Compare commits

..

16 Commits

Author SHA1 Message Date
Discord9
f995204060 test: more reduce tests 2023-09-06 16:38:51 +08:00
Discord9
93561291e4 support more binary function 2023-09-06 16:38:51 +08:00
Discord9
9f59d68391 eval func 2023-09-06 16:37:49 +08:00
Discord9
51083b12bd reduce_bucketed 2023-09-06 16:37:49 +08:00
Discord9
c80165c377 test: simple render 2023-09-06 16:37:49 +08:00
Discord9
76d8709774 sink&source 2023-09-06 16:37:49 +08:00
Discord9
2cf7d6d569 feat: build_accumulable 2023-09-06 16:37:49 +08:00
Discord9
045c8079e6 feat: flow util func 2023-09-06 16:37:49 +08:00
Discord9
54f2f6495f mfp & reduce partially 2023-09-06 16:37:49 +08:00
Discord9
2798d266f5 feat: render plan partially writen 2023-09-06 16:37:49 +08:00
Discord9
824d03a642 working on reduce 2023-09-06 16:36:41 +08:00
Discord9
47f41371d0 Arrangement&types 2023-09-06 16:36:41 +08:00
Discord9
d702b6e5c4 use newer DD 2023-09-06 16:36:41 +08:00
Discord9
13c02f3f92 basic skeleton 2023-09-06 16:36:41 +08:00
Discord9
b52eb2313e renamed as greptime-flow 2023-09-06 16:36:41 +08:00
Discord9
d422bc8401 basic demo 2023-09-06 16:36:41 +08:00
961 changed files with 44878 additions and 48168 deletions

View File

@@ -0,0 +1,93 @@
name: Build and push dev-builder image
description: Build and push dev-builder image to DockerHub and ACR
inputs:
dockerhub-image-registry:
description: The dockerhub image registry to store the images
required: false
default: docker.io
dockerhub-image-registry-username:
description: The dockerhub username to login to the image registry
required: true
dockerhub-image-registry-token:
description: The dockerhub token to login to the image registry
required: true
dockerhub-image-namespace:
description: The dockerhub namespace of the image registry to store the images
required: false
default: greptime
acr-image-registry:
description: The ACR image registry to store the images
required: true
acr-image-registry-username:
description: The ACR username to login to the image registry
required: true
acr-image-registry-password:
description: The ACR password to login to the image registry
required: true
acr-image-namespace:
description: The ACR namespace of the image registry to store the images
required: false
default: greptime
version:
description: Version of the dev-builder
required: false
default: latest
runs:
using: composite
steps:
- name: Login to Dockerhub
uses: docker/login-action@v2
with:
registry: ${{ inputs.dockerhub-image-registry }}
username: ${{ inputs.dockerhub-image-registry-username }}
password: ${{ inputs.dockerhub-image-registry-token }}
- name: Build and push ubuntu dev builder image to dockerhub
shell: bash
run:
make dev-builder \
BASE_IMAGE=ubuntu \
BUILDX_MULTI_PLATFORM_BUILD=true \
IMAGE_REGISTRY=${{ inputs.dockerhub-image-registry }} \
IMAGE_NAMESPACE=${{ inputs.dockerhub-image-namespace }} \
IMAGE_TAG=${{ inputs.version }}
- name: Build and push centos dev builder image to dockerhub
shell: bash
run:
make dev-builder \
BASE_IMAGE=centos \
BUILDX_MULTI_PLATFORM_BUILD=true \
IMAGE_REGISTRY=${{ inputs.dockerhub-image-registry }} \
IMAGE_NAMESPACE=${{ inputs.dockerhub-image-namespace }} \
IMAGE_TAG=${{ inputs.version }}
- name: Login to ACR
uses: docker/login-action@v2
continue-on-error: true
with:
registry: ${{ inputs.acr-image-registry }}
username: ${{ inputs.acr-image-registry-username }}
password: ${{ inputs.acr-image-registry-password }}
- name: Build and push ubuntu dev builder image to ACR
shell: bash
continue-on-error: true
run: # buildx will cache the images that already built, so it will not take long time to build the images again.
make dev-builder \
BASE_IMAGE=ubuntu \
BUILDX_MULTI_PLATFORM_BUILD=true \
IMAGE_REGISTRY=${{ inputs.acr-image-registry }} \
IMAGE_NAMESPACE=${{ inputs.acr-image-namespace }} \
IMAGE_TAG=${{ inputs.version }}
- name: Build and push centos dev builder image to ACR
shell: bash
continue-on-error: true
run: # buildx will cache the images that already built, so it will not take long time to build the images again.
make dev-builder \
BASE_IMAGE=centos \
BUILDX_MULTI_PLATFORM_BUILD=true \
IMAGE_REGISTRY=${{ inputs.acr-image-registry }} \
IMAGE_NAMESPACE=${{ inputs.acr-image-namespace }} \
IMAGE_TAG=${{ inputs.version }}

View File

@@ -1,76 +0,0 @@
name: Build and push dev-builder images
description: Build and push dev-builder images to DockerHub and ACR
inputs:
dockerhub-image-registry:
description: The dockerhub image registry to store the images
required: false
default: docker.io
dockerhub-image-registry-username:
description: The dockerhub username to login to the image registry
required: true
dockerhub-image-registry-token:
description: The dockerhub token to login to the image registry
required: true
dockerhub-image-namespace:
description: The dockerhub namespace of the image registry to store the images
required: false
default: greptime
version:
description: Version of the dev-builder
required: false
default: latest
build-dev-builder-ubuntu:
description: Build dev-builder-ubuntu image
required: false
default: 'true'
build-dev-builder-centos:
description: Build dev-builder-centos image
required: false
default: 'true'
build-dev-builder-android:
description: Build dev-builder-android image
required: false
default: 'true'
runs:
using: composite
steps:
- name: Login to Dockerhub
uses: docker/login-action@v2
with:
registry: ${{ inputs.dockerhub-image-registry }}
username: ${{ inputs.dockerhub-image-registry-username }}
password: ${{ inputs.dockerhub-image-registry-token }}
- name: Build and push dev-builder-ubuntu image
shell: bash
if: ${{ inputs.build-dev-builder-ubuntu == 'true' }}
run: |
make dev-builder \
BASE_IMAGE=ubuntu \
BUILDX_MULTI_PLATFORM_BUILD=true \
IMAGE_REGISTRY=${{ inputs.dockerhub-image-registry }} \
IMAGE_NAMESPACE=${{ inputs.dockerhub-image-namespace }} \
IMAGE_TAG=${{ inputs.version }}
- name: Build and push dev-builder-centos image
shell: bash
if: ${{ inputs.build-dev-builder-centos == 'true' }}
run: |
make dev-builder \
BASE_IMAGE=centos \
BUILDX_MULTI_PLATFORM_BUILD=true \
IMAGE_REGISTRY=${{ inputs.dockerhub-image-registry }} \
IMAGE_NAMESPACE=${{ inputs.dockerhub-image-namespace }} \
IMAGE_TAG=${{ inputs.version }}
- name: Build and push dev-builder-android image # Only build image for amd64 platform.
shell: bash
if: ${{ inputs.build-dev-builder-android == 'true' }}
run: |
make dev-builder \
BASE_IMAGE=android \
IMAGE_REGISTRY=${{ inputs.dockerhub-image-registry }} \
IMAGE_NAMESPACE=${{ inputs.dockerhub-image-namespace }} \
IMAGE_TAG=${{ inputs.version }} && \
docker push ${{ inputs.dockerhub-image-registry }}/${{ inputs.dockerhub-image-namespace }}/dev-builder-android:${{ inputs.version }}

View File

@@ -16,20 +16,35 @@ inputs:
version:
description: Version of the artifact
required: true
release-to-s3-bucket:
description: S3 bucket to store released artifacts
required: true
aws-access-key-id:
description: AWS access key id
required: true
aws-secret-access-key:
description: AWS secret access key
required: true
aws-region:
description: AWS region
required: true
upload-to-s3:
description: Upload to S3
required: false
default: 'true'
upload-latest-artifacts:
description: Upload the latest artifacts to S3
required: false
default: 'true'
working-dir:
description: Working directory to build the artifacts
required: false
default: .
build-android-artifacts:
description: Build android artifacts
required: false
default: 'false'
runs:
using: composite
steps:
- name: Build greptime binary
shell: bash
if: ${{ inputs.build-android-artifacts == 'false' }}
run: |
cd ${{ inputs.working-dir }} && \
make build-by-dev-builder \
@@ -39,25 +54,14 @@ runs:
- name: Upload artifacts
uses: ./.github/actions/upload-artifacts
if: ${{ inputs.build-android-artifacts == 'false' }}
with:
artifacts-dir: ${{ inputs.artifacts-dir }}
target-file: ./target/${{ inputs.cargo-profile }}/greptime
version: ${{ inputs.version }}
working-dir: ${{ inputs.working-dir }}
# TODO(zyy17): We can remove build-android-artifacts flag in the future.
- name: Build greptime binary
shell: bash
if: ${{ inputs.build-android-artifacts == 'true' }}
run: |
cd ${{ inputs.working-dir }} && make strip-android-bin
- name: Upload android artifacts
uses: ./.github/actions/upload-artifacts
if: ${{ inputs.build-android-artifacts == 'true' }}
with:
artifacts-dir: ${{ inputs.artifacts-dir }}
target-file: ./target/aarch64-linux-android/release/greptime
version: ${{ inputs.version }}
release-to-s3-bucket: ${{ inputs.release-to-s3-bucket }}
aws-access-key-id: ${{ inputs.aws-access-key-id }}
aws-secret-access-key: ${{ inputs.aws-secret-access-key }}
aws-region: ${{ inputs.aws-region }}
upload-to-s3: ${{ inputs.upload-to-s3 }}
upload-latest-artifacts: ${{ inputs.upload-latest-artifacts }}
working-dir: ${{ inputs.working-dir }}

View File

@@ -13,10 +13,30 @@ inputs:
disable-run-tests:
description: Disable running integration tests
required: true
release-to-s3-bucket:
description: S3 bucket to store released artifacts
required: true
aws-access-key-id:
description: AWS access key id
required: true
aws-secret-access-key:
description: AWS secret access key
required: true
aws-region:
description: AWS region
required: true
dev-mode:
description: Enable dev mode, only build standard greptime
required: false
default: 'false'
upload-to-s3:
description: Upload to S3
required: false
default: 'true'
upload-latest-artifacts:
description: Upload the latest artifacts to S3
required: false
default: 'true'
working-dir:
description: Working directory to build the artifacts
required: false
@@ -48,6 +68,12 @@ runs:
cargo-profile: ${{ inputs.cargo-profile }}
artifacts-dir: greptime-linux-${{ inputs.arch }}-pyo3-${{ inputs.version }}
version: ${{ inputs.version }}
release-to-s3-bucket: ${{ inputs.release-to-s3-bucket }}
aws-access-key-id: ${{ inputs.aws-access-key-id }}
aws-secret-access-key: ${{ inputs.aws-secret-access-key }}
aws-region: ${{ inputs.aws-region }}
upload-to-s3: ${{ inputs.upload-to-s3 }}
upload-latest-artifacts: ${{ inputs.upload-latest-artifacts }}
working-dir: ${{ inputs.working-dir }}
- name: Build greptime without pyo3
@@ -59,6 +85,12 @@ runs:
cargo-profile: ${{ inputs.cargo-profile }}
artifacts-dir: greptime-linux-${{ inputs.arch }}-${{ inputs.version }}
version: ${{ inputs.version }}
release-to-s3-bucket: ${{ inputs.release-to-s3-bucket }}
aws-access-key-id: ${{ inputs.aws-access-key-id }}
aws-secret-access-key: ${{ inputs.aws-secret-access-key }}
aws-region: ${{ inputs.aws-region }}
upload-to-s3: ${{ inputs.upload-to-s3 }}
upload-latest-artifacts: ${{ inputs.upload-latest-artifacts }}
working-dir: ${{ inputs.working-dir }}
- name: Clean up the target directory # Clean up the target directory for the centos7 base image, or it will still use the objects of last build.
@@ -75,14 +107,10 @@ runs:
cargo-profile: ${{ inputs.cargo-profile }}
artifacts-dir: greptime-linux-${{ inputs.arch }}-centos-${{ inputs.version }}
version: ${{ inputs.version }}
release-to-s3-bucket: ${{ inputs.release-to-s3-bucket }}
aws-access-key-id: ${{ inputs.aws-access-key-id }}
aws-secret-access-key: ${{ inputs.aws-secret-access-key }}
aws-region: ${{ inputs.aws-region }}
upload-to-s3: ${{ inputs.upload-to-s3 }}
upload-latest-artifacts: ${{ inputs.upload-latest-artifacts }}
working-dir: ${{ inputs.working-dir }}
- name: Build greptime on android base image
uses: ./.github/actions/build-greptime-binary
if: ${{ inputs.arch == 'amd64' && inputs.dev-mode == 'false' }} # Only build android base image on amd64.
with:
base-image: android
artifacts-dir: greptime-android-arm64-${{ inputs.version }}
version: ${{ inputs.version }}
working-dir: ${{ inputs.working-dir }}
build-android-artifacts: true

View File

@@ -19,9 +19,21 @@ inputs:
disable-run-tests:
description: Disable running integration tests
required: true
release-to-s3-bucket:
description: S3 bucket to store released artifacts
required: true
artifacts-dir:
description: Directory to store artifacts
required: true
aws-access-key-id:
description: AWS access key id
required: true
aws-secret-access-key:
description: AWS secret access key
required: true
aws-region:
description: AWS region
required: true
runs:
using: composite
steps:
@@ -87,3 +99,7 @@ runs:
artifacts-dir: ${{ inputs.artifacts-dir }}
target-file: target/${{ inputs.arch }}/${{ inputs.cargo-profile }}/greptime
version: ${{ inputs.version }}
release-to-s3-bucket: ${{ inputs.release-to-s3-bucket }}
aws-access-key-id: ${{ inputs.aws-access-key-id }}
aws-secret-access-key: ${{ inputs.aws-secret-access-key }}
aws-region: ${{ inputs.aws-region }}

View File

@@ -1,80 +0,0 @@
name: Build Windows artifacts
description: Build Windows artifacts
inputs:
arch:
description: Architecture to build
required: true
rust-toolchain:
description: Rust toolchain to use
required: true
cargo-profile:
description: Cargo profile to build
required: true
features:
description: Cargo features to build
required: true
version:
description: Version of the artifact
required: true
disable-run-tests:
description: Disable running integration tests
required: true
artifacts-dir:
description: Directory to store artifacts
required: true
runs:
using: composite
steps:
- uses: arduino/setup-protoc@v1
- name: Install rust toolchain
uses: dtolnay/rust-toolchain@master
with:
toolchain: ${{ inputs.rust-toolchain }}
targets: ${{ inputs.arch }}
components: llvm-tools-preview
- name: Rust Cache
uses: Swatinem/rust-cache@v2
- name: Install Python
uses: actions/setup-python@v4
with:
python-version: '3.10'
- name: Install PyArrow Package
shell: pwsh
run: pip install pyarrow
- name: Install WSL distribution
uses: Vampire/setup-wsl@v2
with:
distribution: Ubuntu-22.04
- name: Install latest nextest release # For integration tests.
if: ${{ inputs.disable-run-tests == 'false' }}
uses: taiki-e/install-action@nextest
- name: Run integration tests
if: ${{ inputs.disable-run-tests == 'false' }}
shell: pwsh
run: make test sqlness-test
- name: Upload sqlness logs
if: ${{ failure() }} # Only upload logs when the integration tests failed.
uses: actions/upload-artifact@v3
with:
name: sqlness-logs
path: ${{ runner.temp }}/greptime-*.log
retention-days: 3
- name: Build greptime binary
shell: pwsh
run: cargo build --profile ${{ inputs.cargo-profile }} --features ${{ inputs.features }} --target ${{ inputs.arch }}
- name: Upload artifacts
uses: ./.github/actions/upload-artifacts
with:
artifacts-dir: ${{ inputs.artifacts-dir }}
target-file: target/${{ inputs.arch }}/${{ inputs.cargo-profile }}/greptime
version: ${{ inputs.version }}

View File

@@ -1,5 +1,5 @@
name: Publish GitHub release
description: Publish GitHub release
name: Release artifacts
description: Release artifacts
inputs:
version:
description: Version to release

View File

@@ -1,138 +0,0 @@
name: Release CN artifacts
description: Release artifacts to CN region
inputs:
src-image-registry:
description: The source image registry to store the images
required: true
default: docker.io
src-image-namespace:
description: The namespace of the source image registry to store the images
required: true
default: greptime
src-image-name:
description: The name of the source image
required: false
default: greptimedb
dst-image-registry:
description: The destination image registry to store the images
required: true
dst-image-namespace:
description: The namespace of the destination image registry to store the images
required: true
default: greptime
dst-image-registry-username:
description: The username to login to the image registry
required: true
dst-image-registry-password:
description: The password to login to the image registry
required: true
version:
description: Version of the artifact
required: true
dev-mode:
description: Enable dev mode, only push standard greptime
required: false
default: 'false'
push-latest-tag:
description: Whether to push the latest tag of the image
required: false
default: 'true'
aws-cn-s3-bucket:
description: S3 bucket to store released artifacts in CN region
required: true
aws-cn-access-key-id:
description: AWS access key id in CN region
required: true
aws-cn-secret-access-key:
description: AWS secret access key in CN region
required: true
aws-cn-region:
description: AWS region in CN
required: true
upload-to-s3:
description: Upload to S3
required: false
default: 'true'
artifacts-dir:
description: Directory to store artifacts
required: false
default: 'artifacts'
update-version-info:
description: Update the version info in S3
required: false
default: 'true'
upload-max-retry-times:
description: Max retry times for uploading artifacts to S3
required: false
default: "20"
upload-retry-timeout:
description: Timeout for uploading artifacts to S3
required: false
default: "30" # minutes
runs:
using: composite
steps:
- name: Download artifacts
uses: actions/download-artifact@v3
with:
path: ${{ inputs.artifacts-dir }}
- name: Release artifacts to cn region
uses: nick-invision/retry@v2
if: ${{ inputs.upload-to-s3 == 'true' }}
env:
AWS_ACCESS_KEY_ID: ${{ inputs.aws-cn-access-key-id }}
AWS_SECRET_ACCESS_KEY: ${{ inputs.aws-cn-secret-access-key }}
AWS_DEFAULT_REGION: ${{ inputs.aws-cn-region }}
UPDATE_VERSION_INFO: ${{ inputs.update-version-info }}
with:
max_attempts: ${{ inputs.upload-max-retry-times }}
timeout_minutes: ${{ inputs.upload-retry-timeout }}
command: |
./.github/scripts/upload-artifacts-to-s3.sh \
${{ inputs.artifacts-dir }} \
${{ inputs.version }} \
${{ inputs.aws-cn-s3-bucket }}
- name: Push greptimedb image from Dockerhub to ACR
shell: bash
env:
DST_REGISTRY_USERNAME: ${{ inputs.dst-image-registry-username }}
DST_REGISTRY_PASSWORD: ${{ inputs.dst-image-registry-password }}
run: |
./.github/scripts/copy-image.sh \
${{ inputs.src-image-registry }}/${{ inputs.src-image-namespace }}/${{ inputs.src-image-name }}:${{ inputs.version }} \
${{ inputs.dst-image-registry }}/${{ inputs.dst-image-namespace }}
- name: Push latest greptimedb image from Dockerhub to ACR
shell: bash
if: ${{ inputs.push-latest-tag == 'true' }}
env:
DST_REGISTRY_USERNAME: ${{ inputs.dst-image-registry-username }}
DST_REGISTRY_PASSWORD: ${{ inputs.dst-image-registry-password }}
run: |
./.github/scripts/copy-image.sh \
${{ inputs.src-image-registry }}/${{ inputs.src-image-namespace }}/${{ inputs.src-image-name }}:latest \
${{ inputs.dst-image-registry }}/${{ inputs.dst-image-namespace }}
- name: Push greptimedb-centos image from DockerHub to ACR
shell: bash
if: ${{ inputs.dev-mode == 'false' }}
env:
DST_REGISTRY_USERNAME: ${{ inputs.dst-image-registry-username }}
DST_REGISTRY_PASSWORD: ${{ inputs.dst-image-registry-password }}
run: |
./.github/scripts/copy-image.sh \
${{ inputs.src-image-registry }}/${{ inputs.src-image-namespace }}/${{ inputs.src-image-name }}-centos:latest \
${{ inputs.dst-image-registry }}/${{ inputs.dst-image-namespace }}
- name: Push greptimedb-centos image from DockerHub to ACR
shell: bash
if: ${{ inputs.dev-mode == 'false' && inputs.push-latest-tag == 'true' }}
env:
DST_REGISTRY_USERNAME: ${{ inputs.dst-image-registry-username }}
DST_REGISTRY_PASSWORD: ${{ inputs.dst-image-registry-password }}
run: |
./.github/scripts/copy-image.sh \
${{ inputs.src-image-registry }}/${{ inputs.src-image-namespace }}/${{ inputs.src-image-name }}-centos:latest \
${{ inputs.dst-image-registry }}/${{ inputs.dst-image-namespace }}

View File

@@ -10,6 +10,34 @@ inputs:
version:
description: Version of the artifact
required: true
release-to-s3-bucket:
description: S3 bucket to store released artifacts
required: true
aws-access-key-id:
description: AWS access key id
required: true
aws-secret-access-key:
description: AWS secret access key
required: true
aws-region:
description: AWS region
required: true
upload-to-s3:
description: Upload to S3
required: false
default: 'true'
upload-latest-artifacts:
description: Upload the latest artifacts to S3
required: false
default: 'true'
upload-max-retry-times:
description: Max retry times for uploading artifacts to S3
required: false
default: "20"
upload-retry-timeout:
description: Timeout for uploading artifacts to S3
required: false
default: "10" # minutes
working-dir:
description: Working directory to upload the artifacts
required: false
@@ -33,21 +61,9 @@ runs:
working-directory: ${{ inputs.working-dir }}
shell: bash
run: |
tar -zcvf ${{ inputs.artifacts-dir }}.tar.gz ${{ inputs.artifacts-dir }}
- name: Calculate checksum
if: runner.os != 'Windows'
working-directory: ${{ inputs.working-dir }}
shell: bash
run: |
tar -zcvf ${{ inputs.artifacts-dir }}.tar.gz ${{ inputs.artifacts-dir }} && \
echo $(shasum -a 256 ${{ inputs.artifacts-dir }}.tar.gz | cut -f1 -d' ') > ${{ inputs.artifacts-dir }}.sha256sum
- name: Calculate checksum on Windows
if: runner.os == 'Windows'
working-directory: ${{ inputs.working-dir }}
shell: pwsh
run: Get-FileHash ${{ inputs.artifacts-dir }}.tar.gz -Algorithm SHA256 | select -ExpandProperty Hash > ${{ inputs.artifacts-dir }}.sha256sum
# Note: The artifacts will be double zip compressed(related issue: https://github.com/actions/upload-artifact/issues/39).
# However, when we use 'actions/download-artifact@v3' to download the artifacts, it will be automatically unzipped.
- name: Upload artifacts
@@ -61,3 +77,49 @@ runs:
with:
name: ${{ inputs.artifacts-dir }}.sha256sum
path: ${{ inputs.working-dir }}/${{ inputs.artifacts-dir }}.sha256sum
- name: Upload artifacts to S3
if: ${{ inputs.upload-to-s3 == 'true' }}
uses: nick-invision/retry@v2
env:
AWS_ACCESS_KEY_ID: ${{ inputs.aws-access-key-id }}
AWS_SECRET_ACCESS_KEY: ${{ inputs.aws-secret-access-key }}
AWS_DEFAULT_REGION: ${{ inputs.aws-region }}
with:
max_attempts: ${{ inputs.upload-max-retry-times }}
timeout_minutes: ${{ inputs.upload-retry-timeout }}
# The bucket layout will be:
# releases/greptimedb
# ├── v0.1.0
# │ ├── greptime-darwin-amd64-pyo3-v0.1.0.sha256sum
# │ └── greptime-darwin-amd64-pyo3-v0.1.0.tar.gz
# └── v0.2.0
# ├── greptime-darwin-amd64-pyo3-v0.2.0.sha256sum
# └── greptime-darwin-amd64-pyo3-v0.2.0.tar.gz
command: |
cd ${{ inputs.working-dir }} && \
aws s3 cp \
${{ inputs.artifacts-dir }}.tar.gz \
s3://${{ inputs.release-to-s3-bucket }}/releases/greptimedb/${{ inputs.version }}/${{ inputs.artifacts-dir }}.tar.gz && \
aws s3 cp \
${{ inputs.artifacts-dir }}.sha256sum \
s3://${{ inputs.release-to-s3-bucket }}/releases/greptimedb/${{ inputs.version }}/${{ inputs.artifacts-dir }}.sha256sum
- name: Upload latest artifacts to S3
if: ${{ inputs.upload-to-s3 == 'true' && inputs.upload-latest-artifacts == 'true' }} # We'll also upload the latest artifacts to S3 in the scheduled and formal release.
uses: nick-invision/retry@v2
env:
AWS_ACCESS_KEY_ID: ${{ inputs.aws-access-key-id }}
AWS_SECRET_ACCESS_KEY: ${{ inputs.aws-secret-access-key }}
AWS_DEFAULT_REGION: ${{ inputs.aws-region }}
with:
max_attempts: ${{ inputs.upload-max-retry-times }}
timeout_minutes: ${{ inputs.upload-retry-timeout }}
command: |
cd ${{ inputs.working-dir }} && \
aws s3 cp \
${{ inputs.artifacts-dir }}.tar.gz \
s3://${{ inputs.release-to-s3-bucket }}/releases/greptimedb/latest/${{ inputs.artifacts-dir }}.tar.gz && \
aws s3 cp \
${{ inputs.artifacts-dir }}.sha256sum \
s3://${{ inputs.release-to-s3-bucket }}/releases/greptimedb/latest/${{ inputs.artifacts-dir }}.sha256sum

View File

@@ -1,47 +0,0 @@
#!/usr/bin/env bash
set -e
set -o pipefail
SRC_IMAGE=$1
DST_REGISTRY=$2
SKOPEO_STABLE_IMAGE="quay.io/skopeo/stable:latest"
# Check if necessary variables are set.
function check_vars() {
for var in DST_REGISTRY_USERNAME DST_REGISTRY_PASSWORD DST_REGISTRY SRC_IMAGE; do
if [ -z "${!var}" ]; then
echo "$var is not set or empty."
echo "Usage: DST_REGISTRY_USERNAME=<your-dst-registry-username> DST_REGISTRY_PASSWORD=<your-dst-registry-password> $0 <dst-registry> <src-image>"
exit 1
fi
done
}
# Copies images from DockerHub to the destination registry.
function copy_images_from_dockerhub() {
# Check if docker is installed.
if ! command -v docker &> /dev/null; then
echo "docker is not installed. Please install docker to continue."
exit 1
fi
# Extract the name and tag of the source image.
IMAGE_NAME=$(echo "$SRC_IMAGE" | sed "s/.*\///")
echo "Copying $SRC_IMAGE to $DST_REGISTRY/$IMAGE_NAME"
docker run "$SKOPEO_STABLE_IMAGE" copy -a docker://"$SRC_IMAGE" \
--dest-creds "$DST_REGISTRY_USERNAME":"$DST_REGISTRY_PASSWORD" \
docker://"$DST_REGISTRY/$IMAGE_NAME"
}
function main() {
check_vars
copy_images_from_dockerhub
}
# Usage example:
# DST_REGISTRY_USERNAME=123 DST_REGISTRY_PASSWORD=456 \
# ./copy-image.sh greptime/greptimedb:v0.4.0 greptime-registry.cn-hangzhou.cr.aliyuncs.com
main

View File

@@ -1,102 +0,0 @@
#!/usr/bin/env bash
set -e
set -o pipefail
ARTIFACTS_DIR=$1
VERSION=$2
AWS_S3_BUCKET=$3
RELEASE_DIRS="releases/greptimedb"
GREPTIMEDB_REPO="GreptimeTeam/greptimedb"
# Check if necessary variables are set.
function check_vars() {
for var in AWS_S3_BUCKET VERSION ARTIFACTS_DIR; do
if [ -z "${!var}" ]; then
echo "$var is not set or empty."
echo "Usage: $0 <artifacts-dir> <version> <aws-s3-bucket>"
exit 1
fi
done
}
# Uploads artifacts to AWS S3 bucket.
function upload_artifacts() {
# The bucket layout will be:
# releases/greptimedb
# ├── latest-version.txt
# ├── latest-nightly-version.txt
# ├── v0.1.0
# │ ├── greptime-darwin-amd64-pyo3-v0.1.0.sha256sum
# │ └── greptime-darwin-amd64-pyo3-v0.1.0.tar.gz
# └── v0.2.0
# ├── greptime-darwin-amd64-pyo3-v0.2.0.sha256sum
# └── greptime-darwin-amd64-pyo3-v0.2.0.tar.gz
find "$ARTIFACTS_DIR" -type f \( -name "*.tar.gz" -o -name "*.sha256sum" \) | while IFS= read -r file; do
aws s3 cp \
"$file" "s3://$AWS_S3_BUCKET/$RELEASE_DIRS/$VERSION/$(basename "$file")"
done
}
# Updates the latest version information in AWS S3 if UPDATE_VERSION_INFO is true.
function update_version_info() {
if [ "$UPDATE_VERSION_INFO" == "true" ]; then
# If it's the officail release(like v1.0.0, v1.0.1, v1.0.2, etc.), update latest-version.txt.
if [[ "$VERSION" =~ ^v[0-9]+\.[0-9]+\.[0-9]+$ ]]; then
echo "Updating latest-version.txt"
echo "$VERSION" > latest-version.txt
aws s3 cp \
latest-version.txt "s3://$AWS_S3_BUCKET/$RELEASE_DIRS/latest-version.txt"
fi
# If it's the nightly release, update latest-nightly-version.txt.
if [[ "$VERSION" == *"nightly"* ]]; then
echo "Updating latest-nightly-version.txt"
echo "$VERSION" > latest-nightly-version.txt
aws s3 cp \
latest-nightly-version.txt "s3://$AWS_S3_BUCKET/$RELEASE_DIRS/latest-nightly-version.txt"
fi
fi
}
# Downloads artifacts from Github if DOWNLOAD_ARTIFACTS_FROM_GITHUB is true.
function download_artifacts_from_github() {
if [ "$DOWNLOAD_ARTIFACTS_FROM_GITHUB" == "true" ]; then
# Check if jq is installed.
if ! command -v jq &> /dev/null; then
echo "jq is not installed. Please install jq to continue."
exit 1
fi
# Get the latest release API response.
RELEASES_API_RESPONSE=$(curl -s -H "Accept: application/vnd.github.v3+json" "https://api.github.com/repos/$GREPTIMEDB_REPO/releases/latest")
# Extract download URLs for the artifacts.
# Exclude source code archives which are typically named as 'greptimedb-<version>.zip' or 'greptimedb-<version>.tar.gz'.
ASSET_URLS=$(echo "$RELEASES_API_RESPONSE" | jq -r '.assets[] | select(.name | test("greptimedb-.*\\.(zip|tar\\.gz)$") | not) | .browser_download_url')
# Download each asset.
while IFS= read -r url; do
if [ -n "$url" ]; then
curl -LJO "$url"
echo "Downloaded: $url"
fi
done <<< "$ASSET_URLS"
fi
}
function main() {
check_vars
download_artifacts_from_github
upload_artifacts
update_version_info
}
# Usage example:
# AWS_ACCESS_KEY_ID=<your_access_key_id> \
# AWS_SECRET_ACCESS_KEY=<your_secret_access_key> \
# AWS_DEFAULT_REGION=<your_region> \
# UPDATE_VERSION_INFO=true \
# DOWNLOAD_ARTIFACTS_FROM_GITHUB=false \
# ./upload-artifacts-to-s3.sh <artifacts-dir> <version> <aws-s3-bucket>
main

View File

@@ -17,7 +17,7 @@ env:
jobs:
apidoc:
runs-on: ubuntu-20.04
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v3
- uses: arduino/setup-protoc@v1

View File

@@ -16,11 +16,11 @@ on:
description: The runner uses to build linux-amd64 artifacts
default: ec2-c6i.4xlarge-amd64
options:
- ubuntu-20.04
- ubuntu-20.04-8-cores
- ubuntu-20.04-16-cores
- ubuntu-20.04-32-cores
- ubuntu-20.04-64-cores
- ubuntu-latest
- ubuntu-latest-8-cores
- ubuntu-latest-16-cores
- ubuntu-latest-32-cores
- ubuntu-latest-64-cores
- ec2-c6i.xlarge-amd64 # 4C8G
- ec2-c6i.2xlarge-amd64 # 8C16G
- ec2-c6i.4xlarge-amd64 # 16C32G
@@ -78,7 +78,7 @@ jobs:
allocate-runners:
name: Allocate runners
if: ${{ github.repository == 'GreptimeTeam/greptimedb' }}
runs-on: ubuntu-20.04
runs-on: ubuntu-latest
outputs:
linux-amd64-runner: ${{ steps.start-linux-amd64-runner.outputs.label }}
linux-arm64-runner: ${{ steps.start-linux-arm64-runner.outputs.label }}
@@ -164,7 +164,12 @@ jobs:
cargo-profile: ${{ env.CARGO_PROFILE }}
version: ${{ needs.allocate-runners.outputs.version }}
disable-run-tests: ${{ env.DISABLE_RUN_TESTS }}
release-to-s3-bucket: ${{ vars.AWS_RELEASE_BUCKET }}
aws-access-key-id: ${{ secrets.AWS_CN_ACCESS_KEY_ID }}
aws-secret-access-key: ${{ secrets.AWS_CN_SECRET_ACCESS_KEY }}
aws-region: ${{ vars.AWS_RELEASE_BUCKET_REGION }}
dev-mode: true # Only build the standard greptime binary.
upload-to-s3: false # No need to upload to S3.
working-dir: ${{ env.CHECKOUT_GREPTIMEDB_PATH }}
build-linux-arm64-artifacts:
@@ -193,7 +198,12 @@ jobs:
cargo-profile: ${{ env.CARGO_PROFILE }}
version: ${{ needs.allocate-runners.outputs.version }}
disable-run-tests: ${{ env.DISABLE_RUN_TESTS }}
release-to-s3-bucket: ${{ vars.AWS_RELEASE_BUCKET }}
aws-access-key-id: ${{ secrets.AWS_CN_ACCESS_KEY_ID }}
aws-secret-access-key: ${{ secrets.AWS_CN_SECRET_ACCESS_KEY }}
aws-region: ${{ vars.AWS_RELEASE_BUCKET_REGION }}
dev-mode: true # Only build the standard greptime binary.
upload-to-s3: false # No need to upload to S3.
working-dir: ${{ env.CHECKOUT_GREPTIMEDB_PATH }}
release-images-to-dockerhub:
@@ -204,7 +214,7 @@ jobs:
build-linux-amd64-artifacts,
build-linux-arm64-artifacts,
]
runs-on: ubuntu-20.04
runs-on: ubuntu-latest
outputs:
build-result: ${{ steps.set-build-result.outputs.build-result }}
steps:
@@ -229,44 +239,41 @@ jobs:
run: |
echo "build-result=success" >> $GITHUB_OUTPUT
release-cn-artifacts:
name: Release artifacts to CN region
release-images-to-acr:
name: Build and push images to ACR
if: ${{ inputs.release_images || github.event_name == 'schedule' }}
needs: [
allocate-runners,
release-images-to-dockerhub,
build-linux-amd64-artifacts,
build-linux-arm64-artifacts,
]
runs-on: ubuntu-20.04
runs-on: ubuntu-latest
# When we push to ACR, it's easy to fail due to some unknown network issues.
# However, we don't want to fail the whole workflow because of this.
# The ACR have daily sync with DockerHub, so don't worry about the image not being updated.
continue-on-error: true
steps:
- uses: actions/checkout@v3
with:
fetch-depth: 0
- name: Release artifacts to CN region
uses: ./.github/actions/release-cn-artifacts
- name: Build and push images to ACR
uses: ./.github/actions/build-images
with:
src-image-registry: docker.io
src-image-namespace: ${{ vars.IMAGE_NAMESPACE }}
src-image-name: ${{ env.IMAGE_NAME }}
dst-image-registry-username: ${{ secrets.ALICLOUD_USERNAME }}
dst-image-registry-password: ${{ secrets.ALICLOUD_PASSWORD }}
dst-image-registry: ${{ vars.ACR_IMAGE_REGISTRY }}
dst-image-namespace: ${{ vars.IMAGE_NAMESPACE }}
image-registry: ${{ vars.ACR_IMAGE_REGISTRY }}
image-namespace: ${{ vars.IMAGE_NAMESPACE }}
image-name: ${{ env.IMAGE_NAME }}
image-registry-username: ${{ secrets.ALICLOUD_USERNAME }}
image-registry-password: ${{ secrets.ALICLOUD_PASSWORD }}
version: ${{ needs.allocate-runners.outputs.version }}
aws-cn-s3-bucket: ${{ vars.AWS_RELEASE_BUCKET }}
aws-cn-access-key-id: ${{ secrets.AWS_CN_ACCESS_KEY_ID }}
aws-cn-secret-access-key: ${{ secrets.AWS_CN_SECRET_ACCESS_KEY }}
aws-cn-region: ${{ vars.AWS_RELEASE_BUCKET_REGION }}
dev-mode: true # Only build the standard images(exclude centos images).
push-latest-tag: false # Don't push the latest tag to registry.
update-version-info: false # Don't update the version info in S3.
push-latest-tag: false # Don't push the latest tag to registry.
dev-mode: true # Only build the standard images.
stop-linux-amd64-runner: # It's always run as the last job in the workflow to make sure that the runner is released.
name: Stop linux-amd64 runner
# Only run this job when the runner is allocated.
if: ${{ always() }}
runs-on: ubuntu-20.04
runs-on: ubuntu-latest
needs: [
allocate-runners,
build-linux-amd64-artifacts,
@@ -291,7 +298,7 @@ jobs:
name: Stop linux-arm64 runner
# Only run this job when the runner is allocated.
if: ${{ always() }}
runs-on: ubuntu-20.04
runs-on: ubuntu-latest
needs: [
allocate-runners,
build-linux-arm64-artifacts,
@@ -318,7 +325,7 @@ jobs:
needs: [
release-images-to-dockerhub
]
runs-on: ubuntu-20.04
runs-on: ubuntu-latest
env:
SLACK_WEBHOOK_URL: ${{ secrets.SLACK_WEBHOOK_URL_DEVELOP_CHANNEL }}
steps:

View File

@@ -34,7 +34,7 @@ env:
jobs:
typos:
name: Spell Check with Typos
runs-on: ubuntu-20.04
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v3
- uses: crate-ci/typos@v1.13.10
@@ -42,7 +42,7 @@ jobs:
check:
name: Check
if: github.event.pull_request.draft == false
runs-on: ubuntu-20.04
runs-on: ubuntu-latest
timeout-minutes: 60
steps:
- uses: actions/checkout@v3
@@ -60,7 +60,7 @@ jobs:
toml:
name: Toml Check
if: github.event.pull_request.draft == false
runs-on: ubuntu-20.04
runs-on: ubuntu-latest
timeout-minutes: 60
steps:
- uses: actions/checkout@v3
@@ -80,7 +80,7 @@ jobs:
runs-on: ${{ matrix.os }}
strategy:
matrix:
os: [ ubuntu-20.04-8-cores ]
os: [ ubuntu-latest-8-cores, windows-latest-8-cores ]
timeout-minutes: 60
steps:
- uses: actions/checkout@v3
@@ -105,7 +105,7 @@ jobs:
fmt:
name: Rustfmt
if: github.event.pull_request.draft == false
runs-on: ubuntu-20.04
runs-on: ubuntu-latest
timeout-minutes: 60
steps:
- uses: actions/checkout@v3
@@ -124,7 +124,7 @@ jobs:
clippy:
name: Clippy
if: github.event.pull_request.draft == false
runs-on: ubuntu-20.04
runs-on: ubuntu-latest
timeout-minutes: 60
steps:
- uses: actions/checkout@v3
@@ -142,7 +142,7 @@ jobs:
coverage:
if: github.event.pull_request.draft == false
runs-on: ubuntu-20.04-8-cores
runs-on: ubuntu-latest-8-cores
timeout-minutes: 60
steps:
- uses: actions/checkout@v3
@@ -188,3 +188,43 @@ jobs:
flags: rust
fail_ci_if_error: false
verbose: true
test-on-windows:
if: github.event.pull_request.draft == false
runs-on: windows-latest-8-cores
timeout-minutes: 60
steps:
- run: git config --global core.autocrlf false
- uses: actions/checkout@v3
- uses: arduino/setup-protoc@v1
with:
repo-token: ${{ secrets.GITHUB_TOKEN }}
- name: Install Rust toolchain
uses: dtolnay/rust-toolchain@master
with:
toolchain: ${{ env.RUST_TOOLCHAIN }}
components: llvm-tools-preview
- name: Rust Cache
uses: Swatinem/rust-cache@v2
- name: Install Cargo Nextest
uses: taiki-e/install-action@nextest
- name: Install Python
uses: actions/setup-python@v4
with:
python-version: '3.10'
- name: Install PyArrow Package
run: pip install pyarrow
- name: Install WSL distribution
uses: Vampire/setup-wsl@v2
with:
distribution: Ubuntu-22.04
- name: Running tests
run: cargo nextest run -F pyo3_backend,dashboard
env:
RUST_BACKTRACE: 1
CARGO_INCREMENTAL: 0
GT_S3_BUCKET: ${{ secrets.S3_BUCKET }}
GT_S3_ACCESS_KEY_ID: ${{ secrets.S3_ACCESS_KEY_ID }}
GT_S3_ACCESS_KEY: ${{ secrets.S3_ACCESS_KEY }}
GT_S3_REGION: ${{ secrets.S3_REGION }}
UNITTEST_LOG_DIR: "__unittest_logs"

View File

@@ -11,7 +11,7 @@ on:
jobs:
doc_issue:
if: github.event.label.name == 'doc update required'
runs-on: ubuntu-20.04
runs-on: ubuntu-latest
steps:
- name: create an issue in doc repo
uses: dacbd/create-issue-action@main
@@ -25,7 +25,7 @@ jobs:
${{ github.event.issue.html_url || github.event.pull_request.html_url }}
cloud_issue:
if: github.event.label.name == 'cloud followup required'
runs-on: ubuntu-20.04
runs-on: ubuntu-latest
steps:
- name: create an issue in cloud repo
uses: dacbd/create-issue-action@main

View File

@@ -30,7 +30,7 @@ name: CI
jobs:
typos:
name: Spell Check with Typos
runs-on: ubuntu-20.04
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v3
- uses: crate-ci/typos@v1.13.10
@@ -38,33 +38,33 @@ jobs:
check:
name: Check
if: github.event.pull_request.draft == false
runs-on: ubuntu-20.04
runs-on: ubuntu-latest
steps:
- run: 'echo "No action required"'
fmt:
name: Rustfmt
if: github.event.pull_request.draft == false
runs-on: ubuntu-20.04
runs-on: ubuntu-latest
steps:
- run: 'echo "No action required"'
clippy:
name: Clippy
if: github.event.pull_request.draft == false
runs-on: ubuntu-20.04
runs-on: ubuntu-latest
steps:
- run: 'echo "No action required"'
coverage:
if: github.event.pull_request.draft == false
runs-on: ubuntu-20.04
runs-on: ubuntu-latest
steps:
- run: 'echo "No action required"'
sqlness:
name: Sqlness Test
if: github.event.pull_request.draft == false
runs-on: ubuntu-20.04
runs-on: ubuntu-latest
steps:
- run: 'echo "No action required"'

View File

@@ -8,7 +8,7 @@ on:
types: [opened, synchronize, reopened, ready_for_review]
jobs:
license-header-check:
runs-on: ubuntu-20.04
runs-on: ubuntu-latest
name: license-header-check
steps:
- uses: actions/checkout@v2

View File

@@ -14,11 +14,11 @@ on:
description: The runner uses to build linux-amd64 artifacts
default: ec2-c6i.2xlarge-amd64
options:
- ubuntu-20.04
- ubuntu-20.04-8-cores
- ubuntu-20.04-16-cores
- ubuntu-20.04-32-cores
- ubuntu-20.04-64-cores
- ubuntu-latest
- ubuntu-latest-8-cores
- ubuntu-latest-16-cores
- ubuntu-latest-32-cores
- ubuntu-latest-64-cores
- ec2-c6i.xlarge-amd64 # 4C8G
- ec2-c6i.2xlarge-amd64 # 8C16G
- ec2-c6i.4xlarge-amd64 # 16C32G
@@ -70,7 +70,7 @@ jobs:
allocate-runners:
name: Allocate runners
if: ${{ github.repository == 'GreptimeTeam/greptimedb' }}
runs-on: ubuntu-20.04
runs-on: ubuntu-latest
outputs:
linux-amd64-runner: ${{ steps.start-linux-amd64-runner.outputs.label }}
linux-arm64-runner: ${{ steps.start-linux-arm64-runner.outputs.label }}
@@ -147,6 +147,11 @@ jobs:
cargo-profile: ${{ env.CARGO_PROFILE }}
version: ${{ needs.allocate-runners.outputs.version }}
disable-run-tests: ${{ env.DISABLE_RUN_TESTS }}
release-to-s3-bucket: ${{ vars.AWS_RELEASE_BUCKET }}
aws-access-key-id: ${{ secrets.AWS_CN_ACCESS_KEY_ID }}
aws-secret-access-key: ${{ secrets.AWS_CN_SECRET_ACCESS_KEY }}
aws-region: ${{ vars.AWS_RELEASE_BUCKET_REGION }}
upload-latest-artifacts: false
build-linux-arm64-artifacts:
name: Build linux-arm64 artifacts
@@ -166,6 +171,11 @@ jobs:
cargo-profile: ${{ env.CARGO_PROFILE }}
version: ${{ needs.allocate-runners.outputs.version }}
disable-run-tests: ${{ env.DISABLE_RUN_TESTS }}
release-to-s3-bucket: ${{ vars.AWS_RELEASE_BUCKET }}
aws-access-key-id: ${{ secrets.AWS_CN_ACCESS_KEY_ID }}
aws-secret-access-key: ${{ secrets.AWS_CN_SECRET_ACCESS_KEY }}
aws-region: ${{ vars.AWS_RELEASE_BUCKET_REGION }}
upload-latest-artifacts: false
release-images-to-dockerhub:
name: Build and push images to DockerHub
@@ -175,7 +185,7 @@ jobs:
build-linux-amd64-artifacts,
build-linux-arm64-artifacts,
]
runs-on: ubuntu-20.04
runs-on: ubuntu-latest
outputs:
nightly-build-result: ${{ steps.set-nightly-build-result.outputs.nightly-build-result }}
steps:
@@ -198,14 +208,15 @@ jobs:
run: |
echo "nightly-build-result=success" >> $GITHUB_OUTPUT
release-cn-artifacts:
name: Release artifacts to CN region
release-images-to-acr:
name: Build and push images to ACR
if: ${{ inputs.release_images || github.event_name == 'schedule' }}
needs: [
allocate-runners,
release-images-to-dockerhub,
build-linux-amd64-artifacts,
build-linux-arm64-artifacts,
]
runs-on: ubuntu-20.04
runs-on: ubuntu-latest
# When we push to ACR, it's easy to fail due to some unknown network issues.
# However, we don't want to fail the whole workflow because of this.
# The ACR have daily sync with DockerHub, so don't worry about the image not being updated.
@@ -215,30 +226,21 @@ jobs:
with:
fetch-depth: 0
- name: Release artifacts to CN region
uses: ./.github/actions/release-cn-artifacts
- name: Build and push images to ACR
uses: ./.github/actions/build-images
with:
src-image-registry: docker.io
src-image-namespace: ${{ vars.IMAGE_NAMESPACE }}
src-image-name: greptimedb
dst-image-registry-username: ${{ secrets.ALICLOUD_USERNAME }}
dst-image-registry-password: ${{ secrets.ALICLOUD_PASSWORD }}
dst-image-registry: ${{ vars.ACR_IMAGE_REGISTRY }}
dst-image-namespace: ${{ vars.IMAGE_NAMESPACE }}
image-registry: ${{ vars.ACR_IMAGE_REGISTRY }}
image-namespace: ${{ vars.IMAGE_NAMESPACE }}
image-registry-username: ${{ secrets.ALICLOUD_USERNAME }}
image-registry-password: ${{ secrets.ALICLOUD_PASSWORD }}
version: ${{ needs.allocate-runners.outputs.version }}
aws-cn-s3-bucket: ${{ vars.AWS_RELEASE_BUCKET }}
aws-cn-access-key-id: ${{ secrets.AWS_CN_ACCESS_KEY_ID }}
aws-cn-secret-access-key: ${{ secrets.AWS_CN_SECRET_ACCESS_KEY }}
aws-cn-region: ${{ vars.AWS_RELEASE_BUCKET_REGION }}
dev-mode: false
update-version-info: false # Don't update version info in S3.
push-latest-tag: false # Don't push the latest tag to registry.
push-latest-tag: false # Don't push the latest tag to registry.
stop-linux-amd64-runner: # It's always run as the last job in the workflow to make sure that the runner is released.
name: Stop linux-amd64 runner
# Only run this job when the runner is allocated.
if: ${{ always() }}
runs-on: ubuntu-20.04
runs-on: ubuntu-latest
needs: [
allocate-runners,
build-linux-amd64-artifacts,
@@ -263,7 +265,7 @@ jobs:
name: Stop linux-arm64 runner
# Only run this job when the runner is allocated.
if: ${{ always() }}
runs-on: ubuntu-20.04
runs-on: ubuntu-latest
needs: [
allocate-runners,
build-linux-arm64-artifacts,
@@ -290,7 +292,7 @@ jobs:
needs: [
release-images-to-dockerhub
]
runs-on: ubuntu-20.04
runs-on: ubuntu-latest
env:
SLACK_WEBHOOK_URL: ${{ secrets.SLACK_WEBHOOK_URL_DEVELOP_CHANNEL }}
steps:

View File

@@ -1,98 +0,0 @@
# Nightly CI: runs tests every night for our second tier plaforms (Windows)
on:
schedule:
- cron: '0 23 * * 1-5'
workflow_dispatch:
name: Nightly CI
concurrency:
group: ${{ github.workflow }}-${{ github.head_ref || github.run_id }}
cancel-in-progress: true
env:
RUST_TOOLCHAIN: nightly-2023-08-07
jobs:
sqlness:
name: Sqlness Test
runs-on: ${{ matrix.os }}
strategy:
matrix:
os: [ windows-latest-8-cores ]
timeout-minutes: 60
steps:
- uses: actions/checkout@v4.1.0
- uses: arduino/setup-protoc@v1
with:
repo-token: ${{ secrets.GITHUB_TOKEN }}
- uses: dtolnay/rust-toolchain@master
with:
toolchain: ${{ env.RUST_TOOLCHAIN }}
- name: Rust Cache
uses: Swatinem/rust-cache@v2
- name: Run sqlness
run: cargo sqlness
- name: Notify slack if failed
if: failure()
uses: slackapi/slack-github-action@v1.23.0
env:
SLACK_WEBHOOK_URL: ${{ secrets.SLACK_WEBHOOK_URL_DEVELOP_CHANNEL }}
with:
payload: |
{"text": "Nightly CI failed for sqlness tests"}
- name: Upload sqlness logs
if: always()
uses: actions/upload-artifact@v3
with:
name: sqlness-logs
path: ${{ runner.temp }}/greptime-*.log
retention-days: 3
test-on-windows:
runs-on: windows-latest-8-cores
timeout-minutes: 60
steps:
- run: git config --global core.autocrlf false
- uses: actions/checkout@v4.1.0
- uses: arduino/setup-protoc@v1
with:
repo-token: ${{ secrets.GITHUB_TOKEN }}
- name: Install Rust toolchain
uses: dtolnay/rust-toolchain@master
with:
toolchain: ${{ env.RUST_TOOLCHAIN }}
components: llvm-tools-preview
- name: Rust Cache
uses: Swatinem/rust-cache@v2
- name: Install Cargo Nextest
uses: taiki-e/install-action@nextest
- name: Install Python
uses: actions/setup-python@v4
with:
python-version: '3.10'
- name: Install PyArrow Package
run: pip install pyarrow
- name: Install WSL distribution
uses: Vampire/setup-wsl@v2
with:
distribution: Ubuntu-22.04
- name: Running tests
run: cargo nextest run -F pyo3_backend,dashboard
env:
RUST_BACKTRACE: 1
CARGO_INCREMENTAL: 0
GT_S3_BUCKET: ${{ secrets.S3_BUCKET }}
GT_S3_ACCESS_KEY_ID: ${{ secrets.S3_ACCESS_KEY_ID }}
GT_S3_ACCESS_KEY: ${{ secrets.S3_ACCESS_KEY }}
GT_S3_REGION: ${{ secrets.S3_REGION }}
UNITTEST_LOG_DIR: "__unittest_logs"
- name: Notify slack if failed
if: failure()
uses: slackapi/slack-github-action@v1.23.0
env:
SLACK_WEBHOOK_URL: ${{ secrets.SLACK_WEBHOOK_URL_DEVELOP_CHANNEL }}
with:
payload: |
{"text": "Nightly CI failed for cargo test"}

View File

@@ -10,7 +10,7 @@ on:
jobs:
check:
runs-on: ubuntu-20.04
runs-on: ubuntu-latest
timeout-minutes: 10
steps:
- uses: thehanimo/pr-title-checker@v1.3.4
@@ -19,7 +19,7 @@ jobs:
pass_on_octokit_error: false
configuration_path: ".github/pr-title-checker-config.json"
breaking:
runs-on: ubuntu-20.04
runs-on: ubuntu-latest
timeout-minutes: 10
steps:
- uses: thehanimo/pr-title-checker@v1.3.4

View File

@@ -1,85 +0,0 @@
name: Release dev-builder images
on:
workflow_dispatch: # Allows you to run this workflow manually.
inputs:
version:
description: Version of the dev-builder
required: false
default: latest
release_dev_builder_ubuntu_image:
type: boolean
description: Release dev-builder-ubuntu image
required: false
default: false
release_dev_builder_centos_image:
type: boolean
description: Release dev-builder-centos image
required: false
default: false
release_dev_builder_android_image:
type: boolean
description: Release dev-builder-android image
required: false
default: false
jobs:
release-dev-builder-images:
name: Release dev builder images
if: ${{ inputs.release_dev_builder_ubuntu_image || inputs.release_dev_builder_centos_image || inputs.release_dev_builder_android_image }} # Only manually trigger this job.
runs-on: ubuntu-20.04-16-cores
steps:
- name: Checkout
uses: actions/checkout@v3
with:
fetch-depth: 0
- name: Build and push dev builder images
uses: ./.github/actions/build-dev-builder-images
with:
version: ${{ inputs.version }}
dockerhub-image-registry-username: ${{ secrets.DOCKERHUB_USERNAME }}
dockerhub-image-registry-token: ${{ secrets.DOCKERHUB_TOKEN }}
build-dev-builder-ubuntu: ${{ inputs.release_dev_builder_ubuntu_image }}
build-dev-builder-centos: ${{ inputs.release_dev_builder_centos_image }}
build-dev-builder-android: ${{ inputs.release_dev_builder_android_image }}
release-dev-builder-images-cn: # Note: Be careful issue: https://github.com/containers/skopeo/issues/1874 and we decide to use the latest stable skopeo container.
name: Release dev builder images to CN region
runs-on: ubuntu-20.04
needs: [
release-dev-builder-images
]
steps:
- name: Push dev-builder-ubuntu image
shell: bash
if: ${{ inputs.release_dev_builder_ubuntu_image }}
env:
DST_REGISTRY_USERNAME: ${{ secrets.ALICLOUD_USERNAME }}
DST_REGISTRY_PASSWORD: ${{ secrets.ALICLOUD_PASSWORD }}
run: |
docker run quay.io/skopeo/stable:latest copy -a docker://docker.io/${{ vars.IMAGE_NAMESPACE }}/dev-builder-ubuntu:${{ inputs.version }} \
--dest-creds "$DST_REGISTRY_USERNAME":"$DST_REGISTRY_PASSWORD" \
docker://${{ vars.ACR_IMAGE_REGISTRY }}/${{ vars.IMAGE_NAMESPACE }}/dev-builder-ubuntu:${{ inputs.version }}
- name: Push dev-builder-centos image
shell: bash
if: ${{ inputs.release_dev_builder_centos_image }}
env:
DST_REGISTRY_USERNAME: ${{ secrets.ALICLOUD_USERNAME }}
DST_REGISTRY_PASSWORD: ${{ secrets.ALICLOUD_PASSWORD }}
run: |
docker run quay.io/skopeo/stable:latest copy -a docker://docker.io/${{ vars.IMAGE_NAMESPACE }}/dev-builder-centos:${{ inputs.version }} \
--dest-creds "$DST_REGISTRY_USERNAME":"$DST_REGISTRY_PASSWORD" \
docker://${{ vars.ACR_IMAGE_REGISTRY }}/${{ vars.IMAGE_NAMESPACE }}/dev-builder-centos:${{ inputs.version }}
- name: Push dev-builder-android image
shell: bash
if: ${{ inputs.release_dev_builder_android_image }}
env:
DST_REGISTRY_USERNAME: ${{ secrets.ALICLOUD_USERNAME }}
DST_REGISTRY_PASSWORD: ${{ secrets.ALICLOUD_PASSWORD }}
run: |
docker run quay.io/skopeo/stable:latest copy -a docker://docker.io/${{ vars.IMAGE_NAMESPACE }}/dev-builder-android:${{ inputs.version }} \
--dest-creds "$DST_REGISTRY_USERNAME":"$DST_REGISTRY_PASSWORD" \
docker://${{ vars.ACR_IMAGE_REGISTRY }}/${{ vars.IMAGE_NAMESPACE }}/dev-builder-android:${{ inputs.version }}

View File

@@ -18,11 +18,11 @@ on:
description: The runner uses to build linux-amd64 artifacts
default: ec2-c6i.4xlarge-amd64
options:
- ubuntu-20.04
- ubuntu-20.04-8-cores
- ubuntu-20.04-16-cores
- ubuntu-20.04-32-cores
- ubuntu-20.04-64-cores
- ubuntu-latest
- ubuntu-latest-8-cores
- ubuntu-latest-16-cores
- ubuntu-latest-32-cores
- ubuntu-latest-64-cores
- ec2-c6i.xlarge-amd64 # 4C8G
- ec2-c6i.2xlarge-amd64 # 8C16G
- ec2-c6i.4xlarge-amd64 # 16C32G
@@ -63,12 +63,7 @@ on:
description: Build macos artifacts
required: false
default: false
build_windows_artifacts:
type: boolean
description: Build Windows artifacts
required: false
default: false
publish_github_release:
release_artifacts:
type: boolean
description: Create GitHub release and upload artifacts
required: false
@@ -78,6 +73,11 @@ on:
description: Build and push images to DockerHub and ACR
required: false
default: false
release_dev_builder_image:
type: boolean
description: Release dev-builder image
required: false
default: false
# Use env variables to control all the release process.
env:
@@ -91,18 +91,17 @@ env:
# The scheduled version is '${{ env.NEXT_RELEASE_VERSION }}-nightly-YYYYMMDD', like v0.2.0-nigthly-20230313;
NIGHTLY_RELEASE_PREFIX: nightly
# Note: The NEXT_RELEASE_VERSION should be modified manually by every formal release.
NEXT_RELEASE_VERSION: v0.5.0
NEXT_RELEASE_VERSION: v0.4.0
jobs:
allocate-runners:
name: Allocate runners
if: ${{ github.repository == 'GreptimeTeam/greptimedb' }}
runs-on: ubuntu-20.04
runs-on: ubuntu-latest
outputs:
linux-amd64-runner: ${{ steps.start-linux-amd64-runner.outputs.label }}
linux-arm64-runner: ${{ steps.start-linux-arm64-runner.outputs.label }}
macos-runner: ${{ inputs.macos_runner || vars.DEFAULT_MACOS_RUNNER }}
windows-runner: windows-latest-8-cores
# The following EC2 resource id will be used for resource releasing.
linux-amd64-ec2-runner-label: ${{ steps.start-linux-amd64-runner.outputs.label }}
@@ -178,6 +177,10 @@ jobs:
cargo-profile: ${{ env.CARGO_PROFILE }}
version: ${{ needs.allocate-runners.outputs.version }}
disable-run-tests: ${{ env.DISABLE_RUN_TESTS }}
release-to-s3-bucket: ${{ vars.AWS_RELEASE_BUCKET }}
aws-access-key-id: ${{ secrets.AWS_CN_ACCESS_KEY_ID }}
aws-secret-access-key: ${{ secrets.AWS_CN_SECRET_ACCESS_KEY }}
aws-region: ${{ vars.AWS_RELEASE_BUCKET_REGION }}
build-linux-arm64-artifacts:
name: Build linux-arm64 artifacts
@@ -197,6 +200,10 @@ jobs:
cargo-profile: ${{ env.CARGO_PROFILE }}
version: ${{ needs.allocate-runners.outputs.version }}
disable-run-tests: ${{ env.DISABLE_RUN_TESTS }}
release-to-s3-bucket: ${{ vars.AWS_RELEASE_BUCKET }}
aws-access-key-id: ${{ secrets.AWS_CN_ACCESS_KEY_ID }}
aws-secret-access-key: ${{ secrets.AWS_CN_SECRET_ACCESS_KEY }}
aws-region: ${{ vars.AWS_RELEASE_BUCKET_REGION }}
build-macos-artifacts:
name: Build macOS artifacts
@@ -238,43 +245,11 @@ jobs:
features: ${{ matrix.features }}
version: ${{ needs.allocate-runners.outputs.version }}
disable-run-tests: ${{ env.DISABLE_RUN_TESTS }}
release-to-s3-bucket: ${{ vars.AWS_RELEASE_BUCKET }}
artifacts-dir: ${{ matrix.artifacts-dir-prefix }}-${{ needs.allocate-runners.outputs.version }}
build-windows-artifacts:
name: Build Windows artifacts
strategy:
fail-fast: false
matrix:
include:
- os: ${{ needs.allocate-runners.outputs.windows-runner }}
arch: x86_64-pc-windows-msvc
features: servers/dashboard
artifacts-dir-prefix: greptime-windows-amd64
- os: ${{ needs.allocate-runners.outputs.windows-runner }}
arch: x86_64-pc-windows-msvc
features: pyo3_backend,servers/dashboard
artifacts-dir-prefix: greptime-windows-amd64-pyo3
runs-on: ${{ matrix.os }}
needs: [
allocate-runners,
]
if: ${{ inputs.build_windows_artifacts || github.event_name == 'push' || github.event_name == 'schedule' }}
steps:
- run: git config --global core.autocrlf false
- uses: actions/checkout@v3
with:
fetch-depth: 0
- uses: ./.github/actions/build-windows-artifacts
with:
arch: ${{ matrix.arch }}
rust-toolchain: ${{ env.RUST_TOOLCHAIN }}
cargo-profile: ${{ env.CARGO_PROFILE }}
features: ${{ matrix.features }}
version: ${{ needs.allocate-runners.outputs.version }}
disable-run-tests: ${{ env.DISABLE_RUN_TESTS }}
artifacts-dir: ${{ matrix.artifacts-dir-prefix }}-${{ needs.allocate-runners.outputs.version }}
aws-access-key-id: ${{ secrets.AWS_CN_ACCESS_KEY_ID }}
aws-secret-access-key: ${{ secrets.AWS_CN_SECRET_ACCESS_KEY }}
aws-region: ${{ vars.AWS_RELEASE_BUCKET_REGION }}
release-images-to-dockerhub:
name: Build and push images to DockerHub
@@ -299,18 +274,15 @@ jobs:
image-registry-password: ${{ secrets.DOCKERHUB_TOKEN }}
version: ${{ needs.allocate-runners.outputs.version }}
release-cn-artifacts:
name: Release artifacts to CN region
release-images-to-acr:
name: Build and push images to ACR
if: ${{ inputs.release_images || github.event_name == 'push' || github.event_name == 'schedule' }}
needs: [ # The job have to wait for all the artifacts are built.
needs: [
allocate-runners,
build-linux-amd64-artifacts,
build-linux-arm64-artifacts,
build-macos-artifacts,
build-windows-artifacts,
release-images-to-dockerhub,
]
runs-on: ubuntu-20.04
runs-on: ubuntu-2004-16-cores
# When we push to ACR, it's easy to fail due to some unknown network issues.
# However, we don't want to fail the whole workflow because of this.
# The ACR have daily sync with DockerHub, so don't worry about the image not being updated.
@@ -320,47 +292,55 @@ jobs:
with:
fetch-depth: 0
- name: Release artifacts to CN region
uses: ./.github/actions/release-cn-artifacts
- name: Build and push images to ACR
uses: ./.github/actions/build-images
with:
src-image-registry: docker.io
src-image-namespace: ${{ vars.IMAGE_NAMESPACE }}
src-image-name: greptimedb
dst-image-registry-username: ${{ secrets.ALICLOUD_USERNAME }}
dst-image-registry-password: ${{ secrets.ALICLOUD_PASSWORD }}
dst-image-registry: ${{ vars.ACR_IMAGE_REGISTRY }}
dst-image-namespace: ${{ vars.IMAGE_NAMESPACE }}
image-registry: ${{ vars.ACR_IMAGE_REGISTRY }}
image-namespace: ${{ vars.IMAGE_NAMESPACE }}
image-registry-username: ${{ secrets.ALICLOUD_USERNAME }}
image-registry-password: ${{ secrets.ALICLOUD_PASSWORD }}
version: ${{ needs.allocate-runners.outputs.version }}
aws-cn-s3-bucket: ${{ vars.AWS_RELEASE_BUCKET }}
aws-cn-access-key-id: ${{ secrets.AWS_CN_ACCESS_KEY_ID }}
aws-cn-secret-access-key: ${{ secrets.AWS_CN_SECRET_ACCESS_KEY }}
aws-cn-region: ${{ vars.AWS_RELEASE_BUCKET_REGION }}
dev-mode: false
update-version-info: true
push-latest-tag: true
publish-github-release:
release-artifacts:
name: Create GitHub release and upload artifacts
if: ${{ inputs.publish_github_release || github.event_name == 'push' || github.event_name == 'schedule' }}
needs: [ # The job have to wait for all the artifacts are built.
if: ${{ inputs.release_artifacts || github.event_name == 'push' || github.event_name == 'schedule' }}
needs: [
allocate-runners,
build-linux-amd64-artifacts,
build-linux-arm64-artifacts,
build-macos-artifacts,
build-windows-artifacts,
release-images-to-dockerhub,
]
runs-on: ubuntu-20.04
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v3
with:
fetch-depth: 0
- name: Publish GitHub release
uses: ./.github/actions/publish-github-release
- name: Release artifacts
uses: ./.github/actions/release-artifacts
with:
version: ${{ needs.allocate-runners.outputs.version }}
release-dev-builder-image:
name: Release dev builder image
if: ${{ inputs.release_dev_builder_image }} # Only manually trigger this job.
runs-on: ubuntu-latest-16-cores
steps:
- name: Checkout
uses: actions/checkout@v3
with:
fetch-depth: 0
- name: Build and push dev builder image
uses: ./.github/actions/build-dev-builder-image
with:
dockerhub-image-registry-username: ${{ secrets.DOCKERHUB_USERNAME }}
dockerhub-image-registry-token: ${{ secrets.DOCKERHUB_TOKEN }}
acr-image-registry: ${{ vars.ACR_IMAGE_REGISTRY }}
acr-image-registry-username: ${{ secrets.ALICLOUD_USERNAME }}
acr-image-registry-password: ${{ secrets.ALICLOUD_PASSWORD }}
### Stop runners ###
# It's very necessary to split the job of releasing runners into 'stop-linux-amd64-runner' and 'stop-linux-arm64-runner'.
# Because we can terminate the specified EC2 instance immediately after the job is finished without uncessary waiting.
@@ -368,7 +348,7 @@ jobs:
name: Stop linux-amd64 runner
# Only run this job when the runner is allocated.
if: ${{ always() }}
runs-on: ubuntu-20.04
runs-on: ubuntu-latest
needs: [
allocate-runners,
build-linux-amd64-artifacts,
@@ -393,7 +373,7 @@ jobs:
name: Stop linux-arm64 runner
# Only run this job when the runner is allocated.
if: ${{ always() }}
runs-on: ubuntu-20.04
runs-on: ubuntu-latest
needs: [
allocate-runners,
build-linux-arm64-artifacts,

View File

@@ -1,26 +0,0 @@
name: size-labeler
on: [pull_request]
jobs:
labeler:
runs-on: ubuntu-latest
name: Label the PR size
steps:
- uses: codelytv/pr-size-labeler@v1
with:
GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
s_label: 'Size: S'
s_max_size: '100'
m_label: 'Size: M'
m_max_size: '500'
l_label: 'Size: L'
l_max_size: '1000'
xl_label: 'Size: XL'
fail_if_xl: 'false'
message_if_xl: >
This PR exceeds the recommended size of 1000 lines.
Please make sure you are NOT addressing multiple issues with one PR.
Note this PR might be rejected due to its size.
github_api_url: 'api.github.com'
files_to_ignore: 'Cargo.lock'

1450
Cargo.lock generated

File diff suppressed because it is too large Load Diff

View File

@@ -8,11 +8,10 @@ members = [
"src/cmd",
"src/common/base",
"src/common/catalog",
"src/common/config",
"src/common/datasource",
"src/common/error",
"src/common/function",
"src/common/macro",
"src/common/function-macro",
"src/common/greptimedb-telemetry",
"src/common/grpc",
"src/common/grpc-expr",
@@ -30,16 +29,15 @@ members = [
"src/common/version",
"src/datanode",
"src/datatypes",
"src/file-engine",
"src/file-table-engine",
"src/frontend",
"src/log-store",
"src/meta-client",
"src/meta-srv",
"src/mito",
"src/mito2",
"src/object-store",
"src/operator",
"src/partition",
"src/plugins",
"src/promql",
"src/query",
"src/script",
@@ -48,62 +46,53 @@ members = [
"src/sql",
"src/storage",
"src/store-api",
"src/flow",
"src/table",
"src/table-procedure",
"tests-integration",
"tests/runner",
]
resolver = "2"
[workspace.package]
version = "0.4.2"
version = "0.4.0-nightly"
edition = "2021"
license = "Apache-2.0"
[workspace.dependencies]
aquamarine = "0.3"
arrow = { version = "43.0" }
etcd-client = "0.11"
arrow-array = "43.0"
arrow-flight = "43.0"
arrow-schema = { version = "43.0", features = ["serde"] }
async-stream = "0.3"
async-trait = "0.1"
chrono = { version = "0.4", features = ["serde"] }
datafusion = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "b6f3b28b6fe91924cc8dd3d83726b766f2a706ec" }
datafusion-common = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "b6f3b28b6fe91924cc8dd3d83726b766f2a706ec" }
datafusion-expr = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "b6f3b28b6fe91924cc8dd3d83726b766f2a706ec" }
datafusion-optimizer = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "b6f3b28b6fe91924cc8dd3d83726b766f2a706ec" }
datafusion-physical-expr = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "b6f3b28b6fe91924cc8dd3d83726b766f2a706ec" }
datafusion-sql = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "b6f3b28b6fe91924cc8dd3d83726b766f2a706ec" }
datafusion-substrait = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "b6f3b28b6fe91924cc8dd3d83726b766f2a706ec" }
datafusion = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "c0b0fca548e99d020c76e1a1cd7132aab26000e1" }
datafusion-common = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "c0b0fca548e99d020c76e1a1cd7132aab26000e1" }
datafusion-expr = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "c0b0fca548e99d020c76e1a1cd7132aab26000e1" }
datafusion-optimizer = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "c0b0fca548e99d020c76e1a1cd7132aab26000e1" }
datafusion-physical-expr = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "c0b0fca548e99d020c76e1a1cd7132aab26000e1" }
datafusion-sql = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "c0b0fca548e99d020c76e1a1cd7132aab26000e1" }
datafusion-substrait = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "c0b0fca548e99d020c76e1a1cd7132aab26000e1" }
derive_builder = "0.12"
etcd-client = "0.11"
futures = "0.3"
futures-util = "0.3"
greptime-proto = { git = "https://github.com/GreptimeTeam/greptime-proto.git", rev = "1f1dd532a111e3834cc3019c5605e2993ffb9dc3" }
greptime-proto = { git = "https://github.com/GreptimeTeam/greptime-proto.git", rev = "4a277f27caa035a801d5b9c020a0449777736614" }
humantime-serde = "1.1"
itertools = "0.10"
lazy_static = "1.4"
meter-core = { git = "https://github.com/GreptimeTeam/greptime-meter.git", rev = "abbd357c1e193cd270ea65ee7652334a150b628f" }
metrics = "0.20"
moka = "0.12"
once_cell = "1.18"
opentelemetry-proto = { version = "0.2", features = ["gen-tonic", "metrics", "traces"] }
opentelemetry-proto = { version = "0.2", features = ["gen-tonic", "metrics"] }
parquet = "43.0"
paste = "1.0"
prost = "0.11"
raft-engine = { git = "https://github.com/tikv/raft-engine.git", rev = "22dfb426cd994602b57725ef080287d3e53db479" }
rand = "0.8"
regex = "1.8"
reqwest = { version = "0.11", default-features = false, features = [
"json",
"rustls-tls-native-roots",
"stream",
] }
serde = { version = "1.0", features = ["derive"] }
serde_json = "1.0"
smallvec = "1"
snafu = { version = "0.7", features = ["backtraces"] }
sqlparser = { git = "https://github.com/GreptimeTeam/sqlparser-rs.git", rev = "6cf9d23d5b8fbecd65efc1d9afb7e80ad7a424da", features = [
sqlparser = { git = "https://github.com/GreptimeTeam/sqlparser-rs.git", rev = "296a4f6c73b129d6f565a42a2e5e53c6bc2b9da4", features = [
"visitor",
] }
strum = { version = "0.25", features = ["derive"] }
@@ -113,6 +102,8 @@ tokio-util = { version = "0.7", features = ["io-util", "compat"] }
toml = "0.7"
tonic = { version = "0.9", features = ["tls"] }
uuid = { version = "1", features = ["serde", "v4", "fast-rng"] }
metrics = "0.20"
meter-core = { git = "https://github.com/GreptimeTeam/greptime-meter.git", rev = "abbd357c1e193cd270ea65ee7652334a150b628f" }
## workspaces members
api = { path = "src/api" }
auth = { path = "src/auth" }
@@ -121,29 +112,29 @@ client = { path = "src/client" }
cmd = { path = "src/cmd" }
common-base = { path = "src/common/base" }
common-catalog = { path = "src/common/catalog" }
common-config = { path = "src/common/config" }
common-datasource = { path = "src/common/datasource" }
common-error = { path = "src/common/error" }
common-function = { path = "src/common/function" }
common-function-macro = { path = "src/common/function-macro" }
common-greptimedb-telemetry = { path = "src/common/greptimedb-telemetry" }
common-grpc = { path = "src/common/grpc" }
common-grpc-expr = { path = "src/common/grpc-expr" }
common-macro = { path = "src/common/macro" }
common-mem-prof = { path = "src/common/mem-prof" }
common-meta = { path = "src/common/meta" }
common-pprof = { path = "src/common/pprof" }
common-procedure = { path = "src/common/procedure" }
common-procedure-test = { path = "src/common/procedure-test" }
common-pprof = { path = "src/common/pprof" }
common-query = { path = "src/common/query" }
common-recordbatch = { path = "src/common/recordbatch" }
common-runtime = { path = "src/common/runtime" }
substrait = { path = "src/common/substrait" }
common-telemetry = { path = "src/common/telemetry" }
common-test-util = { path = "src/common/test-util" }
common-time = { path = "src/common/time" }
common-version = { path = "src/common/version" }
datanode = { path = "src/datanode" }
datatypes = { path = "src/datatypes" }
file-engine = { path = "src/file-engine" }
file-table-engine = { path = "src/file-table-engine" }
frontend = { path = "src/frontend" }
log-store = { path = "src/log-store" }
meta-client = { path = "src/meta-client" }
@@ -151,9 +142,7 @@ meta-srv = { path = "src/meta-srv" }
mito = { path = "src/mito" }
mito2 = { path = "src/mito2" }
object-store = { path = "src/object-store" }
operator = { path = "src/operator" }
partition = { path = "src/partition" }
plugins = { path = "src/plugins" }
promql = { path = "src/promql" }
query = { path = "src/query" }
script = { path = "src/script" }
@@ -162,8 +151,8 @@ session = { path = "src/session" }
sql = { path = "src/sql" }
storage = { path = "src/storage" }
store-api = { path = "src/store-api" }
substrait = { path = "src/common/substrait" }
table = { path = "src/table" }
table-procedure = { path = "src/table-procedure" }
[workspace.dependencies.meter-macros]
git = "https://github.com/GreptimeTeam/greptime-meter.git"

View File

@@ -55,15 +55,11 @@ else
BUILDX_MULTI_PLATFORM_BUILD_OPTS := -o type=docker
endif
ifneq ($(strip $(CARGO_BUILD_EXTRA_OPTS)),)
CARGO_BUILD_OPTS += ${CARGO_BUILD_EXTRA_OPTS}
endif
##@ Build
.PHONY: build
build: ## Build debug version greptime.
cargo ${CARGO_EXTENSION} build ${CARGO_BUILD_OPTS}
cargo build ${CARGO_BUILD_OPTS}
.POHNY: build-by-dev-builder
build-by-dev-builder: ## Build greptime by dev-builder.
@@ -71,34 +67,11 @@ build-by-dev-builder: ## Build greptime by dev-builder.
-v ${PWD}:/greptimedb -v ${CARGO_REGISTRY_CACHE}:/root/.cargo/registry \
-w /greptimedb ${IMAGE_REGISTRY}/${IMAGE_NAMESPACE}/dev-builder-${BASE_IMAGE}:latest \
make build \
CARGO_EXTENSION="${CARGO_EXTENSION}" \
CARGO_PROFILE=${CARGO_PROFILE} \
FEATURES=${FEATURES} \
TARGET_DIR=${TARGET_DIR} \
TARGET=${TARGET} \
RELEASE=${RELEASE} \
CARGO_BUILD_EXTRA_OPTS="${CARGO_BUILD_EXTRA_OPTS}"
.PHONY: build-android-bin
build-android-bin: ## Build greptime binary for android.
docker run --network=host \
-v ${PWD}:/greptimedb -v ${CARGO_REGISTRY_CACHE}:/root/.cargo/registry \
-w /greptimedb ${IMAGE_REGISTRY}/${IMAGE_NAMESPACE}/dev-builder-android:latest \
make build \
CARGO_EXTENSION="ndk --platform 23 -t aarch64-linux-android" \
CARGO_PROFILE=release \
FEATURES="${FEATURES}" \
TARGET_DIR="${TARGET_DIR}" \
TARGET="${TARGET}" \
RELEASE="${RELEASE}" \
CARGO_BUILD_EXTRA_OPTS="--bin greptime --no-default-features"
.PHONY: strip-android-bin
strip-android-bin: build-android-bin ## Strip greptime binary for android.
docker run --network=host \
-v ${PWD}:/greptimedb \
-w /greptimedb ${IMAGE_REGISTRY}/${IMAGE_NAMESPACE}/dev-builder-android:latest \
bash -c '$${NDK_ROOT}/toolchains/llvm/prebuilt/linux-x86_64/bin/llvm-strip /greptimedb/target/aarch64-linux-android/release/greptime'
RELEASE=${RELEASE}
.PHONY: clean
clean: ## Clean the project.

View File

@@ -27,14 +27,6 @@
<a href="https://greptime.com/slack"><img src="https://img.shields.io/badge/slack-GreptimeDB-0abd59?logo=slack" alt="slack" /></a>
</p>
## Upcoming Event
Come and meet us in **KubeCon + CloudNativeCon North America 2023!**
<p align="center">
<picture>
<img alt="KubeCon + CloudNativeCon North Logo" src="./docs/banner/KCCNC_NA_2023_1000x200_Email Banner.png" width="800px">
</picture>
</p>
## What is GreptimeDB
GreptimeDB is an open-source time-series database with a special focus on
@@ -104,11 +96,11 @@ Or if you built from docker:
docker run -p 4002:4002 -v "$(pwd):/tmp/greptimedb" greptime/greptimedb standalone start
```
Please see the online document site for more installation options and [operations info](https://docs.greptime.com/user-guide/operations/overview).
Please see [the online document site](https://docs.greptime.com/getting-started/overview#install-greptimedb) for more installation options and [operations info](https://docs.greptime.com/user-guide/operations/overview).
### Get started
Read the [complete getting started guide](https://docs.greptime.com/getting-started/try-out-greptimedb) on our [official document site](https://docs.greptime.com/).
Read the [complete getting started guide](https://docs.greptime.com/getting-started/overview#connect) on our [official document site](https://docs.greptime.com/).
To write and query data, GreptimeDB is compatible with multiple [protocols and clients](https://docs.greptime.com/user-guide/clients/overview).

View File

@@ -6,10 +6,8 @@ license.workspace = true
[dependencies]
arrow.workspace = true
chrono.workspace = true
clap = { version = "4.0", features = ["derive"] }
client = { workspace = true }
futures-util.workspace = true
indicatif = "0.17.1"
itertools.workspace = true
parquet.workspace = true

View File

@@ -27,16 +27,16 @@ use arrow::record_batch::RecordBatch;
use clap::Parser;
use client::api::v1::column::Values;
use client::api::v1::{
Column, ColumnDataType, ColumnDef, CreateTableExpr, InsertRequest, InsertRequests, SemanticType,
Column, ColumnDataType, ColumnDef, CreateTableExpr, InsertRequest, InsertRequests,
};
use client::{Client, Database, Output, DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME};
use futures_util::TryStreamExt;
use client::{Client, Database, DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME};
use indicatif::{MultiProgress, ProgressBar, ProgressStyle};
use parquet::arrow::arrow_reader::ParquetRecordBatchReaderBuilder;
use tokio::task::JoinSet;
const CATALOG_NAME: &str = "greptime";
const SCHEMA_NAME: &str = "public";
const TABLE_NAME: &str = "nyc_taxi";
#[derive(Parser)]
#[command(name = "NYC benchmark runner")]
@@ -74,12 +74,7 @@ fn get_file_list<P: AsRef<Path>>(path: P) -> Vec<PathBuf> {
.collect()
}
fn new_table_name() -> String {
format!("nyc_taxi_{}", chrono::Utc::now().timestamp())
}
async fn write_data(
table_name: &str,
batch_size: usize,
db: &Database,
path: PathBuf,
@@ -109,7 +104,8 @@ async fn write_data(
}
let (columns, row_count) = convert_record_batch(record_batch);
let request = InsertRequest {
table_name: table_name.to_string(),
table_name: TABLE_NAME.to_string(),
region_number: 0,
columns,
row_count,
};
@@ -118,7 +114,7 @@ async fn write_data(
};
let now = Instant::now();
db.insert(requests).await.unwrap();
let _ = db.insert(requests).await.unwrap();
let elapsed = now.elapsed();
total_rpc_elapsed_ms += elapsed.as_millis();
progress_bar.inc(row_count as _);
@@ -136,11 +132,6 @@ fn convert_record_batch(record_batch: RecordBatch) -> (Vec<Column>, u32) {
for (array, field) in record_batch.columns().iter().zip(fields.iter()) {
let (values, datatype) = build_values(array);
let semantic_type = match field.name().as_str() {
"VendorID" => SemanticType::Tag,
"tpep_pickup_datetime" => SemanticType::Timestamp,
_ => SemanticType::Field,
};
let column = Column {
column_name: field.name().clone(),
@@ -151,7 +142,8 @@ fn convert_record_batch(record_batch: RecordBatch) -> (Vec<Column>, u32) {
.map(|bitmap| bitmap.buffer().as_slice().to_vec())
.unwrap_or_default(),
datatype: datatype.into(),
semantic_type: semantic_type as i32,
// datatype and semantic_type are set to default
..Default::default()
};
columns.push(column);
}
@@ -197,7 +189,7 @@ fn build_values(column: &ArrayRef) -> (Values, ColumnDataType) {
let values = array.values();
(
Values {
timestamp_microsecond_values: values.to_vec(),
ts_microsecond_values: values.to_vec(),
..Default::default()
},
ColumnDataType::TimestampMicrosecond,
@@ -252,193 +244,156 @@ fn is_record_batch_full(batch: &RecordBatch) -> bool {
batch.columns().iter().all(|col| col.null_count() == 0)
}
fn create_table_expr(table_name: &str) -> CreateTableExpr {
fn create_table_expr() -> CreateTableExpr {
CreateTableExpr {
catalog_name: CATALOG_NAME.to_string(),
schema_name: SCHEMA_NAME.to_string(),
table_name: table_name.to_string(),
table_name: TABLE_NAME.to_string(),
desc: "".to_string(),
column_defs: vec![
ColumnDef {
name: "VendorID".to_string(),
data_type: ColumnDataType::Int64 as i32,
datatype: ColumnDataType::Int64 as i32,
is_nullable: true,
default_constraint: vec![],
semantic_type: SemanticType::Tag as i32,
comment: String::new(),
},
ColumnDef {
name: "tpep_pickup_datetime".to_string(),
data_type: ColumnDataType::TimestampMicrosecond as i32,
is_nullable: false,
datatype: ColumnDataType::TimestampMicrosecond as i32,
is_nullable: true,
default_constraint: vec![],
semantic_type: SemanticType::Timestamp as i32,
comment: String::new(),
},
ColumnDef {
name: "tpep_dropoff_datetime".to_string(),
data_type: ColumnDataType::TimestampMicrosecond as i32,
datatype: ColumnDataType::TimestampMicrosecond as i32,
is_nullable: true,
default_constraint: vec![],
semantic_type: SemanticType::Field as i32,
comment: String::new(),
},
ColumnDef {
name: "passenger_count".to_string(),
data_type: ColumnDataType::Float64 as i32,
datatype: ColumnDataType::Float64 as i32,
is_nullable: true,
default_constraint: vec![],
semantic_type: SemanticType::Field as i32,
comment: String::new(),
},
ColumnDef {
name: "trip_distance".to_string(),
data_type: ColumnDataType::Float64 as i32,
datatype: ColumnDataType::Float64 as i32,
is_nullable: true,
default_constraint: vec![],
semantic_type: SemanticType::Field as i32,
comment: String::new(),
},
ColumnDef {
name: "RatecodeID".to_string(),
data_type: ColumnDataType::Float64 as i32,
datatype: ColumnDataType::Float64 as i32,
is_nullable: true,
default_constraint: vec![],
semantic_type: SemanticType::Field as i32,
comment: String::new(),
},
ColumnDef {
name: "store_and_fwd_flag".to_string(),
data_type: ColumnDataType::String as i32,
datatype: ColumnDataType::String as i32,
is_nullable: true,
default_constraint: vec![],
semantic_type: SemanticType::Field as i32,
comment: String::new(),
},
ColumnDef {
name: "PULocationID".to_string(),
data_type: ColumnDataType::Int64 as i32,
datatype: ColumnDataType::Int64 as i32,
is_nullable: true,
default_constraint: vec![],
semantic_type: SemanticType::Field as i32,
comment: String::new(),
},
ColumnDef {
name: "DOLocationID".to_string(),
data_type: ColumnDataType::Int64 as i32,
datatype: ColumnDataType::Int64 as i32,
is_nullable: true,
default_constraint: vec![],
semantic_type: SemanticType::Field as i32,
comment: String::new(),
},
ColumnDef {
name: "payment_type".to_string(),
data_type: ColumnDataType::Int64 as i32,
datatype: ColumnDataType::Int64 as i32,
is_nullable: true,
default_constraint: vec![],
semantic_type: SemanticType::Field as i32,
comment: String::new(),
},
ColumnDef {
name: "fare_amount".to_string(),
data_type: ColumnDataType::Float64 as i32,
datatype: ColumnDataType::Float64 as i32,
is_nullable: true,
default_constraint: vec![],
semantic_type: SemanticType::Field as i32,
comment: String::new(),
},
ColumnDef {
name: "extra".to_string(),
data_type: ColumnDataType::Float64 as i32,
datatype: ColumnDataType::Float64 as i32,
is_nullable: true,
default_constraint: vec![],
semantic_type: SemanticType::Field as i32,
comment: String::new(),
},
ColumnDef {
name: "mta_tax".to_string(),
data_type: ColumnDataType::Float64 as i32,
datatype: ColumnDataType::Float64 as i32,
is_nullable: true,
default_constraint: vec![],
semantic_type: SemanticType::Field as i32,
comment: String::new(),
},
ColumnDef {
name: "tip_amount".to_string(),
data_type: ColumnDataType::Float64 as i32,
datatype: ColumnDataType::Float64 as i32,
is_nullable: true,
default_constraint: vec![],
semantic_type: SemanticType::Field as i32,
comment: String::new(),
},
ColumnDef {
name: "tolls_amount".to_string(),
data_type: ColumnDataType::Float64 as i32,
datatype: ColumnDataType::Float64 as i32,
is_nullable: true,
default_constraint: vec![],
semantic_type: SemanticType::Field as i32,
comment: String::new(),
},
ColumnDef {
name: "improvement_surcharge".to_string(),
data_type: ColumnDataType::Float64 as i32,
datatype: ColumnDataType::Float64 as i32,
is_nullable: true,
default_constraint: vec![],
semantic_type: SemanticType::Field as i32,
comment: String::new(),
},
ColumnDef {
name: "total_amount".to_string(),
data_type: ColumnDataType::Float64 as i32,
datatype: ColumnDataType::Float64 as i32,
is_nullable: true,
default_constraint: vec![],
semantic_type: SemanticType::Field as i32,
comment: String::new(),
},
ColumnDef {
name: "congestion_surcharge".to_string(),
data_type: ColumnDataType::Float64 as i32,
datatype: ColumnDataType::Float64 as i32,
is_nullable: true,
default_constraint: vec![],
semantic_type: SemanticType::Field as i32,
comment: String::new(),
},
ColumnDef {
name: "airport_fee".to_string(),
data_type: ColumnDataType::Float64 as i32,
datatype: ColumnDataType::Float64 as i32,
is_nullable: true,
default_constraint: vec![],
semantic_type: SemanticType::Field as i32,
comment: String::new(),
},
],
time_index: "tpep_pickup_datetime".to_string(),
primary_keys: vec!["VendorID".to_string()],
create_if_not_exists: true,
create_if_not_exists: false,
table_options: Default::default(),
region_numbers: vec![0],
table_id: None,
engine: "mito".to_string(),
}
}
fn query_set(table_name: &str) -> HashMap<String, String> {
fn query_set() -> HashMap<String, String> {
HashMap::from([
(
"count_all".to_string(),
format!("SELECT COUNT(*) FROM {table_name};"),
format!("SELECT COUNT(*) FROM {TABLE_NAME};"),
),
(
"fare_amt_by_passenger".to_string(),
format!("SELECT passenger_count, MIN(fare_amount), MAX(fare_amount), SUM(fare_amount) FROM {table_name} GROUP BY passenger_count"),
format!("SELECT passenger_count, MIN(fare_amount), MAX(fare_amount), SUM(fare_amount) FROM {TABLE_NAME} GROUP BY passenger_count"),
)
])
}
async fn do_write(args: &Args, db: &Database, table_name: &str) {
async fn do_write(args: &Args, db: &Database) {
let mut file_list = get_file_list(args.path.clone().expect("Specify data path in argument"));
let mut write_jobs = JoinSet::new();
let create_table_result = db.create(create_table_expr(table_name)).await;
let create_table_result = db.create(create_table_expr()).await;
println!("Create table result: {create_table_result:?}");
let progress_bar_style = ProgressStyle::with_template(
@@ -456,10 +411,8 @@ async fn do_write(args: &Args, db: &Database, table_name: &str) {
let db = db.clone();
let mpb = multi_progress_bar.clone();
let pb_style = progress_bar_style.clone();
let table_name = table_name.to_string();
let _ = write_jobs.spawn(async move {
write_data(&table_name, batch_size, &db, path, mpb, pb_style).await
});
let _ = write_jobs
.spawn(async move { write_data(batch_size, &db, path, mpb, pb_style).await });
}
}
while write_jobs.join_next().await.is_some() {
@@ -468,32 +421,24 @@ async fn do_write(args: &Args, db: &Database, table_name: &str) {
let db = db.clone();
let mpb = multi_progress_bar.clone();
let pb_style = progress_bar_style.clone();
let table_name = table_name.to_string();
let _ = write_jobs.spawn(async move {
write_data(&table_name, batch_size, &db, path, mpb, pb_style).await
});
let _ = write_jobs
.spawn(async move { write_data(batch_size, &db, path, mpb, pb_style).await });
}
}
}
async fn do_query(num_iter: usize, db: &Database, table_name: &str) {
for (query_name, query) in query_set(table_name) {
async fn do_query(num_iter: usize, db: &Database) {
for (query_name, query) in query_set() {
println!("Running query: {query}");
for i in 0..num_iter {
let now = Instant::now();
let res = db.sql(&query).await.unwrap();
match res {
Output::AffectedRows(_) | Output::RecordBatches(_) => (),
Output::Stream(stream) => {
stream.try_collect::<Vec<_>>().await.unwrap();
}
}
let _res = db.sql(&query).await.unwrap();
let elapsed = now.elapsed();
println!(
"query {}, iteration {}: {}ms",
query_name,
i,
elapsed.as_millis(),
elapsed.as_millis()
);
}
}
@@ -510,14 +455,13 @@ fn main() {
.block_on(async {
let client = Client::with_urls(vec![&args.endpoint]);
let db = Database::new(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, client);
let table_name = new_table_name();
if !args.skip_write {
do_write(&args, &db, &table_name).await;
do_write(&args, &db).await;
}
if !args.skip_read {
do_query(args.iter_num, &db, &table_name).await;
do_query(args.iter_num, &db).await;
}
})
}

View File

@@ -1,5 +1,7 @@
# Node running mode, see `standalone.example.toml`.
mode = "distributed"
# Whether to use in-memory catalog, see `standalone.example.toml`.
enable_memory_catalog = false
# The datanode identifier, should be unique.
node_id = 42
# gRPC server address, "127.0.0.1:3001" by default.
@@ -8,24 +10,19 @@ rpc_addr = "127.0.0.1:3001"
rpc_hostname = "127.0.0.1"
# The number of gRPC server worker threads, 8 by default.
rpc_runtime_size = 8
# Start services after regions have obtained leases.
# It will block the datanode start if it can't receive leases in the heartbeat from metasrv.
require_lease_before_startup = false
[heartbeat]
# Interval for sending heartbeat messages to the Metasrv, 3 seconds by default.
interval = "3s"
# Interval for sending heartbeat messages to the Metasrv in milliseconds, 5000 by default.
interval_millis = 5000
# Metasrv client options.
[meta_client]
[meta_client_options]
# Metasrv address list.
metasrv_addrs = ["127.0.0.1:3002"]
# Heartbeat timeout, 500 milliseconds by default.
heartbeat_timeout = "500ms"
# Operation timeout, 3 seconds by default.
timeout = "3s"
# Connect server timeout, 1 second by default.
connect_timeout = "1s"
# Operation timeout in milliseconds, 3000 by default.
timeout_millis = 3000
# Connect server timeout in milliseconds, 5000 by default.
connect_timeout_millis = 5000
# `TCP_NODELAY` option for accepted connections, true by default.
tcp_nodelay = true
@@ -47,12 +44,6 @@ type = "File"
# TTL for all tables. Disabled by default.
# global_ttl = "7d"
# Cache configuration for object storage such as 'S3' etc.
# The local file cache directory
# cache_path = "/path/local_cache"
# The local file cache capacity in bytes.
# cache_capacity = "256MB"
# Compaction options, see `standalone.example.toml`.
[storage.compaction]
max_inflight_tasks = 4
@@ -80,27 +71,10 @@ auto_flush_interval = "1h"
# Global write buffer size for all regions.
global_write_buffer_size = "1GB"
# Mito engine options
[[region_engine]]
[region_engine.mito]
# Number of region workers
num_workers = 8
# Request channel size of each worker
worker_channel_size = 128
# Max batch size for a worker to handle requests
worker_request_batch_size = 64
# Number of meta action updated to trigger a new checkpoint for the manifest
manifest_checkpoint_distance = 10
# Manifest compression type
manifest_compress_type = "Uncompressed"
# Max number of running background jobs
max_background_jobs = 4
# Interval to auto flush a region if it has not flushed yet.
auto_flush_interval = "1h"
# Global write buffer size for all regions.
global_write_buffer_size = "1GB"
# Global write buffer size threshold to reject write requests (default 2G).
global_write_buffer_reject_size = "2GB"
# Procedure storage options, see `standalone.example.toml`.
[procedure]
max_retry_times = 3
retry_delay = "500ms"
# Log options
# [logging]

View File

@@ -2,67 +2,64 @@
mode = "distributed"
[heartbeat]
# Interval for sending heartbeat task to the Metasrv, 5 seconds by default.
interval = "5s"
# Interval for retry sending heartbeat task, 5 seconds by default.
retry_interval = "5s"
# Interval for sending heartbeat task to the Metasrv in milliseconds, 5000 by default.
interval_millis = 5000
# Interval for retry sending heartbeat task in milliseconds, 5000 by default.
retry_interval_millis = 5000
# HTTP server options, see `standalone.example.toml`.
[http]
[http_options]
addr = "127.0.0.1:4000"
timeout = "30s"
body_limit = "64MB"
# gRPC server options, see `standalone.example.toml`.
[grpc]
[grpc_options]
addr = "127.0.0.1:4001"
runtime_size = 8
# MySQL server options, see `standalone.example.toml`.
[mysql]
enable = true
[mysql_options]
addr = "127.0.0.1:4002"
runtime_size = 2
# MySQL server TLS options, see `standalone.example.toml`.
[mysql.tls]
[mysql_options.tls]
mode = "disable"
cert_path = ""
key_path = ""
# PostgresSQL server options, see `standalone.example.toml`.
[postgres]
enable = true
[postgres_options]
addr = "127.0.0.1:4003"
runtime_size = 2
# PostgresSQL server TLS options, see `standalone.example.toml`.
[postgres.tls]
[postgres_options.tls]
mode = "disable"
cert_path = ""
key_path = ""
# OpenTSDB protocol options, see `standalone.example.toml`.
[opentsdb]
enable = true
[opentsdb_options]
addr = "127.0.0.1:4242"
runtime_size = 2
# InfluxDB protocol options, see `standalone.example.toml`.
[influxdb]
[influxdb_options]
enable = true
# Prometheus remote storage options, see `standalone.example.toml`.
[prom_store]
[prom_store_options]
enable = true
# Metasrv client options, see `datanode.example.toml`.
[meta_client]
[meta_client_options]
metasrv_addrs = ["127.0.0.1:3002"]
timeout = "3s"
timeout_millis = 3000
# DDL timeouts options.
ddl_timeout = "10s"
connect_timeout = "1s"
ddl_timeout_millis = 10000
connect_timeout_millis = 5000
tcp_nodelay = true
# Log options, see `standalone.example.toml`

View File

@@ -6,6 +6,8 @@ bind_addr = "127.0.0.1:3002"
server_addr = "127.0.0.1:3002"
# Etcd server address, "127.0.0.1:2379" by default.
store_addr = "127.0.0.1:2379"
# Datanode lease in seconds, 15 seconds by default.
datanode_lease_secs = 15
# Datanode selector type.
# - "LeaseBased" (default value).
# - "LoadBased"
@@ -32,6 +34,6 @@ retry_delay = "500ms"
# [datanode]
# # Datanode client options.
# [datanode.client_options]
# timeout = "10s"
# connect_timeout = "10s"
# timeout_millis = 10000
# connect_timeout_millis = 10000
# tcp_nodelay = true

View File

@@ -1,10 +1,12 @@
# Node running mode, "standalone" or "distributed".
mode = "standalone"
# Whether to use in-memory catalog, `false` by default.
enable_memory_catalog = false
# Whether to enable greptimedb telemetry, true by default.
enable_telemetry = true
# HTTP server options.
[http]
[http_options]
# Server address, "127.0.0.1:4000" by default.
addr = "127.0.0.1:4000"
# HTTP request timeout, 30s by default.
@@ -14,23 +16,21 @@ timeout = "30s"
body_limit = "64MB"
# gRPC server options.
[grpc]
[grpc_options]
# Server address, "127.0.0.1:4001" by default.
addr = "127.0.0.1:4001"
# The number of server worker threads, 8 by default.
runtime_size = 8
# MySQL server options.
[mysql]
# Whether to enable
enable = true
[mysql_options]
# Server address, "127.0.0.1:4002" by default.
addr = "127.0.0.1:4002"
# The number of server worker threads, 2 by default.
runtime_size = 2
# MySQL server TLS options.
[mysql.tls]
[mysql_options.tls]
# TLS mode, refer to https://www.postgresql.org/docs/current/libpq-ssl.html
# - "disable" (default value)
# - "prefer"
@@ -44,16 +44,14 @@ cert_path = ""
key_path = ""
# PostgresSQL server options.
[postgres]
# Whether to enable
enable = true
[postgres_options]
# Server address, "127.0.0.1:4003" by default.
addr = "127.0.0.1:4003"
# The number of server worker threads, 2 by default.
runtime_size = 2
# PostgresSQL server TLS options, see `[mysql_options.tls]` section.
[postgres.tls]
[postgres_options.tls]
# TLS mode.
mode = "disable"
# certificate file path.
@@ -62,21 +60,19 @@ cert_path = ""
key_path = ""
# OpenTSDB protocol options.
[opentsdb]
# Whether to enable
enable = true
[opentsdb_options]
# OpenTSDB telnet API server address, "127.0.0.1:4242" by default.
addr = "127.0.0.1:4242"
# The number of server worker threads, 2 by default.
runtime_size = 2
# InfluxDB protocol options.
[influxdb]
[influxdb_options]
# Whether to enable InfluxDB protocol in HTTP API, true by default.
enable = true
# Prometheus remote storage options
[prom_store]
[prom_store_options]
# Whether to enable Prometheus remote write and read in HTTP API, true by default.
enable = true
@@ -95,20 +91,6 @@ read_batch_size = 128
# Whether to sync log file after every write.
sync_write = false
# Metadata storage options.
[metadata_store]
# Kv file size in bytes.
file_size = "256MB"
# Kv purge threshold.
purge_threshold = "4GB"
# Procedure storage options.
[procedure]
# Procedure max retry time.
max_retry_times = 3
# Initial retry delay of procedures, increases exponentially
retry_delay = "500ms"
# Storage options.
[storage]
# The working home directory.
@@ -117,10 +99,6 @@ data_home = "/tmp/greptimedb/"
type = "File"
# TTL for all tables. Disabled by default.
# global_ttl = "7d"
# Cache configuration for object storage such as 'S3' etc.
# cache_path = "/path/local_cache"
# The local file cache capacity in bytes.
# cache_capacity = "256MB"
# Compaction options.
[storage.compaction]
@@ -152,6 +130,13 @@ auto_flush_interval = "1h"
# Global write buffer size for all regions.
global_write_buffer_size = "1GB"
# Procedure storage options.
[procedure]
# Procedure max retry time.
max_retry_times = 3
# Initial retry delay of procedures, increases exponentially
retry_delay = "500ms"
# Log options
# [logging]
# Specify logs directory.

View File

@@ -1,4 +1,4 @@
FROM ubuntu:20.04 as builder
FROM ubuntu:22.04 as builder
ARG CARGO_PROFILE
ARG FEATURES
@@ -7,11 +7,6 @@ ARG OUTPUT_DIR
ENV LANG en_US.utf8
WORKDIR /greptimedb
# Add PPA for Python 3.10.
RUN apt-get update && \
DEBIAN_FRONTEND=noninteractive apt-get install -y software-properties-common && \
add-apt-repository ppa:deadsnakes/ppa -y
# Install dependencies.
RUN --mount=type=cache,target=/var/cache/apt \
apt-get update && apt-get install -y \

View File

@@ -1,41 +0,0 @@
FROM --platform=linux/amd64 saschpe/android-ndk:34-jdk17.0.8_7-ndk25.2.9519653-cmake3.22.1
ENV LANG en_US.utf8
WORKDIR /greptimedb
# Rename libunwind to libgcc
RUN cp ${NDK_ROOT}/toolchains/llvm/prebuilt/linux-x86_64/lib64/clang/14.0.7/lib/linux/aarch64/libunwind.a ${NDK_ROOT}/toolchains/llvm/prebuilt/linux-x86_64/lib64/clang/14.0.7/lib/linux/aarch64/libgcc.a
# Install dependencies.
RUN apt-get update && apt-get install -y \
libssl-dev \
protobuf-compiler \
curl \
git \
build-essential \
pkg-config \
python3 \
python3-dev \
python3-pip \
&& pip3 install --upgrade pip \
&& pip3 install pyarrow
# Trust workdir
RUN git config --global --add safe.directory /greptimedb
# Install Rust.
SHELL ["/bin/bash", "-c"]
RUN curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- --no-modify-path --default-toolchain none -y
ENV PATH /root/.cargo/bin/:$PATH
# Add android toolchains
ARG RUST_TOOLCHAIN
RUN rustup toolchain install ${RUST_TOOLCHAIN}
RUN rustup target add aarch64-linux-android
# Install cargo-ndk
RUN cargo install cargo-ndk
ENV ANDROID_NDK_HOME $NDK_ROOT
# Builder entrypoint.
CMD ["cargo", "ndk", "--platform", "23", "-t", "aarch64-linux-android", "build", "--bin", "greptime", "--profile", "release", "--no-default-features"]

View File

@@ -1,13 +1,8 @@
FROM ubuntu:20.04
FROM ubuntu:22.04
ENV LANG en_US.utf8
WORKDIR /greptimedb
# Add PPA for Python 3.10.
RUN apt-get update && \
DEBIAN_FRONTEND=noninteractive apt-get install -y software-properties-common && \
add-apt-repository ppa:deadsnakes/ppa -y
# Install dependencies.
RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y \
libssl-dev \
@@ -19,13 +14,8 @@ RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y \
build-essential \
pkg-config \
python3.10 \
python3.10-dev
# Remove Python 3.8 and install pip.
RUN apt-get -y purge python3.8 && \
apt-get -y autoremove && \
ln -s /usr/bin/python3.10 /usr/bin/python3 && \
curl -sS https://bootstrap.pypa.io/get-pip.py | python3.10
python3.10-dev \
python3-pip
RUN git config --global --add safe.directory /greptimedb

Binary file not shown.

Before

Width:  |  Height:  |  Size: 51 KiB

View File

@@ -1,61 +0,0 @@
# TSBS benchmark - v0.4.0
## Environment
### Local
| | |
| ------ | ---------------------------------- |
| CPU | AMD Ryzen 7 7735HS (8 core 3.2GHz) |
| Memory | 32GB |
| Disk | SOLIDIGM SSDPFKNU010TZ |
| OS | Ubuntu 22.04.2 LTS |
### Aliyun amd64
| | |
| ------- | -------------- |
| Machine | ecs.g7.4xlarge |
| CPU | 16 core |
| Memory | 64GB |
| Disk | 100G |
| OS | Ubuntu 22.04 |
### Aliyun arm64
| | |
| ------- | ----------------- |
| Machine | ecs.g8y.4xlarge |
| CPU | 16 core |
| Memory | 64GB |
| Disk | 100G |
| OS | Ubuntu 22.04 ARM |
## Write performance
| Environment | Ingest raterows/s |
| ------------------ | --------------------- |
| Local | 365280.60 |
| Aliyun g7.4xlarge | 341368.72 |
| Aliyun g8y.4xlarge | 320907.29 |
## Query performance
| Query type | Local (ms) | Aliyun g7.4xlarge (ms) | Aliyun g8y.4xlarge (ms) |
| --------------------- | ---------- | ---------------------- | ----------------------- |
| cpu-max-all-1 | 50.70 | 31.46 | 47.61 |
| cpu-max-all-8 | 262.16 | 129.26 | 152.43 |
| double-groupby-1 | 2512.71 | 1408.19 | 1586.10 |
| double-groupby-5 | 3896.15 | 2304.29 | 2585.29 |
| double-groupby-all | 5404.67 | 3337.61 | 3773.91 |
| groupby-orderby-limit | 3786.98 | 2065.72 | 2312.57 |
| high-cpu-1 | 71.96 | 37.29 | 54.01 |
| high-cpu-all | 9468.75 | 7595.69 | 8467.46 |
| lastpoint | 13379.43 | 11253.76 | 12949.40 |
| single-groupby-1-1-1 | 20.72 | 12.16 | 13.35 |
| single-groupby-1-1-12 | 28.53 | 15.67 | 21.62 |
| single-groupby-1-8-1 | 72.23 | 37.90 | 43.52 |
| single-groupby-5-1-1 | 26.75 | 15.59 | 17.48 |
| single-groupby-5-1-12 | 45.41 | 22.90 | 31.96 |
| single-groupby-5-8-1 | 107.96 | 59.76 | 69.58 |

View File

@@ -7,7 +7,6 @@ license.workspace = true
[dependencies]
common-base = { workspace = true }
common-error = { workspace = true }
common-macro = { workspace = true }
common-time = { workspace = true }
datatypes = { workspace = true }
greptime-proto.workspace = true

View File

@@ -16,16 +16,14 @@ use std::any::Any;
use common_error::ext::ErrorExt;
use common_error::status_code::StatusCode;
use common_macro::stack_trace_debug;
use datatypes::prelude::ConcreteDataType;
use snafu::prelude::*;
use snafu::Location;
pub type Result<T> = std::result::Result<T, Error>;
#[derive(Snafu)]
#[derive(Debug, Snafu)]
#[snafu(visibility(pub))]
#[stack_trace_debug]
pub enum Error {
#[snafu(display("Unknown proto column datatype: {}", datatype))]
UnknownColumnDataType { datatype: i32, location: Location },
@@ -36,14 +34,22 @@ pub enum Error {
location: Location,
},
#[snafu(display("Failed to convert column default constraint, column: {}", column))]
#[snafu(display(
"Failed to convert column default constraint, column: {}, source: {}",
column,
source
))]
ConvertColumnDefaultConstraint {
column: String,
location: Location,
source: datatypes::error::Error,
},
#[snafu(display("Invalid column default constraint, column: {}", column))]
#[snafu(display(
"Invalid column default constraint, column: {}, source: {}",
column,
source
))]
InvalidColumnDefaultConstraint {
column: String,
location: Location,

View File

@@ -18,32 +18,32 @@ use common_base::BitVec;
use common_time::interval::IntervalUnit;
use common_time::time::Time;
use common_time::timestamp::TimeUnit;
use common_time::{Date, DateTime, Duration, Interval, Timestamp};
use common_time::{Date, DateTime, Interval, Timestamp};
use datatypes::prelude::{ConcreteDataType, ValueRef};
use datatypes::scalars::ScalarVector;
use datatypes::types::{
DurationType, Int16Type, Int8Type, IntervalType, TimeType, TimestampType, UInt16Type, UInt8Type,
Int16Type, Int8Type, IntervalType, TimeType, TimestampType, UInt16Type, UInt8Type,
};
use datatypes::value::{OrderedF32, OrderedF64, Value};
use datatypes::vectors::{
BinaryVector, BooleanVector, DateTimeVector, DateVector, DurationMicrosecondVector,
DurationMillisecondVector, DurationNanosecondVector, DurationSecondVector, Float32Vector,
Float64Vector, Int32Vector, Int64Vector, IntervalDayTimeVector, IntervalMonthDayNanoVector,
BinaryVector, BooleanVector, DateTimeVector, DateVector, Float32Vector, Float64Vector,
Int32Vector, Int64Vector, IntervalDayTimeVector, IntervalMonthDayNanoVector,
IntervalYearMonthVector, PrimitiveVector, StringVector, TimeMicrosecondVector,
TimeMillisecondVector, TimeNanosecondVector, TimeSecondVector, TimestampMicrosecondVector,
TimestampMillisecondVector, TimestampNanosecondVector, TimestampSecondVector, UInt32Vector,
UInt64Vector, VectorRef,
};
use greptime_proto::v1;
use greptime_proto::v1::ddl_request::Expr;
use greptime_proto::v1::greptime_request::Request;
use greptime_proto::v1::query_request::Query;
use greptime_proto::v1::value::ValueData;
use greptime_proto::v1::{self, DdlRequest, IntervalMonthDayNano, QueryRequest, Row, SemanticType};
use greptime_proto::v1::{DdlRequest, IntervalMonthDayNano, QueryRequest, SemanticType};
use snafu::prelude::*;
use crate::error::{self, Result};
use crate::v1::column::Values;
use crate::v1::{Column, ColumnDataType, Value as GrpcValue};
use crate::v1::{Column, ColumnDataType};
#[derive(Debug, PartialEq, Eq)]
pub struct ColumnDataTypeWrapper(ColumnDataType);
@@ -101,14 +101,6 @@ impl From<ColumnDataTypeWrapper> for ConcreteDataType {
ColumnDataType::IntervalMonthDayNano => {
ConcreteDataType::interval_month_day_nano_datatype()
}
ColumnDataType::DurationSecond => ConcreteDataType::duration_second_datatype(),
ColumnDataType::DurationMillisecond => {
ConcreteDataType::duration_millisecond_datatype()
}
ColumnDataType::DurationMicrosecond => {
ConcreteDataType::duration_microsecond_datatype()
}
ColumnDataType::DurationNanosecond => ConcreteDataType::duration_nanosecond_datatype(),
}
}
}
@@ -150,12 +142,6 @@ impl TryFrom<ConcreteDataType> for ColumnDataTypeWrapper {
IntervalType::DayTime(_) => ColumnDataType::IntervalDayTime,
IntervalType::MonthDayNano(_) => ColumnDataType::IntervalMonthDayNano,
},
ConcreteDataType::Duration(d) => match d {
DurationType::Second(_) => ColumnDataType::DurationSecond,
DurationType::Millisecond(_) => ColumnDataType::DurationMillisecond,
DurationType::Microsecond(_) => ColumnDataType::DurationMicrosecond,
DurationType::Nanosecond(_) => ColumnDataType::DurationNanosecond,
},
ConcreteDataType::Null(_)
| ConcreteDataType::List(_)
| ConcreteDataType::Dictionary(_) => {
@@ -229,19 +215,19 @@ pub fn values_with_capacity(datatype: ColumnDataType, capacity: usize) -> Values
..Default::default()
},
ColumnDataType::TimestampSecond => Values {
timestamp_second_values: Vec::with_capacity(capacity),
ts_second_values: Vec::with_capacity(capacity),
..Default::default()
},
ColumnDataType::TimestampMillisecond => Values {
timestamp_millisecond_values: Vec::with_capacity(capacity),
ts_millisecond_values: Vec::with_capacity(capacity),
..Default::default()
},
ColumnDataType::TimestampMicrosecond => Values {
timestamp_microsecond_values: Vec::with_capacity(capacity),
ts_microsecond_values: Vec::with_capacity(capacity),
..Default::default()
},
ColumnDataType::TimestampNanosecond => Values {
timestamp_nanosecond_values: Vec::with_capacity(capacity),
ts_nanosecond_values: Vec::with_capacity(capacity),
..Default::default()
},
ColumnDataType::TimeSecond => Values {
@@ -272,22 +258,6 @@ pub fn values_with_capacity(datatype: ColumnDataType, capacity: usize) -> Values
interval_month_day_nano_values: Vec::with_capacity(capacity),
..Default::default()
},
ColumnDataType::DurationSecond => Values {
duration_second_values: Vec::with_capacity(capacity),
..Default::default()
},
ColumnDataType::DurationMillisecond => Values {
duration_millisecond_values: Vec::with_capacity(capacity),
..Default::default()
},
ColumnDataType::DurationMicrosecond => Values {
duration_microsecond_values: Vec::with_capacity(capacity),
..Default::default()
},
ColumnDataType::DurationNanosecond => Values {
duration_nanosecond_values: Vec::with_capacity(capacity),
..Default::default()
},
}
}
@@ -317,10 +287,10 @@ pub fn push_vals(column: &mut Column, origin_count: usize, vector: VectorRef) {
Value::Date(val) => values.date_values.push(val.val()),
Value::DateTime(val) => values.datetime_values.push(val.val()),
Value::Timestamp(val) => match val.unit() {
TimeUnit::Second => values.timestamp_second_values.push(val.value()),
TimeUnit::Millisecond => values.timestamp_millisecond_values.push(val.value()),
TimeUnit::Microsecond => values.timestamp_microsecond_values.push(val.value()),
TimeUnit::Nanosecond => values.timestamp_nanosecond_values.push(val.value()),
TimeUnit::Second => values.ts_second_values.push(val.value()),
TimeUnit::Millisecond => values.ts_millisecond_values.push(val.value()),
TimeUnit::Microsecond => values.ts_microsecond_values.push(val.value()),
TimeUnit::Nanosecond => values.ts_nanosecond_values.push(val.value()),
},
Value::Time(val) => match val.unit() {
TimeUnit::Second => values.time_second_values.push(val.value()),
@@ -335,12 +305,6 @@ pub fn push_vals(column: &mut Column, origin_count: usize, vector: VectorRef) {
.interval_month_day_nano_values
.push(convert_i128_to_interval(val.to_i128())),
},
Value::Duration(val) => match val.unit() {
TimeUnit::Second => values.duration_second_values.push(val.value()),
TimeUnit::Millisecond => values.duration_millisecond_values.push(val.value()),
TimeUnit::Microsecond => values.duration_microsecond_values.push(val.value()),
TimeUnit::Nanosecond => values.duration_nanosecond_values.push(val.value()),
},
Value::List(_) => unreachable!(),
});
column.null_mask = null_mask.into_vec();
@@ -375,6 +339,8 @@ fn ddl_request_type(request: &DdlRequest) -> &'static str {
Some(Expr::CreateTable(_)) => "ddl.create_table",
Some(Expr::Alter(_)) => "ddl.alter",
Some(Expr::DropTable(_)) => "ddl.drop_table",
Some(Expr::FlushTable(_)) => "ddl.flush_table",
Some(Expr::CompactTable(_)) => "ddl.compact_table",
Some(Expr::TruncateTable(_)) => "ddl.truncate_table",
None => "ddl.empty",
}
@@ -412,16 +378,10 @@ pub fn pb_value_to_value_ref(value: &v1::Value) -> ValueRef {
ValueData::StringValue(string) => ValueRef::String(string.as_str()),
ValueData::DateValue(d) => ValueRef::Date(Date::from(*d)),
ValueData::DatetimeValue(d) => ValueRef::DateTime(DateTime::new(*d)),
ValueData::TimestampSecondValue(t) => ValueRef::Timestamp(Timestamp::new_second(*t)),
ValueData::TimestampMillisecondValue(t) => {
ValueRef::Timestamp(Timestamp::new_millisecond(*t))
}
ValueData::TimestampMicrosecondValue(t) => {
ValueRef::Timestamp(Timestamp::new_microsecond(*t))
}
ValueData::TimestampNanosecondValue(t) => {
ValueRef::Timestamp(Timestamp::new_nanosecond(*t))
}
ValueData::TsSecondValue(t) => ValueRef::Timestamp(Timestamp::new_second(*t)),
ValueData::TsMillisecondValue(t) => ValueRef::Timestamp(Timestamp::new_millisecond(*t)),
ValueData::TsMicrosecondValue(t) => ValueRef::Timestamp(Timestamp::new_microsecond(*t)),
ValueData::TsNanosecondValue(t) => ValueRef::Timestamp(Timestamp::new_nanosecond(*t)),
ValueData::TimeSecondValue(t) => ValueRef::Time(Time::new_second(*t)),
ValueData::TimeMillisecondValue(t) => ValueRef::Time(Time::new_millisecond(*t)),
ValueData::TimeMicrosecondValue(t) => ValueRef::Time(Time::new_microsecond(*t)),
@@ -432,10 +392,6 @@ pub fn pb_value_to_value_ref(value: &v1::Value) -> ValueRef {
let interval = Interval::from_month_day_nano(v.months, v.days, v.nanoseconds);
ValueRef::Interval(interval)
}
ValueData::DurationSecondValue(v) => ValueRef::Duration(Duration::new_second(*v)),
ValueData::DurationMillisecondValue(v) => ValueRef::Duration(Duration::new_millisecond(*v)),
ValueData::DurationMicrosecondValue(v) => ValueRef::Duration(Duration::new_microsecond(*v)),
ValueData::DurationNanosecondValue(v) => ValueRef::Duration(Duration::new_nanosecond(*v)),
}
}
@@ -465,17 +421,17 @@ pub fn pb_values_to_vector_ref(data_type: &ConcreteDataType, values: Values) ->
ConcreteDataType::Date(_) => Arc::new(DateVector::from_vec(values.date_values)),
ConcreteDataType::DateTime(_) => Arc::new(DateTimeVector::from_vec(values.datetime_values)),
ConcreteDataType::Timestamp(unit) => match unit {
TimestampType::Second(_) => Arc::new(TimestampSecondVector::from_vec(
values.timestamp_second_values,
)),
TimestampType::Second(_) => {
Arc::new(TimestampSecondVector::from_vec(values.ts_second_values))
}
TimestampType::Millisecond(_) => Arc::new(TimestampMillisecondVector::from_vec(
values.timestamp_millisecond_values,
values.ts_millisecond_values,
)),
TimestampType::Microsecond(_) => Arc::new(TimestampMicrosecondVector::from_vec(
values.timestamp_microsecond_values,
values.ts_microsecond_values,
)),
TimestampType::Nanosecond(_) => Arc::new(TimestampNanosecondVector::from_vec(
values.timestamp_nanosecond_values,
values.ts_nanosecond_values,
)),
},
ConcreteDataType::Time(unit) => match unit {
@@ -508,20 +464,6 @@ pub fn pb_values_to_vector_ref(data_type: &ConcreteDataType, values: Values) ->
))
}
},
ConcreteDataType::Duration(unit) => match unit {
DurationType::Second(_) => Arc::new(DurationSecondVector::from_vec(
values.duration_second_values,
)),
DurationType::Millisecond(_) => Arc::new(DurationMillisecondVector::from_vec(
values.duration_millisecond_values,
)),
DurationType::Microsecond(_) => Arc::new(DurationMicrosecondVector::from_vec(
values.duration_microsecond_values,
)),
DurationType::Nanosecond(_) => Arc::new(DurationNanosecondVector::from_vec(
values.duration_nanosecond_values,
)),
},
ConcreteDataType::Null(_) | ConcreteDataType::List(_) | ConcreteDataType::Dictionary(_) => {
unreachable!()
}
@@ -611,22 +553,22 @@ pub fn pb_values_to_values(data_type: &ConcreteDataType, values: Values) -> Vec<
.map(|v| Value::Date(v.into()))
.collect(),
ConcreteDataType::Timestamp(TimestampType::Second(_)) => values
.timestamp_second_values
.ts_second_values
.into_iter()
.map(|v| Value::Timestamp(Timestamp::new_second(v)))
.collect(),
ConcreteDataType::Timestamp(TimestampType::Millisecond(_)) => values
.timestamp_millisecond_values
.ts_millisecond_values
.into_iter()
.map(|v| Value::Timestamp(Timestamp::new_millisecond(v)))
.collect(),
ConcreteDataType::Timestamp(TimestampType::Microsecond(_)) => values
.timestamp_microsecond_values
.ts_microsecond_values
.into_iter()
.map(|v| Value::Timestamp(Timestamp::new_microsecond(v)))
.collect(),
ConcreteDataType::Timestamp(TimestampType::Nanosecond(_)) => values
.timestamp_nanosecond_values
.ts_nanosecond_values
.into_iter()
.map(|v| Value::Timestamp(Timestamp::new_nanosecond(v)))
.collect(),
@@ -672,26 +614,6 @@ pub fn pb_values_to_values(data_type: &ConcreteDataType, values: Values) -> Vec<
))
})
.collect(),
ConcreteDataType::Duration(DurationType::Second(_)) => values
.duration_second_values
.into_iter()
.map(|v| Value::Duration(Duration::new_second(v)))
.collect(),
ConcreteDataType::Duration(DurationType::Millisecond(_)) => values
.duration_millisecond_values
.into_iter()
.map(|v| Value::Duration(Duration::new_millisecond(v)))
.collect(),
ConcreteDataType::Duration(DurationType::Microsecond(_)) => values
.duration_microsecond_values
.into_iter()
.map(|v| Value::Duration(Duration::new_microsecond(v)))
.collect(),
ConcreteDataType::Duration(DurationType::Nanosecond(_)) => values
.duration_nanosecond_values
.into_iter()
.map(|v| Value::Duration(Duration::new_nanosecond(v)))
.collect(),
ConcreteDataType::Null(_) | ConcreteDataType::List(_) | ConcreteDataType::Dictionary(_) => {
unreachable!()
}
@@ -763,16 +685,16 @@ pub fn to_proto_value(value: Value) -> Option<v1::Value> {
},
Value::Timestamp(v) => match v.unit() {
TimeUnit::Second => v1::Value {
value_data: Some(ValueData::TimestampSecondValue(v.value())),
value_data: Some(ValueData::TsSecondValue(v.value())),
},
TimeUnit::Millisecond => v1::Value {
value_data: Some(ValueData::TimestampMillisecondValue(v.value())),
value_data: Some(ValueData::TsMillisecondValue(v.value())),
},
TimeUnit::Microsecond => v1::Value {
value_data: Some(ValueData::TimestampMicrosecondValue(v.value())),
value_data: Some(ValueData::TsMicrosecondValue(v.value())),
},
TimeUnit::Nanosecond => v1::Value {
value_data: Some(ValueData::TimestampNanosecondValue(v.value())),
value_data: Some(ValueData::TsNanosecondValue(v.value())),
},
},
Value::Time(v) => match v.unit() {
@@ -802,20 +724,6 @@ pub fn to_proto_value(value: Value) -> Option<v1::Value> {
)),
},
},
Value::Duration(v) => match v.unit() {
TimeUnit::Second => v1::Value {
value_data: Some(ValueData::DurationSecondValue(v.value())),
},
TimeUnit::Millisecond => v1::Value {
value_data: Some(ValueData::DurationMillisecondValue(v.value())),
},
TimeUnit::Microsecond => v1::Value {
value_data: Some(ValueData::DurationMicrosecondValue(v.value())),
},
TimeUnit::Nanosecond => v1::Value {
value_data: Some(ValueData::DurationNanosecondValue(v.value())),
},
},
Value::List(_) => return None,
};
@@ -842,10 +750,10 @@ pub fn proto_value_type(value: &v1::Value) -> Option<ColumnDataType> {
ValueData::StringValue(_) => ColumnDataType::String,
ValueData::DateValue(_) => ColumnDataType::Date,
ValueData::DatetimeValue(_) => ColumnDataType::Datetime,
ValueData::TimestampSecondValue(_) => ColumnDataType::TimestampSecond,
ValueData::TimestampMillisecondValue(_) => ColumnDataType::TimestampMillisecond,
ValueData::TimestampMicrosecondValue(_) => ColumnDataType::TimestampMicrosecond,
ValueData::TimestampNanosecondValue(_) => ColumnDataType::TimestampNanosecond,
ValueData::TsSecondValue(_) => ColumnDataType::TimestampSecond,
ValueData::TsMillisecondValue(_) => ColumnDataType::TimestampMillisecond,
ValueData::TsMicrosecondValue(_) => ColumnDataType::TimestampMicrosecond,
ValueData::TsNanosecondValue(_) => ColumnDataType::TimestampNanosecond,
ValueData::TimeSecondValue(_) => ColumnDataType::TimeSecond,
ValueData::TimeMillisecondValue(_) => ColumnDataType::TimeMillisecond,
ValueData::TimeMicrosecondValue(_) => ColumnDataType::TimeMicrosecond,
@@ -853,10 +761,6 @@ pub fn proto_value_type(value: &v1::Value) -> Option<ColumnDataType> {
ValueData::IntervalYearMonthValues(_) => ColumnDataType::IntervalYearMonth,
ValueData::IntervalDayTimeValues(_) => ColumnDataType::IntervalDayTime,
ValueData::IntervalMonthDayNanoValues(_) => ColumnDataType::IntervalMonthDayNano,
ValueData::DurationSecondValue(_) => ColumnDataType::DurationSecond,
ValueData::DurationMillisecondValue(_) => ColumnDataType::DurationMillisecond,
ValueData::DurationMicrosecondValue(_) => ColumnDataType::DurationMicrosecond,
ValueData::DurationNanosecondValue(_) => ColumnDataType::DurationNanosecond,
};
Some(value_type)
}
@@ -893,92 +797,15 @@ pub fn to_column_data_type(data_type: &ConcreteDataType) -> Option<ColumnDataTyp
ConcreteDataType::Time(TimeType::Millisecond(_)) => ColumnDataType::TimeMillisecond,
ConcreteDataType::Time(TimeType::Microsecond(_)) => ColumnDataType::TimeMicrosecond,
ConcreteDataType::Time(TimeType::Nanosecond(_)) => ColumnDataType::TimeNanosecond,
ConcreteDataType::Duration(DurationType::Second(_)) => ColumnDataType::DurationSecond,
ConcreteDataType::Duration(DurationType::Millisecond(_)) => {
ColumnDataType::DurationMillisecond
}
ConcreteDataType::Duration(DurationType::Microsecond(_)) => {
ColumnDataType::DurationMicrosecond
}
ConcreteDataType::Duration(DurationType::Nanosecond(_)) => {
ColumnDataType::DurationNanosecond
}
ConcreteDataType::Interval(IntervalType::YearMonth(_)) => ColumnDataType::IntervalYearMonth,
ConcreteDataType::Interval(IntervalType::MonthDayNano(_)) => {
ColumnDataType::IntervalMonthDayNano
}
ConcreteDataType::Interval(IntervalType::DayTime(_)) => ColumnDataType::IntervalDayTime,
ConcreteDataType::Null(_) | ConcreteDataType::List(_) | ConcreteDataType::Dictionary(_) => {
return None
}
ConcreteDataType::Null(_)
| ConcreteDataType::Interval(_)
| ConcreteDataType::List(_)
| ConcreteDataType::Dictionary(_) => return None,
};
Some(column_data_type)
}
pub fn vectors_to_rows<'a>(
columns: impl Iterator<Item = &'a VectorRef>,
row_count: usize,
) -> Vec<Row> {
let mut rows = vec![Row { values: vec![] }; row_count];
for column in columns {
for (row_index, row) in rows.iter_mut().enumerate() {
row.values.push(value_to_grpc_value(column.get(row_index)))
}
}
rows
}
pub fn value_to_grpc_value(value: Value) -> GrpcValue {
GrpcValue {
value_data: match value {
Value::Null => None,
Value::Boolean(v) => Some(ValueData::BoolValue(v)),
Value::UInt8(v) => Some(ValueData::U8Value(v as _)),
Value::UInt16(v) => Some(ValueData::U16Value(v as _)),
Value::UInt32(v) => Some(ValueData::U32Value(v)),
Value::UInt64(v) => Some(ValueData::U64Value(v)),
Value::Int8(v) => Some(ValueData::I8Value(v as _)),
Value::Int16(v) => Some(ValueData::I16Value(v as _)),
Value::Int32(v) => Some(ValueData::I32Value(v)),
Value::Int64(v) => Some(ValueData::I64Value(v)),
Value::Float32(v) => Some(ValueData::F32Value(*v)),
Value::Float64(v) => Some(ValueData::F64Value(*v)),
Value::String(v) => Some(ValueData::StringValue(v.as_utf8().to_string())),
Value::Binary(v) => Some(ValueData::BinaryValue(v.to_vec())),
Value::Date(v) => Some(ValueData::DateValue(v.val())),
Value::DateTime(v) => Some(ValueData::DatetimeValue(v.val())),
Value::Timestamp(v) => Some(match v.unit() {
TimeUnit::Second => ValueData::TimestampSecondValue(v.value()),
TimeUnit::Millisecond => ValueData::TimestampMillisecondValue(v.value()),
TimeUnit::Microsecond => ValueData::TimestampMicrosecondValue(v.value()),
TimeUnit::Nanosecond => ValueData::TimestampNanosecondValue(v.value()),
}),
Value::Time(v) => Some(match v.unit() {
TimeUnit::Second => ValueData::TimeSecondValue(v.value()),
TimeUnit::Millisecond => ValueData::TimeMillisecondValue(v.value()),
TimeUnit::Microsecond => ValueData::TimeMicrosecondValue(v.value()),
TimeUnit::Nanosecond => ValueData::TimeNanosecondValue(v.value()),
}),
Value::Interval(v) => Some(match v.unit() {
IntervalUnit::YearMonth => ValueData::IntervalYearMonthValues(v.to_i32()),
IntervalUnit::DayTime => ValueData::IntervalDayTimeValues(v.to_i64()),
IntervalUnit::MonthDayNano => {
ValueData::IntervalMonthDayNanoValues(convert_i128_to_interval(v.to_i128()))
}
}),
Value::Duration(v) => Some(match v.unit() {
TimeUnit::Second => ValueData::DurationSecondValue(v.value()),
TimeUnit::Millisecond => ValueData::DurationMillisecondValue(v.value()),
TimeUnit::Microsecond => ValueData::DurationMicrosecondValue(v.value()),
TimeUnit::Nanosecond => ValueData::DurationNanosecondValue(v.value()),
}),
Value::List(_) => unreachable!(),
},
}
}
/// Returns true if the column type is equal to expected type.
fn is_column_type_eq(column_type: ColumnDataType, expect_type: &ConcreteDataType) -> bool {
if let Some(expect) = to_column_data_type(expect_type) {
@@ -993,16 +820,14 @@ mod tests {
use std::sync::Arc;
use datatypes::types::{
DurationMillisecondType, DurationSecondType, Int32Type, IntervalDayTimeType,
IntervalMonthDayNanoType, IntervalYearMonthType, TimeMillisecondType, TimeSecondType,
TimestampMillisecondType, TimestampSecondType, UInt32Type,
IntervalDayTimeType, IntervalMonthDayNanoType, IntervalYearMonthType, TimeMillisecondType,
TimeSecondType, TimestampMillisecondType, TimestampSecondType,
};
use datatypes::vectors::{
BooleanVector, DurationMicrosecondVector, DurationMillisecondVector,
DurationNanosecondVector, DurationSecondVector, IntervalDayTimeVector,
IntervalMonthDayNanoVector, IntervalYearMonthVector, TimeMicrosecondVector,
TimeMillisecondVector, TimeNanosecondVector, TimeSecondVector, TimestampMicrosecondVector,
TimestampMillisecondVector, TimestampNanosecondVector, TimestampSecondVector, Vector,
BooleanVector, IntervalDayTimeVector, IntervalMonthDayNanoVector, IntervalYearMonthVector,
TimeMicrosecondVector, TimeMillisecondVector, TimeNanosecondVector, TimeSecondVector,
TimestampMicrosecondVector, TimestampMillisecondVector, TimestampNanosecondVector,
TimestampSecondVector, Vector,
};
use paste::paste;
@@ -1063,7 +888,7 @@ mod tests {
assert_eq!(2, values.capacity());
let values = values_with_capacity(ColumnDataType::TimestampMillisecond, 2);
let values = values.timestamp_millisecond_values;
let values = values.ts_millisecond_values;
assert_eq!(2, values.capacity());
let values = values_with_capacity(ColumnDataType::TimeMillisecond, 2);
@@ -1077,10 +902,6 @@ mod tests {
let values = values_with_capacity(ColumnDataType::IntervalMonthDayNano, 2);
let values = values.interval_month_day_nano_values;
assert_eq!(2, values.capacity());
let values = values_with_capacity(ColumnDataType::DurationMillisecond, 2);
let values = values.duration_millisecond_values;
assert_eq!(2, values.capacity());
}
#[test]
@@ -1165,10 +986,6 @@ mod tests {
ConcreteDataType::interval_datatype(IntervalUnit::MonthDayNano),
ColumnDataTypeWrapper(ColumnDataType::IntervalMonthDayNano).into()
);
assert_eq!(
ConcreteDataType::duration_millisecond_datatype(),
ColumnDataTypeWrapper(ColumnDataType::DurationMillisecond).into()
)
}
#[test]
@@ -1257,12 +1074,6 @@ mod tests {
.try_into()
.unwrap()
);
assert_eq!(
ColumnDataTypeWrapper(ColumnDataType::DurationMillisecond),
ConcreteDataType::duration_millisecond_datatype()
.try_into()
.unwrap()
);
let result: Result<ColumnDataTypeWrapper> = ConcreteDataType::null_datatype().try_into();
assert!(result.is_err());
@@ -1296,28 +1107,28 @@ mod tests {
push_vals(&mut column, 3, vector);
assert_eq!(
vec![1, 2, 3],
column.values.as_ref().unwrap().timestamp_nanosecond_values
column.values.as_ref().unwrap().ts_nanosecond_values
);
let vector = Arc::new(TimestampMillisecondVector::from_vec(vec![4, 5, 6]));
push_vals(&mut column, 3, vector);
assert_eq!(
vec![4, 5, 6],
column.values.as_ref().unwrap().timestamp_millisecond_values
column.values.as_ref().unwrap().ts_millisecond_values
);
let vector = Arc::new(TimestampMicrosecondVector::from_vec(vec![7, 8, 9]));
push_vals(&mut column, 3, vector);
assert_eq!(
vec![7, 8, 9],
column.values.as_ref().unwrap().timestamp_microsecond_values
column.values.as_ref().unwrap().ts_microsecond_values
);
let vector = Arc::new(TimestampSecondVector::from_vec(vec![10, 11, 12]));
push_vals(&mut column, 3, vector);
assert_eq!(
vec![10, 11, 12],
column.values.as_ref().unwrap().timestamp_second_values
column.values.as_ref().unwrap().ts_second_values
);
}
@@ -1406,47 +1217,6 @@ mod tests {
});
}
#[test]
fn test_column_put_duration_values() {
let mut column = Column {
column_name: "test".to_string(),
semantic_type: 0,
values: Some(Values {
..Default::default()
}),
null_mask: vec![],
datatype: 0,
};
let vector = Arc::new(DurationNanosecondVector::from_vec(vec![1, 2, 3]));
push_vals(&mut column, 3, vector);
assert_eq!(
vec![1, 2, 3],
column.values.as_ref().unwrap().duration_nanosecond_values
);
let vector = Arc::new(DurationMicrosecondVector::from_vec(vec![7, 8, 9]));
push_vals(&mut column, 3, vector);
assert_eq!(
vec![7, 8, 9],
column.values.as_ref().unwrap().duration_microsecond_values
);
let vector = Arc::new(DurationMillisecondVector::from_vec(vec![4, 5, 6]));
push_vals(&mut column, 3, vector);
assert_eq!(
vec![4, 5, 6],
column.values.as_ref().unwrap().duration_millisecond_values
);
let vector = Arc::new(DurationSecondVector::from_vec(vec![10, 11, 12]));
push_vals(&mut column, 3, vector);
assert_eq!(
vec![10, 11, 12],
column.values.as_ref().unwrap().duration_second_values
);
}
#[test]
fn test_column_put_vector() {
use crate::v1::SemanticType;
@@ -1487,7 +1257,7 @@ mod tests {
let actual = pb_values_to_values(
&ConcreteDataType::Timestamp(TimestampType::Second(TimestampSecondType)),
Values {
timestamp_second_values: vec![1_i64, 2_i64, 3_i64],
ts_second_values: vec![1_i64, 2_i64, 3_i64],
..Default::default()
},
);
@@ -1502,7 +1272,7 @@ mod tests {
let actual = pb_values_to_values(
&ConcreteDataType::Timestamp(TimestampType::Millisecond(TimestampMillisecondType)),
Values {
timestamp_millisecond_values: vec![1_i64, 2_i64, 3_i64],
ts_millisecond_values: vec![1_i64, 2_i64, 3_i64],
..Default::default()
},
);
@@ -1547,39 +1317,6 @@ mod tests {
assert_eq!(expect, actual);
}
#[test]
fn test_convert_duration_values() {
// second
let actual = pb_values_to_values(
&ConcreteDataType::Duration(DurationType::Second(DurationSecondType)),
Values {
duration_second_values: vec![1_i64, 2_i64, 3_i64],
..Default::default()
},
);
let expect = vec![
Value::Duration(Duration::new_second(1_i64)),
Value::Duration(Duration::new_second(2_i64)),
Value::Duration(Duration::new_second(3_i64)),
];
assert_eq!(expect, actual);
// millisecond
let actual = pb_values_to_values(
&ConcreteDataType::Duration(DurationType::Millisecond(DurationMillisecondType)),
Values {
duration_millisecond_values: vec![1_i64, 2_i64, 3_i64],
..Default::default()
},
);
let expect = vec![
Value::Duration(Duration::new_millisecond(1_i64)),
Value::Duration(Duration::new_millisecond(2_i64)),
Value::Duration(Duration::new_millisecond(3_i64)),
];
assert_eq!(expect, actual);
}
#[test]
fn test_convert_interval_values() {
// year_month
@@ -1787,76 +1524,4 @@ mod tests {
Value::DateTime(3.into())
]
);
#[test]
fn test_vectors_to_rows_for_different_types() {
let boolean_vec = BooleanVector::from_vec(vec![true, false, true]);
let int8_vec = PrimitiveVector::<Int8Type>::from_iter_values(vec![1, 2, 3]);
let int32_vec = PrimitiveVector::<Int32Type>::from_iter_values(vec![100, 200, 300]);
let uint8_vec = PrimitiveVector::<UInt8Type>::from_iter_values(vec![10, 20, 30]);
let uint32_vec = PrimitiveVector::<UInt32Type>::from_iter_values(vec![1000, 2000, 3000]);
let float32_vec = Float32Vector::from_vec(vec![1.1, 2.2, 3.3]);
let date_vec = DateVector::from_vec(vec![10, 20, 30]);
let string_vec = StringVector::from_vec(vec!["a", "b", "c"]);
let vector_refs: Vec<VectorRef> = vec![
Arc::new(boolean_vec),
Arc::new(int8_vec),
Arc::new(int32_vec),
Arc::new(uint8_vec),
Arc::new(uint32_vec),
Arc::new(float32_vec),
Arc::new(date_vec),
Arc::new(string_vec),
];
let result = vectors_to_rows(vector_refs.iter(), 3);
assert_eq!(result.len(), 3);
assert_eq!(result[0].values.len(), 8);
let values = result[0]
.values
.iter()
.map(|v| v.value_data.clone().unwrap())
.collect::<Vec<_>>();
assert_eq!(values[0], ValueData::BoolValue(true));
assert_eq!(values[1], ValueData::I8Value(1));
assert_eq!(values[2], ValueData::I32Value(100));
assert_eq!(values[3], ValueData::U8Value(10));
assert_eq!(values[4], ValueData::U32Value(1000));
assert_eq!(values[5], ValueData::F32Value(1.1));
assert_eq!(values[6], ValueData::DateValue(10));
assert_eq!(values[7], ValueData::StringValue("a".to_string()));
assert_eq!(result[1].values.len(), 8);
let values = result[1]
.values
.iter()
.map(|v| v.value_data.clone().unwrap())
.collect::<Vec<_>>();
assert_eq!(values[0], ValueData::BoolValue(false));
assert_eq!(values[1], ValueData::I8Value(2));
assert_eq!(values[2], ValueData::I32Value(200));
assert_eq!(values[3], ValueData::U8Value(20));
assert_eq!(values[4], ValueData::U32Value(2000));
assert_eq!(values[5], ValueData::F32Value(2.2));
assert_eq!(values[6], ValueData::DateValue(20));
assert_eq!(values[7], ValueData::StringValue("b".to_string()));
assert_eq!(result[2].values.len(), 8);
let values = result[2]
.values
.iter()
.map(|v| v.value_data.clone().unwrap())
.collect::<Vec<_>>();
assert_eq!(values[0], ValueData::BoolValue(true));
assert_eq!(values[1], ValueData::I8Value(3));
assert_eq!(values[2], ValueData::I32Value(300));
assert_eq!(values[3], ValueData::U8Value(30));
assert_eq!(values[4], ValueData::U32Value(3000));
assert_eq!(values[5], ValueData::F32Value(3.3));
assert_eq!(values[6], ValueData::DateValue(30));
assert_eq!(values[7], ValueData::StringValue("c".to_string()));
}
}

View File

@@ -12,9 +12,7 @@
// See the License for the specific language governing permissions and
// limitations under the License.
use std::collections::HashMap;
use datatypes::schema::{ColumnDefaultConstraint, ColumnSchema, COMMENT_KEY};
use datatypes::schema::{ColumnDefaultConstraint, ColumnSchema};
use snafu::ResultExt;
use crate::error::{self, Result};
@@ -22,7 +20,7 @@ use crate::helper::ColumnDataTypeWrapper;
use crate::v1::ColumnDef;
pub fn try_as_column_schema(column_def: &ColumnDef) -> Result<ColumnSchema> {
let data_type = ColumnDataTypeWrapper::try_new(column_def.data_type)?;
let data_type = ColumnDataTypeWrapper::try_new(column_def.datatype)?;
let constraint = if column_def.default_constraint.is_empty() {
None
@@ -36,17 +34,9 @@ pub fn try_as_column_schema(column_def: &ColumnDef) -> Result<ColumnSchema> {
)
};
let mut metadata = HashMap::new();
if !column_def.comment.is_empty() {
metadata.insert(COMMENT_KEY.to_string(), column_def.comment.clone());
}
Ok(
ColumnSchema::new(&column_def.name, data_type.into(), column_def.is_nullable)
.with_default_constraint(constraint)
.context(error::InvalidColumnDefaultConstraintSnafu {
column: &column_def.name,
})?
.with_metadata(metadata),
)
ColumnSchema::new(&column_def.name, data_type.into(), column_def.is_nullable)
.with_default_constraint(constraint)
.context(error::InvalidColumnDefaultConstraintSnafu {
column: &column_def.name,
})
}

View File

@@ -4,6 +4,8 @@ version.workspace = true
edition.workspace = true
license.workspace = true
# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
[features]
default = []
testing = []
@@ -12,7 +14,6 @@ testing = []
api.workspace = true
async-trait.workspace = true
common-error.workspace = true
common-macro.workspace = true
digest = "0.10"
hex = { version = "0.4" }
secrecy = { version = "0.8", features = ["serde", "alloc"] }

View File

@@ -26,7 +26,7 @@ use crate::{UserInfoRef, UserProviderRef};
pub(crate) const DEFAULT_USERNAME: &str = "greptime";
/// construct a [`UserInfo`](crate::user_info::UserInfo) impl with name
/// construct a [`UserInfo`] impl with name
/// use default username `greptime` if None is provided
pub fn userinfo_by_name(username: Option<String>) -> UserInfoRef {
DefaultUserInfo::with_name(username.unwrap_or_else(|| DEFAULT_USERNAME.to_string()))

View File

@@ -14,12 +14,10 @@
use common_error::ext::{BoxedError, ErrorExt};
use common_error::status_code::StatusCode;
use common_macro::stack_trace_debug;
use snafu::{Location, Snafu};
#[derive(Snafu)]
#[derive(Debug, Snafu)]
#[snafu(visibility(pub))]
#[stack_trace_debug]
pub enum Error {
#[snafu(display("Invalid config value: {}, {}", value, msg))]
InvalidConfig { value: String, msg: String },
@@ -30,14 +28,13 @@ pub enum Error {
#[snafu(display("Internal state error: {}", msg))]
InternalState { msg: String },
#[snafu(display("IO error"))]
#[snafu(display("IO error, source: {}", source))]
Io {
#[snafu(source)]
error: std::io::Error,
source: std::io::Error,
location: Location,
},
#[snafu(display("Auth failed"))]
#[snafu(display("Auth failed, source: {}", source))]
AuthBackend {
location: Location,
source: BoxedError,

View File

@@ -22,15 +22,15 @@ use crate::UserInfoRef;
pub trait UserProvider: Send + Sync {
fn name(&self) -> &str;
/// Checks whether a user is valid and allowed to access the database.
/// [`authenticate`] checks whether a user is valid and allowed to access the database.
async fn authenticate(&self, id: Identity<'_>, password: Password<'_>) -> Result<UserInfoRef>;
/// Checks whether a connection request
/// [`authorize`] checks whether a connection request
/// from a certain user to a certain catalog/schema is legal.
/// This method should be called after [authenticate()](UserProvider::authenticate()).
/// This method should be called after [`authenticate`].
async fn authorize(&self, catalog: &str, schema: &str, user_info: &UserInfoRef) -> Result<()>;
/// Combination of [authenticate()](UserProvider::authenticate()) and [authorize()](UserProvider::authorize()).
/// [`auth`] is a combination of [`authenticate`] and [`authorize`].
/// In most cases it's preferred for both convenience and performance.
async fn auth(
&self,

View File

@@ -16,7 +16,6 @@ async-trait = "0.1"
common-catalog = { workspace = true }
common-error = { workspace = true }
common-grpc = { workspace = true }
common-macro = { workspace = true }
common-meta = { workspace = true }
common-query = { workspace = true }
common-recordbatch = { workspace = true }
@@ -31,9 +30,8 @@ futures-util.workspace = true
lazy_static.workspace = true
meta-client = { workspace = true }
metrics.workspace = true
moka = { workspace = true, features = ["future"] }
moka = { version = "0.11", features = ["future"] }
parking_lot = "0.12"
partition.workspace = true
regex.workspace = true
serde.workspace = true
serde_json = "1.0"
@@ -48,6 +46,7 @@ catalog = { workspace = true, features = ["testing"] }
chrono.workspace = true
common-test-util = { workspace = true }
log-store = { workspace = true }
mito = { workspace = true, features = ["test"] }
object-store = { workspace = true }
storage = { workspace = true }
tokio.workspace = true

View File

@@ -17,48 +17,54 @@ use std::fmt::Debug;
use common_error::ext::{BoxedError, ErrorExt};
use common_error::status_code::StatusCode;
use common_macro::stack_trace_debug;
use datafusion::error::DataFusionError;
use datatypes::prelude::ConcreteDataType;
use snafu::{Location, Snafu};
use table::metadata::TableId;
use tokio::task::JoinError;
#[derive(Snafu)]
use crate::DeregisterTableRequest;
#[derive(Debug, Snafu)]
#[snafu(visibility(pub))]
#[stack_trace_debug]
pub enum Error {
#[snafu(display("Failed to list catalogs"))]
#[snafu(display("Failed to list catalogs, source: {}", source))]
ListCatalogs {
location: Location,
source: BoxedError,
},
#[snafu(display("Failed to list {}'s schemas", catalog))]
#[snafu(display("Failed to list {}'s schemas, source: {}", catalog, source))]
ListSchemas {
location: Location,
catalog: String,
source: BoxedError,
},
#[snafu(display("Failed to re-compile script due to internal error"))]
#[snafu(display(
"Failed to re-compile script due to internal error, source: {}",
source
))]
CompileScriptInternal {
location: Location,
source: BoxedError,
},
#[snafu(display("Failed to open system catalog table"))]
#[snafu(display("Failed to open system catalog table, source: {}", source))]
OpenSystemCatalog {
location: Location,
source: table::error::Error,
},
#[snafu(display("Failed to create system catalog table"))]
#[snafu(display("Failed to create system catalog table, source: {}", source))]
CreateSystemCatalog {
location: Location,
source: table::error::Error,
},
#[snafu(display("Failed to create table, table info: {}", table_info))]
#[snafu(display(
"Failed to create table, table info: {}, source: {}",
table_info,
source
))]
CreateTable {
table_info: String,
location: Location,
@@ -92,14 +98,13 @@ pub enum Error {
#[snafu(display("Catalog value is not present"))]
EmptyValue { location: Location },
#[snafu(display("Failed to deserialize value"))]
#[snafu(display("Failed to deserialize value, source: {}", source))]
ValueDeserialize {
#[snafu(source)]
error: serde_json::error::Error,
source: serde_json::error::Error,
location: Location,
},
#[snafu(display("Table engine not found: {}", engine_name))]
#[snafu(display("Table engine not found: {}, source: {}", engine_name, source))]
TableEngineNotFound {
engine_name: String,
location: Location,
@@ -137,18 +142,15 @@ pub enum Error {
#[snafu(display("Operation {} not supported", op))]
NotSupported { op: String, location: Location },
#[snafu(display("Failed to open table {table_id}"))]
#[snafu(display("Failed to open table, table info: {}, source: {}", table_info, source))]
OpenTable {
table_id: TableId,
table_info: String,
location: Location,
source: table::error::Error,
},
#[snafu(display("Failed to open table in parallel"))]
ParallelOpenTable {
#[snafu(source)]
error: JoinError,
},
#[snafu(display("Failed to open table in parallel, source: {}", source))]
ParallelOpenTable { source: JoinError },
#[snafu(display("Table not found while opening table, table info: {}", table_info))]
TableNotFound {
@@ -162,52 +164,72 @@ pub enum Error {
source: common_recordbatch::error::Error,
},
#[snafu(display("Failed to create recordbatch"))]
#[snafu(display("Failed to create recordbatch, source: {}", source))]
CreateRecordBatch {
location: Location,
source: common_recordbatch::error::Error,
},
#[snafu(display("Failed to insert table creation record to system catalog"))]
#[snafu(display(
"Failed to insert table creation record to system catalog, source: {}",
source
))]
InsertCatalogRecord {
location: Location,
source: table::error::Error,
},
#[snafu(display("Failed to scan system catalog table"))]
#[snafu(display(
"Failed to deregister table, request: {:?}, source: {}",
request,
source
))]
DeregisterTable {
request: DeregisterTableRequest,
location: Location,
source: table::error::Error,
},
#[snafu(display("Illegal catalog manager state: {}", msg))]
IllegalManagerState { location: Location, msg: String },
#[snafu(display("Failed to scan system catalog table, source: {}", source))]
SystemCatalogTableScan {
location: Location,
source: table::error::Error,
},
#[snafu(display(""))]
#[snafu(display("{source}"))]
Internal {
location: Location,
source: BoxedError,
},
#[snafu(display("Failed to upgrade weak catalog manager reference"))]
#[snafu(display(
"Failed to upgrade weak catalog manager reference. location: {}",
location
))]
UpgradeWeakCatalogManagerRef { location: Location },
#[snafu(display("Failed to execute system catalog table scan"))]
#[snafu(display("Failed to execute system catalog table scan, source: {}", source))]
SystemCatalogTableScanExec {
location: Location,
source: common_query::error::Error,
},
#[snafu(display("Cannot parse catalog value"))]
#[snafu(display("Cannot parse catalog value, source: {}", source))]
InvalidCatalogValue {
location: Location,
source: common_catalog::error::Error,
},
#[snafu(display("Failed to perform metasrv operation"))]
#[snafu(display("Failed to perform metasrv operation, source: {}", source))]
MetaSrv {
location: Location,
source: meta_client::error::Error,
},
#[snafu(display("Invalid table info in catalog"))]
#[snafu(display("Invalid table info in catalog, source: {}", source))]
InvalidTableInfoInCatalog {
location: Location,
source: datatypes::error::Error,
@@ -216,14 +238,17 @@ pub enum Error {
#[snafu(display("Illegal access to catalog: {} and schema: {}", catalog, schema))]
QueryAccessDenied { catalog: String, schema: String },
#[snafu(display(""))]
#[snafu(display("Invalid system table definition: {err_msg}"))]
InvalidSystemTableDef { err_msg: String, location: Location },
#[snafu(display("{}: {}", msg, source))]
Datafusion {
#[snafu(source)]
error: DataFusionError,
msg: String,
source: DataFusionError,
location: Location,
},
#[snafu(display("Table schema mismatch"))]
#[snafu(display("Table schema mismatch, source: {}", source))]
TableSchemaMismatch {
location: Location,
source: table::error::Error,
@@ -232,7 +257,7 @@ pub enum Error {
#[snafu(display("A generic error has occurred, msg: {}", msg))]
Generic { msg: String, location: Location },
#[snafu(display("Table metadata manager error"))]
#[snafu(display("Table metadata manager error: {}", source))]
TableMetadataManager {
source: common_meta::error::Error,
location: Location,
@@ -247,8 +272,10 @@ impl ErrorExt for Error {
Error::InvalidKey { .. }
| Error::SchemaNotFound { .. }
| Error::TableNotFound { .. }
| Error::IllegalManagerState { .. }
| Error::CatalogNotFound { .. }
| Error::InvalidEntryType { .. }
| Error::InvalidSystemTableDef { .. }
| Error::ParallelOpenTable { .. } => StatusCode::Unexpected,
Error::SystemCatalog { .. }
@@ -279,6 +306,7 @@ impl ErrorExt for Error {
| Error::InsertCatalogRecord { source, .. }
| Error::OpenTable { source, .. }
| Error::CreateTable { source, .. }
| Error::DeregisterTable { source, .. }
| Error::TableSchemaMismatch { source, .. } => source.status_code(),
Error::MetaSrv { source, .. } => source.status_code(),

View File

@@ -158,7 +158,7 @@ impl InformationSchemaColumnsBuilder {
for schema_name in catalog_manager.schema_names(&catalog_name).await? {
if !catalog_manager
.schema_exists(&catalog_name, &schema_name)
.schema_exist(&catalog_name, &schema_name)
.await?
{
continue;

View File

@@ -154,7 +154,7 @@ impl InformationSchemaTablesBuilder {
for schema_name in catalog_manager.schema_names(&catalog_name).await? {
if !catalog_manager
.schema_exists(&catalog_name, &schema_name)
.schema_exist(&catalog_name, &schema_name)
.await?
{
continue;

View File

@@ -1,292 +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::collections::BTreeSet;
use std::sync::{Arc, Weak};
use common_catalog::consts::{DEFAULT_SCHEMA_NAME, INFORMATION_SCHEMA_NAME, NUMBERS_TABLE_ID};
use common_error::ext::BoxedError;
use common_meta::cache_invalidator::{CacheInvalidator, CacheInvalidatorRef, Context};
use common_meta::datanode_manager::DatanodeManagerRef;
use common_meta::error::Result as MetaResult;
use common_meta::key::catalog_name::CatalogNameKey;
use common_meta::key::schema_name::SchemaNameKey;
use common_meta::key::table_name::TableNameKey;
use common_meta::key::{TableMetadataManager, TableMetadataManagerRef};
use common_meta::kv_backend::KvBackendRef;
use common_meta::table_name::TableName;
use futures_util::TryStreamExt;
use partition::manager::{PartitionRuleManager, PartitionRuleManagerRef};
use snafu::prelude::*;
use table::dist_table::DistTable;
use table::metadata::TableId;
use table::table::numbers::{NumbersTable, NUMBERS_TABLE_NAME};
use table::TableRef;
use crate::error::{
self as catalog_err, ListCatalogsSnafu, ListSchemasSnafu, Result as CatalogResult,
TableMetadataManagerSnafu,
};
use crate::information_schema::{InformationSchemaProvider, COLUMNS, TABLES};
use crate::CatalogManager;
/// Access all existing catalog, schema and tables.
///
/// The result comes from two source, all the user tables are presented in
/// a kv-backend which persists the metadata of a table. And system tables
/// comes from `SystemCatalog`, which is static and read-only.
#[derive(Clone)]
pub struct KvBackendCatalogManager {
// TODO(LFC): Maybe use a real implementation for Standalone mode.
// Now we use `NoopKvCacheInvalidator` for Standalone mode. In Standalone mode, the KV backend
// is implemented by RaftEngine. Maybe we need a cache for it?
cache_invalidator: CacheInvalidatorRef,
partition_manager: PartitionRuleManagerRef,
table_metadata_manager: TableMetadataManagerRef,
datanode_manager: DatanodeManagerRef,
/// A sub-CatalogManager that handles system tables
system_catalog: SystemCatalog,
}
#[async_trait::async_trait]
impl CacheInvalidator for KvBackendCatalogManager {
async fn invalidate_table_name(&self, ctx: &Context, table_name: TableName) -> MetaResult<()> {
self.cache_invalidator
.invalidate_table_name(ctx, table_name)
.await
}
async fn invalidate_table_id(&self, ctx: &Context, table_id: TableId) -> MetaResult<()> {
self.cache_invalidator
.invalidate_table_id(ctx, table_id)
.await
}
}
impl KvBackendCatalogManager {
pub fn new(
backend: KvBackendRef,
cache_invalidator: CacheInvalidatorRef,
datanode_manager: DatanodeManagerRef,
) -> Arc<Self> {
Arc::new_cyclic(|me| Self {
partition_manager: Arc::new(PartitionRuleManager::new(backend.clone())),
table_metadata_manager: Arc::new(TableMetadataManager::new(backend)),
cache_invalidator,
datanode_manager,
system_catalog: SystemCatalog {
catalog_manager: me.clone(),
},
})
}
pub fn partition_manager(&self) -> PartitionRuleManagerRef {
self.partition_manager.clone()
}
pub fn table_metadata_manager_ref(&self) -> &TableMetadataManagerRef {
&self.table_metadata_manager
}
pub fn datanode_manager(&self) -> DatanodeManagerRef {
self.datanode_manager.clone()
}
}
#[async_trait::async_trait]
impl CatalogManager for KvBackendCatalogManager {
async fn catalog_names(&self) -> CatalogResult<Vec<String>> {
let stream = self
.table_metadata_manager
.catalog_manager()
.catalog_names()
.await;
let keys = stream
.try_collect::<Vec<_>>()
.await
.map_err(BoxedError::new)
.context(ListCatalogsSnafu)?;
Ok(keys)
}
async fn schema_names(&self, catalog: &str) -> CatalogResult<Vec<String>> {
let stream = self
.table_metadata_manager
.schema_manager()
.schema_names(catalog)
.await;
let mut keys = stream
.try_collect::<BTreeSet<_>>()
.await
.map_err(BoxedError::new)
.context(ListSchemasSnafu { catalog })?
.into_iter()
.collect::<Vec<_>>();
keys.extend_from_slice(&self.system_catalog.schema_names());
Ok(keys)
}
async fn table_names(&self, catalog: &str, schema: &str) -> CatalogResult<Vec<String>> {
let mut tables = self
.table_metadata_manager
.table_name_manager()
.tables(catalog, schema)
.await
.context(TableMetadataManagerSnafu)?
.into_iter()
.map(|(k, _)| k)
.collect::<Vec<String>>();
tables.extend_from_slice(&self.system_catalog.table_names(schema));
Ok(tables)
}
async fn catalog_exists(&self, catalog: &str) -> CatalogResult<bool> {
self.table_metadata_manager
.catalog_manager()
.exists(CatalogNameKey::new(catalog))
.await
.context(TableMetadataManagerSnafu)
}
async fn schema_exists(&self, catalog: &str, schema: &str) -> CatalogResult<bool> {
if self.system_catalog.schema_exist(schema) {
return Ok(true);
}
self.table_metadata_manager
.schema_manager()
.exists(SchemaNameKey::new(catalog, schema))
.await
.context(TableMetadataManagerSnafu)
}
async fn table_exists(&self, catalog: &str, schema: &str, table: &str) -> CatalogResult<bool> {
if self.system_catalog.table_exist(schema, table) {
return Ok(true);
}
let key = TableNameKey::new(catalog, schema, table);
self.table_metadata_manager
.table_name_manager()
.get(key)
.await
.context(TableMetadataManagerSnafu)
.map(|x| x.is_some())
}
async fn table(
&self,
catalog: &str,
schema: &str,
table_name: &str,
) -> CatalogResult<Option<TableRef>> {
if let Some(table) = self.system_catalog.table(catalog, schema, table_name) {
return Ok(Some(table));
}
let key = TableNameKey::new(catalog, schema, table_name);
let Some(table_name_value) = self
.table_metadata_manager
.table_name_manager()
.get(key)
.await
.context(TableMetadataManagerSnafu)?
else {
return Ok(None);
};
let table_id = table_name_value.table_id();
let Some(table_info_value) = self
.table_metadata_manager
.table_info_manager()
.get(table_id)
.await
.context(TableMetadataManagerSnafu)?
.map(|v| v.into_inner())
else {
return Ok(None);
};
let table_info = Arc::new(
table_info_value
.table_info
.try_into()
.context(catalog_err::InvalidTableInfoInCatalogSnafu)?,
);
Ok(Some(DistTable::table(table_info)))
}
fn as_any(&self) -> &dyn Any {
self
}
}
// TODO: This struct can hold a static map of all system tables when
// the upper layer (e.g., procedure) can inform the catalog manager
// a new catalog is created.
/// Existing system tables:
/// - public.numbers
/// - information_schema.tables
/// - information_schema.columns
#[derive(Clone)]
struct SystemCatalog {
catalog_manager: Weak<KvBackendCatalogManager>,
}
impl SystemCatalog {
fn schema_names(&self) -> Vec<String> {
vec![INFORMATION_SCHEMA_NAME.to_string()]
}
fn table_names(&self, schema: &str) -> Vec<String> {
if schema == INFORMATION_SCHEMA_NAME {
vec![TABLES.to_string(), COLUMNS.to_string()]
} else if schema == DEFAULT_SCHEMA_NAME {
vec![NUMBERS_TABLE_NAME.to_string()]
} else {
vec![]
}
}
fn schema_exist(&self, schema: &str) -> bool {
schema == INFORMATION_SCHEMA_NAME
}
fn table_exist(&self, schema: &str, table: &str) -> bool {
if schema == INFORMATION_SCHEMA_NAME {
table == TABLES || table == COLUMNS
} else if schema == DEFAULT_SCHEMA_NAME {
table == NUMBERS_TABLE_NAME
} else {
false
}
}
fn table(&self, catalog: &str, schema: &str, table_name: &str) -> Option<TableRef> {
if schema == INFORMATION_SCHEMA_NAME {
let information_schema_provider =
InformationSchemaProvider::new(catalog.to_string(), self.catalog_manager.clone());
information_schema_provider.table(table_name)
} else if schema == DEFAULT_SCHEMA_NAME && table_name == NUMBERS_TABLE_NAME {
Some(NumbersTable::table(NUMBERS_TABLE_ID))
} else {
None
}
}
}

View File

@@ -17,38 +17,79 @@
#![feature(try_blocks)]
use std::any::Any;
use std::collections::HashMap;
use std::fmt::{Debug, Formatter};
use std::sync::Arc;
use futures::future::BoxFuture;
use table::metadata::TableId;
use api::v1::meta::{RegionStat, TableIdent, TableName};
use common_telemetry::{info, warn};
use snafu::ResultExt;
use table::engine::{EngineContext, TableEngineRef};
use table::metadata::{TableId, TableType};
use table::requests::CreateTableRequest;
use table::TableRef;
use crate::error::Result;
use crate::error::{CreateTableSnafu, Result};
pub mod error;
pub mod information_schema;
pub mod kvbackend;
pub mod memory;
pub mod local;
mod metrics;
pub mod remote;
pub mod system;
pub mod table_source;
pub mod tables;
#[async_trait::async_trait]
pub trait CatalogManager: Send + Sync {
fn as_any(&self) -> &dyn Any;
/// Starts a catalog manager.
async fn start(&self) -> Result<()>;
/// Registers a catalog to catalog manager, returns whether the catalog exist before.
async fn register_catalog(self: Arc<Self>, name: String) -> Result<bool>;
/// Register a schema with catalog name and schema name. Retuens whether the
/// schema registered.
///
/// # Errors
///
/// This method will/should fail if catalog not exist
async fn register_schema(&self, request: RegisterSchemaRequest) -> Result<bool>;
/// Deregisters a database within given catalog/schema to catalog manager
async fn deregister_schema(&self, request: DeregisterSchemaRequest) -> Result<bool>;
/// Registers a table within given catalog/schema to catalog manager,
/// returns whether the table registered.
///
/// # Errors
///
/// This method will/should fail if catalog or schema not exist
async fn register_table(&self, request: RegisterTableRequest) -> Result<bool>;
/// Deregisters a table within given catalog/schema to catalog manager
async fn deregister_table(&self, request: DeregisterTableRequest) -> Result<()>;
/// Rename a table to [RenameTableRequest::new_table_name], returns whether the table is renamed.
async fn rename_table(&self, request: RenameTableRequest) -> Result<bool>;
/// Register a system table, should be called before starting the manager.
async fn register_system_table(&self, request: RegisterSystemTableRequest)
-> error::Result<()>;
async fn catalog_names(&self) -> Result<Vec<String>>;
async fn schema_names(&self, catalog: &str) -> Result<Vec<String>>;
async fn table_names(&self, catalog: &str, schema: &str) -> Result<Vec<String>>;
async fn catalog_exists(&self, catalog: &str) -> Result<bool>;
async fn catalog_exist(&self, catalog: &str) -> Result<bool>;
async fn schema_exists(&self, catalog: &str, schema: &str) -> Result<bool>;
async fn schema_exist(&self, catalog: &str, schema: &str) -> Result<bool>;
async fn table_exists(&self, catalog: &str, schema: &str, table: &str) -> Result<bool>;
async fn table_exist(&self, catalog: &str, schema: &str, table: &str) -> Result<bool>;
/// Returns the table by catalog, schema and table name.
async fn table(
@@ -62,8 +103,7 @@ pub trait CatalogManager: Send + Sync {
pub type CatalogManagerRef = Arc<dyn CatalogManager>;
/// Hook called after system table opening.
pub type OpenSystemTableHook =
Box<dyn Fn(TableRef) -> BoxFuture<'static, Result<()>> + Send + Sync>;
pub type OpenSystemTableHook = Arc<dyn Fn(TableRef) -> Result<()> + Send + Sync>;
/// Register system table request:
/// - When system table is already created and registered, the hook will be called
@@ -122,3 +162,120 @@ pub struct RegisterSchemaRequest {
pub catalog: String,
pub schema: String,
}
pub(crate) async fn handle_system_table_request<'a, M: CatalogManager>(
manager: &'a M,
engine: TableEngineRef,
sys_table_requests: &'a mut Vec<RegisterSystemTableRequest>,
) -> Result<()> {
for req in sys_table_requests.drain(..) {
let catalog_name = &req.create_table_request.catalog_name;
let schema_name = &req.create_table_request.schema_name;
let table_name = &req.create_table_request.table_name;
let table_id = req.create_table_request.id;
let table = manager.table(catalog_name, schema_name, table_name).await?;
let table = if let Some(table) = table {
table
} else {
let table = engine
.create_table(&EngineContext::default(), req.create_table_request.clone())
.await
.with_context(|_| CreateTableSnafu {
table_info: common_catalog::format_full_table_name(
catalog_name,
schema_name,
table_name,
),
})?;
let _ = manager
.register_table(RegisterTableRequest {
catalog: catalog_name.clone(),
schema: schema_name.clone(),
table_name: table_name.clone(),
table_id,
table: table.clone(),
})
.await?;
info!("Created and registered system table: {table_name}");
table
};
if let Some(hook) = req.open_hook {
(hook)(table)?;
}
}
Ok(())
}
/// The stat of regions in the datanode node.
/// The number of regions can be got from len of vec.
///
/// Ignores any errors occurred during iterating regions. The intention of this method is to
/// collect region stats that will be carried in Datanode's heartbeat to Metasrv, so it's a
/// "try our best" job.
pub async fn datanode_stat(catalog_manager: &CatalogManagerRef) -> (u64, Vec<RegionStat>) {
let mut region_number: u64 = 0;
let mut region_stats = Vec::new();
let Ok(catalog_names) = catalog_manager.catalog_names().await else {
return (region_number, region_stats);
};
for catalog_name in catalog_names {
let Ok(schema_names) = catalog_manager.schema_names(&catalog_name).await else {
continue;
};
for schema_name in schema_names {
let Ok(table_names) = catalog_manager
.table_names(&catalog_name, &schema_name)
.await
else {
continue;
};
for table_name in table_names {
let Ok(Some(table)) = catalog_manager
.table(&catalog_name, &schema_name, &table_name)
.await
else {
continue;
};
if table.table_type() != TableType::Base {
continue;
}
let table_info = table.table_info();
let region_numbers = &table_info.meta.region_numbers;
region_number += region_numbers.len() as u64;
let engine = &table_info.meta.engine;
let table_id = table_info.ident.table_id;
match table.region_stats() {
Ok(stats) => {
let stats = stats.into_iter().map(|stat| RegionStat {
region_id: stat.region_id,
table_ident: Some(TableIdent {
table_id,
table_name: Some(TableName {
catalog_name: catalog_name.clone(),
schema_name: schema_name.clone(),
table_name: table_name.clone(),
}),
engine: engine.clone(),
}),
approximate_bytes: stat.disk_usage_bytes as i64,
attrs: HashMap::from([("engine_name".to_owned(), engine.clone())]),
..Default::default()
});
region_stats.extend(stats);
}
Err(e) => {
warn!("Failed to get region status, err: {:?}", e);
}
};
}
}
}
(region_number, region_stats)
}

19
src/catalog/src/local.rs Normal file
View File

@@ -0,0 +1,19 @@
// Copyright 2023 Greptime Team
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
pub mod manager;
pub mod memory;
pub use manager::LocalCatalogManager;
pub use memory::{new_memory_catalog_manager, MemoryCatalogManager};

View File

@@ -0,0 +1,633 @@
// 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::sync::atomic::{AtomicU32, Ordering};
use std::sync::Arc;
use common_catalog::consts::{
DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, INFORMATION_SCHEMA_NAME, MIN_USER_TABLE_ID,
MITO_ENGINE, NUMBERS_TABLE_ID, SYSTEM_CATALOG_NAME, SYSTEM_CATALOG_TABLE_ID,
SYSTEM_CATALOG_TABLE_NAME,
};
use common_catalog::format_full_table_name;
use common_recordbatch::{RecordBatch, SendableRecordBatchStream};
use common_telemetry::{error, info};
use datatypes::prelude::ScalarVector;
use datatypes::vectors::{BinaryVector, UInt8Vector};
use futures_util::lock::Mutex;
use metrics::increment_gauge;
use snafu::{ensure, OptionExt, ResultExt};
use table::engine::manager::TableEngineManagerRef;
use table::engine::EngineContext;
use table::metadata::TableId;
use table::requests::OpenTableRequest;
use table::table::numbers::{NumbersTable, NUMBERS_TABLE_NAME};
use table::table::TableIdProvider;
use table::TableRef;
use crate::error::{
self, CatalogNotFoundSnafu, IllegalManagerStateSnafu, OpenTableSnafu, ReadSystemCatalogSnafu,
Result, SchemaExistsSnafu, SchemaNotFoundSnafu, SystemCatalogSnafu,
SystemCatalogTypeMismatchSnafu, TableEngineNotFoundSnafu, TableExistsSnafu, TableNotExistSnafu,
TableNotFoundSnafu, UnimplementedSnafu,
};
use crate::local::memory::MemoryCatalogManager;
use crate::system::{
decode_system_catalog, Entry, SystemCatalogTable, TableEntry, ENTRY_TYPE_INDEX, KEY_INDEX,
VALUE_INDEX,
};
use crate::tables::SystemCatalog;
use crate::{
handle_system_table_request, CatalogManager, DeregisterSchemaRequest, DeregisterTableRequest,
RegisterSchemaRequest, RegisterSystemTableRequest, RegisterTableRequest, RenameTableRequest,
};
/// A `CatalogManager` consists of a system catalog and a bunch of user catalogs.
pub struct LocalCatalogManager {
system: Arc<SystemCatalog>,
catalogs: Arc<MemoryCatalogManager>,
engine_manager: TableEngineManagerRef,
next_table_id: AtomicU32,
init_lock: Mutex<bool>,
register_lock: Mutex<()>,
system_table_requests: Mutex<Vec<RegisterSystemTableRequest>>,
}
impl LocalCatalogManager {
/// Create a new [CatalogManager] with given user catalogs and mito engine
pub async fn try_new(engine_manager: TableEngineManagerRef) -> Result<Self> {
let engine = engine_manager
.engine(MITO_ENGINE)
.context(TableEngineNotFoundSnafu {
engine_name: MITO_ENGINE,
})?;
let table = SystemCatalogTable::new(engine.clone()).await?;
let memory_catalog_manager = crate::local::memory::new_memory_catalog_manager()?;
let system_catalog = Arc::new(SystemCatalog::new(table));
Ok(Self {
system: system_catalog,
catalogs: memory_catalog_manager,
engine_manager,
next_table_id: AtomicU32::new(MIN_USER_TABLE_ID),
init_lock: Mutex::new(false),
register_lock: Mutex::new(()),
system_table_requests: Mutex::new(Vec::default()),
})
}
/// Scan all entries from system catalog table
pub async fn init(&self) -> Result<()> {
self.init_system_catalog().await?;
let system_records = self.system.information_schema.system.records().await?;
let entries = self.collect_system_catalog_entries(system_records).await?;
let max_table_id = self.handle_system_catalog_entries(entries).await?;
info!(
"All system catalog entries processed, max table id: {}",
max_table_id
);
self.next_table_id
.store((max_table_id + 1).max(MIN_USER_TABLE_ID), Ordering::Relaxed);
*self.init_lock.lock().await = true;
// Processing system table hooks
let mut sys_table_requests = self.system_table_requests.lock().await;
let engine = self
.engine_manager
.engine(MITO_ENGINE)
.context(TableEngineNotFoundSnafu {
engine_name: MITO_ENGINE,
})?;
handle_system_table_request(self, engine, &mut sys_table_requests).await?;
Ok(())
}
async fn init_system_catalog(&self) -> Result<()> {
// register default catalog and default schema
self.catalogs
.register_catalog_sync(DEFAULT_CATALOG_NAME.to_string())?;
self.catalogs.register_schema_sync(RegisterSchemaRequest {
catalog: DEFAULT_CATALOG_NAME.to_string(),
schema: DEFAULT_SCHEMA_NAME.to_string(),
})?;
// register SystemCatalogTable
self.catalogs
.register_catalog_sync(SYSTEM_CATALOG_NAME.to_string())?;
self.catalogs.register_schema_sync(RegisterSchemaRequest {
catalog: SYSTEM_CATALOG_NAME.to_string(),
schema: INFORMATION_SCHEMA_NAME.to_string(),
})?;
let register_table_req = RegisterTableRequest {
catalog: SYSTEM_CATALOG_NAME.to_string(),
schema: INFORMATION_SCHEMA_NAME.to_string(),
table_name: SYSTEM_CATALOG_TABLE_NAME.to_string(),
table_id: SYSTEM_CATALOG_TABLE_ID,
table: self.system.information_schema.system.as_table_ref(),
};
self.catalogs.register_table(register_table_req).await?;
// Add numbers table for test
let register_number_table_req = RegisterTableRequest {
catalog: DEFAULT_CATALOG_NAME.to_string(),
schema: DEFAULT_SCHEMA_NAME.to_string(),
table_name: NUMBERS_TABLE_NAME.to_string(),
table_id: NUMBERS_TABLE_ID,
table: NumbersTable::table(NUMBERS_TABLE_ID),
};
self.catalogs
.register_table(register_number_table_req)
.await?;
Ok(())
}
/// Collect stream of system catalog entries to `Vec<Entry>`
async fn collect_system_catalog_entries(
&self,
stream: SendableRecordBatchStream,
) -> Result<Vec<Entry>> {
let record_batch = common_recordbatch::util::collect(stream)
.await
.context(ReadSystemCatalogSnafu)?;
let rbs = record_batch
.into_iter()
.map(Self::record_batch_to_entry)
.collect::<Result<Vec<_>>>()?;
Ok(rbs.into_iter().flat_map(Vec::into_iter).collect::<_>())
}
/// Convert `RecordBatch` to a vector of `Entry`.
fn record_batch_to_entry(rb: RecordBatch) -> Result<Vec<Entry>> {
ensure!(
rb.num_columns() >= 6,
SystemCatalogSnafu {
msg: format!("Length mismatch: {}", rb.num_columns())
}
);
let entry_type = rb
.column(ENTRY_TYPE_INDEX)
.as_any()
.downcast_ref::<UInt8Vector>()
.with_context(|| SystemCatalogTypeMismatchSnafu {
data_type: rb.column(ENTRY_TYPE_INDEX).data_type(),
})?;
let key = rb
.column(KEY_INDEX)
.as_any()
.downcast_ref::<BinaryVector>()
.with_context(|| SystemCatalogTypeMismatchSnafu {
data_type: rb.column(KEY_INDEX).data_type(),
})?;
let value = rb
.column(VALUE_INDEX)
.as_any()
.downcast_ref::<BinaryVector>()
.with_context(|| SystemCatalogTypeMismatchSnafu {
data_type: rb.column(VALUE_INDEX).data_type(),
})?;
let mut res = Vec::with_capacity(rb.num_rows());
for ((t, k), v) in entry_type
.iter_data()
.zip(key.iter_data())
.zip(value.iter_data())
{
let entry = decode_system_catalog(t, k, v)?;
res.push(entry);
}
Ok(res)
}
/// Processes records from system catalog table and returns the max table id persisted
/// in system catalog table.
async fn handle_system_catalog_entries(&self, entries: Vec<Entry>) -> Result<TableId> {
let entries = Self::sort_entries(entries);
let mut max_table_id = 0;
for entry in entries {
match entry {
Entry::Catalog(c) => {
self.catalogs
.register_catalog_sync(c.catalog_name.clone())?;
info!("Register catalog: {}", c.catalog_name);
}
Entry::Schema(s) => {
let req = RegisterSchemaRequest {
catalog: s.catalog_name.clone(),
schema: s.schema_name.clone(),
};
let _ = self.catalogs.register_schema_sync(req)?;
info!("Registered schema: {:?}", s);
}
Entry::Table(t) => {
max_table_id = max_table_id.max(t.table_id);
if t.is_deleted {
continue;
}
self.open_and_register_table(&t).await?;
info!("Registered table: {:?}", t);
}
}
}
Ok(max_table_id)
}
/// Sort catalog entries to ensure catalog entries comes first, then schema entries,
/// and table entries is the last.
fn sort_entries(mut entries: Vec<Entry>) -> Vec<Entry> {
entries.sort();
entries
}
async fn open_and_register_table(&self, t: &TableEntry) -> Result<()> {
self.check_catalog_schema_exist(&t.catalog_name, &t.schema_name)
.await?;
let context = EngineContext {};
let open_request = OpenTableRequest {
catalog_name: t.catalog_name.clone(),
schema_name: t.schema_name.clone(),
table_name: t.table_name.clone(),
table_id: t.table_id,
region_numbers: vec![0],
};
let engine = self
.engine_manager
.engine(&t.engine)
.context(TableEngineNotFoundSnafu {
engine_name: &t.engine,
})?;
let table_ref = engine
.open_table(&context, open_request)
.await
.with_context(|_| OpenTableSnafu {
table_info: format!(
"{}.{}.{}, id: {}",
&t.catalog_name, &t.schema_name, &t.table_name, t.table_id
),
})?
.with_context(|| TableNotFoundSnafu {
table_info: format!(
"{}.{}.{}, id: {}",
&t.catalog_name, &t.schema_name, &t.table_name, t.table_id
),
})?;
let register_request = RegisterTableRequest {
catalog: t.catalog_name.clone(),
schema: t.schema_name.clone(),
table_name: t.table_name.clone(),
table_id: t.table_id,
table: table_ref,
};
let _ = self.catalogs.register_table(register_request).await?;
Ok(())
}
async fn check_state(&self) -> Result<()> {
let started = self.init_lock.lock().await;
ensure!(
*started,
IllegalManagerStateSnafu {
msg: "Catalog manager not started",
}
);
Ok(())
}
async fn check_catalog_schema_exist(
&self,
catalog_name: &str,
schema_name: &str,
) -> Result<()> {
if !self.catalogs.catalog_exist(catalog_name).await? {
return CatalogNotFoundSnafu { catalog_name }.fail()?;
}
if !self
.catalogs
.schema_exist(catalog_name, schema_name)
.await?
{
return SchemaNotFoundSnafu {
catalog: catalog_name,
schema: schema_name,
}
.fail()?;
}
Ok(())
}
}
#[async_trait::async_trait]
impl TableIdProvider for LocalCatalogManager {
async fn next_table_id(&self) -> table::Result<TableId> {
Ok(self.next_table_id.fetch_add(1, Ordering::Relaxed))
}
}
#[async_trait::async_trait]
impl CatalogManager for LocalCatalogManager {
/// Start [LocalCatalogManager] to load all information from system catalog table.
/// Make sure table engine is initialized before starting [MemoryCatalogManager].
async fn start(&self) -> Result<()> {
self.init().await
}
async fn register_table(&self, request: RegisterTableRequest) -> Result<bool> {
self.check_state().await?;
let catalog_name = request.catalog.clone();
let schema_name = request.schema.clone();
self.check_catalog_schema_exist(&catalog_name, &schema_name)
.await?;
{
let _lock = self.register_lock.lock().await;
if let Some(existing) = self
.catalogs
.table(&request.catalog, &request.schema, &request.table_name)
.await?
{
if existing.table_info().ident.table_id != request.table_id {
error!(
"Unexpected table register request: {:?}, existing: {:?}",
request,
existing.table_info()
);
return TableExistsSnafu {
table: format_full_table_name(
&catalog_name,
&schema_name,
&request.table_name,
),
}
.fail();
}
// Try to register table with same table id, just ignore.
Ok(false)
} else {
// table does not exist
let engine = request.table.table_info().meta.engine.to_string();
let table_name = request.table_name.clone();
let table_id = request.table_id;
let _ = self.catalogs.register_table(request).await?;
let _ = self
.system
.register_table(
catalog_name.clone(),
schema_name.clone(),
table_name,
table_id,
engine,
)
.await?;
increment_gauge!(
crate::metrics::METRIC_CATALOG_MANAGER_TABLE_COUNT,
1.0,
&[crate::metrics::db_label(&catalog_name, &schema_name)],
);
Ok(true)
}
}
}
async fn rename_table(&self, request: RenameTableRequest) -> Result<bool> {
self.check_state().await?;
let catalog_name = &request.catalog;
let schema_name = &request.schema;
self.check_catalog_schema_exist(catalog_name, schema_name)
.await?;
ensure!(
self.catalogs
.table(catalog_name, schema_name, &request.new_table_name)
.await?
.is_none(),
TableExistsSnafu {
table: &request.new_table_name
}
);
let _lock = self.register_lock.lock().await;
let old_table = self
.catalogs
.table(catalog_name, schema_name, &request.table_name)
.await?
.context(TableNotExistSnafu {
table: &request.table_name,
})?;
let engine = old_table.table_info().meta.engine.to_string();
// rename table in system catalog
let _ = self
.system
.register_table(
catalog_name.clone(),
schema_name.clone(),
request.new_table_name.clone(),
request.table_id,
engine,
)
.await?;
self.catalogs.rename_table(request).await
}
async fn deregister_table(&self, request: DeregisterTableRequest) -> Result<()> {
self.check_state().await?;
{
let _ = self.register_lock.lock().await;
let DeregisterTableRequest {
catalog,
schema,
table_name,
} = &request;
let table_id = self
.catalogs
.table(catalog, schema, table_name)
.await?
.with_context(|| error::TableNotExistSnafu {
table: format_full_table_name(catalog, schema, table_name),
})?
.table_info()
.ident
.table_id;
self.system.deregister_table(&request, table_id).await?;
self.catalogs.deregister_table(request).await
}
}
async fn register_schema(&self, request: RegisterSchemaRequest) -> Result<bool> {
self.check_state().await?;
let catalog_name = &request.catalog;
let schema_name = &request.schema;
if !self.catalogs.catalog_exist(catalog_name).await? {
return CatalogNotFoundSnafu { catalog_name }.fail()?;
}
{
let _lock = self.register_lock.lock().await;
ensure!(
!self
.catalogs
.schema_exist(catalog_name, schema_name)
.await?,
SchemaExistsSnafu {
schema: schema_name,
}
);
let _ = self
.system
.register_schema(request.catalog.clone(), schema_name.clone())
.await?;
self.catalogs.register_schema_sync(request)
}
}
async fn deregister_schema(&self, _request: DeregisterSchemaRequest) -> Result<bool> {
UnimplementedSnafu {
operation: "deregister schema",
}
.fail()
}
async fn register_system_table(&self, request: RegisterSystemTableRequest) -> Result<()> {
let catalog_name = request.create_table_request.catalog_name.clone();
let schema_name = request.create_table_request.schema_name.clone();
let mut sys_table_requests = self.system_table_requests.lock().await;
sys_table_requests.push(request);
increment_gauge!(
crate::metrics::METRIC_CATALOG_MANAGER_TABLE_COUNT,
1.0,
&[crate::metrics::db_label(&catalog_name, &schema_name)],
);
Ok(())
}
async fn schema_exist(&self, catalog: &str, schema: &str) -> Result<bool> {
self.catalogs.schema_exist(catalog, schema).await
}
async fn table(
&self,
catalog_name: &str,
schema_name: &str,
table_name: &str,
) -> Result<Option<TableRef>> {
self.catalogs
.table(catalog_name, schema_name, table_name)
.await
}
async fn catalog_exist(&self, catalog: &str) -> Result<bool> {
if catalog.eq_ignore_ascii_case(SYSTEM_CATALOG_NAME) {
Ok(true)
} else {
self.catalogs.catalog_exist(catalog).await
}
}
async fn table_exist(&self, catalog: &str, schema: &str, table: &str) -> Result<bool> {
self.catalogs.table_exist(catalog, schema, table).await
}
async fn catalog_names(&self) -> Result<Vec<String>> {
self.catalogs.catalog_names().await
}
async fn schema_names(&self, catalog_name: &str) -> Result<Vec<String>> {
self.catalogs.schema_names(catalog_name).await
}
async fn table_names(&self, catalog_name: &str, schema_name: &str) -> Result<Vec<String>> {
self.catalogs.table_names(catalog_name, schema_name).await
}
async fn register_catalog(self: Arc<Self>, name: String) -> Result<bool> {
self.catalogs.clone().register_catalog(name).await
}
fn as_any(&self) -> &dyn Any {
self
}
}
#[cfg(test)]
mod tests {
use std::assert_matches::assert_matches;
use mito::engine::MITO_ENGINE;
use super::*;
use crate::system::{CatalogEntry, SchemaEntry};
#[test]
fn test_sort_entry() {
let vec = vec![
Entry::Table(TableEntry {
catalog_name: "C1".to_string(),
schema_name: "S1".to_string(),
table_name: "T1".to_string(),
table_id: 1,
engine: MITO_ENGINE.to_string(),
is_deleted: false,
}),
Entry::Catalog(CatalogEntry {
catalog_name: "C2".to_string(),
}),
Entry::Schema(SchemaEntry {
catalog_name: "C1".to_string(),
schema_name: "S1".to_string(),
}),
Entry::Schema(SchemaEntry {
catalog_name: "C2".to_string(),
schema_name: "S2".to_string(),
}),
Entry::Catalog(CatalogEntry {
catalog_name: "".to_string(),
}),
Entry::Table(TableEntry {
catalog_name: "C1".to_string(),
schema_name: "S1".to_string(),
table_name: "T2".to_string(),
table_id: 2,
engine: MITO_ENGINE.to_string(),
is_deleted: false,
}),
];
let res = LocalCatalogManager::sort_entries(vec);
assert_matches!(res[0], Entry::Catalog(..));
assert_matches!(res[1], Entry::Catalog(..));
assert_matches!(res[2], Entry::Schema(..));
assert_matches!(res[3], Entry::Schema(..));
assert_matches!(res[4], Entry::Table(..));
assert_matches!(res[5], Entry::Table(..));
}
}

View File

@@ -15,29 +15,129 @@
use std::any::Any;
use std::collections::hash_map::Entry;
use std::collections::HashMap;
use std::sync::atomic::{AtomicU32, Ordering};
use std::sync::{Arc, RwLock, Weak};
use common_catalog::consts::{DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, INFORMATION_SCHEMA_NAME};
use common_catalog::consts::{
DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, INFORMATION_SCHEMA_NAME, MIN_USER_TABLE_ID,
};
use metrics::{decrement_gauge, increment_gauge};
use snafu::OptionExt;
use table::metadata::TableId;
use table::table::TableIdProvider;
use table::TableRef;
use crate::error::{CatalogNotFoundSnafu, Result, SchemaNotFoundSnafu, TableExistsSnafu};
use crate::error::{
CatalogNotFoundSnafu, Result, SchemaNotFoundSnafu, TableExistsSnafu, TableNotFoundSnafu,
};
use crate::information_schema::InformationSchemaProvider;
use crate::{CatalogManager, DeregisterTableRequest, RegisterSchemaRequest, RegisterTableRequest};
use crate::{
CatalogManager, DeregisterSchemaRequest, DeregisterTableRequest, RegisterSchemaRequest,
RegisterSystemTableRequest, RegisterTableRequest, RenameTableRequest,
};
type SchemaEntries = HashMap<String, HashMap<String, TableRef>>;
/// Simple in-memory list of catalogs
#[derive(Clone)]
pub struct MemoryCatalogManager {
/// Collection of catalogs containing schemas and ultimately Tables
catalogs: Arc<RwLock<HashMap<String, SchemaEntries>>>,
pub catalogs: RwLock<HashMap<String, SchemaEntries>>,
pub table_id: AtomicU32,
}
#[async_trait::async_trait]
impl TableIdProvider for MemoryCatalogManager {
async fn next_table_id(&self) -> table::error::Result<TableId> {
Ok(self.table_id.fetch_add(1, Ordering::Relaxed))
}
}
#[async_trait::async_trait]
impl CatalogManager for MemoryCatalogManager {
async fn schema_exists(&self, catalog: &str, schema: &str) -> Result<bool> {
async fn start(&self) -> Result<()> {
self.table_id.store(MIN_USER_TABLE_ID, Ordering::Relaxed);
Ok(())
}
async fn register_table(&self, request: RegisterTableRequest) -> Result<bool> {
self.register_table_sync(request)
}
async fn rename_table(&self, request: RenameTableRequest) -> Result<bool> {
let mut catalogs = self.catalogs.write().unwrap();
let schema = catalogs
.get_mut(&request.catalog)
.with_context(|| CatalogNotFoundSnafu {
catalog_name: &request.catalog,
})?
.get_mut(&request.schema)
.with_context(|| SchemaNotFoundSnafu {
catalog: &request.catalog,
schema: &request.schema,
})?;
// check old and new table names
if !schema.contains_key(&request.table_name) {
return TableNotFoundSnafu {
table_info: request.table_name.to_string(),
}
.fail()?;
}
if schema.contains_key(&request.new_table_name) {
return TableExistsSnafu {
table: &request.new_table_name,
}
.fail();
}
let table = schema.remove(&request.table_name).unwrap();
let _ = schema.insert(request.new_table_name, table);
Ok(true)
}
async fn deregister_table(&self, request: DeregisterTableRequest) -> Result<()> {
self.deregister_table_sync(request)
}
async fn register_schema(&self, request: RegisterSchemaRequest) -> Result<bool> {
self.register_schema_sync(request)
}
async fn deregister_schema(&self, request: DeregisterSchemaRequest) -> Result<bool> {
let mut catalogs = self.catalogs.write().unwrap();
let schemas = catalogs
.get_mut(&request.catalog)
.with_context(|| CatalogNotFoundSnafu {
catalog_name: &request.catalog,
})?;
let table_count = schemas
.remove(&request.schema)
.with_context(|| SchemaNotFoundSnafu {
catalog: &request.catalog,
schema: &request.schema,
})?
.len();
decrement_gauge!(
crate::metrics::METRIC_CATALOG_MANAGER_TABLE_COUNT,
table_count as f64,
&[crate::metrics::db_label(&request.catalog, &request.schema)],
);
decrement_gauge!(
crate::metrics::METRIC_CATALOG_MANAGER_SCHEMA_COUNT,
1.0,
&[crate::metrics::db_label(&request.catalog, &request.schema)],
);
Ok(true)
}
async fn register_system_table(&self, _request: RegisterSystemTableRequest) -> Result<()> {
// TODO(ruihang): support register system table request
Ok(())
}
async fn schema_exist(&self, catalog: &str, schema: &str) -> Result<bool> {
self.schema_exist_sync(catalog, schema)
}
@@ -59,11 +159,11 @@ impl CatalogManager for MemoryCatalogManager {
Ok(result)
}
async fn catalog_exists(&self, catalog: &str) -> Result<bool> {
async fn catalog_exist(&self, catalog: &str) -> Result<bool> {
self.catalog_exist_sync(catalog)
}
async fn table_exists(&self, catalog: &str, schema: &str, table: &str) -> Result<bool> {
async fn table_exist(&self, catalog: &str, schema: &str, table: &str) -> Result<bool> {
let catalogs = self.catalogs.read().unwrap();
Ok(catalogs
.get(catalog)
@@ -108,27 +208,28 @@ impl CatalogManager for MemoryCatalogManager {
.collect())
}
async fn register_catalog(self: Arc<Self>, name: String) -> Result<bool> {
self.register_catalog_sync(name)
}
fn as_any(&self) -> &dyn Any {
self
}
}
impl MemoryCatalogManager {
pub fn new() -> Arc<Self> {
Arc::new(Self {
catalogs: Default::default(),
})
}
/// Creates a manager with some default setups
/// (e.g. default catalog/schema and information schema)
pub fn with_default_setup() -> Arc<Self> {
let manager = Arc::new(Self {
table_id: AtomicU32::new(MIN_USER_TABLE_ID),
catalogs: Default::default(),
});
// Safety: default catalog/schema is registered in order so no CatalogNotFound error will occur
manager.register_catalog_sync(DEFAULT_CATALOG_NAME).unwrap();
manager
.register_catalog_sync(DEFAULT_CATALOG_NAME.to_string())
.unwrap();
manager
.register_schema_sync(RegisterSchemaRequest {
catalog: DEFAULT_CATALOG_NAME.to_string(),
@@ -156,15 +257,12 @@ impl MemoryCatalogManager {
}
/// Registers a catalog if it does not exist and returns false if the schema exists.
pub fn register_catalog_sync(&self, name: &str) -> Result<bool> {
let name = name.to_string();
pub fn register_catalog_sync(self: &Arc<Self>, name: String) -> Result<bool> {
let mut catalogs = self.catalogs.write().unwrap();
match catalogs.entry(name.clone()) {
Entry::Vacant(e) => {
let arc_self = Arc::new(self.clone());
let catalog = arc_self.create_catalog_entry(name);
let catalog = self.create_catalog_entry(name);
e.insert(catalog);
increment_gauge!(crate::metrics::METRIC_CATALOG_MANAGER_CATALOG_COUNT, 1.0);
Ok(true)
@@ -263,7 +361,7 @@ impl MemoryCatalogManager {
let schema = &table.table_info().schema_name;
if !manager.catalog_exist_sync(catalog).unwrap() {
manager.register_catalog_sync(catalog).unwrap();
manager.register_catalog_sync(catalog.to_string()).unwrap();
}
if !manager.schema_exist_sync(catalog, schema).unwrap() {
@@ -295,6 +393,8 @@ pub fn new_memory_catalog_manager() -> Result<Arc<MemoryCatalogManager>> {
#[cfg(test)]
mod tests {
use common_catalog::consts::*;
use common_error::ext::ErrorExt;
use common_error::status_code::StatusCode;
use table::table::numbers::{NumbersTable, NUMBERS_TABLE_NAME};
use super::*;
@@ -311,7 +411,7 @@ mod tests {
table: NumbersTable::table(NUMBERS_TABLE_ID),
};
catalog_list.register_table_sync(register_request).unwrap();
let _ = catalog_list.register_table(register_request).await.unwrap();
let table = catalog_list
.table(
DEFAULT_CATALOG_NAME,
@@ -328,11 +428,130 @@ mod tests {
.is_none());
}
#[tokio::test]
async fn test_mem_manager_rename_table() {
let catalog = MemoryCatalogManager::with_default_setup();
let table_name = "test_table";
assert!(!catalog
.table_exist(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, table_name)
.await
.unwrap());
// register test table
let table_id = 2333;
let register_request = RegisterTableRequest {
catalog: DEFAULT_CATALOG_NAME.to_string(),
schema: DEFAULT_SCHEMA_NAME.to_string(),
table_name: table_name.to_string(),
table_id,
table: NumbersTable::table(table_id),
};
assert!(catalog.register_table(register_request).await.unwrap());
assert!(catalog
.table_exist(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, table_name)
.await
.unwrap());
// rename test table
let new_table_name = "test_table_renamed";
let rename_request = RenameTableRequest {
catalog: DEFAULT_CATALOG_NAME.to_string(),
schema: DEFAULT_SCHEMA_NAME.to_string(),
table_name: table_name.to_string(),
new_table_name: new_table_name.to_string(),
table_id,
};
let _ = catalog.rename_table(rename_request).await.unwrap();
// test old table name not exist
assert!(!catalog
.table_exist(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, table_name)
.await
.unwrap());
// test new table name exists
assert!(catalog
.table_exist(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, new_table_name)
.await
.unwrap());
let registered_table = catalog
.table(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, new_table_name)
.await
.unwrap()
.unwrap();
assert_eq!(registered_table.table_info().ident.table_id, table_id);
let dup_register_request = RegisterTableRequest {
catalog: DEFAULT_CATALOG_NAME.to_string(),
schema: DEFAULT_SCHEMA_NAME.to_string(),
table_name: new_table_name.to_string(),
table_id: table_id + 1,
table: NumbersTable::table(table_id + 1),
};
let result = catalog.register_table(dup_register_request).await;
let err = result.err().unwrap();
assert_eq!(StatusCode::TableAlreadyExists, err.status_code());
}
#[tokio::test]
async fn test_catalog_rename_table() {
let catalog = MemoryCatalogManager::with_default_setup();
let table_name = "num";
let table_id = 2333;
let table = NumbersTable::table(table_id);
// register table
let register_table_req = RegisterTableRequest {
catalog: DEFAULT_CATALOG_NAME.to_string(),
schema: DEFAULT_SCHEMA_NAME.to_string(),
table_name: table_name.to_string(),
table_id,
table,
};
assert!(catalog.register_table(register_table_req).await.unwrap());
assert!(catalog
.table(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, table_name)
.await
.unwrap()
.is_some());
// rename table
let new_table_name = "numbers_new";
let rename_table_req = RenameTableRequest {
catalog: DEFAULT_CATALOG_NAME.to_string(),
schema: DEFAULT_SCHEMA_NAME.to_string(),
table_name: table_name.to_string(),
new_table_name: new_table_name.to_string(),
table_id,
};
assert!(catalog.rename_table(rename_table_req).await.unwrap());
assert!(catalog
.table(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, table_name)
.await
.unwrap()
.is_none());
assert!(catalog
.table(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, new_table_name)
.await
.unwrap()
.is_some());
let registered_table = catalog
.table(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, new_table_name)
.await
.unwrap()
.unwrap();
assert_eq!(registered_table.table_info().ident.table_id, table_id);
}
#[test]
pub fn test_register_catalog_sync() {
let list = MemoryCatalogManager::with_default_setup();
assert!(list.register_catalog_sync("test_catalog").unwrap());
assert!(!list.register_catalog_sync("test_catalog").unwrap());
assert!(list
.register_catalog_sync("test_catalog".to_string())
.unwrap());
assert!(!list
.register_catalog_sync("test_catalog".to_string())
.unwrap());
}
#[tokio::test]
@@ -347,7 +566,7 @@ mod tests {
table_id: 2333,
table: NumbersTable::table(2333),
};
catalog.register_table_sync(register_table_req).unwrap();
let _ = catalog.register_table(register_table_req).await.unwrap();
assert!(catalog
.table(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, table_name)
.await
@@ -359,11 +578,53 @@ mod tests {
schema: DEFAULT_SCHEMA_NAME.to_string(),
table_name: table_name.to_string(),
};
catalog.deregister_table_sync(deregister_table_req).unwrap();
catalog
.deregister_table(deregister_table_req)
.await
.unwrap();
assert!(catalog
.table(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, table_name)
.await
.unwrap()
.is_none());
}
#[tokio::test]
async fn test_catalog_deregister_schema() {
let catalog = MemoryCatalogManager::with_default_setup();
// Registers a catalog, a schema, and a table.
let catalog_name = "foo_catalog".to_string();
let schema_name = "foo_schema".to_string();
let table_name = "foo_table".to_string();
let schema = RegisterSchemaRequest {
catalog: catalog_name.clone(),
schema: schema_name.clone(),
};
let table = RegisterTableRequest {
catalog: catalog_name.clone(),
schema: schema_name.clone(),
table_name,
table_id: 0,
table: NumbersTable::table(0),
};
catalog
.clone()
.register_catalog(catalog_name.clone())
.await
.unwrap();
catalog.register_schema(schema).await.unwrap();
catalog.register_table(table).await.unwrap();
let request = DeregisterSchemaRequest {
catalog: catalog_name.clone(),
schema: schema_name.clone(),
};
assert!(catalog.deregister_schema(request).await.unwrap());
assert!(!catalog
.schema_exist(&catalog_name, &schema_name)
.await
.unwrap());
}
}

View File

@@ -12,11 +12,21 @@
// See the License for the specific language governing permissions and
// limitations under the License.
use std::sync::Arc;
pub use client::{CachedMetaKvBackend, MetaKvBackend};
pub use manager::RemoteCatalogManager;
mod client;
mod manager;
#[cfg(feature = "testing")]
pub mod mock;
pub use manager::KvBackendCatalogManager;
pub mod region_alive_keeper;
#[async_trait::async_trait]
pub trait KvCacheInvalidator: Send + Sync {
async fn invalidate_key(&self, key: &[u8]);
}
pub type KvCacheInvalidatorRef = Arc<dyn KvCacheInvalidator>;

View File

@@ -18,9 +18,8 @@ use std::sync::Arc;
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::{CacheNotGetSnafu, Error, MetaSrvSnafu, Result};
use common_meta::kv_backend::{KvBackend, KvBackendRef, TxnService};
use common_meta::rpc::store::{
BatchDeleteRequest, BatchDeleteResponse, BatchGetRequest, BatchGetResponse, BatchPutRequest,
@@ -29,11 +28,12 @@ use common_meta::rpc::store::{
RangeRequest, RangeResponse,
};
use common_meta::rpc::KeyValue;
use common_telemetry::{debug, timer};
use common_telemetry::timer;
use meta_client::client::MetaClient;
use moka::future::{Cache, CacheBuilder};
use snafu::{OptionExt, ResultExt};
use super::KvCacheInvalidator;
use crate::metrics::{METRIC_CATALOG_KV_GET, METRIC_CATALOG_KV_REMOTE_GET};
const CACHE_MAX_CAPACITY: u64 = 10000;
@@ -197,8 +197,7 @@ impl KvBackend for CachedMetaKvBackend {
#[async_trait::async_trait]
impl KvCacheInvalidator for CachedMetaKvBackend {
async fn invalidate_key(&self, key: &[u8]) {
self.cache.invalidate(key).await;
debug!("invalidated cache key: {}", String::from_utf8_lossy(key));
self.cache.invalidate(key).await
}
}
@@ -252,7 +251,7 @@ impl KvBackend for MetaKvBackend {
.range(req)
.await
.map_err(BoxedError::new)
.context(ExternalSnafu)
.context(MetaSrvSnafu)
}
async fn get(&self, key: &[u8]) -> Result<Option<KeyValue>> {
@@ -261,7 +260,7 @@ impl KvBackend for MetaKvBackend {
.range(RangeRequest::new().with_key(key))
.await
.map_err(BoxedError::new)
.context(ExternalSnafu)?;
.context(MetaSrvSnafu)?;
Ok(response.take_kvs().get_mut(0).map(|kv| KeyValue {
key: kv.take_key(),
value: kv.take_value(),
@@ -273,7 +272,7 @@ impl KvBackend for MetaKvBackend {
.batch_put(req)
.await
.map_err(BoxedError::new)
.context(ExternalSnafu)
.context(MetaSrvSnafu)
}
async fn put(&self, req: PutRequest) -> Result<PutResponse> {
@@ -281,7 +280,7 @@ impl KvBackend for MetaKvBackend {
.put(req)
.await
.map_err(BoxedError::new)
.context(ExternalSnafu)
.context(MetaSrvSnafu)
}
async fn delete_range(&self, req: DeleteRangeRequest) -> Result<DeleteRangeResponse> {
@@ -289,7 +288,7 @@ impl KvBackend for MetaKvBackend {
.delete_range(req)
.await
.map_err(BoxedError::new)
.context(ExternalSnafu)
.context(MetaSrvSnafu)
}
async fn batch_delete(&self, req: BatchDeleteRequest) -> Result<BatchDeleteResponse> {
@@ -297,7 +296,7 @@ impl KvBackend for MetaKvBackend {
.batch_delete(req)
.await
.map_err(BoxedError::new)
.context(ExternalSnafu)
.context(MetaSrvSnafu)
}
async fn batch_get(&self, req: BatchGetRequest) -> Result<BatchGetResponse> {
@@ -305,7 +304,7 @@ impl KvBackend for MetaKvBackend {
.batch_get(req)
.await
.map_err(BoxedError::new)
.context(ExternalSnafu)
.context(MetaSrvSnafu)
}
async fn compare_and_put(
@@ -316,7 +315,7 @@ impl KvBackend for MetaKvBackend {
.compare_and_put(request)
.await
.map_err(BoxedError::new)
.context(ExternalSnafu)
.context(MetaSrvSnafu)
}
async fn move_value(&self, req: MoveValueRequest) -> Result<MoveValueResponse> {
@@ -324,7 +323,7 @@ impl KvBackend for MetaKvBackend {
.move_value(req)
.await
.map_err(BoxedError::new)
.context(ExternalSnafu)
.context(MetaSrvSnafu)
}
fn as_any(&self) -> &dyn Any {

View File

@@ -0,0 +1,436 @@
// 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::sync::Arc;
use async_trait::async_trait;
use common_catalog::consts::MITO_ENGINE;
use common_meta::ident::TableIdent;
use common_meta::key::catalog_name::CatalogNameKey;
use common_meta::key::datanode_table::DatanodeTableValue;
use common_meta::key::schema_name::SchemaNameKey;
use common_meta::key::TableMetadataManagerRef;
use common_telemetry::{error, info, warn};
use futures_util::TryStreamExt;
use metrics::increment_gauge;
use snafu::{ensure, OptionExt, ResultExt};
use table::engine::manager::TableEngineManagerRef;
use table::engine::EngineContext;
use table::requests::OpenTableRequest;
use table::TableRef;
use tokio::sync::Mutex;
use crate::error::{
OpenTableSnafu, ParallelOpenTableSnafu, Result, TableEngineNotFoundSnafu, TableExistsSnafu,
TableMetadataManagerSnafu, TableNotFoundSnafu, UnimplementedSnafu,
};
use crate::local::MemoryCatalogManager;
use crate::remote::region_alive_keeper::RegionAliveKeepers;
use crate::{
handle_system_table_request, CatalogManager, DeregisterSchemaRequest, DeregisterTableRequest,
RegisterSchemaRequest, RegisterSystemTableRequest, RegisterTableRequest, RenameTableRequest,
};
/// Catalog manager based on metasrv.
pub struct RemoteCatalogManager {
node_id: u64,
engine_manager: TableEngineManagerRef,
system_table_requests: Mutex<Vec<RegisterSystemTableRequest>>,
region_alive_keepers: Arc<RegionAliveKeepers>,
memory_catalog_manager: Arc<MemoryCatalogManager>,
table_metadata_manager: TableMetadataManagerRef,
}
impl RemoteCatalogManager {
pub fn new(
engine_manager: TableEngineManagerRef,
node_id: u64,
region_alive_keepers: Arc<RegionAliveKeepers>,
table_metadata_manager: TableMetadataManagerRef,
) -> Self {
Self {
engine_manager,
node_id,
system_table_requests: Default::default(),
region_alive_keepers,
memory_catalog_manager: MemoryCatalogManager::with_default_setup(),
table_metadata_manager,
}
}
async fn initiate_catalogs(&self) -> Result<()> {
let tables = self
.table_metadata_manager
.datanode_table_manager()
.tables(self.node_id)
.try_collect::<Vec<_>>()
.await
.context(TableMetadataManagerSnafu)?;
let joins = tables
.into_iter()
.map(|datanode_table_value| {
let engine_manager = self.engine_manager.clone();
let memory_catalog_manager = self.memory_catalog_manager.clone();
let table_metadata_manager = self.table_metadata_manager.clone();
let region_alive_keepers = self.region_alive_keepers.clone();
common_runtime::spawn_bg(async move {
let table_id = datanode_table_value.table_id;
if let Err(e) = open_and_register_table(
engine_manager,
datanode_table_value,
memory_catalog_manager,
table_metadata_manager,
region_alive_keepers,
)
.await
{
// Note that we don't return error here if table opened failed. This is because
// we don't want those broken tables to impede the startup of Datanode.
// However, this could be changed in the future.
error!(e; "Failed to open or register table, id = {table_id}")
}
})
})
.collect::<Vec<_>>();
let _ = futures::future::try_join_all(joins)
.await
.context(ParallelOpenTableSnafu)?;
Ok(())
}
}
async fn open_and_register_table(
engine_manager: TableEngineManagerRef,
datanode_table_value: DatanodeTableValue,
memory_catalog_manager: Arc<MemoryCatalogManager>,
table_metadata_manager: TableMetadataManagerRef,
region_alive_keepers: Arc<RegionAliveKeepers>,
) -> Result<()> {
let context = EngineContext {};
let table_id = datanode_table_value.table_id;
let region_numbers = datanode_table_value.regions;
let table_info_value = table_metadata_manager
.table_info_manager()
.get(table_id)
.await
.context(TableMetadataManagerSnafu)?
.context(TableNotFoundSnafu {
table_info: format!("table id: {table_id}"),
})?;
let table_info = &table_info_value.table_info;
let catalog_name = table_info.catalog_name.clone();
let schema_name = table_info.schema_name.clone();
let table_name = table_info.name.clone();
let request = OpenTableRequest {
catalog_name: catalog_name.clone(),
schema_name: schema_name.clone(),
table_name: table_name.clone(),
table_id,
region_numbers: region_numbers.clone(),
};
let engine =
engine_manager
.engine(&table_info.meta.engine)
.context(TableEngineNotFoundSnafu {
engine_name: &table_info.meta.engine,
})?;
let table_ident = TableIdent {
catalog: catalog_name,
schema: schema_name,
table: table_name,
table_id,
engine: table_info.meta.engine.clone(),
};
let table = engine
.open_table(&context, request)
.await
.with_context(|_| OpenTableSnafu {
table_info: table_ident.to_string(),
})?
.with_context(|| TableNotFoundSnafu {
table_info: table_ident.to_string(),
})?;
info!("Successfully opened table, {table_ident}");
if !memory_catalog_manager
.catalog_exist(&table_ident.catalog)
.await?
{
memory_catalog_manager.register_catalog_sync(table_ident.catalog.clone())?;
}
if !memory_catalog_manager
.schema_exist(&table_ident.catalog, &table_ident.schema)
.await?
{
memory_catalog_manager.register_schema_sync(RegisterSchemaRequest {
catalog: table_ident.catalog.clone(),
schema: table_ident.schema.clone(),
})?;
}
let request = RegisterTableRequest {
catalog: table_ident.catalog.clone(),
schema: table_ident.schema.clone(),
table_name: table_ident.table.clone(),
table_id,
table,
};
let registered =
register_table(&memory_catalog_manager, &region_alive_keepers, request).await?;
ensure!(
registered,
TableExistsSnafu {
table: table_ident.to_string(),
}
);
info!("Successfully registered table, {table_ident}");
Ok(())
}
async fn register_table(
memory_catalog_manager: &Arc<MemoryCatalogManager>,
region_alive_keepers: &Arc<RegionAliveKeepers>,
request: RegisterTableRequest,
) -> Result<bool> {
let table = request.table.clone();
let registered = memory_catalog_manager.register_table_sync(request)?;
if registered {
let table_info = table.table_info();
let table_ident = TableIdent {
catalog: table_info.catalog_name.clone(),
schema: table_info.schema_name.clone(),
table: table_info.name.clone(),
table_id: table_info.table_id(),
engine: table_info.meta.engine.clone(),
};
region_alive_keepers
.register_table(table_ident, table, memory_catalog_manager.clone())
.await?;
}
Ok(registered)
}
#[async_trait]
impl CatalogManager for RemoteCatalogManager {
async fn start(&self) -> Result<()> {
self.initiate_catalogs().await?;
let mut system_table_requests = self.system_table_requests.lock().await;
let engine = self
.engine_manager
.engine(MITO_ENGINE)
.context(TableEngineNotFoundSnafu {
engine_name: MITO_ENGINE,
})?;
handle_system_table_request(self, engine, &mut system_table_requests).await?;
info!("All system table opened");
Ok(())
}
async fn register_table(&self, request: RegisterTableRequest) -> Result<bool> {
register_table(
&self.memory_catalog_manager,
&self.region_alive_keepers,
request,
)
.await
}
async fn deregister_table(&self, request: DeregisterTableRequest) -> Result<()> {
let Some(table) = self
.memory_catalog_manager
.table(&request.catalog, &request.schema, &request.table_name)
.await?
else {
return Ok(());
};
let table_info = table.table_info();
let table_ident = TableIdent {
catalog: request.catalog.clone(),
schema: request.schema.clone(),
table: request.table_name.clone(),
table_id: table_info.ident.table_id,
engine: table_info.meta.engine.clone(),
};
if let Some(keeper) = self
.region_alive_keepers
.deregister_table(&table_ident)
.await
{
warn!(
"Table {} is deregistered from region alive keepers",
keeper.table_ident(),
);
}
self.memory_catalog_manager.deregister_table(request).await
}
async fn register_schema(&self, request: RegisterSchemaRequest) -> Result<bool> {
self.memory_catalog_manager.register_schema_sync(request)
}
async fn deregister_schema(&self, _request: DeregisterSchemaRequest) -> Result<bool> {
UnimplementedSnafu {
operation: "deregister schema",
}
.fail()
}
async fn rename_table(&self, request: RenameTableRequest) -> Result<bool> {
self.memory_catalog_manager.rename_table(request).await
}
async fn register_system_table(&self, request: RegisterSystemTableRequest) -> Result<()> {
let catalog_name = request.create_table_request.catalog_name.clone();
let schema_name = request.create_table_request.schema_name.clone();
let mut requests = self.system_table_requests.lock().await;
requests.push(request);
increment_gauge!(
crate::metrics::METRIC_CATALOG_MANAGER_TABLE_COUNT,
1.0,
&[crate::metrics::db_label(&catalog_name, &schema_name)],
);
Ok(())
}
async fn schema_exist(&self, catalog: &str, schema: &str) -> Result<bool> {
if !self.catalog_exist(catalog).await? {
return Ok(false);
}
if self
.memory_catalog_manager
.schema_exist(catalog, schema)
.await?
{
return Ok(true);
}
let remote_schema_exists = self
.table_metadata_manager
.schema_manager()
.exist(SchemaNameKey::new(catalog, schema))
.await
.context(TableMetadataManagerSnafu)?;
// Create schema locally if remote schema exists. Since local schema is managed by memory
// catalog manager, creating a local schema is relatively cheap (just a HashMap).
// Besides, if this method ("schema_exist) is called, it's very likely that someone wants to
// create a table in this schema. We should create the schema now.
if remote_schema_exists
&& self
.memory_catalog_manager
.register_schema(RegisterSchemaRequest {
catalog: catalog.to_string(),
schema: schema.to_string(),
})
.await?
{
info!("register schema '{catalog}/{schema}' on demand");
}
Ok(remote_schema_exists)
}
async fn table(
&self,
catalog_name: &str,
schema_name: &str,
table_name: &str,
) -> Result<Option<TableRef>> {
self.memory_catalog_manager
.table(catalog_name, schema_name, table_name)
.await
}
async fn catalog_exist(&self, catalog: &str) -> Result<bool> {
if self.memory_catalog_manager.catalog_exist(catalog).await? {
return Ok(true);
}
let key = CatalogNameKey::new(catalog);
let remote_catalog_exists = self
.table_metadata_manager
.catalog_manager()
.exist(key)
.await
.context(TableMetadataManagerSnafu)?;
// Create catalog locally if remote catalog exists. Since local catalog is managed by memory
// catalog manager, creating a local catalog is relatively cheap (just a HashMap).
// Besides, if this method ("catalog_exist) is called, it's very likely that someone wants to
// create a table in this catalog. We should create the catalog now.
if remote_catalog_exists
&& self
.memory_catalog_manager
.clone()
.register_catalog(catalog.to_string())
.await?
{
info!("register catalog '{catalog}' on demand");
}
Ok(remote_catalog_exists)
}
async fn table_exist(&self, catalog: &str, schema: &str, table: &str) -> Result<bool> {
if !self.catalog_exist(catalog).await? {
return Ok(false);
}
if !self.schema_exist(catalog, schema).await? {
return Ok(false);
}
self.memory_catalog_manager
.table_exist(catalog, schema, table)
.await
}
async fn catalog_names(&self) -> Result<Vec<String>> {
self.memory_catalog_manager.catalog_names().await
}
async fn schema_names(&self, catalog_name: &str) -> Result<Vec<String>> {
self.memory_catalog_manager.schema_names(catalog_name).await
}
async fn table_names(&self, catalog_name: &str, schema_name: &str) -> Result<Vec<String>> {
self.memory_catalog_manager
.table_names(catalog_name, schema_name)
.await
}
async fn register_catalog(self: Arc<Self>, name: String) -> Result<bool> {
self.memory_catalog_manager.register_catalog_sync(name)
}
fn as_any(&self) -> &dyn Any {
self
}
}

View File

@@ -55,14 +55,14 @@ impl TableEngine for MockTableEngine {
let data = vec![Arc::new(StringVector::from(vec!["a", "b", "c"])) as _];
let record_batch = RecordBatch::new(schema, data).unwrap();
let table = MemTable::new_with_catalog(
let table: TableRef = Arc::new(MemTable::new_with_catalog(
&request.table_name,
record_batch,
table_id,
request.catalog_name,
request.schema_name,
vec![0],
);
)) as Arc<_>;
let mut tables = self.tables.write().unwrap();
let _ = tables.insert(table_id, table.clone() as TableRef);

View File

@@ -0,0 +1,867 @@
// Copyright 2023 Greptime Team
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
use std::collections::HashMap;
use std::future::Future;
use std::sync::atomic::{AtomicBool, Ordering};
use std::sync::Arc;
use async_trait::async_trait;
use common_meta::error::InvalidProtoMsgSnafu;
use common_meta::heartbeat::handler::{
HandleControl, HeartbeatResponseHandler, HeartbeatResponseHandlerContext,
};
use common_meta::ident::TableIdent;
use common_meta::RegionIdent;
use common_telemetry::{debug, error, info, warn};
use snafu::{OptionExt, ResultExt};
use store_api::storage::RegionNumber;
use table::engine::manager::TableEngineManagerRef;
use table::engine::{CloseTableResult, EngineContext, TableEngineRef};
use table::metadata::TableId;
use table::requests::CloseTableRequest;
use table::TableRef;
use tokio::sync::{mpsc, oneshot, Mutex};
use tokio::task::JoinHandle;
use tokio::time::{Duration, Instant};
use crate::error::{Result, TableEngineNotFoundSnafu};
use crate::local::MemoryCatalogManager;
use crate::DeregisterTableRequest;
/// [RegionAliveKeepers] manages all [RegionAliveKeeper] in a scope of tables.
pub struct RegionAliveKeepers {
table_engine_manager: TableEngineManagerRef,
keepers: Arc<Mutex<HashMap<TableId, Arc<RegionAliveKeeper>>>>,
heartbeat_interval_millis: u64,
started: AtomicBool,
/// The epoch when [RegionAliveKeepers] is created. It's used to get a monotonically non-decreasing
/// elapsed time when submitting heartbeats to Metasrv (because [Instant] is monotonically
/// non-decreasing). The heartbeat request will carry the duration since this epoch, and the
/// duration acts like an "invariant point" for region's keep alive lease.
epoch: Instant,
}
impl RegionAliveKeepers {
pub fn new(
table_engine_manager: TableEngineManagerRef,
heartbeat_interval_millis: u64,
) -> Self {
Self {
table_engine_manager,
keepers: Arc::new(Mutex::new(HashMap::new())),
heartbeat_interval_millis,
started: AtomicBool::new(false),
epoch: Instant::now(),
}
}
pub async fn find_keeper(&self, table_id: TableId) -> Option<Arc<RegionAliveKeeper>> {
self.keepers.lock().await.get(&table_id).cloned()
}
pub async fn register_table(
&self,
table_ident: TableIdent,
table: TableRef,
catalog_manager: Arc<MemoryCatalogManager>,
) -> Result<()> {
let table_id = table_ident.table_id;
let keeper = self.find_keeper(table_id).await;
if keeper.is_some() {
return Ok(());
}
let table_engine = self
.table_engine_manager
.engine(&table_ident.engine)
.context(TableEngineNotFoundSnafu {
engine_name: &table_ident.engine,
})?;
let keeper = Arc::new(RegionAliveKeeper::new(
table_engine,
catalog_manager,
table_ident.clone(),
self.heartbeat_interval_millis,
));
for r in table.table_info().meta.region_numbers.iter() {
keeper.register_region(*r).await;
}
let mut keepers = self.keepers.lock().await;
let _ = keepers.insert(table_id, keeper.clone());
if self.started.load(Ordering::Relaxed) {
keeper.start().await;
info!("RegionAliveKeeper for table {table_ident} is started!");
} else {
info!("RegionAliveKeeper for table {table_ident} is registered but not started yet!");
}
Ok(())
}
pub async fn deregister_table(
&self,
table_ident: &TableIdent,
) -> Option<Arc<RegionAliveKeeper>> {
let table_id = table_ident.table_id;
self.keepers.lock().await.remove(&table_id).map(|x| {
info!("Deregister RegionAliveKeeper for table {table_ident}");
x
})
}
pub async fn register_region(&self, region_ident: &RegionIdent) {
let table_id = region_ident.table_ident.table_id;
let Some(keeper) = self.find_keeper(table_id).await else {
// Alive keeper could be affected by lagging msg, just warn and ignore.
warn!("Alive keeper for region {region_ident} is not found!");
return;
};
keeper.register_region(region_ident.region_number).await
}
pub async fn deregister_region(&self, region_ident: &RegionIdent) {
let table_id = region_ident.table_ident.table_id;
let Some(keeper) = self.find_keeper(table_id).await else {
// Alive keeper could be affected by lagging msg, just warn and ignore.
warn!("Alive keeper for region {region_ident} is not found!");
return;
};
let _ = keeper.deregister_region(region_ident.region_number).await;
}
pub async fn start(&self) {
let keepers = self.keepers.lock().await;
for keeper in keepers.values() {
keeper.start().await;
}
self.started.store(true, Ordering::Relaxed);
info!(
"RegionAliveKeepers for tables {:?} are started!",
keepers.keys().map(|x| x.to_string()).collect::<Vec<_>>(),
);
}
pub fn epoch(&self) -> Instant {
self.epoch
}
}
#[async_trait]
impl HeartbeatResponseHandler for RegionAliveKeepers {
fn is_acceptable(&self, ctx: &HeartbeatResponseHandlerContext) -> bool {
!ctx.response.region_leases.is_empty()
}
async fn handle(
&self,
ctx: &mut HeartbeatResponseHandlerContext,
) -> common_meta::error::Result<HandleControl> {
let leases = ctx.response.region_leases.drain(..).collect::<Vec<_>>();
for lease in leases {
let table_ident: TableIdent = match lease
.table_ident
.context(InvalidProtoMsgSnafu {
err_msg: "'table_ident' is missing in RegionLease",
})
.and_then(|x| x.try_into())
{
Ok(x) => x,
Err(e) => {
error!(e; "");
continue;
}
};
let table_id = table_ident.table_id;
let Some(keeper) = self.keepers.lock().await.get(&table_id).cloned() else {
// Alive keeper could be affected by lagging msg, just warn and ignore.
warn!("Alive keeper for table {table_ident} is not found!");
continue;
};
let start_instant = self.epoch + Duration::from_millis(lease.duration_since_epoch);
let deadline = start_instant + Duration::from_secs(lease.lease_seconds);
keeper.keep_lived(lease.regions, deadline).await;
}
Ok(HandleControl::Continue)
}
}
/// [RegionAliveKeeper] starts a countdown for each region in a table. When deadline is reached,
/// the region will be closed.
/// The deadline is controlled by Metasrv. It works like "lease" for regions: a Datanode submits its
/// opened regions to Metasrv, in heartbeats. If Metasrv decides some region could be resided in this
/// Datanode, it will "extend" the region's "lease", with a deadline for [RegionAliveKeeper] to
/// countdown.
pub struct RegionAliveKeeper {
catalog_manager: Arc<MemoryCatalogManager>,
table_engine: TableEngineRef,
table_ident: TableIdent,
countdown_task_handles: Arc<Mutex<HashMap<RegionNumber, Arc<CountdownTaskHandle>>>>,
heartbeat_interval_millis: u64,
started: AtomicBool,
}
impl RegionAliveKeeper {
fn new(
table_engine: TableEngineRef,
catalog_manager: Arc<MemoryCatalogManager>,
table_ident: TableIdent,
heartbeat_interval_millis: u64,
) -> Self {
Self {
catalog_manager,
table_engine,
table_ident,
countdown_task_handles: Arc::new(Mutex::new(HashMap::new())),
heartbeat_interval_millis,
started: AtomicBool::new(false),
}
}
async fn find_handle(&self, region: &RegionNumber) -> Option<Arc<CountdownTaskHandle>> {
self.countdown_task_handles
.lock()
.await
.get(region)
.cloned()
}
async fn register_region(&self, region: RegionNumber) {
if self.find_handle(&region).await.is_some() {
return;
}
let countdown_task_handles = Arc::downgrade(&self.countdown_task_handles);
let on_task_finished = async move {
if let Some(x) = countdown_task_handles.upgrade() {
let _ = x.lock().await.remove(&region);
} // Else the countdown task handles map could be dropped because the keeper is dropped.
};
let catalog_manager = self.catalog_manager.clone();
let ident = self.table_ident.clone();
let handle = Arc::new(CountdownTaskHandle::new(
self.table_engine.clone(),
self.table_ident.clone(),
region,
move |result: Option<CloseTableResult>| {
if matches!(result, Some(CloseTableResult::Released(_))) {
let result = catalog_manager.deregister_table_sync(DeregisterTableRequest {
catalog: ident.catalog.to_string(),
schema: ident.schema.to_string(),
table_name: ident.table.to_string(),
});
info!(
"Deregister table: {} after countdown task finished, result: {result:?}",
ident.table_id
);
} else {
debug!("Countdown task returns: {result:?}");
}
on_task_finished
},
));
let mut handles = self.countdown_task_handles.lock().await;
let _ = handles.insert(region, handle.clone());
if self.started.load(Ordering::Relaxed) {
handle.start(self.heartbeat_interval_millis).await;
info!(
"Region alive countdown for region {region} in table {} is started!",
self.table_ident
);
} else {
info!(
"Region alive countdown for region {region} in table {} is registered but not started yet!",
self.table_ident
);
}
}
async fn deregister_region(&self, region: RegionNumber) -> Option<Arc<CountdownTaskHandle>> {
self.countdown_task_handles
.lock()
.await
.remove(&region)
.map(|x| {
info!(
"Deregister alive countdown for region {region} in table {}",
self.table_ident
);
x
})
}
async fn start(&self) {
let handles = self.countdown_task_handles.lock().await;
for handle in handles.values() {
handle.start(self.heartbeat_interval_millis).await;
}
self.started.store(true, Ordering::Relaxed);
info!(
"Region alive countdowns for regions {:?} in table {} are started!",
handles.keys().copied().collect::<Vec<_>>(),
self.table_ident
);
}
async fn keep_lived(&self, designated_regions: Vec<RegionNumber>, deadline: Instant) {
for region in designated_regions {
if let Some(handle) = self.find_handle(&region).await {
handle.reset_deadline(deadline).await;
}
// Else the region alive keeper might be triggered by lagging messages, we can safely ignore it.
}
}
pub async fn deadline(&self, region: RegionNumber) -> Option<Instant> {
let mut deadline = None;
if let Some(handle) = self.find_handle(&region).await {
let (s, r) = oneshot::channel();
if handle.tx.send(CountdownCommand::Deadline(s)).await.is_ok() {
deadline = r.await.ok()
}
}
deadline
}
pub fn table_ident(&self) -> &TableIdent {
&self.table_ident
}
}
#[derive(Debug)]
enum CountdownCommand {
Start(u64),
Reset(Instant),
Deadline(oneshot::Sender<Instant>),
}
struct CountdownTaskHandle {
tx: mpsc::Sender<CountdownCommand>,
handler: JoinHandle<()>,
table_ident: TableIdent,
region: RegionNumber,
}
impl CountdownTaskHandle {
/// Creates a new [CountdownTaskHandle] and starts the countdown task.
/// # Params
/// - `on_task_finished`: a callback to be invoked when the task is finished. Note that it will not
/// be invoked if the task is cancelled (by dropping the handle). This is because we want something
/// meaningful to be done when the task is finished, e.g. deregister the handle from the map.
/// While dropping the handle does not necessarily mean the task is finished.
fn new<Fut>(
table_engine: TableEngineRef,
table_ident: TableIdent,
region: RegionNumber,
on_task_finished: impl FnOnce(Option<CloseTableResult>) -> Fut + Send + 'static,
) -> Self
where
Fut: Future<Output = ()> + Send,
{
let (tx, rx) = mpsc::channel(1024);
let mut countdown_task = CountdownTask {
table_engine,
table_ident: table_ident.clone(),
region,
rx,
};
let handler = common_runtime::spawn_bg(async move {
let result = countdown_task.run().await;
on_task_finished(result).await;
});
Self {
tx,
handler,
table_ident,
region,
}
}
async fn start(&self, heartbeat_interval_millis: u64) {
if let Err(e) = self
.tx
.send(CountdownCommand::Start(heartbeat_interval_millis))
.await
{
warn!(
"Failed to start region alive keeper countdown: {e}. \
Maybe the task is stopped due to region been closed."
);
}
}
async fn reset_deadline(&self, deadline: Instant) {
if let Err(e) = self.tx.send(CountdownCommand::Reset(deadline)).await {
warn!(
"Failed to reset region alive keeper deadline: {e}. \
Maybe the task is stopped due to region been closed."
);
}
}
}
impl Drop for CountdownTaskHandle {
fn drop(&mut self) {
debug!(
"Aborting region alive countdown task for region {} in table {}",
self.region, self.table_ident,
);
self.handler.abort();
}
}
struct CountdownTask {
table_engine: TableEngineRef,
table_ident: TableIdent,
region: RegionNumber,
rx: mpsc::Receiver<CountdownCommand>,
}
impl CountdownTask {
// returns true if
async fn run(&mut self) -> Option<CloseTableResult> {
// 30 years. See `Instant::far_future`.
let far_future = Instant::now() + Duration::from_secs(86400 * 365 * 30);
// Make sure the alive countdown is not gonna happen before heartbeat task is started (the
// "start countdown" command will be sent from heartbeat task).
let countdown = tokio::time::sleep_until(far_future);
tokio::pin!(countdown);
let region = &self.region;
let table_ident = &self.table_ident;
loop {
tokio::select! {
command = self.rx.recv() => {
match command {
Some(CountdownCommand::Start(heartbeat_interval_millis)) => {
// Set first deadline in 4 heartbeats (roughly after 20 seconds from now if heartbeat
// interval is set to default 5 seconds), to make Datanode and Metasrv more tolerable to
// network or other jitters during startup.
let first_deadline = Instant::now() + Duration::from_millis(heartbeat_interval_millis) * 4;
countdown.set(tokio::time::sleep_until(first_deadline));
},
Some(CountdownCommand::Reset(deadline)) => {
if countdown.deadline() < deadline {
debug!(
"Reset deadline of region {region} of table {table_ident} to approximately {} seconds later",
(deadline - Instant::now()).as_secs_f32(),
);
countdown.set(tokio::time::sleep_until(deadline));
}
// Else the countdown could be either:
// - not started yet;
// - during startup protection;
// - received a lagging heartbeat message.
// All can be safely ignored.
},
None => {
info!(
"The handle of countdown task for region {region} of table {table_ident} \
is dropped, RegionAliveKeeper out."
);
break;
},
Some(CountdownCommand::Deadline(tx)) => {
let _ = tx.send(countdown.deadline());
}
}
}
() = &mut countdown => {
let result = self.close_region().await;
warn!(
"Region {region} of table {table_ident} is closed, result: {result:?}. \
RegionAliveKeeper out.",
);
return Some(result);
}
}
}
None
}
async fn close_region(&self) -> CloseTableResult {
let ctx = EngineContext::default();
let region = self.region;
let table_ident = &self.table_ident;
loop {
let request = CloseTableRequest {
catalog_name: table_ident.catalog.clone(),
schema_name: table_ident.schema.clone(),
table_name: table_ident.table.clone(),
table_id: table_ident.table_id,
region_numbers: vec![region],
flush: true,
};
match self.table_engine.close_table(&ctx, request).await {
Ok(result) => return result,
// If region is failed to close, immediately retry. Maybe we should panic instead?
Err(e) => error!(e;
"Failed to close region {region} of table {table_ident}. \
For the integrity of data, retry closing and retry without wait.",
),
}
}
}
}
#[cfg(test)]
mod test {
use std::sync::atomic::{AtomicBool, Ordering};
use std::sync::Arc;
use api::v1::meta::{HeartbeatResponse, RegionLease};
use common_meta::heartbeat::mailbox::HeartbeatMailbox;
use datatypes::schema::RawSchema;
use table::engine::manager::MemoryTableEngineManager;
use table::engine::TableEngine;
use table::requests::{CreateTableRequest, TableOptions};
use table::test_util::EmptyTable;
use super::*;
use crate::remote::mock::MockTableEngine;
async fn prepare_keepers() -> (TableIdent, RegionAliveKeepers) {
let table_engine = Arc::new(MockTableEngine::default());
let table_engine_manager = Arc::new(MemoryTableEngineManager::new(table_engine));
let keepers = RegionAliveKeepers::new(table_engine_manager, 5000);
let catalog = "my_catalog";
let schema = "my_schema";
let table = "my_table";
let table_ident = TableIdent {
catalog: catalog.to_string(),
schema: schema.to_string(),
table: table.to_string(),
table_id: 1,
engine: "MockTableEngine".to_string(),
};
let table = Arc::new(EmptyTable::new(CreateTableRequest {
id: 1,
catalog_name: catalog.to_string(),
schema_name: schema.to_string(),
table_name: table.to_string(),
desc: None,
schema: RawSchema {
column_schemas: vec![],
timestamp_index: None,
version: 0,
},
region_numbers: vec![1, 2, 3],
primary_key_indices: vec![],
create_if_not_exists: false,
table_options: TableOptions::default(),
engine: "MockTableEngine".to_string(),
}));
let catalog_manager = MemoryCatalogManager::new_with_table(table.clone());
keepers
.register_table(table_ident.clone(), table, catalog_manager)
.await
.unwrap();
assert!(keepers
.keepers
.lock()
.await
.contains_key(&table_ident.table_id));
(table_ident, keepers)
}
#[tokio::test(flavor = "multi_thread")]
async fn test_handle_heartbeat_response() {
let (table_ident, keepers) = prepare_keepers().await;
keepers.start().await;
let startup_protection_until = Instant::now() + Duration::from_secs(21);
let duration_since_epoch = (Instant::now() - keepers.epoch).as_millis() as _;
let lease_seconds = 100;
let response = HeartbeatResponse {
region_leases: vec![RegionLease {
table_ident: Some(table_ident.clone().into()),
regions: vec![1, 3], // Not extending region 2's lease time.
duration_since_epoch,
lease_seconds,
}],
..Default::default()
};
let keep_alive_until = keepers.epoch
+ Duration::from_millis(duration_since_epoch)
+ Duration::from_secs(lease_seconds);
let (tx, _) = mpsc::channel(8);
let mailbox = Arc::new(HeartbeatMailbox::new(tx));
let mut ctx = HeartbeatResponseHandlerContext::new(mailbox, response);
assert!(keepers.handle(&mut ctx).await.unwrap() == HandleControl::Continue);
// sleep to wait for background task spawned in `handle`
tokio::time::sleep(Duration::from_secs(1)).await;
async fn test(
keeper: &Arc<RegionAliveKeeper>,
region_number: RegionNumber,
startup_protection_until: Instant,
keep_alive_until: Instant,
is_kept_live: bool,
) {
let deadline = keeper.deadline(region_number).await.unwrap();
if is_kept_live {
assert!(deadline > startup_protection_until && deadline == keep_alive_until);
} else {
assert!(deadline <= startup_protection_until);
}
}
let keeper = &keepers
.keepers
.lock()
.await
.get(&table_ident.table_id)
.cloned()
.unwrap();
// Test region 1 and 3 is kept lived. Their deadlines are updated to desired instant.
test(keeper, 1, startup_protection_until, keep_alive_until, true).await;
test(keeper, 3, startup_protection_until, keep_alive_until, true).await;
// Test region 2 is not kept lived. It's deadline is not updated: still during startup protection period.
test(keeper, 2, startup_protection_until, keep_alive_until, false).await;
}
#[tokio::test(flavor = "multi_thread")]
async fn test_region_alive_keepers() {
let (table_ident, keepers) = prepare_keepers().await;
keepers
.register_region(&RegionIdent {
cluster_id: 1,
datanode_id: 1,
table_ident: table_ident.clone(),
region_number: 4,
})
.await;
keepers.start().await;
for keeper in keepers.keepers.lock().await.values() {
let regions = {
let handles = keeper.countdown_task_handles.lock().await;
handles.keys().copied().collect::<Vec<_>>()
};
for region in regions {
// assert countdown tasks are started
let deadline = keeper.deadline(region).await.unwrap();
assert!(deadline <= Instant::now() + Duration::from_secs(20));
}
}
keepers
.deregister_region(&RegionIdent {
cluster_id: 1,
datanode_id: 1,
table_ident: table_ident.clone(),
region_number: 1,
})
.await;
let mut regions = keepers
.find_keeper(table_ident.table_id)
.await
.unwrap()
.countdown_task_handles
.lock()
.await
.keys()
.copied()
.collect::<Vec<_>>();
regions.sort();
assert_eq!(regions, vec![2, 3, 4]);
let keeper = keepers.deregister_table(&table_ident).await.unwrap();
assert!(Arc::try_unwrap(keeper).is_ok(), "keeper is not dropped");
assert!(keepers.keepers.lock().await.is_empty());
}
#[tokio::test(flavor = "multi_thread")]
async fn test_region_alive_keeper() {
let table_engine = Arc::new(MockTableEngine::default());
let table_ident = TableIdent {
catalog: "my_catalog".to_string(),
schema: "my_schema".to_string(),
table: "my_table".to_string(),
table_id: 1024,
engine: "mito".to_string(),
};
let catalog_manager = MemoryCatalogManager::with_default_setup();
let keeper = RegionAliveKeeper::new(table_engine, catalog_manager, table_ident, 1000);
let region = 1;
assert!(keeper.find_handle(&region).await.is_none());
keeper.register_region(region).await;
let _ = keeper.find_handle(&region).await.unwrap();
let ten_seconds_later = || Instant::now() + Duration::from_secs(10);
keeper.keep_lived(vec![1, 2, 3], ten_seconds_later()).await;
assert!(keeper.find_handle(&2).await.is_none());
assert!(keeper.find_handle(&3).await.is_none());
let far_future = Instant::now() + Duration::from_secs(86400 * 365 * 29);
// assert if keeper is not started, keep_lived is of no use
assert!(keeper.deadline(region).await.unwrap() > far_future);
keeper.start().await;
keeper.keep_lived(vec![1, 2, 3], ten_seconds_later()).await;
// assert keep_lived works if keeper is started
assert!(keeper.deadline(region).await.unwrap() <= ten_seconds_later());
let handle = keeper.deregister_region(region).await.unwrap();
assert!(Arc::try_unwrap(handle).is_ok(), "handle is not dropped");
assert!(keeper.find_handle(&region).await.is_none());
}
#[tokio::test(flavor = "multi_thread")]
async fn test_countdown_task_handle() {
let table_engine = Arc::new(MockTableEngine::default());
let table_ident = TableIdent {
catalog: "my_catalog".to_string(),
schema: "my_schema".to_string(),
table: "my_table".to_string(),
table_id: 1024,
engine: "mito".to_string(),
};
let finished = Arc::new(AtomicBool::new(false));
let finished_clone = finished.clone();
let handle = CountdownTaskHandle::new(
table_engine.clone(),
table_ident.clone(),
1,
|_| async move { finished_clone.store(true, Ordering::Relaxed) },
);
let tx = handle.tx.clone();
// assert countdown task is running
tx.send(CountdownCommand::Start(5000)).await.unwrap();
assert!(!finished.load(Ordering::Relaxed));
drop(handle);
tokio::time::sleep(Duration::from_secs(1)).await;
// assert countdown task is stopped
assert!(tx
.try_send(CountdownCommand::Reset(
Instant::now() + Duration::from_secs(10)
))
.is_err());
// assert `on_task_finished` is not called (because the task is aborted by the handle's drop)
assert!(!finished.load(Ordering::Relaxed));
let finished = Arc::new(AtomicBool::new(false));
let finished_clone = finished.clone();
let handle = CountdownTaskHandle::new(table_engine, table_ident, 1, |_| async move {
finished_clone.store(true, Ordering::Relaxed)
});
handle.tx.send(CountdownCommand::Start(100)).await.unwrap();
tokio::time::sleep(Duration::from_secs(1)).await;
// assert `on_task_finished` is called when task is finished normally
assert!(finished.load(Ordering::Relaxed));
}
#[tokio::test(flavor = "multi_thread")]
async fn test_countdown_task_run() {
let ctx = &EngineContext::default();
let catalog = "my_catalog";
let schema = "my_schema";
let table = "my_table";
let table_id = 1;
let request = CreateTableRequest {
id: table_id,
catalog_name: catalog.to_string(),
schema_name: schema.to_string(),
table_name: table.to_string(),
desc: None,
schema: RawSchema {
column_schemas: vec![],
timestamp_index: None,
version: 0,
},
region_numbers: vec![],
primary_key_indices: vec![],
create_if_not_exists: false,
table_options: TableOptions::default(),
engine: "mito".to_string(),
};
let table_engine = Arc::new(MockTableEngine::default());
let _ = table_engine.create_table(ctx, request).await.unwrap();
let table_ident = TableIdent {
catalog: catalog.to_string(),
schema: schema.to_string(),
table: table.to_string(),
table_id,
engine: "mito".to_string(),
};
let (tx, rx) = mpsc::channel(10);
let mut task = CountdownTask {
table_engine: table_engine.clone(),
table_ident,
region: 1,
rx,
};
let _handle = common_runtime::spawn_bg(async move {
task.run().await;
});
async fn deadline(tx: &mpsc::Sender<CountdownCommand>) -> Instant {
let (s, r) = oneshot::channel();
tx.send(CountdownCommand::Deadline(s)).await.unwrap();
r.await.unwrap()
}
// if countdown task is not started, its deadline is set to far future
assert!(deadline(&tx).await > Instant::now() + Duration::from_secs(86400 * 365 * 29));
// start countdown in 250ms * 4 = 1s
tx.send(CountdownCommand::Start(250)).await.unwrap();
// assert deadline is correctly set
assert!(deadline(&tx).await <= Instant::now() + Duration::from_secs(1));
// reset countdown in 1.5s
tx.send(CountdownCommand::Reset(
Instant::now() + Duration::from_millis(1500),
))
.await
.unwrap();
// assert the table is closed after deadline is reached
assert!(table_engine.table_exists(ctx, table_id));
// spare 500ms for the task to close the table
tokio::time::sleep(Duration::from_millis(2000)).await;
assert!(!table_engine.table_exists(ctx, table_id));
}
}

634
src/catalog/src/system.rs Normal file
View File

@@ -0,0 +1,634 @@
// Copyright 2023 Greptime Team
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
use std::collections::HashMap;
use std::sync::Arc;
use common_catalog::consts::{
DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, INFORMATION_SCHEMA_NAME, MITO_ENGINE,
SYSTEM_CATALOG_NAME, SYSTEM_CATALOG_TABLE_ID, SYSTEM_CATALOG_TABLE_NAME,
};
use common_recordbatch::SendableRecordBatchStream;
use common_telemetry::{debug, warn};
use common_time::util;
use datatypes::prelude::{ConcreteDataType, ScalarVector, VectorRef};
use datatypes::schema::{ColumnSchema, RawSchema};
use datatypes::vectors::{BinaryVector, TimestampMillisecondVector, UInt8Vector};
use serde::{Deserialize, Serialize};
use snafu::{ensure, OptionExt, ResultExt};
use store_api::storage::ScanRequest;
use table::engine::{EngineContext, TableEngineRef};
use table::metadata::TableId;
use table::requests::{CreateTableRequest, InsertRequest, OpenTableRequest, TableOptions};
use table::TableRef;
use crate::error::{
self, CreateSystemCatalogSnafu, DeregisterTableSnafu, EmptyValueSnafu, Error,
InsertCatalogRecordSnafu, InvalidEntryTypeSnafu, InvalidKeySnafu, OpenSystemCatalogSnafu,
Result, ValueDeserializeSnafu,
};
use crate::DeregisterTableRequest;
pub const ENTRY_TYPE_INDEX: usize = 0;
pub const KEY_INDEX: usize = 1;
pub const VALUE_INDEX: usize = 3;
pub struct SystemCatalogTable(TableRef);
impl SystemCatalogTable {
pub async fn new(engine: TableEngineRef) -> Result<Self> {
let request = OpenTableRequest {
catalog_name: SYSTEM_CATALOG_NAME.to_string(),
schema_name: INFORMATION_SCHEMA_NAME.to_string(),
table_name: SYSTEM_CATALOG_TABLE_NAME.to_string(),
table_id: SYSTEM_CATALOG_TABLE_ID,
region_numbers: vec![0],
};
let schema = build_system_catalog_schema();
let ctx = EngineContext::default();
if let Some(table) = engine
.open_table(&ctx, request)
.await
.context(OpenSystemCatalogSnafu)?
{
Ok(Self(table))
} else {
// system catalog table is not yet created, try to create
let request = CreateTableRequest {
id: SYSTEM_CATALOG_TABLE_ID,
catalog_name: SYSTEM_CATALOG_NAME.to_string(),
schema_name: INFORMATION_SCHEMA_NAME.to_string(),
table_name: SYSTEM_CATALOG_TABLE_NAME.to_string(),
desc: Some("System catalog table".to_string()),
schema,
region_numbers: vec![0],
primary_key_indices: vec![ENTRY_TYPE_INDEX, KEY_INDEX],
create_if_not_exists: true,
table_options: TableOptions::default(),
engine: engine.name().to_string(),
};
let table = engine
.create_table(&ctx, request)
.await
.context(CreateSystemCatalogSnafu)?;
Ok(Self(table))
}
}
pub async fn register_table(
&self,
catalog: String,
schema: String,
table_name: String,
table_id: TableId,
engine: String,
) -> Result<usize> {
let insert_request =
build_table_insert_request(catalog, schema, table_name, table_id, engine);
self.0
.insert(insert_request)
.await
.context(InsertCatalogRecordSnafu)
}
pub(crate) async fn deregister_table(
&self,
request: &DeregisterTableRequest,
table_id: TableId,
) -> Result<()> {
let deletion_request = build_table_deletion_request(request, table_id);
self.0
.insert(deletion_request)
.await
.map(|x| {
if x != 1 {
let table = common_catalog::format_full_table_name(
&request.catalog,
&request.schema,
&request.table_name
);
warn!("Failed to delete table record from information_schema, unexpected returned result: {x}, table: {table}");
}
})
.with_context(|_| DeregisterTableSnafu {
request: request.clone(),
})
}
pub async fn register_schema(&self, catalog: String, schema: String) -> Result<usize> {
let insert_request = build_schema_insert_request(catalog, schema);
self.0
.insert(insert_request)
.await
.context(InsertCatalogRecordSnafu)
}
/// Create a stream of all entries inside system catalog table
pub async fn records(&self) -> Result<SendableRecordBatchStream> {
let full_projection = None;
let scan_req = ScanRequest {
sequence: None,
projection: full_projection,
filters: vec![],
output_ordering: None,
limit: None,
};
let stream = self
.0
.scan_to_stream(scan_req)
.await
.context(error::SystemCatalogTableScanSnafu)?;
Ok(stream)
}
pub fn as_table_ref(&self) -> TableRef {
self.0.clone()
}
}
/// Build system catalog table schema.
/// A system catalog table consists of 6 columns, namely
/// - entry_type: type of entry in current row, can be any variant of [EntryType].
/// - key: a binary encoded key of entry, differs according to different entry type.
/// - timestamp: currently not used.
/// - value: JSON-encoded value of entry's metadata.
/// - gmt_created: create time of this metadata.
/// - gmt_modified: last updated time of this metadata.
fn build_system_catalog_schema() -> RawSchema {
let cols = vec![
ColumnSchema::new(
"entry_type".to_string(),
ConcreteDataType::uint8_datatype(),
false,
),
ColumnSchema::new(
"key".to_string(),
ConcreteDataType::binary_datatype(),
false,
),
ColumnSchema::new(
"timestamp".to_string(),
ConcreteDataType::timestamp_millisecond_datatype(),
false,
)
.with_time_index(true),
ColumnSchema::new(
"value".to_string(),
ConcreteDataType::binary_datatype(),
false,
),
ColumnSchema::new(
"gmt_created".to_string(),
ConcreteDataType::timestamp_millisecond_datatype(),
false,
),
ColumnSchema::new(
"gmt_modified".to_string(),
ConcreteDataType::timestamp_millisecond_datatype(),
false,
),
];
RawSchema::new(cols)
}
/// Formats key string for table entry in system catalog
#[inline]
pub fn format_table_entry_key(catalog: &str, schema: &str, table_id: TableId) -> String {
format!("{catalog}.{schema}.{table_id}")
}
pub fn build_table_insert_request(
catalog: String,
schema: String,
table_name: String,
table_id: TableId,
engine: String,
) -> InsertRequest {
let entry_key = format_table_entry_key(&catalog, &schema, table_id);
build_insert_request(
EntryType::Table,
entry_key.as_bytes(),
serde_json::to_string(&TableEntryValue {
table_name,
engine,
is_deleted: false,
})
.unwrap()
.as_bytes(),
)
}
pub(crate) fn build_table_deletion_request(
request: &DeregisterTableRequest,
table_id: TableId,
) -> InsertRequest {
let entry_key = format_table_entry_key(&request.catalog, &request.schema, table_id);
build_insert_request(
EntryType::Table,
entry_key.as_bytes(),
serde_json::to_string(&TableEntryValue {
table_name: "".to_string(),
engine: "".to_string(),
is_deleted: true,
})
.unwrap()
.as_bytes(),
)
}
fn build_primary_key_columns(entry_type: EntryType, key: &[u8]) -> HashMap<String, VectorRef> {
HashMap::from([
(
"entry_type".to_string(),
Arc::new(UInt8Vector::from_slice([entry_type as u8])) as VectorRef,
),
(
"key".to_string(),
Arc::new(BinaryVector::from_slice(&[key])) as VectorRef,
),
(
"timestamp".to_string(),
// Timestamp in key part is intentionally left to 0
Arc::new(TimestampMillisecondVector::from_slice([0])) as VectorRef,
),
])
}
pub fn build_schema_insert_request(catalog_name: String, schema_name: String) -> InsertRequest {
let full_schema_name = format!("{catalog_name}.{schema_name}");
build_insert_request(
EntryType::Schema,
full_schema_name.as_bytes(),
serde_json::to_string(&SchemaEntryValue {})
.unwrap()
.as_bytes(),
)
}
pub fn build_insert_request(entry_type: EntryType, key: &[u8], value: &[u8]) -> InsertRequest {
let primary_key_columns = build_primary_key_columns(entry_type, key);
let mut columns_values = HashMap::with_capacity(6);
columns_values.extend(primary_key_columns);
let _ = columns_values.insert(
"value".to_string(),
Arc::new(BinaryVector::from_slice(&[value])) as _,
);
let now = util::current_time_millis();
let _ = columns_values.insert(
"gmt_created".to_string(),
Arc::new(TimestampMillisecondVector::from_slice([now])) as _,
);
let _ = columns_values.insert(
"gmt_modified".to_string(),
Arc::new(TimestampMillisecondVector::from_slice([now])) as _,
);
InsertRequest {
catalog_name: DEFAULT_CATALOG_NAME.to_string(),
schema_name: DEFAULT_SCHEMA_NAME.to_string(),
table_name: SYSTEM_CATALOG_TABLE_NAME.to_string(),
columns_values,
region_number: 0, // system catalog table has only one region
}
}
pub fn decode_system_catalog(
entry_type: Option<u8>,
key: Option<&[u8]>,
value: Option<&[u8]>,
) -> Result<Entry> {
debug!(
"Decode system catalog entry: {:?}, {:?}, {:?}",
entry_type, key, value
);
let entry_type = entry_type.context(InvalidKeySnafu { key: None })?;
let key = String::from_utf8_lossy(key.context(InvalidKeySnafu { key: None })?);
match EntryType::try_from(entry_type)? {
EntryType::Catalog => {
// As for catalog entry, the key is a string with format: `<catalog_name>`
// and the value is current not used.
let catalog_name = key.to_string();
Ok(Entry::Catalog(CatalogEntry { catalog_name }))
}
EntryType::Schema => {
// As for schema entry, the key is a string with format: `<catalog_name>.<schema_name>`
// and the value is current not used.
let schema_parts = key.split('.').collect::<Vec<_>>();
ensure!(
schema_parts.len() == 2,
InvalidKeySnafu {
key: Some(key.to_string())
}
);
Ok(Entry::Schema(SchemaEntry {
catalog_name: schema_parts[0].to_string(),
schema_name: schema_parts[1].to_string(),
}))
}
EntryType::Table => {
// As for table entry, the key is a string with format: `<catalog_name>.<schema_name>.<table_id>`
// and the value is a JSON string with format: `{"table_name": <table_name>}`
let table_parts = key.split('.').collect::<Vec<_>>();
ensure!(
table_parts.len() >= 3,
InvalidKeySnafu {
key: Some(key.to_string())
}
);
let value = value.context(EmptyValueSnafu)?;
debug!("Table meta value: {}", String::from_utf8_lossy(value));
let table_meta: TableEntryValue =
serde_json::from_slice(value).context(ValueDeserializeSnafu)?;
let table_id = table_parts[2].parse::<TableId>().unwrap();
Ok(Entry::Table(TableEntry {
catalog_name: table_parts[0].to_string(),
schema_name: table_parts[1].to_string(),
table_name: table_meta.table_name,
table_id,
engine: table_meta.engine,
is_deleted: table_meta.is_deleted,
}))
}
}
}
#[derive(Debug, Copy, Clone, PartialEq, Eq)]
pub enum EntryType {
Catalog = 1,
Schema = 2,
Table = 3,
}
impl TryFrom<u8> for EntryType {
type Error = Error;
fn try_from(value: u8) -> std::result::Result<Self, Self::Error> {
match value {
b if b == Self::Catalog as u8 => Ok(Self::Catalog),
b if b == Self::Schema as u8 => Ok(Self::Schema),
b if b == Self::Table as u8 => Ok(Self::Table),
b => InvalidEntryTypeSnafu {
entry_type: Some(b),
}
.fail(),
}
}
}
#[derive(Debug, PartialEq, Eq, Ord, PartialOrd)]
pub enum Entry {
Catalog(CatalogEntry),
Schema(SchemaEntry),
Table(TableEntry),
}
#[derive(Debug, PartialEq, Eq, Ord, PartialOrd)]
pub struct CatalogEntry {
pub catalog_name: String,
}
#[derive(Debug, PartialEq, Eq, Ord, PartialOrd)]
pub struct SchemaEntry {
pub catalog_name: String,
pub schema_name: String,
}
#[derive(Debug, Serialize, Deserialize, PartialEq, Eq)]
pub struct SchemaEntryValue;
#[derive(Debug, PartialEq, Eq, Ord, PartialOrd)]
pub struct TableEntry {
pub catalog_name: String,
pub schema_name: String,
pub table_name: String,
pub table_id: TableId,
pub engine: String,
pub is_deleted: bool,
}
#[derive(Debug, Serialize, Deserialize, PartialEq, Eq)]
pub struct TableEntryValue {
pub table_name: String,
#[serde(default = "mito_engine")]
pub engine: String,
#[serde(default = "not_deleted")]
pub is_deleted: bool,
}
fn mito_engine() -> String {
MITO_ENGINE.to_string()
}
fn not_deleted() -> bool {
false
}
#[cfg(test)]
mod tests {
use common_recordbatch::RecordBatches;
use common_test_util::temp_dir::{create_temp_dir, TempDir};
use datatypes::value::Value;
use log_store::NoopLogStore;
use mito::config::EngineConfig;
use mito::engine::{MitoEngine, MITO_ENGINE};
use object_store::ObjectStore;
use storage::compaction::noop::NoopCompactionScheduler;
use storage::config::EngineConfig as StorageEngineConfig;
use storage::EngineImpl;
use table::metadata::TableType;
use table::metadata::TableType::Base;
use super::*;
#[test]
pub fn test_decode_catalog_entry() {
let entry = decode_system_catalog(
Some(EntryType::Catalog as u8),
Some("some_catalog".as_bytes()),
None,
)
.unwrap();
if let Entry::Catalog(e) = entry {
assert_eq!("some_catalog", e.catalog_name);
} else {
panic!("Unexpected type: {entry:?}");
}
}
#[test]
pub fn test_decode_schema_entry() {
let entry = decode_system_catalog(
Some(EntryType::Schema as u8),
Some("some_catalog.some_schema".as_bytes()),
None,
)
.unwrap();
if let Entry::Schema(e) = entry {
assert_eq!("some_catalog", e.catalog_name);
assert_eq!("some_schema", e.schema_name);
} else {
panic!("Unexpected type: {entry:?}");
}
}
#[test]
pub fn test_decode_table() {
let entry = decode_system_catalog(
Some(EntryType::Table as u8),
Some("some_catalog.some_schema.42".as_bytes()),
Some("{\"table_name\":\"some_table\"}".as_bytes()),
)
.unwrap();
if let Entry::Table(e) = entry {
assert_eq!("some_catalog", e.catalog_name);
assert_eq!("some_schema", e.schema_name);
assert_eq!("some_table", e.table_name);
assert_eq!(42, e.table_id);
} else {
panic!("Unexpected type: {entry:?}");
}
}
#[test]
pub fn test_decode_mismatch() {
assert!(decode_system_catalog(
Some(EntryType::Table as u8),
Some("some_catalog.some_schema.42".as_bytes()),
None,
)
.is_err());
}
#[test]
pub fn test_entry_type() {
assert_eq!(EntryType::Catalog, EntryType::try_from(1).unwrap());
assert_eq!(EntryType::Schema, EntryType::try_from(2).unwrap());
assert_eq!(EntryType::Table, EntryType::try_from(3).unwrap());
assert!(EntryType::try_from(4).is_err());
}
pub async fn prepare_table_engine() -> (TempDir, TableEngineRef) {
let dir = create_temp_dir("system-table-test");
let store_dir = dir.path().to_string_lossy();
let mut builder = object_store::services::Fs::default();
let _ = builder.root(&store_dir);
let object_store = ObjectStore::new(builder).unwrap().finish();
let noop_compaction_scheduler = Arc::new(NoopCompactionScheduler::default());
let table_engine = Arc::new(MitoEngine::new(
EngineConfig::default(),
EngineImpl::new(
StorageEngineConfig::default(),
Arc::new(NoopLogStore),
object_store.clone(),
noop_compaction_scheduler,
)
.unwrap(),
object_store,
));
(dir, table_engine)
}
#[tokio::test]
async fn test_system_table_type() {
let (_dir, table_engine) = prepare_table_engine().await;
let system_table = SystemCatalogTable::new(table_engine).await.unwrap();
assert_eq!(Base, system_table.as_table_ref().table_type());
}
#[tokio::test]
async fn test_system_table_info() {
let (_dir, table_engine) = prepare_table_engine().await;
let system_table = SystemCatalogTable::new(table_engine).await.unwrap();
let info = system_table.as_table_ref().table_info();
assert_eq!(TableType::Base, info.table_type);
assert_eq!(SYSTEM_CATALOG_TABLE_NAME, info.name);
assert_eq!(SYSTEM_CATALOG_TABLE_ID, info.ident.table_id);
assert_eq!(SYSTEM_CATALOG_NAME, info.catalog_name);
assert_eq!(INFORMATION_SCHEMA_NAME, info.schema_name);
}
#[tokio::test]
async fn test_system_catalog_table_records() {
let (_, table_engine) = prepare_table_engine().await;
let catalog_table = SystemCatalogTable::new(table_engine).await.unwrap();
let result = catalog_table
.register_table(
DEFAULT_CATALOG_NAME.to_string(),
DEFAULT_SCHEMA_NAME.to_string(),
"my_table".to_string(),
1,
MITO_ENGINE.to_string(),
)
.await
.unwrap();
assert_eq!(result, 1);
let records = catalog_table.records().await.unwrap();
let mut batches = RecordBatches::try_collect(records).await.unwrap().take();
assert_eq!(batches.len(), 1);
let batch = batches.remove(0);
assert_eq!(batch.num_rows(), 1);
let row = batch.rows().next().unwrap();
let Value::UInt8(entry_type) = row[0] else {
unreachable!()
};
let Value::Binary(key) = row[1].clone() else {
unreachable!()
};
let Value::Binary(value) = row[3].clone() else {
unreachable!()
};
let entry = decode_system_catalog(Some(entry_type), Some(&*key), Some(&*value)).unwrap();
let expected = Entry::Table(TableEntry {
catalog_name: DEFAULT_CATALOG_NAME.to_string(),
schema_name: DEFAULT_SCHEMA_NAME.to_string(),
table_name: "my_table".to_string(),
table_id: 1,
engine: MITO_ENGINE.to_string(),
is_deleted: false,
});
assert_eq!(entry, expected);
catalog_table
.deregister_table(
&DeregisterTableRequest {
catalog: DEFAULT_CATALOG_NAME.to_string(),
schema: DEFAULT_SCHEMA_NAME.to_string(),
table_name: "my_table".to_string(),
},
1,
)
.await
.unwrap();
let records = catalog_table.records().await.unwrap();
let batches = RecordBatches::try_collect(records).await.unwrap().take();
assert_eq!(batches.len(), 1);
}
}

View File

@@ -123,7 +123,7 @@ mod tests {
use session::context::QueryContext;
use super::*;
use crate::memory::MemoryCatalogManager;
use crate::local::MemoryCatalogManager;
#[test]
fn test_validate_table_ref() {

77
src/catalog/src/tables.rs Normal file
View File

@@ -0,0 +1,77 @@
// 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.
// The `tables` table in system catalog keeps a record of all tables created by user.
use std::sync::Arc;
use table::metadata::TableId;
use crate::system::SystemCatalogTable;
use crate::DeregisterTableRequest;
pub struct InformationSchema {
pub system: Arc<SystemCatalogTable>,
}
pub struct SystemCatalog {
pub information_schema: Arc<InformationSchema>,
}
impl SystemCatalog {
pub(crate) fn new(system: SystemCatalogTable) -> Self {
let schema = InformationSchema {
system: Arc::new(system),
};
Self {
information_schema: Arc::new(schema),
}
}
pub async fn register_table(
&self,
catalog: String,
schema: String,
table_name: String,
table_id: TableId,
engine: String,
) -> crate::error::Result<usize> {
self.information_schema
.system
.register_table(catalog, schema, table_name, table_id, engine)
.await
}
pub(crate) async fn deregister_table(
&self,
request: &DeregisterTableRequest,
table_id: TableId,
) -> crate::error::Result<()> {
self.information_schema
.system
.deregister_table(request, table_id)
.await
}
pub async fn register_schema(
&self,
catalog: String,
schema: String,
) -> crate::error::Result<usize> {
self.information_schema
.system
.register_schema(catalog, schema)
.await
}
}

View File

@@ -0,0 +1,175 @@
// 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.
#[cfg(test)]
mod tests {
use std::sync::Arc;
use catalog::local::LocalCatalogManager;
use catalog::{CatalogManager, RegisterTableRequest, RenameTableRequest};
use common_catalog::consts::{DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME};
use common_telemetry::{error, info};
use common_test_util::temp_dir::TempDir;
use mito::config::EngineConfig;
use table::engine::manager::MemoryTableEngineManager;
use table::table::numbers::NumbersTable;
use tokio::sync::Mutex;
async fn create_local_catalog_manager(
) -> Result<(TempDir, LocalCatalogManager), catalog::error::Error> {
let (dir, object_store) =
mito::table::test_util::new_test_object_store("setup_mock_engine_and_table").await;
let mock_engine = Arc::new(mito::table::test_util::MockMitoEngine::new(
EngineConfig::default(),
mito::table::test_util::MockEngine::default(),
object_store,
));
let engine_manager = Arc::new(MemoryTableEngineManager::new(mock_engine.clone()));
let catalog_manager = LocalCatalogManager::try_new(engine_manager).await.unwrap();
catalog_manager.start().await?;
Ok((dir, catalog_manager))
}
#[tokio::test]
async fn test_rename_table() {
common_telemetry::init_default_ut_logging();
let (_dir, catalog_manager) = create_local_catalog_manager().await.unwrap();
// register table
let table_name = "test_table";
let table_id = 42;
let request = RegisterTableRequest {
catalog: DEFAULT_CATALOG_NAME.to_string(),
schema: DEFAULT_SCHEMA_NAME.to_string(),
table_name: table_name.to_string(),
table_id,
table: NumbersTable::table(table_id),
};
assert!(catalog_manager.register_table(request).await.unwrap());
// rename table
let new_table_name = "table_t";
let rename_table_req = RenameTableRequest {
catalog: DEFAULT_CATALOG_NAME.to_string(),
schema: DEFAULT_SCHEMA_NAME.to_string(),
table_name: table_name.to_string(),
new_table_name: new_table_name.to_string(),
table_id,
};
assert!(catalog_manager
.rename_table(rename_table_req)
.await
.unwrap());
let registered_table = catalog_manager
.table(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, new_table_name)
.await
.unwrap()
.unwrap();
assert_eq!(registered_table.table_info().ident.table_id, table_id);
}
#[tokio::test]
async fn test_duplicate_register() {
let (_dir, catalog_manager) = create_local_catalog_manager().await.unwrap();
let request = RegisterTableRequest {
catalog: DEFAULT_CATALOG_NAME.to_string(),
schema: DEFAULT_SCHEMA_NAME.to_string(),
table_name: "test_table".to_string(),
table_id: 42,
table: NumbersTable::table(42),
};
assert!(catalog_manager
.register_table(request.clone())
.await
.unwrap());
// register table with same table id will succeed with 0 as return val.
assert!(!catalog_manager.register_table(request).await.unwrap());
let err = catalog_manager
.register_table(RegisterTableRequest {
catalog: DEFAULT_CATALOG_NAME.to_string(),
schema: DEFAULT_SCHEMA_NAME.to_string(),
table_name: "test_table".to_string(),
table_id: 43,
table: NumbersTable::table(43),
})
.await
.unwrap_err();
assert!(
err.to_string()
.contains("Table `greptime.public.test_table` already exists"),
"Actual error message: {err}",
);
}
#[test]
fn test_concurrent_register() {
common_telemetry::init_default_ut_logging();
let rt = Arc::new(tokio::runtime::Builder::new_multi_thread().build().unwrap());
let (_dir, catalog_manager) =
rt.block_on(async { create_local_catalog_manager().await.unwrap() });
let catalog_manager = Arc::new(catalog_manager);
let succeed = Arc::new(Mutex::new(None));
let mut handles = Vec::with_capacity(8);
for i in 0..8 {
let catalog = catalog_manager.clone();
let succeed = succeed.clone();
let handle = rt.spawn(async move {
let table_id = 42 + i;
let table = NumbersTable::table(table_id);
let table_info = table.table_info();
let req = RegisterTableRequest {
catalog: DEFAULT_CATALOG_NAME.to_string(),
schema: DEFAULT_SCHEMA_NAME.to_string(),
table_name: "test_table".to_string(),
table_id,
table,
};
match catalog.register_table(req).await {
Ok(res) => {
if res {
let mut succeed = succeed.lock().await;
info!("Successfully registered table: {}", table_id);
*succeed = Some(table_info);
}
}
Err(_) => {
error!("Failed to register table {}", table_id);
}
}
});
handles.push(handle);
}
rt.block_on(async move {
for handle in handles {
handle.await.unwrap();
}
let guard = succeed.lock().await;
let table_info = guard.as_ref().unwrap();
let table_registered = catalog_manager
.table(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, "test_table")
.await
.unwrap()
.unwrap();
assert_eq!(
table_registered.table_info().ident.table_id,
table_info.ident.table_id
);
});
}
}

View File

@@ -0,0 +1,453 @@
// 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 std::collections::HashSet;
use std::sync::Arc;
use std::time::Duration;
use catalog::remote::mock::MockTableEngine;
use catalog::remote::region_alive_keeper::RegionAliveKeepers;
use catalog::remote::{CachedMetaKvBackend, RemoteCatalogManager};
use catalog::{CatalogManager, RegisterSchemaRequest, RegisterTableRequest};
use common_catalog::consts::{
DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, INFORMATION_SCHEMA_NAME, MITO_ENGINE,
};
use common_meta::helper::CatalogValue;
use common_meta::ident::TableIdent;
use common_meta::key::catalog_name::CatalogNameKey;
use common_meta::key::TableMetadataManager;
use common_meta::kv_backend::memory::MemoryKvBackend;
use common_meta::kv_backend::KvBackend;
use common_meta::rpc::store::{CompareAndPutRequest, PutRequest};
use datatypes::schema::RawSchema;
use table::engine::manager::{MemoryTableEngineManager, TableEngineManagerRef};
use table::engine::{EngineContext, TableEngineRef};
use table::requests::CreateTableRequest;
use table::test_util::EmptyTable;
use tokio::time::Instant;
struct TestingComponents {
catalog_manager: Arc<RemoteCatalogManager>,
table_engine_manager: TableEngineManagerRef,
region_alive_keepers: Arc<RegionAliveKeepers>,
}
impl TestingComponents {
fn table_engine(&self) -> TableEngineRef {
self.table_engine_manager.engine(MITO_ENGINE).unwrap()
}
}
#[tokio::test]
async fn test_cached_backend() {
let backend = CachedMetaKvBackend::wrap(Arc::new(MemoryKvBackend::default()));
let default_catalog_key = CatalogNameKey::new(DEFAULT_CATALOG_NAME).to_string();
let req = PutRequest::new()
.with_key(default_catalog_key.as_bytes())
.with_value(CatalogValue.as_bytes().unwrap());
backend.put(req).await.unwrap();
let ret = backend.get(b"__catalog_name/greptime").await.unwrap();
let _ = ret.unwrap();
let req = CompareAndPutRequest::new()
.with_key(b"__catalog_name/greptime".to_vec())
.with_expect(CatalogValue.as_bytes().unwrap())
.with_value(b"123".to_vec());
let _ = backend.compare_and_put(req).await.unwrap();
let ret = backend.get(b"__catalog_name/greptime").await.unwrap();
assert_eq!(b"123", ret.as_ref().unwrap().value.as_slice());
let req = PutRequest::new()
.with_key(b"__catalog_name/greptime".to_vec())
.with_value(b"1234".to_vec());
let _ = backend.put(req).await;
let ret = backend.get(b"__catalog_name/greptime").await.unwrap();
assert_eq!(b"1234", ret.unwrap().value.as_slice());
backend
.delete(b"__catalog_name/greptime", false)
.await
.unwrap();
let ret = backend.get(b"__catalog_name/greptime").await.unwrap();
assert!(ret.is_none());
}
async fn prepare_components(node_id: u64) -> TestingComponents {
let backend = Arc::new(MemoryKvBackend::default());
let req = PutRequest::new()
.with_key(b"__catalog_name/greptime".to_vec())
.with_value(b"".to_vec());
backend.put(req).await.unwrap();
let req = PutRequest::new()
.with_key(b"__schema_name/greptime-public".to_vec())
.with_value(b"".to_vec());
backend.put(req).await.unwrap();
let cached_backend = Arc::new(CachedMetaKvBackend::wrap(backend));
let table_engine = Arc::new(MockTableEngine::default());
let engine_manager = Arc::new(MemoryTableEngineManager::alias(
MITO_ENGINE.to_string(),
table_engine,
));
let region_alive_keepers = Arc::new(RegionAliveKeepers::new(engine_manager.clone(), 5000));
let catalog_manager = RemoteCatalogManager::new(
engine_manager.clone(),
node_id,
region_alive_keepers.clone(),
Arc::new(TableMetadataManager::new(cached_backend)),
);
catalog_manager.start().await.unwrap();
TestingComponents {
catalog_manager: Arc::new(catalog_manager),
table_engine_manager: engine_manager,
region_alive_keepers,
}
}
#[tokio::test]
async fn test_remote_catalog_default() {
common_telemetry::init_default_ut_logging();
let node_id = 42;
let TestingComponents {
catalog_manager, ..
} = prepare_components(node_id).await;
assert_eq!(
vec![DEFAULT_CATALOG_NAME.to_string()],
catalog_manager.catalog_names().await.unwrap()
);
let mut schema_names = catalog_manager
.schema_names(DEFAULT_CATALOG_NAME)
.await
.unwrap();
schema_names.sort_unstable();
assert_eq!(
vec![
INFORMATION_SCHEMA_NAME.to_string(),
DEFAULT_SCHEMA_NAME.to_string()
],
schema_names
);
}
#[tokio::test]
async fn test_remote_catalog_register_nonexistent() {
common_telemetry::init_default_ut_logging();
let node_id = 42;
let components = prepare_components(node_id).await;
// register a new table with an nonexistent catalog
let catalog_name = "nonexistent_catalog".to_string();
let schema_name = "nonexistent_schema".to_string();
let table_name = "fail_table".to_string();
// this schema has no effect
let table_schema = RawSchema::new(vec![]);
let table = components
.table_engine()
.create_table(
&EngineContext {},
CreateTableRequest {
id: 1,
catalog_name: catalog_name.clone(),
schema_name: schema_name.clone(),
table_name: table_name.clone(),
desc: None,
schema: table_schema,
region_numbers: vec![0],
primary_key_indices: vec![],
create_if_not_exists: false,
table_options: Default::default(),
engine: MITO_ENGINE.to_string(),
},
)
.await
.unwrap();
let reg_req = RegisterTableRequest {
catalog: catalog_name,
schema: schema_name,
table_name,
table_id: 1,
table,
};
let res = components.catalog_manager.register_table(reg_req).await;
// because nonexistent_catalog does not exist yet.
assert_matches!(
res.err().unwrap(),
catalog::error::Error::CatalogNotFound { .. }
);
}
#[tokio::test]
async fn test_register_table() {
let node_id = 42;
let components = prepare_components(node_id).await;
let mut schema_names = components
.catalog_manager
.schema_names(DEFAULT_CATALOG_NAME)
.await
.unwrap();
schema_names.sort_unstable();
assert_eq!(
vec![
INFORMATION_SCHEMA_NAME.to_string(),
DEFAULT_SCHEMA_NAME.to_string(),
],
schema_names
);
// register a new table with an nonexistent catalog
let catalog_name = DEFAULT_CATALOG_NAME.to_string();
let schema_name = DEFAULT_SCHEMA_NAME.to_string();
let table_name = "test_table".to_string();
let table_id = 1;
// this schema has no effect
let table_schema = RawSchema::new(vec![]);
let table = components
.table_engine()
.create_table(
&EngineContext {},
CreateTableRequest {
id: table_id,
catalog_name: catalog_name.clone(),
schema_name: schema_name.clone(),
table_name: table_name.clone(),
desc: None,
schema: table_schema,
region_numbers: vec![0],
primary_key_indices: vec![],
create_if_not_exists: false,
table_options: Default::default(),
engine: MITO_ENGINE.to_string(),
},
)
.await
.unwrap();
let reg_req = RegisterTableRequest {
catalog: catalog_name,
schema: schema_name,
table_name: table_name.clone(),
table_id,
table,
};
assert!(components
.catalog_manager
.register_table(reg_req)
.await
.unwrap());
assert_eq!(
vec![table_name],
components
.catalog_manager
.table_names(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME)
.await
.unwrap()
);
}
#[tokio::test]
async fn test_register_catalog_schema_table() {
let node_id = 42;
let components = prepare_components(node_id).await;
let catalog_name = "test_catalog".to_string();
let schema_name = "nonexistent_schema".to_string();
// register catalog to catalog manager
assert!(components
.catalog_manager
.clone()
.register_catalog(catalog_name.clone())
.await
.is_ok());
assert_eq!(
HashSet::<String>::from_iter(vec![
DEFAULT_CATALOG_NAME.to_string(),
catalog_name.clone()
]),
HashSet::from_iter(components.catalog_manager.catalog_names().await.unwrap())
);
let table_to_register = components
.table_engine()
.create_table(
&EngineContext {},
CreateTableRequest {
id: 2,
catalog_name: catalog_name.clone(),
schema_name: schema_name.clone(),
table_name: "".to_string(),
desc: None,
schema: RawSchema::new(vec![]),
region_numbers: vec![0],
primary_key_indices: vec![],
create_if_not_exists: false,
table_options: Default::default(),
engine: MITO_ENGINE.to_string(),
},
)
.await
.unwrap();
let reg_req = RegisterTableRequest {
catalog: catalog_name.clone(),
schema: schema_name.clone(),
table_name: " fail_table".to_string(),
table_id: 2,
table: table_to_register,
};
// this register will fail since schema does not exist yet
assert_matches!(
components
.catalog_manager
.register_table(reg_req.clone())
.await
.unwrap_err(),
catalog::error::Error::SchemaNotFound { .. }
);
let register_schema_request = RegisterSchemaRequest {
catalog: catalog_name.to_string(),
schema: schema_name.to_string(),
};
assert!(components
.catalog_manager
.register_schema(register_schema_request)
.await
.expect("Register schema should not fail"));
assert!(components
.catalog_manager
.register_table(reg_req)
.await
.unwrap());
assert_eq!(
HashSet::from([schema_name.clone(), INFORMATION_SCHEMA_NAME.to_string()]),
components
.catalog_manager
.schema_names(&catalog_name)
.await
.unwrap()
.into_iter()
.collect()
)
}
#[tokio::test]
async fn test_register_table_before_and_after_region_alive_keeper_started() {
let components = prepare_components(42).await;
let catalog_manager = &components.catalog_manager;
let region_alive_keepers = &components.region_alive_keepers;
let table_before = TableIdent {
catalog: DEFAULT_CATALOG_NAME.to_string(),
schema: DEFAULT_SCHEMA_NAME.to_string(),
table: "table_before".to_string(),
table_id: 1,
engine: MITO_ENGINE.to_string(),
};
let request = RegisterTableRequest {
catalog: table_before.catalog.clone(),
schema: table_before.schema.clone(),
table_name: table_before.table.clone(),
table_id: table_before.table_id,
table: Arc::new(EmptyTable::new(CreateTableRequest {
id: table_before.table_id,
catalog_name: table_before.catalog.clone(),
schema_name: table_before.schema.clone(),
table_name: table_before.table.clone(),
desc: None,
schema: RawSchema::new(vec![]),
region_numbers: vec![0],
primary_key_indices: vec![],
create_if_not_exists: false,
table_options: Default::default(),
engine: MITO_ENGINE.to_string(),
})),
};
assert!(catalog_manager.register_table(request).await.unwrap());
let keeper = region_alive_keepers
.find_keeper(table_before.table_id)
.await
.unwrap();
let deadline = keeper.deadline(0).await.unwrap();
let far_future = Instant::now() + Duration::from_secs(86400 * 365 * 29);
// assert region alive countdown is not started
assert!(deadline > far_future);
region_alive_keepers.start().await;
let table_after = TableIdent {
catalog: DEFAULT_CATALOG_NAME.to_string(),
schema: DEFAULT_SCHEMA_NAME.to_string(),
table: "table_after".to_string(),
table_id: 2,
engine: MITO_ENGINE.to_string(),
};
let request = RegisterTableRequest {
catalog: table_after.catalog.clone(),
schema: table_after.schema.clone(),
table_name: table_after.table.clone(),
table_id: table_after.table_id,
table: Arc::new(EmptyTable::new(CreateTableRequest {
id: table_after.table_id,
catalog_name: table_after.catalog.clone(),
schema_name: table_after.schema.clone(),
table_name: table_after.table.clone(),
desc: None,
schema: RawSchema::new(vec![]),
region_numbers: vec![0],
primary_key_indices: vec![],
create_if_not_exists: false,
table_options: Default::default(),
engine: MITO_ENGINE.to_string(),
})),
};
assert!(catalog_manager.register_table(request).await.unwrap());
let keeper = region_alive_keepers
.find_keeper(table_after.table_id)
.await
.unwrap();
let deadline = keeper.deadline(0).await.unwrap();
// assert countdown is started for the table registered after [RegionAliveKeepers] started
assert!(deadline <= Instant::now() + Duration::from_secs(20));
let keeper = region_alive_keepers
.find_keeper(table_before.table_id)
.await
.unwrap();
let deadline = keeper.deadline(0).await.unwrap();
// assert countdown is started for the table registered before [RegionAliveKeepers] started, too
assert!(deadline <= Instant::now() + Duration::from_secs(20));
}
}

View File

@@ -11,12 +11,10 @@ testing = []
api = { workspace = true }
arrow-flight.workspace = true
async-stream.workspace = true
async-trait.workspace = true
common-base = { workspace = true }
common-catalog = { workspace = true }
common-error = { workspace = true }
common-grpc = { workspace = true }
common-macro = { workspace = true }
common-meta = { workspace = true }
common-query = { workspace = true }
common-recordbatch = { workspace = true }
@@ -27,11 +25,10 @@ datatypes = { workspace = true }
derive_builder.workspace = true
enum_dispatch = "0.3"
futures-util.workspace = true
moka = { workspace = true, features = ["future"] }
moka = { version = "0.9", features = ["future"] }
parking_lot = "0.12"
prost.workspace = true
rand.workspace = true
session = { workspace = true }
snafu.workspace = true
tokio-stream = { version = "0.1", features = ["net"] }
tokio.workspace = true

View File

@@ -12,7 +12,7 @@
// See the License for the specific language governing permissions and
// limitations under the License.
use api::v1::{ColumnDataType, ColumnDef, CreateTableExpr, SemanticType, TableId};
use api::v1::{ColumnDataType, ColumnDef, CreateTableExpr, TableId};
use client::{Client, Database};
use common_catalog::consts::{DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, MITO_ENGINE};
use prost::Message;
@@ -41,27 +41,21 @@ async fn run() {
column_defs: vec![
ColumnDef {
name: "timestamp".to_string(),
data_type: ColumnDataType::TimestampMillisecond as i32,
datatype: ColumnDataType::TimestampMillisecond as i32,
is_nullable: false,
default_constraint: vec![],
semantic_type: SemanticType::Timestamp as i32,
comment: String::new(),
},
ColumnDef {
name: "key".to_string(),
data_type: ColumnDataType::Uint64 as i32,
datatype: ColumnDataType::Uint64 as i32,
is_nullable: false,
default_constraint: vec![],
semantic_type: SemanticType::Tag as i32,
comment: String::new(),
},
ColumnDef {
name: "value".to_string(),
data_type: ColumnDataType::Uint64 as i32,
datatype: ColumnDataType::Uint64 as i32,
is_nullable: false,
default_constraint: vec![],
semantic_type: SemanticType::Field as i32,
comment: String::new(),
},
],
time_index: "timestamp".to_string(),
@@ -69,6 +63,7 @@ async fn run() {
create_if_not_exists: false,
table_options: Default::default(),
table_id: Some(TableId { id: 1024 }),
region_numbers: vec![0],
engine: MITO_ENGINE.to_string(),
};
@@ -78,7 +73,7 @@ async fn run() {
let logical = mock_logical_plan();
event!(Level::INFO, "plan size: {:#?}", logical.len());
let result = db.logical_plan(logical, 0).await.unwrap();
let result = db.logical_plan(logical, None).await.unwrap();
event!(Level::INFO, "result: {:#?}", result);
}

View File

@@ -42,14 +42,14 @@ async fn run() {
.insert(vec![to_insert_request(weather_records_1())])
.await
{
error!("Error: {e:?}");
error!("Error: {e}");
}
if let Err(e) = stream_inserter
.insert(vec![to_insert_request(weather_records_2())])
.await
{
error!("Error: {e:?}");
error!("Error: {e}");
}
let result = stream_inserter.finish().await;
@@ -59,7 +59,7 @@ async fn run() {
info!("Rows written: {rows}");
}
Err(e) => {
error!("Error: {e:?}");
error!("Error: {e}");
}
};
}
@@ -131,7 +131,7 @@ fn to_insert_request(records: Vec<WeatherRecord>) -> InsertRequest {
Column {
column_name: "ts".to_owned(),
values: Some(column::Values {
timestamp_millisecond_values: timestamp_millis,
ts_millisecond_values: timestamp_millis,
..Default::default()
}),
semantic_type: SemanticType::Timestamp as i32,
@@ -177,5 +177,6 @@ fn to_insert_request(records: Vec<WeatherRecord>) -> InsertRequest {
table_name: "weather_demo".to_owned(),
columns,
row_count: rows as u32,
..Default::default()
}
}

View File

@@ -138,46 +138,24 @@ impl Client {
Ok((addr, channel))
}
fn max_grpc_recv_message_size(&self) -> usize {
self.inner
.channel_manager
.config()
.max_recv_message_size
.as_bytes() as usize
}
fn max_grpc_send_message_size(&self) -> usize {
self.inner
.channel_manager
.config()
.max_send_message_size
.as_bytes() as usize
}
pub(crate) fn make_flight_client(&self) -> Result<FlightClient> {
let (addr, channel) = self.find_channel()?;
Ok(FlightClient {
addr,
client: FlightServiceClient::new(channel)
.max_decoding_message_size(self.max_grpc_recv_message_size())
.max_encoding_message_size(self.max_grpc_send_message_size()),
client: FlightServiceClient::new(channel),
})
}
pub(crate) fn make_database_client(&self) -> Result<DatabaseClient> {
let (_, channel) = self.find_channel()?;
Ok(DatabaseClient {
inner: GreptimeDatabaseClient::new(channel)
.max_decoding_message_size(self.max_grpc_recv_message_size())
.max_encoding_message_size(self.max_grpc_send_message_size()),
inner: GreptimeDatabaseClient::new(channel),
})
}
pub(crate) fn raw_region_client(&self) -> Result<PbRegionClient<Channel>> {
let (_, channel) = self.find_channel()?;
Ok(PbRegionClient::new(channel)
.max_decoding_message_size(self.max_grpc_recv_message_size())
.max_encoding_message_size(self.max_grpc_send_message_size()))
Ok(PbRegionClient::new(channel))
}
pub fn make_prometheus_gateway_client(&self) -> Result<PrometheusGatewayClient<Channel>> {

View File

@@ -13,15 +13,12 @@
// limitations under the License.
use std::fmt::{Debug, Formatter};
use std::sync::Arc;
use std::time::Duration;
use common_grpc::channel_manager::{ChannelConfig, ChannelManager};
use common_meta::datanode_manager::{Datanode, DatanodeManager};
use common_meta::peer::Peer;
use moka::future::{Cache, CacheBuilder};
use crate::region::RegionRequester;
use crate::Client;
pub struct DatanodeClients {
@@ -43,15 +40,6 @@ impl Debug for DatanodeClients {
}
}
#[async_trait::async_trait]
impl DatanodeManager for DatanodeClients {
async fn datanode(&self, datanode: &Peer) -> Arc<dyn Datanode> {
let client = self.get_client(datanode).await;
Arc::new(RegionRequester::new(client))
}
}
impl DatanodeClients {
pub fn new(config: ChannelConfig) -> Self {
Self {

View File

@@ -17,9 +17,9 @@ use api::v1::ddl_request::Expr as DdlExpr;
use api::v1::greptime_request::Request;
use api::v1::query_request::Query;
use api::v1::{
AlterExpr, AuthHeader, CreateTableExpr, DdlRequest, DeleteRequests, DropTableExpr,
GreptimeRequest, InsertRequests, PromRangeQuery, QueryRequest, RequestHeader,
RowInsertRequests, TruncateTableExpr,
AlterExpr, AuthHeader, CompactTableExpr, CreateTableExpr, DdlRequest, DeleteRequests,
DropTableExpr, FlushTableExpr, GreptimeRequest, InsertRequests, PromRangeQuery, QueryRequest,
RequestHeader, RowInsertRequests, TruncateTableExpr,
};
use arrow_flight::Ticket;
use async_stream::stream;
@@ -147,13 +147,13 @@ impl Database {
async fn handle(&self, request: Request) -> Result<u32> {
let mut client = self.client.make_database_client()?.inner;
let request = self.to_rpc_request(request, 0);
let request = self.to_rpc_request(request, None);
let response = client.handle(request).await?.into_inner();
from_grpc_response(response)
}
#[inline]
fn to_rpc_request(&self, request: Request, trace_id: u64) -> GreptimeRequest {
fn to_rpc_request(&self, request: Request, trace_id: Option<u64>) -> GreptimeRequest {
GreptimeRequest {
header: Some(RequestHeader {
catalog: self.catalog.clone(),
@@ -161,27 +161,28 @@ impl Database {
authorization: self.ctx.auth_header.clone(),
dbname: self.dbname.clone(),
trace_id,
span_id: 0,
span_id: None,
}),
request: Some(request),
}
}
pub async fn sql<S>(&self, sql: S) -> Result<Output>
where
S: AsRef<str>,
{
pub async fn sql(&self, sql: &str) -> Result<Output> {
let _timer = timer!(metrics::METRIC_GRPC_SQL);
self.do_get(
Request::Query(QueryRequest {
query: Some(Query::Sql(sql.as_ref().to_string())),
query: Some(Query::Sql(sql.to_string())),
}),
0,
None,
)
.await
}
pub async fn logical_plan(&self, logical_plan: Vec<u8>, trace_id: u64) -> Result<Output> {
pub async fn logical_plan(
&self,
logical_plan: Vec<u8>,
trace_id: Option<u64>,
) -> Result<Output> {
let _timer = timer!(metrics::METRIC_GRPC_LOGICAL_PLAN);
self.do_get(
Request::Query(QueryRequest {
@@ -209,7 +210,7 @@ impl Database {
step: step.to_string(),
})),
}),
0,
None,
)
.await
}
@@ -220,7 +221,7 @@ impl Database {
Request::Ddl(DdlRequest {
expr: Some(DdlExpr::CreateTable(expr)),
}),
0,
None,
)
.await
}
@@ -231,7 +232,7 @@ impl Database {
Request::Ddl(DdlRequest {
expr: Some(DdlExpr::Alter(expr)),
}),
0,
None,
)
.await
}
@@ -242,7 +243,29 @@ impl Database {
Request::Ddl(DdlRequest {
expr: Some(DdlExpr::DropTable(expr)),
}),
0,
None,
)
.await
}
pub async fn flush_table(&self, expr: FlushTableExpr) -> Result<Output> {
let _timer = timer!(metrics::METRIC_GRPC_FLUSH_TABLE);
self.do_get(
Request::Ddl(DdlRequest {
expr: Some(DdlExpr::FlushTable(expr)),
}),
None,
)
.await
}
pub async fn compact_table(&self, expr: CompactTableExpr) -> Result<Output> {
let _timer = timer!(metrics::METRIC_GRPC_COMPACT_TABLE);
self.do_get(
Request::Ddl(DdlRequest {
expr: Some(DdlExpr::CompactTable(expr)),
}),
None,
)
.await
}
@@ -253,12 +276,12 @@ impl Database {
Request::Ddl(DdlRequest {
expr: Some(DdlExpr::TruncateTable(expr)),
}),
0,
None,
)
.await
}
async fn do_get(&self, request: Request, trace_id: u64) -> Result<Output> {
async fn do_get(&self, request: Request, trace_id: Option<u64>) -> Result<Output> {
// FIXME(paomian): should be added some labels for metrics
let _timer = timer!(metrics::METRIC_GRPC_DO_GET);
let request = self.to_rpc_request(request, trace_id);
@@ -279,7 +302,7 @@ impl Database {
source: BoxedError::new(ServerSnafu { code, msg }.build()),
};
logging::error!(
"Failed to do Flight get, addr: {}, code: {}, source: {:?}",
"Failed to do Flight get, addr: {}, code: {}, source: {}",
client.addr(),
tonic_code,
error

View File

@@ -17,37 +17,29 @@ use std::any::Any;
use common_error::ext::{BoxedError, ErrorExt};
use common_error::status_code::StatusCode;
use common_error::{GREPTIME_ERROR_CODE, GREPTIME_ERROR_MSG};
use common_macro::stack_trace_debug;
use snafu::{Location, Snafu};
use tonic::{Code, Status};
#[derive(Snafu)]
#[derive(Debug, Snafu)]
#[snafu(visibility(pub))]
#[stack_trace_debug]
pub enum Error {
#[snafu(display("Illegal Flight messages, reason: {}", reason))]
IllegalFlightMessages { reason: String, location: Location },
#[snafu(display("Failed to do Flight get, code: {}", tonic_code))]
#[snafu(display("Failed to do Flight get, code: {}, source: {}", tonic_code, source))]
FlightGet {
addr: String,
tonic_code: Code,
source: BoxedError,
},
#[snafu(display("Failure occurs during handling request"))]
HandleRequest {
location: Location,
source: BoxedError,
},
#[snafu(display("Failed to convert FlightData"))]
#[snafu(display("Failed to convert FlightData, source: {}", source))]
ConvertFlightData {
location: Location,
source: common_grpc::Error,
},
#[snafu(display("Column datatype error"))]
#[snafu(display("Column datatype error, source: {}", source))]
ColumnDataType {
location: Location,
source: api::error::Error,
@@ -59,16 +51,17 @@ pub enum Error {
#[snafu(display("Missing required field in protobuf, field: {}", field))]
MissingField { field: String, location: Location },
#[snafu(display("Failed to create gRPC channel, peer address: {}", addr))]
#[snafu(display(
"Failed to create gRPC channel, peer address: {}, source: {}",
addr,
source
))]
CreateChannel {
addr: String,
location: Location,
source: common_grpc::error::Error,
},
#[snafu(display("Failed to request RegionServer, code: {}", code))]
RegionServer { code: Code, source: BoxedError },
// Server error carried in Tonic Status's metadata.
#[snafu(display("{}", msg))]
Server { code: StatusCode, msg: String },
@@ -92,9 +85,7 @@ impl ErrorExt for Error {
| Error::ClientStreaming { .. } => StatusCode::Internal,
Error::Server { code, .. } => *code,
Error::FlightGet { source, .. }
| Error::HandleRequest { source, .. }
| Error::RegionServer { source, .. } => source.status_code(),
Error::FlightGet { source, .. } => source.status_code(),
Error::CreateChannel { source, .. } | Error::ConvertFlightData { source, .. } => {
source.status_code()
}

View File

@@ -26,8 +26,6 @@ use api::v1::greptime_response::Response;
use api::v1::{AffectedRows, GreptimeResponse};
pub use common_catalog::consts::{DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME};
use common_error::status_code::StatusCode;
pub use common_query::Output;
pub use common_recordbatch::{RecordBatches, SendableRecordBatchStream};
use snafu::OptionExt;
pub use self::client::Client;

View File

@@ -21,6 +21,8 @@ pub const METRIC_GRPC_SQL: &str = "grpc.sql";
pub const METRIC_GRPC_LOGICAL_PLAN: &str = "grpc.logical_plan";
pub const METRIC_GRPC_ALTER: &str = "grpc.alter";
pub const METRIC_GRPC_DROP_TABLE: &str = "grpc.drop_table";
pub const METRIC_GRPC_FLUSH_TABLE: &str = "grpc.flush_table";
pub const METRIC_GRPC_COMPACT_TABLE: &str = "grpc.compact_table";
pub const METRIC_GRPC_TRUNCATE_TABLE: &str = "grpc.truncate_table";
pub const METRIC_GRPC_DO_GET: &str = "grpc.do_get";
pub(crate) const METRIC_REGION_REQUEST_GRPC: &str = "grpc.region_request";

View File

@@ -12,146 +12,44 @@
// See the License for the specific language governing permissions and
// limitations under the License.
use api::v1::region::{QueryRequest, RegionRequest, RegionResponse};
use api::v1::region::{region_request, RegionRequest, RegionRequestHeader, RegionResponse};
use api::v1::ResponseHeader;
use arrow_flight::Ticket;
use async_stream::stream;
use async_trait::async_trait;
use common_error::ext::{BoxedError, ErrorExt};
use common_error::status_code::StatusCode;
use common_grpc::flight::{FlightDecoder, FlightMessage};
use common_meta::datanode_manager::{AffectedRows, Datanode};
use common_meta::error::{self as meta_error, Result as MetaResult};
use common_recordbatch::error::ExternalSnafu;
use common_recordbatch::{RecordBatchStreamAdaptor, SendableRecordBatchStream};
use common_telemetry::{error, timer};
use prost::Message;
use snafu::{location, Location, OptionExt, ResultExt};
use tokio_stream::StreamExt;
use common_telemetry::timer;
use snafu::OptionExt;
use crate::error::Error::RegionServer;
use crate::error::{
self, ConvertFlightDataSnafu, IllegalDatabaseResponseSnafu, IllegalFlightMessagesSnafu,
MissingFieldSnafu, Result, ServerSnafu,
};
use crate::{metrics, Client, Error};
use crate::error::{IllegalDatabaseResponseSnafu, Result, ServerSnafu};
use crate::{metrics, Client};
type AffectedRows = u64;
#[derive(Debug)]
pub struct RegionRequester {
trace_id: Option<u64>,
span_id: Option<u64>,
client: Client,
}
#[async_trait]
impl Datanode for RegionRequester {
async fn handle(&self, request: RegionRequest) -> MetaResult<AffectedRows> {
self.handle_inner(request).await.map_err(|err| {
if matches!(err, RegionServer { .. }) {
meta_error::Error::RetryLater {
source: BoxedError::new(err),
}
} else {
meta_error::Error::External {
source: BoxedError::new(err),
location: location!(),
}
}
})
}
async fn handle_query(&self, request: QueryRequest) -> MetaResult<SendableRecordBatchStream> {
let ticket = Ticket {
ticket: request.encode_to_vec().into(),
};
self.do_get_inner(ticket)
.await
.map_err(BoxedError::new)
.context(meta_error::ExternalSnafu)
}
}
impl RegionRequester {
pub fn new(client: Client) -> Self {
Self { client }
// TODO(LFC): Pass in trace_id and span_id from some context when we have it.
Self {
trace_id: None,
span_id: None,
client,
}
}
pub async fn do_get_inner(&self, ticket: Ticket) -> Result<SendableRecordBatchStream> {
let mut flight_client = self.client.make_flight_client()?;
let response = flight_client
.mut_inner()
.do_get(ticket)
.await
.map_err(|e| {
let tonic_code = e.code();
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()),
};
error!(
e; "Failed to do Flight get, addr: {}, code: {}",
flight_client.addr(),
tonic_code
);
error
})?;
pub async fn handle(self, request: region_request::Body) -> Result<AffectedRows> {
let request_type = request.as_ref().to_string();
let flight_data_stream = response.into_inner();
let mut decoder = FlightDecoder::default();
let mut flight_message_stream = flight_data_stream.map(move |flight_data| {
flight_data
.map_err(Error::from)
.and_then(|data| decoder.try_decode(data).context(ConvertFlightDataSnafu))
});
let Some(first_flight_message) = flight_message_stream.next().await else {
return IllegalFlightMessagesSnafu {
reason: "Expect the response not to be empty",
}
.fail();
let request = RegionRequest {
header: Some(RegionRequestHeader {
trace_id: self.trace_id,
span_id: self.span_id,
}),
body: Some(request),
};
let FlightMessage::Schema(schema) = first_flight_message? else {
return IllegalFlightMessagesSnafu {
reason: "Expect schema to be the first flight message",
}
.fail();
};
let stream = Box::pin(stream!({
while let Some(flight_message) = flight_message_stream.next().await {
let flight_message = flight_message
.map_err(BoxedError::new)
.context(ExternalSnafu)?;
let FlightMessage::Recordbatch(record_batch) = flight_message else {
yield IllegalFlightMessagesSnafu {
reason: "A Schema message must be succeeded exclusively by a set of RecordBatch messages"
}
.fail()
.map_err(BoxedError::new)
.context(ExternalSnafu);
break;
};
yield Ok(record_batch);
}
}));
let record_batch_stream = RecordBatchStreamAdaptor {
schema,
stream,
output_ordering: None,
};
Ok(Box::pin(record_batch_stream))
}
async fn handle_inner(&self, request: RegionRequest) -> Result<AffectedRows> {
let request_type = request
.body
.as_ref()
.with_context(|| MissingFieldSnafu { field: "body" })?
.as_ref()
.to_string();
let _timer = timer!(
metrics::METRIC_REGION_REQUEST_GRPC,
@@ -163,31 +61,15 @@ impl RegionRequester {
let RegionResponse {
header,
affected_rows,
} = client
.handle(request)
.await
.map_err(|e| {
let code = e.code();
let err: error::Error = e.into();
// Uses `Error::RegionServer` instead of `Error::Server`
error::Error::RegionServer {
code,
source: BoxedError::new(err),
}
})?
.into_inner();
} = client.handle(request).await?.into_inner();
check_response_header(header)?;
Ok(affected_rows)
}
pub async fn handle(&self, request: RegionRequest) -> Result<AffectedRows> {
self.handle_inner(request).await
}
}
pub fn check_response_header(header: Option<ResponseHeader>) -> Result<()> {
fn check_response_header(header: Option<ResponseHeader>) -> Result<()> {
let status = header
.and_then(|header| header.status)
.context(IllegalDatabaseResponseSnafu {

View File

@@ -23,12 +23,8 @@ chrono.workspace = true
clap = { version = "3.1", features = ["derive"] }
client = { workspace = true }
common-base = { workspace = true }
common-catalog = { workspace = true }
common-config = { workspace = true }
common-error = { workspace = true }
common-macro = { workspace = true }
common-meta = { workspace = true }
common-procedure = { workspace = true }
common-query = { workspace = true }
common-recordbatch = { workspace = true }
common-telemetry = { workspace = true, features = [
@@ -39,24 +35,18 @@ datanode = { workspace = true }
datatypes = { workspace = true }
either = "1.8"
etcd-client.workspace = true
file-engine = { workspace = true }
frontend = { workspace = true }
futures.workspace = true
lazy_static.workspace = true
meta-client = { workspace = true }
meta-srv = { workspace = true }
metrics.workspace = true
mito2 = { workspace = true }
nu-ansi-term = "0.46"
partition = { workspace = true }
plugins.workspace = true
prost.workspace = true
query = { workspace = true }
rand.workspace = true
regex.workspace = true
rustyline = "10.1"
serde.workspace = true
serde_json.workspace = true
servers = { workspace = true }
session = { workspace = true }
snafu.workspace = true

View File

@@ -116,7 +116,7 @@ impl SubCommand {
Ok(Application::Metasrv(app))
}
(SubCommand::Standalone(cmd), Options::Standalone(opts)) => {
let app = cmd.build(*opts).await?;
let app = cmd.build(opts.fe_opts, opts.dn_opts).await?;
Ok(Application::Standalone(app))
}
(SubCommand::Cli(cmd), Options::Cli(_)) => {

View File

@@ -14,7 +14,6 @@
mod bench;
mod cmd;
mod export;
mod helper;
mod repl;
// TODO(weny): Removes it
@@ -28,7 +27,6 @@ use common_telemetry::logging::LoggingOptions;
pub use repl::Repl;
use upgrade::UpgradeCommand;
use self::export::ExportCommand;
use crate::error::Result;
use crate::options::{Options, TopLevelOptions};
@@ -80,19 +78,17 @@ impl Command {
#[derive(Parser)]
enum SubCommand {
// Attach(AttachCommand),
Attach(AttachCommand),
Upgrade(UpgradeCommand),
Bench(BenchTableMetadataCommand),
Export(ExportCommand),
}
impl SubCommand {
async fn build(self) -> Result<Instance> {
match self {
// SubCommand::Attach(cmd) => cmd.build().await,
SubCommand::Attach(cmd) => cmd.build().await,
SubCommand::Upgrade(cmd) => cmd.build().await,
SubCommand::Bench(cmd) => cmd.build().await,
SubCommand::Export(cmd) => cmd.build().await,
}
}
}
@@ -108,9 +104,51 @@ pub(crate) struct AttachCommand {
}
impl AttachCommand {
#[allow(dead_code)]
async fn build(self) -> Result<Instance> {
let repl = Repl::try_new(&self).await?;
Ok(Instance::Repl(repl))
}
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn test_load_options() {
let cmd = Command {
cmd: SubCommand::Attach(AttachCommand {
grpc_addr: String::from(""),
meta_addr: None,
disable_helper: false,
}),
};
let opts = cmd.load_options(TopLevelOptions::default()).unwrap();
let logging_opts = opts.logging_options();
assert_eq!("/tmp/greptimedb/logs", logging_opts.dir);
assert!(logging_opts.level.is_none());
assert!(!logging_opts.enable_jaeger_tracing);
}
#[test]
fn test_top_level_options() {
let cmd = Command {
cmd: SubCommand::Attach(AttachCommand {
grpc_addr: String::from(""),
meta_addr: None,
disable_helper: false,
}),
};
let opts = cmd
.load_options(TopLevelOptions {
log_dir: Some("/tmp/greptimedb/test/logs".to_string()),
log_level: Some("debug".to_string()),
})
.unwrap();
let logging_opts = opts.logging_options();
assert_eq!("/tmp/greptimedb/test/logs", logging_opts.dir);
assert_eq!("debug", logging_opts.level.as_ref().unwrap());
}
}

View File

@@ -120,6 +120,7 @@ fn create_table_info(table_id: TableId, table_name: TableName) -> RawTableInfo {
created_on: chrono::DateTime::default(),
primary_key_indices: vec![],
next_column_id: columns as u32 + 1,
engine_options: Default::default(),
value_indices: vec![],
options: Default::default(),
region_numbers: (1..=100).collect(),

View File

@@ -1,395 +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::path::Path;
use std::sync::Arc;
use async_trait::async_trait;
use clap::{Parser, ValueEnum};
use client::{Client, Database, DEFAULT_SCHEMA_NAME};
use common_query::Output;
use common_recordbatch::util::collect;
use common_telemetry::{debug, error, info, warn};
use datatypes::scalars::ScalarVector;
use datatypes::vectors::{StringVector, Vector};
use snafu::{OptionExt, ResultExt};
use tokio::fs::File;
use tokio::io::AsyncWriteExt;
use tokio::sync::Semaphore;
use crate::cli::{Instance, Tool};
use crate::error::{
CollectRecordBatchesSnafu, ConnectServerSnafu, EmptyResultSnafu, Error, FileIoSnafu,
InvalidDatabaseNameSnafu, NotDataFromOutputSnafu, RequestDatabaseSnafu, Result,
};
type TableReference = (String, String, String);
#[derive(Debug, Default, Clone, ValueEnum)]
enum ExportTarget {
/// Corresponding to `SHOW CREATE TABLE`
#[default]
CreateTable,
/// Corresponding to `EXPORT TABLE`
TableData,
}
#[derive(Debug, Default, Parser)]
pub struct ExportCommand {
/// Server address to connect
#[clap(long)]
addr: String,
/// Directory to put the exported data. E.g.: /tmp/greptimedb-export
#[clap(long)]
output_dir: String,
/// The name of the catalog to export. Default to "greptime-*"".
#[clap(long, default_value = "")]
database: String,
/// Parallelism of the export.
#[clap(long, short = 'j', default_value = "1")]
export_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)]
target: ExportTarget,
}
impl ExportCommand {
pub async fn build(&self) -> Result<Instance> {
let client = Client::with_urls([self.addr.clone()]);
client
.health_check()
.await
.with_context(|_| ConnectServerSnafu {
addr: self.addr.clone(),
})?;
let (catalog, schema) = split_database(&self.database)?;
let database_client = Database::new(
catalog.clone(),
schema.clone().unwrap_or(DEFAULT_SCHEMA_NAME.to_string()),
client,
);
Ok(Instance::Tool(Box::new(Export {
client: database_client,
catalog,
schema,
output_dir: self.output_dir.clone(),
parallelism: self.export_jobs,
target: self.target.clone(),
})))
}
}
pub struct Export {
client: Database,
catalog: String,
schema: Option<String>,
output_dir: String,
parallelism: usize,
target: ExportTarget,
}
impl Export {
/// 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 mut client = self.client.clone();
client.set_catalog(self.catalog.clone());
let result =
client
.sql("show databases")
.await
.with_context(|_| RequestDatabaseSnafu {
sql: "show databases".to_string(),
})?;
let Output::Stream(stream) = result else {
NotDataFromOutputSnafu.fail()?
};
let record_batch = collect(stream)
.await
.context(CollectRecordBatchesSnafu)?
.pop()
.context(EmptyResultSnafu)?;
let schemas = record_batch
.column(0)
.as_any()
.downcast_ref::<StringVector>()
.unwrap();
let mut result = Vec::with_capacity(schemas.len());
for i in 0..schemas.len() {
let schema = schemas.get_data(i).unwrap().to_owned();
result.push((self.catalog.clone(), schema));
}
Ok(result)
}
}
/// Return a list of [`TableReference`] to be exported.
/// Includes all tables under the given `catalog` and `schema`
async fn get_table_list(&self, catalog: &str, schema: &str) -> Result<Vec<TableReference>> {
// TODO: SQL injection hurts
let sql = format!(
"select table_catalog, table_schema, table_name from \
information_schema.tables where table_type = \'BASE TABLE\'\
and table_catalog = \'{catalog}\' and table_schema = \'{schema}\'",
);
let mut client = self.client.clone();
client.set_catalog(catalog);
client.set_schema(schema);
let result = client
.sql(&sql)
.await
.with_context(|_| RequestDatabaseSnafu { sql })?;
let Output::Stream(stream) = result else {
NotDataFromOutputSnafu.fail()?
};
let Some(record_batch) = collect(stream)
.await
.context(CollectRecordBatchesSnafu)?
.pop()
else {
return Ok(vec![]);
};
debug!("Fetched table list: {}", record_batch.pretty_print());
if record_batch.num_rows() == 0 {
return Ok(vec![]);
}
let mut result = Vec::with_capacity(record_batch.num_rows());
let catalog_column = record_batch
.column(0)
.as_any()
.downcast_ref::<StringVector>()
.unwrap();
let schema_column = record_batch
.column(1)
.as_any()
.downcast_ref::<StringVector>()
.unwrap();
let table_column = record_batch
.column(2)
.as_any()
.downcast_ref::<StringVector>()
.unwrap();
for i in 0..record_batch.num_rows() {
let catalog = catalog_column.get_data(i).unwrap().to_owned();
let schema = schema_column.get_data(i).unwrap().to_owned();
let table = table_column.get_data(i).unwrap().to_owned();
result.push((catalog, schema, table));
}
Ok(result)
}
async fn show_create_table(&self, catalog: &str, schema: &str, table: &str) -> Result<String> {
let sql = format!("show create table {}.{}.{}", catalog, schema, table);
let mut client = self.client.clone();
client.set_catalog(catalog);
client.set_schema(schema);
let result = client
.sql(&sql)
.await
.with_context(|_| RequestDatabaseSnafu { sql })?;
let Output::Stream(stream) = result else {
NotDataFromOutputSnafu.fail()?
};
let record_batch = collect(stream)
.await
.context(CollectRecordBatchesSnafu)?
.pop()
.context(EmptyResultSnafu)?;
let create_table = record_batch
.column(1)
.as_any()
.downcast_ref::<StringVector>()
.unwrap()
.get_data(0)
.unwrap();
Ok(format!("{create_table};\n"))
}
async fn export_create_table(&self) -> Result<()> {
let semaphore = Arc::new(Semaphore::new(self.parallelism));
let db_names = self.iter_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 semaphore_moved = semaphore.clone();
tasks.push(async move {
let _permit = semaphore_moved.acquire().await.unwrap();
let table_list = self.get_table_list(&catalog, &schema).await?;
let table_count = table_list.len();
tokio::fs::create_dir_all(&self.output_dir)
.await
.context(FileIoSnafu)?;
let output_file =
Path::new(&self.output_dir).join(format!("{catalog}-{schema}.sql"));
let mut file = File::create(output_file).await.context(FileIoSnafu)?;
for (c, s, t) in table_list {
match self.show_create_table(&c, &s, &t).await {
Err(e) => {
error!(e; "Failed to export table {}.{}.{}", c, s, t)
}
Ok(create_table) => {
file.write_all(create_table.as_bytes())
.await
.context(FileIoSnafu)?;
}
}
}
info!("finished exporting {catalog}.{schema} with {table_count} tables",);
Ok::<(), Error>(())
});
}
let success = futures::future::join_all(tasks)
.await
.into_iter()
.filter(|r| match r {
Ok(_) => true,
Err(e) => {
error!(e; "export job failed");
false
}
})
.count();
info!("success {success}/{db_count} jobs");
Ok(())
}
async fn export_table_data(&self) -> Result<()> {
let semaphore = Arc::new(Semaphore::new(self.parallelism));
let db_names = self.iter_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 semaphore_moved = semaphore.clone();
tasks.push(async move {
let _permit = semaphore_moved.acquire().await.unwrap();
tokio::fs::create_dir_all(&self.output_dir)
.await
.context(FileIoSnafu)?;
let output_dir = Path::new(&self.output_dir).join(format!("{catalog}-{schema}/"));
let mut client = self.client.clone();
client.set_catalog(catalog.clone());
client.set_schema(schema.clone());
// copy database to
let sql = format!(
"copy database {} to '{}' with (format='parquet');",
schema,
output_dir.to_str().unwrap()
);
client
.sql(sql.clone())
.await
.context(RequestDatabaseSnafu { sql })?;
info!("finished exporting {catalog}.{schema} data");
// export copy from sql
let dir_filenames = match output_dir.read_dir() {
Ok(dir) => dir,
Err(_) => {
warn!("empty database {catalog}.{schema}");
return Ok(());
}
};
let copy_from_file =
Path::new(&self.output_dir).join(format!("{catalog}-{schema}_copy_from.sql"));
let mut file = File::create(copy_from_file).await.context(FileIoSnafu)?;
let copy_from_sql = dir_filenames
.into_iter()
.map(|file| {
let file = file.unwrap();
let filename = file.file_name().into_string().unwrap();
format!(
"copy {} from '{}' with (format='parquet');\n",
filename.replace(".parquet", ""),
file.path().to_str().unwrap()
)
})
.collect::<Vec<_>>()
.join("");
file.write_all(copy_from_sql.as_bytes())
.await
.context(FileIoSnafu)?;
info!("finished exporting {catalog}.{schema} copy_from.sql");
Ok::<(), Error>(())
});
}
let success = futures::future::join_all(tasks)
.await
.into_iter()
.filter(|r| match r {
Ok(_) => true,
Err(e) => {
error!(e; "export job failed");
false
}
})
.count();
info!("success {success}/{db_count} jobs");
Ok(())
}
}
#[async_trait]
impl Tool for Export {
async fn do_work(&self) -> Result<()> {
match self.target {
ExportTarget::CreateTable => self.export_create_table().await,
ExportTarget::TableData => self.export_table_data().await,
}
}
}
/// Split at `-`.
fn split_database(database: &str) -> Result<(String, Option<String>)> {
let (catalog, schema) = database
.split_once('-')
.with_context(|| InvalidDatabaseNameSnafu {
database: database.to_string(),
})?;
if schema == "*" {
Ok((catalog.to_string(), None))
} else {
Ok((catalog.to_string(), Some(schema.to_string())))
}
}

View File

@@ -16,16 +16,20 @@ use std::path::PathBuf;
use std::sync::Arc;
use std::time::Instant;
use catalog::kvbackend::{CachedMetaKvBackend, KvBackendCatalogManager};
use catalog::remote::CachedMetaKvBackend;
use client::client_manager::DatanodeClients;
use client::{Client, Database, DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME};
use common_base::Plugins;
use common_error::ext::ErrorExt;
use common_meta::key::TableMetadataManager;
use common_query::Output;
use common_recordbatch::RecordBatches;
use common_telemetry::logging;
use either::Either;
use frontend::catalog::FrontendCatalogManager;
use meta_client::client::MetaClientBuilder;
use partition::manager::PartitionRuleManager;
use partition::route::TableRoutes;
use query::datafusion::DatafusionQueryEngine;
use query::logical_optimizer::LogicalOptimizer;
use query::parser::QueryLanguageParser;
@@ -35,7 +39,7 @@ use query::QueryEngine;
use rustyline::error::ReadlineError;
use rustyline::Editor;
use session::context::QueryContext;
use snafu::ResultExt;
use snafu::{ErrorCompat, ResultExt};
use substrait::{DFLogicalSubstraitConvertor, SubstraitPlan};
use crate::cli::cmd::ReplCommand;
@@ -148,7 +152,7 @@ impl Repl {
.await
.map_err(|e| {
let status_code = e.status_code();
let root_cause = e.output_msg();
let root_cause = e.iter_chain().last().unwrap();
println!("Error: {}({status_code}), {root_cause}", status_code as u32)
})
.is_ok()
@@ -176,7 +180,7 @@ impl Repl {
.encode(&plan)
.context(SubstraitEncodeLogicalPlanSnafu)?;
self.database.logical_plan(plan.to_vec(), 0).await
self.database.logical_plan(plan.to_vec(), None).await
} else {
self.database.sql(&sql).await
}
@@ -250,19 +254,24 @@ async fn create_query_engine(meta_addr: &str) -> Result<DatafusionQueryEngine> {
let cached_meta_backend = Arc::new(CachedMetaKvBackend::new(meta_client.clone()));
let table_routes = Arc::new(TableRoutes::new(meta_client));
let partition_manager = Arc::new(PartitionRuleManager::new(table_routes));
let datanode_clients = Arc::new(DatanodeClients::default());
let catalog_list = KvBackendCatalogManager::new(
let catalog_list = Arc::new(FrontendCatalogManager::new(
cached_meta_backend.clone(),
cached_meta_backend.clone(),
partition_manager,
datanode_clients,
);
let plugins: Plugins = Default::default();
Arc::new(TableMetadataManager::new(cached_meta_backend)),
));
let plugins: Arc<Plugins> = Default::default();
let state = Arc::new(QueryEngineState::new(
catalog_list,
None,
None,
false,
None,
None,
plugins.clone(),
));

View File

@@ -17,15 +17,15 @@ use std::sync::Arc;
use async_trait::async_trait;
use clap::Parser;
use client::api::v1::meta::TableRouteValue;
use common_meta::ddl::utils::region_storage_path;
use common_meta::error as MetaError;
use common_meta::helper::{CatalogKey as v1CatalogKey, SchemaKey as v1SchemaKey, TableGlobalValue};
use common_meta::key::catalog_name::{CatalogNameKey, CatalogNameValue};
use common_meta::key::datanode_table::{DatanodeTableKey, DatanodeTableValue, RegionInfo};
use common_meta::key::datanode_table::{DatanodeTableKey, DatanodeTableValue};
use common_meta::key::schema_name::{SchemaNameKey, SchemaNameValue};
use common_meta::key::table_info::{TableInfoKey, TableInfoValue};
use common_meta::key::table_name::{TableNameKey, TableNameValue};
use common_meta::key::table_region::{TableRegionKey, TableRegionValue};
use common_meta::key::table_route::{TableRouteKey, TableRouteValue as NextTableRouteValue};
use common_meta::key::table_route::{NextTableRouteKey, TableRouteValue as NextTableRouteValue};
use common_meta::key::{RegionDistribution, TableMetaKey};
use common_meta::range_stream::PaginationStream;
use common_meta::rpc::router::TableRoute;
@@ -39,7 +39,6 @@ use meta_srv::service::store::etcd::EtcdStore;
use meta_srv::service::store::kv::{KvBackendAdapter, KvStoreRef};
use prost::Message;
use snafu::ResultExt;
use v1_helper::{CatalogKey as v1CatalogKey, SchemaKey as v1SchemaKey, TableGlobalValue};
use crate::cli::{Instance, Tool};
use crate::error::{self, ConnectEtcdSnafu, Result};
@@ -155,7 +154,7 @@ impl MigrateTableMetadata {
let new_table_value = NextTableRouteValue::new(table_route.region_routes);
let table_id = table_route.table.id as u32;
let new_key = TableRouteKey::new(table_id);
let new_key = NextTableRouteKey::new(table_id);
info!("Creating '{new_key}'");
if self.dryrun {
@@ -387,11 +386,6 @@ impl MigrateTableMetadata {
async fn create_datanode_table_keys(&self, value: &TableGlobalValue) {
let table_id = value.table_id();
let engine = value.table_info.meta.engine.as_str();
let region_storage_path = region_storage_path(
&value.table_info.catalog_name,
&value.table_info.schema_name,
);
let region_distribution: RegionDistribution =
value.regions_id_map.clone().into_iter().collect();
@@ -400,18 +394,7 @@ impl MigrateTableMetadata {
.map(|(datanode_id, regions)| {
let k = DatanodeTableKey::new(datanode_id, table_id);
info!("Creating DatanodeTableKey '{k}' => {regions:?}");
(
k,
DatanodeTableValue::new(
table_id,
regions,
RegionInfo {
engine: engine.to_string(),
region_storage_path: region_storage_path.clone(),
region_options: (&value.table_info.meta.options).into(),
},
),
)
(k, DatanodeTableValue::new(table_id, regions))
})
.collect::<Vec<_>>();
@@ -426,148 +409,3 @@ impl MigrateTableMetadata {
}
}
}
#[deprecated(since = "0.4.0", note = "Used for migrate old version(v0.3) metadata")]
mod v1_helper {
use std::collections::HashMap;
use std::fmt::{Display, Formatter};
use err::{DeserializeCatalogEntryValueSnafu, Error, InvalidCatalogSnafu};
use lazy_static::lazy_static;
use regex::Regex;
use serde::{Deserialize, Serialize};
use snafu::{ensure, OptionExt, ResultExt};
use table::metadata::{RawTableInfo, TableId};
pub const CATALOG_KEY_PREFIX: &str = "__c";
pub const SCHEMA_KEY_PREFIX: &str = "__s";
/// The pattern of a valid catalog, schema or table name.
const NAME_PATTERN: &str = "[a-zA-Z_:][a-zA-Z0-9_:]*";
lazy_static! {
static ref CATALOG_KEY_PATTERN: Regex =
Regex::new(&format!("^{CATALOG_KEY_PREFIX}-({NAME_PATTERN})$")).unwrap();
}
lazy_static! {
static ref SCHEMA_KEY_PATTERN: Regex = Regex::new(&format!(
"^{SCHEMA_KEY_PREFIX}-({NAME_PATTERN})-({NAME_PATTERN})$"
))
.unwrap();
}
/// Table global info contains necessary info for a datanode to create table regions, including
/// table id, table meta(schema...), region id allocation across datanodes.
#[derive(Debug, Clone, Serialize, Deserialize, PartialEq, Eq)]
pub struct TableGlobalValue {
/// Id of datanode that created the global table info kv. only for debugging.
pub node_id: u64,
/// Allocation of region ids across all datanodes.
pub regions_id_map: HashMap<u64, Vec<u32>>,
pub table_info: RawTableInfo,
}
impl TableGlobalValue {
pub fn table_id(&self) -> TableId {
self.table_info.ident.table_id
}
}
pub struct CatalogKey {
pub catalog_name: String,
}
impl Display for CatalogKey {
fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
f.write_str(CATALOG_KEY_PREFIX)?;
f.write_str("-")?;
f.write_str(&self.catalog_name)
}
}
impl CatalogKey {
pub fn parse(s: impl AsRef<str>) -> Result<Self, Error> {
let key = s.as_ref();
let captures = CATALOG_KEY_PATTERN
.captures(key)
.context(InvalidCatalogSnafu { key })?;
ensure!(captures.len() == 2, InvalidCatalogSnafu { key });
Ok(Self {
catalog_name: captures[1].to_string(),
})
}
}
#[derive(Debug, Serialize, Deserialize)]
pub struct CatalogValue;
pub struct SchemaKey {
pub catalog_name: String,
pub schema_name: String,
}
impl Display for SchemaKey {
fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
f.write_str(SCHEMA_KEY_PREFIX)?;
f.write_str("-")?;
f.write_str(&self.catalog_name)?;
f.write_str("-")?;
f.write_str(&self.schema_name)
}
}
impl SchemaKey {
pub fn parse(s: impl AsRef<str>) -> Result<Self, Error> {
let key = s.as_ref();
let captures = SCHEMA_KEY_PATTERN
.captures(key)
.context(InvalidCatalogSnafu { key })?;
ensure!(captures.len() == 3, InvalidCatalogSnafu { key });
Ok(Self {
catalog_name: captures[1].to_string(),
schema_name: captures[2].to_string(),
})
}
}
#[derive(Debug, Serialize, Deserialize)]
pub struct SchemaValue;
macro_rules! define_catalog_value {
( $($val_ty: ty), *) => {
$(
impl $val_ty {
pub fn parse(s: impl AsRef<str>) -> Result<Self, Error> {
serde_json::from_str(s.as_ref())
.context(DeserializeCatalogEntryValueSnafu { raw: s.as_ref() })
}
pub fn from_bytes(bytes: impl AsRef<[u8]>) -> Result<Self, Error> {
Self::parse(&String::from_utf8_lossy(bytes.as_ref()))
}
}
)*
}
}
define_catalog_value!(TableGlobalValue);
mod err {
use snafu::{Location, Snafu};
#[derive(Debug, Snafu)]
#[snafu(visibility(pub))]
pub enum Error {
#[snafu(display("Invalid catalog info: {}", key))]
InvalidCatalog { key: String, location: Location },
#[snafu(display("Failed to deserialize catalog entry value: {}", raw))]
DeserializeCatalogEntryValue {
raw: String,
location: Location,
source: serde_json::error::Error,
},
}
}
}

View File

@@ -16,8 +16,7 @@ use std::time::Duration;
use clap::Parser;
use common_telemetry::logging;
use datanode::config::DatanodeOptions;
use datanode::datanode::{Datanode, DatanodeBuilder};
use datanode::datanode::{Datanode, DatanodeOptions};
use meta_client::MetaClientOptions;
use servers::Mode;
use snafu::ResultExt;
@@ -31,10 +30,6 @@ pub struct Instance {
impl Instance {
pub async fn start(&mut self) -> Result<()> {
plugins::start_datanode_plugins(self.datanode.plugins())
.await
.context(StartDatanodeSnafu)?;
self.datanode.start().await.context(StartDatanodeSnafu)
}
@@ -134,7 +129,7 @@ impl StartCommand {
}
if let Some(metasrv_addrs) = &self.metasrv_addr {
opts.meta_client
opts.meta_client_options
.get_or_insert_with(MetaClientOptions::default)
.metasrv_addrs = metasrv_addrs.clone();
opts.mode = Mode::Distributed;
@@ -156,29 +151,24 @@ impl StartCommand {
}
if let Some(http_addr) = &self.http_addr {
opts.http.addr = http_addr.clone();
opts.http_opts.addr = http_addr.clone();
}
if let Some(http_timeout) = self.http_timeout {
opts.http.timeout = Duration::from_secs(http_timeout)
opts.http_opts.timeout = Duration::from_secs(http_timeout)
}
// Disable dashboard in datanode.
opts.http.disable_dashboard = true;
opts.http_opts.disable_dashboard = true;
Ok(Options::Datanode(Box::new(opts)))
}
async fn build(self, mut opts: DatanodeOptions) -> Result<Instance> {
let plugins = plugins::setup_datanode_plugins(&mut opts)
.await
.context(StartDatanodeSnafu)?;
async fn build(self, opts: DatanodeOptions) -> Result<Instance> {
logging::info!("Datanode start command: {:#?}", self);
logging::info!("Datanode options: {:#?}", opts);
let datanode = DatanodeBuilder::new(opts, None, plugins)
.build()
let datanode = Datanode::new(opts, Default::default())
.await
.context(StartDatanodeSnafu)?;
@@ -193,8 +183,9 @@ mod tests {
use common_base::readable_size::ReadableSize;
use common_test_util::temp_dir::create_named_temp_file;
use datanode::config::{CompactionConfig, FileConfig, ObjectStoreConfig, RegionManifestConfig};
use servers::heartbeat_options::HeartbeatOptions;
use datanode::datanode::{
CompactionConfig, FileConfig, ObjectStoreConfig, RegionManifestConfig,
};
use servers::Mode;
use super::*;
@@ -211,14 +202,11 @@ mod tests {
rpc_hostname = "127.0.0.1"
rpc_runtime_size = 8
[heartbeat]
interval = "300ms"
[meta_client]
[meta_client_options]
metasrv_addrs = ["127.0.0.1:3002"]
timeout = "3s"
connect_timeout = "5s"
ddl_timeout = "10s"
timeout_millis = 3000
connect_timeout_millis = 5000
ddl_timeout_millis= 10000
tcp_nodelay = true
[wal]
@@ -261,33 +249,25 @@ mod tests {
assert_eq!("127.0.0.1:3001".to_string(), options.rpc_addr);
assert_eq!(Some(42), options.node_id);
assert_eq!("/other/wal", options.wal.dir.unwrap());
assert_eq!("/other/wal", options.wal.dir.unwrap());
assert_eq!(Duration::from_secs(600), options.wal.purge_interval);
assert_eq!(1024 * 1024 * 1024, options.wal.file_size.0);
assert_eq!(1024 * 1024 * 1024 * 50, options.wal.purge_threshold.0);
assert!(!options.wal.sync_write);
let HeartbeatOptions {
interval: heart_beat_interval,
..
} = options.heartbeat;
assert_eq!(300, heart_beat_interval.as_millis());
let MetaClientOptions {
metasrv_addrs: metasrv_addr,
timeout,
connect_timeout,
ddl_timeout,
timeout_millis,
connect_timeout_millis,
tcp_nodelay,
..
} = options.meta_client.unwrap();
ddl_timeout_millis,
} = options.meta_client_options.unwrap();
assert_eq!(vec!["127.0.0.1:3002".to_string()], metasrv_addr);
assert_eq!(5000, connect_timeout.as_millis());
assert_eq!(10000, ddl_timeout.as_millis());
assert_eq!(3000, timeout.as_millis());
assert_eq!(5000, connect_timeout_millis);
assert_eq!(10000, ddl_timeout_millis);
assert_eq!(3000, timeout_millis);
assert!(tcp_nodelay);
assert_eq!("/tmp/greptimedb/", options.storage.data_home);
assert!(matches!(
@@ -380,9 +360,9 @@ mod tests {
rpc_hostname = "127.0.0.1"
rpc_runtime_size = 8
[meta_client]
timeout = "3s"
connect_timeout = "5s"
[meta_client_options]
timeout_millis = 3000
connect_timeout_millis = 5000
tcp_nodelay = true
[wal]
@@ -433,10 +413,10 @@ mod tests {
Some("99"),
),
(
// meta_client.metasrv_addrs = 127.0.0.1:3001,127.0.0.1:3002,127.0.0.1:3003
// meta_client_options.metasrv_addrs = 127.0.0.1:3001,127.0.0.1:3002,127.0.0.1:3003
[
env_prefix.to_string(),
"meta_client".to_uppercase(),
"meta_client_options".to_uppercase(),
"metasrv_addrs".to_uppercase(),
]
.join(ENV_VAR_SEP),
@@ -463,7 +443,7 @@ mod tests {
Some(Duration::from_secs(9))
);
assert_eq!(
opts.meta_client.unwrap().metasrv_addrs,
opts.meta_client_options.unwrap().metasrv_addrs,
vec![
"127.0.0.1:3001".to_string(),
"127.0.0.1:3002".to_string(),

View File

@@ -16,76 +16,56 @@ use std::any::Any;
use common_error::ext::ErrorExt;
use common_error::status_code::StatusCode;
use common_macro::stack_trace_debug;
use config::ConfigError;
use rustyline::error::ReadlineError;
use snafu::{Location, Snafu};
#[derive(Snafu)]
#[derive(Debug, Snafu)]
#[snafu(visibility(pub))]
#[stack_trace_debug]
pub enum Error {
#[snafu(display("Failed to create default catalog and schema"))]
InitMetadata {
location: Location,
source: common_meta::error::Error,
},
#[snafu(display("Failed to iter stream"))]
#[snafu(display("Failed to iter stream, source: {}", source))]
IterStream {
location: Location,
source: common_meta::error::Error,
},
#[snafu(display("Failed to start procedure manager"))]
StartProcedureManager {
location: Location,
source: common_procedure::error::Error,
},
#[snafu(display("Failed to stop procedure manager"))]
StopProcedureManager {
location: Location,
source: common_procedure::error::Error,
},
#[snafu(display("Failed to start datanode"))]
#[snafu(display("Failed to start datanode, source: {}", source))]
StartDatanode {
location: Location,
source: datanode::error::Error,
},
#[snafu(display("Failed to shutdown datanode"))]
#[snafu(display("Failed to shutdown datanode, source: {}", source))]
ShutdownDatanode {
location: Location,
source: datanode::error::Error,
},
#[snafu(display("Failed to start frontend"))]
#[snafu(display("Failed to start frontend, source: {}", source))]
StartFrontend {
location: Location,
source: frontend::error::Error,
},
#[snafu(display("Failed to shutdown frontend"))]
#[snafu(display("Failed to shutdown frontend, source: {}", source))]
ShutdownFrontend {
location: Location,
source: frontend::error::Error,
},
#[snafu(display("Failed to build meta server"))]
#[snafu(display("Failed to build meta server, source: {}", source))]
BuildMetaServer {
location: Location,
source: meta_srv::error::Error,
},
#[snafu(display("Failed to start meta server"))]
#[snafu(display("Failed to start meta server, source: {}", source))]
StartMetaServer {
location: Location,
source: meta_srv::error::Error,
},
#[snafu(display("Failed to shutdown meta server"))]
#[snafu(display("Failed to shutdown meta server, source: {}", source))]
ShutdownMetaServer {
location: Location,
source: meta_srv::error::Error,
@@ -97,7 +77,13 @@ pub enum Error {
#[snafu(display("Illegal config: {}", msg))]
IllegalConfig { msg: String, location: Location },
#[snafu(display("Unsupported selector type: {}", selector_type))]
#[snafu(display("Illegal auth config: {}", source))]
IllegalAuthConfig {
location: Location,
source: auth::error::Error,
},
#[snafu(display("Unsupported selector type, {} source: {}", selector_type, source))]
UnsupportedSelectorType {
selector_type: String,
location: Location,
@@ -107,124 +93,80 @@ pub enum Error {
#[snafu(display("Invalid REPL command: {reason}"))]
InvalidReplCommand { reason: String },
#[snafu(display("Cannot create REPL"))]
#[snafu(display("Cannot create REPL: {}", source))]
ReplCreation {
#[snafu(source)]
error: ReadlineError,
source: ReadlineError,
location: Location,
},
#[snafu(display("Error reading command"))]
#[snafu(display("Error reading command: {}", source))]
Readline {
#[snafu(source)]
error: ReadlineError,
source: ReadlineError,
location: Location,
},
#[snafu(display("Failed to request database, sql: {sql}"))]
#[snafu(display("Failed to request database, sql: {sql}, source: {source}"))]
RequestDatabase {
sql: String,
location: Location,
source: client::Error,
},
#[snafu(display("Failed to collect RecordBatches"))]
#[snafu(display("Failed to collect RecordBatches, source: {source}"))]
CollectRecordBatches {
location: Location,
source: common_recordbatch::error::Error,
},
#[snafu(display("Failed to pretty print Recordbatches"))]
#[snafu(display("Failed to pretty print Recordbatches, source: {source}"))]
PrettyPrintRecordBatches {
location: Location,
source: common_recordbatch::error::Error,
},
#[snafu(display("Failed to start Meta client"))]
#[snafu(display("Failed to start Meta client, source: {}", source))]
StartMetaClient {
location: Location,
source: meta_client::error::Error,
},
#[snafu(display("Failed to parse SQL: {}", sql))]
#[snafu(display("Failed to parse SQL: {}, source: {}", sql, source))]
ParseSql {
sql: String,
location: Location,
source: query::error::Error,
},
#[snafu(display("Failed to plan statement"))]
#[snafu(display("Failed to plan statement, source: {}", source))]
PlanStatement {
location: Location,
source: query::error::Error,
},
#[snafu(display("Failed to encode logical plan in substrait"))]
#[snafu(display("Failed to encode logical plan in substrait, source: {}", source))]
SubstraitEncodeLogicalPlan {
location: Location,
source: substrait::error::Error,
},
#[snafu(display("Failed to load layered config"))]
#[snafu(display("Failed to load layered config, source: {}", source))]
LoadLayeredConfig {
#[snafu(source)]
error: ConfigError,
source: ConfigError,
location: Location,
},
#[snafu(display("Failed to start catalog manager"))]
#[snafu(display("Failed to start catalog manager, source: {}", source))]
StartCatalogManager {
location: Location,
source: catalog::error::Error,
},
#[snafu(display("Failed to connect to Etcd at {etcd_addr}"))]
#[snafu(display("Failed to connect to Etcd at {etcd_addr}, source: {}", source))]
ConnectEtcd {
etcd_addr: String,
#[snafu(source)]
error: etcd_client::Error,
source: etcd_client::Error,
location: Location,
},
#[snafu(display("Failed to connect server at {addr}"))]
ConnectServer {
addr: String,
source: client::error::Error,
location: Location,
},
#[snafu(display("Failed to serde json"))]
SerdeJson {
#[snafu(source)]
error: serde_json::error::Error,
location: Location,
},
#[snafu(display("Expect data from output, but got another thing"))]
NotDataFromOutput { location: Location },
#[snafu(display("Empty result from output"))]
EmptyResult { location: Location },
#[snafu(display("Failed to manipulate file"))]
FileIo {
location: Location,
#[snafu(source)]
error: std::io::Error,
},
#[snafu(display("Invalid database name: {}", database))]
InvalidDatabaseName {
location: Location,
database: String,
},
#[snafu(display("Failed to create directory {}", dir))]
CreateDir {
dir: String,
#[snafu(source)]
error: std::io::Error,
},
}
pub type Result<T> = std::result::Result<T, Error>;
@@ -240,21 +182,14 @@ impl ErrorExt for Error {
Error::ShutdownMetaServer { source, .. } => source.status_code(),
Error::BuildMetaServer { source, .. } => source.status_code(),
Error::UnsupportedSelectorType { source, .. } => source.status_code(),
Error::IterStream { source, .. } | Error::InitMetadata { source, .. } => {
source.status_code()
}
Error::ConnectServer { source, .. } => source.status_code(),
Error::IterStream { source, .. } => source.status_code(),
Error::MissingConfig { .. }
| Error::LoadLayeredConfig { .. }
| Error::IllegalConfig { .. }
| Error::InvalidReplCommand { .. }
| Error::ConnectEtcd { .. }
| Error::NotDataFromOutput { .. }
| Error::CreateDir { .. }
| Error::EmptyResult { .. }
| Error::InvalidDatabaseName { .. } => StatusCode::InvalidArguments,
Error::StartProcedureManager { source, .. }
| Error::StopProcedureManager { source, .. } => source.status_code(),
| Error::IllegalAuthConfig { .. }
| Error::ConnectEtcd { .. } => StatusCode::InvalidArguments,
Error::ReplCreation { .. } | Error::Readline { .. } => StatusCode::Internal,
Error::RequestDatabase { source, .. } => source.status_code(),
Error::CollectRecordBatches { source, .. }
@@ -265,8 +200,6 @@ impl ErrorExt for Error {
}
Error::SubstraitEncodeLogicalPlan { source, .. } => source.status_code(),
Error::StartCatalogManager { source, .. } => source.status_code(),
Error::SerdeJson { .. } | Error::FileIo { .. } => StatusCode::Unexpected,
}
}

View File

@@ -12,18 +12,21 @@
// See the License for the specific language governing permissions and
// limitations under the License.
use std::time::Duration;
use std::sync::Arc;
use auth::UserProviderRef;
use clap::Parser;
use common_base::Plugins;
use common_telemetry::logging;
use frontend::frontend::FrontendOptions;
use frontend::instance::{FrontendInstance, Instance as FeInstance};
use frontend::service_config::InfluxdbOptions;
use meta_client::MetaClientOptions;
use servers::tls::{TlsMode, TlsOption};
use servers::Mode;
use snafu::ResultExt;
use crate::error::{self, Result, StartFrontendSnafu};
use crate::error::{self, IllegalAuthConfigSnafu, Result, StartCatalogManagerSnafu};
use crate::options::{Options, TopLevelOptions};
pub struct Instance {
@@ -32,11 +35,16 @@ pub struct Instance {
impl Instance {
pub async fn start(&mut self) -> Result<()> {
plugins::start_frontend_plugins(self.frontend.plugins().clone())
self.frontend
.catalog_manager()
.start()
.await
.context(StartFrontendSnafu)?;
.context(StartCatalogManagerSnafu)?;
self.frontend.start().await.context(StartFrontendSnafu)
self.frontend
.start()
.await
.context(error::StartFrontendSnafu)
}
pub async fn stop(&self) -> Result<()> {
@@ -87,9 +95,7 @@ pub struct StartCommand {
#[clap(long)]
http_addr: Option<String>,
#[clap(long)]
http_timeout: Option<u64>,
#[clap(long)]
rpc_addr: Option<String>,
grpc_addr: Option<String>,
#[clap(long)]
mysql_addr: Option<String>,
#[clap(long)]
@@ -139,75 +145,86 @@ impl StartCommand {
);
if let Some(addr) = &self.http_addr {
opts.http.addr = addr.clone()
}
if let Some(http_timeout) = self.http_timeout {
opts.http.timeout = Duration::from_secs(http_timeout)
if let Some(http_opts) = &mut opts.http_options {
http_opts.addr = addr.clone()
}
}
if let Some(disable_dashboard) = self.disable_dashboard {
opts.http.disable_dashboard = disable_dashboard;
opts.http_options
.get_or_insert_with(Default::default)
.disable_dashboard = disable_dashboard;
}
if let Some(addr) = &self.rpc_addr {
opts.grpc.addr = addr.clone()
if let Some(addr) = &self.grpc_addr {
if let Some(grpc_opts) = &mut opts.grpc_options {
grpc_opts.addr = addr.clone()
}
}
if let Some(addr) = &self.mysql_addr {
opts.mysql.enable = true;
opts.mysql.addr = addr.clone();
opts.mysql.tls = tls_opts.clone();
if let Some(mysql_opts) = &mut opts.mysql_options {
mysql_opts.addr = addr.clone();
mysql_opts.tls = tls_opts.clone();
}
}
if let Some(addr) = &self.postgres_addr {
opts.postgres.enable = true;
opts.postgres.addr = addr.clone();
opts.postgres.tls = tls_opts;
if let Some(postgres_opts) = &mut opts.postgres_options {
postgres_opts.addr = addr.clone();
postgres_opts.tls = tls_opts;
}
}
if let Some(addr) = &self.opentsdb_addr {
opts.opentsdb.enable = true;
opts.opentsdb.addr = addr.clone();
if let Some(opentsdb_addr) = &mut opts.opentsdb_options {
opentsdb_addr.addr = addr.clone();
}
}
if let Some(enable) = self.influxdb_enable {
opts.influxdb.enable = enable;
opts.influxdb_options = Some(InfluxdbOptions { enable });
}
if let Some(metasrv_addrs) = &self.metasrv_addr {
opts.meta_client
opts.meta_client_options
.get_or_insert_with(MetaClientOptions::default)
.metasrv_addrs = metasrv_addrs.clone();
opts.mode = Mode::Distributed;
}
opts.user_provider = self.user_provider.clone();
Ok(Options::Frontend(Box::new(opts)))
}
async fn build(self, mut opts: FrontendOptions) -> Result<Instance> {
let plugins = plugins::setup_frontend_plugins(&mut opts)
.await
.context(StartFrontendSnafu)?;
async fn build(self, opts: FrontendOptions) -> Result<Instance> {
logging::info!("Frontend start command: {:#?}", self);
logging::info!("Frontend options: {:#?}", opts);
let plugins = Arc::new(load_frontend_plugins(&self.user_provider)?);
let mut instance = FeInstance::try_new_distributed(&opts, plugins.clone())
.await
.context(StartFrontendSnafu)?;
.context(error::StartFrontendSnafu)?;
instance
.build_servers(&opts)
.await
.context(StartFrontendSnafu)?;
.context(error::StartFrontendSnafu)?;
Ok(Instance { frontend: instance })
}
}
pub fn load_frontend_plugins(user_provider: &Option<String>) -> Result<Plugins> {
let plugins = Plugins::new();
if let Some(provider) = user_provider {
let provider = auth::user_provider_from_option(provider).context(IllegalAuthConfigSnafu)?;
plugins.insert::<UserProviderRef>(provider);
}
Ok(plugins)
}
#[cfg(test)]
mod tests {
use std::io::Write;
@@ -217,7 +234,6 @@ mod tests {
use common_base::readable_size::ReadableSize;
use common_test_util::temp_dir::create_named_temp_file;
use frontend::service_config::GrpcOptions;
use servers::http::HttpOptions;
use super::*;
use crate::options::ENV_VAR_SEP;
@@ -239,29 +255,40 @@ mod tests {
unreachable!()
};
assert_eq!(opts.http.addr, "127.0.0.1:1234");
assert_eq!(ReadableSize::mb(64), opts.http.body_limit);
assert_eq!(opts.mysql.addr, "127.0.0.1:5678");
assert_eq!(opts.postgres.addr, "127.0.0.1:5432");
assert_eq!(opts.opentsdb.addr, "127.0.0.1:4321");
assert_eq!(opts.http_options.as_ref().unwrap().addr, "127.0.0.1:1234");
assert_eq!(
ReadableSize::mb(64),
opts.http_options.as_ref().unwrap().body_limit
);
assert_eq!(opts.mysql_options.as_ref().unwrap().addr, "127.0.0.1:5678");
assert_eq!(
opts.postgres_options.as_ref().unwrap().addr,
"127.0.0.1:5432"
);
assert_eq!(
opts.opentsdb_options.as_ref().unwrap().addr,
"127.0.0.1:4321"
);
let default_opts = FrontendOptions::default();
assert_eq!(opts.grpc.addr, default_opts.grpc.addr);
assert!(opts.mysql.enable);
assert_eq!(opts.mysql.runtime_size, default_opts.mysql.runtime_size);
assert!(opts.postgres.enable);
assert_eq!(
opts.postgres.runtime_size,
default_opts.postgres.runtime_size
opts.grpc_options.unwrap().addr,
default_opts.grpc_options.unwrap().addr
);
assert!(opts.opentsdb.enable);
assert_eq!(
opts.opentsdb.runtime_size,
default_opts.opentsdb.runtime_size
opts.mysql_options.as_ref().unwrap().runtime_size,
default_opts.mysql_options.as_ref().unwrap().runtime_size
);
assert_eq!(
opts.postgres_options.as_ref().unwrap().runtime_size,
default_opts.postgres_options.as_ref().unwrap().runtime_size
);
assert_eq!(
opts.opentsdb_options.as_ref().unwrap().runtime_size,
default_opts.opentsdb_options.as_ref().unwrap().runtime_size
);
assert!(!opts.influxdb.enable);
assert!(!opts.influxdb_options.unwrap().enable);
}
#[test]
@@ -270,7 +297,7 @@ mod tests {
let toml_str = r#"
mode = "distributed"
[http]
[http_options]
addr = "127.0.0.1:4000"
timeout = "30s"
body_limit = "2GB"
@@ -292,10 +319,19 @@ mod tests {
unreachable!()
};
assert_eq!(Mode::Distributed, fe_opts.mode);
assert_eq!("127.0.0.1:4000".to_string(), fe_opts.http.addr);
assert_eq!(Duration::from_secs(30), fe_opts.http.timeout);
assert_eq!(
"127.0.0.1:4000".to_string(),
fe_opts.http_options.as_ref().unwrap().addr
);
assert_eq!(
Duration::from_secs(30),
fe_opts.http_options.as_ref().unwrap().timeout
);
assert_eq!(ReadableSize::gb(2), fe_opts.http.body_limit);
assert_eq!(
ReadableSize::gb(2),
fe_opts.http_options.as_ref().unwrap().body_limit
);
assert_eq!("debug", fe_opts.logging.level.as_ref().unwrap());
assert_eq!("/tmp/greptimedb/test/logs".to_string(), fe_opts.logging.dir);
@@ -303,17 +339,14 @@ mod tests {
#[tokio::test]
async fn test_try_from_start_command_to_anymap() {
let mut fe_opts = FrontendOptions {
http: HttpOptions {
disable_dashboard: false,
..Default::default()
},
let command = StartCommand {
user_provider: Some("static_user_provider:cmd:test=test".to_string()),
disable_dashboard: Some(false),
..Default::default()
};
let plugins = plugins::setup_frontend_plugins(&mut fe_opts).await.unwrap();
let plugins = load_frontend_plugins(&command.user_provider);
let plugins = plugins.unwrap();
let provider = plugins.get::<UserProviderRef>().unwrap();
let result = provider
.authenticate(
@@ -349,15 +382,15 @@ mod tests {
let toml_str = r#"
mode = "distributed"
[http]
[http_options]
addr = "127.0.0.1:4000"
[meta_client]
timeout = "3s"
connect_timeout = "5s"
[meta_client_options]
timeout_millis = 3000
connect_timeout_millis = 5000
tcp_nodelay = true
[mysql]
[mysql_options]
addr = "127.0.0.1:4002"
"#;
write!(file, "{}", toml_str).unwrap();
@@ -366,40 +399,40 @@ mod tests {
temp_env::with_vars(
[
(
// mysql.addr = 127.0.0.1:14002
// mysql_options.addr = 127.0.0.1:14002
[
env_prefix.to_string(),
"mysql".to_uppercase(),
"mysql_options".to_uppercase(),
"addr".to_uppercase(),
]
.join(ENV_VAR_SEP),
Some("127.0.0.1:14002"),
),
(
// mysql.runtime_size = 11
// mysql_options.runtime_size = 11
[
env_prefix.to_string(),
"mysql".to_uppercase(),
"mysql_options".to_uppercase(),
"runtime_size".to_uppercase(),
]
.join(ENV_VAR_SEP),
Some("11"),
),
(
// http.addr = 127.0.0.1:24000
// http_options.addr = 127.0.0.1:24000
[
env_prefix.to_string(),
"http".to_uppercase(),
"http_options".to_uppercase(),
"addr".to_uppercase(),
]
.join(ENV_VAR_SEP),
Some("127.0.0.1:24000"),
),
(
// meta_client.metasrv_addrs = 127.0.0.1:3001,127.0.0.1:3002,127.0.0.1:3003
// meta_client_options.metasrv_addrs = 127.0.0.1:3001,127.0.0.1:3002,127.0.0.1:3003
[
env_prefix.to_string(),
"meta_client".to_uppercase(),
"meta_client_options".to_uppercase(),
"metasrv_addrs".to_uppercase(),
]
.join(ENV_VAR_SEP),
@@ -424,9 +457,9 @@ mod tests {
};
// Should be read from env, env > default values.
assert_eq!(fe_opts.mysql.runtime_size, 11);
assert_eq!(fe_opts.mysql_options.as_ref().unwrap().runtime_size, 11);
assert_eq!(
fe_opts.meta_client.unwrap().metasrv_addrs,
fe_opts.meta_client_options.unwrap().metasrv_addrs,
vec![
"127.0.0.1:3001".to_string(),
"127.0.0.1:3002".to_string(),
@@ -435,13 +468,22 @@ mod tests {
);
// Should be read from config file, config file > env > default values.
assert_eq!(fe_opts.mysql.addr, "127.0.0.1:4002");
assert_eq!(
fe_opts.mysql_options.as_ref().unwrap().addr,
"127.0.0.1:4002"
);
// Should be read from cli, cli > config file > env > default values.
assert_eq!(fe_opts.http.addr, "127.0.0.1:14000");
assert_eq!(
fe_opts.http_options.as_ref().unwrap().addr,
"127.0.0.1:14000"
);
// Should be default value.
assert_eq!(fe_opts.grpc.addr, GrpcOptions::default().addr);
assert_eq!(
fe_opts.grpc_options.as_ref().unwrap().addr,
GrpcOptions::default().addr
);
},
);
}

View File

@@ -20,7 +20,7 @@ use meta_srv::bootstrap::MetaSrvInstance;
use meta_srv::metasrv::MetaSrvOptions;
use snafu::ResultExt;
use crate::error::{self, Result, StartMetaServerSnafu};
use crate::error::{self, Result};
use crate::options::{Options, TopLevelOptions};
pub struct Instance {
@@ -29,10 +29,10 @@ pub struct Instance {
impl Instance {
pub async fn start(&mut self) -> Result<()> {
plugins::start_meta_srv_plugins(self.instance.plugins())
self.instance
.start()
.await
.context(StartMetaServerSnafu)?;
self.instance.start().await.context(StartMetaServerSnafu)
.context(error::StartMetaServerSnafu)
}
pub async fn stop(&self) -> Result<()> {
@@ -93,7 +93,7 @@ struct StartCommand {
#[clap(long)]
use_memory_store: Option<bool>,
#[clap(long)]
enable_region_failover: Option<bool>,
disable_region_failover: Option<bool>,
#[clap(long)]
http_addr: Option<String>,
#[clap(long)]
@@ -140,33 +140,30 @@ impl StartCommand {
opts.use_memory_store = use_memory_store;
}
if let Some(enable_region_failover) = self.enable_region_failover {
opts.enable_region_failover = enable_region_failover;
if let Some(disable_region_failover) = self.disable_region_failover {
opts.disable_region_failover = disable_region_failover;
}
if let Some(http_addr) = &self.http_addr {
opts.http.addr = http_addr.clone();
opts.http_opts.addr = http_addr.clone();
}
if let Some(http_timeout) = self.http_timeout {
opts.http.timeout = Duration::from_secs(http_timeout);
opts.http_opts.timeout = Duration::from_secs(http_timeout);
}
// Disable dashboard in metasrv.
opts.http.disable_dashboard = true;
opts.http_opts.disable_dashboard = true;
Ok(Options::Metasrv(Box::new(opts)))
}
async fn build(self, mut opts: MetaSrvOptions) -> Result<Instance> {
let plugins = plugins::setup_meta_srv_plugins(&mut opts)
.await
.context(StartMetaServerSnafu)?;
async fn build(self, opts: MetaSrvOptions) -> Result<Instance> {
logging::info!("MetaSrv start command: {:#?}", self);
logging::info!("MetaSrv options: {:#?}", opts);
let instance = MetaSrvInstance::new(opts, plugins)
let instance = MetaSrvInstance::new(opts)
.await
.context(error::BuildMetaServerSnafu)?;
@@ -210,6 +207,7 @@ mod tests {
bind_addr = "127.0.0.1:3002"
server_addr = "127.0.0.1:3002"
store_addr = "127.0.0.1:2379"
datanode_lease_secs = 15
selector = "LeaseBased"
use_memory_store = false
@@ -231,6 +229,7 @@ mod tests {
assert_eq!("127.0.0.1:3002".to_string(), options.bind_addr);
assert_eq!("127.0.0.1:3002".to_string(), options.server_addr);
assert_eq!("127.0.0.1:2379".to_string(), options.store_addr);
assert_eq!(15, options.datanode_lease_secs);
assert_eq!(SelectorType::LeaseBased, options.selector);
assert_eq!("debug", options.logging.level.as_ref().unwrap());
assert_eq!("/tmp/greptimedb/test/logs".to_string(), options.logging.dir);
@@ -267,7 +266,7 @@ mod tests {
selector = "LeaseBased"
use_memory_store = false
[http]
[http_options]
addr = "127.0.0.1:4000"
[logging]
@@ -290,10 +289,10 @@ mod tests {
Some("127.0.0.1:13002"),
),
(
// http.addr = 127.0.0.1:24000
// http_options.addr = 127.0.0.1:24000
[
env_prefix.to_string(),
"http".to_uppercase(),
"http_options".to_uppercase(),
"addr".to_uppercase(),
]
.join(ENV_VAR_SEP),
@@ -321,7 +320,7 @@ mod tests {
assert_eq!(opts.server_addr, "127.0.0.1:3002");
// Should be read from cli, cli > config file > env > default values.
assert_eq!(opts.http.addr, "127.0.0.1:14000");
assert_eq!(opts.http_opts.addr, "127.0.0.1:14000");
// Should be default value.
assert_eq!(opts.store_addr, "127.0.0.1:2379");

View File

@@ -12,27 +12,22 @@
// See the License for the specific language governing permissions and
// limitations under the License.
use common_config::KvStoreConfig;
use common_telemetry::logging::LoggingOptions;
use config::{Config, Environment, File, FileFormat};
use datanode::config::{DatanodeOptions, ProcedureConfig};
use datanode::datanode::DatanodeOptions;
use frontend::frontend::FrontendOptions;
use meta_srv::metasrv::MetaSrvOptions;
use serde::{Deserialize, Serialize};
use snafu::ResultExt;
use crate::error::{LoadLayeredConfigSnafu, Result, SerdeJsonSnafu};
use crate::error::{LoadLayeredConfigSnafu, Result};
pub const ENV_VAR_SEP: &str = "__";
pub const ENV_LIST_SEP: &str = ",";
/// Options mixed up from datanode, frontend and metasrv.
pub struct MixOptions {
pub data_home: String,
pub procedure: ProcedureConfig,
pub metadata_store: KvStoreConfig,
pub frontend: FrontendOptions,
pub datanode: DatanodeOptions,
pub fe_opts: FrontendOptions,
pub dn_opts: DatanodeOptions,
pub logging: LoggingOptions,
}
@@ -94,16 +89,9 @@ impl Options {
.ignore_empty(true)
};
// Workaround: Replacement for `Config::try_from(&default_opts)` due to
// `ConfigSerializer` cannot handle the case of an empty struct contained
// within an iterative structure.
// See: https://github.com/mehcode/config-rs/issues/461
let json_str = serde_json::to_string(&default_opts).context(SerdeJsonSnafu)?;
let default_config = File::from_str(&json_str, FileFormat::Json);
// Add default values and environment variables as the sources of the configuration.
let mut layered_config = Config::builder()
.add_source(default_config)
.add_source(Config::try_from(&default_opts).context(LoadLayeredConfigSnafu)?)
.add_source(env_source);
// Add config file as the source of the configuration if it is specified.
@@ -127,7 +115,7 @@ mod tests {
use std::time::Duration;
use common_test_util::temp_dir::create_named_temp_file;
use datanode::config::{DatanodeOptions, ObjectStoreConfig};
use datanode::datanode::{DatanodeOptions, ObjectStoreConfig};
use super::*;
@@ -143,9 +131,9 @@ mod tests {
mysql_addr = "127.0.0.1:4406"
mysql_runtime_size = 2
[meta_client]
timeout = "3s"
connect_timeout = "5s"
[meta_client_options]
timeout_millis = 3000
connect_timeout_millis = 5000
tcp_nodelay = true
[wal]
@@ -224,10 +212,10 @@ mod tests {
Some("/other/wal/dir"),
),
(
// meta_client.metasrv_addrs = 127.0.0.1:3001,127.0.0.1:3002,127.0.0.1:3003
// meta_client_options.metasrv_addrs = 127.0.0.1:3001,127.0.0.1:3002,127.0.0.1:3003
[
env_prefix.to_string(),
"meta_client".to_uppercase(),
"meta_client_options".to_uppercase(),
"metasrv_addrs".to_uppercase(),
]
.join(ENV_VAR_SEP),
@@ -255,7 +243,7 @@ mod tests {
Some(Duration::from_secs(42))
);
assert_eq!(
opts.meta_client.unwrap().metasrv_addrs,
opts.meta_client_options.unwrap().metasrv_addrs,
vec![
"127.0.0.1:3001".to_string(),
"127.0.0.1:3002".to_string(),

View File

@@ -13,28 +13,18 @@
// limitations under the License.
use std::sync::Arc;
use std::{fs, path};
use catalog::kvbackend::KvBackendCatalogManager;
use catalog::CatalogManagerRef;
use clap::Parser;
use common_base::Plugins;
use common_config::{metadata_store_dir, KvStoreConfig, WalConfig};
use common_meta::cache_invalidator::DummyKvCacheInvalidator;
use common_meta::kv_backend::KvBackendRef;
use common_procedure::ProcedureManagerRef;
use common_telemetry::info;
use common_telemetry::logging::LoggingOptions;
use datanode::config::{DatanodeOptions, ProcedureConfig, RegionEngineConfig, StorageConfig};
use datanode::datanode::{Datanode, DatanodeBuilder};
use datanode::region_server::RegionServer;
use file_engine::config::EngineConfig as FileEngineConfig;
use datanode::datanode::{Datanode, DatanodeOptions, ProcedureConfig, StorageConfig, WalConfig};
use datanode::instance::InstanceRef;
use frontend::frontend::FrontendOptions;
use frontend::instance::{FrontendInstance, Instance as FeInstance, StandaloneDatanodeManager};
use frontend::instance::{FrontendInstance, Instance as FeInstance};
use frontend::service_config::{
GrpcOptions, InfluxdbOptions, MysqlOptions, OpentsdbOptions, PostgresOptions, PromStoreOptions,
};
use mito2::config::MitoConfig;
use serde::{Deserialize, Serialize};
use servers::http::HttpOptions;
use servers::tls::{TlsMode, TlsOption};
@@ -42,10 +32,10 @@ use servers::Mode;
use snafu::ResultExt;
use crate::error::{
CreateDirSnafu, IllegalConfigSnafu, InitMetadataSnafu, Result, ShutdownDatanodeSnafu,
ShutdownFrontendSnafu, StartDatanodeSnafu, StartFrontendSnafu, StartProcedureManagerSnafu,
StopProcedureManagerSnafu,
IllegalConfigSnafu, Result, ShutdownDatanodeSnafu, ShutdownFrontendSnafu, StartDatanodeSnafu,
StartFrontendSnafu,
};
use crate::frontend::load_frontend_plugins;
use crate::options::{MixOptions, Options, TopLevelOptions};
#[derive(Parser)]
@@ -55,8 +45,12 @@ pub struct Command {
}
impl Command {
pub async fn build(self, opts: MixOptions) -> Result<Instance> {
self.subcmd.build(opts).await
pub async fn build(
self,
fe_opts: FrontendOptions,
dn_opts: DatanodeOptions,
) -> Result<Instance> {
self.subcmd.build(fe_opts, dn_opts).await
}
pub fn load_options(&self, top_level_options: TopLevelOptions) -> Result<Options> {
@@ -70,9 +64,9 @@ enum SubCommand {
}
impl SubCommand {
async fn build(self, opts: MixOptions) -> Result<Instance> {
async fn build(self, fe_opts: FrontendOptions, dn_opts: DatanodeOptions) -> Result<Instance> {
match self {
SubCommand::Start(cmd) => cmd.build(opts).await,
SubCommand::Start(cmd) => cmd.build(fe_opts, dn_opts).await,
}
}
@@ -87,46 +81,38 @@ impl SubCommand {
#[serde(default)]
pub struct StandaloneOptions {
pub mode: Mode,
pub enable_memory_catalog: bool,
pub enable_telemetry: bool,
pub http: HttpOptions,
pub grpc: GrpcOptions,
pub mysql: MysqlOptions,
pub postgres: PostgresOptions,
pub opentsdb: OpentsdbOptions,
pub influxdb: InfluxdbOptions,
pub prom_store: PromStoreOptions,
pub http_options: Option<HttpOptions>,
pub grpc_options: Option<GrpcOptions>,
pub mysql_options: Option<MysqlOptions>,
pub postgres_options: Option<PostgresOptions>,
pub opentsdb_options: Option<OpentsdbOptions>,
pub influxdb_options: Option<InfluxdbOptions>,
pub prom_store_options: Option<PromStoreOptions>,
pub wal: WalConfig,
pub storage: StorageConfig,
pub metadata_store: KvStoreConfig,
pub procedure: ProcedureConfig,
pub logging: LoggingOptions,
pub user_provider: Option<String>,
/// Options for different store engines.
pub region_engine: Vec<RegionEngineConfig>,
}
impl Default for StandaloneOptions {
fn default() -> Self {
Self {
mode: Mode::Standalone,
enable_memory_catalog: false,
enable_telemetry: true,
http: HttpOptions::default(),
grpc: GrpcOptions::default(),
mysql: MysqlOptions::default(),
postgres: PostgresOptions::default(),
opentsdb: OpentsdbOptions::default(),
influxdb: InfluxdbOptions::default(),
prom_store: PromStoreOptions::default(),
http_options: Some(HttpOptions::default()),
grpc_options: Some(GrpcOptions::default()),
mysql_options: Some(MysqlOptions::default()),
postgres_options: Some(PostgresOptions::default()),
opentsdb_options: Some(OpentsdbOptions::default()),
influxdb_options: Some(InfluxdbOptions::default()),
prom_store_options: Some(PromStoreOptions::default()),
wal: WalConfig::default(),
storage: StorageConfig::default(),
metadata_store: KvStoreConfig::default(),
procedure: ProcedureConfig::default(),
logging: LoggingOptions::default(),
user_provider: None,
region_engine: vec![
RegionEngineConfig::Mito(MitoConfig::default()),
RegionEngineConfig::File(FileEngineConfig::default()),
],
}
}
}
@@ -135,27 +121,26 @@ impl StandaloneOptions {
fn frontend_options(self) -> FrontendOptions {
FrontendOptions {
mode: self.mode,
http: self.http,
grpc: self.grpc,
mysql: self.mysql,
postgres: self.postgres,
opentsdb: self.opentsdb,
influxdb: self.influxdb,
prom_store: self.prom_store,
meta_client: None,
http_options: self.http_options,
grpc_options: self.grpc_options,
mysql_options: self.mysql_options,
postgres_options: self.postgres_options,
opentsdb_options: self.opentsdb_options,
influxdb_options: self.influxdb_options,
prom_store_options: self.prom_store_options,
meta_client_options: None,
logging: self.logging,
user_provider: self.user_provider,
..Default::default()
}
}
fn datanode_options(self) -> DatanodeOptions {
DatanodeOptions {
node_id: Some(0),
enable_memory_catalog: self.enable_memory_catalog,
enable_telemetry: self.enable_telemetry,
wal: self.wal,
storage: self.storage,
region_engine: self.region_engine,
procedure: self.procedure,
..Default::default()
}
}
@@ -164,19 +149,16 @@ impl StandaloneOptions {
pub struct Instance {
datanode: Datanode,
frontend: FeInstance,
procedure_manager: ProcedureManagerRef,
}
impl Instance {
pub async fn start(&mut self) -> Result<()> {
// Start datanode instance before starting services, to avoid requests come in before internal components are started.
self.datanode.start().await.context(StartDatanodeSnafu)?;
info!("Datanode instance started");
self.procedure_manager
.start()
self.datanode
.start_instance()
.await
.context(StartProcedureManagerSnafu)?;
.context(StartDatanodeSnafu)?;
info!("Datanode instance started");
self.frontend.start().await.context(StartFrontendSnafu)?;
Ok(())
@@ -188,13 +170,8 @@ impl Instance {
.await
.context(ShutdownFrontendSnafu)?;
self.procedure_manager
.stop()
.await
.context(StopProcedureManagerSnafu)?;
self.datanode
.shutdown()
.shutdown_instance()
.await
.context(ShutdownDatanodeSnafu)?;
info!("Datanode instance stopped.");
@@ -219,6 +196,8 @@ struct StartCommand {
influxdb_enable: bool,
#[clap(short, long)]
config_file: Option<String>,
#[clap(short = 'm', long = "memory-catalog")]
enable_memory_catalog: bool,
#[clap(long)]
tls_mode: Option<TlsMode>,
#[clap(long)]
@@ -239,6 +218,8 @@ impl StartCommand {
None,
)?;
opts.enable_memory_catalog = self.enable_memory_catalog;
opts.mode = Mode::Standalone;
if let Some(dir) = top_level_options.log_dir {
@@ -256,7 +237,9 @@ impl StartCommand {
);
if let Some(addr) = &self.http_addr {
opts.http.addr = addr.clone()
if let Some(http_opts) = &mut opts.http_options {
http_opts.addr = addr.clone()
}
}
if let Some(addr) = &self.rpc_addr {
@@ -270,58 +253,48 @@ impl StartCommand {
}
.fail();
}
opts.grpc.addr = addr.clone()
if let Some(grpc_opts) = &mut opts.grpc_options {
grpc_opts.addr = addr.clone()
}
}
if let Some(addr) = &self.mysql_addr {
opts.mysql.enable = true;
opts.mysql.addr = addr.clone();
opts.mysql.tls = tls_opts.clone();
if let Some(mysql_opts) = &mut opts.mysql_options {
mysql_opts.addr = addr.clone();
mysql_opts.tls = tls_opts.clone();
}
}
if let Some(addr) = &self.postgres_addr {
opts.postgres.enable = true;
opts.postgres.addr = addr.clone();
opts.postgres.tls = tls_opts;
if let Some(postgres_opts) = &mut opts.postgres_options {
postgres_opts.addr = addr.clone();
postgres_opts.tls = tls_opts;
}
}
if let Some(addr) = &self.opentsdb_addr {
opts.opentsdb.enable = true;
opts.opentsdb.addr = addr.clone();
if let Some(opentsdb_addr) = &mut opts.opentsdb_options {
opentsdb_addr.addr = addr.clone();
}
}
if self.influxdb_enable {
opts.influxdb.enable = self.influxdb_enable;
opts.influxdb_options = Some(InfluxdbOptions { enable: true });
}
opts.user_provider = self.user_provider.clone();
let metadata_store = opts.metadata_store.clone();
let procedure = opts.procedure.clone();
let frontend = opts.clone().frontend_options();
let fe_opts = opts.clone().frontend_options();
let logging = opts.logging.clone();
let datanode = opts.datanode_options();
let dn_opts = opts.datanode_options();
Ok(Options::Standalone(Box::new(MixOptions {
procedure,
metadata_store,
data_home: datanode.storage.data_home.to_string(),
frontend,
datanode,
fe_opts,
dn_opts,
logging,
})))
}
#[allow(unreachable_code)]
#[allow(unused_variables)]
#[allow(clippy::diverging_sub_expression)]
async fn build(self, opts: MixOptions) -> Result<Instance> {
let mut fe_opts = opts.frontend;
let fe_plugins = plugins::setup_frontend_plugins(&mut fe_opts)
.await
.context(StartFrontendSnafu)?;
let dn_opts = opts.datanode;
async fn build(self, fe_opts: FrontendOptions, dn_opts: DatanodeOptions) -> Result<Instance> {
let plugins = Arc::new(load_frontend_plugins(&self.user_provider)?);
info!("Standalone start command: {:#?}", self);
info!(
@@ -329,79 +302,30 @@ impl StartCommand {
fe_opts, dn_opts
);
// Ensure the data_home directory exists.
fs::create_dir_all(path::Path::new(&opts.data_home)).context(CreateDirSnafu {
dir: &opts.data_home,
})?;
let metadata_dir = metadata_store_dir(&opts.data_home);
let (kv_store, procedure_manager) = FeInstance::try_build_standalone_components(
metadata_dir,
opts.metadata_store,
opts.procedure,
)
.await
.context(StartFrontendSnafu)?;
let datanode =
DatanodeBuilder::new(dn_opts.clone(), Some(kv_store.clone()), Default::default())
.build()
.await
.context(StartDatanodeSnafu)?;
let region_server = datanode.region_server();
let catalog_manager = KvBackendCatalogManager::new(
kv_store.clone(),
Arc::new(DummyKvCacheInvalidator),
Arc::new(StandaloneDatanodeManager(region_server.clone())),
);
catalog_manager
.table_metadata_manager_ref()
.init()
let datanode = Datanode::new(dn_opts.clone(), Default::default())
.await
.context(InitMetadataSnafu)?;
.context(StartDatanodeSnafu)?;
// TODO: build frontend instance like in distributed mode
let mut frontend = build_frontend(
fe_plugins,
kv_store,
procedure_manager.clone(),
catalog_manager,
region_server,
)
.await?;
let mut frontend = build_frontend(plugins.clone(), datanode.get_instance()).await?;
frontend
.build_servers(&fe_opts)
.await
.context(StartFrontendSnafu)?;
Ok(Instance {
datanode,
frontend,
procedure_manager,
})
Ok(Instance { datanode, frontend })
}
}
/// Build frontend instance in standalone mode
async fn build_frontend(
plugins: Plugins,
kv_store: KvBackendRef,
procedure_manager: ProcedureManagerRef,
catalog_manager: CatalogManagerRef,
region_server: RegionServer,
plugins: Arc<Plugins>,
datanode_instance: InstanceRef,
) -> Result<FeInstance> {
let frontend_instance = FeInstance::try_new_standalone(
kv_store,
procedure_manager,
catalog_manager,
plugins,
region_server,
)
.await
.context(StartFrontendSnafu)?;
let mut frontend_instance = FeInstance::try_new_standalone(datanode_instance.clone())
.await
.context(StartFrontendSnafu)?;
frontend_instance.set_plugins(plugins.clone());
Ok(frontend_instance)
}
@@ -421,13 +345,13 @@ mod tests {
#[tokio::test]
async fn test_try_from_start_command_to_anymap() {
let mut fe_opts = FrontendOptions {
let command = StartCommand {
user_provider: Some("static_user_provider:cmd:test=test".to_string()),
..Default::default()
};
let plugins = plugins::setup_frontend_plugins(&mut fe_opts).await.unwrap();
let plugins = load_frontend_plugins(&command.user_provider);
let plugins = plugins.unwrap();
let provider = plugins.get::<UserProviderRef>().unwrap();
let result = provider
.authenticate(
@@ -475,9 +399,9 @@ mod tests {
checkpoint_margin = 9
gc_duration = '7s'
[http]
[http_options]
addr = "127.0.0.1:4000"
timeout = "33s"
timeout = "30s"
body_limit = "128MB"
[logging]
@@ -495,24 +419,40 @@ mod tests {
else {
unreachable!()
};
let fe_opts = options.frontend;
let dn_opts = options.datanode;
let fe_opts = options.fe_opts;
let dn_opts = options.dn_opts;
let logging_opts = options.logging;
assert_eq!(Mode::Standalone, fe_opts.mode);
assert_eq!("127.0.0.1:4000".to_string(), fe_opts.http.addr);
assert_eq!(Duration::from_secs(33), fe_opts.http.timeout);
assert_eq!(ReadableSize::mb(128), fe_opts.http.body_limit);
assert_eq!("127.0.0.1:4001".to_string(), fe_opts.grpc.addr);
assert!(fe_opts.mysql.enable);
assert_eq!("127.0.0.1:4002", fe_opts.mysql.addr);
assert_eq!(2, fe_opts.mysql.runtime_size);
assert_eq!(None, fe_opts.mysql.reject_no_database);
assert!(fe_opts.influxdb.enable);
assert_eq!(
"127.0.0.1:4000".to_string(),
fe_opts.http_options.as_ref().unwrap().addr
);
assert_eq!(
Duration::from_secs(30),
fe_opts.http_options.as_ref().unwrap().timeout
);
assert_eq!(
ReadableSize::mb(128),
fe_opts.http_options.as_ref().unwrap().body_limit
);
assert_eq!(
"127.0.0.1:4001".to_string(),
fe_opts.grpc_options.unwrap().addr
);
assert_eq!(
"127.0.0.1:4002",
fe_opts.mysql_options.as_ref().unwrap().addr
);
assert_eq!(2, fe_opts.mysql_options.as_ref().unwrap().runtime_size);
assert_eq!(
None,
fe_opts.mysql_options.as_ref().unwrap().reject_no_database
);
assert!(fe_opts.influxdb_options.as_ref().unwrap().enable);
assert_eq!("/tmp/greptimedb/test/wal", dn_opts.wal.dir.unwrap());
match &dn_opts.storage.store {
datanode::config::ObjectStoreConfig::S3(s3_config) => {
datanode::datanode::ObjectStoreConfig::S3(s3_config) => {
assert_eq!(
"Secret([REDACTED alloc::string::String])".to_string(),
format!("{:?}", s3_config.access_key_id)
@@ -554,7 +494,7 @@ mod tests {
let toml_str = r#"
mode = "standalone"
[http]
[http_options]
addr = "127.0.0.1:4000"
[logging]
@@ -586,10 +526,10 @@ mod tests {
Some("info"),
),
(
// http.addr = 127.0.0.1:24000
// http_options.addr = 127.0.0.1:24000
[
env_prefix.to_string(),
"http".to_uppercase(),
"http_options".to_uppercase(),
"addr".to_uppercase(),
]
.join(ENV_VAR_SEP),
@@ -620,33 +560,21 @@ mod tests {
assert_eq!(opts.logging.level.as_ref().unwrap(), "debug");
// Should be read from cli, cli > config file > env > default values.
assert_eq!(opts.frontend.http.addr, "127.0.0.1:14000");
assert_eq!(ReadableSize::mb(64), opts.frontend.http.body_limit);
assert_eq!(
opts.fe_opts.http_options.as_ref().unwrap().addr,
"127.0.0.1:14000"
);
assert_eq!(
ReadableSize::mb(64),
opts.fe_opts.http_options.as_ref().unwrap().body_limit
);
// Should be default value.
assert_eq!(opts.frontend.grpc.addr, GrpcOptions::default().addr);
assert_eq!(
opts.fe_opts.grpc_options.unwrap().addr,
GrpcOptions::default().addr
);
},
);
}
#[test]
fn test_load_default_standalone_options() {
let options: StandaloneOptions =
Options::load_layered_options(None, "GREPTIMEDB_FRONTEND", None).unwrap();
let default_options = StandaloneOptions::default();
assert_eq!(options.mode, default_options.mode);
assert_eq!(options.enable_telemetry, default_options.enable_telemetry);
assert_eq!(options.http, default_options.http);
assert_eq!(options.grpc, default_options.grpc);
assert_eq!(options.mysql, default_options.mysql);
assert_eq!(options.postgres, default_options.postgres);
assert_eq!(options.opentsdb, default_options.opentsdb);
assert_eq!(options.influxdb, default_options.influxdb);
assert_eq!(options.prom_store, default_options.prom_store);
assert_eq!(options.wal, default_options.wal);
assert_eq!(options.metadata_store, default_options.metadata_store);
assert_eq!(options.procedure, default_options.procedure);
assert_eq!(options.logging, default_options.logging);
assert_eq!(options.region_engine, default_options.region_engine);
}
}

View File

@@ -9,7 +9,6 @@ anymap = "1.0.0-beta.2"
bitvec = "1.0"
bytes = { version = "1.1", features = ["serde"] }
common-error = { workspace = true }
common-macro = { workspace = true }
paste = "1.0"
serde = { version = "1.0", features = ["derive"] }
snafu.workspace = true

View File

@@ -17,13 +17,11 @@ 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)]
#[derive(Debug, Snafu)]
#[snafu(visibility(pub))]
#[stack_trace_debug]
pub enum Error {
#[snafu(display(
"Destination buffer overflow, src_len: {}, dst_len: {}",
@@ -39,10 +37,9 @@ pub enum Error {
#[snafu(display("Buffer underflow"))]
Underflow { location: Location },
#[snafu(display("IO operation reach EOF"))]
#[snafu(display("IO operation reach EOF, source: {}", source))]
Eof {
#[snafu(source)]
error: std::io::Error,
source: std::io::Error,
location: Location,
},
}

View File

@@ -15,6 +15,7 @@
pub mod bit_vec;
pub mod buffer;
pub mod bytes;
pub mod paths;
#[allow(clippy::all)]
pub mod readable_size;
@@ -23,8 +24,6 @@ use std::sync::{Arc, Mutex, MutexGuard};
pub use bit_vec::BitVec;
/// [`Plugins`] is a wrapper of Arc contents.
/// Make it Cloneable and we can treat it like an Arc struct.
#[derive(Default, Clone)]
pub struct Plugins {
inner: Arc<Mutex<anymap::Map<dyn Any + Send + Sync>>>,

View File

@@ -12,14 +12,14 @@
// See the License for the specific language governing permissions and
// limitations under the License.
use common_base::Plugins;
use datanode::config::DatanodeOptions;
use datanode::error::Result;
//! Path constants for table engines, cluster states and WAL
/// All paths relative to data_home(file storage) or root path(S3, OSS etc).
pub async fn setup_datanode_plugins(_opts: &mut DatanodeOptions) -> Result<Plugins> {
Ok(Plugins::new())
}
/// WAL dir for local file storage
pub const WAL_DIR: &str = "wal/";
pub async fn start_datanode_plugins(_plugins: Plugins) -> Result<()> {
Ok(())
}
/// Data dir for table engines
pub const DATA_DIR: &str = "data/";
/// Cluster state dir
pub const CLUSTER_DIR: &str = "cluster/";

View File

@@ -14,9 +14,10 @@
// See the License for the specific language governing permissions and
// limitations under the License.
// This file is copied from https://github.com/tikv/raft-engine/blob/8dd2a39f359ff16f5295f35343f626e0c10132fa/src/util.rs
// This file is copied from https://github.com/tikv/raft-engine/blob/8dd2a39f359ff16f5295f35343f626e0c10132fa/src/util.rs without any modification.
use std::fmt::{self, Debug, Display, Write};
use std::fmt;
use std::fmt::{Display, Write};
use std::ops::{Div, Mul};
use std::str::FromStr;
@@ -33,7 +34,7 @@ pub const GIB: u64 = MIB * BINARY_DATA_MAGNITUDE;
pub const TIB: u64 = GIB * BINARY_DATA_MAGNITUDE;
pub const PIB: u64 = TIB * BINARY_DATA_MAGNITUDE;
#[derive(Clone, Copy, PartialEq, Eq, PartialOrd)]
#[derive(Clone, Debug, Copy, PartialEq, Eq, PartialOrd)]
pub struct ReadableSize(pub u64);
impl ReadableSize {
@@ -154,12 +155,6 @@ impl FromStr for ReadableSize {
}
}
impl Debug for ReadableSize {
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
write!(f, "{}", self)
}
}
impl Display for ReadableSize {
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
if self.0 >= PIB {

View File

@@ -6,7 +6,6 @@ license.workspace = true
[dependencies]
common-error = { workspace = true }
common-macro = { workspace = true }
serde.workspace = true
serde_json = "1.0"
snafu = { version = "0.7", features = ["backtraces"] }

View File

@@ -41,7 +41,7 @@ pub fn default_engine() -> &'static str {
MITO_ENGINE
}
pub const FILE_ENGINE: &str = "file";
pub const IMMUTABLE_FILE_ENGINE: &str = "file";
pub const SEMANTIC_TYPE_PRIMARY_KEY: &str = "TAG";
pub const SEMANTIC_TYPE_FIELD: &str = "FIELD";

View File

@@ -16,24 +16,34 @@ 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)]
#[derive(Debug, Snafu)]
#[snafu(visibility(pub))]
#[stack_trace_debug]
pub enum Error {
#[snafu(display("Invalid full table name: {}", table_name))]
InvalidFullTableName {
table_name: String,
#[snafu(display("Invalid catalog info: {}", key))]
InvalidCatalog { key: String, location: Location },
#[snafu(display("Failed to deserialize catalog entry value: {}", raw))]
DeserializeCatalogEntryValue {
raw: String,
location: Location,
source: serde_json::error::Error,
},
#[snafu(display("Failed to serialize catalog entry value"))]
SerializeCatalogEntryValue {
location: Location,
source: serde_json::error::Error,
},
}
impl ErrorExt for Error {
fn status_code(&self) -> StatusCode {
match self {
Error::InvalidFullTableName { .. } => StatusCode::Unexpected,
Error::InvalidCatalog { .. }
| Error::DeserializeCatalogEntryValue { .. }
| Error::SerializeCatalogEntryValue { .. } => StatusCode::Unexpected,
}
}

View File

@@ -1,10 +0,0 @@
[package]
name = "common-config"
version.workspace = true
edition.workspace = true
license.workspace = true
[dependencies]
common-base.workspace = true
humantime-serde.workspace = true
serde.workspace = true

View File

@@ -1,73 +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::time::Duration;
use common_base::readable_size::ReadableSize;
use serde::{Deserialize, Serialize};
#[derive(Debug, Clone, Serialize, Deserialize, PartialEq, Eq)]
#[serde(default)]
pub struct WalConfig {
// wal directory
pub dir: Option<String>,
// wal file size in bytes
pub file_size: ReadableSize,
// wal purge threshold in bytes
pub purge_threshold: ReadableSize,
// purge interval in seconds
#[serde(with = "humantime_serde")]
pub purge_interval: Duration,
// read batch size
pub read_batch_size: usize,
// whether to sync log file after every write
pub sync_write: bool,
}
impl Default for WalConfig {
fn default() -> Self {
Self {
dir: None,
file_size: ReadableSize::mb(256), // log file size 256MB
purge_threshold: ReadableSize::gb(4), // purge threshold 4GB
purge_interval: Duration::from_secs(600),
read_batch_size: 128,
sync_write: false,
}
}
}
pub fn metadata_store_dir(store_dir: &str) -> String {
format!("{store_dir}/metadata")
}
#[derive(Debug, Clone, Serialize, Deserialize, PartialEq, Eq)]
#[serde(default)]
pub struct KvStoreConfig {
// Kv file size in bytes
pub file_size: ReadableSize,
// Kv purge threshold in bytes
pub purge_threshold: ReadableSize,
}
impl Default for KvStoreConfig {
fn default() -> Self {
Self {
// log file size 256MB
file_size: ReadableSize::mb(256),
// purge threshold 4GB
purge_threshold: ReadableSize::gb(4),
}
}
}

View File

@@ -18,17 +18,14 @@ async-compression = { version = "0.3", features = [
async-trait.workspace = true
bytes = "1.1"
common-error = { workspace = true }
common-macro = { workspace = true }
common-runtime = { workspace = true }
datafusion.workspace = true
derive_builder.workspace = true
futures.workspace = true
lazy_static.workspace = true
object-store = { workspace = true }
orc-rust = "0.2"
paste = "1.0"
regex = "1.7"
serde.workspace = true
snafu.workspace = true
strum.workspace = true
tokio-util.workspace = true

Some files were not shown because too many files have changed in this diff Show More