chore: add deprecate develop branch warning

Signed-off-by: Ruihang Xia <waynestxia@gmail.com>

.cargo/config.toml

@@ -12,9 +12,4 @@ rustflags = [
     "-Wclippy::print_stdout",
     "-Wclippy::print_stderr",
     "-Wclippy::implicit_clone",
-
-    # It seems clippy has made a false positive decision here when upgrading rust toolchain to
-    # nightly-2023-08-07, we do need it to be borrowed mutably.
-    # Allow it for now; try disallow it when the toolchain is upgraded in the future.
-    "-Aclippy::needless_pass_by_ref_mut",
 ]

.github/ISSUE_TEMPLATE/bug_report.yml

@@ -41,13 +41,27 @@ body:
       required: true
 
   - type: textarea
-    id: what-happened
+    id: reproduce
     attributes:
-      label: What happened?
+      label: Minimal reproduce step
       description: |
-        Tell us what happened and also what you would have expected to
-        happen instead.
-      placeholder: "Describe the bug"
+        Please walk us through and provide steps and details on how
+        to reproduce the issue. If possible, provide scripts that we
+        can run to trigger the bug.
+    validations:
+      required: true
+
+  - type: textarea
+    id: expected-manner
+    attributes:
+      label: What did you expect to see?
+    validations:
+      required: true
+
+  - type: textarea
+    id: actual-manner
+    attributes:
+      label: What did you see instead?
     validations:
       required: true
 
@@ -72,14 +86,3 @@ body:
         trace. This will be automatically formatted into code, so no
         need for backticks.
       render: bash
-
-  - type: textarea
-    id: reproduce
-    attributes:
-      label: How can we reproduce the bug?
-      description: |
-        Please walk us through and provide steps and details on how
-        to reproduce the issue. If possible, provide scripts that we
-        can run to trigger the bug.
-    validations:
-      required: true

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

@@ -40,9 +40,11 @@ runs:
     - name: Upload artifacts
       uses: ./.github/actions/upload-artifacts
       if: ${{ inputs.build-android-artifacts == 'false' }}
+      env:
+        PROFILE_TARGET: ${{ inputs.cargo-profile == 'dev' && 'debug' || inputs.cargo-profile }}
       with:
         artifacts-dir: ${{ inputs.artifacts-dir }}
-        target-file: ./target/${{ inputs.cargo-profile }}/greptime
+        target-file: ./target/$PROFILE_TARGET/greptime
         version: ${{ inputs.version }}
         working-dir: ${{ inputs.working-dir }}
 

.github/actions/deploy-greptimedb/action.yml

@@ -0,0 +1,31 @@
+name: Deploy GreptimeDB cluster
+description: Deploy GreptimeDB cluster on Kubernetes
+inputs:
+  aws-ci-test-bucket:
+    description: 'AWS S3 bucket name for testing'
+    required: true
+  aws-region:
+    description: 'AWS region for testing'
+    required: true
+  data-root:
+    description: 'Data root for testing'
+    required: true
+  aws-access-key-id:
+    description: 'AWS access key id for testing'
+    required: true
+  aws-secret-access-key:
+    description: 'AWS secret access key for testing'
+    required: true
+runs:
+  using: composite
+  steps:
+    - name: Deploy GreptimeDB by Helm
+      shell: bash
+      env:
+        DATA_ROOT: ${{ inputs.data-root }}
+        AWS_CI_TEST_BUCKET: ${{ inputs.aws-ci-test-bucket }}
+        AWS_REGION: ${{ inputs.aws-region }}
+        AWS_ACCESS_KEY_ID: ${{ inputs.aws-access-key-id }}
+        AWS_SECRET_ACCESS_KEY: ${{ inputs.aws-secret-access-key }}
+      run: |
+        ./.github/scripts/deploy-greptimedb.sh

.github/actions/publish-github-release/action.yml

@@ -31,10 +31,12 @@ runs:
           echo "prerelease=false" >> $GITHUB_ENV
           echo "makeLatest=true" >> $GITHUB_ENV
           echo "generateReleaseNotes=false" >> $GITHUB_ENV
+          echo "omitBody=true" >> $GITHUB_ENV
         else
           echo "prerelease=true" >> $GITHUB_ENV
           echo "makeLatest=false" >> $GITHUB_ENV
           echo "generateReleaseNotes=true" >> $GITHUB_ENV
+          echo "omitBody=false" >> $GITHUB_ENV
         fi
 
     - name: Publish release
@@ -45,6 +47,7 @@ runs:
         makeLatest: ${{ env.makeLatest }}
         tag: ${{ inputs.version }}
         generateReleaseNotes: ${{ env.generateReleaseNotes }}
+        omitBody: ${{ env.omitBody }} # omitBody is true when the release is a official release.
         allowUpdates: true
         artifacts: |
           **/greptime-*/*

.github/actions/sqlness-test/action.yml

@@ -0,0 +1,59 @@
+name: Run sqlness test
+description: Run sqlness test on GreptimeDB
+
+inputs:
+  aws-ci-test-bucket:
+    description: 'AWS S3 bucket name for testing'
+    required: true
+  aws-region:
+    description: 'AWS region for testing'
+    required: true
+  data-root:
+    description: 'Data root for testing'
+    required: true
+  aws-access-key-id:
+    description: 'AWS access key id for testing'
+    required: true
+  aws-secret-access-key:
+    description: 'AWS secret access key for testing'
+    required: true
+
+runs:
+  using: composite
+  steps:
+    - name: Deploy GreptimeDB cluster by Helm
+      uses: ./.github/actions/deploy-greptimedb
+      with:
+        data-root: ${{ inputs.data-root }}
+        aws-ci-test-bucket: ${{ inputs.aws-ci-test-bucket }}
+        aws-region: ${{ inputs.aws-region }}
+        aws-access-key-id: ${{ inputs.aws-access-key-id }}
+        aws-secret-access-key: ${{ inputs.aws-secret-access-key }}
+
+    # TODO(zyy17): The following tests will be replaced by the real sqlness test.
+    - name: Run tests on greptimedb cluster
+      shell: bash
+      run: |
+        mysql -h 127.0.0.1 -P 14002 -e "CREATE TABLE IF NOT EXISTS system_metrics (host VARCHAR(255), idc VARCHAR(255), cpu_util DOUBLE, memory_util DOUBLE, disk_util DOUBLE, ts TIMESTAMP DEFAULT CURRENT_TIMESTAMP, PRIMARY KEY(host, idc), TIME INDEX(ts));" && \
+        mysql -h 127.0.0.1 -P 14002 -e "SHOW TABLES;"
+
+    - name: Run tests on greptimedb cluster that uses S3
+      shell: bash
+      run: |
+        mysql -h 127.0.0.1 -P 24002 -e "CREATE TABLE IF NOT EXISTS system_metrics (host VARCHAR(255), idc VARCHAR(255), cpu_util DOUBLE, memory_util DOUBLE, disk_util DOUBLE, ts TIMESTAMP DEFAULT CURRENT_TIMESTAMP, PRIMARY KEY(host, idc), TIME INDEX(ts));" && \
+        mysql -h 127.0.0.1 -P 24002 -e "SHOW TABLES;"
+
+    - name: Run tests on standalone greptimedb
+      shell: bash
+      run: |
+        mysql -h 127.0.0.1 -P 34002 -e "CREATE TABLE IF NOT EXISTS system_metrics (host VARCHAR(255), idc VARCHAR(255), cpu_util DOUBLE, memory_util DOUBLE, disk_util DOUBLE, ts TIMESTAMP DEFAULT CURRENT_TIMESTAMP, PRIMARY KEY(host, idc), TIME INDEX(ts));" && \
+        mysql -h 127.0.0.1 -P 34002 -e "SHOW TABLES;"
+
+    - name: Clean S3 data
+      shell: bash
+      env:
+        AWS_DEFAULT_REGION: ${{ inputs.aws-region }}
+        AWS_ACCESS_KEY_ID: ${{ inputs.aws-access-key-id }}
+        AWS_SECRET_ACCESS_KEY: ${{ inputs.aws-secret-access-key }}
+      run: |
+        aws s3 rm s3://${{ inputs.aws-ci-test-bucket }}/${{ inputs.data-root }} --recursive

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

@@ -22,7 +22,7 @@ runs:
       shell: bash
       run: |
         mkdir -p ${{ inputs.artifacts-dir }} && \
-        mv ${{ inputs.target-file }} ${{ inputs.artifacts-dir }}
+        cp ${{ inputs.target-file }} ${{ inputs.artifacts-dir }}
 
     # The compressed artifacts will use the following layout:
     # greptime-linux-amd64-pyo3-v0.3.0sha256sum

.github/doc-label-config.yml

@@ -0,0 +1,4 @@
+Doc not needed:
+    - '- \[x\]  This PR does not require documentation updates.'
+Doc update required:
+    - '- \[ \]  This PR does not require documentation updates.'

.github/pull_request_template.md

@@ -15,5 +15,6 @@ Please explain IN DETAIL what the changes are in this PR and why they are needed
 
 - [ ]  I have written the necessary rustdoc comments.
 - [ ]  I have added the necessary unit tests and integration tests.
+- [ ]  This PR does not require documentation updates.
 
 ## Refer to a related PR or issue link (optional)

.github/scripts/deploy-greptimedb.sh

@@ -0,0 +1,169 @@
+#!/usr/bin/env bash
+
+set -e
+set -o pipefail
+
+KUBERNETES_VERSION="${KUBERNETES_VERSION:-v1.24.0}"
+ENABLE_STANDALONE_MODE="${ENABLE_STANDALONE_MODE:-true}"
+DEFAULT_INSTALL_NAMESPACE=${DEFAULT_INSTALL_NAMESPACE:-default}
+GREPTIMEDB_IMAGE_TAG=${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.
+function create_kind_cluster() {
+  cat <<EOF | kind create cluster --name "${CLUSTER}" --image kindest/node:"$KUBERNETES_VERSION" --config=-
+kind: Cluster
+apiVersion: kind.x-k8s.io/v1alpha4
+nodes:
+- role: control-plane
+- role: worker
+- role: worker
+- role: worker
+- role: worker
+- role: worker
+EOF
+}
+
+# Add greptime Helm chart repo.
+function add_greptime_chart() {
+  helm repo add greptime "$GREPTIME_CHART"
+  helm repo update
+}
+
+# Deploy a etcd cluster with 3 members.
+function deploy_etcd_cluster() {
+  local namespace="$1"
+
+  helm install etcd "$ETCD_CHART" \
+    --set replicaCount=3 \
+    --set auth.rbac.create=false \
+    --set auth.rbac.token.enabled=false \
+    -n "$namespace"
+
+  # Wait for etcd cluster to be ready.
+  kubectl rollout status statefulset/etcd -n "$namespace"
+}
+
+# Deploy greptimedb-operator.
+function deploy_greptimedb_operator() {
+  # Use the latest chart and image.
+  helm install greptimedb-operator greptime/greptimedb-operator \
+    --set image.tag=latest \
+    -n "$DEFAULT_INSTALL_NAMESPACE"
+
+  # Wait for greptimedb-operator to be ready.
+  kubectl rollout status deployment/greptimedb-operator -n "$DEFAULT_INSTALL_NAMESPACE"
+}
+
+# Deploy greptimedb cluster by using local storage.
+# It will expose cluster service ports as '14000', '14001', '14002', '14003' to local access.
+function deploy_greptimedb_cluster() {
+  local cluster_name=$1
+  local install_namespace=$2
+
+  kubectl create ns "$install_namespace"
+
+  deploy_etcd_cluster "$install_namespace"
+
+  helm install "$cluster_name" greptime/greptimedb-cluster \
+    --set image.tag="$GREPTIMEDB_IMAGE_TAG" \
+    --set meta.etcdEndpoints="etcd.$install_namespace:2379" \
+    -n "$install_namespace"
+
+  # Wait for greptimedb cluster to be ready.
+  while true; do
+    PHASE=$(kubectl -n "$install_namespace" get gtc "$cluster_name" -o jsonpath='{.status.clusterPhase}')
+    if [ "$PHASE" == "Running" ]; then
+      echo "Cluster is ready"
+      break
+    else
+      echo "Cluster is not ready yet: Current phase: $PHASE"
+      sleep 5 # wait for 5 seconds before check again.
+    fi
+  done
+
+  # Expose greptimedb cluster to local access.
+  kubectl -n "$install_namespace" port-forward svc/"$cluster_name"-frontend \
+    14000:4000 \
+    14001:4001 \
+    14002:4002 \
+    14003:4003 > /tmp/connections.out &
+}
+
+# Deploy greptimedb cluster by using S3.
+# It will expose cluster service ports as '24000', '24001', '24002', '24003' to local access.
+function deploy_greptimedb_cluster_with_s3_storage() {
+  local cluster_name=$1
+  local install_namespace=$2
+
+  kubectl create ns "$install_namespace"
+
+  deploy_etcd_cluster "$install_namespace"
+
+  helm install "$cluster_name" greptime/greptimedb-cluster -n "$install_namespace" \
+    --set image.tag="$GREPTIMEDB_IMAGE_TAG" \
+    --set meta.etcdEndpoints="etcd.$install_namespace:2379" \
+    --set storage.s3.bucket="$AWS_CI_TEST_BUCKET" \
+    --set storage.s3.region="$AWS_REGION" \
+    --set storage.s3.root="$DATA_ROOT" \
+    --set storage.credentials.secretName=s3-credentials \
+    --set storage.credentials.accessKeyId="$AWS_ACCESS_KEY_ID" \
+    --set storage.credentials.secretAccessKey="$AWS_SECRET_ACCESS_KEY"
+
+  # Wait for greptimedb cluster to be ready.
+  while true; do
+    PHASE=$(kubectl -n "$install_namespace" get gtc "$cluster_name" -o jsonpath='{.status.clusterPhase}')
+    if [ "$PHASE" == "Running" ]; then
+      echo "Cluster is ready"
+      break
+    else
+      echo "Cluster is not ready yet: Current phase: $PHASE"
+      sleep 5 # wait for 5 seconds before check again.
+    fi
+  done
+
+  # Expose greptimedb cluster to local access.
+  kubectl -n "$install_namespace" port-forward svc/"$cluster_name"-frontend \
+    24000:4000 \
+    24001:4001 \
+    24002:4002 \
+    24003:4003 > /tmp/connections.out &
+}
+
+# Deploy standalone greptimedb.
+# It will expose cluster service ports as '34000', '34001', '34002', '34003' to local access.
+function deploy_standalone_greptimedb() {
+  helm install greptimedb-standalone greptime/greptimedb-standalone \
+    --set image.tag="$GREPTIMEDB_IMAGE_TAG" \
+    -n "$DEFAULT_INSTALL_NAMESPACE"
+
+  # Wait for etcd cluster to be ready.
+  kubectl rollout status statefulset/greptimedb-standalone -n "$DEFAULT_INSTALL_NAMESPACE"
+
+  # Expose greptimedb to local access.
+  kubectl -n "$DEFAULT_INSTALL_NAMESPACE" port-forward svc/greptimedb-standalone \
+    34000:4000 \
+    34001:4001 \
+    34002:4002 \
+    34003:4003 > /tmp/connections.out &
+}
+
+# Entrypoint of the script.
+function main() {
+  create_kind_cluster
+  add_greptime_chart
+
+  # Deploy standalone greptimedb in the same K8s.
+  if [ "$ENABLE_STANDALONE_MODE" == "true" ]; then
+    deploy_standalone_greptimedb
+  fi
+
+  deploy_greptimedb_operator
+  deploy_greptimedb_cluster testcluster testcluster
+  deploy_greptimedb_cluster_with_s3_storage testcluster-s3 testcluster-s3
+}
+
+# Usages:
+# - Deploy greptimedb cluster: ./deploy-greptimedb.sh
+main

.github/workflows/apidoc.yml

@@ -13,7 +13,7 @@ on:
 name: Build API docs
 
 env:
-  RUST_TOOLCHAIN: nightly-2023-08-07
+  RUST_TOOLCHAIN: nightly-2023-12-19
 
 jobs:
   apidoc:

.github/workflows/dev-build.yml

@@ -55,10 +55,18 @@ on:
         description: Build and push images to DockerHub and ACR
         required: false
         default: true
+      cargo_profile:
+        type: choice
+        description: The cargo profile to use in building GreptimeDB.
+        default: nightly
+        options:
+          - dev
+          - release
+          - nightly
 
 # Use env variables to control all the release process.
 env:
-  CARGO_PROFILE: nightly
+  CARGO_PROFILE: ${{ inputs.cargo_profile }}
 
   # Controls whether to run tests, include unit-test, integration-test and sqlness.
   DISABLE_RUN_TESTS: ${{ inputs.skip_test || vars.DEFAULT_SKIP_TEST }}

.github/workflows/develop.yml

@@ -29,7 +29,7 @@ concurrency:
   cancel-in-progress: true
 
 env:
-  RUST_TOOLCHAIN: nightly-2023-08-07
+  RUST_TOOLCHAIN: nightly-2023-12-19
 
 jobs:
   typos:
@@ -42,7 +42,10 @@ jobs:
   check:
     name: Check
     if: github.event.pull_request.draft == false
-    runs-on: ubuntu-20.04
+    runs-on: ${{ matrix.os }}
+    strategy:
+      matrix:
+        os: [ windows-latest-8-cores, ubuntu-20.04 ]
     timeout-minutes: 60
     steps:
       - uses: actions/checkout@v3
@@ -161,15 +164,21 @@ jobs:
         uses: Swatinem/rust-cache@v2
       - name: Install latest nextest release
         uses: taiki-e/install-action@nextest
+      - name: Install cargo-llvm-cov
+        uses: taiki-e/install-action@cargo-llvm-cov
       - name: Install Python
         uses: actions/setup-python@v4
         with:
           python-version: '3.10'
       - name: Install PyArrow Package
         run: pip install pyarrow
-      - name: Install cargo-llvm-cov
-        uses: taiki-e/install-action@cargo-llvm-cov
-      - name: Collect coverage data
+      - name: Setup etcd server
+        working-directory: tests-integration/fixtures/etcd
+        run: docker compose -f docker-compose-standalone.yml up -d --wait
+      - name: Setup kafka server
+        working-directory: tests-integration/fixtures/kafka
+        run: docker compose -f docker-compose-standalone.yml up -d --wait
+      - name: Run nextest cases
         run: cargo llvm-cov nextest --workspace --lcov --output-path lcov.info -F pyo3_backend -F dashboard
         env:
           CARGO_BUILD_RUSTFLAGS: "-C link-arg=-fuse-ld=lld"
@@ -179,6 +188,8 @@ jobs:
           GT_S3_ACCESS_KEY_ID: ${{ secrets.S3_ACCESS_KEY_ID }}
           GT_S3_ACCESS_KEY: ${{ secrets.S3_ACCESS_KEY }}
           GT_S3_REGION: ${{ secrets.S3_REGION }}
+          GT_ETCD_ENDPOINTS: http://127.0.0.1:2379
+          GT_KAFKA_ENDPOINTS: 127.0.0.1:9092
           UNITTEST_LOG_DIR: "__unittest_logs"
       - name: Codecov upload
         uses: codecov/codecov-action@v2

.github/workflows/doc-label.yml

@@ -0,0 +1,20 @@
+name: "PR Doc Labeler"
+on:
+  pull_request_target:
+    types: [opened, edited, synchronize, ready_for_review, auto_merge_enabled, labeled, unlabeled]
+
+permissions:
+  pull-requests: write
+  contents: read
+
+jobs:
+  triage:
+    if: ${{ github.repository == 'GreptimeTeam/greptimedb' }}
+    runs-on: ubuntu-latest
+    steps:
+    - uses: github/issue-labeler@v3.3
+      with:
+        configuration-path: .github/doc-label-config.yml
+        enable-versioned-regex: false
+        repo-token: ${{ secrets.GITHUB_TOKEN }}
+        sync-labels: 1

.github/workflows/license.yaml

@@ -13,4 +13,4 @@ jobs:
     steps:
     - uses: actions/checkout@v2
     - name: Check License Header
-      uses: apache/skywalking-eyes/header@df70871af1a8109c9a5b1dc824faaf65246c5236
+      uses: korandoru/hawkeye@v3

.github/workflows/nightly-ci.yml

@@ -12,11 +12,12 @@ concurrency:
   cancel-in-progress: true
 
 env:
-  RUST_TOOLCHAIN: nightly-2023-08-07
+  RUST_TOOLCHAIN: nightly-2023-12-19
 
 jobs:
   sqlness:
     name: Sqlness Test
+    if: ${{ github.repository == 'GreptimeTeam/greptimedb' }}
     runs-on: ${{ matrix.os }}
     strategy:
       matrix:
@@ -34,6 +35,14 @@ jobs:
         uses: Swatinem/rust-cache@v2
       - name: Run sqlness
         run: cargo sqlness
+      - name: Notify slack if failed
+        if: failure()
+        uses: slackapi/slack-github-action@v1.23.0
+        env:
+          SLACK_WEBHOOK_URL: ${{ secrets.SLACK_WEBHOOK_URL_DEVELOP_CHANNEL }}
+        with:
+          payload: |
+            {"text": "Nightly CI failed for sqlness tests"}
       - name: Upload sqlness logs
         if: always()
         uses: actions/upload-artifact@v3
@@ -43,6 +52,7 @@ jobs:
           retention-days: 3
 
   test-on-windows:
+    if: ${{ github.repository == 'GreptimeTeam/greptimedb' }}
     runs-on: windows-latest-8-cores
     timeout-minutes: 60
     steps:
@@ -80,3 +90,11 @@ jobs:
           GT_S3_ACCESS_KEY: ${{ secrets.S3_ACCESS_KEY }}
           GT_S3_REGION: ${{ secrets.S3_REGION }}
           UNITTEST_LOG_DIR: "__unittest_logs"
+      - name: Notify slack if failed
+        if: failure()
+        uses: slackapi/slack-github-action@v1.23.0
+        env:
+          SLACK_WEBHOOK_URL: ${{ secrets.SLACK_WEBHOOK_URL_DEVELOP_CHANNEL }}
+        with:
+          payload: |
+            {"text": "Nightly CI failed for cargo test"}

.github/workflows/nightly-funtional-tests.yml

@@ -0,0 +1,27 @@
+name: Nightly functional tests
+
+on:
+  schedule:
+    # At 00:00 on Tuesday.
+    - cron: '0 0 * * 2'
+  workflow_dispatch:
+
+jobs:
+  sqlness-test:
+    name: Run sqlness test
+    if: ${{ github.repository == 'GreptimeTeam/greptimedb' }}
+    runs-on: ubuntu-22.04
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v3
+        with:
+          fetch-depth: 0
+
+      - name: Run sqlness test
+        uses: ./.github/actions/sqlness-test
+        with:
+          data-root: sqlness-test
+          aws-ci-test-bucket: ${{ vars.AWS_CI_TEST_BUCKET }}
+          aws-region: ${{ vars.AWS_CI_TEST_BUCKET_REGION }}
+          aws-access-key-id: ${{ secrets.AWS_CI_TEST_ACCESS_KEY_ID }}
+          aws-secret-access-key: ${{ secrets.AWS_CI_TEST_SECRET_ACCESS_KEY }}

.github/workflows/release.yml

@@ -82,7 +82,7 @@ on:
 # Use env variables to control all the release process.
 env:
   # The arguments of building greptime.
-  RUST_TOOLCHAIN: nightly-2023-08-07
+  RUST_TOOLCHAIN: nightly-2023-12-19
   CARGO_PROFILE: nightly
 
   # Controls whether to run tests, include unit-test, integration-test and sqlness.
@@ -91,7 +91,7 @@ env:
   # The scheduled version is '${{ env.NEXT_RELEASE_VERSION }}-nightly-YYYYMMDD', like v0.2.0-nigthly-20230313;
   NIGHTLY_RELEASE_PREFIX: nightly
   # Note: The NEXT_RELEASE_VERSION should be modified manually by every formal release.
-  NEXT_RELEASE_VERSION: v0.5.0
+  NEXT_RELEASE_VERSION: v0.6.0
 
 jobs:
   allocate-runners:
@@ -302,8 +302,12 @@ jobs:
   release-cn-artifacts:
     name: Release artifacts to CN region
     if: ${{ inputs.release_images || github.event_name == 'push' || github.event_name == 'schedule' }}
-    needs: [
+    needs: [ # The job have to wait for all the artifacts are built.
       allocate-runners,
+      build-linux-amd64-artifacts,
+      build-linux-arm64-artifacts,
+      build-macos-artifacts,
+      build-windows-artifacts,
       release-images-to-dockerhub,
     ]
     runs-on: ubuntu-20.04
@@ -338,11 +342,12 @@ jobs:
   publish-github-release:
     name: Create GitHub release and upload artifacts
     if: ${{ inputs.publish_github_release || github.event_name == 'push' || github.event_name == 'schedule' }}
-    needs: [
+    needs: [ # The job have to wait for all the artifacts are built.
       allocate-runners,
       build-linux-amd64-artifacts,
       build-linux-arm64-artifacts,
       build-macos-artifacts,
+      build-windows-artifacts,
       release-images-to-dockerhub,
     ]
     runs-on: ubuntu-20.04

.github/workflows/size-label.yml

@@ -0,0 +1,25 @@
+name: size-labeler
+
+on: [pull_request_target]
+
+jobs:
+  labeler:
+    runs-on: ubuntu-latest
+    name: Label the PR size
+    permissions:
+      issues: write
+      pull-requests: write
+    steps:
+      - uses: codelytv/pr-size-labeler@v1
+        with:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          s_label: 'Size: S'
+          s_max_size: '100'
+          m_label: 'Size: M'
+          m_max_size: '500'
+          l_label: 'Size: L'
+          l_max_size: '1000'
+          xl_label: 'Size: XL'
+          fail_if_xl: 'false'
+          message_if_xl: ""
+          files_to_ignore: 'Cargo.lock'

.github/workflows/user-doc-label-checker.yml

@@ -0,0 +1,19 @@
+name: Check user doc labels
+on:
+  pull_request:
+    types:
+      - opened
+      - reopened
+      - labeled
+      - unlabeled
+
+jobs:
+
+  check_labels:
+    name: Check doc labels
+    runs-on: ubuntu-latest
+    steps:
+      - uses: docker://agilepathway/pull-request-label-checker:latest
+        with:
+          one_of: Doc update required,Doc not needed
+          repo_token: ${{ secrets.GITHUB_TOKEN }}

.licenserc.yaml

@@ -1,14 +0,0 @@
-header:
-  license:
-    spdx-id: Apache-2.0
-    copyright-owner: Greptime Team
-
-  paths:
-    - "**/*.rs"
-    - "**/*.py"
-
-  comment: on-failure
-
-dependency:
-  files:
-    - Cargo.toml

CONTRIBUTING.md

@@ -2,7 +2,7 @@
 
 Thanks a lot for considering contributing to GreptimeDB. We believe people like you would make GreptimeDB a great product. We intend to build a community where individuals can have open talks, show respect for one another, and speak with true ❤️. Meanwhile, we are to keep transparency and make your effort count here.
 
-Read the guidelines, and they can help you get started. Communicate with respect to developers maintaining and developing the project. In return, they should reciprocate that respect by addressing your issue, reviewing changes, as well as helping finalize and merge your pull requests.
+Please read the guidelines, and they can help you get started. Communicate with respect to developers maintaining and developing the project. In return, they should reciprocate that respect by addressing your issue, reviewing changes, as well as helping finalize and merge your pull requests.
 
 Follow our [README](https://github.com/GreptimeTeam/greptimedb#readme) to get the whole picture of the project. To learn about the design of GreptimeDB, please refer to the [design docs](https://github.com/GrepTimeTeam/docs).
 
@@ -21,7 +21,7 @@ Pull requests are great, but we accept all kinds of other help if you like. Such
 - Write tutorials or blog posts. Blog, speak about, or create tutorials about one of GreptimeDB's many features. Mention [@greptime](https://twitter.com/greptime) on Twitter and email info@greptime.com so we can give pointers and tips and help you spread the word by promoting your content on Greptime communication channels.
 - Improve the documentation. [Submit documentation](http://github.com/greptimeTeam/docs/) updates, enhancements, designs, or bug fixes, and fixing any spelling or grammar errors will be very much appreciated.
 - Present at meetups and conferences about your GreptimeDB projects. Your unique challenges and successes in building things with GreptimeDB can provide great speaking material. We'd love to review your talk abstract, so get in touch with us if you'd like some help!
-- Submit bug reports. To report a bug or a security issue, you can [open a new GitHub issue](https://github.com/GrepTimeTeam/greptimedb/issues/new).
+- Submitting bug reports. To report a bug or a security issue, you can [open a new GitHub issue](https://github.com/GrepTimeTeam/greptimedb/issues/new).
 - Speak up feature requests. Send feedback is a great way for us to understand your different use cases of GreptimeDB better. If you want to share your experience with GreptimeDB, or if you want to discuss any ideas, you can start a discussion on [GitHub discussions](https://github.com/GreptimeTeam/greptimedb/discussions), chat with the Greptime team on [Slack](https://greptime.com/slack), or you can tweet [@greptime](https://twitter.com/greptime) on Twitter.
 
 ## Code of Conduct
@@ -49,6 +49,7 @@ GreptimeDB uses the [Apache 2.0 license](https://github.com/GreptimeTeam/greptim
 ### Before PR
 
 - To ensure that community is free and confident in its ability to use your contributions, please sign the Contributor License Agreement (CLA) which will be incorporated in the pull request process.
+- Make sure all files have proper license header (running `docker run --rm -v $(pwd):/github/workspace ghcr.io/korandoru/hawkeye-native:v3 format` from the project root).
 - Make sure all your codes are formatted and follow the [coding style](https://pingcap.github.io/style-guide/rust/).
 - Make sure all unit tests are passed (using `cargo test --workspace` or [nextest](https://nexte.st/index.html) `cargo nextest run`).
 - Make sure all clippy warnings are fixed (you can check it locally by running `cargo clippy --workspace --all-targets -- -D warnings`).
@@ -81,7 +82,7 @@ Now, `pre-commit` will run automatically on `git commit`.
 ### Title
 
 The titles of pull requests should be prefixed with category names listed in [Conventional Commits specification](https://www.conventionalcommits.org/en/v1.0.0)
-like `feat`/`fix`/`docs`, with a concise summary of code change following. DO NOT use last commit message as pull request title.
+like `feat`/`fix`/`docs`, with a concise summary of code change following. AVOID using the last commit message as pull request title.
 
 ### Description
 
@@ -100,7 +101,7 @@ of what you were trying to do and what went wrong. You can also reach for help i
 
 ## Community
 
-The core team will be thrilled if you participate in any way you like. When you are stuck, try 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:
+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)

Cargo.toml

@@ -27,6 +27,7 @@ members = [
     "src/common/telemetry",
     "src/common/test-util",
     "src/common/time",
+    "src/common/decimal",
     "src/common/version",
     "src/datanode",
     "src/datatypes",
@@ -35,84 +36,106 @@ members = [
     "src/log-store",
     "src/meta-client",
     "src/meta-srv",
+    "src/metric-engine",
     "src/mito2",
     "src/object-store",
     "src/operator",
     "src/partition",
     "src/plugins",
     "src/promql",
+    "src/puffin",
     "src/query",
     "src/script",
     "src/servers",
     "src/session",
     "src/sql",
-    "src/storage",
     "src/store-api",
     "src/table",
+    "src/index",
     "tests-integration",
     "tests/runner",
 ]
 resolver = "2"
 
 [workspace.package]
-version = "0.4.1"
+version = "0.5.0"
 edition = "2021"
 license = "Apache-2.0"
 
 [workspace.dependencies]
+ahash = { version = "0.8", features = ["compile-time-rng"] }
 aquamarine = "0.3"
-arrow = { version = "43.0" }
-arrow-array = "43.0"
-arrow-flight = "43.0"
-arrow-schema = { version = "43.0", features = ["serde"] }
+arrow = { version = "47.0" }
+arrow-array = "47.0"
+arrow-flight = "47.0"
+arrow-schema = { version = "47.0", features = ["serde"] }
 async-stream = "0.3"
 async-trait = "0.1"
+base64 = "0.21"
+bigdecimal = "0.4.2"
+bitflags = "2.4.1"
+bytemuck = "1.12"
+bytes = { version = "1.5", features = ["serde"] }
 chrono = { version = "0.4", features = ["serde"] }
-datafusion = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "b6f3b28b6fe91924cc8dd3d83726b766f2a706ec" }
-datafusion-common = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "b6f3b28b6fe91924cc8dd3d83726b766f2a706ec" }
-datafusion-expr = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "b6f3b28b6fe91924cc8dd3d83726b766f2a706ec" }
-datafusion-optimizer = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "b6f3b28b6fe91924cc8dd3d83726b766f2a706ec" }
-datafusion-physical-expr = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "b6f3b28b6fe91924cc8dd3d83726b766f2a706ec" }
-datafusion-sql = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "b6f3b28b6fe91924cc8dd3d83726b766f2a706ec" }
-datafusion-substrait = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "b6f3b28b6fe91924cc8dd3d83726b766f2a706ec" }
+dashmap = "5.4"
+datafusion = { git = "https://github.com/apache/arrow-datafusion.git", rev = "26e43acac3a96cec8dd4c8365f22dfb1a84306e9" }
+datafusion-common = { git = "https://github.com/apache/arrow-datafusion.git", rev = "26e43acac3a96cec8dd4c8365f22dfb1a84306e9" }
+datafusion-expr = { git = "https://github.com/apache/arrow-datafusion.git", rev = "26e43acac3a96cec8dd4c8365f22dfb1a84306e9" }
+datafusion-optimizer = { git = "https://github.com/apache/arrow-datafusion.git", rev = "26e43acac3a96cec8dd4c8365f22dfb1a84306e9" }
+datafusion-physical-expr = { git = "https://github.com/apache/arrow-datafusion.git", rev = "26e43acac3a96cec8dd4c8365f22dfb1a84306e9" }
+datafusion-sql = { git = "https://github.com/apache/arrow-datafusion.git", rev = "26e43acac3a96cec8dd4c8365f22dfb1a84306e9" }
+datafusion-substrait = { git = "https://github.com/apache/arrow-datafusion.git", rev = "26e43acac3a96cec8dd4c8365f22dfb1a84306e9" }
 derive_builder = "0.12"
-etcd-client = "0.11"
+etcd-client = "0.12"
+fst = "0.4.7"
 futures = "0.3"
 futures-util = "0.3"
-greptime-proto = { git = "https://github.com/GreptimeTeam/greptime-proto.git", rev = "1f1dd532a111e3834cc3019c5605e2993ffb9dc3" }
+greptime-proto = { git = "https://github.com/GreptimeTeam/greptime-proto.git", rev = "a31ea166fc015ea7ff111ac94e26c3a5d64364d2" }
 humantime-serde = "1.1"
 itertools = "0.10"
 lazy_static = "1.4"
 meter-core = { git = "https://github.com/GreptimeTeam/greptime-meter.git", rev = "abbd357c1e193cd270ea65ee7652334a150b628f" }
-metrics = "0.20"
+mockall = "0.11.4"
 moka = "0.12"
 once_cell = "1.18"
-opentelemetry-proto = { version = "0.2", features = ["gen-tonic", "metrics"] }
-parquet = "43.0"
+opentelemetry-proto = { git = "https://github.com/waynexia/opentelemetry-rust.git", rev = "33841b38dda79b15f2024952be5f32533325ca02", features = [
+    "gen-tonic",
+    "metrics",
+    "trace",
+] }
+parquet = "47.0"
 paste = "1.0"
-prost = "0.11"
+pin-project = "1.0"
+prometheus = { version = "0.13.3", features = ["process"] }
+prost = "0.12"
 raft-engine = { git = "https://github.com/tikv/raft-engine.git", rev = "22dfb426cd994602b57725ef080287d3e53db479" }
 rand = "0.8"
 regex = "1.8"
+regex-automata = { version = "0.1", features = ["transducer"] }
 reqwest = { version = "0.11", default-features = false, features = [
     "json",
     "rustls-tls-native-roots",
     "stream",
 ] }
+rskafka = "0.5"
+rust_decimal = "1.33"
 serde = { version = "1.0", features = ["derive"] }
 serde_json = "1.0"
 smallvec = "1"
-snafu = { version = "0.7", features = ["backtraces"] }
-sqlparser = { git = "https://github.com/GreptimeTeam/sqlparser-rs.git", rev = "6cf9d23d5b8fbecd65efc1d9afb7e80ad7a424da", features = [
+snafu = "0.7"
+# on branch v0.38.x
+sqlparser = { git = "https://github.com/GreptimeTeam/sqlparser-rs.git", rev = "6a93567ae38d42be5c8d08b13c8ff4dde26502ef", features = [
     "visitor",
 ] }
 strum = { version = "0.25", features = ["derive"] }
 tempfile = "3"
 tokio = { version = "1.28", features = ["full"] }
+tokio-stream = { version = "0.1" }
 tokio-util = { version = "0.7", features = ["io-util", "compat"] }
-toml = "0.7"
-tonic = { version = "0.9", features = ["tls"] }
+toml = "0.8.8"
+tonic = { version = "0.10", features = ["tls"] }
 uuid = { version = "1", features = ["serde", "v4", "fast-rng"] }
+
 ## workspaces members
 api = { path = "src/api" }
 auth = { path = "src/auth" }
@@ -123,6 +146,7 @@ common-base = { path = "src/common/base" }
 common-catalog = { path = "src/common/catalog" }
 common-config = { path = "src/common/config" }
 common-datasource = { path = "src/common/datasource" }
+common-decimal = { path = "src/common/decimal" }
 common-error = { path = "src/common/error" }
 common-function = { path = "src/common/function" }
 common-greptimedb-telemetry = { path = "src/common/greptimedb-telemetry" }
@@ -148,7 +172,7 @@ frontend = { path = "src/frontend" }
 log-store = { path = "src/log-store" }
 meta-client = { path = "src/meta-client" }
 meta-srv = { path = "src/meta-srv" }
-mito = { path = "src/mito" }
+metric-engine = { path = "src/metric-engine" }
 mito2 = { path = "src/mito2" }
 object-store = { path = "src/object-store" }
 operator = { path = "src/operator" }
@@ -160,7 +184,6 @@ script = { path = "src/script" }
 servers = { path = "src/servers" }
 session = { path = "src/session" }
 sql = { path = "src/sql" }
-storage = { path = "src/storage" }
 store-api = { path = "src/store-api" }
 substrait = { path = "src/common/substrait" }
 table = { path = "src/table" }
@@ -170,7 +193,7 @@ git = "https://github.com/GreptimeTeam/greptime-meter.git"
 rev = "abbd357c1e193cd270ea65ee7652334a150b628f"
 
 [profile.release]
-debug = true
+debug = 1
 
 [profile.nightly]
 inherits = "release"

LICENSE

@@ -186,7 +186,7 @@
       same "printed page" as the copyright notice for easier
       identification within third-party archives.
 
-   Copyright 2022 Greptime Team
+   Copyright [yyyy] [name of copyright owner]
 
    Licensed under the Apache License, Version 2.0 (the "License");
    you may not use this file except in compliance with the License.

Makefile

@@ -157,11 +157,11 @@ sqlness-test: ## Run sqlness test.
 
 .PHONY: check
 check: ## Cargo check all the targets.
-	cargo check --workspace --all-targets
+	cargo check --workspace --all-targets --all-features
 
 .PHONY: clippy
 clippy: ## Check clippy rules.
-	cargo clippy --workspace --all-targets -F pyo3_backend -- -D warnings
+	cargo clippy --workspace --all-targets --all-features -- -D warnings
 
 .PHONY: fmt-check
 fmt-check: ## Check code format.

README.md

@@ -27,13 +27,8 @@
     <a href="https://greptime.com/slack"><img src="https://img.shields.io/badge/slack-GreptimeDB-0abd59?logo=slack" alt="slack" /></a>
 </p>
 
-## Upcoming Event
-Come and meet us in **KubeCon + CloudNativeCon North America 2023!**
-<p align="center">
-  <picture>
-    <img alt="KubeCon + CloudNativeCon North Logo" src="./docs/banner/KCCNC_NA_2023_1000x200_Email Banner.png" width="800px">
-  </picture>
-</p>
+> [!WARNING]  
+> Our default branch has changed from `develop` to `main` (issue [#3025](https://github.com/GreptimeTeam/greptimedb/issues/3025)). Please update your local repository to use the `main` branch.
 
 ## What is GreptimeDB
 
@@ -108,7 +103,7 @@ Please see the online document site for more installation options and [operation
 
 ### Get started
 
-Read the [complete getting started guide](https://docs.greptime.com/getting-started/try-out-greptimedb) on our [official document site](https://docs.greptime.com/).
+Read the [complete getting started guide](https://docs.greptime.com/getting-started/overview) on our [official document site](https://docs.greptime.com/).
 
 To write and query data, GreptimeDB is compatible with multiple [protocols and clients](https://docs.greptime.com/user-guide/clients/overview).
 
@@ -117,7 +112,7 @@ To write and query data, GreptimeDB is compatible with multiple [protocols and c
 ### Installation
 
 - [Pre-built Binaries](https://greptime.com/download):
-  For Linux and macOS, you can easily download pre-built binaries including official releases and nightly builds that are ready to use. 
+  For Linux and macOS, you can easily download pre-built binaries including official releases and nightly builds that are ready to use.
   In most cases, downloading the version without PyO3 is sufficient. However, if you plan to run scripts in CPython (and use Python packages like NumPy and Pandas), you will need to download the version with PyO3 and install a Python with the same version as the Python in the PyO3 version.
   We recommend using virtualenv for the installation process to manage multiple Python versions.
 - [Docker Images](https://hub.docker.com/r/greptime/greptimedb)(**recommended**): pre-built
@@ -143,6 +138,7 @@ To write and query data, GreptimeDB is compatible with multiple [protocols and c
 - [GreptimeDB Java Client](https://github.com/GreptimeTeam/greptimedb-client-java)
 - [GreptimeDB Python Client](https://github.com/GreptimeTeam/greptimedb-client-py) (WIP)
 - [GreptimeDB Rust Client](https://github.com/GreptimeTeam/greptimedb-client-rust)
+- [GreptimeDB JavaScript Client](https://github.com/GreptimeTeam/greptime-js-sdk)
 
 ## Project Status
 
@@ -184,6 +180,6 @@ Please refer to [contribution guidelines](CONTRIBUTING.md) for more information.
 ## Acknowledgement
 - 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://github.com/apache/arrow-datafusion).
-- [OpenDAL](https://github.com/datafuselabs/opendal) from [Datafuse Labs](https://github.com/datafuselabs) gives GreptimeDB a very general and elegant data access abstraction layer.
-- GreptimeDB’s meta service is based on [etcd](https://etcd.io/).
+- [Apache OpenDAL (incubating)](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/).
 - GreptimeDB uses [RustPython](https://github.com/RustPython/RustPython) for experimental embedded python scripting.

benchmarks/Cargo.toml

@@ -8,7 +8,7 @@ license.workspace = true
 arrow.workspace = true
 chrono.workspace = true
 clap = { version = "4.0", features = ["derive"] }
-client = { workspace = true }
+client.workspace = true
 futures-util.workspace = true
 indicatif = "0.17.1"
 itertools.workspace = true

benchmarks/src/bin/nyc-taxi.rs

@@ -152,6 +152,7 @@ fn convert_record_batch(record_batch: RecordBatch) -> (Vec<Column>, u32) {
                 .unwrap_or_default(),
             datatype: datatype.into(),
             semantic_type: semantic_type as i32,
+            ..Default::default()
         };
         columns.push(column);
     }
@@ -266,6 +267,7 @@ fn create_table_expr(table_name: &str) -> CreateTableExpr {
                 default_constraint: vec![],
                 semantic_type: SemanticType::Tag as i32,
                 comment: String::new(),
+                ..Default::default()
             },
             ColumnDef {
                 name: "tpep_pickup_datetime".to_string(),
@@ -274,6 +276,7 @@ fn create_table_expr(table_name: &str) -> CreateTableExpr {
                 default_constraint: vec![],
                 semantic_type: SemanticType::Timestamp as i32,
                 comment: String::new(),
+                ..Default::default()
             },
             ColumnDef {
                 name: "tpep_dropoff_datetime".to_string(),
@@ -282,6 +285,7 @@ fn create_table_expr(table_name: &str) -> CreateTableExpr {
                 default_constraint: vec![],
                 semantic_type: SemanticType::Field as i32,
                 comment: String::new(),
+                ..Default::default()
             },
             ColumnDef {
                 name: "passenger_count".to_string(),
@@ -290,6 +294,7 @@ fn create_table_expr(table_name: &str) -> CreateTableExpr {
                 default_constraint: vec![],
                 semantic_type: SemanticType::Field as i32,
                 comment: String::new(),
+                ..Default::default()
             },
             ColumnDef {
                 name: "trip_distance".to_string(),
@@ -298,6 +303,7 @@ fn create_table_expr(table_name: &str) -> CreateTableExpr {
                 default_constraint: vec![],
                 semantic_type: SemanticType::Field as i32,
                 comment: String::new(),
+                ..Default::default()
             },
             ColumnDef {
                 name: "RatecodeID".to_string(),
@@ -306,6 +312,7 @@ fn create_table_expr(table_name: &str) -> CreateTableExpr {
                 default_constraint: vec![],
                 semantic_type: SemanticType::Field as i32,
                 comment: String::new(),
+                ..Default::default()
             },
             ColumnDef {
                 name: "store_and_fwd_flag".to_string(),
@@ -314,6 +321,7 @@ fn create_table_expr(table_name: &str) -> CreateTableExpr {
                 default_constraint: vec![],
                 semantic_type: SemanticType::Field as i32,
                 comment: String::new(),
+                ..Default::default()
             },
             ColumnDef {
                 name: "PULocationID".to_string(),
@@ -322,6 +330,7 @@ fn create_table_expr(table_name: &str) -> CreateTableExpr {
                 default_constraint: vec![],
                 semantic_type: SemanticType::Field as i32,
                 comment: String::new(),
+                ..Default::default()
             },
             ColumnDef {
                 name: "DOLocationID".to_string(),
@@ -330,6 +339,7 @@ fn create_table_expr(table_name: &str) -> CreateTableExpr {
                 default_constraint: vec![],
                 semantic_type: SemanticType::Field as i32,
                 comment: String::new(),
+                ..Default::default()
             },
             ColumnDef {
                 name: "payment_type".to_string(),
@@ -338,6 +348,7 @@ fn create_table_expr(table_name: &str) -> CreateTableExpr {
                 default_constraint: vec![],
                 semantic_type: SemanticType::Field as i32,
                 comment: String::new(),
+                ..Default::default()
             },
             ColumnDef {
                 name: "fare_amount".to_string(),
@@ -346,6 +357,7 @@ fn create_table_expr(table_name: &str) -> CreateTableExpr {
                 default_constraint: vec![],
                 semantic_type: SemanticType::Field as i32,
                 comment: String::new(),
+                ..Default::default()
             },
             ColumnDef {
                 name: "extra".to_string(),
@@ -354,6 +366,7 @@ fn create_table_expr(table_name: &str) -> CreateTableExpr {
                 default_constraint: vec![],
                 semantic_type: SemanticType::Field as i32,
                 comment: String::new(),
+                ..Default::default()
             },
             ColumnDef {
                 name: "mta_tax".to_string(),
@@ -362,6 +375,7 @@ fn create_table_expr(table_name: &str) -> CreateTableExpr {
                 default_constraint: vec![],
                 semantic_type: SemanticType::Field as i32,
                 comment: String::new(),
+                ..Default::default()
             },
             ColumnDef {
                 name: "tip_amount".to_string(),
@@ -370,6 +384,7 @@ fn create_table_expr(table_name: &str) -> CreateTableExpr {
                 default_constraint: vec![],
                 semantic_type: SemanticType::Field as i32,
                 comment: String::new(),
+                ..Default::default()
             },
             ColumnDef {
                 name: "tolls_amount".to_string(),
@@ -378,6 +393,7 @@ fn create_table_expr(table_name: &str) -> CreateTableExpr {
                 default_constraint: vec![],
                 semantic_type: SemanticType::Field as i32,
                 comment: String::new(),
+                ..Default::default()
             },
             ColumnDef {
                 name: "improvement_surcharge".to_string(),
@@ -386,6 +402,7 @@ fn create_table_expr(table_name: &str) -> CreateTableExpr {
                 default_constraint: vec![],
                 semantic_type: SemanticType::Field as i32,
                 comment: String::new(),
+                ..Default::default()
             },
             ColumnDef {
                 name: "total_amount".to_string(),
@@ -394,6 +411,7 @@ fn create_table_expr(table_name: &str) -> CreateTableExpr {
                 default_constraint: vec![],
                 semantic_type: SemanticType::Field as i32,
                 comment: String::new(),
+                ..Default::default()
             },
             ColumnDef {
                 name: "congestion_surcharge".to_string(),
@@ -402,6 +420,7 @@ fn create_table_expr(table_name: &str) -> CreateTableExpr {
                 default_constraint: vec![],
                 semantic_type: SemanticType::Field as i32,
                 comment: String::new(),
+                ..Default::default()
             },
             ColumnDef {
                 name: "airport_fee".to_string(),
@@ -410,6 +429,7 @@ fn create_table_expr(table_name: &str) -> CreateTableExpr {
                 default_constraint: vec![],
                 semantic_type: SemanticType::Field as i32,
                 comment: String::new(),
+                ..Default::default()
             },
         ],
         time_index: "tpep_pickup_datetime".to_string(),

config/datanode.example.toml

@@ -12,6 +12,10 @@ rpc_runtime_size = 8
 # It will block the datanode start if it can't receive leases in the heartbeat from metasrv.
 require_lease_before_startup = false
 
+# Initialize all regions in the background during the startup.
+# By default, it provides services after all regions have been initialized.
+initialize_region_in_background = false
+
 [heartbeat]
 # Interval for sending heartbeat messages to the Metasrv, 3 seconds by default.
 interval = "3s"
@@ -29,9 +33,15 @@ connect_timeout = "1s"
 # `TCP_NODELAY` option for accepted connections, true by default.
 tcp_nodelay = true
 
-# WAL options, see `standalone.example.toml`.
+# WAL options.
+# Currently, users are expected to choose the wal through the provider field. 
+# When a wal provider is chose, the user should comment out all other wal config 
+# except those corresponding to the chosen one.
 [wal]
 # WAL data directory
+provider = "raft_engine"
+
+# Raft-engine wal options, see `standalone.example.toml`.
 # dir = "/tmp/greptimedb/wal"
 file_size = "256MB"
 purge_threshold = "4GB"
@@ -39,10 +49,21 @@ purge_interval = "10m"
 read_batch_size = 128
 sync_write = false
 
+# Kafka wal options, see `standalone.example.toml`.
+# broker_endpoints = ["127.0.0.1:9092"]
+# max_batch_size = "4MB"
+# linger = "200ms"
+# produce_record_timeout = "100ms"
+# backoff_init = "500ms"
+# backoff_max = "10s"
+# backoff_base = 2
+# backoff_deadline = "5mins"
+
 # Storage options, see `standalone.example.toml`.
 [storage]
 # The working home directory.
 data_home = "/tmp/greptimedb/"
+# Storage type.
 type = "File"
 # TTL for all tables. Disabled by default.
 # global_ttl = "7d"
@@ -53,32 +74,11 @@ type = "File"
 # The local file cache capacity in bytes.
 # cache_capacity = "256MB"
 
-# Compaction options, see `standalone.example.toml`.
-[storage.compaction]
-max_inflight_tasks = 4
-max_files_in_level0 = 8
-max_purge_tasks = 32
-
-# Storage manifest options
-[storage.manifest]
-# Region checkpoint actions margin.
-# Create a checkpoint every <checkpoint_margin> actions.
-checkpoint_margin = 10
-# Region manifest logs and checkpoints gc execution duration
-gc_duration = '10m'
-
-# Storage flush options
-[storage.flush]
-# Max inflight flush tasks.
-max_flush_tasks = 8
-# Default write buffer size for a region.
-region_write_buffer_size = "32MB"
-# Interval to check whether a region needs flush.
-picker_schedule_interval = "5m"
-# Interval to auto flush a region if it has not flushed yet.
-auto_flush_interval = "1h"
-# Global write buffer size for all regions.
-global_write_buffer_size = "1GB"
+# Custom storage options
+#[[storage.providers]]
+#type = "S3"
+#[[storage.providers]]
+#type = "Gcs"
 
 # Mito engine options
 [[region_engine]]
@@ -91,8 +91,8 @@ worker_channel_size = 128
 worker_request_batch_size = 64
 # Number of meta action updated to trigger a new checkpoint for the manifest
 manifest_checkpoint_distance = 10
-# Manifest compression type
-manifest_compress_type = "Uncompressed"
+# Whether to compress manifest and checkpoint file by gzip (default false).
+compress_manifest = false
 # Max number of running background jobs
 max_background_jobs = 4
 # Interval to auto flush a region if it has not flushed yet.
@@ -101,10 +101,39 @@ auto_flush_interval = "1h"
 global_write_buffer_size = "1GB"
 # Global write buffer size threshold to reject write requests (default 2G).
 global_write_buffer_reject_size = "2GB"
+# Cache size for SST metadata (default 128MB). Setting it to 0 to disable the cache.
+sst_meta_cache_size = "128MB"
+# Cache size for vectors and arrow arrays (default 512MB). Setting it to 0 to disable the cache.
+vector_cache_size = "512MB"
+# Cache size for pages of SST row groups (default 512MB). Setting it to 0 to disable the cache.
+page_cache_size = "512MB"
+# Buffer size for SST writing.
+sst_write_buffer_size = "8MB"
+# Parallelism to scan a region (default: 1/4 of cpu cores).
+# - 0: using the default value (1/4 of cpu cores).
+# - 1: scan in current thread.
+# - n: scan in parallelism n.
+scan_parallelism = 0
+# Capacity of the channel to send data from parallel scan tasks to the main task (default 32).
+parallel_scan_channel_size = 32
 
-# Log options
+# Log options, see `standalone.example.toml`
 # [logging]
-# Specify logs directory.
 # dir = "/tmp/greptimedb/logs"
-# Specify the log level [info | debug | error | warn]
 # level = "info"
+
+# Datanode export the metrics generated by itself
+# encoded to Prometheus remote-write format
+# and send to Prometheus remote-write compatible receiver (e.g. send to `greptimedb` itself)
+# This is only used for `greptimedb` to export its own metrics internally. It's different from prometheus scrape.
+# [export_metrics]
+# whether enable export metrics, default is false
+# enable = false
+# The url of metrics export endpoint, default is `frontend` default HTTP endpoint.
+# endpoint = "127.0.0.1:4000"
+# The database name of exported metrics stores, user needs to specify a valid database
+# db = ""
+# The interval of export metrics
+# write_interval = "30s"
+# HTTP headers of Prometheus remote-write carry
+# headers = {}

config/frontend.example.toml

@@ -1,5 +1,7 @@
 # Node running mode, see `standalone.example.toml`.
 mode = "distributed"
+# The default timezone of the server
+# default_timezone = "UTC"
 
 [heartbeat]
 # Interval for sending heartbeat task to the Metasrv, 5 seconds by default.
@@ -77,3 +79,19 @@ tcp_nodelay = true
 timeout = "10s"
 connect_timeout = "10s"
 tcp_nodelay = true
+
+# Frontend export the metrics generated by itself
+# encoded to Prometheus remote-write format
+# and send to Prometheus remote-write compatible receiver (e.g. send to `greptimedb` itself)
+# This is only used for `greptimedb` to export its own metrics internally. It's different from prometheus scrape.
+# [export_metrics]
+# whether enable export metrics, default is false
+# enable = false
+# The url of metrics export endpoint, default is `frontend` default HTTP endpoint.
+# endpoint = "127.0.0.1:4000"
+# The database name of exported metrics stores, user needs to specify a valid database
+# db = ""
+# The interval of export metrics
+# write_interval = "30s"
+# HTTP headers of Prometheus remote-write carry
+# headers = {}

config/metasrv.example.toml

@@ -7,14 +7,16 @@ server_addr = "127.0.0.1:3002"
 # Etcd server address, "127.0.0.1:2379" by default.
 store_addr = "127.0.0.1:2379"
 # Datanode selector type.
-# - "LeaseBased" (default value).
-# - "LoadBased"
-# For details, please see "https://docs.greptime.com/developer-guide/meta/selector".
-selector = "LeaseBased"
+# - "lease_based" (default value).
+# - "load_based"
+# For details, please see "https://docs.greptime.com/developer-guide/metasrv/selector".
+selector = "lease_based"
 # Store data in memory, false by default.
 use_memory_store = false
 # Whether to enable greptimedb telemetry, true by default.
 enable_telemetry = true
+# If it's not empty, the metasrv will store all data with this key prefix.
+store_key_prefix = ""
 
 # Log options, see `standalone.example.toml`
 # [logging]
@@ -28,6 +30,13 @@ max_retry_times = 12
 # Initial retry delay of procedures, increases exponentially
 retry_delay = "500ms"
 
+# Failure detectors options.
+[failure_detector]
+threshold = 8.0
+min_std_deviation = "100ms"
+acceptable_heartbeat_pause = "3000ms"
+first_heartbeat_estimate = "1000ms"
+
 # # Datanode options.
 # [datanode]
 # # Datanode client options.
@@ -35,3 +44,53 @@ retry_delay = "500ms"
 # timeout = "10s"
 # connect_timeout = "10s"
 # tcp_nodelay = true
+
+[wal]
+# Available wal providers:
+# - "raft_engine" (default)
+# - "kafka"
+provider = "raft_engine"
+
+# There're none raft-engine wal config since meta srv only involves in remote wal currently.
+
+# Kafka wal config.
+# The broker endpoints of the Kafka cluster. ["127.0.0.1:9092"] by default.
+# broker_endpoints = ["127.0.0.1:9092"]
+# Number of topics to be created upon start.
+# num_topics = 64
+# Topic selector type.
+# Available selector types: 
+# - "round_robin" (default)
+# selector_type = "round_robin"
+# A Kafka topic is constructed by concatenating `topic_name_prefix` and `topic_id`.
+# topic_name_prefix = "greptimedb_wal_topic"
+# Number of partitions per topic.
+# num_partitions = 1
+# Expected number of replicas of each partition.
+# replication_factor = 1
+# Above which a topic creation operation will be cancelled.
+# create_topic_timeout = "30s"
+# The initial backoff for kafka clients.
+# backoff_init = "500ms"
+# The maximum backoff for kafka clients.
+# backoff_max = "10s"
+# Exponential backoff rate, i.e. next backoff = base * current backoff.
+# backoff_base = 2
+# Stop reconnecting if the total wait time reaches the deadline. If this config is missing, the reconnecting won't terminate.
+# backoff_deadline = "5mins"
+
+# Metasrv export the metrics generated by itself
+# encoded to Prometheus remote-write format
+# and send to Prometheus remote-write compatible receiver (e.g. send to `greptimedb` itself)
+# This is only used for `greptimedb` to export its own metrics internally. It's different from prometheus scrape.
+# [export_metrics]
+# whether enable export metrics, default is false
+# enable = false
+# The url of metrics export endpoint, default is `frontend` default HTTP endpoint.
+# endpoint = "127.0.0.1:4000"
+# The database name of exported metrics stores, user needs to specify a valid database
+# db = ""
+# The interval of export metrics
+# write_interval = "30s"
+# HTTP headers of Prometheus remote-write carry
+# headers = {}

config/standalone.example.toml

@@ -2,6 +2,8 @@
 mode = "standalone"
 # Whether to enable greptimedb telemetry, true by default.
 enable_telemetry = true
+# The default timezone of the server
+# default_timezone = "UTC"
 
 # HTTP server options.
 [http]
@@ -80,8 +82,49 @@ enable = true
 # Whether to enable Prometheus remote write and read in HTTP API, true by default.
 enable = true
 
-# WAL options.
 [wal]
+# Available wal providers:
+# - "raft_engine" (default)
+# - "kafka"
+provider = "raft_engine"
+
+# There're none raft-engine wal config since meta srv only involves in remote wal currently.
+
+# Kafka wal options.
+# The broker endpoints of the Kafka cluster. ["127.0.0.1:9092"] by default.
+# broker_endpoints = ["127.0.0.1:9092"]
+
+# Number of topics to be created upon start.
+# num_topics = 64
+# Topic selector type.
+# Available selector types: 
+# - "round_robin" (default)
+# selector_type = "round_robin"
+# A Kafka topic is constructed by concatenating `topic_name_prefix` and `topic_id`.
+# topic_name_prefix = "greptimedb_wal_topic"
+# Number of partitions per topic.
+# num_partitions = 1
+# Expected number of replicas of each partition.
+# replication_factor = 1
+
+# The maximum log size a kafka batch producer could buffer.
+# max_batch_size = "4MB"
+# The linger duration of a kafka batch producer.
+# linger = "200ms"
+# The maximum amount of time (in milliseconds) to wait for Kafka records to be returned.
+# produce_record_timeout = "100ms"
+# Above which a topic creation operation will be cancelled.
+# create_topic_timeout = "30s"
+
+# The initial backoff for kafka clients.
+# backoff_init = "500ms"
+# The maximum backoff for kafka clients.
+# backoff_max = "10s"
+# Exponential backoff rate, i.e. next backoff = base * current backoff.
+# backoff_base = 2
+# Stop reconnecting if the total wait time reaches the deadline. If this config is missing, the reconnecting won't terminate.
+# backoff_deadline = "5mins"
+
 # WAL data directory
 # dir = "/tmp/greptimedb/wal"
 # WAL file size in bytes.
@@ -122,35 +165,48 @@ type = "File"
 # The local file cache capacity in bytes.
 # cache_capacity = "256MB"
 
-# Compaction options.
-[storage.compaction]
-# Max task number that can concurrently run.
-max_inflight_tasks = 4
-# Max files in level 0 to trigger compaction.
-max_files_in_level0 = 8
-# Max task number for SST purge task after compaction.
-max_purge_tasks = 32
+# Custom storage options
+#[[storage.providers]]
+#type = "S3"
+#[[storage.providers]]
+#type = "Gcs"
 
-# Storage manifest options
-[storage.manifest]
-# Region checkpoint actions margin.
-# Create a checkpoint every <checkpoint_margin> actions.
-checkpoint_margin = 10
-# Region manifest logs and checkpoints gc execution duration
-gc_duration = '10m'
-
-# Storage flush options
-[storage.flush]
-# Max inflight flush tasks.
-max_flush_tasks = 8
-# Default write buffer size for a region.
-region_write_buffer_size = "32MB"
-# Interval to check whether a region needs flush.
-picker_schedule_interval = "5m"
+# Mito engine options
+[[region_engine]]
+[region_engine.mito]
+# Number of region workers
+num_workers = 8
+# Request channel size of each worker
+worker_channel_size = 128
+# Max batch size for a worker to handle requests
+worker_request_batch_size = 64
+# Number of meta action updated to trigger a new checkpoint for the manifest
+manifest_checkpoint_distance = 10
+# Whether to compress manifest and checkpoint file by gzip (default false).
+compress_manifest = false
+# Max number of running background jobs
+max_background_jobs = 4
 # Interval to auto flush a region if it has not flushed yet.
 auto_flush_interval = "1h"
 # Global write buffer size for all regions.
 global_write_buffer_size = "1GB"
+# Global write buffer size threshold to reject write requests (default 2G).
+global_write_buffer_reject_size = "2GB"
+# Cache size for SST metadata (default 128MB). Setting it to 0 to disable the cache.
+sst_meta_cache_size = "128MB"
+# Cache size for vectors and arrow arrays (default 512MB). Setting it to 0 to disable the cache.
+vector_cache_size = "512MB"
+# Cache size for pages of SST row groups (default 512MB). Setting it to 0 to disable the cache.
+page_cache_size = "512MB"
+# Buffer size for SST writing.
+sst_write_buffer_size = "8MB"
+# Parallelism to scan a region (default: 1/4 of cpu cores).
+# - 0: using the default value (1/4 of cpu cores).
+# - 1: scan in current thread.
+# - n: scan in parallelism n.
+scan_parallelism = 0
+# Capacity of the channel to send data from parallel scan tasks to the main task (default 32).
+parallel_scan_channel_size = 32
 
 # Log options
 # [logging]
@@ -158,3 +214,27 @@ global_write_buffer_size = "1GB"
 # dir = "/tmp/greptimedb/logs"
 # Specify the log level [info | debug | error | warn]
 # level = "info"
+# whether enable tracing, default is false
+# enable_otlp_tracing = false
+# tracing exporter endpoint with format `ip:port`, we use grpc oltp as exporter, default endpoint is `localhost:4317`
+# otlp_endpoint = "localhost:4317"
+# The percentage of tracing will be sampled and exported. Valid range `[0, 1]`, 1 means all traces are sampled, 0 means all traces are not sampled, the default value is 1. ratio > 1 are treated as 1. Fractions < 0 are treated as 0
+# tracing_sample_ratio = 1.0
+# Whether to append logs to stdout. Defaults to true.
+# append_stdout = true
+
+# Standalone export the metrics generated by itself
+# encoded to Prometheus remote-write format
+# and send to Prometheus remote-write compatible receiver (e.g. send to `greptimedb` itself)
+# This is only used for `greptimedb` to export its own metrics internally. It's different from prometheus scrape.
+# [export_metrics]
+# whether enable export metrics, default is false
+# enable = false
+# The url of metrics export endpoint, default is `frontend` default HTTP endpoint.
+# endpoint = "127.0.0.1:4000"
+# The database name of exported metrics stores, user needs to specify a valid database
+# db = ""
+# The interval of export metrics
+# write_interval = "30s"
+# HTTP headers of Prometheus remote-write carry
+# headers = {}

docker/dev-builder/centos/Dockerfile

@@ -26,4 +26,5 @@ ARG RUST_TOOLCHAIN
 RUN rustup toolchain install ${RUST_TOOLCHAIN}
 
 # Install nextest.
-RUN cargo install cargo-nextest --locked
+RUN cargo install cargo-binstall --locked
+RUN cargo binstall cargo-nextest --no-confirm

docker/dev-builder/ubuntu/Dockerfile

@@ -19,8 +19,13 @@ RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y \
     build-essential \
     pkg-config \
     python3.10 \
-    python3.10-dev \
-    python3-pip
+    python3.10-dev
+
+# Remove Python 3.8 and install pip.
+RUN apt-get -y purge python3.8 && \
+    apt-get -y autoremove && \
+    ln -s /usr/bin/python3.10 /usr/bin/python3 && \
+    curl -sS https://bootstrap.pypa.io/get-pip.py | python3.10
 
 RUN git config --global --add safe.directory /greptimedb
 
@@ -38,4 +43,5 @@ ARG RUST_TOOLCHAIN
 RUN rustup toolchain install ${RUST_TOOLCHAIN}
 
 # Install nextest.
-RUN cargo install cargo-nextest --locked
+RUN cargo install cargo-binstall --locked
+RUN cargo binstall cargo-nextest --no-confirm

docker/dev-builder/ubuntu/Dockerfile-18.10

@@ -0,0 +1,48 @@
+# Use the legacy glibc 2.28.
+FROM ubuntu:18.10
+
+ENV LANG en_US.utf8
+WORKDIR /greptimedb
+
+# Use old-releases.ubuntu.com to avoid 404s: https://help.ubuntu.com/community/EOLUpgrades.
+RUN echo "deb http://old-releases.ubuntu.com/ubuntu/ cosmic main restricted universe multiverse\n\
+deb http://old-releases.ubuntu.com/ubuntu/ cosmic-updates main restricted universe multiverse\n\
+deb http://old-releases.ubuntu.com/ubuntu/ cosmic-security main restricted universe multiverse" > /etc/apt/sources.list
+
+# Install dependencies.
+RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y \
+    libssl-dev \
+    tzdata \
+    curl \
+    ca-certificates \
+    git \
+    build-essential \
+    unzip \
+    pkg-config
+
+# Install protoc.
+ENV PROTOC_VERSION=25.1
+RUN if [ "$(uname -m)" = "x86_64" ]; then \
+        PROTOC_ZIP=protoc-${PROTOC_VERSION}-linux-x86_64.zip; \
+    elif [ "$(uname -m)" = "aarch64" ]; then \
+        PROTOC_ZIP=protoc-${PROTOC_VERSION}-linux-aarch_64.zip; \
+    else \
+        echo "Unsupported architecture"; exit 1; \
+    fi && \
+    curl -OL https://github.com/protocolbuffers/protobuf/releases/download/v${PROTOC_VERSION}/${PROTOC_ZIP} && \
+    unzip -o ${PROTOC_ZIP} -d /usr/local bin/protoc && \
+    unzip -o ${PROTOC_ZIP} -d /usr/local 'include/*' && \
+    rm -f ${PROTOC_ZIP}
+
+# Install Rust.
+SHELL ["/bin/bash", "-c"]
+RUN curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- --no-modify-path --default-toolchain none -y
+ENV PATH /root/.cargo/bin/:$PATH
+
+# Install Rust toolchains.
+ARG RUST_TOOLCHAIN
+RUN rustup toolchain install ${RUST_TOOLCHAIN}
+
+# Install nextest.
+RUN cargo install cargo-binstall --locked
+RUN cargo binstall cargo-nextest --no-confirm

docs/rfcs/2023-07-10-metric-engine.md

@@ -50,10 +50,10 @@ The concept "Table" in GreptimeDB is a bit "heavy" compared to other time-series
 	```
 
 The following parts will describe these implementation details:
-    - How to route these metric region tables and how those table are distributed
-    - How to maintain the schema and other metadata of the underlying mito engine table
-    - How to maintain the schema of metric engine table
-    - How the query goes
+- How to route these metric region tables and how those table are distributed
+- How to maintain the schema and other metadata of the underlying mito engine table
+- How to maintain the schema of metric engine table
+- How the query goes
 
 ## Routing
 

docs/rfcs/2023-11-03-inverted-index.md

@@ -0,0 +1,113 @@
+---
+Feature Name: Inverted Index for SST File
+Tracking Issue: TBD
+Date: 2023-11-03
+Author: "Zhong Zhenchi <zhongzc_arch@outlook.com>"
+---
+
+# Summary
+This RFC proposes an optimization towards the storage engine by introducing an inverted indexing methodology aimed at optimizing label selection queries specifically pertaining to Metrics with tag columns as the target for optimization.
+
+# Introduction
+In the current system setup, in the Mito Engine, the first column of Primary Keys has a Min-Max index, which significantly optimizes the outcome. However, there are limitations when it comes to other columns, primarily tags. This RFC suggests the implementation of an inverted index to provide enhanced filtering benefits to bridge these limitations and improve overall system performance.
+
+# Design Detail
+
+## Inverted Index
+
+The primary aim of the proposed inverted index is to optimize tag columns in the SST Parquet Files within the Mito Engine. The mapping and construction of an inverted index, from Tag Values to Row Groups, enables efficient logical structures that provide faster and more flexible queries.
+
+When scanning SST Files, pushed-down filters applied to a respective Tag's inverted index, determine the final Row Groups to be indexed and scanned, further bolstering the speed and efficiency of data retrieval processes.
+
+## Index Format
+
+The Inverted Index for each SST file follows the format shown below:
+
+```
+inverted_index₀ inverted_index₁ ... inverted_indexₙ footer
+```
+
+The structure inside each Inverted Index is as followed:
+
+```
+bitmap₀ bitmap₁ bitmap₂ ... bitmapₙ null_bitmap fst
+```
+
+The format is encapsulated by a footer:
+
+```
+footer_payload footer_payload_size
+```
+
+The `footer_payload` is presented in protobuf encoding of `InvertedIndexFooter`.
+
+The complete format is containerized in [Puffin](https://iceberg.apache.org/puffin-spec/) with the type defined as `greptime-inverted-index-v1`.
+
+## Protobuf Details
+
+The `InvertedIndexFooter` is defined in the following protobuf structure:
+
+```protobuf
+message InvertedIndexFooter {
+    repeated InvertedIndexMeta metas;
+}
+
+message InvertedIndexMeta {
+    string name;
+    uint64 row_count_in_group;
+    uint64 fst_offset;
+    uint64 fst_size;
+    uint64 null_bitmap_offset;
+    uint64 null_bitmap_size;
+    InvertedIndexStats stats;
+}
+
+message InvertedIndexStats {
+    uint64 null_count;
+    uint64 distinct_count;
+    bytes min_value;
+    bytes max_value;
+}
+```
+
+## Bitmap
+
+Bitmaps are used to represent indices of fixed-size groups. Rows are divided into groups of a fixed size, defined in the `InvertedIndexMeta` as `row_count_in_group`.
+
+For example, when `row_count_in_group` is `4096`, it means each group has `4096` rows. If there are a total of `10000` rows, there will be `3` groups in total. The first two groups will have `4096` rows each, and the last group will have `1808` rows. If the indexed values are found in row `200` and `9000`, they will correspond to groups `0` and `2`, respectively. Therefore, the bitmap should show `0` and `2`.
+
+Bitmap is implemented using [BitVec](https://docs.rs/bitvec/latest/bitvec/), selected due to its efficient representation of dense data arrays typical of indices of groups.
+
+
+## Finite State Transducer (FST)
+
+[FST](https://docs.rs/fst/latest/fst/) is a highly efficient data structure ideal for in-memory indexing. It represents ordered sets or maps where the keys are bytes. The choice of the FST effectively balances the need for performance, space efficiency, and the ability to perform complex analyses such as regular expression matching.
+
+The conventional usage of FST and `u64` values has been adapted to facilitate indirect indexing to row groups. As the row groups are represented as Bitmaps, we utilize the `u64` values split into bitmap's offset (higher 32 bits) and size (lower 32 bits) to represent the location of these Bitmaps. 
+
+## API Design
+
+Two APIs `InvertedIndexBuilder` for building indexes and  `InvertedIndexSearcher` for querying indexes are designed:
+
+```rust
+type Bytes = Vec<u8>;
+type GroupId = u64;
+
+trait InvertedIndexBuilder {
+    fn add(&mut self, name: &str, value: Option<&Bytes>, group_id: GroupId) -> Result<()>;
+    fn finish(&mut self) -> Result<()>;
+}
+
+enum Predicate {
+    Gt(Bytes),
+    GtEq(Bytes),
+    Lt(Bytes),
+    LtEq(Bytes),
+    InList(Vec<Bytes>),
+    RegexMatch(String),
+}
+
+trait InvertedIndexSearcher {
+    fn search(&mut self, name: &str, predicates: &[Predicate]) -> Result<impl IntoIterator<GroupId>>;
+}
+```

docs/rfcs/2023-11-07-region-migration.md

@@ -0,0 +1,169 @@
+---
+Feature Name: Region Migration Procedure
+Tracking Issue: https://github.com/GreptimeTeam/greptimedb/issues/2700
+Date: 2023-11-03
+Author: "Xu Wenkang <wenymedia@gmail.com>"
+---
+
+# Summary
+This RFC proposes a way that brings the ability of Meta Server to move regions between the Datanodes.
+
+# Motivation
+Typically, We need this ability in the following scenarios:
+- Migrate hot-spot Regions to idle Datanode
+- Move the failure Regions to an available Datanode
+
+# Details
+
+```mermaid
+flowchart TD
+    style Start fill:#85CB90,color:#fff
+    style End fill:#85CB90,color:#fff
+    style SelectCandidate fill:#F38488,color:#fff
+    style OpenCandidate fill:#F38488,color:#fff
+    style UpdateMetadataDown fill:#F38488,color:#fff
+    style UpdateMetadataUp fill:#F38488,color:#fff
+    style UpdateMetadataRollback fill:#F38488,color:#fff
+    style DowngradeLeader fill:#F38488,color:#fff
+    style UpgradeCandidate fill:#F38488,color:#fff
+
+    Start[Start] 
+    SelectCandidate[Select Candidate] 
+    UpdateMetadataDown["`Update Metadata(Down)
+        1. Downgrade Leader
+    `"]
+    DowngradeLeader["`Downgrade Leader
+    1. Become Follower
+    2. Return **last_entry_id**
+    `"]
+    UpgradeCandidate["`Upgrade Candidate
+    1. Replay to **last_entry_id**
+    2. Become Leader
+    `"]
+    UpdateMetadataUp["`Update Metadata(Up)
+        1. Switch Leader
+        2.1. Remove Old Leader(Opt.)
+        2.2. Move Old Leader to Follower(Opt.)
+    `"]
+    UpdateMetadataRollback["`Update Metadata(Rollback)
+        1. Upgrade old Leader
+    `"]
+    End
+    AnyCandidate{Available?}
+    OpenCandidate["Open Candidate"]
+    CloseOldLeader["Close Old Leader"]
+
+    Start 
+    --> SelectCandidate
+    --> AnyCandidate
+    --> |Yes| UpdateMetadataDown 
+    --> I1["Invalid Frontend Cache"]
+    --> DowngradeLeader 
+    --> UpgradeCandidate
+    --> UpdateMetadataUp
+    --> I2["Invalid Frontend Cache"]
+    --> End
+    
+    UpgradeCandidate
+    --> UpdateMetadataRollback
+    --> I3["Invalid Frontend Cache"]
+    --> End
+
+    I2 
+    --> CloseOldLeader
+    --> End
+
+    AnyCandidate 
+    --> |No| OpenCandidate
+    --> UpdateMetadataDown
+```
+
+**Only the red nodes will persist state after it has succeeded**, and other nodes won't persist state. (excluding the Start and End nodes).
+
+## Steps
+
+**The persistent context:** It's shared in each step and available after recovering. It will only be updated/stored after the Red node has succeeded.
+
+Values: 
+- `region_id`: The target leader region.
+- `peer`: The target datanode.
+- `close_old_leader`: Indicates whether close the region. 
+- `leader_may_unreachable`: It's used to support the failover procedure.
+
+**The Volatile context:** It's shared in each step and available in executing (including retrying). It will be dropped if the procedure runner crashes.
+
+### Select Candidate
+
+The Persistent state: Selected Candidate Region.
+
+### Update Metadata(Down)
+
+**The Persistent context:**
+- The (latest/updated) `version` of `TableRouteValue`, It will be used in the step of `Update Metadata(Up)`.
+
+### Downgrade Leader
+This step sends an instruction via heartbeat and performs:
+1. Downgrades leader region.
+2. Retrieves the `last_entry_id` (if available).
+
+If the target leader region is not found: 
+- Sets `close_old_leader` to true.
+- Sets `leader_may_unreachable` to true.
+
+If the target Datanode is unreachable:
+- Waits for region lease expired.
+- Sets `close_old_leader` to true.
+- Sets `leader_may_unreachable` to true.
+
+**The Persistent context:**
+None
+
+**The Persistent state:** 
+- `last_entry_id`
+
+*Passes to next step.
+
+
+### Upgrade Candidate
+This step sends an instruction via heartbeat and performs:
+1. Replays the WAL to latest(`last_entry_id`).
+2. Upgrades the candidate region.
+
+If the target region is not found: 
+- Rollbacks.
+- Notifies the failover detector if `leader_may_unreachable` == true.
+- Exits procedure.
+
+If the target Datanode is unreachable:
+- Rollbacks.
+- Notifies the failover detector if `leader_may_unreachable` == true.
+- Exits procedure.
+
+**The Persistent context:**
+None
+
+### Update Metadata(Up)
+This step performs
+1. Switches Leader.
+2. Removes Old Leader(Opt.).
+3. Moves Old Leader to follower(Opt.).
+
+The `TableRouteValue` version should equal the `TableRouteValue`'s `version` in Persistent context. Otherwise, verifies whether `TableRouteValue` already updated.
+
+**The Persistent context:**
+None
+
+### Close Old Leader(Opt.)
+This step sends a close region instruction via heartbeat.
+
+If the target leader region is not found: 
+- Ignore.
+
+If the target Datanode is unreachable: 
+- Ignore.
+
+### Open Candidate(Opt.)
+This step sends an open region instruction via heartbeat and waits for conditions to be met (typically, the condition is that the `last_entry_id` of the Candidate Region is very close to that of the Leader Region or the latest).
+
+If the target Datanode is unreachable: 
+- Exits procedure.

docs/rfcs/2023-12-22-enclose-column-id.md

@@ -0,0 +1,44 @@
+---
+Feature Name: Enclose Column Id
+Tracking Issue: https://github.com/GreptimeTeam/greptimedb/issues/2982
+Date: 2023-12-22
+Author: "Ruihang Xia <waynestxia@gmail.com>"
+---
+
+# Summary
+This RFC proposes to enclose the usage of `ColumnId` into the region engine only.
+
+# Motivation
+`ColumnId` is an identifier for columns. It's assigned by meta server, stored in `TableInfo` and `RegionMetadata` and used in region engine to distinguish columns.
+
+At present, Both Frontend, Datanode and Metasrv are aware of `ColumnId` but it's only used in region engine. Thus this RFC proposes to remove it from Frontend (mainly used in `TableInfo`) and Metasrv.
+
+# Details
+
+`ColumnId` is used widely on both read and write paths. Removing it from Frontend and Metasrv implies several things:
+
+- A column may have different column id in different regions.
+- A column is identified by its name in all components.
+- Column order in the region engine is not restricted, i.e., no need to be in the same order with table info.
+
+The first thing doesn't matter IMO. This concept doesn't exist anymore outside of region server, and each region is autonomous and independent -- the only guarantee it should hold is those columns exist. But if we consider region repartition, where the SST file would be re-assign to different regions, things would become a bit more complicated. A possible solution is store the relation between name and ColumnId in the manifest, but it's out of the scope of this RFC. We can likely give a workaround by introducing a indirection mapping layer of different version of partitions.
+
+And more importantly, we can still assume columns have the same column ids across regions. We have procedure to maintain consistency between regions and the region engine should ensure alterations are idempotent. So it is possible that region repartition doesn't need to consider column ids or other region metadata in the future.
+
+Users write and query column by their names, not by ColumnId or something else. The second point also means to change the column reference in ScanRequest from index to name. This change can hugely alleviate the misuse of the column index, which has given us many surprises.
+
+And for the last one, column order only matters in table info. This order is used in user-faced table structure operation, like add column, describe column or as the default order of INSERT clause. None of them is connected with the order in storage.
+
+# Drawback
+Firstly, this is a breaking change. Delivering this change requires a full upgrade of the cluster. Secondly, this change may introduce some performance regression. For example, we have to pass the full table name in the `ScanRequest` instead of the `ColumnId`. But this influence is very limited, since the column index is only used in the region engine.
+
+# Alternatives
+
+There are two alternatives from the perspective of "what can be used as the column identifier":
+
+- Index of column to the table schema
+- `ColumnId` of that column
+
+The first one is what we are using now. By choosing this way, it's required to keep the column order in the region engine the same as the table info. This is not hard to achieve, but it's a bit annoying. And things become tricky when there is internal column or different schemas like those stored in file format. And this is the initial purpose of this RFC, which is trying to decouple the table schema and region schema.
+
+The second one, in other hand, requires the `ColumnId` should be identical in all regions and `TableInfo`. It has the same drawback with the previous alternative, that the `TableInfo` and `RegionMetadata` are tighted together. Another point is that the `ColumnId` is assigned by the Metasrv, who doesn't need it but have to maintain it. And this also limits the functionality of `ColumnId`, by taking the ability of assigning it from concrete region engine.

licenserc.toml

@@ -0,0 +1,24 @@
+# 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.
+
+headerPath = "Apache-2.0.txt"
+
+includes = [
+    "*.rs",
+    "*.py",
+]
+
+[properties]
+inceptionYear = 2023
+copyrightOwner = "Greptime Team"

rust-toolchain.toml

@@ -1,2 +1,2 @@
 [toolchain]
-channel = "nightly-2023-08-07"
+channel = "nightly-2023-12-19"

src/api/Cargo.toml

@@ -5,14 +5,16 @@ edition.workspace = true
 license.workspace = true
 
 [dependencies]
-common-base = { workspace = true }
-common-error = { workspace = true }
-common-macro = { workspace = true }
-common-time = { workspace = true }
-datatypes = { workspace = true }
+common-base.workspace = true
+common-decimal.workspace = true
+common-error.workspace = true
+common-macro.workspace = true
+common-time.workspace = true
+datatypes.workspace = true
 greptime-proto.workspace = true
+paste = "1.0"
 prost.workspace = true
-snafu = { version = "0.7", features = ["backtraces"] }
+snafu.workspace = true
 tonic.workspace = true
 
 [build-dependencies]

src/api/src/error.rs

@@ -28,7 +28,12 @@ pub type Result<T> = std::result::Result<T, Error>;
 #[stack_trace_debug]
 pub enum Error {
     #[snafu(display("Unknown proto column datatype: {}", datatype))]
-    UnknownColumnDataType { datatype: i32, location: Location },
+    UnknownColumnDataType {
+        datatype: i32,
+        location: Location,
+        #[snafu(source)]
+        error: prost::DecodeError,
+    },
 
     #[snafu(display("Failed to create column datatype from {:?}", from))]
     IntoColumnDataType {

src/api/src/helper.rs

@@ -15,6 +15,8 @@
 use std::sync::Arc;
 
 use common_base::BitVec;
+use common_decimal::decimal128::{DECIMAL128_DEFAULT_SCALE, DECIMAL128_MAX_PRECISION};
+use common_decimal::Decimal128;
 use common_time::interval::IntervalUnit;
 use common_time::time::Time;
 use common_time::timestamp::TimeUnit;
@@ -26,47 +28,71 @@ use datatypes::types::{
 };
 use datatypes::value::{OrderedF32, OrderedF64, Value};
 use datatypes::vectors::{
-    BinaryVector, BooleanVector, DateTimeVector, DateVector, DurationMicrosecondVector,
-    DurationMillisecondVector, DurationNanosecondVector, DurationSecondVector, Float32Vector,
-    Float64Vector, Int32Vector, Int64Vector, IntervalDayTimeVector, IntervalMonthDayNanoVector,
-    IntervalYearMonthVector, PrimitiveVector, StringVector, TimeMicrosecondVector,
-    TimeMillisecondVector, TimeNanosecondVector, TimeSecondVector, TimestampMicrosecondVector,
-    TimestampMillisecondVector, TimestampNanosecondVector, TimestampSecondVector, UInt32Vector,
-    UInt64Vector, VectorRef,
+    BinaryVector, BooleanVector, DateTimeVector, DateVector, Decimal128Vector,
+    DurationMicrosecondVector, DurationMillisecondVector, DurationNanosecondVector,
+    DurationSecondVector, Float32Vector, Float64Vector, Int32Vector, Int64Vector,
+    IntervalDayTimeVector, IntervalMonthDayNanoVector, IntervalYearMonthVector, PrimitiveVector,
+    StringVector, TimeMicrosecondVector, TimeMillisecondVector, TimeNanosecondVector,
+    TimeSecondVector, TimestampMicrosecondVector, TimestampMillisecondVector,
+    TimestampNanosecondVector, TimestampSecondVector, UInt32Vector, UInt64Vector, VectorRef,
 };
+use greptime_proto::v1;
+use greptime_proto::v1::column_data_type_extension::TypeExt;
 use greptime_proto::v1::ddl_request::Expr;
 use greptime_proto::v1::greptime_request::Request;
 use greptime_proto::v1::query_request::Query;
 use greptime_proto::v1::value::ValueData;
-use greptime_proto::v1::{self, DdlRequest, IntervalMonthDayNano, QueryRequest, Row, SemanticType};
+use greptime_proto::v1::{
+    ColumnDataTypeExtension, DdlRequest, DecimalTypeExtension, QueryRequest, Row, SemanticType,
+};
+use paste::paste;
 use snafu::prelude::*;
 
 use crate::error::{self, Result};
 use crate::v1::column::Values;
 use crate::v1::{Column, ColumnDataType, Value as GrpcValue};
 
-#[derive(Debug, PartialEq, Eq)]
-pub struct ColumnDataTypeWrapper(ColumnDataType);
+/// ColumnDataTypeWrapper is a wrapper of ColumnDataType and ColumnDataTypeExtension.
+/// It could be used to convert with ConcreteDataType.
+#[derive(Debug, PartialEq)]
+pub struct ColumnDataTypeWrapper {
+    datatype: ColumnDataType,
+    datatype_ext: Option<ColumnDataTypeExtension>,
+}
 
 impl ColumnDataTypeWrapper {
-    pub fn try_new(datatype: i32) -> Result<Self> {
-        let datatype = ColumnDataType::from_i32(datatype)
+    /// Try to create a ColumnDataTypeWrapper from i32(ColumnDataType) and ColumnDataTypeExtension.
+    pub fn try_new(datatype: i32, datatype_ext: Option<ColumnDataTypeExtension>) -> Result<Self> {
+        let datatype = ColumnDataType::try_from(datatype)
             .context(error::UnknownColumnDataTypeSnafu { datatype })?;
-        Ok(Self(datatype))
+        Ok(Self {
+            datatype,
+            datatype_ext,
+        })
     }
 
-    pub fn new(datatype: ColumnDataType) -> Self {
-        Self(datatype)
+    /// Create a ColumnDataTypeWrapper from ColumnDataType and ColumnDataTypeExtension.
+    pub fn new(datatype: ColumnDataType, datatype_ext: Option<ColumnDataTypeExtension>) -> Self {
+        Self {
+            datatype,
+            datatype_ext,
+        }
     }
 
+    /// Get the ColumnDataType.
     pub fn datatype(&self) -> ColumnDataType {
-        self.0
+        self.datatype
+    }
+
+    /// Get a tuple of ColumnDataType and ColumnDataTypeExtension.
+    pub fn to_parts(&self) -> (ColumnDataType, Option<ColumnDataTypeExtension>) {
+        (self.datatype, self.datatype_ext.clone())
     }
 }
 
 impl From<ColumnDataTypeWrapper> for ConcreteDataType {
-    fn from(datatype: ColumnDataTypeWrapper) -> Self {
-        match datatype.0 {
+    fn from(datatype_wrapper: ColumnDataTypeWrapper) -> Self {
+        match datatype_wrapper.datatype {
             ColumnDataType::Boolean => ConcreteDataType::boolean_datatype(),
             ColumnDataType::Int8 => ConcreteDataType::int8_datatype(),
             ColumnDataType::Int16 => ConcreteDataType::int16_datatype(),
@@ -109,6 +135,100 @@ impl From<ColumnDataTypeWrapper> for ConcreteDataType {
                 ConcreteDataType::duration_microsecond_datatype()
             }
             ColumnDataType::DurationNanosecond => ConcreteDataType::duration_nanosecond_datatype(),
+            ColumnDataType::Decimal128 => {
+                if let Some(TypeExt::DecimalType(d)) = datatype_wrapper
+                    .datatype_ext
+                    .as_ref()
+                    .and_then(|datatype_ext| datatype_ext.type_ext.as_ref())
+                {
+                    ConcreteDataType::decimal128_datatype(d.precision as u8, d.scale as i8)
+                } else {
+                    ConcreteDataType::decimal128_default_datatype()
+                }
+            }
+        }
+    }
+}
+
+/// This macro is used to generate datatype functions
+/// with lower style for ColumnDataTypeWrapper.
+///
+///
+/// For example: we can use `ColumnDataTypeWrapper::int8_datatype()`,
+/// to get a ColumnDataTypeWrapper with datatype `ColumnDataType::Int8`.
+macro_rules! impl_column_type_functions {
+    ($($Type: ident), +) => {
+        paste! {
+            impl ColumnDataTypeWrapper {
+                $(
+                    pub fn [<$Type:lower _datatype>]() -> ColumnDataTypeWrapper {
+                        ColumnDataTypeWrapper {
+                            datatype: ColumnDataType::$Type,
+                            datatype_ext: None,
+                        }
+                    }
+                )+
+            }
+        }
+    }
+}
+
+/// This macro is used to generate datatype functions
+/// with snake style for ColumnDataTypeWrapper.
+///
+///
+/// For example: we can use `ColumnDataTypeWrapper::duration_second_datatype()`,
+/// to get a ColumnDataTypeWrapper with datatype `ColumnDataType::DurationSecond`.
+macro_rules! impl_column_type_functions_with_snake {
+    ($($TypeName: ident), +) => {
+        paste!{
+            impl ColumnDataTypeWrapper {
+                $(
+                    pub fn [<$TypeName:snake _datatype>]() -> ColumnDataTypeWrapper {
+                        ColumnDataTypeWrapper {
+                            datatype: ColumnDataType::$TypeName,
+                            datatype_ext: None,
+                        }
+                    }
+                )+
+            }
+        }
+    };
+}
+
+impl_column_type_functions!(
+    Boolean, Uint8, Uint16, Uint32, Uint64, Int8, Int16, Int32, Int64, Float32, Float64, Binary,
+    Date, Datetime, String
+);
+
+impl_column_type_functions_with_snake!(
+    TimestampSecond,
+    TimestampMillisecond,
+    TimestampMicrosecond,
+    TimestampNanosecond,
+    TimeSecond,
+    TimeMillisecond,
+    TimeMicrosecond,
+    TimeNanosecond,
+    IntervalYearMonth,
+    IntervalDayTime,
+    IntervalMonthDayNano,
+    DurationSecond,
+    DurationMillisecond,
+    DurationMicrosecond,
+    DurationNanosecond
+);
+
+impl ColumnDataTypeWrapper {
+    pub fn decimal128_datatype(precision: i32, scale: i32) -> Self {
+        ColumnDataTypeWrapper {
+            datatype: ColumnDataType::Decimal128,
+            datatype_ext: Some(ColumnDataTypeExtension {
+                type_ext: Some(TypeExt::DecimalType(DecimalTypeExtension {
+                    precision,
+                    scale,
+                })),
+            }),
         }
     }
 }
@@ -117,7 +237,7 @@ impl TryFrom<ConcreteDataType> for ColumnDataTypeWrapper {
     type Error = error::Error;
 
     fn try_from(datatype: ConcreteDataType) -> Result<Self> {
-        let datatype = ColumnDataTypeWrapper(match datatype {
+        let column_datatype = match datatype {
             ConcreteDataType::Boolean(_) => ColumnDataType::Boolean,
             ConcreteDataType::Int8(_) => ColumnDataType::Int8,
             ConcreteDataType::Int16(_) => ColumnDataType::Int16,
@@ -156,13 +276,30 @@ impl TryFrom<ConcreteDataType> for ColumnDataTypeWrapper {
                 DurationType::Microsecond(_) => ColumnDataType::DurationMicrosecond,
                 DurationType::Nanosecond(_) => ColumnDataType::DurationNanosecond,
             },
+            ConcreteDataType::Decimal128(_) => ColumnDataType::Decimal128,
             ConcreteDataType::Null(_)
             | ConcreteDataType::List(_)
             | ConcreteDataType::Dictionary(_) => {
                 return error::IntoColumnDataTypeSnafu { from: datatype }.fail()
             }
-        });
-        Ok(datatype)
+        };
+        let datatype_extension = match column_datatype {
+            ColumnDataType::Decimal128 => {
+                datatype
+                    .as_decimal128()
+                    .map(|decimal_type| ColumnDataTypeExtension {
+                        type_ext: Some(TypeExt::DecimalType(DecimalTypeExtension {
+                            precision: decimal_type.precision() as i32,
+                            scale: decimal_type.scale() as i32,
+                        })),
+                    })
+            }
+            _ => None,
+        };
+        Ok(Self {
+            datatype: column_datatype,
+            datatype_ext: datatype_extension,
+        })
     }
 }
 
@@ -288,6 +425,10 @@ pub fn values_with_capacity(datatype: ColumnDataType, capacity: usize) -> Values
             duration_nanosecond_values: Vec::with_capacity(capacity),
             ..Default::default()
         },
+        ColumnDataType::Decimal128 => Values {
+            decimal128_values: Vec::with_capacity(capacity),
+            ..Default::default()
+        },
     }
 }
 
@@ -341,6 +482,7 @@ pub fn push_vals(column: &mut Column, origin_count: usize, vector: VectorRef) {
             TimeUnit::Microsecond => values.duration_microsecond_values.push(val.value()),
             TimeUnit::Nanosecond => values.duration_nanosecond_values.push(val.value()),
         },
+        Value::Decimal128(val) => values.decimal128_values.push(convert_to_pb_decimal128(val)),
         Value::List(_) => unreachable!(),
     });
     column.null_mask = null_mask.into_vec();
@@ -381,17 +523,26 @@ fn ddl_request_type(request: &DdlRequest) -> &'static str {
 }
 
 /// Converts an i128 value to google protobuf type [IntervalMonthDayNano].
-pub fn convert_i128_to_interval(v: i128) -> IntervalMonthDayNano {
+pub fn convert_i128_to_interval(v: i128) -> v1::IntervalMonthDayNano {
     let interval = Interval::from_i128(v);
     let (months, days, nanoseconds) = interval.to_month_day_nano();
-    IntervalMonthDayNano {
+    v1::IntervalMonthDayNano {
         months,
         days,
         nanoseconds,
     }
 }
 
-pub fn pb_value_to_value_ref(value: &v1::Value) -> ValueRef {
+/// Convert common decimal128 to grpc decimal128 without precision and scale.
+pub fn convert_to_pb_decimal128(v: Decimal128) -> v1::Decimal128 {
+    let (hi, lo) = v.split_value();
+    v1::Decimal128 { hi, lo }
+}
+
+pub fn pb_value_to_value_ref<'a>(
+    value: &'a v1::Value,
+    datatype_ext: &'a Option<ColumnDataTypeExtension>,
+) -> ValueRef<'a> {
     let Some(value) = &value.value_data else {
         return ValueRef::Null;
     };
@@ -426,9 +577,9 @@ pub fn pb_value_to_value_ref(value: &v1::Value) -> ValueRef {
         ValueData::TimeMillisecondValue(t) => ValueRef::Time(Time::new_millisecond(*t)),
         ValueData::TimeMicrosecondValue(t) => ValueRef::Time(Time::new_microsecond(*t)),
         ValueData::TimeNanosecondValue(t) => ValueRef::Time(Time::new_nanosecond(*t)),
-        ValueData::IntervalYearMonthValues(v) => ValueRef::Interval(Interval::from_i32(*v)),
-        ValueData::IntervalDayTimeValues(v) => ValueRef::Interval(Interval::from_i64(*v)),
-        ValueData::IntervalMonthDayNanoValues(v) => {
+        ValueData::IntervalYearMonthValue(v) => ValueRef::Interval(Interval::from_i32(*v)),
+        ValueData::IntervalDayTimeValue(v) => ValueRef::Interval(Interval::from_i64(*v)),
+        ValueData::IntervalMonthDayNanoValue(v) => {
             let interval = Interval::from_month_day_nano(v.months, v.days, v.nanoseconds);
             ValueRef::Interval(interval)
         }
@@ -436,6 +587,28 @@ pub fn pb_value_to_value_ref(value: &v1::Value) -> ValueRef {
         ValueData::DurationMillisecondValue(v) => ValueRef::Duration(Duration::new_millisecond(*v)),
         ValueData::DurationMicrosecondValue(v) => ValueRef::Duration(Duration::new_microsecond(*v)),
         ValueData::DurationNanosecondValue(v) => ValueRef::Duration(Duration::new_nanosecond(*v)),
+        ValueData::Decimal128Value(v) => {
+            // get precision and scale from datatype_extension
+            if let Some(TypeExt::DecimalType(d)) = datatype_ext
+                .as_ref()
+                .and_then(|column_ext| column_ext.type_ext.as_ref())
+            {
+                ValueRef::Decimal128(Decimal128::from_value_precision_scale(
+                    v.hi,
+                    v.lo,
+                    d.precision as u8,
+                    d.scale as i8,
+                ))
+            } else {
+                // If the precision and scale are not set, use the default value.
+                ValueRef::Decimal128(Decimal128::from_value_precision_scale(
+                    v.hi,
+                    v.lo,
+                    DECIMAL128_MAX_PRECISION,
+                    DECIMAL128_DEFAULT_SCALE,
+                ))
+            }
+        }
     }
 }
 
@@ -522,6 +695,11 @@ pub fn pb_values_to_vector_ref(data_type: &ConcreteDataType, values: Values) ->
                 values.duration_nanosecond_values,
             )),
         },
+        ConcreteDataType::Decimal128(d) => Arc::new(Decimal128Vector::from_values(
+            values.decimal128_values.iter().map(|x| {
+                Decimal128::from_value_precision_scale(x.hi, x.lo, d.precision(), d.scale()).into()
+            }),
+        )),
         ConcreteDataType::Null(_) | ConcreteDataType::List(_) | ConcreteDataType::Dictionary(_) => {
             unreachable!()
         }
@@ -692,6 +870,18 @@ pub fn pb_values_to_values(data_type: &ConcreteDataType, values: Values) -> Vec<
             .into_iter()
             .map(|v| Value::Duration(Duration::new_nanosecond(v)))
             .collect(),
+        ConcreteDataType::Decimal128(d) => values
+            .decimal128_values
+            .into_iter()
+            .map(|v| {
+                Value::Decimal128(Decimal128::from_value_precision_scale(
+                    v.hi,
+                    v.lo,
+                    d.precision(),
+                    d.scale(),
+                ))
+            })
+            .collect(),
         ConcreteDataType::Null(_) | ConcreteDataType::List(_) | ConcreteDataType::Dictionary(_) => {
             unreachable!()
         }
@@ -704,12 +894,14 @@ pub fn is_semantic_type_eq(type_value: i32, semantic_type: SemanticType) -> bool
 }
 
 /// Returns true if the pb type value is valid.
-pub fn is_column_type_value_eq(type_value: i32, expect_type: &ConcreteDataType) -> bool {
-    let Some(column_type) = ColumnDataType::from_i32(type_value) else {
-        return false;
-    };
-
-    is_column_type_eq(column_type, expect_type)
+pub fn is_column_type_value_eq(
+    type_value: i32,
+    type_extension: Option<ColumnDataTypeExtension>,
+    expect_type: &ConcreteDataType,
+) -> bool {
+    ColumnDataTypeWrapper::try_new(type_value, type_extension)
+        .map(|wrapper| ConcreteDataType::from(wrapper) == *expect_type)
+        .unwrap_or(false)
 }
 
 /// Convert value into proto's value.
@@ -791,13 +983,13 @@ pub fn to_proto_value(value: Value) -> Option<v1::Value> {
         },
         Value::Interval(v) => match v.unit() {
             IntervalUnit::YearMonth => v1::Value {
-                value_data: Some(ValueData::IntervalYearMonthValues(v.to_i32())),
+                value_data: Some(ValueData::IntervalYearMonthValue(v.to_i32())),
             },
             IntervalUnit::DayTime => v1::Value {
-                value_data: Some(ValueData::IntervalDayTimeValues(v.to_i64())),
+                value_data: Some(ValueData::IntervalDayTimeValue(v.to_i64())),
             },
             IntervalUnit::MonthDayNano => v1::Value {
-                value_data: Some(ValueData::IntervalMonthDayNanoValues(
+                value_data: Some(ValueData::IntervalMonthDayNanoValue(
                     convert_i128_to_interval(v.to_i128()),
                 )),
             },
@@ -816,13 +1008,16 @@ pub fn to_proto_value(value: Value) -> Option<v1::Value> {
                 value_data: Some(ValueData::DurationNanosecondValue(v.value())),
             },
         },
+        Value::Decimal128(v) => v1::Value {
+            value_data: Some(ValueData::Decimal128Value(convert_to_pb_decimal128(v))),
+        },
         Value::List(_) => return None,
     };
 
     Some(proto_value)
 }
 
-/// Returns the [ColumnDataType] of the value.
+/// Returns the [ColumnDataTypeWrapper] of the value.
 ///
 /// If value is null, returns `None`.
 pub fn proto_value_type(value: &v1::Value) -> Option<ColumnDataType> {
@@ -850,72 +1045,18 @@ pub fn proto_value_type(value: &v1::Value) -> Option<ColumnDataType> {
         ValueData::TimeMillisecondValue(_) => ColumnDataType::TimeMillisecond,
         ValueData::TimeMicrosecondValue(_) => ColumnDataType::TimeMicrosecond,
         ValueData::TimeNanosecondValue(_) => ColumnDataType::TimeNanosecond,
-        ValueData::IntervalYearMonthValues(_) => ColumnDataType::IntervalYearMonth,
-        ValueData::IntervalDayTimeValues(_) => ColumnDataType::IntervalDayTime,
-        ValueData::IntervalMonthDayNanoValues(_) => ColumnDataType::IntervalMonthDayNano,
+        ValueData::IntervalYearMonthValue(_) => ColumnDataType::IntervalYearMonth,
+        ValueData::IntervalDayTimeValue(_) => ColumnDataType::IntervalDayTime,
+        ValueData::IntervalMonthDayNanoValue(_) => ColumnDataType::IntervalMonthDayNano,
         ValueData::DurationSecondValue(_) => ColumnDataType::DurationSecond,
         ValueData::DurationMillisecondValue(_) => ColumnDataType::DurationMillisecond,
         ValueData::DurationMicrosecondValue(_) => ColumnDataType::DurationMicrosecond,
         ValueData::DurationNanosecondValue(_) => ColumnDataType::DurationNanosecond,
+        ValueData::Decimal128Value(_) => ColumnDataType::Decimal128,
     };
     Some(value_type)
 }
 
-/// Convert [ConcreteDataType] to [ColumnDataType].
-pub fn to_column_data_type(data_type: &ConcreteDataType) -> Option<ColumnDataType> {
-    let column_data_type = match data_type {
-        ConcreteDataType::Boolean(_) => ColumnDataType::Boolean,
-        ConcreteDataType::Int8(_) => ColumnDataType::Int8,
-        ConcreteDataType::Int16(_) => ColumnDataType::Int16,
-        ConcreteDataType::Int32(_) => ColumnDataType::Int32,
-        ConcreteDataType::Int64(_) => ColumnDataType::Int64,
-        ConcreteDataType::UInt8(_) => ColumnDataType::Uint8,
-        ConcreteDataType::UInt16(_) => ColumnDataType::Uint16,
-        ConcreteDataType::UInt32(_) => ColumnDataType::Uint32,
-        ConcreteDataType::UInt64(_) => ColumnDataType::Uint64,
-        ConcreteDataType::Float32(_) => ColumnDataType::Float32,
-        ConcreteDataType::Float64(_) => ColumnDataType::Float64,
-        ConcreteDataType::Binary(_) => ColumnDataType::Binary,
-        ConcreteDataType::String(_) => ColumnDataType::String,
-        ConcreteDataType::Date(_) => ColumnDataType::Date,
-        ConcreteDataType::DateTime(_) => ColumnDataType::Datetime,
-        ConcreteDataType::Timestamp(TimestampType::Second(_)) => ColumnDataType::TimestampSecond,
-        ConcreteDataType::Timestamp(TimestampType::Millisecond(_)) => {
-            ColumnDataType::TimestampMillisecond
-        }
-        ConcreteDataType::Timestamp(TimestampType::Microsecond(_)) => {
-            ColumnDataType::TimestampMicrosecond
-        }
-        ConcreteDataType::Timestamp(TimestampType::Nanosecond(_)) => {
-            ColumnDataType::TimestampNanosecond
-        }
-        ConcreteDataType::Time(TimeType::Second(_)) => ColumnDataType::TimeSecond,
-        ConcreteDataType::Time(TimeType::Millisecond(_)) => ColumnDataType::TimeMillisecond,
-        ConcreteDataType::Time(TimeType::Microsecond(_)) => ColumnDataType::TimeMicrosecond,
-        ConcreteDataType::Time(TimeType::Nanosecond(_)) => ColumnDataType::TimeNanosecond,
-        ConcreteDataType::Duration(DurationType::Second(_)) => ColumnDataType::DurationSecond,
-        ConcreteDataType::Duration(DurationType::Millisecond(_)) => {
-            ColumnDataType::DurationMillisecond
-        }
-        ConcreteDataType::Duration(DurationType::Microsecond(_)) => {
-            ColumnDataType::DurationMicrosecond
-        }
-        ConcreteDataType::Duration(DurationType::Nanosecond(_)) => {
-            ColumnDataType::DurationNanosecond
-        }
-        ConcreteDataType::Interval(IntervalType::YearMonth(_)) => ColumnDataType::IntervalYearMonth,
-        ConcreteDataType::Interval(IntervalType::MonthDayNano(_)) => {
-            ColumnDataType::IntervalMonthDayNano
-        }
-        ConcreteDataType::Interval(IntervalType::DayTime(_)) => ColumnDataType::IntervalDayTime,
-        ConcreteDataType::Null(_) | ConcreteDataType::List(_) | ConcreteDataType::Dictionary(_) => {
-            return None
-        }
-    };
-
-    Some(column_data_type)
-}
-
 pub fn vectors_to_rows<'a>(
     columns: impl Iterator<Item = &'a VectorRef>,
     row_count: usize,
@@ -962,10 +1103,10 @@ pub fn value_to_grpc_value(value: Value) -> GrpcValue {
                 TimeUnit::Nanosecond => ValueData::TimeNanosecondValue(v.value()),
             }),
             Value::Interval(v) => Some(match v.unit() {
-                IntervalUnit::YearMonth => ValueData::IntervalYearMonthValues(v.to_i32()),
-                IntervalUnit::DayTime => ValueData::IntervalDayTimeValues(v.to_i64()),
+                IntervalUnit::YearMonth => ValueData::IntervalYearMonthValue(v.to_i32()),
+                IntervalUnit::DayTime => ValueData::IntervalDayTimeValue(v.to_i64()),
                 IntervalUnit::MonthDayNano => {
-                    ValueData::IntervalMonthDayNanoValues(convert_i128_to_interval(v.to_i128()))
+                    ValueData::IntervalMonthDayNanoValue(convert_i128_to_interval(v.to_i128()))
                 }
             }),
             Value::Duration(v) => Some(match v.unit() {
@@ -974,20 +1115,12 @@ pub fn value_to_grpc_value(value: Value) -> GrpcValue {
                 TimeUnit::Microsecond => ValueData::DurationMicrosecondValue(v.value()),
                 TimeUnit::Nanosecond => ValueData::DurationNanosecondValue(v.value()),
             }),
+            Value::Decimal128(v) => Some(ValueData::Decimal128Value(convert_to_pb_decimal128(v))),
             Value::List(_) => unreachable!(),
         },
     }
 }
 
-/// Returns true if the column type is equal to expected type.
-fn is_column_type_eq(column_type: ColumnDataType, expect_type: &ConcreteDataType) -> bool {
-    if let Some(expect) = to_column_data_type(expect_type) {
-        column_type == expect
-    } else {
-        false
-    }
-}
-
 #[cfg(test)]
 mod tests {
     use std::sync::Arc;
@@ -1081,189 +1214,204 @@ mod tests {
         let values = values_with_capacity(ColumnDataType::DurationMillisecond, 2);
         let values = values.duration_millisecond_values;
         assert_eq!(2, values.capacity());
+
+        let values = values_with_capacity(ColumnDataType::Decimal128, 2);
+        let values = values.decimal128_values;
+        assert_eq!(2, values.capacity());
     }
 
     #[test]
     fn test_concrete_datatype_from_column_datatype() {
         assert_eq!(
             ConcreteDataType::boolean_datatype(),
-            ColumnDataTypeWrapper(ColumnDataType::Boolean).into()
+            ColumnDataTypeWrapper::boolean_datatype().into()
         );
         assert_eq!(
             ConcreteDataType::int8_datatype(),
-            ColumnDataTypeWrapper(ColumnDataType::Int8).into()
+            ColumnDataTypeWrapper::int8_datatype().into()
         );
         assert_eq!(
             ConcreteDataType::int16_datatype(),
-            ColumnDataTypeWrapper(ColumnDataType::Int16).into()
+            ColumnDataTypeWrapper::int16_datatype().into()
         );
         assert_eq!(
             ConcreteDataType::int32_datatype(),
-            ColumnDataTypeWrapper(ColumnDataType::Int32).into()
+            ColumnDataTypeWrapper::int32_datatype().into()
         );
         assert_eq!(
             ConcreteDataType::int64_datatype(),
-            ColumnDataTypeWrapper(ColumnDataType::Int64).into()
+            ColumnDataTypeWrapper::int64_datatype().into()
         );
         assert_eq!(
             ConcreteDataType::uint8_datatype(),
-            ColumnDataTypeWrapper(ColumnDataType::Uint8).into()
+            ColumnDataTypeWrapper::uint8_datatype().into()
         );
         assert_eq!(
             ConcreteDataType::uint16_datatype(),
-            ColumnDataTypeWrapper(ColumnDataType::Uint16).into()
+            ColumnDataTypeWrapper::uint16_datatype().into()
         );
         assert_eq!(
             ConcreteDataType::uint32_datatype(),
-            ColumnDataTypeWrapper(ColumnDataType::Uint32).into()
+            ColumnDataTypeWrapper::uint32_datatype().into()
         );
         assert_eq!(
             ConcreteDataType::uint64_datatype(),
-            ColumnDataTypeWrapper(ColumnDataType::Uint64).into()
+            ColumnDataTypeWrapper::uint64_datatype().into()
         );
         assert_eq!(
             ConcreteDataType::float32_datatype(),
-            ColumnDataTypeWrapper(ColumnDataType::Float32).into()
+            ColumnDataTypeWrapper::float32_datatype().into()
         );
         assert_eq!(
             ConcreteDataType::float64_datatype(),
-            ColumnDataTypeWrapper(ColumnDataType::Float64).into()
+            ColumnDataTypeWrapper::float64_datatype().into()
         );
         assert_eq!(
             ConcreteDataType::binary_datatype(),
-            ColumnDataTypeWrapper(ColumnDataType::Binary).into()
+            ColumnDataTypeWrapper::binary_datatype().into()
         );
         assert_eq!(
             ConcreteDataType::string_datatype(),
-            ColumnDataTypeWrapper(ColumnDataType::String).into()
+            ColumnDataTypeWrapper::string_datatype().into()
         );
         assert_eq!(
             ConcreteDataType::date_datatype(),
-            ColumnDataTypeWrapper(ColumnDataType::Date).into()
+            ColumnDataTypeWrapper::date_datatype().into()
         );
         assert_eq!(
             ConcreteDataType::datetime_datatype(),
-            ColumnDataTypeWrapper(ColumnDataType::Datetime).into()
+            ColumnDataTypeWrapper::datetime_datatype().into()
         );
         assert_eq!(
             ConcreteDataType::timestamp_millisecond_datatype(),
-            ColumnDataTypeWrapper(ColumnDataType::TimestampMillisecond).into()
+            ColumnDataTypeWrapper::timestamp_millisecond_datatype().into()
         );
         assert_eq!(
             ConcreteDataType::time_datatype(TimeUnit::Millisecond),
-            ColumnDataTypeWrapper(ColumnDataType::TimeMillisecond).into()
+            ColumnDataTypeWrapper::time_millisecond_datatype().into()
         );
         assert_eq!(
             ConcreteDataType::interval_datatype(IntervalUnit::DayTime),
-            ColumnDataTypeWrapper(ColumnDataType::IntervalDayTime).into()
+            ColumnDataTypeWrapper::interval_day_time_datatype().into()
         );
         assert_eq!(
             ConcreteDataType::interval_datatype(IntervalUnit::YearMonth),
-            ColumnDataTypeWrapper(ColumnDataType::IntervalYearMonth).into()
+            ColumnDataTypeWrapper::interval_year_month_datatype().into()
         );
         assert_eq!(
             ConcreteDataType::interval_datatype(IntervalUnit::MonthDayNano),
-            ColumnDataTypeWrapper(ColumnDataType::IntervalMonthDayNano).into()
+            ColumnDataTypeWrapper::interval_month_day_nano_datatype().into()
         );
         assert_eq!(
             ConcreteDataType::duration_millisecond_datatype(),
-            ColumnDataTypeWrapper(ColumnDataType::DurationMillisecond).into()
+            ColumnDataTypeWrapper::duration_millisecond_datatype().into()
+        );
+        assert_eq!(
+            ConcreteDataType::decimal128_datatype(10, 2),
+            ColumnDataTypeWrapper::decimal128_datatype(10, 2).into()
         )
     }
 
     #[test]
     fn test_column_datatype_from_concrete_datatype() {
         assert_eq!(
-            ColumnDataTypeWrapper(ColumnDataType::Boolean),
+            ColumnDataTypeWrapper::boolean_datatype(),
             ConcreteDataType::boolean_datatype().try_into().unwrap()
         );
         assert_eq!(
-            ColumnDataTypeWrapper(ColumnDataType::Int8),
+            ColumnDataTypeWrapper::int8_datatype(),
             ConcreteDataType::int8_datatype().try_into().unwrap()
         );
         assert_eq!(
-            ColumnDataTypeWrapper(ColumnDataType::Int16),
+            ColumnDataTypeWrapper::int16_datatype(),
             ConcreteDataType::int16_datatype().try_into().unwrap()
         );
         assert_eq!(
-            ColumnDataTypeWrapper(ColumnDataType::Int32),
+            ColumnDataTypeWrapper::int32_datatype(),
             ConcreteDataType::int32_datatype().try_into().unwrap()
         );
         assert_eq!(
-            ColumnDataTypeWrapper(ColumnDataType::Int64),
+            ColumnDataTypeWrapper::int64_datatype(),
             ConcreteDataType::int64_datatype().try_into().unwrap()
         );
         assert_eq!(
-            ColumnDataTypeWrapper(ColumnDataType::Uint8),
+            ColumnDataTypeWrapper::uint8_datatype(),
             ConcreteDataType::uint8_datatype().try_into().unwrap()
         );
         assert_eq!(
-            ColumnDataTypeWrapper(ColumnDataType::Uint16),
+            ColumnDataTypeWrapper::uint16_datatype(),
             ConcreteDataType::uint16_datatype().try_into().unwrap()
         );
         assert_eq!(
-            ColumnDataTypeWrapper(ColumnDataType::Uint32),
+            ColumnDataTypeWrapper::uint32_datatype(),
             ConcreteDataType::uint32_datatype().try_into().unwrap()
         );
         assert_eq!(
-            ColumnDataTypeWrapper(ColumnDataType::Uint64),
+            ColumnDataTypeWrapper::uint64_datatype(),
             ConcreteDataType::uint64_datatype().try_into().unwrap()
         );
         assert_eq!(
-            ColumnDataTypeWrapper(ColumnDataType::Float32),
+            ColumnDataTypeWrapper::float32_datatype(),
             ConcreteDataType::float32_datatype().try_into().unwrap()
         );
         assert_eq!(
-            ColumnDataTypeWrapper(ColumnDataType::Float64),
+            ColumnDataTypeWrapper::float64_datatype(),
             ConcreteDataType::float64_datatype().try_into().unwrap()
         );
         assert_eq!(
-            ColumnDataTypeWrapper(ColumnDataType::Binary),
+            ColumnDataTypeWrapper::binary_datatype(),
             ConcreteDataType::binary_datatype().try_into().unwrap()
         );
         assert_eq!(
-            ColumnDataTypeWrapper(ColumnDataType::String),
+            ColumnDataTypeWrapper::string_datatype(),
             ConcreteDataType::string_datatype().try_into().unwrap()
         );
         assert_eq!(
-            ColumnDataTypeWrapper(ColumnDataType::Date),
+            ColumnDataTypeWrapper::date_datatype(),
             ConcreteDataType::date_datatype().try_into().unwrap()
         );
         assert_eq!(
-            ColumnDataTypeWrapper(ColumnDataType::Datetime),
+            ColumnDataTypeWrapper::datetime_datatype(),
             ConcreteDataType::datetime_datatype().try_into().unwrap()
         );
         assert_eq!(
-            ColumnDataTypeWrapper(ColumnDataType::TimestampMillisecond),
+            ColumnDataTypeWrapper::timestamp_millisecond_datatype(),
             ConcreteDataType::timestamp_millisecond_datatype()
                 .try_into()
                 .unwrap()
         );
         assert_eq!(
-            ColumnDataTypeWrapper(ColumnDataType::IntervalYearMonth),
+            ColumnDataTypeWrapper::interval_year_month_datatype(),
             ConcreteDataType::interval_datatype(IntervalUnit::YearMonth)
                 .try_into()
                 .unwrap()
         );
         assert_eq!(
-            ColumnDataTypeWrapper(ColumnDataType::IntervalDayTime),
+            ColumnDataTypeWrapper::interval_day_time_datatype(),
             ConcreteDataType::interval_datatype(IntervalUnit::DayTime)
                 .try_into()
                 .unwrap()
         );
         assert_eq!(
-            ColumnDataTypeWrapper(ColumnDataType::IntervalMonthDayNano),
+            ColumnDataTypeWrapper::interval_month_day_nano_datatype(),
             ConcreteDataType::interval_datatype(IntervalUnit::MonthDayNano)
                 .try_into()
                 .unwrap()
         );
         assert_eq!(
-            ColumnDataTypeWrapper(ColumnDataType::DurationMillisecond),
+            ColumnDataTypeWrapper::duration_millisecond_datatype(),
             ConcreteDataType::duration_millisecond_datatype()
                 .try_into()
                 .unwrap()
         );
 
+        assert_eq!(
+            ColumnDataTypeWrapper::decimal128_datatype(10, 2),
+            ConcreteDataType::decimal128_datatype(10, 2)
+                .try_into()
+                .unwrap()
+        );
+
         let result: Result<ColumnDataTypeWrapper> = ConcreteDataType::null_datatype().try_into();
         assert!(result.is_err());
         assert_eq!(
@@ -1290,6 +1438,7 @@ mod tests {
             }),
             null_mask: vec![],
             datatype: 0,
+            ..Default::default()
         };
 
         let vector = Arc::new(TimestampNanosecondVector::from_vec(vec![1, 2, 3]));
@@ -1331,6 +1480,7 @@ mod tests {
             }),
             null_mask: vec![],
             datatype: 0,
+            ..Default::default()
         };
 
         let vector = Arc::new(TimeNanosecondVector::from_vec(vec![1, 2, 3]));
@@ -1372,6 +1522,7 @@ mod tests {
             }),
             null_mask: vec![],
             datatype: 0,
+            ..Default::default()
         };
 
         let vector = Arc::new(IntervalYearMonthVector::from_vec(vec![1, 2, 3]));
@@ -1416,6 +1567,7 @@ mod tests {
             }),
             null_mask: vec![],
             datatype: 0,
+            ..Default::default()
         };
 
         let vector = Arc::new(DurationNanosecondVector::from_vec(vec![1, 2, 3]));
@@ -1460,6 +1612,7 @@ mod tests {
             }),
             null_mask: vec![2],
             datatype: ColumnDataType::Boolean as i32,
+            ..Default::default()
         };
         let row_count = 4;
 
@@ -1617,17 +1770,17 @@ mod tests {
             &ConcreteDataType::Interval(IntervalType::MonthDayNano(IntervalMonthDayNanoType)),
             Values {
                 interval_month_day_nano_values: vec![
-                    IntervalMonthDayNano {
+                    v1::IntervalMonthDayNano {
                         months: 1,
                         days: 2,
                         nanoseconds: 3,
                     },
-                    IntervalMonthDayNano {
+                    v1::IntervalMonthDayNano {
                         months: 5,
                         days: 6,
                         nanoseconds: 7,
                     },
-                    IntervalMonthDayNano {
+                    v1::IntervalMonthDayNano {
                         months: 9,
                         days: 10,
                         nanoseconds: 11,
@@ -1859,4 +2012,33 @@ mod tests {
         assert_eq!(values[6], ValueData::DateValue(30));
         assert_eq!(values[7], ValueData::StringValue("c".to_string()));
     }
+
+    #[test]
+    fn test_is_column_type_value_eq() {
+        // test column type eq
+        let column1 = Column {
+            column_name: "test".to_string(),
+            semantic_type: 0,
+            values: Some(Values {
+                bool_values: vec![false, true, true],
+                ..Default::default()
+            }),
+            null_mask: vec![2],
+            datatype: ColumnDataType::Boolean as i32,
+            datatype_extension: None,
+        };
+        assert!(is_column_type_value_eq(
+            column1.datatype,
+            column1.datatype_extension,
+            &ConcreteDataType::boolean_datatype(),
+        ));
+    }
+
+    #[test]
+    fn test_convert_to_pb_decimal128() {
+        let decimal = Decimal128::new(123, 3, 1);
+        let pb_decimal = convert_to_pb_decimal128(decimal);
+        assert_eq!(pb_decimal.lo, 123);
+        assert_eq!(pb_decimal.hi, 0);
+    }
 }

src/api/src/v1/column_def.rs

@@ -22,7 +22,10 @@ use crate::helper::ColumnDataTypeWrapper;
 use crate::v1::ColumnDef;
 
 pub fn try_as_column_schema(column_def: &ColumnDef) -> Result<ColumnSchema> {
-    let data_type = ColumnDataTypeWrapper::try_new(column_def.data_type)?;
+    let data_type = ColumnDataTypeWrapper::try_new(
+        column_def.data_type,
+        column_def.datatype_extension.clone(),
+    )?;
 
     let constraint = if column_def.default_constraint.is_empty() {
         None

src/auth/src/common.rs

@@ -4,7 +4,7 @@
 // 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
+//     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,

src/auth/src/error.rs

@@ -4,7 +4,7 @@
 // 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
+//     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,

src/auth/src/lib.rs

@@ -4,7 +4,7 @@
 // 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
+//     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,

src/auth/src/permission.rs

@@ -4,7 +4,7 @@
 // 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
+//     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,

src/auth/src/tests.rs

@@ -4,13 +4,14 @@
 // 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
+//     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 secrecy::ExposeSecret;
 
 use crate::error::{

src/auth/src/user_info.rs

@@ -4,7 +4,7 @@
 // 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
+//     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,

src/auth/src/user_provider.rs

@@ -4,7 +4,7 @@
 // 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
+//     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,

src/auth/tests/mod.rs

@@ -4,7 +4,7 @@
 // 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
+//     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,

src/catalog/Cargo.toml

@@ -8,46 +8,47 @@ license.workspace = true
 testing = []
 
 [dependencies]
-api = { workspace = true }
+api.workspace = true
 arc-swap = "1.0"
 arrow-schema.workspace = true
 async-stream.workspace = true
 async-trait = "0.1"
-common-catalog = { workspace = true }
-common-error = { workspace = true }
-common-grpc = { workspace = true }
-common-macro = { workspace = true }
-common-meta = { workspace = true }
-common-query = { workspace = true }
-common-recordbatch = { workspace = true }
-common-runtime = { workspace = true }
-common-telemetry = { workspace = true }
-common-time = { workspace = true }
-dashmap = "5.4"
+build-data = "0.1"
+common-catalog.workspace = true
+common-error.workspace = true
+common-grpc.workspace = true
+common-macro.workspace = true
+common-meta.workspace = true
+common-query.workspace = true
+common-recordbatch.workspace = true
+common-runtime.workspace = true
+common-telemetry.workspace = true
+common-time.workspace = true
+dashmap.workspace = true
 datafusion.workspace = true
-datatypes = { workspace = true }
+datatypes.workspace = true
 futures = "0.3"
 futures-util.workspace = true
 lazy_static.workspace = true
-meta-client = { workspace = true }
-metrics.workspace = true
+meta-client.workspace = true
 moka = { workspace = true, features = ["future"] }
 parking_lot = "0.12"
 partition.workspace = true
+paste = "1.0"
+prometheus.workspace = true
 regex.workspace = true
 serde.workspace = true
-serde_json = "1.0"
-session = { workspace = true }
-snafu = { version = "0.7", features = ["backtraces"] }
-store-api = { workspace = true }
-table = { workspace = true }
+serde_json.workspace = true
+session.workspace = true
+snafu.workspace = true
+store-api.workspace = true
+table.workspace = true
 tokio.workspace = true
 
 [dev-dependencies]
 catalog = { workspace = true, features = ["testing"] }
 chrono.workspace = true
-common-test-util = { workspace = true }
-log-store = { workspace = true }
-object-store = { workspace = true }
-storage = { workspace = true }
+common-test-util.workspace = true
+log-store.workspace = true
+object-store.workspace = true
 tokio.workspace = true

src/catalog/src/error.rs

@@ -180,7 +180,7 @@ pub enum Error {
         source: table::error::Error,
     },
 
-    #[snafu(display(""))]
+    #[snafu(display("Internal error"))]
     Internal {
         location: Location,
         source: BoxedError,
@@ -216,7 +216,7 @@ pub enum Error {
     #[snafu(display("Illegal access to catalog: {} and schema: {}", catalog, schema))]
     QueryAccessDenied { catalog: String, schema: String },
 
-    #[snafu(display(""))]
+    #[snafu(display("DataFusion error"))]
     Datafusion {
         #[snafu(source)]
         error: DataFusionError,

src/catalog/src/information_schema.rs

@@ -13,16 +13,20 @@
 // limitations under the License.
 
 mod columns;
+mod memory_table;
+mod table_names;
 mod tables;
 
 use std::collections::HashMap;
 use std::sync::{Arc, Weak};
 
-use common_catalog::consts::INFORMATION_SCHEMA_NAME;
+use common_catalog::consts::{self, INFORMATION_SCHEMA_NAME};
 use common_error::ext::BoxedError;
-use common_recordbatch::{RecordBatchStreamAdaptor, SendableRecordBatchStream};
+use common_recordbatch::{RecordBatchStreamWrapper, SendableRecordBatchStream};
 use datatypes::schema::SchemaRef;
 use futures_util::StreamExt;
+use lazy_static::lazy_static;
+use paste::paste;
 use snafu::ResultExt;
 use store_api::data_source::DataSource;
 use store_api::storage::{ScanRequest, TableId};
@@ -32,43 +36,102 @@ use table::metadata::{
 };
 use table::thin_table::{ThinTable, ThinTableAdapter};
 use table::TableRef;
+pub use table_names::*;
 
 use self::columns::InformationSchemaColumns;
 use crate::error::Result;
+use crate::information_schema::memory_table::{get_schema_columns, MemoryTable};
 use crate::information_schema::tables::InformationSchemaTables;
 use crate::CatalogManager;
 
-pub const TABLES: &str = "tables";
-pub const COLUMNS: &str = "columns";
+lazy_static! {
+    // Memory tables in `information_schema`.
+    static ref MEMORY_TABLES: &'static [&'static str] = &[
+        ENGINES,
+        COLUMN_PRIVILEGES,
+        COLUMN_STATISTICS,
+        BUILD_INFO,
+    ];
+}
 
+macro_rules! setup_memory_table {
+    ($name: expr) => {
+        paste! {
+            {
+                let (schema, columns) = get_schema_columns($name);
+                Some(Arc::new(MemoryTable::new(
+                    consts::[<INFORMATION_SCHEMA_ $name  _TABLE_ID>],
+                    $name,
+                    schema,
+                    columns
+                )) as _)
+            }
+        }
+    };
+}
+
+/// The `information_schema` tables info provider.
 pub struct InformationSchemaProvider {
     catalog_name: String,
     catalog_manager: Weak<dyn CatalogManager>,
+    tables: HashMap<String, TableRef>,
 }
 
 impl InformationSchemaProvider {
     pub fn new(catalog_name: String, catalog_manager: Weak<dyn CatalogManager>) -> Self {
-        Self {
+        let mut provider = Self {
             catalog_name,
             catalog_manager,
-        }
+            tables: HashMap::new(),
+        };
+
+        provider.build_tables();
+
+        provider
     }
 
-    /// Build a map of [TableRef] in information schema.
-    /// Including `tables` and `columns`.
-    pub fn build(
-        catalog_name: String,
-        catalog_manager: Weak<dyn CatalogManager>,
-    ) -> HashMap<String, TableRef> {
-        let provider = Self::new(catalog_name, catalog_manager);
+    /// Returns table names in the order of table id.
+    pub fn table_names(&self) -> Vec<String> {
+        let mut tables = self.tables.values().clone().collect::<Vec<_>>();
 
-        let mut schema = HashMap::new();
-        schema.insert(TABLES.to_owned(), provider.table(TABLES).unwrap());
-        schema.insert(COLUMNS.to_owned(), provider.table(COLUMNS).unwrap());
-        schema
+        tables.sort_by(|t1, t2| {
+            t1.table_info()
+                .table_id()
+                .partial_cmp(&t2.table_info().table_id())
+                .unwrap()
+        });
+        tables
+            .into_iter()
+            .map(|t| t.table_info().name.clone())
+            .collect()
     }
 
+    /// Returns a map of [TableRef] in information schema.
+    pub fn tables(&self) -> &HashMap<String, TableRef> {
+        assert!(!self.tables.is_empty());
+
+        &self.tables
+    }
+
+    /// Returns the [TableRef] by table name.
     pub fn table(&self, name: &str) -> Option<TableRef> {
+        self.tables.get(name).cloned()
+    }
+
+    fn build_tables(&mut self) {
+        let mut tables = HashMap::new();
+        tables.insert(TABLES.to_string(), self.build_table(TABLES).unwrap());
+        tables.insert(COLUMNS.to_string(), self.build_table(COLUMNS).unwrap());
+
+        // Add memory tables
+        for name in MEMORY_TABLES.iter() {
+            tables.insert((*name).to_string(), self.build_table(name).unwrap());
+        }
+
+        self.tables = tables;
+    }
+
+    fn build_table(&self, name: &str) -> Option<TableRef> {
         self.information_table(name).map(|table| {
             let table_info = Self::table_info(self.catalog_name.clone(), &table);
             let filter_pushdown = FilterPushDownType::Unsupported;
@@ -89,6 +152,10 @@ impl InformationSchemaProvider {
                 self.catalog_name.clone(),
                 self.catalog_manager.clone(),
             )) as _),
+            ENGINES => setup_memory_table!(ENGINES),
+            COLUMN_PRIVILEGES => setup_memory_table!(COLUMN_PRIVILEGES),
+            COLUMN_STATISTICS => setup_memory_table!(COLUMN_STATISTICS),
+            BUILD_INFO => setup_memory_table!(BUILD_INFO),
             _ => None,
         }
     }
@@ -102,9 +169,9 @@ impl InformationSchemaProvider {
             .unwrap();
         let table_info = TableInfoBuilder::default()
             .table_id(table.table_id())
-            .name(table.table_name().to_owned())
+            .name(table.table_name().to_string())
             .catalog_name(catalog_name)
-            .schema_name(INFORMATION_SCHEMA_NAME.to_owned())
+            .schema_name(INFORMATION_SCHEMA_NAME.to_string())
             .meta(table_meta)
             .table_type(table.table_type())
             .build()
@@ -171,11 +238,12 @@ impl DataSource for InformationTableDataSource {
                 None => batch,
             });
 
-        let stream = RecordBatchStreamAdaptor {
+        let stream = RecordBatchStreamWrapper {
             schema: projected_schema,
             stream: Box::pin(stream),
             output_ordering: None,
         };
+
         Ok(Box::pin(stream))
     }
 }

src/catalog/src/information_schema/columns.rs

@@ -16,8 +16,8 @@ use std::sync::{Arc, Weak};
 
 use arrow_schema::SchemaRef as ArrowSchemaRef;
 use common_catalog::consts::{
-    INFORMATION_SCHEMA_COLUMNS_TABLE_ID, INFORMATION_SCHEMA_NAME, SEMANTIC_TYPE_FIELD,
-    SEMANTIC_TYPE_PRIMARY_KEY, SEMANTIC_TYPE_TIME_INDEX,
+    INFORMATION_SCHEMA_COLUMNS_TABLE_ID, SEMANTIC_TYPE_FIELD, SEMANTIC_TYPE_PRIMARY_KEY,
+    SEMANTIC_TYPE_TIME_INDEX,
 };
 use common_error::ext::BoxedError;
 use common_query::physical_plan::TaskContext;
@@ -33,8 +33,7 @@ use datatypes::vectors::{StringVectorBuilder, VectorRef};
 use snafu::{OptionExt, ResultExt};
 use store_api::storage::TableId;
 
-use super::tables::InformationSchemaTables;
-use super::{InformationTable, COLUMNS, TABLES};
+use super::{InformationTable, COLUMNS};
 use crate::error::{
     CreateRecordBatchSnafu, InternalSnafu, Result, UpgradeWeakCatalogManagerRefSnafu,
 };
@@ -102,7 +101,7 @@ impl InformationTable for InformationSchemaColumns {
             schema,
             futures::stream::once(async move {
                 builder
-                    .make_tables()
+                    .make_columns()
                     .await
                     .map(|x| x.into_df_record_batch())
                     .map_err(Into::into)
@@ -148,8 +147,8 @@ impl InformationSchemaColumnsBuilder {
         }
     }
 
-    /// Construct the `information_schema.tables` virtual table
-    async fn make_tables(&mut self) -> Result<RecordBatch> {
+    /// Construct the `information_schema.columns` virtual table
+    async fn make_columns(&mut self) -> Result<RecordBatch> {
         let catalog_name = self.catalog_name.clone();
         let catalog_manager = self
             .catalog_manager
@@ -163,48 +162,38 @@ impl InformationSchemaColumnsBuilder {
             {
                 continue;
             }
+
             for table_name in catalog_manager
                 .table_names(&catalog_name, &schema_name)
                 .await?
             {
-                let (keys, schema) = if let Some(table) = catalog_manager
+                if let Some(table) = catalog_manager
                     .table(&catalog_name, &schema_name, &table_name)
                     .await?
                 {
                     let keys = &table.table_info().meta.primary_key_indices;
                     let schema = table.schema();
-                    (keys.clone(), schema)
-                } else {
-                    // TODO: this specific branch is only a workaround for FrontendCatalogManager.
-                    if schema_name == INFORMATION_SCHEMA_NAME {
-                        if table_name == COLUMNS {
-                            (vec![], InformationSchemaColumns::schema())
-                        } else if table_name == TABLES {
-                            (vec![], InformationSchemaTables::schema())
-                        } else {
-                            continue;
-                        }
-                    } else {
-                        continue;
-                    }
-                };
 
-                for (idx, column) in schema.column_schemas().iter().enumerate() {
-                    let semantic_type = if column.is_time_index() {
-                        SEMANTIC_TYPE_TIME_INDEX
-                    } else if keys.contains(&idx) {
-                        SEMANTIC_TYPE_PRIMARY_KEY
-                    } else {
-                        SEMANTIC_TYPE_FIELD
-                    };
-                    self.add_column(
-                        &catalog_name,
-                        &schema_name,
-                        &table_name,
-                        &column.name,
-                        column.data_type.name(),
-                        semantic_type,
-                    );
+                    for (idx, column) in schema.column_schemas().iter().enumerate() {
+                        let semantic_type = if column.is_time_index() {
+                            SEMANTIC_TYPE_TIME_INDEX
+                        } else if keys.contains(&idx) {
+                            SEMANTIC_TYPE_PRIMARY_KEY
+                        } else {
+                            SEMANTIC_TYPE_FIELD
+                        };
+
+                        self.add_column(
+                            &catalog_name,
+                            &schema_name,
+                            &table_name,
+                            &column.name,
+                            &column.data_type.name(),
+                            semantic_type,
+                        );
+                    }
+                } else {
+                    unreachable!();
                 }
             }
         }
@@ -238,6 +227,7 @@ impl InformationSchemaColumnsBuilder {
             Arc::new(self.data_types.finish()),
             Arc::new(self.semantic_types.finish()),
         ];
+
         RecordBatch::new(self.schema.clone(), columns).context(CreateRecordBatchSnafu)
     }
 }
@@ -254,7 +244,7 @@ impl DfPartitionStream for InformationSchemaColumns {
             schema,
             futures::stream::once(async move {
                 builder
-                    .make_tables()
+                    .make_columns()
                     .await
                     .map(|x| x.into_df_record_batch())
                     .map_err(Into::into)

src/catalog/src/information_schema/memory_table.rs

@@ -0,0 +1,214 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+mod tables;
+use std::sync::Arc;
+
+use arrow_schema::SchemaRef as ArrowSchemaRef;
+use common_error::ext::BoxedError;
+use common_query::physical_plan::TaskContext;
+use common_recordbatch::adapter::RecordBatchStreamAdapter;
+use common_recordbatch::{RecordBatch, SendableRecordBatchStream};
+use datafusion::physical_plan::stream::RecordBatchStreamAdapter as DfRecordBatchStreamAdapter;
+use datafusion::physical_plan::streaming::PartitionStream as DfPartitionStream;
+use datafusion::physical_plan::SendableRecordBatchStream as DfSendableRecordBatchStream;
+use datatypes::schema::SchemaRef;
+use datatypes::vectors::VectorRef;
+use snafu::ResultExt;
+use store_api::storage::TableId;
+pub use tables::get_schema_columns;
+
+use crate::error::{CreateRecordBatchSnafu, InternalSnafu, Result};
+use crate::information_schema::InformationTable;
+
+/// A memory table with specified schema and columns.
+pub(super) struct MemoryTable {
+    table_id: TableId,
+    table_name: &'static str,
+    schema: SchemaRef,
+    columns: Vec<VectorRef>,
+}
+
+impl MemoryTable {
+    /// Creates a memory table with table id, name, schema and columns.
+    pub(super) fn new(
+        table_id: TableId,
+        table_name: &'static str,
+        schema: SchemaRef,
+        columns: Vec<VectorRef>,
+    ) -> Self {
+        Self {
+            table_id,
+            table_name,
+            schema,
+            columns,
+        }
+    }
+
+    fn builder(&self) -> MemoryTableBuilder {
+        MemoryTableBuilder::new(self.schema.clone(), self.columns.clone())
+    }
+}
+
+impl InformationTable for MemoryTable {
+    fn table_id(&self) -> TableId {
+        self.table_id
+    }
+
+    fn table_name(&self) -> &'static str {
+        self.table_name
+    }
+
+    fn schema(&self) -> SchemaRef {
+        self.schema.clone()
+    }
+
+    fn to_stream(&self) -> Result<SendableRecordBatchStream> {
+        let schema = self.schema.arrow_schema().clone();
+        let mut builder = self.builder();
+        let stream = Box::pin(DfRecordBatchStreamAdapter::new(
+            schema,
+            futures::stream::once(async move {
+                builder
+                    .memory_records()
+                    .await
+                    .map(|x| x.into_df_record_batch())
+                    .map_err(Into::into)
+            }),
+        ));
+        Ok(Box::pin(
+            RecordBatchStreamAdapter::try_new(stream)
+                .map_err(BoxedError::new)
+                .context(InternalSnafu)?,
+        ))
+    }
+}
+
+struct MemoryTableBuilder {
+    schema: SchemaRef,
+    columns: Vec<VectorRef>,
+}
+
+impl MemoryTableBuilder {
+    fn new(schema: SchemaRef, columns: Vec<VectorRef>) -> Self {
+        Self { schema, columns }
+    }
+
+    /// Construct the `information_schema.{table_name}` virtual table
+    async fn memory_records(&mut self) -> Result<RecordBatch> {
+        if self.columns.is_empty() {
+            RecordBatch::new_empty(self.schema.clone()).context(CreateRecordBatchSnafu)
+        } else {
+            RecordBatch::new(self.schema.clone(), std::mem::take(&mut self.columns))
+                .context(CreateRecordBatchSnafu)
+        }
+    }
+}
+
+impl DfPartitionStream for MemoryTable {
+    fn schema(&self) -> &ArrowSchemaRef {
+        self.schema.arrow_schema()
+    }
+
+    fn execute(&self, _: Arc<TaskContext>) -> DfSendableRecordBatchStream {
+        let schema = self.schema.arrow_schema().clone();
+        let mut builder = self.builder();
+        Box::pin(DfRecordBatchStreamAdapter::new(
+            schema,
+            futures::stream::once(async move {
+                builder
+                    .memory_records()
+                    .await
+                    .map(|x| x.into_df_record_batch())
+                    .map_err(Into::into)
+            }),
+        ))
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use std::sync::Arc;
+
+    use common_recordbatch::RecordBatches;
+    use datatypes::prelude::ConcreteDataType;
+    use datatypes::schema::{ColumnSchema, Schema};
+    use datatypes::vectors::StringVector;
+
+    use super::*;
+
+    #[tokio::test]
+    async fn test_memory_table() {
+        let schema = Arc::new(Schema::new(vec![
+            ColumnSchema::new("a", ConcreteDataType::string_datatype(), false),
+            ColumnSchema::new("b", ConcreteDataType::string_datatype(), false),
+        ]));
+
+        let table = MemoryTable::new(
+            42,
+            "test",
+            schema.clone(),
+            vec![
+                Arc::new(StringVector::from(vec!["a1", "a2"])),
+                Arc::new(StringVector::from(vec!["b1", "b2"])),
+            ],
+        );
+
+        assert_eq!(42, table.table_id());
+        assert_eq!("test", table.table_name());
+        assert_eq!(schema, InformationTable::schema(&table));
+
+        let stream = table.to_stream().unwrap();
+
+        let batches = RecordBatches::try_collect(stream).await.unwrap();
+
+        assert_eq!(
+            "\
++----+----+
+| a  | b  |
++----+----+
+| a1 | b1 |
+| a2 | b2 |
++----+----+",
+            batches.pretty_print().unwrap()
+        );
+    }
+
+    #[tokio::test]
+    async fn test_empty_memory_table() {
+        let schema = Arc::new(Schema::new(vec![
+            ColumnSchema::new("a", ConcreteDataType::string_datatype(), false),
+            ColumnSchema::new("b", ConcreteDataType::string_datatype(), false),
+        ]));
+
+        let table = MemoryTable::new(42, "test", schema.clone(), vec![]);
+
+        assert_eq!(42, table.table_id());
+        assert_eq!("test", table.table_name());
+        assert_eq!(schema, InformationTable::schema(&table));
+
+        let stream = table.to_stream().unwrap();
+
+        let batches = RecordBatches::try_collect(stream).await.unwrap();
+
+        assert_eq!(
+            "\
++---+---+
+| a | b |
++---+---+
++---+---+",
+            batches.pretty_print().unwrap()
+        );
+    }
+}

src/catalog/src/information_schema/memory_table/tables.rs

@@ -0,0 +1,135 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use std::sync::Arc;
+
+use common_catalog::consts::MITO_ENGINE;
+use datatypes::prelude::{ConcreteDataType, VectorRef};
+use datatypes::schema::{ColumnSchema, Schema, SchemaRef};
+use datatypes::vectors::StringVector;
+
+use crate::information_schema::table_names::*;
+
+const UNKNOWN: &str = "unknown";
+
+/// Find the schema and columns by the table_name, only valid for memory tables.
+/// Safety: the user MUST ensure the table schema exists, panic otherwise.
+pub fn get_schema_columns(table_name: &str) -> (SchemaRef, Vec<VectorRef>) {
+    let (column_schemas, columns): (_, Vec<VectorRef>) = match table_name {
+        COLUMN_PRIVILEGES => (
+            string_columns(&[
+                "GRANTEE",
+                "TABLE_CATALOG",
+                "TABLE_SCHEMA",
+                "TABLE_NAME",
+                "COLUMN_NAME",
+                "PRIVILEGE_TYPE",
+                "IS_GRANTABLE",
+            ]),
+            vec![],
+        ),
+
+        COLUMN_STATISTICS => (
+            string_columns(&[
+                "SCHEMA_NAME",
+                "TABLE_NAME",
+                "COLUMN_NAME",
+                // TODO(dennis): It must be a JSON type, but we don't support it yet
+                "HISTOGRAM",
+            ]),
+            vec![],
+        ),
+
+        ENGINES => (
+            string_columns(&[
+                "ENGINE",
+                "SUPPORT",
+                "COMMENT",
+                "TRANSACTIONS",
+                "XA",
+                "SAVEPOINTS",
+            ]),
+            vec![
+                Arc::new(StringVector::from(vec![MITO_ENGINE])),
+                Arc::new(StringVector::from(vec!["DEFAULT"])),
+                Arc::new(StringVector::from(vec![
+                    "Storage engine for time-series data",
+                ])),
+                Arc::new(StringVector::from(vec!["NO"])),
+                Arc::new(StringVector::from(vec!["NO"])),
+                Arc::new(StringVector::from(vec!["NO"])),
+            ],
+        ),
+
+        BUILD_INFO => (
+            string_columns(&[
+                "GIT_BRANCH",
+                "GIT_COMMIT",
+                "GIT_COMMIT_SHORT",
+                "GIT_DIRTY",
+                "PKG_VERSION",
+            ]),
+            vec![
+                Arc::new(StringVector::from(vec![
+                    build_data::get_git_branch().unwrap_or_else(|_| UNKNOWN.to_string())
+                ])),
+                Arc::new(StringVector::from(vec![
+                    build_data::get_git_commit().unwrap_or_else(|_| UNKNOWN.to_string())
+                ])),
+                Arc::new(StringVector::from(vec![
+                    build_data::get_git_commit_short().unwrap_or_else(|_| UNKNOWN.to_string())
+                ])),
+                Arc::new(StringVector::from(vec![
+                    build_data::get_git_dirty().map_or(UNKNOWN.to_string(), |v| v.to_string())
+                ])),
+                Arc::new(StringVector::from(vec![option_env!("CARGO_PKG_VERSION")])),
+            ],
+        ),
+
+        _ => unreachable!("Unknown table in information_schema: {}", table_name),
+    };
+
+    (Arc::new(Schema::new(column_schemas)), columns)
+}
+
+fn string_columns(names: &[&'static str]) -> Vec<ColumnSchema> {
+    names.iter().map(|name| string_column(name)).collect()
+}
+
+fn string_column(name: &str) -> ColumnSchema {
+    ColumnSchema::new(
+        str::to_lowercase(name),
+        ConcreteDataType::string_datatype(),
+        false,
+    )
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_string_columns() {
+        let columns = ["a", "b", "c"];
+        let column_schemas = string_columns(&columns);
+
+        assert_eq!(3, column_schemas.len());
+        for (i, name) in columns.iter().enumerate() {
+            let cs = column_schemas.get(i).unwrap();
+
+            assert_eq!(*name, cs.name);
+            assert_eq!(ConcreteDataType::string_datatype(), cs.data_type);
+        }
+    }
+}

src/catalog/src/information_schema/table_names.rs

@@ -12,13 +12,11 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-use crate::storage::SchemaRef;
+/// All table names in `information_schema`.
 
-/// Metadata of a region.
-pub trait RegionMeta: Send + Sync {
-    /// Returns the schema of the region.
-    fn schema(&self) -> &SchemaRef;
-
-    /// Returns the version of the region metadata.
-    fn version(&self) -> u32;
-}
+pub const TABLES: &str = "tables";
+pub const COLUMNS: &str = "columns";
+pub const ENGINES: &str = "engines";
+pub const COLUMN_PRIVILEGES: &str = "column_privileges";
+pub const COLUMN_STATISTICS: &str = "column_statistics";
+pub const BUILD_INFO: &str = "build_info";

src/catalog/src/information_schema/tables.rs

@@ -15,10 +15,7 @@
 use std::sync::{Arc, Weak};
 
 use arrow_schema::SchemaRef as ArrowSchemaRef;
-use common_catalog::consts::{
-    INFORMATION_SCHEMA_COLUMNS_TABLE_ID, INFORMATION_SCHEMA_NAME,
-    INFORMATION_SCHEMA_TABLES_TABLE_ID,
-};
+use common_catalog::consts::INFORMATION_SCHEMA_TABLES_TABLE_ID;
 use common_error::ext::BoxedError;
 use common_query::physical_plan::TaskContext;
 use common_recordbatch::adapter::RecordBatchStreamAdapter;
@@ -33,7 +30,7 @@ use snafu::{OptionExt, ResultExt};
 use store_api::storage::TableId;
 use table::metadata::TableType;
 
-use super::{COLUMNS, TABLES};
+use super::TABLES;
 use crate::error::{
     CreateRecordBatchSnafu, InternalSnafu, Result, UpgradeWeakCatalogManagerRefSnafu,
 };
@@ -178,29 +175,8 @@ impl InformationSchemaTablesBuilder {
                         Some(&table_info.meta.engine),
                     );
                 } else {
-                    // TODO: this specific branch is only a workaround for FrontendCatalogManager.
-                    if schema_name == INFORMATION_SCHEMA_NAME {
-                        if table_name == COLUMNS {
-                            self.add_table(
-                                &catalog_name,
-                                &schema_name,
-                                &table_name,
-                                TableType::Temporary,
-                                Some(INFORMATION_SCHEMA_COLUMNS_TABLE_ID),
-                                None,
-                            );
-                        } else if table_name == TABLES {
-                            self.add_table(
-                                &catalog_name,
-                                &schema_name,
-                                &table_name,
-                                TableType::Temporary,
-                                Some(INFORMATION_SCHEMA_TABLES_TABLE_ID),
-                                None,
-                            );
-                        }
-                    }
-                };
+                    unreachable!();
+                }
             }
         }
 

src/catalog/src/kvbackend/client.rs

@@ -25,11 +25,10 @@ use common_meta::kv_backend::{KvBackend, KvBackendRef, TxnService};
 use common_meta::rpc::store::{
     BatchDeleteRequest, BatchDeleteResponse, BatchGetRequest, BatchGetResponse, BatchPutRequest,
     BatchPutResponse, CompareAndPutRequest, CompareAndPutResponse, DeleteRangeRequest,
-    DeleteRangeResponse, MoveValueRequest, MoveValueResponse, PutRequest, PutResponse,
-    RangeRequest, RangeResponse,
+    DeleteRangeResponse, PutRequest, PutResponse, RangeRequest, RangeResponse,
 };
 use common_meta::rpc::KeyValue;
-use common_telemetry::{debug, timer};
+use common_telemetry::debug;
 use meta_client::client::MetaClient;
 use moka::future::{Cache, CacheBuilder};
 use snafu::{OptionExt, ResultExt};
@@ -152,25 +151,11 @@ impl KvBackend for CachedMetaKvBackend {
         }
     }
 
-    async fn move_value(&self, req: MoveValueRequest) -> Result<MoveValueResponse> {
-        let from_key = &req.from_key.clone();
-        let to_key = &req.to_key.clone();
-
-        let ret = self.kv_backend.move_value(req).await;
-
-        if ret.is_ok() {
-            self.invalidate_key(from_key).await;
-            self.invalidate_key(to_key).await;
-        }
-
-        ret
-    }
-
     async fn get(&self, key: &[u8]) -> Result<Option<KeyValue>> {
-        let _timer = timer!(METRIC_CATALOG_KV_GET);
+        let _timer = METRIC_CATALOG_KV_GET.start_timer();
 
         let init = async {
-            let _timer = timer!(METRIC_CATALOG_KV_REMOTE_GET);
+            let _timer = METRIC_CATALOG_KV_REMOTE_GET.start_timer();
             self.kv_backend.get(key).await.map(|val| {
                 val.with_context(|| CacheNotGetSnafu {
                     key: String::from_utf8_lossy(key),
@@ -319,14 +304,6 @@ impl KvBackend for MetaKvBackend {
             .context(ExternalSnafu)
     }
 
-    async fn move_value(&self, req: MoveValueRequest) -> Result<MoveValueResponse> {
-        self.client
-            .move_value(req)
-            .await
-            .map_err(BoxedError::new)
-            .context(ExternalSnafu)
-    }
-
     fn as_any(&self) -> &dyn Any {
         self
     }

src/catalog/src/kvbackend/manager.rs

@@ -19,7 +19,6 @@ use std::sync::{Arc, Weak};
 use common_catalog::consts::{DEFAULT_SCHEMA_NAME, INFORMATION_SCHEMA_NAME, NUMBERS_TABLE_ID};
 use common_error::ext::BoxedError;
 use common_meta::cache_invalidator::{CacheInvalidator, CacheInvalidatorRef, Context};
-use common_meta::datanode_manager::DatanodeManagerRef;
 use common_meta::error::Result as MetaResult;
 use common_meta::key::catalog_name::CatalogNameKey;
 use common_meta::key::schema_name::SchemaNameKey;
@@ -39,7 +38,7 @@ use crate::error::{
     self as catalog_err, ListCatalogsSnafu, ListSchemasSnafu, Result as CatalogResult,
     TableMetadataManagerSnafu,
 };
-use crate::information_schema::{InformationSchemaProvider, COLUMNS, TABLES};
+use crate::information_schema::InformationSchemaProvider;
 use crate::CatalogManager;
 
 /// Access all existing catalog, schema and tables.
@@ -55,7 +54,6 @@ pub struct KvBackendCatalogManager {
     cache_invalidator: CacheInvalidatorRef,
     partition_manager: PartitionRuleManagerRef,
     table_metadata_manager: TableMetadataManagerRef,
-    datanode_manager: DatanodeManagerRef,
     /// A sub-CatalogManager that handles system tables
     system_catalog: SystemCatalog,
 }
@@ -76,18 +74,18 @@ impl CacheInvalidator for KvBackendCatalogManager {
 }
 
 impl KvBackendCatalogManager {
-    pub fn new(
-        backend: KvBackendRef,
-        cache_invalidator: CacheInvalidatorRef,
-        datanode_manager: DatanodeManagerRef,
-    ) -> Arc<Self> {
+    pub fn new(backend: KvBackendRef, cache_invalidator: CacheInvalidatorRef) -> Arc<Self> {
         Arc::new_cyclic(|me| Self {
             partition_manager: Arc::new(PartitionRuleManager::new(backend.clone())),
             table_metadata_manager: Arc::new(TableMetadataManager::new(backend)),
             cache_invalidator,
-            datanode_manager,
             system_catalog: SystemCatalog {
                 catalog_manager: me.clone(),
+                information_schema_provider: Arc::new(InformationSchemaProvider::new(
+                    // The catalog name is not used in system_catalog, so let it empty
+                    "".to_string(),
+                    me.clone(),
+                )),
             },
         })
     }
@@ -99,10 +97,6 @@ impl KvBackendCatalogManager {
     pub fn table_metadata_manager_ref(&self) -> &TableMetadataManagerRef {
         &self.table_metadata_manager
     }
-
-    pub fn datanode_manager(&self) -> DatanodeManagerRef {
-        self.datanode_manager.clone()
-    }
 }
 
 #[async_trait::async_trait]
@@ -133,13 +127,11 @@ impl CatalogManager for KvBackendCatalogManager {
             .try_collect::<BTreeSet<_>>()
             .await
             .map_err(BoxedError::new)
-            .context(ListSchemasSnafu { catalog })?
-            .into_iter()
-            .collect::<Vec<_>>();
+            .context(ListSchemasSnafu { catalog })?;
 
-        keys.extend_from_slice(&self.system_catalog.schema_names());
+        keys.extend(self.system_catalog.schema_names());
 
-        Ok(keys)
+        Ok(keys.into_iter().collect())
     }
 
     async fn table_names(&self, catalog: &str, schema: &str) -> CatalogResult<Vec<String>> {
@@ -242,11 +234,11 @@ impl CatalogManager for KvBackendCatalogManager {
 // a new catalog is created.
 /// Existing system tables:
 /// - public.numbers
-/// - information_schema.tables
-/// - information_schema.columns
+/// - information_schema.{tables}
 #[derive(Clone)]
 struct SystemCatalog {
     catalog_manager: Weak<KvBackendCatalogManager>,
+    information_schema_provider: Arc<InformationSchemaProvider>,
 }
 
 impl SystemCatalog {
@@ -256,7 +248,7 @@ impl SystemCatalog {
 
     fn table_names(&self, schema: &str) -> Vec<String> {
         if schema == INFORMATION_SCHEMA_NAME {
-            vec![TABLES.to_string(), COLUMNS.to_string()]
+            self.information_schema_provider.table_names()
         } else if schema == DEFAULT_SCHEMA_NAME {
             vec![NUMBERS_TABLE_NAME.to_string()]
         } else {
@@ -270,7 +262,7 @@ impl SystemCatalog {
 
     fn table_exist(&self, schema: &str, table: &str) -> bool {
         if schema == INFORMATION_SCHEMA_NAME {
-            table == TABLES || table == COLUMNS
+            self.information_schema_provider.table(table).is_some()
         } else if schema == DEFAULT_SCHEMA_NAME {
             table == NUMBERS_TABLE_NAME
         } else {

src/catalog/src/lib.rs

@@ -12,7 +12,6 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-#![feature(trait_upcasting)]
 #![feature(assert_matches)]
 #![feature(try_blocks)]
 

src/catalog/src/memory/manager.rs

@@ -17,8 +17,10 @@ use std::collections::hash_map::Entry;
 use std::collections::HashMap;
 use std::sync::{Arc, RwLock, Weak};
 
-use common_catalog::consts::{DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, INFORMATION_SCHEMA_NAME};
-use metrics::{decrement_gauge, increment_gauge};
+use common_catalog::build_db_string;
+use common_catalog::consts::{
+    DEFAULT_CATALOG_NAME, DEFAULT_PRIVATE_SCHEMA_NAME, DEFAULT_SCHEMA_NAME, INFORMATION_SCHEMA_NAME,
+};
 use snafu::OptionExt;
 use table::TableRef;
 
@@ -135,6 +137,18 @@ impl MemoryCatalogManager {
                 schema: DEFAULT_SCHEMA_NAME.to_string(),
             })
             .unwrap();
+        manager
+            .register_schema_sync(RegisterSchemaRequest {
+                catalog: DEFAULT_CATALOG_NAME.to_string(),
+                schema: DEFAULT_PRIVATE_SCHEMA_NAME.to_string(),
+            })
+            .unwrap();
+        manager
+            .register_schema_sync(RegisterSchemaRequest {
+                catalog: DEFAULT_CATALOG_NAME.to_string(),
+                schema: INFORMATION_SCHEMA_NAME.to_string(),
+            })
+            .unwrap();
 
         manager
     }
@@ -166,7 +180,7 @@ impl MemoryCatalogManager {
                 let arc_self = Arc::new(self.clone());
                 let catalog = arc_self.create_catalog_entry(name);
                 e.insert(catalog);
-                increment_gauge!(crate::metrics::METRIC_CATALOG_MANAGER_CATALOG_COUNT, 1.0);
+                crate::metrics::METRIC_CATALOG_MANAGER_CATALOG_COUNT.inc();
                 Ok(true)
             }
             Entry::Occupied(_) => Ok(false),
@@ -187,11 +201,9 @@ impl MemoryCatalogManager {
             })?;
         let result = schema.remove(&request.table_name);
         if result.is_some() {
-            decrement_gauge!(
-                crate::metrics::METRIC_CATALOG_MANAGER_TABLE_COUNT,
-                1.0,
-                &[crate::metrics::db_label(&request.catalog, &request.schema)],
-            );
+            crate::metrics::METRIC_CATALOG_MANAGER_TABLE_COUNT
+                .with_label_values(&[build_db_string(&request.catalog, &request.schema).as_str()])
+                .dec();
         }
         Ok(())
     }
@@ -210,7 +222,7 @@ impl MemoryCatalogManager {
         match catalog.entry(request.schema) {
             Entry::Vacant(e) => {
                 e.insert(HashMap::new());
-                increment_gauge!(crate::metrics::METRIC_CATALOG_MANAGER_SCHEMA_COUNT, 1.0);
+                crate::metrics::METRIC_CATALOG_MANAGER_SCHEMA_COUNT.inc();
                 Ok(true)
             }
             Entry::Occupied(_) => Ok(false),
@@ -238,19 +250,19 @@ impl MemoryCatalogManager {
             .fail();
         }
         schema.insert(request.table_name, request.table);
-        increment_gauge!(
-            crate::metrics::METRIC_CATALOG_MANAGER_TABLE_COUNT,
-            1.0,
-            &[crate::metrics::db_label(&request.catalog, &request.schema)],
-        );
+        crate::metrics::METRIC_CATALOG_MANAGER_TABLE_COUNT
+            .with_label_values(&[build_db_string(&request.catalog, &request.schema).as_str()])
+            .inc();
         Ok(true)
     }
 
     fn create_catalog_entry(self: &Arc<Self>, catalog: String) -> SchemaEntries {
-        let information_schema = InformationSchemaProvider::build(
+        let information_schema_provider = InformationSchemaProvider::new(
             catalog,
             Arc::downgrade(self) as Weak<dyn CatalogManager>,
         );
+        let information_schema = information_schema_provider.tables().clone();
+
         let mut catalog = HashMap::new();
         catalog.insert(INFORMATION_SCHEMA_NAME.to_string(), information_schema);
         catalog

src/catalog/src/metrics.rs

@@ -12,18 +12,24 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-use common_catalog::build_db_string;
-
 pub(crate) const METRIC_DB_LABEL: &str = "db";
 
-pub(crate) const METRIC_CATALOG_MANAGER_CATALOG_COUNT: &str = "catalog.catalog_count";
-pub(crate) const METRIC_CATALOG_MANAGER_SCHEMA_COUNT: &str = "catalog.schema_count";
-pub(crate) const METRIC_CATALOG_MANAGER_TABLE_COUNT: &str = "catalog.table_count";
+use lazy_static::lazy_static;
+use prometheus::*;
 
-pub(crate) const METRIC_CATALOG_KV_REMOTE_GET: &str = "catalog.kv.get.remote";
-pub(crate) const METRIC_CATALOG_KV_GET: &str = "catalog.kv.get";
-
-#[inline]
-pub(crate) fn db_label(catalog: &str, schema: &str) -> (&'static str, String) {
-    (METRIC_DB_LABEL, build_db_string(catalog, schema))
+lazy_static! {
+    pub static ref METRIC_CATALOG_MANAGER_CATALOG_COUNT: IntGauge =
+        register_int_gauge!("catalog_catalog_count", "catalog catalog count").unwrap();
+    pub static ref METRIC_CATALOG_MANAGER_SCHEMA_COUNT: IntGauge =
+        register_int_gauge!("catalog_schema_count", "catalog schema count").unwrap();
+    pub static ref METRIC_CATALOG_MANAGER_TABLE_COUNT: IntGaugeVec = register_int_gauge_vec!(
+        "catalog_table_count",
+        "catalog table count",
+        &[METRIC_DB_LABEL]
+    )
+    .unwrap();
+    pub static ref METRIC_CATALOG_KV_REMOTE_GET: Histogram =
+        register_histogram!("catalog_kv_get_remote", "catalog kv get remote").unwrap();
+    pub static ref METRIC_CATALOG_KV_GET: Histogram =
+        register_histogram!("catalog_kv_get", "catalog kv get").unwrap();
 }

src/client/Cargo.toml

@@ -8,44 +8,45 @@ license.workspace = true
 testing = []
 
 [dependencies]
-api = { workspace = true }
+api.workspace = true
 arrow-flight.workspace = true
 async-stream.workspace = true
 async-trait.workspace = true
-common-base = { workspace = true }
-common-catalog = { workspace = true }
-common-error = { workspace = true }
-common-grpc = { workspace = true }
-common-macro = { workspace = true }
-common-meta = { workspace = true }
-common-query = { workspace = true }
-common-recordbatch = { workspace = true }
-common-telemetry = { workspace = true }
-common-time = { workspace = true }
+common-base.workspace = true
+common-catalog.workspace = true
+common-error.workspace = true
+common-grpc.workspace = true
+common-macro.workspace = true
+common-meta.workspace = true
+common-query.workspace = true
+common-recordbatch.workspace = true
+common-telemetry.workspace = true
+common-time.workspace = true
 datafusion.workspace = true
-datatypes = { workspace = true }
+datatypes.workspace = true
 derive_builder.workspace = true
 enum_dispatch = "0.3"
 futures-util.workspace = true
+lazy_static.workspace = true
 moka = { workspace = true, features = ["future"] }
 parking_lot = "0.12"
+prometheus.workspace = true
 prost.workspace = true
 rand.workspace = true
-session = { workspace = true }
+session.workspace = true
 snafu.workspace = true
-tokio-stream = { version = "0.1", features = ["net"] }
+tokio-stream = { workspace = true, features = ["net"] }
 tokio.workspace = true
 tonic.workspace = true
 
 [dev-dependencies]
-common-grpc-expr = { workspace = true }
-datanode = { workspace = true }
+common-grpc-expr.workspace = true
+datanode.workspace = true
 derive-new = "0.5"
-prost.workspace = true
-substrait = { workspace = true }
+substrait.workspace = true
 tracing = "0.1"
 tracing-subscriber = { version = "0.3", features = ["env-filter"] }
 
 [dev-dependencies.substrait_proto]
 package = "substrait"
-version = "0.7"
+version = "0.17"

src/client/examples/logical.rs

@@ -46,6 +46,7 @@ async fn run() {
                 default_constraint: vec![],
                 semantic_type: SemanticType::Timestamp as i32,
                 comment: String::new(),
+                ..Default::default()
             },
             ColumnDef {
                 name: "key".to_string(),
@@ -54,6 +55,7 @@ async fn run() {
                 default_constraint: vec![],
                 semantic_type: SemanticType::Tag as i32,
                 comment: String::new(),
+                ..Default::default()
             },
             ColumnDef {
                 name: "value".to_string(),
@@ -62,6 +64,7 @@ async fn run() {
                 default_constraint: vec![],
                 semantic_type: SemanticType::Field as i32,
                 comment: String::new(),
+                ..Default::default()
             },
         ],
         time_index: "timestamp".to_string(),
@@ -78,7 +81,7 @@ async fn run() {
 
     let logical = mock_logical_plan();
     event!(Level::INFO, "plan size: {:#?}", logical.len());
-    let result = db.logical_plan(logical, 0).await.unwrap();
+    let result = db.logical_plan(logical).await.unwrap();
 
     event!(Level::INFO, "result: {:#?}", result);
 }

src/client/src/database.rs

@@ -27,8 +27,9 @@ use common_error::ext::{BoxedError, ErrorExt};
 use common_grpc::flight::{FlightDecoder, FlightMessage};
 use common_query::Output;
 use common_recordbatch::error::ExternalSnafu;
-use common_recordbatch::RecordBatchStreamAdaptor;
-use common_telemetry::{logging, timer};
+use common_recordbatch::RecordBatchStreamWrapper;
+use common_telemetry::logging;
+use common_telemetry::tracing_context::W3cTrace;
 use futures_util::StreamExt;
 use prost::Message;
 use snafu::{ensure, ResultExt};
@@ -111,12 +112,12 @@ impl Database {
     }
 
     pub async fn insert(&self, requests: InsertRequests) -> Result<u32> {
-        let _timer = timer!(metrics::METRIC_GRPC_INSERT);
+        let _timer = metrics::METRIC_GRPC_INSERT.start_timer();
         self.handle(Request::Inserts(requests)).await
     }
 
     pub async fn row_insert(&self, requests: RowInsertRequests) -> Result<u32> {
-        let _timer = timer!(metrics::METRIC_GRPC_INSERT);
+        let _timer = metrics::METRIC_GRPC_INSERT.start_timer();
         self.handle(Request::RowInserts(requests)).await
     }
 
@@ -141,27 +142,27 @@ impl Database {
     }
 
     pub async fn delete(&self, request: DeleteRequests) -> Result<u32> {
-        let _timer = timer!(metrics::METRIC_GRPC_DELETE);
+        let _timer = metrics::METRIC_GRPC_DELETE.start_timer();
         self.handle(Request::Deletes(request)).await
     }
 
     async fn handle(&self, request: Request) -> Result<u32> {
         let mut client = self.client.make_database_client()?.inner;
-        let request = self.to_rpc_request(request, 0);
+        let request = self.to_rpc_request(request);
         let response = client.handle(request).await?.into_inner();
         from_grpc_response(response)
     }
 
     #[inline]
-    fn to_rpc_request(&self, request: Request, trace_id: u64) -> GreptimeRequest {
+    fn to_rpc_request(&self, request: Request) -> GreptimeRequest {
         GreptimeRequest {
             header: Some(RequestHeader {
                 catalog: self.catalog.clone(),
                 schema: self.schema.clone(),
                 authorization: self.ctx.auth_header.clone(),
                 dbname: self.dbname.clone(),
-                trace_id,
-                span_id: 0,
+                // TODO(Taylor-lagrange): add client grpc tracing
+                tracing_context: W3cTrace::new(),
             }),
             request: Some(request),
         }
@@ -171,24 +172,18 @@ impl Database {
     where
         S: AsRef<str>,
     {
-        let _timer = timer!(metrics::METRIC_GRPC_SQL);
-        self.do_get(
-            Request::Query(QueryRequest {
-                query: Some(Query::Sql(sql.as_ref().to_string())),
-            }),
-            0,
-        )
+        let _timer = metrics::METRIC_GRPC_SQL.start_timer();
+        self.do_get(Request::Query(QueryRequest {
+            query: Some(Query::Sql(sql.as_ref().to_string())),
+        }))
         .await
     }
 
-    pub async fn logical_plan(&self, logical_plan: Vec<u8>, trace_id: u64) -> Result<Output> {
-        let _timer = timer!(metrics::METRIC_GRPC_LOGICAL_PLAN);
-        self.do_get(
-            Request::Query(QueryRequest {
-                query: Some(Query::LogicalPlan(logical_plan)),
-            }),
-            trace_id,
-        )
+    pub async fn logical_plan(&self, logical_plan: Vec<u8>) -> Result<Output> {
+        let _timer = metrics::METRIC_GRPC_LOGICAL_PLAN.start_timer();
+        self.do_get(Request::Query(QueryRequest {
+            query: Some(Query::LogicalPlan(logical_plan)),
+        }))
         .await
     }
 
@@ -199,69 +194,54 @@ impl Database {
         end: &str,
         step: &str,
     ) -> Result<Output> {
-        let _timer = timer!(metrics::METRIC_GRPC_PROMQL_RANGE_QUERY);
-        self.do_get(
-            Request::Query(QueryRequest {
-                query: Some(Query::PromRangeQuery(PromRangeQuery {
-                    query: promql.to_string(),
-                    start: start.to_string(),
-                    end: end.to_string(),
-                    step: step.to_string(),
-                })),
-            }),
-            0,
-        )
+        let _timer = metrics::METRIC_GRPC_PROMQL_RANGE_QUERY.start_timer();
+        self.do_get(Request::Query(QueryRequest {
+            query: Some(Query::PromRangeQuery(PromRangeQuery {
+                query: promql.to_string(),
+                start: start.to_string(),
+                end: end.to_string(),
+                step: step.to_string(),
+            })),
+        }))
         .await
     }
 
     pub async fn create(&self, expr: CreateTableExpr) -> Result<Output> {
-        let _timer = timer!(metrics::METRIC_GRPC_CREATE_TABLE);
-        self.do_get(
-            Request::Ddl(DdlRequest {
-                expr: Some(DdlExpr::CreateTable(expr)),
-            }),
-            0,
-        )
+        let _timer = metrics::METRIC_GRPC_CREATE_TABLE.start_timer();
+        self.do_get(Request::Ddl(DdlRequest {
+            expr: Some(DdlExpr::CreateTable(expr)),
+        }))
         .await
     }
 
     pub async fn alter(&self, expr: AlterExpr) -> Result<Output> {
-        let _timer = timer!(metrics::METRIC_GRPC_ALTER);
-        self.do_get(
-            Request::Ddl(DdlRequest {
-                expr: Some(DdlExpr::Alter(expr)),
-            }),
-            0,
-        )
+        let _timer = metrics::METRIC_GRPC_ALTER.start_timer();
+        self.do_get(Request::Ddl(DdlRequest {
+            expr: Some(DdlExpr::Alter(expr)),
+        }))
         .await
     }
 
     pub async fn drop_table(&self, expr: DropTableExpr) -> Result<Output> {
-        let _timer = timer!(metrics::METRIC_GRPC_DROP_TABLE);
-        self.do_get(
-            Request::Ddl(DdlRequest {
-                expr: Some(DdlExpr::DropTable(expr)),
-            }),
-            0,
-        )
+        let _timer = metrics::METRIC_GRPC_DROP_TABLE.start_timer();
+        self.do_get(Request::Ddl(DdlRequest {
+            expr: Some(DdlExpr::DropTable(expr)),
+        }))
         .await
     }
 
     pub async fn truncate_table(&self, expr: TruncateTableExpr) -> Result<Output> {
-        let _timer = timer!(metrics::METRIC_GRPC_TRUNCATE_TABLE);
-        self.do_get(
-            Request::Ddl(DdlRequest {
-                expr: Some(DdlExpr::TruncateTable(expr)),
-            }),
-            0,
-        )
+        let _timer = metrics::METRIC_GRPC_TRUNCATE_TABLE.start_timer();
+        self.do_get(Request::Ddl(DdlRequest {
+            expr: Some(DdlExpr::TruncateTable(expr)),
+        }))
         .await
     }
 
-    async fn do_get(&self, request: Request, trace_id: u64) -> Result<Output> {
+    async fn do_get(&self, request: Request) -> Result<Output> {
         // FIXME(paomian): should be added some labels for metrics
-        let _timer = timer!(metrics::METRIC_GRPC_DO_GET);
-        let request = self.to_rpc_request(request, trace_id);
+        let _timer = metrics::METRIC_GRPC_DO_GET.start_timer();
+        let request = self.to_rpc_request(request);
         let request = Ticket {
             ticket: request.encode_to_vec().into(),
         };
@@ -335,7 +315,7 @@ impl Database {
                         yield Ok(record_batch);
                     }
                 }));
-                let record_batch_stream = RecordBatchStreamAdaptor {
+                let record_batch_stream = RecordBatchStreamWrapper {
                     schema,
                     stream,
                     output_ordering: None,

src/client/src/error.rs

@@ -131,3 +131,15 @@ impl From<Status> for Error {
         Self::Server { code, msg }
     }
 }
+
+impl Error {
+    pub fn should_retry(&self) -> bool {
+        !matches!(
+            self,
+            Self::RegionServer {
+                code: Code::InvalidArgument,
+                ..
+            }
+        )
+    }
+}

src/client/src/metrics.rs

@@ -12,15 +12,34 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-//! client metrics
-pub const METRIC_GRPC_CREATE_TABLE: &str = "grpc.create_table";
-pub const METRIC_GRPC_PROMQL_RANGE_QUERY: &str = "grpc.promql.range_query";
-pub const METRIC_GRPC_INSERT: &str = "grpc.insert";
-pub const METRIC_GRPC_DELETE: &str = "grpc.delete";
-pub const METRIC_GRPC_SQL: &str = "grpc.sql";
-pub const METRIC_GRPC_LOGICAL_PLAN: &str = "grpc.logical_plan";
-pub const METRIC_GRPC_ALTER: &str = "grpc.alter";
-pub const METRIC_GRPC_DROP_TABLE: &str = "grpc.drop_table";
-pub const METRIC_GRPC_TRUNCATE_TABLE: &str = "grpc.truncate_table";
-pub const METRIC_GRPC_DO_GET: &str = "grpc.do_get";
-pub(crate) const METRIC_REGION_REQUEST_GRPC: &str = "grpc.region_request";
+use lazy_static::lazy_static;
+use prometheus::*;
+
+lazy_static! {
+    pub static ref METRIC_GRPC_CREATE_TABLE: Histogram =
+        register_histogram!("grpc_create_table", "grpc create table").unwrap();
+    pub static ref METRIC_GRPC_PROMQL_RANGE_QUERY: Histogram =
+        register_histogram!("grpc_promql_range_query", "grpc promql range query").unwrap();
+    pub static ref METRIC_GRPC_INSERT: Histogram =
+        register_histogram!("grpc_insert", "grpc insert").unwrap();
+    pub static ref METRIC_GRPC_DELETE: Histogram =
+        register_histogram!("grpc_delete", "grpc delete").unwrap();
+    pub static ref METRIC_GRPC_SQL: Histogram =
+        register_histogram!("grpc_sql", "grpc sql").unwrap();
+    pub static ref METRIC_GRPC_LOGICAL_PLAN: Histogram =
+        register_histogram!("grpc_logical_plan", "grpc logical plan").unwrap();
+    pub static ref METRIC_GRPC_ALTER: Histogram =
+        register_histogram!("grpc_alter", "grpc alter").unwrap();
+    pub static ref METRIC_GRPC_DROP_TABLE: Histogram =
+        register_histogram!("grpc_drop_table", "grpc drop table").unwrap();
+    pub static ref METRIC_GRPC_TRUNCATE_TABLE: Histogram =
+        register_histogram!("grpc_truncate_table", "grpc truncate table").unwrap();
+    pub static ref METRIC_GRPC_DO_GET: Histogram =
+        register_histogram!("grpc_do_get", "grpc do get").unwrap();
+    pub static ref METRIC_REGION_REQUEST_GRPC: HistogramVec = register_histogram_vec!(
+        "grpc_region_request",
+        "grpc region request",
+        &["request_type"]
+    )
+    .unwrap();
+}

src/client/src/region.rs

@@ -23,13 +23,12 @@ use common_grpc::flight::{FlightDecoder, FlightMessage};
 use common_meta::datanode_manager::{AffectedRows, Datanode};
 use common_meta::error::{self as meta_error, Result as MetaResult};
 use common_recordbatch::error::ExternalSnafu;
-use common_recordbatch::{RecordBatchStreamAdaptor, SendableRecordBatchStream};
-use common_telemetry::{error, timer};
+use common_recordbatch::{RecordBatchStreamWrapper, SendableRecordBatchStream};
+use common_telemetry::error;
 use prost::Message;
 use snafu::{location, Location, OptionExt, ResultExt};
 use tokio_stream::StreamExt;
 
-use crate::error::Error::RegionServer;
 use crate::error::{
     self, ConvertFlightDataSnafu, IllegalDatabaseResponseSnafu, IllegalFlightMessagesSnafu,
     MissingFieldSnafu, Result, ServerSnafu,
@@ -45,7 +44,7 @@ pub struct RegionRequester {
 impl Datanode for RegionRequester {
     async fn handle(&self, request: RegionRequest) -> MetaResult<AffectedRows> {
         self.handle_inner(request).await.map_err(|err| {
-            if matches!(err, RegionServer { .. }) {
+            if err.should_retry() {
                 meta_error::Error::RetryLater {
                     source: BoxedError::new(err),
                 }
@@ -137,7 +136,7 @@ impl RegionRequester {
                 yield Ok(record_batch);
             }
         }));
-        let record_batch_stream = RecordBatchStreamAdaptor {
+        let record_batch_stream = RecordBatchStreamWrapper {
             schema,
             stream,
             output_ordering: None,
@@ -152,11 +151,9 @@ impl RegionRequester {
             .with_context(|| MissingFieldSnafu { field: "body" })?
             .as_ref()
             .to_string();
-
-        let _timer = timer!(
-            metrics::METRIC_REGION_REQUEST_GRPC,
-            &[("request_type", request_type)]
-        );
+        let _timer = metrics::METRIC_REGION_REQUEST_GRPC
+            .with_label_values(&[request_type.as_str()])
+            .start_timer();
 
         let mut client = self.client.raw_region_client()?;
 

src/cmd/Cargo.toml

@@ -10,71 +10,72 @@ name = "greptime"
 path = "src/bin/greptime.rs"
 
 [features]
-default = ["metrics-process"]
 tokio-console = ["common-telemetry/tokio-console"]
-metrics-process = ["servers/metrics-process"]
 
 [dependencies]
 anymap = "1.0.0-beta.2"
 async-trait.workspace = true
 auth.workspace = true
-catalog = { workspace = true }
+catalog.workspace = true
 chrono.workspace = true
-clap = { version = "3.1", features = ["derive"] }
-client = { workspace = true }
-common-base = { workspace = true }
-common-catalog = { workspace = true }
-common-config = { workspace = true }
-common-error = { workspace = true }
-common-macro = { workspace = true }
-common-meta = { workspace = true }
-common-procedure = { workspace = true }
-common-query = { workspace = true }
-common-recordbatch = { workspace = true }
+clap = { version = "4.4", features = ["derive"] }
+client.workspace = true
+common-base.workspace = true
+common-catalog.workspace = true
+common-config.workspace = true
+common-error.workspace = true
+common-macro.workspace = true
+common-meta.workspace = true
+common-procedure.workspace = true
+common-query.workspace = true
+common-recordbatch.workspace = true
 common-telemetry = { workspace = true, features = [
     "deadlock_detection",
 ] }
+common-time.workspace = true
 config = "0.13"
-datanode = { workspace = true }
-datatypes = { workspace = true }
+datanode.workspace = true
+datatypes.workspace = true
 either = "1.8"
 etcd-client.workspace = true
-file-engine = { workspace = true }
-frontend = { workspace = true }
+file-engine.workspace = true
+frontend.workspace = true
 futures.workspace = true
+human-panic = "1.2.2"
 lazy_static.workspace = true
-meta-client = { workspace = true }
-meta-srv = { workspace = true }
-metrics.workspace = true
-mito2 = { workspace = true }
+meta-client.workspace = true
+meta-srv.workspace = true
+mito2.workspace = true
 nu-ansi-term = "0.46"
-partition = { workspace = true }
+partition.workspace = true
 plugins.workspace = true
+prometheus.workspace = true
 prost.workspace = true
-query = { workspace = true }
+query.workspace = true
 rand.workspace = true
 regex.workspace = true
 rustyline = "10.1"
 serde.workspace = true
 serde_json.workspace = true
-servers = { workspace = true }
-session = { workspace = true }
+servers.workspace = true
+session.workspace = true
 snafu.workspace = true
-substrait = { workspace = true }
-table = { workspace = true }
+store-api.workspace = true
+substrait.workspace = true
+table.workspace = true
 tokio.workspace = true
+toml.workspace = true
 
 [target.'cfg(not(windows))'.dependencies]
 tikv-jemallocator = "0.5"
 
 [dev-dependencies]
-common-test-util = { workspace = true }
+common-test-util.workspace = true
 serde.workspace = true
 temp-env = "0.3"
-toml.workspace = true
 
 [target.'cfg(not(windows))'.dev-dependencies]
 rexpect = "0.5"
 
 [build-dependencies]
-common-version = { workspace = true }
+common-version.workspace = true

src/cmd/src/bin/greptime.rs

@@ -16,75 +16,12 @@
 
 use std::fmt;
 
-use clap::Parser;
+use clap::{FromArgMatches, Parser, Subcommand};
 use cmd::error::Result;
-use cmd::options::{Options, TopLevelOptions};
-use cmd::{cli, datanode, frontend, metasrv, standalone};
-use common_telemetry::logging::{error, info, TracingOptions};
-use metrics::gauge;
-
-#[derive(Parser)]
-#[clap(name = "greptimedb", version = print_version())]
-struct Command {
-    #[clap(long)]
-    log_dir: Option<String>,
-    #[clap(long)]
-    log_level: Option<String>,
-    #[clap(subcommand)]
-    subcmd: SubCommand,
-
-    #[cfg(feature = "tokio-console")]
-    #[clap(long)]
-    tokio_console_addr: Option<String>,
-}
-
-pub enum Application {
-    Datanode(datanode::Instance),
-    Frontend(frontend::Instance),
-    Metasrv(metasrv::Instance),
-    Standalone(standalone::Instance),
-    Cli(cli::Instance),
-}
-
-impl Application {
-    async fn start(&mut self) -> Result<()> {
-        match self {
-            Application::Datanode(instance) => instance.start().await,
-            Application::Frontend(instance) => instance.start().await,
-            Application::Metasrv(instance) => instance.start().await,
-            Application::Standalone(instance) => instance.start().await,
-            Application::Cli(instance) => instance.start().await,
-        }
-    }
-
-    async fn stop(&self) -> Result<()> {
-        match self {
-            Application::Datanode(instance) => instance.stop().await,
-            Application::Frontend(instance) => instance.stop().await,
-            Application::Metasrv(instance) => instance.stop().await,
-            Application::Standalone(instance) => instance.stop().await,
-            Application::Cli(instance) => instance.stop().await,
-        }
-    }
-}
-
-impl Command {
-    async fn build(self, opts: Options) -> Result<Application> {
-        self.subcmd.build(opts).await
-    }
-
-    fn load_options(&self) -> Result<Options> {
-        let top_level_opts = self.top_level_options();
-        self.subcmd.load_options(top_level_opts)
-    }
-
-    fn top_level_options(&self) -> TopLevelOptions {
-        TopLevelOptions {
-            log_dir: self.log_dir.clone(),
-            log_level: self.log_level.clone(),
-        }
-    }
-}
+use cmd::options::{CliOptions, Options};
+use cmd::{
+    cli, datanode, frontend, greptimedb_cli, log_versions, metasrv, standalone, start_app, App,
+};
 
 #[derive(Parser)]
 enum SubCommand {
@@ -101,40 +38,41 @@ enum SubCommand {
 }
 
 impl SubCommand {
-    async fn build(self, opts: Options) -> Result<Application> {
-        match (self, opts) {
+    async fn build(self, opts: Options) -> Result<Box<dyn App>> {
+        let app: Box<dyn App> = match (self, opts) {
             (SubCommand::Datanode(cmd), Options::Datanode(dn_opts)) => {
                 let app = cmd.build(*dn_opts).await?;
-                Ok(Application::Datanode(app))
+                Box::new(app) as _
             }
             (SubCommand::Frontend(cmd), Options::Frontend(fe_opts)) => {
                 let app = cmd.build(*fe_opts).await?;
-                Ok(Application::Frontend(app))
+                Box::new(app) as _
             }
             (SubCommand::Metasrv(cmd), Options::Metasrv(meta_opts)) => {
                 let app = cmd.build(*meta_opts).await?;
-                Ok(Application::Metasrv(app))
+                Box::new(app) as _
             }
             (SubCommand::Standalone(cmd), Options::Standalone(opts)) => {
                 let app = cmd.build(*opts).await?;
-                Ok(Application::Standalone(app))
+                Box::new(app) as _
             }
             (SubCommand::Cli(cmd), Options::Cli(_)) => {
                 let app = cmd.build().await?;
-                Ok(Application::Cli(app))
+                Box::new(app) as _
             }
 
             _ => unreachable!(),
-        }
+        };
+        Ok(app)
     }
 
-    fn load_options(&self, top_level_opts: TopLevelOptions) -> Result<Options> {
+    fn load_options(&self, cli_options: &CliOptions) -> Result<Options> {
         match self {
-            SubCommand::Datanode(cmd) => cmd.load_options(top_level_opts),
-            SubCommand::Frontend(cmd) => cmd.load_options(top_level_opts),
-            SubCommand::Metasrv(cmd) => cmd.load_options(top_level_opts),
-            SubCommand::Standalone(cmd) => cmd.load_options(top_level_opts),
-            SubCommand::Cli(cmd) => cmd.load_options(top_level_opts),
+            SubCommand::Datanode(cmd) => cmd.load_options(cli_options),
+            SubCommand::Frontend(cmd) => cmd.load_options(cli_options),
+            SubCommand::Metasrv(cmd) => cmd.load_options(cli_options),
+            SubCommand::Standalone(cmd) => cmd.load_options(cli_options),
+            SubCommand::Cli(cmd) => cmd.load_options(cli_options),
         }
     }
 }
@@ -151,88 +89,49 @@ impl fmt::Display for SubCommand {
     }
 }
 
-fn print_version() -> &'static str {
-    concat!(
-        "\nbranch: ",
-        env!("GIT_BRANCH"),
-        "\ncommit: ",
-        env!("GIT_COMMIT"),
-        "\ndirty: ",
-        env!("GIT_DIRTY"),
-        "\nversion: ",
-        env!("CARGO_PKG_VERSION")
-    )
-}
-
-fn short_version() -> &'static str {
-    env!("CARGO_PKG_VERSION")
-}
-
-// {app_name}-{branch_name}-{commit_short}
-// The branch name (tag) of a release build should already contain the short
-// version so the full version doesn't concat the short version explicitly.
-fn full_version() -> &'static str {
-    concat!(
-        "greptimedb-",
-        env!("GIT_BRANCH"),
-        "-",
-        env!("GIT_COMMIT_SHORT")
-    )
-}
-
-fn log_env_flags() {
-    info!("command line arguments");
-    for argument in std::env::args() {
-        info!("argument: {}", argument);
-    }
-}
-
 #[cfg(not(windows))]
 #[global_allocator]
 static ALLOC: tikv_jemallocator::Jemalloc = tikv_jemallocator::Jemalloc;
 
 #[tokio::main]
 async fn main() -> Result<()> {
-    let cmd = Command::parse();
-    let app_name = &cmd.subcmd.to_string();
-
-    let opts = cmd.load_options()?;
-    let logging_opts = opts.logging_options();
-    let tracing_opts = TracingOptions {
-        #[cfg(feature = "tokio-console")]
-        tokio_console_addr: cmd.tokio_console_addr.clone(),
+    let metadata = human_panic::Metadata {
+        version: env!("CARGO_PKG_VERSION").into(),
+        name: "GreptimeDB".into(),
+        authors: Default::default(),
+        homepage: "https://github.com/GreptimeTeam/greptimedb/discussions".into(),
     };
+    human_panic::setup_panic!(metadata);
 
     common_telemetry::set_panic_hook();
-    common_telemetry::init_default_metrics_recorder();
-    let _guard = common_telemetry::init_global_logging(app_name, logging_opts, tracing_opts);
 
-    // Report app version as gauge.
-    gauge!("app_version", 1.0, "short_version" => short_version(), "version" => full_version());
+    let cli = greptimedb_cli();
 
-    // Log version and argument flags.
-    info!(
-        "short_version: {}, full_version: {}",
-        short_version(),
-        full_version()
+    let cli = SubCommand::augment_subcommands(cli);
+
+    let args = cli.get_matches();
+
+    let subcmd = match SubCommand::from_arg_matches(&args) {
+        Ok(subcmd) => subcmd,
+        Err(e) => e.exit(),
+    };
+
+    let app_name = subcmd.to_string();
+
+    let cli_options = CliOptions::new(&args);
+
+    let opts = subcmd.load_options(&cli_options)?;
+
+    let _guard = common_telemetry::init_global_logging(
+        &app_name,
+        opts.logging_options(),
+        cli_options.tracing_options(),
+        opts.node_id(),
     );
-    log_env_flags();
 
-    let mut app = cmd.build(opts).await?;
+    log_versions();
 
-    tokio::select! {
-        result = app.start() => {
-            if let Err(err) = result {
-                error!(err; "Fatal error occurs!");
-            }
-        }
-        _ = tokio::signal::ctrl_c() => {
-            if let Err(err) = app.stop().await {
-                error!(err; "Fatal error occurs!");
-            }
-            info!("Goodbye!");
-        }
-    }
+    let app = subcmd.build(opts).await?;
 
-    Ok(())
+    start_app(app).await
 }

src/cmd/src/cli.rs

@@ -13,9 +13,15 @@
 // limitations under the License.
 
 mod bench;
+
+// Wait for https://github.com/GreptimeTeam/greptimedb/issues/2373
+#[allow(unused)]
 mod cmd;
 mod export;
 mod helper;
+
+// Wait for https://github.com/GreptimeTeam/greptimedb/issues/2373
+#[allow(unused)]
 mod repl;
 // TODO(weny): Removes it
 #[allow(deprecated)]
@@ -30,27 +36,35 @@ use upgrade::UpgradeCommand;
 
 use self::export::ExportCommand;
 use crate::error::Result;
-use crate::options::{Options, TopLevelOptions};
+use crate::options::{CliOptions, Options};
+use crate::App;
 
 #[async_trait]
-pub trait Tool {
+pub trait Tool: Send + Sync {
     async fn do_work(&self) -> Result<()>;
 }
 
-pub enum Instance {
-    Repl(Repl),
-    Tool(Box<dyn Tool>),
+pub struct Instance {
+    tool: Box<dyn Tool>,
 }
 
 impl Instance {
-    pub async fn start(&mut self) -> Result<()> {
-        match self {
-            Instance::Repl(repl) => repl.run().await,
-            Instance::Tool(tool) => tool.do_work().await,
-        }
+    fn new(tool: Box<dyn Tool>) -> Self {
+        Self { tool }
+    }
+}
+
+#[async_trait]
+impl App for Instance {
+    fn name(&self) -> &str {
+        "greptime-cli"
     }
 
-    pub async fn stop(&self) -> Result<()> {
+    async fn start(&mut self) -> Result<()> {
+        self.tool.do_work().await
+    }
+
+    async fn stop(&self) -> Result<()> {
         Ok(())
     }
 }
@@ -66,14 +80,15 @@ impl Command {
         self.cmd.build().await
     }
 
-    pub fn load_options(&self, top_level_opts: TopLevelOptions) -> Result<Options> {
+    pub fn load_options(&self, cli_options: &CliOptions) -> Result<Options> {
         let mut logging_opts = LoggingOptions::default();
-        if let Some(dir) = top_level_opts.log_dir {
-            logging_opts.dir = dir;
-        }
-        if top_level_opts.log_level.is_some() {
-            logging_opts.level = top_level_opts.log_level;
+
+        if let Some(dir) = &cli_options.log_dir {
+            logging_opts.dir = dir.clone();
         }
+
+        logging_opts.level = cli_options.log_level.clone();
+
         Ok(Options::Cli(Box::new(logging_opts)))
     }
 }
@@ -110,7 +125,6 @@ pub(crate) struct AttachCommand {
 impl AttachCommand {
     #[allow(dead_code)]
     async fn build(self) -> Result<Instance> {
-        let repl = Repl::try_new(&self).await?;
-        Ok(Instance::Repl(repl))
+        unimplemented!("Wait for https://github.com/GreptimeTeam/greptimedb/issues/2373")
     }
 }

src/cmd/src/cli/bench.rs

@@ -12,7 +12,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-use std::collections::BTreeMap;
+use std::collections::{BTreeMap, HashMap};
 use std::future::Future;
 use std::sync::Arc;
 use std::time::Duration;
@@ -20,15 +20,15 @@ use std::time::Duration;
 use async_trait::async_trait;
 use clap::Parser;
 use common_meta::key::{TableMetadataManager, TableMetadataManagerRef};
+use common_meta::kv_backend::etcd::EtcdStore;
 use common_meta::peer::Peer;
 use common_meta::rpc::router::{Region, RegionRoute};
 use common_meta::table_name::TableName;
 use common_telemetry::info;
 use datatypes::data_type::ConcreteDataType;
 use datatypes::schema::{ColumnSchema, RawSchema};
-use meta_srv::service::store::etcd::EtcdStore;
-use meta_srv::service::store::kv::KvBackendAdapter;
 use rand::Rng;
+use store_api::storage::RegionNumber;
 use table::metadata::{RawTableInfo, RawTableMeta, TableId, TableIdent, TableType};
 
 use self::metadata::TableMetadataBencher;
@@ -64,15 +64,13 @@ impl BenchTableMetadataCommand {
     pub async fn build(&self) -> Result<Instance> {
         let etcd_store = EtcdStore::with_endpoints([&self.etcd_addr]).await.unwrap();
 
-        let table_metadata_manager = Arc::new(TableMetadataManager::new(KvBackendAdapter::wrap(
-            etcd_store,
-        )));
+        let table_metadata_manager = Arc::new(TableMetadataManager::new(etcd_store));
 
         let tool = BenchTableMetadata {
             table_metadata_manager,
             count: self.count,
         };
-        Ok(Instance::Tool(Box::new(tool)))
+        Ok(Instance::new(Box::new(tool)))
     }
 }
 
@@ -140,12 +138,12 @@ fn create_table_info(table_id: TableId, table_name: TableName) -> RawTableInfo {
     }
 }
 
-fn create_region_routes() -> Vec<RegionRoute> {
-    let mut regions = Vec::with_capacity(100);
+fn create_region_routes(regions: Vec<RegionNumber>) -> Vec<RegionRoute> {
+    let mut region_routes = Vec::with_capacity(100);
     let mut rng = rand::thread_rng();
 
-    for region_id in 0..64u64 {
-        regions.push(RegionRoute {
+    for region_id in regions.into_iter().map(u64::from) {
+        region_routes.push(RegionRoute {
             region: Region {
                 id: region_id.into(),
                 name: String::new(),
@@ -157,8 +155,15 @@ fn create_region_routes() -> Vec<RegionRoute> {
                 addr: String::new(),
             }),
             follower_peers: vec![],
+            leader_status: None,
         });
     }
 
-    regions
+    region_routes
+}
+
+fn create_region_wal_options(regions: Vec<RegionNumber>) -> HashMap<RegionNumber, String> {
+    // TODO(niebayes): construct region wal options for benchmark.
+    let _ = regions;
+    HashMap::default()
 }

src/cmd/src/cli/bench/metadata.rs

@@ -14,10 +14,13 @@
 
 use std::time::Instant;
 
+use common_meta::key::table_route::TableRouteValue;
 use common_meta::key::TableMetadataManagerRef;
 use common_meta::table_name::TableName;
 
-use super::{bench_self_recorded, create_region_routes, create_table_info};
+use crate::cli::bench::{
+    bench_self_recorded, create_region_routes, create_region_wal_options, create_table_info,
+};
 
 pub struct TableMetadataBencher {
     table_metadata_manager: TableMetadataManagerRef,
@@ -43,12 +46,19 @@ impl TableMetadataBencher {
                 let table_name = format!("bench_table_name_{}", i);
                 let table_name = TableName::new("bench_catalog", "bench_schema", table_name);
                 let table_info = create_table_info(i, table_name);
-                let region_routes = create_region_routes();
+
+                let regions: Vec<_> = (0..64).collect();
+                let region_routes = create_region_routes(regions.clone());
+                let region_wal_options = create_region_wal_options(regions);
 
                 let start = Instant::now();
 
                 self.table_metadata_manager
-                    .create_table_metadata(table_info, region_routes)
+                    .create_table_metadata(
+                        table_info,
+                        TableRouteValue::physical(region_routes),
+                        region_wal_options,
+                    )
                     .await
                     .unwrap();
 

src/cmd/src/cli/export.rs

@@ -17,6 +17,8 @@ use std::sync::Arc;
 
 use async_trait::async_trait;
 use clap::{Parser, ValueEnum};
+use client::api::v1::auth_header::AuthScheme;
+use client::api::v1::Basic;
 use client::{Client, Database, DEFAULT_SCHEMA_NAME};
 use common_query::Output;
 use common_recordbatch::util::collect;
@@ -25,13 +27,14 @@ use datatypes::scalars::ScalarVector;
 use datatypes::vectors::{StringVector, Vector};
 use snafu::{OptionExt, ResultExt};
 use tokio::fs::File;
-use tokio::io::AsyncWriteExt;
+use tokio::io::{AsyncWriteExt, BufWriter};
 use tokio::sync::Semaphore;
 
 use crate::cli::{Instance, Tool};
 use crate::error::{
     CollectRecordBatchesSnafu, ConnectServerSnafu, EmptyResultSnafu, Error, FileIoSnafu,
-    InvalidDatabaseNameSnafu, NotDataFromOutputSnafu, RequestDatabaseSnafu, Result,
+    IllegalConfigSnafu, InvalidDatabaseNameSnafu, NotDataFromOutputSnafu, RequestDatabaseSnafu,
+    Result,
 };
 
 type TableReference = (String, String, String);
@@ -70,6 +73,10 @@ pub struct ExportCommand {
     /// Things to export
     #[clap(long, short = 't', value_enum)]
     target: ExportTarget,
+
+    /// basic authentication for connecting to the server
+    #[clap(long)]
+    auth_basic: Option<String>,
 }
 
 impl ExportCommand {
@@ -82,13 +89,23 @@ impl ExportCommand {
                 addr: self.addr.clone(),
             })?;
         let (catalog, schema) = split_database(&self.database)?;
-        let database_client = Database::new(
+        let mut database_client = Database::new(
             catalog.clone(),
             schema.clone().unwrap_or(DEFAULT_SCHEMA_NAME.to_string()),
             client,
         );
 
-        Ok(Instance::Tool(Box::new(Export {
+        if let Some(auth_basic) = &self.auth_basic {
+            let (username, password) = auth_basic.split_once(':').context(IllegalConfigSnafu {
+                msg: "auth_basic cannot be split by ':'".to_string(),
+            })?;
+            database_client.set_auth(AuthScheme::Basic(Basic {
+                username: username.to_string(),
+                password: password.to_string(),
+            }));
+        }
+
+        Ok(Instance::new(Box::new(Export {
             client: database_client,
             catalog,
             schema,
@@ -141,6 +158,9 @@ impl Export {
             let mut result = Vec::with_capacity(schemas.len());
             for i in 0..schemas.len() {
                 let schema = schemas.get_data(i).unwrap().to_owned();
+                if schema == common_catalog::consts::INFORMATION_SCHEMA_NAME {
+                    continue;
+                }
                 result.push((self.catalog.clone(), schema));
             }
             Ok(result)
@@ -326,25 +346,30 @@ impl Export {
 
                 let copy_from_file =
                     Path::new(&self.output_dir).join(format!("{catalog}-{schema}_copy_from.sql"));
-                let mut file = File::create(copy_from_file).await.context(FileIoSnafu)?;
+                let mut writer =
+                    BufWriter::new(File::create(copy_from_file).await.context(FileIoSnafu)?);
 
-                let copy_from_sql = dir_filenames
-                    .into_iter()
-                    .map(|file| {
-                        let file = file.unwrap();
-                        let filename = file.file_name().into_string().unwrap();
+                for table_file in dir_filenames {
+                    let table_file = table_file.unwrap();
+                    let table_name = table_file
+                        .file_name()
+                        .into_string()
+                        .unwrap()
+                        .replace(".parquet", "");
 
-                        format!(
-                            "copy {} from '{}' with (format='parquet');\n",
-                            filename.replace(".parquet", ""),
-                            file.path().to_str().unwrap()
+                    writer
+                        .write(
+                            format!(
+                                "copy {} from '{}' with (format='parquet');\n",
+                                table_name,
+                                table_file.path().to_str().unwrap()
+                            )
+                            .as_bytes(),
                         )
-                    })
-                    .collect::<Vec<_>>()
-                    .join("");
-                file.write_all(copy_from_sql.as_bytes())
-                    .await
-                    .context(FileIoSnafu)?;
+                        .await
+                        .context(FileIoSnafu)?;
+                }
+                writer.flush().await.context(FileIoSnafu)?;
 
                 info!("finished exporting {catalog}.{schema} copy_from.sql");
 

src/cmd/src/cli/repl.rs

@@ -17,7 +17,6 @@ use std::sync::Arc;
 use std::time::Instant;
 
 use catalog::kvbackend::{CachedMetaKvBackend, KvBackendCatalogManager};
-use client::client_manager::DatanodeClients;
 use client::{Client, Database, DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME};
 use common_base::Plugins;
 use common_error::ext::ErrorExt;
@@ -176,7 +175,7 @@ impl Repl {
                 .encode(&plan)
                 .context(SubstraitEncodeLogicalPlanSnafu)?;
 
-            self.database.logical_plan(plan.to_vec(), 0).await
+            self.database.logical_plan(plan.to_vec()).await
         } else {
             self.database.sql(&sql).await
         }
@@ -250,13 +249,8 @@ async fn create_query_engine(meta_addr: &str) -> Result<DatafusionQueryEngine> {
 
     let cached_meta_backend = Arc::new(CachedMetaKvBackend::new(meta_client.clone()));
 
-    let datanode_clients = Arc::new(DatanodeClients::default());
-
-    let catalog_list = KvBackendCatalogManager::new(
-        cached_meta_backend.clone(),
-        cached_meta_backend.clone(),
-        datanode_clients,
-    );
+    let catalog_list =
+        KvBackendCatalogManager::new(cached_meta_backend.clone(), cached_meta_backend);
     let plugins: Plugins = Default::default();
     let state = Arc::new(QueryEngineState::new(
         catalog_list,

src/cmd/src/cli/upgrade.rs

@@ -12,6 +12,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+use std::collections::HashMap;
 use std::sync::Arc;
 
 use async_trait::async_trait;
@@ -26,7 +27,9 @@ use common_meta::key::table_info::{TableInfoKey, TableInfoValue};
 use common_meta::key::table_name::{TableNameKey, TableNameValue};
 use common_meta::key::table_region::{TableRegionKey, TableRegionValue};
 use common_meta::key::table_route::{TableRouteKey, TableRouteValue as NextTableRouteValue};
-use common_meta::key::{RegionDistribution, TableMetaKey};
+use common_meta::key::{RegionDistribution, TableMetaKey, TableMetaValue};
+use common_meta::kv_backend::etcd::EtcdStore;
+use common_meta::kv_backend::KvBackendRef;
 use common_meta::range_stream::PaginationStream;
 use common_meta::rpc::router::TableRoute;
 use common_meta::rpc::store::{BatchDeleteRequest, BatchPutRequest, PutRequest, RangeRequest};
@@ -35,8 +38,6 @@ use common_meta::util::get_prefix_end_key;
 use common_telemetry::info;
 use etcd_client::Client;
 use futures::TryStreamExt;
-use meta_srv::service::store::etcd::EtcdStore;
-use meta_srv::service::store::kv::{KvBackendAdapter, KvStoreRef};
 use prost::Message;
 use snafu::ResultExt;
 use v1_helper::{CatalogKey as v1CatalogKey, SchemaKey as v1SchemaKey, TableGlobalValue};
@@ -76,12 +77,12 @@ impl UpgradeCommand {
             skip_schema_keys: self.skip_schema_keys,
             skip_table_route_keys: self.skip_table_route_keys,
         };
-        Ok(Instance::Tool(Box::new(tool)))
+        Ok(Instance::new(Box::new(tool)))
     }
 }
 
 struct MigrateTableMetadata {
-    etcd_store: KvStoreRef,
+    etcd_store: KvBackendRef,
     dryrun: bool,
 
     skip_table_global_keys: bool,
@@ -123,7 +124,7 @@ impl MigrateTableMetadata {
         info!("Start scanning key from: {}", String::from_utf8_lossy(&key));
 
         let mut stream = PaginationStream::new(
-            KvBackendAdapter::wrap(self.etcd_store.clone()),
+            self.etcd_store.clone(),
             RangeRequest::new().with_range(key, range_end),
             PAGE_SIZE,
             Arc::new(|kv: KeyValue| {
@@ -152,7 +153,7 @@ impl MigrateTableMetadata {
         )
         .unwrap();
 
-        let new_table_value = NextTableRouteValue::new(table_route.region_routes);
+        let new_table_value = NextTableRouteValue::physical(table_route.region_routes);
 
         let table_id = table_route.table.id as u32;
         let new_key = TableRouteKey::new(table_id);
@@ -182,7 +183,7 @@ impl MigrateTableMetadata {
         let mut keys = Vec::new();
         info!("Start scanning key from: {}", String::from_utf8_lossy(&key));
         let mut stream = PaginationStream::new(
-            KvBackendAdapter::wrap(self.etcd_store.clone()),
+            self.etcd_store.clone(),
             RangeRequest::new().with_range(key, range_end),
             PAGE_SIZE,
             Arc::new(|kv: KeyValue| {
@@ -234,7 +235,7 @@ impl MigrateTableMetadata {
         let mut keys = Vec::new();
         info!("Start scanning key from: {}", String::from_utf8_lossy(&key));
         let mut stream = PaginationStream::new(
-            KvBackendAdapter::wrap(self.etcd_store.clone()),
+            self.etcd_store.clone(),
             RangeRequest::new().with_range(key, range_end),
             PAGE_SIZE,
             Arc::new(|kv: KeyValue| {
@@ -284,7 +285,7 @@ impl MigrateTableMetadata {
 
         info!("Start scanning key from: {}", String::from_utf8_lossy(&key));
         let mut stream = PaginationStream::new(
-            KvBackendAdapter::wrap(self.etcd_store.clone()),
+            self.etcd_store.clone(),
             RangeRequest::new().with_range(key, range_end.clone()),
             PAGE_SIZE,
             Arc::new(|kv: KeyValue| {
@@ -395,6 +396,9 @@ impl MigrateTableMetadata {
         let region_distribution: RegionDistribution =
             value.regions_id_map.clone().into_iter().collect();
 
+        // TODO(niebayes): properly fetch or construct wal options.
+        let region_wal_options = HashMap::default();
+
         let datanode_table_kvs = region_distribution
             .into_iter()
             .map(|(datanode_id, regions)| {
@@ -409,6 +413,7 @@ impl MigrateTableMetadata {
                             engine: engine.to_string(),
                             region_storage_path: region_storage_path.clone(),
                             region_options: (&value.table_info.meta.options).into(),
+                            region_wal_options: region_wal_options.clone(),
                         },
                     ),
                 )

src/cmd/src/datanode.rs

@@ -12,25 +12,41 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+use std::sync::Arc;
 use std::time::Duration;
 
+use async_trait::async_trait;
+use catalog::kvbackend::MetaKvBackend;
 use clap::Parser;
-use common_telemetry::logging;
+use common_config::WalConfig;
+use common_telemetry::{info, logging};
 use datanode::config::DatanodeOptions;
 use datanode::datanode::{Datanode, DatanodeBuilder};
 use meta_client::MetaClientOptions;
 use servers::Mode;
-use snafu::ResultExt;
+use snafu::{OptionExt, ResultExt};
 
 use crate::error::{MissingConfigSnafu, Result, ShutdownDatanodeSnafu, StartDatanodeSnafu};
-use crate::options::{Options, TopLevelOptions};
+use crate::options::{CliOptions, Options};
+use crate::App;
 
 pub struct Instance {
     datanode: Datanode,
 }
 
 impl Instance {
-    pub async fn start(&mut self) -> Result<()> {
+    fn new(datanode: Datanode) -> Self {
+        Self { datanode }
+    }
+}
+
+#[async_trait]
+impl App for Instance {
+    fn name(&self) -> &str {
+        "greptime-datanode"
+    }
+
+    async fn start(&mut self) -> Result<()> {
         plugins::start_datanode_plugins(self.datanode.plugins())
             .await
             .context(StartDatanodeSnafu)?;
@@ -38,7 +54,7 @@ impl Instance {
         self.datanode.start().await.context(StartDatanodeSnafu)
     }
 
-    pub async fn stop(&self) -> Result<()> {
+    async fn stop(&self) -> Result<()> {
         self.datanode
             .shutdown()
             .await
@@ -57,8 +73,8 @@ impl Command {
         self.subcmd.build(opts).await
     }
 
-    pub fn load_options(&self, top_level_opts: TopLevelOptions) -> Result<Options> {
-        self.subcmd.load_options(top_level_opts)
+    pub fn load_options(&self, cli_options: &CliOptions) -> Result<Options> {
+        self.subcmd.load_options(cli_options)
     }
 }
 
@@ -74,9 +90,9 @@ impl SubCommand {
         }
     }
 
-    fn load_options(&self, top_level_opts: TopLevelOptions) -> Result<Options> {
+    fn load_options(&self, cli_options: &CliOptions) -> Result<Options> {
         match self {
-            SubCommand::Start(cmd) => cmd.load_options(top_level_opts),
+            SubCommand::Start(cmd) => cmd.load_options(cli_options),
         }
     }
 }
@@ -89,7 +105,7 @@ struct StartCommand {
     rpc_addr: Option<String>,
     #[clap(long)]
     rpc_hostname: Option<String>,
-    #[clap(long, multiple = true, value_delimiter = ',')]
+    #[clap(long, value_delimiter = ',', num_args = 1..)]
     metasrv_addr: Option<Vec<String>>,
     #[clap(short, long)]
     config_file: Option<String>,
@@ -106,19 +122,19 @@ struct StartCommand {
 }
 
 impl StartCommand {
-    fn load_options(&self, top_level_opts: TopLevelOptions) -> Result<Options> {
+    fn load_options(&self, cli_options: &CliOptions) -> Result<Options> {
         let mut opts: DatanodeOptions = Options::load_layered_options(
             self.config_file.as_deref(),
             self.env_prefix.as_ref(),
             DatanodeOptions::env_list_keys(),
         )?;
 
-        if let Some(dir) = top_level_opts.log_dir {
-            opts.logging.dir = dir;
+        if let Some(dir) = &cli_options.log_dir {
+            opts.logging.dir = dir.clone();
         }
 
-        if top_level_opts.log_level.is_some() {
-            opts.logging.level = top_level_opts.log_level;
+        if cli_options.log_level.is_some() {
+            opts.logging.level = cli_options.log_level.clone();
         }
 
         if let Some(addr) = &self.rpc_addr {
@@ -151,8 +167,18 @@ impl StartCommand {
             opts.storage.data_home = data_home.clone();
         }
 
-        if let Some(wal_dir) = &self.wal_dir {
-            opts.wal.dir = Some(wal_dir.clone());
+        // `wal_dir` only affects raft-engine config.
+        if let Some(wal_dir) = &self.wal_dir
+            && let WalConfig::RaftEngine(raft_engine_config) = &mut opts.wal
+        {
+            if raft_engine_config
+                .dir
+                .as_ref()
+                .is_some_and(|original_dir| original_dir != wal_dir)
+            {
+                info!("The wal dir of raft-engine is altered to {wal_dir}");
+            }
+            raft_engine_config.dir.replace(wal_dir.clone());
         }
 
         if let Some(http_addr) = &self.http_addr {
@@ -177,12 +203,32 @@ impl StartCommand {
         logging::info!("Datanode start command: {:#?}", self);
         logging::info!("Datanode options: {:#?}", opts);
 
-        let datanode = DatanodeBuilder::new(opts, None, plugins)
+        let node_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 = datanode::heartbeat::new_metasrv_client(node_id, meta_config)
+            .await
+            .context(StartDatanodeSnafu)?;
+
+        let meta_backend = Arc::new(MetaKvBackend {
+            client: Arc::new(meta_client.clone()),
+        });
+
+        let datanode = DatanodeBuilder::new(opts, plugins)
+            .with_meta_client(meta_client)
+            .with_kv_backend(meta_backend)
+            .enable_region_server_service()
+            .enable_http_service()
             .build()
             .await
             .context(StartDatanodeSnafu)?;
 
-        Ok(Instance { datanode })
+        Ok(Instance::new(datanode))
     }
 }
 
@@ -191,14 +237,13 @@ mod tests {
     use std::io::Write;
     use std::time::Duration;
 
-    use common_base::readable_size::ReadableSize;
     use common_test_util::temp_dir::create_named_temp_file;
-    use datanode::config::{CompactionConfig, FileConfig, ObjectStoreConfig, RegionManifestConfig};
+    use datanode::config::{FileConfig, GcsConfig, ObjectStoreConfig, S3Config};
     use servers::heartbeat_options::HeartbeatOptions;
     use servers::Mode;
 
     use super::*;
-    use crate::options::ENV_VAR_SEP;
+    use crate::options::{CliOptions, ENV_VAR_SEP};
 
     #[test]
     fn test_read_from_config_file() {
@@ -222,6 +267,7 @@ mod tests {
             tcp_nodelay = true
 
             [wal]
+            provider = "raft_engine"
             dir = "/other/wal"
             file_size = "1GB"
             purge_threshold = "50GB"
@@ -230,18 +276,17 @@ mod tests {
             sync_write = false
 
             [storage]
-            type = "File"
             data_home = "/tmp/greptimedb/"
+            type = "File"
 
-            [storage.compaction]
-            max_inflight_tasks = 3
-            max_files_in_level0 = 7
-            max_purge_tasks = 32
+            [[storage.providers]]
+            type = "Gcs"
+            bucket = "foo"
+            endpoint = "bar"
 
-            [storage.manifest]
-            checkpoint_margin = 9
-            gc_duration = '7s'
-            compress = true
+            [[storage.providers]]
+            type = "S3"
+            bucket = "foo"
 
             [logging]
             level = "debug"
@@ -254,19 +299,24 @@ mod tests {
             ..Default::default()
         };
 
-        let Options::Datanode(options) = cmd.load_options(TopLevelOptions::default()).unwrap()
-        else {
+        let Options::Datanode(options) = cmd.load_options(&CliOptions::default()).unwrap() else {
             unreachable!()
         };
 
         assert_eq!("127.0.0.1:3001".to_string(), options.rpc_addr);
         assert_eq!(Some(42), options.node_id);
-        assert_eq!("/other/wal", options.wal.dir.unwrap());
 
-        assert_eq!(Duration::from_secs(600), options.wal.purge_interval);
-        assert_eq!(1024 * 1024 * 1024, options.wal.file_size.0);
-        assert_eq!(1024 * 1024 * 1024 * 50, options.wal.purge_threshold.0);
-        assert!(!options.wal.sync_write);
+        let WalConfig::RaftEngine(raft_engine_config) = options.wal else {
+            unreachable!()
+        };
+        assert_eq!("/other/wal", raft_engine_config.dir.unwrap());
+        assert_eq!(Duration::from_secs(600), raft_engine_config.purge_interval);
+        assert_eq!(1024 * 1024 * 1024, raft_engine_config.file_size.0);
+        assert_eq!(
+            1024 * 1024 * 1024 * 50,
+            raft_engine_config.purge_threshold.0
+        );
+        assert!(!raft_engine_config.sync_write);
 
         let HeartbeatOptions {
             interval: heart_beat_interval,
@@ -294,24 +344,15 @@ mod tests {
             &options.storage.store,
             ObjectStoreConfig::File(FileConfig { .. })
         ));
-
-        assert_eq!(
-            CompactionConfig {
-                max_inflight_tasks: 3,
-                max_files_in_level0: 7,
-                max_purge_tasks: 32,
-                sst_write_buffer_size: ReadableSize::mb(8),
-            },
-            options.storage.compaction,
-        );
-        assert_eq!(
-            RegionManifestConfig {
-                checkpoint_margin: Some(9),
-                gc_duration: Some(Duration::from_secs(7)),
-                compress: true
-            },
-            options.storage.manifest,
-        );
+        assert_eq!(options.storage.providers.len(), 2);
+        assert!(matches!(
+            options.storage.providers[0],
+            ObjectStoreConfig::Gcs(GcsConfig { .. })
+        ));
+        assert!(matches!(
+            options.storage.providers[1],
+            ObjectStoreConfig::S3(S3Config { .. })
+        ));
 
         assert_eq!("debug", options.logging.level.unwrap());
         assert_eq!("/tmp/greptimedb/test/logs".to_string(), options.logging.dir);
@@ -320,7 +361,7 @@ mod tests {
     #[test]
     fn test_try_from_cmd() {
         if let Options::Datanode(opt) = StartCommand::default()
-            .load_options(TopLevelOptions::default())
+            .load_options(&CliOptions::default())
             .unwrap()
         {
             assert_eq!(Mode::Standalone, opt.mode)
@@ -331,7 +372,7 @@ mod tests {
             metasrv_addr: Some(vec!["127.0.0.1:3002".to_string()]),
             ..Default::default()
         })
-        .load_options(TopLevelOptions::default())
+        .load_options(&CliOptions::default())
         .unwrap()
         {
             assert_eq!(Mode::Distributed, opt.mode)
@@ -341,7 +382,7 @@ mod tests {
             metasrv_addr: Some(vec!["127.0.0.1:3002".to_string()]),
             ..Default::default()
         })
-        .load_options(TopLevelOptions::default())
+        .load_options(&CliOptions::default())
         .is_err());
 
         // Providing node_id but leave metasrv_addr absent is ok since metasrv_addr has default value
@@ -349,18 +390,21 @@ mod tests {
             node_id: Some(42),
             ..Default::default()
         })
-        .load_options(TopLevelOptions::default())
+        .load_options(&CliOptions::default())
         .is_ok());
     }
 
     #[test]
-    fn test_top_level_options() {
+    fn test_load_log_options_from_cli() {
         let cmd = StartCommand::default();
 
         let options = cmd
-            .load_options(TopLevelOptions {
+            .load_options(&CliOptions {
                 log_dir: Some("/tmp/greptimedb/test/logs".to_string()),
                 log_level: Some("debug".to_string()),
+
+                #[cfg(feature = "tokio-console")]
+                tokio_console_addr: None,
             })
             .unwrap();
 
@@ -386,21 +430,16 @@ mod tests {
             tcp_nodelay = true
 
             [wal]
+            provider = "raft_engine"
             file_size = "1GB"
             purge_threshold = "50GB"
-            purge_interval = "10m"
-            read_batch_size = 128
+            purge_interval = "5m"
             sync_write = false
 
             [storage]
             type = "File"
             data_home = "/tmp/greptimedb/"
 
-            [storage.compaction]
-            max_inflight_tasks = 3
-            max_files_in_level0 = 7
-            max_purge_tasks = 32
-
             [logging]
             level = "debug"
             dir = "/tmp/greptimedb/test/logs"
@@ -411,26 +450,24 @@ mod tests {
         temp_env::with_vars(
             [
                 (
-                    // storage.manifest.gc_duration = 9s
+                    // wal.purge_interval = 1m
                     [
                         env_prefix.to_string(),
-                        "storage".to_uppercase(),
-                        "manifest".to_uppercase(),
-                        "gc_duration".to_uppercase(),
+                        "wal".to_uppercase(),
+                        "purge_interval".to_uppercase(),
                     ]
                     .join(ENV_VAR_SEP),
-                    Some("9s"),
+                    Some("1m"),
                 ),
                 (
-                    // storage.compaction.max_purge_tasks = 99
+                    // wal.read_batch_size = 100
                     [
                         env_prefix.to_string(),
-                        "storage".to_uppercase(),
-                        "compaction".to_uppercase(),
-                        "max_purge_tasks".to_uppercase(),
+                        "wal".to_uppercase(),
+                        "read_batch_size".to_uppercase(),
                     ]
                     .join(ENV_VAR_SEP),
-                    Some("99"),
+                    Some("100"),
                 ),
                 (
                     // meta_client.metasrv_addrs = 127.0.0.1:3001,127.0.0.1:3002,127.0.0.1:3003
@@ -451,17 +488,16 @@ mod tests {
                     ..Default::default()
                 };
 
-                let Options::Datanode(opts) =
-                    command.load_options(TopLevelOptions::default()).unwrap()
+                let Options::Datanode(opts) = command.load_options(&CliOptions::default()).unwrap()
                 else {
                     unreachable!()
                 };
 
                 // Should be read from env, env > default values.
-                assert_eq!(
-                    opts.storage.manifest.gc_duration,
-                    Some(Duration::from_secs(9))
-                );
+                let WalConfig::RaftEngine(raft_engine_config) = opts.wal else {
+                    unreachable!()
+                };
+                assert_eq!(raft_engine_config.read_batch_size, 100);
                 assert_eq!(
                     opts.meta_client.unwrap().metasrv_addrs,
                     vec![
@@ -472,19 +508,16 @@ mod tests {
                 );
 
                 // Should be read from config file, config file > env > default values.
-                assert_eq!(opts.storage.compaction.max_purge_tasks, 32);
+                assert_eq!(
+                    raft_engine_config.purge_interval,
+                    Duration::from_secs(60 * 5)
+                );
 
                 // Should be read from cli, cli > config file > env > default values.
-                assert_eq!(opts.wal.dir.unwrap(), "/other/wal/dir");
+                assert_eq!(raft_engine_config.dir.unwrap(), "/other/wal/dir");
 
                 // Should be default value.
-                assert_eq!(
-                    opts.storage.manifest.checkpoint_margin,
-                    DatanodeOptions::default()
-                        .storage
-                        .manifest
-                        .checkpoint_margin
-                );
+                assert_eq!(opts.http.addr, DatanodeOptions::default().http.addr);
             },
         );
     }

src/cmd/src/error.rs

@@ -14,7 +14,7 @@
 
 use std::any::Any;
 
-use common_error::ext::ErrorExt;
+use common_error::ext::{BoxedError, ErrorExt};
 use common_error::status_code::StatusCode;
 use common_macro::stack_trace_debug;
 use config::ConfigError;
@@ -37,6 +37,18 @@ pub enum Error {
         source: common_meta::error::Error,
     },
 
+    #[snafu(display("Failed to init DDL manager"))]
+    InitDdlManager {
+        location: Location,
+        source: common_meta::error::Error,
+    },
+
+    #[snafu(display("Failed to init default timezone"))]
+    InitTimezone {
+        location: Location,
+        source: common_time::error::Error,
+    },
+
     #[snafu(display("Failed to start procedure manager"))]
     StartProcedureManager {
         location: Location,
@@ -49,6 +61,12 @@ pub enum Error {
         source: common_procedure::error::Error,
     },
 
+    #[snafu(display("Failed to start wal options allocator"))]
+    StartWalOptionsAllocator {
+        location: Location,
+        source: common_meta::error::Error,
+    },
+
     #[snafu(display("Failed to start datanode"))]
     StartDatanode {
         location: Location,
@@ -225,6 +243,12 @@ pub enum Error {
         #[snafu(source)]
         error: std::io::Error,
     },
+
+    #[snafu(display("Other error"))]
+    Other {
+        source: BoxedError,
+        location: Location,
+    },
 }
 
 pub type Result<T> = std::result::Result<T, Error>;
@@ -240,21 +264,26 @@ impl ErrorExt for Error {
             Error::ShutdownMetaServer { source, .. } => source.status_code(),
             Error::BuildMetaServer { source, .. } => source.status_code(),
             Error::UnsupportedSelectorType { source, .. } => source.status_code(),
-            Error::IterStream { source, .. } | Error::InitMetadata { source, .. } => {
-                source.status_code()
-            }
+
+            Error::IterStream { source, .. }
+            | Error::InitMetadata { source, .. }
+            | Error::InitDdlManager { source, .. } => source.status_code(),
+
             Error::ConnectServer { source, .. } => source.status_code(),
             Error::MissingConfig { .. }
             | Error::LoadLayeredConfig { .. }
             | Error::IllegalConfig { .. }
             | Error::InvalidReplCommand { .. }
+            | Error::InitTimezone { .. }
             | Error::ConnectEtcd { .. }
             | Error::NotDataFromOutput { .. }
             | Error::CreateDir { .. }
             | Error::EmptyResult { .. }
             | Error::InvalidDatabaseName { .. } => StatusCode::InvalidArguments,
+
             Error::StartProcedureManager { source, .. }
             | Error::StopProcedureManager { source, .. } => source.status_code(),
+            Error::StartWalOptionsAllocator { source, .. } => source.status_code(),
             Error::ReplCreation { .. } | Error::Readline { .. } => StatusCode::Internal,
             Error::RequestDatabase { source, .. } => source.status_code(),
             Error::CollectRecordBatches { source, .. }
@@ -267,6 +296,8 @@ impl ErrorExt for Error {
             Error::StartCatalogManager { source, .. } => source.status_code(),
 
             Error::SerdeJson { .. } | Error::FileIo { .. } => StatusCode::Unexpected,
+
+            Error::Other { source, .. } => source.status_code(),
         }
     }
 

src/cmd/src/frontend.rs

@@ -12,26 +12,48 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+use std::sync::Arc;
 use std::time::Duration;
 
+use async_trait::async_trait;
+use catalog::kvbackend::CachedMetaKvBackend;
 use clap::Parser;
+use client::client_manager::DatanodeClients;
+use common_meta::heartbeat::handler::parse_mailbox_message::ParseMailboxMessageHandler;
+use common_meta::heartbeat::handler::HandlerGroupExecutor;
 use common_telemetry::logging;
+use common_time::timezone::set_default_timezone;
 use frontend::frontend::FrontendOptions;
+use frontend::heartbeat::handler::invalidate_table_cache::InvalidateTableCacheHandler;
+use frontend::heartbeat::HeartbeatTask;
+use frontend::instance::builder::FrontendBuilder;
 use frontend::instance::{FrontendInstance, Instance as FeInstance};
 use meta_client::MetaClientOptions;
 use servers::tls::{TlsMode, TlsOption};
 use servers::Mode;
-use snafu::ResultExt;
+use snafu::{OptionExt, ResultExt};
 
-use crate::error::{self, Result, StartFrontendSnafu};
-use crate::options::{Options, TopLevelOptions};
+use crate::error::{self, InitTimezoneSnafu, MissingConfigSnafu, Result, StartFrontendSnafu};
+use crate::options::{CliOptions, Options};
+use crate::App;
 
 pub struct Instance {
     frontend: FeInstance,
 }
 
 impl Instance {
-    pub async fn start(&mut self) -> Result<()> {
+    fn new(frontend: FeInstance) -> Self {
+        Self { frontend }
+    }
+}
+
+#[async_trait]
+impl App for Instance {
+    fn name(&self) -> &str {
+        "greptime-frontend"
+    }
+
+    async fn start(&mut self) -> Result<()> {
         plugins::start_frontend_plugins(self.frontend.plugins().clone())
             .await
             .context(StartFrontendSnafu)?;
@@ -39,7 +61,7 @@ impl Instance {
         self.frontend.start().await.context(StartFrontendSnafu)
     }
 
-    pub async fn stop(&self) -> Result<()> {
+    async fn stop(&self) -> Result<()> {
         self.frontend
             .shutdown()
             .await
@@ -58,8 +80,8 @@ impl Command {
         self.subcmd.build(opts).await
     }
 
-    pub fn load_options(&self, top_level_opts: TopLevelOptions) -> Result<Options> {
-        self.subcmd.load_options(top_level_opts)
+    pub fn load_options(&self, cli_options: &CliOptions) -> Result<Options> {
+        self.subcmd.load_options(cli_options)
     }
 }
 
@@ -75,9 +97,9 @@ impl SubCommand {
         }
     }
 
-    fn load_options(&self, top_level_opts: TopLevelOptions) -> Result<Options> {
+    fn load_options(&self, cli_options: &CliOptions) -> Result<Options> {
         match self {
-            SubCommand::Start(cmd) => cmd.load_options(top_level_opts),
+            SubCommand::Start(cmd) => cmd.load_options(cli_options),
         }
     }
 }
@@ -100,7 +122,7 @@ pub struct StartCommand {
     config_file: Option<String>,
     #[clap(short, long)]
     influxdb_enable: Option<bool>,
-    #[clap(long, multiple = true, value_delimiter = ',')]
+    #[clap(long, value_delimiter = ',', num_args = 1..)]
     metasrv_addr: Option<Vec<String>>,
     #[clap(long)]
     tls_mode: Option<TlsMode>,
@@ -117,19 +139,19 @@ pub struct StartCommand {
 }
 
 impl StartCommand {
-    fn load_options(&self, top_level_opts: TopLevelOptions) -> Result<Options> {
+    fn load_options(&self, cli_options: &CliOptions) -> Result<Options> {
         let mut opts: FrontendOptions = Options::load_layered_options(
             self.config_file.as_deref(),
             self.env_prefix.as_ref(),
             FrontendOptions::env_list_keys(),
         )?;
 
-        if let Some(dir) = top_level_opts.log_dir {
-            opts.logging.dir = dir;
+        if let Some(dir) = &cli_options.log_dir {
+            opts.logging.dir = dir.clone();
         }
 
-        if top_level_opts.log_level.is_some() {
-            opts.logging.level = top_level_opts.log_level;
+        if cli_options.log_level.is_some() {
+            opts.logging.level = cli_options.log_level.clone();
         }
 
         let tls_opts = TlsOption::new(
@@ -188,6 +210,7 @@ impl StartCommand {
     }
 
     async fn build(self, mut opts: FrontendOptions) -> Result<Instance> {
+        #[allow(clippy::unnecessary_mut_passed)]
         let plugins = plugins::setup_frontend_plugins(&mut opts)
             .await
             .context(StartFrontendSnafu)?;
@@ -195,16 +218,50 @@ impl StartCommand {
         logging::info!("Frontend start command: {:#?}", self);
         logging::info!("Frontend options: {:#?}", opts);
 
-        let mut instance = FeInstance::try_new_distributed(&opts, plugins.clone())
+        set_default_timezone(opts.default_timezone.as_deref()).context(InitTimezoneSnafu)?;
+
+        let meta_client_options = opts.meta_client.as_ref().context(MissingConfigSnafu {
+            msg: "'meta_client'",
+        })?;
+        let meta_client = FeInstance::create_meta_client(meta_client_options)
             .await
             .context(StartFrontendSnafu)?;
 
+        let meta_backend = Arc::new(CachedMetaKvBackend::new(meta_client.clone()));
+
+        let executor = HandlerGroupExecutor::new(vec![
+            Arc::new(ParseMailboxMessageHandler),
+            Arc::new(InvalidateTableCacheHandler::new(meta_backend.clone())),
+        ]);
+
+        let heartbeat_task = HeartbeatTask::new(
+            meta_client.clone(),
+            opts.heartbeat.clone(),
+            Arc::new(executor),
+        );
+
+        let mut instance = FrontendBuilder::new(
+            meta_backend.clone(),
+            Arc::new(DatanodeClients::default()),
+            meta_client,
+        )
+        .with_cache_invalidator(meta_backend)
+        .with_plugin(plugins)
+        .with_heartbeat_task(heartbeat_task)
+        .try_build()
+        .await
+        .context(StartFrontendSnafu)?;
+
+        instance
+            .build_export_metrics_task(&opts.export_metrics)
+            .context(StartFrontendSnafu)?;
+
         instance
-            .build_servers(&opts)
+            .build_servers(opts)
             .await
             .context(StartFrontendSnafu)?;
 
-        Ok(Instance { frontend: instance })
+        Ok(Instance::new(instance))
     }
 }
 
@@ -220,7 +277,7 @@ mod tests {
     use servers::http::HttpOptions;
 
     use super::*;
-    use crate::options::ENV_VAR_SEP;
+    use crate::options::{CliOptions, ENV_VAR_SEP};
 
     #[test]
     fn test_try_from_start_command() {
@@ -234,8 +291,7 @@ mod tests {
             ..Default::default()
         };
 
-        let Options::Frontend(opts) = command.load_options(TopLevelOptions::default()).unwrap()
-        else {
+        let Options::Frontend(opts) = command.load_options(&CliOptions::default()).unwrap() else {
             unreachable!()
         };
 
@@ -287,7 +343,7 @@ mod tests {
             ..Default::default()
         };
 
-        let Options::Frontend(fe_opts) = command.load_options(TopLevelOptions::default()).unwrap()
+        let Options::Frontend(fe_opts) = command.load_options(&CliOptions::default()).unwrap()
         else {
             unreachable!()
         };
@@ -312,6 +368,7 @@ mod tests {
             ..Default::default()
         };
 
+        #[allow(clippy::unnecessary_mut_passed)]
         let plugins = plugins::setup_frontend_plugins(&mut fe_opts).await.unwrap();
 
         let provider = plugins.get::<UserProviderRef>().unwrap();
@@ -325,16 +382,19 @@ mod tests {
     }
 
     #[test]
-    fn test_top_level_options() {
+    fn test_load_log_options_from_cli() {
         let cmd = StartCommand {
             disable_dashboard: Some(false),
             ..Default::default()
         };
 
         let options = cmd
-            .load_options(TopLevelOptions {
+            .load_options(&CliOptions {
                 log_dir: Some("/tmp/greptimedb/test/logs".to_string()),
                 log_level: Some("debug".to_string()),
+
+                #[cfg(feature = "tokio-console")]
+                tokio_console_addr: None,
             })
             .unwrap();
 
@@ -414,11 +474,8 @@ mod tests {
                     ..Default::default()
                 };
 
-                let top_level_opts = TopLevelOptions {
-                    log_dir: None,
-                    log_level: Some("error".to_string()),
-                };
-                let Options::Frontend(fe_opts) = command.load_options(top_level_opts).unwrap()
+                let Options::Frontend(fe_opts) =
+                    command.load_options(&CliOptions::default()).unwrap()
                 else {
                     unreachable!()
                 };

src/cmd/src/lib.rs

@@ -12,7 +12,11 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-#![feature(assert_matches)]
+#![feature(assert_matches, let_chains)]
+
+use async_trait::async_trait;
+use clap::arg;
+use common_telemetry::{error, info};
 
 pub mod cli;
 pub mod datanode;
@@ -21,3 +25,100 @@ pub mod frontend;
 pub mod metasrv;
 pub mod options;
 pub mod standalone;
+
+lazy_static::lazy_static! {
+    static ref APP_VERSION: prometheus::IntGaugeVec =
+        prometheus::register_int_gauge_vec!("app_version", "app version", &["short_version", "version"]).unwrap();
+}
+
+#[async_trait]
+pub trait App {
+    fn name(&self) -> &str;
+
+    async fn start(&mut self) -> error::Result<()>;
+
+    async fn stop(&self) -> error::Result<()>;
+}
+
+pub async fn start_app(mut app: Box<dyn App>) -> error::Result<()> {
+    let name = app.name().to_string();
+
+    tokio::select! {
+        result = app.start() => {
+            if let Err(err) = result {
+                error!(err; "Failed to start app {name}!");
+            }
+        }
+        _ = tokio::signal::ctrl_c() => {
+            if let Err(err) = app.stop().await {
+                error!(err; "Failed to stop app {name}!");
+            }
+            info!("Goodbye!");
+        }
+    }
+
+    Ok(())
+}
+
+pub fn log_versions() {
+    // Report app version as gauge.
+    APP_VERSION
+        .with_label_values(&[short_version(), full_version()])
+        .inc();
+
+    // Log version and argument flags.
+    info!(
+        "short_version: {}, full_version: {}",
+        short_version(),
+        full_version()
+    );
+
+    log_env_flags();
+}
+
+pub fn greptimedb_cli() -> clap::Command {
+    let cmd = clap::Command::new("greptimedb")
+        .version(print_version())
+        .subcommand_required(true);
+
+    #[cfg(feature = "tokio-console")]
+    let cmd = cmd.arg(arg!(--"tokio-console-addr"[TOKIO_CONSOLE_ADDR]));
+
+    cmd.args([arg!(--"log-dir"[LOG_DIR]), arg!(--"log-level"[LOG_LEVEL])])
+}
+
+fn print_version() -> &'static str {
+    concat!(
+        "\nbranch: ",
+        env!("GIT_BRANCH"),
+        "\ncommit: ",
+        env!("GIT_COMMIT"),
+        "\ndirty: ",
+        env!("GIT_DIRTY"),
+        "\nversion: ",
+        env!("CARGO_PKG_VERSION")
+    )
+}
+
+fn short_version() -> &'static str {
+    env!("CARGO_PKG_VERSION")
+}
+
+// {app_name}-{branch_name}-{commit_short}
+// The branch name (tag) of a release build should already contain the short
+// version so the full version doesn't concat the short version explicitly.
+fn full_version() -> &'static str {
+    concat!(
+        "greptimedb-",
+        env!("GIT_BRANCH"),
+        "-",
+        env!("GIT_COMMIT_SHORT")
+    )
+}
+
+fn log_env_flags() {
+    info!("command line arguments");
+    for argument in std::env::args() {
+        info!("argument: {}", argument);
+    }
+}

src/cmd/src/metasrv.rs

@@ -14,6 +14,7 @@
 
 use std::time::Duration;
 
+use async_trait::async_trait;
 use clap::Parser;
 use common_telemetry::logging;
 use meta_srv::bootstrap::MetaSrvInstance;
@@ -21,21 +22,34 @@ use meta_srv::metasrv::MetaSrvOptions;
 use snafu::ResultExt;
 
 use crate::error::{self, Result, StartMetaServerSnafu};
-use crate::options::{Options, TopLevelOptions};
+use crate::options::{CliOptions, Options};
+use crate::App;
 
 pub struct Instance {
     instance: MetaSrvInstance,
 }
 
 impl Instance {
-    pub async fn start(&mut self) -> Result<()> {
+    fn new(instance: MetaSrvInstance) -> Self {
+        Self { instance }
+    }
+}
+
+#[async_trait]
+impl App for Instance {
+    fn name(&self) -> &str {
+        "greptime-metasrv"
+    }
+
+    async fn start(&mut self) -> Result<()> {
         plugins::start_meta_srv_plugins(self.instance.plugins())
             .await
             .context(StartMetaServerSnafu)?;
+
         self.instance.start().await.context(StartMetaServerSnafu)
     }
 
-    pub async fn stop(&self) -> Result<()> {
+    async fn stop(&self) -> Result<()> {
         self.instance
             .shutdown()
             .await
@@ -54,8 +68,8 @@ impl Command {
         self.subcmd.build(opts).await
     }
 
-    pub fn load_options(&self, top_level_opts: TopLevelOptions) -> Result<Options> {
-        self.subcmd.load_options(top_level_opts)
+    pub fn load_options(&self, cli_options: &CliOptions) -> Result<Options> {
+        self.subcmd.load_options(cli_options)
     }
 }
 
@@ -71,9 +85,9 @@ impl SubCommand {
         }
     }
 
-    fn load_options(&self, top_level_opts: TopLevelOptions) -> Result<Options> {
+    fn load_options(&self, cli_options: &CliOptions) -> Result<Options> {
         match self {
-            SubCommand::Start(cmd) => cmd.load_options(top_level_opts),
+            SubCommand::Start(cmd) => cmd.load_options(cli_options),
         }
     }
 }
@@ -100,22 +114,29 @@ struct StartCommand {
     http_timeout: Option<u64>,
     #[clap(long, default_value = "GREPTIMEDB_METASRV")]
     env_prefix: String,
+    /// The working home directory of this metasrv instance.
+    #[clap(long)]
+    data_home: Option<String>,
+
+    /// If it's not empty, the metasrv will store all data with this key prefix.
+    #[clap(long, default_value = "")]
+    store_key_prefix: String,
 }
 
 impl StartCommand {
-    fn load_options(&self, top_level_opts: TopLevelOptions) -> Result<Options> {
+    fn load_options(&self, cli_options: &CliOptions) -> Result<Options> {
         let mut opts: MetaSrvOptions = Options::load_layered_options(
             self.config_file.as_deref(),
             self.env_prefix.as_ref(),
             None,
         )?;
 
-        if let Some(dir) = top_level_opts.log_dir {
-            opts.logging.dir = dir;
+        if let Some(dir) = &cli_options.log_dir {
+            opts.logging.dir = dir.clone();
         }
 
-        if top_level_opts.log_level.is_some() {
-            opts.logging.level = top_level_opts.log_level;
+        if cli_options.log_level.is_some() {
+            opts.logging.level = cli_options.log_level.clone();
         }
 
         if let Some(addr) = &self.bind_addr {
@@ -152,6 +173,14 @@ impl StartCommand {
             opts.http.timeout = Duration::from_secs(http_timeout);
         }
 
+        if let Some(data_home) = &self.data_home {
+            opts.data_home = data_home.clone();
+        }
+
+        if !self.store_key_prefix.is_empty() {
+            opts.store_key_prefix = self.store_key_prefix.clone()
+        }
+
         // Disable dashboard in metasrv.
         opts.http.disable_dashboard = true;
 
@@ -166,11 +195,16 @@ impl StartCommand {
         logging::info!("MetaSrv start command: {:#?}", self);
         logging::info!("MetaSrv options: {:#?}", opts);
 
-        let instance = MetaSrvInstance::new(opts, plugins)
+        let builder = meta_srv::bootstrap::metasrv_builder(&opts, plugins.clone(), None)
+            .await
+            .context(error::BuildMetaServerSnafu)?;
+        let metasrv = builder.build().await.context(error::BuildMetaServerSnafu)?;
+
+        let instance = MetaSrvInstance::new(opts, plugins, metasrv)
             .await
             .context(error::BuildMetaServerSnafu)?;
 
-        Ok(Instance { instance })
+        Ok(Instance::new(instance))
     }
 }
 
@@ -194,8 +228,7 @@ mod tests {
             ..Default::default()
         };
 
-        let Options::Metasrv(options) = cmd.load_options(TopLevelOptions::default()).unwrap()
-        else {
+        let Options::Metasrv(options) = cmd.load_options(&CliOptions::default()).unwrap() else {
             unreachable!()
         };
         assert_eq!("127.0.0.1:3002".to_string(), options.bind_addr);
@@ -216,6 +249,12 @@ mod tests {
             [logging]
             level = "debug"
             dir = "/tmp/greptimedb/test/logs"
+            
+            [failure_detector]
+            threshold = 8.0
+            min_std_deviation = "100ms"
+            acceptable_heartbeat_pause = "3000ms"
+            first_heartbeat_estimate = "1000ms"
         "#;
         write!(file, "{}", toml_str).unwrap();
 
@@ -224,8 +263,7 @@ mod tests {
             ..Default::default()
         };
 
-        let Options::Metasrv(options) = cmd.load_options(TopLevelOptions::default()).unwrap()
-        else {
+        let Options::Metasrv(options) = cmd.load_options(&CliOptions::default()).unwrap() else {
             unreachable!()
         };
         assert_eq!("127.0.0.1:3002".to_string(), options.bind_addr);
@@ -234,10 +272,29 @@ mod tests {
         assert_eq!(SelectorType::LeaseBased, options.selector);
         assert_eq!("debug", options.logging.level.as_ref().unwrap());
         assert_eq!("/tmp/greptimedb/test/logs".to_string(), options.logging.dir);
+        assert_eq!(8.0, options.failure_detector.threshold);
+        assert_eq!(
+            100.0,
+            options.failure_detector.min_std_deviation.as_millis() as f32
+        );
+        assert_eq!(
+            3000,
+            options
+                .failure_detector
+                .acceptable_heartbeat_pause
+                .as_millis()
+        );
+        assert_eq!(
+            1000,
+            options
+                .failure_detector
+                .first_heartbeat_estimate
+                .as_millis()
+        );
     }
 
     #[test]
-    fn test_top_level_options() {
+    fn test_load_log_options_from_cli() {
         let cmd = StartCommand {
             bind_addr: Some("127.0.0.1:3002".to_string()),
             server_addr: Some("127.0.0.1:3002".to_string()),
@@ -247,9 +304,12 @@ mod tests {
         };
 
         let options = cmd
-            .load_options(TopLevelOptions {
+            .load_options(&CliOptions {
                 log_dir: Some("/tmp/greptimedb/test/logs".to_string()),
                 log_level: Some("debug".to_string()),
+
+                #[cfg(feature = "tokio-console")]
+                tokio_console_addr: None,
             })
             .unwrap();
 
@@ -308,8 +368,7 @@ mod tests {
                     ..Default::default()
                 };
 
-                let Options::Metasrv(opts) =
-                    command.load_options(TopLevelOptions::default()).unwrap()
+                let Options::Metasrv(opts) = command.load_options(&CliOptions::default()).unwrap()
                 else {
                     unreachable!()
                 };

src/cmd/src/options.rs

@@ -12,11 +12,14 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-use common_config::KvStoreConfig;
-use common_telemetry::logging::LoggingOptions;
+use clap::ArgMatches;
+use common_config::KvBackendConfig;
+use common_meta::wal::WalConfig as MetaSrvWalConfig;
+use common_telemetry::logging::{LoggingOptions, TracingOptions};
 use config::{Config, Environment, File, FileFormat};
 use datanode::config::{DatanodeOptions, ProcedureConfig};
-use frontend::frontend::FrontendOptions;
+use frontend::error::{Result as FeResult, TomlFormatSnafu};
+use frontend::frontend::{FrontendOptions, TomlSerializable};
 use meta_srv::metasrv::MetaSrvOptions;
 use serde::{Deserialize, Serialize};
 use snafu::ResultExt;
@@ -27,13 +30,27 @@ pub const ENV_VAR_SEP: &str = "__";
 pub const ENV_LIST_SEP: &str = ",";
 
 /// Options mixed up from datanode, frontend and metasrv.
+#[derive(Serialize, Debug, Clone)]
 pub struct MixOptions {
     pub data_home: String,
     pub procedure: ProcedureConfig,
-    pub metadata_store: KvStoreConfig,
+    pub metadata_store: KvBackendConfig,
     pub frontend: FrontendOptions,
     pub datanode: DatanodeOptions,
     pub logging: LoggingOptions,
+    pub wal_meta: MetaSrvWalConfig,
+}
+
+impl From<MixOptions> for FrontendOptions {
+    fn from(value: MixOptions) -> Self {
+        value.frontend
+    }
+}
+
+impl TomlSerializable for MixOptions {
+    fn to_toml(&self) -> FeResult<String> {
+        toml::to_string(self).context(TomlFormatSnafu)
+    }
 }
 
 pub enum Options {
@@ -44,10 +61,32 @@ pub enum Options {
     Cli(Box<LoggingOptions>),
 }
 
-#[derive(Clone, Debug, Default)]
-pub struct TopLevelOptions {
+#[derive(Default)]
+pub struct CliOptions {
     pub log_dir: Option<String>,
     pub log_level: Option<String>,
+
+    #[cfg(feature = "tokio-console")]
+    pub tokio_console_addr: Option<String>,
+}
+
+impl CliOptions {
+    pub fn new(args: &ArgMatches) -> Self {
+        Self {
+            log_dir: args.get_one::<String>("log-dir").cloned(),
+            log_level: args.get_one::<String>("log-level").cloned(),
+
+            #[cfg(feature = "tokio-console")]
+            tokio_console_addr: args.get_one::<String>("tokio-console-addr").cloned(),
+        }
+    }
+
+    pub fn tracing_options(&self) -> TracingOptions {
+        TracingOptions {
+            #[cfg(feature = "tokio-console")]
+            tokio_console_addr: self.tokio_console_addr.clone(),
+        }
+    }
 }
 
 impl Options {
@@ -119,13 +158,22 @@ impl Options {
 
         Ok(opts)
     }
+
+    pub fn node_id(&self) -> Option<String> {
+        match self {
+            Options::Metasrv(_) | Options::Cli(_) => None,
+            Options::Datanode(opt) => opt.node_id.map(|x| x.to_string()),
+            Options::Frontend(opt) => opt.node_id.clone(),
+            Options::Standalone(opt) => opt.frontend.node_id.clone(),
+        }
+    }
 }
 
 #[cfg(test)]
 mod tests {
     use std::io::Write;
-    use std::time::Duration;
 
+    use common_config::WalConfig;
     use common_test_util::temp_dir::create_named_temp_file;
     use datanode::config::{DatanodeOptions, ObjectStoreConfig};
 
@@ -149,6 +197,7 @@ mod tests {
             tcp_nodelay = true
 
             [wal]
+            provider = "raft_engine"
             dir = "/tmp/greptimedb/wal"
             file_size = "1GB"
             purge_threshold = "50GB"
@@ -156,11 +205,6 @@ mod tests {
             read_batch_size = 128
             sync_write = false
 
-            [storage.compaction]
-            max_inflight_tasks = 3
-            max_files_in_level0 = 7
-            max_purge_tasks = 32
-
             [logging]
             level = "debug"
             dir = "/tmp/greptimedb/test/logs"
@@ -171,17 +215,6 @@ mod tests {
         temp_env::with_vars(
             // The following environment variables will be used to override the values in the config file.
             [
-                (
-                    // storage.manifest.checkpoint_margin = 99
-                    [
-                        env_prefix.to_string(),
-                        "storage".to_uppercase(),
-                        "manifest".to_uppercase(),
-                        "checkpoint_margin".to_uppercase(),
-                    ]
-                    .join(ENV_VAR_SEP),
-                    Some("99"),
-                ),
                 (
                     // storage.type = S3
                     [
@@ -202,17 +235,6 @@ mod tests {
                     .join(ENV_VAR_SEP),
                     Some("mybucket"),
                 ),
-                (
-                    // storage.manifest.gc_duration = 42s
-                    [
-                        env_prefix.to_string(),
-                        "storage".to_uppercase(),
-                        "manifest".to_uppercase(),
-                        "gc_duration".to_uppercase(),
-                    ]
-                    .join(ENV_VAR_SEP),
-                    Some("42s"),
-                ),
                 (
                     // wal.dir = /other/wal/dir
                     [
@@ -243,17 +265,12 @@ mod tests {
                 .unwrap();
 
                 // Check the configs from environment variables.
-                assert_eq!(opts.storage.manifest.checkpoint_margin, Some(99));
-                match opts.storage.store {
+                match &opts.storage.store {
                     ObjectStoreConfig::S3(s3_config) => {
                         assert_eq!(s3_config.bucket, "mybucket".to_string());
                     }
                     _ => panic!("unexpected store type"),
                 }
-                assert_eq!(
-                    opts.storage.manifest.gc_duration,
-                    Some(Duration::from_secs(42))
-                );
                 assert_eq!(
                     opts.meta_client.unwrap().metasrv_addrs,
                     vec![
@@ -264,7 +281,10 @@ mod tests {
                 );
 
                 // Should be the values from config file, not environment variables.
-                assert_eq!(opts.wal.dir.unwrap(), "/tmp/greptimedb/wal");
+                let WalConfig::RaftEngine(raft_engine_config) = opts.wal else {
+                    unreachable!()
+                };
+                assert_eq!(raft_engine_config.dir.unwrap(), "/tmp/greptimedb/wal");
 
                 // Should be default values.
                 assert_eq!(opts.node_id, None);

src/cmd/src/standalone.rs

@@ -15,38 +15,49 @@
 use std::sync::Arc;
 use std::{fs, path};
 
-use catalog::kvbackend::KvBackendCatalogManager;
-use catalog::CatalogManagerRef;
+use async_trait::async_trait;
 use clap::Parser;
-use common_base::Plugins;
-use common_config::{metadata_store_dir, KvStoreConfig, WalConfig};
-use common_meta::cache_invalidator::DummyKvCacheInvalidator;
+use common_catalog::consts::MIN_USER_TABLE_ID;
+use common_config::wal::StandaloneWalConfig;
+use common_config::{metadata_store_dir, KvBackendConfig};
+use common_meta::cache_invalidator::DummyCacheInvalidator;
+use common_meta::datanode_manager::DatanodeManagerRef;
+use common_meta::ddl::{DdlTaskExecutorRef, TableMetadataAllocatorRef};
+use common_meta::ddl_manager::DdlManager;
+use common_meta::key::{TableMetadataManager, TableMetadataManagerRef};
 use common_meta::kv_backend::KvBackendRef;
+use common_meta::region_keeper::MemoryRegionKeeper;
+use common_meta::sequence::SequenceBuilder;
+use common_meta::wal::{WalOptionsAllocator, WalOptionsAllocatorRef};
 use common_procedure::ProcedureManagerRef;
 use common_telemetry::info;
 use common_telemetry::logging::LoggingOptions;
+use common_time::timezone::set_default_timezone;
 use datanode::config::{DatanodeOptions, ProcedureConfig, RegionEngineConfig, StorageConfig};
 use datanode::datanode::{Datanode, DatanodeBuilder};
-use datanode::region_server::RegionServer;
 use file_engine::config::EngineConfig as FileEngineConfig;
 use frontend::frontend::FrontendOptions;
+use frontend::instance::builder::FrontendBuilder;
+use frontend::instance::standalone::StandaloneTableMetadataAllocator;
 use frontend::instance::{FrontendInstance, Instance as FeInstance, StandaloneDatanodeManager};
 use frontend::service_config::{
     GrpcOptions, InfluxdbOptions, MysqlOptions, OpentsdbOptions, PostgresOptions, PromStoreOptions,
 };
 use mito2::config::MitoConfig;
 use serde::{Deserialize, Serialize};
+use servers::export_metrics::ExportMetricsOption;
 use servers::http::HttpOptions;
 use servers::tls::{TlsMode, TlsOption};
 use servers::Mode;
 use snafu::ResultExt;
 
 use crate::error::{
-    CreateDirSnafu, IllegalConfigSnafu, InitMetadataSnafu, Result, ShutdownDatanodeSnafu,
-    ShutdownFrontendSnafu, StartDatanodeSnafu, StartFrontendSnafu, StartProcedureManagerSnafu,
-    StopProcedureManagerSnafu,
+    CreateDirSnafu, IllegalConfigSnafu, InitDdlManagerSnafu, InitMetadataSnafu, InitTimezoneSnafu,
+    Result, ShutdownDatanodeSnafu, ShutdownFrontendSnafu, StartDatanodeSnafu, StartFrontendSnafu,
+    StartProcedureManagerSnafu, StartWalOptionsAllocatorSnafu, StopProcedureManagerSnafu,
 };
-use crate::options::{MixOptions, Options, TopLevelOptions};
+use crate::options::{CliOptions, MixOptions, Options};
+use crate::App;
 
 #[derive(Parser)]
 pub struct Command {
@@ -59,8 +70,8 @@ impl Command {
         self.subcmd.build(opts).await
     }
 
-    pub fn load_options(&self, top_level_options: TopLevelOptions) -> Result<Options> {
-        self.subcmd.load_options(top_level_options)
+    pub fn load_options(&self, cli_options: &CliOptions) -> Result<Options> {
+        self.subcmd.load_options(cli_options)
     }
 }
 
@@ -76,9 +87,9 @@ impl SubCommand {
         }
     }
 
-    fn load_options(&self, top_level_options: TopLevelOptions) -> Result<Options> {
+    fn load_options(&self, cli_options: &CliOptions) -> Result<Options> {
         match self {
-            SubCommand::Start(cmd) => cmd.load_options(top_level_options),
+            SubCommand::Start(cmd) => cmd.load_options(cli_options),
         }
     }
 }
@@ -88,6 +99,7 @@ impl SubCommand {
 pub struct StandaloneOptions {
     pub mode: Mode,
     pub enable_telemetry: bool,
+    pub default_timezone: Option<String>,
     pub http: HttpOptions,
     pub grpc: GrpcOptions,
     pub mysql: MysqlOptions,
@@ -95,14 +107,15 @@ pub struct StandaloneOptions {
     pub opentsdb: OpentsdbOptions,
     pub influxdb: InfluxdbOptions,
     pub prom_store: PromStoreOptions,
-    pub wal: WalConfig,
+    pub wal: StandaloneWalConfig,
     pub storage: StorageConfig,
-    pub metadata_store: KvStoreConfig,
+    pub metadata_store: KvBackendConfig,
     pub procedure: ProcedureConfig,
     pub logging: LoggingOptions,
     pub user_provider: Option<String>,
     /// Options for different store engines.
     pub region_engine: Vec<RegionEngineConfig>,
+    pub export_metrics: ExportMetricsOption,
 }
 
 impl Default for StandaloneOptions {
@@ -110,6 +123,7 @@ impl Default for StandaloneOptions {
         Self {
             mode: Mode::Standalone,
             enable_telemetry: true,
+            default_timezone: None,
             http: HttpOptions::default(),
             grpc: GrpcOptions::default(),
             mysql: MysqlOptions::default(),
@@ -117,11 +131,12 @@ impl Default for StandaloneOptions {
             opentsdb: OpentsdbOptions::default(),
             influxdb: InfluxdbOptions::default(),
             prom_store: PromStoreOptions::default(),
-            wal: WalConfig::default(),
+            wal: StandaloneWalConfig::default(),
             storage: StorageConfig::default(),
-            metadata_store: KvStoreConfig::default(),
+            metadata_store: KvBackendConfig::default(),
             procedure: ProcedureConfig::default(),
             logging: LoggingOptions::default(),
+            export_metrics: ExportMetricsOption::default(),
             user_provider: None,
             region_engine: vec![
                 RegionEngineConfig::Mito(MitoConfig::default()),
@@ -135,6 +150,7 @@ impl StandaloneOptions {
     fn frontend_options(self) -> FrontendOptions {
         FrontendOptions {
             mode: self.mode,
+            default_timezone: self.default_timezone,
             http: self.http,
             grpc: self.grpc,
             mysql: self.mysql,
@@ -145,6 +161,8 @@ impl StandaloneOptions {
             meta_client: None,
             logging: self.logging,
             user_provider: self.user_provider,
+            // Handle the export metrics task run by standalone to frontend for execution
+            export_metrics: self.export_metrics,
             ..Default::default()
         }
     }
@@ -153,9 +171,10 @@ impl StandaloneOptions {
         DatanodeOptions {
             node_id: Some(0),
             enable_telemetry: self.enable_telemetry,
-            wal: self.wal,
+            wal: self.wal.into(),
             storage: self.storage,
             region_engine: self.region_engine,
+            rpc_addr: self.grpc.addr,
             ..Default::default()
         }
     }
@@ -165,24 +184,33 @@ pub struct Instance {
     datanode: Datanode,
     frontend: FeInstance,
     procedure_manager: ProcedureManagerRef,
+    wal_options_allocator: WalOptionsAllocatorRef,
 }
 
-impl Instance {
-    pub async fn start(&mut self) -> Result<()> {
-        // Start datanode instance before starting services, to avoid requests come in before internal components are started.
-        self.datanode.start().await.context(StartDatanodeSnafu)?;
-        info!("Datanode instance started");
+#[async_trait]
+impl App for Instance {
+    fn name(&self) -> &str {
+        "greptime-standalone"
+    }
+
+    async fn start(&mut self) -> Result<()> {
+        self.datanode.start_telemetry();
 
         self.procedure_manager
             .start()
             .await
             .context(StartProcedureManagerSnafu)?;
 
+        self.wal_options_allocator
+            .start()
+            .await
+            .context(StartWalOptionsAllocatorSnafu)?;
+
         self.frontend.start().await.context(StartFrontendSnafu)?;
         Ok(())
     }
 
-    pub async fn stop(&self) -> Result<()> {
+    async fn stop(&self) -> Result<()> {
         self.frontend
             .shutdown()
             .await
@@ -204,7 +232,7 @@ impl Instance {
 }
 
 #[derive(Debug, Default, Parser)]
-struct StartCommand {
+pub struct StartCommand {
     #[clap(long)]
     http_addr: Option<String>,
     #[clap(long)]
@@ -218,7 +246,7 @@ struct StartCommand {
     #[clap(short, long)]
     influxdb_enable: bool,
     #[clap(short, long)]
-    config_file: Option<String>,
+    pub config_file: Option<String>,
     #[clap(long)]
     tls_mode: Option<TlsMode>,
     #[clap(long)]
@@ -228,25 +256,36 @@ struct StartCommand {
     #[clap(long)]
     user_provider: Option<String>,
     #[clap(long, default_value = "GREPTIMEDB_STANDALONE")]
-    env_prefix: String,
+    pub env_prefix: String,
+    /// The working home directory of this standalone instance.
+    #[clap(long)]
+    data_home: Option<String>,
 }
 
 impl StartCommand {
-    fn load_options(&self, top_level_options: TopLevelOptions) -> Result<Options> {
-        let mut opts: StandaloneOptions = Options::load_layered_options(
+    fn load_options(&self, cli_options: &CliOptions) -> Result<Options> {
+        let opts: StandaloneOptions = Options::load_layered_options(
             self.config_file.as_deref(),
             self.env_prefix.as_ref(),
             None,
         )?;
 
+        self.convert_options(cli_options, opts)
+    }
+
+    pub fn convert_options(
+        &self,
+        cli_options: &CliOptions,
+        mut opts: StandaloneOptions,
+    ) -> Result<Options> {
         opts.mode = Mode::Standalone;
 
-        if let Some(dir) = top_level_options.log_dir {
-            opts.logging.dir = dir;
+        if let Some(dir) = &cli_options.log_dir {
+            opts.logging.dir = dir.clone();
         }
 
-        if top_level_options.log_level.is_some() {
-            opts.logging.level = top_level_options.log_level;
+        if cli_options.log_level.is_some() {
+            opts.logging.level = cli_options.log_level.clone();
         }
 
         let tls_opts = TlsOption::new(
@@ -259,6 +298,10 @@ impl StartCommand {
             opts.http.addr = addr.clone()
         }
 
+        if let Some(data_home) = &self.data_home {
+            opts.storage.data_home = data_home.clone();
+        }
+
         if let Some(addr) = &self.rpc_addr {
             // frontend grpc addr conflict with datanode default grpc addr
             let datanode_grpc_addr = DatanodeOptions::default().rpc_addr;
@@ -300,7 +343,8 @@ impl StartCommand {
         let procedure = opts.procedure.clone();
         let frontend = opts.clone().frontend_options();
         let logging = opts.logging.clone();
-        let datanode = opts.datanode_options();
+        let wal_meta = opts.wal.clone().into();
+        let datanode = opts.datanode_options().clone();
 
         Ok(Options::Standalone(Box::new(MixOptions {
             procedure,
@@ -309,6 +353,7 @@ impl StartCommand {
             frontend,
             datanode,
             logging,
+            wal_meta,
         })))
     }
 
@@ -316,18 +361,20 @@ impl StartCommand {
     #[allow(unused_variables)]
     #[allow(clippy::diverging_sub_expression)]
     async fn build(self, opts: MixOptions) -> Result<Instance> {
-        let mut fe_opts = opts.frontend;
-        let fe_plugins = plugins::setup_frontend_plugins(&mut fe_opts)
+        let mut fe_opts = opts.frontend.clone();
+        #[allow(clippy::unnecessary_mut_passed)]
+        let fe_plugins = plugins::setup_frontend_plugins(&mut fe_opts) // mut ref is MUST, DO NOT change it
             .await
             .context(StartFrontendSnafu)?;
 
-        let dn_opts = opts.datanode;
+        let dn_opts = opts.datanode.clone();
 
         info!("Standalone start command: {:#?}", self);
-        info!(
-            "Standalone frontend options: {:#?}, datanode options: {:#?}",
-            fe_opts, dn_opts
-        );
+
+        info!("Building standalone instance with {opts:#?}");
+
+        set_default_timezone(opts.frontend.default_timezone.as_deref())
+            .context(InitTimezoneSnafu)?;
 
         // Ensure the data_home directory exists.
         fs::create_dir_all(path::Path::new(&opts.data_home)).context(CreateDirSnafu {
@@ -335,45 +382,55 @@ impl StartCommand {
         })?;
 
         let metadata_dir = metadata_store_dir(&opts.data_home);
-        let (kv_store, procedure_manager) = FeInstance::try_build_standalone_components(
+        let (kv_backend, procedure_manager) = FeInstance::try_build_standalone_components(
             metadata_dir,
-            opts.metadata_store,
-            opts.procedure,
+            opts.metadata_store.clone(),
+            opts.procedure.clone(),
         )
         .await
         .context(StartFrontendSnafu)?;
 
-        let datanode =
-            DatanodeBuilder::new(dn_opts.clone(), Some(kv_store.clone()), Default::default())
-                .build()
-                .await
-                .context(StartDatanodeSnafu)?;
-        let region_server = datanode.region_server();
+        let builder =
+            DatanodeBuilder::new(dn_opts, fe_plugins.clone()).with_kv_backend(kv_backend.clone());
+        let datanode = builder.build().await.context(StartDatanodeSnafu)?;
 
-        let catalog_manager = KvBackendCatalogManager::new(
-            kv_store.clone(),
-            Arc::new(DummyKvCacheInvalidator),
-            Arc::new(StandaloneDatanodeManager(region_server.clone())),
+        let datanode_manager = Arc::new(StandaloneDatanodeManager(datanode.region_server()));
+
+        let table_id_sequence = Arc::new(
+            SequenceBuilder::new("table_id", kv_backend.clone())
+                .initial(MIN_USER_TABLE_ID as u64)
+                .step(10)
+                .build(),
         );
+        let wal_options_allocator = Arc::new(WalOptionsAllocator::new(
+            opts.wal_meta.clone(),
+            kv_backend.clone(),
+        ));
+        let table_meta_allocator = Arc::new(StandaloneTableMetadataAllocator::new(
+            table_id_sequence,
+            wal_options_allocator.clone(),
+        ));
 
-        catalog_manager
-            .table_metadata_manager_ref()
-            .init()
-            .await
-            .context(InitMetadataSnafu)?;
-
-        // TODO: build frontend instance like in distributed mode
-        let mut frontend = build_frontend(
-            fe_plugins,
-            kv_store,
+        let ddl_task_executor = Self::create_ddl_task_executor(
+            kv_backend.clone(),
             procedure_manager.clone(),
-            catalog_manager,
-            region_server,
+            datanode_manager.clone(),
+            table_meta_allocator,
         )
         .await?;
 
+        let mut frontend = FrontendBuilder::new(kv_backend, datanode_manager, ddl_task_executor)
+            .with_plugin(fe_plugins)
+            .try_build()
+            .await
+            .context(StartFrontendSnafu)?;
+
         frontend
-            .build_servers(&fe_opts)
+            .build_export_metrics_task(&opts.frontend.export_metrics)
+            .context(StartFrontendSnafu)?;
+
+        frontend
+            .build_servers(opts)
             .await
             .context(StartFrontendSnafu)?;
 
@@ -381,28 +438,46 @@ impl StartCommand {
             datanode,
             frontend,
             procedure_manager,
+            wal_options_allocator,
         })
     }
-}
 
-/// Build frontend instance in standalone mode
-async fn build_frontend(
-    plugins: Plugins,
-    kv_store: KvBackendRef,
-    procedure_manager: ProcedureManagerRef,
-    catalog_manager: CatalogManagerRef,
-    region_server: RegionServer,
-) -> Result<FeInstance> {
-    let frontend_instance = FeInstance::try_new_standalone(
-        kv_store,
-        procedure_manager,
-        catalog_manager,
-        plugins,
-        region_server,
-    )
-    .await
-    .context(StartFrontendSnafu)?;
-    Ok(frontend_instance)
+    pub async fn create_ddl_task_executor(
+        kv_backend: KvBackendRef,
+        procedure_manager: ProcedureManagerRef,
+        datanode_manager: DatanodeManagerRef,
+        table_meta_allocator: TableMetadataAllocatorRef,
+    ) -> Result<DdlTaskExecutorRef> {
+        let table_metadata_manager =
+            Self::create_table_metadata_manager(kv_backend.clone()).await?;
+
+        let ddl_task_executor: DdlTaskExecutorRef = Arc::new(
+            DdlManager::try_new(
+                procedure_manager,
+                datanode_manager,
+                Arc::new(DummyCacheInvalidator),
+                table_metadata_manager,
+                table_meta_allocator,
+                Arc::new(MemoryRegionKeeper::default()),
+            )
+            .context(InitDdlManagerSnafu)?,
+        );
+
+        Ok(ddl_task_executor)
+    }
+
+    async fn create_table_metadata_manager(
+        kv_backend: KvBackendRef,
+    ) -> Result<TableMetadataManagerRef> {
+        let table_metadata_manager = Arc::new(TableMetadataManager::new(kv_backend));
+
+        table_metadata_manager
+            .init()
+            .await
+            .context(InitMetadataSnafu)?;
+
+        Ok(table_metadata_manager)
+    }
 }
 
 #[cfg(test)]
@@ -413,11 +488,13 @@ mod tests {
 
     use auth::{Identity, Password, UserProviderRef};
     use common_base::readable_size::ReadableSize;
+    use common_config::WalConfig;
     use common_test_util::temp_dir::create_named_temp_file;
+    use datanode::config::{FileConfig, GcsConfig};
     use servers::Mode;
 
     use super::*;
-    use crate::options::ENV_VAR_SEP;
+    use crate::options::{CliOptions, ENV_VAR_SEP};
 
     #[tokio::test]
     async fn test_try_from_start_command_to_anymap() {
@@ -426,6 +503,7 @@ mod tests {
             ..Default::default()
         };
 
+        #[allow(clippy::unnecessary_mut_passed)]
         let plugins = plugins::setup_frontend_plugins(&mut fe_opts).await.unwrap();
 
         let provider = plugins.get::<UserProviderRef>().unwrap();
@@ -454,6 +532,7 @@ mod tests {
             enable_memory_catalog = true
 
             [wal]
+            provider = "raft_engine"
             dir = "/tmp/greptimedb/test/wal"
             file_size = "1GB"
             purge_threshold = "50GB"
@@ -462,6 +541,15 @@ mod tests {
             sync_write = false
 
             [storage]
+            data_home = "/tmp/greptimedb/"
+            type = "File"
+
+            [[storage.providers]]
+            type = "Gcs"
+            bucket = "foo"
+            endpoint = "bar"
+
+            [[storage.providers]]
             type = "S3"
             access_key_id = "access_key_id"
             secret_access_key = "secret_access_key"
@@ -491,8 +579,7 @@ mod tests {
             ..Default::default()
         };
 
-        let Options::Standalone(options) = cmd.load_options(TopLevelOptions::default()).unwrap()
-        else {
+        let Options::Standalone(options) = cmd.load_options(&CliOptions::default()).unwrap() else {
             unreachable!()
         };
         let fe_opts = options.frontend;
@@ -509,9 +596,21 @@ mod tests {
         assert_eq!(None, fe_opts.mysql.reject_no_database);
         assert!(fe_opts.influxdb.enable);
 
-        assert_eq!("/tmp/greptimedb/test/wal", dn_opts.wal.dir.unwrap());
+        let WalConfig::RaftEngine(raft_engine_config) = dn_opts.wal else {
+            unreachable!()
+        };
+        assert_eq!("/tmp/greptimedb/test/wal", raft_engine_config.dir.unwrap());
 
-        match &dn_opts.storage.store {
+        assert!(matches!(
+            &dn_opts.storage.store,
+            datanode::config::ObjectStoreConfig::File(FileConfig { .. })
+        ));
+        assert_eq!(dn_opts.storage.providers.len(), 2);
+        assert!(matches!(
+            dn_opts.storage.providers[0],
+            datanode::config::ObjectStoreConfig::Gcs(GcsConfig { .. })
+        ));
+        match &dn_opts.storage.providers[1] {
             datanode::config::ObjectStoreConfig::S3(s3_config) => {
                 assert_eq!(
                     "Secret([REDACTED alloc::string::String])".to_string(),
@@ -528,16 +627,19 @@ mod tests {
     }
 
     #[test]
-    fn test_top_level_options() {
+    fn test_load_log_options_from_cli() {
         let cmd = StartCommand {
             user_provider: Some("static_user_provider:cmd:test=test".to_string()),
             ..Default::default()
         };
 
         let Options::Standalone(opts) = cmd
-            .load_options(TopLevelOptions {
+            .load_options(&CliOptions {
                 log_dir: Some("/tmp/greptimedb/test/logs".to_string()),
                 log_level: Some("debug".to_string()),
+
+                #[cfg(feature = "tokio-console")]
+                tokio_console_addr: None,
             })
             .unwrap()
         else {
@@ -604,11 +706,8 @@ mod tests {
                     ..Default::default()
                 };
 
-                let top_level_opts = TopLevelOptions {
-                    log_dir: None,
-                    log_level: None,
-                };
-                let Options::Standalone(opts) = command.load_options(top_level_opts).unwrap()
+                let Options::Standalone(opts) =
+                    command.load_options(&CliOptions::default()).unwrap()
                 else {
                     unreachable!()
                 };

src/common/base/Cargo.toml

@@ -7,9 +7,9 @@ license.workspace = true
 [dependencies]
 anymap = "1.0.0-beta.2"
 bitvec = "1.0"
-bytes = { version = "1.1", features = ["serde"] }
-common-error = { workspace = true }
-common-macro = { workspace = true }
+bytes.workspace = true
+common-error.workspace = true
+common-macro.workspace = true
 paste = "1.0"
 serde = { version = "1.0", features = ["derive"] }
 snafu.workspace = true

src/common/catalog/Cargo.toml

@@ -5,11 +5,9 @@ edition.workspace = true
 license.workspace = true
 
 [dependencies]
-common-error = { workspace = true }
-common-macro = { workspace = true }
-serde.workspace = true
-serde_json = "1.0"
-snafu = { version = "0.7", features = ["backtraces"] }
+common-error.workspace = true
+common-macro.workspace = true
+snafu.workspace = true
 
 [dev-dependencies]
 chrono.workspace = true

src/common/catalog/src/consts.rs

@@ -17,6 +17,7 @@ pub const INFORMATION_SCHEMA_NAME: &str = "information_schema";
 pub const SYSTEM_CATALOG_TABLE_NAME: &str = "system_catalog";
 pub const DEFAULT_CATALOG_NAME: &str = "greptime";
 pub const DEFAULT_SCHEMA_NAME: &str = "public";
+pub const DEFAULT_PRIVATE_SCHEMA_NAME: &str = "greptime_private";
 
 /// Reserves [0,MIN_USER_TABLE_ID) for internal usage.
 /// User defined table id starts from this value.
@@ -29,13 +30,25 @@ pub const SYSTEM_CATALOG_TABLE_ID: u32 = 0;
 pub const SCRIPTS_TABLE_ID: u32 = 1;
 /// numbers table id
 pub const NUMBERS_TABLE_ID: u32 = 2;
+
+/// ----- Begin of information_schema tables -----
 /// id for information_schema.tables
 pub const INFORMATION_SCHEMA_TABLES_TABLE_ID: u32 = 3;
 /// id for information_schema.columns
 pub const INFORMATION_SCHEMA_COLUMNS_TABLE_ID: u32 = 4;
+/// id for information_schema.engines
+pub const INFORMATION_SCHEMA_ENGINES_TABLE_ID: u32 = 5;
+/// id for information_schema.column_privileges
+pub const INFORMATION_SCHEMA_COLUMN_PRIVILEGES_TABLE_ID: u32 = 6;
+/// id for information_schema.column_statistics
+pub const INFORMATION_SCHEMA_COLUMN_STATISTICS_TABLE_ID: u32 = 7;
+/// id for information_schema.build_info
+pub const INFORMATION_SCHEMA_BUILD_INFO_TABLE_ID: u32 = 8;
+/// ----- End of information_schema tables -----
 
 pub const MITO_ENGINE: &str = "mito";
 pub const MITO2_ENGINE: &str = "mito2";
+pub const METRIC_ENGINE: &str = "metric";
 
 pub fn default_engine() -> &'static str {
     MITO_ENGINE

src/common/config/Cargo.toml

@@ -7,4 +7,8 @@ license.workspace = true
 [dependencies]
 common-base.workspace = true
 humantime-serde.workspace = true
+rskafka.workspace = true
 serde.workspace = true
+serde_json.workspace = true
+serde_with = "3"
+toml.workspace = true

src/common/config/src/lib.rs

@@ -12,41 +12,12 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-use std::time::Duration;
+pub mod wal;
 
 use common_base::readable_size::ReadableSize;
 use serde::{Deserialize, Serialize};
 
-#[derive(Debug, Clone, Serialize, Deserialize, PartialEq, Eq)]
-#[serde(default)]
-pub struct WalConfig {
-    // wal directory
-    pub dir: Option<String>,
-    // wal file size in bytes
-    pub file_size: ReadableSize,
-    // wal purge threshold in bytes
-    pub purge_threshold: ReadableSize,
-    // purge interval in seconds
-    #[serde(with = "humantime_serde")]
-    pub purge_interval: Duration,
-    // read batch size
-    pub read_batch_size: usize,
-    // whether to sync log file after every write
-    pub sync_write: bool,
-}
-
-impl Default for WalConfig {
-    fn default() -> Self {
-        Self {
-            dir: None,
-            file_size: ReadableSize::mb(256), // log file size 256MB
-            purge_threshold: ReadableSize::gb(4), // purge threshold 4GB
-            purge_interval: Duration::from_secs(600),
-            read_batch_size: 128,
-            sync_write: false,
-        }
-    }
-}
+pub use crate::wal::{KafkaWalOptions, WalConfig, WalOptions, WAL_OPTIONS_KEY};
 
 pub fn metadata_store_dir(store_dir: &str) -> String {
     format!("{store_dir}/metadata")
@@ -54,14 +25,14 @@ pub fn metadata_store_dir(store_dir: &str) -> String {
 
 #[derive(Debug, Clone, Serialize, Deserialize, PartialEq, Eq)]
 #[serde(default)]
-pub struct KvStoreConfig {
+pub struct KvBackendConfig {
     // Kv file size in bytes
     pub file_size: ReadableSize,
     // Kv purge threshold in bytes
     pub purge_threshold: ReadableSize,
 }
 
-impl Default for KvStoreConfig {
+impl Default for KvBackendConfig {
     fn default() -> Self {
         Self {
             // log file size 256MB

src/common/config/src/wal.rs

@@ -0,0 +1,142 @@
+// 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.
+
+pub mod kafka;
+pub mod raft_engine;
+
+use serde::{Deserialize, Serialize};
+use serde_with::with_prefix;
+
+pub use crate::wal::kafka::{
+    KafkaConfig, KafkaOptions as KafkaWalOptions, StandaloneKafkaConfig, Topic as KafkaWalTopic,
+};
+pub use crate::wal::raft_engine::RaftEngineConfig;
+
+/// An encoded wal options will be wrapped into a (WAL_OPTIONS_KEY, encoded wal options) key-value pair
+/// and inserted into the options of a `RegionCreateRequest`.
+pub const WAL_OPTIONS_KEY: &str = "wal_options";
+
+/// Wal config for datanode.
+#[derive(Debug, Clone, Serialize, Deserialize, PartialEq, Eq)]
+#[serde(tag = "provider", rename_all = "snake_case")]
+pub enum WalConfig {
+    RaftEngine(RaftEngineConfig),
+    Kafka(KafkaConfig),
+}
+
+impl From<StandaloneWalConfig> for WalConfig {
+    fn from(value: StandaloneWalConfig) -> Self {
+        match value {
+            StandaloneWalConfig::RaftEngine(config) => WalConfig::RaftEngine(config),
+            StandaloneWalConfig::Kafka(config) => WalConfig::Kafka(config.base),
+        }
+    }
+}
+
+impl Default for WalConfig {
+    fn default() -> Self {
+        WalConfig::RaftEngine(RaftEngineConfig::default())
+    }
+}
+
+/// Wal config for datanode.
+#[derive(Debug, Clone, Serialize, Deserialize, PartialEq, Eq)]
+#[serde(tag = "provider", rename_all = "snake_case")]
+pub enum StandaloneWalConfig {
+    RaftEngine(RaftEngineConfig),
+    Kafka(StandaloneKafkaConfig),
+}
+
+impl Default for StandaloneWalConfig {
+    fn default() -> Self {
+        StandaloneWalConfig::RaftEngine(RaftEngineConfig::default())
+    }
+}
+
+/// Wal options allocated to a region.
+/// A wal options is encoded by metasrv with `serde_json::to_string`, and then decoded
+/// by datanode with `serde_json::from_str`.
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize, Default)]
+#[serde(tag = "wal.provider", rename_all = "snake_case")]
+pub enum WalOptions {
+    #[default]
+    RaftEngine,
+    #[serde(with = "prefix_wal_kafka")]
+    Kafka(KafkaWalOptions),
+}
+
+with_prefix!(prefix_wal_kafka "wal.kafka.");
+
+#[cfg(test)]
+mod tests {
+    use std::time::Duration;
+
+    use common_base::readable_size::ReadableSize;
+    use rskafka::client::partition::Compression as RsKafkaCompression;
+
+    use crate::wal::kafka::KafkaBackoffConfig;
+    use crate::wal::{KafkaConfig, KafkaWalOptions, WalOptions};
+
+    #[test]
+    fn test_serde_kafka_config() {
+        let toml_str = r#"
+            broker_endpoints = ["127.0.0.1:9092"]
+            max_batch_size = "4MB"
+            linger = "200ms"
+            produce_record_timeout = "100ms"
+            backoff_init = "500ms"
+            backoff_max = "10s"
+            backoff_base = 2
+            backoff_deadline = "5mins"
+        "#;
+        let decoded: KafkaConfig = toml::from_str(toml_str).unwrap();
+        let expected = KafkaConfig {
+            broker_endpoints: vec!["127.0.0.1:9092".to_string()],
+            compression: RsKafkaCompression::default(),
+            max_batch_size: ReadableSize::mb(4),
+            linger: Duration::from_millis(200),
+            produce_record_timeout: Duration::from_millis(100),
+            backoff: KafkaBackoffConfig {
+                init: Duration::from_millis(500),
+                max: Duration::from_secs(10),
+                base: 2,
+                deadline: Some(Duration::from_secs(60 * 5)),
+            },
+        };
+        assert_eq!(decoded, expected);
+    }
+
+    #[test]
+    fn test_serde_wal_options() {
+        // Test serde raft-engine wal options.
+        let wal_options = WalOptions::RaftEngine;
+        let encoded = serde_json::to_string(&wal_options).unwrap();
+        let expected = r#"{"wal.provider":"raft_engine"}"#;
+        assert_eq!(&encoded, expected);
+
+        let decoded: WalOptions = serde_json::from_str(&encoded).unwrap();
+        assert_eq!(decoded, wal_options);
+
+        // Test serde kafka wal options.
+        let wal_options = WalOptions::Kafka(KafkaWalOptions {
+            topic: "test_topic".to_string(),
+        });
+        let encoded = serde_json::to_string(&wal_options).unwrap();
+        let expected = r#"{"wal.provider":"kafka","wal.kafka.topic":"test_topic"}"#;
+        assert_eq!(&encoded, expected);
+
+        let decoded: WalOptions = serde_json::from_str(&encoded).unwrap();
+        assert_eq!(decoded, wal_options);
+    }
+}

src/common/config/src/wal/kafka.rs

@@ -0,0 +1,144 @@
+// 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::time::Duration;
+
+use common_base::readable_size::ReadableSize;
+use rskafka::client::partition::Compression as RsKafkaCompression;
+use serde::{Deserialize, Serialize};
+use serde_with::with_prefix;
+
+/// Topic name prefix.
+pub const TOPIC_NAME_PREFIX: &str = "greptimedb_wal_topic";
+/// Kafka wal topic.
+pub type Topic = String;
+
+/// The type of the topic selector, i.e. with which strategy to select a topic.
+#[derive(Default, Debug, Clone, Serialize, Deserialize, PartialEq, Eq)]
+#[serde(rename_all = "snake_case")]
+pub enum TopicSelectorType {
+    #[default]
+    RoundRobin,
+}
+
+/// Configurations for kafka wal.
+#[derive(Debug, Clone, Serialize, Deserialize, PartialEq, Eq)]
+#[serde(default)]
+pub struct KafkaConfig {
+    /// The broker endpoints of the Kafka cluster.
+    pub broker_endpoints: Vec<String>,
+    /// The compression algorithm used to compress log entries.
+    #[serde(skip)]
+    #[serde(default)]
+    pub compression: RsKafkaCompression,
+    /// The maximum log size a kakfa batch producer could buffer.
+    pub max_batch_size: ReadableSize,
+    /// The linger duration of a kafka batch producer.
+    #[serde(with = "humantime_serde")]
+    pub linger: Duration,
+    /// The maximum amount of time (in milliseconds) to wait for Kafka records to be returned.
+    #[serde(with = "humantime_serde")]
+    pub produce_record_timeout: Duration,
+    /// The backoff config.
+    #[serde(flatten, with = "kafka_backoff")]
+    pub backoff: KafkaBackoffConfig,
+}
+
+impl Default for KafkaConfig {
+    fn default() -> Self {
+        Self {
+            broker_endpoints: vec!["127.0.0.1:9092".to_string()],
+            compression: RsKafkaCompression::NoCompression,
+            max_batch_size: ReadableSize::mb(4),
+            linger: Duration::from_millis(200),
+            produce_record_timeout: Duration::from_millis(100),
+            backoff: KafkaBackoffConfig::default(),
+        }
+    }
+}
+
+with_prefix!(pub kafka_backoff "backoff_");
+
+#[derive(Debug, Clone, Serialize, Deserialize, PartialEq, Eq)]
+#[serde(default)]
+pub struct KafkaBackoffConfig {
+    /// The initial backoff for kafka clients.
+    #[serde(with = "humantime_serde")]
+    pub init: Duration,
+    /// The maximum backoff for kafka clients.
+    #[serde(with = "humantime_serde")]
+    pub max: Duration,
+    /// Exponential backoff rate, i.e. next backoff = base * current backoff.
+    // Sets to u32 type since some structs containing the KafkaConfig need to derive the Eq trait.
+    pub base: u32,
+    /// Stop reconnecting if the total wait time reaches the deadline.
+    /// If it's None, the reconnecting won't terminate.
+    #[serde(with = "humantime_serde")]
+    pub deadline: Option<Duration>,
+}
+
+impl Default for KafkaBackoffConfig {
+    fn default() -> Self {
+        Self {
+            init: Duration::from_millis(500),
+            max: Duration::from_secs(10),
+            base: 2,
+            deadline: Some(Duration::from_secs(60 * 5)), // 5 mins
+        }
+    }
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize, PartialEq, Eq)]
+#[serde(default)]
+pub struct StandaloneKafkaConfig {
+    #[serde(flatten)]
+    pub base: KafkaConfig,
+    /// Number of topics to be created upon start.
+    pub num_topics: usize,
+    /// The type of the topic selector with which to select a topic for a region.
+    pub selector_type: TopicSelectorType,
+    /// Topic name prefix.
+    pub topic_name_prefix: String,
+    /// Number of partitions per topic.
+    pub num_partitions: i32,
+    /// The replication factor of each topic.
+    pub replication_factor: i16,
+    /// Above which a topic creation operation will be cancelled.
+    #[serde(with = "humantime_serde")]
+    pub create_topic_timeout: Duration,
+}
+
+impl Default for StandaloneKafkaConfig {
+    fn default() -> Self {
+        let base = KafkaConfig::default();
+        let replication_factor = base.broker_endpoints.len() as i16;
+
+        Self {
+            base,
+            num_topics: 64,
+            selector_type: TopicSelectorType::RoundRobin,
+            topic_name_prefix: "greptimedb_wal_topic".to_string(),
+            num_partitions: 1,
+            replication_factor,
+            create_topic_timeout: Duration::from_secs(30),
+        }
+    }
+}
+
+/// Kafka wal options allocated to a region.
+#[derive(Debug, Clone, Serialize, Deserialize, PartialEq, Eq)]
+pub struct KafkaOptions {
+    /// Kafka wal topic.
+    pub topic: Topic,
+}

src/common/config/src/wal/raft_engine.rs

@@ -0,0 +1,50 @@
+// 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::time::Duration;
+
+use common_base::readable_size::ReadableSize;
+use serde::{Deserialize, Serialize};
+
+/// Configurations for raft-engine wal.
+#[derive(Debug, Clone, Serialize, Deserialize, PartialEq, Eq)]
+#[serde(default)]
+pub struct RaftEngineConfig {
+    // wal directory
+    pub dir: Option<String>,
+    // wal file size in bytes
+    pub file_size: ReadableSize,
+    // wal purge threshold in bytes
+    pub purge_threshold: ReadableSize,
+    // purge interval in seconds
+    #[serde(with = "humantime_serde")]
+    pub purge_interval: Duration,
+    // read batch size
+    pub read_batch_size: usize,
+    // whether to sync log file after every write
+    pub sync_write: bool,
+}
+
+impl Default for RaftEngineConfig {
+    fn default() -> Self {
+        Self {
+            dir: None,
+            file_size: ReadableSize::mb(256),
+            purge_threshold: ReadableSize::gb(4),
+            purge_interval: Duration::from_secs(600),
+            read_batch_size: 128,
+            sync_write: false,
+        }
+    }
+}

src/common/datasource/Cargo.toml

@@ -16,16 +16,18 @@ async-compression = { version = "0.3", features = [
     "tokio",
 ] }
 async-trait.workspace = true
-bytes = "1.1"
-common-error = { workspace = true }
-common-macro = { workspace = true }
-common-runtime = { workspace = true }
+bytes.workspace = true
+common-error.workspace = true
+common-macro.workspace = true
+common-runtime.workspace = true
 datafusion.workspace = true
+datatypes.workspace = true
 derive_builder.workspace = true
 futures.workspace = true
 lazy_static.workspace = true
-object-store = { workspace = true }
+object-store.workspace = true
 orc-rust = "0.2"
+parquet.workspace = true
 paste = "1.0"
 regex = "1.7"
 serde.workspace = true
@@ -36,4 +38,4 @@ tokio.workspace = true
 url = "2.3"
 
 [dev-dependencies]
-common-test-util = { workspace = true }
+common-test-util.workspace = true

src/common/datasource/src/compression.rs

@@ -26,7 +26,9 @@ use tokio::io::{AsyncRead, AsyncWriteExt, BufReader};
 use tokio_util::io::{ReaderStream, StreamReader};
 
 use crate::error::{self, Error, Result};
+
 #[derive(Debug, Clone, Copy, PartialEq, Eq, Hash, EnumIter, Serialize, Deserialize)]
+#[serde(rename_all = "lowercase")]
 pub enum CompressionType {
     /// Gzip-ed file
     Gzip,

src/common/datasource/src/error.rs

@@ -166,6 +166,14 @@ pub enum Error {
 
     #[snafu(display("Buffered writer closed"))]
     BufferedWriterClosed { location: Location },
+
+    #[snafu(display("Failed to write parquet file, path: {}", path))]
+    WriteParquet {
+        path: String,
+        location: Location,
+        #[snafu(source)]
+        error: parquet::errors::ParquetError,
+    },
 }
 
 pub type Result<T> = std::result::Result<T, Error>;
@@ -178,7 +186,8 @@ impl ErrorExt for Error {
             | ListObjects { .. }
             | ReadObject { .. }
             | WriteObject { .. }
-            | AsyncWrite { .. } => StatusCode::StorageUnavailable,
+            | AsyncWrite { .. }
+            | WriteParquet { .. } => StatusCode::StorageUnavailable,
 
             UnsupportedBackendProtocol { .. }
             | UnsupportedCompressionType { .. }
@@ -231,6 +240,7 @@ impl ErrorExt for Error {
             InvalidConnection { location, .. } => Some(*location),
             UnsupportedCompressionType { location, .. } => Some(*location),
             UnsupportedFormat { location, .. } => Some(*location),
+            WriteParquet { location, .. } => Some(*location),
         }
     }
 }

src/common/datasource/src/file_format/parquet.rs

@@ -12,11 +12,13 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+use std::future::Future;
+use std::pin::Pin;
 use std::result;
 use std::sync::Arc;
 
 use arrow::record_batch::RecordBatch;
-use arrow_schema::Schema;
+use arrow_schema::{Schema, SchemaRef};
 use async_trait::async_trait;
 use datafusion::datasource::physical_plan::{FileMeta, ParquetFileReaderFactory};
 use datafusion::error::Result as DatafusionResult;
@@ -26,11 +28,15 @@ use datafusion::parquet::errors::{ParquetError, Result as ParquetResult};
 use datafusion::parquet::file::metadata::ParquetMetaData;
 use datafusion::parquet::format::FileMetaData;
 use datafusion::physical_plan::metrics::ExecutionPlanMetricsSet;
+use datafusion::physical_plan::SendableRecordBatchStream;
 use futures::future::BoxFuture;
+use futures::StreamExt;
 use object_store::{ObjectStore, Reader};
+use parquet::basic::{Compression, ZstdLevel};
+use parquet::file::properties::WriterProperties;
 use snafu::ResultExt;
 
-use crate::buffered_writer::{ArrowWriterCloser, DfRecordBatchEncoder};
+use crate::buffered_writer::{ArrowWriterCloser, DfRecordBatchEncoder, LazyBufferedWriter};
 use crate::error::{self, Result};
 use crate::file_format::FileFormat;
 use crate::share_buffer::SharedBuffer;
@@ -156,6 +162,103 @@ impl ArrowWriterCloser for ArrowWriter<SharedBuffer> {
     }
 }
 
+/// Parquet writer that buffers row groups in memory and writes buffered data to an underlying
+/// storage by chunks to reduce memory consumption.
+pub struct BufferedWriter {
+    inner: InnerBufferedWriter,
+}
+
+type InnerBufferedWriter = LazyBufferedWriter<
+    object_store::Writer,
+    ArrowWriter<SharedBuffer>,
+    Box<
+        dyn FnMut(
+                String,
+            )
+                -> Pin<Box<dyn Future<Output = error::Result<object_store::Writer>> + Send>>
+            + Send,
+    >,
+>;
+
+impl BufferedWriter {
+    pub async fn try_new(
+        path: String,
+        store: ObjectStore,
+        arrow_schema: SchemaRef,
+        props: Option<WriterProperties>,
+        buffer_threshold: usize,
+    ) -> error::Result<Self> {
+        let buffer = SharedBuffer::with_capacity(buffer_threshold);
+
+        let arrow_writer = ArrowWriter::try_new(buffer.clone(), arrow_schema.clone(), props)
+            .context(error::WriteParquetSnafu { path: &path })?;
+
+        Ok(Self {
+            inner: LazyBufferedWriter::new(
+                buffer_threshold,
+                buffer,
+                arrow_writer,
+                &path,
+                Box::new(move |path| {
+                    let store = store.clone();
+                    Box::pin(async move {
+                        store
+                            .writer(&path)
+                            .await
+                            .context(error::WriteObjectSnafu { path })
+                    })
+                }),
+            ),
+        })
+    }
+
+    /// Write a record batch to stream writer.
+    pub async fn write(&mut self, arrow_batch: &RecordBatch) -> error::Result<()> {
+        self.inner.write(arrow_batch).await?;
+        self.inner.try_flush(false).await?;
+
+        Ok(())
+    }
+
+    /// Close parquet writer.
+    ///
+    /// Return file metadata and bytes written.
+    pub async fn close(self) -> error::Result<(FileMetaData, u64)> {
+        self.inner.close_with_arrow_writer().await
+    }
+}
+
+/// Output the stream to a parquet file.
+///
+/// Returns number of rows written.
+pub async fn stream_to_parquet(
+    mut stream: SendableRecordBatchStream,
+    store: ObjectStore,
+    path: &str,
+    threshold: usize,
+) -> Result<usize> {
+    let write_props = WriterProperties::builder()
+        .set_compression(Compression::ZSTD(ZstdLevel::default()))
+        .build();
+    let schema = stream.schema();
+    let mut buffered_writer = BufferedWriter::try_new(
+        path.to_string(),
+        store,
+        schema,
+        Some(write_props),
+        threshold,
+    )
+    .await?;
+    let mut rows_written = 0;
+    while let Some(batch) = stream.next().await {
+        let batch = batch.context(error::ReadRecordBatchSnafu)?;
+        buffered_writer.write(&batch).await?;
+        rows_written += batch.num_rows();
+    }
+    buffered_writer.close().await?;
+    Ok(rows_written)
+}
+
 #[cfg(test)]
 mod tests {
     use common_test_util::find_workspace_path;