Update version.h to 8.11.5 and update HISTORY.md

* Backport new value types

.circleci/config.yml

@@ -54,6 +54,7 @@ commands:
           command: |
             echo "Installing CMake..."
             choco install cmake --installargs 'ADD_CMAKE_TO_PATH=System' -y
+            choco install liberica8jdk -y
             mkdir $Env:THIRDPARTY_HOME
             cd $Env:THIRDPARTY_HOME
             echo "Building Snappy dependency..."
@@ -67,9 +68,10 @@ commands:
       - run:
           name: "Build RocksDB"
           command: |
+            $env:Path = $env:JAVA_HOME + ";" + $env:Path
             mkdir build
             cd build
-            & $Env:CMAKE_BIN -G "$Env:CMAKE_GENERATOR" -DCMAKE_BUILD_TYPE=Debug -DOPTDBG=1 -DPORTABLE=1 -DSNAPPY=1 -DJNI=1 ..
+            & $Env:CMAKE_BIN -G "$Env:CMAKE_GENERATOR" -DCMAKE_BUILD_TYPE=Debug -DOPTDBG=1 -DPORTABLE="$Env:CMAKE_PORTABLE" -DSNAPPY=1 -DJNI=1 ..
             cd ..
             echo "Building with VS version: $Env:CMAKE_GENERATOR"
             msbuild.exe build/rocksdb.sln -maxCpuCount -property:Configuration=Debug -property:Platform=x64
@@ -144,9 +146,12 @@ commands:
   install-maven:
     steps:
       - run:
-          name: Install maven
+          name: Install Maven
           command: |
-            sudo apt-get update -y && sudo apt-get install -y maven
+            wget --no-check-certificate https://dlcdn.apache.org/maven/maven-3/3.9.6/binaries/apache-maven-3.9.6-bin.tar.gz
+            tar zxf apache-maven-3.9.6-bin.tar.gz
+            echo "export M2_HOME=$(pwd)/apache-maven-3.9.6" >> $BASH_ENV
+            echo 'export PATH=$M2_HOME/bin:$PATH' >> $BASH_ENV
 
   setup-folly:
     steps:
@@ -227,6 +232,10 @@ executors:
       #  $ docker run --cap-add=SYS_PTRACE --security-opt seccomp=unconfined -it zjay437/rocksdb:0.5 bash
       # option `--cap-add=SYS_PTRACE --security-opt seccomp=unconfined` is used to enable gdb to attach an existing process
       - image: zjay437/rocksdb:0.6
+  linux-java-docker:
+    docker:
+      # This is the Docker Image used for building RocksJava releases, see: https://github.com/evolvedbinary/docker-rocksjava
+      - image: evolvedbinary/rocksjava:centos6_x64-be
 
 jobs:
   build-macos:
@@ -239,7 +248,7 @@ jobs:
       - increase-max-open-files-on-macos
       - install-gflags-on-macos
       - pre-steps-macos
-      - run: ulimit -S -n `ulimit -H -n` && OPT=-DCIRCLECI make V=1 J=16 -j16 all
+      - run: ulimit -S -n `ulimit -H -n` && make V=1 J=16 -j16 all
       - post-steps
 
   build-macos-cmake:
@@ -563,6 +572,24 @@ jobs:
           no_output_timeout: 100m
       - post-steps
 
+  build-windows-vs2022-avx2:
+    executor:
+      name: win/server-2022
+      size: 2xlarge
+    environment:
+      THIRDPARTY_HOME: C:/Users/circleci/thirdparty
+      CMAKE_HOME: C:/Program Files/CMake
+      CMAKE_BIN: C:/Program Files/CMake/bin/cmake.exe
+      CTEST_BIN: C:/Program Files/CMake/bin/ctest.exe
+      JAVA_HOME: C:/Program Files/BellSoft/LibericaJDK-8
+      SNAPPY_HOME: C:/Users/circleci/thirdparty/snappy-1.1.8
+      SNAPPY_INCLUDE: C:/Users/circleci/thirdparty/snappy-1.1.8;C:/Users/circleci/thirdparty/snappy-1.1.8/build
+      SNAPPY_LIB_DEBUG: C:/Users/circleci/thirdparty/snappy-1.1.8/build/Debug/snappy.lib
+      CMAKE_GENERATOR: Visual Studio 17 2022
+      CMAKE_PORTABLE: AVX2
+    steps:
+      - windows-build-steps
+
   build-windows-vs2022:
     executor:
       name: win/server-2022
@@ -572,10 +599,12 @@ jobs:
       CMAKE_HOME: C:/Program Files/CMake
       CMAKE_BIN: C:/Program Files/CMake/bin/cmake.exe
       CTEST_BIN: C:/Program Files/CMake/bin/ctest.exe
+      JAVA_HOME: C:/Program Files/BellSoft/LibericaJDK-8
       SNAPPY_HOME: C:/Users/circleci/thirdparty/snappy-1.1.8
       SNAPPY_INCLUDE: C:/Users/circleci/thirdparty/snappy-1.1.8;C:/Users/circleci/thirdparty/snappy-1.1.8/build
       SNAPPY_LIB_DEBUG: C:/Users/circleci/thirdparty/snappy-1.1.8/build/Debug/snappy.lib
       CMAKE_GENERATOR: Visual Studio 17 2022
+      CMAKE_PORTABLE: 1
     steps:
       - windows-build-steps
 
@@ -588,15 +617,17 @@ jobs:
       CMAKE_HOME: C:/Program Files/CMake
       CMAKE_BIN: C:/Program Files/CMake/bin/cmake.exe
       CTEST_BIN: C:/Program Files/CMake/bin/ctest.exe
+      JAVA_HOME: C:/Program Files/BellSoft/LibericaJDK-8
       SNAPPY_HOME: C:/Users/circleci/thirdparty/snappy-1.1.8
       SNAPPY_INCLUDE: C:/Users/circleci/thirdparty/snappy-1.1.8;C:/Users/circleci/thirdparty/snappy-1.1.8/build
       SNAPPY_LIB_DEBUG: C:/Users/circleci/thirdparty/snappy-1.1.8/build/Debug/snappy.lib
       CMAKE_GENERATOR: Visual Studio 16 2019
+      CMAKE_PORTABLE: 1
     steps:
       - windows-build-steps
 
   build-linux-java:
-    executor: linux-docker
+    executor: linux-java-docker
     resource_class: large
     steps:
       - pre-steps
@@ -609,17 +640,13 @@ jobs:
             which javac && javac -version
       - run:
           name: "Test RocksDBJava"
-          command: make V=1 J=8 -j8 jtest
+          command: scl enable devtoolset-7 'make V=1 J=8 -j8 jtest'
       - post-steps
 
   build-linux-java-pmd:
-    machine: 
-      image: ubuntu-2004:202111-02
+    executor: linux-java-docker
     resource_class: large
-    environment:
-      JAVA_HOME: /usr/lib/jvm/java-8-openjdk-amd64
     steps:
-      - install-maven
       - pre-steps
       - run:
           name: "Set Java Environment"
@@ -628,13 +655,14 @@ jobs:
             echo 'export PATH=$JAVA_HOME/bin:$PATH' >> $BASH_ENV
             which java && java -version
             which javac && javac -version
+      - install-maven
       - run:
           name: "PMD RocksDBJava"
-          command: make V=1 J=8 -j8 jpmd
+          command: scl enable devtoolset-7 'make V=1 J=8 -j8 jpmd'
       - post-pmd-steps
 
   build-linux-java-static:
-    executor: linux-docker
+    executor: linux-java-docker
     resource_class: large
     steps:
       - pre-steps
@@ -647,7 +675,7 @@ jobs:
             which javac && javac -version
       - run:
           name: "Build RocksDBJava Static Library"
-          command: make V=1 J=8 -j8 rocksdbjavastatic
+          command: scl enable devtoolset-7 'make V=1 J=8 -j8 rocksdbjavastatic'
       - post-steps
 
   build-macos-java:
@@ -850,67 +878,11 @@ jobs:
       - perform-benchmarks
       - post-benchmarks
 
-workflows:
+workflows: # Only jobs that haven't been successfully migrated to GitHub Actions
   version: 2
-  jobs-linux-run-tests:
-    jobs:
-      - build-linux
-      - build-linux-cmake-with-folly
-      - build-linux-cmake-with-folly-lite-no-test
-      - build-linux-gcc-7-with-folly
-      - build-linux-gcc-7-with-folly-lite-no-test
-      - build-linux-cmake-with-folly-coroutines
-      - build-linux-cmake-with-benchmark
-      - build-linux-encrypted_env-no_compression
-  jobs-linux-run-tests-san:
-    jobs:
-      - build-linux-clang10-asan
-      - build-linux-clang10-ubsan
-      - build-linux-clang10-mini-tsan
-      - build-linux-static_lib-alt_namespace-status_checked
-  jobs-linux-no-test-run:
-    jobs:
-      - build-linux-release
-      - build-linux-release-rtti
-      - build-examples
-      - build-fuzzers
-      - build-linux-clang-no_test_run
-      - build-linux-clang-13-no_test_run
-      - build-linux-gcc-8-no_test_run
-      - build-linux-gcc-10-cxx20-no_test_run
-      - build-linux-gcc-11-no_test_run
-      - build-linux-arm-cmake-no_test_run
-  jobs-linux-other-checks:
-    jobs:
-      - build-linux-clang10-clang-analyze
-      - build-linux-unity-and-headers
-      - build-linux-mini-crashtest
-  jobs-windows:
-    jobs:
-      - build-windows-vs2022
-      - build-windows-vs2019
-      - build-cmake-mingw
-  jobs-java:
-    jobs:
-      - build-linux-java
-      - build-linux-java-static
-      - build-macos-java
-      - build-macos-java-static
-      - build-macos-java-static-universal
-      - build-linux-java-pmd
-  jobs-macos:
-    jobs:
-      - build-macos
-      - build-macos-cmake:
-          run_even_tests: true
-      - build-macos-cmake:
-          run_even_tests: false
   jobs-linux-arm:
     jobs:
       - build-linux-arm
-  build-fuzzers:
-    jobs:
-      - build-fuzzers
   benchmark-linux:
     triggers:
       - schedule:
@@ -930,9 +902,4 @@ workflows:
               only:
                 - main
     jobs:
-      - build-format-compatible
       - build-linux-arm-test-full
-      - build-linux-run-microbench
-      - build-linux-non-shm
-      - build-linux-clang-13-asan-ubsan-with-folly
-      - build-linux-valgrind

.github/actions/build-folly/action.yml

@@ -0,0 +1,7 @@
+name: build-folly
+runs:
+  using: composite
+  steps:
+  - name: Build folly and dependencies
+    run: make build_folly
+    shell: bash

.github/actions/build-for-benchmarks/action.yml

@@ -0,0 +1,8 @@
+name: build-for-benchmarks
+runs:
+  using: composite
+  steps:
+  - uses: "./.github/actions/pre-steps"
+  - name: Linux build for benchmarks
+    run: make V=1 J=8 -j8 release
+    shell: bash

.github/actions/increase-max-open-files-on-macos/action.yml

@@ -0,0 +1,10 @@
+name: increase-max-open-files-on-macos
+runs:
+  using: composite
+  steps:
+  - name: Increase max open files
+    run: |-
+      sudo sysctl -w kern.maxfiles=1048576
+      sudo sysctl -w kern.maxfilesperproc=1048576
+      sudo launchctl limit maxfiles 1048576
+    shell: bash

.github/actions/install-gflags-on-macos/action.yml

@@ -0,0 +1,7 @@
+name: install-gflags-on-macos
+runs:
+  using: composite
+  steps:
+  - name: Install gflags on macos
+    run: HOMEBREW_NO_AUTO_UPDATE=1 brew install gflags
+    shell: bash

.github/actions/install-gflags/action.yml

@@ -0,0 +1,7 @@
+name: install-gflags
+runs:
+  using: composite
+  steps:
+  - name: Install gflags
+    run: sudo apt-get update -y && sudo apt-get install -y libgflags-dev
+    shell: bash

.github/actions/install-jdk8-on-macos/action.yml

@@ -0,0 +1,9 @@
+name: install-jdk8-on-macos
+runs:
+  using: composite
+  steps:
+  - name: Install JDK 8 on macos
+    run: |-
+      HOMEBREW_NO_AUTO_UPDATE=1 brew tap bell-sw/liberica
+      HOMEBREW_NO_AUTO_UPDATE=1 brew install --cask liberica-jdk8
+    shell: bash

.github/actions/install-maven/action.yml

@@ -0,0 +1,11 @@
+name: install-maven
+runs:
+  using: composite
+  steps:
+  - name: Install Maven
+    run: |
+      wget --no-check-certificate https://dlcdn.apache.org/maven/maven-3/3.9.6/binaries/apache-maven-3.9.6-bin.tar.gz
+      tar zxf apache-maven-3.9.6-bin.tar.gz
+      echo "export M2_HOME=$(pwd)/apache-maven-3.9.6" >> $GITHUB_ENV
+      echo "$(pwd)/apache-maven-3.9.6/bin" >> $GITHUB_PATH
+    shell: bash

.github/actions/perform-benchmarks/action.yml

@@ -0,0 +1,22 @@
+name: perform-benchmarks
+runs:
+  using: composite
+  steps:
+  - name: Test low-variance benchmarks
+    run: "./tools/benchmark_ci.py --db_dir ${{ runner.temp }}/rocksdb-benchmark-datadir --output_dir ${{ runner.temp }}/benchmark-results --num_keys 20000000"
+    env:
+      LD_LIBRARY_PATH: "/usr/local/lib"
+      DURATION_RO: 300
+      DURATION_RW: 500
+      NUM_THREADS: 1
+      MAX_BACKGROUND_JOBS: 4
+      CI_TESTS_ONLY: 'true'
+      WRITE_BUFFER_SIZE_MB: 16
+      TARGET_FILE_SIZE_BASE_MB: 16
+      MAX_BYTES_FOR_LEVEL_BASE_MB: 64
+      COMPRESSION_TYPE: none
+      CACHE_INDEX_AND_FILTER_BLOCKS: 1
+      MIN_LEVEL_TO_COMPRESS: 3
+      CACHE_SIZE_MB: 10240
+      MB_WRITE_PER_SEC: 2
+    shell: bash

.github/actions/post-benchmarks/action.yml

@@ -0,0 +1,17 @@
+name: post-benchmarks
+runs:
+  using: composite
+  steps:
+  - name: Upload Benchmark Results artifact
+    uses: actions/upload-artifact@v4.0.0
+    with:
+      name: benchmark-results
+      path: "${{ runner.temp }}/benchmark-results/**"
+      if-no-files-found: error
+  - name: Send benchmark report to visualisation
+    run: |-
+      set +e
+      set +o pipefail
+      ./build_tools/benchmark_log_tool.py --tsvfile ${{ runner.temp }}/benchmark-results/report.tsv --esdocument https://search-rocksdb-bench-k2izhptfeap2hjfxteolsgsynm.us-west-2.es.amazonaws.com/bench_test3_rix/_doc
+      true
+    shell: bash

.github/actions/post-steps/action.yml

@@ -0,0 +1,38 @@
+name: post-steps
+description: Steps that are taken after a RocksDB job
+inputs:
+  artifact-prefix:
+    description: Prefix to append to the name of artifacts that are uploaded
+    required: true
+    default: "${{ github.job }}"
+runs:
+  using: composite
+  steps:
+  - name: Upload Test Results artifact
+    uses: actions/upload-artifact@v4.0.0
+    with:
+      name: "${{ inputs.artifact-prefix }}-test-results"
+      path: "${{ runner.temp }}/test-results/**"
+  - name: Upload DB LOG file artifact
+    uses: actions/upload-artifact@v4.0.0
+    with:
+      name: "${{ inputs.artifact-prefix }}-db-log-file"
+      path: LOG
+  - name: Copy Test Logs (on Failure)
+    if: ${{ failure() }}
+    run: |
+      mkdir -p ${{ runner.temp }}/failure-test-logs
+      cp -r t/* ${{ runner.temp }}/failure-test-logs
+    shell: bash
+  - name: Upload Test Logs (on Failure) artifact
+    uses: actions/upload-artifact@v4.0.0
+    with:
+      name: "${{ inputs.artifact-prefix }}-failure-test-logs"
+      path: ${{ runner.temp }}/failure-test-logs/**
+      if-no-files-found: ignore
+  - name: Upload Core Dumps artifact
+    uses: actions/upload-artifact@v4.0.0
+    with:
+      name: "${{ inputs.artifact-prefix }}-core-dumps"
+      path: "core.*"
+      if-no-files-found: ignore

.github/actions/pre-steps-macos/action.yml

@@ -0,0 +1,5 @@
+name: pre-steps-macos
+runs:
+  using: composite
+  steps:
+  - uses: "./.github/actions/pre-steps"

.github/actions/pre-steps/action.yml

@@ -0,0 +1,18 @@
+name: pre-steps
+runs:
+  using: composite
+  steps:
+  - name: Setup Environment Variables
+    run: |-
+      echo "GTEST_THROW_ON_FAILURE=0" >> "$GITHUB_ENV"
+      echo "GTEST_OUTPUT=\"xml:${{ runner.temp }}/test-results/\"" >> "$GITHUB_ENV"
+      echo "SKIP_FORMAT_BUCK_CHECKS=1" >> "$GITHUB_ENV"
+      echo "GTEST_COLOR=1" >> "$GITHUB_ENV"
+      echo "CTEST_OUTPUT_ON_FAILURE=1" >> "$GITHUB_ENV"
+      echo "CTEST_TEST_TIMEOUT=300" >> "$GITHUB_ENV"
+      echo "ZLIB_DOWNLOAD_BASE=https://rocksdb-deps.s3.us-west-2.amazonaws.com/pkgs/zlib" >> "$GITHUB_ENV"
+      echo "BZIP2_DOWNLOAD_BASE=https://rocksdb-deps.s3.us-west-2.amazonaws.com/pkgs/bzip2" >> "$GITHUB_ENV"
+      echo "SNAPPY_DOWNLOAD_BASE=https://rocksdb-deps.s3.us-west-2.amazonaws.com/pkgs/snappy" >> "$GITHUB_ENV"
+      echo "LZ4_DOWNLOAD_BASE=https://rocksdb-deps.s3.us-west-2.amazonaws.com/pkgs/lz4" >> "$GITHUB_ENV"
+      echo "ZSTD_DOWNLOAD_BASE=https://rocksdb-deps.s3.us-west-2.amazonaws.com/pkgs/zstd" >> "$GITHUB_ENV"
+    shell: bash

.github/actions/setup-folly/action.yml

@@ -0,0 +1,7 @@
+name: setup-folly
+runs:
+  using: composite
+  steps:
+  - name: Checkout folly sources
+    run: make checkout_folly
+    shell: bash

.github/actions/setup-upstream/action.yml

@@ -0,0 +1,20 @@
+name: build-folly
+runs:
+  using: composite
+  steps:
+  - name: Fix repo ownership
+    # Needed in some cases, as safe.directory setting doesn't take effect
+    # under env -i
+    run: chown `whoami` . || true
+    shell: bash
+  - name: Set upstream
+    run: git remote add upstream https://github.com/facebook/rocksdb.git
+    shell: bash
+  - name: Fetch upstream
+    run: git fetch upstream
+    shell: bash
+  - name: Git status
+    # NOTE: some old branch builds under check_format_compatible.sh invoke
+    # git under env -i
+    run: git status && git remote -v && env -i git branch
+    shell: bash

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

@@ -0,0 +1,54 @@
+name: windows-build-steps
+runs:
+  using: composite
+  steps:
+  - name: Add msbuild to PATH
+    uses: microsoft/setup-msbuild@v1.3.1
+  - name: Custom steps
+    env:
+      THIRDPARTY_HOME: ${{ github.workspace }}/thirdparty
+      CMAKE_HOME: C:/Program Files/CMake
+      CMAKE_BIN: C:/Program Files/CMake/bin/cmake.exe
+      CTEST_BIN: C:/Program Files/CMake/bin/ctest.exe
+      JAVA_HOME: C:/Program Files/BellSoft/LibericaJDK-8
+      SNAPPY_HOME: ${{ github.workspace }}/thirdparty/snappy-1.1.8
+      SNAPPY_INCLUDE: ${{ github.workspace }}/thirdparty/snappy-1.1.8;${{ github.workspace }}/thirdparty/snappy-1.1.8/build
+      SNAPPY_LIB_DEBUG: ${{ github.workspace }}/thirdparty/snappy-1.1.8/build/Debug/snappy.lib
+    run: |-
+      # NOTE: if ... Exit $LASTEXITCODE lines needed to exit and report failure
+      echo ===================== Install Dependencies =====================
+      choco install liberica8jdk -y
+      if(!$?) { Exit $LASTEXITCODE }
+      mkdir $Env:THIRDPARTY_HOME
+      cd $Env:THIRDPARTY_HOME
+      echo "Building Snappy dependency..."
+      curl -Lo snappy-1.1.8.zip https://github.com/google/snappy/archive/refs/tags/1.1.8.zip
+      if(!$?) { Exit $LASTEXITCODE }
+      unzip -q snappy-1.1.8.zip
+      if(!$?) { Exit $LASTEXITCODE }
+      cd snappy-1.1.8
+      mkdir build
+      cd build
+      & cmake -G "$Env:CMAKE_GENERATOR" ..
+      if(!$?) { Exit $LASTEXITCODE }
+      msbuild Snappy.sln -maxCpuCount -property:Configuration=Debug -property:Platform=x64
+      if(!$?) { Exit $LASTEXITCODE }
+      echo ======================== Build RocksDB =========================
+      cd ${{ github.workspace }}
+      $env:Path = $env:JAVA_HOME + ";" + $env:Path
+      mkdir build
+      cd build
+      & cmake -G "$Env:CMAKE_GENERATOR" -DCMAKE_BUILD_TYPE=Debug -DOPTDBG=1 -DPORTABLE="$Env:CMAKE_PORTABLE" -DSNAPPY=1 -DJNI=1 ..
+      if(!$?) { Exit $LASTEXITCODE }
+      cd ..
+      echo "Building with VS version: $Env:CMAKE_GENERATOR"
+      msbuild build/rocksdb.sln -maxCpuCount -property:Configuration=Debug -property:Platform=x64
+      if(!$?) { Exit $LASTEXITCODE }
+      echo ========================= Test RocksDB =========================
+      build_tools\run_ci_db_test.ps1 -SuiteRun arena_test,db_basic_test,db_test,db_test2,db_merge_operand_test,bloom_test,c_test,coding_test,crc32c_test,dynamic_bloom_test,env_basic_test,env_test,hash_test,random_test -Concurrency 16
+      if(!$?) { Exit $LASTEXITCODE }
+      echo ======================== Test RocksJava ========================
+      cd build\java
+      & ctest -C Debug -j 16
+      if(!$?) { Exit $LASTEXITCODE }
+    shell: pwsh

.github/workflows/benchmark-linux.yml

@@ -0,0 +1,15 @@
+name: facebook/rocksdb/benchmark-linux
+on: workflow_dispatch
+jobs:
+  # FIXME: when this job is fixed, it should be given a cron schedule like
+  # schedule:
+  # - cron: 0 * * * *
+  # workflow_dispatch:
+  benchmark-linux:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on: ubuntu-latest
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: "./.github/actions/build-for-benchmarks"
+    - uses: "./.github/actions/perform-benchmarks"
+    - uses: "./.github/actions/post-benchmarks"

.github/workflows/nightly-candidate.yml

@@ -0,0 +1,18 @@
+name: facebook/rocksdb/nightly
+on: workflow_dispatch
+jobs:
+  # These jobs would be in nightly but are failing or otherwise broken for
+  # some reason.
+  build-linux-arm-test-full:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: arm64large
+    container:
+      image: ubuntu-2004:202111-02
+      options: --shm-size=16gb
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: "./.github/actions/pre-steps"
+    - uses: "./.github/actions/install-gflags"
+    - run: make V=1 J=4 -j4 check
+    - uses: "./.github/actions/post-steps"

.github/workflows/nightly.yml

@@ -0,0 +1,98 @@
+name: facebook/rocksdb/nightly
+on:
+  schedule:
+  - cron: 0 9 * * *
+  workflow_dispatch:
+jobs:
+  build-format-compatible:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 16-core-ubuntu
+    container:
+      image: zjay437/rocksdb:0.6
+      options: --shm-size=16gb
+    steps:
+    - uses: actions/checkout@v4.1.0
+      with:
+        fetch-depth: 0 # Need full repo history
+        fetch-tags: true
+    - uses: "./.github/actions/setup-upstream"
+    - uses: "./.github/actions/pre-steps"
+    - name: test
+      run: |-
+        export TEST_TMPDIR=/dev/shm/rocksdb
+        rm -rf /dev/shm/rocksdb
+        mkdir /dev/shm/rocksdb
+        git config --global --add safe.directory /__w/rocksdb/rocksdb
+        tools/check_format_compatible.sh
+    - uses: "./.github/actions/post-steps"
+  build-linux-run-microbench:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 16-core-ubuntu
+    container:
+      image: zjay437/rocksdb:0.6
+      options: --shm-size=16gb
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: "./.github/actions/pre-steps"
+    - run: DEBUG_LEVEL=0 make -j32 run_microbench
+    - uses: "./.github/actions/post-steps"
+  build-linux-non-shm:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 16-core-ubuntu
+    container:
+      image: zjay437/rocksdb:0.6
+      options: --shm-size=16gb
+    env:
+      TEST_TMPDIR: "/tmp/rocksdb_test_tmp"
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: "./.github/actions/pre-steps"
+    - run: make V=1 -j32 check
+    - uses: "./.github/actions/post-steps"
+  build-linux-clang-13-asan-ubsan-with-folly:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 16-core-ubuntu
+    container:
+      image: zjay437/rocksdb:0.6
+      options: --shm-size=16gb
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: "./.github/actions/pre-steps"
+    - uses: "./.github/actions/setup-folly"
+    - uses: "./.github/actions/build-folly"
+    - run: CC=clang-13 CXX=clang++-13 LIB_MODE=static USE_CLANG=1 USE_FOLLY=1 COMPILE_WITH_UBSAN=1 COMPILE_WITH_ASAN=1 make -j32 check
+    - uses: "./.github/actions/post-steps"
+  build-linux-valgrind:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 16-core-ubuntu
+    container:
+      image: zjay437/rocksdb:0.6
+      options: --shm-size=16gb
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: "./.github/actions/pre-steps"
+    - run: PORTABLE=1 make V=1 -j32 valgrind_test
+    - uses: "./.github/actions/post-steps"
+  build-windows-vs2022-avx2:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on: windows-2022
+    env:
+      CMAKE_GENERATOR: Visual Studio 17 2022
+      CMAKE_PORTABLE: AVX2
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: "./.github/actions/windows-build-steps"
+  build-windows-vs2022:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on: windows-2022
+    env:
+      CMAKE_GENERATOR: Visual Studio 17 2022
+      CMAKE_PORTABLE: 1
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: "./.github/actions/windows-build-steps"

.github/workflows/pr-jobs-candidate.yml

@@ -0,0 +1,46 @@
+name: facebook/rocksdb/pr-jobs-candidate
+on: workflow_dispatch
+jobs:
+  # These jobs would be in pr-jobs but are failing or otherwise broken for
+  # some reason.
+  # =========================== ARM Jobs ============================ #
+  build-linux-arm:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: arm64large # GitHub hosted ARM runners do not yet exist
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: "./.github/actions/pre-steps"
+    - uses: "./.github/actions/install-gflags"
+    - run: ROCKSDBTESTS_PLATFORM_DEPENDENT=only make V=1 J=4 -j4 all_but_some_tests check_some
+    - uses: "./.github/actions/post-steps"
+  build-linux-arm-cmake-no_test_run:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: arm64large # GitHub hosted ARM runners do not yet exist
+    env:
+      JAVA_HOME: "/usr/lib/jvm/java-8-openjdk-arm64"
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: "./.github/actions/pre-steps"
+    - uses: "./.github/actions/install-gflags"
+    - name: Set Java Environment
+      run: |-
+        echo "JAVA_HOME=${JAVA_HOME}"
+        echo 'export PATH=$JAVA_HOME/bin:$PATH' >> $BASH_ENV
+        which java && java -version
+        which javac && javac -version
+    - name: Build with cmake
+      run: |-
+        mkdir build
+        cd build
+        cmake -DCMAKE_BUILD_TYPE=Release -DWITH_TESTS=0 -DWITH_GFLAGS=1 -DWITH_BENCHMARK_TOOLS=0 -DWITH_TOOLS=0 -DWITH_CORE_TOOLS=1 ..
+        make -j4
+    - name: Build Java with cmake
+      run: |-
+        rm -rf build
+        mkdir build
+        cd build
+        cmake -DJNI=1 -DCMAKE_BUILD_TYPE=Release -DWITH_GFLAGS=1 ..
+        make -j4 rocksdb rocksdbjni
+    - uses: "./.github/actions/post-steps"

.github/workflows/pr-jobs.yml

@@ -0,0 +1,608 @@
+name: facebook/rocksdb/pr-jobs
+on: [push, pull_request]
+jobs:
+  # NOTE: multiple workflows would be recommended, but the current GHA UI in
+  # PRs doesn't make it clear when there's an overall error with a workflow,
+  # making it easy to overlook something broken. Grouping everything into one
+  # workflow minimizes the problem because it will be suspicious if there are
+  # no GHA results.
+  #
+  # The if: ${{ github.repository_owner == 'facebook' }} lines prevent the
+  # jobs from attempting to run on repo forks, because of a few problems:
+  # * runs-on labels are repository (owner) specific, so the job might wait
+  # for days waiting for a runner that simply isn't available.
+  # * Pushes to branches on forks for pull requests (the normal process) would
+  # run the workflow jobs twice: once in the pull-from fork and once for the PR
+  # destination repo. This is wasteful and dumb.
+  # * It is not known how to avoid copy-pasting the line to each job,
+  # increasing the risk of misconfiguration, especially on forks that might
+  # want to run with this GHA setup.
+  #
+  # DEBUGGING WITH SSH: Temporarily add this as a job step, either before the
+  # step of interest without the "if:" line or after the failing step with the
+  # "if:" line. Then use ssh command printed in CI output.
+  #  - name: Setup tmate session # TEMPORARY!
+  #    if: ${{ failure() }}
+  #    uses: mxschmitt/action-tmate@v3
+  #    with:
+  #      limit-access-to-actor: true
+
+  # ======================== Fast Initial Checks ====================== #
+  check-format-and-targets:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on: ubuntu-latest
+    steps:
+    - uses: actions/checkout@v4.1.0
+      with:
+        fetch-depth: 0 # Need full checkout to determine merge base
+        fetch-tags: true
+    - uses: "./.github/actions/setup-upstream"
+    - name: Setup Python
+      uses: actions/setup-python@v5
+    - name: Install Dependencies
+      run: python -m pip install --upgrade pip
+    - name: Install argparse
+      run: pip install argparse
+    - name: Download clang-format-diff.py
+      run: wget https://raw.githubusercontent.com/llvm/llvm-project/release/12.x/clang/tools/clang-format/clang-format-diff.py
+    - name: Check format
+      run: VERBOSE_CHECK=1 make check-format
+    - name: Compare buckify output
+      run: make check-buck-targets
+    - name: Simple source code checks
+      run: make check-sources
+  # ========================= Linux With Tests ======================== #
+  build-linux:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 16-core-ubuntu
+    container:
+      image: zjay437/rocksdb:0.6
+      options: --shm-size=16gb
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: "./.github/actions/pre-steps"
+    - run: make V=1 J=32 -j32 check
+    - uses: "./.github/actions/post-steps"
+  build-linux-cmake-mingw:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 4-core-ubuntu
+    container:
+      image: zjay437/rocksdb:0.6
+      options: --shm-size=16gb
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: "./.github/actions/pre-steps"
+    - run: update-alternatives --set x86_64-w64-mingw32-g++ /usr/bin/x86_64-w64-mingw32-g++-posix
+    - name: Build cmake-mingw
+      run: |-
+        export PATH=$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
+    - uses: "./.github/actions/post-steps"
+  build-linux-cmake-with-folly:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 16-core-ubuntu
+    container:
+      image: zjay437/rocksdb:0.6
+      options: --shm-size=16gb
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: "./.github/actions/pre-steps"
+    - uses: "./.github/actions/setup-folly"
+    - uses: "./.github/actions/build-folly"
+    - run: "(mkdir build && cd build && cmake -DUSE_FOLLY=1 -DWITH_GFLAGS=1 -DROCKSDB_BUILD_SHARED=0 .. && make V=1 -j20 && ctest -j20)"
+    - uses: "./.github/actions/post-steps"
+  build-linux-cmake-with-folly-lite-no-test:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 16-core-ubuntu
+    container:
+      image: zjay437/rocksdb:0.6
+      options: --shm-size=16gb
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: "./.github/actions/pre-steps"
+    - uses: "./.github/actions/setup-folly"
+    - run: "(mkdir build && cd build && cmake -DUSE_FOLLY_LITE=1 -DWITH_GFLAGS=1 .. && make V=1 -j20)"
+    - uses: "./.github/actions/post-steps"
+  build-linux-gcc-7-with-folly:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 16-core-ubuntu
+    container:
+      image: zjay437/rocksdb:0.6
+      options: --shm-size=16gb
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: "./.github/actions/pre-steps"
+    - uses: "./.github/actions/setup-folly"
+    - uses: "./.github/actions/build-folly"
+    - run: USE_FOLLY=1 LIB_MODE=static CC=gcc-7 CXX=g++-7 V=1 make -j32 check
+    - uses: "./.github/actions/post-steps"
+  build-linux-gcc-7-with-folly-lite-no-test:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 16-core-ubuntu
+    container:
+      image: zjay437/rocksdb:0.6
+      options: --shm-size=16gb
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: "./.github/actions/pre-steps"
+    - uses: "./.github/actions/setup-folly"
+    - run: USE_FOLLY_LITE=1 CC=gcc-7 CXX=g++-7 V=1 make -j32 all
+    - uses: "./.github/actions/post-steps"
+  build-linux-cmake-with-folly-coroutines:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 16-core-ubuntu
+    container:
+      image: zjay437/rocksdb:0.6
+      options: --shm-size=16gb
+    env:
+      CC: gcc-10
+      CXX: g++-10
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: "./.github/actions/pre-steps"
+    - uses: "./.github/actions/setup-folly"
+    - uses: "./.github/actions/build-folly"
+    - run: "(mkdir build && cd build && cmake -DUSE_COROUTINES=1 -DWITH_GFLAGS=1 -DROCKSDB_BUILD_SHARED=0 .. && make V=1 -j20 && ctest -j20)"
+    - uses: "./.github/actions/post-steps"
+  build-linux-cmake-with-benchmark:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 16-core-ubuntu
+    container:
+      image: zjay437/rocksdb:0.6
+      options: --shm-size=16gb
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: "./.github/actions/pre-steps"
+    - run: mkdir build && cd build && cmake -DWITH_GFLAGS=1 -DWITH_BENCHMARK=1 .. && make V=1 -j20 && ctest -j20
+    - uses: "./.github/actions/post-steps"
+  build-linux-encrypted_env-no_compression:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 16-core-ubuntu
+    container:
+      image: zjay437/rocksdb:0.6
+      options: --shm-size=16gb
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: "./.github/actions/pre-steps"
+    - run: ENCRYPTED_ENV=1 ROCKSDB_DISABLE_SNAPPY=1 ROCKSDB_DISABLE_ZLIB=1 ROCKSDB_DISABLE_BZIP=1 ROCKSDB_DISABLE_LZ4=1 ROCKSDB_DISABLE_ZSTD=1 make V=1 J=32 -j32 check
+    - run: "./sst_dump --help | grep -E -q 'Supported compression types: kNoCompression$' # Verify no compiled in compression\n"
+    - uses: "./.github/actions/post-steps"
+  # ======================== Linux No Test Runs ======================= #
+  build-linux-release:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 16-core-ubuntu
+    container:
+      image: zjay437/rocksdb:0.6
+      options: --shm-size=16gb
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - run: make V=1 -j32 LIB_MODE=shared release
+    - run: ls librocksdb.so
+    - run: "./db_stress --version"
+    - run: make clean
+    - run: make V=1 -j32 release
+    - run: ls librocksdb.a
+    - run: "./db_stress --version"
+    - run: make clean
+    - run: apt-get remove -y libgflags-dev
+    - run: make V=1 -j32 LIB_MODE=shared release
+    - run: ls librocksdb.so
+    - run: if ./db_stress --version; then false; else true; fi
+    - run: make clean
+    - run: make V=1 -j32 release
+    - run: ls librocksdb.a
+    - run: if ./db_stress --version; then false; else true; fi
+    - uses: "./.github/actions/post-steps"
+  build-linux-release-rtti:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 8-core-ubuntu
+    container:
+      image: zjay437/rocksdb:0.6
+      options: --shm-size=16gb
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - run: USE_RTTI=1 DEBUG_LEVEL=0 make V=1 -j16 static_lib tools db_bench
+    - run: "./db_stress --version"
+    - run: make clean
+    - run: apt-get remove -y libgflags-dev
+    - run: USE_RTTI=1 DEBUG_LEVEL=0 make V=1 -j16 static_lib tools db_bench
+    - run: if ./db_stress --version; then false; else true; fi
+  build-examples:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 4-core-ubuntu
+    container:
+      image: zjay437/rocksdb:0.6
+      options: --shm-size=16gb
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: "./.github/actions/pre-steps"
+    - name: Build examples
+      run: make V=1 -j4 static_lib && cd examples && make V=1 -j4
+    - uses: "./.github/actions/post-steps"
+  build-fuzzers:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 4-core-ubuntu
+    container:
+      image: zjay437/rocksdb:0.6
+      options: --shm-size=16gb
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: "./.github/actions/pre-steps"
+    - name: Build rocksdb lib
+      run: CC=clang-13 CXX=clang++-13 USE_CLANG=1 make -j4 static_lib
+    - name: Build fuzzers
+      run: cd fuzz && make sst_file_writer_fuzzer db_fuzzer db_map_fuzzer
+    - uses: "./.github/actions/post-steps"
+  build-linux-clang-no_test_run:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 8-core-ubuntu
+    container:
+      image: zjay437/rocksdb:0.6
+      options: --shm-size=16gb
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - run: CC=clang CXX=clang++ USE_CLANG=1 PORTABLE=1 make V=1 -j16 all
+    - uses: "./.github/actions/post-steps"
+  build-linux-clang-13-no_test_run:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 16-core-ubuntu
+    container:
+      image: zjay437/rocksdb:0.6
+      options: --shm-size=16gb
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: "./.github/actions/pre-steps"
+    - run: CC=clang-13 CXX=clang++-13 USE_CLANG=1 make -j32 all microbench
+    - uses: "./.github/actions/post-steps"
+  build-linux-gcc-8-no_test_run:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 16-core-ubuntu
+    container:
+      image: zjay437/rocksdb:0.6
+      options: --shm-size=16gb
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: "./.github/actions/pre-steps"
+    - run: CC=gcc-8 CXX=g++-8 V=1 make -j32 all
+    - uses: "./.github/actions/post-steps"
+  build-linux-gcc-10-cxx20-no_test_run:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 16-core-ubuntu
+    container:
+      image: zjay437/rocksdb:0.6
+      options: --shm-size=16gb
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: "./.github/actions/pre-steps"
+    - run: CC=gcc-10 CXX=g++-10 V=1 ROCKSDB_CXX_STANDARD=c++20 make -j32 all
+    - uses: "./.github/actions/post-steps"
+  build-linux-gcc-11-no_test_run:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 16-core-ubuntu
+    container:
+      image: zjay437/rocksdb:0.6
+      options: --shm-size=16gb
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: "./.github/actions/pre-steps"
+    - run: LIB_MODE=static CC=gcc-11 CXX=g++-11 V=1 make -j32 all microbench
+    - uses: "./.github/actions/post-steps"
+  # ======================== Linux Other Checks ======================= #
+  build-linux-clang10-clang-analyze:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 16-core-ubuntu
+    container:
+      image: zjay437/rocksdb:0.6
+      options: --shm-size=16gb
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: "./.github/actions/pre-steps"
+    - 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
+    - uses: "./.github/actions/post-steps"
+    - name: compress test report
+      run: tar -cvzf scan_build_report.tar.gz scan_build_report
+      if: failure()
+    - uses: actions/upload-artifact@v4.0.0
+      with:
+        name: scan-build-report
+        path: scan_build_report.tar.gz
+  build-linux-unity-and-headers:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 4-core-ubuntu
+    container:
+      image: gcc:latest
+      options: --shm-size=16gb
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - run: apt-get update -y && apt-get install -y libgflags-dev
+    - name: Unity build
+      run: make V=1 -j8 unity_test
+    - run: make V=1 -j8 -k check-headers
+    - uses: "./.github/actions/post-steps"
+  build-linux-mini-crashtest:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 4-core-ubuntu
+    container:
+      image: zjay437/rocksdb:0.6
+      options: --shm-size=16gb
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: "./.github/actions/pre-steps"
+    - run: ulimit -S -n `ulimit -H -n` && make V=1 -j8 CRASH_TEST_EXT_ARGS='--duration=960 --max_key=2500000 --use_io_uring=0' blackbox_crash_test_with_atomic_flush
+    - uses: "./.github/actions/post-steps"
+  # ======================= Linux with Sanitizers ===================== #
+  build-linux-clang10-asan:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 32-core-ubuntu
+    container:
+      image: zjay437/rocksdb:0.6
+      options: --shm-size=16gb
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: "./.github/actions/pre-steps"
+    - run: COMPILE_WITH_ASAN=1 CC=clang-10 CXX=clang++-10 ROCKSDB_DISABLE_ALIGNED_NEW=1 USE_CLANG=1 make V=1 -j32 check
+    - uses: "./.github/actions/post-steps"
+  build-linux-clang10-ubsan:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 16-core-ubuntu
+    container:
+      image: zjay437/rocksdb:0.6
+      options: --shm-size=16gb
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: "./.github/actions/pre-steps"
+    - run: COMPILE_WITH_UBSAN=1 OPT="-fsanitize-blacklist=.circleci/ubsan_suppression_list.txt" CC=clang-10 CXX=clang++-10 ROCKSDB_DISABLE_ALIGNED_NEW=1 USE_CLANG=1 make V=1 -j32 ubsan_check
+    - uses: "./.github/actions/post-steps"
+  build-linux-clang10-mini-tsan:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 32-core-ubuntu
+    container:
+      image: zjay437/rocksdb:0.6
+      options: --shm-size=16gb
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: "./.github/actions/pre-steps"
+    - run: COMPILE_WITH_TSAN=1 CC=clang-13 CXX=clang++-13 ROCKSDB_DISABLE_ALIGNED_NEW=1 USE_CLANG=1 make V=1 -j32 check
+    - uses: "./.github/actions/post-steps"
+  build-linux-static_lib-alt_namespace-status_checked:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 16-core-ubuntu
+    container:
+      image: zjay437/rocksdb:0.6
+      options: --shm-size=16gb
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: "./.github/actions/pre-steps"
+    - run: ASSERT_STATUS_CHECKED=1 TEST_UINT128_COMPAT=1 ROCKSDB_MODIFY_NPHASH=1 LIB_MODE=static OPT="-DROCKSDB_NAMESPACE=alternative_rocksdb_ns" make V=1 -j24 check
+    - uses: "./.github/actions/post-steps"
+  # ========================= MacOS with Tests ======================== #
+  build-macos:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on: macos-13
+    env:
+      ROCKSDB_DISABLE_JEMALLOC: 1
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: maxim-lobanov/setup-xcode@v1.6.0
+      with:
+        xcode-version: 14.3.1
+    - uses: "./.github/actions/increase-max-open-files-on-macos"
+    - uses: "./.github/actions/install-gflags-on-macos"
+    - uses: "./.github/actions/pre-steps-macos"
+    - name: Run all tests
+      run: ulimit -S -n `ulimit -H -n` && make V=1 J=16 -j16 all
+    - uses: "./.github/actions/post-steps"
+  build-macos-cmake:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on: macos-13
+    strategy:
+      matrix:
+        run_even_tests: [true, false]
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: maxim-lobanov/setup-xcode@v1.6.0
+      with:
+        xcode-version: 14.3.1
+    - uses: "./.github/actions/increase-max-open-files-on-macos"
+    - uses: "./.github/actions/install-gflags-on-macos"
+    - uses: "./.github/actions/pre-steps-macos"
+    - name: cmake generate project file
+      run: ulimit -S -n `ulimit -H -n` && mkdir build && cd build && cmake -DWITH_GFLAGS=1 ..
+    - name: Build tests
+      run: cd build && make V=1 -j16
+    - name: Run even tests
+      run: ulimit -S -n `ulimit -H -n` && cd build && ctest -j16 -I 0,,2
+      if: ${{ matrix.run_even_tests }}
+    - name: Run odd tests
+      run: ulimit -S -n `ulimit -H -n` && cd build && ctest -j16 -I 1,,2
+      if: ${{ ! matrix.run_even_tests  }}
+    - uses: "./.github/actions/post-steps"
+  # ======================== Windows with Tests ======================= #
+  # NOTE: some windows jobs are in "nightly" to save resources
+  build-windows-vs2019:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on: windows-2019
+    env:
+      CMAKE_GENERATOR: Visual Studio 16 2019
+      CMAKE_PORTABLE: 1
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: "./.github/actions/windows-build-steps"
+  # ============================ Java Jobs ============================ #
+  build-linux-java:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 4-core-ubuntu
+    container:
+      image: evolvedbinary/rocksjava:centos6_x64-be
+      options: --shm-size=16gb
+    steps:
+    # The docker image is intentionally based on an OS that has an older GLIBC version.
+    # That GLIBC is incompatibile with GitHub's actions/checkout. Thus we implement a manual checkout step.
+    - name: Checkout
+      env:
+        GH_TOKEN: ${{ github.token }}
+      run: |
+        chown `whoami` . || true
+        git clone --no-checkout https://oath2:$GH_TOKEN@github.com/${{ github.repository }}.git .
+        git -c protocol.version=2 fetch --update-head-ok --no-tags --prune --no-recurse-submodules --depth=1 origin +${{ github.sha }}:${{ github.ref }}
+        git checkout --progress --force ${{ github.ref }}
+        git log -1 --format='%H'
+    - uses: "./.github/actions/pre-steps"
+    - name: Set Java Environment
+      run: |-
+        echo "JAVA_HOME=${JAVA_HOME}"
+        which java && java -version
+        which javac && javac -version
+    - name: Test RocksDBJava
+      run: scl enable devtoolset-7 'make V=1 J=8 -j8 jtest'
+    # NOTE: post-steps skipped because of compatibility issues with docker image
+  build-linux-java-static:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 4-core-ubuntu
+    container:
+      image: evolvedbinary/rocksjava:centos6_x64-be
+      options: --shm-size=16gb
+    steps:
+    # The docker image is intentionally based on an OS that has an older GLIBC version.
+    # That GLIBC is incompatibile with GitHub's actions/checkout. Thus we implement a manual checkout step.
+    - name: Checkout
+      env:
+        GH_TOKEN: ${{ github.token }}
+      run: |
+        chown `whoami` . || true
+        git clone --no-checkout https://oath2:$GH_TOKEN@github.com/${{ github.repository }}.git .
+        git -c protocol.version=2 fetch --update-head-ok --no-tags --prune --no-recurse-submodules --depth=1 origin +${{ github.sha }}:${{ github.ref }}
+        git checkout --progress --force ${{ github.ref }}
+        git log -1 --format='%H'
+    - uses: "./.github/actions/pre-steps"
+    - name: Set Java Environment
+      run: |-
+        echo "JAVA_HOME=${JAVA_HOME}"
+        which java && java -version
+        which javac && javac -version
+    - name: Build RocksDBJava Static Library
+      run: scl enable devtoolset-7 'make V=1 J=8 -j8 rocksdbjavastatic'
+    # NOTE: post-steps skipped because of compatibility issues with docker image
+  build-macos-java:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on: macos-13
+    env:
+      JAVA_HOME: "/Library/Java/JavaVirtualMachines/liberica-jdk-8.jdk/Contents/Home"
+      ROCKSDB_DISABLE_JEMALLOC: 1
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: maxim-lobanov/setup-xcode@v1.6.0
+      with:
+        xcode-version: 14.3.1
+    - uses: "./.github/actions/increase-max-open-files-on-macos"
+    - uses: "./.github/actions/install-gflags-on-macos"
+    - uses: "./.github/actions/install-jdk8-on-macos"
+    - uses: "./.github/actions/pre-steps-macos"
+    - name: Set Java Environment
+      run: |-
+        echo "JAVA_HOME=${JAVA_HOME}"
+        which java && java -version
+        which javac && javac -version
+    - name: Test RocksDBJava
+      run: make V=1 J=16 -j16 jtest
+    - uses: "./.github/actions/post-steps"
+  build-macos-java-static:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on: macos-13
+    env:
+      JAVA_HOME: "/Library/Java/JavaVirtualMachines/liberica-jdk-8.jdk/Contents/Home"
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: maxim-lobanov/setup-xcode@v1.6.0
+      with:
+        xcode-version: 14.3.1
+    - uses: "./.github/actions/increase-max-open-files-on-macos"
+    - uses: "./.github/actions/install-gflags-on-macos"
+    - uses: "./.github/actions/install-jdk8-on-macos"
+    - uses: "./.github/actions/pre-steps-macos"
+    - name: Set Java Environment
+      run: |-
+        echo "JAVA_HOME=${JAVA_HOME}"
+        which java && java -version
+        which javac && javac -version
+    - name: Build RocksDBJava x86 and ARM Static Libraries
+      run: make V=1 J=16 -j16 rocksdbjavastaticosx
+    - uses: "./.github/actions/post-steps"
+  build-macos-java-static-universal:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on: macos-13
+    env:
+      JAVA_HOME: "/Library/Java/JavaVirtualMachines/liberica-jdk-8.jdk/Contents/Home"
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: maxim-lobanov/setup-xcode@v1.6.0
+      with:
+        xcode-version: 14.3.1
+    - uses: "./.github/actions/increase-max-open-files-on-macos"
+    - uses: "./.github/actions/install-gflags-on-macos"
+    - uses: "./.github/actions/install-jdk8-on-macos"
+    - uses: "./.github/actions/pre-steps-macos"
+    - name: Set Java Environment
+      run: |-
+        echo "JAVA_HOME=${JAVA_HOME}"
+        which java && java -version
+        which javac && javac -version
+    - name: Build RocksDBJava Universal Binary Static Library
+      run: make V=1 J=16 -j16 rocksdbjavastaticosx_ub
+    - uses: "./.github/actions/post-steps"
+  build-linux-java-pmd:
+    if: ${{ github.repository_owner == 'facebook' }}
+    runs-on:
+      labels: 4-core-ubuntu
+    container:
+      image: evolvedbinary/rocksjava:rockylinux8_x64-be
+      options: --shm-size=16gb
+    steps:
+    - uses: actions/checkout@v4.1.0
+    - uses: "./.github/actions/install-maven"
+    - uses: "./.github/actions/pre-steps"
+    - name: Set Java Environment
+      run: |-
+        echo "JAVA_HOME=${JAVA_HOME}"
+        which java && java -version
+        which javac && javac -version
+    - name: PMD RocksDBJava
+      run: make V=1 J=8 -j8 jpmd
+    - uses: actions/upload-artifact@v4.0.0
+      with:
+        name: pmd-report
+        path: "${{ github.workspace }}/java/target/pmd.xml"
+    - uses: actions/upload-artifact@v4.0.0
+      with:
+        name: maven-site
+        path: "${{ github.workspace }}/java/target/site"

.github/workflows/sanity_check.yml

@@ -1,45 +0,0 @@
-name: Check buck targets and code format
-on: [push, pull_request]
-permissions:
-  contents: read
-
-jobs:
-  check:
-    name: Check TARGETS file and code format
-    runs-on: ubuntu-latest
-    steps:
-    - name: Checkout feature branch
-      uses: actions/checkout@v2
-      with:
-        fetch-depth: 0
-
-    - name: Fetch from upstream
-      run: |
-        git remote add upstream https://github.com/facebook/rocksdb.git && git fetch upstream
-
-    - name: Where am I
-      run: |
-        echo git status && git status
-        echo "git remote -v" && git remote -v
-        echo git branch && git branch
-
-    - name: Setup Python
-      uses: actions/setup-python@v1
-
-    - name: Install Dependencies
-      run: python -m pip install --upgrade pip
-
-    - name: Install argparse
-      run: pip install argparse
-
-    - name: Download clang-format-diff.py
-      run: wget https://raw.githubusercontent.com/llvm/llvm-project/release/12.x/clang/tools/clang-format/clang-format-diff.py
-
-    - name: Check format
-      run: VERBOSE_CHECK=1 make check-format
-
-    - name: Compare buckify output
-      run: make check-buck-targets
-
-    - name: Simple source code checks
-      run: make check-sources

CMakeLists.txt

@@ -76,11 +76,6 @@ if (WITH_WINDOWS_UTF8_FILENAMES)
 endif()
 option(ROCKSDB_BUILD_SHARED "Build shared versions of the RocksDB libraries" ON)
 
-if ($ENV{CIRCLECI})
-  message(STATUS "Build for CircieCI env, a few tests may be disabled")
-  add_definitions(-DCIRCLECI)
-endif()
-
 if( NOT DEFINED CMAKE_CXX_STANDARD )
   set(CMAKE_CXX_STANDARD 17)
 endif()
@@ -772,6 +767,7 @@ set(SOURCES
         options/configurable.cc
         options/customizable.cc
         options/db_options.cc
+        options/offpeak_time_info.cc
         options/options.cc
         options/options_helper.cc
         options/options_parser.cc
@@ -1105,11 +1101,15 @@ set(BUILD_VERSION_CC ${CMAKE_BINARY_DIR}/build_version.cc)
 configure_file(util/build_version.cc.in ${BUILD_VERSION_CC} @ONLY)
 
 add_library(${ROCKSDB_STATIC_LIB} STATIC ${SOURCES} ${BUILD_VERSION_CC})
+target_include_directories(${ROCKSDB_STATIC_LIB} PUBLIC
+  $<BUILD_INTERFACE:${PROJECT_SOURCE_DIR}/include>)
 target_link_libraries(${ROCKSDB_STATIC_LIB} PRIVATE
   ${THIRDPARTY_LIBS} ${SYSTEM_LIBS})
 
 if(ROCKSDB_BUILD_SHARED)
   add_library(${ROCKSDB_SHARED_LIB} SHARED ${SOURCES} ${BUILD_VERSION_CC})
+  target_include_directories(${ROCKSDB_SHARED_LIB} PUBLIC
+    $<BUILD_INTERFACE:${PROJECT_SOURCE_DIR}/include>)
   target_link_libraries(${ROCKSDB_SHARED_LIB} PRIVATE
     ${THIRDPARTY_LIBS} ${SYSTEM_LIBS})
 
@@ -1423,6 +1423,7 @@ if(WITH_TESTS)
         util/ribbon_test.cc
         util/slice_test.cc
         util/slice_transform_test.cc
+        util/string_util_test.cc
         util/timer_queue_test.cc
         util/timer_test.cc
         util/thread_list_test.cc
@@ -1472,7 +1473,7 @@ if(WITH_TESTS)
       utilities/cassandra/test_utils.cc
   )
   enable_testing()
-  add_custom_target(check COMMAND ${CMAKE_CTEST_COMMAND})
+  add_custom_target(rocksdb_check COMMAND ${CMAKE_CTEST_COMMAND})
   set(TESTUTILLIB testutillib${ARTIFACT_SUFFIX})
   add_library(${TESTUTILLIB} STATIC ${TESTUTIL_SOURCE})
   target_link_libraries(${TESTUTILLIB} ${ROCKSDB_LIB} ${FOLLY_LIBS})
@@ -1497,7 +1498,7 @@ if(WITH_TESTS)
       target_link_libraries(${exename}${ARTIFACT_SUFFIX} testutillib${ARTIFACT_SUFFIX} testharness gtest ${THIRDPARTY_LIBS} ${ROCKSDB_LIB})
       if(NOT "${exename}" MATCHES "db_sanity_test")
         gtest_discover_tests(${exename} DISCOVERY_TIMEOUT 120)
-        add_dependencies(check ${exename}${ARTIFACT_SUFFIX})
+        add_dependencies(rocksdb_check ${exename}${ARTIFACT_SUFFIX})
       endif()
   endforeach(sourcefile ${TESTS})
 
@@ -1517,7 +1518,7 @@ if(WITH_TESTS)
     add_executable(c_test db/c_test.c)
     target_link_libraries(c_test ${ROCKSDB_LIB_FOR_C} testharness)
     add_test(NAME c_test COMMAND c_test${ARTIFACT_SUFFIX})
-    add_dependencies(check c_test)
+    add_dependencies(rocksdb_check c_test)
   endif()
 endif()
 
@@ -1596,6 +1597,3 @@ option(WITH_BENCHMARK "build benchmark tests" OFF)
 if(WITH_BENCHMARK)
   add_subdirectory(${PROJECT_SOURCE_DIR}/microbench/)
 endif()
-
-target_include_directories(${PROJECT_NAME} PUBLIC
-  $<BUILD_INTERFACE:${PROJECT_SOURCE_DIR}/include>)

HISTORY.md

@@ -1,6 +1,137 @@
 # Rocksdb Change Log
 > NOTE: Entries for next release do not go here. Follow instructions in `unreleased_history/README.txt`
 
+## 8.11.5 (11/26/2025)
+### Behavior Changes
+* Backport new value types `kTypeValuePreferredSeqno` and `kTypeColumnFamilyValuePreferredSeqno` introduced in RocksDB 9.1. This fixes a forward compatibility issue to allow reading MANIFEST with entries containing kTypeMaxValid written by RocksDB 9.1+.
+
+## 8.11.4 (04/09/2024)
+### Bug Fixes
+* Fixed CMake Javadoc build
+* Fixed Java `SstFileMetaData` to prevent throwing `java.lang.NoSuchMethodError`
+
+## 8.11.3 (02/27/2024)
+* Correct CMake Javadoc and source jar builds
+
+## 8.11.2 (02/16/2024)
+* Update zlib to 1.3.1 for Java builds
+
+## 8.11.1 (01/25/2024)
+### Bug Fixes
+* Fix a bug where older data of an ingested key can be returned for read when universal compaction is used
+* Apply appropriate rate limiting and priorities in more places.
+
+## 8.11.0 (01/19/2024)
+### New Features
+* Add new statistics: `rocksdb.sst.write.micros` measures time of each write to SST file; `rocksdb.file.write.{flush|compaction|db.open}.micros` measure time of each write to SST table (currently only block-based table format) and blob file for flush, compaction and db open.
+
+### Public API Changes
+* Added another enumerator `kVerify` to enum class `FileOperationType` in listener.h. Update your `switch` statements as needed.
+* Add CompressionOptions to the CompressedSecondaryCacheOptions structure to allow users to specify library specific options when creating the compressed secondary cache.
+* Deprecated several options: `level_compaction_dynamic_file_size`, `ignore_max_compaction_bytes_for_input`, `check_flush_compaction_key_order`, `flush_verify_memtable_count`, `compaction_verify_record_count`, `fail_if_options_file_error`, and `enforce_single_del_contracts`
+* Exposed options ttl via c api.
+
+### Behavior Changes
+* `rocksdb.blobdb.blob.file.write.micros` expands to also measure time writing the header and footer. Therefore the COUNT may be higher and values may be smaller than before. For stacked BlobDB, it no longer measures the time of explictly flushing blob file.
+* Files will be compacted to the next level if the data age exceeds periodic_compaction_seconds except for the last level.
+* Reduced the compaction debt ratio trigger for scheduling parallel compactions
+* For leveled compaction with default compaction pri (kMinOverlappingRatio), files marked for compaction will be prioritized over files not marked when picking a file from a level for compaction.
+
+### Bug Fixes
+* Fix bug in auto_readahead_size that combined with IndexType::kBinarySearchWithFirstKey + fails or iterator lands at a wrong key
+* Fixed some cases in which DB file corruption was detected but ignored on creating a backup with BackupEngine.
+* Fix bugs where `rocksdb.blobdb.blob.file.synced` includes blob files failed to get synced and `rocksdb.blobdb.blob.file.bytes.written` includes blob bytes failed to get written.
+* Fixed a possible memory leak or crash on a failure (such as I/O error) in automatic atomic flush of multiple column families.
+* Fixed some cases of in-memory data corruption using mmap reads with `BackupEngine`, `sst_dump`, or `ldb`.
+* Fixed issues with experimental `preclude_last_level_data_seconds` option that could interfere with expected data tiering.
+* Fixed the handling of the edge case when all existing blob files become unreferenced. Such files are now correctly deleted.
+
+## 8.10.0 (12/15/2023)
+### New Features
+* Provide support for async_io to trim readahead_size by doing block cache lookup
+* Added initial wide-column support in `WriteBatchWithIndex`. This includes the `PutEntity` API and support for wide columns in the existing read APIs (`GetFromBatch`, `GetFromBatchAndDB`, `MultiGetFromBatchAndDB`, and `BaseDeltaIterator`).
+
+### Public API Changes
+* Custom implementations of `TablePropertiesCollectorFactory` may now return a `nullptr` collector to decline processing a file, reducing callback overheads in such cases.
+
+### Behavior Changes
+* Make ReadOptions.auto_readahead_size default true which does prefetching optimizations for forward scans if iterate_upper_bound and block_cache is also specified.
+* Compactions can be scheduled in parallel in an additional scenario: high compaction debt relative to the data size
+* HyperClockCache now has built-in protection against excessive CPU consumption under the extreme stress condition of no (or very few) evictable cache entries, which can slightly increase memory usage such conditions. New option `HyperClockCacheOptions::eviction_effort_cap` controls the space-time trade-off of the response. The default should be generally well-balanced, with no measurable affect on normal operation.
+
+### Bug Fixes
+* Fix a corner case with auto_readahead_size where Prev Operation returns NOT SUPPORTED error when scans direction is changed from forward to backward.
+* Avoid destroying the periodic task scheduler's default timer in order to prevent static destruction order issues.
+* Fix double counting of BYTES_WRITTEN ticker when doing writes with transactions.
+* Fix a WRITE_STALL counter that was reporting wrong value in few cases.
+* A lookup by MultiGet in a TieredCache that goes to the local flash cache and finishes with very low latency, i.e before the subsequent call to WaitAll, is ignored, resulting in a false negative and a memory leak.
+
+### Performance Improvements
+* Java API extensions to improve consistency and completeness of APIs
+1 Extended `RocksDB.get([ColumnFamilyHandle columnFamilyHandle,] ReadOptions opt, ByteBuffer key, ByteBuffer value)` which now accepts indirect buffer parameters as well as direct buffer parameters
+2 Extended `RocksDB.put( [ColumnFamilyHandle columnFamilyHandle,] WriteOptions writeOpts, final ByteBuffer key, final ByteBuffer value)` which now accepts indirect buffer parameters as well as direct buffer parameters
+3 Added `RocksDB.merge([ColumnFamilyHandle columnFamilyHandle,] WriteOptions writeOptions, ByteBuffer key, ByteBuffer value)` methods with the same parameter options as `put(...)` - direct and indirect buffers are supported
+4 Added `RocksIterator.key( byte[] key [, int offset, int len])` methods which retrieve the iterator key into the supplied buffer
+5 Added `RocksIterator.value( byte[] value [, int offset, int len])` methods which retrieve the iterator value into the supplied buffer
+6 Deprecated `get(final ColumnFamilyHandle columnFamilyHandle, final ReadOptions readOptions, byte[])` in favour of `get(final ReadOptions readOptions, final ColumnFamilyHandle columnFamilyHandle, byte[])` which has consistent parameter ordering with other methods in the same class
+7 Added `Transaction.get( ReadOptions opt, [ColumnFamilyHandle columnFamilyHandle, ] byte[] key, byte[] value)` methods which retrieve the requested value into the supplied buffer
+8 Added `Transaction.get( ReadOptions opt, [ColumnFamilyHandle columnFamilyHandle, ] ByteBuffer key, ByteBuffer value)` methods which retrieve the requested value into the supplied buffer
+9 Added `Transaction.getForUpdate( ReadOptions readOptions, [ColumnFamilyHandle columnFamilyHandle, ] byte[] key, byte[] value, boolean exclusive [, boolean doValidate])` methods which retrieve the requested value into the supplied buffer
+10 Added `Transaction.getForUpdate( ReadOptions readOptions, [ColumnFamilyHandle columnFamilyHandle, ] ByteBuffer key, ByteBuffer value, boolean exclusive [, boolean doValidate])` methods which retrieve the requested value into the supplied buffer
+11 Added `Transaction.getIterator()` method as a convenience which defaults the `ReadOptions` value supplied to existing `Transaction.iterator()` methods. This mirrors the existing `RocksDB.iterator()` method.
+12 Added `Transaction.put([ColumnFamilyHandle columnFamilyHandle, ]  ByteBuffer key, ByteBuffer value [, boolean assumeTracked])` methods which supply the key, and the value to be written in a `ByteBuffer` parameter
+13 Added `Transaction.merge([ColumnFamilyHandle columnFamilyHandle, ] ByteBuffer key, ByteBuffer value [,  boolean assumeTracked])` methods which supply the key, and the value to be written/merged in a `ByteBuffer` parameter
+14 Added `Transaction.mergeUntracked([ColumnFamilyHandle columnFamilyHandle, ] ByteBuffer key, ByteBuffer value)` methods which supply the key, and the value to be written/merged in a `ByteBuffer` parameter
+
+
+## 8.9.0 (11/17/2023)
+### New Features
+* Add GetEntity() and PutEntity() API implementation for Attribute Group support. Through the use of Column Families, AttributeGroup enables users to logically group wide-column entities.
+
+### Public API Changes
+* Added rocksdb_ratelimiter_create_auto_tuned API to create an auto-tuned GenericRateLimiter.
+* Added clipColumnFamily() to the Java API to clip the entries in the CF according to the range [begin_key, end_key).
+* Make the `EnableFileDeletion` API not default to force enabling. For users that rely on this default behavior and still
+want to continue to use force enabling, they need to explicitly pass a `true` to `EnableFileDeletion`.
+* Add new Cache APIs GetSecondaryCacheCapacity() and GetSecondaryCachePinnedUsage() to return the configured capacity, and cache reservation charged to the secondary cache.
+
+### Behavior Changes
+* During off-peak hours defined by `daily_offpeak_time_utc`, the compaction picker will select a larger number of files for periodic compaction. This selection will include files that are projected to expire by the next off-peak start time, ensuring that these files are not chosen for periodic compaction outside of off-peak hours.
+* If an error occurs when writing to a trace file after `DB::StartTrace()`, the subsequent trace writes are skipped to avoid writing to a file that has previously seen error. In this case, `DB::EndTrace()` will also return a non-ok status with info about the error occured previously in its status message.
+* Deleting stale files upon recovery are delegated to SstFileManger if available so they can be rate limited.
+* Make RocksDB only call `TablePropertiesCollector::Finish()` once.
+* When `WAL_ttl_seconds > 0`, we now process archived WALs for deletion at least every `WAL_ttl_seconds / 2` seconds. Previously it could be less frequent in case of small `WAL_ttl_seconds` values when size-based expiration (`WAL_size_limit_MB > 0 `) was simultaneously enabled.
+
+### Bug Fixes
+* Fixed a crash or assertion failure bug in experimental new HyperClockCache variant, especially when running with a SecondaryCache.
+* Fix a race between flush error recovery and db destruction that can lead to db crashing.
+* Fixed some bugs in the index builder/reader path for user-defined timestamps in Memtable only feature.
+
+## 8.8.0 (10/23/2023)
+### New Features
+* Introduce AttributeGroup by adding the first AttributeGroup support API, MultiGetEntity(). Through the use of Column Families, AttributeGroup enables users to logically group wide-column entities. More APIs to support AttributeGroup will come soon, including GetEntity, PutEntity, and others.
+* Added new tickers `rocksdb.fifo.{max.size|ttl}.compactions` to count FIFO compactions that drop files for different reasons
+* Add an experimental offpeak duration awareness by setting `DBOptions::daily_offpeak_time_utc` in "HH:mm-HH:mm" format. This information will be used for resource optimization in the future
+* Users can now change the max bytes granted in a single refill period (i.e, burst) during runtime by `SetSingleBurstBytes()` for RocksDB rate limiter
+
+### Public API Changes
+* The default value of `DBOptions::fail_if_options_file_error` changed from `false` to `true`. Operations that set in-memory options (e.g., `DB::Open*()`, `DB::SetOptions()`, `DB::CreateColumnFamily*()`, and `DB::DropColumnFamily()`) but fail to persist the change will now return a non-OK `Status` by default.
+
+### Behavior Changes
+* For non direct IO, eliminate the file system prefetching attempt for compaction read when `Options::compaction_readahead_size` is 0
+* During a write stop, writes now block on in-progress recovery attempts
+
+### Bug Fixes
+* Fix a bug in auto_readahead_size where first_internal_key of index blocks wasn't copied properly resulting in corruption error when first_internal_key was used for comparison.
+* Fixed a bug where compaction read under non direct IO still falls back to RocksDB internal prefetching after file system's prefetching returns non-OK status other than `Status::NotSupported()`
+* Add bounds check in WBWIIteratorImpl and make BaseDeltaIterator, WriteUnpreparedTxn and WritePreparedTxn respect the upper bound and lower bound in ReadOption. See 11680.
+* Fixed the handling of wide-column base values in the `max_successive_merges` logic.
+* Fixed a rare race bug involving a concurrent combination of Create/DropColumnFamily and/or Set(DB)Options that could lead to inconsistency between (a) the DB's reported options state, (b) the DB options in effect, and (c) the latest persisted OPTIONS file.
+* Fixed a possible underflow when computing the compressed secondary cache share of memory reservations while updating the compressed secondary to total block cache ratio.
+
+### Performance Improvements
+* Improved the I/O efficiency of DB::Open a new DB with `create_missing_column_families=true` and many column families.
+
 ## 8.7.0 (09/22/2023)
 ### New Features
 * Added an experimental new "automatic" variant of HyperClockCache that does not require a prior estimate of the average size of cache entries. This variant is activated when HyperClockCacheOptions::estimated\_entry\_charge = 0 and has essentially the same concurrency benefits as the existing HyperClockCache.

Makefile

@@ -539,7 +539,7 @@ endif
 
 ifdef USE_CLANG
 	# Used by some teams in Facebook
-	WARNING_FLAGS += -Wshift-sign-overflow -Wambiguous-reversed-operator
+	WARNING_FLAGS += -Wshift-sign-overflow -Wambiguous-reversed-operator -Wimplicit-fallthrough
 endif
 
 ifeq ($(PLATFORM), OS_OPENBSD)
@@ -994,7 +994,7 @@ endif
 
 .PHONY: check_0
 check_0:
-	printf '%s\n' ''						\
+	@printf '%s\n' ''						\
 	  'To monitor subtest <duration,pass/fail,name>,'		\
 	  '  run "make watch-log" in a separate window' '';		\
 	{ \
@@ -1016,7 +1016,7 @@ valgrind-exclude-regexp = InlineSkipTest.ConcurrentInsert|TransactionStressTest.
 .PHONY: valgrind_check_0
 valgrind_check_0: test_log_prefix := valgrind_
 valgrind_check_0:
-	printf '%s\n' ''						\
+	@printf '%s\n' ''						\
 	  'To monitor subtest <duration,pass/fail,name>,'		\
 	  '  run "make watch-log" in a separate window' '';		\
 	{								\
@@ -1768,6 +1768,9 @@ cuckoo_table_db_test: $(OBJ_DIR)/db/cuckoo_table_db_test.o $(TEST_LIBRARY) $(LIB
 listener_test: $(OBJ_DIR)/db/listener_test.o $(TEST_LIBRARY) $(LIBRARY)
 	$(AM_LINK)
 
+string_util_test: $(OBJ_DIR)/util/string_util_test.o $(TEST_LIBRARY) $(LIBRARY)
+	$(AM_LINK)
+
 thread_list_test: $(OBJ_DIR)/util/thread_list_test.o $(TEST_LIBRARY) $(LIBRARY)
 	$(AM_LINK)
 
@@ -2060,7 +2063,7 @@ JAVA_INCLUDE = -I$(JAVA_HOME)/include/ -I$(JAVA_HOME)/include/linux
 ifeq ($(PLATFORM), OS_SOLARIS)
 	ARCH := $(shell isainfo -b)
 else ifeq ($(PLATFORM), OS_OPENBSD)
-	ifneq (,$(filter amd64 ppc64 ppc64le s390x arm64 aarch64 sparc64 loongarch64, $(MACHINE)))
+	ifneq (,$(filter amd64 ppc64 ppc64le s390x arm64 aarch64 riscv64 sparc64 loongarch64, $(MACHINE)))
 		ARCH := 64
 	else
 		ARCH := 32
@@ -2081,7 +2084,7 @@ ifneq ($(origin JNI_LIBC), undefined)
 endif
 
 ifeq (,$(ROCKSDBJNILIB))
-ifneq (,$(filter ppc% s390x arm64 aarch64 sparc64 loongarch64, $(MACHINE)))
+ifneq (,$(filter ppc% s390x arm64 aarch64 riscv64 sparc64 loongarch64, $(MACHINE)))
 	ROCKSDBJNILIB = librocksdbjni-linux-$(MACHINE)$(JNI_LIBC_POSTFIX).so
 else
 	ROCKSDBJNILIB = librocksdbjni-linux$(ARCH)$(JNI_LIBC_POSTFIX).so
@@ -2094,8 +2097,8 @@ ROCKSDB_JAVADOCS_JAR = rocksdbjni-$(ROCKSDB_JAVA_VERSION)-javadoc.jar
 ROCKSDB_SOURCES_JAR = rocksdbjni-$(ROCKSDB_JAVA_VERSION)-sources.jar
 SHA256_CMD = sha256sum
 
-ZLIB_VER ?= 1.3
-ZLIB_SHA256 ?= ff0ba4c292013dbc27530b3a81e1f9a813cd39de01ca5e0f8bf355702efa593e
+ZLIB_VER ?= 1.3.1
+ZLIB_SHA256 ?= 9a93b2b7dfdac77ceba5a558a580e74667dd6fede4585b91eefb60f03b72df23
 ZLIB_DOWNLOAD_BASE ?= http://zlib.net
 BZIP2_VER ?= 1.0.8
 BZIP2_SHA256 ?= ab5a03176ee106d3f0fa90e381da478ddae405918153cca248e682cd0c4a2269
@@ -2107,7 +2110,7 @@ LZ4_VER ?= 1.9.4
 LZ4_SHA256 ?= 0b0e3aa07c8c063ddf40b082bdf7e37a1562bda40a0ff5272957f3e987e0e54b
 LZ4_DOWNLOAD_BASE ?= https://github.com/lz4/lz4/archive
 ZSTD_VER ?= 1.5.5
-ZSTD_SHA256 ?= 9c4396cc829cfae319a6e2615202e82aad41372073482fce286fac78646d3ee4
+ZSTD_SHA256 ?= 98e9c3d949d1b924e28e01eccb7deed865eefebf25c2f21c702e5cd5b63b85e1
 ZSTD_DOWNLOAD_BASE ?= https://github.com/facebook/zstd/archive
 CURL_SSL_OPTS ?= --tlsv1
 
@@ -2346,6 +2349,10 @@ rocksdbjavastaticdockers390x:
 	mkdir -p java/target
 	docker run --rm --name rocksdb_linux_s390x-be --attach stdin --attach stdout --attach stderr --volume $(HOME)/.m2:/root/.m2:ro --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) evolvedbinary/rocksjava:ubuntu18_s390x-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh
 
+rocksdbjavastaticdockerriscv64:
+	mkdir -p java/target
+	docker run --rm --name rocksdb_linux_riscv64-be --attach stdin --attach stdout --attach stderr --volume $(HOME)/.m2:/root/.m2:ro --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) evolvedbinary/rocksjava:ubuntu20_riscv64-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh
+
 rocksdbjavastaticdockerx86musl:
 	mkdir -p java/target
 	docker run --rm --name rocksdb_linux_x86-musl-be --platform linux/386 --attach stdin --attach stdout --attach stderr --volume $(HOME)/.m2:/root/.m2:ro --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) evolvedbinary/rocksjava:alpine3_x86-be /rocksdb-host/java/crossbuild/docker-build-linux-alpine.sh

PLUGINS.md

@@ -5,4 +5,5 @@ This is the list of all known third-party plugins for RocksDB. If something is m
 * [ZenFS](https://github.com/westerndigitalcorporation/zenfs): a file system for zoned block devices
 * [RADOS](https://github.com/riversand963/rocksdb-rados-env): an Env used for interacting with RADOS. Migrated from RocksDB main repo.
 * [PMEM](https://github.com/pmem/pmem-rocksdb-plugin): a collection of plugins to enable Persistent Memory on RocksDB.
-* [IPPCP](https://github.com/intel/ippcp-plugin-rocksdb): a plugin to enable encryption on RocksDB based on Intel optimized open source IPP-Crypto library.
+* [IPPCP](https://github.com/intel/ippcp-plugin-rocksdb): a plugin to enable encryption on RocksDB based on Intel optimized open source IPP-Crypto library.
+* [encfs](https://github.com/pegasus-kv/encfs): a plugin to enable encryption on RocksDB based on OpenSSL library.

TARGETS

@@ -163,6 +163,7 @@ cpp_library_wrapper(name="rocksdb_lib", srcs=[
         "options/configurable.cc",
         "options/customizable.cc",
         "options/db_options.cc",
+        "options/offpeak_time_info.cc",
         "options/options.cc",
         "options/options_helper.cc",
         "options/options_parser.cc",
@@ -5620,3 +5621,5 @@ cpp_unittest_wrapper(name="write_unprepared_transaction_test",
             deps=[":rocksdb_test_lib"],
             extra_compiler_flags=[])
 
+
+export_file(name = "tools/db_crashtest.py")

USERS.md

@@ -152,6 +152,9 @@ LzLabs is using RocksDB as a storage engine in their multi-database distributed
 ## 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 storage engine.
 
+## Qdrant
+[Qdrant](https://qdrant.tech/) is an open source vector database, it [uses](https://qdrant.tech/documentation/concepts/storage/) RocksDB as its persistent storage.
+
 ## Milvus
 [Milvus](https://milvus.io/) is an open source vector database for unstructured data. It uses RocksDB not only as one of the supported kv storage engines, but also as a message queue.
 

buckifier/buckify_rocksdb.py

@@ -303,6 +303,7 @@ def generate_targets(repo_path, deps_map):
                     deps=json.dumps(deps["extra_deps"] + [":rocksdb_test_lib"]),
                     extra_compiler_flags=json.dumps(deps["extra_compiler_flags"]),
                 )
+    TARGETS.export_file("tools/db_crashtest.py")
 
     print(ColorString.info("Generated TARGETS Summary:"))
     print(ColorString.info("- %d libs" % TARGETS.total_lib))

buckifier/targets_builder.py

@@ -148,3 +148,9 @@ add_c_test_wrapper()
                 ).encode("utf-8")
             )
         self.total_test = self.total_test + 1
+
+    def export_file(self, name):
+        with open(self.path, "a") as targets_file:
+            targets_file.write(
+                targets_cfg.export_file_template.format(name=name)
+            )

buckifier/targets_cfg.py

@@ -37,3 +37,7 @@ fancy_bench_template = """
 fancy_bench_wrapper(suite_name="{name}", binary_to_bench_to_metric_list_map={bench_config}, slow={slow}, expected_runtime={expected_runtime}, sl_iterations={sl_iterations}, regression_threshold={regression_threshold})
 
 """
+
+export_file_template = """
+export_file(name = "{name}")
+"""

build_tools/build_detect_platform

@@ -647,8 +647,10 @@ if [ "$PORTABLE" == "" ] || [ "$PORTABLE" == 0 ]; then
     fi
     COMMON_FLAGS="$COMMON_FLAGS"
   elif test -n "`echo $TARGET_ARCHITECTURE | grep ^riscv64`"; then
-    RISC_ISA=$(cat /proc/cpuinfo | grep isa | head -1 | cut --delimiter=: -f 2 | cut -b 2-)
-    COMMON_FLAGS="$COMMON_FLAGS -march=${RISC_ISA}"
+    RISC_ISA=$(cat /proc/cpuinfo | grep -E '^isa\s*:' | head -1 | cut --delimiter=: -f 2 | cut -b 2-)
+    if [ -n "${RISCV_ISA}" ]; then
+      COMMON_FLAGS="$COMMON_FLAGS -march=${RISC_ISA}"
+    fi
   elif [ "$TARGET_OS" == "IOS" ]; then
     COMMON_FLAGS="$COMMON_FLAGS"
   else
@@ -660,8 +662,7 @@ else
     if test -n "`echo $TARGET_ARCHITECTURE | grep ^s390x`"; then
       COMMON_FLAGS="$COMMON_FLAGS -march=z196 "
     elif test -n "`echo $TARGET_ARCHITECTURE | grep ^riscv64`"; then
-      RISC_ISA=$(cat /proc/cpuinfo | grep isa | head -1 | cut --delimiter=: -f 2 | cut -b 2-)
-      COMMON_FLAGS="$COMMON_FLAGS -march=${RISC_ISA}"
+      COMMON_FLAGS="$COMMON_FLAGS -march=rv64gc"
     elif test "$USE_SSE"; then
       # USE_SSE is DEPRECATED
       # This is a rough approximation of the old USE_SSE behavior

build_tools/gnu_parallel

@@ -1561,7 +1561,7 @@ sub save_stdin_stdout_stderr {
 	::die_bug("Can't dup STDERR: $!");
     open $Global::original_stdin, "<&", "STDIN" or
 	::die_bug("Can't dup STDIN: $!");
-    $Global::is_terminal = (-t $Global::original_stderr) && !$ENV{'CIRCLECI'} && !$ENV{'TRAVIS'};
+    $Global::is_terminal = (-t $Global::original_stderr) && !$ENV{'CIRCLECI'}&& !$ENV{'GITHUB_ACTIONS'}  && !$ENV{'TRAVIS'};
 }
 
 sub enough_file_handles {

cache/cache_bench_tool.cc

@@ -48,6 +48,10 @@ DEFINE_uint64(cache_size, 1 * GiB,
               "Number of bytes to use as a cache of uncompressed data.");
 DEFINE_int32(num_shard_bits, -1,
              "ShardedCacheOptions::shard_bits. Default = auto");
+DEFINE_int32(
+    eviction_effort_cap,
+    ROCKSDB_NAMESPACE::HyperClockCacheOptions(1, 1).eviction_effort_cap,
+    "HyperClockCacheOptions::eviction_effort_cap");
 
 DEFINE_double(resident_ratio, 0.25,
               "Ratio of keys fitting in cache to keyspace.");
@@ -63,11 +67,22 @@ DEFINE_int32(
 DEFINE_uint32(skew, 5, "Degree of skew in key selection. 0 = no skew");
 DEFINE_bool(populate_cache, true, "Populate cache before operations");
 
-DEFINE_uint32(lookup_insert_percent, 87,
+DEFINE_double(pinned_ratio, 0.25,
+              "Keep roughly this portion of entries pinned in cache.");
+DEFINE_double(
+    vary_capacity_ratio, 0.0,
+    "If greater than 0.0, will periodically vary the capacity between this "
+    "ratio less than full size and full size. If vary_capacity_ratio + "
+    "pinned_ratio is close to or exceeds 1.0, the cache might thrash.");
+
+DEFINE_uint32(lookup_insert_percent, 82,
               "Ratio of lookup (+ insert on not found) to total workload "
               "(expressed as a percentage)");
 DEFINE_uint32(insert_percent, 2,
               "Ratio of insert to total workload (expressed as a percentage)");
+DEFINE_uint32(blind_insert_percent, 5,
+              "Ratio of insert without keeping handle to total workload "
+              "(expressed as a percentage)");
 DEFINE_uint32(lookup_percent, 10,
               "Ratio of lookup to total workload (expressed as a percentage)");
 DEFINE_uint32(erase_percent, 1,
@@ -101,7 +116,6 @@ DEFINE_uint32(seed, 0, "Hashing/random seed to use. 0 = choose at random");
 
 DEFINE_string(secondary_cache_uri, "",
               "Full URI for creating a custom secondary cache object");
-static class std::shared_ptr<ROCKSDB_NAMESPACE::SecondaryCache> secondary_cache;
 
 DEFINE_string(cache_type, "lru_cache", "Type of block cache.");
 
@@ -200,16 +214,19 @@ class SharedState {
 
   bool Started() const { return start_; }
 
-  void AddLookupStats(uint64_t hits, uint64_t misses) {
+  void AddLookupStats(uint64_t hits, uint64_t misses, size_t pinned_count) {
     MutexLock l(&mu_);
     lookup_count_ += hits + misses;
     lookup_hits_ += hits;
+    pinned_count_ += pinned_count;
   }
 
   double GetLookupHitRatio() const {
     return 1.0 * lookup_hits_ / lookup_count_;
   }
 
+  size_t GetPinnedCount() const { return pinned_count_; }
+
  private:
   port::Mutex mu_;
   port::CondVar cv_;
@@ -221,6 +238,7 @@ class SharedState {
   uint64_t num_done_ = 0;
   uint64_t lookup_count_ = 0;
   uint64_t lookup_hits_ = 0;
+  size_t pinned_count_ = 0;
 };
 
 // Per-thread state for concurrent executions of the same benchmark.
@@ -313,6 +331,28 @@ Cache::CacheItemHelper helper2(CacheEntryRole::kIndexBlock, DeleteFn, SizeFn,
 Cache::CacheItemHelper helper3_wos(CacheEntryRole::kFilterBlock, DeleteFn);
 Cache::CacheItemHelper helper3(CacheEntryRole::kFilterBlock, DeleteFn, SizeFn,
                                SaveToFn, CreateFn, &helper3_wos);
+
+void ConfigureSecondaryCache(ShardedCacheOptions& opts) {
+  if (!FLAGS_secondary_cache_uri.empty()) {
+    std::shared_ptr<SecondaryCache> secondary_cache;
+    Status s = SecondaryCache::CreateFromString(
+        ConfigOptions(), FLAGS_secondary_cache_uri, &secondary_cache);
+    if (secondary_cache == nullptr) {
+      fprintf(stderr,
+              "No secondary cache registered matching string: %s status=%s\n",
+              FLAGS_secondary_cache_uri.c_str(), s.ToString().c_str());
+      exit(1);
+    }
+    opts.secondary_cache = secondary_cache;
+  }
+}
+
+ShardedCacheBase* AsShardedCache(Cache* c) {
+  if (!FLAGS_secondary_cache_uri.empty()) {
+    c = static_cast_with_check<CacheWrapper>(c)->GetTarget().get();
+  }
+  return static_cast_with_check<ShardedCacheBase>(c);
+}
 }  // namespace
 
 class CacheBench {
@@ -327,7 +367,9 @@ class CacheBench {
                                  FLAGS_lookup_insert_percent),
         insert_threshold_(lookup_insert_threshold_ +
                           kHundredthUint64 * FLAGS_insert_percent),
-        lookup_threshold_(insert_threshold_ +
+        blind_insert_threshold_(insert_threshold_ +
+                                kHundredthUint64 * FLAGS_blind_insert_percent),
+        lookup_threshold_(blind_insert_threshold_ +
                           kHundredthUint64 * FLAGS_lookup_percent),
         erase_threshold_(lookup_threshold_ +
                          kHundredthUint64 * FLAGS_erase_percent) {
@@ -353,6 +395,7 @@ class CacheBench {
           FLAGS_cache_size, /*estimated_entry_charge=*/0, FLAGS_num_shard_bits);
       opts.hash_seed = BitwiseAnd(FLAGS_seed, INT32_MAX);
       opts.memory_allocator = allocator;
+      opts.eviction_effort_cap = FLAGS_eviction_effort_cap;
       if (FLAGS_cache_type == "fixed_hyper_clock_cache" ||
           FLAGS_cache_type == "hyper_clock_cache") {
         opts.estimated_entry_charge = FLAGS_value_bytes_estimate > 0
@@ -366,6 +409,7 @@ class CacheBench {
         fprintf(stderr, "Cache type not supported.\n");
         exit(1);
       }
+      ConfigureSecondaryCache(opts);
       cache_ = opts.MakeSharedCache();
     } else if (FLAGS_cache_type == "lru_cache") {
       LRUCacheOptions opts(FLAGS_cache_size, FLAGS_num_shard_bits,
@@ -373,19 +417,7 @@ class CacheBench {
                            0.5 /* high_pri_pool_ratio */);
       opts.hash_seed = BitwiseAnd(FLAGS_seed, INT32_MAX);
       opts.memory_allocator = allocator;
-      if (!FLAGS_secondary_cache_uri.empty()) {
-        Status s = SecondaryCache::CreateFromString(
-            ConfigOptions(), FLAGS_secondary_cache_uri, &secondary_cache);
-        if (secondary_cache == nullptr) {
-          fprintf(
-              stderr,
-              "No secondary cache registered matching string: %s status=%s\n",
-              FLAGS_secondary_cache_uri.c_str(), s.ToString().c_str());
-          exit(1);
-        }
-        opts.secondary_cache = secondary_cache;
-      }
-
+      ConfigureSecondaryCache(opts);
       cache_ = NewLRUCache(opts);
     } else {
       fprintf(stderr, "Cache type not supported.\n");
@@ -505,6 +537,8 @@ class CacheBench {
     size_t slot = cache_->GetTableAddressCount();
     printf("Final load factor: %g (%zu / %zu)\n", 1.0 * occ / slot, occ, slot);
 
+    printf("Final pinned count: %zu\n", shared.GetPinnedCount());
+
     if (FLAGS_histograms) {
       printf("\nOperation latency (ns):\n");
       HistogramImpl combined;
@@ -536,6 +570,7 @@ class CacheBench {
   // Cumulative thresholds in the space of a random uint64_t
   const uint64_t lookup_insert_threshold_;
   const uint64_t insert_threshold_;
+  const uint64_t blind_insert_threshold_;
   const uint64_t lookup_threshold_;
   const uint64_t erase_threshold_;
 
@@ -651,29 +686,44 @@ class CacheBench {
     uint64_t lookup_misses = 0;
     uint64_t lookup_hits = 0;
     // To hold handles for a non-trivial amount of time
-    Cache::Handle* handle = nullptr;
+    std::deque<Cache::Handle*> pinned;
+    size_t total_pin_count = static_cast<size_t>(
+        (FLAGS_cache_size * FLAGS_pinned_ratio) / FLAGS_value_bytes + 0.999999);
+    // For this thread. Some round up, some round down, as appropriate
+    size_t pin_count = (total_pin_count + thread->tid) / FLAGS_threads;
+
     KeyGen gen;
     const auto clock = SystemClock::Default().get();
     uint64_t start_time = clock->NowMicros();
     StopWatchNano timer(clock);
     auto system_clock = SystemClock::Default();
+    size_t steps_to_next_capacity_change = 0;
 
     for (uint64_t i = 0; i < FLAGS_ops_per_thread; i++) {
       Slice key = gen.GetRand(thread->rnd, max_key_, FLAGS_skew);
       uint64_t random_op = thread->rnd.Next();
 
+      if (FLAGS_vary_capacity_ratio > 0.0 && thread->tid == 0) {
+        if (steps_to_next_capacity_change == 0) {
+          double cut_ratio = static_cast<double>(thread->rnd.Next()) /
+                             static_cast<double>(UINT64_MAX) *
+                             FLAGS_vary_capacity_ratio;
+          cache_->SetCapacity(FLAGS_cache_size * (1.0 - cut_ratio));
+          steps_to_next_capacity_change =
+              static_cast<size_t>(FLAGS_ops_per_thread / 100);
+        } else {
+          --steps_to_next_capacity_change;
+        }
+      }
+
       if (FLAGS_histograms) {
         timer.Start();
       }
 
       if (random_op < lookup_insert_threshold_) {
-        if (handle) {
-          cache_->Release(handle);
-          handle = nullptr;
-        }
         // do lookup
-        handle = cache_->Lookup(key, &helper2, /*context*/ nullptr,
-                                Cache::Priority::LOW);
+        auto handle = cache_->Lookup(key, &helper2, /*context*/ nullptr,
+                                     Cache::Priority::LOW);
         if (handle) {
           ++lookup_hits;
           if (!FLAGS_lean) {
@@ -681,32 +731,31 @@ class CacheBench {
             result += NPHash64(static_cast<char*>(cache_->Value(handle)),
                                FLAGS_value_bytes);
           }
+          pinned.push_back(handle);
         } else {
           ++lookup_misses;
           // do insert
           Status s = cache_->Insert(
               key, createValue(thread->rnd, cache_->memory_allocator()),
-              &helper2, FLAGS_value_bytes, &handle);
+              &helper2, FLAGS_value_bytes, &pinned.emplace_back());
           assert(s.ok());
         }
       } else if (random_op < insert_threshold_) {
-        if (handle) {
-          cache_->Release(handle);
-          handle = nullptr;
-        }
         // do insert
         Status s = cache_->Insert(
             key, createValue(thread->rnd, cache_->memory_allocator()), &helper3,
-            FLAGS_value_bytes, &handle);
+            FLAGS_value_bytes, &pinned.emplace_back());
+        assert(s.ok());
+      } else if (random_op < blind_insert_threshold_) {
+        // insert without keeping a handle
+        Status s = cache_->Insert(
+            key, createValue(thread->rnd, cache_->memory_allocator()), &helper3,
+            FLAGS_value_bytes);
         assert(s.ok());
       } else if (random_op < lookup_threshold_) {
-        if (handle) {
-          cache_->Release(handle);
-          handle = nullptr;
-        }
         // do lookup
-        handle = cache_->Lookup(key, &helper2, /*context*/ nullptr,
-                                Cache::Priority::LOW);
+        auto handle = cache_->Lookup(key, &helper2, /*context*/ nullptr,
+                                     Cache::Priority::LOW);
         if (handle) {
           ++lookup_hits;
           if (!FLAGS_lean) {
@@ -714,6 +763,7 @@ class CacheBench {
             result += NPHash64(static_cast<char*>(cache_->Value(handle)),
                                FLAGS_value_bytes);
           }
+          pinned.push_back(handle);
         } else {
           ++lookup_misses;
         }
@@ -727,7 +777,6 @@ class CacheBench {
       if (FLAGS_histograms) {
         thread->latency_ns_hist.Add(timer.ElapsedNanos());
       }
-      thread->shared->AddLookupStats(lookup_hits, lookup_misses);
       if (FLAGS_usleep > 0) {
         unsigned us =
             static_cast<unsigned>(thread->rnd.Uniform(FLAGS_usleep + 1));
@@ -735,12 +784,17 @@ class CacheBench {
           system_clock->SleepForMicroseconds(us);
         }
       }
+      while (pinned.size() > pin_count) {
+        cache_->Release(pinned.front());
+        pinned.pop_front();
+      }
     }
     if (FLAGS_early_exit) {
       MutexLock l(thread->shared->GetMutex());
       exit(0);
     }
-    if (handle) {
+    thread->shared->AddLookupStats(lookup_hits, lookup_misses, pinned.size());
+    for (auto handle : pinned) {
       cache_->Release(handle);
       handle = nullptr;
     }
@@ -769,8 +823,7 @@ class CacheBench {
     printf("Cache size          : %s\n",
            BytesToHumanString(FLAGS_cache_size).c_str());
     printf("Num shard bits      : %d\n",
-           static_cast_with_check<ShardedCacheBase>(cache_.get())
-               ->GetNumShardBits());
+           AsShardedCache(cache_.get())->GetNumShardBits());
     printf("Max key             : %" PRIu64 "\n", max_key_);
     printf("Resident ratio      : %g\n", FLAGS_resident_ratio);
     printf("Skew degree         : %u\n", FLAGS_skew);

cache/clock_cache.h

@@ -11,6 +11,7 @@
 
 #include <array>
 #include <atomic>
+#include <climits>
 #include <cstddef>
 #include <cstdint>
 #include <memory>
@@ -24,6 +25,7 @@
 #include "port/port.h"
 #include "rocksdb/cache.h"
 #include "rocksdb/secondary_cache.h"
+#include "util/atomic.h"
 #include "util/autovector.h"
 #include "util/math.h"
 
@@ -368,11 +370,19 @@ struct ClockHandle : public ClockHandleBasicData {
   // TODO: make these coundown values tuning parameters for eviction?
 
   // See above. Mutable for read reference counting.
-  mutable std::atomic<uint64_t> meta{};
+  mutable AcqRelAtomic<uint64_t> meta{};
 };  // struct ClockHandle
 
 class BaseClockTable {
  public:
+  struct BaseOpts {
+    explicit BaseOpts(int _eviction_effort_cap)
+        : eviction_effort_cap(_eviction_effort_cap) {}
+    explicit BaseOpts(const HyperClockCacheOptions& opts)
+        : BaseOpts(opts.eviction_effort_cap) {}
+    int eviction_effort_cap;
+  };
+
   BaseClockTable(CacheMetadataChargePolicy metadata_charge_policy,
                  MemoryAllocator* allocator,
                  const Cache::EvictionCallback* eviction_callback,
@@ -385,36 +395,37 @@ class BaseClockTable {
   template <class Table>
   typename Table::HandleImpl* CreateStandalone(ClockHandleBasicData& proto,
                                                size_t capacity,
-                                               bool strict_capacity_limit,
+                                               uint32_t eec_and_scl,
                                                bool allow_uncharged);
 
   template <class Table>
   Status Insert(const ClockHandleBasicData& proto,
                 typename Table::HandleImpl** handle, Cache::Priority priority,
-                size_t capacity, bool strict_capacity_limit);
+                size_t capacity, uint32_t eec_and_scl);
 
   void Ref(ClockHandle& handle);
 
-  size_t GetOccupancy() const {
-    return occupancy_.load(std::memory_order_relaxed);
-  }
+  size_t GetOccupancy() const { return occupancy_.LoadRelaxed(); }
 
-  size_t GetUsage() const { return usage_.load(std::memory_order_relaxed); }
+  size_t GetUsage() const { return usage_.LoadRelaxed(); }
 
-  size_t GetStandaloneUsage() const {
-    return standalone_usage_.load(std::memory_order_relaxed);
-  }
+  size_t GetStandaloneUsage() const { return standalone_usage_.LoadRelaxed(); }
 
   uint32_t GetHashSeed() const { return hash_seed_; }
 
-  uint64_t GetYieldCount() const { return yield_count_.load(); }
+  uint64_t GetYieldCount() const { return yield_count_.LoadRelaxed(); }
+
+  uint64_t GetEvictionEffortExceededCount() const {
+    return eviction_effort_exceeded_count_.LoadRelaxed();
+  }
 
   struct EvictionData {
     size_t freed_charge = 0;
     size_t freed_count = 0;
+    size_t seen_pinned_count = 0;
   };
 
-  void TrackAndReleaseEvictedEntry(ClockHandle* h, EvictionData* data);
+  void TrackAndReleaseEvictedEntry(ClockHandle* h);
 
 #ifndef NDEBUG
   // Acquire N references
@@ -439,6 +450,7 @@ class BaseClockTable {
   template <class Table>
   Status ChargeUsageMaybeEvictStrict(size_t total_charge, size_t capacity,
                                      bool need_evict_for_occupancy,
+                                     uint32_t eviction_effort_cap,
                                      typename Table::InsertState& state);
 
   // Helper for updating `usage_` for new entry with given `total_charge`
@@ -452,6 +464,7 @@ class BaseClockTable {
   template <class Table>
   bool ChargeUsageMaybeEvictNonStrict(size_t total_charge, size_t capacity,
                                       bool need_evict_for_occupancy,
+                                      uint32_t eviction_effort_cap,
                                       typename Table::InsertState& state);
 
  protected:  // data
@@ -460,21 +473,29 @@ class BaseClockTable {
   // operations in ClockCacheShard.
 
   // Clock algorithm sweep pointer.
-  std::atomic<uint64_t> clock_pointer_{};
+  // (Relaxed: only needs to be consistent with itself.)
+  RelaxedAtomic<uint64_t> clock_pointer_{};
 
   // Counter for number of times we yield to wait on another thread.
-  std::atomic<uint64_t> yield_count_{};
+  // It is normal for this to occur rarely in normal operation.
+  // (Relaxed: a simple stat counter.)
+  RelaxedAtomic<uint64_t> yield_count_{};
+
+  // Counter for number of times eviction effort cap is exceeded.
+  // It is normal for this to occur rarely in normal operation.
+  // (Relaxed: a simple stat counter.)
+  RelaxedAtomic<uint64_t> eviction_effort_exceeded_count_{};
 
   // TODO: is this separation needed if we don't do background evictions?
   ALIGN_AS(CACHE_LINE_SIZE)
   // Number of elements in the table.
-  std::atomic<size_t> occupancy_{};
+  AcqRelAtomic<size_t> occupancy_{};
 
   // Memory usage by entries tracked by the cache (including standalone)
-  std::atomic<size_t> usage_{};
+  AcqRelAtomic<size_t> usage_{};
 
   // Part of usage by standalone entries (not in table)
-  std::atomic<size_t> standalone_usage_{};
+  AcqRelAtomic<size_t> standalone_usage_{};
 
   ALIGN_AS(CACHE_LINE_SIZE)
   const CacheMetadataChargePolicy metadata_charge_policy_;
@@ -500,7 +521,11 @@ class FixedHyperClockTable : public BaseClockTable {
   struct ALIGN_AS(64U) HandleImpl : public ClockHandle {
     // The number of elements that hash to this slot or a lower one, but wind
     // up in this slot or a higher one.
-    std::atomic<uint32_t> displacements{};
+    // (Relaxed: within a Cache op, does not need consistency with entries
+    // inserted/removed during that op. For example, a Lookup() that
+    // happens-after an Insert() will see an appropriate displacements value
+    // for the entry to be in a published state.)
+    RelaxedAtomic<uint32_t> displacements{};
 
     // Whether this is a "deteched" handle that is independently allocated
     // with `new` (so must be deleted with `delete`).
@@ -514,17 +539,19 @@ class FixedHyperClockTable : public BaseClockTable {
     inline void SetStandalone() { standalone = true; }
   };  // struct HandleImpl
 
-  struct Opts {
-    explicit Opts(size_t _estimated_value_size)
-        : estimated_value_size(_estimated_value_size) {}
-    explicit Opts(const HyperClockCacheOptions& opts) {
+  struct Opts : public BaseOpts {
+    explicit Opts(size_t _estimated_value_size, int _eviction_effort_cap)
+        : BaseOpts(_eviction_effort_cap),
+          estimated_value_size(_estimated_value_size) {}
+    explicit Opts(const HyperClockCacheOptions& opts)
+        : BaseOpts(opts.eviction_effort_cap) {
       assert(opts.estimated_entry_charge > 0);
       estimated_value_size = opts.estimated_entry_charge;
     }
     size_t estimated_value_size;
   };
 
-  FixedHyperClockTable(size_t capacity, bool strict_capacity_limit,
+  FixedHyperClockTable(size_t capacity,
                        CacheMetadataChargePolicy metadata_charge_policy,
                        MemoryAllocator* allocator,
                        const Cache::EvictionCallback* eviction_callback,
@@ -546,7 +573,8 @@ class FixedHyperClockTable : public BaseClockTable {
   // Runs the clock eviction algorithm trying to reclaim at least
   // requested_charge. Returns how much is evicted, which could be less
   // if it appears impossible to evict the requested amount without blocking.
-  void Evict(size_t requested_charge, InsertState& state, EvictionData* data);
+  void Evict(size_t requested_charge, InsertState& state, EvictionData* data,
+             uint32_t eviction_effort_cap);
 
   HandleImpl* Lookup(const UniqueId64x2& hashed_key);
 
@@ -787,32 +815,33 @@ class AutoHyperClockTable : public BaseClockTable {
 
     // See above. The head pointer is logically independent of the rest of
     // the entry, including the chain next pointer.
-    std::atomic<uint64_t> head_next_with_shift{kUnusedMarker};
-    std::atomic<uint64_t> chain_next_with_shift{kUnusedMarker};
+    AcqRelAtomic<uint64_t> head_next_with_shift{kUnusedMarker};
+    AcqRelAtomic<uint64_t> chain_next_with_shift{kUnusedMarker};
 
     // For supporting CreateStandalone and some fallback cases.
     inline bool IsStandalone() const {
-      return head_next_with_shift.load(std::memory_order_acquire) ==
-             kStandaloneMarker;
+      return head_next_with_shift.Load() == kStandaloneMarker;
     }
 
     inline void SetStandalone() {
-      head_next_with_shift.store(kStandaloneMarker, std::memory_order_release);
+      head_next_with_shift.Store(kStandaloneMarker);
     }
   };  // struct HandleImpl
 
-  struct Opts {
-    explicit Opts(size_t _min_avg_value_size)
-        : min_avg_value_size(_min_avg_value_size) {}
+  struct Opts : public BaseOpts {
+    explicit Opts(size_t _min_avg_value_size, int _eviction_effort_cap)
+        : BaseOpts(_eviction_effort_cap),
+          min_avg_value_size(_min_avg_value_size) {}
 
-    explicit Opts(const HyperClockCacheOptions& opts) {
+    explicit Opts(const HyperClockCacheOptions& opts)
+        : BaseOpts(opts.eviction_effort_cap) {
       assert(opts.estimated_entry_charge == 0);
       min_avg_value_size = opts.min_avg_entry_charge;
     }
     size_t min_avg_value_size;
   };
 
-  AutoHyperClockTable(size_t capacity, bool strict_capacity_limit,
+  AutoHyperClockTable(size_t capacity,
                       CacheMetadataChargePolicy metadata_charge_policy,
                       MemoryAllocator* allocator,
                       const Cache::EvictionCallback* eviction_callback,
@@ -839,7 +868,8 @@ class AutoHyperClockTable : public BaseClockTable {
   // Runs the clock eviction algorithm trying to reclaim at least
   // requested_charge. Returns how much is evicted, which could be less
   // if it appears impossible to evict the requested amount without blocking.
-  void Evict(size_t requested_charge, InsertState& state, EvictionData* data);
+  void Evict(size_t requested_charge, InsertState& state, EvictionData* data,
+             uint32_t eviction_effort_cap);
 
   HandleImpl* Lookup(const UniqueId64x2& hashed_key);
 
@@ -904,7 +934,8 @@ class AutoHyperClockTable : public BaseClockTable {
   // with proper handling to ensure all existing data is seen even in the
   // presence of concurrent insertions, etc. (See implementation.)
   template <class OpData>
-  void PurgeImpl(OpData* op_data, size_t home = SIZE_MAX);
+  void PurgeImpl(OpData* op_data, size_t home = SIZE_MAX,
+                 EvictionData* data = nullptr);
 
   // An RAII wrapper for locking a chain of entries for removals. See
   // implementation.
@@ -914,7 +945,7 @@ class AutoHyperClockTable : public BaseClockTable {
   // implementation.
   template <class OpData>
   void PurgeImplLocked(OpData* op_data, ChainRewriteLock& rewrite_lock,
-                       size_t home);
+                       size_t home, EvictionData* data);
 
   // Update length_info_ as much as possible without waiting, given a known
   // usable (ready for inserts and lookups) grow_home. (Previous grow_homes
@@ -939,22 +970,25 @@ class AutoHyperClockTable : public BaseClockTable {
   // log time to find the correct chain, but normally this value enables
   // readers to find the correct chain on the first try.
   //
-  // NOTES: length_info_ is only updated at the end of a Grow operation,
-  // so that waiting in Grow operations isn't done while entries are pinned
-  // for internal operation purposes. Thus, Lookup and Insert have to
-  // detect and support cases where length_info hasn't caught up to updated
-  // chains. Winning grow thread is the one that transitions
-  // head_next_with_shift from zeros. Grow threads can spin/yield wait for
-  // preconditions and postconditions to be met.
-  std::atomic<uint64_t> length_info_;
+  // To maximize parallelization of Grow() operations, this field is only
+  // updated opportunistically after Grow() operations and in DoInsert() where
+  // it is found to be out-of-date. See CatchUpLengthInfoNoWait().
+  AcqRelAtomic<uint64_t> length_info_;
 
   // An already-computed version of the usable length times the max load
   // factor. Could be slightly out of date but GrowIfNeeded()/Grow() handle
   // that internally.
-  std::atomic<size_t> occupancy_limit_;
+  // (Relaxed: allowed to lag behind length_info_ by a little)
+  RelaxedAtomic<size_t> occupancy_limit_;
+
+  // The next index to use from array_ upon the next Grow(). Might be ahead of
+  // length_info_.
+  // (Relaxed: self-contained source of truth for next grow home)
+  RelaxedAtomic<size_t> grow_frontier_;
 
   // See explanation in AutoHyperClockTable::Evict
-  std::atomic<size_t> clock_pointer_mask_;
+  // (Relaxed: allowed to lag behind clock_pointer_ and length_info_ state)
+  RelaxedAtomic<size_t> clock_pointer_mask_;
 };  // class AutoHyperClockTable
 
 // A single shard of sharded cache.
@@ -1070,10 +1104,13 @@ class ALIGN_AS(CACHE_LINE_SIZE) ClockCacheShard final : public CacheShardBase {
   Table table_;
 
   // Maximum total charge of all elements stored in the table.
-  std::atomic<size_t> capacity_;
+  // (Relaxed: eventual consistency/update is OK)
+  RelaxedAtomic<size_t> capacity_;
 
-  // Whether to reject insertion if cache reaches its full capacity.
-  std::atomic<bool> strict_capacity_limit_;
+  // Encodes eviction_effort_cap (bottom 31 bits) and strict_capacity_limit
+  // (top bit). See HyperClockCacheOptions::eviction_effort_cap etc.
+  // (Relaxed: eventual consistency/update is OK)
+  RelaxedAtomic<uint32_t> eec_and_scl_;
 };  // class ClockCacheShard
 
 template <class Table>

cache/compressed_secondary_cache.cc

@@ -31,7 +31,7 @@ CompressedSecondaryCache::~CompressedSecondaryCache() {}
 std::unique_ptr<SecondaryCacheResultHandle> CompressedSecondaryCache::Lookup(
     const Slice& key, const Cache::CacheItemHelper* helper,
     Cache::CreateContext* create_context, bool /*wait*/, bool advise_erase,
-    bool& kept_in_sec_cache) {
+    Statistics* stats, bool& kept_in_sec_cache) {
   assert(helper);
   // This is a minor optimization. Its ok to skip it in TSAN in order to
   // avoid a false positive.
@@ -51,6 +51,7 @@ std::unique_ptr<SecondaryCacheResultHandle> CompressedSecondaryCache::Lookup(
   void* handle_value = cache_->Value(lru_handle);
   if (handle_value == nullptr) {
     cache_->Release(lru_handle, /*erase_if_last_ref=*/false);
+    RecordTick(stats, COMPRESSED_SECONDARY_CACHE_DUMMY_HITS);
     return nullptr;
   }
 
@@ -137,6 +138,7 @@ std::unique_ptr<SecondaryCacheResultHandle> CompressedSecondaryCache::Lookup(
     cache_->Release(lru_handle, /*erase_if_last_ref=*/false);
   }
   handle.reset(new CompressedSecondaryCacheResultHandle(value, charge));
+  RecordTick(stats, COMPRESSED_SECONDARY_CACHE_HITS);
   return handle;
 }
 
@@ -190,13 +192,13 @@ Status CompressedSecondaryCache::InsertInternal(
       type == kNoCompression &&
       !cache_options_.do_not_compress_roles.Contains(helper->role)) {
     PERF_COUNTER_ADD(compressed_sec_cache_uncompressed_bytes, data_size);
-    CompressionOptions compression_opts;
     CompressionContext compression_context(cache_options_.compression_type,
-                                           compression_opts);
+                                           cache_options_.compression_opts);
     uint64_t sample_for_compression{0};
     CompressionInfo compression_info(
-        compression_opts, compression_context, CompressionDict::GetEmptyDict(),
-        cache_options_.compression_type, sample_for_compression);
+        cache_options_.compression_opts, compression_context,
+        CompressionDict::GetEmptyDict(), cache_options_.compression_type,
+        sample_for_compression);
 
     bool success =
         CompressData(val, compression_info,
@@ -289,6 +291,11 @@ std::string CompressedSecondaryCache::GetPrintableOptions() const {
   snprintf(buffer, kBufferSize, "    compression_type : %s\n",
            CompressionTypeToString(cache_options_.compression_type).c_str());
   ret.append(buffer);
+  snprintf(buffer, kBufferSize, "    compression_opts : %s\n",
+           CompressionOptionsToString(
+               const_cast<CompressionOptions&>(cache_options_.compression_opts))
+               .c_str());
+  ret.append(buffer);
   snprintf(buffer, kBufferSize, "    compress_format_version : %d\n",
            cache_options_.compress_format_version);
   ret.append(buffer);

cache/compressed_secondary_cache.h

@@ -86,7 +86,7 @@ class CompressedSecondaryCache : public SecondaryCache {
   std::unique_ptr<SecondaryCacheResultHandle> Lookup(
       const Slice& key, const Cache::CacheItemHelper* helper,
       Cache::CreateContext* create_context, bool /*wait*/, bool advise_erase,
-      bool& kept_in_sec_cache) override;
+      Statistics* stats, bool& kept_in_sec_cache) override;
 
   bool SupportForceErase() const override { return true; }
 

cache/compressed_secondary_cache_test.cc

@@ -44,7 +44,7 @@ class CompressedSecondaryCacheTestBase : public testing::Test,
     // Lookup an non-existent key.
     std::unique_ptr<SecondaryCacheResultHandle> handle0 =
         sec_cache->Lookup(key0, GetHelper(), this, true, /*advise_erase=*/true,
-                          kept_in_sec_cache);
+                          /*stats=*/nullptr, kept_in_sec_cache);
     ASSERT_EQ(handle0, nullptr);
 
     Random rnd(301);
@@ -59,7 +59,7 @@ class CompressedSecondaryCacheTestBase : public testing::Test,
 
     std::unique_ptr<SecondaryCacheResultHandle> handle1_1 =
         sec_cache->Lookup(key1, GetHelper(), this, true, /*advise_erase=*/false,
-                          kept_in_sec_cache);
+                          /*stats=*/nullptr, kept_in_sec_cache);
     ASSERT_EQ(handle1_1, nullptr);
 
     // Insert and Lookup the item k1 for the second time and advise erasing it.
@@ -68,7 +68,7 @@ class CompressedSecondaryCacheTestBase : public testing::Test,
 
     std::unique_ptr<SecondaryCacheResultHandle> handle1_2 =
         sec_cache->Lookup(key1, GetHelper(), this, true, /*advise_erase=*/true,
-                          kept_in_sec_cache);
+                          /*stats=*/nullptr, kept_in_sec_cache);
     ASSERT_NE(handle1_2, nullptr);
     ASSERT_FALSE(kept_in_sec_cache);
     if (sec_cache_is_compressed) {
@@ -89,7 +89,7 @@ class CompressedSecondaryCacheTestBase : public testing::Test,
     // Lookup the item k1 again.
     std::unique_ptr<SecondaryCacheResultHandle> handle1_3 =
         sec_cache->Lookup(key1, GetHelper(), this, true, /*advise_erase=*/true,
-                          kept_in_sec_cache);
+                          /*stats=*/nullptr, kept_in_sec_cache);
     ASSERT_EQ(handle1_3, nullptr);
 
     // Insert and Lookup the item k2.
@@ -99,7 +99,7 @@ class CompressedSecondaryCacheTestBase : public testing::Test,
     ASSERT_EQ(get_perf_context()->compressed_sec_cache_insert_dummy_count, 2);
     std::unique_ptr<SecondaryCacheResultHandle> handle2_1 =
         sec_cache->Lookup(key2, GetHelper(), this, true, /*advise_erase=*/false,
-                          kept_in_sec_cache);
+                          /*stats=*/nullptr, kept_in_sec_cache);
     ASSERT_EQ(handle2_1, nullptr);
 
     ASSERT_OK(sec_cache->Insert(key2, &item2, GetHelper(), false));
@@ -115,7 +115,7 @@ class CompressedSecondaryCacheTestBase : public testing::Test,
     }
     std::unique_ptr<SecondaryCacheResultHandle> handle2_2 =
         sec_cache->Lookup(key2, GetHelper(), this, true, /*advise_erase=*/false,
-                          kept_in_sec_cache);
+                          /*stats=*/nullptr, kept_in_sec_cache);
     ASSERT_NE(handle2_2, nullptr);
     std::unique_ptr<TestItem> val2 =
         std::unique_ptr<TestItem>(static_cast<TestItem*>(handle2_2->Value()));
@@ -196,14 +196,14 @@ class CompressedSecondaryCacheTestBase : public testing::Test,
     bool kept_in_sec_cache{false};
     std::unique_ptr<SecondaryCacheResultHandle> handle1 =
         sec_cache->Lookup(key1, GetHelper(), this, true, /*advise_erase=*/false,
-                          kept_in_sec_cache);
+                          /*stats=*/nullptr, kept_in_sec_cache);
     ASSERT_EQ(handle1, nullptr);
 
     // Insert k2 and k1 is evicted.
     ASSERT_OK(sec_cache->Insert(key2, &item2, GetHelper(), false));
     std::unique_ptr<SecondaryCacheResultHandle> handle2 =
         sec_cache->Lookup(key2, GetHelper(), this, true, /*advise_erase=*/false,
-                          kept_in_sec_cache);
+                          /*stats=*/nullptr, kept_in_sec_cache);
     ASSERT_NE(handle2, nullptr);
     std::unique_ptr<TestItem> val2 =
         std::unique_ptr<TestItem>(static_cast<TestItem*>(handle2->Value()));
@@ -215,14 +215,14 @@ class CompressedSecondaryCacheTestBase : public testing::Test,
 
     std::unique_ptr<SecondaryCacheResultHandle> handle1_1 =
         sec_cache->Lookup(key1, GetHelper(), this, true, /*advise_erase=*/false,
-                          kept_in_sec_cache);
+                          /*stats=*/nullptr, kept_in_sec_cache);
     ASSERT_EQ(handle1_1, nullptr);
 
     // Create Fails.
     SetFailCreate(true);
     std::unique_ptr<SecondaryCacheResultHandle> handle2_1 =
         sec_cache->Lookup(key2, GetHelper(), this, true, /*advise_erase=*/true,
-                          kept_in_sec_cache);
+                          /*stats=*/nullptr, kept_in_sec_cache);
     ASSERT_EQ(handle2_1, nullptr);
 
     // Save Fails.
@@ -912,9 +912,9 @@ TEST_P(CompressedSecondaryCacheTestWithCompressionParam, EntryRoles) {
     ASSERT_EQ(get_perf_context()->compressed_sec_cache_insert_real_count, 1U);
 
     bool kept_in_sec_cache{true};
-    std::unique_ptr<SecondaryCacheResultHandle> handle =
-        sec_cache->Lookup(ith_key, GetHelper(role), this, true,
-                          /*advise_erase=*/true, kept_in_sec_cache);
+    std::unique_ptr<SecondaryCacheResultHandle> handle = sec_cache->Lookup(
+        ith_key, GetHelper(role), this, true,
+        /*advise_erase=*/true, /*stats=*/nullptr, kept_in_sec_cache);
     ASSERT_NE(handle, nullptr);
 
     // Lookup returns the right data
@@ -992,6 +992,8 @@ class CompressedSecCacheTestWithTiered
         /*_capacity=*/0,
         /*_estimated_entry_charge=*/256 << 10,
         /*_num_shard_bits=*/0);
+    // eviction_effort_cap setting simply to avoid churn in existing test
+    hcc_opts.eviction_effort_cap = 100;
     TieredCacheOptions opts;
     lru_opts.capacity = 0;
     lru_opts.num_shard_bits = 0;
@@ -1224,13 +1226,11 @@ TEST_P(CompressedSecCacheTestWithTiered, DynamicUpdate) {
   ASSERT_OK(sec_cache->GetCapacity(sec_capacity));
   ASSERT_EQ(sec_capacity, 0);
 
-  ASSERT_NOK(UpdateTieredCache(tiered_cache, -1, 0.3));
-  // Only check usage for LRU cache. HCC shows a 64KB usage for some reason
-  if (std::get<0>(GetParam()) == PrimaryCacheType::kCacheTypeLRU) {
-    ASSERT_EQ(GetCache()->GetUsage(), 0);
-  }
+  ASSERT_OK(UpdateTieredCache(tiered_cache, -1, 0.3));
+  EXPECT_PRED3(CacheUsageWithinBounds, GetCache()->GetUsage(), (30 << 20),
+               GetPercent(30 << 20, 1));
   ASSERT_OK(sec_cache->GetCapacity(sec_capacity));
-  ASSERT_EQ(sec_capacity, 0);
+  ASSERT_EQ(sec_capacity, (30 << 20));
 }
 
 TEST_P(CompressedSecCacheTestWithTiered, DynamicUpdateWithReservation) {
@@ -1276,6 +1276,13 @@ TEST_P(CompressedSecCacheTestWithTiered, DynamicUpdateWithReservation) {
   ASSERT_OK(sec_cache->GetCapacity(sec_capacity));
   ASSERT_EQ(sec_capacity, (30 << 20));
 
+  ASSERT_OK(tiered_cache->GetSecondaryCacheCapacity(sec_capacity));
+  ASSERT_EQ(sec_capacity, 30 << 20);
+  size_t sec_usage;
+  ASSERT_OK(tiered_cache->GetSecondaryCachePinnedUsage(sec_usage));
+  EXPECT_PRED3(CacheUsageWithinBounds, sec_usage, 3 << 20,
+               GetPercent(3 << 20, 1));
+
   ASSERT_OK(UpdateTieredCache(tiered_cache, -1, 0.39));
   EXPECT_PRED3(CacheUsageWithinBounds, GetCache()->GetUsage(), (45 << 20),
                GetPercent(45 << 20, 1));
@@ -1309,28 +1316,50 @@ TEST_P(CompressedSecCacheTestWithTiered, DynamicUpdateWithReservation) {
   ASSERT_OK(sec_cache->GetCapacity(sec_capacity));
   ASSERT_EQ(sec_capacity, 0);
 
+  ASSERT_OK(UpdateTieredCache(tiered_cache, -1, 0.3));
+  EXPECT_PRED3(CacheUsageWithinBounds, GetCache()->GetUsage(), (37 << 20),
+               GetPercent(37 << 20, 1));
+  EXPECT_PRED3(CacheUsageWithinBounds, sec_cache->TEST_GetUsage(), (3 << 20),
+               GetPercent(3 << 20, 1));
+  ASSERT_OK(sec_cache->GetCapacity(sec_capacity));
+  ASSERT_EQ(sec_capacity, 30 << 20);
+
   ASSERT_OK(cache_res_mgr()->UpdateCacheReservation(0));
 }
 
-TEST_P(CompressedSecCacheTestWithTiered,
-       DynamicUpdateWithReservationUnderflow) {
+TEST_P(CompressedSecCacheTestWithTiered, ReservationOverCapacity) {
+  CompressedSecondaryCache* sec_cache =
+      reinterpret_cast<CompressedSecondaryCache*>(GetSecondaryCache());
   std::shared_ptr<Cache> tiered_cache = GetTieredCache();
-  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->LoadDependency(
-      {{"CacheWithSecondaryAdapter::Release:ChargeSecCache1",
-        "CacheWithSecondaryAdapter::UpdateCacheReservationRatio:Begin"},
-       {"CacheWithSecondaryAdapter::UpdateCacheReservationRatio:End",
-        "CacheWithSecondaryAdapter::Release:ChargeSecCache2"}});
-  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
 
-  port::Thread reserve_release_thread([&]() {
-    EXPECT_EQ(cache_res_mgr()->UpdateCacheReservation(50), Status::OK());
-    EXPECT_EQ(cache_res_mgr()->UpdateCacheReservation(0), Status::OK());
-  });
-  ASSERT_OK(UpdateTieredCache(tiered_cache, 100 << 20, 0.01));
-  reserve_release_thread.join();
-  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
+  ASSERT_OK(cache_res_mgr()->UpdateCacheReservation(110 << 20));
+  // Use EXPECT_PRED3 instead of EXPECT_NEAR to void too many size_t to
+  // double explicit casts
+  EXPECT_PRED3(CacheUsageWithinBounds, GetCache()->GetUsage(), (110 << 20),
+               GetPercent(110 << 20, 1));
+  EXPECT_PRED3(CacheUsageWithinBounds, sec_cache->TEST_GetUsage(), (30 << 20),
+               GetPercent(30 << 20, 1));
+  size_t sec_capacity;
+  ASSERT_OK(sec_cache->GetCapacity(sec_capacity));
+  ASSERT_EQ(sec_capacity, (30 << 20));
 
-  ASSERT_OK(UpdateTieredCache(tiered_cache, 100 << 20, 0.3));
+  ASSERT_OK(UpdateTieredCache(tiered_cache, -1, 0.39));
+  EXPECT_PRED3(CacheUsageWithinBounds, GetCache()->GetUsage(), (110 << 20),
+               GetPercent(110 << 20, 1));
+  EXPECT_PRED3(CacheUsageWithinBounds, sec_cache->TEST_GetUsage(), (39 << 20),
+               GetPercent(39 << 20, 1));
+  ASSERT_OK(sec_cache->GetCapacity(sec_capacity));
+  ASSERT_EQ(sec_capacity, (39 << 20));
+
+  ASSERT_OK(cache_res_mgr()->UpdateCacheReservation(90 << 20));
+  EXPECT_PRED3(CacheUsageWithinBounds, GetCache()->GetUsage(), (94 << 20),
+               GetPercent(94 << 20, 1));
+  EXPECT_PRED3(CacheUsageWithinBounds, sec_cache->TEST_GetUsage(), (35 << 20),
+               GetPercent(35 << 20, 1));
+  ASSERT_OK(sec_cache->GetCapacity(sec_capacity));
+  ASSERT_EQ(sec_capacity, (39 << 20));
+
+  ASSERT_OK(cache_res_mgr()->UpdateCacheReservation(0));
 }
 
 INSTANTIATE_TEST_CASE_P(

cache/lru_cache_test.cc

@@ -389,12 +389,13 @@ class ClockCacheTest : public testing::Test {
     }
   }
 
-  void NewShard(size_t capacity, bool strict_capacity_limit = true) {
+  void NewShard(size_t capacity, bool strict_capacity_limit = true,
+                int eviction_effort_cap = 30) {
     DeleteShard();
     shard_ =
         reinterpret_cast<Shard*>(port::cacheline_aligned_alloc(sizeof(Shard)));
 
-    TableOpts opts{1 /*value_size*/};
+    TableOpts opts{1 /*value_size*/, eviction_effort_cap};
     new (shard_)
         Shard(capacity, strict_capacity_limit, kDontChargeCacheMetadata,
               /*allocator*/ nullptr, &eviction_callback_, &hash_seed_, opts);
@@ -445,12 +446,20 @@ class ClockCacheTest : public testing::Test {
     return Slice(reinterpret_cast<const char*>(&hashed_key), 16U);
   }
 
+  // A bad hash function for testing / stressing collision handling
   static inline UniqueId64x2 TestHashedKey(char key) {
     // For testing hash near-collision behavior, put the variance in
     // hashed_key in bits that are unlikely to be used as hash bits.
     return {(static_cast<uint64_t>(key) << 56) + 1234U, 5678U};
   }
 
+  // A reasonable hash function, for testing "typical behavior" etc.
+  template <typename T>
+  static inline UniqueId64x2 CheapHash(T i) {
+    return {static_cast<uint64_t>(i) * uint64_t{0x85EBCA77C2B2AE63},
+            static_cast<uint64_t>(i) * uint64_t{0xC2B2AE3D27D4EB4F}};
+  }
+
   Shard* shard_ = nullptr;
 
  private:
@@ -683,6 +692,53 @@ TYPED_TEST(ClockCacheTest, ClockEvictionTest) {
   }
 }
 
+TYPED_TEST(ClockCacheTest, ClockEvictionEffortCapTest) {
+  using HandleImpl = typename ClockCacheTest<TypeParam>::Shard::HandleImpl;
+  for (bool strict_capacity_limit : {true, false}) {
+    SCOPED_TRACE("strict_capacity_limit = " +
+                 std::to_string(strict_capacity_limit));
+    for (int eec : {-42, 0, 1, 10, 100, 1000}) {
+      SCOPED_TRACE("eviction_effort_cap = " + std::to_string(eec));
+      constexpr size_t kCapacity = 1000;
+      // Start with much larger capacity to ensure that we can go way over
+      // capacity without reaching table occupancy limit.
+      this->NewShard(3 * kCapacity, strict_capacity_limit, eec);
+      auto& shard = *this->shard_;
+      shard.SetCapacity(kCapacity);
+
+      // Nearly fill the cache with pinned entries, then add a bunch of
+      // non-pinned entries. eviction_effort_cap should affect how many
+      // evictable entries are present beyond the cache capacity, despite
+      // being evictable.
+      constexpr size_t kCount = kCapacity - 1;
+      std::unique_ptr<HandleImpl* []> ha { new HandleImpl* [kCount] {} };
+      for (size_t i = 0; i < 2 * kCount; ++i) {
+        UniqueId64x2 hkey = this->CheapHash(i);
+        ASSERT_OK(shard.Insert(
+            this->TestKey(hkey), hkey, nullptr /*value*/, &kNoopCacheItemHelper,
+            1 /*charge*/, i < kCount ? &ha[i] : nullptr, Cache::Priority::LOW));
+      }
+
+      if (strict_capacity_limit) {
+        // If strict_capacity_limit is enabled, the cache will never exceed its
+        // capacity
+        EXPECT_EQ(shard.GetOccupancyCount(), kCapacity);
+      } else {
+        // Rough inverse relationship between cap and possible memory
+        // explosion, which shows up as increased table occupancy count.
+        int effective_eec = std::max(int{1}, eec) + 1;
+        EXPECT_NEAR(shard.GetOccupancyCount() * 1.0,
+                    kCount * (1 + 1.4 / effective_eec),
+                    kCount * (0.6 / effective_eec) + 1.0);
+      }
+
+      for (size_t i = 0; i < kCount; ++i) {
+        shard.Release(ha[i]);
+      }
+    }
+  }
+}
+
 namespace {
 struct DeleteCounter {
   int deleted = 0;
@@ -1035,7 +1091,8 @@ class TestSecondaryCache : public SecondaryCache {
   std::unique_ptr<SecondaryCacheResultHandle> Lookup(
       const Slice& key, const Cache::CacheItemHelper* helper,
       Cache::CreateContext* create_context, bool /*wait*/,
-      bool /*advise_erase*/, bool& kept_in_sec_cache) override {
+      bool /*advise_erase*/, Statistics* /*stats*/,
+      bool& kept_in_sec_cache) override {
     std::string key_str = key.ToString();
     TEST_SYNC_POINT_CALLBACK("TestSecondaryCache::Lookup", &key_str);
 

cache/secondary_cache_adapter.cc

@@ -83,7 +83,10 @@ CacheWithSecondaryAdapter::CacheWithSecondaryAdapter(
     : CacheWrapper(std::move(target)),
       secondary_cache_(std::move(secondary_cache)),
       adm_policy_(adm_policy),
-      distribute_cache_res_(distribute_cache_res) {
+      distribute_cache_res_(distribute_cache_res),
+      placeholder_usage_(0),
+      reserved_usage_(0),
+      sec_reserved_(0) {
   target_->SetEvictionCallback(
       [this](const Slice& key, Handle* handle, bool was_hit) {
         return EvictionHandler(key, handle, was_hit);
@@ -103,8 +106,7 @@ CacheWithSecondaryAdapter::CacheWithSecondaryAdapter(
     // secondary cache is freed from the reservation.
     s = pri_cache_res_->UpdateCacheReservation(sec_capacity);
     assert(s.ok());
-    sec_cache_res_ratio_.store((double)sec_capacity / target_->GetCapacity(),
-                               std::memory_order_relaxed);
+    sec_cache_res_ratio_ = (double)sec_capacity / target_->GetCapacity();
   }
 }
 
@@ -113,10 +115,12 @@ CacheWithSecondaryAdapter::~CacheWithSecondaryAdapter() {
   // use after free
   target_->SetEvictionCallback({});
 #ifndef NDEBUG
-  if (distribute_cache_res_ && !ratio_changed_) {
+  if (distribute_cache_res_) {
     size_t sec_capacity = 0;
     Status s = secondary_cache_->GetCapacity(sec_capacity);
     assert(s.ok());
+    assert(placeholder_usage_ == 0);
+    assert(reserved_usage_ == 0);
     assert(pri_cache_res_->GetTotalMemoryUsed() == sec_capacity);
   }
 #endif  // NDEBUG
@@ -236,13 +240,31 @@ Status CacheWithSecondaryAdapter::Insert(const Slice& key, ObjectPtr value,
                                          const Slice& compressed_value,
                                          CompressionType type) {
   Status s = target_->Insert(key, value, helper, charge, handle, priority);
-  if (s.ok() && value == nullptr && distribute_cache_res_) {
-    size_t sec_charge = static_cast<size_t>(
-        charge * (sec_cache_res_ratio_.load(std::memory_order_relaxed)));
-    s = secondary_cache_->Deflate(sec_charge);
-    assert(s.ok());
-    s = pri_cache_res_->UpdateCacheReservation(sec_charge, /*increase=*/false);
-    assert(s.ok());
+  if (s.ok() && value == nullptr && distribute_cache_res_ && handle) {
+    charge = target_->GetCharge(*handle);
+
+    MutexLock l(&cache_res_mutex_);
+    placeholder_usage_ += charge;
+    // Check if total placeholder reservation is more than the overall
+    // cache capacity. If it is, then we don't try to charge the
+    // secondary cache because we don't want to overcharge it (beyond
+    // its capacity).
+    // In order to make this a bit more lightweight, we also check if
+    // the difference between placeholder_usage_ and reserved_usage_ is
+    // atleast kReservationChunkSize and avoid any adjustments if not.
+    if ((placeholder_usage_ <= target_->GetCapacity()) &&
+        ((placeholder_usage_ - reserved_usage_) >= kReservationChunkSize)) {
+      reserved_usage_ = placeholder_usage_ & ~(kReservationChunkSize - 1);
+      size_t new_sec_reserved =
+          static_cast<size_t>(reserved_usage_ * sec_cache_res_ratio_);
+      size_t sec_charge = new_sec_reserved - sec_reserved_;
+      s = secondary_cache_->Deflate(sec_charge);
+      assert(s.ok());
+      s = pri_cache_res_->UpdateCacheReservation(sec_charge,
+                                                 /*increase=*/false);
+      assert(s.ok());
+      sec_reserved_ += sec_charge;
+    }
   }
   // Warm up the secondary cache with the compressed block. The secondary
   // cache may choose to ignore it based on the admission policy.
@@ -272,7 +294,8 @@ Cache::Handle* CacheWithSecondaryAdapter::Lookup(const Slice& key,
     bool kept_in_sec_cache = false;
     std::unique_ptr<SecondaryCacheResultHandle> secondary_handle =
         secondary_cache_->Lookup(key, helper, create_context, /*wait*/ true,
-                                 found_dummy_entry, /*out*/ kept_in_sec_cache);
+                                 found_dummy_entry, stats,
+                                 /*out*/ kept_in_sec_cache);
     if (secondary_handle) {
       result = Promote(std::move(secondary_handle), key, helper, priority,
                        stats, found_dummy_entry, kept_in_sec_cache);
@@ -287,14 +310,27 @@ bool CacheWithSecondaryAdapter::Release(Handle* handle,
     ObjectPtr v = target_->Value(handle);
     if (v == nullptr && distribute_cache_res_) {
       size_t charge = target_->GetCharge(handle);
-      size_t sec_charge = static_cast<size_t>(
-          charge * (sec_cache_res_ratio_.load(std::memory_order_relaxed)));
-      TEST_SYNC_POINT("CacheWithSecondaryAdapter::Release:ChargeSecCache1");
-      TEST_SYNC_POINT("CacheWithSecondaryAdapter::Release:ChargeSecCache2");
-      Status s = secondary_cache_->Inflate(sec_charge);
-      assert(s.ok());
-      s = pri_cache_res_->UpdateCacheReservation(sec_charge, /*increase=*/true);
-      assert(s.ok());
+
+      MutexLock l(&cache_res_mutex_);
+      placeholder_usage_ -= charge;
+      // Check if total placeholder reservation is more than the overall
+      // cache capacity. If it is, then we do nothing as reserved_usage_ must
+      // be already maxed out
+      if ((placeholder_usage_ <= target_->GetCapacity()) &&
+          (placeholder_usage_ < reserved_usage_)) {
+        // Adjust reserved_usage_ in chunks of kReservationChunkSize, so
+        // we don't hit this slow path too often.
+        reserved_usage_ = placeholder_usage_ & ~(kReservationChunkSize - 1);
+        size_t new_sec_reserved =
+            static_cast<size_t>(reserved_usage_ * sec_cache_res_ratio_);
+        size_t sec_charge = sec_reserved_ - new_sec_reserved;
+        Status s = secondary_cache_->Inflate(sec_charge);
+        assert(s.ok());
+        s = pri_cache_res_->UpdateCacheReservation(sec_charge,
+                                                   /*increase=*/true);
+        assert(s.ok());
+        sec_reserved_ -= sec_charge;
+      }
     }
   }
   return target_->Release(handle, erase_if_last_ref);
@@ -313,10 +349,10 @@ void CacheWithSecondaryAdapter::StartAsyncLookupOnMySecondary(
   assert(async_handle.result_handle == nullptr);
 
   std::unique_ptr<SecondaryCacheResultHandle> secondary_handle =
-      secondary_cache_->Lookup(async_handle.key, async_handle.helper,
-                               async_handle.create_context, /*wait*/ false,
-                               async_handle.found_dummy_entry,
-                               /*out*/ async_handle.kept_in_sec_cache);
+      secondary_cache_->Lookup(
+          async_handle.key, async_handle.helper, async_handle.create_context,
+          /*wait*/ false, async_handle.found_dummy_entry, async_handle.stats,
+          /*out*/ async_handle.kept_in_sec_cache);
   if (secondary_handle) {
     // TODO with stacked secondaries: Check & process if already ready?
     async_handle.pending_handle = secondary_handle.release();
@@ -441,13 +477,11 @@ const char* CacheWithSecondaryAdapter::Name() const {
 // where the new capacity < total cache reservations.
 void CacheWithSecondaryAdapter::SetCapacity(size_t capacity) {
   size_t sec_capacity = static_cast<size_t>(
-      capacity * (distribute_cache_res_
-                      ? sec_cache_res_ratio_.load(std::memory_order_relaxed)
-                      : 0.0));
+      capacity * (distribute_cache_res_ ? sec_cache_res_ratio_ : 0.0));
   size_t old_sec_capacity = 0;
 
   if (distribute_cache_res_) {
-    MutexLock m(&mutex_);
+    MutexLock m(&cache_res_mutex_);
 
     Status s = secondary_cache_->GetCapacity(old_sec_capacity);
     if (!s.ok()) {
@@ -462,9 +496,17 @@ void CacheWithSecondaryAdapter::SetCapacity(size_t capacity) {
       // 3. Decrease the primary cache capacity to the total budget
       s = secondary_cache_->SetCapacity(sec_capacity);
       if (s.ok()) {
+        if (placeholder_usage_ > capacity) {
+          // Adjust reserved_usage_ down
+          reserved_usage_ = capacity & ~(kReservationChunkSize - 1);
+        }
+        size_t new_sec_reserved =
+            static_cast<size_t>(reserved_usage_ * sec_cache_res_ratio_);
         s = pri_cache_res_->UpdateCacheReservation(
-            old_sec_capacity - sec_capacity,
+            (old_sec_capacity - sec_capacity) -
+                (sec_reserved_ - new_sec_reserved),
             /*increase=*/false);
+        sec_reserved_ = new_sec_reserved;
         assert(s.ok());
         target_->SetCapacity(capacity);
       }
@@ -489,6 +531,29 @@ void CacheWithSecondaryAdapter::SetCapacity(size_t capacity) {
   }
 }
 
+Status CacheWithSecondaryAdapter::GetSecondaryCacheCapacity(
+    size_t& size) const {
+  return secondary_cache_->GetCapacity(size);
+}
+
+Status CacheWithSecondaryAdapter::GetSecondaryCachePinnedUsage(
+    size_t& size) const {
+  Status s;
+  if (distribute_cache_res_) {
+    MutexLock m(&cache_res_mutex_);
+    size_t capacity = 0;
+    s = secondary_cache_->GetCapacity(capacity);
+    if (s.ok()) {
+      size = capacity - pri_cache_res_->GetTotalMemoryUsed();
+    } else {
+      size = 0;
+    }
+  } else {
+    size = 0;
+  }
+  return s;
+}
+
 // Update the secondary/primary allocation ratio (remember, the primary
 // capacity is the total memory budget when distribute_cache_res_ is true).
 // When the ratio changes, we may accumulate some error in the calculations
@@ -503,12 +568,11 @@ void CacheWithSecondaryAdapter::SetCapacity(size_t capacity) {
 // in the future.
 Status CacheWithSecondaryAdapter::UpdateCacheReservationRatio(
     double compressed_secondary_ratio) {
-  if (!distribute_cache_res_ ||
-      sec_cache_res_ratio_.load(std::memory_order_relaxed) == 0.0) {
+  if (!distribute_cache_res_) {
     return Status::NotSupported();
   }
 
-  MutexLock m(&mutex_);
+  MutexLock m(&cache_res_mutex_);
   size_t pri_capacity = target_->GetCapacity();
   size_t sec_capacity =
       static_cast<size_t>(pri_capacity * compressed_secondary_ratio);
@@ -518,38 +582,17 @@ Status CacheWithSecondaryAdapter::UpdateCacheReservationRatio(
     return s;
   }
 
-  TEST_SYNC_POINT(
-      "CacheWithSecondaryAdapter::UpdateCacheReservationRatio:Begin");
-
-  // There's a possible race condition here. Since the read of pri_cache_res_
-  // memory used (secondary cache usage charged to primary cache), and the
-  // change to sec_cache_res_ratio_ are not guarded by a mutex, its possible
-  // that an Insert/Release in another thread might decrease/increase the
-  // pri_cache_res_ reservation by the wrong amount. This should not be a
-  // problem because updating the sec/pri ratio is a rare operation, and
-  // the worst that can happen is we may over/under charge the secondary
-  // cache usage by a little bit. But we do need to protect against
-  // underflow of old_sec_reserved.
-  // TODO: Make the accounting more accurate by tracking the total memory
-  // reservation on the primary cache. This will also allow us to remove
-  // the restriction of not being able to change the sec/pri ratio from
-  // 0.0 to higher.
-  size_t sec_charge_to_pri = pri_cache_res_->GetTotalMemoryUsed();
-  size_t old_sec_reserved = (old_sec_capacity > sec_charge_to_pri)
-                                ? (old_sec_capacity - sec_charge_to_pri)
-                                : 0;
   // Calculate the new secondary cache reservation
-  size_t sec_reserved = static_cast<size_t>(
-      old_sec_reserved *
-      (double)(compressed_secondary_ratio /
-               sec_cache_res_ratio_.load(std::memory_order_relaxed)));
-  sec_cache_res_ratio_.store(compressed_secondary_ratio,
-                             std::memory_order_relaxed);
+  // reserved_usage_ will never be > the cache capacity, so we don't
+  // have to worry about adjusting it here.
+  sec_cache_res_ratio_ = compressed_secondary_ratio;
+  size_t new_sec_reserved =
+      static_cast<size_t>(reserved_usage_ * sec_cache_res_ratio_);
   if (sec_capacity > old_sec_capacity) {
     // We're increasing the ratio, thus ending up with a larger secondary
     // cache and a smaller usable primary cache capacity. Similar to
     // SetCapacity(), we try to avoid a temporary increase in total usage
-    // beyond teh configured capacity -
+    // beyond the configured capacity -
     // 1. A higher secondary cache ratio means it gets a higher share of
     //    cache reservations. So first account for that by deflating the
     //    secondary cache
@@ -557,12 +600,13 @@ Status CacheWithSecondaryAdapter::UpdateCacheReservationRatio(
     //    cache utilization (increase in capacity - increase in share of cache
     //    reservation)
     // 3. Increase secondary cache capacity
-    s = secondary_cache_->Deflate(sec_reserved - old_sec_reserved);
+    s = secondary_cache_->Deflate(new_sec_reserved - sec_reserved_);
     assert(s.ok());
     s = pri_cache_res_->UpdateCacheReservation(
-        (sec_capacity - old_sec_capacity) - (sec_reserved - old_sec_reserved),
+        (sec_capacity - old_sec_capacity) - (new_sec_reserved - sec_reserved_),
         /*increase=*/true);
     assert(s.ok());
+    sec_reserved_ = new_sec_reserved;
     s = secondary_cache_->SetCapacity(sec_capacity);
     assert(s.ok());
   } else {
@@ -576,21 +620,16 @@ Status CacheWithSecondaryAdapter::UpdateCacheReservationRatio(
     s = secondary_cache_->SetCapacity(sec_capacity);
     if (s.ok()) {
       s = pri_cache_res_->UpdateCacheReservation(
-          (old_sec_capacity - sec_capacity) - (old_sec_reserved - sec_reserved),
+          (old_sec_capacity - sec_capacity) -
+              (sec_reserved_ - new_sec_reserved),
           /*increase=*/false);
       assert(s.ok());
-      s = secondary_cache_->Inflate(old_sec_reserved - sec_reserved);
+      s = secondary_cache_->Inflate(sec_reserved_ - new_sec_reserved);
       assert(s.ok());
+      sec_reserved_ = new_sec_reserved;
     }
   }
 
-  TEST_SYNC_POINT("CacheWithSecondaryAdapter::UpdateCacheReservationRatio:End");
-#ifndef NDEBUG
-  // As mentioned in the function comments, we may accumulate some erros when
-  // the ratio is changed. We set a flag here which disables some assertions
-  // in the destructor
-  ratio_changed_ = true;
-#endif
   return s;
 }
 
@@ -645,12 +684,14 @@ std::shared_ptr<Cache> NewTieredCache(const TieredCacheOptions& _opts) {
         *(static_cast_with_check<LRUCacheOptions, ShardedCacheOptions>(
             opts.cache_opts));
     cache_opts.capacity = opts.total_capacity;
+    cache_opts.secondary_cache = nullptr;
     cache = cache_opts.MakeSharedCache();
   } else if (opts.cache_type == PrimaryCacheType::kCacheTypeHCC) {
     HyperClockCacheOptions cache_opts =
         *(static_cast_with_check<HyperClockCacheOptions, ShardedCacheOptions>(
             opts.cache_opts));
     cache_opts.capacity = opts.total_capacity;
+    cache_opts.secondary_cache = nullptr;
     cache = cache_opts.MakeSharedCache();
   } else {
     return nullptr;

cache/secondary_cache_adapter.h

@@ -47,6 +47,10 @@ class CacheWithSecondaryAdapter : public CacheWrapper {
 
   void SetCapacity(size_t capacity) override;
 
+  Status GetSecondaryCacheCapacity(size_t& size) const override;
+
+  Status GetSecondaryCachePinnedUsage(size_t& size) const override;
+
   Status UpdateCacheReservationRatio(double ratio);
 
   Status UpdateAdmissionPolicy(TieredAdmissionPolicy adm_policy);
@@ -56,6 +60,8 @@ class CacheWithSecondaryAdapter : public CacheWrapper {
   SecondaryCache* TEST_GetSecondaryCache() { return secondary_cache_.get(); }
 
  private:
+  static constexpr size_t kReservationChunkSize = 1 << 20;
+
   bool EvictionHandler(const Slice& key, Handle* handle, bool was_hit);
 
   void StartAsyncLookupOnMySecondary(AsyncLookupHandle& async_handle);
@@ -80,11 +86,18 @@ class CacheWithSecondaryAdapter : public CacheWrapper {
   std::shared_ptr<ConcurrentCacheReservationManager> pri_cache_res_;
   // Fraction of a cache memory reservation to be assigned to the secondary
   // cache
-  std::atomic<double> sec_cache_res_ratio_;
-  port::Mutex mutex_;
-#ifndef NDEBUG
-  bool ratio_changed_ = false;
-#endif
+  double sec_cache_res_ratio_;
+  // Mutex for use when managing cache memory reservations. Should not be used
+  // for other purposes, as it may risk causing deadlocks.
+  mutable port::Mutex cache_res_mutex_;
+  // Total memory reserved by placeholder entriesin the cache
+  size_t placeholder_usage_;
+  // Total placeholoder memory charged to both the primary and secondary
+  // caches. Will be <= placeholder_usage_.
+  size_t reserved_usage_;
+  // Amount of memory reserved in the secondary cache. This should be
+  // reserved_usage_ * sec_cache_res_ratio_ in steady state.
+  size_t sec_reserved_;
 };
 
 }  // namespace ROCKSDB_NAMESPACE

cache/sharded_cache.cc

@@ -83,6 +83,16 @@ size_t ShardedCacheBase::GetCapacity() const {
   return capacity_;
 }
 
+Status ShardedCacheBase::GetSecondaryCacheCapacity(size_t& size) const {
+  size = 0;
+  return Status::OK();
+}
+
+Status ShardedCacheBase::GetSecondaryCachePinnedUsage(size_t& size) const {
+  size = 0;
+  return Status::OK();
+}
+
 bool ShardedCacheBase::HasStrictCapacityLimit() const {
   MutexLock l(&config_mutex_);
   return strict_capacity_limit_;

cache/sharded_cache.h

@@ -99,6 +99,8 @@ class ShardedCacheBase : public Cache {
 
   bool HasStrictCapacityLimit() const override;
   size_t GetCapacity() const override;
+  Status GetSecondaryCacheCapacity(size_t& size) const override;
+  Status GetSecondaryCachePinnedUsage(size_t& size) const override;
 
   using Cache::GetUsage;
   size_t GetUsage(Handle* handle) const override;

cache/tiered_secondary_cache.cc

@@ -5,6 +5,8 @@
 
 #include "cache/tiered_secondary_cache.h"
 
+#include "monitoring/statistics_impl.h"
+
 namespace ROCKSDB_NAMESPACE {
 
 // Creation callback for use in the lookup path. It calls the upper layer
@@ -29,6 +31,9 @@ Status TieredSecondaryCache::MaybeInsertAndCreate(
     // TODO: Don't hardcode the source
     context->comp_sec_cache->InsertSaved(*context->key, data, type, source)
         .PermitUncheckedError();
+    RecordTick(context->stats, COMPRESSED_SECONDARY_CACHE_PROMOTIONS);
+  } else {
+    RecordTick(context->stats, COMPRESSED_SECONDARY_CACHE_PROMOTION_SKIPS);
   }
   // Primary cache will accept the object, so call its helper to create
   // the object
@@ -43,10 +48,10 @@ Status TieredSecondaryCache::MaybeInsertAndCreate(
 std::unique_ptr<SecondaryCacheResultHandle> TieredSecondaryCache::Lookup(
     const Slice& key, const Cache::CacheItemHelper* helper,
     Cache::CreateContext* create_context, bool wait, bool advise_erase,
-    bool& kept_in_sec_cache) {
+    Statistics* stats, bool& kept_in_sec_cache) {
   bool dummy = false;
   std::unique_ptr<SecondaryCacheResultHandle> result =
-      target()->Lookup(key, helper, create_context, wait, advise_erase,
+      target()->Lookup(key, helper, create_context, wait, advise_erase, stats,
                        /*kept_in_sec_cache=*/dummy);
   // We never want the item to spill back into the secondary cache
   kept_in_sec_cache = true;
@@ -66,9 +71,10 @@ std::unique_ptr<SecondaryCacheResultHandle> TieredSecondaryCache::Lookup(
     ctx.helper = helper;
     ctx.inner_ctx = create_context;
     ctx.comp_sec_cache = target();
+    ctx.stats = stats;
 
     return nvm_sec_cache_->Lookup(key, outer_helper, &ctx, wait, advise_erase,
-                                  kept_in_sec_cache);
+                                  stats, kept_in_sec_cache);
   }
 
   // If wait is false, i.e its an async lookup, we have to allocate a result
@@ -80,8 +86,10 @@ std::unique_ptr<SecondaryCacheResultHandle> TieredSecondaryCache::Lookup(
   handle->ctx()->helper = helper;
   handle->ctx()->inner_ctx = create_context;
   handle->ctx()->comp_sec_cache = target();
-  handle->SetInnerHandle(nvm_sec_cache_->Lookup(
-      key, outer_helper, handle->ctx(), wait, advise_erase, kept_in_sec_cache));
+  handle->ctx()->stats = stats;
+  handle->SetInnerHandle(
+      nvm_sec_cache_->Lookup(key, outer_helper, handle->ctx(), wait,
+                             advise_erase, stats, kept_in_sec_cache));
   if (!handle->inner_handle()) {
     handle.reset();
   } else {
@@ -109,10 +117,8 @@ void TieredSecondaryCache::WaitAll(
   }
   nvm_sec_cache_->WaitAll(nvm_handles);
   for (auto handle : my_handles) {
-    assert(handle->IsReady());
-    auto nvm_handle = handle->inner_handle();
-    handle->SetSize(nvm_handle->Size());
-    handle->SetValue(nvm_handle->Value());
+    assert(handle->inner_handle()->IsReady());
+    handle->Complete();
   }
 }
 

cache/tiered_secondary_cache.h

@@ -59,7 +59,7 @@ class TieredSecondaryCache : public SecondaryCacheWrapper {
   virtual std::unique_ptr<SecondaryCacheResultHandle> Lookup(
       const Slice& key, const Cache::CacheItemHelper* helper,
       Cache::CreateContext* create_context, bool wait, bool advise_erase,
-      bool& kept_in_sec_cache) override;
+      Statistics* stats, bool& kept_in_sec_cache) override;
 
   virtual void WaitAll(
       std::vector<SecondaryCacheResultHandle*> handles) override;
@@ -72,6 +72,7 @@ class TieredSecondaryCache : public SecondaryCacheWrapper {
     Cache::CreateContext* inner_ctx;
     std::shared_ptr<SecondaryCacheResultHandle> inner_handle;
     SecondaryCache* comp_sec_cache;
+    Statistics* stats;
   };
 
   class ResultHandle : public SecondaryCacheResultHandle {
@@ -79,7 +80,10 @@ class TieredSecondaryCache : public SecondaryCacheWrapper {
     ~ResultHandle() override {}
 
     bool IsReady() override {
-      return !inner_handle_ || inner_handle_->IsReady();
+      if (inner_handle_ && inner_handle_->IsReady()) {
+        Complete();
+      }
+      return ready_;
     }
 
     void Wait() override {
@@ -92,10 +96,10 @@ class TieredSecondaryCache : public SecondaryCacheWrapper {
     Cache::ObjectPtr Value() override { return value_; }
 
     void Complete() {
-      assert(IsReady());
       size_ = inner_handle_->Size();
       value_ = inner_handle_->Value();
       inner_handle_.reset();
+      ready_ = true;
     }
 
     void SetInnerHandle(std::unique_ptr<SecondaryCacheResultHandle>&& handle) {
@@ -115,6 +119,7 @@ class TieredSecondaryCache : public SecondaryCacheWrapper {
     CreateContext ctx_;
     size_t size_;
     Cache::ObjectPtr value_;
+    bool ready_ = false;
   };
 
   static void NoopDelete(Cache::ObjectPtr /*obj*/,

cache/tiered_secondary_cache_test.cc

@@ -15,10 +15,11 @@ namespace ROCKSDB_NAMESPACE {
 
 class TestSecondaryCache : public SecondaryCache {
  public:
-  explicit TestSecondaryCache(size_t capacity)
+  explicit TestSecondaryCache(size_t capacity, bool ready_before_wait)
       : cache_(NewLRUCache(capacity, 0, false, 0.5 /* high_pri_pool_ratio */,
                            nullptr, kDefaultToAdaptiveMutex,
                            kDontChargeCacheMetadata)),
+        ready_before_wait_(ready_before_wait),
         num_insert_saved_(0),
         num_hits_(0),
         num_misses_(0) {}
@@ -61,7 +62,7 @@ class TestSecondaryCache : public SecondaryCache {
   std::unique_ptr<SecondaryCacheResultHandle> Lookup(
       const Slice& key, const Cache::CacheItemHelper* helper,
       Cache::CreateContext* create_context, bool wait, bool /*advise_erase*/,
-      bool& kept_in_sec_cache) override {
+      Statistics* /*stats*/, bool& kept_in_sec_cache) override {
     std::string key_str = key.ToString();
     TEST_SYNC_POINT_CALLBACK("TestSecondaryCache::Lookup", &key_str);
 
@@ -88,7 +89,8 @@ class TestSecondaryCache : public SecondaryCache {
                             /*alloc*/ nullptr, &value, &charge);
       if (s.ok()) {
         secondary_handle.reset(new TestSecondaryCacheResultHandle(
-            cache_.get(), handle, value, charge, /*ready=*/wait));
+            cache_.get(), handle, value, charge,
+            /*ready=*/wait || ready_before_wait_));
         kept_in_sec_cache = true;
       } else {
         cache_.Release(handle);
@@ -168,6 +170,7 @@ class TestSecondaryCache : public SecondaryCache {
       BasicTypedSharedCacheInterface<char[], CacheEntryRole::kMisc>;
   using TypedHandle = SharedCache::TypedHandle;
   SharedCache cache_;
+  bool ready_before_wait_;
   uint32_t num_insert_saved_;
   uint32_t num_hits_;
   uint32_t num_misses_;
@@ -179,11 +182,10 @@ class DBTieredSecondaryCacheTest : public DBTestBase {
   DBTieredSecondaryCacheTest()
       : DBTestBase("db_tiered_secondary_cache_test", /*env_do_fsync=*/true) {}
 
-  std::shared_ptr<Cache> NewCache(size_t pri_capacity,
-                                  size_t compressed_capacity,
-                                  size_t nvm_capacity,
-                                  TieredAdmissionPolicy adm_policy =
-                                      TieredAdmissionPolicy::kAdmPolicyAuto) {
+  std::shared_ptr<Cache> NewCache(
+      size_t pri_capacity, size_t compressed_capacity, size_t nvm_capacity,
+      TieredAdmissionPolicy adm_policy = TieredAdmissionPolicy::kAdmPolicyAuto,
+      bool ready_before_wait = false) {
     LRUCacheOptions lru_opts;
     TieredCacheOptions opts;
     lru_opts.capacity = 0;
@@ -194,10 +196,11 @@ class DBTieredSecondaryCacheTest : public DBTestBase {
     opts.comp_cache_opts.capacity = 0;
     opts.comp_cache_opts.num_shard_bits = 0;
     opts.total_capacity = pri_capacity + compressed_capacity;
-    opts.compressed_secondary_ratio =
+    opts.compressed_secondary_ratio = compressed_secondary_ratio_ =
         (double)compressed_capacity / opts.total_capacity;
     if (nvm_capacity > 0) {
-      nvm_sec_cache_.reset(new TestSecondaryCache(nvm_capacity));
+      nvm_sec_cache_.reset(
+          new TestSecondaryCache(nvm_capacity, ready_before_wait));
       opts.nvm_sec_cache = nvm_sec_cache_;
     }
     opts.adm_policy = adm_policy;
@@ -207,6 +210,12 @@ class DBTieredSecondaryCacheTest : public DBTestBase {
     return cache_;
   }
 
+  void ClearPrimaryCache() {
+    ASSERT_EQ(UpdateTieredCache(cache_, -1, 1.0), Status::OK());
+    ASSERT_EQ(UpdateTieredCache(cache_, -1, compressed_secondary_ratio_),
+              Status::OK());
+  }
+
   TestSecondaryCache* nvm_sec_cache() { return nvm_sec_cache_.get(); }
 
   CompressedSecondaryCache* compressed_secondary_cache() {
@@ -218,6 +227,7 @@ class DBTieredSecondaryCacheTest : public DBTestBase {
  private:
   std::shared_ptr<Cache> cache_;
   std::shared_ptr<TestSecondaryCache> nvm_sec_cache_;
+  double compressed_secondary_ratio_;
 };
 
 // In this test, the block size is set to 4096. Each value is 1007 bytes, so
@@ -582,6 +592,116 @@ TEST_F(DBTieredSecondaryCacheTest, WaitAllTest) {
   Destroy(options);
 }
 
+TEST_F(DBTieredSecondaryCacheTest, ReadyBeforeWaitAllTest) {
+  if (!LZ4_Supported()) {
+    ROCKSDB_GTEST_SKIP("This test requires LZ4 support.");
+    return;
+  }
+
+  BlockBasedTableOptions table_options;
+  table_options.block_cache = NewCache(250 * 1024, 20 * 1024, 256 * 1024,
+                                       TieredAdmissionPolicy::kAdmPolicyAuto,
+                                       /*ready_before_wait=*/true);
+  table_options.block_size = 4 * 1024;
+  table_options.cache_index_and_filter_blocks = false;
+  Options options = GetDefaultOptions();
+  options.create_if_missing = true;
+  options.table_factory.reset(NewBlockBasedTableFactory(table_options));
+  options.statistics = CreateDBStatistics();
+
+  options.paranoid_file_checks = false;
+  DestroyAndReopen(options);
+  Random rnd(301);
+  const int N = 256;
+  for (int i = 0; i < N; i++) {
+    std::string p_v;
+    test::CompressibleString(&rnd, 0.5, 1007, &p_v);
+    ASSERT_OK(Put(Key(i), p_v));
+  }
+
+  ASSERT_OK(Flush());
+
+  std::vector<std::string> keys;
+  std::vector<std::string> values;
+
+  keys.push_back(Key(0));
+  keys.push_back(Key(4));
+  keys.push_back(Key(8));
+  values = MultiGet(keys, /*snapshot=*/nullptr, /*async=*/true);
+  ASSERT_EQ(values.size(), keys.size());
+  for (auto value : values) {
+    ASSERT_EQ(1007, value.size());
+  }
+  ASSERT_EQ(nvm_sec_cache()->num_insert_saved(), 3u);
+  ASSERT_EQ(nvm_sec_cache()->num_misses(), 3u);
+  ASSERT_EQ(nvm_sec_cache()->num_hits(), 0u);
+  ASSERT_EQ(options.statistics->getTickerCount(BLOCK_CACHE_MISS), 3u);
+
+  keys.clear();
+  values.clear();
+  keys.push_back(Key(12));
+  keys.push_back(Key(16));
+  keys.push_back(Key(20));
+  values = MultiGet(keys, /*snapshot=*/nullptr, /*async=*/true);
+  ASSERT_EQ(values.size(), keys.size());
+  for (auto value : values) {
+    ASSERT_EQ(1007, value.size());
+  }
+  ASSERT_EQ(nvm_sec_cache()->num_insert_saved(), 6u);
+  ASSERT_EQ(nvm_sec_cache()->num_misses(), 6u);
+  ASSERT_EQ(nvm_sec_cache()->num_hits(), 0u);
+  ASSERT_EQ(options.statistics->getTickerCount(BLOCK_CACHE_MISS), 6u);
+
+  keys.clear();
+  values.clear();
+  keys.push_back(Key(0));
+  keys.push_back(Key(4));
+  keys.push_back(Key(8));
+  values = MultiGet(keys, /*snapshot=*/nullptr, /*async=*/true);
+  ASSERT_EQ(values.size(), keys.size());
+  for (auto value : values) {
+    ASSERT_EQ(1007, value.size());
+  }
+  ASSERT_EQ(nvm_sec_cache()->num_insert_saved(), 6u);
+  ASSERT_EQ(nvm_sec_cache()->num_misses(), 6u);
+  ASSERT_EQ(nvm_sec_cache()->num_hits(), 3u);
+  ASSERT_EQ(options.statistics->getTickerCount(BLOCK_CACHE_MISS), 6u);
+
+  ClearPrimaryCache();
+
+  keys.clear();
+  values.clear();
+  keys.push_back(Key(0));
+  keys.push_back(Key(32));
+  keys.push_back(Key(36));
+  values = MultiGet(keys, /*snapshot=*/nullptr, /*async=*/true);
+  ASSERT_EQ(values.size(), keys.size());
+  for (auto value : values) {
+    ASSERT_EQ(1007, value.size());
+  }
+  ASSERT_EQ(nvm_sec_cache()->num_insert_saved(), 8u);
+  ASSERT_EQ(nvm_sec_cache()->num_misses(), 8u);
+  ASSERT_EQ(nvm_sec_cache()->num_hits(), 4u);
+  ASSERT_EQ(options.statistics->getTickerCount(BLOCK_CACHE_MISS), 8u);
+
+  keys.clear();
+  values.clear();
+  keys.push_back(Key(0));
+  keys.push_back(Key(32));
+  keys.push_back(Key(36));
+  values = MultiGet(keys, /*snapshot=*/nullptr, /*async=*/true);
+  ASSERT_EQ(values.size(), keys.size());
+  for (auto value : values) {
+    ASSERT_EQ(1007, value.size());
+  }
+  ASSERT_EQ(nvm_sec_cache()->num_insert_saved(), 8u);
+  ASSERT_EQ(nvm_sec_cache()->num_misses(), 8u);
+  ASSERT_EQ(nvm_sec_cache()->num_hits(), 4u);
+  ASSERT_EQ(options.statistics->getTickerCount(BLOCK_CACHE_MISS), 8u);
+
+  Destroy(options);
+}
+
 // This test is for iteration. It iterates through a set of keys in two
 // passes. First pass loads the compressed blocks into the nvm tier, and
 // the second pass should hit all of those blocks.

db/blob/blob_file_builder.cc

@@ -34,9 +34,9 @@ BlobFileBuilder::BlobFileBuilder(
     VersionSet* versions, FileSystem* fs,
     const ImmutableOptions* immutable_options,
     const MutableCFOptions* mutable_cf_options, const FileOptions* file_options,
-    std::string db_id, std::string db_session_id, int job_id,
-    uint32_t column_family_id, const std::string& column_family_name,
-    Env::IOPriority io_priority, Env::WriteLifeTimeHint write_hint,
+    const WriteOptions* write_options, std::string db_id,
+    std::string db_session_id, int job_id, uint32_t column_family_id,
+    const std::string& column_family_name, Env::WriteLifeTimeHint write_hint,
     const std::shared_ptr<IOTracer>& io_tracer,
     BlobFileCompletionCallback* blob_callback,
     BlobFileCreationReason creation_reason,
@@ -44,18 +44,18 @@ BlobFileBuilder::BlobFileBuilder(
     std::vector<BlobFileAddition>* blob_file_additions)
     : BlobFileBuilder([versions]() { return versions->NewFileNumber(); }, fs,
                       immutable_options, mutable_cf_options, file_options,
-                      db_id, db_session_id, job_id, column_family_id,
-                      column_family_name, io_priority, write_hint, io_tracer,
-                      blob_callback, creation_reason, blob_file_paths,
-                      blob_file_additions) {}
+                      write_options, db_id, db_session_id, job_id,
+                      column_family_id, column_family_name, write_hint,
+                      io_tracer, blob_callback, creation_reason,
+                      blob_file_paths, blob_file_additions) {}
 
 BlobFileBuilder::BlobFileBuilder(
     std::function<uint64_t()> file_number_generator, FileSystem* fs,
     const ImmutableOptions* immutable_options,
     const MutableCFOptions* mutable_cf_options, const FileOptions* file_options,
-    std::string db_id, std::string db_session_id, int job_id,
-    uint32_t column_family_id, const std::string& column_family_name,
-    Env::IOPriority io_priority, Env::WriteLifeTimeHint write_hint,
+    const WriteOptions* write_options, std::string db_id,
+    std::string db_session_id, int job_id, uint32_t column_family_id,
+    const std::string& column_family_name, Env::WriteLifeTimeHint write_hint,
     const std::shared_ptr<IOTracer>& io_tracer,
     BlobFileCompletionCallback* blob_callback,
     BlobFileCreationReason creation_reason,
@@ -69,12 +69,12 @@ BlobFileBuilder::BlobFileBuilder(
       blob_compression_type_(mutable_cf_options->blob_compression_type),
       prepopulate_blob_cache_(mutable_cf_options->prepopulate_blob_cache),
       file_options_(file_options),
+      write_options_(write_options),
       db_id_(std::move(db_id)),
       db_session_id_(std::move(db_session_id)),
       job_id_(job_id),
       column_family_id_(column_family_id),
       column_family_name_(column_family_name),
-      io_priority_(io_priority),
       write_hint_(write_hint),
       io_tracer_(io_tracer),
       blob_callback_(blob_callback),
@@ -87,6 +87,7 @@ BlobFileBuilder::BlobFileBuilder(
   assert(fs_);
   assert(immutable_options_);
   assert(file_options_);
+  assert(write_options_);
   assert(blob_file_paths_);
   assert(blob_file_paths_->empty());
   assert(blob_file_additions_);
@@ -207,14 +208,14 @@ Status BlobFileBuilder::OpenBlobFileIfNeeded() {
   blob_file_paths_->emplace_back(std::move(blob_file_path));
 
   assert(file);
-  file->SetIOPriority(io_priority_);
+  file->SetIOPriority(write_options_->rate_limiter_priority);
   file->SetWriteLifeTimeHint(write_hint_);
   FileTypeSet tmp_set = immutable_options_->checksum_handoff_file_types;
   Statistics* const statistics = immutable_options_->stats;
   std::unique_ptr<WritableFileWriter> file_writer(new WritableFileWriter(
       std::move(file), blob_file_paths_->back(), *file_options_,
       immutable_options_->clock, io_tracer_, statistics,
-      immutable_options_->listeners,
+      Histograms::BLOB_DB_BLOB_FILE_WRITE_MICROS, immutable_options_->listeners,
       immutable_options_->file_checksum_gen_factory.get(),
       tmp_set.Contains(FileType::kBlobFile), false));
 
@@ -231,7 +232,7 @@ Status BlobFileBuilder::OpenBlobFileIfNeeded() {
                        expiration_range);
 
   {
-    Status s = blob_log_writer->WriteHeader(header);
+    Status s = blob_log_writer->WriteHeader(*write_options_, header);
 
     TEST_SYNC_POINT_CALLBACK(
         "BlobFileBuilder::OpenBlobFileIfNeeded:WriteHeader", &s);
@@ -296,7 +297,8 @@ Status BlobFileBuilder::WriteBlobToFile(const Slice& key, const Slice& blob,
 
   uint64_t key_offset = 0;
 
-  Status s = writer_->AddRecord(key, blob, &key_offset, blob_offset);
+  Status s =
+      writer_->AddRecord(*write_options_, key, blob, &key_offset, blob_offset);
 
   TEST_SYNC_POINT_CALLBACK("BlobFileBuilder::WriteBlobToFile:AddRecord", &s);
 
@@ -321,7 +323,8 @@ Status BlobFileBuilder::CloseBlobFile() {
   std::string checksum_method;
   std::string checksum_value;
 
-  Status s = writer_->AppendFooter(footer, &checksum_method, &checksum_value);
+  Status s = writer_->AppendFooter(*write_options_, footer, &checksum_method,
+                                   &checksum_value);
 
   TEST_SYNC_POINT_CALLBACK("BlobFileBuilder::WriteBlobToFile:AppendFooter", &s);
 

db/blob/blob_file_builder.h

@@ -13,6 +13,7 @@
 #include "rocksdb/advanced_options.h"
 #include "rocksdb/compression_type.h"
 #include "rocksdb/env.h"
+#include "rocksdb/options.h"
 #include "rocksdb/rocksdb_namespace.h"
 #include "rocksdb/types.h"
 
@@ -36,11 +37,11 @@ class BlobFileBuilder {
   BlobFileBuilder(VersionSet* versions, FileSystem* fs,
                   const ImmutableOptions* immutable_options,
                   const MutableCFOptions* mutable_cf_options,
-                  const FileOptions* file_options, std::string db_id,
+                  const FileOptions* file_options,
+                  const WriteOptions* write_options, std::string db_id,
                   std::string db_session_id, 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,
                   BlobFileCompletionCallback* blob_callback,
@@ -51,11 +52,11 @@ class BlobFileBuilder {
   BlobFileBuilder(std::function<uint64_t()> file_number_generator,
                   FileSystem* fs, const ImmutableOptions* immutable_options,
                   const MutableCFOptions* mutable_cf_options,
-                  const FileOptions* file_options, std::string db_id,
+                  const FileOptions* file_options,
+                  const WriteOptions* write_options, std::string db_id,
                   std::string db_session_id, 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,
                   BlobFileCompletionCallback* blob_callback,
@@ -92,12 +93,12 @@ class BlobFileBuilder {
   CompressionType blob_compression_type_;
   PrepopulateBlobCache prepopulate_blob_cache_;
   const FileOptions* file_options_;
+  const WriteOptions* write_options_;
   const std::string db_id_;
   const std::string db_session_id_;
   int job_id_;
   uint32_t column_family_id_;
   std::string column_family_name_;
-  Env::IOPriority io_priority_;
   Env::WriteLifeTimeHint write_hint_;
   std::shared_ptr<IOTracer> io_tracer_;
   BlobFileCompletionCallback* blob_callback_;

db/blob/blob_file_builder_test.cc

@@ -43,6 +43,7 @@ class BlobFileBuilderTest : public testing::Test {
     mock_env_.reset(MockEnv::Create(Env::Default()));
     fs_ = mock_env_->GetFileSystem().get();
     clock_ = mock_env_->GetSystemClock().get();
+    write_options_.rate_limiter_priority = Env::IO_HIGH;
   }
 
   void VerifyBlobFile(uint64_t blob_file_number,
@@ -113,6 +114,7 @@ class BlobFileBuilderTest : public testing::Test {
   FileSystem* fs_;
   SystemClock* clock_;
   FileOptions file_options_;
+  WriteOptions write_options_;
 };
 
 TEST_F(BlobFileBuilderTest, BuildAndCheckOneFile) {
@@ -136,7 +138,6 @@ TEST_F(BlobFileBuilderTest, BuildAndCheckOneFile) {
   constexpr int job_id = 1;
   constexpr uint32_t column_family_id = 123;
   constexpr char column_family_name[] = "foobar";
-  constexpr Env::IOPriority io_priority = Env::IO_HIGH;
   constexpr Env::WriteLifeTimeHint write_hint = Env::WLTH_MEDIUM;
 
   std::vector<std::string> blob_file_paths;
@@ -144,8 +145,8 @@ TEST_F(BlobFileBuilderTest, BuildAndCheckOneFile) {
 
   BlobFileBuilder builder(
       TestFileNumberGenerator(), fs_, &immutable_options, &mutable_cf_options,
-      &file_options_, "" /*db_id*/, "" /*db_session_id*/, job_id,
-      column_family_id, column_family_name, io_priority, write_hint,
+      &file_options_, &write_options_, "" /*db_id*/, "" /*db_session_id*/,
+      job_id, column_family_id, column_family_name, write_hint,
       nullptr /*IOTracer*/, nullptr /*BlobFileCompletionCallback*/,
       BlobFileCreationReason::kFlush, &blob_file_paths, &blob_file_additions);
 
@@ -221,7 +222,6 @@ TEST_F(BlobFileBuilderTest, BuildAndCheckMultipleFiles) {
   constexpr int job_id = 1;
   constexpr uint32_t column_family_id = 123;
   constexpr char column_family_name[] = "foobar";
-  constexpr Env::IOPriority io_priority = Env::IO_HIGH;
   constexpr Env::WriteLifeTimeHint write_hint = Env::WLTH_MEDIUM;
 
   std::vector<std::string> blob_file_paths;
@@ -229,8 +229,8 @@ TEST_F(BlobFileBuilderTest, BuildAndCheckMultipleFiles) {
 
   BlobFileBuilder builder(
       TestFileNumberGenerator(), fs_, &immutable_options, &mutable_cf_options,
-      &file_options_, "" /*db_id*/, "" /*db_session_id*/, job_id,
-      column_family_id, column_family_name, io_priority, write_hint,
+      &file_options_, &write_options_, "" /*db_id*/, "" /*db_session_id*/,
+      job_id, column_family_id, column_family_name, write_hint,
       nullptr /*IOTracer*/, nullptr /*BlobFileCompletionCallback*/,
       BlobFileCreationReason::kFlush, &blob_file_paths, &blob_file_additions);
 
@@ -309,7 +309,6 @@ TEST_F(BlobFileBuilderTest, InlinedValues) {
   constexpr int job_id = 1;
   constexpr uint32_t column_family_id = 123;
   constexpr char column_family_name[] = "foobar";
-  constexpr Env::IOPriority io_priority = Env::IO_HIGH;
   constexpr Env::WriteLifeTimeHint write_hint = Env::WLTH_MEDIUM;
 
   std::vector<std::string> blob_file_paths;
@@ -317,8 +316,8 @@ TEST_F(BlobFileBuilderTest, InlinedValues) {
 
   BlobFileBuilder builder(
       TestFileNumberGenerator(), fs_, &immutable_options, &mutable_cf_options,
-      &file_options_, "" /*db_id*/, "" /*db_session_id*/, job_id,
-      column_family_id, column_family_name, io_priority, write_hint,
+      &file_options_, &write_options_, "" /*db_id*/, "" /*db_session_id*/,
+      job_id, column_family_id, column_family_name, write_hint,
       nullptr /*IOTracer*/, nullptr /*BlobFileCompletionCallback*/,
       BlobFileCreationReason::kFlush, &blob_file_paths, &blob_file_additions);
 
@@ -364,7 +363,6 @@ TEST_F(BlobFileBuilderTest, Compression) {
   constexpr int job_id = 1;
   constexpr uint32_t column_family_id = 123;
   constexpr char column_family_name[] = "foobar";
-  constexpr Env::IOPriority io_priority = Env::IO_HIGH;
   constexpr Env::WriteLifeTimeHint write_hint = Env::WLTH_MEDIUM;
 
   std::vector<std::string> blob_file_paths;
@@ -372,8 +370,8 @@ TEST_F(BlobFileBuilderTest, Compression) {
 
   BlobFileBuilder builder(
       TestFileNumberGenerator(), fs_, &immutable_options, &mutable_cf_options,
-      &file_options_, "" /*db_id*/, "" /*db_session_id*/, job_id,
-      column_family_id, column_family_name, io_priority, write_hint,
+      &file_options_, &write_options_, "" /*db_id*/, "" /*db_session_id*/,
+      job_id, column_family_id, column_family_name, write_hint,
       nullptr /*IOTracer*/, nullptr /*BlobFileCompletionCallback*/,
       BlobFileCreationReason::kFlush, &blob_file_paths, &blob_file_additions);
 
@@ -448,7 +446,6 @@ TEST_F(BlobFileBuilderTest, CompressionError) {
   constexpr int job_id = 1;
   constexpr uint32_t column_family_id = 123;
   constexpr char column_family_name[] = "foobar";
-  constexpr Env::IOPriority io_priority = Env::IO_HIGH;
   constexpr Env::WriteLifeTimeHint write_hint = Env::WLTH_MEDIUM;
 
   std::vector<std::string> blob_file_paths;
@@ -456,8 +453,8 @@ TEST_F(BlobFileBuilderTest, CompressionError) {
 
   BlobFileBuilder builder(
       TestFileNumberGenerator(), fs_, &immutable_options, &mutable_cf_options,
-      &file_options_, "" /*db_id*/, "" /*db_session_id*/, job_id,
-      column_family_id, column_family_name, io_priority, write_hint,
+      &file_options_, &write_options_, "" /*db_id*/, "" /*db_session_id*/,
+      job_id, column_family_id, column_family_name, write_hint,
       nullptr /*IOTracer*/, nullptr /*BlobFileCompletionCallback*/,
       BlobFileCreationReason::kFlush, &blob_file_paths, &blob_file_additions);
 
@@ -528,7 +525,6 @@ TEST_F(BlobFileBuilderTest, Checksum) {
   constexpr int job_id = 1;
   constexpr uint32_t column_family_id = 123;
   constexpr char column_family_name[] = "foobar";
-  constexpr Env::IOPriority io_priority = Env::IO_HIGH;
   constexpr Env::WriteLifeTimeHint write_hint = Env::WLTH_MEDIUM;
 
   std::vector<std::string> blob_file_paths;
@@ -536,8 +532,8 @@ TEST_F(BlobFileBuilderTest, Checksum) {
 
   BlobFileBuilder builder(
       TestFileNumberGenerator(), fs_, &immutable_options, &mutable_cf_options,
-      &file_options_, "" /*db_id*/, "" /*db_session_id*/, job_id,
-      column_family_id, column_family_name, io_priority, write_hint,
+      &file_options_, &write_options_, "" /*db_id*/, "" /*db_session_id*/,
+      job_id, column_family_id, column_family_name, write_hint,
       nullptr /*IOTracer*/, nullptr /*BlobFileCompletionCallback*/,
       BlobFileCreationReason::kFlush, &blob_file_paths, &blob_file_additions);
 
@@ -589,11 +585,13 @@ class BlobFileBuilderIOErrorTest
   BlobFileBuilderIOErrorTest() : sync_point_(GetParam()) {
     mock_env_.reset(MockEnv::Create(Env::Default()));
     fs_ = mock_env_->GetFileSystem().get();
+    write_options_.rate_limiter_priority = Env::IO_HIGH;
   }
 
   std::unique_ptr<Env> mock_env_;
   FileSystem* fs_;
   FileOptions file_options_;
+  WriteOptions write_options_;
   std::string sync_point_;
 };
 
@@ -626,7 +624,6 @@ TEST_P(BlobFileBuilderIOErrorTest, IOError) {
   constexpr int job_id = 1;
   constexpr uint32_t column_family_id = 123;
   constexpr char column_family_name[] = "foobar";
-  constexpr Env::IOPriority io_priority = Env::IO_HIGH;
   constexpr Env::WriteLifeTimeHint write_hint = Env::WLTH_MEDIUM;
 
   std::vector<std::string> blob_file_paths;
@@ -634,8 +631,8 @@ TEST_P(BlobFileBuilderIOErrorTest, IOError) {
 
   BlobFileBuilder builder(
       TestFileNumberGenerator(), fs_, &immutable_options, &mutable_cf_options,
-      &file_options_, "" /*db_id*/, "" /*db_session_id*/, job_id,
-      column_family_id, column_family_name, io_priority, write_hint,
+      &file_options_, &write_options_, "" /*db_id*/, "" /*db_session_id*/,
+      job_id, column_family_id, column_family_name, write_hint,
       nullptr /*IOTracer*/, nullptr /*BlobFileCompletionCallback*/,
       BlobFileCreationReason::kFlush, &blob_file_paths, &blob_file_additions);
 

db/blob/blob_file_cache_test.cc

@@ -57,7 +57,7 @@ void WriteBlobFile(uint32_t column_family_id,
   BlobLogHeader header(column_family_id, kNoCompression, has_ttl,
                        expiration_range);
 
-  ASSERT_OK(blob_log_writer.WriteHeader(header));
+  ASSERT_OK(blob_log_writer.WriteHeader(WriteOptions(), header));
 
   constexpr char key[] = "key";
   constexpr char blob[] = "blob";
@@ -67,7 +67,8 @@ void WriteBlobFile(uint32_t column_family_id,
   uint64_t key_offset = 0;
   uint64_t blob_offset = 0;
 
-  ASSERT_OK(blob_log_writer.AddRecord(key, blob, &key_offset, &blob_offset));
+  ASSERT_OK(blob_log_writer.AddRecord(WriteOptions(), key, blob, &key_offset,
+                                      &blob_offset));
 
   BlobLogFooter footer;
   footer.blob_count = 1;
@@ -76,8 +77,8 @@ void WriteBlobFile(uint32_t column_family_id,
   std::string checksum_method;
   std::string checksum_value;
 
-  ASSERT_OK(
-      blob_log_writer.AppendFooter(footer, &checksum_method, &checksum_value));
+  ASSERT_OK(blob_log_writer.AppendFooter(WriteOptions(), footer,
+                                         &checksum_method, &checksum_value));
 }
 
 }  // anonymous namespace

db/blob/blob_file_reader_test.cc

@@ -63,7 +63,7 @@ void WriteBlobFile(const ImmutableOptions& immutable_options,
   BlobLogHeader header(column_family_id, compression, has_ttl,
                        expiration_range_header);
 
-  ASSERT_OK(blob_log_writer.WriteHeader(header));
+  ASSERT_OK(blob_log_writer.WriteHeader(WriteOptions(), header));
 
   std::vector<std::string> compressed_blobs(num);
   std::vector<Slice> blobs_to_write(num);
@@ -91,7 +91,8 @@ void WriteBlobFile(const ImmutableOptions& immutable_options,
 
   for (size_t i = 0; i < num; ++i) {
     uint64_t key_offset = 0;
-    ASSERT_OK(blob_log_writer.AddRecord(keys[i], blobs_to_write[i], &key_offset,
+    ASSERT_OK(blob_log_writer.AddRecord(WriteOptions(), keys[i],
+                                        blobs_to_write[i], &key_offset,
                                         &blob_offsets[i]));
   }
 
@@ -101,8 +102,8 @@ void WriteBlobFile(const ImmutableOptions& immutable_options,
 
   std::string checksum_method;
   std::string checksum_value;
-  ASSERT_OK(
-      blob_log_writer.AppendFooter(footer, &checksum_method, &checksum_value));
+  ASSERT_OK(blob_log_writer.AppendFooter(WriteOptions(), footer,
+                                         &checksum_method, &checksum_value));
 }
 
 // Creates a test blob file with a single blob in it. Note: this method
@@ -473,7 +474,7 @@ TEST_F(BlobFileReaderTest, Malformed) {
     BlobLogHeader header(column_family_id, kNoCompression, has_ttl,
                          expiration_range);
 
-    ASSERT_OK(blob_log_writer.WriteHeader(header));
+    ASSERT_OK(blob_log_writer.WriteHeader(WriteOptions(), header));
   }
 
   constexpr HistogramImpl* blob_file_read_hist = nullptr;

db/blob/blob_log_writer.cc

@@ -33,35 +33,49 @@ BlobLogWriter::BlobLogWriter(std::unique_ptr<WritableFileWriter>&& dest,
 
 BlobLogWriter::~BlobLogWriter() = default;
 
-Status BlobLogWriter::Sync() {
+Status BlobLogWriter::Sync(const WriteOptions& write_options) {
   TEST_SYNC_POINT("BlobLogWriter::Sync");
 
   StopWatch sync_sw(clock_, statistics_, BLOB_DB_BLOB_FILE_SYNC_MICROS);
-  Status s = dest_->Sync(use_fsync_);
-  RecordTick(statistics_, BLOB_DB_BLOB_FILE_SYNCED);
+  IOOptions opts;
+  Status s = WritableFileWriter::PrepareIOOptions(write_options, opts);
+  if (s.ok()) {
+    s = dest_->Sync(opts, use_fsync_);
+  }
+  if (s.ok()) {
+    RecordTick(statistics_, BLOB_DB_BLOB_FILE_SYNCED);
+  }
   return s;
 }
 
-Status BlobLogWriter::WriteHeader(BlobLogHeader& header) {
+Status BlobLogWriter::WriteHeader(const WriteOptions& write_options,
+                                  BlobLogHeader& header) {
   assert(block_offset_ == 0);
   assert(last_elem_type_ == kEtNone);
   std::string str;
   header.EncodeTo(&str);
 
-  Status s = dest_->Append(Slice(str));
+  IOOptions opts;
+  Status s = WritableFileWriter::PrepareIOOptions(write_options, opts);
+  if (s.ok()) {
+    s = dest_->Append(opts, Slice(str));
+  }
   if (s.ok()) {
     block_offset_ += str.size();
     if (do_flush_) {
-      s = dest_->Flush();
+      s = dest_->Flush(opts);
     }
   }
   last_elem_type_ = kEtFileHdr;
-  RecordTick(statistics_, BLOB_DB_BLOB_FILE_BYTES_WRITTEN,
-             BlobLogHeader::kSize);
+  if (s.ok()) {
+    RecordTick(statistics_, BLOB_DB_BLOB_FILE_BYTES_WRITTEN,
+               BlobLogHeader::kSize);
+  }
   return s;
 }
 
-Status BlobLogWriter::AppendFooter(BlobLogFooter& footer,
+Status BlobLogWriter::AppendFooter(const WriteOptions& write_options,
+                                   BlobLogFooter& footer,
                                    std::string* checksum_method,
                                    std::string* checksum_value) {
   assert(block_offset_ != 0);
@@ -75,14 +89,17 @@ Status BlobLogWriter::AppendFooter(BlobLogFooter& footer,
     s.PermitUncheckedError();
     return Status::IOError("Seen Error. Skip closing.");
   } else {
-    s = dest_->Append(Slice(str));
+    IOOptions opts;
+    s = WritableFileWriter::PrepareIOOptions(write_options, opts);
+    if (s.ok()) {
+      s = dest_->Append(opts, Slice(str));
+    }
     if (s.ok()) {
       block_offset_ += str.size();
-
-      s = Sync();
+      s = Sync(write_options);
 
       if (s.ok()) {
-        s = dest_->Close();
+        s = dest_->Close(opts);
 
         if (s.ok()) {
           assert(!!checksum_method == !!checksum_value);
@@ -111,12 +128,15 @@ Status BlobLogWriter::AppendFooter(BlobLogFooter& footer,
   }
 
   last_elem_type_ = kEtFileFooter;
-  RecordTick(statistics_, BLOB_DB_BLOB_FILE_BYTES_WRITTEN,
-             BlobLogFooter::kSize);
+  if (s.ok()) {
+    RecordTick(statistics_, BLOB_DB_BLOB_FILE_BYTES_WRITTEN,
+               BlobLogFooter::kSize);
+  }
   return s;
 }
 
-Status BlobLogWriter::AddRecord(const Slice& key, const Slice& val,
+Status BlobLogWriter::AddRecord(const WriteOptions& write_options,
+                                const Slice& key, const Slice& val,
                                 uint64_t expiration, uint64_t* key_offset,
                                 uint64_t* blob_offset) {
   assert(block_offset_ != 0);
@@ -125,11 +145,13 @@ Status BlobLogWriter::AddRecord(const Slice& key, const Slice& val,
   std::string buf;
   ConstructBlobHeader(&buf, key, val, expiration);
 
-  Status s = EmitPhysicalRecord(buf, key, val, key_offset, blob_offset);
+  Status s =
+      EmitPhysicalRecord(write_options, buf, key, val, key_offset, blob_offset);
   return s;
 }
 
-Status BlobLogWriter::AddRecord(const Slice& key, const Slice& val,
+Status BlobLogWriter::AddRecord(const WriteOptions& write_options,
+                                const Slice& key, const Slice& val,
                                 uint64_t* key_offset, uint64_t* blob_offset) {
   assert(block_offset_ != 0);
   assert(last_elem_type_ == kEtFileHdr || last_elem_type_ == kEtRecord);
@@ -137,7 +159,8 @@ Status BlobLogWriter::AddRecord(const Slice& key, const Slice& val,
   std::string buf;
   ConstructBlobHeader(&buf, key, val, 0);
 
-  Status s = EmitPhysicalRecord(buf, key, val, key_offset, blob_offset);
+  Status s =
+      EmitPhysicalRecord(write_options, buf, key, val, key_offset, blob_offset);
   return s;
 }
 
@@ -150,28 +173,34 @@ void BlobLogWriter::ConstructBlobHeader(std::string* buf, const Slice& key,
   record.EncodeHeaderTo(buf);
 }
 
-Status BlobLogWriter::EmitPhysicalRecord(const std::string& headerbuf,
+Status BlobLogWriter::EmitPhysicalRecord(const WriteOptions& write_options,
+                                         const std::string& headerbuf,
                                          const Slice& key, const Slice& val,
                                          uint64_t* key_offset,
                                          uint64_t* blob_offset) {
-  StopWatch write_sw(clock_, statistics_, BLOB_DB_BLOB_FILE_WRITE_MICROS);
-  Status s = dest_->Append(Slice(headerbuf));
+  IOOptions opts;
+  Status s = WritableFileWriter::PrepareIOOptions(write_options, opts);
   if (s.ok()) {
-    s = dest_->Append(key);
+    s = dest_->Append(opts, Slice(headerbuf));
   }
   if (s.ok()) {
-    s = dest_->Append(val);
+    s = dest_->Append(opts, key);
+  }
+  if (s.ok()) {
+    s = dest_->Append(opts, val);
   }
   if (do_flush_ && s.ok()) {
-    s = dest_->Flush();
+    s = dest_->Flush(opts);
   }
 
   *key_offset = block_offset_ + BlobLogRecord::kHeaderSize;
   *blob_offset = *key_offset + key.size();
   block_offset_ = *blob_offset + val.size();
   last_elem_type_ = kEtRecord;
-  RecordTick(statistics_, BLOB_DB_BLOB_FILE_BYTES_WRITTEN,
-             BlobLogRecord::kHeaderSize + key.size() + val.size());
+  if (s.ok()) {
+    RecordTick(statistics_, BLOB_DB_BLOB_FILE_BYTES_WRITTEN,
+               BlobLogRecord::kHeaderSize + key.size() + val.size());
+  }
   return s;
 }
 

db/blob/blob_log_writer.h

@@ -43,20 +43,24 @@ class BlobLogWriter {
   static void ConstructBlobHeader(std::string* buf, const Slice& key,
                                   const Slice& val, uint64_t expiration);
 
-  Status AddRecord(const Slice& key, const Slice& val, uint64_t* key_offset,
+  Status AddRecord(const WriteOptions& write_options, const Slice& key,
+                   const Slice& val, uint64_t* key_offset,
                    uint64_t* blob_offset);
 
-  Status AddRecord(const Slice& key, const Slice& val, uint64_t expiration,
-                   uint64_t* key_offset, uint64_t* blob_offset);
+  Status AddRecord(const WriteOptions& write_options, const Slice& key,
+                   const Slice& val, uint64_t expiration, uint64_t* key_offset,
+                   uint64_t* blob_offset);
 
-  Status EmitPhysicalRecord(const std::string& headerbuf, const Slice& key,
+  Status EmitPhysicalRecord(const WriteOptions& write_options,
+                            const std::string& headerbuf, const Slice& key,
                             const Slice& val, uint64_t* key_offset,
                             uint64_t* blob_offset);
 
-  Status AppendFooter(BlobLogFooter& footer, std::string* checksum_method,
+  Status AppendFooter(const WriteOptions& write_options, BlobLogFooter& footer,
+                      std::string* checksum_method,
                       std::string* checksum_value);
 
-  Status WriteHeader(BlobLogHeader& header);
+  Status WriteHeader(const WriteOptions& write_options, BlobLogHeader& header);
 
   WritableFileWriter* file() { return dest_.get(); }
 
@@ -64,7 +68,7 @@ class BlobLogWriter {
 
   uint64_t get_log_number() const { return log_number_; }
 
-  Status Sync();
+  Status Sync(const WriteOptions& write_options);
 
  private:
   std::unique_ptr<WritableFileWriter> dest_;

db/blob/blob_source_test.cc

@@ -65,7 +65,7 @@ void WriteBlobFile(const ImmutableOptions& immutable_options,
   BlobLogHeader header(column_family_id, compression, has_ttl,
                        expiration_range_header);
 
-  ASSERT_OK(blob_log_writer.WriteHeader(header));
+  ASSERT_OK(blob_log_writer.WriteHeader(WriteOptions(), header));
 
   std::vector<std::string> compressed_blobs(num);
   std::vector<Slice> blobs_to_write(num);
@@ -93,7 +93,8 @@ void WriteBlobFile(const ImmutableOptions& immutable_options,
 
   for (size_t i = 0; i < num; ++i) {
     uint64_t key_offset = 0;
-    ASSERT_OK(blob_log_writer.AddRecord(keys[i], blobs_to_write[i], &key_offset,
+    ASSERT_OK(blob_log_writer.AddRecord(WriteOptions(), keys[i],
+                                        blobs_to_write[i], &key_offset,
                                         &blob_offsets[i]));
   }
 
@@ -103,8 +104,8 @@ void WriteBlobFile(const ImmutableOptions& immutable_options,
 
   std::string checksum_method;
   std::string checksum_value;
-  ASSERT_OK(
-      blob_log_writer.AppendFooter(footer, &checksum_method, &checksum_value));
+  ASSERT_OK(blob_log_writer.AppendFooter(WriteOptions(), footer,
+                                         &checksum_method, &checksum_value));
 }
 
 }  // anonymous namespace
@@ -1220,7 +1221,7 @@ TEST_F(BlobSecondaryCacheTest, GetBlobsFromSecondaryCache) {
       auto sec_handle0 = secondary_cache->Lookup(
           key0, BlobSource::SharedCacheInterface::GetFullHelper(),
           /*context*/ nullptr, true,
-          /*advise_erase=*/true, kept_in_sec_cache);
+          /*advise_erase=*/true, /*stats=*/nullptr, kept_in_sec_cache);
       ASSERT_FALSE(kept_in_sec_cache);
       ASSERT_NE(sec_handle0, nullptr);
       ASSERT_TRUE(sec_handle0->IsReady());
@@ -1248,7 +1249,7 @@ TEST_F(BlobSecondaryCacheTest, GetBlobsFromSecondaryCache) {
       auto sec_handle1 = secondary_cache->Lookup(
           key1, BlobSource::SharedCacheInterface::GetFullHelper(),
           /*context*/ nullptr, true,
-          /*advise_erase=*/true, kept_in_sec_cache);
+          /*advise_erase=*/true, /*stats=*/nullptr, kept_in_sec_cache);
       ASSERT_FALSE(kept_in_sec_cache);
       ASSERT_EQ(sec_handle1, nullptr);
 

db/blob/prefetch_buffer_collection.cc

@@ -11,8 +11,10 @@ FilePrefetchBuffer* PrefetchBufferCollection::GetOrCreatePrefetchBuffer(
     uint64_t file_number) {
   auto& prefetch_buffer = prefetch_buffers_[file_number];
   if (!prefetch_buffer) {
-    prefetch_buffer.reset(
-        new FilePrefetchBuffer(readahead_size_, readahead_size_));
+    ReadaheadParams readahead_params;
+    readahead_params.initial_readahead_size = readahead_size_;
+    readahead_params.max_readahead_size = readahead_size_;
+    prefetch_buffer.reset(new FilePrefetchBuffer(readahead_params));
   }
 
   return prefetch_buffer.get();

db/builder.cc

@@ -32,9 +32,11 @@
 #include "options/options_helper.h"
 #include "rocksdb/db.h"
 #include "rocksdb/env.h"
+#include "rocksdb/file_system.h"
 #include "rocksdb/iterator.h"
 #include "rocksdb/options.h"
 #include "rocksdb/table.h"
+#include "seqno_to_time_mapping.h"
 #include "table/block_based/block_based_table_builder.h"
 #include "table/format.h"
 #include "table/internal_iterator.h"
@@ -57,8 +59,8 @@ TableBuilder* NewTableBuilder(const TableBuilderOptions& tboptions,
 Status BuildTable(
     const std::string& dbname, VersionSet* versions,
     const ImmutableDBOptions& db_options, const TableBuilderOptions& tboptions,
-    const FileOptions& file_options, const ReadOptions& read_options,
-    TableCache* table_cache, InternalIterator* iter,
+    const FileOptions& file_options, TableCache* table_cache,
+    InternalIterator* iter,
     std::vector<std::unique_ptr<FragmentedRangeTombstoneIterator>>
         range_del_iters,
     FileMetaData* meta, std::vector<BlobFileAddition>* blob_file_additions,
@@ -69,9 +71,8 @@ Status BuildTable(
     IOStatus* io_status, const std::shared_ptr<IOTracer>& io_tracer,
     BlobFileCreationReason blob_creation_reason,
     const SeqnoToTimeMapping& seqno_to_time_mapping, EventLogger* event_logger,
-    int job_id, const Env::IOPriority io_priority,
-    TableProperties* table_properties, Env::WriteLifeTimeHint write_hint,
-    const std::string* full_history_ts_low,
+    int job_id, TableProperties* table_properties,
+    Env::WriteLifeTimeHint write_hint, const std::string* full_history_ts_low,
     BlobFileCompletionCallback* blob_callback, Version* version,
     uint64_t* num_input_entries, uint64_t* memtable_payload_bytes,
     uint64_t* memtable_garbage_bytes) {
@@ -164,11 +165,11 @@ Status BuildTable(
 
       table_file_created = true;
       FileTypeSet tmp_set = ioptions.checksum_handoff_file_types;
-      file->SetIOPriority(io_priority);
+      file->SetIOPriority(tboptions.write_options.rate_limiter_priority);
       file->SetWriteLifeTimeHint(write_hint);
       file_writer.reset(new WritableFileWriter(
           std::move(file), fname, file_options, ioptions.clock, io_tracer,
-          ioptions.stats, ioptions.listeners,
+          ioptions.stats, Histograms::SST_WRITE_MICROS, ioptions.listeners,
           ioptions.file_checksum_gen_factory.get(),
           tmp_set.Contains(FileType::kTableFile), false));
 
@@ -188,10 +189,11 @@ Status BuildTable(
          blob_file_additions)
             ? new BlobFileBuilder(
                   versions, fs, &ioptions, &mutable_cf_options, &file_options,
-                  tboptions.db_id, tboptions.db_session_id, job_id,
-                  tboptions.column_family_id, tboptions.column_family_name,
-                  io_priority, write_hint, io_tracer, blob_callback,
-                  blob_creation_reason, &blob_file_paths, blob_file_additions)
+                  &(tboptions.write_options), tboptions.db_id,
+                  tboptions.db_session_id, job_id, tboptions.column_family_id,
+                  tboptions.column_family_name, write_hint, io_tracer,
+                  blob_callback, blob_creation_reason, &blob_file_paths,
+                  blob_file_additions)
             : nullptr);
 
     const std::atomic<bool> kManualCompactionCanceledFalse{false};
@@ -244,7 +246,11 @@ Status BuildTable(
       }
 
       // TODO(noetzli): Update stats after flush, too.
-      if (io_priority == Env::IO_HIGH &&
+      // TODO(hx235): Replace `rate_limiter_priority` with `io_activity` for
+      // flush IO in repair when we have an `Env::IOActivity` enum for it
+      if ((tboptions.write_options.io_activity == Env::IOActivity::kFlush ||
+           tboptions.write_options.io_activity == Env::IOActivity::kDBOpen ||
+           tboptions.write_options.rate_limiter_priority == Env::IO_HIGH) &&
           IOSTATS(bytes_written) >= kReportFlushIOStatsEvery) {
         ThreadStatusUtil::SetThreadOperationProperty(
             ThreadStatus::FLUSH_BYTES_WRITTEN, IOSTATS(bytes_written));
@@ -275,7 +281,7 @@ Status BuildTable(
             SizeApproximationOptions approx_opts;
             approx_opts.files_size_error_margin = 0.1;
             meta->compensated_range_deletion_size += versions->ApproximateSize(
-                approx_opts, read_options, version, kv.first.Encode(),
+                approx_opts, tboptions.read_options, version, kv.first.Encode(),
                 tombstone_end.Encode(), 0 /* start_level */, -1 /* end_level */,
                 TableReaderCaller::kFlush);
           }
@@ -294,12 +300,14 @@ Status BuildTable(
     if (!s.ok() || empty) {
       builder->Abandon();
     } else {
-      std::string seqno_to_time_mapping_str;
-      seqno_to_time_mapping.Encode(
-          seqno_to_time_mapping_str, meta->fd.smallest_seqno,
-          meta->fd.largest_seqno, meta->file_creation_time);
+      SeqnoToTimeMapping relevant_mapping;
+      relevant_mapping.CopyFromSeqnoRange(seqno_to_time_mapping,
+                                          meta->fd.smallest_seqno,
+                                          meta->fd.largest_seqno);
+      relevant_mapping.SetCapacity(kMaxSeqnoTimePairsPerSST);
+      relevant_mapping.Enforce(tboptions.file_creation_time);
       builder->SetSeqnoTimeTableProperties(
-          seqno_to_time_mapping_str,
+          relevant_mapping,
           ioptions.compaction_style == CompactionStyle::kCompactionStyleFIFO
               ? meta->file_creation_time
               : meta->oldest_ancester_time);
@@ -346,13 +354,16 @@ Status BuildTable(
 
     // Finish and check for file errors
     TEST_SYNC_POINT("BuildTable:BeforeSyncTable");
-    if (s.ok() && !empty) {
+    IOOptions opts;
+    *io_status =
+        WritableFileWriter::PrepareIOOptions(tboptions.write_options, opts);
+    if (s.ok() && io_status->ok() && !empty) {
       StopWatch sw(ioptions.clock, ioptions.stats, TABLE_SYNC_MICROS);
-      *io_status = file_writer->Sync(ioptions.use_fsync);
+      *io_status = file_writer->Sync(opts, ioptions.use_fsync);
     }
     TEST_SYNC_POINT("BuildTable:BeforeCloseTableFile");
     if (s.ok() && io_status->ok() && !empty) {
-      *io_status = file_writer->Close();
+      *io_status = file_writer->Close(opts);
     }
     if (s.ok() && io_status->ok() && !empty) {
       // Add the checksum information to file metadata.
@@ -396,9 +407,9 @@ Status BuildTable(
       // No matter whether use_direct_io_for_flush_and_compaction is true,
       // the goal is to cache it here for further user reads.
       std::unique_ptr<InternalIterator> it(table_cache->NewIterator(
-          read_options, file_options, tboptions.internal_comparator, *meta,
-          nullptr /* range_del_agg */, mutable_cf_options.prefix_extractor,
-          nullptr,
+          tboptions.read_options, file_options, tboptions.internal_comparator,
+          *meta, nullptr /* range_del_agg */,
+          mutable_cf_options.prefix_extractor, nullptr,
           (internal_stats == nullptr) ? nullptr
                                       : internal_stats->GetFileReadHist(0),
           TableReaderCaller::kFlush, /*arena=*/nullptr,
@@ -436,8 +447,13 @@ Status BuildTable(
     constexpr IODebugContext* dbg = nullptr;
 
     if (table_file_created) {
-      Status ignored = fs->DeleteFile(fname, IOOptions(), dbg);
-      ignored.PermitUncheckedError();
+      IOOptions opts;
+      Status prepare =
+          WritableFileWriter::PrepareIOOptions(tboptions.write_options, opts);
+      if (prepare.ok()) {
+        Status ignored = fs->DeleteFile(fname, opts, dbg);
+        ignored.PermitUncheckedError();
+      }
     }
 
     assert(blob_file_additions || blob_file_paths.empty());

db/builder.h

@@ -53,8 +53,8 @@ TableBuilder* NewTableBuilder(const TableBuilderOptions& tboptions,
 extern Status BuildTable(
     const std::string& dbname, VersionSet* versions,
     const ImmutableDBOptions& db_options, const TableBuilderOptions& tboptions,
-    const FileOptions& file_options, const ReadOptions& read_options,
-    TableCache* table_cache, InternalIterator* iter,
+    const FileOptions& file_options, TableCache* table_cache,
+    InternalIterator* iter,
     std::vector<std::unique_ptr<FragmentedRangeTombstoneIterator>>
         range_del_iters,
     FileMetaData* meta, std::vector<BlobFileAddition>* blob_file_additions,
@@ -66,7 +66,6 @@ extern Status BuildTable(
     BlobFileCreationReason blob_creation_reason,
     const SeqnoToTimeMapping& seqno_to_time_mapping,
     EventLogger* event_logger = nullptr, int job_id = 0,
-    const Env::IOPriority io_priority = Env::IO_HIGH,
     TableProperties* table_properties = nullptr,
     Env::WriteLifeTimeHint write_hint = Env::WLTH_NOT_SET,
     const std::string* full_history_ts_low = nullptr,

db/c.cc

@@ -45,6 +45,7 @@
 #include "rocksdb/utilities/transaction_db.h"
 #include "rocksdb/utilities/write_batch_with_index.h"
 #include "rocksdb/write_batch.h"
+#include "rocksdb/write_buffer_manager.h"
 #include "utilities/merge_operators.h"
 
 using ROCKSDB_NAMESPACE::BackupEngine;
@@ -125,6 +126,7 @@ using ROCKSDB_NAMESPACE::WALRecoveryMode;
 using ROCKSDB_NAMESPACE::WritableFile;
 using ROCKSDB_NAMESPACE::WriteBatch;
 using ROCKSDB_NAMESPACE::WriteBatchWithIndex;
+using ROCKSDB_NAMESPACE::WriteBufferManager;
 using ROCKSDB_NAMESPACE::WriteOptions;
 
 using std::unordered_set;
@@ -219,6 +221,9 @@ struct rocksdb_memory_allocator_t {
 struct rocksdb_cache_t {
   std::shared_ptr<Cache> rep;
 };
+struct rocksdb_write_buffer_manager_t {
+  std::shared_ptr<WriteBufferManager> rep;
+};
 struct rocksdb_livefiles_t {
   std::vector<LiveFileMetaData> rep;
 };
@@ -2904,6 +2909,16 @@ void rocksdb_options_set_db_paths(rocksdb_options_t* opt,
   opt->rep.db_paths = db_paths;
 }
 
+void rocksdb_options_set_cf_paths(rocksdb_options_t* opt,
+                                  const rocksdb_dbpath_t** dbpath_values,
+                                  size_t num_paths) {
+  std::vector<DbPath> cf_paths(num_paths);
+  for (size_t i = 0; i < num_paths; ++i) {
+    cf_paths[i] = dbpath_values[i]->rep;
+  }
+  opt->rep.cf_paths = cf_paths;
+}
+
 void rocksdb_options_set_env(rocksdb_options_t* opt, rocksdb_env_t* env) {
   opt->rep.env = (env ? env->rep : nullptr);
 }
@@ -2935,6 +2950,11 @@ void rocksdb_options_set_write_buffer_size(rocksdb_options_t* opt, size_t s) {
   opt->rep.write_buffer_size = s;
 }
 
+void rocksdb_options_set_write_buffer_manager(
+    rocksdb_options_t* opt, rocksdb_write_buffer_manager_t* wbm) {
+  opt->rep.write_buffer_manager = wbm->rep;
+}
+
 size_t rocksdb_options_get_write_buffer_size(rocksdb_options_t* opt) {
   return opt->rep.write_buffer_size;
 }
@@ -3029,6 +3049,24 @@ void rocksdb_options_set_max_bytes_for_level_multiplier_additional(
   }
 }
 
+void rocksdb_options_set_ttl(rocksdb_options_t* opt, uint64_t seconds) {
+  opt->rep.ttl = seconds;
+}
+
+uint64_t rocksdb_options_get_ttl(rocksdb_options_t* opt) {
+  return opt->rep.ttl;
+}
+
+void rocksdb_options_set_periodic_compaction_seconds(rocksdb_options_t* opt,
+                                                     uint64_t seconds) {
+  opt->rep.periodic_compaction_seconds = seconds;
+}
+
+uint64_t rocksdb_options_get_periodic_compaction_seconds(
+    rocksdb_options_t* opt) {
+  return opt->rep.periodic_compaction_seconds;
+}
+
 void rocksdb_options_enable_statistics(rocksdb_options_t* opt) {
   opt->rep.statistics = ROCKSDB_NAMESPACE::CreateDBStatistics();
 }
@@ -3957,6 +3995,16 @@ rocksdb_ratelimiter_t* rocksdb_ratelimiter_create(int64_t rate_bytes_per_sec,
   return rate_limiter;
 }
 
+rocksdb_ratelimiter_t* rocksdb_ratelimiter_create_auto_tuned(
+    int64_t rate_bytes_per_sec, int64_t refill_period_us, int32_t fairness) {
+  rocksdb_ratelimiter_t* rate_limiter = new rocksdb_ratelimiter_t;
+  rate_limiter->rep.reset(NewGenericRateLimiter(rate_bytes_per_sec,
+                                                refill_period_us, fairness,
+                                                RateLimiter::Mode::kWritesOnly,
+                                                true));  // auto_tuned
+  return rate_limiter;
+}
+
 void rocksdb_ratelimiter_destroy(rocksdb_ratelimiter_t* limiter) {
   delete limiter;
 }
@@ -4866,6 +4914,60 @@ size_t rocksdb_cache_get_occupancy_count(const rocksdb_cache_t* cache) {
   return cache->rep->GetOccupancyCount();
 }
 
+rocksdb_write_buffer_manager_t* rocksdb_write_buffer_manager_create(
+    size_t buffer_size, bool allow_stall) {
+  rocksdb_write_buffer_manager_t* wbm = new rocksdb_write_buffer_manager_t;
+  wbm->rep.reset(new WriteBufferManager(buffer_size, {}, allow_stall));
+  return wbm;
+}
+
+rocksdb_write_buffer_manager_t* rocksdb_write_buffer_manager_create_with_cache(
+    size_t buffer_size, const rocksdb_cache_t* cache, bool allow_stall) {
+  rocksdb_write_buffer_manager_t* wbm = new rocksdb_write_buffer_manager_t;
+  wbm->rep.reset(new WriteBufferManager(buffer_size, cache->rep, allow_stall));
+  return wbm;
+}
+
+void rocksdb_write_buffer_manager_destroy(rocksdb_write_buffer_manager_t* wbm) {
+  delete wbm;
+}
+
+bool rocksdb_write_buffer_manager_enabled(rocksdb_write_buffer_manager_t* wbm) {
+  return wbm->rep->enabled();
+}
+
+bool rocksdb_write_buffer_manager_cost_to_cache(
+    rocksdb_write_buffer_manager_t* wbm) {
+  return wbm->rep->cost_to_cache();
+}
+
+size_t rocksdb_write_buffer_manager_memory_usage(
+    rocksdb_write_buffer_manager_t* wbm) {
+  return wbm->rep->memory_usage();
+}
+
+size_t rocksdb_write_buffer_manager_mutable_memtable_memory_usage(
+    rocksdb_write_buffer_manager_t* wbm) {
+  return wbm->rep->mutable_memtable_memory_usage();
+}
+
+size_t rocksdb_write_buffer_manager_dummy_entries_in_cache_usage(
+    rocksdb_write_buffer_manager_t* wbm) {
+  return wbm->rep->dummy_entries_in_cache_usage();
+}
+size_t rocksdb_write_buffer_manager_buffer_size(
+    rocksdb_write_buffer_manager_t* wbm) {
+  return wbm->rep->buffer_size();
+}
+void rocksdb_write_buffer_manager_set_buffer_size(
+    rocksdb_write_buffer_manager_t* wbm, size_t new_size) {
+  wbm->rep->SetBufferSize(new_size);
+}
+ROCKSDB_LIBRARY_API void rocksdb_write_buffer_manager_set_allow_stall(
+    rocksdb_write_buffer_manager_t* wbm, bool new_allow_stall) {
+  wbm->rep->SetAllowStall(new_allow_stall);
+}
+
 rocksdb_dbpath_t* rocksdb_dbpath_create(const char* path,
                                         uint64_t target_size) {
   rocksdb_dbpath_t* result = new rocksdb_dbpath_t;
@@ -5464,6 +5566,11 @@ char* rocksdb_sst_file_metadata_get_relative_filename(
   return strdup(file_meta->rep->relative_filename.c_str());
 }
 
+char* rocksdb_sst_file_metadata_get_directory(
+    rocksdb_sst_file_metadata_t* file_meta) {
+  return strdup(file_meta->rep->directory.c_str());
+}
+
 uint64_t rocksdb_sst_file_metadata_get_size(
     rocksdb_sst_file_metadata_t* file_meta) {
   return file_meta->rep->size;

db/c_test.c

@@ -713,6 +713,11 @@ int main(int argc, char** argv) {
   rocksdb_options_set_ratelimiter(options, rate_limiter);
   rocksdb_ratelimiter_destroy(rate_limiter);
 
+  rate_limiter =
+      rocksdb_ratelimiter_create_auto_tuned(1000 * 1024 * 1024, 100 * 1000, 10);
+  rocksdb_options_set_ratelimiter(options, rate_limiter);
+  rocksdb_ratelimiter_destroy(rate_limiter);
+
   roptions = rocksdb_readoptions_create();
   rocksdb_readoptions_set_verify_checksums(roptions, 1);
   rocksdb_readoptions_set_fill_cache(roptions, 1);
@@ -1474,10 +1479,20 @@ int main(int argc, char** argv) {
     CheckCondition(cflen == 2);
     rocksdb_list_column_families_destroy(column_fams, cflen);
 
-    rocksdb_options_t* cf_options = rocksdb_options_create();
+    rocksdb_options_t* cf_options_1 = rocksdb_options_create();
+    rocksdb_options_t* cf_options_2 = rocksdb_options_create();
+
+    // use dbpathname2 as the cf_path for "cf1"
+    rocksdb_dbpath_t* dbpath2;
+    char dbpathname2[200];
+    snprintf(dbpathname2, sizeof(dbpathname2), "%s/rocksdb_c_test-%d-dbpath2",
+             GetTempDir(), ((int)geteuid()));
+    dbpath2 = rocksdb_dbpath_create(dbpathname2, 1024 * 1024);
+    const rocksdb_dbpath_t* cf_paths[1] = {dbpath2};
+    rocksdb_options_set_cf_paths(cf_options_2, cf_paths, 1);
 
     const char* cf_names[2] = {"default", "cf1"};
-    const rocksdb_options_t* cf_opts[2] = {cf_options, cf_options};
+    const rocksdb_options_t* cf_opts[2] = {cf_options_1, cf_options_2};
     rocksdb_column_family_handle_t* handles[2];
 
     LoadAndCheckLatestOptions(dbname, env, false, cache, NULL, 2, cf_names,
@@ -1505,6 +1520,37 @@ int main(int argc, char** argv) {
     rocksdb_flushoptions_t* flush_options = rocksdb_flushoptions_create();
     rocksdb_flushoptions_set_wait(flush_options, 1);
     rocksdb_flush_cf(db, flush_options, handles[1], &err);
+
+    // make sure all files in "cf1" are under the specified cf path
+    {
+      rocksdb_column_family_metadata_t* cf_meta =
+          rocksdb_get_column_family_metadata_cf(db, handles[1]);
+      size_t cf_file_count = rocksdb_column_family_metadata_get_size(cf_meta);
+      assert(cf_file_count > 0);
+      size_t level_count =
+          rocksdb_column_family_metadata_get_level_count(cf_meta);
+      assert(level_count > 0);
+      for (size_t l = 0; l < level_count; ++l) {
+        rocksdb_level_metadata_t* level_meta =
+            rocksdb_column_family_metadata_get_level_metadata(cf_meta, l);
+        assert(level_meta);
+
+        size_t file_count = rocksdb_level_metadata_get_file_count(level_meta);
+        for (size_t f = 0; f < file_count; ++f) {
+          rocksdb_sst_file_metadata_t* file_meta =
+              rocksdb_level_metadata_get_sst_file_metadata(level_meta, f);
+          assert(file_meta);
+          char* file_path = rocksdb_sst_file_metadata_get_directory(file_meta);
+          assert(strcmp(file_path, dbpathname2) == 0);
+          Free(&file_path);
+          rocksdb_sst_file_metadata_destroy(file_meta);
+        }
+        rocksdb_level_metadata_destroy(level_meta);
+      }
+
+      rocksdb_column_family_metadata_destroy(cf_meta);
+    }
+
     CheckNoError(err) rocksdb_flushoptions_destroy(flush_options);
 
     CheckGetCF(db, roptions, handles[1], "foo", "hello");
@@ -1668,7 +1714,9 @@ int main(int argc, char** argv) {
     }
     rocksdb_destroy_db(options, dbname, &err);
     rocksdb_options_destroy(db_options);
-    rocksdb_options_destroy(cf_options);
+    rocksdb_options_destroy(cf_options_1);
+    rocksdb_options_destroy(cf_options_2);
+    rocksdb_dbpath_destroy(dbpath2);
   }
 
   StartPhase("prefix");
@@ -1873,6 +1921,13 @@ int main(int argc, char** argv) {
     CheckCondition(2.0 ==
                    rocksdb_options_get_max_bytes_for_level_multiplier(o));
 
+    rocksdb_options_set_periodic_compaction_seconds(o, 100000);
+    CheckCondition(100000 ==
+                   rocksdb_options_get_periodic_compaction_seconds(o));
+
+    rocksdb_options_set_ttl(o, 5000);
+    CheckCondition(5000 == rocksdb_options_get_ttl(o));
+
     rocksdb_options_set_skip_stats_update_on_db_open(o, 1);
     CheckCondition(1 == rocksdb_options_get_skip_stats_update_on_db_open(o));
 
@@ -2303,6 +2358,16 @@ int main(int argc, char** argv) {
     CheckCondition(2.0 ==
                    rocksdb_options_get_max_bytes_for_level_multiplier(o));
 
+    rocksdb_options_set_periodic_compaction_seconds(copy, 8000);
+    CheckCondition(8000 ==
+                   rocksdb_options_get_periodic_compaction_seconds(copy));
+    CheckCondition(100000 ==
+                   rocksdb_options_get_periodic_compaction_seconds(o));
+
+    rocksdb_options_set_ttl(copy, 8000);
+    CheckCondition(8000 == rocksdb_options_get_ttl(copy));
+    CheckCondition(5000 == rocksdb_options_get_ttl(o));
+
     rocksdb_options_set_skip_stats_update_on_db_open(copy, 0);
     CheckCondition(0 == rocksdb_options_get_skip_stats_update_on_db_open(copy));
     CheckCondition(1 == rocksdb_options_get_skip_stats_update_on_db_open(o));
@@ -3723,6 +3788,30 @@ int main(int argc, char** argv) {
     rocksdb_wait_for_compact_options_destroy(wco);
   }
 
+  StartPhase("write_buffer_manager");
+  {
+    rocksdb_cache_t* lru;
+    lru = rocksdb_cache_create_lru(100);
+
+    rocksdb_write_buffer_manager_t* write_buffer_manager;
+    write_buffer_manager =
+        rocksdb_write_buffer_manager_create_with_cache(200, lru, false);
+
+    CheckCondition(true ==
+                   rocksdb_write_buffer_manager_enabled(write_buffer_manager));
+    CheckCondition(true == rocksdb_write_buffer_manager_cost_to_cache(
+                               write_buffer_manager));
+    CheckCondition(
+        200 == rocksdb_write_buffer_manager_buffer_size(write_buffer_manager));
+
+    rocksdb_write_buffer_manager_set_buffer_size(write_buffer_manager, 300);
+    CheckCondition(
+        300 == rocksdb_write_buffer_manager_buffer_size(write_buffer_manager));
+
+    rocksdb_write_buffer_manager_destroy(write_buffer_manager);
+    rocksdb_cache_destroy(lru);
+  }
+
   StartPhase("cancel_all_background_work");
   rocksdb_cancel_all_background_work(db, 1);
 

db/column_family.cc

@@ -322,8 +322,8 @@ ColumnFamilyOptions SanitizeOptions(const ImmutableDBOptions& db_options,
     }
     ROCKS_LOG_WARN(db_options.logger,
                    "Adjust the value to "
-                   "level0_stop_writes_trigger(%d)"
-                   "level0_slowdown_writes_trigger(%d)"
+                   "level0_stop_writes_trigger(%d) "
+                   "level0_slowdown_writes_trigger(%d) "
                    "level0_file_num_compaction_trigger(%d)",
                    result.level0_stop_writes_trigger,
                    result.level0_slowdown_writes_trigger,
@@ -358,16 +358,16 @@ ColumnFamilyOptions SanitizeOptions(const ImmutableDBOptions& db_options,
 
   if (result.level_compaction_dynamic_level_bytes) {
     if (result.compaction_style != kCompactionStyleLevel) {
-      ROCKS_LOG_WARN(db_options.info_log.get(),
-                     "level_compaction_dynamic_level_bytes only makes sense"
+      ROCKS_LOG_INFO(db_options.info_log.get(),
+                     "level_compaction_dynamic_level_bytes only makes sense "
                      "for level-based compaction");
       result.level_compaction_dynamic_level_bytes = false;
     } else if (result.cf_paths.size() > 1U) {
       // we don't yet know how to make both of this feature and multiple
       // DB path work.
       ROCKS_LOG_WARN(db_options.info_log.get(),
-                     "multiple cf_paths/db_paths and"
-                     "level_compaction_dynamic_level_bytes"
+                     "multiple cf_paths/db_paths and "
+                     "level_compaction_dynamic_level_bytes "
                      "can't be used together");
       result.level_compaction_dynamic_level_bytes = false;
     }
@@ -833,8 +833,8 @@ std::unique_ptr<WriteControllerToken> SetupDelay(
   return write_controller->GetDelayToken(write_rate);
 }
 
-int GetL0ThresholdSpeedupCompaction(int level0_file_num_compaction_trigger,
-                                    int level0_slowdown_writes_trigger) {
+int GetL0FileCountForCompactionSpeedup(int level0_file_num_compaction_trigger,
+                                       int level0_slowdown_writes_trigger) {
   // SanitizeOptions() ensures it.
   assert(level0_file_num_compaction_trigger <= level0_slowdown_writes_trigger);
 
@@ -864,6 +864,35 @@ int GetL0ThresholdSpeedupCompaction(int level0_file_num_compaction_trigger,
     return static_cast<int>(res);
   }
 }
+
+uint64_t GetPendingCompactionBytesForCompactionSpeedup(
+    const MutableCFOptions& mutable_cf_options,
+    const VersionStorageInfo* vstorage) {
+  // Compaction debt relatively large compared to the stable (bottommost) data
+  // size indicates compaction fell behind.
+  const uint64_t kBottommostSizeDivisor = 8;
+  // Meaningful progress toward the slowdown trigger is another good indication.
+  const uint64_t kSlowdownTriggerDivisor = 4;
+
+  uint64_t bottommost_files_size = 0;
+  for (const auto& level_and_file : vstorage->BottommostFiles()) {
+    bottommost_files_size += level_and_file.second->fd.GetFileSize();
+  }
+
+  // Slowdown trigger might be zero but that means compaction speedup should
+  // always happen (undocumented/historical), so no special treatment is needed.
+  uint64_t slowdown_threshold =
+      mutable_cf_options.soft_pending_compaction_bytes_limit /
+      kSlowdownTriggerDivisor;
+
+  // Size of zero, however, should not be used to decide to speedup compaction.
+  if (bottommost_files_size == 0) {
+    return slowdown_threshold;
+  }
+
+  uint64_t size_threshold = bottommost_files_size / kBottommostSizeDivisor;
+  return std::min(size_threshold, slowdown_threshold);
+}
 }  // anonymous namespace
 
 std::pair<WriteStallCondition, WriteStallCause>
@@ -1019,7 +1048,7 @@ WriteStallCondition ColumnFamilyData::RecalculateWriteStallConditions(
     } else {
       assert(write_stall_condition == WriteStallCondition::kNormal);
       if (vstorage->l0_delay_trigger_count() >=
-          GetL0ThresholdSpeedupCompaction(
+          GetL0FileCountForCompactionSpeedup(
               mutable_cf_options.level0_file_num_compaction_trigger,
               mutable_cf_options.level0_slowdown_writes_trigger)) {
         write_controller_token_ =
@@ -1029,22 +1058,22 @@ WriteStallCondition ColumnFamilyData::RecalculateWriteStallConditions(
             "[%s] Increasing compaction threads because we have %d level-0 "
             "files ",
             name_.c_str(), vstorage->l0_delay_trigger_count());
-      } else if (vstorage->estimated_compaction_needed_bytes() >=
-                 mutable_cf_options.soft_pending_compaction_bytes_limit / 4) {
-        // Increase compaction threads if bytes needed for compaction exceeds
-        // 1/4 of threshold for slowing down.
+      } else if (mutable_cf_options.soft_pending_compaction_bytes_limit == 0) {
         // If soft pending compaction byte limit is not set, always speed up
         // compaction.
         write_controller_token_ =
             write_controller->GetCompactionPressureToken();
-        if (mutable_cf_options.soft_pending_compaction_bytes_limit > 0) {
-          ROCKS_LOG_INFO(
-              ioptions_.logger,
-              "[%s] Increasing compaction threads because of estimated pending "
-              "compaction "
-              "bytes %" PRIu64,
-              name_.c_str(), vstorage->estimated_compaction_needed_bytes());
-        }
+      } else if (vstorage->estimated_compaction_needed_bytes() >=
+                 GetPendingCompactionBytesForCompactionSpeedup(
+                     mutable_cf_options, vstorage)) {
+        write_controller_token_ =
+            write_controller->GetCompactionPressureToken();
+        ROCKS_LOG_INFO(
+            ioptions_.logger,
+            "[%s] Increasing compaction threads because of estimated pending "
+            "compaction "
+            "bytes %" PRIu64,
+            name_.c_str(), vstorage->estimated_compaction_needed_bytes());
       } else {
         write_controller_token_.reset();
       }
@@ -1138,7 +1167,7 @@ Status ColumnFamilyData::RangesOverlapWithMemtables(
   *overlap = false;
   // Create an InternalIterator over all unflushed memtables
   Arena arena;
-  // TODO: plumb Env::IOActivity
+  // TODO: plumb Env::IOActivity, Env::IOPriority
   ReadOptions read_opts;
   read_opts.total_order_seek = true;
   MergeIteratorBuilder merge_iter_builder(&internal_comparator_, &arena);

db/column_family_test.cc

@@ -2971,6 +2971,43 @@ TEST_P(ColumnFamilyTest, CompactionSpeedupTwoColumnFamilies) {
   ASSERT_EQ(1, dbfull()->TEST_BGCompactionsAllowed());
 }
 
+TEST_P(ColumnFamilyTest, CompactionSpeedupForCompactionDebt) {
+  db_options_.max_background_compactions = 6;
+  Open();
+  ColumnFamilyData* cfd =
+      static_cast<ColumnFamilyHandleImpl*>(db_->DefaultColumnFamily())->cfd();
+  MutableCFOptions mutable_cf_options(column_family_options_);
+  mutable_cf_options.soft_pending_compaction_bytes_limit =
+      std::numeric_limits<uint64_t>::max();
+
+  {
+    // No bottommost data, so debt ratio cannot trigger speedup.
+    VersionStorageInfo* vstorage = cfd->current()->storage_info();
+    vstorage->TEST_set_estimated_compaction_needed_bytes(1048576 /* 1MB */);
+    RecalculateWriteStallConditions(cfd, mutable_cf_options);
+    ASSERT_EQ(1, dbfull()->TEST_BGCompactionsAllowed());
+  }
+
+  // Add a tiny amount of bottommost data.
+  ASSERT_OK(db_->Put(WriteOptions(), "foo", "bar"));
+  ASSERT_OK(db_->Flush(FlushOptions()));
+
+  {
+    // 1MB debt is way bigger than bottommost data so definitely triggers
+    // speedup.
+    VersionStorageInfo* vstorage = cfd->current()->storage_info();
+    vstorage->TEST_set_estimated_compaction_needed_bytes(1048576 /* 1MB */);
+    RecalculateWriteStallConditions(cfd, mutable_cf_options);
+    ASSERT_EQ(6, dbfull()->TEST_BGCompactionsAllowed());
+
+    // Eight bytes is way smaller than bottommost data so definitely does not
+    // trigger speedup.
+    vstorage->TEST_set_estimated_compaction_needed_bytes(8);
+    RecalculateWriteStallConditions(cfd, mutable_cf_options);
+    ASSERT_EQ(1, dbfull()->TEST_BGCompactionsAllowed());
+  }
+}
+
 TEST_P(ColumnFamilyTest, CreateAndDestroyOptions) {
   std::unique_ptr<ColumnFamilyOptions> cfo(new ColumnFamilyOptions());
   ColumnFamilyHandle* cfh;

db/compaction/compaction.cc

@@ -115,6 +115,42 @@ void Compaction::GetBoundaryKeys(
   }
 }
 
+void Compaction::GetBoundaryInternalKeys(
+    VersionStorageInfo* vstorage,
+    const std::vector<CompactionInputFiles>& inputs, InternalKey* smallest_key,
+    InternalKey* largest_key, int exclude_level) {
+  bool initialized = false;
+  const InternalKeyComparator* icmp = vstorage->InternalComparator();
+  for (size_t i = 0; i < inputs.size(); ++i) {
+    if (inputs[i].files.empty() || inputs[i].level == exclude_level) {
+      continue;
+    }
+    if (inputs[i].level == 0) {
+      // we need to consider all files on level 0
+      for (const auto* f : inputs[i].files) {
+        if (!initialized || icmp->Compare(f->smallest, *smallest_key) < 0) {
+          *smallest_key = f->smallest;
+        }
+        if (!initialized || icmp->Compare(f->largest, *largest_key) > 0) {
+          *largest_key = f->largest;
+        }
+        initialized = true;
+      }
+    } else {
+      // we only need to consider the first and last file
+      if (!initialized ||
+          icmp->Compare(inputs[i].files[0]->smallest, *smallest_key) < 0) {
+        *smallest_key = inputs[i].files[0]->smallest;
+      }
+      if (!initialized ||
+          icmp->Compare(inputs[i].files.back()->largest, *largest_key) > 0) {
+        *largest_key = inputs[i].files.back()->largest;
+      }
+      initialized = true;
+    }
+  }
+}
+
 std::vector<CompactionInputFiles> Compaction::PopulateWithAtomicBoundaries(
     VersionStorageInfo* vstorage, std::vector<CompactionInputFiles> inputs) {
   const Comparator* ucmp = vstorage->InternalComparator()->user_comparator();
@@ -399,9 +435,14 @@ void Compaction::PopulatePenultimateLevelOutputRange() {
     }
   }
 
-  GetBoundaryKeys(input_vstorage_, inputs_,
-                  &penultimate_level_smallest_user_key_,
-                  &penultimate_level_largest_user_key_, exclude_level);
+  // FIXME: should make use of `penultimate_output_range_type_`.
+  // FIXME: when last level's input range does not overlap with
+  //  penultimate level, and penultimate level input is empty,
+  //  this call will not set penultimate_level_smallest_ or
+  //  penultimate_level_largest_. No keys will be compacted up.
+  GetBoundaryInternalKeys(input_vstorage_, inputs_,
+                          &penultimate_level_smallest_,
+                          &penultimate_level_largest_, exclude_level);
 }
 
 Compaction::~Compaction() {
@@ -426,33 +467,40 @@ bool Compaction::OverlapPenultimateLevelOutputRange(
   if (!SupportsPerKeyPlacement()) {
     return false;
   }
+
+  // See FIXME in Compaction::PopulatePenultimateLevelOutputRange().
+  // We do not compact any key up in this case.
+  if (penultimate_level_smallest_.size() == 0 ||
+      penultimate_level_largest_.size() == 0) {
+    return false;
+  }
+
   const Comparator* ucmp =
       input_vstorage_->InternalComparator()->user_comparator();
 
   return ucmp->CompareWithoutTimestamp(
-             smallest_key, penultimate_level_largest_user_key_) <= 0 &&
+             smallest_key, penultimate_level_largest_.user_key()) <= 0 &&
          ucmp->CompareWithoutTimestamp(
-             largest_key, penultimate_level_smallest_user_key_) >= 0;
+             largest_key, penultimate_level_smallest_.user_key()) >= 0;
 }
 
 // key includes timestamp if user-defined timestamp is enabled.
-bool Compaction::WithinPenultimateLevelOutputRange(const Slice& key) const {
+bool Compaction::WithinPenultimateLevelOutputRange(
+    const ParsedInternalKey& ikey) const {
   if (!SupportsPerKeyPlacement()) {
     return false;
   }
 
-  if (penultimate_level_smallest_user_key_.empty() ||
-      penultimate_level_largest_user_key_.empty()) {
+  if (penultimate_level_smallest_.size() == 0 ||
+      penultimate_level_largest_.size() == 0) {
     return false;
   }
 
-  const Comparator* ucmp =
-      input_vstorage_->InternalComparator()->user_comparator();
+  const InternalKeyComparator* icmp = input_vstorage_->InternalComparator();
 
-  return ucmp->CompareWithoutTimestamp(
-             key, penultimate_level_smallest_user_key_) >= 0 &&
-         ucmp->CompareWithoutTimestamp(
-             key, penultimate_level_largest_user_key_) <= 0;
+  // op_type of a key can change during compaction, e.g. Merge -> Put.
+  return icmp->CompareKeySeq(ikey, penultimate_level_smallest_.Encode()) >= 0 &&
+         icmp->CompareKeySeq(ikey, penultimate_level_largest_.Encode()) <= 0;
 }
 
 bool Compaction::InputCompressionMatchesOutput() const {

db/compaction/compaction.h

@@ -353,14 +353,6 @@ class Compaction {
 
   Slice GetLargestUserKey() const { return largest_user_key_; }
 
-  Slice GetPenultimateLevelSmallestUserKey() const {
-    return penultimate_level_smallest_user_key_;
-  }
-
-  Slice GetPenultimateLevelLargestUserKey() const {
-    return penultimate_level_largest_user_key_;
-  }
-
   PenultimateOutputRangeType GetPenultimateOutputRangeType() const {
     return penultimate_output_range_type_;
   }
@@ -383,10 +375,8 @@ class Compaction {
   // per_key_placement feature, which is safe to place the key to the
   // penultimate level. different compaction strategy has different rules.
   // If per_key_placement is not supported, always return false.
-  // TODO: currently it doesn't support moving data from the last level to the
-  //  penultimate level
   //  key includes timestamp if user-defined timestamp is enabled.
-  bool WithinPenultimateLevelOutputRange(const Slice& key) const;
+  bool WithinPenultimateLevelOutputRange(const ParsedInternalKey& ikey) const;
 
   CompactionReason compaction_reason() const { return compaction_reason_; }
 
@@ -456,6 +446,13 @@ class Compaction {
                               Slice* smallest_key, Slice* largest_key,
                               int exclude_level = -1);
 
+  // get the smallest and largest internal key present in files to be compacted
+  static void GetBoundaryInternalKeys(
+      VersionStorageInfo* vstorage,
+      const std::vector<CompactionInputFiles>& inputs,
+      InternalKey* smallest_key, InternalKey* largest_key,
+      int exclude_level = -1);
+
   // populate penultimate level output range, which will be used to determine if
   // a key is safe to output to the penultimate level (details see
   // `Compaction::WithinPenultimateLevelOutputRange()`.
@@ -568,8 +565,8 @@ class Compaction {
   // Key range for penultimate level output
   // includes timestamp if user-defined timestamp is enabled.
   // penultimate_output_range_type_ shows the range type
-  Slice penultimate_level_smallest_user_key_;
-  Slice penultimate_level_largest_user_key_;
+  InternalKey penultimate_level_smallest_;
+  InternalKey penultimate_level_largest_;
   PenultimateOutputRangeType penultimate_output_range_type_ =
       PenultimateOutputRangeType::kNotSupported;
 };

db/compaction/compaction_iterator.cc

@@ -1227,7 +1227,7 @@ void CompactionIterator::DecideOutputLevel() {
     // not from this compaction.
     // TODO: add statistic for declined output_to_penultimate_level
     bool safe_to_penultimate_level =
-        compaction_->WithinPenultimateLevelOutputRange(ikey_.user_key);
+        compaction_->WithinPenultimateLevelOutputRange(ikey_);
     if (!safe_to_penultimate_level) {
       output_to_penultimate_level_ = false;
       // It could happen when disable/enable `last_level_temperature` while
@@ -1256,10 +1256,13 @@ void CompactionIterator::PrepareOutput() {
       } else if (ikey_.type == kTypeBlobIndex) {
         GarbageCollectBlobIfNeeded();
       }
-    }
 
-    if (compaction_ != nullptr && compaction_->SupportsPerKeyPlacement()) {
-      DecideOutputLevel();
+      // For range del sentinel, we don't use it to cut files for bottommost
+      // compaction. So it should not make a difference which output level we
+      // decide.
+      if (compaction_ != nullptr && compaction_->SupportsPerKeyPlacement()) {
+        DecideOutputLevel();
+      }
     }
 
     // Zeroing out the sequence number leads to better compression.

db/compaction/compaction_iterator.h

@@ -119,7 +119,8 @@ class CompactionIterator {
     virtual bool SupportsPerKeyPlacement() const = 0;
 
     // `key` includes timestamp if user-defined timestamp is enabled.
-    virtual bool WithinPenultimateLevelOutputRange(const Slice& key) const = 0;
+    virtual bool WithinPenultimateLevelOutputRange(
+        const ParsedInternalKey&) const = 0;
   };
 
   class RealCompaction : public CompactionProxy {
@@ -186,8 +187,9 @@ class CompactionIterator {
     // Check if key is within penultimate level output range, to see if it's
     // safe to output to the penultimate level for per_key_placement feature.
     // `key` includes timestamp if user-defined timestamp is enabled.
-    bool WithinPenultimateLevelOutputRange(const Slice& key) const override {
-      return compaction_->WithinPenultimateLevelOutputRange(key);
+    bool WithinPenultimateLevelOutputRange(
+        const ParsedInternalKey& ikey) const override {
+      return compaction_->WithinPenultimateLevelOutputRange(ikey);
     }
 
    private:

db/compaction/compaction_iterator_test.cc

@@ -184,8 +184,9 @@ class FakeCompaction : public CompactionIterator::CompactionProxy {
     return supports_per_key_placement;
   }
 
-  bool WithinPenultimateLevelOutputRange(const Slice& key) const override {
-    return (!key.starts_with("unsafe_pb"));
+  bool WithinPenultimateLevelOutputRange(
+      const ParsedInternalKey& key) const override {
+    return (!key.user_key.starts_with("unsafe_pb"));
   }
 
   bool key_not_exists_beyond_output_level = false;

db/compaction/compaction_job.cc

@@ -288,39 +288,37 @@ void CompactionJob::Prepare() {
 
   if (preserve_time_duration > 0) {
     const ReadOptions read_options(Env::IOActivity::kCompaction);
-    // setup seqno_to_time_mapping_
-    seqno_to_time_mapping_.SetMaxTimeDuration(preserve_time_duration);
+    // Setup seqno_to_time_mapping_ with relevant time range.
+    seqno_to_time_mapping_.SetMaxTimeSpan(preserve_time_duration);
     for (const auto& each_level : *c->inputs()) {
       for (const auto& fmd : each_level.files) {
         std::shared_ptr<const TableProperties> tp;
         Status s =
             cfd->current()->GetTableProperties(read_options, &tp, fmd, nullptr);
         if (s.ok()) {
-          seqno_to_time_mapping_.Add(tp->seqno_to_time_mapping)
-              .PermitUncheckedError();
-          seqno_to_time_mapping_.Add(fmd->fd.smallest_seqno,
-                                     fmd->oldest_ancester_time);
+          s = seqno_to_time_mapping_.DecodeFrom(tp->seqno_to_time_mapping);
+        }
+        if (!s.ok()) {
+          ROCKS_LOG_WARN(
+              db_options_.info_log,
+              "Problem reading or processing seqno-to-time mapping: %s",
+              s.ToString().c_str());
         }
       }
     }
 
-    auto status = seqno_to_time_mapping_.Sort();
-    if (!status.ok()) {
-      ROCKS_LOG_WARN(db_options_.info_log,
-                     "Invalid sequence number to time mapping: Status: %s",
-                     status.ToString().c_str());
-    }
     int64_t _current_time = 0;
-    status = db_options_.clock->GetCurrentTime(&_current_time);
-    if (!status.ok()) {
+    Status s = db_options_.clock->GetCurrentTime(&_current_time);
+    if (!s.ok()) {
       ROCKS_LOG_WARN(db_options_.info_log,
                      "Failed to get current time in compaction: Status: %s",
-                     status.ToString().c_str());
+                     s.ToString().c_str());
       // preserve all time information
       preserve_time_min_seqno_ = 0;
       preclude_last_level_min_seqno_ = 0;
+      seqno_to_time_mapping_.Enforce();
     } else {
-      seqno_to_time_mapping_.TruncateOldEntries(_current_time);
+      seqno_to_time_mapping_.Enforce(_current_time);
       uint64_t preserve_time =
           static_cast<uint64_t>(_current_time) > preserve_time_duration
               ? _current_time - preserve_time_duration
@@ -344,6 +342,16 @@ void CompactionJob::Prepare() {
             1;
       }
     }
+    // For accuracy of the GetProximalSeqnoBeforeTime queries above, we only
+    // limit the capacity after them.
+    // Here If we set capacity to the per-SST limit, we could be throwing away
+    // fidelity when a compaction output file has a narrower seqno range than
+    // all the inputs. If we only limit capacity for each compaction output, we
+    // could be doing a lot of unnecessary recomputation in a large compaction
+    // (up to quadratic in number of files). Thus, we do soemthing in the
+    // middle: enforce a resonably large constant size limit substantially
+    // larger than kMaxSeqnoTimePairsPerSST.
+    seqno_to_time_mapping_.SetCapacity(kMaxSeqnoToTimeEntries);
   }
 }
 
@@ -476,7 +484,8 @@ void CompactionJob::GenSubcompactionBoundaries() {
   // overlap with N-1 other ranges. Since we requested a relatively large number
   // (128) of ranges from each input files, even N range overlapping would
   // cause relatively small inaccuracy.
-  const ReadOptions read_options(Env::IOActivity::kCompaction);
+  ReadOptions read_options(Env::IOActivity::kCompaction);
+  read_options.rate_limiter_priority = GetRateLimiterPriority();
   auto* c = compact_->compaction;
   if (c->max_subcompactions() <= 1 &&
       !(c->immutable_options()->compaction_pri == kRoundRobin &&
@@ -728,8 +737,9 @@ Status CompactionJob::Run() {
         // use_direct_io_for_flush_and_compaction is true, we will regard this
         // verification as user reads since the goal is to cache it here for
         // further user reads
-        const ReadOptions verify_table_read_options(
-            Env::IOActivity::kCompaction);
+        ReadOptions verify_table_read_options(Env::IOActivity::kCompaction);
+        verify_table_read_options.rate_limiter_priority =
+            GetRateLimiterPriority();
         InternalIterator* iter = cfd->table_cache()->NewIterator(
             verify_table_read_options, file_options_,
             cfd->internal_comparator(), files_output[file_idx]->meta,
@@ -1130,6 +1140,9 @@ void CompactionJob::ProcessKeyValueCompaction(SubcompactionState* sub_compact) {
   // (b) CompactionFilter::Decision::kRemoveAndSkipUntil.
   read_options.total_order_seek = true;
 
+  const WriteOptions write_options(Env::IOPriority::IO_LOW,
+                                   Env::IOActivity::kCompaction);
+
   // Remove the timestamps from boundaries because boundaries created in
   // GenSubcompactionBoundaries doesn't strip away the timestamp.
   size_t ts_sz = cfd->user_comparator()->timestamp_size();
@@ -1264,8 +1277,8 @@ void CompactionJob::ProcessKeyValueCompaction(SubcompactionState* sub_compact) {
           ? new BlobFileBuilder(
                 versions_, fs_.get(),
                 sub_compact->compaction->immutable_options(),
-                mutable_cf_options, &file_options_, db_id_, db_session_id_,
-                job_id_, cfd->GetID(), cfd->GetName(), Env::IOPriority::IO_LOW,
+                mutable_cf_options, &file_options_, &write_options, db_id_,
+                db_session_id_, job_id_, cfd->GetID(), cfd->GetName(),
                 write_hint_, io_tracer_, blob_callback_,
                 BlobFileCreationReason::kCompaction, &blob_file_paths,
                 sub_compact->Current().GetBlobFileAdditionsPtr())
@@ -1710,6 +1723,8 @@ Status CompactionJob::InstallCompactionResults(
   db_mutex_->AssertHeld();
 
   const ReadOptions read_options(Env::IOActivity::kCompaction);
+  const WriteOptions write_options(Env::IOActivity::kCompaction);
+
   auto* compaction = compact_->compaction;
   assert(compaction);
 
@@ -1792,8 +1807,9 @@ Status CompactionJob::InstallCompactionResults(
   };
 
   return versions_->LogAndApply(
-      compaction->column_family_data(), mutable_cf_options, read_options, edit,
-      db_mutex_, db_directory_, /*new_descriptor_log=*/false,
+      compaction->column_family_data(), mutable_cf_options, read_options,
+      write_options, edit, db_mutex_, db_directory_,
+      /*new_descriptor_log=*/false,
       /*column_family_options=*/nullptr, manifest_wcb);
 }
 
@@ -1900,6 +1916,8 @@ Status CompactionJob::OpenCompactionOutputFile(SubcompactionState* sub_compact,
           sub_compact->start.has_value() ? &tmp_start : nullptr,
           sub_compact->end.has_value() ? &tmp_end : nullptr);
   if (oldest_ancester_time == std::numeric_limits<uint64_t>::max()) {
+    // TODO: fix DBSSTTest.GetTotalSstFilesSize and use
+    //  kUnknownOldestAncesterTime
     oldest_ancester_time = current_time;
   }
 
@@ -1941,13 +1959,17 @@ Status CompactionJob::OpenCompactionOutputFile(SubcompactionState* sub_compact,
       sub_compact->compaction->immutable_options()->listeners;
   outputs.AssignFileWriter(new WritableFileWriter(
       std::move(writable_file), fname, fo_copy, db_options_.clock, io_tracer_,
-      db_options_.stats, listeners, db_options_.file_checksum_gen_factory.get(),
+      db_options_.stats, Histograms::SST_WRITE_MICROS, listeners,
+      db_options_.file_checksum_gen_factory.get(),
       tmp_set.Contains(FileType::kTableFile), false));
 
   // TODO(hx235): pass in the correct `oldest_key_time` instead of `0`
+  const ReadOptions read_options(Env::IOActivity::kCompaction);
+  const WriteOptions write_options(Env::IOActivity::kCompaction);
   TableBuilderOptions tboptions(
       *cfd->ioptions(), *(sub_compact->compaction->mutable_cf_options()),
-      cfd->internal_comparator(), cfd->int_tbl_prop_collector_factories(),
+      read_options, write_options, cfd->internal_comparator(),
+      cfd->int_tbl_prop_collector_factories(),
       sub_compact->compaction->output_compression(),
       sub_compact->compaction->output_compression_opts(), cfd->GetID(),
       cfd->GetName(), sub_compact->compaction->output_level(),

db/compaction/compaction_job_test.cc

@@ -215,7 +215,9 @@ class CompactionJobTestBase : public testing::Test {
             dbname_, &db_options_, env_options_, table_cache_.get(),
             &write_buffer_manager_, &write_controller_,
             /*block_cache_tracer=*/nullptr,
-            /*io_tracer=*/nullptr, /*db_id*/ "", /*db_session_id*/ "")),
+            /*io_tracer=*/nullptr, /*db_id=*/"", /*db_session_id=*/"",
+            /*daily_offpeak_time_utc=*/"",
+            /*error_handler=*/nullptr, /*read_only=*/false)),
         shutting_down_(false),
         mock_table_factory_(new mock::MockTableFactory()),
         error_handler_(nullptr, db_options_, &mutex_),
@@ -293,9 +295,12 @@ class CompactionJobTestBase : public testing::Test {
     Status s = WritableFileWriter::Create(fs_, table_name, FileOptions(),
                                           &file_writer, nullptr);
     ASSERT_OK(s);
+    const ReadOptions read_options;
+    const WriteOptions write_options;
     std::unique_ptr<TableBuilder> table_builder(
         cf_options_.table_factory->NewTableBuilder(
             TableBuilderOptions(*cfd_->ioptions(), mutable_cf_options_,
+                                read_options, write_options,
                                 cfd_->internal_comparator(),
                                 cfd_->int_tbl_prop_collector_factories(),
                                 CompressionType::kNoCompression,
@@ -392,7 +397,7 @@ class CompactionJobTestBase : public testing::Test {
     mutex_.Lock();
     EXPECT_OK(versions_->LogAndApply(
         versions_->GetColumnFamilySet()->GetDefault(), mutable_cf_options_,
-        read_options_, &edit, &mutex_, nullptr));
+        read_options_, write_options_, &edit, &mutex_, nullptr));
     mutex_.Unlock();
   }
 
@@ -540,13 +545,14 @@ class CompactionJobTestBase : public testing::Test {
     ASSERT_OK(s);
     db_options_.info_log = info_log;
 
-    versions_.reset(
-        new VersionSet(dbname_, &db_options_, env_options_, table_cache_.get(),
-                       &write_buffer_manager_, &write_controller_,
-                       /*block_cache_tracer=*/nullptr, /*io_tracer=*/nullptr,
-                       /*db_id*/ "", /*db_session_id*/ ""));
+    versions_.reset(new VersionSet(
+        dbname_, &db_options_, env_options_, table_cache_.get(),
+        &write_buffer_manager_, &write_controller_,
+        /*block_cache_tracer=*/nullptr, /*io_tracer=*/nullptr,
+        /*db_id=*/"", /*db_session_id=*/"", /*daily_offpeak_time_utc=*/"",
+        /*error_handler=*/nullptr, /*read_only=*/false));
     compaction_job_stats_.Reset();
-    ASSERT_OK(SetIdentityFile(env_, dbname_));
+    ASSERT_OK(SetIdentityFile(WriteOptions(), env_, dbname_));
 
     VersionEdit new_db;
     new_db.SetLogNumber(0);
@@ -565,11 +571,11 @@ class CompactionJobTestBase : public testing::Test {
       log::Writer log(std::move(file_writer), 0, false);
       std::string record;
       new_db.EncodeTo(&record);
-      s = log.AddRecord(record);
+      s = log.AddRecord(WriteOptions(), record);
     }
     ASSERT_OK(s);
     // Make "CURRENT" file that points to the new manifest file.
-    s = SetCurrentFile(fs_.get(), dbname_, 1, nullptr);
+    s = SetCurrentFile(WriteOptions(), fs_.get(), dbname_, 1, nullptr);
 
     ASSERT_OK(s);
 
@@ -733,6 +739,7 @@ class CompactionJobTestBase : public testing::Test {
   MutableCFOptions mutable_cf_options_;
   MutableDBOptions mutable_db_options_;
   const ReadOptions read_options_;
+  const WriteOptions write_options_;
   std::shared_ptr<Cache> table_cache_;
   WriteController write_controller_;
   WriteBufferManager write_buffer_manager_;
@@ -1524,13 +1531,15 @@ TEST_F(CompactionJobTest, VerifyPenultimateLevelOutput) {
       {files0, files1, files2, files3}, input_levels,
       /*verify_func=*/[&](Compaction& comp) {
         for (char c = 'a'; c <= 'z'; c++) {
-          std::string c_str;
-          c_str = c;
-          const Slice key(c_str);
           if (c == 'a') {
-            ASSERT_FALSE(comp.WithinPenultimateLevelOutputRange(key));
+            ParsedInternalKey pik("a", 0U, kTypeValue);
+            ASSERT_FALSE(comp.WithinPenultimateLevelOutputRange(pik));
           } else {
-            ASSERT_TRUE(comp.WithinPenultimateLevelOutputRange(key));
+            std::string c_str{c};
+            // WithinPenultimateLevelOutputRange checks internal key range.
+            // 'z' is the last key, so set seqno properly.
+            ParsedInternalKey pik(c_str, c == 'z' ? 12U : 0U, kTypeValue);
+            ASSERT_TRUE(comp.WithinPenultimateLevelOutputRange(pik));
           }
         }
       });

db/compaction/compaction_outputs.cc

@@ -25,11 +25,11 @@ Status CompactionOutputs::Finish(
   assert(meta != nullptr);
   Status s = intput_status;
   if (s.ok()) {
-    std::string seqno_to_time_mapping_str;
-    seqno_to_time_mapping.Encode(
-        seqno_to_time_mapping_str, meta->fd.smallest_seqno,
-        meta->fd.largest_seqno, meta->file_creation_time);
-    builder_->SetSeqnoTimeTableProperties(seqno_to_time_mapping_str,
+    SeqnoToTimeMapping relevant_mapping;
+    relevant_mapping.CopyFromSeqnoRange(
+        seqno_to_time_mapping, meta->fd.smallest_seqno, meta->fd.largest_seqno);
+    relevant_mapping.SetCapacity(kMaxSeqnoTimePairsPerSST);
+    builder_->SetSeqnoTimeTableProperties(relevant_mapping,
                                           meta->oldest_ancester_time);
     s = builder_->Finish();
 
@@ -62,12 +62,15 @@ IOStatus CompactionOutputs::WriterSyncClose(const Status& input_status,
                                             Statistics* statistics,
                                             bool use_fsync) {
   IOStatus io_s;
-  if (input_status.ok()) {
+  IOOptions opts;
+  io_s = WritableFileWriter::PrepareIOOptions(
+      WriteOptions(Env::IOActivity::kCompaction), opts);
+  if (input_status.ok() && io_s.ok()) {
     StopWatch sw(clock, statistics, COMPACTION_OUTFILE_SYNC_MICROS);
-    io_s = file_writer_->Sync(use_fsync);
+    io_s = file_writer_->Sync(opts, use_fsync);
   }
   if (input_status.ok() && io_s.ok()) {
-    io_s = file_writer_->Close();
+    io_s = file_writer_->Close(opts);
   }
 
   if (input_status.ok() && io_s.ok()) {

db/compaction/compaction_picker_level.cc

@@ -45,6 +45,13 @@ bool LevelCompactionPicker::NeedsCompaction(
 }
 
 namespace {
+
+enum class CompactToNextLevel {
+  kNo,   // compact to the same level as the input file
+  kYes,  // compact to the next level except the last level to the same level
+  kSkipLastLevel,  // compact to the next level but skip the last level
+};
+
 // A class to build a leveled compaction step-by-step.
 class LevelCompactionBuilder {
  public:
@@ -115,9 +122,10 @@ class LevelCompactionBuilder {
   // level_files is a vector of (level, file metadata) in ascending order of
   // level. If compact_to_next_level is true, compact the file to the next
   // level, otherwise, compact to the same level as the input file.
+  // If skip_last_level is true, skip the last level.
   void PickFileToCompact(
       const autovector<std::pair<int, FileMetaData*>>& level_files,
-      bool compact_to_next_level);
+      CompactToNextLevel compact_to_next_level);
 
   const std::string& cf_name_;
   VersionStorageInfo* vstorage_;
@@ -149,20 +157,24 @@ class LevelCompactionBuilder {
 
 void LevelCompactionBuilder::PickFileToCompact(
     const autovector<std::pair<int, FileMetaData*>>& level_files,
-    bool compact_to_next_level) {
+    CompactToNextLevel compact_to_next_level) {
   for (auto& level_file : level_files) {
     // If it's being compacted it has nothing to do here.
     // If this assert() fails that means that some function marked some
     // files as being_compacted, but didn't call ComputeCompactionScore()
     assert(!level_file.second->being_compacted);
     start_level_ = level_file.first;
-    if ((compact_to_next_level &&
+    if ((compact_to_next_level == CompactToNextLevel::kSkipLastLevel &&
          start_level_ == vstorage_->num_non_empty_levels() - 1) ||
         (start_level_ == 0 &&
          !compaction_picker_->level0_compactions_in_progress()->empty())) {
       continue;
     }
-    if (compact_to_next_level) {
+
+    // Compact to the next level only if the file is not in the last level and
+    // compact_to_next_level is kYes or kSkipLastLevel.
+    if (compact_to_next_level != CompactToNextLevel::kNo &&
+        (start_level_ < vstorage_->num_non_empty_levels() - 1)) {
       output_level_ =
           (start_level_ == 0) ? vstorage_->base_level() : start_level_ + 1;
     } else {
@@ -248,7 +260,8 @@ void LevelCompactionBuilder::SetupInitialFiles() {
   }
 
   // Bottommost Files Compaction on deleting tombstones
-  PickFileToCompact(vstorage_->BottommostFilesMarkedForCompaction(), false);
+  PickFileToCompact(vstorage_->BottommostFilesMarkedForCompaction(),
+                    CompactToNextLevel::kNo);
   if (!start_level_inputs_.empty()) {
     compaction_reason_ = CompactionReason::kBottommostFiles;
     return;
@@ -274,21 +287,26 @@ void LevelCompactionBuilder::SetupInitialFiles() {
     }
   }
 
-  PickFileToCompact(vstorage_->ExpiredTtlFiles(), true);
+  PickFileToCompact(vstorage_->ExpiredTtlFiles(),
+                    CompactToNextLevel::kSkipLastLevel);
   if (!start_level_inputs_.empty()) {
     compaction_reason_ = CompactionReason::kTtl;
     return;
   }
 
   // Periodic Compaction
-  PickFileToCompact(vstorage_->FilesMarkedForPeriodicCompaction(), false);
+  PickFileToCompact(vstorage_->FilesMarkedForPeriodicCompaction(),
+                    ioptions_.level_compaction_dynamic_level_bytes
+                        ? CompactToNextLevel::kYes
+                        : CompactToNextLevel::kNo);
   if (!start_level_inputs_.empty()) {
     compaction_reason_ = CompactionReason::kPeriodicCompaction;
     return;
   }
 
   // Forced blob garbage collection
-  PickFileToCompact(vstorage_->FilesMarkedForForcedBlobGC(), false);
+  PickFileToCompact(vstorage_->FilesMarkedForForcedBlobGC(),
+                    CompactToNextLevel::kNo);
   if (!start_level_inputs_.empty()) {
     compaction_reason_ = CompactionReason::kForcedBlobGC;
     return;

db/compaction/compaction_picker_test.cc

@@ -85,7 +85,8 @@ class CompactionPickerTestBase : public testing::Test {
     vstorage_.reset(new VersionStorageInfo(
         &icmp_, ucmp_, options_.num_levels, style, nullptr, false,
         EpochNumberRequirement::kMustPresent, ioptions_.clock,
-        options_.bottommost_file_compaction_delay));
+        options_.bottommost_file_compaction_delay,
+        OffpeakTimeOption(mutable_db_options_.daily_offpeak_time_utc)));
     vstorage_->PrepareForVersionAppend(ioptions_, mutable_cf_options_);
   }
 
@@ -95,7 +96,8 @@ class CompactionPickerTestBase : public testing::Test {
     temp_vstorage_.reset(new VersionStorageInfo(
         &icmp_, ucmp_, options_.num_levels, ioptions_.compaction_style,
         vstorage_.get(), false, EpochNumberRequirement::kMustPresent,
-        ioptions_.clock, options_.bottommost_file_compaction_delay));
+        ioptions_.clock, options_.bottommost_file_compaction_delay,
+        OffpeakTimeOption(mutable_db_options_.daily_offpeak_time_utc)));
   }
 
   void DeleteVersionStorage() {
@@ -4147,6 +4149,122 @@ TEST_P(PerKeyPlacementCompactionPickerTest,
   }
 }
 
+TEST_F(CompactionPickerTest,
+       LevelCompactionPrioritizeFilesMarkedForCompaction1) {
+  int num_levels = ioptions_.num_levels;
+  ioptions_.level_compaction_dynamic_level_bytes = true;
+  ioptions_.compaction_pri = kMinOverlappingRatio;
+  mutable_cf_options_.max_bytes_for_level_base = 200;
+  mutable_cf_options_.max_bytes_for_level_multiplier = 10;
+  NewVersionStorage(num_levels, kCompactionStyleLevel);
+  // L5
+  // F4 [100, 200] size:100, F5 [300, 400] size:100,  F6 [500, 600] size:200
+  // F5 is marked for compaction
+  // L6
+  // F1 [100, 200] size:100, F2 [300, 400] size:1000, F3 [500, 600] size:100
+  //
+  // First compaction should pick F5 since it's marked for compaction.
+  // Second compaction should pick F6 since it has min overlap ratio.
+  Add(/*level=*/num_levels - 1, /*file_number=*/1U, /*smallest=*/"100",
+      /*largest=*/"200",
+      /*file_size=*/100, /*path_id=*/0, /*smallest_seq=*/0,
+      /*largest_seq=*/0, /*compensated_file_size=*/100,
+      /*marked_for_compact=*/false);
+  Add(/*level=*/num_levels - 1, /*file_number=*/2U, /*smallest=*/"300",
+      /*largest=*/"400",
+      /*file_size=*/1000, /*path_id=*/0, /*smallest_seq=*/0,
+      /*largest_seq=*/0, /*compensated_file_size=*/1000,
+      /*marked_for_compact=*/false);
+  Add(/*level=*/num_levels - 1, /*file_number=*/3U, /*smallest=*/"500",
+      /*largest=*/"600",
+      /*file_size=*/100, /*path_id=*/0, /*smallest_seq=*/0,
+      /*largest_seq=*/0, /*compensated_file_size=*/100,
+      /*marked_for_compact=*/false);
+
+  Add(/*level=*/num_levels - 2, /*file_number=*/4U, /*smallest=*/"100",
+      /*largest=*/"200",
+      /*file_size=*/100, /*path_id=*/0, /*smallest_seq=*/100,
+      /*largest_seq=*/200, /*compensated_file_size=*/100,
+      /*marked_for_compact=*/false);
+  // Marked for compaction, but with a larger overlap ratio.
+  Add(/*level=*/num_levels - 2, /*file_number=*/5U, /*smallest=*/"300",
+      /*largest=*/"400",
+      /*file_size=*/100, /*path_id=*/0, /*smallest_seq=*/300,
+      /*largest_seq=*/400, /*compensated_file_size=*/100,
+      /*marked_for_compact=*/true);
+  Add(/*level=*/num_levels - 2, /*file_number=*/6U, /*smallest=*/"500",
+      /*largest=*/"600",
+      /*file_size=*/200, /*path_id=*/0, /*smallest_seq=*/400,
+      /*largest_seq=*/500, /*compensated_file_size=*/200,
+      /*marked_for_compact=*/false);
+  UpdateVersionStorageInfo();
+
+  std::unique_ptr<Compaction> compaction(level_compaction_picker.PickCompaction(
+      cf_name_, mutable_cf_options_, mutable_db_options_, vstorage_.get(),
+      &log_buffer_));
+  ASSERT_TRUE(compaction);
+  ASSERT_EQ(num_levels - 2, compaction->start_level());
+  ASSERT_EQ(num_levels - 1, compaction->output_level());
+  ASSERT_EQ(1U, compaction->num_input_files(0));
+  ASSERT_EQ(5U, compaction->input(0, 0)->fd.GetNumber());
+
+  std::unique_ptr<Compaction> second_compaction(
+      level_compaction_picker.PickCompaction(cf_name_, mutable_cf_options_,
+                                             mutable_db_options_,
+                                             vstorage_.get(), &log_buffer_));
+  ASSERT_TRUE(second_compaction);
+  ASSERT_EQ(num_levels - 1, compaction->output_level());
+  ASSERT_EQ(num_levels - 2, compaction->start_level());
+  ASSERT_EQ(1U, compaction->num_input_files(0));
+  ASSERT_EQ(6U, second_compaction->input(0, 0)->fd.GetNumber());
+}
+
+TEST_F(CompactionPickerTest,
+       LevelCompactionPrioritizeFilesMarkedForCompaction2) {
+  int num_levels = ioptions_.num_levels;
+  ioptions_.level_compaction_dynamic_level_bytes = true;
+  ioptions_.compaction_pri = kMinOverlappingRatio;
+  mutable_cf_options_.max_bytes_for_level_base = 200;
+  mutable_cf_options_.max_bytes_for_level_multiplier = 10;
+  // L4
+  // F3 [100, 200] size:2000
+  // L5
+  // F2 [100, 200] size:2000, marked for compaction
+  // L6
+  // F1 [100, 200] size: 20000
+  //
+  // L4 should be prioritized over L5 since L4 has a higher compaction score.
+  // Files marked for compaction do not affect level picking order.
+  NewVersionStorage(num_levels, kCompactionStyleLevel);
+  Add(/*level=*/num_levels - 1, /*file_number=*/1U, /*smallest=*/"100",
+      /*largest=*/"200",
+      /*file_size=*/20000, /*path_id=*/0, /*smallest_seq=*/0,
+      /*largest_seq=*/0, /*compensated_file_size=*/100,
+      /*marked_for_compact=*/false);
+  // Level score should be 1.
+  Add(/*level=*/num_levels - 2, /*file_number=*/2U, /*smallest=*/"100",
+      /*largest=*/"200",
+      /*file_size=*/2000, /*path_id=*/0, /*smallest_seq=*/100,
+      /*largest_seq=*/200, /*compensated_file_size=*/2000,
+      /*marked_for_compact=*/true);
+  // Level score should be larger than L5.
+  Add(/*level=*/num_levels - 3, /*file_number=*/3U, /*smallest=*/"100",
+      /*largest=*/"200",
+      /*file_size=*/2000, /*path_id=*/0, /*smallest_seq=*/300,
+      /*largest_seq=*/400, /*compensated_file_size=*/2000,
+      /*marked_for_compact=*/false);
+  UpdateVersionStorageInfo();
+
+  std::unique_ptr<Compaction> compaction(level_compaction_picker.PickCompaction(
+      cf_name_, mutable_cf_options_, mutable_db_options_, vstorage_.get(),
+      &log_buffer_));
+  ASSERT_TRUE(compaction);
+  ASSERT_EQ(num_levels - 3, compaction->start_level());
+  ASSERT_EQ(num_levels - 2, compaction->output_level());
+  ASSERT_EQ(1U, compaction->num_input_files(0));
+  ASSERT_EQ(3U, compaction->input(0, 0)->fd.GetNumber());
+}
+
 INSTANTIATE_TEST_CASE_P(PerKeyPlacementCompactionPickerTest,
                         PerKeyPlacementCompactionPickerTest, ::testing::Bool());
 

db/compaction/tiered_compaction_test.cc

@@ -1202,15 +1202,120 @@ TEST_P(TieredCompactionTest, RangeBasedTieredStorageLevel) {
   ASSERT_EQ("0,0,0,0,0,0,1", FilesPerLevel());
   ASSERT_EQ(GetSstSizeHelper(Temperature::kUnknown), 0);
   ASSERT_GT(GetSstSizeHelper(Temperature::kCold), 0);
-
   ASSERT_EQ(
       options.statistics->getTickerCount(COMPACTION_RANGE_DEL_DROP_OBSOLETE),
       1);
+
+  // Tests that we only compact keys up to penultimate level
+  // that are within penultimate level input's internal key range.
+  {
+    MutexLock l(&mutex);
+    hot_start = Key(0);
+    hot_end = Key(100);
+  }
+  const Snapshot* temp_snap = db_->GetSnapshot();
+  // Key(0) and Key(1) here are inserted with higher sequence number
+  // than Key(0) and Key(1) inserted above.
+  // Only Key(0) in last level will be compacted up, not Key(1).
+  ASSERT_OK(Put(Key(0), "value" + std::to_string(0)));
+  ASSERT_OK(Put(Key(1), "value" + std::to_string(100)));
+  ASSERT_OK(db_->CompactRange(cro, nullptr, nullptr));
+  ASSERT_EQ("0,0,0,0,0,1,1", FilesPerLevel());
+  {
+    std::vector<LiveFileMetaData> metas;
+    db_->GetLiveFilesMetaData(&metas);
+    for (const auto& f : metas) {
+      if (f.temperature == Temperature::kUnknown) {
+        // Expect Key(0), Key(0), Key(1)
+        ASSERT_EQ(f.num_entries, 3);
+        ASSERT_EQ(f.smallestkey, Key(0));
+        ASSERT_EQ(f.largestkey, Key(1));
+      } else {
+        ASSERT_EQ(f.temperature, Temperature::kCold);
+        // Key(2)-Key(49) and Key(100).
+        ASSERT_EQ(f.num_entries, 50);
+      }
+    }
+  }
+  db_->ReleaseSnapshot(temp_snap);
 }
 
 INSTANTIATE_TEST_CASE_P(TieredCompactionTest, TieredCompactionTest,
                         testing::Bool());
 
+TEST_P(TieredCompactionTest, CheckInternalKeyRange) {
+  // When compacting keys from the last level to penultimate level,
+  // output to penultimate level should be within internal key range
+  // of input files from penultimate level.
+  // Set up:
+  // L5:
+  //  File 1: DeleteRange[1, 3)@4, File 2: [3@5, 100@6]
+  // L6:
+  // File 3: [2@1, 3@2], File 4: [50@3]
+  //
+  // When File 1 and File 3 are being compacted,
+  // Key(3) cannot be compacted up, otherwise it causes
+  // inconsistency where File 3's Key(3) has a lower sequence number
+  // than File 2's Key(3).
+  const int kNumLevels = 7;
+  auto options = CurrentOptions();
+  SetColdTemperature(options);
+  options.level_compaction_dynamic_level_bytes = true;
+  options.num_levels = kNumLevels;
+  options.statistics = CreateDBStatistics();
+  options.max_subcompactions = 10;
+  options.preclude_last_level_data_seconds = 10000;
+  DestroyAndReopen(options);
+  auto cmp = options.comparator;
+
+  std::string hot_start = Key(0);
+  std::string hot_end = Key(0);
+  SyncPoint::GetInstance()->SetCallBack(
+      "CompactionIterator::PrepareOutput.context", [&](void* arg) {
+        auto context = static_cast<PerKeyPlacementContext*>(arg);
+        context->output_to_penultimate_level =
+            cmp->Compare(context->key, hot_start) >= 0 &&
+            cmp->Compare(context->key, hot_end) < 0;
+      });
+  SyncPoint::GetInstance()->EnableProcessing();
+  // File 1
+  ASSERT_OK(Put(Key(2), "val2"));
+  ASSERT_OK(Put(Key(3), "val3"));
+  ASSERT_OK(Flush());
+  MoveFilesToLevel(6);
+  // File 2
+  ASSERT_OK(Put(Key(50), "val50"));
+  ASSERT_OK(Flush());
+  MoveFilesToLevel(6);
+
+  const Snapshot* snapshot = db_->GetSnapshot();
+  hot_end = Key(100);
+  std::string start = Key(1);
+  std::string end = Key(3);
+  ASSERT_OK(
+      db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(), start, end));
+  ASSERT_OK(Flush());
+  MoveFilesToLevel(5);
+  // File 3
+  ASSERT_OK(Put(Key(3), "vall"));
+  ASSERT_OK(Put(Key(100), "val100"));
+  ASSERT_OK(Flush());
+  MoveFilesToLevel(5);
+  // Try to compact keys up
+  CompactRangeOptions cro;
+  cro.bottommost_level_compaction = BottommostLevelCompaction::kForce;
+  start = Key(1);
+  end = Key(2);
+  Slice begin_slice(start);
+  Slice end_slice(end);
+  ASSERT_OK(db_->CompactRange(cro, &begin_slice, &end_slice));
+  // Without internal key range checking, we get the following error:
+  // Corruption: force_consistency_checks(DEBUG): VersionBuilder: L5 has
+  // overlapping ranges: file #18 largest key: '6B6579303030303033' seq:102,
+  // type:1 vs. file #15 smallest key: '6B6579303030303033' seq:104, type:1
+  db_->ReleaseSnapshot(snapshot);
+}
+
 class PrecludeLastLevelTest : public DBTestBase {
  public:
   PrecludeLastLevelTest()
@@ -1470,9 +1575,8 @@ TEST_F(PrecludeLastLevelTest, SmallPrecludeTime) {
   ASSERT_EQ(tables_props.size(), 1);
   ASSERT_FALSE(tables_props.begin()->second->seqno_to_time_mapping.empty());
   SeqnoToTimeMapping tp_mapping;
-  ASSERT_OK(
-      tp_mapping.Add(tables_props.begin()->second->seqno_to_time_mapping));
-  ASSERT_OK(tp_mapping.Sort());
+  ASSERT_OK(tp_mapping.DecodeFrom(
+      tables_props.begin()->second->seqno_to_time_mapping));
   ASSERT_FALSE(tp_mapping.Empty());
   auto seqs = tp_mapping.TEST_GetInternalMapping();
   ASSERT_FALSE(seqs.empty());
@@ -1971,13 +2075,7 @@ TEST_F(PrecludeLastLevelTest, PartialPenultimateLevelCompaction) {
   Close();
 }
 
-// FIXME broken test:
-// dbfull()->TEST_WaitForCompact()
-// Corruption: force_consistency_checks(DEBUG): VersionBuilder: L5 has
-// overlapping ranges:
-// file #14 largest key: '6B6579303030303134' seq:32, type:1 vs.
-// file #19 smallest key: '6B6579303030303130' seq:10, type:1
-TEST_F(PrecludeLastLevelTest, DISABLED_RangeDelsCauseFileEndpointsToOverlap) {
+TEST_F(PrecludeLastLevelTest, RangeDelsCauseFileEndpointsToOverlap) {
   const int kNumLevels = 7;
   const int kSecondsPerKey = 10;
   const int kNumFiles = 3;

db/convenience.cc

@@ -34,7 +34,7 @@ Status DeleteFilesInRanges(DB* db, ColumnFamilyHandle* column_family,
 Status VerifySstFileChecksum(const Options& options,
                              const EnvOptions& env_options,
                              const std::string& file_path) {
-  // TODO: plumb Env::IOActivity
+  // TODO: plumb Env::IOActivity, Env::IOPriority
   const ReadOptions read_options;
   return VerifySstFileChecksum(options, env_options, read_options, file_path);
 }

db/db_basic_test.cc

@@ -1853,6 +1853,56 @@ TEST_P(DBMultiGetTestWithParam, MultiGetBatchedMultiLevel) {
   }
 }
 
+TEST_P(DBMultiGetTestWithParam, MultiGetBatchedEmptyLevel) {
+#ifndef USE_COROUTINES
+  if (std::get<1>(GetParam())) {
+    ROCKSDB_GTEST_BYPASS("This test requires coroutine support");
+    return;
+  }
+#endif  // USE_COROUTINES
+  // Skip for unbatched MultiGet
+  if (!std::get<0>(GetParam())) {
+    ROCKSDB_GTEST_BYPASS("This test is only for batched MultiGet");
+    return;
+  }
+  Options options = CurrentOptions();
+  options.disable_auto_compactions = true;
+  options.merge_operator = MergeOperators::CreateStringAppendOperator();
+  Reopen(options);
+  int key;
+
+  key = 9;
+  ASSERT_OK(Put("key_" + std::to_string(key), "val_l2_" + std::to_string(key)));
+  ASSERT_OK(Flush());
+  MoveFilesToLevel(4);
+
+  key = 5;
+  ASSERT_OK(Put("key_" + std::to_string(key), "val_l2_" + std::to_string(key)));
+  key = 9;
+  ASSERT_OK(
+      Merge("key_" + std::to_string(key), "val_l2_" + std::to_string(key)));
+  ASSERT_OK(Flush());
+  // Leave level 3 empty
+  MoveFilesToLevel(2);
+
+  key = 2;
+  ASSERT_OK(Put("key_" + std::to_string(key), "val_l2_" + std::to_string(key)));
+  key = 6;
+  ASSERT_OK(
+      Merge("key_" + std::to_string(key), "val_l2_" + std::to_string(key)));
+  ASSERT_OK(Flush());
+  MoveFilesToLevel(1);
+
+  std::vector<std::string> keys;
+  std::vector<std::string> values;
+
+  keys.push_back("key_" + std::to_string(9));
+  keys.push_back("key_" + std::to_string(9));
+
+  values = MultiGet(keys, nullptr, std::get<1>(GetParam()));
+  ASSERT_EQ(values.size(), 2);
+}
+
 TEST_P(DBMultiGetTestWithParam, MultiGetBatchedMultiLevelMerge) {
 #ifndef USE_COROUTINES
   if (std::get<1>(GetParam())) {
@@ -3126,7 +3176,8 @@ TEST_F(DBBasicTest, LastSstFileNotInManifest) {
   // Manually add a sst file.
   constexpr uint64_t kSstFileNumber = 100;
   const std::string kSstFile = MakeTableFileName(dbname_, kSstFileNumber);
-  ASSERT_OK(WriteStringToFile(env_, /* data = */ "bad sst file content",
+  ASSERT_OK(WriteStringToFile(env_,
+                              /* data = */ "bad sst file content",
                               /* fname = */ kSstFile,
                               /* should_sync = */ true));
   ASSERT_OK(env_->FileExists(kSstFile));
@@ -4415,6 +4466,8 @@ TEST_F(DBBasicTest, ManifestWriteFailure) {
   options.create_if_missing = true;
   options.disable_auto_compactions = true;
   options.env = env_;
+  options.enable_blob_files = true;
+  options.blob_file_size = 0;
   DestroyAndReopen(options);
   ASSERT_OK(Put("foo", "bar"));
   ASSERT_OK(Flush());
@@ -4435,6 +4488,11 @@ TEST_F(DBBasicTest, ManifestWriteFailure) {
   SyncPoint::GetInstance()->ClearAllCallBacks();
   SyncPoint::GetInstance()->EnableProcessing();
   Reopen(options);
+  // The IO error was a mocked one from the `AfterSyncManifest` callback. The
+  // Flush's VersionEdit actually made it into the Manifest. So these keys can
+  // be read back. Read them to check all live sst files and blob files.
+  ASSERT_EQ("bar", Get("foo"));
+  ASSERT_EQ("value", Get("key"));
 }
 
 TEST_F(DBBasicTest, DestroyDefaultCfHandle) {

db/db_compaction_test.cc

@@ -19,7 +19,7 @@
 #include "rocksdb/concurrent_task_limiter.h"
 #include "rocksdb/experimental.h"
 #include "rocksdb/sst_file_writer.h"
-#include "rocksdb/utilities/convenience.h"
+#include "test_util/mock_time_env.h"
 #include "test_util/sync_point.h"
 #include "test_util/testutil.h"
 #include "util/concurrent_task_limiter_impl.h"
@@ -76,7 +76,7 @@ class CompactionStatsCollector : public EventListener {
 class DBCompactionTest : public DBTestBase {
  public:
   DBCompactionTest()
-      : DBTestBase("db_compaction_test", /*env_do_fsync=*/true) {}
+      : DBTestBase("db_compaction_test", /*env_do_fsync=*/false) {}
 
  protected:
   /*
@@ -121,7 +121,7 @@ class DBCompactionTestWithParam
       public testing::WithParamInterface<std::tuple<uint32_t, bool>> {
  public:
   DBCompactionTestWithParam()
-      : DBTestBase("db_compaction_test", /*env_do_fsync=*/true) {
+      : DBTestBase("db_compaction_test", /*env_do_fsync=*/false) {
     max_subcompactions_ = std::get<0>(GetParam());
     exclusive_manual_compaction_ = std::get<1>(GetParam());
   }
@@ -140,7 +140,7 @@ class DBCompactionTestWithBottommostParam
           std::tuple<BottommostLevelCompaction, bool>> {
  public:
   DBCompactionTestWithBottommostParam()
-      : DBTestBase("db_compaction_test", /*env_do_fsync=*/true) {
+      : DBTestBase("db_compaction_test", /*env_do_fsync=*/false) {
     bottommost_level_compaction_ = std::get<0>(GetParam());
   }
 
@@ -160,7 +160,7 @@ class DBCompactionWaitForCompactTest
           std::tuple<bool, bool, bool, std::chrono::microseconds>> {
  public:
   DBCompactionWaitForCompactTest()
-      : DBTestBase("db_compaction_test", /*env_do_fsync=*/true) {
+      : DBTestBase("db_compaction_test", /*env_do_fsync=*/false) {
     abort_on_pause_ = std::get<0>(GetParam());
     flush_ = std::get<1>(GetParam());
     close_db_ = std::get<2>(GetParam());
@@ -845,6 +845,20 @@ TEST_F(DBCompactionTest, BGCompactionsAllowed) {
   options.memtable_factory.reset(
       test::NewSpecialSkipListFactory(kNumKeysPerFile));
 
+  CreateAndReopenWithCF({"one", "two", "three"}, options);
+
+  Random rnd(301);
+  for (int cf = 0; cf < 4; cf++) {
+    // Make a trivial L1 for L0 to compact into. L2 will be large so debt ratio
+    // will not cause compaction pressure.
+    ASSERT_OK(Put(cf, Key(0), rnd.RandomString(102400)));
+    ASSERT_OK(Flush(cf));
+    MoveFilesToLevel(2, cf);
+    ASSERT_OK(Put(cf, Key(0), ""));
+    ASSERT_OK(Flush(cf));
+    MoveFilesToLevel(1, cf);
+  }
+
   // Block all threads in thread pool.
   const size_t kTotalTasks = 4;
   env_->SetBackgroundThreads(4, Env::LOW);
@@ -855,9 +869,6 @@ TEST_F(DBCompactionTest, BGCompactionsAllowed) {
     sleeping_tasks[i].WaitUntilSleeping();
   }
 
-  CreateAndReopenWithCF({"one", "two", "three"}, options);
-
-  Random rnd(301);
   for (int cf = 0; cf < 4; cf++) {
     for (int num = 0; num < options.level0_file_num_compaction_trigger; num++) {
       for (int i = 0; i < kNumKeysPerFile; i++) {
@@ -4794,9 +4805,9 @@ TEST_F(DBCompactionTest, LevelTtlCascadingCompactions) {
       ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
           "VersionEdit::EncodeTo:VarintOldestAncesterTime", [&](void* arg) {
             if (if_restart && if_open_all_files) {
-              std::string* encoded_fieled = static_cast<std::string*>(arg);
-              *encoded_fieled = "";
-              PutVarint64(encoded_fieled, 0);
+              std::string* encoded_field = static_cast<std::string*>(arg);
+              *encoded_field = "";
+              PutVarint64(encoded_field, 0);
             }
           });
       ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
@@ -4942,9 +4953,9 @@ TEST_F(DBCompactionTest, LevelPeriodicCompaction) {
       ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
           "VersionEdit::EncodeTo:VarintFileCreationTime", [&](void* arg) {
             if (if_restart && if_open_all_files) {
-              std::string* encoded_fieled = static_cast<std::string*>(arg);
-              *encoded_fieled = "";
-              PutVarint64(encoded_fieled, 0);
+              std::string* encoded_field = static_cast<std::string*>(arg);
+              *encoded_field = "";
+              PutVarint64(encoded_field, 0);
             }
           });
       ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
@@ -5020,6 +5031,106 @@ TEST_F(DBCompactionTest, LevelPeriodicCompaction) {
   }
 }
 
+TEST_F(DBCompactionTest, LevelPeriodicCompactionOffpeak) {
+  // This test simply checks if offpeak adjustment works in Leveled
+  // Compactions. For testing offpeak periodic compactions in various
+  // scenarios, please refer to
+  // DBTestUniversalCompaction2::PeriodicCompactionOffpeak
+  constexpr int kNumKeysPerFile = 32;
+  constexpr int kNumLevelFiles = 2;
+  constexpr int kValueSize = 100;
+  constexpr int kSecondsPerDay = 86400;
+  constexpr int kSecondsPerHour = 3600;
+  constexpr int kSecondsPerMinute = 60;
+
+  for (bool if_restart : {false, true}) {
+    SCOPED_TRACE("if_restart=" + std::to_string(if_restart));
+    Options options = CurrentOptions();
+    options.ttl = 0;
+    options.periodic_compaction_seconds = 5 * kSecondsPerDay;  // 5 days
+    // In the case where all files are opened and doing DB restart
+    // forcing the file creation time in manifest file to be 0 to
+    // simulate the case of reading from an old version.
+    ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
+        "VersionEdit::EncodeTo:VarintFileCreationTime", [&](void* arg) {
+          if (if_restart) {
+            std::string* encoded_field = static_cast<std::string*>(arg);
+            *encoded_field = "";
+            PutVarint64(encoded_field, 0);
+          }
+        });
+    ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
+
+    // Just to add some extra random days to current time
+    Random rnd(test::RandomSeed());
+    int days = rnd.Uniform(100);
+
+    int periodic_compactions = 0;
+    ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
+        "LevelCompactionPicker::PickCompaction:Return", [&](void* arg) {
+          Compaction* compaction = static_cast<Compaction*>(arg);
+          auto compaction_reason = compaction->compaction_reason();
+          if (compaction_reason == CompactionReason::kPeriodicCompaction) {
+            periodic_compactions++;
+          }
+        });
+
+    // Starting at 12:15AM
+    int now_hour = 0;
+    int now_minute = 15;
+    auto mock_clock = std::make_shared<MockSystemClock>(env_->GetSystemClock());
+    auto mock_env = std::make_unique<CompositeEnvWrapper>(env_, mock_clock);
+    options.env = mock_env.get();
+    mock_clock->SetCurrentTime(days * kSecondsPerDay +
+                               now_hour * kSecondsPerHour +
+                               now_minute * kSecondsPerMinute);
+    // Offpeak is set from 12:30AM to 4:30AM
+    options.daily_offpeak_time_utc = "00:30-04:30";
+    Reopen(options);
+
+    for (int i = 0; i < kNumLevelFiles; ++i) {
+      for (int j = 0; j < kNumKeysPerFile; ++j) {
+        ASSERT_OK(
+            Put(Key(i * kNumKeysPerFile + j), rnd.RandomString(kValueSize)));
+      }
+      ASSERT_OK(Flush());
+    }
+    ASSERT_OK(dbfull()->TEST_WaitForCompact());
+    ASSERT_EQ("2", FilesPerLevel());
+    ASSERT_EQ(0, periodic_compactions);
+
+    // Move clock forward by 1 hour. Now at 1:15AM Day 0. No compaction.
+    mock_clock->MockSleepForSeconds(1 * kSecondsPerHour);
+    ASSERT_OK(Put("a", "1"));
+    ASSERT_OK(Flush());
+    ASSERT_OK(dbfull()->TEST_WaitForCompact());
+    // Assert that the files stay in the same level
+    ASSERT_EQ("3", FilesPerLevel());
+    ASSERT_EQ(0, periodic_compactions);
+    MoveFilesToLevel(1);
+    ASSERT_EQ("0,3", FilesPerLevel());
+
+    // Move clock forward by 4 days and check if it triggers periodic
+    // comapaction at 1:15AM Day 4. Files created on Day 0 at 12:15AM is
+    // expected to expire before the offpeak starts next day at 12:30AM
+    mock_clock->MockSleepForSeconds(4 * kSecondsPerDay);
+    ASSERT_OK(Put("b", "2"));
+    if (if_restart) {
+      Reopen(options);
+    } else {
+      ASSERT_OK(Flush());
+    }
+    ASSERT_OK(dbfull()->TEST_WaitForCompact());
+    ASSERT_EQ("1,3", FilesPerLevel());
+    // The two old files go through the periodic compaction process
+    ASSERT_EQ(2, periodic_compactions);
+
+    Destroy(options);
+
+    ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
+  }
+}
+
 TEST_F(DBCompactionTest, LevelPeriodicCompactionWithOldDB) {
   // This test makes sure that periodic compactions are working with a DB
   // where file_creation_time of some files is 0.
@@ -6050,7 +6161,7 @@ class CompactionPriTest : public DBTestBase,
                           public testing::WithParamInterface<uint32_t> {
  public:
   CompactionPriTest()
-      : DBTestBase("compaction_pri_test", /*env_do_fsync=*/true) {
+      : DBTestBase("compaction_pri_test", /*env_do_fsync=*/false) {
     compaction_pri_ = GetParam();
   }
 
@@ -6170,26 +6281,30 @@ TEST_F(DBCompactionTest, PersistRoundRobinCompactCursor) {
 
 TEST_P(RoundRobinSubcompactionsAgainstPressureToken, PressureTokenTest) {
   const int kKeysPerBuffer = 100;
+  const int kNumSubcompactions = 2;
+  const int kFilesPerLevel = 50;
   Options options = CurrentOptions();
-  options.num_levels = 4;
+  options.num_levels = 3;
   options.max_bytes_for_level_multiplier = 2;
   options.level0_file_num_compaction_trigger = 4;
   options.target_file_size_base = kKeysPerBuffer * 1024;
   options.compaction_pri = CompactionPri::kRoundRobin;
-  options.max_bytes_for_level_base = 8 * kKeysPerBuffer * 1024;
+  // Target size is chosen so that filling the level with `kFilesPerLevel` files
+  // will make it oversized by `kNumSubcompactions` files.
+  options.max_bytes_for_level_base =
+      (kFilesPerLevel - kNumSubcompactions) * kKeysPerBuffer * 1024;
   options.disable_auto_compactions = true;
-  // Setup 7 threads but limited subcompactions so that
-  // RoundRobin requires extra compactions from reserved threads
+  // Setup `kNumSubcompactions` threads but limited subcompactions so
+  // that RoundRobin requires extra compactions from reserved threads
   options.max_subcompactions = 1;
-  options.max_background_compactions = 7;
+  options.max_background_compactions = kNumSubcompactions;
   options.max_compaction_bytes = 100000000;
   DestroyAndReopen(options);
-  env_->SetBackgroundThreads(7, Env::LOW);
+  env_->SetBackgroundThreads(kNumSubcompactions, Env::LOW);
 
   Random rnd(301);
-  const std::vector<int> files_per_level = {0, 15, 25};
   for (int lvl = 2; lvl > 0; lvl--) {
-    for (int i = 0; i < files_per_level[lvl]; i++) {
+    for (int i = 0; i < kFilesPerLevel; i++) {
       for (int j = 0; j < kKeysPerBuffer; j++) {
         // Add (lvl-1) to ensure nearly equivallent number of files
         // in L2 are overlapped with fils selected to compact from
@@ -6200,9 +6315,8 @@ TEST_P(RoundRobinSubcompactionsAgainstPressureToken, PressureTokenTest) {
       ASSERT_OK(Flush());
     }
     MoveFilesToLevel(lvl);
-    ASSERT_EQ(files_per_level[lvl], NumTableFilesAtLevel(lvl, 0));
+    ASSERT_EQ(kFilesPerLevel, NumTableFilesAtLevel(lvl, 0));
   }
-  // 15 files in L1; 25 files in L2
 
   // This is a variable for making sure the following callback is called
   // and the assertions in it are indeed excuted.
@@ -6211,10 +6325,10 @@ TEST_P(RoundRobinSubcompactionsAgainstPressureToken, PressureTokenTest) {
       "CompactionJob::GenSubcompactionBoundaries:0", [&](void* arg) {
         uint64_t num_planned_subcompactions = *(static_cast<uint64_t*>(arg));
         if (grab_pressure_token_) {
-          // 7 files are selected for round-robin under auto
+          // `kNumSubcompactions` files are selected for round-robin under auto
           // compaction. The number of planned subcompaction is restricted by
           // the limited number of max_background_compactions
-          ASSERT_EQ(num_planned_subcompactions, 7);
+          ASSERT_EQ(num_planned_subcompactions, kNumSubcompactions);
         } else {
           ASSERT_EQ(num_planned_subcompactions, 1);
         }

db/db_filesnapshot.cc

@@ -121,7 +121,7 @@ Status DBImpl::GetSortedWalFiles(VectorLogPtr& files) {
 
   // DisableFileDeletions / EnableFileDeletions not supported in read-only DB
   if (deletions_disabled.ok()) {
-    Status s2 = EnableFileDeletions(/*force*/ false);
+    Status s2 = EnableFileDeletions(/*force=*/false);
     assert(s2.ok());
     s2.PermitUncheckedError();
   } else {

db/db_flush_test.cc

@@ -3035,6 +3035,39 @@ TEST_P(DBAtomicFlushTest, RollbackAfterFailToInstallResults) {
   SyncPoint::GetInstance()->ClearAllCallBacks();
 }
 
+TEST_P(DBAtomicFlushTest, FailureInMultiCfAutomaticFlush) {
+  bool atomic_flush = GetParam();
+  auto fault_injection_env = std::make_shared<FaultInjectionTestEnv>(env_);
+  Options options = CurrentOptions();
+  options.env = fault_injection_env.get();
+  options.create_if_missing = true;
+  options.atomic_flush = atomic_flush;
+  const int kNumKeysTriggerFlush = 4;
+  options.memtable_factory.reset(
+      test::NewSpecialSkipListFactory(kNumKeysTriggerFlush));
+  CreateAndReopenWithCF({"pikachu"}, options);
+  ASSERT_EQ(2, handles_.size());
+  for (size_t cf = 0; cf < handles_.size(); ++cf) {
+    ASSERT_OK(Put(static_cast<int>(cf), "a", "value"));
+  }
+  SyncPoint::GetInstance()->DisableProcessing();
+  SyncPoint::GetInstance()->ClearAllCallBacks();
+  SyncPoint::GetInstance()->SetCallBack(
+      "DBImpl::ScheduleFlushes:PreSwitchMemtable",
+      [&](void* /*arg*/) { fault_injection_env->SetFilesystemActive(false); });
+  SyncPoint::GetInstance()->EnableProcessing();
+
+  for (int i = 1; i < kNumKeysTriggerFlush; ++i) {
+    ASSERT_OK(Put(0, "key" + std::to_string(i), "value" + std::to_string(i)));
+  }
+  ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable());
+  // Next write after failed flush should fail.
+  ASSERT_NOK(Put(0, "x", "y"));
+  fault_injection_env->SetFilesystemActive(true);
+  Close();
+  SyncPoint::GetInstance()->ClearAllCallBacks();
+}
+
 // In atomic flush, concurrent bg flush threads commit to the MANIFEST in
 // serial, in the order of their picked memtables for each column family.
 // Only when a bg flush thread finds out that its memtables are the earliest

db/db_impl/db_impl.cc

@@ -276,10 +276,11 @@ DBImpl::DBImpl(const DBOptions& options, const std::string& dbname,
         this->RecordSeqnoToTimeMapping(/*populate_historical_seconds=*/0);
       });
 
-  versions_.reset(new VersionSet(dbname_, &immutable_db_options_, file_options_,
-                                 table_cache_.get(), write_buffer_manager_,
-                                 &write_controller_, &block_cache_tracer_,
-                                 io_tracer_, db_id_, db_session_id_));
+  versions_.reset(new VersionSet(
+      dbname_, &immutable_db_options_, file_options_, table_cache_.get(),
+      write_buffer_manager_, &write_controller_, &block_cache_tracer_,
+      io_tracer_, db_id_, db_session_id_, options.daily_offpeak_time_utc,
+      &error_handler_, read_only));
   column_family_memtables_.reset(
       new ColumnFamilyMemTablesImpl(versions_->GetColumnFamilySet()));
 
@@ -332,8 +333,10 @@ Status DBImpl::Resume() {
 Status DBImpl::ResumeImpl(DBRecoverContext context) {
   mutex_.AssertHeld();
 
-  // TODO: plumb Env::IOActivity
+  // TODO: plumb Env::IOActivity, Env::IOPriority
   const ReadOptions read_options;
+  const WriteOptions write_options;
+
   WaitForBackgroundWork();
 
   Status s;
@@ -359,10 +362,8 @@ Status DBImpl::ResumeImpl(DBRecoverContext context) {
     if (io_s.IsIOError()) {
       // If resuming from IOError resulted from MANIFEST write, then assert
       // that we must have already set the MANIFEST writer to nullptr during
-      // clean-up phase MANIFEST writing. We must have also disabled file
-      // deletions.
+      // clean-up phase MANIFEST writing.
       assert(!versions_->descriptor_log_);
-      assert(!IsFileDeletionsEnabled());
       // Since we are trying to recover from MANIFEST write error, we need to
       // switch to a new MANIFEST anyway. The old MANIFEST can be corrupted.
       // Therefore, force writing a dummy version edit because we do not know
@@ -374,8 +375,8 @@ Status DBImpl::ResumeImpl(DBRecoverContext context) {
       assert(cfh);
       ColumnFamilyData* cfd = cfh->cfd();
       const MutableCFOptions& cf_opts = *cfd->GetLatestMutableCFOptions();
-      s = versions_->LogAndApply(cfd, cf_opts, read_options, &edit, &mutex_,
-                                 directories_.GetDbDir());
+      s = versions_->LogAndApply(cfd, cf_opts, read_options, write_options,
+                                 &edit, &mutex_, directories_.GetDbDir());
       if (!s.ok()) {
         io_s = versions_->io_status();
         if (!io_s.ok()) {
@@ -658,6 +659,18 @@ Status DBImpl::CloseHelper() {
 
   // versions need to be destroyed before table_cache since it can hold
   // references to table_cache.
+  {
+    Status s = versions_->Close(directories_.GetDbDir(), &mutex_);
+    if (!s.ok()) {
+      ROCKS_LOG_ERROR(immutable_db_options_.info_log,
+                      "Unable to close MANIFEST with error -- %s",
+                      s.ToString().c_str());
+      if (ret.ok()) {
+        ret = s;
+      }
+    }
+  }
+
   versions_.reset();
   mutex_.Unlock();
   if (db_lock_ != nullptr) {
@@ -705,23 +718,26 @@ Status DBImpl::CloseHelper() {
 Status DBImpl::CloseImpl() { return CloseHelper(); }
 
 DBImpl::~DBImpl() {
+  ThreadStatus::OperationType cur_op_type =
+      ThreadStatusUtil::GetThreadOperation();
+  ThreadStatusUtil::SetThreadOperation(ThreadStatus::OperationType::OP_UNKNOWN);
+
   // TODO: remove this.
   init_logger_creation_s_.PermitUncheckedError();
 
   InstrumentedMutexLock closing_lock_guard(&closing_mutex_);
-  if (closed_) {
-    return;
+  if (!closed_) {
+    closed_ = true;
+
+    {
+      const Status s = MaybeReleaseTimestampedSnapshotsAndCheck();
+      s.PermitUncheckedError();
+    }
+
+    closing_status_ = CloseImpl();
+    closing_status_.PermitUncheckedError();
   }
-
-  closed_ = true;
-
-  {
-    const Status s = MaybeReleaseTimestampedSnapshotsAndCheck();
-    s.PermitUncheckedError();
-  }
-
-  closing_status_ = CloseImpl();
-  closing_status_.PermitUncheckedError();
+  ThreadStatusUtil::SetThreadOperation(cur_op_type);
 }
 
 void DBImpl::MaybeIgnoreError(Status* s) const {
@@ -796,7 +812,9 @@ Status DBImpl::StartPeriodicTaskScheduler() {
   return s;
 }
 
-Status DBImpl::RegisterRecordSeqnoTimeWorker(bool is_new_db) {
+Status DBImpl::RegisterRecordSeqnoTimeWorker(const ReadOptions& read_options,
+                                             const WriteOptions& write_options,
+                                             bool is_new_db) {
   options_mutex_.AssertHeld();
 
   uint64_t min_preserve_seconds = std::numeric_limits<uint64_t>::max();
@@ -816,9 +834,15 @@ Status DBImpl::RegisterRecordSeqnoTimeWorker(bool is_new_db) {
       }
     }
     if (min_preserve_seconds == std::numeric_limits<uint64_t>::max()) {
-      seqno_to_time_mapping_.Resize(0, 0);
+      // Don't track
+      seqno_to_time_mapping_.SetCapacity(0);
+      seqno_to_time_mapping_.SetMaxTimeSpan(UINT64_MAX);
     } else {
-      seqno_to_time_mapping_.Resize(min_preserve_seconds, max_preserve_seconds);
+      uint64_t cap = std::min(kMaxSeqnoToTimeEntries,
+                              max_preserve_seconds * kMaxSeqnoTimePairsPerCF /
+                                  min_preserve_seconds);
+      seqno_to_time_mapping_.SetCapacity(cap);
+      seqno_to_time_mapping_.SetMaxTimeSpan(max_preserve_seconds);
     }
     mapping_was_empty = seqno_to_time_mapping_.Empty();
   }
@@ -827,9 +851,8 @@ Status DBImpl::RegisterRecordSeqnoTimeWorker(bool is_new_db) {
   if (min_preserve_seconds != std::numeric_limits<uint64_t>::max()) {
     // round up to 1 when the time_duration is smaller than
     // kMaxSeqnoTimePairsPerCF
-    seqno_time_cadence = (min_preserve_seconds +
-                          SeqnoToTimeMapping::kMaxSeqnoTimePairsPerCF - 1) /
-                         SeqnoToTimeMapping::kMaxSeqnoTimePairsPerCF;
+    seqno_time_cadence = (min_preserve_seconds + kMaxSeqnoTimePairsPerCF - 1) /
+                         kMaxSeqnoTimePairsPerCF;
   }
 
   TEST_SYNC_POINT_CALLBACK(
@@ -866,7 +889,7 @@ Status DBImpl::RegisterRecordSeqnoTimeWorker(bool is_new_db) {
       assert(mapping_was_empty);
 
       // We can simply modify these, before writes are allowed
-      constexpr uint64_t kMax = SeqnoToTimeMapping::kMaxSeqnoTimePairsPerSST;
+      constexpr uint64_t kMax = kMaxSeqnoTimePairsPerSST;
       versions_->SetLastAllocatedSequence(kMax);
       versions_->SetLastPublishedSequence(kMax);
       versions_->SetLastSequence(kMax);
@@ -879,7 +902,8 @@ Status DBImpl::RegisterRecordSeqnoTimeWorker(bool is_new_db) {
         VersionEdit edit;
         edit.SetLastSequence(kMax);
         s = versions_->LogAndApplyToDefaultColumnFamily(
-            {}, &edit, &mutex_, directories_.GetDbDir());
+            read_options, write_options, &edit, &mutex_,
+            directories_.GetDbDir());
         if (!s.ok() && versions_->io_status().IsIOError()) {
           s = error_handler_.SetBGError(versions_->io_status(),
                                         BackgroundErrorReason::kManifestWrite);
@@ -989,6 +1013,7 @@ void DBImpl::PersistStats() {
     stats_slice_initialized_ = true;
     std::swap(stats_slice_, stats_map);
     if (s.ok()) {
+      // TODO: plumb Env::IOActivity, Env::IOPriority
       WriteOptions wo;
       wo.low_pri = true;
       wo.no_slowdown = true;
@@ -1203,8 +1228,10 @@ FSDirectory* DBImpl::GetDataDir(ColumnFamilyData* cfd, size_t path_id) const {
 Status DBImpl::SetOptions(
     ColumnFamilyHandle* column_family,
     const std::unordered_map<std::string, std::string>& options_map) {
-  // TODO: plumb Env::IOActivity
+  // TODO: plumb Env::IOActivity, Env::IOPriority
   const ReadOptions read_options;
+  const WriteOptions write_options;
+
   auto* cfd =
       static_cast_with_check<ColumnFamilyHandleImpl>(column_family)->cfd();
   if (options_map.empty()) {
@@ -1227,14 +1254,15 @@ Status DBImpl::SetOptions(
       new_options = *cfd->GetLatestMutableCFOptions();
       // Append new version to recompute compaction score.
       VersionEdit dummy_edit;
-      s = versions_->LogAndApply(cfd, new_options, read_options, &dummy_edit,
-                                 &mutex_, directories_.GetDbDir());
+      s = versions_->LogAndApply(cfd, new_options, read_options, write_options,
+                                 &dummy_edit, &mutex_, directories_.GetDbDir());
       // Trigger possible flush/compactions. This has to be before we persist
       // options to file, otherwise there will be a deadlock with writer
       // thread.
       InstallSuperVersionAndScheduleWork(cfd, &sv_context, new_options);
 
-      persist_options_status = WriteOptionsFile(true /*db_mutex_already_held*/);
+      persist_options_status =
+          WriteOptionsFile(write_options, true /*db_mutex_already_held*/);
       bg_cv_.SignalAll();
     }
   }
@@ -1328,17 +1356,24 @@ Status DBImpl::SetDBOptions(
       const bool max_compactions_increased =
           new_bg_job_limits.max_compactions >
           current_bg_job_limits.max_compactions;
+      const bool offpeak_time_changed =
+          versions_->offpeak_time_option().daily_offpeak_time_utc !=
+          new_db_options.daily_offpeak_time_utc;
 
-      if (max_flushes_increased || max_compactions_increased) {
+      if (max_flushes_increased || max_compactions_increased ||
+          offpeak_time_changed) {
         if (max_flushes_increased) {
           env_->IncBackgroundThreadsIfNeeded(new_bg_job_limits.max_flushes,
                                              Env::Priority::HIGH);
         }
-
         if (max_compactions_increased) {
           env_->IncBackgroundThreadsIfNeeded(new_bg_job_limits.max_compactions,
                                              Env::Priority::LOW);
         }
+        if (offpeak_time_changed) {
+          versions_->ChangeOffpeakTimeOption(
+              new_db_options.daily_offpeak_time_utc);
+        }
 
         MaybeScheduleFlushOrCompaction();
       }
@@ -1406,7 +1441,8 @@ Status DBImpl::SetDBOptions(
         }
         write_thread_.ExitUnbatched(&w);
       }
-      persist_options_status = WriteOptionsFile(true /*db_mutex_already_held*/);
+      persist_options_status =
+          WriteOptionsFile(WriteOptions(), true /*db_mutex_already_held*/);
     } else {
       // To get here, we must have had invalid options and will not attempt to
       // persist the options, which means the status is "OK/Uninitialized.
@@ -1458,14 +1494,14 @@ int DBImpl::FindMinimumEmptyLevelFitting(
   return minimum_level;
 }
 
-Status DBImpl::FlushWAL(bool sync) {
+Status DBImpl::FlushWAL(const WriteOptions& write_options, bool sync) {
   if (manual_wal_flush_) {
     IOStatus io_s;
     {
       // We need to lock log_write_mutex_ since logs_ might change concurrently
       InstrumentedMutexLock wl(&log_write_mutex_);
       log::Writer* cur_log_writer = logs_.back().writer;
-      io_s = cur_log_writer->WriteBuffer();
+      io_s = cur_log_writer->WriteBuffer(write_options);
     }
     if (!io_s.ok()) {
       ROCKS_LOG_ERROR(immutable_db_options_.info_log, "WAL flush error %s",
@@ -1538,11 +1574,22 @@ Status DBImpl::SyncWAL() {
   RecordTick(stats_, WAL_FILE_SYNCED);
   Status status;
   IOStatus io_s;
-  for (log::Writer* log : logs_to_sync) {
-    io_s = log->file()->SyncWithoutFlush(immutable_db_options_.use_fsync);
-    if (!io_s.ok()) {
-      status = io_s;
-      break;
+  // TODO: plumb Env::IOActivity, Env::IOPriority
+  const ReadOptions read_options;
+  const WriteOptions write_options;
+  IOOptions opts;
+  io_s = WritableFileWriter::PrepareIOOptions(write_options, opts);
+  if (!io_s.ok()) {
+    status = io_s;
+  }
+  if (io_s.ok()) {
+    for (log::Writer* log : logs_to_sync) {
+      io_s =
+          log->file()->SyncWithoutFlush(opts, immutable_db_options_.use_fsync);
+      if (!io_s.ok()) {
+        status = io_s;
+        break;
+      }
     }
   }
   if (!io_s.ok()) {
@@ -1571,9 +1618,7 @@ Status DBImpl::SyncWAL() {
   }
   if (status.ok() && synced_wals.IsWalAddition()) {
     InstrumentedMutexLock l(&mutex_);
-    // TODO: plumb Env::IOActivity
-    const ReadOptions read_options;
-    status = ApplyWALToManifest(read_options, &synced_wals);
+    status = ApplyWALToManifest(read_options, write_options, &synced_wals);
   }
 
   TEST_SYNC_POINT("DBImpl::SyncWAL:BeforeMarkLogsSynced:2");
@@ -1582,12 +1627,14 @@ Status DBImpl::SyncWAL() {
 }
 
 Status DBImpl::ApplyWALToManifest(const ReadOptions& read_options,
+                                  const WriteOptions& write_options,
                                   VersionEdit* synced_wals) {
   // not empty, write to MANIFEST.
   mutex_.AssertHeld();
 
   Status status = versions_->LogAndApplyToDefaultColumnFamily(
-      read_options, synced_wals, &mutex_, directories_.GetDbDir());
+      read_options, write_options, synced_wals, &mutex_,
+      directories_.GetDbDir());
   if (!status.ok() && versions_->io_status().IsIOError()) {
     status = error_handler_.SetBGError(versions_->io_status(),
                                        BackgroundErrorReason::kManifestWrite);
@@ -2050,6 +2097,73 @@ Status DBImpl::GetEntity(const ReadOptions& _read_options,
   return GetImpl(read_options, key, get_impl_options);
 }
 
+Status DBImpl::GetEntity(const ReadOptions& _read_options, const Slice& key,
+                         PinnableAttributeGroups* result) {
+  if (!result) {
+    return Status::InvalidArgument(
+        "Cannot call GetEntity without PinnableAttributeGroups object");
+  }
+  Status s;
+  const size_t num_column_families = result->size();
+  if (_read_options.io_activity != Env::IOActivity::kUnknown &&
+      _read_options.io_activity != Env::IOActivity::kGetEntity) {
+    s = Status::InvalidArgument(
+        "Cannot call GetEntity with `ReadOptions::io_activity` != "
+        "`Env::IOActivity::kUnknown` or `Env::IOActivity::kGetEntity`");
+    for (size_t i = 0; i < num_column_families; ++i) {
+      (*result)[i].SetStatus(s);
+    }
+    return s;
+  }
+  // return early if no CF was passed in
+  if (num_column_families == 0) {
+    return s;
+  }
+  ReadOptions read_options(_read_options);
+  if (read_options.io_activity == Env::IOActivity::kUnknown) {
+    read_options.io_activity = Env::IOActivity::kGetEntity;
+  }
+  std::vector<Slice> keys;
+  std::vector<ColumnFamilyHandle*> column_families;
+  for (size_t i = 0; i < num_column_families; ++i) {
+    // If any of the CFH is null, break early since the entire query will fail
+    if (!(*result)[i].column_family()) {
+      s = Status::InvalidArgument(
+          "DB failed to query because one or more group(s) have null column "
+          "family handle");
+      (*result)[i].SetStatus(
+          Status::InvalidArgument("Column family handle cannot be null"));
+      break;
+    }
+    // Adding the same key slice for different CFs
+    keys.emplace_back(key);
+    column_families.emplace_back((*result)[i].column_family());
+  }
+  if (!s.ok()) {
+    for (size_t i = 0; i < num_column_families; ++i) {
+      if ((*result)[i].status().ok()) {
+        (*result)[i].SetStatus(
+            Status::Incomplete("DB not queried due to invalid argument(s) in "
+                               "one or more of the attribute groups"));
+      }
+    }
+    return s;
+  }
+  std::vector<PinnableWideColumns> columns(num_column_families);
+  std::vector<Status> statuses(num_column_families);
+  MultiGetCommon(
+      read_options, num_column_families, column_families.data(), keys.data(),
+      /* values */ nullptr, columns.data(),
+      /* timestamps */ nullptr, statuses.data(), /* sorted_input */ false);
+  // Set results
+  for (size_t i = 0; i < num_column_families; ++i) {
+    (*result)[i].Reset();
+    (*result)[i].SetStatus(statuses[i]);
+    (*result)[i].SetColumns(std::move(columns[i]));
+  }
+  return s;
+}
+
 bool DBImpl::ShouldReferenceSuperVersion(const MergeContext& merge_context) {
   // If both thresholds are reached, a function returning merge operands as
   // `PinnableSlice`s should reference the `SuperVersion` to avoid large and/or
@@ -2829,7 +2943,6 @@ void DBImpl::MultiGetCommon(const ReadOptions& read_options,
   bool should_fail = false;
   for (size_t i = 0; i < num_keys; ++i) {
     ColumnFamilyHandle* cfh = column_families[i];
-    assert(cfh);
     if (read_options.timestamp) {
       statuses[i] = FailIfTsMismatchCf(cfh, *(read_options.timestamp));
       if (!statuses[i].ok()) {
@@ -3402,6 +3515,7 @@ void DBImpl::MultiGetEntity(const ReadOptions& _read_options, size_t num_keys,
 }
 
 Status DBImpl::WrapUpCreateColumnFamilies(
+    const ReadOptions& read_options, const WriteOptions& write_options,
     const std::vector<const ColumnFamilyOptions*>& cf_options) {
   // NOTE: this function is skipped for create_missing_column_families and
   // DB::Open, so new functionality here might need to go into Open also.
@@ -3414,26 +3528,32 @@ Status DBImpl::WrapUpCreateColumnFamilies(
     }
   }
   // Attempt both follow-up actions even if one fails
-  Status s = WriteOptionsFile(false /*db_mutex_already_held*/);
+  Status s = WriteOptionsFile(write_options, false /*db_mutex_already_held*/);
   if (register_worker) {
-    s.UpdateIfOk(RegisterRecordSeqnoTimeWorker(/*from_db_open=*/false));
+    s.UpdateIfOk(RegisterRecordSeqnoTimeWorker(read_options, write_options,
+                                               /* is_new_db */ false));
   }
   return s;
 }
 
-Status DBImpl::CreateColumnFamily(const ColumnFamilyOptions& cf_options,
+Status DBImpl::CreateColumnFamily(const ReadOptions& read_options,
+                                  const WriteOptions& write_options,
+                                  const ColumnFamilyOptions& cf_options,
                                   const std::string& column_family,
                                   ColumnFamilyHandle** handle) {
   assert(handle != nullptr);
   InstrumentedMutexLock ol(&options_mutex_);
-  Status s = CreateColumnFamilyImpl(cf_options, column_family, handle);
+  Status s = CreateColumnFamilyImpl(read_options, write_options, cf_options,
+                                    column_family, handle);
   if (s.ok()) {
-    s.UpdateIfOk(WrapUpCreateColumnFamilies({&cf_options}));
+    s.UpdateIfOk(
+        WrapUpCreateColumnFamilies(read_options, write_options, {&cf_options}));
   }
   return s;
 }
 
 Status DBImpl::CreateColumnFamilies(
+    const ReadOptions& read_options, const WriteOptions& write_options,
     const ColumnFamilyOptions& cf_options,
     const std::vector<std::string>& column_family_names,
     std::vector<ColumnFamilyHandle*>* handles) {
@@ -3445,7 +3565,8 @@ Status DBImpl::CreateColumnFamilies(
   bool success_once = false;
   for (size_t i = 0; i < num_cf; i++) {
     ColumnFamilyHandle* handle;
-    s = CreateColumnFamilyImpl(cf_options, column_family_names[i], &handle);
+    s = CreateColumnFamilyImpl(read_options, write_options, cf_options,
+                               column_family_names[i], &handle);
     if (!s.ok()) {
       break;
     }
@@ -3453,12 +3574,14 @@ Status DBImpl::CreateColumnFamilies(
     success_once = true;
   }
   if (success_once) {
-    s.UpdateIfOk(WrapUpCreateColumnFamilies({&cf_options}));
+    s.UpdateIfOk(
+        WrapUpCreateColumnFamilies(read_options, write_options, {&cf_options}));
   }
   return s;
 }
 
 Status DBImpl::CreateColumnFamilies(
+    const ReadOptions& read_options, const WriteOptions& write_options,
     const std::vector<ColumnFamilyDescriptor>& column_families,
     std::vector<ColumnFamilyHandle*>* handles) {
   assert(handles != nullptr);
@@ -3471,7 +3594,8 @@ Status DBImpl::CreateColumnFamilies(
   cf_opts.reserve(num_cf);
   for (size_t i = 0; i < num_cf; i++) {
     ColumnFamilyHandle* handle;
-    s = CreateColumnFamilyImpl(column_families[i].options,
+    s = CreateColumnFamilyImpl(read_options, write_options,
+                               column_families[i].options,
                                column_families[i].name, &handle);
     if (!s.ok()) {
       break;
@@ -3481,17 +3605,18 @@ Status DBImpl::CreateColumnFamilies(
     cf_opts.push_back(&column_families[i].options);
   }
   if (success_once) {
-    s.UpdateIfOk(WrapUpCreateColumnFamilies(cf_opts));
+    s.UpdateIfOk(
+        WrapUpCreateColumnFamilies(read_options, write_options, cf_opts));
   }
   return s;
 }
 
-Status DBImpl::CreateColumnFamilyImpl(const ColumnFamilyOptions& cf_options,
+Status DBImpl::CreateColumnFamilyImpl(const ReadOptions& read_options,
+                                      const WriteOptions& write_options,
+                                      const ColumnFamilyOptions& cf_options,
                                       const std::string& column_family_name,
                                       ColumnFamilyHandle** handle) {
   options_mutex_.AssertHeld();
-  // TODO: plumb Env::IOActivity
-  const ReadOptions read_options;
   Status s;
   *handle = nullptr;
 
@@ -3535,7 +3660,7 @@ Status DBImpl::CreateColumnFamilyImpl(const ColumnFamilyOptions& cf_options,
       // LogAndApply will both write the creation in MANIFEST and create
       // ColumnFamilyData object
       s = versions_->LogAndApply(nullptr, MutableCFOptions(cf_options),
-                                 read_options, &edit, &mutex_,
+                                 read_options, write_options, &edit, &mutex_,
                                  directories_.GetDbDir(), false, &cf_options);
       write_thread_.ExitUnbatched(&w);
     }
@@ -3584,7 +3709,8 @@ Status DBImpl::DropColumnFamily(ColumnFamilyHandle* column_family) {
   InstrumentedMutexLock ol(&options_mutex_);
   Status s = DropColumnFamilyImpl(column_family);
   if (s.ok()) {
-    s = WriteOptionsFile(false /*db_mutex_already_held*/);
+    // TODO: plumb Env::IOActivity, Env::IOPriority
+    s = WriteOptionsFile(WriteOptions(), false /*db_mutex_already_held*/);
   }
   return s;
 }
@@ -3602,8 +3728,9 @@ Status DBImpl::DropColumnFamilies(
     success_once = true;
   }
   if (success_once) {
+    // TODO: plumb Env::IOActivity, Env::IOPriority
     Status persist_options_status =
-        WriteOptionsFile(false /*db_mutex_already_held*/);
+        WriteOptionsFile(WriteOptions(), false /*db_mutex_already_held*/);
     if (s.ok() && !persist_options_status.ok()) {
       s = persist_options_status;
     }
@@ -3612,8 +3739,10 @@ Status DBImpl::DropColumnFamilies(
 }
 
 Status DBImpl::DropColumnFamilyImpl(ColumnFamilyHandle* column_family) {
-  // TODO: plumb Env::IOActivity
+  // TODO: plumb Env::IOActivity, Env::IOPriority
   const ReadOptions read_options;
+  const WriteOptions write_options;
+
   auto cfh = static_cast_with_check<ColumnFamilyHandleImpl>(column_family);
   auto cfd = cfh->cfd();
   if (cfd->GetID() == 0) {
@@ -3637,7 +3766,7 @@ Status DBImpl::DropColumnFamilyImpl(ColumnFamilyHandle* column_family) {
       WriteThread::Writer w;
       write_thread_.EnterUnbatched(&w, &mutex_);
       s = versions_->LogAndApply(cfd, *cfd->GetLatestMutableCFOptions(),
-                                 read_options, &edit, &mutex_,
+                                 read_options, write_options, &edit, &mutex_,
                                  directories_.GetDbDir());
       write_thread_.ExitUnbatched(&w);
     }
@@ -3664,7 +3793,8 @@ Status DBImpl::DropColumnFamilyImpl(ColumnFamilyHandle* column_family) {
 
   if (cfd->ioptions()->preserve_internal_time_seconds > 0 ||
       cfd->ioptions()->preclude_last_level_data_seconds > 0) {
-    s = RegisterRecordSeqnoTimeWorker(/*from_db_open=*/false);
+    s = RegisterRecordSeqnoTimeWorker(read_options, write_options,
+                                      /* is_new_db */ false);
   }
 
   if (s.ok()) {
@@ -3695,7 +3825,7 @@ bool DBImpl::KeyMayExist(const ReadOptions& read_options,
     // falsify later if key-may-exist but can't fetch value
     *value_found = true;
   }
-  // TODO: plumb Env::IOActivity
+  // TODO: plumb Env::IOActivity, Env::IOPriority
   ReadOptions roptions = read_options;
   roptions.read_tier = kBlockCacheTier;  // read from block cache only
   PinnableSlice pinnable_val;
@@ -4214,7 +4344,7 @@ Status DBImpl::GetPropertiesOfAllTables(ColumnFamilyHandle* column_family,
   version->Ref();
   mutex_.Unlock();
 
-  // TODO: plumb Env::IOActivity
+  // TODO: plumb Env::IOActivity, Env::IOPriority
   const ReadOptions read_options;
   auto s = version->GetPropertiesOfAllTables(read_options, props);
 
@@ -4238,7 +4368,7 @@ Status DBImpl::GetPropertiesOfTablesInRange(ColumnFamilyHandle* column_family,
   version->Ref();
   mutex_.Unlock();
 
-  // TODO: plumb Env::IOActivity
+  // TODO: plumb Env::IOActivity, Env::IOPriority
   const ReadOptions read_options;
   auto s = version->GetPropertiesOfTablesInRange(read_options, range, n, props);
 
@@ -4580,7 +4710,7 @@ Status DBImpl::GetApproximateSizes(const SizeApproximationOptions& options,
   SuperVersion* sv = GetAndRefSuperVersion(cfd);
   v = sv->current;
 
-  // TODO: plumb Env::IOActivity
+  // TODO: plumb Env::IOActivity, Env::IOPriority
   const ReadOptions read_options;
   for (int i = 0; i < n; i++) {
     // Add timestamp if needed
@@ -4644,8 +4774,10 @@ Status DBImpl::GetUpdatesSince(
 }
 
 Status DBImpl::DeleteFile(std::string name) {
-  // TODO: plumb Env::IOActivity
+  // TODO: plumb Env::IOActivity, Env::IOPriority
   const ReadOptions read_options;
+  const WriteOptions write_options;
+
   uint64_t number;
   FileType type;
   WalFileType log_type;
@@ -4725,7 +4857,7 @@ Status DBImpl::DeleteFile(std::string name) {
     edit.SetColumnFamily(cfd->GetID());
     edit.DeleteFile(level, number);
     status = versions_->LogAndApply(cfd, *cfd->GetLatestMutableCFOptions(),
-                                    read_options, &edit, &mutex_,
+                                    read_options, write_options, &edit, &mutex_,
                                     directories_.GetDbDir());
     if (status.ok()) {
       InstallSuperVersionAndScheduleWork(cfd,
@@ -4748,8 +4880,10 @@ Status DBImpl::DeleteFile(std::string name) {
 Status DBImpl::DeleteFilesInRanges(ColumnFamilyHandle* column_family,
                                    const RangePtr* ranges, size_t n,
                                    bool include_end) {
-  // TODO: plumb Env::IOActivity
+  // TODO: plumb Env::IOActivity, Env::IOPriority
   const ReadOptions read_options;
+  const WriteOptions write_options;
+
   Status status = Status::OK();
   auto cfh = static_cast_with_check<ColumnFamilyHandleImpl>(column_family);
   ColumnFamilyData* cfd = cfh->cfd();
@@ -4817,7 +4951,7 @@ Status DBImpl::DeleteFilesInRanges(ColumnFamilyHandle* column_family,
     }
     input_version->Ref();
     status = versions_->LogAndApply(cfd, *cfd->GetLatestMutableCFOptions(),
-                                    read_options, &edit, &mutex_,
+                                    read_options, write_options, &edit, &mutex_,
                                     directories_.GetDbDir());
     if (status.ok()) {
       InstallSuperVersionAndScheduleWork(cfd,
@@ -5231,7 +5365,8 @@ Status DestroyDB(const std::string& dbname, const Options& options,
   return result;
 }
 
-Status DBImpl::WriteOptionsFile(bool db_mutex_already_held) {
+Status DBImpl::WriteOptionsFile(const WriteOptions& write_options,
+                                bool db_mutex_already_held) {
   options_mutex_.AssertHeld();
 
   if (db_mutex_already_held) {
@@ -5265,8 +5400,8 @@ Status DBImpl::WriteOptionsFile(bool db_mutex_already_held) {
 
   std::string file_name =
       TempOptionsFileName(GetName(), versions_->NewFileNumber());
-  Status s = PersistRocksDBOptions(db_options, cf_names, cf_opts, file_name,
-                                   fs_.get());
+  Status s = PersistRocksDBOptions(write_options, db_options, cf_names, cf_opts,
+                                   file_name, fs_.get());
 
   if (s.ok()) {
     s = RenameTempFileToOptionsFile(file_name);
@@ -5459,7 +5594,7 @@ Status DBImpl::GetLatestSequenceForKey(
   MergeContext merge_context;
   SequenceNumber max_covering_tombstone_seq = 0;
 
-  // TODO: plumb Env::IOActivity
+  // TODO: plumb Env::IOActivity, Env::IOPriority
   ReadOptions read_options;
   SequenceNumber current_seq = versions_->LastSequence();
 
@@ -5615,8 +5750,10 @@ Status DBImpl::IngestExternalFile(
 
 Status DBImpl::IngestExternalFiles(
     const std::vector<IngestExternalFileArg>& args) {
-  // TODO: plumb Env::IOActivity
+  // TODO: plumb Env::IOActivity, Env::IOPriority
   const ReadOptions read_options;
+  const WriteOptions write_options;
+
   if (args.empty()) {
     return Status::InvalidArgument("ingestion arg list is empty");
   }
@@ -5834,9 +5971,10 @@ Status DBImpl::IngestExternalFiles(
         }
         assert(0 == num_entries);
       }
-      status = versions_->LogAndApply(cfds_to_commit, mutable_cf_options_list,
-                                      read_options, edit_lists, &mutex_,
-                                      directories_.GetDbDir());
+      status = versions_->LogAndApply(
+          cfds_to_commit, mutable_cf_options_list, read_options, write_options,
+
+          edit_lists, &mutex_, directories_.GetDbDir());
       // It is safe to update VersionSet last seqno here after LogAndApply since
       // LogAndApply persists last sequence number from VersionEdits,
       // which are from file's largest seqno and not from VersionSet.
@@ -5938,8 +6076,10 @@ Status DBImpl::CreateColumnFamilyWithImport(
     ColumnFamilyHandle** handle) {
   assert(handle != nullptr);
   assert(*handle == nullptr);
-  // TODO: plumb Env::IOActivity
+  // TODO: plumb Env::IOActivity, Env::IOPriority
   const ReadOptions read_options;
+  const WriteOptions write_options;
+
   std::string cf_comparator_name = options.comparator->Name();
 
   size_t total_file_num = 0;
@@ -5955,7 +6095,8 @@ Status DBImpl::CreateColumnFamilyWithImport(
   }
 
   // Create column family.
-  auto status = CreateColumnFamily(options, column_family_name, handle);
+  auto status = CreateColumnFamily(read_options, write_options, options,
+                                   column_family_name, handle);
   if (!status.ok()) {
     return status;
   }
@@ -5991,8 +6132,8 @@ Status DBImpl::CreateColumnFamilyWithImport(
       next_file_number = versions_->FetchAddFileNumber(total_file_num);
       auto cf_options = cfd->GetLatestMutableCFOptions();
       status =
-          versions_->LogAndApply(cfd, *cf_options, read_options, &dummy_edit,
-                                 &mutex_, directories_.GetDbDir());
+          versions_->LogAndApply(cfd, *cf_options, read_options, write_options,
+                                 &dummy_edit, &mutex_, directories_.GetDbDir());
       if (status.ok()) {
         InstallSuperVersionAndScheduleWork(cfd, &dummy_sv_ctx, *cf_options);
       }
@@ -6029,8 +6170,8 @@ Status DBImpl::CreateColumnFamilyWithImport(
       if (status.ok()) {
         auto cf_options = cfd->GetLatestMutableCFOptions();
         status = versions_->LogAndApply(cfd, *cf_options, read_options,
-                                        import_job.edit(), &mutex_,
-                                        directories_.GetDbDir());
+                                        write_options, import_job.edit(),
+                                        &mutex_, directories_.GetDbDir());
         if (status.ok()) {
           InstallSuperVersionAndScheduleWork(cfd, &sv_context, *cf_options);
         }
@@ -6114,6 +6255,7 @@ Status DBImpl::ClipColumnFamily(ColumnFamilyHandle* column_family,
       empty_after_delete = true;
     } else {
       const Comparator* const ucmp = column_family->GetComparator();
+      // TODO: plumb Env::IOActivity, Env::IOPriority
       WriteOptions wo;
       // Delete [smallest_user_key, clip_begin_key)
       if (ucmp->Compare(smallest_user_key, begin_key) < 0) {
@@ -6143,7 +6285,7 @@ Status DBImpl::ClipColumnFamily(ColumnFamilyHandle* column_family,
     // last level to compact to and that range tombstones are not dropped
     // during non-bottommost compactions, calling CompactRange() on these two
     // ranges may not clear all range tombstones.
-    status = CompactRange(compact_options, nullptr, nullptr);
+    status = CompactRange(compact_options, column_family, nullptr, nullptr);
   }
   return status;
 }
@@ -6434,8 +6576,10 @@ Status DBImpl::ReserveFileNumbersBeforeIngestion(
     ColumnFamilyData* cfd, uint64_t num,
     std::unique_ptr<std::list<uint64_t>::iterator>& pending_output_elem,
     uint64_t* next_file_number) {
-  // TODO: plumb Env::IOActivity
+  // TODO: plumb Env::IOActivity, Env::IOPriority
   const ReadOptions read_options;
+  const WriteOptions write_options;
+
   Status s;
   SuperVersionContext dummy_sv_ctx(true /* create_superversion */);
   assert(nullptr != next_file_number);
@@ -6453,8 +6597,8 @@ Status DBImpl::ReserveFileNumbersBeforeIngestion(
   // reuse the file number that has already assigned to the internal file,
   // and this will overwrite the external file. To protect the external
   // file, we have to make sure the file number will never being reused.
-  s = versions_->LogAndApply(cfd, *cf_options, read_options, &dummy_edit,
-                             &mutex_, directories_.GetDbDir());
+  s = versions_->LogAndApply(cfd, *cf_options, read_options, write_options,
+                             &dummy_edit, &mutex_, directories_.GetDbDir());
   if (s.ok()) {
     InstallSuperVersionAndScheduleWork(cfd, &dummy_sv_ctx, *cf_options);
   }
@@ -6500,28 +6644,24 @@ void DBImpl::RecordSeqnoToTimeMapping(uint64_t populate_historical_seconds) {
   immutable_db_options_.clock->GetCurrentTime(&unix_time_signed)
       .PermitUncheckedError();  // Ignore error
   uint64_t unix_time = static_cast<uint64_t>(unix_time_signed);
-  bool appended = false;
-  {
-    InstrumentedMutexLock l(&mutex_);
-    if (populate_historical_seconds > 0) {
+  if (populate_historical_seconds > 0) {
+    bool success = true;
+    {
+      InstrumentedMutexLock l(&mutex_);
       if (seqno > 1 && unix_time > populate_historical_seconds) {
         // seqno=0 is reserved
         SequenceNumber from_seqno = 1;
-        appended = seqno_to_time_mapping_.PrePopulate(
+        success = seqno_to_time_mapping_.PrePopulate(
             from_seqno, seqno, unix_time - populate_historical_seconds,
             unix_time);
       } else {
         // One of these will fail
         assert(seqno > 1);
         assert(unix_time > populate_historical_seconds);
+        success = false;
       }
-    } else {
-      // FIXME: assert(seqno > 0);
-      appended = seqno_to_time_mapping_.Append(seqno, unix_time);
     }
-  }
-  if (populate_historical_seconds > 0) {
-    if (appended) {
+    if (success) {
       ROCKS_LOG_INFO(
           immutable_db_options_.info_log,
           "Pre-populated sequence number to time entries: [1,%" PRIu64
@@ -6534,11 +6674,11 @@ void DBImpl::RecordSeqnoToTimeMapping(uint64_t populate_historical_seconds) {
           "] -> [%" PRIu64 ",%" PRIu64 "]",
           seqno, unix_time - populate_historical_seconds, unix_time);
     }
-  } else if (!appended) {
-    ROCKS_LOG_WARN(immutable_db_options_.info_log,
-                   "Failed to insert sequence number to time entry: %" PRIu64
-                   " -> %" PRIu64,
-                   seqno, unix_time);
+  } else {
+    InstrumentedMutexLock l(&mutex_);
+    // FIXME: assert(seqno > 0);
+    // Always successful assuming seqno never go backwards
+    seqno_to_time_mapping_.Append(seqno, unix_time);
   }
 }
 

db/db_impl/db_impl.h

@@ -197,6 +197,8 @@ class DBImpl : public DB {
   Status PutEntity(const WriteOptions& options,
                    ColumnFamilyHandle* column_family, const Slice& key,
                    const WideColumns& columns) override;
+  Status PutEntity(const WriteOptions& options, const Slice& key,
+                   const AttributeGroups& attribute_groups) override;
 
   using DB::Merge;
   Status Merge(const WriteOptions& options, ColumnFamilyHandle* column_family,
@@ -242,6 +244,8 @@ class DBImpl : public DB {
   Status GetEntity(const ReadOptions& options,
                    ColumnFamilyHandle* column_family, const Slice& key,
                    PinnableWideColumns* columns) override;
+  Status GetEntity(const ReadOptions& options, const Slice& key,
+                   PinnableAttributeGroups* result) override;
 
   using DB::GetMergeOperands;
   Status GetMergeOperands(const ReadOptions& options,
@@ -317,14 +321,41 @@ class DBImpl : public DB {
 
   virtual Status CreateColumnFamily(const ColumnFamilyOptions& cf_options,
                                     const std::string& column_family,
-                                    ColumnFamilyHandle** handle) override;
+                                    ColumnFamilyHandle** handle) override {
+    // TODO: plumb Env::IOActivity, Env::IOPriority
+    return CreateColumnFamily(ReadOptions(), WriteOptions(), cf_options,
+                              column_family, handle);
+  }
+  virtual Status CreateColumnFamily(const ReadOptions& read_options,
+                                    const WriteOptions& write_options,
+                                    const ColumnFamilyOptions& cf_options,
+                                    const std::string& column_family,
+                                    ColumnFamilyHandle** handle);
   virtual Status CreateColumnFamilies(
       const ColumnFamilyOptions& cf_options,
       const std::vector<std::string>& column_family_names,
-      std::vector<ColumnFamilyHandle*>* handles) override;
+      std::vector<ColumnFamilyHandle*>* handles) override {
+    // TODO: plumb Env::IOActivity, Env::IOPriority
+    return CreateColumnFamilies(ReadOptions(), WriteOptions(), cf_options,
+                                column_family_names, handles);
+  }
+  virtual Status CreateColumnFamilies(
+      const ReadOptions& read_options, const WriteOptions& write_options,
+      const ColumnFamilyOptions& cf_options,
+      const std::vector<std::string>& column_family_names,
+      std::vector<ColumnFamilyHandle*>* handles);
+
   virtual Status CreateColumnFamilies(
       const std::vector<ColumnFamilyDescriptor>& column_families,
-      std::vector<ColumnFamilyHandle*>* handles) override;
+      std::vector<ColumnFamilyHandle*>* handles) override {
+    // TODO: plumb Env::IOActivity, Env::IOPriority
+    return CreateColumnFamilies(ReadOptions(), WriteOptions(), column_families,
+                                handles);
+  }
+  virtual Status CreateColumnFamilies(
+      const ReadOptions& read_options, const WriteOptions& write_options,
+      const std::vector<ColumnFamilyDescriptor>& column_families,
+      std::vector<ColumnFamilyHandle*>* handles);
   virtual Status DropColumnFamily(ColumnFamilyHandle* column_family) override;
   virtual Status DropColumnFamilies(
       const std::vector<ColumnFamilyHandle*>& column_families) override;
@@ -436,7 +467,12 @@ class DBImpl : public DB {
   virtual Status Flush(
       const FlushOptions& options,
       const std::vector<ColumnFamilyHandle*>& column_families) override;
-  virtual Status FlushWAL(bool sync) override;
+  virtual Status FlushWAL(bool sync) override {
+    // TODO: plumb Env::IOActivity, Env::IOPriority
+    return FlushWAL(WriteOptions(), sync);
+  }
+
+  virtual Status FlushWAL(const WriteOptions& write_options, bool sync);
   bool WALBufferIsEmpty();
   virtual Status SyncWAL() override;
   virtual Status LockWAL() override;
@@ -1186,6 +1222,7 @@ class DBImpl : public DB {
   size_t TEST_GetWalPreallocateBlockSize(uint64_t write_buffer_size) const;
   void TEST_WaitForPeriodicTaskRun(std::function<void()> callback) const;
   SeqnoToTimeMapping TEST_GetSeqnoToTimeMapping() const;
+  const autovector<uint64_t>& TEST_GetFilesToQuarantine() const;
   size_t TEST_EstimateInMemoryStatsHistorySize() const;
 
   uint64_t TEST_GetCurrentLogNumber() const {
@@ -1393,14 +1430,16 @@ class DBImpl : public DB {
     autovector<ColumnFamilyData*> cfds_;
     autovector<const MutableCFOptions*> mutable_cf_opts_;
     autovector<autovector<VersionEdit*>> edit_lists_;
-    // files_to_delete_ contains sst files
-    std::unordered_set<std::string> files_to_delete_;
+    // Stale SST files to delete found upon recovery. This stores a mapping from
+    // such a file's absolute path to its parent directory.
+    std::unordered_map<std::string, std::string> files_to_delete_;
     bool is_new_db_ = false;
   };
 
   // Persist options to options file. Must be holding options_mutex_.
   // Will lock DB mutex if !db_mutex_already_held.
-  Status WriteOptionsFile(bool db_mutex_already_held);
+  Status WriteOptionsFile(const WriteOptions& write_options,
+                          bool db_mutex_already_held);
 
   Status CompactRangeInternal(const CompactRangeOptions& options,
                               ColumnFamilyHandle* column_family,
@@ -1526,7 +1565,8 @@ class DBImpl : public DB {
   virtual bool OwnTablesAndLogs() const { return true; }
 
   // Setup DB identity file, and write DB ID to manifest if necessary.
-  Status SetupDBId(bool read_only, RecoveryContext* recovery_ctx);
+  Status SetupDBId(const WriteOptions& write_options, bool read_only,
+                   RecoveryContext* recovery_ctx);
   // Assign db_id_ and write DB ID to manifest if necessary.
   void SetDBId(std::string&& id, bool read_only, RecoveryContext* recovery_ctx);
 
@@ -1653,7 +1693,8 @@ class DBImpl : public DB {
       return w;
     }
     Status ClearWriter() {
-      Status s = writer->WriteBuffer();
+      // TODO: plumb Env::IOActivity, Env::IOPriority
+      Status s = writer->WriteBuffer(WriteOptions());
       delete writer;
       writer = nullptr;
       return s;
@@ -1829,12 +1870,15 @@ class DBImpl : public DB {
   const Status CreateArchivalDirectory();
 
   // Create a column family, without some of the follow-up work yet
-  Status CreateColumnFamilyImpl(const ColumnFamilyOptions& cf_options,
+  Status CreateColumnFamilyImpl(const ReadOptions& read_options,
+                                const WriteOptions& write_options,
+                                const ColumnFamilyOptions& cf_options,
                                 const std::string& cf_name,
                                 ColumnFamilyHandle** handle);
 
   // Follow-up work to user creating a column family or (families)
   Status WrapUpCreateColumnFamilies(
+      const ReadOptions& read_options, const WriteOptions& write_options,
       const std::vector<const ColumnFamilyOptions*>& cf_options);
 
   Status DropColumnFamilyImpl(ColumnFamilyHandle* column_family);
@@ -1866,7 +1910,8 @@ class DBImpl : public DB {
   void ReleaseFileNumberFromPendingOutputs(
       std::unique_ptr<std::list<uint64_t>::iterator>& v);
 
-  IOStatus SyncClosedLogs(JobContext* job_context, VersionEdit* synced_wals,
+  IOStatus SyncClosedLogs(const WriteOptions& write_options,
+                          JobContext* job_context, VersionEdit* synced_wals,
                           bool error_recovery_in_prog);
 
   // Flush the in-memory write buffer to storage.  Switches to a new
@@ -2052,12 +2097,10 @@ class DBImpl : public DB {
                     WriteBatch* tmp_batch, WriteBatch** merged_batch,
                     size_t* write_with_wal, WriteBatch** to_be_cached_state);
 
-  // rate_limiter_priority is used to charge `DBOptions::rate_limiter`
-  // for automatic WAL flush (`Options::manual_wal_flush` == false)
-  // associated with this WriteToWAL
-  IOStatus WriteToWAL(const WriteBatch& merged_batch, log::Writer* log_writer,
-                      uint64_t* log_used, uint64_t* log_size,
-                      Env::IOPriority rate_limiter_priority,
+  IOStatus WriteToWAL(const WriteBatch& merged_batch,
+                      const WriteOptions& write_options,
+                      log::Writer* log_writer, uint64_t* log_used,
+                      uint64_t* log_size,
                       LogFileNumberSize& log_file_number_size);
 
   IOStatus WriteToWAL(const WriteThread::WriteGroup& write_group,
@@ -2169,7 +2212,9 @@ class DBImpl : public DB {
   // Cancel scheduled periodic tasks
   Status CancelPeriodicTaskScheduler();
 
-  Status RegisterRecordSeqnoTimeWorker(bool is_new_db);
+  Status RegisterRecordSeqnoTimeWorker(const ReadOptions& read_options,
+                                       const WriteOptions& write_options,
+                                       bool is_new_db);
 
   void PrintStatistics();
 
@@ -2197,7 +2242,9 @@ class DBImpl : public DB {
 
   // helper function to call after some of the logs_ were synced
   void MarkLogsSynced(uint64_t up_to, bool synced_dir, VersionEdit* edit);
-  Status ApplyWALToManifest(const ReadOptions& read_options, VersionEdit* edit);
+  Status ApplyWALToManifest(const ReadOptions& read_options,
+                            const WriteOptions& write_options,
+                            VersionEdit* edit);
   // WALs with log number up to up_to are not synced successfully.
   void MarkLogsNotSynced(uint64_t up_to);
 
@@ -2269,8 +2316,9 @@ class DBImpl : public DB {
   size_t GetWalPreallocateBlockSize(uint64_t write_buffer_size) const;
   Env::WriteLifeTimeHint CalculateWALWriteHint() { return Env::WLTH_SHORT; }
 
-  IOStatus CreateWAL(uint64_t log_file_num, uint64_t recycle_log_number,
-                     size_t preallocate_block_size, log::Writer** new_log);
+  IOStatus CreateWAL(const WriteOptions& write_options, uint64_t log_file_num,
+                     uint64_t recycle_log_number, size_t preallocate_block_size,
+                     log::Writer** new_log);
 
   // Validate self-consistency of DB options
   static Status ValidateOptions(const DBOptions& db_options);
@@ -2375,10 +2423,6 @@ class DBImpl : public DB {
 
   Status DisableFileDeletionsWithLock();
 
-  // Safely decrease `disable_delete_obsolete_files_` by one while holding lock
-  // and return its remaning value.
-  int EnableFileDeletionsWithLock();
-
   Status IncreaseFullHistoryTsLowImpl(ColumnFamilyData* cfd,
                                       std::string ts_low);
 
@@ -2868,7 +2912,9 @@ static void ClipToRange(T* ptr, V minvalue, V maxvalue) {
 
 inline Status DBImpl::FailIfCfHasTs(
     const ColumnFamilyHandle* column_family) const {
-  column_family = column_family ? column_family : DefaultColumnFamily();
+  if (!column_family) {
+    return Status::InvalidArgument("column family handle cannot be null");
+  }
   assert(column_family);
   const Comparator* const ucmp = column_family->GetComparator();
   assert(ucmp);

db/db_impl/db_impl_compaction_flush.cc

@@ -19,6 +19,10 @@
 #include "monitoring/perf_context_imp.h"
 #include "monitoring/thread_status_updater.h"
 #include "monitoring/thread_status_util.h"
+#include "rocksdb/file_system.h"
+#include "rocksdb/io_status.h"
+#include "rocksdb/options.h"
+#include "rocksdb/table.h"
 #include "test_util/sync_point.h"
 #include "util/cast_util.h"
 #include "util/coding.h"
@@ -112,7 +116,8 @@ bool DBImpl::ShouldRescheduleFlushRequestToRetainUDT(
   return true;
 }
 
-IOStatus DBImpl::SyncClosedLogs(JobContext* job_context,
+IOStatus DBImpl::SyncClosedLogs(const WriteOptions& write_options,
+                                JobContext* job_context,
                                 VersionEdit* synced_wals,
                                 bool error_recovery_in_prog) {
   TEST_SYNC_POINT("DBImpl::SyncClosedLogs:Start");
@@ -143,7 +148,13 @@ IOStatus DBImpl::SyncClosedLogs(JobContext* job_context,
       if (error_recovery_in_prog) {
         log->file()->reset_seen_error();
       }
-      io_s = log->file()->Sync(immutable_db_options_.use_fsync);
+
+      IOOptions io_options;
+      io_s = WritableFileWriter::PrepareIOOptions(write_options, io_options);
+      if (!io_s.ok()) {
+        break;
+      }
+      io_s = log->file()->Sync(io_options, immutable_db_options_.use_fsync);
       if (!io_s.ok()) {
         break;
       }
@@ -152,16 +163,21 @@ IOStatus DBImpl::SyncClosedLogs(JobContext* job_context,
         if (error_recovery_in_prog) {
           log->file()->reset_seen_error();
         }
-        io_s = log->Close();
+        // TODO: plumb Env::IOActivity, Env::IOPriority
+        io_s = log->Close(WriteOptions());
         if (!io_s.ok()) {
           break;
         }
       }
     }
     if (io_s.ok()) {
-      io_s = directories_.GetWalDir()->FsyncWithDirOptions(
-          IOOptions(), nullptr,
-          DirFsyncOptions(DirFsyncOptions::FsyncReason::kNewFileSynced));
+      IOOptions io_options;
+      io_s = WritableFileWriter::PrepareIOOptions(write_options, io_options);
+      if (io_s.ok()) {
+        io_s = directories_.GetWalDir()->FsyncWithDirOptions(
+            io_options, nullptr,
+            DirFsyncOptions(DirFsyncOptions::FsyncReason::kNewFileSynced));
+      }
     }
 
     TEST_SYNC_POINT_CALLBACK("DBImpl::SyncClosedLogs:BeforeReLock",
@@ -199,6 +215,8 @@ Status DBImpl::FlushMemTableToOutputFile(
   assert(cfd->imm()->IsFlushPending());
   assert(versions_);
   assert(versions_->GetColumnFamilySet());
+  const ReadOptions read_options(Env::IOActivity::kFlush);
+  const WriteOptions write_options(Env::IOActivity::kFlush);
   // If there are more than one column families, we need to make sure that
   // all the log files except the most recent one are synced. Otherwise if
   // the host crashes after flushing and before WAL is persistent, the
@@ -265,13 +283,12 @@ Status DBImpl::FlushMemTableToOutputFile(
     VersionEdit synced_wals;
     bool error_recovery_in_prog = error_handler_.IsRecoveryInProgress();
     mutex_.Unlock();
-    log_io_s =
-        SyncClosedLogs(job_context, &synced_wals, error_recovery_in_prog);
+    log_io_s = SyncClosedLogs(write_options, job_context, &synced_wals,
+                              error_recovery_in_prog);
     mutex_.Lock();
     if (log_io_s.ok() && synced_wals.IsWalAddition()) {
-      const ReadOptions read_options(Env::IOActivity::kFlush);
-      log_io_s =
-          status_to_io_status(ApplyWALToManifest(read_options, &synced_wals));
+      log_io_s = status_to_io_status(
+          ApplyWALToManifest(read_options, write_options, &synced_wals));
       TEST_SYNC_POINT_CALLBACK("DBImpl::FlushMemTableToOutputFile:CommitWal:1",
                                nullptr);
     }
@@ -465,6 +482,8 @@ Status DBImpl::AtomicFlushMemTablesToOutputFiles(
     const autovector<BGFlushArg>& bg_flush_args, bool* made_progress,
     JobContext* job_context, LogBuffer* log_buffer, Env::Priority thread_pri) {
   mutex_.AssertHeld();
+  const ReadOptions read_options(Env::IOActivity::kFlush);
+  const WriteOptions write_options(Env::IOActivity::kFlush);
 
   autovector<ColumnFamilyData*> cfds;
   for (const auto& arg : bg_flush_args) {
@@ -552,13 +571,12 @@ Status DBImpl::AtomicFlushMemTablesToOutputFiles(
     VersionEdit synced_wals;
     bool error_recovery_in_prog = error_handler_.IsRecoveryInProgress();
     mutex_.Unlock();
-    log_io_s =
-        SyncClosedLogs(job_context, &synced_wals, error_recovery_in_prog);
+    log_io_s = SyncClosedLogs(write_options, job_context, &synced_wals,
+                              error_recovery_in_prog);
     mutex_.Lock();
     if (log_io_s.ok() && synced_wals.IsWalAddition()) {
-      const ReadOptions read_options(Env::IOActivity::kFlush);
-      log_io_s =
-          status_to_io_status(ApplyWALToManifest(read_options, &synced_wals));
+      log_io_s = status_to_io_status(
+          ApplyWALToManifest(read_options, write_options, &synced_wals));
     }
 
     if (!log_io_s.ok() && !log_io_s.IsShutdownInProgress() &&
@@ -653,9 +671,14 @@ Status DBImpl::AtomicFlushMemTablesToOutputFiles(
     // Sync on all distinct output directories.
     for (auto dir : distinct_output_dirs) {
       if (dir != nullptr) {
-        Status error_status = dir->FsyncWithDirOptions(
-            IOOptions(), nullptr,
-            DirFsyncOptions(DirFsyncOptions::FsyncReason::kNewFileSynced));
+        IOOptions io_options;
+        Status error_status =
+            WritableFileWriter::PrepareIOOptions(write_options, io_options);
+        if (error_status.ok()) {
+          error_status = dir->FsyncWithDirOptions(
+              io_options, nullptr,
+              DirFsyncOptions(DirFsyncOptions::FsyncReason::kNewFileSynced));
+        }
         if (!error_status.ok()) {
           s = error_status;
           break;
@@ -1049,8 +1072,10 @@ Status DBImpl::IncreaseFullHistoryTsLowImpl(ColumnFamilyData* cfd,
   edit.SetColumnFamily(cfd->GetID());
   edit.SetFullHistoryTsLow(ts_low);
 
-  // TODO: plumb Env::IOActivity
+  // TODO: plumb Env::IOActivity, Env::IOPriority
   const ReadOptions read_options;
+  const WriteOptions write_options;
+
   TEST_SYNC_POINT_CALLBACK("DBImpl::IncreaseFullHistoryTsLowImpl:BeforeEdit",
                            &edit);
 
@@ -1064,7 +1089,7 @@ Status DBImpl::IncreaseFullHistoryTsLowImpl(ColumnFamilyData* cfd,
   }
 
   Status s = versions_->LogAndApply(cfd, *cfd->GetLatestMutableCFOptions(),
-                                    read_options, &edit, &mutex_,
+                                    read_options, write_options, &edit, &mutex_,
                                     directories_.GetDbDir());
   if (!s.ok()) {
     return s;
@@ -1754,6 +1779,7 @@ Status DBImpl::ReFitLevel(ColumnFamilyData* cfd, int level, int target_level) {
   }
 
   const ReadOptions read_options(Env::IOActivity::kCompaction);
+  const WriteOptions write_options(Env::IOActivity::kCompaction);
 
   SuperVersionContext sv_context(/* create_superversion */ true);
 
@@ -1870,9 +1896,9 @@ Status DBImpl::ReFitLevel(ColumnFamilyData* cfd, int level, int target_level) {
                     "[%s] Apply version edit:\n%s", cfd->GetName().c_str(),
                     edit.DebugString().data());
 
-    Status status =
-        versions_->LogAndApply(cfd, mutable_cf_options, read_options, &edit,
-                               &mutex_, directories_.GetDbDir());
+    Status status = versions_->LogAndApply(cfd, mutable_cf_options,
+                                           read_options, write_options, &edit,
+                                           &mutex_, directories_.GetDbDir());
 
     cfd->compaction_picker()->UnregisterCompaction(c.get());
     c.reset();
@@ -2783,6 +2809,7 @@ void DBImpl::EnableManualCompaction() {
 
 void DBImpl::MaybeScheduleFlushOrCompaction() {
   mutex_.AssertHeld();
+  TEST_SYNC_POINT("DBImpl::MaybeScheduleFlushOrCompaction:Start");
   if (!opened_successfully_) {
     // Compaction may introduce data race to DB open
     return;
@@ -3300,7 +3327,7 @@ void DBImpl::BackgroundCallFlush(Env::Priority thread_pri) {
 
     TEST_SYNC_POINT("DBImpl::BackgroundCallFlush:FlushFinish:0");
     ReleaseFileNumberFromPendingOutputs(pending_outputs_inserted_elem);
-    // There is no need to do these clean up if the flush job is rescheduled
+    // There is no need to find obsolete files if the flush job is rescheduled
     // to retain user-defined timestamps because the job doesn't get to the
     // stage of actually flushing the MemTables.
     if (!flush_rescheduled_to_retain_udt) {
@@ -3308,25 +3335,25 @@ void DBImpl::BackgroundCallFlush(Env::Priority thread_pri) {
       // have created. Thus, we force full scan in FindObsoleteFiles()
       FindObsoleteFiles(&job_context, !s.ok() && !s.IsShutdownInProgress() &&
                                           !s.IsColumnFamilyDropped());
-      // delete unnecessary files if any, this is done outside the mutex
-      if (job_context.HaveSomethingToClean() ||
-          job_context.HaveSomethingToDelete() || !log_buffer.IsEmpty()) {
-        mutex_.Unlock();
-        TEST_SYNC_POINT("DBImpl::BackgroundCallFlush:FilesFound");
-        // Have to flush the info logs before bg_flush_scheduled_--
-        // because if bg_flush_scheduled_ becomes 0 and the lock is
-        // released, the deconstructor of DB can kick in and destroy all the
-        // states of DB so info_log might not be available after that point.
-        // It also applies to access other states that DB owns.
-        log_buffer.FlushBufferToLog();
-        if (job_context.HaveSomethingToDelete()) {
-          PurgeObsoleteFiles(job_context);
-        }
-        job_context.Clean();
-        mutex_.Lock();
-      }
-      TEST_SYNC_POINT("DBImpl::BackgroundCallFlush:ContextCleanedUp");
     }
+    // delete unnecessary files if any, this is done outside the mutex
+    if (job_context.HaveSomethingToClean() ||
+        job_context.HaveSomethingToDelete() || !log_buffer.IsEmpty()) {
+      mutex_.Unlock();
+      TEST_SYNC_POINT("DBImpl::BackgroundCallFlush:FilesFound");
+      // Have to flush the info logs before bg_flush_scheduled_--
+      // because if bg_flush_scheduled_ becomes 0 and the lock is
+      // released, the deconstructor of DB can kick in and destroy all the
+      // states of DB so info_log might not be available after that point.
+      // It also applies to access other states that DB owns.
+      log_buffer.FlushBufferToLog();
+      if (job_context.HaveSomethingToDelete()) {
+        PurgeObsoleteFiles(job_context);
+      }
+      job_context.Clean();
+      mutex_.Lock();
+    }
+    TEST_SYNC_POINT("DBImpl::BackgroundCallFlush:ContextCleanedUp");
 
     assert(num_running_flushes_ > 0);
     num_running_flushes_--;
@@ -3479,6 +3506,7 @@ Status DBImpl::BackgroundCompaction(bool* made_progress,
   TEST_SYNC_POINT("DBImpl::BackgroundCompaction:Start");
 
   const ReadOptions read_options(Env::IOActivity::kCompaction);
+  const WriteOptions write_options(Env::IOActivity::kCompaction);
 
   bool is_manual = (manual_compaction != nullptr);
   std::unique_ptr<Compaction> c;
@@ -3691,7 +3719,7 @@ Status DBImpl::BackgroundCompaction(bool* made_progress,
     }
     status = versions_->LogAndApply(
         c->column_family_data(), *c->mutable_cf_options(), read_options,
-        c->edit(), &mutex_, directories_.GetDbDir(),
+        write_options, c->edit(), &mutex_, directories_.GetDbDir(),
         /*new_descriptor_log=*/false, /*column_family_options=*/nullptr,
         [&c, &compaction_released](const Status& s) {
           c->ReleaseCompactionFiles(s);
@@ -3765,7 +3793,7 @@ Status DBImpl::BackgroundCompaction(bool* made_progress,
     }
     status = versions_->LogAndApply(
         c->column_family_data(), *c->mutable_cf_options(), read_options,
-        c->edit(), &mutex_, directories_.GetDbDir(),
+        write_options, c->edit(), &mutex_, directories_.GetDbDir(),
         /*new_descriptor_log=*/false, /*column_family_options=*/nullptr,
         [&c, &compaction_released](const Status& s) {
           c->ReleaseCompactionFiles(s);

db/db_impl/db_impl_debug.cc

@@ -309,6 +309,10 @@ SeqnoToTimeMapping DBImpl::TEST_GetSeqnoToTimeMapping() const {
   return seqno_to_time_mapping_;
 }
 
+const autovector<uint64_t>& DBImpl::TEST_GetFilesToQuarantine() const {
+  InstrumentedMutexLock l(&mutex_);
+  return error_handler_.GetFilesToQuarantine();
+}
 
 size_t DBImpl::TEST_EstimateInMemoryStatsHistorySize() const {
   InstrumentedMutexLock l(&const_cast<DBImpl*>(this)->stats_history_mutex_);

db/db_impl/db_impl_experimental.cc

@@ -61,8 +61,10 @@ Status DBImpl::PromoteL0(ColumnFamilyHandle* column_family, int target_level) {
                    "PromoteL0 FAILED. Invalid target level %d\n", target_level);
     return Status::InvalidArgument("Invalid target level");
   }
-  // TODO: plumb Env::IOActivity
+  // TODO: plumb Env::IOActivity, Env::IOPriority
   const ReadOptions read_options;
+  const WriteOptions write_options;
+
   Status status;
   VersionEdit edit;
   JobContext job_context(next_job_id_.fetch_add(1), true);
@@ -143,7 +145,7 @@ Status DBImpl::PromoteL0(ColumnFamilyHandle* column_family, int target_level) {
     }
 
     status = versions_->LogAndApply(cfd, *cfd->GetLatestMutableCFOptions(),
-                                    read_options, &edit, &mutex_,
+                                    read_options, write_options, &edit, &mutex_,
                                     directories_.GetDbDir());
     if (status.ok()) {
       InstallSuperVersionAndScheduleWork(cfd,

db/db_impl/db_impl_files.cc

@@ -18,6 +18,7 @@
 #include "file/sst_file_manager_impl.h"
 #include "logging/logging.h"
 #include "port/port.h"
+#include "rocksdb/options.h"
 #include "util/autovector.h"
 #include "util/defer.h"
 
@@ -100,14 +101,6 @@ Status DBImpl::EnableFileDeletions(bool force) {
   return Status::OK();
 }
 
-int DBImpl::EnableFileDeletionsWithLock() {
-  mutex_.AssertHeld();
-  // In case others have called EnableFileDeletions(true /* force */) in between
-  disable_delete_obsolete_files_ =
-      std::max(0, disable_delete_obsolete_files_ - 1);
-  return disable_delete_obsolete_files_;
-}
-
 bool DBImpl::IsFileDeletionsEnabled() const {
   return 0 == disable_delete_obsolete_files_;
 }
@@ -154,6 +147,7 @@ void DBImpl::FindObsoleteFiles(JobContext* job_context, bool force,
   // mutex_ cannot be released. Otherwise, we might see no min_pending_output
   // here but later find newer generated unfinalized files while scanning.
   job_context->min_pending_output = MinObsoleteSstNumberToKeep();
+  job_context->files_to_quarantine = error_handler_.GetFilesToQuarantine();
 
   // Get obsolete files.  This function will also update the list of
   // pending files in VersionSet().
@@ -427,6 +421,8 @@ void DBImpl::PurgeObsoleteFiles(JobContext& state, bool schedule_only) {
                                              state.blob_live.end());
   std::unordered_set<uint64_t> log_recycle_files_set(
       state.log_recycle_files.begin(), state.log_recycle_files.end());
+  std::unordered_set<uint64_t> quarantine_files_set(
+      state.files_to_quarantine.begin(), state.files_to_quarantine.end());
 
   auto candidate_files = state.full_scan_candidate_files;
   candidate_files.reserve(
@@ -515,7 +511,8 @@ void DBImpl::PurgeObsoleteFiles(JobContext& state, bool schedule_only) {
   // Close WALs before trying to delete them.
   for (const auto w : state.logs_to_free) {
     // TODO: maybe check the return value of Close.
-    auto s = w->Close();
+    // TODO: plumb Env::IOActivity, Env::IOPriority
+    auto s = w->Close(WriteOptions());
     s.PermitUncheckedError();
   }
 
@@ -530,6 +527,10 @@ void DBImpl::PurgeObsoleteFiles(JobContext& state, bool schedule_only) {
       continue;
     }
 
+    if (quarantine_files_set.find(number) != quarantine_files_set.end()) {
+      continue;
+    }
+
     bool keep = true;
     switch (type) {
       case kWalFile:
@@ -926,7 +927,8 @@ void DBImpl::SetDBId(std::string&& id, bool read_only,
   }
 }
 
-Status DBImpl::SetupDBId(bool read_only, RecoveryContext* recovery_ctx) {
+Status DBImpl::SetupDBId(const WriteOptions& write_options, bool read_only,
+                         RecoveryContext* recovery_ctx) {
   Status s;
   // Check for the IDENTITY file and create it if not there or
   // broken or not matching manifest
@@ -959,7 +961,7 @@ Status DBImpl::SetupDBId(bool read_only, RecoveryContext* recovery_ctx) {
   }
   // Persist it to IDENTITY file if allowed
   if (!read_only) {
-    s = SetIdentityFile(env_, dbname_, db_id_);
+    s = SetIdentityFile(write_options, env_, dbname_, db_id_);
   }
   return s;
 }
@@ -1003,7 +1005,7 @@ Status DBImpl::DeleteUnreferencedSstFiles(RecoveryContext* recovery_ctx) {
       if (type == kTableFile && number >= next_file_number &&
           recovery_ctx->files_to_delete_.find(normalized_fpath) ==
               recovery_ctx->files_to_delete_.end()) {
-        recovery_ctx->files_to_delete_.emplace(normalized_fpath);
+        recovery_ctx->files_to_delete_.emplace(normalized_fpath, path);
       }
     }
   }

db/db_impl/db_impl_open.cc

@@ -21,6 +21,7 @@
 #include "monitoring/persistent_stats_history.h"
 #include "monitoring/thread_status_util.h"
 #include "options/options_helper.h"
+#include "rocksdb/options.h"
 #include "rocksdb/table.h"
 #include "rocksdb/wal_filter.h"
 #include "test_util/sync_point.h"
@@ -309,7 +310,8 @@ Status DBImpl::ValidateOptions(const DBOptions& db_options) {
 
 Status DBImpl::NewDB(std::vector<std::string>* new_filenames) {
   VersionEdit new_db;
-  Status s = SetIdentityFile(env_, dbname_);
+  const WriteOptions write_options(Env::IOActivity::kDBOpen);
+  Status s = SetIdentityFile(write_options, env_, dbname_);
   if (!s.ok()) {
     return s;
   }
@@ -339,20 +341,23 @@ Status DBImpl::NewDB(std::vector<std::string>* new_filenames) {
         immutable_db_options_.manifest_preallocation_size);
     std::unique_ptr<WritableFileWriter> file_writer(new WritableFileWriter(
         std::move(file), manifest, file_options, immutable_db_options_.clock,
-        io_tracer_, nullptr /* stats */, immutable_db_options_.listeners,
-        nullptr, tmp_set.Contains(FileType::kDescriptorFile),
+        io_tracer_, nullptr /* stats */,
+        Histograms::HISTOGRAM_ENUM_MAX /* hist_type */,
+        immutable_db_options_.listeners, nullptr,
+        tmp_set.Contains(FileType::kDescriptorFile),
         tmp_set.Contains(FileType::kDescriptorFile)));
     log::Writer log(std::move(file_writer), 0, false);
     std::string record;
     new_db.EncodeTo(&record);
-    s = log.AddRecord(record);
+    s = log.AddRecord(write_options, record);
     if (s.ok()) {
-      s = SyncManifest(&immutable_db_options_, log.file());
+      s = SyncManifest(&immutable_db_options_, write_options, log.file());
     }
   }
   if (s.ok()) {
     // Make "CURRENT" file that points to the new manifest file.
-    s = SetCurrentFile(fs_.get(), dbname_, 1, directories_.GetDbDir());
+    s = SetCurrentFile(write_options, fs_.get(), dbname_, 1,
+                       directories_.GetDbDir());
     if (new_filenames) {
       new_filenames->emplace_back(
           manifest.substr(manifest.find_last_of("/\\") + 1));
@@ -418,6 +423,7 @@ Status DBImpl::Recover(
     uint64_t* recovered_seq, RecoveryContext* recovery_ctx) {
   mutex_.AssertHeld();
 
+  const WriteOptions write_options(Env::IOActivity::kDBOpen);
   bool tmp_is_new_db = false;
   bool& is_new_db = recovery_ctx ? recovery_ctx->is_new_db_ : tmp_is_new_db;
   assert(db_lock_ == nullptr);
@@ -642,7 +648,7 @@ Status DBImpl::Recover(
       }
     }
   }
-  s = SetupDBId(read_only, recovery_ctx);
+  s = SetupDBId(write_options, read_only, recovery_ctx);
   ROCKS_LOG_INFO(immutable_db_options_.info_log, "DB ID: %s\n", db_id_.c_str());
   if (s.ok() && !read_only) {
     s = DeleteUnreferencedSstFiles(recovery_ctx);
@@ -872,8 +878,9 @@ Status DBImpl::PersistentStatsProcessFormatVersion() {
       if (s.ok()) {
         ColumnFamilyOptions cfo;
         OptimizeForPersistentStats(&cfo);
-        s = CreateColumnFamilyImpl(cfo, kPersistentStatsColumnFamilyName,
-                                   &handle);
+        s = CreateColumnFamilyImpl(ReadOptions(Env::IOActivity::kDBOpen),
+                                   WriteOptions(Env::IOActivity::kDBOpen), cfo,
+                                   kPersistentStatsColumnFamilyName, &handle);
       }
       if (s.ok()) {
         persist_stats_cf_handle_ = static_cast<ColumnFamilyHandleImpl*>(handle);
@@ -895,6 +902,7 @@ Status DBImpl::PersistentStatsProcessFormatVersion() {
                     std::to_string(kStatsCFCompatibleFormatVersion));
     }
     if (s.ok()) {
+      // TODO: plumb Env::IOActivity, Env::IOPriority
       WriteOptions wo;
       wo.low_pri = true;
       wo.no_slowdown = true;
@@ -926,7 +934,9 @@ Status DBImpl::InitPersistStatsColumnFamily() {
     ColumnFamilyHandle* handle = nullptr;
     ColumnFamilyOptions cfo;
     OptimizeForPersistentStats(&cfo);
-    s = CreateColumnFamilyImpl(cfo, kPersistentStatsColumnFamilyName, &handle);
+    s = CreateColumnFamilyImpl(ReadOptions(Env::IOActivity::kDBOpen),
+                               WriteOptions(Env::IOActivity::kDBOpen), cfo,
+                               kPersistentStatsColumnFamilyName, &handle);
     persist_stats_cf_handle_ = static_cast<ColumnFamilyHandleImpl*>(handle);
     mutex_.Lock();
   }
@@ -937,13 +947,19 @@ Status DBImpl::LogAndApplyForRecovery(const RecoveryContext& recovery_ctx) {
   mutex_.AssertHeld();
   assert(versions_->descriptor_log_ == nullptr);
   const ReadOptions read_options(Env::IOActivity::kDBOpen);
-  Status s = versions_->LogAndApply(
-      recovery_ctx.cfds_, recovery_ctx.mutable_cf_opts_, read_options,
-      recovery_ctx.edit_lists_, &mutex_, directories_.GetDbDir());
+  const WriteOptions write_options(Env::IOActivity::kDBOpen);
+
+  Status s = versions_->LogAndApply(recovery_ctx.cfds_,
+                                    recovery_ctx.mutable_cf_opts_, read_options,
+                                    write_options, recovery_ctx.edit_lists_,
+                                    &mutex_, directories_.GetDbDir());
   if (s.ok() && !(recovery_ctx.files_to_delete_.empty())) {
     mutex_.Unlock();
-    for (const auto& fname : recovery_ctx.files_to_delete_) {
-      s = env_->DeleteFile(fname);
+    for (const auto& stale_sst_file : recovery_ctx.files_to_delete_) {
+      s = DeleteDBFile(&immutable_db_options_, stale_sst_file.first,
+                       stale_sst_file.second,
+                       /*force_bg=*/false,
+                       /*force_fg=*/false);
       if (!s.ok()) {
         break;
       }
@@ -1248,6 +1264,13 @@ Status DBImpl::RecoverLogFiles(const std::vector<uint64_t>& wal_numbers,
       }
 
       SequenceNumber sequence = WriteBatchInternal::Sequence(batch_to_use);
+      if (sequence > kMaxSequenceNumber) {
+        reporter.Corruption(
+            record.size(),
+            Status::Corruption("sequence " + std::to_string(sequence) +
+                               " is too large"));
+        continue;
+      }
 
       if (immutable_db_options_.wal_recovery_mode ==
           WALRecoveryMode::kPointInTimeRecovery) {
@@ -1655,9 +1678,11 @@ Status DBImpl::WriteLevel0TableForRecovery(int job_id, ColumnFamilyData* cfd,
       }
 
       IOStatus io_s;
+      const ReadOptions read_option(Env::IOActivity::kDBOpen);
+      const WriteOptions write_option(Env::IO_HIGH, Env::IOActivity::kDBOpen);
       TableBuilderOptions tboptions(
-          *cfd->ioptions(), mutable_cf_options, cfd->internal_comparator(),
-          cfd->int_tbl_prop_collector_factories(),
+          *cfd->ioptions(), mutable_cf_options, read_option, write_option,
+          cfd->internal_comparator(), cfd->int_tbl_prop_collector_factories(),
           GetCompressionFlush(*cfd->ioptions(), mutable_cf_options),
           mutable_cf_options.compression_opts, cfd->GetID(), cfd->GetName(),
           0 /* level */, false /* is_bottommost */,
@@ -1667,16 +1692,15 @@ Status DBImpl::WriteLevel0TableForRecovery(int job_id, ColumnFamilyData* cfd,
       SeqnoToTimeMapping empty_seqno_to_time_mapping;
       Version* version = cfd->current();
       version->Ref();
-      const ReadOptions read_option(Env::IOActivity::kDBOpen);
       uint64_t num_input_entries = 0;
       s = BuildTable(
           dbname_, versions_.get(), immutable_db_options_, tboptions,
-          file_options_for_compaction_, read_option, cfd->table_cache(),
-          iter.get(), std::move(range_del_iters), &meta, &blob_file_additions,
+          file_options_for_compaction_, cfd->table_cache(), iter.get(),
+          std::move(range_del_iters), &meta, &blob_file_additions,
           snapshot_seqs, earliest_write_conflict_snapshot, kMaxSequenceNumber,
           snapshot_checker, paranoid_file_checks, cfd->internal_stats(), &io_s,
           io_tracer_, BlobFileCreationReason::kRecovery,
-          empty_seqno_to_time_mapping, &event_logger_, job_id, Env::IO_HIGH,
+          empty_seqno_to_time_mapping, &event_logger_, job_id,
           nullptr /* table_properties */, write_hint,
           nullptr /*full_history_ts_low*/, &blob_callback_, version,
           &num_input_entries);
@@ -1878,7 +1902,8 @@ Status DB::OpenAndTrimHistory(
   return s;
 }
 
-IOStatus DBImpl::CreateWAL(uint64_t log_file_num, uint64_t recycle_log_number,
+IOStatus DBImpl::CreateWAL(const WriteOptions& write_options,
+                           uint64_t log_file_num, uint64_t recycle_log_number,
                            size_t preallocate_block_size,
                            log::Writer** new_log) {
   IOStatus io_s;
@@ -1912,14 +1937,15 @@ IOStatus DBImpl::CreateWAL(uint64_t log_file_num, uint64_t recycle_log_number,
     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,
-        immutable_db_options_.clock, io_tracer_, nullptr /* stats */, listeners,
-        nullptr, tmp_set.Contains(FileType::kWalFile),
+        immutable_db_options_.clock, io_tracer_, nullptr /* stats */,
+        Histograms::HISTOGRAM_ENUM_MAX /* hist_type */, listeners, nullptr,
+        tmp_set.Contains(FileType::kWalFile),
         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,
                                immutable_db_options_.wal_compression);
-    io_s = (*new_log)->AddCompressionTypeRecord();
+    io_s = (*new_log)->AddCompressionTypeRecord(write_options);
   }
   return io_s;
 }
@@ -1928,6 +1954,9 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
                     const std::vector<ColumnFamilyDescriptor>& column_families,
                     std::vector<ColumnFamilyHandle*>* handles, DB** dbptr,
                     const bool seq_per_batch, const bool batch_per_txn) {
+  const WriteOptions write_options(Env::IOActivity::kDBOpen);
+  const ReadOptions read_options(Env::IOActivity::kDBOpen);
+
   Status s = ValidateOptionsByTable(db_options, column_families);
   if (!s.ok()) {
     return s;
@@ -2004,7 +2033,7 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
     log::Writer* new_log = nullptr;
     const size_t preallocate_block_size =
         impl->GetWalPreallocateBlockSize(max_write_buffer_size);
-    s = impl->CreateWAL(new_log_number, 0 /*recycle_log_number*/,
+    s = impl->CreateWAL(write_options, new_log_number, 0 /*recycle_log_number*/,
                         preallocate_block_size, &new_log);
     if (s.ok()) {
       InstrumentedMutexLock wl(&impl->log_write_mutex_);
@@ -2029,21 +2058,25 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
       if (recovered_seq != kMaxSequenceNumber) {
         WriteBatch empty_batch;
         WriteBatchInternal::SetSequence(&empty_batch, recovered_seq);
-        WriteOptions write_options;
         uint64_t log_used, log_size;
         log::Writer* log_writer = impl->logs_.back().writer;
         LogFileNumberSize& log_file_number_size = impl->alive_log_files_.back();
 
         assert(log_writer->get_log_number() == log_file_number_size.number);
         impl->mutex_.AssertHeld();
-        s = impl->WriteToWAL(empty_batch, log_writer, &log_used, &log_size,
-                             Env::IO_TOTAL, log_file_number_size);
+        s = impl->WriteToWAL(empty_batch, write_options, log_writer, &log_used,
+                             &log_size, log_file_number_size);
         if (s.ok()) {
           // Need to fsync, otherwise it might get lost after a power reset.
-          s = impl->FlushWAL(false);
+          s = impl->FlushWAL(write_options, false);
           TEST_SYNC_POINT_CALLBACK("DBImpl::Open::BeforeSyncWAL", /*arg=*/&s);
+          IOOptions opts;
           if (s.ok()) {
-            s = log_writer->file()->Sync(impl->immutable_db_options_.use_fsync);
+            s = WritableFileWriter::PrepareIOOptions(write_options, opts);
+          }
+          if (s.ok()) {
+            s = log_writer->file()->Sync(opts,
+                                         impl->immutable_db_options_.use_fsync);
           }
         }
       }
@@ -2074,7 +2107,8 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
           impl->mutex_.Unlock();
           // NOTE: the work normally done in WrapUpCreateColumnFamilies will
           // be done separately below.
-          s = impl->CreateColumnFamilyImpl(cf.options, cf.name, &handle);
+          s = impl->CreateColumnFamilyImpl(read_options, write_options,
+                                           cf.options, cf.name, &handle);
           impl->mutex_.Lock();
           if (s.ok()) {
             handles->push_back(handle);
@@ -2126,7 +2160,7 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
     // Persist RocksDB Options before scheduling the compaction.
     // The WriteOptionsFile() will release and lock the mutex internally.
     persist_options_status =
-        impl->WriteOptionsFile(true /*db_mutex_already_held*/);
+        impl->WriteOptionsFile(write_options, true /*db_mutex_already_held*/);
     *dbptr = impl;
     impl->opened_successfully_ = true;
     impl->DeleteObsoleteFiles();
@@ -2226,12 +2260,17 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
                      impl);
     LogFlush(impl->immutable_db_options_.info_log);
     if (!impl->WALBufferIsEmpty()) {
-      s = impl->FlushWAL(false);
+      s = impl->FlushWAL(write_options, false);
       if (s.ok()) {
         // Sync is needed otherwise WAL buffered data might get lost after a
         // power reset.
         log::Writer* log_writer = impl->logs_.back().writer;
-        s = log_writer->file()->Sync(impl->immutable_db_options_.use_fsync);
+        IOOptions opts;
+        s = WritableFileWriter::PrepareIOOptions(write_options, opts);
+        if (s.ok()) {
+          s = log_writer->file()->Sync(opts,
+                                       impl->immutable_db_options_.use_fsync);
+        }
       }
     }
     if (s.ok() && !persist_options_status.ok()) {
@@ -2248,7 +2287,8 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
     s = impl->StartPeriodicTaskScheduler();
   }
   if (s.ok()) {
-    s = impl->RegisterRecordSeqnoTimeWorker(recovery_ctx.is_new_db_);
+    s = impl->RegisterRecordSeqnoTimeWorker(read_options, write_options,
+                                            recovery_ctx.is_new_db_);
   }
   impl->options_mutex_.Unlock();
   if (!s.ok()) {

db/db_impl/db_impl_readonly.h

@@ -53,6 +53,10 @@ class DBImplReadOnly : public DBImpl {
                    const WideColumns& /* columns */) override {
     return Status::NotSupported("Not supported operation in read only mode.");
   }
+  Status PutEntity(const WriteOptions& /* options */, const Slice& /* key */,
+                   const AttributeGroups& /* attribute_groups */) override {
+    return Status::NotSupported("Not supported operation in read only mode.");
+  }
 
   using DBImpl::Merge;
   virtual Status Merge(const WriteOptions& /*options*/,

db/db_impl/db_impl_secondary.h

@@ -136,6 +136,10 @@ class DBImplSecondary : public DBImpl {
                    const WideColumns& /* columns */) override {
     return Status::NotSupported("Not supported operation in secondary mode.");
   }
+  Status PutEntity(const WriteOptions& /* options */, const Slice& /* key */,
+                   const AttributeGroups& /* attribute_groups */) override {
+    return Status::NotSupported("Not supported operation in secondary mode.");
+  }
 
   using DBImpl::Merge;
   Status Merge(const WriteOptions& /*options*/,

db/db_impl/db_impl_write.cc

@@ -48,6 +48,17 @@ Status DBImpl::PutEntity(const WriteOptions& options,
   return DB::PutEntity(options, column_family, key, columns);
 }
 
+Status DBImpl::PutEntity(const WriteOptions& options, const Slice& key,
+                         const AttributeGroups& attribute_groups) {
+  for (const AttributeGroup& ag : attribute_groups) {
+    const Status s = FailIfCfHasTs(ag.column_family());
+    if (!s.ok()) {
+      return s;
+    }
+  }
+  return DB::PutEntity(options, key, attribute_groups);
+}
+
 Status DBImpl::Merge(const WriteOptions& o, ColumnFamilyHandle* column_family,
                      const Slice& key, const Slice& val) {
   const Status s = FailIfCfHasTs(column_family);
@@ -426,10 +437,10 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
         valid_batches += writer->batch_cnt;
         if (writer->ShouldWriteToMemtable()) {
           total_count += WriteBatchInternal::Count(writer->batch);
+          total_byte_size = WriteBatchInternal::AppendedByteSize(
+              total_byte_size, WriteBatchInternal::ByteSize(writer->batch));
           parallel = parallel && !writer->batch->HasMerge();
         }
-        total_byte_size = WriteBatchInternal::AppendedByteSize(
-            total_byte_size, WriteBatchInternal::ByteSize(writer->batch));
         if (writer->pre_release_callback) {
           pre_release_callback_cnt++;
         }
@@ -609,9 +620,9 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
     log_write_mutex_.Unlock();
     if (status.ok() && synced_wals.IsWalAddition()) {
       InstrumentedMutexLock l(&mutex_);
-      // TODO: plumb Env::IOActivity
+      // TODO: plumb Env::IOActivity, Env::IOPriority
       const ReadOptions read_options;
-      status = ApplyWALToManifest(read_options, &synced_wals);
+      status = ApplyWALToManifest(read_options, write_options, &synced_wals);
     }
 
     // Requesting sync with two_write_queues_ is expected to be very rare. We
@@ -709,11 +720,11 @@ Status DBImpl::PipelinedWriteImpl(const WriteOptions& write_options,
           if (writer->ShouldWriteToMemtable()) {
             writer->sequence = next_sequence;
             size_t count = WriteBatchInternal::Count(writer->batch);
+            total_byte_size = WriteBatchInternal::AppendedByteSize(
+                total_byte_size, WriteBatchInternal::ByteSize(writer->batch));
             next_sequence += count;
             total_count += count;
           }
-          total_byte_size = WriteBatchInternal::AppendedByteSize(
-              total_byte_size, WriteBatchInternal::ByteSize(writer->batch));
         }
       }
       if (w.disable_wal) {
@@ -772,9 +783,9 @@ Status DBImpl::PipelinedWriteImpl(const WriteOptions& write_options,
     }
     if (w.status.ok() && synced_wals.IsWalAddition()) {
       InstrumentedMutexLock l(&mutex_);
-      // TODO: plumb Env::IOActivity
+      // TODO: plumb Env::IOActivity, Env::IOPriority
       const ReadOptions read_options;
-      w.status = ApplyWALToManifest(read_options, &synced_wals);
+      w.status = ApplyWALToManifest(read_options, write_options, &synced_wals);
     }
     write_thread_.ExitAsBatchGroupLeader(wal_write_group, w.status);
   }
@@ -1307,9 +1318,9 @@ Status DBImpl::MergeBatch(const WriteThread::WriteGroup& write_group,
 // When two_write_queues_ is disabled, this function is called from the only
 // write thread. Otherwise this must be called holding log_write_mutex_.
 IOStatus DBImpl::WriteToWAL(const WriteBatch& merged_batch,
+                            const WriteOptions& write_options,
                             log::Writer* log_writer, uint64_t* log_used,
                             uint64_t* log_size,
-                            Env::IOPriority rate_limiter_priority,
                             LogFileNumberSize& log_file_number_size) {
   assert(log_size != nullptr);
 
@@ -1332,12 +1343,11 @@ IOStatus DBImpl::WriteToWAL(const WriteBatch& merged_batch,
     log_write_mutex_.Lock();
   }
   IOStatus io_s = log_writer->MaybeAddUserDefinedTimestampSizeRecord(
-      versions_->GetColumnFamiliesTimestampSizeForRecord(),
-      rate_limiter_priority);
+      write_options, versions_->GetColumnFamiliesTimestampSizeForRecord());
   if (!io_s.ok()) {
     return io_s;
   }
-  io_s = log_writer->AddRecord(log_entry, rate_limiter_priority);
+  io_s = log_writer->AddRecord(write_options, log_entry);
 
   if (UNLIKELY(needs_locking)) {
     log_write_mutex_.Unlock();
@@ -1380,9 +1390,13 @@ IOStatus DBImpl::WriteToWAL(const WriteThread::WriteGroup& write_group,
   WriteBatchInternal::SetSequence(merged_batch, sequence);
 
   uint64_t log_size;
-  io_s = WriteToWAL(*merged_batch, log_writer, log_used, &log_size,
-                    write_group.leader->rate_limiter_priority,
-                    log_file_number_size);
+
+  // TODO: plumb Env::IOActivity, Env::IOPriority
+  WriteOptions write_options;
+  write_options.rate_limiter_priority =
+      write_group.leader->rate_limiter_priority;
+  io_s = WriteToWAL(*merged_batch, write_options, log_writer, log_used,
+                    &log_size, log_file_number_size);
   if (to_be_cached_state) {
     cached_recoverable_state_ = *to_be_cached_state;
     cached_recoverable_state_empty_ = false;
@@ -1409,10 +1423,17 @@ IOStatus DBImpl::WriteToWAL(const WriteThread::WriteGroup& write_group,
       log_write_mutex_.Lock();
     }
 
-    for (auto& log : logs_) {
-      io_s = log.writer->file()->Sync(immutable_db_options_.use_fsync);
-      if (!io_s.ok()) {
-        break;
+    if (io_s.ok()) {
+      for (auto& log : logs_) {
+        IOOptions opts;
+        io_s = WritableFileWriter::PrepareIOOptions(write_options, opts);
+        if (!io_s.ok()) {
+          break;
+        }
+        io_s = log.writer->file()->Sync(opts, immutable_db_options_.use_fsync);
+        if (!io_s.ok()) {
+          break;
+        }
       }
     }
 
@@ -1485,9 +1506,13 @@ IOStatus DBImpl::ConcurrentWriteToWAL(
   assert(log_writer->get_log_number() == log_file_number_size.number);
 
   uint64_t log_size;
-  io_s = WriteToWAL(*merged_batch, log_writer, log_used, &log_size,
-                    write_group.leader->rate_limiter_priority,
-                    log_file_number_size);
+
+  // TODO: plumb Env::IOActivity, Env::IOPriority
+  WriteOptions write_options;
+  write_options.rate_limiter_priority =
+      write_group.leader->rate_limiter_priority;
+  io_s = WriteToWAL(*merged_batch, write_options, log_writer, log_used,
+                    &log_size, log_file_number_size);
   if (to_be_cached_state) {
     cached_recoverable_state_ = *to_be_cached_state;
     cached_recoverable_state_empty_ = false;
@@ -1552,6 +1577,8 @@ Status DBImpl::WriteRecoverableState() {
     if (status.ok()) {
       cached_recoverable_state_.Clear();
       cached_recoverable_state_empty_ = true;
+    } else {
+      // FIXME: !ok status is untested
     }
     return status;
   }
@@ -1814,11 +1841,9 @@ uint64_t DBImpl::GetMaxTotalWalSize() const {
 Status DBImpl::DelayWrite(uint64_t num_bytes, WriteThread& write_thread,
                           const WriteOptions& write_options) {
   mutex_.AssertHeld();
-  uint64_t time_delayed = 0;
+  uint64_t start_time = 0;
   bool delayed = false;
   {
-    StopWatch sw(immutable_db_options_.clock, stats_, WRITE_STALL,
-                 Histograms::HISTOGRAM_ENUM_MAX, &time_delayed);
     // To avoid parallel timed delays (bad throttling), only support them
     // on the primary write queue.
     uint64_t delay;
@@ -1830,6 +1855,8 @@ Status DBImpl::DelayWrite(uint64_t num_bytes, WriteThread& write_thread,
       delay = 0;
     }
     TEST_SYNC_POINT("DBImpl::DelayWrite:Start");
+    start_time = immutable_db_options_.clock->NowMicros();
+
     if (delay > 0) {
       if (write_options.no_slowdown) {
         return Status::Incomplete("Write stall");
@@ -1846,7 +1873,7 @@ Status DBImpl::DelayWrite(uint64_t num_bytes, WriteThread& write_thread,
       // (slightly longer because WriteController minimum delay is 1ms, in
       // case of sleep imprecision, rounding, etc.)
       const uint64_t kDelayInterval = 1001;
-      uint64_t stall_end = sw.start_time() + delay;
+      uint64_t stall_end = start_time + delay;
       while (write_controller_.NeedsDelay()) {
         if (immutable_db_options_.clock->NowMicros() >= stall_end) {
           // We already delayed this write `delay` microseconds
@@ -1887,9 +1914,11 @@ Status DBImpl::DelayWrite(uint64_t num_bytes, WriteThread& write_thread,
   }
   assert(!delayed || !write_options.no_slowdown);
   if (delayed) {
+    auto time_delayed = immutable_db_options_.clock->NowMicros() - start_time;
     default_cf_internal_stats_->AddDBStats(
         InternalStats::kIntStatsWriteStallMicros, time_delayed);
     RecordTick(stats_, STALL_MICROS, time_delayed);
+    RecordInHistogram(stats_, WRITE_STALL, time_delayed);
   }
 
   // If DB is not in read-only mode and write_controller is not stopping
@@ -1956,9 +1985,13 @@ Status DBImpl::ThrottleLowPriWritesIfNeeded(const WriteOptions& write_options,
       // a chance to run. Now we guarantee we are still slowly making
       // progress.
       PERF_TIMER_GUARD(write_delay_time);
-      write_controller_.low_pri_rate_limiter()->Request(
-          my_batch->GetDataSize(), Env::IO_HIGH, nullptr /* stats */,
-          RateLimiter::OpType::kWrite);
+      auto data_size = my_batch->GetDataSize();
+      while (data_size > 0) {
+        size_t allowed = write_controller_.low_pri_rate_limiter()->RequestToken(
+            data_size, 0 /* alignment */, Env::IO_HIGH, nullptr /* stats */,
+            RateLimiter::OpType::kWrite);
+        data_size -= allowed;
+      }
     }
   }
   return Status::OK();
@@ -2043,16 +2076,15 @@ Status DBImpl::ScheduleFlushes(WriteContext* context) {
     nonmem_write_thread_.EnterUnbatched(&nonmem_w, &mutex_);
   }
 
+  TEST_SYNC_POINT_CALLBACK("DBImpl::ScheduleFlushes:PreSwitchMemtable",
+                           nullptr);
   for (auto& cfd : cfds) {
-    if (!cfd->mem()->IsEmpty()) {
+    if (status.ok() && !cfd->mem()->IsEmpty()) {
       status = SwitchMemtable(cfd, context);
     }
     if (cfd->UnrefAndTryDelete()) {
       cfd = nullptr;
     }
-    if (!status.ok()) {
-      break;
-    }
   }
 
   if (two_write_queues_) {
@@ -2099,8 +2131,10 @@ void DBImpl::NotifyOnMemTableSealed(ColumnFamilyData* /*cfd*/,
 // two_write_queues_ is true (This is to simplify the reasoning.)
 Status DBImpl::SwitchMemtable(ColumnFamilyData* cfd, WriteContext* context) {
   mutex_.AssertHeld();
-  // TODO: plumb Env::IOActivity
+  // TODO: plumb Env::IOActivity, Env::IOPriority
   const ReadOptions read_options;
+  const WriteOptions write_options;
+
   log::Writer* new_log = nullptr;
   MemTable* new_mem = nullptr;
   IOStatus io_s;
@@ -2147,8 +2181,8 @@ Status DBImpl::SwitchMemtable(ColumnFamilyData* cfd, WriteContext* context) {
   if (creating_new_log) {
     // TODO: Write buffer size passed in should be max of all CF's instead
     // of mutable_cf_options.write_buffer_size.
-    io_s = CreateWAL(new_log_number, recycle_log_number, preallocate_block_size,
-                     &new_log);
+    io_s = CreateWAL(write_options, new_log_number, recycle_log_number,
+                     preallocate_block_size, &new_log);
     if (s.ok()) {
       s = io_s;
     }
@@ -2185,7 +2219,7 @@ Status DBImpl::SwitchMemtable(ColumnFamilyData* cfd, WriteContext* context) {
         // In recovery path, we force another try of writing WAL buffer.
         cur_log_writer->file()->reset_seen_error();
       }
-      io_s = cur_log_writer->WriteBuffer();
+      io_s = cur_log_writer->WriteBuffer(write_options);
       if (s.ok()) {
         s = io_s;
       }
@@ -2253,7 +2287,8 @@ Status DBImpl::SwitchMemtable(ColumnFamilyData* cfd, WriteContext* context) {
       VersionEdit wal_deletion;
       wal_deletion.DeleteWalsBefore(min_wal_number_to_keep);
       s = versions_->LogAndApplyToDefaultColumnFamily(
-          read_options, &wal_deletion, &mutex_, directories_.GetDbDir());
+          read_options, write_options, &wal_deletion, &mutex_,
+          directories_.GetDbDir());
       if (!s.ok() && versions_->io_status().IsIOError()) {
         s = error_handler_.SetBGError(versions_->io_status(),
                                       BackgroundErrorReason::kManifestWrite);
@@ -2381,6 +2416,22 @@ Status DB::PutEntity(const WriteOptions& options,
   return Write(options, &batch);
 }
 
+Status DB::PutEntity(const WriteOptions& options, const Slice& key,
+                     const AttributeGroups& attribute_groups) {
+  ColumnFamilyHandle* default_cf = DefaultColumnFamily();
+  assert(default_cf);
+  const Comparator* const default_cf_ucmp = default_cf->GetComparator();
+  assert(default_cf_ucmp);
+  WriteBatch batch(0 /* reserved_bytes */, 0 /* max_bytes */,
+                   options.protection_bytes_per_key,
+                   default_cf_ucmp->timestamp_size());
+  const Status s = batch.PutEntity(key, attribute_groups);
+  if (!s.ok()) {
+    return s;
+  }
+  return Write(options, &batch);
+}
+
 Status DB::Delete(const WriteOptions& opt, ColumnFamilyHandle* column_family,
                   const Slice& key) {
   WriteBatch batch(0 /* reserved_bytes */, 0 /* max_bytes */,

db/db_info_dumper.cc

@@ -5,10 +5,9 @@
 
 #include "db/db_info_dumper.h"
 
-#include <stdio.h>
-
 #include <algorithm>
 #include <cinttypes>
+#include <cstdio>
 #include <string>
 #include <vector>
 

db/db_iter.cc

@@ -83,8 +83,7 @@ DBIter::DBIter(Env* _env, const ReadOptions& read_options,
       cfd_(cfd),
       timestamp_ub_(read_options.timestamp),
       timestamp_lb_(read_options.iter_start_ts),
-      timestamp_size_(timestamp_ub_ ? timestamp_ub_->size() : 0),
-      auto_readahead_size_(read_options.auto_readahead_size) {
+      timestamp_size_(timestamp_ub_ ? timestamp_ub_->size() : 0) {
   RecordTick(statistics_, NO_ITERATOR_CREATED);
   if (pin_thru_lifetime_) {
     pinned_iters_mgr_.StartPinning();
@@ -114,6 +113,9 @@ Status DBIter::GetProperty(std::string prop_name, std::string* prop) {
   } else if (prop_name == "rocksdb.iterator.internal-key") {
     *prop = saved_key_.GetUserKey().ToString();
     return Status::OK();
+  } else if (prop_name == "rocksdb.iterator.write-time") {
+    // TODO(yuzhangyu): implement return the actual write time.
+    return Status::NotSupported("write time property is under construction");
   }
   return Status::InvalidArgument("Unidentified property.");
 }
@@ -199,6 +201,7 @@ bool DBIter::SetBlobValueIfNeeded(const Slice& user_key,
 
   // TODO: consider moving ReadOptions from ArenaWrappedDBIter to DBIter to
   // avoid having to copy options back and forth.
+  // TODO: plumb Env::IOActivity, Env::IOPriority
   ReadOptions read_options;
   read_options.read_tier = read_tier_;
   read_options.fill_cache = fill_cache_;
@@ -392,9 +395,7 @@ bool DBIter::FindNextUserEntryInternal(bool skipping_saved_key,
           case kTypeValue:
           case kTypeBlobIndex:
           case kTypeWideColumnEntity:
-            if (!iter_.PrepareValue()) {
-              assert(!iter_.status().ok());
-              valid_ = false;
+            if (!PrepareValue()) {
               return false;
             }
             if (timestamp_lb_) {
@@ -425,9 +426,7 @@ bool DBIter::FindNextUserEntryInternal(bool skipping_saved_key,
             return true;
             break;
           case kTypeMerge:
-            if (!iter_.PrepareValue()) {
-              assert(!iter_.status().ok());
-              valid_ = false;
+            if (!PrepareValue()) {
               return false;
             }
             saved_key_.SetUserKey(
@@ -572,8 +571,7 @@ bool DBIter::MergeValuesNewToOld() {
       iter_.Next();
       break;
     }
-    if (!iter_.PrepareValue()) {
-      valid_ = false;
+    if (!PrepareValue()) {
       return false;
     }
 
@@ -744,22 +742,15 @@ bool DBIter::ReverseToBackward() {
   // When current_entry_is_merged_ is true, iter_ may be positioned on the next
   // key, which may not exist or may have prefix different from current.
   // If that's the case, seek to saved_key_.
-  //
-  // In case of auto_readahead_size enabled, index_iter moves forward during
-  // forward scan for block cache lookup and points to different block. If Prev
-  // op is called, it needs to call SeekForPrev to point to right index_iter_ in
-  // BlockBasedTableIterator. This only happens when direction is changed from
-  // forward to backward.
-  if ((current_entry_is_merged_ &&
-       (!expect_total_order_inner_iter() || !iter_.Valid())) ||
-      auto_readahead_size_) {
+  if (current_entry_is_merged_ &&
+      (!expect_total_order_inner_iter() || !iter_.Valid())) {
     IterKey last_key;
     // Using kMaxSequenceNumber and kValueTypeForSeek
     // (not kValueTypeForSeekForPrev) to seek to a key strictly smaller
     // than saved_key_.
     last_key.SetInternalKey(ParsedInternalKey(
         saved_key_.GetUserKey(), kMaxSequenceNumber, kValueTypeForSeek));
-    if (!expect_total_order_inner_iter() || auto_readahead_size_) {
+    if (!expect_total_order_inner_iter()) {
       iter_.SeekForPrev(last_key.GetInternalKey());
     } else {
       // Some iterators may not support SeekForPrev(), so we avoid using it
@@ -906,8 +897,7 @@ bool DBIter::FindValueForCurrentKey() {
       return FindValueForCurrentKeyUsingSeek();
     }
 
-    if (!iter_.PrepareValue()) {
-      valid_ = false;
+    if (!PrepareValue()) {
       return false;
     }
 
@@ -1146,8 +1136,7 @@ bool DBIter::FindValueForCurrentKeyUsingSeek() {
     }
     return true;
   }
-  if (!iter_.PrepareValue()) {
-    valid_ = false;
+  if (!PrepareValue()) {
     return false;
   }
   if (timestamp_size_ > 0) {
@@ -1212,8 +1201,7 @@ bool DBIter::FindValueForCurrentKeyUsingSeek() {
         ikey.type == kTypeDeletionWithTimestamp) {
       break;
     }
-    if (!iter_.PrepareValue()) {
-      valid_ = false;
+    if (!PrepareValue()) {
       return false;
     }
 
@@ -1290,8 +1278,8 @@ bool DBIter::MergeWithNoBaseValue(const Slice& user_key) {
   const Status s = MergeHelper::TimedFullMerge(
       merge_operator_, user_key, MergeHelper::kNoBaseValue,
       merge_context_.GetOperands(), logger_, statistics_, clock_,
-      /* update_num_ops_stats */ true, &saved_value_, &pinned_value_,
-      &result_type, /* op_failure_scope */ nullptr);
+      /* update_num_ops_stats */ true, /* op_failure_scope */ nullptr,
+      &saved_value_, &pinned_value_, &result_type);
   return SetValueAndColumnsFromMergeResult(s, result_type);
 }
 
@@ -1303,8 +1291,8 @@ bool DBIter::MergeWithPlainBaseValue(const Slice& value,
   const Status s = MergeHelper::TimedFullMerge(
       merge_operator_, user_key, MergeHelper::kPlainBaseValue, value,
       merge_context_.GetOperands(), logger_, statistics_, clock_,
-      /* update_num_ops_stats */ true, &saved_value_, &pinned_value_,
-      &result_type, /* op_failure_scope */ nullptr);
+      /* update_num_ops_stats */ true, /* op_failure_scope */ nullptr,
+      &saved_value_, &pinned_value_, &result_type);
   return SetValueAndColumnsFromMergeResult(s, result_type);
 }
 
@@ -1316,8 +1304,8 @@ bool DBIter::MergeWithWideColumnBaseValue(const Slice& entity,
   const Status s = MergeHelper::TimedFullMerge(
       merge_operator_, user_key, MergeHelper::kWideBaseValue, entity,
       merge_context_.GetOperands(), logger_, statistics_, clock_,
-      /* update_num_ops_stats */ true, &saved_value_, &pinned_value_,
-      &result_type, /* op_failure_scope */ nullptr);
+      /* update_num_ops_stats */ true, /* op_failure_scope */ nullptr,
+      &saved_value_, &pinned_value_, &result_type);
   return SetValueAndColumnsFromMergeResult(s, result_type);
 }
 

db/db_iter.h

@@ -126,6 +126,10 @@ class DBIter final : public Iterator {
   void operator=(const DBIter&) = delete;
 
   ~DBIter() override {
+    ThreadStatus::OperationType cur_op_type =
+        ThreadStatusUtil::GetThreadOperation();
+    ThreadStatusUtil::SetThreadOperation(
+        ThreadStatus::OperationType::OP_UNKNOWN);
     // Release pinned data if any
     if (pinned_iters_mgr_.PinningEnabled()) {
       pinned_iters_mgr_.ReleasePinnedData();
@@ -134,6 +138,7 @@ class DBIter final : public Iterator {
     ResetInternalKeysSkippedCounter();
     local_stats_.BumpGlobalStatistics(statistics_);
     iter_.DeleteIter(arena_mode_);
+    ThreadStatusUtil::SetThreadOperation(cur_op_type);
   }
   void SetIter(InternalIterator* iter) {
     assert(iter_.iter() == nullptr);
@@ -328,6 +333,22 @@ class DBIter final : public Iterator {
   bool MergeWithPlainBaseValue(const Slice& value, const Slice& user_key);
   bool MergeWithWideColumnBaseValue(const Slice& entity, const Slice& user_key);
 
+  bool PrepareValue() {
+    if (!iter_.PrepareValue()) {
+      assert(!iter_.status().ok());
+      valid_ = false;
+      return false;
+    }
+    // ikey_ could change as BlockBasedTableIterator does Block cache
+    // lookup and index_iter_ could point to different block resulting
+    // in ikey_ pointing to wrong key. So ikey_ needs to be updated in
+    // case of Seek/Next calls to point to right key again.
+    if (!ParseKey(&ikey_)) {
+      return false;
+    }
+    return true;
+  }
+
   const SliceTransform* prefix_extractor_;
   Env* const env_;
   SystemClock* clock_;
@@ -402,7 +423,6 @@ class DBIter final : public Iterator {
   const Slice* const timestamp_lb_;
   const size_t timestamp_size_;
   std::string saved_timestamp_;
-  bool auto_readahead_size_;
 };
 
 // Return a new iterator that converts internal keys (yielded by

db/db_iter_stress_test.cc

@@ -50,7 +50,9 @@ struct Entry {
   bool visible = true;
 
   bool operator<(const Entry& e) const {
-    if (key != e.key) return key < e.key;
+    if (key != e.key) {
+      return key < e.key;
+    }
     return std::tie(sequence, type) > std::tie(e.sequence, e.type);
   }
 };
@@ -177,7 +179,9 @@ struct StressTestIterator : public InternalIterator {
   }
 
   void SeekToFirst() override {
-    if (MaybeFail()) return;
+    if (MaybeFail()) {
+      return;
+    }
     MaybeMutate();
 
     status_ = Status::OK();
@@ -185,7 +189,9 @@ struct StressTestIterator : public InternalIterator {
     SkipForward();
   }
   void SeekToLast() override {
-    if (MaybeFail()) return;
+    if (MaybeFail()) {
+      return;
+    }
     MaybeMutate();
 
     status_ = Status::OK();
@@ -194,7 +200,9 @@ struct StressTestIterator : public InternalIterator {
   }
 
   void Seek(const Slice& target) override {
-    if (MaybeFail()) return;
+    if (MaybeFail()) {
+      return;
+    }
     MaybeMutate();
 
     status_ = Status::OK();
@@ -206,7 +214,9 @@ struct StressTestIterator : public InternalIterator {
     SkipForward();
   }
   void SeekForPrev(const Slice& target) override {
-    if (MaybeFail()) return;
+    if (MaybeFail()) {
+      return;
+    }
     MaybeMutate();
 
     status_ = Status::OK();
@@ -221,14 +231,18 @@ struct StressTestIterator : public InternalIterator {
 
   void Next() override {
     assert(Valid());
-    if (MaybeFail()) return;
+    if (MaybeFail()) {
+      return;
+    }
     MaybeMutate();
     ++iter;
     SkipForward();
   }
   void Prev() override {
     assert(Valid());
-    if (MaybeFail()) return;
+    if (MaybeFail()) {
+      return;
+    }
     MaybeMutate();
     --iter;
     SkipBackward();
@@ -318,7 +332,9 @@ struct ReferenceIterator {
         return false;
       }
       assert(e.sequence <= sequence);
-      if (!e.visible) continue;
+      if (!e.visible) {
+        continue;
+      }
       if (e.type == kTypeDeletion) {
         return false;
       }
@@ -339,11 +355,13 @@ struct ReferenceIterator {
         break;
       }
       assert(e.sequence <= sequence);
-      if (!e.visible) continue;
+      if (!e.visible) {
+        continue;
+      }
       if (e.type == kTypeDeletion) {
         break;
       }
-      operands.push_back(e.value);
+      operands.emplace_back(e.value);
       if (e.type == kTypeValue) {
         break;
       }
@@ -588,15 +606,17 @@ TEST_F(DBIteratorStressTest, StressTest) {
 
                     // Check that the key moved in the right direction.
                     if (forward) {
-                      if (seek)
+                      if (seek) {
                         ASSERT_GE(db_iter->key().ToString(), old_key);
-                      else
+                      } else {
                         ASSERT_GT(db_iter->key().ToString(), old_key);
+                      }
                     } else {
-                      if (seek)
+                      if (seek) {
                         ASSERT_LE(db_iter->key().ToString(), old_key);
-                      else
+                      } else {
                         ASSERT_LT(db_iter->key().ToString(), old_key);
+                      }
                     }
 
                     if (ref_iter->Valid()) {

db/db_iter_test.cc

@@ -65,8 +65,7 @@ class TestIterator : public InternalIterator {
            size_t seq_num, bool update_iter = false) {
     valid_ = true;
     ParsedInternalKey internal_key(argkey, seq_num, type);
-    data_.push_back(
-        std::pair<std::string, std::string>(std::string(), argvalue));
+    data_.emplace_back(std::string(), argvalue);
     AppendInternalKey(&data_.back().first, internal_key);
     if (update_iter && valid_ && cmp.Compare(data_.back().first, key()) < 0) {
       // insert a key smaller than current key
@@ -2617,7 +2616,7 @@ class DBIterWithMergeIterTest : public testing::Test {
     child_iters.push_back(internal_iter2_);
     InternalKeyComparator icomp(BytewiseComparator());
     InternalIterator* merge_iter =
-        NewMergingIterator(&icomp_, &child_iters[0], 2u);
+        NewMergingIterator(&icomp_, child_iters.data(), 2u);
 
     db_iter_.reset(NewDBIterator(
         env_, ro_, ImmutableOptions(options_), MutableCFOptions(options_),

db/db_iterator_test.cc

@@ -86,7 +86,7 @@ TEST_F(DBIteratorBaseTest, APICallsWithPerfContext) {
 class DBIteratorTest : public DBIteratorBaseTest,
                        public testing::WithParamInterface<bool> {
  public:
-  DBIteratorTest() {}
+  DBIteratorTest() = default;
 
   Iterator* NewIterator(const ReadOptions& read_options,
                         ColumnFamilyHandle* column_family = nullptr) {

db/db_kv_checksum_test.cc

@@ -658,7 +658,7 @@ TEST_F(DbKVChecksumWALToWriteBatchTest, WriteBatchChecksumHandoff) {
   Options options = CurrentOptions();
   Reopen(options);
   ASSERT_OK(db_->Put(WriteOptions(), "key", "val"));
-  std::string content = "";
+  std::string content;
   SyncPoint::GetInstance()->SetCallBack(
       "DBImpl::RecoverLogFiles:BeforeUpdateProtectionInfo:batch",
       [&](void* batch_ptr) {

db/db_log_iter_test.cc

@@ -236,7 +236,7 @@ TEST_F(DBTestXactLogIterator, TransactionLogIteratorCorruptedLog) {
     ASSERT_OK(test::TruncateFile(env_, logfile_path,
                                  wal_files.front()->SizeFileBytes() / 2));
 
-    ASSERT_OK(db_->EnableFileDeletions());
+    ASSERT_OK(db_->EnableFileDeletions(/*force=*/false));
 
     // Insert a new entry to a new log file
     ASSERT_OK(Put("key1025", DummyString(10)));

db/db_options_test.cc

@@ -1099,7 +1099,7 @@ TEST_F(DBOptionsTest, SetFIFOCompactionOptions) {
   ASSERT_EQ(fifo_temp_opt[1].age, 30000);
 }
 
-TEST_F(DBOptionsTest, OffPeakTimes) {
+TEST_F(DBOptionsTest, OffpeakTimes) {
   Options options;
   options.create_if_missing = true;
   Random rnd(test::RandomSeed());
@@ -1146,148 +1146,134 @@ TEST_F(DBOptionsTest, OffPeakTimes) {
       "1:0000000000000-2:000000000042",  // Weird, but we can parse the int.
   };
 
-  for (std::string invalid_case : invalid_cases) {
+  for (const std::string& invalid_case : invalid_cases) {
     options.daily_offpeak_time_utc = invalid_case;
     verify_invalid();
   }
-  for (std::string valid_case : valid_cases) {
+  for (const std::string& valid_case : valid_cases) {
     options.daily_offpeak_time_utc = valid_case;
     verify_valid();
   }
 
-  auto verify_is_now_offpeak = [&](bool expected, int now_utc_hour,
-                                   int now_utc_minute, int now_utc_second = 0) {
+  auto verify_offpeak_info = [&](bool expected_is_now_off_peak,
+                                 int expected_seconds_till_next_offpeak_start,
+                                 int now_utc_hour, int now_utc_minute,
+                                 int now_utc_second = 0) {
     auto mock_clock = std::make_shared<MockSystemClock>(env_->GetSystemClock());
     // Add some extra random days to current time
     int days = rnd.Uniform(100);
-    mock_clock->SetCurrentTime(days * 86400 + now_utc_hour * 3600 +
-                               now_utc_minute * 60 + now_utc_second);
+    mock_clock->SetCurrentTime(
+        days * OffpeakTimeOption::kSecondsPerDay +
+        now_utc_hour * OffpeakTimeOption::kSecondsPerHour +
+        now_utc_minute * OffpeakTimeOption::kSecondsPerMinute + now_utc_second);
     Status s = DBImpl::TEST_ValidateOptions(options);
     ASSERT_OK(s);
-    auto db_options = MutableDBOptions(options);
-    ASSERT_EQ(expected, db_options.IsNowOffPeak(mock_clock.get()));
+    auto offpeak_option = OffpeakTimeOption(options.daily_offpeak_time_utc);
+    int64_t now;
+    ASSERT_OK(mock_clock.get()->GetCurrentTime(&now));
+    auto offpeak_info = offpeak_option.GetOffpeakTimeInfo(now);
+    ASSERT_EQ(expected_is_now_off_peak, offpeak_info.is_now_offpeak);
+    ASSERT_EQ(expected_seconds_till_next_offpeak_start,
+              offpeak_info.seconds_till_next_offpeak_start);
   };
 
   options.daily_offpeak_time_utc = "";
-  verify_is_now_offpeak(false, 12, 30);
+  verify_offpeak_info(false, 0, 12, 30);
 
   options.daily_offpeak_time_utc = "06:30-11:30";
-  verify_is_now_offpeak(false, 5, 30);
-  verify_is_now_offpeak(true, 6, 30);
-  verify_is_now_offpeak(true, 10, 30);
-  verify_is_now_offpeak(true, 11, 30);
-  verify_is_now_offpeak(false, 13, 30);
+  verify_offpeak_info(false, 1 * OffpeakTimeOption::kSecondsPerHour, 5, 30);
+  verify_offpeak_info(true, 24 * OffpeakTimeOption::kSecondsPerHour, 6, 30);
+  verify_offpeak_info(true, 20 * OffpeakTimeOption::kSecondsPerHour, 10, 30);
+  verify_offpeak_info(true, 19 * OffpeakTimeOption::kSecondsPerHour, 11, 30);
+  verify_offpeak_info(false, 17 * OffpeakTimeOption::kSecondsPerHour, 13, 30);
 
   options.daily_offpeak_time_utc = "23:30-04:30";
-  verify_is_now_offpeak(false, 6, 30);
-  verify_is_now_offpeak(true, 23, 30);
-  verify_is_now_offpeak(true, 0, 0);
-  verify_is_now_offpeak(true, 1, 0);
-  verify_is_now_offpeak(true, 4, 30);
-  verify_is_now_offpeak(false, 4, 31);
+  verify_offpeak_info(false, 17 * OffpeakTimeOption::kSecondsPerHour, 6, 30);
+  verify_offpeak_info(true, 24 * OffpeakTimeOption::kSecondsPerHour, 23, 30);
+  verify_offpeak_info(true,
+                      23 * OffpeakTimeOption::kSecondsPerHour +
+                          30 * OffpeakTimeOption::kSecondsPerMinute,
+                      0, 0);
+  verify_offpeak_info(true,
+                      22 * OffpeakTimeOption::kSecondsPerHour +
+                          30 * OffpeakTimeOption::kSecondsPerMinute,
+                      1, 0);
+  verify_offpeak_info(true, 19 * OffpeakTimeOption::kSecondsPerHour, 4, 30);
+  verify_offpeak_info(false,
+                      18 * OffpeakTimeOption::kSecondsPerHour +
+                          59 * OffpeakTimeOption::kSecondsPerMinute,
+                      4, 31);
 
   // Entire day offpeak
   options.daily_offpeak_time_utc = "00:00-23:59";
-  verify_is_now_offpeak(true, 0, 0);
-  verify_is_now_offpeak(true, 12, 00);
-  verify_is_now_offpeak(true, 23, 59);
-  verify_is_now_offpeak(true, 23, 59, 1);
-  verify_is_now_offpeak(true, 23, 59, 59);
+  verify_offpeak_info(true, 24 * OffpeakTimeOption::kSecondsPerHour, 0, 0);
+  verify_offpeak_info(true, 12 * OffpeakTimeOption::kSecondsPerHour, 12, 00);
+  verify_offpeak_info(true, 1 * OffpeakTimeOption::kSecondsPerMinute, 23, 59);
+  verify_offpeak_info(true, 59, 23, 59, 1);
+  verify_offpeak_info(true, 1, 23, 59, 59);
 
-  // Open the db and test by Get/SetDBOptions
-  options.daily_offpeak_time_utc = "";
+  // Start with a valid option
+  options.daily_offpeak_time_utc = "01:30-04:15";
   DestroyAndReopen(options);
-  ASSERT_EQ("", dbfull()->GetDBOptions().daily_offpeak_time_utc);
+  ASSERT_EQ("01:30-04:15", dbfull()->GetDBOptions().daily_offpeak_time_utc);
+
+  int may_schedule_compaction_called = 0;
+  SyncPoint::GetInstance()->SetCallBack(
+      "DBImpl::MaybeScheduleFlushOrCompaction:Start",
+      [&](void*) { may_schedule_compaction_called++; });
+  SyncPoint::GetInstance()->EnableProcessing();
+
+  // Make sure calling SetDBOptions with invalid option does not change the
+  // value nor call MaybeScheduleFlushOrCompaction()
   for (std::string invalid_case : invalid_cases) {
     ASSERT_NOK(
         dbfull()->SetDBOptions({{"daily_offpeak_time_utc", invalid_case}}));
+    ASSERT_EQ("01:30-04:15", dbfull()
+                                 ->GetVersionSet()
+                                 ->offpeak_time_option()
+                                 .daily_offpeak_time_utc);
+    ASSERT_EQ(1 * kSecondInHour + 30 * kSecondInMinute,
+              dbfull()
+                  ->GetVersionSet()
+                  ->offpeak_time_option()
+                  .daily_offpeak_start_time_utc);
+    ASSERT_EQ(4 * kSecondInHour + 15 * kSecondInMinute,
+              dbfull()
+                  ->GetVersionSet()
+                  ->offpeak_time_option()
+                  .daily_offpeak_end_time_utc);
   }
+  ASSERT_EQ(0, may_schedule_compaction_called);
+
+  // Changing to new valid values should call MaybeScheduleFlushOrCompaction()
+  // and sets the offpeak_time_option in VersionSet
+  int expected_count = 0;
   for (std::string valid_case : valid_cases) {
+    if (dbfull()
+            ->GetVersionSet()
+            ->offpeak_time_option()
+            .daily_offpeak_time_utc != valid_case) {
+      expected_count++;
+    }
     ASSERT_OK(dbfull()->SetDBOptions({{"daily_offpeak_time_utc", valid_case}}));
     ASSERT_EQ(valid_case, dbfull()->GetDBOptions().daily_offpeak_time_utc);
+    ASSERT_EQ(valid_case, dbfull()
+                              ->GetVersionSet()
+                              ->offpeak_time_option()
+                              .daily_offpeak_time_utc);
   }
-  Close();
+  ASSERT_EQ(expected_count, may_schedule_compaction_called);
 
-  // Sets off-peak time from 11:30PM to 4:30AM next day.
-  // Starting at 1:30PM, use mock sleep to make time pass
-  // and see if IsNowOffPeak() returns correctly per time changes
-  int now_hour = 13;
-  int now_minute = 30;
-  options.daily_offpeak_time_utc = "23:30-04:30";
-  auto mock_clock = std::make_shared<MockSystemClock>(env_->GetSystemClock());
-  auto mock_env = std::make_unique<CompositeEnvWrapper>(env_, mock_clock);
-  // Add some extra random days to current time
-  int days = rnd.Uniform(100);
-  mock_clock->SetCurrentTime(days * 86400 + now_hour * 3600 + now_minute * 60);
-  options.env = mock_env.get();
-
-  // Starting at 1:30PM. It's not off-peak
-  DestroyAndReopen(options);
-  ASSERT_FALSE(MutableDBOptions(dbfull()->GetDBOptions())
-                   .IsNowOffPeak(mock_clock.get()));
-
-  // Now it's at 4:30PM. Still not off-peak
-  mock_clock->MockSleepForSeconds(3 * 3600);
-  ASSERT_FALSE(MutableDBOptions(dbfull()->GetDBOptions())
-                   .IsNowOffPeak(mock_clock.get()));
-
-  // Now it's at 11:30PM. It's off-peak
-  mock_clock->MockSleepForSeconds(7 * 3600);
-  ASSERT_TRUE(MutableDBOptions(dbfull()->GetDBOptions())
-                  .IsNowOffPeak(mock_clock.get()));
-
-  // Now it's at 2:30AM next day. It's still off-peak
-  mock_clock->MockSleepForSeconds(3 * 3600);
-  ASSERT_TRUE(MutableDBOptions(dbfull()->GetDBOptions())
-                  .IsNowOffPeak(mock_clock.get()));
-
-  // Now it's at 4:30AM. It's still off-peak
-  mock_clock->MockSleepForSeconds(2 * 3600);
-  ASSERT_TRUE(MutableDBOptions(dbfull()->GetDBOptions())
-                  .IsNowOffPeak(mock_clock.get()));
-
-  // Sleep for one more minute. It's at 4:31AM It's no longer off-peak
-  mock_clock->MockSleepForSeconds(60);
-  ASSERT_FALSE(MutableDBOptions(dbfull()->GetDBOptions())
-                   .IsNowOffPeak(mock_clock.get()));
-  Close();
-
-  // Entire day offpeak
-  options.daily_offpeak_time_utc = "00:00-23:59";
-  DestroyAndReopen(options);
-  // It doesn't matter what time it is. It should be just offpeak.
-  ASSERT_TRUE(MutableDBOptions(dbfull()->GetDBOptions())
-                  .IsNowOffPeak(mock_clock.get()));
-
-  // Mock Sleep for 3 hours. It's still off-peak
-  mock_clock->MockSleepForSeconds(3 * 3600);
-  ASSERT_TRUE(MutableDBOptions(dbfull()->GetDBOptions())
-                  .IsNowOffPeak(mock_clock.get()));
-
-  // Mock Sleep for 20 hours. It's still off-peak
-  mock_clock->MockSleepForSeconds(20 * 3600);
-  ASSERT_TRUE(MutableDBOptions(dbfull()->GetDBOptions())
-                  .IsNowOffPeak(mock_clock.get()));
-
-  // Mock Sleep for 59 minutes. It's still off-peak
-  mock_clock->MockSleepForSeconds(59 * 60);
-  ASSERT_TRUE(MutableDBOptions(dbfull()->GetDBOptions())
-                  .IsNowOffPeak(mock_clock.get()));
-
-  // Mock Sleep for 59 seconds. It's still off-peak
-  mock_clock->MockSleepForSeconds(59);
-  ASSERT_TRUE(MutableDBOptions(dbfull()->GetDBOptions())
-                  .IsNowOffPeak(mock_clock.get()));
-
-  // Mock Sleep for 1 second (exactly 24h passed). It's still off-peak
-  mock_clock->MockSleepForSeconds(1);
-  ASSERT_TRUE(MutableDBOptions(dbfull()->GetDBOptions())
-                  .IsNowOffPeak(mock_clock.get()));
-  // Another second for sanity check
-  mock_clock->MockSleepForSeconds(1);
-  ASSERT_TRUE(MutableDBOptions(dbfull()->GetDBOptions())
-                  .IsNowOffPeak(mock_clock.get()));
+  // Changing to the same value should not call MaybeScheduleFlushOrCompaction()
+  ASSERT_OK(
+      dbfull()->SetDBOptions({{"daily_offpeak_time_utc", "06:30-11:30"}}));
+  may_schedule_compaction_called = 0;
+  ASSERT_OK(
+      dbfull()->SetDBOptions({{"daily_offpeak_time_utc", "06:30-11:30"}}));
+  ASSERT_EQ(0, may_schedule_compaction_called);
 
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->ClearAllCallBacks();
   Close();
 }
 

db/db_properties_test.cc

@@ -7,9 +7,8 @@
 // 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 <stdio.h>
-
 #include <algorithm>
+#include <cstdio>
 #include <string>
 
 #include "db/db_test_util.h"
@@ -107,12 +106,12 @@ TEST_F(DBPropertiesTest, Empty) {
         dbfull()->GetProperty("rocksdb.is-file-deletions-enabled", &num));
     ASSERT_EQ("0", num);
 
-    ASSERT_OK(db_->EnableFileDeletions(false));
+    ASSERT_OK(db_->EnableFileDeletions(/*force=*/false));
     ASSERT_TRUE(
         dbfull()->GetProperty("rocksdb.is-file-deletions-enabled", &num));
     ASSERT_EQ("0", num);
 
-    ASSERT_OK(db_->EnableFileDeletions());
+    ASSERT_OK(db_->EnableFileDeletions(/*force=*/true));
     ASSERT_TRUE(
         dbfull()->GetProperty("rocksdb.is-file-deletions-enabled", &num));
     ASSERT_EQ("1", num);
@@ -1084,12 +1083,14 @@ class CountingUserTblPropCollector : public TablePropertiesCollector {
   const char* Name() const override { return "CountingUserTblPropCollector"; }
 
   Status Finish(UserCollectedProperties* properties) override {
+    assert(!finish_called_);
     std::string encoded;
     PutVarint32(&encoded, count_);
     *properties = UserCollectedProperties{
         {"CountingUserTblPropCollector", message_},
         {"Count", encoded},
     };
+    finish_called_ = true;
     return Status::OK();
   }
 
@@ -1101,12 +1102,14 @@ class CountingUserTblPropCollector : public TablePropertiesCollector {
   }
 
   UserCollectedProperties GetReadableProperties() const override {
+    assert(finish_called_);
     return UserCollectedProperties{};
   }
 
  private:
   std::string message_ = "Rocksdb";
   uint32_t count_ = 0;
+  bool finish_called_ = false;
 };
 
 class CountingUserTblPropCollectorFactory
@@ -1740,7 +1743,7 @@ TEST_F(DBPropertiesTest, SstFilesSize) {
   ASSERT_EQ(obsolete_sst_size, sst_size);
 
   // Let the obsolete files be deleted.
-  ASSERT_OK(db_->EnableFileDeletions());
+  ASSERT_OK(db_->EnableFileDeletions(/*force=*/false));
   ASSERT_TRUE(db_->GetIntProperty(DB::Properties::kObsoleteSstFilesSize,
                                   &obsolete_sst_size));
   ASSERT_EQ(obsolete_sst_size, 0);

db/db_range_del_test.cc

@@ -22,7 +22,7 @@ class DBRangeDelTest : public DBTestBase {
     uint64_t uint64_key = static_cast<uint64_t>(key);
     std::string str;
     str.resize(8);
-    memcpy(&str[0], static_cast<void*>(&uint64_key), 8);
+    memcpy(str.data(), static_cast<void*>(&uint64_key), 8);
     return str;
   }
 };

db/db_secondary_test.cc

@@ -103,7 +103,7 @@ void DBSecondaryTestBase::CheckFileTypeCounts(const std::string& dir,
   ASSERT_OK(env_->GetChildren(dir, &filenames));
 
   int log_cnt = 0, sst_cnt = 0, manifest_cnt = 0;
-  for (auto file : filenames) {
+  for (const auto& file : filenames) {
     uint64_t number;
     FileType type;
     if (ParseFileName(file, &number, &type)) {