fix(mito): use 1day as default time partition duration (#6202)

* fix unit tests

* fix: sqlness

* fix/default-time-window:
 ## Add Helper Functions and Enhance Compaction Tests

 - **Refactor Compaction Logic**: Introduced helper functions `flush` and `compact` in `compaction_test.rs` to streamline compaction operations.
 - **Enhance Compaction Tests**: Added a new test `test_infer_compaction_time_window` in `compaction_test.rs` to verify compaction time window inference.
 - **Testing Improvements**: Added `#[cfg(test)]` attribute to `new_multi_partitions` in `time_partition.rs` to ensure it's only included in test builds.

* fix/default-time-window:
 - **Refactor `TimePartition` Struct**: Removed unnecessary comments regarding `time_range` in `time_partition.rs`.
 - **Enhance `TimePartitions` Functionality**: Added a method `part_duration_or_default` to provide a default partition duration in `time_partition.rs`.
 - **Update SQL Test Cases**: Modified SQL operations and expected results in `scan_big_varchar.result` and `scan_big_varchar.sql` to reflect changes in data manipulation logic.

* fix/default-time-window:
 ### Update Time Partition Default Duration

 - **Refactor Default Duration**: Introduced `INITIAL_TIME_WINDOW` constant to define the default time window duration as `Duration::from_days(1)`. This change replaces multiple instances of the hardcoded default duration across the `time_partition.rs` file.
 - **Files Affected**: `time_partition.rs`

* fix/default-time-window:
 ## Update Partition Duration Handling

 - **`time_partition.rs`**: Refactored `part_duration` to be non-optional, removing `Option` wrapper. Updated logic to use `unwrap_or` with `INITIAL_TIME_WINDOW` where necessary. Adjusted related methods and tests to accommodate this change.
 - **`version.rs` (memtable and region)**: Updated handling of `part_duration` to align with changes in `time_partition.rs`, ensuring consistent use of non-optional `Duration`.

* fix/default-time-window:
 ### Improve Error Context in `time_partition.rs`

 - Enhanced error context message in `time_partition.rs` to provide clearer information on partition time range issues, including bucket size details.

Signed-off-by: Lei, HUANG <lhuang@greptime.com>

---------

Signed-off-by: Lei, HUANG <lhuang@greptime.com>

.github/CODEOWNERS

@@ -4,7 +4,7 @@
 
 * @GreptimeTeam/db-approver
 
-## [Module] Databse Engine
+## [Module] Database Engine
 /src/index @zhongzc
 /src/mito2 @evenyag @v0y4g3r @waynexia
 /src/query @evenyag

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

@@ -52,7 +52,7 @@ runs:
       uses: ./.github/actions/build-greptime-binary
       with:
         base-image: ubuntu
-        features: servers/dashboard,pg_kvbackend,mysql_kvbackend
+        features: servers/dashboard
         cargo-profile: ${{ inputs.cargo-profile }}
         artifacts-dir: greptime-linux-${{ inputs.arch }}-${{ inputs.version }}
         version: ${{ inputs.version }}
@@ -70,7 +70,7 @@ runs:
       if: ${{ inputs.arch == 'amd64' && inputs.dev-mode == 'false' }} # Builds greptime for centos if the host machine is amd64.
       with:
         base-image: centos
-        features: servers/dashboard,pg_kvbackend,mysql_kvbackend
+        features: servers/dashboard
         cargo-profile: ${{ inputs.cargo-profile }}
         artifacts-dir: greptime-linux-${{ inputs.arch }}-centos-${{ inputs.version }}
         version: ${{ inputs.version }}

.github/actions/release-cn-artifacts/action.yaml

@@ -64,11 +64,11 @@ inputs:
   upload-max-retry-times:
     description: Max retry times for uploading artifacts to S3
     required: false
-    default: "20"
+    default: "30"
   upload-retry-timeout:
     description: Timeout for uploading artifacts to S3
     required: false
-    default: "30" # minutes
+    default: "120" # minutes
 runs:
   using: composite
   steps:

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

@@ -22,7 +22,6 @@ datanode:
     [wal]
     provider = "kafka"
     broker_endpoints = ["kafka.kafka-cluster.svc.cluster.local:9092"]
-    linger = "2ms"
     overwrite_entry_start_id = true
 frontend:
   configData: |-

.github/scripts/create-version.sh

@@ -8,19 +8,20 @@ set -e
 # - If it's a nightly build, the version is 'nightly-YYYYMMDD-$(git rev-parse --short HEAD)', like 'nightly-20230712-e5b243c'.
 # create_version ${GIHUB_EVENT_NAME} ${NEXT_RELEASE_VERSION} ${NIGHTLY_RELEASE_PREFIX}
 function create_version() {
-  # Read from envrionment variables.
+  # Read from environment variables.
   if [ -z "$GITHUB_EVENT_NAME" ]; then
-      echo "GITHUB_EVENT_NAME is empty"
+      echo "GITHUB_EVENT_NAME is empty" >&2
       exit 1
   fi
 
   if [ -z "$NEXT_RELEASE_VERSION" ]; then
-      echo "NEXT_RELEASE_VERSION is empty"
-      exit 1
+      echo "NEXT_RELEASE_VERSION is empty, use version from Cargo.toml" >&2
+      # NOTE: Need a `v` prefix for the version string.
+      export NEXT_RELEASE_VERSION=v$(grep '^version = ' Cargo.toml | cut -d '"' -f 2 | head -n 1)
   fi
 
   if [ -z "$NIGHTLY_RELEASE_PREFIX" ]; then
-      echo "NIGHTLY_RELEASE_PREFIX is empty"
+      echo "NIGHTLY_RELEASE_PREFIX is empty" >&2
       exit 1
   fi
 
@@ -35,7 +36,7 @@ function create_version() {
   # It will be like 'dev-2023080819-f0e7216c'.
   if [ "$NEXT_RELEASE_VERSION" = dev ]; then
     if [ -z "$COMMIT_SHA" ]; then
-      echo "COMMIT_SHA is empty in dev build"
+      echo "COMMIT_SHA is empty in dev build" >&2
       exit 1
     fi
     echo "dev-$(date "+%Y%m%d-%s")-$(echo "$COMMIT_SHA" | cut -c1-8)"
@@ -45,7 +46,7 @@ function create_version() {
   # Note: Only output 'version=xxx' to stdout when everything is ok, so that it can be used in GitHub Actions Outputs.
   if [ "$GITHUB_EVENT_NAME" = push ]; then
     if [ -z "$GITHUB_REF_NAME" ]; then
-      echo "GITHUB_REF_NAME is empty in push event"
+      echo "GITHUB_REF_NAME is empty in push event" >&2
       exit 1
     fi
     echo "$GITHUB_REF_NAME"
@@ -54,7 +55,7 @@ function create_version() {
   elif [ "$GITHUB_EVENT_NAME" = schedule ]; then
     echo "$NEXT_RELEASE_VERSION-$NIGHTLY_RELEASE_PREFIX-$(date "+%Y%m%d")"
   else
-    echo "Unsupported GITHUB_EVENT_NAME: $GITHUB_EVENT_NAME"
+    echo "Unsupported GITHUB_EVENT_NAME: $GITHUB_EVENT_NAME" >&2
     exit 1
   fi
 }

.github/scripts/deploy-greptimedb.sh

@@ -10,7 +10,7 @@ GREPTIMEDB_IMAGE_TAG=${GREPTIMEDB_IMAGE_TAG:-latest}
 ETCD_CHART="oci://registry-1.docker.io/bitnamicharts/etcd"
 GREPTIME_CHART="https://greptimeteam.github.io/helm-charts/"
 
-# Ceate a cluster with 1 control-plane node and 5 workers.
+# Create a cluster with 1 control-plane node and 5 workers.
 function create_kind_cluster() {
   cat <<EOF | kind create cluster --name "${CLUSTER}" --image kindest/node:"$KUBERNETES_VERSION" --config=-
 kind: Cluster

.github/scripts/update-dev-builder-version.sh

@@ -0,0 +1,37 @@
+#!/bin/bash
+
+DEV_BUILDER_IMAGE_TAG=$1
+
+update_dev_builder_version() {
+  if [ -z "$DEV_BUILDER_IMAGE_TAG" ]; then
+    echo "Error: Should specify the dev-builder image tag"
+    exit 1
+  fi
+
+  # Configure Git configs.
+  git config --global user.email greptimedb-ci@greptime.com
+  git config --global user.name greptimedb-ci
+
+  # Checkout a new branch.
+  BRANCH_NAME="ci/update-dev-builder-$(date +%Y%m%d%H%M%S)"
+  git checkout -b $BRANCH_NAME
+
+  # Update the dev-builder image tag in the Makefile.
+  sed -i "s/DEV_BUILDER_IMAGE_TAG ?=.*/DEV_BUILDER_IMAGE_TAG ?= ${DEV_BUILDER_IMAGE_TAG}/g" Makefile
+
+  # Commit the changes.
+  git add Makefile
+  git commit -m "ci: update dev-builder image tag"
+  git push origin $BRANCH_NAME
+
+  # Create a Pull Request.
+  gh pr create \
+      --title "ci: update dev-builder image tag" \
+      --body "This PR updates the dev-builder image tag" \
+      --base main \
+      --head $BRANCH_NAME \
+      --reviewer zyy17 \
+      --reviewer daviderli614
+}
+
+update_dev_builder_version

.github/scripts/update-helm-charts-version.sh

@@ -0,0 +1,46 @@
+#!/bin/bash
+
+set -e
+
+VERSION=${VERSION}
+GITHUB_TOKEN=${GITHUB_TOKEN}
+
+update_helm_charts_version() {
+  # Configure Git configs.
+  git config --global user.email update-helm-charts-version@greptime.com
+  git config --global user.name update-helm-charts-version
+
+  # Clone helm-charts repository.
+  git clone "https://x-access-token:${GITHUB_TOKEN}@github.com/GreptimeTeam/helm-charts.git"
+  cd helm-charts
+
+  # Set default remote for gh CLI
+  gh repo set-default GreptimeTeam/helm-charts
+
+  # Checkout a new branch.
+  BRANCH_NAME="chore/greptimedb-${VERSION}"
+  git checkout -b $BRANCH_NAME
+
+  # Update version.
+  make update-version CHART=greptimedb-cluster VERSION=${VERSION}
+  make update-version CHART=greptimedb-standalone VERSION=${VERSION}
+
+  # Update docs.
+  make docs
+
+  # Commit the changes.
+  git add .
+  git commit -m "chore: Update GreptimeDB version to ${VERSION}"
+  git push origin $BRANCH_NAME
+
+  # Create a Pull Request.
+  gh pr create \
+    --title "chore: Update GreptimeDB version to ${VERSION}" \
+    --body "This PR updates the GreptimeDB version." \
+    --base main \
+    --head $BRANCH_NAME \
+    --reviewer zyy17 \
+    --reviewer daviderli614
+}
+
+update_helm_charts_version

.github/scripts/update-homebrew-greptme-version.sh

@@ -0,0 +1,42 @@
+#!/bin/bash
+
+set -e
+
+VERSION=${VERSION}
+GITHUB_TOKEN=${GITHUB_TOKEN}
+
+update_homebrew_greptime_version() {
+  # Configure Git configs.
+  git config --global user.email update-greptime-version@greptime.com
+  git config --global user.name update-greptime-version
+
+  # Clone helm-charts repository.
+  git clone "https://x-access-token:${GITHUB_TOKEN}@github.com/GreptimeTeam/homebrew-greptime.git"
+  cd homebrew-greptime
+
+  # Set default remote for gh CLI
+  gh repo set-default GreptimeTeam/homebrew-greptime
+
+  # Checkout a new branch.
+  BRANCH_NAME="chore/greptimedb-${VERSION}"
+  git checkout -b $BRANCH_NAME
+
+  # Update version.
+  make update-greptime-version VERSION=${VERSION}
+
+  # Commit the changes.
+  git add .
+  git commit -m "chore: Update GreptimeDB version to ${VERSION}"
+  git push origin $BRANCH_NAME
+
+  # Create a Pull Request.
+  gh pr create \
+    --title "chore: Update GreptimeDB version to ${VERSION}" \
+    --body "This PR updates the GreptimeDB version." \
+    --base main \
+    --head $BRANCH_NAME \
+    --reviewer zyy17 \
+    --reviewer daviderli614
+}
+
+update_homebrew_greptime_version

.github/scripts/upload-artifacts-to-s3.sh

@@ -41,7 +41,7 @@ function upload_artifacts() {
 # Updates the latest version information in AWS S3 if UPDATE_VERSION_INFO is true.
 function update_version_info() {
   if [ "$UPDATE_VERSION_INFO" == "true" ]; then
-    # If it's the officail release(like v1.0.0, v1.0.1, v1.0.2, etc.), update latest-version.txt.
+    # If it's the official release(like v1.0.0, v1.0.1, v1.0.2, etc.), update latest-version.txt.
     if [[ "$VERSION" =~ ^v[0-9]+\.[0-9]+\.[0-9]+$ ]]; then
       echo "Updating latest-version.txt"
       echo "$VERSION" > latest-version.txt

.github/workflows/dev-build.yml

@@ -55,6 +55,11 @@ on:
         description: Build and push images to DockerHub and ACR
         required: false
         default: true
+      upload_artifacts_to_s3:
+        type: boolean
+        description: Whether upload artifacts to s3
+        required: false
+        default: false
       cargo_profile:
         type: choice
         description: The cargo profile to use in building GreptimeDB.
@@ -238,7 +243,7 @@ jobs:
           version: ${{ needs.allocate-runners.outputs.version }}
           push-latest-tag: false # Don't push the latest tag to registry.
           dev-mode: true # Only build the standard images.
-          
+
       - name: Echo Docker image tag to step summary
         run: |
           echo "## Docker Image Tag" >> $GITHUB_STEP_SUMMARY
@@ -281,7 +286,7 @@ jobs:
           aws-cn-access-key-id: ${{ secrets.AWS_CN_ACCESS_KEY_ID }}
           aws-cn-secret-access-key: ${{ secrets.AWS_CN_SECRET_ACCESS_KEY }}
           aws-cn-region: ${{ vars.AWS_RELEASE_BUCKET_REGION }}
-          upload-to-s3: false
+          upload-to-s3: ${{ inputs.upload_artifacts_to_s3 }}
           dev-mode: true                     # Only build the standard images(exclude centos images).
           push-latest-tag: false             # Don't push the latest tag to registry.
           update-version-info: false         # Don't update the version info in S3.

.github/workflows/develop.yml

@@ -22,6 +22,7 @@ concurrency:
 
 jobs:
   check-typos-and-docs:
+    if: ${{ github.repository == 'GreptimeTeam/greptimedb' }}
     name: Check typos and docs
     runs-on: ubuntu-latest
     steps:
@@ -36,6 +37,7 @@ jobs:
           || (echo "'config/config.md' is not up-to-date, please run 'make config-docs'." && exit 1)
 
   license-header-check:
+    if: ${{ github.repository == 'GreptimeTeam/greptimedb' }}
     runs-on: ubuntu-latest
     name: Check License Header
     steps:
@@ -45,6 +47,7 @@ jobs:
       - uses: korandoru/hawkeye@v5
 
   check:
+    if: ${{ github.repository == 'GreptimeTeam/greptimedb' }}
     name: Check
     runs-on: ${{ matrix.os }}
     strategy:
@@ -71,6 +74,7 @@ jobs:
         run: cargo check --locked --workspace --all-targets
 
   toml:
+    if: ${{ github.repository == 'GreptimeTeam/greptimedb' }}
     name: Toml Check
     runs-on: ubuntu-latest
     timeout-minutes: 60
@@ -85,6 +89,7 @@ jobs:
         run: taplo format --check
 
   build:
+    if: ${{ github.repository == 'GreptimeTeam/greptimedb' }}
     name: Build GreptimeDB binaries
     runs-on: ${{ matrix.os }}
     strategy:
@@ -127,6 +132,7 @@ jobs:
           version: current
 
   fuzztest:
+    if: ${{ github.repository == 'GreptimeTeam/greptimedb' }}
     name: Fuzz Test
     needs: build
     runs-on: ubuntu-latest
@@ -183,11 +189,13 @@ jobs:
           max-total-time: 120
 
   unstable-fuzztest:
+    if: ${{ github.repository == 'GreptimeTeam/greptimedb' }}
     name: Unstable Fuzz Test
     needs: build-greptime-ci
     runs-on: ubuntu-latest
     timeout-minutes: 60
     strategy:
+      fail-fast: false
       matrix:
         target: [ "unstable_fuzz_create_table_standalone" ]
     steps:
@@ -215,12 +223,12 @@ jobs:
         run: |
           sudo apt update && sudo apt install -y libfuzzer-14-dev
           cargo install cargo-fuzz cargo-gc-bin --force
-      - name: Download pre-built binariy
+      - name: Download pre-built binary
         uses: actions/download-artifact@v4
         with:
           name: bin
           path: .
-      - name: Unzip bianry
+      - name: Unzip binary
         run: |
           tar -xvf ./bin.tar.gz
           rm ./bin.tar.gz
@@ -244,6 +252,7 @@ jobs:
           retention-days: 3
 
   build-greptime-ci:
+    if: ${{ github.repository == 'GreptimeTeam/greptimedb' }}
     name: Build GreptimeDB binary (profile-CI)
     runs-on: ${{ matrix.os }}
     strategy:
@@ -267,7 +276,7 @@ jobs:
       - name: Install cargo-gc-bin
         shell: bash
         run: cargo install cargo-gc-bin --force
-      - name: Build greptime bianry
+      - name: Build greptime binary
         shell: bash
         # `cargo gc` will invoke `cargo build` with specified args
         run: cargo gc --profile ci -- --bin greptime --features "pg_kvbackend,mysql_kvbackend"
@@ -285,11 +294,13 @@ jobs:
           version: current
 
   distributed-fuzztest:
+    if: ${{ github.repository == 'GreptimeTeam/greptimedb' }}
     name: Fuzz Test (Distributed, ${{ matrix.mode.name }}, ${{ matrix.target }})
     runs-on: ubuntu-latest
     needs:  build-greptime-ci
     timeout-minutes: 60
     strategy:
+      fail-fast: false
       matrix:
         target: [ "fuzz_create_table", "fuzz_alter_table", "fuzz_create_database", "fuzz_create_logical_table", "fuzz_alter_logical_table", "fuzz_insert", "fuzz_insert_logical_table" ]
         mode:
@@ -319,9 +330,9 @@ jobs:
         name: Setup Minio
         uses: ./.github/actions/setup-minio
       - if: matrix.mode.kafka
-        name: Setup Kafka cluser
+        name: Setup Kafka cluster
         uses: ./.github/actions/setup-kafka-cluster
-      - name: Setup Etcd cluser
+      - name: Setup Etcd cluster
         uses: ./.github/actions/setup-etcd-cluster
       # Prepares for fuzz tests
       - uses: arduino/setup-protoc@v3
@@ -416,11 +427,13 @@ jobs:
           docker system prune -f
 
   distributed-fuzztest-with-chaos:
+    if: ${{ github.repository == 'GreptimeTeam/greptimedb' }}
     name: Fuzz Test with Chaos (Distributed, ${{ matrix.mode.name }}, ${{ matrix.target }})
     runs-on: ubuntu-latest
     needs:  build-greptime-ci
     timeout-minutes: 60
     strategy:
+      fail-fast: false
       matrix:
         target: ["fuzz_migrate_mito_regions", "fuzz_migrate_metric_regions", "fuzz_failover_mito_regions", "fuzz_failover_metric_regions"]
         mode:
@@ -465,9 +478,9 @@ jobs:
         name: Setup Minio
         uses: ./.github/actions/setup-minio
       - if: matrix.mode.kafka
-        name: Setup Kafka cluser
+        name: Setup Kafka cluster
         uses: ./.github/actions/setup-kafka-cluster
-      - name: Setup Etcd cluser
+      - name: Setup Etcd cluster
         uses: ./.github/actions/setup-etcd-cluster
       # Prepares for fuzz tests
       - uses: arduino/setup-protoc@v3
@@ -563,10 +576,12 @@ jobs:
           docker system prune -f
 
   sqlness:
+    if: ${{ github.repository == 'GreptimeTeam/greptimedb' }}
     name: Sqlness Test (${{ matrix.mode.name }})
     needs: build
     runs-on: ${{ matrix.os }}
     strategy:
+      fail-fast: false
       matrix:
         os: [ ubuntu-latest ]
         mode:
@@ -609,6 +624,7 @@ jobs:
           retention-days: 3
 
   fmt:
+    if: ${{ github.repository == 'GreptimeTeam/greptimedb' }}
     name: Rustfmt
     runs-on: ubuntu-latest
     timeout-minutes: 60
@@ -626,6 +642,7 @@ jobs:
         run: make fmt-check
 
   clippy:
+    if: ${{ github.repository == 'GreptimeTeam/greptimedb' }}
     name: Clippy
     runs-on: ubuntu-latest
     timeout-minutes: 60
@@ -651,6 +668,7 @@ jobs:
         run: make clippy
 
   conflict-check:
+    if: ${{ github.repository == 'GreptimeTeam/greptimedb' }}
     name: Check for conflict
     runs-on: ubuntu-latest
     steps:
@@ -661,7 +679,7 @@ jobs:
         uses: olivernybroe/action-conflict-finder@v4.0
 
   test:
-    if: github.event_name != 'merge_group'
+    if: ${{ github.repository == 'GreptimeTeam/greptimedb' && github.event_name != 'merge_group' }}
     runs-on: ubuntu-22.04-arm
     timeout-minutes: 60
     needs:  [conflict-check, clippy, fmt]
@@ -713,7 +731,7 @@ jobs:
           UNITTEST_LOG_DIR: "__unittest_logs"
 
   coverage:
-    if: github.event_name == 'merge_group'
+    if: ${{ github.repository == 'GreptimeTeam/greptimedb' && github.event_name == 'merge_group' }}
     runs-on: ubuntu-22.04-8-cores
     timeout-minutes: 60
     steps:
@@ -773,6 +791,7 @@ jobs:
           verbose: true
 
   # compat:
+  #   if: ${{ github.repository == 'GreptimeTeam/greptimedb' }}
   #   name: Compatibility Test
   #   needs: build
   #   runs-on: ubuntu-22.04

.github/workflows/nightly-ci.yml

@@ -117,16 +117,16 @@ jobs:
     name: Run clean build on Linux
     runs-on: ubuntu-latest
     if: ${{ github.repository == 'GreptimeTeam/greptimedb' }}
-    timeout-minutes: 60
+    timeout-minutes: 45
     steps:
       - uses: actions/checkout@v4
         with:
           fetch-depth: 0
           persist-credentials: false
-      - uses: cachix/install-nix-action@v27
-        with:
-          nix_path: nixpkgs=channel:nixos-24.11
-      - run: nix develop --command cargo build
+      - uses: cachix/install-nix-action@v31
+      - run: nix develop --command cargo check --bin greptime
+        env:
+          CARGO_BUILD_RUSTFLAGS: "-C link-arg=-fuse-ld=mold"
 
   check-status:
     name: Check status

.github/workflows/release-dev-builder-images.yaml

@@ -24,11 +24,19 @@ on:
         description: Release dev-builder-android image
         required: false
         default: false
+      update_dev_builder_image_tag:
+        type: boolean
+        description: Update the DEV_BUILDER_IMAGE_TAG in Makefile and create a PR
+        required: false
+        default: false
 
 jobs:
   release-dev-builder-images:
     name: Release dev builder images
-    if: ${{ inputs.release_dev_builder_ubuntu_image || inputs.release_dev_builder_centos_image || inputs.release_dev_builder_android_image }} # Only manually trigger this job.
+    # The jobs are triggered by the following events:
+    # 1. Manually triggered workflow_dispatch event
+    # 2. Push event when the PR that modifies the `rust-toolchain.toml` or `docker/dev-builder/**` is merged to main
+    if: ${{ github.event_name == 'push' || inputs.release_dev_builder_ubuntu_image || inputs.release_dev_builder_centos_image || inputs.release_dev_builder_android_image }}
     runs-on: ubuntu-latest
     outputs:
       version: ${{ steps.set-version.outputs.version }}
@@ -57,9 +65,9 @@ jobs:
           version: ${{ env.VERSION }}
           dockerhub-image-registry-username: ${{ secrets.DOCKERHUB_USERNAME }}
           dockerhub-image-registry-token: ${{ secrets.DOCKERHUB_TOKEN }}
-          build-dev-builder-ubuntu: ${{ inputs.release_dev_builder_ubuntu_image }}
-          build-dev-builder-centos: ${{ inputs.release_dev_builder_centos_image }}
-          build-dev-builder-android: ${{ inputs.release_dev_builder_android_image }}
+          build-dev-builder-ubuntu: ${{ inputs.release_dev_builder_ubuntu_image || github.event_name == 'push' }}
+          build-dev-builder-centos: ${{ inputs.release_dev_builder_centos_image || github.event_name == 'push' }}
+          build-dev-builder-android: ${{ inputs.release_dev_builder_android_image || github.event_name == 'push' }}
 
   release-dev-builder-images-ecr:
     name: Release dev builder images to AWS ECR
@@ -85,7 +93,7 @@ jobs:
 
       - name: Push dev-builder-ubuntu image
         shell: bash
-        if: ${{ inputs.release_dev_builder_ubuntu_image }}
+        if: ${{ inputs.release_dev_builder_ubuntu_image || github.event_name == 'push' }}
         env:
           IMAGE_VERSION: ${{ needs.release-dev-builder-images.outputs.version }}
           IMAGE_NAMESPACE: ${{ vars.IMAGE_NAMESPACE }}
@@ -106,7 +114,7 @@ jobs:
 
       - name: Push dev-builder-centos image
         shell: bash
-        if: ${{ inputs.release_dev_builder_centos_image }}
+        if: ${{ inputs.release_dev_builder_centos_image || github.event_name == 'push' }}
         env:
           IMAGE_VERSION: ${{ needs.release-dev-builder-images.outputs.version }}
           IMAGE_NAMESPACE: ${{ vars.IMAGE_NAMESPACE }}
@@ -127,7 +135,7 @@ jobs:
 
       - name: Push dev-builder-android image
         shell: bash
-        if: ${{ inputs.release_dev_builder_android_image }}
+        if: ${{ inputs.release_dev_builder_android_image || github.event_name == 'push' }}
         env:
           IMAGE_VERSION: ${{ needs.release-dev-builder-images.outputs.version }}
           IMAGE_NAMESPACE: ${{ vars.IMAGE_NAMESPACE }}
@@ -162,7 +170,7 @@ jobs:
 
       - name: Push dev-builder-ubuntu image
         shell: bash
-        if: ${{ inputs.release_dev_builder_ubuntu_image }}
+        if: ${{ inputs.release_dev_builder_ubuntu_image || github.event_name == 'push' }}
         env:
           IMAGE_VERSION: ${{ needs.release-dev-builder-images.outputs.version }}
           IMAGE_NAMESPACE: ${{ vars.IMAGE_NAMESPACE }}
@@ -176,7 +184,7 @@ jobs:
 
       - name: Push dev-builder-centos image
         shell: bash
-        if: ${{ inputs.release_dev_builder_centos_image }}
+        if: ${{ inputs.release_dev_builder_centos_image || github.event_name == 'push' }}
         env:
           IMAGE_VERSION: ${{ needs.release-dev-builder-images.outputs.version }}
           IMAGE_NAMESPACE: ${{ vars.IMAGE_NAMESPACE }}
@@ -190,7 +198,7 @@ jobs:
 
       - name: Push dev-builder-android image
         shell: bash
-        if: ${{ inputs.release_dev_builder_android_image }}
+        if: ${{ inputs.release_dev_builder_android_image || github.event_name == 'push' }}
         env:
           IMAGE_VERSION: ${{ needs.release-dev-builder-images.outputs.version }}
           IMAGE_NAMESPACE: ${{ vars.IMAGE_NAMESPACE }}
@@ -201,3 +209,24 @@ jobs:
             quay.io/skopeo/stable:latest \
             copy -a docker://docker.io/$IMAGE_NAMESPACE/dev-builder-android:$IMAGE_VERSION \
             docker://$ACR_IMAGE_REGISTRY/$IMAGE_NAMESPACE/dev-builder-android:$IMAGE_VERSION
+  
+  update-dev-builder-image-tag:
+    name: Update dev-builder image tag
+    runs-on: ubuntu-latest
+    permissions:
+      contents: write
+      pull-requests: write
+    if: ${{ github.event_name == 'push' || inputs.update_dev_builder_image_tag }}
+    needs: [
+      release-dev-builder-images
+    ]
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v4
+
+      - name: Update dev-builder image tag
+        shell: bash
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+        run: |
+          ./.github/scripts/update-dev-builder-version.sh ${{ needs.release-dev-builder-images.outputs.version }}

.github/workflows/release.yml

@@ -88,10 +88,8 @@ env:
   # Controls whether to run tests, include unit-test, integration-test and sqlness.
   DISABLE_RUN_TESTS: ${{ inputs.skip_test || vars.DEFAULT_SKIP_TEST }}
 
-  # The scheduled version is '${{ env.NEXT_RELEASE_VERSION }}-nightly-YYYYMMDD', like v0.2.0-nigthly-20230313;
+  # The scheduled version is '${{ env.NEXT_RELEASE_VERSION }}-nightly-YYYYMMDD', like v0.2.0-nightly-20230313;
   NIGHTLY_RELEASE_PREFIX: nightly
-  # Note: The NEXT_RELEASE_VERSION should be modified manually by every formal release.
-  NEXT_RELEASE_VERSION: v0.14.0
 
 jobs:
   allocate-runners:
@@ -126,7 +124,7 @@ jobs:
 
       # The create-version will create a global variable named 'version' in the global workflows.
       # - If it's a tag push release, the version is the tag name(${{ github.ref_name }});
-      # - If it's a scheduled release, the version is '${{ env.NEXT_RELEASE_VERSION }}-nightly-$buildTime', like v0.2.0-nigthly-20230313;
+      # - If it's a scheduled release, the version is '${{ env.NEXT_RELEASE_VERSION }}-nightly-$buildTime', like v0.2.0-nightly-20230313;
       # - If it's a manual release, the version is '${{ env.NEXT_RELEASE_VERSION }}-<short-git-sha>-YYYYMMDDSS', like v0.2.0-e5b243c-2023071245;
       - name: Create version
         id: create-version
@@ -135,7 +133,6 @@ jobs:
         env:
           GITHUB_EVENT_NAME: ${{ github.event_name }}
           GITHUB_REF_NAME: ${{ github.ref_name }}
-          NEXT_RELEASE_VERSION: ${{ env.NEXT_RELEASE_VERSION }}
           NIGHTLY_RELEASE_PREFIX: ${{ env.NIGHTLY_RELEASE_PREFIX }}
 
       - name: Allocate linux-amd64 runner
@@ -391,7 +388,7 @@ jobs:
 
   ### Stop runners ###
   # It's very necessary to split the job of releasing runners into 'stop-linux-amd64-runner' and 'stop-linux-arm64-runner'.
-  # Because we can terminate the specified EC2 instance immediately after the job is finished without uncessary waiting.
+  # Because we can terminate the specified EC2 instance immediately after the job is finished without unnecessary waiting.
   stop-linux-amd64-runner: # It's always run as the last job in the workflow to make sure that the runner is released.
     name: Stop linux-amd64 runner
     # Only run this job when the runner is allocated.
@@ -447,7 +444,7 @@ jobs:
   bump-doc-version:
     name: Bump doc version
     if: ${{ github.event_name == 'push' || github.event_name == 'schedule' }}
-    needs: [allocate-runners]
+    needs: [allocate-runners, publish-github-release]
     runs-on: ubuntu-latest
     # Permission reference: https://docs.github.com/en/actions/using-jobs/assigning-permissions-to-jobs
     permissions:
@@ -467,6 +464,71 @@ jobs:
           GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
           DOCS_REPO_TOKEN: ${{ secrets.DOCS_REPO_TOKEN }}
 
+  bump-website-version:
+    name: Bump website version
+    if: ${{ github.ref_type == 'tag' && !contains(github.ref_name, 'nightly') && github.event_name != 'schedule' }}
+    needs: [allocate-runners, publish-github-release]
+    runs-on: ubuntu-latest
+    # Permission reference: https://docs.github.com/en/actions/using-jobs/assigning-permissions-to-jobs
+    permissions:
+      issues: write # Allows the action to create issues for cyborg.
+      contents: write # Allows the action to create a release.
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+          persist-credentials: false
+      - uses: ./.github/actions/setup-cyborg
+      - name: Bump website version
+        working-directory: cyborg
+        run: pnpm tsx bin/bump-website-version.ts
+        env:
+          VERSION: ${{ needs.allocate-runners.outputs.version }}
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          WEBSITE_REPO_TOKEN: ${{ secrets.WEBSITE_REPO_TOKEN }}
+
+  bump-helm-charts-version:
+    name: Bump helm charts version
+    if: ${{ github.ref_type == 'tag' && !contains(github.ref_name, 'nightly') && github.event_name != 'schedule' }}
+    needs: [allocate-runners, publish-github-release]
+    runs-on: ubuntu-latest
+    permissions:
+      contents: write
+      pull-requests: write
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+
+      - name: Bump helm charts version
+        env:
+          GITHUB_TOKEN: ${{ secrets.HELM_CHARTS_REPO_TOKEN }}
+          VERSION: ${{ needs.allocate-runners.outputs.version }}
+        run: |
+          ./.github/scripts/update-helm-charts-version.sh
+
+  bump-homebrew-greptime-version:
+    name: Bump homebrew greptime version
+    if: ${{ github.ref_type == 'tag' && !contains(github.ref_name, 'nightly') && github.event_name != 'schedule' }}
+    needs: [allocate-runners, publish-github-release]
+    runs-on: ubuntu-latest
+    permissions:
+      contents: write
+      pull-requests: write
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+
+      - name: Bump homebrew greptime version
+        env:
+          GITHUB_TOKEN: ${{ secrets.HOMEBREW_GREPTIME_REPO_TOKEN }}
+          VERSION: ${{ needs.allocate-runners.outputs.version }}
+        run: |
+          ./.github/scripts/update-homebrew-greptme-version.sh
+
   notification:
     if: ${{ github.repository == 'GreptimeTeam/greptimedb' && (github.event_name == 'push' || github.event_name == 'schedule') && always() }}
     name: Send notification to Greptime team

.github/workflows/semantic-pull-request.yml

@@ -14,6 +14,9 @@ concurrency:
 jobs:
   check:
     runs-on: ubuntu-latest
+    permissions:
+      pull-requests: write  # Add permissions to modify PRs
+      issues: write
     timeout-minutes: 10
     steps:
       - uses: actions/checkout@v4

.gitignore

@@ -28,6 +28,7 @@ debug/
 # Logs
 **/__unittest_logs
 logs/
+!grafana/dashboards/logs/
 
 # cpython's generated python byte code
 **/__pycache__/
@@ -57,3 +58,6 @@ tests-fuzz/corpus/
 
 ## default data home
 greptimedb_data
+
+# github
+!/.github

CONTRIBUTING.md

@@ -108,7 +108,7 @@ of what you were trying to do and what went wrong. You can also reach for help i
 The core team will be thrilled if you would like to participate in any way you like. When you are stuck, try to ask for help by filing an issue, with a detailed description of what you were trying to do and what went wrong. If you have any questions or if you would like to get involved in our community, please check out:
 
 - [GreptimeDB Community Slack](https://greptime.com/slack)
-- [GreptimeDB Github Discussions](https://github.com/GreptimeTeam/greptimedb/discussions)
+- [GreptimeDB GitHub Discussions](https://github.com/GreptimeTeam/greptimedb/discussions)
 
 Also, see some extra GreptimeDB content:
 

Cargo.toml

@@ -30,12 +30,14 @@ members = [
     "src/common/recordbatch",
     "src/common/runtime",
     "src/common/session",
+    "src/common/stat",
     "src/common/substrait",
     "src/common/telemetry",
     "src/common/test-util",
     "src/common/time",
     "src/common/version",
     "src/common/wal",
+    "src/common/workload",
     "src/datanode",
     "src/datatypes",
     "src/file-engine",
@@ -68,15 +70,17 @@ members = [
 resolver = "2"
 
 [workspace.package]
-version = "0.14.0"
+version = "0.15.0"
 edition = "2021"
 license = "Apache-2.0"
 
 [workspace.lints]
-clippy.print_stdout = "warn"
-clippy.print_stderr = "warn"
 clippy.dbg_macro = "warn"
 clippy.implicit_clone = "warn"
+clippy.result_large_err = "allow"
+clippy.large_enum_variant = "allow"
+clippy.doc_overindented_list_items = "allow"
+clippy.uninlined_format_args = "allow"
 rust.unknown_lints = "deny"
 rust.unexpected_cfgs = { level = "warn", check-cfg = ['cfg(tokio_unstable)'] }
 
@@ -129,7 +133,7 @@ etcd-client = "0.14"
 fst = "0.4.7"
 futures = "0.3"
 futures-util = "0.3"
-greptime-proto = { git = "https://github.com/GreptimeTeam/greptime-proto.git", rev = "e82b0158cd38d4021edb4e4c0ae77f999051e62f" }
+greptime-proto = { git = "https://github.com/GreptimeTeam/greptime-proto.git", rev = "454c52634c3bac27de10bf0d85d5533eed1cf03f" }
 hex = "0.4"
 http = "1"
 humantime = "2.1"
@@ -145,6 +149,7 @@ meter-core = { git = "https://github.com/GreptimeTeam/greptime-meter.git", rev =
 mockall = "0.13"
 moka = "0.12"
 nalgebra = "0.33"
+nix = { version = "0.30.1", default-features = false, features = ["event", "fs", "process"] }
 notify = "8.0"
 num_cpus = "1.16"
 object_store_opendal = "0.50"
@@ -161,8 +166,10 @@ parquet = { version = "54.2", default-features = false, features = ["arrow", "as
 paste = "1.0"
 pin-project = "1.0"
 prometheus = { version = "0.13.3", features = ["process"] }
-promql-parser = { version = "0.5.1", features = ["ser"] }
-prost = "0.13"
+promql-parser = { git = "https://github.com/GreptimeTeam/promql-parser.git", rev = "0410e8b459dda7cb222ce9596f8bf3971bd07bd2", features = [
+    "ser",
+] }
+prost = { version = "0.13", features = ["no-recursion-limit"] }
 raft-engine = { version = "0.4.1", default-features = false }
 rand = "0.9"
 ratelimit = "0.10"
@@ -174,7 +181,7 @@ reqwest = { version = "0.12", default-features = false, features = [
     "stream",
     "multipart",
 ] }
-rskafka = { git = "https://github.com/influxdata/rskafka.git", rev = "75535b5ad9bae4a5dbb582c82e44dfd81ec10105", features = [
+rskafka = { git = "https://github.com/influxdata/rskafka.git", rev = "8dbd01ed809f5a791833a594e85b144e36e45820", features = [
     "transport-tls",
 ] }
 rstest = "0.25"
@@ -255,6 +262,7 @@ common-test-util = { path = "src/common/test-util" }
 common-time = { path = "src/common/time" }
 common-version = { path = "src/common/version" }
 common-wal = { path = "src/common/wal" }
+common-workload = { path = "src/common/workload" }
 datanode = { path = "src/datanode" }
 datatypes = { path = "src/datatypes" }
 file-engine = { path = "src/file-engine" }
@@ -281,6 +289,7 @@ query = { path = "src/query" }
 servers = { path = "src/servers" }
 session = { path = "src/session" }
 sql = { path = "src/sql" }
+stat = { path = "src/common/stat" }
 store-api = { path = "src/store-api" }
 substrait = { path = "src/common/substrait" }
 table = { path = "src/table" }

Makefile

@@ -8,7 +8,7 @@ CARGO_BUILD_OPTS := --locked
 IMAGE_REGISTRY ?= docker.io
 IMAGE_NAMESPACE ?= greptime
 IMAGE_TAG ?= latest
-DEV_BUILDER_IMAGE_TAG ?= 2024-12-25-a71b93dd-20250305072908
+DEV_BUILDER_IMAGE_TAG ?= 2025-05-19-b2377d4b-20250520045554
 BUILDX_MULTI_PLATFORM_BUILD ?= false
 BUILDX_BUILDER_NAME ?= gtbuilder
 BASE_IMAGE ?= ubuntu

README.md

@@ -8,6 +8,8 @@
 
 <h2 align="center">Real-Time & Cloud-Native Observability  Database<br/>for metrics, logs, and traces</h2>
 
+>  Delivers sub-second querying at PB scale and exceptional cost efficiency from edge to cloud.
+
 <div align="center">
 <h3 align="center">
   <a href="https://greptime.com/product/cloud">GreptimeCloud</a> |
@@ -49,74 +51,77 @@
 </div>
 
 - [Introduction](#introduction)
-- [**Features: Why GreptimeDB**](#why-greptimedb)
-- [Architecture](https://docs.greptime.com/contributor-guide/overview/#architecture)
-- [Try it for free](#try-greptimedb)
+- [⭐ Key Features](#features)
+- [Quick Comparison](#quick-comparison)
+- [Architecture](#architecture)
+- [Try GreptimeDB](#try-greptimedb)
 - [Getting Started](#getting-started)
-- [Project Status](#project-status)
-- [Join the community](#community)
-  - [Contributing](#contributing)
+- [Build From Source](#build-from-source)
 - [Tools & Extensions](#tools--extensions)
+- [Project Status](#project-status)
+- [Community](#community)
 - [License](#license)
+- [Commercial Support](#commercial-support)
+- [Contributing](#contributing)
 - [Acknowledgement](#acknowledgement)
 
 ## Introduction
 
-**GreptimeDB** is an open-source, cloud-native, unified & cost-effective observability database for **Metrics**, **Logs**, and **Traces**. You can gain real-time insights from Edge to Cloud at Any Scale.
+**GreptimeDB** is an open-source, cloud-native database purpose-built for the unified collection and analysis of observability data (metrics, logs, and traces). Whether you’re operating on the edge, in the cloud, or across hybrid environments, GreptimeDB empowers real-time insights at massive scale — all in one system.
 
-## News
+## Features
 
-**[GreptimeDB tops JSONBench's billion-record cold run test!](https://greptime.com/blogs/2025-03-18-jsonbench-greptimedb-performance)**
+|   Feature  | Description |
+| --------- | ----------- |
+| [Unified Observability Data](https://docs.greptime.com/user-guide/concepts/why-greptimedb) | Store metrics, logs, and traces as timestamped, contextual wide events. Query via [SQL](https://docs.greptime.com/user-guide/query-data/sql), [PromQL](https://docs.greptime.com/user-guide/query-data/promql), and [streaming](https://docs.greptime.com/user-guide/flow-computation/overview). |
+| [High Performance & Cost Effective](https://docs.greptime.com/user-guide/manage-data/data-index) | Written in Rust, with a distributed query engine, [rich indexing](https://docs.greptime.com/user-guide/manage-data/data-index), and optimized columnar storage, delivering sub-second responses at PB scale. |
+| [Cloud-Native Architecture](https://docs.greptime.com/user-guide/concepts/architecture) | Designed for [Kubernetes](https://docs.greptime.com/user-guide/deployments/deploy-on-kubernetes/greptimedb-operator-management), with compute/storage separation, native object storage (AWS S3, Azure Blob, etc.) and seamless cross-cloud access. |
+| [Developer-Friendly](https://docs.greptime.com/user-guide/protocols/overview) | Access via SQL/PromQL interfaces, REST API, MySQL/PostgreSQL protocols, and popular ingestion [protocols](https://docs.greptime.com/user-guide/protocols/overview). |
+| [Flexible Deployment](https://docs.greptime.com/user-guide/deployments/overview) | Deploy anywhere: edge (including ARM/[Android](https://docs.greptime.com/user-guide/deployments/run-on-android)) or cloud, with unified APIs and efficient data sync. |
 
-## Why GreptimeDB
+Learn more in [Why GreptimeDB](https://docs.greptime.com/user-guide/concepts/why-greptimedb) and [Observability 2.0 and the Database for It](https://greptime.com/blogs/2025-04-25-greptimedb-observability2-new-database).
 
-Our core developers have been building observability data platforms for years. Based on our best practices, GreptimeDB was born to give you:
+## Quick Comparison
 
-* **Unified Processing of Observability Data**
+| Feature                         | GreptimeDB            | Traditional TSDB   | Log Stores      |
+|----------------------------------|-----------------------|--------------------|-----------------|
+| Data Types                      | Metrics, Logs, Traces | Metrics only       | Logs only       |
+| Query Language                  | SQL, PromQL, Streaming|  Custom/PromQL     | Custom/DSL      |
+| Deployment                      | Edge + Cloud          | Cloud/On-prem      | Mostly central  |
+| Indexing & Performance          | PB-Scale, Sub-second  | Varies             | Varies          |
+| Integration                     | REST, SQL, Common protocols | Varies     | Varies          |
 
-  A unified database that treats metrics, logs, and traces as timestamped wide events with context, supporting [SQL](https://docs.greptime.com/user-guide/query-data/sql)/[PromQL](https://docs.greptime.com/user-guide/query-data/promql) queries and [stream processing](https://docs.greptime.com/user-guide/flow-computation/overview) to simplify complex data stacks.
+**Performance:**
+* [GreptimeDB tops JSONBench's billion-record cold run test!](https://greptime.com/blogs/2025-03-18-jsonbench-greptimedb-performance)
+* [TSBS Benchmark](https://github.com/GreptimeTeam/greptimedb/tree/main/docs/benchmarks/tsbs)
 
-* **High Performance and Cost-effective**
+Read [more benchmark reports](https://docs.greptime.com/user-guide/concepts/features-that-you-concern#how-is-greptimedbs-performance-compared-to-other-solutions).
 
-   Written in Rust, combines a distributed query engine with [rich indexing](https://docs.greptime.com/user-guide/manage-data/data-index) (inverted, fulltext, skip data, and vector) and optimized columnar storage to deliver sub-second responses on petabyte-scale data and high-cost efficiency.
+## Architecture
 
-* **Cloud-native Distributed Database**
-
-  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.
-
-* **Developer-Friendly**
-
-  Access standardized SQL/PromQL interfaces through built-in web dashboard, REST API, and MySQL/PostgreSQL protocols. Supports widely adopted data ingestion [protocols](https://docs.greptime.com/user-guide/protocols/overview) for seamless migration and integration.
-
-* **Flexible Deployment Options**
-
-  Deploy GreptimeDB anywhere from ARM-based edge devices to cloud environments with unified APIs and bandwidth-efficient data synchronization. Query edge and cloud data seamlessly through identical APIs. [Learn how to run on Android](https://docs.greptime.com/user-guide/deployments/run-on-android/).
-
-For more detailed info please read  [Why GreptimeDB](https://docs.greptime.com/user-guide/concepts/why-greptimedb).
+* Read the [architecture](https://docs.greptime.com/contributor-guide/overview/#architecture) document.
+* [DeepWiki](https://deepwiki.com/GreptimeTeam/greptimedb/1-overview) provides an in-depth look at GreptimeDB:
+  <img alt="GreptimeDB System Overview" src="docs/architecture.png">
 
 ## Try GreptimeDB
 
 ### 1. [Live Demo](https://greptime.com/playground)
 
-Try out the features of GreptimeDB right from your browser.
+Experience GreptimeDB directly in your browser.
 
 ### 2. [GreptimeCloud](https://console.greptime.cloud/)
 
 Start instantly with a free cluster.
 
-### 3. Docker Image
-
-To install GreptimeDB locally, the recommended way is via Docker:
+### 3. Docker (Local Quickstart)
 
 ```shell
 docker pull greptime/greptimedb
 ```
 
-Start a GreptimeDB container with:
-
 ```shell
 docker run -p 127.0.0.1:4000-4003:4000-4003 \
-  -v "$(pwd)/greptimedb:./greptimedb_data" \
+  -v "$(pwd)/greptimedb_data:/greptimedb_data" \
   --name greptime --rm \
   greptime/greptimedb:latest standalone start \
   --http-addr 0.0.0.0:4000 \
@@ -124,114 +129,89 @@ docker run -p 127.0.0.1:4000-4003:4000-4003 \
   --mysql-addr 0.0.0.0:4002 \
   --postgres-addr 0.0.0.0:4003
 ```
+Dashboard: [http://localhost:4000/dashboard](http://localhost:4000/dashboard)
+[Full Install Guide](https://docs.greptime.com/getting-started/installation/overview)
 
-Access the dashboard via `http://localhost:4000/dashboard`.
-
-Read more about [Installation](https://docs.greptime.com/getting-started/installation/overview) on docs.
+**Troubleshooting:**
+* Cannot connect to the database? Ensure that ports `4000`, `4001`, `4002`, and `4003` are not blocked by a firewall or used by other services.
+* Failed to start? Check the container logs with `docker logs greptime` for further details.
 
 ## Getting Started
 
-* [Quickstart](https://docs.greptime.com/getting-started/quick-start)
-* [User Guide](https://docs.greptime.com/user-guide/overview)
-* [Demos](https://github.com/GreptimeTeam/demo-scene)
-* [FAQ](https://docs.greptime.com/faq-and-others/faq)
+- [Quickstart](https://docs.greptime.com/getting-started/quick-start)
+- [User Guide](https://docs.greptime.com/user-guide/overview)
+- [Demo Scenes](https://github.com/GreptimeTeam/demo-scene)
+- [FAQ](https://docs.greptime.com/faq-and-others/faq)
 
-## Build
-
-Check the prerequisite:
+## Build From Source
 
+**Prerequisites:**
 * [Rust toolchain](https://www.rust-lang.org/tools/install) (nightly)
 * [Protobuf compiler](https://grpc.io/docs/protoc-installation/) (>= 3.15)
 * C/C++ building essentials, including `gcc`/`g++`/`autoconf` and glibc library (eg. `libc6-dev` on Ubuntu and `glibc-devel` on Fedora)
 * Python toolchain (optional): Required only if using some test scripts.
 
-Build GreptimeDB binary:
-
-```shell
+**Build and Run:**
+```bash
 make
-```
-
-Run a standalone server:
-
-```shell
 cargo run -- standalone start
 ```
 
 ## Tools & Extensions
 
-### Kubernetes
-
-- [GreptimeDB Operator](https://github.com/GrepTimeTeam/greptimedb-operator)
-
-### Dashboard
-
-- [The dashboard UI for GreptimeDB](https://github.com/GreptimeTeam/dashboard)
-
-### SDK
-
-- [GreptimeDB Go Ingester](https://github.com/GreptimeTeam/greptimedb-ingester-go)
-- [GreptimeDB Java Ingester](https://github.com/GreptimeTeam/greptimedb-ingester-java)
-- [GreptimeDB C++ Ingester](https://github.com/GreptimeTeam/greptimedb-ingester-cpp)
-- [GreptimeDB Erlang Ingester](https://github.com/GreptimeTeam/greptimedb-ingester-erl)
-- [GreptimeDB Rust Ingester](https://github.com/GreptimeTeam/greptimedb-ingester-rust)
-- [GreptimeDB JavaScript Ingester](https://github.com/GreptimeTeam/greptimedb-ingester-js)
-
-### Grafana Dashboard
-
-Our official Grafana dashboard for monitoring GreptimeDB is available at [grafana](grafana/README.md) directory.
+- **Kubernetes:** [GreptimeDB Operator](https://github.com/GrepTimeTeam/greptimedb-operator)
+- **Helm Charts:** [Greptime Helm Charts](https://github.com/GreptimeTeam/helm-charts)
+- **Dashboard:** [Web UI](https://github.com/GreptimeTeam/dashboard)
+- **SDKs/Ingester:** [Go](https://github.com/GreptimeTeam/greptimedb-ingester-go), [Java](https://github.com/GreptimeTeam/greptimedb-ingester-java), [C++](https://github.com/GreptimeTeam/greptimedb-ingester-cpp), [Erlang](https://github.com/GreptimeTeam/greptimedb-ingester-erl), [Rust](https://github.com/GreptimeTeam/greptimedb-ingester-rust), [JS](https://github.com/GreptimeTeam/greptimedb-ingester-js)
+- **Grafana**: [Official Dashboard](https://github.com/GreptimeTeam/greptimedb/blob/main/grafana/README.md)
 
 ## Project Status
 
-GreptimeDB is currently in Beta. We are targeting GA (General Availability) with v1.0 release by Early 2025.
+> **Status:** Beta.
+> **GA (v1.0):** Targeted for mid 2025.
 
-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
+- Being used in production by early adopters
+- Stable, actively maintained, with regular releases ([version info](https://docs.greptime.com/nightly/reference/about-greptimedb-version))
+- Suitable for evaluation and pilot deployments
 
 For production use, we recommend using the latest stable release.
+[![Star History Chart](https://api.star-history.com/svg?repos=GreptimeTeam/GreptimeDB&type=Date)](https://www.star-history.com/#GreptimeTeam/GreptimeDB&Date)
+
+If you find this project useful, a ⭐ would mean a lot to us!
+<img alt="Known Users" src="https://greptime.com/logo/img/users.png"/>
 
 ## Community
 
-Our core team is thrilled to see you participate in any ways you like. When you are stuck, try to
-ask for help by filling an issue with a detailed description of what you were trying to do
-and what went wrong. If you have any questions or if you would like to get involved in our
-community, please check out:
+We invite you to engage and contribute!
 
-- GreptimeDB Community on [Slack](https://greptime.com/slack)
-- GreptimeDB [GitHub Discussions forum](https://github.com/GreptimeTeam/greptimedb/discussions)
-- Greptime official [website](https://greptime.com)
-
-In addition, you may:
-
-- View our official [Blog](https://greptime.com/blogs/)
-- Connect us with [Linkedin](https://www.linkedin.com/company/greptime/)
-- Follow us on [Twitter](https://twitter.com/greptime)
-
-## Commercial Support
-
-If you are running GreptimeDB OSS in your organization, we offer additional
-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 commercial license.
+- [Slack](https://greptime.com/slack)
+- [Discussions](https://github.com/GreptimeTeam/greptimedb/discussions)
+- [Official Website](https://greptime.com/)
+- [Blog](https://greptime.com/blogs/)
+- [LinkedIn](https://www.linkedin.com/company/greptime/)
+- [Twitter](https://twitter.com/greptime)
 
 ## License
 
-GreptimeDB uses the [Apache License 2.0](https://apache.org/licenses/LICENSE-2.0.txt) to strike a balance between
-open contributions and allowing you to use the software however you want.
+GreptimeDB is licensed under the [Apache License 2.0](https://apache.org/licenses/LICENSE-2.0.txt).
+
+## Commercial Support
+
+Running GreptimeDB in your organization?
+We offer enterprise add-ons, services, training, and consulting.
+[Contact us](https://greptime.com/contactus) for details.
 
 ## Contributing
 
-Please refer to [contribution guidelines](CONTRIBUTING.md) and [internal concepts docs](https://docs.greptime.com/contributor-guide/overview.html) for more information.
+- Read our [Contribution Guidelines](https://github.com/GreptimeTeam/greptimedb/blob/main/CONTRIBUTING.md).
+- Explore [Internal Concepts](https://docs.greptime.com/contributor-guide/overview.html) and [DeepWiki](https://deepwiki.com/GreptimeTeam/greptimedb).
+- Pick up a [good first issue](https://github.com/GreptimeTeam/greptimedb/issues?q=is%3Aissue+is%3Aopen+label%3A%22good+first+issue%22) and join the #contributors [Slack](https://greptime.com/slack) channel.
 
 ## Acknowledgement
 
-Special thanks to all the contributors who have propelled GreptimeDB forward. For a complete list of contributors, please refer to [AUTHOR.md](AUTHOR.md).
+Special thanks to all contributors! See [AUTHORS.md](https://github.com/GreptimeTeam/greptimedb/blob/main/AUTHOR.md).
 
-- GreptimeDB uses [Apache Arrow™](https://arrow.apache.org/) as the memory model and [Apache Parquet™](https://parquet.apache.org/) as the persistent file format.
-- GreptimeDB's query engine is powered by [Apache Arrow DataFusion™](https://arrow.apache.org/datafusion/).
-- [Apache OpenDAL™](https://opendal.apache.org) gives GreptimeDB a very general and elegant data access abstraction layer.
-- GreptimeDB's meta service is based on [etcd](https://etcd.io/).
-
-<img alt="Known Users" src="https://greptime.com/logo/img/users.png"/>
+- Uses [Apache Arrow™](https://arrow.apache.org/) (memory model)
+- [Apache Parquet™](https://parquet.apache.org/) (file storage)
+- [Apache Arrow DataFusion™](https://arrow.apache.org/datafusion/) (query engine)
+- [Apache OpenDAL™](https://opendal.apache.org/) (data access abstraction)

config/config.md

@@ -27,6 +27,7 @@
 | `http.body_limit` | String | `64MB` | HTTP request body limit.<br/>The following units are supported: `B`, `KB`, `KiB`, `MB`, `MiB`, `GB`, `GiB`, `TB`, `TiB`, `PB`, `PiB`.<br/>Set to 0 to disable limit. |
 | `http.enable_cors` | Bool | `true` | HTTP CORS support, it's turned on by default<br/>This allows browser to access http APIs without CORS restrictions |
 | `http.cors_allowed_origins` | Array | Unset | Customize allowed origins for HTTP CORS. |
+| `http.prom_validation_mode` | String | `strict` | Whether to enable validation for Prometheus remote write requests.<br/>Available options:<br/>- strict: deny invalid UTF-8 strings (default).<br/>- lossy: allow invalid UTF-8 strings, replace invalid characters with REPLACEMENT_CHARACTER(U+FFFD).<br/>- unchecked: do not valid strings. |
 | `grpc` | -- | -- | The gRPC server options. |
 | `grpc.bind_addr` | String | `127.0.0.1:4001` | The address to bind the gRPC server. |
 | `grpc.runtime_size` | Integer | `8` | The number of server worker threads. |
@@ -154,6 +155,7 @@
 | `region_engine.mito.index.metadata_cache_size` | String | `64MiB` | Cache size for inverted index metadata. |
 | `region_engine.mito.index.content_cache_size` | String | `128MiB` | Cache size for inverted index content. |
 | `region_engine.mito.index.content_cache_page_size` | String | `64KiB` | Page size for inverted index content cache. |
+| `region_engine.mito.index.result_cache_size` | String | `128MiB` | Cache size for index result. |
 | `region_engine.mito.inverted_index` | -- | -- | The options for inverted index in Mito engine. |
 | `region_engine.mito.inverted_index.create_on_flush` | String | `auto` | Whether to create the index on flush.<br/>- `auto`: automatically (default)<br/>- `disable`: never |
 | `region_engine.mito.inverted_index.create_on_compaction` | String | `auto` | Whether to create the index on compaction.<br/>- `auto`: automatically (default)<br/>- `disable`: never |
@@ -188,10 +190,11 @@
 | `logging.max_log_files` | Integer | `720` | The maximum amount of log files. |
 | `logging.tracing_sample_ratio` | -- | -- | The percentage of tracing will be sampled and exported.<br/>Valid range `[0, 1]`, 1 means all traces are sampled, 0 means all traces are not sampled, the default value is 1.<br/>ratio > 1 are treated as 1. Fractions < 0 are treated as 0 |
 | `logging.tracing_sample_ratio.default_ratio` | Float | `1.0` | -- |
-| `logging.slow_query` | -- | -- | The slow query log options. |
-| `logging.slow_query.enable` | Bool | `false` | Whether to enable slow query log. |
-| `logging.slow_query.threshold` | String | Unset | The threshold of slow query. |
-| `logging.slow_query.sample_ratio` | Float | Unset | The sampling ratio of slow query log. The value should be in the range of (0, 1]. |
+| `slow_query` | -- | -- | The slow query log options. |
+| `slow_query.enable` | Bool | `false` | Whether to enable slow query log. |
+| `slow_query.record_type` | String | Unset | The record type of slow queries. It can be `system_table` or `log`. |
+| `slow_query.threshold` | String | Unset | The threshold of slow query. |
+| `slow_query.sample_ratio` | Float | Unset | The sampling ratio of slow query log. The value should be in the range of (0, 1]. |
 | `export_metrics` | -- | -- | The datanode can export its metrics and send to Prometheus compatible service (e.g. send to `greptimedb` itself) from remote-write API.<br/>This is only used for `greptimedb` to export its own metrics internally. It's different from prometheus scrape. |
 | `export_metrics.enable` | Bool | `false` | whether enable export metrics. |
 | `export_metrics.write_interval` | String | `30s` | The interval of export metrics. |
@@ -224,6 +227,7 @@
 | `http.body_limit` | String | `64MB` | HTTP request body limit.<br/>The following units are supported: `B`, `KB`, `KiB`, `MB`, `MiB`, `GB`, `GiB`, `TB`, `TiB`, `PB`, `PiB`.<br/>Set to 0 to disable limit. |
 | `http.enable_cors` | Bool | `true` | HTTP CORS support, it's turned on by default<br/>This allows browser to access http APIs without CORS restrictions |
 | `http.cors_allowed_origins` | Array | Unset | Customize allowed origins for HTTP CORS. |
+| `http.prom_validation_mode` | String | `strict` | Whether to enable validation for Prometheus remote write requests.<br/>Available options:<br/>- strict: deny invalid UTF-8 strings (default).<br/>- lossy: allow invalid UTF-8 strings, replace invalid characters with REPLACEMENT_CHARACTER(U+FFFD).<br/>- unchecked: do not valid strings. |
 | `grpc` | -- | -- | The gRPC server options. |
 | `grpc.bind_addr` | String | `127.0.0.1:4001` | The address to bind the gRPC server. |
 | `grpc.server_addr` | String | `127.0.0.1:4001` | The address advertised to the metasrv, and used for connections from outside the host.<br/>If left empty or unset, the server will automatically use the IP address of the first network interface<br/>on the host, with the same port number as the one specified in `grpc.bind_addr`. |
@@ -288,10 +292,12 @@
 | `logging.max_log_files` | Integer | `720` | The maximum amount of log files. |
 | `logging.tracing_sample_ratio` | -- | -- | The percentage of tracing will be sampled and exported.<br/>Valid range `[0, 1]`, 1 means all traces are sampled, 0 means all traces are not sampled, the default value is 1.<br/>ratio > 1 are treated as 1. Fractions < 0 are treated as 0 |
 | `logging.tracing_sample_ratio.default_ratio` | Float | `1.0` | -- |
-| `logging.slow_query` | -- | -- | The slow query log options. |
-| `logging.slow_query.enable` | Bool | `false` | Whether to enable slow query log. |
-| `logging.slow_query.threshold` | String | Unset | The threshold of slow query. |
-| `logging.slow_query.sample_ratio` | Float | Unset | The sampling ratio of slow query log. The value should be in the range of (0, 1]. |
+| `slow_query` | -- | -- | The slow query log options. |
+| `slow_query.enable` | Bool | `true` | Whether to enable slow query log. |
+| `slow_query.record_type` | String | `system_table` | The record type of slow queries. It can be `system_table` or `log`.<br/>If `system_table` is selected, the slow queries will be recorded in a system table `greptime_private.slow_queries`.<br/>If `log` is selected, the slow queries will be logged in a log file `greptimedb-slow-queries.*`. |
+| `slow_query.threshold` | String | `30s` | The threshold of slow query. It can be human readable time string, for example: `10s`, `100ms`, `1s`. |
+| `slow_query.sample_ratio` | Float | `1.0` | The sampling ratio of slow query log. The value should be in the range of (0, 1]. For example, `0.1` means 10% of the slow queries will be logged and `1.0` means all slow queries will be logged. |
+| `slow_query.ttl` | String | `30d` | The TTL of the `slow_queries` system table. Default is `30d` when `record_type` is `system_table`. |
 | `export_metrics` | -- | -- | The datanode can export its metrics and send to Prometheus compatible service (e.g. send to `greptimedb` itself) from remote-write API.<br/>This is only used for `greptimedb` to export its own metrics internally. It's different from prometheus scrape. |
 | `export_metrics.enable` | Bool | `false` | whether enable export metrics. |
 | `export_metrics.write_interval` | String | `30s` | The interval of export metrics. |
@@ -325,6 +331,10 @@
 | `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. |
+| `http` | -- | -- | The HTTP server options. |
+| `http.addr` | String | `127.0.0.1:4000` | The address to bind the HTTP server. |
+| `http.timeout` | String | `0s` | HTTP request timeout. Set to 0 to disable timeout. |
+| `http.body_limit` | String | `64MB` | HTTP request body limit.<br/>The following units are supported: `B`, `KB`, `KiB`, `MB`, `MiB`, `GB`, `GiB`, `TB`, `TiB`, `PB`, `PiB`.<br/>Set to 0 to disable limit. |
 | `procedure` | -- | -- | Procedure storage options. |
 | `procedure.max_retry_times` | Integer | `12` | Procedure max retry time. |
 | `procedure.retry_delay` | String | `500ms` | Initial retry delay of procedures, increases exponentially |
@@ -362,10 +372,6 @@
 | `logging.max_log_files` | Integer | `720` | The maximum amount of log files. |
 | `logging.tracing_sample_ratio` | -- | -- | The percentage of tracing will be sampled and exported.<br/>Valid range `[0, 1]`, 1 means all traces are sampled, 0 means all traces are not sampled, the default value is 1.<br/>ratio > 1 are treated as 1. Fractions < 0 are treated as 0 |
 | `logging.tracing_sample_ratio.default_ratio` | Float | `1.0` | -- |
-| `logging.slow_query` | -- | -- | The slow query log options. |
-| `logging.slow_query.enable` | Bool | `false` | Whether to enable slow query log. |
-| `logging.slow_query.threshold` | String | Unset | The threshold of slow query. |
-| `logging.slow_query.sample_ratio` | Float | Unset | The sampling ratio of slow query log. The value should be in the range of (0, 1]. |
 | `export_metrics` | -- | -- | The datanode can export its metrics and send to Prometheus compatible service (e.g. send to `greptimedb` itself) from remote-write API.<br/>This is only used for `greptimedb` to export its own metrics internally. It's different from prometheus scrape. |
 | `export_metrics.enable` | Bool | `false` | whether enable export metrics. |
 | `export_metrics.write_interval` | String | `30s` | The interval of export metrics. |
@@ -495,6 +501,7 @@
 | `region_engine.mito.index.metadata_cache_size` | String | `64MiB` | Cache size for inverted index metadata. |
 | `region_engine.mito.index.content_cache_size` | String | `128MiB` | Cache size for inverted index content. |
 | `region_engine.mito.index.content_cache_page_size` | String | `64KiB` | Page size for inverted index content cache. |
+| `region_engine.mito.index.result_cache_size` | String | `128MiB` | Cache size for index result. |
 | `region_engine.mito.inverted_index` | -- | -- | The options for inverted index in Mito engine. |
 | `region_engine.mito.inverted_index.create_on_flush` | String | `auto` | Whether to create the index on flush.<br/>- `auto`: automatically (default)<br/>- `disable`: never |
 | `region_engine.mito.inverted_index.create_on_compaction` | String | `auto` | Whether to create the index on compaction.<br/>- `auto`: automatically (default)<br/>- `disable`: never |
@@ -529,10 +536,6 @@
 | `logging.max_log_files` | Integer | `720` | The maximum amount of log files. |
 | `logging.tracing_sample_ratio` | -- | -- | The percentage of tracing will be sampled and exported.<br/>Valid range `[0, 1]`, 1 means all traces are sampled, 0 means all traces are not sampled, the default value is 1.<br/>ratio > 1 are treated as 1. Fractions < 0 are treated as 0 |
 | `logging.tracing_sample_ratio.default_ratio` | Float | `1.0` | -- |
-| `logging.slow_query` | -- | -- | The slow query log options. |
-| `logging.slow_query.enable` | Bool | `false` | Whether to enable slow query log. |
-| `logging.slow_query.threshold` | String | Unset | The threshold of slow query. |
-| `logging.slow_query.sample_ratio` | Float | Unset | The sampling ratio of slow query log. The value should be in the range of (0, 1]. |
 | `export_metrics` | -- | -- | The datanode can export its metrics and send to Prometheus compatible service (e.g. send to `greptimedb` itself) from remote-write API.<br/>This is only used for `greptimedb` to export its own metrics internally. It's different from prometheus scrape. |
 | `export_metrics.enable` | Bool | `false` | whether enable export metrics. |
 | `export_metrics.write_interval` | String | `30s` | The interval of export metrics. |
@@ -585,9 +588,5 @@
 | `logging.max_log_files` | Integer | `720` | The maximum amount of log files. |
 | `logging.tracing_sample_ratio` | -- | -- | The percentage of tracing will be sampled and exported.<br/>Valid range `[0, 1]`, 1 means all traces are sampled, 0 means all traces are not sampled, the default value is 1.<br/>ratio > 1 are treated as 1. Fractions < 0 are treated as 0 |
 | `logging.tracing_sample_ratio.default_ratio` | Float | `1.0` | -- |
-| `logging.slow_query` | -- | -- | The slow query log options. |
-| `logging.slow_query.enable` | Bool | `false` | Whether to enable slow query log. |
-| `logging.slow_query.threshold` | String | Unset | The threshold of slow query. |
-| `logging.slow_query.sample_ratio` | Float | Unset | The sampling ratio of slow query log. The value should be in the range of (0, 1]. |
 | `tracing` | -- | -- | The tracing options. Only effect when compiled with `tokio-console` feature. |
 | `tracing.tokio_console_addr` | String | Unset | The tokio console address. |

config/datanode.example.toml

@@ -499,6 +499,9 @@ content_cache_size = "128MiB"
 ## Page size for inverted index content cache.
 content_cache_page_size = "64KiB"
 
+## Cache size for index result.
+result_cache_size = "128MiB"
+
 ## The options for inverted index in Mito engine.
 [region_engine.mito.inverted_index]
 
@@ -632,19 +635,6 @@ max_log_files = 720
 [logging.tracing_sample_ratio]
 default_ratio = 1.0
 
-## The slow query log options.
-[logging.slow_query]
-## Whether to enable slow query log.
-enable = false
-
-## The threshold of slow query.
-## @toml2docs:none-default
-threshold = "10s"
-
-## The sampling ratio of slow query log. The value should be in the range of (0, 1].
-## @toml2docs:none-default
-sample_ratio = 1.0
-
 ## The datanode can export its metrics and send to Prometheus compatible service (e.g. send to `greptimedb` itself) from remote-write API.
 ## This is only used for `greptimedb` to export its own metrics internally. It's different from prometheus scrape.
 [export_metrics]

config/flownode.example.toml

@@ -100,19 +100,6 @@ max_log_files = 720
 [logging.tracing_sample_ratio]
 default_ratio = 1.0
 
-## The slow query log options.
-[logging.slow_query]
-## Whether to enable slow query log.
-enable = false
-
-## The threshold of slow query.
-## @toml2docs:none-default
-threshold = "10s"
-
-## The sampling ratio of slow query log. The value should be in the range of (0, 1].
-## @toml2docs:none-default
-sample_ratio = 1.0
-
 ## The tracing options. Only effect when compiled with `tokio-console` feature.
 #+ [tracing]
 ## The tokio console address.

config/frontend.example.toml

@@ -37,6 +37,12 @@ enable_cors = true
 ## Customize allowed origins for HTTP CORS.
 ## @toml2docs:none-default
 cors_allowed_origins = ["https://example.com"]
+## Whether to enable validation for Prometheus remote write requests.
+## Available options:
+## - strict: deny invalid UTF-8 strings (default).
+## - lossy: allow invalid UTF-8 strings, replace invalid characters with REPLACEMENT_CHARACTER(U+FFFD).
+## - unchecked: do not valid strings.
+prom_validation_mode = "strict"
 
 ## The gRPC server options.
 [grpc]
@@ -223,18 +229,24 @@ max_log_files = 720
 default_ratio = 1.0
 
 ## The slow query log options.
-[logging.slow_query]
+[slow_query]
 ## Whether to enable slow query log.
-enable = false
+enable = true
 
-## The threshold of slow query.
-## @toml2docs:none-default
-threshold = "10s"
+## The record type of slow queries. It can be `system_table` or `log`.
+## If `system_table` is selected, the slow queries will be recorded in a system table `greptime_private.slow_queries`.
+## If `log` is selected, the slow queries will be logged in a log file `greptimedb-slow-queries.*`.
+record_type = "system_table"
 
-## The sampling ratio of slow query log. The value should be in the range of (0, 1].
-## @toml2docs:none-default
+## The threshold of slow query. It can be human readable time string, for example: `10s`, `100ms`, `1s`.
+threshold = "30s"
+
+## The sampling ratio of slow query log. The value should be in the range of (0, 1]. For example, `0.1` means 10% of the slow queries will be logged and `1.0` means all slow queries will be logged.
 sample_ratio = 1.0
 
+## The TTL of the `slow_queries` system table. Default is `30d` when `record_type` is `system_table`.
+ttl = "30d"
+
 ## The datanode can export its metrics and send to Prometheus compatible service (e.g. send to `greptimedb` itself) from remote-write API.
 ## This is only used for `greptimedb` to export its own metrics internally. It's different from prometheus scrape.
 [export_metrics]

config/metasrv.example.toml

@@ -67,6 +67,17 @@ node_max_idle_time = "24hours"
 ## The number of threads to execute the runtime for global write operations.
 #+ compact_rt_size = 4
 
+## The HTTP server options.
+[http]
+## The address to bind the HTTP server.
+addr = "127.0.0.1:4000"
+## HTTP request timeout. Set to 0 to disable timeout.
+timeout = "0s"
+## HTTP request body limit.
+## The following units are supported: `B`, `KB`, `KiB`, `MB`, `MiB`, `GB`, `GiB`, `TB`, `TiB`, `PB`, `PiB`.
+## Set to 0 to disable limit.
+body_limit = "64MB"
+
 ## Procedure storage options.
 [procedure]
 
@@ -218,19 +229,6 @@ max_log_files = 720
 [logging.tracing_sample_ratio]
 default_ratio = 1.0
 
-## The slow query log options.
-[logging.slow_query]
-## Whether to enable slow query log.
-enable = false
-
-## The threshold of slow query.
-## @toml2docs:none-default
-threshold = "10s"
-
-## The sampling ratio of slow query log. The value should be in the range of (0, 1].
-## @toml2docs:none-default
-sample_ratio = 1.0
-
 ## The datanode can export its metrics and send to Prometheus compatible service (e.g. send to `greptimedb` itself) from remote-write API.
 ## This is only used for `greptimedb` to export its own metrics internally. It's different from prometheus scrape.
 [export_metrics]

config/standalone.example.toml

@@ -43,6 +43,13 @@ enable_cors = true
 ## @toml2docs:none-default
 cors_allowed_origins = ["https://example.com"]
 
+## Whether to enable validation for Prometheus remote write requests.
+## Available options:
+## - strict: deny invalid UTF-8 strings (default).
+## - lossy: allow invalid UTF-8 strings, replace invalid characters with REPLACEMENT_CHARACTER(U+FFFD).
+## - unchecked: do not valid strings.
+prom_validation_mode = "strict"
+
 ## The gRPC server options.
 [grpc]
 ## The address to bind the gRPC server.
@@ -590,6 +597,9 @@ content_cache_size = "128MiB"
 ## Page size for inverted index content cache.
 content_cache_page_size = "64KiB"
 
+## Cache size for index result.
+result_cache_size = "128MiB"
+
 ## The options for inverted index in Mito engine.
 [region_engine.mito.inverted_index]
 
@@ -724,17 +734,21 @@ max_log_files = 720
 default_ratio = 1.0
 
 ## The slow query log options.
-[logging.slow_query]
+[slow_query]
 ## Whether to enable slow query log.
-enable = false
+#+ enable = false
+
+## The record type of slow queries. It can be `system_table` or `log`.
+## @toml2docs:none-default
+#+ record_type = "system_table"
 
 ## The threshold of slow query.
 ## @toml2docs:none-default
-threshold = "10s"
+#+ threshold = "10s"
 
 ## The sampling ratio of slow query log. The value should be in the range of (0, 1].
 ## @toml2docs:none-default
-sample_ratio = 1.0
+#+ sample_ratio = 1.0
 
 ## The datanode can export its metrics and send to Prometheus compatible service (e.g. send to `greptimedb` itself) from remote-write API.
 ## This is only used for `greptimedb` to export its own metrics internally. It's different from prometheus scrape.

cyborg/bin/bump-website-version.ts

@@ -0,0 +1,57 @@
+/*
+ * 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.
+ */
+
+import * as core from "@actions/core";
+import {obtainClient} from "@/common";
+
+async function triggerWorkflow(workflowId: string, version: string) {
+  const websiteClient = obtainClient("WEBSITE_REPO_TOKEN")
+  try {
+    await websiteClient.rest.actions.createWorkflowDispatch({
+      owner: "GreptimeTeam",
+      repo: "website",
+      workflow_id: workflowId,
+      ref: "main",
+      inputs: {
+        version,
+      },
+    });
+    console.log(`Successfully triggered ${workflowId} workflow with version ${version}`);
+  } catch (error) {
+    core.setFailed(`Failed to trigger workflow: ${error.message}`);
+  }
+}
+
+const version = process.env.VERSION;
+if (!version) {
+  core.setFailed("VERSION environment variable is required");
+  process.exit(1);
+}
+
+// Remove 'v' prefix if exists
+const cleanVersion = version.startsWith('v') ? version.slice(1) : version;
+
+if (cleanVersion.includes('nightly')) {
+  console.log('Nightly version detected, skipping workflow trigger.');
+  process.exit(0);
+}
+
+try {
+  triggerWorkflow('bump-patch-version.yml', cleanVersion);
+} catch (error) {
+  core.setFailed(`Error processing version: ${error.message}`);
+  process.exit(1);
+}

docs/how-to/how-to-change-log-level-on-the-fly.md

@@ -11,6 +11,6 @@ And database will reply with something like:
 Log Level changed from Some("info") to "trace,flow=debug"%
 ```
 
-The data is a string in the format of `global_level,module1=level1,module2=level2,...` that follow the same rule of `RUST_LOG`. 
+The data is a string in the format of `global_level,module1=level1,module2=level2,...` that follows the same rule of `RUST_LOG`. 
 
 The module is the module name of the log, and the level is the log level. The log level can be one of the following: `trace`, `debug`, `info`, `warn`, `error`, `off`(case insensitive).

docs/how-to/how-to-implement-sql-statement.md

@@ -14,7 +14,7 @@ impl SqlQueryHandler for Instance {
 ```
 
 Normally, when a SQL query arrives at GreptimeDB, the `do_query` method will be called. After some parsing work, the SQL
-will be feed into `StatementExecutor`:
+will be fed into `StatementExecutor`:
 
 ```rust
 // in Frontend Instance:
@@ -27,7 +27,7 @@ an example.
 
 Now, what if the statements should be handled differently for GreptimeDB Standalone and Cluster? You can see there's
 a `SqlStatementExecutor` field in `StatementExecutor`. Each GreptimeDB Standalone and Cluster has its own implementation
-of `SqlStatementExecutor`. If you are going to implement the statements differently in the two mode (
+of `SqlStatementExecutor`. If you are going to implement the statements differently in the two modes (
 like `CREATE TABLE`), you have to implement them in their own `SqlStatementExecutor`s.
 
 Summarize as the diagram below:

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

@@ -1,6 +1,6 @@
 # Profile memory usage of GreptimeDB
 
-This crate provides an easy approach to dump memory profiling info. A set of ready to use scripts is provided in [docs/how-to/memory-profile-scripts](docs/how-to/memory-profile-scripts).
+This crate provides an easy approach to dump memory profiling info. A set of ready to use scripts is provided in [docs/how-to/memory-profile-scripts](./memory-profile-scripts/scripts).
 
 ## Prerequisites
 ### jemalloc
@@ -44,6 +44,10 @@ Dump memory profiling data through HTTP API:
 
 ```bash
 curl -X POST localhost:4000/debug/prof/mem > greptime.hprof
+# or output flamegraph directly
+curl -X POST "localhost:4000/debug/prof/mem?output=flamegraph" > greptime.svg
+# or output pprof format
+curl -X POST "localhost:4000/debug/prof/mem?output=proto" > greptime.pprof
 ```
 
 You can periodically dump profiling data and compare them to find the delta memory usage.

docs/how-to/how-to-write-aggregate-function.md

@@ -1,8 +1,8 @@
 Currently, our query engine is based on DataFusion, so all aggregate function is executed by DataFusion, through its UDAF interface. You can find DataFusion's UDAF example [here](https://github.com/apache/arrow-datafusion/blob/arrow2/datafusion-examples/examples/simple_udaf.rs). Basically, we provide the same way as DataFusion to write aggregate functions: both are centered in a struct called "Accumulator" to accumulates states along the way in aggregation.
 
-However, DataFusion's UDAF implementation has a huge restriction, that it requires user to provide a concrete "Accumulator". Take `Median` aggregate function for example, to aggregate a `u32` datatype column, you have to write a `MedianU32`, and use `SELECT MEDIANU32(x)` in SQL. `MedianU32` cannot be used to aggregate a `i32` datatype column. Or, there's another way: you can use a special type that can hold all kinds of data (like our `Value` enum or Arrow's `ScalarValue`), and `match` all the way up to do aggregate calculations. It might work, though rather tedious. (But I think it's DataFusion's prefer way to write UDAF.)
+However, DataFusion's UDAF implementation has a huge restriction, that it requires user to provide a concrete "Accumulator". Take `Median` aggregate function for example, to aggregate a `u32` datatype column, you have to write a `MedianU32`, and use `SELECT MEDIANU32(x)` in SQL. `MedianU32` cannot be used to aggregate a `i32` datatype column. Or, there's another way: you can use a special type that can hold all kinds of data (like our `Value` enum or Arrow's `ScalarValue`), and `match` all the way up to do aggregate calculations. It might work, though rather tedious. (But I think it's DataFusion's preferred way to write UDAF.)
 
-So is there a way we can make an aggregate function that automatically match the input data's type? For example, a `Median` aggregator that can work on both `u32` column and `i32`? The answer is yes until we found a way to bypassing DataFusion's restriction, a restriction that DataFusion simply don't pass the input data's type when creating an Accumulator.
+So is there a way we can make an aggregate function that automatically match the input data's type? For example, a `Median` aggregator that can work on both `u32` column and `i32`? The answer is yes until we find a way to bypass DataFusion's restriction, a restriction that DataFusion simply doesn't pass the input data's type when creating an Accumulator.
 
 > There's an example in `my_sum_udaf_example.rs`, take that as quick start.
 
@@ -16,7 +16,7 @@ You must first define a struct that will be used to create your accumulator. For
 struct MySumAccumulatorCreator {}
 ```
 
-Attribute macro `#[as_aggr_func_creator]` and derive macro `#[derive(Debug, AggrFuncTypeStore)]` must both annotated on the struct. They work together to provide a storage of aggregate function's input data types, which are needed for creating generic accumulator later.  
+Attribute macro `#[as_aggr_func_creator]` and derive macro `#[derive(Debug, AggrFuncTypeStore)]` must both be annotated on the struct. They work together to provide a storage of aggregate function's input data types, which are needed for creating generic accumulator later.  
 
 > Note that the `as_aggr_func_creator` macro will add fields to the struct, so the struct cannot be defined as an empty struct without field like `struct Foo;`, neither as a new type like `struct Foo(bar)`.
 
@@ -32,11 +32,11 @@ pub trait AggregateFunctionCreator: Send + Sync + Debug {
 
 You can use input data's type in methods that return output type and state types (just invoke `input_types()`).
 
-The output type is aggregate function's output data's type. For example, `SUM` aggregate function's output type is `u64` for a `u32` datatype column. The state types are accumulator's internal states' types. Take `AVG` aggregate function on a `i32` column as example, it's state types are `i64` (for sum) and `u64` (for count).
+The output type is aggregate function's output data's type. For example, `SUM` aggregate function's output type is `u64` for a `u32` datatype column. The state types are accumulator's internal states' types. Take `AVG` aggregate function on a `i32` column as example, its state types are `i64` (for sum) and `u64` (for count).
 
 The `creator` function is where you define how an accumulator (that will be used in DataFusion) is created. You define "how" to create the accumulator (instead of "what" to create), using the input data's type as arguments. With input datatype known, you can create accumulator generically.
 
-# 2. Impl `Accumulator` trait for you accumulator.
+# 2. Impl `Accumulator` trait for your accumulator.
 
 The accumulator is where you store the aggregate calculation states and evaluate a result. You must impl `Accumulator` trait for it. The trait's definition is:
 
@@ -49,7 +49,7 @@ pub trait Accumulator: Send + Sync + Debug {
 }
 ```
 
-The DataFusion basically execute aggregate like this:
+The DataFusion basically executes aggregate like this:
 
 1. Partitioning all input data for aggregate. Create an accumulator for each part.
 2. Call `update_batch` on each accumulator with partitioned data, to let you update your aggregate calculation.
@@ -57,16 +57,16 @@ The DataFusion basically execute aggregate like this:
 4. Call `merge_batch` to merge all accumulator's internal state to one.
 5. Execute `evaluate` on the chosen one to get the final calculation result.
 
-Once you know the meaning of each method, you can easily write your accumulator. You can refer to `Median` accumulator or `SUM` accumulator defined in  file `my_sum_udaf_example.rs` for more details.
+Once you know the meaning of each method, you can easily write your accumulator. You can refer to `Median` accumulator or `SUM` accumulator defined in file `my_sum_udaf_example.rs` for more details.
 
 # 3. Register your aggregate function to our query engine.
 
 You can call `register_aggregate_function` method in query engine to register your aggregate function. To do that, you have to new an instance of struct `AggregateFunctionMeta`. The struct has three fields, first is the name of your aggregate function's name. The function name is case-sensitive due to DataFusion's restriction. We strongly recommend using lowercase for your name. If you have to use uppercase name, wrap your aggregate function with quotation marks. For example, if you define an aggregate function named "my_aggr", you can use "`SELECT MY_AGGR(x)`"; if you define "my_AGGR", you have to use "`SELECT "my_AGGR"(x)`".
 
-The second field is arg_counts ,the count of the arguments. Like accumulator `percentile`, calculating the p_number of the column. We need to input the value of column and the value of p to cacalate, and so the count of the arguments is two.
+The second field is arg_counts ,the count of the arguments. Like accumulator `percentile`, calculating the p_number of the column. We need to input the value of column and the value of p to calculate, and so the count of the arguments is two.
 
 The third field is a function about how to create your accumulator creator that you defined in step 1 above. Create creator, that's a bit intertwined, but it is how we make DataFusion use a newly created aggregate function each time it executes a SQL, preventing the stored input types from affecting each other. The key detail can be starting looking at our `DfContextProviderAdapter` struct's `get_aggregate_meta` method.
 
 # (Optional) 4. Make your aggregate function automatically registered.
 
-If you've written a great aggregate function that want to let everyone use it, you can make it automatically registered to our query engine at start time. It's quick simple, just refer to the `AggregateFunctions::register` function in `common/function/src/scalars/aggregate/mod.rs`.
+If you've written a great aggregate function that wants to let everyone use it, you can make it automatically register to our query engine at start time. It's quick and simple, just refer to the `AggregateFunctions::register` function in `common/function/src/scalars/aggregate/mod.rs`.

docs/how-to/how-to-write-fuzz-tests.md

@@ -3,7 +3,7 @@
 This document introduces how to write fuzz tests in GreptimeDB.
 
 ## What is a fuzz test
-Fuzz test is tool that leverage deterministic random generation to assist in finding bugs. The goal of fuzz tests is to identify inputs generated by the fuzzer that cause system panics, crashes, or unexpected behaviors to occur. And we are using the [cargo-fuzz](https://github.com/rust-fuzz/cargo-fuzz) to run our fuzz test targets.
+Fuzz test is tool that leverages deterministic random generation to assist in finding bugs. The goal of fuzz tests is to identify inputs generated by the fuzzer that cause system panics, crashes, or unexpected behaviors to occur. And we are using the [cargo-fuzz](https://github.com/rust-fuzz/cargo-fuzz) to run our fuzz test targets.
 
 ## Why we need them
 - Find bugs by leveraging random generation

flake.lock

@@ -8,11 +8,11 @@
         "rust-analyzer-src": "rust-analyzer-src"
       },
       "locked": {
-        "lastModified": 1737613896,
-        "narHash": "sha256-ldqXIglq74C7yKMFUzrS9xMT/EVs26vZpOD68Sh7OcU=",
+        "lastModified": 1745735608,
+        "narHash": "sha256-L0jzm815XBFfF2wCFmR+M1CF+beIEFj6SxlqVKF59Ec=",
         "owner": "nix-community",
         "repo": "fenix",
-        "rev": "303a062fdd8e89f233db05868468975d17855d80",
+        "rev": "c39a78eba6ed2a022cc3218db90d485077101496",
         "type": "github"
       },
       "original": {
@@ -41,16 +41,16 @@
     },
     "nixpkgs": {
       "locked": {
-        "lastModified": 1737569578,
-        "narHash": "sha256-6qY0pk2QmUtBT9Mywdvif0i/CLVgpCjMUn6g9vB+f3M=",
+        "lastModified": 1748162331,
+        "narHash": "sha256-rqc2RKYTxP3tbjA+PB3VMRQNnjesrT0pEofXQTrMsS8=",
         "owner": "NixOS",
         "repo": "nixpkgs",
-        "rev": "47addd76727f42d351590c905d9d1905ca895b82",
+        "rev": "7c43f080a7f28b2774f3b3f43234ca11661bf334",
         "type": "github"
       },
       "original": {
         "owner": "NixOS",
-        "ref": "nixos-24.11",
+        "ref": "nixos-25.05",
         "repo": "nixpkgs",
         "type": "github"
       }
@@ -65,11 +65,11 @@
     "rust-analyzer-src": {
       "flake": false,
       "locked": {
-        "lastModified": 1737581772,
-        "narHash": "sha256-t1P2Pe3FAX9TlJsCZbmJ3wn+C4qr6aSMypAOu8WNsN0=",
+        "lastModified": 1745694049,
+        "narHash": "sha256-fxvRYH/tS7hGQeg9zCVh5RBcSWT+JGJet7RA8Ss+rC0=",
         "owner": "rust-lang",
         "repo": "rust-analyzer",
-        "rev": "582af7ee9c8d84f5d534272fc7de9f292bd849be",
+        "rev": "d8887c0758bbd2d5f752d5bd405d4491e90e7ed6",
         "type": "github"
       },
       "original": {

flake.nix

@@ -2,7 +2,7 @@
   description = "Development environment flake";
 
   inputs = {
-    nixpkgs.url = "github:NixOS/nixpkgs/nixos-24.11";
+    nixpkgs.url = "github:NixOS/nixpkgs/nixos-25.05";
     fenix = {
       url = "github:nix-community/fenix";
       inputs.nixpkgs.follows = "nixpkgs";
@@ -21,7 +21,7 @@
         lib = nixpkgs.lib;
         rustToolchain = fenix.packages.${system}.fromToolchainName {
           name = (lib.importTOML ./rust-toolchain.toml).toolchain.channel;
-          sha256 = "sha256-f/CVA1EC61EWbh0SjaRNhLL0Ypx2ObupbzigZp8NmL4=";
+          sha256 = "sha256-tJJr8oqX3YD+ohhPK7jlt/7kvKBnBqJVjYtoFr520d4=";
         };
       in
       {
@@ -51,6 +51,7 @@
           ];
 
           LD_LIBRARY_PATH = pkgs.lib.makeLibraryPath buildInputs;
+          NIX_HARDENING_ENABLE = "";
         };
       });
 }

grafana/README.md

@@ -2,30 +2,63 @@
 
 ## Overview
 
-This repository maintains the Grafana dashboards for GreptimeDB. It has two types of dashboards:
+This repository contains Grafana dashboards for visualizing metrics and logs of GreptimeDB instances running in either cluster or standalone mode. **The Grafana version should be greater than 9.0**.
 
-- `cluster/dashboard.json`: The Grafana dashboard for the GreptimeDB cluster. Read the [dashboard.md](./dashboards/cluster/dashboard.md) for more details.
-- `standalone/dashboard.json`: The Grafana dashboard for the standalone GreptimeDB instance. **It's generated from the `cluster/dashboard.json` by removing the instance filter through the `make dashboards` command**. Read the [dashboard.md](./dashboards/standalone/dashboard.md) for more details.
+We highly recommend using the self-monitoring feature provided by [GreptimeDB Operator](https://github.com/GrepTimeTeam/greptimedb-operator) to automatically collect metrics and logs from your GreptimeDB instances and store them in a dedicated GreptimeDB instance.
 
-As the rapid development of GreptimeDB, the metrics may be changed, and please feel free to submit your feedback and/or contribution to this dashboard 🤗
+- **Metrics Dashboards**
 
-**NOTE**: 
+  - `dashboards/metrics/cluster/dashboard.json`: The Grafana dashboard for the GreptimeDB cluster. Read the [dashboard.md](./dashboards/metrics/cluster/dashboard.md) for more details.
+  
+  - `dashboards/metrics/standalone/dashboard.json`: The Grafana dashboard for the standalone GreptimeDB instance. **It's generated from the `cluster/dashboard.json` by removing the instance filter through the `make dashboards` command**. Read the [dashboard.md](./dashboards/metrics/standalone/dashboard.md) for more details.
 
-- The Grafana version should be greater than 9.0.
+- **Logs Dashboard**
 
-- If you want to modify the dashboards, you only need to modify the `cluster/dashboard.json` and run the `make dashboards` command to generate the `standalone/dashboard.json` and other related files.
+  The `dashboards/logs/dashboard.json` provides a comprehensive Grafana dashboard for visualizing GreptimeDB logs. To utilize this dashboard effectively, you need to collect logs in JSON format from your GreptimeDB instances and store them in a dedicated GreptimeDB instance.
 
-To maintain the dashboards easily, we use the [`dac`](https://github.com/zyy17/dac) tool to generate the intermediate dashboards and markdown documents:
+  For proper integration, the logs table must adhere to the following schema design with the table name `_gt_logs`:
 
-- `cluster/dashboard.yaml`: The intermediate dashboard for the GreptimeDB cluster.
-- `standalone/dashboard.yaml`: The intermediate dashboard for the standalone GreptimeDB instance.
+  ```sql
+  CREATE TABLE IF NOT EXISTS `_gt_logs` (
+    `pod_ip` STRING NULL,
+    `namespace` STRING NULL,
+    `cluster` STRING NULL,
+    `file` STRING NULL,
+    `module_path` STRING NULL,
+    `level` STRING NULL,
+    `target` STRING NULL,
+    `role` STRING NULL,
+    `pod` STRING NULL SKIPPING INDEX WITH(granularity = '10240', type = 'BLOOM'),
+    `message` STRING NULL FULLTEXT INDEX WITH(analyzer = 'English', backend = 'bloom', case_sensitive = 'false'),
+    `err` STRING NULL FULLTEXT INDEX WITH(analyzer = 'English', backend = 'bloom', case_sensitive = 'false'),
+    `timestamp` TIMESTAMP(9) NOT NULL,
+    TIME INDEX (`timestamp`),
+    PRIMARY KEY (`level`, `target`, `role`)
+  )
+    ENGINE=mito
+  WITH (
+    append_mode = 'true'
+  )
+  ```
+
+## Development
+
+As GreptimeDB evolves rapidly, metrics may change over time. We welcome your feedback and contributions to improve these dashboards 🤗
+
+To modify the metrics dashboards, simply edit the `dashboards/metrics/cluster/dashboard.json` file and run the `make dashboards` command. This will automatically generate the updated `dashboards/metrics/standalone/dashboard.json` and other related files.
+
+For easier dashboard maintenance, we utilize the [`dac`](https://github.com/zyy17/dac) tool to generate human-readable intermediate dashboards and documentation:
+
+- `dashboards/metrics/cluster/dashboard.yaml`: The intermediate dashboard file for the GreptimeDB cluster.
+- `dashboards/metrics/standalone/dashboard.yaml`: The intermediate dashboard file for standalone GreptimeDB instances.
 
 ## Data Sources
 
-There are two data sources for the dashboards to fetch the metrics:
+The following data sources are used to fetch metrics and logs:
 
-- **Prometheus**: Expose the metrics of GreptimeDB.
-- **Information Schema**: It is the MySQL port of the current monitored instance. The `overview` dashboard will use this datasource to show the information schema of the current instance.
+- **`${metrics}`**: Prometheus data source for providing the GreptimeDB metrics.
+- **`${logs}`**: MySQL data source for providing the GreptimeDB logs.
+- **`${information_schema}`**: MySQL data source for providing the information schema of the current instance and used for the `overview` panel. It is the MySQL port of the current monitored instance.
 
 ## Instance Filters
 
@@ -43,9 +76,9 @@ And the legend will be like: `[{{instance}}]-[{{ pod }}]`.
 
 ## Deployment
 
-### Helm
+### (Recommended) Helm Chart
 
-If you use the Helm [chart](https://github.com/GreptimeTeam/helm-charts) to deploy a GreptimeDB cluster, you can enable self-monitoring by setting the following values in your Helm chart:
+If you use the [Helm Chart](https://github.com/GreptimeTeam/helm-charts) to deploy a GreptimeDB cluster, you can enable self-monitoring by setting the following values in your Helm chart:
 
 - `monitoring.enabled=true`: Deploys a standalone GreptimeDB instance dedicated to monitoring the cluster;
 - `grafana.enabled=true`: Deploys Grafana and automatically imports the monitoring dashboard;
@@ -85,5 +118,5 @@ The standalone GreptimeDB instance will collect metrics from your cluster, and t
 
 3. **Import the dashboards based on your deployment scenario**
 
-   - **Cluster**: Import the `cluster/dashboard.json` dashboard.
-   - **Standalone**: Import the `standalone/dashboard.json` dashboard.
+   - **Cluster**: Import the `dashboards/metrics/cluster/dashboard.json` dashboard.
+   - **Standalone**: Import the `dashboards/metrics/standalone/dashboard.json` dashboard.

grafana/dashboards/logs/dashboard.json

@@ -0,0 +1,292 @@
+{
+  "annotations": {
+    "list": [
+      {
+        "builtIn": 1,
+        "datasource": {
+          "type": "grafana",
+          "uid": "-- Grafana --"
+        },
+        "enable": true,
+        "hide": true,
+        "iconColor": "rgba(0, 211, 255, 1)",
+        "name": "Annotations & Alerts",
+        "type": "dashboard"
+      }
+    ]
+  },
+  "editable": true,
+  "fiscalYearStartMonth": 0,
+  "graphTooltip": 0,
+  "id": 12,
+  "links": [],
+  "panels": [
+    {
+      "datasource": {
+        "default": false,
+        "type": "mysql",
+        "uid": "${datasource}"
+      },
+      "fieldConfig": {
+        "defaults": {},
+        "overrides": []
+      },
+      "gridPos": {
+        "h": 20,
+        "w": 24,
+        "x": 0,
+        "y": 0
+      },
+      "id": 1,
+      "options": {
+        "dedupStrategy": "none",
+        "enableInfiniteScrolling": true,
+        "enableLogDetails": true,
+        "prettifyLogMessage": false,
+        "showCommonLabels": false,
+        "showLabels": false,
+        "showTime": true,
+        "sortOrder": "Descending",
+        "wrapLogMessage": false
+      },
+      "pluginVersion": "11.6.0",
+      "targets": [
+        {
+          "dataset": "greptime_private",
+          "datasource": {
+            "type": "mysql",
+            "uid": "${datasource}"
+          },
+          "editorMode": "code",
+          "format": "table",
+          "rawQuery": true,
+          "rawSql": "SELECT `timestamp`, CONCAT('[', `level`, ']', ' ', '<', `target`, '>', ' ', `message`),\n  `role`,\n  `pod`,\n  `pod_ip`,\n  `namespace`,\n  `cluster`,\n  `err`,\n  `file`,\n  `module_path`\nFROM\n  `_gt_logs`\nWHERE\n  (\n    \"$level\" = \"'all'\"\n    OR `level` IN ($level)\n  ) \n  AND (\n    \"$role\" = \"'all'\"\n    OR `role` IN ($role)\n  )\n  AND (\n    \"$pod\" = \"\"\n    OR `pod` = '$pod'\n  )\n  AND (\n    \"$target\" = \"\"\n    OR `target` = '$target'\n  )\n  AND (\n    \"$search\" = \"\"\n    OR matches_term(`message`, '$search')\n  )\n  AND (\n    \"$exclude\" = \"\"\n    OR NOT matches_term(`message`, '$exclude')\n  )\n  AND $__timeFilter(`timestamp`)\nORDER BY `timestamp` DESC\nLIMIT $limit;\n",
+          "refId": "A",
+          "sql": {
+            "columns": [
+              {
+                "parameters": [],
+                "type": "function"
+              }
+            ],
+            "groupBy": [
+              {
+                "property": {
+                  "type": "string"
+                },
+                "type": "groupBy"
+              }
+            ],
+            "limit": 50
+          }
+        }
+      ],
+      "title": "Logs",
+      "type": "logs"
+    }
+  ],
+  "preload": false,
+  "refresh": "",
+  "schemaVersion": 41,
+  "tags": [],
+  "templating": {
+    "list": [
+      {
+        "current": {
+          "text": "logs",
+          "value": "P98F38F12DB221A8C"
+        },
+        "includeAll": false,
+        "name": "datasource",
+        "options": [],
+        "query": "mysql",
+        "refresh": 1,
+        "regex": "",
+        "type": "datasource"
+      },
+      {
+        "allValue": "'all'",
+        "current": {
+          "text": [
+            "$__all"
+          ],
+          "value": [
+            "$__all"
+          ]
+        },
+        "includeAll": true,
+        "label": "level",
+        "multi": true,
+        "name": "level",
+        "options": [
+          {
+            "selected": false,
+            "text": "INFO",
+            "value": "INFO"
+          },
+          {
+            "selected": false,
+            "text": "ERROR",
+            "value": "ERROR"
+          },
+          {
+            "selected": false,
+            "text": "WARN",
+            "value": "WARN"
+          },
+          {
+            "selected": false,
+            "text": "DEBUG",
+            "value": "DEBUG"
+          },
+          {
+            "selected": false,
+            "text": "TRACE",
+            "value": "TRACE"
+          }
+        ],
+        "query": "INFO,ERROR,WARN,DEBUG,TRACE",
+        "type": "custom"
+      },
+      {
+        "allValue": "'all'",
+        "current": {
+          "text": [
+            "$__all"
+          ],
+          "value": [
+            "$__all"
+          ]
+        },
+        "includeAll": true,
+        "label": "role",
+        "multi": true,
+        "name": "role",
+        "options": [
+          {
+            "selected": false,
+            "text": "datanode",
+            "value": "datanode"
+          },
+          {
+            "selected": false,
+            "text": "frontend",
+            "value": "frontend"
+          },
+          {
+            "selected": false,
+            "text": "meta",
+            "value": "meta"
+          }
+        ],
+        "query": "datanode,frontend,meta",
+        "type": "custom"
+      },
+      {
+        "current": {
+          "text": "",
+          "value": ""
+        },
+        "label": "pod",
+        "name": "pod",
+        "options": [
+          {
+            "selected": true,
+            "text": "",
+            "value": ""
+          }
+        ],
+        "query": "",
+        "type": "textbox"
+      },
+      {
+        "current": {
+          "text": "",
+          "value": ""
+        },
+        "label": "target",
+        "name": "target",
+        "options": [
+          {
+            "selected": true,
+            "text": "",
+            "value": ""
+          }
+        ],
+        "query": "",
+        "type": "textbox"
+      },
+      {
+        "current": {
+          "text": "",
+          "value": ""
+        },
+        "label": "search",
+        "name": "search",
+        "options": [
+          {
+            "selected": true,
+            "text": "",
+            "value": ""
+          }
+        ],
+        "query": "",
+        "type": "textbox"
+      },
+      {
+        "current": {
+          "text": "",
+          "value": ""
+        },
+        "label": "exclude",
+        "name": "exclude",
+        "options": [
+          {
+            "selected": true,
+            "text": "",
+            "value": ""
+          }
+        ],
+        "query": "",
+        "type": "textbox"
+      },
+      {
+        "current": {
+          "text": "2000",
+          "value": "2000"
+        },
+        "includeAll": false,
+        "label": "limit",
+        "name": "limit",
+        "options": [
+          {
+            "selected": true,
+            "text": "2000",
+            "value": "2000"
+          },
+          {
+            "selected": false,
+            "text": "5000",
+            "value": "5000"
+          },
+          {
+            "selected": false,
+            "text": "8000",
+            "value": "8000"
+          }
+        ],
+        "query": "2000,5000,8000",
+        "type": "custom"
+      }
+    ]
+  },
+  "time": {
+    "from": "now-6h",
+    "to": "now"
+  },
+  "timepicker": {},
+  "timezone": "browser",
+  "title": "GreptimeDB Logs",
+  "uid": "edx5veo4rd3wge2",
+  "version": 1
+}

grafana/dashboards/metrics/cluster/dashboard.md

@@ -46,6 +46,7 @@
 | Ingest Rows per Instance | `sum by(instance, pod)(rate(greptime_table_operator_ingest_rows{instance=~"$frontend"}[$__rate_interval]))` | `timeseries` | Ingestion rate by row as in each frontend | `prometheus` | `rowsps` | `[{{instance}}]-[{{pod}}]` |
 | Region Call QPS per Instance | `sum by(instance, pod, request_type) (rate(greptime_grpc_region_request_count{instance=~"$frontend"}[$__rate_interval]))` | `timeseries` | Region Call QPS per Instance. | `prometheus` | `ops` | `[{{instance}}]-[{{pod}}]-[{{request_type}}]` |
 | Region Call P99 per Instance | `histogram_quantile(0.99, sum by(instance, pod, le, request_type) (rate(greptime_grpc_region_request_bucket{instance=~"$frontend"}[$__rate_interval])))` | `timeseries` | Region Call P99 per Instance. | `prometheus` | `s` | `[{{instance}}]-[{{pod}}]-[{{request_type}}]` |
+| Frontend Handle Bulk Insert Elapsed Time  | `sum by(instance, pod, stage) (rate(greptime_table_operator_handle_bulk_insert_sum[$__rate_interval]))/sum by(instance, pod, stage) (rate(greptime_table_operator_handle_bulk_insert_count[$__rate_interval]))`<br/>`histogram_quantile(0.99, sum by(instance, pod, stage, le) (rate(greptime_table_operator_handle_bulk_insert_bucket[$__rate_interval])))` | `timeseries` | Per-stage time for frontend to handle bulk insert requests | `prometheus` | `s` | `[{{instance}}]-[{{pod}}]-[{{stage}}]-AVG` |
 # Mito Engine
 | Title | Query | Type | Description | Datasource | Unit | Legend Format |
 | --- | --- | --- | --- | --- | --- | --- |
@@ -59,7 +60,7 @@
 | Read Stage P99 per Instance | `histogram_quantile(0.99, sum by(instance, pod, le, stage) (rate(greptime_mito_read_stage_elapsed_bucket{instance=~"$datanode"}[$__rate_interval])))` | `timeseries` | Read Stage P99 per Instance. | `prometheus` | `s` | `[{{instance}}]-[{{pod}}]-[{{stage}}]` |
 | Write Stage P99 per Instance | `histogram_quantile(0.99, sum by(instance, pod, le, stage) (rate(greptime_mito_write_stage_elapsed_bucket{instance=~"$datanode"}[$__rate_interval])))` | `timeseries` | Write Stage P99 per Instance. | `prometheus` | `s` | `[{{instance}}]-[{{pod}}]-[{{stage}}]` |
 | Compaction OPS per Instance | `sum by(instance, pod) (rate(greptime_mito_compaction_total_elapsed_count{instance=~"$datanode"}[$__rate_interval]))` | `timeseries` | Compaction OPS per Instance. | `prometheus` | `ops` | `[{{ instance }}]-[{{pod}}]` |
-| Compaction P99 per Instance by Stage | `histogram_quantile(0.99, sum by(instance, pod, le, stage) (rate(greptime_mito_compaction_stage_elapsed_bucket{instance=~"$datanode"}[$__rate_interval])))` | `timeseries` | Compaction latency by stage | `prometheus` | `s` | `[{{instance}}]-[{{pod}}]-[{{stage}}]-p99` |
+| Compaction Elapsed Time per Instance by Stage | `histogram_quantile(0.99, sum by(instance, pod, le, stage) (rate(greptime_mito_compaction_stage_elapsed_bucket{instance=~"$datanode"}[$__rate_interval])))`<br/>`sum by(instance, pod, le, stage) (rate(greptime_mito_compaction_stage_elapsed_sum{instance=~"$datanode"}[$__rate_interval]))/sum by(instance, pod, le, stage) (rate(greptime_mito_compaction_stage_elapsed_count{instance=~"$datanode"}[$__rate_interval]))` | `timeseries` | Compaction latency by stage | `prometheus` | `s` | `[{{instance}}]-[{{pod}}]-[{{stage}}]-p99` |
 | Compaction P99 per Instance | `histogram_quantile(0.99, sum by(instance, pod, le,stage) (rate(greptime_mito_compaction_total_elapsed_bucket{instance=~"$datanode"}[$__rate_interval])))` | `timeseries` | Compaction P99 per Instance. | `prometheus` | `s` | `[{{instance}}]-[{{pod}}]-[{{stage}}]-compaction` |
 | WAL write size | `histogram_quantile(0.95, sum by(le,instance, pod) (rate(raft_engine_write_size_bucket[$__rate_interval])))`<br/>`histogram_quantile(0.99, sum by(le,instance,pod) (rate(raft_engine_write_size_bucket[$__rate_interval])))`<br/>`sum by (instance, pod)(rate(raft_engine_write_size_sum[$__rate_interval]))` | `timeseries` | Write-ahead logs write size as bytes. This chart includes stats of p95 and p99 size by instance, total WAL write rate. | `prometheus` | `bytes` | `[{{instance}}]-[{{pod}}]-req-size-p95` |
 | Cached Bytes per Instance | `greptime_mito_cache_bytes{instance=~"$datanode"}` | `timeseries` | Cached Bytes per Instance. | `prometheus` | `decbytes` | `[{{instance}}]-[{{pod}}]-[{{type}}]` |
@@ -67,6 +68,9 @@
 | WAL sync duration seconds | `histogram_quantile(0.99, sum by(le, type, node, instance, pod) (rate(raft_engine_sync_log_duration_seconds_bucket[$__rate_interval])))` | `timeseries` | Raft engine (local disk) log store sync latency, p99 | `prometheus` | `s` | `[{{instance}}]-[{{pod}}]-p99` |
 | Log Store op duration seconds | `histogram_quantile(0.99, sum by(le,logstore,optype,instance, pod) (rate(greptime_logstore_op_elapsed_bucket[$__rate_interval])))` | `timeseries` | Write-ahead log operations latency at p99 | `prometheus` | `s` | `[{{instance}}]-[{{pod}}]-[{{logstore}}]-[{{optype}}]-p99` |
 | Inflight Flush | `greptime_mito_inflight_flush_count` | `timeseries` | Ongoing flush task count | `prometheus` | `none` | `[{{instance}}]-[{{pod}}]` |
+| Compaction Input/Output Bytes | `sum by(instance, pod) (greptime_mito_compaction_input_bytes)`<br/>`sum by(instance, pod) (greptime_mito_compaction_output_bytes)` | `timeseries` | Compaction oinput output bytes | `prometheus` | `bytes` | `[{{instance}}]-[{{pod}}]-input` |
+| Region Worker Handle Bulk Insert Requests | `histogram_quantile(0.95, sum by(le,instance, stage, pod) (rate(greptime_region_worker_handle_write_bucket[$__rate_interval])))`<br/>`sum by(le,instance, stage, pod) (rate(greptime_region_worker_handle_write_sum[$__rate_interval]))/sum by(le,instance, stage, pod) (rate(greptime_region_worker_handle_write_count[$__rate_interval]))` | `timeseries` | Per-stage elapsed time for region worker to handle bulk insert region requests. | `prometheus` | `s` | `[{{instance}}]-[{{pod}}]-[{{stage}}]-P95` |
+| Region Worker Convert Requests | `histogram_quantile(0.95, sum by(le, instance, stage, pod) (rate(greptime_datanode_convert_region_request_bucket[$__rate_interval])))`<br/>`sum by(le,instance, stage, pod) (rate(greptime_datanode_convert_region_request_sum[$__rate_interval]))/sum by(le,instance, stage, pod) (rate(greptime_datanode_convert_region_request_count[$__rate_interval]))` | `timeseries` | Per-stage elapsed time for region worker to decode requests. | `prometheus` | `s` | `[{{instance}}]-[{{pod}}]-[{{stage}}]-P95` |
 # OpenDAL
 | Title | Query | Type | Description | Datasource | Unit | Legend Format |
 | --- | --- | --- | --- | --- | --- | --- |

grafana/dashboards/metrics/cluster/dashboard.yaml

@@ -371,6 +371,21 @@ groups:
                 type: prometheus
                 uid: ${metrics}
               legendFormat: '[{{instance}}]-[{{pod}}]-[{{request_type}}]'
+        - title: 'Frontend Handle Bulk Insert Elapsed Time '
+          type: timeseries
+          description: Per-stage time for frontend to handle bulk insert requests
+          unit: s
+          queries:
+            - expr: sum by(instance, pod, stage) (rate(greptime_table_operator_handle_bulk_insert_sum[$__rate_interval]))/sum by(instance, pod, stage) (rate(greptime_table_operator_handle_bulk_insert_count[$__rate_interval]))
+              datasource:
+                type: prometheus
+                uid: ${metrics}
+              legendFormat: '[{{instance}}]-[{{pod}}]-[{{stage}}]-AVG'
+            - expr: histogram_quantile(0.99, sum by(instance, pod, stage, le) (rate(greptime_table_operator_handle_bulk_insert_bucket[$__rate_interval])))
+              datasource:
+                type: prometheus
+                uid: ${metrics}
+              legendFormat: '[{{instance}}]-[{{pod}}]-[{{stage}}]-P95'
     - title: Mito Engine
       panels:
         - title: Request OPS per Instance
@@ -472,7 +487,7 @@ groups:
                 type: prometheus
                 uid: ${metrics}
               legendFormat: '[{{ instance }}]-[{{pod}}]'
-        - title: Compaction P99 per Instance by Stage
+        - title: Compaction Elapsed Time per Instance by Stage
           type: timeseries
           description: Compaction latency by stage
           unit: s
@@ -482,6 +497,11 @@ groups:
                 type: prometheus
                 uid: ${metrics}
               legendFormat: '[{{instance}}]-[{{pod}}]-[{{stage}}]-p99'
+            - expr: sum by(instance, pod, le, stage) (rate(greptime_mito_compaction_stage_elapsed_sum{instance=~"$datanode"}[$__rate_interval]))/sum by(instance, pod, le, stage) (rate(greptime_mito_compaction_stage_elapsed_count{instance=~"$datanode"}[$__rate_interval]))
+              datasource:
+                type: prometheus
+                uid: ${metrics}
+              legendFormat: '[{{instance}}]-[{{pod}}]-[{{stage}}]-avg'
         - title: Compaction P99 per Instance
           type: timeseries
           description: Compaction P99 per Instance.
@@ -562,6 +582,51 @@ groups:
                 type: prometheus
                 uid: ${metrics}
               legendFormat: '[{{instance}}]-[{{pod}}]'
+        - title: Compaction Input/Output Bytes
+          type: timeseries
+          description: Compaction oinput output bytes
+          unit: bytes
+          queries:
+            - expr: sum by(instance, pod) (greptime_mito_compaction_input_bytes)
+              datasource:
+                type: prometheus
+                uid: ${metrics}
+              legendFormat: '[{{instance}}]-[{{pod}}]-input'
+            - expr: sum by(instance, pod) (greptime_mito_compaction_output_bytes)
+              datasource:
+                type: prometheus
+                uid: ${metrics}
+              legendFormat: '[{{instance}}]-[{{pod}}]-output'
+        - title: Region Worker Handle Bulk Insert Requests
+          type: timeseries
+          description: Per-stage elapsed time for region worker to handle bulk insert region requests.
+          unit: s
+          queries:
+            - expr: histogram_quantile(0.95, sum by(le,instance, stage, pod) (rate(greptime_region_worker_handle_write_bucket[$__rate_interval])))
+              datasource:
+                type: prometheus
+                uid: ${metrics}
+              legendFormat: '[{{instance}}]-[{{pod}}]-[{{stage}}]-P95'
+            - expr: sum by(le,instance, stage, pod) (rate(greptime_region_worker_handle_write_sum[$__rate_interval]))/sum by(le,instance, stage, pod) (rate(greptime_region_worker_handle_write_count[$__rate_interval]))
+              datasource:
+                type: prometheus
+                uid: ${metrics}
+              legendFormat: '[{{instance}}]-[{{pod}}]-[{{stage}}]-AVG'
+        - title: Region Worker Convert Requests
+          type: timeseries
+          description: Per-stage elapsed time for region worker to decode requests.
+          unit: s
+          queries:
+            - expr: histogram_quantile(0.95, sum by(le, instance, stage, pod) (rate(greptime_datanode_convert_region_request_bucket[$__rate_interval])))
+              datasource:
+                type: prometheus
+                uid: ${metrics}
+              legendFormat: '[{{instance}}]-[{{pod}}]-[{{stage}}]-P95'
+            - expr: sum by(le,instance, stage, pod) (rate(greptime_datanode_convert_region_request_sum[$__rate_interval]))/sum by(le,instance, stage, pod) (rate(greptime_datanode_convert_region_request_count[$__rate_interval]))
+              datasource:
+                type: prometheus
+                uid: ${metrics}
+              legendFormat: '[{{instance}}]-[{{pod}}]-[{{stage}}]-AVG'
     - title: OpenDAL
       panels:
         - title: QPS per Instance

grafana/dashboards/metrics/standalone/dashboard.md

@@ -46,6 +46,7 @@
 | Ingest Rows per Instance | `sum by(instance, pod)(rate(greptime_table_operator_ingest_rows{}[$__rate_interval]))` | `timeseries` | Ingestion rate by row as in each frontend | `prometheus` | `rowsps` | `[{{instance}}]-[{{pod}}]` |
 | Region Call QPS per Instance | `sum by(instance, pod, request_type) (rate(greptime_grpc_region_request_count{}[$__rate_interval]))` | `timeseries` | Region Call QPS per Instance. | `prometheus` | `ops` | `[{{instance}}]-[{{pod}}]-[{{request_type}}]` |
 | Region Call P99 per Instance | `histogram_quantile(0.99, sum by(instance, pod, le, request_type) (rate(greptime_grpc_region_request_bucket{}[$__rate_interval])))` | `timeseries` | Region Call P99 per Instance. | `prometheus` | `s` | `[{{instance}}]-[{{pod}}]-[{{request_type}}]` |
+| Frontend Handle Bulk Insert Elapsed Time  | `sum by(instance, pod, stage) (rate(greptime_table_operator_handle_bulk_insert_sum[$__rate_interval]))/sum by(instance, pod, stage) (rate(greptime_table_operator_handle_bulk_insert_count[$__rate_interval]))`<br/>`histogram_quantile(0.99, sum by(instance, pod, stage, le) (rate(greptime_table_operator_handle_bulk_insert_bucket[$__rate_interval])))` | `timeseries` | Per-stage time for frontend to handle bulk insert requests | `prometheus` | `s` | `[{{instance}}]-[{{pod}}]-[{{stage}}]-AVG` |
 # Mito Engine
 | Title | Query | Type | Description | Datasource | Unit | Legend Format |
 | --- | --- | --- | --- | --- | --- | --- |
@@ -59,7 +60,7 @@
 | Read Stage P99 per Instance | `histogram_quantile(0.99, sum by(instance, pod, le, stage) (rate(greptime_mito_read_stage_elapsed_bucket{}[$__rate_interval])))` | `timeseries` | Read Stage P99 per Instance. | `prometheus` | `s` | `[{{instance}}]-[{{pod}}]-[{{stage}}]` |
 | Write Stage P99 per Instance | `histogram_quantile(0.99, sum by(instance, pod, le, stage) (rate(greptime_mito_write_stage_elapsed_bucket{}[$__rate_interval])))` | `timeseries` | Write Stage P99 per Instance. | `prometheus` | `s` | `[{{instance}}]-[{{pod}}]-[{{stage}}]` |
 | Compaction OPS per Instance | `sum by(instance, pod) (rate(greptime_mito_compaction_total_elapsed_count{}[$__rate_interval]))` | `timeseries` | Compaction OPS per Instance. | `prometheus` | `ops` | `[{{ instance }}]-[{{pod}}]` |
-| Compaction P99 per Instance by Stage | `histogram_quantile(0.99, sum by(instance, pod, le, stage) (rate(greptime_mito_compaction_stage_elapsed_bucket{}[$__rate_interval])))` | `timeseries` | Compaction latency by stage | `prometheus` | `s` | `[{{instance}}]-[{{pod}}]-[{{stage}}]-p99` |
+| Compaction Elapsed Time per Instance by Stage | `histogram_quantile(0.99, sum by(instance, pod, le, stage) (rate(greptime_mito_compaction_stage_elapsed_bucket{}[$__rate_interval])))`<br/>`sum by(instance, pod, le, stage) (rate(greptime_mito_compaction_stage_elapsed_sum{}[$__rate_interval]))/sum by(instance, pod, le, stage) (rate(greptime_mito_compaction_stage_elapsed_count{}[$__rate_interval]))` | `timeseries` | Compaction latency by stage | `prometheus` | `s` | `[{{instance}}]-[{{pod}}]-[{{stage}}]-p99` |
 | Compaction P99 per Instance | `histogram_quantile(0.99, sum by(instance, pod, le,stage) (rate(greptime_mito_compaction_total_elapsed_bucket{}[$__rate_interval])))` | `timeseries` | Compaction P99 per Instance. | `prometheus` | `s` | `[{{instance}}]-[{{pod}}]-[{{stage}}]-compaction` |
 | WAL write size | `histogram_quantile(0.95, sum by(le,instance, pod) (rate(raft_engine_write_size_bucket[$__rate_interval])))`<br/>`histogram_quantile(0.99, sum by(le,instance,pod) (rate(raft_engine_write_size_bucket[$__rate_interval])))`<br/>`sum by (instance, pod)(rate(raft_engine_write_size_sum[$__rate_interval]))` | `timeseries` | Write-ahead logs write size as bytes. This chart includes stats of p95 and p99 size by instance, total WAL write rate. | `prometheus` | `bytes` | `[{{instance}}]-[{{pod}}]-req-size-p95` |
 | Cached Bytes per Instance | `greptime_mito_cache_bytes{}` | `timeseries` | Cached Bytes per Instance. | `prometheus` | `decbytes` | `[{{instance}}]-[{{pod}}]-[{{type}}]` |
@@ -67,6 +68,9 @@
 | WAL sync duration seconds | `histogram_quantile(0.99, sum by(le, type, node, instance, pod) (rate(raft_engine_sync_log_duration_seconds_bucket[$__rate_interval])))` | `timeseries` | Raft engine (local disk) log store sync latency, p99 | `prometheus` | `s` | `[{{instance}}]-[{{pod}}]-p99` |
 | Log Store op duration seconds | `histogram_quantile(0.99, sum by(le,logstore,optype,instance, pod) (rate(greptime_logstore_op_elapsed_bucket[$__rate_interval])))` | `timeseries` | Write-ahead log operations latency at p99 | `prometheus` | `s` | `[{{instance}}]-[{{pod}}]-[{{logstore}}]-[{{optype}}]-p99` |
 | Inflight Flush | `greptime_mito_inflight_flush_count` | `timeseries` | Ongoing flush task count | `prometheus` | `none` | `[{{instance}}]-[{{pod}}]` |
+| Compaction Input/Output Bytes | `sum by(instance, pod) (greptime_mito_compaction_input_bytes)`<br/>`sum by(instance, pod) (greptime_mito_compaction_output_bytes)` | `timeseries` | Compaction oinput output bytes | `prometheus` | `bytes` | `[{{instance}}]-[{{pod}}]-input` |
+| Region Worker Handle Bulk Insert Requests | `histogram_quantile(0.95, sum by(le,instance, stage, pod) (rate(greptime_region_worker_handle_write_bucket[$__rate_interval])))`<br/>`sum by(le,instance, stage, pod) (rate(greptime_region_worker_handle_write_sum[$__rate_interval]))/sum by(le,instance, stage, pod) (rate(greptime_region_worker_handle_write_count[$__rate_interval]))` | `timeseries` | Per-stage elapsed time for region worker to handle bulk insert region requests. | `prometheus` | `s` | `[{{instance}}]-[{{pod}}]-[{{stage}}]-P95` |
+| Region Worker Convert Requests | `histogram_quantile(0.95, sum by(le, instance, stage, pod) (rate(greptime_datanode_convert_region_request_bucket[$__rate_interval])))`<br/>`sum by(le,instance, stage, pod) (rate(greptime_datanode_convert_region_request_sum[$__rate_interval]))/sum by(le,instance, stage, pod) (rate(greptime_datanode_convert_region_request_count[$__rate_interval]))` | `timeseries` | Per-stage elapsed time for region worker to decode requests. | `prometheus` | `s` | `[{{instance}}]-[{{pod}}]-[{{stage}}]-P95` |
 # OpenDAL
 | Title | Query | Type | Description | Datasource | Unit | Legend Format |
 | --- | --- | --- | --- | --- | --- | --- |

grafana/dashboards/metrics/standalone/dashboard.yaml

@@ -371,6 +371,21 @@ groups:
                 type: prometheus
                 uid: ${metrics}
               legendFormat: '[{{instance}}]-[{{pod}}]-[{{request_type}}]'
+        - title: 'Frontend Handle Bulk Insert Elapsed Time '
+          type: timeseries
+          description: Per-stage time for frontend to handle bulk insert requests
+          unit: s
+          queries:
+            - expr: sum by(instance, pod, stage) (rate(greptime_table_operator_handle_bulk_insert_sum[$__rate_interval]))/sum by(instance, pod, stage) (rate(greptime_table_operator_handle_bulk_insert_count[$__rate_interval]))
+              datasource:
+                type: prometheus
+                uid: ${metrics}
+              legendFormat: '[{{instance}}]-[{{pod}}]-[{{stage}}]-AVG'
+            - expr: histogram_quantile(0.99, sum by(instance, pod, stage, le) (rate(greptime_table_operator_handle_bulk_insert_bucket[$__rate_interval])))
+              datasource:
+                type: prometheus
+                uid: ${metrics}
+              legendFormat: '[{{instance}}]-[{{pod}}]-[{{stage}}]-P95'
     - title: Mito Engine
       panels:
         - title: Request OPS per Instance
@@ -472,7 +487,7 @@ groups:
                 type: prometheus
                 uid: ${metrics}
               legendFormat: '[{{ instance }}]-[{{pod}}]'
-        - title: Compaction P99 per Instance by Stage
+        - title: Compaction Elapsed Time per Instance by Stage
           type: timeseries
           description: Compaction latency by stage
           unit: s
@@ -482,6 +497,11 @@ groups:
                 type: prometheus
                 uid: ${metrics}
               legendFormat: '[{{instance}}]-[{{pod}}]-[{{stage}}]-p99'
+            - expr: sum by(instance, pod, le, stage) (rate(greptime_mito_compaction_stage_elapsed_sum{}[$__rate_interval]))/sum by(instance, pod, le, stage) (rate(greptime_mito_compaction_stage_elapsed_count{}[$__rate_interval]))
+              datasource:
+                type: prometheus
+                uid: ${metrics}
+              legendFormat: '[{{instance}}]-[{{pod}}]-[{{stage}}]-avg'
         - title: Compaction P99 per Instance
           type: timeseries
           description: Compaction P99 per Instance.
@@ -562,6 +582,51 @@ groups:
                 type: prometheus
                 uid: ${metrics}
               legendFormat: '[{{instance}}]-[{{pod}}]'
+        - title: Compaction Input/Output Bytes
+          type: timeseries
+          description: Compaction oinput output bytes
+          unit: bytes
+          queries:
+            - expr: sum by(instance, pod) (greptime_mito_compaction_input_bytes)
+              datasource:
+                type: prometheus
+                uid: ${metrics}
+              legendFormat: '[{{instance}}]-[{{pod}}]-input'
+            - expr: sum by(instance, pod) (greptime_mito_compaction_output_bytes)
+              datasource:
+                type: prometheus
+                uid: ${metrics}
+              legendFormat: '[{{instance}}]-[{{pod}}]-output'
+        - title: Region Worker Handle Bulk Insert Requests
+          type: timeseries
+          description: Per-stage elapsed time for region worker to handle bulk insert region requests.
+          unit: s
+          queries:
+            - expr: histogram_quantile(0.95, sum by(le,instance, stage, pod) (rate(greptime_region_worker_handle_write_bucket[$__rate_interval])))
+              datasource:
+                type: prometheus
+                uid: ${metrics}
+              legendFormat: '[{{instance}}]-[{{pod}}]-[{{stage}}]-P95'
+            - expr: sum by(le,instance, stage, pod) (rate(greptime_region_worker_handle_write_sum[$__rate_interval]))/sum by(le,instance, stage, pod) (rate(greptime_region_worker_handle_write_count[$__rate_interval]))
+              datasource:
+                type: prometheus
+                uid: ${metrics}
+              legendFormat: '[{{instance}}]-[{{pod}}]-[{{stage}}]-AVG'
+        - title: Region Worker Convert Requests
+          type: timeseries
+          description: Per-stage elapsed time for region worker to decode requests.
+          unit: s
+          queries:
+            - expr: histogram_quantile(0.95, sum by(le, instance, stage, pod) (rate(greptime_datanode_convert_region_request_bucket[$__rate_interval])))
+              datasource:
+                type: prometheus
+                uid: ${metrics}
+              legendFormat: '[{{instance}}]-[{{pod}}]-[{{stage}}]-P95'
+            - expr: sum by(le,instance, stage, pod) (rate(greptime_datanode_convert_region_request_sum[$__rate_interval]))/sum by(le,instance, stage, pod) (rate(greptime_datanode_convert_region_request_count[$__rate_interval]))
+              datasource:
+                type: prometheus
+                uid: ${metrics}
+              legendFormat: '[{{instance}}]-[{{pod}}]-[{{stage}}]-AVG'
     - title: OpenDAL
       panels:
         - title: QPS per Instance

grafana/scripts/check.sh

@@ -1,6 +1,6 @@
 #!/usr/bin/env bash
 
-DASHBOARD_DIR=${1:-grafana/dashboards}
+DASHBOARD_DIR=${1:-grafana/dashboards/metrics}
 
 check_dashboard_description() {
   for dashboard in $(find $DASHBOARD_DIR -name "*.json"); do
@@ -25,7 +25,7 @@ check_dashboard_description() {
 check_dashboards_generation() {
   ./grafana/scripts/gen-dashboards.sh
 
-  if [[ -n "$(git diff --name-only grafana/dashboards)" ]]; then
+  if [[ -n "$(git diff --name-only grafana/dashboards/metrics)" ]]; then
     echo "Error: The dashboards are not generated correctly. You should execute the `make dashboards` command."
     exit 1
   fi

grafana/scripts/gen-dashboards.sh

@@ -1,12 +1,12 @@
 #! /usr/bin/env bash
 
-CLUSTER_DASHBOARD_DIR=${1:-grafana/dashboards/cluster}
-STANDALONE_DASHBOARD_DIR=${2:-grafana/dashboards/standalone}
+CLUSTER_DASHBOARD_DIR=${1:-grafana/dashboards/metrics/cluster}
+STANDALONE_DASHBOARD_DIR=${2:-grafana/dashboards/metrics/standalone}
 DAC_IMAGE=ghcr.io/zyy17/dac:20250423-522bd35
 
 remove_instance_filters() {
   # Remove the instance filters for the standalone dashboards.
-  sed 's/instance=~\\"$datanode\\",//; s/instance=~\\"$datanode\\"//; s/instance=~\\"$frontend\\",//; s/instance=~\\"$frontend\\"//; s/instance=~\\"$metasrv\\",//; s/instance=~\\"$metasrv\\"//; s/instance=~\\"$flownode\\",//; s/instance=~\\"$flownode\\"//;' $CLUSTER_DASHBOARD_DIR/dashboard.json > $STANDALONE_DASHBOARD_DIR/dashboard.json
+  sed -E 's/instance=~\\"(\$datanode|\$frontend|\$metasrv|\$flownode)\\",?//g' "$CLUSTER_DASHBOARD_DIR/dashboard.json" > "$STANDALONE_DASHBOARD_DIR/dashboard.json"
 }
 
 generate_intermediate_dashboards_and_docs() {

licenserc.toml

@@ -26,6 +26,13 @@ excludes = [
     "src/common/base/src/secrets.rs",
     "src/servers/src/repeated_field.rs",
     "src/servers/src/http/test_helpers.rs",
+    # enterprise
+    "src/common/meta/src/rpc/ddl/trigger.rs",
+    "src/operator/src/expr_helper/trigger.rs",
+    "src/sql/src/statements/create/trigger.rs",
+    "src/sql/src/statements/show/trigger.rs",
+    "src/sql/src/parsers/create_parser/trigger.rs",
+    "src/sql/src/parsers/show_parser/trigger.rs",
 ]
 
 [properties]

rust-toolchain.toml

@@ -1,2 +1,2 @@
 [toolchain]
-channel = "nightly-2024-12-25"
+channel = "nightly-2025-05-19"

src/api/src/helper.rs

@@ -1050,7 +1050,7 @@ pub fn value_to_grpc_value(value: Value) -> GrpcValue {
             Value::Int64(v) => Some(ValueData::I64Value(v)),
             Value::Float32(v) => Some(ValueData::F32Value(*v)),
             Value::Float64(v) => Some(ValueData::F64Value(*v)),
-            Value::String(v) => Some(ValueData::StringValue(v.as_utf8().to_string())),
+            Value::String(v) => Some(ValueData::StringValue(v.into_string())),
             Value::Binary(v) => Some(ValueData::BinaryValue(v.to_vec())),
             Value::Date(v) => Some(ValueData::DateValue(v.val())),
             Value::Timestamp(v) => Some(match v.unit() {

src/auth/src/common.rs

@@ -36,7 +36,7 @@ pub fn userinfo_by_name(username: Option<String>) -> UserInfoRef {
 }
 
 pub fn user_provider_from_option(opt: &String) -> Result<UserProviderRef> {
-    let (name, content) = opt.split_once(':').context(InvalidConfigSnafu {
+    let (name, content) = opt.split_once(':').with_context(|| InvalidConfigSnafu {
         value: opt.to_string(),
         msg: "UserProviderOption must be in format `<option>:<value>`",
     })?;
@@ -57,6 +57,24 @@ pub fn user_provider_from_option(opt: &String) -> Result<UserProviderRef> {
     }
 }
 
+pub fn static_user_provider_from_option(opt: &String) -> Result<StaticUserProvider> {
+    let (name, content) = opt.split_once(':').with_context(|| InvalidConfigSnafu {
+        value: opt.to_string(),
+        msg: "UserProviderOption must be in format `<option>:<value>`",
+    })?;
+    match name {
+        STATIC_USER_PROVIDER => {
+            let provider = StaticUserProvider::new(content)?;
+            Ok(provider)
+        }
+        _ => InvalidConfigSnafu {
+            value: name.to_string(),
+            msg: format!("Invalid UserProviderOption, expect only {STATIC_USER_PROVIDER}"),
+        }
+        .fail(),
+    }
+}
+
 type Username<'a> = &'a str;
 type HostOrIp<'a> = &'a str;
 

src/auth/src/error.rs

@@ -38,6 +38,14 @@ pub enum Error {
         location: Location,
     },
 
+    #[snafu(display("Failed to convert to utf8"))]
+    FromUtf8 {
+        #[snafu(source)]
+        error: std::string::FromUtf8Error,
+        #[snafu(implicit)]
+        location: Location,
+    },
+
     #[snafu(display("Authentication source failure"))]
     AuthBackend {
         #[snafu(implicit)]
@@ -85,7 +93,7 @@ impl ErrorExt for Error {
     fn status_code(&self) -> StatusCode {
         match self {
             Error::InvalidConfig { .. } => StatusCode::InvalidArguments,
-            Error::IllegalParam { .. } => StatusCode::InvalidArguments,
+            Error::IllegalParam { .. } | Error::FromUtf8 { .. } => StatusCode::InvalidArguments,
             Error::FileWatch { .. } => StatusCode::InvalidArguments,
             Error::InternalState { .. } => StatusCode::Unexpected,
             Error::Io { .. } => StatusCode::StorageUnavailable,

src/auth/src/lib.rs

@@ -22,10 +22,12 @@ mod user_provider;
 pub mod tests;
 
 pub use common::{
-    auth_mysql, user_provider_from_option, userinfo_by_name, HashedPassword, Identity, Password,
+    auth_mysql, static_user_provider_from_option, user_provider_from_option, userinfo_by_name,
+    HashedPassword, Identity, Password,
 };
 pub use permission::{PermissionChecker, PermissionReq, PermissionResp};
 pub use user_info::UserInfo;
+pub use user_provider::static_user_provider::StaticUserProvider;
 pub use user_provider::UserProvider;
 
 /// pub type alias

src/auth/src/user_provider/static_user_provider.rs

@@ -15,15 +15,15 @@
 use std::collections::HashMap;
 
 use async_trait::async_trait;
-use snafu::OptionExt;
+use snafu::{OptionExt, ResultExt};
 
-use crate::error::{InvalidConfigSnafu, Result};
+use crate::error::{FromUtf8Snafu, InvalidConfigSnafu, Result};
 use crate::user_provider::{authenticate_with_credential, load_credential_from_file};
 use crate::{Identity, Password, UserInfoRef, UserProvider};
 
 pub(crate) const STATIC_USER_PROVIDER: &str = "static_user_provider";
 
-pub(crate) struct StaticUserProvider {
+pub struct StaticUserProvider {
     users: HashMap<String, Vec<u8>>,
 }
 
@@ -60,6 +60,18 @@ impl StaticUserProvider {
                 .fail(),
         }
     }
+
+    /// Return a random username/password pair
+    /// This is useful for invoking from other components in the cluster
+    pub fn get_one_user_pwd(&self) -> Result<(String, String)> {
+        let kv = self.users.iter().next().context(InvalidConfigSnafu {
+            value: "",
+            msg: "Expect at least one pair of username and password",
+        })?;
+        let username = kv.0;
+        let pwd = String::from_utf8(kv.1.clone()).context(FromUtf8Snafu)?;
+        Ok((username.clone(), pwd))
+    }
 }
 
 #[async_trait]

src/catalog/src/system_schema/pg_catalog/pg_namespace/oid_map.rs

@@ -84,12 +84,6 @@ mod tests {
         let key1 = "3178510";
         let key2 = "4215648";
 
-        // have collision
-        assert_eq!(
-            oid_map.hasher.hash_one(key1) as u32,
-            oid_map.hasher.hash_one(key2) as u32
-        );
-
         // insert them into oid_map
         let oid1 = oid_map.get_oid(key1);
         let oid2 = oid_map.get_oid(key2);

src/cli/Cargo.toml

@@ -5,8 +5,12 @@ edition.workspace = true
 license.workspace = true
 
 [features]
-pg_kvbackend = ["common-meta/pg_kvbackend"]
-mysql_kvbackend = ["common-meta/mysql_kvbackend"]
+default = [
+    "pg_kvbackend",
+    "mysql_kvbackend",
+]
+pg_kvbackend = ["common-meta/pg_kvbackend", "meta-srv/pg_kvbackend"]
+mysql_kvbackend = ["common-meta/mysql_kvbackend", "meta-srv/mysql_kvbackend"]
 
 [lints]
 workspace = true
@@ -43,15 +47,12 @@ etcd-client.workspace = true
 futures.workspace = true
 humantime.workspace = true
 meta-client.workspace = true
+meta-srv.workspace = true
 nu-ansi-term = "0.46"
-opendal = { version = "0.51.1", features = [
-    "services-fs",
-    "services-s3",
-] }
+object-store.workspace = true
 query.workspace = true
 rand.workspace = true
 reqwest.workspace = true
-rustyline = "10.1"
 serde.workspace = true
 serde_json.workspace = true
 servers.workspace = true

src/cli/src/cmd.rs

@@ -1,154 +0,0 @@
-// 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 crate::error::{Error, InvalidReplCommandSnafu, Result};
-
-/// Represents the parsed command from the user (which may be over many lines)
-#[derive(Debug, PartialEq)]
-pub(crate) enum ReplCommand {
-    Help,
-    UseDatabase { db_name: String },
-    Sql { sql: String },
-    Exit,
-}
-
-impl TryFrom<&str> for ReplCommand {
-    type Error = Error;
-
-    fn try_from(input: &str) -> Result<Self> {
-        let input = input.trim();
-        if input.is_empty() {
-            return InvalidReplCommandSnafu {
-                reason: "No command specified".to_string(),
-            }
-            .fail();
-        }
-
-        // If line ends with ';', it must be treated as a complete input.
-        // However, the opposite is not true.
-        let input_is_completed = input.ends_with(';');
-
-        let input = input.strip_suffix(';').map(|x| x.trim()).unwrap_or(input);
-        let lowercase = input.to_lowercase();
-        match lowercase.as_str() {
-            "help" => Ok(Self::Help),
-            "exit" | "quit" => Ok(Self::Exit),
-            _ => match input.split_once(' ') {
-                Some((maybe_use, database)) if maybe_use.to_lowercase() == "use" => {
-                    Ok(Self::UseDatabase {
-                        db_name: database.trim().to_string(),
-                    })
-                }
-                // Any valid SQL must contains at least one whitespace.
-                Some(_) if input_is_completed => Ok(Self::Sql {
-                    sql: input.to_string(),
-                }),
-                _ => InvalidReplCommandSnafu {
-                    reason: format!("unknown command '{input}', maybe input is not completed"),
-                }
-                .fail(),
-            },
-        }
-    }
-}
-
-impl ReplCommand {
-    pub fn help() -> &'static str {
-        r#"
-Available commands (case insensitive):
-- 'help': print this help
-- 'exit' or 'quit': exit the REPL
-- 'use <your database name>': switch to another database/schema context
-- Other typed in text will be treated as SQL.
-  You can enter new line while typing, just remember to end it with ';'.
-"#
-    }
-}
-
-#[cfg(test)]
-mod tests {
-    use super::*;
-    use crate::error::Error::InvalidReplCommand;
-
-    #[test]
-    fn test_from_str() {
-        fn test_ok(s: &str, expected: ReplCommand) {
-            let actual: ReplCommand = s.try_into().unwrap();
-            assert_eq!(expected, actual, "'{}'", s);
-        }
-
-        fn test_err(s: &str) {
-            let result: Result<ReplCommand> = s.try_into();
-            assert!(matches!(result, Err(InvalidReplCommand { .. })))
-        }
-
-        test_err("");
-        test_err("  ");
-        test_err("\t");
-
-        test_ok("help", ReplCommand::Help);
-        test_ok("help", ReplCommand::Help);
-        test_ok("  help", ReplCommand::Help);
-        test_ok("  help  ", ReplCommand::Help);
-        test_ok("  HELP  ", ReplCommand::Help);
-        test_ok("  Help;  ", ReplCommand::Help);
-        test_ok("  help  ; ", ReplCommand::Help);
-
-        test_ok("exit", ReplCommand::Exit);
-        test_ok("exit;", ReplCommand::Exit);
-        test_ok("exit ;", ReplCommand::Exit);
-        test_ok("EXIT", ReplCommand::Exit);
-
-        test_ok("quit", ReplCommand::Exit);
-        test_ok("quit;", ReplCommand::Exit);
-        test_ok("quit ;", ReplCommand::Exit);
-        test_ok("QUIT", ReplCommand::Exit);
-
-        test_ok(
-            "use Foo",
-            ReplCommand::UseDatabase {
-                db_name: "Foo".to_string(),
-            },
-        );
-        test_ok(
-            "  use Foo ;  ",
-            ReplCommand::UseDatabase {
-                db_name: "Foo".to_string(),
-            },
-        );
-        // ensure that database name is case sensitive
-        test_ok(
-            "  use FOO ;  ",
-            ReplCommand::UseDatabase {
-                db_name: "FOO".to_string(),
-            },
-        );
-
-        // ensure that we aren't messing with capitalization
-        test_ok(
-            "SELECT * from foo;",
-            ReplCommand::Sql {
-                sql: "SELECT * from foo".to_string(),
-            },
-        );
-        // Input line (that don't belong to any other cases above) must ends with ';' to make it a valid SQL.
-        test_err("insert blah");
-        test_ok(
-            "insert blah;",
-            ReplCommand::Sql {
-                sql: "insert blah".to_string(),
-            },
-        );
-    }
-}

src/cli/src/error.rs

@@ -17,6 +17,7 @@ use std::any::Any;
 use common_error::ext::{BoxedError, ErrorExt};
 use common_error::status_code::StatusCode;
 use common_macro::stack_trace_debug;
+use object_store::Error as ObjectStoreError;
 use snafu::{Location, Snafu};
 
 #[derive(Snafu)]
@@ -101,9 +102,6 @@ pub enum Error {
         error: reqwest::Error,
     },
 
-    #[snafu(display("Invalid REPL command: {reason}"))]
-    InvalidReplCommand { reason: String },
-
     #[snafu(display("Failed to parse SQL: {}", sql))]
     ParseSql {
         sql: String,
@@ -228,7 +226,7 @@ pub enum Error {
         #[snafu(implicit)]
         location: Location,
         #[snafu(source)]
-        error: opendal::Error,
+        error: ObjectStoreError,
     },
     #[snafu(display("S3 config need be set"))]
     S3ConfigNotSet {
@@ -240,6 +238,12 @@ pub enum Error {
         #[snafu(implicit)]
         location: Location,
     },
+    #[snafu(display("KV backend not set: {}", backend))]
+    KvBackendNotSet {
+        backend: String,
+        #[snafu(implicit)]
+        location: Location,
+    },
 }
 
 pub type Result<T> = std::result::Result<T, Error>;
@@ -254,7 +258,6 @@ impl ErrorExt for Error {
             Error::MissingConfig { .. }
             | Error::LoadLayeredConfig { .. }
             | Error::IllegalConfig { .. }
-            | Error::InvalidReplCommand { .. }
             | Error::InitTimezone { .. }
             | Error::ConnectEtcd { .. }
             | Error::CreateDir { .. }
@@ -277,8 +280,9 @@ impl ErrorExt for Error {
 
             Error::Other { source, .. } => source.status_code(),
             Error::OpenDal { .. } => StatusCode::Internal,
-            Error::S3ConfigNotSet { .. } => StatusCode::InvalidArguments,
-            Error::OutputDirNotSet { .. } => StatusCode::InvalidArguments,
+            Error::S3ConfigNotSet { .. }
+            | Error::OutputDirNotSet { .. }
+            | Error::KvBackendNotSet { .. } => StatusCode::InvalidArguments,
 
             Error::BuildRuntime { source, .. } => source.status_code(),
 

src/cli/src/export.rs

@@ -19,10 +19,12 @@ use std::time::Duration;
 
 use async_trait::async_trait;
 use clap::{Parser, ValueEnum};
+use common_base::secrets::{ExposeSecret, SecretString};
 use common_error::ext::BoxedError;
 use common_telemetry::{debug, error, info};
-use opendal::layers::LoggingLayer;
-use opendal::{services, Operator};
+use object_store::layers::LoggingLayer;
+use object_store::services::Oss;
+use object_store::{services, ObjectStore};
 use serde_json::Value;
 use snafu::{OptionExt, ResultExt};
 use tokio::sync::Semaphore;
@@ -110,11 +112,26 @@ pub struct ExportCommand {
     #[clap(long)]
     s3: bool,
 
+    /// if both `ddl_local_dir` and remote storage (s3/oss) are set, `ddl_local_dir` will be only used for
+    /// exported SQL files, and the data will be exported to remote storage.
+    ///
+    /// Note that `ddl_local_dir` export sql files to **LOCAL** file system, this is useful if export client don't have
+    /// direct access to remote storage.
+    ///
+    /// if remote storage is set but `ddl_local_dir` is not set, both SQL&data will be exported to remote storage.
+    #[clap(long)]
+    ddl_local_dir: Option<String>,
+
     /// The s3 bucket name
     /// if s3 is set, this is required
     #[clap(long)]
     s3_bucket: Option<String>,
 
+    // The s3 root path
+    /// if s3 is set, this is required
+    #[clap(long)]
+    s3_root: Option<String>,
+
     /// The s3 endpoint
     /// if s3 is set, this is required
     #[clap(long)]
@@ -134,6 +151,30 @@ pub struct ExportCommand {
     /// if s3 is set, this is required
     #[clap(long)]
     s3_region: Option<String>,
+
+    /// if export data to oss
+    #[clap(long)]
+    oss: bool,
+
+    /// The oss bucket name
+    /// if oss is set, this is required
+    #[clap(long)]
+    oss_bucket: Option<String>,
+
+    /// The oss endpoint
+    /// if oss is set, this is required
+    #[clap(long)]
+    oss_endpoint: Option<String>,
+
+    /// The oss access key id
+    /// if oss is set, this is required
+    #[clap(long)]
+    oss_access_key_id: Option<String>,
+
+    /// The oss access key secret
+    /// if oss is set, this is required
+    #[clap(long)]
+    oss_access_key_secret: Option<String>,
 }
 
 impl ExportCommand {
@@ -147,7 +188,7 @@ impl ExportCommand {
         {
             return Err(BoxedError::new(S3ConfigNotSetSnafu {}.build()));
         }
-        if !self.s3 && self.output_dir.is_none() {
+        if !self.s3 && !self.oss && self.output_dir.is_none() {
             return Err(BoxedError::new(OutputDirNotSetSnafu {}.build()));
         }
         let (catalog, schema) =
@@ -172,11 +213,32 @@ impl ExportCommand {
             start_time: self.start_time.clone(),
             end_time: self.end_time.clone(),
             s3: self.s3,
+            ddl_local_dir: self.ddl_local_dir.clone(),
             s3_bucket: self.s3_bucket.clone(),
+            s3_root: self.s3_root.clone(),
             s3_endpoint: self.s3_endpoint.clone(),
-            s3_access_key: self.s3_access_key.clone(),
-            s3_secret_key: self.s3_secret_key.clone(),
+            // Wrap sensitive values in SecretString
+            s3_access_key: self
+                .s3_access_key
+                .as_ref()
+                .map(|k| SecretString::from(k.clone())),
+            s3_secret_key: self
+                .s3_secret_key
+                .as_ref()
+                .map(|k| SecretString::from(k.clone())),
             s3_region: self.s3_region.clone(),
+            oss: self.oss,
+            oss_bucket: self.oss_bucket.clone(),
+            oss_endpoint: self.oss_endpoint.clone(),
+            // Wrap sensitive values in SecretString
+            oss_access_key_id: self
+                .oss_access_key_id
+                .as_ref()
+                .map(|k| SecretString::from(k.clone())),
+            oss_access_key_secret: self
+                .oss_access_key_secret
+                .as_ref()
+                .map(|k| SecretString::from(k.clone())),
         }))
     }
 }
@@ -192,21 +254,30 @@ pub struct Export {
     start_time: Option<String>,
     end_time: Option<String>,
     s3: bool,
+    ddl_local_dir: Option<String>,
     s3_bucket: Option<String>,
+    s3_root: Option<String>,
     s3_endpoint: Option<String>,
-    s3_access_key: Option<String>,
-    s3_secret_key: Option<String>,
+    // Changed to SecretString for sensitive data
+    s3_access_key: Option<SecretString>,
+    s3_secret_key: Option<SecretString>,
     s3_region: Option<String>,
+    oss: bool,
+    oss_bucket: Option<String>,
+    oss_endpoint: Option<String>,
+    // Changed to SecretString for sensitive data
+    oss_access_key_id: Option<SecretString>,
+    oss_access_key_secret: Option<SecretString>,
 }
 
 impl Export {
     fn catalog_path(&self) -> PathBuf {
-        if self.s3 {
+        if self.s3 || self.oss {
             PathBuf::from(&self.catalog)
         } else if let Some(dir) = &self.output_dir {
             PathBuf::from(dir).join(&self.catalog)
         } else {
-            unreachable!("catalog_path: output_dir must be set when not using s3")
+            unreachable!("catalog_path: output_dir must be set when not using remote storage")
         }
     }
 
@@ -364,7 +435,7 @@ impl Export {
         let timer = Instant::now();
         let db_names = self.get_db_names().await?;
         let db_count = db_names.len();
-        let operator = self.build_operator().await?;
+        let operator = self.build_prefer_fs_operator().await?;
 
         for schema in db_names {
             let create_database = self
@@ -394,7 +465,7 @@ impl Export {
         let semaphore = Arc::new(Semaphore::new(self.parallelism));
         let db_names = self.get_db_names().await?;
         let db_count = db_names.len();
-        let operator = Arc::new(self.build_operator().await?);
+        let operator = Arc::new(self.build_prefer_fs_operator().await?);
         let mut tasks = Vec::with_capacity(db_names.len());
 
         for schema in db_names {
@@ -408,7 +479,7 @@ impl Export {
                     .await?;
 
                 // Create directory if needed for file system storage
-                if !export_self.s3 {
+                if !export_self.s3 && !export_self.oss {
                     let db_dir = format!("{}/{}/", export_self.catalog, schema);
                     operator.create_dir(&db_dir).await.context(OpenDalSnafu)?;
                 }
@@ -451,21 +522,45 @@ impl Export {
         Ok(())
     }
 
-    async fn build_operator(&self) -> Result<Operator> {
+    async fn build_operator(&self) -> Result<ObjectStore> {
         if self.s3 {
             self.build_s3_operator().await
+        } else if self.oss {
+            self.build_oss_operator().await
         } else {
             self.build_fs_operator().await
         }
     }
 
-    async fn build_s3_operator(&self) -> Result<Operator> {
-        let mut builder = services::S3::default().root("").bucket(
+    /// build operator with preference for file system
+    async fn build_prefer_fs_operator(&self) -> Result<ObjectStore> {
+        if (self.s3 || self.oss) && self.ddl_local_dir.is_some() {
+            let root = self.ddl_local_dir.as_ref().unwrap().clone();
+            let op = ObjectStore::new(services::Fs::default().root(&root))
+                .context(OpenDalSnafu)?
+                .layer(LoggingLayer::default())
+                .finish();
+            Ok(op)
+        } else if self.s3 {
+            self.build_s3_operator().await
+        } else if self.oss {
+            self.build_oss_operator().await
+        } else {
+            self.build_fs_operator().await
+        }
+    }
+
+    async fn build_s3_operator(&self) -> Result<ObjectStore> {
+        let mut builder = services::S3::default().bucket(
             self.s3_bucket
                 .as_ref()
                 .expect("s3_bucket must be provided when s3 is enabled"),
         );
 
+        if let Some(root) = self.s3_root.as_ref() {
+            builder = builder.root(root);
+        }
+
         if let Some(endpoint) = self.s3_endpoint.as_ref() {
             builder = builder.endpoint(endpoint);
         }
@@ -475,27 +570,51 @@ impl Export {
         }
 
         if let Some(key_id) = self.s3_access_key.as_ref() {
-            builder = builder.access_key_id(key_id);
+            builder = builder.access_key_id(key_id.expose_secret());
         }
 
         if let Some(secret_key) = self.s3_secret_key.as_ref() {
-            builder = builder.secret_access_key(secret_key);
+            builder = builder.secret_access_key(secret_key.expose_secret());
         }
 
-        let op = Operator::new(builder)
+        let op = ObjectStore::new(builder)
             .context(OpenDalSnafu)?
             .layer(LoggingLayer::default())
             .finish();
         Ok(op)
     }
 
-    async fn build_fs_operator(&self) -> Result<Operator> {
+    async fn build_oss_operator(&self) -> Result<ObjectStore> {
+        let mut builder = Oss::default()
+            .bucket(self.oss_bucket.as_ref().expect("oss_bucket must be set"))
+            .endpoint(
+                self.oss_endpoint
+                    .as_ref()
+                    .expect("oss_endpoint must be set"),
+            );
+
+        // Use expose_secret() to access the actual secret value
+        if let Some(key_id) = self.oss_access_key_id.as_ref() {
+            builder = builder.access_key_id(key_id.expose_secret());
+        }
+        if let Some(secret_key) = self.oss_access_key_secret.as_ref() {
+            builder = builder.access_key_secret(secret_key.expose_secret());
+        }
+
+        let op = ObjectStore::new(builder)
+            .context(OpenDalSnafu)?
+            .layer(LoggingLayer::default())
+            .finish();
+        Ok(op)
+    }
+
+    async fn build_fs_operator(&self) -> Result<ObjectStore> {
         let root = self
             .output_dir
             .as_ref()
             .context(OutputDirNotSetSnafu)?
             .clone();
-        let op = Operator::new(services::Fs::default().root(&root))
+        let op = ObjectStore::new(services::Fs::default().root(&root))
             .context(OpenDalSnafu)?
             .layer(LoggingLayer::default())
             .finish();
@@ -509,6 +628,7 @@ impl Export {
         let db_count = db_names.len();
         let mut tasks = Vec::with_capacity(db_count);
         let operator = Arc::new(self.build_operator().await?);
+        let fs_first_operator = Arc::new(self.build_prefer_fs_operator().await?);
         let with_options = build_with_options(&self.start_time, &self.end_time);
 
         for schema in db_names {
@@ -516,12 +636,13 @@ impl Export {
             let export_self = self.clone();
             let with_options_clone = with_options.clone();
             let operator = operator.clone();
+            let fs_first_operator = fs_first_operator.clone();
 
             tasks.push(async move {
                 let _permit = semaphore_moved.acquire().await.unwrap();
 
-                // Create directory if not using S3
-                if !export_self.s3 {
+                // Create directory if not using remote storage
+                if !export_self.s3 && !export_self.oss {
                     let db_dir = format!("{}/{}/", export_self.catalog, schema);
                     operator.create_dir(&db_dir).await.context(OpenDalSnafu)?;
                 }
@@ -533,7 +654,11 @@ impl Export {
                     r#"COPY DATABASE "{}"."{}" TO '{}' WITH ({}){};"#,
                     export_self.catalog, schema, path, with_options_clone, connection_part
                 );
-                info!("Executing sql: {sql}");
+
+                // Log SQL command but mask sensitive information
+                let safe_sql = export_self.mask_sensitive_sql(&sql);
+                info!("Executing sql: {}", safe_sql);
+
                 export_self.database_client.sql_in_public(&sql).await?;
                 info!(
                     "Finished exporting {}.{} data to {}",
@@ -549,7 +674,7 @@ impl Export {
                 let copy_from_path = export_self.get_file_path(&schema, "copy_from.sql");
                 export_self
                     .write_to_storage(
-                        &operator,
+                        &fs_first_operator,
                         &copy_from_path,
                         copy_database_from_sql.into_bytes(),
                     )
@@ -573,6 +698,29 @@ impl Export {
         Ok(())
     }
 
+    /// Mask sensitive information in SQL commands for safe logging
+    fn mask_sensitive_sql(&self, sql: &str) -> String {
+        let mut masked_sql = sql.to_string();
+
+        // Mask S3 credentials
+        if let Some(access_key) = &self.s3_access_key {
+            masked_sql = masked_sql.replace(access_key.expose_secret(), "[REDACTED]");
+        }
+        if let Some(secret_key) = &self.s3_secret_key {
+            masked_sql = masked_sql.replace(secret_key.expose_secret(), "[REDACTED]");
+        }
+
+        // Mask OSS credentials
+        if let Some(access_key_id) = &self.oss_access_key_id {
+            masked_sql = masked_sql.replace(access_key_id.expose_secret(), "[REDACTED]");
+        }
+        if let Some(access_key_secret) = &self.oss_access_key_secret {
+            masked_sql = masked_sql.replace(access_key_secret.expose_secret(), "[REDACTED]");
+        }
+
+        masked_sql
+    }
+
     fn get_file_path(&self, schema: &str, file_name: &str) -> String {
         format!("{}/{}/{}", self.catalog, schema, file_name)
     }
@@ -580,8 +728,20 @@ impl Export {
     fn format_output_path(&self, file_path: &str) -> String {
         if self.s3 {
             format!(
-                "s3://{}/{}",
+                "s3://{}{}/{}",
                 self.s3_bucket.as_ref().unwrap_or(&String::new()),
+                if let Some(root) = &self.s3_root {
+                    format!("/{}", root)
+                } else {
+                    String::new()
+                },
+                file_path
+            )
+        } else if self.oss {
+            format!(
+                "oss://{}/{}/{}",
+                self.oss_bucket.as_ref().unwrap_or(&String::new()),
+                self.catalog,
                 file_path
             )
         } else {
@@ -595,19 +755,27 @@ impl Export {
 
     async fn write_to_storage(
         &self,
-        op: &Operator,
+        op: &ObjectStore,
         file_path: &str,
         content: Vec<u8>,
     ) -> Result<()> {
-        op.write(file_path, content).await.context(OpenDalSnafu)
+        op.write(file_path, content)
+            .await
+            .context(OpenDalSnafu)
+            .map(|_| ())
     }
 
     fn get_storage_params(&self, schema: &str) -> (String, String) {
         if self.s3 {
             let s3_path = format!(
-                "s3://{}/{}/{}/",
+                "s3://{}{}/{}/{}/",
                 // Safety: s3_bucket is required when s3 is enabled
                 self.s3_bucket.as_ref().unwrap(),
+                if let Some(root) = &self.s3_root {
+                    format!("/{}", root)
+                } else {
+                    String::new()
+                },
                 self.catalog,
                 schema
             );
@@ -620,15 +788,36 @@ impl Export {
             };
 
             // Safety: All s3 options are required
+            // Use expose_secret() to access the actual secret values
             let connection_options = format!(
                 "ACCESS_KEY_ID='{}', SECRET_ACCESS_KEY='{}', REGION='{}'{}",
-                self.s3_access_key.as_ref().unwrap(),
-                self.s3_secret_key.as_ref().unwrap(),
+                self.s3_access_key.as_ref().unwrap().expose_secret(),
+                self.s3_secret_key.as_ref().unwrap().expose_secret(),
                 self.s3_region.as_ref().unwrap(),
                 endpoint_option
             );
 
             (s3_path, format!(" CONNECTION ({})", connection_options))
+        } else if self.oss {
+            let oss_path = format!(
+                "oss://{}/{}/{}/",
+                self.oss_bucket.as_ref().unwrap(),
+                self.catalog,
+                schema
+            );
+            let endpoint_option = if let Some(endpoint) = self.oss_endpoint.as_ref() {
+                format!(", ENDPOINT='{}'", endpoint)
+            } else {
+                String::new()
+            };
+
+            let connection_options = format!(
+                "ACCESS_KEY_ID='{}', ACCESS_KEY_SECRET='{}'{}",
+                self.oss_access_key_id.as_ref().unwrap().expose_secret(),
+                self.oss_access_key_secret.as_ref().unwrap().expose_secret(),
+                endpoint_option
+            );
+            (oss_path, format!(" CONNECTION ({})", connection_options))
         } else {
             (
                 self.catalog_path()

src/cli/src/helper.rs

@@ -1,112 +0,0 @@
-// 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 rustyline::completion::Completer;
-use rustyline::highlight::{Highlighter, MatchingBracketHighlighter};
-use rustyline::hint::{Hinter, HistoryHinter};
-use rustyline::validate::{ValidationContext, ValidationResult, Validator};
-
-use crate::cmd::ReplCommand;
-
-pub(crate) struct RustylineHelper {
-    hinter: HistoryHinter,
-    highlighter: MatchingBracketHighlighter,
-}
-
-impl Default for RustylineHelper {
-    fn default() -> Self {
-        Self {
-            hinter: HistoryHinter {},
-            highlighter: MatchingBracketHighlighter::default(),
-        }
-    }
-}
-
-impl rustyline::Helper for RustylineHelper {}
-
-impl Validator for RustylineHelper {
-    fn validate(&self, ctx: &mut ValidationContext<'_>) -> rustyline::Result<ValidationResult> {
-        let input = ctx.input();
-        match ReplCommand::try_from(input) {
-            Ok(_) => Ok(ValidationResult::Valid(None)),
-            Err(e) => {
-                if input.trim_end().ends_with(';') {
-                    // If line ends with ';', it HAS to be a valid command.
-                    Ok(ValidationResult::Invalid(Some(e.to_string())))
-                } else {
-                    Ok(ValidationResult::Incomplete)
-                }
-            }
-        }
-    }
-}
-
-impl Hinter for RustylineHelper {
-    type Hint = String;
-
-    fn hint(&self, line: &str, pos: usize, ctx: &rustyline::Context<'_>) -> Option<Self::Hint> {
-        self.hinter.hint(line, pos, ctx)
-    }
-}
-
-impl Highlighter for RustylineHelper {
-    fn highlight<'l>(&self, line: &'l str, pos: usize) -> Cow<'l, str> {
-        self.highlighter.highlight(line, pos)
-    }
-
-    fn highlight_prompt<'b, 's: 'b, 'p: 'b>(
-        &'s self,
-        prompt: &'p str,
-        default: bool,
-    ) -> Cow<'b, str> {
-        self.highlighter.highlight_prompt(prompt, default)
-    }
-
-    fn highlight_hint<'h>(&self, hint: &'h str) -> Cow<'h, str> {
-        use nu_ansi_term::Style;
-        Cow::Owned(Style::new().dimmed().paint(hint).to_string())
-    }
-
-    fn highlight_candidate<'c>(
-        &self,
-        candidate: &'c str,
-        completion: rustyline::CompletionType,
-    ) -> Cow<'c, str> {
-        self.highlighter.highlight_candidate(candidate, completion)
-    }
-
-    fn highlight_char(&self, line: &str, pos: usize) -> bool {
-        self.highlighter.highlight_char(line, pos)
-    }
-}
-
-impl Completer for RustylineHelper {
-    type Candidate = String;
-
-    fn complete(
-        &self,
-        line: &str,
-        pos: usize,
-        ctx: &rustyline::Context<'_>,
-    ) -> rustyline::Result<(usize, Vec<Self::Candidate>)> {
-        // If there is a hint, use that as the auto-complete when user hits `tab`
-        if let Some(hint) = self.hinter.hint(line, pos, ctx) {
-            Ok((pos, vec![hint]))
-        } else {
-            Ok((0, vec![]))
-        }
-    }
-}

src/cli/src/lib.rs

@@ -13,16 +13,11 @@
 // 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;
+pub mod error;
+mod export;
 mod import;
+mod meta_snapshot;
 
 use async_trait::async_trait;
 use clap::Parser;
@@ -33,6 +28,7 @@ use error::Result;
 pub use crate::bench::BenchTableMetadataCommand;
 pub use crate::export::ExportCommand;
 pub use crate::import::ImportCommand;
+pub use crate::meta_snapshot::{MetaRestoreCommand, MetaSnapshotCommand};
 
 #[async_trait]
 pub trait Tool: Send + Sync {

src/cli/src/meta_snapshot.rs

@@ -0,0 +1,329 @@
+// 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 async_trait::async_trait;
+use clap::Parser;
+use common_base::secrets::{ExposeSecret, SecretString};
+use common_error::ext::BoxedError;
+use common_meta::kv_backend::chroot::ChrootKvBackend;
+use common_meta::kv_backend::etcd::EtcdStore;
+use common_meta::kv_backend::KvBackendRef;
+use common_meta::snapshot::MetadataSnapshotManager;
+use meta_srv::bootstrap::create_etcd_client;
+use meta_srv::metasrv::BackendImpl;
+use object_store::services::{Fs, S3};
+use object_store::ObjectStore;
+use snafu::ResultExt;
+
+use crate::error::{KvBackendNotSetSnafu, OpenDalSnafu, S3ConfigNotSetSnafu};
+use crate::Tool;
+#[derive(Debug, Default, Parser)]
+struct MetaConnection {
+    /// The endpoint of store. one of etcd, pg or mysql.
+    #[clap(long, alias = "store-addr", value_delimiter = ',', num_args = 1..)]
+    store_addrs: Vec<String>,
+    /// The database backend.
+    #[clap(long, value_enum)]
+    backend: Option<BackendImpl>,
+    #[clap(long, default_value = "")]
+    store_key_prefix: String,
+    #[cfg(any(feature = "pg_kvbackend", feature = "mysql_kvbackend"))]
+    #[clap(long,default_value = common_meta::kv_backend::DEFAULT_META_TABLE_NAME)]
+    meta_table_name: String,
+    #[clap(long, default_value = "128")]
+    max_txn_ops: usize,
+}
+
+impl MetaConnection {
+    pub async fn build(&self) -> Result<KvBackendRef, BoxedError> {
+        let max_txn_ops = self.max_txn_ops;
+        let store_addrs = &self.store_addrs;
+        if store_addrs.is_empty() {
+            KvBackendNotSetSnafu { backend: "all" }
+                .fail()
+                .map_err(BoxedError::new)
+        } else {
+            let kvbackend = match self.backend {
+                Some(BackendImpl::EtcdStore) => {
+                    let etcd_client = create_etcd_client(store_addrs)
+                        .await
+                        .map_err(BoxedError::new)?;
+                    Ok(EtcdStore::with_etcd_client(etcd_client, max_txn_ops))
+                }
+                #[cfg(feature = "pg_kvbackend")]
+                Some(BackendImpl::PostgresStore) => {
+                    let table_name = &self.meta_table_name;
+                    let pool = meta_srv::bootstrap::create_postgres_pool(store_addrs)
+                        .await
+                        .map_err(BoxedError::new)?;
+                    Ok(common_meta::kv_backend::rds::PgStore::with_pg_pool(
+                        pool,
+                        table_name,
+                        max_txn_ops,
+                    )
+                    .await
+                    .map_err(BoxedError::new)?)
+                }
+                #[cfg(feature = "mysql_kvbackend")]
+                Some(BackendImpl::MysqlStore) => {
+                    let table_name = &self.meta_table_name;
+                    let pool = meta_srv::bootstrap::create_mysql_pool(store_addrs)
+                        .await
+                        .map_err(BoxedError::new)?;
+                    Ok(common_meta::kv_backend::rds::MySqlStore::with_mysql_pool(
+                        pool,
+                        table_name,
+                        max_txn_ops,
+                    )
+                    .await
+                    .map_err(BoxedError::new)?)
+                }
+                _ => KvBackendNotSetSnafu { backend: "all" }
+                    .fail()
+                    .map_err(BoxedError::new),
+            };
+            if self.store_key_prefix.is_empty() {
+                kvbackend
+            } else {
+                let chroot_kvbackend =
+                    ChrootKvBackend::new(self.store_key_prefix.as_bytes().to_vec(), kvbackend?);
+                Ok(Arc::new(chroot_kvbackend))
+            }
+        }
+    }
+}
+
+// TODO(qtang): Abstract a generic s3 config for export import meta snapshot restore
+#[derive(Debug, Default, Parser)]
+struct S3Config {
+    /// whether to use s3 as the output directory. default is false.
+    #[clap(long, default_value = "false")]
+    s3: bool,
+    /// The s3 bucket name.
+    #[clap(long)]
+    s3_bucket: Option<String>,
+    /// The s3 region.
+    #[clap(long)]
+    s3_region: Option<String>,
+    /// The s3 access key.
+    #[clap(long)]
+    s3_access_key: Option<SecretString>,
+    /// The s3 secret key.
+    #[clap(long)]
+    s3_secret_key: Option<SecretString>,
+    /// The s3 endpoint. we will automatically use the default s3 decided by the region if not set.
+    #[clap(long)]
+    s3_endpoint: Option<String>,
+}
+
+impl S3Config {
+    pub fn build(&self, root: &str) -> Result<Option<ObjectStore>, BoxedError> {
+        if !self.s3 {
+            Ok(None)
+        } else {
+            if self.s3_region.is_none()
+                || self.s3_access_key.is_none()
+                || self.s3_secret_key.is_none()
+                || self.s3_bucket.is_none()
+            {
+                return S3ConfigNotSetSnafu.fail().map_err(BoxedError::new);
+            }
+            // Safety, unwrap is safe because we have checked the options above.
+            let mut config = S3::default()
+                .bucket(self.s3_bucket.as_ref().unwrap())
+                .region(self.s3_region.as_ref().unwrap())
+                .access_key_id(self.s3_access_key.as_ref().unwrap().expose_secret())
+                .secret_access_key(self.s3_secret_key.as_ref().unwrap().expose_secret());
+
+            if !root.is_empty() && root != "." {
+                config = config.root(root);
+            }
+
+            if let Some(endpoint) = &self.s3_endpoint {
+                config = config.endpoint(endpoint);
+            }
+            Ok(Some(
+                ObjectStore::new(config)
+                    .context(OpenDalSnafu)
+                    .map_err(BoxedError::new)?
+                    .finish(),
+            ))
+        }
+    }
+}
+
+/// Export metadata snapshot tool.
+/// This tool is used to export metadata snapshot from etcd, pg or mysql.
+/// It will dump the metadata snapshot to local file or s3 bucket.
+/// The snapshot file will be in binary format.
+#[derive(Debug, Default, Parser)]
+pub struct MetaSnapshotCommand {
+    /// The connection to the metadata store.
+    #[clap(flatten)]
+    connection: MetaConnection,
+    /// The s3 config.
+    #[clap(flatten)]
+    s3_config: S3Config,
+    /// The name of the target snapshot file. we will add the file extension automatically.
+    #[clap(long, default_value = "metadata_snapshot")]
+    file_name: String,
+    /// The directory to store the snapshot file.
+    /// if target output is s3 bucket, this is the root directory in the bucket.
+    /// if target output is local file, this is the local directory.
+    #[clap(long, default_value = "")]
+    output_dir: String,
+}
+
+fn create_local_file_object_store(root: &str) -> Result<ObjectStore, BoxedError> {
+    let root = if root.is_empty() { "." } else { root };
+    let object_store = ObjectStore::new(Fs::default().root(root))
+        .context(OpenDalSnafu)
+        .map_err(BoxedError::new)?
+        .finish();
+    Ok(object_store)
+}
+
+impl MetaSnapshotCommand {
+    pub async fn build(&self) -> Result<Box<dyn Tool>, BoxedError> {
+        let kvbackend = self.connection.build().await?;
+        let output_dir = &self.output_dir;
+        let object_store = self.s3_config.build(output_dir).map_err(BoxedError::new)?;
+        if let Some(store) = object_store {
+            let tool = MetaSnapshotTool {
+                inner: MetadataSnapshotManager::new(kvbackend, store),
+                target_file: self.file_name.clone(),
+            };
+            Ok(Box::new(tool))
+        } else {
+            let object_store = create_local_file_object_store(output_dir)?;
+            let tool = MetaSnapshotTool {
+                inner: MetadataSnapshotManager::new(kvbackend, object_store),
+                target_file: self.file_name.clone(),
+            };
+            Ok(Box::new(tool))
+        }
+    }
+}
+
+pub struct MetaSnapshotTool {
+    inner: MetadataSnapshotManager,
+    target_file: String,
+}
+
+#[async_trait]
+impl Tool for MetaSnapshotTool {
+    async fn do_work(&self) -> std::result::Result<(), BoxedError> {
+        self.inner
+            .dump("", &self.target_file)
+            .await
+            .map_err(BoxedError::new)?;
+        Ok(())
+    }
+}
+
+/// Restore metadata snapshot tool.
+/// This tool is used to restore metadata snapshot from etcd, pg or mysql.
+/// It will restore the metadata snapshot from local file or s3 bucket.
+#[derive(Debug, Default, Parser)]
+pub struct MetaRestoreCommand {
+    /// The connection to the metadata store.
+    #[clap(flatten)]
+    connection: MetaConnection,
+    /// The s3 config.
+    #[clap(flatten)]
+    s3_config: S3Config,
+    /// The name of the target snapshot file.
+    #[clap(long, default_value = "metadata_snapshot.metadata.fb")]
+    file_name: String,
+    /// The directory to store the snapshot file.
+    #[clap(long, default_value = ".")]
+    input_dir: String,
+    #[clap(long, default_value = "false")]
+    force: bool,
+}
+
+impl MetaRestoreCommand {
+    pub async fn build(&self) -> Result<Box<dyn Tool>, BoxedError> {
+        let kvbackend = self.connection.build().await?;
+        let input_dir = &self.input_dir;
+        let object_store = self.s3_config.build(input_dir).map_err(BoxedError::new)?;
+        if let Some(store) = object_store {
+            let tool = MetaRestoreTool::new(
+                MetadataSnapshotManager::new(kvbackend, store),
+                self.file_name.clone(),
+                self.force,
+            );
+            Ok(Box::new(tool))
+        } else {
+            let object_store = create_local_file_object_store(input_dir)?;
+            let tool = MetaRestoreTool::new(
+                MetadataSnapshotManager::new(kvbackend, object_store),
+                self.file_name.clone(),
+                self.force,
+            );
+            Ok(Box::new(tool))
+        }
+    }
+}
+
+pub struct MetaRestoreTool {
+    inner: MetadataSnapshotManager,
+    source_file: String,
+    force: bool,
+}
+
+impl MetaRestoreTool {
+    pub fn new(inner: MetadataSnapshotManager, source_file: String, force: bool) -> Self {
+        Self {
+            inner,
+            source_file,
+            force,
+        }
+    }
+}
+
+#[async_trait]
+impl Tool for MetaRestoreTool {
+    async fn do_work(&self) -> std::result::Result<(), BoxedError> {
+        let clean = self
+            .inner
+            .check_target_source_clean()
+            .await
+            .map_err(BoxedError::new)?;
+        if clean {
+            common_telemetry::info!(
+                "The target source is clean, we will restore the metadata snapshot."
+            );
+            self.inner
+                .restore(&self.source_file)
+                .await
+                .map_err(BoxedError::new)?;
+            Ok(())
+        } else if !self.force {
+            common_telemetry::warn!(
+                 "The target source is not clean, if you want to restore the metadata snapshot forcefully, please use --force option."
+             );
+            Ok(())
+        } else {
+            common_telemetry::info!("The target source is not clean, We will restore the metadata snapshot with --force.");
+            self.inner
+                .restore(&self.source_file)
+                .await
+                .map_err(BoxedError::new)?;
+            Ok(())
+        }
+    }
+}

src/client/Cargo.toml

@@ -25,6 +25,7 @@ common-meta.workspace = true
 common-query.workspace = true
 common-recordbatch.workspace = true
 common-telemetry.workspace = true
+datatypes.workspace = true
 enum_dispatch = "0.3"
 futures.workspace = true
 futures-util.workspace = true

src/client/src/database.rs

@@ -14,6 +14,7 @@
 
 use std::pin::Pin;
 use std::str::FromStr;
+use std::sync::Arc;
 
 use api::v1::auth_header::AuthScheme;
 use api::v1::ddl_request::Expr as DdlExpr;
@@ -35,21 +36,21 @@ use common_grpc::flight::do_put::DoPutResponse;
 use common_grpc::flight::{FlightDecoder, FlightMessage};
 use common_query::Output;
 use common_recordbatch::error::ExternalSnafu;
-use common_recordbatch::RecordBatchStreamWrapper;
-use common_telemetry::error;
+use common_recordbatch::{RecordBatch, RecordBatchStreamWrapper};
 use common_telemetry::tracing_context::W3cTrace;
+use common_telemetry::{error, warn};
 use futures::future;
 use futures_util::{Stream, StreamExt, TryStreamExt};
 use prost::Message;
 use snafu::{ensure, ResultExt};
-use tonic::metadata::{AsciiMetadataKey, MetadataValue};
+use tonic::metadata::{AsciiMetadataKey, AsciiMetadataValue, MetadataMap, MetadataValue};
 use tonic::transport::Channel;
 
 use crate::error::{
-    ConvertFlightDataSnafu, Error, FlightGetSnafu, IllegalFlightMessagesSnafu, InvalidAsciiSnafu,
+    ConvertFlightDataSnafu, Error, FlightGetSnafu, IllegalFlightMessagesSnafu,
     InvalidTonicMetadataValueSnafu, ServerSnafu,
 };
-use crate::{from_grpc_response, Client, Result};
+use crate::{error, from_grpc_response, Client, Result};
 
 type FlightDataStream = Pin<Box<dyn Stream<Item = FlightData> + Send>>;
 
@@ -165,26 +166,27 @@ impl Database {
 
         let mut request = tonic::Request::new(request);
         let metadata = request.metadata_mut();
-        for (key, value) in hints {
-            let key = AsciiMetadataKey::from_bytes(format!("x-greptime-hint-{}", key).as_bytes())
-                .map_err(|_| {
-                InvalidAsciiSnafu {
-                    value: key.to_string(),
-                }
-                .build()
-            })?;
-            let value = value.parse().map_err(|_| {
-                InvalidAsciiSnafu {
-                    value: value.to_string(),
-                }
-                .build()
-            })?;
-            metadata.insert(key, value);
-        }
+        Self::put_hints(metadata, hints)?;
+
         let response = client.handle(request).await?.into_inner();
         from_grpc_response(response)
     }
 
+    fn put_hints(metadata: &mut MetadataMap, hints: &[(&str, &str)]) -> Result<()> {
+        let Some(value) = hints
+            .iter()
+            .map(|(k, v)| format!("{}={}", k, v))
+            .reduce(|a, b| format!("{},{}", a, b))
+        else {
+            return Ok(());
+        };
+
+        let key = AsciiMetadataKey::from_static("x-greptime-hints");
+        let value = AsciiMetadataValue::from_str(&value).context(InvalidTonicMetadataValueSnafu)?;
+        metadata.insert(key, value);
+        Ok(())
+    }
+
     pub async fn handle(&self, request: Request) -> Result<u32> {
         let mut client = make_database_client(&self.client)?.inner;
         let request = self.to_rpc_request(request);
@@ -192,6 +194,36 @@ impl Database {
         from_grpc_response(response)
     }
 
+    /// Retry if connection fails, max_retries is the max number of retries, so the total wait time
+    /// is `max_retries * GRPC_CONN_TIMEOUT`
+    pub async fn handle_with_retry(&self, request: Request, max_retries: u32) -> Result<u32> {
+        let mut client = make_database_client(&self.client)?.inner;
+        let mut retries = 0;
+        let request = self.to_rpc_request(request);
+        loop {
+            let raw_response = client.handle(request.clone()).await;
+            match (raw_response, retries < max_retries) {
+                (Ok(resp), _) => return from_grpc_response(resp.into_inner()),
+                (Err(err), true) => {
+                    // determine if the error is retryable
+                    if is_grpc_retryable(&err) {
+                        // retry
+                        retries += 1;
+                        warn!("Retrying {} times with error = {:?}", retries, err);
+                        continue;
+                    }
+                }
+                (Err(err), false) => {
+                    error!(
+                        "Failed to send request to grpc handle after {} retries, error = {:?}",
+                        retries, err
+                    );
+                    return Err(err.into());
+                }
+            }
+        }
+    }
+
     #[inline]
     fn to_rpc_request(&self, request: Request) -> GreptimeRequest {
         GreptimeRequest {
@@ -212,39 +244,49 @@ impl Database {
     where
         S: AsRef<str>,
     {
-        self.do_get(Request::Query(QueryRequest {
+        self.sql_with_hint(sql, &[]).await
+    }
+
+    pub async fn sql_with_hint<S>(&self, sql: S, hints: &[(&str, &str)]) -> Result<Output>
+    where
+        S: AsRef<str>,
+    {
+        let request = Request::Query(QueryRequest {
             query: Some(Query::Sql(sql.as_ref().to_string())),
-        }))
-        .await
+        });
+        self.do_get(request, hints).await
     }
 
     pub async fn logical_plan(&self, logical_plan: Vec<u8>) -> Result<Output> {
-        self.do_get(Request::Query(QueryRequest {
+        let request = Request::Query(QueryRequest {
             query: Some(Query::LogicalPlan(logical_plan)),
-        }))
-        .await
+        });
+        self.do_get(request, &[]).await
     }
 
     pub async fn create(&self, expr: CreateTableExpr) -> Result<Output> {
-        self.do_get(Request::Ddl(DdlRequest {
+        let request = Request::Ddl(DdlRequest {
             expr: Some(DdlExpr::CreateTable(expr)),
-        }))
-        .await
+        });
+        self.do_get(request, &[]).await
     }
 
     pub async fn alter(&self, expr: AlterTableExpr) -> Result<Output> {
-        self.do_get(Request::Ddl(DdlRequest {
+        let request = Request::Ddl(DdlRequest {
             expr: Some(DdlExpr::AlterTable(expr)),
-        }))
-        .await
+        });
+        self.do_get(request, &[]).await
     }
 
-    async fn do_get(&self, request: Request) -> Result<Output> {
+    async fn do_get(&self, request: Request, hints: &[(&str, &str)]) -> Result<Output> {
         let request = self.to_rpc_request(request);
         let request = Ticket {
             ticket: request.encode_to_vec().into(),
         };
 
+        let mut request = tonic::Request::new(request);
+        Self::put_hints(request.metadata_mut(), hints)?;
+
         let mut client = self.client.make_flight_client()?;
 
         let response = client.mut_inner().do_get(request).await.or_else(|e| {
@@ -274,7 +316,7 @@ impl Database {
         let mut flight_message_stream = flight_data_stream.map(move |flight_data| {
             flight_data
                 .map_err(Error::from)
-                .and_then(|data| decoder.try_decode(data).context(ConvertFlightDataSnafu))
+                .and_then(|data| decoder.try_decode(&data).context(ConvertFlightDataSnafu))
         });
 
         let Some(first_flight_message) = flight_message_stream.next().await else {
@@ -296,20 +338,30 @@ impl Database {
                 );
                 Ok(Output::new_with_affected_rows(rows))
             }
-            FlightMessage::Recordbatch(_) | FlightMessage::Metrics(_) => {
+            FlightMessage::RecordBatch(_) | FlightMessage::Metrics(_) => {
                 IllegalFlightMessagesSnafu {
                     reason: "The first flight message cannot be a RecordBatch or Metrics message",
                 }
                 .fail()
             }
             FlightMessage::Schema(schema) => {
+                let schema = Arc::new(
+                    datatypes::schema::Schema::try_from(schema)
+                        .context(error::ConvertSchemaSnafu)?,
+                );
+                let schema_cloned = schema.clone();
                 let stream = Box::pin(stream!({
                     while let Some(flight_message) = flight_message_stream.next().await {
                         let flight_message = flight_message
                             .map_err(BoxedError::new)
                             .context(ExternalSnafu)?;
                         match flight_message {
-                            FlightMessage::Recordbatch(record_batch) => yield Ok(record_batch),
+                            FlightMessage::RecordBatch(arrow_batch) => {
+                                yield RecordBatch::try_from_df_record_batch(
+                                    schema_cloned.clone(),
+                                    arrow_batch,
+                                )
+                            }
                             FlightMessage::Metrics(_) => {}
                             FlightMessage::AffectedRows(_) | FlightMessage::Schema(_) => {
                                 yield IllegalFlightMessagesSnafu {reason: format!("A Schema message must be succeeded exclusively by a set of RecordBatch messages, flight_message: {:?}", flight_message)}
@@ -368,6 +420,11 @@ impl Database {
     }
 }
 
+/// by grpc standard, only `Unavailable` is retryable, see: https://github.com/grpc/grpc/blob/master/doc/statuscodes.md#status-codes-and-their-use-in-grpc
+pub fn is_grpc_retryable(err: &tonic::Status) -> bool {
+    matches!(err.code(), tonic::Code::Unavailable)
+}
+
 #[derive(Default, Debug, Clone)]
 struct FlightContext {
     auth_header: Option<AuthHeader>,

src/client/src/error.rs

@@ -110,13 +110,6 @@ pub enum Error {
         location: Location,
     },
 
-    #[snafu(display("Failed to parse ascii string: {}", value))]
-    InvalidAscii {
-        value: String,
-        #[snafu(implicit)]
-        location: Location,
-    },
-
     #[snafu(display("Invalid Tonic metadata value"))]
     InvalidTonicMetadataValue {
         #[snafu(source)]
@@ -124,6 +117,13 @@ pub enum Error {
         #[snafu(implicit)]
         location: Location,
     },
+
+    #[snafu(display("Failed to convert Schema"))]
+    ConvertSchema {
+        #[snafu(implicit)]
+        location: Location,
+        source: datatypes::error::Error,
+    },
 }
 
 pub type Result<T> = std::result::Result<T, Error>;
@@ -143,10 +143,8 @@ impl ErrorExt for Error {
             | Error::ConvertFlightData { source, .. }
             | Error::CreateTlsChannel { source, .. } => source.status_code(),
             Error::IllegalGrpcClientState { .. } => StatusCode::Unexpected,
-
-            Error::InvalidAscii { .. } | Error::InvalidTonicMetadataValue { .. } => {
-                StatusCode::InvalidArguments
-            }
+            Error::InvalidTonicMetadataValue { .. } => StatusCode::InvalidArguments,
+            Error::ConvertSchema { source, .. } => source.status_code(),
         }
     }
 

src/client/src/region.rs

@@ -28,7 +28,7 @@ use common_meta::error::{self as meta_error, Result as MetaResult};
 use common_meta::node_manager::Datanode;
 use common_query::request::QueryRequest;
 use common_recordbatch::error::ExternalSnafu;
-use common_recordbatch::{RecordBatchStreamWrapper, SendableRecordBatchStream};
+use common_recordbatch::{RecordBatch, RecordBatchStreamWrapper, SendableRecordBatchStream};
 use common_telemetry::error;
 use common_telemetry::tracing_context::TracingContext;
 use prost::Message;
@@ -55,6 +55,7 @@ impl Datanode for RegionRequester {
             if err.should_retry() {
                 meta_error::Error::RetryLater {
                     source: BoxedError::new(err),
+                    clean_poisons: false,
                 }
             } else {
                 meta_error::Error::External {
@@ -125,7 +126,7 @@ impl RegionRequester {
         let mut flight_message_stream = flight_data_stream.map(move |flight_data| {
             flight_data
                 .map_err(Error::from)
-                .and_then(|data| decoder.try_decode(data).context(ConvertFlightDataSnafu))
+                .and_then(|data| decoder.try_decode(&data).context(ConvertFlightDataSnafu))
         });
 
         let Some(first_flight_message) = flight_message_stream.next().await else {
@@ -146,6 +147,10 @@ impl RegionRequester {
 
         let tracing_context = TracingContext::from_current_span();
 
+        let schema = Arc::new(
+            datatypes::schema::Schema::try_from(schema).context(error::ConvertSchemaSnafu)?,
+        );
+        let schema_cloned = schema.clone();
         let stream = Box::pin(stream!({
             let _span = tracing_context.attach(common_telemetry::tracing::info_span!(
                 "poll_flight_data_stream"
@@ -156,7 +161,12 @@ impl RegionRequester {
                     .context(ExternalSnafu)?;
 
                 match flight_message {
-                    FlightMessage::Recordbatch(record_batch) => yield Ok(record_batch),
+                    FlightMessage::RecordBatch(record_batch) => {
+                        yield RecordBatch::try_from_df_record_batch(
+                            schema_cloned.clone(),
+                            record_batch,
+                        )
+                    }
                     FlightMessage::Metrics(s) => {
                         let m = serde_json::from_str(&s).ok().map(Arc::new);
                         metrics_ref.swap(m);

src/cmd/Cargo.toml

@@ -10,7 +10,13 @@ name = "greptime"
 path = "src/bin/greptime.rs"
 
 [features]
-default = ["servers/pprof", "servers/mem-prof"]
+default = [
+    "servers/pprof",
+    "servers/mem-prof",
+    "meta-srv/pg_kvbackend",
+    "meta-srv/mysql_kvbackend",
+]
+enterprise = ["common-meta/enterprise", "frontend/enterprise", "meta-srv/enterprise"]
 tokio-console = ["common-telemetry/tokio-console"]
 
 [lints]
@@ -74,6 +80,7 @@ servers.workspace = true
 session.workspace = true
 similar-asserts.workspace = true
 snafu.workspace = true
+stat.workspace = true
 store-api.workspace = true
 substrait.workspace = true
 table.workspace = true

src/cmd/src/bin/greptime.rs

@@ -15,9 +15,11 @@
 #![doc = include_str!("../../../../README.md")]
 
 use clap::{Parser, Subcommand};
+use cmd::datanode::builder::InstanceBuilder;
 use cmd::error::{InitTlsProviderSnafu, Result};
 use cmd::options::GlobalOptions;
 use cmd::{cli, datanode, flownode, frontend, metasrv, standalone, App};
+use common_base::Plugins;
 use common_version::version;
 use servers::install_ring_crypto_provider;
 
@@ -102,10 +104,10 @@ async fn main_body() -> Result<()> {
 async fn start(cli: Command) -> Result<()> {
     match cli.subcmd {
         SubCommand::Datanode(cmd) => {
-            cmd.build(cmd.load_options(&cli.global_options)?)
-                .await?
-                .run()
-                .await
+            let opts = cmd.load_options(&cli.global_options)?;
+            let plugins = Plugins::new();
+            let builder = InstanceBuilder::try_new_with_init(opts, plugins).await?;
+            cmd.build_with(builder).await?.run().await
         }
         SubCommand::Flownode(cmd) => {
             cmd.build(cmd.load_options(&cli.global_options)?)

src/cmd/src/cli.rs

@@ -58,7 +58,7 @@ impl App for Instance {
         false
     }
 
-    async fn stop(&self) -> Result<()> {
+    async fn stop(&mut self) -> Result<()> {
         Ok(())
     }
 }
@@ -76,6 +76,7 @@ impl Command {
             &opts,
             &TracingOptions::default(),
             None,
+            None,
         );
 
         let tool = self.cmd.build().await.context(error::BuildCliSnafu)?;

src/cmd/src/datanode.rs

@@ -12,33 +12,27 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-use std::sync::Arc;
+pub mod builder;
+
 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};
 use common_wal::config::DatanodeWalConfig;
-use datanode::datanode::{Datanode, DatanodeBuilder};
-use datanode::service::DatanodeServiceBuilder;
-use meta_client::{MetaClientOptions, MetaClientType};
-use servers::Mode;
-use snafu::{ensure, OptionExt, ResultExt};
+use datanode::datanode::Datanode;
+use meta_client::MetaClientOptions;
+use snafu::{ensure, ResultExt};
 use tracing_appender::non_blocking::WorkerGuard;
 
+use crate::datanode::builder::InstanceBuilder;
 use crate::error::{
-    LoadLayeredConfigSnafu, MetaClientInitSnafu, MissingConfigSnafu, Result, ShutdownDatanodeSnafu,
-    StartDatanodeSnafu,
+    LoadLayeredConfigSnafu, MissingConfigSnafu, Result, ShutdownDatanodeSnafu, StartDatanodeSnafu,
 };
 use crate::options::{GlobalOptions, GreptimeOptions};
-use crate::{log_versions, App};
+use crate::App;
 
 pub const APP_NAME: &str = "greptime-datanode";
 
@@ -83,7 +77,7 @@ impl App for Instance {
         self.datanode.start().await.context(StartDatanodeSnafu)
     }
 
-    async fn stop(&self) -> Result<()> {
+    async fn stop(&mut self) -> Result<()> {
         self.datanode
             .shutdown()
             .await
@@ -98,8 +92,8 @@ pub struct Command {
 }
 
 impl Command {
-    pub async fn build(&self, opts: DatanodeOptions) -> Result<Instance> {
-        self.subcmd.build(opts).await
+    pub async fn build_with(&self, builder: InstanceBuilder) -> Result<Instance> {
+        self.subcmd.build_with(builder).await
     }
 
     pub fn load_options(&self, global_options: &GlobalOptions) -> Result<DatanodeOptions> {
@@ -115,9 +109,12 @@ enum SubCommand {
 }
 
 impl SubCommand {
-    async fn build(&self, opts: DatanodeOptions) -> Result<Instance> {
+    async fn build_with(&self, builder: InstanceBuilder) -> Result<Instance> {
         match self {
-            SubCommand::Start(cmd) => cmd.build(opts).await,
+            SubCommand::Start(cmd) => {
+                info!("Building datanode with {:#?}", cmd);
+                builder.build().await
+            }
         }
     }
 }
@@ -159,6 +156,7 @@ impl StartCommand {
         .context(LoadLayeredConfigSnafu)?;
 
         self.merge_with_cli_options(global_options, &mut opts)?;
+        opts.component.sanitize();
 
         Ok(opts)
     }
@@ -263,74 +261,6 @@ impl StartCommand {
 
         Ok(())
     }
-
-    async fn build(&self, opts: DatanodeOptions) -> Result<Instance> {
-        common_runtime::init_global_runtimes(&opts.runtime);
-
-        let guard = common_telemetry::init_global_logging(
-            APP_NAME,
-            &opts.component.logging,
-            &opts.component.tracing,
-            opts.component.node_id.map(|x| x.to_string()),
-        );
-        log_versions(version(), short_version(), APP_NAME);
-
-        info!("Datanode start command: {:#?}", self);
-        info!("Datanode options: {:#?}", opts);
-
-        let plugin_opts = opts.plugins;
-        let mut opts = opts.component;
-        opts.grpc.detect_server_addr();
-        let mut plugins = Plugins::new();
-        plugins::setup_datanode_plugins(&mut plugins, &plugin_opts, &opts)
-            .await
-            .context(StartDatanodeSnafu)?;
-
-        let member_id = opts
-            .node_id
-            .context(MissingConfigSnafu { msg: "'node_id'" })?;
-
-        let meta_config = opts.meta_client.as_ref().context(MissingConfigSnafu {
-            msg: "'meta_client_options'",
-        })?;
-
-        let meta_client = meta_client::create_meta_client(
-            MetaClientType::Datanode { member_id },
-            meta_config,
-            None,
-        )
-        .await
-        .context(MetaClientInitSnafu)?;
-
-        let meta_backend = Arc::new(MetaKvBackend {
-            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, Mode::Distributed)
-            .with_meta_client(meta_client)
-            .with_kv_backend(meta_backend)
-            .with_cache_registry(layered_cache_registry)
-            .build()
-            .await
-            .context(StartDatanodeSnafu)?;
-
-        let services = DatanodeServiceBuilder::new(&opts)
-            .with_default_grpc_server(&datanode.region_server())
-            .enable_http_service()
-            .build()
-            .await
-            .context(StartDatanodeSnafu)?;
-        datanode.setup_services(services);
-
-        Ok(Instance::new(datanode, guard))
-    }
 }
 
 #[cfg(test)]
@@ -352,7 +282,6 @@ mod tests {
         common_telemetry::init_default_ut_logging();
         let mut file = create_named_temp_file();
         let toml_str = r#"
-            mode = "distributed"
             enable_memory_catalog = false
             node_id = 42
 
@@ -379,7 +308,6 @@ mod tests {
     fn test_read_from_config_file() {
         let mut file = create_named_temp_file();
         let toml_str = r#"
-            mode = "distributed"
             enable_memory_catalog = false
             node_id = 42
 
@@ -545,7 +473,6 @@ mod tests {
     fn test_config_precedence_order() {
         let mut file = create_named_temp_file();
         let toml_str = r#"
-            mode = "distributed"
             enable_memory_catalog = false
             node_id = 42
             rpc_addr = "127.0.0.1:3001"

src/cmd/src/datanode/builder.rs

@@ -0,0 +1,139 @@
+// 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 cache::build_datanode_cache_registry;
+use catalog::kvbackend::MetaKvBackend;
+use common_base::Plugins;
+use common_meta::cache::LayeredCacheRegistryBuilder;
+use common_telemetry::info;
+use common_version::{short_version, version};
+use datanode::datanode::DatanodeBuilder;
+use datanode::service::DatanodeServiceBuilder;
+use meta_client::MetaClientType;
+use snafu::{OptionExt, ResultExt};
+use tracing_appender::non_blocking::WorkerGuard;
+
+use crate::datanode::{DatanodeOptions, Instance, APP_NAME};
+use crate::error::{MetaClientInitSnafu, MissingConfigSnafu, Result, StartDatanodeSnafu};
+use crate::{create_resource_limit_metrics, log_versions};
+
+/// Builder for Datanode instance.
+pub struct InstanceBuilder {
+    guard: Vec<WorkerGuard>,
+    opts: DatanodeOptions,
+    datanode_builder: DatanodeBuilder,
+}
+
+impl InstanceBuilder {
+    /// Try to create a new [InstanceBuilder], and do some initialization work like allocating
+    /// runtime resources, setting up global logging and plugins, etc.
+    pub async fn try_new_with_init(
+        mut opts: DatanodeOptions,
+        mut plugins: Plugins,
+    ) -> Result<Self> {
+        let guard = Self::init(&mut opts, &mut plugins).await?;
+
+        let datanode_builder = Self::datanode_builder(&opts, plugins).await?;
+
+        Ok(Self {
+            guard,
+            opts,
+            datanode_builder,
+        })
+    }
+
+    async fn init(opts: &mut DatanodeOptions, plugins: &mut Plugins) -> Result<Vec<WorkerGuard>> {
+        common_runtime::init_global_runtimes(&opts.runtime);
+
+        let dn_opts = &mut opts.component;
+        let guard = common_telemetry::init_global_logging(
+            APP_NAME,
+            &dn_opts.logging,
+            &dn_opts.tracing,
+            dn_opts.node_id.map(|x| x.to_string()),
+            None,
+        );
+
+        log_versions(version(), short_version(), APP_NAME);
+        create_resource_limit_metrics(APP_NAME);
+
+        plugins::setup_datanode_plugins(plugins, &opts.plugins, dn_opts)
+            .await
+            .context(StartDatanodeSnafu)?;
+
+        dn_opts.grpc.detect_server_addr();
+
+        info!("Initialized Datanode instance with {:#?}", opts);
+        Ok(guard)
+    }
+
+    async fn datanode_builder(opts: &DatanodeOptions, plugins: Plugins) -> Result<DatanodeBuilder> {
+        let dn_opts = &opts.component;
+
+        let member_id = dn_opts
+            .node_id
+            .context(MissingConfigSnafu { msg: "'node_id'" })?;
+        let meta_client_options = dn_opts.meta_client.as_ref().context(MissingConfigSnafu {
+            msg: "meta client options",
+        })?;
+        let client = meta_client::create_meta_client(
+            MetaClientType::Datanode { member_id },
+            meta_client_options,
+            Some(&plugins),
+        )
+        .await
+        .context(MetaClientInitSnafu)?;
+
+        let backend = Arc::new(MetaKvBackend {
+            client: client.clone(),
+        });
+        let mut builder = DatanodeBuilder::new(dn_opts.clone(), plugins.clone(), backend.clone());
+
+        let registry = Arc::new(
+            LayeredCacheRegistryBuilder::default()
+                .add_cache_registry(build_datanode_cache_registry(backend))
+                .build(),
+        );
+        builder
+            .with_cache_registry(registry)
+            .with_meta_client(client.clone());
+        Ok(builder)
+    }
+
+    /// Get the mutable builder for Datanode, in case you want to change some fields before the
+    /// final construction.
+    pub fn mut_datanode_builder(&mut self) -> &mut DatanodeBuilder {
+        &mut self.datanode_builder
+    }
+
+    /// Try to build the Datanode instance.
+    pub async fn build(self) -> Result<Instance> {
+        let mut datanode = self
+            .datanode_builder
+            .build()
+            .await
+            .context(StartDatanodeSnafu)?;
+
+        let services = DatanodeServiceBuilder::new(&self.opts.component)
+            .with_default_grpc_server(&datanode.region_server())
+            .enable_http_service()
+            .build()
+            .context(StartDatanodeSnafu)?;
+        datanode.setup_services(services);
+
+        Ok(Instance::new(datanode, self.guard))
+    }
+}

src/cmd/src/error.rs

@@ -177,9 +177,6 @@ pub enum Error {
         source: meta_srv::error::Error,
     },
 
-    #[snafu(display("Invalid REPL command: {reason}"))]
-    InvalidReplCommand { reason: String },
-
     #[snafu(display("Failed to parse SQL: {}", sql))]
     ParseSql {
         sql: String,
@@ -331,7 +328,6 @@ impl ErrorExt for Error {
             Error::MissingConfig { .. }
             | Error::LoadLayeredConfig { .. }
             | Error::IllegalConfig { .. }
-            | Error::InvalidReplCommand { .. }
             | Error::InitTimezone { .. }
             | Error::ConnectEtcd { .. }
             | Error::CreateDir { .. }

src/cmd/src/flownode.rs

@@ -33,7 +33,8 @@ use common_telemetry::info;
 use common_telemetry::logging::TracingOptions;
 use common_version::{short_version, version};
 use flow::{
-    FlownodeBuilder, FlownodeInstance, FlownodeServiceBuilder, FrontendClient, FrontendInvoker,
+    get_flow_auth_options, FlownodeBuilder, FlownodeInstance, FlownodeServiceBuilder,
+    FrontendClient, FrontendInvoker,
 };
 use meta_client::{MetaClientOptions, MetaClientType};
 use snafu::{ensure, OptionExt, ResultExt};
@@ -44,7 +45,7 @@ use crate::error::{
     MissingConfigSnafu, Result, ShutdownFlownodeSnafu, StartFlownodeSnafu,
 };
 use crate::options::{GlobalOptions, GreptimeOptions};
-use crate::{log_versions, App};
+use crate::{create_resource_limit_metrics, log_versions, App};
 
 pub const APP_NAME: &str = "greptime-flownode";
 
@@ -82,10 +83,14 @@ impl App for Instance {
     }
 
     async fn start(&mut self) -> Result<()> {
+        plugins::start_flownode_plugins(self.flownode.flow_engine().plugins().clone())
+            .await
+            .context(StartFlownodeSnafu)?;
+
         self.flownode.start().await.context(StartFlownodeSnafu)
     }
 
-    async fn stop(&self) -> Result<()> {
+    async fn stop(&mut self) -> Result<()> {
         self.flownode
             .shutdown()
             .await
@@ -151,6 +156,9 @@ struct StartCommand {
     /// HTTP request timeout in seconds.
     #[clap(long)]
     http_timeout: Option<u64>,
+    /// User Provider cfg, for auth, currently only support static user provider
+    #[clap(long)]
+    user_provider: Option<String>,
 }
 
 impl StartCommand {
@@ -214,6 +222,10 @@ impl StartCommand {
             opts.http.timeout = Duration::from_secs(http_timeout);
         }
 
+        if let Some(user_provider) = &self.user_provider {
+            opts.user_provider = Some(user_provider.clone());
+        }
+
         ensure!(
             opts.node_id.is_some(),
             MissingConfigSnafu {
@@ -232,15 +244,24 @@ impl StartCommand {
             &opts.component.logging,
             &opts.component.tracing,
             opts.component.node_id.map(|x| x.to_string()),
+            None,
         );
+
         log_versions(version(), short_version(), APP_NAME);
+        create_resource_limit_metrics(APP_NAME);
 
         info!("Flownode start command: {:#?}", self);
         info!("Flownode options: {:#?}", opts);
 
+        let plugin_opts = opts.plugins;
         let mut opts = opts.component;
         opts.grpc.detect_server_addr();
 
+        let mut plugins = Plugins::new();
+        plugins::setup_flownode_plugins(&mut plugins, &plugin_opts, &opts)
+            .await
+            .context(StartFlownodeSnafu)?;
+
         let member_id = opts
             .node_id
             .context(MissingConfigSnafu { msg: "'node_id'" })?;
@@ -315,10 +336,12 @@ impl StartCommand {
         );
 
         let flow_metadata_manager = Arc::new(FlowMetadataManager::new(cached_meta_backend.clone()));
-        let frontend_client = FrontendClient::from_meta_client(meta_client.clone());
+        let flow_auth_header = get_flow_auth_options(&opts).context(StartFlownodeSnafu)?;
+        let frontend_client =
+            FrontendClient::from_meta_client(meta_client.clone(), flow_auth_header);
         let flownode_builder = FlownodeBuilder::new(
             opts.clone(),
-            Plugins::new(),
+            plugins,
             table_metadata_manager,
             catalog_manager.clone(),
             flow_metadata_manager,
@@ -331,7 +354,6 @@ impl StartCommand {
             .with_grpc_server(flownode.flownode_server().clone())
             .enable_http_service()
             .build()
-            .await
             .context(StartFlownodeSnafu)?;
         flownode.setup_services(services);
         let flownode = flownode;

src/cmd/src/frontend.rs

@@ -37,7 +37,6 @@ use frontend::heartbeat::HeartbeatTask;
 use frontend::instance::builder::FrontendBuilder;
 use frontend::server::Services;
 use meta_client::{MetaClientOptions, MetaClientType};
-use query::stats::StatementStatistics;
 use servers::export_metrics::ExportMetricsTask;
 use servers::tls::{TlsMode, TlsOption};
 use snafu::{OptionExt, ResultExt};
@@ -45,7 +44,7 @@ use tracing_appender::non_blocking::WorkerGuard;
 
 use crate::error::{self, Result};
 use crate::options::{GlobalOptions, GreptimeOptions};
-use crate::{log_versions, App};
+use crate::{create_resource_limit_metrics, log_versions, App};
 
 type FrontendOptions = GreptimeOptions<frontend::frontend::FrontendOptions>;
 
@@ -89,7 +88,7 @@ impl App for Instance {
             .context(error::StartFrontendSnafu)
     }
 
-    async fn stop(&self) -> Result<()> {
+    async fn stop(&mut self) -> Result<()> {
         self.frontend
             .shutdown()
             .await
@@ -269,8 +268,11 @@ impl StartCommand {
             &opts.component.logging,
             &opts.component.tracing,
             opts.component.node_id.clone(),
+            opts.component.slow_query.as_ref(),
         );
+
         log_versions(version(), short_version(), APP_NAME);
+        create_resource_limit_metrics(APP_NAME);
 
         info!("Frontend start command: {:#?}", self);
         info!("Frontend options: {:#?}", opts);
@@ -368,7 +370,6 @@ impl StartCommand {
             catalog_manager,
             Arc::new(client),
             meta_client,
-            StatementStatistics::new(opts.logging.slow_query.clone()),
         )
         .with_plugin(plugins.clone())
         .with_local_cache_invalidator(layered_cache_registry)
@@ -382,7 +383,6 @@ impl StartCommand {
 
         let servers = Services::new(opts, instance.clone(), plugins)
             .build()
-            .await
             .context(error::StartFrontendSnafu)?;
 
         let frontend = Frontend {
@@ -448,8 +448,6 @@ mod tests {
     fn test_read_from_config_file() {
         let mut file = create_named_temp_file();
         let toml_str = r#"
-            mode = "distributed"
-
             [http]
             addr = "127.0.0.1:4000"
             timeout = "0s"
@@ -538,8 +536,6 @@ mod tests {
     fn test_config_precedence_order() {
         let mut file = create_named_temp_file();
         let toml_str = r#"
-            mode = "distributed"
-
             [http]
             addr = "127.0.0.1:4000"
 

src/cmd/src/lib.rs

@@ -16,6 +16,7 @@
 
 use async_trait::async_trait;
 use common_telemetry::{error, info};
+use stat::{get_cpu_limit, get_memory_limit};
 
 use crate::error::Result;
 
@@ -31,6 +32,12 @@ pub mod standalone;
 lazy_static::lazy_static! {
     static ref APP_VERSION: prometheus::IntGaugeVec =
         prometheus::register_int_gauge_vec!("greptime_app_version", "app version", &["version", "short_version", "app"]).unwrap();
+
+    static ref CPU_LIMIT: prometheus::IntGaugeVec =
+        prometheus::register_int_gauge_vec!("greptime_cpu_limit_in_millicores", "cpu limit in millicores", &["app"]).unwrap();
+
+    static ref MEMORY_LIMIT: prometheus::IntGaugeVec =
+        prometheus::register_int_gauge_vec!("greptime_memory_limit_in_bytes", "memory limit in bytes", &["app"]).unwrap();
 }
 
 /// wait for the close signal, for unix platform it's SIGINT or SIGTERM
@@ -74,7 +81,7 @@ pub trait App: Send {
         true
     }
 
-    async fn stop(&self) -> Result<()>;
+    async fn stop(&mut self) -> Result<()>;
 
     async fn run(&mut self) -> Result<()> {
         info!("Starting app: {}", self.name());
@@ -114,6 +121,24 @@ pub fn log_versions(version: &str, short_version: &str, app: &str) {
     log_env_flags();
 }
 
+pub fn create_resource_limit_metrics(app: &str) {
+    if let Some(cpu_limit) = get_cpu_limit() {
+        info!(
+            "GreptimeDB start with cpu limit in millicores: {}",
+            cpu_limit
+        );
+        CPU_LIMIT.with_label_values(&[app]).set(cpu_limit);
+    }
+
+    if let Some(memory_limit) = get_memory_limit() {
+        info!(
+            "GreptimeDB start with memory limit in bytes: {}",
+            memory_limit
+        );
+        MEMORY_LIMIT.with_label_values(&[app]).set(memory_limit);
+    }
+}
+
 fn log_env_flags() {
     info!("command line arguments");
     for argument in std::env::args() {

src/cmd/src/metasrv.rs

@@ -12,6 +12,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+use std::fmt;
 use std::time::Duration;
 
 use async_trait::async_trait;
@@ -28,7 +29,7 @@ use tracing_appender::non_blocking::WorkerGuard;
 
 use crate::error::{self, LoadLayeredConfigSnafu, Result, StartMetaServerSnafu};
 use crate::options::{GlobalOptions, GreptimeOptions};
-use crate::{log_versions, App};
+use crate::{create_resource_limit_metrics, log_versions, App};
 
 type MetasrvOptions = GreptimeOptions<meta_srv::metasrv::MetasrvOptions>;
 
@@ -68,7 +69,7 @@ impl App for Instance {
         self.instance.start().await.context(StartMetaServerSnafu)
     }
 
-    async fn stop(&self) -> Result<()> {
+    async fn stop(&mut self) -> Result<()> {
         self.instance
             .shutdown()
             .await
@@ -131,7 +132,7 @@ impl SubCommand {
     }
 }
 
-#[derive(Debug, Default, Parser)]
+#[derive(Default, Parser)]
 pub struct StartCommand {
     /// The address to bind the gRPC server.
     #[clap(long, alias = "bind-addr")]
@@ -171,6 +172,27 @@ pub struct StartCommand {
     backend: Option<BackendImpl>,
 }
 
+impl fmt::Debug for StartCommand {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        f.debug_struct("StartCommand")
+            .field("rpc_bind_addr", &self.rpc_bind_addr)
+            .field("rpc_server_addr", &self.rpc_server_addr)
+            .field("store_addrs", &self.sanitize_store_addrs())
+            .field("config_file", &self.config_file)
+            .field("selector", &self.selector)
+            .field("use_memory_store", &self.use_memory_store)
+            .field("enable_region_failover", &self.enable_region_failover)
+            .field("http_addr", &self.http_addr)
+            .field("http_timeout", &self.http_timeout)
+            .field("env_prefix", &self.env_prefix)
+            .field("data_home", &self.data_home)
+            .field("store_key_prefix", &self.store_key_prefix)
+            .field("max_txn_ops", &self.max_txn_ops)
+            .field("backend", &self.backend)
+            .finish()
+    }
+}
+
 impl StartCommand {
     pub fn load_options(&self, global_options: &GlobalOptions) -> Result<MetasrvOptions> {
         let mut opts = MetasrvOptions::load_layered_options(
@@ -184,6 +206,15 @@ impl StartCommand {
         Ok(opts)
     }
 
+    fn sanitize_store_addrs(&self) -> Option<Vec<String>> {
+        self.store_addrs.as_ref().map(|addrs| {
+            addrs
+                .iter()
+                .map(|addr| common_meta::kv_backend::util::sanitize_connection_string(addr))
+                .collect()
+        })
+    }
+
     // The precedence order is: cli > config file > environment variables > default values.
     fn merge_with_cli_options(
         &self,
@@ -269,8 +300,11 @@ impl StartCommand {
             &opts.component.logging,
             &opts.component.tracing,
             None,
+            None,
         );
+
         log_versions(version(), short_version(), APP_NAME);
+        create_resource_limit_metrics(APP_NAME);
 
         info!("Metasrv start command: {:#?}", self);
 

src/cmd/src/standalone.rs

@@ -35,6 +35,8 @@ 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;
+#[cfg(feature = "enterprise")]
+use common_meta::ddl_manager::TriggerDdlManagerRef;
 use common_meta::key::flow::flow_state::FlowStat;
 use common_meta::key::flow::{FlowMetadataManager, FlowMetadataManagerRef};
 use common_meta::key::{TableMetadataManager, TableMetadataManagerRef};
@@ -47,7 +49,7 @@ use common_meta::sequence::SequenceBuilder;
 use common_meta::wal_options_allocator::{build_wal_options_allocator, WalOptionsAllocatorRef};
 use common_procedure::{ProcedureInfo, ProcedureManagerRef};
 use common_telemetry::info;
-use common_telemetry::logging::{LoggingOptions, TracingOptions};
+use common_telemetry::logging::{LoggingOptions, SlowQueryOptions, TracingOptions};
 use common_time::timezone::set_default_timezone;
 use common_version::{short_version, version};
 use common_wal::config::DatanodeWalConfig;
@@ -69,20 +71,19 @@ use frontend::service_config::{
 };
 use meta_srv::metasrv::{FLOW_ID_SEQ, TABLE_ID_SEQ};
 use mito2::config::MitoConfig;
-use query::stats::StatementStatistics;
+use query::options::QueryOptions;
 use serde::{Deserialize, Serialize};
 use servers::export_metrics::{ExportMetricsOption, ExportMetricsTask};
 use servers::grpc::GrpcOptions;
 use servers::http::HttpOptions;
 use servers::tls::{TlsMode, TlsOption};
-use servers::Mode;
 use snafu::ResultExt;
 use tokio::sync::RwLock;
 use tracing_appender::non_blocking::WorkerGuard;
 
 use crate::error::{Result, StartFlownodeSnafu};
 use crate::options::{GlobalOptions, GreptimeOptions};
-use crate::{error, log_versions, App};
+use crate::{create_resource_limit_metrics, error, log_versions, App};
 
 pub const APP_NAME: &str = "greptime-standalone";
 
@@ -154,6 +155,8 @@ pub struct StandaloneOptions {
     pub init_regions_in_background: bool,
     pub init_regions_parallelism: usize,
     pub max_in_flight_write_bytes: Option<ReadableSize>,
+    pub slow_query: Option<SlowQueryOptions>,
+    pub query: QueryOptions,
 }
 
 impl Default for StandaloneOptions {
@@ -185,6 +188,8 @@ impl Default for StandaloneOptions {
             init_regions_in_background: false,
             init_regions_parallelism: 16,
             max_in_flight_write_bytes: None,
+            slow_query: Some(SlowQueryOptions::default()),
+            query: QueryOptions::default(),
         }
     }
 }
@@ -224,6 +229,7 @@ impl StandaloneOptions {
             // 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,
+            slow_query: cloned_opts.slow_query,
             ..Default::default()
         }
     }
@@ -239,6 +245,7 @@ impl StandaloneOptions {
             grpc: cloned_opts.grpc,
             init_regions_in_background: cloned_opts.init_regions_in_background,
             init_regions_parallelism: cloned_opts.init_regions_parallelism,
+            query: cloned_opts.query,
             ..Default::default()
         }
     }
@@ -256,8 +263,8 @@ pub struct Instance {
 
 impl Instance {
     /// Find the socket addr of a server by its `name`.
-    pub async fn server_addr(&self, name: &str) -> Option<SocketAddr> {
-        self.frontend.server_handlers().addr(name).await
+    pub fn server_addr(&self, name: &str) -> Option<SocketAddr> {
+        self.frontend.server_handlers().addr(name)
     }
 }
 
@@ -294,7 +301,7 @@ impl App for Instance {
         Ok(())
     }
 
-    async fn stop(&self) -> Result<()> {
+    async fn stop(&mut self) -> Result<()> {
         self.frontend
             .shutdown()
             .await
@@ -448,8 +455,11 @@ impl StartCommand {
             &opts.component.logging,
             &opts.component.tracing,
             None,
+            opts.component.slow_query.as_ref(),
         );
+
         log_versions(version(), short_version(), APP_NAME);
+        create_resource_limit_metrics(APP_NAME);
 
         info!("Standalone start command: {:#?}", self);
         info!("Standalone options: {opts:#?}");
@@ -497,12 +507,9 @@ impl StartCommand {
             .build(),
         );
 
-        let datanode = DatanodeBuilder::new(dn_opts, plugins.clone(), Mode::Standalone)
-            .with_kv_backend(kv_backend.clone())
-            .with_cache_registry(layered_cache_registry.clone())
-            .build()
-            .await
-            .context(error::StartDatanodeSnafu)?;
+        let mut builder = DatanodeBuilder::new(dn_opts, plugins.clone(), kv_backend.clone());
+        builder.with_cache_registry(layered_cache_registry.clone());
+        let datanode = builder.build().await.context(error::StartDatanodeSnafu)?;
 
         let information_extension = Arc::new(StandaloneInformationExtension::new(
             datanode.region_server(),
@@ -580,6 +587,8 @@ impl StartCommand {
             flow_id_sequence,
         ));
 
+        #[cfg(feature = "enterprise")]
+        let trigger_ddl_manager: Option<TriggerDdlManagerRef> = plugins.get();
         let ddl_task_executor = Self::create_ddl_task_executor(
             procedure_manager.clone(),
             node_manager.clone(),
@@ -588,6 +597,8 @@ impl StartCommand {
             table_meta_allocator,
             flow_metadata_manager,
             flow_meta_allocator,
+            #[cfg(feature = "enterprise")]
+            trigger_ddl_manager,
         )
         .await?;
 
@@ -598,7 +609,6 @@ impl StartCommand {
             catalog_manager.clone(),
             node_manager.clone(),
             ddl_task_executor.clone(),
-            StatementStatistics::new(opts.logging.slow_query.clone()),
         )
         .with_plugin(plugins.clone())
         .try_build()
@@ -634,7 +644,6 @@ impl StartCommand {
 
         let servers = Services::new(opts, fe_instance.clone(), plugins)
             .build()
-            .await
             .context(error::StartFrontendSnafu)?;
 
         let frontend = Frontend {
@@ -654,6 +663,7 @@ impl StartCommand {
         })
     }
 
+    #[allow(clippy::too_many_arguments)]
     pub async fn create_ddl_task_executor(
         procedure_manager: ProcedureManagerRef,
         node_manager: NodeManagerRef,
@@ -662,6 +672,7 @@ impl StartCommand {
         table_metadata_allocator: TableMetadataAllocatorRef,
         flow_metadata_manager: FlowMetadataManagerRef,
         flow_metadata_allocator: FlowMetadataAllocatorRef,
+        #[cfg(feature = "enterprise")] trigger_ddl_manager: Option<TriggerDdlManagerRef>,
     ) -> Result<ProcedureExecutorRef> {
         let procedure_executor: ProcedureExecutorRef = Arc::new(
             DdlManager::try_new(
@@ -678,6 +689,8 @@ impl StartCommand {
                 },
                 procedure_manager,
                 true,
+                #[cfg(feature = "enterprise")]
+                trigger_ddl_manager,
             )
             .context(error::InitDdlManagerSnafu)?,
         );
@@ -858,8 +871,6 @@ mod tests {
     fn test_read_from_config_file() {
         let mut file = create_named_temp_file();
         let toml_str = r#"
-            mode = "distributed"
-
             enable_memory_catalog = true
 
             [wal]
@@ -990,8 +1001,6 @@ mod tests {
     fn test_config_precedence_order() {
         let mut file = create_named_temp_file();
         let toml_str = r#"
-            mode = "standalone"
-
             [http]
             addr = "127.0.0.1:4000"
 

src/cmd/tests/load_config_test.rs

@@ -18,7 +18,7 @@ use cmd::options::GreptimeOptions;
 use cmd::standalone::StandaloneOptions;
 use common_config::Configurable;
 use common_options::datanode::{ClientOptions, DatanodeClientOptions};
-use common_telemetry::logging::{LoggingOptions, SlowQueryOptions, DEFAULT_OTLP_ENDPOINT};
+use common_telemetry::logging::{LoggingOptions, DEFAULT_OTLP_ENDPOINT};
 use common_wal::config::raft_engine::RaftEngineConfig;
 use common_wal::config::DatanodeWalConfig;
 use datanode::config::{DatanodeOptions, RegionEngineConfig, StorageConfig};
@@ -167,11 +167,6 @@ fn test_load_metasrv_example_config() {
                 level: Some("info".to_string()),
                 otlp_endpoint: Some(DEFAULT_OTLP_ENDPOINT.to_string()),
                 tracing_sample_ratio: Some(Default::default()),
-                slow_query: SlowQueryOptions {
-                    enable: false,
-                    threshold: None,
-                    sample_ratio: None,
-                },
                 ..Default::default()
             },
             datanode: DatanodeClientOptions {

src/common/config/src/config.rs

@@ -111,11 +111,9 @@ mod tests {
     use serde::{Deserialize, Serialize};
 
     use super::*;
-    use crate::Mode;
 
-    #[derive(Debug, Serialize, Deserialize)]
+    #[derive(Debug, Serialize, Deserialize, Default)]
     struct TestDatanodeConfig {
-        mode: Mode,
         node_id: Option<u64>,
         logging: LoggingOptions,
         meta_client: Option<MetaClientOptions>,
@@ -123,19 +121,6 @@ mod tests {
         storage: StorageConfig,
     }
 
-    impl Default for TestDatanodeConfig {
-        fn default() -> Self {
-            Self {
-                mode: Mode::Distributed,
-                node_id: None,
-                logging: LoggingOptions::default(),
-                meta_client: None,
-                wal: DatanodeWalConfig::default(),
-                storage: StorageConfig::default(),
-            }
-        }
-    }
-
     impl Configurable for TestDatanodeConfig {
         fn env_list_keys() -> Option<&'static [&'static str]> {
             Some(&["meta_client.metasrv_addrs"])
@@ -146,7 +131,6 @@ mod tests {
     fn test_load_layered_options() {
         let mut file = create_named_temp_file();
         let toml_str = r#"
-            mode = "distributed"
             enable_memory_catalog = false
             rpc_addr = "127.0.0.1:3001"
             rpc_hostname = "127.0.0.1"

src/common/config/src/lib.rs

@@ -26,16 +26,6 @@ pub fn metadata_store_dir(store_dir: &str) -> String {
     format!("{store_dir}/metadata")
 }
 
-/// The Server running mode
-#[derive(Clone, Debug, Serialize, Deserialize, Eq, PartialEq, Copy)]
-#[serde(rename_all = "lowercase")]
-pub enum Mode {
-    // The single process mode.
-    Standalone,
-    // The distributed cluster mode.
-    Distributed,
-}
-
 #[derive(Debug, Clone, Copy, Serialize, Deserialize, PartialEq, Eq)]
 #[serde(default)]
 pub struct KvBackendConfig {

src/common/datasource/src/object_store.rs

@@ -13,7 +13,9 @@
 // limitations under the License.
 
 pub mod fs;
+pub mod oss;
 pub mod s3;
+
 use std::collections::HashMap;
 
 use lazy_static::lazy_static;
@@ -25,10 +27,12 @@ use url::{ParseError, Url};
 use self::fs::build_fs_backend;
 use self::s3::build_s3_backend;
 use crate::error::{self, Result};
+use crate::object_store::oss::build_oss_backend;
 use crate::util::find_dir_and_filename;
 
 pub const FS_SCHEMA: &str = "FS";
 pub const S3_SCHEMA: &str = "S3";
+pub const OSS_SCHEMA: &str = "OSS";
 
 /// Returns `(schema, Option<host>, path)`
 pub fn parse_url(url: &str) -> Result<(String, Option<String>, String)> {
@@ -64,6 +68,12 @@ pub fn build_backend(url: &str, connection: &HashMap<String, String>) -> Result<
             })?;
             Ok(build_s3_backend(&host, &root, connection)?)
         }
+        OSS_SCHEMA => {
+            let host = host.context(error::EmptyHostPathSnafu {
+                url: url.to_string(),
+            })?;
+            Ok(build_oss_backend(&host, &root, connection)?)
+        }
         FS_SCHEMA => Ok(build_fs_backend(&root)?),
 
         _ => error::UnsupportedBackendProtocolSnafu {

src/common/datasource/src/object_store/oss.rs

@@ -0,0 +1,118 @@
+// 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::collections::HashMap;
+
+use object_store::services::Oss;
+use object_store::ObjectStore;
+use snafu::ResultExt;
+
+use crate::error::{self, Result};
+
+const BUCKET: &str = "bucket";
+const ENDPOINT: &str = "endpoint";
+const ACCESS_KEY_ID: &str = "access_key_id";
+const ACCESS_KEY_SECRET: &str = "access_key_secret";
+const ROOT: &str = "root";
+const ALLOW_ANONYMOUS: &str = "allow_anonymous";
+
+/// Check if the key is supported in OSS configuration.
+pub fn is_supported_in_oss(key: &str) -> bool {
+    [
+        ROOT,
+        ALLOW_ANONYMOUS,
+        BUCKET,
+        ENDPOINT,
+        ACCESS_KEY_ID,
+        ACCESS_KEY_SECRET,
+    ]
+    .contains(&key)
+}
+
+/// Build an OSS backend using the provided bucket, root, and connection parameters.
+pub fn build_oss_backend(
+    bucket: &str,
+    root: &str,
+    connection: &HashMap<String, String>,
+) -> Result<ObjectStore> {
+    let mut builder = Oss::default().bucket(bucket).root(root);
+
+    if let Some(endpoint) = connection.get(ENDPOINT) {
+        builder = builder.endpoint(endpoint);
+    }
+
+    if let Some(access_key_id) = connection.get(ACCESS_KEY_ID) {
+        builder = builder.access_key_id(access_key_id);
+    }
+
+    if let Some(access_key_secret) = connection.get(ACCESS_KEY_SECRET) {
+        builder = builder.access_key_secret(access_key_secret);
+    }
+
+    if let Some(allow_anonymous) = connection.get(ALLOW_ANONYMOUS) {
+        let allow = allow_anonymous.as_str().parse::<bool>().map_err(|e| {
+            error::InvalidConnectionSnafu {
+                msg: format!(
+                    "failed to parse the option {}={}, {}",
+                    ALLOW_ANONYMOUS, allow_anonymous, e
+                ),
+            }
+            .build()
+        })?;
+        if allow {
+            builder = builder.allow_anonymous();
+        }
+    }
+
+    let op = ObjectStore::new(builder)
+        .context(error::BuildBackendSnafu)?
+        .layer(object_store::layers::LoggingLayer::default())
+        .layer(object_store::layers::TracingLayer)
+        .layer(object_store::layers::build_prometheus_metrics_layer(true))
+        .finish();
+
+    Ok(op)
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_is_supported_in_oss() {
+        assert!(is_supported_in_oss(ROOT));
+        assert!(is_supported_in_oss(ALLOW_ANONYMOUS));
+        assert!(is_supported_in_oss(BUCKET));
+        assert!(is_supported_in_oss(ENDPOINT));
+        assert!(is_supported_in_oss(ACCESS_KEY_ID));
+        assert!(is_supported_in_oss(ACCESS_KEY_SECRET));
+        assert!(!is_supported_in_oss("foo"));
+        assert!(!is_supported_in_oss("BAR"));
+    }
+
+    #[test]
+    fn test_build_oss_backend_all_fields_valid() {
+        let mut connection = HashMap::new();
+        connection.insert(
+            ENDPOINT.to_string(),
+            "http://oss-ap-southeast-1.aliyuncs.com".to_string(),
+        );
+        connection.insert(ACCESS_KEY_ID.to_string(), "key_id".to_string());
+        connection.insert(ACCESS_KEY_SECRET.to_string(), "key_secret".to_string());
+        connection.insert(ALLOW_ANONYMOUS.to_string(), "true".to_string());
+
+        let result = build_oss_backend("my-bucket", "my-root", &connection);
+        assert!(result.is_ok());
+    }
+}

src/common/function/Cargo.toml

@@ -13,7 +13,7 @@ default = ["geo"]
 geo = ["geohash", "h3o", "s2", "wkt", "geo-types", "dep:geo"]
 
 [dependencies]
-ahash = "0.8"
+ahash.workspace = true
 api.workspace = true
 arc-swap = "1.0"
 async-trait.workspace = true

src/common/function/src/aggr.rs

@@ -19,4 +19,4 @@ mod uddsketch_state;
 pub use geo_path::{GeoPathAccumulator, GEO_PATH_NAME};
 pub(crate) use hll::HllStateType;
 pub use hll::{HllState, HLL_MERGE_NAME, HLL_NAME};
-pub use uddsketch_state::{UddSketchState, UDDSKETCH_STATE_NAME};
+pub use uddsketch_state::{UddSketchState, UDDSKETCH_MERGE_NAME, UDDSKETCH_STATE_NAME};

src/common/function/src/aggr/uddsketch_state.rs

@@ -31,23 +31,28 @@ use datafusion::physical_plan::expressions::Literal;
 use datafusion::prelude::create_udaf;
 use datatypes::arrow::array::ArrayRef;
 use datatypes::arrow::datatypes::{DataType, Float64Type};
+use serde::{Deserialize, Serialize};
 use uddsketch::{SketchHashKey, UDDSketch};
 
 pub const UDDSKETCH_STATE_NAME: &str = "uddsketch_state";
 
-#[derive(Debug)]
+pub const UDDSKETCH_MERGE_NAME: &str = "uddsketch_merge";
+
+#[derive(Debug, Serialize, Deserialize)]
 pub struct UddSketchState {
     uddsketch: UDDSketch,
+    error_rate: f64,
 }
 
 impl UddSketchState {
     pub fn new(bucket_size: u64, error_rate: f64) -> Self {
         Self {
             uddsketch: UDDSketch::new(bucket_size, error_rate),
+            error_rate,
         }
     }
 
-    pub fn udf_impl() -> AggregateUDF {
+    pub fn state_udf_impl() -> AggregateUDF {
         create_udaf(
             UDDSKETCH_STATE_NAME,
             vec![DataType::Int64, DataType::Float64, DataType::Float64],
@@ -61,18 +66,55 @@ impl UddSketchState {
         )
     }
 
+    /// Create a UDF for the `uddsketch_merge` function.
+    ///
+    /// `uddsketch_merge` accepts bucket size, error rate, and a binary column of states generated by `uddsketch_state`
+    /// and merges them into a single state.
+    ///
+    /// The bucket size and error rate must be the same as the original state.
+    pub fn merge_udf_impl() -> AggregateUDF {
+        create_udaf(
+            UDDSKETCH_MERGE_NAME,
+            vec![DataType::Int64, DataType::Float64, DataType::Binary],
+            Arc::new(DataType::Binary),
+            Volatility::Immutable,
+            Arc::new(|args| {
+                let (bucket_size, error_rate) = downcast_accumulator_args(args)?;
+                Ok(Box::new(UddSketchState::new(bucket_size, error_rate)))
+            }),
+            Arc::new(vec![DataType::Binary]),
+        )
+    }
+
     fn update(&mut self, value: f64) {
         self.uddsketch.add_value(value);
     }
 
-    fn merge(&mut self, raw: &[u8]) {
-        if let Ok(uddsketch) = bincode::deserialize::<UDDSketch>(raw) {
-            if uddsketch.count() != 0 {
-                self.uddsketch.merge_sketch(&uddsketch);
+    fn merge(&mut self, raw: &[u8]) -> DfResult<()> {
+        if let Ok(uddsketch) = bincode::deserialize::<Self>(raw) {
+            if uddsketch.uddsketch.count() != 0 {
+                if self.uddsketch.max_allowed_buckets() != uddsketch.uddsketch.max_allowed_buckets()
+                    || (self.error_rate - uddsketch.error_rate).abs() >= 1e-9
+                {
+                    return Err(DataFusionError::Plan(format!(
+                        "Merging UDDSketch with different parameters: arguments={:?} vs actual input={:?}",
+                        (
+                            self.uddsketch.max_allowed_buckets(),
+                            self.error_rate
+                        ),
+                        (uddsketch.uddsketch.max_allowed_buckets(), uddsketch.error_rate)
+                    )));
+                }
+                self.uddsketch.merge_sketch(&uddsketch.uddsketch);
             }
         } else {
             trace!("Warning: Failed to deserialize UDDSketch from {:?}", raw);
+            return Err(DataFusionError::Plan(
+                "Failed to deserialize UDDSketch from binary".to_string(),
+            ));
         }
+
+        Ok(())
     }
 }
 
@@ -113,9 +155,21 @@ fn downcast_accumulator_args(args: AccumulatorArgs) -> DfResult<(u64, f64)> {
 impl DfAccumulator for UddSketchState {
     fn update_batch(&mut self, values: &[ArrayRef]) -> DfResult<()> {
         let array = &values[2]; // the third column is data value
-        let f64_array = as_primitive_array::<Float64Type>(array)?;
-        for v in f64_array.iter().flatten() {
-            self.update(v);
+        match array.data_type() {
+            DataType::Float64 => {
+                let f64_array = as_primitive_array::<Float64Type>(array)?;
+                for v in f64_array.iter().flatten() {
+                    self.update(v);
+                }
+            }
+            // meaning instantiate as `uddsketch_merge`
+            DataType::Binary => self.merge_batch(std::slice::from_ref(array))?,
+            _ => {
+                return not_impl_err!(
+                    "UDDSketch functions do not support data type: {}",
+                    array.data_type()
+                )
+            }
         }
 
         Ok(())
@@ -123,7 +177,7 @@ impl DfAccumulator for UddSketchState {
 
     fn evaluate(&mut self) -> DfResult<ScalarValue> {
         Ok(ScalarValue::Binary(Some(
-            bincode::serialize(&self.uddsketch).map_err(|e| {
+            bincode::serialize(&self).map_err(|e| {
                 DataFusionError::Internal(format!("Failed to serialize UDDSketch: {}", e))
             })?,
         )))
@@ -150,7 +204,7 @@ impl DfAccumulator for UddSketchState {
 
     fn state(&mut self) -> DfResult<Vec<ScalarValue>> {
         Ok(vec![ScalarValue::Binary(Some(
-            bincode::serialize(&self.uddsketch).map_err(|e| {
+            bincode::serialize(&self).map_err(|e| {
                 DataFusionError::Internal(format!("Failed to serialize UDDSketch: {}", e))
             })?,
         ))])
@@ -160,7 +214,7 @@ impl DfAccumulator for UddSketchState {
         let array = &states[0];
         let binary_array = as_binary_array(array)?;
         for v in binary_array.iter().flatten() {
-            self.merge(v);
+            self.merge(v)?;
         }
 
         Ok(())
@@ -182,8 +236,8 @@ mod tests {
 
         let result = state.evaluate().unwrap();
         if let ScalarValue::Binary(Some(bytes)) = result {
-            let deserialized: UDDSketch = bincode::deserialize(&bytes).unwrap();
-            assert_eq!(deserialized.count(), 3);
+            let deserialized: UddSketchState = bincode::deserialize(&bytes).unwrap();
+            assert_eq!(deserialized.uddsketch.count(), 3);
         } else {
             panic!("Expected binary scalar value");
         }
@@ -201,13 +255,15 @@ mod tests {
         // Create new state and merge the serialized data
         let mut new_state = UddSketchState::new(10, 0.01);
         if let ScalarValue::Binary(Some(bytes)) = &serialized {
-            new_state.merge(bytes);
+            new_state.merge(bytes).unwrap();
 
             // Verify the merged state matches original by comparing deserialized values
-            let original_sketch: UDDSketch = bincode::deserialize(bytes).unwrap();
+            let original_sketch: UddSketchState = bincode::deserialize(bytes).unwrap();
+            let original_sketch = original_sketch.uddsketch;
             let new_result = new_state.evaluate().unwrap();
             if let ScalarValue::Binary(Some(new_bytes)) = new_result {
-                let new_sketch: UDDSketch = bincode::deserialize(&new_bytes).unwrap();
+                let new_sketch: UddSketchState = bincode::deserialize(&new_bytes).unwrap();
+                let new_sketch = new_sketch.uddsketch;
                 assert_eq!(original_sketch.count(), new_sketch.count());
                 assert_eq!(original_sketch.sum(), new_sketch.sum());
                 assert_eq!(original_sketch.mean(), new_sketch.mean());
@@ -244,7 +300,8 @@ mod tests {
 
         let result = state.evaluate().unwrap();
         if let ScalarValue::Binary(Some(bytes)) = result {
-            let deserialized: UDDSketch = bincode::deserialize(&bytes).unwrap();
+            let deserialized: UddSketchState = bincode::deserialize(&bytes).unwrap();
+            let deserialized = deserialized.uddsketch;
             assert_eq!(deserialized.count(), 3);
         } else {
             panic!("Expected binary scalar value");
@@ -273,7 +330,8 @@ mod tests {
 
             let result = merged_state.evaluate().unwrap();
             if let ScalarValue::Binary(Some(bytes)) = result {
-                let deserialized: UDDSketch = bincode::deserialize(&bytes).unwrap();
+                let deserialized: UddSketchState = bincode::deserialize(&bytes).unwrap();
+                let deserialized = deserialized.uddsketch;
                 assert_eq!(deserialized.count(), 2);
             } else {
                 panic!("Expected binary scalar value");

src/common/function/src/scalars/ip/cidr.rs

@@ -468,8 +468,8 @@ mod tests {
         let empty_values = vec![""];
         let empty_input = Arc::new(StringVector::from_slice(&empty_values)) as VectorRef;
 
-        let ipv4_result = ipv4_func.eval(&ctx, &[empty_input.clone()]);
-        let ipv6_result = ipv6_func.eval(&ctx, &[empty_input.clone()]);
+        let ipv4_result = ipv4_func.eval(&ctx, std::slice::from_ref(&empty_input));
+        let ipv6_result = ipv6_func.eval(&ctx, std::slice::from_ref(&empty_input));
 
         assert!(ipv4_result.is_err());
         assert!(ipv6_result.is_err());
@@ -478,7 +478,7 @@ mod tests {
         let invalid_values = vec!["not an ip", "192.168.1.256", "zzzz::ffff"];
         let invalid_input = Arc::new(StringVector::from_slice(&invalid_values)) as VectorRef;
 
-        let ipv4_result = ipv4_func.eval(&ctx, &[invalid_input.clone()]);
+        let ipv4_result = ipv4_func.eval(&ctx, std::slice::from_ref(&invalid_input));
 
         assert!(ipv4_result.is_err());
     }

src/common/function/src/scalars/ip/ipv6.rs

@@ -294,7 +294,7 @@ mod tests {
         let input = Arc::new(StringVector::from_slice(&values)) as VectorRef;
 
         // Convert IPv6 addresses to binary
-        let binary_result = to_num.eval(&ctx, &[input.clone()]).unwrap();
+        let binary_result = to_num.eval(&ctx, std::slice::from_ref(&input)).unwrap();
 
         // Convert binary to hex string representation (for ipv6_num_to_string)
         let mut hex_strings = Vec::new();

src/common/function/src/scalars/matches_term.rs

@@ -12,8 +12,9 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+use std::fmt;
+use std::iter::repeat_n;
 use std::sync::Arc;
-use std::{fmt, iter};
 
 use common_query::error::{InvalidFuncArgsSnafu, Result};
 use common_query::prelude::Volatility;
@@ -126,9 +127,10 @@ impl Function for MatchesTermFunction {
             let term = term_column.get_ref(0).as_string().unwrap();
             match term {
                 None => {
-                    return Ok(Arc::new(BooleanVector::from_iter(
-                        iter::repeat(None).take(text_column.len()),
-                    )));
+                    return Ok(Arc::new(BooleanVector::from_iter(repeat_n(
+                        None,
+                        text_column.len(),
+                    ))));
                 }
                 Some(term) => Some(MatchesTermFinder::new(term)),
             }
@@ -217,7 +219,7 @@ impl MatchesTermFinder {
         }
 
         let mut pos = 0;
-        while let Some(found_pos) = self.finder.find(text[pos..].as_bytes()) {
+        while let Some(found_pos) = self.finder.find(&text.as_bytes()[pos..]) {
             let actual_pos = pos + found_pos;
 
             let prev_ok = self.starts_with_non_alnum

src/common/function/src/scalars/math.rs

@@ -12,7 +12,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-mod clamp;
+pub mod clamp;
 mod modulo;
 mod pow;
 mod rate;
@@ -20,7 +20,7 @@ mod rate;
 use std::fmt;
 use std::sync::Arc;
 
-pub use clamp::ClampFunction;
+pub use clamp::{ClampFunction, ClampMaxFunction, ClampMinFunction};
 use common_query::error::{GeneralDataFusionSnafu, Result};
 use common_query::prelude::Signature;
 use datafusion::error::DataFusionError;
@@ -44,6 +44,8 @@ impl MathFunction {
         registry.register(Arc::new(RateFunction));
         registry.register(Arc::new(RangeFunction));
         registry.register(Arc::new(ClampFunction));
+        registry.register(Arc::new(ClampMinFunction));
+        registry.register(Arc::new(ClampMaxFunction));
     }
 }
 

src/common/function/src/scalars/math/clamp.rs

@@ -155,6 +155,182 @@ fn clamp_impl<T: LogicalPrimitiveType, const CLAMP_MIN: bool, const CLAMP_MAX: b
     Ok(Arc::new(PrimitiveVector::<T>::from(result)))
 }
 
+#[derive(Clone, Debug, Default)]
+pub struct ClampMinFunction;
+
+const CLAMP_MIN_NAME: &str = "clamp_min";
+
+impl Function for ClampMinFunction {
+    fn name(&self) -> &str {
+        CLAMP_MIN_NAME
+    }
+
+    fn return_type(&self, input_types: &[ConcreteDataType]) -> Result<ConcreteDataType> {
+        Ok(input_types[0].clone())
+    }
+
+    fn signature(&self) -> Signature {
+        // input, min
+        Signature::uniform(2, ConcreteDataType::numerics(), Volatility::Immutable)
+    }
+
+    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 2, have: {}",
+                    columns.len()
+                ),
+            }
+        );
+        ensure!(
+            columns[0].data_type().is_numeric(),
+            InvalidFuncArgsSnafu {
+                err_msg: format!(
+                    "The first arg's type is not numeric, have: {}",
+                    columns[0].data_type()
+                ),
+            }
+        );
+        ensure!(
+            columns[0].data_type() == columns[1].data_type(),
+            InvalidFuncArgsSnafu {
+                err_msg: format!(
+                    "Arguments don't have identical types: {}, {}",
+                    columns[0].data_type(),
+                    columns[1].data_type()
+                ),
+            }
+        );
+        ensure!(
+            columns[1].len() == 1,
+            InvalidFuncArgsSnafu {
+                err_msg: format!(
+                    "The second arg (min) should be scalar, have: {:?}",
+                    columns[1]
+                ),
+            }
+        );
+
+        with_match_primitive_type_id!(columns[0].data_type().logical_type_id(), |$S| {
+            let input_array = columns[0].to_arrow_array();
+            let input = input_array
+                .as_any()
+                .downcast_ref::<PrimitiveArray<<$S as LogicalPrimitiveType>::ArrowPrimitive>>()
+                .unwrap();
+
+            let min = TryAsPrimitive::<$S>::try_as_primitive(&columns[1].get(0))
+                .with_context(|| {
+                    InvalidFuncArgsSnafu {
+                        err_msg: "The second arg (min) should not be none",
+                    }
+                })?;
+            // For clamp_min, max is effectively infinity, so we don't use it in the clamp_impl logic.
+            // We pass a default/dummy value for max.
+            let max_dummy = <$S as LogicalPrimitiveType>::Native::default();
+
+            clamp_impl::<$S, true, false>(input, min, max_dummy)
+        },{
+            unreachable!()
+        })
+    }
+}
+
+impl Display for ClampMinFunction {
+    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+        write!(f, "{}", CLAMP_MIN_NAME.to_ascii_uppercase())
+    }
+}
+
+#[derive(Clone, Debug, Default)]
+pub struct ClampMaxFunction;
+
+const CLAMP_MAX_NAME: &str = "clamp_max";
+
+impl Function for ClampMaxFunction {
+    fn name(&self) -> &str {
+        CLAMP_MAX_NAME
+    }
+
+    fn return_type(&self, input_types: &[ConcreteDataType]) -> Result<ConcreteDataType> {
+        Ok(input_types[0].clone())
+    }
+
+    fn signature(&self) -> Signature {
+        // input, max
+        Signature::uniform(2, ConcreteDataType::numerics(), Volatility::Immutable)
+    }
+
+    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 2, have: {}",
+                    columns.len()
+                ),
+            }
+        );
+        ensure!(
+            columns[0].data_type().is_numeric(),
+            InvalidFuncArgsSnafu {
+                err_msg: format!(
+                    "The first arg's type is not numeric, have: {}",
+                    columns[0].data_type()
+                ),
+            }
+        );
+        ensure!(
+            columns[0].data_type() == columns[1].data_type(),
+            InvalidFuncArgsSnafu {
+                err_msg: format!(
+                    "Arguments don't have identical types: {}, {}",
+                    columns[0].data_type(),
+                    columns[1].data_type()
+                ),
+            }
+        );
+        ensure!(
+            columns[1].len() == 1,
+            InvalidFuncArgsSnafu {
+                err_msg: format!(
+                    "The second arg (max) should be scalar, have: {:?}",
+                    columns[1]
+                ),
+            }
+        );
+
+        with_match_primitive_type_id!(columns[0].data_type().logical_type_id(), |$S| {
+            let input_array = columns[0].to_arrow_array();
+            let input = input_array
+                .as_any()
+                .downcast_ref::<PrimitiveArray<<$S as LogicalPrimitiveType>::ArrowPrimitive>>()
+                .unwrap();
+
+            let max = TryAsPrimitive::<$S>::try_as_primitive(&columns[1].get(0))
+                .with_context(|| {
+                    InvalidFuncArgsSnafu {
+                        err_msg: "The second arg (max) should not be none",
+                    }
+                })?;
+            // For clamp_max, min is effectively -infinity, so we don't use it in the clamp_impl logic.
+            // We pass a default/dummy value for min.
+            let min_dummy = <$S as LogicalPrimitiveType>::Native::default();
+
+            clamp_impl::<$S, false, true>(input, min_dummy, max)
+        },{
+            unreachable!()
+        })
+    }
+}
+
+impl Display for ClampMaxFunction {
+    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+        write!(f, "{}", CLAMP_MAX_NAME.to_ascii_uppercase())
+    }
+}
+
 #[cfg(test)]
 mod test {
 
@@ -394,4 +570,134 @@ mod test {
         let result = func.eval(&FunctionContext::default(), args.as_slice());
         assert!(result.is_err());
     }
+
+    #[test]
+    fn clamp_min_i64() {
+        let inputs = [
+            (
+                vec![Some(-3), Some(-2), Some(-1), Some(0), Some(1), Some(2)],
+                -1,
+                vec![Some(-1), Some(-1), Some(-1), Some(0), Some(1), Some(2)],
+            ),
+            (
+                vec![Some(-3), None, Some(-1), None, None, Some(2)],
+                -2,
+                vec![Some(-2), None, Some(-1), None, None, Some(2)],
+            ),
+        ];
+
+        let func = ClampMinFunction;
+        for (in_data, min, expected) in inputs {
+            let args = [
+                Arc::new(Int64Vector::from(in_data)) as _,
+                Arc::new(Int64Vector::from_vec(vec![min])) as _,
+            ];
+            let result = func
+                .eval(&FunctionContext::default(), args.as_slice())
+                .unwrap();
+            let expected: VectorRef = Arc::new(Int64Vector::from(expected));
+            assert_eq!(expected, result);
+        }
+    }
+
+    #[test]
+    fn clamp_max_i64() {
+        let inputs = [
+            (
+                vec![Some(-3), Some(-2), Some(-1), Some(0), Some(1), Some(2)],
+                1,
+                vec![Some(-3), Some(-2), Some(-1), Some(0), Some(1), Some(1)],
+            ),
+            (
+                vec![Some(-3), None, Some(-1), None, None, Some(2)],
+                0,
+                vec![Some(-3), None, Some(-1), None, None, Some(0)],
+            ),
+        ];
+
+        let func = ClampMaxFunction;
+        for (in_data, max, expected) in inputs {
+            let args = [
+                Arc::new(Int64Vector::from(in_data)) as _,
+                Arc::new(Int64Vector::from_vec(vec![max])) as _,
+            ];
+            let result = func
+                .eval(&FunctionContext::default(), args.as_slice())
+                .unwrap();
+            let expected: VectorRef = Arc::new(Int64Vector::from(expected));
+            assert_eq!(expected, result);
+        }
+    }
+
+    #[test]
+    fn clamp_min_f64() {
+        let inputs = [(
+            vec![Some(-3.0), Some(-2.0), Some(-1.0), Some(0.0), Some(1.0)],
+            -1.0,
+            vec![Some(-1.0), Some(-1.0), Some(-1.0), Some(0.0), Some(1.0)],
+        )];
+
+        let func = ClampMinFunction;
+        for (in_data, min, expected) in inputs {
+            let args = [
+                Arc::new(Float64Vector::from(in_data)) as _,
+                Arc::new(Float64Vector::from_vec(vec![min])) as _,
+            ];
+            let result = func
+                .eval(&FunctionContext::default(), args.as_slice())
+                .unwrap();
+            let expected: VectorRef = Arc::new(Float64Vector::from(expected));
+            assert_eq!(expected, result);
+        }
+    }
+
+    #[test]
+    fn clamp_max_f64() {
+        let inputs = [(
+            vec![Some(-3.0), Some(-2.0), Some(-1.0), Some(0.0), Some(1.0)],
+            0.0,
+            vec![Some(-3.0), Some(-2.0), Some(-1.0), Some(0.0), Some(0.0)],
+        )];
+
+        let func = ClampMaxFunction;
+        for (in_data, max, expected) in inputs {
+            let args = [
+                Arc::new(Float64Vector::from(in_data)) as _,
+                Arc::new(Float64Vector::from_vec(vec![max])) as _,
+            ];
+            let result = func
+                .eval(&FunctionContext::default(), args.as_slice())
+                .unwrap();
+            let expected: VectorRef = Arc::new(Float64Vector::from(expected));
+            assert_eq!(expected, result);
+        }
+    }
+
+    #[test]
+    fn clamp_min_type_not_match() {
+        let input = vec![Some(-3.0), Some(-2.0), Some(-1.0), Some(0.0), Some(1.0)];
+        let min = -1;
+
+        let func = ClampMinFunction;
+        let args = [
+            Arc::new(Float64Vector::from(input)) as _,
+            Arc::new(Int64Vector::from_vec(vec![min])) as _,
+        ];
+        let result = func.eval(&FunctionContext::default(), args.as_slice());
+        assert!(result.is_err());
+    }
+
+    #[test]
+    fn clamp_max_type_not_match() {
+        let input = vec![Some(-3.0), Some(-2.0), Some(-1.0), Some(0.0), Some(1.0)];
+        let max = 1;
+
+        let func = ClampMaxFunction;
+        let args = [
+            Arc::new(Float64Vector::from(input)) as _,
+            Arc::new(Int64Vector::from_vec(vec![max])) as _,
+        ];
+        let result = func.eval(&FunctionContext::default(), args.as_slice());
+        assert!(result.is_err());
+    }
 }

src/common/function/src/scalars/math/rate.rs

@@ -37,7 +37,7 @@ impl fmt::Display for RateFunction {
 
 impl Function for RateFunction {
     fn name(&self) -> &str {
-        "prom_rate"
+        "rate"
     }
 
     fn return_type(&self, _input_types: &[ConcreteDataType]) -> Result<ConcreteDataType> {
@@ -82,7 +82,7 @@ mod tests {
     #[test]
     fn test_rate_function() {
         let rate = RateFunction;
-        assert_eq!("prom_rate", rate.name());
+        assert_eq!("rate", rate.name());
         assert_eq!(
             ConcreteDataType::float64_datatype(),
             rate.return_type(&[]).unwrap()

src/common/function/src/scalars/timestamp.rs

@@ -13,10 +13,8 @@
 // limitations under the License.
 
 use std::sync::Arc;
-mod greatest;
 mod to_unixtime;
 
-use greatest::GreatestFunction;
 use to_unixtime::ToUnixtimeFunction;
 
 use crate::function_registry::FunctionRegistry;
@@ -26,6 +24,5 @@ pub(crate) struct TimestampFunction;
 impl TimestampFunction {
     pub fn register(registry: &FunctionRegistry) {
         registry.register(Arc::new(ToUnixtimeFunction));
-        registry.register(Arc::new(GreatestFunction));
     }
 }

src/common/function/src/scalars/timestamp/greatest.rs

@@ -1,328 +0,0 @@
-// 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::fmt::{self};
-
-use common_query::error::{
-    self, ArrowComputeSnafu, InvalidFuncArgsSnafu, Result, UnsupportedInputDataTypeSnafu,
-};
-use common_query::prelude::{Signature, Volatility};
-use datafusion::arrow::compute::kernels::cmp::gt;
-use datatypes::arrow::array::AsArray;
-use datatypes::arrow::compute::cast;
-use datatypes::arrow::compute::kernels::zip;
-use datatypes::arrow::datatypes::{
-    DataType as ArrowDataType, Date32Type, TimeUnit, TimestampMicrosecondType,
-    TimestampMillisecondType, TimestampNanosecondType, TimestampSecondType,
-};
-use datatypes::prelude::ConcreteDataType;
-use datatypes::types::TimestampType;
-use datatypes::vectors::{Helper, VectorRef};
-use snafu::{ensure, ResultExt};
-
-use crate::function::{Function, FunctionContext};
-
-#[derive(Clone, Debug, Default)]
-pub struct GreatestFunction;
-
-const NAME: &str = "greatest";
-
-macro_rules! gt_time_types {
-    ($ty: ident, $columns:expr) => {{
-        let column1 = $columns[0].to_arrow_array();
-        let column2 = $columns[1].to_arrow_array();
-
-        let column1 = column1.as_primitive::<$ty>();
-        let column2 = column2.as_primitive::<$ty>();
-        let boolean_array = gt(&column1, &column2).context(ArrowComputeSnafu)?;
-
-        let result = zip::zip(&boolean_array, &column1, &column2).context(ArrowComputeSnafu)?;
-        Helper::try_into_vector(&result).context(error::FromArrowArraySnafu)
-    }};
-}
-
-impl Function for GreatestFunction {
-    fn name(&self) -> &str {
-        NAME
-    }
-
-    fn return_type(&self, input_types: &[ConcreteDataType]) -> Result<ConcreteDataType> {
-        ensure!(
-            input_types.len() == 2,
-            InvalidFuncArgsSnafu {
-                err_msg: format!(
-                    "The length of the args is not correct, expect exactly two, have: {}",
-                    input_types.len()
-                )
-            }
-        );
-
-        match &input_types[0] {
-            ConcreteDataType::String(_) => Ok(ConcreteDataType::timestamp_millisecond_datatype()),
-            ConcreteDataType::Date(_) => Ok(ConcreteDataType::date_datatype()),
-            ConcreteDataType::Timestamp(ts_type) => Ok(ConcreteDataType::Timestamp(*ts_type)),
-            _ => UnsupportedInputDataTypeSnafu {
-                function: NAME,
-                datatypes: input_types,
-            }
-            .fail(),
-        }
-    }
-
-    fn signature(&self) -> Signature {
-        Signature::uniform(
-            2,
-            vec![
-                ConcreteDataType::string_datatype(),
-                ConcreteDataType::date_datatype(),
-                ConcreteDataType::timestamp_nanosecond_datatype(),
-                ConcreteDataType::timestamp_microsecond_datatype(),
-                ConcreteDataType::timestamp_millisecond_datatype(),
-                ConcreteDataType::timestamp_second_datatype(),
-            ],
-            Volatility::Immutable,
-        )
-    }
-
-    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()
-                ),
-            }
-        );
-        match columns[0].data_type() {
-            ConcreteDataType::String(_) => {
-                let column1 = cast(
-                    &columns[0].to_arrow_array(),
-                    &ArrowDataType::Timestamp(TimeUnit::Millisecond, None),
-                )
-                .context(ArrowComputeSnafu)?;
-                let column1 = column1.as_primitive::<TimestampMillisecondType>();
-                let column2 = cast(
-                    &columns[1].to_arrow_array(),
-                    &ArrowDataType::Timestamp(TimeUnit::Millisecond, None),
-                )
-                .context(ArrowComputeSnafu)?;
-                let column2 = column2.as_primitive::<TimestampMillisecondType>();
-                let boolean_array = gt(&column1, &column2).context(ArrowComputeSnafu)?;
-                let result =
-                    zip::zip(&boolean_array, &column1, &column2).context(ArrowComputeSnafu)?;
-                Ok(Helper::try_into_vector(&result).context(error::FromArrowArraySnafu)?)
-            }
-            ConcreteDataType::Date(_) => gt_time_types!(Date32Type, columns),
-            ConcreteDataType::Timestamp(ts_type) => match ts_type {
-                TimestampType::Second(_) => gt_time_types!(TimestampSecondType, columns),
-                TimestampType::Millisecond(_) => {
-                    gt_time_types!(TimestampMillisecondType, columns)
-                }
-                TimestampType::Microsecond(_) => {
-                    gt_time_types!(TimestampMicrosecondType, columns)
-                }
-                TimestampType::Nanosecond(_) => {
-                    gt_time_types!(TimestampNanosecondType, columns)
-                }
-            },
-            _ => UnsupportedInputDataTypeSnafu {
-                function: NAME,
-                datatypes: columns.iter().map(|c| c.data_type()).collect::<Vec<_>>(),
-            }
-            .fail(),
-        }
-    }
-}
-
-impl fmt::Display for GreatestFunction {
-    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
-        write!(f, "GREATEST")
-    }
-}
-
-#[cfg(test)]
-mod tests {
-    use std::sync::Arc;
-
-    use common_time::timestamp::TimeUnit;
-    use common_time::{Date, Timestamp};
-    use datatypes::types::{
-        DateType, TimestampMicrosecondType, TimestampMillisecondType, TimestampNanosecondType,
-        TimestampSecondType,
-    };
-    use datatypes::value::Value;
-    use datatypes::vectors::{
-        DateVector, StringVector, TimestampMicrosecondVector, TimestampMillisecondVector,
-        TimestampNanosecondVector, TimestampSecondVector, Vector,
-    };
-    use paste::paste;
-
-    use super::*;
-    #[test]
-    fn test_greatest_takes_string_vector() {
-        let function = GreatestFunction;
-        assert_eq!(
-            function
-                .return_type(&[
-                    ConcreteDataType::string_datatype(),
-                    ConcreteDataType::string_datatype()
-                ])
-                .unwrap(),
-            ConcreteDataType::timestamp_millisecond_datatype()
-        );
-        let columns = vec![
-            Arc::new(StringVector::from(vec![
-                "1970-01-01".to_string(),
-                "2012-12-23".to_string(),
-            ])) as _,
-            Arc::new(StringVector::from(vec![
-                "2001-02-01".to_string(),
-                "1999-01-01".to_string(),
-            ])) as _,
-        ];
-
-        let result = function
-            .eval(&FunctionContext::default(), &columns)
-            .unwrap();
-        let result = result
-            .as_any()
-            .downcast_ref::<TimestampMillisecondVector>()
-            .unwrap();
-        assert_eq!(result.len(), 2);
-        assert_eq!(
-            result.get(0),
-            Value::Timestamp(Timestamp::from_str("2001-02-01 00:00:00", None).unwrap())
-        );
-        assert_eq!(
-            result.get(1),
-            Value::Timestamp(Timestamp::from_str("2012-12-23 00:00:00", None).unwrap())
-        );
-    }
-
-    #[test]
-    fn test_greatest_takes_date_vector() {
-        let function = GreatestFunction;
-        assert_eq!(
-            function
-                .return_type(&[
-                    ConcreteDataType::date_datatype(),
-                    ConcreteDataType::date_datatype()
-                ])
-                .unwrap(),
-            ConcreteDataType::Date(DateType)
-        );
-
-        let columns = vec![
-            Arc::new(DateVector::from_slice(vec![-1, 2])) as _,
-            Arc::new(DateVector::from_slice(vec![0, 1])) as _,
-        ];
-
-        let result = function
-            .eval(&FunctionContext::default(), &columns)
-            .unwrap();
-        let result = result.as_any().downcast_ref::<DateVector>().unwrap();
-        assert_eq!(result.len(), 2);
-        assert_eq!(
-            result.get(0),
-            Value::Date(Date::from_str_utc("1970-01-01").unwrap())
-        );
-        assert_eq!(
-            result.get(1),
-            Value::Date(Date::from_str_utc("1970-01-03").unwrap())
-        );
-    }
-
-    #[test]
-    fn test_greatest_takes_datetime_vector() {
-        let function = GreatestFunction;
-        assert_eq!(
-            function
-                .return_type(&[
-                    ConcreteDataType::timestamp_millisecond_datatype(),
-                    ConcreteDataType::timestamp_millisecond_datatype()
-                ])
-                .unwrap(),
-            ConcreteDataType::timestamp_millisecond_datatype()
-        );
-
-        let columns = vec![
-            Arc::new(TimestampMillisecondVector::from_slice(vec![-1, 2])) as _,
-            Arc::new(TimestampMillisecondVector::from_slice(vec![0, 1])) as _,
-        ];
-
-        let result = function
-            .eval(&FunctionContext::default(), &columns)
-            .unwrap();
-        let result = result
-            .as_any()
-            .downcast_ref::<TimestampMillisecondVector>()
-            .unwrap();
-        assert_eq!(result.len(), 2);
-        assert_eq!(
-            result.get(0),
-            Value::Timestamp(Timestamp::from_str("1970-01-01 00:00:00", None).unwrap())
-        );
-        assert_eq!(
-            result.get(1),
-            Value::Timestamp(Timestamp::from_str("1970-01-01 00:00:00.002", None).unwrap())
-        );
-    }
-
-    macro_rules! test_timestamp {
-        ($type: expr,$unit: ident) => {
-            paste! {
-                #[test]
-                fn [<test_greatest_takes_ $unit:lower _vector>]() {
-                    let function = GreatestFunction;
-                    assert_eq!(
-                        function.return_type(&[$type, $type]).unwrap(),
-                        ConcreteDataType::Timestamp(TimestampType::$unit([<Timestamp $unit Type>]))
-                    );
-
-                    let columns = vec![
-                        Arc::new([<Timestamp $unit Vector>]::from_slice(vec![-1, 2])) as _,
-                        Arc::new([<Timestamp $unit Vector>]::from_slice(vec![0, 1])) as _,
-                    ];
-
-                    let result = function.eval(&FunctionContext::default(), &columns).unwrap();
-                    let result = result.as_any().downcast_ref::<[<Timestamp $unit Vector>]>().unwrap();
-                    assert_eq!(result.len(), 2);
-                    assert_eq!(
-                        result.get(0),
-                        Value::Timestamp(Timestamp::new(0, TimeUnit::$unit))
-                    );
-                    assert_eq!(
-                        result.get(1),
-                        Value::Timestamp(Timestamp::new(2, TimeUnit::$unit))
-                    );
-                }
-            }
-        }
-    }
-
-    test_timestamp!(
-        ConcreteDataType::timestamp_nanosecond_datatype(),
-        Nanosecond
-    );
-    test_timestamp!(
-        ConcreteDataType::timestamp_microsecond_datatype(),
-        Microsecond
-    );
-    test_timestamp!(
-        ConcreteDataType::timestamp_millisecond_datatype(),
-        Millisecond
-    );
-    test_timestamp!(ConcreteDataType::timestamp_second_datatype(), Second);
-}

src/common/grpc/Cargo.toml

@@ -10,6 +10,7 @@ workspace = true
 [dependencies]
 api.workspace = true
 arrow-flight.workspace = true
+bytes.workspace = true
 common-base.workspace = true
 common-error.workspace = true
 common-macro.workspace = true

src/common/grpc/benches/bench_flight_decoder.rs

@@ -0,0 +1,146 @@
+// 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 arrow_flight::FlightData;
+use bytes::Bytes;
+use common_grpc::flight::{FlightDecoder, FlightEncoder, FlightMessage};
+use common_recordbatch::DfRecordBatch;
+use criterion::{criterion_group, criterion_main, Criterion};
+use datatypes::arrow;
+use datatypes::arrow::array::{ArrayRef, Int64Array, StringArray, TimestampMillisecondArray};
+use datatypes::arrow::datatypes::DataType;
+use datatypes::data_type::ConcreteDataType;
+use datatypes::schema::{ColumnSchema, Schema};
+use prost::Message;
+
+fn schema() -> arrow::datatypes::SchemaRef {
+    let schema = Schema::new(vec![
+        ColumnSchema::new("k0", ConcreteDataType::string_datatype(), false),
+        ColumnSchema::new("k1", ConcreteDataType::string_datatype(), false),
+        ColumnSchema::new(
+            "ts",
+            ConcreteDataType::timestamp_millisecond_datatype(),
+            false,
+        ),
+        ColumnSchema::new("v0", ConcreteDataType::int64_datatype(), false),
+        ColumnSchema::new("v1", ConcreteDataType::int64_datatype(), false),
+    ]);
+    schema.arrow_schema().clone()
+}
+
+/// Generate record batch according to provided schema and num rows.
+fn prepare_random_record_batch(
+    schema: arrow::datatypes::SchemaRef,
+    num_rows: usize,
+) -> DfRecordBatch {
+    let tag_candidates = (0..10000).map(|i| i.to_string()).collect::<Vec<_>>();
+
+    let columns: Vec<ArrayRef> = schema
+        .fields
+        .iter()
+        .map(|col| match col.data_type() {
+            DataType::Utf8 => {
+                let array = StringArray::from(
+                    (0..num_rows)
+                        .map(|_| {
+                            let idx: usize = rand::random_range(0..10000);
+                            format!("tag-{}", tag_candidates[idx])
+                        })
+                        .collect::<Vec<_>>(),
+                );
+                Arc::new(array) as ArrayRef
+            }
+            DataType::Timestamp(_, _) => {
+                let now = common_time::util::current_time_millis();
+                let array = TimestampMillisecondArray::from(
+                    (0..num_rows).map(|i| now + i as i64).collect::<Vec<_>>(),
+                );
+                Arc::new(array) as ArrayRef
+            }
+            DataType::Int64 => {
+                let array = Int64Array::from((0..num_rows).map(|i| i as i64).collect::<Vec<_>>());
+                Arc::new(array) as ArrayRef
+            }
+            _ => unreachable!(),
+        })
+        .collect();
+
+    DfRecordBatch::try_new(schema, columns).unwrap()
+}
+
+fn prepare_flight_data(num_rows: usize) -> (FlightData, FlightData) {
+    let schema = schema();
+    let mut encoder = FlightEncoder::default();
+    let schema_data = encoder.encode(FlightMessage::Schema(schema.clone()));
+    let rb = prepare_random_record_batch(schema, num_rows);
+    let rb_data = encoder.encode(FlightMessage::RecordBatch(rb));
+    (schema_data, rb_data)
+}
+
+fn decode_flight_data_from_protobuf(schema: &Bytes, payload: &Bytes) -> DfRecordBatch {
+    let schema = FlightData::decode(&schema[..]).unwrap();
+    let payload = FlightData::decode(&payload[..]).unwrap();
+    let mut decoder = FlightDecoder::default();
+    let _schema = decoder.try_decode(&schema).unwrap();
+    let message = decoder.try_decode(&payload).unwrap();
+    let FlightMessage::RecordBatch(batch) = message else {
+        unreachable!("unexpected message");
+    };
+    batch
+}
+
+fn decode_flight_data_from_header_and_body(
+    schema: &Bytes,
+    data_header: &Bytes,
+    data_body: &Bytes,
+) -> DfRecordBatch {
+    let mut decoder = FlightDecoder::try_from_schema_bytes(schema).unwrap();
+    decoder
+        .try_decode_record_batch(data_header, data_body)
+        .unwrap()
+}
+
+fn bench_decode_flight_data(c: &mut Criterion) {
+    let row_counts = [100000, 200000, 1000000];
+
+    for row_count in row_counts {
+        let (schema, payload) = prepare_flight_data(row_count);
+
+        // arguments for decode_flight_data_from_protobuf
+        let schema_bytes = Bytes::from(schema.encode_to_vec());
+        let payload_bytes = Bytes::from(payload.encode_to_vec());
+
+        let mut group = c.benchmark_group(format!("flight_decoder_{}_rows", row_count));
+        group.bench_function("decode_from_protobuf", |b| {
+            b.iter(|| decode_flight_data_from_protobuf(&schema_bytes, &payload_bytes));
+        });
+
+        group.bench_function("decode_from_header_and_body", |b| {
+            b.iter(|| {
+                decode_flight_data_from_header_and_body(
+                    &schema.data_header,
+                    &payload.data_header,
+                    &payload.data_body,
+                )
+            });
+        });
+
+        group.finish();
+    }
+}
+
+criterion_group!(benches, bench_decode_flight_data);
+criterion_main!(benches);

src/common/grpc/benches/bench_main.rs

@@ -14,8 +14,10 @@
 
 use criterion::criterion_main;
 
+mod bench_flight_decoder;
 mod channel_manager;
 
 criterion_main! {
-    channel_manager::benches
+    channel_manager::benches,
+    bench_flight_decoder::benches
 }

src/common/grpc/src/error.rs

@@ -18,6 +18,7 @@ use std::io;
 use common_error::ext::ErrorExt;
 use common_error::status_code::StatusCode;
 use common_macro::stack_trace_debug;
+use datatypes::arrow::error::ArrowError;
 use snafu::{Location, Snafu};
 
 pub type Result<T> = std::result::Result<T, Error>;
@@ -59,13 +60,6 @@ pub enum Error {
         location: Location,
     },
 
-    #[snafu(display("Failed to create RecordBatch"))]
-    CreateRecordBatch {
-        #[snafu(implicit)]
-        location: Location,
-        source: common_recordbatch::error::Error,
-    },
-
     #[snafu(display("Failed to convert Arrow type: {}", from))]
     Conversion {
         from: String,
@@ -88,13 +82,6 @@ pub enum Error {
         location: Location,
     },
 
-    #[snafu(display("Failed to convert Arrow Schema"))]
-    ConvertArrowSchema {
-        #[snafu(implicit)]
-        location: Location,
-        source: datatypes::error::Error,
-    },
-
     #[snafu(display("Not supported: {}", feat))]
     NotSupported { feat: String },
 
@@ -105,6 +92,14 @@ pub enum Error {
         #[snafu(implicit)]
         location: Location,
     },
+
+    #[snafu(display("Failed arrow operation"))]
+    Arrow {
+        #[snafu(implicit)]
+        location: Location,
+        #[snafu(source)]
+        error: ArrowError,
+    },
 }
 
 impl ErrorExt for Error {
@@ -121,8 +116,7 @@ impl ErrorExt for Error {
             | Error::DecodeFlightData { .. }
             | Error::SerdeJson { .. } => StatusCode::Internal,
 
-            Error::CreateRecordBatch { source, .. } => source.status_code(),
-            Error::ConvertArrowSchema { source, .. } => source.status_code(),
+            Error::Arrow { .. } => StatusCode::Internal,
         }
     }
 

src/common/grpc/src/flight.rs

@@ -21,25 +21,24 @@ use api::v1::{AffectedRows, FlightMetadata, Metrics};
 use arrow_flight::utils::flight_data_to_arrow_batch;
 use arrow_flight::{FlightData, SchemaAsIpc};
 use common_base::bytes::Bytes;
-use common_recordbatch::{RecordBatch, RecordBatches};
+use common_recordbatch::DfRecordBatch;
 use datatypes::arrow;
-use datatypes::arrow::datatypes::Schema as ArrowSchema;
-use datatypes::arrow::ipc::{root_as_message, writer, MessageHeader};
-use datatypes::schema::{Schema, SchemaRef};
+use datatypes::arrow::buffer::Buffer;
+use datatypes::arrow::datatypes::{Schema as ArrowSchema, SchemaRef};
+use datatypes::arrow::error::ArrowError;
+use datatypes::arrow::ipc::{convert, reader, root_as_message, writer, MessageHeader};
 use flatbuffers::FlatBufferBuilder;
 use prost::bytes::Bytes as ProstBytes;
 use prost::Message;
 use snafu::{OptionExt, ResultExt};
 
-use crate::error::{
-    ConvertArrowSchemaSnafu, CreateRecordBatchSnafu, DecodeFlightDataSnafu, InvalidFlightDataSnafu,
-    Result,
-};
+use crate::error;
+use crate::error::{DecodeFlightDataSnafu, InvalidFlightDataSnafu, Result};
 
 #[derive(Debug, Clone)]
 pub enum FlightMessage {
     Schema(SchemaRef),
-    Recordbatch(RecordBatch),
+    RecordBatch(DfRecordBatch),
     AffectedRows(usize),
     Metrics(String),
 }
@@ -67,14 +66,12 @@ impl Default for FlightEncoder {
 impl FlightEncoder {
     pub fn encode(&mut self, flight_message: FlightMessage) -> FlightData {
         match flight_message {
-            FlightMessage::Schema(schema) => {
-                SchemaAsIpc::new(schema.arrow_schema(), &self.write_options).into()
-            }
-            FlightMessage::Recordbatch(recordbatch) => {
+            FlightMessage::Schema(schema) => SchemaAsIpc::new(&schema, &self.write_options).into(),
+            FlightMessage::RecordBatch(record_batch) => {
                 let (encoded_dictionaries, encoded_batch) = self
                     .data_gen
                     .encoded_batch(
-                        recordbatch.df_record_batch(),
+                        &record_batch,
                         &mut self.dictionary_tracker,
                         &self.write_options,
                     )
@@ -124,10 +121,59 @@ impl FlightEncoder {
 #[derive(Default)]
 pub struct FlightDecoder {
     schema: Option<SchemaRef>,
+    schema_bytes: Option<bytes::Bytes>,
 }
 
 impl FlightDecoder {
-    pub fn try_decode(&mut self, flight_data: FlightData) -> Result<FlightMessage> {
+    /// Build a [FlightDecoder] instance from provided schema bytes.
+    pub fn try_from_schema_bytes(schema_bytes: &bytes::Bytes) -> Result<Self> {
+        let arrow_schema = convert::try_schema_from_flatbuffer_bytes(&schema_bytes[..])
+            .context(error::ArrowSnafu)?;
+        Ok(Self {
+            schema: Some(Arc::new(arrow_schema)),
+            schema_bytes: Some(schema_bytes.clone()),
+        })
+    }
+
+    pub fn try_decode_record_batch(
+        &mut self,
+        data_header: &bytes::Bytes,
+        data_body: &bytes::Bytes,
+    ) -> Result<DfRecordBatch> {
+        let schema = self
+            .schema
+            .as_ref()
+            .context(InvalidFlightDataSnafu {
+                reason: "Should have decoded schema first!",
+            })?
+            .clone();
+        let message = root_as_message(&data_header[..])
+            .map_err(|err| {
+                ArrowError::ParseError(format!("Unable to get root as message: {err:?}"))
+            })
+            .context(error::ArrowSnafu)?;
+        let result = message
+            .header_as_record_batch()
+            .ok_or_else(|| {
+                ArrowError::ParseError(
+                    "Unable to convert flight data header to a record batch".to_string(),
+                )
+            })
+            .and_then(|batch| {
+                reader::read_record_batch(
+                    &Buffer::from(data_body.as_ref()),
+                    batch,
+                    schema,
+                    &HashMap::new(),
+                    None,
+                    &message.version(),
+                )
+            })
+            .context(error::ArrowSnafu)?;
+        Ok(result)
+    }
+
+    pub fn try_decode(&mut self, flight_data: &FlightData) -> Result<FlightMessage> {
         let message = root_as_message(&flight_data.data_header).map_err(|e| {
             InvalidFlightDataSnafu {
                 reason: e.to_string(),
@@ -136,7 +182,7 @@ impl FlightDecoder {
         })?;
         match message.header_type() {
             MessageHeader::NONE => {
-                let metadata = FlightMetadata::decode(flight_data.app_metadata)
+                let metadata = FlightMetadata::decode(flight_data.app_metadata.clone())
                     .context(DecodeFlightDataSnafu)?;
                 if let Some(AffectedRows { value }) = metadata.affected_rows {
                     return Ok(FlightMessage::AffectedRows(value as _));
@@ -152,36 +198,29 @@ impl FlightDecoder {
                 .fail()
             }
             MessageHeader::Schema => {
-                let arrow_schema = ArrowSchema::try_from(&flight_data).map_err(|e| {
+                let arrow_schema = Arc::new(ArrowSchema::try_from(flight_data).map_err(|e| {
                     InvalidFlightDataSnafu {
                         reason: e.to_string(),
                     }
                     .build()
-                })?;
-                let schema =
-                    Arc::new(Schema::try_from(arrow_schema).context(ConvertArrowSchemaSnafu)?);
-
-                self.schema = Some(schema.clone());
-
-                Ok(FlightMessage::Schema(schema))
+                })?);
+                self.schema = Some(arrow_schema.clone());
+                self.schema_bytes = Some(flight_data.data_header.clone());
+                Ok(FlightMessage::Schema(arrow_schema))
             }
             MessageHeader::RecordBatch => {
                 let schema = self.schema.clone().context(InvalidFlightDataSnafu {
                     reason: "Should have decoded schema first!",
                 })?;
-                let arrow_schema = schema.arrow_schema().clone();
-
                 let arrow_batch =
-                    flight_data_to_arrow_batch(&flight_data, arrow_schema, &HashMap::new())
+                    flight_data_to_arrow_batch(flight_data, schema.clone(), &HashMap::new())
                         .map_err(|e| {
                             InvalidFlightDataSnafu {
                                 reason: e.to_string(),
                             }
                             .build()
                         })?;
-                let recordbatch = RecordBatch::try_from_df_record_batch(schema, arrow_batch)
-                    .context(CreateRecordBatchSnafu)?;
-                Ok(FlightMessage::Recordbatch(recordbatch))
+                Ok(FlightMessage::RecordBatch(arrow_batch))
             }
             other => {
                 let name = other.variant_name().unwrap_or("UNKNOWN");
@@ -192,16 +231,26 @@ impl FlightDecoder {
             }
         }
     }
+
+    pub fn schema(&self) -> Option<&SchemaRef> {
+        self.schema.as_ref()
+    }
+
+    pub fn schema_bytes(&self) -> Option<bytes::Bytes> {
+        self.schema_bytes.clone()
+    }
 }
 
-pub fn flight_messages_to_recordbatches(messages: Vec<FlightMessage>) -> Result<RecordBatches> {
+pub fn flight_messages_to_recordbatches(
+    messages: Vec<FlightMessage>,
+) -> Result<Vec<DfRecordBatch>> {
     if messages.is_empty() {
-        Ok(RecordBatches::empty())
+        Ok(vec![])
     } else {
         let mut recordbatches = Vec::with_capacity(messages.len() - 1);
 
-        let schema = match &messages[0] {
-            FlightMessage::Schema(schema) => schema.clone(),
+        match &messages[0] {
+            FlightMessage::Schema(_schema) => {}
             _ => {
                 return InvalidFlightDataSnafu {
                     reason: "First Flight Message must be schema!",
@@ -212,7 +261,7 @@ pub fn flight_messages_to_recordbatches(messages: Vec<FlightMessage>) -> Result<
 
         for message in messages.into_iter().skip(1) {
             match message {
-                FlightMessage::Recordbatch(recordbatch) => recordbatches.push(recordbatch),
+                FlightMessage::RecordBatch(recordbatch) => recordbatches.push(recordbatch),
                 _ => {
                     return InvalidFlightDataSnafu {
                         reason: "Expect the following Flight Messages are all Recordbatches!",
@@ -222,7 +271,7 @@ pub fn flight_messages_to_recordbatches(messages: Vec<FlightMessage>) -> Result<
             }
         }
 
-        RecordBatches::try_new(schema, recordbatches).context(CreateRecordBatchSnafu)
+        Ok(recordbatches)
     }
 }
 
@@ -243,38 +292,33 @@ fn build_none_flight_msg() -> Bytes {
 #[cfg(test)]
 mod test {
     use arrow_flight::utils::batches_to_flight_data;
-    use datatypes::arrow::datatypes::{DataType, Field};
-    use datatypes::prelude::ConcreteDataType;
-    use datatypes::schema::ColumnSchema;
-    use datatypes::vectors::Int32Vector;
+    use datatypes::arrow::array::Int32Array;
+    use datatypes::arrow::datatypes::{DataType, Field, Schema};
 
     use super::*;
     use crate::Error;
 
     #[test]
     fn test_try_decode() {
-        let arrow_schema = ArrowSchema::new(vec![Field::new("n", DataType::Int32, true)]);
-        let schema = Arc::new(Schema::try_from(arrow_schema.clone()).unwrap());
+        let schema = Arc::new(ArrowSchema::new(vec![Field::new(
+            "n",
+            DataType::Int32,
+            true,
+        )]));
 
-        let batch1 = RecordBatch::new(
+        let batch1 = DfRecordBatch::try_new(
             schema.clone(),
-            vec![Arc::new(Int32Vector::from(vec![Some(1), None, Some(3)])) as _],
+            vec![Arc::new(Int32Array::from(vec![Some(1), None, Some(3)])) as _],
         )
         .unwrap();
-        let batch2 = RecordBatch::new(
+        let batch2 = DfRecordBatch::try_new(
             schema.clone(),
-            vec![Arc::new(Int32Vector::from(vec![None, Some(5)])) as _],
+            vec![Arc::new(Int32Array::from(vec![None, Some(5)])) as _],
         )
         .unwrap();
 
-        let flight_data = batches_to_flight_data(
-            &arrow_schema,
-            vec![
-                batch1.clone().into_df_record_batch(),
-                batch2.clone().into_df_record_batch(),
-            ],
-        )
-        .unwrap();
+        let flight_data =
+            batches_to_flight_data(&schema, vec![batch1.clone(), batch2.clone()]).unwrap();
         assert_eq!(flight_data.len(), 3);
         let [d1, d2, d3] = flight_data.as_slice() else {
             unreachable!()
@@ -283,14 +327,14 @@ mod test {
         let decoder = &mut FlightDecoder::default();
         assert!(decoder.schema.is_none());
 
-        let result = decoder.try_decode(d2.clone());
+        let result = decoder.try_decode(d2);
         assert!(matches!(result, Err(Error::InvalidFlightData { .. })));
         assert!(result
             .unwrap_err()
             .to_string()
             .contains("Should have decoded schema first!"));
 
-        let message = decoder.try_decode(d1.clone()).unwrap();
+        let message = decoder.try_decode(d1).unwrap();
         assert!(matches!(message, FlightMessage::Schema(_)));
         let FlightMessage::Schema(decoded_schema) = message else {
             unreachable!()
@@ -299,16 +343,16 @@ mod test {
 
         let _ = decoder.schema.as_ref().unwrap();
 
-        let message = decoder.try_decode(d2.clone()).unwrap();
-        assert!(matches!(message, FlightMessage::Recordbatch(_)));
-        let FlightMessage::Recordbatch(actual_batch) = message else {
+        let message = decoder.try_decode(d2).unwrap();
+        assert!(matches!(message, FlightMessage::RecordBatch(_)));
+        let FlightMessage::RecordBatch(actual_batch) = message else {
             unreachable!()
         };
         assert_eq!(actual_batch, batch1);
 
-        let message = decoder.try_decode(d3.clone()).unwrap();
-        assert!(matches!(message, FlightMessage::Recordbatch(_)));
-        let FlightMessage::Recordbatch(actual_batch) = message else {
+        let message = decoder.try_decode(d3).unwrap();
+        assert!(matches!(message, FlightMessage::RecordBatch(_)));
+        let FlightMessage::RecordBatch(actual_batch) = message else {
             unreachable!()
         };
         assert_eq!(actual_batch, batch2);
@@ -316,27 +360,22 @@ mod test {
 
     #[test]
     fn test_flight_messages_to_recordbatches() {
-        let schema = Arc::new(Schema::new(vec![ColumnSchema::new(
-            "m",
-            ConcreteDataType::int32_datatype(),
-            true,
-        )]));
-        let batch1 = RecordBatch::new(
+        let schema = Arc::new(Schema::new(vec![Field::new("m", DataType::Int32, true)]));
+        let batch1 = DfRecordBatch::try_new(
             schema.clone(),
-            vec![Arc::new(Int32Vector::from(vec![Some(2), None, Some(4)])) as _],
+            vec![Arc::new(Int32Array::from(vec![Some(2), None, Some(4)])) as _],
         )
         .unwrap();
-        let batch2 = RecordBatch::new(
+        let batch2 = DfRecordBatch::try_new(
             schema.clone(),
-            vec![Arc::new(Int32Vector::from(vec![None, Some(6)])) as _],
+            vec![Arc::new(Int32Array::from(vec![None, Some(6)])) as _],
         )
         .unwrap();
-        let recordbatches =
-            RecordBatches::try_new(schema.clone(), vec![batch1.clone(), batch2.clone()]).unwrap();
+        let recordbatches = vec![batch1.clone(), batch2.clone()];
 
         let m1 = FlightMessage::Schema(schema);
-        let m2 = FlightMessage::Recordbatch(batch1);
-        let m3 = FlightMessage::Recordbatch(batch2);
+        let m2 = FlightMessage::RecordBatch(batch1);
+        let m3 = FlightMessage::RecordBatch(batch2);
 
         let result = flight_messages_to_recordbatches(vec![m2.clone(), m1.clone(), m3.clone()]);
         assert!(matches!(result, Err(Error::InvalidFlightData { .. })));

src/common/grpc/src/lib.rs

@@ -18,4 +18,5 @@ pub mod flight;
 pub mod precision;
 pub mod select;
 
+pub use arrow_flight::FlightData;
 pub use error::Error;

src/common/mem-prof/Cargo.toml

@@ -8,6 +8,7 @@ license.workspace = true
 workspace = true
 
 [dependencies]
+anyhow = "1"
 common-error.workspace = true
 common-macro.workspace = true
 snafu.workspace = true
@@ -16,6 +17,11 @@ tokio.workspace = true
 
 [target.'cfg(not(windows))'.dependencies]
 tikv-jemalloc-ctl = { version = "0.6", features = ["use_std", "stats"] }
+jemalloc-pprof-utils = { version = "0.7", package = "pprof_util", features = [
+    "flamegraph",
+    "symbolize",
+] } # for parsing jemalloc prof dump
+jemalloc-pprof-mappings = { version = "0.7", package = "mappings" } # for get the name of functions in the prof dump
 
 [target.'cfg(not(windows))'.dependencies.tikv-jemalloc-sys]
 features = ["stats", "profiling", "unprefixed_malloc_on_supported_platforms"]