   TEST_SYNC_POINT("DBWriteTest::WriteStallRemoveNoSlowdownWrite:2");
   // The leader is going to create missing newer links. When the leader
@@ -178,14 +179,15 @@ TEST_P(DBWriteTest, WriteThreadHangOnWriteStall) {
     std::string key = "foo" + std::to_string(a);
     WriteOptions wo;
     wo.no_slowdown = false;
-    dbfull()->Put(wo, key, "bar");
+    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;
-    dbfull()->Put(wo, key, "bar");
+    Status s = dbfull()->Put(wo, key, "bar");
+    ASSERT_TRUE(s.ok() || s.IsIncomplete());
   };
   std::function<void(void *)> unblock_main_thread_func = [&](void *) {
     mutex.Lock();
@@ -195,13 +197,13 @@ TEST_P(DBWriteTest, WriteThreadHangOnWriteStall) {
   };
 
   // Create 3 L0 files and schedule 4th without waiting
-  Put("foo" + std::to_string(thread_num.fetch_add(1)), "bar");
-  Flush();
-  Put("foo" + std::to_string(thread_num.fetch_add(1)), "bar");
-  Flush();
-  Put("foo" + std::to_string(thread_num.fetch_add(1)), "bar");
-  Flush();
-  Put("foo" + std::to_string(thread_num.fetch_add(1)), "bar");
+  ASSERT_OK(Put("foo" + std::to_string(thread_num.fetch_add(1)), "bar"));
+  ASSERT_OK(Flush());
+  ASSERT_OK(Put("foo" + std::to_string(thread_num.fetch_add(1)), "bar"));
+  ASSERT_OK(Flush());
+  ASSERT_OK(Put("foo" + std::to_string(thread_num.fetch_add(1)), "bar"));
+  ASSERT_OK(Flush());
+  ASSERT_OK(Put("foo" + std::to_string(thread_num.fetch_add(1)), "bar"));
 
   ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
       "WriteThread::JoinBatchGroup:Start", unblock_main_thread_func);
@@ -222,7 +224,7 @@ TEST_P(DBWriteTest, WriteThreadHangOnWriteStall) {
   // write_thread
   FlushOptions fopt;
   fopt.wait = false;
-  dbfull()->Flush(fopt);
+  ASSERT_OK(dbfull()->Flush(fopt));
 
   // Create a mix of slowdown/no_slowdown write threads
   mutex.Lock();
@@ -243,7 +245,7 @@ TEST_P(DBWriteTest, WriteThreadHangOnWriteStall) {
   mutex.Unlock();
 
   TEST_SYNC_POINT("DBWriteTest::WriteThreadHangOnWriteStall:1");
-  dbfull()->TEST_WaitForFlushMemTable(nullptr);
+  ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable(nullptr));
   // This would have triggered a write stall. Unblock the write group leader
   TEST_SYNC_POINT("DBWriteTest::WriteThreadHangOnWriteStall:2");
   // The leader is going to create missing newer links. When the leader finishes,
@@ -307,6 +309,11 @@ TEST_P(DBWriteTest, IOErrorOnWALWritePropagateToWriteThreadFollower) {
     threads[i].join();
   }
   ASSERT_EQ(1, leader_count);
+
+  // The Failed PUT operations can cause a BG error to be set.
+  // Mark it as Checked for the ASSERT_STATUS_CHECKED
+  dbfull()->Resume().PermitUncheckedError();
+
   // Close before mock_env destruct.
   Close();
 }
@@ -320,7 +327,7 @@ TEST_P(DBWriteTest, ManualWalFlushInEffect) {
   ASSERT_TRUE(dbfull()->FlushWAL(false).ok());
   ASSERT_TRUE(dbfull()->TEST_WALBufferIsEmpty());
   // try the 2nd wal created during SwitchWAL
-  dbfull()->TEST_SwitchWAL();
+  ASSERT_OK(dbfull()->TEST_SwitchWAL());
   ASSERT_TRUE(Put("key" + ToString(0), "value").ok());
   ASSERT_TRUE(options.manual_wal_flush != dbfull()->TEST_WALBufferIsEmpty());
   ASSERT_TRUE(dbfull()->FlushWAL(false).ok());
@@ -351,7 +358,9 @@ TEST_P(DBWriteTest, IOErrorOnWALWriteTriggersReadOnlyMode) {
     }
     */
     if (!options.manual_wal_flush) {
-      ASSERT_FALSE(res.ok());
+      ASSERT_NOK(res);
+    } else {
+      ASSERT_OK(res);
     }
   }
   // Close before mock_env destruct.
@@ -395,7 +404,7 @@ TEST_P(DBWriteTest, LockWalInEffect) {
   ASSERT_TRUE(dbfull()->TEST_WALBufferIsEmpty(false));
   ASSERT_OK(dbfull()->UnlockWAL());
   // try the 2nd wal created during SwitchWAL
-  dbfull()->TEST_SwitchWAL();
+  ASSERT_OK(dbfull()->TEST_SwitchWAL());
   ASSERT_OK(Put("key" + ToString(0), "value"));
   ASSERT_TRUE(options.manual_wal_flush != dbfull()->TEST_WALBufferIsEmpty());
   ASSERT_OK(dbfull()->LockWAL());
@@ -423,13 +432,14 @@ TEST_P(DBWriteTest, ConcurrentlyDisabledWAL) {
               ROCKSDB_NAMESPACE::WriteOptions write_option_default;
               std::string no_wal_key = no_wal_key_prefix + std::to_string(t) +
                                        "_" + std::to_string(i);
-              this->Put(no_wal_key, no_wal_value, write_option_disable);
+              ASSERT_OK(
+                  this->Put(no_wal_key, no_wal_value, write_option_disable));
               std::string wal_key =
                   wal_key_prefix + std::to_string(i) + "_" + std::to_string(i);
-              this->Put(wal_key, wal_value, write_option_default);
-              dbfull()->SyncWAL();
+              ASSERT_OK(this->Put(wal_key, wal_value, write_option_default));
+              ASSERT_OK(dbfull()->SyncWAL());
             }
-            return 0;
+            return;
         });
     }
     for (auto& t: threads) {

db/dbformat.cc

@@ -78,6 +78,22 @@ void AppendInternalKeyFooter(std::string* result, SequenceNumber s,
   PutFixed64(result, PackSequenceAndType(s, t));
 }
 
+void AppendKeyWithMinTimestamp(std::string* result, const Slice& key,
+                               size_t ts_sz) {
+  assert(ts_sz > 0);
+  const std::string kTsMin(ts_sz, static_cast<unsigned char>(0));
+  result->append(key.data(), key.size());
+  result->append(kTsMin.data(), ts_sz);
+}
+
+void AppendKeyWithMaxTimestamp(std::string* result, const Slice& key,
+                               size_t ts_sz) {
+  assert(ts_sz > 0);
+  const std::string kTsMax(ts_sz, static_cast<unsigned char>(0xff));
+  result->append(key.data(), key.size());
+  result->append(kTsMax.data(), ts_sz);
+}
+
 std::string ParsedInternalKey::DebugString(bool log_err_key, bool hex) const {
   std::string result = "'";
   if (log_err_key) {

db/dbformat.h

@@ -146,8 +146,10 @@ inline void UnPackSequenceAndType(uint64_t packed, uint64_t* seq,
   *seq = packed >> 8;
   *t = static_cast<ValueType>(packed & 0xff);
 
-  assert(*seq <= kMaxSequenceNumber);
-  assert(IsExtendedValueType(*t));
+  // Commented the following two assertions in order to test key-value checksum
+  // on corrupted keys without crashing ("DbKvChecksumTest").
+  // assert(*seq <= kMaxSequenceNumber);
+  // assert(IsExtendedValueType(*t));
 }
 
 EntryType GetEntryType(ValueType value_type);
@@ -167,6 +169,14 @@ extern void AppendInternalKeyWithDifferentTimestamp(
 extern void AppendInternalKeyFooter(std::string* result, SequenceNumber s,
                                     ValueType t);
 
+// Append the key and a minimal timestamp to *result
+extern void AppendKeyWithMinTimestamp(std::string* result, const Slice& key,
+                                      size_t ts_sz);
+
+// Append the key and a maximal timestamp to *result
+extern void AppendKeyWithMaxTimestamp(std::string* result, const Slice& key,
+                                      size_t ts_sz);
+
 // Attempt to parse an internal key from "internal_key".  On success,
 // stores the parsed data in "*result", and returns true.
 //

db/deletefile_test.cc

@@ -40,7 +40,7 @@ class DeleteFileTest : public DBTestBase {
         wal_dir_(dbname_ + "/wal_files") {}
 
   void SetOptions(Options* options) {
-    assert(options);
+    ASSERT_NE(options, nullptr);
     options->delete_obsolete_files_period_micros = 0;  // always do full purge
     options->enable_thread_tracking = true;
     options->write_buffer_size = 1024 * 1024 * 1000;
@@ -105,7 +105,7 @@ class DeleteFileTest : public DBTestBase {
   void CheckFileTypeCounts(const std::string& dir, int required_log,
                            int required_sst, int required_manifest) {
     std::vector<std::string> filenames;
-    env_->GetChildren(dir, &filenames);
+    ASSERT_OK(env_->GetChildren(dir, &filenames));
 
     int log_cnt = 0, sst_cnt = 0, manifest_cnt = 0;
     for (auto file : filenames) {
@@ -180,7 +180,8 @@ TEST_F(DeleteFileTest, AddKeysAndQueryLevels) {
   ASSERT_TRUE(status.IsInvalidArgument());
 
   // Lowest level file deletion should succeed.
-  ASSERT_OK(db_->DeleteFile(level2file));
+  status = db_->DeleteFile(level2file);
+  ASSERT_OK(status);
 }
 
 TEST_F(DeleteFileTest, PurgeObsoleteFilesTest) {
@@ -201,7 +202,7 @@ TEST_F(DeleteFileTest, PurgeObsoleteFilesTest) {
   compact_options.change_level = true;
   compact_options.target_level = 2;
   Slice first_slice(first), last_slice(last);
-  db_->CompactRange(compact_options, &first_slice, &last_slice);
+  ASSERT_OK(db_->CompactRange(compact_options, &first_slice, &last_slice));
   // 1 sst after compaction
   CheckFileTypeCounts(dbname_, 0, 1, 1);
 
@@ -210,7 +211,9 @@ TEST_F(DeleteFileTest, PurgeObsoleteFilesTest) {
   Iterator *itr = nullptr;
   CreateTwoLevels();
   itr = db_->NewIterator(ReadOptions());
-  db_->CompactRange(compact_options, &first_slice, &last_slice);
+  ASSERT_OK(itr->status());
+  ASSERT_OK(db_->CompactRange(compact_options, &first_slice, &last_slice));
+  ASSERT_OK(itr->status());
   // 3 sst after compaction with live iterator
   CheckFileTypeCounts(dbname_, 0, 3, 1);
   delete itr;
@@ -237,7 +240,8 @@ TEST_F(DeleteFileTest, BackgroundPurgeIteratorTest) {
   ReadOptions read_options;
   read_options.background_purge_on_iterator_cleanup = true;
   itr = db_->NewIterator(read_options);
-  db_->CompactRange(compact_options, &first_slice, &last_slice);
+  ASSERT_OK(itr->status());
+  ASSERT_OK(db_->CompactRange(compact_options, &first_slice, &last_slice));
   // 3 sst after compaction with live iterator
   CheckFileTypeCounts(dbname_, 0, 3, 1);
   test::SleepingBackgroundTask sleeping_task_before;
@@ -344,11 +348,12 @@ TEST_F(DeleteFileTest, BackgroundPurgeCopyOptions) {
     ReadOptions read_options;
     read_options.background_purge_on_iterator_cleanup = true;
     itr = db_->NewIterator(read_options);
+    ASSERT_OK(itr->status());
     // ReadOptions is deleted, but iterator cleanup function should not be
     // affected
   }
 
-  db_->CompactRange(compact_options, &first_slice, &last_slice);
+  ASSERT_OK(db_->CompactRange(compact_options, &first_slice, &last_slice));
   // 3 sst after compaction with live iterator
   CheckFileTypeCounts(dbname_, 0, 3, 1);
   delete itr;
@@ -382,9 +387,11 @@ TEST_F(DeleteFileTest, BackgroundPurgeTestMultipleJobs) {
   ReadOptions read_options;
   read_options.background_purge_on_iterator_cleanup = true;
   Iterator* itr1 = db_->NewIterator(read_options);
+  ASSERT_OK(itr1->status());
   CreateTwoLevels();
   Iterator* itr2 = db_->NewIterator(read_options);
-  db_->CompactRange(compact_options, &first_slice, &last_slice);
+  ASSERT_OK(itr2->status());
+  ASSERT_OK(db_->CompactRange(compact_options, &first_slice, &last_slice));
   // 5 sst files after 2 compactions with 2 live iterators
   CheckFileTypeCounts(dbname_, 0, 5, 1);
 
@@ -417,6 +424,7 @@ TEST_F(DeleteFileTest, DeleteFileWithIterator) {
   CreateTwoLevels();
   ReadOptions read_options;
   Iterator* it = db_->NewIterator(read_options);
+  ASSERT_OK(it->status());
   std::vector<LiveFileMetaData> metadata;
   db_->GetLiveFilesMetaData(&metadata);
 
@@ -432,7 +440,7 @@ TEST_F(DeleteFileTest, DeleteFileWithIterator) {
   Status status = db_->DeleteFile(level2file);
   fprintf(stdout, "Deletion status %s: %s\n",
           level2file.c_str(), status.ToString().c_str());
-  ASSERT_TRUE(status.ok());
+  ASSERT_OK(status);
   it->SeekToFirst();
   int numKeysIterated = 0;
   while(it->Valid()) {
@@ -452,7 +460,7 @@ TEST_F(DeleteFileTest, DeleteLogFiles) {
 
   AddKeys(10, 0);
   VectorLogPtr logfiles;
-  db_->GetSortedWalFiles(logfiles);
+  ASSERT_OK(db_->GetSortedWalFiles(logfiles));
   ASSERT_GT(logfiles.size(), 0UL);
   // Take the last log file which is expected to be alive and try to delete it
   // Should not succeed because live logs are not allowed to be deleted
@@ -461,7 +469,7 @@ TEST_F(DeleteFileTest, DeleteLogFiles) {
   ASSERT_OK(env_->FileExists(wal_dir_ + "/" + alive_log->PathName()));
   fprintf(stdout, "Deleting alive log file %s\n",
           alive_log->PathName().c_str());
-  ASSERT_TRUE(!db_->DeleteFile(alive_log->PathName()).ok());
+  ASSERT_NOK(db_->DeleteFile(alive_log->PathName()));
   ASSERT_OK(env_->FileExists(wal_dir_ + "/" + alive_log->PathName()));
   logfiles.clear();
 
@@ -469,10 +477,10 @@ TEST_F(DeleteFileTest, DeleteLogFiles) {
   // Call Flush again to flush out memtable and move alive log to archived log
   // and try to delete the archived log file
   FlushOptions fopts;
-  db_->Flush(fopts);
+  ASSERT_OK(db_->Flush(fopts));
   AddKeys(10, 0);
-  db_->Flush(fopts);
-  db_->GetSortedWalFiles(logfiles);
+  ASSERT_OK(db_->Flush(fopts));
+  ASSERT_OK(db_->GetSortedWalFiles(logfiles));
   ASSERT_GT(logfiles.size(), 0UL);
   std::unique_ptr<LogFile> archived_log = std::move(logfiles.front());
   ASSERT_EQ(archived_log->Type(), kArchivedLogFile);
@@ -480,8 +488,8 @@ TEST_F(DeleteFileTest, DeleteLogFiles) {
   fprintf(stdout, "Deleting archived log file %s\n",
           archived_log->PathName().c_str());
   ASSERT_OK(db_->DeleteFile(archived_log->PathName()));
-  ASSERT_EQ(Status::NotFound(),
-            env_->FileExists(wal_dir_ + "/" + archived_log->PathName()));
+  ASSERT_TRUE(
+      env_->FileExists(wal_dir_ + "/" + archived_log->PathName()).IsNotFound());
 }
 
 TEST_F(DeleteFileTest, DeleteNonDefaultColumnFamily) {
@@ -520,6 +528,7 @@ TEST_F(DeleteFileTest, DeleteNonDefaultColumnFamily) {
 
   {
     std::unique_ptr<Iterator> itr(db_->NewIterator(ReadOptions(), handles_[1]));
+    ASSERT_OK(itr->status());
     int count = 0;
     for (itr->SeekToFirst(); itr->Valid(); itr->Next()) {
       ASSERT_OK(itr->status());

db/error_handler.cc

@@ -111,6 +111,23 @@ std::map<std::tuple<BackgroundErrorReason, Status::Code, Status::SubCode, bool>,
                          Status::Code::kIOError, Status::SubCode::kIOFenced,
                          false),
          Status::Severity::kFatalError},
+        // Errors during MANIFEST write when WAL is disabled
+        {std::make_tuple(BackgroundErrorReason::kManifestWriteNoWAL,
+                         Status::Code::kIOError, Status::SubCode::kNoSpace,
+                         true),
+         Status::Severity::kHardError},
+        {std::make_tuple(BackgroundErrorReason::kManifestWriteNoWAL,
+                         Status::Code::kIOError, Status::SubCode::kNoSpace,
+                         false),
+         Status::Severity::kHardError},
+        {std::make_tuple(BackgroundErrorReason::kManifestWriteNoWAL,
+                         Status::Code::kIOError, Status::SubCode::kIOFenced,
+                         true),
+         Status::Severity::kFatalError},
+        {std::make_tuple(BackgroundErrorReason::kManifestWriteNoWAL,
+                         Status::Code::kIOError, Status::SubCode::kIOFenced,
+                         false),
+         Status::Severity::kFatalError},
 
 };
 
@@ -175,6 +192,12 @@ std::map<std::tuple<BackgroundErrorReason, Status::Code, bool>,
         {std::make_tuple(BackgroundErrorReason::kFlushNoWAL,
                          Status::Code::kIOError, false),
          Status::Severity::kNoError},
+        {std::make_tuple(BackgroundErrorReason::kManifestWriteNoWAL,
+                         Status::Code::kIOError, true),
+         Status::Severity::kFatalError},
+        {std::make_tuple(BackgroundErrorReason::kManifestWriteNoWAL,
+                         Status::Code::kIOError, false),
+         Status::Severity::kFatalError},
 };
 
 std::map<std::tuple<BackgroundErrorReason, bool>, Status::Severity>
@@ -244,10 +267,11 @@ void ErrorHandler::CancelErrorRecovery() {
 // This can also get called as part of a recovery operation. In that case, we
 // also track the error separately in recovery_error_ so we can tell in the
 // end whether recovery succeeded or not
-Status ErrorHandler::SetBGError(const Status& bg_err, BackgroundErrorReason reason) {
+const Status& ErrorHandler::SetBGError(const Status& bg_err,
+                                       BackgroundErrorReason reason) {
   db_mutex_->AssertHeld();
   if (bg_err.ok()) {
-    return Status::OK();
+    return bg_err;
   }
 
   bool paranoid = db_options_.paranoid_checks;
@@ -324,11 +348,32 @@ Status ErrorHandler::SetBGError(const Status& bg_err, BackgroundErrorReason reas
   return bg_error_;
 }
 
-Status ErrorHandler::SetBGError(const IOStatus& bg_io_err,
-                                BackgroundErrorReason reason) {
+// This is the main function for looking at IO related error during the
+// background operations. The main logic is:
+// 1) File scope IO error is treated as retryable IO error in the write
+//    path. In RocksDB, If a file has write IO error and it is at file scope,
+//    RocksDB never write to the same file again. RocksDB will create a new
+//    file and rewrite the whole content. Thus, it is retryable.
+// 1) if the error is caused by data loss, the error is mapped to
+//    unrecoverable error. Application/user must take action to handle
+//    this situation (File scope case is excluded).
+// 2) if the error is a Retryable IO error (i.e., it is a file scope IO error,
+//     or its retryable flag is set and not a data loss error), auto resume
+//     will be called and the auto resume can be controlled by resume count
+//     and resume interval options. There are three sub-cases:
+//    a) if the error happens during compaction, it is mapped to a soft error.
+//       the compaction thread will reschedule a new compaction.
+//    b) if the error happens during flush and also WAL is empty, it is mapped
+//       to a soft error. Note that, it includes the case that IO error happens
+//       in SST or manifest write during flush.
+//    c) all other errors are mapped to hard error.
+// 3) for other cases, SetBGError(const Status& bg_err, BackgroundErrorReason
+//    reason) will be called to handle other error cases.
+const Status& ErrorHandler::SetBGError(const IOStatus& bg_io_err,
+                                       BackgroundErrorReason reason) {
   db_mutex_->AssertHeld();
   if (bg_io_err.ok()) {
-    return Status::OK();
+    return bg_io_err;
   }
   ROCKS_LOG_WARN(db_options_.info_log, "Background IO error %s",
                  bg_io_err.ToString().c_str());
@@ -336,45 +381,50 @@ Status ErrorHandler::SetBGError(const IOStatus& bg_io_err,
   if (recovery_in_prog_ && recovery_io_error_.ok()) {
     recovery_io_error_ = bg_io_err;
   }
-  if (BackgroundErrorReason::kManifestWrite == reason) {
+  if (BackgroundErrorReason::kManifestWrite == reason ||
+      BackgroundErrorReason::kManifestWriteNoWAL == reason) {
     // Always returns ok
     db_->DisableFileDeletionsWithLock().PermitUncheckedError();
   }
 
   Status new_bg_io_err = bg_io_err;
-  Status s;
   DBRecoverContext context;
-  if (bg_io_err.GetDataLoss()) {
-    // First, data loss is treated as unrecoverable error. So it can directly
-    // overwrite any existing bg_error_.
+  if (bg_io_err.GetScope() != IOStatus::IOErrorScope::kIOErrorScopeFile &&
+      bg_io_err.GetDataLoss()) {
+    // First, data loss (non file scope) is treated as unrecoverable error. So
+    // it can directly overwrite any existing bg_error_.
     bool auto_recovery = false;
     Status bg_err(new_bg_io_err, Status::Severity::kUnrecoverableError);
     bg_error_ = bg_err;
     if (recovery_in_prog_ && recovery_error_.ok()) {
       recovery_error_ = bg_err;
     }
-    EventHelpers::NotifyOnBackgroundError(db_options_.listeners, reason, &s,
-                                          db_mutex_, &auto_recovery);
+    EventHelpers::NotifyOnBackgroundError(db_options_.listeners, reason,
+                                          &bg_err, db_mutex_, &auto_recovery);
     recover_context_ = context;
     return bg_error_;
-  } else if (bg_io_err.GetRetryable()) {
-    // Second, check if the error is a retryable IO error or not. if it is
-    // retryable error and its severity is higher than bg_error_, overwrite
-    // the bg_error_ with new error.
-    // In current stage, for retryable IO error of compaction, treat it as
-    // soft error. In other cases, treat the retryable IO error as hard
-    // error.
+  } else if (bg_io_err.GetScope() ==
+                 IOStatus::IOErrorScope::kIOErrorScopeFile ||
+             bg_io_err.GetRetryable()) {
+    // Second, check if the error is a retryable IO error (file scope IO error
+    // is also treated as retryable IO error in RocksDB write path). if it is
+    // retryable error and its severity is higher than bg_error_, overwrite the
+    // bg_error_ with new error. In current stage, for retryable IO error of
+    // compaction, treat it as soft error. In other cases, treat the retryable
+    // IO error as hard error.
     bool auto_recovery = false;
-    EventHelpers::NotifyOnBackgroundError(db_options_.listeners, reason, &s,
-                                          db_mutex_, &auto_recovery);
+    EventHelpers::NotifyOnBackgroundError(db_options_.listeners, reason,
+                                          &new_bg_io_err, db_mutex_,
+                                          &auto_recovery);
     if (BackgroundErrorReason::kCompaction == reason) {
-      Status bg_err(new_bg_io_err, Status::Severity::kSoftError);
-      if (bg_err.severity() > bg_error_.severity()) {
-        bg_error_ = bg_err;
-      }
-      recover_context_ = context;
+      // We map the retryable IO error during compaction to soft error. Since
+      // compaction can reschedule by itself. We will not set the BG error in
+      // this case
+      // TODO:  a better way to set or clean the retryable IO error which
+      // happens during compaction SST file write.
       return bg_error_;
-    } else if (BackgroundErrorReason::kFlushNoWAL == reason) {
+    } else if (BackgroundErrorReason::kFlushNoWAL == reason ||
+               BackgroundErrorReason::kManifestWriteNoWAL == reason) {
       // When the BG Retryable IO error reason is flush without WAL,
       // We map it to a soft error. At the same time, all the background work
       // should be stopped except the BG work from recovery. Therefore, we
@@ -405,12 +455,11 @@ Status ErrorHandler::SetBGError(const IOStatus& bg_io_err,
       return StartRecoverFromRetryableBGIOError(bg_io_err);
     }
   } else {
-    s = SetBGError(new_bg_io_err, reason);
+    return SetBGError(new_bg_io_err, reason);
   }
-  return s;
 }
 
-Status ErrorHandler::OverrideNoSpaceError(Status bg_error,
+Status ErrorHandler::OverrideNoSpaceError(const Status& bg_error,
                                           bool* auto_recovery) {
 #ifndef ROCKSDB_LITE
   if (bg_error.severity() >= Status::Severity::kFatalError) {
@@ -466,7 +515,11 @@ Status ErrorHandler::ClearBGError() {
   // Signal that recovery succeeded
   if (recovery_error_.ok()) {
     Status old_bg_error = bg_error_;
+    // Clear and check the recovery IO and BG error
     bg_error_ = Status::OK();
+    recovery_io_error_ = IOStatus::OK();
+    bg_error_.PermitUncheckedError();
+    recovery_io_error_.PermitUncheckedError();
     recovery_in_prog_ = false;
     soft_error_no_bg_work_ = false;
     EventHelpers::NotifyOnErrorRecoveryCompleted(db_options_.listeners,
@@ -516,6 +569,7 @@ Status ErrorHandler::RecoverFromBGError(bool is_manual) {
   // during the recovery process. While recovering, the only operations that
   // can generate background errors should be the flush operations
   recovery_error_ = Status::OK();
+  recovery_error_.PermitUncheckedError();
   Status s = db_->ResumeImpl(recover_context_);
   if (s.ok()) {
     soft_error_no_bg_work_ = false;
@@ -537,24 +591,34 @@ Status ErrorHandler::RecoverFromBGError(bool is_manual) {
 #endif
 }
 
-Status ErrorHandler::StartRecoverFromRetryableBGIOError(IOStatus io_error) {
+const Status& ErrorHandler::StartRecoverFromRetryableBGIOError(
+    const IOStatus& io_error) {
 #ifndef ROCKSDB_LITE
   db_mutex_->AssertHeld();
-  if (bg_error_.ok() || io_error.ok()) {
-    return Status::OK();
-  }
-  if (db_options_.max_bgerror_resume_count <= 0 || recovery_in_prog_ ||
-      recovery_thread_) {
+  if (bg_error_.ok()) {
+    return bg_error_;
+  } else if (io_error.ok()) {
+    return io_error;
+  } else if (db_options_.max_bgerror_resume_count <= 0 || recovery_in_prog_) {
     // Auto resume BG error is not enabled, directly return bg_error_.
     return bg_error_;
   }
 
+  if (recovery_thread_) {
+    // In this case, if recovery_in_prog_ is false, current thread should
+    // wait the previous recover thread to finish and create a new thread
+    // to recover from the bg error.
+    db_mutex_->Unlock();
+    recovery_thread_->join();
+    db_mutex_->Lock();
+  }
+
   recovery_in_prog_ = true;
   recovery_thread_.reset(
       new port::Thread(&ErrorHandler::RecoverFromRetryableBGIOError, this));
 
   if (recovery_io_error_.ok() && recovery_error_.ok()) {
-    return Status::OK();
+    return recovery_error_;
   } else {
     TEST_SYNC_POINT("StartRecoverRetryableBGIOError:RecoverFail");
     return bg_error_;
@@ -619,6 +683,7 @@ void ErrorHandler::RecoverFromRetryableBGIOError() {
         TEST_SYNC_POINT("RecoverFromRetryableBGIOError:RecoverSuccess");
         Status old_bg_error = bg_error_;
         bg_error_ = Status::OK();
+        bg_error_.PermitUncheckedError();
         EventHelpers::NotifyOnErrorRecoveryCompleted(db_options_.listeners,
                                                      old_bg_error, db_mutex_);
         recovery_in_prog_ = false;