fix(sqlness): redact all volatile text (#4583)

Commit Message:

 Add SQLNESS replacements for RoundRobinBatch and region patterns

.coderabbit.yaml

@@ -0,0 +1,15 @@
+# yaml-language-server: $schema=https://coderabbit.ai/integrations/schema.v2.json
+language: "en-US"
+early_access: false
+reviews:
+  profile: "chill"
+  request_changes_workflow: false
+  high_level_summary: true
+  poem: true
+  review_status: true
+  collapse_walkthrough: false
+  auto_review:
+    enabled: false
+    drafts: false
+chat:
+  auto_reply: true

.env.example

@@ -14,10 +14,11 @@ GT_AZBLOB_CONTAINER=AZBLOB container
 GT_AZBLOB_ACCOUNT_NAME=AZBLOB account name
 GT_AZBLOB_ACCOUNT_KEY=AZBLOB account key
 GT_AZBLOB_ENDPOINT=AZBLOB endpoint
-# Settings for gcs test 
-GT_GCS_BUCKET = GCS bucket 
+# Settings for gcs test
+GT_GCS_BUCKET = GCS bucket
 GT_GCS_SCOPE  = GCS scope
-GT_GCS_CREDENTIAL_PATH = GCS credential path 
+GT_GCS_CREDENTIAL_PATH = GCS credential path
+GT_GCS_CREDENTIAL = GCS credential
 GT_GCS_ENDPOINT = GCS end point
 # Settings for kafka wal test
 GT_KAFKA_ENDPOINTS = localhost:9092
@@ -28,3 +29,8 @@ GT_MYSQL_ADDR = localhost:4002
 # Setting for unstable fuzz tests
 GT_FUZZ_BINARY_PATH=/path/to/
 GT_FUZZ_INSTANCE_ROOT_DIR=/tmp/unstable_greptime
+GT_FUZZ_INPUT_MAX_ROWS=2048
+GT_FUZZ_INPUT_MAX_TABLES=32
+GT_FUZZ_INPUT_MAX_COLUMNS=32
+GT_FUZZ_INPUT_MAX_ALTER_ACTIONS=256
+GT_FUZZ_INPUT_MAX_INSERT_ACTIONS=8

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

@@ -62,12 +62,13 @@ runs:
     # Get proper backtraces in mac Sonoma. Currently there's an issue with the new
     # linker that prevents backtraces from getting printed correctly.
     #
-    # <https://github.com/rust-lang/rust/issues/113783> 
+    # <https://github.com/rust-lang/rust/issues/113783>
     - name: Run integration tests
       if: ${{ inputs.disable-run-tests == 'false' }}
       shell: bash
-      env: 
+      env:
         CARGO_BUILD_RUSTFLAGS: "-Clink-arg=-Wl,-ld_classic"
+        SQLNESS_OPTS: "--preserve-state"
       run: |
         make test sqlness-test
 
@@ -81,7 +82,7 @@ runs:
 
     - name: Build greptime binary
       shell: bash
-      env: 
+      env:
         CARGO_BUILD_RUSTFLAGS: "-Clink-arg=-Wl,-ld_classic"
       run: |
         make build \

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

@@ -40,7 +40,7 @@ runs:
     - name: Install Python
       uses: actions/setup-python@v5
       with:
-        python-version: '3.10'
+        python-version: "3.10"
 
     - name: Install PyArrow Package
       shell: pwsh
@@ -62,13 +62,14 @@ runs:
       env:
         RUSTUP_WINDOWS_PATH_ADD_BIN: 1 # Workaround for https://github.com/nextest-rs/nextest/issues/1493
         RUST_BACKTRACE: 1
+        SQLNESS_OPTS: "--preserve-state"
 
     - name: Upload sqlness logs
       if: ${{ failure() }} # Only upload logs when the integration tests failed.
       uses: actions/upload-artifact@v4
       with:
         name: sqlness-logs
-        path: /tmp/greptime-*.log
+        path: C:\Users\RUNNER~1\AppData\Local\Temp\sqlness*
         retention-days: 3
 
     - name: Build greptime binary

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

@@ -123,10 +123,10 @@ runs:
         DST_REGISTRY_PASSWORD: ${{ inputs.dst-image-registry-password }}
       run: |
         ./.github/scripts/copy-image.sh \
-         ${{ inputs.src-image-registry }}/${{ inputs.src-image-namespace }}/${{ inputs.src-image-name }}-centos:latest \
+         ${{ inputs.src-image-registry }}/${{ inputs.src-image-namespace }}/${{ inputs.src-image-name }}-centos:${{ inputs.version }} \
          ${{ inputs.dst-image-registry }}/${{ inputs.dst-image-namespace }}
 
-    - name: Push greptimedb-centos image from DockerHub to ACR
+    - name: Push latest greptimedb-centos image from DockerHub to ACR
       shell: bash
       if: ${{ inputs.dev-mode == 'false' && inputs.push-latest-tag == 'true' }}
       env:

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

@@ -2,7 +2,7 @@ name: Setup Etcd cluster
 description: Deploy Etcd cluster on Kubernetes
 inputs:
   etcd-replicas:
-    default: 3
+    default: 1
     description: "Etcd replicas"
   namespace:
     default: "etcd-cluster"

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

@@ -31,17 +31,21 @@ runs:
   using: composite
   steps:
   - name: Install GreptimeDB operator
-    shell: bash
-    run: |
-      helm repo add greptime https://greptimeteam.github.io/helm-charts/ 
-      helm repo update
-      helm upgrade \
-        --install \
-        --create-namespace \
-        greptimedb-operator greptime/greptimedb-operator \
-        -n greptimedb-admin \
-        --wait \
-        --wait-for-jobs
+    uses: nick-fields/retry@v3
+    with: 
+      timeout_minutes: 3
+      max_attempts: 3
+      shell: bash
+      command: |
+        helm repo add greptime https://greptimeteam.github.io/helm-charts/ 
+        helm repo update
+        helm upgrade \
+          --install \
+          --create-namespace \
+          greptimedb-operator greptime/greptimedb-operator \
+          -n greptimedb-admin \
+          --wait \
+          --wait-for-jobs
   - name: Install GreptimeDB cluster
     shell: bash
     run: | 

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

@@ -1,18 +1,13 @@
 meta:
-  config: |-
+  configData: |-
     [runtime]
-    read_rt_size = 8
-    write_rt_size = 8
-    bg_rt_size = 8
+    global_rt_size = 4
 datanode:
-  config: |-
+  configData: |-
     [runtime]
-    read_rt_size = 8
-    write_rt_size = 8
-    bg_rt_size = 8
+    global_rt_size = 4
+    compact_rt_size = 2
 frontend:
-  config: |-
+  configData: |-
     [runtime]
-    read_rt_size = 8
-    write_rt_size = 8
-    bg_rt_size = 8
+    global_rt_size = 4

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

@@ -1,29 +1,24 @@
 meta:
-  config: |-
+  configData: |-
     [runtime]
-    read_rt_size = 8
-    write_rt_size = 8
-    bg_rt_size = 8
-    
+    global_rt_size = 4
+
     [datanode]
     [datanode.client]
     timeout = "60s"
 datanode:
-  config: |-
+  configData: |-
     [runtime]
-    read_rt_size = 8
-    write_rt_size = 8
-    bg_rt_size = 8
+    global_rt_size = 4
+    compact_rt_size = 2
     
     [storage]
     cache_path = "/data/greptimedb/s3cache"
     cache_capacity = "256MB"
 frontend:
-  config: |-
+  configData: |-
     [runtime]
-    read_rt_size = 8
-    write_rt_size = 8
-    bg_rt_size = 8
+    global_rt_size = 4
 
     [meta_client]
     ddl_timeout = "60s"

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

@@ -1,25 +1,20 @@
 meta:
-  config: |-
+  configData: |-
     [runtime]
-    read_rt_size = 8
-    write_rt_size = 8
-    bg_rt_size = 8
+    global_rt_size = 4
     
     [datanode]
     [datanode.client]
     timeout = "60s"
 datanode:
-  config: |-
+  configData: |-
     [runtime]
-    read_rt_size = 8
-    write_rt_size = 8
-    bg_rt_size = 8
+    global_rt_size = 4
+    compact_rt_size = 2
 frontend:
-  config: |-
+  configData: |-
     [runtime]
-    read_rt_size = 8
-    write_rt_size = 8
-    bg_rt_size = 8
+    global_rt_size = 4
 
     [meta_client]
     ddl_timeout = "60s"

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

@@ -1,9 +1,7 @@
 meta:
-  config: |-
+  configData: |-
     [runtime]
-    read_rt_size = 8
-    write_rt_size = 8
-    bg_rt_size = 8
+    global_rt_size = 4
     
     [wal]
     provider = "kafka"
@@ -15,22 +13,19 @@ meta:
     [datanode.client]
     timeout = "60s"
 datanode:
-  config: |-
+  configData: |-
     [runtime]
-    read_rt_size = 8
-    write_rt_size = 8
-    bg_rt_size = 8
+    global_rt_size = 4
+    compact_rt_size = 2
 
     [wal]
     provider = "kafka"
     broker_endpoints = ["kafka.kafka-cluster.svc.cluster.local:9092"]
     linger = "2ms"
 frontend:
-  config: |-
+  configData: |-
     [runtime]
-    read_rt_size = 8
-    write_rt_size = 8
-    bg_rt_size = 8
+    global_rt_size = 4
 
     [meta_client]
     ddl_timeout = "60s"
@@ -43,3 +38,8 @@ objectStorage:
   credentials:
     accessKeyId: rootuser
     secretAccessKey: rootpass123
+remoteWal:
+   enabled: true
+   kafka:
+     brokerEndpoints: 
+      - "kafka.kafka-cluster.svc.cluster.local:9092"

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

@@ -0,0 +1,30 @@
+name: Setup PostgreSQL
+description: Deploy PostgreSQL on Kubernetes
+inputs:
+  postgres-replicas:
+    default: 1
+    description: "Number of PostgreSQL replicas"
+  namespace:
+    default: "postgres-namespace"
+  postgres-version:
+    default: "14.2"
+    description: "PostgreSQL version"
+  storage-size:
+    default: "1Gi"
+    description: "Storage size for PostgreSQL"
+
+runs:
+  using: composite
+  steps:
+  - name: Install PostgreSQL
+    shell: bash
+    run: |
+      helm upgrade \
+        --install postgresql oci://registry-1.docker.io/bitnamicharts/postgresql \
+        --set replicaCount=${{ inputs.postgres-replicas }} \
+        --set image.tag=${{ inputs.postgres-version }} \
+        --set persistence.size=${{ inputs.storage-size }} \
+        --set postgresql.username=greptimedb \
+        --set postgresql.password=admin \
+        --create-namespace \
+        -n ${{ inputs.namespace }}

.github/workflows/develop.yml

@@ -139,10 +139,24 @@ jobs:
     name: Fuzz Test
     needs: build
     runs-on: ubuntu-latest
+    timeout-minutes: 60
     strategy:
+      fail-fast: false
       matrix:
         target: [ "fuzz_create_table", "fuzz_alter_table", "fuzz_create_database", "fuzz_create_logical_table", "fuzz_alter_logical_table", "fuzz_insert", "fuzz_insert_logical_table" ]
     steps:
+      - name: Remove unused software
+        run: |
+          echo "Disk space before:"
+          df -h
+          [[ -d /usr/share/dotnet ]] && sudo rm -rf /usr/share/dotnet
+          [[ -d /usr/local/lib/android ]] && sudo rm -rf /usr/local/lib/android
+          [[ -d /opt/ghc ]] && sudo rm -rf /opt/ghc
+          [[ -d /opt/hostedtoolcache/CodeQL ]] && sudo rm -rf /opt/hostedtoolcache/CodeQL
+          sudo docker image prune --all --force
+          sudo docker builder prune -a
+          echo "Disk space after:"
+          df -h
       - uses: actions/checkout@v4
       - uses: arduino/setup-protoc@v3
         with:
@@ -186,10 +200,23 @@ jobs:
     name: Unstable Fuzz Test
     needs: build-greptime-ci
     runs-on: ubuntu-latest
+    timeout-minutes: 60
     strategy:
       matrix:
         target: [ "unstable_fuzz_create_table_standalone" ]
     steps:
+      - name: Remove unused software
+        run: |
+          echo "Disk space before:"
+          df -h
+          [[ -d /usr/share/dotnet ]] && sudo rm -rf /usr/share/dotnet
+          [[ -d /usr/local/lib/android ]] && sudo rm -rf /usr/local/lib/android
+          [[ -d /opt/ghc ]] && sudo rm -rf /opt/ghc
+          [[ -d /opt/hostedtoolcache/CodeQL ]] && sudo rm -rf /opt/hostedtoolcache/CodeQL
+          sudo docker image prune --all --force
+          sudo docker builder prune -a
+          echo "Disk space after:"
+          df -h
       - uses: actions/checkout@v4
       - uses: arduino/setup-protoc@v3
         with:
@@ -278,27 +305,28 @@ jobs:
     name: Fuzz Test (Distributed, ${{ matrix.mode.name }}, ${{ matrix.target }})
     runs-on: ubuntu-latest
     needs:  build-greptime-ci
+    timeout-minutes: 60
     strategy:
       matrix:
         target: [ "fuzz_create_table", "fuzz_alter_table", "fuzz_create_database", "fuzz_create_logical_table", "fuzz_alter_logical_table", "fuzz_insert", "fuzz_insert_logical_table" ]
-        mode: 
-          - name: "Disk"
-            minio: false
-            kafka: false
-            values: "with-disk.yaml"
-          - name: "Minio"
-            minio: true
-            kafka: false
-            values: "with-minio.yaml"
-          - name: "Minio with Cache"
-            minio: true
-            kafka: false
-            values: "with-minio-and-cache.yaml"
+        mode:
           - name: "Remote WAL"
             minio: true
             kafka: true
             values: "with-remote-wal.yaml"
     steps:
+      - name: Remove unused software
+        run: |
+          echo "Disk space before:"
+          df -h
+          [[ -d /usr/share/dotnet ]] && sudo rm -rf /usr/share/dotnet
+          [[ -d /usr/local/lib/android ]] && sudo rm -rf /usr/local/lib/android
+          [[ -d /opt/ghc ]] && sudo rm -rf /opt/ghc
+          [[ -d /opt/hostedtoolcache/CodeQL ]] && sudo rm -rf /opt/hostedtoolcache/CodeQL
+          sudo docker image prune --all --force
+          sudo docker builder prune -a
+          echo "Disk space after:"
+          df -h
       - uses: actions/checkout@v4
       - name: Setup Kind
         uses: ./.github/actions/setup-kind
@@ -310,6 +338,8 @@ jobs:
         uses: ./.github/actions/setup-kafka-cluster
       - name: Setup Etcd cluser
         uses: ./.github/actions/setup-etcd-cluster
+      - name: Setup Postgres cluser
+        uses: ./.github/actions/setup-postgres-cluster
       # Prepares for fuzz tests
       - uses: arduino/setup-protoc@v3
         with:
@@ -413,15 +443,28 @@ jobs:
     name: Fuzz Test with Chaos (Distributed, ${{ matrix.mode.name }}, ${{ matrix.target }})
     runs-on: ubuntu-latest
     needs:  build-greptime-ci
+    timeout-minutes: 60
     strategy:
       matrix:
-        target: ["fuzz_failover_mito_regions"]
+        target: ["fuzz_migrate_mito_regions", "fuzz_failover_mito_regions", "fuzz_failover_metric_regions"]
         mode: 
           - name: "Remote WAL"
             minio: true
             kafka: true
             values: "with-remote-wal.yaml"
     steps:
+      - name: Remove unused software
+        run: |
+          echo "Disk space before:"
+          df -h
+          [[ -d /usr/share/dotnet ]] && sudo rm -rf /usr/share/dotnet
+          [[ -d /usr/local/lib/android ]] && sudo rm -rf /usr/local/lib/android
+          [[ -d /opt/ghc ]] && sudo rm -rf /opt/ghc
+          [[ -d /opt/hostedtoolcache/CodeQL ]] && sudo rm -rf /opt/hostedtoolcache/CodeQL
+          sudo docker image prune --all --force
+          sudo docker builder prune -a
+          echo "Disk space after:"
+          df -h
       - uses: actions/checkout@v4
       - name: Setup Kind
         uses: ./.github/actions/setup-kind
@@ -435,6 +478,8 @@ jobs:
         uses: ./.github/actions/setup-kafka-cluster
       - name: Setup Etcd cluser
         uses: ./.github/actions/setup-etcd-cluster
+      - name: Setup Postgres cluser
+        uses: ./.github/actions/setup-postgres-cluster
       # Prepares for fuzz tests
       - uses: arduino/setup-protoc@v3
         with:
@@ -552,6 +597,10 @@ jobs:
     timeout-minutes: 60
     steps:
       - uses: actions/checkout@v4
+      - if: matrix.mode.kafka
+        name: Setup kafka server
+        working-directory: tests-integration/fixtures/kafka
+        run: docker compose -f docker-compose-standalone.yml up -d --wait
       - name: Download pre-built binaries
         uses: actions/download-artifact@v4
         with:
@@ -559,10 +608,6 @@ jobs:
           path: .
       - name: Unzip binaries
         run: tar -xvf ./bins.tar.gz
-      - if: matrix.mode.kafka
-        name: Setup kafka server
-        working-directory: tests-integration/fixtures/kafka
-        run: docker compose -f docker-compose-standalone.yml up -d --wait
       - name: Run sqlness
         run: RUST_BACKTRACE=1 ./bins/sqlness-runner ${{ matrix.mode.opts }} -c ./tests/cases --bins-dir ./bins --preserve-state
       - name: Upload sqlness logs
@@ -661,6 +706,9 @@ jobs:
       - name: Setup minio
         working-directory: tests-integration/fixtures/minio
         run: docker compose -f docker-compose-standalone.yml up -d --wait
+      - name: Setup postgres server
+        working-directory: tests-integration/fixtures/postgres
+        run: docker compose -f docker-compose-standalone.yml up -d --wait
       - name: Run nextest cases
         run: cargo llvm-cov nextest --workspace --lcov --output-path lcov.info -F pyo3_backend -F dashboard
         env:
@@ -677,7 +725,9 @@ jobs:
           GT_MINIO_REGION: us-west-2
           GT_MINIO_ENDPOINT_URL: http://127.0.0.1:9000
           GT_ETCD_ENDPOINTS: http://127.0.0.1:2379
+          GT_POSTGRES_ENDPOINTS: postgres://greptimedb:admin@127.0.0.1:5432/postgres
           GT_KAFKA_ENDPOINTS: 127.0.0.1:9092
+          GT_KAFKA_SASL_ENDPOINTS: 127.0.0.1:9093
           UNITTEST_LOG_DIR: "__unittest_logs"
       - name: Codecov upload
         uses: codecov/codecov-action@v4

.github/workflows/nightly-build.yml

@@ -199,7 +199,7 @@ jobs:
           image-registry-username: ${{ secrets.DOCKERHUB_USERNAME }}
           image-registry-password: ${{ secrets.DOCKERHUB_TOKEN }}
           version: ${{ needs.allocate-runners.outputs.version }}
-          push-latest-tag: false # Don't push the latest tag to registry.
+          push-latest-tag: true
 
       - name: Set nightly build result
         id: set-nightly-build-result
@@ -240,7 +240,7 @@ jobs:
           aws-cn-region: ${{ vars.AWS_RELEASE_BUCKET_REGION }}
           dev-mode: false
           update-version-info: false  # Don't update version info in S3.
-          push-latest-tag: false      # Don't push the latest tag to registry.
+          push-latest-tag: true
 
   stop-linux-amd64-runner: # It's always run as the last job in the workflow to make sure that the runner is released.
     name: Stop linux-amd64 runner

.github/workflows/nightly-ci.yml

@@ -33,6 +33,13 @@ jobs:
           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 }}
+      - name: Upload sqlness logs
+        if: failure()
+        uses: actions/upload-artifact@v4
+        with:
+          name: sqlness-logs-kind
+          path: /tmp/kind/
+          retention-days: 3
 
   sqlness-windows:
     name: Sqlness tests on Windows
@@ -51,13 +58,15 @@ jobs:
       - name: Rust Cache
         uses: Swatinem/rust-cache@v2
       - name: Run sqlness
-        run: cargo sqlness
+        run: make sqlness-test
+        env:
+          SQLNESS_OPTS: "--preserve-state"
       - name: Upload sqlness logs
-        if: always()
+        if: failure()
         uses: actions/upload-artifact@v4
         with:
           name: sqlness-logs
-          path: /tmp/greptime-*.log
+          path: C:\Users\RUNNER~1\AppData\Local\Temp\sqlness*
           retention-days: 3
 
   test-on-windows:
@@ -109,11 +118,7 @@ jobs:
 
   check-status:
     name: Check status
-    needs: [
-      sqlness-test,
-      sqlness-windows,
-      test-on-windows,
-    ]
+    needs: [sqlness-test, sqlness-windows, test-on-windows]
     if: ${{ github.repository == 'GreptimeTeam/greptimedb' }}
     runs-on: ubuntu-20.04
     outputs:
@@ -127,9 +132,7 @@ jobs:
   notification:
     if: ${{ github.repository == 'GreptimeTeam/greptimedb' && always() }} # Not requiring successful dependent jobs, always run.
     name: Send notification to Greptime team
-    needs: [
-      check-status
-    ]
+    needs: [check-status]
     runs-on: ubuntu-20.04
     env:
       SLACK_WEBHOOK_URL: ${{ secrets.SLACK_WEBHOOK_URL_DEVELOP_CHANNEL }}

.github/workflows/release.yml

@@ -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: v0.9.0
+  NEXT_RELEASE_VERSION: v0.10.0
 
 # Permission reference: https://docs.github.com/en/actions/using-jobs/assigning-permissions-to-jobs
 permissions:

.pre-commit-config.yaml

@@ -16,6 +16,7 @@ repos:
     hooks:
     -    id: fmt
     -    id: clippy
-         args: ["--workspace", "--all-targets", "--", "-D", "warnings", "-D", "clippy::print_stdout", "-D", "clippy::print_stderr"]
+         args: ["--workspace", "--all-targets", "--all-features", "--", "-D", "warnings"]
          stages: [push]
     -    id: cargo-check
+         args: ["--workspace", "--all-targets", "--all-features"]

AUTHOR.md

@@ -0,0 +1,43 @@
+# GreptimeDB Authors
+
+## Individual Committers (in alphabetical order)
+
+* [CookiePieWw](https://github.com/CookiePieWw)
+* [KKould](https://github.com/KKould)
+* [NiwakaDev](https://github.com/NiwakaDev)
+* [etolbakov](https://github.com/etolbakov)
+* [irenjj](https://github.com/irenjj)
+
+## Team Members (in alphabetical order)
+
+* [Breeze-P](https://github.com/Breeze-P)
+* [GrepTime](https://github.com/GrepTime)
+* [MichaelScofield](https://github.com/MichaelScofield)
+* [Wenjie0329](https://github.com/Wenjie0329)
+* [WenyXu](https://github.com/WenyXu)
+* [ZonaHex](https://github.com/ZonaHex)
+* [apdong2022](https://github.com/apdong2022)
+* [beryl678](https://github.com/beryl678)
+* [daviderli614](https://github.com/daviderli614)
+* [discord9](https://github.com/discord9)
+* [evenyag](https://github.com/evenyag)
+* [fengjiachun](https://github.com/fengjiachun)
+* [fengys1996](https://github.com/fengys1996)
+* [holalengyu](https://github.com/holalengyu)
+* [killme2008](https://github.com/killme2008)
+* [nicecui](https://github.com/nicecui)
+* [paomian](https://github.com/paomian)
+* [shuiyisong](https://github.com/shuiyisong)
+* [sunchanglong](https://github.com/sunchanglong)
+* [sunng87](https://github.com/sunng87)
+* [tisonkun](https://github.com/tisonkun)
+* [v0y4g3r](https://github.com/v0y4g3r)
+* [waynexia](https://github.com/waynexia)
+* [xtang](https://github.com/xtang)
+* [zhaoyingnan01](https://github.com/zhaoyingnan01)
+* [zhongzc](https://github.com/zhongzc)
+* [zyy17](https://github.com/zyy17)
+
+## All Contributors
+
+[![All Contributors](https://contrib.rocks/image?repo=GreptimeTeam/greptimedb)](https://github.com/GreptimeTeam/greptimedb/graphs/contributors)

CONTRIBUTING.md

@@ -4,10 +4,7 @@ Thanks a lot for considering contributing to GreptimeDB. We believe people like
 
 You can find our contributors at https://github.com/GreptimeTeam/greptimedb/graphs/contributors. When you dedicate to GreptimeDB for a few months and keep bringing high-quality contributions (code, docs, advocate, etc.), you will be a candidate of a committer.
 
-A committer will be granted both read & write access to GreptimeDB repos. Here is a list of current committers except GreptimeDB team members:
-
-* [Eugene Tolbakov](https://github.com/etolbakov): PromQL support, SQL engine, InfluxDB APIs, and more.
-* [@NiwakaDev](https://github.com/NiwakaDev): SQL engine and storage layer.
+A committer will be granted both read & write access to GreptimeDB repos. Check the [AUTHOR.md](AUTHOR.md) file for all current individual committers.
 
 Please read the guidelines, and they can help you get started. Communicate respectfully with the 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.
 
@@ -17,7 +14,7 @@ Follow our [README](https://github.com/GreptimeTeam/greptimedb#readme) to get th
 
 It can feel intimidating to contribute to a complex project, but it can also be exciting and fun. These general notes will help everyone participate in this communal activity.
 
-- Follow the [Code of Conduct](https://github.com/GreptimeTeam/greptimedb/blob/main/CODE_OF_CONDUCT.md)
+- Follow the [Code of Conduct](https://github.com/GreptimeTeam/.github/blob/main/.github/CODE_OF_CONDUCT.md)
 - Small changes make huge differences. We will happily accept a PR making a single character change if it helps move forward. Don't wait to have everything working.
 - Check the closed issues before opening your issue.
 - Try to follow the existing style of the code.
@@ -33,7 +30,7 @@ Pull requests are great, but we accept all kinds of other help if you like. Such
 
 ## Code of Conduct
 
-Also, there are things that we are not looking for because they don't match the goals of the product or benefit the community. Please read [Code of Conduct](https://github.com/GreptimeTeam/greptimedb/blob/main/CODE_OF_CONDUCT.md); we hope everyone can keep good manners and become an honored member.
+Also, there are things that we are not looking for because they don't match the goals of the product or benefit the community. Please read [Code of Conduct](https://github.com/GreptimeTeam/.github/blob/main/.github/CODE_OF_CONDUCT.md); we hope everyone can keep good manners and become an honored member.
 
 ## License
 
@@ -58,7 +55,7 @@ GreptimeDB uses the [Apache 2.0 license](https://github.com/GreptimeTeam/greptim
 - 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/) and [style guide](docs/style-guide.md).
-- Make sure all unit tests are passed (using `cargo test --workspace` or [nextest](https://nexte.st/index.html) `cargo nextest run`).
+- Make sure all unit tests are passed using [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`).
 
 #### `pre-commit` Hooks

Cargo.toml

@@ -64,7 +64,7 @@ members = [
 resolver = "2"
 
 [workspace.package]
-version = "0.8.2"
+version = "0.9.2"
 edition = "2021"
 license = "Apache-2.0"
 
@@ -104,28 +104,27 @@ clap = { version = "4.4", features = ["derive"] }
 config = "0.13.0"
 crossbeam-utils = "0.8"
 dashmap = "5.4"
-datafusion = { git = "https://github.com/apache/datafusion.git", rev = "729b356ef543ffcda6813c7b5373507a04ae0109" }
-datafusion-common = { git = "https://github.com/apache/datafusion.git", rev = "729b356ef543ffcda6813c7b5373507a04ae0109" }
-datafusion-expr = { git = "https://github.com/apache/datafusion.git", rev = "729b356ef543ffcda6813c7b5373507a04ae0109" }
-datafusion-functions = { git = "https://github.com/apache/datafusion.git", rev = "729b356ef543ffcda6813c7b5373507a04ae0109" }
-datafusion-optimizer = { git = "https://github.com/apache/datafusion.git", rev = "729b356ef543ffcda6813c7b5373507a04ae0109" }
-datafusion-physical-expr = { git = "https://github.com/apache/datafusion.git", rev = "729b356ef543ffcda6813c7b5373507a04ae0109" }
-datafusion-physical-plan = { git = "https://github.com/apache/datafusion.git", rev = "729b356ef543ffcda6813c7b5373507a04ae0109" }
-datafusion-sql = { git = "https://github.com/apache/datafusion.git", rev = "729b356ef543ffcda6813c7b5373507a04ae0109" }
-datafusion-substrait = { git = "https://github.com/apache/datafusion.git", rev = "729b356ef543ffcda6813c7b5373507a04ae0109" }
+datafusion = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "7823ef2f63663907edab46af0d51359900f608d6" }
+datafusion-common = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "7823ef2f63663907edab46af0d51359900f608d6" }
+datafusion-expr = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "7823ef2f63663907edab46af0d51359900f608d6" }
+datafusion-functions = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "7823ef2f63663907edab46af0d51359900f608d6" }
+datafusion-optimizer = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "7823ef2f63663907edab46af0d51359900f608d6" }
+datafusion-physical-expr = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "7823ef2f63663907edab46af0d51359900f608d6" }
+datafusion-physical-plan = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "7823ef2f63663907edab46af0d51359900f608d6" }
+datafusion-sql = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "7823ef2f63663907edab46af0d51359900f608d6" }
+datafusion-substrait = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "7823ef2f63663907edab46af0d51359900f608d6" }
 derive_builder = "0.12"
 dotenv = "0.15"
-# TODO(LFC): Wait for https://github.com/etcdv3/etcd-client/pull/76
-etcd-client = { git = "https://github.com/MichaelScofield/etcd-client.git", rev = "4c371e9b3ea8e0a8ee2f9cbd7ded26e54a45df3b" }
+etcd-client = { version = "0.13" }
 fst = "0.4.7"
 futures = "0.3"
 futures-util = "0.3"
-greptime-proto = { git = "https://github.com/GreptimeTeam/greptime-proto.git", rev = "a70a6af9c69e40f9a918936a48717343402b4393" }
+greptime-proto = { git = "https://github.com/GreptimeTeam/greptime-proto.git", rev = "c437b55725b7f5224fe9d46db21072b4a682ee4b" }
 humantime = "2.1"
 humantime-serde = "1.1"
 itertools = "0.10"
 lazy_static = "1.4"
-meter-core = { git = "https://github.com/GreptimeTeam/greptime-meter.git", rev = "80b72716dcde47ec4161478416a5c6c21343364d" }
+meter-core = { git = "https://github.com/GreptimeTeam/greptime-meter.git", rev = "80eb97c24c88af4dd9a86f8bbaf50e741d4eb8cd" }
 mockall = "0.11.4"
 moka = "0.12"
 notify = "6.1"
@@ -152,14 +151,19 @@ reqwest = { version = "0.12", default-features = false, features = [
     "stream",
     "multipart",
 ] }
-rskafka = "0.5"
+# SCRAM-SHA-512 requires https://github.com/dequbed/rsasl/pull/48, https://github.com/influxdata/rskafka/pull/247 
+rskafka = { git = "https://github.com/WenyXu/rskafka.git", rev = "940c6030012c5b746fad819fb72e3325b26e39de", features = [
+    "transport-tls",
+] }
 rstest = "0.21"
 rstest_reuse = "0.7"
 rust_decimal = "1.33"
+rustc-hash = "2.0"
 schemars = "0.8"
 serde = { version = "1.0", features = ["derive"] }
 serde_json = { version = "1.0", features = ["float_roundtrip"] }
 serde_with = "3"
+shadow-rs = "0.31"
 smallvec = { version = "1", features = ["serde"] }
 snafu = "0.8"
 sysinfo = "0.30"
@@ -170,6 +174,7 @@ sqlparser = { git = "https://github.com/GreptimeTeam/sqlparser-rs.git", rev = "5
 strum = { version = "0.25", features = ["derive"] }
 tempfile = "3"
 tokio = { version = "1.36", features = ["full"] }
+tokio-postgres = "0.7"
 tokio-stream = { version = "0.1" }
 tokio-util = { version = "0.7", features = ["io-util", "compat"] }
 toml = "0.8.8"
@@ -184,7 +189,7 @@ auth = { path = "src/auth" }
 cache = { path = "src/cache" }
 catalog = { path = "src/catalog" }
 client = { path = "src/client" }
-cmd = { path = "src/cmd" }
+cmd = { path = "src/cmd", default-features = false }
 common-base = { path = "src/common/base" }
 common-catalog = { path = "src/common/catalog" }
 common-config = { path = "src/common/config" }
@@ -214,7 +219,7 @@ datanode = { path = "src/datanode" }
 datatypes = { path = "src/datatypes" }
 file-engine = { path = "src/file-engine" }
 flow = { path = "src/flow" }
-frontend = { path = "src/frontend" }
+frontend = { path = "src/frontend", default-features = false }
 index = { path = "src/index" }
 log-store = { path = "src/log-store" }
 meta-client = { path = "src/meta-client" }
@@ -239,7 +244,7 @@ table = { path = "src/table" }
 
 [workspace.dependencies.meter-macros]
 git = "https://github.com/GreptimeTeam/greptime-meter.git"
-rev = "80b72716dcde47ec4161478416a5c6c21343364d"
+rev = "80eb97c24c88af4dd9a86f8bbaf50e741d4eb8cd"
 
 [profile.release]
 debug = 1

Makefile

@@ -15,6 +15,7 @@ RUST_TOOLCHAIN ?= $(shell cat rust-toolchain.toml | grep channel | cut -d'"' -f2
 CARGO_REGISTRY_CACHE ?= ${HOME}/.cargo/registry
 ARCH := $(shell uname -m | sed 's/x86_64/amd64/' | sed 's/aarch64/arm64/')
 OUTPUT_DIR := $(shell if [ "$(RELEASE)" = "true" ]; then echo "release"; elif [ ! -z "$(CARGO_PROFILE)" ]; then echo "$(CARGO_PROFILE)" ; else echo "debug"; fi)
+SQLNESS_OPTS ?=
 
 # The arguments for running integration tests.
 ETCD_VERSION ?= v3.5.9
@@ -161,7 +162,7 @@ nextest: ## Install nextest tools.
 
 .PHONY: sqlness-test
 sqlness-test: ## Run sqlness test.
-	cargo sqlness
+	cargo sqlness ${SQLNESS_OPTS}
 
 # Run fuzz test ${FUZZ_TARGET}.
 RUNS ?= 1
@@ -172,7 +173,7 @@ fuzz:
 
 .PHONY: fuzz-ls
 fuzz-ls:
-	cargo fuzz list --fuzz-dir tests-fuzz 
+	cargo fuzz list --fuzz-dir tests-fuzz
 
 .PHONY: check
 check: ## Cargo check all the targets.
@@ -205,10 +206,14 @@ run-it-in-container: start-etcd ## Run integration tests in dev-builder.
 	-w /greptimedb ${IMAGE_REGISTRY}/${IMAGE_NAMESPACE}/dev-builder-${BASE_IMAGE}:latest \
 	make test sqlness-test BUILD_JOBS=${BUILD_JOBS}
 
-.PHONY: run-cluster-with-etcd
-run-cluster-with-etcd: ## Run greptime cluster with etcd in docker-compose.
+.PHONY: start-cluster
+start-cluster: ## Start the greptimedb cluster with etcd by using docker compose.
 	 docker compose -f ./docker/docker-compose/cluster-with-etcd.yaml up
 
+.PHONY: stop-cluster
+stop-cluster: ## Stop the greptimedb cluster that created by docker compose.
+	docker compose -f ./docker/docker-compose/cluster-with-etcd.yaml stop
+
 ##@ Docs
 config-docs: ## Generate configuration documentation from toml files.
 	docker run --rm \

README.md

@@ -6,12 +6,12 @@
   </picture>
 </p>
 
-<h1 align="center">Cloud-scale, Fast and Efficient Time Series Database</h1>
+<h2 align="center">Unified Time Series Database for Metrics, Logs, and Events</h2>
 
 <div align="center">
 <h3 align="center">
   <a href="https://greptime.com/product/cloud">GreptimeCloud</a> |
-  <a href="https://docs.greptime.com/">User guide</a> |
+  <a href="https://docs.greptime.com/">User Guide</a> |
   <a href="https://greptimedb.rs/">API Docs</a> |
   <a href="https://github.com/GreptimeTeam/greptimedb/issues/3412">Roadmap 2024</a>
 </h4>
@@ -50,24 +50,23 @@
 
 ## Introduction
 
-**GreptimeDB** is an open-source time-series database focusing on efficiency, scalability, and analytical capabilities.
-Designed to work on infrastructure of the cloud era, GreptimeDB benefits users with its elasticity and commodity storage, offering a fast and cost-effective **alternative to InfluxDB** and a **long-term storage for Prometheus**.
+**GreptimeDB** is an open-source unified time-series database for **Metrics**, **Logs**, and **Events** (also **Traces** in plan). You can gain real-time insights from Edge to Cloud at any scale.
 
 ## Why GreptimeDB
 
 Our core developers have been building time-series data platforms for years. Based on our best-practices, GreptimeDB is born to give you:
 
-* **Easy horizontal scaling**
+* **Unified all kinds of time series**
 
-  Seamless scalability from a standalone binary at edge to a robust, highly available distributed cluster in cloud, with a transparent experience for both developers and administrators.
+  GreptimeDB treats all time series as contextual events with timestamp, and thus unifies the processing of metrics, logs, and events. It supports analyzing metrics, logs, and events with SQL and PromQL, and doing streaming with continuous aggregation.
 
-* **Analyzing time-series data**
+* **Cloud-Edge collaboration**
 
-  Query your time-series data with SQL and PromQL. Use Python scripts to facilitate complex analytical tasks.
+  GreptimeDB can be deployed on ARM architecture-compatible Android/Linux systems as well as cloud environments from various vendors. Both sides run the same software, providing identical APIs and control planes, so your application can run at the edge or on the cloud without modification, and data synchronization also becomes extremely easy and efficient.
 
 * **Cloud-native distributed database**
 
-  Fully open-source distributed cluster architecture that harnesses the power of cloud-native elastic computing resources.
+  By leveraging object storage (S3 and others), separating compute and storage, scaling stateless compute nodes arbitrarily, GreptimeDB implements seamless scalability. It also supports cross-cloud deployment with a built-in unified data access layer over different object storages.
 
 * **Performance and Cost-effective**
 
@@ -105,10 +104,10 @@ Read more about [Installation](https://docs.greptime.com/getting-started/install
 
 ## Getting Started
 
-* [Quickstart](https://docs.greptime.com/getting-started/quick-start/overview)
-* [Write Data](https://docs.greptime.com/user-guide/clients/overview)
-* [Query Data](https://docs.greptime.com/user-guide/query-data/overview)
-* [Operations](https://docs.greptime.com/user-guide/operations/overview)
+* [Quickstart](https://docs.greptime.com/getting-started/quick-start)
+* [User Guide](https://docs.greptime.com/user-guide/overview)
+* [Demos](https://github.com/GreptimeTeam/demo-scene)
+* [FAQ](https://docs.greptime.com/faq-and-others/faq)
 
 ## Build
 
@@ -151,9 +150,10 @@ Our official Grafana dashboard is available at [grafana](grafana/README.md) dire
 
 ## Project Status
 
-The current version has not yet reached General Availability version standards.
-In line with our Greptime 2024 Roadmap, we plan to achieve a production-level
-version with the update to v1.0 in August. [[Join Force]](https://github.com/GreptimeTeam/greptimedb/issues/3412)
+The current version has not yet reached the standards for General Availability. 
+According to our Greptime 2024 Roadmap, we aim to achieve a production-level version with the release of v1.0 by the end of 2024. [Join Us](https://github.com/GreptimeTeam/greptimedb/issues/3412)
+
+We welcome you to test and use GreptimeDB. Some users have already adopted it in their production environments. If you're interested in trying it out, please use the latest stable release available.
 
 ## Community
 
@@ -183,6 +183,8 @@ Please refer to [contribution guidelines](CONTRIBUTING.md) and [internal concept
 
 ## Acknowledgement
 
+Special thanks to all the contributors who have propelled GreptimeDB forward. For a complete list of contributors, please refer to [AUTHOR.md](AUTHOR.md).
+
 - GreptimeDB uses [Apache Arrow™](https://arrow.apache.org/) as the memory model and [Apache Parquet™](https://parquet.apache.org/) as the persistent file format.
 - GreptimeDB's query engine is powered by [Apache Arrow DataFusion™](https://arrow.apache.org/datafusion/).
 - [Apache OpenDAL™](https://opendal.apache.org) gives GreptimeDB a very general and elegant data access abstraction layer.

config/config-docs-template.md

@@ -1,10 +1,12 @@
 # Configurations
 
-- [Standalone Mode](#standalone-mode)
-- [Distributed Mode](#distributed-mode)
+- [Configurations](#configurations)
+  - [Standalone Mode](#standalone-mode)
+  - [Distributed Mode](#distributed-mode)
     - [Frontend](#frontend)
     - [Metasrv](#metasrv)
     - [Datanode](#datanode)
+    - [Flownode](#flownode)
 
 ## Standalone Mode
 
@@ -23,3 +25,7 @@
 ### Datanode
 
 {{ toml2docs "./datanode.example.toml" }}
+
+### Flownode
+
+{{ toml2docs "./flownode.example.toml"}}

config/config.md

@@ -1,10 +1,12 @@
 # Configurations
 
-- [Standalone Mode](#standalone-mode)
-- [Distributed Mode](#distributed-mode)
+- [Configurations](#configurations)
+  - [Standalone Mode](#standalone-mode)
+  - [Distributed Mode](#distributed-mode)
     - [Frontend](#frontend)
     - [Metasrv](#metasrv)
     - [Datanode](#datanode)
+    - [Flownode](#flownode)
 
 ## Standalone Mode
 
@@ -14,9 +16,8 @@
 | `enable_telemetry` | Bool | `true` | Enable telemetry to collect anonymous usage data. |
 | `default_timezone` | String | `None` | The default timezone of the server. |
 | `runtime` | -- | -- | The runtime options. |
-| `runtime.read_rt_size` | Integer | `8` | The number of threads to execute the runtime for global read operations. |
-| `runtime.write_rt_size` | Integer | `8` | The number of threads to execute the runtime for global write operations. |
-| `runtime.bg_rt_size` | Integer | `4` | The number of threads to execute the runtime for global background operations. |
+| `runtime.global_rt_size` | Integer | `8` | The number of threads to execute the runtime for global read operations. |
+| `runtime.compact_rt_size` | Integer | `4` | The number of threads to execute the runtime for global write operations. |
 | `http` | -- | -- | The HTTP server options. |
 | `http.addr` | String | `127.0.0.1:4000` | The address to bind the HTTP server. |
 | `http.timeout` | String | `30s` | HTTP request timeout. Set to 0 to disable timeout. |
@@ -66,6 +67,11 @@
 | `wal.prefill_log_files` | Bool | `false` | Whether to pre-create log files on start up.<br/>**It's only used when the provider is `raft_engine`**. |
 | `wal.sync_period` | String | `10s` | Duration for fsyncing log files.<br/>**It's only used when the provider is `raft_engine`**. |
 | `wal.broker_endpoints` | Array | -- | The Kafka broker endpoints.<br/>**It's only used when the provider is `kafka`**. |
+| `wal.num_topics` | Integer | `64` | Number of topics to be created upon start.<br/>**It's only used when the provider is `kafka`**. |
+| `wal.selector_type` | String | `round_robin` | Topic selector type.<br/>Available selector types:<br/>- `round_robin` (default)<br/>**It's only used when the provider is `kafka`**. |
+| `wal.topic_name_prefix` | String | `greptimedb_wal_topic` | A Kafka topic is constructed by concatenating `topic_name_prefix` and `topic_id`.<br/>**It's only used when the provider is `kafka`**. |
+| `wal.replication_factor` | Integer | `1` | Expected number of replicas of each partition.<br/>**It's only used when the provider is `kafka`**. |
+| `wal.create_topic_timeout` | String | `30s` | Above which a topic creation operation will be cancelled.<br/>**It's only used when the provider is `kafka`**. |
 | `wal.max_batch_bytes` | String | `1MB` | The max size of a single producer batch.<br/>Warning: Kafka has a default limit of 1MB per message in a topic.<br/>**It's only used when the provider is `kafka`**. |
 | `wal.consumer_wait_timeout` | String | `100ms` | The consumer wait timeout.<br/>**It's only used when the provider is `kafka`**. |
 | `wal.backoff_init` | String | `500ms` | The initial backoff delay.<br/>**It's only used when the provider is `kafka`**. |
@@ -92,6 +98,7 @@
 | `storage.account_key` | String | `None` | The account key of the azure account.<br/>**It's only used when the storage type is `Azblob`**. |
 | `storage.scope` | String | `None` | The scope of the google cloud storage.<br/>**It's only used when the storage type is `Gcs`**. |
 | `storage.credential_path` | String | `None` | The credential path of the google cloud storage.<br/>**It's only used when the storage type is `Gcs`**. |
+| `storage.credential` | String | `None` | The credential of the google cloud storage.<br/>**It's only used when the storage type is `Gcs`**. |
 | `storage.container` | String | `None` | The container of the azure account.<br/>**It's only used when the storage type is `Azblob`**. |
 | `storage.sas_token` | String | `None` | The sas token of the azure account.<br/>**It's only used when the storage type is `Azblob`**. |
 | `storage.endpoint` | String | `None` | The endpoint of the S3 service.<br/>**It's only used when the storage type is `S3`, `Oss`, `Gcs` and `Azblob`**. |
@@ -118,22 +125,33 @@
 | `region_engine.mito.scan_parallelism` | Integer | `0` | Parallelism to scan a region (default: 1/4 of cpu cores).<br/>- `0`: using the default value (1/4 of cpu cores).<br/>- `1`: scan in current thread.<br/>- `n`: scan in parallelism n. |
 | `region_engine.mito.parallel_scan_channel_size` | Integer | `32` | Capacity of the channel to send data from parallel scan tasks to the main task. |
 | `region_engine.mito.allow_stale_entries` | Bool | `false` | Whether to allow stale WAL entries read during replay. |
+| `region_engine.mito.index` | -- | -- | The options for index in Mito engine. |
+| `region_engine.mito.index.aux_path` | String | `""` | Auxiliary directory path for the index in filesystem, used to store intermediate files for<br/>creating the index and staging files for searching the index, defaults to `{data_home}/index_intermediate`.<br/>The default name for this directory is `index_intermediate` for backward compatibility.<br/><br/>This path contains two subdirectories:<br/>- `__intm`: for storing intermediate files used during creating index.<br/>- `staging`: for storing staging files used during searching index. |
+| `region_engine.mito.index.staging_size` | String | `2GB` | The max capacity of the staging directory. |
 | `region_engine.mito.inverted_index` | -- | -- | The options for inverted index in Mito engine. |
-| `region_engine.mito.inverted_index.create_on_flush` | String | `auto` | Whether to create the index on flush.<br/>- `auto`: automatically<br/>- `disable`: never |
-| `region_engine.mito.inverted_index.create_on_compaction` | String | `auto` | Whether to create the index on compaction.<br/>- `auto`: automatically<br/>- `disable`: never |
-| `region_engine.mito.inverted_index.apply_on_query` | String | `auto` | Whether to apply the index on query<br/>- `auto`: automatically<br/>- `disable`: never |
-| `region_engine.mito.inverted_index.mem_threshold_on_create` | String | `64M` | Memory threshold for performing an external sort during index creation.<br/>Setting to empty will disable external sorting, forcing all sorting operations to happen in memory. |
-| `region_engine.mito.inverted_index.intermediate_path` | String | `""` | File system path to store intermediate files for external sorting (default `{data_home}/index_intermediate`). |
+| `region_engine.mito.inverted_index.create_on_flush` | String | `auto` | Whether to create the index on flush.<br/>- `auto`: automatically (default)<br/>- `disable`: never |
+| `region_engine.mito.inverted_index.create_on_compaction` | String | `auto` | Whether to create the index on compaction.<br/>- `auto`: automatically (default)<br/>- `disable`: never |
+| `region_engine.mito.inverted_index.apply_on_query` | String | `auto` | Whether to apply the index on query<br/>- `auto`: automatically (default)<br/>- `disable`: never |
+| `region_engine.mito.inverted_index.mem_threshold_on_create` | String | `auto` | Memory threshold for performing an external sort during index creation.<br/>- `auto`: automatically determine the threshold based on the system memory size (default)<br/>- `unlimited`: no memory limit<br/>- `[size]` e.g. `64MB`: fixed memory threshold |
+| `region_engine.mito.inverted_index.intermediate_path` | String | `""` | Deprecated, use `region_engine.mito.index.aux_path` instead. |
+| `region_engine.mito.inverted_index.metadata_cache_size` | String | `64MiB` | Cache size for inverted index metadata. |
+| `region_engine.mito.inverted_index.content_cache_size` | String | `128MiB` | Cache size for inverted index content. |
+| `region_engine.mito.fulltext_index` | -- | -- | The options for full-text index in Mito engine. |
+| `region_engine.mito.fulltext_index.create_on_flush` | String | `auto` | Whether to create the index on flush.<br/>- `auto`: automatically (default)<br/>- `disable`: never |
+| `region_engine.mito.fulltext_index.create_on_compaction` | String | `auto` | Whether to create the index on compaction.<br/>- `auto`: automatically (default)<br/>- `disable`: never |
+| `region_engine.mito.fulltext_index.apply_on_query` | String | `auto` | Whether to apply the index on query<br/>- `auto`: automatically (default)<br/>- `disable`: never |
+| `region_engine.mito.fulltext_index.mem_threshold_on_create` | String | `auto` | Memory threshold for index creation.<br/>- `auto`: automatically determine the threshold based on the system memory size (default)<br/>- `unlimited`: no memory limit<br/>- `[size]` e.g. `64MB`: fixed memory threshold |
 | `region_engine.mito.memtable` | -- | -- | -- |
 | `region_engine.mito.memtable.type` | String | `time_series` | Memtable type.<br/>- `time_series`: time-series memtable<br/>- `partition_tree`: partition tree memtable (experimental) |
 | `region_engine.mito.memtable.index_max_keys_per_shard` | Integer | `8192` | The max number of keys in one shard.<br/>Only available for `partition_tree` memtable. |
 | `region_engine.mito.memtable.data_freeze_threshold` | Integer | `32768` | The max rows of data inside the actively writing buffer in one shard.<br/>Only available for `partition_tree` memtable. |
 | `region_engine.mito.memtable.fork_dictionary_bytes` | String | `1GiB` | Max dictionary bytes.<br/>Only available for `partition_tree` memtable. |
+| `region_engine.file` | -- | -- | Enable the file engine. |
 | `logging` | -- | -- | The logging options. |
 | `logging.dir` | String | `/tmp/greptimedb/logs` | The directory to store the log files. |
 | `logging.level` | String | `None` | The log level. Can be `info`/`debug`/`warn`/`error`. |
 | `logging.enable_otlp_tracing` | Bool | `false` | Enable OTLP tracing. |
-| `logging.otlp_endpoint` | String | `None` | The OTLP tracing endpoint. |
+| `logging.otlp_endpoint` | String | `http://localhost:4317` | The OTLP tracing endpoint. |
 | `logging.append_stdout` | Bool | `true` | Whether to append logs to stdout. |
 | `logging.tracing_sample_ratio` | -- | -- | The percentage of tracing will be sampled and exported.<br/>Valid range `[0, 1]`, 1 means all traces are sampled, 0 means all traces are not sampled, the default value is 1.<br/>ratio > 1 are treated as 1. Fractions < 0 are treated as 0 |
 | `logging.tracing_sample_ratio.default_ratio` | Float | `1.0` | -- |
@@ -155,12 +173,10 @@
 
 | Key | Type | Default | Descriptions |
 | --- | -----| ------- | ----------- |
-| `mode` | String | `standalone` | The running mode of the datanode. It can be `standalone` or `distributed`. |
 | `default_timezone` | String | `None` | The default timezone of the server. |
 | `runtime` | -- | -- | The runtime options. |
-| `runtime.read_rt_size` | Integer | `8` | The number of threads to execute the runtime for global read operations. |
-| `runtime.write_rt_size` | Integer | `8` | The number of threads to execute the runtime for global write operations. |
-| `runtime.bg_rt_size` | Integer | `4` | The number of threads to execute the runtime for global background operations. |
+| `runtime.global_rt_size` | Integer | `8` | The number of threads to execute the runtime for global read operations. |
+| `runtime.compact_rt_size` | Integer | `4` | The number of threads to execute the runtime for global write operations. |
 | `heartbeat` | -- | -- | The heartbeat options. |
 | `heartbeat.interval` | String | `18s` | Interval for sending heartbeat messages to the metasrv. |
 | `heartbeat.retry_interval` | String | `3s` | Interval for retrying to send heartbeat messages to the metasrv. |
@@ -220,7 +236,7 @@
 | `logging.dir` | String | `/tmp/greptimedb/logs` | The directory to store the log files. |
 | `logging.level` | String | `None` | The log level. Can be `info`/`debug`/`warn`/`error`. |
 | `logging.enable_otlp_tracing` | Bool | `false` | Enable OTLP tracing. |
-| `logging.otlp_endpoint` | String | `None` | The OTLP tracing endpoint. |
+| `logging.otlp_endpoint` | String | `http://localhost:4317` | The OTLP tracing endpoint. |
 | `logging.append_stdout` | Bool | `true` | Whether to append logs to stdout. |
 | `logging.tracing_sample_ratio` | -- | -- | The percentage of tracing will be sampled and exported.<br/>Valid range `[0, 1]`, 1 means all traces are sampled, 0 means all traces are not sampled, the default value is 1.<br/>ratio > 1 are treated as 1. Fractions < 0 are treated as 0 |
 | `logging.tracing_sample_ratio.default_ratio` | Float | `1.0` | -- |
@@ -244,28 +260,28 @@
 | `bind_addr` | String | `127.0.0.1:3002` | The bind address of metasrv. |
 | `server_addr` | String | `127.0.0.1:3002` | The communication server address for frontend and datanode to connect to metasrv,  "127.0.0.1:3002" by default for localhost. |
 | `store_addr` | String | `127.0.0.1:2379` | Etcd server address. |
-| `selector` | String | `lease_based` | Datanode selector type.<br/>- `lease_based` (default value).<br/>- `load_based`<br/>For details, please see "https://docs.greptime.com/developer-guide/metasrv/selector". |
+| `selector` | String | `round_robin` | Datanode selector type.<br/>- `round_robin` (default value)<br/>- `lease_based`<br/>- `load_based`<br/>For details, please see "https://docs.greptime.com/developer-guide/metasrv/selector". |
 | `use_memory_store` | Bool | `false` | Store data in memory. |
 | `enable_telemetry` | Bool | `true` | Whether to enable greptimedb telemetry. |
 | `store_key_prefix` | String | `""` | If it's not empty, the metasrv will store all data with this key prefix. |
+| `enable_region_failover` | Bool | `false` | Whether to enable region failover.<br/>This feature is only available on GreptimeDB running on cluster mode and<br/>- Using Remote WAL<br/>- Using shared storage (e.g., s3). |
 | `runtime` | -- | -- | The runtime options. |
-| `runtime.read_rt_size` | Integer | `8` | The number of threads to execute the runtime for global read operations. |
-| `runtime.write_rt_size` | Integer | `8` | The number of threads to execute the runtime for global write operations. |
-| `runtime.bg_rt_size` | Integer | `4` | The number of threads to execute the runtime for global background operations. |
+| `runtime.global_rt_size` | Integer | `8` | The number of threads to execute the runtime for global read operations. |
+| `runtime.compact_rt_size` | Integer | `4` | The number of threads to execute the runtime for global write operations. |
 | `procedure` | -- | -- | Procedure storage options. |
 | `procedure.max_retry_times` | Integer | `12` | Procedure max retry time. |
 | `procedure.retry_delay` | String | `500ms` | Initial retry delay of procedures, increases exponentially |
 | `procedure.max_metadata_value_size` | String | `1500KiB` | Auto split large value<br/>GreptimeDB procedure uses etcd as the default metadata storage backend.<br/>The etcd the maximum size of any request is 1.5 MiB<br/>1500KiB = 1536KiB (1.5MiB) - 36KiB (reserved size of key)<br/>Comments out the `max_metadata_value_size`, for don't split large value (no limit). |
 | `failure_detector` | -- | -- | -- |
-| `failure_detector.threshold` | Float | `8.0` | -- |
-| `failure_detector.min_std_deviation` | String | `100ms` | -- |
-| `failure_detector.acceptable_heartbeat_pause` | String | `3000ms` | -- |
-| `failure_detector.first_heartbeat_estimate` | String | `1000ms` | -- |
+| `failure_detector.threshold` | Float | `8.0` | The threshold value used by the failure detector to determine failure conditions. |
+| `failure_detector.min_std_deviation` | String | `100ms` | The minimum standard deviation of the heartbeat intervals, used to calculate acceptable variations. |
+| `failure_detector.acceptable_heartbeat_pause` | String | `10000ms` | The acceptable pause duration between heartbeats, used to determine if a heartbeat interval is acceptable. |
+| `failure_detector.first_heartbeat_estimate` | String | `1000ms` | The initial estimate of the heartbeat interval used by the failure detector. |
 | `datanode` | -- | -- | Datanode options. |
 | `datanode.client` | -- | -- | Datanode client options. |
-| `datanode.client.timeout` | String | `10s` | -- |
-| `datanode.client.connect_timeout` | String | `10s` | -- |
-| `datanode.client.tcp_nodelay` | Bool | `true` | -- |
+| `datanode.client.timeout` | String | `10s` | Operation timeout. |
+| `datanode.client.connect_timeout` | String | `10s` | Connect server timeout. |
+| `datanode.client.tcp_nodelay` | Bool | `true` | `TCP_NODELAY` option for accepted connections. |
 | `wal` | -- | -- | -- |
 | `wal.provider` | String | `raft_engine` | -- |
 | `wal.broker_endpoints` | Array | -- | The broker endpoints of the Kafka cluster. |
@@ -282,7 +298,7 @@
 | `logging.dir` | String | `/tmp/greptimedb/logs` | The directory to store the log files. |
 | `logging.level` | String | `None` | The log level. Can be `info`/`debug`/`warn`/`error`. |
 | `logging.enable_otlp_tracing` | Bool | `false` | Enable OTLP tracing. |
-| `logging.otlp_endpoint` | String | `None` | The OTLP tracing endpoint. |
+| `logging.otlp_endpoint` | String | `http://localhost:4317` | The OTLP tracing endpoint. |
 | `logging.append_stdout` | Bool | `true` | Whether to append logs to stdout. |
 | `logging.tracing_sample_ratio` | -- | -- | The percentage of tracing will be sampled and exported.<br/>Valid range `[0, 1]`, 1 means all traces are sampled, 0 means all traces are not sampled, the default value is 1.<br/>ratio > 1 are treated as 1. Fractions < 0 are treated as 0 |
 | `logging.tracing_sample_ratio.default_ratio` | Float | `1.0` | -- |
@@ -325,9 +341,8 @@
 | `grpc.tls.key_path` | String | `None` | Private key file path. |
 | `grpc.tls.watch` | Bool | `false` | Watch for Certificate and key file change and auto reload.<br/>For now, gRPC tls config does not support auto reload. |
 | `runtime` | -- | -- | The runtime options. |
-| `runtime.read_rt_size` | Integer | `8` | The number of threads to execute the runtime for global read operations. |
-| `runtime.write_rt_size` | Integer | `8` | The number of threads to execute the runtime for global write operations. |
-| `runtime.bg_rt_size` | Integer | `4` | The number of threads to execute the runtime for global background operations. |
+| `runtime.global_rt_size` | Integer | `8` | The number of threads to execute the runtime for global read operations. |
+| `runtime.compact_rt_size` | Integer | `4` | The number of threads to execute the runtime for global write operations. |
 | `heartbeat` | -- | -- | The heartbeat options. |
 | `heartbeat.interval` | String | `3s` | Interval for sending heartbeat messages to the metasrv. |
 | `heartbeat.retry_interval` | String | `3s` | Interval for retrying to send heartbeat messages to the metasrv. |
@@ -359,6 +374,8 @@
 | `wal.backoff_max` | String | `10s` | The maximum backoff delay.<br/>**It's only used when the provider is `kafka`**. |
 | `wal.backoff_base` | Integer | `2` | The exponential backoff rate, i.e. next backoff = base * current backoff.<br/>**It's only used when the provider is `kafka`**. |
 | `wal.backoff_deadline` | String | `5mins` | The deadline of retries.<br/>**It's only used when the provider is `kafka`**. |
+| `wal.create_index` | Bool | `true` | Whether to enable WAL index creation.<br/>**It's only used when the provider is `kafka`**. |
+| `wal.dump_index_interval` | String | `60s` | The interval for dumping WAL indexes.<br/>**It's only used when the provider is `kafka`**. |
 | `storage` | -- | -- | The data storage options. |
 | `storage.data_home` | String | `/tmp/greptimedb/` | The working home directory. |
 | `storage.type` | String | `File` | The storage type used to store the data.<br/>- `File`: the data is stored in the local file system.<br/>- `S3`: the data is stored in the S3 object storage.<br/>- `Gcs`: the data is stored in the Google Cloud Storage.<br/>- `Azblob`: the data is stored in the Azure Blob Storage.<br/>- `Oss`: the data is stored in the Aliyun OSS. |
@@ -373,6 +390,7 @@
 | `storage.account_key` | String | `None` | The account key of the azure account.<br/>**It's only used when the storage type is `Azblob`**. |
 | `storage.scope` | String | `None` | The scope of the google cloud storage.<br/>**It's only used when the storage type is `Gcs`**. |
 | `storage.credential_path` | String | `None` | The credential path of the google cloud storage.<br/>**It's only used when the storage type is `Gcs`**. |
+| `storage.credential` | String | `None` | The credential of the google cloud storage.<br/>**It's only used when the storage type is `Gcs`**. |
 | `storage.container` | String | `None` | The container of the azure account.<br/>**It's only used when the storage type is `Azblob`**. |
 | `storage.sas_token` | String | `None` | The sas token of the azure account.<br/>**It's only used when the storage type is `Azblob`**. |
 | `storage.endpoint` | String | `None` | The endpoint of the S3 service.<br/>**It's only used when the storage type is `S3`, `Oss`, `Gcs` and `Azblob`**. |
@@ -399,22 +417,31 @@
 | `region_engine.mito.scan_parallelism` | Integer | `0` | Parallelism to scan a region (default: 1/4 of cpu cores).<br/>- `0`: using the default value (1/4 of cpu cores).<br/>- `1`: scan in current thread.<br/>- `n`: scan in parallelism n. |
 | `region_engine.mito.parallel_scan_channel_size` | Integer | `32` | Capacity of the channel to send data from parallel scan tasks to the main task. |
 | `region_engine.mito.allow_stale_entries` | Bool | `false` | Whether to allow stale WAL entries read during replay. |
+| `region_engine.mito.index` | -- | -- | The options for index in Mito engine. |
+| `region_engine.mito.index.aux_path` | String | `""` | Auxiliary directory path for the index in filesystem, used to store intermediate files for<br/>creating the index and staging files for searching the index, defaults to `{data_home}/index_intermediate`.<br/>The default name for this directory is `index_intermediate` for backward compatibility.<br/><br/>This path contains two subdirectories:<br/>- `__intm`: for storing intermediate files used during creating index.<br/>- `staging`: for storing staging files used during searching index. |
+| `region_engine.mito.index.staging_size` | String | `2GB` | The max capacity of the staging directory. |
 | `region_engine.mito.inverted_index` | -- | -- | The options for inverted index in Mito engine. |
-| `region_engine.mito.inverted_index.create_on_flush` | String | `auto` | Whether to create the index on flush.<br/>- `auto`: automatically<br/>- `disable`: never |
-| `region_engine.mito.inverted_index.create_on_compaction` | String | `auto` | Whether to create the index on compaction.<br/>- `auto`: automatically<br/>- `disable`: never |
-| `region_engine.mito.inverted_index.apply_on_query` | String | `auto` | Whether to apply the index on query<br/>- `auto`: automatically<br/>- `disable`: never |
-| `region_engine.mito.inverted_index.mem_threshold_on_create` | String | `64M` | Memory threshold for performing an external sort during index creation.<br/>Setting to empty will disable external sorting, forcing all sorting operations to happen in memory. |
-| `region_engine.mito.inverted_index.intermediate_path` | String | `""` | File system path to store intermediate files for external sorting (default `{data_home}/index_intermediate`). |
+| `region_engine.mito.inverted_index.create_on_flush` | String | `auto` | Whether to create the index on flush.<br/>- `auto`: automatically (default)<br/>- `disable`: never |
+| `region_engine.mito.inverted_index.create_on_compaction` | String | `auto` | Whether to create the index on compaction.<br/>- `auto`: automatically (default)<br/>- `disable`: never |
+| `region_engine.mito.inverted_index.apply_on_query` | String | `auto` | Whether to apply the index on query<br/>- `auto`: automatically (default)<br/>- `disable`: never |
+| `region_engine.mito.inverted_index.mem_threshold_on_create` | String | `auto` | Memory threshold for performing an external sort during index creation.<br/>- `auto`: automatically determine the threshold based on the system memory size (default)<br/>- `unlimited`: no memory limit<br/>- `[size]` e.g. `64MB`: fixed memory threshold |
+| `region_engine.mito.inverted_index.intermediate_path` | String | `""` | Deprecated, use `region_engine.mito.index.aux_path` instead. |
+| `region_engine.mito.fulltext_index` | -- | -- | The options for full-text index in Mito engine. |
+| `region_engine.mito.fulltext_index.create_on_flush` | String | `auto` | Whether to create the index on flush.<br/>- `auto`: automatically (default)<br/>- `disable`: never |
+| `region_engine.mito.fulltext_index.create_on_compaction` | String | `auto` | Whether to create the index on compaction.<br/>- `auto`: automatically (default)<br/>- `disable`: never |
+| `region_engine.mito.fulltext_index.apply_on_query` | String | `auto` | Whether to apply the index on query<br/>- `auto`: automatically (default)<br/>- `disable`: never |
+| `region_engine.mito.fulltext_index.mem_threshold_on_create` | String | `auto` | Memory threshold for index creation.<br/>- `auto`: automatically determine the threshold based on the system memory size (default)<br/>- `unlimited`: no memory limit<br/>- `[size]` e.g. `64MB`: fixed memory threshold |
 | `region_engine.mito.memtable` | -- | -- | -- |
 | `region_engine.mito.memtable.type` | String | `time_series` | Memtable type.<br/>- `time_series`: time-series memtable<br/>- `partition_tree`: partition tree memtable (experimental) |
 | `region_engine.mito.memtable.index_max_keys_per_shard` | Integer | `8192` | The max number of keys in one shard.<br/>Only available for `partition_tree` memtable. |
 | `region_engine.mito.memtable.data_freeze_threshold` | Integer | `32768` | The max rows of data inside the actively writing buffer in one shard.<br/>Only available for `partition_tree` memtable. |
 | `region_engine.mito.memtable.fork_dictionary_bytes` | String | `1GiB` | Max dictionary bytes.<br/>Only available for `partition_tree` memtable. |
+| `region_engine.file` | -- | -- | Enable the file engine. |
 | `logging` | -- | -- | The logging options. |
 | `logging.dir` | String | `/tmp/greptimedb/logs` | The directory to store the log files. |
 | `logging.level` | String | `None` | The log level. Can be `info`/`debug`/`warn`/`error`. |
 | `logging.enable_otlp_tracing` | Bool | `false` | Enable OTLP tracing. |
-| `logging.otlp_endpoint` | String | `None` | The OTLP tracing endpoint. |
+| `logging.otlp_endpoint` | String | `http://localhost:4317` | The OTLP tracing endpoint. |
 | `logging.append_stdout` | Bool | `true` | Whether to append logs to stdout. |
 | `logging.tracing_sample_ratio` | -- | -- | The percentage of tracing will be sampled and exported.<br/>Valid range `[0, 1]`, 1 means all traces are sampled, 0 means all traces are not sampled, the default value is 1.<br/>ratio > 1 are treated as 1. Fractions < 0 are treated as 0 |
 | `logging.tracing_sample_ratio.default_ratio` | Float | `1.0` | -- |
@@ -428,3 +455,40 @@
 | `export_metrics.remote_write.headers` | InlineTable | -- | HTTP headers of Prometheus remote-write carry. |
 | `tracing` | -- | -- | The tracing options. Only effect when compiled with `tokio-console` feature. |
 | `tracing.tokio_console_addr` | String | `None` | The tokio console address. |
+
+
+### Flownode
+
+| Key | Type | Default | Descriptions |
+| --- | -----| ------- | ----------- |
+| `mode` | String | `distributed` | The running mode of the flownode. It can be `standalone` or `distributed`. |
+| `node_id` | Integer | `None` | The flownode identifier and should be unique in the cluster. |
+| `grpc` | -- | -- | The gRPC server options. |
+| `grpc.addr` | String | `127.0.0.1:6800` | The address to bind the gRPC server. |
+| `grpc.hostname` | String | `127.0.0.1` | The hostname advertised to the metasrv,<br/>and used for connections from outside the host |
+| `grpc.runtime_size` | Integer | `2` | The number of server worker threads. |
+| `grpc.max_recv_message_size` | String | `512MB` | The maximum receive message size for gRPC server. |
+| `grpc.max_send_message_size` | String | `512MB` | The maximum send message size for gRPC server. |
+| `meta_client` | -- | -- | The metasrv client options. |
+| `meta_client.metasrv_addrs` | Array | -- | The addresses of the metasrv. |
+| `meta_client.timeout` | String | `3s` | Operation timeout. |
+| `meta_client.heartbeat_timeout` | String | `500ms` | Heartbeat timeout. |
+| `meta_client.ddl_timeout` | String | `10s` | DDL timeout. |
+| `meta_client.connect_timeout` | String | `1s` | Connect server timeout. |
+| `meta_client.tcp_nodelay` | Bool | `true` | `TCP_NODELAY` option for accepted connections. |
+| `meta_client.metadata_cache_max_capacity` | Integer | `100000` | The configuration about the cache of the metadata. |
+| `meta_client.metadata_cache_ttl` | String | `10m` | TTL of the metadata cache. |
+| `meta_client.metadata_cache_tti` | String | `5m` | -- |
+| `heartbeat` | -- | -- | The heartbeat options. |
+| `heartbeat.interval` | String | `3s` | Interval for sending heartbeat messages to the metasrv. |
+| `heartbeat.retry_interval` | String | `3s` | Interval for retrying to send heartbeat messages to the metasrv. |
+| `logging` | -- | -- | The logging options. |
+| `logging.dir` | String | `/tmp/greptimedb/logs` | The directory to store the log files. |
+| `logging.level` | String | `None` | The log level. Can be `info`/`debug`/`warn`/`error`. |
+| `logging.enable_otlp_tracing` | Bool | `false` | Enable OTLP tracing. |
+| `logging.otlp_endpoint` | String | `http://localhost:4317` | The OTLP tracing endpoint. |
+| `logging.append_stdout` | Bool | `true` | Whether to append logs to stdout. |
+| `logging.tracing_sample_ratio` | -- | -- | The percentage of tracing will be sampled and exported.<br/>Valid range `[0, 1]`, 1 means all traces are sampled, 0 means all traces are not sampled, the default value is 1.<br/>ratio > 1 are treated as 1. Fractions < 0 are treated as 0 |
+| `logging.tracing_sample_ratio.default_ratio` | Float | `1.0` | -- |
+| `tracing` | -- | -- | The tracing options. Only effect when compiled with `tokio-console` feature. |
+| `tracing.tokio_console_addr` | String | `None` | The tokio console address. |

config/datanode.example.toml

@@ -73,11 +73,9 @@ watch = false
 ## The runtime options.
 [runtime]
 ## The number of threads to execute the runtime for global read operations.
-read_rt_size = 8
+global_rt_size = 8
 ## The number of threads to execute the runtime for global write operations.
-write_rt_size = 8
-## The number of threads to execute the runtime for global background operations.
-bg_rt_size = 4
+compact_rt_size = 4
 
 ## The heartbeat options.
 [heartbeat]
@@ -189,6 +187,32 @@ backoff_base = 2
 ## **It's only used when the provider is `kafka`**.
 backoff_deadline = "5mins"
 
+## Whether to enable WAL index creation.
+## **It's only used when the provider is `kafka`**.
+create_index = true
+
+## The interval for dumping WAL indexes.
+## **It's only used when the provider is `kafka`**.
+dump_index_interval = "60s"
+
+# The Kafka SASL configuration.
+# **It's only used when the provider is `kafka`**.
+# Available SASL mechanisms:
+# - `PLAIN`
+# - `SCRAM-SHA-256`
+# - `SCRAM-SHA-512`
+# [wal.sasl]
+# type = "SCRAM-SHA-512"
+# username = "user_kafka"
+# password = "secret"
+
+# The Kafka TLS configuration.
+# **It's only used when the provider is `kafka`**.
+# [wal.tls]
+# server_ca_cert_path = "/path/to/server_cert"
+# client_cert_path = "/path/to/client_cert"
+# client_key_path = "/path/to/key"
+
 # Example of using S3 as the storage.
 # [storage]
 # type = "S3"
@@ -225,6 +249,7 @@ backoff_deadline = "5mins"
 # root = "data"
 # scope = "test"
 # credential_path = "123456"
+# credential = "base64-credential"
 # endpoint = "https://storage.googleapis.com"
 
 ## The data storage options.
@@ -296,6 +321,11 @@ scope = "test"
 ## +toml2docs:none-default
 credential_path = "test"
 
+## The credential of the google cloud storage.
+## **It's only used when the storage type is `Gcs`**.
+## +toml2docs:none-default
+credential= "base64-credential"
+
 ## The container of the azure account.
 ## **It's only used when the storage type is `Azblob`**.
 ## +toml2docs:none-default
@@ -394,31 +424,72 @@ parallel_scan_channel_size = 32
 ## Whether to allow stale WAL entries read during replay.
 allow_stale_entries = false
 
+## The options for index in Mito engine.
+[region_engine.mito.index]
+
+## Auxiliary directory path for the index in filesystem, used to store intermediate files for
+## creating the index and staging files for searching the index, defaults to `{data_home}/index_intermediate`.
+## The default name for this directory is `index_intermediate` for backward compatibility.
+##
+## This path contains two subdirectories:
+## - `__intm`: for storing intermediate files used during creating index.
+## - `staging`: for storing staging files used during searching index.
+aux_path = ""
+
+## The max capacity of the staging directory.
+staging_size = "2GB"
+
 ## The options for inverted index in Mito engine.
 [region_engine.mito.inverted_index]
 
 ## Whether to create the index on flush.
-## - `auto`: automatically
+## - `auto`: automatically (default)
 ## - `disable`: never
 create_on_flush = "auto"
 
 ## Whether to create the index on compaction.
-## - `auto`: automatically
+## - `auto`: automatically (default)
 ## - `disable`: never
 create_on_compaction = "auto"
 
 ## Whether to apply the index on query
-## - `auto`: automatically
+## - `auto`: automatically (default)
 ## - `disable`: never
 apply_on_query = "auto"
 
 ## Memory threshold for performing an external sort during index creation.
-## Setting to empty will disable external sorting, forcing all sorting operations to happen in memory.
-mem_threshold_on_create = "64M"
+## - `auto`: automatically determine the threshold based on the system memory size (default)
+## - `unlimited`: no memory limit
+## - `[size]` e.g. `64MB`: fixed memory threshold
+mem_threshold_on_create = "auto"
 
-## File system path to store intermediate files for external sorting (default `{data_home}/index_intermediate`).
+## Deprecated, use `region_engine.mito.index.aux_path` instead.
 intermediate_path = ""
 
+## The options for full-text index in Mito engine.
+[region_engine.mito.fulltext_index]
+
+## Whether to create the index on flush.
+## - `auto`: automatically (default)
+## - `disable`: never
+create_on_flush = "auto"
+
+## Whether to create the index on compaction.
+## - `auto`: automatically (default)
+## - `disable`: never
+create_on_compaction = "auto"
+
+## Whether to apply the index on query
+## - `auto`: automatically (default)
+## - `disable`: never
+apply_on_query = "auto"
+
+## Memory threshold for index creation.
+## - `auto`: automatically determine the threshold based on the system memory size (default)
+## - `unlimited`: no memory limit
+## - `[size]` e.g. `64MB`: fixed memory threshold
+mem_threshold_on_create = "auto"
+
 [region_engine.mito.memtable]
 ## Memtable type.
 ## - `time_series`: time-series memtable
@@ -437,6 +508,10 @@ data_freeze_threshold = 32768
 ## Only available for `partition_tree` memtable.
 fork_dictionary_bytes = "1GiB"
 
+[[region_engine]]
+## Enable the file engine.
+[region_engine.file]
+
 ## The logging options.
 [logging]
 ## The directory to store the log files.
@@ -450,8 +525,7 @@ level = "info"
 enable_otlp_tracing = false
 
 ## The OTLP tracing endpoint.
-## +toml2docs:none-default
-otlp_endpoint = ""
+otlp_endpoint = "http://localhost:4317"
 
 ## Whether to append logs to stdout.
 append_stdout = true

config/flownode.example.toml

@@ -0,0 +1,89 @@
+## The running mode of the flownode. It can be `standalone` or `distributed`.
+mode = "distributed"
+
+## The flownode identifier and should be unique in the cluster.
+## +toml2docs:none-default
+node_id = 14
+
+## The gRPC server options.
+[grpc]
+## The address to bind the gRPC server.
+addr = "127.0.0.1:6800"
+## The hostname advertised to the metasrv,
+## and used for connections from outside the host
+hostname = "127.0.0.1"
+## The number of server worker threads.
+runtime_size = 2
+## The maximum receive message size for gRPC server.
+max_recv_message_size = "512MB"
+## The maximum send message size for gRPC server.
+max_send_message_size = "512MB"
+
+
+## The metasrv client options.
+[meta_client]
+## The addresses of the metasrv.
+metasrv_addrs = ["127.0.0.1:3002"]
+
+## Operation timeout.
+timeout = "3s"
+
+## Heartbeat timeout.
+heartbeat_timeout = "500ms"
+
+## DDL timeout.
+ddl_timeout = "10s"
+
+## Connect server timeout.
+connect_timeout = "1s"
+
+## `TCP_NODELAY` option for accepted connections.
+tcp_nodelay = true
+
+## The configuration about the cache of the metadata.
+metadata_cache_max_capacity = 100000
+
+## TTL of the metadata cache.
+metadata_cache_ttl = "10m"
+
+# TTI of the metadata cache.
+metadata_cache_tti = "5m"
+
+## The heartbeat options.
+[heartbeat]
+## Interval for sending heartbeat messages to the metasrv.
+interval = "3s"
+
+## Interval for retrying to send heartbeat messages to the metasrv.
+retry_interval = "3s"
+
+## The logging options.
+[logging]
+## The directory to store the log files.
+dir = "/tmp/greptimedb/logs"
+
+## The log level. Can be `info`/`debug`/`warn`/`error`.
+## +toml2docs:none-default
+level = "info"
+
+## Enable OTLP tracing.
+enable_otlp_tracing = false
+
+## The OTLP tracing endpoint.
+otlp_endpoint = "http://localhost:4317"
+
+## Whether to append logs to stdout.
+append_stdout = true
+
+## 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
+[logging.tracing_sample_ratio]
+default_ratio = 1.0
+
+## The tracing options. Only effect when compiled with `tokio-console` feature.
+[tracing]
+## The tokio console address.
+## +toml2docs:none-default
+tokio_console_addr = "127.0.0.1"
+

config/frontend.example.toml

@@ -1,6 +1,3 @@
-## The running mode of the datanode. It can be `standalone` or `distributed`.
-mode = "standalone"
-
 ## The default timezone of the server.
 ## +toml2docs:none-default
 default_timezone = "UTC"
@@ -8,11 +5,9 @@ default_timezone = "UTC"
 ## The runtime options.
 [runtime]
 ## The number of threads to execute the runtime for global read operations.
-read_rt_size = 8
+global_rt_size = 8
 ## The number of threads to execute the runtime for global write operations.
-write_rt_size = 8
-## The number of threads to execute the runtime for global background operations.
-bg_rt_size = 4
+compact_rt_size = 4
 
 ## The heartbeat options.
 [heartbeat]
@@ -182,8 +177,7 @@ level = "info"
 enable_otlp_tracing = false
 
 ## The OTLP tracing endpoint.
-## +toml2docs:none-default
-otlp_endpoint = ""
+otlp_endpoint = "http://localhost:4317"
 
 ## Whether to append logs to stdout.
 append_stdout = true

config/metasrv.example.toml

@@ -11,10 +11,11 @@ server_addr = "127.0.0.1:3002"
 store_addr = "127.0.0.1:2379"
 
 ## Datanode selector type.
-## - `lease_based` (default value).
+## - `round_robin` (default value)
+## - `lease_based`
 ## - `load_based`
 ## For details, please see "https://docs.greptime.com/developer-guide/metasrv/selector".
-selector = "lease_based"
+selector = "round_robin"
 
 ## Store data in memory.
 use_memory_store = false
@@ -25,14 +26,18 @@ enable_telemetry = true
 ## If it's not empty, the metasrv will store all data with this key prefix.
 store_key_prefix = ""
 
+## Whether to enable region failover.
+## This feature is only available on GreptimeDB running on cluster mode and
+## - Using Remote WAL
+## - Using shared storage (e.g., s3).
+enable_region_failover = false
+
 ## The runtime options.
 [runtime]
 ## The number of threads to execute the runtime for global read operations.
-read_rt_size = 8
+global_rt_size = 8
 ## The number of threads to execute the runtime for global write operations.
-write_rt_size = 8
-## The number of threads to execute the runtime for global background operations.
-bg_rt_size = 4
+compact_rt_size = 4
 
 ## Procedure storage options.
 [procedure]
@@ -52,17 +57,32 @@ max_metadata_value_size = "1500KiB"
 
 # Failure detectors options.
 [failure_detector]
+
+## The threshold value used by the failure detector to determine failure conditions.
 threshold = 8.0
+
+## The minimum standard deviation of the heartbeat intervals, used to calculate acceptable variations.
 min_std_deviation = "100ms"
-acceptable_heartbeat_pause = "3000ms"
+
+## The acceptable pause duration between heartbeats, used to determine if a heartbeat interval is acceptable.
+acceptable_heartbeat_pause = "10000ms"
+
+## The initial estimate of the heartbeat interval used by the failure detector.
 first_heartbeat_estimate = "1000ms"
 
 ## Datanode options.
 [datanode]
+
 ## Datanode client options.
 [datanode.client]
+
+## Operation timeout.
 timeout = "10s"
+
+## Connect server timeout.
 connect_timeout = "10s"
+
+## `TCP_NODELAY` option for accepted connections.
 tcp_nodelay = true
 
 [wal]
@@ -104,6 +124,24 @@ 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"
 
+# The Kafka SASL configuration.
+# **It's only used when the provider is `kafka`**.
+# Available SASL mechanisms:
+# - `PLAIN`
+# - `SCRAM-SHA-256`
+# - `SCRAM-SHA-512`
+# [wal.sasl]
+# type = "SCRAM-SHA-512"
+# username = "user_kafka"
+# password = "secret"
+
+# The Kafka TLS configuration.
+# **It's only used when the provider is `kafka`**.
+# [wal.tls]
+# server_ca_cert_path = "/path/to/server_cert"
+# client_cert_path = "/path/to/client_cert"
+# client_key_path = "/path/to/key"
+
 ## The logging options.
 [logging]
 ## The directory to store the log files.
@@ -117,8 +155,7 @@ level = "info"
 enable_otlp_tracing = false
 
 ## The OTLP tracing endpoint.
-## +toml2docs:none-default
-otlp_endpoint = ""
+otlp_endpoint = "http://localhost:4317"
 
 ## Whether to append logs to stdout.
 append_stdout = true

config/standalone.example.toml

@@ -11,11 +11,9 @@ default_timezone = "UTC"
 ## The runtime options.
 [runtime]
 ## The number of threads to execute the runtime for global read operations.
-read_rt_size = 8
+global_rt_size = 8
 ## The number of threads to execute the runtime for global write operations.
-write_rt_size = 8
-## The number of threads to execute the runtime for global background operations.
-bg_rt_size = 4
+compact_rt_size = 4
 
 ## The HTTP server options.
 [http]
@@ -173,6 +171,28 @@ sync_period = "10s"
 ## **It's only used when the provider is `kafka`**.
 broker_endpoints = ["127.0.0.1:9092"]
 
+## Number of topics to be created upon start.
+## **It's only used when the provider is `kafka`**.
+num_topics = 64
+
+## Topic selector type.
+## Available selector types:
+## - `round_robin` (default)
+## **It's only used when the provider is `kafka`**.
+selector_type = "round_robin"
+
+## A Kafka topic is constructed by concatenating `topic_name_prefix` and `topic_id`.
+## **It's only used when the provider is `kafka`**.
+topic_name_prefix = "greptimedb_wal_topic"
+
+## Expected number of replicas of each partition.
+## **It's only used when the provider is `kafka`**.
+replication_factor = 1
+
+## Above which a topic creation operation will be cancelled.
+## **It's only used when the provider is `kafka`**.
+create_topic_timeout = "30s"
+
 ## The max size of a single producer batch.
 ## Warning: Kafka has a default limit of 1MB per message in a topic.
 ## **It's only used when the provider is `kafka`**.
@@ -198,6 +218,24 @@ backoff_base = 2
 ## **It's only used when the provider is `kafka`**.
 backoff_deadline = "5mins"
 
+# The Kafka SASL configuration.
+# **It's only used when the provider is `kafka`**.
+# Available SASL mechanisms:
+# - `PLAIN`
+# - `SCRAM-SHA-256`
+# - `SCRAM-SHA-512`
+# [wal.sasl]
+# type = "SCRAM-SHA-512"
+# username = "user_kafka"
+# password = "secret"
+
+# The Kafka TLS configuration.
+# **It's only used when the provider is `kafka`**.
+# [wal.tls]
+# server_ca_cert_path = "/path/to/server_cert"
+# client_cert_path = "/path/to/client_cert"
+# client_key_path = "/path/to/key"
+
 ## Metadata storage options.
 [metadata_store]
 ## Kv file size in bytes.
@@ -248,6 +286,7 @@ retry_delay = "500ms"
 # root = "data"
 # scope = "test"
 # credential_path = "123456"
+# credential = "base64-credential"
 # endpoint = "https://storage.googleapis.com"
 
 ## The data storage options.
@@ -319,6 +358,11 @@ scope = "test"
 ## +toml2docs:none-default
 credential_path = "test"
 
+## The credential of the google cloud storage.
+## **It's only used when the storage type is `Gcs`**.
+## +toml2docs:none-default
+credential = "base64-credential"
+
 ## The container of the azure account.
 ## **It's only used when the storage type is `Azblob`**.
 ## +toml2docs:none-default
@@ -417,31 +461,78 @@ parallel_scan_channel_size = 32
 ## Whether to allow stale WAL entries read during replay.
 allow_stale_entries = false
 
+## The options for index in Mito engine.
+[region_engine.mito.index]
+
+## Auxiliary directory path for the index in filesystem, used to store intermediate files for
+## creating the index and staging files for searching the index, defaults to `{data_home}/index_intermediate`.
+## The default name for this directory is `index_intermediate` for backward compatibility.
+##
+## This path contains two subdirectories:
+## - `__intm`: for storing intermediate files used during creating index.
+## - `staging`: for storing staging files used during searching index.
+aux_path = ""
+
+## The max capacity of the staging directory.
+staging_size = "2GB"
+
 ## The options for inverted index in Mito engine.
 [region_engine.mito.inverted_index]
 
 ## Whether to create the index on flush.
-## - `auto`: automatically
+## - `auto`: automatically (default)
 ## - `disable`: never
 create_on_flush = "auto"
 
 ## Whether to create the index on compaction.
-## - `auto`: automatically
+## - `auto`: automatically (default)
 ## - `disable`: never
 create_on_compaction = "auto"
 
 ## Whether to apply the index on query
-## - `auto`: automatically
+## - `auto`: automatically (default)
 ## - `disable`: never
 apply_on_query = "auto"
 
 ## Memory threshold for performing an external sort during index creation.
-## Setting to empty will disable external sorting, forcing all sorting operations to happen in memory.
-mem_threshold_on_create = "64M"
+## - `auto`: automatically determine the threshold based on the system memory size (default)
+## - `unlimited`: no memory limit
+## - `[size]` e.g. `64MB`: fixed memory threshold
+mem_threshold_on_create = "auto"
 
-## File system path to store intermediate files for external sorting (default `{data_home}/index_intermediate`).
+## Deprecated, use `region_engine.mito.index.aux_path` instead.
 intermediate_path = ""
 
+## Cache size for inverted index metadata.
+metadata_cache_size = "64MiB"
+
+## Cache size for inverted index content.
+content_cache_size = "128MiB"
+
+## The options for full-text index in Mito engine.
+[region_engine.mito.fulltext_index]
+
+## Whether to create the index on flush.
+## - `auto`: automatically (default)
+## - `disable`: never
+create_on_flush = "auto"
+
+## Whether to create the index on compaction.
+## - `auto`: automatically (default)
+## - `disable`: never
+create_on_compaction = "auto"
+
+## Whether to apply the index on query
+## - `auto`: automatically (default)
+## - `disable`: never
+apply_on_query = "auto"
+
+## Memory threshold for index creation.
+## - `auto`: automatically determine the threshold based on the system memory size (default)
+## - `unlimited`: no memory limit
+## - `[size]` e.g. `64MB`: fixed memory threshold
+mem_threshold_on_create = "auto"
+
 [region_engine.mito.memtable]
 ## Memtable type.
 ## - `time_series`: time-series memtable
@@ -460,6 +551,10 @@ data_freeze_threshold = 32768
 ## Only available for `partition_tree` memtable.
 fork_dictionary_bytes = "1GiB"
 
+[[region_engine]]
+## Enable the file engine.
+[region_engine.file]
+
 ## The logging options.
 [logging]
 ## The directory to store the log files.
@@ -473,8 +568,7 @@ level = "info"
 enable_otlp_tracing = false
 
 ## The OTLP tracing endpoint.
-## +toml2docs:none-default
-otlp_endpoint = ""
+otlp_endpoint = "http://localhost:4317"
 
 ## Whether to append logs to stdout.
 append_stdout = true

docker/ci/centos/Dockerfile

@@ -1,5 +1,9 @@
 FROM centos:7
 
+# Note: CentOS 7 has reached EOL since 2024-07-01 thus `mirror.centos.org` is no longer available and we need to use `vault.centos.org` instead.
+RUN sed -i s/mirror.centos.org/vault.centos.org/g /etc/yum.repos.d/*.repo
+RUN sed -i s/^#.*baseurl=http/baseurl=http/g /etc/yum.repos.d/*.repo
+
 RUN yum install -y epel-release \
     openssl \
     openssl-devel  \

docker/dev-builder/centos/Dockerfile

@@ -2,6 +2,10 @@ FROM centos:7 as builder
 
 ENV LANG en_US.utf8
 
+# Note: CentOS 7 has reached EOL since 2024-07-01 thus `mirror.centos.org` is no longer available and we need to use `vault.centos.org` instead.
+RUN sed -i s/mirror.centos.org/vault.centos.org/g /etc/yum.repos.d/*.repo
+RUN sed -i s/^#.*baseurl=http/baseurl=http/g /etc/yum.repos.d/*.repo
+
 # Install dependencies
 RUN ulimit -n 1024000 && yum groupinstall -y 'Development Tools'
 RUN yum install -y epel-release  \
@@ -25,6 +29,10 @@ ENV PATH /opt/rh/rh-python38/root/usr/bin:/usr/local/bin:/root/.cargo/bin/:$PATH
 ARG RUST_TOOLCHAIN
 RUN rustup toolchain install ${RUST_TOOLCHAIN}
 
+
+# Install cargo-binstall with a specific version to adapt the current rust toolchain.
+# Note: if we use the latest version, we may encounter the following `use of unstable library feature 'io_error_downcast'` error.
+RUN cargo install cargo-binstall --version 1.6.6 --locked
+
 # Install nextest.
-RUN cargo install cargo-binstall --locked
 RUN cargo binstall cargo-nextest --no-confirm

docker/dev-builder/ubuntu/Dockerfile

@@ -55,6 +55,9 @@ ENV PATH /root/.cargo/bin/:$PATH
 ARG RUST_TOOLCHAIN
 RUN rustup toolchain install ${RUST_TOOLCHAIN}
 
+# Install cargo-binstall with a specific version to adapt the current rust toolchain.
+# Note: if we use the latest version, we may encounter the following `use of unstable library feature 'io_error_downcast'` error.
+RUN cargo install cargo-binstall --version 1.6.6 --locked
+
 # Install nextest.
-RUN cargo install cargo-binstall --locked
 RUN cargo binstall cargo-nextest --no-confirm

docker/dev-builder/ubuntu/Dockerfile-18.10

@@ -43,6 +43,9 @@ ENV PATH /root/.cargo/bin/:$PATH
 ARG RUST_TOOLCHAIN
 RUN rustup toolchain install ${RUST_TOOLCHAIN}
 
+# Install cargo-binstall with a specific version to adapt the current rust toolchain.
+# Note: if we use the latest version, we may encounter the following `use of unstable library feature 'io_error_downcast'` error.
+RUN cargo install cargo-binstall --version 1.6.6 --locked
+
 # Install nextest.
-RUN cargo install cargo-binstall --locked
 RUN cargo binstall cargo-nextest --no-confirm

docker/docker-compose/cluster-with-etcd.yaml

@@ -1,12 +1,13 @@
 x-custom:
-  initial_cluster_token: &initial_cluster_token "--initial-cluster-token=etcd-cluster"
-  common_settings: &common_settings
-    image: quay.io/coreos/etcd:v3.5.10
+  etcd_initial_cluster_token: &etcd_initial_cluster_token "--initial-cluster-token=etcd-cluster"
+  etcd_common_settings: &etcd_common_settings
+    image: "${ETCD_REGISTRY:-quay.io}/${ETCD_NAMESPACE:-coreos}/etcd:${ETCD_VERSION:-v3.5.10}"
     entrypoint: /usr/local/bin/etcd
+  greptimedb_image: &greptimedb_image "${GREPTIMEDB_REGISTRY:-docker.io}/${GREPTIMEDB_NAMESPACE:-greptime}/greptimedb:${GREPTIMEDB_VERSION:-latest}"
 
 services:
   etcd0:
-    <<: *common_settings
+    <<: *etcd_common_settings
     container_name: etcd0
     ports:
       - 2379:2379
@@ -22,7 +23,7 @@ services:
       - --election-timeout=1250
       - --initial-cluster=etcd0=http://etcd0:2380
       - --initial-cluster-state=new
-      - *initial_cluster_token
+      - *etcd_initial_cluster_token
     volumes:
       - /tmp/greptimedb-cluster-docker-compose/etcd0:/var/lib/etcd
     healthcheck:
@@ -34,7 +35,7 @@ services:
       - greptimedb
 
   metasrv:
-    image: docker.io/greptime/greptimedb:latest
+    image: *greptimedb_image
     container_name: metasrv
     ports:
       - 3002:3002
@@ -56,19 +57,26 @@ services:
       - greptimedb
 
   datanode0:
-    image: docker.io/greptime/greptimedb:latest
+    image: *greptimedb_image
     container_name: datanode0
     ports:
       - 3001:3001
+      - 5000:5000
     command:
       - datanode
       - start
       - --node-id=0
       - --rpc-addr=0.0.0.0:3001
       - --rpc-hostname=datanode0:3001
-      - --metasrv-addr=metasrv:3002
+      - --metasrv-addrs=metasrv:3002
+      - --http-addr=0.0.0.0:5000
     volumes:
       - /tmp/greptimedb-cluster-docker-compose/datanode0:/tmp/greptimedb
+    healthcheck:
+      test: [ "CMD", "curl", "-f", "http://datanode0:5000/health" ]
+      interval: 5s
+      timeout: 3s
+      retries: 5
     depends_on:
       metasrv:
         condition: service_healthy
@@ -76,7 +84,7 @@ services:
       - greptimedb
 
   frontend0:
-    image: docker.io/greptime/greptimedb:latest
+    image: *greptimedb_image
     container_name: frontend0
     ports:
       - 4000:4000
@@ -91,8 +99,31 @@ services:
       - --rpc-addr=0.0.0.0:4001
       - --mysql-addr=0.0.0.0:4002
       - --postgres-addr=0.0.0.0:4003
+    healthcheck:
+      test: [ "CMD", "curl", "-f", "http://frontend0:4000/health" ]
+      interval: 5s
+      timeout: 3s
+      retries: 5
     depends_on:
-      metasrv:
+      datanode0:
+        condition: service_healthy
+    networks:
+      - greptimedb
+
+  flownode0:
+    image: *greptimedb_image
+    container_name: flownode0
+    ports:
+      - 4004:4004
+    command:
+      - flownode
+      - start
+      - --node-id=0
+      - --metasrv-addrs=metasrv:3002
+      - --rpc-addr=0.0.0.0:4004
+      - --rpc-hostname=flownode0:4004
+    depends_on:
+      frontend0:
         condition: service_healthy
     networks:
       - greptimedb

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

@@ -105,7 +105,7 @@ use tests_fuzz::utils::{init_greptime_connections, Connections};
 
 fuzz_target!(|input: FuzzInput| {
     common_telemetry::init_default_ut_logging();
-    common_runtime::block_on_write(async {
+    common_runtime::block_on_global(async {
         let Connections { mysql } = init_greptime_connections().await;
             let mut rng = ChaChaRng::seed_from_u64(input.seed);
             let columns = rng.gen_range(2..30);

grafana/README.md

@@ -25,7 +25,7 @@ Please ensure the following configuration before importing the dashboard into Gr
 
 __1. Prometheus scrape config__
 
-Assign `greptime_pod` label to each host target. We use this label to identify each node instance.
+Configure Prometheus to scrape the cluster.
 
 ```yml
 # example config
@@ -34,27 +34,15 @@ Assign `greptime_pod` label to each host target. We use this label to identify e
 scrape_configs:
   - job_name: metasrv
     static_configs:
-    - targets: ['<ip>:<port>']
-      labels:
-        greptime_pod: metasrv
+    - targets: ['<metasrv-ip>:<port>']
 
   - job_name: datanode
     static_configs:
-    - targets: ['<ip>:<port>']
-      labels:
-        greptime_pod: datanode1
-    - targets: ['<ip>:<port>']
-      labels:
-        greptime_pod: datanode2
-    - targets: ['<ip>:<port>']
-      labels:
-        greptime_pod: datanode3
+    - targets: ['<datanode0-ip>:<port>', '<datanode1-ip>:<port>', '<datanode2-ip>:<port>']
 
   - job_name: frontend
     static_configs:
-    - targets: ['<ip>:<port>']
-      labels:
-        greptime_pod: frontend
+    - targets: ['<frontend-ip>:<port>']
 ```
 
 __2. Grafana config__
@@ -63,4 +51,4 @@ Create a Prometheus data source in Grafana before using this dashboard. We use `
 
 ### Usage
 
-Use `datasource` or `greptime_pod` on the upper-left corner to filter data from certain node.
+Use `datasource` or `instance` on the upper-left corner to filter data from certain node.

scripts/install.sh

@@ -1,62 +1,72 @@
-#!/bin/sh
+#!/usr/bin/env bash
 
 set -ue
 
 OS_TYPE=
 ARCH_TYPE=
+
+# Set the GitHub token to avoid GitHub API rate limit.
+# You can run with `GITHUB_TOKEN`:
+#  GITHUB_TOKEN=<your_token> ./scripts/install.sh
+GITHUB_TOKEN=${GITHUB_TOKEN:-}
+
 VERSION=${1:-latest}
 GITHUB_ORG=GreptimeTeam
 GITHUB_REPO=greptimedb
 BIN=greptime
 
-get_os_type() {
-    os_type="$(uname -s)"
+function get_os_type() {
+  os_type="$(uname -s)"
 
-    case "$os_type" in
+  case "$os_type" in
     Darwin)
-        OS_TYPE=darwin
-        ;;
+      OS_TYPE=darwin
+      ;;
     Linux)
-        OS_TYPE=linux
-        ;;
+      OS_TYPE=linux
+      ;;
     *)
-        echo "Error: Unknown OS type: $os_type"
-        exit 1
-    esac
+      echo "Error: Unknown OS type: $os_type"
+      exit 1
+  esac
 }
 
-get_arch_type() {
-    arch_type="$(uname -m)"
+function get_arch_type() {
+  arch_type="$(uname -m)"
 
-    case "$arch_type" in
+  case "$arch_type" in
     arm64)
-        ARCH_TYPE=arm64
-        ;;
+      ARCH_TYPE=arm64
+      ;;
     aarch64)
-        ARCH_TYPE=arm64
-        ;;
+      ARCH_TYPE=arm64
+      ;;
     x86_64)
-        ARCH_TYPE=amd64
-        ;;
+      ARCH_TYPE=amd64
+      ;;
     amd64)
-        ARCH_TYPE=amd64
-        ;;
+      ARCH_TYPE=amd64
+      ;;
     *)
-        echo "Error: Unknown CPU type: $arch_type"
-        exit 1
-    esac
+      echo "Error: Unknown CPU type: $arch_type"
+      exit 1
+  esac
 }
 
-get_os_type
-get_arch_type
-
-if [ -n "${OS_TYPE}" ] && [ -n "${ARCH_TYPE}" ]; then
-    # Use the latest nightly version.
+function download_artifact() {
+  if [ -n "${OS_TYPE}" ] && [ -n "${ARCH_TYPE}" ]; then
+    # Use the latest stable released version.
+    # GitHub API reference: https://docs.github.com/en/rest/releases/releases?apiVersion=2022-11-28#get-the-latest-release.
     if [ "${VERSION}" = "latest" ]; then
-        VERSION=$(curl -s -XGET "https://api.github.com/repos/${GITHUB_ORG}/${GITHUB_REPO}/releases" | grep tag_name | grep nightly | cut -d: -f 2 | sed 's/.*"\(.*\)".*/\1/' | uniq | sort -r | head -n 1)
-        if [ -z "${VERSION}" ]; then
-            echo "Failed to get the latest version."
-            exit 1
+      # To avoid other tools dependency, we choose to use `curl` to get the version metadata and parsed by `sed`.
+      VERSION=$(curl -sL \
+        -H "Accept: application/vnd.github+json" \
+        -H "X-GitHub-Api-Version: 2022-11-28" \
+        ${GITHUB_TOKEN:+-H "Authorization: Bearer $GITHUB_TOKEN"} \
+        "https://api.github.com/repos/${GITHUB_ORG}/${GITHUB_REPO}/releases/latest" | sed -n 's/.*"tag_name": "\([^"]*\)".*/\1/p')
+      if [ -z "${VERSION}" ]; then
+        echo "Failed to get the latest stable released version."
+        exit 1
         fi
     fi
 
@@ -73,4 +83,9 @@ if [ -n "${OS_TYPE}" ] && [ -n "${ARCH_TYPE}" ]; then
       rm -r "${PACKAGE_NAME%.tar.gz}" && \
       echo "Run './${BIN} --help' to get started"
     fi
-fi
+  fi
+}
+
+get_os_type
+get_arch_type
+download_artifact

src/api/Cargo.toml

@@ -17,10 +17,11 @@ datatypes.workspace = true
 greptime-proto.workspace = true
 paste = "1.0"
 prost.workspace = true
+serde_json.workspace = true
 snafu.workspace = true
 
 [build-dependencies]
-tonic-build = "0.9"
+tonic-build = "0.11"
 
 [dev-dependencies]
 paste = "1.0"

src/api/src/error.rs

@@ -58,13 +58,23 @@ pub enum Error {
         location: Location,
         source: datatypes::error::Error,
     },
+
+    #[snafu(display("Failed to serialize JSON"))]
+    SerializeJson {
+        #[snafu(source)]
+        error: serde_json::Error,
+        #[snafu(implicit)]
+        location: Location,
+    },
 }
 
 impl ErrorExt for Error {
     fn status_code(&self) -> StatusCode {
         match self {
             Error::UnknownColumnDataType { .. } => StatusCode::InvalidArguments,
-            Error::IntoColumnDataType { .. } => StatusCode::Unexpected,
+            Error::IntoColumnDataType { .. } | Error::SerializeJson { .. } => {
+                StatusCode::Unexpected
+            }
             Error::ConvertColumnDefaultConstraint { source, .. }
             | Error::InvalidColumnDefaultConstraint { source, .. } => source.status_code(),
         }

src/api/src/helper.rs

@@ -1843,6 +1843,7 @@ mod tests {
             null_mask: vec![2],
             datatype: ColumnDataType::Boolean as i32,
             datatype_extension: None,
+            options: None,
         };
         assert!(is_column_type_value_eq(
             column1.datatype,

src/api/src/lib.rs

@@ -12,6 +12,8 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+#![feature(let_chains)]
+
 pub mod error;
 pub mod helper;
 

src/api/src/v1/column_def.rs

@@ -14,13 +14,19 @@
 
 use std::collections::HashMap;
 
-use datatypes::schema::{ColumnDefaultConstraint, ColumnSchema, COMMENT_KEY};
+use datatypes::schema::{
+    ColumnDefaultConstraint, ColumnSchema, FulltextOptions, COMMENT_KEY, FULLTEXT_KEY,
+};
 use snafu::ResultExt;
 
 use crate::error::{self, Result};
 use crate::helper::ColumnDataTypeWrapper;
-use crate::v1::ColumnDef;
+use crate::v1::{ColumnDef, ColumnOptions, SemanticType};
 
+/// Key used to store fulltext options in gRPC column options.
+const FULLTEXT_GRPC_KEY: &str = "fulltext";
+
+/// Tries to construct a `ColumnSchema` from the given  `ColumnDef`.
 pub fn try_as_column_schema(column_def: &ColumnDef) -> Result<ColumnSchema> {
     let data_type = ColumnDataTypeWrapper::try_new(
         column_def.data_type,
@@ -43,13 +49,147 @@ pub fn try_as_column_schema(column_def: &ColumnDef) -> Result<ColumnSchema> {
     if !column_def.comment.is_empty() {
         metadata.insert(COMMENT_KEY.to_string(), column_def.comment.clone());
     }
+    if let Some(options) = column_def.options.as_ref()
+        && let Some(fulltext) = options.options.get(FULLTEXT_GRPC_KEY)
+    {
+        metadata.insert(FULLTEXT_KEY.to_string(), fulltext.to_string());
+    }
 
-    Ok(
-        ColumnSchema::new(&column_def.name, data_type.into(), column_def.is_nullable)
-            .with_default_constraint(constraint)
-            .context(error::InvalidColumnDefaultConstraintSnafu {
-                column: &column_def.name,
-            })?
-            .with_metadata(metadata),
-    )
+    ColumnSchema::new(&column_def.name, data_type.into(), column_def.is_nullable)
+        .with_metadata(metadata)
+        .with_time_index(column_def.semantic_type() == SemanticType::Timestamp)
+        .with_default_constraint(constraint)
+        .context(error::InvalidColumnDefaultConstraintSnafu {
+            column: &column_def.name,
+        })
+}
+
+/// Constructs a `ColumnOptions` from the given `ColumnSchema`.
+pub fn options_from_column_schema(column_schema: &ColumnSchema) -> Option<ColumnOptions> {
+    let mut options = ColumnOptions::default();
+    if let Some(fulltext) = column_schema.metadata().get(FULLTEXT_KEY) {
+        options
+            .options
+            .insert(FULLTEXT_GRPC_KEY.to_string(), fulltext.to_string());
+    }
+
+    (!options.options.is_empty()).then_some(options)
+}
+
+/// Checks if the `ColumnOptions` contains fulltext options.
+pub fn contains_fulltext(options: &Option<ColumnOptions>) -> bool {
+    options
+        .as_ref()
+        .map_or(false, |o| o.options.contains_key(FULLTEXT_GRPC_KEY))
+}
+
+/// Tries to construct a `ColumnOptions` from the given `FulltextOptions`.
+pub fn options_from_fulltext(fulltext: &FulltextOptions) -> Result<Option<ColumnOptions>> {
+    let mut options = ColumnOptions::default();
+
+    let v = serde_json::to_string(fulltext).context(error::SerializeJsonSnafu)?;
+    options.options.insert(FULLTEXT_GRPC_KEY.to_string(), v);
+
+    Ok((!options.options.is_empty()).then_some(options))
+}
+
+#[cfg(test)]
+mod tests {
+
+    use datatypes::data_type::ConcreteDataType;
+    use datatypes::schema::FulltextAnalyzer;
+
+    use super::*;
+    use crate::v1::ColumnDataType;
+
+    #[test]
+    fn test_try_as_column_schema() {
+        let column_def = ColumnDef {
+            name: "test".to_string(),
+            data_type: ColumnDataType::String as i32,
+            is_nullable: true,
+            default_constraint: ColumnDefaultConstraint::Value("test_default".into())
+                .try_into()
+                .unwrap(),
+            semantic_type: SemanticType::Field as i32,
+            comment: "test_comment".to_string(),
+            datatype_extension: None,
+            options: Some(ColumnOptions {
+                options: HashMap::from([(
+                    FULLTEXT_GRPC_KEY.to_string(),
+                    "{\"enable\":true}".to_string(),
+                )]),
+            }),
+        };
+
+        let schema = try_as_column_schema(&column_def).unwrap();
+        assert_eq!(schema.name, "test");
+        assert_eq!(schema.data_type, ConcreteDataType::string_datatype());
+        assert!(!schema.is_time_index());
+        assert!(schema.is_nullable());
+        assert_eq!(
+            schema.default_constraint().unwrap(),
+            &ColumnDefaultConstraint::Value("test_default".into())
+        );
+        assert_eq!(schema.metadata().get(COMMENT_KEY).unwrap(), "test_comment");
+        assert_eq!(
+            schema.fulltext_options().unwrap().unwrap(),
+            FulltextOptions {
+                enable: true,
+                ..Default::default()
+            }
+        );
+    }
+
+    #[test]
+    fn test_options_from_column_schema() {
+        let schema = ColumnSchema::new("test", ConcreteDataType::string_datatype(), true);
+        let options = options_from_column_schema(&schema);
+        assert!(options.is_none());
+
+        let schema = ColumnSchema::new("test", ConcreteDataType::string_datatype(), true)
+            .with_fulltext_options(FulltextOptions {
+                enable: true,
+                analyzer: FulltextAnalyzer::English,
+                case_sensitive: false,
+            })
+            .unwrap();
+        let options = options_from_column_schema(&schema).unwrap();
+        assert_eq!(
+            options.options.get(FULLTEXT_GRPC_KEY).unwrap(),
+            "{\"enable\":true,\"analyzer\":\"English\",\"case-sensitive\":false}"
+        );
+    }
+
+    #[test]
+    fn test_options_with_fulltext() {
+        let fulltext = FulltextOptions {
+            enable: true,
+            analyzer: FulltextAnalyzer::English,
+            case_sensitive: false,
+        };
+        let options = options_from_fulltext(&fulltext).unwrap().unwrap();
+        assert_eq!(
+            options.options.get(FULLTEXT_GRPC_KEY).unwrap(),
+            "{\"enable\":true,\"analyzer\":\"English\",\"case-sensitive\":false}"
+        );
+    }
+
+    #[test]
+    fn test_contains_fulltext() {
+        let options = ColumnOptions {
+            options: HashMap::from([(
+                FULLTEXT_GRPC_KEY.to_string(),
+                "{\"enable\":true}".to_string(),
+            )]),
+        };
+        assert!(contains_fulltext(&Some(options)));
+
+        let options = ColumnOptions {
+            options: HashMap::new(),
+        };
+        assert!(!contains_fulltext(&Some(options)));
+
+        assert!(!contains_fulltext(&None));
+    }
 }

src/auth/src/error.rs

@@ -38,10 +38,11 @@ pub enum Error {
         location: Location,
     },
 
-    #[snafu(display("Auth failed"))]
+    #[snafu(display("Authentication source failure"))]
     AuthBackend {
         #[snafu(implicit)]
         location: Location,
+        #[snafu(source)]
         source: BoxedError,
     },
 
@@ -87,7 +88,7 @@ impl ErrorExt for Error {
             Error::IllegalParam { .. } => StatusCode::InvalidArguments,
             Error::FileWatch { .. } => StatusCode::InvalidArguments,
             Error::InternalState { .. } => StatusCode::Unexpected,
-            Error::Io { .. } => StatusCode::Internal,
+            Error::Io { .. } => StatusCode::StorageUnavailable,
             Error::AuthBackend { .. } => StatusCode::Internal,
 
             Error::UserNotFound { .. } => StatusCode::UserNotFound,

src/auth/src/permission.rs

@@ -42,7 +42,7 @@ pub enum PermissionResp {
 pub trait PermissionChecker: Send + Sync {
     fn check_permission(
         &self,
-        user_info: Option<UserInfoRef>,
+        user_info: UserInfoRef,
         req: PermissionReq,
     ) -> Result<PermissionResp>;
 }
@@ -50,7 +50,7 @@ pub trait PermissionChecker: Send + Sync {
 impl PermissionChecker for Option<&PermissionCheckerRef> {
     fn check_permission(
         &self,
-        user_info: Option<UserInfoRef>,
+        user_info: UserInfoRef,
         req: PermissionReq,
     ) -> Result<PermissionResp> {
         match self {

src/auth/tests/mod.rs

@@ -27,7 +27,7 @@ struct DummyPermissionChecker;
 impl PermissionChecker for DummyPermissionChecker {
     fn check_permission(
         &self,
-        _user_info: Option<UserInfoRef>,
+        _user_info: UserInfoRef,
         req: PermissionReq,
     ) -> auth::error::Result<PermissionResp> {
         match req {
@@ -45,13 +45,13 @@ fn test_permission_checker() {
     let checker: PermissionCheckerRef = Arc::new(DummyPermissionChecker);
 
     let grpc_result = checker.check_permission(
-        None,
+        auth::userinfo_by_name(None),
         PermissionReq::GrpcRequest(&Request::Query(Default::default())),
     );
     assert_matches!(grpc_result, Ok(PermissionResp::Allow));
 
     let sql_result = checker.check_permission(
-        None,
+        auth::userinfo_by_name(None),
         PermissionReq::SqlStatement(&Statement::ShowDatabases(ShowDatabases::new(
             ShowKind::All,
             false,
@@ -59,6 +59,7 @@ fn test_permission_checker() {
     );
     assert_matches!(sql_result, Ok(PermissionResp::Reject));
 
-    let err_result = checker.check_permission(None, PermissionReq::Opentsdb);
+    let err_result =
+        checker.check_permission(auth::userinfo_by_name(None), PermissionReq::Opentsdb);
     assert_matches!(err_result, Err(InternalState { msg }) if msg == "testing");
 }

src/cache/src/error.rs

@@ -34,7 +34,7 @@ pub type Result<T> = std::result::Result<T, Error>;
 impl ErrorExt for Error {
     fn status_code(&self) -> StatusCode {
         match self {
-            Error::CacheRequired { .. } => StatusCode::Internal,
+            Error::CacheRequired { .. } => StatusCode::Unexpected,
         }
     }
 

src/catalog/Cargo.toml

@@ -40,6 +40,7 @@ moka = { workspace = true, features = ["future", "sync"] }
 partition.workspace = true
 paste = "1.0"
 prometheus.workspace = true
+rustc-hash.workspace = true
 serde_json.workspace = true
 session.workspace = true
 snafu.workspace = true

src/catalog/src/error.rs

@@ -18,6 +18,7 @@ use std::fmt::Debug;
 use common_error::ext::{BoxedError, ErrorExt};
 use common_error::status_code::StatusCode;
 use common_macro::stack_trace_debug;
+use common_query::error::datafusion_status_code;
 use datafusion::error::DataFusionError;
 use snafu::{Location, Snafu};
 
@@ -56,6 +57,31 @@ pub enum Error {
         source: BoxedError,
     },
 
+    #[snafu(display("Failed to list flows in catalog {catalog}"))]
+    ListFlows {
+        #[snafu(implicit)]
+        location: Location,
+        catalog: String,
+        source: BoxedError,
+    },
+
+    #[snafu(display("Flow info not found: {flow_name} in catalog {catalog_name}"))]
+    FlowInfoNotFound {
+        flow_name: String,
+        catalog_name: String,
+        #[snafu(implicit)]
+        location: Location,
+    },
+
+    #[snafu(display("Can't convert value to json, input={input}"))]
+    Json {
+        input: String,
+        #[snafu(source)]
+        error: serde_json::error::Error,
+        #[snafu(implicit)]
+        location: Location,
+    },
+
     #[snafu(display("Failed to re-compile script due to internal error"))]
     CompileScriptInternal {
         #[snafu(implicit)]
@@ -114,6 +140,18 @@ pub enum Error {
         location: Location,
     },
 
+    #[snafu(display(
+        "View plan columns changed from: {} to: {}",
+        origin_names,
+        actual_names
+    ))]
+    ViewPlanColumnsChanged {
+        origin_names: String,
+        actual_names: String,
+        #[snafu(implicit)]
+        location: Location,
+    },
+
     #[snafu(display("Failed to find table partitions"))]
     FindPartitions { source: partition::error::Error },
 
@@ -173,6 +211,14 @@ pub enum Error {
         location: Location,
     },
 
+    #[snafu(display("Failed to project view columns"))]
+    ProjectViewColumns {
+        #[snafu(source)]
+        error: DataFusionError,
+        #[snafu(implicit)]
+        location: Location,
+    },
+
     #[snafu(display("Table metadata manager error"))]
     TableMetadataManager {
         source: common_meta::error::Error,
@@ -208,6 +254,21 @@ pub enum Error {
     },
 }
 
+impl Error {
+    pub fn should_fail(&self) -> bool {
+        use Error::*;
+
+        matches!(
+            self,
+            GetViewCache { .. }
+                | ViewInfoNotFound { .. }
+                | DecodePlan { .. }
+                | ViewPlanColumnsChanged { .. }
+                | ProjectViewColumns { .. }
+        )
+    }
+}
+
 pub type Result<T> = std::result::Result<T, Error>;
 
 impl ErrorExt for Error {
@@ -218,10 +279,15 @@ impl ErrorExt for Error {
             | Error::FindPartitions { .. }
             | Error::FindRegionRoutes { .. }
             | Error::CacheNotFound { .. }
-            | Error::CastManager { .. } => StatusCode::Unexpected,
+            | Error::CastManager { .. }
+            | Error::Json { .. } => StatusCode::Unexpected,
+
+            Error::ViewPlanColumnsChanged { .. } => StatusCode::InvalidArguments,
 
             Error::ViewInfoNotFound { .. } => StatusCode::TableNotFound,
 
+            Error::FlowInfoNotFound { .. } => StatusCode::FlowNotFound,
+
             Error::SystemCatalog { .. } => StatusCode::StorageUnavailable,
 
             Error::UpgradeWeakCatalogManagerRef { .. } => StatusCode::Internal,
@@ -232,7 +298,8 @@ impl ErrorExt for Error {
             Error::ListCatalogs { source, .. }
             | Error::ListNodes { source, .. }
             | Error::ListSchemas { source, .. }
-            | Error::ListTables { source, .. } => source.status_code(),
+            | Error::ListTables { source, .. }
+            | Error::ListFlows { source, .. } => source.status_code(),
 
             Error::CreateTable { source, .. } => source.status_code(),
 
@@ -245,7 +312,8 @@ impl ErrorExt for Error {
             }
 
             Error::QueryAccessDenied { .. } => StatusCode::AccessDenied,
-            Error::Datafusion { .. } => StatusCode::EngineExecuteQuery,
+            Error::Datafusion { error, .. } => datafusion_status_code::<Self>(error, None),
+            Error::ProjectViewColumns { .. } => StatusCode::EngineExecuteQuery,
             Error::TableMetadataManager { source, .. } => source.status_code(),
             Error::GetViewCache { source, .. } | Error::GetTableCache { source, .. } => {
                 source.status_code()
@@ -260,7 +328,7 @@ impl ErrorExt for Error {
 
 impl From<Error> for DataFusionError {
     fn from(e: Error) -> Self {
-        DataFusionError::Internal(e.to_string())
+        DataFusionError::External(Box::new(e))
     }
 }
 
@@ -299,7 +367,7 @@ mod tests {
         }
         .into();
         match e {
-            DataFusionError::Internal(_) => {}
+            DataFusionError::External(_) => {}
             _ => {
                 panic!("catalog error should be converted to DataFusionError::Internal")
             }

src/catalog/src/kvbackend/manager.rs

@@ -19,11 +19,13 @@ use std::sync::{Arc, Weak};
 use async_stream::try_stream;
 use common_catalog::consts::{
     DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, INFORMATION_SCHEMA_NAME, NUMBERS_TABLE_ID,
+    PG_CATALOG_NAME,
 };
 use common_config::Mode;
 use common_error::ext::BoxedError;
 use common_meta::cache::{LayeredCacheRegistryRef, ViewInfoCacheRef};
 use common_meta::key::catalog_name::CatalogNameKey;
+use common_meta::key::flow::FlowMetadataManager;
 use common_meta::key::schema_name::SchemaNameKey;
 use common_meta::key::table_info::TableInfoValue;
 use common_meta::key::table_name::TableNameKey;
@@ -46,6 +48,8 @@ use crate::error::{
 };
 use crate::information_schema::InformationSchemaProvider;
 use crate::kvbackend::TableCacheRef;
+use crate::system_schema::pg_catalog::PGCatalogProvider;
+use crate::system_schema::SystemSchemaProvider;
 use crate::CatalogManager;
 
 /// Access all existing catalog, schema and tables.
@@ -82,14 +86,21 @@ impl KvBackendCatalogManager {
                     .get()
                     .expect("Failed to get table_route_cache"),
             )),
-            table_metadata_manager: Arc::new(TableMetadataManager::new(backend)),
+            table_metadata_manager: Arc::new(TableMetadataManager::new(backend.clone())),
             system_catalog: SystemCatalog {
                 catalog_manager: me.clone(),
                 catalog_cache: Cache::new(CATALOG_CACHE_MAX_CAPACITY),
+                pg_catalog_cache: Cache::new(CATALOG_CACHE_MAX_CAPACITY),
                 information_schema_provider: Arc::new(InformationSchemaProvider::new(
                     DEFAULT_CATALOG_NAME.to_string(),
                     me.clone(),
+                    Arc::new(FlowMetadataManager::new(backend.clone())),
                 )),
+                pg_catalog_provider: Arc::new(PGCatalogProvider::new(
+                    DEFAULT_CATALOG_NAME.to_string(),
+                    me.clone(),
+                )),
+                backend,
             },
             cache_registry,
         })
@@ -295,30 +306,41 @@ fn build_table(table_info_value: TableInfoValue) -> Result<TableRef> {
 /// Existing system tables:
 /// - public.numbers
 /// - information_schema.{tables}
+/// - pg_catalog.{tables}
 #[derive(Clone)]
 struct SystemCatalog {
     catalog_manager: Weak<KvBackendCatalogManager>,
     catalog_cache: Cache<String, Arc<InformationSchemaProvider>>,
+    pg_catalog_cache: Cache<String, Arc<PGCatalogProvider>>,
+
+    // system_schema_provier for default catalog
     information_schema_provider: Arc<InformationSchemaProvider>,
+    pg_catalog_provider: Arc<PGCatalogProvider>,
+    backend: KvBackendRef,
 }
 
 impl SystemCatalog {
+    // TODO(j0hn50n133): remove the duplicated hard-coded table names logic
     fn schema_names(&self) -> Vec<String> {
-        vec![INFORMATION_SCHEMA_NAME.to_string()]
+        vec![
+            INFORMATION_SCHEMA_NAME.to_string(),
+            PG_CATALOG_NAME.to_string(),
+        ]
     }
 
     fn table_names(&self, schema: &str) -> Vec<String> {
-        if schema == INFORMATION_SCHEMA_NAME {
-            self.information_schema_provider.table_names()
-        } else if schema == DEFAULT_SCHEMA_NAME {
-            vec![NUMBERS_TABLE_NAME.to_string()]
-        } else {
-            vec![]
+        match schema {
+            INFORMATION_SCHEMA_NAME => self.information_schema_provider.table_names(),
+            PG_CATALOG_NAME => self.pg_catalog_provider.table_names(),
+            DEFAULT_SCHEMA_NAME => {
+                vec![NUMBERS_TABLE_NAME.to_string()]
+            }
+            _ => vec![],
         }
     }
 
     fn schema_exists(&self, schema: &str) -> bool {
-        schema == INFORMATION_SCHEMA_NAME
+        schema == INFORMATION_SCHEMA_NAME || schema == PG_CATALOG_NAME
     }
 
     fn table_exists(&self, schema: &str, table: &str) -> bool {
@@ -326,6 +348,8 @@ impl SystemCatalog {
             self.information_schema_provider.table(table).is_some()
         } else if schema == DEFAULT_SCHEMA_NAME {
             table == NUMBERS_TABLE_NAME
+        } else if schema == PG_CATALOG_NAME {
+            self.pg_catalog_provider.table(table).is_some()
         } else {
             false
         }
@@ -338,9 +362,23 @@ impl SystemCatalog {
                     Arc::new(InformationSchemaProvider::new(
                         catalog.to_string(),
                         self.catalog_manager.clone(),
+                        Arc::new(FlowMetadataManager::new(self.backend.clone())),
                     ))
                 });
             information_schema_provider.table(table_name)
+        } else if schema == PG_CATALOG_NAME {
+            if catalog == DEFAULT_CATALOG_NAME {
+                self.pg_catalog_provider.table(table_name)
+            } else {
+                let pg_catalog_provider =
+                    self.pg_catalog_cache.get_with_by_ref(catalog, move || {
+                        Arc::new(PGCatalogProvider::new(
+                            catalog.to_string(),
+                            self.catalog_manager.clone(),
+                        ))
+                    });
+                pg_catalog_provider.table(table_name)
+            }
         } else if schema == DEFAULT_SCHEMA_NAME && table_name == NUMBERS_TABLE_NAME {
             Some(NumbersTable::table(NUMBERS_TABLE_ID))
         } else {

src/catalog/src/lib.rs

@@ -28,12 +28,16 @@ use table::TableRef;
 use crate::error::Result;
 
 pub mod error;
-pub mod information_schema;
 pub mod kvbackend;
 pub mod memory;
 mod metrics;
-pub mod table_source;
+pub mod system_schema;
+pub mod information_schema {
+    // TODO(j0hn50n133): re-export to make it compatible with the legacy code, migrate to the new path later
+    pub use crate::system_schema::information_schema::*;
+}
 
+pub mod table_source;
 #[async_trait::async_trait]
 pub trait CatalogManager: Send + Sync {
     fn as_any(&self) -> &dyn Any;

src/catalog/src/memory/manager.rs

@@ -20,14 +20,18 @@ use std::sync::{Arc, RwLock, Weak};
 use async_stream::{stream, try_stream};
 use common_catalog::build_db_string;
 use common_catalog::consts::{
-    DEFAULT_CATALOG_NAME, DEFAULT_PRIVATE_SCHEMA_NAME, DEFAULT_SCHEMA_NAME, INFORMATION_SCHEMA_NAME,
+    DEFAULT_CATALOG_NAME, DEFAULT_PRIVATE_SCHEMA_NAME, DEFAULT_SCHEMA_NAME,
+    INFORMATION_SCHEMA_NAME, PG_CATALOG_NAME,
 };
+use common_meta::key::flow::FlowMetadataManager;
+use common_meta::kv_backend::memory::MemoryKvBackend;
 use futures_util::stream::BoxStream;
 use snafu::OptionExt;
 use table::TableRef;
 
 use crate::error::{CatalogNotFoundSnafu, Result, SchemaNotFoundSnafu, TableExistsSnafu};
 use crate::information_schema::InformationSchemaProvider;
+use crate::system_schema::SystemSchemaProvider;
 use crate::{CatalogManager, DeregisterTableRequest, RegisterSchemaRequest, RegisterTableRequest};
 
 type SchemaEntries = HashMap<String, HashMap<String, TableRef>>;
@@ -173,6 +177,12 @@ impl MemoryCatalogManager {
                 schema: DEFAULT_PRIVATE_SCHEMA_NAME.to_string(),
             })
             .unwrap();
+        manager
+            .register_schema_sync(RegisterSchemaRequest {
+                catalog: DEFAULT_CATALOG_NAME.to_string(),
+                schema: PG_CATALOG_NAME.to_string(),
+            })
+            .unwrap();
         manager
             .register_schema_sync(RegisterSchemaRequest {
                 catalog: DEFAULT_CATALOG_NAME.to_string(),
@@ -196,7 +206,7 @@ impl MemoryCatalogManager {
     }
 
     fn catalog_exist_sync(&self, catalog: &str) -> Result<bool> {
-        Ok(self.catalogs.read().unwrap().get(catalog).is_some())
+        Ok(self.catalogs.read().unwrap().contains_key(catalog))
     }
 
     /// Registers a catalog if it does not exist and returns false if the schema exists.
@@ -290,6 +300,7 @@ impl MemoryCatalogManager {
         let information_schema_provider = InformationSchemaProvider::new(
             catalog,
             Arc::downgrade(self) as Weak<dyn CatalogManager>,
+            Arc::new(FlowMetadataManager::new(Arc::new(MemoryKvBackend::new()))),
         );
         let information_schema = information_schema_provider.tables().clone();
 

src/catalog/src/system_schema.rs

@@ -0,0 +1,166 @@
+// 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 information_schema;
+mod memory_table;
+pub mod pg_catalog;
+mod predicate;
+mod utils;
+
+use std::collections::HashMap;
+use std::sync::Arc;
+
+use common_error::ext::BoxedError;
+use common_recordbatch::{RecordBatchStreamWrapper, SendableRecordBatchStream};
+use datatypes::schema::SchemaRef;
+use futures_util::StreamExt;
+use snafu::ResultExt;
+use store_api::data_source::DataSource;
+use store_api::storage::ScanRequest;
+use table::error::{SchemaConversionSnafu, TablesRecordBatchSnafu};
+use table::metadata::{
+    FilterPushDownType, TableId, TableInfoBuilder, TableInfoRef, TableMetaBuilder, TableType,
+};
+use table::{Table, TableRef};
+
+use crate::error::Result;
+
+pub trait SystemSchemaProvider {
+    /// Returns a map of [TableRef] in information schema.
+    fn tables(&self) -> &HashMap<String, TableRef>;
+
+    /// Returns the [TableRef] by table name.
+    fn table(&self, name: &str) -> Option<TableRef> {
+        self.tables().get(name).cloned()
+    }
+
+    /// Returns table names in the order of table id.
+    fn table_names(&self) -> Vec<String> {
+        let mut tables = self.tables().values().clone().collect::<Vec<_>>();
+
+        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()
+    }
+}
+
+trait SystemSchemaProviderInner {
+    fn catalog_name(&self) -> &str;
+    fn schema_name() -> &'static str;
+    fn build_table(&self, name: &str) -> Option<TableRef> {
+        self.system_table(name).map(|table| {
+            let table_info = Self::table_info(self.catalog_name().to_string(), &table);
+            let filter_pushdown = FilterPushDownType::Inexact;
+            let data_source = Arc::new(SystemTableDataSource::new(table));
+            let table = Table::new(table_info, filter_pushdown, data_source);
+            Arc::new(table)
+        })
+    }
+    fn system_table(&self, name: &str) -> Option<SystemTableRef>;
+
+    fn table_info(catalog_name: String, table: &SystemTableRef) -> TableInfoRef {
+        let table_meta = TableMetaBuilder::default()
+            .schema(table.schema())
+            .primary_key_indices(vec![])
+            .next_column_id(0)
+            .build()
+            .unwrap();
+        let table_info = TableInfoBuilder::default()
+            .table_id(table.table_id())
+            .name(table.table_name().to_string())
+            .catalog_name(catalog_name)
+            .schema_name(Self::schema_name().to_string())
+            .meta(table_meta)
+            .table_type(table.table_type())
+            .build()
+            .unwrap();
+        Arc::new(table_info)
+    }
+}
+
+pub(crate) trait SystemTable {
+    fn table_id(&self) -> TableId;
+
+    fn table_name(&self) -> &'static str;
+
+    fn schema(&self) -> SchemaRef;
+
+    fn to_stream(&self, request: ScanRequest) -> Result<SendableRecordBatchStream>;
+
+    fn table_type(&self) -> TableType {
+        TableType::Temporary
+    }
+}
+
+pub(crate) type SystemTableRef = Arc<dyn SystemTable + Send + Sync>;
+
+struct SystemTableDataSource {
+    table: SystemTableRef,
+}
+
+impl SystemTableDataSource {
+    fn new(table: SystemTableRef) -> Self {
+        Self { table }
+    }
+
+    fn try_project(&self, projection: &[usize]) -> std::result::Result<SchemaRef, BoxedError> {
+        let schema = self
+            .table
+            .schema()
+            .try_project(projection)
+            .context(SchemaConversionSnafu)
+            .map_err(BoxedError::new)?;
+        Ok(Arc::new(schema))
+    }
+}
+
+impl DataSource for SystemTableDataSource {
+    fn get_stream(
+        &self,
+        request: ScanRequest,
+    ) -> std::result::Result<SendableRecordBatchStream, BoxedError> {
+        let projection = request.projection.clone();
+        let projected_schema = match &projection {
+            Some(projection) => self.try_project(projection)?,
+            None => self.table.schema(),
+        };
+
+        let stream = self
+            .table
+            .to_stream(request)
+            .map_err(BoxedError::new)
+            .context(TablesRecordBatchSnafu)
+            .map_err(BoxedError::new)?
+            .map(move |batch| match &projection {
+                Some(p) => batch.and_then(|b| b.try_project(p)),
+                None => batch,
+            });
+
+        let stream = RecordBatchStreamWrapper {
+            schema: projected_schema,
+            stream: Box::pin(stream),
+            output_ordering: None,
+            metrics: Default::default(),
+        };
+
+        Ok(Box::pin(stream))
+    }
+}

src/catalog/src/system_schema/information_schema.rs

@@ -14,50 +14,49 @@
 
 mod cluster_info;
 pub mod columns;
+pub mod flows;
+mod information_memory_table;
 pub mod key_column_usage;
-mod memory_table;
 mod partitions;
-mod predicate;
 mod region_peers;
 mod runtime_metrics;
 pub mod schemata;
 mod table_constraints;
 mod table_names;
 pub mod tables;
-pub(crate) mod utils;
+mod views;
 
 use std::collections::HashMap;
 use std::sync::{Arc, Weak};
 
 use common_catalog::consts::{self, DEFAULT_CATALOG_NAME, INFORMATION_SCHEMA_NAME};
-use common_error::ext::BoxedError;
-use common_recordbatch::{RecordBatchStreamWrapper, SendableRecordBatchStream};
+use common_meta::key::flow::FlowMetadataManager;
+use common_recordbatch::SendableRecordBatchStream;
 use datatypes::schema::SchemaRef;
-use futures_util::StreamExt;
 use lazy_static::lazy_static;
 use paste::paste;
-pub(crate) use predicate::Predicates;
-use snafu::ResultExt;
-use store_api::data_source::DataSource;
 use store_api::storage::{ScanRequest, TableId};
-use table::error::{SchemaConversionSnafu, TablesRecordBatchSnafu};
-use table::metadata::{
-    FilterPushDownType, TableInfoBuilder, TableInfoRef, TableMetaBuilder, TableType,
-};
-use table::{Table, TableRef};
+use table::metadata::TableType;
+use table::TableRef;
 pub use table_names::*;
+use views::InformationSchemaViews;
 
 use self::columns::InformationSchemaColumns;
+use super::{SystemSchemaProviderInner, SystemTable, SystemTableRef};
 use crate::error::Result;
-use crate::information_schema::cluster_info::InformationSchemaClusterInfo;
-use crate::information_schema::key_column_usage::InformationSchemaKeyColumnUsage;
-use crate::information_schema::memory_table::{get_schema_columns, MemoryTable};
-use crate::information_schema::partitions::InformationSchemaPartitions;
-use crate::information_schema::region_peers::InformationSchemaRegionPeers;
-use crate::information_schema::runtime_metrics::InformationSchemaMetrics;
-use crate::information_schema::schemata::InformationSchemaSchemata;
-use crate::information_schema::table_constraints::InformationSchemaTableConstraints;
-use crate::information_schema::tables::InformationSchemaTables;
+use crate::system_schema::information_schema::cluster_info::InformationSchemaClusterInfo;
+use crate::system_schema::information_schema::flows::InformationSchemaFlows;
+use crate::system_schema::information_schema::information_memory_table::get_schema_columns;
+use crate::system_schema::information_schema::key_column_usage::InformationSchemaKeyColumnUsage;
+use crate::system_schema::information_schema::partitions::InformationSchemaPartitions;
+use crate::system_schema::information_schema::region_peers::InformationSchemaRegionPeers;
+use crate::system_schema::information_schema::runtime_metrics::InformationSchemaMetrics;
+use crate::system_schema::information_schema::schemata::InformationSchemaSchemata;
+use crate::system_schema::information_schema::table_constraints::InformationSchemaTableConstraints;
+use crate::system_schema::information_schema::tables::InformationSchemaTables;
+use crate::system_schema::memory_table::MemoryTable;
+pub(crate) use crate::system_schema::predicate::Predicates;
+use crate::system_schema::SystemSchemaProvider;
 use crate::CatalogManager;
 
 lazy_static! {
@@ -106,107 +105,26 @@ macro_rules! setup_memory_table {
 pub struct InformationSchemaProvider {
     catalog_name: String,
     catalog_manager: Weak<dyn CatalogManager>,
+    flow_metadata_manager: Arc<FlowMetadataManager>,
     tables: HashMap<String, TableRef>,
 }
 
-impl InformationSchemaProvider {
-    pub fn new(catalog_name: String, catalog_manager: Weak<dyn CatalogManager>) -> Self {
-        let mut provider = Self {
-            catalog_name,
-            catalog_manager,
-            tables: HashMap::new(),
-        };
-
-        provider.build_tables();
-
-        provider
-    }
-
-    /// 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<_>>();
-
-        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> {
+impl SystemSchemaProvider for InformationSchemaProvider {
+    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()
+}
+impl SystemSchemaProviderInner for InformationSchemaProvider {
+    fn catalog_name(&self) -> &str {
+        &self.catalog_name
+    }
+    fn schema_name() -> &'static str {
+        INFORMATION_SCHEMA_NAME
     }
 
-    fn build_tables(&mut self) {
-        let mut tables = HashMap::new();
-
-        // SECURITY NOTE:
-        // Carefully consider the tables that may expose sensitive cluster configurations,
-        // authentication details, and other critical information.
-        // Only put these tables under `greptime` catalog to prevent info leak.
-        if self.catalog_name == DEFAULT_CATALOG_NAME {
-            tables.insert(
-                RUNTIME_METRICS.to_string(),
-                self.build_table(RUNTIME_METRICS).unwrap(),
-            );
-            tables.insert(
-                BUILD_INFO.to_string(),
-                self.build_table(BUILD_INFO).unwrap(),
-            );
-            tables.insert(
-                REGION_PEERS.to_string(),
-                self.build_table(REGION_PEERS).unwrap(),
-            );
-            tables.insert(
-                CLUSTER_INFO.to_string(),
-                self.build_table(CLUSTER_INFO).unwrap(),
-            );
-        }
-
-        tables.insert(TABLES.to_string(), self.build_table(TABLES).unwrap());
-        tables.insert(SCHEMATA.to_string(), self.build_table(SCHEMATA).unwrap());
-        tables.insert(COLUMNS.to_string(), self.build_table(COLUMNS).unwrap());
-        tables.insert(
-            KEY_COLUMN_USAGE.to_string(),
-            self.build_table(KEY_COLUMN_USAGE).unwrap(),
-        );
-        tables.insert(
-            TABLE_CONSTRAINTS.to_string(),
-            self.build_table(TABLE_CONSTRAINTS).unwrap(),
-        );
-
-        // Add memory tables
-        for name in MEMORY_TABLES.iter() {
-            tables.insert((*name).to_string(), self.build_table(name).expect(name));
-        }
-
-        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::Inexact;
-            let data_source = Arc::new(InformationTableDataSource::new(table));
-            let table = Table::new(table_info, filter_pushdown, data_source);
-            Arc::new(table)
-        })
-    }
-
-    fn information_table(&self, name: &str) -> Option<InformationTableRef> {
+    fn system_table(&self, name: &str) -> Option<SystemTableRef> {
         match name.to_ascii_lowercase().as_str() {
             TABLES => Some(Arc::new(InformationSchemaTables::new(
                 self.catalog_name.clone(),
@@ -262,27 +180,83 @@ impl InformationSchemaProvider {
             CLUSTER_INFO => Some(Arc::new(InformationSchemaClusterInfo::new(
                 self.catalog_manager.clone(),
             )) as _),
+            VIEWS => Some(Arc::new(InformationSchemaViews::new(
+                self.catalog_name.clone(),
+                self.catalog_manager.clone(),
+            )) as _),
+            FLOWS => Some(Arc::new(InformationSchemaFlows::new(
+                self.catalog_name.clone(),
+                self.flow_metadata_manager.clone(),
+            )) as _),
             _ => None,
         }
     }
+}
 
-    fn table_info(catalog_name: String, table: &InformationTableRef) -> TableInfoRef {
-        let table_meta = TableMetaBuilder::default()
-            .schema(table.schema())
-            .primary_key_indices(vec![])
-            .next_column_id(0)
-            .build()
-            .unwrap();
-        let table_info = TableInfoBuilder::default()
-            .table_id(table.table_id())
-            .name(table.table_name().to_string())
-            .catalog_name(catalog_name)
-            .schema_name(INFORMATION_SCHEMA_NAME.to_string())
-            .meta(table_meta)
-            .table_type(table.table_type())
-            .build()
-            .unwrap();
-        Arc::new(table_info)
+impl InformationSchemaProvider {
+    pub fn new(
+        catalog_name: String,
+        catalog_manager: Weak<dyn CatalogManager>,
+        flow_metadata_manager: Arc<FlowMetadataManager>,
+    ) -> Self {
+        let mut provider = Self {
+            catalog_name,
+            catalog_manager,
+            flow_metadata_manager,
+            tables: HashMap::new(),
+        };
+
+        provider.build_tables();
+
+        provider
+    }
+
+    fn build_tables(&mut self) {
+        let mut tables = HashMap::new();
+
+        // SECURITY NOTE:
+        // Carefully consider the tables that may expose sensitive cluster configurations,
+        // authentication details, and other critical information.
+        // Only put these tables under `greptime` catalog to prevent info leak.
+        if self.catalog_name == DEFAULT_CATALOG_NAME {
+            tables.insert(
+                RUNTIME_METRICS.to_string(),
+                self.build_table(RUNTIME_METRICS).unwrap(),
+            );
+            tables.insert(
+                BUILD_INFO.to_string(),
+                self.build_table(BUILD_INFO).unwrap(),
+            );
+            tables.insert(
+                REGION_PEERS.to_string(),
+                self.build_table(REGION_PEERS).unwrap(),
+            );
+            tables.insert(
+                CLUSTER_INFO.to_string(),
+                self.build_table(CLUSTER_INFO).unwrap(),
+            );
+        }
+
+        tables.insert(TABLES.to_string(), self.build_table(TABLES).unwrap());
+        tables.insert(VIEWS.to_string(), self.build_table(VIEWS).unwrap());
+        tables.insert(SCHEMATA.to_string(), self.build_table(SCHEMATA).unwrap());
+        tables.insert(COLUMNS.to_string(), self.build_table(COLUMNS).unwrap());
+        tables.insert(
+            KEY_COLUMN_USAGE.to_string(),
+            self.build_table(KEY_COLUMN_USAGE).unwrap(),
+        );
+        tables.insert(
+            TABLE_CONSTRAINTS.to_string(),
+            self.build_table(TABLE_CONSTRAINTS).unwrap(),
+        );
+        tables.insert(FLOWS.to_string(), self.build_table(FLOWS).unwrap());
+
+        // Add memory tables
+        for name in MEMORY_TABLES.iter() {
+            tables.insert((*name).to_string(), self.build_table(name).expect(name));
+        }
+
+        self.tables = tables;
     }
 }
 
@@ -300,57 +274,28 @@ trait InformationTable {
     }
 }
 
-type InformationTableRef = Arc<dyn InformationTable + Send + Sync>;
-
-struct InformationTableDataSource {
-    table: InformationTableRef,
-}
-
-impl InformationTableDataSource {
-    fn new(table: InformationTableRef) -> Self {
-        Self { table }
+// Provide compatibility for legacy `information_schema` code.
+impl<T> SystemTable for T
+where
+    T: InformationTable,
+{
+    fn table_id(&self) -> TableId {
+        InformationTable::table_id(self)
     }
 
-    fn try_project(&self, projection: &[usize]) -> std::result::Result<SchemaRef, BoxedError> {
-        let schema = self
-            .table
-            .schema()
-            .try_project(projection)
-            .context(SchemaConversionSnafu)
-            .map_err(BoxedError::new)?;
-        Ok(Arc::new(schema))
-    }
-}
-
-impl DataSource for InformationTableDataSource {
-    fn get_stream(
-        &self,
-        request: ScanRequest,
-    ) -> std::result::Result<SendableRecordBatchStream, BoxedError> {
-        let projection = request.projection.clone();
-        let projected_schema = match &projection {
-            Some(projection) => self.try_project(projection)?,
-            None => self.table.schema(),
-        };
-
-        let stream = self
-            .table
-            .to_stream(request)
-            .map_err(BoxedError::new)
-            .context(TablesRecordBatchSnafu)
-            .map_err(BoxedError::new)?
-            .map(move |batch| match &projection {
-                Some(p) => batch.and_then(|b| b.try_project(p)),
-                None => batch,
-            });
-
-        let stream = RecordBatchStreamWrapper {
-            schema: projected_schema,
-            stream: Box::pin(stream),
-            output_ordering: None,
-            metrics: Default::default(),
-        };
-
-        Ok(Box::pin(stream))
+    fn table_name(&self) -> &'static str {
+        InformationTable::table_name(self)
+    }
+
+    fn schema(&self) -> SchemaRef {
+        InformationTable::schema(self)
+    }
+
+    fn table_type(&self) -> TableType {
+        InformationTable::table_type(self)
+    }
+
+    fn to_stream(&self, request: ScanRequest) -> Result<SendableRecordBatchStream> {
+        InformationTable::to_stream(self, request)
     }
 }

src/catalog/src/system_schema/information_schema/cluster_info.rs

@@ -41,7 +41,8 @@ use store_api::storage::{ScanRequest, TableId};
 
 use super::CLUSTER_INFO;
 use crate::error::{CreateRecordBatchSnafu, InternalSnafu, ListNodesSnafu, Result};
-use crate::information_schema::{utils, InformationTable, Predicates};
+use crate::system_schema::information_schema::{InformationTable, Predicates};
+use crate::system_schema::utils;
 use crate::CatalogManager;
 
 const PEER_ID: &str = "peer_id";

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

@@ -0,0 +1,305 @@
+// 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::INFORMATION_SCHEMA_FLOW_TABLE_ID;
+use common_error::ext::BoxedError;
+use common_meta::key::flow::flow_info::FlowInfoValue;
+use common_meta::key::flow::FlowMetadataManager;
+use common_meta::key::FlowId;
+use common_recordbatch::adapter::RecordBatchStreamAdapter;
+use common_recordbatch::{DfSendableRecordBatchStream, RecordBatch, SendableRecordBatchStream};
+use datafusion::execution::TaskContext;
+use datafusion::physical_plan::stream::RecordBatchStreamAdapter as DfRecordBatchStreamAdapter;
+use datafusion::physical_plan::streaming::PartitionStream as DfPartitionStream;
+use datatypes::prelude::ConcreteDataType as CDT;
+use datatypes::scalars::ScalarVectorBuilder;
+use datatypes::schema::{ColumnSchema, Schema, SchemaRef};
+use datatypes::value::Value;
+use datatypes::vectors::{Int64VectorBuilder, StringVectorBuilder, UInt32VectorBuilder, VectorRef};
+use futures::TryStreamExt;
+use snafu::{OptionExt, ResultExt};
+use store_api::storage::{ScanRequest, TableId};
+
+use crate::error::{
+    CreateRecordBatchSnafu, FlowInfoNotFoundSnafu, InternalSnafu, JsonSnafu, ListFlowsSnafu, Result,
+};
+use crate::information_schema::{Predicates, FLOWS};
+use crate::system_schema::information_schema::InformationTable;
+
+const INIT_CAPACITY: usize = 42;
+
+// rows of information_schema.flows
+// pk is (flow_name, flow_id, table_catalog)
+pub const FLOW_NAME: &str = "flow_name";
+pub const FLOW_ID: &str = "flow_id";
+pub const TABLE_CATALOG: &str = "table_catalog";
+pub const FLOW_DEFINITION: &str = "flow_definition";
+pub const COMMENT: &str = "comment";
+pub const EXPIRE_AFTER: &str = "expire_after";
+pub const SOURCE_TABLE_IDS: &str = "source_table_ids";
+pub const SINK_TABLE_NAME: &str = "sink_table_name";
+pub const FLOWNODE_IDS: &str = "flownode_ids";
+pub const OPTIONS: &str = "options";
+
+/// The `information_schema.flows` to provides information about flows in databases.
+pub(super) struct InformationSchemaFlows {
+    schema: SchemaRef,
+    catalog_name: String,
+    flow_metadata_manager: Arc<FlowMetadataManager>,
+}
+
+impl InformationSchemaFlows {
+    pub(super) fn new(
+        catalog_name: String,
+        flow_metadata_manager: Arc<FlowMetadataManager>,
+    ) -> Self {
+        Self {
+            schema: Self::schema(),
+            catalog_name,
+            flow_metadata_manager,
+        }
+    }
+
+    /// for complex fields(including [`SOURCE_TABLE_IDS`], [`FLOWNODE_IDS`] and [`OPTIONS`]), it will be serialized to json string for now
+    /// TODO(discord9): use a better way to store complex fields like json type
+    pub(crate) fn schema() -> SchemaRef {
+        Arc::new(Schema::new(
+            vec![
+                (FLOW_NAME, CDT::string_datatype(), false),
+                (FLOW_ID, CDT::uint32_datatype(), false),
+                (TABLE_CATALOG, CDT::string_datatype(), false),
+                (FLOW_DEFINITION, CDT::string_datatype(), false),
+                (COMMENT, CDT::string_datatype(), true),
+                (EXPIRE_AFTER, CDT::int64_datatype(), true),
+                (SOURCE_TABLE_IDS, CDT::string_datatype(), true),
+                (SINK_TABLE_NAME, CDT::string_datatype(), false),
+                (FLOWNODE_IDS, CDT::string_datatype(), true),
+                (OPTIONS, CDT::string_datatype(), true),
+            ]
+            .into_iter()
+            .map(|(name, ty, nullable)| ColumnSchema::new(name, ty, nullable))
+            .collect(),
+        ))
+    }
+
+    fn builder(&self) -> InformationSchemaFlowsBuilder {
+        InformationSchemaFlowsBuilder::new(
+            self.schema.clone(),
+            self.catalog_name.clone(),
+            &self.flow_metadata_manager,
+        )
+    }
+}
+
+impl InformationTable for InformationSchemaFlows {
+    fn table_id(&self) -> TableId {
+        INFORMATION_SCHEMA_FLOW_TABLE_ID
+    }
+
+    fn table_name(&self) -> &'static str {
+        FLOWS
+    }
+
+    fn schema(&self) -> SchemaRef {
+        self.schema.clone()
+    }
+
+    fn to_stream(&self, request: ScanRequest) -> 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
+                    .make_flows(Some(request))
+                    .await
+                    .map(|x| x.into_df_record_batch())
+                    .map_err(|err| datafusion::error::DataFusionError::External(Box::new(err)))
+            }),
+        ));
+        Ok(Box::pin(
+            RecordBatchStreamAdapter::try_new(stream)
+                .map_err(BoxedError::new)
+                .context(InternalSnafu)?,
+        ))
+    }
+}
+
+/// Builds the `information_schema.FLOWS` table row by row
+///
+/// columns are based on [`FlowInfoValue`]
+struct InformationSchemaFlowsBuilder {
+    schema: SchemaRef,
+    catalog_name: String,
+    flow_metadata_manager: Arc<FlowMetadataManager>,
+
+    flow_names: StringVectorBuilder,
+    flow_ids: UInt32VectorBuilder,
+    table_catalogs: StringVectorBuilder,
+    raw_sqls: StringVectorBuilder,
+    comments: StringVectorBuilder,
+    expire_afters: Int64VectorBuilder,
+    source_table_id_groups: StringVectorBuilder,
+    sink_table_names: StringVectorBuilder,
+    flownode_id_groups: StringVectorBuilder,
+    option_groups: StringVectorBuilder,
+}
+
+impl InformationSchemaFlowsBuilder {
+    fn new(
+        schema: SchemaRef,
+        catalog_name: String,
+        flow_metadata_manager: &Arc<FlowMetadataManager>,
+    ) -> Self {
+        Self {
+            schema,
+            catalog_name,
+            flow_metadata_manager: flow_metadata_manager.clone(),
+
+            flow_names: StringVectorBuilder::with_capacity(INIT_CAPACITY),
+            flow_ids: UInt32VectorBuilder::with_capacity(INIT_CAPACITY),
+            table_catalogs: StringVectorBuilder::with_capacity(INIT_CAPACITY),
+            raw_sqls: StringVectorBuilder::with_capacity(INIT_CAPACITY),
+            comments: StringVectorBuilder::with_capacity(INIT_CAPACITY),
+            expire_afters: Int64VectorBuilder::with_capacity(INIT_CAPACITY),
+            source_table_id_groups: StringVectorBuilder::with_capacity(INIT_CAPACITY),
+            sink_table_names: StringVectorBuilder::with_capacity(INIT_CAPACITY),
+            flownode_id_groups: StringVectorBuilder::with_capacity(INIT_CAPACITY),
+            option_groups: StringVectorBuilder::with_capacity(INIT_CAPACITY),
+        }
+    }
+
+    /// Construct the `information_schema.flows` virtual table
+    async fn make_flows(&mut self, request: Option<ScanRequest>) -> Result<RecordBatch> {
+        let catalog_name = self.catalog_name.clone();
+        let predicates = Predicates::from_scan_request(&request);
+
+        let flow_info_manager = self.flow_metadata_manager.clone();
+
+        // TODO(discord9): use `AsyncIterator` once it's stable-ish
+        let mut stream = flow_info_manager
+            .flow_name_manager()
+            .flow_names(&catalog_name)
+            .await;
+
+        while let Some((flow_name, flow_id)) = stream
+            .try_next()
+            .await
+            .map_err(BoxedError::new)
+            .context(ListFlowsSnafu {
+                catalog: &catalog_name,
+            })?
+        {
+            let flow_info = flow_info_manager
+                .flow_info_manager()
+                .get(flow_id.flow_id())
+                .await
+                .map_err(BoxedError::new)
+                .context(InternalSnafu)?
+                .context(FlowInfoNotFoundSnafu {
+                    catalog_name: catalog_name.to_string(),
+                    flow_name: flow_name.to_string(),
+                })?;
+            self.add_flow(&predicates, flow_id.flow_id(), flow_info)?;
+        }
+
+        self.finish()
+    }
+
+    fn add_flow(
+        &mut self,
+        predicates: &Predicates,
+        flow_id: FlowId,
+        flow_info: FlowInfoValue,
+    ) -> Result<()> {
+        let row = [
+            (FLOW_NAME, &Value::from(flow_info.flow_name().to_string())),
+            (FLOW_ID, &Value::from(flow_id)),
+            (
+                TABLE_CATALOG,
+                &Value::from(flow_info.catalog_name().to_string()),
+            ),
+        ];
+        if !predicates.eval(&row) {
+            return Ok(());
+        }
+        self.flow_names.push(Some(flow_info.flow_name()));
+        self.flow_ids.push(Some(flow_id));
+        self.table_catalogs.push(Some(flow_info.catalog_name()));
+        self.raw_sqls.push(Some(flow_info.raw_sql()));
+        self.comments.push(Some(flow_info.comment()));
+        self.expire_afters.push(flow_info.expire_after());
+        self.source_table_id_groups.push(Some(
+            &serde_json::to_string(flow_info.source_table_ids()).context(JsonSnafu {
+                input: format!("{:?}", flow_info.source_table_ids()),
+            })?,
+        ));
+        self.sink_table_names
+            .push(Some(&flow_info.sink_table_name().to_string()));
+        self.flownode_id_groups.push(Some(
+            &serde_json::to_string(flow_info.flownode_ids()).context({
+                JsonSnafu {
+                    input: format!("{:?}", flow_info.flownode_ids()),
+                }
+            })?,
+        ));
+        self.option_groups
+            .push(Some(&serde_json::to_string(flow_info.options()).context(
+                JsonSnafu {
+                    input: format!("{:?}", flow_info.options()),
+                },
+            )?));
+
+        Ok(())
+    }
+
+    fn finish(&mut self) -> Result<RecordBatch> {
+        let columns: Vec<VectorRef> = vec![
+            Arc::new(self.flow_names.finish()),
+            Arc::new(self.flow_ids.finish()),
+            Arc::new(self.table_catalogs.finish()),
+            Arc::new(self.raw_sqls.finish()),
+            Arc::new(self.comments.finish()),
+            Arc::new(self.expire_afters.finish()),
+            Arc::new(self.source_table_id_groups.finish()),
+            Arc::new(self.sink_table_names.finish()),
+            Arc::new(self.flownode_id_groups.finish()),
+            Arc::new(self.option_groups.finish()),
+        ];
+        RecordBatch::new(self.schema.clone(), columns).context(CreateRecordBatchSnafu)
+    }
+}
+
+impl DfPartitionStream for InformationSchemaFlows {
+    fn schema(&self) -> &arrow_schema::SchemaRef {
+        self.schema.arrow_schema()
+    }
+
+    fn execute(&self, _: Arc<TaskContext>) -> DfSendableRecordBatchStream {
+        let schema: Arc<arrow_schema::Schema> = self.schema.arrow_schema().clone();
+        let mut builder = self.builder();
+        Box::pin(DfRecordBatchStreamAdapter::new(
+            schema,
+            futures::stream::once(async move {
+                builder
+                    .make_flows(None)
+                    .await
+                    .map(|x| x.into_df_record_batch())
+                    .map_err(Into::into)
+            }),
+        ))
+    }
+}

src/catalog/src/system_schema/information_schema/information_memory_table.rs

@@ -15,17 +15,19 @@
 use std::sync::Arc;
 
 use common_catalog::consts::{METRIC_ENGINE, MITO_ENGINE};
-use datatypes::prelude::{ConcreteDataType, VectorRef};
-use datatypes::schema::{ColumnSchema, Schema, SchemaRef};
-use datatypes::vectors::{Int64Vector, StringVector};
+use datatypes::schema::{Schema, SchemaRef};
+use datatypes::vectors::{Int64Vector, StringVector, VectorRef};
 
-use crate::information_schema::table_names::*;
+use super::table_names::*;
+use crate::system_schema::utils::tables::{
+    bigint_column, datetime_column, string_column, string_columns,
+};
 
 const NO_VALUE: &str = "NO";
 
 /// 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>) {
+pub(super) fn get_schema_columns(table_name: &str) -> (SchemaRef, Vec<VectorRef>) {
     let (column_schemas, columns): (_, Vec<VectorRef>) = match table_name {
         COLUMN_PRIVILEGES => (
             string_columns(&[
@@ -80,7 +82,7 @@ pub fn get_schema_columns(table_name: &str) -> (SchemaRef, Vec<VectorRef>) {
                     "GIT_BRANCH",
                     "GIT_COMMIT",
                     "GIT_COMMIT_SHORT",
-                    "GIT_DIRTY",
+                    "GIT_CLEAN",
                     "PKG_VERSION",
                 ]),
                 vec![
@@ -89,7 +91,7 @@ pub fn get_schema_columns(table_name: &str) -> (SchemaRef, Vec<VectorRef>) {
                     Arc::new(StringVector::from(vec![build_info
                         .commit_short
                         .to_string()])),
-                    Arc::new(StringVector::from(vec![build_info.dirty.to_string()])),
+                    Arc::new(StringVector::from(vec![build_info.clean.to_string()])),
                     Arc::new(StringVector::from(vec![build_info.version.to_string()])),
                 ],
             )
@@ -414,50 +416,3 @@ pub fn get_schema_columns(table_name: &str) -> (SchemaRef, Vec<VectorRef>) {
 
     (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,
-    )
-}
-
-fn bigint_column(name: &str) -> ColumnSchema {
-    ColumnSchema::new(
-        str::to_lowercase(name),
-        ConcreteDataType::int64_datatype(),
-        false,
-    )
-}
-
-fn datetime_column(name: &str) -> ColumnSchema {
-    ColumnSchema::new(
-        str::to_lowercase(name),
-        ConcreteDataType::datetime_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/system_schema/information_schema/key_column_usage.rs

@@ -27,6 +27,7 @@ use datatypes::prelude::{ConcreteDataType, MutableVector, ScalarVectorBuilder, V
 use datatypes::schema::{ColumnSchema, Schema, SchemaRef};
 use datatypes::value::Value;
 use datatypes::vectors::{ConstantVector, StringVector, StringVectorBuilder, UInt32VectorBuilder};
+use futures_util::TryStreamExt;
 use snafu::{OptionExt, ResultExt};
 use store_api::storage::{ScanRequest, TableId};
 
@@ -34,7 +35,7 @@ use super::KEY_COLUMN_USAGE;
 use crate::error::{
     CreateRecordBatchSnafu, InternalSnafu, Result, UpgradeWeakCatalogManagerRefSnafu,
 };
-use crate::information_schema::{InformationTable, Predicates};
+use crate::system_schema::information_schema::{InformationTable, Predicates};
 use crate::CatalogManager;
 
 pub const CONSTRAINT_SCHEMA: &str = "constraint_schema";
@@ -211,71 +212,58 @@ impl InformationSchemaKeyColumnUsageBuilder {
             .context(UpgradeWeakCatalogManagerRefSnafu)?;
         let predicates = Predicates::from_scan_request(&request);
 
-        let mut primary_constraints = vec![];
-
         for schema_name in catalog_manager.schema_names(&catalog_name).await? {
-            if !catalog_manager
-                .schema_exists(&catalog_name, &schema_name)
-                .await?
-            {
-                continue;
-            }
+            let mut stream = catalog_manager.tables(&catalog_name, &schema_name);
 
-            for table_name in catalog_manager
-                .table_names(&catalog_name, &schema_name)
-                .await?
-            {
-                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();
+            while let Some(table) = stream.try_next().await? {
+                let mut primary_constraints = vec![];
 
-                    for (idx, column) in schema.column_schemas().iter().enumerate() {
-                        if column.is_time_index() {
-                            self.add_key_column_usage(
-                                &predicates,
-                                &schema_name,
-                                TIME_INDEX_CONSTRAINT_NAME,
-                                &catalog_name,
-                                &schema_name,
-                                &table_name,
-                                &column.name,
-                                1, //always 1 for time index
-                            );
-                        }
-                        if keys.contains(&idx) {
-                            primary_constraints.push((
-                                catalog_name.clone(),
-                                schema_name.clone(),
-                                table_name.clone(),
-                                column.name.clone(),
-                            ));
-                        }
-                        // TODO(dimbtp): foreign key constraint not supported yet
+                let table_info = table.table_info();
+                let table_name = &table_info.name;
+                let keys = &table_info.meta.primary_key_indices;
+                let schema = table.schema();
+
+                for (idx, column) in schema.column_schemas().iter().enumerate() {
+                    if column.is_time_index() {
+                        self.add_key_column_usage(
+                            &predicates,
+                            &schema_name,
+                            TIME_INDEX_CONSTRAINT_NAME,
+                            &catalog_name,
+                            &schema_name,
+                            table_name,
+                            &column.name,
+                            1, //always 1 for time index
+                        );
                     }
-                } else {
-                    unreachable!();
+                    if keys.contains(&idx) {
+                        primary_constraints.push((
+                            catalog_name.clone(),
+                            schema_name.clone(),
+                            table_name.to_string(),
+                            column.name.clone(),
+                        ));
+                    }
+                    // TODO(dimbtp): foreign key constraint not supported yet
+                }
+
+                for (i, (catalog_name, schema_name, table_name, column_name)) in
+                    primary_constraints.into_iter().enumerate()
+                {
+                    self.add_key_column_usage(
+                        &predicates,
+                        &schema_name,
+                        PRI_CONSTRAINT_NAME,
+                        &catalog_name,
+                        &schema_name,
+                        &table_name,
+                        &column_name,
+                        i as u32 + 1,
+                    );
                 }
             }
         }
 
-        for (i, (catalog_name, schema_name, table_name, column_name)) in
-            primary_constraints.into_iter().enumerate()
-        {
-            self.add_key_column_usage(
-                &predicates,
-                &schema_name,
-                PRI_CONSTRAINT_NAME,
-                &catalog_name,
-                &schema_name,
-                &table_name,
-                &column_name,
-                i as u32 + 1,
-            );
-        }
-
         self.finish()
     }
 

src/catalog/src/system_schema/information_schema/partitions.rs

@@ -44,8 +44,8 @@ use crate::error::{
     CreateRecordBatchSnafu, FindPartitionsSnafu, InternalSnafu, Result,
     UpgradeWeakCatalogManagerRefSnafu,
 };
-use crate::information_schema::{InformationTable, Predicates};
 use crate::kvbackend::KvBackendCatalogManager;
+use crate::system_schema::information_schema::{InformationTable, Predicates};
 use crate::CatalogManager;
 
 const TABLE_CATALOG: &str = "table_catalog";

src/catalog/src/system_schema/information_schema/region_peers.rs

@@ -39,8 +39,8 @@ use crate::error::{
     CreateRecordBatchSnafu, FindRegionRoutesSnafu, InternalSnafu, Result,
     UpgradeWeakCatalogManagerRefSnafu,
 };
-use crate::information_schema::{InformationTable, Predicates};
 use crate::kvbackend::KvBackendCatalogManager;
+use crate::system_schema::information_schema::{InformationTable, Predicates};
 use crate::CatalogManager;
 
 const REGION_ID: &str = "region_id";

src/catalog/src/system_schema/information_schema/schemata.rs

@@ -33,10 +33,11 @@ use store_api::storage::{ScanRequest, TableId};
 
 use super::SCHEMATA;
 use crate::error::{
-    CreateRecordBatchSnafu, InternalSnafu, Result, SchemaNotFoundSnafu, TableMetadataManagerSnafu,
+    CreateRecordBatchSnafu, InternalSnafu, Result, TableMetadataManagerSnafu,
     UpgradeWeakCatalogManagerRefSnafu,
 };
-use crate::information_schema::{utils, InformationTable, Predicates};
+use crate::system_schema::information_schema::{InformationTable, Predicates};
+use crate::system_schema::utils;
 use crate::CatalogManager;
 
 pub const CATALOG_NAME: &str = "catalog_name";
@@ -172,17 +173,14 @@ impl InformationSchemaSchemataBuilder {
 
         for schema_name in catalog_manager.schema_names(&catalog_name).await? {
             let opts = if let Some(table_metadata_manager) = &table_metadata_manager {
-                let schema_opts = table_metadata_manager
+                table_metadata_manager
                     .schema_manager()
                     .get(SchemaNameKey::new(&catalog_name, &schema_name))
                     .await
                     .context(TableMetadataManagerSnafu)?
-                    .context(SchemaNotFoundSnafu {
-                        catalog: &catalog_name,
-                        schema: &schema_name,
-                    })?;
-
-                Some(format!("{schema_opts}"))
+                    // information_schema is not available from this
+                    // table_metadata_manager and we return None
+                    .map(|schema_opts| format!("{schema_opts}"))
             } else {
                 None
             };

src/catalog/src/system_schema/information_schema/table_names.rs

@@ -43,3 +43,5 @@ pub const PARTITIONS: &str = "partitions";
 pub const REGION_PEERS: &str = "region_peers";
 pub const TABLE_CONSTRAINTS: &str = "table_constraints";
 pub const CLUSTER_INFO: &str = "cluster_info";
+pub const VIEWS: &str = "views";
+pub const FLOWS: &str = "flows";

src/catalog/src/system_schema/information_schema/tables.rs

@@ -0,0 +1,372 @@
+// 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, Weak};
+
+use arrow_schema::SchemaRef as ArrowSchemaRef;
+use common_catalog::consts::INFORMATION_SCHEMA_TABLES_TABLE_ID;
+use common_error::ext::BoxedError;
+use common_recordbatch::adapter::RecordBatchStreamAdapter;
+use common_recordbatch::{RecordBatch, SendableRecordBatchStream};
+use datafusion::execution::TaskContext;
+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::prelude::{ConcreteDataType, ScalarVectorBuilder, VectorRef};
+use datatypes::schema::{ColumnSchema, Schema, SchemaRef};
+use datatypes::value::Value;
+use datatypes::vectors::{
+    DateTimeVectorBuilder, StringVectorBuilder, UInt32VectorBuilder, UInt64VectorBuilder,
+};
+use futures::TryStreamExt;
+use snafu::{OptionExt, ResultExt};
+use store_api::storage::{ScanRequest, TableId};
+use table::metadata::{TableInfo, TableType};
+
+use super::TABLES;
+use crate::error::{
+    CreateRecordBatchSnafu, InternalSnafu, Result, UpgradeWeakCatalogManagerRefSnafu,
+};
+use crate::system_schema::information_schema::{InformationTable, Predicates};
+use crate::CatalogManager;
+
+pub const TABLE_CATALOG: &str = "table_catalog";
+pub const TABLE_SCHEMA: &str = "table_schema";
+pub const TABLE_NAME: &str = "table_name";
+pub const TABLE_TYPE: &str = "table_type";
+pub const VERSION: &str = "version";
+pub const ROW_FORMAT: &str = "row_format";
+pub const TABLE_ROWS: &str = "table_rows";
+pub const DATA_LENGTH: &str = "data_length";
+pub const INDEX_LENGTH: &str = "index_length";
+pub const MAX_DATA_LENGTH: &str = "max_data_length";
+pub const AVG_ROW_LENGTH: &str = "avg_row_length";
+pub const DATA_FREE: &str = "data_free";
+pub const AUTO_INCREMENT: &str = "auto_increment";
+pub const CREATE_TIME: &str = "create_time";
+pub const UPDATE_TIME: &str = "update_time";
+pub const CHECK_TIME: &str = "check_time";
+pub const TABLE_COLLATION: &str = "table_collation";
+pub const CHECKSUM: &str = "checksum";
+pub const CREATE_OPTIONS: &str = "create_options";
+pub const TABLE_COMMENT: &str = "table_comment";
+pub const MAX_INDEX_LENGTH: &str = "max_index_length";
+pub const TEMPORARY: &str = "temporary";
+const TABLE_ID: &str = "table_id";
+pub const ENGINE: &str = "engine";
+const INIT_CAPACITY: usize = 42;
+
+pub(super) struct InformationSchemaTables {
+    schema: SchemaRef,
+    catalog_name: String,
+    catalog_manager: Weak<dyn CatalogManager>,
+}
+
+impl InformationSchemaTables {
+    pub(super) fn new(catalog_name: String, catalog_manager: Weak<dyn CatalogManager>) -> Self {
+        Self {
+            schema: Self::schema(),
+            catalog_name,
+            catalog_manager,
+        }
+    }
+
+    pub(crate) fn schema() -> SchemaRef {
+        Arc::new(Schema::new(vec![
+            ColumnSchema::new(TABLE_CATALOG, ConcreteDataType::string_datatype(), false),
+            ColumnSchema::new(TABLE_SCHEMA, ConcreteDataType::string_datatype(), false),
+            ColumnSchema::new(TABLE_NAME, ConcreteDataType::string_datatype(), false),
+            ColumnSchema::new(TABLE_TYPE, ConcreteDataType::string_datatype(), false),
+            ColumnSchema::new(TABLE_ID, ConcreteDataType::uint32_datatype(), true),
+            ColumnSchema::new(DATA_LENGTH, ConcreteDataType::uint64_datatype(), true),
+            ColumnSchema::new(MAX_DATA_LENGTH, ConcreteDataType::uint64_datatype(), true),
+            ColumnSchema::new(INDEX_LENGTH, ConcreteDataType::uint64_datatype(), true),
+            ColumnSchema::new(MAX_INDEX_LENGTH, ConcreteDataType::uint64_datatype(), true),
+            ColumnSchema::new(AVG_ROW_LENGTH, ConcreteDataType::uint64_datatype(), true),
+            ColumnSchema::new(ENGINE, ConcreteDataType::string_datatype(), true),
+            ColumnSchema::new(VERSION, ConcreteDataType::uint64_datatype(), true),
+            ColumnSchema::new(ROW_FORMAT, ConcreteDataType::string_datatype(), true),
+            ColumnSchema::new(TABLE_ROWS, ConcreteDataType::uint64_datatype(), true),
+            ColumnSchema::new(DATA_FREE, ConcreteDataType::uint64_datatype(), true),
+            ColumnSchema::new(AUTO_INCREMENT, ConcreteDataType::uint64_datatype(), true),
+            ColumnSchema::new(CREATE_TIME, ConcreteDataType::datetime_datatype(), true),
+            ColumnSchema::new(UPDATE_TIME, ConcreteDataType::datetime_datatype(), true),
+            ColumnSchema::new(CHECK_TIME, ConcreteDataType::datetime_datatype(), true),
+            ColumnSchema::new(TABLE_COLLATION, ConcreteDataType::string_datatype(), true),
+            ColumnSchema::new(CHECKSUM, ConcreteDataType::uint64_datatype(), true),
+            ColumnSchema::new(CREATE_OPTIONS, ConcreteDataType::string_datatype(), true),
+            ColumnSchema::new(TABLE_COMMENT, ConcreteDataType::string_datatype(), true),
+            ColumnSchema::new(TEMPORARY, ConcreteDataType::string_datatype(), true),
+        ]))
+    }
+
+    fn builder(&self) -> InformationSchemaTablesBuilder {
+        InformationSchemaTablesBuilder::new(
+            self.schema.clone(),
+            self.catalog_name.clone(),
+            self.catalog_manager.clone(),
+        )
+    }
+}
+
+impl InformationTable for InformationSchemaTables {
+    fn table_id(&self) -> TableId {
+        INFORMATION_SCHEMA_TABLES_TABLE_ID
+    }
+
+    fn table_name(&self) -> &'static str {
+        TABLES
+    }
+
+    fn schema(&self) -> SchemaRef {
+        self.schema.clone()
+    }
+
+    fn to_stream(&self, request: ScanRequest) -> 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
+                    .make_tables(Some(request))
+                    .await
+                    .map(|x| x.into_df_record_batch())
+                    .map_err(|err| datafusion::error::DataFusionError::External(Box::new(err)))
+            }),
+        ));
+        Ok(Box::pin(
+            RecordBatchStreamAdapter::try_new(stream)
+                .map_err(BoxedError::new)
+                .context(InternalSnafu)?,
+        ))
+    }
+}
+
+/// Builds the `information_schema.TABLE` table row by row
+///
+/// Columns are based on <https://www.postgresql.org/docs/current/infoschema-columns.html>
+struct InformationSchemaTablesBuilder {
+    schema: SchemaRef,
+    catalog_name: String,
+    catalog_manager: Weak<dyn CatalogManager>,
+
+    catalog_names: StringVectorBuilder,
+    schema_names: StringVectorBuilder,
+    table_names: StringVectorBuilder,
+    table_types: StringVectorBuilder,
+    table_ids: UInt32VectorBuilder,
+    version: UInt64VectorBuilder,
+    row_format: StringVectorBuilder,
+    table_rows: UInt64VectorBuilder,
+    data_length: UInt64VectorBuilder,
+    max_data_length: UInt64VectorBuilder,
+    index_length: UInt64VectorBuilder,
+    avg_row_length: UInt64VectorBuilder,
+    max_index_length: UInt64VectorBuilder,
+    data_free: UInt64VectorBuilder,
+    auto_increment: UInt64VectorBuilder,
+    create_time: DateTimeVectorBuilder,
+    update_time: DateTimeVectorBuilder,
+    check_time: DateTimeVectorBuilder,
+    table_collation: StringVectorBuilder,
+    checksum: UInt64VectorBuilder,
+    create_options: StringVectorBuilder,
+    table_comment: StringVectorBuilder,
+    engines: StringVectorBuilder,
+    temporary: StringVectorBuilder,
+}
+
+impl InformationSchemaTablesBuilder {
+    fn new(
+        schema: SchemaRef,
+        catalog_name: String,
+        catalog_manager: Weak<dyn CatalogManager>,
+    ) -> Self {
+        Self {
+            schema,
+            catalog_name,
+            catalog_manager,
+            catalog_names: StringVectorBuilder::with_capacity(INIT_CAPACITY),
+            schema_names: StringVectorBuilder::with_capacity(INIT_CAPACITY),
+            table_names: StringVectorBuilder::with_capacity(INIT_CAPACITY),
+            table_types: StringVectorBuilder::with_capacity(INIT_CAPACITY),
+            table_ids: UInt32VectorBuilder::with_capacity(INIT_CAPACITY),
+            data_length: UInt64VectorBuilder::with_capacity(INIT_CAPACITY),
+            max_data_length: UInt64VectorBuilder::with_capacity(INIT_CAPACITY),
+            index_length: UInt64VectorBuilder::with_capacity(INIT_CAPACITY),
+            avg_row_length: UInt64VectorBuilder::with_capacity(INIT_CAPACITY),
+            engines: StringVectorBuilder::with_capacity(INIT_CAPACITY),
+            version: UInt64VectorBuilder::with_capacity(INIT_CAPACITY),
+            row_format: StringVectorBuilder::with_capacity(INIT_CAPACITY),
+            table_rows: UInt64VectorBuilder::with_capacity(INIT_CAPACITY),
+            max_index_length: UInt64VectorBuilder::with_capacity(INIT_CAPACITY),
+            data_free: UInt64VectorBuilder::with_capacity(INIT_CAPACITY),
+            auto_increment: UInt64VectorBuilder::with_capacity(INIT_CAPACITY),
+            create_time: DateTimeVectorBuilder::with_capacity(INIT_CAPACITY),
+            update_time: DateTimeVectorBuilder::with_capacity(INIT_CAPACITY),
+            check_time: DateTimeVectorBuilder::with_capacity(INIT_CAPACITY),
+            table_collation: StringVectorBuilder::with_capacity(INIT_CAPACITY),
+            checksum: UInt64VectorBuilder::with_capacity(INIT_CAPACITY),
+            create_options: StringVectorBuilder::with_capacity(INIT_CAPACITY),
+            table_comment: StringVectorBuilder::with_capacity(INIT_CAPACITY),
+            temporary: StringVectorBuilder::with_capacity(INIT_CAPACITY),
+        }
+    }
+
+    /// Construct the `information_schema.tables` virtual table
+    async fn make_tables(&mut self, request: Option<ScanRequest>) -> Result<RecordBatch> {
+        let catalog_name = self.catalog_name.clone();
+        let catalog_manager = self
+            .catalog_manager
+            .upgrade()
+            .context(UpgradeWeakCatalogManagerRefSnafu)?;
+        let predicates = Predicates::from_scan_request(&request);
+
+        for schema_name in catalog_manager.schema_names(&catalog_name).await? {
+            let mut stream = catalog_manager.tables(&catalog_name, &schema_name);
+
+            while let Some(table) = stream.try_next().await? {
+                let table_info = table.table_info();
+                self.add_table(
+                    &predicates,
+                    &catalog_name,
+                    &schema_name,
+                    table_info,
+                    table.table_type(),
+                );
+            }
+        }
+
+        self.finish()
+    }
+
+    #[allow(clippy::too_many_arguments)]
+    fn add_table(
+        &mut self,
+        predicates: &Predicates,
+        catalog_name: &str,
+        schema_name: &str,
+        table_info: Arc<TableInfo>,
+        table_type: TableType,
+    ) {
+        let table_name = table_info.name.as_ref();
+        let table_id = table_info.table_id();
+        let engine = table_info.meta.engine.as_ref();
+
+        let table_type_text = match table_type {
+            TableType::Base => "BASE TABLE",
+            TableType::View => "VIEW",
+            TableType::Temporary => "LOCAL TEMPORARY",
+        };
+
+        let row = [
+            (TABLE_CATALOG, &Value::from(catalog_name)),
+            (TABLE_SCHEMA, &Value::from(schema_name)),
+            (TABLE_NAME, &Value::from(table_name)),
+            (TABLE_TYPE, &Value::from(table_type_text)),
+        ];
+
+        if !predicates.eval(&row) {
+            return;
+        }
+
+        self.catalog_names.push(Some(catalog_name));
+        self.schema_names.push(Some(schema_name));
+        self.table_names.push(Some(table_name));
+        self.table_types.push(Some(table_type_text));
+        self.table_ids.push(Some(table_id));
+        // TODO(sunng87): use real data for these fields
+        self.data_length.push(Some(0));
+        self.max_data_length.push(Some(0));
+        self.index_length.push(Some(0));
+        self.avg_row_length.push(Some(0));
+        self.max_index_length.push(Some(0));
+        self.checksum.push(Some(0));
+        self.table_rows.push(Some(0));
+        self.data_free.push(Some(0));
+        self.auto_increment.push(Some(0));
+        self.row_format.push(Some("Fixed"));
+        self.table_collation.push(Some("utf8_bin"));
+        self.update_time.push(None);
+        self.check_time.push(None);
+
+        // use mariadb default table version number here
+        self.version.push(Some(11));
+        self.table_comment.push(table_info.desc.as_deref());
+        self.create_options
+            .push(Some(table_info.meta.options.to_string().as_ref()));
+        self.create_time
+            .push(Some(table_info.meta.created_on.timestamp_millis().into()));
+
+        self.temporary
+            .push(if matches!(table_type, TableType::Temporary) {
+                Some("Y")
+            } else {
+                Some("N")
+            });
+        self.engines.push(Some(engine));
+    }
+
+    fn finish(&mut self) -> Result<RecordBatch> {
+        let columns: Vec<VectorRef> = vec![
+            Arc::new(self.catalog_names.finish()),
+            Arc::new(self.schema_names.finish()),
+            Arc::new(self.table_names.finish()),
+            Arc::new(self.table_types.finish()),
+            Arc::new(self.table_ids.finish()),
+            Arc::new(self.data_length.finish()),
+            Arc::new(self.max_data_length.finish()),
+            Arc::new(self.index_length.finish()),
+            Arc::new(self.max_index_length.finish()),
+            Arc::new(self.avg_row_length.finish()),
+            Arc::new(self.engines.finish()),
+            Arc::new(self.version.finish()),
+            Arc::new(self.row_format.finish()),
+            Arc::new(self.table_rows.finish()),
+            Arc::new(self.data_free.finish()),
+            Arc::new(self.auto_increment.finish()),
+            Arc::new(self.create_time.finish()),
+            Arc::new(self.update_time.finish()),
+            Arc::new(self.check_time.finish()),
+            Arc::new(self.table_collation.finish()),
+            Arc::new(self.checksum.finish()),
+            Arc::new(self.create_options.finish()),
+            Arc::new(self.table_comment.finish()),
+            Arc::new(self.temporary.finish()),
+        ];
+        RecordBatch::new(self.schema.clone(), columns).context(CreateRecordBatchSnafu)
+    }
+}
+
+impl DfPartitionStream for InformationSchemaTables {
+    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
+                    .make_tables(None)
+                    .await
+                    .map(|x| x.into_df_record_batch())
+                    .map_err(Into::into)
+            }),
+        ))
+    }
+}

src/catalog/src/system_schema/information_schema/views.rs

@@ -15,7 +15,7 @@
 use std::sync::{Arc, Weak};
 
 use arrow_schema::SchemaRef as ArrowSchemaRef;
-use common_catalog::consts::INFORMATION_SCHEMA_TABLES_TABLE_ID;
+use common_catalog::consts::INFORMATION_SCHEMA_VIEW_TABLE_ID;
 use common_error::ext::BoxedError;
 use common_recordbatch::adapter::RecordBatchStreamAdapter;
 use common_recordbatch::{RecordBatch, SendableRecordBatchStream};
@@ -26,34 +26,41 @@ use datafusion::physical_plan::SendableRecordBatchStream as DfSendableRecordBatc
 use datatypes::prelude::{ConcreteDataType, ScalarVectorBuilder, VectorRef};
 use datatypes::schema::{ColumnSchema, Schema, SchemaRef};
 use datatypes::value::Value;
-use datatypes::vectors::{StringVectorBuilder, UInt32VectorBuilder};
+use datatypes::vectors::StringVectorBuilder;
 use futures::TryStreamExt;
 use snafu::{OptionExt, ResultExt};
 use store_api::storage::{ScanRequest, TableId};
 use table::metadata::TableType;
 
-use super::TABLES;
+use super::VIEWS;
 use crate::error::{
-    CreateRecordBatchSnafu, InternalSnafu, Result, UpgradeWeakCatalogManagerRefSnafu,
+    CastManagerSnafu, CreateRecordBatchSnafu, GetViewCacheSnafu, InternalSnafu, Result,
+    UpgradeWeakCatalogManagerRefSnafu, ViewInfoNotFoundSnafu,
 };
-use crate::information_schema::{InformationTable, Predicates};
+use crate::kvbackend::KvBackendCatalogManager;
+use crate::system_schema::information_schema::{InformationTable, Predicates};
 use crate::CatalogManager;
+const INIT_CAPACITY: usize = 42;
 
 pub const TABLE_CATALOG: &str = "table_catalog";
 pub const TABLE_SCHEMA: &str = "table_schema";
 pub const TABLE_NAME: &str = "table_name";
-pub const TABLE_TYPE: &str = "table_type";
-const TABLE_ID: &str = "table_id";
-const ENGINE: &str = "engine";
-const INIT_CAPACITY: usize = 42;
+pub const VIEW_DEFINITION: &str = "view_definition";
+pub const CHECK_OPTION: &str = "check_option";
+pub const IS_UPDATABLE: &str = "is_updatable";
+pub const DEFINER: &str = "definer";
+pub const SECURITY_TYPE: &str = "security_type";
+pub const CHARACTER_SET_CLIENT: &str = "character_set_client";
+pub const COLLATION_CONNECTION: &str = "collation_connection";
 
-pub(super) struct InformationSchemaTables {
+/// The `information_schema.views` to provides information about views in databases.
+pub(super) struct InformationSchemaViews {
     schema: SchemaRef,
     catalog_name: String,
     catalog_manager: Weak<dyn CatalogManager>,
 }
 
-impl InformationSchemaTables {
+impl InformationSchemaViews {
     pub(super) fn new(catalog_name: String, catalog_manager: Weak<dyn CatalogManager>) -> Self {
         Self {
             schema: Self::schema(),
@@ -67,14 +74,26 @@ impl InformationSchemaTables {
             ColumnSchema::new(TABLE_CATALOG, ConcreteDataType::string_datatype(), false),
             ColumnSchema::new(TABLE_SCHEMA, ConcreteDataType::string_datatype(), false),
             ColumnSchema::new(TABLE_NAME, ConcreteDataType::string_datatype(), false),
-            ColumnSchema::new(TABLE_TYPE, ConcreteDataType::string_datatype(), false),
-            ColumnSchema::new(TABLE_ID, ConcreteDataType::uint32_datatype(), true),
-            ColumnSchema::new(ENGINE, ConcreteDataType::string_datatype(), true),
+            ColumnSchema::new(VIEW_DEFINITION, ConcreteDataType::string_datatype(), false),
+            ColumnSchema::new(CHECK_OPTION, ConcreteDataType::string_datatype(), true),
+            ColumnSchema::new(IS_UPDATABLE, ConcreteDataType::string_datatype(), true),
+            ColumnSchema::new(DEFINER, ConcreteDataType::string_datatype(), true),
+            ColumnSchema::new(SECURITY_TYPE, ConcreteDataType::string_datatype(), true),
+            ColumnSchema::new(
+                CHARACTER_SET_CLIENT,
+                ConcreteDataType::string_datatype(),
+                true,
+            ),
+            ColumnSchema::new(
+                COLLATION_CONNECTION,
+                ConcreteDataType::string_datatype(),
+                true,
+            ),
         ]))
     }
 
-    fn builder(&self) -> InformationSchemaTablesBuilder {
-        InformationSchemaTablesBuilder::new(
+    fn builder(&self) -> InformationSchemaViewsBuilder {
+        InformationSchemaViewsBuilder::new(
             self.schema.clone(),
             self.catalog_name.clone(),
             self.catalog_manager.clone(),
@@ -82,13 +101,13 @@ impl InformationSchemaTables {
     }
 }
 
-impl InformationTable for InformationSchemaTables {
+impl InformationTable for InformationSchemaViews {
     fn table_id(&self) -> TableId {
-        INFORMATION_SCHEMA_TABLES_TABLE_ID
+        INFORMATION_SCHEMA_VIEW_TABLE_ID
     }
 
     fn table_name(&self) -> &'static str {
-        TABLES
+        VIEWS
     }
 
     fn schema(&self) -> SchemaRef {
@@ -102,12 +121,10 @@ impl InformationTable for InformationSchemaTables {
             schema,
             futures::stream::once(async move {
                 builder
-                    .make_tables(Some(request))
+                    .make_views(Some(request))
                     .await
                     .map(|x| x.into_df_record_batch())
-                    .map_err(|err| {
-                        datafusion::error::DataFusionError::External(format!("{err:?}").into())
-                    })
+                    .map_err(|err| datafusion::error::DataFusionError::External(Box::new(err)))
             }),
         ));
         Ok(Box::pin(
@@ -118,10 +135,10 @@ impl InformationTable for InformationSchemaTables {
     }
 }
 
-/// Builds the `information_schema.TABLE` table row by row
+/// Builds the `information_schema.VIEWS` table row by row
 ///
-/// Columns are based on <https://www.postgresql.org/docs/current/infoschema-columns.html>
-struct InformationSchemaTablesBuilder {
+/// Columns are based on <https://dev.mysql.com/doc/refman/8.4/en/information-schema-views-table.html>
+struct InformationSchemaViewsBuilder {
     schema: SchemaRef,
     catalog_name: String,
     catalog_manager: Weak<dyn CatalogManager>,
@@ -129,12 +146,16 @@ struct InformationSchemaTablesBuilder {
     catalog_names: StringVectorBuilder,
     schema_names: StringVectorBuilder,
     table_names: StringVectorBuilder,
-    table_types: StringVectorBuilder,
-    table_ids: UInt32VectorBuilder,
-    engines: StringVectorBuilder,
+    view_definitions: StringVectorBuilder,
+    check_options: StringVectorBuilder,
+    is_updatable: StringVectorBuilder,
+    definer: StringVectorBuilder,
+    security_type: StringVectorBuilder,
+    character_set_client: StringVectorBuilder,
+    collation_connection: StringVectorBuilder,
 }
 
-impl InformationSchemaTablesBuilder {
+impl InformationSchemaViewsBuilder {
     fn new(
         schema: SchemaRef,
         catalog_name: String,
@@ -147,75 +168,85 @@ impl InformationSchemaTablesBuilder {
             catalog_names: StringVectorBuilder::with_capacity(INIT_CAPACITY),
             schema_names: StringVectorBuilder::with_capacity(INIT_CAPACITY),
             table_names: StringVectorBuilder::with_capacity(INIT_CAPACITY),
-            table_types: StringVectorBuilder::with_capacity(INIT_CAPACITY),
-            table_ids: UInt32VectorBuilder::with_capacity(INIT_CAPACITY),
-            engines: StringVectorBuilder::with_capacity(INIT_CAPACITY),
+            view_definitions: StringVectorBuilder::with_capacity(INIT_CAPACITY),
+            check_options: StringVectorBuilder::with_capacity(INIT_CAPACITY),
+            is_updatable: StringVectorBuilder::with_capacity(INIT_CAPACITY),
+            definer: StringVectorBuilder::with_capacity(INIT_CAPACITY),
+            security_type: StringVectorBuilder::with_capacity(INIT_CAPACITY),
+            character_set_client: StringVectorBuilder::with_capacity(INIT_CAPACITY),
+            collation_connection: StringVectorBuilder::with_capacity(INIT_CAPACITY),
         }
     }
 
-    /// Construct the `information_schema.tables` virtual table
-    async fn make_tables(&mut self, request: Option<ScanRequest>) -> Result<RecordBatch> {
+    /// Construct the `information_schema.views` virtual table
+    async fn make_views(&mut self, request: Option<ScanRequest>) -> Result<RecordBatch> {
         let catalog_name = self.catalog_name.clone();
         let catalog_manager = self
             .catalog_manager
             .upgrade()
             .context(UpgradeWeakCatalogManagerRefSnafu)?;
         let predicates = Predicates::from_scan_request(&request);
+        let view_info_cache = catalog_manager
+            .as_any()
+            .downcast_ref::<KvBackendCatalogManager>()
+            .context(CastManagerSnafu)?
+            .view_info_cache()?;
 
         for schema_name in catalog_manager.schema_names(&catalog_name).await? {
             let mut stream = catalog_manager.tables(&catalog_name, &schema_name);
 
             while let Some(table) = stream.try_next().await? {
                 let table_info = table.table_info();
-                self.add_table(
-                    &predicates,
-                    &catalog_name,
-                    &schema_name,
-                    &table_info.name,
-                    table.table_type(),
-                    Some(table_info.ident.table_id),
-                    Some(&table_info.meta.engine),
-                );
+                if table_info.table_type == TableType::View {
+                    let view_info = view_info_cache
+                        .get(table_info.ident.table_id)
+                        .await
+                        .context(GetViewCacheSnafu)?
+                        .context(ViewInfoNotFoundSnafu {
+                            name: &table_info.name,
+                        })?;
+                    self.add_view(
+                        &predicates,
+                        &catalog_name,
+                        &schema_name,
+                        &table_info.name,
+                        &view_info.definition,
+                    )
+                }
             }
         }
 
         self.finish()
     }
 
-    #[allow(clippy::too_many_arguments)]
-    fn add_table(
+    fn add_view(
         &mut self,
         predicates: &Predicates,
         catalog_name: &str,
         schema_name: &str,
         table_name: &str,
-        table_type: TableType,
-        table_id: Option<u32>,
-        engine: Option<&str>,
+        definition: &str,
     ) {
-        let table_type = match table_type {
-            TableType::Base => "BASE TABLE",
-            TableType::View => "VIEW",
-            TableType::Temporary => "LOCAL TEMPORARY",
-        };
-
         let row = [
             (TABLE_CATALOG, &Value::from(catalog_name)),
             (TABLE_SCHEMA, &Value::from(schema_name)),
             (TABLE_NAME, &Value::from(table_name)),
-            (TABLE_TYPE, &Value::from(table_type)),
         ];
 
         if !predicates.eval(&row) {
             return;
         }
-
         self.catalog_names.push(Some(catalog_name));
         self.schema_names.push(Some(schema_name));
         self.table_names.push(Some(table_name));
-        self.table_types.push(Some(table_type));
-        self.table_ids.push(table_id);
-        self.engines.push(engine);
+        self.view_definitions.push(Some(definition));
+        self.check_options.push(None);
+        // View is not updatable, statements such UPDATE , DELETE , and INSERT are illegal and are rejected.
+        self.is_updatable.push(Some("NO"));
+        self.definer.push(None);
+        self.security_type.push(None);
+        self.character_set_client.push(Some("utf8"));
+        self.collation_connection.push(Some("utf8_bin"));
     }
 
     fn finish(&mut self) -> Result<RecordBatch> {
@@ -223,15 +254,19 @@ impl InformationSchemaTablesBuilder {
             Arc::new(self.catalog_names.finish()),
             Arc::new(self.schema_names.finish()),
             Arc::new(self.table_names.finish()),
-            Arc::new(self.table_types.finish()),
-            Arc::new(self.table_ids.finish()),
-            Arc::new(self.engines.finish()),
+            Arc::new(self.view_definitions.finish()),
+            Arc::new(self.check_options.finish()),
+            Arc::new(self.is_updatable.finish()),
+            Arc::new(self.definer.finish()),
+            Arc::new(self.security_type.finish()),
+            Arc::new(self.character_set_client.finish()),
+            Arc::new(self.collation_connection.finish()),
         ];
         RecordBatch::new(self.schema.clone(), columns).context(CreateRecordBatchSnafu)
     }
 }
 
-impl DfPartitionStream for InformationSchemaTables {
+impl DfPartitionStream for InformationSchemaViews {
     fn schema(&self) -> &ArrowSchemaRef {
         self.schema.arrow_schema()
     }
@@ -243,7 +278,7 @@ impl DfPartitionStream for InformationSchemaTables {
             schema,
             futures::stream::once(async move {
                 builder
-                    .make_tables(None)
+                    .make_views(None)
                     .await
                     .map(|x| x.into_df_record_batch())
                     .map_err(Into::into)

src/catalog/src/system_schema/memory_table.rs

@@ -12,7 +12,8 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-mod tables;
+mod table_columns;
+
 use std::sync::Arc;
 
 use arrow_schema::SchemaRef as ArrowSchemaRef;
@@ -27,22 +28,21 @@ use datatypes::schema::SchemaRef;
 use datatypes::vectors::VectorRef;
 use snafu::ResultExt;
 use store_api::storage::{ScanRequest, TableId};
-pub use tables::get_schema_columns;
 
+use super::SystemTable;
 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>,
+pub(crate) struct MemoryTable {
+    pub(crate) table_id: TableId,
+    pub(crate) table_name: &'static str,
+    pub(crate) schema: SchemaRef,
+    pub(crate) columns: Vec<VectorRef>,
 }
 
 impl MemoryTable {
     /// Creates a memory table with table id, name, schema and columns.
-    pub(super) fn new(
+    pub fn new(
         table_id: TableId,
         table_name: &'static str,
         schema: SchemaRef,
@@ -56,12 +56,54 @@ impl MemoryTable {
         }
     }
 
-    fn builder(&self) -> MemoryTableBuilder {
+    pub fn builder(&self) -> MemoryTableBuilder {
         MemoryTableBuilder::new(self.schema.clone(), self.columns.clone())
     }
 }
 
-impl InformationTable for MemoryTable {
+pub(crate) 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
+    pub 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)
+            }),
+        ))
+    }
+}
+
+impl SystemTable for MemoryTable {
     fn table_id(&self) -> TableId {
         self.table_id
     }
@@ -95,48 +137,6 @@ impl InformationTable for MemoryTable {
     }
 }
 
-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;
@@ -147,6 +147,7 @@ mod tests {
     use datatypes::vectors::StringVector;
 
     use super::*;
+    use crate::system_schema::SystemTable;
 
     #[tokio::test]
     async fn test_memory_table() {
@@ -166,8 +167,8 @@ mod tests {
         );
 
         assert_eq!(42, table.table_id());
-        assert_eq!("test", table.table_name());
-        assert_eq!(schema, InformationTable::schema(&table));
+        assert_eq!("test", table.table_name);
+        assert_eq!(schema, SystemTable::schema(&table));
 
         let stream = table.to_stream(ScanRequest::default()).unwrap();
 
@@ -196,7 +197,7 @@ mod tests {
 
         assert_eq!(42, table.table_id());
         assert_eq!("test", table.table_name());
-        assert_eq!(schema, InformationTable::schema(&table));
+        assert_eq!(schema, SystemTable::schema(&table));
 
         let stream = table.to_stream(ScanRequest::default()).unwrap();
 

src/catalog/src/system_schema/memory_table/table_columns.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.
+
+#[macro_export]
+macro_rules! memory_table_cols{
+    ([$($colname:ident),*], $t:expr) => {
+        let t = &$t;
+        $(
+            let mut $colname = Vec::with_capacity(t.len());
+        )*
+        paste::paste!{
+            for &($([<r_ $colname>]),*) in t {
+                    $(
+                        $colname.push([<r_ $colname>]);
+                    )*
+            }
+        }
+    };
+}
+
+#[cfg(test)]
+mod tests {
+
+    #[test]
+    fn test_memory_table_columns() {
+        memory_table_cols!(
+            [oid, typname, typlen],
+            [
+                (1, "String", -1),
+                (2, "Binary", -1),
+                (3, "Time", 8),
+                (4, "Datetime", 8)
+            ]
+        );
+        assert_eq!(&oid[..], &[1, 2, 3, 4]);
+        assert_eq!(&typname[..], &["String", "Binary", "Time", "Datetime"]);
+        assert_eq!(&typlen[..], &[-1, -1, 8, 8]);
+    }
+}

src/catalog/src/system_schema/pg_catalog.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.
+
+mod pg_catalog_memory_table;
+mod pg_class;
+mod pg_namespace;
+mod table_names;
+
+use std::collections::HashMap;
+use std::sync::{Arc, Weak};
+
+use common_catalog::consts::{self, PG_CATALOG_NAME};
+use datatypes::schema::ColumnSchema;
+use lazy_static::lazy_static;
+use paste::paste;
+use pg_catalog_memory_table::get_schema_columns;
+use pg_class::PGClass;
+use pg_namespace::PGNamespace;
+use table::TableRef;
+pub use table_names::*;
+
+use self::pg_namespace::oid_map::{PGNamespaceOidMap, PGNamespaceOidMapRef};
+use super::memory_table::MemoryTable;
+use super::utils::tables::u32_column;
+use super::{SystemSchemaProvider, SystemSchemaProviderInner, SystemTableRef};
+use crate::CatalogManager;
+
+lazy_static! {
+    static ref MEMORY_TABLES: &'static [&'static str] = &[table_names::PG_TYPE];
+}
+
+/// The column name for the OID column.
+/// The OID column is a unique identifier of type u32 for each object in the database.
+const OID_COLUMN_NAME: &str = "oid";
+
+fn oid_column() -> ColumnSchema {
+    u32_column(OID_COLUMN_NAME)
+}
+
+/// [`PGCatalogProvider`] is the provider for a schema named `pg_catalog`, it is not a catalog.
+pub struct PGCatalogProvider {
+    catalog_name: String,
+    catalog_manager: Weak<dyn CatalogManager>,
+    tables: HashMap<String, TableRef>,
+
+    // Workaround to store mapping of schema_name to a numeric id
+    namespace_oid_map: PGNamespaceOidMapRef,
+}
+
+impl SystemSchemaProvider for PGCatalogProvider {
+    fn tables(&self) -> &HashMap<String, TableRef> {
+        assert!(!self.tables.is_empty());
+
+        &self.tables
+    }
+}
+
+// TODO(j0hn50n133): Not sure whether to avoid duplication with `information_schema` or not.
+macro_rules! setup_memory_table {
+    ($name: expr) => {
+        paste! {
+            {
+                let (schema, columns) = get_schema_columns($name);
+                Some(Arc::new(MemoryTable::new(
+                    consts::[<PG_CATALOG_ $name  _TABLE_ID>],
+                    $name,
+                    schema,
+                    columns
+                )) as _)
+            }
+        }
+    };
+}
+
+impl PGCatalogProvider {
+    pub fn new(catalog_name: String, catalog_manager: Weak<dyn CatalogManager>) -> Self {
+        let mut provider = Self {
+            catalog_name,
+            catalog_manager,
+            tables: HashMap::new(),
+            namespace_oid_map: Arc::new(PGNamespaceOidMap::new()),
+        };
+        provider.build_tables();
+        provider
+    }
+
+    fn build_tables(&mut self) {
+        // SECURITY NOTE:
+        // Must follow the same security rules as [`InformationSchemaProvider::build_tables`].
+        let mut tables = HashMap::new();
+        // TODO(J0HN50N133): modeling the table_name as a enum type to get rid of expect/unwrap here
+        // It's safe to unwrap here because we are sure that the constants have been handle correctly inside system_table.
+        for name in MEMORY_TABLES.iter() {
+            tables.insert(name.to_string(), self.build_table(name).expect(name));
+        }
+        tables.insert(
+            PG_NAMESPACE.to_string(),
+            self.build_table(PG_NAMESPACE).expect(PG_NAMESPACE),
+        );
+        tables.insert(
+            PG_CLASS.to_string(),
+            self.build_table(PG_CLASS).expect(PG_NAMESPACE),
+        );
+        self.tables = tables;
+    }
+}
+
+impl SystemSchemaProviderInner for PGCatalogProvider {
+    fn schema_name() -> &'static str {
+        PG_CATALOG_NAME
+    }
+
+    fn system_table(&self, name: &str) -> Option<SystemTableRef> {
+        match name {
+            table_names::PG_TYPE => setup_memory_table!(PG_TYPE),
+            table_names::PG_NAMESPACE => Some(Arc::new(PGNamespace::new(
+                self.catalog_name.clone(),
+                self.catalog_manager.clone(),
+                self.namespace_oid_map.clone(),
+            ))),
+            table_names::PG_CLASS => Some(Arc::new(PGClass::new(
+                self.catalog_name.clone(),
+                self.catalog_manager.clone(),
+                self.namespace_oid_map.clone(),
+            ))),
+            _ => None,
+        }
+    }
+
+    fn catalog_name(&self) -> &str {
+        &self.catalog_name
+    }
+}

src/catalog/src/system_schema/pg_catalog/pg_catalog_memory_table.rs

@@ -0,0 +1,69 @@
+// 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 datatypes::schema::{ColumnSchema, Schema, SchemaRef};
+use datatypes::vectors::{Int16Vector, StringVector, UInt32Vector, VectorRef};
+
+use super::oid_column;
+use super::table_names::PG_TYPE;
+use crate::memory_table_cols;
+use crate::system_schema::utils::tables::{i16_column, string_column};
+
+fn pg_type_schema_columns() -> (Vec<ColumnSchema>, Vec<VectorRef>) {
+    // TODO(j0hn50n133): acquire this information from `DataType` instead of hardcoding it to avoid regression.
+    memory_table_cols!(
+        [oid, typname, typlen],
+        [
+            (1, "String", -1),
+            (2, "Binary", -1),
+            (3, "Int8", 1),
+            (4, "Int16", 2),
+            (5, "Int32", 4),
+            (6, "Int64", 8),
+            (7, "UInt8", 1),
+            (8, "UInt16", 2),
+            (9, "UInt32", 4),
+            (10, "UInt64", 8),
+            (11, "Float32", 4),
+            (12, "Float64", 8),
+            (13, "Decimal", 16),
+            (14, "Date", 4),
+            (15, "DateTime", 8),
+            (16, "Timestamp", 8),
+            (17, "Time", 8),
+            (18, "Duration", 8),
+            (19, "Interval", 16),
+            (20, "List", -1),
+        ]
+    );
+    (
+        // not quiet identical with pg, we only follow the definition in pg
+        vec![oid_column(), string_column("typname"), i16_column("typlen")],
+        vec![
+            Arc::new(UInt32Vector::from_vec(oid)), // oid
+            Arc::new(StringVector::from(typname)),
+            Arc::new(Int16Vector::from_vec(typlen)), // typlen in bytes
+        ],
+    )
+}
+
+pub(super) fn get_schema_columns(table_name: &str) -> (SchemaRef, Vec<VectorRef>) {
+    let (column_schemas, columns): (_, Vec<VectorRef>) = match table_name {
+        PG_TYPE => pg_type_schema_columns(),
+        _ => unreachable!("Unknown table in pg_catalog: {}", table_name),
+    };
+    (Arc::new(Schema::new(column_schemas)), columns)
+}

src/catalog/src/system_schema/pg_catalog/pg_class.rs

@@ -0,0 +1,263 @@
+// 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, Weak};
+
+use arrow_schema::SchemaRef as ArrowSchemaRef;
+use common_catalog::consts::PG_CATALOG_PG_CLASS_TABLE_ID;
+use common_error::ext::BoxedError;
+use common_recordbatch::adapter::RecordBatchStreamAdapter;
+use common_recordbatch::{DfSendableRecordBatchStream, RecordBatch};
+use datafusion::execution::TaskContext;
+use datafusion::physical_plan::stream::RecordBatchStreamAdapter as DfRecordBatchStreamAdapter;
+use datafusion::physical_plan::streaming::PartitionStream as DfPartitionStream;
+use datatypes::scalars::ScalarVectorBuilder;
+use datatypes::schema::{Schema, SchemaRef};
+use datatypes::value::Value;
+use datatypes::vectors::{StringVectorBuilder, UInt32VectorBuilder, VectorRef};
+use futures::TryStreamExt;
+use snafu::{OptionExt, ResultExt};
+use store_api::storage::ScanRequest;
+use table::metadata::TableType;
+
+use super::pg_namespace::oid_map::PGNamespaceOidMapRef;
+use super::{OID_COLUMN_NAME, PG_CLASS};
+use crate::error::{
+    CreateRecordBatchSnafu, InternalSnafu, Result, UpgradeWeakCatalogManagerRefSnafu,
+};
+use crate::information_schema::Predicates;
+use crate::system_schema::utils::tables::{string_column, u32_column};
+use crate::system_schema::SystemTable;
+use crate::CatalogManager;
+
+// === column name ===
+pub const RELNAME: &str = "relname";
+pub const RELNAMESPACE: &str = "relnamespace";
+pub const RELKIND: &str = "relkind";
+pub const RELOWNER: &str = "relowner";
+
+// === enum value of relkind ===
+pub const RELKIND_TABLE: &str = "r";
+pub const RELKIND_VIEW: &str = "v";
+
+/// The initial capacity of the vector builders.
+const INIT_CAPACITY: usize = 42;
+/// The dummy owner id for the namespace.
+const DUMMY_OWNER_ID: u32 = 0;
+
+/// The `pg_catalog.pg_class` table implementation.
+pub(super) struct PGClass {
+    schema: SchemaRef,
+    catalog_name: String,
+    catalog_manager: Weak<dyn CatalogManager>,
+
+    // Workaround to convert schema_name to a numeric id
+    namespace_oid_map: PGNamespaceOidMapRef,
+}
+
+impl PGClass {
+    pub(super) fn new(
+        catalog_name: String,
+        catalog_manager: Weak<dyn CatalogManager>,
+        namespace_oid_map: PGNamespaceOidMapRef,
+    ) -> Self {
+        Self {
+            schema: Self::schema(),
+            catalog_name,
+            catalog_manager,
+            namespace_oid_map,
+        }
+    }
+
+    fn schema() -> SchemaRef {
+        Arc::new(Schema::new(vec![
+            u32_column(OID_COLUMN_NAME),
+            string_column(RELNAME),
+            u32_column(RELNAMESPACE),
+            string_column(RELKIND),
+            u32_column(RELOWNER),
+        ]))
+    }
+
+    fn builder(&self) -> PGClassBuilder {
+        PGClassBuilder::new(
+            self.schema.clone(),
+            self.catalog_name.clone(),
+            self.catalog_manager.clone(),
+            self.namespace_oid_map.clone(),
+        )
+    }
+}
+
+impl SystemTable for PGClass {
+    fn table_id(&self) -> table::metadata::TableId {
+        PG_CATALOG_PG_CLASS_TABLE_ID
+    }
+
+    fn table_name(&self) -> &'static str {
+        PG_CLASS
+    }
+
+    fn schema(&self) -> SchemaRef {
+        self.schema.clone()
+    }
+
+    fn to_stream(
+        &self,
+        request: ScanRequest,
+    ) -> Result<common_recordbatch::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
+                    .make_class(Some(request))
+                    .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)?,
+        ))
+    }
+}
+
+impl DfPartitionStream for PGClass {
+    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
+                    .make_class(None)
+                    .await
+                    .map(|x| x.into_df_record_batch())
+                    .map_err(Into::into)
+            }),
+        ))
+    }
+}
+
+/// Builds the `pg_catalog.pg_class` table row by row
+/// TODO(J0HN50N133): `relowner` is always the [`DUMMY_OWNER_ID`] cuz we don't have user.
+/// Once we have user system, make it the actual owner of the table.
+struct PGClassBuilder {
+    schema: SchemaRef,
+    catalog_name: String,
+    catalog_manager: Weak<dyn CatalogManager>,
+    namespace_oid_map: PGNamespaceOidMapRef,
+
+    oid: UInt32VectorBuilder,
+    relname: StringVectorBuilder,
+    relnamespace: UInt32VectorBuilder,
+    relkind: StringVectorBuilder,
+    relowner: UInt32VectorBuilder,
+}
+
+impl PGClassBuilder {
+    fn new(
+        schema: SchemaRef,
+        catalog_name: String,
+        catalog_manager: Weak<dyn CatalogManager>,
+        namespace_oid_map: PGNamespaceOidMapRef,
+    ) -> Self {
+        Self {
+            schema,
+            catalog_name,
+            catalog_manager,
+            namespace_oid_map,
+
+            oid: UInt32VectorBuilder::with_capacity(INIT_CAPACITY),
+            relname: StringVectorBuilder::with_capacity(INIT_CAPACITY),
+            relnamespace: UInt32VectorBuilder::with_capacity(INIT_CAPACITY),
+            relkind: StringVectorBuilder::with_capacity(INIT_CAPACITY),
+            relowner: UInt32VectorBuilder::with_capacity(INIT_CAPACITY),
+        }
+    }
+
+    async fn make_class(&mut self, request: Option<ScanRequest>) -> Result<RecordBatch> {
+        let catalog_name = self.catalog_name.clone();
+        let catalog_manager = self
+            .catalog_manager
+            .upgrade()
+            .context(UpgradeWeakCatalogManagerRefSnafu)?;
+        let predicates = Predicates::from_scan_request(&request);
+        for schema_name in catalog_manager.schema_names(&catalog_name).await? {
+            let mut stream = catalog_manager.tables(&catalog_name, &schema_name);
+            while let Some(table) = stream.try_next().await? {
+                let table_info = table.table_info();
+                self.add_class(
+                    &predicates,
+                    table_info.table_id(),
+                    &schema_name,
+                    &table_info.name,
+                    if table_info.table_type == TableType::View {
+                        RELKIND_VIEW
+                    } else {
+                        RELKIND_TABLE
+                    },
+                );
+            }
+        }
+        self.finish()
+    }
+
+    fn add_class(
+        &mut self,
+        predicates: &Predicates,
+        oid: u32,
+        schema: &str,
+        table: &str,
+        kind: &str,
+    ) {
+        let namespace_oid = self.namespace_oid_map.get_oid(schema);
+        let row = [
+            (OID_COLUMN_NAME, &Value::from(oid)),
+            (RELNAMESPACE, &Value::from(schema)),
+            (RELNAME, &Value::from(table)),
+            (RELKIND, &Value::from(kind)),
+            (RELOWNER, &Value::from(DUMMY_OWNER_ID)),
+        ];
+
+        if !predicates.eval(&row) {
+            return;
+        }
+
+        self.oid.push(Some(oid));
+        self.relnamespace.push(Some(namespace_oid));
+        self.relname.push(Some(table));
+        self.relkind.push(Some(kind));
+        self.relowner.push(Some(DUMMY_OWNER_ID));
+    }
+
+    fn finish(&mut self) -> Result<RecordBatch> {
+        let columns: Vec<VectorRef> = vec![
+            Arc::new(self.oid.finish()),
+            Arc::new(self.relname.finish()),
+            Arc::new(self.relnamespace.finish()),
+            Arc::new(self.relkind.finish()),
+            Arc::new(self.relowner.finish()),
+        ];
+        RecordBatch::new(self.schema.clone(), columns).context(CreateRecordBatchSnafu)
+    }
+}

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

@@ -0,0 +1,206 @@
+// 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(super) mod oid_map;
+
+use std::sync::{Arc, Weak};
+
+use arrow_schema::SchemaRef as ArrowSchemaRef;
+use common_catalog::consts::PG_CATALOG_PG_NAMESPACE_TABLE_ID;
+use common_error::ext::BoxedError;
+use common_recordbatch::adapter::RecordBatchStreamAdapter;
+use common_recordbatch::{DfSendableRecordBatchStream, RecordBatch, SendableRecordBatchStream};
+use datafusion::execution::TaskContext;
+use datafusion::physical_plan::stream::RecordBatchStreamAdapter as DfRecordBatchStreamAdapter;
+use datafusion::physical_plan::streaming::PartitionStream as DfPartitionStream;
+use datatypes::scalars::ScalarVectorBuilder;
+use datatypes::schema::{Schema, SchemaRef};
+use datatypes::value::Value;
+use datatypes::vectors::{StringVectorBuilder, UInt32VectorBuilder, VectorRef};
+use snafu::{OptionExt, ResultExt};
+use store_api::storage::ScanRequest;
+
+use super::{PGNamespaceOidMapRef, OID_COLUMN_NAME, PG_NAMESPACE};
+use crate::error::{
+    CreateRecordBatchSnafu, InternalSnafu, Result, UpgradeWeakCatalogManagerRefSnafu,
+};
+use crate::information_schema::Predicates;
+use crate::system_schema::utils::tables::{string_column, u32_column};
+use crate::system_schema::SystemTable;
+use crate::CatalogManager;
+
+/// The `pg_catalog.pg_namespace` table implementation.
+/// namespace is a schema in greptime
+
+const NSPNAME: &str = "nspname";
+const INIT_CAPACITY: usize = 42;
+
+pub(super) struct PGNamespace {
+    schema: SchemaRef,
+    catalog_name: String,
+    catalog_manager: Weak<dyn CatalogManager>,
+
+    // Workaround to convert schema_name to a numeric id
+    oid_map: PGNamespaceOidMapRef,
+}
+
+impl PGNamespace {
+    pub(super) fn new(
+        catalog_name: String,
+        catalog_manager: Weak<dyn CatalogManager>,
+        oid_map: PGNamespaceOidMapRef,
+    ) -> Self {
+        Self {
+            schema: Self::schema(),
+            catalog_name,
+            catalog_manager,
+            oid_map,
+        }
+    }
+
+    fn schema() -> SchemaRef {
+        Arc::new(Schema::new(vec![
+            // TODO(J0HN50N133): we do not have a numeric schema id, use schema name as a workaround. Use a proper schema id once we have it.
+            u32_column(OID_COLUMN_NAME),
+            string_column(NSPNAME),
+        ]))
+    }
+
+    fn builder(&self) -> PGNamespaceBuilder {
+        PGNamespaceBuilder::new(
+            self.schema.clone(),
+            self.catalog_name.clone(),
+            self.catalog_manager.clone(),
+            self.oid_map.clone(),
+        )
+    }
+}
+
+impl SystemTable for PGNamespace {
+    fn schema(&self) -> SchemaRef {
+        self.schema.clone()
+    }
+
+    fn table_id(&self) -> table::metadata::TableId {
+        PG_CATALOG_PG_NAMESPACE_TABLE_ID
+    }
+
+    fn table_name(&self) -> &'static str {
+        PG_NAMESPACE
+    }
+
+    fn to_stream(&self, request: ScanRequest) -> 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
+                    .make_namespace(Some(request))
+                    .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)?,
+        ))
+    }
+}
+
+impl DfPartitionStream for PGNamespace {
+    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
+                    .make_namespace(None)
+                    .await
+                    .map(|x| x.into_df_record_batch())
+                    .map_err(Into::into)
+            }),
+        ))
+    }
+}
+
+/// Builds the `pg_catalog.pg_namespace` table row by row
+/// `oid` use schema name as a workaround since we don't have numeric schema id.
+/// `nspname` is the schema name.
+struct PGNamespaceBuilder {
+    schema: SchemaRef,
+    catalog_name: String,
+    catalog_manager: Weak<dyn CatalogManager>,
+    namespace_oid_map: PGNamespaceOidMapRef,
+
+    oid: UInt32VectorBuilder,
+    nspname: StringVectorBuilder,
+}
+
+impl PGNamespaceBuilder {
+    fn new(
+        schema: SchemaRef,
+        catalog_name: String,
+        catalog_manager: Weak<dyn CatalogManager>,
+        namespace_oid_map: PGNamespaceOidMapRef,
+    ) -> Self {
+        Self {
+            schema,
+            catalog_name,
+            catalog_manager,
+            namespace_oid_map,
+            oid: UInt32VectorBuilder::with_capacity(INIT_CAPACITY),
+            nspname: StringVectorBuilder::with_capacity(INIT_CAPACITY),
+        }
+    }
+
+    /// Construct the `pg_catalog.pg_namespace` virtual table
+    async fn make_namespace(&mut self, request: Option<ScanRequest>) -> Result<RecordBatch> {
+        let catalog_name = self.catalog_name.clone();
+        let catalog_manager = self
+            .catalog_manager
+            .upgrade()
+            .context(UpgradeWeakCatalogManagerRefSnafu)?;
+        let predicates = Predicates::from_scan_request(&request);
+        for schema_name in catalog_manager.schema_names(&catalog_name).await? {
+            self.add_namespace(&predicates, &schema_name);
+        }
+        self.finish()
+    }
+    fn finish(&mut self) -> Result<RecordBatch> {
+        let columns: Vec<VectorRef> =
+            vec![Arc::new(self.oid.finish()), Arc::new(self.nspname.finish())];
+        RecordBatch::new(self.schema.clone(), columns).context(CreateRecordBatchSnafu)
+    }
+
+    fn add_namespace(&mut self, predicates: &Predicates, schema_name: &str) {
+        let oid = self.namespace_oid_map.get_oid(schema_name);
+        let row = [
+            (OID_COLUMN_NAME, &Value::from(oid)),
+            (NSPNAME, &Value::from(schema_name)),
+        ];
+        if !predicates.eval(&row) {
+            return;
+        }
+        self.oid.push(Some(oid));
+        self.nspname.push(Some(schema_name));
+    }
+}

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

@@ -0,0 +1,100 @@
+// 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::hash::BuildHasher;
+use std::sync::Arc;
+
+use dashmap::DashMap;
+use rustc_hash::FxSeededState;
+
+pub type PGNamespaceOidMapRef = Arc<PGNamespaceOidMap>;
+// Workaround to convert schema_name to a numeric id,
+// remove this when we have numeric schema id in greptime
+pub struct PGNamespaceOidMap {
+    oid_map: DashMap<String, u32>,
+
+    // Rust use SipHasher by default, which provides resistance against DOS attacks.
+    // This will produce different hash value between each greptime instance. This will
+    // cause the sqlness test fail. We need a deterministic hash here to provide
+    // same oid for the same schema name with best effort and DOS attacks aren't concern here.
+    hasher: FxSeededState,
+}
+
+impl PGNamespaceOidMap {
+    pub fn new() -> Self {
+        Self {
+            oid_map: DashMap::new(),
+            hasher: FxSeededState::with_seed(0), // PLEASE DO NOT MODIFY THIS SEED VALUE!!!
+        }
+    }
+
+    fn oid_is_used(&self, oid: u32) -> bool {
+        self.oid_map.iter().any(|e| *e.value() == oid)
+    }
+
+    pub fn get_oid(&self, schema_name: &str) -> u32 {
+        if let Some(oid) = self.oid_map.get(schema_name) {
+            *oid
+        } else {
+            let mut oid = self.hasher.hash_one(schema_name) as u32;
+            while self.oid_is_used(oid) {
+                oid = self.hasher.hash_one(oid) as u32;
+            }
+            self.oid_map.insert(schema_name.to_string(), oid);
+            oid
+        }
+    }
+}
+
+#[cfg(test)]
+mod tests {
+
+    use super::*;
+
+    #[test]
+    fn oid_is_stable() {
+        let oid_map_1 = PGNamespaceOidMap::new();
+        let oid_map_2 = PGNamespaceOidMap::new();
+
+        let schema = "schema";
+        let oid = oid_map_1.get_oid(schema);
+
+        // oid keep stable in the same instance
+        assert_eq!(oid, oid_map_1.get_oid(schema));
+
+        // oid keep stable between different instances
+        assert_eq!(oid, oid_map_2.get_oid(schema));
+    }
+
+    #[test]
+    fn oid_collision() {
+        let oid_map = PGNamespaceOidMap::new();
+
+        let key1 = "3178510";
+        let key2 = "4215648";
+
+        // have collision
+        assert_eq!(
+            oid_map.hasher.hash_one(key1) as u32,
+            oid_map.hasher.hash_one(key2) as u32
+        );
+
+        // insert them into oid_map
+        let oid1 = oid_map.get_oid(key1);
+        let oid2 = oid_map.get_oid(key2);
+
+        // they should have different id
+        assert_ne!(oid1, oid2);
+    }
+}

src/catalog/src/system_schema/pg_catalog/table_names.rs

@@ -0,0 +1,18 @@
+// 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 const PG_DATABASE: &str = "pg_databases";
+pub const PG_NAMESPACE: &str = "pg_namespace";
+pub const PG_CLASS: &str = "pg_class";
+pub const PG_TYPE: &str = "pg_type";

src/catalog/src/system_schema/predicate.rs

@@ -25,7 +25,7 @@ type ColumnName = String;
 /// we only support these simple predicates currently.
 /// TODO(dennis): supports more predicate types.
 #[derive(Clone, PartialEq, Eq, Debug)]
-enum Predicate {
+pub(crate) enum Predicate {
     Eq(ColumnName, Value),
     Like(ColumnName, String, bool),
     NotEq(ColumnName, Value),

src/catalog/src/system_schema/utils.rs

@@ -12,6 +12,8 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+pub mod tables;
+
 use std::sync::{Arc, Weak};
 
 use common_config::Mode;

src/catalog/src/system_schema/utils/tables.rs

@@ -0,0 +1,79 @@
+// 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 datatypes::prelude::ConcreteDataType;
+use datatypes::schema::ColumnSchema;
+
+pub fn string_columns(names: &[&'static str]) -> Vec<ColumnSchema> {
+    names.iter().map(|name| string_column(name)).collect()
+}
+
+pub fn string_column(name: &str) -> ColumnSchema {
+    ColumnSchema::new(
+        str::to_lowercase(name),
+        ConcreteDataType::string_datatype(),
+        false,
+    )
+}
+
+pub fn u32_column(name: &str) -> ColumnSchema {
+    ColumnSchema::new(
+        str::to_lowercase(name),
+        ConcreteDataType::uint32_datatype(),
+        false,
+    )
+}
+
+pub fn i16_column(name: &str) -> ColumnSchema {
+    ColumnSchema::new(
+        str::to_lowercase(name),
+        ConcreteDataType::int16_datatype(),
+        false,
+    )
+}
+
+pub fn bigint_column(name: &str) -> ColumnSchema {
+    ColumnSchema::new(
+        str::to_lowercase(name),
+        ConcreteDataType::int64_datatype(),
+        false,
+    )
+}
+
+pub fn datetime_column(name: &str) -> ColumnSchema {
+    ColumnSchema::new(
+        str::to_lowercase(name),
+        ConcreteDataType::datetime_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/table_source.rs

@@ -17,21 +17,24 @@ use std::sync::Arc;
 
 use bytes::Bytes;
 use common_catalog::format_full_table_name;
-use common_query::logical_plan::SubstraitPlanDecoderRef;
+use common_query::logical_plan::{rename_logical_plan_columns, SubstraitPlanDecoderRef};
 use datafusion::common::{ResolvedTableReference, TableReference};
 use datafusion::datasource::view::ViewTable;
 use datafusion::datasource::{provider_as_source, TableProvider};
 use datafusion::logical_expr::TableSource;
+use itertools::Itertools;
 use session::context::QueryContext;
 use snafu::{ensure, OptionExt, ResultExt};
 use table::metadata::TableType;
 use table::table::adapter::DfTableProviderAdapter;
 mod dummy_catalog;
 use dummy_catalog::DummyCatalogList;
+use table::TableRef;
 
 use crate::error::{
-    CastManagerSnafu, DatafusionSnafu, DecodePlanSnafu, GetViewCacheSnafu, QueryAccessDeniedSnafu,
-    Result, TableNotExistSnafu, ViewInfoNotFoundSnafu,
+    CastManagerSnafu, DatafusionSnafu, DecodePlanSnafu, GetViewCacheSnafu, ProjectViewColumnsSnafu,
+    QueryAccessDeniedSnafu, Result, TableNotExistSnafu, ViewInfoNotFoundSnafu,
+    ViewPlanColumnsChangedSnafu,
 };
 use crate::kvbackend::KvBackendCatalogManager;
 use crate::CatalogManagerRef;
@@ -43,6 +46,7 @@ pub struct DfTableSourceProvider {
     default_catalog: String,
     default_schema: String,
     plan_decoder: SubstraitPlanDecoderRef,
+    enable_ident_normalization: bool,
 }
 
 impl DfTableSourceProvider {
@@ -51,14 +55,16 @@ impl DfTableSourceProvider {
         disallow_cross_catalog_query: bool,
         query_ctx: &QueryContext,
         plan_decoder: SubstraitPlanDecoderRef,
+        enable_ident_normalization: bool,
     ) -> Self {
         Self {
             catalog_manager,
             disallow_cross_catalog_query,
             resolved_tables: HashMap::new(),
             default_catalog: query_ctx.current_catalog().to_owned(),
-            default_schema: query_ctx.current_schema().to_owned(),
+            default_schema: query_ctx.current_schema(),
             plan_decoder,
+            enable_ident_normalization,
         }
     }
 
@@ -108,32 +114,7 @@ impl DfTableSourceProvider {
             })?;
 
         let provider: Arc<dyn TableProvider> = if table.table_info().table_type == TableType::View {
-            let catalog_manager = self
-                .catalog_manager
-                .as_any()
-                .downcast_ref::<KvBackendCatalogManager>()
-                .context(CastManagerSnafu)?;
-
-            let view_info = catalog_manager
-                .view_info_cache()?
-                .get(table.table_info().ident.table_id)
-                .await
-                .context(GetViewCacheSnafu)?
-                .context(ViewInfoNotFoundSnafu {
-                    name: &table.table_info().name,
-                })?;
-
-            // Build the catalog list provider for deserialization.
-            let catalog_list = Arc::new(DummyCatalogList::new(self.catalog_manager.clone()));
-            let logical_plan = self
-                .plan_decoder
-                .decode(Bytes::from(view_info.view_info.clone()), catalog_list, true)
-                .await
-                .context(DecodePlanSnafu {
-                    name: &table.table_info().name,
-                })?;
-
-            Arc::new(ViewTable::try_new(logical_plan, None).context(DatafusionSnafu)?)
+            self.create_view_provider(&table).await?
         } else {
             Arc::new(DfTableProviderAdapter::new(table))
         };
@@ -143,6 +124,80 @@ impl DfTableSourceProvider {
         let _ = self.resolved_tables.insert(resolved_name, source.clone());
         Ok(source)
     }
+
+    async fn create_view_provider(&self, table: &TableRef) -> Result<Arc<dyn TableProvider>> {
+        let catalog_manager = self
+            .catalog_manager
+            .as_any()
+            .downcast_ref::<KvBackendCatalogManager>()
+            .context(CastManagerSnafu)?;
+
+        let view_info = catalog_manager
+            .view_info_cache()?
+            .get(table.table_info().ident.table_id)
+            .await
+            .context(GetViewCacheSnafu)?
+            .context(ViewInfoNotFoundSnafu {
+                name: &table.table_info().name,
+            })?;
+
+        // Build the catalog list provider for deserialization.
+        let catalog_list = Arc::new(DummyCatalogList::new(self.catalog_manager.clone()));
+        let logical_plan = self
+            .plan_decoder
+            .decode(Bytes::from(view_info.view_info.clone()), catalog_list, true)
+            .await
+            .context(DecodePlanSnafu {
+                name: &table.table_info().name,
+            })?;
+
+        let columns: Vec<_> = view_info.columns.iter().map(|c| c.as_str()).collect();
+
+        let original_plan_columns: Vec<_> =
+            view_info.plan_columns.iter().map(|c| c.as_str()).collect();
+
+        let plan_columns: Vec<_> = logical_plan
+            .schema()
+            .columns()
+            .into_iter()
+            .map(|c| c.name)
+            .collect();
+
+        // Only check columns number, because substrait doesn't include aliases currently.
+        // See https://github.com/apache/datafusion/issues/10815#issuecomment-2158666881
+        // and https://github.com/apache/datafusion/issues/6489
+        // TODO(dennis): check column names
+        ensure!(
+            original_plan_columns.len() == plan_columns.len(),
+            ViewPlanColumnsChangedSnafu {
+                origin_names: original_plan_columns.iter().join(","),
+                actual_names: plan_columns.iter().join(","),
+            }
+        );
+
+        // We have to do `columns` projection here, because
+        // substrait doesn't include aliases neither for tables nor for columns:
+        // https://github.com/apache/datafusion/issues/10815#issuecomment-2158666881
+        let logical_plan = if !columns.is_empty() {
+            rename_logical_plan_columns(
+                self.enable_ident_normalization,
+                logical_plan,
+                plan_columns
+                    .iter()
+                    .map(|c| c.as_str())
+                    .zip(columns.into_iter())
+                    .collect(),
+            )
+            .context(ProjectViewColumnsSnafu)?
+        } else {
+            logical_plan
+        };
+
+        Ok(Arc::new(
+            ViewTable::try_new(logical_plan, Some(view_info.definition.to_string()))
+                .context(DatafusionSnafu)?,
+        ))
+    }
 }
 
 #[cfg(test)]
@@ -162,6 +217,7 @@ mod tests {
             true,
             query_ctx,
             DummyDecoder::arc(),
+            true,
         );
 
         let table_ref = TableReference::bare("table_name");
@@ -277,12 +333,19 @@ mod tests {
         let logical_plan = vec![1, 2, 3];
         // Create view metadata
         table_metadata_manager
-            .create_view_metadata(view_info.clone().into(), logical_plan, HashSet::new())
+            .create_view_metadata(
+                view_info.clone().into(),
+                logical_plan,
+                HashSet::new(),
+                vec!["a".to_string(), "b".to_string()],
+                vec!["id".to_string(), "name".to_string()],
+                "definition".to_string(),
+            )
             .await
             .unwrap();
 
         let mut table_provider =
-            DfTableSourceProvider::new(catalog_manager, true, query_ctx, MockDecoder::arc());
+            DfTableSourceProvider::new(catalog_manager, true, query_ctx, MockDecoder::arc(), true);
 
         // View not found
         let table_ref = TableReference::bare("not_exists_view");
@@ -290,6 +353,12 @@ mod tests {
 
         let table_ref = TableReference::bare(view_info.name);
         let source = table_provider.resolve_table(table_ref).await.unwrap();
-        assert_eq!(*source.get_logical_plan().unwrap(), mock_plan());
+        assert_eq!(
+            r#"
+Projection: person.id AS a, person.name AS b
+  Filter: person.id > Int32(500)
+    TableScan: person"#,
+            format!("\n{:?}", source.get_logical_plan().unwrap())
+        );
     }
 }

src/client/Cargo.toml

@@ -49,4 +49,4 @@ tracing-subscriber = { version = "0.3", features = ["env-filter"] }
 
 [dev-dependencies.substrait_proto]
 package = "substrait"
-version = "0.17"
+version = "0.37"

src/client/src/database.rs

@@ -33,9 +33,12 @@ use common_telemetry::tracing_context::W3cTrace;
 use futures_util::StreamExt;
 use prost::Message;
 use snafu::{ensure, ResultExt};
+use tonic::metadata::AsciiMetadataKey;
 use tonic::transport::Channel;
 
-use crate::error::{ConvertFlightDataSnafu, Error, IllegalFlightMessagesSnafu, ServerSnafu};
+use crate::error::{
+    ConvertFlightDataSnafu, Error, IllegalFlightMessagesSnafu, InvalidAsciiSnafu, ServerSnafu,
+};
 use crate::{from_grpc_response, Client, Result};
 
 #[derive(Clone, Debug, Default)]
@@ -130,6 +133,36 @@ impl Database {
         self.handle(Request::Inserts(requests)).await
     }
 
+    pub async fn insert_with_hints(
+        &self,
+        requests: InsertRequests,
+        hints: &[(&str, &str)],
+    ) -> Result<u32> {
+        let mut client = make_database_client(&self.client)?.inner;
+        let request = self.to_rpc_request(Request::Inserts(requests));
+
+        let mut request = tonic::Request::new(request);
+        let metadata = request.metadata_mut();
+        for (key, value) in hints {
+            let key = AsciiMetadataKey::from_bytes(format!("x-greptime-hint-{}", key).as_bytes())
+                .map_err(|_| {
+                InvalidAsciiSnafu {
+                    value: key.to_string(),
+                }
+                .build()
+            })?;
+            let value = value.parse().map_err(|_| {
+                InvalidAsciiSnafu {
+                    value: value.to_string(),
+                }
+                .build()
+            })?;
+            metadata.insert(key, value);
+        }
+        let response = client.handle(request).await?.into_inner();
+        from_grpc_response(response)
+    }
+
     async fn handle(&self, request: Request) -> Result<u32> {
         let mut client = make_database_client(&self.client)?.inner;
         let request = self.to_rpc_request(request);

src/client/src/error.rs

@@ -53,13 +53,6 @@ pub enum Error {
         source: common_grpc::Error,
     },
 
-    #[snafu(display("Column datatype error"))]
-    ColumnDataType {
-        #[snafu(implicit)]
-        location: Location,
-        source: api::error::Error,
-    },
-
     #[snafu(display("Illegal GRPC client state: {}", err_msg))]
     IllegalGrpcClientState {
         err_msg: String,
@@ -129,6 +122,13 @@ pub enum Error {
         #[snafu(implicit)]
         location: Location,
     },
+
+    #[snafu(display("Failed to parse ascii string: {}", value))]
+    InvalidAscii {
+        value: String,
+        #[snafu(implicit)]
+        location: Location,
+    },
 }
 
 pub type Result<T> = std::result::Result<T, Error>;
@@ -137,7 +137,6 @@ impl ErrorExt for Error {
     fn status_code(&self) -> StatusCode {
         match self {
             Error::IllegalFlightMessages { .. }
-            | Error::ColumnDataType { .. }
             | Error::MissingField { .. }
             | Error::IllegalDatabaseResponse { .. }
             | Error::ClientStreaming { .. } => StatusCode::Internal,
@@ -151,6 +150,8 @@ impl ErrorExt for Error {
             | Error::ConvertFlightData { source, .. }
             | Error::CreateTlsChannel { source, .. } => source.status_code(),
             Error::IllegalGrpcClientState { .. } => StatusCode::Unexpected,
+
+            Error::InvalidAscii { .. } => StatusCode::InvalidArguments,
         }
     }
 

src/client/src/flow.rs

@@ -16,7 +16,7 @@ use api::v1::flow::{FlowRequest, FlowResponse};
 use api::v1::region::InsertRequests;
 use common_error::ext::BoxedError;
 use common_meta::node_manager::Flownode;
-use snafu::{location, Location, ResultExt};
+use snafu::{location, ResultExt};
 
 use crate::error::Result;
 use crate::Client;

src/client/src/region.rs

@@ -33,7 +33,7 @@ use common_telemetry::error;
 use common_telemetry::tracing_context::TracingContext;
 use prost::Message;
 use query::query_engine::DefaultSerializer;
-use snafu::{location, Location, OptionExt, ResultExt};
+use snafu::{location, OptionExt, ResultExt};
 use substrait::{DFLogicalSubstraitConvertor, SubstraitPlan};
 use tokio_stream::StreamExt;
 

src/cmd/Cargo.toml

@@ -10,7 +10,9 @@ name = "greptime"
 path = "src/bin/greptime.rs"
 
 [features]
+default = ["python"]
 tokio-console = ["common-telemetry/tokio-console"]
+python = ["frontend/python"]
 
 [lints]
 workspace = true
@@ -47,7 +49,7 @@ either = "1.8"
 etcd-client.workspace = true
 file-engine.workspace = true
 flow.workspace = true
-frontend.workspace = true
+frontend = { workspace = true, default-features = false }
 futures.workspace = true
 human-panic = "1.2.2"
 lazy_static.workspace = true
@@ -74,6 +76,7 @@ substrait.workspace = true
 table.workspace = true
 tokio.workspace = true
 toml.workspace = true
+tonic.workspace = true
 tracing-appender = "0.2"
 
 [target.'cfg(not(windows))'.dependencies]
@@ -82,12 +85,10 @@ tikv-jemallocator = "0.5"
 [dev-dependencies]
 client = { workspace = true, features = ["testing"] }
 common-test-util.workspace = true
+common-version.workspace = true
 serde.workspace = true
 temp-env = "0.3"
 tempfile.workspace = true
 
 [target.'cfg(not(windows))'.dev-dependencies]
 rexpect = "0.5"
-
-[build-dependencies]
-common-version.workspace = true

src/cmd/build.rs

@@ -1,20 +0,0 @@
-// Copyright 2023 Greptime Team
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-
-fn main() {
-    // Trigger this script if the git branch/commit changes
-    println!("cargo:rerun-if-changed=.git/refs/heads");
-
-    common_version::setup_build_info();
-}

src/cmd/src/bin/greptime.rs

@@ -17,11 +17,11 @@
 use clap::{Parser, Subcommand};
 use cmd::error::Result;
 use cmd::options::GlobalOptions;
-use cmd::{cli, datanode, frontend, metasrv, standalone, App};
+use cmd::{cli, datanode, flownode, frontend, metasrv, standalone, App};
 use common_version::version;
 
 #[derive(Parser)]
-#[command(name = "greptime", author, version, long_version = version!(), about)]
+#[command(name = "greptime", author, version, long_version = version(), about)]
 #[command(propagate_version = true)]
 pub(crate) struct Command {
     #[clap(subcommand)]
@@ -37,6 +37,10 @@ enum SubCommand {
     #[clap(name = "datanode")]
     Datanode(datanode::Command),
 
+    /// Start flownode service.
+    #[clap(name = "flownode")]
+    Flownode(flownode::Command),
+
     /// Start frontend service.
     #[clap(name = "frontend")]
     Frontend(frontend::Command),
@@ -58,8 +62,37 @@ enum SubCommand {
 #[global_allocator]
 static ALLOC: tikv_jemallocator::Jemalloc = tikv_jemallocator::Jemalloc;
 
+#[cfg(debug_assertions)]
+fn main() -> Result<()> {
+    use snafu::ResultExt;
+    // Set the stack size to 8MB for the thread so it wouldn't overflow on large stack usage in debug mode
+    // see https://github.com/GreptimeTeam/greptimedb/pull/4317
+    // and https://github.com/rust-lang/rust/issues/34283
+    std::thread::Builder::new()
+        .name("main_spawn".to_string())
+        .stack_size(8 * 1024 * 1024)
+        .spawn(|| {
+            {
+                tokio::runtime::Builder::new_multi_thread()
+                    .thread_stack_size(8 * 1024 * 1024)
+                    .enable_all()
+                    .build()
+                    .expect("Failed building the Runtime")
+                    .block_on(main_body())
+            }
+        })
+        .context(cmd::error::SpawnThreadSnafu)?
+        .join()
+        .expect("Couldn't join on the associated thread")
+}
+
+#[cfg(not(debug_assertions))]
 #[tokio::main]
 async fn main() -> Result<()> {
+    main_body().await
+}
+
+async fn main_body() -> Result<()> {
     setup_human_panic();
     start(Command::parse()).await
 }
@@ -72,6 +105,12 @@ async fn start(cli: Command) -> Result<()> {
                 .run()
                 .await
         }
+        SubCommand::Flownode(cmd) => {
+            cmd.build(cmd.load_options(&cli.global_options)?)
+                .await?
+                .run()
+                .await
+        }
         SubCommand::Frontend(cmd) => {
             cmd.build(cmd.load_options(&cli.global_options)?)
                 .await?

src/cmd/src/cli/export.rs

@@ -21,11 +21,12 @@ use base64::engine::general_purpose;
 use base64::Engine;
 use clap::{Parser, ValueEnum};
 use client::DEFAULT_SCHEMA_NAME;
-use common_telemetry::{debug, error, info, warn};
+use common_catalog::consts::DEFAULT_CATALOG_NAME;
+use common_telemetry::{debug, error, info};
 use serde_json::Value;
 use servers::http::greptime_result_v1::GreptimedbV1Response;
 use servers::http::GreptimeQueryOutput;
-use snafu::{OptionExt, ResultExt};
+use snafu::ResultExt;
 use tokio::fs::File;
 use tokio::io::{AsyncWriteExt, BufWriter};
 use tokio::sync::Semaphore;
@@ -34,19 +35,20 @@ use tracing_appender::non_blocking::WorkerGuard;
 
 use crate::cli::{Instance, Tool};
 use crate::error::{
-    EmptyResultSnafu, Error, FileIoSnafu, HttpQuerySqlSnafu, InvalidDatabaseNameSnafu, Result,
-    SerdeJsonSnafu,
+    EmptyResultSnafu, Error, FileIoSnafu, HttpQuerySqlSnafu, Result, SerdeJsonSnafu,
 };
 
 type TableReference = (String, String, String);
 
 #[derive(Debug, Default, Clone, ValueEnum)]
 enum ExportTarget {
-    /// Corresponding to `SHOW CREATE TABLE`
+    /// Export all table schemas, corresponding to `SHOW CREATE TABLE`.
+    Schema,
+    /// Export all table data, corresponding to `COPY DATABASE TO`.
+    Data,
+    /// Export all table schemas and data at once.
     #[default]
-    CreateTable,
-    /// Corresponding to `EXPORT TABLE`
-    TableData,
+    All,
 }
 
 #[derive(Debug, Default, Parser)]
@@ -72,10 +74,20 @@ pub struct ExportCommand {
     max_retry: usize,
 
     /// Things to export
-    #[clap(long, short = 't', value_enum)]
+    #[clap(long, short = 't', value_enum, default_value = "all")]
     target: ExportTarget,
 
-    /// basic authentication for connecting to the server
+    /// A half-open time range: [start_time, end_time).
+    /// The start of the time range (time-index column) for data export.
+    #[clap(long)]
+    start_time: Option<String>,
+
+    /// A half-open time range: [start_time, end_time).
+    /// The end of the time range (time-index column) for data export.
+    #[clap(long)]
+    end_time: Option<String>,
+
+    /// The basic authentication for connecting to the server
     #[clap(long)]
     auth_basic: Option<String>,
 }
@@ -99,6 +111,8 @@ impl ExportCommand {
                 output_dir: self.output_dir.clone(),
                 parallelism: self.export_jobs,
                 target: self.target.clone(),
+                start_time: self.start_time.clone(),
+                end_time: self.end_time.clone(),
                 auth_header,
             }),
             guard,
@@ -113,6 +127,8 @@ pub struct Export {
     output_dir: String,
     parallelism: usize,
     target: ExportTarget,
+    start_time: Option<String>,
+    end_time: Option<String>,
     auth_header: Option<String>,
 }
 
@@ -161,13 +177,13 @@ impl Export {
         if let Some(schema) = &self.schema {
             Ok(vec![(self.catalog.clone(), schema.clone())])
         } else {
-            let result = self.sql("show databases").await?;
+            let result = self.sql("SHOW DATABASES").await?;
             let Some(records) = result else {
                 EmptyResultSnafu.fail()?
             };
             let mut result = Vec::with_capacity(records.len());
             for value in records {
-                let serde_json::Value::String(schema) = &value[0] else {
+                let Value::String(schema) = &value[0] else {
                     unreachable!()
                 };
                 if schema == common_catalog::consts::INFORMATION_SCHEMA_NAME {
@@ -188,9 +204,11 @@ impl Export {
     ) -> Result<(Vec<TableReference>, Vec<TableReference>)> {
         // Puts all metric table first
         let sql = format!(
-            "select table_catalog, table_schema, table_name from \
-            information_schema.columns where column_name = '__tsid' \
-            and table_catalog = \'{catalog}\' and table_schema = \'{schema}\'"
+            "SELECT table_catalog, table_schema, table_name \
+            FROM information_schema.columns \
+            WHERE column_name = '__tsid' \
+                and table_catalog = \'{catalog}\' \
+                and table_schema = \'{schema}\'"
         );
         let result = self.sql(&sql).await?;
         let Some(records) = result else {
@@ -210,9 +228,11 @@ impl Export {
 
         // TODO: SQL injection hurts
         let sql = format!(
-            "select table_catalog, table_schema, table_name from \
-            information_schema.tables where table_type = \'BASE TABLE\' \
-            and table_catalog = \'{catalog}\' and table_schema = \'{schema}\'",
+            "SELECT table_catalog, table_schema, table_name \
+            FROM information_schema.tables \
+            WHERE table_type = \'BASE TABLE\' \
+                and table_catalog = \'{catalog}\' \
+                and table_schema = \'{schema}\'",
         );
         let result = self.sql(&sql).await?;
         let Some(records) = result else {
@@ -249,14 +269,14 @@ impl Export {
 
     async fn show_create_table(&self, catalog: &str, schema: &str, table: &str) -> Result<String> {
         let sql = format!(
-            r#"show create table "{}"."{}"."{}""#,
+            r#"SHOW CREATE TABLE "{}"."{}"."{}""#,
             catalog, schema, table
         );
         let result = self.sql(&sql).await?;
         let Some(records) = result else {
             EmptyResultSnafu.fail()?
         };
-        let serde_json::Value::String(create_table) = &records[0][1] else {
+        let Value::String(create_table) = &records[0][1] else {
             unreachable!()
         };
 
@@ -276,11 +296,13 @@ impl Export {
                 let (metric_physical_tables, remaining_tables) =
                     self.get_table_list(&catalog, &schema).await?;
                 let table_count = metric_physical_tables.len() + remaining_tables.len();
-                tokio::fs::create_dir_all(&self.output_dir)
+                let output_dir = Path::new(&self.output_dir)
+                    .join(&catalog)
+                    .join(format!("{schema}/"));
+                tokio::fs::create_dir_all(&output_dir)
                     .await
                     .context(FileIoSnafu)?;
-                let output_file =
-                    Path::new(&self.output_dir).join(format!("{catalog}-{schema}.sql"));
+                let output_file = Path::new(&output_dir).join("create_tables.sql");
                 let mut file = File::create(output_file).await.context(FileIoSnafu)?;
                 for (c, s, t) in metric_physical_tables.into_iter().chain(remaining_tables) {
                     match self.show_create_table(&c, &s, &t).await {
@@ -294,7 +316,12 @@ impl Export {
                         }
                     }
                 }
-                info!("finished exporting {catalog}.{schema} with {table_count} tables",);
+
+                info!(
+                    "Finished exporting {catalog}.{schema} with {table_count} table schemas to path: {}",
+                    output_dir.to_string_lossy()
+                );
+
                 Ok::<(), Error>(())
             });
         }
@@ -317,7 +344,7 @@ impl Export {
         Ok(())
     }
 
-    async fn export_table_data(&self) -> Result<()> {
+    async fn export_database_data(&self) -> Result<()> {
         let timer = Instant::now();
         let semaphore = Arc::new(Semaphore::new(self.parallelism));
         let db_names = self.iter_db_names().await?;
@@ -327,70 +354,66 @@ impl Export {
             let semaphore_moved = semaphore.clone();
             tasks.push(async move {
                 let _permit = semaphore_moved.acquire().await.unwrap();
-                tokio::fs::create_dir_all(&self.output_dir)
+                let output_dir = Path::new(&self.output_dir)
+                    .join(&catalog)
+                    .join(format!("{schema}/"));
+                tokio::fs::create_dir_all(&output_dir)
                     .await
                     .context(FileIoSnafu)?;
-                let output_dir = Path::new(&self.output_dir).join(format!("{catalog}-{schema}/"));
-                // Ignores metric physical tables
-                let (metrics_tables, table_list) = self.get_table_list(&catalog, &schema).await?;
-                for (_, _, table_name) in metrics_tables {
-                    warn!("Ignores metric physical table: {table_name}");
-                }
-                for (catalog_name, schema_name, table_name) in table_list {
-                    // copy table to
-                    let sql = format!(
-                        r#"Copy "{}"."{}"."{}" TO '{}{}.parquet' WITH (format='parquet');"#,
-                        catalog_name,
-                        schema_name,
-                        table_name,
-                        output_dir.to_str().unwrap(),
-                        table_name,
-                    );
-                    info!("Executing sql: {sql}");
-                    self.sql(&sql).await?;
-                }
-                info!("Finished exporting {catalog}.{schema} data");
 
-                // export copy from sql
-                let dir_filenames = match output_dir.read_dir() {
-                    Ok(dir) => dir,
-                    Err(_) => {
-                        warn!("empty database {catalog}.{schema}");
-                        return Ok(());
+                let with_options = match (&self.start_time, &self.end_time) {
+                    (Some(start_time), Some(end_time)) => {
+                        format!(
+                            "WITH (FORMAT='parquet', start_time='{}', end_time='{}')",
+                            start_time, end_time
+                        )
                     }
+                    (Some(start_time), None) => {
+                        format!("WITH (FORMAT='parquet', start_time='{}')", start_time)
+                    }
+                    (None, Some(end_time)) => {
+                        format!("WITH (FORMAT='parquet', end_time='{}')", end_time)
+                    }
+                    (None, None) => "WITH (FORMAT='parquet')".to_string(),
                 };
 
-                let copy_from_file =
-                    Path::new(&self.output_dir).join(format!("{catalog}-{schema}_copy_from.sql"));
+                let sql = format!(
+                    r#"COPY DATABASE "{}"."{}" TO '{}' {};"#,
+                    catalog,
+                    schema,
+                    output_dir.to_str().unwrap(),
+                    with_options
+                );
+
+                info!("Executing sql: {sql}");
+
+                self.sql(&sql).await?;
+
+                info!(
+                    "Finished exporting {catalog}.{schema} data into path: {}",
+                    output_dir.to_string_lossy()
+                );
+
+                // The export copy from sql
+                let copy_from_file = output_dir.join("copy_from.sql");
                 let mut writer =
                     BufWriter::new(File::create(copy_from_file).await.context(FileIoSnafu)?);
-
-                for table_file in dir_filenames {
-                    let table_file = table_file.unwrap();
-                    let table_name = table_file
-                        .file_name()
-                        .into_string()
-                        .unwrap()
-                        .replace(".parquet", "");
-
-                    writer
-                        .write(
-                            format!(
-                                "copy {} from '{}' with (format='parquet');\n",
-                                table_name,
-                                table_file.path().to_str().unwrap()
-                            )
-                            .as_bytes(),
-                        )
-                        .await
-                        .context(FileIoSnafu)?;
-                }
+                let copy_database_from_sql = format!(
+                    r#"COPY DATABASE "{}"."{}" FROM '{}' WITH (FORMAT='parquet');"#,
+                    catalog,
+                    schema,
+                    output_dir.to_str().unwrap()
+                );
+                writer
+                    .write(copy_database_from_sql.as_bytes())
+                    .await
+                    .context(FileIoSnafu)?;
                 writer.flush().await.context(FileIoSnafu)?;
 
-                info!("finished exporting {catalog}.{schema} copy_from.sql");
+                info!("Finished exporting {catalog}.{schema} copy_from.sql");
 
                 Ok::<(), Error>(())
-            });
+            })
         }
 
         let success = futures::future::join_all(tasks)
@@ -399,35 +422,41 @@ impl Export {
             .filter(|r| match r {
                 Ok(_) => true,
                 Err(e) => {
-                    error!(e; "export job failed");
+                    error!(e; "export database job failed");
                     false
                 }
             })
             .count();
         let elapsed = timer.elapsed();
+
         info!("Success {success}/{db_count} jobs, costs: {:?}", elapsed);
 
         Ok(())
     }
 }
 
+#[allow(deprecated)]
 #[async_trait]
 impl Tool for Export {
     async fn do_work(&self) -> Result<()> {
         match self.target {
-            ExportTarget::CreateTable => self.export_create_table().await,
-            ExportTarget::TableData => self.export_table_data().await,
+            ExportTarget::Schema => self.export_create_table().await,
+            ExportTarget::Data => self.export_database_data().await,
+            ExportTarget::All => {
+                self.export_create_table().await?;
+                self.export_database_data().await
+            }
         }
     }
 }
 
 /// Split at `-`.
 fn split_database(database: &str) -> Result<(String, Option<String>)> {
-    let (catalog, schema) = database
-        .split_once('-')
-        .with_context(|| InvalidDatabaseNameSnafu {
-            database: database.to_string(),
-        })?;
+    let (catalog, schema) = match database.split_once('-') {
+        Some((catalog, schema)) => (catalog, schema),
+        None => (DEFAULT_CATALOG_NAME, database),
+    };
+
     if schema == "*" {
         Ok((catalog.to_string(), None))
     } else {
@@ -442,10 +471,26 @@ mod tests {
     use common_catalog::consts::{DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME};
     use common_telemetry::logging::LoggingOptions;
 
+    use crate::cli::export::split_database;
     use crate::error::Result as CmdResult;
     use crate::options::GlobalOptions;
     use crate::{cli, standalone, App};
 
+    #[test]
+    fn test_split_database() {
+        let result = split_database("catalog-schema").unwrap();
+        assert_eq!(result, ("catalog".to_string(), Some("schema".to_string())));
+
+        let result = split_database("schema").unwrap();
+        assert_eq!(result, ("greptime".to_string(), Some("schema".to_string())));
+
+        let result = split_database("catalog-*").unwrap();
+        assert_eq!(result, ("catalog".to_string(), None));
+
+        let result = split_database("*").unwrap();
+        assert_eq!(result, ("greptime".to_string(), None));
+    }
+
     #[tokio::test(flavor = "multi_thread")]
     async fn test_export_create_table_with_quoted_names() -> CmdResult<()> {
         let output_dir = tempfile::tempdir().unwrap();
@@ -487,7 +532,7 @@ mod tests {
             "--output-dir",
             &*output_dir.path().to_string_lossy(),
             "--target",
-            "create-table",
+            "schema",
         ]);
         let mut cli_app = cli.build(LoggingOptions::default()).await?;
         cli_app.start().await?;
@@ -496,7 +541,9 @@ mod tests {
 
         let output_file = output_dir
             .path()
-            .join("greptime-cli.export.create_table.sql");
+            .join("greptime")
+            .join("cli.export.create_table")
+            .join("create_tables.sql");
         let res = std::fs::read_to_string(output_file).unwrap();
         let expect = r#"CREATE TABLE IF NOT EXISTS "a.b.c" (
   "ts" TIMESTAMP(3) NOT NULL,

src/cmd/src/cli/repl.rs

@@ -34,7 +34,6 @@ use common_telemetry::debug;
 use either::Either;
 use meta_client::client::MetaClientBuilder;
 use query::datafusion::DatafusionQueryEngine;
-use query::logical_optimizer::LogicalOptimizer;
 use query::parser::QueryLanguageParser;
 use query::plan::LogicalPlan;
 use query::query_engine::{DefaultSerializer, QueryEngineState};
@@ -289,6 +288,7 @@ async fn create_query_engine(meta_addr: &str) -> Result<DatafusionQueryEngine> {
         None,
         None,
         None,
+        None,
         false,
         plugins.clone(),
     ));

src/cmd/src/datanode.rs

@@ -18,6 +18,7 @@ use std::time::Duration;
 use async_trait::async_trait;
 use catalog::kvbackend::MetaKvBackend;
 use clap::Parser;
+use common_base::Plugins;
 use common_config::Configurable;
 use common_telemetry::logging::TracingOptions;
 use common_telemetry::{info, warn};
@@ -25,13 +26,14 @@ use common_version::{short_version, version};
 use common_wal::config::DatanodeWalConfig;
 use datanode::datanode::{Datanode, DatanodeBuilder};
 use datanode::service::DatanodeServiceBuilder;
-use meta_client::MetaClientOptions;
+use meta_client::{MetaClientOptions, MetaClientType};
 use servers::Mode;
 use snafu::{OptionExt, ResultExt};
 use tracing_appender::non_blocking::WorkerGuard;
 
 use crate::error::{
-    LoadLayeredConfigSnafu, MissingConfigSnafu, Result, ShutdownDatanodeSnafu, StartDatanodeSnafu,
+    LoadLayeredConfigSnafu, MetaClientInitSnafu, MissingConfigSnafu, Result, ShutdownDatanodeSnafu,
+    StartDatanodeSnafu,
 };
 use crate::options::{GlobalOptions, GreptimeOptions};
 use crate::{log_versions, App};
@@ -125,7 +127,7 @@ struct StartCommand {
     rpc_addr: Option<String>,
     #[clap(long)]
     rpc_hostname: Option<String>,
-    #[clap(long, aliases = ["metasrv-addr"], value_delimiter = ',', num_args = 1..)]
+    #[clap(long, value_delimiter = ',', num_args = 1..)]
     metasrv_addrs: Option<Vec<String>>,
     #[clap(short, long)]
     config_file: Option<String>,
@@ -265,17 +267,19 @@ impl StartCommand {
             &opts.component.tracing,
             opts.component.node_id.map(|x| x.to_string()),
         );
-        log_versions(version!(), short_version!());
+        log_versions(version(), short_version());
 
         info!("Datanode start command: {:#?}", self);
         info!("Datanode options: {:#?}", opts);
 
-        let mut opts = opts.component;
-        let plugins = plugins::setup_datanode_plugins(&mut opts)
+        let opts = opts.component;
+        let mut plugins = Plugins::new();
+        plugins::setup_datanode_plugins(&mut plugins, &opts)
             .await
             .context(StartDatanodeSnafu)?;
 
-        let node_id = opts
+        let cluster_id = 0; // TODO(hl): read from config
+        let member_id = opts
             .node_id
             .context(MissingConfigSnafu { msg: "'node_id'" })?;
 
@@ -283,12 +287,16 @@ impl StartCommand {
             msg: "'meta_client_options'",
         })?;
 
-        let meta_client = datanode::heartbeat::new_metasrv_client(node_id, meta_config)
-            .await
-            .context(StartDatanodeSnafu)?;
+        let meta_client = meta_client::create_meta_client(
+            cluster_id,
+            MetaClientType::Datanode { member_id },
+            meta_config,
+        )
+        .await
+        .context(MetaClientInitSnafu)?;
 
         let meta_backend = Arc::new(MetaKvBackend {
-            client: Arc::new(meta_client.clone()),
+            client: meta_client.clone(),
         });
 
         let mut datanode = DatanodeBuilder::new(opts.clone(), plugins)
@@ -332,7 +340,7 @@ mod tests {
             mode = "distributed"
             enable_memory_catalog = false
             node_id = 42
-            
+
             rpc_addr = "127.0.0.1:4001"
             rpc_hostname = "192.168.0.1"
             [grpc]
@@ -359,7 +367,7 @@ mod tests {
             mode = "distributed"
             enable_memory_catalog = false
             node_id = 42
-            
+
             [grpc]
             addr = "127.0.0.1:3001"
             hostname = "127.0.0.1"

src/cmd/src/error.rs

@@ -87,6 +87,20 @@ pub enum Error {
         source: datanode::error::Error,
     },
 
+    #[snafu(display("Failed to start flownode"))]
+    StartFlownode {
+        #[snafu(implicit)]
+        location: Location,
+        source: flow::Error,
+    },
+
+    #[snafu(display("Failed to shutdown flownode"))]
+    ShutdownFlownode {
+        #[snafu(implicit)]
+        location: Location,
+        source: flow::Error,
+    },
+
     #[snafu(display("Failed to start frontend"))]
     StartFrontend {
         #[snafu(implicit)]
@@ -284,13 +298,6 @@ pub enum Error {
         error: std::io::Error,
     },
 
-    #[snafu(display("Invalid database name: {}", database))]
-    InvalidDatabaseName {
-        #[snafu(implicit)]
-        location: Location,
-        database: String,
-    },
-
     #[snafu(display("Failed to create directory {}", dir))]
     CreateDir {
         dir: String,
@@ -298,6 +305,12 @@ pub enum Error {
         error: std::io::Error,
     },
 
+    #[snafu(display("Failed to spawn thread"))]
+    SpawnThread {
+        #[snafu(source)]
+        error: std::io::Error,
+    },
+
     #[snafu(display("Other error"))]
     Other {
         source: BoxedError,
@@ -325,6 +338,22 @@ pub enum Error {
         location: Location,
         source: cache::error::Error,
     },
+
+    #[snafu(display("Failed to initialize meta client"))]
+    MetaClientInit {
+        #[snafu(implicit)]
+        location: Location,
+        source: meta_client::error::Error,
+    },
+
+    #[snafu(display("Tonic transport error: {error:?} with msg: {msg:?}"))]
+    TonicTransport {
+        #[snafu(implicit)]
+        location: Location,
+        #[snafu(source)]
+        error: tonic::transport::Error,
+        msg: Option<String>,
+    },
 }
 
 pub type Result<T> = std::result::Result<T, Error>;
@@ -354,8 +383,7 @@ impl ErrorExt for Error {
             | Error::ConnectEtcd { .. }
             | Error::NotDataFromOutput { .. }
             | Error::CreateDir { .. }
-            | Error::EmptyResult { .. }
-            | Error::InvalidDatabaseName { .. } => StatusCode::InvalidArguments,
+            | Error::EmptyResult { .. } => StatusCode::InvalidArguments,
 
             Error::StartProcedureManager { source, .. }
             | Error::StopProcedureManager { source, .. } => source.status_code(),
@@ -373,13 +401,20 @@ impl ErrorExt for Error {
             Error::SubstraitEncodeLogicalPlan { source, .. } => source.status_code(),
             Error::StartCatalogManager { source, .. } => source.status_code(),
 
-            Error::SerdeJson { .. } | Error::FileIo { .. } => StatusCode::Unexpected,
+            Error::SerdeJson { .. } | Error::FileIo { .. } | Error::SpawnThread { .. } => {
+                StatusCode::Unexpected
+            }
 
             Error::Other { source, .. } => source.status_code(),
 
             Error::BuildRuntime { source, .. } => source.status_code(),
 
             Error::CacheRequired { .. } | Error::BuildCacheRegistry { .. } => StatusCode::Internal,
+            Self::StartFlownode { source, .. } | Self::ShutdownFlownode { source, .. } => {
+                source.status_code()
+            }
+            Error::MetaClientInit { source, .. } => source.status_code(),
+            Error::TonicTransport { .. } => StatusCode::Internal,
         }
     }
 

src/cmd/src/flownode.rs

@@ -0,0 +1,337 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use std::sync::Arc;
+
+use cache::{build_fundamental_cache_registry, with_default_composite_cache_registry};
+use catalog::kvbackend::{CachedMetaKvBackendBuilder, KvBackendCatalogManager, MetaKvBackend};
+use clap::Parser;
+use client::client_manager::NodeClients;
+use common_base::Plugins;
+use common_config::Configurable;
+use common_grpc::channel_manager::ChannelConfig;
+use common_meta::cache::{CacheRegistryBuilder, LayeredCacheRegistryBuilder};
+use common_meta::heartbeat::handler::parse_mailbox_message::ParseMailboxMessageHandler;
+use common_meta::heartbeat::handler::HandlerGroupExecutor;
+use common_meta::key::flow::FlowMetadataManager;
+use common_meta::key::TableMetadataManager;
+use common_telemetry::info;
+use common_telemetry::logging::TracingOptions;
+use common_version::{short_version, version};
+use flow::{FlownodeBuilder, FlownodeInstance, FrontendInvoker};
+use frontend::heartbeat::handler::invalidate_table_cache::InvalidateTableCacheHandler;
+use meta_client::{MetaClientOptions, MetaClientType};
+use servers::Mode;
+use snafu::{OptionExt, ResultExt};
+use tracing_appender::non_blocking::WorkerGuard;
+
+use crate::error::{
+    BuildCacheRegistrySnafu, InitMetadataSnafu, LoadLayeredConfigSnafu, MetaClientInitSnafu,
+    MissingConfigSnafu, Result, ShutdownFlownodeSnafu, StartFlownodeSnafu,
+};
+use crate::options::{GlobalOptions, GreptimeOptions};
+use crate::{log_versions, App};
+
+pub const APP_NAME: &str = "greptime-flownode";
+
+type FlownodeOptions = GreptimeOptions<flow::FlownodeOptions>;
+
+pub struct Instance {
+    flownode: FlownodeInstance,
+
+    // Keep the logging guard to prevent the worker from being dropped.
+    _guard: Vec<WorkerGuard>,
+}
+
+impl Instance {
+    pub fn new(flownode: FlownodeInstance, guard: Vec<WorkerGuard>) -> Self {
+        Self {
+            flownode,
+            _guard: guard,
+        }
+    }
+
+    pub fn flownode_mut(&mut self) -> &mut FlownodeInstance {
+        &mut self.flownode
+    }
+
+    pub fn flownode(&self) -> &FlownodeInstance {
+        &self.flownode
+    }
+}
+
+#[async_trait::async_trait]
+impl App for Instance {
+    fn name(&self) -> &str {
+        APP_NAME
+    }
+
+    async fn start(&mut self) -> Result<()> {
+        self.flownode.start().await.context(StartFlownodeSnafu)
+    }
+
+    async fn stop(&self) -> Result<()> {
+        self.flownode
+            .shutdown()
+            .await
+            .context(ShutdownFlownodeSnafu)
+    }
+}
+
+#[derive(Parser)]
+pub struct Command {
+    #[clap(subcommand)]
+    subcmd: SubCommand,
+}
+
+impl Command {
+    pub async fn build(&self, opts: FlownodeOptions) -> Result<Instance> {
+        self.subcmd.build(opts).await
+    }
+
+    pub fn load_options(&self, global_options: &GlobalOptions) -> Result<FlownodeOptions> {
+        match &self.subcmd {
+            SubCommand::Start(cmd) => cmd.load_options(global_options),
+        }
+    }
+}
+
+#[derive(Parser)]
+enum SubCommand {
+    Start(StartCommand),
+}
+
+impl SubCommand {
+    async fn build(&self, opts: FlownodeOptions) -> Result<Instance> {
+        match self {
+            SubCommand::Start(cmd) => cmd.build(opts).await,
+        }
+    }
+}
+
+#[derive(Debug, Parser, Default)]
+struct StartCommand {
+    /// Flownode's id
+    #[clap(long)]
+    node_id: Option<u64>,
+    /// Bind address for the gRPC server.
+    #[clap(long)]
+    rpc_addr: Option<String>,
+    /// Hostname for the gRPC server.
+    #[clap(long)]
+    rpc_hostname: Option<String>,
+    /// Metasrv address list;
+    #[clap(long, value_delimiter = ',', num_args = 1..)]
+    metasrv_addrs: Option<Vec<String>>,
+    /// The configuration file for flownode
+    #[clap(short, long)]
+    config_file: Option<String>,
+    /// The prefix of environment variables, default is `GREPTIMEDB_FLOWNODE`;
+    #[clap(long, default_value = "GREPTIMEDB_FLOWNODE")]
+    env_prefix: String,
+}
+
+impl StartCommand {
+    fn load_options(&self, global_options: &GlobalOptions) -> Result<FlownodeOptions> {
+        let mut opts = FlownodeOptions::load_layered_options(
+            self.config_file.as_deref(),
+            self.env_prefix.as_ref(),
+        )
+        .context(LoadLayeredConfigSnafu)?;
+
+        self.merge_with_cli_options(global_options, &mut opts)?;
+
+        Ok(opts)
+    }
+
+    // The precedence order is: cli > config file > environment variables > default values.
+    fn merge_with_cli_options(
+        &self,
+        global_options: &GlobalOptions,
+        opts: &mut FlownodeOptions,
+    ) -> Result<()> {
+        let opts = &mut opts.component;
+
+        if let Some(dir) = &global_options.log_dir {
+            opts.logging.dir.clone_from(dir);
+        }
+
+        if global_options.log_level.is_some() {
+            opts.logging.level.clone_from(&global_options.log_level);
+        }
+
+        opts.tracing = TracingOptions {
+            #[cfg(feature = "tokio-console")]
+            tokio_console_addr: global_options.tokio_console_addr.clone(),
+        };
+
+        if let Some(addr) = &self.rpc_addr {
+            opts.grpc.addr.clone_from(addr);
+        }
+
+        if let Some(hostname) = &self.rpc_hostname {
+            opts.grpc.hostname.clone_from(hostname);
+        }
+
+        if let Some(node_id) = self.node_id {
+            opts.node_id = Some(node_id);
+        }
+
+        if let Some(metasrv_addrs) = &self.metasrv_addrs {
+            opts.meta_client
+                .get_or_insert_with(MetaClientOptions::default)
+                .metasrv_addrs
+                .clone_from(metasrv_addrs);
+            opts.mode = Mode::Distributed;
+        }
+
+        if let (Mode::Distributed, None) = (&opts.mode, &opts.node_id) {
+            return MissingConfigSnafu {
+                msg: "Missing node id option",
+            }
+            .fail();
+        }
+
+        Ok(())
+    }
+
+    async fn build(&self, opts: FlownodeOptions) -> Result<Instance> {
+        common_runtime::init_global_runtimes(&opts.runtime);
+
+        let guard = common_telemetry::init_global_logging(
+            APP_NAME,
+            &opts.component.logging,
+            &opts.component.tracing,
+            opts.component.node_id.map(|x| x.to_string()),
+        );
+        log_versions(version(), short_version());
+
+        info!("Flownode start command: {:#?}", self);
+        info!("Flownode options: {:#?}", opts);
+
+        let opts = opts.component;
+
+        // TODO(discord9): make it not optionale after cluster id is required
+        let cluster_id = opts.cluster_id.unwrap_or(0);
+
+        let member_id = opts
+            .node_id
+            .context(MissingConfigSnafu { msg: "'node_id'" })?;
+
+        let meta_config = opts.meta_client.as_ref().context(MissingConfigSnafu {
+            msg: "'meta_client_options'",
+        })?;
+
+        let meta_client = meta_client::create_meta_client(
+            cluster_id,
+            MetaClientType::Flownode { member_id },
+            meta_config,
+        )
+        .await
+        .context(MetaClientInitSnafu)?;
+
+        let cache_max_capacity = meta_config.metadata_cache_max_capacity;
+        let cache_ttl = meta_config.metadata_cache_ttl;
+        let cache_tti = meta_config.metadata_cache_tti;
+
+        // TODO(discord9): add helper function to ease the creation of cache registry&such
+        let cached_meta_backend = CachedMetaKvBackendBuilder::new(meta_client.clone())
+            .cache_max_capacity(cache_max_capacity)
+            .cache_ttl(cache_ttl)
+            .cache_tti(cache_tti)
+            .build();
+        let cached_meta_backend = Arc::new(cached_meta_backend);
+
+        // Builds cache registry
+        let layered_cache_builder = LayeredCacheRegistryBuilder::default().add_cache_registry(
+            CacheRegistryBuilder::default()
+                .add_cache(cached_meta_backend.clone())
+                .build(),
+        );
+        let fundamental_cache_registry =
+            build_fundamental_cache_registry(Arc::new(MetaKvBackend::new(meta_client.clone())));
+        let layered_cache_registry = Arc::new(
+            with_default_composite_cache_registry(
+                layered_cache_builder.add_cache_registry(fundamental_cache_registry),
+            )
+            .context(BuildCacheRegistrySnafu)?
+            .build(),
+        );
+
+        let catalog_manager = KvBackendCatalogManager::new(
+            opts.mode,
+            Some(meta_client.clone()),
+            cached_meta_backend.clone(),
+            layered_cache_registry.clone(),
+        );
+
+        let table_metadata_manager =
+            Arc::new(TableMetadataManager::new(cached_meta_backend.clone()));
+        table_metadata_manager
+            .init()
+            .await
+            .context(InitMetadataSnafu)?;
+
+        let executor = HandlerGroupExecutor::new(vec![
+            Arc::new(ParseMailboxMessageHandler),
+            Arc::new(InvalidateTableCacheHandler::new(
+                layered_cache_registry.clone(),
+            )),
+        ]);
+
+        let heartbeat_task = flow::heartbeat::HeartbeatTask::new(
+            &opts,
+            meta_client.clone(),
+            opts.heartbeat.clone(),
+            Arc::new(executor),
+        );
+
+        let flow_metadata_manager = Arc::new(FlowMetadataManager::new(cached_meta_backend.clone()));
+        let flownode_builder = FlownodeBuilder::new(
+            opts,
+            Plugins::new(),
+            table_metadata_manager,
+            catalog_manager.clone(),
+            flow_metadata_manager,
+        )
+        .with_heartbeat_task(heartbeat_task);
+
+        let flownode = flownode_builder.build().await.context(StartFlownodeSnafu)?;
+
+        // flownode's frontend to datanode need not timeout.
+        // Some queries are expected to take long time.
+        let channel_config = ChannelConfig {
+            timeout: None,
+            ..Default::default()
+        };
+        let client = Arc::new(NodeClients::new(channel_config));
+
+        let invoker = FrontendInvoker::build_from(
+            flownode.flow_worker_manager().clone(),
+            catalog_manager.clone(),
+            cached_meta_backend.clone(),
+            layered_cache_registry.clone(),
+            meta_client.clone(),
+            client,
+        )
+        .await
+        .context(StartFlownodeSnafu)?;
+        flownode
+            .flow_worker_manager()
+            .set_frontend_invoker(invoker)
+            .await;
+
+        Ok(Instance::new(flownode, guard))
+    }
+}

src/cmd/src/frontend.rs

@@ -20,6 +20,7 @@ use cache::{build_fundamental_cache_registry, with_default_composite_cache_regis
 use catalog::kvbackend::{CachedMetaKvBackendBuilder, KvBackendCatalogManager, MetaKvBackend};
 use clap::Parser;
 use client::client_manager::NodeClients;
+use common_base::Plugins;
 use common_config::Configurable;
 use common_grpc::channel_manager::ChannelConfig;
 use common_meta::cache::{CacheRegistryBuilder, LayeredCacheRegistryBuilder};
@@ -34,14 +35,15 @@ use frontend::heartbeat::HeartbeatTask;
 use frontend::instance::builder::FrontendBuilder;
 use frontend::instance::{FrontendInstance, Instance as FeInstance};
 use frontend::server::Services;
-use meta_client::MetaClientOptions;
+use meta_client::{MetaClientOptions, MetaClientType};
 use servers::tls::{TlsMode, TlsOption};
 use servers::Mode;
 use snafu::{OptionExt, ResultExt};
 use tracing_appender::non_blocking::WorkerGuard;
 
 use crate::error::{
-    self, InitTimezoneSnafu, LoadLayeredConfigSnafu, MissingConfigSnafu, Result, StartFrontendSnafu,
+    self, InitTimezoneSnafu, LoadLayeredConfigSnafu, MetaClientInitSnafu, MissingConfigSnafu,
+    Result, StartFrontendSnafu,
 };
 use crate::options::{GlobalOptions, GreptimeOptions};
 use crate::{log_versions, App};
@@ -147,7 +149,7 @@ pub struct StartCommand {
     config_file: Option<String>,
     #[clap(short, long)]
     influxdb_enable: Option<bool>,
-    #[clap(long, aliases = ["metasrv-addr"], value_delimiter = ',', num_args = 1..)]
+    #[clap(long, value_delimiter = ',', num_args = 1..)]
     metasrv_addrs: Option<Vec<String>>,
     #[clap(long)]
     tls_mode: Option<TlsMode>,
@@ -241,10 +243,11 @@ impl StartCommand {
                 .get_or_insert_with(MetaClientOptions::default)
                 .metasrv_addrs
                 .clone_from(metasrv_addrs);
-            opts.mode = Mode::Distributed;
         }
 
-        opts.user_provider.clone_from(&self.user_provider);
+        if let Some(user_provider) = &self.user_provider {
+            opts.user_provider = Some(user_provider.clone());
+        }
 
         Ok(())
     }
@@ -258,14 +261,14 @@ impl StartCommand {
             &opts.component.tracing,
             opts.component.node_id.clone(),
         );
-        log_versions(version!(), short_version!());
+        log_versions(version(), short_version());
 
         info!("Frontend start command: {:#?}", self);
         info!("Frontend options: {:#?}", opts);
 
-        let mut opts = opts.component;
-        #[allow(clippy::unnecessary_mut_passed)]
-        let plugins = plugins::setup_frontend_plugins(&mut opts)
+        let opts = opts.component;
+        let mut plugins = Plugins::new();
+        plugins::setup_frontend_plugins(&mut plugins, &opts)
             .await
             .context(StartFrontendSnafu)?;
 
@@ -279,10 +282,16 @@ impl StartCommand {
         let cache_ttl = meta_client_options.metadata_cache_ttl;
         let cache_tti = meta_client_options.metadata_cache_tti;
 
-        let meta_client = FeInstance::create_meta_client(meta_client_options)
-            .await
-            .context(StartFrontendSnafu)?;
+        let cluster_id = 0; // (TODO: jeremy): It is currently a reserved field and has not been enabled.
+        let meta_client = meta_client::create_meta_client(
+            cluster_id,
+            MetaClientType::Frontend,
+            meta_client_options,
+        )
+        .await
+        .context(MetaClientInitSnafu)?;
 
+        // TODO(discord9): add helper function to ease the creation of cache registry&such
         let cached_meta_backend = CachedMetaKvBackendBuilder::new(meta_client.clone())
             .cache_max_capacity(cache_max_capacity)
             .cache_ttl(cache_ttl)
@@ -307,7 +316,7 @@ impl StartCommand {
         );
 
         let catalog_manager = KvBackendCatalogManager::new(
-            opts.mode,
+            Mode::Distributed,
             Some(meta_client.clone()),
             cached_meta_backend.clone(),
             layered_cache_registry.clone(),
@@ -336,6 +345,7 @@ impl StartCommand {
         let client = NodeClients::new(channel_config);
 
         let mut instance = FrontendBuilder::new(
+            opts.clone(),
             cached_meta_backend.clone(),
             layered_cache_registry.clone(),
             catalog_manager,
@@ -349,12 +359,12 @@ impl StartCommand {
         .await
         .context(StartFrontendSnafu)?;
 
-        let servers = Services::new(opts.clone(), Arc::new(instance.clone()), plugins)
+        let servers = Services::new(opts, Arc::new(instance.clone()), plugins)
             .build()
             .await
             .context(StartFrontendSnafu)?;
         instance
-            .build_servers(opts, servers)
+            .build_servers(servers)
             .context(StartFrontendSnafu)?;
 
         Ok(Instance::new(instance, guard))
@@ -437,7 +447,6 @@ mod tests {
 
         let fe_opts = command.load_options(&Default::default()).unwrap().component;
 
-        assert_eq!(Mode::Distributed, fe_opts.mode);
         assert_eq!("127.0.0.1:4000".to_string(), fe_opts.http.addr);
         assert_eq!(Duration::from_secs(30), fe_opts.http.timeout);
 
@@ -450,7 +459,7 @@ mod tests {
 
     #[tokio::test]
     async fn test_try_from_start_command_to_anymap() {
-        let mut fe_opts = frontend::frontend::FrontendOptions {
+        let fe_opts = frontend::frontend::FrontendOptions {
             http: HttpOptions {
                 disable_dashboard: false,
                 ..Default::default()
@@ -459,8 +468,10 @@ mod tests {
             ..Default::default()
         };
 
-        #[allow(clippy::unnecessary_mut_passed)]
-        let plugins = plugins::setup_frontend_plugins(&mut fe_opts).await.unwrap();
+        let mut plugins = Plugins::new();
+        plugins::setup_frontend_plugins(&mut plugins, &fe_opts)
+            .await
+            .unwrap();
 
         let provider = plugins.get::<UserProviderRef>().unwrap();
         let result = provider

src/cmd/src/lib.rs

@@ -22,6 +22,7 @@ use crate::error::Result;
 pub mod cli;
 pub mod datanode;
 pub mod error;
+pub mod flownode;
 pub mod frontend;
 pub mod metasrv;
 pub mod options;
@@ -29,7 +30,7 @@ pub mod standalone;
 
 lazy_static::lazy_static! {
     static ref APP_VERSION: prometheus::IntGaugeVec =
-        prometheus::register_int_gauge_vec!("greptime_app_version", "app version", &["short_version", "version"]).unwrap();
+        prometheus::register_int_gauge_vec!("greptime_app_version", "app version", &["version", "short_version"]).unwrap();
 }
 
 #[async_trait]
@@ -73,16 +74,16 @@ pub trait App: Send {
 }
 
 /// Log the versions of the application, and the arguments passed to the cli.
-/// `version_string` should be the same as the output of cli "--version";
-/// and the `app_version` is the short version of the codes, often consist of git branch and commit.
-pub fn log_versions(version_string: &str, app_version: &str) {
+/// `version` should be the same as the output of cli "--version";
+/// and the `short_version` is the short version of the codes, often consist of git branch and commit.
+pub fn log_versions(version: &str, short_version: &str) {
     // Report app version as gauge.
     APP_VERSION
-        .with_label_values(&[env!("CARGO_PKG_VERSION"), app_version])
+        .with_label_values(&[env!("CARGO_PKG_VERSION"), short_version])
         .inc();
 
     // Log version and argument flags.
-    info!("GreptimeDB version: {}", version_string);
+    info!("GreptimeDB version: {}", version);
 
     log_env_flags();
 }

src/cmd/src/metasrv.rs

@@ -16,11 +16,13 @@ use std::time::Duration;
 
 use async_trait::async_trait;
 use clap::Parser;
+use common_base::Plugins;
 use common_config::Configurable;
 use common_telemetry::info;
 use common_telemetry::logging::TracingOptions;
 use common_version::{short_version, version};
 use meta_srv::bootstrap::MetasrvInstance;
+use meta_srv::metasrv::BackendImpl;
 use snafu::ResultExt;
 use tracing_appender::non_blocking::WorkerGuard;
 
@@ -136,6 +138,9 @@ struct StartCommand {
     /// The max operations per txn
     #[clap(long)]
     max_txn_ops: Option<usize>,
+    /// The database backend.
+    #[clap(long, value_enum)]
+    backend: Option<BackendImpl>,
 }
 
 impl StartCommand {
@@ -218,6 +223,12 @@ impl StartCommand {
             opts.max_txn_ops = max_txn_ops;
         }
 
+        if let Some(backend) = &self.backend {
+            opts.backend.clone_from(backend);
+        } else {
+            opts.backend = BackendImpl::default()
+        }
+
         // Disable dashboard in metasrv.
         opts.http.disable_dashboard = true;
 
@@ -233,13 +244,14 @@ impl StartCommand {
             &opts.component.tracing,
             None,
         );
-        log_versions(version!(), short_version!());
+        log_versions(version(), short_version());
 
         info!("Metasrv start command: {:#?}", self);
         info!("Metasrv options: {:#?}", opts);
 
-        let mut opts = opts.component;
-        let plugins = plugins::setup_metasrv_plugins(&mut opts)
+        let opts = opts.component;
+        let mut plugins = Plugins::new();
+        plugins::setup_metasrv_plugins(&mut plugins, &opts)
             .await
             .context(StartMetaServerSnafu)?;
 
@@ -296,7 +308,7 @@ mod tests {
             [logging]
             level = "debug"
             dir = "/tmp/greptimedb/test/logs"
-            
+
             [failure_detector]
             threshold = 8.0
             min_std_deviation = "100ms"

src/cmd/src/standalone.rs

@@ -19,19 +19,20 @@ use async_trait::async_trait;
 use cache::{build_fundamental_cache_registry, with_default_composite_cache_registry};
 use catalog::kvbackend::KvBackendCatalogManager;
 use clap::Parser;
+use common_base::Plugins;
 use common_catalog::consts::{MIN_USER_FLOW_ID, MIN_USER_TABLE_ID};
 use common_config::{metadata_store_dir, Configurable, KvBackendConfig};
+use common_error::ext::BoxedError;
 use common_meta::cache::LayeredCacheRegistryBuilder;
 use common_meta::cache_invalidator::CacheInvalidatorRef;
 use common_meta::ddl::flow_meta::{FlowMetadataAllocator, FlowMetadataAllocatorRef};
 use common_meta::ddl::table_meta::{TableMetadataAllocator, TableMetadataAllocatorRef};
-use common_meta::ddl::{DdlContext, ProcedureExecutorRef};
+use common_meta::ddl::{DdlContext, NoopRegionFailureDetectorControl, ProcedureExecutorRef};
 use common_meta::ddl_manager::DdlManager;
 use common_meta::key::flow::{FlowMetadataManager, FlowMetadataManagerRef};
 use common_meta::key::{TableMetadataManager, TableMetadataManagerRef};
 use common_meta::kv_backend::KvBackendRef;
 use common_meta::node_manager::NodeManagerRef;
-use common_meta::peer::StandalonePeerLookupService;
 use common_meta::region_keeper::MemoryRegionKeeper;
 use common_meta::sequence::SequenceBuilder;
 use common_meta::wal_options_allocator::{WalOptionsAllocator, WalOptionsAllocatorRef};
@@ -40,11 +41,11 @@ use common_telemetry::info;
 use common_telemetry::logging::{LoggingOptions, TracingOptions};
 use common_time::timezone::set_default_timezone;
 use common_version::{short_version, version};
-use common_wal::config::StandaloneWalConfig;
+use common_wal::config::DatanodeWalConfig;
 use datanode::config::{DatanodeOptions, ProcedureConfig, RegionEngineConfig, StorageConfig};
 use datanode::datanode::{Datanode, DatanodeBuilder};
 use file_engine::config::EngineConfig as FileEngineConfig;
-use flow::FlownodeBuilder;
+use flow::{FlowWorkerManager, FlownodeBuilder, FrontendInvoker};
 use frontend::frontend::FrontendOptions;
 use frontend::instance::builder::FrontendBuilder;
 use frontend::instance::{FrontendInstance, Instance as FeInstance, StandaloneDatanodeManager};
@@ -61,12 +62,14 @@ use servers::http::HttpOptions;
 use servers::tls::{TlsMode, TlsOption};
 use servers::Mode;
 use snafu::ResultExt;
+use tokio::sync::broadcast;
 use tracing_appender::non_blocking::WorkerGuard;
 
 use crate::error::{
     BuildCacheRegistrySnafu, CreateDirSnafu, IllegalConfigSnafu, InitDdlManagerSnafu,
-    InitMetadataSnafu, InitTimezoneSnafu, LoadLayeredConfigSnafu, Result, ShutdownDatanodeSnafu,
-    ShutdownFrontendSnafu, StartDatanodeSnafu, StartFrontendSnafu, StartProcedureManagerSnafu,
+    InitMetadataSnafu, InitTimezoneSnafu, LoadLayeredConfigSnafu, OtherSnafu, Result,
+    ShutdownDatanodeSnafu, ShutdownFlownodeSnafu, ShutdownFrontendSnafu, StartDatanodeSnafu,
+    StartFlownodeSnafu, StartFrontendSnafu, StartProcedureManagerSnafu,
     StartWalOptionsAllocatorSnafu, StopProcedureManagerSnafu,
 };
 use crate::options::{GlobalOptions, GreptimeOptions};
@@ -128,7 +131,7 @@ pub struct StandaloneOptions {
     pub opentsdb: OpentsdbOptions,
     pub influxdb: InfluxdbOptions,
     pub prom_store: PromStoreOptions,
-    pub wal: StandaloneWalConfig,
+    pub wal: DatanodeWalConfig,
     pub storage: StorageConfig,
     pub metadata_store: KvBackendConfig,
     pub procedure: ProcedureConfig,
@@ -153,7 +156,7 @@ impl Default for StandaloneOptions {
             opentsdb: OpentsdbOptions::default(),
             influxdb: InfluxdbOptions::default(),
             prom_store: PromStoreOptions::default(),
-            wal: StandaloneWalConfig::default(),
+            wal: DatanodeWalConfig::default(),
             storage: StorageConfig::default(),
             metadata_store: KvBackendConfig::default(),
             procedure: ProcedureConfig::default(),
@@ -179,7 +182,6 @@ impl StandaloneOptions {
     pub fn frontend_options(&self) -> FrontendOptions {
         let cloned_opts = self.clone();
         FrontendOptions {
-            mode: cloned_opts.mode,
             default_timezone: cloned_opts.default_timezone,
             http: cloned_opts.http,
             grpc: cloned_opts.grpc,
@@ -202,7 +204,7 @@ impl StandaloneOptions {
         DatanodeOptions {
             node_id: Some(0),
             enable_telemetry: cloned_opts.enable_telemetry,
-            wal: cloned_opts.wal.into(),
+            wal: cloned_opts.wal,
             storage: cloned_opts.storage,
             region_engine: cloned_opts.region_engine,
             grpc: cloned_opts.grpc,
@@ -214,6 +216,9 @@ impl StandaloneOptions {
 pub struct Instance {
     datanode: Datanode,
     frontend: FeInstance,
+    // TODO(discord9): wrapped it in flownode instance instead
+    flow_worker_manager: Arc<FlowWorkerManager>,
+    flow_shutdown: broadcast::Sender<()>,
     procedure_manager: ProcedureManagerRef,
     wal_options_allocator: WalOptionsAllocatorRef,
 
@@ -245,6 +250,9 @@ impl App for Instance {
             .context(StartFrontendSnafu)?;
 
         self.frontend.start().await.context(StartFrontendSnafu)?;
+        self.flow_worker_manager
+            .clone()
+            .run_background(Some(self.flow_shutdown.subscribe()));
         Ok(())
     }
 
@@ -263,6 +271,15 @@ impl App for Instance {
             .shutdown()
             .await
             .context(ShutdownDatanodeSnafu)?;
+        self.flow_shutdown
+            .send(())
+            .map_err(|_e| {
+                flow::error::InternalSnafu {
+                    reason: "Failed to send shutdown signal to flow worker manager, all receiver end already closed".to_string(),
+                }
+                .build()
+            })
+            .context(ShutdownFlownodeSnafu)?;
         info!("Datanode instance stopped.");
 
         Ok(())
@@ -379,7 +396,9 @@ impl StartCommand {
             opts.influxdb.enable = self.influxdb_enable;
         }
 
-        opts.user_provider.clone_from(&self.user_provider);
+        if let Some(user_provider) = &self.user_provider {
+            opts.user_provider = Some(user_provider.clone());
+        }
 
         Ok(())
     }
@@ -396,19 +415,23 @@ impl StartCommand {
             &opts.component.tracing,
             None,
         );
-        log_versions(version!(), short_version!());
+        log_versions(version(), short_version());
 
         info!("Standalone start command: {:#?}", self);
         info!("Standalone options: {opts:#?}");
 
+        let mut plugins = Plugins::new();
         let opts = opts.component;
-        let mut fe_opts = opts.frontend_options();
-        #[allow(clippy::unnecessary_mut_passed)]
-        let fe_plugins = plugins::setup_frontend_plugins(&mut fe_opts) // mut ref is MUST, DO NOT change it
+        let fe_opts = opts.frontend_options();
+        let dn_opts = opts.datanode_options();
+
+        plugins::setup_frontend_plugins(&mut plugins, &fe_opts)
             .await
             .context(StartFrontendSnafu)?;
 
-        let dn_opts = opts.datanode_options();
+        plugins::setup_datanode_plugins(&mut plugins, &dn_opts)
+            .await
+            .context(StartDatanodeSnafu)?;
 
         set_default_timezone(fe_opts.default_timezone.as_deref()).context(InitTimezoneSnafu)?;
 
@@ -447,24 +470,31 @@ impl StartCommand {
         let table_metadata_manager =
             Self::create_table_metadata_manager(kv_backend.clone()).await?;
 
-        let flow_builder = FlownodeBuilder::new(
-            1,
-            Default::default(),
-            fe_plugins.clone(),
-            table_metadata_manager.clone(),
-            catalog_manager.clone(),
-        );
-        let flownode = Arc::new(flow_builder.build().await);
-
-        let datanode = DatanodeBuilder::new(dn_opts, fe_plugins.clone())
+        let datanode = DatanodeBuilder::new(dn_opts, plugins.clone())
             .with_kv_backend(kv_backend.clone())
             .build()
             .await
             .context(StartDatanodeSnafu)?;
 
+        let flow_metadata_manager = Arc::new(FlowMetadataManager::new(kv_backend.clone()));
+        let flow_builder = FlownodeBuilder::new(
+            Default::default(),
+            plugins.clone(),
+            table_metadata_manager.clone(),
+            catalog_manager.clone(),
+            flow_metadata_manager.clone(),
+        );
+        let flownode = Arc::new(
+            flow_builder
+                .build()
+                .await
+                .map_err(BoxedError::new)
+                .context(OtherSnafu)?,
+        );
+
         let node_manager = Arc::new(StandaloneDatanodeManager {
             region_server: datanode.region_server(),
-            flow_server: flownode.clone(),
+            flow_server: flownode.flow_worker_manager(),
         });
 
         let table_id_sequence = Arc::new(
@@ -483,7 +513,6 @@ impl StartCommand {
             opts.wal.into(),
             kv_backend.clone(),
         ));
-        let flow_metadata_manager = Arc::new(FlowMetadataManager::new(kv_backend.clone()));
         let table_meta_allocator = Arc::new(TableMetadataAllocator::new(
             table_id_sequence,
             wal_options_allocator.clone(),
@@ -504,35 +533,47 @@ impl StartCommand {
         .await?;
 
         let mut frontend = FrontendBuilder::new(
-            kv_backend,
-            layered_cache_registry,
-            catalog_manager,
-            node_manager,
-            ddl_task_executor,
+            fe_opts.clone(),
+            kv_backend.clone(),
+            layered_cache_registry.clone(),
+            catalog_manager.clone(),
+            node_manager.clone(),
+            ddl_task_executor.clone(),
         )
-        .with_plugin(fe_plugins.clone())
+        .with_plugin(plugins.clone())
         .try_build()
         .await
         .context(StartFrontendSnafu)?;
 
+        let flow_worker_manager = flownode.flow_worker_manager();
         // flow server need to be able to use frontend to write insert requests back
-        flownode
-            .set_frontend_invoker(Box::new(frontend.clone()))
-            .await;
-        // TODO(discord9): unify with adding `start` and `shutdown` method to flownode too.
-        let _handle = flownode.clone().run_background();
+        let invoker = FrontendInvoker::build_from(
+            flow_worker_manager.clone(),
+            catalog_manager.clone(),
+            kv_backend.clone(),
+            layered_cache_registry.clone(),
+            ddl_task_executor.clone(),
+            node_manager,
+        )
+        .await
+        .context(StartFlownodeSnafu)?;
+        flow_worker_manager.set_frontend_invoker(invoker).await;
 
-        let servers = Services::new(fe_opts.clone(), Arc::new(frontend.clone()), fe_plugins)
+        let (tx, _rx) = broadcast::channel(1);
+
+        let servers = Services::new(fe_opts, Arc::new(frontend.clone()), plugins)
             .build()
             .await
             .context(StartFrontendSnafu)?;
         frontend
-            .build_servers(fe_opts, servers)
+            .build_servers(servers)
             .context(StartFrontendSnafu)?;
 
         Ok(Instance {
             datanode,
             frontend,
+            flow_worker_manager,
+            flow_shutdown: tx,
             procedure_manager,
             wal_options_allocator,
             _guard: guard,
@@ -558,7 +599,7 @@ impl StartCommand {
                     table_metadata_allocator,
                     flow_metadata_manager,
                     flow_metadata_allocator,
-                    peer_lookup_service: Arc::new(StandalonePeerLookupService::new()),
+                    region_failure_detector_controller: Arc::new(NoopRegionFailureDetectorControl),
                 },
                 procedure_manager,
                 true,
@@ -595,20 +636,21 @@ mod tests {
     use common_test_util::temp_dir::create_named_temp_file;
     use common_wal::config::DatanodeWalConfig;
     use datanode::config::{FileConfig, GcsConfig};
-    use servers::Mode;
 
     use super::*;
     use crate::options::GlobalOptions;
 
     #[tokio::test]
     async fn test_try_from_start_command_to_anymap() {
-        let mut fe_opts = FrontendOptions {
+        let fe_opts = FrontendOptions {
             user_provider: Some("static_user_provider:cmd:test=test".to_string()),
             ..Default::default()
         };
 
-        #[allow(clippy::unnecessary_mut_passed)]
-        let plugins = plugins::setup_frontend_plugins(&mut fe_opts).await.unwrap();
+        let mut plugins = Plugins::new();
+        plugins::setup_frontend_plugins(&mut plugins, &fe_opts)
+            .await
+            .unwrap();
 
         let provider = plugins.get::<UserProviderRef>().unwrap();
         let result = provider
@@ -693,7 +735,6 @@ mod tests {
         let fe_opts = options.frontend_options();
         let dn_opts = options.datanode_options();
         let logging_opts = options.logging;
-        assert_eq!(Mode::Standalone, fe_opts.mode);
         assert_eq!("127.0.0.1:4000".to_string(), fe_opts.http.addr);
         assert_eq!(Duration::from_secs(33), fe_opts.http.timeout);
         assert_eq!(ReadableSize::mb(128), fe_opts.http.body_limit);

src/cmd/tests/load_config_test.rs

@@ -22,10 +22,11 @@ use common_grpc::channel_manager::{
     DEFAULT_MAX_GRPC_RECV_MESSAGE_SIZE, DEFAULT_MAX_GRPC_SEND_MESSAGE_SIZE,
 };
 use common_runtime::global::RuntimeOptions;
-use common_telemetry::logging::LoggingOptions;
+use common_telemetry::logging::{LoggingOptions, DEFAULT_OTLP_ENDPOINT};
 use common_wal::config::raft_engine::RaftEngineConfig;
-use common_wal::config::{DatanodeWalConfig, StandaloneWalConfig};
+use common_wal::config::DatanodeWalConfig;
 use datanode::config::{DatanodeOptions, RegionEngineConfig, StorageConfig};
+use file_engine::config::EngineConfig;
 use frontend::frontend::FrontendOptions;
 use frontend::service_config::datanode::DatanodeClientOptions;
 use meta_client::MetaClientOptions;
@@ -45,9 +46,8 @@ fn test_load_datanode_example_config() {
 
     let expected = GreptimeOptions::<DatanodeOptions> {
         runtime: RuntimeOptions {
-            read_rt_size: 8,
-            write_rt_size: 8,
-            bg_rt_size: 4,
+            global_rt_size: 8,
+            compact_rt_size: 4,
         },
         component: DatanodeOptions {
             node_id: Some(42),
@@ -71,21 +71,24 @@ fn test_load_datanode_example_config() {
                 data_home: "/tmp/greptimedb/".to_string(),
                 ..Default::default()
             },
-            region_engine: vec![RegionEngineConfig::Mito(MitoConfig {
-                num_workers: 8,
-                auto_flush_interval: Duration::from_secs(3600),
-                scan_parallelism: 0,
-                global_write_buffer_size: ReadableSize::gb(1),
-                global_write_buffer_reject_size: ReadableSize::gb(2),
-                sst_meta_cache_size: ReadableSize::mb(128),
-                vector_cache_size: ReadableSize::mb(512),
-                page_cache_size: ReadableSize::mb(512),
-                max_background_jobs: 4,
-                ..Default::default()
-            })],
+            region_engine: vec![
+                RegionEngineConfig::Mito(MitoConfig {
+                    num_workers: 8,
+                    auto_flush_interval: Duration::from_secs(3600),
+                    scan_parallelism: 0,
+                    global_write_buffer_size: ReadableSize::gb(1),
+                    global_write_buffer_reject_size: ReadableSize::gb(2),
+                    sst_meta_cache_size: ReadableSize::mb(128),
+                    vector_cache_size: ReadableSize::mb(512),
+                    page_cache_size: ReadableSize::mb(512),
+                    max_background_jobs: 4,
+                    ..Default::default()
+                }),
+                RegionEngineConfig::File(EngineConfig {}),
+            ],
             logging: LoggingOptions {
                 level: Some("info".to_string()),
-                otlp_endpoint: Some("".to_string()),
+                otlp_endpoint: Some(DEFAULT_OTLP_ENDPOINT.to_string()),
                 tracing_sample_ratio: Some(Default::default()),
                 ..Default::default()
             },
@@ -115,9 +118,8 @@ fn test_load_frontend_example_config() {
             .unwrap();
     let expected = GreptimeOptions::<FrontendOptions> {
         runtime: RuntimeOptions {
-            read_rt_size: 8,
-            write_rt_size: 8,
-            bg_rt_size: 4,
+            global_rt_size: 8,
+            compact_rt_size: 4,
         },
         component: FrontendOptions {
             default_timezone: Some("UTC".to_string()),
@@ -134,7 +136,7 @@ fn test_load_frontend_example_config() {
             }),
             logging: LoggingOptions {
                 level: Some("info".to_string()),
-                otlp_endpoint: Some("".to_string()),
+                otlp_endpoint: Some(DEFAULT_OTLP_ENDPOINT.to_string()),
                 tracing_sample_ratio: Some(Default::default()),
                 ..Default::default()
             },
@@ -163,17 +165,16 @@ fn test_load_metasrv_example_config() {
             .unwrap();
     let expected = GreptimeOptions::<MetasrvOptions> {
         runtime: RuntimeOptions {
-            read_rt_size: 8,
-            write_rt_size: 8,
-            bg_rt_size: 4,
+            global_rt_size: 8,
+            compact_rt_size: 4,
         },
         component: MetasrvOptions {
-            selector: SelectorType::LeaseBased,
+            selector: SelectorType::default(),
             data_home: "/tmp/metasrv/".to_string(),
             logging: LoggingOptions {
                 dir: "/tmp/greptimedb/logs".to_string(),
                 level: Some("info".to_string()),
-                otlp_endpoint: Some("".to_string()),
+                otlp_endpoint: Some(DEFAULT_OTLP_ENDPOINT.to_string()),
                 tracing_sample_ratio: Some(Default::default()),
                 ..Default::default()
             },
@@ -196,36 +197,38 @@ fn test_load_standalone_example_config() {
             .unwrap();
     let expected = GreptimeOptions::<StandaloneOptions> {
         runtime: RuntimeOptions {
-            read_rt_size: 8,
-            write_rt_size: 8,
-            bg_rt_size: 4,
+            global_rt_size: 8,
+            compact_rt_size: 4,
         },
         component: StandaloneOptions {
             default_timezone: Some("UTC".to_string()),
-            wal: StandaloneWalConfig::RaftEngine(RaftEngineConfig {
+            wal: DatanodeWalConfig::RaftEngine(RaftEngineConfig {
                 dir: Some("/tmp/greptimedb/wal".to_string()),
                 sync_period: Some(Duration::from_secs(10)),
                 ..Default::default()
             }),
-            region_engine: vec![RegionEngineConfig::Mito(MitoConfig {
-                num_workers: 8,
-                auto_flush_interval: Duration::from_secs(3600),
-                scan_parallelism: 0,
-                global_write_buffer_size: ReadableSize::gb(1),
-                global_write_buffer_reject_size: ReadableSize::gb(2),
-                sst_meta_cache_size: ReadableSize::mb(128),
-                vector_cache_size: ReadableSize::mb(512),
-                page_cache_size: ReadableSize::mb(512),
-                max_background_jobs: 4,
-                ..Default::default()
-            })],
+            region_engine: vec![
+                RegionEngineConfig::Mito(MitoConfig {
+                    num_workers: 8,
+                    auto_flush_interval: Duration::from_secs(3600),
+                    scan_parallelism: 0,
+                    global_write_buffer_size: ReadableSize::gb(1),
+                    global_write_buffer_reject_size: ReadableSize::gb(2),
+                    sst_meta_cache_size: ReadableSize::mb(128),
+                    vector_cache_size: ReadableSize::mb(512),
+                    page_cache_size: ReadableSize::mb(512),
+                    max_background_jobs: 4,
+                    ..Default::default()
+                }),
+                RegionEngineConfig::File(EngineConfig {}),
+            ],
             storage: StorageConfig {
                 data_home: "/tmp/greptimedb/".to_string(),
                 ..Default::default()
             },
             logging: LoggingOptions {
                 level: Some("info".to_string()),
-                otlp_endpoint: Some("".to_string()),
+                otlp_endpoint: Some(DEFAULT_OTLP_ENDPOINT.to_string()),
                 tracing_sample_ratio: Some(Default::default()),
                 ..Default::default()
             },

src/common/catalog/src/consts.rs

@@ -14,6 +14,7 @@
 
 pub const SYSTEM_CATALOG_NAME: &str = "system";
 pub const INFORMATION_SCHEMA_NAME: &str = "information_schema";
+pub const PG_CATALOG_NAME: &str = "pg_catalog";
 pub const SYSTEM_CATALOG_TABLE_NAME: &str = "system_catalog";
 pub const DEFAULT_CATALOG_NAME: &str = "greptime";
 pub const DEFAULT_SCHEMA_NAME: &str = "public";
@@ -93,8 +94,18 @@ pub const INFORMATION_SCHEMA_REGION_PEERS_TABLE_ID: u32 = 29;
 pub const INFORMATION_SCHEMA_TABLE_CONSTRAINTS_TABLE_ID: u32 = 30;
 /// id for information_schema.cluster_info
 pub const INFORMATION_SCHEMA_CLUSTER_INFO_TABLE_ID: u32 = 31;
+/// id for information_schema.VIEWS
+pub const INFORMATION_SCHEMA_VIEW_TABLE_ID: u32 = 32;
+/// id for information_schema.FLOWS
+pub const INFORMATION_SCHEMA_FLOW_TABLE_ID: u32 = 33;
 /// ----- End of information_schema tables -----
 
+/// ----- Begin of pg_catalog tables -----
+pub const PG_CATALOG_PG_CLASS_TABLE_ID: u32 = 256;
+pub const PG_CATALOG_PG_TYPE_TABLE_ID: u32 = 257;
+pub const PG_CATALOG_PG_NAMESPACE_TABLE_ID: u32 = 258;
+
+/// ----- End of pg_catalog tables -----
 pub const MITO_ENGINE: &str = "mito";
 pub const MITO2_ENGINE: &str = "mito2";
 pub const METRIC_ENGINE: &str = "metric";
@@ -108,3 +119,7 @@ pub const FILE_ENGINE: &str = "file";
 pub const SEMANTIC_TYPE_PRIMARY_KEY: &str = "TAG";
 pub const SEMANTIC_TYPE_FIELD: &str = "FIELD";
 pub const SEMANTIC_TYPE_TIME_INDEX: &str = "TIMESTAMP";
+
+pub fn is_readonly_schema(schema: &str) -> bool {
+    matches!(schema, INFORMATION_SCHEMA_NAME)
+}