rust/neon
1
0
Fork 0
mirror of https://github.com/neondatabase/neon.git synced 2026-01-16 09:52:54 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge pull request #10875 from neondatabase/rc/release-compute/2025-02-18

Browse Source 
Compute release 2025-02-18
This commit is contained in:
Heikki Linnakangas
2025-02-19 20:57:42 +02:00
committed by GitHub
parent 156c18e1ad 81367a6bbc
commit 2b7243bd37
226 changed files with 8840 additions and 4646 deletions
Download Patch File Download Diff File Expand all files Collapse all files

26
.github/workflows/_build-and-test-locally.yml vendored
View File

@@ -20,9 +20,14 @@ on:
required: true
type: string
test-cfg:
description: 'a json object of postgres versions and lfc/sanitizers states to build and run regression tests on'
description: 'a json object of postgres versions and lfc states to run regression tests on'
required: true
type: string
sanitizers:
description: 'enabled or disabled'
required: false
default: 'disabled'
type: string
defaults:
run:
@@ -48,8 +53,6 @@ jobs:
# io_uring will account the memory of the CQ and SQ as locked.
# More details: https://github.com/neondatabase/neon/issues/6373#issuecomment-1905814391
options: --init --shm-size=512mb --ulimit memlock=67108864:67108864
strategy:
matrix: ${{ fromJSON(format('{{"include":{0}}}', inputs.test-cfg)) }}
env:
BUILD_TYPE: ${{ inputs.build-type }}
GIT_VERSION: ${{ github.event.pull_request.head.sha || github.sha }}
@@ -89,7 +92,7 @@ jobs:
- name: Set env variables
env:
ARCH: ${{ inputs.arch }}
SANITIZERS: ${{ matrix.sanitizers }}
SANITIZERS: ${{ inputs.sanitizers }}
run: |
CARGO_FEATURES="--features testing"
if [[ $BUILD_TYPE == "debug" && $ARCH == 'x64' ]]; then
@@ -167,7 +170,7 @@ jobs:
- name: Run cargo build
env:
WITH_TESTS: ${{ matrix.sanitizers != 'enabled' && '--tests' || '' }}
WITH_TESTS: ${{ inputs.sanitizers != 'enabled' && '--tests' || '' }}
run: |
export ASAN_OPTIONS=detect_leaks=0
${cov_prefix} mold -run cargo build $CARGO_FLAGS $CARGO_FEATURES --bins ${WITH_TESTS}
@@ -177,7 +180,7 @@ jobs:
- name: Install rust binaries
env:
ARCH: ${{ inputs.arch }}
SANITIZERS: ${{ matrix.sanitizers }}
SANITIZERS: ${{ inputs.sanitizers }}
run: |
# Install target binaries
mkdir -p /tmp/neon/bin/
@@ -225,7 +228,7 @@ jobs:
role-duration-seconds: 18000 # 5 hours
- name: Run rust tests
if: ${{ matrix.sanitizers != 'enabled' }}
if: ${{ inputs.sanitizers != 'enabled' }}
env:
NEXTEST_RETRIES: 3
run: |
@@ -287,6 +290,7 @@ jobs:
DATABASE_URL: postgresql://localhost:1235/storage_controller
POSTGRES_DISTRIB_DIR: /tmp/neon/pg_install
run: |
export ASAN_OPTIONS=detect_leaks=0
/tmp/neon/bin/neon_local init
/tmp/neon/bin/neon_local storage_controller start
@@ -333,7 +337,7 @@ jobs:
- name: Pytest regression tests
continue-on-error: ${{ matrix.lfc_state == 'with-lfc' && inputs.build-type == 'debug' }}
uses: ./.github/actions/run-python-test-set
timeout-minutes: ${{ matrix.sanitizers != 'enabled' && 60 || 180 }}
timeout-minutes: ${{ inputs.sanitizers != 'enabled' && 60 || 180 }}
with:
build_type: ${{ inputs.build-type }}
test_selection: regress
@@ -344,6 +348,10 @@ jobs:
rerun_failed: true
pg_version: ${{ matrix.pg_version }}
aws-oicd-role-arn: ${{ vars.DEV_AWS_OIDC_ROLE_ARN }}
# `--session-timeout` is equal to (timeout-minutes - 10 minutes) * 60 seconds.
# Attempt to stop tests gracefully to generate test reports
# until they are forcibly stopped by the stricter `timeout-minutes` limit.
extra_params: --session-timeout=${{ inputs.sanitizers != 'enabled' && 3000 || 10200 }}
env:
TEST_RESULT_CONNSTR: ${{ secrets.REGRESS_TEST_RESULT_CONNSTR_NEW }}
CHECK_ONDISK_DATA_COMPATIBILITY: nonempty
@@ -351,7 +359,7 @@ jobs:
PAGESERVER_VIRTUAL_FILE_IO_ENGINE: tokio-epoll-uring
PAGESERVER_GET_VECTORED_CONCURRENT_IO: sidecar-task
USE_LFC: ${{ matrix.lfc_state == 'with-lfc' && 'true' || 'false' }}
SANITIZERS: ${{ matrix.sanitizers }}
SANITIZERS: ${{ inputs.sanitizers }}
# Temporary disable this step until we figure out why it's so flaky
# Ref https://github.com/neondatabase/neon/issues/4540

56
.github/workflows/_push-to-acr.yml vendored
View File

@@ -1,56 +0,0 @@
name: Push images to ACR
on:
workflow_call:
inputs:
client_id:
description: Client ID of Azure managed identity or Entra app
required: true
type: string
image_tag:
description: Tag for the container image
required: true
type: string
images:
description: Images to push
required: true
type: string
registry_name:
description: Name of the container registry
required: true
type: string
subscription_id:
description: Azure subscription ID
required: true
type: string
tenant_id:
description: Azure tenant ID
required: true
type: string
jobs:
push-to-acr:
runs-on: ubuntu-22.04
permissions:
contents: read # This is required for actions/checkout
id-token: write # This is required for Azure Login to work.
steps:
- name: Azure login
uses: azure/login@6c251865b4e6290e7b78be643ea2d005bc51f69a # @v2.1.1
with:
client-id: ${{ inputs.client_id }}
subscription-id: ${{ inputs.subscription_id }}
tenant-id: ${{ inputs.tenant_id }}
- name: Login to ACR
run: |
az acr login --name=${{ inputs.registry_name }}
- name: Copy docker images to ACR ${{ inputs.registry_name }}
run: |
images='${{ inputs.images }}'
for image in ${images}; do
docker buildx imagetools create \
-t ${{ inputs.registry_name }}.azurecr.io/neondatabase/${image}:${{ inputs.image_tag }} \
neondatabase/${image}:${{ inputs.image_tag }}
done

101
.github/workflows/_push-to-container-registry.yml vendored Normal file
View File

@@ -0,0 +1,101 @@
name: Push images to Container Registry
on:
workflow_call:
inputs:
# Example: {"docker.io/neondatabase/neon:13196061314":["369495373322.dkr.ecr.eu-central-1.amazonaws.com/neon:13196061314","neoneastus2.azurecr.io/neondatabase/neon:13196061314"]}
image-map:
description: JSON map of images, mapping from a source image to an array of target images that should be pushed.
required: true
type: string
aws-region:
description: AWS region to log in to. Required when pushing to ECR.
required: false
type: string
aws-account-ids:
description: Comma separated AWS account IDs to log in to for pushing to ECR. Required when pushing to ECR.
required: false
type: string
azure-client-id:
description: Client ID of Azure managed identity or Entra app. Required when pushing to ACR.
required: false
type: string
azure-subscription-id:
description: Azure subscription ID. Required when pushing to ACR.
required: false
type: string
azure-tenant-id:
description: Azure tenant ID. Required when pushing to ACR.
required: false
type: string
acr-registry-name:
description: ACR registry name. Required when pushing to ACR.
required: false
type: string
secrets:
docker-hub-username:
description: Docker Hub username. Required when pushing to Docker Hub.
required: false
docker-hub-password:
description: Docker Hub password. Required when pushing to Docker Hub.
required: false
aws-role-to-assume:
description: AWS role to assume. Required when pushing to ECR.
required: false
permissions: {}
defaults:
run:
shell: bash -euo pipefail {0}
jobs:
push-to-container-registry:
runs-on: ubuntu-22.04
permissions:
id-token: write # Required for aws/azure login
steps:
- uses: actions/checkout@v4
with:
sparse-checkout: scripts/push_with_image_map.py
sparse-checkout-cone-mode: false
- name: Print image-map
run: echo '${{ inputs.image-map }}' | jq
- name: Configure AWS credentials
if: contains(inputs.image-map, 'amazonaws.com/')
uses: aws-actions/configure-aws-credentials@v4
with:
aws-region: "${{ inputs.aws-region }}"
role-to-assume: "${{ secrets.aws-role-to-assume }}"
role-duration-seconds: 3600
- name: Login to ECR
if: contains(inputs.image-map, 'amazonaws.com/')
uses: aws-actions/amazon-ecr-login@v2
with:
registries: "${{ inputs.aws-account-ids }}"
- name: Configure Azure credentials
if: contains(inputs.image-map, 'azurecr.io/')
uses: azure/login@6c251865b4e6290e7b78be643ea2d005bc51f69a # @v2.1.1
with:
client-id: ${{ inputs.azure-client-id }}
subscription-id: ${{ inputs.azure-subscription-id }}
tenant-id: ${{ inputs.azure-tenant-id }}
- name: Login to ACR
if: contains(inputs.image-map, 'azurecr.io/')
run: |
az acr login --name=${{ inputs.acr-registry-name }}
- name: Log in to Docker Hub
uses: docker/login-action@v3
with:
username: ${{ secrets.docker-hub-username }}
password: ${{ secrets.docker-hub-password }}
- name: Copy docker images to target registries
run: python scripts/push_with_image_map.py
env:
IMAGE_MAP: ${{ inputs.image-map }}

269
.github/workflows/build_and_test.yml vendored
View File

@@ -263,8 +263,9 @@ jobs:
echo "json=$(jq --compact-output '.' /tmp/benchmark_durations.json)" >> $GITHUB_OUTPUT
benchmarks:
if: github.ref_name == 'main' || contains(github.event.pull_request.labels.*.name, 'run-benchmarks')
needs: [ check-permissions, build-and-test-locally, build-build-tools-image, get-benchmarks-durations ]
# `!failure() && !cancelled()` is required because the workflow depends on the job that can be skipped: `deploy` in PRs
if: github.ref_name == 'main' || (contains(github.event.pull_request.labels.*.name, 'run-benchmarks') && !failure() && !cancelled())
needs: [ check-permissions, build-build-tools-image, get-benchmarks-durations, deploy ]
permissions:
id-token: write # aws-actions/configure-aws-credentials
statuses: write
@@ -497,7 +498,7 @@ jobs:
trigger-e2e-tests:
if: ${{ !github.event.pull_request.draft || contains( github.event.pull_request.labels.*.name, 'run-e2e-tests-in-draft') || github.ref_name == 'main' || github.ref_name == 'release' || github.ref_name == 'release-proxy' || github.ref_name == 'release-compute' }}
needs: [ check-permissions, promote-images-dev, tag ]
needs: [ check-permissions, push-neon-image-dev, push-compute-image-dev, tag ]
uses: ./.github/workflows/trigger-e2e-tests.yml
secrets: inherit
@@ -571,21 +572,6 @@ jobs:
neondatabase/neon:${{ needs.tag.outputs.build-tag }}-bookworm-x64 \
neondatabase/neon:${{ needs.tag.outputs.build-tag }}-bookworm-arm64
- name: Configure AWS credentials
uses: aws-actions/configure-aws-credentials@v4
with:
aws-region: eu-central-1
role-to-assume: ${{ vars.DEV_AWS_OIDC_ROLE_ARN }}
role-duration-seconds: 3600
- name: Login to Amazon Dev ECR
uses: aws-actions/amazon-ecr-login@v2
- name: Push multi-arch image to ECR
run: |
docker buildx imagetools create -t 369495373322.dkr.ecr.eu-central-1.amazonaws.com/neon:${{ needs.tag.outputs.build-tag }} \
neondatabase/neon:${{ needs.tag.outputs.build-tag }}
compute-node-image-arch:
needs: [ check-permissions, build-build-tools-image, tag ]
permissions:
@@ -632,16 +618,6 @@ jobs:
username: ${{ secrets.NEON_DOCKERHUB_USERNAME }}
password: ${{ secrets.NEON_DOCKERHUB_PASSWORD }}
- name: Configure AWS credentials
uses: aws-actions/configure-aws-credentials@v4
with:
aws-region: eu-central-1
role-to-assume: ${{ vars.DEV_AWS_OIDC_ROLE_ARN }}
role-duration-seconds: 3600
- name: Login to Amazon Dev ECR
uses: aws-actions/amazon-ecr-login@v2
- uses: docker/login-action@v3
with:
registry: cache.neon.build
@@ -729,21 +705,6 @@ jobs:
neondatabase/neon-test-extensions-${{ matrix.version.pg }}:${{ needs.tag.outputs.build-tag }}-${{ matrix.version.debian }}-x64 \
neondatabase/neon-test-extensions-${{ matrix.version.pg }}:${{ needs.tag.outputs.build-tag }}-${{ matrix.version.debian }}-arm64
- name: Configure AWS credentials
uses: aws-actions/configure-aws-credentials@v4
with:
aws-region: eu-central-1
role-to-assume: ${{ vars.DEV_AWS_OIDC_ROLE_ARN }}
role-duration-seconds: 3600
- name: Login to Amazon Dev ECR
uses: aws-actions/amazon-ecr-login@v2
- name: Push multi-arch compute-node-${{ matrix.version.pg }} image to ECR
run: |
docker buildx imagetools create -t 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node-${{ matrix.version.pg }}:${{ needs.tag.outputs.build-tag }} \
neondatabase/compute-node-${{ matrix.version.pg }}:${{ needs.tag.outputs.build-tag }}
vm-compute-node-image:
needs: [ check-permissions, tag, compute-node-image ]
runs-on: [ self-hosted, large ]
@@ -876,133 +837,109 @@ jobs:
docker compose --profile test-extensions -f ./docker-compose/docker-compose.yml logs || true
docker compose --profile test-extensions -f ./docker-compose/docker-compose.yml down
promote-images-dev:
needs: [ check-permissions, tag, vm-compute-node-image, neon-image ]
generate-image-maps:
needs: [ tag ]
runs-on: ubuntu-22.04
permissions:
id-token: write # aws-actions/configure-aws-credentials
statuses: write
contents: read
env:
VERSIONS: v14 v15 v16 v17
outputs:
neon-dev: ${{ steps.generate.outputs.neon-dev }}
neon-prod: ${{ steps.generate.outputs.neon-prod }}
compute-dev: ${{ steps.generate.outputs.compute-dev }}
compute-prod: ${{ steps.generate.outputs.compute-prod }}
steps:
- uses: docker/login-action@v3
- uses: actions/checkout@v4
with:
username: ${{ secrets.NEON_DOCKERHUB_USERNAME }}
password: ${{ secrets.NEON_DOCKERHUB_PASSWORD }}
sparse-checkout: scripts/generate_image_maps.py
sparse-checkout-cone-mode: false
- name: Configure AWS credentials
uses: aws-actions/configure-aws-credentials@v4
with:
aws-region: eu-central-1
role-to-assume: ${{ vars.DEV_AWS_OIDC_ROLE_ARN }}
role-duration-seconds: 3600
- name: Generate Image Maps
id: generate
run: python scripts/generate_image_maps.py
env:
BUILD_TAG: "${{ needs.tag.outputs.build-tag }}"
BRANCH: "${{ github.ref_name }}"
DEV_ACR: "${{ vars.AZURE_DEV_REGISTRY_NAME }}"
PROD_ACR: "${{ vars.AZURE_PROD_REGISTRY_NAME }}"
- name: Login to Amazon Dev ECR
uses: aws-actions/amazon-ecr-login@v2
- name: Copy vm-compute-node images to ECR
run: |
for version in ${VERSIONS}; do
docker buildx imagetools create -t 369495373322.dkr.ecr.eu-central-1.amazonaws.com/vm-compute-node-${version}:${{ needs.tag.outputs.build-tag }} \
neondatabase/vm-compute-node-${version}:${{ needs.tag.outputs.build-tag }}
done
promote-images-prod:
needs: [ check-permissions, tag, test-images, promote-images-dev ]
runs-on: ubuntu-22.04
if: github.ref_name == 'main' || github.ref_name == 'release' || github.ref_name == 'release-proxy' || github.ref_name == 'release-compute'
permissions:
id-token: write # aws-actions/configure-aws-credentials
statuses: write
contents: read
env:
VERSIONS: v14 v15 v16 v17
steps:
- name: Configure AWS credentials
uses: aws-actions/configure-aws-credentials@v4
with:
aws-region: eu-central-1
role-to-assume: ${{ vars.DEV_AWS_OIDC_ROLE_ARN }}
role-duration-seconds: 3600
- name: Login to Amazon Dev ECR
uses: aws-actions/amazon-ecr-login@v2
- uses: docker/login-action@v3
with:
username: ${{ secrets.NEON_DOCKERHUB_USERNAME }}
password: ${{ secrets.NEON_DOCKERHUB_PASSWORD }}
- name: Add latest tag to images
if: github.ref_name == 'main'
run: |
for repo in neondatabase 369495373322.dkr.ecr.eu-central-1.amazonaws.com; do
docker buildx imagetools create -t $repo/neon:latest \
$repo/neon:${{ needs.tag.outputs.build-tag }}
for version in ${VERSIONS}; do
docker buildx imagetools create -t $repo/compute-node-${version}:latest \
$repo/compute-node-${version}:${{ needs.tag.outputs.build-tag }}
docker buildx imagetools create -t $repo/vm-compute-node-${version}:latest \
$repo/vm-compute-node-${version}:${{ needs.tag.outputs.build-tag }}
done
done
docker buildx imagetools create -t neondatabase/neon-test-extensions-v16:latest \
neondatabase/neon-test-extensions-v16:${{ needs.tag.outputs.build-tag }}
- name: Configure AWS-prod credentials
if: github.ref_name == 'release'|| github.ref_name == 'release-proxy' || github.ref_name == 'release-compute'
uses: aws-actions/configure-aws-credentials@v4
with:
aws-region: eu-central-1
mask-aws-account-id: true
role-to-assume: ${{ secrets.PROD_GHA_OIDC_ROLE }}
- name: Login to prod ECR
uses: docker/login-action@v3
if: github.ref_name == 'release'|| github.ref_name == 'release-proxy' || github.ref_name == 'release-compute'
with:
registry: 093970136003.dkr.ecr.eu-central-1.amazonaws.com
- name: Copy all images to prod ECR
if: github.ref_name == 'release' || github.ref_name == 'release-proxy' || github.ref_name == 'release-compute'
run: |
for image in neon {vm-,}compute-node-{v14,v15,v16,v17}; do
docker buildx imagetools create -t 093970136003.dkr.ecr.eu-central-1.amazonaws.com/${image}:${{ needs.tag.outputs.build-tag }} \
369495373322.dkr.ecr.eu-central-1.amazonaws.com/${image}:${{ needs.tag.outputs.build-tag }}
done
push-to-acr-dev:
if: github.ref_name == 'main'
needs: [ tag, promote-images-dev ]
uses: ./.github/workflows/_push-to-acr.yml
push-neon-image-dev:
needs: [ generate-image-maps, neon-image ]
uses: ./.github/workflows/_push-to-container-registry.yml
with:
client_id: ${{ vars.AZURE_DEV_CLIENT_ID }}
image_tag: ${{ needs.tag.outputs.build-tag }}
images: neon vm-compute-node-v14 vm-compute-node-v15 vm-compute-node-v16 vm-compute-node-v17 compute-node-v14 compute-node-v15 compute-node-v16 compute-node-v17
registry_name: ${{ vars.AZURE_DEV_REGISTRY_NAME }}
subscription_id: ${{ vars.AZURE_DEV_SUBSCRIPTION_ID }}
tenant_id: ${{ vars.AZURE_TENANT_ID }}
image-map: '${{ needs.generate-image-maps.outputs.neon-dev }}'
aws-region: eu-central-1
aws-account-ids: "369495373322"
azure-client-id: ${{ vars.AZURE_DEV_CLIENT_ID }}
azure-subscription-id: ${{ vars.AZURE_DEV_SUBSCRIPTION_ID }}
azure-tenant-id: ${{ vars.AZURE_TENANT_ID }}
acr-registry-name: ${{ vars.AZURE_DEV_REGISTRY_NAME }}
secrets:
aws-role-to-assume: "${{ vars.DEV_AWS_OIDC_ROLE_ARN }}"
docker-hub-username: ${{ secrets.NEON_DOCKERHUB_USERNAME }}
docker-hub-password: ${{ secrets.NEON_DOCKERHUB_PASSWORD }}
push-to-acr-prod:
push-compute-image-dev:
needs: [ generate-image-maps, vm-compute-node-image ]
uses: ./.github/workflows/_push-to-container-registry.yml
with:
image-map: '${{ needs.generate-image-maps.outputs.compute-dev }}'
aws-region: eu-central-1
aws-account-ids: "369495373322"
azure-client-id: ${{ vars.AZURE_DEV_CLIENT_ID }}
azure-subscription-id: ${{ vars.AZURE_DEV_SUBSCRIPTION_ID }}
azure-tenant-id: ${{ vars.AZURE_TENANT_ID }}
acr-registry-name: ${{ vars.AZURE_DEV_REGISTRY_NAME }}
secrets:
aws-role-to-assume: "${{ vars.DEV_AWS_OIDC_ROLE_ARN }}"
docker-hub-username: ${{ secrets.NEON_DOCKERHUB_USERNAME }}
docker-hub-password: ${{ secrets.NEON_DOCKERHUB_PASSWORD }}
push-neon-image-prod:
if: github.ref_name == 'release' || github.ref_name == 'release-proxy' || github.ref_name == 'release-compute'
needs: [ tag, promote-images-prod ]
uses: ./.github/workflows/_push-to-acr.yml
needs: [ generate-image-maps, neon-image, test-images ]
uses: ./.github/workflows/_push-to-container-registry.yml
with:
client_id: ${{ vars.AZURE_PROD_CLIENT_ID }}
image_tag: ${{ needs.tag.outputs.build-tag }}
images: neon vm-compute-node-v14 vm-compute-node-v15 vm-compute-node-v16 vm-compute-node-v17 compute-node-v14 compute-node-v15 compute-node-v16 compute-node-v17
registry_name: ${{ vars.AZURE_PROD_REGISTRY_NAME }}
subscription_id: ${{ vars.AZURE_PROD_SUBSCRIPTION_ID }}
tenant_id: ${{ vars.AZURE_TENANT_ID }}
image-map: '${{ needs.generate-image-maps.outputs.neon-prod }}'
aws-region: eu-central-1
aws-account-ids: "093970136003"
azure-client-id: ${{ vars.AZURE_PROD_CLIENT_ID }}
azure-subscription-id: ${{ vars.AZURE_PROD_SUBSCRIPTION_ID }}
azure-tenant-id: ${{ vars.AZURE_TENANT_ID }}
acr-registry-name: ${{ vars.AZURE_PROD_REGISTRY_NAME }}
secrets:
aws-role-to-assume: "${{ secrets.PROD_GHA_OIDC_ROLE }}"
docker-hub-username: ${{ secrets.NEON_DOCKERHUB_USERNAME }}
docker-hub-password: ${{ secrets.NEON_DOCKERHUB_PASSWORD }}
push-compute-image-prod:
if: github.ref_name == 'release' || github.ref_name == 'release-proxy' || github.ref_name == 'release-compute'
needs: [ generate-image-maps, vm-compute-node-image, test-images ]
uses: ./.github/workflows/_push-to-container-registry.yml
with:
image-map: '${{ needs.generate-image-maps.outputs.compute-prod }}'
aws-region: eu-central-1
aws-account-ids: "093970136003"
azure-client-id: ${{ vars.AZURE_PROD_CLIENT_ID }}
azure-subscription-id: ${{ vars.AZURE_PROD_SUBSCRIPTION_ID }}
azure-tenant-id: ${{ vars.AZURE_TENANT_ID }}
acr-registry-name: ${{ vars.AZURE_PROD_REGISTRY_NAME }}
secrets:
aws-role-to-assume: "${{ secrets.PROD_GHA_OIDC_ROLE }}"
docker-hub-username: ${{ secrets.NEON_DOCKERHUB_USERNAME }}
docker-hub-password: ${{ secrets.NEON_DOCKERHUB_PASSWORD }}
# This is a bit of a special case so we're not using a generated image map.
add-latest-tag-to-neon-extensions-test-image:
if: github.ref_name == 'main'
needs: [ tag, compute-node-image ]
uses: ./.github/workflows/_push-to-container-registry.yml
with:
image-map: |
{
"docker.io/neondatabase/neon-test-extensions-v16:${{ needs.tag.outputs.build-tag }}": ["docker.io/neondatabase/neon-test-extensions-v16:latest"],
"docker.io/neondatabase/neon-test-extensions-v17:${{ needs.tag.outputs.build-tag }}": ["docker.io/neondatabase/neon-test-extensions-v17:latest"]
}
secrets:
docker-hub-username: ${{ secrets.NEON_DOCKERHUB_USERNAME }}
docker-hub-password: ${{ secrets.NEON_DOCKERHUB_PASSWORD }}
trigger-custom-extensions-build-and-wait:
needs: [ check-permissions, tag ]
@@ -1084,7 +1021,7 @@ jobs:
exit 1
deploy:
needs: [ check-permissions, promote-images-prod, tag, build-and-test-locally, trigger-custom-extensions-build-and-wait, push-to-acr-dev, push-to-acr-prod ]
needs: [ check-permissions, push-neon-image-prod, push-compute-image-prod, tag, build-and-test-locally, trigger-custom-extensions-build-and-wait ]
# `!failure() && !cancelled()` is required because the workflow depends on the job that can be skipped: `push-to-acr-dev` and `push-to-acr-prod`
if: (github.ref_name == 'main' || github.ref_name == 'release' || github.ref_name == 'release-proxy' || github.ref_name == 'release-compute') && !failure() && !cancelled()
permissions:
@@ -1337,7 +1274,7 @@ jobs:
done
pin-build-tools-image:
needs: [ build-build-tools-image, promote-images-prod, build-and-test-locally ]
needs: [ build-build-tools-image, push-compute-image-prod, push-neon-image-prod, build-and-test-locally ]
if: github.ref_name == 'main'
uses: ./.github/workflows/pin-build-tools-image.yml
with:
@@ -1362,7 +1299,8 @@ jobs:
- check-codestyle-rust
- check-dependencies-rust
- files-changed
- promote-images-dev
- push-compute-image-dev
- push-neon-image-dev
- test-images
- trigger-custom-extensions-build-and-wait
runs-on: ubuntu-22.04
@@ -1379,6 +1317,7 @@ jobs:
|| needs.check-codestyle-python.result == 'skipped'
|| needs.check-codestyle-rust.result == 'skipped'
|| needs.files-changed.result == 'skipped'
|| needs.promote-images-dev.result == 'skipped'
|| needs.push-compute-image-dev.result == 'skipped'
|| needs.push-neon-image-dev.result == 'skipped'
|| needs.test-images.result == 'skipped'
|| needs.trigger-custom-extensions-build-and-wait.result == 'skipped'

3
.github/workflows/build_and_test_with_sanitizers.yml vendored
View File

@@ -74,7 +74,8 @@ jobs:
build-tools-image: ${{ needs.build-build-tools-image.outputs.image }}-bookworm
build-tag: ${{ needs.tag.outputs.build-tag }}
build-type: ${{ matrix.build-type }}
test-cfg: '[{"pg_version":"v17", "sanitizers": "enabled"}]'
test-cfg: '[{"pg_version":"v17"}]'
sanitizers: enabled
secrets: inherit

76
.github/workflows/force-test-extensions-upgrade.yml vendored Normal file
View File

@@ -0,0 +1,76 @@
name: Force Test Upgrading of Extension
on:
schedule:
# * is a special character in YAML so you have to quote this string
# ┌───────────── minute (0 - 59)
# │ ┌───────────── hour (0 - 23)
# │ │ ┌───────────── day of the month (1 - 31)
# │ │ │ ┌───────────── month (1 - 12 or JAN-DEC)
# │ │ │ │ ┌───────────── day of the week (0 - 6 or SUN-SAT)
- cron: '45 2 * * *' # run once a day, timezone is utc
workflow_dispatch: # adds ability to run this manually
defaults:
run:
shell: bash -euxo pipefail {0}
concurrency:
# Allow only one workflow
group: ${{ github.workflow }}
cancel-in-progress: true
permissions:
id-token: write # aws-actions/configure-aws-credentials
statuses: write
contents: read
jobs:
regress:
strategy:
fail-fast: false
matrix:
pg-version: [16, 17]
runs-on: small
steps:
- uses: actions/checkout@v4
with:
submodules: false
- name: Get the last compute release tag
id: get-last-compute-release-tag
env:
GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
run: |
tag=$(gh api -q '[.[].tag_name | select(startswith("release-compute"))][0]'\
-H "Accept: application/vnd.github+json" \
-H "X-GitHub-Api-Version: 2022-11-28" \
"/repos/${GITHUB_REPOSITORY}/releases")
echo tag=${tag} >> ${GITHUB_OUTPUT}
- name: Test extension upgrade
timeout-minutes: 20
env:
NEWTAG: latest
OLDTAG: ${{ steps.get-last-compute-release-tag.outputs.tag }}
PG_VERSION: ${{ matrix.pg-version }}
FORCE_ALL_UPGRADE_TESTS: true
run: ./docker-compose/test_extensions_upgrade.sh
- name: Print logs and clean up
if: always()
run: |
docker compose --profile test-extensions -f ./docker-compose/docker-compose.yml logs || true
docker compose --profile test-extensions -f ./docker-compose/docker-compose.yml down
- name: Post to the Slack channel
if: ${{ github.event.schedule && failure() }}
uses: slackapi/slack-github-action@v1
with:
channel-id: ${{ vars.SLACK_ON_CALL_QA_STAGING_STREAM }}
slack-message: |
Test upgrading of extensions: ${{ job.status }}
<${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}|GitHub Run>
env:
SLACK_BOT_TOKEN: ${{ secrets.SLACK_BOT_TOKEN }}

41
.github/workflows/regenerate-pg-setting.yml vendored Normal file
View File

@@ -0,0 +1,41 @@
name: Regenerate Postgres Settings
on:
pull_request:
types:
- opened
- synchronize
- reopened
paths:
- pgxn/neon/**.c
- vendor/postgres-v*
- vendor/revisions.json
concurrency:
group: ${{ github.workflow }}-${{ github.head_ref }}
cancel-in-progress: true
permissions:
pull-requests: write
jobs:
regenerate-pg-settings:
runs-on: ubuntu-22.04
steps:
- name: Add comment
uses: thollander/actions-comment-pull-request@v3
with:
comment-tag: ${{ github.job }}
pr-number: ${{ github.event.number }}
message: |
If this PR added a GUC in the Postgres fork or `neon` extension,
please regenerate the Postgres settings in the `cloud` repo:
```
make NEON_WORKDIR=path/to/neon/checkout \
-C goapp/internal/shareddomain/postgres generate
```
If you're an external contributor, a Neon employee will assist in
making sure this step is done.

38
.github/workflows/trigger-e2e-tests.yml vendored
View File

@@ -15,7 +15,14 @@ env:
E2E_CONCURRENCY_GROUP: ${{ github.repository }}-e2e-tests-${{ github.ref_name }}-${{ github.ref_name == 'main' && github.sha || 'anysha' }}
jobs:
check-permissions:
if: ${{ !contains(github.event.pull_request.labels.*.name, 'run-no-ci') }}
uses: ./.github/workflows/check-permissions.yml
with:
github-event-name: ${{ github.event_name }}
cancel-previous-e2e-tests:
needs: [ check-permissions ]
if: github.event_name == 'pull_request'
runs-on: ubuntu-22.04
@@ -29,6 +36,7 @@ jobs:
--field concurrency_group="${{ env.E2E_CONCURRENCY_GROUP }}"
tag:
needs: [ check-permissions ]
runs-on: ubuntu-22.04
outputs:
build-tag: ${{ steps.build-tag.outputs.tag }}
@@ -68,7 +76,7 @@ jobs:
GH_TOKEN: ${{ secrets.CI_ACCESS_TOKEN }}
TAG: ${{ needs.tag.outputs.build-tag }}
steps:
- name: Wait for `promote-images-dev` job to finish
- name: Wait for `push-{neon,compute}-image-dev` job to finish
# It's important to have a timeout here, the script in the step can run infinitely
timeout-minutes: 60
run: |
@@ -79,20 +87,20 @@ jobs:
# For PRs we use the run id as the tag
BUILD_AND_TEST_RUN_ID=${TAG}
while true; do
conclusion=$(gh run --repo ${GITHUB_REPOSITORY} view ${BUILD_AND_TEST_RUN_ID} --json jobs --jq '.jobs[] | select(.name == "promote-images-dev") | .conclusion')
case "$conclusion" in
success)
break
;;
failure | cancelled | skipped)
echo "The 'promote-images-dev' job didn't succeed: '${conclusion}'. Exiting..."
exit 1
;;
*)
echo "The 'promote-images-dev' hasn't succeed yet. Waiting..."
sleep 60
;;
esac
gh run --repo ${GITHUB_REPOSITORY} view ${BUILD_AND_TEST_RUN_ID} --json jobs --jq '[.jobs[] | select((.name | startswith("push-neon-image-dev")) or (.name | startswith("push-compute-image-dev"))) | {"name": .name, "conclusion": .conclusion, "url": .url}]' > jobs.json
if [ $(jq '[.[] | select(.conclusion == "success")] | length' jobs.json) -eq 2 ]; then
break
fi
jq -c '.[]' jobs.json | while read -r job; do
case $(echo $job | jq .conclusion) in
failure | cancelled | skipped)
echo "The '$(echo $job | jq .name)' job didn't succeed: '$(echo $job | jq .conclusion)'. See log in '$(echo $job | jq .url)' Exiting..."
exit 1
;;
esac
done
echo "The 'push-{neon,compute}-image-dev' jobs haven't succeeded yet. Waiting..."
sleep 60
done
- name: Set e2e-platforms

164
Cargo.lock generated
View File

@@ -300,9 +300,9 @@ checksum = "d468802bab17cbc0cc575e9b053f41e72aa36bfa6b7f55e3529ffa43161b97fa"
[[package]]
name = "aws-config"
version = "1.5.15"
version = "1.5.10"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "dc47e70fc35d054c8fcd296d47a61711f043ac80534a10b4f741904f81e73a90"
checksum = "9b49afaa341e8dd8577e1a2200468f98956d6eda50bcf4a53246cc00174ba924"
dependencies = [
"aws-credential-types",
"aws-runtime",
@@ -311,7 +311,7 @@ dependencies = [
"aws-sdk-sts",
"aws-smithy-async",
"aws-smithy-http",
"aws-smithy-json",
"aws-smithy-json 0.60.7",
"aws-smithy-runtime",
"aws-smithy-runtime-api",
"aws-smithy-types",
@@ -342,9 +342,9 @@ dependencies = [
[[package]]
name = "aws-runtime"
version = "1.5.4"
version = "1.4.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "bee7643696e7fdd74c10f9eb42848a87fe469d35eae9c3323f80aa98f350baac"
checksum = "b5ac934720fbb46206292d2c75b57e67acfc56fe7dfd34fb9a02334af08409ea"
dependencies = [
"aws-credential-types",
"aws-sigv4",
@@ -368,15 +368,15 @@ dependencies = [
[[package]]
name = "aws-sdk-iam"
version = "1.60.0"
version = "1.53.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a43daa438f8e7e4ebbbcb5c712b3b85db50d62e637a7da4ba9da51095d327460"
checksum = "fb8a6fea8d335cde419176b1f2c6d2d6e97997719e7df4b51e59064310f48e4a"
dependencies = [
"aws-credential-types",
"aws-runtime",
"aws-smithy-async",
"aws-smithy-http",
"aws-smithy-json",
"aws-smithy-json 0.61.1",
"aws-smithy-query",
"aws-smithy-runtime",
"aws-smithy-runtime-api",
@@ -391,15 +391,15 @@ dependencies = [
[[package]]
name = "aws-sdk-kms"
version = "1.58.0"
version = "1.51.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "40b7a24700ac548025a47a5c579886f5198895bb1eccd8964dfd71cd66c16912"
checksum = "3c30f6fd5646b99d9b45ec3a0c22e67112c175b2383100c960d7ee39d96c8d96"
dependencies = [
"aws-credential-types",
"aws-runtime",
"aws-smithy-async",
"aws-smithy-http",
"aws-smithy-json",
"aws-smithy-json 0.61.1",
"aws-smithy-runtime",
"aws-smithy-runtime-api",
"aws-smithy-types",
@@ -413,9 +413,9 @@ dependencies = [
[[package]]
name = "aws-sdk-s3"
version = "1.68.0"
version = "1.65.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "bc5ddf1dc70287dc9a2f953766a1fe15e3e74aef02fd1335f2afa475c9b4f4fc"
checksum = "d3ba2c5c0f2618937ce3d4a5ad574b86775576fa24006bcb3128c6e2cbf3c34e"
dependencies = [
"aws-credential-types",
"aws-runtime",
@@ -424,7 +424,7 @@ dependencies = [
"aws-smithy-checksums",
"aws-smithy-eventstream",
"aws-smithy-http",
"aws-smithy-json",
"aws-smithy-json 0.61.1",
"aws-smithy-runtime",
"aws-smithy-runtime-api",
"aws-smithy-types",
@@ -447,15 +447,15 @@ dependencies = [
[[package]]
name = "aws-sdk-sso"
version = "1.57.0"
version = "1.50.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "c54bab121fe1881a74c338c5f723d1592bf3b53167f80268a1274f404e1acc38"
checksum = "05ca43a4ef210894f93096039ef1d6fa4ad3edfabb3be92b80908b9f2e4b4eab"
dependencies = [
"aws-credential-types",
"aws-runtime",
"aws-smithy-async",
"aws-smithy-http",
"aws-smithy-json",
"aws-smithy-json 0.61.1",
"aws-smithy-runtime",
"aws-smithy-runtime-api",
"aws-smithy-types",
@@ -469,15 +469,15 @@ dependencies = [
[[package]]
name = "aws-sdk-ssooidc"
version = "1.58.0"
version = "1.51.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8c8234fd024f7ac61c4e44ea008029bde934250f371efe7d4a39708397b1080c"
checksum = "abaf490c2e48eed0bb8e2da2fb08405647bd7f253996e0f93b981958ea0f73b0"
dependencies = [
"aws-credential-types",
"aws-runtime",
"aws-smithy-async",
"aws-smithy-http",
"aws-smithy-json",
"aws-smithy-json 0.61.1",
"aws-smithy-runtime",
"aws-smithy-runtime-api",
"aws-smithy-types",
@@ -491,15 +491,15 @@ dependencies = [
[[package]]
name = "aws-sdk-sts"
version = "1.58.0"
version = "1.51.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ba60e1d519d6f23a9df712c04fdeadd7872ac911c84b2f62a8bda92e129b7962"
checksum = "b68fde0d69c8bfdc1060ea7da21df3e39f6014da316783336deff0a9ec28f4bf"
dependencies = [
"aws-credential-types",
"aws-runtime",
"aws-smithy-async",
"aws-smithy-http",
"aws-smithy-json",
"aws-smithy-json 0.61.1",
"aws-smithy-query",
"aws-smithy-runtime",
"aws-smithy-runtime-api",
@@ -514,9 +514,9 @@ dependencies = [
[[package]]
name = "aws-sigv4"
version = "1.2.8"
version = "1.2.6"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "0bc5bbd1e4a2648fd8c5982af03935972c24a2f9846b396de661d351ee3ce837"
checksum = "7d3820e0c08d0737872ff3c7c1f21ebbb6693d832312d6152bf18ef50a5471c2"
dependencies = [
"aws-credential-types",
"aws-smithy-eventstream",
@@ -543,9 +543,9 @@ dependencies = [
[[package]]
name = "aws-smithy-async"
version = "1.2.4"
version = "1.2.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "fa59d1327d8b5053c54bf2eaae63bf629ba9e904434d0835a28ed3c0ed0a614e"
checksum = "62220bc6e97f946ddd51b5f1361f78996e704677afc518a4ff66b7a72ea1378c"
dependencies = [
"futures-util",
"pin-project-lite",
@@ -575,9 +575,9 @@ dependencies = [
[[package]]
name = "aws-smithy-eventstream"
version = "0.60.6"
version = "0.60.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8b18559a41e0c909b77625adf2b8c50de480a8041e5e4a3f5f7d177db70abc5a"
checksum = "cef7d0a272725f87e51ba2bf89f8c21e4df61b9e49ae1ac367a6d69916ef7c90"
dependencies = [
"aws-smithy-types",
"bytes",
@@ -586,9 +586,9 @@ dependencies = [
[[package]]
name = "aws-smithy-http"
version = "0.60.12"
version = "0.60.11"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "7809c27ad8da6a6a68c454e651d4962479e81472aa19ae99e59f9aba1f9713cc"
checksum = "5c8bc3e8fdc6b8d07d976e301c02fe553f72a39b7a9fea820e023268467d7ab6"
dependencies = [
"aws-smithy-eventstream",
"aws-smithy-runtime-api",
@@ -607,9 +607,18 @@ dependencies = [
[[package]]
name = "aws-smithy-json"
version = "0.61.2"
version = "0.60.7"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "623a51127f24c30776c8b374295f2df78d92517386f77ba30773f15a30ce1422"
checksum = "4683df9469ef09468dad3473d129960119a0d3593617542b7d52086c8486f2d6"
dependencies = [
"aws-smithy-types",
]
[[package]]
name = "aws-smithy-json"
version = "0.61.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ee4e69cc50921eb913c6b662f8d909131bb3e6ad6cb6090d3a39b66fc5c52095"
dependencies = [
"aws-smithy-types",
]
@@ -626,9 +635,9 @@ dependencies = [
[[package]]
name = "aws-smithy-runtime"
version = "1.7.7"
version = "1.7.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "865f7050bbc7107a6c98a397a9fcd9413690c27fa718446967cf03b2d3ac517e"
checksum = "9f20685047ca9d6f17b994a07f629c813f08b5bce65523e47124879e60103d45"
dependencies = [
"aws-smithy-async",
"aws-smithy-http",
@@ -670,9 +679,9 @@ dependencies = [
[[package]]
name = "aws-smithy-types"
version = "1.2.13"
version = "1.2.9"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "c7b8a53819e42f10d0821f56da995e1470b199686a1809168db6ca485665f042"
checksum = "4fbd94a32b3a7d55d3806fe27d98d3ad393050439dd05eb53ece36ec5e3d3510"
dependencies = [
"base64-simd",
"bytes",
@@ -705,9 +714,9 @@ dependencies = [
[[package]]
name = "aws-types"
version = "1.3.5"
version = "1.3.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "dfbd0a668309ec1f66c0f6bda4840dd6d4796ae26d699ebc266d7cc95c6d040f"
checksum = "5221b91b3e441e6675310829fd8984801b772cb1546ef6c0e54dec9f1ac13fef"
dependencies = [
"aws-credential-types",
"aws-smithy-async",
@@ -777,7 +786,7 @@ dependencies = [
[[package]]
name = "azure_core"
version = "0.21.0"
source = "git+https://github.com/neondatabase/azure-sdk-for-rust.git?branch=neon#66e77bdd87bf87e773acf3b0c84b532c1124367d"
source = "git+https://github.com/neondatabase/azure-sdk-for-rust.git?branch=neon#f64bd57262ced51afce5d8909c06dcb11a6dd85a"
dependencies = [
"async-trait",
"base64 0.22.1",
@@ -806,7 +815,7 @@ dependencies = [
[[package]]
name = "azure_identity"
version = "0.21.0"
source = "git+https://github.com/neondatabase/azure-sdk-for-rust.git?branch=neon#66e77bdd87bf87e773acf3b0c84b532c1124367d"
source = "git+https://github.com/neondatabase/azure-sdk-for-rust.git?branch=neon#f64bd57262ced51afce5d8909c06dcb11a6dd85a"
dependencies = [
"async-lock",
"async-trait",
@@ -825,7 +834,7 @@ dependencies = [
[[package]]
name = "azure_storage"
version = "0.21.0"
source = "git+https://github.com/neondatabase/azure-sdk-for-rust.git?branch=neon#66e77bdd87bf87e773acf3b0c84b532c1124367d"
source = "git+https://github.com/neondatabase/azure-sdk-for-rust.git?branch=neon#f64bd57262ced51afce5d8909c06dcb11a6dd85a"
dependencies = [
"RustyXML",
"async-lock",
@@ -843,7 +852,7 @@ dependencies = [
[[package]]
name = "azure_storage_blobs"
version = "0.21.0"
source = "git+https://github.com/neondatabase/azure-sdk-for-rust.git?branch=neon#66e77bdd87bf87e773acf3b0c84b532c1124367d"
source = "git+https://github.com/neondatabase/azure-sdk-for-rust.git?branch=neon#f64bd57262ced51afce5d8909c06dcb11a6dd85a"
dependencies = [
"RustyXML",
"azure_core",
@@ -863,7 +872,7 @@ dependencies = [
[[package]]
name = "azure_svc_blobstorage"
version = "0.21.0"
source = "git+https://github.com/neondatabase/azure-sdk-for-rust.git?branch=neon#66e77bdd87bf87e773acf3b0c84b532c1124367d"
source = "git+https://github.com/neondatabase/azure-sdk-for-rust.git?branch=neon#f64bd57262ced51afce5d8909c06dcb11a6dd85a"
dependencies = [
"azure_core",
"bytes",
@@ -1020,12 +1029,6 @@ dependencies = [
"generic-array",
]
[[package]]
name = "boxcar"
version = "0.2.8"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "2721c3c5a6f0e7f7e607125d963fedeb765f545f67adc9d71ed934693881eb42"
[[package]]
name = "bstr"
version = "1.5.0"
@@ -1284,6 +1287,7 @@ version = "0.1.0"
dependencies = [
"anyhow",
"chrono",
"jsonwebtoken",
"regex",
"remote_storage",
"serde",
@@ -1299,6 +1303,7 @@ dependencies = [
"aws-config",
"aws-sdk-kms",
"aws-sdk-s3",
"aws-smithy-types",
"axum",
"base64 0.13.1",
"bytes",
@@ -1311,6 +1316,7 @@ dependencies = [
"flate2",
"futures",
"http 1.1.0",
"jsonwebtoken",
"metrics",
"nix 0.27.1",
"notify",
@@ -1346,6 +1352,7 @@ dependencies = [
"utils",
"uuid",
"vm_monitor",
"walkdir",
"workspace_hack",
"zstd",
]
@@ -1424,6 +1431,7 @@ dependencies = [
"comfy-table",
"compute_api",
"futures",
"http-utils",
"humantime",
"humantime-serde",
"hyper 0.14.30",
@@ -2748,6 +2756,38 @@ dependencies = [
"url",
]
[[package]]
name = "http-utils"
version = "0.1.0"
dependencies = [
"anyhow",
"backtrace",
"bytes",
"fail",
"flate2",
"hyper 0.14.30",
"inferno 0.12.0",
"itertools 0.10.5",
"jemalloc_pprof",
"metrics",
"once_cell",
"pprof",
"regex",
"routerify",
"serde",
"serde_json",
"serde_path_to_error",
"thiserror 1.0.69",
"tokio",
"tokio-stream",
"tokio-util",
"tracing",
"url",
"utils",
"uuid",
"workspace_hack",
]
[[package]]
name = "httparse"
version = "1.8.0"
@@ -4102,6 +4142,7 @@ dependencies = [
"futures",
"hex",
"hex-literal",
"http-utils",
"humantime",
"humantime-serde",
"hyper 0.14.30",
@@ -4202,6 +4243,7 @@ dependencies = [
"anyhow",
"bytes",
"futures",
"http-utils",
"pageserver_api",
"postgres",
"reqwest",
@@ -4883,7 +4925,6 @@ dependencies = [
"aws-sdk-iam",
"aws-sigv4",
"base64 0.13.1",
"boxcar",
"bstr",
"bytes",
"camino",
@@ -4908,6 +4949,7 @@ dependencies = [
"hostname",
"http 1.1.0",
"http-body-util",
"http-utils",
"humantime",
"humantime-serde",
"hyper 0.14.30",
@@ -4934,7 +4976,6 @@ dependencies = [
"postgres-protocol2",
"postgres_backend",
"pq_proto",
"prometheus",
"rand 0.8.5",
"rand_distr",
"rcgen",
@@ -4959,7 +5000,6 @@ dependencies = [
"smallvec",
"smol_str",
"socket2",
"strum",
"strum_macros",
"subtle",
"thiserror 1.0.69",
@@ -4974,7 +5014,6 @@ dependencies = [
"tracing",
"tracing-log",
"tracing-opentelemetry",
"tracing-serde",
"tracing-subscriber",
"tracing-utils",
"try-lock",
@@ -5755,6 +5794,7 @@ dependencies = [
"futures",
"hex",
"http 1.1.0",
"http-utils",
"humantime",
"hyper 0.14.30",
"itertools 0.10.5",
@@ -5819,6 +5859,7 @@ dependencies = [
name = "safekeeper_client"
version = "0.1.0"
dependencies = [
"http-utils",
"reqwest",
"safekeeper_api",
"serde",
@@ -6401,6 +6442,7 @@ dependencies = [
"fail",
"futures",
"hex",
"http-utils",
"humantime",
"hyper 0.14.30",
"itertools 0.10.5",
@@ -6412,10 +6454,13 @@ dependencies = [
"pageserver_client",
"postgres_connection",
"rand 0.8.5",
"regex",
"reqwest",
"routerify",
"rustls 0.23.18",
"rustls-native-certs 0.8.0",
"safekeeper_api",
"safekeeper_client",
"scoped-futures",
"scopeguard",
"serde",
@@ -7565,48 +7610,38 @@ dependencies = [
"criterion",
"diatomic-waker",
"fail",
"flate2",
"futures",
"git-version",
"hex",
"hex-literal",
"humantime",
"hyper 0.14.30",
"inferno 0.12.0",
"itertools 0.10.5",
"jemalloc_pprof",
"jsonwebtoken",
"metrics",
"nix 0.27.1",
"once_cell",
"pin-project-lite",
"postgres_connection",
"pprof",
"pq_proto",
"rand 0.8.5",
"regex",
"routerify",
"scopeguard",
"sentry",
"serde",
"serde_assert",
"serde_json",
"serde_path_to_error",
"serde_with",
"signal-hook",
"strum",
"strum_macros",
"thiserror 1.0.69",
"tokio",
"tokio-stream",
"tokio-tar",
"tokio-util",
"toml_edit",
"tracing",
"tracing-error",
"tracing-subscriber",
"url",
"uuid",
"walkdir",
]
@@ -8201,6 +8236,7 @@ dependencies = [
"tracing-core",
"tracing-log",
"url",
"uuid",
"zerocopy",
"zeroize",
"zstd",

2
Cargo.toml
View File

@@ -18,6 +18,7 @@ members = [
"storage_scrubber",
"workspace_hack",
"libs/compute_api",
"libs/http-utils",
"libs/pageserver_api",
"libs/postgres_ffi",
"libs/safekeeper_api",
@@ -229,6 +230,7 @@ azure_storage_blobs = { git = "https://github.com/neondatabase/azure-sdk-for-rus
## Local libraries
compute_api = { version = "0.1", path = "./libs/compute_api/" }
consumption_metrics = { version = "0.1", path = "./libs/consumption_metrics/" }
http-utils = { version = "0.1", path = "./libs/http-utils/" }
metrics = { version = "0.1", path = "./libs/metrics/" }
pageserver = { path = "./pageserver" }
pageserver_api = { version = "0.1", path = "./libs/pageserver_api/" }

41
Dockerfile
View File

@@ -10,6 +10,28 @@ ARG STABLE_PG_VERSION=16
ARG DEBIAN_VERSION=bookworm
ARG DEBIAN_FLAVOR=${DEBIAN_VERSION}-slim
# Here are the INDEX DIGESTS for the images we use.
# You can get them following next steps for now:
# 1. Get an authentication token from DockerHub:
# TOKEN=$(curl -s "https://auth.docker.io/token?service=registry.docker.io&scope=repository:library/debian:pull" | jq -r .token)
# 2. Using that token, query index for the given tag:
# curl -s -H "Authorization: Bearer $TOKEN" \
# -H "Accept: application/vnd.docker.distribution.manifest.list.v2+json" \
# "https://registry.hub.docker.com/v2/library/debian/manifests/bullseye-slim" \
# -I | grep -i docker-content-digest
# 3. As a next step, TODO(fedordikarev): create script and schedule workflow to run these checks
# and updates on regular bases and in automated way.
ARG BOOKWORM_SLIM_SHA=sha256:40b107342c492725bc7aacbe93a49945445191ae364184a6d24fedb28172f6f7
ARG BULLSEYE_SLIM_SHA=sha256:e831d9a884d63734fe3dd9c491ed9a5a3d4c6a6d32c5b14f2067357c49b0b7e1
# Here we use ${var/search/replace} syntax, to check
# if base image is one of the images, we pin image index for.
# If var will match one the known images, we will replace it with the known sha.
# If no match, than value will be unaffected, and will process with no-pinned image.
ARG BASE_IMAGE_SHA=debian:${DEBIAN_FLAVOR}
ARG BASE_IMAGE_SHA=${BASE_IMAGE_SHA/debian:bookworm-slim/debian@$BOOKWORM_SLIM_SHA}
ARG BASE_IMAGE_SHA=${BASE_IMAGE_SHA/debian:bullseye-slim/debian@$BULLSEYE_SLIM_SHA}
# Build Postgres
FROM $REPOSITORY/$IMAGE:$TAG AS pg-build
WORKDIR /home/nonroot
@@ -28,6 +50,14 @@ RUN set -e \
&& rm -rf pg_install/build \
&& tar -C pg_install -czf /home/nonroot/postgres_install.tar.gz .
# Prepare cargo-chef recipe
FROM $REPOSITORY/$IMAGE:$TAG AS plan
WORKDIR /home/nonroot
COPY --chown=nonroot . .
RUN cargo chef prepare --recipe-path recipe.json
# Build neon binaries
FROM $REPOSITORY/$IMAGE:$TAG AS build
WORKDIR /home/nonroot
@@ -41,9 +71,15 @@ COPY --from=pg-build /home/nonroot/pg_install/v16/include/postgresql/server pg_i
COPY --from=pg-build /home/nonroot/pg_install/v17/include/postgresql/server pg_install/v17/include/postgresql/server
COPY --from=pg-build /home/nonroot/pg_install/v16/lib pg_install/v16/lib
COPY --from=pg-build /home/nonroot/pg_install/v17/lib pg_install/v17/lib
COPY --from=plan /home/nonroot/recipe.json recipe.json
ARG ADDITIONAL_RUSTFLAGS=""
RUN set -e \
&& RUSTFLAGS="-Clinker=clang -Clink-arg=-fuse-ld=mold -Clink-arg=-Wl,--no-rosegment -Cforce-frame-pointers=yes ${ADDITIONAL_RUSTFLAGS}" cargo chef cook --locked --release --recipe-path recipe.json
COPY --chown=nonroot . .
ARG ADDITIONAL_RUSTFLAGS
RUN set -e \
&& RUSTFLAGS="-Clinker=clang -Clink-arg=-fuse-ld=mold -Clink-arg=-Wl,--no-rosegment -Cforce-frame-pointers=yes ${ADDITIONAL_RUSTFLAGS}" cargo build \
--bin pg_sni_router \
@@ -59,7 +95,7 @@ RUN set -e \
# Build final image
#
FROM debian:${DEBIAN_FLAVOR}
FROM $BASE_IMAGE_SHA
ARG DEFAULT_PG_VERSION
WORKDIR /data
@@ -112,4 +148,3 @@ EXPOSE 6400
EXPOSE 9898
CMD ["/usr/local/bin/pageserver", "-D", "/data/.neon"]

36
build-tools.Dockerfile
View File

@@ -1,6 +1,29 @@
ARG DEBIAN_VERSION=bookworm
ARG DEBIAN_FLAVOR=${DEBIAN_VERSION}-slim
FROM debian:bookworm-slim AS pgcopydb_builder
# Here are the INDEX DIGESTS for the images we use.
# You can get them following next steps for now:
# 1. Get an authentication token from DockerHub:
# TOKEN=$(curl -s "https://auth.docker.io/token?service=registry.docker.io&scope=repository:library/debian:pull" | jq -r .token)
# 2. Using that token, query index for the given tag:
# curl -s -H "Authorization: Bearer $TOKEN" \
# -H "Accept: application/vnd.docker.distribution.manifest.list.v2+json" \
# "https://registry.hub.docker.com/v2/library/debian/manifests/bullseye-slim" \
# -I | grep -i docker-content-digest
# 3. As a next step, TODO(fedordikarev): create script and schedule workflow to run these checks
# and updates on regular bases and in automated way.
ARG BOOKWORM_SLIM_SHA=sha256:40b107342c492725bc7aacbe93a49945445191ae364184a6d24fedb28172f6f7
ARG BULLSEYE_SLIM_SHA=sha256:e831d9a884d63734fe3dd9c491ed9a5a3d4c6a6d32c5b14f2067357c49b0b7e1
# Here we use ${var/search/replace} syntax, to check
# if base image is one of the images, we pin image index for.
# If var will match one the known images, we will replace it with the known sha.
# If no match, than value will be unaffected, and will process with no-pinned image.
ARG BASE_IMAGE_SHA=debian:${DEBIAN_FLAVOR}
ARG BASE_IMAGE_SHA=${BASE_IMAGE_SHA/debian:bookworm-slim/debian@$BOOKWORM_SLIM_SHA}
ARG BASE_IMAGE_SHA=${BASE_IMAGE_SHA/debian:bullseye-slim/debian@$BULLSEYE_SLIM_SHA}
FROM $BASE_IMAGE_SHA AS pgcopydb_builder
ARG DEBIAN_VERSION
# Use strict mode for bash to catch errors early
@@ -9,7 +32,7 @@ SHELL ["/bin/bash", "-euo", "pipefail", "-c"]
# By default, /bin/sh used in debian images will treat '\n' as eol,
# but as we use bash as SHELL, and built-in echo in bash requires '-e' flag for that.
RUN echo 'Acquire::Retries "5";' > /etc/apt/apt.conf.d/80-retries && \
echo -e "retry_connrefused = on\ntimeout=15\ntries=5\n" > /root/.wgetrc && \
echo -e "retry_connrefused=on\ntimeout=15\ntries=5\nretry-on-host-error=on\n" > /root/.wgetrc && \
echo -e "--retry-connrefused\n--connect-timeout 15\n--retry 5\n--max-time 300\n" > /root/.curlrc
COPY build_tools/patches/pgcopydbv017.patch /pgcopydbv017.patch
@@ -58,7 +81,7 @@ RUN if [ "${DEBIAN_VERSION}" = "bookworm" ]; then \
mkdir -p mkdir -p /pgcopydb/lib && touch /pgcopydb/lib/libpq.so.5; \
fi
FROM debian:${DEBIAN_VERSION}-slim AS build_tools
FROM $BASE_IMAGE_SHA AS build_tools
ARG DEBIAN_VERSION
# Add nonroot user
@@ -75,7 +98,7 @@ COPY --from=pgcopydb_builder /usr/lib/postgresql/16/bin/pgcopydb /pgcopydb/bin/p
COPY --from=pgcopydb_builder /pgcopydb/lib/libpq.so.5 /pgcopydb/lib/libpq.so.5
RUN echo 'Acquire::Retries "5";' > /etc/apt/apt.conf.d/80-retries && \
echo -e "retry_connrefused = on\ntimeout=15\ntries=5\n" > /root/.wgetrc && \
echo -e "retry_connrefused=on\ntimeout=15\ntries=5\nretry-on-host-error=on\n" > /root/.wgetrc && \
echo -e "--retry-connrefused\n--connect-timeout 15\n--retry 5\n--max-time 300\n" > /root/.curlrc
# System deps
@@ -138,7 +161,8 @@ RUN curl -fsSL \
--output sql_exporter.tar.gz \
&& mkdir /tmp/sql_exporter \
&& tar xzvf sql_exporter.tar.gz -C /tmp/sql_exporter --strip-components=1 \
&& mv /tmp/sql_exporter/sql_exporter /usr/local/bin/sql_exporter
&& mv /tmp/sql_exporter/sql_exporter /usr/local/bin/sql_exporter \
&& rm sql_exporter.tar.gz
# protobuf-compiler (protoc)
ENV PROTOC_VERSION=25.1
@@ -276,6 +300,7 @@ ARG CARGO_HAKARI_VERSION=0.9.33
ARG CARGO_DENY_VERSION=0.16.2
ARG CARGO_HACK_VERSION=0.6.33
ARG CARGO_NEXTEST_VERSION=0.9.85
ARG CARGO_CHEF_VERSION=0.1.71
ARG CARGO_DIESEL_CLI_VERSION=2.2.6
RUN curl -sSO https://static.rust-lang.org/rustup/dist/$(uname -m)-unknown-linux-gnu/rustup-init && whoami && \
chmod +x rustup-init && \
@@ -290,6 +315,7 @@ RUN curl -sSO https://static.rust-lang.org/rustup/dist/$(uname -m)-unknown-linux
cargo install cargo-deny --locked --version ${CARGO_DENY_VERSION} && \
cargo install cargo-hack --version ${CARGO_HACK_VERSION} && \
cargo install cargo-nextest --version ${CARGO_NEXTEST_VERSION} && \
cargo install cargo-chef --locked --version ${CARGO_CHEF_VERSION} && \
cargo install diesel_cli --version ${CARGO_DIESEL_CLI_VERSION} \
--features postgres-bundled --no-default-features && \
rm -rf /home/nonroot/.cargo/registry && \

292
compute/compute-node.Dockerfile
View File

@@ -83,7 +83,28 @@ ARG TAG=pinned
ARG BUILD_TAG
ARG DEBIAN_VERSION=bookworm
ARG DEBIAN_FLAVOR=${DEBIAN_VERSION}-slim
ARG ALPINE_CURL_VERSION=8.11.1
# Here are the INDEX DIGESTS for the images we use.
# You can get them following next steps for now:
# 1. Get an authentication token from DockerHub:
# TOKEN=$(curl -s "https://auth.docker.io/token?service=registry.docker.io&scope=repository:library/debian:pull" | jq -r .token)
# 2. Using that token, query index for the given tag:
# curl -s -H "Authorization: Bearer $TOKEN" \
# -H "Accept: application/vnd.docker.distribution.manifest.list.v2+json" \
# "https://registry.hub.docker.com/v2/library/debian/manifests/bullseye-slim" \
# -I | grep -i docker-content-digest
# 3. As a next step, TODO(fedordikarev): create script and schedule workflow to run these checks
# and updates on regular bases and in automated way.
ARG BOOKWORM_SLIM_SHA=sha256:40b107342c492725bc7aacbe93a49945445191ae364184a6d24fedb28172f6f7
ARG BULLSEYE_SLIM_SHA=sha256:e831d9a884d63734fe3dd9c491ed9a5a3d4c6a6d32c5b14f2067357c49b0b7e1
# Here we use ${var/search/replace} syntax, to check
# if base image is one of the images, we pin image index for.
# If var will match one the known images, we will replace it with the known sha.
# If no match, than value will be unaffected, and will process with no-pinned image.
ARG BASE_IMAGE_SHA=debian:${DEBIAN_FLAVOR}
ARG BASE_IMAGE_SHA=${BASE_IMAGE_SHA/debian:bookworm-slim/debian@$BOOKWORM_SLIM_SHA}
ARG BASE_IMAGE_SHA=${BASE_IMAGE_SHA/debian:bullseye-slim/debian@$BULLSEYE_SLIM_SHA}
# By default, build all PostgreSQL extensions. For quick local testing when you don't
# care about the extensions, pass EXTENSIONS=none or EXTENSIONS=minimal
@@ -94,7 +115,7 @@ ARG EXTENSIONS=all
# Layer "build-deps"
#
#########################################################################################
FROM debian:$DEBIAN_FLAVOR AS build-deps
FROM $BASE_IMAGE_SHA AS build-deps
ARG DEBIAN_VERSION
# Use strict mode for bash to catch errors early
@@ -103,7 +124,7 @@ SHELL ["/bin/bash", "-euo", "pipefail", "-c"]
# By default, /bin/sh used in debian images will treat '\n' as eol,
# but as we use bash as SHELL, and built-in echo in bash requires '-e' flag for that.
RUN echo 'Acquire::Retries "5";' > /etc/apt/apt.conf.d/80-retries && \
echo -e "retry_connrefused = on\ntimeout=15\ntries=5\n" > /root/.wgetrc && \
echo -e "retry_connrefused = on\ntimeout=15\ntries=5\nretry-on-host-error=on\n" > /root/.wgetrc && \
echo -e "--retry-connrefused\n--connect-timeout 15\n--retry 5\n--max-time 300\n" > /root/.curlrc
RUN case $DEBIAN_VERSION in \
@@ -127,7 +148,7 @@ RUN case $DEBIAN_VERSION in \
apt install --no-install-recommends --no-install-suggests -y \
ninja-build git autoconf automake libtool build-essential bison flex libreadline-dev \
zlib1g-dev libxml2-dev libcurl4-openssl-dev libossp-uuid-dev wget ca-certificates pkg-config libssl-dev \
libicu-dev libxslt1-dev liblz4-dev libzstd-dev zstd \
libicu-dev libxslt1-dev liblz4-dev libzstd-dev zstd curl unzip g++ \
$VERSION_INSTALLS \
&& apt clean && rm -rf /var/lib/apt/lists/*
@@ -139,11 +160,11 @@ RUN case $DEBIAN_VERSION in \
#########################################################################################
FROM build-deps AS pg-build
ARG PG_VERSION
COPY vendor/postgres-${PG_VERSION} postgres
COPY vendor/postgres-${PG_VERSION:?} postgres
RUN cd postgres && \
export CONFIGURE_CMD="./configure CFLAGS='-O2 -g3' --enable-debug --with-openssl --with-uuid=ossp \
--with-icu --with-libxml --with-libxslt --with-lz4" && \
if [ "${PG_VERSION}" != "v14" ]; then \
if [ "${PG_VERSION:?}" != "v14" ]; then \
# zstd is available only from PG15
export CONFIGURE_CMD="${CONFIGURE_CMD} --with-zstd"; \
fi && \
@@ -237,7 +258,7 @@ RUN case "${DEBIAN_VERSION}" in \
# Postgis 3.5.0 supports v17
WORKDIR /ext-src
RUN case "${PG_VERSION}" in \
RUN case "${PG_VERSION:?}" in \
"v17") \
export POSTGIS_VERSION=3.5.0 \
export POSTGIS_CHECKSUM=ca698a22cc2b2b3467ac4e063b43a28413f3004ddd505bdccdd74c56a647f510 \
@@ -312,7 +333,7 @@ FROM build-deps AS pgrouting-src
ARG DEBIAN_VERSION
ARG PG_VERSION
WORKDIR /ext-src
RUN case "${PG_VERSION}" in \
RUN case "${PG_VERSION:?}" in \
"v17") \
export PGROUTING_VERSION=3.6.2 \
export PGROUTING_CHECKSUM=f4a1ed79d6f714e52548eca3bb8e5593c6745f1bde92eb5fb858efd8984dffa2 \
@@ -358,7 +379,7 @@ COPY compute/patches/plv8-3.1.10.patch .
#
# Use new version only for v17
# because since v3.2, plv8 doesn't include plcoffee and plls extensions
RUN case "${PG_VERSION}" in \
RUN case "${PG_VERSION:?}" in \
"v17") \
export PLV8_TAG=v3.2.3 \
;; \
@@ -372,7 +393,7 @@ RUN case "${PG_VERSION}" in \
git clone --recurse-submodules --depth 1 --branch ${PLV8_TAG} https://github.com/plv8/plv8.git plv8-src && \
tar -czf plv8.tar.gz --exclude .git plv8-src && \
cd plv8-src && \
if [[ "${PG_VERSION}" < "v17" ]]; then patch -p1 < /ext-src/plv8-3.1.10.patch; fi
if [[ "${PG_VERSION:?}" < "v17" ]]; then patch -p1 < /ext-src/plv8-3.1.10.patch; fi
FROM pg-build AS plv8-build
ARG PG_VERSION
@@ -392,7 +413,7 @@ RUN \
find /usr/local/pgsql/ -name "plv8-*.so" | xargs strip && \
# don't break computes with installed old version of plv8
cd /usr/local/pgsql/lib/ && \
case "${PG_VERSION}" in \
case "${PG_VERSION:?}" in \
"v17") \
ln -s plv8-3.2.3.so plv8-3.1.8.so && \
ln -s plv8-3.2.3.so plv8-3.1.5.so && \
@@ -729,7 +750,7 @@ FROM build-deps AS timescaledb-src
ARG PG_VERSION
WORKDIR /ext-src
RUN case "${PG_VERSION}" in \
RUN case "${PG_VERSION:?}" in \
"v14" | "v15") \
export TIMESCALEDB_VERSION=2.10.1 \
export TIMESCALEDB_CHECKSUM=6fca72a6ed0f6d32d2b3523951ede73dc5f9b0077b38450a029a5f411fdb8c73 \
@@ -767,7 +788,7 @@ ARG PG_VERSION
# version-specific, has separate releases for each version
WORKDIR /ext-src
RUN case "${PG_VERSION}" in \
RUN case "${PG_VERSION:?}" in \
"v14") \
export PG_HINT_PLAN_VERSION=14_1_4_1 \
export PG_HINT_PLAN_CHECKSUM=c3501becf70ead27f70626bce80ea401ceac6a77e2083ee5f3ff1f1444ec1ad1 \
@@ -843,7 +864,7 @@ ARG PG_VERSION
# https://github.com/rdkit/rdkit/releases/tag/Release_2024_09_1
WORKDIR /ext-src
RUN case "${PG_VERSION}" in \
RUN case "${PG_VERSION:?}" in \
"v17") \
export RDKIT_VERSION=Release_2024_09_1 \
export RDKIT_CHECKSUM=034c00d6e9de323506834da03400761ed8c3721095114369d06805409747a60f \
@@ -970,7 +991,7 @@ ARG PG_VERSION
#
# last release v0.40.0 - Jul 22, 2024
WORKDIR /ext-src
RUN case "${PG_VERSION}" in \
RUN case "${PG_VERSION:?}" in \
"v17") \
export SEMVER_VERSION=0.40.0 \
export SEMVER_CHECKSUM=3e50bcc29a0e2e481e7b6d2bc937cadc5f5869f55d983b5a1aafeb49f5425cfc \
@@ -1006,7 +1027,7 @@ ARG PG_VERSION
# This is our extension, support stopped in favor of pgvector
# TODO: deprecate it
WORKDIR /ext-src
RUN case "${PG_VERSION}" in \
RUN case "${PG_VERSION:?}" in \
"v14" | "v15") \
export PG_EMBEDDING_VERSION=0.3.5 \
export PG_EMBEDDING_CHECKSUM=0e95b27b8b6196e2cf0a0c9ec143fe2219b82e54c5bb4ee064e76398cbe69ae9 \
@@ -1039,7 +1060,7 @@ ARG PG_VERSION
# This is an experimental extension, never got to real production.
# !Do not remove! It can be present in shared_preload_libraries and compute will fail to start if library is not found.
WORKDIR /ext-src
RUN case "${PG_VERSION}" in "v17") \
RUN case "${PG_VERSION:?}" in "v17") \
echo "postgresql_anonymizer does not yet support PG17" && exit 0;; \
esac && \
wget https://github.com/neondatabase/postgresql_anonymizer/archive/refs/tags/neon_1.1.1.tar.gz -O pg_anon.tar.gz && \
@@ -1091,7 +1112,7 @@ RUN curl -sSO https://static.rust-lang.org/rustup/dist/$(uname -m)-unknown-linux
FROM pg-build-nonroot-with-cargo AS rust-extensions-build
ARG PG_VERSION
RUN case "${PG_VERSION}" in \
RUN case "${PG_VERSION:?}" in \
'v17') \
echo 'v17 is not supported yet by pgrx. Quit' && exit 0;; \
esac && \
@@ -1270,7 +1291,7 @@ FROM build-deps AS pgx_ulid-src
ARG PG_VERSION
WORKDIR /ext-src
RUN case "${PG_VERSION}" in \
RUN case "${PG_VERSION:?}" in \
"v14" | "v15" | "v16") \
;; \
*) \
@@ -1302,7 +1323,7 @@ FROM build-deps AS pgx_ulid-pgrx12-src
ARG PG_VERSION
WORKDIR /ext-src
RUN case "${PG_VERSION}" in \
RUN case "${PG_VERSION:?}" in \
"v17") \
;; \
*) \
@@ -1430,8 +1451,8 @@ RUN make -j $(getconf _NPROCESSORS_ONLN) && \
FROM build-deps AS pg_mooncake-src
ARG PG_VERSION
WORKDIR /ext-src
RUN wget https://github.com/Mooncake-Labs/pg_mooncake/releases/download/v0.1.1/pg_mooncake-0.1.1.tar.gz -O pg_mooncake.tar.gz && \
echo "a2d16eff7948dde64f072609ca5d2962d6b4d07cb89d45952add473529c55f55 pg_mooncake.tar.gz" | sha256sum --check && \
RUN wget https://github.com/Mooncake-Labs/pg_mooncake/releases/download/v0.1.2/pg_mooncake-0.1.2.tar.gz -O pg_mooncake.tar.gz && \
echo "4550473784fcdd2e1e18062bc01eb9c286abd27cdf5e11a4399be6c0a426ba90 pg_mooncake.tar.gz" | sha256sum --check && \
mkdir pg_mooncake-src && cd pg_mooncake-src && tar xzf ../pg_mooncake.tar.gz --strip-components=1 -C . && \
echo "make -f pg_mooncake-src/Makefile.build installcheck TEST_DIR=./test SQL_DIR=./sql SRC_DIR=./src" > neon-test.sh && \
chmod a+x neon-test.sh
@@ -1443,6 +1464,31 @@ RUN make release -j $(getconf _NPROCESSORS_ONLN) && \
make install -j $(getconf _NPROCESSORS_ONLN) && \
echo 'trusted = true' >> /usr/local/pgsql/share/extension/pg_mooncake.control
#########################################################################################
#
# Layer "pg-duckdb-pg-build"
# compile pg_duckdb extension
#
#########################################################################################
FROM build-deps AS pg_duckdb-src
WORKDIR /ext-src
COPY compute/patches/pg_duckdb_v031.patch .
# pg_duckdb build requires source dir to be a git repo to get submodules
# allow neon_superuser to execute some functions that in pg_duckdb are available to superuser only:
# - extension management function duckdb.install_extension()
# - access to duckdb.extensions table and its sequence
RUN git clone --depth 1 --branch v0.3.1 https://github.com/duckdb/pg_duckdb.git pg_duckdb-src && \
cd pg_duckdb-src && \
git submodule update --init --recursive && \
patch -p1 < /ext-src/pg_duckdb_v031.patch
FROM pg-build AS pg_duckdb-build
ARG PG_VERSION
COPY --from=pg_duckdb-src /ext-src/ /ext-src/
WORKDIR /ext-src/pg_duckdb-src
RUN make install -j $(getconf _NPROCESSORS_ONLN) && \
echo 'trusted = true' >> /usr/local/pgsql/share/extension/pg_duckdb.control
#########################################################################################
#
# Layer "pg_repack"
@@ -1463,6 +1509,73 @@ WORKDIR /ext-src/pg_repack-src
RUN make -j $(getconf _NPROCESSORS_ONLN) && \
make -j $(getconf _NPROCESSORS_ONLN) install
#########################################################################################
#
# Layer "pgaudit"
# compile pgaudit extension
#
#########################################################################################
FROM build-deps AS pgaudit-src
ARG PG_VERSION
WORKDIR /ext-src
RUN case "${PG_VERSION}" in \
"v14") \
export PGAUDIT_VERSION=1.6.2 \
export PGAUDIT_CHECKSUM=1f350d70a0cbf488c0f2b485e3a5c9b11f78ad9e3cbb95ef6904afa1eb3187eb \
;; \
"v15") \
export PGAUDIT_VERSION=1.7.0 \
export PGAUDIT_CHECKSUM=8f4a73e451c88c567e516e6cba7dc1e23bc91686bb6f1f77f8f3126d428a8bd8 \
;; \
"v16") \
export PGAUDIT_VERSION=16.0 \
export PGAUDIT_CHECKSUM=d53ef985f2d0b15ba25c512c4ce967dce07b94fd4422c95bd04c4c1a055fe738 \
;; \
"v17") \
export PGAUDIT_VERSION=17.0 \
export PGAUDIT_CHECKSUM=7d0d08d030275d525f36cd48b38c6455f1023da863385badff0cec44965bfd8c \
;; \
*) \
echo "pgaudit is not supported on this PostgreSQL version" && exit 1;; \
esac && \
wget https://github.com/pgaudit/pgaudit/archive/refs/tags/${PGAUDIT_VERSION}.tar.gz -O pgaudit.tar.gz && \
echo "${PGAUDIT_CHECKSUM} pgaudit.tar.gz" | sha256sum --check && \
mkdir pgaudit-src && cd pgaudit-src && tar xzf ../pgaudit.tar.gz --strip-components=1 -C .
FROM pg-build AS pgaudit-build
COPY --from=pgaudit-src /ext-src/ /ext-src/
WORKDIR /ext-src/pgaudit-src
RUN make install USE_PGXS=1 -j $(getconf _NPROCESSORS_ONLN)
#########################################################################################
#
# Layer "pgauditlogtofile"
# compile pgauditlogtofile extension
#
#########################################################################################
FROM build-deps AS pgauditlogtofile-src
ARG PG_VERSION
WORKDIR /ext-src
RUN case "${PG_VERSION}" in \
"v14" | "v15" | "v16" | "v17") \
export PGAUDITLOGTOFILE_VERSION=v1.6.4 \
export PGAUDITLOGTOFILE_CHECKSUM=ef801eb09c26aaa935c0dabd92c81eb9ebe338930daa9674d420a280c6bc2d70 \
;; \
*) \
echo "pgauditlogtofile is not supported on this PostgreSQL version" && exit 1;; \
esac && \
wget https://github.com/fmbiete/pgauditlogtofile/archive/refs/tags/${PGAUDITLOGTOFILE_VERSION}.tar.gz -O pgauditlogtofile.tar.gz && \
echo "${PGAUDITLOGTOFILE_CHECKSUM} pgauditlogtofile.tar.gz" | sha256sum --check && \
mkdir pgauditlogtofile-src && cd pgauditlogtofile-src && tar xzf ../pgauditlogtofile.tar.gz --strip-components=1 -C .
FROM pg-build AS pgauditlogtofile-build
COPY --from=pgauditlogtofile-src /ext-src/ /ext-src/
WORKDIR /ext-src/pgauditlogtofile-src
RUN make install USE_PGXS=1 -j $(getconf _NPROCESSORS_ONLN)
#########################################################################################
#
# Layer "neon-ext-build"
@@ -1556,7 +1669,10 @@ COPY --from=pg_anon-build /usr/local/pgsql/ /usr/local/pgsql/
COPY --from=pg_ivm-build /usr/local/pgsql/ /usr/local/pgsql/
COPY --from=pg_partman-build /usr/local/pgsql/ /usr/local/pgsql/
COPY --from=pg_mooncake-build /usr/local/pgsql/ /usr/local/pgsql/
COPY --from=pg_duckdb-build /usr/local/pgsql/ /usr/local/pgsql/
COPY --from=pg_repack-build /usr/local/pgsql/ /usr/local/pgsql/
COPY --from=pgaudit-build /usr/local/pgsql/ /usr/local/pgsql/
COPY --from=pgauditlogtofile-build /usr/local/pgsql/ /usr/local/pgsql/
#########################################################################################
#
@@ -1578,7 +1694,15 @@ ENV BUILD_TAG=$BUILD_TAG
USER nonroot
# Copy entire project to get Cargo.* files with proper dependencies for the whole project
COPY --chown=nonroot . .
RUN mold -run cargo build --locked --profile release-line-debug-size-lto --bin compute_ctl --bin fast_import --bin local_proxy
RUN --mount=type=cache,uid=1000,target=/home/nonroot/.cargo/registry \
--mount=type=cache,uid=1000,target=/home/nonroot/.cargo/git \
--mount=type=cache,uid=1000,target=/home/nonroot/target \
mold -run cargo build --locked --profile release-line-debug-size-lto --bin compute_ctl --bin fast_import --bin local_proxy && \
mkdir target-bin && \
cp target/release-line-debug-size-lto/compute_ctl \
target/release-line-debug-size-lto/fast_import \
target/release-line-debug-size-lto/local_proxy \
target-bin
#########################################################################################
#
@@ -1586,7 +1710,7 @@ RUN mold -run cargo build --locked --profile release-line-debug-size-lto --bin c
#
#########################################################################################
FROM debian:$DEBIAN_FLAVOR AS pgbouncer
FROM $BASE_IMAGE_SHA AS pgbouncer
RUN set -e \
&& echo 'Acquire::Retries "5";' > /etc/apt/apt.conf.d/80-retries \
&& apt update \
@@ -1607,7 +1731,7 @@ RUN set -e \
&& git clone --recurse-submodules --depth 1 --branch ${PGBOUNCER_TAG} https://github.com/pgbouncer/pgbouncer.git pgbouncer \
&& cd pgbouncer \
&& ./autogen.sh \
&& LDFLAGS=-static ./configure --prefix=/usr/local/pgbouncer --without-openssl \
&& ./configure --prefix=/usr/local/pgbouncer --without-openssl \
&& make -j $(nproc) dist_man_MANS= \
&& make install dist_man_MANS=
@@ -1616,13 +1740,12 @@ RUN set -e \
# Layer "exporters"
#
#########################################################################################
FROM alpine/curl:${ALPINE_CURL_VERSION} AS exporters
FROM build-deps AS exporters
ARG TARGETARCH
# Keep sql_exporter version same as in build-tools.Dockerfile and
# test_runner/regress/test_compute_metrics.py
# See comment on the top of the file regading `echo`, `-e` and `\n`
RUN echo -e "--retry-connrefused\n--connect-timeout 15\n--retry 5\n--max-time 300\n" > /root/.curlrc; \
if [ "$TARGETARCH" = "amd64" ]; then\
RUN if [ "$TARGETARCH" = "amd64" ]; then\
postgres_exporter_sha256='027e75dda7af621237ff8f5ac66b78a40b0093595f06768612b92b1374bd3105';\
pgbouncer_exporter_sha256='c9f7cf8dcff44f0472057e9bf52613d93f3ffbc381ad7547a959daa63c5e84ac';\
sql_exporter_sha256='38e439732bbf6e28ca4a94d7bc3686d3fa1abdb0050773d5617a9efdb9e64d08';\
@@ -1673,7 +1796,7 @@ USER nonroot
COPY --chown=nonroot compute compute
RUN make PG_VERSION="${PG_VERSION}" -C compute
RUN make PG_VERSION="${PG_VERSION:?}" -C compute
#########################################################################################
#
@@ -1699,15 +1822,15 @@ COPY --from=pg_graphql-src /ext-src/ /ext-src/
COPY --from=hypopg-src /ext-src/ /ext-src/
COPY --from=pg_hashids-src /ext-src/ /ext-src/
COPY --from=rum-src /ext-src/ /ext-src/
#COPY --from=pgtap-src /ext-src/ /ext-src/
COPY --from=pgtap-src /ext-src/ /ext-src/
COPY --from=ip4r-src /ext-src/ /ext-src/
COPY --from=prefix-src /ext-src/ /ext-src/
COPY --from=hll-src /ext-src/ /ext-src/
COPY --from=plpgsql_check-src /ext-src/ /ext-src/
#COPY --from=timescaledb-src /ext-src/ /ext-src/
COPY --from=pg_hint_plan-src /ext-src/ /ext-src/
COPY compute/patches/pg_hint_plan_${PG_VERSION}.patch /ext-src
RUN cd /ext-src/pg_hint_plan-src && patch -p1 < /ext-src/pg_hint_plan_${PG_VERSION}.patch
COPY compute/patches/pg_hint_plan_${PG_VERSION:?}.patch /ext-src
RUN cd /ext-src/pg_hint_plan-src && patch -p1 < /ext-src/pg_hint_plan_${PG_VERSION:?}.patch
COPY --from=pg_cron-src /ext-src/ /ext-src/
#COPY --from=pgx_ulid-src /ext-src/ /ext-src/
#COPY --from=pgx_ulid-pgrx12-src /ext-src/ /ext-src/
@@ -1736,51 +1859,12 @@ ENV PGDATABASE=postgres
# Put it all together into the final image
#
#########################################################################################
FROM debian:$DEBIAN_FLAVOR
FROM $BASE_IMAGE_SHA
ARG DEBIAN_VERSION
# Use strict mode for bash to catch errors early
SHELL ["/bin/bash", "-euo", "pipefail", "-c"]
# Add user postgres
RUN mkdir /var/db && useradd -m -d /var/db/postgres postgres && \
echo "postgres:test_console_pass" | chpasswd && \
mkdir /var/db/postgres/compute && mkdir /var/db/postgres/specs && \
mkdir /var/db/postgres/pgbouncer && \
chown -R postgres:postgres /var/db/postgres && \
chmod 0750 /var/db/postgres/compute && \
chmod 0750 /var/db/postgres/pgbouncer && \
echo '/usr/local/lib' >> /etc/ld.so.conf && /sbin/ldconfig && \
# create folder for file cache
mkdir -p -m 777 /neon/cache
COPY --from=postgres-cleanup-layer --chown=postgres /usr/local/pgsql /usr/local
COPY --from=compute-tools --chown=postgres /home/nonroot/target/release-line-debug-size-lto/compute_ctl /usr/local/bin/compute_ctl
COPY --from=compute-tools --chown=postgres /home/nonroot/target/release-line-debug-size-lto/fast_import /usr/local/bin/fast_import
# pgbouncer and its config
COPY --from=pgbouncer /usr/local/pgbouncer/bin/pgbouncer /usr/local/bin/pgbouncer
COPY --chmod=0666 --chown=postgres compute/etc/pgbouncer.ini /etc/pgbouncer.ini
# local_proxy and its config
COPY --from=compute-tools --chown=postgres /home/nonroot/target/release-line-debug-size-lto/local_proxy /usr/local/bin/local_proxy
RUN mkdir -p /etc/local_proxy && chown postgres:postgres /etc/local_proxy
# Metrics exporter binaries and configuration files
COPY --from=exporters ./postgres_exporter /bin/postgres_exporter
COPY --from=exporters ./pgbouncer_exporter /bin/pgbouncer_exporter
COPY --from=exporters ./sql_exporter /bin/sql_exporter
COPY --chown=postgres compute/etc/postgres_exporter.yml /etc/postgres_exporter.yml
COPY --from=sql_exporter_preprocessor --chmod=0644 /home/nonroot/compute/etc/sql_exporter.yml /etc/sql_exporter.yml
COPY --from=sql_exporter_preprocessor --chmod=0644 /home/nonroot/compute/etc/neon_collector.yml /etc/neon_collector.yml
COPY --from=sql_exporter_preprocessor --chmod=0644 /home/nonroot/compute/etc/sql_exporter_autoscaling.yml /etc/sql_exporter_autoscaling.yml
COPY --from=sql_exporter_preprocessor --chmod=0644 /home/nonroot/compute/etc/neon_collector_autoscaling.yml /etc/neon_collector_autoscaling.yml
# Create remote extension download directory
RUN mkdir /usr/local/download_extensions && chown -R postgres:postgres /usr/local/download_extensions
# Install:
# libreadline8 for psql
# liblz4-1 for lz4
@@ -1790,10 +1874,9 @@ RUN mkdir /usr/local/download_extensions && chown -R postgres:postgres /usr/loca
# libzstd1 for zstd
# libboost* for rdkit
# ca-certificates for communicating with s3 by compute_ctl
# libevent for pgbouncer
RUN echo 'Acquire::Retries "5";' > /etc/apt/apt.conf.d/80-retries && \
echo -e "retry_connrefused = on\ntimeout=15\ntries=5\n" > /root/.wgetrc
RUN apt update && \
case $DEBIAN_VERSION in \
# Version-specific installs for Bullseye (PG14-PG16):
@@ -1828,33 +1911,54 @@ RUN apt update && \
libxslt1.1 \
libzstd1 \
libcurl4 \
libevent-2.1-7 \
locales \
procps \
ca-certificates \
curl \
unzip \
$VERSION_INSTALLS && \
apt clean && rm -rf /var/lib/apt/lists/* /tmp/* /var/tmp/* && \
localedef -i en_US -c -f UTF-8 -A /usr/share/locale/locale.alias en_US.UTF-8
# aws cli is used by fast_import (curl and unzip above are at this time only used for this installation step)
ARG TARGETARCH
RUN set -ex; \
if [ "${TARGETARCH}" = "amd64" ]; then \
TARGETARCH_ALT="x86_64"; \
CHECKSUM="c9a9df3770a3ff9259cb469b6179e02829687a464e0824d5c32d378820b53a00"; \
elif [ "${TARGETARCH}" = "arm64" ]; then \
TARGETARCH_ALT="aarch64"; \
CHECKSUM="8181730be7891582b38b028112e81b4899ca817e8c616aad807c9e9d1289223a"; \
else \
echo "Unsupported architecture: ${TARGETARCH}"; exit 1; \
fi; \
curl --retry 5 -L "https://awscli.amazonaws.com/awscli-exe-linux-${TARGETARCH_ALT}-2.17.5.zip" -o /tmp/awscliv2.zip; \
echo "${CHECKSUM} /tmp/awscliv2.zip" | sha256sum -c -; \
unzip /tmp/awscliv2.zip -d /tmp/awscliv2; \
/tmp/awscliv2/aws/install; \
rm -rf /tmp/awscliv2.zip /tmp/awscliv2; \
true
# Add user postgres
RUN mkdir /var/db && useradd -m -d /var/db/postgres postgres && \
echo "postgres:test_console_pass" | chpasswd && \
mkdir /var/db/postgres/compute && mkdir /var/db/postgres/specs && \
mkdir /var/db/postgres/pgbouncer && \
chown -R postgres:postgres /var/db/postgres && \
chmod 0750 /var/db/postgres/compute && \
chmod 0750 /var/db/postgres/pgbouncer && \
# create folder for file cache
mkdir -p -m 777 /neon/cache && \
# Create remote extension download directory
mkdir /usr/local/download_extensions && \
chown -R postgres:postgres /usr/local/download_extensions
# pgbouncer and its config
COPY --from=pgbouncer /usr/local/pgbouncer/bin/pgbouncer /usr/local/bin/pgbouncer
COPY --chmod=0666 --chown=postgres compute/etc/pgbouncer.ini /etc/pgbouncer.ini
COPY --from=postgres-cleanup-layer --chown=postgres /usr/local/pgsql /usr/local
COPY --from=compute-tools --chown=postgres /home/nonroot/target-bin/compute_ctl /usr/local/bin/compute_ctl
COPY --from=compute-tools --chown=postgres /home/nonroot/target-bin/fast_import /usr/local/bin/fast_import
# local_proxy and its config
COPY --from=compute-tools --chown=postgres /home/nonroot/target-bin/local_proxy /usr/local/bin/local_proxy
RUN mkdir -p /etc/local_proxy && chown postgres:postgres /etc/local_proxy
# Metrics exporter binaries and configuration files
COPY --from=exporters ./postgres_exporter /bin/postgres_exporter
COPY --from=exporters ./pgbouncer_exporter /bin/pgbouncer_exporter
COPY --from=exporters ./sql_exporter /bin/sql_exporter
COPY --chown=postgres compute/etc/postgres_exporter.yml /etc/postgres_exporter.yml
COPY --from=sql_exporter_preprocessor --chmod=0644 /home/nonroot/compute/etc/sql_exporter.yml /etc/sql_exporter.yml
COPY --from=sql_exporter_preprocessor --chmod=0644 /home/nonroot/compute/etc/neon_collector.yml /etc/neon_collector.yml
COPY --from=sql_exporter_preprocessor --chmod=0644 /home/nonroot/compute/etc/sql_exporter_autoscaling.yml /etc/sql_exporter_autoscaling.yml
COPY --from=sql_exporter_preprocessor --chmod=0644 /home/nonroot/compute/etc/neon_collector_autoscaling.yml /etc/neon_collector_autoscaling.yml
# Make the libraries we built available
RUN echo '/usr/local/lib' >> /etc/ld.so.conf && /sbin/ldconfig
ENV LANG=en_US.utf8
USER postgres

11
compute/patches/pg_duckdb_v031.patch Normal file
View File

@@ -0,0 +1,11 @@
diff --git a/sql/pg_duckdb--0.2.0--0.3.0.sql b/sql/pg_duckdb--0.2.0--0.3.0.sql
index d777d76..af60106 100644
--- a/sql/pg_duckdb--0.2.0--0.3.0.sql
+++ b/sql/pg_duckdb--0.2.0--0.3.0.sql
@@ -1056,3 +1056,6 @@ GRANT ALL ON FUNCTION duckdb.cache(TEXT, TEXT) TO PUBLIC;
GRANT ALL ON FUNCTION duckdb.cache_info() TO PUBLIC;
GRANT ALL ON FUNCTION duckdb.cache_delete(TEXT) TO PUBLIC;
GRANT ALL ON PROCEDURE duckdb.recycle_ddb() TO PUBLIC;
+GRANT ALL ON FUNCTION duckdb.install_extension(TEXT) TO neon_superuser;
+GRANT ALL ON TABLE duckdb.extensions TO neon_superuser;
+GRANT ALL ON SEQUENCE duckdb.extensions_table_seq TO neon_superuser;

8
compute/patches/pg_hint_plan_v16.patch
View File

@@ -6,16 +6,16 @@ index da723b8..5328114 100644
----
-- No.A-1-1-3
CREATE EXTENSION pg_hint_plan;
+LOG: Sending request to compute_ctl: http://localhost:3080/extension_server/pg_hint_plan
+LOG: Sending request to compute_ctl: http://localhost:3081/extension_server/pg_hint_plan
-- No.A-1-2-3
DROP EXTENSION pg_hint_plan;
-- No.A-1-1-4
CREATE SCHEMA other_schema;
CREATE EXTENSION pg_hint_plan SCHEMA other_schema;
+LOG: Sending request to compute_ctl: http://localhost:3080/extension_server/pg_hint_plan
+LOG: Sending request to compute_ctl: http://localhost:3081/extension_server/pg_hint_plan
ERROR: extension "pg_hint_plan" must be installed in schema "hint_plan"
CREATE EXTENSION pg_hint_plan;
+LOG: Sending request to compute_ctl: http://localhost:3080/extension_server/pg_hint_plan
+LOG: Sending request to compute_ctl: http://localhost:3081/extension_server/pg_hint_plan
DROP SCHEMA other_schema;
----
---- No. A-5-1 comment pattern
@@ -35,7 +35,7 @@ index d372459..6282afe 100644
SET client_min_messages TO LOG;
SET pg_hint_plan.enable_hint TO on;
CREATE EXTENSION file_fdw;
+LOG: Sending request to compute_ctl: http://localhost:3080/extension_server/file_fdw
+LOG: Sending request to compute_ctl: http://localhost:3081/extension_server/file_fdw
CREATE SERVER file_server FOREIGN DATA WRAPPER file_fdw;
CREATE USER MAPPING FOR PUBLIC SERVER file_server;
CREATE FOREIGN TABLE ft1 (id int, val int) SERVER file_server OPTIONS (format 'csv', filename :'filename');

8
compute/patches/pg_hint_plan_v17.patch
View File

@@ -6,16 +6,16 @@ index e7d68a1..65a056c 100644
----
-- No.A-1-1-3
CREATE EXTENSION pg_hint_plan;
+LOG: Sending request to compute_ctl: http://localhost:3080/extension_server/pg_hint_plan
+LOG: Sending request to compute_ctl: http://localhost:3081/extension_server/pg_hint_plan
-- No.A-1-2-3
DROP EXTENSION pg_hint_plan;
-- No.A-1-1-4
CREATE SCHEMA other_schema;
CREATE EXTENSION pg_hint_plan SCHEMA other_schema;
+LOG: Sending request to compute_ctl: http://localhost:3080/extension_server/pg_hint_plan
+LOG: Sending request to compute_ctl: http://localhost:3081/extension_server/pg_hint_plan
ERROR: extension "pg_hint_plan" must be installed in schema "hint_plan"
CREATE EXTENSION pg_hint_plan;
+LOG: Sending request to compute_ctl: http://localhost:3080/extension_server/pg_hint_plan
+LOG: Sending request to compute_ctl: http://localhost:3081/extension_server/pg_hint_plan
DROP SCHEMA other_schema;
----
---- No. A-5-1 comment pattern
@@ -168,7 +168,7 @@ index 017fa4b..98d989b 100644
SET client_min_messages TO LOG;
SET pg_hint_plan.enable_hint TO on;
CREATE EXTENSION file_fdw;
+LOG: Sending request to compute_ctl: http://localhost:3080/extension_server/file_fdw
+LOG: Sending request to compute_ctl: http://localhost:3081/extension_server/file_fdw
CREATE SERVER file_server FOREIGN DATA WRAPPER file_fdw;
CREATE USER MAPPING FOR PUBLIC SERVER file_server;
CREATE FOREIGN TABLE ft1 (id int, val int) SERVER file_server OPTIONS (format 'csv', filename :'filename');

8
compute/vm-image-spec-bookworm.yaml
View File

@@ -47,7 +47,9 @@ files:
# Allow postgres user (which is what compute_ctl runs as) to run /neonvm/bin/resize-swap
# and /neonvm/bin/set-disk-quota as root without requiring entering a password (NOPASSWD),
# regardless of hostname (ALL)
postgres ALL=(root) NOPASSWD: /neonvm/bin/resize-swap, /neonvm/bin/set-disk-quota
#
# Also allow it to shut down the VM. The fast_import job does that when it's finished.
postgres ALL=(root) NOPASSWD: /neonvm/bin/resize-swap, /neonvm/bin/set-disk-quota, /neonvm/bin/poweroff
- filename: cgconfig.conf
content: |
# Configuration for cgroups in VM compute nodes
@@ -72,8 +74,8 @@ build: |
# At time of migration to bookworm (2024-10-09), debian has a version of libcgroup/cgroup-tools 2.0.2,
# and it _probably_ can be used as-is. However, we'll build it ourselves to minimise the changeset
# for debian version migration.
#
FROM debian:bookworm-slim as libcgroup-builder
ARG BOOKWORM_SLIM_SHA=sha256:40b107342c492725bc7aacbe93a49945445191ae364184a6d24fedb28172f6f7
FROM debian@$BOOKWORM_SLIM_SHA as libcgroup-builder
ENV LIBCGROUP_VERSION=v2.0.3
RUN set -exu \

7
compute/vm-image-spec-bullseye.yaml
View File

@@ -47,7 +47,9 @@ files:
# Allow postgres user (which is what compute_ctl runs as) to run /neonvm/bin/resize-swap
# and /neonvm/bin/set-disk-quota as root without requiring entering a password (NOPASSWD),
# regardless of hostname (ALL)
postgres ALL=(root) NOPASSWD: /neonvm/bin/resize-swap, /neonvm/bin/set-disk-quota
#
# Also allow it to shut down the VM. The fast_import job does that when it's finished.
postgres ALL=(root) NOPASSWD: /neonvm/bin/resize-swap, /neonvm/bin/set-disk-quota, /neonvm/bin/poweroff
- filename: cgconfig.conf
content: |
# Configuration for cgroups in VM compute nodes
@@ -68,7 +70,8 @@ build: |
# At time of writing (2023-03-14), debian bullseye has a version of cgroup-tools (technically
# libcgroup) that doesn't support cgroup v2 (version 0.41-11). Unfortunately, the vm-monitor
# requires cgroup v2, so we'll build cgroup-tools ourselves.
FROM debian:bullseye-slim as libcgroup-builder
ARG BULLSEYE_SLIM_SHA=sha256:e831d9a884d63734fe3dd9c491ed9a5a3d4c6a6d32c5b14f2067357c49b0b7e1
FROM debian@$BULLSEYE_SLIM_SHA as libcgroup-builder
ENV LIBCGROUP_VERSION=v2.0.3
RUN set -exu \

3
compute_tools/Cargo.toml
View File

@@ -14,6 +14,7 @@ base64.workspace = true
aws-config.workspace = true
aws-sdk-s3.workspace = true
aws-sdk-kms.workspace = true
aws-smithy-types.workspace = true
anyhow.workspace = true
axum = { workspace = true, features = [] }
camino.workspace = true
@@ -24,6 +25,7 @@ fail.workspace = true
flate2.workspace = true
futures.workspace = true
http.workspace = true
jsonwebtoken.workspace = true
metrics.workspace = true
nix.workspace = true
notify.workspace = true
@@ -53,6 +55,7 @@ thiserror.workspace = true
url.workspace = true
uuid.workspace = true
prometheus.workspace = true
walkdir.workspace = true
postgres_initdb.workspace = true
compute_api.workspace = true

139
compute_tools/src/bin/compute_ctl.rs
View File

@@ -41,19 +41,21 @@ use std::process::exit;
use std::str::FromStr;
use std::sync::atomic::Ordering;
use std::sync::{mpsc, Arc, Condvar, Mutex, RwLock};
use std::time::SystemTime;
use std::{thread, time::Duration};
use anyhow::{Context, Result};
use chrono::Utc;
use clap::Parser;
use compute_tools::disk_quota::set_disk_quota;
use compute_tools::http::server::Server;
use compute_tools::lsn_lease::launch_lsn_lease_bg_task_for_static;
use signal_hook::consts::{SIGQUIT, SIGTERM};
use signal_hook::{consts::SIGINT, iterator::Signals};
use tracing::{error, info, warn};
use url::Url;
use compute_api::responses::ComputeStatus;
use compute_api::responses::{ComputeCtlConfig, ComputeStatus};
use compute_api::spec::ComputeSpec;
use compute_tools::compute::{
@@ -61,7 +63,6 @@ use compute_tools::compute::{
};
use compute_tools::configurator::launch_configurator;
use compute_tools::extension_server::get_pg_version_string;
use compute_tools::http::launch_http_server;
use compute_tools::logger::*;
use compute_tools::monitor::launch_monitor;
use compute_tools::params::*;
@@ -85,6 +86,19 @@ fn parse_remote_ext_config(arg: &str) -> Result<String> {
}
}
/// Generate a compute ID if one is not supplied. This exists to keep forward
/// compatibility tests working, but will be removed in a future iteration.
fn generate_compute_id() -> String {
let now = SystemTime::now();
format!(
"compute-{}",
now.duration_since(SystemTime::UNIX_EPOCH)
.unwrap()
.as_secs()
)
}
#[derive(Parser)]
#[command(rename_all = "kebab-case")]
struct Cli {
@@ -94,8 +108,20 @@ struct Cli {
#[arg(short = 'r', long, value_parser = parse_remote_ext_config)]
pub remote_ext_config: Option<String>,
#[arg(long, default_value_t = 3080)]
pub http_port: u16,
/// The port to bind the external listening HTTP server to. Clients running
/// outside the compute will talk to the compute through this port. Keep
/// the previous name for this argument around for a smoother release
/// with the control plane.
///
/// TODO: Remove the alias after the control plane release which teaches the
/// control plane about the renamed argument.
#[arg(long, alias = "http-port", default_value_t = 3080)]
pub external_http_port: u16,
/// The port to bind the internal listening HTTP server to. Clients like
/// the neon extension (for installing remote extensions) and local_proxy.
#[arg(long)]
pub internal_http_port: Option<u16>,
#[arg(short = 'D', long, value_name = "DATADIR")]
pub pgdata: String,
@@ -130,17 +156,26 @@ struct Cli {
#[arg(short = 'S', long, group = "spec-path")]
pub spec_path: Option<OsString>,
#[arg(short = 'i', long, group = "compute-id", conflicts_with_all = ["spec", "spec-path"])]
pub compute_id: Option<String>,
#[arg(short = 'i', long, group = "compute-id", default_value = generate_compute_id())]
pub compute_id: String,
#[arg(short = 'p', long, conflicts_with_all = ["spec", "spec-path"], requires = "compute-id", value_name = "CONTROL_PLANE_API_BASE_URL")]
#[arg(short = 'p', long, conflicts_with_all = ["spec", "spec-path"], value_name = "CONTROL_PLANE_API_BASE_URL")]
pub control_plane_uri: Option<String>,
}
fn main() -> Result<()> {
let cli = Cli::parse();
let build_tag = init()?;
// For historical reasons, the main thread that processes the spec and launches postgres
// is synchronous, but we always have this tokio runtime available and we "enter" it so
// that you can use tokio::spawn() and tokio::runtime::Handle::current().block_on(...)
// from all parts of compute_ctl.
let runtime = tokio::runtime::Builder::new_multi_thread()
.enable_all()
.build()?;
let _rt_guard = runtime.enter();
let build_tag = runtime.block_on(init())?;
let scenario = failpoint_support::init();
@@ -172,8 +207,8 @@ fn main() -> Result<()> {
deinit_and_exit(wait_pg_result);
}
fn init() -> Result<String> {
init_tracing_and_logging(DEFAULT_LOG_LEVEL)?;
async fn init() -> Result<String> {
init_tracing_and_logging(DEFAULT_LOG_LEVEL).await?;
let mut signals = Signals::new([SIGINT, SIGTERM, SIGQUIT])?;
thread::spawn(move || {
@@ -246,6 +281,7 @@ fn try_spec_from_cli(cli: &Cli) -> Result<CliSpecParams> {
info!("got spec from cli argument {}", spec_json);
return Ok(CliSpecParams {
spec: Some(serde_json::from_str(spec_json)?),
compute_ctl_config: ComputeCtlConfig::default(),
live_config_allowed: false,
});
}
@@ -255,26 +291,19 @@ fn try_spec_from_cli(cli: &Cli) -> Result<CliSpecParams> {
let file = File::open(Path::new(spec_path))?;
return Ok(CliSpecParams {
spec: Some(serde_json::from_reader(file)?),
compute_ctl_config: ComputeCtlConfig::default(),
live_config_allowed: true,
});
}
if cli.compute_id.is_none() {
panic!(
"compute spec should be provided by one of the following ways: \
--spec OR --spec-path OR --control-plane-uri and --compute-id"
);
};
if cli.control_plane_uri.is_none() {
panic!("must specify both --control-plane-uri and --compute-id or none");
panic!("must specify --control-plane-uri");
};
match get_spec_from_control_plane(
cli.control_plane_uri.as_ref().unwrap(),
cli.compute_id.as_ref().unwrap(),
) {
Ok(spec) => Ok(CliSpecParams {
spec,
match get_spec_from_control_plane(cli.control_plane_uri.as_ref().unwrap(), &cli.compute_id) {
Ok(resp) => Ok(CliSpecParams {
spec: resp.0,
compute_ctl_config: resp.1,
live_config_allowed: true,
}),
Err(e) => {
@@ -291,6 +320,8 @@ fn try_spec_from_cli(cli: &Cli) -> Result<CliSpecParams> {
struct CliSpecParams {
/// If a spec was provided via CLI or file, the [`ComputeSpec`]
spec: Option<ComputeSpec>,
#[allow(dead_code)]
compute_ctl_config: ComputeCtlConfig,
live_config_allowed: bool,
}
@@ -300,6 +331,7 @@ fn wait_spec(
CliSpecParams {
spec,
live_config_allowed,
compute_ctl_config: _,
}: CliSpecParams,
) -> Result<Arc<ComputeNode>> {
let mut new_state = ComputeState::new();
@@ -319,13 +351,15 @@ fn wait_spec(
let tokio_conn_conf = tokio_postgres::config::Config::from_str(connstr.as_str())
.context("cannot build tokio postgres config from connstr")?;
let compute_node = ComputeNode {
compute_id: cli.compute_id.clone(),
connstr,
conn_conf,
tokio_conn_conf,
pgdata: cli.pgdata.clone(),
pgbin: cli.pgbin.clone(),
pgversion: get_pg_version_string(&cli.pgbin),
http_port: cli.http_port,
external_http_port: cli.external_http_port,
internal_http_port: cli.internal_http_port.unwrap_or(cli.external_http_port + 1),
live_config_allowed,
state: Mutex::new(new_state),
state_changed: Condvar::new(),
@@ -343,10 +377,13 @@ fn wait_spec(
compute.prewarm_postgres()?;
}
// Launch http service first, so that we can serve control-plane requests
// while configuration is still in progress.
let _http_handle =
launch_http_server(cli.http_port, &compute).expect("cannot launch http endpoint thread");
// Launch the external HTTP server first, so that we can serve control plane
// requests while configuration is still in progress.
Server::External(cli.external_http_port).launch(&compute);
// The internal HTTP server could be launched later, but there isn't much
// sense in waiting.
Server::Internal(cli.internal_http_port.unwrap_or(cli.external_http_port + 1)).launch(&compute);
if !spec_set {
// No spec provided, hang waiting for it.
@@ -484,21 +521,6 @@ fn start_postgres(
use std::env;
use tokio_util::sync::CancellationToken;
// Note: it seems like you can make a runtime in an inner scope and
// if you start a task in it it won't be dropped. However, make it
// in the outermost scope just to be safe.
let rt = if env::var_os("AUTOSCALING").is_some() {
Some(
tokio::runtime::Builder::new_multi_thread()
.worker_threads(4)
.enable_all()
.build()
.expect("failed to create tokio runtime for monitor")
)
} else {
None
};
// This token is used internally by the monitor to clean up all threads
let token = CancellationToken::new();
@@ -509,16 +531,19 @@ fn start_postgres(
Some(cli.filecache_connstr.clone())
};
let vm_monitor = rt.as_ref().map(|rt| {
rt.spawn(vm_monitor::start(
let vm_monitor = if env::var_os("AUTOSCALING").is_some() {
let vm_monitor = tokio::spawn(vm_monitor::start(
Box::leak(Box::new(vm_monitor::Args {
cgroup: Some(cli.cgroup.clone()),
pgconnstr,
addr: cli.vm_monitor_addr.clone(),
})),
token.clone(),
))
});
));
Some(vm_monitor)
} else {
None
};
}
}
@@ -528,8 +553,6 @@ fn start_postgres(
delay_exit,
compute,
#[cfg(target_os = "linux")]
rt,
#[cfg(target_os = "linux")]
token,
#[cfg(target_os = "linux")]
vm_monitor,
@@ -537,15 +560,13 @@ fn start_postgres(
))
}
type PostgresHandle = (std::process::Child, std::thread::JoinHandle<()>);
type PostgresHandle = (std::process::Child, tokio::task::JoinHandle<Result<()>>);
struct StartPostgresResult {
delay_exit: bool,
// passed through from WaitSpecResult
compute: Arc<ComputeNode>,
#[cfg(target_os = "linux")]
rt: Option<tokio::runtime::Runtime>,
#[cfg(target_os = "linux")]
token: tokio_util::sync::CancellationToken,
#[cfg(target_os = "linux")]
@@ -564,10 +585,10 @@ fn wait_postgres(pg: Option<PostgresHandle>) -> Result<WaitPostgresResult> {
.expect("failed to start waiting on Postgres process");
PG_PID.store(0, Ordering::SeqCst);
// Process has exited, so we can join the logs thread.
let _ = logs_handle
.join()
.map_err(|e| tracing::error!("log thread panicked: {:?}", e));
// Process has exited. Wait for the log collecting task to finish.
let _ = tokio::runtime::Handle::current()
.block_on(logs_handle)
.map_err(|e| tracing::error!("log task panicked: {:?}", e));
info!("Postgres exited with code {}, shutting down", ecode);
exit_code = ecode.code()
@@ -588,8 +609,6 @@ fn cleanup_after_postgres_exit(
vm_monitor,
#[cfg(target_os = "linux")]
token,
#[cfg(target_os = "linux")]
rt,
}: StartPostgresResult,
) -> Result<bool> {
// Terminate the vm_monitor so it releases the file watcher on
@@ -602,10 +621,6 @@ fn cleanup_after_postgres_exit(
token.cancel();
// Kills the actual task running the monitor
handle.abort();
// If handle is some, rt must have been used to produce it, and
// hence is also some
rt.unwrap().shutdown_timeout(Duration::from_secs(2));
}
}
}

713
compute_tools/src/bin/fast_import.rs
View File

@@ -25,10 +25,10 @@
//! docker push localhost:3030/localregistry/compute-node-v14:latest
//! ```
use anyhow::Context;
use anyhow::{bail, Context};
use aws_config::BehaviorVersion;
use camino::{Utf8Path, Utf8PathBuf};
use clap::Parser;
use clap::{Parser, Subcommand};
use compute_tools::extension_server::{get_pg_version, PostgresMajorVersion};
use nix::unistd::Pid;
use tracing::{error, info, info_span, warn, Instrument};
@@ -44,22 +44,59 @@ mod s3_uri;
const PG_WAIT_TIMEOUT: std::time::Duration = std::time::Duration::from_secs(600);
const PG_WAIT_RETRY_INTERVAL: std::time::Duration = std::time::Duration::from_millis(300);
#[derive(Subcommand, Debug)]
enum Command {
/// Runs local postgres (neon binary), restores into it,
/// uploads pgdata to s3 to be consumed by pageservers
Pgdata {
/// Raw connection string to the source database. Used only in tests,
/// real scenario uses encrypted connection string in spec.json from s3.
#[clap(long)]
source_connection_string: Option<String>,
/// If specified, will not shut down the local postgres after the import. Used in local testing
#[clap(short, long)]
interactive: bool,
/// Port to run postgres on. Default is 5432.
#[clap(long, default_value_t = 5432)]
pg_port: u16, // port to run postgres on, 5432 is default
/// Number of CPUs in the system. This is used to configure # of
/// parallel worker processes, for index creation.
#[clap(long, env = "NEON_IMPORTER_NUM_CPUS")]
num_cpus: Option<usize>,
/// Amount of RAM in the system. This is used to configure shared_buffers
/// and maintenance_work_mem.
#[clap(long, env = "NEON_IMPORTER_MEMORY_MB")]
memory_mb: Option<usize>,
},
/// Runs pg_dump-pg_restore from source to destination without running local postgres.
DumpRestore {
/// Raw connection string to the source database. Used only in tests,
/// real scenario uses encrypted connection string in spec.json from s3.
#[clap(long)]
source_connection_string: Option<String>,
/// Raw connection string to the destination database. Used only in tests,
/// real scenario uses encrypted connection string in spec.json from s3.
#[clap(long)]
destination_connection_string: Option<String>,
},
}
#[derive(clap::Parser)]
struct Args {
#[clap(long)]
#[clap(long, env = "NEON_IMPORTER_WORKDIR")]
working_directory: Utf8PathBuf,
#[clap(long, env = "NEON_IMPORTER_S3_PREFIX")]
s3_prefix: Option<s3_uri::S3Uri>,
#[clap(long)]
source_connection_string: Option<String>,
#[clap(short, long)]
interactive: bool,
#[clap(long)]
#[clap(long, env = "NEON_IMPORTER_PG_BIN_DIR")]
pg_bin_dir: Utf8PathBuf,
#[clap(long)]
#[clap(long, env = "NEON_IMPORTER_PG_LIB_DIR")]
pg_lib_dir: Utf8PathBuf,
#[clap(long)]
pg_port: Option<u16>, // port to run postgres on, 5432 is default
#[clap(subcommand)]
command: Command,
}
#[serde_with::serde_as]
@@ -68,6 +105,8 @@ struct Spec {
encryption_secret: EncryptionSecret,
#[serde_as(as = "serde_with::base64::Base64")]
source_connstring_ciphertext_base64: Vec<u8>,
#[serde_as(as = "Option<serde_with::base64::Base64>")]
destination_connstring_ciphertext_base64: Option<Vec<u8>>,
}
#[derive(serde::Deserialize)]
@@ -83,180 +122,150 @@ const DEFAULT_LOCALE: &str = if cfg!(target_os = "macos") {
"C.UTF-8"
};
#[tokio::main]
pub(crate) async fn main() -> anyhow::Result<()> {
utils::logging::init(
utils::logging::LogFormat::Plain,
utils::logging::TracingErrorLayerEnablement::EnableWithRustLogFilter,
utils::logging::Output::Stdout,
)?;
info!("starting");
let args = Args::parse();
// Validate arguments
if args.s3_prefix.is_none() && args.source_connection_string.is_none() {
anyhow::bail!("either s3_prefix or source_connection_string must be specified");
}
if args.s3_prefix.is_some() && args.source_connection_string.is_some() {
anyhow::bail!("only one of s3_prefix or source_connection_string can be specified");
}
let working_directory = args.working_directory;
let pg_bin_dir = args.pg_bin_dir;
let pg_lib_dir = args.pg_lib_dir;
let pg_port = args.pg_port.unwrap_or_else(|| {
info!("pg_port not specified, using default 5432");
5432
});
// Initialize AWS clients only if s3_prefix is specified
let (aws_config, kms_client) = if args.s3_prefix.is_some() {
let config = aws_config::load_defaults(BehaviorVersion::v2024_03_28()).await;
let kms = aws_sdk_kms::Client::new(&config);
(Some(config), Some(kms))
} else {
(None, None)
};
// Get source connection string either from S3 spec or direct argument
let source_connection_string = if let Some(s3_prefix) = &args.s3_prefix {
let spec: Spec = {
let spec_key = s3_prefix.append("/spec.json");
let s3_client = aws_sdk_s3::Client::new(aws_config.as_ref().unwrap());
let object = s3_client
.get_object()
.bucket(&spec_key.bucket)
.key(spec_key.key)
.send()
.await
.context("get spec from s3")?
.body
.collect()
.await
.context("download spec body")?;
serde_json::from_slice(&object.into_bytes()).context("parse spec as json")?
};
match spec.encryption_secret {
EncryptionSecret::KMS { key_id } => {
let mut output = kms_client
.unwrap()
.decrypt()
.key_id(key_id)
.ciphertext_blob(aws_sdk_s3::primitives::Blob::new(
spec.source_connstring_ciphertext_base64,
))
.send()
.await
.context("decrypt source connection string")?;
let plaintext = output
.plaintext
.take()
.context("get plaintext source connection string")?;
String::from_utf8(plaintext.into_inner())
.context("parse source connection string as utf8")?
}
}
} else {
args.source_connection_string.unwrap()
};
match tokio::fs::create_dir(&working_directory).await {
Ok(()) => {}
Err(e) if e.kind() == std::io::ErrorKind::AlreadyExists => {
if !is_directory_empty(&working_directory)
.await
.context("check if working directory is empty")?
{
anyhow::bail!("working directory is not empty");
} else {
// ok
}
}
Err(e) => return Err(anyhow::Error::new(e).context("create working directory")),
}
let pgdata_dir = working_directory.join("pgdata");
tokio::fs::create_dir(&pgdata_dir)
async fn decode_connstring(
kms_client: &aws_sdk_kms::Client,
key_id: &String,
connstring_ciphertext_base64: Vec<u8>,
) -> Result<String, anyhow::Error> {
let mut output = kms_client
.decrypt()
.key_id(key_id)
.ciphertext_blob(aws_sdk_s3::primitives::Blob::new(
connstring_ciphertext_base64,
))
.send()
.await
.context("create pgdata directory")?;
.context("decrypt connection string")?;
let pgbin = pg_bin_dir.join("postgres");
let pg_version = match get_pg_version(pgbin.as_ref()) {
PostgresMajorVersion::V14 => 14,
PostgresMajorVersion::V15 => 15,
PostgresMajorVersion::V16 => 16,
PostgresMajorVersion::V17 => 17,
};
let superuser = "cloud_admin"; // XXX: this shouldn't be hard-coded
postgres_initdb::do_run_initdb(postgres_initdb::RunInitdbArgs {
superuser,
locale: DEFAULT_LOCALE, // XXX: this shouldn't be hard-coded,
pg_version,
initdb_bin: pg_bin_dir.join("initdb").as_ref(),
library_search_path: &pg_lib_dir, // TODO: is this right? Prob works in compute image, not sure about neon_local.
pgdata: &pgdata_dir,
})
.await
.context("initdb")?;
let plaintext = output
.plaintext
.take()
.context("get plaintext connection string")?;
let nproc = num_cpus::get();
String::from_utf8(plaintext.into_inner()).context("parse connection string as utf8")
}
//
// Launch postgres process
//
let mut postgres_proc = tokio::process::Command::new(pgbin)
.arg("-D")
.arg(&pgdata_dir)
.args(["-p", &format!("{pg_port}")])
.args(["-c", "wal_level=minimal"])
.args(["-c", "shared_buffers=10GB"])
.args(["-c", "max_wal_senders=0"])
.args(["-c", "fsync=off"])
.args(["-c", "full_page_writes=off"])
.args(["-c", "synchronous_commit=off"])
.args(["-c", "maintenance_work_mem=8388608"])
.args(["-c", &format!("max_parallel_maintenance_workers={nproc}")])
.args(["-c", &format!("max_parallel_workers={nproc}")])
.args(["-c", &format!("max_parallel_workers_per_gather={nproc}")])
.args(["-c", &format!("max_worker_processes={nproc}")])
.args([
"-c",
&format!(
"effective_io_concurrency={}",
if cfg!(target_os = "macos") { 0 } else { 100 }
),
])
.env_clear()
.env("LD_LIBRARY_PATH", &pg_lib_dir)
.env(
"ASAN_OPTIONS",
std::env::var("ASAN_OPTIONS").unwrap_or_default(),
struct PostgresProcess {
pgdata_dir: Utf8PathBuf,
pg_bin_dir: Utf8PathBuf,
pgbin: Utf8PathBuf,
pg_lib_dir: Utf8PathBuf,
postgres_proc: Option<tokio::process::Child>,
}
impl PostgresProcess {
fn new(pgdata_dir: Utf8PathBuf, pg_bin_dir: Utf8PathBuf, pg_lib_dir: Utf8PathBuf) -> Self {
Self {
pgdata_dir,
pgbin: pg_bin_dir.join("postgres"),
pg_bin_dir,
pg_lib_dir,
postgres_proc: None,
}
}
async fn prepare(&self, initdb_user: &str) -> Result<(), anyhow::Error> {
tokio::fs::create_dir(&self.pgdata_dir)
.await
.context("create pgdata directory")?;
let pg_version = match get_pg_version(self.pgbin.as_ref()) {
PostgresMajorVersion::V14 => 14,
PostgresMajorVersion::V15 => 15,
PostgresMajorVersion::V16 => 16,
PostgresMajorVersion::V17 => 17,
};
postgres_initdb::do_run_initdb(postgres_initdb::RunInitdbArgs {
superuser: initdb_user,
locale: DEFAULT_LOCALE, // XXX: this shouldn't be hard-coded,
pg_version,
initdb_bin: self.pg_bin_dir.join("initdb").as_ref(),
library_search_path: &self.pg_lib_dir, // TODO: is this right? Prob works in compute image, not sure about neon_local.
pgdata: &self.pgdata_dir,
})
.await
.context("initdb")
}
async fn start(
&mut self,
initdb_user: &str,
port: u16,
nproc: usize,
memory_mb: usize,
) -> Result<&tokio::process::Child, anyhow::Error> {
self.prepare(initdb_user).await?;
// Somewhat arbitrarily, use 10 % of memory for shared buffer cache, 70% for
// maintenance_work_mem (i.e. for sorting during index creation), and leave the rest
// available for misc other stuff that PostgreSQL uses memory for.
let shared_buffers_mb = ((memory_mb as f32) * 0.10) as usize;
let maintenance_work_mem_mb = ((memory_mb as f32) * 0.70) as usize;
//
// Launch postgres process
//
let mut proc = tokio::process::Command::new(&self.pgbin)
.arg("-D")
.arg(&self.pgdata_dir)
.args(["-p", &format!("{port}")])
.args(["-c", "wal_level=minimal"])
.args(["-c", &format!("shared_buffers={shared_buffers_mb}MB")])
.args(["-c", "max_wal_senders=0"])
.args(["-c", "fsync=off"])
.args(["-c", "full_page_writes=off"])
.args(["-c", "synchronous_commit=off"])
.args([
"-c",
&format!("maintenance_work_mem={maintenance_work_mem_mb}MB"),
])
.args(["-c", &format!("max_parallel_maintenance_workers={nproc}")])
.args(["-c", &format!("max_parallel_workers={nproc}")])
.args(["-c", &format!("max_parallel_workers_per_gather={nproc}")])
.args(["-c", &format!("max_worker_processes={nproc}")])
.args(["-c", "effective_io_concurrency=100"])
.env_clear()
.env("LD_LIBRARY_PATH", &self.pg_lib_dir)
.env(
"ASAN_OPTIONS",
std::env::var("ASAN_OPTIONS").unwrap_or_default(),
)
.env(
"UBSAN_OPTIONS",
std::env::var("UBSAN_OPTIONS").unwrap_or_default(),
)
.stdout(std::process::Stdio::piped())
.stderr(std::process::Stdio::piped())
.spawn()
.context("spawn postgres")?;
info!("spawned postgres, waiting for it to become ready");
tokio::spawn(
child_stdio_to_log::relay_process_output(proc.stdout.take(), proc.stderr.take())
.instrument(info_span!("postgres")),
);
self.postgres_proc = Some(proc);
Ok(self.postgres_proc.as_ref().unwrap())
}
async fn shutdown(&mut self) -> Result<(), anyhow::Error> {
let proc: &mut tokio::process::Child = self.postgres_proc.as_mut().unwrap();
info!("shutdown postgres");
nix::sys::signal::kill(
Pid::from_raw(i32::try_from(proc.id().unwrap()).expect("convert child pid to i32")),
nix::sys::signal::SIGTERM,
)
.env(
"UBSAN_OPTIONS",
std::env::var("UBSAN_OPTIONS").unwrap_or_default(),
)
.stdout(std::process::Stdio::piped())
.stderr(std::process::Stdio::piped())
.spawn()
.context("spawn postgres")?;
info!("spawned postgres, waiting for it to become ready");
tokio::spawn(
child_stdio_to_log::relay_process_output(
postgres_proc.stdout.take(),
postgres_proc.stderr.take(),
)
.instrument(info_span!("postgres")),
);
.context("signal postgres to shut down")?;
proc.wait()
.await
.context("wait for postgres to shut down")
.map(|_| ())
}
}
async fn wait_until_ready(connstring: String, create_dbname: String) {
// Create neondb database in the running postgres
let restore_pg_connstring =
format!("host=localhost port={pg_port} user={superuser} dbname=postgres");
let start_time = std::time::Instant::now();
loop {
@@ -267,7 +276,12 @@ pub(crate) async fn main() -> anyhow::Result<()> {
std::process::exit(1);
}
match tokio_postgres::connect(&restore_pg_connstring, tokio_postgres::NoTls).await {
match tokio_postgres::connect(
&connstring.replace("dbname=neondb", "dbname=postgres"),
tokio_postgres::NoTls,
)
.await
{
Ok((client, connection)) => {
// Spawn the connection handling task to maintain the connection
tokio::spawn(async move {
@@ -276,9 +290,12 @@ pub(crate) async fn main() -> anyhow::Result<()> {
}
});
match client.simple_query("CREATE DATABASE neondb;").await {
match client
.simple_query(format!("CREATE DATABASE {create_dbname};").as_str())
.await
{
Ok(_) => {
info!("created neondb database");
info!("created {} database", create_dbname);
break;
}
Err(e) => {
@@ -302,10 +319,16 @@ pub(crate) async fn main() -> anyhow::Result<()> {
}
}
}
}
let restore_pg_connstring = restore_pg_connstring.replace("dbname=postgres", "dbname=neondb");
let dumpdir = working_directory.join("dumpdir");
async fn run_dump_restore(
workdir: Utf8PathBuf,
pg_bin_dir: Utf8PathBuf,
pg_lib_dir: Utf8PathBuf,
source_connstring: String,
destination_connstring: String,
) -> Result<(), anyhow::Error> {
let dumpdir = workdir.join("dumpdir");
let common_args = [
// schema mapping (prob suffices to specify them on one side)
@@ -334,10 +357,18 @@ pub(crate) async fn main() -> anyhow::Result<()> {
.arg("--no-sync")
// POSITIONAL args
// source db (db name included in connection string)
.arg(&source_connection_string)
.arg(&source_connstring)
// how we run it
.env_clear()
.env("LD_LIBRARY_PATH", &pg_lib_dir)
.env(
"ASAN_OPTIONS",
std::env::var("ASAN_OPTIONS").unwrap_or_default(),
)
.env(
"UBSAN_OPTIONS",
std::env::var("UBSAN_OPTIONS").unwrap_or_default(),
)
.kill_on_drop(true)
.stdout(std::process::Stdio::piped())
.stderr(std::process::Stdio::piped())
@@ -354,24 +385,31 @@ pub(crate) async fn main() -> anyhow::Result<()> {
let st = pg_dump.wait().await.context("wait for pg_dump")?;
info!(status=?st, "pg_dump exited");
if !st.success() {
warn!(status=%st, "pg_dump failed, restore will likely fail as well");
error!(status=%st, "pg_dump failed, restore will likely fail as well");
bail!("pg_dump failed");
}
}
// TODO: do it in a streaming way, plenty of internal research done on this already
// TODO: maybe do it in a streaming way, plenty of internal research done on this already
// TODO: do the unlogged table trick
info!("restore from working directory into vanilla postgres");
{
let mut pg_restore = tokio::process::Command::new(pg_bin_dir.join("pg_restore"))
.args(&common_args)
.arg("-d")
.arg(&restore_pg_connstring)
.arg(&destination_connstring)
// POSITIONAL args
.arg(&dumpdir)
// how we run it
.env_clear()
.env("LD_LIBRARY_PATH", &pg_lib_dir)
.env(
"ASAN_OPTIONS",
std::env::var("ASAN_OPTIONS").unwrap_or_default(),
)
.env(
"UBSAN_OPTIONS",
std::env::var("UBSAN_OPTIONS").unwrap_or_default(),
)
.kill_on_drop(true)
.stdout(std::process::Stdio::piped())
.stderr(std::process::Stdio::piped())
@@ -389,48 +427,259 @@ pub(crate) async fn main() -> anyhow::Result<()> {
let st = pg_restore.wait().await.context("wait for pg_restore")?;
info!(status=?st, "pg_restore exited");
if !st.success() {
warn!(status=%st, "pg_restore failed, restore will likely fail as well");
}
}
// If interactive mode, wait for Ctrl+C
if args.interactive {
info!("Running in interactive mode. Press Ctrl+C to shut down.");
tokio::signal::ctrl_c().await.context("wait for ctrl-c")?;
}
info!("shutdown postgres");
{
nix::sys::signal::kill(
Pid::from_raw(
i32::try_from(postgres_proc.id().unwrap()).expect("convert child pid to i32"),
),
nix::sys::signal::SIGTERM,
)
.context("signal postgres to shut down")?;
postgres_proc
.wait()
.await
.context("wait for postgres to shut down")?;
}
// Only sync if s3_prefix was specified
if let Some(s3_prefix) = args.s3_prefix {
info!("upload pgdata");
aws_s3_sync::sync(Utf8Path::new(&pgdata_dir), &s3_prefix.append("/pgdata/"))
.await
.context("sync dump directory to destination")?;
info!("write status");
{
let status_dir = working_directory.join("status");
std::fs::create_dir(&status_dir).context("create status directory")?;
let status_file = status_dir.join("pgdata");
std::fs::write(&status_file, serde_json::json!({"done": true}).to_string())
.context("write status file")?;
aws_s3_sync::sync(&status_dir, &s3_prefix.append("/status/"))
.await
.context("sync status directory to destination")?;
error!(status=%st, "pg_restore failed, restore will likely fail as well");
bail!("pg_restore failed");
}
}
Ok(())
}
#[allow(clippy::too_many_arguments)]
async fn cmd_pgdata(
s3_client: Option<aws_sdk_s3::Client>,
kms_client: Option<aws_sdk_kms::Client>,
maybe_s3_prefix: Option<s3_uri::S3Uri>,
maybe_spec: Option<Spec>,
source_connection_string: Option<String>,
interactive: bool,
pg_port: u16,
workdir: Utf8PathBuf,
pg_bin_dir: Utf8PathBuf,
pg_lib_dir: Utf8PathBuf,
num_cpus: Option<usize>,
memory_mb: Option<usize>,
) -> Result<(), anyhow::Error> {
if maybe_spec.is_none() && source_connection_string.is_none() {
bail!("spec must be provided for pgdata command");
}
if maybe_spec.is_some() && source_connection_string.is_some() {
bail!("only one of spec or source_connection_string can be provided");
}
let source_connection_string = if let Some(spec) = maybe_spec {
match spec.encryption_secret {
EncryptionSecret::KMS { key_id } => {
decode_connstring(
kms_client.as_ref().unwrap(),
&key_id,
spec.source_connstring_ciphertext_base64,
)
.await?
}
}
} else {
source_connection_string.unwrap()
};
let superuser = "cloud_admin";
let destination_connstring = format!(
"host=localhost port={} user={} dbname=neondb",
pg_port, superuser
);
let pgdata_dir = workdir.join("pgdata");
let mut proc = PostgresProcess::new(pgdata_dir.clone(), pg_bin_dir.clone(), pg_lib_dir.clone());
let nproc = num_cpus.unwrap_or_else(num_cpus::get);
let memory_mb = memory_mb.unwrap_or(256);
proc.start(superuser, pg_port, nproc, memory_mb).await?;
wait_until_ready(destination_connstring.clone(), "neondb".to_string()).await;
run_dump_restore(
workdir.clone(),
pg_bin_dir,
pg_lib_dir,
source_connection_string,
destination_connstring,
)
.await?;
// If interactive mode, wait for Ctrl+C
if interactive {
info!("Running in interactive mode. Press Ctrl+C to shut down.");
tokio::signal::ctrl_c().await.context("wait for ctrl-c")?;
}
proc.shutdown().await?;
// Only sync if s3_prefix was specified
if let Some(s3_prefix) = maybe_s3_prefix {
info!("upload pgdata");
aws_s3_sync::upload_dir_recursive(
s3_client.as_ref().unwrap(),
Utf8Path::new(&pgdata_dir),
&s3_prefix.append("/pgdata/"),
)
.await
.context("sync dump directory to destination")?;
info!("write status");
{
let status_dir = workdir.join("status");
std::fs::create_dir(&status_dir).context("create status directory")?;
let status_file = status_dir.join("pgdata");
std::fs::write(&status_file, serde_json::json!({"done": true}).to_string())
.context("write status file")?;
aws_s3_sync::upload_dir_recursive(
s3_client.as_ref().unwrap(),
&status_dir,
&s3_prefix.append("/status/"),
)
.await
.context("sync status directory to destination")?;
}
}
Ok(())
}
async fn cmd_dumprestore(
kms_client: Option<aws_sdk_kms::Client>,
maybe_spec: Option<Spec>,
source_connection_string: Option<String>,
destination_connection_string: Option<String>,
workdir: Utf8PathBuf,
pg_bin_dir: Utf8PathBuf,
pg_lib_dir: Utf8PathBuf,
) -> Result<(), anyhow::Error> {
let (source_connstring, destination_connstring) = if let Some(spec) = maybe_spec {
match spec.encryption_secret {
EncryptionSecret::KMS { key_id } => {
let source = decode_connstring(
kms_client.as_ref().unwrap(),
&key_id,
spec.source_connstring_ciphertext_base64,
)
.await?;
let dest = if let Some(dest_ciphertext) =
spec.destination_connstring_ciphertext_base64
{
decode_connstring(kms_client.as_ref().unwrap(), &key_id, dest_ciphertext)
.await?
} else {
bail!("destination connection string must be provided in spec for dump_restore command");
};
(source, dest)
}
}
} else {
(
source_connection_string.unwrap(),
if let Some(val) = destination_connection_string {
val
} else {
bail!("destination connection string must be provided for dump_restore command");
},
)
};
run_dump_restore(
workdir,
pg_bin_dir,
pg_lib_dir,
source_connstring,
destination_connstring,
)
.await
}
#[tokio::main]
pub(crate) async fn main() -> anyhow::Result<()> {
utils::logging::init(
utils::logging::LogFormat::Json,
utils::logging::TracingErrorLayerEnablement::EnableWithRustLogFilter,
utils::logging::Output::Stdout,
)?;
info!("starting");
let args = Args::parse();
// Initialize AWS clients only if s3_prefix is specified
let (s3_client, kms_client) = if args.s3_prefix.is_some() {
let config = aws_config::load_defaults(BehaviorVersion::v2024_03_28()).await;
let s3_client = aws_sdk_s3::Client::new(&config);
let kms = aws_sdk_kms::Client::new(&config);
(Some(s3_client), Some(kms))
} else {
(None, None)
};
let spec: Option<Spec> = if let Some(s3_prefix) = &args.s3_prefix {
let spec_key = s3_prefix.append("/spec.json");
let object = s3_client
.as_ref()
.unwrap()
.get_object()
.bucket(&spec_key.bucket)
.key(spec_key.key)
.send()
.await
.context("get spec from s3")?
.body
.collect()
.await
.context("download spec body")?;
serde_json::from_slice(&object.into_bytes()).context("parse spec as json")?
} else {
None
};
match tokio::fs::create_dir(&args.working_directory).await {
Ok(()) => {}
Err(e) if e.kind() == std::io::ErrorKind::AlreadyExists => {
if !is_directory_empty(&args.working_directory)
.await
.context("check if working directory is empty")?
{
bail!("working directory is not empty");
} else {
// ok
}
}
Err(e) => return Err(anyhow::Error::new(e).context("create working directory")),
}
match args.command {
Command::Pgdata {
source_connection_string,
interactive,
pg_port,
num_cpus,
memory_mb,
} => {
cmd_pgdata(
s3_client,
kms_client,
args.s3_prefix,
spec,
source_connection_string,
interactive,
pg_port,
args.working_directory,
args.pg_bin_dir,
args.pg_lib_dir,
num_cpus,
memory_mb,
)
.await?;
}
Command::DumpRestore {
source_connection_string,
destination_connection_string,
} => {
cmd_dumprestore(
kms_client,
spec,
source_connection_string,
destination_connection_string,
args.working_directory,
args.pg_bin_dir,
args.pg_lib_dir,
)
.await?;
}
}

116
compute_tools/src/bin/fast_import/aws_s3_sync.rs
View File

@@ -1,24 +1,102 @@
use anyhow::Context;
use camino::Utf8Path;
use camino::{Utf8Path, Utf8PathBuf};
use tokio::task::JoinSet;
use walkdir::WalkDir;
use super::s3_uri::S3Uri;
pub(crate) async fn sync(local: &Utf8Path, remote: &S3Uri) -> anyhow::Result<()> {
let mut builder = tokio::process::Command::new("aws");
builder
.arg("s3")
.arg("sync")
.arg(local.as_str())
.arg(remote.to_string());
let st = builder
.spawn()
.context("spawn aws s3 sync")?
.wait()
.await
.context("wait for aws s3 sync")?;
if st.success() {
Ok(())
} else {
Err(anyhow::anyhow!("aws s3 sync failed"))
use tracing::{info, warn};
const MAX_PARALLEL_UPLOADS: usize = 10;
/// Upload all files from 'local' to 'remote'
pub(crate) async fn upload_dir_recursive(
s3_client: &aws_sdk_s3::Client,
local: &Utf8Path,
remote: &S3Uri,
) -> anyhow::Result<()> {
// Recursively scan directory
let mut dirwalker = WalkDir::new(local)
.into_iter()
.map(|entry| {
let entry = entry?;
let file_type = entry.file_type();
let path = <&Utf8Path>::try_from(entry.path())?.to_path_buf();
Ok((file_type, path))
})
.filter_map(|e: anyhow::Result<(std::fs::FileType, Utf8PathBuf)>| {
match e {
Ok((file_type, path)) if file_type.is_file() => Some(Ok(path)),
Ok((file_type, _path)) if file_type.is_dir() => {
// The WalkDir iterator will recurse into directories, but we don't want
// to do anything with directories as such. There's no concept of uploading
// an empty directory to S3.
None
}
Ok((file_type, path)) if file_type.is_symlink() => {
// huh, didn't expect a symlink. Can't upload that to S3. Warn and skip.
warn!("cannot upload symlink ({})", path);
None
}
Ok((_file_type, path)) => {
// should not happen
warn!("directory entry has unexpected type ({})", path);
None
}
Err(e) => Some(Err(e)),
}
});
// Spawn upload tasks for each file, keeping MAX_PARALLEL_UPLOADS active in
// parallel.
let mut joinset = JoinSet::new();
loop {
// Could we upload more?
while joinset.len() < MAX_PARALLEL_UPLOADS {
if let Some(full_local_path) = dirwalker.next() {
let full_local_path = full_local_path?;
let relative_local_path = full_local_path
.strip_prefix(local)
.expect("all paths start from the walkdir root");
let remote_path = remote.append(relative_local_path.as_str());
info!(
"starting upload of {} to {}",
&full_local_path, &remote_path
);
let upload_task = upload_file(s3_client.clone(), full_local_path, remote_path);
joinset.spawn(upload_task);
} else {
info!("draining upload tasks");
break;
}
}
// Wait for an upload to complete
if let Some(res) = joinset.join_next().await {
let _ = res?;
} else {
// all done!
break;
}
}
Ok(())
}
pub(crate) async fn upload_file(
s3_client: aws_sdk_s3::Client,
local_path: Utf8PathBuf,
remote: S3Uri,
) -> anyhow::Result<()> {
use aws_smithy_types::byte_stream::ByteStream;
let stream = ByteStream::from_path(&local_path).await?;
let _result = s3_client
.put_object()
.bucket(remote.bucket)
.key(&remote.key)
.body(stream)
.send()
.await?;
info!("upload of {} to {} finished", &local_path, &remote.key);
Ok(())
}

31
compute_tools/src/catalog.rs
View File

@@ -140,5 +140,34 @@ pub async fn get_database_schema(
warn!("pg_dump stderr: {}", line)
}
});
Ok(initial_stream.chain(stdout_reader.map(|res| res.map(|b| b.freeze()))))
#[allow(dead_code)]
struct SchemaStream<S> {
// We keep a reference to the child process to ensure it stays alive
// while the stream is being consumed. When SchemaStream is dropped,
// cmd will be dropped, which triggers kill_on_drop and terminates pg_dump
cmd: tokio::process::Child,
stream: S,
}
impl<S> Stream for SchemaStream<S>
where
S: Stream<Item = Result<bytes::Bytes, std::io::Error>> + Unpin,
{
type Item = Result<bytes::Bytes, std::io::Error>;
fn poll_next(
mut self: std::pin::Pin<&mut Self>,
cx: &mut std::task::Context<'_>,
) -> std::task::Poll<Option<Self::Item>> {
Stream::poll_next(std::pin::Pin::new(&mut self.stream), cx)
}
}
let schema_stream = SchemaStream {
cmd,
stream: initial_stream.chain(stdout_reader.map(|res| res.map(|b| b.freeze()))),
};
Ok(schema_stream)
}

147
compute_tools/src/compute.rs
View File

@@ -9,7 +9,6 @@ use std::str::FromStr;
use std::sync::atomic::AtomicU32;
use std::sync::atomic::Ordering;
use std::sync::{Arc, Condvar, Mutex, RwLock};
use std::thread;
use std::time::Duration;
use std::time::Instant;
@@ -59,6 +58,8 @@ pub static PG_PID: AtomicU32 = AtomicU32::new(0);
/// Compute node info shared across several `compute_ctl` threads.
pub struct ComputeNode {
/// The ID of the compute
pub compute_id: String,
// Url type maintains proper escaping
pub connstr: url::Url,
// We connect to Postgres from many different places, so build configs once
@@ -81,8 +82,10 @@ pub struct ComputeNode {
/// - we push spec and it does configuration
/// - but then it is restarted without any spec again
pub live_config_allowed: bool,
/// The port that the compute's HTTP server listens on
pub http_port: u16,
/// The port that the compute's external HTTP server listens on
pub external_http_port: u16,
/// The port that the compute's internal HTTP server listens on
pub internal_http_port: u16,
/// Volatile part of the `ComputeNode`, which should be used under `Mutex`.
/// To allow HTTP API server to serving status requests, while configuration
/// is in progress, lock should be held only for short periods of time to do
@@ -546,11 +549,7 @@ impl ComputeNode {
pub fn check_safekeepers_synced(&self, compute_state: &ComputeState) -> Result<Option<Lsn>> {
let start_time = Utc::now();
// Run actual work with new tokio runtime
let rt = tokio::runtime::Builder::new_current_thread()
.enable_all()
.build()
.expect("failed to create rt");
let rt = tokio::runtime::Handle::current();
let result = rt.block_on(self.check_safekeepers_synced_async(compute_state));
// Record runtime
@@ -597,9 +596,9 @@ impl ComputeNode {
SYNC_SAFEKEEPERS_PID.store(0, Ordering::SeqCst);
// Process has exited, so we can join the logs thread.
let _ = logs_handle
.join()
.map_err(|e| tracing::error!("log thread panicked: {:?}", e));
let _ = tokio::runtime::Handle::current()
.block_on(logs_handle)
.map_err(|e| tracing::error!("log task panicked: {:?}", e));
if !sync_output.status.success() {
anyhow::bail!(
@@ -634,7 +633,7 @@ impl ComputeNode {
config::write_postgres_conf(
&pgdata_path.join("postgresql.conf"),
&pspec.spec,
self.http_port,
self.internal_http_port,
)?;
// Syncing safekeepers is only safe with primary nodes: if a primary
@@ -784,7 +783,7 @@ impl ComputeNode {
pub fn start_postgres(
&self,
storage_auth_token: Option<String>,
) -> Result<(std::process::Child, std::thread::JoinHandle<()>)> {
) -> Result<(std::process::Child, tokio::task::JoinHandle<Result<()>>)> {
let pgdata_path = Path::new(&self.pgdata);
// Run postgres as a child process.
@@ -800,7 +799,7 @@ impl ComputeNode {
.expect("cannot start postgres process");
PG_PID.store(pg.id(), Ordering::SeqCst);
// Start a thread to collect logs from stderr.
// Start a task to collect logs from stderr.
let stderr = pg.stderr.take().expect("stderr should be captured");
let logs_handle = handle_postgres_logs(stderr);
@@ -809,20 +808,28 @@ impl ComputeNode {
Ok((pg, logs_handle))
}
/// Do post configuration of the already started Postgres. This function spawns a background thread to
/// Do post configuration of the already started Postgres. This function spawns a background task to
/// configure the database after applying the compute spec. Currently, it upgrades the neon extension
/// version. In the future, it may upgrade all 3rd-party extensions.
#[instrument(skip_all)]
pub fn post_apply_config(&self) -> Result<()> {
let conf = self.get_conn_conf(Some("compute_ctl:post_apply_config"));
thread::spawn(move || {
let func = || {
let mut client = conf.connect(NoTls)?;
let conf = self.get_tokio_conn_conf(Some("compute_ctl:post_apply_config"));
tokio::spawn(async move {
let res = async {
let (mut client, connection) = conf.connect(NoTls).await?;
tokio::spawn(async move {
if let Err(e) = connection.await {
eprintln!("connection error: {}", e);
}
});
handle_neon_extension_upgrade(&mut client)
.await
.context("handle_neon_extension_upgrade")?;
Ok::<_, anyhow::Error>(())
};
if let Err(err) = func() {
}
.await;
if let Err(err) = res {
error!("error while post_apply_config: {err:#}");
}
});
@@ -919,13 +926,10 @@ impl ComputeNode {
conf: Arc<tokio_postgres::Config>,
concurrency: usize,
) -> Result<()> {
let rt = tokio::runtime::Builder::new_multi_thread()
.enable_all()
.build()?;
info!("Applying config with max {} concurrency", concurrency);
debug!("Config: {:?}", spec);
let rt = tokio::runtime::Handle::current();
rt.block_on(async {
// Proceed with post-startup configuration. Note, that order of operations is important.
let client = Self::get_maintenance_client(&conf).await?;
@@ -1319,14 +1323,18 @@ impl ComputeNode {
}
// Run migrations separately to not hold up cold starts
thread::spawn(move || {
let conf = conf.as_ref().clone();
let mut conf = postgres::config::Config::from(conf);
tokio::spawn(async move {
let mut conf = conf.as_ref().clone();
conf.application_name("compute_ctl:migrations");
match conf.connect(NoTls) {
Ok(mut client) => {
if let Err(e) = handle_migrations(&mut client) {
match conf.connect(NoTls).await {
Ok((mut client, connection)) => {
tokio::spawn(async move {
if let Err(e) = connection.await {
eprintln!("connection error: {}", e);
}
});
if let Err(e) = handle_migrations(&mut client).await {
error!("Failed to run migrations: {}", e);
}
}
@@ -1363,16 +1371,11 @@ impl ComputeNode {
if let Some(ref pgbouncer_settings) = spec.pgbouncer_settings {
info!("tuning pgbouncer");
let rt = tokio::runtime::Builder::new_current_thread()
.enable_all()
.build()
.expect("failed to create rt");
// Spawn a thread to do the tuning,
// Spawn a background task to do the tuning,
// so that we don't block the main thread that starts Postgres.
let pgbouncer_settings = pgbouncer_settings.clone();
let _handle = thread::spawn(move || {
let res = rt.block_on(tune_pgbouncer(pgbouncer_settings));
tokio::spawn(async move {
let res = tune_pgbouncer(pgbouncer_settings).await;
if let Err(err) = res {
error!("error while tuning pgbouncer: {err:?}");
}
@@ -1382,41 +1385,42 @@ impl ComputeNode {
if let Some(ref local_proxy) = spec.local_proxy_config {
info!("configuring local_proxy");
// Spawn a thread to do the configuration,
// Spawn a background task to do the configuration,
// so that we don't block the main thread that starts Postgres.
let local_proxy = local_proxy.clone();
let _handle = Some(thread::spawn(move || {
tokio::spawn(async move {
if let Err(err) = local_proxy::configure(&local_proxy) {
error!("error while configuring local_proxy: {err:?}");
}
}));
});
}
// Write new config
let pgdata_path = Path::new(&self.pgdata);
let postgresql_conf_path = pgdata_path.join("postgresql.conf");
config::write_postgres_conf(&postgresql_conf_path, &spec, self.http_port)?;
config::write_postgres_conf(&postgresql_conf_path, &spec, self.internal_http_port)?;
let max_concurrent_connections = spec.reconfigure_concurrency;
if !spec.skip_pg_catalog_updates {
let max_concurrent_connections = spec.reconfigure_concurrency;
// Temporarily reset max_cluster_size in config
// to avoid the possibility of hitting the limit, while we are reconfiguring:
// creating new extensions, roles, etc.
config::with_compute_ctl_tmp_override(pgdata_path, "neon.max_cluster_size=-1", || {
self.pg_reload_conf()?;
// Temporarily reset max_cluster_size in config
// to avoid the possibility of hitting the limit, while we are reconfiguring:
// creating new extensions, roles, etc.
config::with_compute_ctl_tmp_override(pgdata_path, "neon.max_cluster_size=-1", || {
self.pg_reload_conf()?;
if spec.mode == ComputeMode::Primary {
let mut conf = tokio_postgres::Config::from_str(self.connstr.as_str()).unwrap();
conf.application_name("apply_config");
let conf = Arc::new(conf);
if spec.mode == ComputeMode::Primary {
let mut conf = tokio_postgres::Config::from_str(self.connstr.as_str()).unwrap();
conf.application_name("apply_config");
let conf = Arc::new(conf);
let spec = Arc::new(spec.clone());
let spec = Arc::new(spec.clone());
self.apply_spec_sql(spec, conf, max_concurrent_connections)?;
}
self.apply_spec_sql(spec, conf, max_concurrent_connections)?;
}
Ok(())
})?;
Ok(())
})?;
}
self.pg_reload_conf()?;
@@ -1431,7 +1435,9 @@ impl ComputeNode {
}
#[instrument(skip_all)]
pub fn start_compute(&self) -> Result<(std::process::Child, std::thread::JoinHandle<()>)> {
pub fn start_compute(
&self,
) -> Result<(std::process::Child, tokio::task::JoinHandle<Result<()>>)> {
let compute_state = self.state.lock().unwrap().clone();
let pspec = compute_state.pspec.as_ref().expect("spec must be set");
info!(
@@ -1446,16 +1452,11 @@ impl ComputeNode {
if let Some(pgbouncer_settings) = &pspec.spec.pgbouncer_settings {
info!("tuning pgbouncer");
let rt = tokio::runtime::Builder::new_current_thread()
.enable_all()
.build()
.expect("failed to create rt");
// Spawn a thread to do the tuning,
// Spawn a background task to do the tuning,
// so that we don't block the main thread that starts Postgres.
let pgbouncer_settings = pgbouncer_settings.clone();
let _handle = thread::spawn(move || {
let res = rt.block_on(tune_pgbouncer(pgbouncer_settings));
let _handle = tokio::spawn(async move {
let res = tune_pgbouncer(pgbouncer_settings).await;
if let Err(err) = res {
error!("error while tuning pgbouncer: {err:?}");
}
@@ -1465,10 +1466,10 @@ impl ComputeNode {
if let Some(local_proxy) = &pspec.spec.local_proxy_config {
info!("configuring local_proxy");
// Spawn a thread to do the configuration,
// Spawn a background task to do the configuration,
// so that we don't block the main thread that starts Postgres.
let local_proxy = local_proxy.clone();
let _handle = thread::spawn(move || {
let _handle = tokio::spawn(async move {
if let Err(err) = local_proxy::configure(&local_proxy) {
error!("error while configuring local_proxy: {err:?}");
}
@@ -1487,7 +1488,8 @@ impl ComputeNode {
extension_server::create_control_files(remote_extensions, &self.pgbin);
let library_load_start_time = Utc::now();
let remote_ext_metrics = self.prepare_preload_libraries(&pspec.spec)?;
let rt = tokio::runtime::Handle::current();
let remote_ext_metrics = rt.block_on(self.prepare_preload_libraries(&pspec.spec))?;
let library_load_time = Utc::now()
.signed_duration_since(library_load_start_time)
@@ -1542,7 +1544,7 @@ impl ComputeNode {
self.post_apply_config()?;
let conf = self.get_conn_conf(None);
thread::spawn(move || {
tokio::task::spawn_blocking(|| {
let res = get_installed_extensions(conf);
match res {
Ok(extensions) => {
@@ -1891,7 +1893,6 @@ LIMIT 100",
Ok(ext_version)
}
#[tokio::main]
pub async fn prepare_preload_libraries(
&self,
spec: &ComputeSpec,

3
compute_tools/src/configurator.rs
View File

@@ -51,9 +51,12 @@ fn configurator_main_loop(compute: &Arc<ComputeNode>) {
pub fn launch_configurator(compute: &Arc<ComputeNode>) -> thread::JoinHandle<()> {
let compute = Arc::clone(compute);
let runtime = tokio::runtime::Handle::current();
thread::Builder::new()
.name("compute-configurator".into())
.spawn(move || {
let _rt_guard = runtime.enter();
configurator_main_loop(&compute);
info!("configurator thread is exited");
})

4
compute_tools/src/http/mod.rs
View File

@@ -4,11 +4,9 @@ use http::{header::CONTENT_TYPE, StatusCode};
use serde::Serialize;
use tracing::error;
pub use server::launch_http_server;
mod extract;
mod routes;
mod server;
pub mod server;
/// Convenience response builder for JSON responses
struct JsonResponse;

16
compute_tools/src/http/routes/failpoints.rs
View File

@@ -1,7 +1,21 @@
use axum::response::{IntoResponse, Response};
use http::StatusCode;
use serde::{Deserialize, Serialize};
use tracing::info;
use utils::failpoint_support::{apply_failpoint, ConfigureFailpointsRequest};
use utils::failpoint_support::apply_failpoint;
pub type ConfigureFailpointsRequest = Vec<FailpointConfig>;
/// Information for configuring a single fail point
#[derive(Debug, Serialize, Deserialize)]
pub struct FailpointConfig {
/// Name of the fail point
pub name: String,
/// List of actions to take, using the format described in `fail::cfg`
///
/// We also support `actions = "exit"` to cause the fail point to immediately exit.
pub actions: String,
}
use crate::http::{extract::Json, JsonResponse};

198
compute_tools/src/http/server.rs
View File

@@ -1,7 +1,7 @@
use std::{
fmt::Display,
net::{IpAddr, Ipv6Addr, SocketAddr},
sync::Arc,
thread,
time::Duration,
};
@@ -26,46 +26,65 @@ use super::routes::{
};
use crate::compute::ComputeNode;
async fn handle_404() -> Response {
StatusCode::NOT_FOUND.into_response()
}
const X_REQUEST_ID: &str = "x-request-id";
/// This middleware function allows compute_ctl to generate its own request ID
/// if one isn't supplied. The control plane will always send one as a UUID. The
/// neon Postgres extension on the other hand does not send one.
async fn maybe_add_request_id_header(mut request: Request, next: Next) -> Response {
let headers = request.headers_mut();
if headers.get(X_REQUEST_ID).is_none() {
headers.append(X_REQUEST_ID, Uuid::new_v4().to_string().parse().unwrap());
}
next.run(request).await
/// `compute_ctl` has two servers: internal and external. The internal server
/// binds to the loopback interface and handles communication from clients on
/// the compute. The external server is what receives communication from the
/// control plane, the metrics scraper, etc. We make the distinction because
/// certain routes in `compute_ctl` only need to be exposed to local processes
/// like Postgres via the neon extension and local_proxy.
#[derive(Clone, Copy, Debug)]
pub enum Server {
Internal(u16),
External(u16),
}
/// Run the HTTP server and wait on it forever.
#[tokio::main]
async fn serve(port: u16, compute: Arc<ComputeNode>) {
let mut app = Router::new()
.route("/check_writability", post(check_writability::is_writable))
.route("/configure", post(configure::configure))
.route("/database_schema", get(database_schema::get_schema_dump))
.route("/dbs_and_roles", get(dbs_and_roles::get_catalog_objects))
.route(
"/extension_server/{*filename}",
post(extension_server::download_extension),
)
.route("/extensions", post(extensions::install_extension))
.route("/grants", post(grants::add_grant))
.route("/insights", get(insights::get_insights))
.route("/metrics", get(metrics::get_metrics))
.route("/metrics.json", get(metrics_json::get_metrics))
.route("/status", get(status::get_status))
.route("/terminate", post(terminate::terminate))
.fallback(handle_404)
.layer(
impl Display for Server {
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
match self {
Server::Internal(_) => f.write_str("internal"),
Server::External(_) => f.write_str("external"),
}
}
}
impl From<Server> for Router<Arc<ComputeNode>> {
fn from(server: Server) -> Self {
let mut router = Router::<Arc<ComputeNode>>::new();
router = match server {
Server::Internal(_) => {
router = router
.route(
"/extension_server/{*filename}",
post(extension_server::download_extension),
)
.route("/extensions", post(extensions::install_extension))
.route("/grants", post(grants::add_grant));
// Add in any testing support
if cfg!(feature = "testing") {
use super::routes::failpoints;
router = router.route("/failpoints", post(failpoints::configure_failpoints));
}
router
}
Server::External(_) => router
.route("/check_writability", post(check_writability::is_writable))
.route("/configure", post(configure::configure))
.route("/database_schema", get(database_schema::get_schema_dump))
.route("/dbs_and_roles", get(dbs_and_roles::get_catalog_objects))
.route("/insights", get(insights::get_insights))
.route("/metrics", get(metrics::get_metrics))
.route("/metrics.json", get(metrics_json::get_metrics))
.route("/status", get(status::get_status))
.route("/terminate", post(terminate::terminate)),
};
router.fallback(Server::handle_404).method_not_allowed_fallback(Server::handle_405).layer(
ServiceBuilder::new()
// Add this middleware since we assume the request ID exists
.layer(middleware::from_fn(maybe_add_request_id_header))
@@ -105,45 +124,88 @@ async fn serve(port: u16, compute: Arc<ComputeNode>) {
)
.layer(PropagateRequestIdLayer::x_request_id()),
)
.with_state(compute);
}
}
// Add in any testing support
if cfg!(feature = "testing") {
use super::routes::failpoints;
app = app.route("/failpoints", post(failpoints::configure_failpoints))
impl Server {
async fn handle_404() -> impl IntoResponse {
StatusCode::NOT_FOUND
}
// This usually binds to both IPv4 and IPv6 on Linux, see
// https://github.com/rust-lang/rust/pull/34440 for more information
let addr = SocketAddr::new(IpAddr::from(Ipv6Addr::UNSPECIFIED), port);
let listener = match TcpListener::bind(&addr).await {
Ok(listener) => listener,
Err(e) => {
error!(
"failed to bind the compute_ctl HTTP server to port {}: {}",
port, e
);
return;
async fn handle_405() -> impl IntoResponse {
StatusCode::METHOD_NOT_ALLOWED
}
async fn listener(&self) -> Result<TcpListener> {
let addr = SocketAddr::new(self.ip(), self.port());
let listener = TcpListener::bind(&addr).await?;
Ok(listener)
}
fn ip(&self) -> IpAddr {
match self {
// TODO: Change this to Ipv6Addr::LOCALHOST when the GitHub runners
// allow binding to localhost
Server::Internal(_) => IpAddr::from(Ipv6Addr::UNSPECIFIED),
Server::External(_) => IpAddr::from(Ipv6Addr::UNSPECIFIED),
}
};
if let Ok(local_addr) = listener.local_addr() {
info!("compute_ctl HTTP server listening on {}", local_addr);
} else {
info!("compute_ctl HTTP server listening on port {}", port);
}
if let Err(e) = axum::serve(listener, app).await {
error!("compute_ctl HTTP server error: {}", e);
fn port(self) -> u16 {
match self {
Server::Internal(port) => port,
Server::External(port) => port,
}
}
async fn serve(self, compute: Arc<ComputeNode>) {
let listener = self.listener().await.unwrap_or_else(|e| {
// If we can't bind, the compute cannot operate correctly
panic!(
"failed to bind the compute_ctl {} HTTP server to {}: {}",
self,
SocketAddr::new(self.ip(), self.port()),
e
);
});
if tracing::enabled!(tracing::Level::INFO) {
let local_addr = match listener.local_addr() {
Ok(local_addr) => local_addr,
Err(_) => SocketAddr::new(self.ip(), self.port()),
};
info!(
"compute_ctl {} HTTP server listening at {}",
self, local_addr
);
}
let router = Router::from(self).with_state(compute);
if let Err(e) = axum::serve(listener, router).await {
error!("compute_ctl {} HTTP server error: {}", self, e);
}
}
pub fn launch(self, compute: &Arc<ComputeNode>) {
let state = Arc::clone(compute);
info!("Launching the {} server", self);
tokio::spawn(self.serve(state));
}
}
/// Launch a separate HTTP server thread and return its `JoinHandle`.
pub fn launch_http_server(port: u16, state: &Arc<ComputeNode>) -> Result<thread::JoinHandle<()>> {
let state = Arc::clone(state);
/// This middleware function allows compute_ctl to generate its own request ID
/// if one isn't supplied. The control plane will always send one as a UUID. The
/// neon Postgres extension on the other hand does not send one.
async fn maybe_add_request_id_header(mut request: Request, next: Next) -> Response {
let headers = request.headers_mut();
if headers.get(X_REQUEST_ID).is_none() {
headers.append(X_REQUEST_ID, Uuid::new_v4().to_string().parse().unwrap());
}
Ok(thread::Builder::new()
.name("http-server".into())
.spawn(move || serve(port, state))?)
next.run(request).await
}

4
compute_tools/src/logger.rs
View File

@@ -11,7 +11,7 @@ use tracing_subscriber::prelude::*;
/// set `OTEL_EXPORTER_OTLP_ENDPOINT=http://jaeger:4318`. See
/// `tracing-utils` package description.
///
pub fn init_tracing_and_logging(default_log_level: &str) -> anyhow::Result<()> {
pub async fn init_tracing_and_logging(default_log_level: &str) -> anyhow::Result<()> {
// Initialize Logging
let env_filter = tracing_subscriber::EnvFilter::try_from_default_env()
.unwrap_or_else(|_| tracing_subscriber::EnvFilter::new(default_log_level));
@@ -22,7 +22,7 @@ pub fn init_tracing_and_logging(default_log_level: &str) -> anyhow::Result<()> {
.with_writer(std::io::stderr);
// Initialize OpenTelemetry
let otlp_layer = tracing_utils::init_tracing_without_runtime("compute_ctl");
let otlp_layer = tracing_utils::init_tracing("compute_ctl").await;
// Put it all together
tracing_subscriber::registry()

47
compute_tools/src/migration.rs
View File

@@ -1,6 +1,6 @@
use anyhow::{Context, Result};
use fail::fail_point;
use postgres::{Client, Transaction};
use tokio_postgres::{Client, Transaction};
use tracing::{error, info};
use crate::metrics::DB_MIGRATION_FAILED;
@@ -21,10 +21,11 @@ impl<'m> MigrationRunner<'m> {
}
/// Get the current value neon_migration.migration_id
fn get_migration_id(&mut self) -> Result<i64> {
async fn get_migration_id(&mut self) -> Result<i64> {
let row = self
.client
.query_one("SELECT id FROM neon_migration.migration_id", &[])?;
.query_one("SELECT id FROM neon_migration.migration_id", &[])
.await?;
Ok(row.get::<&str, i64>("id"))
}
@@ -34,7 +35,7 @@ impl<'m> MigrationRunner<'m> {
/// This function has a fail point called compute-migration, which can be
/// used if you would like to fail the application of a series of migrations
/// at some point.
fn update_migration_id(txn: &mut Transaction, migration_id: i64) -> Result<()> {
async fn update_migration_id(txn: &mut Transaction<'_>, migration_id: i64) -> Result<()> {
// We use this fail point in order to check that failing in the
// middle of applying a series of migrations fails in an expected
// manner
@@ -59,31 +60,38 @@ impl<'m> MigrationRunner<'m> {
"UPDATE neon_migration.migration_id SET id = $1",
&[&migration_id],
)
.await
.with_context(|| format!("update neon_migration.migration_id to {migration_id}"))?;
Ok(())
}
/// Prepare the migrations the target database for handling migrations
fn prepare_database(&mut self) -> Result<()> {
async fn prepare_database(&mut self) -> Result<()> {
self.client
.simple_query("CREATE SCHEMA IF NOT EXISTS neon_migration")?;
self.client.simple_query("CREATE TABLE IF NOT EXISTS neon_migration.migration_id (key INT NOT NULL PRIMARY KEY, id bigint NOT NULL DEFAULT 0)")?;
self.client.simple_query(
"INSERT INTO neon_migration.migration_id VALUES (0, 0) ON CONFLICT DO NOTHING",
)?;
.simple_query("CREATE SCHEMA IF NOT EXISTS neon_migration")
.await?;
self.client.simple_query("CREATE TABLE IF NOT EXISTS neon_migration.migration_id (key INT NOT NULL PRIMARY KEY, id bigint NOT NULL DEFAULT 0)").await?;
self.client
.simple_query("ALTER SCHEMA neon_migration OWNER TO cloud_admin")?;
.simple_query(
"INSERT INTO neon_migration.migration_id VALUES (0, 0) ON CONFLICT DO NOTHING",
)
.await?;
self.client
.simple_query("REVOKE ALL ON SCHEMA neon_migration FROM PUBLIC")?;
.simple_query("ALTER SCHEMA neon_migration OWNER TO cloud_admin")
.await?;
self.client
.simple_query("REVOKE ALL ON SCHEMA neon_migration FROM PUBLIC")
.await?;
Ok(())
}
/// Run an individual migration in a separate transaction block.
fn run_migration(client: &mut Client, migration_id: i64, migration: &str) -> Result<()> {
async fn run_migration(client: &mut Client, migration_id: i64, migration: &str) -> Result<()> {
let mut txn = client
.transaction()
.await
.with_context(|| format!("begin transaction for migration {migration_id}"))?;
if migration.starts_with("-- SKIP") {
@@ -92,35 +100,38 @@ impl<'m> MigrationRunner<'m> {
// Even though we are skipping the migration, updating the
// migration ID should help keep logic easy to understand when
// trying to understand the state of a cluster.
Self::update_migration_id(&mut txn, migration_id)?;
Self::update_migration_id(&mut txn, migration_id).await?;
} else {
info!("Running migration id={}:\n{}\n", migration_id, migration);
txn.simple_query(migration)
.await
.with_context(|| format!("apply migration {migration_id}"))?;
Self::update_migration_id(&mut txn, migration_id)?;
Self::update_migration_id(&mut txn, migration_id).await?;
}
txn.commit()
.await
.with_context(|| format!("commit transaction for migration {migration_id}"))?;
Ok(())
}
/// Run the configured set of migrations
pub fn run_migrations(mut self) -> Result<()> {
pub async fn run_migrations(mut self) -> Result<()> {
self.prepare_database()
.await
.context("prepare database to handle migrations")?;
let mut current_migration = self.get_migration_id()? as usize;
let mut current_migration = self.get_migration_id().await? as usize;
while current_migration < self.migrations.len() {
// The index lags the migration ID by 1, so the current migration
// ID is also the next index
let migration_id = (current_migration + 1) as i64;
let migration = self.migrations[current_migration];
match Self::run_migration(self.client, migration_id, migration) {
match Self::run_migration(self.client, migration_id, migration).await {
Ok(_) => {
info!("Finished migration id={}", migration_id);
}

22
compute_tools/src/pg_helpers.rs
View File

@@ -7,7 +7,6 @@ use std::os::unix::fs::PermissionsExt;
use std::path::Path;
use std::process::Child;
use std::str::FromStr;
use std::thread::JoinHandle;
use std::time::{Duration, Instant};
use anyhow::{bail, Result};
@@ -16,6 +15,7 @@ use ini::Ini;
use notify::{RecursiveMode, Watcher};
use postgres::config::Config;
use tokio::io::AsyncBufReadExt;
use tokio::task::JoinHandle;
use tokio::time::timeout;
use tokio_postgres;
use tokio_postgres::NoTls;
@@ -477,23 +477,13 @@ pub async fn tune_pgbouncer(pgbouncer_config: HashMap<String, String>) -> Result
Ok(())
}
/// Spawn a thread that will read Postgres logs from `stderr`, join multiline logs
/// Spawn a task that will read Postgres logs from `stderr`, join multiline logs
/// and send them to the logger. In the future we may also want to add context to
/// these logs.
pub fn handle_postgres_logs(stderr: std::process::ChildStderr) -> JoinHandle<()> {
std::thread::spawn(move || {
let runtime = tokio::runtime::Builder::new_current_thread()
.enable_all()
.build()
.expect("failed to build tokio runtime");
let res = runtime.block_on(async move {
let stderr = tokio::process::ChildStderr::from_std(stderr)?;
handle_postgres_logs_async(stderr).await
});
if let Err(e) = res {
tracing::error!("error while processing postgres logs: {}", e);
}
pub fn handle_postgres_logs(stderr: std::process::ChildStderr) -> JoinHandle<Result<()>> {
tokio::spawn(async move {
let stderr = tokio::process::ChildStderr::from_std(stderr)?;
handle_postgres_logs_async(stderr).await
})
}

55
compute_tools/src/spec.rs
View File

@@ -1,8 +1,8 @@
use anyhow::{anyhow, bail, Result};
use postgres::Client;
use reqwest::StatusCode;
use std::fs::File;
use std::path::Path;
use tokio_postgres::Client;
use tracing::{error, info, instrument, warn};
use crate::config;
@@ -11,7 +11,9 @@ use crate::migration::MigrationRunner;
use crate::params::PG_HBA_ALL_MD5;
use crate::pg_helpers::*;
use compute_api::responses::{ControlPlaneComputeStatus, ControlPlaneSpecResponse};
use compute_api::responses::{
ComputeCtlConfig, ControlPlaneComputeStatus, ControlPlaneSpecResponse,
};
use compute_api::spec::ComputeSpec;
// Do control plane request and return response if any. In case of error it
@@ -73,14 +75,13 @@ fn do_control_plane_request(
pub fn get_spec_from_control_plane(
base_uri: &str,
compute_id: &str,
) -> Result<Option<ComputeSpec>> {
) -> Result<(Option<ComputeSpec>, ComputeCtlConfig)> {
let cp_uri = format!("{base_uri}/compute/api/v2/computes/{compute_id}/spec");
let jwt: String = match std::env::var("NEON_CONTROL_PLANE_TOKEN") {
Ok(v) => v,
Err(_) => "".to_string(),
};
let mut attempt = 1;
let mut spec: Result<Option<ComputeSpec>> = Ok(None);
info!("getting spec from control plane: {}", cp_uri);
@@ -90,7 +91,7 @@ pub fn get_spec_from_control_plane(
// - no spec for compute yet (Empty state) -> return Ok(None)
// - got spec -> return Ok(Some(spec))
while attempt < 4 {
spec = match do_control_plane_request(&cp_uri, &jwt) {
let result = match do_control_plane_request(&cp_uri, &jwt) {
Ok(spec_resp) => {
CPLANE_REQUESTS_TOTAL
.with_label_values(&[
@@ -99,10 +100,10 @@ pub fn get_spec_from_control_plane(
])
.inc();
match spec_resp.status {
ControlPlaneComputeStatus::Empty => Ok(None),
ControlPlaneComputeStatus::Empty => Ok((None, spec_resp.compute_ctl_config)),
ControlPlaneComputeStatus::Attached => {
if let Some(spec) = spec_resp.spec {
Ok(Some(spec))
Ok((Some(spec), spec_resp.compute_ctl_config))
} else {
bail!("compute is attached, but spec is empty")
}
@@ -121,10 +122,10 @@ pub fn get_spec_from_control_plane(
}
};
if let Err(e) = &spec {
if let Err(e) = &result {
error!("attempt {} to get spec failed with: {}", attempt, e);
} else {
return spec;
return result;
}
attempt += 1;
@@ -132,7 +133,9 @@ pub fn get_spec_from_control_plane(
}
// All attempts failed, return error.
spec
Err(anyhow::anyhow!(
"Exhausted all attempts to retrieve the spec from the control plane"
))
}
/// Check `pg_hba.conf` and update if needed to allow external connections.
@@ -166,17 +169,17 @@ pub fn add_standby_signal(pgdata_path: &Path) -> Result<()> {
}
#[instrument(skip_all)]
pub fn handle_neon_extension_upgrade(client: &mut Client) -> Result<()> {
pub async fn handle_neon_extension_upgrade(client: &mut Client) -> Result<()> {
info!("handle neon extension upgrade");
let query = "ALTER EXTENSION neon UPDATE";
info!("update neon extension version with query: {}", query);
client.simple_query(query)?;
client.simple_query(query).await?;
Ok(())
}
#[instrument(skip_all)]
pub fn handle_migrations(client: &mut Client) -> Result<()> {
pub async fn handle_migrations(client: &mut Client) -> Result<()> {
info!("handle migrations");
// !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
@@ -206,7 +209,9 @@ pub fn handle_migrations(client: &mut Client) -> Result<()> {
),
];
MigrationRunner::new(client, &migrations).run_migrations()?;
MigrationRunner::new(client, &migrations)
.run_migrations()
.await?;
Ok(())
}
@@ -214,7 +219,7 @@ pub fn handle_migrations(client: &mut Client) -> Result<()> {
/// Connect to the database as superuser and pre-create anon extension
/// if it is present in shared_preload_libraries
#[instrument(skip_all)]
pub fn handle_extension_anon(
pub async fn handle_extension_anon(
spec: &ComputeSpec,
db_owner: &str,
db_client: &mut Client,
@@ -227,7 +232,7 @@ pub fn handle_extension_anon(
if !grants_only {
// check if extension is already initialized using anon.is_initialized()
let query = "SELECT anon.is_initialized()";
match db_client.query(query, &[]) {
match db_client.query(query, &[]).await {
Ok(rows) => {
if !rows.is_empty() {
let is_initialized: bool = rows[0].get(0);
@@ -249,7 +254,7 @@ pub fn handle_extension_anon(
// Users cannot create it themselves, because superuser is required.
let mut query = "CREATE EXTENSION IF NOT EXISTS anon CASCADE";
info!("creating anon extension with query: {}", query);
match db_client.query(query, &[]) {
match db_client.query(query, &[]).await {
Ok(_) => {}
Err(e) => {
error!("anon extension creation failed with error: {}", e);
@@ -259,7 +264,7 @@ pub fn handle_extension_anon(
// check that extension is installed
query = "SELECT extname FROM pg_extension WHERE extname = 'anon'";
let rows = db_client.query(query, &[])?;
let rows = db_client.query(query, &[]).await?;
if rows.is_empty() {
error!("anon extension is not installed");
return Ok(());
@@ -268,7 +273,7 @@ pub fn handle_extension_anon(
// Initialize anon extension
// This also requires superuser privileges, so users cannot do it themselves.
query = "SELECT anon.init()";
match db_client.query(query, &[]) {
match db_client.query(query, &[]).await {
Ok(_) => {}
Err(e) => {
error!("anon.init() failed with error: {}", e);
@@ -279,7 +284,7 @@ pub fn handle_extension_anon(
// check that extension is installed, if not bail early
let query = "SELECT extname FROM pg_extension WHERE extname = 'anon'";
match db_client.query(query, &[]) {
match db_client.query(query, &[]).await {
Ok(rows) => {
if rows.is_empty() {
error!("anon extension is not installed");
@@ -294,12 +299,12 @@ pub fn handle_extension_anon(
let query = format!("GRANT ALL ON SCHEMA anon TO {}", db_owner);
info!("granting anon extension permissions with query: {}", query);
db_client.simple_query(&query)?;
db_client.simple_query(&query).await?;
// Grant permissions to db_owner to use anon extension functions
let query = format!("GRANT ALL ON ALL FUNCTIONS IN SCHEMA anon TO {}", db_owner);
info!("granting anon extension permissions with query: {}", query);
db_client.simple_query(&query)?;
db_client.simple_query(&query).await?;
// This is needed, because some functions are defined as SECURITY DEFINER.
// In Postgres SECURITY DEFINER functions are executed with the privileges
@@ -314,16 +319,16 @@ pub fn handle_extension_anon(
where nsp.nspname = 'anon';", db_owner);
info!("change anon extension functions owner to db owner");
db_client.simple_query(&query)?;
db_client.simple_query(&query).await?;
// affects views as well
let query = format!("GRANT ALL ON ALL TABLES IN SCHEMA anon TO {}", db_owner);
info!("granting anon extension permissions with query: {}", query);
db_client.simple_query(&query)?;
db_client.simple_query(&query).await?;
let query = format!("GRANT ALL ON ALL SEQUENCES IN SCHEMA anon TO {}", db_owner);
info!("granting anon extension permissions with query: {}", query);
db_client.simple_query(&query)?;
db_client.simple_query(&query).await?;
}
}

1
control_plane/Cargo.toml
View File

@@ -33,6 +33,7 @@ postgres_backend.workspace = true
safekeeper_api.workspace = true
postgres_connection.workspace = true
storage_broker.workspace = true
http-utils.workspace = true
utils.workspace = true
whoami.workspace = true

7
control_plane/src/bin/neon_local.rs
View File

@@ -552,8 +552,10 @@ struct EndpointCreateCmdArgs {
lsn: Option<Lsn>,
#[clap(long)]
pg_port: Option<u16>,
#[clap(long, alias = "http-port")]
external_http_port: Option<u16>,
#[clap(long)]
http_port: Option<u16>,
internal_http_port: Option<u16>,
#[clap(long = "pageserver-id")]
endpoint_pageserver_id: Option<NodeId>,
@@ -1353,7 +1355,8 @@ async fn handle_endpoint(subcmd: &EndpointCmd, env: &local_env::LocalEnv) -> Res
tenant_id,
timeline_id,
args.pg_port,
args.http_port,
args.external_http_port,
args.internal_http_port,
args.pg_version,
mode,
!args.update_catalog,

127
control_plane/src/endpoint.rs
View File

@@ -37,6 +37,8 @@
//! ```
//!
use std::collections::BTreeMap;
use std::net::IpAddr;
use std::net::Ipv4Addr;
use std::net::SocketAddr;
use std::net::TcpStream;
use std::path::PathBuf;
@@ -46,6 +48,8 @@ use std::sync::Arc;
use std::time::Duration;
use anyhow::{anyhow, bail, Context, Result};
use compute_api::requests::ConfigurationRequest;
use compute_api::responses::ComputeCtlConfig;
use compute_api::spec::Database;
use compute_api::spec::PgIdent;
use compute_api::spec::RemoteExtSpec;
@@ -73,7 +77,8 @@ pub struct EndpointConf {
timeline_id: TimelineId,
mode: ComputeMode,
pg_port: u16,
http_port: u16,
external_http_port: u16,
internal_http_port: u16,
pg_version: u32,
skip_pg_catalog_updates: bool,
drop_subscriptions_before_start: bool,
@@ -128,7 +133,7 @@ impl ComputeControlPlane {
1 + self
.endpoints
.values()
.map(|ep| std::cmp::max(ep.pg_address.port(), ep.http_address.port()))
.map(|ep| std::cmp::max(ep.pg_address.port(), ep.external_http_address.port()))
.max()
.unwrap_or(self.base_port)
}
@@ -140,18 +145,27 @@ impl ComputeControlPlane {
tenant_id: TenantId,
timeline_id: TimelineId,
pg_port: Option<u16>,
http_port: Option<u16>,
external_http_port: Option<u16>,
internal_http_port: Option<u16>,
pg_version: u32,
mode: ComputeMode,
skip_pg_catalog_updates: bool,
drop_subscriptions_before_start: bool,
) -> Result<Arc<Endpoint>> {
let pg_port = pg_port.unwrap_or_else(|| self.get_port());
let http_port = http_port.unwrap_or_else(|| self.get_port() + 1);
let external_http_port = external_http_port.unwrap_or_else(|| self.get_port() + 1);
let internal_http_port = internal_http_port.unwrap_or_else(|| external_http_port + 1);
let ep = Arc::new(Endpoint {
endpoint_id: endpoint_id.to_owned(),
pg_address: SocketAddr::new("127.0.0.1".parse().unwrap(), pg_port),
http_address: SocketAddr::new("127.0.0.1".parse().unwrap(), http_port),
pg_address: SocketAddr::new(IpAddr::from(Ipv4Addr::LOCALHOST), pg_port),
external_http_address: SocketAddr::new(
IpAddr::from(Ipv4Addr::UNSPECIFIED),
external_http_port,
),
internal_http_address: SocketAddr::new(
IpAddr::from(Ipv4Addr::LOCALHOST),
internal_http_port,
),
env: self.env.clone(),
timeline_id,
mode,
@@ -176,7 +190,8 @@ impl ComputeControlPlane {
tenant_id,
timeline_id,
mode,
http_port,
external_http_port,
internal_http_port,
pg_port,
pg_version,
skip_pg_catalog_updates,
@@ -230,9 +245,10 @@ pub struct Endpoint {
pub timeline_id: TimelineId,
pub mode: ComputeMode,
// port and address of the Postgres server and `compute_ctl`'s HTTP API
// port and address of the Postgres server and `compute_ctl`'s HTTP APIs
pub pg_address: SocketAddr,
pub http_address: SocketAddr,
pub external_http_address: SocketAddr,
pub internal_http_address: SocketAddr,
// postgres major version in the format: 14, 15, etc.
pg_version: u32,
@@ -287,8 +303,15 @@ impl Endpoint {
serde_json::from_slice(&std::fs::read(entry.path().join("endpoint.json"))?)?;
Ok(Endpoint {
pg_address: SocketAddr::new("127.0.0.1".parse().unwrap(), conf.pg_port),
http_address: SocketAddr::new("127.0.0.1".parse().unwrap(), conf.http_port),
pg_address: SocketAddr::new(IpAddr::from(Ipv4Addr::LOCALHOST), conf.pg_port),
external_http_address: SocketAddr::new(
IpAddr::from(Ipv4Addr::UNSPECIFIED),
conf.external_http_port,
),
internal_http_address: SocketAddr::new(
IpAddr::from(Ipv4Addr::LOCALHOST),
conf.internal_http_port,
),
endpoint_id,
env: env.clone(),
timeline_id: conf.timeline_id,
@@ -650,24 +673,51 @@ impl Endpoint {
println!("Also at '{}'", conn_str);
}
let mut cmd = Command::new(self.env.neon_distrib_dir.join("compute_ctl"));
cmd.args(["--http-port", &self.http_address.port().to_string()])
.args(["--pgdata", self.pgdata().to_str().unwrap()])
.args(["--connstr", &conn_str])
.args([
"--spec-path",
self.endpoint_path().join("spec.json").to_str().unwrap(),
])
.args([
"--pgbin",
self.env
.pg_bin_dir(self.pg_version)?
.join("postgres")
.to_str()
.unwrap(),
])
.stdin(std::process::Stdio::null())
.stderr(logfile.try_clone()?)
.stdout(logfile);
//cmd.args([
// "--external-http-port",
// &self.external_http_address.port().to_string(),
//])
//.args([
// "--internal-http-port",
// &self.internal_http_address.port().to_string(),
//])
cmd.args([
"--http-port",
&self.external_http_address.port().to_string(),
])
.args(["--pgdata", self.pgdata().to_str().unwrap()])
.args(["--connstr", &conn_str])
.args([
"--spec-path",
self.endpoint_path().join("spec.json").to_str().unwrap(),
])
.args([
"--pgbin",
self.env
.pg_bin_dir(self.pg_version)?
.join("postgres")
.to_str()
.unwrap(),
])
// TODO: It would be nice if we generated compute IDs with the same
// algorithm as the real control plane.
//
// TODO: Add this back when
// https://github.com/neondatabase/neon/pull/10747 is merged.
//
//.args([
// "--compute-id",
// &format!(
// "compute-{}",
// SystemTime::now()
// .duration_since(UNIX_EPOCH)
// .unwrap()
// .as_secs()
// ),
//])
.stdin(std::process::Stdio::null())
.stderr(logfile.try_clone()?)
.stdout(logfile);
if let Some(remote_ext_config) = remote_ext_config {
cmd.args(["--remote-ext-config", remote_ext_config]);
@@ -754,8 +804,8 @@ impl Endpoint {
reqwest::Method::GET,
format!(
"http://{}:{}/status",
self.http_address.ip(),
self.http_address.port()
self.external_http_address.ip(),
self.external_http_address.port()
),
)
.send()
@@ -828,14 +878,17 @@ impl Endpoint {
let response = client
.post(format!(
"http://{}:{}/configure",
self.http_address.ip(),
self.http_address.port()
self.external_http_address.ip(),
self.external_http_address.port()
))
.header(CONTENT_TYPE.as_str(), "application/json")
.body(format!(
"{{\"spec\":{}}}",
serde_json::to_string_pretty(&spec)?
))
.body(
serde_json::to_string(&ConfigurationRequest {
spec,
compute_ctl_config: ComputeCtlConfig::default(),
})
.unwrap(),
)
.send()
.await?;

10
control_plane/src/pageserver.rs
View File

@@ -357,6 +357,16 @@ impl PageServerNode {
.map(serde_json::from_str)
.transpose()
.context("Failed to parse 'compaction_algorithm' json")?,
compaction_l0_first: settings
.remove("compaction_l0_first")
.map(|x| x.parse::<bool>())
.transpose()
.context("Failed to parse 'compaction_l0_first' as a bool")?,
compaction_l0_semaphore: settings
.remove("compaction_l0_semaphore")
.map(|x| x.parse::<bool>())
.transpose()
.context("Failed to parse 'compaction_l0_semaphore' as a bool")?,
l0_flush_delay_threshold: settings
.remove("l0_flush_delay_threshold")
.map(|x| x.parse::<usize>())

4
control_plane/src/safekeeper.rs
View File

@@ -17,8 +17,10 @@ use camino::Utf8PathBuf;
use postgres_connection::PgConnectionConfig;
use reqwest::{IntoUrl, Method};
use thiserror::Error;
use http_utils::error::HttpErrorBody;
use utils::auth::{Claims, Scope};
use utils::{http::error::HttpErrorBody, id::NodeId};
use utils::id::NodeId;
use crate::{
background_process,

5
control_plane/src/storage_controller.rs
View File

@@ -838,7 +838,10 @@ impl StorageController {
self.dispatch(
Method::PUT,
format!("control/v1/tenant/{tenant_shard_id}/migrate"),
Some(TenantShardMigrateRequest { node_id }),
Some(TenantShardMigrateRequest {
node_id,
migration_config: None,
}),
)
.await
}

48
control_plane/storcon_cli/src/main.rs
View File

@@ -22,7 +22,7 @@ use pageserver_api::{
};
use pageserver_client::mgmt_api::{self};
use reqwest::{Method, StatusCode, Url};
use utils::id::{NodeId, TenantId};
use utils::id::{NodeId, TenantId, TimelineId};
use pageserver_api::controller_api::{
NodeConfigureRequest, NodeRegisterRequest, NodeSchedulingPolicy, PlacementPolicy,
@@ -239,6 +239,19 @@ enum Command {
#[arg(long)]
scheduling_policy: SkSchedulingPolicyArg,
},
/// Downloads any missing heatmap layers for all shard for a given timeline
DownloadHeatmapLayers {
/// Tenant ID or tenant shard ID. When an unsharded tenant ID is specified,
/// the operation is performed on all shards. When a sharded tenant ID is
/// specified, the operation is only performed on the specified shard.
#[arg(long)]
tenant_shard_id: TenantShardId,
#[arg(long)]
timeline_id: TimelineId,
/// Optional: Maximum download concurrency (default is 16)
#[arg(long)]
concurrency: Option<usize>,
},
}
#[derive(Parser)]
@@ -609,7 +622,10 @@ async fn main() -> anyhow::Result<()> {
tenant_shard_id,
node,
} => {
let req = TenantShardMigrateRequest { node_id: node };
let req = TenantShardMigrateRequest {
node_id: node,
migration_config: None,
};
storcon_client
.dispatch::<TenantShardMigrateRequest, TenantShardMigrateResponse>(
@@ -623,7 +639,10 @@ async fn main() -> anyhow::Result<()> {
tenant_shard_id,
node,
} => {
let req = TenantShardMigrateRequest { node_id: node };
let req = TenantShardMigrateRequest {
node_id: node,
migration_config: None,
};
storcon_client
.dispatch::<TenantShardMigrateRequest, TenantShardMigrateResponse>(
@@ -1082,7 +1101,10 @@ async fn main() -> anyhow::Result<()> {
.dispatch::<TenantShardMigrateRequest, TenantShardMigrateResponse>(
Method::PUT,
format!("control/v1/tenant/{}/migrate", mv.tenant_shard_id),
Some(TenantShardMigrateRequest { node_id: mv.to }),
Some(TenantShardMigrateRequest {
node_id: mv.to,
migration_config: None,
}),
)
.await
.map_err(|e| (mv.tenant_shard_id, mv.from, mv.to, e))
@@ -1238,6 +1260,24 @@ async fn main() -> anyhow::Result<()> {
String::from(scheduling_policy)
);
}
Command::DownloadHeatmapLayers {
tenant_shard_id,
timeline_id,
concurrency,
} => {
let mut path = format!(
"/v1/tenant/{}/timeline/{}/download_heatmap_layers",
tenant_shard_id, timeline_id,
);
if let Some(c) = concurrency {
path = format!("{path}?concurrency={c}");
}
storcon_client
.dispatch::<(), ()>(Method::POST, path, None)
.await?;
}
}
Ok(())

2
docker-compose/docker_compose_test.sh
View File

@@ -71,7 +71,7 @@ for pg_version in ${TEST_VERSION_ONLY-14 15 16 17}; do
cat ../compute/patches/contrib_pg${pg_version}.patch | docker exec -i $TEST_CONTAINER_NAME bash -c "(cd /postgres && patch -p1)"
# We are running tests now
rm -f testout.txt testout_contrib.txt
docker exec -e USE_PGXS=1 -e SKIP=timescaledb-src,rdkit-src,postgis-src,pgx_ulid-src,pgtap-src,pg_tiktoken-src,pg_jsonschema-src,kq_imcx-src,wal2json_2_5-src \
docker exec -e USE_PGXS=1 -e SKIP=timescaledb-src,rdkit-src,postgis-src,pgx_ulid-src,pg_tiktoken-src,pg_jsonschema-src,kq_imcx-src,wal2json_2_5-src \
$TEST_CONTAINER_NAME /run-tests.sh /ext-src | tee testout.txt && EXT_SUCCESS=1 || EXT_SUCCESS=0
docker exec -e SKIP=start-scripts,postgres_fdw,ltree_plpython,jsonb_plpython,jsonb_plperl,hstore_plpython,hstore_plperl,dblink,bool_plperl \
$TEST_CONTAINER_NAME /run-tests.sh /postgres/contrib | tee testout_contrib.txt && CONTRIB_SUCCESS=1 || CONTRIB_SUCCESS=0

2
docker-compose/ext-src/pgjwt-src/test-upgrade.sh
View File

@@ -2,4 +2,4 @@
set -ex
cd "$(dirname ${0})"
patch -p1 <test-upgrade.patch
pg_prove test.sql
pg_prove -d contrib_regression test.sql

15
docker-compose/ext-src/pgtap-src/test-upgrade.patch Normal file
View File

@@ -0,0 +1,15 @@
diff --git a/test/schedule/create.sql b/test/schedule/create.sql
index ba355ed..7e250f5 100644
--- a/test/schedule/create.sql
+++ b/test/schedule/create.sql
@@ -1,3 +1,2 @@
\unset ECHO
\i test/psql.sql
-CREATE EXTENSION pgtap;
diff --git a/test/schedule/main.sch b/test/schedule/main.sch
index a8a5fbc..0463fc4 100644
--- a/test/schedule/main.sch
+++ b/test/schedule/main.sch
@@ -1,2 +1 @@
-test: build
test: create

6
docker-compose/ext-src/pgtap-src/test-upgrade.sh Executable file
View File

@@ -0,0 +1,6 @@
#!/bin/sh
set -ex
cd "$(dirname ${0})"
patch -p1 <test-upgrade.patch
PG_REGRESS=$(dirname "$(pg_config --pgxs)")/../test/regress/pg_regress
${PG_REGRESS} --inputdir=./ --bindir='/usr/local/pgsql/bin' --inputdir=test --max-connections=86 --schedule test/schedule/main.sch --schedule test/build/run.sch --dbname contrib_regression --use-existing

17
docker-compose/test_extensions_upgrade.sh
View File

@@ -11,6 +11,7 @@ if [ -z ${OLDTAG+x} ] || [ -z ${NEWTAG+x} ] || [ -z "${OLDTAG}" ] || [ -z "${NEW
exit 1
fi
export PG_VERSION=${PG_VERSION:-16}
export PG_TEST_VERSION=${PG_VERSION}
function wait_for_ready {
TIME=0
while ! docker compose logs compute_is_ready | grep -q "accepting connections" && [ ${TIME} -le 300 ] ; do
@@ -41,7 +42,8 @@ EXTENSIONS='[
{"extname": "roaringbitmap", "extdir": "pg_roaringbitmap-src"},
{"extname": "semver", "extdir": "pg_semver-src"},
{"extname": "pg_ivm", "extdir": "pg_ivm-src"},
{"extname": "pgjwt", "extdir": "pgjwt-src"}
{"extname": "pgjwt", "extdir": "pgjwt-src"},
{"extname": "pgtap", "extdir": "pgtap-src"}
]'
EXTNAMES=$(echo ${EXTENSIONS} | jq -r '.[].extname' | paste -sd ' ' -)
TAG=${NEWTAG} docker compose --profile test-extensions up --quiet-pull --build -d
@@ -58,8 +60,12 @@ docker compose cp ext-src neon-test-extensions:/
docker compose exec neon-test-extensions psql -c "DROP DATABASE IF EXISTS contrib_regression"
docker compose exec neon-test-extensions psql -c "CREATE DATABASE contrib_regression"
create_extensions "${EXTNAMES}"
query="select pge.extname from pg_extension pge join (select key as extname, value as extversion from json_each_text('${new_vers}')) x on pge.extname=x.extname and pge.extversion <> x.extversion"
exts=$(docker compose exec neon-test-extensions psql -Aqt -d contrib_regression -c "$query")
if [ "${FORCE_ALL_UPGRADE_TESTS:-false}" = true ]; then
exts="${EXTNAMES}"
else
query="select pge.extname from pg_extension pge join (select key as extname, value as extversion from json_each_text('${new_vers}')) x on pge.extname=x.extname and pge.extversion <> x.extversion"
exts=$(docker compose exec neon-test-extensions psql -Aqt -d contrib_regression -c "$query")
fi
if [ -z "${exts}" ]; then
echo "No extensions were upgraded"
else
@@ -87,7 +93,10 @@ else
exit 1
fi
docker compose exec neon-test-extensions psql -d contrib_regression -c "\dx ${ext}"
docker compose exec neon-test-extensions sh -c /ext-src/${EXTDIR}/test-upgrade.sh
if ! docker compose exec neon-test-extensions sh -c /ext-src/${EXTDIR}/test-upgrade.sh; then
docker compose exec neon-test-extensions cat /ext-src/${EXTDIR}/regression.diffs
exit 1
fi
docker compose exec neon-test-extensions psql -d contrib_regression -c "alter extension ${ext} update"
docker compose exec neon-test-extensions psql -d contrib_regression -c "\dx ${ext}"
done

215
docs/rfcs/035-safekeeper-dynamic-membership-change.md
View File

@@ -285,10 +285,10 @@ To summarize, list of cplane changes:
### storage_controller implementation
Current 'load everything on startup and keep in memory' easy design is fine.
Single timeline shouldn't take more than 100 bytes (it's 16 byte tenant_id, 16
byte timeline_id, int generation, vec of ~3 safekeeper ids plus some flags), so
10^6 of timelines shouldn't take more than 100MB.
If desired, we may continue using current 'load everything on startup and keep
in memory' approach: single timeline shouldn't take more than 100 bytes (it's 16
byte tenant_id, 16 byte timeline_id, int generation, vec of ~3 safekeeper ids
plus some flags), so 10^6 of timelines shouldn't take more than 100MB.
Similar to pageserver attachment Intents storage_controller would have in-memory
`MigrationRequest` (or its absense) for each timeline and pool of tasks trying
@@ -296,7 +296,7 @@ to make these request reality; this ensures one instance of storage_controller
won't do several migrations on the same timeline concurrently. In the first
version it is simpler to have more manual control and no retries, i.e. migration
failure removes the request. Later we can build retries and automatic
scheduling/migration. `MigrationRequest` is
scheduling/migration around. `MigrationRequest` is
```
enum MigrationRequest {
To(Vec<NodeId>),
@@ -313,9 +313,9 @@ similarly, in the first version it is ok to trigger it manually).
#### Schema
`safekeepers` table mirroring current `nodes` should be added, except that for
`scheduling_policy` field (seems like `status` is a better name for it): it is enough
to have at least in the beginning only 3 fields: 1) `active` 2) `offline` 3)
`decomissioned`.
`scheduling_policy`: it is enough to have at least in the beginning only 3
fields: 1) `active` 2) `paused` (initially means only not assign new tlis there
3) `decomissioned` (node is removed).
`timelines` table:
```
@@ -324,18 +324,24 @@ table! {
timelines (tenant_id, timeline_id) {
timeline_id -> Varchar,
tenant_id -> Varchar,
start_lsn -> pg_lsn,
generation -> Int4,
sk_set -> Array<Int4>, // list of safekeeper ids
new_sk_set -> Nullable<Array<Int4>>, // list of safekeeper ids, null if not joint conf
new_sk_set -> Nullable<Array<Int8>>, // list of safekeeper ids, null if not joint conf
cplane_notified_generation -> Int4,
deleted_at -> Nullable<Timestamptz>,
}
}
```
`start_lsn` is needed to create timeline on safekeepers properly, see below. We
might also want to add ancestor_timeline_id to preserve the hierarchy, but for
this RFC it is not needed.
#### API
Node management is similar to pageserver:
1) POST `/control/v1/safekeepers` upserts safekeeper.
1) POST `/control/v1/safekeepers` inserts safekeeper.
2) GET `/control/v1/safekeepers` lists safekeepers.
3) GET `/control/v1/safekeepers/:node_id` gets safekeeper.
4) PUT `/control/v1/safekepers/:node_id/status` changes status to e.g.
@@ -345,25 +351,15 @@ Node management is similar to pageserver:
Safekeeper deploy scripts should register safekeeper at storage_contorller as
they currently do with cplane, under the same id.
Timeline creation/deletion: already existing POST `tenant/:tenant_id/timeline`
would 1) choose initial set of safekeepers; 2) write to the db initial
`Configuration` with `INSERT ON CONFLICT DO NOTHING` returning existing row in
case of conflict; 3) create timeline on the majority of safekeepers (already
created is ok).
Timeline creation/deletion will work through already existing POST and DELETE
`tenant/:tenant_id/timeline`. Cplane is expected to retry both until they
succeed. See next section on the implementation details.
We don't want to block timeline creation when one safekeeper is down. Currently
this is solved by compute implicitly creating timeline on any safekeeper it is
connected to. This creates ugly timeline state on safekeeper when timeline is
created, but start LSN is not defined yet. It would be nice to remove this; to
do that, controller can in the background retry to create timeline on
safekeeper(s) which missed that during initial creation call. It can do that
through `pull_timeline` from majority so it doesn't need to remember
`parent_lsn` in its db.
Timeline deletion removes the row from the db and forwards deletion to the
current configuration members. Without additional actions deletions might leak,
see below on this; initially let's ignore these, reporting to cplane success if
at least one safekeeper deleted the timeline (this will remove s3 data).
We don't want to block timeline creation/deletion when one safekeeper is down.
Currently this is crutched by compute implicitly creating timeline on any
safekeeper it is connected to. This creates ugly timeline state on safekeeper
when timeline is created, but start LSN is not defined yet. Next section
describes dealing with this.
Tenant deletion repeats timeline deletion for all timelines.
@@ -395,26 +391,6 @@ Similar call should be added for the tenant.
It would be great to have some way of subscribing to the results (apart from
looking at logs/metrics).
Migration is executed as described above. One subtlety is that (local) deletion on
source safekeeper might fail, which is not a problem if we are going to
decomission the node but leaves garbage otherwise. I'd propose in the first version
1) Don't attempt deletion at all if node status is `offline`.
2) If it failed, just issue warning.
And add PUT `/control/v1/safekeepers/:node_id/scrub` endpoint which would find and
remove garbage timelines for manual use. It will 1) list all timelines on the
safekeeper 2) compare each one against configuration storage: if timeline
doesn't exist at all (had been deleted), it can be deleted. Otherwise, it can
be deleted under generation number if node is not member of current generation.
Automating this is untrivial; we'd need to register all potential missing
deletions <tenant_id, timeline_id, generation, node_id> in the same transaction
which switches configurations. Similarly when timeline is fully deleted to
prevent cplane operation from blocking when some safekeeper is not available
deletion should be also registered.
One more task pool should infinitely retry notifying control plane about changed
safekeeper sets.
3) GET `/control/v1/tenant/:tenant_id/timeline/:timeline_id/` should return
current in memory state of the timeline and pending `MigrationRequest`,
if any.
@@ -423,12 +399,153 @@ safekeeper sets.
migration by switching configuration from the joint to the one with (previous) `sk_set` under CAS
(incrementing generation as always).
#### API implementation and reconciliation
For timeline creation/deletion we want to preserve the basic assumption that
unreachable minority (1 sk of 3) doesn't block their completion, but eventually
we want to finish creation/deletion on nodes which missed it (unless they are
removed). Similarly for migration; it may and should finish even though excluded
members missed their exclusion. And of course e.g. such pending exclusion on
node C after migration ABC -> ABD must not prevent next migration ABD -> ABE. As
another example, if some node missed timeline creation it clearly must not block
migration from it. Hence it is natural to have per safekeeper background
reconciler which retries these ops until they succeed. There are 3 possible
operation types, and the type is defined by timeline state (membership
configuration and whether it is deleted) and safekeeper id: we may need to
create timeline on sk (node added), locally delete it (node excluded, somewhat
similar to detach) or globally delete it (timeline is deleted).
Next, on storage controller restart in principle these pending operations can be
figured out by comparing safekeepers state against storcon state. But it seems
better to me to materialize them in the database; it is not expensive, avoids
these startup scans which themselves can fail etc and makes it very easy to see
outstanding work directly at the source of truth -- the db. So we can add table
`safekeeper_timeline_pending_ops`
```
table! {
// timeline_id, sk_id is primary key
safekeeper_timeline_pending_ops (sk_id, tenant_id, timeline_id) {
sk_id -> int8,
tenant_id -> Varchar,
timeline_id -> Varchar,
generation -> Int4,
op_type -> Varchar,
}
}
```
`op_type` can be `include` (seed from peers and ensure generation is up to
date), `exclude` (remove locally) and `delete`. Field is actually not strictly
needed as it can be computed from current configuration, but gives more explicit
observability.
`generation` is necessary there because after op is done reconciler must remove
it and not remove another row with higher gen which in theory might appear.
Any insert of row should overwrite (remove) all rows with the same sk and
timeline id but lower `generation` as next op makes previous obsolete. Insertion
of `op_type` `delete` overwrites all rows.
About `exclude`: rather than adding explicit safekeeper http endpoint, it is
reasonable to reuse membership switch endpoint: if safekeeper is not member
of the configuration it locally removes the timeline on the switch. In this case
404 should also be considered an 'ok' answer by the caller.
So, main loop of per sk reconcile reads `safekeeper_timeline_pending_ops`
joined with timeline configuration to get current conf (with generation `n`)
for the safekeeper and does the jobs, infinitely retrying failures:
1) If node is member (`include`):
- Check if timeline exists on it, if not, call pull_timeline on it from
other members
- Call switch configuration to the current
2) If node is not member (`exclude`):
- Call switch configuration to the current, 404 is ok.
3) If timeline is deleted (`delete`), call delete.
In cases 1 and 2 remove `safekeeper_timeline_pending_ops` for the sk and
timeline with generation <= `n` if `op_type` is not `delete`.
In case 3 also remove `safekeeper_timeline_pending_ops`
entry + remove `timelines` entry if there is nothing left in `safekeeper_timeline_pending_ops` for the timeline.
Let's consider in details how APIs can be implemented from this angle.
Timeline creation. It is assumed that cplane retries it until success, so all
actions must be idempotent. Now, a tricky point here is timeline start LSN. For
the initial (tenant creation) call cplane doesn't know it. However, setting
start_lsn on safekeepers during creation is a good thing -- it provides a
guarantee that walproposer can always find a common point in WAL histories of
safekeeper and its own, and so absense of it would be a clear sign of
corruption. The following sequence works:
1) Create timeline (or observe that it exists) on pageserver,
figuring out last_record_lsn in response.
2) Choose safekeepers and insert (ON CONFLICT DO NOTHING) timeline row into the
db. Note that last_record_lsn returned on the previous step is movable as it
changes once ingestion starts, insert must not overwrite it (as well as other
fields like membership conf). On the contrary, start_lsn used in the next
step must be set to the value in the db. cplane_notified_generation can be set
to 1 (initial generation) in insert to avoid notifying cplane about initial
conf as cplane will receive it in timeline creation request anyway.
3) Issue timeline creation calls to at least majority of safekeepers. Using
majority here is not necessary but handy because it guarantees that any live
majority will have at least one sk with created timeline and so
reconciliation task can use pull_timeline shared with migration instead of
create timeline special init case. OFC if timeline is already exists call is
ignored.
4) For minority of safekeepers which could have missed creation insert
entries to `safekeeper_timeline_pending_ops`. We won't miss this insertion
because response to cplane is sent only after it has happened, and cplane
retries the call until 200 response.
There is a small question how request handler (timeline creation in this
case) would interact with per sk reconciler. As always I prefer to do the
simplest possible thing and here it seems to be just waking it up so it
re-reads the db for work to do. Passing work in memory is faster, but
that shouldn't matter, and path to scan db for work will exist anyway,
simpler to reuse it.
For pg version / wal segment size: while we may persist them in `timelines`
table, it is not necessary as initial creation at step 3 can take them from
pageserver or cplane creation call and later pull_timeline will carry them
around.
Timeline migration.
1) CAS to the db to create joint conf, and in the same transaction create
`safekeeper_timeline_pending_ops` `include` entries to initialize new members
as well as deliver this conf to current ones; poke per sk reconcilers to work
on it. Also any conf change should also poke cplane notifier task(s).
2) Once it becomes possible per alg description above, get out of joint conf
with another CAS. Task should get wakeups from per sk reconcilers because
conf switch is required for advancement; however retries should be sleep
based as well as LSN advancement might be needed, though in happy path
it isn't. To see whether further transition is possible on wakup migration
executor polls safekeepers per the algorithm. CAS creating new conf with only
new members should again insert entries to `safekeeper_timeline_pending_ops`
to switch them there, as well as `exclude` rows to remove timeline from
old members.
Timeline deletion: just set `deleted_at` on the timeline row and insert
`safekeeper_timeline_pending_ops` entries in the same xact, the rest is done by
per sk reconcilers.
When node is removed (set to `decomissioned`), `safekeeper_timeline_pending_ops`
for it must be cleared in the same transaction.
One more task pool should infinitely retry notifying control plane about changed
safekeeper sets (trying making `cplane_notified_generation` equal `generation`).
#### Dealing with multiple instances of storage_controller
Operations described above executed concurrently might create some errors but do
not prevent progress, so while we normally don't want to run multiple instances
of storage_controller it is fine to have it temporarily, e.g. during redeploy.
To harden against some controller instance creating some work in
`safekeeper_timeline_pending_ops` and then disappearing without anyone pickup up
the job per sk reconcilers apart from explicit wakups should scan for work
periodically. It is possible to remove that though if all db updates are
protected with leadership token/term -- then such scans are needed only after
leadership is acquired.
Any interactions with db update in-memory controller state, e.g. if migration
request failed because different one is in progress, controller remembers that
and tries to finish it.
@@ -545,7 +662,7 @@ Aurora does this but similarly I don't think this is needed.
We should use Compute <-> safekeeper protocol change to include other (long
yearned) modifications:
- send data in network order to make arm work.
- send data in network order without putting whole structs to be arch independent
- remove term_start_lsn from AppendRequest
- add horizon to TermHistory
- add to ProposerGreeting number of connection from this wp to sk

1
libs/compute_api/Cargo.toml
View File

@@ -7,6 +7,7 @@ license.workspace = true
[dependencies]
anyhow.workspace = true
chrono.workspace = true
jsonwebtoken.workspace = true
serde.workspace = true
serde_json.workspace = true
regex.workspace = true

6
libs/compute_api/src/requests.rs
View File

@@ -1,18 +1,20 @@
//! Structs representing the JSON formats used in the compute_ctl's HTTP API.
use crate::{
privilege::Privilege,
responses::ComputeCtlConfig,
spec::{ComputeSpec, ExtVersion, PgIdent},
};
use serde::Deserialize;
use serde::{Deserialize, Serialize};
/// Request of the /configure API
///
/// We now pass only `spec` in the configuration request, but later we can
/// extend it and something like `restart: bool` or something else. So put
/// `spec` into a struct initially to be more flexible in the future.
#[derive(Deserialize, Debug)]
#[derive(Debug, Deserialize, Serialize)]
pub struct ConfigurationRequest {
pub spec: ComputeSpec,
pub compute_ctl_config: ComputeCtlConfig,
}
#[derive(Deserialize, Debug)]

19
libs/compute_api/src/responses.rs
View File

@@ -3,6 +3,7 @@
use std::fmt::Display;
use chrono::{DateTime, Utc};
use jsonwebtoken::jwk::JwkSet;
use serde::{Deserialize, Serialize, Serializer};
use crate::{
@@ -135,13 +136,27 @@ pub struct CatalogObjects {
pub databases: Vec<Database>,
}
#[derive(Debug, Deserialize, Serialize)]
pub struct ComputeCtlConfig {
pub jwks: JwkSet,
}
impl Default for ComputeCtlConfig {
fn default() -> Self {
Self {
jwks: JwkSet {
keys: Vec::default(),
},
}
}
}
/// Response of the `/computes/{compute_id}/spec` control-plane API.
/// This is not actually a compute API response, so consider moving
/// to a different place.
#[derive(Deserialize, Debug)]
pub struct ControlPlaneSpecResponse {
pub spec: Option<ComputeSpec>,
pub status: ControlPlaneComputeStatus,
pub compute_ctl_config: ComputeCtlConfig,
}
#[derive(Deserialize, Clone, Copy, Debug, PartialEq, Eq)]

37
libs/http-utils/Cargo.toml Normal file
View File

@@ -0,0 +1,37 @@
[package]
name = "http-utils"
version = "0.1.0"
edition.workspace = true
license.workspace = true
[dependencies]
anyhow.workspace = true
backtrace.workspace = true
bytes.workspace = true
inferno.workspace = true
fail.workspace = true
flate2.workspace = true
hyper0.workspace = true
itertools.workspace = true
jemalloc_pprof.workspace = true
once_cell.workspace = true
pprof.workspace = true
regex.workspace = true
routerify.workspace = true
serde.workspace = true
serde_json.workspace = true
serde_path_to_error.workspace = true
thiserror.workspace = true
tracing.workspace = true
tokio.workspace = true
tokio-util.workspace = true
url.workspace = true
uuid.workspace = true
# to use tokio channels as streams, this is faster to compile than async_stream
# why is it only here? no other crate should use it, streams are rarely needed.
tokio-stream = { version = "0.1.14" }
metrics.workspace = true
utils.workspace = true
workspace_hack = { version = "0.1", path = "../../workspace_hack" }

8
libs/utils/src/http/endpoint.rs → libs/http-utils/src/endpoint.rs
View File

@@ -1,7 +1,6 @@
use crate::auth::{AuthError, Claims, SwappableJwtAuth};
use crate::http::error::{api_error_handler, route_error_handler, ApiError};
use crate::http::request::{get_query_param, parse_query_param};
use crate::error::{api_error_handler, route_error_handler, ApiError};
use crate::pprof;
use crate::request::{get_query_param, parse_query_param};
use ::pprof::protos::Message as _;
use ::pprof::ProfilerGuardBuilder;
use anyhow::{anyhow, Context};
@@ -19,6 +18,7 @@ use tokio::sync::{mpsc, Mutex, Notify};
use tokio_stream::wrappers::ReceiverStream;
use tokio_util::io::ReaderStream;
use tracing::{debug, info, info_span, warn, Instrument};
use utils::auth::{AuthError, Claims, SwappableJwtAuth};
use std::future::Future;
use std::io::Write as _;
@@ -718,9 +718,9 @@ pub fn check_permission_with(
#[cfg(test)]
mod tests {
use super::*;
use futures::future::poll_fn;
use hyper::service::Service;
use routerify::RequestServiceBuilder;
use std::future::poll_fn;
use std::net::{IpAddr, SocketAddr};
#[tokio::test]

11
libs/utils/src/http/error.rs → libs/http-utils/src/error.rs
View File

@@ -5,6 +5,8 @@ use std::error::Error as StdError;
use thiserror::Error;
use tracing::{error, info, warn};
use utils::auth::AuthError;
#[derive(Debug, Error)]
pub enum ApiError {
#[error("Bad request: {0:#?}")]
@@ -96,6 +98,15 @@ impl ApiError {
}
}
impl From<AuthError> for ApiError {
fn from(_value: AuthError) -> Self {
// Don't pass on the value of the AuthError as a precautionary measure.
// Being intentionally vague in public error communication hurts debugability
// but it is more secure.
ApiError::Forbidden("JWT authentication error".to_string())
}
}
#[derive(Serialize, Deserialize)]
pub struct HttpErrorBody {
pub msg: String,

50
libs/http-utils/src/failpoints.rs Normal file
View File

@@ -0,0 +1,50 @@
use crate::error::ApiError;
use crate::json::{json_request, json_response};
use hyper::{Body, Request, Response, StatusCode};
use serde::{Deserialize, Serialize};
use tokio_util::sync::CancellationToken;
use utils::failpoint_support::apply_failpoint;
pub type ConfigureFailpointsRequest = Vec<FailpointConfig>;
/// Information for configuring a single fail point
#[derive(Debug, Serialize, Deserialize)]
pub struct FailpointConfig {
/// Name of the fail point
pub name: String,
/// List of actions to take, using the format described in `fail::cfg`
///
/// We also support `actions = "exit"` to cause the fail point to immediately exit.
pub actions: String,
}
/// Configure failpoints through http.
pub async fn failpoints_handler(
mut request: Request<Body>,
_cancel: CancellationToken,
) -> Result<Response<Body>, ApiError> {
if !fail::has_failpoints() {
return Err(ApiError::BadRequest(anyhow::anyhow!(
"Cannot manage failpoints because neon was compiled without failpoints support"
)));
}
let failpoints: ConfigureFailpointsRequest = json_request(&mut request).await?;
for fp in failpoints {
tracing::info!("cfg failpoint: {} {}", fp.name, fp.actions);
// We recognize one extra "action" that's not natively recognized
// by the failpoints crate: exit, to immediately kill the process
let cfg_result = apply_failpoint(&fp.name, &fp.actions);
if let Err(err_msg) = cfg_result {
return Err(ApiError::BadRequest(anyhow::anyhow!(
"Failed to configure failpoints: {err_msg}"
)));
}
}
json_response(StatusCode::OK, ())
}

0
libs/utils/src/http/json.rs → libs/http-utils/src/json.rs
View File

4
libs/utils/src/http/mod.rs → libs/http-utils/src/lib.rs
View File

@@ -1,8 +1,12 @@
pub mod endpoint;
pub mod error;
pub mod failpoints;
pub mod json;
pub mod pprof;
pub mod request;
extern crate hyper0 as hyper;
/// Current fast way to apply simple http routing in various Neon binaries.
/// Re-exported for sake of uniform approach, that could be later replaced with better alternatives, if needed.
pub use routerify::{ext::RequestExt, RouterBuilder, RouterService};

0
libs/utils/src/pprof.rs → libs/http-utils/src/pprof.rs
View File

0
libs/utils/src/http/request.rs → libs/http-utils/src/request.rs
View File

23
libs/pageserver_api/src/config.rs
View File

@@ -121,6 +121,7 @@ pub struct ConfigToml {
pub wal_receiver_protocol: PostgresClientProtocol,
pub page_service_pipelining: PageServicePipeliningConfig,
pub get_vectored_concurrent_io: GetVectoredConcurrentIo,
pub enable_read_path_debugging: Option<bool>,
}
#[derive(Debug, Clone, PartialEq, Eq, serde::Serialize, serde::Deserialize)]
@@ -262,6 +263,11 @@ pub struct TenantConfigToml {
/// size exceeds `compaction_upper_limit * checkpoint_distance`.
pub compaction_upper_limit: usize,
pub compaction_algorithm: crate::models::CompactionAlgorithmSettings,
/// If true, compact down L0 across all tenant timelines before doing regular compaction.
pub compaction_l0_first: bool,
/// If true, use a separate semaphore (i.e. concurrency limit) for the L0 compaction pass. Only
/// has an effect if `compaction_l0_first` is `true`.
pub compaction_l0_semaphore: bool,
/// Level0 delta layer threshold at which to delay layer flushes for compaction backpressure,
/// such that they take 2x as long, and start waiting for layer flushes during ephemeral layer
/// rolls. This helps compaction keep up with WAL ingestion, and avoids read amplification
@@ -345,7 +351,7 @@ pub struct TenantConfigToml {
/// Enable rel_size_v2 for this tenant. Once enabled, the tenant will persist this information into
/// `index_part.json`, and it cannot be reversed.
pub rel_size_v2_enabled: Option<bool>,
pub rel_size_v2_enabled: bool,
// gc-compaction related configs
/// Enable automatic gc-compaction trigger on this tenant.
@@ -490,7 +496,7 @@ impl Default for ConfigToml {
NonZeroUsize::new(DEFAULT_MAX_VECTORED_READ_BYTES).unwrap(),
)),
image_compression: (DEFAULT_IMAGE_COMPRESSION),
timeline_offloading: false,
timeline_offloading: true,
ephemeral_bytes_per_memory_kb: (DEFAULT_EPHEMERAL_BYTES_PER_MEMORY_KB),
l0_flush: None,
virtual_file_io_mode: None,
@@ -510,6 +516,11 @@ impl Default for ConfigToml {
} else {
GetVectoredConcurrentIo::SidecarTask
},
enable_read_path_debugging: if cfg!(test) || cfg!(feature = "testing") {
Some(true)
} else {
None
},
}
}
}
@@ -537,6 +548,8 @@ pub mod tenant_conf_defaults {
// most of our pageservers. Compaction ~50 layers requires about 2GB memory (could be reduced later by optimizing L0 hole
// calculation to avoid loading all keys into the memory). So with this config, we can get a maximum peak compaction usage of 18GB.
pub const DEFAULT_COMPACTION_UPPER_LIMIT: usize = 50;
pub const DEFAULT_COMPACTION_L0_FIRST: bool = false;
pub const DEFAULT_COMPACTION_L0_SEMAPHORE: bool = true;
pub const DEFAULT_COMPACTION_ALGORITHM: crate::models::CompactionAlgorithm =
crate::models::CompactionAlgorithm::Legacy;
@@ -586,6 +599,8 @@ impl Default for TenantConfigToml {
compaction_algorithm: crate::models::CompactionAlgorithmSettings {
kind: DEFAULT_COMPACTION_ALGORITHM,
},
compaction_l0_first: DEFAULT_COMPACTION_L0_FIRST,
compaction_l0_semaphore: DEFAULT_COMPACTION_L0_SEMAPHORE,
l0_flush_delay_threshold: None,
l0_flush_stall_threshold: None,
l0_flush_wait_upload: DEFAULT_L0_FLUSH_WAIT_UPLOAD,
@@ -616,9 +631,9 @@ impl Default for TenantConfigToml {
image_creation_preempt_threshold: DEFAULT_IMAGE_CREATION_PREEMPT_THRESHOLD,
lsn_lease_length: LsnLease::DEFAULT_LENGTH,
lsn_lease_length_for_ts: LsnLease::DEFAULT_LENGTH_FOR_TS,
timeline_offloading: false,
timeline_offloading: true,
wal_receiver_protocol_override: None,
rel_size_v2_enabled: None,
rel_size_v2_enabled: false,
gc_compaction_enabled: DEFAULT_GC_COMPACTION_ENABLED,
gc_compaction_initial_threshold_kb: DEFAULT_GC_COMPACTION_INITIAL_THRESHOLD_KB,
gc_compaction_ratio_percent: DEFAULT_GC_COMPACTION_RATIO_PERCENT,

12
libs/pageserver_api/src/controller_api.rs
View File

@@ -182,6 +182,18 @@ pub struct TenantDescribeResponseShard {
#[derive(Serialize, Deserialize, Debug)]
pub struct TenantShardMigrateRequest {
pub node_id: NodeId,
#[serde(default)]
pub migration_config: Option<MigrationConfig>,
}
#[derive(Serialize, Deserialize, Debug)]
pub struct MigrationConfig {
#[serde(default)]
#[serde(with = "humantime_serde")]
pub secondary_warmup_timeout: Option<Duration>,
#[serde(default)]
#[serde(with = "humantime_serde")]
pub secondary_download_request_timeout: Option<Duration>,
}
#[derive(Serialize, Clone, Debug)]

112
libs/pageserver_api/src/key.rs
View File

@@ -1,10 +1,12 @@
use anyhow::{bail, Result};
use byteorder::{ByteOrder, BE};
use bytes::Bytes;
use postgres_ffi::relfile_utils::{FSM_FORKNUM, VISIBILITYMAP_FORKNUM};
use postgres_ffi::Oid;
use postgres_ffi::RepOriginId;
use serde::{Deserialize, Serialize};
use std::{fmt, ops::Range};
use utils::const_assert;
use crate::reltag::{BlockNumber, RelTag, SlruKind};
@@ -49,6 +51,64 @@ pub const AUX_KEY_PREFIX: u8 = 0x62;
/// The key prefix of ReplOrigin keys.
pub const REPL_ORIGIN_KEY_PREFIX: u8 = 0x63;
/// The key prefix of db directory keys.
pub const DB_DIR_KEY_PREFIX: u8 = 0x64;
/// The key prefix of rel directory keys.
pub const REL_DIR_KEY_PREFIX: u8 = 0x65;
#[derive(Debug, Clone, Copy, Hash, PartialEq, Eq)]
pub enum RelDirExists {
Exists,
Removed,
}
#[derive(Debug)]
pub struct DecodeError;
impl fmt::Display for DecodeError {
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
write!(f, "invalid marker")
}
}
impl std::error::Error for DecodeError {}
impl RelDirExists {
/// The value of the rel directory keys that indicates the existence of a relation.
const REL_EXISTS_MARKER: Bytes = Bytes::from_static(b"r");
pub fn encode(&self) -> Bytes {
match self {
Self::Exists => Self::REL_EXISTS_MARKER.clone(),
Self::Removed => SPARSE_TOMBSTONE_MARKER.clone(),
}
}
pub fn decode_option(data: Option<impl AsRef<[u8]>>) -> Result<Self, DecodeError> {
match data {
Some(marker) if marker.as_ref() == Self::REL_EXISTS_MARKER => Ok(Self::Exists),
// Any other marker is invalid
Some(_) => Err(DecodeError),
None => Ok(Self::Removed),
}
}
pub fn decode(data: impl AsRef<[u8]>) -> Result<Self, DecodeError> {
let data = data.as_ref();
if data == Self::REL_EXISTS_MARKER {
Ok(Self::Exists)
} else if data == SPARSE_TOMBSTONE_MARKER {
Ok(Self::Removed)
} else {
Err(DecodeError)
}
}
}
/// A tombstone in the sparse keyspace, which is an empty buffer.
pub const SPARSE_TOMBSTONE_MARKER: Bytes = Bytes::from_static(b"");
/// Check if the key falls in the range of metadata keys.
pub const fn is_metadata_key_slice(key: &[u8]) -> bool {
key[0] >= METADATA_KEY_BEGIN_PREFIX && key[0] < METADATA_KEY_END_PREFIX
@@ -110,6 +170,24 @@ impl Key {
}
}
pub fn rel_dir_sparse_key_range() -> Range<Self> {
Key {
field1: REL_DIR_KEY_PREFIX,
field2: 0,
field3: 0,
field4: 0,
field5: 0,
field6: 0,
}..Key {
field1: REL_DIR_KEY_PREFIX + 1,
field2: 0,
field3: 0,
field4: 0,
field5: 0,
field6: 0,
}
}
/// This function checks more extensively what keys we can take on the write path.
/// If a key beginning with 00 does not have a global/default tablespace OID, it
/// will be rejected on the write path.
@@ -440,6 +518,36 @@ pub fn rel_dir_to_key(spcnode: Oid, dbnode: Oid) -> Key {
}
}
#[inline(always)]
pub fn rel_tag_sparse_key(spcnode: Oid, dbnode: Oid, relnode: Oid, forknum: u8) -> Key {
Key {
field1: REL_DIR_KEY_PREFIX,
field2: spcnode,
field3: dbnode,
field4: relnode,
field5: forknum,
field6: 1,
}
}
pub fn rel_tag_sparse_key_range(spcnode: Oid, dbnode: Oid) -> Range<Key> {
Key {
field1: REL_DIR_KEY_PREFIX,
field2: spcnode,
field3: dbnode,
field4: 0,
field5: 0,
field6: 0,
}..Key {
field1: REL_DIR_KEY_PREFIX,
field2: spcnode,
field3: dbnode,
field4: u32::MAX,
field5: u8::MAX,
field6: u32::MAX,
} // it's fine to exclude the last key b/c we only use field6 == 1
}
#[inline(always)]
pub fn rel_block_to_key(rel: RelTag, blknum: BlockNumber) -> Key {
Key {
@@ -734,9 +842,9 @@ impl Key {
self.field1 == RELATION_SIZE_PREFIX
}
pub fn sparse_non_inherited_keyspace() -> Range<Key> {
pub const fn sparse_non_inherited_keyspace() -> Range<Key> {
// The two keys are adjacent; if we will have non-adjancent keys in the future, we should return a keyspace
debug_assert_eq!(AUX_KEY_PREFIX + 1, REPL_ORIGIN_KEY_PREFIX);
const_assert!(AUX_KEY_PREFIX + 1 == REPL_ORIGIN_KEY_PREFIX);
Key {
field1: AUX_KEY_PREFIX,
field2: 0,

36
libs/pageserver_api/src/models.rs
View File

@@ -464,6 +464,10 @@ pub struct TenantConfigPatch {
#[serde(skip_serializing_if = "FieldPatch::is_noop")]
pub compaction_algorithm: FieldPatch<CompactionAlgorithmSettings>,
#[serde(skip_serializing_if = "FieldPatch::is_noop")]
pub compaction_l0_first: FieldPatch<bool>,
#[serde(skip_serializing_if = "FieldPatch::is_noop")]
pub compaction_l0_semaphore: FieldPatch<bool>,
#[serde(skip_serializing_if = "FieldPatch::is_noop")]
pub l0_flush_delay_threshold: FieldPatch<usize>,
#[serde(skip_serializing_if = "FieldPatch::is_noop")]
pub l0_flush_stall_threshold: FieldPatch<usize>,
@@ -529,6 +533,8 @@ pub struct TenantConfig {
pub compaction_upper_limit: Option<usize>,
// defer parsing compaction_algorithm, like eviction_policy
pub compaction_algorithm: Option<CompactionAlgorithmSettings>,
pub compaction_l0_first: Option<bool>,
pub compaction_l0_semaphore: Option<bool>,
pub l0_flush_delay_threshold: Option<usize>,
pub l0_flush_stall_threshold: Option<usize>,
pub l0_flush_wait_upload: Option<bool>,
@@ -567,6 +573,8 @@ impl TenantConfig {
mut compaction_threshold,
mut compaction_upper_limit,
mut compaction_algorithm,
mut compaction_l0_first,
mut compaction_l0_semaphore,
mut l0_flush_delay_threshold,
mut l0_flush_stall_threshold,
mut l0_flush_wait_upload,
@@ -606,6 +614,10 @@ impl TenantConfig {
.compaction_upper_limit
.apply(&mut compaction_upper_limit);
patch.compaction_algorithm.apply(&mut compaction_algorithm);
patch.compaction_l0_first.apply(&mut compaction_l0_first);
patch
.compaction_l0_semaphore
.apply(&mut compaction_l0_semaphore);
patch
.l0_flush_delay_threshold
.apply(&mut l0_flush_delay_threshold);
@@ -669,6 +681,8 @@ impl TenantConfig {
compaction_threshold,
compaction_upper_limit,
compaction_algorithm,
compaction_l0_first,
compaction_l0_semaphore,
l0_flush_delay_threshold,
l0_flush_stall_threshold,
l0_flush_wait_upload,
@@ -1066,8 +1080,7 @@ pub struct TenantInfo {
/// Opaque explanation if gc is being blocked.
///
/// Only looked up for the individual tenant detail, not the listing. This is purely for
/// debugging, not included in openapi.
/// Only looked up for the individual tenant detail, not the listing.
#[serde(skip_serializing_if = "Option::is_none")]
pub gc_blocking: Option<String>,
}
@@ -1122,7 +1135,26 @@ pub struct TimelineInfo {
pub ancestor_lsn: Option<Lsn>,
pub last_record_lsn: Lsn,
pub prev_record_lsn: Option<Lsn>,
/// Legacy field for compat with control plane. Synonym of `min_readable_lsn`.
/// TODO: remove once control plane no longer reads it.
pub latest_gc_cutoff_lsn: Lsn,
/// The LSN up to which GC has advanced: older data may still exist but it is not available for clients.
/// This LSN is not suitable for deciding where to create branches etc: use [`TimelineInfo::min_readable_lsn`] instead,
/// as it is easier to reason about.
#[serde(default)]
pub applied_gc_cutoff_lsn: Lsn,
/// The upper bound of data which is either already GC'ed, or elegible to be GC'ed at any time based on PITR interval.
/// This LSN represents the "end of history" for this timeline, and callers should use it to figure out the oldest
/// LSN at which it is legal to create a branch or ephemeral endpoint.
///
/// Note that holders of valid LSN leases may be able to create branches and read pages earlier
/// than this LSN, but new leases may not be taken out earlier than this LSN.
#[serde(default)]
pub min_readable_lsn: Lsn,
pub disk_consistent_lsn: Lsn,
/// The LSN that we have succesfully uploaded to remote storage

7
libs/postgres_backend/src/lib.rs
View File

@@ -9,6 +9,8 @@ use bytes::Bytes;
use serde::{Deserialize, Serialize};
use std::io::ErrorKind;
use std::net::SocketAddr;
use std::os::fd::AsRawFd;
use std::os::fd::RawFd;
use std::pin::Pin;
use std::sync::Arc;
use std::task::{ready, Poll};
@@ -268,6 +270,7 @@ impl<IO: AsyncRead + AsyncWrite + Unpin> MaybeWriteOnly<IO> {
}
pub struct PostgresBackend<IO> {
pub socket_fd: RawFd,
framed: MaybeWriteOnly<IO>,
pub state: ProtoState,
@@ -293,9 +296,11 @@ impl PostgresBackend<tokio::net::TcpStream> {
tls_config: Option<Arc<rustls::ServerConfig>>,
) -> io::Result<Self> {
let peer_addr = socket.peer_addr()?;
let socket_fd = socket.as_raw_fd();
let stream = MaybeTlsStream::Unencrypted(socket);
Ok(Self {
socket_fd,
framed: MaybeWriteOnly::Full(Framed::new(stream)),
state: ProtoState::Initialization,
auth_type,
@@ -307,6 +312,7 @@ impl PostgresBackend<tokio::net::TcpStream> {
impl<IO: AsyncRead + AsyncWrite + Unpin> PostgresBackend<IO> {
pub fn new_from_io(
socket_fd: RawFd,
socket: IO,
peer_addr: SocketAddr,
auth_type: AuthType,
@@ -315,6 +321,7 @@ impl<IO: AsyncRead + AsyncWrite + Unpin> PostgresBackend<IO> {
let stream = MaybeTlsStream::Unencrypted(socket);
Ok(Self {
socket_fd,
framed: MaybeWriteOnly::Full(Framed::new(stream)),
state: ProtoState::Initialization,
auth_type,

98
libs/proxy/tokio-postgres2/src/client.rs
View File

@@ -10,8 +10,8 @@ use crate::simple_query::SimpleQueryStream;
use crate::types::{Oid, ToSql, Type};
use crate::{
prepare, query, simple_query, slice_iter, CancelToken, Error, ReadyForQueryStatus, Row,
SimpleQueryMessage, Statement, ToStatement, Transaction, TransactionBuilder,
query, simple_query, slice_iter, CancelToken, Error, ReadyForQueryStatus, Row,
SimpleQueryMessage, Statement, Transaction, TransactionBuilder,
};
use bytes::BytesMut;
use fallible_iterator::FallibleIterator;
@@ -54,18 +54,18 @@ impl Responses {
}
/// A cache of type info and prepared statements for fetching type info
/// (corresponding to the queries in the [prepare] module).
/// (corresponding to the queries in the [crate::prepare] module).
#[derive(Default)]
struct CachedTypeInfo {
/// A statement for basic information for a type from its
/// OID. Corresponds to [TYPEINFO_QUERY](prepare::TYPEINFO_QUERY) (or its
/// OID. Corresponds to [TYPEINFO_QUERY](crate::prepare::TYPEINFO_QUERY) (or its
/// fallback).
typeinfo: Option<Statement>,
/// A statement for getting information for a composite type from its OID.
/// Corresponds to [TYPEINFO_QUERY](prepare::TYPEINFO_COMPOSITE_QUERY).
/// Corresponds to [TYPEINFO_QUERY](crate::prepare::TYPEINFO_COMPOSITE_QUERY).
typeinfo_composite: Option<Statement>,
/// A statement for getting information for a composite type from its OID.
/// Corresponds to [TYPEINFO_QUERY](prepare::TYPEINFO_COMPOSITE_QUERY) (or
/// Corresponds to [TYPEINFO_QUERY](crate::prepare::TYPEINFO_COMPOSITE_QUERY) (or
/// its fallback).
typeinfo_enum: Option<Statement>,
@@ -190,26 +190,6 @@ impl Client {
&self.inner
}
/// Creates a new prepared statement.
///
/// Prepared statements can be executed repeatedly, and may contain query parameters (indicated by `$1`, `$2`, etc),
/// which are set when executed. Prepared statements can only be used with the connection that created them.
pub async fn prepare(&self, query: &str) -> Result<Statement, Error> {
self.prepare_typed(query, &[]).await
}
/// Like `prepare`, but allows the types of query parameters to be explicitly specified.
///
/// The list of types may be smaller than the number of parameters - the types of the remaining parameters will be
/// inferred. For example, `client.prepare_typed(query, &[])` is equivalent to `client.prepare(query)`.
pub async fn prepare_typed(
&self,
query: &str,
parameter_types: &[Type],
) -> Result<Statement, Error> {
prepare::prepare(&self.inner, query, parameter_types).await
}
/// Executes a statement, returning a vector of the resulting rows.
///
/// A statement may contain parameters, specified by `$n`, where `n` is the index of the parameter of the list
@@ -222,14 +202,11 @@ impl Client {
/// # Panics
///
/// Panics if the number of parameters provided does not match the number expected.
pub async fn query<T>(
pub async fn query(
&self,
statement: &T,
statement: Statement,
params: &[&(dyn ToSql + Sync)],
) -> Result<Vec<Row>, Error>
where
T: ?Sized + ToStatement,
{
) -> Result<Vec<Row>, Error> {
self.query_raw(statement, slice_iter(params))
.await?
.try_collect()
@@ -250,13 +227,15 @@ impl Client {
/// Panics if the number of parameters provided does not match the number expected.
///
/// [`query`]: #method.query
pub async fn query_raw<'a, T, I>(&self, statement: &T, params: I) -> Result<RowStream, Error>
pub async fn query_raw<'a, I>(
&self,
statement: Statement,
params: I,
) -> Result<RowStream, Error>
where
T: ?Sized + ToStatement,
I: IntoIterator<Item = &'a (dyn ToSql + Sync)>,
I::IntoIter: ExactSizeIterator,
{
let statement = statement.__convert().into_statement(self).await?;
query::query(&self.inner, statement, params).await
}
@@ -271,55 +250,6 @@ impl Client {
query::query_txt(&self.inner, statement, params).await
}
/// Executes a statement, returning the number of rows modified.
///
/// A statement may contain parameters, specified by `$n`, where `n` is the index of the parameter of the list
/// provided, 1-indexed.
///
/// The `statement` argument can either be a `Statement`, or a raw query string. If the same statement will be
/// repeatedly executed (perhaps with different query parameters), consider preparing the statement up front
/// with the `prepare` method.
///
/// If the statement does not modify any rows (e.g. `SELECT`), 0 is returned.
///
/// # Panics
///
/// Panics if the number of parameters provided does not match the number expected.
pub async fn execute<T>(
&self,
statement: &T,
params: &[&(dyn ToSql + Sync)],
) -> Result<u64, Error>
where
T: ?Sized + ToStatement,
{
self.execute_raw(statement, slice_iter(params)).await
}
/// The maximally flexible version of [`execute`].
///
/// A statement may contain parameters, specified by `$n`, where `n` is the index of the parameter of the list
/// provided, 1-indexed.
///
/// The `statement` argument can either be a `Statement`, or a raw query string. If the same statement will be
/// repeatedly executed (perhaps with different query parameters), consider preparing the statement up front
/// with the `prepare` method.
///
/// # Panics
///
/// Panics if the number of parameters provided does not match the number expected.
///
/// [`execute`]: #method.execute
pub async fn execute_raw<'a, T, I>(&self, statement: &T, params: I) -> Result<u64, Error>
where
T: ?Sized + ToStatement,
I: IntoIterator<Item = &'a (dyn ToSql + Sync)>,
I::IntoIter: ExactSizeIterator,
{
let statement = statement.__convert().into_statement(self).await?;
query::execute(self.inner(), statement, params).await
}
/// Executes a sequence of SQL statements using the simple query protocol, returning the resulting rows.
///
/// Statements should be separated by semicolons. If an error occurs, execution of the sequence will stop at that

9
libs/proxy/tokio-postgres2/src/generic_client.rs
View File

@@ -1,7 +1,8 @@
#![allow(async_fn_in_trait)]
use crate::query::RowStream;
use crate::types::Type;
use crate::{Client, Error, Transaction};
use async_trait::async_trait;
use postgres_protocol2::Oid;
mod private {
@@ -11,7 +12,6 @@ mod private {
/// A trait allowing abstraction over connections and transactions.
///
/// This trait is "sealed", and cannot be implemented outside of this crate.
#[async_trait]
pub trait GenericClient: private::Sealed {
/// Like `Client::query_raw_txt`.
async fn query_raw_txt<S, I>(&self, statement: &str, params: I) -> Result<RowStream, Error>
@@ -26,7 +26,6 @@ pub trait GenericClient: private::Sealed {
impl private::Sealed for Client {}
#[async_trait]
impl GenericClient for Client {
async fn query_raw_txt<S, I>(&self, statement: &str, params: I) -> Result<RowStream, Error>
where
@@ -39,14 +38,12 @@ impl GenericClient for Client {
/// Query for type information
async fn get_type(&self, oid: Oid) -> Result<Type, Error> {
self.get_type(oid).await
crate::prepare::get_type(self.inner(), oid).await
}
}
impl private::Sealed for Transaction<'_> {}
#[async_trait]
#[allow(clippy::needless_lifetimes)]
impl GenericClient for Transaction<'_> {
async fn query_raw_txt<S, I>(&self, statement: &str, params: I) -> Result<RowStream, Error>
where

2
libs/proxy/tokio-postgres2/src/lib.rs
View File

@@ -14,7 +14,6 @@ pub use crate::row::{Row, SimpleQueryRow};
pub use crate::simple_query::SimpleQueryStream;
pub use crate::statement::{Column, Statement};
pub use crate::tls::NoTls;
pub use crate::to_statement::ToStatement;
pub use crate::transaction::Transaction;
pub use crate::transaction_builder::{IsolationLevel, TransactionBuilder};
use crate::types::ToSql;
@@ -65,7 +64,6 @@ pub mod row;
mod simple_query;
mod statement;
pub mod tls;
mod to_statement;
mod transaction;
mod transaction_builder;
pub mod types;

48
libs/proxy/tokio-postgres2/src/prepare.rs
View File

@@ -1,7 +1,6 @@
use crate::client::InnerClient;
use crate::codec::FrontendMessage;
use crate::connection::RequestMessages;
use crate::error::SqlState;
use crate::types::{Field, Kind, Oid, Type};
use crate::{query, slice_iter};
use crate::{Column, Error, Statement};
@@ -13,7 +12,6 @@ use postgres_protocol2::message::backend::Message;
use postgres_protocol2::message::frontend;
use std::future::Future;
use std::pin::Pin;
use std::sync::atomic::{AtomicUsize, Ordering};
use std::sync::Arc;
pub(crate) const TYPEINFO_QUERY: &str = "\
@@ -24,14 +22,6 @@ INNER JOIN pg_catalog.pg_namespace n ON t.typnamespace = n.oid
WHERE t.oid = $1
";
// Range types weren't added until Postgres 9.2, so pg_range may not exist
const TYPEINFO_FALLBACK_QUERY: &str = "\
SELECT t.typname, t.typtype, t.typelem, NULL::OID, t.typbasetype, n.nspname, t.typrelid
FROM pg_catalog.pg_type t
INNER JOIN pg_catalog.pg_namespace n ON t.typnamespace = n.oid
WHERE t.oid = $1
";
const TYPEINFO_ENUM_QUERY: &str = "\
SELECT enumlabel
FROM pg_catalog.pg_enum
@@ -39,14 +29,6 @@ WHERE enumtypid = $1
ORDER BY enumsortorder
";
// Postgres 9.0 didn't have enumsortorder
const TYPEINFO_ENUM_FALLBACK_QUERY: &str = "\
SELECT enumlabel
FROM pg_catalog.pg_enum
WHERE enumtypid = $1
ORDER BY oid
";
pub(crate) const TYPEINFO_COMPOSITE_QUERY: &str = "\
SELECT attname, atttypid
FROM pg_catalog.pg_attribute
@@ -56,15 +38,13 @@ AND attnum > 0
ORDER BY attnum
";
static NEXT_ID: AtomicUsize = AtomicUsize::new(0);
pub async fn prepare(
client: &Arc<InnerClient>,
name: &'static str,
query: &str,
types: &[Type],
) -> Result<Statement, Error> {
let name = format!("s{}", NEXT_ID.fetch_add(1, Ordering::SeqCst));
let buf = encode(client, &name, query, types)?;
let buf = encode(client, name, query, types)?;
let mut responses = client.send(RequestMessages::Single(FrontendMessage::Raw(buf)))?;
match responses.next().await? {
@@ -105,10 +85,11 @@ pub async fn prepare(
fn prepare_rec<'a>(
client: &'a Arc<InnerClient>,
name: &'static str,
query: &'a str,
types: &'a [Type],
) -> Pin<Box<dyn Future<Output = Result<Statement, Error>> + 'a + Send>> {
Box::pin(prepare(client, query, types))
Box::pin(prepare(client, name, query, types))
}
fn encode(client: &InnerClient, name: &str, query: &str, types: &[Type]) -> Result<Bytes, Error> {
@@ -192,13 +173,8 @@ async fn typeinfo_statement(client: &Arc<InnerClient>) -> Result<Statement, Erro
return Ok(stmt);
}
let stmt = match prepare_rec(client, TYPEINFO_QUERY, &[]).await {
Ok(stmt) => stmt,
Err(ref e) if e.code() == Some(&SqlState::UNDEFINED_TABLE) => {
prepare_rec(client, TYPEINFO_FALLBACK_QUERY, &[]).await?
}
Err(e) => return Err(e),
};
let typeinfo = "neon_proxy_typeinfo";
let stmt = prepare_rec(client, typeinfo, TYPEINFO_QUERY, &[]).await?;
client.set_typeinfo(&stmt);
Ok(stmt)
@@ -219,13 +195,8 @@ async fn typeinfo_enum_statement(client: &Arc<InnerClient>) -> Result<Statement,
return Ok(stmt);
}
let stmt = match prepare_rec(client, TYPEINFO_ENUM_QUERY, &[]).await {
Ok(stmt) => stmt,
Err(ref e) if e.code() == Some(&SqlState::UNDEFINED_COLUMN) => {
prepare_rec(client, TYPEINFO_ENUM_FALLBACK_QUERY, &[]).await?
}
Err(e) => return Err(e),
};
let typeinfo = "neon_proxy_typeinfo_enum";
let stmt = prepare_rec(client, typeinfo, TYPEINFO_ENUM_QUERY, &[]).await?;
client.set_typeinfo_enum(&stmt);
Ok(stmt)
@@ -255,7 +226,8 @@ async fn typeinfo_composite_statement(client: &Arc<InnerClient>) -> Result<State
return Ok(stmt);
}
let stmt = prepare_rec(client, TYPEINFO_COMPOSITE_QUERY, &[]).await?;
let typeinfo = "neon_proxy_typeinfo_composite";
let stmt = prepare_rec(client, typeinfo, TYPEINFO_COMPOSITE_QUERY, &[]).await?;
client.set_typeinfo_composite(&stmt);
Ok(stmt)

43
libs/proxy/tokio-postgres2/src/query.rs
View File

@@ -157,49 +157,6 @@ where
})
}
pub async fn execute<'a, I>(
client: &InnerClient,
statement: Statement,
params: I,
) -> Result<u64, Error>
where
I: IntoIterator<Item = &'a (dyn ToSql + Sync)>,
I::IntoIter: ExactSizeIterator,
{
let buf = if log_enabled!(Level::Debug) {
let params = params.into_iter().collect::<Vec<_>>();
debug!(
"executing statement {} with parameters: {:?}",
statement.name(),
BorrowToSqlParamsDebug(params.as_slice()),
);
encode(client, &statement, params)?
} else {
encode(client, &statement, params)?
};
let mut responses = start(client, buf).await?;
let mut rows = 0;
loop {
match responses.next().await? {
Message::DataRow(_) => {}
Message::CommandComplete(body) => {
rows = body
.tag()
.map_err(Error::parse)?
.rsplit(' ')
.next()
.unwrap()
.parse()
.unwrap_or(0);
}
Message::EmptyQueryResponse => rows = 0,
Message::ReadyForQuery(_) => return Ok(rows),
_ => return Err(Error::unexpected_message()),
}
}
}
async fn start(client: &InnerClient, buf: Bytes) -> Result<Responses, Error> {
let mut responses = client.send(RequestMessages::Single(FrontendMessage::Raw(buf)))?;

10
libs/proxy/tokio-postgres2/src/statement.rs
View File

@@ -13,7 +13,7 @@ use std::{
struct StatementInner {
client: Weak<InnerClient>,
name: String,
name: &'static str,
params: Vec<Type>,
columns: Vec<Column>,
}
@@ -22,7 +22,7 @@ impl Drop for StatementInner {
fn drop(&mut self) {
if let Some(client) = self.client.upgrade() {
let buf = client.with_buf(|buf| {
frontend::close(b'S', &self.name, buf).unwrap();
frontend::close(b'S', self.name, buf).unwrap();
frontend::sync(buf);
buf.split().freeze()
});
@@ -40,7 +40,7 @@ pub struct Statement(Arc<StatementInner>);
impl Statement {
pub(crate) fn new(
inner: &Arc<InnerClient>,
name: String,
name: &'static str,
params: Vec<Type>,
columns: Vec<Column>,
) -> Statement {
@@ -55,14 +55,14 @@ impl Statement {
pub(crate) fn new_anonymous(params: Vec<Type>, columns: Vec<Column>) -> Statement {
Statement(Arc::new(StatementInner {
client: Weak::new(),
name: String::new(),
name: "<anonymous>",
params,
columns,
}))
}
pub(crate) fn name(&self) -> &str {
&self.0.name
self.0.name
}
/// Returns the expected types of the statement's parameters.

57
libs/proxy/tokio-postgres2/src/to_statement.rs
View File

@@ -1,57 +0,0 @@
use crate::to_statement::private::{Sealed, ToStatementType};
use crate::Statement;
mod private {
use crate::{Client, Error, Statement};
pub trait Sealed {}
pub enum ToStatementType<'a> {
Statement(&'a Statement),
Query(&'a str),
}
impl ToStatementType<'_> {
pub async fn into_statement(self, client: &Client) -> Result<Statement, Error> {
match self {
ToStatementType::Statement(s) => Ok(s.clone()),
ToStatementType::Query(s) => client.prepare(s).await,
}
}
}
}
/// A trait abstracting over prepared and unprepared statements.
///
/// Many methods are generic over this bound, so that they support both a raw query string as well as a statement which
/// was prepared previously.
///
/// This trait is "sealed" and cannot be implemented by anything outside this crate.
pub trait ToStatement: Sealed {
#[doc(hidden)]
fn __convert(&self) -> ToStatementType<'_>;
}
impl ToStatement for Statement {
fn __convert(&self) -> ToStatementType<'_> {
ToStatementType::Statement(self)
}
}
impl Sealed for Statement {}
impl ToStatement for str {
fn __convert(&self) -> ToStatementType<'_> {
ToStatementType::Query(self)
}
}
impl Sealed for str {}
impl ToStatement for String {
fn __convert(&self) -> ToStatementType<'_> {
ToStatementType::Query(self)
}
}
impl Sealed for String {}

42
libs/safekeeper_api/src/membership.rs
View File

@@ -9,13 +9,43 @@ use anyhow::bail;
use serde::{Deserialize, Serialize};
use utils::id::NodeId;
/// Number uniquely identifying safekeeper configuration.
/// Note: it is a part of sk control file.
pub type Generation = u32;
/// 1 is the first valid generation, 0 is used as
/// a placeholder before we fully migrate to generations.
pub const INVALID_GENERATION: Generation = 0;
pub const INITIAL_GENERATION: Generation = 1;
pub const INVALID_GENERATION: SafekeeperGeneration = SafekeeperGeneration::new(0);
pub const INITIAL_GENERATION: SafekeeperGeneration = SafekeeperGeneration::new(1);
/// Number uniquely identifying safekeeper configuration.
/// Note: it is a part of sk control file.
///
/// Like tenant generations, but for safekeepers.
#[derive(Debug, Copy, Clone, PartialEq, Eq, PartialOrd, Ord, Serialize, Deserialize)]
pub struct SafekeeperGeneration(u32);
impl SafekeeperGeneration {
pub const fn new(v: u32) -> Self {
Self(v)
}
#[track_caller]
pub fn previous(&self) -> Option<Self> {
Some(Self(self.0.checked_sub(1)?))
}
#[track_caller]
pub fn next(&self) -> Self {
Self(self.0 + 1)
}
pub fn into_inner(self) -> u32 {
self.0
}
}
impl Display for SafekeeperGeneration {
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
write!(f, "{}", self.0)
}
}
/// Membership is defined by ids so e.g. walproposer uses them to figure out
/// quorums, but we also carry host and port to give wp idea where to connect.
@@ -89,7 +119,7 @@ impl Display for MemberSet {
#[derive(Debug, Clone, Serialize, Deserialize, PartialEq)]
pub struct Configuration {
/// Unique id.
pub generation: Generation,
pub generation: SafekeeperGeneration,
/// Current members of the configuration.
pub members: MemberSet,
/// Some means it is a joint conf.

15
libs/safekeeper_api/src/models.rs
View File

@@ -282,3 +282,18 @@ pub struct TimelineTermBumpResponse {
pub struct SafekeeperUtilization {
pub timeline_count: u64,
}
/// pull_timeline request body.
#[derive(Debug, Deserialize, Serialize)]
pub struct PullTimelineRequest {
pub tenant_id: TenantId,
pub timeline_id: TimelineId,
pub http_hosts: Vec<String>,
}
#[derive(Debug, Serialize, Deserialize)]
pub struct PullTimelineResponse {
// Donor safekeeper host
pub safekeeper_host: String,
// TODO: add more fields?
}

16
libs/utils/Cargo.toml
View File

@@ -21,23 +21,17 @@ bytes.workspace = true
camino.workspace = true
chrono.workspace = true
diatomic-waker.workspace = true
flate2.workspace = true
git-version.workspace = true
hex = { workspace = true, features = ["serde"] }
humantime.workspace = true
hyper0 = { workspace = true, features = ["full"] }
inferno.workspace = true
itertools.workspace = true
fail.workspace = true
futures = { workspace = true }
jemalloc_pprof.workspace = true
jsonwebtoken.workspace = true
nix.workspace = true
nix = {workspace = true, features = [ "ioctl" ] }
once_cell.workspace = true
pin-project-lite.workspace = true
pprof.workspace = true
regex.workspace = true
routerify.workspace = true
serde.workspace = true
serde_with.workspace = true
serde_json.workspace = true
@@ -54,8 +48,6 @@ rand.workspace = true
scopeguard.workspace = true
strum.workspace = true
strum_macros.workspace = true
url.workspace = true
uuid.workspace = true
walkdir.workspace = true
pq_proto.workspace = true
@@ -64,12 +56,6 @@ metrics.workspace = true
const_format.workspace = true
# to use tokio channels as streams, this is faster to compile than async_stream
# why is it only here? no other crate should use it, streams are rarely needed.
tokio-stream = { version = "0.1.14" }
serde_path_to_error.workspace = true
[dev-dependencies]
byteorder.workspace = true
bytes.workspace = true

11
libs/utils/src/auth.rs
View File

@@ -10,7 +10,7 @@ use jsonwebtoken::{
};
use serde::{Deserialize, Serialize};
use crate::{http::error::ApiError, id::TenantId};
use crate::id::TenantId;
/// Algorithm to use. We require EdDSA.
const STORAGE_TOKEN_ALGORITHM: Algorithm = Algorithm::EdDSA;
@@ -90,15 +90,6 @@ impl Display for AuthError {
}
}
impl From<AuthError> for ApiError {
fn from(_value: AuthError) -> Self {
// Don't pass on the value of the AuthError as a precautionary measure.
// Being intentionally vague in public error communication hurts debugability
// but it is more secure.
ApiError::Forbidden("JWT authentication error".to_string())
}
}
pub struct JwtAuth {
decoding_keys: Vec<DecodingKey>,
validation: Validation,

6
libs/utils/src/backoff.rs
View File

@@ -1,4 +1,5 @@
use std::fmt::{Debug, Display};
use std::time::Duration;
use futures::Future;
use tokio_util::sync::CancellationToken;
@@ -29,6 +30,11 @@ pub async fn exponential_backoff(
}
}
pub fn exponential_backoff_duration(n: u32, base_increment: f64, max_seconds: f64) -> Duration {
let seconds = exponential_backoff_duration_seconds(n, base_increment, max_seconds);
Duration::from_secs_f64(seconds)
}
pub fn exponential_backoff_duration_seconds(n: u32, base_increment: f64, max_seconds: f64) -> f64 {
if n == 0 {
0.0

43
libs/utils/src/bin_ser.rs
View File

@@ -286,6 +286,11 @@ mod tests {
const SHORT2_ENC_LE: &[u8] = &[8, 0, 0, 3, 7];
const SHORT2_ENC_LE_TRAILING: &[u8] = &[8, 0, 0, 3, 7, 0xff, 0xff, 0xff];
#[derive(Debug, PartialEq, Eq, Serialize, Deserialize)]
struct NewTypeStruct(u32);
const NT1: NewTypeStruct = NewTypeStruct(414243);
const NT1_INNER: u32 = 414243;
#[derive(Debug, PartialEq, Eq, Serialize, Deserialize)]
pub struct LongMsg {
pub tag: u8,
@@ -408,4 +413,42 @@ mod tests {
let msg2 = LongMsg::des(&encoded).unwrap();
assert_eq!(msg, msg2);
}
#[test]
/// Ensure that newtype wrappers around u32 don't change the serialization format
fn be_nt() {
use super::BeSer;
assert_eq!(NT1.serialized_size().unwrap(), 4);
let msg = NT1;
let encoded = msg.ser().unwrap();
let expected = hex_literal::hex!("0006 5223");
assert_eq!(encoded, expected);
assert_eq!(encoded, NT1_INNER.ser().unwrap());
let msg2 = NewTypeStruct::des(&encoded).unwrap();
assert_eq!(msg, msg2);
}
#[test]
/// Ensure that newtype wrappers around u32 don't change the serialization format
fn le_nt() {
use super::LeSer;
assert_eq!(NT1.serialized_size().unwrap(), 4);
let msg = NT1;
let encoded = msg.ser().unwrap();
let expected = hex_literal::hex!("2352 0600");
assert_eq!(encoded, expected);
assert_eq!(encoded, NT1_INNER.ser().unwrap());
let msg2 = NewTypeStruct::des(&encoded).unwrap();
assert_eq!(msg, msg2);
}
}

49
libs/utils/src/failpoint_support.rs
View File

@@ -1,13 +1,6 @@
//! Failpoint support code shared between pageserver and safekeepers.
use crate::http::{
error::ApiError,
json::{json_request, json_response},
};
use hyper::{Body, Request, Response, StatusCode};
use serde::{Deserialize, Serialize};
use tokio_util::sync::CancellationToken;
use tracing::*;
/// Declare a failpoint that can use to `pause` failpoint action.
/// We don't want to block the executor thread, hence, spawn_blocking + await.
@@ -184,45 +177,3 @@ fn exit_failpoint() {
tracing::info!("Exit requested by failpoint");
std::process::exit(1);
}
pub type ConfigureFailpointsRequest = Vec<FailpointConfig>;
/// Information for configuring a single fail point
#[derive(Debug, Serialize, Deserialize)]
pub struct FailpointConfig {
/// Name of the fail point
pub name: String,
/// List of actions to take, using the format described in `fail::cfg`
///
/// We also support `actions = "exit"` to cause the fail point to immediately exit.
pub actions: String,
}
/// Configure failpoints through http.
pub async fn failpoints_handler(
mut request: Request<Body>,
_cancel: CancellationToken,
) -> Result<Response<Body>, ApiError> {
if !fail::has_failpoints() {
return Err(ApiError::BadRequest(anyhow::anyhow!(
"Cannot manage failpoints because neon was compiled without failpoints support"
)));
}
let failpoints: ConfigureFailpointsRequest = json_request(&mut request).await?;
for fp in failpoints {
info!("cfg failpoint: {} {}", fp.name, fp.actions);
// We recognize one extra "action" that's not natively recognized
// by the failpoints crate: exit, to immediately kill the process
let cfg_result = apply_failpoint(&fp.name, &fp.actions);
if let Err(err_msg) = cfg_result {
return Err(ApiError::BadRequest(anyhow::anyhow!(
"Failed to configure failpoints: {err_msg}"
)));
}
}
json_response(StatusCode::OK, ())
}

10
libs/utils/src/lib.rs
View File

@@ -2,8 +2,6 @@
//! between other crates in this repository.
#![deny(clippy::undocumented_unsafe_blocks)]
extern crate hyper0 as hyper;
pub mod backoff;
/// `Lsn` type implements common tasks on Log Sequence Numbers
@@ -33,9 +31,6 @@ pub mod shard;
mod hex;
pub use hex::Hex;
// http endpoint utils
pub mod http;
// definition of the Generation type for pageserver attachment APIs
pub mod generation;
@@ -96,10 +91,11 @@ pub mod circuit_breaker;
pub mod try_rcu;
pub mod pprof;
pub mod guard_arc_swap;
#[cfg(target_os = "linux")]
pub mod linux_socket_ioctl;
// Re-export used in macro. Avoids adding git-version as dep in target crates.
#[doc(hidden)]
pub use git_version;

35
libs/utils/src/linux_socket_ioctl.rs Normal file
View File

@@ -0,0 +1,35 @@
//! Linux-specific socket ioctls.
//!
//! <https://elixir.bootlin.com/linux/v6.1.128/source/include/uapi/linux/sockios.h#L25-L27>
use std::{
io,
mem::MaybeUninit,
os::{fd::RawFd, raw::c_int},
};
use nix::libc::{FIONREAD, TIOCOUTQ};
unsafe fn do_ioctl(socket_fd: RawFd, cmd: nix::libc::Ioctl) -> io::Result<c_int> {
let mut inq: MaybeUninit<c_int> = MaybeUninit::uninit();
let err = nix::libc::ioctl(socket_fd, cmd, inq.as_mut_ptr());
if err == 0 {
Ok(inq.assume_init())
} else {
Err(io::Error::last_os_error())
}
}
/// # Safety
///
/// Caller must ensure that `socket_fd` is a valid TCP socket file descriptor.
pub unsafe fn inq(socket_fd: RawFd) -> io::Result<c_int> {
do_ioctl(socket_fd, FIONREAD)
}
/// # Safety
///
/// Caller must ensure that `socket_fd` is a valid TCP socket file descriptor.
pub unsafe fn outq(socket_fd: RawFd) -> io::Result<c_int> {
do_ioctl(socket_fd, TIOCOUTQ)
}

4
libs/utils/src/shard.rs
View File

@@ -117,6 +117,10 @@ impl TenantShardId {
)
}
pub fn range(&self) -> RangeInclusive<Self> {
RangeInclusive::new(*self, *self)
}
pub fn shard_slug(&self) -> impl std::fmt::Display + '_ {
ShardSlug(self)
}

1
pageserver/Cargo.toml
View File

@@ -79,6 +79,7 @@ pq_proto.workspace = true
remote_storage.workspace = true
storage_broker.workspace = true
tenant_size_model.workspace = true
http-utils.workspace = true
utils.workspace = true
workspace_hack.workspace = true
reqwest.workspace = true

1
pageserver/client/Cargo.toml
View File

@@ -11,6 +11,7 @@ testing = [ "pageserver_api/testing" ]
pageserver_api.workspace = true
thiserror.workspace = true
reqwest = { workspace = true, features = [ "stream" ] }
http-utils.workspace = true
utils.workspace = true
serde.workspace = true
workspace_hack = { version = "0.1", path = "../../workspace_hack" }

27
pageserver/client/src/mgmt_api.rs
View File

@@ -1,11 +1,12 @@
use std::{collections::HashMap, error::Error as _};
use bytes::Bytes;
use detach_ancestor::AncestorDetached;
use pageserver_api::{models::*, shard::TenantShardId};
use reqwest::{IntoUrl, Method, StatusCode};
use detach_ancestor::AncestorDetached;
use http_utils::error::HttpErrorBody;
use pageserver_api::{models::*, shard::TenantShardId};
use utils::{
http::error::HttpErrorBody,
id::{TenantId, TimelineId},
lsn::Lsn,
};
@@ -476,6 +477,26 @@ impl Client {
self.request(Method::POST, &uri, ()).await.map(|_| ())
}
pub async fn timeline_download_heatmap_layers(
&self,
tenant_shard_id: TenantShardId,
timeline_id: TimelineId,
concurrency: Option<usize>,
) -> Result<()> {
let mut path = reqwest::Url::parse(&format!(
"{}/v1/tenant/{}/timeline/{}/download_heatmap_layers",
self.mgmt_api_endpoint, tenant_shard_id, timeline_id
))
.expect("Cannot build URL");
if let Some(concurrency) = concurrency {
path.query_pairs_mut()
.append_pair("concurrency", &format!("{}", concurrency));
}
self.request(Method::POST, path, ()).await.map(|_| ())
}
pub async fn tenant_reset(&self, tenant_shard_id: TenantShardId) -> Result<()> {
let uri = format!(
"{}/v1/tenant/{}/reset",

69
pageserver/src/basebackup.rs
View File

@@ -13,7 +13,7 @@
use anyhow::{anyhow, Context};
use bytes::{BufMut, Bytes, BytesMut};
use fail::fail_point;
use pageserver_api::key::Key;
use pageserver_api::key::{rel_block_to_key, Key};
use postgres_ffi::pg_constants;
use std::fmt::Write as FmtWrite;
use std::time::{Instant, SystemTime};
@@ -42,8 +42,8 @@ use utils::lsn::Lsn;
pub enum BasebackupError {
#[error("basebackup pageserver error {0:#}")]
Server(#[from] anyhow::Error),
#[error("basebackup client error {0:#}")]
Client(#[source] io::Error),
#[error("basebackup client error {0:#} when {1}")]
Client(#[source] io::Error, &'static str),
}
/// Create basebackup with non-rel data in it.
@@ -234,7 +234,7 @@ where
self.ar
.append(&header, self.buf.as_slice())
.await
.map_err(BasebackupError::Client)?;
.map_err(|e| BasebackupError::Client(e, "flush"))?;
self.total_blocks += nblocks;
debug!("Added to basebackup slru {} relsize {}", segname, nblocks);
@@ -273,9 +273,9 @@ where
for dir in subdirs.iter() {
let header = new_tar_header_dir(dir)?;
self.ar
.append(&header, &mut io::empty())
.append(&header, io::empty())
.await
.context("could not add directory to basebackup tarball")?;
.map_err(|e| BasebackupError::Client(e, "send_tarball"))?;
}
// Send config files.
@@ -286,13 +286,13 @@ where
self.ar
.append(&header, data)
.await
.context("could not add config file to basebackup tarball")?;
.map_err(|e| BasebackupError::Client(e, "send_tarball,pg_hba.conf"))?;
} else {
let header = new_tar_header(filepath, 0)?;
self.ar
.append(&header, &mut io::empty())
.append(&header, io::empty())
.await
.context("could not add config file to basebackup tarball")?;
.map_err(|e| BasebackupError::Client(e, "send_tarball,add_config_file"))?;
}
}
if !lazy_slru_download {
@@ -406,7 +406,7 @@ where
self.ar
.append(&header, &*content)
.await
.context("could not add aux file to basebackup tarball")?;
.map_err(|e| BasebackupError::Client(e, "send_tarball,add_aux_file"))?;
}
if min_restart_lsn != Lsn::MAX {
@@ -419,7 +419,7 @@ where
self.ar
.append(&header, &data[..])
.await
.context("could not add restart.lsn file to basebackup tarball")?;
.map_err(|e| BasebackupError::Client(e, "send_tarball,restart.lsn"))?;
}
for xid in self
.timeline
@@ -451,9 +451,9 @@ where
let crc32 = crc32c::crc32c(&content);
content.extend_from_slice(&crc32.to_le_bytes());
let header = new_tar_header("pg_logical/replorigin_checkpoint", content.len() as u64)?;
self.ar.append(&header, &*content).await.context(
"could not add pg_logical/replorigin_checkpoint file to basebackup tarball",
)?;
self.ar.append(&header, &*content).await.map_err(|e| {
BasebackupError::Client(e, "send_tarball,pg_logical/replorigin_checkpoint")
})?;
}
fail_point!("basebackup-before-control-file", |_| {
@@ -464,7 +464,10 @@ where
// Generate pg_control and bootstrap WAL segment.
self.add_pgcontrol_file().await?;
self.ar.finish().await.map_err(BasebackupError::Client)?;
self.ar
.finish()
.await
.map_err(|e| BasebackupError::Client(e, "send_tarball,finish"))?;
debug!("all tarred up!");
Ok(())
}
@@ -482,9 +485,9 @@ where
let file_name = dst.to_segfile_name(0);
let header = new_tar_header(&file_name, 0)?;
self.ar
.append(&header, &mut io::empty())
.append(&header, io::empty())
.await
.map_err(BasebackupError::Client)?;
.map_err(|e| BasebackupError::Client(e, "add_rel,empty"))?;
return Ok(());
}
@@ -498,13 +501,9 @@ where
for blknum in startblk..endblk {
let img = self
.timeline
.get_rel_page_at_lsn(
src,
blknum,
Version::Lsn(self.lsn),
self.ctx,
self.io_concurrency.clone(),
)
// TODO: investigate using get_vectored for the entire startblk..endblk range.
// But this code path is not on the critical path for most basebackups (?).
.get(rel_block_to_key(src, blknum), self.lsn, self.ctx)
.await
.map_err(|e| BasebackupError::Server(e.into()))?;
segment_data.extend_from_slice(&img[..]);
@@ -515,7 +514,7 @@ where
self.ar
.append(&header, segment_data.as_slice())
.await
.map_err(BasebackupError::Client)?;
.map_err(|e| BasebackupError::Client(e, "add_rel,segment"))?;
seg += 1;
startblk = endblk;
@@ -566,7 +565,7 @@ where
self.ar
.append(&header, pg_version_str.as_bytes())
.await
.map_err(BasebackupError::Client)?;
.map_err(|e| BasebackupError::Client(e, "add_dbdir,PG_VERSION"))?;
info!("timeline.pg_version {}", self.timeline.pg_version);
@@ -576,7 +575,7 @@ where
self.ar
.append(&header, &img[..])
.await
.map_err(BasebackupError::Client)?;
.map_err(|e| BasebackupError::Client(e, "add_dbdir,global/pg_filenode.map"))?;
} else {
warn!("global/pg_filenode.map is missing");
}
@@ -612,9 +611,9 @@ where
let path = format!("base/{}", dbnode);
let header = new_tar_header_dir(&path)?;
self.ar
.append(&header, &mut io::empty())
.append(&header, io::empty())
.await
.map_err(BasebackupError::Client)?;
.map_err(|e| BasebackupError::Client(e, "add_dbdir,base"))?;
if let Some(img) = relmap_img {
let dst_path = format!("base/{}/PG_VERSION", dbnode);
@@ -627,14 +626,14 @@ where
self.ar
.append(&header, pg_version_str.as_bytes())
.await
.map_err(BasebackupError::Client)?;
.map_err(|e| BasebackupError::Client(e, "add_dbdir,base/PG_VERSION"))?;
let relmap_path = format!("base/{}/pg_filenode.map", dbnode);
let header = new_tar_header(&relmap_path, img.len() as u64)?;
self.ar
.append(&header, &img[..])
.await
.map_err(BasebackupError::Client)?;
.map_err(|e| BasebackupError::Client(e, "add_dbdir,base/pg_filenode.map"))?;
}
};
Ok(())
@@ -663,7 +662,7 @@ where
self.ar
.append(&header, &buf[..])
.await
.map_err(BasebackupError::Client)?;
.map_err(|e| BasebackupError::Client(e, "add_twophase_file"))?;
Ok(())
}
@@ -693,7 +692,7 @@ where
zenith_signal.as_bytes(),
)
.await
.map_err(BasebackupError::Client)?;
.map_err(|e| BasebackupError::Client(e, "add_pgcontrol_file,zenith.signal"))?;
let checkpoint_bytes = self
.timeline
@@ -718,7 +717,7 @@ where
self.ar
.append(&header, &pg_control_bytes[..])
.await
.map_err(BasebackupError::Client)?;
.map_err(|e| BasebackupError::Client(e, "add_pgcontrol_file,pg_control"))?;
//send wal segment
let segno = self.lsn.segment_number(WAL_SEGMENT_SIZE);
@@ -742,7 +741,7 @@ where
self.ar
.append(&header, &wal_seg[..])
.await
.map_err(BasebackupError::Client)?;
.map_err(|e| BasebackupError::Client(e, "add_pgcontrol_file,wal_segment"))?;
Ok(())
}
}

2
pageserver/src/bin/pageserver.rs
View File

@@ -592,7 +592,7 @@ fn start_pageserver(
let router = http::make_router(router_state, launch_ts, http_auth.clone())?
.build()
.map_err(|err| anyhow!(err))?;
let service = utils::http::RouterService::new(router).unwrap();
let service = http_utils::RouterService::new(router).unwrap();
let server = hyper0::Server::from_tcp(http_listener)?
.serve(service)
.with_graceful_shutdown({

6
pageserver/src/config.rs
View File

@@ -193,6 +193,10 @@ pub struct PageServerConf {
pub page_service_pipelining: pageserver_api::config::PageServicePipeliningConfig,
pub get_vectored_concurrent_io: pageserver_api::config::GetVectoredConcurrentIo,
/// Enable read path debugging. If enabled, read key errors will print a backtrace of the layer
/// files read.
pub enable_read_path_debugging: bool,
}
/// Token for authentication to safekeepers
@@ -355,6 +359,7 @@ impl PageServerConf {
wal_receiver_protocol,
page_service_pipelining,
get_vectored_concurrent_io,
enable_read_path_debugging,
} = config_toml;
let mut conf = PageServerConf {
@@ -440,6 +445,7 @@ impl PageServerConf {
.unwrap_or_default(),
virtual_file_io_mode: virtual_file_io_mode.unwrap_or(virtual_file::IoMode::preferred()),
no_sync: no_sync.unwrap_or(false),
enable_read_path_debugging: enable_read_path_debugging.unwrap_or(false),
};
// ------------------------------------------------------------

109
pageserver/src/deletion_queue.rs
View File

@@ -8,7 +8,6 @@ use std::time::Duration;
use crate::controller_upcall_client::ControlPlaneGenerationsApi;
use crate::metrics;
use crate::tenant::remote_timeline_client::remote_layer_path;
use crate::tenant::remote_timeline_client::remote_timeline_path;
use crate::tenant::remote_timeline_client::LayerFileMetadata;
use crate::virtual_file::MaybeFatalIo;
@@ -463,45 +462,18 @@ impl DeletionQueueClient {
///
/// The `current_generation` is the generation of this pageserver's current attachment. The
/// generations in `layers` are the generations in which those layers were written.
pub(crate) async fn push_layers(
pub(crate) fn push_layers(
&self,
tenant_shard_id: TenantShardId,
timeline_id: TimelineId,
current_generation: Generation,
layers: Vec<(LayerName, LayerFileMetadata)>,
) -> Result<(), DeletionQueueError> {
if current_generation.is_none() {
debug!("Enqueuing deletions in legacy mode, skipping queue");
// None generations are not valid for attached tenants: they must always be attached in
// a known generation. None generations are still permitted for layers in the index because
// they may be historical.
assert!(!current_generation.is_none());
let mut layer_paths = Vec::new();
for (layer, meta) in layers {
layer_paths.push(remote_layer_path(
&tenant_shard_id.tenant_id,
&timeline_id,
meta.shard,
&layer,
meta.generation,
));
}
self.push_immediate(layer_paths).await?;
return self.flush_immediate().await;
}
self.push_layers_sync(tenant_shard_id, timeline_id, current_generation, layers)
}
/// When a Tenant has a generation, push_layers is always synchronous because
/// the ListValidator channel is an unbounded channel.
///
/// This can be merged into push_layers when we remove the Generation-less mode
/// support (`<https://github.com/neondatabase/neon/issues/5395>`)
pub(crate) fn push_layers_sync(
&self,
tenant_shard_id: TenantShardId,
timeline_id: TimelineId,
current_generation: Generation,
layers: Vec<(LayerName, LayerFileMetadata)>,
) -> Result<(), DeletionQueueError> {
metrics::DELETION_QUEUE
.keys_submitted
.inc_by(layers.len() as u64);
@@ -957,14 +929,12 @@ mod test {
// File should still be there after we push it to the queue (we haven't pushed enough to flush anything)
info!("Pushing");
client
.push_layers(
tenant_shard_id,
TIMELINE_ID,
now_generation,
[(layer_file_name_1.clone(), layer_metadata)].to_vec(),
)
.await?;
client.push_layers(
tenant_shard_id,
TIMELINE_ID,
now_generation,
[(layer_file_name_1.clone(), layer_metadata)].to_vec(),
)?;
assert_remote_files(&[&remote_layer_file_name_1], &remote_timeline_path);
assert_local_files(&[], &deletion_prefix);
@@ -1017,14 +987,12 @@ mod test {
assert_remote_files(&[&remote_layer_name], &remote_timeline_path);
tracing::debug!("Pushing...");
client
.push_layers(
tenant_shard_id,
TIMELINE_ID,
stale_generation,
[(EXAMPLE_LAYER_NAME.clone(), layer_metadata.clone())].to_vec(),
)
.await?;
client.push_layers(
tenant_shard_id,
TIMELINE_ID,
stale_generation,
[(EXAMPLE_LAYER_NAME.clone(), layer_metadata.clone())].to_vec(),
)?;
// We enqueued the operation in a stale generation: it should have failed validation
tracing::debug!("Flushing...");
@@ -1032,14 +1000,12 @@ mod test {
assert_remote_files(&[&remote_layer_name], &remote_timeline_path);
tracing::debug!("Pushing...");
client
.push_layers(
tenant_shard_id,
TIMELINE_ID,
latest_generation,
[(EXAMPLE_LAYER_NAME.clone(), layer_metadata.clone())].to_vec(),
)
.await?;
client.push_layers(
tenant_shard_id,
TIMELINE_ID,
latest_generation,
[(EXAMPLE_LAYER_NAME.clone(), layer_metadata.clone())].to_vec(),
)?;
// We enqueued the operation in a fresh generation: it should have passed validation
tracing::debug!("Flushing...");
@@ -1074,28 +1040,24 @@ mod test {
// generation gets that treatment)
let remote_layer_file_name_historical =
ctx.write_remote_layer(EXAMPLE_LAYER_NAME, layer_generation)?;
client
.push_layers(
tenant_shard_id,
TIMELINE_ID,
now_generation.previous(),
[(EXAMPLE_LAYER_NAME.clone(), layer_metadata.clone())].to_vec(),
)
.await?;
client.push_layers(
tenant_shard_id,
TIMELINE_ID,
now_generation.previous(),
[(EXAMPLE_LAYER_NAME.clone(), layer_metadata.clone())].to_vec(),
)?;
// Inject a deletion in the generation before generation_now: after restart,
// this deletion should get executed, because we execute deletions in the
// immediately previous generation on the same node.
let remote_layer_file_name_previous =
ctx.write_remote_layer(EXAMPLE_LAYER_NAME_ALT, layer_generation)?;
client
.push_layers(
tenant_shard_id,
TIMELINE_ID,
now_generation,
[(EXAMPLE_LAYER_NAME_ALT.clone(), layer_metadata.clone())].to_vec(),
)
.await?;
client.push_layers(
tenant_shard_id,
TIMELINE_ID,
now_generation,
[(EXAMPLE_LAYER_NAME_ALT.clone(), layer_metadata.clone())].to_vec(),
)?;
client.flush().await?;
assert_remote_files(
@@ -1139,6 +1101,7 @@ pub(crate) mod mock {
use tracing::info;
use super::*;
use crate::tenant::remote_timeline_client::remote_layer_path;
use std::sync::atomic::{AtomicUsize, Ordering};
pub struct ConsumerState {

11
pageserver/src/disk_usage_eviction_task.rs
View File

@@ -61,6 +61,7 @@ use crate::{
remote_timeline_client::LayerFileMetadata,
secondary::SecondaryTenant,
storage_layer::{AsLayerDesc, EvictionError, Layer, LayerName, LayerVisibilityHint},
tasks::sleep_random,
},
CancellableTask, DiskUsageEvictionTask,
};
@@ -210,14 +211,8 @@ async fn disk_usage_eviction_task(
info!("disk usage based eviction task finishing");
};
use crate::tenant::tasks::random_init_delay;
{
if random_init_delay(task_config.period, &cancel)
.await
.is_err()
{
return;
}
if sleep_random(task_config.period, &cancel).await.is_err() {
return;
}
let mut iteration_no = 0;

40
pageserver/src/http/openapi_spec.yml
View File

@@ -824,6 +824,38 @@ paths:
schema:
$ref: "#/components/schemas/TenantConfigResponse"
/v1/tenant/{tenant_shard_id}/timeline/{timeline_id}/download_heatmap_layers:
parameters:
- name: tenant_shard_id
in: path
required: true
schema:
type: string
- name: timeline_id
in: path
required: true
schema:
type: string
- name: concurrency
description: Maximum number of concurrent downloads (capped at remote storage concurrency)
in: query
required: false
schema:
type: integer
post:
description: |
Download all layers in the specified timeline's heatmap. The `tenant_shard_id` parameter
may be used to target all shards of a tenant when the unsharded form is used, or a specific
tenant shard with the sharded form.
responses:
"200":
description: Success
delete:
description: Stop any on-going background downloads of heatmap layers for the specified timeline.
responses:
"200":
description: Success
/v1/utilization:
get:
description: |
@@ -882,6 +914,8 @@ components:
properties:
reason:
type: string
gc_blocking:
type: string
TenantCreateRequest:
allOf:
@@ -1080,9 +1114,15 @@ components:
type: integer
state:
type: string
min_readable_lsn:
type: string
format: hex
latest_gc_cutoff_lsn:
type: string
format: hex
applied_gc_cutoff_lsn:
type: string
format: hex
SyntheticSizeResponse:
type: object

103
pageserver/src/http/routes.rs
View File

@@ -13,6 +13,12 @@ use enumset::EnumSet;
use futures::future::join_all;
use futures::StreamExt;
use futures::TryFutureExt;
use http_utils::endpoint::{
profile_cpu_handler, profile_heap_handler, prometheus_metrics_handler, request_span,
};
use http_utils::failpoints::failpoints_handler;
use http_utils::request::must_parse_query_param;
use http_utils::request::{get_request_param, must_get_query_param, parse_query_param};
use humantime::format_rfc3339;
use hyper::header;
use hyper::StatusCode;
@@ -60,13 +66,6 @@ use tokio::time::Instant;
use tokio_util::io::StreamReader;
use tokio_util::sync::CancellationToken;
use tracing::*;
use utils::auth::JwtAuth;
use utils::failpoint_support::failpoints_handler;
use utils::http::endpoint::{
profile_cpu_handler, profile_heap_handler, prometheus_metrics_handler, request_span,
};
use utils::http::request::must_parse_query_param;
use utils::http::request::{get_request_param, must_get_query_param, parse_query_param};
use crate::config::PageServerConf;
use crate::context::{DownloadBehavior, RequestContext};
@@ -104,6 +103,13 @@ use crate::tenant::OffloadedTimeline;
use crate::tenant::{LogicalSizeCalculationCause, PageReconstructError};
use crate::DEFAULT_PG_VERSION;
use crate::{disk_usage_eviction_task, tenant};
use http_utils::{
endpoint::{self, attach_openapi_ui, auth_middleware, check_permission_with},
error::{ApiError, HttpErrorBody},
json::{json_request, json_request_maybe, json_response},
request::parse_request_param,
RequestExt, RouterBuilder,
};
use pageserver_api::models::{
StatusResponse, TenantConfigRequest, TenantInfo, TimelineCreateRequest, TimelineGcRequest,
TimelineInfo,
@@ -111,13 +117,6 @@ use pageserver_api::models::{
use utils::{
auth::SwappableJwtAuth,
generation::Generation,
http::{
endpoint::{self, attach_openapi_ui, auth_middleware, check_permission_with},
error::{ApiError, HttpErrorBody},
json::{json_request, json_request_maybe, json_response},
request::parse_request_param,
RequestExt, RouterBuilder,
},
id::{TenantId, TimelineId},
lsn::Lsn,
};
@@ -483,6 +482,11 @@ async fn build_timeline_info_common(
let (pitr_history_size, within_ancestor_pitr) = timeline.get_pitr_history_stats();
let min_readable_lsn = std::cmp::max(
timeline.get_gc_cutoff_lsn(),
*timeline.get_applied_gc_cutoff_lsn(),
);
let info = TimelineInfo {
tenant_id: timeline.tenant_shard_id,
timeline_id: timeline.timeline_id,
@@ -494,7 +498,12 @@ async fn build_timeline_info_common(
initdb_lsn,
last_record_lsn,
prev_record_lsn: Some(timeline.get_prev_record_lsn()),
latest_gc_cutoff_lsn: *timeline.get_latest_gc_cutoff_lsn(),
// Externally, expose the lowest LSN that can be used to create a branch as the "GC cutoff", although internally
// we distinguish between the "planned" GC cutoff (PITR point) and the "latest" GC cutoff (where we
// actually trimmed data to), which can pass each other when PITR is changed.
latest_gc_cutoff_lsn: min_readable_lsn,
min_readable_lsn,
applied_gc_cutoff_lsn: *timeline.get_applied_gc_cutoff_lsn(),
current_logical_size: current_logical_size.size_dont_care_about_accuracy(),
current_logical_size_is_accurate: match current_logical_size.accuracy() {
tenant::timeline::logical_size::Accuracy::Approximate => false,
@@ -561,7 +570,7 @@ async fn reload_auth_validation_keys_handler(
let key_path = config.auth_validation_public_key_path.as_ref().unwrap();
info!("Reloading public key(s) for verifying JWT tokens from {key_path:?}");
match JwtAuth::from_key_path(key_path) {
match utils::auth::JwtAuth::from_key_path(key_path) {
Ok(new_auth) => {
shared_auth.swap(new_auth);
json_response(StatusCode::OK, ())
@@ -1454,6 +1463,59 @@ async fn timeline_layer_scan_disposable_keys(
)
}
async fn timeline_download_heatmap_layers_handler(
request: Request<Body>,
_cancel: CancellationToken,
) -> Result<Response<Body>, ApiError> {
// Only used in the case where remote storage is not configured.
const DEFAULT_MAX_CONCURRENCY: usize = 100;
// A conservative default.
const DEFAULT_CONCURRENCY: usize = 16;
let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
let timeline_id: TimelineId = parse_request_param(&request, "timeline_id")?;
let desired_concurrency =
parse_query_param(&request, "concurrency")?.unwrap_or(DEFAULT_CONCURRENCY);
check_permission(&request, Some(tenant_shard_id.tenant_id))?;
let state = get_state(&request);
let timeline =
active_timeline_of_active_tenant(&state.tenant_manager, tenant_shard_id, timeline_id)
.await?;
let max_concurrency = get_config(&request)
.remote_storage_config
.as_ref()
.map(|c| c.concurrency_limit())
.unwrap_or(DEFAULT_MAX_CONCURRENCY);
let concurrency = std::cmp::min(max_concurrency, desired_concurrency);
timeline.start_heatmap_layers_download(concurrency).await?;
json_response(StatusCode::ACCEPTED, ())
}
async fn timeline_shutdown_download_heatmap_layers_handler(
request: Request<Body>,
_cancel: CancellationToken,
) -> Result<Response<Body>, ApiError> {
let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
let timeline_id: TimelineId = parse_request_param(&request, "timeline_id")?;
check_permission(&request, Some(tenant_shard_id.tenant_id))?;
let state = get_state(&request);
let timeline =
active_timeline_of_active_tenant(&state.tenant_manager, tenant_shard_id, timeline_id)
.await?;
timeline.stop_and_drain_heatmap_layers_download().await;
json_response(StatusCode::OK, ())
}
async fn layer_download_handler(
request: Request<Body>,
_cancel: CancellationToken,
@@ -2152,6 +2214,7 @@ async fn timeline_compact_handler(
let state = get_state(&request);
let mut flags = EnumSet::empty();
flags |= CompactFlags::NoYield; // run compaction to completion
if Some(true) == parse_query_param::<_, bool>(&request, "force_l0_compaction")? {
flags |= CompactFlags::ForceL0Compaction;
@@ -3616,6 +3679,14 @@ pub fn make_router(
"/v1/tenant/:tenant_shard_id/timeline/:timeline_id/layer",
|r| api_handler(r, layer_map_info_handler),
)
.post(
"/v1/tenant/:tenant_shard_id/timeline/:timeline_id/download_heatmap_layers",
|r| api_handler(r, timeline_download_heatmap_layers_handler),
)
.delete(
"/v1/tenant/:tenant_shard_id/timeline/:timeline_id/download_heatmap_layers",
|r| api_handler(r, timeline_shutdown_download_heatmap_layers_handler),
)
.get(
"/v1/tenant/:tenant_shard_id/timeline/:timeline_id/layer/:layer_file_name",
|r| api_handler(r, layer_download_handler),

78
pageserver/src/metrics.rs
View File

@@ -1,5 +1,6 @@
use std::collections::HashMap;
use std::num::NonZeroUsize;
use std::os::fd::RawFd;
use std::pin::Pin;
use std::sync::atomic::AtomicU64;
use std::sync::{Arc, Mutex};
@@ -129,7 +130,7 @@ pub(crate) static LAYERS_PER_READ: Lazy<HistogramVec> = Lazy::new(|| {
"Layers visited to serve a single read (read amplification). In a batch, all visited layers count towards every read.",
&["tenant_id", "shard_id", "timeline_id"],
// Low resolution to reduce cardinality.
vec![1.0, 5.0, 10.0, 25.0, 50.0, 100.0],
vec![4.0, 8.0, 16.0, 32.0, 64.0, 128.0, 256.0],
)
.expect("failed to define a metric")
});
@@ -1366,10 +1367,7 @@ impl SmgrOpTimer {
/// The first callers receives Some, subsequent ones None.
///
/// See [`SmgrOpTimerState`] for more context.
pub(crate) fn observe_execution_end_flush_start(
&mut self,
at: Instant,
) -> Option<SmgrOpFlushInProgress> {
pub(crate) fn observe_execution_end(&mut self, at: Instant) -> Option<SmgrOpFlushInProgress> {
// NB: unlike the other observe_* methods, this one take()s.
#[allow(clippy::question_mark)] // maintain similar code pattern.
let Some(mut inner) = self.0.take() else {
@@ -1403,7 +1401,6 @@ impl SmgrOpTimer {
..
} = inner;
Some(SmgrOpFlushInProgress {
flush_started_at: at,
global_micros: global_flush_in_progress_micros,
per_timeline_micros: per_timeline_flush_in_progress_micros,
})
@@ -1419,7 +1416,6 @@ impl SmgrOpTimer {
/// add another `observe_*` method to [`SmgrOpTimer`], follow the existing pattern there,
/// and remove this struct from the code base.
pub(crate) struct SmgrOpFlushInProgress {
flush_started_at: Instant,
global_micros: IntCounter,
per_timeline_micros: IntCounter,
}
@@ -1438,32 +1434,72 @@ impl Drop for SmgrOpTimer {
self.observe_throttle_start(now);
self.observe_throttle_done(ThrottleResult::NotThrottled { end: now });
self.observe_execution_start(now);
self.observe_execution_end_flush_start(now);
let maybe_flush_timer = self.observe_execution_end(now);
drop(maybe_flush_timer);
}
}
impl SmgrOpFlushInProgress {
pub(crate) async fn measure<Fut, O>(mut self, mut fut: Fut) -> O
/// The caller must guarantee that `socket_fd`` outlives this function.
pub(crate) async fn measure<Fut, O>(
self,
started_at: Instant,
mut fut: Fut,
socket_fd: RawFd,
) -> O
where
Fut: std::future::Future<Output = O>,
{
let mut fut = std::pin::pin!(fut);
// Whenever observe_guard gets called, or dropped,
// it adds the time elapsed since its last call to metrics.
// Last call is tracked in `now`.
let mut logged = false;
let mut last_counter_increment_at = started_at;
let mut observe_guard = scopeguard::guard(
|| {
|is_timeout| {
let now = Instant::now();
let elapsed = now - self.flush_started_at;
self.global_micros
.inc_by(u64::try_from(elapsed.as_micros()).unwrap());
self.per_timeline_micros
.inc_by(u64::try_from(elapsed.as_micros()).unwrap());
self.flush_started_at = now;
// Increment counter
{
let elapsed_since_last_observe = now - last_counter_increment_at;
self.global_micros
.inc_by(u64::try_from(elapsed_since_last_observe.as_micros()).unwrap());
self.per_timeline_micros
.inc_by(u64::try_from(elapsed_since_last_observe.as_micros()).unwrap());
last_counter_increment_at = now;
}
// Log something on every timeout, and on completion but only if we hit a timeout.
if is_timeout || logged {
logged = true;
let elapsed_total = now - started_at;
let msg = if is_timeout {
"slow flush ongoing"
} else {
"slow flush completed or cancelled"
};
let (inq, outq) = {
// SAFETY: caller guarantees that `socket_fd` outlives this function.
#[cfg(target_os = "linux")]
unsafe {
(
utils::linux_socket_ioctl::inq(socket_fd).unwrap_or(-2),
utils::linux_socket_ioctl::outq(socket_fd).unwrap_or(-2),
)
}
#[cfg(not(target_os = "linux"))]
{
_ = socket_fd; // appease unused lint on macOS
(-1, -1)
}
};
let elapsed_total_secs = format!("{:.6}", elapsed_total.as_secs_f64());
tracing::info!(elapsed_total_secs, inq, outq, msg);
}
},
|mut observe| {
observe();
observe(false);
},
);
@@ -1471,7 +1507,7 @@ impl SmgrOpFlushInProgress {
match tokio::time::timeout(Duration::from_secs(10), &mut fut).await {
Ok(v) => return v,
Err(_timeout) => {
(*observe_guard)();
(*observe_guard)(true);
}
}
}

207
pageserver/src/page_service.rs
View File

@@ -73,6 +73,7 @@ use pageserver_api::models::PageTraceEvent;
use pageserver_api::reltag::SlruKind;
use postgres_ffi::pg_constants::DEFAULTTABLESPACE_OID;
use postgres_ffi::BLCKSZ;
use std::os::fd::AsRawFd;
/// How long we may wait for a [`crate::tenant::mgr::TenantSlot::InProgress`]` and/or a [`crate::tenant::Tenant`] which
/// is not yet in state [`TenantState::Active`].
@@ -236,7 +237,7 @@ pub async fn libpq_listener_main(
type ConnectionHandlerResult = anyhow::Result<()>;
#[instrument(skip_all, fields(peer_addr))]
#[instrument(skip_all, fields(peer_addr, application_name))]
#[allow(clippy::too_many_arguments)]
async fn page_service_conn_main(
conf: &'static PageServerConf,
@@ -257,6 +258,8 @@ async fn page_service_conn_main(
.set_nodelay(true)
.context("could not set TCP_NODELAY")?;
let socket_fd = socket.as_raw_fd();
let peer_addr = socket.peer_addr().context("get peer address")?;
tracing::Span::current().record("peer_addr", field::display(peer_addr));
@@ -305,7 +308,7 @@ async fn page_service_conn_main(
cancel.clone(),
gate_guard,
);
let pgbackend = PostgresBackend::new_from_io(socket, peer_addr, auth_type, None)?;
let pgbackend = PostgresBackend::new_from_io(socket_fd, socket, peer_addr, auth_type, None)?;
match pgbackend.run(&mut conn_handler, &cancel).await {
Ok(()) => {
@@ -489,7 +492,6 @@ impl timeline::handle::TenantManager<TenantManagerTypes> for TenantManagerWrappe
let timeline = tenant_shard
.get_timeline(timeline_id, true)
.map_err(GetActiveTimelineError::Timeline)?;
set_tracing_field_shard_id(&timeline);
Ok(timeline)
}
}
@@ -774,11 +776,11 @@ impl PageServerHandler {
let batched_msg = match neon_fe_msg {
PagestreamFeMessage::Exists(req) => {
let span = tracing::info_span!(parent: parent_span, "handle_get_rel_exists_request", rel = %req.rel, req_lsn = %req.hdr.request_lsn);
let shard = timeline_handles
.get(tenant_id, timeline_id, ShardSelector::Zero)
.instrument(span.clone()) // sets `shard_id` field
.await?;
debug_assert_current_span_has_tenant_and_timeline_id_no_shard_id();
let span = tracing::info_span!(parent: &parent_span, "handle_get_rel_exists_request", rel = %req.rel, req_lsn = %req.hdr.request_lsn, shard_id = %shard.tenant_shard_id.shard_slug());
let timer = record_op_start_and_throttle(
&shard,
metrics::SmgrQueryType::GetRelExists,
@@ -793,11 +795,10 @@ impl PageServerHandler {
}
}
PagestreamFeMessage::Nblocks(req) => {
let span = tracing::info_span!(parent: parent_span, "handle_get_nblocks_request", rel = %req.rel, req_lsn = %req.hdr.request_lsn);
let shard = timeline_handles
.get(tenant_id, timeline_id, ShardSelector::Zero)
.instrument(span.clone()) // sets `shard_id` field
.await?;
let span = tracing::info_span!(parent: &parent_span, "handle_get_nblocks_request", rel = %req.rel, req_lsn = %req.hdr.request_lsn, shard_id = %shard.tenant_shard_id.shard_slug());
let timer = record_op_start_and_throttle(
&shard,
metrics::SmgrQueryType::GetRelSize,
@@ -812,11 +813,10 @@ impl PageServerHandler {
}
}
PagestreamFeMessage::DbSize(req) => {
let span = tracing::info_span!(parent: parent_span, "handle_db_size_request", dbnode = %req.dbnode, req_lsn = %req.hdr.request_lsn);
let shard = timeline_handles
.get(tenant_id, timeline_id, ShardSelector::Zero)
.instrument(span.clone()) // sets `shard_id` field
.await?;
let span = tracing::info_span!(parent: &parent_span, "handle_db_size_request", dbnode = %req.dbnode, req_lsn = %req.hdr.request_lsn, shard_id = %shard.tenant_shard_id.shard_slug());
let timer = record_op_start_and_throttle(
&shard,
metrics::SmgrQueryType::GetDbSize,
@@ -831,11 +831,10 @@ impl PageServerHandler {
}
}
PagestreamFeMessage::GetSlruSegment(req) => {
let span = tracing::info_span!(parent: parent_span, "handle_get_slru_segment_request", kind = %req.kind, segno = %req.segno, req_lsn = %req.hdr.request_lsn);
let shard = timeline_handles
.get(tenant_id, timeline_id, ShardSelector::Zero)
.instrument(span.clone()) // sets `shard_id` field
.await?;
let span = tracing::info_span!(parent: &parent_span, "handle_get_slru_segment_request", kind = %req.kind, segno = %req.segno, req_lsn = %req.hdr.request_lsn, shard_id = %shard.tenant_shard_id.shard_slug());
let timer = record_op_start_and_throttle(
&shard,
metrics::SmgrQueryType::GetSlruSegment,
@@ -850,12 +849,20 @@ impl PageServerHandler {
}
}
PagestreamFeMessage::GetPage(req) => {
let span = tracing::info_span!(parent: parent_span, "handle_get_page_at_lsn_request_batched", req_lsn = %req.hdr.request_lsn);
// avoid a somewhat costly Span::record() by constructing the entire span in one go.
macro_rules! mkspan {
(before shard routing) => {{
tracing::info_span!(parent: &parent_span, "handle_get_page_request", rel = %req.rel, blkno = %req.blkno, req_lsn = %req.hdr.request_lsn)
}};
($shard_id:expr) => {{
tracing::info_span!(parent: &parent_span, "handle_get_page_request", rel = %req.rel, blkno = %req.blkno, req_lsn = %req.hdr.request_lsn, shard_id = %$shard_id)
}};
}
macro_rules! respond_error {
($error:expr) => {{
($span:expr, $error:expr) => {{
let error = BatchedFeMessage::RespondError {
span,
span: $span,
error: BatchedPageStreamError {
req: req.hdr,
err: $error,
@@ -868,27 +875,35 @@ impl PageServerHandler {
let key = rel_block_to_key(req.rel, req.blkno);
let shard = match timeline_handles
.get(tenant_id, timeline_id, ShardSelector::Page(key))
.instrument(span.clone()) // sets `shard_id` field
.await
{
Ok(tl) => tl,
Err(GetActiveTimelineError::Tenant(GetActiveTenantError::NotFound(_))) => {
// We already know this tenant exists in general, because we resolved it at
// start of connection. Getting a NotFound here indicates that the shard containing
// the requested page is not present on this node: the client's knowledge of shard->pageserver
// mapping is out of date.
//
// Closing the connection by returning ``::Reconnect` has the side effect of rate-limiting above message, via
// client's reconnect backoff, as well as hopefully prompting the client to load its updated configuration
// and talk to a different pageserver.
return respond_error!(PageStreamError::Reconnect(
"getpage@lsn request routed to wrong shard".into()
));
}
Err(e) => {
return respond_error!(e.into());
let span = mkspan!(before shard routing);
match e {
GetActiveTimelineError::Tenant(GetActiveTenantError::NotFound(_)) => {
// We already know this tenant exists in general, because we resolved it at
// start of connection. Getting a NotFound here indicates that the shard containing
// the requested page is not present on this node: the client's knowledge of shard->pageserver
// mapping is out of date.
//
// Closing the connection by returning ``::Reconnect` has the side effect of rate-limiting above message, via
// client's reconnect backoff, as well as hopefully prompting the client to load its updated configuration
// and talk to a different pageserver.
return respond_error!(
span,
PageStreamError::Reconnect(
"getpage@lsn request routed to wrong shard".into()
)
);
}
e => {
return respond_error!(span, e.into());
}
}
}
};
let span = mkspan!(shard.tenant_shard_id.shard_slug());
let timer = record_op_start_and_throttle(
&shard,
@@ -902,7 +917,7 @@ impl PageServerHandler {
&shard,
req.hdr.request_lsn,
req.hdr.not_modified_since,
&shard.get_latest_gc_cutoff_lsn(),
&shard.get_applied_gc_cutoff_lsn(),
ctx,
)
// TODO: if we actually need to wait for lsn here, it delays the entire batch which doesn't need to wait
@@ -910,7 +925,7 @@ impl PageServerHandler {
{
Ok(lsn) => lsn,
Err(e) => {
return respond_error!(e);
return respond_error!(span, e);
}
};
BatchedFeMessage::GetPage {
@@ -922,11 +937,10 @@ impl PageServerHandler {
}
#[cfg(feature = "testing")]
PagestreamFeMessage::Test(req) => {
let span = tracing::info_span!(parent: parent_span, "handle_test_request");
let shard = timeline_handles
.get(tenant_id, timeline_id, ShardSelector::Zero)
.instrument(span.clone()) // sets `shard_id` field
.await?;
let span = tracing::info_span!(parent: &parent_span, "handle_test_request", shard_id = %shard.tenant_shard_id.shard_slug());
let timer =
record_op_start_and_throttle(&shard, metrics::SmgrQueryType::Test, received_at)
.await?;
@@ -1063,7 +1077,7 @@ impl PageServerHandler {
};
// invoke handler function
let (handler_results, span): (
let (mut handler_results, span): (
Vec<Result<(PagestreamBeMessage, SmgrOpTimer), BatchedPageStreamError>>,
_,
) = match batch {
@@ -1190,11 +1204,49 @@ impl PageServerHandler {
}
};
// We purposefully don't count flush time into the smgr operation timer.
//
// The reason is that current compute client will not perform protocol processing
// if the postgres backend process is doing things other than `->smgr_read()`.
// This is especially the case for prefetch.
//
// If the compute doesn't read from the connection, eventually TCP will backpressure
// all the way into our flush call below.
//
// The timer's underlying metric is used for a storage-internal latency SLO and
// we don't want to include latency in it that we can't control.
// And as pointed out above, in this case, we don't control the time that flush will take.
//
// We put each response in the batch onto the wire in a separate pgb_writer.flush()
// call, which (all unmeasured) adds syscall overhead but reduces time to first byte
// and avoids building up a "giant" contiguous userspace buffer to hold the entire response.
// TODO: vectored socket IO would be great, but pgb_writer doesn't support that.
let flush_timers = {
let flushing_start_time = Instant::now();
let mut flush_timers = Vec::with_capacity(handler_results.len());
for handler_result in &mut handler_results {
let flush_timer = match handler_result {
Ok((_, timer)) => Some(
timer
.observe_execution_end(flushing_start_time)
.expect("we are the first caller"),
),
Err(_) => {
// TODO: measure errors
None
}
};
flush_timers.push(flush_timer);
}
assert_eq!(flush_timers.len(), handler_results.len());
flush_timers
};
// Map handler result to protocol behavior.
// Some handler errors cause exit from pagestream protocol.
// Other handler errors are sent back as an error message and we stay in pagestream protocol.
for handler_result in handler_results {
let (response_msg, timer) = match handler_result {
for (handler_result, flushing_timer) in handler_results.into_iter().zip(flush_timers) {
let response_msg = match handler_result {
Err(e) => match &e.err {
PageStreamError::Shutdown => {
// If we fail to fulfil a request during shutdown, which may be _because_ of
@@ -1218,16 +1270,14 @@ impl PageServerHandler {
span.in_scope(|| {
error!("error reading relation or page version: {full:#}")
});
(
PagestreamBeMessage::Error(PagestreamErrorResponse {
req: e.req,
message: e.err.to_string(),
}),
None, // TODO: measure errors
)
PagestreamBeMessage::Error(PagestreamErrorResponse {
req: e.req,
message: e.err.to_string(),
})
}
},
Ok((response_msg, timer)) => (response_msg, Some(timer)),
Ok((response_msg, _op_timer_already_observed)) => response_msg,
};
//
@@ -1238,31 +1288,16 @@ impl PageServerHandler {
&response_msg.serialize(protocol_version),
))?;
// We purposefully don't count flush time into the timer.
//
// The reason is that current compute client will not perform protocol processing
// if the postgres backend process is doing things other than `->smgr_read()`.
// This is especially the case for prefetch.
//
// If the compute doesn't read from the connection, eventually TCP will backpressure
// all the way into our flush call below.
//
// The timer's underlying metric is used for a storage-internal latency SLO and
// we don't want to include latency in it that we can't control.
// And as pointed out above, in this case, we don't control the time that flush will take.
let flushing_timer = timer.map(|mut timer| {
timer
.observe_execution_end_flush_start(Instant::now())
.expect("we are the first caller")
});
// what we want to do
let socket_fd = pgb_writer.socket_fd;
let flush_fut = pgb_writer.flush();
// metric for how long flushing takes
let flush_fut = match flushing_timer {
Some(flushing_timer) => {
futures::future::Either::Left(flushing_timer.measure(flush_fut))
}
Some(flushing_timer) => futures::future::Either::Left(flushing_timer.measure(
Instant::now(),
flush_fut,
socket_fd,
)),
None => futures::future::Either::Right(flush_fut),
};
// do it while respecting cancellation
@@ -1340,7 +1375,7 @@ impl PageServerHandler {
.take()
.expect("implementation error: timeline_handles should not be locked");
let request_span = info_span!("request", shard_id = tracing::field::Empty);
let request_span = info_span!("request");
let ((pgb_reader, timeline_handles), result) = match self.pipelining_config.clone() {
PageServicePipeliningConfig::Pipelined(pipelining_config) => {
self.handle_pagerequests_pipelined(
@@ -1764,6 +1799,13 @@ impl PageServerHandler {
.as_millis()
.to_string()
});
info!(
"acquired lease for {} until {}",
lsn,
valid_until_str.as_deref().unwrap_or("<unknown>")
);
let bytes = valid_until_str.as_ref().map(|x| x.as_bytes());
pgb.write_message_noflush(&BeMessage::RowDescription(&[RowDescriptor::text_col(
@@ -1781,7 +1823,7 @@ impl PageServerHandler {
req: &PagestreamExistsRequest,
ctx: &RequestContext,
) -> Result<PagestreamBeMessage, PageStreamError> {
let latest_gc_cutoff_lsn = timeline.get_latest_gc_cutoff_lsn();
let latest_gc_cutoff_lsn = timeline.get_applied_gc_cutoff_lsn();
let lsn = Self::wait_or_get_last_lsn(
timeline,
req.hdr.request_lsn,
@@ -1808,7 +1850,7 @@ impl PageServerHandler {
req: &PagestreamNblocksRequest,
ctx: &RequestContext,
) -> Result<PagestreamBeMessage, PageStreamError> {
let latest_gc_cutoff_lsn = timeline.get_latest_gc_cutoff_lsn();
let latest_gc_cutoff_lsn = timeline.get_applied_gc_cutoff_lsn();
let lsn = Self::wait_or_get_last_lsn(
timeline,
req.hdr.request_lsn,
@@ -1835,7 +1877,7 @@ impl PageServerHandler {
req: &PagestreamDbSizeRequest,
ctx: &RequestContext,
) -> Result<PagestreamBeMessage, PageStreamError> {
let latest_gc_cutoff_lsn = timeline.get_latest_gc_cutoff_lsn();
let latest_gc_cutoff_lsn = timeline.get_applied_gc_cutoff_lsn();
let lsn = Self::wait_or_get_last_lsn(
timeline,
req.hdr.request_lsn,
@@ -1925,7 +1967,7 @@ impl PageServerHandler {
req: &PagestreamGetSlruSegmentRequest,
ctx: &RequestContext,
) -> Result<PagestreamBeMessage, PageStreamError> {
let latest_gc_cutoff_lsn = timeline.get_latest_gc_cutoff_lsn();
let latest_gc_cutoff_lsn = timeline.get_applied_gc_cutoff_lsn();
let lsn = Self::wait_or_get_last_lsn(
timeline,
req.hdr.request_lsn,
@@ -2021,7 +2063,8 @@ impl PageServerHandler {
{
fn map_basebackup_error(err: BasebackupError) -> QueryError {
match err {
BasebackupError::Client(e) => QueryError::Disconnected(ConnectionError::Io(e)),
// TODO: passthrough the error site to the final error message?
BasebackupError::Client(e, _) => QueryError::Disconnected(ConnectionError::Io(e)),
BasebackupError::Server(e) => QueryError::Other(e),
}
}
@@ -2034,6 +2077,7 @@ impl PageServerHandler {
.unwrap()
.get(tenant_id, timeline_id, ShardSelector::Zero)
.await?;
set_tracing_field_shard_id(&timeline);
if timeline.is_archived() == Some(true) {
// TODO after a grace period, turn this log line into a hard error
@@ -2041,7 +2085,7 @@ impl PageServerHandler {
//return Err(QueryError::NotFound("timeline is archived".into()))
}
let latest_gc_cutoff_lsn = timeline.get_latest_gc_cutoff_lsn();
let latest_gc_cutoff_lsn = timeline.get_applied_gc_cutoff_lsn();
if let Some(lsn) = lsn {
// Backup was requested at a particular LSN. Wait for it to arrive.
info!("waiting for {}", lsn);
@@ -2121,10 +2165,12 @@ impl PageServerHandler {
.await
.map_err(map_basebackup_error)?;
}
writer
.flush()
.await
.map_err(|e| map_basebackup_error(BasebackupError::Client(e)))?;
writer.flush().await.map_err(|e| {
map_basebackup_error(BasebackupError::Client(
e,
"handle_basebackup_request,flush",
))
})?;
}
pgb.write_message_noflush(&BeMessage::CopyDone)
@@ -2424,9 +2470,16 @@ where
fn startup(
&mut self,
_pgb: &mut PostgresBackend<IO>,
_sm: &FeStartupPacket,
sm: &FeStartupPacket,
) -> Result<(), QueryError> {
fail::fail_point!("ps::connection-start::startup-packet");
if let FeStartupPacket::StartupMessage { params, .. } = sm {
if let Some(app_name) = params.get("application_name") {
Span::current().record("application_name", field::display(app_name));
}
};
Ok(())
}

324
pageserver/src/pgdatadir_mapping.rs
View File

@@ -23,13 +23,14 @@ use anyhow::{ensure, Context};
use bytes::{Buf, Bytes, BytesMut};
use enum_map::Enum;
use itertools::Itertools;
use pageserver_api::key::Key;
use pageserver_api::key::{
dbdir_key_range, rel_block_to_key, rel_dir_to_key, rel_key_range, rel_size_to_key,
relmap_file_key, repl_origin_key, repl_origin_key_range, slru_block_to_key, slru_dir_to_key,
slru_segment_key_range, slru_segment_size_to_key, twophase_file_key, twophase_key_range,
CompactKey, AUX_FILES_KEY, CHECKPOINT_KEY, CONTROLFILE_KEY, DBDIR_KEY, TWOPHASEDIR_KEY,
rel_tag_sparse_key_range, relmap_file_key, repl_origin_key, repl_origin_key_range,
slru_block_to_key, slru_dir_to_key, slru_segment_key_range, slru_segment_size_to_key,
twophase_file_key, twophase_key_range, CompactKey, RelDirExists, AUX_FILES_KEY, CHECKPOINT_KEY,
CONTROLFILE_KEY, DBDIR_KEY, TWOPHASEDIR_KEY,
};
use pageserver_api::key::{rel_tag_sparse_key, Key};
use pageserver_api::keyspace::SparseKeySpace;
use pageserver_api::record::NeonWalRecord;
use pageserver_api::reltag::{BlockNumber, RelTag, SlruKind};
@@ -490,12 +491,33 @@ impl Timeline {
if !dbdirs.contains_key(&(tag.spcnode, tag.dbnode)) {
return Ok(false);
}
// fetch directory listing
// Read path: first read the new reldir keyspace. Early return if the relation exists.
// Otherwise, read the old reldir keyspace.
// TODO: if IndexPart::rel_size_migration is `Migrated`, we only need to read from v2.
if self.get_rel_size_v2_enabled() {
// fetch directory listing (new)
let key = rel_tag_sparse_key(tag.spcnode, tag.dbnode, tag.relnode, tag.forknum);
let buf = RelDirExists::decode_option(version.sparse_get(self, key, ctx).await?)
.map_err(|_| PageReconstructError::Other(anyhow::anyhow!("invalid reldir key")))?;
let exists_v2 = buf == RelDirExists::Exists;
// Fast path: if the relation exists in the new format, return true.
// TODO: we should have a verification mode that checks both keyspaces
// to ensure the relation only exists in one of them.
if exists_v2 {
return Ok(true);
}
}
// fetch directory listing (old)
let key = rel_dir_to_key(tag.spcnode, tag.dbnode);
let buf = version.get(self, key, ctx).await?;
let dir = RelDirectory::des(&buf)?;
Ok(dir.rels.contains(&(tag.relnode, tag.forknum)))
let exists_v1 = dir.rels.contains(&(tag.relnode, tag.forknum));
Ok(exists_v1)
}
/// Get a list of all existing relations in given tablespace and database.
@@ -513,12 +535,12 @@ impl Timeline {
version: Version<'_>,
ctx: &RequestContext,
) -> Result<HashSet<RelTag>, PageReconstructError> {
// fetch directory listing
// fetch directory listing (old)
let key = rel_dir_to_key(spcnode, dbnode);
let buf = version.get(self, key, ctx).await?;
let dir = RelDirectory::des(&buf)?;
let rels: HashSet<RelTag> =
let rels_v1: HashSet<RelTag> =
HashSet::from_iter(dir.rels.iter().map(|(relnode, forknum)| RelTag {
spcnode,
dbnode,
@@ -526,6 +548,46 @@ impl Timeline {
forknum: *forknum,
}));
if !self.get_rel_size_v2_enabled() {
return Ok(rels_v1);
}
// scan directory listing (new), merge with the old results
let key_range = rel_tag_sparse_key_range(spcnode, dbnode);
let io_concurrency = IoConcurrency::spawn_from_conf(
self.conf,
self.gate
.enter()
.map_err(|_| PageReconstructError::Cancelled)?,
);
let results = self
.scan(
KeySpace::single(key_range),
version.get_lsn(),
ctx,
io_concurrency,
)
.await?;
let mut rels = rels_v1;
for (key, val) in results {
let val = RelDirExists::decode(&val?)
.map_err(|_| PageReconstructError::Other(anyhow::anyhow!("invalid reldir key")))?;
assert_eq!(key.field6, 1);
assert_eq!(key.field2, spcnode);
assert_eq!(key.field3, dbnode);
let tag = RelTag {
spcnode,
dbnode,
relnode: key.field4,
forknum: key.field5,
};
if val == RelDirExists::Removed {
debug_assert!(!rels.contains(&tag), "removed reltag in v2");
continue;
}
let did_not_contain = rels.insert(tag);
debug_assert!(did_not_contain, "duplicate reltag in v2");
}
Ok(rels)
}
@@ -611,7 +673,7 @@ impl Timeline {
) -> Result<LsnForTimestamp, PageReconstructError> {
pausable_failpoint!("find-lsn-for-timestamp-pausable");
let gc_cutoff_lsn_guard = self.get_latest_gc_cutoff_lsn();
let gc_cutoff_lsn_guard = self.get_applied_gc_cutoff_lsn();
let gc_cutoff_planned = {
let gc_info = self.gc_info.read().unwrap();
gc_info.min_cutoff()
@@ -1144,7 +1206,11 @@ impl Timeline {
let dense_keyspace = result.to_keyspace();
let sparse_keyspace = SparseKeySpace(KeySpace {
ranges: vec![Key::metadata_aux_key_range(), repl_origin_key_range()],
ranges: vec![
Key::metadata_aux_key_range(),
repl_origin_key_range(),
Key::rel_dir_sparse_key_range(),
],
});
if cfg!(debug_assertions) {
@@ -1274,12 +1340,22 @@ pub struct DatadirModification<'a> {
/// For special "directory" keys that store key-value maps, track the size of the map
/// if it was updated in this modification.
pending_directory_entries: Vec<(DirectoryKind, usize)>,
pending_directory_entries: Vec<(DirectoryKind, MetricsUpdate)>,
/// An **approximation** of how many metadata bytes will be written to the EphemeralFile.
pending_metadata_bytes: usize,
}
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum MetricsUpdate {
/// Set the metrics to this value
Set(u64),
/// Increment the metrics by this value
Add(u64),
/// Decrement the metrics by this value
Sub(u64),
}
impl DatadirModification<'_> {
// When a DatadirModification is committed, we do a monolithic serialization of all its contents. WAL records can
// contain multiple pages, so the pageserver's record-based batch size isn't sufficient to bound this allocation: we
@@ -1359,7 +1435,8 @@ impl DatadirModification<'_> {
let buf = DbDirectory::ser(&DbDirectory {
dbdirs: HashMap::new(),
})?;
self.pending_directory_entries.push((DirectoryKind::Db, 0));
self.pending_directory_entries
.push((DirectoryKind::Db, MetricsUpdate::Set(0)));
self.put(DBDIR_KEY, Value::Image(buf.into()));
let buf = if self.tline.pg_version >= 17 {
@@ -1372,7 +1449,7 @@ impl DatadirModification<'_> {
})
}?;
self.pending_directory_entries
.push((DirectoryKind::TwoPhase, 0));
.push((DirectoryKind::TwoPhase, MetricsUpdate::Set(0)));
self.put(TWOPHASEDIR_KEY, Value::Image(buf.into()));
let buf: Bytes = SlruSegmentDirectory::ser(&SlruSegmentDirectory::default())?.into();
@@ -1382,17 +1459,23 @@ impl DatadirModification<'_> {
// harmless but they'd just be dropped on later compaction.
if self.tline.tenant_shard_id.is_shard_zero() {
self.put(slru_dir_to_key(SlruKind::Clog), empty_dir.clone());
self.pending_directory_entries
.push((DirectoryKind::SlruSegment(SlruKind::Clog), 0));
self.pending_directory_entries.push((
DirectoryKind::SlruSegment(SlruKind::Clog),
MetricsUpdate::Set(0),
));
self.put(
slru_dir_to_key(SlruKind::MultiXactMembers),
empty_dir.clone(),
);
self.pending_directory_entries
.push((DirectoryKind::SlruSegment(SlruKind::Clog), 0));
self.pending_directory_entries.push((
DirectoryKind::SlruSegment(SlruKind::Clog),
MetricsUpdate::Set(0),
));
self.put(slru_dir_to_key(SlruKind::MultiXactOffsets), empty_dir);
self.pending_directory_entries
.push((DirectoryKind::SlruSegment(SlruKind::MultiXactOffsets), 0));
self.pending_directory_entries.push((
DirectoryKind::SlruSegment(SlruKind::MultiXactOffsets),
MetricsUpdate::Set(0),
));
}
Ok(())
@@ -1658,10 +1741,16 @@ impl DatadirModification<'_> {
}
if r.is_none() {
// Create RelDirectory
// TODO: if we have fully migrated to v2, no need to create this directory
let buf = RelDirectory::ser(&RelDirectory {
rels: HashSet::new(),
})?;
self.pending_directory_entries.push((DirectoryKind::Rel, 0));
self.pending_directory_entries
.push((DirectoryKind::Rel, MetricsUpdate::Set(0)));
if self.tline.get_rel_size_v2_enabled() {
self.pending_directory_entries
.push((DirectoryKind::RelV2, MetricsUpdate::Set(0)));
}
self.put(
rel_dir_to_key(spcnode, dbnode),
Value::Image(Bytes::from(buf)),
@@ -1685,8 +1774,10 @@ impl DatadirModification<'_> {
if !dir.xids.insert(xid) {
anyhow::bail!("twophase file for xid {} already exists", xid);
}
self.pending_directory_entries
.push((DirectoryKind::TwoPhase, dir.xids.len()));
self.pending_directory_entries.push((
DirectoryKind::TwoPhase,
MetricsUpdate::Set(dir.xids.len() as u64),
));
Bytes::from(TwoPhaseDirectoryV17::ser(&dir)?)
} else {
let xid = xid as u32;
@@ -1694,8 +1785,10 @@ impl DatadirModification<'_> {
if !dir.xids.insert(xid) {
anyhow::bail!("twophase file for xid {} already exists", xid);
}
self.pending_directory_entries
.push((DirectoryKind::TwoPhase, dir.xids.len()));
self.pending_directory_entries.push((
DirectoryKind::TwoPhase,
MetricsUpdate::Set(dir.xids.len() as u64),
));
Bytes::from(TwoPhaseDirectory::ser(&dir)?)
};
self.put(TWOPHASEDIR_KEY, Value::Image(newdirbuf));
@@ -1744,8 +1837,10 @@ impl DatadirModification<'_> {
let mut dir = DbDirectory::des(&buf)?;
if dir.dbdirs.remove(&(spcnode, dbnode)).is_some() {
let buf = DbDirectory::ser(&dir)?;
self.pending_directory_entries
.push((DirectoryKind::Db, dir.dbdirs.len()));
self.pending_directory_entries.push((
DirectoryKind::Db,
MetricsUpdate::Set(dir.dbdirs.len() as u64),
));
self.put(DBDIR_KEY, Value::Image(buf.into()));
} else {
warn!(
@@ -1778,39 +1873,85 @@ impl DatadirModification<'_> {
// tablespace. Create the reldir entry for it if so.
let mut dbdir = DbDirectory::des(&self.get(DBDIR_KEY, ctx).await.context("read db")?)
.context("deserialize db")?;
let rel_dir_key = rel_dir_to_key(rel.spcnode, rel.dbnode);
let mut rel_dir =
let dbdir_exists =
if let hash_map::Entry::Vacant(e) = dbdir.dbdirs.entry((rel.spcnode, rel.dbnode)) {
// Didn't exist. Update dbdir
e.insert(false);
let buf = DbDirectory::ser(&dbdir).context("serialize db")?;
self.pending_directory_entries
.push((DirectoryKind::Db, dbdir.dbdirs.len()));
self.pending_directory_entries.push((
DirectoryKind::Db,
MetricsUpdate::Set(dbdir.dbdirs.len() as u64),
));
self.put(DBDIR_KEY, Value::Image(buf.into()));
// and create the RelDirectory
RelDirectory::default()
false
} else {
// reldir already exists, fetch it
RelDirectory::des(&self.get(rel_dir_key, ctx).await.context("read db")?)
.context("deserialize db")?
true
};
let rel_dir_key = rel_dir_to_key(rel.spcnode, rel.dbnode);
let mut rel_dir = if !dbdir_exists {
// Create the RelDirectory
RelDirectory::default()
} else {
// reldir already exists, fetch it
RelDirectory::des(&self.get(rel_dir_key, ctx).await.context("read db")?)
.context("deserialize db")?
};
// Add the new relation to the rel directory entry, and write it back
if !rel_dir.rels.insert((rel.relnode, rel.forknum)) {
return Err(RelationError::AlreadyExists);
}
self.pending_directory_entries
.push((DirectoryKind::Rel, rel_dir.rels.len()));
self.put(
rel_dir_key,
Value::Image(Bytes::from(
RelDirectory::ser(&rel_dir).context("serialize")?,
)),
);
if self.tline.get_rel_size_v2_enabled() {
let sparse_rel_dir_key =
rel_tag_sparse_key(rel.spcnode, rel.dbnode, rel.relnode, rel.forknum);
// check if the rel_dir_key exists in v2
let val = self
.sparse_get(sparse_rel_dir_key, ctx)
.await
.map_err(|e| RelationError::Other(e.into()))?;
let val = RelDirExists::decode_option(val)
.map_err(|_| RelationError::Other(anyhow::anyhow!("invalid reldir key")))?;
if val == RelDirExists::Exists {
return Err(RelationError::AlreadyExists);
}
self.put(
sparse_rel_dir_key,
Value::Image(RelDirExists::Exists.encode()),
);
if !dbdir_exists {
self.pending_directory_entries
.push((DirectoryKind::Rel, MetricsUpdate::Set(0)));
self.pending_directory_entries
.push((DirectoryKind::RelV2, MetricsUpdate::Set(0)));
// We don't write `rel_dir_key -> rel_dir.rels` back to the storage in the v2 path unless it's the initial creation.
// TODO: if we have fully migrated to v2, no need to create this directory. Otherwise, there
// will be key not found errors if we don't create an empty one for rel_size_v2.
self.put(
rel_dir_key,
Value::Image(Bytes::from(
RelDirectory::ser(&RelDirectory::default()).context("serialize")?,
)),
);
}
self.pending_directory_entries
.push((DirectoryKind::RelV2, MetricsUpdate::Add(1)));
} else {
if !dbdir_exists {
self.pending_directory_entries
.push((DirectoryKind::Rel, MetricsUpdate::Set(0)))
}
self.pending_directory_entries
.push((DirectoryKind::Rel, MetricsUpdate::Add(1)));
self.put(
rel_dir_key,
Value::Image(Bytes::from(
RelDirectory::ser(&rel_dir).context("serialize")?,
)),
);
}
// Put size
let size_key = rel_size_to_key(rel);
let buf = nblocks.to_le_bytes();
@@ -1896,9 +2037,34 @@ impl DatadirModification<'_> {
let mut dirty = false;
for rel_tag in rel_tags {
if dir.rels.remove(&(rel_tag.relnode, rel_tag.forknum)) {
let found = if dir.rels.remove(&(rel_tag.relnode, rel_tag.forknum)) {
self.pending_directory_entries
.push((DirectoryKind::Rel, MetricsUpdate::Sub(1)));
dirty = true;
true
} else if self.tline.get_rel_size_v2_enabled() {
// The rel is not found in the old reldir key, so we need to check the new sparse keyspace.
// Note that a relation can only exist in one of the two keyspaces (guaranteed by the ingestion
// logic).
let key =
rel_tag_sparse_key(spc_node, db_node, rel_tag.relnode, rel_tag.forknum);
let val = RelDirExists::decode_option(self.sparse_get(key, ctx).await?)
.map_err(|_| RelationError::Other(anyhow::anyhow!("invalid reldir key")))?;
if val == RelDirExists::Exists {
self.pending_directory_entries
.push((DirectoryKind::RelV2, MetricsUpdate::Sub(1)));
// put tombstone
self.put(key, Value::Image(RelDirExists::Removed.encode()));
// no need to set dirty to true
true
} else {
false
}
} else {
false
};
if found {
// update logical size
let size_key = rel_size_to_key(rel_tag);
let old_size = self.get(size_key, ctx).await?.get_u32_le();
@@ -1914,8 +2080,6 @@ impl DatadirModification<'_> {
if dirty {
self.put(dir_key, Value::Image(Bytes::from(RelDirectory::ser(&dir)?)));
self.pending_directory_entries
.push((DirectoryKind::Rel, dir.rels.len()));
}
}
@@ -1939,8 +2103,10 @@ impl DatadirModification<'_> {
if !dir.segments.insert(segno) {
anyhow::bail!("slru segment {kind:?}/{segno} already exists");
}
self.pending_directory_entries
.push((DirectoryKind::SlruSegment(kind), dir.segments.len()));
self.pending_directory_entries.push((
DirectoryKind::SlruSegment(kind),
MetricsUpdate::Set(dir.segments.len() as u64),
));
self.put(
dir_key,
Value::Image(Bytes::from(SlruSegmentDirectory::ser(&dir)?)),
@@ -1987,8 +2153,10 @@ impl DatadirModification<'_> {
if !dir.segments.remove(&segno) {
warn!("slru segment {:?}/{} does not exist", kind, segno);
}
self.pending_directory_entries
.push((DirectoryKind::SlruSegment(kind), dir.segments.len()));
self.pending_directory_entries.push((
DirectoryKind::SlruSegment(kind),
MetricsUpdate::Set(dir.segments.len() as u64),
));
self.put(
dir_key,
Value::Image(Bytes::from(SlruSegmentDirectory::ser(&dir)?)),
@@ -2020,8 +2188,10 @@ impl DatadirModification<'_> {
if !dir.xids.remove(&xid) {
warn!("twophase file for xid {} does not exist", xid);
}
self.pending_directory_entries
.push((DirectoryKind::TwoPhase, dir.xids.len()));
self.pending_directory_entries.push((
DirectoryKind::TwoPhase,
MetricsUpdate::Set(dir.xids.len() as u64),
));
Bytes::from(TwoPhaseDirectoryV17::ser(&dir)?)
} else {
let xid: u32 = u32::try_from(xid)?;
@@ -2030,8 +2200,10 @@ impl DatadirModification<'_> {
if !dir.xids.remove(&xid) {
warn!("twophase file for xid {} does not exist", xid);
}
self.pending_directory_entries
.push((DirectoryKind::TwoPhase, dir.xids.len()));
self.pending_directory_entries.push((
DirectoryKind::TwoPhase,
MetricsUpdate::Set(dir.xids.len() as u64),
));
Bytes::from(TwoPhaseDirectory::ser(&dir)?)
};
self.put(TWOPHASEDIR_KEY, Value::Image(newdirbuf));
@@ -2147,7 +2319,7 @@ impl DatadirModification<'_> {
}
for (kind, count) in std::mem::take(&mut self.pending_directory_entries) {
writer.update_directory_entries_count(kind, count as u64);
writer.update_directory_entries_count(kind, count);
}
Ok(())
@@ -2233,7 +2405,7 @@ impl DatadirModification<'_> {
}
for (kind, count) in std::mem::take(&mut self.pending_directory_entries) {
writer.update_directory_entries_count(kind, count as u64);
writer.update_directory_entries_count(kind, count);
}
self.pending_metadata_bytes = 0;
@@ -2297,6 +2469,22 @@ impl DatadirModification<'_> {
self.tline.get(key, lsn, ctx).await
}
/// Get a key from the sparse keyspace. Automatically converts the missing key error
/// and the empty value into None.
async fn sparse_get(
&self,
key: Key,
ctx: &RequestContext,
) -> Result<Option<Bytes>, PageReconstructError> {
let val = self.get(key, ctx).await;
match val {
Ok(val) if val.is_empty() => Ok(None),
Ok(val) => Ok(Some(val)),
Err(PageReconstructError::MissingKey(_)) => Ok(None),
Err(e) => Err(e),
}
}
fn put(&mut self, key: Key, val: Value) {
if Self::is_data_key(&key) {
self.put_data(key.to_compact(), val)
@@ -2379,6 +2567,23 @@ impl Version<'_> {
}
}
/// Get a key from the sparse keyspace. Automatically converts the missing key error
/// and the empty value into None.
async fn sparse_get(
&self,
timeline: &Timeline,
key: Key,
ctx: &RequestContext,
) -> Result<Option<Bytes>, PageReconstructError> {
let val = self.get(timeline, key, ctx).await;
match val {
Ok(val) if val.is_empty() => Ok(None),
Ok(val) => Ok(Some(val)),
Err(PageReconstructError::MissingKey(_)) => Ok(None),
Err(e) => Err(e),
}
}
fn get_lsn(&self) -> Lsn {
match self {
Version::Lsn(lsn) => *lsn,
@@ -2438,6 +2643,7 @@ pub(crate) enum DirectoryKind {
Rel,
AuxFiles,
SlruSegment(SlruKind),
RelV2,
}
impl DirectoryKind {

4
pageserver/src/task_mgr.rs
View File

@@ -328,8 +328,8 @@ pub enum TaskKind {
// Eviction. One per timeline.
Eviction,
// Ingest housekeeping (flushing ephemeral layers on time threshold or disk pressure)
IngestHousekeeping,
// Tenant housekeeping (flush idle ephemeral layers, shut down idle walredo, etc.).
TenantHousekeeping,
/// See [`crate::disk_usage_eviction_task`].
DiskUsageEviction,

512
pageserver/src/tenant.rs
View File

@@ -20,6 +20,7 @@ use chrono::NaiveDateTime;
use enumset::EnumSet;
use futures::stream::FuturesUnordered;
use futures::StreamExt;
use itertools::Itertools as _;
use pageserver_api::models;
use pageserver_api::models::CompactInfoResponse;
use pageserver_api::models::LsnLease;
@@ -39,6 +40,8 @@ use remote_timeline_client::manifest::{
use remote_timeline_client::UploadQueueNotReadyError;
use remote_timeline_client::FAILED_REMOTE_OP_RETRIES;
use remote_timeline_client::FAILED_UPLOAD_WARN_THRESHOLD;
use secondary::heatmap::HeatMapTenant;
use secondary::heatmap::HeatMapTimeline;
use std::collections::BTreeMap;
use std::fmt;
use std::future::Future;
@@ -51,10 +54,14 @@ use timeline::compaction::GcCompactionQueue;
use timeline::import_pgdata;
use timeline::offload::offload_timeline;
use timeline::offload::OffloadError;
use timeline::CompactFlags;
use timeline::CompactOptions;
use timeline::CompactionError;
use timeline::PreviousHeatmap;
use timeline::ShutdownMode;
use tokio::io::BufReader;
use tokio::sync::watch;
use tokio::sync::Notify;
use tokio::task::JoinSet;
use tokio_util::sync::CancellationToken;
use tracing::*;
@@ -258,6 +265,7 @@ struct TimelinePreload {
timeline_id: TimelineId,
client: RemoteTimelineClient,
index_part: Result<MaybeDeletedIndexPart, DownloadError>,
previous_heatmap: Option<PreviousHeatmap>,
}
pub(crate) struct TenantPreload {
@@ -349,6 +357,9 @@ pub struct Tenant {
/// Overhead of mutex is acceptable because compaction is done with a multi-second period.
compaction_circuit_breaker: std::sync::Mutex<CircuitBreaker>,
/// Signals the tenant compaction loop that there is L0 compaction work to be done.
pub(crate) l0_compaction_trigger: Arc<Notify>,
/// Scheduled gc-compaction tasks.
scheduled_compaction_tasks: std::sync::Mutex<HashMap<TimelineId, Arc<GcCompactionQueue>>>,
@@ -1121,6 +1132,7 @@ impl Tenant {
resources: TimelineResources,
mut index_part: IndexPart,
metadata: TimelineMetadata,
previous_heatmap: Option<PreviousHeatmap>,
ancestor: Option<Arc<Timeline>>,
cause: LoadTimelineCause,
ctx: &RequestContext,
@@ -1151,6 +1163,7 @@ impl Tenant {
let timeline = self.create_timeline_struct(
timeline_id,
&metadata,
previous_heatmap,
ancestor.clone(),
resources,
CreateTimelineCause::Load,
@@ -1550,8 +1563,18 @@ impl Tenant {
}
}
// TODO(vlad): Could go to S3 if the secondary is freezing cold and hasn't even
// pulled the first heatmap. Not entirely necessary since the storage controller
// will kick the secondary in any case and cause a download.
let maybe_heatmap_at = self.read_on_disk_heatmap().await;
let timelines = self
.load_timelines_metadata(remote_timeline_ids, remote_storage, cancel)
.load_timelines_metadata(
remote_timeline_ids,
remote_storage,
maybe_heatmap_at,
cancel,
)
.await?;
Ok(TenantPreload {
@@ -1564,6 +1587,26 @@ impl Tenant {
})
}
async fn read_on_disk_heatmap(&self) -> Option<(HeatMapTenant, std::time::Instant)> {
let on_disk_heatmap_path = self.conf.tenant_heatmap_path(&self.tenant_shard_id);
match tokio::fs::read_to_string(on_disk_heatmap_path).await {
Ok(heatmap) => match serde_json::from_str::<HeatMapTenant>(&heatmap) {
Ok(heatmap) => Some((heatmap, std::time::Instant::now())),
Err(err) => {
error!("Failed to deserialize old heatmap: {err}");
None
}
},
Err(err) => match err.kind() {
std::io::ErrorKind::NotFound => None,
_ => {
error!("Unexpected IO error reading old heatmap: {err}");
None
}
},
}
}
///
/// Background task that downloads all data for a tenant and brings it to Active state.
///
@@ -1651,7 +1694,10 @@ impl Tenant {
match index_part {
MaybeDeletedIndexPart::IndexPart(index_part) => {
timeline_ancestors.insert(timeline_id, index_part.metadata.clone());
remote_index_and_client.insert(timeline_id, (index_part, preload.client));
remote_index_and_client.insert(
timeline_id,
(index_part, preload.client, preload.previous_heatmap),
);
}
MaybeDeletedIndexPart::Deleted(index_part) => {
info!(
@@ -1670,7 +1716,7 @@ impl Tenant {
// layer file.
let sorted_timelines = tree_sort_timelines(timeline_ancestors, |m| m.ancestor_timeline())?;
for (timeline_id, remote_metadata) in sorted_timelines {
let (index_part, remote_client) = remote_index_and_client
let (index_part, remote_client, previous_heatmap) = remote_index_and_client
.remove(&timeline_id)
.expect("just put it in above");
@@ -1690,12 +1736,8 @@ impl Tenant {
timeline_id,
index_part,
remote_metadata,
TimelineResources {
remote_client,
pagestream_throttle: self.pagestream_throttle.clone(),
pagestream_throttle_metrics: self.pagestream_throttle_metrics.clone(),
l0_flush_global_state: self.l0_flush_global_state.clone(),
},
previous_heatmap,
self.get_timeline_resources_for(remote_client),
LoadTimelineCause::Attach,
ctx,
)
@@ -1844,11 +1886,13 @@ impl Tenant {
}
#[instrument(skip_all, fields(timeline_id=%timeline_id))]
#[allow(clippy::too_many_arguments)]
async fn load_remote_timeline(
self: &Arc<Self>,
timeline_id: TimelineId,
index_part: IndexPart,
remote_metadata: TimelineMetadata,
previous_heatmap: Option<PreviousHeatmap>,
resources: TimelineResources,
cause: LoadTimelineCause,
ctx: &RequestContext,
@@ -1878,6 +1922,7 @@ impl Tenant {
resources,
index_part,
remote_metadata,
previous_heatmap,
ancestor,
cause,
ctx,
@@ -1889,14 +1934,29 @@ impl Tenant {
self: &Arc<Tenant>,
timeline_ids: HashSet<TimelineId>,
remote_storage: &GenericRemoteStorage,
heatmap: Option<(HeatMapTenant, std::time::Instant)>,
cancel: CancellationToken,
) -> anyhow::Result<HashMap<TimelineId, TimelinePreload>> {
let mut timeline_heatmaps = heatmap.map(|h| (h.0.into_timelines_index(), h.1));
let mut part_downloads = JoinSet::new();
for timeline_id in timeline_ids {
let cancel_clone = cancel.clone();
let previous_timeline_heatmap = timeline_heatmaps.as_mut().and_then(|hs| {
hs.0.remove(&timeline_id).map(|h| PreviousHeatmap::Active {
heatmap: h,
read_at: hs.1,
})
});
part_downloads.spawn(
self.load_timeline_metadata(timeline_id, remote_storage.clone(), cancel_clone)
.instrument(info_span!("download_index_part", %timeline_id)),
self.load_timeline_metadata(
timeline_id,
remote_storage.clone(),
previous_timeline_heatmap,
cancel_clone,
)
.instrument(info_span!("download_index_part", %timeline_id)),
);
}
@@ -1944,6 +2004,7 @@ impl Tenant {
self: &Arc<Tenant>,
timeline_id: TimelineId,
remote_storage: GenericRemoteStorage,
previous_heatmap: Option<PreviousHeatmap>,
cancel: CancellationToken,
) -> impl Future<Output = TimelinePreload> {
let client = self.build_timeline_client(timeline_id, remote_storage);
@@ -1959,6 +2020,7 @@ impl Tenant {
client,
timeline_id,
index_part,
previous_heatmap,
}
}
}
@@ -2070,7 +2132,12 @@ impl Tenant {
})?;
let timeline_preload = self
.load_timeline_metadata(timeline_id, self.remote_storage.clone(), cancel.clone())
.load_timeline_metadata(
timeline_id,
self.remote_storage.clone(),
None,
cancel.clone(),
)
.await;
let index_part = match timeline_preload.index_part {
@@ -2104,6 +2171,7 @@ impl Tenant {
timeline_id,
index_part,
remote_metadata,
None,
timeline_resources,
LoadTimelineCause::Unoffload,
&ctx,
@@ -2819,7 +2887,7 @@ impl Tenant {
};
let metadata = index_part.metadata.clone();
self
.load_remote_timeline(timeline_id, index_part, metadata, resources, LoadTimelineCause::ImportPgdata{
.load_remote_timeline(timeline_id, index_part, metadata, None, resources, LoadTimelineCause::ImportPgdata{
create_guard: timeline_create_guard, activate, }, &ctx)
.await?
.ready_to_activate()
@@ -2898,150 +2966,197 @@ impl Tenant {
.await
}
/// Perform one compaction iteration.
/// This function is periodically called by compactor task.
/// Also it can be explicitly requested per timeline through page server
/// api's 'compact' command.
/// Performs one compaction iteration. Called periodically from the compaction loop. Returns
/// whether another compaction is needed, if we still have pending work or if we yield for
/// immediate L0 compaction.
///
/// Returns whether we have pending compaction task.
/// Compaction can also be explicitly requested for a timeline via the HTTP API.
async fn compaction_iteration(
self: &Arc<Self>,
cancel: &CancellationToken,
ctx: &RequestContext,
) -> Result<CompactionOutcome, timeline::CompactionError> {
// Don't start doing work during shutdown, or when broken, we do not need those in the logs
) -> Result<CompactionOutcome, CompactionError> {
// Don't compact inactive tenants.
if !self.is_active() {
return Ok(CompactionOutcome::Done);
return Ok(CompactionOutcome::Skipped);
}
{
let conf = self.tenant_conf.load();
// Note that compaction usually requires deletions, but we don't respect
// may_delete_layers_hint here: that is because tenants in AttachedMulti
// should proceed with compaction even if they can't do deletion, to avoid
// accumulating dangerously deep stacks of L0 layers. Deletions will be
// enqueued inside RemoteTimelineClient, and executed layer if/when we transition
// to AttachedSingle state.
if !conf.location.may_upload_layers_hint() {
info!("Skipping compaction in location state {:?}", conf.location);
return Ok(CompactionOutcome::Done);
}
// Don't compact tenants that can't upload layers. We don't check `may_delete_layers_hint`,
// since we need to compact L0 even in AttachedMulti to bound read amplification.
let location = self.tenant_conf.load().location;
if !location.may_upload_layers_hint() {
info!("skipping compaction in location state {location:?}");
return Ok(CompactionOutcome::Skipped);
}
// Scan through the hashmap and collect a list of all the timelines,
// while holding the lock. Then drop the lock and actually perform the
// compactions. We don't want to block everything else while the
// compaction runs.
let timelines_to_compact_or_offload;
{
let timelines = self.timelines.lock().unwrap();
timelines_to_compact_or_offload = timelines
.iter()
.filter_map(|(timeline_id, timeline)| {
let (is_active, (can_offload, _)) =
(timeline.is_active(), timeline.can_offload());
let has_no_unoffloaded_children = {
!timelines
.iter()
.any(|(_id, tl)| tl.get_ancestor_timeline_id() == Some(*timeline_id))
};
let config_allows_offload = self.conf.timeline_offloading
|| self
.tenant_conf
.load()
.tenant_conf
.timeline_offloading
.unwrap_or_default();
let can_offload =
can_offload && has_no_unoffloaded_children && config_allows_offload;
if (is_active, can_offload) == (false, false) {
None
} else {
Some((*timeline_id, timeline.clone(), (is_active, can_offload)))
}
})
.collect::<Vec<_>>();
drop(timelines);
}
// Before doing any I/O work, check our circuit breaker
// Don't compact if the circuit breaker is tripped.
if self.compaction_circuit_breaker.lock().unwrap().is_broken() {
info!("Skipping compaction due to previous failures");
return Ok(CompactionOutcome::Done);
info!("skipping compaction due to previous failures");
return Ok(CompactionOutcome::Skipped);
}
let mut has_pending_task = false;
// Collect all timelines to compact, along with offload instructions and L0 counts.
let mut compact: Vec<Arc<Timeline>> = Vec::new();
let mut offload: HashSet<TimelineId> = HashSet::new();
let mut l0_counts: HashMap<TimelineId, usize> = HashMap::new();
for (timeline_id, timeline, (can_compact, can_offload)) in &timelines_to_compact_or_offload
{
// pending_task_left == None: cannot compact, maybe still pending tasks
// pending_task_left == Some(Pending): compaction task left
// pending_task_left == Some(Done): no compaction task left
let pending_task_left = if *can_compact {
let compaction_outcome = timeline
.compact(cancel, EnumSet::empty(), ctx)
.instrument(info_span!("compact_timeline", %timeline_id))
.await
.inspect_err(|e| match e {
timeline::CompactionError::ShuttingDown => (),
timeline::CompactionError::Offload(_) => {
// Failures to offload timelines do not trip the circuit breaker, because
// they do not do lots of writes the way compaction itself does: it is cheap
// to retry, and it would be bad to stop all compaction because of an issue with offloading.
}
timeline::CompactionError::Other(e) => {
self.compaction_circuit_breaker
.lock()
.unwrap()
.fail(&CIRCUIT_BREAKERS_BROKEN, e);
}
})?;
if let CompactionOutcome::Pending = compaction_outcome {
Some(CompactionOutcome::Pending)
} else {
let queue = {
let guard = self.scheduled_compaction_tasks.lock().unwrap();
guard.get(timeline_id).cloned()
};
if let Some(queue) = queue {
let outcome = queue
.iteration(cancel, ctx, &self.gc_block, timeline)
.await?;
Some(outcome)
} else {
Some(CompactionOutcome::Done)
}
let offload_enabled = self.get_timeline_offloading_enabled();
let timelines = self.timelines.lock().unwrap();
for (&timeline_id, timeline) in timelines.iter() {
// Skip inactive timelines.
if !timeline.is_active() {
continue;
}
} else {
None
};
has_pending_task |= pending_task_left == Some(CompactionOutcome::Pending);
if pending_task_left == Some(CompactionOutcome::Done) && *can_offload {
pausable_failpoint!("before-timeline-auto-offload");
match offload_timeline(self, timeline)
.instrument(info_span!("offload_timeline", %timeline_id))
.await
{
Err(OffloadError::NotArchived) => {
// Ignore this, we likely raced with unarchival
Ok(())
}
other => other,
}?;
// Schedule the timeline for compaction.
compact.push(timeline.clone());
// Schedule the timeline for offloading if eligible.
let can_offload = offload_enabled
&& timeline.can_offload().0
&& !timelines
.iter()
.any(|(_, tli)| tli.get_ancestor_timeline_id() == Some(timeline_id));
if can_offload {
offload.insert(timeline_id);
}
}
} // release timelines lock
for timeline in &compact {
// Collect L0 counts. Can't await while holding lock above.
if let Ok(lm) = timeline.layers.read().await.layer_map() {
l0_counts.insert(timeline.timeline_id, lm.level0_deltas().len());
}
}
// Pass 1: L0 compaction across all timelines, in order of L0 count. We prioritize this to
// bound read amplification.
//
// TODO: this may spin on one or more ingest-heavy timelines, starving out image/GC
// compaction and offloading. We leave that as a potential problem to solve later. Consider
// splitting L0 and image/GC compaction to separate background jobs.
if self.get_compaction_l0_first() {
let compaction_threshold = self.get_compaction_threshold();
let compact_l0 = compact
.iter()
.map(|tli| (tli, l0_counts.get(&tli.timeline_id).copied().unwrap_or(0)))
.filter(|&(_, l0)| l0 >= compaction_threshold)
.sorted_by_key(|&(_, l0)| l0)
.rev()
.map(|(tli, _)| tli.clone())
.collect_vec();
let mut has_pending_l0 = false;
for timeline in compact_l0 {
let outcome = timeline
.compact(cancel, CompactFlags::OnlyL0Compaction.into(), ctx)
.instrument(info_span!("compact_timeline", timeline_id = %timeline.timeline_id))
.await
.inspect_err(|err| self.maybe_trip_compaction_breaker(err))?;
match outcome {
CompactionOutcome::Done => {}
CompactionOutcome::Skipped => {}
CompactionOutcome::Pending => has_pending_l0 = true,
CompactionOutcome::YieldForL0 => has_pending_l0 = true,
}
}
if has_pending_l0 {
return Ok(CompactionOutcome::YieldForL0); // do another pass
}
}
// Pass 2: image compaction and timeline offloading. If any timelines have accumulated
// more L0 layers, they may also be compacted here.
//
// NB: image compaction may yield if there is pending L0 compaction.
//
// TODO: it will only yield if there is pending L0 compaction on the same timeline. If a
// different timeline needs compaction, it won't. It should check `l0_compaction_trigger`.
// We leave this for a later PR.
//
// TODO: consider ordering timelines by some priority, e.g. time since last full compaction,
// amount of L1 delta debt or garbage, offload-eligible timelines first, etc.
let mut has_pending = false;
for timeline in compact {
if !timeline.is_active() {
continue;
}
let mut outcome = timeline
.compact(cancel, EnumSet::default(), ctx)
.instrument(info_span!("compact_timeline", timeline_id = %timeline.timeline_id))
.await
.inspect_err(|err| self.maybe_trip_compaction_breaker(err))?;
// If we're done compacting, check the scheduled GC compaction queue for more work.
if outcome == CompactionOutcome::Done {
let queue = self
.scheduled_compaction_tasks
.lock()
.unwrap()
.get(&timeline.timeline_id)
.cloned();
if let Some(queue) = queue {
outcome = queue
.iteration(cancel, ctx, &self.gc_block, &timeline)
.instrument(
info_span!("gc_compact_timeline", timeline_id = %timeline.timeline_id),
)
.await?;
}
}
// If we're done compacting, offload the timeline if requested.
if outcome == CompactionOutcome::Done && offload.contains(&timeline.timeline_id) {
pausable_failpoint!("before-timeline-auto-offload");
offload_timeline(self, &timeline)
.instrument(info_span!("offload_timeline", timeline_id = %timeline.timeline_id))
.await
.or_else(|err| match err {
// Ignore this, we likely raced with unarchival.
OffloadError::NotArchived => Ok(()),
err => Err(err),
})?;
}
match outcome {
CompactionOutcome::Done => {}
CompactionOutcome::Skipped => {}
CompactionOutcome::Pending => has_pending = true,
// This mostly makes sense when the L0-only pass above is enabled, since there's
// otherwise no guarantee that we'll start with the timeline that has high L0.
CompactionOutcome::YieldForL0 => return Ok(CompactionOutcome::YieldForL0),
}
}
// Success! Untrip the breaker if necessary.
self.compaction_circuit_breaker
.lock()
.unwrap()
.success(&CIRCUIT_BREAKERS_UNBROKEN);
Ok(if has_pending_task {
CompactionOutcome::Pending
} else {
CompactionOutcome::Done
})
match has_pending {
true => Ok(CompactionOutcome::Pending),
false => Ok(CompactionOutcome::Done),
}
}
/// Trips the compaction circuit breaker if appropriate.
pub(crate) fn maybe_trip_compaction_breaker(&self, err: &CompactionError) {
match err {
CompactionError::ShuttingDown => (),
// Offload failures don't trip the circuit breaker, since they're cheap to retry and
// shouldn't block compaction.
CompactionError::Offload(_) => {}
CompactionError::Other(err) => {
self.compaction_circuit_breaker
.lock()
.unwrap()
.fail(&CIRCUIT_BREAKERS_BROKEN, err);
}
}
}
/// Cancel scheduled compaction tasks
@@ -3088,32 +3203,28 @@ impl Tenant {
Ok(rx)
}
// Call through to all timelines to freeze ephemeral layers if needed. Usually
// this happens during ingest: this background housekeeping is for freezing layers
// that are open but haven't been written to for some time.
async fn ingest_housekeeping(&self) {
// Scan through the hashmap and collect a list of all the timelines,
// while holding the lock. Then drop the lock and actually perform the
// compactions. We don't want to block everything else while the
// compaction runs.
let timelines = {
self.timelines
.lock()
.unwrap()
.values()
.filter_map(|timeline| {
if timeline.is_active() {
Some(timeline.clone())
} else {
None
}
})
.collect::<Vec<_>>()
};
/// Performs periodic housekeeping, via the tenant housekeeping background task.
async fn housekeeping(&self) {
// Call through to all timelines to freeze ephemeral layers as needed. This usually happens
// during ingest, but we don't want idle timelines to hold open layers for too long.
let timelines = self
.timelines
.lock()
.unwrap()
.values()
.filter(|tli| tli.is_active())
.cloned()
.collect_vec();
for timeline in &timelines {
for timeline in timelines {
timeline.maybe_freeze_ephemeral_layer().await;
}
// Shut down walredo if idle.
const WALREDO_IDLE_TIMEOUT: Duration = Duration::from_secs(180);
if let Some(ref walredo_mgr) = self.walredo_mgr {
walredo_mgr.maybe_quiesce(WALREDO_IDLE_TIMEOUT);
}
}
pub fn timeline_has_no_attached_children(&self, timeline_id: TimelineId) -> bool {
@@ -3816,6 +3927,13 @@ impl Tenant {
.unwrap_or(self.conf.default_tenant_conf.compaction_threshold)
}
pub fn get_rel_size_v2_enabled(&self) -> bool {
let tenant_conf = self.tenant_conf.load().tenant_conf.clone();
tenant_conf
.rel_size_v2_enabled
.unwrap_or(self.conf.default_tenant_conf.rel_size_v2_enabled)
}
pub fn get_compaction_upper_limit(&self) -> usize {
let tenant_conf = self.tenant_conf.load().tenant_conf.clone();
tenant_conf
@@ -3823,6 +3941,13 @@ impl Tenant {
.unwrap_or(self.conf.default_tenant_conf.compaction_upper_limit)
}
pub fn get_compaction_l0_first(&self) -> bool {
let tenant_conf = self.tenant_conf.load().tenant_conf.clone();
tenant_conf
.compaction_l0_first
.unwrap_or(self.conf.default_tenant_conf.compaction_l0_first)
}
pub fn get_gc_horizon(&self) -> u64 {
let tenant_conf = self.tenant_conf.load().tenant_conf.clone();
tenant_conf
@@ -3877,6 +4002,16 @@ impl Tenant {
.unwrap_or(self.conf.default_tenant_conf.lsn_lease_length)
}
pub fn get_timeline_offloading_enabled(&self) -> bool {
if self.conf.timeline_offloading {
return true;
}
let tenant_conf = self.tenant_conf.load().tenant_conf.clone();
tenant_conf
.timeline_offloading
.unwrap_or(self.conf.default_tenant_conf.timeline_offloading)
}
/// Generate an up-to-date TenantManifest based on the state of this Tenant.
fn build_tenant_manifest(&self) -> TenantManifest {
let timelines_offloaded = self.timelines_offloaded.lock().unwrap();
@@ -3971,6 +4106,7 @@ impl Tenant {
&self,
new_timeline_id: TimelineId,
new_metadata: &TimelineMetadata,
previous_heatmap: Option<PreviousHeatmap>,
ancestor: Option<Arc<Timeline>>,
resources: TimelineResources,
cause: CreateTimelineCause,
@@ -3994,6 +4130,7 @@ impl Tenant {
self.conf,
Arc::clone(&self.tenant_conf),
new_metadata,
previous_heatmap,
ancestor,
new_timeline_id,
self.tenant_shard_id,
@@ -4115,6 +4252,7 @@ impl Tenant {
// use an extremely long backoff.
Some(Duration::from_secs(3600 * 24)),
)),
l0_compaction_trigger: Arc::new(Notify::new()),
scheduled_compaction_tasks: Mutex::new(Default::default()),
activate_now_sem: tokio::sync::Semaphore::new(0),
attach_wal_lag_cooldown: Arc::new(std::sync::OnceLock::new()),
@@ -4635,24 +4773,24 @@ impl Tenant {
// We check it against both the planned GC cutoff stored in 'gc_info',
// and the 'latest_gc_cutoff' of the last GC that was performed. The
// planned GC cutoff in 'gc_info' is normally larger than
// 'latest_gc_cutoff_lsn', but beware of corner cases like if you just
// 'applied_gc_cutoff_lsn', but beware of corner cases like if you just
// changed the GC settings for the tenant to make the PITR window
// larger, but some of the data was already removed by an earlier GC
// iteration.
// check against last actual 'latest_gc_cutoff' first
let latest_gc_cutoff_lsn = src_timeline.get_latest_gc_cutoff_lsn();
let applied_gc_cutoff_lsn = src_timeline.get_applied_gc_cutoff_lsn();
{
let gc_info = src_timeline.gc_info.read().unwrap();
let planned_cutoff = gc_info.min_cutoff();
if gc_info.lsn_covered_by_lease(start_lsn) {
tracing::info!("skipping comparison of {start_lsn} with gc cutoff {} and planned gc cutoff {planned_cutoff} due to lsn lease", *latest_gc_cutoff_lsn);
tracing::info!("skipping comparison of {start_lsn} with gc cutoff {} and planned gc cutoff {planned_cutoff} due to lsn lease", *applied_gc_cutoff_lsn);
} else {
src_timeline
.check_lsn_is_in_scope(start_lsn, &latest_gc_cutoff_lsn)
.check_lsn_is_in_scope(start_lsn, &applied_gc_cutoff_lsn)
.context(format!(
"invalid branch start lsn: less than latest GC cutoff {}",
*latest_gc_cutoff_lsn,
*applied_gc_cutoff_lsn,
))
.map_err(CreateTimelineError::AncestorLsn)?;
@@ -4691,7 +4829,7 @@ impl Tenant {
dst_prev,
Some(src_id),
start_lsn,
*src_timeline.latest_gc_cutoff_lsn.read(), // FIXME: should we hold onto this guard longer?
*src_timeline.applied_gc_cutoff_lsn.read(), // FIXME: should we hold onto this guard longer?
src_timeline.initdb_lsn,
src_timeline.pg_version,
);
@@ -5023,12 +5161,19 @@ impl Tenant {
)
}
/// Call this before constructing a timeline, to build its required structures
/// Builds required resources for a new timeline.
fn build_timeline_resources(&self, timeline_id: TimelineId) -> TimelineResources {
let remote_client = self.build_timeline_remote_client(timeline_id);
self.get_timeline_resources_for(remote_client)
}
/// Builds timeline resources for the given remote client.
fn get_timeline_resources_for(&self, remote_client: RemoteTimelineClient) -> TimelineResources {
TimelineResources {
remote_client: self.build_timeline_remote_client(timeline_id),
remote_client,
pagestream_throttle: self.pagestream_throttle.clone(),
pagestream_throttle_metrics: self.pagestream_throttle_metrics.clone(),
l0_compaction_trigger: self.l0_compaction_trigger.clone(),
l0_flush_global_state: self.l0_flush_global_state.clone(),
}
}
@@ -5057,6 +5202,7 @@ impl Tenant {
.create_timeline_struct(
new_timeline_id,
new_metadata,
None,
ancestor,
resources,
CreateTimelineCause::Load,
@@ -5474,6 +5620,8 @@ pub(crate) mod harness {
compaction_threshold: Some(tenant_conf.compaction_threshold),
compaction_upper_limit: Some(tenant_conf.compaction_upper_limit),
compaction_algorithm: Some(tenant_conf.compaction_algorithm),
compaction_l0_first: Some(tenant_conf.compaction_l0_first),
compaction_l0_semaphore: Some(tenant_conf.compaction_l0_semaphore),
l0_flush_delay_threshold: tenant_conf.l0_flush_delay_threshold,
l0_flush_stall_threshold: tenant_conf.l0_flush_stall_threshold,
l0_flush_wait_upload: Some(tenant_conf.l0_flush_wait_upload),
@@ -5502,7 +5650,7 @@ pub(crate) mod harness {
lsn_lease_length_for_ts: Some(tenant_conf.lsn_lease_length_for_ts),
timeline_offloading: Some(tenant_conf.timeline_offloading),
wal_receiver_protocol_override: tenant_conf.wal_receiver_protocol_override,
rel_size_v2_enabled: tenant_conf.rel_size_v2_enabled,
rel_size_v2_enabled: Some(tenant_conf.rel_size_v2_enabled),
gc_compaction_enabled: Some(tenant_conf.gc_compaction_enabled),
gc_compaction_initial_threshold_kb: Some(
tenant_conf.gc_compaction_initial_threshold_kb,
@@ -6061,8 +6209,8 @@ mod tests {
make_some_layers(tline.as_ref(), Lsn(0x20), &ctx).await?;
repo.gc_iteration(Some(TIMELINE_ID), 0x10, Duration::ZERO)?;
let latest_gc_cutoff_lsn = tline.get_latest_gc_cutoff_lsn();
assert!(*latest_gc_cutoff_lsn > Lsn(0x25));
let applied_gc_cutoff_lsn = tline.get_applied_gc_cutoff_lsn();
assert!(*applied_gc_cutoff_lsn > Lsn(0x25));
match tline.get(*TEST_KEY, Lsn(0x25)) {
Ok(_) => panic!("request for page should have failed"),
Err(err) => assert!(err.to_string().contains("not found at")),
@@ -8344,7 +8492,7 @@ mod tests {
.await?;
{
tline
.latest_gc_cutoff_lsn
.applied_gc_cutoff_lsn
.lock_for_write()
.store_and_unlock(Lsn(0x30))
.wait()
@@ -8452,7 +8600,7 @@ mod tests {
// increase GC horizon and compact again
{
tline
.latest_gc_cutoff_lsn
.applied_gc_cutoff_lsn
.lock_for_write()
.store_and_unlock(Lsn(0x40))
.wait()
@@ -8620,8 +8768,8 @@ mod tests {
// Force set disk consistent lsn so we can get the cutoff at `end_lsn`.
info!(
"latest_gc_cutoff_lsn: {}",
*timeline.get_latest_gc_cutoff_lsn()
"applied_gc_cutoff_lsn: {}",
*timeline.get_applied_gc_cutoff_lsn()
);
timeline.force_set_disk_consistent_lsn(end_lsn);
@@ -8647,7 +8795,7 @@ mod tests {
// Make lease on a already GC-ed LSN.
// 0/80 does not have a valid lease + is below latest_gc_cutoff
assert!(Lsn(0x80) < *timeline.get_latest_gc_cutoff_lsn());
assert!(Lsn(0x80) < *timeline.get_applied_gc_cutoff_lsn());
timeline
.init_lsn_lease(Lsn(0x80), timeline.get_lsn_lease_length(), &ctx)
.expect_err("lease request on GC-ed LSN should fail");
@@ -8838,7 +8986,7 @@ mod tests {
};
{
tline
.latest_gc_cutoff_lsn
.applied_gc_cutoff_lsn
.lock_for_write()
.store_and_unlock(Lsn(0x30))
.wait()
@@ -8925,7 +9073,7 @@ mod tests {
// increase GC horizon and compact again
{
tline
.latest_gc_cutoff_lsn
.applied_gc_cutoff_lsn
.lock_for_write()
.store_and_unlock(Lsn(0x40))
.wait()
@@ -9378,7 +9526,7 @@ mod tests {
.await?;
{
tline
.latest_gc_cutoff_lsn
.applied_gc_cutoff_lsn
.lock_for_write()
.store_and_unlock(Lsn(0x30))
.wait()
@@ -9525,7 +9673,7 @@ mod tests {
// increase GC horizon and compact again
{
tline
.latest_gc_cutoff_lsn
.applied_gc_cutoff_lsn
.lock_for_write()
.store_and_unlock(Lsn(0x38))
.wait()
@@ -9626,7 +9774,7 @@ mod tests {
.await?;
{
tline
.latest_gc_cutoff_lsn
.applied_gc_cutoff_lsn
.lock_for_write()
.store_and_unlock(Lsn(0x30))
.wait()
@@ -9877,7 +10025,7 @@ mod tests {
{
parent_tline
.latest_gc_cutoff_lsn
.applied_gc_cutoff_lsn
.lock_for_write()
.store_and_unlock(Lsn(0x10))
.wait()
@@ -9897,7 +10045,7 @@ mod tests {
{
branch_tline
.latest_gc_cutoff_lsn
.applied_gc_cutoff_lsn
.lock_for_write()
.store_and_unlock(Lsn(0x50))
.wait()
@@ -10253,7 +10401,7 @@ mod tests {
{
tline
.latest_gc_cutoff_lsn
.applied_gc_cutoff_lsn
.lock_for_write()
.store_and_unlock(Lsn(0x30))
.wait()
@@ -10638,7 +10786,7 @@ mod tests {
.await?;
{
tline
.latest_gc_cutoff_lsn
.applied_gc_cutoff_lsn
.lock_for_write()
.store_and_unlock(Lsn(0x30))
.wait()
@@ -10889,7 +11037,7 @@ mod tests {
.await?;
{
tline
.latest_gc_cutoff_lsn
.applied_gc_cutoff_lsn
.lock_for_write()
.store_and_unlock(Lsn(0x30))
.wait()

30
pageserver/src/tenant/config.rs
View File

@@ -285,6 +285,14 @@ pub struct TenantConfOpt {
#[serde(default)]
pub compaction_algorithm: Option<CompactionAlgorithmSettings>,
#[serde(skip_serializing_if = "Option::is_none")]
#[serde(default)]
pub compaction_l0_first: Option<bool>,
#[serde(skip_serializing_if = "Option::is_none")]
#[serde(default)]
pub compaction_l0_semaphore: Option<bool>,
#[serde(skip_serializing_if = "Option::is_none")]
#[serde(default)]
pub l0_flush_delay_threshold: Option<usize>,
@@ -416,6 +424,12 @@ impl TenantConfOpt {
.as_ref()
.unwrap_or(&global_conf.compaction_algorithm)
.clone(),
compaction_l0_first: self
.compaction_l0_first
.unwrap_or(global_conf.compaction_l0_first),
compaction_l0_semaphore: self
.compaction_l0_semaphore
.unwrap_or(global_conf.compaction_l0_semaphore),
l0_flush_delay_threshold: self
.l0_flush_delay_threshold
.or(global_conf.l0_flush_delay_threshold),
@@ -466,12 +480,14 @@ impl TenantConfOpt {
.lsn_lease_length_for_ts
.unwrap_or(global_conf.lsn_lease_length_for_ts),
timeline_offloading: self
.lazy_slru_download
.timeline_offloading
.unwrap_or(global_conf.timeline_offloading),
wal_receiver_protocol_override: self
.wal_receiver_protocol_override
.or(global_conf.wal_receiver_protocol_override),
rel_size_v2_enabled: self.rel_size_v2_enabled.or(global_conf.rel_size_v2_enabled),
rel_size_v2_enabled: self
.rel_size_v2_enabled
.unwrap_or(global_conf.rel_size_v2_enabled),
gc_compaction_enabled: self
.gc_compaction_enabled
.unwrap_or(global_conf.gc_compaction_enabled),
@@ -493,6 +509,8 @@ impl TenantConfOpt {
mut compaction_threshold,
mut compaction_upper_limit,
mut compaction_algorithm,
mut compaction_l0_first,
mut compaction_l0_semaphore,
mut l0_flush_delay_threshold,
mut l0_flush_stall_threshold,
mut l0_flush_wait_upload,
@@ -538,6 +556,10 @@ impl TenantConfOpt {
.compaction_upper_limit
.apply(&mut compaction_upper_limit);
patch.compaction_algorithm.apply(&mut compaction_algorithm);
patch.compaction_l0_first.apply(&mut compaction_l0_first);
patch
.compaction_l0_semaphore
.apply(&mut compaction_l0_semaphore);
patch
.l0_flush_delay_threshold
.apply(&mut l0_flush_delay_threshold);
@@ -619,6 +641,8 @@ impl TenantConfOpt {
compaction_threshold,
compaction_upper_limit,
compaction_algorithm,
compaction_l0_first,
compaction_l0_semaphore,
l0_flush_delay_threshold,
l0_flush_stall_threshold,
l0_flush_wait_upload,
@@ -681,6 +705,8 @@ impl From<TenantConfOpt> for models::TenantConfig {
compaction_period: value.compaction_period.map(humantime),
compaction_threshold: value.compaction_threshold,
compaction_upper_limit: value.compaction_upper_limit,
compaction_l0_first: value.compaction_l0_first,
compaction_l0_semaphore: value.compaction_l0_semaphore,
l0_flush_delay_threshold: value.l0_flush_delay_threshold,
l0_flush_stall_threshold: value.l0_flush_stall_threshold,
l0_flush_wait_upload: value.l0_flush_wait_upload,

3
pageserver/src/tenant/metadata.rs
View File

@@ -130,7 +130,10 @@ struct TimelineMetadataBodyV2 {
prev_record_lsn: Option<Lsn>,
ancestor_timeline: Option<TimelineId>,
ancestor_lsn: Lsn,
// The LSN at which GC was last executed. Synonym of [`Timeline::applied_gc_cutoff_lsn`].
latest_gc_cutoff_lsn: Lsn,
initdb_lsn: Lsn,
pg_version: u32,
}

4
pageserver/src/tenant/mgr.rs
View File

@@ -2816,8 +2816,8 @@ where
}
use {
crate::tenant::gc_result::GcResult, pageserver_api::models::TimelineGcRequest,
utils::http::error::ApiError,
crate::tenant::gc_result::GcResult, http_utils::error::ApiError,
pageserver_api::models::TimelineGcRequest,
};
#[cfg(test)]

3
pageserver/src/tenant/remote_timeline_client.rs
View File

@@ -517,7 +517,7 @@ impl RemoteTimelineClient {
if let Ok(queue) = queue_locked.initialized_mut() {
let blocked_deletions = std::mem::take(&mut queue.blocked_deletions);
for d in blocked_deletions {
if let Err(e) = self.deletion_queue_client.push_layers_sync(
if let Err(e) = self.deletion_queue_client.push_layers(
self.tenant_shard_id,
self.timeline_id,
self.generation,
@@ -2151,7 +2151,6 @@ impl RemoteTimelineClient {
self.generation,
delete.layers.clone(),
)
.await
.map_err(|e| anyhow::anyhow!(e))
}
}

19
pageserver/src/tenant/secondary/heatmap.rs
View File

@@ -1,4 +1,4 @@
use std::time::SystemTime;
use std::{collections::HashMap, time::SystemTime};
use crate::tenant::{remote_timeline_client::index::LayerFileMetadata, storage_layer::LayerName};
@@ -8,7 +8,7 @@ use serde_with::{serde_as, DisplayFromStr, TimestampSeconds};
use utils::{generation::Generation, id::TimelineId};
#[derive(Serialize, Deserialize)]
pub(super) struct HeatMapTenant {
pub(crate) struct HeatMapTenant {
/// Generation of the attached location that uploaded the heatmap: this is not required
/// for correctness, but acts as a hint to secondary locations in order to detect thrashing
/// in the unlikely event that two attached locations are both uploading conflicting heatmaps.
@@ -25,8 +25,17 @@ pub(super) struct HeatMapTenant {
pub(super) upload_period_ms: Option<u128>,
}
impl HeatMapTenant {
pub(crate) fn into_timelines_index(self) -> HashMap<TimelineId, HeatMapTimeline> {
self.timelines
.into_iter()
.map(|htl| (htl.timeline_id, htl))
.collect()
}
}
#[serde_as]
#[derive(Serialize, Deserialize)]
#[derive(Serialize, Deserialize, Clone)]
pub(crate) struct HeatMapTimeline {
#[serde_as(as = "DisplayFromStr")]
pub(crate) timeline_id: TimelineId,
@@ -35,13 +44,13 @@ pub(crate) struct HeatMapTimeline {
}
#[serde_as]
#[derive(Serialize, Deserialize)]
#[derive(Serialize, Deserialize, Clone)]
pub(crate) struct HeatMapLayer {
pub(crate) name: LayerName,
pub(crate) metadata: LayerFileMetadata,
#[serde_as(as = "TimestampSeconds<i64>")]
pub(super) access_time: SystemTime,
pub(crate) access_time: SystemTime,
// TODO: an actual 'heat' score that would let secondary locations prioritize downloading
// the hottest layers, rather than trying to simply mirror whatever layers are on-disk on the primary.
}

2
pageserver/src/tenant/size.rs
View File

@@ -394,7 +394,7 @@ pub(super) async fn gather_inputs(
ancestor_lsn,
last_record: last_record_lsn,
// this is not used above, because it might not have updated recently enough
latest_gc_cutoff: *timeline.get_latest_gc_cutoff_lsn(),
latest_gc_cutoff: *timeline.get_applied_gc_cutoff_lsn(),
next_pitr_cutoff,
retention_param_cutoff,
lease_points,

5
pageserver/src/tenant/storage_layer.rs
View File

@@ -44,7 +44,7 @@ pub(crate) use layer::{EvictionError, Layer, ResidentLayer};
use self::inmemory_layer::InMemoryLayerFileId;
use super::timeline::GetVectoredError;
use super::timeline::{GetVectoredError, ReadPath};
use super::PageReconstructError;
pub fn range_overlaps<T>(a: &Range<T>, b: &Range<T>) -> bool
@@ -262,6 +262,8 @@ pub(crate) struct ValuesReconstructState {
pub(crate) io_concurrency: IoConcurrency,
num_active_ios: Arc<AtomicUsize>,
pub(crate) read_path: Option<ReadPath>,
}
/// The level of IO concurrency to be used on the read path
@@ -609,6 +611,7 @@ impl ValuesReconstructState {
delta_layers_visited: 0,
io_concurrency,
num_active_ios: Arc::new(AtomicUsize::new(0)),
read_path: None,
}
}

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