Compare commits

..

10 Commits

Author SHA1 Message Date
Konstantin Knizhnik
cca517fb94 Add test for prewarm under workload 2024-12-23 11:36:47 +02:00
Konstantin Knizhnik
5f74ff1307 Make ruffhappy 2024-12-15 08:02:35 +02:00
Konstantin Knizhnik
1d77fb0dea Eliminate stale reads from LFC in case of prewarm conflict 2024-12-14 21:58:26 +02:00
Konstantin Knizhnik
07027bde7d Do not run test_lfc_prewarm test without LFC 2024-12-14 21:11:20 +02:00
Konstantin Knizhnik
1f2b47c70f Set LFC path in test-lfc_prewarm test 2024-12-14 21:09:58 +02:00
Konstantin Knizhnik
7b80ad4950 Fix format warning 2024-12-14 21:09:58 +02:00
Konstantin Knizhnik
e07eedca5d correctly handle PS disconect duriug prewarm 2024-12-14 21:09:58 +02:00
Konstantin Knizhnik
7e2fb10cca Fix handling zero neon.file_cache_prewarm_limit 2024-12-14 21:09:57 +02:00
Konstantin Knizhnik
dc1684efcc Add delay between upgrade of extension version 2024-12-14 21:09:57 +02:00
Konstantin Knizhnik
ec8b8b941d Add functions to get LFC state, prewarm LFC and monitor prewarm process 2024-12-14 21:09:57 +02:00
379 changed files with 7918 additions and 19604 deletions

View File

@@ -3,16 +3,6 @@
# by the RUSTDOCFLAGS env var in CI.
rustdocflags = ["-Arustdoc::private_intra_doc_links"]
# Enable frame pointers. This may have a minor performance overhead, but makes it easier and more
# efficient to obtain stack traces (and thus CPU/heap profiles). It may also avoid seg faults that
# we've seen with libunwind-based profiling. See also:
#
# * <https://www.brendangregg.com/blog/2024-03-17/the-return-of-the-frame-pointers.html>
# * <https://github.com/rust-lang/rust/pull/122646>
#
# NB: the RUSTFLAGS envvar will replace this. Make sure to update e.g. Dockerfile as well.
rustflags = ["-Cforce-frame-pointers=yes"]
[alias]
build_testing = ["build", "--features", "testing"]
neon = ["run", "--bin", "neon_local"]

View File

@@ -25,4 +25,3 @@ config-variables:
- PGREGRESS_PG17_PROJECT_ID
- SLACK_ON_CALL_QA_STAGING_STREAM
- DEV_AWS_OIDC_ROLE_MANAGE_BENCHMARK_EC2_VMS_ARN
- SLACK_ON_CALL_STORAGE_STAGING_STREAM

View File

@@ -1,12 +0,0 @@
rust_code: ['**/*.rs', '**/Cargo.toml', '**/Cargo.lock']
v14: ['vendor/postgres-v14/**', 'Makefile', 'pgxn/**']
v15: ['vendor/postgres-v15/**', 'Makefile', 'pgxn/**']
v16: ['vendor/postgres-v16/**', 'Makefile', 'pgxn/**']
v17: ['vendor/postgres-v17/**', 'Makefile', 'pgxn/**']
rebuild_neon_extra:
- .github/workflows/neon_extra_builds.yml
rebuild_macos:
- .github/workflows/build-macos.yml

View File

@@ -1,91 +0,0 @@
name: Check Codestyle Rust
on:
workflow_call:
inputs:
build-tools-image:
description: "build-tools image"
required: true
type: string
archs:
description: "Json array of architectures to run on"
type: string
defaults:
run:
shell: bash -euxo pipefail {0}
jobs:
check-codestyle-rust:
strategy:
matrix:
arch: ${{ fromJson(inputs.archs) }}
runs-on: ${{ fromJson(format('["self-hosted", "{0}"]', matrix.arch == 'arm64' && 'small-arm64' || 'small')) }}
container:
image: ${{ inputs.build-tools-image }}
credentials:
username: ${{ secrets.NEON_DOCKERHUB_USERNAME }}
password: ${{ secrets.NEON_DOCKERHUB_PASSWORD }}
options: --init
steps:
- name: Checkout
uses: actions/checkout@v4
with:
submodules: true
- name: Cache cargo deps
uses: actions/cache@v4
with:
path: |
~/.cargo/registry
!~/.cargo/registry/src
~/.cargo/git
target
key: v1-${{ runner.os }}-${{ runner.arch }}-cargo-${{ hashFiles('./Cargo.lock') }}-${{ hashFiles('./rust-toolchain.toml') }}-rust
# Some of our rust modules use FFI and need those to be checked
- name: Get postgres headers
run: make postgres-headers -j$(nproc)
# cargo hack runs the given cargo subcommand (clippy in this case) for all feature combinations.
# This will catch compiler & clippy warnings in all feature combinations.
# TODO: use cargo hack for build and test as well, but, that's quite expensive.
# NB: keep clippy args in sync with ./run_clippy.sh
#
# The only difference between "clippy --debug" and "clippy --release" is that in --release mode,
# #[cfg(debug_assertions)] blocks are not built. It's not worth building everything for second
# time just for that, so skip "clippy --release".
- run: |
CLIPPY_COMMON_ARGS="$( source .neon_clippy_args; echo "$CLIPPY_COMMON_ARGS")"
if [ "$CLIPPY_COMMON_ARGS" = "" ]; then
echo "No clippy args found in .neon_clippy_args"
exit 1
fi
echo "CLIPPY_COMMON_ARGS=${CLIPPY_COMMON_ARGS}" >> $GITHUB_ENV
- name: Run cargo clippy (debug)
run: cargo hack --features default --ignore-unknown-features --feature-powerset clippy $CLIPPY_COMMON_ARGS
- name: Check documentation generation
run: cargo doc --workspace --no-deps --document-private-items
env:
RUSTDOCFLAGS: "-Dwarnings -Arustdoc::private_intra_doc_links"
# Use `${{ !cancelled() }}` to run quck tests after the longer clippy run
- name: Check formatting
if: ${{ !cancelled() }}
run: cargo fmt --all -- --check
# https://github.com/facebookincubator/cargo-guppy/tree/bec4e0eb29dcd1faac70b1b5360267fc02bf830e/tools/cargo-hakari#2-keep-the-workspace-hack-up-to-date-in-ci
- name: Check rust dependencies
if: ${{ !cancelled() }}
run: |
cargo hakari generate --diff # workspace-hack Cargo.toml is up-to-date
cargo hakari manage-deps --dry-run # all workspace crates depend on workspace-hack
# https://github.com/EmbarkStudios/cargo-deny
- name: Check rust licenses/bans/advisories/sources
if: ${{ !cancelled() }}
run: cargo deny check --hide-inclusion-graph

View File

@@ -33,7 +33,7 @@ jobs:
# SC2086 - Double quote to prevent globbing and word splitting. - https://www.shellcheck.net/wiki/SC2086
SHELLCHECK_OPTS: --exclude=SC2046,SC2086
with:
fail_level: error
fail_on_error: true
filter_mode: nofilter
level: error

View File

@@ -308,7 +308,6 @@ jobs:
"image": [ "'"$image_default"'" ],
"include": [{ "pg_version": 16, "region_id": "'"$region_id_default"'", "platform": "neonvm-captest-freetier", "db_size": "3gb" ,"runner": '"$runner_default"', "image": "'"$image_default"'" },
{ "pg_version": 16, "region_id": "'"$region_id_default"'", "platform": "neonvm-captest-new", "db_size": "10gb","runner": '"$runner_default"', "image": "'"$image_default"'" },
{ "pg_version": 16, "region_id": "'"$region_id_default"'", "platform": "neonvm-captest-new-many-tables","db_size": "10gb","runner": '"$runner_default"', "image": "'"$image_default"'" },
{ "pg_version": 16, "region_id": "'"$region_id_default"'", "platform": "neonvm-captest-new", "db_size": "50gb","runner": '"$runner_default"', "image": "'"$image_default"'" },
{ "pg_version": 16, "region_id": "azure-eastus2", "platform": "neonvm-azure-captest-freetier", "db_size": "3gb" ,"runner": '"$runner_azure"', "image": "neondatabase/build-tools:pinned-bookworm" },
{ "pg_version": 16, "region_id": "azure-eastus2", "platform": "neonvm-azure-captest-new", "db_size": "10gb","runner": '"$runner_azure"', "image": "neondatabase/build-tools:pinned-bookworm" },
@@ -411,7 +410,7 @@ jobs:
aws-oicd-role-arn: ${{ vars.DEV_AWS_OIDC_ROLE_ARN }}
- name: Create Neon Project
if: contains(fromJson('["neonvm-captest-new", "neonvm-captest-new-many-tables", "neonvm-captest-freetier", "neonvm-azure-captest-freetier", "neonvm-azure-captest-new"]'), matrix.platform)
if: contains(fromJson('["neonvm-captest-new", "neonvm-captest-freetier", "neonvm-azure-captest-freetier", "neonvm-azure-captest-new"]'), matrix.platform)
id: create-neon-project
uses: ./.github/actions/neon-project-create
with:
@@ -430,7 +429,7 @@ jobs:
neonvm-captest-sharding-reuse)
CONNSTR=${{ secrets.BENCHMARK_CAPTEST_SHARDING_CONNSTR }}
;;
neonvm-captest-new | neonvm-captest-new-many-tables | neonvm-captest-freetier | neonvm-azure-captest-new | neonvm-azure-captest-freetier)
neonvm-captest-new | neonvm-captest-freetier | neonvm-azure-captest-new | neonvm-azure-captest-freetier)
CONNSTR=${{ steps.create-neon-project.outputs.dsn }}
;;
rds-aurora)
@@ -447,26 +446,6 @@ jobs:
echo "connstr=${CONNSTR}" >> $GITHUB_OUTPUT
# we want to compare Neon project OLTP throughput and latency at scale factor 10 GB
# without (neonvm-captest-new)
# and with (neonvm-captest-new-many-tables) many relations in the database
- name: Create many relations before the run
if: contains(fromJson('["neonvm-captest-new-many-tables"]'), matrix.platform)
uses: ./.github/actions/run-python-test-set
with:
build_type: ${{ env.BUILD_TYPE }}
test_selection: performance
run_in_parallel: false
save_perf_report: ${{ env.SAVE_PERF_REPORT }}
extra_params: -m remote_cluster --timeout 21600 -k test_perf_many_relations
pg_version: ${{ env.DEFAULT_PG_VERSION }}
aws-oicd-role-arn: ${{ vars.DEV_AWS_OIDC_ROLE_ARN }}
env:
BENCHMARK_CONNSTR: ${{ steps.set-up-connstr.outputs.connstr }}
VIP_VAP_ACCESS_TOKEN: "${{ secrets.VIP_VAP_ACCESS_TOKEN }}"
PERF_TEST_RESULT_CONNSTR: "${{ secrets.PERF_TEST_RESULT_CONNSTR }}"
TEST_NUM_RELATIONS: 10000
- name: Benchmark init
uses: ./.github/actions/run-python-test-set
with:

View File

@@ -1,241 +0,0 @@
name: Check neon with MacOS builds
on:
workflow_call:
inputs:
pg_versions:
description: "Array of the pg versions to build for, for example: ['v14', 'v17']"
type: string
default: '[]'
required: false
rebuild_rust_code:
description: "Rebuild Rust code"
type: boolean
default: false
required: false
rebuild_everything:
description: "If true, rebuild for all versions"
type: boolean
default: false
required: false
env:
RUST_BACKTRACE: 1
COPT: '-Werror'
# TODO: move `check-*` and `files-changed` jobs to the "Caller" Workflow
# We should care about that as Github has limitations:
# - You can connect up to four levels of workflows
# - You can call a maximum of 20 unique reusable workflows from a single workflow file.
# https://docs.github.com/en/actions/sharing-automations/reusing-workflows#limitations
jobs:
build-pgxn:
if: |
(inputs.pg_versions != '[]' || inputs.rebuild_everything) && (
contains(github.event.pull_request.labels.*.name, 'run-extra-build-macos') ||
contains(github.event.pull_request.labels.*.name, 'run-extra-build-*') ||
github.ref_name == 'main'
)
timeout-minutes: 30
runs-on: macos-15
strategy:
matrix:
postgres-version: ${{ inputs.rebuild_everything && fromJson('["v14", "v15", "v16", "v17"]') || fromJSON(inputs.pg_versions) }}
env:
# Use release build only, to have less debug info around
# Hence keeping target/ (and general cache size) smaller
BUILD_TYPE: release
steps:
- name: Checkout main repo
uses: actions/checkout@v4
- name: Set pg ${{ matrix.postgres-version }} for caching
id: pg_rev
run: echo pg_rev=$(git rev-parse HEAD:vendor/postgres-${{ matrix.postgres-version }}) | tee -a "${GITHUB_OUTPUT}"
- name: Cache postgres ${{ matrix.postgres-version }} build
id: cache_pg
uses: actions/cache@v4
with:
path: pg_install/${{ matrix.postgres-version }}
key: v1-${{ runner.os }}-${{ runner.arch }}-${{ env.BUILD_TYPE }}-pg-${{ matrix.postgres-version }}-${{ steps.pg_rev.outputs.pg_rev }}-${{ hashFiles('Makefile') }}
- name: Checkout submodule vendor/postgres-${{ matrix.postgres-version }}
if: steps.cache_pg.outputs.cache-hit != 'true'
run: |
git submodule init vendor/postgres-${{ matrix.postgres-version }}
git submodule update --depth 1 --recursive
- name: Install build dependencies
if: steps.cache_pg.outputs.cache-hit != 'true'
run: |
brew install flex bison openssl protobuf icu4c
- name: Set extra env for macOS
if: steps.cache_pg.outputs.cache-hit != 'true'
run: |
echo 'LDFLAGS=-L/usr/local/opt/openssl@3/lib' >> $GITHUB_ENV
echo 'CPPFLAGS=-I/usr/local/opt/openssl@3/include' >> $GITHUB_ENV
- name: Build Postgres ${{ matrix.postgres-version }}
if: steps.cache_pg.outputs.cache-hit != 'true'
run: |
make postgres-${{ matrix.postgres-version }} -j$(sysctl -n hw.ncpu)
- name: Build Neon Pg Ext ${{ matrix.postgres-version }}
if: steps.cache_pg.outputs.cache-hit != 'true'
run: |
make "neon-pg-ext-${{ matrix.postgres-version }}" -j$(sysctl -n hw.ncpu)
- name: Get postgres headers ${{ matrix.postgres-version }}
if: steps.cache_pg.outputs.cache-hit != 'true'
run: |
make postgres-headers-${{ matrix.postgres-version }} -j$(sysctl -n hw.ncpu)
build-walproposer-lib:
if: |
(inputs.pg_versions != '[]' || inputs.rebuild_everything) && (
contains(github.event.pull_request.labels.*.name, 'run-extra-build-macos') ||
contains(github.event.pull_request.labels.*.name, 'run-extra-build-*') ||
github.ref_name == 'main'
)
timeout-minutes: 30
runs-on: macos-15
needs: [build-pgxn]
env:
# Use release build only, to have less debug info around
# Hence keeping target/ (and general cache size) smaller
BUILD_TYPE: release
steps:
- name: Checkout main repo
uses: actions/checkout@v4
- name: Set pg v17 for caching
id: pg_rev
run: echo pg_rev=$(git rev-parse HEAD:vendor/postgres-v17) | tee -a "${GITHUB_OUTPUT}"
- name: Cache postgres v17 build
id: cache_pg
uses: actions/cache@v4
with:
path: pg_install/v17
key: v1-${{ runner.os }}-${{ runner.arch }}-${{ env.BUILD_TYPE }}-pg-v17-${{ steps.pg_rev.outputs.pg_rev }}-${{ hashFiles('Makefile') }}
- name: Cache walproposer-lib
id: cache_walproposer_lib
uses: actions/cache@v4
with:
path: pg_install/build/walproposer-lib
key: v1-${{ runner.os }}-${{ runner.arch }}-${{ env.BUILD_TYPE }}-walproposer_lib-v17-${{ steps.pg_rev.outputs.pg_rev }}-${{ hashFiles('Makefile') }}
- name: Checkout submodule vendor/postgres-v17
if: steps.cache_walproposer_lib.outputs.cache-hit != 'true'
run: |
git submodule init vendor/postgres-v17
git submodule update --depth 1 --recursive
- name: Install build dependencies
if: steps.cache_walproposer_lib.outputs.cache-hit != 'true'
run: |
brew install flex bison openssl protobuf icu4c
- name: Set extra env for macOS
if: steps.cache_walproposer_lib.outputs.cache-hit != 'true'
run: |
echo 'LDFLAGS=-L/usr/local/opt/openssl@3/lib' >> $GITHUB_ENV
echo 'CPPFLAGS=-I/usr/local/opt/openssl@3/include' >> $GITHUB_ENV
- name: Build walproposer-lib (only for v17)
if: steps.cache_walproposer_lib.outputs.cache-hit != 'true'
run:
make walproposer-lib -j$(sysctl -n hw.ncpu)
cargo-build:
if: |
(inputs.pg_versions != '[]' || inputs.rebuild_rust_code || inputs.rebuild_everything) && (
contains(github.event.pull_request.labels.*.name, 'run-extra-build-macos') ||
contains(github.event.pull_request.labels.*.name, 'run-extra-build-*') ||
github.ref_name == 'main'
)
timeout-minutes: 30
runs-on: macos-15
needs: [build-pgxn, build-walproposer-lib]
env:
# Use release build only, to have less debug info around
# Hence keeping target/ (and general cache size) smaller
BUILD_TYPE: release
steps:
- name: Checkout main repo
uses: actions/checkout@v4
with:
submodules: true
- name: Set pg v14 for caching
id: pg_rev_v14
run: echo pg_rev=$(git rev-parse HEAD:vendor/postgres-v14) | tee -a "${GITHUB_OUTPUT}"
- name: Set pg v15 for caching
id: pg_rev_v15
run: echo pg_rev=$(git rev-parse HEAD:vendor/postgres-v15) | tee -a "${GITHUB_OUTPUT}"
- name: Set pg v16 for caching
id: pg_rev_v16
run: echo pg_rev=$(git rev-parse HEAD:vendor/postgres-v16) | tee -a "${GITHUB_OUTPUT}"
- name: Set pg v17 for caching
id: pg_rev_v17
run: echo pg_rev=$(git rev-parse HEAD:vendor/postgres-v17) | tee -a "${GITHUB_OUTPUT}"
- name: Cache postgres v14 build
id: cache_pg
uses: actions/cache@v4
with:
path: pg_install/v14
key: v1-${{ runner.os }}-${{ runner.arch }}-${{ env.BUILD_TYPE }}-pg-v14-${{ steps.pg_rev_v14.outputs.pg_rev }}-${{ hashFiles('Makefile') }}
- name: Cache postgres v15 build
id: cache_pg_v15
uses: actions/cache@v4
with:
path: pg_install/v15
key: v1-${{ runner.os }}-${{ runner.arch }}-${{ env.BUILD_TYPE }}-pg-v15-${{ steps.pg_rev_v15.outputs.pg_rev }}-${{ hashFiles('Makefile') }}
- name: Cache postgres v16 build
id: cache_pg_v16
uses: actions/cache@v4
with:
path: pg_install/v16
key: v1-${{ runner.os }}-${{ runner.arch }}-${{ env.BUILD_TYPE }}-pg-v16-${{ steps.pg_rev_v16.outputs.pg_rev }}-${{ hashFiles('Makefile') }}
- name: Cache postgres v17 build
id: cache_pg_v17
uses: actions/cache@v4
with:
path: pg_install/v17
key: v1-${{ runner.os }}-${{ runner.arch }}-${{ env.BUILD_TYPE }}-pg-v17-${{ steps.pg_rev_v17.outputs.pg_rev }}-${{ hashFiles('Makefile') }}
- name: Cache cargo deps (only for v17)
uses: actions/cache@v4
with:
path: |
~/.cargo/registry
!~/.cargo/registry/src
~/.cargo/git
target
key: v1-${{ runner.os }}-${{ runner.arch }}-cargo-${{ hashFiles('./Cargo.lock') }}-${{ hashFiles('./rust-toolchain.toml') }}-rust
- name: Cache walproposer-lib
id: cache_walproposer_lib
uses: actions/cache@v4
with:
path: pg_install/build/walproposer-lib
key: v1-${{ runner.os }}-${{ runner.arch }}-${{ env.BUILD_TYPE }}-walproposer_lib-v17-${{ steps.pg_rev_v17.outputs.pg_rev }}-${{ hashFiles('Makefile') }}
- name: Install build dependencies
run: |
brew install flex bison openssl protobuf icu4c
- name: Set extra env for macOS
run: |
echo 'LDFLAGS=-L/usr/local/opt/openssl@3/lib' >> $GITHUB_ENV
echo 'CPPFLAGS=-I/usr/local/opt/openssl@3/include' >> $GITHUB_ENV
- name: Run cargo build (only for v17)
run: PQ_LIB_DIR=$(pwd)/pg_install/v17/lib cargo build --all --release -j$(sysctl -n hw.ncpu)
- name: Check that no warnings are produced (only for v17)
run: ./run_clippy.sh

View File

@@ -164,11 +164,77 @@ jobs:
check-codestyle-rust:
needs: [ check-permissions, build-build-tools-image ]
uses: ./.github/workflows/_check-codestyle-rust.yml
with:
build-tools-image: ${{ needs.build-build-tools-image.outputs.image }}-bookworm
archs: '["x64", "arm64"]'
secrets: inherit
strategy:
matrix:
arch: [ x64, arm64 ]
runs-on: ${{ fromJson(format('["self-hosted", "{0}"]', matrix.arch == 'arm64' && 'small-arm64' || 'small')) }}
container:
image: ${{ needs.build-build-tools-image.outputs.image }}-bookworm
credentials:
username: ${{ secrets.NEON_DOCKERHUB_USERNAME }}
password: ${{ secrets.NEON_DOCKERHUB_PASSWORD }}
options: --init
steps:
- name: Checkout
uses: actions/checkout@v4
with:
submodules: true
- name: Cache cargo deps
uses: actions/cache@v4
with:
path: |
~/.cargo/registry
!~/.cargo/registry/src
~/.cargo/git
target
key: v1-${{ runner.os }}-${{ runner.arch }}-cargo-${{ hashFiles('./Cargo.lock') }}-${{ hashFiles('./rust-toolchain.toml') }}-rust
# Some of our rust modules use FFI and need those to be checked
- name: Get postgres headers
run: make postgres-headers -j$(nproc)
# cargo hack runs the given cargo subcommand (clippy in this case) for all feature combinations.
# This will catch compiler & clippy warnings in all feature combinations.
# TODO: use cargo hack for build and test as well, but, that's quite expensive.
# NB: keep clippy args in sync with ./run_clippy.sh
#
# The only difference between "clippy --debug" and "clippy --release" is that in --release mode,
# #[cfg(debug_assertions)] blocks are not built. It's not worth building everything for second
# time just for that, so skip "clippy --release".
- run: |
CLIPPY_COMMON_ARGS="$( source .neon_clippy_args; echo "$CLIPPY_COMMON_ARGS")"
if [ "$CLIPPY_COMMON_ARGS" = "" ]; then
echo "No clippy args found in .neon_clippy_args"
exit 1
fi
echo "CLIPPY_COMMON_ARGS=${CLIPPY_COMMON_ARGS}" >> $GITHUB_ENV
- name: Run cargo clippy (debug)
run: cargo hack --feature-powerset clippy $CLIPPY_COMMON_ARGS
- name: Check documentation generation
run: cargo doc --workspace --no-deps --document-private-items
env:
RUSTDOCFLAGS: "-Dwarnings -Arustdoc::private_intra_doc_links"
# Use `${{ !cancelled() }}` to run quck tests after the longer clippy run
- name: Check formatting
if: ${{ !cancelled() }}
run: cargo fmt --all -- --check
# https://github.com/facebookincubator/cargo-guppy/tree/bec4e0eb29dcd1faac70b1b5360267fc02bf830e/tools/cargo-hakari#2-keep-the-workspace-hack-up-to-date-in-ci
- name: Check rust dependencies
if: ${{ !cancelled() }}
run: |
cargo hakari generate --diff # workspace-hack Cargo.toml is up-to-date
cargo hakari manage-deps --dry-run # all workspace crates depend on workspace-hack
# https://github.com/EmbarkStudios/cargo-deny
- name: Check rust licenses/bans/advisories/sources
if: ${{ !cancelled() }}
run: cargo deny check --hide-inclusion-graph
build-and-test-locally:
needs: [ tag, build-build-tools-image ]
@@ -280,22 +346,25 @@ jobs:
# XXX: no coverage data handling here, since benchmarks are run on release builds,
# while coverage is currently collected for the debug ones
report-benchmarks-results-to-slack:
report-benchmarks-failures:
needs: [ benchmarks, create-test-report ]
if: github.ref_name == 'main' && !cancelled() && contains(fromJSON('["success", "failure"]'), needs.benchmarks.result)
if: github.ref_name == 'main' && failure() && needs.benchmarks.result == 'failure'
permissions:
id-token: write # aws-actions/configure-aws-credentials
statuses: write
contents: write
pull-requests: write
runs-on: ubuntu-22.04
steps:
- uses: slackapi/slack-github-action@v2
- uses: slackapi/slack-github-action@v1
with:
method: chat.postMessage
token: ${{ secrets.SLACK_BOT_TOKEN }}
payload: |
channel: "${{ vars.SLACK_ON_CALL_STORAGE_STAGING_STREAM }}"
text: |
Benchmarks on main: *${{ needs.benchmarks.result }}*
- <${{ needs.create-test-report.outputs.report-url }}|Allure report>
- <${{ github.event.head_commit.url }}|${{ github.sha }}>
channel-id: C060CNA47S9 # on-call-staging-storage-stream
slack-message: |
Benchmarks failed on main <${{ github.event.head_commit.url }}|${{ github.sha }}>
<${{ needs.create-test-report.outputs.report-url }}|Allure report>
env:
SLACK_BOT_TOKEN: ${{ secrets.SLACK_BOT_TOKEN }}
create-test-report:
needs: [ check-permissions, build-and-test-locally, coverage-report, build-build-tools-image, benchmarks ]
@@ -469,7 +538,7 @@ jobs:
trigger-e2e-tests:
if: ${{ !github.event.pull_request.draft || contains( github.event.pull_request.labels.*.name, 'run-e2e-tests-in-draft') || github.ref_name == 'main' || github.ref_name == 'release' || github.ref_name == 'release-proxy' || github.ref_name == 'release-compute' }}
needs: [ check-permissions, promote-images-dev, tag ]
needs: [ check-permissions, promote-images, tag ]
uses: ./.github/workflows/trigger-e2e-tests.yml
secrets: inherit
@@ -659,6 +728,30 @@ jobs:
tags: |
neondatabase/neon-test-extensions-${{ matrix.version.pg }}:${{needs.tag.outputs.build-tag}}-${{ matrix.version.debian }}-${{ matrix.arch }}
- name: Build compute-tools image
# compute-tools are Postgres independent, so build it only once
# We pick 16, because that builds on debian 11 with older glibc (and is
# thus compatible with newer glibc), rather than 17 on Debian 12, as
# that isn't guaranteed to be compatible with Debian 11
if: matrix.version.pg == 'v16'
uses: docker/build-push-action@v6
with:
target: compute-tools-image
context: .
build-args: |
GIT_VERSION=${{ github.event.pull_request.head.sha || github.sha }}
BUILD_TAG=${{ needs.tag.outputs.build-tag }}
TAG=${{ needs.build-build-tools-image.outputs.image-tag }}-${{ matrix.version.debian }}
DEBIAN_VERSION=${{ matrix.version.debian }}
provenance: false
push: true
pull: true
file: compute/compute-node.Dockerfile
cache-from: type=registry,ref=cache.neon.build/compute-node-${{ matrix.version.pg }}:cache-${{ matrix.version.debian }}-${{ matrix.arch }}
cache-to: ${{ github.ref_name == 'main' && format('type=registry,ref=cache.neon.build/compute-tools-{0}:cache-{1}-{2},mode=max', matrix.version.pg, matrix.version.debian, matrix.arch) || '' }}
tags: |
neondatabase/compute-tools:${{ needs.tag.outputs.build-tag }}-${{ matrix.version.debian }}-${{ matrix.arch }}
compute-node-image:
needs: [ compute-node-image-arch, tag ]
permissions:
@@ -701,6 +794,14 @@ 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: Create multi-arch compute-tools image
if: matrix.version.pg == 'v16'
run: |
docker buildx imagetools create -t neondatabase/compute-tools:${{ needs.tag.outputs.build-tag }} \
-t neondatabase/compute-tools:${{ needs.tag.outputs.build-tag }}-${{ matrix.version.debian }} \
neondatabase/compute-tools:${{ needs.tag.outputs.build-tag }}-${{ matrix.version.debian }}-x64 \
neondatabase/compute-tools:${{ needs.tag.outputs.build-tag }}-${{ matrix.version.debian }}-arm64
- name: Configure AWS credentials
uses: aws-actions/configure-aws-credentials@v4
with:
@@ -716,6 +817,12 @@ jobs:
docker buildx imagetools create -t 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node-${{ matrix.version.pg }}:${{ needs.tag.outputs.build-tag }} \
neondatabase/compute-node-${{ matrix.version.pg }}:${{ needs.tag.outputs.build-tag }}
- name: Push multi-arch compute-tools image to ECR
if: matrix.version.pg == 'v16'
run: |
docker buildx imagetools create -t 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-tools:${{ needs.tag.outputs.build-tag }} \
neondatabase/compute-tools:${{ needs.tag.outputs.build-tag }}
vm-compute-node-image:
needs: [ check-permissions, tag, compute-node-image ]
runs-on: [ self-hosted, large ]
@@ -823,8 +930,8 @@ jobs:
docker compose -f ./docker-compose/docker-compose.yml logs || 0
docker compose -f ./docker-compose/docker-compose.yml down
promote-images-dev:
needs: [ check-permissions, tag, vm-compute-node-image, neon-image ]
promote-images:
needs: [ check-permissions, tag, test-images, vm-compute-node-image ]
runs-on: ubuntu-22.04
permissions:
@@ -858,35 +965,6 @@ jobs:
neondatabase/vm-compute-node-${version}:${{ needs.tag.outputs.build-tag }}
done
promote-images-prod:
needs: [ check-permissions, tag, test-images, vm-compute-node-image ]
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: |
@@ -894,6 +972,9 @@ jobs:
docker buildx imagetools create -t $repo/neon:latest \
$repo/neon:${{ needs.tag.outputs.build-tag }}
docker buildx imagetools create -t $repo/compute-tools:latest \
$repo/compute-tools:${{ needs.tag.outputs.build-tag }}
for version in ${VERSIONS}; do
docker buildx imagetools create -t $repo/compute-node-${version}:latest \
$repo/compute-node-${version}:${{ needs.tag.outputs.build-tag }}
@@ -922,31 +1003,31 @@ jobs:
- name: Copy all images to prod ECR
if: github.ref_name == 'release' || github.ref_name == 'release-proxy' || github.ref_name == 'release-compute'
run: |
for image in neon {vm-,}compute-node-{v14,v15,v16,v17}; do
for image in neon compute-tools {vm-,}compute-node-{v14,v15,v16,v17}; do
docker buildx imagetools create -t 093970136003.dkr.ecr.eu-central-1.amazonaws.com/${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 ]
needs: [ tag, promote-images ]
uses: ./.github/workflows/_push-to-acr.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
images: neon compute-tools 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 }}
push-to-acr-prod:
if: github.ref_name == 'release' || github.ref_name == 'release-proxy' || github.ref_name == 'release-compute'
needs: [ tag, promote-images-prod ]
needs: [ tag, promote-images ]
uses: ./.github/workflows/_push-to-acr.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
images: neon compute-tools 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 }}
@@ -1031,7 +1112,7 @@ 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, promote-images, tag, build-and-test-locally, trigger-custom-extensions-build-and-wait, push-to-acr-dev, push-to-acr-prod ]
# `!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:
@@ -1078,6 +1159,12 @@ jobs:
console.log(`Tag ${tag} created successfully.`);
}
// TODO: check how GitHub releases looks for proxy/compute releases and enable them if they're ok
if (context.ref !== 'refs/heads/release') {
console.log(`GitHub release skipped for ${context.ref}.`);
return;
}
try {
const existingRelease = await github.rest.repos.getReleaseByTag({
owner: context.repo.owner,
@@ -1096,8 +1183,7 @@ jobs:
owner: context.repo.owner,
repo: context.repo.repo,
tag_name: tag,
// TODO: Automate release notes properly
generate_release_notes: false,
generate_release_notes: true,
});
console.log(`Release for tag ${tag} created successfully.`);
}
@@ -1247,7 +1333,7 @@ jobs:
done
pin-build-tools-image:
needs: [ build-build-tools-image, promote-images-prod, build-and-test-locally ]
needs: [ build-build-tools-image, promote-images, build-and-test-locally ]
if: github.ref_name == 'main'
uses: ./.github/workflows/pin-build-tools-image.yml
with:
@@ -1270,7 +1356,7 @@ jobs:
- build-and-test-locally
- check-codestyle-python
- check-codestyle-rust
- promote-images-dev
- promote-images
- test-images
- trigger-custom-extensions-build-and-wait
runs-on: ubuntu-22.04

View File

@@ -21,8 +21,6 @@ concurrency:
permissions:
id-token: write # aws-actions/configure-aws-credentials
statuses: write
contents: write
jobs:
regress:

View File

@@ -31,15 +31,19 @@ jobs:
uses: ./.github/workflows/build-build-tools-image.yml
secrets: inherit
files-changed:
name: Detect what files changed
runs-on: ubuntu-22.04
timeout-minutes: 3
outputs:
v17: ${{ steps.files_changed.outputs.v17 }}
postgres_changes: ${{ steps.postgres_changes.outputs.changes }}
rebuild_rust_code: ${{ steps.files_changed.outputs.rust_code }}
rebuild_everything: ${{ steps.files_changed.outputs.rebuild_neon_extra || steps.files_changed.outputs.rebuild_macos }}
check-macos-build:
needs: [ check-permissions ]
if: |
contains(github.event.pull_request.labels.*.name, 'run-extra-build-macos') ||
contains(github.event.pull_request.labels.*.name, 'run-extra-build-*') ||
github.ref_name == 'main'
timeout-minutes: 90
runs-on: macos-15
env:
# Use release build only, to have less debug info around
# Hence keeping target/ (and general cache size) smaller
BUILD_TYPE: release
steps:
- name: Checkout
@@ -47,45 +51,106 @@ jobs:
with:
submodules: true
- name: Check for Postgres changes
uses: dorny/paths-filter@1441771bbfdd59dcd748680ee64ebd8faab1a242 #v3
id: files_changed
- name: Install macOS postgres dependencies
run: brew install flex bison openssl protobuf icu4c
- name: Set pg 14 revision for caching
id: pg_v14_rev
run: echo pg_rev=$(git rev-parse HEAD:vendor/postgres-v14) >> $GITHUB_OUTPUT
- name: Set pg 15 revision for caching
id: pg_v15_rev
run: echo pg_rev=$(git rev-parse HEAD:vendor/postgres-v15) >> $GITHUB_OUTPUT
- name: Set pg 16 revision for caching
id: pg_v16_rev
run: echo pg_rev=$(git rev-parse HEAD:vendor/postgres-v16) >> $GITHUB_OUTPUT
- name: Set pg 17 revision for caching
id: pg_v17_rev
run: echo pg_rev=$(git rev-parse HEAD:vendor/postgres-v17) >> $GITHUB_OUTPUT
- name: Cache postgres v14 build
id: cache_pg_14
uses: actions/cache@v4
with:
token: ${{ github.token }}
filters: .github/file-filters.yaml
base: ${{ github.event_name != 'pull_request' && (github.event.merge_group.base_ref || github.ref_name) || '' }}
ref: ${{ github.event_name != 'pull_request' && (github.event.merge_group.head_ref || github.ref) || '' }}
path: pg_install/v14
key: v1-${{ runner.os }}-${{ runner.arch }}-${{ env.BUILD_TYPE }}-pg-${{ steps.pg_v14_rev.outputs.pg_rev }}-${{ hashFiles('Makefile') }}
- name: Filter out only v-string for build matrix
id: postgres_changes
- name: Cache postgres v15 build
id: cache_pg_15
uses: actions/cache@v4
with:
path: pg_install/v15
key: v1-${{ runner.os }}-${{ runner.arch }}-${{ env.BUILD_TYPE }}-pg-${{ steps.pg_v15_rev.outputs.pg_rev }}-${{ hashFiles('Makefile') }}
- name: Cache postgres v16 build
id: cache_pg_16
uses: actions/cache@v4
with:
path: pg_install/v16
key: v1-${{ runner.os }}-${{ runner.arch }}-${{ env.BUILD_TYPE }}-pg-${{ steps.pg_v16_rev.outputs.pg_rev }}-${{ hashFiles('Makefile') }}
- name: Cache postgres v17 build
id: cache_pg_17
uses: actions/cache@v4
with:
path: pg_install/v17
key: v1-${{ runner.os }}-${{ runner.arch }}-${{ env.BUILD_TYPE }}-pg-${{ steps.pg_v17_rev.outputs.pg_rev }}-${{ hashFiles('Makefile') }}
- name: Set extra env for macOS
run: |
v_strings_only_as_json_array=$(echo ${{ steps.files_changed.outputs.chnages }} | jq '.[]|select(test("v\\d+"))' | jq --slurp -c)
echo "changes=${v_strings_only_as_json_array}" | tee -a "${GITHUB_OUTPUT}"
echo 'LDFLAGS=-L/usr/local/opt/openssl@3/lib' >> $GITHUB_ENV
echo 'CPPFLAGS=-I/usr/local/opt/openssl@3/include' >> $GITHUB_ENV
check-macos-build:
needs: [ check-permissions, files-changed ]
if: |
contains(github.event.pull_request.labels.*.name, 'run-extra-build-macos') ||
contains(github.event.pull_request.labels.*.name, 'run-extra-build-*') ||
github.ref_name == 'main'
uses: ./.github/workflows/build-macos.yml
with:
pg_versions: ${{ needs.files-changed.outputs.postgres_changes }}
rebuild_rust_code: ${{ needs.files-changed.outputs.rebuild_rust_code }}
rebuild_everything: ${{ fromJson(needs.files-changed.outputs.rebuild_everything) }}
- name: Cache cargo deps
uses: actions/cache@v4
with:
path: |
~/.cargo/registry
!~/.cargo/registry/src
~/.cargo/git
target
key: v1-${{ runner.os }}-${{ runner.arch }}-cargo-${{ hashFiles('./Cargo.lock') }}-${{ hashFiles('./rust-toolchain.toml') }}-rust
- name: Build postgres v14
if: steps.cache_pg_14.outputs.cache-hit != 'true'
run: make postgres-v14 -j$(sysctl -n hw.ncpu)
- name: Build postgres v15
if: steps.cache_pg_15.outputs.cache-hit != 'true'
run: make postgres-v15 -j$(sysctl -n hw.ncpu)
- name: Build postgres v16
if: steps.cache_pg_16.outputs.cache-hit != 'true'
run: make postgres-v16 -j$(sysctl -n hw.ncpu)
- name: Build postgres v17
if: steps.cache_pg_17.outputs.cache-hit != 'true'
run: make postgres-v17 -j$(sysctl -n hw.ncpu)
- name: Build neon extensions
run: make neon-pg-ext -j$(sysctl -n hw.ncpu)
- name: Build walproposer-lib
run: make walproposer-lib -j$(sysctl -n hw.ncpu)
- name: Run cargo build
run: PQ_LIB_DIR=$(pwd)/pg_install/v16/lib cargo build --all --release
- name: Check that no warnings are produced
run: ./run_clippy.sh
gather-rust-build-stats:
needs: [ check-permissions, build-build-tools-image, files-changed ]
needs: [ check-permissions, build-build-tools-image ]
permissions:
id-token: write # aws-actions/configure-aws-credentials
statuses: write
contents: write
if: |
(needs.files-changed.outputs.v17 == 'true' || needs.files-changed.outputs.rebuild_everything == 'true') && (
contains(github.event.pull_request.labels.*.name, 'run-extra-build-stats') ||
contains(github.event.pull_request.labels.*.name, 'run-extra-build-*') ||
github.ref_name == 'main'
)
contains(github.event.pull_request.labels.*.name, 'run-extra-build-stats') ||
contains(github.event.pull_request.labels.*.name, 'run-extra-build-*') ||
github.ref_name == 'main'
runs-on: [ self-hosted, large ]
container:
image: ${{ needs.build-build-tools-image.outputs.image }}-bookworm

View File

@@ -1,12 +1,6 @@
name: Pre-merge checks
on:
pull_request:
paths:
- .github/workflows/_check-codestyle-python.yml
- .github/workflows/_check-codestyle-rust.yml
- .github/workflows/build-build-tools-image.yml
- .github/workflows/pre-merge-checks.yml
merge_group:
branches:
- main
@@ -23,10 +17,8 @@ jobs:
runs-on: ubuntu-22.04
outputs:
python-changed: ${{ steps.python-src.outputs.any_changed }}
rust-changed: ${{ steps.rust-src.outputs.any_changed }}
steps:
- uses: actions/checkout@v4
- uses: tj-actions/changed-files@4edd678ac3f81e2dc578756871e4d00c19191daf # v45.0.4
id: python-src
with:
@@ -38,25 +30,11 @@ jobs:
poetry.lock
pyproject.toml
- uses: tj-actions/changed-files@4edd678ac3f81e2dc578756871e4d00c19191daf # v45.0.4
id: rust-src
with:
files: |
.github/workflows/_check-codestyle-rust.yml
.github/workflows/build-build-tools-image.yml
.github/workflows/pre-merge-checks.yml
**/**.rs
**/Cargo.toml
Cargo.toml
Cargo.lock
- name: PRINT ALL CHANGED FILES FOR DEBUG PURPOSES
env:
PYTHON_CHANGED_FILES: ${{ steps.python-src.outputs.all_changed_files }}
RUST_CHANGED_FILES: ${{ steps.rust-src.outputs.all_changed_files }}
run: |
echo "${PYTHON_CHANGED_FILES}"
echo "${RUST_CHANGED_FILES}"
build-build-tools-image:
if: needs.get-changed-files.outputs.python-changed == 'true'
@@ -77,16 +55,6 @@ jobs:
build-tools-image: ${{ needs.build-build-tools-image.outputs.image }}-bookworm-x64
secrets: inherit
check-codestyle-rust:
if: needs.get-changed-files.outputs.rust-changed == 'true'
needs: [ get-changed-files, build-build-tools-image ]
uses: ./.github/workflows/_check-codestyle-rust.yml
with:
# `-bookworm-x64` suffix should match the combination in `build-build-tools-image`
build-tools-image: ${{ needs.build-build-tools-image.outputs.image }}-bookworm-x64
archs: '["x64"]'
secrets: inherit
# To get items from the merge queue merged into main we need to satisfy "Status checks that are required".
# Currently we require 2 jobs (checks with exact name):
# - conclusion
@@ -99,7 +67,6 @@ jobs:
needs:
- get-changed-files
- check-codestyle-python
- check-codestyle-rust
runs-on: ubuntu-22.04
steps:
- name: Create fake `neon-cloud-e2e` check

View File

@@ -68,7 +68,7 @@ jobs:
GH_TOKEN: ${{ secrets.CI_ACCESS_TOKEN }}
TAG: ${{ needs.tag.outputs.build-tag }}
steps:
- name: Wait for `promote-images-dev` job to finish
- name: Wait for `promote-images` job to finish
# It's important to have a timeout here, the script in the step can run infinitely
timeout-minutes: 60
run: |
@@ -79,17 +79,17 @@ jobs:
# For PRs we use the run id as the tag
BUILD_AND_TEST_RUN_ID=${TAG}
while true; do
conclusion=$(gh run --repo ${GITHUB_REPOSITORY} view ${BUILD_AND_TEST_RUN_ID} --json jobs --jq '.jobs[] | select(.name == "promote-images-dev") | .conclusion')
conclusion=$(gh run --repo ${GITHUB_REPOSITORY} view ${BUILD_AND_TEST_RUN_ID} --json jobs --jq '.jobs[] | select(.name == "promote-images") | .conclusion')
case "$conclusion" in
success)
break
;;
failure | cancelled | skipped)
echo "The 'promote-images-dev' job didn't succeed: '${conclusion}'. Exiting..."
echo "The 'promote-images' job didn't succeed: '${conclusion}'. Exiting..."
exit 1
;;
*)
echo "The 'promote-images-dev' hasn't succeed yet. Waiting..."
echo "The 'promote-images' hasn't succeed yet. Waiting..."
sleep 60
;;
esac

420
Cargo.lock generated
View File

@@ -10,9 +10,9 @@ checksum = "8b5ace29ee3216de37c0546865ad08edef58b0f9e76838ed8959a84a990e58c5"
[[package]]
name = "addr2line"
version = "0.24.2"
version = "0.21.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "dfbe277e56a376000877090da837660b4427aad530e3028d44e0bffe4f89a1c1"
checksum = "8a30b2e23b9e17a9f90641c7ab1549cd9b44f296d3ccbf309d2863cfe398a0cb"
dependencies = [
"gimli",
]
@@ -23,12 +23,6 @@ version = "1.0.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f26201604c87b1e01bd3d98f8d5d9a8fcbb815e8cedb41ffccbeb4bf593a35fe"
[[package]]
name = "adler2"
version = "2.0.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "512761e0bb2578dd7380c6baaa0f4ce03e84f95e960231d1dec8bf4d7d6e2627"
[[package]]
name = "ahash"
version = "0.8.11"
@@ -718,13 +712,13 @@ dependencies = [
[[package]]
name = "axum"
version = "0.7.9"
version = "0.7.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "edca88bc138befd0323b20752846e6587272d3b03b0343c8ea28a6f819e6e71f"
checksum = "3a6c9af12842a67734c9a2e355436e5d03b22383ed60cf13cd0c18fbfe3dcbcf"
dependencies = [
"async-trait",
"axum-core",
"base64 0.22.1",
"base64 0.21.1",
"bytes",
"futures-util",
"http 1.1.0",
@@ -746,8 +740,8 @@ dependencies = [
"sha1",
"sync_wrapper 1.0.1",
"tokio",
"tokio-tungstenite 0.24.0",
"tower 0.5.2",
"tokio-tungstenite",
"tower",
"tower-layer",
"tower-service",
"tracing",
@@ -877,17 +871,17 @@ dependencies = [
[[package]]
name = "backtrace"
version = "0.3.74"
version = "0.3.69"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8d82cb332cdfaed17ae235a638438ac4d4839913cc2af585c3c6746e8f8bee1a"
checksum = "2089b7e3f35b9dd2d0ed921ead4f6d318c27680d4a5bd167b3ee120edb105837"
dependencies = [
"addr2line",
"cc",
"cfg-if",
"libc",
"miniz_oxide 0.8.0",
"miniz_oxide",
"object",
"rustc-demangle",
"windows-targets 0.52.6",
]
[[package]]
@@ -1133,7 +1127,7 @@ dependencies = [
"num-traits",
"serde",
"wasm-bindgen",
"windows-targets 0.52.6",
"windows-targets 0.52.4",
]
[[package]]
@@ -1267,7 +1261,6 @@ dependencies = [
"aws-config",
"aws-sdk-kms",
"aws-sdk-s3",
"axum",
"base64 0.13.1",
"bytes",
"camino",
@@ -1275,10 +1268,9 @@ dependencies = [
"chrono",
"clap",
"compute_api",
"fail",
"flate2",
"futures",
"http 1.1.0",
"hyper 0.14.30",
"metrics",
"nix 0.27.1",
"notify",
@@ -1304,15 +1296,12 @@ dependencies = [
"tokio-postgres",
"tokio-stream",
"tokio-util",
"tower 0.5.2",
"tower-http",
"tracing",
"tracing-opentelemetry",
"tracing-subscriber",
"tracing-utils",
"url",
"utils",
"uuid",
"vm_monitor",
"workspace_hack",
"zstd",
@@ -1606,32 +1595,6 @@ dependencies = [
"typenum",
]
[[package]]
name = "curve25519-dalek"
version = "4.1.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "97fb8b7c4503de7d6ae7b42ab72a5a59857b4c937ec27a3d4539dba95b5ab2be"
dependencies = [
"cfg-if",
"cpufeatures",
"curve25519-dalek-derive",
"digest",
"fiat-crypto",
"rustc_version",
"subtle",
]
[[package]]
name = "curve25519-dalek-derive"
version = "0.1.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f46882e17999c6cc590af592290432be3bce0428cb0d5f8b6715e4dc7b383eb3"
dependencies = [
"proc-macro2",
"quote",
"syn 2.0.90",
]
[[package]]
name = "darling"
version = "0.20.1"
@@ -1680,20 +1643,6 @@ dependencies = [
"parking_lot_core 0.9.8",
]
[[package]]
name = "dashmap"
version = "6.1.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "5041cc499144891f3790297212f32a74fb938e5136a14943f338ef9e0ae276cf"
dependencies = [
"cfg-if",
"crossbeam-utils",
"hashbrown 0.14.5",
"lock_api",
"once_cell",
"parking_lot_core 0.9.8",
]
[[package]]
name = "data-encoding"
version = "2.4.0"
@@ -1777,9 +1726,9 @@ checksum = "ab03c107fafeb3ee9f5925686dbb7a73bc76e3932abb0d2b365cb64b169cf04c"
[[package]]
name = "diesel"
version = "2.2.6"
version = "2.2.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ccf1bedf64cdb9643204a36dd15b19a6ce8e7aa7f7b105868e9f1fad5ffa7d12"
checksum = "65e13bab2796f412722112327f3e575601a3e9cdcbe426f0d30dbf43f3f5dc71"
dependencies = [
"bitflags 2.4.1",
"byteorder",
@@ -1902,28 +1851,6 @@ dependencies = [
"spki 0.7.3",
]
[[package]]
name = "ed25519"
version = "2.2.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "115531babc129696a58c64a4fef0a8bf9e9698629fb97e9e40767d235cfbcd53"
dependencies = [
"signature 2.2.0",
]
[[package]]
name = "ed25519-dalek"
version = "2.1.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4a3daa8e81a3963a60642bcc1f90a670680bd4a77535faa384e9d1c79d620871"
dependencies = [
"curve25519-dalek",
"ed25519",
"rand_core 0.6.4",
"sha2",
"subtle",
]
[[package]]
name = "either"
version = "1.8.1"
@@ -2015,15 +1942,6 @@ dependencies = [
"syn 2.0.90",
]
[[package]]
name = "env_filter"
version = "0.1.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "186e05a59d4c50738528153b83b0b0194d3a29507dfec16eccd4b342903397d0"
dependencies = [
"log",
]
[[package]]
name = "env_logger"
version = "0.10.2"
@@ -2037,16 +1955,6 @@ dependencies = [
"termcolor",
]
[[package]]
name = "env_logger"
version = "0.11.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "6c012a26a7f605efc424dd53697843a72be7dc86ad2d01f7814337794a12231d"
dependencies = [
"env_filter",
"log",
]
[[package]]
name = "equator"
version = "0.2.2"
@@ -2162,12 +2070,6 @@ dependencies = [
"subtle",
]
[[package]]
name = "fiat-crypto"
version = "0.2.9"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "28dea519a9695b9977216879a3ebfddf92f1c08c05d984f8996aecd6ecdc811d"
[[package]]
name = "filetime"
version = "0.2.22"
@@ -2205,7 +2107,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "3b9429470923de8e8cbd4d2dc513535400b4b3fef0319fb5c4e1f520a7bef743"
dependencies = [
"crc32fast",
"miniz_oxide 0.7.1",
"miniz_oxide",
]
[[package]]
@@ -2406,9 +2308,9 @@ dependencies = [
[[package]]
name = "gimli"
version = "0.31.1"
version = "0.28.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "07e28edb80900c19c28f1072f2e8aeca7fa06b23cd4169cefe1af5aa3260783f"
checksum = "4271d37baee1b8c7e4b708028c57d816cf9d2434acb33a549475f78c181f6253"
[[package]]
name = "git-version"
@@ -2811,7 +2713,7 @@ dependencies = [
"pin-project-lite",
"socket2",
"tokio",
"tower 0.4.13",
"tower",
"tower-service",
"tracing",
]
@@ -3036,28 +2938,6 @@ dependencies = [
"str_stack",
]
[[package]]
name = "inferno"
version = "0.12.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "75a5d75fee4d36809e6b021e4b96b686e763d365ffdb03af2bd00786353f84fe"
dependencies = [
"ahash",
"clap",
"crossbeam-channel",
"crossbeam-utils",
"dashmap 6.1.0",
"env_logger 0.11.2",
"indexmap 2.0.1",
"itoa",
"log",
"num-format",
"once_cell",
"quick-xml 0.37.1",
"rgb",
"str_stack",
]
[[package]]
name = "inotify"
version = "0.9.6"
@@ -3265,7 +3145,7 @@ version = "0.7.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4644821e1c3d7a560fe13d842d13f587c07348a1a05d3a797152d41c90c56df2"
dependencies = [
"dashmap 5.5.0",
"dashmap",
"hashbrown 0.13.2",
]
@@ -3373,9 +3253,9 @@ checksum = "b87248edafb776e59e6ee64a79086f65890d3510f2c656c000bf2a7e8a0aea40"
[[package]]
name = "matchit"
version = "0.8.4"
version = "0.8.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "47e1ffaa40ddd1f3ed91f717a33c8c0ee23fff369e3aa8772b9605cc1d22f4c3"
checksum = "540f1c43aed89909c0cc0cc604e3bb2f7e7a341a3728a9e6cfe760e733cd11ed"
[[package]]
name = "md-5"
@@ -3524,15 +3404,6 @@ dependencies = [
"adler",
]
[[package]]
name = "miniz_oxide"
version = "0.8.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e2d80299ef12ff69b16a84bb182e3b9df68b5a91574d3d4fa6e41b65deec4df1"
dependencies = [
"adler2",
]
[[package]]
name = "mio"
version = "0.8.11"
@@ -3767,9 +3638,9 @@ dependencies = [
[[package]]
name = "object"
version = "0.36.5"
version = "0.32.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "aedf0a2d09c573ed1d8d85b30c119153926a2b36dce0ab28322c09a117a4683e"
checksum = "a6a622008b6e321afc04970976f62ee297fdbaa6f95318ca343e3eebb9648441"
dependencies = [
"memchr",
]
@@ -3803,23 +3674,23 @@ checksum = "ff011a302c396a5197692431fc1948019154afc178baf7d8e37367442a4601cf"
[[package]]
name = "opentelemetry"
version = "0.27.1"
version = "0.26.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ab70038c28ed37b97d8ed414b6429d343a8bbf44c9f79ec854f3a643029ba6d7"
checksum = "570074cc999d1a58184080966e5bd3bf3a9a4af650c3b05047c2621e7405cd17"
dependencies = [
"futures-core",
"futures-sink",
"js-sys",
"once_cell",
"pin-project-lite",
"thiserror",
"tracing",
]
[[package]]
name = "opentelemetry-http"
version = "0.27.0"
version = "0.26.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "10a8a7f5f6ba7c1b286c2fbca0454eaba116f63bbe69ed250b642d36fbb04d80"
checksum = "6351496aeaa49d7c267fb480678d85d1cd30c5edb20b497c48c56f62a8c14b99"
dependencies = [
"async-trait",
"bytes",
@@ -3830,9 +3701,9 @@ dependencies = [
[[package]]
name = "opentelemetry-otlp"
version = "0.27.0"
version = "0.26.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "91cf61a1868dacc576bf2b2a1c3e9ab150af7272909e80085c3173384fe11f76"
checksum = "29e1f9c8b032d4f635c730c0efcf731d5e2530ea13fa8bef7939ddc8420696bd"
dependencies = [
"async-trait",
"futures-core",
@@ -3848,9 +3719,9 @@ dependencies = [
[[package]]
name = "opentelemetry-proto"
version = "0.27.0"
version = "0.26.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a6e05acbfada5ec79023c85368af14abd0b307c015e9064d249b2a950ef459a6"
checksum = "c9d3968ce3aefdcca5c27e3c4ea4391b37547726a70893aab52d3de95d5f8b34"
dependencies = [
"opentelemetry",
"opentelemetry_sdk",
@@ -3860,21 +3731,22 @@ dependencies = [
[[package]]
name = "opentelemetry-semantic-conventions"
version = "0.27.0"
version = "0.26.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "bc1b6902ff63b32ef6c489e8048c5e253e2e4a803ea3ea7e783914536eb15c52"
checksum = "db945c1eaea8ac6a9677185357480d215bb6999faa9f691d0c4d4d641eab7a09"
[[package]]
name = "opentelemetry_sdk"
version = "0.27.1"
version = "0.26.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "231e9d6ceef9b0b2546ddf52335785ce41252bc7474ee8ba05bfad277be13ab8"
checksum = "d2c627d9f4c9cdc1f21a29ee4bfbd6028fcb8bcf2a857b43f3abdf72c9c862f3"
dependencies = [
"async-trait",
"futures-channel",
"futures-executor",
"futures-util",
"glob",
"once_cell",
"opentelemetry",
"percent-encoding",
"rand 0.8.5",
@@ -3882,7 +3754,6 @@ dependencies = [
"thiserror",
"tokio",
"tokio-stream",
"tracing",
]
[[package]]
@@ -3982,11 +3853,9 @@ name = "pagectl"
version = "0.1.0"
dependencies = [
"anyhow",
"bincode",
"camino",
"clap",
"humantime",
"itertools 0.10.5",
"pageserver",
"pageserver_api",
"postgres_ffi",
@@ -4008,7 +3877,6 @@ dependencies = [
"arc-swap",
"async-compression",
"async-stream",
"bincode",
"bit_field",
"byteorder",
"bytes",
@@ -4048,7 +3916,6 @@ dependencies = [
"postgres_connection",
"postgres_ffi",
"postgres_initdb",
"pprof",
"pq_proto",
"procfs",
"rand 0.8.5",
@@ -4534,13 +4401,11 @@ dependencies = [
"bindgen",
"bytes",
"crc32c",
"criterion",
"env_logger 0.10.2",
"env_logger",
"log",
"memoffset 0.9.0",
"once_cell",
"postgres",
"pprof",
"regex",
"serde",
"thiserror",
@@ -4576,7 +4441,7 @@ dependencies = [
"cfg-if",
"criterion",
"findshlibs",
"inferno 0.11.21",
"inferno",
"libc",
"log",
"nix 0.26.4",
@@ -4611,9 +4476,9 @@ checksum = "5b40af805b3121feab8a3c29f04d8ad262fa8e0561883e7653e024ae4479e6de"
[[package]]
name = "pq-sys"
version = "0.6.3"
version = "0.4.8"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f6cc05d7ea95200187117196eee9edd0644424911821aeb28a18ce60ea0b8793"
checksum = "31c0052426df997c0cbd30789eb44ca097e3541717a7b8fa36b1c464ee7edebd"
dependencies = [
"vcpkg",
]
@@ -4802,10 +4667,9 @@ dependencies = [
"clap",
"compute_api",
"consumption_metrics",
"dashmap 5.5.0",
"dashmap",
"ecdsa 0.16.9",
"ed25519-dalek",
"env_logger 0.10.2",
"env_logger",
"fallible-iterator",
"flate2",
"framed-websockets",
@@ -4876,7 +4740,7 @@ dependencies = [
"tokio-postgres",
"tokio-postgres2",
"tokio-rustls 0.26.0",
"tokio-tungstenite 0.21.0",
"tokio-tungstenite",
"tokio-util",
"tracing",
"tracing-subscriber",
@@ -4912,15 +4776,6 @@ dependencies = [
"serde",
]
[[package]]
name = "quick-xml"
version = "0.37.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f22f29bdff3987b4d8632ef95fd6424ec7e4e0a57e2f4fc63e489e75357f6a03"
dependencies = [
"memchr",
]
[[package]]
name = "quote"
version = "1.0.37"
@@ -5207,7 +5062,6 @@ dependencies = [
"once_cell",
"pin-project-lite",
"rand 0.8.5",
"reqwest",
"scopeguard",
"serde",
"serde_json",
@@ -5305,15 +5159,15 @@ dependencies = [
[[package]]
name = "reqwest-tracing"
version = "0.5.5"
version = "0.5.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "73e6153390585f6961341b50e5a1931d6be6dee4292283635903c26ef9d980d2"
checksum = "ff82cf5730a1311fb9413b0bc2b8e743e0157cd73f010ab4ec374a923873b6a2"
dependencies = [
"anyhow",
"async-trait",
"getrandom 0.2.11",
"http 1.1.0",
"matchit 0.8.4",
"matchit 0.8.2",
"opentelemetry",
"reqwest",
"reqwest-middleware",
@@ -5466,9 +5320,9 @@ dependencies = [
[[package]]
name = "rustc-demangle"
version = "0.1.24"
version = "0.1.23"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "719b953e2095829ee67db738b3bfa9fa368c94900df327b3f07fe6e794d2fe1f"
checksum = "d626bb9dae77e28219937af045c257c28bfd3f69333c512553507f5f9798cb76"
[[package]]
name = "rustc-hash"
@@ -5659,7 +5513,6 @@ dependencies = [
"crc32c",
"criterion",
"desim",
"env_logger 0.10.2",
"fail",
"futures",
"hex",
@@ -5682,13 +5535,11 @@ dependencies = [
"remote_storage",
"reqwest",
"safekeeper_api",
"safekeeper_client",
"scopeguard",
"sd-notify",
"serde",
"serde_json",
"sha2",
"smallvec",
"storage_broker",
"strum",
"strum_macros",
@@ -5713,29 +5564,14 @@ dependencies = [
name = "safekeeper_api"
version = "0.1.0"
dependencies = [
"anyhow",
"const_format",
"pageserver_api",
"postgres_ffi",
"pq_proto",
"serde",
"serde_json",
"tokio",
"utils",
]
[[package]]
name = "safekeeper_client"
version = "0.1.0"
dependencies = [
"reqwest",
"safekeeper_api",
"serde",
"thiserror",
"utils",
"workspace_hack",
]
[[package]]
name = "same-file"
version = "1.0.6"
@@ -6932,19 +6768,7 @@ dependencies = [
"futures-util",
"log",
"tokio",
"tungstenite 0.21.0",
]
[[package]]
name = "tokio-tungstenite"
version = "0.24.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "edc5f74e248dc973e0dbb7b74c7e0d6fcc301c694ff50049504004ef4d0cdcd9"
dependencies = [
"futures-util",
"log",
"tokio",
"tungstenite 0.24.0",
"tungstenite",
]
[[package]]
@@ -7025,7 +6849,7 @@ dependencies = [
"tokio",
"tokio-rustls 0.26.0",
"tokio-stream",
"tower 0.4.13",
"tower",
"tower-layer",
"tower-service",
"tracing",
@@ -7065,50 +6889,17 @@ dependencies = [
"tracing",
]
[[package]]
name = "tower"
version = "0.5.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d039ad9159c98b70ecfd540b2573b97f7f52c3e8d9f8ad57a24b916a536975f9"
dependencies = [
"futures-core",
"futures-util",
"pin-project-lite",
"sync_wrapper 1.0.1",
"tokio",
"tower-layer",
"tower-service",
"tracing",
]
[[package]]
name = "tower-http"
version = "0.6.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "403fa3b783d4b626a8ad51d766ab03cb6d2dbfc46b1c5d4448395e6628dc9697"
dependencies = [
"bitflags 2.4.1",
"bytes",
"http 1.1.0",
"http-body 1.0.0",
"pin-project-lite",
"tower-layer",
"tower-service",
"tracing",
"uuid",
]
[[package]]
name = "tower-layer"
version = "0.3.3"
version = "0.3.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "121c2a6cda46980bb0fcd1647ffaf6cd3fc79a013de288782836f6df9c48780e"
checksum = "c20c8dbed6283a09604c3e69b4b7eeb54e298b8a600d4d5ecb5ad39de609f1d0"
[[package]]
name = "tower-service"
version = "0.3.3"
version = "0.3.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8df9b6e13f2d32c91b9bd719c00d1958837bc7dec474d94952798cc8e69eeec3"
checksum = "b6bc1c9ce2b5135ac7f93c72918fc37feb872bdc6a5533a8b85eb4b86bfdae52"
[[package]]
name = "tracing"
@@ -7177,9 +6968,9 @@ dependencies = [
[[package]]
name = "tracing-opentelemetry"
version = "0.28.0"
version = "0.27.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "97a971f6058498b5c0f1affa23e7ea202057a7301dbff68e968b2d578bcbd053"
checksum = "dc58af5d3f6c5811462cabb3289aec0093f7338e367e5a33d28c0433b3c7360b"
dependencies = [
"js-sys",
"once_cell",
@@ -7263,24 +7054,6 @@ dependencies = [
"utf-8",
]
[[package]]
name = "tungstenite"
version = "0.24.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "18e5b8366ee7a95b16d32197d0b2604b43a0be89dc5fac9f8e96ccafbaedda8a"
dependencies = [
"byteorder",
"bytes",
"data-encoding",
"http 1.1.0",
"httparse",
"log",
"rand 0.8.5",
"sha1",
"thiserror",
"utf-8",
]
[[package]]
name = "twox-hash"
version = "1.6.3"
@@ -7430,7 +7203,6 @@ dependencies = [
"anyhow",
"arc-swap",
"async-compression",
"backtrace",
"bincode",
"byteorder",
"bytes",
@@ -7441,15 +7213,12 @@ 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",
@@ -7552,7 +7321,7 @@ dependencies = [
"anyhow",
"camino-tempfile",
"clap",
"env_logger 0.10.2",
"env_logger",
"log",
"postgres",
"postgres_ffi",
@@ -7567,21 +7336,12 @@ dependencies = [
"anyhow",
"async-compression",
"bytes",
"camino",
"camino-tempfile",
"criterion",
"futures",
"pageserver_api",
"postgres_ffi",
"pprof",
"prost",
"remote_storage",
"serde",
"serde_json",
"thiserror",
"tikv-jemallocator",
"tokio",
"tokio-util",
"tonic",
"tonic-build",
"tracing",
@@ -7815,7 +7575,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e48a53791691ab099e5e2ad123536d0fff50652600abaf43bbf952894110d0be"
dependencies = [
"windows-core",
"windows-targets 0.52.6",
"windows-targets 0.52.4",
]
[[package]]
@@ -7824,7 +7584,7 @@ version = "0.52.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "33ab640c8d7e35bf8ba19b884ba838ceb4fba93a4e8c65a9059d08afcfc683d9"
dependencies = [
"windows-targets 0.52.6",
"windows-targets 0.52.4",
]
[[package]]
@@ -7842,7 +7602,7 @@ version = "0.52.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "282be5f36a8ce781fad8c8ae18fa3f9beff57ec1b52cb3de0789201425d9a33d"
dependencies = [
"windows-targets 0.52.6",
"windows-targets 0.52.4",
]
[[package]]
@@ -7862,18 +7622,17 @@ dependencies = [
[[package]]
name = "windows-targets"
version = "0.52.6"
version = "0.52.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "9b724f72796e036ab90c1021d4780d4d3d648aca59e491e6b98e725b84e99973"
checksum = "7dd37b7e5ab9018759f893a1952c9420d060016fc19a472b4bb20d1bdd694d1b"
dependencies = [
"windows_aarch64_gnullvm 0.52.6",
"windows_aarch64_msvc 0.52.6",
"windows_i686_gnu 0.52.6",
"windows_i686_gnullvm",
"windows_i686_msvc 0.52.6",
"windows_x86_64_gnu 0.52.6",
"windows_x86_64_gnullvm 0.52.6",
"windows_x86_64_msvc 0.52.6",
"windows_aarch64_gnullvm 0.52.4",
"windows_aarch64_msvc 0.52.4",
"windows_i686_gnu 0.52.4",
"windows_i686_msvc 0.52.4",
"windows_x86_64_gnu 0.52.4",
"windows_x86_64_gnullvm 0.52.4",
"windows_x86_64_msvc 0.52.4",
]
[[package]]
@@ -7884,9 +7643,9 @@ checksum = "91ae572e1b79dba883e0d315474df7305d12f569b400fcf90581b06062f7e1bc"
[[package]]
name = "windows_aarch64_gnullvm"
version = "0.52.6"
version = "0.52.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "32a4622180e7a0ec044bb555404c800bc9fd9ec262ec147edd5989ccd0c02cd3"
checksum = "bcf46cf4c365c6f2d1cc93ce535f2c8b244591df96ceee75d8e83deb70a9cac9"
[[package]]
name = "windows_aarch64_msvc"
@@ -7896,9 +7655,9 @@ checksum = "b2ef27e0d7bdfcfc7b868b317c1d32c641a6fe4629c171b8928c7b08d98d7cf3"
[[package]]
name = "windows_aarch64_msvc"
version = "0.52.6"
version = "0.52.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "09ec2a7bb152e2252b53fa7803150007879548bc709c039df7627cabbd05d469"
checksum = "da9f259dd3bcf6990b55bffd094c4f7235817ba4ceebde8e6d11cd0c5633b675"
[[package]]
name = "windows_i686_gnu"
@@ -7908,15 +7667,9 @@ checksum = "622a1962a7db830d6fd0a69683c80a18fda201879f0f447f065a3b7467daa241"
[[package]]
name = "windows_i686_gnu"
version = "0.52.6"
version = "0.52.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8e9b5ad5ab802e97eb8e295ac6720e509ee4c243f69d781394014ebfe8bbfa0b"
[[package]]
name = "windows_i686_gnullvm"
version = "0.52.6"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "0eee52d38c090b3caa76c563b86c3a4bd71ef1a819287c19d586d7334ae8ed66"
checksum = "b474d8268f99e0995f25b9f095bc7434632601028cf86590aea5c8a5cb7801d3"
[[package]]
name = "windows_i686_msvc"
@@ -7926,9 +7679,9 @@ checksum = "4542c6e364ce21bf45d69fdd2a8e455fa38d316158cfd43b3ac1c5b1b19f8e00"
[[package]]
name = "windows_i686_msvc"
version = "0.52.6"
version = "0.52.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "240948bc05c5e7c6dabba28bf89d89ffce3e303022809e73deaefe4f6ec56c66"
checksum = "1515e9a29e5bed743cb4415a9ecf5dfca648ce85ee42e15873c3cd8610ff8e02"
[[package]]
name = "windows_x86_64_gnu"
@@ -7938,9 +7691,9 @@ checksum = "ca2b8a661f7628cbd23440e50b05d705db3686f894fc9580820623656af974b1"
[[package]]
name = "windows_x86_64_gnu"
version = "0.52.6"
version = "0.52.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "147a5c80aabfbf0c7d901cb5895d1de30ef2907eb21fbbab29ca94c5b08b1a78"
checksum = "5eee091590e89cc02ad514ffe3ead9eb6b660aedca2183455434b93546371a03"
[[package]]
name = "windows_x86_64_gnullvm"
@@ -7950,9 +7703,9 @@ checksum = "7896dbc1f41e08872e9d5e8f8baa8fdd2677f29468c4e156210174edc7f7b953"
[[package]]
name = "windows_x86_64_gnullvm"
version = "0.52.6"
version = "0.52.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "24d5b23dc417412679681396f2b49f3de8c1473deb516bd34410872eff51ed0d"
checksum = "77ca79f2451b49fa9e2af39f0747fe999fcda4f5e241b2898624dca97a1f2177"
[[package]]
name = "windows_x86_64_msvc"
@@ -7962,9 +7715,9 @@ checksum = "1a515f5799fe4961cb532f983ce2b23082366b898e52ffbce459c86f67c8378a"
[[package]]
name = "windows_x86_64_msvc"
version = "0.52.6"
version = "0.52.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "589f6da84c646204747d1270a2a5661ea66ed1cced2631d546fdfb155959f9ec"
checksum = "32b752e52a2da0ddfbdbcc6fceadfeede4c939ed16d13e648833a61dfb611ed8"
[[package]]
name = "winnow"
@@ -8072,8 +7825,7 @@ dependencies = [
"tokio-util",
"toml_edit",
"tonic",
"tower 0.4.13",
"tower 0.5.2",
"tower",
"tracing",
"tracing-core",
"url",

View File

@@ -11,7 +11,6 @@ members = [
"pageserver/pagebench",
"proxy",
"safekeeper",
"safekeeper/client",
"storage_broker",
"storage_controller",
"storage_controller/client",
@@ -52,7 +51,6 @@ anyhow = { version = "1.0", features = ["backtrace"] }
arc-swap = "1.6"
async-compression = { version = "0.4.0", features = ["tokio", "gzip", "zstd"] }
atomic-take = "1.1.0"
backtrace = "0.3.74"
flate2 = "1.0.26"
async-stream = "0.3"
async-trait = "0.1"
@@ -65,7 +63,7 @@ aws-smithy-types = "1.2"
aws-credential-types = "1.2.0"
aws-sigv4 = { version = "1.2", features = ["sign-http"] }
aws-types = "1.3"
axum = { version = "0.7.9", features = ["ws"] }
axum = { version = "0.7.5", features = ["ws"] }
base64 = "0.13.0"
bincode = "1.3"
bindgen = "0.70"
@@ -110,7 +108,6 @@ hyper-util = "0.1"
tokio-tungstenite = "0.21.0"
indexmap = "2"
indoc = "2"
inferno = "0.12.0"
ipnet = "2.10.0"
itertools = "0.10"
itoa = "1.0.11"
@@ -127,16 +124,16 @@ notify = "6.0.0"
num_cpus = "1.15"
num-traits = "0.2.15"
once_cell = "1.13"
opentelemetry = "0.27"
opentelemetry_sdk = "0.27"
opentelemetry-otlp = { version = "0.27", default-features = false, features = ["http-proto", "trace", "http", "reqwest-client"] }
opentelemetry-semantic-conventions = "0.27"
opentelemetry = "0.26"
opentelemetry_sdk = "0.26"
opentelemetry-otlp = { version = "0.26", default-features=false, features = ["http-proto", "trace", "http", "reqwest-client"] }
opentelemetry-semantic-conventions = "0.26"
parking_lot = "0.12"
parquet = { version = "53", default-features = false, features = ["zstd"] }
parquet_derive = "53"
pbkdf2 = { version = "0.12.1", features = ["simple", "std"] }
pin-project-lite = "0.2"
pprof = { version = "0.14", features = ["criterion", "flamegraph", "frame-pointer", "protobuf", "protobuf-codec"] }
pprof = { version = "0.14", features = ["criterion", "flamegraph", "protobuf", "protobuf-codec"] }
procfs = "0.16"
prometheus = {version = "0.13", default-features=false, features = ["process"]} # removes protobuf dependency
prost = "0.13"
@@ -144,7 +141,7 @@ rand = "0.8"
redis = { version = "0.25.2", features = ["tokio-rustls-comp", "keep-alive"] }
regex = "1.10.2"
reqwest = { version = "0.12", default-features = false, features = ["rustls-tls"] }
reqwest-tracing = { version = "0.5", features = ["opentelemetry_0_27"] }
reqwest-tracing = { version = "0.5", features = ["opentelemetry_0_26"] }
reqwest-middleware = "0.4"
reqwest-retry = "0.7"
routerify = "3"
@@ -188,12 +185,10 @@ tokio-util = { version = "0.7.10", features = ["io", "rt"] }
toml = "0.8"
toml_edit = "0.22"
tonic = {version = "0.12.3", features = ["tls", "tls-roots"]}
tower = { version = "0.5.2", default-features = false }
tower-http = { version = "0.6.2", features = ["request-id", "trace"] }
tower-service = "0.3.3"
tower-service = "0.3.2"
tracing = "0.1"
tracing-error = "0.2"
tracing-opentelemetry = "0.28"
tracing-opentelemetry = "0.27"
tracing-subscriber = { version = "0.3", default-features = false, features = ["smallvec", "fmt", "tracing-log", "std", "env-filter", "json"] }
try-lock = "0.2.5"
twox-hash = { version = "1.6.3", default-features = false }
@@ -238,7 +233,6 @@ postgres_initdb = { path = "./libs/postgres_initdb" }
pq_proto = { version = "0.1", path = "./libs/pq_proto/" }
remote_storage = { version = "0.1", path = "./libs/remote_storage/" }
safekeeper_api = { version = "0.1", path = "./libs/safekeeper_api" }
safekeeper_client = { path = "./safekeeper/client" }
desim = { version = "0.1", path = "./libs/desim" }
storage_broker = { version = "0.1", path = "./storage_broker/" } # Note: main broker code is inside the binary crate, so linking with the library shouldn't be heavy.
storage_controller_client = { path = "./storage_controller/client" }
@@ -269,8 +263,6 @@ tokio-postgres = { git = "https://github.com/neondatabase/rust-postgres.git", br
[profile.release]
# This is useful for profiling and, to some extent, debug.
# Besides, debug info should not affect the performance.
#
# NB: we also enable frame pointers for improved profiling, see .cargo/config.toml.
debug = true
# disable debug symbols for all packages except this one to decrease binaries size

View File

@@ -45,7 +45,7 @@ COPY --chown=nonroot . .
ARG ADDITIONAL_RUSTFLAGS
RUN set -e \
&& PQ_LIB_DIR=$(pwd)/pg_install/v${STABLE_PG_VERSION}/lib RUSTFLAGS="-Clinker=clang -Clink-arg=-fuse-ld=mold -Clink-arg=-Wl,--no-rosegment -Cforce-frame-pointers=yes ${ADDITIONAL_RUSTFLAGS}" cargo build \
&& PQ_LIB_DIR=$(pwd)/pg_install/v${STABLE_PG_VERSION}/lib RUSTFLAGS="-Clinker=clang -Clink-arg=-fuse-ld=mold -Clink-arg=-Wl,--no-rosegment ${ADDITIONAL_RUSTFLAGS}" cargo build \
--bin pg_sni_router \
--bin pageserver \
--bin pagectl \
@@ -69,9 +69,6 @@ RUN set -e \
libreadline-dev \
libseccomp-dev \
ca-certificates \
# System postgres for use with client libraries (e.g. in storage controller)
postgresql-15 \
openssl \
&& rm -rf /var/lib/apt/lists/* /tmp/* /var/tmp/* \
&& useradd -d /data neon \
&& chown -R neon:neon /data
@@ -104,6 +101,11 @@ RUN mkdir -p /data/.neon/ && \
> /data/.neon/pageserver.toml && \
chown -R neon:neon /data/.neon
# When running a binary that links with libpq, default to using our most recent postgres version. Binaries
# that want a particular postgres version will select it explicitly: this is just a default.
ENV LD_LIBRARY_PATH=/usr/local/v${DEFAULT_PG_VERSION}/lib
VOLUME ["/data"]
USER neon
EXPOSE 6400

View File

@@ -3,6 +3,7 @@ ROOT_PROJECT_DIR := $(dir $(abspath $(lastword $(MAKEFILE_LIST))))
# Where to install Postgres, default is ./pg_install, maybe useful for package managers
POSTGRES_INSTALL_DIR ?= $(ROOT_PROJECT_DIR)/pg_install/
OPENSSL_PREFIX_DIR := /usr/local/openssl
ICU_PREFIX_DIR := /usr/local/icu
#
@@ -25,9 +26,11 @@ endif
ifeq ($(shell test -e /home/nonroot/.docker_build && echo -n yes),yes)
# Exclude static build openssl, icu for local build (MacOS, Linux)
# Only keep for build type release and debug
PG_CFLAGS += -I$(OPENSSL_PREFIX_DIR)/include
PG_CONFIGURE_OPTS += --with-icu
PG_CONFIGURE_OPTS += ICU_CFLAGS='-I/$(ICU_PREFIX_DIR)/include -DU_STATIC_IMPLEMENTATION'
PG_CONFIGURE_OPTS += ICU_LIBS='-L$(ICU_PREFIX_DIR)/lib -L$(ICU_PREFIX_DIR)/lib64 -licui18n -licuuc -licudata -lstdc++ -Wl,-Bdynamic -lm'
PG_CONFIGURE_OPTS += LDFLAGS='-L$(OPENSSL_PREFIX_DIR)/lib -L$(OPENSSL_PREFIX_DIR)/lib64 -L$(ICU_PREFIX_DIR)/lib -L$(ICU_PREFIX_DIR)/lib64 -Wl,-Bstatic -lssl -lcrypto -Wl,-Bdynamic -lrt -lm -ldl -lpthread'
endif
UNAME_S := $(shell uname -s)

View File

@@ -115,7 +115,7 @@ RUN set -e \
# Keep the version the same as in compute/compute-node.Dockerfile and
# test_runner/regress/test_compute_metrics.py.
ENV SQL_EXPORTER_VERSION=0.17.0
ENV SQL_EXPORTER_VERSION=0.16.0
RUN curl -fsSL \
"https://github.com/burningalchemist/sql_exporter/releases/download/${SQL_EXPORTER_VERSION}/sql_exporter-${SQL_EXPORTER_VERSION}.linux-$(case "$(uname -m)" in x86_64) echo amd64;; aarch64) echo arm64;; esac).tar.gz" \
--output sql_exporter.tar.gz \
@@ -190,6 +190,21 @@ RUN for package in Capture::Tiny DateTime Devel::Cover Digest::MD5 File::Spec JS
&& make install \
&& rm -rf ../lcov.tar.gz
# Compile and install the static OpenSSL library
ENV OPENSSL_VERSION=1.1.1w
ENV OPENSSL_PREFIX=/usr/local/openssl
RUN wget -O /tmp/openssl-${OPENSSL_VERSION}.tar.gz https://www.openssl.org/source/openssl-${OPENSSL_VERSION}.tar.gz && \
echo "cf3098950cb4d853ad95c0841f1f9c6d3dc102dccfcacd521d93925208b76ac8 /tmp/openssl-${OPENSSL_VERSION}.tar.gz" | sha256sum --check && \
cd /tmp && \
tar xzvf /tmp/openssl-${OPENSSL_VERSION}.tar.gz && \
rm /tmp/openssl-${OPENSSL_VERSION}.tar.gz && \
cd /tmp/openssl-${OPENSSL_VERSION} && \
./config --prefix=${OPENSSL_PREFIX} -static --static no-shared -fPIC && \
make -j "$(nproc)" && \
make install && \
cd /tmp && \
rm -rf /tmp/openssl-${OPENSSL_VERSION}
# Use the same version of libicu as the compute nodes so that
# clusters created using inidb on pageserver can be used by computes.
#
@@ -243,7 +258,7 @@ WORKDIR /home/nonroot
# Rust
# Please keep the version of llvm (installed above) in sync with rust llvm (`rustc --version --verbose | grep LLVM`)
ENV RUSTC_VERSION=1.84.0
ENV RUSTC_VERSION=1.83.0
ENV RUSTUP_HOME="/home/nonroot/.rustup"
ENV PATH="/home/nonroot/.cargo/bin:${PATH}"
ARG RUSTFILT_VERSION=0.2.1

File diff suppressed because it is too large Load Diff

View File

@@ -3,7 +3,7 @@
metrics: [
import 'sql_exporter/checkpoints_req.libsonnet',
import 'sql_exporter/checkpoints_timed.libsonnet',
import 'sql_exporter/compute_backpressure_throttling_seconds_total.libsonnet',
import 'sql_exporter/compute_backpressure_throttling_seconds.libsonnet',
import 'sql_exporter/compute_current_lsn.libsonnet',
import 'sql_exporter/compute_logical_snapshot_files.libsonnet',
import 'sql_exporter/compute_logical_snapshots_bytes.libsonnet',

View File

@@ -1,10 +1,10 @@
{
metric_name: 'compute_backpressure_throttling_seconds_total',
type: 'counter',
metric_name: 'compute_backpressure_throttling_seconds',
type: 'gauge',
help: 'Time compute has spent throttled',
key_labels: null,
values: [
'throttled',
],
query: importstr 'sql_exporter/compute_backpressure_throttling_seconds_total.sql',
query: importstr 'sql_exporter/compute_backpressure_throttling_seconds.sql',
}

View File

@@ -7,7 +7,7 @@ license.workspace = true
[features]
default = []
# Enables test specific features.
testing = ["fail/failpoints"]
testing = []
[dependencies]
base64.workspace = true
@@ -15,15 +15,13 @@ aws-config.workspace = true
aws-sdk-s3.workspace = true
aws-sdk-kms.workspace = true
anyhow.workspace = true
axum = { workspace = true, features = [] }
camino.workspace = true
chrono.workspace = true
cfg-if.workspace = true
clap.workspace = true
fail.workspace = true
flate2.workspace = true
futures.workspace = true
http.workspace = true
hyper0 = { workspace = true, features = ["full"] }
metrics.workspace = true
nix.workspace = true
notify.workspace = true
@@ -38,8 +36,6 @@ serde_with.workspace = true
serde_json.workspace = true
signal-hook.workspace = true
tar.workspace = true
tower.workspace = true
tower-http.workspace = true
reqwest = { workspace = true, features = ["json"] }
tokio = { workspace = true, features = ["rt", "rt-multi-thread"] }
tokio-postgres.workspace = true
@@ -51,7 +47,6 @@ tracing-subscriber.workspace = true
tracing-utils.workspace = true
thiserror.workspace = true
url.workspace = true
uuid.workspace = true
prometheus.workspace = true
postgres_initdb.workspace = true

View File

@@ -60,22 +60,19 @@ 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::http::api::launch_http_server;
use compute_tools::logger::*;
use compute_tools::monitor::launch_monitor;
use compute_tools::params::*;
use compute_tools::spec::*;
use compute_tools::swap::resize_swap;
use rlimit::{setrlimit, Resource};
use utils::failpoint_support;
// this is an arbitrary build tag. Fine as a default / for testing purposes
// in-case of not-set environment var
const BUILD_TAG_DEFAULT: &str = "latest";
fn main() -> Result<()> {
let scenario = failpoint_support::init();
let (build_tag, clap_args) = init()?;
// enable core dumping for all child processes
@@ -103,14 +100,17 @@ fn main() -> Result<()> {
maybe_delay_exit(delay_exit);
scenario.teardown();
deinit_and_exit(wait_pg_result);
}
fn init() -> Result<(String, clap::ArgMatches)> {
init_tracing_and_logging(DEFAULT_LOG_LEVEL)?;
opentelemetry::global::set_error_handler(|err| {
tracing::info!("OpenTelemetry error: {err}");
})
.expect("global error handler lock poisoned");
let mut signals = Signals::new([SIGINT, SIGTERM, SIGQUIT])?;
thread::spawn(move || {
for sig in signals.forever() {
@@ -419,14 +419,9 @@ fn start_postgres(
"running compute with features: {:?}",
state.pspec.as_ref().unwrap().spec.features
);
// before we release the mutex, fetch some parameters for later.
let &ComputeSpec {
swap_size_bytes,
disk_quota_bytes,
#[cfg(target_os = "linux")]
disable_lfc_resizing,
..
} = &state.pspec.as_ref().unwrap().spec;
// before we release the mutex, fetch the swap size (if any) for later.
let swap_size_bytes = state.pspec.as_ref().unwrap().spec.swap_size_bytes;
let disk_quota_bytes = state.pspec.as_ref().unwrap().spec.disk_quota_bytes;
drop(state);
// Launch remaining service threads
@@ -488,10 +483,7 @@ fn start_postgres(
let mut pg = None;
if !prestartup_failed {
pg = match compute.start_compute() {
Ok(pg) => {
info!(postmaster_pid = %pg.0.id(), "Postgres was started");
Some(pg)
}
Ok(pg) => Some(pg),
Err(err) => {
error!("could not start the compute node: {:#}", err);
compute.set_failed_status(err);
@@ -534,18 +526,11 @@ fn start_postgres(
// This token is used internally by the monitor to clean up all threads
let token = CancellationToken::new();
// don't pass postgres connection string to vm-monitor if we don't want it to resize LFC
let pgconnstr = if disable_lfc_resizing.unwrap_or(false) {
None
} else {
file_cache_connstr.cloned()
};
let vm_monitor = rt.as_ref().map(|rt| {
rt.spawn(vm_monitor::start(
Box::leak(Box::new(vm_monitor::Args {
cgroup: cgroup.cloned(),
pgconnstr,
pgconnstr: file_cache_connstr.cloned(),
addr: vm_monitor_addr.clone(),
})),
token.clone(),
@@ -589,8 +574,6 @@ fn wait_postgres(pg: Option<PostgresHandle>) -> Result<WaitPostgresResult> {
// propagate to Postgres and it will be shut down as well.
let mut exit_code = None;
if let Some((mut pg, logs_handle)) = pg {
info!(postmaster_pid = %pg.id(), "Waiting for Postgres to exit");
let ecode = pg
.wait()
.expect("failed to start waiting on Postgres process");

View File

@@ -17,7 +17,7 @@
//!
//! # Local Testing
//!
//! - Comment out most of the pgxns in compute-node.Dockerfile to speed up the build.
//! - Comment out most of the pgxns in The Dockerfile.compute-tools to speed up the build.
//! - Build the image with the following command:
//!
//! ```bash
@@ -31,35 +31,26 @@ use camino::{Utf8Path, Utf8PathBuf};
use clap::Parser;
use compute_tools::extension_server::{get_pg_version, PostgresMajorVersion};
use nix::unistd::Pid;
use tracing::{error, info, info_span, warn, Instrument};
use tracing::{info, info_span, warn, Instrument};
use utils::fs_ext::is_directory_empty;
#[path = "fast_import/aws_s3_sync.rs"]
mod aws_s3_sync;
#[path = "fast_import/child_stdio_to_log.rs"]
mod child_stdio_to_log;
#[path = "fast_import/s3_uri.rs"]
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);
#[path = "fast_import/s5cmd.rs"]
mod s5cmd;
#[derive(clap::Parser)]
struct Args {
#[clap(long)]
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,
s3_prefix: s3_uri::S3Uri,
#[clap(long)]
pg_bin_dir: Utf8PathBuf,
#[clap(long)]
pg_lib_dir: Utf8PathBuf,
#[clap(long)]
pg_port: Option<u16>, // port to run postgres on, 5432 is default
}
#[serde_with::serde_as]
@@ -76,13 +67,6 @@ enum EncryptionSecret {
KMS { key_id: String },
}
// copied from pageserver_api::config::defaults::DEFAULT_LOCALE to avoid dependency just for a constant
const DEFAULT_LOCALE: &str = if cfg!(target_os = "macos") {
"C"
} else {
"C.UTF-8"
};
#[tokio::main]
pub(crate) async fn main() -> anyhow::Result<()> {
utils::logging::init(
@@ -93,74 +77,30 @@ pub(crate) async fn main() -> anyhow::Result<()> {
info!("starting");
let args = Args::parse();
let Args {
working_directory,
s3_prefix,
pg_bin_dir,
pg_lib_dir,
} = 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 aws_config = aws_config::load_defaults(BehaviorVersion::v2024_03_28()).await;
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()
let spec: Spec = {
let spec_key = s3_prefix.append("/spec.json");
let s3_client = aws_sdk_s3::Client::new(&aws_config);
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 tokio::fs::create_dir(&working_directory).await {
@@ -183,6 +123,15 @@ pub(crate) async fn main() -> anyhow::Result<()> {
.await
.context("create pgdata directory")?;
//
// Setup clients
//
let aws_config = aws_config::load_defaults(BehaviorVersion::v2024_03_28()).await;
let kms_client = aws_sdk_kms::Client::new(&aws_config);
//
// Initialize pgdata
//
let pgbin = pg_bin_dir.join("postgres");
let pg_version = match get_pg_version(pgbin.as_ref()) {
PostgresMajorVersion::V14 => 14,
@@ -193,7 +142,7 @@ pub(crate) async fn main() -> anyhow::Result<()> {
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,
locale: "en_US.UTF-8", // 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.
@@ -210,7 +159,6 @@ pub(crate) async fn main() -> anyhow::Result<()> {
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"])
@@ -222,15 +170,8 @@ pub(crate) async fn main() -> anyhow::Result<()> {
.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 }
),
])
.args(["-c", "effective_io_concurrency=100"])
.env_clear()
.env("LD_LIBRARY_PATH", &pg_lib_dir)
.stdout(std::process::Stdio::piped())
.stderr(std::process::Stdio::piped())
.spawn()
@@ -244,58 +185,44 @@ pub(crate) async fn main() -> anyhow::Result<()> {
)
.instrument(info_span!("postgres")),
);
// 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();
format!("host=localhost port=5432 user={superuser} dbname=postgres");
loop {
if start_time.elapsed() > PG_WAIT_TIMEOUT {
error!(
"timeout exceeded: failed to poll postgres and create database within 10 minutes"
);
std::process::exit(1);
}
match tokio_postgres::connect(&restore_pg_connstring, tokio_postgres::NoTls).await {
Ok((client, connection)) => {
// Spawn the connection handling task to maintain the connection
tokio::spawn(async move {
if let Err(e) = connection.await {
warn!("connection error: {}", e);
}
});
match client.simple_query("CREATE DATABASE neondb;").await {
Ok(_) => {
info!("created neondb database");
break;
}
Err(e) => {
warn!(
"failed to create database: {}, retying in {}s",
e,
PG_WAIT_RETRY_INTERVAL.as_secs_f32()
);
tokio::time::sleep(PG_WAIT_RETRY_INTERVAL).await;
continue;
}
}
}
Err(_) => {
info!(
"postgres not ready yet, retrying in {}s",
PG_WAIT_RETRY_INTERVAL.as_secs_f32()
);
tokio::time::sleep(PG_WAIT_RETRY_INTERVAL).await;
continue;
}
let res = tokio_postgres::connect(&restore_pg_connstring, tokio_postgres::NoTls).await;
if res.is_ok() {
info!("postgres is ready, could connect to it");
break;
}
}
let restore_pg_connstring = restore_pg_connstring.replace("dbname=postgres", "dbname=neondb");
//
// Decrypt connection string
//
let source_connection_string = {
match spec.encryption_secret {
EncryptionSecret::KMS { key_id } => {
let mut output = kms_client
.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")?
}
}
};
//
// Start the work
//
let dumpdir = working_directory.join("dumpdir");
@@ -329,7 +256,6 @@ pub(crate) async fn main() -> anyhow::Result<()> {
.arg(&source_connection_string)
// how we run it
.env_clear()
.env("LD_LIBRARY_PATH", &pg_lib_dir)
.kill_on_drop(true)
.stdout(std::process::Stdio::piped())
.stderr(std::process::Stdio::piped())
@@ -363,7 +289,6 @@ pub(crate) async fn main() -> anyhow::Result<()> {
.arg(&dumpdir)
// how we run it
.env_clear()
.env("LD_LIBRARY_PATH", &pg_lib_dir)
.kill_on_drop(true)
.stdout(std::process::Stdio::piped())
.stderr(std::process::Stdio::piped())
@@ -385,12 +310,6 @@ pub(crate) async fn main() -> anyhow::Result<()> {
}
}
// 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(
@@ -406,24 +325,21 @@ pub(crate) async fn main() -> anyhow::Result<()> {
.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!("upload pgdata");
s5cmd::sync(Utf8Path::new(&pgdata_dir), &s3_prefix.append("/"))
.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")?;
}
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("status");
std::fs::write(&status_file, serde_json::json!({"done": true}).to_string())
.context("write status file")?;
s5cmd::sync(&status_file, &s3_prefix.append("/status/pgdata"))
.await
.context("sync status directory to destination")?;
}
Ok(())

View File

@@ -4,21 +4,24 @@ use camino::Utf8Path;
use super::s3_uri::S3Uri;
pub(crate) async fn sync(local: &Utf8Path, remote: &S3Uri) -> anyhow::Result<()> {
let mut builder = tokio::process::Command::new("aws");
let mut builder = tokio::process::Command::new("s5cmd");
// s5cmd uses aws-sdk-go v1, hence doesn't support AWS_ENDPOINT_URL
if let Some(val) = std::env::var_os("AWS_ENDPOINT_URL") {
builder.arg("--endpoint-url").arg(val);
}
builder
.arg("s3")
.arg("sync")
.arg(local.as_str())
.arg(remote.to_string());
let st = builder
.spawn()
.context("spawn aws s3 sync")?
.context("spawn s5cmd")?
.wait()
.await
.context("wait for aws s3 sync")?;
.context("wait for s5cmd")?;
if st.success() {
Ok(())
} else {
Err(anyhow::anyhow!("aws s3 sync failed"))
Err(anyhow::anyhow!("s5cmd failed"))
}
}

View File

@@ -36,11 +36,11 @@ pub async fn get_dbs_and_roles(compute: &Arc<ComputeNode>) -> anyhow::Result<Cat
#[derive(Debug, thiserror::Error)]
pub enum SchemaDumpError {
#[error("database does not exist")]
#[error("Database does not exist.")]
DatabaseDoesNotExist,
#[error("failed to execute pg_dump")]
#[error("Failed to execute pg_dump.")]
IO(#[from] std::io::Error),
#[error("unexpected I/O error")]
#[error("Unexpected error.")]
Unexpected,
}

View File

@@ -15,7 +15,7 @@ use std::time::Instant;
use anyhow::{Context, Result};
use chrono::{DateTime, Utc};
use compute_api::spec::{Database, PgIdent, Role};
use compute_api::spec::{PgIdent, Role};
use futures::future::join_all;
use futures::stream::FuturesUnordered;
use futures::StreamExt;
@@ -45,10 +45,8 @@ use crate::spec_apply::ApplySpecPhase::{
DropInvalidDatabases, DropRoles, HandleNeonExtension, HandleOtherExtensions,
RenameAndDeleteDatabases, RenameRoles, RunInEachDatabase,
};
use crate::spec_apply::PerDatabasePhase;
use crate::spec_apply::PerDatabasePhase::{
ChangeSchemaPerms, DeleteDBRoleReferences, DropSubscriptionsForDeletedDatabases,
HandleAnonExtension,
ChangeSchemaPerms, DeleteDBRoleReferences, HandleAnonExtension,
};
use crate::spec_apply::{apply_operations, MutableApplyContext, DB};
use crate::sync_sk::{check_if_synced, ping_safekeeper};
@@ -836,7 +834,7 @@ impl ComputeNode {
conf
}
pub async fn get_maintenance_client(
async fn get_maintenance_client(
conf: &tokio_postgres::Config,
) -> Result<tokio_postgres::Client> {
let mut conf = conf.clone();
@@ -945,78 +943,6 @@ impl ComputeNode {
dbs: databases,
}));
// Apply special pre drop database phase.
// NOTE: we use the code of RunInEachDatabase phase for parallelism
// and connection management, but we don't really run it in *each* database,
// only in databases, we're about to drop.
info!("Applying PerDatabase (pre-dropdb) phase");
let concurrency_token = Arc::new(tokio::sync::Semaphore::new(concurrency));
// Run the phase for each database that we're about to drop.
let db_processes = spec
.delta_operations
.iter()
.flatten()
.filter_map(move |op| {
if op.action.as_str() == "delete_db" {
Some(op.name.clone())
} else {
None
}
})
.map(|dbname| {
let spec = spec.clone();
let ctx = ctx.clone();
let jwks_roles = jwks_roles.clone();
let mut conf = conf.as_ref().clone();
let concurrency_token = concurrency_token.clone();
// We only need dbname field for this phase, so set other fields to dummy values
let db = DB::UserDB(Database {
name: dbname.clone(),
owner: "cloud_admin".to_string(),
options: None,
restrict_conn: false,
invalid: false,
});
debug!("Applying per-database phases for Database {:?}", &db);
match &db {
DB::SystemDB => {}
DB::UserDB(db) => {
conf.dbname(db.name.as_str());
}
}
let conf = Arc::new(conf);
let fut = Self::apply_spec_sql_db(
spec.clone(),
conf,
ctx.clone(),
jwks_roles.clone(),
concurrency_token.clone(),
db,
[DropSubscriptionsForDeletedDatabases].to_vec(),
);
Ok(spawn(fut))
})
.collect::<Vec<Result<_, anyhow::Error>>>();
for process in db_processes.into_iter() {
let handle = process?;
if let Err(e) = handle.await? {
// Handle the error case where the database does not exist
// We do not check whether the DB exists or not in the deletion phase,
// so we shouldn't be strict about it in pre-deletion cleanup as well.
if e.to_string().contains("does not exist") {
warn!("Error dropping subscription: {}", e);
} else {
return Err(e);
}
};
}
for phase in [
CreateSuperUser,
DropInvalidDatabases,
@@ -1036,7 +962,7 @@ impl ComputeNode {
.await?;
}
info!("Applying RunInEachDatabase2 phase");
info!("Applying RunInEachDatabase phase");
let concurrency_token = Arc::new(tokio::sync::Semaphore::new(concurrency));
let db_processes = spec
@@ -1071,12 +997,6 @@ impl ComputeNode {
jwks_roles.clone(),
concurrency_token.clone(),
db,
[
DeleteDBRoleReferences,
ChangeSchemaPerms,
HandleAnonExtension,
]
.to_vec(),
);
Ok(spawn(fut))
@@ -1123,13 +1043,16 @@ impl ComputeNode {
jwks_roles: Arc<HashSet<String>>,
concurrency_token: Arc<tokio::sync::Semaphore>,
db: DB,
subphases: Vec<PerDatabasePhase>,
) -> Result<()> {
let _permit = concurrency_token.acquire().await?;
let mut client_conn = None;
for subphase in subphases {
for subphase in [
DeleteDBRoleReferences,
ChangeSchemaPerms,
HandleAnonExtension,
] {
apply_operations(
spec.clone(),
ctx.clone(),
@@ -1258,19 +1181,8 @@ impl ComputeNode {
let mut conf = postgres::config::Config::from(conf);
conf.application_name("compute_ctl:migrations");
match conf.connect(NoTls) {
Ok(mut client) => {
if let Err(e) = handle_migrations(&mut client) {
error!("Failed to run migrations: {}", e);
}
}
Err(e) => {
error!(
"Failed to connect to the compute for running migrations: {}",
e
);
}
};
let mut client = conf.connect(NoTls)?;
handle_migrations(&mut client).context("apply_config handle_migrations")
});
Ok::<(), anyhow::Error>(())

View File

@@ -0,0 +1,591 @@
use std::convert::Infallible;
use std::net::IpAddr;
use std::net::Ipv6Addr;
use std::net::SocketAddr;
use std::sync::Arc;
use std::thread;
use crate::catalog::SchemaDumpError;
use crate::catalog::{get_database_schema, get_dbs_and_roles};
use crate::compute::forward_termination_signal;
use crate::compute::{ComputeNode, ComputeState, ParsedSpec};
use crate::installed_extensions;
use compute_api::requests::{ConfigurationRequest, ExtensionInstallRequest, SetRoleGrantsRequest};
use compute_api::responses::{
ComputeStatus, ComputeStatusResponse, ExtensionInstallResult, GenericAPIError,
SetRoleGrantsResponse,
};
use anyhow::Result;
use hyper::header::CONTENT_TYPE;
use hyper::service::{make_service_fn, service_fn};
use hyper::{Body, Method, Request, Response, Server, StatusCode};
use metrics::proto::MetricFamily;
use metrics::Encoder;
use metrics::TextEncoder;
use tokio::task;
use tracing::{debug, error, info, warn};
use tracing_utils::http::OtelName;
use utils::http::request::must_get_query_param;
fn status_response_from_state(state: &ComputeState) -> ComputeStatusResponse {
ComputeStatusResponse {
start_time: state.start_time,
tenant: state
.pspec
.as_ref()
.map(|pspec| pspec.tenant_id.to_string()),
timeline: state
.pspec
.as_ref()
.map(|pspec| pspec.timeline_id.to_string()),
status: state.status,
last_active: state.last_active,
error: state.error.clone(),
}
}
// Service function to handle all available routes.
async fn routes(req: Request<Body>, compute: &Arc<ComputeNode>) -> Response<Body> {
//
// NOTE: The URI path is currently included in traces. That's OK because
// it doesn't contain any variable parts or sensitive information. But
// please keep that in mind if you change the routing here.
//
match (req.method(), req.uri().path()) {
// Serialized compute state.
(&Method::GET, "/status") => {
debug!("serving /status GET request");
let state = compute.state.lock().unwrap();
let status_response = status_response_from_state(&state);
Response::new(Body::from(serde_json::to_string(&status_response).unwrap()))
}
// Startup metrics in JSON format. Keep /metrics reserved for a possible
// future use for Prometheus metrics format.
(&Method::GET, "/metrics.json") => {
info!("serving /metrics.json GET request");
let metrics = compute.state.lock().unwrap().metrics.clone();
Response::new(Body::from(serde_json::to_string(&metrics).unwrap()))
}
// Prometheus metrics
(&Method::GET, "/metrics") => {
debug!("serving /metrics GET request");
// When we call TextEncoder::encode() below, it will immediately
// return an error if a metric family has no metrics, so we need to
// preemptively filter out metric families with no metrics.
let metrics = installed_extensions::collect()
.into_iter()
.filter(|m| !m.get_metric().is_empty())
.collect::<Vec<MetricFamily>>();
let encoder = TextEncoder::new();
let mut buffer = vec![];
if let Err(err) = encoder.encode(&metrics, &mut buffer) {
let msg = format!("error handling /metrics request: {err}");
error!(msg);
return render_json_error(&msg, StatusCode::INTERNAL_SERVER_ERROR);
}
match Response::builder()
.status(StatusCode::OK)
.header(CONTENT_TYPE, encoder.format_type())
.body(Body::from(buffer))
{
Ok(response) => response,
Err(err) => {
let msg = format!("error handling /metrics request: {err}");
error!(msg);
render_json_error(&msg, StatusCode::INTERNAL_SERVER_ERROR)
}
}
}
// Collect Postgres current usage insights
(&Method::GET, "/insights") => {
info!("serving /insights GET request");
let status = compute.get_status();
if status != ComputeStatus::Running {
let msg = format!("compute is not running, current status: {:?}", status);
error!(msg);
return Response::new(Body::from(msg));
}
let insights = compute.collect_insights().await;
Response::new(Body::from(insights))
}
(&Method::POST, "/check_writability") => {
info!("serving /check_writability POST request");
let status = compute.get_status();
if status != ComputeStatus::Running {
let msg = format!(
"invalid compute status for check_writability request: {:?}",
status
);
error!(msg);
return Response::new(Body::from(msg));
}
let res = crate::checker::check_writability(compute).await;
match res {
Ok(_) => Response::new(Body::from("true")),
Err(e) => {
error!("check_writability failed: {}", e);
Response::new(Body::from(e.to_string()))
}
}
}
(&Method::POST, "/extensions") => {
info!("serving /extensions POST request");
let status = compute.get_status();
if status != ComputeStatus::Running {
let msg = format!(
"invalid compute status for extensions request: {:?}",
status
);
error!(msg);
return render_json_error(&msg, StatusCode::PRECONDITION_FAILED);
}
let request = hyper::body::to_bytes(req.into_body()).await.unwrap();
let request = serde_json::from_slice::<ExtensionInstallRequest>(&request).unwrap();
let res = compute
.install_extension(&request.extension, &request.database, request.version)
.await;
match res {
Ok(version) => render_json(Body::from(
serde_json::to_string(&ExtensionInstallResult {
extension: request.extension,
version,
})
.unwrap(),
)),
Err(e) => {
error!("install_extension failed: {}", e);
render_json_error(&e.to_string(), StatusCode::INTERNAL_SERVER_ERROR)
}
}
}
(&Method::GET, "/info") => {
let num_cpus = num_cpus::get_physical();
info!("serving /info GET request. num_cpus: {}", num_cpus);
Response::new(Body::from(
serde_json::json!({
"num_cpus": num_cpus,
})
.to_string(),
))
}
// Accept spec in JSON format and request compute configuration. If
// anything goes wrong after we set the compute status to `ConfigurationPending`
// and update compute state with new spec, we basically leave compute
// in the potentially wrong state. That said, it's control-plane's
// responsibility to watch compute state after reconfiguration request
// and to clean restart in case of errors.
(&Method::POST, "/configure") => {
info!("serving /configure POST request");
match handle_configure_request(req, compute).await {
Ok(msg) => Response::new(Body::from(msg)),
Err((msg, code)) => {
error!("error handling /configure request: {msg}");
render_json_error(&msg, code)
}
}
}
(&Method::POST, "/terminate") => {
info!("serving /terminate POST request");
match handle_terminate_request(compute).await {
Ok(()) => Response::new(Body::empty()),
Err((msg, code)) => {
error!("error handling /terminate request: {msg}");
render_json_error(&msg, code)
}
}
}
(&Method::GET, "/dbs_and_roles") => {
info!("serving /dbs_and_roles GET request",);
match get_dbs_and_roles(compute).await {
Ok(res) => render_json(Body::from(serde_json::to_string(&res).unwrap())),
Err(_) => {
render_json_error("can't get dbs and roles", StatusCode::INTERNAL_SERVER_ERROR)
}
}
}
(&Method::GET, "/database_schema") => {
let database = match must_get_query_param(&req, "database") {
Err(e) => return e.into_response(),
Ok(database) => database,
};
info!("serving /database_schema GET request with database: {database}",);
match get_database_schema(compute, &database).await {
Ok(res) => render_plain(Body::wrap_stream(res)),
Err(SchemaDumpError::DatabaseDoesNotExist) => {
render_json_error("database does not exist", StatusCode::NOT_FOUND)
}
Err(e) => {
error!("can't get schema dump: {}", e);
render_json_error("can't get schema dump", StatusCode::INTERNAL_SERVER_ERROR)
}
}
}
(&Method::POST, "/grants") => {
info!("serving /grants POST request");
let status = compute.get_status();
if status != ComputeStatus::Running {
let msg = format!(
"invalid compute status for set_role_grants request: {:?}",
status
);
error!(msg);
return render_json_error(&msg, StatusCode::PRECONDITION_FAILED);
}
let request = hyper::body::to_bytes(req.into_body()).await.unwrap();
let request = serde_json::from_slice::<SetRoleGrantsRequest>(&request).unwrap();
let res = compute
.set_role_grants(
&request.database,
&request.schema,
&request.privileges,
&request.role,
)
.await;
match res {
Ok(()) => render_json(Body::from(
serde_json::to_string(&SetRoleGrantsResponse {
database: request.database,
schema: request.schema,
role: request.role,
privileges: request.privileges,
})
.unwrap(),
)),
Err(e) => render_json_error(
&format!("could not grant role privileges to the schema: {e}"),
// TODO: can we filter on role/schema not found errors
// and return appropriate error code?
StatusCode::INTERNAL_SERVER_ERROR,
),
}
}
// get the list of installed extensions
// currently only used in python tests
// TODO: call it from cplane
(&Method::GET, "/installed_extensions") => {
info!("serving /installed_extensions GET request");
let status = compute.get_status();
if status != ComputeStatus::Running {
let msg = format!(
"invalid compute status for extensions request: {:?}",
status
);
error!(msg);
return Response::new(Body::from(msg));
}
let conf = compute.get_conn_conf(None);
let res =
task::spawn_blocking(move || installed_extensions::get_installed_extensions(conf))
.await
.unwrap();
match res {
Ok(res) => render_json(Body::from(serde_json::to_string(&res).unwrap())),
Err(e) => render_json_error(
&format!("could not get list of installed extensions: {}", e),
StatusCode::INTERNAL_SERVER_ERROR,
),
}
}
// download extension files from remote extension storage on demand
(&Method::POST, route) if route.starts_with("/extension_server/") => {
info!("serving {:?} POST request", route);
info!("req.uri {:?}", req.uri());
// don't even try to download extensions
// if no remote storage is configured
if compute.ext_remote_storage.is_none() {
info!("no extensions remote storage configured");
let mut resp = Response::new(Body::from("no remote storage configured"));
*resp.status_mut() = StatusCode::INTERNAL_SERVER_ERROR;
return resp;
}
let mut is_library = false;
if let Some(params) = req.uri().query() {
info!("serving {:?} POST request with params: {}", route, params);
if params == "is_library=true" {
is_library = true;
} else {
let mut resp = Response::new(Body::from("Wrong request parameters"));
*resp.status_mut() = StatusCode::BAD_REQUEST;
return resp;
}
}
let filename = route.split('/').last().unwrap().to_string();
info!("serving /extension_server POST request, filename: {filename:?} is_library: {is_library}");
// get ext_name and path from spec
// don't lock compute_state for too long
let ext = {
let compute_state = compute.state.lock().unwrap();
let pspec = compute_state.pspec.as_ref().expect("spec must be set");
let spec = &pspec.spec;
// debug only
info!("spec: {:?}", spec);
let remote_extensions = match spec.remote_extensions.as_ref() {
Some(r) => r,
None => {
info!("no remote extensions spec was provided");
let mut resp = Response::new(Body::from("no remote storage configured"));
*resp.status_mut() = StatusCode::INTERNAL_SERVER_ERROR;
return resp;
}
};
remote_extensions.get_ext(
&filename,
is_library,
&compute.build_tag,
&compute.pgversion,
)
};
match ext {
Ok((ext_name, ext_path)) => {
match compute.download_extension(ext_name, ext_path).await {
Ok(_) => Response::new(Body::from("OK")),
Err(e) => {
error!("extension download failed: {}", e);
let mut resp = Response::new(Body::from(e.to_string()));
*resp.status_mut() = StatusCode::INTERNAL_SERVER_ERROR;
resp
}
}
}
Err(e) => {
warn!("extension download failed to find extension: {}", e);
let mut resp = Response::new(Body::from("failed to find file"));
*resp.status_mut() = StatusCode::INTERNAL_SERVER_ERROR;
resp
}
}
}
// Return the `404 Not Found` for any other routes.
_ => {
let mut not_found = Response::new(Body::from("404 Not Found"));
*not_found.status_mut() = StatusCode::NOT_FOUND;
not_found
}
}
}
async fn handle_configure_request(
req: Request<Body>,
compute: &Arc<ComputeNode>,
) -> Result<String, (String, StatusCode)> {
if !compute.live_config_allowed {
return Err((
"live configuration is not allowed for this compute node".to_string(),
StatusCode::PRECONDITION_FAILED,
));
}
let body_bytes = hyper::body::to_bytes(req.into_body()).await.unwrap();
let spec_raw = String::from_utf8(body_bytes.to_vec()).unwrap();
if let Ok(request) = serde_json::from_str::<ConfigurationRequest>(&spec_raw) {
let spec = request.spec;
let parsed_spec = match ParsedSpec::try_from(spec) {
Ok(ps) => ps,
Err(msg) => return Err((msg, StatusCode::BAD_REQUEST)),
};
// XXX: wrap state update under lock in code blocks. Otherwise,
// we will try to `Send` `mut state` into the spawned thread
// bellow, which will cause error:
// ```
// error: future cannot be sent between threads safely
// ```
{
let mut state = compute.state.lock().unwrap();
if state.status != ComputeStatus::Empty && state.status != ComputeStatus::Running {
let msg = format!(
"invalid compute status for configuration request: {:?}",
state.status.clone()
);
return Err((msg, StatusCode::PRECONDITION_FAILED));
}
state.pspec = Some(parsed_spec);
state.set_status(ComputeStatus::ConfigurationPending, &compute.state_changed);
drop(state);
info!("set new spec and notified waiters");
}
// Spawn a blocking thread to wait for compute to become Running.
// This is needed to do not block the main pool of workers and
// be able to serve other requests while some particular request
// is waiting for compute to finish configuration.
let c = compute.clone();
task::spawn_blocking(move || {
let mut state = c.state.lock().unwrap();
while state.status != ComputeStatus::Running {
state = c.state_changed.wait(state).unwrap();
info!(
"waiting for compute to become Running, current status: {:?}",
state.status
);
if state.status == ComputeStatus::Failed {
let err = state.error.as_ref().map_or("unknown error", |x| x);
let msg = format!("compute configuration failed: {:?}", err);
return Err((msg, StatusCode::INTERNAL_SERVER_ERROR));
}
}
Ok(())
})
.await
.unwrap()?;
// Return current compute state if everything went well.
let state = compute.state.lock().unwrap().clone();
let status_response = status_response_from_state(&state);
Ok(serde_json::to_string(&status_response).unwrap())
} else {
Err(("invalid spec".to_string(), StatusCode::BAD_REQUEST))
}
}
fn render_json_error(e: &str, status: StatusCode) -> Response<Body> {
let error = GenericAPIError {
error: e.to_string(),
};
Response::builder()
.status(status)
.header(CONTENT_TYPE, "application/json")
.body(Body::from(serde_json::to_string(&error).unwrap()))
.unwrap()
}
fn render_json(body: Body) -> Response<Body> {
Response::builder()
.header(CONTENT_TYPE, "application/json")
.body(body)
.unwrap()
}
fn render_plain(body: Body) -> Response<Body> {
Response::builder()
.header(CONTENT_TYPE, "text/plain")
.body(body)
.unwrap()
}
async fn handle_terminate_request(compute: &Arc<ComputeNode>) -> Result<(), (String, StatusCode)> {
{
let mut state = compute.state.lock().unwrap();
if state.status == ComputeStatus::Terminated {
return Ok(());
}
if state.status != ComputeStatus::Empty && state.status != ComputeStatus::Running {
let msg = format!(
"invalid compute status for termination request: {}",
state.status
);
return Err((msg, StatusCode::PRECONDITION_FAILED));
}
state.set_status(ComputeStatus::TerminationPending, &compute.state_changed);
drop(state);
}
forward_termination_signal();
info!("sent signal and notified waiters");
// Spawn a blocking thread to wait for compute to become Terminated.
// This is needed to do not block the main pool of workers and
// be able to serve other requests while some particular request
// is waiting for compute to finish configuration.
let c = compute.clone();
task::spawn_blocking(move || {
let mut state = c.state.lock().unwrap();
while state.status != ComputeStatus::Terminated {
state = c.state_changed.wait(state).unwrap();
info!(
"waiting for compute to become {}, current status: {:?}",
ComputeStatus::Terminated,
state.status
);
}
Ok(())
})
.await
.unwrap()?;
info!("terminated Postgres");
Ok(())
}
// Main Hyper HTTP server function that runs it and blocks waiting on it forever.
#[tokio::main]
async fn serve(port: u16, state: Arc<ComputeNode>) {
// this usually binds to both IPv4 and IPv6 on linux
// see e.g. https://github.com/rust-lang/rust/pull/34440
let addr = SocketAddr::new(IpAddr::from(Ipv6Addr::UNSPECIFIED), port);
let make_service = make_service_fn(move |_conn| {
let state = state.clone();
async move {
Ok::<_, Infallible>(service_fn(move |req: Request<Body>| {
let state = state.clone();
async move {
Ok::<_, Infallible>(
// NOTE: We include the URI path in the string. It
// doesn't contain any variable parts or sensitive
// information in this API.
tracing_utils::http::tracing_handler(
req,
|req| routes(req, &state),
OtelName::UriPath,
)
.await,
)
}
}))
}
});
info!("starting HTTP server on {}", addr);
let server = Server::bind(&addr).serve(make_service);
// Run this server forever
if let Err(e) = server.await {
error!("server error: {}", e);
}
}
/// Launch a separate Hyper HTTP API server thread and return its `JoinHandle`.
pub fn launch_http_server(port: u16, state: &Arc<ComputeNode>) -> Result<thread::JoinHandle<()>> {
let state = Arc::clone(state);
Ok(thread::Builder::new()
.name("http-endpoint".into())
.spawn(move || serve(port, state))?)
}

View File

@@ -1,48 +0,0 @@
use std::ops::{Deref, DerefMut};
use axum::{
async_trait,
extract::{rejection::JsonRejection, FromRequest, Request},
};
use compute_api::responses::GenericAPIError;
use http::StatusCode;
/// Custom `Json` extractor, so that we can format errors into
/// `JsonResponse<GenericAPIError>`.
#[derive(Debug, Clone, Copy, Default)]
pub(crate) struct Json<T>(pub T);
#[async_trait]
impl<S, T> FromRequest<S> for Json<T>
where
axum::Json<T>: FromRequest<S, Rejection = JsonRejection>,
S: Send + Sync,
{
type Rejection = (StatusCode, axum::Json<GenericAPIError>);
async fn from_request(req: Request, state: &S) -> Result<Self, Self::Rejection> {
match axum::Json::<T>::from_request(req, state).await {
Ok(value) => Ok(Self(value.0)),
Err(rejection) => Err((
rejection.status(),
axum::Json(GenericAPIError {
error: rejection.body_text().to_lowercase(),
}),
)),
}
}
}
impl<T> Deref for Json<T> {
type Target = T;
fn deref(&self) -> &Self::Target {
&self.0
}
}
impl<T> DerefMut for Json<T> {
fn deref_mut(&mut self) -> &mut Self::Target {
&mut self.0
}
}

View File

@@ -1,7 +0,0 @@
pub(crate) mod json;
pub(crate) mod path;
pub(crate) mod query;
pub(crate) use json::Json;
pub(crate) use path::Path;
pub(crate) use query::Query;

View File

@@ -1,48 +0,0 @@
use std::ops::{Deref, DerefMut};
use axum::{
async_trait,
extract::{rejection::PathRejection, FromRequestParts},
};
use compute_api::responses::GenericAPIError;
use http::{request::Parts, StatusCode};
/// Custom `Path` extractor, so that we can format errors into
/// `JsonResponse<GenericAPIError>`.
#[derive(Debug, Clone, Copy, Default)]
pub(crate) struct Path<T>(pub T);
#[async_trait]
impl<S, T> FromRequestParts<S> for Path<T>
where
axum::extract::Path<T>: FromRequestParts<S, Rejection = PathRejection>,
S: Send + Sync,
{
type Rejection = (StatusCode, axum::Json<GenericAPIError>);
async fn from_request_parts(parts: &mut Parts, state: &S) -> Result<Self, Self::Rejection> {
match axum::extract::Path::<T>::from_request_parts(parts, state).await {
Ok(value) => Ok(Self(value.0)),
Err(rejection) => Err((
rejection.status(),
axum::Json(GenericAPIError {
error: rejection.body_text().to_ascii_lowercase(),
}),
)),
}
}
}
impl<T> Deref for Path<T> {
type Target = T;
fn deref(&self) -> &Self::Target {
&self.0
}
}
impl<T> DerefMut for Path<T> {
fn deref_mut(&mut self) -> &mut Self::Target {
&mut self.0
}
}

View File

@@ -1,48 +0,0 @@
use std::ops::{Deref, DerefMut};
use axum::{
async_trait,
extract::{rejection::QueryRejection, FromRequestParts},
};
use compute_api::responses::GenericAPIError;
use http::{request::Parts, StatusCode};
/// Custom `Query` extractor, so that we can format errors into
/// `JsonResponse<GenericAPIError>`.
#[derive(Debug, Clone, Copy, Default)]
pub(crate) struct Query<T>(pub T);
#[async_trait]
impl<S, T> FromRequestParts<S> for Query<T>
where
axum::extract::Query<T>: FromRequestParts<S, Rejection = QueryRejection>,
S: Send + Sync,
{
type Rejection = (StatusCode, axum::Json<GenericAPIError>);
async fn from_request_parts(parts: &mut Parts, state: &S) -> Result<Self, Self::Rejection> {
match axum::extract::Query::<T>::from_request_parts(parts, state).await {
Ok(value) => Ok(Self(value.0)),
Err(rejection) => Err((
rejection.status(),
axum::Json(GenericAPIError {
error: rejection.body_text().to_ascii_lowercase(),
}),
)),
}
}
}
impl<T> Deref for Query<T> {
type Target = T;
fn deref(&self) -> &Self::Target {
&self.0
}
}
impl<T> DerefMut for Query<T> {
fn deref_mut(&mut self) -> &mut Self::Target {
&mut self.0
}
}

View File

@@ -1,56 +1 @@
use axum::{body::Body, response::Response};
use compute_api::responses::{ComputeStatus, GenericAPIError};
use http::{header::CONTENT_TYPE, StatusCode};
use serde::Serialize;
use tracing::error;
pub use server::launch_http_server;
mod extract;
mod routes;
mod server;
/// Convenience response builder for JSON responses
struct JsonResponse;
impl JsonResponse {
/// Helper for actually creating a response
fn create_response(code: StatusCode, body: impl Serialize) -> Response {
Response::builder()
.status(code)
.header(CONTENT_TYPE.as_str(), "application/json")
.body(Body::from(serde_json::to_string(&body).unwrap()))
.unwrap()
}
/// Create a successful error response
pub(self) fn success(code: StatusCode, body: impl Serialize) -> Response {
assert!({
let code = code.as_u16();
(200..300).contains(&code)
});
Self::create_response(code, body)
}
/// Create an error response
pub(self) fn error(code: StatusCode, error: impl ToString) -> Response {
assert!(code.as_u16() >= 400);
let message = error.to_string();
error!(message);
Self::create_response(code, &GenericAPIError { error: message })
}
/// Create an error response related to the compute being in an invalid state
pub(self) fn invalid_status(status: ComputeStatus) -> Response {
Self::create_response(
StatusCode::PRECONDITION_FAILED,
&GenericAPIError {
error: format!("invalid compute status: {status}"),
},
)
}
}
pub mod api;

View File

@@ -37,7 +37,7 @@ paths:
schema:
$ref: "#/components/schemas/ComputeMetrics"
/metrics:
/metrics
get:
tags:
- Info

View File

@@ -1,20 +0,0 @@
use std::sync::Arc;
use axum::{extract::State, response::Response};
use compute_api::responses::ComputeStatus;
use http::StatusCode;
use crate::{checker::check_writability, compute::ComputeNode, http::JsonResponse};
/// Check that the compute is currently running.
pub(in crate::http) async fn is_writable(State(compute): State<Arc<ComputeNode>>) -> Response {
let status = compute.get_status();
if status != ComputeStatus::Running {
return JsonResponse::invalid_status(status);
}
match check_writability(&compute).await {
Ok(_) => JsonResponse::success(StatusCode::OK, true),
Err(e) => JsonResponse::error(StatusCode::INTERNAL_SERVER_ERROR, e),
}
}

View File

@@ -1,91 +0,0 @@
use std::sync::Arc;
use axum::{extract::State, response::Response};
use compute_api::{
requests::ConfigurationRequest,
responses::{ComputeStatus, ComputeStatusResponse},
};
use http::StatusCode;
use tokio::task;
use tracing::info;
use crate::{
compute::{ComputeNode, ParsedSpec},
http::{extract::Json, JsonResponse},
};
// Accept spec in JSON format and request compute configuration. If anything
// goes wrong after we set the compute status to `ConfigurationPending` and
// update compute state with new spec, we basically leave compute in the
// potentially wrong state. That said, it's control-plane's responsibility to
// watch compute state after reconfiguration request and to clean restart in
// case of errors.
pub(in crate::http) async fn configure(
State(compute): State<Arc<ComputeNode>>,
request: Json<ConfigurationRequest>,
) -> Response {
if !compute.live_config_allowed {
return JsonResponse::error(
StatusCode::PRECONDITION_FAILED,
"live configuration is not allowed for this compute node".to_string(),
);
}
let pspec = match ParsedSpec::try_from(request.spec.clone()) {
Ok(p) => p,
Err(e) => return JsonResponse::error(StatusCode::BAD_REQUEST, e),
};
// XXX: wrap state update under lock in a code block. Otherwise, we will try
// to `Send` `mut state` into the spawned thread bellow, which will cause
// the following rustc error:
//
// error: future cannot be sent between threads safely
{
let mut state = compute.state.lock().unwrap();
if !matches!(state.status, ComputeStatus::Empty | ComputeStatus::Running) {
return JsonResponse::invalid_status(state.status);
}
state.pspec = Some(pspec);
state.set_status(ComputeStatus::ConfigurationPending, &compute.state_changed);
drop(state);
}
// Spawn a blocking thread to wait for compute to become Running. This is
// needed to do not block the main pool of workers and be able to serve
// other requests while some particular request is waiting for compute to
// finish configuration.
let c = compute.clone();
let completed = task::spawn_blocking(move || {
let mut state = c.state.lock().unwrap();
while state.status != ComputeStatus::Running {
state = c.state_changed.wait(state).unwrap();
info!(
"waiting for compute to become {}, current status: {}",
ComputeStatus::Running,
state.status
);
if state.status == ComputeStatus::Failed {
let err = state.error.as_ref().map_or("unknown error", |x| x);
let msg = format!("compute configuration failed: {:?}", err);
return Err(msg);
}
}
Ok(())
})
.await
.unwrap();
if let Err(e) = completed {
return JsonResponse::error(StatusCode::INTERNAL_SERVER_ERROR, e);
}
// Return current compute state if everything went well.
let state = compute.state.lock().unwrap().clone();
let body = ComputeStatusResponse::from(&state);
JsonResponse::success(StatusCode::OK, body)
}

View File

@@ -1,34 +0,0 @@
use std::sync::Arc;
use axum::{body::Body, extract::State, response::Response};
use http::{header::CONTENT_TYPE, StatusCode};
use serde::Deserialize;
use crate::{
catalog::{get_database_schema, SchemaDumpError},
compute::ComputeNode,
http::{extract::Query, JsonResponse},
};
#[derive(Debug, Clone, Deserialize)]
pub(in crate::http) struct DatabaseSchemaParams {
database: String,
}
/// Get a schema dump of the requested database.
pub(in crate::http) async fn get_schema_dump(
params: Query<DatabaseSchemaParams>,
State(compute): State<Arc<ComputeNode>>,
) -> Response {
match get_database_schema(&compute, &params.database).await {
Ok(schema) => Response::builder()
.status(StatusCode::OK)
.header(CONTENT_TYPE.as_str(), "application/json")
.body(Body::from_stream(schema))
.unwrap(),
Err(SchemaDumpError::DatabaseDoesNotExist) => {
JsonResponse::error(StatusCode::NOT_FOUND, SchemaDumpError::DatabaseDoesNotExist)
}
Err(e) => JsonResponse::error(StatusCode::INTERNAL_SERVER_ERROR, e),
}
}

View File

@@ -1,16 +0,0 @@
use std::sync::Arc;
use axum::{extract::State, response::Response};
use http::StatusCode;
use crate::{catalog::get_dbs_and_roles, compute::ComputeNode, http::JsonResponse};
/// Get the databases and roles from the compute.
pub(in crate::http) async fn get_catalog_objects(
State(compute): State<Arc<ComputeNode>>,
) -> Response {
match get_dbs_and_roles(&compute).await {
Ok(catalog_objects) => JsonResponse::success(StatusCode::OK, catalog_objects),
Err(e) => JsonResponse::error(StatusCode::INTERNAL_SERVER_ERROR, e),
}
}

View File

@@ -1,68 +0,0 @@
use std::sync::Arc;
use axum::{
extract::State,
response::{IntoResponse, Response},
};
use http::StatusCode;
use serde::Deserialize;
use crate::{
compute::ComputeNode,
http::{
extract::{Path, Query},
JsonResponse,
},
};
#[derive(Debug, Clone, Deserialize)]
pub(in crate::http) struct ExtensionServerParams {
#[serde(default)]
is_library: bool,
}
/// Download a remote extension.
pub(in crate::http) async fn download_extension(
Path(filename): Path<String>,
params: Query<ExtensionServerParams>,
State(compute): State<Arc<ComputeNode>>,
) -> Response {
// Don't even try to download extensions if no remote storage is configured
if compute.ext_remote_storage.is_none() {
return JsonResponse::error(
StatusCode::PRECONDITION_FAILED,
"remote storage is not configured",
);
}
let ext = {
let state = compute.state.lock().unwrap();
let pspec = state.pspec.as_ref().unwrap();
let spec = &pspec.spec;
let remote_extensions = match spec.remote_extensions.as_ref() {
Some(r) => r,
None => {
return JsonResponse::error(
StatusCode::CONFLICT,
"information about remote extensions is unavailable",
);
}
};
remote_extensions.get_ext(
&filename,
params.is_library,
&compute.build_tag,
&compute.pgversion,
)
};
match ext {
Ok((ext_name, ext_path)) => match compute.download_extension(ext_name, ext_path).await {
Ok(_) => StatusCode::OK.into_response(),
Err(e) => JsonResponse::error(StatusCode::INTERNAL_SERVER_ERROR, e),
},
Err(e) => JsonResponse::error(StatusCode::NOT_FOUND, e),
}
}

View File

@@ -1,45 +0,0 @@
use std::sync::Arc;
use axum::{extract::State, response::Response};
use compute_api::{
requests::ExtensionInstallRequest,
responses::{ComputeStatus, ExtensionInstallResponse},
};
use http::StatusCode;
use crate::{
compute::ComputeNode,
http::{extract::Json, JsonResponse},
};
/// Install a extension.
pub(in crate::http) async fn install_extension(
State(compute): State<Arc<ComputeNode>>,
request: Json<ExtensionInstallRequest>,
) -> Response {
let status = compute.get_status();
if status != ComputeStatus::Running {
return JsonResponse::invalid_status(status);
}
match compute
.install_extension(
&request.extension,
&request.database,
request.version.to_string(),
)
.await
{
Ok(version) => JsonResponse::success(
StatusCode::CREATED,
Some(ExtensionInstallResponse {
extension: request.extension.clone(),
version,
}),
),
Err(e) => JsonResponse::error(
StatusCode::INTERNAL_SERVER_ERROR,
format!("failed to install extension: {e}"),
),
}
}

View File

@@ -1,35 +0,0 @@
use axum::response::{IntoResponse, Response};
use http::StatusCode;
use tracing::info;
use utils::failpoint_support::{apply_failpoint, ConfigureFailpointsRequest};
use crate::http::{extract::Json, JsonResponse};
/// Configure failpoints for testing purposes.
pub(in crate::http) async fn configure_failpoints(
failpoints: Json<ConfigureFailpointsRequest>,
) -> Response {
if !fail::has_failpoints() {
return JsonResponse::error(
StatusCode::PRECONDITION_FAILED,
"Cannot manage failpoints because neon was compiled without failpoints support",
);
}
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(e) = cfg_result {
return JsonResponse::error(
StatusCode::BAD_REQUEST,
format!("failed to configure failpoints: {e}"),
);
}
}
StatusCode::OK.into_response()
}

View File

@@ -1,48 +0,0 @@
use std::sync::Arc;
use axum::{extract::State, response::Response};
use compute_api::{
requests::SetRoleGrantsRequest,
responses::{ComputeStatus, SetRoleGrantsResponse},
};
use http::StatusCode;
use crate::{
compute::ComputeNode,
http::{extract::Json, JsonResponse},
};
/// Add grants for a role.
pub(in crate::http) async fn add_grant(
State(compute): State<Arc<ComputeNode>>,
request: Json<SetRoleGrantsRequest>,
) -> Response {
let status = compute.get_status();
if status != ComputeStatus::Running {
return JsonResponse::invalid_status(status);
}
match compute
.set_role_grants(
&request.database,
&request.schema,
&request.privileges,
&request.role,
)
.await
{
Ok(()) => JsonResponse::success(
StatusCode::CREATED,
Some(SetRoleGrantsResponse {
database: request.database.clone(),
schema: request.schema.clone(),
role: request.role.clone(),
privileges: request.privileges.clone(),
}),
),
Err(e) => JsonResponse::error(
StatusCode::INTERNAL_SERVER_ERROR,
format!("failed to grant role privileges to the schema: {e}"),
),
}
}

View File

@@ -1,11 +0,0 @@
use axum::response::Response;
use compute_api::responses::InfoResponse;
use http::StatusCode;
use crate::http::JsonResponse;
/// Get information about the physical characteristics about the compute.
pub(in crate::http) async fn get_info() -> Response {
let num_cpus = num_cpus::get_physical();
JsonResponse::success(StatusCode::OK, &InfoResponse { num_cpus })
}

View File

@@ -1,18 +0,0 @@
use std::sync::Arc;
use axum::{extract::State, response::Response};
use compute_api::responses::ComputeStatus;
use http::StatusCode;
use crate::{compute::ComputeNode, http::JsonResponse};
/// Collect current Postgres usage insights.
pub(in crate::http) async fn get_insights(State(compute): State<Arc<ComputeNode>>) -> Response {
let status = compute.get_status();
if status != ComputeStatus::Running {
return JsonResponse::invalid_status(status);
}
let insights = compute.collect_insights().await;
JsonResponse::success(StatusCode::OK, insights)
}

View File

@@ -1,33 +0,0 @@
use std::sync::Arc;
use axum::{extract::State, response::Response};
use compute_api::responses::ComputeStatus;
use http::StatusCode;
use tokio::task;
use crate::{compute::ComputeNode, http::JsonResponse, installed_extensions};
/// Get a list of installed extensions.
pub(in crate::http) async fn get_installed_extensions(
State(compute): State<Arc<ComputeNode>>,
) -> Response {
let status = compute.get_status();
if status != ComputeStatus::Running {
return JsonResponse::invalid_status(status);
}
let conf = compute.get_conn_conf(None);
let res = task::spawn_blocking(move || installed_extensions::get_installed_extensions(conf))
.await
.unwrap();
match res {
Ok(installed_extensions) => {
JsonResponse::success(StatusCode::OK, Some(installed_extensions))
}
Err(e) => JsonResponse::error(
StatusCode::INTERNAL_SERVER_ERROR,
format!("failed to get list of installed extensions: {e}"),
),
}
}

View File

@@ -1,32 +0,0 @@
use axum::{body::Body, response::Response};
use http::header::CONTENT_TYPE;
use http::StatusCode;
use metrics::proto::MetricFamily;
use metrics::Encoder;
use metrics::TextEncoder;
use crate::{http::JsonResponse, installed_extensions};
/// Expose Prometheus metrics.
pub(in crate::http) async fn get_metrics() -> Response {
// When we call TextEncoder::encode() below, it will immediately return an
// error if a metric family has no metrics, so we need to preemptively
// filter out metric families with no metrics.
let metrics = installed_extensions::collect()
.into_iter()
.filter(|m| !m.get_metric().is_empty())
.collect::<Vec<MetricFamily>>();
let encoder = TextEncoder::new();
let mut buffer = vec![];
if let Err(e) = encoder.encode(&metrics, &mut buffer) {
return JsonResponse::error(StatusCode::INTERNAL_SERVER_ERROR, e);
}
Response::builder()
.status(StatusCode::OK)
.header(CONTENT_TYPE, encoder.format_type())
.body(Body::from(buffer))
.unwrap()
}

View File

@@ -1,12 +0,0 @@
use std::sync::Arc;
use axum::{extract::State, response::Response};
use http::StatusCode;
use crate::{compute::ComputeNode, http::JsonResponse};
/// Get startup metrics.
pub(in crate::http) async fn get_metrics(State(compute): State<Arc<ComputeNode>>) -> Response {
let metrics = compute.state.lock().unwrap().metrics.clone();
JsonResponse::success(StatusCode::OK, metrics)
}

View File

@@ -1,38 +0,0 @@
use compute_api::responses::ComputeStatusResponse;
use crate::compute::ComputeState;
pub(in crate::http) mod check_writability;
pub(in crate::http) mod configure;
pub(in crate::http) mod database_schema;
pub(in crate::http) mod dbs_and_roles;
pub(in crate::http) mod extension_server;
pub(in crate::http) mod extensions;
pub(in crate::http) mod failpoints;
pub(in crate::http) mod grants;
pub(in crate::http) mod info;
pub(in crate::http) mod insights;
pub(in crate::http) mod installed_extensions;
pub(in crate::http) mod metrics;
pub(in crate::http) mod metrics_json;
pub(in crate::http) mod status;
pub(in crate::http) mod terminate;
impl From<&ComputeState> for ComputeStatusResponse {
fn from(state: &ComputeState) -> Self {
ComputeStatusResponse {
start_time: state.start_time,
tenant: state
.pspec
.as_ref()
.map(|pspec| pspec.tenant_id.to_string()),
timeline: state
.pspec
.as_ref()
.map(|pspec| pspec.timeline_id.to_string()),
status: state.status,
last_active: state.last_active,
error: state.error.clone(),
}
}
}

View File

@@ -1,14 +0,0 @@
use std::{ops::Deref, sync::Arc};
use axum::{extract::State, http::StatusCode, response::Response};
use compute_api::responses::ComputeStatusResponse;
use crate::{compute::ComputeNode, http::JsonResponse};
/// Retrieve the state of the comute.
pub(in crate::http) async fn get_status(State(compute): State<Arc<ComputeNode>>) -> Response {
let state = compute.state.lock().unwrap();
let body = ComputeStatusResponse::from(state.deref());
JsonResponse::success(StatusCode::OK, body)
}

View File

@@ -1,58 +0,0 @@
use std::sync::Arc;
use axum::{
extract::State,
response::{IntoResponse, Response},
};
use compute_api::responses::ComputeStatus;
use http::StatusCode;
use tokio::task;
use tracing::info;
use crate::{
compute::{forward_termination_signal, ComputeNode},
http::JsonResponse,
};
/// Terminate the compute.
pub(in crate::http) async fn terminate(State(compute): State<Arc<ComputeNode>>) -> Response {
{
let mut state = compute.state.lock().unwrap();
if state.status == ComputeStatus::Terminated {
return StatusCode::CREATED.into_response();
}
if !matches!(state.status, ComputeStatus::Empty | ComputeStatus::Running) {
return JsonResponse::invalid_status(state.status);
}
state.set_status(ComputeStatus::TerminationPending, &compute.state_changed);
drop(state);
}
forward_termination_signal();
info!("sent signal and notified waiters");
// Spawn a blocking thread to wait for compute to become Terminated.
// This is needed to do not block the main pool of workers and
// be able to serve other requests while some particular request
// is waiting for compute to finish configuration.
let c = compute.clone();
task::spawn_blocking(move || {
let mut state = c.state.lock().unwrap();
while state.status != ComputeStatus::Terminated {
state = c.state_changed.wait(state).unwrap();
info!(
"waiting for compute to become {}, current status: {:?}",
ComputeStatus::Terminated,
state.status
);
}
})
.await
.unwrap();
info!("terminated Postgres");
StatusCode::OK.into_response()
}

View File

@@ -1,155 +0,0 @@
use std::{
net::{IpAddr, Ipv6Addr, SocketAddr},
sync::Arc,
thread,
time::Duration,
};
use anyhow::Result;
use axum::{
extract::Request,
middleware::{self, Next},
response::{IntoResponse, Response},
routing::{get, post},
Router,
};
use http::StatusCode;
use tokio::net::TcpListener;
use tower::ServiceBuilder;
use tower_http::{request_id::PropagateRequestIdLayer, trace::TraceLayer};
use tracing::{debug, error, info, Span};
use uuid::Uuid;
use super::routes::{
check_writability, configure, database_schema, dbs_and_roles, extension_server, extensions,
grants, info as info_route, insights, installed_extensions, metrics, metrics_json, status,
terminate,
};
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
}
/// 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("/info", get(info_route::get_info))
.route("/insights", get(insights::get_insights))
.route(
"/installed_extensions",
get(installed_extensions::get_installed_extensions),
)
.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(
ServiceBuilder::new()
// Add this middleware since we assume the request ID exists
.layer(middleware::from_fn(maybe_add_request_id_header))
.layer(
TraceLayer::new_for_http()
.on_request(|request: &http::Request<_>, _span: &Span| {
let request_id = request
.headers()
.get(X_REQUEST_ID)
.unwrap()
.to_str()
.unwrap();
match request.uri().path() {
"/metrics" => {
debug!(%request_id, "{} {}", request.method(), request.uri())
}
_ => info!(%request_id, "{} {}", request.method(), request.uri()),
};
})
.on_response(
|response: &http::Response<_>, latency: Duration, _span: &Span| {
let request_id = response
.headers()
.get(X_REQUEST_ID)
.unwrap()
.to_str()
.unwrap();
info!(
%request_id,
code = response.status().as_u16(),
latency = latency.as_millis()
)
},
),
)
.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))
}
// 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;
}
};
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);
}
}
/// 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);
Ok(thread::Builder::new()
.name("http-server".into())
.spawn(move || serve(port, state))?)
}

View File

@@ -3,6 +3,8 @@
#![deny(unsafe_code)]
#![deny(clippy::undocumented_unsafe_blocks)]
extern crate hyper0 as hyper;
pub mod checker;
pub mod config;
pub mod configurator;

View File

@@ -1,16 +1,13 @@
use anyhow::{Context, Result};
use fail::fail_point;
use postgres::{Client, Transaction};
use postgres::Client;
use tracing::info;
/// Runs a series of migrations on a target database
pub(crate) struct MigrationRunner<'m> {
client: &'m mut Client,
migrations: &'m [&'m str],
}
impl<'m> MigrationRunner<'m> {
/// Create a new migration runner
pub fn new(client: &'m mut Client, migrations: &'m [&'m str]) -> Self {
// The neon_migration.migration_id::id column is a bigint, which is equivalent to an i64
assert!(migrations.len() + 1 < i64::MAX as usize);
@@ -18,110 +15,87 @@ impl<'m> MigrationRunner<'m> {
Self { client, migrations }
}
/// Get the current value neon_migration.migration_id
fn get_migration_id(&mut self) -> Result<i64> {
let query = "SELECT id FROM neon_migration.migration_id";
let row = self
.client
.query_one("SELECT id FROM neon_migration.migration_id", &[])?;
.query_one(query, &[])
.context("run_migrations get migration_id")?;
Ok(row.get::<&str, i64>("id"))
}
/// Update the neon_migration.migration_id value
///
/// 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<()> {
// 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
if cfg!(feature = "testing") {
let fail = (|| {
fail_point!("compute-migration", |fail_migration_id| {
migration_id == fail_migration_id.unwrap().parse::<i64>().unwrap()
});
fn update_migration_id(&mut self, migration_id: i64) -> Result<()> {
let setval = format!("UPDATE neon_migration.migration_id SET id={}", migration_id);
false
})();
if fail {
return Err(anyhow::anyhow!(format!(
"migration {} was configured to fail because of a failpoint",
migration_id
)));
}
}
txn.query(
"UPDATE neon_migration.migration_id SET id = $1",
&[&migration_id],
)
.with_context(|| format!("update neon_migration.migration_id to {migration_id}"))?;
self.client
.simple_query(&setval)
.context("run_migrations update id")?;
Ok(())
}
/// Prepare the migrations the target database for handling migrations
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",
)?;
self.client
.simple_query("ALTER SCHEMA neon_migration OWNER TO cloud_admin")?;
self.client
.simple_query("REVOKE ALL ON SCHEMA neon_migration FROM PUBLIC")?;
fn prepare_migrations(&mut self) -> Result<()> {
let query = "CREATE SCHEMA IF NOT EXISTS neon_migration";
self.client.simple_query(query)?;
let 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(query)?;
let query = "INSERT INTO neon_migration.migration_id VALUES (0, 0) ON CONFLICT DO NOTHING";
self.client.simple_query(query)?;
let query = "ALTER SCHEMA neon_migration OWNER TO cloud_admin";
self.client.simple_query(query)?;
let query = "REVOKE ALL ON SCHEMA neon_migration FROM PUBLIC";
self.client.simple_query(query)?;
Ok(())
}
/// Run an individual migration
fn run_migration(txn: &mut Transaction, migration_id: i64, migration: &str) -> Result<()> {
if migration.starts_with("-- SKIP") {
info!("Skipping migration id={}", migration_id);
// 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(txn, migration_id)?;
} else {
info!("Running migration id={}:\n{}\n", migration_id, migration);
txn.simple_query(migration)
.with_context(|| format!("apply migration {migration_id}"))?;
Self::update_migration_id(txn, migration_id)?;
}
Ok(())
}
/// Run the configured set of migrations
pub fn run_migrations(mut self) -> Result<()> {
self.prepare_database()
.context("prepare database to handle migrations")?;
self.prepare_migrations()?;
let mut current_migration = self.get_migration_id()? 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;
macro_rules! migration_id {
($cm:expr) => {
($cm + 1) as i64
};
}
let mut txn = self
.client
.transaction()
.with_context(|| format!("begin transaction for migration {migration_id}"))?;
let migration = self.migrations[current_migration];
Self::run_migration(&mut txn, migration_id, self.migrations[current_migration])
.with_context(|| format!("running migration {migration_id}"))?;
if migration.starts_with("-- SKIP") {
info!("Skipping migration id={}", migration_id!(current_migration));
} else {
info!(
"Running migration id={}:\n{}\n",
migration_id!(current_migration),
migration
);
txn.commit()
.with_context(|| format!("commit transaction for migration {migration_id}"))?;
self.client
.simple_query("BEGIN")
.context("begin migration")?;
info!("Finished migration id={}", migration_id);
self.client.simple_query(migration).with_context(|| {
format!(
"run_migrations migration id={}",
migration_id!(current_migration)
)
})?;
// Migration IDs start at 1
self.update_migration_id(migration_id!(current_migration))?;
self.client
.simple_query("COMMIT")
.context("commit migration")?;
info!("Finished migration id={}", migration_id!(current_migration));
}
current_migration += 1;
}

View File

@@ -1,9 +0,0 @@
DO $$
DECLARE
bypassrls boolean;
BEGIN
SELECT rolbypassrls INTO bypassrls FROM pg_roles WHERE rolname = 'neon_superuser';
IF NOT bypassrls THEN
RAISE EXCEPTION 'neon_superuser cannot bypass RLS';
END IF;
END $$;

View File

@@ -1,25 +0,0 @@
DO $$
DECLARE
role record;
BEGIN
FOR role IN
SELECT rolname AS name, rolinherit AS inherit
FROM pg_roles
WHERE pg_has_role(rolname, 'neon_superuser', 'member')
LOOP
IF NOT role.inherit THEN
RAISE EXCEPTION '% cannot inherit', quote_ident(role.name);
END IF;
END LOOP;
FOR role IN
SELECT rolname AS name, rolbypassrls AS bypassrls
FROM pg_roles
WHERE NOT pg_has_role(rolname, 'neon_superuser', 'member')
AND NOT starts_with(rolname, 'pg_')
LOOP
IF role.bypassrls THEN
RAISE EXCEPTION '% can bypass RLS', quote_ident(role.name);
END IF;
END LOOP;
END $$;

View File

@@ -1,10 +0,0 @@
DO $$
BEGIN
IF (SELECT current_setting('server_version_num')::numeric < 160000) THEN
RETURN;
END IF;
IF NOT (SELECT pg_has_role('neon_superuser', 'pg_create_subscription', 'member')) THEN
RAISE EXCEPTION 'neon_superuser cannot execute pg_create_subscription';
END IF;
END $$;

View File

@@ -1,19 +0,0 @@
DO $$
DECLARE
monitor record;
BEGIN
SELECT pg_has_role('neon_superuser', 'pg_monitor', 'member') AS member,
admin_option AS admin
INTO monitor
FROM pg_auth_members
WHERE roleid = 'pg_monitor'::regrole
AND member = 'pg_monitor'::regrole;
IF NOT monitor.member THEN
RAISE EXCEPTION 'neon_superuser is not a member of pg_monitor';
END IF;
IF NOT monitor.admin THEN
RAISE EXCEPTION 'neon_superuser cannot grant pg_monitor';
END IF;
END $$;

View File

@@ -1,2 +0,0 @@
-- This test was never written becuase at the time migration tests were added
-- the accompanying migration was already skipped.

View File

@@ -1,2 +0,0 @@
-- This test was never written becuase at the time migration tests were added
-- the accompanying migration was already skipped.

View File

@@ -1,2 +0,0 @@
-- This test was never written becuase at the time migration tests were added
-- the accompanying migration was already skipped.

View File

@@ -1,2 +0,0 @@
-- This test was never written becuase at the time migration tests were added
-- the accompanying migration was already skipped.

View File

@@ -1,2 +0,0 @@
-- This test was never written becuase at the time migration tests were added
-- the accompanying migration was already skipped.

View File

@@ -1,13 +0,0 @@
DO $$
DECLARE
can_execute boolean;
BEGIN
SELECT bool_and(has_function_privilege('neon_superuser', oid, 'execute'))
INTO can_execute
FROM pg_proc
WHERE proname IN ('pg_export_snapshot', 'pg_log_standby_snapshot')
AND pronamespace = 'pg_catalog'::regnamespace;
IF NOT can_execute THEN
RAISE EXCEPTION 'neon_superuser cannot execute both pg_export_snapshot and pg_log_standby_snapshot';
END IF;
END $$;

View File

@@ -1,13 +0,0 @@
DO $$
DECLARE
can_execute boolean;
BEGIN
SELECT has_function_privilege('neon_superuser', oid, 'execute')
INTO can_execute
FROM pg_proc
WHERE proname = 'pg_show_replication_origin_status'
AND pronamespace = 'pg_catalog'::regnamespace;
IF NOT can_execute THEN
RAISE EXCEPTION 'neon_superuser cannot execute pg_show_replication_origin_status';
END IF;
END $$;

View File

@@ -47,7 +47,6 @@ pub enum PerDatabasePhase {
DeleteDBRoleReferences,
ChangeSchemaPerms,
HandleAnonExtension,
DropSubscriptionsForDeletedDatabases,
}
#[derive(Clone, Debug)]
@@ -75,7 +74,7 @@ pub struct MutableApplyContext {
pub dbs: HashMap<String, Database>,
}
/// Apply the operations that belong to the given spec apply phase.
/// Appply the operations that belong to the given spec apply phase.
///
/// Commands within a single phase are executed in order of Iterator yield.
/// Commands of ApplySpecPhase::RunInEachDatabase will execute in the database
@@ -327,12 +326,13 @@ async fn get_operations<'a>(
// Use FORCE to drop database even if there are active connections.
// We run this from `cloud_admin`, so it should have enough privileges.
//
// NB: there could be other db states, which prevent us from dropping
// the database. For example, if db is used by any active subscription
// or replication slot.
// Such cases are handled in the DropSubscriptionsForDeletedDatabases
// phase. We do all the cleanup before actually dropping the database.
// TODO: deal with it once we allow logical replication. Proper fix should
// involve returning an error code to the control plane, so it could
// figure out that this is a non-retryable error, return it to the user
// and fail operation permanently.
let drop_db_query: String = format!(
"DROP DATABASE IF EXISTS {} WITH (FORCE)",
&op.name.pg_quote()
@@ -444,30 +444,6 @@ async fn get_operations<'a>(
}
ApplySpecPhase::RunInEachDatabase { db, subphase } => {
match subphase {
PerDatabasePhase::DropSubscriptionsForDeletedDatabases => {
match &db {
DB::UserDB(db) => {
let drop_subscription_query: String = format!(
include_str!("sql/drop_subscription_for_drop_dbs.sql"),
datname_str = escape_literal(&db.name),
);
let operations = vec![Operation {
query: drop_subscription_query,
comment: Some(format!(
"optionally dropping subscriptions for DB {}",
db.name,
)),
}]
.into_iter();
Ok(Box::new(operations))
}
// skip this cleanup for the system databases
// because users can't drop them
DB::SystemDB => Ok(Box::new(empty())),
}
}
PerDatabasePhase::DeleteDBRoleReferences => {
let ctx = ctx.read().await;
@@ -498,19 +474,7 @@ async fn get_operations<'a>(
),
comment: None,
},
// Revoke some potentially blocking privileges (Neon-specific currently)
Operation {
query: format!(
include_str!("sql/pre_drop_role_revoke_privileges.sql"),
role_name = quoted,
),
comment: None,
},
// This now will only drop privileges of the role
// TODO: this is obviously not 100% true because of the above case,
// there could be still some privileges that are not revoked. Maybe this
// only drops privileges that were granted *by this* role, not *to this* role,
// but this has to be checked.
Operation {
query: format!("DROP OWNED BY {}", quoted),
comment: None,

View File

@@ -1,11 +0,0 @@
DO $$
DECLARE
subname TEXT;
BEGIN
FOR subname IN SELECT pg_subscription.subname FROM pg_subscription WHERE subdbid = (SELECT oid FROM pg_database WHERE datname = {datname_str}) LOOP
EXECUTE format('ALTER SUBSCRIPTION %I DISABLE;', subname);
EXECUTE format('ALTER SUBSCRIPTION %I SET (slot_name = NONE);', subname);
EXECUTE format('DROP SUBSCRIPTION %I;', subname);
END LOOP;
END;
$$;

View File

@@ -1,28 +0,0 @@
SET SESSION ROLE neon_superuser;
DO $$
DECLARE
schema TEXT;
revoke_query TEXT;
BEGIN
FOR schema IN
SELECT schema_name
FROM information_schema.schemata
-- So far, we only had issues with 'public' schema. Probably, because we do some additional grants,
-- e.g., make DB owner the owner of 'public' schema automatically (when created via API).
-- See https://github.com/neondatabase/cloud/issues/13582 for the context.
-- Still, keep the loop because i) it efficiently handles the case when there is no 'public' schema,
-- ii) it's easy to add more schemas to the list if needed.
WHERE schema_name IN ('public')
LOOP
revoke_query := format(
'REVOKE ALL PRIVILEGES ON ALL TABLES IN SCHEMA %I FROM {role_name} GRANTED BY neon_superuser;',
schema
);
EXECUTE revoke_query;
END LOOP;
END;
$$;
RESET ROLE;

View File

@@ -19,7 +19,6 @@ use control_plane::storage_controller::{
NeonStorageControllerStartArgs, NeonStorageControllerStopArgs, StorageController,
};
use control_plane::{broker, local_env};
use nix::fcntl::{flock, FlockArg};
use pageserver_api::config::{
DEFAULT_HTTP_LISTEN_PORT as DEFAULT_PAGESERVER_HTTP_PORT,
DEFAULT_PG_LISTEN_PORT as DEFAULT_PAGESERVER_PG_PORT,
@@ -37,8 +36,6 @@ use safekeeper_api::{
};
use std::borrow::Cow;
use std::collections::{BTreeSet, HashMap};
use std::fs::File;
use std::os::fd::AsRawFd;
use std::path::PathBuf;
use std::process::exit;
use std::str::FromStr;
@@ -692,21 +689,6 @@ struct TimelineTreeEl {
pub children: BTreeSet<TimelineId>,
}
/// A flock-based guard over the neon_local repository directory
struct RepoLock {
_file: File,
}
impl RepoLock {
fn new() -> Result<Self> {
let repo_dir = File::open(local_env::base_path())?;
let repo_dir_fd = repo_dir.as_raw_fd();
flock(repo_dir_fd, FlockArg::LockExclusive)?;
Ok(Self { _file: repo_dir })
}
}
// Main entry point for the 'neon_local' CLI utility
//
// This utility helps to manage neon installation. That includes following:
@@ -718,14 +700,9 @@ fn main() -> Result<()> {
let cli = Cli::parse();
// Check for 'neon init' command first.
let (subcommand_result, _lock) = if let NeonLocalCmd::Init(args) = cli.command {
(handle_init(&args).map(|env| Some(Cow::Owned(env))), None)
let subcommand_result = if let NeonLocalCmd::Init(args) = cli.command {
handle_init(&args).map(|env| Some(Cow::Owned(env)))
} else {
// This tool uses a collection of simple files to store its state, and consequently
// it is not generally safe to run multiple commands concurrently. Rather than expect
// all callers to know this, use a lock file to protect against concurrent execution.
let _repo_lock = RepoLock::new().unwrap();
// all other commands need an existing config
let env = LocalEnv::load_config(&local_env::base_path()).context("Error loading config")?;
let original_env = env.clone();
@@ -751,12 +728,11 @@ fn main() -> Result<()> {
NeonLocalCmd::Mappings(subcmd) => handle_mappings(&subcmd, env),
};
let subcommand_result = if &original_env != env {
if &original_env != env {
subcommand_result.map(|()| Some(Cow::Borrowed(env)))
} else {
subcommand_result.map(|()| None)
};
(subcommand_result, Some(_repo_lock))
}
};
match subcommand_result {
@@ -946,7 +922,7 @@ fn handle_init(args: &InitCmdArgs) -> anyhow::Result<LocalEnv> {
} else {
// User (likely interactive) did not provide a description of the environment, give them the default
NeonLocalInitConf {
control_plane_api: Some(DEFAULT_PAGESERVER_CONTROL_PLANE_API.parse().unwrap()),
control_plane_api: Some(Some(DEFAULT_PAGESERVER_CONTROL_PLANE_API.parse().unwrap())),
broker: NeonBroker {
listen_addr: DEFAULT_BROKER_ADDR.parse().unwrap(),
},
@@ -1742,15 +1718,18 @@ async fn handle_start_all_impl(
broker::start_broker_process(env, &retry_timeout).await
});
js.spawn(async move {
let storage_controller = StorageController::from_env(env);
storage_controller
.start(NeonStorageControllerStartArgs::with_default_instance_id(
retry_timeout,
))
.await
.map_err(|e| e.context("start storage_controller"))
});
// Only start the storage controller if the pageserver is configured to need it
if env.control_plane_api.is_some() {
js.spawn(async move {
let storage_controller = StorageController::from_env(env);
storage_controller
.start(NeonStorageControllerStartArgs::with_default_instance_id(
retry_timeout,
))
.await
.map_err(|e| e.context("start storage_controller"))
});
}
for ps_conf in &env.pageservers {
js.spawn(async move {
@@ -1795,6 +1774,10 @@ async fn neon_start_status_check(
const RETRY_INTERVAL: Duration = Duration::from_millis(100);
const NOTICE_AFTER_RETRIES: Duration = Duration::from_secs(5);
if env.control_plane_api.is_none() {
return Ok(());
}
let storcon = StorageController::from_env(env);
let retries = retry_timeout.as_millis() / RETRY_INTERVAL.as_millis();

View File

@@ -62,7 +62,7 @@ use crate::local_env::LocalEnv;
use crate::postgresql_conf::PostgresConf;
use crate::storage_controller::StorageController;
use compute_api::responses::{ComputeStatus, ComputeStatusResponse};
use compute_api::responses::{ComputeState, ComputeStatus};
use compute_api::spec::{Cluster, ComputeFeature, ComputeMode, ComputeSpec};
// contents of a endpoint.json file
@@ -316,10 +316,6 @@ impl Endpoint {
// and can cause errors like 'no unpinned buffers available', see
// <https://github.com/neondatabase/neon/issues/9956>
conf.append("shared_buffers", "1MB");
// Postgres defaults to effective_io_concurrency=1, which does not exercise the pageserver's
// batching logic. Set this to 2 so that we exercise the code a bit without letting
// individual tests do a lot of concurrent work on underpowered test machines
conf.append("effective_io_concurrency", "2");
conf.append("fsync", "off");
conf.append("max_connections", "100");
conf.append("wal_level", "logical");
@@ -585,7 +581,6 @@ impl Endpoint {
features: self.features.clone(),
swap_size_bytes: None,
disk_quota_bytes: None,
disable_lfc_resizing: None,
cluster: Cluster {
cluster_id: None, // project ID: not used
name: None, // project name: not used
@@ -739,7 +734,7 @@ impl Endpoint {
}
// Call the /status HTTP API
pub async fn get_status(&self) -> Result<ComputeStatusResponse> {
pub async fn get_status(&self) -> Result<ComputeState> {
let client = reqwest::Client::new();
let response = client

View File

@@ -76,7 +76,7 @@ pub struct LocalEnv {
// Control plane upcall API for pageserver: if None, we will not run storage_controller If set, this will
// be propagated into each pageserver's configuration.
pub control_plane_api: Url,
pub control_plane_api: Option<Url>,
// Control plane upcall API for storage controller. If set, this will be propagated into the
// storage controller's configuration.
@@ -133,7 +133,7 @@ pub struct NeonLocalInitConf {
pub storage_controller: Option<NeonStorageControllerConf>,
pub pageservers: Vec<NeonLocalInitPageserverConf>,
pub safekeepers: Vec<SafekeeperConf>,
pub control_plane_api: Option<Url>,
pub control_plane_api: Option<Option<Url>>,
pub control_plane_compute_hook_api: Option<Option<Url>>,
}
@@ -180,7 +180,7 @@ impl NeonStorageControllerConf {
const DEFAULT_MAX_WARMING_UP_INTERVAL: std::time::Duration = std::time::Duration::from_secs(30);
// Very tight heartbeat interval to speed up tests
const DEFAULT_HEARTBEAT_INTERVAL: std::time::Duration = std::time::Duration::from_millis(1000);
const DEFAULT_HEARTBEAT_INTERVAL: std::time::Duration = std::time::Duration::from_millis(100);
}
impl Default for NeonStorageControllerConf {
@@ -483,6 +483,7 @@ impl LocalEnv {
.iter()
.find(|(mapped_tenant_id, _)| mapped_tenant_id == &tenant_id)
.map(|&(_, timeline_id)| timeline_id)
.map(TimelineId::from)
}
pub fn timeline_name_mappings(&self) -> HashMap<TenantTimelineId, String> {
@@ -534,7 +535,7 @@ impl LocalEnv {
storage_controller,
pageservers,
safekeepers,
control_plane_api: control_plane_api.unwrap(),
control_plane_api,
control_plane_compute_hook_api,
branch_name_mappings,
}
@@ -637,7 +638,7 @@ impl LocalEnv {
storage_controller: self.storage_controller.clone(),
pageservers: vec![], // it's skip_serializing anyway
safekeepers: self.safekeepers.clone(),
control_plane_api: Some(self.control_plane_api.clone()),
control_plane_api: self.control_plane_api.clone(),
control_plane_compute_hook_api: self.control_plane_compute_hook_api.clone(),
branch_name_mappings: self.branch_name_mappings.clone(),
},
@@ -767,7 +768,7 @@ impl LocalEnv {
storage_controller: storage_controller.unwrap_or_default(),
pageservers: pageservers.iter().map(Into::into).collect(),
safekeepers,
control_plane_api: control_plane_api.unwrap(),
control_plane_api: control_plane_api.unwrap_or_default(),
control_plane_compute_hook_api: control_plane_compute_hook_api.unwrap_or_default(),
branch_name_mappings: Default::default(),
};

View File

@@ -95,19 +95,21 @@ impl PageServerNode {
let mut overrides = vec![pg_distrib_dir_param, broker_endpoint_param];
overrides.push(format!(
"control_plane_api='{}'",
self.env.control_plane_api.as_str()
));
if let Some(control_plane_api) = &self.env.control_plane_api {
overrides.push(format!(
"control_plane_api='{}'",
control_plane_api.as_str()
));
// Storage controller uses the same auth as pageserver: if JWT is enabled
// for us, we will also need it to talk to them.
if matches!(conf.http_auth_type, AuthType::NeonJWT) {
let jwt_token = self
.env
.generate_auth_token(&Claims::new(None, Scope::GenerationsApi))
.unwrap();
overrides.push(format!("control_plane_api_token='{}'", jwt_token));
// Storage controller uses the same auth as pageserver: if JWT is enabled
// for us, we will also need it to talk to them.
if matches!(conf.http_auth_type, AuthType::NeonJWT) {
let jwt_token = self
.env
.generate_auth_token(&Claims::new(None, Scope::GenerationsApi))
.unwrap();
overrides.push(format!("control_plane_api_token='{}'", jwt_token));
}
}
if !conf.other.contains_key("remote_storage") {
@@ -418,11 +420,6 @@ impl PageServerNode {
.map(serde_json::from_str)
.transpose()
.context("parse `wal_receiver_protocol_override` from json")?,
rel_size_v2_enabled: settings
.remove("rel_size_v2_enabled")
.map(|x| x.parse::<bool>())
.transpose()
.context("Failed to parse 'rel_size_v2_enabled' as bool")?,
};
if !settings.is_empty() {
bail!("Unrecognized tenant settings: {settings:?}")

View File

@@ -338,7 +338,7 @@ impl StorageController {
.port(),
)
} else {
let listen_url = self.env.control_plane_api.clone();
let listen_url = self.env.control_plane_api.clone().unwrap();
let listen = format!(
"{}:{}",
@@ -708,7 +708,7 @@ impl StorageController {
} else {
// The configured URL has the /upcall path prefix for pageservers to use: we will strip that out
// for general purpose API access.
let listen_url = self.env.control_plane_api.clone();
let listen_url = self.env.control_plane_api.clone().unwrap();
Url::from_str(&format!(
"http://{}:{}/{path}",
listen_url.host_str().unwrap(),
@@ -822,7 +822,10 @@ impl StorageController {
self.dispatch(
Method::PUT,
format!("control/v1/tenant/{tenant_shard_id}/migrate"),
Some(TenantShardMigrateRequest { node_id }),
Some(TenantShardMigrateRequest {
tenant_shard_id,
node_id,
}),
)
.await
}

View File

@@ -1,17 +1,11 @@
use futures::StreamExt;
use std::{
collections::{HashMap, HashSet},
str::FromStr,
time::Duration,
};
use std::{str::FromStr, time::Duration};
use clap::{Parser, Subcommand};
use pageserver_api::{
controller_api::{
AvailabilityZone, NodeAvailabilityWrapper, NodeDescribeResponse, NodeShardResponse,
SafekeeperDescribeResponse, SafekeeperSchedulingPolicyRequest, ShardSchedulingPolicy,
ShardsPreferredAzsRequest, SkSchedulingPolicy, TenantCreateRequest, TenantDescribeResponse,
TenantPolicyRequest,
ShardSchedulingPolicy, TenantCreateRequest, TenantDescribeResponse, TenantPolicyRequest,
},
models::{
EvictionPolicy, EvictionPolicyLayerAccessThreshold, LocationConfigSecondary,
@@ -117,13 +111,6 @@ enum Command {
#[arg(long)]
node: NodeId,
},
/// Migrate the secondary location for a tenant shard to a specific pageserver.
TenantShardMigrateSecondary {
#[arg(long)]
tenant_shard_id: TenantShardId,
#[arg(long)]
node: NodeId,
},
/// Cancel any ongoing reconciliation for this shard
TenantShardCancelReconcile {
#[arg(long)]
@@ -158,12 +145,6 @@ enum Command {
#[arg(long)]
tenant_id: TenantId,
},
TenantSetPreferredAz {
#[arg(long)]
tenant_id: TenantId,
#[arg(long)]
preferred_az: Option<String>,
},
/// Uncleanly drop a tenant from the storage controller: this doesn't delete anything from pageservers. Appropriate
/// if you e.g. used `tenant-warmup` by mistake on a tenant ID that doesn't really exist, or is in some other region.
TenantDrop {
@@ -230,15 +211,6 @@ enum Command {
#[arg(long)]
timeout: humantime::Duration,
},
/// List safekeepers known to the storage controller
Safekeepers {},
/// Set the scheduling policy of the specified safekeeper
SafekeeperScheduling {
#[arg(long)]
node_id: NodeId,
#[arg(long)]
scheduling_policy: SkSchedulingPolicyArg,
},
}
#[derive(Parser)]
@@ -291,17 +263,6 @@ impl FromStr for PlacementPolicyArg {
}
}
#[derive(Debug, Clone)]
struct SkSchedulingPolicyArg(SkSchedulingPolicy);
impl FromStr for SkSchedulingPolicyArg {
type Err = anyhow::Error;
fn from_str(s: &str) -> Result<Self, Self::Err> {
SkSchedulingPolicy::from_str(s).map(Self)
}
}
#[derive(Debug, Clone)]
struct ShardSchedulingPolicyArg(ShardSchedulingPolicy);
@@ -431,12 +392,11 @@ async fn main() -> anyhow::Result<()> {
resp.sort_by(|a, b| a.listen_http_addr.cmp(&b.listen_http_addr));
let mut table = comfy_table::Table::new();
table.set_header(["Id", "Hostname", "AZ", "Scheduling", "Availability"]);
table.set_header(["Id", "Hostname", "Scheduling", "Availability"]);
for node in resp {
table.add_row([
format!("{}", node.id),
node.listen_http_addr,
node.availability_zone_id,
format!("{:?}", node.scheduling),
format!("{:?}", node.availability),
]);
@@ -496,65 +456,33 @@ async fn main() -> anyhow::Result<()> {
println!("{table}");
}
Command::Tenants { node_id: None } => {
// Set up output formatting
let mut resp = storcon_client
.dispatch::<(), Vec<TenantDescribeResponse>>(
Method::GET,
"control/v1/tenant".to_string(),
None,
)
.await?;
resp.sort_by(|a, b| a.tenant_id.cmp(&b.tenant_id));
let mut table = comfy_table::Table::new();
table.set_header([
"TenantId",
"Preferred AZ",
"ShardCount",
"StripeSize",
"Placement",
"Scheduling",
]);
// Pagination loop over listing API
let mut start_after = None;
const LIMIT: usize = 1000;
loop {
let path = match start_after {
None => format!("control/v1/tenant?limit={LIMIT}"),
Some(start_after) => {
format!("control/v1/tenant?limit={LIMIT}&start_after={start_after}")
}
};
let resp = storcon_client
.dispatch::<(), Vec<TenantDescribeResponse>>(Method::GET, path, None)
.await?;
if resp.is_empty() {
// End of data reached
break;
}
// Give some visual feedback while we're building up the table (comfy_table doesn't have
// streaming output)
if resp.len() >= LIMIT {
eprint!(".");
}
start_after = Some(resp.last().unwrap().tenant_id);
for tenant in resp {
let shard_zero = tenant.shards.into_iter().next().unwrap();
table.add_row([
format!("{}", tenant.tenant_id),
shard_zero
.preferred_az_id
.as_ref()
.cloned()
.unwrap_or("".to_string()),
format!("{}", shard_zero.tenant_shard_id.shard_count.literal()),
format!("{:?}", tenant.stripe_size),
format!("{:?}", tenant.policy),
format!("{:?}", shard_zero.scheduling_policy),
]);
}
}
// Terminate progress dots
if table.row_count() > LIMIT {
eprint!("");
for tenant in resp {
let shard_zero = tenant.shards.into_iter().next().unwrap();
table.add_row([
format!("{}", tenant.tenant_id),
format!("{}", shard_zero.tenant_shard_id.shard_count.literal()),
format!("{:?}", tenant.stripe_size),
format!("{:?}", tenant.policy),
format!("{:?}", shard_zero.scheduling_policy),
]);
}
println!("{table}");
@@ -609,7 +537,10 @@ async fn main() -> anyhow::Result<()> {
tenant_shard_id,
node,
} => {
let req = TenantShardMigrateRequest { node_id: node };
let req = TenantShardMigrateRequest {
tenant_shard_id,
node_id: node,
};
storcon_client
.dispatch::<TenantShardMigrateRequest, TenantShardMigrateResponse>(
@@ -619,20 +550,6 @@ async fn main() -> anyhow::Result<()> {
)
.await?;
}
Command::TenantShardMigrateSecondary {
tenant_shard_id,
node,
} => {
let req = TenantShardMigrateRequest { node_id: node };
storcon_client
.dispatch::<TenantShardMigrateRequest, TenantShardMigrateResponse>(
Method::PUT,
format!("control/v1/tenant/{tenant_shard_id}/migrate_secondary"),
Some(req),
)
.await?;
}
Command::TenantShardCancelReconcile { tenant_shard_id } => {
storcon_client
.dispatch::<(), ()>(
@@ -676,19 +593,6 @@ async fn main() -> anyhow::Result<()> {
None,
)
.await?;
let nodes = storcon_client
.dispatch::<(), Vec<NodeDescribeResponse>>(
Method::GET,
"control/v1/node".to_string(),
None,
)
.await?;
let nodes = nodes
.into_iter()
.map(|n| (n.id, n))
.collect::<HashMap<_, _>>();
println!("Tenant {tenant_id}");
let mut table = comfy_table::Table::new();
table.add_row(["Policy", &format!("{:?}", policy)]);
@@ -697,14 +601,7 @@ async fn main() -> anyhow::Result<()> {
println!("{table}");
println!("Shards:");
let mut table = comfy_table::Table::new();
table.set_header([
"Shard",
"Attached",
"Attached AZ",
"Secondary",
"Last error",
"status",
]);
table.set_header(["Shard", "Attached", "Secondary", "Last error", "status"]);
for shard in shards {
let secondary = shard
.node_secondary
@@ -727,18 +624,11 @@ async fn main() -> anyhow::Result<()> {
}
let status = status_parts.join(",");
let attached_node = shard
.node_attached
.as_ref()
.map(|id| nodes.get(id).expect("Shard references nonexistent node"));
table.add_row([
format!("{}", shard.tenant_shard_id),
attached_node
.map(|n| format!("{} ({})", n.listen_http_addr, n.id))
.unwrap_or(String::new()),
attached_node
.map(|n| n.availability_zone_id.clone())
shard
.node_attached
.map(|n| format!("{}", n))
.unwrap_or(String::new()),
secondary,
shard.last_error,
@@ -747,66 +637,6 @@ async fn main() -> anyhow::Result<()> {
}
println!("{table}");
}
Command::TenantSetPreferredAz {
tenant_id,
preferred_az,
} => {
// First learn about the tenant's shards
let describe_response = storcon_client
.dispatch::<(), TenantDescribeResponse>(
Method::GET,
format!("control/v1/tenant/{tenant_id}"),
None,
)
.await?;
// Learn about nodes to validate the AZ ID
let nodes = storcon_client
.dispatch::<(), Vec<NodeDescribeResponse>>(
Method::GET,
"control/v1/node".to_string(),
None,
)
.await?;
if let Some(preferred_az) = &preferred_az {
let azs = nodes
.into_iter()
.map(|n| (n.availability_zone_id))
.collect::<HashSet<_>>();
if !azs.contains(preferred_az) {
anyhow::bail!(
"AZ {} not found on any node: known AZs are: {:?}",
preferred_az,
azs
);
}
} else {
// Make it obvious to the user that since they've omitted an AZ, we're clearing it
eprintln!("Clearing preferred AZ for tenant {}", tenant_id);
}
// Construct a request that modifies all the tenant's shards
let req = ShardsPreferredAzsRequest {
preferred_az_ids: describe_response
.shards
.into_iter()
.map(|s| {
(
s.tenant_shard_id,
preferred_az.clone().map(AvailabilityZone),
)
})
.collect(),
};
storcon_client
.dispatch::<ShardsPreferredAzsRequest, ()>(
Method::PUT,
"control/v1/preferred_azs".to_string(),
Some(req),
)
.await?;
}
Command::TenantWarmup { tenant_id } => {
let describe_response = storcon_client
.dispatch::<(), TenantDescribeResponse>(
@@ -1082,7 +912,10 @@ 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 {
tenant_shard_id: mv.tenant_shard_id,
node_id: mv.to,
}),
)
.await
.map_err(|e| (mv.tenant_shard_id, mv.from, mv.to, e))
@@ -1187,57 +1020,6 @@ async fn main() -> anyhow::Result<()> {
"Fill was cancelled for node {node_id}. Schedulling policy is now {final_policy:?}"
);
}
Command::Safekeepers {} => {
let mut resp = storcon_client
.dispatch::<(), Vec<SafekeeperDescribeResponse>>(
Method::GET,
"control/v1/safekeeper".to_string(),
None,
)
.await?;
resp.sort_by(|a, b| a.id.cmp(&b.id));
let mut table = comfy_table::Table::new();
table.set_header([
"Id",
"Version",
"Host",
"Port",
"Http Port",
"AZ Id",
"Scheduling",
]);
for sk in resp {
table.add_row([
format!("{}", sk.id),
format!("{}", sk.version),
sk.host,
format!("{}", sk.port),
format!("{}", sk.http_port),
sk.availability_zone_id.clone(),
String::from(sk.scheduling_policy),
]);
}
println!("{table}");
}
Command::SafekeeperScheduling {
node_id,
scheduling_policy,
} => {
let scheduling_policy = scheduling_policy.0;
storcon_client
.dispatch::<SafekeeperSchedulingPolicyRequest, ()>(
Method::POST,
format!("control/v1/safekeeper/{node_id}/scheduling_policy"),
Some(SafekeeperSchedulingPolicyRequest { scheduling_policy }),
)
.await?;
println!(
"Scheduling policy of {node_id} set to {}",
String::from(scheduling_policy)
);
}
}
Ok(())

View File

@@ -132,6 +132,11 @@
"name": "cron.database",
"value": "postgres",
"vartype": "string"
},
{
"name": "session_preload_libraries",
"value": "anon",
"vartype": "string"
}
]
},

View File

@@ -35,11 +35,11 @@ for pg_version in ${TEST_VERSION_ONLY-14 15 16 17}; do
echo "clean up containers if exists"
cleanup
PG_TEST_VERSION=$((pg_version < 16 ? 16 : pg_version))
# The support of pg_anon not yet added to PG17, so we have to add the corresponding option for other PG versions
if [ "${pg_version}" -ne 17 ]; then
# The support of pg_anon not yet added to PG17, so we have to remove the corresponding option
if [ $pg_version -eq 17 ]; then
SPEC_PATH="compute_wrapper/var/db/postgres/specs"
mv $SPEC_PATH/spec.json $SPEC_PATH/spec.bak
jq '.cluster.settings += [{"name": "session_preload_libraries","value": "anon","vartype": "string"}]' "${SPEC_PATH}/spec.bak" > "${SPEC_PATH}/spec.json"
jq 'del(.cluster.settings[] | select (.name == "session_preload_libraries"))' $SPEC_PATH/spec.bak > $SPEC_PATH/spec.json
fi
PG_VERSION=$pg_version PG_TEST_VERSION=$PG_TEST_VERSION docker compose --profile test-extensions -f $COMPOSE_FILE up --build -d
@@ -106,8 +106,8 @@ for pg_version in ${TEST_VERSION_ONLY-14 15 16 17}; do
fi
fi
cleanup
# Restore the original spec.json
if [ "$pg_version" -ne 17 ]; then
mv "$SPEC_PATH/spec.bak" "$SPEC_PATH/spec.json"
# The support of pg_anon not yet added to PG17, so we have to remove the corresponding option
if [ $pg_version -eq 17 ]; then
mv $SPEC_PATH/spec.bak $SPEC_PATH/spec.json
fi
done

View File

@@ -7,11 +7,15 @@ Currently we build two main images:
- [neondatabase/neon](https://hub.docker.com/repository/docker/neondatabase/neon) — image with pre-built `pageserver`, `safekeeper` and `proxy` binaries and all the required runtime dependencies. Built from [/Dockerfile](/Dockerfile).
- [neondatabase/compute-node-v16](https://hub.docker.com/repository/docker/neondatabase/compute-node-v16) — compute node image with pre-built Postgres binaries from [neondatabase/postgres](https://github.com/neondatabase/postgres). Similar images exist for v15 and v14. Built from [/compute-node/Dockerfile](/compute/compute-node.Dockerfile).
And additional intermediate image:
- [neondatabase/compute-tools](https://hub.docker.com/repository/docker/neondatabase/compute-tools) — compute node configuration management tools.
## Build pipeline
We build all images after a successful `release` tests run and push automatically to Docker Hub with two parallel CI jobs
1. `neondatabase/compute-node-v17` (and -16, -v15, -v14)
1. `neondatabase/compute-tools` and `neondatabase/compute-node-v16` (and -v15 and -v14)
2. `neondatabase/neon`

View File

@@ -81,7 +81,7 @@ configuration generation in them is less than its current one. Namely, it
refuses to vote, to truncate WAL in `handle_elected` and to accept WAL. In
response it sends its current configuration generation to let walproposer know.
Safekeeper gets `PUT /v1/tenants/{tenant_id}/timelines/{timeline_id}/configuration`
Safekeeper gets `PUT /v1/tenants/{tenant_id}/timelines/{timeline_id}/configuration`
accepting `Configuration`. Safekeeper switches to the given conf it is higher than its
current one and ignores it otherwise. In any case it replies with
```
@@ -103,7 +103,7 @@ currently and tries to communicate with all of them. However, the list does not
define consensus members. Instead, on start walproposer tracks highest
configuration it receives from `AcceptorGreeting`s. Once it assembles greetings
from majority of `sk_set` and majority of `new_sk_set` (if it is present), it
establishes this configuration as its own and moves to voting.
establishes this configuration as its own and moves to voting.
It should stop talking to safekeepers not listed in the configuration at this
point, though it is not unsafe to continue doing so.
@@ -119,7 +119,7 @@ refusal to accept due to configuration change) it simply restarts.
The following algorithm can be executed anywhere having access to configuration
storage and safekeepers. It is safe to interrupt / restart it and run multiple
instances of it concurrently, though likely one of them won't make
progress then. It accepts `desired_set: Vec<NodeId>` as input.
progress then. It accepts `desired_set: Vec<NodeId>` as input.
Algorithm will refuse to make the change if it encounters previous interrupted
change attempt, but in this case it will try to finish it.
@@ -140,7 +140,7 @@ storage are reachable.
safe. Failed CAS aborts the procedure.
4) Call `PUT` `configuration` on safekeepers from the current set,
delivering them `joint_conf`. Collecting responses from majority is required
to proceed. If any response returned generation higher than
to proceed. If any response returned generation higher than
`joint_conf.generation`, abort (another switch raced us). Otherwise, choose
max `<last_log_term, flush_lsn>` among responses and establish it as
(in memory) `sync_position`. Also choose max `term` and establish it as (in
@@ -149,49 +149,49 @@ storage are reachable.
without ack from the new set. Similarly, we'll bump term on new majority
to `sync_term` so that two computes with the same term are never elected.
4) Initialize timeline on safekeeper(s) from `new_sk_set` where it
doesn't exist yet by doing `pull_timeline` from the majority of the
doesn't exist yet by doing `pull_timeline` from the majority of the
current set. Doing that on majority of `new_sk_set` is enough to
proceed, but it is reasonable to ensure that all `new_sk_set` members
are initialized -- if some of them are down why are we migrating there?
5) Call `POST` `bump_term(sync_term)` on safekeepers from the new set.
5) Call `POST` `bump_term(sync_term)` on safekeepers from the new set.
Success on majority is enough.
6) Repeatedly call `PUT` `configuration` on safekeepers from the new set,
delivering them `joint_conf` and collecting their positions. This will
switch them to the `joint_conf` which generally won't be needed
switch them to the `joint_conf` which generally won't be needed
because `pull_timeline` already includes it and plus additionally would be
broadcast by compute. More importantly, we may proceed to the next step
only when `<last_log_term, flush_lsn>` on the majority of the new set reached
`sync_position`. Similarly, on the happy path no waiting is not needed because
only when `<last_log_term, flush_lsn>` on the majority of the new set reached
`sync_position`. Similarly, on the happy path no waiting is not needed because
`pull_timeline` already includes it. However, we should double
check to be safe. For example, timeline could have been created earlier e.g.
manually or after try-to-migrate, abort, try-to-migrate-again sequence.
7) Create `new_conf: Configuration` incrementing `join_conf` generation and having new
safekeeper set as `sk_set` and None `new_sk_set`. Write it to configuration
manually or after try-to-migrate, abort, try-to-migrate-again sequence.
7) Create `new_conf: Configuration` incrementing `join_conf` generation and having new
safekeeper set as `sk_set` and None `new_sk_set`. Write it to configuration
storage under one more CAS.
8) Call `PUT` `configuration` on safekeepers from the new set,
delivering them `new_conf`. It is enough to deliver it to the majority
delivering them `new_conf`. It is enough to deliver it to the majority
of the new set; the rest can be updated by compute.
I haven't put huge effort to make the description above very precise, because it
is natural language prone to interpretations anyway. Instead I'd like to make TLA+
spec of it.
Description above focuses on safety. To make the flow practical and live, here a few more
Description above focuses on safety. To make the flow practical and live, here a few more
considerations.
1) It makes sense to ping new set to ensure it we are migrating to live node(s) before
1) It makes sense to ping new set to ensure it we are migrating to live node(s) before
step 3.
2) If e.g. accidentally wrong new sk set has been specified, before CAS in step `6` is completed
2) If e.g. accidentally wrong new sk set has been specified, before CAS in step `6` is completed
it is safe to rollback to the old conf with one more CAS.
3) On step 4 timeline might be already created on members of the new set for various reasons;
3) On step 4 timeline might be already created on members of the new set for various reasons;
the simplest is the procedure restart. There are more complicated scenarious like mentioned
in step 5. Deleting and re-doing `pull_timeline` is generally unsafe without involving
generations, so seems simpler to treat existing timeline as success. However, this also
in step 5. Deleting and re-doing `pull_timeline` is generally unsafe without involving
generations, so seems simpler to treat existing timeline as success. However, this also
has a disadvantage: you might imagine an surpassingly unlikely schedule where condition in
the step 5 is never reached until compute is (re)awaken up to synchronize new member(s).
I don't think we'll observe this in practice, but can add waking up compute if needed.
4) In the end timeline should be locally deleted on the safekeeper(s) which are
in the old set but not in the new one, unless they are unreachable. To be
safe this also should be done under generation number (deletion proceeds only if
safe this also should be done under generation number (deletion proceeds only if
current configuration is <= than one in request and safekeeper is not memeber of it).
5) If current conf fetched on step 1 is already not joint and members equal to `desired_set`,
jump to step 7, using it as `new_conf`.
@@ -202,87 +202,47 @@ The procedure ought to be driven from somewhere. Obvious candidates are control
plane and storage_controller; and as each of them already has db we don't want
yet another storage. I propose to manage safekeepers in storage_controller
because 1) since it is in rust it simplifies simulation testing (more on this
below) 2) it already manages pageservers.
below) 2) it already manages pageservers.
This assumes that migration will be fully usable only after we migrate all
tenants/timelines to storage_controller. It is discussible whether we want also
to manage pageserver attachments for all of these, but likely we do.
This requires us to define storcon <-> cplane interface and changes.
This requires us to define storcon <-> cplane interface.
### storage_controller <-> control plane interface and changes
### storage_controller <-> control plane interface
First of all, control plane should
[change](https://neondb.slack.com/archives/C03438W3FLZ/p1719226543199829)
storing safekeepers per timeline instead of per tenant because we can't migrate
tenants atomically.
tenants atomically.
The important question is how updated configuration is delivered from
storage_controller to control plane to provide it to computes. As always, there
are two options, pull and push. Let's do it the same push as with pageserver
`/notify-attach` because 1) it keeps storage_controller out of critical compute
start path 2) uniformity. It makes storage_controller responsible for retrying
notifying control plane until it succeeds.
start path 2) provides easier upgrade: there won't be such a thing as 'timeline
managed by control plane / storcon', cplane just takes the value out of its db
when needed 3) uniformity. It makes storage_controller responsible for retrying notifying
control plane until it succeeds.
It is not needed for the control plane to fully know the `Configuration`. It is
enough for it to only to be aware of the list of safekeepers in the latest
configuration to supply it to compute, plus associated generation number to
protect from stale update requests and to also pass it to compute.
So, cplane `/notify-safekeepers` for the timeline can accept JSON like
```
{
tenant_id: String,
timeline_id: String,
generation: u32,
safekeepers: Vec<SafekeeperId>,
}
```
where `SafekeeperId` is
```
{
node_id: u64,
host: String
}
```
In principle `host` is redundant, but may be useful for observability.
The request updates list of safekeepers in the db if the provided conf
generation is higher (the cplane db should also store generations for this).
Similarly to
[`/notify-attach`](https://www.notion.so/neondatabase/Storage-Controller-Control-Plane-interface-6de56dd310a043bfa5c2f5564fa98365),
it should update db which makes the call successful, and then try to schedule
`apply_config` if possible, it is ok if not. storage_controller should rate
limit calling the endpoint, but likely this won't be needed, as migration
So, cplane `/notify-safekeepers` for the timeline accepts `Configuration` and
updates it in the db if the provided conf generation is higher (the cplane db
should also store generations for this). Similarly to [`/notify-attach`](https://www.notion.so/neondatabase/Storage-Controller-Control-Plane-interface-6de56dd310a043bfa5c2f5564fa98365), it
should update db which makes the call successful, and then try to schedule
`apply_config` if possible, it is ok if not. storage_controller
should rate limit calling the endpoint, but likely this won't be needed, as migration
throughput is limited by `pull_timeline`.
Timeline (branch) creation in cplane should call storage_controller POST
`tenant/:tenant_id/timeline` like it currently does for sharded tenants.
Response should be augmented with `safekeepers_generation` and `safekeepers`
fields like described in `/notify-safekeepers` above. Initially (currently)
these fields may be absent; in this case cplane chooses safekeepers on its own
like it currently does. The call should be retried until succeeds.
Response should be augmented with `safekeeper_conf: Configuration`. The call
should be retried until succeeds.
Timeline deletion and tenant deletion in cplane should call appropriate
storage_controller endpoints like it currently does for sharded tenants. The
calls should be retried until they succeed.
When compute receives safekeepers list from control plane it needs to know the
generation to checked whether it should be updated (note that compute may get
safekeeper list from either cplane or safekeepers). Currently `neon.safekeepers`
GUC is just a comma separates list of `host:port`. Let's prefix it with
`g#<generation>:` to this end, so it will look like
```
g#42:safekeeper-0.eu-central-1.aws.neon.tech:6401,safekeeper-2.eu-central-1.aws.neon.tech:6401,safekeeper-1.eu-central-1.aws.neon.tech:6401
```
To summarize, list of cplane changes:
- per tenant -> per timeline safekeepers management and addition of int `safekeeper_generation` field.
- `/notify-safekeepers` endpoint.
- Branch creation call may return list of safekeepers and when it is
present cplane should adopt it instead of choosing on its own like it does currently.
- `neon.safekeepers` GUC should be prefixed with `g#<generation>:`.
### storage_controller implementation
Current 'load everything on startup and keep in memory' easy design is fine.
@@ -400,10 +360,10 @@ 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
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
@@ -452,8 +412,8 @@ There should be following layers of tests:
3) Since simulation testing injects at relatively high level points (not
syscalls), it omits some code, in particular `pull_timeline`. Thus it is
better to have basic tests covering whole system as well. Extended version of
`test_restarts_under_load` would do: start background load and do migration
under it, then restart endpoint and check that no reported commits
`test_restarts_under_load` would do: start background load and do migration
under it, then restart endpoint and check that no reported commits
had been lost. I'd also add one more creating classic network split scenario, with
one compute talking to AC and another to BD while migration from nodes ABC to ABD
happens.
@@ -462,51 +422,35 @@ There should be following layers of tests:
## Order of implementation and rollout
Note that
Note that
- Control plane parts and integration with it is fully independent from everything else
(tests would use simulation and neon_local).
- It is reasonable to make compute <-> safekeepers protocol change
independent of enabling generations.
- There is a lot of infra work making storage_controller aware of timelines and safekeepers
and its impl/rollout should be separate from migration itself.
- Initially walproposer can just stop working while it observes joint configuration.
- Initially walproposer can just stop working while it observers joint configuration.
Such window would be typically very short anyway.
- Obviously we want to test the whole thing thoroughly on staging and only then
gradually enable in prod.
Let's have the following implementation bits for gradual rollout:
- compute gets `neon.safekeepers_proto_version` flag.
Initially both compute and safekeepers will be able to talk both
versions so that we can delay force restart of them and for
simplicity of rollback in case it is needed.
- storcon gets `-set-safekeepers` config option disabled by
default. Timeline creation request chooses safekeepers
(and returns them in response to cplane) only when it is set to
true.
- control_plane [see above](storage_controller-<->-control-plane interface-and-changes)
prefixes `neon.safekeepers` GUC with generation number. When it is 0
(or prefix not present at all), walproposer behaves as currently, committing on
the provided safekeeper list -- generations are disabled.
If it is non 0 it follows this RFC rules.
- We provide a script for manual migration to storage controller.
It selects timeline(s) from control plane (specified or all of them) db
and calls special import endpoint on storage controller which is very
similar to timeline creation: it inserts into the db, sets
configuration to initial on the safekeepers, calls cplane
`notify-safekeepers`.
To rollout smoothly, both walproposer and safekeeper should have flag
`configurations_enabled`; when set to false, they would work as currently, i.e.
walproposer is able to commit on whatever safekeeper set it is provided. Until
all timelines are managed by storcon we'd need to use current script to migrate
and update/drop entries in the storage_controller database if it has any.
Then the rollout for a region would be:
- Current situation: safekeepers are choosen by control_plane.
- We manually migrate some timelines, test moving them around.
- Then we enable `--set-safekeepers` so that all new timelines
are on storage controller.
- Finally migrate all existing timelines using the script (no
compute should be speaking old proto version at this point).
Safekeepers would need to be able to talk both current and new protocol version
with compute to reduce number of computes restarted in prod once v2 protocol is
deployed (though before completely switching we'd need to force this).
Let's have the following rollout order:
- storage_controller becomes aware of safekeepers;
- storage_controller gets timeline creation for new timelines and deletion requests, but
doesn't manage all timelines yet. Migration can be tested on these new timelines.
To keep control plane and storage_controller databases in sync while control
plane still chooses the safekeepers initially (until all timelines are imported
it can choose better), `TimelineCreateRequest` can get optional safekeepers
field with safekeepers chosen by cplane.
- Then we can import all existing timelines from control plane to
storage_controller and gradually enable configurations region by region.
Until all timelines are managed by storcon we'd need to use current ad hoc
script to migrate if needed. To keep state clean, all storage controller managed
timelines must be migrated before that, or controller db and configurations
state of safekeepers dropped manually.
Very rough implementation order:
- Add concept of configurations to safekeepers (including control file),
@@ -514,10 +458,10 @@ Very rough implementation order:
- Implement walproposer changes, including protocol.
- Implement storconn part. Use it in neon_local (and pytest).
- Make cplane store safekeepers per timeline instead of per tenant.
- Implement cplane/storcon integration. Route branch creation/deletion
- Implement cplane/storcon integration. Route branch creation/deletion
through storcon. Then we can test migration of new branches.
- Finally import existing branches. Then we can drop cplane
safekeeper selection code. Gradually enable configurations at
- Finally import existing branches. Then we can drop cplane
safekeeper selection code. Gradually enable configurations at
computes and safekeepers. Before that, all computes must talk only
v3 protocol version.

View File

@@ -1,247 +0,0 @@
# CPU and Memory Profiling
Created 2025-01-12 by Erik Grinaker.
See also [internal user guide](https://www.notion.so/neondatabase/Storage-CPU-Memory-Profiling-14bf189e004780228ec7d04442742324?pvs=4).
## Summary
This document proposes a standard cross-team pattern for CPU and memory profiling across
applications and languages, using the [pprof](https://github.com/google/pprof) profile format.
It enables both ad hoc profiles via HTTP endpoints, and continuous profiling across the fleet via
[Grafana Cloud Profiles](https://grafana.com/docs/grafana-cloud/monitor-applications/profiles/).
Continuous profiling incurs an overhead of about 0.1% CPU usage and 3% slower heap allocations.
## Motivation
CPU and memory profiles are crucial observability tools for understanding performance issues,
resource exhaustion, and resource costs. They allow answering questions like:
* Why is this process using 100% CPU?
* How do I make this go faster?
* Why did this process run out of memory?
* Why are we paying for all these CPU cores and memory chips?
Go has [first-class support](https://pkg.go.dev/net/http/pprof) for profiling included in its
standard library, using the [pprof](https://github.com/google/pprof) profile format and associated
tooling.
This is not the case for Rust and C, where obtaining profiles can be rather cumbersome. It requires
installing and running additional tools like `perf` as root on production nodes, with analysis tools
that can be hard to use and often don't give good results. This is not only annoying, but can also
significantly affect the resolution time of production incidents.
This proposal will:
* Provide CPU and heap profiles in pprof format via HTTP API.
* Record continuous profiles in Grafana for aggregate historical analysis.
* Make it easy for anyone to see a flamegraph in less than one minute.
* Be reasonably consistent across teams and services (Rust, Go, C).
## Non Goals (For Now)
* [Additional profile types](https://grafana.com/docs/pyroscope/next/configure-client/profile-types/)
like mutexes, locks, goroutines, etc.
* [Runtime trace integration](https://grafana.com/docs/pyroscope/next/configure-client/trace-span-profiles/).
* [Profile-guided optimization](https://en.wikipedia.org/wiki/Profile-guided_optimization).
## Using Profiles
Ready-to-use profiles can be obtained using e.g. `curl`. For Rust services:
```
$ curl localhost:9898/profile/cpu >profile.pb.gz
```
pprof profiles can be explored using the [`pprof`](https://github.com/google/pprof) web UI, which
provides flamegraphs, call graphs, plain text listings, and more:
```
$ pprof -http :6060 <profile>
```
Some endpoints (e.g. Rust-based ones) can also generate flamegraph SVGs directly:
```
$ curl localhost:9898/profile/cpu?format=svg >profile.svg
$ open profile.svg
```
Continuous profiles are available in Grafana under Explore → Profiles → Explore Profiles
(currently only in [staging](https://neonstaging.grafana.net/a/grafana-pyroscope-app/profiles-explorer)).
## API Requirements
* HTTP endpoints that return a profile in pprof format (with symbols).
* CPU: records a profile over the request time interval (`seconds` query parameter).
* Memory: returns the current in-use heap allocations.
* Unauthenticated, as it should not expose user data or pose a denial-of-service risk.
* Default sample frequency should not impact service (maximum 5% CPU overhead).
* Linux-compatibility.
Nice to have:
* Return flamegraph SVG directly from the HTTP endpoint if requested.
* Configurable sample frequency for CPU profiles.
* Historical heap allocations, by count and bytes.
* macOS-compatiblity.
## Rust Profiling
[`libs/utils/src/http/endpoint.rs`](https://github.com/neondatabase/neon/blob/8327f68043e692c77f70d6a6dafa463636c01578/libs/utils/src/http/endpoint.rs)
contains ready-to-use HTTP endpoints for CPU and memory profiling:
[`profile_cpu_handler`](https://github.com/neondatabase/neon/blob/8327f68043e692c77f70d6a6dafa463636c01578/libs/utils/src/http/endpoint.rs#L338) and [`profile_heap_handler`](https://github.com/neondatabase/neon/blob/8327f68043e692c77f70d6a6dafa463636c01578/libs/utils/src/http/endpoint.rs#L416).
### CPU
CPU profiles are provided by [pprof-rs](https://github.com/tikv/pprof-rs) via
[`profile_cpu_handler`](https://github.com/neondatabase/neon/blob/8327f68043e692c77f70d6a6dafa463636c01578/libs/utils/src/http/endpoint.rs#L338).
Expose it unauthenticated at `/profile/cpu`.
Parameters:
* `format`: profile output format (`pprof` or `svg`; default `pprof`).
* `seconds`: duration to collect profile over, in seconds (default `5`).
* `frequency`: how often to sample thread stacks, in Hz (default `99`).
* `force`: if `true`, cancel a running profile and start a new one (default `false`).
Works on Linux and macOS.
### Memory
Use the jemalloc allocator via [`tikv-jemallocator`](https://github.com/tikv/jemallocator),
and enable profiling with samples every 2 MB allocated:
```rust
#[global_allocator]
static GLOBAL: tikv_jemallocator::Jemalloc = tikv_jemallocator::Jemalloc;
#[allow(non_upper_case_globals)]
#[export_name = "malloc_conf"]
pub static malloc_conf: &[u8] = b"prof:true,prof_active:true,lg_prof_sample:21\0";
```
pprof profiles are generated by
[`jemalloc-pprof`](https://github.com/polarsignals/rust-jemalloc-pprof) via
[`profile_heap_handler`](https://github.com/neondatabase/neon/blob/8327f68043e692c77f70d6a6dafa463636c01578/libs/utils/src/http/endpoint.rs#L416).
Expose it unauthenticated at `/profile/heap`.
Parameters:
* `format`: profile output format (`pprof`, `svg`, or `jemalloc`; default `pprof`).
Works on Linux only, due to [jemalloc limitations](https://github.com/jemalloc/jemalloc/issues/26).
## Go Profiling
The Go standard library includes pprof profiling via HTTP API in
[`net/http/pprof`](https://pkg.go.dev/net/http/pprof). Expose it unauthenticated at
`/debug/pprof`.
Works on Linux and macOS.
### CPU
Via `/debug/pprof/profile`. Parameters:
* `debug`: profile output format (`0` is pprof, `1` or above is plaintext; default `0`).
* `seconds`: duration to collect profile over, in seconds (default `30`).
Does not support a frequency parameter (see [#57488](https://github.com/golang/go/issues/57488)),
and defaults to 100 Hz. A lower frequency can be hardcoded via `SetCPUProfileRate`, but the default
is likely ok (estimated 1% overhead).
### Memory
Via `/debug/pprof/heap`. Parameters:
* `seconds`: take a delta profile over the given duration, in seconds (default `0`).
* `gc`: if `1`, garbage collect before taking profile.
## C Profiling
[gperftools](https://github.com/gperftools/gperftools) provides in-process CPU and heap profiling
with pprof output.
However, continuous profiling of PostgreSQL is expensive (many computes), and has limited value
since we don't own the internals anyway.
Ad hoc profiling might still be useful, but the compute team considers existing tooling sufficient,
so this is not a priority at the moment.
## Grafana Continuous Profiling
[Grafana Alloy](https://grafana.com/docs/alloy/latest/) continually scrapes CPU and memory profiles
across the fleet, and archives them as time series. This can be used to analyze resource usage over
time, either in aggregate or zoomed in to specific events and nodes.
Profiles are retained for 30 days. Profile ingestion volume for CPU+heap at 60-second intervals
is about 0.5 GB/node/day, or about $0.25/node/day = $7.5/node/month ($0.50/GB).
It is currently enabled in [staging](https://neonstaging.grafana.net/a/grafana-pyroscope-app/profiles-explorer)
for Pageserver and Safekeeper.
### Scraping
* CPU profiling: 59 seconds at 19 Hz every 60 seconds.
* Heap profiling: heap snapshot with 2 MB frequency every 60 seconds.
There are two main approaches that can be taken for CPU profiles:
* Continuous low-frequency profiles (e.g. 19 Hz for 60 seconds every 60 seconds).
* Occasional high-frequency profiles (e.g. 99 Hz for 5 seconds every 60 seconds).
We choose continuous low-frequency profiles where possible. This has a fixed low overhead, instead
of a spiky high overhead. It likely also gives a more representative view of resource usage.
However, a 19 Hz rate gives a minimum resolution of 52.6 ms per sample, which may be larger than the
actual runtime of small functions. Note that Go does not support a frequency parameter, so we must
use a fixed frequency for all profiles via `SetCPUProfileRate()` (default 100 Hz).
Only one CPU profile can be taken at a time. With continuous profiling, one will always be running.
To allow also taking an ad hoc CPU profile, the Rust endpoint supports a `force` query parameter to
cancel a running profile and start a new one.
### Overhead
With Rust:
* CPU profiles at 19 Hz frequency: 0.1% overhead.
* Heap profiles at 2 MB frequency: 3% allocation overhead.
* Profile call/encoding/symbolization: 20 ms every 60 seconds, or 0.03% of 1 CPU (for Pageserver).
* Profile symbolization caches: 125 MB memory, or 0.4% of 32 GB (for Pageserver).
Benchmarks with pprof-rs showed that the CPU time for taking a stack trace of a 40-frame stack was
11 µs using the `frame-pointer` feature, and 1.4 µs using `libunwind` with DWARF. `libunwind` saw
frequent seg faults, so we use `frame-pointer` and build binaries with frame pointers (negligible
overhead).
CPU profiles work by installing an `ITIMER_PROF` for the process, which triggers a `SIGPROF` signal
after a given amount of cumulative CPU time across all CPUs. The signal handler will run for one
of the currently executing threads and take a stack trace. Thus, a 19 Hz profile will take 1 stack
trace every 52.6 ms CPU time -- assuming 11 µs for a stack trace, this is 0.02% overhead, but
likely 0.1% in practice (given e.g. context switches).
Heap profiles work by probabilistically taking a stack trace on allocations, adjusted for the
allocation size. A 1 MB allocation takes about 15 µs in benchmarks, and a stack trace about 1 µs,
so we can estimate that a 2 MB sampling frequency has about 3% allocation overhead -- this is
consistent with benchmarks. This is significantly larger than CPU profiles, but mitigated by the
fact that performance-sensitive code will avoid allocations as far as possible.
Profile symbolization uses in-memory caches for symbol lookups. These take about 125 MB for
Pageserver.
## Alternatives Considered
* eBPF profiles.
* Don't require instrumenting the binary.
* Use less resources.
* Can profile in kernel space too.
* Supported by Grafana.
* Less information about stack frames and spans.
* Limited tooling for local analysis.
* Does not support heap profiles.
* Does not work on macOS.
* [Polar Signals](https://www.polarsignals.com) instead of Grafana.
* We already use Grafana for everything else. Appears good enough.

View File

@@ -1,255 +0,0 @@
#
Created on Aug 2024
Implemented on Jan 2025
## Summary
Data in large tenants is split up between multiple pageservers according to key hashes, as
introduced in the [sharding RFC](031-sharding-static.md) and [shard splitting RFC](032-shard-splitting.md).
Whereas currently we send all WAL to all pageserver shards, and each shard filters out the data that it needs,
in this RFC we add a mechanism to filter the WAL on the safekeeper, so that each shard receives
only the data it needs.
This will place some extra CPU load on the safekeepers, in exchange for reducing the network bandwidth
for ingesting WAL back to scaling as O(1) with shard count, rather than O(N_shards).
## Motivation
1. Large databases require higher shard counts. Whereas currently we run with up to 8 shards for tenants
with a few TB of storage, the next order of magnitude capacity increase will require tens of shards, such
that sending all WAL to all shards is impractical in terms of bandwidth.
2. For contemporary database sizes (~2TB), the pageserver is the bottleneck for ingest: since each
shard has to decode and process the whole WAL, sharding doesn't fully relieve this bottleneck. To achieve significantly higher ingest speeds, we need to filter the WAL earlier so that each pageserver
only has to process relevant parts.
## Non Goals (if relevant)
We do not seek to introduce multiple WALs per timeline, or to share the work of handling a timeline's
WAL across safekeepers (beyond simple 3x replication). This RFC may be thought of as an incremental
move of the ingestion bottleneck up the stack: instead of high write rates bottlenecking on the
pageserver, they will bottleneck on the safekeeper.
## Impacted components (e.g. pageserver, safekeeper, console, etc)
Safekeeper, pageserver.
There will be no control plane or storage controller coordination needed, as pageservers will directly
indicate their sharding parameters to the safekeeper when subscribing for WAL.
## Proposed implementation
Terminology:
- "Data pages" refers to postgres relation blocks, and SLRU blocks.
- "Metadata pages" refers to everything else the pageserver stores, such as relation sizes and
directories of relations.
### Phase 1: Refactor ingest
Currently, pageserver ingest code is structured approximately as follows:
1. `handle_walreceiver_connection` reads a stream of binary WAL records off a network
socket
2. `WalIngest::ingest_record` to translate the record into a series of page-level modifications
3. `DatadirModification` accumulates page updates from several `ingest_record` calls, and when
its `commit()` method is called, flushes these into a Timeline's open `InMemoryLayer`.
This process currently assumes access to a pageserver `Timeline` throughout `ingest_record` and
from `DatadirModification`, which is used to do read-modify-write cycles on metadata pages
such as relation sizes and the master DBDIR page. It also assumes that records are ingested
strictly one after the other: they cannot be ingested in parallel because each record assumes
that earlier records' changes have already been applied to `Timeline`.
This code will be refactored to disentangle the simple, fast decode of relation page writes
from the more complex logic for updating internal metadata. An intermediate representation
called `InterpretedWalRecords` will be introduced. This is similar to the internal state of
a `DatadirModification`, but does not require access to a Timeline. Instead of storing
metadata updates as materialized writes to pages, it will accumulate these as abstract operations,
for example rather than including a write to a relation size key, this structure will include
an operation that indicates "Update relation _foo_'s size to the max of its current value and
_bar_", such that these may be applied later to a real Timeline.
The `DatadirModification` will be aware of the `EphemeralFile` format, so that as it accumulates
simple page writes of relation blocks, it can write them directly into a buffer in the serialized
format. This will avoid the need to later deserialize/reserialize this data when passing the
structure between safekeeper and pageserver.
The new pipeline will be:
1. `handle_walreceiver_connection` reads a stream of binary WAL records off a network
2. A `InterpretedWalRecords` is generated from the incoming WAL records. This does not
require a reference to a Timeline.
3. The logic that is current spread between `WalIngest` and `DatadirModification` for updating
metadata will be refactored to consume the metadata operations from the `InterpretedWalRecords`
and turn them into literal writes to metadata pages. This part must be done sequentially.
4. The resulting buffer of metadata page writes is combined with the buffer of relation block
writes, and written into the `InMemoryLayer`.
Implemented in:
1. https://github.com/neondatabase/neon/pull/9472
2. https://github.com/neondatabase/neon/pull/9504
3. https://github.com/neondatabase/neon/pull/9524
### Phase 2: Decode & filter on safekeeper
In the previous phase, the ingest code was modified to be able to do most of its work without access to
a Timeline: this first stage of ingest simply converts a series of binary wal records into
a buffer of relation/SLRU page writes, and a buffer of abstract metadata writes.
The modified ingest code may be transplanted from pageserver to safekeeper (probably via a
shared crate). The safekeeper->pageserver network protocol is modified to:
- in subscription requests, send the `ShardIdentity` from the pageserver to the safekeeper
- in responses, transmit a `InterpretedWalRecords` instead of a raw `WalRecord`.
- use the `ShardIdentity` to filter the `ProcessedWalIngest` to relevant content for
the subscribing shard before transmitting it.
The overall behavior of the pageserver->safekeeper interaction remains the same, in terms of
consistent LSN feedback, and connection management. Only the payload of the subscriptions
changes, to express an LSN range of WAL as a filtered `ProcessedWalIngest` instead of the
raw data.
The ingest code on the pageserver can now skip the part where it does the first phase of
processing, as it will receive pre-processed, compressed data off the wire.
Note that `InterpretedWalRecord` batches multiple `InterpretedWalRecord(s)` in the same network
message. Safekeeper reads WAL in chunks of 16 blocks and then decodes as many Postgres WAL records
as possible. Each Postgres WAL record maps to one `InterpretedWalRecord` for potentially multiple shards.
Hence, the size of the batch is given by the number of Postgres WAL records that fit in 16 blocks.
The protocol needs to support evolution. Protobuf was chosen here with the view that, in the future,
we may migrate it to GRPC altogether
Implemented in:
1. https://github.com/neondatabase/neon/pull/9746
2. https://github.com/neondatabase/neon/pull/9821
### Phase 3: Fan out interpreted WAL
In the previous phase, the initial processing of WAL was moved to the safekeeper, but it is still
done once for each shard: this will generate O(N_shards) CPU work on the safekeeper (especially
when considering converting to Protobuf format and compression).
To avoid this, we fan-out WAL from one (tenant, timeline, shard) to all other shards subscribed on
the same safekeeper. Under normal operation, the WAL will be read from disk, decoded and interpreted
_only_ once per (safekeeper, timeline).
When the first shard of a sharded timeline subscribes to a given safekeeper a task is spawned
for the WAL reader (`InterpretedWalReader`). This task reads WAL, decodes, interprets it and sends
it to the sender (`InterpretedWalSender`). The sender is a future that is polled from the connection
task. When further shards subscribe on the safekeeper they will attach themselves to the existing WAL reader.
There's two cases to consider:
1. The shard's requested `start_lsn` is ahead of the current position of the WAL reader. In this case, the shard
will start receiving data when the reader reaches that LSN. The intuition here is that there's little to gain
by letting shards "front-run" since compute backpressure is based on the laggard LSN.
2. The shard's requested `start_lsn` is below the current position of the WAL reader. In this case, the WAL reader
gets reset to this requested position (same intuition). Special care is taken such that advanced shards do not receive
interpreted WAL records below their current position.
The approach above implies that there is at most one WAL reader per (tenant, timeline) on a given safekeeper at any point in time.
If this turns out to be operationally problematic, there's a trick we can deploy: `--max-delta-for-fanout` is an optional safekeeper
argument that controls the max absolute delta between a new shard and the current WAL position of the WAL reader. If the absolute
delta is above that value, a new reader is spawned. Note that there's currently no concurrency control on the number of WAL readers,
so it's recommended to use large values to avoid pushing CPU utilisation too high.
Unsharded tenants do not spawn a separate task for the interpreted WAL reader since there's no benefit to it. Instead they poll
the reader and sender concurrently from the connection task.
Shard splits are interesting here because it is the only case when the same shard might have two subscriptions at the same time.
This is handled by giving readers a unique identifier. Both shards will receive the same data while respecting their requested start
position.
Implemented in:
1. https://github.com/neondatabase/neon/pull/10190
## Deployment
Each phase shall be deployed independently. Special care should be taken around protocol changes.
## Observability Tips
* The safekeeper logs the protocol requested by the pageserver
along with the pageserver ID, tenant, timeline and shard: `starting streaming from`.
* There's metrics for the number of wal readers:
* `safekeeper_wal_readers{kind="task", target=~"pageserver.*"}` gives the number of wal reader tasks for each SK
* `safekeeper_wal_readers{kind="future", target=~"pageserver.*"}` gives the numer of wal readers polled inline by each SK
* `safekeeper_interpreted_wal_reader_tasks` gives the number of wal reader tasks per tenant, timeline
* Interesting log lines for the fan-out reader:
* `Spawning interpreted`: first shard creates the interpreted wal reader
* `Fanning out`: a subsequent shard attaches itself to an interpreted wal reader
* `Aborting interpreted`: all senders have finished and the reader task is being aborted
## Future Optimizations
This sections describes some improvement areas which may be revisited in the future.
### Buffering of Interpreted WAL
The interpreted WAL reader may buffer interpreted WAL records in user space to help with serving
subscriptions that are lagging behind the current position of the reader.
Counterpoints:
* Safekeepers serve many thousands of timelines and allocating a buffer for each might be wasteful,
especially given that it would go unused on the happy path.
* WAL is buffered in the kernel page cache. Usually we'd only pay the CPU cost of decoding and interpreting.
### Tweaking the Pagserver Safekeeper Selection Algorithm
We could make the pageserver aware of which safekeeper's already host shards for the timeline along
with their current WAL positions. The pageserver should then prefer safkeepers that are in the same
AZ _and_ already have a shard with a position close to the desired start position.
We currently run one safekeeper per AZ, so the point is mute until that changes.
### Pipelining first ingest phase
The first ingest phase is a stateless transformation of a binary WAL record into a pre-processed
output per shard. To put multiple CPUs to work, we may pipeline this processing up to some defined buffer
depth.
## Alternatives considered
### Give safekeepers enough state to fully decode WAL
In this RFC, we only do the first phase of ingest on the safekeeper, because this is
the phase that is stateless. Subsequent changes then happen on the pageserver, with
access to the `Timeline` state.
We could do more work on the safekeeper if we transmitted metadata state to the safekeeper
when subscribing to the WAL: for example, by telling the safekeeper all the relation sizes,
so that it could then generate all the metadata writes for relation sizes.
We avoid doing this for several reasons:
1. Complexity: it's a more invasive protocol change
2. Decoupling: having the safekeeper understand the `ProcessedWalIngest` already somewhat
infects it with knowledge of the pageserver, but this is mainly an abstract structure
that describes postgres writes. However, if we taught the safekeeper about the exact
way that pageserver deals with metadata keys, this would be a much tighter coupling.
3. Load: once the WAL has been processed to the point that it can be split between shards,
it is preferable to share out work on the remaining shards rather than adding extra CPU
load to the safekeeper.
### Do pre-processing on the compute instead of the safekeeper
Since our first stage of ingest is stateless, it could be done at any stage in the pipeline,
all the way up to the compute.
We choose not to do this, because it is useful for the safekeeper to store the raw WAL rather
than just the preprocessed WAL:
- The safekeeper still needs to be able to serve raw WAL back to postgres for e.g. physical replication
- It simplifies our paxos implementation to have the offset in the write log be literally
the same as the LSN
- Raw WAL must have a stable protocol since we might have to re-ingest it at arbitrary points in the future.
Storing raw WAL give us more flexibility to evolve the pageserver, safekeeper protocol.
### Do wal pre-processing on shard 0 or a separate service, send it to other shards from there
If we wanted to keep the safekeepers as entirely pure stores of raw WAL bytes, then
we could do the initial decode and shard-splitting in some other location:
- Shard 0 could subscribe to the full WAL and then send writes to other shards
- A new intermediate service between the safekeeper and pageserver could do the splitting.
So why not?
- Extra network hop from shard 0 to the final destination shard
- Clearly there is more infrastructure involved here compared with doing it inline on the safekeeper.
- Safekeepers already have very light CPU load: typical cloud instances shapes with appropriate
disks for the safekeepers effectively have "free" CPU resources.
- Doing extra work on shard 0 would complicate scheduling of shards on pageservers, because
shard 0 would have significantly higher CPU load under write workloads than other shards.

View File

@@ -15,17 +15,6 @@ pub struct GenericAPIError {
pub error: String,
}
#[derive(Debug, Clone, Serialize)]
pub struct InfoResponse {
pub num_cpus: usize,
}
#[derive(Debug, Clone, Serialize)]
pub struct ExtensionInstallResponse {
pub extension: PgIdent,
pub version: ExtVersion,
}
/// Response of the /status API
#[derive(Serialize, Debug, Deserialize)]
#[serde(rename_all = "snake_case")]
@@ -39,6 +28,16 @@ pub struct ComputeStatusResponse {
pub error: Option<String>,
}
#[derive(Deserialize, Serialize)]
#[serde(rename_all = "snake_case")]
pub struct ComputeState {
pub status: ComputeStatus,
/// Timestamp of the last Postgres activity
#[serde(serialize_with = "rfc3339_serialize")]
pub last_active: Option<DateTime<Utc>>,
pub error: Option<String>,
}
#[derive(Serialize, Clone, Copy, Debug, Deserialize, PartialEq, Eq)]
#[serde(rename_all = "snake_case")]
pub enum ComputeStatus {
@@ -79,7 +78,7 @@ impl Display for ComputeStatus {
}
}
pub fn rfc3339_serialize<S>(x: &Option<DateTime<Utc>>, s: S) -> Result<S::Ok, S::Error>
fn rfc3339_serialize<S>(x: &Option<DateTime<Utc>>, s: S) -> Result<S::Ok, S::Error>
where
S: Serializer,
{

View File

@@ -67,15 +67,6 @@ pub struct ComputeSpec {
#[serde(default)]
pub disk_quota_bytes: Option<u64>,
/// Disables the vm-monitor behavior that resizes LFC on upscale/downscale, instead relying on
/// the initial size of LFC.
///
/// This is intended for use when the LFC size is being overridden from the default but
/// autoscaling is still enabled, and we don't want the vm-monitor to interfere with the custom
/// LFC sizing.
#[serde(default)]
pub disable_lfc_resizing: Option<bool>,
/// Expected cluster state at the end of transition process.
pub cluster: Cluster,
pub delta_operations: Option<Vec<DeltaOp>>,

View File

@@ -91,7 +91,7 @@ impl Timing {
/// Return true if there is a ready event.
fn is_event_ready(&self, queue: &mut BinaryHeap<Pending>) -> bool {
queue.peek().is_some_and(|x| x.time <= self.now())
queue.peek().map_or(false, |x| x.time <= self.now())
}
/// Clear all pending events.

View File

@@ -301,10 +301,6 @@ pub struct TenantConfigToml {
pub timeline_offloading: bool,
pub wal_receiver_protocol_override: Option<PostgresClientProtocol>,
/// 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 mod defaults {
@@ -542,7 +538,6 @@ impl Default for TenantConfigToml {
lsn_lease_length_for_ts: LsnLease::DEFAULT_LENGTH_FOR_TS,
timeline_offloading: false,
wal_receiver_protocol_override: None,
rel_size_v2_enabled: None,
}
}
}

View File

@@ -87,7 +87,7 @@ impl Display for AvailabilityZone {
#[derive(Serialize, Deserialize)]
pub struct ShardsPreferredAzsRequest {
#[serde(flatten)]
pub preferred_az_ids: HashMap<TenantShardId, Option<AvailabilityZone>>,
pub preferred_az_ids: HashMap<TenantShardId, AvailabilityZone>,
}
#[derive(Serialize, Deserialize)]
@@ -144,8 +144,6 @@ pub struct NodeDescribeResponse {
pub availability: NodeAvailabilityWrapper,
pub scheduling: NodeSchedulingPolicy,
pub availability_zone_id: String,
pub listen_http_addr: String,
pub listen_http_port: u16,
@@ -181,6 +179,7 @@ pub struct TenantDescribeResponseShard {
/// specifies some constraints, e.g. asking it to get off particular node(s)
#[derive(Serialize, Deserialize, Debug)]
pub struct TenantShardMigrateRequest {
pub tenant_shard_id: TenantShardId,
pub node_id: NodeId,
}
@@ -321,42 +320,6 @@ impl From<NodeSchedulingPolicy> for String {
}
}
#[derive(Serialize, Deserialize, Clone, Copy, Eq, PartialEq, Debug)]
pub enum SkSchedulingPolicy {
Active,
Pause,
Decomissioned,
}
impl FromStr for SkSchedulingPolicy {
type Err = anyhow::Error;
fn from_str(s: &str) -> Result<Self, Self::Err> {
Ok(match s {
"active" => Self::Active,
"pause" => Self::Pause,
"decomissioned" => Self::Decomissioned,
_ => {
return Err(anyhow::anyhow!(
"Unknown scheduling policy '{s}', try active,pause,decomissioned"
))
}
})
}
}
impl From<SkSchedulingPolicy> for String {
fn from(value: SkSchedulingPolicy) -> String {
use SkSchedulingPolicy::*;
match value {
Active => "active",
Pause => "pause",
Decomissioned => "decomissioned",
}
.to_string()
}
}
/// Controls how tenant shards are mapped to locations on pageservers, e.g. whether
/// to create secondary locations.
#[derive(Clone, Serialize, Deserialize, Debug, PartialEq, Eq)]
@@ -373,16 +336,6 @@ pub enum PlacementPolicy {
Detached,
}
impl PlacementPolicy {
pub fn want_secondaries(&self) -> usize {
match self {
PlacementPolicy::Attached(secondary_count) => *secondary_count,
PlacementPolicy::Secondary => 1,
PlacementPolicy::Detached => 0,
}
}
}
#[derive(Serialize, Deserialize, Debug)]
pub struct TenantShardMigrateResponse {}
@@ -419,27 +372,6 @@ pub struct MetadataHealthListOutdatedResponse {
pub health_records: Vec<MetadataHealthRecord>,
}
/// Publicly exposed safekeeper description
#[derive(Serialize, Deserialize, Clone)]
pub struct SafekeeperDescribeResponse {
pub id: NodeId,
pub region_id: String,
/// 1 is special, it means just created (not currently posted to storcon).
/// Zero or negative is not really expected.
/// Otherwise the number from `release-$(number_of_commits_on_branch)` tag.
pub version: i64,
pub host: String,
pub port: i32,
pub http_port: i32,
pub availability_zone_id: String,
pub scheduling_policy: SkSchedulingPolicy,
}
#[derive(Serialize, Deserialize, Clone)]
pub struct SafekeeperSchedulingPolicyRequest {
pub scheduling_policy: SkSchedulingPolicy,
}
#[cfg(test)]
mod test {
use super::*;

View File

@@ -24,9 +24,7 @@ pub struct Key {
/// When working with large numbers of Keys in-memory, it is more efficient to handle them as i128 than as
/// a struct of fields.
#[derive(
Clone, Copy, Default, Hash, PartialEq, Eq, Ord, PartialOrd, Serialize, Deserialize, Debug,
)]
#[derive(Clone, Copy, Hash, PartialEq, Eq, Ord, PartialOrd, Serialize, Deserialize, Debug)]
pub struct CompactKey(i128);
/// The storage key size.
@@ -567,10 +565,6 @@ impl Key {
&& self.field5 == 0
&& self.field6 == u32::MAX
}
pub fn is_slru_dir_key(&self) -> bool {
slru_dir_kind(self).is_some()
}
}
#[inline(always)]
@@ -708,7 +702,7 @@ pub fn repl_origin_key_range() -> Range<Key> {
/// Non inherited range for vectored get.
pub const NON_INHERITED_RANGE: Range<Key> = AUX_FILES_KEY..AUX_FILES_KEY.next();
/// Sparse keyspace range for vectored get. Missing key error will be ignored for this range.
pub const SPARSE_RANGE: Range<Key> = Key::metadata_key_range();
pub const NON_INHERITED_SPARSE_RANGE: Range<Key> = Key::metadata_key_range();
impl Key {
// AUX_FILES currently stores only data for logical replication (slots etc), and
@@ -716,42 +710,7 @@ impl Key {
// switch (and generally it likely should be optional), so ignore these.
#[inline(always)]
pub fn is_inherited_key(self) -> bool {
if self.is_sparse() {
self.is_inherited_sparse_key()
} else {
!NON_INHERITED_RANGE.contains(&self)
}
}
#[inline(always)]
pub fn is_sparse(self) -> bool {
self.field1 >= METADATA_KEY_BEGIN_PREFIX && self.field1 < METADATA_KEY_END_PREFIX
}
/// Check if the key belongs to the inherited keyspace.
fn is_inherited_sparse_key(self) -> bool {
debug_assert!(self.is_sparse());
self.field1 == RELATION_SIZE_PREFIX
}
pub 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);
Key {
field1: AUX_KEY_PREFIX,
field2: 0,
field3: 0,
field4: 0,
field5: 0,
field6: 0,
}..Key {
field1: REPL_ORIGIN_KEY_PREFIX + 1,
field2: 0,
field3: 0,
field4: 0,
field5: 0,
field6: 0,
}
!NON_INHERITED_RANGE.contains(&self) && !NON_INHERITED_SPARSE_RANGE.contains(&self)
}
#[inline(always)]

View File

@@ -6,7 +6,6 @@ pub mod utilization;
use camino::Utf8PathBuf;
pub use utilization::PageserverUtilization;
use core::ops::Range;
use std::{
collections::HashMap,
fmt::Display,
@@ -29,10 +28,10 @@ use utils::{
};
use crate::{
key::{CompactKey, Key},
reltag::RelTag,
shard::{ShardCount, ShardStripeSize, TenantShardId},
};
use anyhow::bail;
use bytes::{Buf, BufMut, Bytes, BytesMut};
/// The state of a tenant in this pageserver.
@@ -211,70 +210,6 @@ pub enum TimelineState {
Broken { reason: String, backtrace: String },
}
#[serde_with::serde_as]
#[derive(Debug, Clone, serde::Deserialize, serde::Serialize)]
pub struct CompactLsnRange {
pub start: Lsn,
pub end: Lsn,
}
#[serde_with::serde_as]
#[derive(Debug, Clone, serde::Deserialize, serde::Serialize)]
pub struct CompactKeyRange {
#[serde_as(as = "serde_with::DisplayFromStr")]
pub start: Key,
#[serde_as(as = "serde_with::DisplayFromStr")]
pub end: Key,
}
impl From<Range<Lsn>> for CompactLsnRange {
fn from(range: Range<Lsn>) -> Self {
Self {
start: range.start,
end: range.end,
}
}
}
impl From<Range<Key>> for CompactKeyRange {
fn from(range: Range<Key>) -> Self {
Self {
start: range.start,
end: range.end,
}
}
}
impl From<CompactLsnRange> for Range<Lsn> {
fn from(range: CompactLsnRange) -> Self {
range.start..range.end
}
}
impl From<CompactKeyRange> for Range<Key> {
fn from(range: CompactKeyRange) -> Self {
range.start..range.end
}
}
impl CompactLsnRange {
pub fn above(lsn: Lsn) -> Self {
Self {
start: lsn,
end: Lsn::MAX,
}
}
}
#[derive(Debug, Clone, Serialize)]
pub struct CompactInfoResponse {
pub compact_key_range: Option<CompactKeyRange>,
pub compact_lsn_range: Option<CompactLsnRange>,
pub sub_compaction: bool,
pub running: bool,
pub job_id: usize,
}
#[derive(Serialize, Deserialize, Clone)]
pub struct TimelineCreateRequest {
pub new_timeline_id: TimelineId,
@@ -497,8 +432,6 @@ pub struct TenantConfigPatch {
pub timeline_offloading: FieldPatch<bool>,
#[serde(skip_serializing_if = "FieldPatch::is_noop")]
pub wal_receiver_protocol_override: FieldPatch<PostgresClientProtocol>,
#[serde(skip_serializing_if = "FieldPatch::is_noop")]
pub rel_size_v2_enabled: FieldPatch<bool>,
}
/// An alternative representation of `pageserver::tenant::TenantConf` with
@@ -530,7 +463,6 @@ pub struct TenantConfig {
pub lsn_lease_length_for_ts: Option<String>,
pub timeline_offloading: Option<bool>,
pub wal_receiver_protocol_override: Option<PostgresClientProtocol>,
pub rel_size_v2_enabled: Option<bool>,
}
impl TenantConfig {
@@ -560,7 +492,6 @@ impl TenantConfig {
mut lsn_lease_length_for_ts,
mut timeline_offloading,
mut wal_receiver_protocol_override,
mut rel_size_v2_enabled,
} = self;
patch.checkpoint_distance.apply(&mut checkpoint_distance);
@@ -605,7 +536,6 @@ impl TenantConfig {
patch
.wal_receiver_protocol_override
.apply(&mut wal_receiver_protocol_override);
patch.rel_size_v2_enabled.apply(&mut rel_size_v2_enabled);
Self {
checkpoint_distance,
@@ -632,7 +562,6 @@ impl TenantConfig {
lsn_lease_length_for_ts,
timeline_offloading,
wal_receiver_protocol_override,
rel_size_v2_enabled,
}
}
}
@@ -1405,8 +1334,6 @@ pub enum PagestreamFeMessage {
GetPage(PagestreamGetPageRequest),
DbSize(PagestreamDbSizeRequest),
GetSlruSegment(PagestreamGetSlruSegmentRequest),
#[cfg(feature = "testing")]
Test(PagestreamTestRequest),
}
// Wrapped in libpq CopyData
@@ -1418,22 +1345,6 @@ pub enum PagestreamBeMessage {
Error(PagestreamErrorResponse),
DbSize(PagestreamDbSizeResponse),
GetSlruSegment(PagestreamGetSlruSegmentResponse),
#[cfg(feature = "testing")]
Test(PagestreamTestResponse),
}
// Keep in sync with `pagestore_client.h`
#[repr(u8)]
enum PagestreamFeMessageTag {
Exists = 0,
Nblocks = 1,
GetPage = 2,
DbSize = 3,
GetSlruSegment = 4,
/* future tags above this line */
/// For testing purposes, not available in production.
#[cfg(feature = "testing")]
Test = 99,
}
// Keep in sync with `pagestore_client.h`
@@ -1445,28 +1356,7 @@ enum PagestreamBeMessageTag {
Error = 103,
DbSize = 104,
GetSlruSegment = 105,
/* future tags above this line */
/// For testing purposes, not available in production.
#[cfg(feature = "testing")]
Test = 199,
}
impl TryFrom<u8> for PagestreamFeMessageTag {
type Error = u8;
fn try_from(value: u8) -> Result<Self, u8> {
match value {
0 => Ok(PagestreamFeMessageTag::Exists),
1 => Ok(PagestreamFeMessageTag::Nblocks),
2 => Ok(PagestreamFeMessageTag::GetPage),
3 => Ok(PagestreamFeMessageTag::DbSize),
4 => Ok(PagestreamFeMessageTag::GetSlruSegment),
#[cfg(feature = "testing")]
99 => Ok(PagestreamFeMessageTag::Test),
_ => Err(value),
}
}
}
impl TryFrom<u8> for PagestreamBeMessageTag {
type Error = u8;
fn try_from(value: u8) -> Result<Self, u8> {
@@ -1477,8 +1367,6 @@ impl TryFrom<u8> for PagestreamBeMessageTag {
103 => Ok(PagestreamBeMessageTag::Error),
104 => Ok(PagestreamBeMessageTag::DbSize),
105 => Ok(PagestreamBeMessageTag::GetSlruSegment),
#[cfg(feature = "testing")]
199 => Ok(PagestreamBeMessageTag::Test),
_ => Err(value),
}
}
@@ -1508,108 +1396,78 @@ impl TryFrom<u8> for PagestreamBeMessageTag {
// interface allows sending both LSNs, and let the pageserver do the right thing. There was no
// difference in the responses between V1 and V2.
//
// V3 version of protocol adds request ID to all requests. This request ID is also included in response
// as well as other fields from requests, which allows to verify that we receive response for our request.
// We copy fields from request to response to make checking more reliable: request ID is formed from process ID
// and local counter, so in principle there can be duplicated requests IDs if process PID is reused.
//
#[derive(Debug, PartialEq, Eq, Clone, Copy)]
#[derive(Clone, Copy)]
pub enum PagestreamProtocolVersion {
V2,
V3,
}
pub type RequestId = u64;
#[derive(Debug, PartialEq, Eq, Clone, Copy)]
pub struct PagestreamRequest {
pub reqid: RequestId,
#[derive(Debug, PartialEq, Eq)]
pub struct PagestreamExistsRequest {
pub request_lsn: Lsn,
pub not_modified_since: Lsn,
}
#[derive(Debug, PartialEq, Eq, Clone, Copy)]
pub struct PagestreamExistsRequest {
pub hdr: PagestreamRequest,
pub rel: RelTag,
}
#[derive(Debug, PartialEq, Eq, Clone, Copy)]
#[derive(Debug, PartialEq, Eq)]
pub struct PagestreamNblocksRequest {
pub hdr: PagestreamRequest,
pub request_lsn: Lsn,
pub not_modified_since: Lsn,
pub rel: RelTag,
}
#[derive(Debug, PartialEq, Eq, Clone, Copy)]
#[derive(Debug, PartialEq, Eq)]
pub struct PagestreamGetPageRequest {
pub hdr: PagestreamRequest,
pub request_lsn: Lsn,
pub not_modified_since: Lsn,
pub rel: RelTag,
pub blkno: u32,
}
#[derive(Debug, PartialEq, Eq, Clone, Copy)]
#[derive(Debug, PartialEq, Eq)]
pub struct PagestreamDbSizeRequest {
pub hdr: PagestreamRequest,
pub request_lsn: Lsn,
pub not_modified_since: Lsn,
pub dbnode: u32,
}
#[derive(Debug, PartialEq, Eq, Clone, Copy)]
#[derive(Debug, PartialEq, Eq)]
pub struct PagestreamGetSlruSegmentRequest {
pub hdr: PagestreamRequest,
pub request_lsn: Lsn,
pub not_modified_since: Lsn,
pub kind: u8,
pub segno: u32,
}
#[derive(Debug)]
pub struct PagestreamExistsResponse {
pub req: PagestreamExistsRequest,
pub exists: bool,
}
#[derive(Debug)]
pub struct PagestreamNblocksResponse {
pub req: PagestreamNblocksRequest,
pub n_blocks: u32,
}
#[derive(Debug)]
pub struct PagestreamGetPageResponse {
pub req: PagestreamGetPageRequest,
pub page: Bytes,
}
#[derive(Debug)]
pub struct PagestreamGetSlruSegmentResponse {
pub req: PagestreamGetSlruSegmentRequest,
pub segment: Bytes,
}
#[derive(Debug)]
pub struct PagestreamErrorResponse {
pub req: PagestreamRequest,
pub message: String,
}
#[derive(Debug)]
pub struct PagestreamDbSizeResponse {
pub req: PagestreamDbSizeRequest,
pub db_size: i64,
}
#[cfg(feature = "testing")]
#[derive(Debug, PartialEq, Eq, Clone)]
pub struct PagestreamTestRequest {
pub hdr: PagestreamRequest,
pub batch_key: u64,
pub message: String,
}
#[cfg(feature = "testing")]
#[derive(Debug)]
pub struct PagestreamTestResponse {
pub req: PagestreamTestRequest,
}
// This is a cut-down version of TenantHistorySize from the pageserver crate, omitting fields
// that require pageserver-internal types. It is sufficient to get the total size.
#[derive(Serialize, Deserialize, Debug)]
@@ -1623,16 +1481,15 @@ pub struct TenantHistorySize {
impl PagestreamFeMessage {
/// Serialize a compute -> pageserver message. This is currently only used in testing
/// tools. Always uses protocol version 3.
/// tools. Always uses protocol version 2.
pub fn serialize(&self) -> Bytes {
let mut bytes = BytesMut::new();
match self {
Self::Exists(req) => {
bytes.put_u8(PagestreamFeMessageTag::Exists as u8);
bytes.put_u64(req.hdr.reqid);
bytes.put_u64(req.hdr.request_lsn.0);
bytes.put_u64(req.hdr.not_modified_since.0);
bytes.put_u8(0);
bytes.put_u64(req.request_lsn.0);
bytes.put_u64(req.not_modified_since.0);
bytes.put_u32(req.rel.spcnode);
bytes.put_u32(req.rel.dbnode);
bytes.put_u32(req.rel.relnode);
@@ -1640,10 +1497,9 @@ impl PagestreamFeMessage {
}
Self::Nblocks(req) => {
bytes.put_u8(PagestreamFeMessageTag::Nblocks as u8);
bytes.put_u64(req.hdr.reqid);
bytes.put_u64(req.hdr.request_lsn.0);
bytes.put_u64(req.hdr.not_modified_since.0);
bytes.put_u8(1);
bytes.put_u64(req.request_lsn.0);
bytes.put_u64(req.not_modified_since.0);
bytes.put_u32(req.rel.spcnode);
bytes.put_u32(req.rel.dbnode);
bytes.put_u32(req.rel.relnode);
@@ -1651,10 +1507,9 @@ impl PagestreamFeMessage {
}
Self::GetPage(req) => {
bytes.put_u8(PagestreamFeMessageTag::GetPage as u8);
bytes.put_u64(req.hdr.reqid);
bytes.put_u64(req.hdr.request_lsn.0);
bytes.put_u64(req.hdr.not_modified_since.0);
bytes.put_u8(2);
bytes.put_u64(req.request_lsn.0);
bytes.put_u64(req.not_modified_since.0);
bytes.put_u32(req.rel.spcnode);
bytes.put_u32(req.rel.dbnode);
bytes.put_u32(req.rel.relnode);
@@ -1663,278 +1518,123 @@ impl PagestreamFeMessage {
}
Self::DbSize(req) => {
bytes.put_u8(PagestreamFeMessageTag::DbSize as u8);
bytes.put_u64(req.hdr.reqid);
bytes.put_u64(req.hdr.request_lsn.0);
bytes.put_u64(req.hdr.not_modified_since.0);
bytes.put_u8(3);
bytes.put_u64(req.request_lsn.0);
bytes.put_u64(req.not_modified_since.0);
bytes.put_u32(req.dbnode);
}
Self::GetSlruSegment(req) => {
bytes.put_u8(PagestreamFeMessageTag::GetSlruSegment as u8);
bytes.put_u64(req.hdr.reqid);
bytes.put_u64(req.hdr.request_lsn.0);
bytes.put_u64(req.hdr.not_modified_since.0);
bytes.put_u8(4);
bytes.put_u64(req.request_lsn.0);
bytes.put_u64(req.not_modified_since.0);
bytes.put_u8(req.kind);
bytes.put_u32(req.segno);
}
#[cfg(feature = "testing")]
Self::Test(req) => {
bytes.put_u8(PagestreamFeMessageTag::Test as u8);
bytes.put_u64(req.hdr.reqid);
bytes.put_u64(req.hdr.request_lsn.0);
bytes.put_u64(req.hdr.not_modified_since.0);
bytes.put_u64(req.batch_key);
let message = req.message.as_bytes();
bytes.put_u64(message.len() as u64);
bytes.put_slice(message);
}
}
bytes.into()
}
pub fn parse<R: std::io::Read>(
body: &mut R,
protocol_version: PagestreamProtocolVersion,
) -> anyhow::Result<PagestreamFeMessage> {
pub fn parse<R: std::io::Read>(body: &mut R) -> anyhow::Result<PagestreamFeMessage> {
// these correspond to the NeonMessageTag enum in pagestore_client.h
//
// TODO: consider using protobuf or serde bincode for less error prone
// serialization.
let msg_tag = body.read_u8()?;
let (reqid, request_lsn, not_modified_since) = match protocol_version {
PagestreamProtocolVersion::V2 => (
0,
Lsn::from(body.read_u64::<BigEndian>()?),
Lsn::from(body.read_u64::<BigEndian>()?),
),
PagestreamProtocolVersion::V3 => (
body.read_u64::<BigEndian>()?,
Lsn::from(body.read_u64::<BigEndian>()?),
Lsn::from(body.read_u64::<BigEndian>()?),
),
};
match PagestreamFeMessageTag::try_from(msg_tag)
.map_err(|tag: u8| anyhow::anyhow!("invalid tag {tag}"))?
{
PagestreamFeMessageTag::Exists => {
Ok(PagestreamFeMessage::Exists(PagestreamExistsRequest {
hdr: PagestreamRequest {
reqid,
request_lsn,
not_modified_since,
},
rel: RelTag {
spcnode: body.read_u32::<BigEndian>()?,
dbnode: body.read_u32::<BigEndian>()?,
relnode: body.read_u32::<BigEndian>()?,
forknum: body.read_u8()?,
},
}))
}
PagestreamFeMessageTag::Nblocks => {
Ok(PagestreamFeMessage::Nblocks(PagestreamNblocksRequest {
hdr: PagestreamRequest {
reqid,
request_lsn,
not_modified_since,
},
rel: RelTag {
spcnode: body.read_u32::<BigEndian>()?,
dbnode: body.read_u32::<BigEndian>()?,
relnode: body.read_u32::<BigEndian>()?,
forknum: body.read_u8()?,
},
}))
}
PagestreamFeMessageTag::GetPage => {
Ok(PagestreamFeMessage::GetPage(PagestreamGetPageRequest {
hdr: PagestreamRequest {
reqid,
request_lsn,
not_modified_since,
},
rel: RelTag {
spcnode: body.read_u32::<BigEndian>()?,
dbnode: body.read_u32::<BigEndian>()?,
relnode: body.read_u32::<BigEndian>()?,
forknum: body.read_u8()?,
},
blkno: body.read_u32::<BigEndian>()?,
}))
}
PagestreamFeMessageTag::DbSize => {
Ok(PagestreamFeMessage::DbSize(PagestreamDbSizeRequest {
hdr: PagestreamRequest {
reqid,
request_lsn,
not_modified_since,
},
// these two fields are the same for every request type
let request_lsn = Lsn::from(body.read_u64::<BigEndian>()?);
let not_modified_since = Lsn::from(body.read_u64::<BigEndian>()?);
match msg_tag {
0 => Ok(PagestreamFeMessage::Exists(PagestreamExistsRequest {
request_lsn,
not_modified_since,
rel: RelTag {
spcnode: body.read_u32::<BigEndian>()?,
dbnode: body.read_u32::<BigEndian>()?,
}))
}
PagestreamFeMessageTag::GetSlruSegment => Ok(PagestreamFeMessage::GetSlruSegment(
relnode: body.read_u32::<BigEndian>()?,
forknum: body.read_u8()?,
},
})),
1 => Ok(PagestreamFeMessage::Nblocks(PagestreamNblocksRequest {
request_lsn,
not_modified_since,
rel: RelTag {
spcnode: body.read_u32::<BigEndian>()?,
dbnode: body.read_u32::<BigEndian>()?,
relnode: body.read_u32::<BigEndian>()?,
forknum: body.read_u8()?,
},
})),
2 => Ok(PagestreamFeMessage::GetPage(PagestreamGetPageRequest {
request_lsn,
not_modified_since,
rel: RelTag {
spcnode: body.read_u32::<BigEndian>()?,
dbnode: body.read_u32::<BigEndian>()?,
relnode: body.read_u32::<BigEndian>()?,
forknum: body.read_u8()?,
},
blkno: body.read_u32::<BigEndian>()?,
})),
3 => Ok(PagestreamFeMessage::DbSize(PagestreamDbSizeRequest {
request_lsn,
not_modified_since,
dbnode: body.read_u32::<BigEndian>()?,
})),
4 => Ok(PagestreamFeMessage::GetSlruSegment(
PagestreamGetSlruSegmentRequest {
hdr: PagestreamRequest {
reqid,
request_lsn,
not_modified_since,
},
request_lsn,
not_modified_since,
kind: body.read_u8()?,
segno: body.read_u32::<BigEndian>()?,
},
)),
#[cfg(feature = "testing")]
PagestreamFeMessageTag::Test => Ok(PagestreamFeMessage::Test(PagestreamTestRequest {
hdr: PagestreamRequest {
reqid,
request_lsn,
not_modified_since,
},
batch_key: body.read_u64::<BigEndian>()?,
message: {
let len = body.read_u64::<BigEndian>()?;
let mut buf = vec![0; len as usize];
body.read_exact(&mut buf)?;
String::from_utf8(buf)?
},
})),
_ => bail!("unknown smgr message tag: {:?}", msg_tag),
}
}
}
impl PagestreamBeMessage {
pub fn serialize(&self, protocol_version: PagestreamProtocolVersion) -> Bytes {
pub fn serialize(&self) -> Bytes {
let mut bytes = BytesMut::new();
use PagestreamBeMessageTag as Tag;
match protocol_version {
PagestreamProtocolVersion::V2 => {
match self {
Self::Exists(resp) => {
bytes.put_u8(Tag::Exists as u8);
bytes.put_u8(resp.exists as u8);
}
Self::Nblocks(resp) => {
bytes.put_u8(Tag::Nblocks as u8);
bytes.put_u32(resp.n_blocks);
}
Self::GetPage(resp) => {
bytes.put_u8(Tag::GetPage as u8);
bytes.put(&resp.page[..])
}
Self::Error(resp) => {
bytes.put_u8(Tag::Error as u8);
bytes.put(resp.message.as_bytes());
bytes.put_u8(0); // null terminator
}
Self::DbSize(resp) => {
bytes.put_u8(Tag::DbSize as u8);
bytes.put_i64(resp.db_size);
}
Self::GetSlruSegment(resp) => {
bytes.put_u8(Tag::GetSlruSegment as u8);
bytes.put_u32((resp.segment.len() / BLCKSZ as usize) as u32);
bytes.put(&resp.segment[..]);
}
#[cfg(feature = "testing")]
Self::Test(resp) => {
bytes.put_u8(Tag::Test as u8);
bytes.put_u64(resp.req.batch_key);
let message = resp.req.message.as_bytes();
bytes.put_u64(message.len() as u64);
bytes.put_slice(message);
}
}
match self {
Self::Exists(resp) => {
bytes.put_u8(Tag::Exists as u8);
bytes.put_u8(resp.exists as u8);
}
PagestreamProtocolVersion::V3 => {
match self {
Self::Exists(resp) => {
bytes.put_u8(Tag::Exists as u8);
bytes.put_u64(resp.req.hdr.reqid);
bytes.put_u64(resp.req.hdr.request_lsn.0);
bytes.put_u64(resp.req.hdr.not_modified_since.0);
bytes.put_u32(resp.req.rel.spcnode);
bytes.put_u32(resp.req.rel.dbnode);
bytes.put_u32(resp.req.rel.relnode);
bytes.put_u8(resp.req.rel.forknum);
bytes.put_u8(resp.exists as u8);
}
Self::Nblocks(resp) => {
bytes.put_u8(Tag::Nblocks as u8);
bytes.put_u64(resp.req.hdr.reqid);
bytes.put_u64(resp.req.hdr.request_lsn.0);
bytes.put_u64(resp.req.hdr.not_modified_since.0);
bytes.put_u32(resp.req.rel.spcnode);
bytes.put_u32(resp.req.rel.dbnode);
bytes.put_u32(resp.req.rel.relnode);
bytes.put_u8(resp.req.rel.forknum);
bytes.put_u32(resp.n_blocks);
}
Self::Nblocks(resp) => {
bytes.put_u8(Tag::Nblocks as u8);
bytes.put_u32(resp.n_blocks);
}
Self::GetPage(resp) => {
bytes.put_u8(Tag::GetPage as u8);
bytes.put_u64(resp.req.hdr.reqid);
bytes.put_u64(resp.req.hdr.request_lsn.0);
bytes.put_u64(resp.req.hdr.not_modified_since.0);
bytes.put_u32(resp.req.rel.spcnode);
bytes.put_u32(resp.req.rel.dbnode);
bytes.put_u32(resp.req.rel.relnode);
bytes.put_u8(resp.req.rel.forknum);
bytes.put_u32(resp.req.blkno);
bytes.put(&resp.page[..])
}
Self::GetPage(resp) => {
bytes.put_u8(Tag::GetPage as u8);
bytes.put(&resp.page[..]);
}
Self::Error(resp) => {
bytes.put_u8(Tag::Error as u8);
bytes.put_u64(resp.req.reqid);
bytes.put_u64(resp.req.request_lsn.0);
bytes.put_u64(resp.req.not_modified_since.0);
bytes.put(resp.message.as_bytes());
bytes.put_u8(0); // null terminator
}
Self::DbSize(resp) => {
bytes.put_u8(Tag::DbSize as u8);
bytes.put_u64(resp.req.hdr.reqid);
bytes.put_u64(resp.req.hdr.request_lsn.0);
bytes.put_u64(resp.req.hdr.not_modified_since.0);
bytes.put_u32(resp.req.dbnode);
bytes.put_i64(resp.db_size);
}
Self::Error(resp) => {
bytes.put_u8(Tag::Error as u8);
bytes.put(resp.message.as_bytes());
bytes.put_u8(0); // null terminator
}
Self::DbSize(resp) => {
bytes.put_u8(Tag::DbSize as u8);
bytes.put_i64(resp.db_size);
}
Self::GetSlruSegment(resp) => {
bytes.put_u8(Tag::GetSlruSegment as u8);
bytes.put_u64(resp.req.hdr.reqid);
bytes.put_u64(resp.req.hdr.request_lsn.0);
bytes.put_u64(resp.req.hdr.not_modified_since.0);
bytes.put_u8(resp.req.kind);
bytes.put_u32(resp.req.segno);
bytes.put_u32((resp.segment.len() / BLCKSZ as usize) as u32);
bytes.put(&resp.segment[..]);
}
#[cfg(feature = "testing")]
Self::Test(resp) => {
bytes.put_u8(Tag::Test as u8);
bytes.put_u64(resp.req.hdr.reqid);
bytes.put_u64(resp.req.hdr.request_lsn.0);
bytes.put_u64(resp.req.hdr.not_modified_since.0);
bytes.put_u64(resp.req.batch_key);
let message = resp.req.message.as_bytes();
bytes.put_u64(message.len() as u64);
bytes.put_slice(message);
}
}
Self::GetSlruSegment(resp) => {
bytes.put_u8(Tag::GetSlruSegment as u8);
bytes.put_u32((resp.segment.len() / BLCKSZ as usize) as u32);
bytes.put(&resp.segment[..]);
}
}
bytes.into()
}
@@ -1946,156 +1646,41 @@ impl PagestreamBeMessage {
let ok =
match Tag::try_from(msg_tag).map_err(|tag: u8| anyhow::anyhow!("invalid tag {tag}"))? {
Tag::Exists => {
let reqid = buf.read_u64::<BigEndian>()?;
let request_lsn = Lsn(buf.read_u64::<BigEndian>()?);
let not_modified_since = Lsn(buf.read_u64::<BigEndian>()?);
let rel = RelTag {
spcnode: buf.read_u32::<BigEndian>()?,
dbnode: buf.read_u32::<BigEndian>()?,
relnode: buf.read_u32::<BigEndian>()?,
forknum: buf.read_u8()?,
};
let exists = buf.read_u8()? != 0;
let exists = buf.read_u8()?;
Self::Exists(PagestreamExistsResponse {
req: PagestreamExistsRequest {
hdr: PagestreamRequest {
reqid,
request_lsn,
not_modified_since,
},
rel,
},
exists,
exists: exists != 0,
})
}
Tag::Nblocks => {
let reqid = buf.read_u64::<BigEndian>()?;
let request_lsn = Lsn(buf.read_u64::<BigEndian>()?);
let not_modified_since = Lsn(buf.read_u64::<BigEndian>()?);
let rel = RelTag {
spcnode: buf.read_u32::<BigEndian>()?,
dbnode: buf.read_u32::<BigEndian>()?,
relnode: buf.read_u32::<BigEndian>()?,
forknum: buf.read_u8()?,
};
let n_blocks = buf.read_u32::<BigEndian>()?;
Self::Nblocks(PagestreamNblocksResponse {
req: PagestreamNblocksRequest {
hdr: PagestreamRequest {
reqid,
request_lsn,
not_modified_since,
},
rel,
},
n_blocks,
})
Self::Nblocks(PagestreamNblocksResponse { n_blocks })
}
Tag::GetPage => {
let reqid = buf.read_u64::<BigEndian>()?;
let request_lsn = Lsn(buf.read_u64::<BigEndian>()?);
let not_modified_since = Lsn(buf.read_u64::<BigEndian>()?);
let rel = RelTag {
spcnode: buf.read_u32::<BigEndian>()?,
dbnode: buf.read_u32::<BigEndian>()?,
relnode: buf.read_u32::<BigEndian>()?,
forknum: buf.read_u8()?,
};
let blkno = buf.read_u32::<BigEndian>()?;
let mut page = vec![0; 8192]; // TODO: use MaybeUninit
buf.read_exact(&mut page)?;
Self::GetPage(PagestreamGetPageResponse {
req: PagestreamGetPageRequest {
hdr: PagestreamRequest {
reqid,
request_lsn,
not_modified_since,
},
rel,
blkno,
},
page: page.into(),
})
PagestreamBeMessage::GetPage(PagestreamGetPageResponse { page: page.into() })
}
Tag::Error => {
let reqid = buf.read_u64::<BigEndian>()?;
let request_lsn = Lsn(buf.read_u64::<BigEndian>()?);
let not_modified_since = Lsn(buf.read_u64::<BigEndian>()?);
let mut msg = Vec::new();
buf.read_until(0, &mut msg)?;
let cstring = std::ffi::CString::from_vec_with_nul(msg)?;
let rust_str = cstring.to_str()?;
Self::Error(PagestreamErrorResponse {
req: PagestreamRequest {
reqid,
request_lsn,
not_modified_since,
},
PagestreamBeMessage::Error(PagestreamErrorResponse {
message: rust_str.to_owned(),
})
}
Tag::DbSize => {
let reqid = buf.read_u64::<BigEndian>()?;
let request_lsn = Lsn(buf.read_u64::<BigEndian>()?);
let not_modified_since = Lsn(buf.read_u64::<BigEndian>()?);
let dbnode = buf.read_u32::<BigEndian>()?;
let db_size = buf.read_i64::<BigEndian>()?;
Self::DbSize(PagestreamDbSizeResponse {
req: PagestreamDbSizeRequest {
hdr: PagestreamRequest {
reqid,
request_lsn,
not_modified_since,
},
dbnode,
},
db_size,
})
Self::DbSize(PagestreamDbSizeResponse { db_size })
}
Tag::GetSlruSegment => {
let reqid = buf.read_u64::<BigEndian>()?;
let request_lsn = Lsn(buf.read_u64::<BigEndian>()?);
let not_modified_since = Lsn(buf.read_u64::<BigEndian>()?);
let kind = buf.read_u8()?;
let segno = buf.read_u32::<BigEndian>()?;
let n_blocks = buf.read_u32::<BigEndian>()?;
let mut segment = vec![0; n_blocks as usize * BLCKSZ as usize];
buf.read_exact(&mut segment)?;
Self::GetSlruSegment(PagestreamGetSlruSegmentResponse {
req: PagestreamGetSlruSegmentRequest {
hdr: PagestreamRequest {
reqid,
request_lsn,
not_modified_since,
},
kind,
segno,
},
segment: segment.into(),
})
}
#[cfg(feature = "testing")]
Tag::Test => {
let reqid = buf.read_u64::<BigEndian>()?;
let request_lsn = Lsn(buf.read_u64::<BigEndian>()?);
let not_modified_since = Lsn(buf.read_u64::<BigEndian>()?);
let batch_key = buf.read_u64::<BigEndian>()?;
let len = buf.read_u64::<BigEndian>()?;
let mut msg = vec![0; len as usize];
buf.read_exact(&mut msg)?;
let message = String::from_utf8(msg)?;
Self::Test(PagestreamTestResponse {
req: PagestreamTestRequest {
hdr: PagestreamRequest {
reqid,
request_lsn,
not_modified_since,
},
batch_key,
message,
},
})
}
};
let remaining = buf.into_inner();
if !remaining.is_empty() {
@@ -2115,25 +1700,6 @@ impl PagestreamBeMessage {
Self::Error(_) => "Error",
Self::DbSize(_) => "DbSize",
Self::GetSlruSegment(_) => "GetSlruSegment",
#[cfg(feature = "testing")]
Self::Test(_) => "Test",
}
}
}
#[derive(Debug, Serialize, Deserialize)]
pub struct PageTraceEvent {
pub key: CompactKey,
pub effective_lsn: Lsn,
pub time: SystemTime,
}
impl Default for PageTraceEvent {
fn default() -> Self {
Self {
key: Default::default(),
effective_lsn: Default::default(),
time: std::time::UNIX_EPOCH,
}
}
}
@@ -2150,11 +1716,8 @@ mod tests {
// Test serialization/deserialization of PagestreamFeMessage
let messages = vec![
PagestreamFeMessage::Exists(PagestreamExistsRequest {
hdr: PagestreamRequest {
reqid: 0,
request_lsn: Lsn(4),
not_modified_since: Lsn(3),
},
request_lsn: Lsn(4),
not_modified_since: Lsn(3),
rel: RelTag {
forknum: 1,
spcnode: 2,
@@ -2163,11 +1726,8 @@ mod tests {
},
}),
PagestreamFeMessage::Nblocks(PagestreamNblocksRequest {
hdr: PagestreamRequest {
reqid: 0,
request_lsn: Lsn(4),
not_modified_since: Lsn(4),
},
request_lsn: Lsn(4),
not_modified_since: Lsn(4),
rel: RelTag {
forknum: 1,
spcnode: 2,
@@ -2176,11 +1736,8 @@ mod tests {
},
}),
PagestreamFeMessage::GetPage(PagestreamGetPageRequest {
hdr: PagestreamRequest {
reqid: 0,
request_lsn: Lsn(4),
not_modified_since: Lsn(3),
},
request_lsn: Lsn(4),
not_modified_since: Lsn(3),
rel: RelTag {
forknum: 1,
spcnode: 2,
@@ -2190,19 +1747,14 @@ mod tests {
blkno: 7,
}),
PagestreamFeMessage::DbSize(PagestreamDbSizeRequest {
hdr: PagestreamRequest {
reqid: 0,
request_lsn: Lsn(4),
not_modified_since: Lsn(3),
},
request_lsn: Lsn(4),
not_modified_since: Lsn(3),
dbnode: 7,
}),
];
for msg in messages {
let bytes = msg.serialize();
let reconstructed =
PagestreamFeMessage::parse(&mut bytes.reader(), PagestreamProtocolVersion::V3)
.unwrap();
let reconstructed = PagestreamFeMessage::parse(&mut bytes.reader()).unwrap();
assert!(msg == reconstructed);
}
}

View File

@@ -31,8 +31,6 @@
//! - In a tenant with 4 shards, each shard has ShardCount(N), ShardNumber(i) where i in 0..N-1 (inclusive),
//! and their slugs are 0004, 0104, 0204, and 0304.
use std::hash::{Hash, Hasher};
use crate::{key::Key, models::ShardParameters};
use postgres_ffi::relfile_utils::INIT_FORKNUM;
use serde::{Deserialize, Serialize};
@@ -50,23 +48,6 @@ pub struct ShardIdentity {
layout: ShardLayout,
}
/// Hash implementation
///
/// The stripe size cannot change dynamically, so it can be ignored for efficiency reasons.
impl Hash for ShardIdentity {
fn hash<H: Hasher>(&self, state: &mut H) {
let ShardIdentity {
number,
count,
stripe_size: _,
layout: _,
} = self;
number.0.hash(state);
count.0.hash(state);
}
}
/// Stripe size in number of pages
#[derive(Clone, Copy, Serialize, Deserialize, Eq, PartialEq, Debug)]
pub struct ShardStripeSize(pub u32);
@@ -78,7 +59,7 @@ impl Default for ShardStripeSize {
}
/// Layout version: for future upgrades where we might change how the key->shard mapping works
#[derive(Clone, Copy, Serialize, Deserialize, Eq, PartialEq, Hash, Debug)]
#[derive(Clone, Copy, Serialize, Deserialize, Eq, PartialEq, Debug)]
pub struct ShardLayout(u8);
const LAYOUT_V1: ShardLayout = ShardLayout(1);
@@ -192,11 +173,7 @@ impl ShardIdentity {
/// Return true if the key should be stored on all shards, not just one.
pub fn is_key_global(&self, key: &Key) -> bool {
if key.is_slru_block_key()
|| key.is_slru_segment_size_key()
|| key.is_aux_file_key()
|| key.is_slru_dir_key()
{
if key.is_slru_block_key() || key.is_slru_segment_size_key() || key.is_aux_file_key() {
// Special keys that are only stored on shard 0
false
} else if key.is_rel_block_key() {

View File

@@ -9,11 +9,9 @@ regex.workspace = true
bytes.workspace = true
anyhow.workspace = true
crc32c.workspace = true
criterion.workspace = true
once_cell.workspace = true
log.workspace = true
memoffset.workspace = true
pprof.workspace = true
thiserror.workspace = true
serde.workspace = true
utils.workspace = true
@@ -26,7 +24,3 @@ postgres.workspace = true
[build-dependencies]
anyhow.workspace = true
bindgen.workspace = true
[[bench]]
name = "waldecoder"
harness = false

View File

@@ -1,26 +0,0 @@
## Benchmarks
To run benchmarks:
```sh
# All benchmarks.
cargo bench --package postgres_ffi
# Specific file.
cargo bench --package postgres_ffi --bench waldecoder
# Specific benchmark.
cargo bench --package postgres_ffi --bench waldecoder complete_record/size=1024
# List available benchmarks.
cargo bench --package postgres_ffi --benches -- --list
# Generate flamegraph profiles using pprof-rs, profiling for 10 seconds.
# Output in target/criterion/*/profile/flamegraph.svg.
cargo bench --package postgres_ffi --bench waldecoder complete_record/size=1024 -- --profile-time 10
```
Additional charts and statistics are available in `target/criterion/report/index.html`.
Benchmarks are automatically compared against the previous run. To compare against other runs, see
`--baseline` and `--save-baseline`.

View File

@@ -1,49 +0,0 @@
use std::ffi::CStr;
use criterion::{criterion_group, criterion_main, Bencher, Criterion};
use postgres_ffi::v17::wal_generator::LogicalMessageGenerator;
use postgres_ffi::v17::waldecoder_handler::WalStreamDecoderHandler;
use postgres_ffi::waldecoder::WalStreamDecoder;
use pprof::criterion::{Output, PProfProfiler};
use utils::lsn::Lsn;
const KB: usize = 1024;
// Register benchmarks with Criterion.
criterion_group!(
name = benches;
config = Criterion::default().with_profiler(PProfProfiler::new(100, Output::Flamegraph(None)));
targets = bench_complete_record,
);
criterion_main!(benches);
/// Benchmarks WalStreamDecoder::complete_record() for a logical message of varying size.
fn bench_complete_record(c: &mut Criterion) {
let mut g = c.benchmark_group("complete_record");
for size in [64, KB, 8 * KB, 128 * KB] {
// Kind of weird to change the group throughput per benchmark, but it's the only way
// to vary it per benchmark. It works.
g.throughput(criterion::Throughput::Bytes(size as u64));
g.bench_function(format!("size={size}"), |b| run_bench(b, size).unwrap());
}
fn run_bench(b: &mut Bencher, size: usize) -> anyhow::Result<()> {
const PREFIX: &CStr = c"";
let value_size = LogicalMessageGenerator::make_value_size(size, PREFIX);
let value = vec![1; value_size];
let mut decoder = WalStreamDecoder::new(Lsn(0), 170000);
let msg = LogicalMessageGenerator::new(PREFIX, &value)
.next()
.unwrap()
.encode(Lsn(0));
assert_eq!(msg.len(), size);
b.iter(|| {
let msg = msg.clone(); // Bytes::clone() is cheap
decoder.complete_record(msg).unwrap();
});
Ok(())
}
}

View File

@@ -106,11 +106,11 @@ impl<R: RecordGenerator> WalGenerator<R> {
const TIMELINE_ID: u32 = 1;
/// Creates a new WAL generator with the given record generator.
pub fn new(record_generator: R, start_lsn: Lsn) -> WalGenerator<R> {
pub fn new(record_generator: R) -> WalGenerator<R> {
Self {
record_generator,
lsn: start_lsn,
prev_lsn: start_lsn,
lsn: Lsn(0),
prev_lsn: Lsn(0),
}
}
@@ -231,22 +231,6 @@ impl LogicalMessageGenerator {
};
[&header.encode(), prefix, message].concat().into()
}
/// Computes how large a value must be to get a record of the given size. Convenience method to
/// construct records of pre-determined size. Panics if the record size is too small.
pub fn make_value_size(record_size: usize, prefix: &CStr) -> usize {
let xlog_header_size = XLOG_SIZE_OF_XLOG_RECORD;
let lm_header_size = size_of::<XlLogicalMessage>();
let prefix_size = prefix.to_bytes_with_nul().len();
let data_header_size = match record_size - xlog_header_size - 2 {
0..=255 => 2,
256..=258 => panic!("impossible record_size {record_size}"),
259.. => 5,
};
record_size
.checked_sub(xlog_header_size + lm_header_size + prefix_size + data_header_size)
.expect("record_size too small")
}
}
impl Iterator for LogicalMessageGenerator {

View File

@@ -16,7 +16,7 @@ use utils::bin_ser::DeserializeError;
use utils::lsn::Lsn;
#[repr(C)]
#[derive(Clone, Debug, Serialize, Deserialize)]
#[derive(Debug, Serialize, Deserialize)]
pub struct XlMultiXactCreate {
pub mid: MultiXactId,
/* new MultiXact's ID */
@@ -46,7 +46,7 @@ impl XlMultiXactCreate {
}
#[repr(C)]
#[derive(Clone, Debug, Serialize, Deserialize)]
#[derive(Debug, Serialize, Deserialize)]
pub struct XlMultiXactTruncate {
pub oldest_multi_db: Oid,
/* to-be-truncated range of multixact offsets */
@@ -72,7 +72,7 @@ impl XlMultiXactTruncate {
}
#[repr(C)]
#[derive(Clone, Debug, Serialize, Deserialize)]
#[derive(Debug, Serialize, Deserialize)]
pub struct XlRelmapUpdate {
pub dbid: Oid, /* database ID, or 0 for shared map */
pub tsid: Oid, /* database's tablespace, or pg_global */
@@ -90,7 +90,7 @@ impl XlRelmapUpdate {
}
#[repr(C)]
#[derive(Clone, Debug, Serialize, Deserialize)]
#[derive(Debug, Serialize, Deserialize)]
pub struct XlReploriginDrop {
pub node_id: RepOriginId,
}
@@ -104,7 +104,7 @@ impl XlReploriginDrop {
}
#[repr(C)]
#[derive(Clone, Debug, Serialize, Deserialize)]
#[derive(Debug, Serialize, Deserialize)]
pub struct XlReploriginSet {
pub remote_lsn: Lsn,
pub node_id: RepOriginId,
@@ -911,7 +911,7 @@ impl XlSmgrCreate {
}
#[repr(C)]
#[derive(Clone, Debug, Serialize, Deserialize)]
#[derive(Debug, Serialize, Deserialize)]
pub struct XlSmgrTruncate {
pub blkno: BlockNumber,
pub rnode: RelFileNode,
@@ -984,7 +984,7 @@ impl XlDropDatabase {
/// xl_xact_parsed_abort structs in PostgreSQL, but we use the same
/// struct for commits and aborts.
///
#[derive(Clone, Debug, Serialize, Deserialize)]
#[derive(Debug, Serialize, Deserialize)]
pub struct XlXactParsedRecord {
pub xid: TransactionId,
pub info: u8,

View File

@@ -81,7 +81,7 @@ fn test_end_of_wal<C: crate::Crafter>(test_name: &str) {
continue;
}
let mut f = File::options().write(true).open(file.path()).unwrap();
static ZEROS: [u8; WAL_SEGMENT_SIZE] = [0u8; WAL_SEGMENT_SIZE];
const ZEROS: [u8; WAL_SEGMENT_SIZE] = [0u8; WAL_SEGMENT_SIZE];
f.write_all(
&ZEROS[0..min(
WAL_SEGMENT_SIZE,

View File

@@ -44,7 +44,7 @@ pub struct ProtocolVersion(u32);
impl ProtocolVersion {
pub const fn new(major: u16, minor: u16) -> Self {
Self(((major as u32) << 16) | minor as u32)
Self((major as u32) << 16 | minor as u32)
}
pub const fn minor(self) -> u16 {
self.0 as u16

View File

@@ -1,7 +1,7 @@
[package]
name = "postgres-protocol2"
version = "0.1.0"
edition = "2021"
edition = "2018"
license = "MIT/Apache-2.0"
[dependencies]

View File

@@ -9,7 +9,8 @@
//!
//! This library assumes that the `client_encoding` backend parameter has been
//! set to `UTF8`. It will most likely not behave properly if that is not the case.
#![warn(missing_docs, clippy::all)]
#![doc(html_root_url = "https://docs.rs/postgres-protocol/0.6")]
#![warn(missing_docs, rust_2018_idioms, clippy::all)]
use byteorder::{BigEndian, ByteOrder};
use bytes::{BufMut, BytesMut};

View File

@@ -3,6 +3,7 @@
use byteorder::{BigEndian, ByteOrder};
use bytes::{Buf, BufMut, BytesMut};
use std::convert::TryFrom;
use std::error::Error;
use std::io;
use std::marker;

View File

@@ -1,7 +1,7 @@
[package]
name = "postgres-types2"
version = "0.1.0"
edition = "2021"
edition = "2018"
license = "MIT/Apache-2.0"
[dependencies]

View File

@@ -2,7 +2,8 @@
//!
//! This crate is used by the `tokio-postgres` and `postgres` crates. You normally don't need to depend directly on it
//! unless you want to define your own `ToSql` or `FromSql` definitions.
#![warn(clippy::all, missing_docs)]
#![doc(html_root_url = "https://docs.rs/postgres-types/0.2")]
#![warn(clippy::all, rust_2018_idioms, missing_docs)]
use fallible_iterator::FallibleIterator;
use postgres_protocol2::types;

View File

@@ -1,7 +1,7 @@
[package]
name = "tokio-postgres2"
version = "0.1.0"
edition = "2021"
edition = "2018"
license = "MIT/Apache-2.0"
[dependencies]

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