docs: json datatype rfc (#4515)

* docs: json datatype rfc

* docs: turn to a jsonb proposal

* chore: fix typo

* feat: add store and query process

* fix: typo

* fix: use query nodes instead of query plans

* feat: a detailed overview of query

* fix: grammar

* fix: use independent cast function

* fix: unify cast function

* fix: refine, make statements clear

* docs: update rfc according to impl

* docs: refine

* docs: fix wrong arrows

* docs: refine

* docs: fix some errors qaq

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

@@ -42,7 +42,7 @@ runs:
     - name: Install rust toolchain
       uses: actions-rust-lang/setup-rust-toolchain@v1
       with:
-        targets: ${{ inputs.arch }}
+        target: ${{ inputs.arch }}
 
     - name: Start etcd # For integration tests.
       if: ${{ inputs.disable-run-tests == 'false' }}

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

@@ -27,7 +27,7 @@ runs:
     - name: Install rust toolchain
       uses: actions-rust-lang/setup-rust-toolchain@v1
       with:
-        targets: ${{ inputs.arch }}
+        target: ${{ inputs.arch }}
         components: llvm-tools-preview
 
     - name: Rust Cache

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

@@ -18,6 +18,8 @@ runs:
         --set replicaCount=${{ inputs.etcd-replicas }} \
         --set resources.requests.cpu=50m \
         --set resources.requests.memory=128Mi \
+        --set resources.limits.cpu=1000m \
+        --set resources.limits.memory=2Gi \
         --set auth.rbac.create=false \
         --set auth.rbac.token.enabled=false \
         --set persistence.size=2Gi \

.github/actions/start-runner/action.yml

@@ -38,7 +38,7 @@ runs:
   steps:
     - name: Configure AWS credentials
       if: startsWith(inputs.runner, 'ec2')
-      uses: aws-actions/configure-aws-credentials@v2
+      uses: aws-actions/configure-aws-credentials@v4
       with:
         aws-access-key-id: ${{ inputs.aws-access-key-id }}
         aws-secret-access-key: ${{ inputs.aws-secret-access-key }}

.github/actions/stop-runner/action.yml

@@ -25,7 +25,7 @@ runs:
   steps:
     - name: Configure AWS credentials
       if: ${{ inputs.label && inputs.ec2-instance-id }}
-      uses: aws-actions/configure-aws-credentials@v2
+      uses: aws-actions/configure-aws-credentials@v4
       with:
         aws-access-key-id: ${{ inputs.aws-access-key-id }}
         aws-secret-access-key: ${{ inputs.aws-secret-access-key }}

.github/workflows/develop.yml

@@ -616,8 +616,8 @@ jobs:
         with:
           # Shares across multiple jobs
           shared-key: "check-rust-fmt"
-      - name: Run cargo fmt
-        run: cargo fmt --all -- --check
+      - name: Check format
+        run: make fmt-check
 
   clippy:
     name: Clippy

.github/workflows/release.yml

@@ -33,6 +33,7 @@ on:
         description: The runner uses to build linux-arm64 artifacts
         default: ec2-c6g.4xlarge-arm64
         options:
+          - ubuntu-2204-32-cores-arm
           - ec2-c6g.xlarge-arm64 # 4C8G
           - ec2-c6g.2xlarge-arm64 # 8C16G
           - ec2-c6g.4xlarge-arm64 # 16C32G
@@ -98,16 +99,6 @@ permissions:
   contents: write # Allows the action to create a release.
 
 jobs:
-  check-builder-rust-version:
-    name: Check rust version in builder
-    runs-on: ubuntu-20.04
-    steps:
-      - uses: actions/checkout@v4
-      - name: Check Rust toolchain version
-        shell: bash
-        run: |
-          ./scripts/check-builder-rust-version.sh
-
   allocate-runners:
     name: Allocate runners
     if: ${{ github.repository == 'GreptimeTeam/greptimedb' }}
@@ -132,6 +123,11 @@ jobs:
         with:
           fetch-depth: 0
 
+      - name: Check Rust toolchain version
+        shell: bash
+        run: |
+          ./scripts/check-builder-rust-version.sh
+
       # The create-version will create a global variable named 'version' in the global workflows.
       # - If it's a tag push release, the version is the tag name(${{ github.ref_name }});
       # - If it's a scheduled release, the version is '${{ env.NEXT_RELEASE_VERSION }}-nightly-$buildTime', like v0.2.0-nigthly-20230313;
@@ -256,7 +252,8 @@ jobs:
           cargo-profile: ${{ env.CARGO_PROFILE }}
           features: ${{ matrix.features }}
           version: ${{ needs.allocate-runners.outputs.version }}
-          disable-run-tests: ${{ env.DISABLE_RUN_TESTS }}
+          # We decide to disable the integration tests on macOS because it's unnecessary and time-consuming.
+          disable-run-tests: true
           artifacts-dir: ${{ matrix.artifacts-dir-prefix }}-${{ needs.allocate-runners.outputs.version }}
 
       - name: Set build macos result

Cargo.toml

@@ -64,7 +64,7 @@ members = [
 resolver = "2"
 
 [workspace.package]
-version = "0.9.2"
+version = "0.9.3"
 edition = "2021"
 license = "Apache-2.0"
 
@@ -90,7 +90,7 @@ aquamarine = "0.3"
 arrow = { version = "51.0.0", features = ["prettyprint"] }
 arrow-array = { version = "51.0.0", default-features = false, features = ["chrono-tz"] }
 arrow-flight = "51.0"
-arrow-ipc = { version = "51.0.0", default-features = false, features = ["lz4"] }
+arrow-ipc = { version = "51.0.0", default-features = false, features = ["lz4", "zstd"] }
 arrow-schema = { version = "51.0", features = ["serde"] }
 async-stream = "0.3"
 async-trait = "0.1"
@@ -99,7 +99,7 @@ base64 = "0.21"
 bigdecimal = "0.4.2"
 bitflags = "2.4.1"
 bytemuck = "1.12"
-bytes = { version = "1.5", features = ["serde"] }
+bytes = { version = "1.7", features = ["serde"] }
 chrono = { version = "0.4", features = ["serde"] }
 clap = { version = "4.4", features = ["derive"] }
 config = "0.13.0"
@@ -120,10 +120,11 @@ 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 = "c437b55725b7f5224fe9d46db21072b4a682ee4b" }
+greptime-proto = { git = "https://github.com/GreptimeTeam/greptime-proto.git", rev = "973f49cde88a582fb65755cc572ebcf6fb93ccf7" }
 humantime = "2.1"
 humantime-serde = "1.1"
 itertools = "0.10"
+jsonb = { git = "https://github.com/CookiePieWw/jsonb.git", rev = "d0166c130fce903bf6c58643417a3173a6172d31", default-features = false }
 lazy_static = "1.4"
 meter-core = { git = "https://github.com/GreptimeTeam/greptime-meter.git", rev = "80eb97c24c88af4dd9a86f8bbaf50e741d4eb8cd" }
 mockall = "0.11.4"
@@ -135,6 +136,7 @@ opentelemetry-proto = { version = "0.5", features = [
     "gen-tonic",
     "metrics",
     "trace",
+    "with-serde",
 ] }
 parquet = { version = "51.0.0", default-features = false, features = ["arrow", "async", "object_store"] }
 paste = "1.0"
@@ -164,6 +166,7 @@ serde = { version = "1.0", features = ["derive"] }
 serde_json = { version = "1.0", features = ["float_roundtrip"] }
 serde_with = "3"
 shadow-rs = "0.31"
+similar-asserts = "1.6.0"
 smallvec = { version = "1", features = ["serde"] }
 snafu = "0.8"
 sysinfo = "0.30"

Makefile

@@ -191,6 +191,7 @@ fix-clippy: ## Fix clippy violations.
 .PHONY: fmt-check
 fmt-check: ## Check code format.
 	cargo fmt --all -- --check
+	python3 scripts/check-snafu.py
 
 .PHONY: start-etcd
 start-etcd: ## Start single node etcd for testing purpose.
@@ -220,7 +221,7 @@ config-docs: ## Generate configuration documentation from toml files.
 	docker run --rm \
     -v ${PWD}:/greptimedb \
     -w /greptimedb/config \
-    toml2docs/toml2docs:v0.1.1 \
+    toml2docs/toml2docs:v0.1.3 \
     -p '##' \
     -t ./config-docs-template.md \
     -o ./config.md

README.md

@@ -74,7 +74,7 @@ Our core developers have been building time-series data platforms for years. Bas
 
 * **Compatible with InfluxDB, Prometheus and more protocols**
 
-  Widely adopted database protocols and APIs, including MySQL, PostgreSQL, and Prometheus Remote Storage, etc. [Read more](https://docs.greptime.com/user-guide/clients/overview).
+  Widely adopted database protocols and APIs, including MySQL, PostgreSQL, and Prometheus Remote Storage, etc. [Read more](https://docs.greptime.com/user-guide/protocols/overview).
 
 ## Try GreptimeDB
 

config/config.md

@@ -14,7 +14,9 @@
 | --- | -----| ------- | ----------- |
 | `mode` | String | `standalone` | The running mode of the datanode. It can be `standalone` or `distributed`. |
 | `enable_telemetry` | Bool | `true` | Enable telemetry to collect anonymous usage data. |
-| `default_timezone` | String | `None` | The default timezone of the server. |
+| `default_timezone` | String | Unset | The default timezone of the server. |
+| `init_regions_in_background` | Bool | `false` | Initialize all regions in the background during the startup.<br/>By default, it provides services after all regions have been initialized. |
+| `init_regions_parallelism` | Integer | `16` | Parallelism of initializing regions. |
 | `runtime` | -- | -- | The runtime options. |
 | `runtime.global_rt_size` | Integer | `8` | The number of threads to execute the runtime for global read operations. |
 | `runtime.compact_rt_size` | Integer | `4` | The number of threads to execute the runtime for global write operations. |
@@ -27,8 +29,8 @@
 | `grpc.runtime_size` | Integer | `8` | The number of server worker threads. |
 | `grpc.tls` | -- | -- | gRPC server TLS options, see `mysql.tls` section. |
 | `grpc.tls.mode` | String | `disable` | TLS mode. |
-| `grpc.tls.cert_path` | String | `None` | Certificate file path. |
-| `grpc.tls.key_path` | String | `None` | Private key file path. |
+| `grpc.tls.cert_path` | String | Unset | Certificate file path. |
+| `grpc.tls.key_path` | String | Unset | 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. |
 | `mysql` | -- | -- | MySQL server options. |
 | `mysql.enable` | Bool | `true` | Whether to enable. |
@@ -36,8 +38,8 @@
 | `mysql.runtime_size` | Integer | `2` | The number of server worker threads. |
 | `mysql.tls` | -- | -- | -- |
 | `mysql.tls.mode` | String | `disable` | TLS mode, refer to https://www.postgresql.org/docs/current/libpq-ssl.html<br/>- `disable` (default value)<br/>- `prefer`<br/>- `require`<br/>- `verify-ca`<br/>- `verify-full` |
-| `mysql.tls.cert_path` | String | `None` | Certificate file path. |
-| `mysql.tls.key_path` | String | `None` | Private key file path. |
+| `mysql.tls.cert_path` | String | Unset | Certificate file path. |
+| `mysql.tls.key_path` | String | Unset | Private key file path. |
 | `mysql.tls.watch` | Bool | `false` | Watch for Certificate and key file change and auto reload |
 | `postgres` | -- | -- | PostgresSQL server options. |
 | `postgres.enable` | Bool | `true` | Whether to enable |
@@ -45,8 +47,8 @@
 | `postgres.runtime_size` | Integer | `2` | The number of server worker threads. |
 | `postgres.tls` | -- | -- | PostgresSQL server TLS options, see `mysql.tls` section. |
 | `postgres.tls.mode` | String | `disable` | TLS mode. |
-| `postgres.tls.cert_path` | String | `None` | Certificate file path. |
-| `postgres.tls.key_path` | String | `None` | Private key file path. |
+| `postgres.tls.cert_path` | String | Unset | Certificate file path. |
+| `postgres.tls.key_path` | String | Unset | Private key file path. |
 | `postgres.tls.watch` | Bool | `false` | Watch for Certificate and key file change and auto reload |
 | `opentsdb` | -- | -- | OpenTSDB protocol options. |
 | `opentsdb.enable` | Bool | `true` | Whether to enable OpenTSDB put in HTTP API. |
@@ -57,7 +59,7 @@
 | `prom_store.with_metric_engine` | Bool | `true` | Whether to store the data from Prometheus remote write in metric engine. |
 | `wal` | -- | -- | The WAL options. |
 | `wal.provider` | String | `raft_engine` | The provider of the WAL.<br/>- `raft_engine`: the wal is stored in the local file system by raft-engine.<br/>- `kafka`: it's remote wal that data is stored in Kafka. |
-| `wal.dir` | String | `None` | The directory to store the WAL files.<br/>**It's only used when the provider is `raft_engine`**. |
+| `wal.dir` | String | Unset | The directory to store the WAL files.<br/>**It's only used when the provider is `raft_engine`**. |
 | `wal.file_size` | String | `256MB` | The size of the WAL segment file.<br/>**It's only used when the provider is `raft_engine`**. |
 | `wal.purge_threshold` | String | `4GB` | The threshold of the WAL size to trigger a flush.<br/>**It's only used when the provider is `raft_engine`**. |
 | `wal.purge_interval` | String | `10m` | The interval to trigger a flush.<br/>**It's only used when the provider is `raft_engine`**. |
@@ -66,6 +68,7 @@
 | `wal.enable_log_recycle` | Bool | `true` | Whether to reuse logically truncated log files.<br/>**It's only used when the provider is `raft_engine`**. |
 | `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.recovery_parallelism` | Integer | `2` | Parallelism during WAL recovery. |
 | `wal.broker_endpoints` | Array | -- | The Kafka broker endpoints.<br/>**It's only used when the provider is `kafka`**. |
 | `wal.auto_create_topics` | Bool | `true` | Automatically create topics for WAL.<br/>Set to `true` to automatically create topics for WAL.<br/>Otherwise, use topics named `topic_name_prefix_[0..num_topics)` |
 | `wal.num_topics` | Integer | `64` | Number of topics.<br/>**It's only used when the provider is `kafka`**. |
@@ -88,22 +91,22 @@
 | `storage` | -- | -- | The data storage options. |
 | `storage.data_home` | String | `/tmp/greptimedb/` | The working home directory. |
 | `storage.type` | String | `File` | The storage type used to store the data.<br/>- `File`: the data is stored in the local file system.<br/>- `S3`: the data is stored in the S3 object storage.<br/>- `Gcs`: the data is stored in the Google Cloud Storage.<br/>- `Azblob`: the data is stored in the Azure Blob Storage.<br/>- `Oss`: the data is stored in the Aliyun OSS. |
-| `storage.cache_path` | String | `None` | Cache configuration for object storage such as 'S3' etc.<br/>The local file cache directory. |
-| `storage.cache_capacity` | String | `None` | The local file cache capacity in bytes. |
-| `storage.bucket` | String | `None` | The S3 bucket name.<br/>**It's only used when the storage type is `S3`, `Oss` and `Gcs`**. |
-| `storage.root` | String | `None` | The S3 data will be stored in the specified prefix, for example, `s3://${bucket}/${root}`.<br/>**It's only used when the storage type is `S3`, `Oss` and `Azblob`**. |
-| `storage.access_key_id` | String | `None` | The access key id of the aws account.<br/>It's **highly recommended** to use AWS IAM roles instead of hardcoding the access key id and secret key.<br/>**It's only used when the storage type is `S3` and `Oss`**. |
-| `storage.secret_access_key` | String | `None` | The secret access key of the aws account.<br/>It's **highly recommended** to use AWS IAM roles instead of hardcoding the access key id and secret key.<br/>**It's only used when the storage type is `S3`**. |
-| `storage.access_key_secret` | String | `None` | The secret access key of the aliyun account.<br/>**It's only used when the storage type is `Oss`**. |
-| `storage.account_name` | String | `None` | The account key of the azure account.<br/>**It's only used when the storage type is `Azblob`**. |
-| `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`**. |
-| `storage.region` | String | `None` | The region of the S3 service.<br/>**It's only used when the storage type is `S3`, `Oss`, `Gcs` and `Azblob`**. |
+| `storage.cache_path` | String | Unset | Cache configuration for object storage such as 'S3' etc.<br/>The local file cache directory. |
+| `storage.cache_capacity` | String | Unset | The local file cache capacity in bytes. |
+| `storage.bucket` | String | Unset | The S3 bucket name.<br/>**It's only used when the storage type is `S3`, `Oss` and `Gcs`**. |
+| `storage.root` | String | Unset | The S3 data will be stored in the specified prefix, for example, `s3://${bucket}/${root}`.<br/>**It's only used when the storage type is `S3`, `Oss` and `Azblob`**. |
+| `storage.access_key_id` | String | Unset | The access key id of the aws account.<br/>It's **highly recommended** to use AWS IAM roles instead of hardcoding the access key id and secret key.<br/>**It's only used when the storage type is `S3` and `Oss`**. |
+| `storage.secret_access_key` | String | Unset | The secret access key of the aws account.<br/>It's **highly recommended** to use AWS IAM roles instead of hardcoding the access key id and secret key.<br/>**It's only used when the storage type is `S3`**. |
+| `storage.access_key_secret` | String | Unset | The secret access key of the aliyun account.<br/>**It's only used when the storage type is `Oss`**. |
+| `storage.account_name` | String | Unset | The account key of the azure account.<br/>**It's only used when the storage type is `Azblob`**. |
+| `storage.account_key` | String | Unset | The account key of the azure account.<br/>**It's only used when the storage type is `Azblob`**. |
+| `storage.scope` | String | Unset | The scope of the google cloud storage.<br/>**It's only used when the storage type is `Gcs`**. |
+| `storage.credential_path` | String | Unset | The credential path of the google cloud storage.<br/>**It's only used when the storage type is `Gcs`**. |
+| `storage.credential` | String | Unset | The credential of the google cloud storage.<br/>**It's only used when the storage type is `Gcs`**. |
+| `storage.container` | String | Unset | The container of the azure account.<br/>**It's only used when the storage type is `Azblob`**. |
+| `storage.sas_token` | String | Unset | The sas token of the azure account.<br/>**It's only used when the storage type is `Azblob`**. |
+| `storage.endpoint` | String | Unset | The endpoint of the S3 service.<br/>**It's only used when the storage type is `S3`, `Oss`, `Gcs` and `Azblob`**. |
+| `storage.region` | String | Unset | The region of the S3 service.<br/>**It's only used when the storage type is `S3`, `Oss`, `Gcs` and `Azblob`**. |
 | `[[region_engine]]` | -- | -- | The region engine options. You can configure multiple region engines. |
 | `region_engine.mito` | -- | -- | The Mito engine options. |
 | `region_engine.mito.num_workers` | Integer | `8` | Number of region workers. |
@@ -113,20 +116,21 @@
 | `region_engine.mito.compress_manifest` | Bool | `false` | Whether to compress manifest and checkpoint file by gzip (default false). |
 | `region_engine.mito.max_background_jobs` | Integer | `4` | Max number of running background jobs |
 | `region_engine.mito.auto_flush_interval` | String | `1h` | Interval to auto flush a region if it has not flushed yet. |
-| `region_engine.mito.global_write_buffer_size` | String | `1GB` | Global write buffer size for all regions. If not set, it's default to 1/8 of OS memory with a max limitation of 1GB. |
-| `region_engine.mito.global_write_buffer_reject_size` | String | `2GB` | Global write buffer size threshold to reject write requests. If not set, it's default to 2 times of `global_write_buffer_size` |
-| `region_engine.mito.sst_meta_cache_size` | String | `128MB` | Cache size for SST metadata. Setting it to 0 to disable the cache.<br/>If not set, it's default to 1/32 of OS memory with a max limitation of 128MB. |
-| `region_engine.mito.vector_cache_size` | String | `512MB` | Cache size for vectors and arrow arrays. Setting it to 0 to disable the cache.<br/>If not set, it's default to 1/16 of OS memory with a max limitation of 512MB. |
-| `region_engine.mito.page_cache_size` | String | `512MB` | Cache size for pages of SST row groups. Setting it to 0 to disable the cache.<br/>If not set, it's default to 1/8 of OS memory. |
-| `region_engine.mito.selector_result_cache_size` | String | `512MB` | Cache size for time series selector (e.g. `last_value()`). Setting it to 0 to disable the cache.<br/>If not set, it's default to 1/16 of OS memory with a max limitation of 512MB. |
+| `region_engine.mito.global_write_buffer_size` | String | Auto | Global write buffer size for all regions. If not set, it's default to 1/8 of OS memory with a max limitation of 1GB. |
+| `region_engine.mito.global_write_buffer_reject_size` | String | Auto | Global write buffer size threshold to reject write requests. If not set, it's default to 2 times of `global_write_buffer_size`. |
+| `region_engine.mito.sst_meta_cache_size` | String | Auto | Cache size for SST metadata. Setting it to 0 to disable the cache.<br/>If not set, it's default to 1/32 of OS memory with a max limitation of 128MB. |
+| `region_engine.mito.vector_cache_size` | String | Auto | Cache size for vectors and arrow arrays. Setting it to 0 to disable the cache.<br/>If not set, it's default to 1/16 of OS memory with a max limitation of 512MB. |
+| `region_engine.mito.page_cache_size` | String | Auto | Cache size for pages of SST row groups. Setting it to 0 to disable the cache.<br/>If not set, it's default to 1/8 of OS memory. |
+| `region_engine.mito.selector_result_cache_size` | String | Auto | Cache size for time series selector (e.g. `last_value()`). Setting it to 0 to disable the cache.<br/>If not set, it's default to 1/16 of OS memory with a max limitation of 512MB. |
 | `region_engine.mito.enable_experimental_write_cache` | Bool | `false` | Whether to enable the experimental write cache. |
 | `region_engine.mito.experimental_write_cache_path` | String | `""` | File system path for write cache, defaults to `{data_home}/write_cache`. |
 | `region_engine.mito.experimental_write_cache_size` | String | `512MB` | Capacity for write cache. |
-| `region_engine.mito.experimental_write_cache_ttl` | String | `None` | TTL for write cache. |
+| `region_engine.mito.experimental_write_cache_ttl` | String | Unset | TTL for write cache. |
 | `region_engine.mito.sst_write_buffer_size` | String | `8MB` | Buffer size for SST writing. |
 | `region_engine.mito.scan_parallelism` | Integer | `0` | Parallelism to scan a region (default: 1/4 of cpu cores).<br/>- `0`: using the default value (1/4 of cpu cores).<br/>- `1`: scan in current thread.<br/>- `n`: scan in parallelism n. |
 | `region_engine.mito.parallel_scan_channel_size` | Integer | `32` | Capacity of the channel to send data from parallel scan tasks to the main task. |
 | `region_engine.mito.allow_stale_entries` | Bool | `false` | Whether to allow stale WAL entries read during replay. |
+| `region_engine.mito.min_compaction_interval` | String | `0m` | Minimum time interval between two compactions.<br/>To align with the old behavior, the default value is 0 (no restrictions). |
 | `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. |
@@ -150,23 +154,24 @@
 | `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.dir` | String | `/tmp/greptimedb/logs` | The directory to store the log files. If set to empty, logs will not be written to files. |
+| `logging.level` | String | Unset | 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.log_format` | String | `text` | The log format. Can be `text`/`json`. |
 | `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` | -- |
 | `export_metrics` | -- | -- | The datanode can export its metrics and send to Prometheus compatible service (e.g. send to `greptimedb` itself) from remote-write API.<br/>This is only used for `greptimedb` to export its own metrics internally. It's different from prometheus scrape. |
 | `export_metrics.enable` | Bool | `false` | whether enable export metrics. |
 | `export_metrics.write_interval` | String | `30s` | The interval of export metrics. |
-| `export_metrics.self_import` | -- | -- | For `standalone` mode, `self_import` is recommend to collect metrics generated by itself<br/>You must create the database before enabling it. |
-| `export_metrics.self_import.db` | String | `None` | -- |
+| `export_metrics.self_import` | -- | -- | For `standalone` mode, `self_import` is recommended to collect metrics generated by itself<br/>You must create the database before enabling it. |
+| `export_metrics.self_import.db` | String | Unset | -- |
 | `export_metrics.remote_write` | -- | -- | -- |
 | `export_metrics.remote_write.url` | String | `""` | The url the metrics send to. The url example can be: `http://127.0.0.1:4000/v1/prometheus/write?db=greptime_metrics`. |
 | `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. |
+| `tracing.tokio_console_addr` | String | Unset | The tokio console address. |
 
 
 ## Distributed Mode
@@ -175,7 +180,7 @@
 
 | Key | Type | Default | Descriptions |
 | --- | -----| ------- | ----------- |
-| `default_timezone` | String | `None` | The default timezone of the server. |
+| `default_timezone` | String | Unset | The default timezone of the server. |
 | `runtime` | -- | -- | The runtime options. |
 | `runtime.global_rt_size` | Integer | `8` | The number of threads to execute the runtime for global read operations. |
 | `runtime.compact_rt_size` | Integer | `4` | The number of threads to execute the runtime for global write operations. |
@@ -192,8 +197,8 @@
 | `grpc.runtime_size` | Integer | `8` | The number of server worker threads. |
 | `grpc.tls` | -- | -- | gRPC server TLS options, see `mysql.tls` section. |
 | `grpc.tls.mode` | String | `disable` | TLS mode. |
-| `grpc.tls.cert_path` | String | `None` | Certificate file path. |
-| `grpc.tls.key_path` | String | `None` | Private key file path. |
+| `grpc.tls.cert_path` | String | Unset | Certificate file path. |
+| `grpc.tls.key_path` | String | Unset | 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. |
 | `mysql` | -- | -- | MySQL server options. |
 | `mysql.enable` | Bool | `true` | Whether to enable. |
@@ -201,8 +206,8 @@
 | `mysql.runtime_size` | Integer | `2` | The number of server worker threads. |
 | `mysql.tls` | -- | -- | -- |
 | `mysql.tls.mode` | String | `disable` | TLS mode, refer to https://www.postgresql.org/docs/current/libpq-ssl.html<br/>- `disable` (default value)<br/>- `prefer`<br/>- `require`<br/>- `verify-ca`<br/>- `verify-full` |
-| `mysql.tls.cert_path` | String | `None` | Certificate file path. |
-| `mysql.tls.key_path` | String | `None` | Private key file path. |
+| `mysql.tls.cert_path` | String | Unset | Certificate file path. |
+| `mysql.tls.key_path` | String | Unset | Private key file path. |
 | `mysql.tls.watch` | Bool | `false` | Watch for Certificate and key file change and auto reload |
 | `postgres` | -- | -- | PostgresSQL server options. |
 | `postgres.enable` | Bool | `true` | Whether to enable |
@@ -210,8 +215,8 @@
 | `postgres.runtime_size` | Integer | `2` | The number of server worker threads. |
 | `postgres.tls` | -- | -- | PostgresSQL server TLS options, see `mysql.tls` section. |
 | `postgres.tls.mode` | String | `disable` | TLS mode. |
-| `postgres.tls.cert_path` | String | `None` | Certificate file path. |
-| `postgres.tls.key_path` | String | `None` | Private key file path. |
+| `postgres.tls.cert_path` | String | Unset | Certificate file path. |
+| `postgres.tls.key_path` | String | Unset | Private key file path. |
 | `postgres.tls.watch` | Bool | `false` | Watch for Certificate and key file change and auto reload |
 | `opentsdb` | -- | -- | OpenTSDB protocol options. |
 | `opentsdb.enable` | Bool | `true` | Whether to enable OpenTSDB put in HTTP API. |
@@ -235,23 +240,24 @@
 | `datanode.client.connect_timeout` | String | `10s` | -- |
 | `datanode.client.tcp_nodelay` | Bool | `true` | -- |
 | `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.dir` | String | `/tmp/greptimedb/logs` | The directory to store the log files. If set to empty, logs will not be written to files. |
+| `logging.level` | String | Unset | 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.log_format` | String | `text` | The log format. Can be `text`/`json`. |
 | `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` | -- |
 | `export_metrics` | -- | -- | The datanode can export its metrics and send to Prometheus compatible service (e.g. send to `greptimedb` itself) from remote-write API.<br/>This is only used for `greptimedb` to export its own metrics internally. It's different from prometheus scrape. |
 | `export_metrics.enable` | Bool | `false` | whether enable export metrics. |
 | `export_metrics.write_interval` | String | `30s` | The interval of export metrics. |
 | `export_metrics.self_import` | -- | -- | For `standalone` mode, `self_import` is recommend to collect metrics generated by itself<br/>You must create the database before enabling it. |
-| `export_metrics.self_import.db` | String | `None` | -- |
+| `export_metrics.self_import.db` | String | Unset | -- |
 | `export_metrics.remote_write` | -- | -- | -- |
 | `export_metrics.remote_write.url` | String | `""` | The url the metrics send to. The url example can be: `http://127.0.0.1:4000/v1/prometheus/write?db=greptime_metrics`. |
 | `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. |
+| `tracing.tokio_console_addr` | String | Unset | The tokio console address. |
 
 
 ### Metasrv
@@ -299,23 +305,24 @@
 | `wal.backoff_base` | Integer | `2` | Exponential backoff rate, i.e. next backoff = base * current backoff. |
 | `wal.backoff_deadline` | String | `5mins` | Stop reconnecting if the total wait time reaches the deadline. If this config is missing, the reconnecting won't terminate. |
 | `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.dir` | String | `/tmp/greptimedb/logs` | The directory to store the log files. If set to empty, logs will not be written to files. |
+| `logging.level` | String | Unset | 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.log_format` | String | `text` | The log format. Can be `text`/`json`. |
 | `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` | -- |
 | `export_metrics` | -- | -- | The datanode can export its metrics and send to Prometheus compatible service (e.g. send to `greptimedb` itself) from remote-write API.<br/>This is only used for `greptimedb` to export its own metrics internally. It's different from prometheus scrape. |
 | `export_metrics.enable` | Bool | `false` | whether enable export metrics. |
 | `export_metrics.write_interval` | String | `30s` | The interval of export metrics. |
 | `export_metrics.self_import` | -- | -- | For `standalone` mode, `self_import` is recommend to collect metrics generated by itself<br/>You must create the database before enabling it. |
-| `export_metrics.self_import.db` | String | `None` | -- |
+| `export_metrics.self_import.db` | String | Unset | -- |
 | `export_metrics.remote_write` | -- | -- | -- |
 | `export_metrics.remote_write.url` | String | `""` | The url the metrics send to. The url example can be: `http://127.0.0.1:4000/v1/prometheus/write?db=greptime_metrics`. |
 | `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. |
+| `tracing.tokio_console_addr` | String | Unset | The tokio console address. |
 
 
 ### Datanode
@@ -323,16 +330,16 @@
 | Key | Type | Default | Descriptions |
 | --- | -----| ------- | ----------- |
 | `mode` | String | `standalone` | The running mode of the datanode. It can be `standalone` or `distributed`. |
-| `node_id` | Integer | `None` | The datanode identifier and should be unique in the cluster. |
+| `node_id` | Integer | Unset | The datanode identifier and should be unique in the cluster. |
 | `require_lease_before_startup` | Bool | `false` | Start services after regions have obtained leases.<br/>It will block the datanode start if it can't receive leases in the heartbeat from metasrv. |
 | `init_regions_in_background` | Bool | `false` | Initialize all regions in the background during the startup.<br/>By default, it provides services after all regions have been initialized. |
 | `enable_telemetry` | Bool | `true` | Enable telemetry to collect anonymous usage data. |
 | `init_regions_parallelism` | Integer | `16` | Parallelism of initializing regions. |
-| `rpc_addr` | String | `None` | Deprecated, use `grpc.addr` instead. |
-| `rpc_hostname` | String | `None` | Deprecated, use `grpc.hostname` instead. |
-| `rpc_runtime_size` | Integer | `None` | Deprecated, use `grpc.runtime_size` instead. |
-| `rpc_max_recv_message_size` | String | `None` | Deprecated, use `grpc.rpc_max_recv_message_size` instead. |
-| `rpc_max_send_message_size` | String | `None` | Deprecated, use `grpc.rpc_max_send_message_size` instead. |
+| `rpc_addr` | String | Unset | Deprecated, use `grpc.addr` instead. |
+| `rpc_hostname` | String | Unset | Deprecated, use `grpc.hostname` instead. |
+| `rpc_runtime_size` | Integer | Unset | Deprecated, use `grpc.runtime_size` instead. |
+| `rpc_max_recv_message_size` | String | Unset | Deprecated, use `grpc.rpc_max_recv_message_size` instead. |
+| `rpc_max_send_message_size` | String | Unset | Deprecated, use `grpc.rpc_max_send_message_size` instead. |
 | `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. |
@@ -345,8 +352,8 @@
 | `grpc.max_send_message_size` | String | `512MB` | The maximum send message size for gRPC server. |
 | `grpc.tls` | -- | -- | gRPC server TLS options, see `mysql.tls` section. |
 | `grpc.tls.mode` | String | `disable` | TLS mode. |
-| `grpc.tls.cert_path` | String | `None` | Certificate file path. |
-| `grpc.tls.key_path` | String | `None` | Private key file path. |
+| `grpc.tls.cert_path` | String | Unset | Certificate file path. |
+| `grpc.tls.key_path` | String | Unset | 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.global_rt_size` | Integer | `8` | The number of threads to execute the runtime for global read operations. |
@@ -366,7 +373,7 @@
 | `meta_client.metadata_cache_tti` | String | `5m` | -- |
 | `wal` | -- | -- | The WAL options. |
 | `wal.provider` | String | `raft_engine` | The provider of the WAL.<br/>- `raft_engine`: the wal is stored in the local file system by raft-engine.<br/>- `kafka`: it's remote wal that data is stored in Kafka. |
-| `wal.dir` | String | `None` | The directory to store the WAL files.<br/>**It's only used when the provider is `raft_engine`**. |
+| `wal.dir` | String | Unset | The directory to store the WAL files.<br/>**It's only used when the provider is `raft_engine`**. |
 | `wal.file_size` | String | `256MB` | The size of the WAL segment file.<br/>**It's only used when the provider is `raft_engine`**. |
 | `wal.purge_threshold` | String | `4GB` | The threshold of the WAL size to trigger a flush.<br/>**It's only used when the provider is `raft_engine`**. |
 | `wal.purge_interval` | String | `10m` | The interval to trigger a flush.<br/>**It's only used when the provider is `raft_engine`**. |
@@ -375,6 +382,7 @@
 | `wal.enable_log_recycle` | Bool | `true` | Whether to reuse logically truncated log files.<br/>**It's only used when the provider is `raft_engine`**. |
 | `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.recovery_parallelism` | Integer | `2` | Parallelism during WAL recovery. |
 | `wal.broker_endpoints` | Array | -- | The Kafka broker endpoints.<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`**. |
@@ -387,22 +395,22 @@
 | `storage` | -- | -- | The data storage options. |
 | `storage.data_home` | String | `/tmp/greptimedb/` | The working home directory. |
 | `storage.type` | String | `File` | The storage type used to store the data.<br/>- `File`: the data is stored in the local file system.<br/>- `S3`: the data is stored in the S3 object storage.<br/>- `Gcs`: the data is stored in the Google Cloud Storage.<br/>- `Azblob`: the data is stored in the Azure Blob Storage.<br/>- `Oss`: the data is stored in the Aliyun OSS. |
-| `storage.cache_path` | String | `None` | Cache configuration for object storage such as 'S3' etc.<br/>The local file cache directory. |
-| `storage.cache_capacity` | String | `None` | The local file cache capacity in bytes. |
-| `storage.bucket` | String | `None` | The S3 bucket name.<br/>**It's only used when the storage type is `S3`, `Oss` and `Gcs`**. |
-| `storage.root` | String | `None` | The S3 data will be stored in the specified prefix, for example, `s3://${bucket}/${root}`.<br/>**It's only used when the storage type is `S3`, `Oss` and `Azblob`**. |
-| `storage.access_key_id` | String | `None` | The access key id of the aws account.<br/>It's **highly recommended** to use AWS IAM roles instead of hardcoding the access key id and secret key.<br/>**It's only used when the storage type is `S3` and `Oss`**. |
-| `storage.secret_access_key` | String | `None` | The secret access key of the aws account.<br/>It's **highly recommended** to use AWS IAM roles instead of hardcoding the access key id and secret key.<br/>**It's only used when the storage type is `S3`**. |
-| `storage.access_key_secret` | String | `None` | The secret access key of the aliyun account.<br/>**It's only used when the storage type is `Oss`**. |
-| `storage.account_name` | String | `None` | The account key of the azure account.<br/>**It's only used when the storage type is `Azblob`**. |
-| `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`**. |
-| `storage.region` | String | `None` | The region of the S3 service.<br/>**It's only used when the storage type is `S3`, `Oss`, `Gcs` and `Azblob`**. |
+| `storage.cache_path` | String | Unset | Cache configuration for object storage such as 'S3' etc.<br/>The local file cache directory. |
+| `storage.cache_capacity` | String | Unset | The local file cache capacity in bytes. |
+| `storage.bucket` | String | Unset | The S3 bucket name.<br/>**It's only used when the storage type is `S3`, `Oss` and `Gcs`**. |
+| `storage.root` | String | Unset | The S3 data will be stored in the specified prefix, for example, `s3://${bucket}/${root}`.<br/>**It's only used when the storage type is `S3`, `Oss` and `Azblob`**. |
+| `storage.access_key_id` | String | Unset | The access key id of the aws account.<br/>It's **highly recommended** to use AWS IAM roles instead of hardcoding the access key id and secret key.<br/>**It's only used when the storage type is `S3` and `Oss`**. |
+| `storage.secret_access_key` | String | Unset | The secret access key of the aws account.<br/>It's **highly recommended** to use AWS IAM roles instead of hardcoding the access key id and secret key.<br/>**It's only used when the storage type is `S3`**. |
+| `storage.access_key_secret` | String | Unset | The secret access key of the aliyun account.<br/>**It's only used when the storage type is `Oss`**. |
+| `storage.account_name` | String | Unset | The account key of the azure account.<br/>**It's only used when the storage type is `Azblob`**. |
+| `storage.account_key` | String | Unset | The account key of the azure account.<br/>**It's only used when the storage type is `Azblob`**. |
+| `storage.scope` | String | Unset | The scope of the google cloud storage.<br/>**It's only used when the storage type is `Gcs`**. |
+| `storage.credential_path` | String | Unset | The credential path of the google cloud storage.<br/>**It's only used when the storage type is `Gcs`**. |
+| `storage.credential` | String | Unset | The credential of the google cloud storage.<br/>**It's only used when the storage type is `Gcs`**. |
+| `storage.container` | String | Unset | The container of the azure account.<br/>**It's only used when the storage type is `Azblob`**. |
+| `storage.sas_token` | String | Unset | The sas token of the azure account.<br/>**It's only used when the storage type is `Azblob`**. |
+| `storage.endpoint` | String | Unset | The endpoint of the S3 service.<br/>**It's only used when the storage type is `S3`, `Oss`, `Gcs` and `Azblob`**. |
+| `storage.region` | String | Unset | The region of the S3 service.<br/>**It's only used when the storage type is `S3`, `Oss`, `Gcs` and `Azblob`**. |
 | `[[region_engine]]` | -- | -- | The region engine options. You can configure multiple region engines. |
 | `region_engine.mito` | -- | -- | The Mito engine options. |
 | `region_engine.mito.num_workers` | Integer | `8` | Number of region workers. |
@@ -412,20 +420,21 @@
 | `region_engine.mito.compress_manifest` | Bool | `false` | Whether to compress manifest and checkpoint file by gzip (default false). |
 | `region_engine.mito.max_background_jobs` | Integer | `4` | Max number of running background jobs |
 | `region_engine.mito.auto_flush_interval` | String | `1h` | Interval to auto flush a region if it has not flushed yet. |
-| `region_engine.mito.global_write_buffer_size` | String | `1GB` | Global write buffer size for all regions. If not set, it's default to 1/8 of OS memory with a max limitation of 1GB. |
-| `region_engine.mito.global_write_buffer_reject_size` | String | `2GB` | Global write buffer size threshold to reject write requests. If not set, it's default to 2 times of `global_write_buffer_size` |
-| `region_engine.mito.sst_meta_cache_size` | String | `128MB` | Cache size for SST metadata. Setting it to 0 to disable the cache.<br/>If not set, it's default to 1/32 of OS memory with a max limitation of 128MB. |
-| `region_engine.mito.vector_cache_size` | String | `512MB` | Cache size for vectors and arrow arrays. Setting it to 0 to disable the cache.<br/>If not set, it's default to 1/16 of OS memory with a max limitation of 512MB. |
-| `region_engine.mito.page_cache_size` | String | `512MB` | Cache size for pages of SST row groups. Setting it to 0 to disable the cache.<br/>If not set, it's default to 1/8 of OS memory. |
-| `region_engine.mito.selector_result_cache_size` | String | `512MB` | Cache size for time series selector (e.g. `last_value()`). Setting it to 0 to disable the cache.<br/>If not set, it's default to 1/16 of OS memory with a max limitation of 512MB. |
+| `region_engine.mito.global_write_buffer_size` | String | Auto | Global write buffer size for all regions. If not set, it's default to 1/8 of OS memory with a max limitation of 1GB. |
+| `region_engine.mito.global_write_buffer_reject_size` | String | Auto | Global write buffer size threshold to reject write requests. If not set, it's default to 2 times of `global_write_buffer_size` |
+| `region_engine.mito.sst_meta_cache_size` | String | Auto | Cache size for SST metadata. Setting it to 0 to disable the cache.<br/>If not set, it's default to 1/32 of OS memory with a max limitation of 128MB. |
+| `region_engine.mito.vector_cache_size` | String | Auto | Cache size for vectors and arrow arrays. Setting it to 0 to disable the cache.<br/>If not set, it's default to 1/16 of OS memory with a max limitation of 512MB. |
+| `region_engine.mito.page_cache_size` | String | Auto | Cache size for pages of SST row groups. Setting it to 0 to disable the cache.<br/>If not set, it's default to 1/8 of OS memory. |
+| `region_engine.mito.selector_result_cache_size` | String | Auto | Cache size for time series selector (e.g. `last_value()`). Setting it to 0 to disable the cache.<br/>If not set, it's default to 1/16 of OS memory with a max limitation of 512MB. |
 | `region_engine.mito.enable_experimental_write_cache` | Bool | `false` | Whether to enable the experimental write cache. |
 | `region_engine.mito.experimental_write_cache_path` | String | `""` | File system path for write cache, defaults to `{data_home}/write_cache`. |
 | `region_engine.mito.experimental_write_cache_size` | String | `512MB` | Capacity for write cache. |
-| `region_engine.mito.experimental_write_cache_ttl` | String | `None` | TTL for write cache. |
+| `region_engine.mito.experimental_write_cache_ttl` | String | Unset | TTL for write cache. |
 | `region_engine.mito.sst_write_buffer_size` | String | `8MB` | Buffer size for SST writing. |
 | `region_engine.mito.scan_parallelism` | Integer | `0` | Parallelism to scan a region (default: 1/4 of cpu cores).<br/>- `0`: using the default value (1/4 of cpu cores).<br/>- `1`: scan in current thread.<br/>- `n`: scan in parallelism n. |
 | `region_engine.mito.parallel_scan_channel_size` | Integer | `32` | Capacity of the channel to send data from parallel scan tasks to the main task. |
 | `region_engine.mito.allow_stale_entries` | Bool | `false` | Whether to allow stale WAL entries read during replay. |
+| `region_engine.mito.min_compaction_interval` | String | `0m` | Minimum time interval between two compactions.<br/>To align with the old behavior, the default value is 0 (no restrictions). |
 | `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. |
@@ -447,23 +456,24 @@
 | `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.dir` | String | `/tmp/greptimedb/logs` | The directory to store the log files. If set to empty, logs will not be written to files. |
+| `logging.level` | String | Unset | 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.log_format` | String | `text` | The log format. Can be `text`/`json`. |
 | `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` | -- |
 | `export_metrics` | -- | -- | The datanode can export its metrics and send to Prometheus compatible service (e.g. send to `greptimedb` itself) from remote-write API.<br/>This is only used for `greptimedb` to export its own metrics internally. It's different from prometheus scrape. |
 | `export_metrics.enable` | Bool | `false` | whether enable export metrics. |
 | `export_metrics.write_interval` | String | `30s` | The interval of export metrics. |
 | `export_metrics.self_import` | -- | -- | For `standalone` mode, `self_import` is recommend to collect metrics generated by itself<br/>You must create the database before enabling it. |
-| `export_metrics.self_import.db` | String | `None` | -- |
+| `export_metrics.self_import.db` | String | Unset | -- |
 | `export_metrics.remote_write` | -- | -- | -- |
 | `export_metrics.remote_write.url` | String | `""` | The url the metrics send to. The url example can be: `http://127.0.0.1:4000/v1/prometheus/write?db=greptime_metrics`. |
 | `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. |
+| `tracing.tokio_console_addr` | String | Unset | The tokio console address. |
 
 
 ### Flownode
@@ -471,7 +481,7 @@
 | 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. |
+| `node_id` | Integer | Unset | 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 |
@@ -492,12 +502,13 @@
 | `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.dir` | String | `/tmp/greptimedb/logs` | The directory to store the log files. If set to empty, logs will not be written to files. |
+| `logging.level` | String | Unset | 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.log_format` | String | `text` | The log format. Can be `text`/`json`. |
 | `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. |
+| `tracing.tokio_console_addr` | String | Unset | The tokio console address. |

config/datanode.example.toml

@@ -2,7 +2,7 @@
 mode = "standalone"
 
 ## The datanode identifier and should be unique in the cluster.
-## +toml2docs:none-default
+## @toml2docs:none-default
 node_id = 42
 
 ## Start services after regions have obtained leases.
@@ -20,23 +20,23 @@ enable_telemetry = true
 init_regions_parallelism = 16
 
 ## Deprecated, use `grpc.addr` instead.
-## +toml2docs:none-default
+## @toml2docs:none-default
 rpc_addr = "127.0.0.1:3001"
 
 ## Deprecated, use `grpc.hostname` instead.
-## +toml2docs:none-default
+## @toml2docs:none-default
 rpc_hostname = "127.0.0.1"
 
 ## Deprecated, use `grpc.runtime_size` instead.
-## +toml2docs:none-default
+## @toml2docs:none-default
 rpc_runtime_size = 8
 
 ## Deprecated, use `grpc.rpc_max_recv_message_size` instead.
-## +toml2docs:none-default
+## @toml2docs:none-default
 rpc_max_recv_message_size = "512MB"
 
 ## Deprecated, use `grpc.rpc_max_send_message_size` instead.
-## +toml2docs:none-default
+## @toml2docs:none-default
 rpc_max_send_message_size = "512MB"
 
 
@@ -71,11 +71,11 @@ max_send_message_size = "512MB"
 mode = "disable"
 
 ## Certificate file path.
-## +toml2docs:none-default
+## @toml2docs:none-default
 cert_path = ""
 
 ## Private key file path.
-## +toml2docs:none-default
+## @toml2docs:none-default
 key_path = ""
 
 ## Watch for Certificate and key file change and auto reload.
@@ -83,11 +83,11 @@ key_path = ""
 watch = false
 
 ## The runtime options.
-[runtime]
+#+ [runtime]
 ## The number of threads to execute the runtime for global read operations.
-global_rt_size = 8
+#+ global_rt_size = 8
 ## The number of threads to execute the runtime for global write operations.
-compact_rt_size = 4
+#+ compact_rt_size = 4
 
 ## The heartbeat options.
 [heartbeat]
@@ -135,7 +135,7 @@ provider = "raft_engine"
 
 ## The directory to store the WAL files.
 ## **It's only used when the provider is `raft_engine`**.
-## +toml2docs:none-default
+## @toml2docs:none-default
 dir = "/tmp/greptimedb/wal"
 
 ## The size of the WAL segment file.
@@ -170,6 +170,9 @@ prefill_log_files = false
 ## **It's only used when the provider is `raft_engine`**.
 sync_period = "10s"
 
+## Parallelism during WAL recovery.
+recovery_parallelism = 2
+
 ## The Kafka broker endpoints.
 ## **It's only used when the provider is `kafka`**.
 broker_endpoints = ["127.0.0.1:9092"]
@@ -279,90 +282,104 @@ type = "File"
 
 ## Cache configuration for object storage such as 'S3' etc.
 ## The local file cache directory.
-## +toml2docs:none-default
+## @toml2docs:none-default
 cache_path = "/path/local_cache"
 
 ## The local file cache capacity in bytes.
-## +toml2docs:none-default
+## @toml2docs:none-default
 cache_capacity = "256MB"
 
 ## The S3 bucket name.
 ## **It's only used when the storage type is `S3`, `Oss` and `Gcs`**.
-## +toml2docs:none-default
+## @toml2docs:none-default
 bucket = "greptimedb"
 
 ## The S3 data will be stored in the specified prefix, for example, `s3://${bucket}/${root}`.
 ## **It's only used when the storage type is `S3`, `Oss` and `Azblob`**.
-## +toml2docs:none-default
+## @toml2docs:none-default
 root = "greptimedb"
 
 ## The access key id of the aws account.
 ## It's **highly recommended** to use AWS IAM roles instead of hardcoding the access key id and secret key.
 ## **It's only used when the storage type is `S3` and `Oss`**.
-## +toml2docs:none-default
+## @toml2docs:none-default
 access_key_id = "test"
 
 ## The secret access key of the aws account.
 ## It's **highly recommended** to use AWS IAM roles instead of hardcoding the access key id and secret key.
 ## **It's only used when the storage type is `S3`**.
-## +toml2docs:none-default
+## @toml2docs:none-default
 secret_access_key = "test"
 
 ## The secret access key of the aliyun account.
 ## **It's only used when the storage type is `Oss`**.
-## +toml2docs:none-default
+## @toml2docs:none-default
 access_key_secret = "test"
 
 ## The account key of the azure account.
 ## **It's only used when the storage type is `Azblob`**.
-## +toml2docs:none-default
+## @toml2docs:none-default
 account_name = "test"
 
 ## The account key of the azure account.
 ## **It's only used when the storage type is `Azblob`**.
-## +toml2docs:none-default
+## @toml2docs:none-default
 account_key = "test"
 
 ## The scope of the google cloud storage.
 ## **It's only used when the storage type is `Gcs`**.
-## +toml2docs:none-default
+## @toml2docs:none-default
 scope = "test"
 
 ## The credential path of the google cloud storage.
 ## **It's only used when the storage type is `Gcs`**.
-## +toml2docs:none-default
+## @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"
+## @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
+## @toml2docs:none-default
 container = "greptimedb"
 
 ## The sas token of the azure account.
 ## **It's only used when the storage type is `Azblob`**.
-## +toml2docs:none-default
+## @toml2docs:none-default
 sas_token = ""
 
 ## The endpoint of the S3 service.
 ## **It's only used when the storage type is `S3`, `Oss`, `Gcs` and `Azblob`**.
-## +toml2docs:none-default
+## @toml2docs:none-default
 endpoint = "https://s3.amazonaws.com"
 
 ## The region of the S3 service.
 ## **It's only used when the storage type is `S3`, `Oss`, `Gcs` and `Azblob`**.
-## +toml2docs:none-default
+## @toml2docs:none-default
 region = "us-west-2"
 
 # Custom storage options
 # [[storage.providers]]
+# name = "S3"
 # type = "S3"
+# bucket = "greptimedb"
+# root = "data"
+# access_key_id = "test"
+# secret_access_key = "123456"
+# endpoint = "https://s3.amazonaws.com"
+# region = "us-west-2"
 # [[storage.providers]]
+# name = "Gcs"
 # type = "Gcs"
+# bucket = "greptimedb"
+# root = "data"
+# scope = "test"
+# credential_path = "123456"
+# credential = "base64-credential"
+# endpoint = "https://storage.googleapis.com"
 
 ## The region engine options. You can configure multiple region engines.
 [[region_engine]]
@@ -371,7 +388,7 @@ region = "us-west-2"
 [region_engine.mito]
 
 ## Number of region workers.
-num_workers = 8
+#+ num_workers = 8
 
 ## Request channel size of each worker.
 worker_channel_size = 128
@@ -392,26 +409,32 @@ max_background_jobs = 4
 auto_flush_interval = "1h"
 
 ## Global write buffer size for all regions. If not set, it's default to 1/8 of OS memory with a max limitation of 1GB.
-global_write_buffer_size = "1GB"
+## @toml2docs:none-default="Auto"
+#+ global_write_buffer_size = "1GB"
 
 ## Global write buffer size threshold to reject write requests. If not set, it's default to 2 times of `global_write_buffer_size`
-global_write_buffer_reject_size = "2GB"
+## @toml2docs:none-default="Auto"
+#+ global_write_buffer_reject_size = "2GB"
 
 ## Cache size for SST metadata. Setting it to 0 to disable the cache.
 ## If not set, it's default to 1/32 of OS memory with a max limitation of 128MB.
-sst_meta_cache_size = "128MB"
+## @toml2docs:none-default="Auto"
+#+ sst_meta_cache_size = "128MB"
 
 ## Cache size for vectors and arrow arrays. Setting it to 0 to disable the cache.
 ## If not set, it's default to 1/16 of OS memory with a max limitation of 512MB.
-vector_cache_size = "512MB"
+## @toml2docs:none-default="Auto"
+#+ vector_cache_size = "512MB"
 
 ## Cache size for pages of SST row groups. Setting it to 0 to disable the cache.
 ## If not set, it's default to 1/8 of OS memory.
-page_cache_size = "512MB"
+## @toml2docs:none-default="Auto"
+#+ page_cache_size = "512MB"
 
 ## Cache size for time series selector (e.g. `last_value()`). Setting it to 0 to disable the cache.
 ## If not set, it's default to 1/16 of OS memory with a max limitation of 512MB.
-selector_result_cache_size = "512MB"
+## @toml2docs:none-default="Auto"
+#+ selector_result_cache_size = "512MB"
 
 ## Whether to enable the experimental write cache.
 enable_experimental_write_cache = false
@@ -423,7 +446,7 @@ experimental_write_cache_path = ""
 experimental_write_cache_size = "512MB"
 
 ## TTL for write cache.
-## +toml2docs:none-default
+## @toml2docs:none-default
 experimental_write_cache_ttl = "8h"
 
 ## Buffer size for SST writing.
@@ -441,6 +464,10 @@ parallel_scan_channel_size = 32
 ## Whether to allow stale WAL entries read during replay.
 allow_stale_entries = false
 
+## Minimum time interval between two compactions.
+## To align with the old behavior, the default value is 0 (no restrictions).
+min_compaction_interval = "0m"
+
 ## The options for index in Mito engine.
 [region_engine.mito.index]
 
@@ -531,11 +558,11 @@ fork_dictionary_bytes = "1GiB"
 
 ## The logging options.
 [logging]
-## The directory to store the log files.
+## The directory to store the log files. If set to empty, logs will not be written to files.
 dir = "/tmp/greptimedb/logs"
 
 ## The log level. Can be `info`/`debug`/`warn`/`error`.
-## +toml2docs:none-default
+## @toml2docs:none-default
 level = "info"
 
 ## Enable OTLP tracing.
@@ -547,6 +574,9 @@ otlp_endpoint = "http://localhost:4317"
 ## Whether to append logs to stdout.
 append_stdout = true
 
+## The log format. Can be `text`/`json`.
+log_format = "text"
+
 ## 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
@@ -566,7 +596,7 @@ write_interval = "30s"
 ## For `standalone` mode, `self_import` is recommend to collect metrics generated by itself
 ## You must create the database before enabling it.
 [export_metrics.self_import]
-## +toml2docs:none-default
+## @toml2docs:none-default
 db = "greptime_metrics"
 
 [export_metrics.remote_write]
@@ -579,5 +609,5 @@ headers = { }
 ## The tracing options. Only effect when compiled with `tokio-console` feature.
 [tracing]
 ## The tokio console address.
-## +toml2docs:none-default
+## @toml2docs:none-default
 tokio_console_addr = "127.0.0.1"

config/flownode.example.toml

@@ -2,7 +2,7 @@
 mode = "distributed"
 
 ## The flownode identifier and should be unique in the cluster.
-## +toml2docs:none-default
+## @toml2docs:none-default
 node_id = 14
 
 ## The gRPC server options.
@@ -59,11 +59,11 @@ retry_interval = "3s"
 
 ## The logging options.
 [logging]
-## The directory to store the log files.
+## The directory to store the log files. If set to empty, logs will not be written to files.
 dir = "/tmp/greptimedb/logs"
 
 ## The log level. Can be `info`/`debug`/`warn`/`error`.
-## +toml2docs:none-default
+## @toml2docs:none-default
 level = "info"
 
 ## Enable OTLP tracing.
@@ -75,6 +75,9 @@ otlp_endpoint = "http://localhost:4317"
 ## Whether to append logs to stdout.
 append_stdout = true
 
+## The log format. Can be `text`/`json`.
+log_format = "text"
+
 ## 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
@@ -84,6 +87,6 @@ default_ratio = 1.0
 ## The tracing options. Only effect when compiled with `tokio-console` feature.
 [tracing]
 ## The tokio console address.
-## +toml2docs:none-default
+## @toml2docs:none-default
 tokio_console_addr = "127.0.0.1"
 

config/frontend.example.toml

@@ -1,13 +1,13 @@
 ## The default timezone of the server.
-## +toml2docs:none-default
+## @toml2docs:none-default
 default_timezone = "UTC"
 
 ## The runtime options.
-[runtime]
+#+ [runtime]
 ## The number of threads to execute the runtime for global read operations.
-global_rt_size = 8
+#+ global_rt_size = 8
 ## The number of threads to execute the runtime for global write operations.
-compact_rt_size = 4
+#+ compact_rt_size = 4
 
 ## The heartbeat options.
 [heartbeat]
@@ -44,11 +44,11 @@ runtime_size = 8
 mode = "disable"
 
 ## Certificate file path.
-## +toml2docs:none-default
+## @toml2docs:none-default
 cert_path = ""
 
 ## Private key file path.
-## +toml2docs:none-default
+## @toml2docs:none-default
 key_path = ""
 
 ## Watch for Certificate and key file change and auto reload.
@@ -76,11 +76,11 @@ runtime_size = 2
 mode = "disable"
 
 ## Certificate file path.
-## +toml2docs:none-default
+## @toml2docs:none-default
 cert_path = ""
 
 ## Private key file path.
-## +toml2docs:none-default
+## @toml2docs:none-default
 key_path = ""
 
 ## Watch for Certificate and key file change and auto reload
@@ -101,11 +101,11 @@ runtime_size = 2
 mode = "disable"
 
 ## Certificate file path.
-## +toml2docs:none-default
+## @toml2docs:none-default
 cert_path = ""
 
 ## Private key file path.
-## +toml2docs:none-default
+## @toml2docs:none-default
 key_path = ""
 
 ## Watch for Certificate and key file change and auto reload
@@ -166,11 +166,11 @@ tcp_nodelay = true
 
 ## The logging options.
 [logging]
-## The directory to store the log files.
+## The directory to store the log files. If set to empty, logs will not be written to files.
 dir = "/tmp/greptimedb/logs"
 
 ## The log level. Can be `info`/`debug`/`warn`/`error`.
-## +toml2docs:none-default
+## @toml2docs:none-default
 level = "info"
 
 ## Enable OTLP tracing.
@@ -182,6 +182,9 @@ otlp_endpoint = "http://localhost:4317"
 ## Whether to append logs to stdout.
 append_stdout = true
 
+## The log format. Can be `text`/`json`.
+log_format = "text"
+
 ## 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
@@ -201,7 +204,7 @@ write_interval = "30s"
 ## For `standalone` mode, `self_import` is recommend to collect metrics generated by itself
 ## You must create the database before enabling it.
 [export_metrics.self_import]
-## +toml2docs:none-default
+## @toml2docs:none-default
 db = "greptime_metrics"
 
 [export_metrics.remote_write]
@@ -214,5 +217,5 @@ headers = { }
 ## The tracing options. Only effect when compiled with `tokio-console` feature.
 [tracing]
 ## The tokio console address.
-## +toml2docs:none-default
+## @toml2docs:none-default
 tokio_console_addr = "127.0.0.1"

config/metasrv.example.toml

@@ -36,11 +36,11 @@ enable_region_failover = false
 backend = "EtcdStore"
 
 ## The runtime options.
-[runtime]
+#+ [runtime]
 ## The number of threads to execute the runtime for global read operations.
-global_rt_size = 8
+#+ global_rt_size = 8
 ## The number of threads to execute the runtime for global write operations.
-compact_rt_size = 4
+#+ compact_rt_size = 4
 
 ## Procedure storage options.
 [procedure]
@@ -153,11 +153,11 @@ backoff_deadline = "5mins"
 
 ## The logging options.
 [logging]
-## The directory to store the log files.
+## The directory to store the log files. If set to empty, logs will not be written to files.
 dir = "/tmp/greptimedb/logs"
 
 ## The log level. Can be `info`/`debug`/`warn`/`error`.
-## +toml2docs:none-default
+## @toml2docs:none-default
 level = "info"
 
 ## Enable OTLP tracing.
@@ -169,6 +169,9 @@ otlp_endpoint = "http://localhost:4317"
 ## Whether to append logs to stdout.
 append_stdout = true
 
+## The log format. Can be `text`/`json`.
+log_format = "text"
+
 ## 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
@@ -188,7 +191,7 @@ write_interval = "30s"
 ## For `standalone` mode, `self_import` is recommend to collect metrics generated by itself
 ## You must create the database before enabling it.
 [export_metrics.self_import]
-## +toml2docs:none-default
+## @toml2docs:none-default
 db = "greptime_metrics"
 
 [export_metrics.remote_write]
@@ -201,5 +204,5 @@ headers = { }
 ## The tracing options. Only effect when compiled with `tokio-console` feature.
 [tracing]
 ## The tokio console address.
-## +toml2docs:none-default
+## @toml2docs:none-default
 tokio_console_addr = "127.0.0.1"

config/standalone.example.toml

@@ -5,15 +5,22 @@ mode = "standalone"
 enable_telemetry = true
 
 ## The default timezone of the server.
-## +toml2docs:none-default
+## @toml2docs:none-default
 default_timezone = "UTC"
 
+## Initialize all regions in the background during the startup.
+## By default, it provides services after all regions have been initialized.
+init_regions_in_background = false
+
+## Parallelism of initializing regions.
+init_regions_parallelism = 16
+
 ## The runtime options.
-[runtime]
+#+ [runtime]
 ## The number of threads to execute the runtime for global read operations.
-global_rt_size = 8
+#+ global_rt_size = 8
 ## The number of threads to execute the runtime for global write operations.
-compact_rt_size = 4
+#+ compact_rt_size = 4
 
 ## The HTTP server options.
 [http]
@@ -39,11 +46,11 @@ runtime_size = 8
 mode = "disable"
 
 ## Certificate file path.
-## +toml2docs:none-default
+## @toml2docs:none-default
 cert_path = ""
 
 ## Private key file path.
-## +toml2docs:none-default
+## @toml2docs:none-default
 key_path = ""
 
 ## Watch for Certificate and key file change and auto reload.
@@ -71,11 +78,11 @@ runtime_size = 2
 mode = "disable"
 
 ## Certificate file path.
-## +toml2docs:none-default
+## @toml2docs:none-default
 cert_path = ""
 
 ## Private key file path.
-## +toml2docs:none-default
+## @toml2docs:none-default
 key_path = ""
 
 ## Watch for Certificate and key file change and auto reload
@@ -96,11 +103,11 @@ runtime_size = 2
 mode = "disable"
 
 ## Certificate file path.
-## +toml2docs:none-default
+## @toml2docs:none-default
 cert_path = ""
 
 ## Private key file path.
-## +toml2docs:none-default
+## @toml2docs:none-default
 key_path = ""
 
 ## Watch for Certificate and key file change and auto reload
@@ -132,7 +139,7 @@ provider = "raft_engine"
 
 ## The directory to store the WAL files.
 ## **It's only used when the provider is `raft_engine`**.
-## +toml2docs:none-default
+## @toml2docs:none-default
 dir = "/tmp/greptimedb/wal"
 
 ## The size of the WAL segment file.
@@ -167,6 +174,9 @@ prefill_log_files = false
 ## **It's only used when the provider is `raft_engine`**.
 sync_period = "10s"
 
+## Parallelism during WAL recovery.
+recovery_parallelism = 2
+
 ## The Kafka broker endpoints.
 ## **It's only used when the provider is `kafka`**.
 broker_endpoints = ["127.0.0.1:9092"]
@@ -310,90 +320,104 @@ type = "File"
 
 ## Cache configuration for object storage such as 'S3' etc.
 ## The local file cache directory.
-## +toml2docs:none-default
+## @toml2docs:none-default
 cache_path = "/path/local_cache"
 
 ## The local file cache capacity in bytes.
-## +toml2docs:none-default
+## @toml2docs:none-default
 cache_capacity = "256MB"
 
 ## The S3 bucket name.
 ## **It's only used when the storage type is `S3`, `Oss` and `Gcs`**.
-## +toml2docs:none-default
+## @toml2docs:none-default
 bucket = "greptimedb"
 
 ## The S3 data will be stored in the specified prefix, for example, `s3://${bucket}/${root}`.
 ## **It's only used when the storage type is `S3`, `Oss` and `Azblob`**.
-## +toml2docs:none-default
+## @toml2docs:none-default
 root = "greptimedb"
 
 ## The access key id of the aws account.
 ## It's **highly recommended** to use AWS IAM roles instead of hardcoding the access key id and secret key.
 ## **It's only used when the storage type is `S3` and `Oss`**.
-## +toml2docs:none-default
+## @toml2docs:none-default
 access_key_id = "test"
 
 ## The secret access key of the aws account.
 ## It's **highly recommended** to use AWS IAM roles instead of hardcoding the access key id and secret key.
 ## **It's only used when the storage type is `S3`**.
-## +toml2docs:none-default
+## @toml2docs:none-default
 secret_access_key = "test"
 
 ## The secret access key of the aliyun account.
 ## **It's only used when the storage type is `Oss`**.
-## +toml2docs:none-default
+## @toml2docs:none-default
 access_key_secret = "test"
 
 ## The account key of the azure account.
 ## **It's only used when the storage type is `Azblob`**.
-## +toml2docs:none-default
+## @toml2docs:none-default
 account_name = "test"
 
 ## The account key of the azure account.
 ## **It's only used when the storage type is `Azblob`**.
-## +toml2docs:none-default
+## @toml2docs:none-default
 account_key = "test"
 
 ## The scope of the google cloud storage.
 ## **It's only used when the storage type is `Gcs`**.
-## +toml2docs:none-default
+## @toml2docs:none-default
 scope = "test"
 
 ## The credential path of the google cloud storage.
 ## **It's only used when the storage type is `Gcs`**.
-## +toml2docs:none-default
+## @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
+## @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
+## @toml2docs:none-default
 container = "greptimedb"
 
 ## The sas token of the azure account.
 ## **It's only used when the storage type is `Azblob`**.
-## +toml2docs:none-default
+## @toml2docs:none-default
 sas_token = ""
 
 ## The endpoint of the S3 service.
 ## **It's only used when the storage type is `S3`, `Oss`, `Gcs` and `Azblob`**.
-## +toml2docs:none-default
+## @toml2docs:none-default
 endpoint = "https://s3.amazonaws.com"
 
 ## The region of the S3 service.
 ## **It's only used when the storage type is `S3`, `Oss`, `Gcs` and `Azblob`**.
-## +toml2docs:none-default
+## @toml2docs:none-default
 region = "us-west-2"
 
 # Custom storage options
 # [[storage.providers]]
+# name = "S3"
 # type = "S3"
+# bucket = "greptimedb"
+# root = "data"
+# access_key_id = "test"
+# secret_access_key = "123456"
+# endpoint = "https://s3.amazonaws.com"
+# region = "us-west-2"
 # [[storage.providers]]
+# name = "Gcs"
 # type = "Gcs"
+# bucket = "greptimedb"
+# root = "data"
+# scope = "test"
+# credential_path = "123456"
+# credential = "base64-credential"
+# endpoint = "https://storage.googleapis.com"
 
 ## The region engine options. You can configure multiple region engines.
 [[region_engine]]
@@ -402,7 +426,7 @@ region = "us-west-2"
 [region_engine.mito]
 
 ## Number of region workers.
-num_workers = 8
+#+ num_workers = 8
 
 ## Request channel size of each worker.
 worker_channel_size = 128
@@ -423,26 +447,32 @@ max_background_jobs = 4
 auto_flush_interval = "1h"
 
 ## Global write buffer size for all regions. If not set, it's default to 1/8 of OS memory with a max limitation of 1GB.
-global_write_buffer_size = "1GB"
+## @toml2docs:none-default="Auto"
+#+ global_write_buffer_size = "1GB"
 
-## Global write buffer size threshold to reject write requests. If not set, it's default to 2 times of `global_write_buffer_size`
-global_write_buffer_reject_size = "2GB"
+## Global write buffer size threshold to reject write requests. If not set, it's default to 2 times of `global_write_buffer_size`.
+## @toml2docs:none-default="Auto"
+#+ global_write_buffer_reject_size = "2GB"
 
 ## Cache size for SST metadata. Setting it to 0 to disable the cache.
 ## If not set, it's default to 1/32 of OS memory with a max limitation of 128MB.
-sst_meta_cache_size = "128MB"
+## @toml2docs:none-default="Auto"
+#+ sst_meta_cache_size = "128MB"
 
 ## Cache size for vectors and arrow arrays. Setting it to 0 to disable the cache.
 ## If not set, it's default to 1/16 of OS memory with a max limitation of 512MB.
-vector_cache_size = "512MB"
+## @toml2docs:none-default="Auto"
+#+ vector_cache_size = "512MB"
 
 ## Cache size for pages of SST row groups. Setting it to 0 to disable the cache.
 ## If not set, it's default to 1/8 of OS memory.
-page_cache_size = "512MB"
+## @toml2docs:none-default="Auto"
+#+ page_cache_size = "512MB"
 
 ## Cache size for time series selector (e.g. `last_value()`). Setting it to 0 to disable the cache.
 ## If not set, it's default to 1/16 of OS memory with a max limitation of 512MB.
-selector_result_cache_size = "512MB"
+## @toml2docs:none-default="Auto"
+#+ selector_result_cache_size = "512MB"
 
 ## Whether to enable the experimental write cache.
 enable_experimental_write_cache = false
@@ -454,7 +484,7 @@ experimental_write_cache_path = ""
 experimental_write_cache_size = "512MB"
 
 ## TTL for write cache.
-## +toml2docs:none-default
+## @toml2docs:none-default
 experimental_write_cache_ttl = "8h"
 
 ## Buffer size for SST writing.
@@ -472,6 +502,10 @@ parallel_scan_channel_size = 32
 ## Whether to allow stale WAL entries read during replay.
 allow_stale_entries = false
 
+## Minimum time interval between two compactions.
+## To align with the old behavior, the default value is 0 (no restrictions).
+min_compaction_interval = "0m"
+
 ## The options for index in Mito engine.
 [region_engine.mito.index]
 
@@ -568,11 +602,11 @@ fork_dictionary_bytes = "1GiB"
 
 ## The logging options.
 [logging]
-## The directory to store the log files.
+## The directory to store the log files. If set to empty, logs will not be written to files.
 dir = "/tmp/greptimedb/logs"
 
 ## The log level. Can be `info`/`debug`/`warn`/`error`.
-## +toml2docs:none-default
+## @toml2docs:none-default
 level = "info"
 
 ## Enable OTLP tracing.
@@ -584,6 +618,9 @@ otlp_endpoint = "http://localhost:4317"
 ## Whether to append logs to stdout.
 append_stdout = true
 
+## The log format. Can be `text`/`json`.
+log_format = "text"
+
 ## 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
@@ -600,10 +637,10 @@ enable = false
 ## The interval of export metrics.
 write_interval = "30s"
 
-## For `standalone` mode, `self_import` is recommend to collect metrics generated by itself
+## For `standalone` mode, `self_import` is recommended to collect metrics generated by itself
 ## You must create the database before enabling it.
 [export_metrics.self_import]
-## +toml2docs:none-default
+## @toml2docs:none-default
 db = "greptime_metrics"
 
 [export_metrics.remote_write]
@@ -616,5 +653,5 @@ headers = { }
 ## The tracing options. Only effect when compiled with `tokio-console` feature.
 [tracing]
 ## The tokio console address.
-## +toml2docs:none-default
+## @toml2docs:none-default
 tokio_console_addr = "127.0.0.1"

docs/rfcs/2024-08-06-json-datatype.md

@@ -0,0 +1,197 @@
+---
+Feature Name: Json Datatype
+Tracking Issue: https://github.com/GreptimeTeam/greptimedb/issues/4230
+Date: 2024-8-6
+Author: "Yuhan Wang <profsyb@gmail.com>"
+---
+
+# Summary
+This RFC proposes a method for storing and querying JSON data in the database.
+
+# Motivation
+JSON is widely used across various scenarios. Direct support for writing and querying JSON can significantly enhance the database's flexibility.
+
+# Details
+
+## Storage and Query
+
+GreptimeDB's type system is built on Arrow/DataFusion, where each data type in GreptimeDB corresponds to a data type in Arrow/DataFusion. The proposed JSON type will be implemented on top of the existing `Binary` type, leveraging the current `datatype::value::Value` and `datatype::vectors::BinaryVector` implementations, utilizing the JSONB format as the encoding of JSON data. JSON data is stored and processed similarly to binary data within the storage layer and query engine.
+
+This approach brings problems when dealing with insertions and queries of JSON columns.
+
+## Insertion
+
+Users commonly write JSON data as strings. Thus we need to make conversions between string and JSONB. There are 2 ways to do this:
+
+1. MySQL and PostgreSQL servers provide auto-conversions between strings and JSONB. When a string is inserted into a JSON column, the server will try to parse the string as JSON and convert it to JSONB. The non-JSON strings will be rejected.
+
+2. A function `parse_json` is provided to convert string to JSONB. If the string is not a valid JSON string, the function will return an error.
+
+For example, in MySQL client:
+```SQL
+CREATE TABLE IF NOT EXISTS test (
+    ts TIMESTAMP TIME INDEX,
+    a INT,
+    b JSON
+);
+
+INSERT INTO test VALUES(
+    0,
+    0,
+    '{
+        "name": "jHl2oDDnPc1i2OzlP5Y",
+        "timestamp": "2024-07-25T04:33:11.369386Z",
+        "attributes": { "event_attributes": 48.28667 }
+    }'
+);
+
+INSERT INTO test VALUES(
+    0,
+    0,
+    parse_json('{
+        "name": "jHl2oDDnPc1i2OzlP5Y",
+        "timestamp": "2024-07-25T04:33:11.369386Z",
+        "attributes": { "event_attributes": 48.28667 }
+    }')
+);
+```
+Are both valid.
+
+The dataflow of the insertion process is as follows:
+```
+Insert JSON strings directly through client:
+                                   Parse                       Insert
+        String(Serialized JSON)┌──────────┐Arrow Binary(JSONB)┌──────┐Arrow Binary(JSONB)
+ Client ---------------------->│  Server  │------------------>│ Mito │------------------> Storage
+                               └──────────┘                   └──────┘
+        (Server identifies JSON type and performs auto-conversion)
+
+Insert JSON strings through parse_json function:
+                                                                   Parse                     Insert
+        String(Serialized JSON)┌──────────┐String(Serialized JSON)┌─────┐Arrow Binary(JSONB)┌──────┐Arrow Binary(JSONB)
+ Client ---------------------->│  Server  │---------------------->│ UDF │------------------>│ Mito │------------------> Storage
+                               └──────────┘                       └─────┘                   └──────┘
+                                            (Conversion is performed by UDF inside Query Engine)
+```
+
+Servers identify JSON column through column schema and perform auto-conversions. But when using prepared statements and binding parameters, the corresponding cached plans in datafusion generated by prepared statements cannot identify JSON columns. Under this circumstance, the servers identify JSON columns through the given parameters and perform auto-conversions.
+
+The following is an example of inserting JSON data through prepared statements:
+```Rust
+sqlx::query(
+    "create table test(ts timestamp time index, j json)",
+)
+.execute(&pool)
+.await
+.unwrap();
+
+let json = serde_json::json!({
+    "code": 200,
+    "success": true,
+    "payload": {
+        "features": [
+            "serde",
+            "json"
+        ],
+        "homepage": null
+    }
+});
+
+// Valid, can identify serde_json::Value as JSON type
+sqlx::query("insert into test values($1, $2)")
+    .bind(i)
+    .bind(json)
+    .execute(&pool)
+    .await
+    .unwrap();
+
+// Invalid, cannot identify String as JSON type
+sqlx::query("insert into test values($1, $2)")
+    .bind(i)
+    .bind(json.to_string())
+    .execute(&pool)
+    .await
+    .unwrap();
+```
+
+## Query
+
+Correspondingly, users prefer to display JSON data as strings. Thus we need to make conversions between JSON data and strings before presenting JSON data. There are also 2 ways to do this: auto-conversions on MySQL and PostgreSQL servers, and function `json_to_string`.
+
+For example, in MySQL client:
+```SQL
+SELECT b FROM test;
+
+SELECT json_to_string(b) FROM test;
+```
+Will both return the JSON as human-readable strings.
+
+Specifically, to perform auto-conversions, we attach a message to JSON data in the `metadata` of `Field` in Arrow/Datafusion schema when scanning a JSON column. Frontend servers could identify JSON data and convert it to strings.
+
+The dataflow of the query process is as follows:
+```
+Query directly through client:
+                                  Decode                            Scan
+        String(Serialized JSON)┌──────────┐Arrow Binary(JSONB)┌──────────────┐Arrow Binary(JSONB)
+ Client <----------------------│  Server  │<------------------│ Query Engine │<----------------- Storage
+                               └──────────┘                   └──────────────┘
+(Server identifies JSON type and performs auto-conversion based on column metadata)
+
+Query through json_to_string function:
+                                                                   Scan & Decode
+        String(Serialized JSON)┌──────────┐String(Serialized JSON)┌──────────────┐Arrow Binary(JSONB)
+ Client <----------------------│  Server  │<----------------------│ Query Engine │<----------------- Storage
+                               └──────────┘                       └──────────────┘
+                                                 (Conversion is performed by UDF inside Query Engine)
+
+```
+
+However, if a function uses JSON type as its return type, the metadata method mentioned above is not applicable. Thus the functions of JSON type should specify the return type explicitly instead of returning a JSON type, such as `json_get_int` and `json_get_float` which return corresponding data of `INT` and `FLOAT` type respectively.
+
+## Functions
+Similar to the common JSON type, JSON data can be queried with functions.
+
+For example:
+```SQL
+CREATE TABLE IF NOT EXISTS test (
+    ts TIMESTAMP TIME INDEX,
+    a INT,
+    b JSON
+);
+
+INSERT INTO test VALUES(
+    0,
+    0,
+    '{
+        "name": "jHl2oDDnPc1i2OzlP5Y",
+        "timestamp": "2024-07-25T04:33:11.369386Z",
+        "attributes": { "event_attributes": 48.28667 }
+    }'
+);
+
+SELECT json_get_string(b, 'name') FROM test;
++---------------------+
+| b.name              |
++---------------------+
+| jHl2oDDnPc1i2OzlP5Y |
++---------------------+
+
+SELECT json_get_float(b, 'attributes.event_attributes') FROM test;
++--------------------------------+
+| b.attributes.event_attributes  |
++--------------------------------+
+| 48.28667                       |
++--------------------------------+
+
+```
+And more functions can be added in the future.
+
+# Drawbacks
+
+As a general purpose JSON data type, JSONB may not be as efficient as specialized data types for specific scenarios.
+
+The auto-conversion mechanism is not supported in all scenarios. We need to find workarounds for these scenarios.
+
+# Alternatives
+
+Extract and flatten JSON schema to store in a structured format through pipeline. For nested data, we can provide nested types like `STRUCT` or `ARRAY`.

scripts/check-snafu.py

@@ -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.
+
+import os
+import re
+
+
+def find_rust_files(directory):
+    error_files = []
+    other_rust_files = []
+    for root, _, files in os.walk(directory):
+        for file in files:
+            if file == "error.rs":
+                error_files.append(os.path.join(root, file))
+            elif file.endswith(".rs"):
+                other_rust_files.append(os.path.join(root, file))
+    return error_files, other_rust_files
+
+
+def extract_branch_names(file_content):
+    pattern = re.compile(r"#\[snafu\(display\([^\)]*\)\)\]\s*(\w+)\s*\{")
+    return pattern.findall(file_content)
+
+
+def check_snafu_in_files(branch_name, rust_files):
+    branch_name_snafu = f"{branch_name}Snafu"
+    for rust_file in rust_files:
+        with open(rust_file, "r") as file:
+            content = file.read()
+            if branch_name_snafu in content:
+                return True
+    return False
+
+
+def main():
+    error_files, other_rust_files = find_rust_files(".")
+    branch_names = []
+
+    for error_file in error_files:
+        with open(error_file, "r") as file:
+            content = file.read()
+            branch_names.extend(extract_branch_names(content))
+
+    unused_snafu = [
+        branch_name
+        for branch_name in branch_names
+        if not check_snafu_in_files(branch_name, other_rust_files)
+    ]
+
+    for name in unused_snafu:
+        print(name)
+
+    if unused_snafu:
+        raise SystemExit(1)
+
+
+if __name__ == "__main__":
+    main()

src/api/src/helper.rs

@@ -42,7 +42,8 @@ use greptime_proto::v1::greptime_request::Request;
 use greptime_proto::v1::query_request::Query;
 use greptime_proto::v1::value::ValueData;
 use greptime_proto::v1::{
-    ColumnDataTypeExtension, DdlRequest, DecimalTypeExtension, QueryRequest, Row, SemanticType,
+    ColumnDataTypeExtension, DdlRequest, DecimalTypeExtension, JsonTypeExtension, QueryRequest,
+    Row, SemanticType,
 };
 use paste::paste;
 use snafu::prelude::*;
@@ -103,7 +104,17 @@ impl From<ColumnDataTypeWrapper> for ConcreteDataType {
             ColumnDataType::Uint64 => ConcreteDataType::uint64_datatype(),
             ColumnDataType::Float32 => ConcreteDataType::float32_datatype(),
             ColumnDataType::Float64 => ConcreteDataType::float64_datatype(),
-            ColumnDataType::Binary => ConcreteDataType::binary_datatype(),
+            ColumnDataType::Binary => {
+                if let Some(TypeExt::JsonType(_)) = datatype_wrapper
+                    .datatype_ext
+                    .as_ref()
+                    .and_then(|datatype_ext| datatype_ext.type_ext.as_ref())
+                {
+                    ConcreteDataType::json_datatype()
+                } else {
+                    ConcreteDataType::binary_datatype()
+                }
+            }
             ColumnDataType::String => ConcreteDataType::string_datatype(),
             ColumnDataType::Date => ConcreteDataType::date_datatype(),
             ColumnDataType::Datetime => ConcreteDataType::datetime_datatype(),
@@ -236,7 +247,7 @@ impl TryFrom<ConcreteDataType> for ColumnDataTypeWrapper {
             ConcreteDataType::UInt64(_) => ColumnDataType::Uint64,
             ConcreteDataType::Float32(_) => ColumnDataType::Float32,
             ConcreteDataType::Float64(_) => ColumnDataType::Float64,
-            ConcreteDataType::Binary(_) => ColumnDataType::Binary,
+            ConcreteDataType::Binary(_) | ConcreteDataType::Json(_) => ColumnDataType::Binary,
             ConcreteDataType::String(_) => ColumnDataType::String,
             ConcreteDataType::Date(_) => ColumnDataType::Date,
             ConcreteDataType::DateTime(_) => ColumnDataType::Datetime,
@@ -276,6 +287,16 @@ impl TryFrom<ConcreteDataType> for ColumnDataTypeWrapper {
                         })),
                     })
             }
+            ColumnDataType::Binary => {
+                if datatype == ConcreteDataType::json_datatype() {
+                    // Json is the same as  binary in proto. The extension marks the binary in proto is actually a json.
+                    Some(ColumnDataTypeExtension {
+                        type_ext: Some(TypeExt::JsonType(JsonTypeExtension::JsonBinary.into())),
+                    })
+                } else {
+                    None
+                }
+            }
             _ => None,
         };
         Ok(Self {
@@ -649,7 +670,8 @@ pub fn pb_values_to_vector_ref(data_type: &ConcreteDataType, values: Values) ->
         ConcreteDataType::Null(_)
         | ConcreteDataType::List(_)
         | ConcreteDataType::Dictionary(_)
-        | ConcreteDataType::Duration(_) => {
+        | ConcreteDataType::Duration(_)
+        | ConcreteDataType::Json(_) => {
             unreachable!()
         }
     }
@@ -813,7 +835,8 @@ pub fn pb_values_to_values(data_type: &ConcreteDataType, values: Values) -> Vec<
         ConcreteDataType::Null(_)
         | ConcreteDataType::List(_)
         | ConcreteDataType::Dictionary(_)
-        | ConcreteDataType::Duration(_) => {
+        | ConcreteDataType::Duration(_)
+        | ConcreteDataType::Json(_) => {
             unreachable!()
         }
     }
@@ -831,7 +854,13 @@ pub fn is_column_type_value_eq(
     expect_type: &ConcreteDataType,
 ) -> bool {
     ColumnDataTypeWrapper::try_new(type_value, type_extension)
-        .map(|wrapper| ConcreteDataType::from(wrapper) == *expect_type)
+        .map(|wrapper| {
+            let datatype = ConcreteDataType::from(wrapper);
+            (datatype == *expect_type)
+            // Json type leverage binary type in pb, so this is valid.
+                || (datatype == ConcreteDataType::binary_datatype()
+                    && *expect_type == ConcreteDataType::json_datatype())
+        })
         .unwrap_or(false)
 }
 

src/api/src/region.rs

@@ -21,14 +21,14 @@ use greptime_proto::v1::region::RegionResponse as RegionResponseV1;
 #[derive(Debug)]
 pub struct RegionResponse {
     pub affected_rows: AffectedRows,
-    pub extension: HashMap<String, Vec<u8>>,
+    pub extensions: HashMap<String, Vec<u8>>,
 }
 
 impl RegionResponse {
     pub fn from_region_response(region_response: RegionResponseV1) -> Self {
         Self {
             affected_rows: region_response.affected_rows as _,
-            extension: region_response.extension,
+            extensions: region_response.extensions,
         }
     }
 
@@ -36,7 +36,7 @@ impl RegionResponse {
     pub fn new(affected_rows: AffectedRows) -> Self {
         Self {
             affected_rows,
-            extension: Default::default(),
+            extensions: Default::default(),
         }
     }
 }

src/auth/src/tests.rs

@@ -13,9 +13,11 @@
 // limitations under the License.
 
 use common_base::secrets::ExposeSecret;
+use common_error::ext::BoxedError;
+use snafu::{OptionExt, ResultExt};
 
 use crate::error::{
-    AccessDeniedSnafu, Result, UnsupportedPasswordTypeSnafu, UserNotFoundSnafu,
+    AccessDeniedSnafu, AuthBackendSnafu, Result, UnsupportedPasswordTypeSnafu, UserNotFoundSnafu,
     UserPasswordMismatchSnafu,
 };
 use crate::user_info::DefaultUserInfo;
@@ -49,6 +51,19 @@ impl MockUserProvider {
         info.schema.clone_into(&mut self.schema);
         info.username.clone_into(&mut self.username);
     }
+
+    // this is a deliberate function to ref AuthBackendSnafu
+    // so that it won't get deleted in the future
+    pub fn ref_auth_backend_snafu(&self) -> Result<()> {
+        let none_option = None;
+
+        none_option
+            .context(UserNotFoundSnafu {
+                username: "no_user".to_string(),
+            })
+            .map_err(BoxedError::new)
+            .context(AuthBackendSnafu)
+    }
 }
 
 #[async_trait::async_trait]

src/auth/tests/mod.rs

@@ -18,6 +18,7 @@ use std::sync::Arc;
 
 use api::v1::greptime_request::Request;
 use auth::error::Error::InternalState;
+use auth::error::InternalStateSnafu;
 use auth::{PermissionChecker, PermissionCheckerRef, PermissionReq, PermissionResp, UserInfoRef};
 use sql::statements::show::{ShowDatabases, ShowKind};
 use sql::statements::statement::Statement;
@@ -33,9 +34,10 @@ impl PermissionChecker for DummyPermissionChecker {
         match req {
             PermissionReq::GrpcRequest(_) => Ok(PermissionResp::Allow),
             PermissionReq::SqlStatement(_) => Ok(PermissionResp::Reject),
-            _ => Err(InternalState {
+            _ => InternalStateSnafu {
                 msg: "testing".to_string(),
-            }),
+            }
+            .fail(),
         }
     }
 }

src/catalog/Cargo.toml

@@ -24,6 +24,7 @@ common-macro.workspace = true
 common-meta.workspace = true
 common-query.workspace = true
 common-recordbatch.workspace = true
+common-runtime.workspace = true
 common-telemetry.workspace = true
 common-time.workspace = true
 common-version.workspace = true
@@ -48,6 +49,7 @@ sql.workspace = true
 store-api.workspace = true
 table.workspace = true
 tokio.workspace = true
+tokio-stream = "0.1"
 
 [dev-dependencies]
 cache.workspace = true

src/catalog/src/error.rs

@@ -97,13 +97,6 @@ pub enum Error {
         source: table::error::Error,
     },
 
-    #[snafu(display("System catalog is not valid: {}", msg))]
-    SystemCatalog {
-        msg: String,
-        #[snafu(implicit)]
-        location: Location,
-    },
-
     #[snafu(display("Cannot find catalog by name: {}", catalog_name))]
     CatalogNotFound {
         catalog_name: String,
@@ -186,13 +179,6 @@ pub enum Error {
         source: common_query::error::Error,
     },
 
-    #[snafu(display("Failed to perform metasrv operation"))]
-    Metasrv {
-        #[snafu(implicit)]
-        location: Location,
-        source: meta_client::error::Error,
-    },
-
     #[snafu(display("Invalid table info in catalog"))]
     InvalidTableInfoInCatalog {
         #[snafu(implicit)]
@@ -288,8 +274,6 @@ impl ErrorExt for Error {
 
             Error::FlowInfoNotFound { .. } => StatusCode::FlowNotFound,
 
-            Error::SystemCatalog { .. } => StatusCode::StorageUnavailable,
-
             Error::UpgradeWeakCatalogManagerRef { .. } => StatusCode::Internal,
 
             Error::CreateRecordBatch { source, .. } => source.status_code(),
@@ -303,7 +287,6 @@ impl ErrorExt for Error {
 
             Error::CreateTable { source, .. } => source.status_code(),
 
-            Error::Metasrv { source, .. } => source.status_code(),
             Error::DecodePlan { source, .. } => source.status_code(),
             Error::InvalidTableInfoInCatalog { source, .. } => source.status_code(),
 
@@ -338,27 +321,6 @@ mod tests {
 
     use super::*;
 
-    #[test]
-    pub fn test_error_status_code() {
-        assert_eq!(
-            StatusCode::TableAlreadyExists,
-            Error::TableExists {
-                table: "some_table".to_string(),
-                location: Location::generate(),
-            }
-            .status_code()
-        );
-
-        assert_eq!(
-            StatusCode::StorageUnavailable,
-            Error::SystemCatalog {
-                msg: String::default(),
-                location: Location::generate(),
-            }
-            .status_code()
-        );
-    }
-
     #[test]
     pub fn test_errors_to_datafusion_error() {
         let e: DataFusionError = Error::TableExists {

src/catalog/src/kvbackend/client.rs

@@ -20,8 +20,8 @@ use std::time::Duration;
 
 use common_error::ext::BoxedError;
 use common_meta::cache_invalidator::KvCacheInvalidator;
-use common_meta::error::Error::{CacheNotGet, GetKvCache};
-use common_meta::error::{CacheNotGetSnafu, Error, ExternalSnafu, Result};
+use common_meta::error::Error::CacheNotGet;
+use common_meta::error::{CacheNotGetSnafu, Error, ExternalSnafu, GetKvCacheSnafu, Result};
 use common_meta::kv_backend::{KvBackend, KvBackendRef, TxnService};
 use common_meta::rpc::store::{
     BatchDeleteRequest, BatchDeleteResponse, BatchGetRequest, BatchGetResponse, BatchPutRequest,
@@ -282,8 +282,11 @@ impl KvBackend for CachedMetaKvBackend {
                 _ => Err(e),
             },
         }
-        .map_err(|e| GetKvCache {
-            err_msg: e.to_string(),
+        .map_err(|e| {
+            GetKvCacheSnafu {
+                err_msg: e.to_string(),
+            }
+            .build()
         });
 
         // "cache.invalidate_key" and "cache.try_get_with_by_ref" are not mutually exclusive. So we need

src/catalog/src/kvbackend/manager.rs

@@ -36,11 +36,14 @@ use futures_util::{StreamExt, TryStreamExt};
 use meta_client::client::MetaClient;
 use moka::sync::Cache;
 use partition::manager::{PartitionRuleManager, PartitionRuleManagerRef};
+use session::context::{Channel, QueryContext};
 use snafu::prelude::*;
 use table::dist_table::DistTable;
 use table::table::numbers::{NumbersTable, NUMBERS_TABLE_NAME};
 use table::table_name::TableName;
 use table::TableRef;
+use tokio::sync::Semaphore;
+use tokio_stream::wrappers::ReceiverStream;
 
 use crate::error::{
     CacheNotFoundSnafu, GetTableCacheSnafu, InvalidTableInfoInCatalogSnafu, ListCatalogsSnafu,
@@ -152,7 +155,11 @@ impl CatalogManager for KvBackendCatalogManager {
         Ok(keys)
     }
 
-    async fn schema_names(&self, catalog: &str) -> Result<Vec<String>> {
+    async fn schema_names(
+        &self,
+        catalog: &str,
+        query_ctx: Option<&QueryContext>,
+    ) -> Result<Vec<String>> {
         let stream = self
             .table_metadata_manager
             .schema_manager()
@@ -163,27 +170,29 @@ impl CatalogManager for KvBackendCatalogManager {
             .map_err(BoxedError::new)
             .context(ListSchemasSnafu { catalog })?;
 
-        keys.extend(self.system_catalog.schema_names());
+        keys.extend(self.system_catalog.schema_names(query_ctx));
 
         Ok(keys.into_iter().collect())
     }
 
-    async fn table_names(&self, catalog: &str, schema: &str) -> Result<Vec<String>> {
-        let stream = self
+    async fn table_names(
+        &self,
+        catalog: &str,
+        schema: &str,
+        query_ctx: Option<&QueryContext>,
+    ) -> Result<Vec<String>> {
+        let mut tables = self
             .table_metadata_manager
             .table_name_manager()
-            .tables(catalog, schema);
-        let mut tables = stream
+            .tables(catalog, schema)
+            .map_ok(|(table_name, _)| table_name)
             .try_collect::<Vec<_>>()
             .await
             .map_err(BoxedError::new)
-            .context(ListTablesSnafu { catalog, schema })?
-            .into_iter()
-            .map(|(k, _)| k)
-            .collect::<Vec<_>>();
-        tables.extend_from_slice(&self.system_catalog.table_names(schema));
+            .context(ListTablesSnafu { catalog, schema })?;
 
-        Ok(tables.into_iter().collect())
+        tables.extend(self.system_catalog.table_names(schema, query_ctx));
+        Ok(tables)
     }
 
     async fn catalog_exists(&self, catalog: &str) -> Result<bool> {
@@ -194,8 +203,13 @@ impl CatalogManager for KvBackendCatalogManager {
             .context(TableMetadataManagerSnafu)
     }
 
-    async fn schema_exists(&self, catalog: &str, schema: &str) -> Result<bool> {
-        if self.system_catalog.schema_exists(schema) {
+    async fn schema_exists(
+        &self,
+        catalog: &str,
+        schema: &str,
+        query_ctx: Option<&QueryContext>,
+    ) -> Result<bool> {
+        if self.system_catalog.schema_exists(schema, query_ctx) {
             return Ok(true);
         }
 
@@ -206,8 +220,14 @@ impl CatalogManager for KvBackendCatalogManager {
             .context(TableMetadataManagerSnafu)
     }
 
-    async fn table_exists(&self, catalog: &str, schema: &str, table: &str) -> Result<bool> {
-        if self.system_catalog.table_exists(schema, table) {
+    async fn table_exists(
+        &self,
+        catalog: &str,
+        schema: &str,
+        table: &str,
+        query_ctx: Option<&QueryContext>,
+    ) -> Result<bool> {
+        if self.system_catalog.table_exists(schema, table, query_ctx) {
             return Ok(true);
         }
 
@@ -225,10 +245,12 @@ impl CatalogManager for KvBackendCatalogManager {
         catalog_name: &str,
         schema_name: &str,
         table_name: &str,
+        query_ctx: Option<&QueryContext>,
     ) -> Result<Option<TableRef>> {
-        if let Some(table) = self
-            .system_catalog
-            .table(catalog_name, schema_name, table_name)
+        let channel = query_ctx.map_or(Channel::Unknown, |ctx| ctx.channel());
+        if let Some(table) =
+            self.system_catalog
+                .table(catalog_name, schema_name, table_name, query_ctx)
         {
             return Ok(Some(table));
         }
@@ -236,58 +258,112 @@ impl CatalogManager for KvBackendCatalogManager {
         let table_cache: TableCacheRef = self.cache_registry.get().context(CacheNotFoundSnafu {
             name: "table_cache",
         })?;
-
-        table_cache
+        if let Some(table) = table_cache
             .get_by_ref(&TableName {
                 catalog_name: catalog_name.to_string(),
                 schema_name: schema_name.to_string(),
                 table_name: table_name.to_string(),
             })
             .await
-            .context(GetTableCacheSnafu)
+            .context(GetTableCacheSnafu)?
+        {
+            return Ok(Some(table));
+        }
+
+        if channel == Channel::Postgres {
+            // falldown to pg_catalog
+            if let Some(table) =
+                self.system_catalog
+                    .table(catalog_name, PG_CATALOG_NAME, table_name, query_ctx)
+            {
+                return Ok(Some(table));
+            }
+        }
+
+        return Ok(None);
     }
 
-    fn tables<'a>(&'a self, catalog: &'a str, schema: &'a str) -> BoxStream<'a, Result<TableRef>> {
+    fn tables<'a>(
+        &'a self,
+        catalog: &'a str,
+        schema: &'a str,
+        query_ctx: Option<&'a QueryContext>,
+    ) -> BoxStream<'a, Result<TableRef>> {
         let sys_tables = try_stream!({
             // System tables
-            let sys_table_names = self.system_catalog.table_names(schema);
+            let sys_table_names = self.system_catalog.table_names(schema, query_ctx);
             for table_name in sys_table_names {
-                if let Some(table) = self.system_catalog.table(catalog, schema, &table_name) {
+                if let Some(table) =
+                    self.system_catalog
+                        .table(catalog, schema, &table_name, query_ctx)
+                {
                     yield table;
                 }
             }
         });
 
-        let table_id_stream = self
-            .table_metadata_manager
-            .table_name_manager()
-            .tables(catalog, schema)
-            .map_ok(|(_, v)| v.table_id());
         const BATCH_SIZE: usize = 128;
-        let user_tables = try_stream!({
+        const CONCURRENCY: usize = 8;
+
+        let (tx, rx) = tokio::sync::mpsc::channel(64);
+        let metadata_manager = self.table_metadata_manager.clone();
+        let catalog = catalog.to_string();
+        let schema = schema.to_string();
+        let semaphore = Arc::new(Semaphore::new(CONCURRENCY));
+
+        common_runtime::spawn_global(async move {
+            let table_id_stream = metadata_manager
+                .table_name_manager()
+                .tables(&catalog, &schema)
+                .map_ok(|(_, v)| v.table_id());
             // Split table ids into chunks
             let mut table_id_chunks = table_id_stream.ready_chunks(BATCH_SIZE);
 
             while let Some(table_ids) = table_id_chunks.next().await {
-                let table_ids = table_ids
+                let table_ids = match table_ids
                     .into_iter()
                     .collect::<std::result::Result<Vec<_>, _>>()
                     .map_err(BoxedError::new)
-                    .context(ListTablesSnafu { catalog, schema })?;
+                    .context(ListTablesSnafu {
+                        catalog: &catalog,
+                        schema: &schema,
+                    }) {
+                    Ok(table_ids) => table_ids,
+                    Err(e) => {
+                        let _ = tx.send(Err(e)).await;
+                        return;
+                    }
+                };
 
-                let table_info_values = self
-                    .table_metadata_manager
-                    .table_info_manager()
-                    .batch_get(&table_ids)
-                    .await
-                    .context(TableMetadataManagerSnafu)?;
+                let metadata_manager = metadata_manager.clone();
+                let tx = tx.clone();
+                let semaphore = semaphore.clone();
+                common_runtime::spawn_global(async move {
+                    // we don't explicitly close the semaphore so just ignore the potential error.
+                    let _ = semaphore.acquire().await;
+                    let table_info_values = match metadata_manager
+                        .table_info_manager()
+                        .batch_get(&table_ids)
+                        .await
+                        .context(TableMetadataManagerSnafu)
+                    {
+                        Ok(table_info_values) => table_info_values,
+                        Err(e) => {
+                            let _ = tx.send(Err(e)).await;
+                            return;
+                        }
+                    };
 
-                for table_info_value in table_info_values.into_values() {
-                    yield build_table(table_info_value)?;
-                }
+                    for table in table_info_values.into_values().map(build_table) {
+                        if tx.send(table).await.is_err() {
+                            return;
+                        }
+                    }
+                });
             }
         });
 
+        let user_tables = ReceiverStream::new(rx);
         Box::pin(sys_tables.chain(user_tables))
     }
 }
@@ -313,25 +389,34 @@ struct SystemCatalog {
     catalog_cache: Cache<String, Arc<InformationSchemaProvider>>,
     pg_catalog_cache: Cache<String, Arc<PGCatalogProvider>>,
 
-    // system_schema_provier for default catalog
+    // system_schema_provider 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(),
-            PG_CATALOG_NAME.to_string(),
-        ]
+    fn schema_names(&self, query_ctx: Option<&QueryContext>) -> Vec<String> {
+        let channel = query_ctx.map_or(Channel::Unknown, |ctx| ctx.channel());
+        match channel {
+            // pg_catalog only visible under postgres protocol
+            Channel::Postgres => vec![
+                INFORMATION_SCHEMA_NAME.to_string(),
+                PG_CATALOG_NAME.to_string(),
+            ],
+            _ => {
+                vec![INFORMATION_SCHEMA_NAME.to_string()]
+            }
+        }
     }
 
-    fn table_names(&self, schema: &str) -> Vec<String> {
+    fn table_names(&self, schema: &str, query_ctx: Option<&QueryContext>) -> Vec<String> {
+        let channel = query_ctx.map_or(Channel::Unknown, |ctx| ctx.channel());
         match schema {
             INFORMATION_SCHEMA_NAME => self.information_schema_provider.table_names(),
-            PG_CATALOG_NAME => self.pg_catalog_provider.table_names(),
+            PG_CATALOG_NAME if channel == Channel::Postgres => {
+                self.pg_catalog_provider.table_names()
+            }
             DEFAULT_SCHEMA_NAME => {
                 vec![NUMBERS_TABLE_NAME.to_string()]
             }
@@ -339,23 +424,35 @@ impl SystemCatalog {
         }
     }
 
-    fn schema_exists(&self, schema: &str) -> bool {
-        schema == INFORMATION_SCHEMA_NAME || schema == PG_CATALOG_NAME
+    fn schema_exists(&self, schema: &str, query_ctx: Option<&QueryContext>) -> bool {
+        let channel = query_ctx.map_or(Channel::Unknown, |ctx| ctx.channel());
+        match channel {
+            Channel::Postgres => schema == PG_CATALOG_NAME || schema == INFORMATION_SCHEMA_NAME,
+            _ => schema == INFORMATION_SCHEMA_NAME,
+        }
     }
 
-    fn table_exists(&self, schema: &str, table: &str) -> bool {
+    fn table_exists(&self, schema: &str, table: &str, query_ctx: Option<&QueryContext>) -> bool {
+        let channel = query_ctx.map_or(Channel::Unknown, |ctx| ctx.channel());
         if schema == INFORMATION_SCHEMA_NAME {
             self.information_schema_provider.table(table).is_some()
         } else if schema == DEFAULT_SCHEMA_NAME {
             table == NUMBERS_TABLE_NAME
-        } else if schema == PG_CATALOG_NAME {
+        } else if schema == PG_CATALOG_NAME && channel == Channel::Postgres {
             self.pg_catalog_provider.table(table).is_some()
         } else {
             false
         }
     }
 
-    fn table(&self, catalog: &str, schema: &str, table_name: &str) -> Option<TableRef> {
+    fn table(
+        &self,
+        catalog: &str,
+        schema: &str,
+        table_name: &str,
+        query_ctx: Option<&QueryContext>,
+    ) -> Option<TableRef> {
+        let channel = query_ctx.map_or(Channel::Unknown, |ctx| ctx.channel());
         if schema == INFORMATION_SCHEMA_NAME {
             let information_schema_provider =
                 self.catalog_cache.get_with_by_ref(catalog, move || {
@@ -366,7 +463,7 @@ impl SystemCatalog {
                     ))
                 });
             information_schema_provider.table(table_name)
-        } else if schema == PG_CATALOG_NAME {
+        } else if schema == PG_CATALOG_NAME && channel == Channel::Postgres {
             if catalog == DEFAULT_CATALOG_NAME {
                 self.pg_catalog_provider.table(table_name)
             } else {

src/catalog/src/lib.rs

@@ -20,8 +20,10 @@ use std::fmt::{Debug, Formatter};
 use std::sync::Arc;
 
 use api::v1::CreateTableExpr;
+use common_catalog::consts::{INFORMATION_SCHEMA_NAME, PG_CATALOG_NAME};
 use futures::future::BoxFuture;
 use futures_util::stream::BoxStream;
+use session::context::QueryContext;
 use table::metadata::TableId;
 use table::TableRef;
 
@@ -44,15 +46,35 @@ pub trait CatalogManager: Send + Sync {
 
     async fn catalog_names(&self) -> Result<Vec<String>>;
 
-    async fn schema_names(&self, catalog: &str) -> Result<Vec<String>>;
+    async fn schema_names(
+        &self,
+        catalog: &str,
+        query_ctx: Option<&QueryContext>,
+    ) -> Result<Vec<String>>;
 
-    async fn table_names(&self, catalog: &str, schema: &str) -> Result<Vec<String>>;
+    async fn table_names(
+        &self,
+        catalog: &str,
+        schema: &str,
+        query_ctx: Option<&QueryContext>,
+    ) -> Result<Vec<String>>;
 
     async fn catalog_exists(&self, catalog: &str) -> Result<bool>;
 
-    async fn schema_exists(&self, catalog: &str, schema: &str) -> Result<bool>;
+    async fn schema_exists(
+        &self,
+        catalog: &str,
+        schema: &str,
+        query_ctx: Option<&QueryContext>,
+    ) -> Result<bool>;
 
-    async fn table_exists(&self, catalog: &str, schema: &str, table: &str) -> Result<bool>;
+    async fn table_exists(
+        &self,
+        catalog: &str,
+        schema: &str,
+        table: &str,
+        query_ctx: Option<&QueryContext>,
+    ) -> Result<bool>;
 
     /// Returns the table by catalog, schema and table name.
     async fn table(
@@ -60,10 +82,25 @@ pub trait CatalogManager: Send + Sync {
         catalog: &str,
         schema: &str,
         table_name: &str,
+        query_ctx: Option<&QueryContext>,
     ) -> Result<Option<TableRef>>;
 
     /// Returns all tables with a stream by catalog and schema.
-    fn tables<'a>(&'a self, catalog: &'a str, schema: &'a str) -> BoxStream<'a, Result<TableRef>>;
+    fn tables<'a>(
+        &'a self,
+        catalog: &'a str,
+        schema: &'a str,
+        query_ctx: Option<&'a QueryContext>,
+    ) -> BoxStream<'a, Result<TableRef>>;
+
+    /// Check if `schema` is a reserved schema name
+    fn is_reserved_schema_name(&self, schema: &str) -> bool {
+        // We have to check whether a schema name is reserved before create schema.
+        // We need this rather than use schema_exists directly because `pg_catalog` is
+        // only visible via postgres protocol. So if we don't check, a mysql client may
+        // create a schema named `pg_catalog` which is somehow malformed.
+        schema == INFORMATION_SCHEMA_NAME || schema == PG_CATALOG_NAME
+    }
 }
 
 pub type CatalogManagerRef = Arc<dyn CatalogManager>;

src/catalog/src/memory/manager.rs

@@ -26,6 +26,7 @@ use common_catalog::consts::{
 use common_meta::key::flow::FlowMetadataManager;
 use common_meta::kv_backend::memory::MemoryKvBackend;
 use futures_util::stream::BoxStream;
+use session::context::QueryContext;
 use snafu::OptionExt;
 use table::TableRef;
 
@@ -53,7 +54,11 @@ impl CatalogManager for MemoryCatalogManager {
         Ok(self.catalogs.read().unwrap().keys().cloned().collect())
     }
 
-    async fn schema_names(&self, catalog: &str) -> Result<Vec<String>> {
+    async fn schema_names(
+        &self,
+        catalog: &str,
+        _query_ctx: Option<&QueryContext>,
+    ) -> Result<Vec<String>> {
         Ok(self
             .catalogs
             .read()
@@ -67,7 +72,12 @@ impl CatalogManager for MemoryCatalogManager {
             .collect())
     }
 
-    async fn table_names(&self, catalog: &str, schema: &str) -> Result<Vec<String>> {
+    async fn table_names(
+        &self,
+        catalog: &str,
+        schema: &str,
+        _query_ctx: Option<&QueryContext>,
+    ) -> Result<Vec<String>> {
         Ok(self
             .catalogs
             .read()
@@ -87,11 +97,22 @@ impl CatalogManager for MemoryCatalogManager {
         self.catalog_exist_sync(catalog)
     }
 
-    async fn schema_exists(&self, catalog: &str, schema: &str) -> Result<bool> {
+    async fn schema_exists(
+        &self,
+        catalog: &str,
+        schema: &str,
+        _query_ctx: Option<&QueryContext>,
+    ) -> Result<bool> {
         self.schema_exist_sync(catalog, schema)
     }
 
-    async fn table_exists(&self, catalog: &str, schema: &str, table: &str) -> Result<bool> {
+    async fn table_exists(
+        &self,
+        catalog: &str,
+        schema: &str,
+        table: &str,
+        _query_ctx: Option<&QueryContext>,
+    ) -> Result<bool> {
         let catalogs = self.catalogs.read().unwrap();
         Ok(catalogs
             .get(catalog)
@@ -108,6 +129,7 @@ impl CatalogManager for MemoryCatalogManager {
         catalog: &str,
         schema: &str,
         table_name: &str,
+        _query_ctx: Option<&QueryContext>,
     ) -> Result<Option<TableRef>> {
         let result = try {
             self.catalogs
@@ -121,7 +143,12 @@ impl CatalogManager for MemoryCatalogManager {
         Ok(result)
     }
 
-    fn tables<'a>(&'a self, catalog: &'a str, schema: &'a str) -> BoxStream<'a, Result<TableRef>> {
+    fn tables<'a>(
+        &'a self,
+        catalog: &'a str,
+        schema: &'a str,
+        _query_ctx: Option<&QueryContext>,
+    ) -> BoxStream<'a, Result<TableRef>> {
         let catalogs = self.catalogs.read().unwrap();
 
         let Some(schemas) = catalogs.get(catalog) else {
@@ -371,11 +398,12 @@ mod tests {
                 DEFAULT_CATALOG_NAME,
                 DEFAULT_SCHEMA_NAME,
                 NUMBERS_TABLE_NAME,
+                None,
             )
             .await
             .unwrap()
             .unwrap();
-        let stream = catalog_list.tables(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME);
+        let stream = catalog_list.tables(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, None);
         let tables = stream.try_collect::<Vec<_>>().await.unwrap();
         assert_eq!(tables.len(), 1);
         assert_eq!(
@@ -384,7 +412,12 @@ mod tests {
         );
 
         assert!(catalog_list
-            .table(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, "not_exists")
+            .table(
+                DEFAULT_CATALOG_NAME,
+                DEFAULT_SCHEMA_NAME,
+                "not_exists",
+                None
+            )
             .await
             .unwrap()
             .is_none());
@@ -411,7 +444,7 @@ mod tests {
         };
         catalog.register_table_sync(register_table_req).unwrap();
         assert!(catalog
-            .table(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, table_name)
+            .table(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, table_name, None)
             .await
             .unwrap()
             .is_some());
@@ -423,7 +456,7 @@ mod tests {
         };
         catalog.deregister_table_sync(deregister_table_req).unwrap();
         assert!(catalog
-            .table(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, table_name)
+            .table(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, table_name, None)
             .await
             .unwrap()
             .is_none());

src/catalog/src/system_schema/information_schema/columns.rs

@@ -257,8 +257,8 @@ impl InformationSchemaColumnsBuilder {
             .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);
+        for schema_name in catalog_manager.schema_names(&catalog_name, None).await? {
+            let mut stream = catalog_manager.tables(&catalog_name, &schema_name, None);
 
             while let Some(table) = stream.try_next().await? {
                 let keys = &table.table_info().meta.primary_key_indices;

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

@@ -212,8 +212,8 @@ impl InformationSchemaKeyColumnUsageBuilder {
             .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);
+        for schema_name in catalog_manager.schema_names(&catalog_name, None).await? {
+            let mut stream = catalog_manager.tables(&catalog_name, &schema_name, None);
 
             while let Some(table) = stream.try_next().await? {
                 let mut primary_constraints = vec![];

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

@@ -240,9 +240,9 @@ impl InformationSchemaPartitionsBuilder {
 
         let predicates = Predicates::from_scan_request(&request);
 
-        for schema_name in catalog_manager.schema_names(&catalog_name).await? {
+        for schema_name in catalog_manager.schema_names(&catalog_name, None).await? {
             let table_info_stream = catalog_manager
-                .tables(&catalog_name, &schema_name)
+                .tables(&catalog_name, &schema_name, None)
                 .try_filter_map(|t| async move {
                     let table_info = t.table_info();
                     if table_info.table_type == TableType::Temporary {

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

@@ -176,9 +176,9 @@ impl InformationSchemaRegionPeersBuilder {
 
         let predicates = Predicates::from_scan_request(&request);
 
-        for schema_name in catalog_manager.schema_names(&catalog_name).await? {
+        for schema_name in catalog_manager.schema_names(&catalog_name, None).await? {
             let table_id_stream = catalog_manager
-                .tables(&catalog_name, &schema_name)
+                .tables(&catalog_name, &schema_name, None)
                 .try_filter_map(|t| async move {
                     let table_info = t.table_info();
                     if table_info.table_type == TableType::Temporary {

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

@@ -171,7 +171,7 @@ impl InformationSchemaSchemataBuilder {
         let table_metadata_manager = utils::table_meta_manager(&self.catalog_manager)?;
         let predicates = Predicates::from_scan_request(&request);
 
-        for schema_name in catalog_manager.schema_names(&catalog_name).await? {
+        for schema_name in catalog_manager.schema_names(&catalog_name, None).await? {
             let opts = if let Some(table_metadata_manager) = &table_metadata_manager {
                 table_metadata_manager
                     .schema_manager()

src/catalog/src/system_schema/information_schema/table_constraints.rs

@@ -176,8 +176,8 @@ impl InformationSchemaTableConstraintsBuilder {
             .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);
+        for schema_name in catalog_manager.schema_names(&catalog_name, None).await? {
+            let mut stream = catalog_manager.tables(&catalog_name, &schema_name, None);
 
             while let Some(table) = stream.try_next().await? {
                 let keys = &table.table_info().meta.primary_key_indices;

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

@@ -234,8 +234,8 @@ impl InformationSchemaTablesBuilder {
             .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);
+        for schema_name in catalog_manager.schema_names(&catalog_name, None).await? {
+            let mut stream = catalog_manager.tables(&catalog_name, &schema_name, None);
 
             while let Some(table) = stream.try_next().await? {
                 let table_info = table.table_info();

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

@@ -192,8 +192,8 @@ impl InformationSchemaViewsBuilder {
             .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);
+        for schema_name in catalog_manager.schema_names(&catalog_name, None).await? {
+            let mut stream = catalog_manager.tables(&catalog_name, &schema_name, None);
 
             while let Some(table) = stream.try_next().await? {
                 let table_info = table.table_info();

src/catalog/src/system_schema/pg_catalog.rs

@@ -18,15 +18,16 @@ mod pg_namespace;
 mod table_names;
 
 use std::collections::HashMap;
-use std::sync::{Arc, Weak};
+use std::sync::{Arc, LazyLock, Weak};
 
-use common_catalog::consts::{self, PG_CATALOG_NAME};
+use common_catalog::consts::{self, DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, 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 session::context::{Channel, QueryContext};
 use table::TableRef;
 pub use table_names::*;
 
@@ -142,3 +143,12 @@ impl SystemSchemaProviderInner for PGCatalogProvider {
         &self.catalog_name
     }
 }
+
+/// Provide query context to call the [`CatalogManager`]'s method.
+static PG_QUERY_CTX: LazyLock<QueryContext> = LazyLock::new(|| {
+    QueryContext::with_channel(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, Channel::Postgres)
+});
+
+fn query_ctx() -> Option<&'static QueryContext> {
+    Some(&PG_QUERY_CTX)
+}

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

@@ -32,7 +32,7 @@ use store_api::storage::ScanRequest;
 use table::metadata::TableType;
 
 use super::pg_namespace::oid_map::PGNamespaceOidMapRef;
-use super::{OID_COLUMN_NAME, PG_CLASS};
+use super::{query_ctx, OID_COLUMN_NAME, PG_CLASS};
 use crate::error::{
     CreateRecordBatchSnafu, InternalSnafu, Result, UpgradeWeakCatalogManagerRefSnafu,
 };
@@ -202,8 +202,11 @@ impl PGClassBuilder {
             .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);
+        for schema_name in catalog_manager
+            .schema_names(&catalog_name, query_ctx())
+            .await?
+        {
+            let mut stream = catalog_manager.tables(&catalog_name, &schema_name, query_ctx());
             while let Some(table) = stream.try_next().await? {
                 let table_info = table.table_info();
                 self.add_class(

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

@@ -31,7 +31,7 @@ use datatypes::vectors::{StringVectorBuilder, UInt32VectorBuilder, VectorRef};
 use snafu::{OptionExt, ResultExt};
 use store_api::storage::ScanRequest;
 
-use super::{PGNamespaceOidMapRef, OID_COLUMN_NAME, PG_NAMESPACE};
+use super::{query_ctx, PGNamespaceOidMapRef, OID_COLUMN_NAME, PG_NAMESPACE};
 use crate::error::{
     CreateRecordBatchSnafu, InternalSnafu, Result, UpgradeWeakCatalogManagerRefSnafu,
 };
@@ -180,7 +180,10 @@ impl PGNamespaceBuilder {
             .upgrade()
             .context(UpgradeWeakCatalogManagerRefSnafu)?;
         let predicates = Predicates::from_scan_request(&request);
-        for schema_name in catalog_manager.schema_names(&catalog_name).await? {
+        for schema_name in catalog_manager
+            .schema_names(&catalog_name, query_ctx())
+            .await?
+        {
             self.add_namespace(&predicates, &schema_name);
         }
         self.finish()

src/catalog/src/table_source.rs

@@ -23,7 +23,7 @@ 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 session::context::QueryContextRef;
 use snafu::{ensure, OptionExt, ResultExt};
 use table::metadata::TableType;
 use table::table::adapter::DfTableProviderAdapter;
@@ -45,6 +45,7 @@ pub struct DfTableSourceProvider {
     disallow_cross_catalog_query: bool,
     default_catalog: String,
     default_schema: String,
+    query_ctx: QueryContextRef,
     plan_decoder: SubstraitPlanDecoderRef,
     enable_ident_normalization: bool,
 }
@@ -53,7 +54,7 @@ impl DfTableSourceProvider {
     pub fn new(
         catalog_manager: CatalogManagerRef,
         disallow_cross_catalog_query: bool,
-        query_ctx: &QueryContext,
+        query_ctx: QueryContextRef,
         plan_decoder: SubstraitPlanDecoderRef,
         enable_ident_normalization: bool,
     ) -> Self {
@@ -63,6 +64,7 @@ impl DfTableSourceProvider {
             resolved_tables: HashMap::new(),
             default_catalog: query_ctx.current_catalog().to_owned(),
             default_schema: query_ctx.current_schema(),
+            query_ctx,
             plan_decoder,
             enable_ident_normalization,
         }
@@ -71,8 +73,7 @@ impl DfTableSourceProvider {
     pub fn resolve_table_ref(&self, table_ref: TableReference) -> Result<ResolvedTableReference> {
         if self.disallow_cross_catalog_query {
             match &table_ref {
-                TableReference::Bare { .. } => (),
-                TableReference::Partial { .. } => {}
+                TableReference::Bare { .. } | TableReference::Partial { .. } => {}
                 TableReference::Full {
                     catalog, schema, ..
                 } => {
@@ -107,7 +108,7 @@ impl DfTableSourceProvider {
 
         let table = self
             .catalog_manager
-            .table(catalog_name, schema_name, table_name)
+            .table(catalog_name, schema_name, table_name, Some(&self.query_ctx))
             .await?
             .with_context(|| TableNotExistSnafu {
                 table: format_full_table_name(catalog_name, schema_name, table_name),
@@ -210,12 +211,12 @@ mod tests {
 
     #[test]
     fn test_validate_table_ref() {
-        let query_ctx = &QueryContext::with("greptime", "public");
+        let query_ctx = Arc::new(QueryContext::with("greptime", "public"));
 
         let table_provider = DfTableSourceProvider::new(
             MemoryCatalogManager::with_default_setup(),
             true,
-            query_ctx,
+            query_ctx.clone(),
             DummyDecoder::arc(),
             true,
         );
@@ -308,7 +309,7 @@ mod tests {
 
     #[tokio::test]
     async fn test_resolve_view() {
-        let query_ctx = &QueryContext::with("greptime", "public");
+        let query_ctx = Arc::new(QueryContext::with("greptime", "public"));
         let backend = Arc::new(MemoryKvBackend::default());
         let layered_cache_builder = LayeredCacheRegistryBuilder::default()
             .add_cache_registry(CacheRegistryBuilder::default().build());
@@ -344,8 +345,13 @@ mod tests {
             .await
             .unwrap();
 
-        let mut table_provider =
-            DfTableSourceProvider::new(catalog_manager, true, query_ctx, MockDecoder::arc(), true);
+        let mut table_provider = DfTableSourceProvider::new(
+            catalog_manager,
+            true,
+            query_ctx.clone(),
+            MockDecoder::arc(),
+            true,
+        );
 
         // View not found
         let table_ref = TableReference::bare("not_exists_view");

src/catalog/src/table_source/dummy_catalog.rs

@@ -112,7 +112,7 @@ impl SchemaProvider for DummySchemaProvider {
     async fn table(&self, name: &str) -> datafusion::error::Result<Option<Arc<dyn TableProvider>>> {
         let table = self
             .catalog_manager
-            .table(&self.catalog_name, &self.schema_name, name)
+            .table(&self.catalog_name, &self.schema_name, name, None)
             .await?
             .with_context(|| TableNotExistSnafu {
                 table: format_full_table_name(&self.catalog_name, &self.schema_name, name),

src/client/src/database.rs

@@ -37,7 +37,8 @@ use tonic::metadata::AsciiMetadataKey;
 use tonic::transport::Channel;
 
 use crate::error::{
-    ConvertFlightDataSnafu, Error, IllegalFlightMessagesSnafu, InvalidAsciiSnafu, ServerSnafu,
+    ConvertFlightDataSnafu, Error, FlightGetSnafu, IllegalFlightMessagesSnafu, InvalidAsciiSnafu,
+    ServerSnafu,
 };
 use crate::{from_grpc_response, Client, Result};
 
@@ -225,16 +226,18 @@ impl Database {
 
         let mut client = self.client.make_flight_client()?;
 
-        let response = client.mut_inner().do_get(request).await.map_err(|e| {
+        let response = client.mut_inner().do_get(request).await.or_else(|e| {
             let tonic_code = e.code();
             let e: Error = e.into();
             let code = e.status_code();
             let msg = e.to_string();
-            let error = Error::FlightGet {
-                tonic_code,
-                addr: client.addr().to_string(),
-                source: BoxedError::new(ServerSnafu { code, msg }.build()),
-            };
+            let error =
+                Err(BoxedError::new(ServerSnafu { code, msg }.build())).with_context(|_| {
+                    FlightGetSnafu {
+                        addr: client.addr().to_string(),
+                        tonic_code,
+                    }
+                });
             error!(
                 "Failed to do Flight get, addr: {}, code: {}, source: {:?}",
                 client.addr(),

src/client/src/error.rs

@@ -39,13 +39,6 @@ pub enum Error {
         source: BoxedError,
     },
 
-    #[snafu(display("Failure occurs during handling request"))]
-    HandleRequest {
-        #[snafu(implicit)]
-        location: Location,
-        source: BoxedError,
-    },
-
     #[snafu(display("Failed to convert FlightData"))]
     ConvertFlightData {
         #[snafu(implicit)]
@@ -116,13 +109,6 @@ pub enum Error {
         location: Location,
     },
 
-    #[snafu(display("Failed to send request with streaming: {}", err_msg))]
-    ClientStreaming {
-        err_msg: String,
-        #[snafu(implicit)]
-        location: Location,
-    },
-
     #[snafu(display("Failed to parse ascii string: {}", value))]
     InvalidAscii {
         value: String,
@@ -138,12 +124,10 @@ impl ErrorExt for Error {
         match self {
             Error::IllegalFlightMessages { .. }
             | Error::MissingField { .. }
-            | Error::IllegalDatabaseResponse { .. }
-            | Error::ClientStreaming { .. } => StatusCode::Internal,
+            | Error::IllegalDatabaseResponse { .. } => StatusCode::Internal,
 
             Error::Server { code, .. } => *code,
             Error::FlightGet { source, .. }
-            | Error::HandleRequest { source, .. }
             | Error::RegionServer { source, .. }
             | Error::FlowServer { source, .. } => source.status_code(),
             Error::CreateChannel { source, .. }

src/client/src/flow.rs

@@ -16,9 +16,9 @@ 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, ResultExt};
+use snafu::ResultExt;
 
-use crate::error::Result;
+use crate::error::{FlowServerSnafu, Result};
 use crate::Client;
 
 #[derive(Debug)]
@@ -57,15 +57,10 @@ impl FlowRequester {
         let response = client
             .handle_create_remove(request)
             .await
-            .map_err(|e| {
+            .or_else(|e| {
                 let code = e.code();
                 let err: crate::error::Error = e.into();
-                crate::error::Error::FlowServer {
-                    addr,
-                    code,
-                    source: BoxedError::new(err),
-                    location: location!(),
-                }
+                Err(BoxedError::new(err)).context(FlowServerSnafu { addr, code })
             })?
             .into_inner();
         Ok(response)
@@ -88,15 +83,10 @@ impl FlowRequester {
         let response = client
             .handle_mirror_request(requests)
             .await
-            .map_err(|e| {
+            .or_else(|e| {
                 let code = e.code();
                 let err: crate::error::Error = e.into();
-                crate::error::Error::FlowServer {
-                    addr,
-                    code,
-                    source: BoxedError::new(err),
-                    location: location!(),
-                }
+                Err(BoxedError::new(err)).context(FlowServerSnafu { addr, code })
             })?
             .into_inner();
         Ok(response)

src/client/src/region.rs

@@ -38,8 +38,8 @@ use substrait::{DFLogicalSubstraitConvertor, SubstraitPlan};
 use tokio_stream::StreamExt;
 
 use crate::error::{
-    self, ConvertFlightDataSnafu, IllegalDatabaseResponseSnafu, IllegalFlightMessagesSnafu,
-    MissingFieldSnafu, Result, ServerSnafu,
+    self, ConvertFlightDataSnafu, FlightGetSnafu, IllegalDatabaseResponseSnafu,
+    IllegalFlightMessagesSnafu, MissingFieldSnafu, Result, ServerSnafu,
 };
 use crate::{metrics, Client, Error};
 
@@ -103,11 +103,14 @@ impl RegionRequester {
                 let e: error::Error = e.into();
                 let code = e.status_code();
                 let msg = e.to_string();
-                let error = Error::FlightGet {
-                    tonic_code,
-                    addr: flight_client.addr().to_string(),
-                    source: BoxedError::new(ServerSnafu { code, msg }.build()),
-                };
+                let error = ServerSnafu { code, msg }
+                    .fail::<()>()
+                    .map_err(BoxedError::new)
+                    .with_context(|_| FlightGetSnafu {
+                        tonic_code,
+                        addr: flight_client.addr().to_string(),
+                    })
+                    .unwrap_err();
                 error!(
                     e; "Failed to do Flight get, addr: {}, code: {}",
                     flight_client.addr(),

src/cmd/Cargo.toml

@@ -70,6 +70,7 @@ serde.workspace = true
 serde_json.workspace = true
 servers.workspace = true
 session.workspace = true
+similar-asserts.workspace = true
 snafu.workspace = true
 store-api.workspace = true
 substrait.workspace = true

src/cmd/src/bin/greptime.rs

@@ -15,10 +15,11 @@
 #![doc = include_str!("../../../../README.md")]
 
 use clap::{Parser, Subcommand};
-use cmd::error::Result;
+use cmd::error::{InitTlsProviderSnafu, Result};
 use cmd::options::GlobalOptions;
 use cmd::{cli, datanode, flownode, frontend, metasrv, standalone, App};
 use common_version::version;
+use servers::install_ring_crypto_provider;
 
 #[derive(Parser)]
 #[command(name = "greptime", author, version, long_version = version(), about)]
@@ -94,6 +95,7 @@ async fn main() -> Result<()> {
 
 async fn main_body() -> Result<()> {
     setup_human_panic();
+    install_ring_crypto_provider().map_err(|msg| InitTlsProviderSnafu { msg }.build())?;
     start(Command::parse()).await
 }
 

src/cmd/src/cli.rs

@@ -21,6 +21,8 @@ mod export;
 mod helper;
 
 // Wait for https://github.com/GreptimeTeam/greptimedb/issues/2373
+mod database;
+mod import;
 #[allow(unused)]
 mod repl;
 
@@ -32,6 +34,7 @@ pub use repl::Repl;
 use tracing_appender::non_blocking::WorkerGuard;
 
 use self::export::ExportCommand;
+use crate::cli::import::ImportCommand;
 use crate::error::Result;
 use crate::options::GlobalOptions;
 use crate::App;
@@ -114,6 +117,7 @@ enum SubCommand {
     // Attach(AttachCommand),
     Bench(BenchTableMetadataCommand),
     Export(ExportCommand),
+    Import(ImportCommand),
 }
 
 impl SubCommand {
@@ -122,6 +126,7 @@ impl SubCommand {
             // SubCommand::Attach(cmd) => cmd.build().await,
             SubCommand::Bench(cmd) => cmd.build(guard).await,
             SubCommand::Export(cmd) => cmd.build(guard).await,
+            SubCommand::Import(cmd) => cmd.build(guard).await,
         }
     }
 }

src/cmd/src/cli/database.rs

@@ -0,0 +1,119 @@
+// 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 base64::engine::general_purpose;
+use base64::Engine;
+use common_catalog::consts::{DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME};
+use serde_json::Value;
+use servers::http::greptime_result_v1::GreptimedbV1Response;
+use servers::http::GreptimeQueryOutput;
+use snafu::ResultExt;
+
+use crate::error::{HttpQuerySqlSnafu, Result, SerdeJsonSnafu};
+
+pub(crate) struct DatabaseClient {
+    addr: String,
+    catalog: String,
+    auth_header: Option<String>,
+}
+
+impl DatabaseClient {
+    pub fn new(addr: String, catalog: String, auth_basic: Option<String>) -> Self {
+        let auth_header = if let Some(basic) = auth_basic {
+            let encoded = general_purpose::STANDARD.encode(basic);
+            Some(format!("basic {}", encoded))
+        } else {
+            None
+        };
+
+        Self {
+            addr,
+            catalog,
+            auth_header,
+        }
+    }
+
+    pub async fn sql_in_public(&self, sql: &str) -> Result<Option<Vec<Vec<Value>>>> {
+        self.sql(sql, DEFAULT_SCHEMA_NAME).await
+    }
+
+    /// Execute sql query.
+    pub async fn sql(&self, sql: &str, schema: &str) -> Result<Option<Vec<Vec<Value>>>> {
+        let url = format!("http://{}/v1/sql", self.addr);
+        let params = [
+            ("db", format!("{}-{}", self.catalog, schema)),
+            ("sql", sql.to_string()),
+        ];
+        let mut request = reqwest::Client::new()
+            .post(&url)
+            .form(&params)
+            .header("Content-Type", "application/x-www-form-urlencoded");
+        if let Some(ref auth) = self.auth_header {
+            request = request.header("Authorization", auth);
+        }
+
+        let response = request.send().await.with_context(|_| HttpQuerySqlSnafu {
+            reason: format!("bad url: {}", url),
+        })?;
+        let response = response
+            .error_for_status()
+            .with_context(|_| HttpQuerySqlSnafu {
+                reason: format!("query failed: {}", sql),
+            })?;
+
+        let text = response.text().await.with_context(|_| HttpQuerySqlSnafu {
+            reason: "cannot get response text".to_string(),
+        })?;
+
+        let body = serde_json::from_str::<GreptimedbV1Response>(&text).context(SerdeJsonSnafu)?;
+        Ok(body.output().first().and_then(|output| match output {
+            GreptimeQueryOutput::Records(records) => Some(records.rows().clone()),
+            GreptimeQueryOutput::AffectedRows(_) => None,
+        }))
+    }
+}
+
+/// Split at `-`.
+pub(crate) fn split_database(database: &str) -> Result<(String, Option<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 {
+        Ok((catalog.to_string(), Some(schema.to_string())))
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[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));
+    }
+}

src/cmd/src/cli/export.rs

@@ -13,30 +13,23 @@
 // limitations under the License.
 
 use std::collections::HashSet;
-use std::path::Path;
+use std::path::PathBuf;
 use std::sync::Arc;
 
 use async_trait::async_trait;
-use base64::engine::general_purpose;
-use base64::Engine;
 use clap::{Parser, ValueEnum};
-use client::DEFAULT_SCHEMA_NAME;
-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::ResultExt;
+use snafu::{OptionExt, ResultExt};
 use tokio::fs::File;
 use tokio::io::{AsyncWriteExt, BufWriter};
 use tokio::sync::Semaphore;
 use tokio::time::Instant;
 use tracing_appender::non_blocking::WorkerGuard;
 
-use crate::cli::{Instance, Tool};
-use crate::error::{
-    EmptyResultSnafu, Error, FileIoSnafu, HttpQuerySqlSnafu, Result, SerdeJsonSnafu,
-};
+use crate::cli::database::DatabaseClient;
+use crate::cli::{database, Instance, Tool};
+use crate::error::{EmptyResultSnafu, Error, FileIoSnafu, Result, SchemaNotFoundSnafu};
 
 type TableReference = (String, String, String);
 
@@ -94,26 +87,21 @@ pub struct ExportCommand {
 
 impl ExportCommand {
     pub async fn build(&self, guard: Vec<WorkerGuard>) -> Result<Instance> {
-        let (catalog, schema) = split_database(&self.database)?;
+        let (catalog, schema) = database::split_database(&self.database)?;
 
-        let auth_header = if let Some(basic) = &self.auth_basic {
-            let encoded = general_purpose::STANDARD.encode(basic);
-            Some(format!("basic {}", encoded))
-        } else {
-            None
-        };
+        let database_client =
+            DatabaseClient::new(self.addr.clone(), catalog.clone(), self.auth_basic.clone());
 
         Ok(Instance::new(
             Box::new(Export {
-                addr: self.addr.clone(),
                 catalog,
                 schema,
+                database_client,
                 output_dir: self.output_dir.clone(),
                 parallelism: self.export_jobs,
                 target: self.target.clone(),
                 start_time: self.start_time.clone(),
                 end_time: self.end_time.clone(),
-                auth_header,
             }),
             guard,
         ))
@@ -121,78 +109,59 @@ impl ExportCommand {
 }
 
 pub struct Export {
-    addr: String,
     catalog: String,
     schema: Option<String>,
+    database_client: DatabaseClient,
     output_dir: String,
     parallelism: usize,
     target: ExportTarget,
     start_time: Option<String>,
     end_time: Option<String>,
-    auth_header: Option<String>,
 }
 
 impl Export {
-    /// Execute one single sql query.
-    async fn sql(&self, sql: &str) -> Result<Option<Vec<Vec<Value>>>> {
-        let url = format!(
-            "http://{}/v1/sql?db={}-{}&sql={}",
-            self.addr,
-            self.catalog,
-            self.schema.as_deref().unwrap_or(DEFAULT_SCHEMA_NAME),
-            sql
-        );
+    fn catalog_path(&self) -> PathBuf {
+        PathBuf::from(&self.output_dir).join(&self.catalog)
+    }
 
-        let mut request = reqwest::Client::new()
-            .get(&url)
-            .header("Content-Type", "application/x-www-form-urlencoded");
-        if let Some(ref auth) = self.auth_header {
-            request = request.header("Authorization", auth);
-        }
+    async fn get_db_names(&self) -> Result<Vec<String>> {
+        let db_names = self.all_db_names().await?;
+        let Some(schema) = &self.schema else {
+            return Ok(db_names);
+        };
 
-        let response = request.send().await.with_context(|_| HttpQuerySqlSnafu {
-            reason: format!("bad url: {}", url),
-        })?;
-        let response = response
-            .error_for_status()
-            .with_context(|_| HttpQuerySqlSnafu {
-                reason: format!("query failed: {}", sql),
-            })?;
-
-        let text = response.text().await.with_context(|_| HttpQuerySqlSnafu {
-            reason: "cannot get response text".to_string(),
-        })?;
-
-        let body = serde_json::from_str::<GreptimedbV1Response>(&text).context(SerdeJsonSnafu)?;
-        Ok(body.output().first().and_then(|output| match output {
-            GreptimeQueryOutput::Records(records) => Some(records.rows().clone()),
-            GreptimeQueryOutput::AffectedRows(_) => None,
-        }))
+        // Check if the schema exists
+        db_names
+            .into_iter()
+            .find(|db_name| db_name.to_lowercase() == schema.to_lowercase())
+            .map(|name| vec![name])
+            .context(SchemaNotFoundSnafu {
+                catalog: &self.catalog,
+                schema,
+            })
     }
 
     /// Iterate over all db names.
-    ///
-    /// Newbie: `db_name` is catalog + schema.
-    async fn iter_db_names(&self) -> Result<Vec<(String, String)>> {
-        if let Some(schema) = &self.schema {
-            Ok(vec![(self.catalog.clone(), schema.clone())])
-        } else {
-            let result = self.sql("SHOW DATABASES").await?;
-            let Some(records) = result else {
-                EmptyResultSnafu.fail()?
+    async fn all_db_names(&self) -> Result<Vec<String>> {
+        let records = self
+            .database_client
+            .sql_in_public("SHOW DATABASES")
+            .await?
+            .context(EmptyResultSnafu)?;
+        let mut result = Vec::with_capacity(records.len());
+        for value in records {
+            let Value::String(schema) = &value[0] else {
+                unreachable!()
             };
-            let mut result = Vec::with_capacity(records.len());
-            for value in records {
-                let Value::String(schema) = &value[0] else {
-                    unreachable!()
-                };
-                if schema == common_catalog::consts::INFORMATION_SCHEMA_NAME {
-                    continue;
-                }
-                result.push((self.catalog.clone(), schema.clone()));
+            if schema == common_catalog::consts::INFORMATION_SCHEMA_NAME {
+                continue;
             }
-            Ok(result)
+            if schema == common_catalog::consts::PG_CATALOG_NAME {
+                continue;
+            }
+            result.push(schema.clone());
         }
+        Ok(result)
     }
 
     /// Return a list of [`TableReference`] to be exported.
@@ -201,7 +170,11 @@ impl Export {
         &self,
         catalog: &str,
         schema: &str,
-    ) -> Result<(Vec<TableReference>, Vec<TableReference>)> {
+    ) -> Result<(
+        Vec<TableReference>,
+        Vec<TableReference>,
+        Vec<TableReference>,
+    )> {
         // Puts all metric table first
         let sql = format!(
             "SELECT table_catalog, table_schema, table_name \
@@ -210,15 +183,16 @@ impl Export {
                 and table_catalog = \'{catalog}\' \
                 and table_schema = \'{schema}\'"
         );
-        let result = self.sql(&sql).await?;
-        let Some(records) = result else {
-            EmptyResultSnafu.fail()?
-        };
+        let records = self
+            .database_client
+            .sql_in_public(&sql)
+            .await?
+            .context(EmptyResultSnafu)?;
         let mut metric_physical_tables = HashSet::with_capacity(records.len());
         for value in records {
             let mut t = Vec::with_capacity(3);
             for v in &value {
-                let serde_json::Value::String(value) = v else {
+                let Value::String(value) = v else {
                     unreachable!()
                 };
                 t.push(value);
@@ -226,100 +200,142 @@ impl Export {
             metric_physical_tables.insert((t[0].clone(), t[1].clone(), t[2].clone()));
         }
 
-        // TODO: SQL injection hurts
         let sql = format!(
-            "SELECT table_catalog, table_schema, table_name \
+            "SELECT table_catalog, table_schema, table_name, table_type \
             FROM information_schema.tables \
-            WHERE table_type = \'BASE TABLE\' \
+            WHERE (table_type = \'BASE TABLE\' OR table_type = \'VIEW\') \
                 and table_catalog = \'{catalog}\' \
                 and table_schema = \'{schema}\'",
         );
-        let result = self.sql(&sql).await?;
-        let Some(records) = result else {
-            EmptyResultSnafu.fail()?
-        };
+        let records = self
+            .database_client
+            .sql_in_public(&sql)
+            .await?
+            .context(EmptyResultSnafu)?;
 
-        debug!("Fetched table list: {:?}", records);
+        debug!("Fetched table/view list: {:?}", records);
 
         if records.is_empty() {
-            return Ok((vec![], vec![]));
+            return Ok((vec![], vec![], vec![]));
         }
 
         let mut remaining_tables = Vec::with_capacity(records.len());
+        let mut views = Vec::new();
         for value in records {
-            let mut t = Vec::with_capacity(3);
+            let mut t = Vec::with_capacity(4);
             for v in &value {
-                let serde_json::Value::String(value) = v else {
+                let Value::String(value) = v else {
                     unreachable!()
                 };
                 t.push(value);
             }
             let table = (t[0].clone(), t[1].clone(), t[2].clone());
+            let table_type = t[3].as_str();
             // Ignores the physical table
             if !metric_physical_tables.contains(&table) {
-                remaining_tables.push(table);
+                if table_type == "VIEW" {
+                    views.push(table);
+                } else {
+                    remaining_tables.push(table);
+                }
             }
         }
 
         Ok((
             metric_physical_tables.into_iter().collect(),
             remaining_tables,
+            views,
         ))
     }
 
-    async fn show_create_table(&self, catalog: &str, schema: &str, table: &str) -> Result<String> {
-        let sql = format!(
-            r#"SHOW CREATE TABLE "{}"."{}"."{}""#,
-            catalog, schema, table
-        );
-        let result = self.sql(&sql).await?;
-        let Some(records) = result else {
-            EmptyResultSnafu.fail()?
+    async fn show_create(
+        &self,
+        show_type: &str,
+        catalog: &str,
+        schema: &str,
+        table: Option<&str>,
+    ) -> Result<String> {
+        let sql = match table {
+            Some(table) => format!(
+                r#"SHOW CREATE {} "{}"."{}"."{}""#,
+                show_type, catalog, schema, table
+            ),
+            None => format!(r#"SHOW CREATE {} "{}"."{}""#, show_type, catalog, schema),
         };
-        let Value::String(create_table) = &records[0][1] else {
+        let records = self
+            .database_client
+            .sql_in_public(&sql)
+            .await?
+            .context(EmptyResultSnafu)?;
+        let Value::String(create) = &records[0][1] else {
             unreachable!()
         };
 
-        Ok(format!("{};\n", create_table))
+        Ok(format!("{};\n", create))
+    }
+
+    async fn export_create_database(&self) -> Result<()> {
+        let timer = Instant::now();
+        let db_names = self.get_db_names().await?;
+        let db_count = db_names.len();
+        for schema in db_names {
+            let db_dir = self.catalog_path().join(format!("{schema}/"));
+            tokio::fs::create_dir_all(&db_dir)
+                .await
+                .context(FileIoSnafu)?;
+            let file = db_dir.join("create_database.sql");
+            let mut file = File::create(file).await.context(FileIoSnafu)?;
+            let create_database = self
+                .show_create("DATABASE", &self.catalog, &schema, None)
+                .await?;
+            file.write_all(create_database.as_bytes())
+                .await
+                .context(FileIoSnafu)?;
+        }
+
+        let elapsed = timer.elapsed();
+        info!("Success {db_count} jobs, cost: {elapsed:?}");
+
+        Ok(())
     }
 
     async fn export_create_table(&self) -> Result<()> {
         let timer = Instant::now();
         let semaphore = Arc::new(Semaphore::new(self.parallelism));
-        let db_names = self.iter_db_names().await?;
+        let db_names = self.get_db_names().await?;
         let db_count = db_names.len();
         let mut tasks = Vec::with_capacity(db_names.len());
-        for (catalog, schema) in db_names {
+        for schema in db_names {
             let semaphore_moved = semaphore.clone();
             tasks.push(async move {
                 let _permit = semaphore_moved.acquire().await.unwrap();
-                let (metric_physical_tables, remaining_tables) =
-                    self.get_table_list(&catalog, &schema).await?;
-                let table_count = metric_physical_tables.len() + remaining_tables.len();
-                let output_dir = Path::new(&self.output_dir)
-                    .join(&catalog)
-                    .join(format!("{schema}/"));
-                tokio::fs::create_dir_all(&output_dir)
+                let (metric_physical_tables, remaining_tables, views) =
+                    self.get_table_list(&self.catalog, &schema).await?;
+                let table_count =
+                    metric_physical_tables.len() + remaining_tables.len() + views.len();
+                let db_dir = self.catalog_path().join(format!("{schema}/"));
+                tokio::fs::create_dir_all(&db_dir)
                     .await
                     .context(FileIoSnafu)?;
-                let output_file = Path::new(&output_dir).join("create_tables.sql");
-                let mut file = File::create(output_file).await.context(FileIoSnafu)?;
+                let file = db_dir.join("create_tables.sql");
+                let mut file = File::create(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 {
-                        Err(e) => {
-                            error!(e; r#"Failed to export table "{}"."{}"."{}""#, c, s, t)
-                        }
-                        Ok(create_table) => {
-                            file.write_all(create_table.as_bytes())
-                                .await
-                                .context(FileIoSnafu)?;
-                        }
-                    }
+                    let create_table = self.show_create("TABLE", &c, &s, Some(&t)).await?;
+                    file.write_all(create_table.as_bytes())
+                        .await
+                        .context(FileIoSnafu)?;
+                }
+                for (c, s, v) in views {
+                    let create_view = self.show_create("VIEW", &c, &s, Some(&v)).await?;
+                    file.write_all(create_view.as_bytes())
+                        .await
+                        .context(FileIoSnafu)?;
                 }
 
                 info!(
-                    "Finished exporting {catalog}.{schema} with {table_count} table schemas to path: {}",
-                    output_dir.to_string_lossy()
+                    "Finished exporting {}.{schema} with {table_count} table schemas to path: {}",
+                    self.catalog,
+                    db_dir.to_string_lossy()
                 );
 
                 Ok::<(), Error>(())
@@ -332,14 +348,14 @@ impl Export {
             .filter(|r| match r {
                 Ok(_) => true,
                 Err(e) => {
-                    error!(e; "export job failed");
+                    error!(e; "export schema job failed");
                     false
                 }
             })
             .count();
 
         let elapsed = timer.elapsed();
-        info!("Success {success}/{db_count} jobs, cost: {:?}", elapsed);
+        info!("Success {success}/{db_count} jobs, cost: {elapsed:?}");
 
         Ok(())
     }
@@ -347,17 +363,15 @@ impl Export {
     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?;
+        let db_names = self.get_db_names().await?;
         let db_count = db_names.len();
-        let mut tasks = Vec::with_capacity(db_names.len());
-        for (catalog, schema) in db_names {
+        let mut tasks = Vec::with_capacity(db_count);
+        for schema in db_names {
             let semaphore_moved = semaphore.clone();
             tasks.push(async move {
                 let _permit = semaphore_moved.acquire().await.unwrap();
-                let output_dir = Path::new(&self.output_dir)
-                    .join(&catalog)
-                    .join(format!("{schema}/"));
-                tokio::fs::create_dir_all(&output_dir)
+                let db_dir = self.catalog_path().join(format!("{schema}/"));
+                tokio::fs::create_dir_all(&db_dir)
                     .await
                     .context(FileIoSnafu)?;
 
@@ -379,30 +393,31 @@ impl Export {
 
                 let sql = format!(
                     r#"COPY DATABASE "{}"."{}" TO '{}' {};"#,
-                    catalog,
+                    self.catalog,
                     schema,
-                    output_dir.to_str().unwrap(),
+                    db_dir.to_str().unwrap(),
                     with_options
                 );
 
                 info!("Executing sql: {sql}");
 
-                self.sql(&sql).await?;
+                self.database_client.sql_in_public(&sql).await?;
 
                 info!(
-                    "Finished exporting {catalog}.{schema} data into path: {}",
-                    output_dir.to_string_lossy()
+                    "Finished exporting {}.{schema} data into path: {}",
+                    self.catalog,
+                    db_dir.to_string_lossy()
                 );
 
                 // The export copy from sql
-                let copy_from_file = output_dir.join("copy_from.sql");
+                let copy_from_file = db_dir.join("copy_from.sql");
                 let mut writer =
                     BufWriter::new(File::create(copy_from_file).await.context(FileIoSnafu)?);
                 let copy_database_from_sql = format!(
                     r#"COPY DATABASE "{}"."{}" FROM '{}' WITH (FORMAT='parquet');"#,
-                    catalog,
+                    self.catalog,
                     schema,
-                    output_dir.to_str().unwrap()
+                    db_dir.to_str().unwrap()
                 );
                 writer
                     .write(copy_database_from_sql.as_bytes())
@@ -410,7 +425,7 @@ impl Export {
                     .context(FileIoSnafu)?;
                 writer.flush().await.context(FileIoSnafu)?;
 
-                info!("Finished exporting {catalog}.{schema} copy_from.sql");
+                info!("Finished exporting {}.{schema} copy_from.sql", self.catalog);
 
                 Ok::<(), Error>(())
             })
@@ -429,20 +444,23 @@ impl Export {
             .count();
         let elapsed = timer.elapsed();
 
-        info!("Success {success}/{db_count} jobs, costs: {:?}", 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::Schema => self.export_create_table().await,
+            ExportTarget::Schema => {
+                self.export_create_database().await?;
+                self.export_create_table().await
+            }
             ExportTarget::Data => self.export_database_data().await,
             ExportTarget::All => {
+                self.export_create_database().await?;
                 self.export_create_table().await?;
                 self.export_database_data().await
             }
@@ -450,20 +468,6 @@ impl Tool for Export {
     }
 }
 
-/// Split at `-`.
-fn split_database(database: &str) -> Result<(String, Option<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 {
-        Ok((catalog.to_string(), Some(schema.to_string())))
-    }
-}
-
 #[cfg(test)]
 mod tests {
     use clap::Parser;
@@ -471,26 +475,10 @@ 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();

src/cmd/src/cli/import.rs

@@ -0,0 +1,218 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use std::path::PathBuf;
+use std::sync::Arc;
+
+use async_trait::async_trait;
+use clap::{Parser, ValueEnum};
+use common_catalog::consts::DEFAULT_SCHEMA_NAME;
+use common_telemetry::{error, info, warn};
+use snafu::{OptionExt, ResultExt};
+use tokio::sync::Semaphore;
+use tokio::time::Instant;
+use tracing_appender::non_blocking::WorkerGuard;
+
+use crate::cli::database::DatabaseClient;
+use crate::cli::{database, Instance, Tool};
+use crate::error::{Error, FileIoSnafu, Result, SchemaNotFoundSnafu};
+
+#[derive(Debug, Default, Clone, ValueEnum)]
+enum ImportTarget {
+    /// Import all table schemas into the database.
+    Schema,
+    /// Import all table data into the database.
+    Data,
+    /// Export all table schemas and data at once.
+    #[default]
+    All,
+}
+
+#[derive(Debug, Default, Parser)]
+pub struct ImportCommand {
+    /// Server address to connect
+    #[clap(long)]
+    addr: String,
+
+    /// Directory of the data. E.g.: /tmp/greptimedb-backup
+    #[clap(long)]
+    input_dir: String,
+
+    /// The name of the catalog to import.
+    #[clap(long, default_value = "greptime-*")]
+    database: String,
+
+    /// Parallelism of the import.
+    #[clap(long, short = 'j', default_value = "1")]
+    import_jobs: usize,
+
+    /// Max retry times for each job.
+    #[clap(long, default_value = "3")]
+    max_retry: usize,
+
+    /// Things to export
+    #[clap(long, short = 't', value_enum, default_value = "all")]
+    target: ImportTarget,
+
+    /// The basic authentication for connecting to the server
+    #[clap(long)]
+    auth_basic: Option<String>,
+}
+
+impl ImportCommand {
+    pub async fn build(&self, guard: Vec<WorkerGuard>) -> Result<Instance> {
+        let (catalog, schema) = database::split_database(&self.database)?;
+        let database_client =
+            DatabaseClient::new(self.addr.clone(), catalog.clone(), self.auth_basic.clone());
+
+        Ok(Instance::new(
+            Box::new(Import {
+                catalog,
+                schema,
+                database_client,
+                input_dir: self.input_dir.clone(),
+                parallelism: self.import_jobs,
+                target: self.target.clone(),
+            }),
+            guard,
+        ))
+    }
+}
+
+pub struct Import {
+    catalog: String,
+    schema: Option<String>,
+    database_client: DatabaseClient,
+    input_dir: String,
+    parallelism: usize,
+    target: ImportTarget,
+}
+
+impl Import {
+    async fn import_create_table(&self) -> Result<()> {
+        // Use default db to creates other dbs
+        self.do_sql_job("create_database.sql", Some(DEFAULT_SCHEMA_NAME))
+            .await?;
+        self.do_sql_job("create_tables.sql", None).await
+    }
+
+    async fn import_database_data(&self) -> Result<()> {
+        self.do_sql_job("copy_from.sql", None).await
+    }
+
+    async fn do_sql_job(&self, filename: &str, exec_db: Option<&str>) -> Result<()> {
+        let timer = Instant::now();
+        let semaphore = Arc::new(Semaphore::new(self.parallelism));
+        let db_names = self.get_db_names().await?;
+        let db_count = db_names.len();
+        let mut tasks = Vec::with_capacity(db_count);
+        for schema in db_names {
+            let semaphore_moved = semaphore.clone();
+            tasks.push(async move {
+                let _permit = semaphore_moved.acquire().await.unwrap();
+                let database_input_dir = self.catalog_path().join(&schema);
+                let sql_file = database_input_dir.join(filename);
+                let sql = tokio::fs::read_to_string(sql_file)
+                    .await
+                    .context(FileIoSnafu)?;
+                if sql.is_empty() {
+                    info!("Empty `{filename}` {database_input_dir:?}");
+                } else {
+                    let db = exec_db.unwrap_or(&schema);
+                    self.database_client.sql(&sql, db).await?;
+                    info!("Imported `{filename}` for database {schema}");
+                }
+
+                Ok::<(), Error>(())
+            })
+        }
+
+        let success = futures::future::join_all(tasks)
+            .await
+            .into_iter()
+            .filter(|r| match r {
+                Ok(_) => true,
+                Err(e) => {
+                    error!(e; "import {filename} job failed");
+                    false
+                }
+            })
+            .count();
+        let elapsed = timer.elapsed();
+        info!("Success {success}/{db_count} `{filename}` jobs, cost: {elapsed:?}");
+
+        Ok(())
+    }
+
+    fn catalog_path(&self) -> PathBuf {
+        PathBuf::from(&self.input_dir).join(&self.catalog)
+    }
+
+    async fn get_db_names(&self) -> Result<Vec<String>> {
+        let db_names = self.all_db_names().await?;
+        let Some(schema) = &self.schema else {
+            return Ok(db_names);
+        };
+
+        // Check if the schema exists
+        db_names
+            .into_iter()
+            .find(|db_name| db_name.to_lowercase() == schema.to_lowercase())
+            .map(|name| vec![name])
+            .context(SchemaNotFoundSnafu {
+                catalog: &self.catalog,
+                schema,
+            })
+    }
+
+    // Get all database names in the input directory.
+    // The directory structure should be like:
+    // /tmp/greptimedb-backup
+    // ├── greptime-1
+    // │   ├── db1
+    // │   └── db2
+    async fn all_db_names(&self) -> Result<Vec<String>> {
+        let mut db_names = vec![];
+        let path = self.catalog_path();
+        let mut entries = tokio::fs::read_dir(path).await.context(FileIoSnafu)?;
+        while let Some(entry) = entries.next_entry().await.context(FileIoSnafu)? {
+            let path = entry.path();
+            if path.is_dir() {
+                let db_name = match path.file_name() {
+                    Some(name) => name.to_string_lossy().to_string(),
+                    None => {
+                        warn!("Failed to get the file name of {:?}", path);
+                        continue;
+                    }
+                };
+                db_names.push(db_name);
+            }
+        }
+        Ok(db_names)
+    }
+}
+
+#[async_trait]
+impl Tool for Import {
+    async fn do_work(&self) -> Result<()> {
+        match self.target {
+            ImportTarget::Schema => self.import_create_table().await,
+            ImportTarget::Data => self.import_database_data().await,
+            ImportTarget::All => {
+                self.import_create_table().await?;
+                self.import_database_data().await
+            }
+        }
+    }
+}

src/cmd/src/error.rs

@@ -24,15 +24,14 @@ use snafu::{Location, Snafu};
 #[snafu(visibility(pub))]
 #[stack_trace_debug]
 pub enum Error {
-    #[snafu(display("Failed to create default catalog and schema"))]
-    InitMetadata {
+    #[snafu(display("Failed to install ring crypto provider: {}", msg))]
+    InitTlsProvider {
         #[snafu(implicit)]
         location: Location,
-        source: common_meta::error::Error,
+        msg: String,
     },
-
-    #[snafu(display("Failed to iter stream"))]
-    IterStream {
+    #[snafu(display("Failed to create default catalog and schema"))]
+    InitMetadata {
         #[snafu(implicit)]
         location: Location,
         source: common_meta::error::Error,
@@ -237,13 +236,6 @@ pub enum Error {
         location: Location,
     },
 
-    #[snafu(display("Failed to start catalog manager"))]
-    StartCatalogManager {
-        #[snafu(implicit)]
-        location: Location,
-        source: catalog::error::Error,
-    },
-
     #[snafu(display("Failed to connect to Etcd at {etcd_addr}"))]
     ConnectEtcd {
         etcd_addr: String,
@@ -253,14 +245,6 @@ pub enum Error {
         location: Location,
     },
 
-    #[snafu(display("Failed to connect server at {addr}"))]
-    ConnectServer {
-        addr: String,
-        source: client::error::Error,
-        #[snafu(implicit)]
-        location: Location,
-    },
-
     #[snafu(display("Failed to serde json"))]
     SerdeJson {
         #[snafu(source)]
@@ -278,12 +262,6 @@ pub enum Error {
         location: Location,
     },
 
-    #[snafu(display("Expect data from output, but got another thing"))]
-    NotDataFromOutput {
-        #[snafu(implicit)]
-        location: Location,
-    },
-
     #[snafu(display("Empty result from output"))]
     EmptyResult {
         #[snafu(implicit)]
@@ -346,13 +324,12 @@ pub enum Error {
         source: meta_client::error::Error,
     },
 
-    #[snafu(display("Tonic transport error: {error:?} with msg: {msg:?}"))]
-    TonicTransport {
+    #[snafu(display("Cannot find schema {schema} in catalog {catalog}"))]
+    SchemaNotFound {
+        catalog: String,
+        schema: String,
         #[snafu(implicit)]
         location: Location,
-        #[snafu(source)]
-        error: tonic::transport::Error,
-        msg: Option<String>,
     },
 }
 
@@ -370,18 +347,16 @@ impl ErrorExt for Error {
             Error::BuildMetaServer { source, .. } => source.status_code(),
             Error::UnsupportedSelectorType { source, .. } => source.status_code(),
 
-            Error::IterStream { source, .. }
-            | Error::InitMetadata { source, .. }
-            | Error::InitDdlManager { source, .. } => source.status_code(),
+            Error::InitMetadata { source, .. } | Error::InitDdlManager { source, .. } => {
+                source.status_code()
+            }
 
-            Error::ConnectServer { source, .. } => source.status_code(),
             Error::MissingConfig { .. }
             | Error::LoadLayeredConfig { .. }
             | Error::IllegalConfig { .. }
             | Error::InvalidReplCommand { .. }
             | Error::InitTimezone { .. }
             | Error::ConnectEtcd { .. }
-            | Error::NotDataFromOutput { .. }
             | Error::CreateDir { .. }
             | Error::EmptyResult { .. } => StatusCode::InvalidArguments,
 
@@ -399,11 +374,11 @@ impl ErrorExt for Error {
                 source.status_code()
             }
             Error::SubstraitEncodeLogicalPlan { source, .. } => source.status_code(),
-            Error::StartCatalogManager { source, .. } => source.status_code(),
 
-            Error::SerdeJson { .. } | Error::FileIo { .. } | Error::SpawnThread { .. } => {
-                StatusCode::Unexpected
-            }
+            Error::SerdeJson { .. }
+            | Error::FileIo { .. }
+            | Error::SpawnThread { .. }
+            | Error::InitTlsProvider { .. } => StatusCode::Unexpected,
 
             Error::Other { source, .. } => source.status_code(),
 
@@ -414,7 +389,7 @@ impl ErrorExt for Error {
                 source.status_code()
             }
             Error::MetaClientInit { source, .. } => source.status_code(),
-            Error::TonicTransport { .. } => StatusCode::Internal,
+            Error::SchemaNotFound { .. } => StatusCode::DatabaseNotFound,
         }
     }
 

src/cmd/src/standalone.rs

@@ -141,6 +141,8 @@ pub struct StandaloneOptions {
     pub region_engine: Vec<RegionEngineConfig>,
     pub export_metrics: ExportMetricsOption,
     pub tracing: TracingOptions,
+    pub init_regions_in_background: bool,
+    pub init_regions_parallelism: usize,
 }
 
 impl Default for StandaloneOptions {
@@ -168,6 +170,8 @@ impl Default for StandaloneOptions {
                 RegionEngineConfig::File(FileEngineConfig::default()),
             ],
             tracing: TracingOptions::default(),
+            init_regions_in_background: false,
+            init_regions_parallelism: 16,
         }
     }
 }
@@ -218,6 +222,9 @@ impl StandaloneOptions {
             storage: cloned_opts.storage,
             region_engine: cloned_opts.region_engine,
             grpc: cloned_opts.grpc,
+            init_regions_in_background: cloned_opts.init_regions_in_background,
+            init_regions_parallelism: cloned_opts.init_regions_parallelism,
+            mode: Mode::Standalone,
             ..Default::default()
         }
     }

src/cmd/tests/load_config_test.rs

@@ -16,12 +16,10 @@ use std::time::Duration;
 
 use cmd::options::GreptimeOptions;
 use cmd::standalone::StandaloneOptions;
-use common_base::readable_size::ReadableSize;
 use common_config::Configurable;
 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, DEFAULT_OTLP_ENDPOINT};
 use common_wal::config::raft_engine::RaftEngineConfig;
 use common_wal::config::DatanodeWalConfig;
@@ -45,10 +43,6 @@ fn test_load_datanode_example_config() {
             .unwrap();
 
     let expected = GreptimeOptions::<DatanodeOptions> {
-        runtime: RuntimeOptions {
-            global_rt_size: 8,
-            compact_rt_size: 4,
-        },
         component: DatanodeOptions {
             node_id: Some(42),
             meta_client: Some(MetaClientOptions {
@@ -65,6 +59,7 @@ fn test_load_datanode_example_config() {
             wal: DatanodeWalConfig::RaftEngine(RaftEngineConfig {
                 dir: Some("/tmp/greptimedb/wal".to_string()),
                 sync_period: Some(Duration::from_secs(10)),
+                recovery_parallelism: 2,
                 ..Default::default()
             }),
             storage: StorageConfig {
@@ -73,15 +68,8 @@ fn test_load_datanode_example_config() {
             },
             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,
                     experimental_write_cache_ttl: Some(Duration::from_secs(60 * 60 * 8)),
                     ..Default::default()
                 }),
@@ -106,9 +94,10 @@ fn test_load_datanode_example_config() {
             rpc_max_send_message_size: Some(DEFAULT_MAX_GRPC_SEND_MESSAGE_SIZE),
             ..Default::default()
         },
+        ..Default::default()
     };
 
-    assert_eq!(options, expected);
+    similar_asserts::assert_eq!(options, expected);
 }
 
 #[test]
@@ -118,10 +107,6 @@ fn test_load_frontend_example_config() {
         GreptimeOptions::<FrontendOptions>::load_layered_options(example_config.to_str(), "")
             .unwrap();
     let expected = GreptimeOptions::<FrontendOptions> {
-        runtime: RuntimeOptions {
-            global_rt_size: 8,
-            compact_rt_size: 4,
-        },
         component: FrontendOptions {
             default_timezone: Some("UTC".to_string()),
             meta_client: Some(MetaClientOptions {
@@ -154,8 +139,9 @@ fn test_load_frontend_example_config() {
             },
             ..Default::default()
         },
+        ..Default::default()
     };
-    assert_eq!(options, expected);
+    similar_asserts::assert_eq!(options, expected);
 }
 
 #[test]
@@ -165,10 +151,6 @@ fn test_load_metasrv_example_config() {
         GreptimeOptions::<MetasrvOptions>::load_layered_options(example_config.to_str(), "")
             .unwrap();
     let expected = GreptimeOptions::<MetasrvOptions> {
-        runtime: RuntimeOptions {
-            global_rt_size: 8,
-            compact_rt_size: 4,
-        },
         component: MetasrvOptions {
             selector: SelectorType::default(),
             data_home: "/tmp/metasrv/".to_string(),
@@ -186,8 +168,9 @@ fn test_load_metasrv_example_config() {
             },
             ..Default::default()
         },
+        ..Default::default()
     };
-    assert_eq!(options, expected);
+    similar_asserts::assert_eq!(options, expected);
 }
 
 #[test]
@@ -197,30 +180,19 @@ fn test_load_standalone_example_config() {
         GreptimeOptions::<StandaloneOptions>::load_layered_options(example_config.to_str(), "")
             .unwrap();
     let expected = GreptimeOptions::<StandaloneOptions> {
-        runtime: RuntimeOptions {
-            global_rt_size: 8,
-            compact_rt_size: 4,
-        },
         component: StandaloneOptions {
             default_timezone: Some("UTC".to_string()),
             wal: DatanodeWalConfig::RaftEngine(RaftEngineConfig {
                 dir: Some("/tmp/greptimedb/wal".to_string()),
                 sync_period: Some(Duration::from_secs(10)),
+                recovery_parallelism: 2,
                 ..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),
-                    selector_result_cache_size: ReadableSize::mb(512),
-                    max_background_jobs: 4,
                     experimental_write_cache_ttl: Some(Duration::from_secs(60 * 60 * 8)),
+                    scan_parallelism: 0,
                     ..Default::default()
                 }),
                 RegionEngineConfig::File(EngineConfig {}),
@@ -242,6 +214,7 @@ fn test_load_standalone_example_config() {
             },
             ..Default::default()
         },
+        ..Default::default()
     };
-    assert_eq!(options, expected);
+    similar_asserts::assert_eq!(options, expected);
 }

src/common/base/Cargo.toml

@@ -9,10 +9,12 @@ workspace = true
 
 [dependencies]
 anymap = "1.0.0-beta.2"
+async-trait.workspace = true
 bitvec = "1.0"
 bytes.workspace = true
 common-error.workspace = true
 common-macro.workspace = true
+futures.workspace = true
 paste = "1.0"
 serde = { version = "1.0", features = ["derive"] }
 snafu.workspace = true

src/common/base/src/buffer.rs

@@ -1,242 +0,0 @@
-// Copyright 2023 Greptime Team
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-
-use std::any::Any;
-use std::io::{Read, Write};
-
-use bytes::{Buf, BufMut, BytesMut};
-use common_error::ext::ErrorExt;
-use common_macro::stack_trace_debug;
-use paste::paste;
-use snafu::{ensure, Location, ResultExt, Snafu};
-
-#[derive(Snafu)]
-#[snafu(visibility(pub))]
-#[stack_trace_debug]
-pub enum Error {
-    #[snafu(display(
-        "Destination buffer overflow, src_len: {}, dst_len: {}",
-        src_len,
-        dst_len
-    ))]
-    Overflow {
-        src_len: usize,
-        dst_len: usize,
-        #[snafu(implicit)]
-        location: Location,
-    },
-
-    #[snafu(display("Buffer underflow"))]
-    Underflow {
-        #[snafu(implicit)]
-        location: Location,
-    },
-
-    #[snafu(display("IO operation reach EOF"))]
-    Eof {
-        #[snafu(source)]
-        error: std::io::Error,
-        #[snafu(implicit)]
-        location: Location,
-    },
-}
-
-pub type Result<T> = std::result::Result<T, Error>;
-
-impl ErrorExt for Error {
-    fn as_any(&self) -> &dyn Any {
-        self
-    }
-}
-
-macro_rules! impl_read_le {
-    ( $($num_ty: ty), *) => {
-        $(
-            paste!{
-                // TODO(hl): default implementation requires allocating a
-                // temp buffer. maybe use more efficient impls in concrete buffers.
-                // see https://github.com/GrepTimeTeam/greptimedb/pull/97#discussion_r930798941
-                fn [<read_ $num_ty _le>](&mut self) -> Result<$num_ty> {
-                    let mut buf = [0u8; std::mem::size_of::<$num_ty>()];
-                    self.read_to_slice(&mut buf)?;
-                    Ok($num_ty::from_le_bytes(buf))
-                }
-
-                fn [<peek_ $num_ty _le>](&mut self) -> Result<$num_ty> {
-                    let mut buf = [0u8; std::mem::size_of::<$num_ty>()];
-                    self.peek_to_slice(&mut buf)?;
-                    Ok($num_ty::from_le_bytes(buf))
-                }
-            }
-        )*
-    }
-}
-
-macro_rules! impl_write_le {
-    ( $($num_ty: ty), *) => {
-        $(
-            paste!{
-                fn [<write_ $num_ty _le>](&mut self, n: $num_ty) -> Result<()> {
-                    self.write_from_slice(&n.to_le_bytes())?;
-                    Ok(())
-                }
-            }
-        )*
-    }
-}
-
-pub trait Buffer {
-    /// Returns remaining data size for read.
-    fn remaining_size(&self) -> usize;
-
-    /// Returns true if buffer has no data for read.
-    fn is_empty(&self) -> bool {
-        self.remaining_size() == 0
-    }
-
-    /// Peeks data into dst. This method should not change internal cursor,
-    /// invoke `advance_by` if needed.
-    /// # Panics
-    /// This method **may** panic if buffer does not have enough data to be copied to dst.
-    fn peek_to_slice(&self, dst: &mut [u8]) -> Result<()>;
-
-    /// Reads data into dst. This method will change internal cursor.
-    /// # Panics
-    /// This method **may** panic if buffer does not have enough data to be copied to dst.
-    fn read_to_slice(&mut self, dst: &mut [u8]) -> Result<()> {
-        self.peek_to_slice(dst)?;
-        self.advance_by(dst.len());
-        Ok(())
-    }
-
-    /// Advances internal cursor for next read.
-    /// # Panics
-    /// This method **may** panic if the offset after advancing exceeds the length of underlying buffer.
-    fn advance_by(&mut self, by: usize);
-
-    impl_read_le![u8, i8, u16, i16, u32, i32, u64, i64, f32, f64];
-}
-
-macro_rules! impl_buffer_for_bytes {
-    ( $($buf_ty:ty), *) => {
-        $(
-        impl Buffer for $buf_ty {
-            fn remaining_size(&self) -> usize{
-                self.len()
-            }
-
-            fn peek_to_slice(&self, dst: &mut [u8]) -> Result<()> {
-                let dst_len = dst.len();
-                ensure!(self.remaining() >= dst.len(), OverflowSnafu {
-                        src_len: self.remaining_size(),
-                        dst_len,
-                    }
-                );
-                dst.copy_from_slice(&self[0..dst_len]);
-                Ok(())
-            }
-
-            #[inline]
-            fn advance_by(&mut self, by: usize) {
-                self.advance(by);
-            }
-        }
-        )*
-    };
-}
-
-impl_buffer_for_bytes![bytes::Bytes, bytes::BytesMut];
-
-impl Buffer for &[u8] {
-    fn remaining_size(&self) -> usize {
-        self.len()
-    }
-
-    fn peek_to_slice(&self, dst: &mut [u8]) -> Result<()> {
-        let dst_len = dst.len();
-        ensure!(
-            self.len() >= dst.len(),
-            OverflowSnafu {
-                src_len: self.remaining_size(),
-                dst_len,
-            }
-        );
-        dst.copy_from_slice(&self[0..dst_len]);
-        Ok(())
-    }
-
-    fn read_to_slice(&mut self, dst: &mut [u8]) -> Result<()> {
-        ensure!(
-            self.len() >= dst.len(),
-            OverflowSnafu {
-                src_len: self.remaining_size(),
-                dst_len: dst.len(),
-            }
-        );
-        self.read_exact(dst).context(EofSnafu)
-    }
-
-    fn advance_by(&mut self, by: usize) {
-        *self = &self[by..];
-    }
-}
-
-/// Mutable buffer.
-pub trait BufferMut {
-    fn as_slice(&self) -> &[u8];
-
-    fn write_from_slice(&mut self, src: &[u8]) -> Result<()>;
-
-    impl_write_le![i8, u8, i16, u16, i32, u32, i64, u64, f32, f64];
-}
-
-impl BufferMut for BytesMut {
-    fn as_slice(&self) -> &[u8] {
-        self
-    }
-
-    fn write_from_slice(&mut self, src: &[u8]) -> Result<()> {
-        self.put_slice(src);
-        Ok(())
-    }
-}
-
-impl BufferMut for &mut [u8] {
-    fn as_slice(&self) -> &[u8] {
-        self
-    }
-
-    fn write_from_slice(&mut self, src: &[u8]) -> Result<()> {
-        // see std::io::Write::write_all
-        // https://doc.rust-lang.org/src/std/io/impls.rs.html#363
-        self.write_all(src).map_err(|_| {
-            OverflowSnafu {
-                src_len: src.len(),
-                dst_len: self.as_slice().len(),
-            }
-            .build()
-        })
-    }
-}
-
-impl BufferMut for Vec<u8> {
-    fn as_slice(&self) -> &[u8] {
-        self
-    }
-
-    fn write_from_slice(&mut self, src: &[u8]) -> Result<()> {
-        self.extend_from_slice(src);
-        Ok(())
-    }
-}

src/common/base/src/bytes.rs

@@ -44,6 +44,12 @@ impl From<Vec<u8>> for Bytes {
     }
 }
 
+impl From<Bytes> for Vec<u8> {
+    fn from(bytes: Bytes) -> Vec<u8> {
+        bytes.0.into()
+    }
+}
+
 impl Deref for Bytes {
     type Target = [u8];
 

src/common/base/src/lib.rs

@@ -13,9 +13,9 @@
 // limitations under the License.
 
 pub mod bit_vec;
-pub mod buffer;
 pub mod bytes;
 pub mod plugins;
+pub mod range_read;
 #[allow(clippy::all)]
 pub mod readable_size;
 pub mod secrets;

src/common/base/src/range_read.rs

@@ -0,0 +1,105 @@
+// 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::io;
+use std::ops::Range;
+
+use async_trait::async_trait;
+use bytes::{BufMut, Bytes};
+use futures::{AsyncReadExt, AsyncSeekExt};
+
+/// `Metadata` contains the metadata of a source.
+pub struct Metadata {
+    /// The length of the source in bytes.
+    pub content_length: u64,
+}
+
+/// `RangeReader` reads a range of bytes from a source.
+#[async_trait]
+pub trait RangeReader: Send + Unpin {
+    /// Returns the metadata of the source.
+    async fn metadata(&mut self) -> io::Result<Metadata>;
+
+    /// Reads the bytes in the given range.
+    async fn read(&mut self, range: Range<u64>) -> io::Result<Bytes>;
+
+    /// Reads the bytes in the given range into the buffer.
+    ///
+    /// Handles the buffer based on its capacity:
+    /// - If the buffer is insufficient to hold the bytes, it will either:
+    ///   - Allocate additional space (e.g., for `Vec<u8>`)
+    ///   - Panic (e.g., for `&mut [u8]`)
+    async fn read_into(
+        &mut self,
+        range: Range<u64>,
+        buf: &mut (impl BufMut + Send),
+    ) -> io::Result<()> {
+        let bytes = self.read(range).await?;
+        buf.put_slice(&bytes);
+        Ok(())
+    }
+
+    /// Reads the bytes in the given ranges.
+    async fn read_vec(&mut self, ranges: &[Range<u64>]) -> io::Result<Vec<Bytes>> {
+        let mut result = Vec::with_capacity(ranges.len());
+        for range in ranges {
+            result.push(self.read(range.clone()).await?);
+        }
+        Ok(result)
+    }
+}
+
+#[async_trait]
+impl<R: RangeReader + Send + Unpin> RangeReader for &mut R {
+    async fn metadata(&mut self) -> io::Result<Metadata> {
+        (*self).metadata().await
+    }
+    async fn read(&mut self, range: Range<u64>) -> io::Result<Bytes> {
+        (*self).read(range).await
+    }
+    async fn read_into(
+        &mut self,
+        range: Range<u64>,
+        buf: &mut (impl BufMut + Send),
+    ) -> io::Result<()> {
+        (*self).read_into(range, buf).await
+    }
+    async fn read_vec(&mut self, ranges: &[Range<u64>]) -> io::Result<Vec<Bytes>> {
+        (*self).read_vec(ranges).await
+    }
+}
+
+/// `RangeReaderAdapter` bridges `RangeReader` and `AsyncRead + AsyncSeek`.
+pub struct RangeReaderAdapter<R>(pub R);
+
+/// Implements `RangeReader` for a type that implements `AsyncRead + AsyncSeek`.
+///
+/// TODO(zhongzc): It's a temporary solution for porting the codebase from `AsyncRead + AsyncSeek` to `RangeReader`.
+/// Until the codebase is fully ported to `RangeReader`, remove this implementation.
+#[async_trait]
+impl<R: futures::AsyncRead + futures::AsyncSeek + Send + Unpin> RangeReader
+    for RangeReaderAdapter<R>
+{
+    async fn metadata(&mut self) -> io::Result<Metadata> {
+        let content_length = self.0.seek(io::SeekFrom::End(0)).await?;
+        Ok(Metadata { content_length })
+    }
+
+    async fn read(&mut self, range: Range<u64>) -> io::Result<Bytes> {
+        let mut buf = vec![0; (range.end - range.start) as usize];
+        self.0.seek(io::SeekFrom::Start(range.start)).await?;
+        self.0.read_exact(&mut buf).await?;
+        Ok(Bytes::from(buf))
+    }
+}

src/common/base/tests/buffer_tests.rs

@@ -1,182 +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.
-
-#![feature(assert_matches)]
-
-#[cfg(test)]
-mod tests {
-    use std::assert_matches::assert_matches;
-
-    use bytes::{Buf, Bytes, BytesMut};
-    use common_base::buffer::Error::Overflow;
-    use common_base::buffer::{Buffer, BufferMut};
-    use paste::paste;
-
-    #[test]
-    pub fn test_buffer_read_write() {
-        let mut buf = BytesMut::with_capacity(16);
-        buf.write_u64_le(1234u64).unwrap();
-        let result = buf.peek_u64_le().unwrap();
-        assert_eq!(1234u64, result);
-        buf.advance_by(8);
-
-        buf.write_from_slice("hello, world".as_bytes()).unwrap();
-        let mut content = vec![0u8; 5];
-        buf.peek_to_slice(&mut content).unwrap();
-        let read = String::from_utf8_lossy(&content);
-        assert_eq!("hello", read);
-        buf.advance_by(5);
-        // after read, buffer should still have 7 bytes to read.
-        assert_eq!(7, buf.remaining());
-
-        let mut content = vec![0u8; 6];
-        buf.read_to_slice(&mut content).unwrap();
-        let read = String::from_utf8_lossy(&content);
-        assert_eq!(", worl", read);
-        // after read, buffer should still have 1 byte to read.
-        assert_eq!(1, buf.remaining());
-    }
-
-    #[test]
-    pub fn test_buffer_read() {
-        let mut bytes = Bytes::from_static("hello".as_bytes());
-        assert_eq!(5, bytes.remaining_size());
-        assert_eq!(b'h', bytes.peek_u8_le().unwrap());
-        bytes.advance_by(1);
-        assert_eq!(4, bytes.remaining_size());
-    }
-
-    macro_rules! test_primitive_read_write {
-        ( $($num_ty: ty), *) => {
-            $(
-                paste!{
-                    #[test]
-                    fn [<test_read_write_ $num_ty>]() {
-                        assert_eq!($num_ty::MAX,(&mut $num_ty::MAX.to_le_bytes() as &[u8]).[<read_ $num_ty _le>]().unwrap());
-                        assert_eq!($num_ty::MIN,(&mut $num_ty::MIN.to_le_bytes() as &[u8]).[<read_ $num_ty _le>]().unwrap());
-                    }
-                }
-            )*
-        }
-    }
-
-    test_primitive_read_write![u8, u16, u32, u64, i8, i16, i32, i64, f32, f64];
-
-    #[test]
-    pub fn test_read_write_from_slice_buffer() {
-        let mut buf = "hello".as_bytes();
-        assert_eq!(104, buf.peek_u8_le().unwrap());
-        buf.advance_by(1);
-        assert_eq!(101, buf.peek_u8_le().unwrap());
-        buf.advance_by(1);
-        assert_eq!(108, buf.peek_u8_le().unwrap());
-        buf.advance_by(1);
-        assert_eq!(108, buf.peek_u8_le().unwrap());
-        buf.advance_by(1);
-        assert_eq!(111, buf.peek_u8_le().unwrap());
-        buf.advance_by(1);
-        assert_matches!(buf.peek_u8_le(), Err(Overflow { .. }));
-    }
-
-    #[test]
-    pub fn test_read_u8_from_slice_buffer() {
-        let mut buf = "hello".as_bytes();
-        assert_eq!(104, buf.read_u8_le().unwrap());
-        assert_eq!(101, buf.read_u8_le().unwrap());
-        assert_eq!(108, buf.read_u8_le().unwrap());
-        assert_eq!(108, buf.read_u8_le().unwrap());
-        assert_eq!(111, buf.read_u8_le().unwrap());
-        assert_matches!(buf.read_u8_le(), Err(Overflow { .. }));
-    }
-
-    #[test]
-    pub fn test_read_write_numbers() {
-        let mut buf: Vec<u8> = vec![];
-        buf.write_u64_le(1234).unwrap();
-        assert_eq!(1234, (&buf[..]).read_u64_le().unwrap());
-
-        buf.write_u32_le(4242).unwrap();
-        let mut p = &buf[..];
-        assert_eq!(1234, p.read_u64_le().unwrap());
-        assert_eq!(4242, p.read_u32_le().unwrap());
-    }
-
-    macro_rules! test_primitive_vec_read_write {
-        ( $($num_ty: ty), *) => {
-            $(
-                paste!{
-                    #[test]
-                    fn [<test_read_write_ $num_ty _from_vec_buffer>]() {
-                        let mut buf = vec![];
-                        let _ = buf.[<write_ $num_ty _le>]($num_ty::MAX).unwrap();
-                        assert_eq!($num_ty::MAX, buf.as_slice().[<read_ $num_ty _le>]().unwrap());
-                    }
-                }
-            )*
-        }
-    }
-
-    test_primitive_vec_read_write![u8, u16, u32, u64, i8, i16, i32, i64, f32, f64];
-
-    #[test]
-    pub fn test_peek_write_from_vec_buffer() {
-        let mut buf: Vec<u8> = vec![];
-        buf.write_from_slice("hello".as_bytes()).unwrap();
-        let mut slice = buf.as_slice();
-        assert_eq!(104, slice.peek_u8_le().unwrap());
-        slice.advance_by(1);
-        assert_eq!(101, slice.peek_u8_le().unwrap());
-        slice.advance_by(1);
-        assert_eq!(108, slice.peek_u8_le().unwrap());
-        slice.advance_by(1);
-        assert_eq!(108, slice.peek_u8_le().unwrap());
-        slice.advance_by(1);
-        assert_eq!(111, slice.peek_u8_le().unwrap());
-        slice.advance_by(1);
-        assert_matches!(slice.read_u8_le(), Err(Overflow { .. }));
-    }
-
-    macro_rules! test_primitive_bytes_read_write {
-        ( $($num_ty: ty), *) => {
-            $(
-                paste!{
-                    #[test]
-                    fn [<test_read_write_ $num_ty _from_bytes>]() {
-                        let mut bytes = bytes::Bytes::from($num_ty::MAX.to_le_bytes().to_vec());
-                        assert_eq!($num_ty::MAX, bytes.[<read_ $num_ty _le>]().unwrap());
-
-                        let mut bytes = bytes::Bytes::from($num_ty::MIN.to_le_bytes().to_vec());
-                        assert_eq!($num_ty::MIN, bytes.[<read_ $num_ty _le>]().unwrap());
-                    }
-                }
-            )*
-        }
-    }
-
-    test_primitive_bytes_read_write![u8, u16, u32, u64, i8, i16, i32, i64, f32, f64];
-
-    #[test]
-    pub fn test_write_overflow() {
-        let mut buf = [0u8; 4];
-        assert_matches!(
-            (&mut buf[..]).write_from_slice("hell".as_bytes()),
-            Ok { .. }
-        );
-
-        assert_matches!(
-            (&mut buf[..]).write_from_slice("hello".as_bytes()),
-            Err(common_base::buffer::Error::Overflow { .. })
-        );
-    }
-}

src/common/catalog/src/error.rs

@@ -1,46 +0,0 @@
-// Copyright 2023 Greptime Team
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-
-use std::any::Any;
-
-use common_error::ext::ErrorExt;
-use common_error::status_code::StatusCode;
-use common_macro::stack_trace_debug;
-use snafu::{Location, Snafu};
-
-#[derive(Snafu)]
-#[snafu(visibility(pub))]
-#[stack_trace_debug]
-pub enum Error {
-    #[snafu(display("Invalid full table name: {}", table_name))]
-    InvalidFullTableName {
-        table_name: String,
-        #[snafu(implicit)]
-        location: Location,
-    },
-}
-
-impl ErrorExt for Error {
-    fn status_code(&self) -> StatusCode {
-        match self {
-            Error::InvalidFullTableName { .. } => StatusCode::Unexpected,
-        }
-    }
-
-    fn as_any(&self) -> &dyn Any {
-        self
-    }
-}
-
-pub type Result<T> = std::result::Result<T, Error>;

src/common/catalog/src/lib.rs

@@ -15,7 +15,6 @@
 use consts::DEFAULT_CATALOG_NAME;
 
 pub mod consts;
-pub mod error;
 
 #[inline]
 pub fn format_schema_name(catalog: &str, schema: &str) -> String {

src/common/function/Cargo.toml

@@ -7,6 +7,10 @@ license.workspace = true
 [lints]
 workspace = true
 
+[features]
+default = ["geo"]
+geo = ["geohash", "h3o"]
+
 [dependencies]
 api.workspace = true
 arc-swap = "1.0"
@@ -23,6 +27,9 @@ common-time.workspace = true
 common-version.workspace = true
 datafusion.workspace = true
 datatypes.workspace = true
+geohash = { version = "0.13", optional = true }
+h3o = { version = "0.6", optional = true }
+jsonb.workspace = true
 num = "0.4"
 num-traits = "0.2"
 once_cell.workspace = true

src/common/function/src/function_registry.rs

@@ -22,6 +22,7 @@ use crate::function::{AsyncFunctionRef, FunctionRef};
 use crate::scalars::aggregate::{AggregateFunctionMetaRef, AggregateFunctions};
 use crate::scalars::date::DateFunction;
 use crate::scalars::expression::ExpressionFunction;
+use crate::scalars::json::JsonFunction;
 use crate::scalars::matches::MatchesFunction;
 use crate::scalars::math::MathFunction;
 use crate::scalars::numpy::NumpyFunction;
@@ -116,6 +117,13 @@ pub static FUNCTION_REGISTRY: Lazy<Arc<FunctionRegistry>> = Lazy::new(|| {
     SystemFunction::register(&function_registry);
     TableFunction::register(&function_registry);
 
+    // Json related functions
+    JsonFunction::register(&function_registry);
+
+    // Geo functions
+    #[cfg(feature = "geo")]
+    crate::scalars::geo::GeoFunctions::register(&function_registry);
+
     Arc::new(function_registry)
 });
 

src/common/function/src/scalars.rs

@@ -15,9 +15,13 @@
 pub mod aggregate;
 pub(crate) mod date;
 pub mod expression;
+#[cfg(feature = "geo")]
+pub mod geo;
+pub mod json;
 pub mod matches;
 pub mod math;
 pub mod numpy;
+
 #[cfg(test)]
 pub(crate) mod test;
 pub(crate) mod timestamp;

src/common/function/src/scalars/geo.rs

@@ -0,0 +1,31 @@
+// 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;
+mod geohash;
+mod h3;
+
+use geohash::GeohashFunction;
+use h3::H3Function;
+
+use crate::function_registry::FunctionRegistry;
+
+pub(crate) struct GeoFunctions;
+
+impl GeoFunctions {
+    pub fn register(registry: &FunctionRegistry) {
+        registry.register(Arc::new(GeohashFunction));
+        registry.register(Arc::new(H3Function));
+    }
+}

src/common/function/src/scalars/geo/geohash.rs

@@ -0,0 +1,135 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use std::fmt;
+
+use common_error::ext::{BoxedError, PlainError};
+use common_error::status_code::StatusCode;
+use common_query::error::{self, InvalidFuncArgsSnafu, Result};
+use common_query::prelude::{Signature, TypeSignature};
+use datafusion::logical_expr::Volatility;
+use datatypes::prelude::ConcreteDataType;
+use datatypes::scalars::ScalarVectorBuilder;
+use datatypes::value::Value;
+use datatypes::vectors::{MutableVector, StringVectorBuilder, VectorRef};
+use geohash::Coord;
+use snafu::{ensure, ResultExt};
+
+use crate::function::{Function, FunctionContext};
+
+/// Function that return geohash string for a given geospatial coordinate.
+#[derive(Clone, Debug, Default)]
+pub struct GeohashFunction;
+
+const NAME: &str = "geohash";
+
+impl Function for GeohashFunction {
+    fn name(&self) -> &str {
+        NAME
+    }
+
+    fn return_type(&self, _input_types: &[ConcreteDataType]) -> Result<ConcreteDataType> {
+        Ok(ConcreteDataType::string_datatype())
+    }
+
+    fn signature(&self) -> Signature {
+        let mut signatures = Vec::new();
+        for coord_type in &[
+            ConcreteDataType::float32_datatype(),
+            ConcreteDataType::float64_datatype(),
+        ] {
+            for resolution_type in &[
+                ConcreteDataType::int8_datatype(),
+                ConcreteDataType::int16_datatype(),
+                ConcreteDataType::int32_datatype(),
+                ConcreteDataType::int64_datatype(),
+                ConcreteDataType::uint8_datatype(),
+                ConcreteDataType::uint16_datatype(),
+                ConcreteDataType::uint32_datatype(),
+                ConcreteDataType::uint64_datatype(),
+            ] {
+                signatures.push(TypeSignature::Exact(vec![
+                    // latitude
+                    coord_type.clone(),
+                    // longitude
+                    coord_type.clone(),
+                    // resolution
+                    resolution_type.clone(),
+                ]));
+            }
+        }
+        Signature::one_of(signatures, Volatility::Stable)
+    }
+
+    fn eval(&self, _func_ctx: FunctionContext, columns: &[VectorRef]) -> Result<VectorRef> {
+        ensure!(
+            columns.len() == 3,
+            InvalidFuncArgsSnafu {
+                err_msg: format!(
+                    "The length of the args is not correct, expect 3, provided : {}",
+                    columns.len()
+                ),
+            }
+        );
+
+        let lat_vec = &columns[0];
+        let lon_vec = &columns[1];
+        let resolution_vec = &columns[2];
+
+        let size = lat_vec.len();
+        let mut results = StringVectorBuilder::with_capacity(size);
+
+        for i in 0..size {
+            let lat = lat_vec.get(i).as_f64_lossy();
+            let lon = lon_vec.get(i).as_f64_lossy();
+            let r = match resolution_vec.get(i) {
+                Value::Int8(v) => v as usize,
+                Value::Int16(v) => v as usize,
+                Value::Int32(v) => v as usize,
+                Value::Int64(v) => v as usize,
+                Value::UInt8(v) => v as usize,
+                Value::UInt16(v) => v as usize,
+                Value::UInt32(v) => v as usize,
+                Value::UInt64(v) => v as usize,
+                _ => unreachable!(),
+            };
+
+            let result = match (lat, lon) {
+                (Some(lat), Some(lon)) => {
+                    let coord = Coord { x: lon, y: lat };
+                    let encoded = geohash::encode(coord, r)
+                        .map_err(|e| {
+                            BoxedError::new(PlainError::new(
+                                format!("Geohash error: {}", e),
+                                StatusCode::EngineExecuteQuery,
+                            ))
+                        })
+                        .context(error::ExecuteSnafu)?;
+                    Some(encoded)
+                }
+                _ => None,
+            };
+
+            results.push(result.as_deref());
+        }
+
+        Ok(results.to_vector())
+    }
+}
+
+impl fmt::Display for GeohashFunction {
+    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+        write!(f, "{}", NAME)
+    }
+}

src/common/function/src/scalars/geo/h3.rs

@@ -0,0 +1,145 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use std::fmt;
+
+use common_error::ext::{BoxedError, PlainError};
+use common_error::status_code::StatusCode;
+use common_query::error::{self, InvalidFuncArgsSnafu, Result};
+use common_query::prelude::{Signature, TypeSignature};
+use datafusion::logical_expr::Volatility;
+use datatypes::prelude::ConcreteDataType;
+use datatypes::scalars::ScalarVectorBuilder;
+use datatypes::value::Value;
+use datatypes::vectors::{MutableVector, StringVectorBuilder, VectorRef};
+use h3o::{LatLng, Resolution};
+use snafu::{ensure, ResultExt};
+
+use crate::function::{Function, FunctionContext};
+
+/// Function that returns [h3] encoding string for a given geospatial coordinate.
+///
+/// [h3]: https://h3geo.org/
+#[derive(Clone, Debug, Default)]
+pub struct H3Function;
+
+const NAME: &str = "h3";
+
+impl Function for H3Function {
+    fn name(&self) -> &str {
+        NAME
+    }
+
+    fn return_type(&self, _input_types: &[ConcreteDataType]) -> Result<ConcreteDataType> {
+        Ok(ConcreteDataType::string_datatype())
+    }
+
+    fn signature(&self) -> Signature {
+        let mut signatures = Vec::new();
+        for coord_type in &[
+            ConcreteDataType::float32_datatype(),
+            ConcreteDataType::float64_datatype(),
+        ] {
+            for resolution_type in &[
+                ConcreteDataType::int8_datatype(),
+                ConcreteDataType::int16_datatype(),
+                ConcreteDataType::int32_datatype(),
+                ConcreteDataType::int64_datatype(),
+                ConcreteDataType::uint8_datatype(),
+                ConcreteDataType::uint16_datatype(),
+                ConcreteDataType::uint32_datatype(),
+                ConcreteDataType::uint64_datatype(),
+            ] {
+                signatures.push(TypeSignature::Exact(vec![
+                    // latitude
+                    coord_type.clone(),
+                    // longitude
+                    coord_type.clone(),
+                    // resolution
+                    resolution_type.clone(),
+                ]));
+            }
+        }
+        Signature::one_of(signatures, Volatility::Stable)
+    }
+
+    fn eval(&self, _func_ctx: FunctionContext, columns: &[VectorRef]) -> Result<VectorRef> {
+        ensure!(
+            columns.len() == 3,
+            InvalidFuncArgsSnafu {
+                err_msg: format!(
+                    "The length of the args is not correct, expect 3, provided : {}",
+                    columns.len()
+                ),
+            }
+        );
+
+        let lat_vec = &columns[0];
+        let lon_vec = &columns[1];
+        let resolution_vec = &columns[2];
+
+        let size = lat_vec.len();
+        let mut results = StringVectorBuilder::with_capacity(size);
+
+        for i in 0..size {
+            let lat = lat_vec.get(i).as_f64_lossy();
+            let lon = lon_vec.get(i).as_f64_lossy();
+            let r = match resolution_vec.get(i) {
+                Value::Int8(v) => v as u8,
+                Value::Int16(v) => v as u8,
+                Value::Int32(v) => v as u8,
+                Value::Int64(v) => v as u8,
+                Value::UInt8(v) => v,
+                Value::UInt16(v) => v as u8,
+                Value::UInt32(v) => v as u8,
+                Value::UInt64(v) => v as u8,
+                _ => unreachable!(),
+            };
+
+            let result = match (lat, lon) {
+                (Some(lat), Some(lon)) => {
+                    let coord = LatLng::new(lat, lon)
+                        .map_err(|e| {
+                            BoxedError::new(PlainError::new(
+                                format!("H3 error: {}", e),
+                                StatusCode::EngineExecuteQuery,
+                            ))
+                        })
+                        .context(error::ExecuteSnafu)?;
+                    let r = Resolution::try_from(r)
+                        .map_err(|e| {
+                            BoxedError::new(PlainError::new(
+                                format!("H3 error: {}", e),
+                                StatusCode::EngineExecuteQuery,
+                            ))
+                        })
+                        .context(error::ExecuteSnafu)?;
+                    let encoded = coord.to_cell(r).to_string();
+                    Some(encoded)
+                }
+                _ => None,
+            };
+
+            results.push(result.as_deref());
+        }
+
+        Ok(results.to_vector())
+    }
+}
+
+impl fmt::Display for H3Function {
+    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+        write!(f, "{}", NAME)
+    }
+}

src/common/function/src/scalars/json.rs

@@ -0,0 +1,50 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use std::sync::Arc;
+mod json_get;
+mod json_is;
+mod json_to_string;
+mod parse_json;
+
+use json_get::{JsonGetBool, JsonGetFloat, JsonGetInt, JsonGetString};
+use json_is::{
+    JsonIsArray, JsonIsBool, JsonIsFloat, JsonIsInt, JsonIsNull, JsonIsObject, JsonIsString,
+};
+use json_to_string::JsonToStringFunction;
+use parse_json::ParseJsonFunction;
+
+use crate::function_registry::FunctionRegistry;
+
+pub(crate) struct JsonFunction;
+
+impl JsonFunction {
+    pub fn register(registry: &FunctionRegistry) {
+        registry.register(Arc::new(JsonToStringFunction));
+        registry.register(Arc::new(ParseJsonFunction));
+
+        registry.register(Arc::new(JsonGetInt));
+        registry.register(Arc::new(JsonGetFloat));
+        registry.register(Arc::new(JsonGetString));
+        registry.register(Arc::new(JsonGetBool));
+
+        registry.register(Arc::new(JsonIsNull));
+        registry.register(Arc::new(JsonIsInt));
+        registry.register(Arc::new(JsonIsFloat));
+        registry.register(Arc::new(JsonIsString));
+        registry.register(Arc::new(JsonIsBool));
+        registry.register(Arc::new(JsonIsArray));
+        registry.register(Arc::new(JsonIsObject));
+    }
+}

src/common/function/src/scalars/json/json_get.rs

@@ -0,0 +1,454 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use std::fmt::{self, Display};
+
+use common_query::error::{InvalidFuncArgsSnafu, Result, UnsupportedInputDataTypeSnafu};
+use common_query::prelude::Signature;
+use datafusion::logical_expr::Volatility;
+use datatypes::data_type::ConcreteDataType;
+use datatypes::prelude::VectorRef;
+use datatypes::scalars::ScalarVectorBuilder;
+use datatypes::vectors::{
+    BooleanVectorBuilder, Float64VectorBuilder, Int64VectorBuilder, MutableVector,
+    StringVectorBuilder,
+};
+use snafu::ensure;
+
+use crate::function::{Function, FunctionContext};
+
+fn get_json_by_path(json: &[u8], path: &str) -> Option<Vec<u8>> {
+    let json_path = jsonb::jsonpath::parse_json_path(path.as_bytes());
+    match json_path {
+        Ok(json_path) => {
+            let mut sub_jsonb = Vec::new();
+            let mut sub_offsets = Vec::new();
+            match jsonb::get_by_path(json, json_path, &mut sub_jsonb, &mut sub_offsets) {
+                Ok(_) => Some(sub_jsonb),
+                Err(_) => None,
+            }
+        }
+        _ => None,
+    }
+}
+
+/// Get the value from the JSONB by the given path and return it as specified type.
+/// If the path does not exist or the value is not the type specified, return `NULL`.
+macro_rules! json_get {
+    // e.g. name = JsonGetInt, type = Int64, rust_type = i64, doc = "Get the value from the JSONB by the given path and return it as an integer."
+    ($name:ident, $type:ident, $rust_type:ident, $doc:expr) => {
+        paste::paste! {
+            #[doc = $doc]
+            #[derive(Clone, Debug, Default)]
+            pub struct $name;
+
+            impl Function for $name {
+                fn name(&self) -> &str {
+                    stringify!([<$name:snake>])
+                }
+
+                fn return_type(&self, _input_types: &[ConcreteDataType]) -> Result<ConcreteDataType> {
+                    Ok(ConcreteDataType::[<$type:snake _datatype>]())
+                }
+
+                fn signature(&self) -> Signature {
+                    Signature::exact(
+                        vec![
+                            ConcreteDataType::json_datatype(),
+                            ConcreteDataType::string_datatype(),
+                        ],
+                        Volatility::Immutable,
+                    )
+                }
+
+                fn eval(&self, _func_ctx: FunctionContext, columns: &[VectorRef]) -> Result<VectorRef> {
+                    ensure!(
+                        columns.len() == 2,
+                        InvalidFuncArgsSnafu {
+                            err_msg: format!(
+                                "The length of the args is not correct, expect exactly two, have: {}",
+                                columns.len()
+                            ),
+                        }
+                    );
+                    let jsons = &columns[0];
+                    let paths = &columns[1];
+
+                    let size = jsons.len();
+                    let datatype = jsons.data_type();
+                    let mut results = [<$type VectorBuilder>]::with_capacity(size);
+
+                    match datatype {
+                        // JSON data type uses binary vector
+                        ConcreteDataType::Binary(_) => {
+                            for i in 0..size {
+                                let json = jsons.get_ref(i);
+                                let path = paths.get_ref(i);
+
+                                let json = json.as_binary();
+                                let path = path.as_string();
+                                let result = match (json, path) {
+                                    (Ok(Some(json)), Ok(Some(path))) => {
+                                        get_json_by_path(json, path)
+                                            .and_then(|json| { jsonb::[<to_ $rust_type>](&json).ok() })
+                                    }
+                                    _ => None,
+                                };
+
+                                results.push(result);
+                            }
+                        }
+                        _ => {
+                            return UnsupportedInputDataTypeSnafu {
+                                function: stringify!([<$name:snake>]),
+                                datatypes: columns.iter().map(|c| c.data_type()).collect::<Vec<_>>(),
+                            }
+                            .fail();
+                        }
+                    }
+
+                    Ok(results.to_vector())
+                }
+            }
+
+            impl Display for $name {
+                fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+                    write!(f, "{}", stringify!([<$name:snake>]).to_ascii_uppercase())
+                }
+            }
+        }
+    };
+}
+
+json_get!(
+    JsonGetInt,
+    Int64,
+    i64,
+    "Get the value from the JSONB by the given path and return it as an integer."
+);
+
+json_get!(
+    JsonGetFloat,
+    Float64,
+    f64,
+    "Get the value from the JSONB by the given path and return it as a float."
+);
+
+json_get!(
+    JsonGetBool,
+    Boolean,
+    bool,
+    "Get the value from the JSONB by the given path and return it as a boolean."
+);
+
+/// Get the value from the JSONB by the given path and return it as a string.
+#[derive(Clone, Debug, Default)]
+pub struct JsonGetString;
+
+impl Function for JsonGetString {
+    fn name(&self) -> &str {
+        "json_get_string"
+    }
+
+    fn return_type(&self, _input_types: &[ConcreteDataType]) -> Result<ConcreteDataType> {
+        Ok(ConcreteDataType::string_datatype())
+    }
+
+    fn signature(&self) -> Signature {
+        Signature::exact(
+            vec![
+                ConcreteDataType::json_datatype(),
+                ConcreteDataType::string_datatype(),
+            ],
+            Volatility::Immutable,
+        )
+    }
+
+    fn eval(&self, _func_ctx: FunctionContext, columns: &[VectorRef]) -> Result<VectorRef> {
+        ensure!(
+            columns.len() == 2,
+            InvalidFuncArgsSnafu {
+                err_msg: format!(
+                    "The length of the args is not correct, expect exactly two, have: {}",
+                    columns.len()
+                ),
+            }
+        );
+        let jsons = &columns[0];
+        let paths = &columns[1];
+
+        let size = jsons.len();
+        let datatype = jsons.data_type();
+        let mut results = StringVectorBuilder::with_capacity(size);
+
+        match datatype {
+            // JSON data type uses binary vector
+            ConcreteDataType::Binary(_) => {
+                for i in 0..size {
+                    let json = jsons.get_ref(i);
+                    let path = paths.get_ref(i);
+
+                    let json = json.as_binary();
+                    let path = path.as_string();
+                    let result = match (json, path) {
+                        (Ok(Some(json)), Ok(Some(path))) => {
+                            get_json_by_path(json, path).and_then(|json| jsonb::to_str(&json).ok())
+                        }
+                        _ => None,
+                    };
+
+                    results.push(result.as_deref());
+                }
+            }
+            _ => {
+                return UnsupportedInputDataTypeSnafu {
+                    function: "json_get_string",
+                    datatypes: columns.iter().map(|c| c.data_type()).collect::<Vec<_>>(),
+                }
+                .fail();
+            }
+        }
+
+        Ok(results.to_vector())
+    }
+}
+
+impl Display for JsonGetString {
+    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+        write!(f, "{}", "json_get_string".to_ascii_uppercase())
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use std::sync::Arc;
+
+    use common_query::prelude::TypeSignature;
+    use datatypes::scalars::ScalarVector;
+    use datatypes::vectors::{BinaryVector, StringVector};
+
+    use super::*;
+
+    #[test]
+    fn test_json_get_int() {
+        let json_get_int = JsonGetInt;
+
+        assert_eq!("json_get_int", json_get_int.name());
+        assert_eq!(
+            ConcreteDataType::int64_datatype(),
+            json_get_int
+                .return_type(&[
+                    ConcreteDataType::json_datatype(),
+                    ConcreteDataType::string_datatype()
+                ])
+                .unwrap()
+        );
+
+        assert!(matches!(json_get_int.signature(),
+                         Signature {
+                             type_signature: TypeSignature::Exact(valid_types),
+                             volatility: Volatility::Immutable
+                         } if  valid_types == vec![ConcreteDataType::json_datatype(), ConcreteDataType::string_datatype()]
+        ));
+
+        let json_strings = [
+            r#"{"a": {"b": 2}, "b": 2, "c": 3}"#,
+            r#"{"a": 4, "b": {"c": 6}, "c": 6}"#,
+            r#"{"a": 7, "b": 8, "c": {"a": 7}}"#,
+        ];
+        let paths = vec!["$.a.b", "$.a", "$.c"];
+        let results = [Some(2), Some(4), None];
+
+        let jsonbs = json_strings
+            .iter()
+            .map(|s| {
+                let value = jsonb::parse_value(s.as_bytes()).unwrap();
+                value.to_vec()
+            })
+            .collect::<Vec<_>>();
+
+        let json_vector = BinaryVector::from_vec(jsonbs);
+        let path_vector = StringVector::from_vec(paths);
+        let args: Vec<VectorRef> = vec![Arc::new(json_vector), Arc::new(path_vector)];
+        let vector = json_get_int
+            .eval(FunctionContext::default(), &args)
+            .unwrap();
+
+        assert_eq!(3, vector.len());
+        for (i, gt) in results.iter().enumerate() {
+            let result = vector.get_ref(i);
+            let result = result.as_i64().unwrap();
+            assert_eq!(*gt, result);
+        }
+    }
+
+    #[test]
+    fn test_json_get_float() {
+        let json_get_float = JsonGetFloat;
+
+        assert_eq!("json_get_float", json_get_float.name());
+        assert_eq!(
+            ConcreteDataType::float64_datatype(),
+            json_get_float
+                .return_type(&[
+                    ConcreteDataType::json_datatype(),
+                    ConcreteDataType::string_datatype()
+                ])
+                .unwrap()
+        );
+
+        assert!(matches!(json_get_float.signature(),
+                         Signature {
+                             type_signature: TypeSignature::Exact(valid_types),
+                             volatility: Volatility::Immutable
+                         } if  valid_types == vec![ConcreteDataType::json_datatype(), ConcreteDataType::string_datatype()]
+        ));
+
+        let json_strings = [
+            r#"{"a": {"b": 2.1}, "b": 2.2, "c": 3.3}"#,
+            r#"{"a": 4.4, "b": {"c": 6.6}, "c": 6.6}"#,
+            r#"{"a": 7.7, "b": 8.8, "c": {"a": 7.7}}"#,
+        ];
+        let paths = vec!["$.a.b", "$.a", "$.c"];
+        let results = [Some(2.1), Some(4.4), None];
+
+        let jsonbs = json_strings
+            .iter()
+            .map(|s| {
+                let value = jsonb::parse_value(s.as_bytes()).unwrap();
+                value.to_vec()
+            })
+            .collect::<Vec<_>>();
+
+        let json_vector = BinaryVector::from_vec(jsonbs);
+        let path_vector = StringVector::from_vec(paths);
+        let args: Vec<VectorRef> = vec![Arc::new(json_vector), Arc::new(path_vector)];
+        let vector = json_get_float
+            .eval(FunctionContext::default(), &args)
+            .unwrap();
+
+        assert_eq!(3, vector.len());
+        for (i, gt) in results.iter().enumerate() {
+            let result = vector.get_ref(i);
+            let result = result.as_f64().unwrap();
+            assert_eq!(*gt, result);
+        }
+    }
+
+    #[test]
+    fn test_json_get_bool() {
+        let json_get_bool = JsonGetBool;
+
+        assert_eq!("json_get_bool", json_get_bool.name());
+        assert_eq!(
+            ConcreteDataType::boolean_datatype(),
+            json_get_bool
+                .return_type(&[
+                    ConcreteDataType::json_datatype(),
+                    ConcreteDataType::string_datatype()
+                ])
+                .unwrap()
+        );
+
+        assert!(matches!(json_get_bool.signature(),
+                         Signature {
+                             type_signature: TypeSignature::Exact(valid_types),
+                             volatility: Volatility::Immutable
+                         } if  valid_types == vec![ConcreteDataType::json_datatype(), ConcreteDataType::string_datatype()]
+        ));
+
+        let json_strings = [
+            r#"{"a": {"b": true}, "b": false, "c": true}"#,
+            r#"{"a": false, "b": {"c": true}, "c": false}"#,
+            r#"{"a": true, "b": false, "c": {"a": true}}"#,
+        ];
+        let paths = vec!["$.a.b", "$.a", "$.c"];
+        let results = [Some(true), Some(false), None];
+
+        let jsonbs = json_strings
+            .iter()
+            .map(|s| {
+                let value = jsonb::parse_value(s.as_bytes()).unwrap();
+                value.to_vec()
+            })
+            .collect::<Vec<_>>();
+
+        let json_vector = BinaryVector::from_vec(jsonbs);
+        let path_vector = StringVector::from_vec(paths);
+        let args: Vec<VectorRef> = vec![Arc::new(json_vector), Arc::new(path_vector)];
+        let vector = json_get_bool
+            .eval(FunctionContext::default(), &args)
+            .unwrap();
+
+        assert_eq!(3, vector.len());
+        for (i, gt) in results.iter().enumerate() {
+            let result = vector.get_ref(i);
+            let result = result.as_boolean().unwrap();
+            assert_eq!(*gt, result);
+        }
+    }
+
+    #[test]
+    fn test_json_get_string() {
+        let json_get_string = JsonGetString;
+
+        assert_eq!("json_get_string", json_get_string.name());
+        assert_eq!(
+            ConcreteDataType::string_datatype(),
+            json_get_string
+                .return_type(&[
+                    ConcreteDataType::json_datatype(),
+                    ConcreteDataType::string_datatype()
+                ])
+                .unwrap()
+        );
+
+        assert!(matches!(json_get_string.signature(),
+                         Signature {
+                             type_signature: TypeSignature::Exact(valid_types),
+                             volatility: Volatility::Immutable
+                         } if  valid_types == vec![ConcreteDataType::json_datatype(), ConcreteDataType::string_datatype()]
+        ));
+
+        let json_strings = [
+            r#"{"a": {"b": "a"}, "b": "b", "c": "c"}"#,
+            r#"{"a": "d", "b": {"c": "e"}, "c": "f"}"#,
+            r#"{"a": "g", "b": "h", "c": {"a": "g"}}"#,
+        ];
+        let paths = vec!["$.a.b", "$.a", ""];
+        let results = [Some("a"), Some("d"), None];
+
+        let jsonbs = json_strings
+            .iter()
+            .map(|s| {
+                let value = jsonb::parse_value(s.as_bytes()).unwrap();
+                value.to_vec()
+            })
+            .collect::<Vec<_>>();
+
+        let json_vector = BinaryVector::from_vec(jsonbs);
+        let path_vector = StringVector::from_vec(paths);
+        let args: Vec<VectorRef> = vec![Arc::new(json_vector), Arc::new(path_vector)];
+        let vector = json_get_string
+            .eval(FunctionContext::default(), &args)
+            .unwrap();
+
+        assert_eq!(3, vector.len());
+        for (i, gt) in results.iter().enumerate() {
+            let result = vector.get_ref(i);
+            let result = result.as_string().unwrap();
+            assert_eq!(*gt, result);
+        }
+    }
+}

src/common/function/src/scalars/json/json_is.rs

@@ -0,0 +1,215 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use std::fmt::{self, Display};
+
+use common_query::error::{InvalidFuncArgsSnafu, Result, UnsupportedInputDataTypeSnafu};
+use common_query::prelude::Signature;
+use datafusion::logical_expr::Volatility;
+use datatypes::data_type::ConcreteDataType;
+use datatypes::prelude::VectorRef;
+use datatypes::scalars::ScalarVectorBuilder;
+use datatypes::vectors::{BooleanVectorBuilder, MutableVector};
+use snafu::ensure;
+
+use crate::function::{Function, FunctionContext};
+
+/// Checks if the input is a JSON object of the given type.
+macro_rules! json_is {
+    ($name:ident, $json_type:ident, $doc:expr) => {
+        paste::paste! {
+            #[derive(Clone, Debug, Default)]
+            pub struct $name;
+
+            impl Function for $name {
+                fn name(&self) -> &str {
+                    stringify!([<$name:snake>])
+                }
+
+                fn return_type(&self, _input_types: &[ConcreteDataType]) -> Result<ConcreteDataType> {
+                    Ok(ConcreteDataType::boolean_datatype())
+                }
+
+                fn signature(&self) -> Signature {
+                    Signature::exact(vec![ConcreteDataType::json_datatype()], Volatility::Immutable)
+                }
+
+                fn eval(&self, _func_ctx: FunctionContext, columns: &[VectorRef]) -> Result<VectorRef> {
+                    ensure!(
+                        columns.len() == 1,
+                        InvalidFuncArgsSnafu {
+                            err_msg: format!(
+                                "The length of the args is not correct, expect exactly one, have: {}",
+                                columns.len()
+                            ),
+                        }
+                    );
+
+                    let jsons = &columns[0];
+                    let size = jsons.len();
+                    let datatype = jsons.data_type();
+                    let mut results = BooleanVectorBuilder::with_capacity(size);
+
+                    match datatype {
+                        // JSON data type uses binary vector
+                        ConcreteDataType::Binary(_) => {
+                            for i in 0..size {
+                                let json = jsons.get_ref(i);
+                                let json = json.as_binary();
+                                let result = match json {
+                                    Ok(Some(json)) => {
+                                        Some(jsonb::[<is_ $json_type>](json))
+                                    }
+                                    _ => None,
+                                };
+                                results.push(result);
+                            }
+                        }
+                        _ => {
+                            return UnsupportedInputDataTypeSnafu {
+                                function: stringify!([<$name:snake>]),
+                                datatypes: columns.iter().map(|c| c.data_type()).collect::<Vec<_>>(),
+                            }
+                            .fail();
+                        }
+                    }
+
+                    Ok(results.to_vector())
+                }
+            }
+
+            impl Display for $name {
+                fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+                    write!(f, "{}", stringify!([<$name:snake>]).to_ascii_uppercase())
+                }
+            }
+        }
+    }
+}
+
+json_is!(JsonIsNull, null, "Checks if the input JSONB is null");
+json_is!(
+    JsonIsBool,
+    boolean,
+    "Checks if the input JSONB is a boolean type JSON value"
+);
+json_is!(
+    JsonIsInt,
+    i64,
+    "Checks if the input JSONB is a integer type JSON value"
+);
+json_is!(
+    JsonIsFloat,
+    number,
+    "Checks if the input JSONB is a JSON float"
+);
+json_is!(
+    JsonIsString,
+    string,
+    "Checks if the input JSONB is a JSON string"
+);
+json_is!(
+    JsonIsArray,
+    array,
+    "Checks if the input JSONB is a JSON array"
+);
+json_is!(
+    JsonIsObject,
+    object,
+    "Checks if the input JSONB is a JSON object"
+);
+
+#[cfg(test)]
+mod tests {
+    use std::sync::Arc;
+
+    use datatypes::scalars::ScalarVector;
+    use datatypes::vectors::BinaryVector;
+
+    use super::*;
+
+    #[test]
+    fn test_json_is_functions() {
+        let json_is_functions: [&dyn Function; 6] = [
+            &JsonIsBool,
+            &JsonIsInt,
+            &JsonIsFloat,
+            &JsonIsString,
+            &JsonIsArray,
+            &JsonIsObject,
+        ];
+        let expected_names = [
+            "json_is_bool",
+            "json_is_int",
+            "json_is_float",
+            "json_is_string",
+            "json_is_array",
+            "json_is_object",
+        ];
+        for (func, expected_name) in json_is_functions.iter().zip(expected_names.iter()) {
+            assert_eq!(func.name(), *expected_name);
+            assert_eq!(
+                func.return_type(&[ConcreteDataType::json_datatype()])
+                    .unwrap(),
+                ConcreteDataType::boolean_datatype()
+            );
+            assert_eq!(
+                func.signature(),
+                Signature::exact(
+                    vec![ConcreteDataType::json_datatype()],
+                    Volatility::Immutable
+                )
+            );
+        }
+
+        let json_strings = [
+            r#"true"#,
+            r#"1"#,
+            r#"1.0"#,
+            r#""The pig fly through a castle, and has been attracted by the princess.""#,
+            r#"[1, 2]"#,
+            r#"{"a": 1}"#,
+        ];
+        let expected_results = [
+            [true, false, false, false, false, false],
+            [false, true, false, false, false, false],
+            // Integers are also floats
+            [false, true, true, false, false, false],
+            [false, false, false, true, false, false],
+            [false, false, false, false, true, false],
+            [false, false, false, false, false, true],
+        ];
+
+        let jsonbs = json_strings
+            .iter()
+            .map(|s| {
+                let value = jsonb::parse_value(s.as_bytes()).unwrap();
+                value.to_vec()
+            })
+            .collect::<Vec<_>>();
+        let json_vector = BinaryVector::from_vec(jsonbs);
+        let args: Vec<VectorRef> = vec![Arc::new(json_vector)];
+
+        for (func, expected_result) in json_is_functions.iter().zip(expected_results.iter()) {
+            let vector = func.eval(FunctionContext::default(), &args).unwrap();
+            assert_eq!(vector.len(), json_strings.len());
+
+            for (i, expected) in expected_result.iter().enumerate() {
+                let result = vector.get_ref(i);
+                let result = result.as_boolean().unwrap().unwrap();
+                assert_eq!(result, *expected);
+            }
+        }
+    }
+}

src/common/function/src/scalars/json/json_to_string.rs

@@ -0,0 +1,174 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use std::fmt::{self, Display};
+
+use common_query::error::{InvalidFuncArgsSnafu, Result, UnsupportedInputDataTypeSnafu};
+use common_query::prelude::Signature;
+use datafusion::logical_expr::Volatility;
+use datatypes::data_type::ConcreteDataType;
+use datatypes::prelude::VectorRef;
+use datatypes::scalars::ScalarVectorBuilder;
+use datatypes::vectors::{MutableVector, StringVectorBuilder};
+use snafu::ensure;
+
+use crate::function::{Function, FunctionContext};
+
+/// Converts the `JSONB` into `String`. It's useful for displaying JSONB content.
+#[derive(Clone, Debug, Default)]
+pub struct JsonToStringFunction;
+
+const NAME: &str = "json_to_string";
+
+impl Function for JsonToStringFunction {
+    fn name(&self) -> &str {
+        NAME
+    }
+
+    fn return_type(&self, _input_types: &[ConcreteDataType]) -> Result<ConcreteDataType> {
+        Ok(ConcreteDataType::string_datatype())
+    }
+
+    fn signature(&self) -> Signature {
+        Signature::exact(
+            vec![ConcreteDataType::json_datatype()],
+            Volatility::Immutable,
+        )
+    }
+
+    fn eval(&self, _func_ctx: FunctionContext, columns: &[VectorRef]) -> Result<VectorRef> {
+        ensure!(
+            columns.len() == 1,
+            InvalidFuncArgsSnafu {
+                err_msg: format!(
+                    "The length of the args is not correct, expect exactly one, have: {}",
+                    columns.len()
+                ),
+            }
+        );
+        let jsons = &columns[0];
+
+        let size = jsons.len();
+        let datatype = jsons.data_type();
+        let mut results = StringVectorBuilder::with_capacity(size);
+
+        match datatype {
+            // JSON data type uses binary vector
+            ConcreteDataType::Binary(_) => {
+                for i in 0..size {
+                    let json = jsons.get_ref(i);
+
+                    let json = json.as_binary();
+                    let result = match json {
+                        Ok(Some(json)) => match jsonb::from_slice(json) {
+                            Ok(json) => {
+                                let json = json.to_string();
+                                Some(json)
+                            }
+                            Err(_) => {
+                                return InvalidFuncArgsSnafu {
+                                    err_msg: format!("Illegal json binary: {:?}", json),
+                                }
+                                .fail()
+                            }
+                        },
+                        _ => None,
+                    };
+
+                    results.push(result.as_deref());
+                }
+            }
+            _ => {
+                return UnsupportedInputDataTypeSnafu {
+                    function: NAME,
+                    datatypes: columns.iter().map(|c| c.data_type()).collect::<Vec<_>>(),
+                }
+                .fail();
+            }
+        }
+
+        Ok(results.to_vector())
+    }
+}
+
+impl Display for JsonToStringFunction {
+    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+        write!(f, "JSON_TO_STRING")
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use std::sync::Arc;
+
+    use common_query::prelude::TypeSignature;
+    use datatypes::scalars::ScalarVector;
+    use datatypes::vectors::BinaryVector;
+
+    use super::*;
+
+    #[test]
+    fn test_json_to_string_function() {
+        let json_to_string = JsonToStringFunction;
+
+        assert_eq!("json_to_string", json_to_string.name());
+        assert_eq!(
+            ConcreteDataType::string_datatype(),
+            json_to_string
+                .return_type(&[ConcreteDataType::json_datatype()])
+                .unwrap()
+        );
+
+        assert!(matches!(json_to_string.signature(),
+                         Signature {
+                             type_signature: TypeSignature::Exact(valid_types),
+                             volatility: Volatility::Immutable
+                         } if  valid_types == vec![ConcreteDataType::json_datatype()]
+        ));
+
+        let json_strings = [
+            r#"{"a": {"b": 2}, "b": 2, "c": 3}"#,
+            r#"{"a": 4, "b": {"c": 6}, "c": 6}"#,
+            r#"{"a": 7, "b": 8, "c": {"a": 7}}"#,
+        ];
+
+        let jsonbs = json_strings
+            .iter()
+            .map(|s| {
+                let value = jsonb::parse_value(s.as_bytes()).unwrap();
+                value.to_vec()
+            })
+            .collect::<Vec<_>>();
+
+        let json_vector = BinaryVector::from_vec(jsonbs);
+        let args: Vec<VectorRef> = vec![Arc::new(json_vector)];
+        let vector = json_to_string
+            .eval(FunctionContext::default(), &args)
+            .unwrap();
+
+        assert_eq!(3, vector.len());
+        for (i, gt) in json_strings.iter().enumerate() {
+            let result = vector.get_ref(i);
+            let result = result.as_string().unwrap().unwrap();
+            // remove whitespaces
+            assert_eq!(gt.replace(" ", ""), result);
+        }
+
+        let invalid_jsonb = vec![b"invalid json"];
+        let invalid_json_vector = BinaryVector::from_vec(invalid_jsonb);
+        let args: Vec<VectorRef> = vec![Arc::new(invalid_json_vector)];
+        let vector = json_to_string.eval(FunctionContext::default(), &args);
+        assert!(vector.is_err());
+    }
+}

src/common/function/src/scalars/json/parse_json.rs

@@ -0,0 +1,164 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use std::fmt::{self, Display};
+
+use common_query::error::{InvalidFuncArgsSnafu, Result, UnsupportedInputDataTypeSnafu};
+use common_query::prelude::Signature;
+use datafusion::logical_expr::Volatility;
+use datatypes::data_type::ConcreteDataType;
+use datatypes::prelude::VectorRef;
+use datatypes::scalars::ScalarVectorBuilder;
+use datatypes::vectors::{BinaryVectorBuilder, MutableVector};
+use snafu::ensure;
+
+use crate::function::{Function, FunctionContext};
+
+/// Parses the `String` into `JSONB`.
+#[derive(Clone, Debug, Default)]
+pub struct ParseJsonFunction;
+
+const NAME: &str = "parse_json";
+
+impl Function for ParseJsonFunction {
+    fn name(&self) -> &str {
+        NAME
+    }
+
+    fn return_type(&self, _input_types: &[ConcreteDataType]) -> Result<ConcreteDataType> {
+        Ok(ConcreteDataType::json_datatype())
+    }
+
+    fn signature(&self) -> Signature {
+        Signature::exact(
+            vec![ConcreteDataType::string_datatype()],
+            Volatility::Immutable,
+        )
+    }
+
+    fn eval(&self, _func_ctx: FunctionContext, columns: &[VectorRef]) -> Result<VectorRef> {
+        ensure!(
+            columns.len() == 1,
+            InvalidFuncArgsSnafu {
+                err_msg: format!(
+                    "The length of the args is not correct, expect exactly one, have: {}",
+                    columns.len()
+                ),
+            }
+        );
+        let json_strings = &columns[0];
+
+        let size = json_strings.len();
+        let datatype = json_strings.data_type();
+        let mut results = BinaryVectorBuilder::with_capacity(size);
+
+        match datatype {
+            ConcreteDataType::String(_) => {
+                for i in 0..size {
+                    let json_string = json_strings.get_ref(i);
+
+                    let json_string = json_string.as_string();
+                    let result = match json_string {
+                        Ok(Some(json_string)) => match jsonb::parse_value(json_string.as_bytes()) {
+                            Ok(json) => Some(json.to_vec()),
+                            Err(_) => {
+                                return InvalidFuncArgsSnafu {
+                                    err_msg: format!(
+                                        "Cannot convert the string to json, have: {}",
+                                        json_string
+                                    ),
+                                }
+                                .fail()
+                            }
+                        },
+                        _ => None,
+                    };
+
+                    results.push(result.as_deref());
+                }
+            }
+            _ => {
+                return UnsupportedInputDataTypeSnafu {
+                    function: NAME,
+                    datatypes: columns.iter().map(|c| c.data_type()).collect::<Vec<_>>(),
+                }
+                .fail();
+            }
+        }
+
+        Ok(results.to_vector())
+    }
+}
+
+impl Display for ParseJsonFunction {
+    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+        write!(f, "PARSE_JSON")
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use std::sync::Arc;
+
+    use common_query::prelude::TypeSignature;
+    use datatypes::scalars::ScalarVector;
+    use datatypes::vectors::StringVector;
+
+    use super::*;
+
+    #[test]
+    fn test_get_by_path_function() {
+        let parse_json = ParseJsonFunction;
+
+        assert_eq!("parse_json", parse_json.name());
+        assert_eq!(
+            ConcreteDataType::json_datatype(),
+            parse_json
+                .return_type(&[ConcreteDataType::json_datatype()])
+                .unwrap()
+        );
+
+        assert!(matches!(parse_json.signature(),
+                         Signature {
+                             type_signature: TypeSignature::Exact(valid_types),
+                             volatility: Volatility::Immutable
+                         } if  valid_types == vec![ConcreteDataType::string_datatype()]
+        ));
+
+        let json_strings = [
+            r#"{"a": {"b": 2}, "b": 2, "c": 3}"#,
+            r#"{"a": 4, "b": {"c": 6}, "c": 6}"#,
+            r#"{"a": 7, "b": 8, "c": {"a": 7}}"#,
+        ];
+
+        let jsonbs = json_strings
+            .iter()
+            .map(|s| {
+                let value = jsonb::parse_value(s.as_bytes()).unwrap();
+                value.to_vec()
+            })
+            .collect::<Vec<_>>();
+
+        let json_string_vector = StringVector::from_vec(json_strings.to_vec());
+        let args: Vec<VectorRef> = vec![Arc::new(json_string_vector)];
+        let vector = parse_json.eval(FunctionContext::default(), &args).unwrap();
+
+        assert_eq!(3, vector.len());
+        for (i, gt) in jsonbs.iter().enumerate() {
+            let result = vector.get_ref(i);
+            let result = result.as_binary().unwrap().unwrap();
+            assert_eq!(gt, result);
+        }
+    }
+}

src/common/function/src/system/version.rs

@@ -19,6 +19,7 @@ use common_query::error::Result;
 use common_query::prelude::{Signature, Volatility};
 use datatypes::data_type::ConcreteDataType;
 use datatypes::vectors::{StringVector, VectorRef};
+use session::context::Channel;
 
 use crate::function::{Function, FunctionContext};
 
@@ -44,11 +45,22 @@ impl Function for VersionFunction {
         Signature::exact(vec![], Volatility::Immutable)
     }
 
-    fn eval(&self, _func_ctx: FunctionContext, _columns: &[VectorRef]) -> Result<VectorRef> {
-        let result = StringVector::from(vec![format!(
-            "5.7.20-greptimedb-{}",
-            env!("CARGO_PKG_VERSION")
-        )]);
+    fn eval(&self, func_ctx: FunctionContext, _columns: &[VectorRef]) -> Result<VectorRef> {
+        let version = match func_ctx.query_ctx.channel() {
+            Channel::Mysql => {
+                format!(
+                    "{}-greptimedb-{}",
+                    std::env::var("GREPTIMEDB_MYSQL_SERVER_VERSION")
+                        .unwrap_or_else(|_| "8.4.2".to_string()),
+                    env!("CARGO_PKG_VERSION")
+                )
+            }
+            Channel::Postgres => {
+                format!("16.3-greptimedb-{}", env!("CARGO_PKG_VERSION"))
+            }
+            _ => env!("CARGO_PKG_VERSION").to_string(),
+        };
+        let result = StringVector::from(vec![version]);
         Ok(Arc::new(result))
     }
 }

src/common/grpc-expr/src/error.rs

@@ -64,12 +64,6 @@ pub enum Error {
         location: Location,
     },
 
-    #[snafu(display("Invalid column proto: {}", err_msg))]
-    InvalidColumnProto {
-        err_msg: String,
-        #[snafu(implicit)]
-        location: Location,
-    },
     #[snafu(display("Failed to create vector"))]
     CreateVector {
         #[snafu(implicit)]
@@ -137,7 +131,6 @@ impl ErrorExt for Error {
             Error::DuplicatedTimestampColumn { .. }
             | Error::DuplicatedColumnName { .. }
             | Error::MissingTimestampColumn { .. } => StatusCode::InvalidArguments,
-            Error::InvalidColumnProto { .. } => StatusCode::InvalidArguments,
             Error::CreateVector { .. } => StatusCode::InvalidArguments,
             Error::MissingField { .. } => StatusCode::InvalidArguments,
             Error::InvalidColumnDef { source, .. } => source.status_code(),

src/common/grpc-expr/src/insert.rs

@@ -14,11 +14,10 @@
 
 use api::helper;
 use api::v1::column::Values;
-use api::v1::{AddColumns, Column, CreateTableExpr};
+use api::v1::{Column, CreateTableExpr};
 use common_base::BitVec;
 use datatypes::data_type::{ConcreteDataType, DataType};
 use datatypes::prelude::VectorRef;
-use datatypes::schema::SchemaRef;
 use snafu::{ensure, ResultExt};
 use table::metadata::TableId;
 use table::table_reference::TableReference;
@@ -27,11 +26,6 @@ use crate::error::{CreateVectorSnafu, Result, UnexpectedValuesLengthSnafu};
 use crate::util;
 use crate::util::ColumnExpr;
 
-pub fn find_new_columns(schema: &SchemaRef, columns: &[Column]) -> Result<Option<AddColumns>> {
-    let column_exprs = ColumnExpr::from_columns(columns);
-    util::extract_new_columns(schema, column_exprs)
-}
-
 /// Try to build create table request from insert data.
 pub fn build_create_expr_from_insertion(
     catalog_name: &str,
@@ -114,7 +108,6 @@ mod tests {
     use super::*;
     use crate::error;
     use crate::error::ColumnDataTypeSnafu;
-    use crate::insert::find_new_columns;
 
     #[inline]
     fn build_column_schema(
@@ -281,11 +274,18 @@ mod tests {
 
         let schema = Arc::new(SchemaBuilder::try_from(columns).unwrap().build().unwrap());
 
-        assert!(find_new_columns(&schema, &[]).unwrap().is_none());
+        assert!(
+            util::extract_new_columns(&schema, ColumnExpr::from_columns(&[]))
+                .unwrap()
+                .is_none()
+        );
 
         let insert_batch = mock_insert_batch();
 
-        let add_columns = find_new_columns(&schema, &insert_batch.0).unwrap().unwrap();
+        let add_columns =
+            util::extract_new_columns(&schema, ColumnExpr::from_columns(&insert_batch.0))
+                .unwrap()
+                .unwrap();
 
         assert_eq!(5, add_columns.add_columns.len());
         let host_column = &add_columns.add_columns[0];

src/common/grpc-expr/src/lib.rs

@@ -19,4 +19,4 @@ pub mod insert;
 pub mod util;
 
 pub use alter::{alter_expr_to_request, create_table_schema};
-pub use insert::{build_create_expr_from_insertion, find_new_columns};
+pub use insert::build_create_expr_from_insertion;

src/common/grpc/src/select.rs

@@ -70,7 +70,7 @@ macro_rules! convert_arrow_array_to_grpc_vals {
                     return Ok(vals);
                 },
             )+
-            ConcreteDataType::Null(_) | ConcreteDataType::List(_) | ConcreteDataType::Dictionary(_) | ConcreteDataType::Duration(_) => unreachable!("Should not send {:?} in gRPC", $data_type),
+            ConcreteDataType::Null(_) | ConcreteDataType::List(_) | ConcreteDataType::Dictionary(_) | ConcreteDataType::Duration(_) | ConcreteDataType::Json(_) => unreachable!("Should not send {:?} in gRPC", $data_type),
         }
     }};
 }

src/common/meta/Cargo.toml

@@ -15,6 +15,7 @@ workspace = true
 anymap2 = "0.13.0"
 api.workspace = true
 async-recursion = "1.0"
+async-stream = "0.3"
 async-trait.workspace = true
 base64.workspace = true
 bytes.workspace = true

src/common/meta/src/cache_invalidator.rs

@@ -24,7 +24,7 @@ use crate::key::table_info::TableInfoKey;
 use crate::key::table_name::TableNameKey;
 use crate::key::table_route::TableRouteKey;
 use crate::key::view_info::ViewInfoKey;
-use crate::key::MetaKey;
+use crate::key::MetadataKey;
 
 /// KvBackend cache invalidator
 #[async_trait::async_trait]

src/common/meta/src/ddl/alter_logical_tables.rs

@@ -39,7 +39,7 @@ use crate::key::DeserializedValueWithBytes;
 use crate::lock_key::{CatalogLock, SchemaLock, TableLock};
 use crate::rpc::ddl::AlterTableTask;
 use crate::rpc::router::find_leaders;
-use crate::{cache_invalidator, metrics, ClusterId};
+use crate::{metrics, ClusterId};
 
 pub struct AlterLogicalTablesProcedure {
     pub context: DdlContext,
@@ -131,7 +131,7 @@ impl AlterLogicalTablesProcedure {
         let phy_raw_schemas = future::join_all(alter_region_tasks)
             .await
             .into_iter()
-            .map(|res| res.map(|mut res| res.extension.remove(ALTER_PHYSICAL_EXTENSION_KEY)))
+            .map(|res| res.map(|mut res| res.extensions.remove(ALTER_PHYSICAL_EXTENSION_KEY)))
             .collect::<Result<Vec<_>>>()?;
 
         if phy_raw_schemas.is_empty() {
@@ -170,12 +170,11 @@ impl AlterLogicalTablesProcedure {
     }
 
     pub(crate) async fn on_invalidate_table_cache(&mut self) -> Result<Status> {
-        let ctx = cache_invalidator::Context::default();
         let to_invalidate = self.build_table_cache_keys_to_invalidate();
 
         self.context
             .cache_invalidator
-            .invalidate(&ctx, &to_invalidate)
+            .invalidate(&Default::default(), &to_invalidate)
             .await?;
         Ok(Status::done())
     }

src/common/meta/src/ddl/create_logical_tables.rs

@@ -157,7 +157,7 @@ impl CreateLogicalTablesProcedure {
         let phy_raw_schemas = join_all(create_region_tasks)
             .await
             .into_iter()
-            .map(|res| res.map(|mut res| res.extension.remove(ALTER_PHYSICAL_EXTENSION_KEY)))
+            .map(|res| res.map(|mut res| res.extensions.remove(ALTER_PHYSICAL_EXTENSION_KEY)))
             .collect::<Result<Vec<_>>>()?;
 
         if phy_raw_schemas.is_empty() {

src/common/meta/src/ddl/utils.rs

@@ -15,12 +15,12 @@
 use common_catalog::consts::METRIC_ENGINE;
 use common_error::ext::BoxedError;
 use common_procedure::error::Error as ProcedureError;
-use snafu::{ensure, location, OptionExt};
+use snafu::{ensure, OptionExt, ResultExt};
 use store_api::metric_engine_consts::LOGICAL_TABLE_METADATA_KEY;
 use table::metadata::TableId;
 
 use crate::ddl::DetectingRegion;
-use crate::error::{Error, Result, TableNotFoundSnafu, UnsupportedSnafu};
+use crate::error::{Error, OperateDatanodeSnafu, Result, TableNotFoundSnafu, UnsupportedSnafu};
 use crate::key::table_name::TableNameKey;
 use crate::key::TableMetadataManagerRef;
 use crate::peer::Peer;
@@ -32,11 +32,9 @@ use crate::ClusterId;
 pub fn add_peer_context_if_needed(datanode: Peer) -> impl FnOnce(Error) -> Error {
     move |err| {
         if !err.is_retry_later() {
-            return Error::OperateDatanode {
-                location: location!(),
-                peer: datanode,
-                source: BoxedError::new(err),
-            };
+            return Err::<(), BoxedError>(BoxedError::new(err))
+                .context(OperateDatanodeSnafu { peer: datanode })
+                .unwrap_err();
         }
         err
     }

src/common/meta/src/ddl_manager.rs

@@ -441,11 +441,9 @@ async fn handle_alter_table_task(
         .table_metadata_manager()
         .table_route_manager()
         .table_route_storage()
-        .get_raw(table_id)
+        .get(table_id)
         .await?
-        .context(TableRouteNotFoundSnafu { table_id })?
-        .into_inner();
-
+        .context(TableRouteNotFoundSnafu { table_id })?;
     ensure!(
         table_route_value.is_physical(),
         UnexpectedLogicalRouteTableSnafu {

src/common/meta/src/error.rs

@@ -21,7 +21,7 @@ use common_macro::stack_trace_debug;
 use common_wal::options::WalOptions;
 use serde_json::error::Error as JsonError;
 use snafu::{Location, Snafu};
-use store_api::storage::{RegionId, RegionNumber};
+use store_api::storage::RegionId;
 use table::metadata::TableId;
 
 use crate::peer::Peer;
@@ -49,20 +49,6 @@ pub enum Error {
         region_id: RegionId,
     },
 
-    #[snafu(display("Invalid result with a txn response: {}", err_msg))]
-    InvalidTxnResult {
-        err_msg: String,
-        #[snafu(implicit)]
-        location: Location,
-    },
-
-    #[snafu(display("Invalid engine type: {}", engine_type))]
-    InvalidEngineType {
-        engine_type: String,
-        #[snafu(implicit)]
-        location: Location,
-    },
-
     #[snafu(display("Failed to connect to Etcd"))]
     ConnectEtcd {
         #[snafu(source)]
@@ -95,15 +81,6 @@ pub enum Error {
         location: Location,
     },
 
-    #[snafu(display("Sequence out of range: {}, start={}, step={}", name, start, step))]
-    SequenceOutOfRange {
-        name: String,
-        start: u64,
-        step: u64,
-        #[snafu(implicit)]
-        location: Location,
-    },
-
     #[snafu(display("Unexpected sequence value: {}", err_msg))]
     UnexpectedSequenceValue {
         err_msg: String,
@@ -327,13 +304,6 @@ pub enum Error {
         location: Location,
     },
 
-    #[snafu(display("Catalog already exists, catalog: {}", catalog))]
-    CatalogAlreadyExists {
-        catalog: String,
-        #[snafu(implicit)]
-        location: Location,
-    },
-
     #[snafu(display("Schema already exists, catalog:{}, schema: {}", catalog, schema))]
     SchemaAlreadyExists {
         catalog: String,
@@ -385,15 +355,8 @@ pub enum Error {
         location: Location,
     },
 
-    #[snafu(display("Failed to rename table, reason: {}", reason))]
-    RenameTable {
-        reason: String,
-        #[snafu(implicit)]
-        location: Location,
-    },
-
-    #[snafu(display("Invalid table metadata, err: {}", err_msg))]
-    InvalidTableMetadata {
+    #[snafu(display("Invalid metadata, err: {}", err_msg))]
+    InvalidMetadata {
         err_msg: String,
         #[snafu(implicit)]
         location: Location,
@@ -423,27 +386,6 @@ pub enum Error {
         location: Location,
     },
 
-    #[snafu(display(
-        "Failed to move region {} in table {}, err: {}",
-        region,
-        table_id,
-        err_msg
-    ))]
-    MoveRegion {
-        table_id: TableId,
-        region: RegionNumber,
-        err_msg: String,
-        #[snafu(implicit)]
-        location: Location,
-    },
-
-    #[snafu(display("Invalid catalog value"))]
-    InvalidCatalogValue {
-        source: common_catalog::error::Error,
-        #[snafu(implicit)]
-        location: Location,
-    },
-
     #[snafu(display("External error"))]
     External {
         #[snafu(implicit)]
@@ -612,13 +554,6 @@ pub enum Error {
         location: Location,
     },
 
-    #[snafu(display("Delimiter not found, key: {}", key))]
-    DelimiterNotFound {
-        key: String,
-        #[snafu(implicit)]
-        location: Location,
-    },
-
     #[snafu(display("Invalid prefix: {}, key: {}", prefix, key))]
     MismatchPrefix {
         prefix: String,
@@ -702,15 +637,12 @@ impl ErrorExt for Error {
             | ParseOption { .. }
             | RouteInfoCorrupted { .. }
             | InvalidProtoMsg { .. }
-            | InvalidTableMetadata { .. }
-            | MoveRegion { .. }
+            | InvalidMetadata { .. }
             | Unexpected { .. }
             | TableInfoNotFound { .. }
             | NextSequence { .. }
-            | SequenceOutOfRange { .. }
             | UnexpectedSequenceValue { .. }
             | InvalidHeartbeatResponse { .. }
-            | InvalidTxnResult { .. }
             | EncodeJson { .. }
             | DecodeJson { .. }
             | PayloadNotExist { .. }
@@ -734,22 +666,17 @@ impl ErrorExt for Error {
             | MetadataCorruption { .. }
             | StrFromUtf8 { .. } => StatusCode::Unexpected,
 
-            SendMessage { .. } | GetKvCache { .. } | CacheNotGet { .. } | RenameTable { .. } => {
-                StatusCode::Internal
-            }
+            SendMessage { .. } | GetKvCache { .. } | CacheNotGet { .. } => StatusCode::Internal,
 
             SchemaAlreadyExists { .. } => StatusCode::DatabaseAlreadyExists,
 
             ProcedureNotFound { .. }
             | InvalidViewInfo { .. }
             | PrimaryKeyNotFound { .. }
-            | CatalogAlreadyExists { .. }
             | EmptyKey { .. }
-            | InvalidEngineType { .. }
             | AlterLogicalTablesInvalidArguments { .. }
             | CreateLogicalTablesInvalidArguments { .. }
             | MismatchPrefix { .. }
-            | DelimiterNotFound { .. }
             | TlsConfig { .. } => StatusCode::InvalidArguments,
 
             FlowNotFound { .. } => StatusCode::FlowNotFound,
@@ -767,7 +694,6 @@ impl ErrorExt for Error {
             OperateDatanode { source, .. } => source.status_code(),
             Table { source, .. } => source.status_code(),
             RetryLater { source, .. } => source.status_code(),
-            InvalidCatalogValue { source, .. } => source.status_code(),
             ConvertAlterTableRequest { source, .. } => source.status_code(),
 
             ParseProcedureId { .. }

src/common/meta/src/key.rs

@@ -90,6 +90,7 @@
 pub mod catalog_name;
 pub mod datanode_table;
 pub mod flow;
+pub mod node_address;
 pub mod schema_name;
 pub mod table_info;
 pub mod table_name;
@@ -102,7 +103,7 @@ pub mod view_info;
 
 use std::collections::{BTreeMap, HashMap, HashSet};
 use std::fmt::Debug;
-use std::ops::Deref;
+use std::ops::{Deref, DerefMut};
 use std::sync::Arc;
 
 use bytes::Bytes;
@@ -134,6 +135,7 @@ use self::table_route::{TableRouteManager, TableRouteValue};
 use self::tombstone::TombstoneManager;
 use crate::ddl::utils::region_storage_path;
 use crate::error::{self, Result, SerdeJsonSnafu};
+use crate::key::node_address::NodeAddressValue;
 use crate::key::table_route::TableRouteKey;
 use crate::key::txn_helper::TxnOpGetResponseSet;
 use crate::kv_backend::txn::{Txn, TxnOp};
@@ -152,12 +154,15 @@ pub const TABLE_NAME_KEY_PREFIX: &str = "__table_name";
 pub const CATALOG_NAME_KEY_PREFIX: &str = "__catalog_name";
 pub const SCHEMA_NAME_KEY_PREFIX: &str = "__schema_name";
 pub const TABLE_ROUTE_PREFIX: &str = "__table_route";
+pub const NODE_ADDRESS_PREFIX: &str = "__node_address";
 
-pub const CACHE_KEY_PREFIXES: [&str; 4] = [
+/// The keys with these prefixes will be loaded into the cache when the leader starts.
+pub const CACHE_KEY_PREFIXES: [&str; 5] = [
     TABLE_NAME_KEY_PREFIX,
     CATALOG_NAME_KEY_PREFIX,
     SCHEMA_NAME_KEY_PREFIX,
     TABLE_ROUTE_PREFIX,
+    NODE_ADDRESS_PREFIX,
 ];
 
 pub type RegionDistribution = BTreeMap<DatanodeId, Vec<RegionNumber>>;
@@ -210,8 +215,13 @@ lazy_static! {
         .unwrap();
 }
 
+lazy_static! {
+    static ref NODE_ADDRESS_PATTERN: Regex =
+        Regex::new(&format!("^{NODE_ADDRESS_PREFIX}/([0-9]+)/([0-9]+)$")).unwrap();
+}
+
 /// The key of metadata.
-pub trait MetaKey<'a, T> {
+pub trait MetadataKey<'a, T> {
     fn to_bytes(&self) -> Vec<u8>;
 
     fn from_bytes(bytes: &'a [u8]) -> Result<T>;
@@ -226,7 +236,7 @@ impl From<Vec<u8>> for BytesAdapter {
     }
 }
 
-impl<'a> MetaKey<'a, BytesAdapter> for BytesAdapter {
+impl<'a> MetadataKey<'a, BytesAdapter> for BytesAdapter {
     fn to_bytes(&self) -> Vec<u8> {
         self.0.clone()
     }
@@ -236,7 +246,7 @@ impl<'a> MetaKey<'a, BytesAdapter> for BytesAdapter {
     }
 }
 
-pub(crate) trait TableMetaKeyGetTxnOp {
+pub(crate) trait MetadataKeyGetTxnOp {
     fn build_get_op(
         &self,
     ) -> (
@@ -245,7 +255,7 @@ pub(crate) trait TableMetaKeyGetTxnOp {
     );
 }
 
-pub trait TableMetaValue {
+pub trait MetadataValue {
     fn try_from_raw_value(raw_value: &[u8]) -> Result<Self>
     where
         Self: Sized;
@@ -306,6 +316,12 @@ impl<T: DeserializeOwned + Serialize> Deref for DeserializedValueWithBytes<T> {
     }
 }
 
+impl<T: DeserializeOwned + Serialize> DerefMut for DeserializedValueWithBytes<T> {
+    fn deref_mut(&mut self) -> &mut Self::Target {
+        &mut self.inner
+    }
+}
+
 impl<T: DeserializeOwned + Serialize + Debug> Debug for DeserializedValueWithBytes<T> {
     fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
         write!(
@@ -330,7 +346,7 @@ impl<T: DeserializeOwned + Serialize> Serialize for DeserializedValueWithBytes<T
     }
 }
 
-impl<'de, T: DeserializeOwned + Serialize + TableMetaValue> Deserialize<'de>
+impl<'de, T: DeserializeOwned + Serialize + MetadataValue> Deserialize<'de>
     for DeserializedValueWithBytes<T>
 {
     /// - Deserialize behaviors:
@@ -359,7 +375,7 @@ impl<T: Serialize + DeserializeOwned + Clone> Clone for DeserializedValueWithByt
     }
 }
 
-impl<T: Serialize + DeserializeOwned + TableMetaValue> DeserializedValueWithBytes<T> {
+impl<T: Serialize + DeserializeOwned + MetadataValue> DeserializedValueWithBytes<T> {
     /// Returns a struct containing a deserialized value and an original `bytes`.
     /// It accepts original bytes of inner.
     pub fn from_inner_bytes(bytes: Bytes) -> Result<Self> {
@@ -1156,10 +1172,10 @@ impl TableMetadataManager {
 }
 
 #[macro_export]
-macro_rules! impl_table_meta_value {
+macro_rules! impl_metadata_value {
     ($($val_ty: ty), *) => {
         $(
-            impl $crate::key::TableMetaValue for $val_ty {
+            impl $crate::key::MetadataValue for $val_ty {
                 fn try_from_raw_value(raw_value: &[u8]) -> Result<Self> {
                     serde_json::from_slice(raw_value).context(SerdeJsonSnafu)
                 }
@@ -1172,10 +1188,10 @@ macro_rules! impl_table_meta_value {
     }
 }
 
-macro_rules! impl_meta_key_get_txn_op {
+macro_rules! impl_metadata_key_get_txn_op {
     ($($key: ty), *) => {
         $(
-            impl $crate::key::TableMetaKeyGetTxnOp for $key {
+            impl $crate::key::MetadataKeyGetTxnOp for $key {
                 /// Returns a [TxnOp] to retrieve the corresponding value
                 /// and a filter to retrieve the value from the [TxnOpGetResponseSet]
                 fn build_get_op(
@@ -1197,7 +1213,7 @@ macro_rules! impl_meta_key_get_txn_op {
     }
 }
 
-impl_meta_key_get_txn_op! {
+impl_metadata_key_get_txn_op! {
     TableNameKey<'_>,
     TableInfoKey,
     ViewInfoKey,
@@ -1206,7 +1222,7 @@ impl_meta_key_get_txn_op! {
 }
 
 #[macro_export]
-macro_rules! impl_optional_meta_value {
+macro_rules! impl_optional_metadata_value {
     ($($val_ty: ty), *) => {
         $(
             impl $val_ty {
@@ -1222,7 +1238,7 @@ macro_rules! impl_optional_meta_value {
     }
 }
 
-impl_table_meta_value! {
+impl_metadata_value! {
     TableNameValue,
     TableInfoValue,
     ViewInfoValue,
@@ -1230,10 +1246,11 @@ impl_table_meta_value! {
     FlowInfoValue,
     FlowNameValue,
     FlowRouteValue,
-    TableFlowValue
+    TableFlowValue,
+    NodeAddressValue
 }
 
-impl_optional_meta_value! {
+impl_optional_metadata_value! {
     CatalogNameValue,
     SchemaNameValue
 }
@@ -1952,7 +1969,7 @@ mod tests {
         let table_route_value = table_metadata_manager
             .table_route_manager
             .table_route_storage()
-            .get_raw(table_id)
+            .get_with_raw_bytes(table_id)
             .await
             .unwrap()
             .unwrap();
@@ -2005,7 +2022,7 @@ mod tests {
         let table_route_value = table_metadata_manager
             .table_route_manager
             .table_route_storage()
-            .get_raw(table_id)
+            .get_with_raw_bytes(table_id)
             .await
             .unwrap()
             .unwrap();

src/common/meta/src/key/catalog_name.rs

@@ -20,8 +20,8 @@ use futures::stream::BoxStream;
 use serde::{Deserialize, Serialize};
 use snafu::{OptionExt, ResultExt};
 
-use crate::error::{self, Error, InvalidTableMetadataSnafu, Result};
-use crate::key::{MetaKey, CATALOG_NAME_KEY_PATTERN, CATALOG_NAME_KEY_PREFIX};
+use crate::error::{self, Error, InvalidMetadataSnafu, Result};
+use crate::key::{MetadataKey, CATALOG_NAME_KEY_PATTERN, CATALOG_NAME_KEY_PREFIX};
 use crate::kv_backend::KvBackendRef;
 use crate::range_stream::{PaginationStream, DEFAULT_PAGE_SIZE};
 use crate::rpc::store::RangeRequest;
@@ -56,14 +56,14 @@ impl<'a> CatalogNameKey<'a> {
     }
 }
 
-impl<'a> MetaKey<'a, CatalogNameKey<'a>> for CatalogNameKey<'_> {
+impl<'a> MetadataKey<'a, CatalogNameKey<'a>> for CatalogNameKey<'_> {
     fn to_bytes(&self) -> Vec<u8> {
         self.to_string().into_bytes()
     }
 
     fn from_bytes(bytes: &'a [u8]) -> Result<CatalogNameKey<'a>> {
         let key = std::str::from_utf8(bytes).map_err(|e| {
-            InvalidTableMetadataSnafu {
+            InvalidMetadataSnafu {
                 err_msg: format!(
                     "CatalogNameKey '{}' is not a valid UTF8 string: {e}",
                     String::from_utf8_lossy(bytes)
@@ -87,7 +87,7 @@ impl<'a> TryFrom<&'a str> for CatalogNameKey<'a> {
     fn try_from(s: &'a str) -> Result<Self> {
         let captures = CATALOG_NAME_KEY_PATTERN
             .captures(s)
-            .context(InvalidTableMetadataSnafu {
+            .context(InvalidMetadataSnafu {
                 err_msg: format!("Illegal CatalogNameKey format: '{s}'"),
             })?;
 
@@ -147,7 +147,8 @@ impl CatalogManager {
             req,
             DEFAULT_PAGE_SIZE,
             Arc::new(catalog_decoder),
-        );
+        )
+        .into_stream();
 
         Box::pin(stream)
     }

src/common/meta/src/key/datanode_table.rs

@@ -22,10 +22,10 @@ use snafu::OptionExt;
 use store_api::storage::RegionNumber;
 use table::metadata::TableId;
 
-use super::MetaKey;
-use crate::error::{InvalidTableMetadataSnafu, Result};
+use super::MetadataKey;
+use crate::error::{InvalidMetadataSnafu, Result};
 use crate::key::{
-    RegionDistribution, TableMetaValue, DATANODE_TABLE_KEY_PATTERN, DATANODE_TABLE_KEY_PREFIX,
+    MetadataValue, RegionDistribution, DATANODE_TABLE_KEY_PATTERN, DATANODE_TABLE_KEY_PREFIX,
 };
 use crate::kv_backend::txn::{Txn, TxnOp};
 use crate::kv_backend::KvBackendRef;
@@ -77,14 +77,14 @@ impl DatanodeTableKey {
     }
 }
 
-impl<'a> MetaKey<'a, DatanodeTableKey> for DatanodeTableKey {
+impl<'a> MetadataKey<'a, DatanodeTableKey> for DatanodeTableKey {
     fn to_bytes(&self) -> Vec<u8> {
         self.to_string().into_bytes()
     }
 
     fn from_bytes(bytes: &[u8]) -> Result<DatanodeTableKey> {
         let key = std::str::from_utf8(bytes).map_err(|e| {
-            InvalidTableMetadataSnafu {
+            InvalidMetadataSnafu {
                 err_msg: format!(
                     "DatanodeTableKey '{}' is not a valid UTF8 string: {e}",
                     String::from_utf8_lossy(bytes)
@@ -92,12 +92,11 @@ impl<'a> MetaKey<'a, DatanodeTableKey> for DatanodeTableKey {
             }
             .build()
         })?;
-        let captures =
-            DATANODE_TABLE_KEY_PATTERN
-                .captures(key)
-                .context(InvalidTableMetadataSnafu {
-                    err_msg: format!("Invalid DatanodeTableKey '{key}'"),
-                })?;
+        let captures = DATANODE_TABLE_KEY_PATTERN
+            .captures(key)
+            .context(InvalidMetadataSnafu {
+                err_msg: format!("Invalid DatanodeTableKey '{key}'"),
+            })?;
         // Safety: pass the regex check above
         let datanode_id = captures[1].parse::<DatanodeId>().unwrap();
         let table_id = captures[2].parse::<TableId>().unwrap();
@@ -168,7 +167,8 @@ impl DatanodeTableManager {
             req,
             DEFAULT_PAGE_SIZE,
             Arc::new(datanode_table_value_decoder),
-        );
+        )
+        .into_stream();
 
         Box::pin(stream)
     }

src/common/meta/src/key/flow.rs

@@ -38,7 +38,7 @@ use crate::key::flow::flow_name::FlowNameManager;
 use crate::key::flow::flownode_flow::FlownodeFlowManager;
 pub use crate::key::flow::table_flow::{TableFlowManager, TableFlowManagerRef};
 use crate::key::txn_helper::TxnOpGetResponseSet;
-use crate::key::{FlowId, MetaKey};
+use crate::key::{FlowId, MetadataKey};
 use crate::kv_backend::txn::Txn;
 use crate::kv_backend::KvBackendRef;
 use crate::rpc::store::BatchDeleteRequest;
@@ -66,7 +66,7 @@ impl<T> FlowScoped<T> {
     }
 }
 
-impl<'a, T: MetaKey<'a, T>> MetaKey<'a, FlowScoped<T>> for FlowScoped<T> {
+impl<'a, T: MetadataKey<'a, T>> MetadataKey<'a, FlowScoped<T>> for FlowScoped<T> {
     fn to_bytes(&self) -> Vec<u8> {
         let prefix = FlowScoped::<T>::PREFIX.as_bytes();
         let inner = self.inner.to_bytes();
@@ -295,7 +295,7 @@ mod tests {
         inner: Vec<u8>,
     }
 
-    impl<'a> MetaKey<'a, MockKey> for MockKey {
+    impl<'a> MetadataKey<'a, MockKey> for MockKey {
         fn to_bytes(&self) -> Vec<u8> {
             self.inner.clone()
         }

src/common/meta/src/key/flow/flow_info.rs

@@ -25,7 +25,7 @@ use table::table_name::TableName;
 use crate::error::{self, Result};
 use crate::key::flow::FlowScoped;
 use crate::key::txn_helper::TxnOpGetResponseSet;
-use crate::key::{DeserializedValueWithBytes, FlowId, FlowPartitionId, MetaKey, TableMetaValue};
+use crate::key::{DeserializedValueWithBytes, FlowId, FlowPartitionId, MetadataKey, MetadataValue};
 use crate::kv_backend::txn::Txn;
 use crate::kv_backend::KvBackendRef;
 use crate::FlownodeId;
@@ -42,7 +42,7 @@ lazy_static! {
 /// The layout: `__flow/info/{flow_id}`.
 pub struct FlowInfoKey(FlowScoped<FlowInfoKeyInner>);
 
-impl<'a> MetaKey<'a, FlowInfoKey> for FlowInfoKey {
+impl<'a> MetadataKey<'a, FlowInfoKey> for FlowInfoKey {
     fn to_bytes(&self) -> Vec<u8> {
         self.0.to_bytes()
     }
@@ -80,14 +80,14 @@ impl FlowInfoKeyInner {
     }
 }
 
-impl<'a> MetaKey<'a, FlowInfoKeyInner> for FlowInfoKeyInner {
+impl<'a> MetadataKey<'a, FlowInfoKeyInner> for FlowInfoKeyInner {
     fn to_bytes(&self) -> Vec<u8> {
         format!("{FLOW_INFO_KEY_PREFIX}/{}", self.flow_id).into_bytes()
     }
 
     fn from_bytes(bytes: &'a [u8]) -> Result<FlowInfoKeyInner> {
         let key = std::str::from_utf8(bytes).map_err(|e| {
-            error::InvalidTableMetadataSnafu {
+            error::InvalidMetadataSnafu {
                 err_msg: format!(
                     "FlowInfoKeyInner '{}' is not a valid UTF8 string: {e}",
                     String::from_utf8_lossy(bytes)
@@ -98,7 +98,7 @@ impl<'a> MetaKey<'a, FlowInfoKeyInner> for FlowInfoKeyInner {
         let captures =
             FLOW_INFO_KEY_PATTERN
                 .captures(key)
-                .context(error::InvalidTableMetadataSnafu {
+                .context(error::InvalidMetadataSnafu {
                     err_msg: format!("Invalid FlowInfoKeyInner '{key}'"),
                 })?;
         // Safety: pass the regex check above

src/common/meta/src/key/flow/flow_name.rs

@@ -24,7 +24,7 @@ use crate::error::{self, Result};
 use crate::key::flow::FlowScoped;
 use crate::key::txn_helper::TxnOpGetResponseSet;
 use crate::key::{
-    BytesAdapter, DeserializedValueWithBytes, FlowId, MetaKey, TableMetaValue, NAME_PATTERN,
+    BytesAdapter, DeserializedValueWithBytes, FlowId, MetadataKey, MetadataValue, NAME_PATTERN,
 };
 use crate::kv_backend::txn::Txn;
 use crate::kv_backend::KvBackendRef;
@@ -76,7 +76,7 @@ impl<'a> FlowNameKey<'a> {
     }
 }
 
-impl<'a> MetaKey<'a, FlowNameKey<'a>> for FlowNameKey<'a> {
+impl<'a> MetadataKey<'a, FlowNameKey<'a>> for FlowNameKey<'a> {
     fn to_bytes(&self) -> Vec<u8> {
         self.0.to_bytes()
     }
@@ -95,7 +95,7 @@ pub struct FlowNameKeyInner<'a> {
     pub flow_name: &'a str,
 }
 
-impl<'a> MetaKey<'a, FlowNameKeyInner<'a>> for FlowNameKeyInner<'_> {
+impl<'a> MetadataKey<'a, FlowNameKeyInner<'a>> for FlowNameKeyInner<'_> {
     fn to_bytes(&self) -> Vec<u8> {
         format!(
             "{FLOW_NAME_KEY_PREFIX}/{}/{}",
@@ -106,7 +106,7 @@ impl<'a> MetaKey<'a, FlowNameKeyInner<'a>> for FlowNameKeyInner<'_> {
 
     fn from_bytes(bytes: &'a [u8]) -> Result<FlowNameKeyInner> {
         let key = std::str::from_utf8(bytes).map_err(|e| {
-            error::InvalidTableMetadataSnafu {
+            error::InvalidMetadataSnafu {
                 err_msg: format!(
                     "FlowNameKeyInner '{}' is not a valid UTF8 string: {e}",
                     String::from_utf8_lossy(bytes)
@@ -117,7 +117,7 @@ impl<'a> MetaKey<'a, FlowNameKeyInner<'a>> for FlowNameKeyInner<'_> {
         let captures =
             FLOW_NAME_KEY_PATTERN
                 .captures(key)
-                .context(error::InvalidTableMetadataSnafu {
+                .context(error::InvalidMetadataSnafu {
                     err_msg: format!("Invalid FlowNameKeyInner '{key}'"),
                 })?;
         // Safety: pass the regex check above
@@ -200,7 +200,8 @@ impl FlowNameManager {
             req,
             DEFAULT_PAGE_SIZE,
             Arc::new(flow_name_decoder),
-        );
+        )
+        .into_stream();
 
         Box::pin(stream)
     }

src/common/meta/src/key/flow/flow_route.rs

@@ -22,7 +22,7 @@ use snafu::OptionExt;
 
 use crate::error::{self, Result};
 use crate::key::flow::FlowScoped;
-use crate::key::{BytesAdapter, FlowId, FlowPartitionId, MetaKey, TableMetaValue};
+use crate::key::{BytesAdapter, FlowId, FlowPartitionId, MetadataKey, MetadataValue};
 use crate::kv_backend::txn::{Txn, TxnOp};
 use crate::kv_backend::KvBackendRef;
 use crate::peer::Peer;
@@ -68,7 +68,7 @@ impl FlowRouteKey {
     }
 }
 
-impl<'a> MetaKey<'a, FlowRouteKey> for FlowRouteKey {
+impl<'a> MetadataKey<'a, FlowRouteKey> for FlowRouteKey {
     fn to_bytes(&self) -> Vec<u8> {
         self.0.to_bytes()
     }
@@ -101,7 +101,7 @@ impl FlowRouteKeyInner {
     }
 }
 
-impl<'a> MetaKey<'a, FlowRouteKeyInner> for FlowRouteKeyInner {
+impl<'a> MetadataKey<'a, FlowRouteKeyInner> for FlowRouteKeyInner {
     fn to_bytes(&self) -> Vec<u8> {
         format!(
             "{FLOW_ROUTE_KEY_PREFIX}/{}/{}",
@@ -112,7 +112,7 @@ impl<'a> MetaKey<'a, FlowRouteKeyInner> for FlowRouteKeyInner {
 
     fn from_bytes(bytes: &'a [u8]) -> Result<FlowRouteKeyInner> {
         let key = std::str::from_utf8(bytes).map_err(|e| {
-            error::InvalidTableMetadataSnafu {
+            error::InvalidMetadataSnafu {
                 err_msg: format!(
                     "FlowInfoKeyInner '{}' is not a valid UTF8 string: {e}",
                     String::from_utf8_lossy(bytes)
@@ -123,7 +123,7 @@ impl<'a> MetaKey<'a, FlowRouteKeyInner> for FlowRouteKeyInner {
         let captures =
             FLOW_ROUTE_KEY_PATTERN
                 .captures(key)
-                .context(error::InvalidTableMetadataSnafu {
+                .context(error::InvalidMetadataSnafu {
                     err_msg: format!("Invalid FlowInfoKeyInner '{key}'"),
                 })?;
         // Safety: pass the regex check above
@@ -180,7 +180,8 @@ impl FlowRouteManager {
             req,
             DEFAULT_PAGE_SIZE,
             Arc::new(flow_route_decoder),
-        );
+        )
+        .into_stream();
 
         Box::pin(stream)
     }
@@ -209,7 +210,7 @@ impl FlowRouteManager {
 #[cfg(test)]
 mod tests {
     use super::FlowRouteKey;
-    use crate::key::MetaKey;
+    use crate::key::MetadataKey;
 
     #[test]
     fn test_key_serialization() {

src/common/meta/src/key/flow/flownode_flow.rs

@@ -22,7 +22,7 @@ use snafu::OptionExt;
 
 use crate::error::{self, Result};
 use crate::key::flow::FlowScoped;
-use crate::key::{BytesAdapter, FlowId, FlowPartitionId, MetaKey};
+use crate::key::{BytesAdapter, FlowId, FlowPartitionId, MetadataKey};
 use crate::kv_backend::txn::{Txn, TxnOp};
 use crate::kv_backend::KvBackendRef;
 use crate::range_stream::{PaginationStream, DEFAULT_PAGE_SIZE};
@@ -44,7 +44,7 @@ const FLOWNODE_FLOW_KEY_PREFIX: &str = "flownode";
 /// The layout `__flow/flownode/{flownode_id}/{flow_id}/{partition_id}`
 pub struct FlownodeFlowKey(FlowScoped<FlownodeFlowKeyInner>);
 
-impl<'a> MetaKey<'a, FlownodeFlowKey> for FlownodeFlowKey {
+impl<'a> MetadataKey<'a, FlownodeFlowKey> for FlownodeFlowKey {
     fn to_bytes(&self) -> Vec<u8> {
         self.0.to_bytes()
     }
@@ -113,7 +113,7 @@ impl FlownodeFlowKeyInner {
     }
 }
 
-impl<'a> MetaKey<'a, FlownodeFlowKeyInner> for FlownodeFlowKeyInner {
+impl<'a> MetadataKey<'a, FlownodeFlowKeyInner> for FlownodeFlowKeyInner {
     fn to_bytes(&self) -> Vec<u8> {
         format!(
             "{FLOWNODE_FLOW_KEY_PREFIX}/{}/{}/{}",
@@ -124,7 +124,7 @@ impl<'a> MetaKey<'a, FlownodeFlowKeyInner> for FlownodeFlowKeyInner {
 
     fn from_bytes(bytes: &'a [u8]) -> Result<FlownodeFlowKeyInner> {
         let key = std::str::from_utf8(bytes).map_err(|e| {
-            error::InvalidTableMetadataSnafu {
+            error::InvalidMetadataSnafu {
                 err_msg: format!(
                     "FlownodeFlowKeyInner '{}' is not a valid UTF8 string: {e}",
                     String::from_utf8_lossy(bytes)
@@ -135,7 +135,7 @@ impl<'a> MetaKey<'a, FlownodeFlowKeyInner> for FlownodeFlowKeyInner {
         let captures =
             FLOWNODE_FLOW_KEY_PATTERN
                 .captures(key)
-                .context(error::InvalidTableMetadataSnafu {
+                .context(error::InvalidMetadataSnafu {
                     err_msg: format!("Invalid FlownodeFlowKeyInner '{key}'"),
                 })?;
         // Safety: pass the regex check above
@@ -180,7 +180,8 @@ impl FlownodeFlowManager {
             req,
             DEFAULT_PAGE_SIZE,
             Arc::new(flownode_flow_key_decoder),
-        );
+        )
+        .into_stream();
 
         Box::pin(stream.map_ok(|key| (key.flow_id(), key.partition_id())))
     }
@@ -208,7 +209,7 @@ impl FlownodeFlowManager {
 #[cfg(test)]
 mod tests {
     use crate::key::flow::flownode_flow::FlownodeFlowKey;
-    use crate::key::MetaKey;
+    use crate::key::MetadataKey;
 
     #[test]
     fn test_key_serialization() {

src/common/meta/src/key/flow/table_flow.rs

@@ -23,7 +23,7 @@ use table::metadata::TableId;
 
 use crate::error::{self, Result};
 use crate::key::flow::FlowScoped;
-use crate::key::{BytesAdapter, FlowId, FlowPartitionId, MetaKey, TableMetaValue};
+use crate::key::{BytesAdapter, FlowId, FlowPartitionId, MetadataKey, MetadataValue};
 use crate::kv_backend::txn::{Txn, TxnOp};
 use crate::kv_backend::KvBackendRef;
 use crate::peer::Peer;
@@ -56,7 +56,7 @@ struct TableFlowKeyInner {
 #[derive(Debug, PartialEq)]
 pub struct TableFlowKey(FlowScoped<TableFlowKeyInner>);
 
-impl<'a> MetaKey<'a, TableFlowKey> for TableFlowKey {
+impl<'a> MetadataKey<'a, TableFlowKey> for TableFlowKey {
     fn to_bytes(&self) -> Vec<u8> {
         self.0.to_bytes()
     }
@@ -129,7 +129,7 @@ impl TableFlowKeyInner {
     }
 }
 
-impl<'a> MetaKey<'a, TableFlowKeyInner> for TableFlowKeyInner {
+impl<'a> MetadataKey<'a, TableFlowKeyInner> for TableFlowKeyInner {
     fn to_bytes(&self) -> Vec<u8> {
         format!(
             "{TABLE_FLOW_KEY_PREFIX}/{}/{}/{}/{}",
@@ -140,7 +140,7 @@ impl<'a> MetaKey<'a, TableFlowKeyInner> for TableFlowKeyInner {
 
     fn from_bytes(bytes: &'a [u8]) -> Result<TableFlowKeyInner> {
         let key = std::str::from_utf8(bytes).map_err(|e| {
-            error::InvalidTableMetadataSnafu {
+            error::InvalidMetadataSnafu {
                 err_msg: format!(
                     "TableFlowKeyInner '{}' is not a valid UTF8 string: {e}",
                     String::from_utf8_lossy(bytes)
@@ -151,7 +151,7 @@ impl<'a> MetaKey<'a, TableFlowKeyInner> for TableFlowKeyInner {
         let captures =
             TABLE_FLOW_KEY_PATTERN
                 .captures(key)
-                .context(error::InvalidTableMetadataSnafu {
+                .context(error::InvalidMetadataSnafu {
                     err_msg: format!("Invalid TableFlowKeyInner '{key}'"),
                 })?;
         // Safety: pass the regex check above
@@ -207,7 +207,8 @@ impl TableFlowManager {
             req,
             DEFAULT_PAGE_SIZE,
             Arc::new(table_flow_decoder),
-        );
+        )
+        .into_stream();
 
         Box::pin(stream)
     }

src/common/meta/src/key/node_address.rs

@@ -0,0 +1,114 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use std::fmt::Display;
+
+use api::v1::meta::Role;
+use serde::{Deserialize, Serialize};
+use snafu::OptionExt;
+
+use crate::error::{InvalidMetadataSnafu, Result};
+use crate::key::{MetadataKey, NODE_ADDRESS_PATTERN, NODE_ADDRESS_PREFIX};
+use crate::peer::Peer;
+
+/// The key stores node address.
+///
+/// The layout: `__node_address/{role}/{node_id}`
+#[derive(Debug, PartialEq)]
+pub struct NodeAddressKey {
+    pub role: Role,
+    pub node_id: u64,
+}
+
+impl NodeAddressKey {
+    pub fn new(role: Role, node_id: u64) -> Self {
+        Self { role, node_id }
+    }
+
+    pub fn with_datanode(node_id: u64) -> Self {
+        Self::new(Role::Datanode, node_id)
+    }
+}
+
+#[derive(Debug, PartialEq, Serialize, Deserialize, Clone)]
+pub struct NodeAddressValue {
+    pub peer: Peer,
+}
+
+impl NodeAddressValue {
+    pub fn new(peer: Peer) -> Self {
+        Self { peer }
+    }
+}
+
+impl<'a> MetadataKey<'a, NodeAddressKey> for NodeAddressKey {
+    fn to_bytes(&self) -> Vec<u8> {
+        self.to_string().into_bytes()
+    }
+
+    fn from_bytes(bytes: &[u8]) -> Result<NodeAddressKey> {
+        let key = std::str::from_utf8(bytes).map_err(|e| {
+            InvalidMetadataSnafu {
+                err_msg: format!(
+                    "NodeAddressKey '{}' is not a valid UTF8 string: {e}",
+                    String::from_utf8_lossy(bytes)
+                ),
+            }
+            .build()
+        })?;
+        let captures = NODE_ADDRESS_PATTERN
+            .captures(key)
+            .context(InvalidMetadataSnafu {
+                err_msg: format!("Invalid NodeAddressKey '{key}'"),
+            })?;
+        // Safety: pass the regex check above
+        let role = captures[1].parse::<i32>().unwrap();
+        let role = Role::try_from(role).map_err(|_| {
+            InvalidMetadataSnafu {
+                err_msg: format!("Invalid Role value: {role}"),
+            }
+            .build()
+        })?;
+        let node_id = captures[2].parse::<u64>().unwrap();
+        Ok(NodeAddressKey::new(role, node_id))
+    }
+}
+
+impl Display for NodeAddressKey {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        write!(
+            f,
+            "{}/{}/{}",
+            NODE_ADDRESS_PREFIX, self.role as i32, self.node_id
+        )
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_node_address_key() {
+        let key = NodeAddressKey::new(Role::Datanode, 1);
+        let bytes = key.to_bytes();
+        let key2 = NodeAddressKey::from_bytes(&bytes).unwrap();
+        assert_eq!(key, key2);
+
+        let key = NodeAddressKey::new(Role::Flownode, 3);
+        let bytes = key.to_bytes();
+        let key2 = NodeAddressKey::from_bytes(&bytes).unwrap();
+        assert_eq!(key, key2);
+    }
+}

src/common/meta/src/key/schema_name.rs

@@ -23,8 +23,8 @@ use humantime_serde::re::humantime;
 use serde::{Deserialize, Serialize};
 use snafu::{OptionExt, ResultExt};
 
-use crate::error::{self, Error, InvalidTableMetadataSnafu, ParseOptionSnafu, Result};
-use crate::key::{MetaKey, SCHEMA_NAME_KEY_PATTERN, SCHEMA_NAME_KEY_PREFIX};
+use crate::error::{self, Error, InvalidMetadataSnafu, ParseOptionSnafu, Result};
+use crate::key::{MetadataKey, SCHEMA_NAME_KEY_PATTERN, SCHEMA_NAME_KEY_PREFIX};
 use crate::kv_backend::KvBackendRef;
 use crate::range_stream::{PaginationStream, DEFAULT_PAGE_SIZE};
 use crate::rpc::store::RangeRequest;
@@ -89,6 +89,19 @@ impl TryFrom<&HashMap<String, String>> for SchemaNameValue {
     }
 }
 
+impl From<SchemaNameValue> for HashMap<String, String> {
+    fn from(value: SchemaNameValue) -> Self {
+        let mut opts = HashMap::new();
+        if let Some(ttl) = value.ttl {
+            opts.insert(
+                OPT_KEY_TTL.to_string(),
+                format!("{}", humantime::format_duration(ttl)),
+            );
+        }
+        opts
+    }
+}
+
 impl<'a> SchemaNameKey<'a> {
     pub fn new(catalog: &'a str, schema: &'a str) -> Self {
         Self { catalog, schema }
@@ -109,14 +122,14 @@ impl Display for SchemaNameKey<'_> {
     }
 }
 
-impl<'a> MetaKey<'a, SchemaNameKey<'a>> for SchemaNameKey<'_> {
+impl<'a> MetadataKey<'a, SchemaNameKey<'a>> for SchemaNameKey<'_> {
     fn to_bytes(&self) -> Vec<u8> {
         self.to_string().into_bytes()
     }
 
     fn from_bytes(bytes: &'a [u8]) -> Result<SchemaNameKey<'a>> {
         let key = std::str::from_utf8(bytes).map_err(|e| {
-            InvalidTableMetadataSnafu {
+            InvalidMetadataSnafu {
                 err_msg: format!(
                     "SchemaNameKey '{}' is not a valid UTF8 string: {e}",
                     String::from_utf8_lossy(bytes)
@@ -142,7 +155,7 @@ impl<'a> TryFrom<&'a str> for SchemaNameKey<'a> {
     fn try_from(s: &'a str) -> Result<Self> {
         let captures = SCHEMA_NAME_KEY_PATTERN
             .captures(s)
-            .context(InvalidTableMetadataSnafu {
+            .context(InvalidMetadataSnafu {
                 err_msg: format!("Illegal SchemaNameKey format: '{s}'"),
             })?;
 
@@ -217,7 +230,8 @@ impl SchemaManager {
             req,
             DEFAULT_PAGE_SIZE,
             Arc::new(schema_decoder),
-        );
+        )
+        .into_stream();
 
         Box::pin(stream)
     }

src/common/meta/src/key/table_info.rs

@@ -23,9 +23,9 @@ use table::table_name::TableName;
 use table::table_reference::TableReference;
 
 use super::TABLE_INFO_KEY_PATTERN;
-use crate::error::{InvalidTableMetadataSnafu, Result};
+use crate::error::{InvalidMetadataSnafu, Result};
 use crate::key::txn_helper::TxnOpGetResponseSet;
-use crate::key::{DeserializedValueWithBytes, MetaKey, TableMetaValue, TABLE_INFO_KEY_PREFIX};
+use crate::key::{DeserializedValueWithBytes, MetadataKey, MetadataValue, TABLE_INFO_KEY_PREFIX};
 use crate::kv_backend::txn::Txn;
 use crate::kv_backend::KvBackendRef;
 use crate::rpc::store::BatchGetRequest;
@@ -51,14 +51,14 @@ impl Display for TableInfoKey {
     }
 }
 
-impl<'a> MetaKey<'a, TableInfoKey> for TableInfoKey {
+impl<'a> MetadataKey<'a, TableInfoKey> for TableInfoKey {
     fn to_bytes(&self) -> Vec<u8> {
         self.to_string().into_bytes()
     }
 
     fn from_bytes(bytes: &[u8]) -> Result<TableInfoKey> {
         let key = std::str::from_utf8(bytes).map_err(|e| {
-            InvalidTableMetadataSnafu {
+            InvalidMetadataSnafu {
                 err_msg: format!(
                     "TableInfoKey '{}' is not a valid UTF8 string: {e}",
                     String::from_utf8_lossy(bytes)
@@ -68,7 +68,7 @@ impl<'a> MetaKey<'a, TableInfoKey> for TableInfoKey {
         })?;
         let captures = TABLE_INFO_KEY_PATTERN
             .captures(key)
-            .context(InvalidTableMetadataSnafu {
+            .context(InvalidMetadataSnafu {
                 err_msg: format!("Invalid TableInfoKey '{key}'"),
             })?;
         // Safety: pass the regex check above

src/common/meta/src/key/table_name.rs

@@ -22,8 +22,8 @@ use snafu::OptionExt;
 use table::metadata::TableId;
 use table::table_name::TableName;
 
-use super::{MetaKey, TableMetaValue, TABLE_NAME_KEY_PATTERN, TABLE_NAME_KEY_PREFIX};
-use crate::error::{Error, InvalidTableMetadataSnafu, Result};
+use super::{MetadataKey, MetadataValue, TABLE_NAME_KEY_PATTERN, TABLE_NAME_KEY_PREFIX};
+use crate::error::{Error, InvalidMetadataSnafu, Result};
 use crate::kv_backend::memory::MemoryKvBackend;
 use crate::kv_backend::txn::{Txn, TxnOp};
 use crate::kv_backend::KvBackendRef;
@@ -63,14 +63,14 @@ impl Display for TableNameKey<'_> {
     }
 }
 
-impl<'a> MetaKey<'a, TableNameKey<'a>> for TableNameKey<'_> {
+impl<'a> MetadataKey<'a, TableNameKey<'a>> for TableNameKey<'_> {
     fn to_bytes(&self) -> Vec<u8> {
         self.to_string().into_bytes()
     }
 
     fn from_bytes(bytes: &'a [u8]) -> Result<TableNameKey<'a>> {
         let key = std::str::from_utf8(bytes).map_err(|e| {
-            InvalidTableMetadataSnafu {
+            InvalidMetadataSnafu {
                 err_msg: format!(
                     "TableNameKey '{}' is not a valid UTF8 string: {e}",
                     String::from_utf8_lossy(bytes)
@@ -80,7 +80,7 @@ impl<'a> MetaKey<'a, TableNameKey<'a>> for TableNameKey<'_> {
         })?;
         let captures = TABLE_NAME_KEY_PATTERN
             .captures(key)
-            .context(InvalidTableMetadataSnafu {
+            .context(InvalidMetadataSnafu {
                 err_msg: format!("Invalid TableNameKey '{key}'"),
             })?;
         let catalog = captures.get(1).unwrap().as_str();
@@ -128,7 +128,7 @@ impl<'a> TryFrom<&'a str> for TableNameKey<'a> {
     fn try_from(s: &'a str) -> Result<Self> {
         let captures = TABLE_NAME_KEY_PATTERN
             .captures(s)
-            .context(InvalidTableMetadataSnafu {
+            .context(InvalidMetadataSnafu {
                 err_msg: format!("Illegal TableNameKey format: '{s}'"),
             })?;
         // Safety: pass the regex check above
@@ -259,7 +259,8 @@ impl TableNameManager {
             req,
             DEFAULT_PAGE_SIZE,
             Arc::new(table_decoder),
-        );
+        )
+        .into_stream();
 
         Box::pin(stream)
     }

src/common/meta/src/key/table_route.rs

@@ -22,12 +22,13 @@ use store_api::storage::{RegionId, RegionNumber};
 use table::metadata::TableId;
 
 use crate::error::{
-    self, InvalidTableMetadataSnafu, MetadataCorruptionSnafu, Result, SerdeJsonSnafu,
-    TableRouteNotFoundSnafu, UnexpectedLogicalRouteTableSnafu,
+    InvalidMetadataSnafu, MetadataCorruptionSnafu, Result, SerdeJsonSnafu, TableRouteNotFoundSnafu,
+    UnexpectedLogicalRouteTableSnafu,
 };
+use crate::key::node_address::{NodeAddressKey, NodeAddressValue};
 use crate::key::txn_helper::TxnOpGetResponseSet;
 use crate::key::{
-    DeserializedValueWithBytes, MetaKey, RegionDistribution, TableMetaValue,
+    DeserializedValueWithBytes, MetadataKey, MetadataValue, RegionDistribution,
     TABLE_ROUTE_KEY_PATTERN, TABLE_ROUTE_PREFIX,
 };
 use crate::kv_backend::txn::Txn;
@@ -85,7 +86,7 @@ impl TableRouteValue {
                     debug_assert_eq!(region.region.id.table_id(), physical_table_id);
                     RegionId::new(table_id, region.region.id.region_number())
                 })
-                .collect::<Vec<_>>();
+                .collect();
             TableRouteValue::logical(physical_table_id, region_routes)
         }
     }
@@ -189,17 +190,17 @@ impl TableRouteValue {
                 .region_routes
                 .iter()
                 .map(|region_route| region_route.region.id.region_number())
-                .collect::<Vec<_>>(),
+                .collect(),
             TableRouteValue::Logical(x) => x
                 .region_ids()
                 .iter()
                 .map(|region_id| region_id.region_number())
-                .collect::<Vec<_>>(),
+                .collect(),
         }
     }
 }
 
-impl TableMetaValue for TableRouteValue {
+impl MetadataValue for TableRouteValue {
     fn try_from_raw_value(raw_value: &[u8]) -> Result<Self> {
         let r = serde_json::from_slice::<TableRouteValue>(raw_value);
         match r {
@@ -244,14 +245,14 @@ impl LogicalTableRouteValue {
     }
 }
 
-impl<'a> MetaKey<'a, TableRouteKey> for TableRouteKey {
+impl<'a> MetadataKey<'a, TableRouteKey> for TableRouteKey {
     fn to_bytes(&self) -> Vec<u8> {
         self.to_string().into_bytes()
     }
 
     fn from_bytes(bytes: &[u8]) -> Result<TableRouteKey> {
         let key = std::str::from_utf8(bytes).map_err(|e| {
-            InvalidTableMetadataSnafu {
+            InvalidMetadataSnafu {
                 err_msg: format!(
                     "TableRouteKey '{}' is not a valid UTF8 string: {e}",
                     String::from_utf8_lossy(bytes)
@@ -259,12 +260,11 @@ impl<'a> MetaKey<'a, TableRouteKey> for TableRouteKey {
             }
             .build()
         })?;
-        let captures =
-            TABLE_ROUTE_KEY_PATTERN
-                .captures(key)
-                .context(InvalidTableMetadataSnafu {
-                    err_msg: format!("Invalid TableRouteKey '{key}'"),
-                })?;
+        let captures = TABLE_ROUTE_KEY_PATTERN
+            .captures(key)
+            .context(InvalidMetadataSnafu {
+                err_msg: format!("Invalid TableRouteKey '{key}'"),
+            })?;
         // Safety: pass the regex check above
         let table_id = captures[1].parse::<TableId>().unwrap();
         Ok(TableRouteKey { table_id })
@@ -302,7 +302,7 @@ impl TableRouteManager {
             Some(route) => {
                 ensure!(
                     route.is_physical(),
-                    error::UnexpectedLogicalRouteTableSnafu {
+                    UnexpectedLogicalRouteTableSnafu {
                         err_msg: format!("{route:?} is a non-physical TableRouteValue.")
                     }
                 );
@@ -322,7 +322,7 @@ impl TableRouteManager {
     ) -> Result<TableId> {
         let table_route = self
             .storage
-            .get(logical_or_physical_table_id)
+            .get_inner(logical_or_physical_table_id)
             .await?
             .context(TableRouteNotFoundSnafu {
                 table_id: logical_or_physical_table_id,
@@ -336,7 +336,7 @@ impl TableRouteManager {
 
     /// Returns the [TableRouteValue::Physical] recursively.
     ///
-    /// Returns a [TableRouteNotFound](crate::error::Error::TableRouteNotFound) Error if:
+    /// Returns a [TableRouteNotFound](error::Error::TableRouteNotFound) Error if:
     /// - the physical table(`logical_or_physical_table_id`) does not exist
     /// - the corresponding physical table of the logical table(`logical_or_physical_table_id`) does not exist.
     pub async fn get_physical_table_route(
@@ -529,6 +529,15 @@ impl TableRouteStorage {
 
     /// Returns the [`TableRouteValue`].
     pub async fn get(&self, table_id: TableId) -> Result<Option<TableRouteValue>> {
+        let mut table_route = self.get_inner(table_id).await?;
+        if let Some(table_route) = &mut table_route {
+            self.remap_route_address(table_route).await?;
+        };
+
+        Ok(table_route)
+    }
+
+    async fn get_inner(&self, table_id: TableId) -> Result<Option<TableRouteValue>> {
         let key = TableRouteKey::new(table_id);
         self.kv_backend
             .get(&key.to_bytes())
@@ -538,7 +547,19 @@ impl TableRouteStorage {
     }
 
     /// Returns the [`TableRouteValue`] wrapped with [`DeserializedValueWithBytes`].
-    pub async fn get_raw(
+    pub async fn get_with_raw_bytes(
+        &self,
+        table_id: TableId,
+    ) -> Result<Option<DeserializedValueWithBytes<TableRouteValue>>> {
+        let mut table_route = self.get_with_raw_bytes_inner(table_id).await?;
+        if let Some(table_route) = &mut table_route {
+            self.remap_route_address(table_route).await?;
+        };
+
+        Ok(table_route)
+    }
+
+    async fn get_with_raw_bytes_inner(
         &self,
         table_id: TableId,
     ) -> Result<Option<DeserializedValueWithBytes<TableRouteValue>>> {
@@ -555,27 +576,27 @@ impl TableRouteStorage {
     /// Returns a [TableRouteNotFound](crate::error::Error::TableRouteNotFound) Error if:
     /// - the physical table(`logical_or_physical_table_id`) does not exist
     /// - the corresponding physical table of the logical table(`logical_or_physical_table_id`) does not exist.
-    pub async fn get_raw_physical_table_route(
+    pub async fn get_physical_table_route_with_raw_bytes(
         &self,
         logical_or_physical_table_id: TableId,
     ) -> Result<(TableId, DeserializedValueWithBytes<TableRouteValue>)> {
-        let table_route =
-            self.get_raw(logical_or_physical_table_id)
-                .await?
-                .context(TableRouteNotFoundSnafu {
-                    table_id: logical_or_physical_table_id,
-                })?;
+        let table_route = self
+            .get_with_raw_bytes(logical_or_physical_table_id)
+            .await?
+            .context(TableRouteNotFoundSnafu {
+                table_id: logical_or_physical_table_id,
+            })?;
 
         match table_route.get_inner_ref() {
             TableRouteValue::Physical(_) => Ok((logical_or_physical_table_id, table_route)),
             TableRouteValue::Logical(x) => {
                 let physical_table_id = x.physical_table_id();
-                let physical_table_route =
-                    self.get_raw(physical_table_id)
-                        .await?
-                        .context(TableRouteNotFoundSnafu {
-                            table_id: physical_table_id,
-                        })?;
+                let physical_table_route = self
+                    .get_with_raw_bytes(physical_table_id)
+                    .await?
+                    .context(TableRouteNotFoundSnafu {
+                        table_id: physical_table_id,
+                    })?;
                 Ok((physical_table_id, physical_table_route))
             }
         }
@@ -583,6 +604,13 @@ impl TableRouteStorage {
 
     /// Returns batch of [`TableRouteValue`] that respects the order of `table_ids`.
     pub async fn batch_get(&self, table_ids: &[TableId]) -> Result<Vec<Option<TableRouteValue>>> {
+        let mut table_routes = self.batch_get_inner(table_ids).await?;
+        self.remap_routes_addresses(&mut table_routes).await?;
+
+        Ok(table_routes)
+    }
+
+    async fn batch_get_inner(&self, table_ids: &[TableId]) -> Result<Vec<Option<TableRouteValue>>> {
         let keys = table_ids
             .iter()
             .map(|id| TableRouteKey::new(*id).to_bytes())
@@ -605,8 +633,107 @@ impl TableRouteStorage {
                     Ok(None)
                 }
             })
-            .collect::<Result<Vec<_>>>()
+            .collect()
     }
+
+    async fn remap_routes_addresses(
+        &self,
+        table_routes: &mut [Option<TableRouteValue>],
+    ) -> Result<()> {
+        let keys = table_routes
+            .iter()
+            .flat_map(|table_route| {
+                table_route
+                    .as_ref()
+                    .map(extract_address_keys)
+                    .unwrap_or_default()
+            })
+            .collect::<HashSet<_>>()
+            .into_iter()
+            .collect();
+        let node_addrs = self.get_node_addresses(keys).await?;
+        for table_route in table_routes.iter_mut().flatten() {
+            set_addresses(&node_addrs, table_route)?;
+        }
+
+        Ok(())
+    }
+
+    async fn remap_route_address(&self, table_route: &mut TableRouteValue) -> Result<()> {
+        let keys = extract_address_keys(table_route).into_iter().collect();
+        let node_addrs = self.get_node_addresses(keys).await?;
+        set_addresses(&node_addrs, table_route)?;
+
+        Ok(())
+    }
+
+    async fn get_node_addresses(
+        &self,
+        keys: Vec<Vec<u8>>,
+    ) -> Result<HashMap<u64, NodeAddressValue>> {
+        if keys.is_empty() {
+            return Ok(HashMap::default());
+        }
+
+        self.kv_backend
+            .batch_get(BatchGetRequest { keys })
+            .await?
+            .kvs
+            .into_iter()
+            .map(|kv| {
+                let node_id = NodeAddressKey::from_bytes(&kv.key)?.node_id;
+                let node_addr = NodeAddressValue::try_from_raw_value(&kv.value)?;
+                Ok((node_id, node_addr))
+            })
+            .collect()
+    }
+}
+
+fn set_addresses(
+    node_addrs: &HashMap<u64, NodeAddressValue>,
+    table_route: &mut TableRouteValue,
+) -> Result<()> {
+    let TableRouteValue::Physical(physical_table_route) = table_route else {
+        return Ok(());
+    };
+
+    for region_route in &mut physical_table_route.region_routes {
+        if let Some(leader) = &mut region_route.leader_peer {
+            if let Some(node_addr) = node_addrs.get(&leader.id) {
+                leader.addr = node_addr.peer.addr.clone();
+            }
+        }
+        for follower in &mut region_route.follower_peers {
+            if let Some(node_addr) = node_addrs.get(&follower.id) {
+                follower.addr = node_addr.peer.addr.clone();
+            }
+        }
+    }
+
+    Ok(())
+}
+
+fn extract_address_keys(table_route: &TableRouteValue) -> HashSet<Vec<u8>> {
+    let TableRouteValue::Physical(physical_table_route) = table_route else {
+        return HashSet::default();
+    };
+
+    physical_table_route
+        .region_routes
+        .iter()
+        .flat_map(|region_route| {
+            region_route
+                .follower_peers
+                .iter()
+                .map(|peer| NodeAddressKey::with_datanode(peer.id).to_bytes())
+                .chain(
+                    region_route
+                        .leader_peer
+                        .as_ref()
+                        .map(|leader| NodeAddressKey::with_datanode(leader.id).to_bytes()),
+                )
+        })
+        .collect()
 }
 
 #[cfg(test)]
@@ -615,7 +742,9 @@ mod tests {
 
     use super::*;
     use crate::kv_backend::memory::MemoryKvBackend;
-    use crate::kv_backend::TxnService;
+    use crate::kv_backend::{KvBackend, TxnService};
+    use crate::peer::Peer;
+    use crate::rpc::store::PutRequest;
 
     #[test]
     fn test_table_route_compatibility() {
@@ -644,18 +773,18 @@ mod tests {
     }
 
     #[tokio::test]
-    async fn test_table_route_storage_get_raw_empty() {
+    async fn test_table_route_storage_get_with_raw_bytes_empty() {
         let kv = Arc::new(MemoryKvBackend::default());
         let table_route_storage = TableRouteStorage::new(kv);
-        let table_route = table_route_storage.get_raw(1024).await.unwrap();
+        let table_route = table_route_storage.get_with_raw_bytes(1024).await.unwrap();
         assert!(table_route.is_none());
     }
 
     #[tokio::test]
-    async fn test_table_route_storage_get_raw() {
+    async fn test_table_route_storage_get_with_raw_bytes() {
         let kv = Arc::new(MemoryKvBackend::default());
         let table_route_storage = TableRouteStorage::new(kv.clone());
-        let table_route = table_route_storage.get_raw(1024).await.unwrap();
+        let table_route = table_route_storage.get_with_raw_bytes(1024).await.unwrap();
         assert!(table_route.is_none());
         let table_route_manager = TableRouteManager::new(kv.clone());
         let table_route_value = TableRouteValue::Logical(LogicalTableRouteValue {
@@ -668,7 +797,7 @@ mod tests {
             .unwrap();
         let r = kv.txn(txn).await.unwrap();
         assert!(r.succeeded);
-        let table_route = table_route_storage.get_raw(1024).await.unwrap();
+        let table_route = table_route_storage.get_with_raw_bytes(1024).await.unwrap();
         assert!(table_route.is_some());
         let got = table_route.unwrap().inner;
         assert_eq!(got, table_route_value);
@@ -719,4 +848,61 @@ mod tests {
         assert!(results[2].is_none());
         assert_eq!(results[3].as_ref().unwrap(), &routes[0].1);
     }
+
+    #[tokio::test]
+    async fn remap_route_address_updates_addresses() {
+        let kv = Arc::new(MemoryKvBackend::default());
+        let table_route_storage = TableRouteStorage::new(kv.clone());
+        let mut table_route = TableRouteValue::Physical(PhysicalTableRouteValue {
+            region_routes: vec![RegionRoute {
+                leader_peer: Some(Peer {
+                    id: 1,
+                    ..Default::default()
+                }),
+                follower_peers: vec![Peer {
+                    id: 2,
+                    ..Default::default()
+                }],
+                ..Default::default()
+            }],
+            version: 0,
+        });
+
+        kv.put(PutRequest {
+            key: NodeAddressKey::with_datanode(1).to_bytes(),
+            value: NodeAddressValue {
+                peer: Peer {
+                    addr: "addr1".to_string(),
+                    ..Default::default()
+                },
+            }
+            .try_as_raw_value()
+            .unwrap(),
+            ..Default::default()
+        })
+        .await
+        .unwrap();
+
+        table_route_storage
+            .remap_route_address(&mut table_route)
+            .await
+            .unwrap();
+
+        if let TableRouteValue::Physical(physical_table_route) = table_route {
+            assert_eq!(
+                physical_table_route.region_routes[0]
+                    .leader_peer
+                    .as_ref()
+                    .unwrap()
+                    .addr,
+                "addr1"
+            );
+            assert_eq!(
+                physical_table_route.region_routes[0].follower_peers[0].addr,
+                ""
+            );
+        } else {
+            panic!("Expected PhysicalTableRouteValue");
+        }
+    }
 }