x

* fix: correct invalid testing feature gate usage

* test: refactor tests to avoid test code leak

* fix: sync main

.github/actions/build-greptime-binary/action.yml

@@ -54,7 +54,7 @@ runs:
         PROFILE_TARGET: ${{ inputs.cargo-profile == 'dev' && 'debug' || inputs.cargo-profile }}
       with:
         artifacts-dir: ${{ inputs.artifacts-dir }}
-        target-file: ./target/$PROFILE_TARGET/greptime
+        target-files: ./target/$PROFILE_TARGET/greptime
         version: ${{ inputs.version }}
         working-dir: ${{ inputs.working-dir }}
 
@@ -72,6 +72,6 @@ runs:
       if: ${{ inputs.build-android-artifacts == 'true' }}
       with:
         artifacts-dir: ${{ inputs.artifacts-dir }}
-        target-file: ./target/aarch64-linux-android/release/greptime
+        target-files: ./target/aarch64-linux-android/release/greptime
         version: ${{ inputs.version }}
         working-dir: ${{ inputs.working-dir }}

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

@@ -41,8 +41,8 @@ runs:
         image-name: ${{ inputs.image-name }}
         image-tag: ${{ inputs.version }}
         docker-file: docker/ci/ubuntu/Dockerfile
-        amd64-artifact-name: greptime-linux-amd64-pyo3-${{ inputs.version }}
-        arm64-artifact-name: greptime-linux-arm64-pyo3-${{ inputs.version }}
+        amd64-artifact-name: greptime-linux-amd64-${{ inputs.version }}
+        arm64-artifact-name: greptime-linux-arm64-${{ inputs.version }}
         platforms: linux/amd64,linux/arm64
         push-latest-tag: ${{ inputs.push-latest-tag }}
 

.github/actions/build-linux-artifacts/action.yml

@@ -48,19 +48,7 @@ runs:
         path: /tmp/greptime-*.log
         retention-days: 3
 
-    - name: Build standard greptime
-      uses: ./.github/actions/build-greptime-binary
-      with:
-        base-image: ubuntu
-        features: pyo3_backend,servers/dashboard
-        cargo-profile: ${{ inputs.cargo-profile }}
-        artifacts-dir: greptime-linux-${{ inputs.arch }}-pyo3-${{ inputs.version }}
-        version: ${{ inputs.version }}
-        working-dir: ${{ inputs.working-dir }}
-        image-registry: ${{ inputs.image-registry }}
-        image-namespace: ${{ inputs.image-namespace }}
-
-    - name: Build greptime without pyo3
+    - name: Build greptime
       if: ${{ inputs.dev-mode == 'false' }}
       uses: ./.github/actions/build-greptime-binary
       with:

.github/actions/build-macos-artifacts/action.yml

@@ -90,5 +90,5 @@ runs:
       uses: ./.github/actions/upload-artifacts
       with:
         artifacts-dir: ${{ inputs.artifacts-dir }}
-        target-file: target/${{ inputs.arch }}/${{ inputs.cargo-profile }}/greptime
+        target-files: target/${{ inputs.arch }}/${{ inputs.cargo-profile }}/greptime
         version: ${{ inputs.version }}

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

@@ -33,15 +33,6 @@ runs:
     - name: Rust Cache
       uses: Swatinem/rust-cache@v2
 
-    - name: Install Python
-      uses: actions/setup-python@v5
-      with:
-        python-version: "3.10"
-
-    - name: Install PyArrow Package
-      shell: pwsh
-      run: pip install pyarrow numpy
-
     - name: Install WSL distribution
       uses: Vampire/setup-wsl@v2
       with:
@@ -76,5 +67,5 @@ runs:
       uses: ./.github/actions/upload-artifacts
       with:
         artifacts-dir: ${{ inputs.artifacts-dir }}
-        target-file: target/${{ inputs.arch }}/${{ inputs.cargo-profile }}/greptime
+        target-files: target/${{ inputs.arch }}/${{ inputs.cargo-profile }}/greptime,target/${{ inputs.arch }}/${{ inputs.cargo-profile }}/greptime.pdb
         version: ${{ inputs.version }}

.github/actions/setup-greptimedb-cluster/action.yml

@@ -8,7 +8,7 @@ inputs:
     default: 2
     description: "Number of Datanode replicas"
   meta-replicas:
-    default: 3
+    default: 1
     description: "Number of Metasrv replicas"
   image-registry: 
     default: "docker.io"
@@ -58,7 +58,7 @@ runs:
         --set image.tag=${{ inputs.image-tag }} \
         --set base.podTemplate.main.resources.requests.cpu=50m \
         --set base.podTemplate.main.resources.requests.memory=256Mi \
-        --set base.podTemplate.main.resources.limits.cpu=1000m \
+        --set base.podTemplate.main.resources.limits.cpu=2000m \
         --set base.podTemplate.main.resources.limits.memory=2Gi \
         --set frontend.replicas=${{ inputs.frontend-replicas }} \
         --set datanode.replicas=${{ inputs.datanode-replicas }} \

.github/actions/setup-greptimedb-cluster/with-minio-and-cache.yaml

@@ -5,7 +5,7 @@ meta:
 
     [datanode]
     [datanode.client]
-    timeout = "60s"
+    timeout = "120s"
 datanode:
   configData: |-
     [runtime]
@@ -21,7 +21,7 @@ frontend:
     global_rt_size = 4
 
     [meta_client]
-    ddl_timeout = "60s"
+    ddl_timeout = "120s"
 objectStorage:
   s3:
     bucket: default

.github/actions/setup-greptimedb-cluster/with-minio.yaml

@@ -5,7 +5,7 @@ meta:
     
     [datanode]
     [datanode.client]
-    timeout = "60s"
+    timeout = "120s"
 datanode:
   configData: |-
     [runtime]
@@ -17,7 +17,7 @@ frontend:
     global_rt_size = 4
 
     [meta_client]
-    ddl_timeout = "60s"
+    ddl_timeout = "120s"
 objectStorage:
   s3:
     bucket: default

.github/actions/setup-greptimedb-cluster/with-remote-wal.yaml

@@ -11,7 +11,7 @@ meta:
         
     [datanode]
     [datanode.client]
-    timeout = "60s"
+    timeout = "120s"
 datanode:
   configData: |-
     [runtime]
@@ -28,7 +28,7 @@ frontend:
     global_rt_size = 4
 
     [meta_client]
-    ddl_timeout = "60s"
+    ddl_timeout = "120s"
 objectStorage:
   s3:
     bucket: default

.github/actions/setup-kafka-cluster/action.yml

@@ -18,6 +18,8 @@ runs:
         --set controller.replicaCount=${{ inputs.controller-replicas }} \
         --set controller.resources.requests.cpu=50m \
         --set controller.resources.requests.memory=128Mi \
+        --set controller.resources.limits.cpu=2000m \
+        --set controller.resources.limits.memory=2Gi \
         --set listeners.controller.protocol=PLAINTEXT \
         --set listeners.client.protocol=PLAINTEXT \
         --create-namespace \

.github/actions/upload-artifacts/action.yml

@@ -4,8 +4,8 @@ inputs:
   artifacts-dir:
     description: Directory to store artifacts
     required: true
-  target-file:
-    description: The path of the target artifact
+  target-files:
+    description: The multiple target files to upload, separated by comma
     required: false
   version:
     description: Version of the artifact
@@ -18,12 +18,16 @@ runs:
   using: composite
   steps:
     - name: Create artifacts directory
-      if: ${{ inputs.target-file != '' }}
+      if: ${{ inputs.target-files != '' }}
       working-directory: ${{ inputs.working-dir }}
       shell: bash
       run: |
-        mkdir -p ${{ inputs.artifacts-dir }} && \
-        cp ${{ inputs.target-file }} ${{ inputs.artifacts-dir }}
+        set -e
+        mkdir -p ${{ inputs.artifacts-dir }}
+        IFS=',' read -ra FILES <<< "${{ inputs.target-files }}"
+        for file in "${FILES[@]}"; do
+          cp "$file" ${{ inputs.artifacts-dir }}/
+        done
 
     # The compressed artifacts will use the following layout:
     # greptime-linux-amd64-pyo3-v0.3.0sha256sum

.github/cargo-blacklist.txt

@@ -0,0 +1,3 @@
+native-tls
+openssl
+aws-lc-sys

.github/pull_request_template.md

@@ -4,7 +4,8 @@ I hereby agree to the terms of the [GreptimeDB CLA](https://github.com/GreptimeT
 
 ## What's changed and what's your intention?
 
-__!!! DO NOT LEAVE THIS BLOCK EMPTY !!!__
+<!--    
+ __!!! DO NOT LEAVE THIS BLOCK EMPTY !!!__
 
 Please explain IN DETAIL what the changes are in this PR and why they are needed:
 
@@ -12,9 +13,14 @@ Please explain IN DETAIL what the changes are in this PR and why they are needed
 - How does this PR work? Need a brief introduction for the changed logic (optional)
 - Describe clearly one logical change and avoid lazy messages (optional)
 - Describe any limitations of the current code (optional)
+- Describe if this PR will break **API or data compatibility**  (optional)
+-->
 
-## Checklist
+## PR Checklist
+Please convert it to a draft if some of the following conditions are not met.
 
 - [ ] I have written the necessary rustdoc comments.
 - [ ] I have added the necessary unit tests and integration tests.
 - [ ] This PR requires documentation updates.
+- [ ] API changes are backward compatible.
+- [ ] Schema or data changes are backward compatible.

.github/workflows/dependency-check.yml

@@ -0,0 +1,36 @@
+name: Check Dependencies
+
+on:
+  push:
+    branches:
+      - main
+  pull_request:
+    branches:
+      - main
+
+jobs:
+  check-dependencies:
+    runs-on: ubuntu-latest
+
+    steps:
+    - name: Checkout code
+      uses: actions/checkout@v4
+
+    - name: Set up Rust
+      uses: actions-rust-lang/setup-rust-toolchain@v1
+
+    - name: Run cargo tree
+      run: cargo tree --prefix none > dependencies.txt
+
+    - name: Extract dependency names
+      run: awk '{print $1}' dependencies.txt > dependency_names.txt
+
+    - name: Check for blacklisted crates
+      run: |
+        while read -r dep; do
+          if grep -qFx "$dep" dependency_names.txt; then
+            echo "Blacklisted crate '$dep' found in dependencies."
+            exit 1
+          fi
+        done < .github/cargo-blacklist.txt
+        echo "No blacklisted crates found."

.github/workflows/develop.yml

@@ -10,17 +10,6 @@ on:
       - 'docker/**'
       - '.gitignore'
       - 'grafana/**'
-  push:
-    branches:
-      - main
-    paths-ignore:
-      - 'docs/**'
-      - 'config/**'
-      - '**.md'
-      - '.dockerignore'
-      - 'docker/**'
-      - '.gitignore'
-      - 'grafana/**'
   workflow_dispatch:
 
 name: CI
@@ -84,7 +73,7 @@ jobs:
           # Shares across multiple jobs
           shared-key: "check-toml"
       - name: Install taplo
-        run: cargo +stable install taplo-cli --version ^0.9 --locked
+        run: cargo +stable install taplo-cli --version ^0.9 --locked --force
       - name: Run taplo
         run: taplo format --check
 
@@ -107,7 +96,7 @@ jobs:
           shared-key: "build-binaries"
       - name: Install cargo-gc-bin
         shell: bash
-        run: cargo install cargo-gc-bin
+        run: cargo install cargo-gc-bin --force
       - name: Build greptime binaries
         shell: bash
         # `cargo gc` will invoke `cargo build` with specified args
@@ -163,7 +152,7 @@ jobs:
         run: |
           sudo apt-get install -y libfuzzer-14-dev
           rustup install nightly
-          cargo +nightly install cargo-fuzz cargo-gc-bin
+          cargo +nightly install cargo-fuzz cargo-gc-bin --force
       - name: Download pre-built binaries
         uses: actions/download-artifact@v4
         with:
@@ -220,7 +209,7 @@ jobs:
         shell: bash
         run: |
           sudo apt update && sudo apt install -y libfuzzer-14-dev
-          cargo install cargo-fuzz cargo-gc-bin
+          cargo install cargo-fuzz cargo-gc-bin --force
       - name: Download pre-built binariy
         uses: actions/download-artifact@v4
         with:
@@ -268,14 +257,7 @@ jobs:
           shared-key: "build-greptime-ci"
       - name: Install cargo-gc-bin
         shell: bash
-        run: cargo install cargo-gc-bin
-      - name: Check aws-lc-sys will not build
-        shell: bash
-        run: |
-             if cargo tree -i aws-lc-sys -e features | grep -q aws-lc-sys; then
-               echo "Found aws-lc-sys, which has compilation problems on older gcc versions. Please replace it with ring until its building experience improves."
-               exit 1
-             fi
+        run: cargo install cargo-gc-bin --force
       - name: Build greptime bianry
         shell: bash
         # `cargo gc` will invoke `cargo build` with specified args
@@ -330,8 +312,6 @@ jobs:
         uses: ./.github/actions/setup-kafka-cluster
       - name: Setup Etcd cluser
         uses: ./.github/actions/setup-etcd-cluster
-      - name: Setup Postgres cluser
-        uses: ./.github/actions/setup-postgres-cluster
       # Prepares for fuzz tests
       - uses: arduino/setup-protoc@v3
         with:
@@ -347,7 +327,7 @@ jobs:
         run: |
           sudo apt-get install -y libfuzzer-14-dev
           rustup install nightly
-          cargo +nightly install cargo-fuzz cargo-gc-bin
+          cargo +nightly install cargo-fuzz cargo-gc-bin --force
       # Downloads ci image
       - name: Download pre-built binariy
         uses: actions/download-artifact@v4
@@ -481,8 +461,6 @@ jobs:
         uses: ./.github/actions/setup-kafka-cluster
       - name: Setup Etcd cluser
         uses: ./.github/actions/setup-etcd-cluster
-      - name: Setup Postgres cluser
-        uses: ./.github/actions/setup-postgres-cluster
       # Prepares for fuzz tests
       - uses: arduino/setup-protoc@v3
         with:
@@ -498,7 +476,7 @@ jobs:
         run: |
           sudo apt-get install -y libfuzzer-14-dev
           rustup install nightly
-          cargo +nightly install cargo-fuzz cargo-gc-bin
+          cargo +nightly install cargo-fuzz cargo-gc-bin --force
       # Downloads ci image
       - name: Download pre-built binariy
         uses: actions/download-artifact@v4
@@ -664,6 +642,7 @@ jobs:
     if: github.event.pull_request.draft == false
     runs-on: ubuntu-20.04-8-cores
     timeout-minutes: 60
+    needs:  [clippy, fmt]
     steps:
       - uses: actions/checkout@v4
       - uses: arduino/setup-protoc@v3
@@ -689,12 +668,6 @@ jobs:
         uses: taiki-e/install-action@nextest
       - name: Install cargo-llvm-cov
         uses: taiki-e/install-action@cargo-llvm-cov
-      - name: Install Python
-        uses: actions/setup-python@v5
-        with:
-          python-version: '3.10'
-      - name: Install PyArrow Package
-        run: pip install pyarrow numpy
       - name: Setup etcd server
         working-directory: tests-integration/fixtures/etcd
         run: docker compose -f docker-compose-standalone.yml up -d --wait
@@ -708,7 +681,7 @@ jobs:
         working-directory: tests-integration/fixtures/postgres
         run: docker compose -f docker-compose-standalone.yml up -d --wait
       - name: Run nextest cases
-        run: cargo llvm-cov nextest --workspace --lcov --output-path lcov.info -F pyo3_backend -F dashboard
+        run: cargo llvm-cov nextest --workspace --lcov --output-path lcov.info -F dashboard -F pg_kvbackend
         env:
           CARGO_BUILD_RUSTFLAGS: "-C link-arg=-fuse-ld=lld"
           RUST_BACKTRACE: 1

.github/workflows/nightly-build.yml

@@ -12,7 +12,7 @@ on:
       linux_amd64_runner:
         type: choice
         description: The runner uses to build linux-amd64 artifacts
-        default: ec2-c6i.2xlarge-amd64
+        default: ec2-c6i.4xlarge-amd64
         options:
           - ubuntu-20.04
           - ubuntu-20.04-8-cores
@@ -27,7 +27,7 @@ on:
       linux_arm64_runner:
         type: choice
         description: The runner uses to build linux-arm64 artifacts
-        default: ec2-c6g.2xlarge-arm64
+        default: ec2-c6g.4xlarge-arm64
         options:
           - ec2-c6g.xlarge-arm64 # 4C8G
           - ec2-c6g.2xlarge-arm64 # 8C16G

.github/workflows/nightly-ci.yml

@@ -1,6 +1,6 @@
 on:
   schedule:
-    - cron: "0 23 * * 1-5"
+    - cron: "0 23 * * 1-4"
   workflow_dispatch:
 
 name: Nightly CI
@@ -91,18 +91,12 @@ jobs:
         uses: Swatinem/rust-cache@v2
       - name: Install Cargo Nextest
         uses: taiki-e/install-action@nextest
-      - name: Install Python
-        uses: actions/setup-python@v5
-        with:
-          python-version: "3.10"
-      - name: Install PyArrow Package
-        run: pip install pyarrow numpy
       - name: Install WSL distribution
         uses: Vampire/setup-wsl@v2
         with:
           distribution: Ubuntu-22.04
       - name: Running tests
-        run: cargo nextest run -F pyo3_backend,dashboard
+        run: cargo nextest run -F dashboard
         env:
           CARGO_BUILD_RUSTFLAGS: "-C linker=lld-link"
           RUST_BACKTRACE: 1
@@ -114,6 +108,16 @@ jobs:
           GT_S3_REGION: ${{ vars.AWS_CI_TEST_BUCKET_REGION }}
           UNITTEST_LOG_DIR: "__unittest_logs"
 
+  cleanbuild-linux-nix:
+    runs-on: ubuntu-latest-8-cores
+    timeout-minutes: 60
+    steps:
+      - uses: actions/checkout@v4
+      - uses: cachix/install-nix-action@v27
+        with:
+          nix_path: nixpkgs=channel:nixos-unstable
+      - run: nix-shell --pure --run "cargo build"
+
   check-status:
     name: Check status
     needs: [sqlness-test, sqlness-windows, test-on-windows]

.github/workflows/release.yml

@@ -31,7 +31,7 @@ on:
       linux_arm64_runner:
         type: choice
         description: The runner uses to build linux-arm64 artifacts
-        default: ec2-c6g.4xlarge-arm64
+        default: ec2-c6g.8xlarge-arm64
         options:
           - ubuntu-2204-32-cores-arm
           - ec2-c6g.xlarge-arm64 # 4C8G
@@ -91,7 +91,7 @@ env:
   # The scheduled version is '${{ env.NEXT_RELEASE_VERSION }}-nightly-YYYYMMDD', like v0.2.0-nigthly-20230313;
   NIGHTLY_RELEASE_PREFIX: nightly
   # Note: The NEXT_RELEASE_VERSION should be modified manually by every formal release.
-  NEXT_RELEASE_VERSION: v0.11.0
+  NEXT_RELEASE_VERSION: v0.12.0
 
 # Permission reference: https://docs.github.com/en/actions/using-jobs/assigning-permissions-to-jobs
 permissions:
@@ -222,18 +222,10 @@ jobs:
             arch: aarch64-apple-darwin
             features: servers/dashboard
             artifacts-dir-prefix: greptime-darwin-arm64
-          - os: ${{ needs.allocate-runners.outputs.macos-runner }}
-            arch: aarch64-apple-darwin
-            features: pyo3_backend,servers/dashboard
-            artifacts-dir-prefix: greptime-darwin-arm64-pyo3
           - os: ${{ needs.allocate-runners.outputs.macos-runner }}
             features: servers/dashboard
             arch: x86_64-apple-darwin
             artifacts-dir-prefix: greptime-darwin-amd64
-          - os: ${{ needs.allocate-runners.outputs.macos-runner }}
-            features: pyo3_backend,servers/dashboard
-            arch: x86_64-apple-darwin
-            artifacts-dir-prefix: greptime-darwin-amd64-pyo3
     runs-on: ${{ matrix.os }}
     outputs:
       build-macos-result: ${{ steps.set-build-macos-result.outputs.build-macos-result }}
@@ -271,10 +263,6 @@ jobs:
             arch: x86_64-pc-windows-msvc
             features: servers/dashboard
             artifacts-dir-prefix: greptime-windows-amd64
-          - os: ${{ needs.allocate-runners.outputs.windows-runner }}
-            arch: x86_64-pc-windows-msvc
-            features: pyo3_backend,servers/dashboard
-            artifacts-dir-prefix: greptime-windows-amd64-pyo3
     runs-on: ${{ matrix.os }}
     outputs:
       build-windows-result: ${{ steps.set-build-windows-result.outputs.build-windows-result }}

.gitignore

@@ -47,6 +47,10 @@ benchmarks/data
 
 venv/
 
-# Fuzz tests 
+# Fuzz tests
 tests-fuzz/artifacts/
 tests-fuzz/corpus/
+
+# Nix
+.direnv
+.envrc

AUTHOR.md

@@ -7,6 +7,8 @@
 * [NiwakaDev](https://github.com/NiwakaDev)
 * [etolbakov](https://github.com/etolbakov)
 * [irenjj](https://github.com/irenjj)
+* [tisonkun](https://github.com/tisonkun)
+* [Lanqing Yang](https://github.com/lyang24)
 
 ## Team Members (in alphabetical order)
 
@@ -30,7 +32,6 @@
 * [shuiyisong](https://github.com/shuiyisong)
 * [sunchanglong](https://github.com/sunchanglong)
 * [sunng87](https://github.com/sunng87)
-* [tisonkun](https://github.com/tisonkun)
 * [v0y4g3r](https://github.com/v0y4g3r)
 * [waynexia](https://github.com/waynexia)
 * [xtang](https://github.com/xtang)

Cargo.toml

@@ -4,6 +4,7 @@ members = [
     "src/auth",
     "src/cache",
     "src/catalog",
+    "src/cli",
     "src/client",
     "src/cmd",
     "src/common/base",
@@ -40,6 +41,7 @@ members = [
     "src/flow",
     "src/frontend",
     "src/index",
+    "src/log-query",
     "src/log-store",
     "src/meta-client",
     "src/meta-srv",
@@ -66,7 +68,7 @@ members = [
 resolver = "2"
 
 [workspace.package]
-version = "0.10.0"
+version = "0.12.0"
 edition = "2021"
 license = "Apache-2.0"
 
@@ -124,6 +126,7 @@ futures = "0.3"
 futures-util = "0.3"
 greptime-proto = { git = "https://github.com/GreptimeTeam/greptime-proto.git", rev = "a875e976441188028353f7274a46a7e6e065c5d4" }
 hex = "0.4"
+http = "0.2"
 humantime = "2.1"
 humantime-serde = "1.1"
 itertools = "0.10"
@@ -132,6 +135,7 @@ lazy_static = "1.4"
 meter-core = { git = "https://github.com/GreptimeTeam/greptime-meter.git", rev = "a10facb353b41460eeb98578868ebf19c2084fac" }
 mockall = "0.11.4"
 moka = "0.12"
+nalgebra = "0.33"
 notify = "6.1"
 num_cpus = "1.16"
 once_cell = "1.18"
@@ -167,7 +171,6 @@ rstest = "0.21"
 rstest_reuse = "0.7"
 rust_decimal = "1.33"
 rustc-hash = "2.0"
-schemars = "0.8"
 serde = { version = "1.0", features = ["derive"] }
 serde_json = { version = "1.0", features = ["float_roundtrip"] }
 serde_with = "3"
@@ -179,6 +182,7 @@ sysinfo = "0.30"
 # on branch v0.44.x
 sqlparser = { git = "https://github.com/GreptimeTeam/sqlparser-rs.git", rev = "54a267ac89c09b11c0c88934690530807185d3e7", features = [
     "visitor",
+    "serde",
 ] }
 strum = { version = "0.25", features = ["derive"] }
 tempfile = "3"
@@ -200,6 +204,7 @@ api = { path = "src/api" }
 auth = { path = "src/auth" }
 cache = { path = "src/cache" }
 catalog = { path = "src/catalog" }
+cli = { path = "src/cli" }
 client = { path = "src/client" }
 cmd = { path = "src/cmd", default-features = false }
 common-base = { path = "src/common/base" }
@@ -235,6 +240,7 @@ file-engine = { path = "src/file-engine" }
 flow = { path = "src/flow" }
 frontend = { path = "src/frontend", default-features = false }
 index = { path = "src/index" }
+log-query = { path = "src/log-query" }
 log-store = { path = "src/log-store" }
 meta-client = { path = "src/meta-client" }
 meta-srv = { path = "src/meta-srv" }

README.md

@@ -56,7 +56,7 @@
 - [Project Status](#project-status)
 - [Join the community](#community)
   - [Contributing](#contributing)
-- [Extension](#extension )
+- [Tools & Extensions](#tools--extensions)
 - [License](#license)
 - [Acknowledgement](#acknowledgement)
 
@@ -66,31 +66,33 @@
 
 ## Why GreptimeDB
 
-Our core developers have been building time-series data platforms for years. Based on our best-practices, GreptimeDB is born to give you:
+Our core developers have been building time-series data platforms for years. Based on our best practices, GreptimeDB was born to give you:
 
-* **Unified all kinds of time series**
+* **Unified Processing of Metrics, Logs, and Events**
 
-  GreptimeDB treats all time series as contextual events with timestamp, and thus unifies the processing of metrics, logs, and events. It supports analyzing metrics, logs, and events with SQL and PromQL, and doing streaming with continuous aggregation.
+  GreptimeDB unifies time series data processing by treating all data - whether metrics, logs, or events - as timestamped events with context. Users can analyze this data using either [SQL](https://docs.greptime.com/user-guide/query-data/sql) or [PromQL](https://docs.greptime.com/user-guide/query-data/promql) and leverage stream processing ([Flow](https://docs.greptime.com/user-guide/flow-computation/overview)) to enable continuous aggregation. [Read more](https://docs.greptime.com/user-guide/concepts/data-model).
 
-* **Cloud-Edge collaboration**
+* **Cloud-native Distributed Database**
 
-  GreptimeDB can be deployed on ARM architecture-compatible Android/Linux systems as well as cloud environments from various vendors. Both sides run the same software, providing identical APIs and control planes, so your application can run at the edge or on the cloud without modification, and data synchronization also becomes extremely easy and efficient.
-
-* **Cloud-native distributed database**
-
-  By leveraging object storage (S3 and others), separating compute and storage, scaling stateless compute nodes arbitrarily, GreptimeDB implements seamless scalability. It also supports cross-cloud deployment with a built-in unified data access layer over different object storages.
+  Built for [Kubernetes](https://docs.greptime.com/user-guide/deployments/deploy-on-kubernetes/greptimedb-operator-management). GreptimeDB achieves seamless scalability with its [cloud-native architecture](https://docs.greptime.com/user-guide/concepts/architecture) of separated compute and storage, built on object storage (AWS S3, Azure Blob Storage, etc.) while enabling cross-cloud deployment through a unified data access layer.
 
 * **Performance and Cost-effective**
 
-  Flexible indexing capabilities and distributed, parallel-processing query engine, tackling high cardinality issues down. Optimized columnar layout for handling time-series data; compacted, compressed, and stored on various storage backends, particularly cloud object storage with 50x cost efficiency.
+  Written in pure Rust for superior performance and reliability. GreptimeDB features a distributed query engine with intelligent indexing to handle high cardinality data efficiently. Its optimized columnar storage achieves 50x cost efficiency on cloud object storage through advanced compression. [Benchmark reports](https://www.greptime.com/blogs/2024-09-09-report-summary).
 
-* **Compatible with InfluxDB, Prometheus and more protocols**
+* **Cloud-Edge Collaboration**
 
-  Widely adopted database protocols and APIs, including MySQL, PostgreSQL, and Prometheus Remote Storage, etc. [Read more](https://docs.greptime.com/user-guide/protocols/overview).
+  GreptimeDB seamlessly operates across cloud and edge (ARM/Android/Linux), providing consistent APIs and control plane for unified data management and efficient synchronization. [Learn how to run on Android](https://docs.greptime.com/user-guide/deployments/run-on-android/).
+
+* **Multi-protocol Ingestion, SQL & PromQL Ready**
+
+  Widely adopted database protocols and APIs, including MySQL, PostgreSQL, InfluxDB, OpenTelemetry, Loki and Prometheus, etc.  Effortless Adoption & Seamless Migration. [Supported Protocols Overview](https://docs.greptime.com/user-guide/protocols/overview).
+
+For more detailed info please read  [Why GreptimeDB](https://docs.greptime.com/user-guide/concepts/why-greptimedb).
 
 ## Try GreptimeDB
 
-### 1. [GreptimePlay](https://greptime.com/playground)
+### 1. [Live Demo](https://greptime.com/playground)
 
 Try out the features of GreptimeDB right from your browser.
 
@@ -109,9 +111,18 @@ docker pull greptime/greptimedb
 Start a GreptimeDB container with:
 
 ```shell
-docker run --rm --name greptime --net=host greptime/greptimedb standalone start
+docker run -p 127.0.0.1:4000-4003:4000-4003 \
+  -v "$(pwd)/greptimedb:/tmp/greptimedb" \
+  --name greptime --rm \
+  greptime/greptimedb:latest standalone start \
+  --http-addr 0.0.0.0:4000 \
+  --rpc-addr 0.0.0.0:4001 \
+  --mysql-addr 0.0.0.0:4002 \
+  --postgres-addr 0.0.0.0:4003
 ```
 
+Access the dashboard via `http://localhost:4000/dashboard`.
+
 Read more about [Installation](https://docs.greptime.com/getting-started/installation/overview) on docs.
 
 ## Getting Started
@@ -127,7 +138,7 @@ Check the prerequisite:
 
 * [Rust toolchain](https://www.rust-lang.org/tools/install) (nightly)
 * [Protobuf compiler](https://grpc.io/docs/protoc-installation/) (>= 3.15)
-* Python toolchain (optional): Required only if built with PyO3 backend. More detail for compiling with PyO3 can be found in its [documentation](https://pyo3.rs/v0.18.1/building_and_distribution#configuring-the-python-version).
+* Python toolchain (optional): Required only if built with PyO3 backend. More details for compiling with PyO3 can be found in its [documentation](https://pyo3.rs/v0.18.1/building_and_distribution#configuring-the-python-version).
 
 Build GreptimeDB binary:
 
@@ -141,7 +152,11 @@ Run a standalone server:
 cargo run -- standalone start
 ```
 
-## Extension
+## Tools & Extensions
+
+### Kubernetes
+
+- [GreptimeDB Operator](https://github.com/GrepTimeTeam/greptimedb-operator)
 
 ### Dashboard
 
@@ -158,14 +173,19 @@ cargo run -- standalone start
 
 ### Grafana Dashboard
 
-Our official Grafana dashboard is available at [grafana](grafana/README.md) directory.
+Our official Grafana dashboard for monitoring GreptimeDB is available at [grafana](grafana/README.md) directory.
 
 ## Project Status
 
-The current version has not yet reached the standards for General Availability.
-According to our Greptime 2024 Roadmap, we aim to achieve a production-level version with the release of v1.0 by the end of 2024. [Join Us](https://github.com/GreptimeTeam/greptimedb/issues/3412)
+GreptimeDB is currently in Beta. We are targeting GA (General Availability) with v1.0 release by Early 2025.
 
-We welcome you to test and use GreptimeDB. Some users have already adopted it in their production environments. If you're interested in trying it out, please use the latest stable release available.
+While in Beta, GreptimeDB is already:
+
+* Being used in production by early adopters
+* Actively maintained with regular releases, [about version number](https://docs.greptime.com/nightly/reference/about-greptimedb-version)
+* Suitable for testing and evaluation
+
+For production use, we recommend using the latest stable release.
 
 ## Community
 
@@ -184,12 +204,12 @@ In addition, you may:
 - Connect us with [Linkedin](https://www.linkedin.com/company/greptime/)
 - Follow us on [Twitter](https://twitter.com/greptime)
 
-## Commerial Support
+## Commercial Support
 
 If you are running GreptimeDB OSS in your organization, we offer additional
-enterprise addons, installation service, training and consulting. [Contact
+enterprise add-ons, installation services, training, and consulting. [Contact
 us](https://greptime.com/contactus) and we will reach out to you with more
-detail of our commerial license.
+detail of our commercial license.
 
 ## License
 

config/config.md

@@ -13,11 +13,12 @@
 | Key | Type | Default | Descriptions |
 | --- | -----| ------- | ----------- |
 | `mode` | String | `standalone` | The running mode of the datanode. It can be `standalone` or `distributed`. |
-| `enable_telemetry` | Bool | `true` | Enable telemetry to collect anonymous usage data. |
 | `default_timezone` | String | Unset | The default timezone of the server. |
 | `init_regions_in_background` | Bool | `false` | Initialize all regions in the background during the startup.<br/>By default, it provides services after all regions have been initialized. |
 | `init_regions_parallelism` | Integer | `16` | Parallelism of initializing regions. |
 | `max_concurrent_queries` | Integer | `0` | The maximum current queries allowed to be executed. Zero means unlimited. |
+| `enable_telemetry` | Bool | `true` | Enable telemetry to collect anonymous usage data. Enabled by default. |
+| `max_in_flight_write_bytes` | String | Unset | The maximum in-flight write bytes. |
 | `runtime` | -- | -- | The runtime options. |
 | `runtime.global_rt_size` | Integer | `8` | The number of threads to execute the runtime for global read operations. |
 | `runtime.compact_rt_size` | Integer | `4` | The number of threads to execute the runtime for global write operations. |
@@ -61,9 +62,9 @@
 | `wal` | -- | -- | The WAL options. |
 | `wal.provider` | String | `raft_engine` | The provider of the WAL.<br/>- `raft_engine`: the wal is stored in the local file system by raft-engine.<br/>- `kafka`: it's remote wal that data is stored in Kafka. |
 | `wal.dir` | String | Unset | The directory to store the WAL files.<br/>**It's only used when the provider is `raft_engine`**. |
-| `wal.file_size` | String | `256MB` | The size of the WAL segment file.<br/>**It's only used when the provider is `raft_engine`**. |
-| `wal.purge_threshold` | String | `4GB` | The threshold of the WAL size to trigger a flush.<br/>**It's only used when the provider is `raft_engine`**. |
-| `wal.purge_interval` | String | `10m` | The interval to trigger a flush.<br/>**It's only used when the provider is `raft_engine`**. |
+| `wal.file_size` | String | `128MB` | The size of the WAL segment file.<br/>**It's only used when the provider is `raft_engine`**. |
+| `wal.purge_threshold` | String | `1GB` | The threshold of the WAL size to trigger a flush.<br/>**It's only used when the provider is `raft_engine`**. |
+| `wal.purge_interval` | String | `1m` | The interval to trigger a flush.<br/>**It's only used when the provider is `raft_engine`**. |
 | `wal.read_batch_size` | Integer | `128` | The read batch size.<br/>**It's only used when the provider is `raft_engine`**. |
 | `wal.sync_write` | Bool | `false` | Whether to use sync write.<br/>**It's only used when the provider is `raft_engine`**. |
 | `wal.enable_log_recycle` | Bool | `true` | Whether to reuse logically truncated log files.<br/>**It's only used when the provider is `raft_engine`**. |
@@ -93,7 +94,7 @@
 | `storage` | -- | -- | The data storage options. |
 | `storage.data_home` | String | `/tmp/greptimedb/` | The working home directory. |
 | `storage.type` | String | `File` | The storage type used to store the data.<br/>- `File`: the data is stored in the local file system.<br/>- `S3`: the data is stored in the S3 object storage.<br/>- `Gcs`: the data is stored in the Google Cloud Storage.<br/>- `Azblob`: the data is stored in the Azure Blob Storage.<br/>- `Oss`: the data is stored in the Aliyun OSS. |
-| `storage.cache_path` | String | Unset | Cache configuration for object storage such as 'S3' etc. It is recommended to configure it when using object storage for better performance.<br/>The local file cache directory. |
+| `storage.cache_path` | String | Unset | Read cache configuration for object storage such as 'S3' etc, it's configured by default when using object storage. It is recommended to configure it when using object storage for better performance.<br/>A local file directory, defaults to `{data_home}/object_cache/read`. An empty string means disabling. |
 | `storage.cache_capacity` | String | Unset | The local file cache capacity in bytes. If your disk space is sufficient, it is recommended to set it larger. |
 | `storage.bucket` | String | Unset | The S3 bucket name.<br/>**It's only used when the storage type is `S3`, `Oss` and `Gcs`**. |
 | `storage.root` | String | Unset | The S3 data will be stored in the specified prefix, for example, `s3://${bucket}/${root}`.<br/>**It's only used when the storage type is `S3`, `Oss` and `Azblob`**. |
@@ -109,6 +110,11 @@
 | `storage.sas_token` | String | Unset | The sas token of the azure account.<br/>**It's only used when the storage type is `Azblob`**. |
 | `storage.endpoint` | String | Unset | The endpoint of the S3 service.<br/>**It's only used when the storage type is `S3`, `Oss`, `Gcs` and `Azblob`**. |
 | `storage.region` | String | Unset | The region of the S3 service.<br/>**It's only used when the storage type is `S3`, `Oss`, `Gcs` and `Azblob`**. |
+| `storage.http_client` | -- | -- | The http client options to the storage.<br/>**It's only used when the storage type is `S3`, `Oss`, `Gcs` and `Azblob`**. |
+| `storage.http_client.pool_max_idle_per_host` | Integer | `1024` | The maximum idle connection per host allowed in the pool. |
+| `storage.http_client.connect_timeout` | String | `30s` | The timeout for only the connect phase of a http client. |
+| `storage.http_client.timeout` | String | `30s` | The total request timeout, applied from when the request starts connecting until the response body has finished.<br/>Also considered a total deadline. |
+| `storage.http_client.pool_idle_timeout` | String | `90s` | The timeout for idle sockets being kept-alive. |
 | `[[region_engine]]` | -- | -- | The region engine options. You can configure multiple region engines. |
 | `region_engine.mito` | -- | -- | The Mito engine options. |
 | `region_engine.mito.num_workers` | Integer | `8` | Number of region workers. |
@@ -126,12 +132,11 @@
 | `region_engine.mito.vector_cache_size` | String | Auto | Cache size for vectors and arrow arrays. Setting it to 0 to disable the cache.<br/>If not set, it's default to 1/16 of OS memory with a max limitation of 512MB. |
 | `region_engine.mito.page_cache_size` | String | Auto | Cache size for pages of SST row groups. Setting it to 0 to disable the cache.<br/>If not set, it's default to 1/8 of OS memory. |
 | `region_engine.mito.selector_result_cache_size` | String | Auto | Cache size for time series selector (e.g. `last_value()`). Setting it to 0 to disable the cache.<br/>If not set, it's default to 1/16 of OS memory with a max limitation of 512MB. |
-| `region_engine.mito.enable_experimental_write_cache` | Bool | `false` | Whether to enable the experimental write cache. It is recommended to enable it when using object storage for better performance. |
-| `region_engine.mito.experimental_write_cache_path` | String | `""` | File system path for write cache, defaults to `{data_home}/write_cache`. |
-| `region_engine.mito.experimental_write_cache_size` | String | `1GiB` | Capacity for write cache. If your disk space is sufficient, it is recommended to set it larger. |
+| `region_engine.mito.enable_experimental_write_cache` | Bool | `false` | Whether to enable the experimental write cache, it's enabled by default when using object storage. It is recommended to enable it when using object storage for better performance. |
+| `region_engine.mito.experimental_write_cache_path` | String | `""` | File system path for write cache, defaults to `{data_home}/object_cache/write`. |
+| `region_engine.mito.experimental_write_cache_size` | String | `5GiB` | Capacity for write cache. If your disk space is sufficient, it is recommended to set it larger. |
 | `region_engine.mito.experimental_write_cache_ttl` | String | Unset | TTL for write cache. |
 | `region_engine.mito.sst_write_buffer_size` | String | `8MB` | Buffer size for SST writing. |
-| `region_engine.mito.scan_parallelism` | Integer | `0` | Parallelism to scan a region (default: 1/4 of cpu cores).<br/>- `0`: using the default value (1/4 of cpu cores).<br/>- `1`: scan in current thread.<br/>- `n`: scan in parallelism n. |
 | `region_engine.mito.parallel_scan_channel_size` | Integer | `32` | Capacity of the channel to send data from parallel scan tasks to the main task. |
 | `region_engine.mito.allow_stale_entries` | Bool | `false` | Whether to allow stale WAL entries read during replay. |
 | `region_engine.mito.min_compaction_interval` | String | `0m` | Minimum time interval between two compactions.<br/>To align with the old behavior, the default value is 0 (no restrictions). |
@@ -146,11 +151,17 @@
 | `region_engine.mito.inverted_index.intermediate_path` | String | `""` | Deprecated, use `region_engine.mito.index.aux_path` instead. |
 | `region_engine.mito.inverted_index.metadata_cache_size` | String | `64MiB` | Cache size for inverted index metadata. |
 | `region_engine.mito.inverted_index.content_cache_size` | String | `128MiB` | Cache size for inverted index content. |
+| `region_engine.mito.inverted_index.content_cache_page_size` | String | `8MiB` | Page size for inverted index content cache. |
 | `region_engine.mito.fulltext_index` | -- | -- | The options for full-text index in Mito engine. |
 | `region_engine.mito.fulltext_index.create_on_flush` | String | `auto` | Whether to create the index on flush.<br/>- `auto`: automatically (default)<br/>- `disable`: never |
 | `region_engine.mito.fulltext_index.create_on_compaction` | String | `auto` | Whether to create the index on compaction.<br/>- `auto`: automatically (default)<br/>- `disable`: never |
 | `region_engine.mito.fulltext_index.apply_on_query` | String | `auto` | Whether to apply the index on query<br/>- `auto`: automatically (default)<br/>- `disable`: never |
 | `region_engine.mito.fulltext_index.mem_threshold_on_create` | String | `auto` | Memory threshold for index creation.<br/>- `auto`: automatically determine the threshold based on the system memory size (default)<br/>- `unlimited`: no memory limit<br/>- `[size]` e.g. `64MB`: fixed memory threshold |
+| `region_engine.mito.bloom_filter_index` | -- | -- | The options for bloom filter in Mito engine. |
+| `region_engine.mito.bloom_filter_index.create_on_flush` | String | `auto` | Whether to create the bloom filter on flush.<br/>- `auto`: automatically (default)<br/>- `disable`: never |
+| `region_engine.mito.bloom_filter_index.create_on_compaction` | String | `auto` | Whether to create the bloom filter on compaction.<br/>- `auto`: automatically (default)<br/>- `disable`: never |
+| `region_engine.mito.bloom_filter_index.apply_on_query` | String | `auto` | Whether to apply the bloom filter on query<br/>- `auto`: automatically (default)<br/>- `disable`: never |
+| `region_engine.mito.bloom_filter_index.mem_threshold_on_create` | String | `auto` | Memory threshold for bloom filter creation.<br/>- `auto`: automatically determine the threshold based on the system memory size (default)<br/>- `unlimited`: no memory limit<br/>- `[size]` e.g. `64MB`: fixed memory threshold |
 | `region_engine.mito.memtable` | -- | -- | -- |
 | `region_engine.mito.memtable.type` | String | `time_series` | Memtable type.<br/>- `time_series`: time-series memtable<br/>- `partition_tree`: partition tree memtable (experimental) |
 | `region_engine.mito.memtable.index_max_keys_per_shard` | Integer | `8192` | The max number of keys in one shard.<br/>Only available for `partition_tree` memtable. |
@@ -190,6 +201,7 @@
 | Key | Type | Default | Descriptions |
 | --- | -----| ------- | ----------- |
 | `default_timezone` | String | Unset | The default timezone of the server. |
+| `max_in_flight_write_bytes` | String | Unset | The maximum in-flight write bytes. |
 | `runtime` | -- | -- | The runtime options. |
 | `runtime.global_rt_size` | Integer | `8` | The number of threads to execute the runtime for global read operations. |
 | `runtime.compact_rt_size` | Integer | `4` | The number of threads to execute the runtime for global write operations. |
@@ -281,13 +293,13 @@
 | `data_home` | String | `/tmp/metasrv/` | The working home directory. |
 | `bind_addr` | String | `127.0.0.1:3002` | The bind address of metasrv. |
 | `server_addr` | String | `127.0.0.1:3002` | The communication server address for frontend and datanode to connect to metasrv,  "127.0.0.1:3002" by default for localhost. |
-| `store_addr` | String | `127.0.0.1:2379` | Store server address default to etcd store. |
+| `store_addrs` | Array | -- | Store server address default to etcd store. |
+| `store_key_prefix` | String | `""` | If it's not empty, the metasrv will store all data with this key prefix. |
+| `backend` | String | `EtcdStore` | The datastore for meta server. |
 | `selector` | String | `round_robin` | Datanode selector type.<br/>- `round_robin` (default value)<br/>- `lease_based`<br/>- `load_based`<br/>For details, please see "https://docs.greptime.com/developer-guide/metasrv/selector". |
 | `use_memory_store` | Bool | `false` | Store data in memory. |
-| `enable_telemetry` | Bool | `true` | Whether to enable greptimedb telemetry. |
-| `store_key_prefix` | String | `""` | If it's not empty, the metasrv will store all data with this key prefix. |
 | `enable_region_failover` | Bool | `false` | Whether to enable region failover.<br/>This feature is only available on GreptimeDB running on cluster mode and<br/>- Using Remote WAL<br/>- Using shared storage (e.g., s3). |
-| `backend` | String | `EtcdStore` | The datastore for meta server. |
+| `enable_telemetry` | Bool | `true` | Whether to enable greptimedb telemetry. Enabled by default. |
 | `runtime` | -- | -- | The runtime options. |
 | `runtime.global_rt_size` | Integer | `8` | The number of threads to execute the runtime for global read operations. |
 | `runtime.compact_rt_size` | Integer | `4` | The number of threads to execute the runtime for global write operations. |
@@ -352,7 +364,6 @@
 | `node_id` | Integer | Unset | The datanode identifier and should be unique in the cluster. |
 | `require_lease_before_startup` | Bool | `false` | Start services after regions have obtained leases.<br/>It will block the datanode start if it can't receive leases in the heartbeat from metasrv. |
 | `init_regions_in_background` | Bool | `false` | Initialize all regions in the background during the startup.<br/>By default, it provides services after all regions have been initialized. |
-| `enable_telemetry` | Bool | `true` | Enable telemetry to collect anonymous usage data. |
 | `init_regions_parallelism` | Integer | `16` | Parallelism of initializing regions. |
 | `max_concurrent_queries` | Integer | `0` | The maximum current queries allowed to be executed. Zero means unlimited. |
 | `rpc_addr` | String | Unset | Deprecated, use `grpc.addr` instead. |
@@ -360,6 +371,7 @@
 | `rpc_runtime_size` | Integer | Unset | Deprecated, use `grpc.runtime_size` instead. |
 | `rpc_max_recv_message_size` | String | Unset | Deprecated, use `grpc.rpc_max_recv_message_size` instead. |
 | `rpc_max_send_message_size` | String | Unset | Deprecated, use `grpc.rpc_max_send_message_size` instead. |
+| `enable_telemetry` | Bool | `true` | Enable telemetry to collect anonymous usage data. Enabled by default. |
 | `http` | -- | -- | The HTTP server options. |
 | `http.addr` | String | `127.0.0.1:4000` | The address to bind the HTTP server. |
 | `http.timeout` | String | `30s` | HTTP request timeout. Set to 0 to disable timeout. |
@@ -394,9 +406,9 @@
 | `wal` | -- | -- | The WAL options. |
 | `wal.provider` | String | `raft_engine` | The provider of the WAL.<br/>- `raft_engine`: the wal is stored in the local file system by raft-engine.<br/>- `kafka`: it's remote wal that data is stored in Kafka. |
 | `wal.dir` | String | Unset | The directory to store the WAL files.<br/>**It's only used when the provider is `raft_engine`**. |
-| `wal.file_size` | String | `256MB` | The size of the WAL segment file.<br/>**It's only used when the provider is `raft_engine`**. |
-| `wal.purge_threshold` | String | `4GB` | The threshold of the WAL size to trigger a flush.<br/>**It's only used when the provider is `raft_engine`**. |
-| `wal.purge_interval` | String | `10m` | The interval to trigger a flush.<br/>**It's only used when the provider is `raft_engine`**. |
+| `wal.file_size` | String | `128MB` | The size of the WAL segment file.<br/>**It's only used when the provider is `raft_engine`**. |
+| `wal.purge_threshold` | String | `1GB` | The threshold of the WAL size to trigger a flush.<br/>**It's only used when the provider is `raft_engine`**. |
+| `wal.purge_interval` | String | `1m` | The interval to trigger a flush.<br/>**It's only used when the provider is `raft_engine`**. |
 | `wal.read_batch_size` | Integer | `128` | The read batch size.<br/>**It's only used when the provider is `raft_engine`**. |
 | `wal.sync_write` | Bool | `false` | Whether to use sync write.<br/>**It's only used when the provider is `raft_engine`**. |
 | `wal.enable_log_recycle` | Bool | `true` | Whether to reuse logically truncated log files.<br/>**It's only used when the provider is `raft_engine`**. |
@@ -416,7 +428,7 @@
 | `storage` | -- | -- | The data storage options. |
 | `storage.data_home` | String | `/tmp/greptimedb/` | The working home directory. |
 | `storage.type` | String | `File` | The storage type used to store the data.<br/>- `File`: the data is stored in the local file system.<br/>- `S3`: the data is stored in the S3 object storage.<br/>- `Gcs`: the data is stored in the Google Cloud Storage.<br/>- `Azblob`: the data is stored in the Azure Blob Storage.<br/>- `Oss`: the data is stored in the Aliyun OSS. |
-| `storage.cache_path` | String | Unset | Cache configuration for object storage such as 'S3' etc. It is recommended to configure it when using object storage for better performance.<br/>The local file cache directory. |
+| `storage.cache_path` | String | Unset | Read cache configuration for object storage such as 'S3' etc, it's configured by default when using object storage. It is recommended to configure it when using object storage for better performance.<br/>A local file directory, defaults to `{data_home}`. An empty string means disabling. |
 | `storage.cache_capacity` | String | Unset | The local file cache capacity in bytes. If your disk space is sufficient, it is recommended to set it larger. |
 | `storage.bucket` | String | Unset | The S3 bucket name.<br/>**It's only used when the storage type is `S3`, `Oss` and `Gcs`**. |
 | `storage.root` | String | Unset | The S3 data will be stored in the specified prefix, for example, `s3://${bucket}/${root}`.<br/>**It's only used when the storage type is `S3`, `Oss` and `Azblob`**. |
@@ -432,6 +444,11 @@
 | `storage.sas_token` | String | Unset | The sas token of the azure account.<br/>**It's only used when the storage type is `Azblob`**. |
 | `storage.endpoint` | String | Unset | The endpoint of the S3 service.<br/>**It's only used when the storage type is `S3`, `Oss`, `Gcs` and `Azblob`**. |
 | `storage.region` | String | Unset | The region of the S3 service.<br/>**It's only used when the storage type is `S3`, `Oss`, `Gcs` and `Azblob`**. |
+| `storage.http_client` | -- | -- | The http client options to the storage.<br/>**It's only used when the storage type is `S3`, `Oss`, `Gcs` and `Azblob`**. |
+| `storage.http_client.pool_max_idle_per_host` | Integer | `1024` | The maximum idle connection per host allowed in the pool. |
+| `storage.http_client.connect_timeout` | String | `30s` | The timeout for only the connect phase of a http client. |
+| `storage.http_client.timeout` | String | `30s` | The total request timeout, applied from when the request starts connecting until the response body has finished.<br/>Also considered a total deadline. |
+| `storage.http_client.pool_idle_timeout` | String | `90s` | The timeout for idle sockets being kept-alive. |
 | `[[region_engine]]` | -- | -- | The region engine options. You can configure multiple region engines. |
 | `region_engine.mito` | -- | -- | The Mito engine options. |
 | `region_engine.mito.num_workers` | Integer | `8` | Number of region workers. |
@@ -449,12 +466,11 @@
 | `region_engine.mito.vector_cache_size` | String | Auto | Cache size for vectors and arrow arrays. Setting it to 0 to disable the cache.<br/>If not set, it's default to 1/16 of OS memory with a max limitation of 512MB. |
 | `region_engine.mito.page_cache_size` | String | Auto | Cache size for pages of SST row groups. Setting it to 0 to disable the cache.<br/>If not set, it's default to 1/8 of OS memory. |
 | `region_engine.mito.selector_result_cache_size` | String | Auto | Cache size for time series selector (e.g. `last_value()`). Setting it to 0 to disable the cache.<br/>If not set, it's default to 1/16 of OS memory with a max limitation of 512MB. |
-| `region_engine.mito.enable_experimental_write_cache` | Bool | `false` | Whether to enable the experimental write cache. It is recommended to enable it when using object storage for better performance. |
-| `region_engine.mito.experimental_write_cache_path` | String | `""` | File system path for write cache, defaults to `{data_home}/write_cache`. |
-| `region_engine.mito.experimental_write_cache_size` | String | `1GiB` | Capacity for write cache. If your disk space is sufficient, it is recommended to set it larger. |
+| `region_engine.mito.enable_experimental_write_cache` | Bool | `false` | Whether to enable the experimental write cache, it's enabled by default when using object storage. It is recommended to enable it when using object storage for better performance. |
+| `region_engine.mito.experimental_write_cache_path` | String | `""` | File system path for write cache, defaults to `{data_home}`. |
+| `region_engine.mito.experimental_write_cache_size` | String | `5GiB` | Capacity for write cache. If your disk space is sufficient, it is recommended to set it larger. |
 | `region_engine.mito.experimental_write_cache_ttl` | String | Unset | TTL for write cache. |
 | `region_engine.mito.sst_write_buffer_size` | String | `8MB` | Buffer size for SST writing. |
-| `region_engine.mito.scan_parallelism` | Integer | `0` | Parallelism to scan a region (default: 1/4 of cpu cores).<br/>- `0`: using the default value (1/4 of cpu cores).<br/>- `1`: scan in current thread.<br/>- `n`: scan in parallelism n. |
 | `region_engine.mito.parallel_scan_channel_size` | Integer | `32` | Capacity of the channel to send data from parallel scan tasks to the main task. |
 | `region_engine.mito.allow_stale_entries` | Bool | `false` | Whether to allow stale WAL entries read during replay. |
 | `region_engine.mito.min_compaction_interval` | String | `0m` | Minimum time interval between two compactions.<br/>To align with the old behavior, the default value is 0 (no restrictions). |
@@ -467,11 +483,19 @@
 | `region_engine.mito.inverted_index.apply_on_query` | String | `auto` | Whether to apply the index on query<br/>- `auto`: automatically (default)<br/>- `disable`: never |
 | `region_engine.mito.inverted_index.mem_threshold_on_create` | String | `auto` | Memory threshold for performing an external sort during index creation.<br/>- `auto`: automatically determine the threshold based on the system memory size (default)<br/>- `unlimited`: no memory limit<br/>- `[size]` e.g. `64MB`: fixed memory threshold |
 | `region_engine.mito.inverted_index.intermediate_path` | String | `""` | Deprecated, use `region_engine.mito.index.aux_path` instead. |
+| `region_engine.mito.inverted_index.metadata_cache_size` | String | `64MiB` | Cache size for inverted index metadata. |
+| `region_engine.mito.inverted_index.content_cache_size` | String | `128MiB` | Cache size for inverted index content. |
+| `region_engine.mito.inverted_index.content_cache_page_size` | String | `8MiB` | Page size for inverted index content cache. |
 | `region_engine.mito.fulltext_index` | -- | -- | The options for full-text index in Mito engine. |
 | `region_engine.mito.fulltext_index.create_on_flush` | String | `auto` | Whether to create the index on flush.<br/>- `auto`: automatically (default)<br/>- `disable`: never |
 | `region_engine.mito.fulltext_index.create_on_compaction` | String | `auto` | Whether to create the index on compaction.<br/>- `auto`: automatically (default)<br/>- `disable`: never |
 | `region_engine.mito.fulltext_index.apply_on_query` | String | `auto` | Whether to apply the index on query<br/>- `auto`: automatically (default)<br/>- `disable`: never |
 | `region_engine.mito.fulltext_index.mem_threshold_on_create` | String | `auto` | Memory threshold for index creation.<br/>- `auto`: automatically determine the threshold based on the system memory size (default)<br/>- `unlimited`: no memory limit<br/>- `[size]` e.g. `64MB`: fixed memory threshold |
+| `region_engine.mito.bloom_filter_index` | -- | -- | The options for bloom filter index in Mito engine. |
+| `region_engine.mito.bloom_filter_index.create_on_flush` | String | `auto` | Whether to create the index on flush.<br/>- `auto`: automatically (default)<br/>- `disable`: never |
+| `region_engine.mito.bloom_filter_index.create_on_compaction` | String | `auto` | Whether to create the index on compaction.<br/>- `auto`: automatically (default)<br/>- `disable`: never |
+| `region_engine.mito.bloom_filter_index.apply_on_query` | String | `auto` | Whether to apply the index on query<br/>- `auto`: automatically (default)<br/>- `disable`: never |
+| `region_engine.mito.bloom_filter_index.mem_threshold_on_create` | String | `auto` | Memory threshold for the index creation.<br/>- `auto`: automatically determine the threshold based on the system memory size (default)<br/>- `unlimited`: no memory limit<br/>- `[size]` e.g. `64MB`: fixed memory threshold |
 | `region_engine.mito.memtable` | -- | -- | -- |
 | `region_engine.mito.memtable.type` | String | `time_series` | Memtable type.<br/>- `time_series`: time-series memtable<br/>- `partition_tree`: partition tree memtable (experimental) |
 | `region_engine.mito.memtable.index_max_keys_per_shard` | Integer | `8192` | The max number of keys in one shard.<br/>Only available for `partition_tree` memtable. |

config/datanode.example.toml

@@ -13,9 +13,6 @@ require_lease_before_startup = false
 ## By default, it provides services after all regions have been initialized.
 init_regions_in_background = false
 
-## Enable telemetry to collect anonymous usage data.
-enable_telemetry = true
-
 ## Parallelism of initializing regions.
 init_regions_parallelism = 16
 
@@ -42,6 +39,8 @@ rpc_max_recv_message_size = "512MB"
 ## @toml2docs:none-default
 rpc_max_send_message_size = "512MB"
 
+## Enable telemetry to collect anonymous usage data. Enabled by default.
+#+ enable_telemetry = true
 
 ## The HTTP server options.
 [http]
@@ -143,15 +142,15 @@ dir = "/tmp/greptimedb/wal"
 
 ## The size of the WAL segment file.
 ## **It's only used when the provider is `raft_engine`**.
-file_size = "256MB"
+file_size = "128MB"
 
 ## The threshold of the WAL size to trigger a flush.
 ## **It's only used when the provider is `raft_engine`**.
-purge_threshold = "4GB"
+purge_threshold = "1GB"
 
 ## The interval to trigger a flush.
 ## **It's only used when the provider is `raft_engine`**.
-purge_interval = "10m"
+purge_interval = "1m"
 
 ## The read batch size.
 ## **It's only used when the provider is `raft_engine`**.
@@ -294,14 +293,14 @@ data_home = "/tmp/greptimedb/"
 ## - `Oss`: the data is stored in the Aliyun OSS.
 type = "File"
 
-## Cache configuration for object storage such as 'S3' etc. It is recommended to configure it when using object storage for better performance.
-## The local file cache directory.
+## Read cache configuration for object storage such as 'S3' etc, it's configured by default when using object storage. It is recommended to configure it when using object storage for better performance.
+## A local file directory, defaults to `{data_home}`. An empty string means disabling.
 ## @toml2docs:none-default
-cache_path = "/path/local_cache"
+#+ cache_path = ""
 
 ## The local file cache capacity in bytes. If your disk space is sufficient, it is recommended to set it larger.
 ## @toml2docs:none-default
-cache_capacity = "1GiB"
+cache_capacity = "5GiB"
 
 ## The S3 bucket name.
 ## **It's only used when the storage type is `S3`, `Oss` and `Gcs`**.
@@ -375,6 +374,23 @@ endpoint = "https://s3.amazonaws.com"
 ## @toml2docs:none-default
 region = "us-west-2"
 
+## The http client options to the storage.
+## **It's only used when the storage type is `S3`, `Oss`, `Gcs` and `Azblob`**.
+[storage.http_client]
+
+## The maximum idle connection per host allowed in the pool.
+pool_max_idle_per_host = 1024
+
+## The timeout for only the connect phase of a http client.
+connect_timeout = "30s"
+
+## The total request timeout, applied from when the request starts connecting until the response body has finished.
+## Also considered a total deadline.
+timeout = "30s"
+
+## The timeout for idle sockets being kept-alive.
+pool_idle_timeout = "90s"
+
 # Custom storage options
 # [[storage.providers]]
 # name = "S3"
@@ -459,14 +475,14 @@ auto_flush_interval = "1h"
 ## @toml2docs:none-default="Auto"
 #+ selector_result_cache_size = "512MB"
 
-## Whether to enable the experimental write cache. It is recommended to enable it when using object storage for better performance.
+## Whether to enable the experimental write cache, it's enabled by default when using object storage. It is recommended to enable it when using object storage for better performance.
 enable_experimental_write_cache = false
 
-## File system path for write cache, defaults to `{data_home}/write_cache`.
+## File system path for write cache, defaults to `{data_home}`.
 experimental_write_cache_path = ""
 
 ## Capacity for write cache. If your disk space is sufficient, it is recommended to set it larger.
-experimental_write_cache_size = "1GiB"
+experimental_write_cache_size = "5GiB"
 
 ## TTL for write cache.
 ## @toml2docs:none-default
@@ -475,12 +491,6 @@ experimental_write_cache_ttl = "8h"
 ## Buffer size for SST writing.
 sst_write_buffer_size = "8MB"
 
-## Parallelism to scan a region (default: 1/4 of cpu cores).
-## - `0`: using the default value (1/4 of cpu cores).
-## - `1`: scan in current thread.
-## - `n`: scan in parallelism n.
-scan_parallelism = 0
-
 ## Capacity of the channel to send data from parallel scan tasks to the main task.
 parallel_scan_channel_size = 32
 
@@ -533,6 +543,15 @@ mem_threshold_on_create = "auto"
 ## Deprecated, use `region_engine.mito.index.aux_path` instead.
 intermediate_path = ""
 
+## Cache size for inverted index metadata.
+metadata_cache_size = "64MiB"
+
+## Cache size for inverted index content.
+content_cache_size = "128MiB"
+
+## Page size for inverted index content cache.
+content_cache_page_size = "8MiB"
+
 ## The options for full-text index in Mito engine.
 [region_engine.mito.fulltext_index]
 
@@ -557,6 +576,30 @@ apply_on_query = "auto"
 ## - `[size]` e.g. `64MB`: fixed memory threshold
 mem_threshold_on_create = "auto"
 
+## The options for bloom filter index in Mito engine.
+[region_engine.mito.bloom_filter_index]
+
+## Whether to create the index on flush.
+## - `auto`: automatically (default)
+## - `disable`: never
+create_on_flush = "auto"
+
+## Whether to create the index on compaction.
+## - `auto`: automatically (default)
+## - `disable`: never
+create_on_compaction = "auto"
+
+## Whether to apply the index on query
+## - `auto`: automatically (default)
+## - `disable`: never
+apply_on_query = "auto"
+
+## Memory threshold for the index creation.
+## - `auto`: automatically determine the threshold based on the system memory size (default)
+## - `unlimited`: no memory limit
+## - `[size]` e.g. `64MB`: fixed memory threshold
+mem_threshold_on_create = "auto"
+
 [region_engine.mito.memtable]
 ## Memtable type.
 ## - `time_series`: time-series memtable

config/frontend.example.toml

@@ -2,6 +2,10 @@
 ## @toml2docs:none-default
 default_timezone = "UTC"
 
+## The maximum in-flight write bytes.
+## @toml2docs:none-default
+#+ max_in_flight_write_bytes = "500MB"
+
 ## The runtime options.
 #+ [runtime]
 ## The number of threads to execute the runtime for global read operations.

config/metasrv.example.toml

@@ -8,7 +8,13 @@ bind_addr = "127.0.0.1:3002"
 server_addr = "127.0.0.1:3002"
 
 ## Store server address default to etcd store.
-store_addr = "127.0.0.1:2379"
+store_addrs = ["127.0.0.1:2379"]
+
+## If it's not empty, the metasrv will store all data with this key prefix.
+store_key_prefix = ""
+
+## The datastore for meta server.
+backend = "EtcdStore"
 
 ## Datanode selector type.
 ## - `round_robin` (default value)
@@ -20,20 +26,14 @@ selector = "round_robin"
 ## Store data in memory.
 use_memory_store = false
 
-## Whether to enable greptimedb telemetry.
-enable_telemetry = true
-
-## If it's not empty, the metasrv will store all data with this key prefix.
-store_key_prefix = ""
-
 ## Whether to enable region failover.
 ## This feature is only available on GreptimeDB running on cluster mode and
 ## - Using Remote WAL
 ## - Using shared storage (e.g., s3).
 enable_region_failover = false
 
-## The datastore for meta server.
-backend = "EtcdStore"
+## Whether to enable greptimedb telemetry. Enabled by default.
+#+ enable_telemetry = true
 
 ## The runtime options.
 #+ [runtime]

config/standalone.example.toml

@@ -1,9 +1,6 @@
 ## The running mode of the datanode. It can be `standalone` or `distributed`.
 mode = "standalone"
 
-## Enable telemetry to collect anonymous usage data.
-enable_telemetry = true
-
 ## The default timezone of the server.
 ## @toml2docs:none-default
 default_timezone = "UTC"
@@ -18,6 +15,13 @@ init_regions_parallelism = 16
 ## The maximum current queries allowed to be executed. Zero means unlimited.
 max_concurrent_queries = 0
 
+## Enable telemetry to collect anonymous usage data. Enabled by default.
+#+ enable_telemetry = true
+
+## The maximum in-flight write bytes.
+## @toml2docs:none-default
+#+ max_in_flight_write_bytes = "500MB"
+
 ## The runtime options.
 #+ [runtime]
 ## The number of threads to execute the runtime for global read operations.
@@ -147,15 +151,15 @@ dir = "/tmp/greptimedb/wal"
 
 ## The size of the WAL segment file.
 ## **It's only used when the provider is `raft_engine`**.
-file_size = "256MB"
+file_size = "128MB"
 
 ## The threshold of the WAL size to trigger a flush.
 ## **It's only used when the provider is `raft_engine`**.
-purge_threshold = "4GB"
+purge_threshold = "1GB"
 
 ## The interval to trigger a flush.
 ## **It's only used when the provider is `raft_engine`**.
-purge_interval = "10m"
+purge_interval = "1m"
 
 ## The read batch size.
 ## **It's only used when the provider is `raft_engine`**.
@@ -332,14 +336,14 @@ data_home = "/tmp/greptimedb/"
 ## - `Oss`: the data is stored in the Aliyun OSS.
 type = "File"
 
-## Cache configuration for object storage such as 'S3' etc. It is recommended to configure it when using object storage for better performance.
-## The local file cache directory.
+## Read cache configuration for object storage such as 'S3' etc, it's configured by default when using object storage. It is recommended to configure it when using object storage for better performance.
+## A local file directory, defaults to `{data_home}/object_cache/read`. An empty string means disabling.
 ## @toml2docs:none-default
-cache_path = "/path/local_cache"
+#+ cache_path = ""
 
 ## The local file cache capacity in bytes. If your disk space is sufficient, it is recommended to set it larger.
 ## @toml2docs:none-default
-cache_capacity = "1GiB"
+cache_capacity = "5GiB"
 
 ## The S3 bucket name.
 ## **It's only used when the storage type is `S3`, `Oss` and `Gcs`**.
@@ -413,6 +417,23 @@ endpoint = "https://s3.amazonaws.com"
 ## @toml2docs:none-default
 region = "us-west-2"
 
+## The http client options to the storage.
+## **It's only used when the storage type is `S3`, `Oss`, `Gcs` and `Azblob`**.
+[storage.http_client]
+
+## The maximum idle connection per host allowed in the pool.
+pool_max_idle_per_host = 1024
+
+## The timeout for only the connect phase of a http client.
+connect_timeout = "30s"
+
+## The total request timeout, applied from when the request starts connecting until the response body has finished.
+## Also considered a total deadline.
+timeout = "30s"
+
+## The timeout for idle sockets being kept-alive.
+pool_idle_timeout = "90s"
+
 # Custom storage options
 # [[storage.providers]]
 # name = "S3"
@@ -497,14 +518,14 @@ auto_flush_interval = "1h"
 ## @toml2docs:none-default="Auto"
 #+ selector_result_cache_size = "512MB"
 
-## Whether to enable the experimental write cache. It is recommended to enable it when using object storage for better performance.
+## Whether to enable the experimental write cache, it's enabled by default when using object storage. It is recommended to enable it when using object storage for better performance.
 enable_experimental_write_cache = false
 
-## File system path for write cache, defaults to `{data_home}/write_cache`.
+## File system path for write cache, defaults to `{data_home}/object_cache/write`.
 experimental_write_cache_path = ""
 
 ## Capacity for write cache. If your disk space is sufficient, it is recommended to set it larger.
-experimental_write_cache_size = "1GiB"
+experimental_write_cache_size = "5GiB"
 
 ## TTL for write cache.
 ## @toml2docs:none-default
@@ -513,12 +534,6 @@ experimental_write_cache_ttl = "8h"
 ## Buffer size for SST writing.
 sst_write_buffer_size = "8MB"
 
-## Parallelism to scan a region (default: 1/4 of cpu cores).
-## - `0`: using the default value (1/4 of cpu cores).
-## - `1`: scan in current thread.
-## - `n`: scan in parallelism n.
-scan_parallelism = 0
-
 ## Capacity of the channel to send data from parallel scan tasks to the main task.
 parallel_scan_channel_size = 32
 
@@ -577,6 +592,9 @@ metadata_cache_size = "64MiB"
 ## Cache size for inverted index content.
 content_cache_size = "128MiB"
 
+## Page size for inverted index content cache.
+content_cache_page_size = "8MiB"
+
 ## The options for full-text index in Mito engine.
 [region_engine.mito.fulltext_index]
 
@@ -601,6 +619,30 @@ apply_on_query = "auto"
 ## - `[size]` e.g. `64MB`: fixed memory threshold
 mem_threshold_on_create = "auto"
 
+## The options for bloom filter in Mito engine.
+[region_engine.mito.bloom_filter_index]
+
+## Whether to create the bloom filter on flush.
+## - `auto`: automatically (default)
+## - `disable`: never
+create_on_flush = "auto"
+
+## Whether to create the bloom filter on compaction.
+## - `auto`: automatically (default)
+## - `disable`: never
+create_on_compaction = "auto"
+
+## Whether to apply the bloom filter on query
+## - `auto`: automatically (default)
+## - `disable`: never
+apply_on_query = "auto"
+
+## Memory threshold for bloom filter creation.
+## - `auto`: automatically determine the threshold based on the system memory size (default)
+## - `unlimited`: no memory limit
+## - `[size]` e.g. `64MB`: fixed memory threshold
+mem_threshold_on_create = "auto"
+
 [region_engine.mito.memtable]
 ## Memtable type.
 ## - `time_series`: time-series memtable

docker/buildx/centos/Dockerfile

@@ -13,8 +13,6 @@ RUN yum install -y epel-release  \
     openssl \
     openssl-devel  \
     centos-release-scl  \
-    rh-python38  \
-    rh-python38-python-devel \
     which
 
 # Install protoc
@@ -43,8 +41,6 @@ RUN yum install -y epel-release \
     openssl \
     openssl-devel  \
     centos-release-scl  \
-    rh-python38  \
-    rh-python38-python-devel \
     which
 
 WORKDIR /greptime

docker/buildx/ubuntu/Dockerfile

@@ -20,10 +20,7 @@ RUN --mount=type=cache,target=/var/cache/apt \
     curl \
     git \
     build-essential \
-    pkg-config \
-    python3.10 \
-    python3.10-dev \
-    python3-pip
+    pkg-config
 
 # Install Rust.
 SHELL ["/bin/bash", "-c"]
@@ -46,15 +43,8 @@ ARG OUTPUT_DIR
 
 RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get \
     -y install ca-certificates \
-    python3.10 \
-    python3.10-dev \
-    python3-pip \
     curl
 
-COPY ./docker/python/requirements.txt /etc/greptime/requirements.txt
-
-RUN python3 -m pip install -r /etc/greptime/requirements.txt
-
 WORKDIR /greptime
 COPY --from=builder /out/target/${OUTPUT_DIR}/greptime /greptime/bin/
 ENV PATH /greptime/bin/:$PATH

docker/ci/centos/Dockerfile

@@ -7,9 +7,7 @@ RUN sed -i s/^#.*baseurl=http/baseurl=http/g /etc/yum.repos.d/*.repo
 RUN yum install -y epel-release \
     openssl \
     openssl-devel  \
-    centos-release-scl  \
-    rh-python38  \
-    rh-python38-python-devel
+    centos-release-scl
 
 ARG TARGETARCH
 

docker/ci/ubuntu/Dockerfile

@@ -8,15 +8,8 @@ ARG TARGET_BIN=greptime
 
 RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y \
     ca-certificates \
-    python3.10 \
-    python3.10-dev \
-    python3-pip \
     curl
 
-COPY $DOCKER_BUILD_ROOT/docker/python/requirements.txt /etc/greptime/requirements.txt
-
-RUN python3 -m pip install -r /etc/greptime/requirements.txt
-
 ARG TARGETARCH
 
 ADD $TARGETARCH/$TARGET_BIN /greptime/bin/

docker/dev-builder/ubuntu/Dockerfile

@@ -15,8 +15,8 @@ RUN apt-get update && \
 RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y \
     libssl-dev \
     tzdata \
-    protobuf-compiler \
     curl \
+    unzip \
     ca-certificates \
     git \
     build-essential \
@@ -24,6 +24,20 @@ RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y \
     python3.10 \
     python3.10-dev
 
+ARG TARGETPLATFORM
+RUN echo "target platform: $TARGETPLATFORM"
+
+# Install protobuf, because the one in the apt is too old (v3.12).
+RUN if [ "$TARGETPLATFORM" = "linux/arm64" ]; then \
+    curl -OL https://github.com/protocolbuffers/protobuf/releases/download/v29.1/protoc-29.1-linux-aarch_64.zip && \
+    unzip protoc-29.1-linux-aarch_64.zip -d protoc3; \
+elif [ "$TARGETPLATFORM" = "linux/amd64" ]; then \
+    curl -OL https://github.com/protocolbuffers/protobuf/releases/download/v29.1/protoc-29.1-linux-x86_64.zip && \
+    unzip protoc-29.1-linux-x86_64.zip -d protoc3; \
+fi
+RUN mv protoc3/bin/* /usr/local/bin/
+RUN mv protoc3/include/* /usr/local/include/
+
 # https://github.com/GreptimeTeam/greptimedb/actions/runs/10935485852/job/30357457188#step:3:7106
 # `aws-lc-sys` require gcc >= 10.3.0 to work, hence alias to use gcc-10
 RUN apt-get remove -y gcc-9 g++-9 cpp-9 && \
@@ -49,7 +63,7 @@ RUN apt-get -y purge python3.8 && \
 # wildcard here. However, that requires the git's config files and the submodules all owned by the very same user.
 # It's troublesome to do this since the dev build runs in Docker, which is under user "root"; while outside the Docker,
 # it can be a different user that have prepared the submodules.
-RUN git config --global --add safe.directory *
+RUN git config --global --add safe.directory '*'
 
 # Install Python dependencies.
 COPY $DOCKER_BUILD_ROOT/docker/python/requirements.txt /etc/greptime/requirements.txt

docs/how-to/how-to-profile-cpu.md

@@ -3,7 +3,7 @@
 ## HTTP API
 Sample at 99 Hertz, for 5 seconds, output report in [protobuf format](https://github.com/google/pprof/blob/master/proto/profile.proto).
 ```bash
-curl -s '0:4000/debug/prof/cpu' > /tmp/pprof.out
+curl -X POST -s '0:4000/debug/prof/cpu' > /tmp/pprof.out
 ```
 
 Then you can use `pprof` command with the protobuf file.
@@ -13,10 +13,38 @@ go tool pprof -top /tmp/pprof.out
 
 Sample at 99 Hertz, for 60 seconds, output report in flamegraph format.
 ```bash
-curl -s '0:4000/debug/prof/cpu?seconds=60&output=flamegraph' > /tmp/pprof.svg
+curl -X POST -s '0:4000/debug/prof/cpu?seconds=60&output=flamegraph' > /tmp/pprof.svg
 ```
 
 Sample at 49 Hertz, for 10 seconds, output report in text format.
 ```bash
-curl -s '0:4000/debug/prof/cpu?seconds=10&frequency=49&output=text' > /tmp/pprof.txt
+curl -X POST -s '0:4000/debug/prof/cpu?seconds=10&frequency=49&output=text' > /tmp/pprof.txt
+```
+
+## Using `perf`
+
+First find the pid of GreptimeDB:
+
+Using `perf record` to profile GreptimeDB, at the sampling frequency of 99 hertz, and a duration of 60 seconds:
+
+```bash
+perf record -p <pid> --call-graph dwarf -F 99 -- sleep 60
+```
+
+The result will be saved to file `perf.data`.
+
+Then
+
+```bash
+perf script --no-inline > perf.out
+```
+
+Produce a flame graph out of it:
+
+```bash
+git clone https://github.com/brendangregg/FlameGraph
+
+FlameGraph/stackcollapse-perf.pl perf.out > perf.folded
+
+FlameGraph/flamegraph.pl perf.folded > perf.svg
 ```

docs/how-to/how-to-profile-memory.md

@@ -23,13 +23,13 @@ curl https://raw.githubusercontent.com/brendangregg/FlameGraph/master/flamegraph
 Start GreptimeDB instance with environment variables:
 
 ```bash
-MALLOC_CONF=prof:true,lg_prof_interval:28 ./target/debug/greptime standalone start
+MALLOC_CONF=prof:true ./target/debug/greptime standalone start
 ```
 
 Dump memory profiling data through HTTP API:
 
 ```bash
-curl localhost:4000/debug/prof/mem > greptime.hprof
+curl -X POST localhost:4000/debug/prof/mem > greptime.hprof
 ```
 
 You can periodically dump profiling data and compare them to find the delta memory usage.

rust-toolchain.toml

@@ -1,2 +1,3 @@
 [toolchain]
 channel = "nightly-2024-10-19"
+components = ["rust-analyzer"]

scripts/check-snafu.py

@@ -58,8 +58,10 @@ def main():
         if not check_snafu_in_files(branch_name, other_rust_files)
     ]
 
-    for name in unused_snafu:
-        print(name)
+    if unused_snafu:
+        print("Unused error variants:")
+        for name in unused_snafu:
+            print(name)
 
     if unused_snafu:
         raise SystemExit(1)

shell.nix

@@ -0,0 +1,27 @@
+let
+  nixpkgs = fetchTarball "https://github.com/NixOS/nixpkgs/tarball/nixos-unstable";
+  fenix = import (fetchTarball "https://github.com/nix-community/fenix/archive/main.tar.gz") {};
+  pkgs = import nixpkgs { config = {}; overlays = []; };
+in
+
+pkgs.mkShell rec {
+  nativeBuildInputs = with pkgs; [
+    pkg-config
+    git
+    clang
+    gcc
+    protobuf
+    mold
+    (fenix.fromToolchainFile {
+      dir = ./.;
+    })
+    cargo-nextest
+    taplo
+  ];
+
+  buildInputs = with pkgs; [
+    libgit2
+  ];
+
+  LD_LIBRARY_PATH = pkgs.lib.makeLibraryPath buildInputs;
+}

src/api/src/v1/column_def.rs

@@ -16,7 +16,7 @@ use std::collections::HashMap;
 
 use datatypes::schema::{
     ColumnDefaultConstraint, ColumnSchema, FulltextAnalyzer, FulltextOptions, COMMENT_KEY,
-    FULLTEXT_KEY, INVERTED_INDEX_KEY,
+    FULLTEXT_KEY, INVERTED_INDEX_KEY, SKIPPING_INDEX_KEY,
 };
 use greptime_proto::v1::Analyzer;
 use snafu::ResultExt;
@@ -29,6 +29,8 @@ use crate::v1::{ColumnDef, ColumnOptions, SemanticType};
 const FULLTEXT_GRPC_KEY: &str = "fulltext";
 /// Key used to store inverted index options in gRPC column options.
 const INVERTED_INDEX_GRPC_KEY: &str = "inverted_index";
+/// Key used to store skip index options in gRPC column options.
+const SKIPPING_INDEX_GRPC_KEY: &str = "skipping_index";
 
 /// Tries to construct a `ColumnSchema` from the given  `ColumnDef`.
 pub fn try_as_column_schema(column_def: &ColumnDef) -> Result<ColumnSchema> {
@@ -60,6 +62,9 @@ pub fn try_as_column_schema(column_def: &ColumnDef) -> Result<ColumnSchema> {
         if let Some(inverted_index) = options.options.get(INVERTED_INDEX_GRPC_KEY) {
             metadata.insert(INVERTED_INDEX_KEY.to_string(), inverted_index.clone());
         }
+        if let Some(skipping_index) = options.options.get(SKIPPING_INDEX_GRPC_KEY) {
+            metadata.insert(SKIPPING_INDEX_KEY.to_string(), skipping_index.clone());
+        }
     }
 
     ColumnSchema::new(&column_def.name, data_type.into(), column_def.is_nullable)
@@ -84,6 +89,11 @@ pub fn options_from_column_schema(column_schema: &ColumnSchema) -> Option<Column
             .options
             .insert(INVERTED_INDEX_GRPC_KEY.to_string(), inverted_index.clone());
     }
+    if let Some(skipping_index) = column_schema.metadata().get(SKIPPING_INDEX_KEY) {
+        options
+            .options
+            .insert(SKIPPING_INDEX_GRPC_KEY.to_string(), skipping_index.clone());
+    }
 
     (!options.options.is_empty()).then_some(options)
 }

src/auth/src/permission.rs

@@ -25,6 +25,7 @@ pub enum PermissionReq<'a> {
     GrpcRequest(&'a Request),
     SqlStatement(&'a Statement),
     PromQuery,
+    LogQuery,
     Opentsdb,
     LineProtocol,
     PromStoreWrite,

src/cache/Cargo.toml

@@ -11,4 +11,3 @@ common-macro.workspace = true
 common-meta.workspace = true
 moka.workspace = true
 snafu.workspace = true
-substrait.workspace = true

src/cache/src/lib.rs

@@ -19,9 +19,9 @@ use std::time::Duration;
 
 use catalog::kvbackend::new_table_cache;
 use common_meta::cache::{
-    new_table_flownode_set_cache, new_table_info_cache, new_table_name_cache,
-    new_table_route_cache, new_view_info_cache, CacheRegistry, CacheRegistryBuilder,
-    LayeredCacheRegistryBuilder,
+    new_schema_cache, new_table_flownode_set_cache, new_table_info_cache, new_table_name_cache,
+    new_table_route_cache, new_table_schema_cache, new_view_info_cache, CacheRegistry,
+    CacheRegistryBuilder, LayeredCacheRegistryBuilder,
 };
 use common_meta::kv_backend::KvBackendRef;
 use moka::future::CacheBuilder;
@@ -37,9 +37,47 @@ pub const TABLE_INFO_CACHE_NAME: &str = "table_info_cache";
 pub const VIEW_INFO_CACHE_NAME: &str = "view_info_cache";
 pub const TABLE_NAME_CACHE_NAME: &str = "table_name_cache";
 pub const TABLE_CACHE_NAME: &str = "table_cache";
+pub const SCHEMA_CACHE_NAME: &str = "schema_cache";
+pub const TABLE_SCHEMA_NAME_CACHE_NAME: &str = "table_schema_name_cache";
 pub const TABLE_FLOWNODE_SET_CACHE_NAME: &str = "table_flownode_set_cache";
 pub const TABLE_ROUTE_CACHE_NAME: &str = "table_route_cache";
 
+/// Builds cache registry for datanode, including:
+/// - Schema cache.
+/// - Table id to schema name cache.
+pub fn build_datanode_cache_registry(kv_backend: KvBackendRef) -> CacheRegistry {
+    // Builds table id schema name cache that never expires.
+    let cache = CacheBuilder::new(DEFAULT_CACHE_MAX_CAPACITY).build();
+    let table_id_schema_cache = Arc::new(new_table_schema_cache(
+        TABLE_SCHEMA_NAME_CACHE_NAME.to_string(),
+        cache,
+        kv_backend.clone(),
+    ));
+
+    // Builds schema cache
+    let cache = CacheBuilder::new(DEFAULT_CACHE_MAX_CAPACITY)
+        .time_to_live(DEFAULT_CACHE_TTL)
+        .time_to_idle(DEFAULT_CACHE_TTI)
+        .build();
+    let schema_cache = Arc::new(new_schema_cache(
+        SCHEMA_CACHE_NAME.to_string(),
+        cache,
+        kv_backend.clone(),
+    ));
+
+    CacheRegistryBuilder::default()
+        .add_cache(table_id_schema_cache)
+        .add_cache(schema_cache)
+        .build()
+}
+
+/// Builds cache registry for frontend and datanode, including:
+/// - Table info cache
+/// - Table name cache
+/// - Table route cache
+/// - Table flow node cache
+/// - View cache
+/// - Schema cache
 pub fn build_fundamental_cache_registry(kv_backend: KvBackendRef) -> CacheRegistry {
     // Builds table info cache
     let cache = CacheBuilder::new(DEFAULT_CACHE_MAX_CAPACITY)
@@ -95,12 +133,30 @@ pub fn build_fundamental_cache_registry(kv_backend: KvBackendRef) -> CacheRegist
         kv_backend.clone(),
     ));
 
+    // Builds schema cache
+    let cache = CacheBuilder::new(DEFAULT_CACHE_MAX_CAPACITY)
+        .time_to_live(DEFAULT_CACHE_TTL)
+        .time_to_idle(DEFAULT_CACHE_TTI)
+        .build();
+    let schema_cache = Arc::new(new_schema_cache(
+        SCHEMA_CACHE_NAME.to_string(),
+        cache,
+        kv_backend.clone(),
+    ));
+
+    let table_id_schema_cache = Arc::new(new_table_schema_cache(
+        TABLE_SCHEMA_NAME_CACHE_NAME.to_string(),
+        CacheBuilder::new(DEFAULT_CACHE_MAX_CAPACITY).build(),
+        kv_backend,
+    ));
     CacheRegistryBuilder::default()
         .add_cache(table_info_cache)
         .add_cache(table_name_cache)
         .add_cache(table_route_cache)
         .add_cache(view_info_cache)
         .add_cache(table_flownode_set_cache)
+        .add_cache(schema_cache)
+        .add_cache(table_id_schema_cache)
         .build()
 }
 

src/catalog/Cargo.toml

@@ -18,7 +18,6 @@ async-stream.workspace = true
 async-trait = "0.1"
 bytes.workspace = true
 common-catalog.workspace = true
-common-config.workspace = true
 common-error.workspace = true
 common-macro.workspace = true
 common-meta.workspace = true
@@ -58,7 +57,5 @@ catalog = { workspace = true, features = ["testing"] }
 chrono.workspace = true
 common-meta = { workspace = true, features = ["testing"] }
 common-query = { workspace = true, features = ["testing"] }
-common-test-util.workspace = true
-log-store.workspace = true
 object-store.workspace = true
 tokio.workspace = true

src/catalog/src/error.rs

@@ -64,6 +64,13 @@ pub enum Error {
         source: BoxedError,
     },
 
+    #[snafu(display("Failed to list flow stats"))]
+    ListFlowStats {
+        #[snafu(implicit)]
+        location: Location,
+        source: BoxedError,
+    },
+
     #[snafu(display("Failed to list flows in catalog {catalog}"))]
     ListFlows {
         #[snafu(implicit)]
@@ -326,6 +333,7 @@ impl ErrorExt for Error {
             | Error::ListSchemas { source, .. }
             | Error::ListTables { source, .. }
             | Error::ListFlows { source, .. }
+            | Error::ListFlowStats { source, .. }
             | Error::ListProcedures { source, .. }
             | Error::ListRegionStats { source, .. }
             | Error::ConvertProtoData { source, .. } => source.status_code(),

src/catalog/src/information_extension.rs

@@ -0,0 +1,101 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use api::v1::meta::ProcedureStatus;
+use common_error::ext::BoxedError;
+use common_meta::cluster::{ClusterInfo, NodeInfo};
+use common_meta::datanode::RegionStat;
+use common_meta::ddl::{ExecutorContext, ProcedureExecutor};
+use common_meta::key::flow::flow_state::FlowStat;
+use common_meta::rpc::procedure;
+use common_procedure::{ProcedureInfo, ProcedureState};
+use meta_client::MetaClientRef;
+use snafu::ResultExt;
+
+use crate::error;
+use crate::information_schema::InformationExtension;
+
+pub struct DistributedInformationExtension {
+    meta_client: MetaClientRef,
+}
+
+impl DistributedInformationExtension {
+    pub fn new(meta_client: MetaClientRef) -> Self {
+        Self { meta_client }
+    }
+}
+
+#[async_trait::async_trait]
+impl InformationExtension for DistributedInformationExtension {
+    type Error = crate::error::Error;
+
+    async fn nodes(&self) -> std::result::Result<Vec<NodeInfo>, Self::Error> {
+        self.meta_client
+            .list_nodes(None)
+            .await
+            .map_err(BoxedError::new)
+            .context(error::ListNodesSnafu)
+    }
+
+    async fn procedures(&self) -> std::result::Result<Vec<(String, ProcedureInfo)>, Self::Error> {
+        let procedures = self
+            .meta_client
+            .list_procedures(&ExecutorContext::default())
+            .await
+            .map_err(BoxedError::new)
+            .context(error::ListProceduresSnafu)?
+            .procedures;
+        let mut result = Vec::with_capacity(procedures.len());
+        for procedure in procedures {
+            let pid = match procedure.id {
+                Some(pid) => pid,
+                None => return error::ProcedureIdNotFoundSnafu {}.fail(),
+            };
+            let pid = procedure::pb_pid_to_pid(&pid)
+                .map_err(BoxedError::new)
+                .context(error::ConvertProtoDataSnafu)?;
+            let status = ProcedureStatus::try_from(procedure.status)
+                .map(|v| v.as_str_name())
+                .unwrap_or("Unknown")
+                .to_string();
+            let procedure_info = ProcedureInfo {
+                id: pid,
+                type_name: procedure.type_name,
+                start_time_ms: procedure.start_time_ms,
+                end_time_ms: procedure.end_time_ms,
+                state: ProcedureState::Running,
+                lock_keys: procedure.lock_keys,
+            };
+            result.push((status, procedure_info));
+        }
+
+        Ok(result)
+    }
+
+    async fn region_stats(&self) -> std::result::Result<Vec<RegionStat>, Self::Error> {
+        self.meta_client
+            .list_region_stats()
+            .await
+            .map_err(BoxedError::new)
+            .context(error::ListRegionStatsSnafu)
+    }
+
+    async fn flow_stats(&self) -> std::result::Result<Option<FlowStat>, Self::Error> {
+        self.meta_client
+            .list_flow_stats()
+            .await
+            .map_err(BoxedError::new)
+            .context(crate::error::ListFlowStatsSnafu)
+    }
+}

src/catalog/src/kvbackend.rs

@@ -12,7 +12,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-pub use client::{CachedMetaKvBackend, CachedMetaKvBackendBuilder, MetaKvBackend};
+pub use client::{CachedKvBackend, CachedKvBackendBuilder, MetaKvBackend};
 
 mod client;
 mod manager;

src/catalog/src/kvbackend/client.rs

@@ -18,10 +18,11 @@ use std::sync::atomic::{AtomicUsize, Ordering};
 use std::sync::{Arc, Mutex};
 use std::time::Duration;
 
-use common_error::ext::BoxedError;
+use common_error::ext::{BoxedError, ErrorExt};
 use common_meta::cache_invalidator::KvCacheInvalidator;
 use common_meta::error::Error::CacheNotGet;
 use common_meta::error::{CacheNotGetSnafu, Error, ExternalSnafu, GetKvCacheSnafu, Result};
+use common_meta::kv_backend::txn::{Txn, TxnResponse};
 use common_meta::kv_backend::{KvBackend, KvBackendRef, TxnService};
 use common_meta::rpc::store::{
     BatchDeleteRequest, BatchDeleteResponse, BatchGetRequest, BatchGetResponse, BatchPutRequest,
@@ -36,26 +37,27 @@ use snafu::{OptionExt, ResultExt};
 
 use crate::metrics::{
     METRIC_CATALOG_KV_BATCH_GET, METRIC_CATALOG_KV_GET, METRIC_CATALOG_KV_REMOTE_GET,
+    METRIC_META_CLIENT_GET,
 };
 
 const DEFAULT_CACHE_MAX_CAPACITY: u64 = 10000;
 const DEFAULT_CACHE_TTL: Duration = Duration::from_secs(10 * 60);
 const DEFAULT_CACHE_TTI: Duration = Duration::from_secs(5 * 60);
 
-pub struct CachedMetaKvBackendBuilder {
+pub struct CachedKvBackendBuilder {
     cache_max_capacity: Option<u64>,
     cache_ttl: Option<Duration>,
     cache_tti: Option<Duration>,
-    meta_client: Arc<MetaClient>,
+    inner: KvBackendRef,
 }
 
-impl CachedMetaKvBackendBuilder {
-    pub fn new(meta_client: Arc<MetaClient>) -> Self {
+impl CachedKvBackendBuilder {
+    pub fn new(inner: KvBackendRef) -> Self {
         Self {
             cache_max_capacity: None,
             cache_ttl: None,
             cache_tti: None,
-            meta_client,
+            inner,
         }
     }
 
@@ -74,7 +76,7 @@ impl CachedMetaKvBackendBuilder {
         self
     }
 
-    pub fn build(self) -> CachedMetaKvBackend {
+    pub fn build(self) -> CachedKvBackend {
         let cache_max_capacity = self
             .cache_max_capacity
             .unwrap_or(DEFAULT_CACHE_MAX_CAPACITY);
@@ -85,14 +87,11 @@ impl CachedMetaKvBackendBuilder {
             .time_to_live(cache_ttl)
             .time_to_idle(cache_tti)
             .build();
-
-        let kv_backend = Arc::new(MetaKvBackend {
-            client: self.meta_client,
-        });
+        let kv_backend = self.inner;
         let name = format!("CachedKvBackend({})", kv_backend.name());
         let version = AtomicUsize::new(0);
 
-        CachedMetaKvBackend {
+        CachedKvBackend {
             kv_backend,
             cache,
             name,
@@ -112,19 +111,29 @@ pub type CacheBackend = Cache<Vec<u8>, KeyValue>;
 /// Therefore, it is recommended to use CachedMetaKvBackend to only read metadata related
 /// information. Note: If you read other information, you may read expired data, which depends on
 /// TTL and TTI for cache.
-pub struct CachedMetaKvBackend {
+pub struct CachedKvBackend {
     kv_backend: KvBackendRef,
     cache: CacheBackend,
     name: String,
     version: AtomicUsize,
 }
 
-impl TxnService for CachedMetaKvBackend {
+#[async_trait::async_trait]
+impl TxnService for CachedKvBackend {
     type Error = Error;
+
+    async fn txn(&self, txn: Txn) -> std::result::Result<TxnResponse, Self::Error> {
+        // TODO(hl): txn of CachedKvBackend simply pass through to inner backend without invalidating caches.
+        self.kv_backend.txn(txn).await
+    }
+
+    fn max_txn_ops(&self) -> usize {
+        self.kv_backend.max_txn_ops()
+    }
 }
 
 #[async_trait::async_trait]
-impl KvBackend for CachedMetaKvBackend {
+impl KvBackend for CachedKvBackend {
     fn name(&self) -> &str {
         &self.name
     }
@@ -284,7 +293,7 @@ impl KvBackend for CachedMetaKvBackend {
         }
         .map_err(|e| {
             GetKvCacheSnafu {
-                err_msg: e.to_string(),
+                err_msg: e.output_msg(),
             }
             .build()
         });
@@ -305,7 +314,7 @@ impl KvBackend for CachedMetaKvBackend {
 }
 
 #[async_trait::async_trait]
-impl KvCacheInvalidator for CachedMetaKvBackend {
+impl KvCacheInvalidator for CachedKvBackend {
     async fn invalidate_key(&self, key: &[u8]) {
         self.create_new_version();
         self.cache.invalidate(key).await;
@@ -313,7 +322,7 @@ impl KvCacheInvalidator for CachedMetaKvBackend {
     }
 }
 
-impl CachedMetaKvBackend {
+impl CachedKvBackend {
     // only for test
     #[cfg(test)]
     fn wrap(kv_backend: KvBackendRef) -> Self {
@@ -437,6 +446,8 @@ impl KvBackend for MetaKvBackend {
     }
 
     async fn get(&self, key: &[u8]) -> Result<Option<KeyValue>> {
+        let _timer = METRIC_META_CLIENT_GET.start_timer();
+
         let mut response = self
             .client
             .range(RangeRequest::new().with_key(key))
@@ -466,7 +477,7 @@ mod tests {
     use common_meta::rpc::KeyValue;
     use dashmap::DashMap;
 
-    use super::CachedMetaKvBackend;
+    use super::CachedKvBackend;
 
     #[derive(Default)]
     pub struct SimpleKvBackend {
@@ -540,7 +551,7 @@ mod tests {
     async fn test_cached_kv_backend() {
         let simple_kv = Arc::new(SimpleKvBackend::default());
         let get_execute_times = simple_kv.get_execute_times.clone();
-        let cached_kv = CachedMetaKvBackend::wrap(simple_kv);
+        let cached_kv = CachedKvBackend::wrap(simple_kv);
 
         add_some_vals(&cached_kv).await;
 

src/catalog/src/kvbackend/table_cache.rs

@@ -38,7 +38,7 @@ pub fn new_table_cache(
 ) -> TableCache {
     let init = init_factory(table_info_cache, table_name_cache);
 
-    CacheContainer::new(name, cache, Box::new(invalidator), init, Box::new(filter))
+    CacheContainer::new(name, cache, Box::new(invalidator), init, filter)
 }
 
 fn init_factory(

src/catalog/src/lib.rs

@@ -30,6 +30,7 @@ use table::TableRef;
 use crate::error::Result;
 
 pub mod error;
+pub mod information_extension;
 pub mod kvbackend;
 pub mod memory;
 mod metrics;

src/catalog/src/metrics.rs

@@ -34,4 +34,6 @@ lazy_static! {
         register_histogram!("greptime_catalog_kv_get", "catalog kv get").unwrap();
     pub static ref METRIC_CATALOG_KV_BATCH_GET: Histogram =
         register_histogram!("greptime_catalog_kv_batch_get", "catalog kv batch get").unwrap();
+    pub static ref METRIC_META_CLIENT_GET: Histogram =
+        register_histogram!("greptime_meta_client_get", "meta client get").unwrap();
 }

src/catalog/src/system_schema/information_schema.rs

@@ -35,6 +35,7 @@ use common_catalog::consts::{self, DEFAULT_CATALOG_NAME, INFORMATION_SCHEMA_NAME
 use common_error::ext::ErrorExt;
 use common_meta::cluster::NodeInfo;
 use common_meta::datanode::RegionStat;
+use common_meta::key::flow::flow_state::FlowStat;
 use common_meta::key::flow::FlowMetadataManager;
 use common_procedure::ProcedureInfo;
 use common_recordbatch::SendableRecordBatchStream;
@@ -192,6 +193,7 @@ impl SystemSchemaProviderInner for InformationSchemaProvider {
             )) as _),
             FLOWS => Some(Arc::new(InformationSchemaFlows::new(
                 self.catalog_name.clone(),
+                self.catalog_manager.clone(),
                 self.flow_metadata_manager.clone(),
             )) as _),
             PROCEDURE_INFO => Some(
@@ -338,6 +340,9 @@ pub trait InformationExtension {
 
     /// Gets the region statistics.
     async fn region_stats(&self) -> std::result::Result<Vec<RegionStat>, Self::Error>;
+
+    /// Get the flow statistics. If no flownode is available, return `None`.
+    async fn flow_stats(&self) -> std::result::Result<Option<FlowStat>, Self::Error>;
 }
 
 pub struct NoopInformationExtension;
@@ -357,4 +362,8 @@ impl InformationExtension for NoopInformationExtension {
     async fn region_stats(&self) -> std::result::Result<Vec<RegionStat>, Self::Error> {
         Ok(vec![])
     }
+
+    async fn flow_stats(&self) -> std::result::Result<Option<FlowStat>, Self::Error> {
+        Ok(None)
+    }
 }

src/catalog/src/system_schema/information_schema/flows.rs

@@ -12,11 +12,12 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-use std::sync::Arc;
+use std::sync::{Arc, Weak};
 
 use common_catalog::consts::INFORMATION_SCHEMA_FLOW_TABLE_ID;
 use common_error::ext::BoxedError;
 use common_meta::key::flow::flow_info::FlowInfoValue;
+use common_meta::key::flow::flow_state::FlowStat;
 use common_meta::key::flow::FlowMetadataManager;
 use common_meta::key::FlowId;
 use common_recordbatch::adapter::RecordBatchStreamAdapter;
@@ -28,7 +29,9 @@ use datatypes::prelude::ConcreteDataType as CDT;
 use datatypes::scalars::ScalarVectorBuilder;
 use datatypes::schema::{ColumnSchema, Schema, SchemaRef};
 use datatypes::value::Value;
-use datatypes::vectors::{Int64VectorBuilder, StringVectorBuilder, UInt32VectorBuilder, VectorRef};
+use datatypes::vectors::{
+    Int64VectorBuilder, StringVectorBuilder, UInt32VectorBuilder, UInt64VectorBuilder, VectorRef,
+};
 use futures::TryStreamExt;
 use snafu::{OptionExt, ResultExt};
 use store_api::storage::{ScanRequest, TableId};
@@ -38,6 +41,8 @@ use crate::error::{
 };
 use crate::information_schema::{Predicates, FLOWS};
 use crate::system_schema::information_schema::InformationTable;
+use crate::system_schema::utils;
+use crate::CatalogManager;
 
 const INIT_CAPACITY: usize = 42;
 
@@ -45,6 +50,7 @@ const INIT_CAPACITY: usize = 42;
 // pk is (flow_name, flow_id, table_catalog)
 pub const FLOW_NAME: &str = "flow_name";
 pub const FLOW_ID: &str = "flow_id";
+pub const STATE_SIZE: &str = "state_size";
 pub const TABLE_CATALOG: &str = "table_catalog";
 pub const FLOW_DEFINITION: &str = "flow_definition";
 pub const COMMENT: &str = "comment";
@@ -55,20 +61,24 @@ pub const FLOWNODE_IDS: &str = "flownode_ids";
 pub const OPTIONS: &str = "options";
 
 /// The `information_schema.flows` to provides information about flows in databases.
+///
 pub(super) struct InformationSchemaFlows {
     schema: SchemaRef,
     catalog_name: String,
+    catalog_manager: Weak<dyn CatalogManager>,
     flow_metadata_manager: Arc<FlowMetadataManager>,
 }
 
 impl InformationSchemaFlows {
     pub(super) fn new(
         catalog_name: String,
+        catalog_manager: Weak<dyn CatalogManager>,
         flow_metadata_manager: Arc<FlowMetadataManager>,
     ) -> Self {
         Self {
             schema: Self::schema(),
             catalog_name,
+            catalog_manager,
             flow_metadata_manager,
         }
     }
@@ -80,6 +90,7 @@ impl InformationSchemaFlows {
             vec![
                 (FLOW_NAME, CDT::string_datatype(), false),
                 (FLOW_ID, CDT::uint32_datatype(), false),
+                (STATE_SIZE, CDT::uint64_datatype(), true),
                 (TABLE_CATALOG, CDT::string_datatype(), false),
                 (FLOW_DEFINITION, CDT::string_datatype(), false),
                 (COMMENT, CDT::string_datatype(), true),
@@ -99,6 +110,7 @@ impl InformationSchemaFlows {
         InformationSchemaFlowsBuilder::new(
             self.schema.clone(),
             self.catalog_name.clone(),
+            self.catalog_manager.clone(),
             &self.flow_metadata_manager,
         )
     }
@@ -144,10 +156,12 @@ impl InformationTable for InformationSchemaFlows {
 struct InformationSchemaFlowsBuilder {
     schema: SchemaRef,
     catalog_name: String,
+    catalog_manager: Weak<dyn CatalogManager>,
     flow_metadata_manager: Arc<FlowMetadataManager>,
 
     flow_names: StringVectorBuilder,
     flow_ids: UInt32VectorBuilder,
+    state_sizes: UInt64VectorBuilder,
     table_catalogs: StringVectorBuilder,
     raw_sqls: StringVectorBuilder,
     comments: StringVectorBuilder,
@@ -162,15 +176,18 @@ impl InformationSchemaFlowsBuilder {
     fn new(
         schema: SchemaRef,
         catalog_name: String,
+        catalog_manager: Weak<dyn CatalogManager>,
         flow_metadata_manager: &Arc<FlowMetadataManager>,
     ) -> Self {
         Self {
             schema,
             catalog_name,
+            catalog_manager,
             flow_metadata_manager: flow_metadata_manager.clone(),
 
             flow_names: StringVectorBuilder::with_capacity(INIT_CAPACITY),
             flow_ids: UInt32VectorBuilder::with_capacity(INIT_CAPACITY),
+            state_sizes: UInt64VectorBuilder::with_capacity(INIT_CAPACITY),
             table_catalogs: StringVectorBuilder::with_capacity(INIT_CAPACITY),
             raw_sqls: StringVectorBuilder::with_capacity(INIT_CAPACITY),
             comments: StringVectorBuilder::with_capacity(INIT_CAPACITY),
@@ -195,6 +212,11 @@ impl InformationSchemaFlowsBuilder {
             .flow_names(&catalog_name)
             .await;
 
+        let flow_stat = {
+            let information_extension = utils::information_extension(&self.catalog_manager)?;
+            information_extension.flow_stats().await?
+        };
+
         while let Some((flow_name, flow_id)) = stream
             .try_next()
             .await
@@ -213,7 +235,7 @@ impl InformationSchemaFlowsBuilder {
                     catalog_name: catalog_name.to_string(),
                     flow_name: flow_name.to_string(),
                 })?;
-            self.add_flow(&predicates, flow_id.flow_id(), flow_info)?;
+            self.add_flow(&predicates, flow_id.flow_id(), flow_info, &flow_stat)?;
         }
 
         self.finish()
@@ -224,6 +246,7 @@ impl InformationSchemaFlowsBuilder {
         predicates: &Predicates,
         flow_id: FlowId,
         flow_info: FlowInfoValue,
+        flow_stat: &Option<FlowStat>,
     ) -> Result<()> {
         let row = [
             (FLOW_NAME, &Value::from(flow_info.flow_name().to_string())),
@@ -238,6 +261,11 @@ impl InformationSchemaFlowsBuilder {
         }
         self.flow_names.push(Some(flow_info.flow_name()));
         self.flow_ids.push(Some(flow_id));
+        self.state_sizes.push(
+            flow_stat
+                .as_ref()
+                .and_then(|state| state.state_size.get(&flow_id).map(|v| *v as u64)),
+        );
         self.table_catalogs.push(Some(flow_info.catalog_name()));
         self.raw_sqls.push(Some(flow_info.raw_sql()));
         self.comments.push(Some(flow_info.comment()));
@@ -270,6 +298,7 @@ impl InformationSchemaFlowsBuilder {
         let columns: Vec<VectorRef> = vec![
             Arc::new(self.flow_names.finish()),
             Arc::new(self.flow_ids.finish()),
+            Arc::new(self.state_sizes.finish()),
             Arc::new(self.table_catalogs.finish()),
             Arc::new(self.raw_sqls.finish()),
             Arc::new(self.comments.finish()),

src/catalog/src/system_schema/information_schema/key_column_usage.rs

@@ -54,6 +54,10 @@ const INIT_CAPACITY: usize = 42;
 pub(crate) const PRI_CONSTRAINT_NAME: &str = "PRIMARY";
 /// Time index constraint name
 pub(crate) const TIME_INDEX_CONSTRAINT_NAME: &str = "TIME INDEX";
+/// Inverted index constraint name
+pub(crate) const INVERTED_INDEX_CONSTRAINT_NAME: &str = "INVERTED INDEX";
+/// Fulltext index constraint name
+pub(crate) const FULLTEXT_INDEX_CONSTRAINT_NAME: &str = "FULLTEXT INDEX";
 
 /// The virtual table implementation for `information_schema.KEY_COLUMN_USAGE`.
 pub(super) struct InformationSchemaKeyColumnUsage {
@@ -216,14 +220,13 @@ impl InformationSchemaKeyColumnUsageBuilder {
             let mut stream = catalog_manager.tables(&catalog_name, &schema_name, None);
 
             while let Some(table) = stream.try_next().await? {
-                let mut primary_constraints = vec![];
-
                 let table_info = table.table_info();
                 let table_name = &table_info.name;
                 let keys = &table_info.meta.primary_key_indices;
                 let schema = table.schema();
 
                 for (idx, column) in schema.column_schemas().iter().enumerate() {
+                    let mut constraints = vec![];
                     if column.is_time_index() {
                         self.add_key_column_usage(
                             &predicates,
@@ -236,30 +239,31 @@ impl InformationSchemaKeyColumnUsageBuilder {
                             1, //always 1 for time index
                         );
                     }
-                    if keys.contains(&idx) {
-                        primary_constraints.push((
-                            catalog_name.clone(),
-                            schema_name.clone(),
-                            table_name.to_string(),
-                            column.name.clone(),
-                        ));
-                    }
                     // TODO(dimbtp): foreign key constraint not supported yet
-                }
+                    if keys.contains(&idx) {
+                        constraints.push(PRI_CONSTRAINT_NAME);
+                    }
+                    if column.is_inverted_indexed() {
+                        constraints.push(INVERTED_INDEX_CONSTRAINT_NAME);
+                    }
 
-                for (i, (catalog_name, schema_name, table_name, column_name)) in
-                    primary_constraints.into_iter().enumerate()
-                {
-                    self.add_key_column_usage(
-                        &predicates,
-                        &schema_name,
-                        PRI_CONSTRAINT_NAME,
-                        &catalog_name,
-                        &schema_name,
-                        &table_name,
-                        &column_name,
-                        i as u32 + 1,
-                    );
+                    if column.has_fulltext_index_key() {
+                        constraints.push(FULLTEXT_INDEX_CONSTRAINT_NAME);
+                    }
+
+                    if !constraints.is_empty() {
+                        let aggregated_constraints = constraints.join(", ");
+                        self.add_key_column_usage(
+                            &predicates,
+                            &schema_name,
+                            &aggregated_constraints,
+                            &catalog_name,
+                            &schema_name,
+                            table_name,
+                            &column.name,
+                            idx as u32 + 1,
+                        );
+                    }
                 }
             }
         }

src/cli/Cargo.toml

@@ -0,0 +1,62 @@
+[package]
+name = "cli"
+version.workspace = true
+edition.workspace = true
+license.workspace = true
+
+[lints]
+workspace = true
+
+[dependencies]
+async-trait.workspace = true
+auth.workspace = true
+base64.workspace = true
+cache.workspace = true
+catalog.workspace = true
+chrono.workspace = true
+clap.workspace = true
+client = { workspace = true, features = ["testing"] }
+common-base.workspace = true
+common-catalog.workspace = true
+common-config.workspace = true
+common-error.workspace = true
+common-grpc.workspace = true
+common-macro.workspace = true
+common-meta.workspace = true
+common-procedure.workspace = true
+common-query.workspace = true
+common-recordbatch.workspace = true
+common-runtime.workspace = true
+common-telemetry = { workspace = true, features = [
+    "deadlock_detection",
+] }
+common-time.workspace = true
+common-version.workspace = true
+common-wal.workspace = true
+datatypes.workspace = true
+either = "1.8"
+etcd-client.workspace = true
+futures.workspace = true
+humantime.workspace = true
+meta-client.workspace = true
+nu-ansi-term = "0.46"
+query.workspace = true
+rand.workspace = true
+reqwest.workspace = true
+rustyline = "10.1"
+serde.workspace = true
+serde_json.workspace = true
+servers.workspace = true
+session.workspace = true
+snafu.workspace = true
+store-api.workspace = true
+substrait.workspace = true
+table.workspace = true
+tokio.workspace = true
+tracing-appender.workspace = true
+
+[dev-dependencies]
+common-test-util.workspace = true
+common-version.workspace = true
+serde.workspace = true
+tempfile.workspace = true

src/cli/src/bench.rs

@@ -19,6 +19,7 @@ use std::time::Duration;
 
 use async_trait::async_trait;
 use clap::Parser;
+use common_error::ext::BoxedError;
 use common_meta::key::{TableMetadataManager, TableMetadataManagerRef};
 use common_meta::kv_backend::etcd::EtcdStore;
 use common_meta::peer::Peer;
@@ -30,11 +31,9 @@ use rand::Rng;
 use store_api::storage::RegionNumber;
 use table::metadata::{RawTableInfo, RawTableMeta, TableId, TableIdent, TableType};
 use table::table_name::TableName;
-use tracing_appender::non_blocking::WorkerGuard;
 
 use self::metadata::TableMetadataBencher;
-use crate::cli::{Instance, Tool};
-use crate::error::Result;
+use crate::Tool;
 
 mod metadata;
 
@@ -62,7 +61,7 @@ pub struct BenchTableMetadataCommand {
 }
 
 impl BenchTableMetadataCommand {
-    pub async fn build(&self, guard: Vec<WorkerGuard>) -> Result<Instance> {
+    pub async fn build(&self) -> std::result::Result<Box<dyn Tool>, BoxedError> {
         let etcd_store = EtcdStore::with_endpoints([&self.etcd_addr], 128)
             .await
             .unwrap();
@@ -73,7 +72,7 @@ impl BenchTableMetadataCommand {
             table_metadata_manager,
             count: self.count,
         };
-        Ok(Instance::new(Box::new(tool), guard))
+        Ok(Box::new(tool))
     }
 }
 
@@ -84,7 +83,7 @@ struct BenchTableMetadata {
 
 #[async_trait]
 impl Tool for BenchTableMetadata {
-    async fn do_work(&self) -> Result<()> {
+    async fn do_work(&self) -> std::result::Result<(), BoxedError> {
         let bencher = TableMetadataBencher::new(self.table_metadata_manager.clone(), self.count);
         bencher.bench_create().await;
         bencher.bench_get().await;

src/cli/src/bench/metadata.rs

@@ -18,7 +18,7 @@ use common_meta::key::table_route::TableRouteValue;
 use common_meta::key::TableMetadataManagerRef;
 use table::table_name::TableName;
 
-use crate::cli::bench::{
+use crate::bench::{
     bench_self_recorded, create_region_routes, create_region_wal_options, create_table_info,
 };
 

src/cli/src/database.rs

@@ -26,7 +26,8 @@ use snafu::ResultExt;
 
 use crate::error::{HttpQuerySqlSnafu, Result, SerdeJsonSnafu};
 
-pub(crate) struct DatabaseClient {
+#[derive(Debug, Clone)]
+pub struct DatabaseClient {
     addr: String,
     catalog: String,
     auth_header: Option<String>,

src/cli/src/error.rs

@@ -0,0 +1,316 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use std::any::Any;
+
+use common_error::ext::{BoxedError, ErrorExt};
+use common_error::status_code::StatusCode;
+use common_macro::stack_trace_debug;
+use rustyline::error::ReadlineError;
+use snafu::{Location, Snafu};
+
+#[derive(Snafu)]
+#[snafu(visibility(pub))]
+#[stack_trace_debug]
+pub enum Error {
+    #[snafu(display("Failed to install ring crypto provider: {}", msg))]
+    InitTlsProvider {
+        #[snafu(implicit)]
+        location: Location,
+        msg: String,
+    },
+    #[snafu(display("Failed to create default catalog and schema"))]
+    InitMetadata {
+        #[snafu(implicit)]
+        location: Location,
+        source: common_meta::error::Error,
+    },
+
+    #[snafu(display("Failed to init DDL manager"))]
+    InitDdlManager {
+        #[snafu(implicit)]
+        location: Location,
+        source: common_meta::error::Error,
+    },
+
+    #[snafu(display("Failed to init default timezone"))]
+    InitTimezone {
+        #[snafu(implicit)]
+        location: Location,
+        source: common_time::error::Error,
+    },
+
+    #[snafu(display("Failed to start procedure manager"))]
+    StartProcedureManager {
+        #[snafu(implicit)]
+        location: Location,
+        source: common_procedure::error::Error,
+    },
+
+    #[snafu(display("Failed to stop procedure manager"))]
+    StopProcedureManager {
+        #[snafu(implicit)]
+        location: Location,
+        source: common_procedure::error::Error,
+    },
+
+    #[snafu(display("Failed to start wal options allocator"))]
+    StartWalOptionsAllocator {
+        #[snafu(implicit)]
+        location: Location,
+        source: common_meta::error::Error,
+    },
+
+    #[snafu(display("Missing config, msg: {}", msg))]
+    MissingConfig {
+        msg: String,
+        #[snafu(implicit)]
+        location: Location,
+    },
+
+    #[snafu(display("Illegal config: {}", msg))]
+    IllegalConfig {
+        msg: String,
+        #[snafu(implicit)]
+        location: Location,
+    },
+
+    #[snafu(display("Invalid REPL command: {reason}"))]
+    InvalidReplCommand { reason: String },
+
+    #[snafu(display("Cannot create REPL"))]
+    ReplCreation {
+        #[snafu(source)]
+        error: ReadlineError,
+        #[snafu(implicit)]
+        location: Location,
+    },
+
+    #[snafu(display("Error reading command"))]
+    Readline {
+        #[snafu(source)]
+        error: ReadlineError,
+        #[snafu(implicit)]
+        location: Location,
+    },
+
+    #[snafu(display("Failed to request database, sql: {sql}"))]
+    RequestDatabase {
+        sql: String,
+        #[snafu(source)]
+        source: client::Error,
+        #[snafu(implicit)]
+        location: Location,
+    },
+
+    #[snafu(display("Failed to collect RecordBatches"))]
+    CollectRecordBatches {
+        #[snafu(implicit)]
+        location: Location,
+        source: common_recordbatch::error::Error,
+    },
+
+    #[snafu(display("Failed to pretty print Recordbatches"))]
+    PrettyPrintRecordBatches {
+        #[snafu(implicit)]
+        location: Location,
+        source: common_recordbatch::error::Error,
+    },
+
+    #[snafu(display("Failed to start Meta client"))]
+    StartMetaClient {
+        #[snafu(implicit)]
+        location: Location,
+        source: meta_client::error::Error,
+    },
+
+    #[snafu(display("Failed to parse SQL: {}", sql))]
+    ParseSql {
+        sql: String,
+        #[snafu(implicit)]
+        location: Location,
+        source: query::error::Error,
+    },
+
+    #[snafu(display("Failed to plan statement"))]
+    PlanStatement {
+        #[snafu(implicit)]
+        location: Location,
+        source: query::error::Error,
+    },
+
+    #[snafu(display("Failed to encode logical plan in substrait"))]
+    SubstraitEncodeLogicalPlan {
+        #[snafu(implicit)]
+        location: Location,
+        source: substrait::error::Error,
+    },
+
+    #[snafu(display("Failed to load layered config"))]
+    LoadLayeredConfig {
+        #[snafu(source(from(common_config::error::Error, Box::new)))]
+        source: Box<common_config::error::Error>,
+        #[snafu(implicit)]
+        location: Location,
+    },
+
+    #[snafu(display("Failed to connect to Etcd at {etcd_addr}"))]
+    ConnectEtcd {
+        etcd_addr: String,
+        #[snafu(source)]
+        error: etcd_client::Error,
+        #[snafu(implicit)]
+        location: Location,
+    },
+
+    #[snafu(display("Failed to serde json"))]
+    SerdeJson {
+        #[snafu(source)]
+        error: serde_json::error::Error,
+        #[snafu(implicit)]
+        location: Location,
+    },
+
+    #[snafu(display("Failed to run http request: {reason}"))]
+    HttpQuerySql {
+        reason: String,
+        #[snafu(source)]
+        error: reqwest::Error,
+        #[snafu(implicit)]
+        location: Location,
+    },
+
+    #[snafu(display("Empty result from output"))]
+    EmptyResult {
+        #[snafu(implicit)]
+        location: Location,
+    },
+
+    #[snafu(display("Failed to manipulate file"))]
+    FileIo {
+        #[snafu(implicit)]
+        location: Location,
+        #[snafu(source)]
+        error: std::io::Error,
+    },
+
+    #[snafu(display("Failed to create directory {}", dir))]
+    CreateDir {
+        dir: String,
+        #[snafu(source)]
+        error: std::io::Error,
+    },
+
+    #[snafu(display("Failed to spawn thread"))]
+    SpawnThread {
+        #[snafu(source)]
+        error: std::io::Error,
+    },
+
+    #[snafu(display("Other error"))]
+    Other {
+        source: BoxedError,
+        #[snafu(implicit)]
+        location: Location,
+    },
+
+    #[snafu(display("Failed to build runtime"))]
+    BuildRuntime {
+        #[snafu(implicit)]
+        location: Location,
+        source: common_runtime::error::Error,
+    },
+
+    #[snafu(display("Failed to get cache from cache registry: {}", name))]
+    CacheRequired {
+        #[snafu(implicit)]
+        location: Location,
+        name: String,
+    },
+
+    #[snafu(display("Failed to build cache registry"))]
+    BuildCacheRegistry {
+        #[snafu(implicit)]
+        location: Location,
+        source: cache::error::Error,
+    },
+
+    #[snafu(display("Failed to initialize meta client"))]
+    MetaClientInit {
+        #[snafu(implicit)]
+        location: Location,
+        source: meta_client::error::Error,
+    },
+
+    #[snafu(display("Cannot find schema {schema} in catalog {catalog}"))]
+    SchemaNotFound {
+        catalog: String,
+        schema: String,
+        #[snafu(implicit)]
+        location: Location,
+    },
+}
+
+pub type Result<T> = std::result::Result<T, Error>;
+
+impl ErrorExt for Error {
+    fn status_code(&self) -> StatusCode {
+        match self {
+            Error::InitMetadata { source, .. } | Error::InitDdlManager { source, .. } => {
+                source.status_code()
+            }
+
+            Error::MissingConfig { .. }
+            | Error::LoadLayeredConfig { .. }
+            | Error::IllegalConfig { .. }
+            | Error::InvalidReplCommand { .. }
+            | Error::InitTimezone { .. }
+            | Error::ConnectEtcd { .. }
+            | Error::CreateDir { .. }
+            | Error::EmptyResult { .. } => StatusCode::InvalidArguments,
+
+            Error::StartProcedureManager { source, .. }
+            | Error::StopProcedureManager { source, .. } => source.status_code(),
+            Error::StartWalOptionsAllocator { source, .. } => source.status_code(),
+            Error::ReplCreation { .. } | Error::Readline { .. } | Error::HttpQuerySql { .. } => {
+                StatusCode::Internal
+            }
+            Error::RequestDatabase { source, .. } => source.status_code(),
+            Error::CollectRecordBatches { source, .. }
+            | Error::PrettyPrintRecordBatches { source, .. } => source.status_code(),
+            Error::StartMetaClient { source, .. } => source.status_code(),
+            Error::ParseSql { source, .. } | Error::PlanStatement { source, .. } => {
+                source.status_code()
+            }
+            Error::SubstraitEncodeLogicalPlan { source, .. } => source.status_code(),
+
+            Error::SerdeJson { .. }
+            | Error::FileIo { .. }
+            | Error::SpawnThread { .. }
+            | Error::InitTlsProvider { .. } => StatusCode::Unexpected,
+
+            Error::Other { source, .. } => source.status_code(),
+
+            Error::BuildRuntime { source, .. } => source.status_code(),
+
+            Error::CacheRequired { .. } | Error::BuildCacheRegistry { .. } => StatusCode::Internal,
+            Error::MetaClientInit { source, .. } => source.status_code(),
+            Error::SchemaNotFound { .. } => StatusCode::DatabaseNotFound,
+        }
+    }
+
+    fn as_any(&self) -> &dyn Any {
+        self
+    }
+}

src/cli/src/export.rs

@@ -19,6 +19,7 @@ use std::time::Duration;
 
 use async_trait::async_trait;
 use clap::{Parser, ValueEnum};
+use common_error::ext::BoxedError;
 use common_telemetry::{debug, error, info};
 use serde_json::Value;
 use snafu::{OptionExt, ResultExt};
@@ -26,11 +27,10 @@ use tokio::fs::File;
 use tokio::io::{AsyncWriteExt, BufWriter};
 use tokio::sync::Semaphore;
 use tokio::time::Instant;
-use tracing_appender::non_blocking::WorkerGuard;
 
-use crate::cli::database::DatabaseClient;
-use crate::cli::{database, Instance, Tool};
+use crate::database::DatabaseClient;
 use crate::error::{EmptyResultSnafu, Error, FileIoSnafu, Result, SchemaNotFoundSnafu};
+use crate::{database, Tool};
 
 type TableReference = (String, String, String);
 
@@ -94,8 +94,9 @@ pub struct ExportCommand {
 }
 
 impl ExportCommand {
-    pub async fn build(&self, guard: Vec<WorkerGuard>) -> Result<Instance> {
-        let (catalog, schema) = database::split_database(&self.database)?;
+    pub async fn build(&self) -> std::result::Result<Box<dyn Tool>, BoxedError> {
+        let (catalog, schema) =
+            database::split_database(&self.database).map_err(BoxedError::new)?;
 
         let database_client = DatabaseClient::new(
             self.addr.clone(),
@@ -105,19 +106,16 @@ impl ExportCommand {
             self.timeout.unwrap_or_default(),
         );
 
-        Ok(Instance::new(
-            Box::new(Export {
-                catalog,
-                schema,
-                database_client,
-                output_dir: self.output_dir.clone(),
-                parallelism: self.export_jobs,
-                target: self.target.clone(),
-                start_time: self.start_time.clone(),
-                end_time: self.end_time.clone(),
-            }),
-            guard,
-        ))
+        Ok(Box::new(Export {
+            catalog,
+            schema,
+            database_client,
+            output_dir: self.output_dir.clone(),
+            parallelism: self.export_jobs,
+            target: self.target.clone(),
+            start_time: self.start_time.clone(),
+            end_time: self.end_time.clone(),
+        }))
     }
 }
 
@@ -465,97 +463,22 @@ impl Export {
 
 #[async_trait]
 impl Tool for Export {
-    async fn do_work(&self) -> Result<()> {
+    async fn do_work(&self) -> std::result::Result<(), BoxedError> {
         match self.target {
             ExportTarget::Schema => {
-                self.export_create_database().await?;
-                self.export_create_table().await
+                self.export_create_database()
+                    .await
+                    .map_err(BoxedError::new)?;
+                self.export_create_table().await.map_err(BoxedError::new)
             }
-            ExportTarget::Data => self.export_database_data().await,
+            ExportTarget::Data => self.export_database_data().await.map_err(BoxedError::new),
             ExportTarget::All => {
-                self.export_create_database().await?;
-                self.export_create_table().await?;
-                self.export_database_data().await
+                self.export_create_database()
+                    .await
+                    .map_err(BoxedError::new)?;
+                self.export_create_table().await.map_err(BoxedError::new)?;
+                self.export_database_data().await.map_err(BoxedError::new)
             }
         }
     }
 }
-
-#[cfg(test)]
-mod tests {
-    use clap::Parser;
-    use client::{Client, Database};
-    use common_catalog::consts::{DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME};
-    use common_telemetry::logging::LoggingOptions;
-
-    use crate::error::Result as CmdResult;
-    use crate::options::GlobalOptions;
-    use crate::{cli, standalone, App};
-
-    #[tokio::test(flavor = "multi_thread")]
-    async fn test_export_create_table_with_quoted_names() -> CmdResult<()> {
-        let output_dir = tempfile::tempdir().unwrap();
-
-        let standalone = standalone::Command::parse_from([
-            "standalone",
-            "start",
-            "--data-home",
-            &*output_dir.path().to_string_lossy(),
-        ]);
-
-        let standalone_opts = standalone.load_options(&GlobalOptions::default()).unwrap();
-        let mut instance = standalone.build(standalone_opts).await?;
-        instance.start().await?;
-
-        let client = Client::with_urls(["127.0.0.1:4001"]);
-        let database = Database::new(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, client);
-        database
-            .sql(r#"CREATE DATABASE "cli.export.create_table";"#)
-            .await
-            .unwrap();
-        database
-            .sql(
-                r#"CREATE TABLE "cli.export.create_table"."a.b.c"(
-                        ts TIMESTAMP,
-                        TIME INDEX (ts)
-                    ) engine=mito;
-                "#,
-            )
-            .await
-            .unwrap();
-
-        let output_dir = tempfile::tempdir().unwrap();
-        let cli = cli::Command::parse_from([
-            "cli",
-            "export",
-            "--addr",
-            "127.0.0.1:4000",
-            "--output-dir",
-            &*output_dir.path().to_string_lossy(),
-            "--target",
-            "schema",
-        ]);
-        let mut cli_app = cli.build(LoggingOptions::default()).await?;
-        cli_app.start().await?;
-
-        instance.stop().await?;
-
-        let output_file = output_dir
-            .path()
-            .join("greptime")
-            .join("cli.export.create_table")
-            .join("create_tables.sql");
-        let res = std::fs::read_to_string(output_file).unwrap();
-        let expect = r#"CREATE TABLE IF NOT EXISTS "a.b.c" (
-  "ts" TIMESTAMP(3) NOT NULL,
-  TIME INDEX ("ts")
-)
-
-ENGINE=mito
-;
-"#;
-        assert_eq!(res.trim(), expect.trim());
-
-        Ok(())
-    }
-}

src/cli/src/helper.rs

@@ -19,7 +19,7 @@ use rustyline::highlight::{Highlighter, MatchingBracketHighlighter};
 use rustyline::hint::{Hinter, HistoryHinter};
 use rustyline::validate::{ValidationContext, ValidationResult, Validator};
 
-use crate::cli::cmd::ReplCommand;
+use crate::cmd::ReplCommand;
 
 pub(crate) struct RustylineHelper {
     hinter: HistoryHinter,

src/cli/src/import.rs

@@ -19,15 +19,15 @@ use std::time::Duration;
 use async_trait::async_trait;
 use clap::{Parser, ValueEnum};
 use common_catalog::consts::DEFAULT_SCHEMA_NAME;
+use common_error::ext::BoxedError;
 use common_telemetry::{error, info, warn};
 use snafu::{OptionExt, ResultExt};
 use tokio::sync::Semaphore;
 use tokio::time::Instant;
-use tracing_appender::non_blocking::WorkerGuard;
 
-use crate::cli::database::DatabaseClient;
-use crate::cli::{database, Instance, Tool};
+use crate::database::DatabaseClient;
 use crate::error::{Error, FileIoSnafu, Result, SchemaNotFoundSnafu};
+use crate::{database, Tool};
 
 #[derive(Debug, Default, Clone, ValueEnum)]
 enum ImportTarget {
@@ -79,8 +79,9 @@ pub struct ImportCommand {
 }
 
 impl ImportCommand {
-    pub async fn build(&self, guard: Vec<WorkerGuard>) -> Result<Instance> {
-        let (catalog, schema) = database::split_database(&self.database)?;
+    pub async fn build(&self) -> std::result::Result<Box<dyn Tool>, BoxedError> {
+        let (catalog, schema) =
+            database::split_database(&self.database).map_err(BoxedError::new)?;
         let database_client = DatabaseClient::new(
             self.addr.clone(),
             catalog.clone(),
@@ -89,17 +90,14 @@ impl ImportCommand {
             self.timeout.unwrap_or_default(),
         );
 
-        Ok(Instance::new(
-            Box::new(Import {
-                catalog,
-                schema,
-                database_client,
-                input_dir: self.input_dir.clone(),
-                parallelism: self.import_jobs,
-                target: self.target.clone(),
-            }),
-            guard,
-        ))
+        Ok(Box::new(Import {
+            catalog,
+            schema,
+            database_client,
+            input_dir: self.input_dir.clone(),
+            parallelism: self.import_jobs,
+            target: self.target.clone(),
+        }))
     }
 }
 
@@ -218,13 +216,13 @@ impl Import {
 
 #[async_trait]
 impl Tool for Import {
-    async fn do_work(&self) -> Result<()> {
+    async fn do_work(&self) -> std::result::Result<(), BoxedError> {
         match self.target {
-            ImportTarget::Schema => self.import_create_table().await,
-            ImportTarget::Data => self.import_database_data().await,
+            ImportTarget::Schema => self.import_create_table().await.map_err(BoxedError::new),
+            ImportTarget::Data => self.import_database_data().await.map_err(BoxedError::new),
             ImportTarget::All => {
-                self.import_create_table().await?;
-                self.import_database_data().await
+                self.import_create_table().await.map_err(BoxedError::new)?;
+                self.import_database_data().await.map_err(BoxedError::new)
             }
         }
     }

src/cli/src/lib.rs

@@ -0,0 +1,60 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+mod bench;
+pub mod error;
+// Wait for https://github.com/GreptimeTeam/greptimedb/issues/2373
+#[allow(unused)]
+mod cmd;
+mod export;
+mod helper;
+
+// Wait for https://github.com/GreptimeTeam/greptimedb/issues/2373
+mod database;
+mod import;
+#[allow(unused)]
+mod repl;
+
+use async_trait::async_trait;
+use clap::Parser;
+use common_error::ext::BoxedError;
+pub use database::DatabaseClient;
+use error::Result;
+pub use repl::Repl;
+
+pub use crate::bench::BenchTableMetadataCommand;
+pub use crate::export::ExportCommand;
+pub use crate::import::ImportCommand;
+
+#[async_trait]
+pub trait Tool: Send + Sync {
+    async fn do_work(&self) -> std::result::Result<(), BoxedError>;
+}
+
+#[derive(Debug, Parser)]
+pub(crate) struct AttachCommand {
+    #[clap(long)]
+    pub(crate) grpc_addr: String,
+    #[clap(long)]
+    pub(crate) meta_addr: Option<String>,
+    #[clap(long, action)]
+    pub(crate) disable_helper: bool,
+}
+
+impl AttachCommand {
+    #[allow(dead_code)]
+    async fn build(self) -> Result<Box<dyn Tool>> {
+        unimplemented!("Wait for https://github.com/GreptimeTeam/greptimedb/issues/2373")
+    }
+}

src/cli/src/repl.rs

@@ -20,19 +20,21 @@ use cache::{
     build_fundamental_cache_registry, with_default_composite_cache_registry, TABLE_CACHE_NAME,
     TABLE_ROUTE_CACHE_NAME,
 };
+use catalog::information_extension::DistributedInformationExtension;
 use catalog::kvbackend::{
-    CachedMetaKvBackend, CachedMetaKvBackendBuilder, KvBackendCatalogManager, MetaKvBackend,
+    CachedKvBackend, CachedKvBackendBuilder, KvBackendCatalogManager, MetaKvBackend,
 };
 use client::{Client, Database, OutputData, DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME};
 use common_base::Plugins;
 use common_config::Mode;
 use common_error::ext::ErrorExt;
 use common_meta::cache::{CacheRegistryBuilder, LayeredCacheRegistryBuilder};
+use common_meta::kv_backend::KvBackendRef;
 use common_query::Output;
 use common_recordbatch::RecordBatches;
 use common_telemetry::debug;
 use either::Either;
-use meta_client::client::MetaClientBuilder;
+use meta_client::client::{ClusterKvBackend, MetaClientBuilder};
 use query::datafusion::DatafusionQueryEngine;
 use query::parser::QueryLanguageParser;
 use query::query_engine::{DefaultSerializer, QueryEngineState};
@@ -43,15 +45,14 @@ use session::context::QueryContext;
 use snafu::{OptionExt, ResultExt};
 use substrait::{DFLogicalSubstraitConvertor, SubstraitPlan};
 
-use crate::cli::cmd::ReplCommand;
-use crate::cli::helper::RustylineHelper;
-use crate::cli::AttachCommand;
+use crate::cmd::ReplCommand;
 use crate::error::{
     CollectRecordBatchesSnafu, ParseSqlSnafu, PlanStatementSnafu, PrettyPrintRecordBatchesSnafu,
     ReadlineSnafu, ReplCreationSnafu, RequestDatabaseSnafu, Result, StartMetaClientSnafu,
     SubstraitEncodeLogicalPlanSnafu,
 };
-use crate::{error, DistributedInformationExtension};
+use crate::helper::RustylineHelper;
+use crate::{error, AttachCommand};
 
 /// Captures the state of the repl, gathers commands and executes them one by one
 pub struct Repl {
@@ -258,8 +259,9 @@ async fn create_query_engine(meta_addr: &str) -> Result<DatafusionQueryEngine> {
         .context(StartMetaClientSnafu)?;
     let meta_client = Arc::new(meta_client);
 
-    let cached_meta_backend =
-        Arc::new(CachedMetaKvBackendBuilder::new(meta_client.clone()).build());
+    let cached_meta_backend = Arc::new(
+        CachedKvBackendBuilder::new(Arc::new(MetaKvBackend::new(meta_client.clone()))).build(),
+    );
     let layered_cache_builder = LayeredCacheRegistryBuilder::default().add_cache_registry(
         CacheRegistryBuilder::default()
             .add_cache(cached_meta_backend.clone())

src/client/Cargo.toml

@@ -42,8 +42,6 @@ tonic.workspace = true
 
 [dev-dependencies]
 common-grpc-expr.workspace = true
-datanode.workspace = true
-derive-new = "0.5"
 tracing = "0.1"
 
 [dev-dependencies.substrait_proto]

src/cmd/Cargo.toml

@@ -25,6 +25,7 @@ cache.workspace = true
 catalog.workspace = true
 chrono.workspace = true
 clap.workspace = true
+cli.workspace = true
 client.workspace = true
 common-base.workspace = true
 common-catalog.workspace = true

src/cmd/src/cli.rs

@@ -12,39 +12,17 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-mod bench;
-
-// Wait for https://github.com/GreptimeTeam/greptimedb/issues/2373
-#[allow(unused)]
-mod cmd;
-mod export;
-mod helper;
-
-// Wait for https://github.com/GreptimeTeam/greptimedb/issues/2373
-mod database;
-mod import;
-#[allow(unused)]
-mod repl;
-
-use async_trait::async_trait;
-use bench::BenchTableMetadataCommand;
 use clap::Parser;
+use cli::Tool;
 use common_telemetry::logging::{LoggingOptions, TracingOptions};
-pub use repl::Repl;
+use plugins::SubCommand;
+use snafu::ResultExt;
 use tracing_appender::non_blocking::WorkerGuard;
 
-use self::export::ExportCommand;
-use crate::cli::import::ImportCommand;
-use crate::error::Result;
 use crate::options::GlobalOptions;
-use crate::App;
-
+use crate::{error, App, Result};
 pub const APP_NAME: &str = "greptime-cli";
-
-#[async_trait]
-pub trait Tool: Send + Sync {
-    async fn do_work(&self) -> Result<()>;
-}
+use async_trait::async_trait;
 
 pub struct Instance {
     tool: Box<dyn Tool>,
@@ -54,12 +32,16 @@ pub struct Instance {
 }
 
 impl Instance {
-    fn new(tool: Box<dyn Tool>, guard: Vec<WorkerGuard>) -> Self {
+    pub fn new(tool: Box<dyn Tool>, guard: Vec<WorkerGuard>) -> Self {
         Self {
             tool,
             _guard: guard,
         }
     }
+
+    pub async fn start(&mut self) -> Result<()> {
+        self.tool.do_work().await.context(error::StartCliSnafu)
+    }
 }
 
 #[async_trait]
@@ -69,7 +51,8 @@ impl App for Instance {
     }
 
     async fn start(&mut self) -> Result<()> {
-        self.tool.do_work().await
+        self.start().await.unwrap();
+        Ok(())
     }
 
     fn wait_signal(&self) -> bool {
@@ -96,7 +79,12 @@ impl Command {
             None,
         );
 
-        self.cmd.build(guard).await
+        let tool = self.cmd.build().await.context(error::BuildCliSnafu)?;
+        let instance = Instance {
+            tool,
+            _guard: guard,
+        };
+        Ok(instance)
     }
 
     pub fn load_options(&self, global_options: &GlobalOptions) -> Result<LoggingOptions> {
@@ -112,38 +100,81 @@ impl Command {
     }
 }
 
-#[derive(Parser)]
-enum SubCommand {
-    // Attach(AttachCommand),
-    Bench(BenchTableMetadataCommand),
-    Export(ExportCommand),
-    Import(ImportCommand),
-}
+#[cfg(test)]
+mod tests {
+    use clap::Parser;
+    use client::{Client, Database};
+    use common_catalog::consts::{DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME};
+    use common_telemetry::logging::LoggingOptions;
 
-impl SubCommand {
-    async fn build(&self, guard: Vec<WorkerGuard>) -> Result<Instance> {
-        match self {
-            // SubCommand::Attach(cmd) => cmd.build().await,
-            SubCommand::Bench(cmd) => cmd.build(guard).await,
-            SubCommand::Export(cmd) => cmd.build(guard).await,
-            SubCommand::Import(cmd) => cmd.build(guard).await,
-        }
-    }
-}
-
-#[derive(Debug, Parser)]
-pub(crate) struct AttachCommand {
-    #[clap(long)]
-    pub(crate) grpc_addr: String,
-    #[clap(long)]
-    pub(crate) meta_addr: Option<String>,
-    #[clap(long, action)]
-    pub(crate) disable_helper: bool,
-}
-
-impl AttachCommand {
-    #[allow(dead_code)]
-    async fn build(self) -> Result<Instance> {
-        unimplemented!("Wait for https://github.com/GreptimeTeam/greptimedb/issues/2373")
+    use crate::error::Result as CmdResult;
+    use crate::options::GlobalOptions;
+    use crate::{cli, standalone, App};
+
+    #[tokio::test(flavor = "multi_thread")]
+    async fn test_export_create_table_with_quoted_names() -> CmdResult<()> {
+        let output_dir = tempfile::tempdir().unwrap();
+
+        let standalone = standalone::Command::parse_from([
+            "standalone",
+            "start",
+            "--data-home",
+            &*output_dir.path().to_string_lossy(),
+        ]);
+
+        let standalone_opts = standalone.load_options(&GlobalOptions::default()).unwrap();
+        let mut instance = standalone.build(standalone_opts).await?;
+        instance.start().await?;
+
+        let client = Client::with_urls(["127.0.0.1:4001"]);
+        let database = Database::new(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, client);
+        database
+            .sql(r#"CREATE DATABASE "cli.export.create_table";"#)
+            .await
+            .unwrap();
+        database
+            .sql(
+                r#"CREATE TABLE "cli.export.create_table"."a.b.c"(
+                        ts TIMESTAMP,
+                        TIME INDEX (ts)
+                    ) engine=mito;
+                "#,
+            )
+            .await
+            .unwrap();
+
+        let output_dir = tempfile::tempdir().unwrap();
+        let cli = cli::Command::parse_from([
+            "cli",
+            "export",
+            "--addr",
+            "127.0.0.1:4000",
+            "--output-dir",
+            &*output_dir.path().to_string_lossy(),
+            "--target",
+            "schema",
+        ]);
+        let mut cli_app = cli.build(LoggingOptions::default()).await?;
+        cli_app.start().await?;
+
+        instance.stop().await?;
+
+        let output_file = output_dir
+            .path()
+            .join("greptime")
+            .join("cli.export.create_table")
+            .join("create_tables.sql");
+        let res = std::fs::read_to_string(output_file).unwrap();
+        let expect = r#"CREATE TABLE IF NOT EXISTS "a.b.c" (
+  "ts" TIMESTAMP(3) NOT NULL,
+  TIME INDEX ("ts")
+)
+
+ENGINE=mito
+;
+"#;
+        assert_eq!(res.trim(), expect.trim());
+
+        Ok(())
     }
 }

src/cmd/src/datanode.rs

@@ -16,10 +16,12 @@ use std::sync::Arc;
 use std::time::Duration;
 
 use async_trait::async_trait;
+use cache::build_datanode_cache_registry;
 use catalog::kvbackend::MetaKvBackend;
 use clap::Parser;
 use common_base::Plugins;
 use common_config::Configurable;
+use common_meta::cache::LayeredCacheRegistryBuilder;
 use common_telemetry::logging::TracingOptions;
 use common_telemetry::{info, warn};
 use common_version::{short_version, version};
@@ -57,13 +59,16 @@ impl Instance {
         }
     }
 
-    pub fn datanode_mut(&mut self) -> &mut Datanode {
-        &mut self.datanode
-    }
-
     pub fn datanode(&self) -> &Datanode {
         &self.datanode
     }
+
+    /// Get mutable Datanode instance for changing some internal state, before starting it.
+    // Useful for wrapping Datanode instance. Please do not remove this method even if you find
+    // nowhere it is called.
+    pub fn datanode_mut(&mut self) -> &mut Datanode {
+        &mut self.datanode
+    }
 }
 
 #[async_trait]
@@ -300,9 +305,17 @@ impl StartCommand {
             client: meta_client.clone(),
         });
 
+        // Builds cache registry for datanode.
+        let layered_cache_registry = Arc::new(
+            LayeredCacheRegistryBuilder::default()
+                .add_cache_registry(build_datanode_cache_registry(meta_backend.clone()))
+                .build(),
+        );
+
         let mut datanode = DatanodeBuilder::new(opts.clone(), plugins)
             .with_meta_client(meta_client)
             .with_kv_backend(meta_backend)
+            .with_cache_registry(layered_cache_registry)
             .build()
             .await
             .context(StartDatanodeSnafu)?;

src/cmd/src/error.rs

@@ -114,6 +114,20 @@ pub enum Error {
         source: frontend::error::Error,
     },
 
+    #[snafu(display("Failed to build cli"))]
+    BuildCli {
+        #[snafu(implicit)]
+        location: Location,
+        source: BoxedError,
+    },
+
+    #[snafu(display("Failed to start cli"))]
+    StartCli {
+        #[snafu(implicit)]
+        location: Location,
+        source: BoxedError,
+    },
+
     #[snafu(display("Failed to build meta server"))]
     BuildMetaServer {
         #[snafu(implicit)]
@@ -346,6 +360,8 @@ impl ErrorExt for Error {
             Error::ShutdownMetaServer { source, .. } => source.status_code(),
             Error::BuildMetaServer { source, .. } => source.status_code(),
             Error::UnsupportedSelectorType { source, .. } => source.status_code(),
+            Error::BuildCli { source, .. } => source.status_code(),
+            Error::StartCli { source, .. } => source.status_code(),
 
             Error::InitMetadata { source, .. } | Error::InitDdlManager { source, .. } => {
                 source.status_code()

src/cmd/src/flownode.rs

@@ -15,13 +15,15 @@
 use std::sync::Arc;
 
 use cache::{build_fundamental_cache_registry, with_default_composite_cache_registry};
-use catalog::kvbackend::{CachedMetaKvBackendBuilder, KvBackendCatalogManager, MetaKvBackend};
+use catalog::information_extension::DistributedInformationExtension;
+use catalog::kvbackend::{CachedKvBackendBuilder, KvBackendCatalogManager, MetaKvBackend};
 use clap::Parser;
 use client::client_manager::NodeClients;
 use common_base::Plugins;
 use common_config::Configurable;
 use common_grpc::channel_manager::ChannelConfig;
 use common_meta::cache::{CacheRegistryBuilder, LayeredCacheRegistryBuilder};
+use common_meta::heartbeat::handler::invalidate_table_cache::InvalidateCacheHandler;
 use common_meta::heartbeat::handler::parse_mailbox_message::ParseMailboxMessageHandler;
 use common_meta::heartbeat::handler::HandlerGroupExecutor;
 use common_meta::key::flow::FlowMetadataManager;
@@ -30,7 +32,6 @@ use common_telemetry::info;
 use common_telemetry::logging::TracingOptions;
 use common_version::{short_version, version};
 use flow::{FlownodeBuilder, FlownodeInstance, FrontendInvoker};
-use frontend::heartbeat::handler::invalidate_table_cache::InvalidateTableCacheHandler;
 use meta_client::{MetaClientOptions, MetaClientType};
 use servers::Mode;
 use snafu::{OptionExt, ResultExt};
@@ -41,7 +42,7 @@ use crate::error::{
     MissingConfigSnafu, Result, ShutdownFlownodeSnafu, StartFlownodeSnafu,
 };
 use crate::options::{GlobalOptions, GreptimeOptions};
-use crate::{log_versions, App, DistributedInformationExtension};
+use crate::{log_versions, App};
 
 pub const APP_NAME: &str = "greptime-flownode";
 
@@ -62,10 +63,6 @@ impl Instance {
         }
     }
 
-    pub fn flownode_mut(&mut self) -> &mut FlownodeInstance {
-        &mut self.flownode
-    }
-
     pub fn flownode(&self) -> &FlownodeInstance {
         &self.flownode
     }
@@ -246,11 +243,12 @@ impl StartCommand {
         let cache_tti = meta_config.metadata_cache_tti;
 
         // TODO(discord9): add helper function to ease the creation of cache registry&such
-        let cached_meta_backend = CachedMetaKvBackendBuilder::new(meta_client.clone())
-            .cache_max_capacity(cache_max_capacity)
-            .cache_ttl(cache_ttl)
-            .cache_tti(cache_tti)
-            .build();
+        let cached_meta_backend =
+            CachedKvBackendBuilder::new(Arc::new(MetaKvBackend::new(meta_client.clone())))
+                .cache_max_capacity(cache_max_capacity)
+                .cache_ttl(cache_ttl)
+                .cache_tti(cache_tti)
+                .build();
         let cached_meta_backend = Arc::new(cached_meta_backend);
 
         // Builds cache registry
@@ -287,9 +285,7 @@ impl StartCommand {
 
         let executor = HandlerGroupExecutor::new(vec![
             Arc::new(ParseMailboxMessageHandler),
-            Arc::new(InvalidateTableCacheHandler::new(
-                layered_cache_registry.clone(),
-            )),
+            Arc::new(InvalidateCacheHandler::new(layered_cache_registry.clone())),
         ]);
 
         let heartbeat_task = flow::heartbeat::HeartbeatTask::new(

src/cmd/src/frontend.rs

@@ -17,20 +17,21 @@ use std::time::Duration;
 
 use async_trait::async_trait;
 use cache::{build_fundamental_cache_registry, with_default_composite_cache_registry};
-use catalog::kvbackend::{CachedMetaKvBackendBuilder, KvBackendCatalogManager, MetaKvBackend};
+use catalog::information_extension::DistributedInformationExtension;
+use catalog::kvbackend::{CachedKvBackendBuilder, KvBackendCatalogManager, MetaKvBackend};
 use clap::Parser;
 use client::client_manager::NodeClients;
 use common_base::Plugins;
 use common_config::Configurable;
 use common_grpc::channel_manager::ChannelConfig;
 use common_meta::cache::{CacheRegistryBuilder, LayeredCacheRegistryBuilder};
+use common_meta::heartbeat::handler::invalidate_table_cache::InvalidateCacheHandler;
 use common_meta::heartbeat::handler::parse_mailbox_message::ParseMailboxMessageHandler;
 use common_meta::heartbeat::handler::HandlerGroupExecutor;
 use common_telemetry::info;
 use common_telemetry::logging::TracingOptions;
 use common_time::timezone::set_default_timezone;
 use common_version::{short_version, version};
-use frontend::heartbeat::handler::invalidate_table_cache::InvalidateTableCacheHandler;
 use frontend::heartbeat::HeartbeatTask;
 use frontend::instance::builder::FrontendBuilder;
 use frontend::instance::{FrontendInstance, Instance as FeInstance};
@@ -46,7 +47,7 @@ use crate::error::{
     Result, StartFrontendSnafu,
 };
 use crate::options::{GlobalOptions, GreptimeOptions};
-use crate::{log_versions, App, DistributedInformationExtension};
+use crate::{log_versions, App};
 
 type FrontendOptions = GreptimeOptions<frontend::frontend::FrontendOptions>;
 
@@ -293,11 +294,12 @@ impl StartCommand {
         .context(MetaClientInitSnafu)?;
 
         // TODO(discord9): add helper function to ease the creation of cache registry&such
-        let cached_meta_backend = CachedMetaKvBackendBuilder::new(meta_client.clone())
-            .cache_max_capacity(cache_max_capacity)
-            .cache_ttl(cache_ttl)
-            .cache_tti(cache_tti)
-            .build();
+        let cached_meta_backend =
+            CachedKvBackendBuilder::new(Arc::new(MetaKvBackend::new(meta_client.clone())))
+                .cache_max_capacity(cache_max_capacity)
+                .cache_ttl(cache_ttl)
+                .cache_tti(cache_tti)
+                .build();
         let cached_meta_backend = Arc::new(cached_meta_backend);
 
         // Builds cache registry
@@ -327,9 +329,7 @@ impl StartCommand {
 
         let executor = HandlerGroupExecutor::new(vec![
             Arc::new(ParseMailboxMessageHandler),
-            Arc::new(InvalidateTableCacheHandler::new(
-                layered_cache_registry.clone(),
-            )),
+            Arc::new(InvalidateCacheHandler::new(layered_cache_registry.clone())),
         ]);
 
         let heartbeat_task = HeartbeatTask::new(

src/cmd/src/lib.rs

@@ -15,17 +15,7 @@
 #![feature(assert_matches, let_chains)]
 
 use async_trait::async_trait;
-use catalog::information_schema::InformationExtension;
-use client::api::v1::meta::ProcedureStatus;
-use common_error::ext::BoxedError;
-use common_meta::cluster::{ClusterInfo, NodeInfo};
-use common_meta::datanode::RegionStat;
-use common_meta::ddl::{ExecutorContext, ProcedureExecutor};
-use common_meta::rpc::procedure;
-use common_procedure::{ProcedureInfo, ProcedureState};
 use common_telemetry::{error, info};
-use meta_client::MetaClientRef;
-use snafu::ResultExt;
 
 use crate::error::Result;
 
@@ -130,69 +120,3 @@ fn log_env_flags() {
         info!("argument: {}", argument);
     }
 }
-
-pub struct DistributedInformationExtension {
-    meta_client: MetaClientRef,
-}
-
-impl DistributedInformationExtension {
-    pub fn new(meta_client: MetaClientRef) -> Self {
-        Self { meta_client }
-    }
-}
-
-#[async_trait::async_trait]
-impl InformationExtension for DistributedInformationExtension {
-    type Error = catalog::error::Error;
-
-    async fn nodes(&self) -> std::result::Result<Vec<NodeInfo>, Self::Error> {
-        self.meta_client
-            .list_nodes(None)
-            .await
-            .map_err(BoxedError::new)
-            .context(catalog::error::ListNodesSnafu)
-    }
-
-    async fn procedures(&self) -> std::result::Result<Vec<(String, ProcedureInfo)>, Self::Error> {
-        let procedures = self
-            .meta_client
-            .list_procedures(&ExecutorContext::default())
-            .await
-            .map_err(BoxedError::new)
-            .context(catalog::error::ListProceduresSnafu)?
-            .procedures;
-        let mut result = Vec::with_capacity(procedures.len());
-        for procedure in procedures {
-            let pid = match procedure.id {
-                Some(pid) => pid,
-                None => return catalog::error::ProcedureIdNotFoundSnafu {}.fail(),
-            };
-            let pid = procedure::pb_pid_to_pid(&pid)
-                .map_err(BoxedError::new)
-                .context(catalog::error::ConvertProtoDataSnafu)?;
-            let status = ProcedureStatus::try_from(procedure.status)
-                .map(|v| v.as_str_name())
-                .unwrap_or("Unknown")
-                .to_string();
-            let procedure_info = ProcedureInfo {
-                id: pid,
-                type_name: procedure.type_name,
-                start_time_ms: procedure.start_time_ms,
-                end_time_ms: procedure.end_time_ms,
-                state: ProcedureState::Running,
-                lock_keys: procedure.lock_keys,
-            };
-            result.push((status, procedure_info));
-        }
-
-        Ok(result)
-    }
-
-    async fn region_stats(&self) -> std::result::Result<Vec<RegionStat>, Self::Error> {
-        self.meta_client
-            .list_region_stats()
-            .await
-            .map_err(BoxedError::new)
-            .context(catalog::error::ListRegionStatsSnafu)
-    }
-}

src/cmd/src/standalone.rs

@@ -22,6 +22,7 @@ use catalog::information_schema::InformationExtension;
 use catalog::kvbackend::KvBackendCatalogManager;
 use clap::Parser;
 use client::api::v1::meta::RegionRole;
+use common_base::readable_size::ReadableSize;
 use common_base::Plugins;
 use common_catalog::consts::{MIN_USER_FLOW_ID, MIN_USER_TABLE_ID};
 use common_config::{metadata_store_dir, Configurable, KvBackendConfig};
@@ -34,6 +35,7 @@ use common_meta::ddl::flow_meta::{FlowMetadataAllocator, FlowMetadataAllocatorRe
 use common_meta::ddl::table_meta::{TableMetadataAllocator, TableMetadataAllocatorRef};
 use common_meta::ddl::{DdlContext, NoopRegionFailureDetectorControl, ProcedureExecutorRef};
 use common_meta::ddl_manager::DdlManager;
+use common_meta::key::flow::flow_state::FlowStat;
 use common_meta::key::flow::{FlowMetadataManager, FlowMetadataManagerRef};
 use common_meta::key::{TableMetadataManager, TableMetadataManagerRef};
 use common_meta::kv_backend::KvBackendRef;
@@ -70,7 +72,7 @@ use servers::http::HttpOptions;
 use servers::tls::{TlsMode, TlsOption};
 use servers::Mode;
 use snafu::ResultExt;
-use tokio::sync::broadcast;
+use tokio::sync::{broadcast, RwLock};
 use tracing_appender::non_blocking::WorkerGuard;
 
 use crate::error::{
@@ -151,6 +153,7 @@ pub struct StandaloneOptions {
     pub tracing: TracingOptions,
     pub init_regions_in_background: bool,
     pub init_regions_parallelism: usize,
+    pub max_in_flight_write_bytes: Option<ReadableSize>,
 }
 
 impl Default for StandaloneOptions {
@@ -180,6 +183,7 @@ impl Default for StandaloneOptions {
             tracing: TracingOptions::default(),
             init_regions_in_background: false,
             init_regions_parallelism: 16,
+            max_in_flight_write_bytes: None,
         }
     }
 }
@@ -217,6 +221,7 @@ impl StandaloneOptions {
             user_provider: cloned_opts.user_provider,
             // Handle the export metrics task run by standalone to frontend for execution
             export_metrics: cloned_opts.export_metrics,
+            max_in_flight_write_bytes: cloned_opts.max_in_flight_write_bytes,
             ..Default::default()
         }
     }
@@ -497,6 +502,7 @@ impl StartCommand {
 
         let datanode = DatanodeBuilder::new(dn_opts, plugins.clone())
             .with_kv_backend(kv_backend.clone())
+            .with_cache_registry(layered_cache_registry.clone())
             .build()
             .await
             .context(StartDatanodeSnafu)?;
@@ -506,7 +512,7 @@ impl StartCommand {
             procedure_manager.clone(),
         ));
         let catalog_manager = KvBackendCatalogManager::new(
-            information_extension,
+            information_extension.clone(),
             kv_backend.clone(),
             layered_cache_registry.clone(),
             Some(procedure_manager.clone()),
@@ -531,6 +537,14 @@ impl StartCommand {
                 .context(OtherSnafu)?,
         );
 
+        // set the ref to query for the local flow state
+        {
+            let flow_worker_manager = flownode.flow_worker_manager();
+            information_extension
+                .set_flow_worker_manager(flow_worker_manager.clone())
+                .await;
+        }
+
         let node_manager = Arc::new(StandaloneDatanodeManager {
             region_server: datanode.region_server(),
             flow_server: flownode.flow_worker_manager(),
@@ -668,6 +682,7 @@ pub struct StandaloneInformationExtension {
     region_server: RegionServer,
     procedure_manager: ProcedureManagerRef,
     start_time_ms: u64,
+    flow_worker_manager: RwLock<Option<Arc<FlowWorkerManager>>>,
 }
 
 impl StandaloneInformationExtension {
@@ -676,8 +691,15 @@ impl StandaloneInformationExtension {
             region_server,
             procedure_manager,
             start_time_ms: common_time::util::current_time_millis() as u64,
+            flow_worker_manager: RwLock::new(None),
         }
     }
+
+    /// Set the flow worker manager for the standalone instance.
+    pub async fn set_flow_worker_manager(&self, flow_worker_manager: Arc<FlowWorkerManager>) {
+        let mut guard = self.flow_worker_manager.write().await;
+        *guard = Some(flow_worker_manager);
+    }
 }
 
 #[async_trait::async_trait]
@@ -749,6 +771,18 @@ impl InformationExtension for StandaloneInformationExtension {
             .collect::<Vec<_>>();
         Ok(stats)
     }
+
+    async fn flow_stats(&self) -> std::result::Result<Option<FlowStat>, Self::Error> {
+        Ok(Some(
+            self.flow_worker_manager
+                .read()
+                .await
+                .as_ref()
+                .unwrap()
+                .gen_state_report()
+                .await,
+        ))
+    }
 }
 
 #[cfg(test)]

src/cmd/tests/load_config_test.rs

@@ -69,7 +69,6 @@ fn test_load_datanode_example_config() {
             region_engine: vec![
                 RegionEngineConfig::Mito(MitoConfig {
                     auto_flush_interval: Duration::from_secs(3600),
-                    scan_parallelism: 0,
                     experimental_write_cache_ttl: Some(Duration::from_secs(60 * 60 * 8)),
                     ..Default::default()
                 }),
@@ -205,7 +204,6 @@ fn test_load_standalone_example_config() {
                 RegionEngineConfig::Mito(MitoConfig {
                     auto_flush_interval: Duration::from_secs(3600),
                     experimental_write_cache_ttl: Some(Duration::from_secs(60 * 60 * 8)),
-                    scan_parallelism: 0,
                     ..Default::default()
                 }),
                 RegionEngineConfig::File(EngineConfig {}),

src/common/base/Cargo.toml

@@ -17,6 +17,7 @@ common-macro.workspace = true
 futures.workspace = true
 paste = "1.0"
 pin-project.workspace = true
+rand.workspace = true
 serde = { version = "1.0", features = ["derive"] }
 snafu.workspace = true
 tokio.workspace = true

src/common/base/src/range_read.rs

@@ -36,6 +36,11 @@ pub struct Metadata {
 /// `RangeReader` reads a range of bytes from a source.
 #[async_trait]
 pub trait RangeReader: Send + Unpin {
+    /// Sets the file size hint for the reader.
+    ///
+    /// It's used to optimize the reading process by reducing the number of remote requests.
+    fn with_file_size_hint(&mut self, file_size_hint: u64);
+
     /// Returns the metadata of the source.
     async fn metadata(&mut self) -> io::Result<Metadata>;
 
@@ -70,6 +75,10 @@ pub trait RangeReader: Send + Unpin {
 
 #[async_trait]
 impl<R: ?Sized + RangeReader> RangeReader for &mut R {
+    fn with_file_size_hint(&mut self, file_size_hint: u64) {
+        (*self).with_file_size_hint(file_size_hint)
+    }
+
     async fn metadata(&mut self) -> io::Result<Metadata> {
         (*self).metadata().await
     }
@@ -186,15 +195,17 @@ impl<R: RangeReader + 'static> AsyncRead for AsyncReadAdapter<R> {
 
 #[async_trait]
 impl RangeReader for Vec<u8> {
+    fn with_file_size_hint(&mut self, _file_size_hint: u64) {
+        // do nothing
+    }
+
     async fn metadata(&mut self) -> io::Result<Metadata> {
         Ok(Metadata {
             content_length: self.len() as u64,
         })
     }
 
-    async fn read(&mut self, mut range: Range<u64>) -> io::Result<Bytes> {
-        range.end = range.end.min(self.len() as u64);
-
+    async fn read(&mut self, range: Range<u64>) -> io::Result<Bytes> {
         let bytes = Bytes::copy_from_slice(&self[range.start as usize..range.end as usize]);
         Ok(bytes)
     }
@@ -222,6 +233,10 @@ impl FileReader {
 
 #[async_trait]
 impl RangeReader for FileReader {
+    fn with_file_size_hint(&mut self, _file_size_hint: u64) {
+        // do nothing
+    }
+
     async fn metadata(&mut self) -> io::Result<Metadata> {
         Ok(Metadata {
             content_length: self.content_length,

src/common/base/src/readable_size.rs

@@ -19,7 +19,7 @@ pub const GIB: u64 = MIB * BINARY_DATA_MAGNITUDE;
 pub const TIB: u64 = GIB * BINARY_DATA_MAGNITUDE;
 pub const PIB: u64 = TIB * BINARY_DATA_MAGNITUDE;
 
-#[derive(Clone, Copy, PartialEq, Eq, Ord, PartialOrd)]
+#[derive(Clone, Copy, PartialEq, Eq, Ord, PartialOrd, Default)]
 pub struct ReadableSize(pub u64);
 
 impl ReadableSize {

src/common/catalog/Cargo.toml

@@ -8,10 +8,5 @@ license.workspace = true
 workspace = true
 
 [dependencies]
-common-error.workspace = true
-common-macro.workspace = true
-snafu.workspace = true
 
 [dev-dependencies]
-chrono.workspace = true
-tokio.workspace = true

src/common/datasource/Cargo.toml

@@ -48,5 +48,4 @@ url = "2.3"
 [dev-dependencies]
 common-telemetry.workspace = true
 common-test-util.workspace = true
-dotenv.workspace = true
 uuid.workspace = true

src/common/datasource/src/object_store/fs.rs

@@ -27,7 +27,7 @@ pub fn build_fs_backend(root: &str) -> Result<ObjectStore> {
             DefaultLoggingInterceptor,
         ))
         .layer(object_store::layers::TracingLayer)
-        .layer(object_store::layers::PrometheusMetricsLayer::new(true))
+        .layer(object_store::layers::build_prometheus_metrics_layer(true))
         .finish();
     Ok(object_store)
 }

src/common/datasource/src/object_store/s3.rs

@@ -89,7 +89,7 @@ pub fn build_s3_backend(
             DefaultLoggingInterceptor,
         ))
         .layer(object_store::layers::TracingLayer)
-        .layer(object_store::layers::PrometheusMetricsLayer::new(true))
+        .layer(object_store::layers::build_prometheus_metrics_layer(true))
         .finish())
 }
 

src/common/error/Cargo.toml

@@ -8,6 +8,7 @@ license.workspace = true
 workspace = true
 
 [dependencies]
+http.workspace = true
 snafu.workspace = true
 strum.workspace = true
 tonic.workspace = true

src/common/error/src/lib.rs

@@ -18,9 +18,30 @@ pub mod ext;
 pub mod mock;
 pub mod status_code;
 
+use http::{HeaderMap, HeaderValue};
 pub use snafu;
 
 // HACK - these headers are here for shared in gRPC services. For common HTTP headers,
 // please define in `src/servers/src/http/header.rs`.
 pub const GREPTIME_DB_HEADER_ERROR_CODE: &str = "x-greptime-err-code";
 pub const GREPTIME_DB_HEADER_ERROR_MSG: &str = "x-greptime-err-msg";
+
+/// Create a http header map from error code and message.
+/// using `GREPTIME_DB_HEADER_ERROR_CODE` and `GREPTIME_DB_HEADER_ERROR_MSG` as keys.
+pub fn from_err_code_msg_to_header(code: u32, msg: &str) -> HeaderMap {
+    let mut header = HeaderMap::new();
+
+    let msg = HeaderValue::from_str(msg).unwrap_or_else(|_| {
+        HeaderValue::from_bytes(
+            &msg.as_bytes()
+                .iter()
+                .flat_map(|b| std::ascii::escape_default(*b))
+                .collect::<Vec<u8>>(),
+        )
+        .expect("Already escaped string should be valid ascii")
+    });
+
+    header.insert(GREPTIME_DB_HEADER_ERROR_CODE, code.into());
+    header.insert(GREPTIME_DB_HEADER_ERROR_MSG, msg);
+    header
+}

src/common/frontend/Cargo.toml

@@ -5,12 +5,7 @@ edition.workspace = true
 license.workspace = true
 
 [dependencies]
-api.workspace = true
 async-trait.workspace = true
-common-base.workspace = true
 common-error.workspace = true
 common-macro.workspace = true
-common-query.workspace = true
-session.workspace = true
 snafu.workspace = true
-sql.workspace = true

src/common/function/Cargo.toml

@@ -33,7 +33,7 @@ geo-types = { version = "0.7", optional = true }
 geohash = { version = "0.13", optional = true }
 h3o = { version = "0.6", optional = true }
 jsonb.workspace = true
-nalgebra = "0.33"
+nalgebra.workspace = true
 num = "0.4"
 num-traits = "0.2"
 once_cell.workspace = true
@@ -51,6 +51,5 @@ wkt = { version = "0.11", optional = true }
 
 [dev-dependencies]
 approx = "0.5"
-ron = "0.7"
 serde = { version = "1.0", features = ["derive"] }
 tokio.workspace = true

src/common/function/src/lib.rs

@@ -26,3 +26,4 @@ pub mod function_registry;
 pub mod handlers;
 pub mod helper;
 pub mod state;
+pub mod utils;

src/common/function/src/scalars/aggregate.rs

@@ -32,6 +32,7 @@ pub use scipy_stats_norm_cdf::ScipyStatsNormCdfAccumulatorCreator;
 pub use scipy_stats_norm_pdf::ScipyStatsNormPdfAccumulatorCreator;
 
 use crate::function_registry::FunctionRegistry;
+use crate::scalars::vector::sum::VectorSumCreator;
 
 /// A function creates `AggregateFunctionCreator`.
 /// "Aggregator" *is* AggregatorFunction. Since the later one is long, we named an short alias for it.
@@ -91,6 +92,7 @@ impl AggregateFunctions {
         register_aggr_func!("argmin", 1, ArgminAccumulatorCreator);
         register_aggr_func!("scipystatsnormcdf", 2, ScipyStatsNormCdfAccumulatorCreator);
         register_aggr_func!("scipystatsnormpdf", 2, ScipyStatsNormPdfAccumulatorCreator);
+        register_aggr_func!("vec_sum", 1, VectorSumCreator);
 
         #[cfg(feature = "geo")]
         register_aggr_func!(

src/common/function/src/scalars/matches.rs

@@ -204,20 +204,10 @@ impl PatternAst {
     fn convert_literal(column: &str, pattern: &str) -> Expr {
         logical_expr::col(column).like(logical_expr::lit(format!(
             "%{}%",
-            Self::escape_pattern(pattern)
+            crate::utils::escape_like_pattern(pattern)
         )))
     }
 
-    fn escape_pattern(pattern: &str) -> String {
-        pattern
-            .chars()
-            .flat_map(|c| match c {
-                '\\' | '%' | '_' => vec!['\\', c],
-                _ => vec![c],
-            })
-            .collect::<String>()
-    }
-
     /// Transform this AST with preset rules to make it correct.
     fn transform_ast(self) -> Result<Self> {
         self.transform_up(Self::collapse_binary_branch_fn)

src/common/function/src/scalars/vector.rs

@@ -14,6 +14,14 @@
 
 mod convert;
 mod distance;
+mod elem_sum;
+pub mod impl_conv;
+mod scalar_add;
+mod scalar_mul;
+mod sub;
+pub(crate) mod sum;
+mod vector_div;
+mod vector_mul;
 
 use std::sync::Arc;
 
@@ -31,5 +39,15 @@ impl VectorFunction {
         registry.register(Arc::new(distance::CosDistanceFunction));
         registry.register(Arc::new(distance::DotProductFunction));
         registry.register(Arc::new(distance::L2SqDistanceFunction));
+
+        // scalar calculation
+        registry.register(Arc::new(scalar_add::ScalarAddFunction));
+        registry.register(Arc::new(scalar_mul::ScalarMulFunction));
+
+        // vector calculation
+        registry.register(Arc::new(vector_mul::VectorMulFunction));
+        registry.register(Arc::new(vector_div::VectorDivFunction));
+        registry.register(Arc::new(sub::SubFunction));
+        registry.register(Arc::new(elem_sum::ElemSumFunction));
     }
 }

src/common/function/src/scalars/vector/distance.rs

@@ -18,18 +18,17 @@ mod l2sq;
 
 use std::borrow::Cow;
 use std::fmt::Display;
-use std::sync::Arc;
 
 use common_query::error::{InvalidFuncArgsSnafu, Result};
 use common_query::prelude::Signature;
 use datatypes::prelude::ConcreteDataType;
 use datatypes::scalars::ScalarVectorBuilder;
-use datatypes::value::ValueRef;
-use datatypes::vectors::{Float32VectorBuilder, MutableVector, Vector, VectorRef};
+use datatypes::vectors::{Float32VectorBuilder, MutableVector, VectorRef};
 use snafu::ensure;
 
 use crate::function::{Function, FunctionContext};
 use crate::helper;
+use crate::scalars::vector::impl_conv::{as_veclit, as_veclit_if_const};
 
 macro_rules! define_distance_function {
     ($StructName:ident, $display_name:expr, $similarity_method:path) => {
@@ -80,17 +79,17 @@ macro_rules! define_distance_function {
                     return Ok(result.to_vector());
                 }
 
-                let arg0_const = parse_if_constant_string(arg0)?;
-                let arg1_const = parse_if_constant_string(arg1)?;
+                let arg0_const = as_veclit_if_const(arg0)?;
+                let arg1_const = as_veclit_if_const(arg1)?;
 
                 for i in 0..size {
                     let vec0 = match arg0_const.as_ref() {
-                        Some(a) => Some(Cow::Borrowed(a.as_slice())),
-                        None => as_vector(arg0.get_ref(i))?,
+                        Some(a) => Some(Cow::Borrowed(a.as_ref())),
+                        None => as_veclit(arg0.get_ref(i))?,
                     };
                     let vec1 = match arg1_const.as_ref() {
-                        Some(b) => Some(Cow::Borrowed(b.as_slice())),
-                        None => as_vector(arg1.get_ref(i))?,
+                        Some(b) => Some(Cow::Borrowed(b.as_ref())),
+                        None => as_veclit(arg1.get_ref(i))?,
                     };
 
                     if let (Some(vec0), Some(vec1)) = (vec0, vec1) {
@@ -129,98 +128,6 @@ define_distance_function!(CosDistanceFunction, "vec_cos_distance", cos::cos);
 define_distance_function!(L2SqDistanceFunction, "vec_l2sq_distance", l2sq::l2sq);
 define_distance_function!(DotProductFunction, "vec_dot_product", dot::dot);
 
-/// Parse a vector value if the value is a constant string.
-fn parse_if_constant_string(arg: &Arc<dyn Vector>) -> Result<Option<Vec<f32>>> {
-    if !arg.is_const() {
-        return Ok(None);
-    }
-    if arg.data_type() != ConcreteDataType::string_datatype() {
-        return Ok(None);
-    }
-    arg.get_ref(0)
-        .as_string()
-        .unwrap() // Safe: checked if it is a string
-        .map(parse_f32_vector_from_string)
-        .transpose()
-}
-
-/// Convert a value to a vector value.
-/// Supported data types are binary and string.
-fn as_vector(arg: ValueRef<'_>) -> Result<Option<Cow<'_, [f32]>>> {
-    match arg.data_type() {
-        ConcreteDataType::Binary(_) => arg
-            .as_binary()
-            .unwrap() // Safe: checked if it is a binary
-            .map(binary_as_vector)
-            .transpose(),
-        ConcreteDataType::String(_) => arg
-            .as_string()
-            .unwrap() // Safe: checked if it is a string
-            .map(|s| Ok(Cow::Owned(parse_f32_vector_from_string(s)?)))
-            .transpose(),
-        ConcreteDataType::Null(_) => Ok(None),
-        _ => InvalidFuncArgsSnafu {
-            err_msg: format!("Unsupported data type: {:?}", arg.data_type()),
-        }
-        .fail(),
-    }
-}
-
-/// Convert a u8 slice to a vector value.
-fn binary_as_vector(bytes: &[u8]) -> Result<Cow<'_, [f32]>> {
-    if bytes.len() % std::mem::size_of::<f32>() != 0 {
-        return InvalidFuncArgsSnafu {
-            err_msg: format!("Invalid binary length of vector: {}", bytes.len()),
-        }
-        .fail();
-    }
-
-    if cfg!(target_endian = "little") {
-        Ok(unsafe {
-            let vec = std::slice::from_raw_parts(
-                bytes.as_ptr() as *const f32,
-                bytes.len() / std::mem::size_of::<f32>(),
-            );
-            Cow::Borrowed(vec)
-        })
-    } else {
-        let v = bytes
-            .chunks_exact(std::mem::size_of::<f32>())
-            .map(|chunk| f32::from_le_bytes(chunk.try_into().unwrap()))
-            .collect::<Vec<f32>>();
-        Ok(Cow::Owned(v))
-    }
-}
-
-/// Parse a string to a vector value.
-/// Valid inputs are strings like "[1.0, 2.0, 3.0]".
-fn parse_f32_vector_from_string(s: &str) -> Result<Vec<f32>> {
-    let trimmed = s.trim();
-    if !trimmed.starts_with('[') || !trimmed.ends_with(']') {
-        return InvalidFuncArgsSnafu {
-            err_msg: format!(
-                "Failed to parse {s} to Vector value: not properly enclosed in brackets"
-            ),
-        }
-        .fail();
-    }
-    let content = trimmed[1..trimmed.len() - 1].trim();
-    if content.is_empty() {
-        return Ok(Vec::new());
-    }
-
-    content
-        .split(',')
-        .map(|s| s.trim().parse::<f32>())
-        .collect::<std::result::Result<_, _>>()
-        .map_err(|e| {
-            InvalidFuncArgsSnafu {
-                err_msg: format!("Failed to parse {s} to Vector value: {e}"),
-            }
-            .build()
-        })
-}
-
 #[cfg(test)]
 mod tests {
     use std::sync::Arc;
@@ -456,27 +363,4 @@ mod tests {
             assert!(result.is_err());
         }
     }
-
-    #[test]
-    fn test_parse_vector_from_string() {
-        let result = parse_f32_vector_from_string("[1.0, 2.0, 3.0]").unwrap();
-        assert_eq!(result, vec![1.0, 2.0, 3.0]);
-
-        let result = parse_f32_vector_from_string("[]").unwrap();
-        assert_eq!(result, Vec::<f32>::new());
-
-        let result = parse_f32_vector_from_string("[1.0, a, 3.0]");
-        assert!(result.is_err());
-    }
-
-    #[test]
-    fn test_binary_as_vector() {
-        let bytes = [0, 0, 128, 63];
-        let result = binary_as_vector(&bytes).unwrap();
-        assert_eq!(result.as_ref(), &[1.0]);
-
-        let invalid_bytes = [0, 0, 128];
-        let result = binary_as_vector(&invalid_bytes);
-        assert!(result.is_err());
-    }
 }

src/common/function/src/scalars/vector/elem_sum.rs

@@ -0,0 +1,129 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use std::borrow::Cow;
+use std::fmt::Display;
+
+use common_query::error::InvalidFuncArgsSnafu;
+use common_query::prelude::{Signature, TypeSignature, Volatility};
+use datatypes::prelude::ConcreteDataType;
+use datatypes::scalars::ScalarVectorBuilder;
+use datatypes::vectors::{Float32VectorBuilder, MutableVector, VectorRef};
+use nalgebra::DVectorView;
+use snafu::ensure;
+
+use crate::function::{Function, FunctionContext};
+use crate::scalars::vector::impl_conv::{as_veclit, as_veclit_if_const};
+
+const NAME: &str = "vec_elem_sum";
+
+#[derive(Debug, Clone, Default)]
+pub struct ElemSumFunction;
+
+impl Function for ElemSumFunction {
+    fn name(&self) -> &str {
+        NAME
+    }
+
+    fn return_type(
+        &self,
+        _input_types: &[ConcreteDataType],
+    ) -> common_query::error::Result<ConcreteDataType> {
+        Ok(ConcreteDataType::float32_datatype())
+    }
+
+    fn signature(&self) -> Signature {
+        Signature::one_of(
+            vec![
+                TypeSignature::Exact(vec![ConcreteDataType::string_datatype()]),
+                TypeSignature::Exact(vec![ConcreteDataType::binary_datatype()]),
+            ],
+            Volatility::Immutable,
+        )
+    }
+
+    fn eval(
+        &self,
+        _func_ctx: FunctionContext,
+        columns: &[VectorRef],
+    ) -> common_query::error::Result<VectorRef> {
+        ensure!(
+            columns.len() == 1,
+            InvalidFuncArgsSnafu {
+                err_msg: format!(
+                    "The length of the args is not correct, expect exactly one, have: {}",
+                    columns.len()
+                )
+            }
+        );
+        let arg0 = &columns[0];
+
+        let len = arg0.len();
+        let mut result = Float32VectorBuilder::with_capacity(len);
+        if len == 0 {
+            return Ok(result.to_vector());
+        }
+
+        let arg0_const = as_veclit_if_const(arg0)?;
+
+        for i in 0..len {
+            let arg0 = match arg0_const.as_ref() {
+                Some(arg0) => Some(Cow::Borrowed(arg0.as_ref())),
+                None => as_veclit(arg0.get_ref(i))?,
+            };
+            let Some(arg0) = arg0 else {
+                result.push_null();
+                continue;
+            };
+            result.push(Some(DVectorView::from_slice(&arg0, arg0.len()).sum()));
+        }
+
+        Ok(result.to_vector())
+    }
+}
+
+impl Display for ElemSumFunction {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        write!(f, "{}", NAME.to_ascii_uppercase())
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use std::sync::Arc;
+
+    use datatypes::vectors::StringVector;
+
+    use super::*;
+    use crate::function::FunctionContext;
+
+    #[test]
+    fn test_elem_sum() {
+        let func = ElemSumFunction;
+
+        let input0 = Arc::new(StringVector::from(vec![
+            Some("[1.0,2.0,3.0]".to_string()),
+            Some("[4.0,5.0,6.0]".to_string()),
+            None,
+        ]));
+
+        let result = func.eval(FunctionContext::default(), &[input0]).unwrap();
+
+        let result = result.as_ref();
+        assert_eq!(result.len(), 3);
+        assert_eq!(result.get_ref(0).as_f32().unwrap(), Some(6.0));
+        assert_eq!(result.get_ref(1).as_f32().unwrap(), Some(15.0));
+        assert_eq!(result.get_ref(2).as_f32().unwrap(), None);
+    }
+}

src/common/function/src/scalars/vector/impl_conv.rs

@@ -0,0 +1,155 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use std::borrow::Cow;
+use std::sync::Arc;
+
+use common_query::error::{InvalidFuncArgsSnafu, Result};
+use datatypes::prelude::ConcreteDataType;
+use datatypes::value::ValueRef;
+use datatypes::vectors::Vector;
+
+/// Convert a constant string or binary literal to a vector literal.
+pub fn as_veclit_if_const(arg: &Arc<dyn Vector>) -> Result<Option<Cow<'_, [f32]>>> {
+    if !arg.is_const() {
+        return Ok(None);
+    }
+    if arg.data_type() != ConcreteDataType::string_datatype()
+        && arg.data_type() != ConcreteDataType::binary_datatype()
+    {
+        return Ok(None);
+    }
+    as_veclit(arg.get_ref(0))
+}
+
+/// Convert a string or binary literal to a vector literal.
+pub fn as_veclit(arg: ValueRef<'_>) -> Result<Option<Cow<'_, [f32]>>> {
+    match arg.data_type() {
+        ConcreteDataType::Binary(_) => arg
+            .as_binary()
+            .unwrap() // Safe: checked if it is a binary
+            .map(binlit_as_veclit)
+            .transpose(),
+        ConcreteDataType::String(_) => arg
+            .as_string()
+            .unwrap() // Safe: checked if it is a string
+            .map(|s| Ok(Cow::Owned(parse_veclit_from_strlit(s)?)))
+            .transpose(),
+        ConcreteDataType::Null(_) => Ok(None),
+        _ => InvalidFuncArgsSnafu {
+            err_msg: format!("Unsupported data type: {:?}", arg.data_type()),
+        }
+        .fail(),
+    }
+}
+
+/// Convert a u8 slice to a vector literal.
+pub fn binlit_as_veclit(bytes: &[u8]) -> Result<Cow<'_, [f32]>> {
+    if bytes.len() % std::mem::size_of::<f32>() != 0 {
+        return InvalidFuncArgsSnafu {
+            err_msg: format!("Invalid binary length of vector: {}", bytes.len()),
+        }
+        .fail();
+    }
+
+    if cfg!(target_endian = "little") {
+        Ok(unsafe {
+            let vec = std::slice::from_raw_parts(
+                bytes.as_ptr() as *const f32,
+                bytes.len() / std::mem::size_of::<f32>(),
+            );
+            Cow::Borrowed(vec)
+        })
+    } else {
+        let v = bytes
+            .chunks_exact(std::mem::size_of::<f32>())
+            .map(|chunk| f32::from_le_bytes(chunk.try_into().unwrap()))
+            .collect::<Vec<f32>>();
+        Ok(Cow::Owned(v))
+    }
+}
+
+/// Parse a string literal to a vector literal.
+/// Valid inputs are strings like "[1.0, 2.0, 3.0]".
+pub fn parse_veclit_from_strlit(s: &str) -> Result<Vec<f32>> {
+    let trimmed = s.trim();
+    if !trimmed.starts_with('[') || !trimmed.ends_with(']') {
+        return InvalidFuncArgsSnafu {
+            err_msg: format!(
+                "Failed to parse {s} to Vector value: not properly enclosed in brackets"
+            ),
+        }
+        .fail();
+    }
+    let content = trimmed[1..trimmed.len() - 1].trim();
+    if content.is_empty() {
+        return Ok(Vec::new());
+    }
+
+    content
+        .split(',')
+        .map(|s| s.trim().parse::<f32>())
+        .collect::<std::result::Result<_, _>>()
+        .map_err(|e| {
+            InvalidFuncArgsSnafu {
+                err_msg: format!("Failed to parse {s} to Vector value: {e}"),
+            }
+            .build()
+        })
+}
+
+/// Convert a vector literal to a binary literal.
+pub fn veclit_to_binlit(vec: &[f32]) -> Vec<u8> {
+    if cfg!(target_endian = "little") {
+        unsafe {
+            std::slice::from_raw_parts(vec.as_ptr() as *const u8, std::mem::size_of_val(vec))
+                .to_vec()
+        }
+    } else {
+        let mut bytes = Vec::with_capacity(std::mem::size_of_val(vec));
+        for e in vec {
+            bytes.extend_from_slice(&e.to_le_bytes());
+        }
+        bytes
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_parse_veclit_from_strlit() {
+        let result = parse_veclit_from_strlit("[1.0, 2.0, 3.0]").unwrap();
+        assert_eq!(result, vec![1.0, 2.0, 3.0]);
+
+        let result = parse_veclit_from_strlit("[]").unwrap();
+        assert_eq!(result, Vec::<f32>::new());
+
+        let result = parse_veclit_from_strlit("[1.0, a, 3.0]");
+        assert!(result.is_err());
+    }
+
+    #[test]
+    fn test_binlit_as_veclit() {
+        let vec = &[1.0, 2.0, 3.0];
+        let bytes = veclit_to_binlit(vec);
+        let result = binlit_as_veclit(&bytes).unwrap();
+        assert_eq!(result.as_ref(), vec);
+
+        let invalid_bytes = [0, 0, 128];
+        let result = binlit_as_veclit(&invalid_bytes);
+        assert!(result.is_err());
+    }
+}

src/common/function/src/scalars/vector/scalar_add.rs

@@ -0,0 +1,173 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use std::borrow::Cow;
+use std::fmt::Display;
+
+use common_query::error::{InvalidFuncArgsSnafu, Result};
+use common_query::prelude::Signature;
+use datatypes::prelude::ConcreteDataType;
+use datatypes::scalars::ScalarVectorBuilder;
+use datatypes::vectors::{BinaryVectorBuilder, MutableVector, VectorRef};
+use nalgebra::DVectorView;
+use snafu::ensure;
+
+use crate::function::{Function, FunctionContext};
+use crate::helper;
+use crate::scalars::vector::impl_conv::{as_veclit, as_veclit_if_const, veclit_to_binlit};
+
+const NAME: &str = "vec_scalar_add";
+
+/// Adds a scalar to each element of a vector.
+///
+/// # Example
+///
+/// ```sql
+/// SELECT vec_to_string(vec_scalar_add(1, "[1, 2, 3]")) as result;
+///
+/// +---------+
+/// | result  |
+/// +---------+
+/// | [2,3,4] |
+/// +---------+
+///
+/// -- Negative scalar to simulate subtraction
+/// SELECT vec_to_string(vec_scalar_add(-1, "[1, 2, 3]")) as result;
+///
+/// +---------+
+/// | result  |
+/// +---------+
+/// | [0,1,2] |
+/// +---------+
+/// ```
+#[derive(Debug, Clone, Default)]
+pub struct ScalarAddFunction;
+
+impl Function for ScalarAddFunction {
+    fn name(&self) -> &str {
+        NAME
+    }
+
+    fn return_type(&self, _input_types: &[ConcreteDataType]) -> Result<ConcreteDataType> {
+        Ok(ConcreteDataType::binary_datatype())
+    }
+
+    fn signature(&self) -> Signature {
+        helper::one_of_sigs2(
+            vec![ConcreteDataType::float64_datatype()],
+            vec![
+                ConcreteDataType::string_datatype(),
+                ConcreteDataType::binary_datatype(),
+            ],
+        )
+    }
+
+    fn eval(&self, _func_ctx: FunctionContext, columns: &[VectorRef]) -> Result<VectorRef> {
+        ensure!(
+            columns.len() == 2,
+            InvalidFuncArgsSnafu {
+                err_msg: format!(
+                    "The length of the args is not correct, expect exactly two, have: {}",
+                    columns.len()
+                ),
+            }
+        );
+        let arg0 = &columns[0];
+        let arg1 = &columns[1];
+
+        let len = arg0.len();
+        let mut result = BinaryVectorBuilder::with_capacity(len);
+        if len == 0 {
+            return Ok(result.to_vector());
+        }
+
+        let arg1_const = as_veclit_if_const(arg1)?;
+
+        for i in 0..len {
+            let arg0 = arg0.get(i).as_f64_lossy();
+            let Some(arg0) = arg0 else {
+                result.push_null();
+                continue;
+            };
+
+            let arg1 = match arg1_const.as_ref() {
+                Some(arg1) => Some(Cow::Borrowed(arg1.as_ref())),
+                None => as_veclit(arg1.get_ref(i))?,
+            };
+            let Some(arg1) = arg1 else {
+                result.push_null();
+                continue;
+            };
+
+            let vec = DVectorView::from_slice(&arg1, arg1.len());
+            let vec_res = vec.add_scalar(arg0 as _);
+
+            let veclit = vec_res.as_slice();
+            let binlit = veclit_to_binlit(veclit);
+            result.push(Some(&binlit));
+        }
+
+        Ok(result.to_vector())
+    }
+}
+
+impl Display for ScalarAddFunction {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        write!(f, "{}", NAME.to_ascii_uppercase())
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use std::sync::Arc;
+
+    use datatypes::vectors::{Float32Vector, StringVector};
+
+    use super::*;
+
+    #[test]
+    fn test_scalar_add() {
+        let func = ScalarAddFunction;
+
+        let input0 = Arc::new(Float32Vector::from(vec![
+            Some(1.0),
+            Some(-1.0),
+            None,
+            Some(3.0),
+        ]));
+        let input1 = Arc::new(StringVector::from(vec![
+            Some("[1.0,2.0,3.0]".to_string()),
+            Some("[4.0,5.0,6.0]".to_string()),
+            Some("[7.0,8.0,9.0]".to_string()),
+            None,
+        ]));
+
+        let result = func
+            .eval(FunctionContext::default(), &[input0, input1])
+            .unwrap();
+
+        let result = result.as_ref();
+        assert_eq!(result.len(), 4);
+        assert_eq!(
+            result.get_ref(0).as_binary().unwrap(),
+            Some(veclit_to_binlit(&[2.0, 3.0, 4.0]).as_slice())
+        );
+        assert_eq!(
+            result.get_ref(1).as_binary().unwrap(),
+            Some(veclit_to_binlit(&[3.0, 4.0, 5.0]).as_slice())
+        );
+        assert!(result.get_ref(2).is_null());
+        assert!(result.get_ref(3).is_null());
+    }
+}

src/common/function/src/scalars/vector/scalar_mul.rs

@@ -0,0 +1,173 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use std::borrow::Cow;
+use std::fmt::Display;
+
+use common_query::error::{InvalidFuncArgsSnafu, Result};
+use common_query::prelude::Signature;
+use datatypes::prelude::ConcreteDataType;
+use datatypes::scalars::ScalarVectorBuilder;
+use datatypes::vectors::{BinaryVectorBuilder, MutableVector, VectorRef};
+use nalgebra::DVectorView;
+use snafu::ensure;
+
+use crate::function::{Function, FunctionContext};
+use crate::helper;
+use crate::scalars::vector::impl_conv::{as_veclit, as_veclit_if_const, veclit_to_binlit};
+
+const NAME: &str = "vec_scalar_mul";
+
+/// Multiples a scalar to each element of a vector.
+///
+/// # Example
+///
+/// ```sql
+/// SELECT vec_to_string(vec_scalar_mul(2, "[1, 2, 3]")) as result;
+///
+/// +---------+
+/// | result  |
+/// +---------+
+/// | [2,4,6] |
+/// +---------+
+///
+/// -- 1/scalar to simulate division
+/// SELECT vec_to_string(vec_scalar_mul(0.5, "[2, 4, 6]")) as result;
+///
+/// +---------+
+/// | result  |
+/// +---------+
+/// | [1,2,3] |
+/// +---------+
+/// ```
+#[derive(Debug, Clone, Default)]
+pub struct ScalarMulFunction;
+
+impl Function for ScalarMulFunction {
+    fn name(&self) -> &str {
+        NAME
+    }
+
+    fn return_type(&self, _input_types: &[ConcreteDataType]) -> Result<ConcreteDataType> {
+        Ok(ConcreteDataType::binary_datatype())
+    }
+
+    fn signature(&self) -> Signature {
+        helper::one_of_sigs2(
+            vec![ConcreteDataType::float64_datatype()],
+            vec![
+                ConcreteDataType::string_datatype(),
+                ConcreteDataType::binary_datatype(),
+            ],
+        )
+    }
+
+    fn eval(&self, _func_ctx: FunctionContext, columns: &[VectorRef]) -> Result<VectorRef> {
+        ensure!(
+            columns.len() == 2,
+            InvalidFuncArgsSnafu {
+                err_msg: format!(
+                    "The length of the args is not correct, expect exactly two, have: {}",
+                    columns.len()
+                ),
+            }
+        );
+        let arg0 = &columns[0];
+        let arg1 = &columns[1];
+
+        let len = arg0.len();
+        let mut result = BinaryVectorBuilder::with_capacity(len);
+        if len == 0 {
+            return Ok(result.to_vector());
+        }
+
+        let arg1_const = as_veclit_if_const(arg1)?;
+
+        for i in 0..len {
+            let arg0 = arg0.get(i).as_f64_lossy();
+            let Some(arg0) = arg0 else {
+                result.push_null();
+                continue;
+            };
+
+            let arg1 = match arg1_const.as_ref() {
+                Some(arg1) => Some(Cow::Borrowed(arg1.as_ref())),
+                None => as_veclit(arg1.get_ref(i))?,
+            };
+            let Some(arg1) = arg1 else {
+                result.push_null();
+                continue;
+            };
+
+            let vec = DVectorView::from_slice(&arg1, arg1.len());
+            let vec_res = vec.scale(arg0 as _);
+
+            let veclit = vec_res.as_slice();
+            let binlit = veclit_to_binlit(veclit);
+            result.push(Some(&binlit));
+        }
+
+        Ok(result.to_vector())
+    }
+}
+
+impl Display for ScalarMulFunction {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        write!(f, "{}", NAME.to_ascii_uppercase())
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use std::sync::Arc;
+
+    use datatypes::vectors::{Float32Vector, StringVector};
+
+    use super::*;
+
+    #[test]
+    fn test_scalar_mul() {
+        let func = ScalarMulFunction;
+
+        let input0 = Arc::new(Float32Vector::from(vec![
+            Some(2.0),
+            Some(-0.5),
+            None,
+            Some(3.0),
+        ]));
+        let input1 = Arc::new(StringVector::from(vec![
+            Some("[1.0,2.0,3.0]".to_string()),
+            Some("[8.0,10.0,12.0]".to_string()),
+            Some("[7.0,8.0,9.0]".to_string()),
+            None,
+        ]));
+
+        let result = func
+            .eval(FunctionContext::default(), &[input0, input1])
+            .unwrap();
+
+        let result = result.as_ref();
+        assert_eq!(result.len(), 4);
+        assert_eq!(
+            result.get_ref(0).as_binary().unwrap(),
+            Some(veclit_to_binlit(&[2.0, 4.0, 6.0]).as_slice())
+        );
+        assert_eq!(
+            result.get_ref(1).as_binary().unwrap(),
+            Some(veclit_to_binlit(&[-4.0, -5.0, -6.0]).as_slice())
+        );
+        assert!(result.get_ref(2).is_null());
+        assert!(result.get_ref(3).is_null());
+    }
+}

src/common/function/src/scalars/vector/sub.rs

@@ -0,0 +1,223 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use std::borrow::Cow;
+use std::fmt::Display;
+
+use common_query::error::InvalidFuncArgsSnafu;
+use common_query::prelude::Signature;
+use datatypes::prelude::ConcreteDataType;
+use datatypes::scalars::ScalarVectorBuilder;
+use datatypes::vectors::{BinaryVectorBuilder, MutableVector, VectorRef};
+use nalgebra::DVectorView;
+use snafu::ensure;
+
+use crate::function::{Function, FunctionContext};
+use crate::helper;
+use crate::scalars::vector::impl_conv::{as_veclit, as_veclit_if_const, veclit_to_binlit};
+
+const NAME: &str = "vec_sub";
+
+/// Subtracts corresponding elements of two vectors, returns a vector.
+///
+/// # Example
+///
+/// ```sql
+/// SELECT vec_to_string(vec_sub("[1.0, 1.0]", "[1.0, 2.0]")) as result;
+///
+/// +---------------------------------------------------------------+
+/// | vec_to_string(vec_sub(Utf8("[1.0, 1.0]"),Utf8("[1.0, 2.0]"))) |
+/// +---------------------------------------------------------------+
+/// | [0,-1]                                                        |
+/// +---------------------------------------------------------------+
+///
+/// -- Negative scalar to simulate subtraction
+/// SELECT vec_to_string(vec_sub('[-1.0, -1.0]', '[1.0, 2.0]'));
+///
+/// +-----------------------------------------------------------------+
+/// | vec_to_string(vec_sub(Utf8("[-1.0, -1.0]"),Utf8("[1.0, 2.0]"))) |
+/// +-----------------------------------------------------------------+
+/// | [-2,-3]                                                         |
+/// +-----------------------------------------------------------------+
+///
+#[derive(Debug, Clone, Default)]
+pub struct SubFunction;
+
+impl Function for SubFunction {
+    fn name(&self) -> &str {
+        NAME
+    }
+
+    fn return_type(
+        &self,
+        _input_types: &[ConcreteDataType],
+    ) -> common_query::error::Result<ConcreteDataType> {
+        Ok(ConcreteDataType::binary_datatype())
+    }
+
+    fn signature(&self) -> Signature {
+        helper::one_of_sigs2(
+            vec![
+                ConcreteDataType::string_datatype(),
+                ConcreteDataType::binary_datatype(),
+            ],
+            vec![
+                ConcreteDataType::string_datatype(),
+                ConcreteDataType::binary_datatype(),
+            ],
+        )
+    }
+
+    fn eval(
+        &self,
+        _func_ctx: FunctionContext,
+        columns: &[VectorRef],
+    ) -> common_query::error::Result<VectorRef> {
+        ensure!(
+            columns.len() == 2,
+            InvalidFuncArgsSnafu {
+                err_msg: format!(
+                    "The length of the args is not correct, expect exactly two, have: {}",
+                    columns.len()
+                )
+            }
+        );
+        let arg0 = &columns[0];
+        let arg1 = &columns[1];
+
+        ensure!(
+            arg0.len() == arg1.len(),
+            InvalidFuncArgsSnafu {
+                err_msg: format!(
+                    "The lengths of the vector are not aligned, args 0: {}, args 1: {}",
+                    arg0.len(),
+                    arg1.len(),
+                )
+            }
+        );
+
+        let len = arg0.len();
+        let mut result = BinaryVectorBuilder::with_capacity(len);
+        if len == 0 {
+            return Ok(result.to_vector());
+        }
+
+        let arg0_const = as_veclit_if_const(arg0)?;
+        let arg1_const = as_veclit_if_const(arg1)?;
+
+        for i in 0..len {
+            let arg0 = match arg0_const.as_ref() {
+                Some(arg0) => Some(Cow::Borrowed(arg0.as_ref())),
+                None => as_veclit(arg0.get_ref(i))?,
+            };
+            let arg1 = match arg1_const.as_ref() {
+                Some(arg1) => Some(Cow::Borrowed(arg1.as_ref())),
+                None => as_veclit(arg1.get_ref(i))?,
+            };
+            let (Some(arg0), Some(arg1)) = (arg0, arg1) else {
+                result.push_null();
+                continue;
+            };
+            let vec0 = DVectorView::from_slice(&arg0, arg0.len());
+            let vec1 = DVectorView::from_slice(&arg1, arg1.len());
+
+            let vec_res = vec0 - vec1;
+            let veclit = vec_res.as_slice();
+            let binlit = veclit_to_binlit(veclit);
+            result.push(Some(&binlit));
+        }
+
+        Ok(result.to_vector())
+    }
+}
+
+impl Display for SubFunction {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        write!(f, "{}", NAME.to_ascii_uppercase())
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use std::sync::Arc;
+
+    use common_query::error::Error;
+    use datatypes::vectors::StringVector;
+
+    use super::*;
+
+    #[test]
+    fn test_sub() {
+        let func = SubFunction;
+
+        let input0 = Arc::new(StringVector::from(vec![
+            Some("[1.0,2.0,3.0]".to_string()),
+            Some("[4.0,5.0,6.0]".to_string()),
+            None,
+            Some("[2.0,3.0,3.0]".to_string()),
+        ]));
+        let input1 = Arc::new(StringVector::from(vec![
+            Some("[1.0,1.0,1.0]".to_string()),
+            Some("[6.0,5.0,4.0]".to_string()),
+            Some("[3.0,2.0,2.0]".to_string()),
+            None,
+        ]));
+
+        let result = func
+            .eval(FunctionContext::default(), &[input0, input1])
+            .unwrap();
+
+        let result = result.as_ref();
+        assert_eq!(result.len(), 4);
+        assert_eq!(
+            result.get_ref(0).as_binary().unwrap(),
+            Some(veclit_to_binlit(&[0.0, 1.0, 2.0]).as_slice())
+        );
+        assert_eq!(
+            result.get_ref(1).as_binary().unwrap(),
+            Some(veclit_to_binlit(&[-2.0, 0.0, 2.0]).as_slice())
+        );
+        assert!(result.get_ref(2).is_null());
+        assert!(result.get_ref(3).is_null());
+    }
+
+    #[test]
+    fn test_sub_error() {
+        let func = SubFunction;
+
+        let input0 = Arc::new(StringVector::from(vec![
+            Some("[1.0,2.0,3.0]".to_string()),
+            Some("[4.0,5.0,6.0]".to_string()),
+            None,
+            Some("[2.0,3.0,3.0]".to_string()),
+        ]));
+        let input1 = Arc::new(StringVector::from(vec![
+            Some("[1.0,1.0,1.0]".to_string()),
+            Some("[6.0,5.0,4.0]".to_string()),
+            Some("[3.0,2.0,2.0]".to_string()),
+        ]));
+
+        let result = func.eval(FunctionContext::default(), &[input0, input1]);
+
+        match result {
+            Err(Error::InvalidFuncArgs { err_msg, .. }) => {
+                assert_eq!(
+                    err_msg,
+                    "The lengths of the vector are not aligned, args 0: 4, args 1: 3"
+                )
+            }
+            _ => unreachable!(),
+        }
+    }
+}

src/common/function/src/scalars/vector/sum.rs

@@ -0,0 +1,202 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use std::sync::Arc;
+
+use common_macro::{as_aggr_func_creator, AggrFuncTypeStore};
+use common_query::error::{CreateAccumulatorSnafu, Error, InvalidFuncArgsSnafu};
+use common_query::logical_plan::{Accumulator, AggregateFunctionCreator};
+use common_query::prelude::AccumulatorCreatorFunction;
+use datatypes::prelude::{ConcreteDataType, Value, *};
+use datatypes::vectors::VectorRef;
+use nalgebra::{Const, DVectorView, Dyn, OVector};
+use snafu::ensure;
+
+use crate::scalars::vector::impl_conv::{as_veclit, as_veclit_if_const, veclit_to_binlit};
+
+#[derive(Debug, Default)]
+pub struct VectorSum {
+    sum: Option<OVector<f32, Dyn>>,
+    has_null: bool,
+}
+
+#[as_aggr_func_creator]
+#[derive(Debug, Default, AggrFuncTypeStore)]
+pub struct VectorSumCreator {}
+
+impl AggregateFunctionCreator for VectorSumCreator {
+    fn creator(&self) -> AccumulatorCreatorFunction {
+        let creator: AccumulatorCreatorFunction = Arc::new(move |types: &[ConcreteDataType]| {
+            ensure!(
+                types.len() == 1,
+                InvalidFuncArgsSnafu {
+                    err_msg: format!(
+                        "The length of the args is not correct, expect exactly one, have: {}",
+                        types.len()
+                    )
+                }
+            );
+            let input_type = &types[0];
+            match input_type {
+                ConcreteDataType::String(_) | ConcreteDataType::Binary(_) => {
+                    Ok(Box::new(VectorSum::default()))
+                }
+                _ => {
+                    let err_msg = format!(
+                        "\"VEC_SUM\" aggregate function not support data type {:?}",
+                        input_type.logical_type_id(),
+                    );
+                    CreateAccumulatorSnafu { err_msg }.fail()?
+                }
+            }
+        });
+        creator
+    }
+
+    fn output_type(&self) -> common_query::error::Result<ConcreteDataType> {
+        Ok(ConcreteDataType::binary_datatype())
+    }
+
+    fn state_types(&self) -> common_query::error::Result<Vec<ConcreteDataType>> {
+        Ok(vec![self.output_type()?])
+    }
+}
+
+impl VectorSum {
+    fn inner(&mut self, len: usize) -> &mut OVector<f32, Dyn> {
+        self.sum
+            .get_or_insert_with(|| OVector::zeros_generic(Dyn(len), Const::<1>))
+    }
+
+    fn update(&mut self, values: &[VectorRef], is_update: bool) -> Result<(), Error> {
+        if values.is_empty() || self.has_null {
+            return Ok(());
+        };
+        let column = &values[0];
+        let len = column.len();
+
+        match as_veclit_if_const(column)? {
+            Some(column) => {
+                let vec_column = DVectorView::from_slice(&column, column.len()).scale(len as f32);
+                *self.inner(vec_column.len()) += vec_column;
+            }
+            None => {
+                for i in 0..len {
+                    let Some(arg0) = as_veclit(column.get_ref(i))? else {
+                        if is_update {
+                            self.has_null = true;
+                            self.sum = None;
+                        }
+                        return Ok(());
+                    };
+                    let vec_column = DVectorView::from_slice(&arg0, arg0.len());
+                    *self.inner(vec_column.len()) += vec_column;
+                }
+            }
+        }
+        Ok(())
+    }
+}
+
+impl Accumulator for VectorSum {
+    fn state(&self) -> common_query::error::Result<Vec<Value>> {
+        self.evaluate().map(|v| vec![v])
+    }
+
+    fn update_batch(&mut self, values: &[VectorRef]) -> common_query::error::Result<()> {
+        self.update(values, true)
+    }
+
+    fn merge_batch(&mut self, states: &[VectorRef]) -> common_query::error::Result<()> {
+        self.update(states, false)
+    }
+
+    fn evaluate(&self) -> common_query::error::Result<Value> {
+        match &self.sum {
+            None => Ok(Value::Null),
+            Some(vector) => Ok(Value::from(veclit_to_binlit(vector.as_slice()))),
+        }
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use std::sync::Arc;
+
+    use datatypes::vectors::{ConstantVector, StringVector};
+
+    use super::*;
+
+    #[test]
+    fn test_update_batch() {
+        // test update empty batch, expect not updating anything
+        let mut vec_sum = VectorSum::default();
+        vec_sum.update_batch(&[]).unwrap();
+        assert!(vec_sum.sum.is_none());
+        assert!(!vec_sum.has_null);
+        assert_eq!(Value::Null, vec_sum.evaluate().unwrap());
+
+        // test update one not-null value
+        let mut vec_sum = VectorSum::default();
+        let v: Vec<VectorRef> = vec![Arc::new(StringVector::from(vec![Some(
+            "[1.0,2.0,3.0]".to_string(),
+        )]))];
+        vec_sum.update_batch(&v).unwrap();
+        assert_eq!(
+            Value::from(veclit_to_binlit(&[1.0, 2.0, 3.0])),
+            vec_sum.evaluate().unwrap()
+        );
+
+        // test update one null value
+        let mut vec_sum = VectorSum::default();
+        let v: Vec<VectorRef> = vec![Arc::new(StringVector::from(vec![Option::<String>::None]))];
+        vec_sum.update_batch(&v).unwrap();
+        assert_eq!(Value::Null, vec_sum.evaluate().unwrap());
+
+        // test update no null-value batch
+        let mut vec_sum = VectorSum::default();
+        let v: Vec<VectorRef> = vec![Arc::new(StringVector::from(vec![
+            Some("[1.0,2.0,3.0]".to_string()),
+            Some("[4.0,5.0,6.0]".to_string()),
+            Some("[7.0,8.0,9.0]".to_string()),
+        ]))];
+        vec_sum.update_batch(&v).unwrap();
+        assert_eq!(
+            Value::from(veclit_to_binlit(&[12.0, 15.0, 18.0])),
+            vec_sum.evaluate().unwrap()
+        );
+
+        // test update null-value batch
+        let mut vec_sum = VectorSum::default();
+        let v: Vec<VectorRef> = vec![Arc::new(StringVector::from(vec![
+            Some("[1.0,2.0,3.0]".to_string()),
+            None,
+            Some("[7.0,8.0,9.0]".to_string()),
+        ]))];
+        vec_sum.update_batch(&v).unwrap();
+        assert_eq!(Value::Null, vec_sum.evaluate().unwrap());
+
+        // test update with constant vector
+        let mut vec_sum = VectorSum::default();
+        let v: Vec<VectorRef> = vec![Arc::new(ConstantVector::new(
+            Arc::new(StringVector::from_vec(vec!["[1.0,2.0,3.0]".to_string()])),
+            4,
+        ))];
+        vec_sum.update_batch(&v).unwrap();
+        assert_eq!(
+            Value::from(veclit_to_binlit(&[4.0, 8.0, 12.0])),
+            vec_sum.evaluate().unwrap()
+        );
+    }
+}

src/common/function/src/scalars/vector/vector_div.rs

@@ -0,0 +1,218 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use std::borrow::Cow;
+use std::fmt::Display;
+
+use common_query::error::{InvalidFuncArgsSnafu, Result};
+use common_query::prelude::Signature;
+use datatypes::prelude::ConcreteDataType;
+use datatypes::scalars::ScalarVectorBuilder;
+use datatypes::vectors::{BinaryVectorBuilder, MutableVector, VectorRef};
+use nalgebra::DVectorView;
+use snafu::ensure;
+
+use crate::function::{Function, FunctionContext};
+use crate::helper;
+use crate::scalars::vector::impl_conv::{as_veclit, as_veclit_if_const, veclit_to_binlit};
+
+const NAME: &str = "vec_div";
+
+/// Divides corresponding elements of two vectors.
+///
+/// # Example
+///
+/// ```sql
+/// SELECT vec_to_string(vec_div("[2, 4, 6]", "[2, 2, 2]")) as result;
+///
+/// +---------+
+/// | result  |
+/// +---------+
+/// | [1,2,3] |
+/// +---------+
+///
+/// ```
+#[derive(Debug, Clone, Default)]
+pub struct VectorDivFunction;
+
+impl Function for VectorDivFunction {
+    fn name(&self) -> &str {
+        NAME
+    }
+
+    fn return_type(&self, _input_types: &[ConcreteDataType]) -> Result<ConcreteDataType> {
+        Ok(ConcreteDataType::binary_datatype())
+    }
+
+    fn signature(&self) -> Signature {
+        helper::one_of_sigs2(
+            vec![
+                ConcreteDataType::string_datatype(),
+                ConcreteDataType::binary_datatype(),
+            ],
+            vec![
+                ConcreteDataType::string_datatype(),
+                ConcreteDataType::binary_datatype(),
+            ],
+        )
+    }
+
+    fn eval(&self, _func_ctx: FunctionContext, columns: &[VectorRef]) -> Result<VectorRef> {
+        ensure!(
+            columns.len() == 2,
+            InvalidFuncArgsSnafu {
+                err_msg: format!(
+                    "The length of the args is not correct, expect exactly two, have: {}",
+                    columns.len()
+                ),
+            }
+        );
+
+        let arg0 = &columns[0];
+        let arg1 = &columns[1];
+
+        let len = arg0.len();
+        let mut result = BinaryVectorBuilder::with_capacity(len);
+        if len == 0 {
+            return Ok(result.to_vector());
+        }
+
+        let arg0_const = as_veclit_if_const(arg0)?;
+        let arg1_const = as_veclit_if_const(arg1)?;
+
+        for i in 0..len {
+            let arg0 = match arg0_const.as_ref() {
+                Some(arg0) => Some(Cow::Borrowed(arg0.as_ref())),
+                None => as_veclit(arg0.get_ref(i))?,
+            };
+
+            let arg1 = match arg1_const.as_ref() {
+                Some(arg1) => Some(Cow::Borrowed(arg1.as_ref())),
+                None => as_veclit(arg1.get_ref(i))?,
+            };
+
+            if let (Some(arg0), Some(arg1)) = (arg0, arg1) {
+                ensure!(
+                    arg0.len() == arg1.len(),
+                    InvalidFuncArgsSnafu {
+                        err_msg: format!(
+                            "The length of the vectors must match for division, have: {} vs {}",
+                            arg0.len(),
+                            arg1.len()
+                        ),
+                    }
+                );
+                let vec0 = DVectorView::from_slice(&arg0, arg0.len());
+                let vec1 = DVectorView::from_slice(&arg1, arg1.len());
+                let vec_res = vec0.component_div(&vec1);
+
+                let veclit = vec_res.as_slice();
+                let binlit = veclit_to_binlit(veclit);
+                result.push(Some(&binlit));
+            } else {
+                result.push_null();
+            }
+        }
+
+        Ok(result.to_vector())
+    }
+}
+
+impl Display for VectorDivFunction {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        write!(f, "{}", NAME.to_ascii_uppercase())
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use std::sync::Arc;
+
+    use common_query::error;
+    use datatypes::vectors::StringVector;
+
+    use super::*;
+
+    #[test]
+    fn test_vector_mul() {
+        let func = VectorDivFunction;
+
+        let vec0 = vec![1.0, 2.0, 3.0];
+        let vec1 = vec![1.0, 1.0];
+        let (len0, len1) = (vec0.len(), vec1.len());
+        let input0 = Arc::new(StringVector::from(vec![Some(format!("{vec0:?}"))]));
+        let input1 = Arc::new(StringVector::from(vec![Some(format!("{vec1:?}"))]));
+
+        let err = func
+            .eval(FunctionContext::default(), &[input0, input1])
+            .unwrap_err();
+
+        match err {
+            error::Error::InvalidFuncArgs { err_msg, .. } => {
+                assert_eq!(
+                    err_msg,
+                    format!(
+                        "The length of the vectors must match for division, have: {} vs {}",
+                        len0, len1
+                    )
+                )
+            }
+            _ => unreachable!(),
+        }
+
+        let input0 = Arc::new(StringVector::from(vec![
+            Some("[1.0,2.0,3.0]".to_string()),
+            Some("[8.0,10.0,12.0]".to_string()),
+            Some("[7.0,8.0,9.0]".to_string()),
+            None,
+        ]));
+
+        let input1 = Arc::new(StringVector::from(vec![
+            Some("[1.0,1.0,1.0]".to_string()),
+            Some("[2.0,2.0,2.0]".to_string()),
+            None,
+            Some("[3.0,3.0,3.0]".to_string()),
+        ]));
+
+        let result = func
+            .eval(FunctionContext::default(), &[input0, input1])
+            .unwrap();
+
+        let result = result.as_ref();
+        assert_eq!(result.len(), 4);
+        assert_eq!(
+            result.get_ref(0).as_binary().unwrap(),
+            Some(veclit_to_binlit(&[1.0, 2.0, 3.0]).as_slice())
+        );
+        assert_eq!(
+            result.get_ref(1).as_binary().unwrap(),
+            Some(veclit_to_binlit(&[4.0, 5.0, 6.0]).as_slice())
+        );
+        assert!(result.get_ref(2).is_null());
+        assert!(result.get_ref(3).is_null());
+
+        let input0 = Arc::new(StringVector::from(vec![Some("[1.0,-2.0]".to_string())]));
+        let input1 = Arc::new(StringVector::from(vec![Some("[0.0,0.0]".to_string())]));
+
+        let result = func
+            .eval(FunctionContext::default(), &[input0, input1])
+            .unwrap();
+
+        let result = result.as_ref();
+        assert_eq!(
+            result.get_ref(0).as_binary().unwrap(),
+            Some(veclit_to_binlit(&[f64::INFINITY as f32, f64::NEG_INFINITY as f32]).as_slice())
+        );
+    }
+}

src/common/function/src/scalars/vector/vector_mul.rs

@@ -0,0 +1,205 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use std::borrow::Cow;
+use std::fmt::Display;
+
+use common_query::error::{InvalidFuncArgsSnafu, Result};
+use common_query::prelude::Signature;
+use datatypes::prelude::ConcreteDataType;
+use datatypes::scalars::ScalarVectorBuilder;
+use datatypes::vectors::{BinaryVectorBuilder, MutableVector, VectorRef};
+use nalgebra::DVectorView;
+use snafu::ensure;
+
+use crate::function::{Function, FunctionContext};
+use crate::helper;
+use crate::scalars::vector::impl_conv::{as_veclit, as_veclit_if_const, veclit_to_binlit};
+
+const NAME: &str = "vec_mul";
+
+/// Multiplies corresponding elements of two vectors.
+///
+/// # Example
+///
+/// ```sql
+/// SELECT vec_to_string(vec_mul("[1, 2, 3]", "[1, 2, 3]")) as result;
+///
+/// +---------+
+/// | result  |
+/// +---------+
+/// | [1,4,9] |
+/// +---------+
+///
+/// ```
+#[derive(Debug, Clone, Default)]
+pub struct VectorMulFunction;
+
+impl Function for VectorMulFunction {
+    fn name(&self) -> &str {
+        NAME
+    }
+
+    fn return_type(&self, _input_types: &[ConcreteDataType]) -> Result<ConcreteDataType> {
+        Ok(ConcreteDataType::binary_datatype())
+    }
+
+    fn signature(&self) -> Signature {
+        helper::one_of_sigs2(
+            vec![
+                ConcreteDataType::string_datatype(),
+                ConcreteDataType::binary_datatype(),
+            ],
+            vec![
+                ConcreteDataType::string_datatype(),
+                ConcreteDataType::binary_datatype(),
+            ],
+        )
+    }
+
+    fn eval(&self, _func_ctx: FunctionContext, columns: &[VectorRef]) -> Result<VectorRef> {
+        ensure!(
+            columns.len() == 2,
+            InvalidFuncArgsSnafu {
+                err_msg: format!(
+                    "The length of the args is not correct, expect exactly two, have: {}",
+                    columns.len()
+                ),
+            }
+        );
+
+        let arg0 = &columns[0];
+        let arg1 = &columns[1];
+
+        let len = arg0.len();
+        let mut result = BinaryVectorBuilder::with_capacity(len);
+        if len == 0 {
+            return Ok(result.to_vector());
+        }
+
+        let arg0_const = as_veclit_if_const(arg0)?;
+        let arg1_const = as_veclit_if_const(arg1)?;
+
+        for i in 0..len {
+            let arg0 = match arg0_const.as_ref() {
+                Some(arg0) => Some(Cow::Borrowed(arg0.as_ref())),
+                None => as_veclit(arg0.get_ref(i))?,
+            };
+
+            let arg1 = match arg1_const.as_ref() {
+                Some(arg1) => Some(Cow::Borrowed(arg1.as_ref())),
+                None => as_veclit(arg1.get_ref(i))?,
+            };
+
+            if let (Some(arg0), Some(arg1)) = (arg0, arg1) {
+                ensure!(
+                    arg0.len() == arg1.len(),
+                    InvalidFuncArgsSnafu {
+                        err_msg: format!(
+                            "The length of the vectors must match for multiplying, have: {} vs {}",
+                            arg0.len(),
+                            arg1.len()
+                        ),
+                    }
+                );
+                let vec0 = DVectorView::from_slice(&arg0, arg0.len());
+                let vec1 = DVectorView::from_slice(&arg1, arg1.len());
+                let vec_res = vec1.component_mul(&vec0);
+
+                let veclit = vec_res.as_slice();
+                let binlit = veclit_to_binlit(veclit);
+                result.push(Some(&binlit));
+            } else {
+                result.push_null();
+            }
+        }
+
+        Ok(result.to_vector())
+    }
+}
+
+impl Display for VectorMulFunction {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        write!(f, "{}", NAME.to_ascii_uppercase())
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use std::sync::Arc;
+
+    use common_query::error;
+    use datatypes::vectors::StringVector;
+
+    use super::*;
+
+    #[test]
+    fn test_vector_mul() {
+        let func = VectorMulFunction;
+
+        let vec0 = vec![1.0, 2.0, 3.0];
+        let vec1 = vec![1.0, 1.0];
+        let (len0, len1) = (vec0.len(), vec1.len());
+        let input0 = Arc::new(StringVector::from(vec![Some(format!("{vec0:?}"))]));
+        let input1 = Arc::new(StringVector::from(vec![Some(format!("{vec1:?}"))]));
+
+        let err = func
+            .eval(FunctionContext::default(), &[input0, input1])
+            .unwrap_err();
+
+        match err {
+            error::Error::InvalidFuncArgs { err_msg, .. } => {
+                assert_eq!(
+                    err_msg,
+                    format!(
+                        "The length of the vectors must match for multiplying, have: {} vs {}",
+                        len0, len1
+                    )
+                )
+            }
+            _ => unreachable!(),
+        }
+
+        let input0 = Arc::new(StringVector::from(vec![
+            Some("[1.0,2.0,3.0]".to_string()),
+            Some("[8.0,10.0,12.0]".to_string()),
+            Some("[7.0,8.0,9.0]".to_string()),
+            None,
+        ]));
+
+        let input1 = Arc::new(StringVector::from(vec![
+            Some("[1.0,1.0,1.0]".to_string()),
+            Some("[2.0,2.0,2.0]".to_string()),
+            None,
+            Some("[3.0,3.0,3.0]".to_string()),
+        ]));
+
+        let result = func
+            .eval(FunctionContext::default(), &[input0, input1])
+            .unwrap();
+
+        let result = result.as_ref();
+        assert_eq!(result.len(), 4);
+        assert_eq!(
+            result.get_ref(0).as_binary().unwrap(),
+            Some(veclit_to_binlit(&[1.0, 2.0, 3.0]).as_slice())
+        );
+        assert_eq!(
+            result.get_ref(1).as_binary().unwrap(),
+            Some(veclit_to_binlit(&[16.0, 20.0, 24.0]).as_slice())
+        );
+        assert!(result.get_ref(2).is_null());
+        assert!(result.get_ref(3).is_null());
+    }
+}

src/common/function/src/utils.rs

@@ -0,0 +1,58 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+/// Escapes special characters in the provided pattern string for `LIKE`.
+///
+/// Specifically, it prefixes the backslash (`\`), percent (`%`), and underscore (`_`)
+/// characters with an additional backslash to ensure they are treated literally.
+///
+/// # Examples
+///
+/// ```rust
+/// let escaped = escape_pattern("100%_some\\path");
+/// assert_eq!(escaped, "100\\%\\_some\\\\path");
+/// ```
+pub fn escape_like_pattern(pattern: &str) -> String {
+    pattern
+        .chars()
+        .flat_map(|c| match c {
+            '\\' | '%' | '_' => vec!['\\', c],
+            _ => vec![c],
+        })
+        .collect::<String>()
+}
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_escape_like_pattern() {
+        assert_eq!(
+            escape_like_pattern("100%_some\\path"),
+            "100\\%\\_some\\\\path"
+        );
+        assert_eq!(escape_like_pattern(""), "");
+        assert_eq!(escape_like_pattern("hello"), "hello");
+        assert_eq!(escape_like_pattern("\\%_"), "\\\\\\%\\_");
+        assert_eq!(escape_like_pattern("%%__\\\\"), "\\%\\%\\_\\_\\\\\\\\");
+        assert_eq!(escape_like_pattern("abc123"), "abc123");
+        assert_eq!(escape_like_pattern("%_\\"), "\\%\\_\\\\");
+        assert_eq!(
+            escape_like_pattern("%%__\\\\another%string"),
+            "\\%\\%\\_\\_\\\\\\\\another\\%string"
+        );
+        assert_eq!(escape_like_pattern("foo%bar_"), "foo\\%bar\\_");
+        assert_eq!(escape_like_pattern("\\_\\%"), "\\\\\\_\\\\\\%");
+    }
+}