DO NOT MERGE: test cache-from

We log something at info level each time we schedule a shard to a
non-secondary location.

Might as well have context for it.

.dockerignore

@@ -13,6 +13,7 @@
 # Directories
 !.cargo/
 !.config/
+!compute/
 !compute_tools/
 !control_plane/
 !libs/

.github/workflows/_build-and-test-locally.yml

@@ -62,7 +62,7 @@ jobs:
           #
           git config --global --add safe.directory ${{ github.workspace }}
           git config --global --add safe.directory ${GITHUB_WORKSPACE}
-          for r in 14 15 16; do
+          for r in 14 15 16 17; do
             git config --global --add safe.directory "${{ github.workspace }}/vendor/postgres-v$r"
             git config --global --add safe.directory "${GITHUB_WORKSPACE}/vendor/postgres-v$r"
           done
@@ -83,6 +83,10 @@ jobs:
         id: pg_v16_rev
         run: echo pg_rev=$(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) >> $GITHUB_OUTPUT
+
       # Set some environment variables used by all the steps.
       #
       # CARGO_FLAGS is extra options to pass to "cargo build", "cargo test" etc.
@@ -136,6 +140,13 @@ jobs:
           path: pg_install/v16
           key: v1-${{ runner.os }}-${{ runner.arch }}-${{ inputs.build-type }}-pg-${{ steps.pg_v16_rev.outputs.pg_rev }}-${{ hashFiles('Makefile', 'Dockerfile.build-tools') }}
 
+      - name: Cache postgres v17 build
+        id: cache_pg_17
+        uses: actions/cache@v4
+        with:
+          path: pg_install/v17
+          key: v1-${{ runner.os }}-${{ runner.arch }}-${{ inputs.build-type }}-pg-${{ steps.pg_v17_rev.outputs.pg_rev }}-${{ hashFiles('Makefile', 'Dockerfile.build-tools') }}
+
       - name: Build postgres v14
         if: steps.cache_pg_14.outputs.cache-hit != 'true'
         run: mold -run make postgres-v14 -j$(nproc)
@@ -148,6 +159,10 @@ jobs:
         if: steps.cache_pg_16.outputs.cache-hit != 'true'
         run: mold -run make postgres-v16 -j$(nproc)
 
+      - name: Build postgres v17
+        if: steps.cache_pg_17.outputs.cache-hit != 'true'
+        run: mold -run make postgres-v17 -j$(nproc)
+
       - name: Build neon extensions
         run: mold -run make neon-pg-ext -j$(nproc)
 
@@ -210,7 +225,7 @@ jobs:
         run: |
           PQ_LIB_DIR=$(pwd)/pg_install/v16/lib
           export PQ_LIB_DIR
-          LD_LIBRARY_PATH=$(pwd)/pg_install/v16/lib
+          LD_LIBRARY_PATH=$(pwd)/pg_install/v17/lib
           export LD_LIBRARY_PATH
 
           #nextest does not yet support running doctests

.github/workflows/_push-to-acr.yml

@@ -52,5 +52,5 @@ jobs:
           for image in ${images}; do
             docker buildx imagetools create \
               -t ${{ inputs.registry_name }}.azurecr.io/neondatabase/${image}:${{ inputs.image_tag }} \
-                                        neondatabase/${image}:${{ inputs.image_tag }}
+                                                        neondatabase/${image}:${{ inputs.image_tag }}
           done

.github/workflows/build-build-tools-image.yml

@@ -36,7 +36,7 @@ jobs:
 
     strategy:
       matrix:
-        arch: [ x64, arm64 ]
+        arch: [ x64 ]
 
     runs-on: ${{ fromJson(format('["self-hosted", "{0}"]', matrix.arch == 'arm64' && 'large-arm64' || 'large')) }}
 
@@ -79,7 +79,9 @@ jobs:
           push: true
           pull: true
           file: Dockerfile.build-tools
-          cache-from: type=registry,ref=cache.neon.build/build-tools:cache-${{ matrix.arch }}
+          cache-from: |
+            type=registry,ref=cache.neon.build/build-tools:cache-${{ matrix.arch }}
+            type=registry,ref=cache.neon.build/build-tools:cache-${{ matrix.arch }}-does-not-exist-2
           cache-to: ${{ github.ref_name == 'main' && format('type=registry,ref=cache.neon.build/build-tools:cache-{0},mode=max', matrix.arch) || '' }}
           tags: neondatabase/build-tools:${{ inputs.image-tag }}-${{ matrix.arch }}
 

.github/workflows/build_and_test.yml

@@ -6,6 +6,7 @@ on:
       - main
       - release
       - release-proxy
+      - bayandin/test
   pull_request:
 
 defaults:
@@ -27,7 +28,7 @@ env:
 
 jobs:
   check-permissions:
-    if: ${{ !contains(github.event.pull_request.labels.*.name, 'run-no-ci') }}
+    if: false
     uses: ./.github/workflows/check-permissions.yml
     with:
       github-event-name: ${{ github.event_name }}
@@ -54,8 +55,8 @@ jobs:
       build-tag: ${{steps.build-tag.outputs.tag}}
 
     steps:
-      - name: Checkout
-        uses: actions/checkout@v4
+      # Need `fetch-depth: 0` to count the number of commits in the branch
+      - uses: actions/checkout@v4
         with:
           fetch-depth: 0
 
@@ -78,7 +79,6 @@ jobs:
         id: build-tag
 
   check-build-tools-image:
-    needs: [ check-permissions ]
     uses: ./.github/workflows/check-build-tools-image.yml
 
   build-build-tools-image:
@@ -120,6 +120,59 @@ jobs:
       - name: Run mypy to check types
         run: poetry run mypy .
 
+  # Check that the vendor/postgres-* submodules point to the
+  # corresponding REL_*_STABLE_neon branches.
+  check-submodules:
+    runs-on: ubuntu-22.04
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+        with:
+          submodules: true
+
+      - uses: dorny/paths-filter@v3
+        id: check-if-submodules-changed
+        with:
+          filters: |
+            vendor:
+              - 'vendor/**'
+
+      - name: Check vendor/postgres-v14 submodule reference
+        if: steps.check-if-submodules-changed.outputs.vendor == 'true'
+        uses: jtmullen/submodule-branch-check-action@v1
+        with:
+          path: "vendor/postgres-v14"
+          fetch_depth: "50"
+          sub_fetch_depth: "50"
+          pass_if_unchanged: true
+
+      - name: Check vendor/postgres-v15 submodule reference
+        if: steps.check-if-submodules-changed.outputs.vendor == 'true'
+        uses: jtmullen/submodule-branch-check-action@v1
+        with:
+          path: "vendor/postgres-v15"
+          fetch_depth: "50"
+          sub_fetch_depth: "50"
+          pass_if_unchanged: true
+
+      - name: Check vendor/postgres-v16 submodule reference
+        if: steps.check-if-submodules-changed.outputs.vendor == 'true'
+        uses: jtmullen/submodule-branch-check-action@v1
+        with:
+          path: "vendor/postgres-v16"
+          fetch_depth: "50"
+          sub_fetch_depth: "50"
+          pass_if_unchanged: true
+
+      - name: Check vendor/postgres-v17 submodule reference
+        if: steps.check-if-submodules-changed.outputs.vendor == 'true'
+        uses: jtmullen/submodule-branch-check-action@v1
+        with:
+          path: "vendor/postgres-v17"
+          fetch_depth: "50"
+          sub_fetch_depth: "50"
+          pass_if_unchanged: true
+
   check-codestyle-rust:
     needs: [ check-permissions, build-build-tools-image ]
     strategy:
@@ -159,6 +212,10 @@ jobs:
       # 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 [ "$CLIPPY_COMMON_ARGS" = "" ]; then
@@ -168,8 +225,6 @@ jobs:
           echo "CLIPPY_COMMON_ARGS=${CLIPPY_COMMON_ARGS}" >> $GITHUB_ENV
       - name: Run cargo clippy (debug)
         run: cargo hack --feature-powerset clippy $CLIPPY_COMMON_ARGS
-      - name: Run cargo clippy (release)
-        run: cargo hack --feature-powerset clippy --release $CLIPPY_COMMON_ARGS
 
       - name: Check documentation generation
         run: cargo doc --workspace --no-deps --document-private-items
@@ -211,7 +266,7 @@ jobs:
       build-tag: ${{ needs.tag.outputs.build-tag }}
       build-type: ${{ matrix.build-type }}
       # Run tests on all Postgres versions in release builds and only on the latest version in debug builds
-      pg-versions: ${{ matrix.build-type == 'release' && '["v14", "v15", "v16"]' || '["v16"]' }}
+      pg-versions: ${{ matrix.build-type == 'release' && '["v14", "v15", "v16", "v17"]' || '["v17"]' }}
     secrets: inherit
 
   # Keep `benchmarks` job outside of `build-and-test-locally` workflow to make job failures non-blocking
@@ -357,6 +412,7 @@ jobs:
             })
 
   coverage-report:
+    if: ${{ !startsWith(github.ref_name, 'release') }}
     needs: [ check-permissions, build-build-tools-image, build-and-test-locally ]
     runs-on: [ self-hosted, small ]
     container:
@@ -373,8 +429,8 @@ jobs:
         coverage-html: ${{ steps.upload-coverage-report-new.outputs.report-url }}
         coverage-json: ${{ steps.upload-coverage-report-new.outputs.summary-json }}
     steps:
-      - name: Checkout
-        uses: actions/checkout@v4
+      # Need `fetch-depth: 0` for differential coverage (to get diff between two commits)
+      - uses: actions/checkout@v4
         with:
           submodules: true
           fetch-depth: 0
@@ -475,11 +531,9 @@ jobs:
     runs-on: ${{ fromJson(format('["self-hosted", "{0}"]', matrix.arch == 'arm64' && 'large-arm64' || 'large')) }}
 
     steps:
-      - name: Checkout
-        uses: actions/checkout@v4
+      - uses: actions/checkout@v4
         with:
           submodules: true
-          fetch-depth: 0
 
       - uses: ./.github/actions/set-docker-config-dir
       - uses: docker/setup-buildx-action@v3
@@ -548,17 +602,15 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        version: [ v14, v15, v16 ]
+        version: [ v14, v15, v16, v17 ]
         arch: [ x64, arm64 ]
 
     runs-on: ${{ fromJson(format('["self-hosted", "{0}"]', matrix.arch == 'arm64' && 'large-arm64' || 'large')) }}
 
     steps:
-      - name: Checkout
-        uses: actions/checkout@v4
+      - uses: actions/checkout@v4
         with:
           submodules: true
-          fetch-depth: 0
 
       - uses: ./.github/actions/set-docker-config-dir
       - uses: docker/setup-buildx-action@v3
@@ -599,7 +651,7 @@ jobs:
           provenance: false
           push: true
           pull: true
-          file: Dockerfile.compute-node
+          file: compute/Dockerfile.compute-node
           cache-from: type=registry,ref=cache.neon.build/compute-node-${{ matrix.version }}:cache-${{ matrix.arch }}
           cache-to: ${{ github.ref_name == 'main' && format('type=registry,ref=cache.neon.build/compute-node-{0}:cache-{1},mode=max', matrix.version, matrix.arch) || '' }}
           tags: |
@@ -618,7 +670,7 @@ jobs:
           provenance: false
           push: true
           pull: true
-          file: Dockerfile.compute-node
+          file: compute/Dockerfile.compute-node
           target: neon-pg-ext-test
           cache-from: type=registry,ref=cache.neon.build/neon-test-extensions-${{ matrix.version }}:cache-${{ matrix.arch }}
           cache-to: ${{ github.ref_name == 'main' && format('type=registry,ref=cache.neon.build/neon-test-extensions-{0}:cache-{1},mode=max', matrix.version, matrix.arch) || '' }}
@@ -627,7 +679,7 @@ jobs:
 
       - name: Build compute-tools image
         # compute-tools are Postgres independent, so build it only once
-        if: matrix.version == 'v16'
+        if: matrix.version == 'v17'
         uses: docker/build-push-action@v6
         with:
           target: compute-tools-image
@@ -639,7 +691,7 @@ jobs:
           provenance: false
           push: true
           pull: true
-          file: Dockerfile.compute-node
+          file: compute/Dockerfile.compute-node
           tags: |
             neondatabase/compute-tools:${{ needs.tag.outputs.build-tag }}-${{ matrix.arch }}
 
@@ -649,7 +701,7 @@ jobs:
 
     strategy:
       matrix:
-        version: [ v14, v15, v16 ]
+        version: [ v14, v15, v16, v17 ]
 
     steps:
       - uses: docker/login-action@v3
@@ -671,7 +723,7 @@ jobs:
                                              neondatabase/neon-test-extensions-${{ matrix.version }}:${{ needs.tag.outputs.build-tag }}-arm64
 
       - name: Create multi-arch compute-tools image
-        if: matrix.version == 'v16'
+        if: matrix.version == 'v17'
         run: |
           docker buildx imagetools create -t neondatabase/compute-tools:${{ needs.tag.outputs.build-tag }} \
                                              neondatabase/compute-tools:${{ needs.tag.outputs.build-tag }}-x64 \
@@ -689,7 +741,7 @@ jobs:
                                                                                 neondatabase/compute-node-${{ matrix.version }}:${{ needs.tag.outputs.build-tag }}
 
       - name: Push multi-arch compute-tools image to ECR
-        if: matrix.version == 'v16'
+        if: matrix.version == 'v17'
         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 }}
@@ -700,15 +752,12 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        version: [ v14, v15, v16 ]
+        version: [ v14, v15, v16, v17 ]
     env:
       VM_BUILDER_VERSION: v0.29.3
 
     steps:
-      - name: Checkout
-        uses: actions/checkout@v4
-        with:
-          fetch-depth: 0
+      - uses: actions/checkout@v4
 
       - name: Downloading vm-builder
         run: |
@@ -730,7 +779,7 @@ jobs:
       - name: Build vm image
         run: |
           ./vm-builder \
-            -spec=vm-image-spec.yaml \
+            -spec=compute/vm-image-spec.yaml \
             -src=neondatabase/compute-node-${{ matrix.version }}:${{ needs.tag.outputs.build-tag }} \
             -dst=neondatabase/vm-compute-node-${{ matrix.version }}:${{ needs.tag.outputs.build-tag }}
 
@@ -748,10 +797,7 @@ jobs:
     runs-on: ${{ fromJson(format('["self-hosted", "{0}"]', matrix.arch == 'arm64' && 'small-arm64' || 'small')) }}
 
     steps:
-      - name: Checkout
-        uses: actions/checkout@v4
-        with:
-          fetch-depth: 0
+      - uses: actions/checkout@v4
 
       - uses: ./.github/actions/set-docker-config-dir
       - uses: docker/login-action@v3
@@ -798,7 +844,7 @@ jobs:
     runs-on: ubuntu-22.04
 
     env:
-      VERSIONS: v14 v15 v16
+      VERSIONS: v14 v15 v16 v17
 
     steps:
       - uses: docker/login-action@v3
@@ -839,7 +885,7 @@ jobs:
             done
           done
           docker buildx imagetools create -t neondatabase/neon-test-extensions-v16:latest \
-                                             neondatabase/neon-test-extensions-v16:${{ needs.tag.outputs.build-tag }}
+                                              neondatabase/neon-test-extensions-v16:${{ needs.tag.outputs.build-tag }}
 
       - name: Login to prod ECR
         uses: docker/login-action@v3
@@ -852,7 +898,7 @@ jobs:
       - name: Copy all images to prod ECR
         if: github.ref_name == 'release'|| github.ref_name == 'release-proxy'
         run: |
-          for image in neon compute-tools {vm-,}compute-node-{v14,v15,v16}; 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
@@ -864,7 +910,7 @@ jobs:
     with:
       client_id: ${{ vars.AZURE_DEV_CLIENT_ID }}
       image_tag: ${{ needs.tag.outputs.build-tag }}
-      images: neon compute-tools vm-compute-node-v14 vm-compute-node-v15 vm-compute-node-v16 compute-node-v14 compute-node-v15 compute-node-v16
+      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 }}
@@ -876,7 +922,7 @@ jobs:
     with:
       client_id: ${{ vars.AZURE_PROD_CLIENT_ID }}
       image_tag: ${{ needs.tag.outputs.build-tag }}
-      images: neon compute-tools vm-compute-node-v14 vm-compute-node-v15 vm-compute-node-v16 compute-node-v14 compute-node-v15 compute-node-v16
+      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 }}
@@ -957,6 +1003,7 @@ jobs:
 
   deploy:
     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') && !failure() && !cancelled()
 
     runs-on: [ self-hosted, small ]
@@ -971,15 +1018,12 @@ jobs:
           #
           git config --global --add safe.directory ${{ github.workspace }}
           git config --global --add safe.directory ${GITHUB_WORKSPACE}
-          for r in 14 15 16; do
+          for r in 14 15 16 17; do
             git config --global --add safe.directory "${{ github.workspace }}/vendor/postgres-v$r"
             git config --global --add safe.directory "${GITHUB_WORKSPACE}/vendor/postgres-v$r"
           done
 
-      - name: Checkout
-        uses: actions/checkout@v4
-        with:
-          fetch-depth: 0
+      - uses: actions/checkout@v4
 
       - name: Trigger deploy workflow
         env:
@@ -1058,7 +1102,8 @@ jobs:
   # The job runs on `release` branch and copies compatibility data and Neon artifact from the last *release PR* to the latest directory
   promote-compatibility-data:
     needs: [ deploy ]
-    if: github.ref_name == 'release'
+    # `!failure() && !cancelled()` is required because the workflow transitively depends on the job that can be skipped: `push-to-acr-dev` and `push-to-acr-prod`
+    if: github.ref_name == 'release' && !failure() && !cancelled()
 
     runs-on: ubuntu-22.04
     steps:
@@ -1117,6 +1162,7 @@ jobs:
 
               files_to_promote+=("s3://${BUCKET}/${s3_key}")
 
+              # TODO Add v17
               for pg_version in v14 v15 v16; do
                 # We run less tests for debug builds, so we don't need to promote them
                 if [ "${build_type}" == "debug" ] && { [ "${arch}" == "ARM64" ] || [ "${pg_version}" != "v16" ] ; }; then

.github/workflows/check-build-tools-image.yml

@@ -33,7 +33,7 @@ jobs:
           IMAGE_TAG: |
             ${{ hashFiles('Dockerfile.build-tools',
                           '.github/workflows/check-build-tools-image.yml',
-                          '.github/workflows/build-build-tools-image.yml') }}
+                          '.github/workflows/build-build-tools-image.yml') }}-test
         run: |
           echo "image-tag=${IMAGE_TAG}" | tee -a $GITHUB_OUTPUT
 

.github/workflows/cloud-regress.yml

@@ -0,0 +1,102 @@
+name: Cloud Regression Test
+on:
+  schedule:
+    # * is a special character in YAML so you have to quote this string
+    #          ┌───────────── minute (0 - 59)
+    #          │ ┌───────────── hour (0 - 23)
+    #          │ │ ┌───────────── day of the month (1 - 31)
+    #          │ │ │ ┌───────────── month (1 - 12 or JAN-DEC)
+    #          │ │ │ │ ┌───────────── day of the week (0 - 6 or SUN-SAT)
+    - cron:  '45 1 * * *' # run once a day, timezone is utc
+  workflow_dispatch: # adds ability to run this manually
+
+defaults:
+  run:
+    shell: bash -euxo pipefail {0}
+
+concurrency:
+  # Allow only one workflow
+  group: ${{ github.workflow }}
+  cancel-in-progress: true
+
+jobs:
+  regress:
+    env:
+      POSTGRES_DISTRIB_DIR: /tmp/neon/pg_install
+      DEFAULT_PG_VERSION: 16
+      TEST_OUTPUT: /tmp/test_output
+      BUILD_TYPE: remote
+      AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_DEV }}
+      AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_KEY_DEV }}
+
+    runs-on: us-east-2
+    container:
+      image: neondatabase/build-tools:pinned
+      options: --init
+
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          submodules: true
+
+      - name: Patch the test
+        run: |
+          cd "vendor/postgres-v${DEFAULT_PG_VERSION}"
+          patch -p1 < "../../compute/patches/cloud_regress_pg${DEFAULT_PG_VERSION}.patch"
+
+      - name: Generate a random password
+        id: pwgen
+        run: |
+          set +x
+          DBPASS=$(dd if=/dev/random bs=48 count=1 2>/dev/null | base64)
+          echo "::add-mask::${DBPASS//\//}"
+          echo DBPASS="${DBPASS//\//}" >> "${GITHUB_OUTPUT}"
+
+      - name: Change tests according to the generated password
+        env:
+          DBPASS: ${{ steps.pwgen.outputs.DBPASS }}
+        run: |
+          cd vendor/postgres-v"${DEFAULT_PG_VERSION}"/src/test/regress
+          for fname in sql/*.sql expected/*.out; do
+            sed -i.bak s/NEON_PASSWORD_PLACEHOLDER/"'${DBPASS}'"/ "${fname}"
+          done
+          for ph in $(grep NEON_MD5_PLACEHOLDER expected/password.out | awk '{print $3;}' | sort | uniq); do
+            USER=$(echo "${ph}" | cut -c 22-)
+            MD5=md5$(echo -n "${DBPASS}${USER}" | md5sum | awk '{print $1;}')
+            sed -i.bak "s/${ph}/${MD5}/" expected/password.out
+          done
+
+      - name: Download Neon artifact
+        uses: ./.github/actions/download
+        with:
+          name: neon-${{ runner.os }}-${{ runner.arch }}-release-artifact
+          path: /tmp/neon/
+          prefix: latest
+
+      - name: Run the regression tests
+        uses: ./.github/actions/run-python-test-set
+        with:
+          build_type: ${{ env.BUILD_TYPE }}
+          test_selection: cloud_regress
+          pg_version: ${{ env.DEFAULT_PG_VERSION }}
+          extra_params: -m remote_cluster
+        env:
+          BENCHMARK_CONNSTR: ${{ secrets.PG_REGRESS_CONNSTR }}
+
+      - name: Create Allure report
+        id: create-allure-report
+        if: ${{ !cancelled() }}
+        uses: ./.github/actions/allure-report-generate
+
+      - name: Post to a Slack channel
+        if: ${{ github.event.schedule && failure() }}
+        uses: slackapi/slack-github-action@v1
+        with:
+          channel-id: "C033QLM5P7D" # on-call-staging-stream
+          slack-message: |
+            Periodic pg_regress on staging: ${{ job.status }}
+            <${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}|GitHub Run>
+            <${{ steps.create-allure-report.outputs.report-url }}|Allure report>
+        env:
+          SLACK_BOT_TOKEN: ${{ secrets.SLACK_BOT_TOKEN }}
+

.github/workflows/neon_extra_builds.yml

@@ -72,6 +72,10 @@ jobs:
         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) >> $GITHUB_OUTPUT
+
       - name: Cache postgres v14 build
         id: cache_pg_14
         uses: actions/cache@v4
@@ -93,6 +97,13 @@ jobs:
           path: pg_install/v16
           key: v1-${{ 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: |
           echo 'LDFLAGS=-L/usr/local/opt/openssl@3/lib' >> $GITHUB_ENV
@@ -120,6 +131,10 @@ jobs:
         if: steps.cache_pg_16.outputs.cache-hit != 'true'
         run: make postgres-v16 -j$(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)
 
@@ -166,7 +181,7 @@ jobs:
         run: make walproposer-lib -j$(nproc)
 
       - name: Produce the build stats
-        run: PQ_LIB_DIR=$(pwd)/pg_install/v16/lib cargo build --all --release --timings -j$(nproc)
+        run: PQ_LIB_DIR=$(pwd)/pg_install/v17/lib cargo build --all --release --timings -j$(nproc)
 
       - name: Upload the build stats
         id: upload-stats

.github/workflows/trigger-e2e-tests.yml

@@ -34,8 +34,8 @@ jobs:
       build-tag: ${{ steps.build-tag.outputs.tag }}
 
     steps:
-      - name: Checkout
-        uses: actions/checkout@v4
+      # Need `fetch-depth: 0` to count the number of commits in the branch
+      - uses: actions/checkout@v4
         with:
           fetch-depth: 0
 
@@ -107,7 +107,7 @@ jobs:
           if [ "$GITHUB_EVENT_NAME" == "pull_request" ]; then
             for f in $(gh api "/repos/${GITHUB_REPOSITORY}/pulls/${PR_NUMBER}/files" --paginate --jq '.[].filename'); do
               case "$f" in
-                vendor/*|pgxn/*|libs/vm_monitor/*|Dockerfile.compute-node)
+                vendor/*|pgxn/*|libs/vm_monitor/*|compute/Dockerfile.compute-node)
                   platforms=$(echo "${platforms}" | jq --compact-output '. += ["k8s-neonvm"] | unique')
                   ;;
                 *)

.gitmodules

@@ -5,8 +5,12 @@
 [submodule "vendor/postgres-v15"]
 	path = vendor/postgres-v15
 	url = https://github.com/neondatabase/postgres.git
-	branch = tristan957/15/pg_upgrade
+	branch = REL_15_STABLE_neon
 [submodule "vendor/postgres-v16"]
 	path = vendor/postgres-v16
 	url = https://github.com/neondatabase/postgres.git
-	branch = tristan957/pg_upgrade
+	branch = REL_16_STABLE_neon
+[submodule "vendor/postgres-v17"]
+	path = vendor/postgres-v17
+	url = https://github.com/neondatabase/postgres.git
+	branch = REL_17_STABLE_neon

Cargo.toml

@@ -76,8 +76,6 @@ clap = { version = "4.0", features = ["derive"] }
 comfy-table = "7.1"
 const_format = "0.2"
 crc32c = "0.6"
-crossbeam-deque = "0.8.5"
-crossbeam-utils = "0.8.5"
 dashmap = { version = "5.5.0", features = ["raw-api"] }
 either = "1.8"
 enum-map = "2.4.2"
@@ -95,7 +93,7 @@ hdrhistogram = "7.5.2"
 hex = "0.4"
 hex-literal = "0.4"
 hmac = "0.12.1"
-hostname = "0.3.1"
+hostname = "0.4"
 http = {version = "1.1.0", features = ["std"]}
 http-types = { version = "2", default-features = false }
 humantime = "2.1"
@@ -104,7 +102,6 @@ hyper = "0.14"
 tokio-tungstenite = "0.20.0"
 indexmap = "2"
 indoc = "2"
-inotify = "0.10.2"
 ipnet = "2.9.0"
 itertools = "0.10"
 jsonwebtoken = "9"
@@ -113,7 +110,7 @@ libc = "0.2"
 md5 = "0.7.0"
 measured = { version = "0.0.22", features=["lasso"] }
 measured-process = { version = "0.0.22" }
-memoffset = "0.8"
+memoffset = "0.9"
 nix = { version = "0.27", features = ["dir", "fs", "process", "socket", "signal", "poll"] }
 notify = "6.0.0"
 num_cpus = "1.15"
@@ -142,7 +139,6 @@ rpds = "0.13"
 rustc-hash = "1.1.0"
 rustls = "0.22"
 rustls-pemfile = "2"
-rustls-split = "0.3"
 scopeguard = "1.1"
 sysinfo = "0.29.2"
 sd-notify = "0.4.1"
@@ -164,7 +160,6 @@ strum_macros = "0.26"
 svg_fmt = "0.4.3"
 sync_wrapper = "0.1.2"
 tar = "0.4"
-task-local-extensions = "0.1.4"
 test-context = "0.3"
 thiserror = "1.0"
 tikv-jemallocator = "0.5"

Dockerfile

@@ -5,6 +5,8 @@
 ARG REPOSITORY=neondatabase
 ARG IMAGE=build-tools
 ARG TAG=pinned
+ARG DEFAULT_PG_VERSION=17
+ARG STABLE_PG_VERSION=16
 
 # Build Postgres
 FROM $REPOSITORY/$IMAGE:$TAG AS pg-build
@@ -13,6 +15,7 @@ WORKDIR /home/nonroot
 COPY --chown=nonroot vendor/postgres-v14 vendor/postgres-v14
 COPY --chown=nonroot vendor/postgres-v15 vendor/postgres-v15
 COPY --chown=nonroot vendor/postgres-v16 vendor/postgres-v16
+COPY --chown=nonroot vendor/postgres-v17 vendor/postgres-v17
 COPY --chown=nonroot pgxn pgxn
 COPY --chown=nonroot Makefile Makefile
 COPY --chown=nonroot scripts/ninstall.sh scripts/ninstall.sh
@@ -28,16 +31,19 @@ FROM $REPOSITORY/$IMAGE:$TAG AS build
 WORKDIR /home/nonroot
 ARG GIT_VERSION=local
 ARG BUILD_TAG
+ARG STABLE_PG_VERSION
 
 COPY --from=pg-build /home/nonroot/pg_install/v14/include/postgresql/server pg_install/v14/include/postgresql/server
 COPY --from=pg-build /home/nonroot/pg_install/v15/include/postgresql/server pg_install/v15/include/postgresql/server
 COPY --from=pg-build /home/nonroot/pg_install/v16/include/postgresql/server pg_install/v16/include/postgresql/server
+COPY --from=pg-build /home/nonroot/pg_install/v17/include/postgresql/server pg_install/v17/include/postgresql/server
 COPY --from=pg-build /home/nonroot/pg_install/v16/lib                       pg_install/v16/lib
+COPY --from=pg-build /home/nonroot/pg_install/v17/lib                       pg_install/v17/lib
 COPY --chown=nonroot . .
 
 ARG ADDITIONAL_RUSTFLAGS
 RUN set -e \
-    && PQ_LIB_DIR=$(pwd)/pg_install/v16/lib RUSTFLAGS="-Clinker=clang -Clink-arg=-fuse-ld=mold -Clink-arg=-Wl,--no-rosegment ${ADDITIONAL_RUSTFLAGS}" cargo build \
+    && PQ_LIB_DIR=$(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  \
@@ -52,6 +58,7 @@ RUN set -e \
 # Build final image
 #
 FROM debian:bullseye-slim
+ARG DEFAULT_PG_VERSION
 WORKDIR /data
 
 RUN set -e \
@@ -77,6 +84,7 @@ COPY --from=build --chown=neon:neon /home/nonroot/target/release/storage_scrubbe
 COPY --from=pg-build /home/nonroot/pg_install/v14 /usr/local/v14/
 COPY --from=pg-build /home/nonroot/pg_install/v15 /usr/local/v15/
 COPY --from=pg-build /home/nonroot/pg_install/v16 /usr/local/v16/
+COPY --from=pg-build /home/nonroot/pg_install/v17 /usr/local/v17/
 COPY --from=pg-build /home/nonroot/postgres_install.tar.gz /data/
 
 # By default, pageserver uses `.neon/` working directory in WORKDIR, so create one and fill it with the dummy config.
@@ -93,7 +101,7 @@ RUN mkdir -p /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/v16/lib
+ENV LD_LIBRARY_PATH=/usr/local/v${DEFAULT_PG_VERSION}/lib
 
 
 VOLUME ["/data"]

Makefile

@@ -119,6 +119,8 @@ $(POSTGRES_INSTALL_DIR)/build/%/config.status:
 # I'm not sure why it wouldn't work, but this is the only place (apart from
 # the "build-all-versions" entry points) where direct mention of PostgreSQL
 # versions is used.
+.PHONY: postgres-configure-v17
+postgres-configure-v17: $(POSTGRES_INSTALL_DIR)/build/v17/config.status
 .PHONY: postgres-configure-v16
 postgres-configure-v16: $(POSTGRES_INSTALL_DIR)/build/v16/config.status
 .PHONY: postgres-configure-v15
@@ -215,29 +217,31 @@ neon-pg-clean-ext-%:
 # they depend on openssl and other libraries that are not included in our
 # Rust build.
 .PHONY: walproposer-lib
-walproposer-lib: neon-pg-ext-v16
+walproposer-lib: neon-pg-ext-v17
 	+@echo "Compiling walproposer-lib"
 	mkdir -p $(POSTGRES_INSTALL_DIR)/build/walproposer-lib
-	$(MAKE) PG_CONFIG=$(POSTGRES_INSTALL_DIR)/v16/bin/pg_config CFLAGS='$(PG_CFLAGS) $(COPT)' \
+	$(MAKE) PG_CONFIG=$(POSTGRES_INSTALL_DIR)/v17/bin/pg_config CFLAGS='$(PG_CFLAGS) $(COPT)' \
 		-C $(POSTGRES_INSTALL_DIR)/build/walproposer-lib \
 		-f $(ROOT_PROJECT_DIR)/pgxn/neon/Makefile walproposer-lib
-	cp $(POSTGRES_INSTALL_DIR)/v16/lib/libpgport.a $(POSTGRES_INSTALL_DIR)/build/walproposer-lib
-	cp $(POSTGRES_INSTALL_DIR)/v16/lib/libpgcommon.a $(POSTGRES_INSTALL_DIR)/build/walproposer-lib
-ifeq ($(UNAME_S),Linux)
+	cp $(POSTGRES_INSTALL_DIR)/v17/lib/libpgport.a $(POSTGRES_INSTALL_DIR)/build/walproposer-lib
+	cp $(POSTGRES_INSTALL_DIR)/v17/lib/libpgcommon.a $(POSTGRES_INSTALL_DIR)/build/walproposer-lib
 	$(AR) d $(POSTGRES_INSTALL_DIR)/build/walproposer-lib/libpgport.a \
 		pg_strong_random.o
 	$(AR) d $(POSTGRES_INSTALL_DIR)/build/walproposer-lib/libpgcommon.a \
-		pg_crc32c.o \
-		hmac_openssl.o \
+		checksum_helper.o \
 		cryptohash_openssl.o \
-		scram-common.o \
+		hmac_openssl.o \
 		md5_common.o \
-		checksum_helper.o
+		parse_manifest.o \
+		scram-common.o
+ifeq ($(UNAME_S),Linux)
+	$(AR) d $(POSTGRES_INSTALL_DIR)/build/walproposer-lib/libpgcommon.a \
+		pg_crc32c.o
 endif
 
 .PHONY: walproposer-lib-clean
 walproposer-lib-clean:
-	$(MAKE) PG_CONFIG=$(POSTGRES_INSTALL_DIR)/v16/bin/pg_config \
+	$(MAKE) PG_CONFIG=$(POSTGRES_INSTALL_DIR)/v17/bin/pg_config \
 		-C $(POSTGRES_INSTALL_DIR)/build/walproposer-lib \
 		-f $(ROOT_PROJECT_DIR)/pgxn/neon/Makefile clean
 
@@ -245,38 +249,44 @@ walproposer-lib-clean:
 neon-pg-ext: \
 	neon-pg-ext-v14 \
 	neon-pg-ext-v15 \
-	neon-pg-ext-v16
+	neon-pg-ext-v16 \
+	neon-pg-ext-v17
 
 .PHONY: neon-pg-clean-ext
 neon-pg-clean-ext: \
 	neon-pg-clean-ext-v14 \
 	neon-pg-clean-ext-v15 \
-	neon-pg-clean-ext-v16
+	neon-pg-clean-ext-v16 \
+	neon-pg-clean-ext-v17
 
 # shorthand to build all Postgres versions
 .PHONY: postgres
 postgres: \
 	postgres-v14 \
 	postgres-v15 \
-	postgres-v16
+	postgres-v16 \
+	postgres-v17
 
 .PHONY: postgres-headers
 postgres-headers: \
 	postgres-headers-v14 \
 	postgres-headers-v15 \
-	postgres-headers-v16
+	postgres-headers-v16 \
+	postgres-headers-v17
 
 .PHONY: postgres-clean
 postgres-clean: \
 	postgres-clean-v14 \
 	postgres-clean-v15 \
-	postgres-clean-v16
+	postgres-clean-v16 \
+	postgres-clean-v17
 
 .PHONY: postgres-check
 postgres-check: \
 	postgres-check-v14 \
 	postgres-check-v15 \
-	postgres-check-v16
+	postgres-check-v16 \
+	postgres-check-v17
 
 # This doesn't remove the effects of 'configure'.
 .PHONY: clean
@@ -321,13 +331,13 @@ postgres-%-pgindent: postgres-%-pg-bsd-indent postgres-%-typedefs.list
 	rm -f pg*.BAK
 
 # Indent pxgn/neon.
-.PHONY: pgindent
-neon-pgindent: postgres-v16-pg-bsd-indent neon-pg-ext-v16
-	$(MAKE) PG_CONFIG=$(POSTGRES_INSTALL_DIR)/v16/bin/pg_config CFLAGS='$(PG_CFLAGS) $(COPT)' \
-		FIND_TYPEDEF=$(ROOT_PROJECT_DIR)/vendor/postgres-v16/src/tools/find_typedef \
-		INDENT=$(POSTGRES_INSTALL_DIR)/build/v16/src/tools/pg_bsd_indent/pg_bsd_indent \
-		PGINDENT_SCRIPT=$(ROOT_PROJECT_DIR)/vendor/postgres-v16/src/tools/pgindent/pgindent \
-		-C $(POSTGRES_INSTALL_DIR)/build/neon-v16 \
+.PHONY: neon-pgindent
+neon-pgindent: postgres-v17-pg-bsd-indent neon-pg-ext-v17
+	$(MAKE) PG_CONFIG=$(POSTGRES_INSTALL_DIR)/v17/bin/pg_config CFLAGS='$(PG_CFLAGS) $(COPT)' \
+		FIND_TYPEDEF=$(ROOT_PROJECT_DIR)/vendor/postgres-v17/src/tools/find_typedef \
+		INDENT=$(POSTGRES_INSTALL_DIR)/build/v17/src/tools/pg_bsd_indent/pg_bsd_indent \
+		PGINDENT_SCRIPT=$(ROOT_PROJECT_DIR)/vendor/postgres-v17/src/tools/pgindent/pgindent \
+		-C $(POSTGRES_INSTALL_DIR)/build/neon-v17 \
 		-f $(ROOT_PROJECT_DIR)/pgxn/neon/Makefile pgindent
 
 

compute/Dockerfile.compute-node

@@ -55,22 +55,27 @@ RUN cd postgres && \
     # We could add the additional grant statements to the postgres repository but it would be hard to maintain,
     # whenever we need to pick up a new postgres version and we want to limit the changes in our postgres fork,
     # so we do it here.
-    old_list="pg_stat_statements--1.0--1.1.sql pg_stat_statements--1.1--1.2.sql pg_stat_statements--1.2--1.3.sql pg_stat_statements--1.3--1.4.sql pg_stat_statements--1.4--1.5.sql pg_stat_statements--1.4.sql pg_stat_statements--1.5--1.6.sql"; \
-    # the first loop is for pg_stat_statement extension version <= 1.6
     for file in /usr/local/pgsql/share/extension/pg_stat_statements--*.sql; do \
         filename=$(basename "$file"); \
-        if echo "$old_list" | grep -q -F "$filename"; then \
+        # Note that there are no downgrade scripts for pg_stat_statements, so we \
+        # don't have to modify any downgrade paths or (much) older versions: we only \
+        # have to make sure every creation of the pg_stat_statements_reset function \
+        # also adds execute permissions to the neon_superuser.
+        case $filename in \
+          pg_stat_statements--1.4.sql) \
+            # pg_stat_statements_reset is first created with 1.4
             echo 'GRANT EXECUTE ON FUNCTION pg_stat_statements_reset() TO neon_superuser;' >> $file; \
-        fi; \
-    done; \
-    # the second loop is for pg_stat_statement extension versions >= 1.7,
-    # where pg_stat_statement_reset() got 3 additional arguments
-    for file in /usr/local/pgsql/share/extension/pg_stat_statements--*.sql; do \
-        filename=$(basename "$file"); \
-        if ! echo "$old_list" | grep -q -F "$filename"; then \
+            ;; \
+          pg_stat_statements--1.6--1.7.sql) \
+            # Then with the 1.6-1.7 migration it is re-created with a new signature, thus add the permissions back
             echo 'GRANT EXECUTE ON FUNCTION pg_stat_statements_reset(Oid, Oid, bigint) TO neon_superuser;' >> $file; \
-        fi; \
-    done
+            ;; \
+          pg_stat_statements--1.10--1.11.sql) \
+            # Then with the 1.10-1.11 migration it is re-created with a new signature again, thus add the permissions back
+            echo 'GRANT EXECUTE ON FUNCTION pg_stat_statements_reset(Oid, Oid, bigint, boolean) TO neon_superuser;' >> $file; \
+            ;; \
+        esac; \
+    done;
 
 #########################################################################################
 #
@@ -79,6 +84,7 @@ RUN cd postgres && \
 #
 #########################################################################################
 FROM build-deps AS postgis-build
+ARG PG_VERSION
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 RUN apt update && \
     apt install -y cmake gdal-bin libboost-dev libboost-thread-dev libboost-filesystem-dev \
@@ -87,7 +93,11 @@ RUN apt update && \
     protobuf-c-compiler xsltproc
 
 # SFCGAL > 1.3 requires CGAL > 5.2, Bullseye's libcgal-dev is 5.2
-RUN wget https://gitlab.com/Oslandia/SFCGAL/-/archive/v1.3.10/SFCGAL-v1.3.10.tar.gz -O SFCGAL.tar.gz && \
+RUN case "${PG_VERSION}" in "v17") \
+    mkdir -p /sfcgal && \
+    echo "Postgis doensn't yet support PG17 (needs 3.4.3, if not higher)" && exit 0;; \
+    esac && \
+    wget https://gitlab.com/Oslandia/SFCGAL/-/archive/v1.3.10/SFCGAL-v1.3.10.tar.gz -O SFCGAL.tar.gz && \
     echo "4e39b3b2adada6254a7bdba6d297bb28e1a9835a9f879b74f37e2dab70203232 SFCGAL.tar.gz" | sha256sum --check && \
     mkdir sfcgal-src && cd sfcgal-src && tar xzf ../SFCGAL.tar.gz --strip-components=1 -C . && \
     cmake -DCMAKE_BUILD_TYPE=Release . && make -j $(getconf _NPROCESSORS_ONLN) && \
@@ -96,7 +106,10 @@ RUN wget https://gitlab.com/Oslandia/SFCGAL/-/archive/v1.3.10/SFCGAL-v1.3.10.tar
 
 ENV PATH="/usr/local/pgsql/bin:$PATH"
 
-RUN wget https://download.osgeo.org/postgis/source/postgis-3.3.3.tar.gz -O postgis.tar.gz && \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "Postgis doensn't yet support PG17 (needs 3.4.3, if not higher)" && exit 0;; \
+    esac && \
+    wget https://download.osgeo.org/postgis/source/postgis-3.3.3.tar.gz -O postgis.tar.gz && \
     echo "74eb356e3f85f14233791013360881b6748f78081cc688ff9d6f0f673a762d13 postgis.tar.gz" | sha256sum --check && \
     mkdir postgis-src && cd postgis-src && tar xzf ../postgis.tar.gz --strip-components=1 -C . && \
     find /usr/local/pgsql -type f | sed 's|^/usr/local/pgsql/||' > /before.txt &&\
@@ -122,7 +135,10 @@ RUN wget https://download.osgeo.org/postgis/source/postgis-3.3.3.tar.gz -O postg
     cp /usr/local/pgsql/share/extension/address_standardizer.control /extensions/postgis && \
     cp /usr/local/pgsql/share/extension/address_standardizer_data_us.control /extensions/postgis
 
-RUN wget https://github.com/pgRouting/pgrouting/archive/v3.4.2.tar.gz -O pgrouting.tar.gz && \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "v17 extensions are not supported yet. Quit" && exit 0;; \
+    esac && \
+    wget https://github.com/pgRouting/pgrouting/archive/v3.4.2.tar.gz -O pgrouting.tar.gz && \
     echo "cac297c07d34460887c4f3b522b35c470138760fe358e351ad1db4edb6ee306e pgrouting.tar.gz" | sha256sum --check && \
     mkdir pgrouting-src && cd pgrouting-src && tar xzf ../pgrouting.tar.gz --strip-components=1 -C . && \
     mkdir build && cd build && \
@@ -142,12 +158,19 @@ RUN wget https://github.com/pgRouting/pgrouting/archive/v3.4.2.tar.gz -O pgrouti
 #
 #########################################################################################
 FROM build-deps AS plv8-build
+ARG PG_VERSION
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
-RUN apt update && \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "v17 extensions are not supported yet. Quit" && exit 0;; \
+    esac && \
+    apt update && \
     apt install -y ninja-build python3-dev libncurses5 binutils clang
 
-RUN wget https://github.com/plv8/plv8/archive/refs/tags/v3.1.10.tar.gz -O plv8.tar.gz && \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "v17 extensions are not supported yet. Quit" && exit 0;; \
+    esac && \
+    wget https://github.com/plv8/plv8/archive/refs/tags/v3.1.10.tar.gz -O plv8.tar.gz && \
     echo "7096c3290928561f0d4901b7a52794295dc47f6303102fae3f8e42dd575ad97d plv8.tar.gz" | sha256sum --check && \
     mkdir plv8-src && cd plv8-src && tar xzf ../plv8.tar.gz --strip-components=1 -C . && \
     # generate and copy upgrade scripts
@@ -172,9 +195,13 @@ RUN wget https://github.com/plv8/plv8/archive/refs/tags/v3.1.10.tar.gz -O plv8.t
 #
 #########################################################################################
 FROM build-deps AS h3-pg-build
+ARG PG_VERSION
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
-RUN case "$(uname -m)" in \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "v17 extensions are not supported yet. Quit" && exit 0;; \
+    esac && \
+    case "$(uname -m)" in \
       "x86_64") \
         export CMAKE_CHECKSUM=739d372726cb23129d57a539ce1432453448816e345e1545f6127296926b6754 \
         ;; \
@@ -192,7 +219,11 @@ RUN case "$(uname -m)" in \
       && /tmp/cmake-install.sh --skip-license --prefix=/usr/local/ \
       && rm /tmp/cmake-install.sh
 
-RUN wget https://github.com/uber/h3/archive/refs/tags/v4.1.0.tar.gz -O h3.tar.gz && \
+RUN case "${PG_VERSION}" in "v17") \
+        mkdir -p /h3/usr/ && \
+    echo "v17 extensions are not supported yet. Quit" && exit 0;; \
+    esac && \
+    wget https://github.com/uber/h3/archive/refs/tags/v4.1.0.tar.gz -O h3.tar.gz && \
     echo "ec99f1f5974846bde64f4513cf8d2ea1b8d172d2218ab41803bf6a63532272bc h3.tar.gz" | sha256sum --check && \
     mkdir h3-src && cd h3-src && tar xzf ../h3.tar.gz --strip-components=1 -C . && \
     mkdir build && cd build && \
@@ -202,7 +233,10 @@ RUN wget https://github.com/uber/h3/archive/refs/tags/v4.1.0.tar.gz -O h3.tar.gz
     cp -R /h3/usr / && \
     rm -rf build
 
-RUN wget https://github.com/zachasme/h3-pg/archive/refs/tags/v4.1.3.tar.gz -O h3-pg.tar.gz && \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "v17 extensions are not supported yet. Quit" && exit 0;; \
+    esac && \
+    wget https://github.com/zachasme/h3-pg/archive/refs/tags/v4.1.3.tar.gz -O h3-pg.tar.gz && \
     echo "5c17f09a820859ffe949f847bebf1be98511fb8f1bd86f94932512c00479e324 h3-pg.tar.gz" | sha256sum --check && \
     mkdir h3-pg-src && cd h3-pg-src && tar xzf ../h3-pg.tar.gz --strip-components=1 -C . && \
     export PATH="/usr/local/pgsql/bin:$PATH" && \
@@ -218,9 +252,13 @@ RUN wget https://github.com/zachasme/h3-pg/archive/refs/tags/v4.1.3.tar.gz -O h3
 #
 #########################################################################################
 FROM build-deps AS unit-pg-build
+ARG PG_VERSION
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
-RUN wget https://github.com/df7cb/postgresql-unit/archive/refs/tags/7.7.tar.gz -O postgresql-unit.tar.gz && \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "v17 extensions are not supported yet. Quit" && exit 0;; \
+    esac && \
+    wget https://github.com/df7cb/postgresql-unit/archive/refs/tags/7.7.tar.gz -O postgresql-unit.tar.gz && \
     echo "411d05beeb97e5a4abf17572bfcfbb5a68d98d1018918feff995f6ee3bb03e79 postgresql-unit.tar.gz" | sha256sum --check && \
     mkdir postgresql-unit-src && cd postgresql-unit-src && tar xzf ../postgresql-unit.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
@@ -239,14 +277,18 @@ RUN wget https://github.com/df7cb/postgresql-unit/archive/refs/tags/7.7.tar.gz -
 #
 #########################################################################################
 FROM build-deps AS vector-pg-build
+ARG PG_VERSION
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
-COPY patches/pgvector.patch /pgvector.patch
+COPY compute/patches/pgvector.patch /pgvector.patch
 
 # By default, pgvector Makefile uses `-march=native`. We don't want that,
 # because we build the images on different machines than where we run them.
 # Pass OPTFLAGS="" to remove it.
-RUN wget https://github.com/pgvector/pgvector/archive/refs/tags/v0.7.2.tar.gz -O pgvector.tar.gz && \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "v17 extensions are not supported yet. Quit" && exit 0;; \
+    esac && \
+    wget https://github.com/pgvector/pgvector/archive/refs/tags/v0.7.2.tar.gz -O pgvector.tar.gz && \
     echo "617fba855c9bcb41a2a9bc78a78567fd2e147c72afd5bf9d37b31b9591632b30 pgvector.tar.gz" | sha256sum --check && \
     mkdir pgvector-src && cd pgvector-src && tar xzf ../pgvector.tar.gz --strip-components=1 -C . && \
     patch -p1 < /pgvector.patch && \
@@ -261,10 +303,14 @@ RUN wget https://github.com/pgvector/pgvector/archive/refs/tags/v0.7.2.tar.gz -O
 #
 #########################################################################################
 FROM build-deps AS pgjwt-pg-build
+ARG PG_VERSION
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
 # 9742dab1b2f297ad3811120db7b21451bca2d3c9 made on 13/11/2021
-RUN wget https://github.com/michelp/pgjwt/archive/9742dab1b2f297ad3811120db7b21451bca2d3c9.tar.gz -O pgjwt.tar.gz && \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "v17 extensions are not supported yet. Quit" && exit 0;; \
+    esac && \
+    wget https://github.com/michelp/pgjwt/archive/9742dab1b2f297ad3811120db7b21451bca2d3c9.tar.gz -O pgjwt.tar.gz && \
     echo "cfdefb15007286f67d3d45510f04a6a7a495004be5b3aecb12cda667e774203f pgjwt.tar.gz" | sha256sum --check && \
     mkdir pgjwt-src && cd pgjwt-src && tar xzf ../pgjwt.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) install PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
@@ -277,9 +323,13 @@ RUN wget https://github.com/michelp/pgjwt/archive/9742dab1b2f297ad3811120db7b214
 #
 #########################################################################################
 FROM build-deps AS hypopg-pg-build
+ARG PG_VERSION
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
-RUN wget https://github.com/HypoPG/hypopg/archive/refs/tags/1.4.0.tar.gz -O hypopg.tar.gz && \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "v17 extensions are not supported yet. Quit" && exit 0;; \
+    esac && \
+    wget https://github.com/HypoPG/hypopg/archive/refs/tags/1.4.0.tar.gz -O hypopg.tar.gz && \
     echo "0821011743083226fc9b813c1f2ef5897a91901b57b6bea85a78e466187c6819 hypopg.tar.gz" | sha256sum --check && \
     mkdir hypopg-src && cd hypopg-src && tar xzf ../hypopg.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
@@ -293,9 +343,13 @@ RUN wget https://github.com/HypoPG/hypopg/archive/refs/tags/1.4.0.tar.gz -O hypo
 #
 #########################################################################################
 FROM build-deps AS pg-hashids-pg-build
+ARG PG_VERSION
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
-RUN wget https://github.com/iCyberon/pg_hashids/archive/refs/tags/v1.2.1.tar.gz -O pg_hashids.tar.gz && \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "v17 extensions are not supported yet. Quit" && exit 0;; \
+    esac && \
+    wget https://github.com/iCyberon/pg_hashids/archive/refs/tags/v1.2.1.tar.gz -O pg_hashids.tar.gz && \
     echo "74576b992d9277c92196dd8d816baa2cc2d8046fe102f3dcd7f3c3febed6822a pg_hashids.tar.gz" | sha256sum --check && \
     mkdir pg_hashids-src && cd pg_hashids-src && tar xzf ../pg_hashids.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) PG_CONFIG=/usr/local/pgsql/bin/pg_config USE_PGXS=1 && \
@@ -309,11 +363,15 @@ RUN wget https://github.com/iCyberon/pg_hashids/archive/refs/tags/v1.2.1.tar.gz
 #
 #########################################################################################
 FROM build-deps AS rum-pg-build
+ARG PG_VERSION
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
-COPY patches/rum.patch /rum.patch
+COPY compute/patches/rum.patch /rum.patch
 
-RUN wget https://github.com/postgrespro/rum/archive/refs/tags/1.3.13.tar.gz -O rum.tar.gz && \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "v17 extensions are not supported yet. Quit" && exit 0;; \
+    esac && \
+    wget https://github.com/postgrespro/rum/archive/refs/tags/1.3.13.tar.gz -O rum.tar.gz && \
     echo "6ab370532c965568df6210bd844ac6ba649f53055e48243525b0b7e5c4d69a7d rum.tar.gz" | sha256sum --check && \
     mkdir rum-src && cd rum-src && tar xzf ../rum.tar.gz --strip-components=1 -C . && \
     patch -p1 < /rum.patch && \
@@ -328,9 +386,13 @@ RUN wget https://github.com/postgrespro/rum/archive/refs/tags/1.3.13.tar.gz -O r
 #
 #########################################################################################
 FROM build-deps AS pgtap-pg-build
+ARG PG_VERSION
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
-RUN wget https://github.com/theory/pgtap/archive/refs/tags/v1.2.0.tar.gz -O pgtap.tar.gz && \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "v17 extensions are not supported yet. Quit" && exit 0;; \
+    esac && \
+    wget https://github.com/theory/pgtap/archive/refs/tags/v1.2.0.tar.gz -O pgtap.tar.gz && \
     echo "9c7c3de67ea41638e14f06da5da57bac6f5bd03fea05c165a0ec862205a5c052 pgtap.tar.gz" | sha256sum --check && \
     mkdir pgtap-src && cd pgtap-src && tar xzf ../pgtap.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
@@ -344,9 +406,13 @@ RUN wget https://github.com/theory/pgtap/archive/refs/tags/v1.2.0.tar.gz -O pgta
 #
 #########################################################################################
 FROM build-deps AS ip4r-pg-build
+ARG PG_VERSION
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
-RUN wget https://github.com/RhodiumToad/ip4r/archive/refs/tags/2.4.2.tar.gz -O ip4r.tar.gz && \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "v17 extensions are not supported yet. Quit" && exit 0;; \
+    esac && \
+    wget https://github.com/RhodiumToad/ip4r/archive/refs/tags/2.4.2.tar.gz -O ip4r.tar.gz && \
     echo "0f7b1f159974f49a47842a8ab6751aecca1ed1142b6d5e38d81b064b2ead1b4b ip4r.tar.gz" | sha256sum --check && \
     mkdir ip4r-src && cd ip4r-src && tar xzf ../ip4r.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
@@ -360,9 +426,13 @@ RUN wget https://github.com/RhodiumToad/ip4r/archive/refs/tags/2.4.2.tar.gz -O i
 #
 #########################################################################################
 FROM build-deps AS prefix-pg-build
+ARG PG_VERSION
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
-RUN wget https://github.com/dimitri/prefix/archive/refs/tags/v1.2.10.tar.gz -O prefix.tar.gz && \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "v17 extensions are not supported yet. Quit" && exit 0;; \
+    esac && \
+    wget https://github.com/dimitri/prefix/archive/refs/tags/v1.2.10.tar.gz -O prefix.tar.gz && \
     echo "4342f251432a5f6fb05b8597139d3ccde8dcf87e8ca1498e7ee931ca057a8575 prefix.tar.gz" | sha256sum --check && \
     mkdir prefix-src && cd prefix-src && tar xzf ../prefix.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
@@ -376,9 +446,13 @@ RUN wget https://github.com/dimitri/prefix/archive/refs/tags/v1.2.10.tar.gz -O p
 #
 #########################################################################################
 FROM build-deps AS hll-pg-build
+ARG PG_VERSION
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
-RUN wget https://github.com/citusdata/postgresql-hll/archive/refs/tags/v2.18.tar.gz -O hll.tar.gz && \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "v17 extensions are not supported yet. Quit" && exit 0;; \
+    esac && \
+    wget https://github.com/citusdata/postgresql-hll/archive/refs/tags/v2.18.tar.gz -O hll.tar.gz && \
     echo "e2f55a6f4c4ab95ee4f1b4a2b73280258c5136b161fe9d059559556079694f0e hll.tar.gz" | sha256sum --check && \
     mkdir hll-src && cd hll-src && tar xzf ../hll.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
@@ -392,9 +466,13 @@ RUN wget https://github.com/citusdata/postgresql-hll/archive/refs/tags/v2.18.tar
 #
 #########################################################################################
 FROM build-deps AS plpgsql-check-pg-build
+ARG PG_VERSION
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
-RUN wget https://github.com/okbob/plpgsql_check/archive/refs/tags/v2.5.3.tar.gz -O plpgsql_check.tar.gz && \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "v17 extensions are not supported yet. Quit" && exit 0;; \
+    esac && \
+    wget https://github.com/okbob/plpgsql_check/archive/refs/tags/v2.5.3.tar.gz -O plpgsql_check.tar.gz && \
     echo "6631ec3e7fb3769eaaf56e3dfedb829aa761abf163d13dba354b4c218508e1c0 plpgsql_check.tar.gz" | sha256sum --check && \
     mkdir plpgsql_check-src && cd plpgsql_check-src && tar xzf ../plpgsql_check.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) PG_CONFIG=/usr/local/pgsql/bin/pg_config USE_PGXS=1 && \
@@ -413,7 +491,10 @@ COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 ARG PG_VERSION
 ENV PATH="/usr/local/pgsql/bin:$PATH"
 
-RUN case "${PG_VERSION}" in \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "v17 extensions are not supported yet. Quit" && exit 0;; \
+    esac && \
+    case "${PG_VERSION}" in \
       "v14" | "v15") \
         export TIMESCALEDB_VERSION=2.10.1 \
         export TIMESCALEDB_CHECKSUM=6fca72a6ed0f6d32d2b3523951ede73dc5f9b0077b38450a029a5f411fdb8c73 \
@@ -446,7 +527,10 @@ COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 ARG PG_VERSION
 ENV PATH="/usr/local/pgsql/bin:$PATH"
 
-RUN case "${PG_VERSION}" in \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "v17 extensions are not supported yet. Quit" && exit 0;; \
+    esac && \
+    case "${PG_VERSION}" in \
       "v14") \
         export PG_HINT_PLAN_VERSION=14_1_4_1 \
         export PG_HINT_PLAN_CHECKSUM=c3501becf70ead27f70626bce80ea401ceac6a77e2083ee5f3ff1f1444ec1ad1 \
@@ -459,6 +543,9 @@ RUN case "${PG_VERSION}" in \
         export PG_HINT_PLAN_VERSION=16_1_6_0 \
         export PG_HINT_PLAN_CHECKSUM=fc85a9212e7d2819d4ae4ac75817481101833c3cfa9f0fe1f980984e12347d00 \
         ;; \
+      "v17") \
+        echo "TODO: PG17 pg_hint_plan support" && exit 0 \
+        ;; \
       *) \
         echo "Export the valid PG_HINT_PLAN_VERSION variable" && exit 1 \
         ;; \
@@ -478,10 +565,14 @@ RUN case "${PG_VERSION}" in \
 #
 #########################################################################################
 FROM build-deps AS pg-cron-pg-build
+ARG PG_VERSION
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
 ENV PATH="/usr/local/pgsql/bin/:$PATH"
-RUN wget https://github.com/citusdata/pg_cron/archive/refs/tags/v1.6.0.tar.gz -O pg_cron.tar.gz && \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "v17 extensions are not supported yet. Quit" && exit 0;; \
+    esac && \
+    wget https://github.com/citusdata/pg_cron/archive/refs/tags/v1.6.0.tar.gz -O pg_cron.tar.gz && \
     echo "383a627867d730222c272bfd25cd5e151c578d73f696d32910c7db8c665cc7db pg_cron.tar.gz" | sha256sum --check && \
     mkdir pg_cron-src && cd pg_cron-src && tar xzf ../pg_cron.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) && \
@@ -495,9 +586,13 @@ RUN wget https://github.com/citusdata/pg_cron/archive/refs/tags/v1.6.0.tar.gz -O
 #
 #########################################################################################
 FROM build-deps AS rdkit-pg-build
+ARG PG_VERSION
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
-RUN apt-get update && \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "v17 extensions are not supported yet. Quit" && exit 0;; \
+    esac && \
+    apt-get update && \
     apt-get install -y \
         cmake \
         libboost-iostreams1.74-dev \
@@ -507,7 +602,10 @@ RUN apt-get update && \
         libeigen3-dev
 
 ENV PATH="/usr/local/pgsql/bin/:/usr/local/pgsql/:$PATH"
-RUN wget https://github.com/rdkit/rdkit/archive/refs/tags/Release_2023_03_3.tar.gz -O rdkit.tar.gz && \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "v17 extensions are not supported yet. Quit" && exit 0;; \
+    esac && \
+    wget https://github.com/rdkit/rdkit/archive/refs/tags/Release_2023_03_3.tar.gz -O rdkit.tar.gz && \
     echo "bdbf9a2e6988526bfeb8c56ce3cdfe2998d60ac289078e2215374288185e8c8d rdkit.tar.gz" | sha256sum --check && \
     mkdir rdkit-src && cd rdkit-src && tar xzf ../rdkit.tar.gz --strip-components=1 -C . && \
     cmake \
@@ -544,10 +642,14 @@ RUN wget https://github.com/rdkit/rdkit/archive/refs/tags/Release_2023_03_3.tar.
 #
 #########################################################################################
 FROM build-deps AS pg-uuidv7-pg-build
+ARG PG_VERSION
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
 ENV PATH="/usr/local/pgsql/bin/:$PATH"
-RUN wget https://github.com/fboulnois/pg_uuidv7/archive/refs/tags/v1.0.1.tar.gz -O pg_uuidv7.tar.gz && \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "v17 extensions are not supported yet. Quit" && exit 0;; \
+    esac && \
+    wget https://github.com/fboulnois/pg_uuidv7/archive/refs/tags/v1.0.1.tar.gz -O pg_uuidv7.tar.gz && \
     echo "0d0759ab01b7fb23851ecffb0bce27822e1868a4a5819bfd276101c716637a7a pg_uuidv7.tar.gz" | sha256sum --check && \
     mkdir pg_uuidv7-src && cd pg_uuidv7-src && tar xzf ../pg_uuidv7.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) && \
@@ -561,10 +663,14 @@ RUN wget https://github.com/fboulnois/pg_uuidv7/archive/refs/tags/v1.0.1.tar.gz
 #
 #########################################################################################
 FROM build-deps AS pg-roaringbitmap-pg-build
+ARG PG_VERSION
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
 ENV PATH="/usr/local/pgsql/bin/:$PATH"
-RUN wget https://github.com/ChenHuajun/pg_roaringbitmap/archive/refs/tags/v0.5.4.tar.gz -O pg_roaringbitmap.tar.gz && \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "v17 extensions is not supported yet by pg_roaringbitmap. Quit" && exit 0;; \
+    esac && \
+    wget https://github.com/ChenHuajun/pg_roaringbitmap/archive/refs/tags/v0.5.4.tar.gz -O pg_roaringbitmap.tar.gz && \
     echo "b75201efcb1c2d1b014ec4ae6a22769cc7a224e6e406a587f5784a37b6b5a2aa pg_roaringbitmap.tar.gz" | sha256sum --check && \
     mkdir pg_roaringbitmap-src && cd pg_roaringbitmap-src && tar xzf ../pg_roaringbitmap.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) && \
@@ -578,10 +684,14 @@ RUN wget https://github.com/ChenHuajun/pg_roaringbitmap/archive/refs/tags/v0.5.4
 #
 #########################################################################################
 FROM build-deps AS pg-semver-pg-build
+ARG PG_VERSION
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
 ENV PATH="/usr/local/pgsql/bin/:$PATH"
-RUN wget https://github.com/theory/pg-semver/archive/refs/tags/v0.32.1.tar.gz -O pg_semver.tar.gz && \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "v17 is not supported yet by pg_semver. Quit" && exit 0;; \
+    esac && \
+    wget https://github.com/theory/pg-semver/archive/refs/tags/v0.32.1.tar.gz -O pg_semver.tar.gz && \
     echo "fbdaf7512026d62eec03fad8687c15ed509b6ba395bff140acd63d2e4fbe25d7 pg_semver.tar.gz" | sha256sum --check && \
     mkdir pg_semver-src && cd pg_semver-src && tar xzf ../pg_semver.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) && \
@@ -620,10 +730,14 @@ RUN case "${PG_VERSION}" in \
 #
 #########################################################################################
 FROM build-deps AS pg-anon-pg-build
+ARG PG_VERSION
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
 ENV PATH="/usr/local/pgsql/bin/:$PATH"
-RUN wget  https://github.com/neondatabase/postgresql_anonymizer/archive/refs/tags/neon_1.1.1.tar.gz -O pg_anon.tar.gz && \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "postgresql_anonymizer does not yet support PG17" && exit 0;; \
+    esac && \
+    wget  https://github.com/neondatabase/postgresql_anonymizer/archive/refs/tags/neon_1.1.1.tar.gz -O pg_anon.tar.gz && \
     echo "321ea8d5c1648880aafde850a2c576e4a9e7b9933a34ce272efc839328999fa9  pg_anon.tar.gz" | sha256sum --check && \
     mkdir pg_anon-src && cd pg_anon-src && tar xzf ../pg_anon.tar.gz --strip-components=1 -C . && \
     find /usr/local/pgsql -type f | sed 's|^/usr/local/pgsql/||' > /before.txt &&\
@@ -641,6 +755,7 @@ RUN wget  https://github.com/neondatabase/postgresql_anonymizer/archive/refs/tag
 #
 #########################################################################################
 FROM build-deps AS rust-extensions-build
+ARG PG_VERSION
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
 RUN apt-get update && \
@@ -651,9 +766,11 @@ ENV HOME=/home/nonroot
 ENV PATH="/home/nonroot/.cargo/bin:/usr/local/pgsql/bin/:$PATH"
 USER nonroot
 WORKDIR /home/nonroot
-ARG PG_VERSION
 
-RUN curl -sSO https://static.rust-lang.org/rustup/dist/$(uname -m)-unknown-linux-gnu/rustup-init && \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "v17 is not supported yet by pgrx. Quit" && exit 0;; \
+    esac && \
+    curl -sSO https://static.rust-lang.org/rustup/dist/$(uname -m)-unknown-linux-gnu/rustup-init && \
     chmod +x rustup-init && \
     ./rustup-init -y --no-modify-path --profile minimal --default-toolchain stable && \
     rm rustup-init && \
@@ -672,7 +789,10 @@ USER root
 FROM rust-extensions-build AS pg-jsonschema-pg-build
 ARG PG_VERSION
 
-RUN wget https://github.com/supabase/pg_jsonschema/archive/refs/tags/v0.3.1.tar.gz -O pg_jsonschema.tar.gz && \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "pg_jsonschema does not yet have a release that supports pg17" && exit 0;; \
+    esac && \
+    wget https://github.com/supabase/pg_jsonschema/archive/refs/tags/v0.3.1.tar.gz -O pg_jsonschema.tar.gz && \
     echo "61df3db1ed83cf24f6aa39c826f8818bfa4f0bd33b587fd6b2b1747985642297 pg_jsonschema.tar.gz" | sha256sum --check && \
     mkdir pg_jsonschema-src && cd pg_jsonschema-src && tar xzf ../pg_jsonschema.tar.gz --strip-components=1 -C . && \
     # see commit 252b3685a27a0f4c31a0f91e983c6314838e89e8
@@ -694,7 +814,10 @@ RUN wget https://github.com/supabase/pg_jsonschema/archive/refs/tags/v0.3.1.tar.
 FROM rust-extensions-build AS pg-graphql-pg-build
 ARG PG_VERSION
 
-RUN wget https://github.com/supabase/pg_graphql/archive/refs/tags/v1.5.7.tar.gz -O pg_graphql.tar.gz && \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "pg_graphql does not yet have a release that supports pg17 as of now" && exit 0;; \
+    esac && \
+    wget https://github.com/supabase/pg_graphql/archive/refs/tags/v1.5.7.tar.gz -O pg_graphql.tar.gz && \
     echo "2b3e567a5b31019cb97ae0e33263c1bcc28580be5a444ac4c8ece5c4be2aea41 pg_graphql.tar.gz" | sha256sum --check && \
     mkdir pg_graphql-src && cd pg_graphql-src && tar xzf ../pg_graphql.tar.gz --strip-components=1 -C . && \
     sed -i 's/pgrx = "=0.11.3"/pgrx = { version = "0.11.3", features = [ "unsafe-postgres" ] }/g' Cargo.toml && \
@@ -714,7 +837,10 @@ FROM rust-extensions-build AS pg-tiktoken-pg-build
 ARG PG_VERSION
 
 # 26806147b17b60763039c6a6878884c41a262318 made on 26/09/2023
-RUN wget https://github.com/kelvich/pg_tiktoken/archive/26806147b17b60763039c6a6878884c41a262318.tar.gz -O pg_tiktoken.tar.gz && \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "pg_tiktoken does not have versions, nor support for pg17" && exit 0;; \
+    esac && \
+    wget https://github.com/kelvich/pg_tiktoken/archive/26806147b17b60763039c6a6878884c41a262318.tar.gz -O pg_tiktoken.tar.gz && \
     echo "e64e55aaa38c259512d3e27c572da22c4637418cf124caba904cd50944e5004e pg_tiktoken.tar.gz" | sha256sum --check && \
     mkdir pg_tiktoken-src && cd pg_tiktoken-src && tar xzf ../pg_tiktoken.tar.gz --strip-components=1 -C . && \
     # TODO update pgrx version in the pg_tiktoken repo and remove this line
@@ -733,7 +859,10 @@ RUN wget https://github.com/kelvich/pg_tiktoken/archive/26806147b17b60763039c6a6
 FROM rust-extensions-build AS pg-pgx-ulid-build
 ARG PG_VERSION
 
-RUN wget https://github.com/pksunkara/pgx_ulid/archive/refs/tags/v0.1.5.tar.gz -O pgx_ulid.tar.gz && \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "pgx_ulid does not support pg17 as of the latest version (0.1.5)" && exit 0;; \
+    esac && \
+    wget https://github.com/pksunkara/pgx_ulid/archive/refs/tags/v0.1.5.tar.gz -O pgx_ulid.tar.gz && \
     echo "9d1659a2da65af0133d5451c454de31b37364e3502087dadf579f790bc8bef17 pgx_ulid.tar.gz" | sha256sum --check && \
     mkdir pgx_ulid-src && cd pgx_ulid-src && tar xzf ../pgx_ulid.tar.gz --strip-components=1 -C . && \
     sed -i 's/pgrx       = "^0.11.2"/pgrx = { version = "=0.11.3", features = [ "unsafe-postgres" ] }/g' Cargo.toml && \
@@ -748,10 +877,14 @@ RUN wget https://github.com/pksunkara/pgx_ulid/archive/refs/tags/v0.1.5.tar.gz -
 #########################################################################################
 
 FROM build-deps AS wal2json-pg-build
+ARG PG_VERSION
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
 ENV PATH="/usr/local/pgsql/bin/:$PATH"
-RUN wget https://github.com/eulerto/wal2json/archive/refs/tags/wal2json_2_5.tar.gz && \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "We'll need to update wal2json to 2.6+ for pg17 support" && exit 0;; \
+    esac && \
+    wget https://github.com/eulerto/wal2json/archive/refs/tags/wal2json_2_5.tar.gz && \
     echo "b516653575541cf221b99cf3f8be9b6821f6dbcfc125675c85f35090f824f00e wal2json_2_5.tar.gz" | sha256sum --check && \
     mkdir wal2json-src && cd wal2json-src && tar xzf ../wal2json_2_5.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) && \
@@ -764,10 +897,14 @@ RUN wget https://github.com/eulerto/wal2json/archive/refs/tags/wal2json_2_5.tar.
 #
 #########################################################################################
 FROM build-deps AS pg-ivm-build
+ARG PG_VERSION
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
 ENV PATH="/usr/local/pgsql/bin/:$PATH"
-RUN wget https://github.com/sraoss/pg_ivm/archive/refs/tags/v1.7.tar.gz -O pg_ivm.tar.gz && \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "We'll need to update pg_ivm to 1.9+ for pg17 support" && exit 0;; \
+    esac && \
+    wget https://github.com/sraoss/pg_ivm/archive/refs/tags/v1.7.tar.gz -O pg_ivm.tar.gz && \
     echo "ebfde04f99203c7be4b0e873f91104090e2e83e5429c32ac242d00f334224d5e pg_ivm.tar.gz" | sha256sum --check && \
     mkdir pg_ivm-src && cd pg_ivm-src && tar xzf ../pg_ivm.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) && \
@@ -781,10 +918,14 @@ RUN wget https://github.com/sraoss/pg_ivm/archive/refs/tags/v1.7.tar.gz -O pg_iv
 #
 #########################################################################################
 FROM build-deps AS pg-partman-build
+ARG PG_VERSION
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
 ENV PATH="/usr/local/pgsql/bin/:$PATH"
-RUN wget https://github.com/pgpartman/pg_partman/archive/refs/tags/v5.0.1.tar.gz -O pg_partman.tar.gz && \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "pg_partman doesn't support PG17 yet" && exit 0;; \
+    esac && \
+    wget https://github.com/pgpartman/pg_partman/archive/refs/tags/v5.0.1.tar.gz -O pg_partman.tar.gz && \
     echo "75b541733a9659a6c90dbd40fccb904a630a32880a6e3044d0c4c5f4c8a65525 pg_partman.tar.gz" | sha256sum --check && \
     mkdir pg_partman-src && cd pg_partman-src && tar xzf ../pg_partman.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) && \
@@ -854,8 +995,8 @@ RUN make -j $(getconf _NPROCESSORS_ONLN) \
     case "${PG_VERSION}" in \
         "v14" | "v15") \
         ;; \
-        "v16") \
-            echo "Skipping HNSW for PostgreSQL 16" && exit 0 \
+        "v16" | "v17") \
+            echo "Skipping HNSW for PostgreSQL ${PG_VERSION}" && exit 0 \
         ;; \
         *) \
             echo "unexpected PostgreSQL version" && exit 1 \
@@ -890,6 +1031,41 @@ FROM debian:bullseye-slim AS compute-tools-image
 
 COPY --from=compute-tools /home/nonroot/target/release-line-debug-size-lto/compute_ctl /usr/local/bin/compute_ctl
 
+#########################################################################################
+#
+# Layer "pgbouncer"
+#
+#########################################################################################
+
+FROM debian:bullseye-slim AS pgbouncer
+RUN set -e \
+    && apt-get update \
+    && apt-get install -y \
+        build-essential \
+        git \
+        libevent-dev \
+        libtool \
+        pkg-config
+
+# Use `dist_man_MANS=` to skip manpage generation (which requires python3/pandoc)
+ENV PGBOUNCER_TAG=pgbouncer_1_22_1
+RUN set -e \
+    && git clone --recurse-submodules --depth 1 --branch ${PGBOUNCER_TAG} https://github.com/pgbouncer/pgbouncer.git pgbouncer \
+    && cd pgbouncer \
+    && ./autogen.sh \
+    && LDFLAGS=-static ./configure --prefix=/usr/local/pgbouncer --without-openssl \
+    && make -j $(nproc) dist_man_MANS= \
+    && make install dist_man_MANS=
+
+#########################################################################################
+#
+# Layers "postgres-exporter" and "sql-exporter"
+#
+#########################################################################################
+
+FROM quay.io/prometheuscommunity/postgres-exporter:v0.12.1 AS postgres-exporter
+FROM burningalchemist/sql_exporter:0.13 AS sql-exporter
+
 #########################################################################################
 #
 # Clean up postgres folder before inclusion
@@ -899,7 +1075,7 @@ FROM neon-pg-ext-build AS postgres-cleanup-layer
 COPY --from=neon-pg-ext-build /usr/local/pgsql /usr/local/pgsql
 
 # Remove binaries from /bin/ that we won't use (or would manually copy & install otherwise)
-RUN cd /usr/local/pgsql/bin && rm ecpg raster2pgsql shp2pgsql pgtopo_export pgtopo_import pgsql2shp
+RUN cd /usr/local/pgsql/bin && rm -f ecpg raster2pgsql shp2pgsql pgtopo_export pgtopo_import pgsql2shp
 
 # Remove headers that we won't need anymore - we've completed installation of all extensions
 RUN rm -r /usr/local/pgsql/include
@@ -918,7 +1094,10 @@ RUN rm /usr/local/pgsql/lib/lib*.a
 
 FROM neon-pg-ext-build AS neon-pg-ext-test
 ARG PG_VERSION
-RUN mkdir /ext-src
+RUN case "${PG_VERSION}" in "v17") \
+    echo "v17 extensions are not supported yet. Quit" && exit 0;; \
+    esac && \
+    mkdir /ext-src
 
 #COPY --from=postgis-build /postgis.tar.gz /ext-src/
 #COPY --from=postgis-build /sfcgal/* /usr
@@ -934,7 +1113,7 @@ COPY --from=pgjwt-pg-build /pgjwt.tar.gz /ext-src
 COPY --from=hypopg-pg-build /hypopg.tar.gz /ext-src
 COPY --from=pg-hashids-pg-build /pg_hashids.tar.gz /ext-src
 COPY --from=rum-pg-build /rum.tar.gz /ext-src
-COPY patches/rum.patch /ext-src
+COPY compute/patches/rum.patch /ext-src
 #COPY --from=pgtap-pg-build /pgtap.tar.gz /ext-src
 COPY --from=ip4r-pg-build /ip4r.tar.gz /ext-src
 COPY --from=prefix-pg-build /prefix.tar.gz /ext-src
@@ -942,9 +1121,9 @@ COPY --from=hll-pg-build /hll.tar.gz /ext-src
 COPY --from=plpgsql-check-pg-build /plpgsql_check.tar.gz /ext-src
 #COPY --from=timescaledb-pg-build /timescaledb.tar.gz /ext-src
 COPY --from=pg-hint-plan-pg-build /pg_hint_plan.tar.gz /ext-src
-COPY patches/pg_hint_plan.patch /ext-src
+COPY compute/patches/pg_hint_plan.patch /ext-src
 COPY --from=pg-cron-pg-build /pg_cron.tar.gz /ext-src
-COPY patches/pg_cron.patch /ext-src
+COPY compute/patches/pg_cron.patch /ext-src
 #COPY --from=pg-pgx-ulid-build /home/nonroot/pgx_ulid.tar.gz /ext-src
 #COPY --from=rdkit-pg-build /rdkit.tar.gz /ext-src
 COPY --from=pg-uuidv7-pg-build /pg_uuidv7.tar.gz /ext-src
@@ -953,21 +1132,42 @@ COPY --from=pg-semver-pg-build /pg_semver.tar.gz /ext-src
 #COPY --from=pg-embedding-pg-build /home/nonroot/pg_embedding-src/ /ext-src
 #COPY --from=wal2json-pg-build /wal2json_2_5.tar.gz /ext-src
 COPY --from=pg-anon-pg-build /pg_anon.tar.gz /ext-src
-COPY patches/pg_anon.patch /ext-src
+COPY compute/patches/pg_anon.patch /ext-src
 COPY --from=pg-ivm-build /pg_ivm.tar.gz /ext-src
 COPY --from=pg-partman-build /pg_partman.tar.gz /ext-src
-RUN cd /ext-src/ && for f in *.tar.gz; \
+RUN case "${PG_VERSION}" in "v17") \
+    echo "v17 extensions are not supported yet. Quit" && exit 0;; \
+    esac && \
+    cd /ext-src/ && for f in *.tar.gz; \
     do echo $f; dname=$(echo $f | sed 's/\.tar.*//')-src; \
     rm -rf $dname; mkdir $dname; tar xzf $f --strip-components=1 -C $dname \
     || exit 1; rm -f $f; done
-RUN cd /ext-src/pgvector-src && patch -p1 <../pgvector.patch
-RUN cd /ext-src/rum-src && patch -p1 <../rum.patch
+RUN case "${PG_VERSION}" in "v17") \
+    echo "v17 extensions are not supported yet. Quit" && exit 0;; \
+    esac && \
+    cd /ext-src/rum-src && patch -p1 <../rum.patch
+RUN case "${PG_VERSION}" in "v17") \
+    echo "v17 extensions are not supported yet. Quit" && exit 0;; \
+    esac && \
+    cd /ext-src/pgvector-src && patch -p1 <../pgvector.patch
 # cmake is required for the h3 test
-RUN apt-get update && apt-get install -y cmake
-RUN cd /ext-src/pg_hint_plan-src && patch -p1 < /ext-src/pg_hint_plan.patch
+RUN case "${PG_VERSION}" in "v17") \
+    echo "v17 extensions are not supported yet. Quit" && exit 0;; \
+    esac && \
+    apt-get update && apt-get install -y cmake
+RUN case "${PG_VERSION}" in "v17") \
+    echo "v17 extensions are not supported yet. Quit" && exit 0;; \
+    esac && \
+    cd /ext-src/pg_hint_plan-src && patch -p1 < /ext-src/pg_hint_plan.patch
 COPY --chmod=755 docker-compose/run-tests.sh /run-tests.sh
-RUN patch -p1 </ext-src/pg_anon.patch
-RUN patch -p1 </ext-src/pg_cron.patch
+RUN case "${PG_VERSION}" in "v17") \
+    echo "v17 extensions are not supported yet. Quit" && exit 0;; \
+    esac && \
+    patch -p1 </ext-src/pg_anon.patch
+RUN case "${PG_VERSION}" in "v17") \
+    echo "v17 extensions are not supported yet. Quit" && exit 0;; \
+    esac && \
+    patch -p1 </ext-src/pg_cron.patch
 ENV PATH=/usr/local/pgsql/bin:$PATH
 ENV PGHOST=compute
 ENV PGPORT=55433
@@ -995,9 +1195,23 @@ RUN mkdir /var/db && useradd -m -d /var/db/postgres postgres && \
 COPY --from=postgres-cleanup-layer --chown=postgres /usr/local/pgsql /usr/local
 COPY --from=compute-tools --chown=postgres /home/nonroot/target/release-line-debug-size-lto/compute_ctl /usr/local/bin/compute_ctl
 
+# pgbouncer and its config
+COPY --from=pgbouncer         /usr/local/pgbouncer/bin/pgbouncer /usr/local/bin/pgbouncer
+COPY --chmod=0666 --chown=postgres compute/etc/pgbouncer.ini /etc/pgbouncer.ini
+
+# Metrics exporter binaries and  configuration files
+COPY --from=postgres-exporter /bin/postgres_exporter /bin/postgres_exporter
+COPY --from=sql-exporter      /bin/sql_exporter      /bin/sql_exporter
+
+COPY --chmod=0644 compute/etc/sql_exporter.yml               /etc/sql_exporter.yml
+COPY --chmod=0644 compute/etc/neon_collector.yml             /etc/neon_collector.yml
+COPY --chmod=0644 compute/etc/sql_exporter_autoscaling.yml   /etc/sql_exporter_autoscaling.yml
+COPY --chmod=0644 compute/etc/neon_collector_autoscaling.yml /etc/neon_collector_autoscaling.yml
+
 # Create remote extension download directory
 RUN mkdir /usr/local/download_extensions && chown -R postgres:postgres /usr/local/download_extensions
 
+
 # Install:
 # libreadline8 for psql
 # libicu67, locales for collations (including ICU and plpgsql_check)

compute/README.md

@@ -0,0 +1,21 @@
+This directory contains files that are needed to build the compute
+images, or included in the compute images.
+
+Dockerfile.compute-node
+	To build the compute image
+
+vm-image-spec.yaml
+	Instructions for vm-builder, to turn the compute-node image into
+	corresponding vm-compute-node image.
+
+etc/
+	Configuration files included in /etc in the compute image
+
+patches/
+	Some extensions need to be patched to work with Neon. This
+	directory contains such patches. They are applied to the extension
+	sources in Dockerfile.compute-node
+
+In addition to these, postgres itself, the neon postgres extension,
+and compute_ctl are built and copied into the compute image by
+Dockerfile.compute-node.

compute/etc/neon_collector.yml

@@ -0,0 +1,247 @@
+collector_name: neon_collector
+metrics:
+- metric_name: lfc_misses
+  type: gauge
+  help: 'lfc_misses'
+  key_labels:
+  values: [lfc_misses]
+  query: |
+    select lfc_value as lfc_misses from neon.neon_lfc_stats where lfc_key='file_cache_misses';
+
+- metric_name: lfc_used
+  type: gauge
+  help: 'LFC chunks used (chunk = 1MB)'
+  key_labels:
+  values: [lfc_used]
+  query: |
+    select lfc_value as lfc_used from neon.neon_lfc_stats where lfc_key='file_cache_used';
+
+- metric_name: lfc_hits
+  type: gauge
+  help: 'lfc_hits'
+  key_labels:
+  values: [lfc_hits]
+  query: |
+    select lfc_value as lfc_hits from neon.neon_lfc_stats where lfc_key='file_cache_hits';
+
+- metric_name: lfc_writes
+  type: gauge
+  help: 'lfc_writes'
+  key_labels:
+  values: [lfc_writes]
+  query: |
+    select lfc_value as lfc_writes from neon.neon_lfc_stats where lfc_key='file_cache_writes';
+
+- metric_name: lfc_cache_size_limit
+  type: gauge
+  help: 'LFC cache size limit in bytes'
+  key_labels:
+  values: [lfc_cache_size_limit]
+  query: |
+    select pg_size_bytes(current_setting('neon.file_cache_size_limit')) as lfc_cache_size_limit;
+
+- metric_name: connection_counts
+  type: gauge
+  help: 'Connection counts'
+  key_labels:
+    - datname
+    - state
+  values: [count]
+  query: |
+    select datname, state, count(*) as count from pg_stat_activity where state <> '' group by datname, state;
+
+- metric_name: pg_stats_userdb
+  type: gauge
+  help: 'Stats for several oldest non-system dbs'
+  key_labels:
+    - datname
+  value_label: kind
+  values:
+    - db_size
+    - deadlocks
+    # Rows
+    - inserted
+    - updated
+    - deleted
+  # We export stats for 10 non-system database. Without this limit
+  # it is too easy to abuse the system by creating lots of databases.
+  query: |
+    select pg_database_size(datname) as db_size, deadlocks,
+       tup_inserted as inserted, tup_updated as updated, tup_deleted as deleted,
+       datname
+     from pg_stat_database
+     where datname IN (
+       select datname
+       from pg_database
+       where datname <> 'postgres' and not datistemplate
+       order by oid
+       limit 10
+     );
+
+- metric_name: max_cluster_size
+  type: gauge
+  help: 'neon.max_cluster_size setting'
+  key_labels:
+  values: [max_cluster_size]
+  query: |
+    select setting::int as max_cluster_size from pg_settings where name = 'neon.max_cluster_size';
+
+- metric_name: db_total_size
+  type: gauge
+  help: 'Size of all databases'
+  key_labels:
+  values: [total]
+  query: |
+    select sum(pg_database_size(datname)) as total from pg_database;
+
+# DEPRECATED
+- metric_name: lfc_approximate_working_set_size
+  type: gauge
+  help: 'Approximate working set size in pages of 8192 bytes'
+  key_labels:
+  values: [approximate_working_set_size]
+  query: |
+    select neon.approximate_working_set_size(false) as approximate_working_set_size;
+
+- metric_name: lfc_approximate_working_set_size_windows
+  type: gauge
+  help: 'Approximate working set size in pages of 8192 bytes'
+  key_labels: [duration]
+  values: [size]
+  # NOTE: This is the "public" / "human-readable" version. Here, we supply a small selection
+  # of durations in a pretty-printed form.
+  query: |
+    select
+      x as duration,
+      neon.approximate_working_set_size_seconds(extract('epoch' from x::interval)::int) as size
+    from
+      (values ('5m'),('15m'),('1h')) as t (x);
+
+- metric_name: compute_current_lsn
+  type: gauge
+  help: 'Current LSN of the database'
+  key_labels:
+  values: [lsn]
+  query: |
+    select
+      case
+        when pg_catalog.pg_is_in_recovery()
+        then (pg_last_wal_replay_lsn() - '0/0')::FLOAT8
+        else (pg_current_wal_lsn() - '0/0')::FLOAT8
+      end as lsn;
+
+- metric_name: compute_receive_lsn
+  type: gauge
+  help: 'Returns the last write-ahead log location that has been received and synced to disk by streaming replication'
+  key_labels:
+  values: [lsn]
+  query: |
+    SELECT
+      CASE
+        WHEN pg_catalog.pg_is_in_recovery()
+        THEN (pg_last_wal_receive_lsn() - '0/0')::FLOAT8
+        ELSE 0
+      END AS lsn;
+
+- metric_name: replication_delay_bytes
+  type: gauge
+  help: 'Bytes between received and replayed LSN'
+  key_labels:
+  values: [replication_delay_bytes]
+  # We use a GREATEST call here because this calculation can be negative.
+  # The calculation is not atomic, meaning after we've gotten the receive
+  # LSN, the replay LSN may have advanced past the receive LSN we
+  # are using for the calculation.
+  query: |
+    SELECT GREATEST(0, pg_wal_lsn_diff(pg_last_wal_receive_lsn(), pg_last_wal_replay_lsn())) AS replication_delay_bytes;
+
+- metric_name: replication_delay_seconds
+  type: gauge
+  help: 'Time since last LSN was replayed'
+  key_labels:
+  values: [replication_delay_seconds]
+  query: |
+    SELECT
+      CASE
+        WHEN pg_last_wal_receive_lsn() = pg_last_wal_replay_lsn() THEN 0
+        ELSE GREATEST (0, EXTRACT (EPOCH FROM now() - pg_last_xact_replay_timestamp()))
+     END AS replication_delay_seconds;
+
+- metric_name: checkpoints_req
+  type: gauge
+  help: 'Number of requested checkpoints'
+  key_labels:
+  values: [checkpoints_req]
+  query: |
+    SELECT checkpoints_req FROM pg_stat_bgwriter;
+
+- metric_name: checkpoints_timed
+  type: gauge
+  help: 'Number of scheduled checkpoints'
+  key_labels:
+  values: [checkpoints_timed]
+  query: |
+    SELECT checkpoints_timed FROM pg_stat_bgwriter;
+
+- metric_name: compute_logical_snapshot_files
+  type: gauge
+  help: 'Number of snapshot files in pg_logical/snapshot'
+  key_labels:
+    - timeline_id
+  values: [num_logical_snapshot_files]
+  query: |
+    SELECT
+      (SELECT setting FROM pg_settings WHERE name = 'neon.timeline_id') AS timeline_id,
+      -- Postgres creates temporary snapshot files of the form %X-%X.snap.%d.tmp. These
+      -- temporary snapshot files are renamed to the actual snapshot files after they are
+      -- completely built. We only WAL-log the completely built snapshot files.
+      (SELECT COUNT(*) FROM pg_ls_logicalsnapdir() WHERE name LIKE '%.snap') AS num_logical_snapshot_files;
+
+# In all the below metrics, we cast LSNs to floats because Prometheus only supports floats.
+# It's probably fine because float64 can store integers from -2^53 to +2^53 exactly.
+
+# Number of slots is limited by max_replication_slots, so collecting position for all of them shouldn't be bad.
+- metric_name: logical_slot_restart_lsn
+  type: gauge
+  help: 'restart_lsn of logical slots'
+  key_labels:
+    - slot_name
+  values: [restart_lsn]
+  query: |
+    select slot_name, (restart_lsn - '0/0')::FLOAT8 as restart_lsn
+    from pg_replication_slots
+    where slot_type = 'logical';
+
+- metric_name: compute_subscriptions_count
+  type: gauge
+  help: 'Number of logical replication subscriptions grouped by enabled/disabled'
+  key_labels:
+    - enabled
+  values: [subscriptions_count]
+  query: |
+    select subenabled::text as enabled, count(*) as subscriptions_count
+    from pg_subscription
+    group by subenabled;
+
+- metric_name: retained_wal
+  type: gauge
+  help: 'Retained WAL in inactive replication slots'
+  key_labels:
+    - slot_name
+  values: [retained_wal]
+  query: |
+    SELECT slot_name, pg_wal_lsn_diff(pg_current_wal_lsn(), restart_lsn)::FLOAT8 AS retained_wal
+    FROM pg_replication_slots
+    WHERE active = false;
+
+- metric_name: wal_is_lost
+  type: gauge
+  help: 'Whether or not the replication slot wal_status is lost'
+  key_labels:
+    - slot_name
+  values: [wal_is_lost]
+  query: |
+    SELECT slot_name,
+           CASE WHEN wal_status = 'lost' THEN 1 ELSE 0 END AS wal_is_lost
+    FROM pg_replication_slots;
+

compute/etc/neon_collector_autoscaling.yml

@@ -0,0 +1,55 @@
+collector_name: neon_collector_autoscaling
+metrics:
+- metric_name: lfc_misses
+  type: gauge
+  help: 'lfc_misses'
+  key_labels:
+  values: [lfc_misses]
+  query: |
+    select lfc_value as lfc_misses from neon.neon_lfc_stats where lfc_key='file_cache_misses';
+
+- metric_name: lfc_used
+  type: gauge
+  help: 'LFC chunks used (chunk = 1MB)'
+  key_labels:
+  values: [lfc_used]
+  query: |
+    select lfc_value as lfc_used from neon.neon_lfc_stats where lfc_key='file_cache_used';
+
+- metric_name: lfc_hits
+  type: gauge
+  help: 'lfc_hits'
+  key_labels:
+  values: [lfc_hits]
+  query: |
+    select lfc_value as lfc_hits from neon.neon_lfc_stats where lfc_key='file_cache_hits';
+
+- metric_name: lfc_writes
+  type: gauge
+  help: 'lfc_writes'
+  key_labels:
+  values: [lfc_writes]
+  query: |
+    select lfc_value as lfc_writes from neon.neon_lfc_stats where lfc_key='file_cache_writes';
+
+- metric_name: lfc_cache_size_limit
+  type: gauge
+  help: 'LFC cache size limit in bytes'
+  key_labels:
+  values: [lfc_cache_size_limit]
+  query: |
+    select pg_size_bytes(current_setting('neon.file_cache_size_limit')) as lfc_cache_size_limit;
+
+- metric_name: lfc_approximate_working_set_size_windows
+  type: gauge
+  help: 'Approximate working set size in pages of 8192 bytes'
+  key_labels: [duration_seconds]
+  values: [size]
+  # NOTE: This is the "internal" / "machine-readable" version. This outputs the working set
+  # size looking back 1..60 minutes, labeled with the number of minutes.
+  query: |
+    select
+      x::text as duration_seconds,
+      neon.approximate_working_set_size_seconds(x) as size
+    from
+      (select generate_series * 60 as x from generate_series(1, 60)) as t (x);

compute/etc/pgbouncer.ini

@@ -0,0 +1,17 @@
+[databases]
+*=host=localhost port=5432 auth_user=cloud_admin
+[pgbouncer]
+listen_port=6432
+listen_addr=0.0.0.0
+auth_type=scram-sha-256
+auth_user=cloud_admin
+auth_dbname=postgres
+client_tls_sslmode=disable
+server_tls_sslmode=disable
+pool_mode=transaction
+max_client_conn=10000
+default_pool_size=64
+max_prepared_statements=0
+admin_users=postgres
+unix_socket_dir=/tmp/
+unix_socket_mode=0777

compute/etc/sql_exporter.yml

@@ -0,0 +1,33 @@
+# Configuration for sql_exporter
+# Global defaults.
+global:
+  # If scrape_timeout <= 0, no timeout is set unless Prometheus provides one. The default is 10s.
+  scrape_timeout: 10s
+  # Subtracted from Prometheus' scrape_timeout to give us some headroom and prevent Prometheus from timing out first.
+  scrape_timeout_offset: 500ms
+  # Minimum interval between collector runs: by default (0s) collectors are executed on every scrape.
+  min_interval: 0s
+  # Maximum number of open connections to any one target. Metric queries will run concurrently on multiple connections,
+  # as will concurrent scrapes.
+  max_connections: 1
+  # Maximum number of idle connections to any one target. Unless you use very long collection intervals, this should
+  # always be the same as max_connections.
+  max_idle_connections: 1
+  # Maximum number of maximum amount of time a connection may be reused. Expired connections may be closed lazily before reuse.
+  # If 0, connections are not closed due to a connection's age.
+  max_connection_lifetime: 5m
+
+# The target to monitor and the collectors to execute on it.
+target:
+  # Data source name always has a URI schema that matches the driver name. In some cases (e.g. MySQL)
+  # the schema gets dropped or replaced to match the driver expected DSN format.
+  data_source_name: 'postgresql://cloud_admin@127.0.0.1:5432/postgres?sslmode=disable&application_name=sql_exporter'
+
+  # Collectors (referenced by name) to execute on the target.
+  # Glob patterns are supported (see <https://pkg.go.dev/path/filepath#Match> for syntax).
+  collectors: [neon_collector]
+
+# Collector files specifies a list of globs. One collector definition is read from each matching file.
+# Glob patterns are supported (see <https://pkg.go.dev/path/filepath#Match> for syntax).
+collector_files:
+  - "neon_collector.yml"

compute/etc/sql_exporter_autoscaling.yml

@@ -0,0 +1,33 @@
+# Configuration for sql_exporter for autoscaling-agent
+# Global defaults.
+global:
+  # If scrape_timeout <= 0, no timeout is set unless Prometheus provides one. The default is 10s.
+  scrape_timeout: 10s
+  # Subtracted from Prometheus' scrape_timeout to give us some headroom and prevent Prometheus from timing out first.
+  scrape_timeout_offset: 500ms
+  # Minimum interval between collector runs: by default (0s) collectors are executed on every scrape.
+  min_interval: 0s
+  # Maximum number of open connections to any one target. Metric queries will run concurrently on multiple connections,
+  # as will concurrent scrapes.
+  max_connections: 1
+  # Maximum number of idle connections to any one target. Unless you use very long collection intervals, this should
+  # always be the same as max_connections.
+  max_idle_connections: 1
+  # Maximum number of maximum amount of time a connection may be reused. Expired connections may be closed lazily before reuse.
+  # If 0, connections are not closed due to a connection's age.
+  max_connection_lifetime: 5m
+
+# The target to monitor and the collectors to execute on it.
+target:
+  # Data source name always has a URI schema that matches the driver name. In some cases (e.g. MySQL)
+  # the schema gets dropped or replaced to match the driver expected DSN format.
+  data_source_name: 'postgresql://cloud_admin@127.0.0.1:5432/postgres?sslmode=disable&application_name=sql_exporter_autoscaling'
+
+  # Collectors (referenced by name) to execute on the target.
+  # Glob patterns are supported (see <https://pkg.go.dev/path/filepath#Match> for syntax).
+  collectors: [neon_collector_autoscaling]
+
+# Collector files specifies a list of globs. One collector definition is read from each matching file.
+# Glob patterns are supported (see <https://pkg.go.dev/path/filepath#Match> for syntax).
+collector_files:
+  - "neon_collector_autoscaling.yml"

compute/vm-image-spec.yaml

@@ -0,0 +1,112 @@
+# Supplemental file for neondatabase/autoscaling's vm-builder, for producing the VM compute image.
+---
+commands:
+  - name: cgconfigparser
+    user: root
+    sysvInitAction: sysinit
+    shell: 'cgconfigparser -l /etc/cgconfig.conf -s 1664'
+  # restrict permissions on /neonvm/bin/resize-swap, because we grant access to compute_ctl for
+  # running it as root.
+  - name: chmod-resize-swap
+    user: root
+    sysvInitAction: sysinit
+    shell: 'chmod 711 /neonvm/bin/resize-swap'
+  - name: pgbouncer
+    user: postgres
+    sysvInitAction: respawn
+    shell: '/usr/local/bin/pgbouncer /etc/pgbouncer.ini'
+  - name: postgres-exporter
+    user: nobody
+    sysvInitAction: respawn
+    shell: 'DATA_SOURCE_NAME="user=cloud_admin sslmode=disable dbname=postgres application_name=postgres-exporter" /bin/postgres_exporter'
+  - name: sql-exporter
+    user: nobody
+    sysvInitAction: respawn
+    shell: '/bin/sql_exporter -config.file=/etc/sql_exporter.yml -web.listen-address=:9399'
+  - name: sql-exporter-autoscaling
+    user: nobody
+    sysvInitAction: respawn
+    shell: '/bin/sql_exporter -config.file=/etc/sql_exporter_autoscaling.yml -web.listen-address=:9499'
+shutdownHook: |
+  su -p postgres --session-command '/usr/local/bin/pg_ctl stop -D /var/db/postgres/compute/pgdata -m fast --wait -t 10'
+files:
+  - filename: compute_ctl-resize-swap
+    content: |
+      # Allow postgres user (which is what compute_ctl runs as) to run /neonvm/bin/resize-swap
+      # as root without requiring entering a password (NOPASSWD), regardless of hostname (ALL)
+      postgres ALL=(root) NOPASSWD: /neonvm/bin/resize-swap
+  - filename: cgconfig.conf
+    content: |
+      # Configuration for cgroups in VM compute nodes
+      group neon-postgres {
+          perm {
+              admin {
+                  uid = postgres;
+              }
+              task {
+                  gid = users;
+              }
+          }
+          memory {}
+      }
+build: |
+  # Build cgroup-tools
+  #
+  # At time of writing (2023-03-14), debian bullseye has a version of cgroup-tools (technically
+  # libcgroup) that doesn't support cgroup v2 (version 0.41-11). Unfortunately, the vm-monitor
+  # requires cgroup v2, so we'll build cgroup-tools ourselves.
+  FROM debian:bullseye-slim as libcgroup-builder
+  ENV LIBCGROUP_VERSION=v2.0.3
+
+  RUN set -exu \
+      && apt update \
+      && apt install --no-install-recommends -y \
+          git \
+          ca-certificates \
+          automake \
+          cmake \
+          make \
+          gcc \
+          byacc \
+          flex \
+          libtool \
+          libpam0g-dev \
+      && git clone --depth 1 -b $LIBCGROUP_VERSION https://github.com/libcgroup/libcgroup \
+      && INSTALL_DIR="/libcgroup-install" \
+      && mkdir -p "$INSTALL_DIR/bin" "$INSTALL_DIR/include" \
+      && cd libcgroup \
+      # extracted from bootstrap.sh, with modified flags:
+      && (test -d m4 || mkdir m4) \
+      && autoreconf -fi \
+      && rm -rf autom4te.cache \
+      && CFLAGS="-O3" ./configure --prefix="$INSTALL_DIR" --sysconfdir=/etc --localstatedir=/var --enable-opaque-hierarchy="name=systemd" \
+      # actually build the thing...
+      && make install
+merge: |
+  # tweak nofile limits
+  RUN set -e \
+      && echo 'fs.file-max = 1048576' >>/etc/sysctl.conf \
+      && test ! -e /etc/security || ( \
+         echo '*    - nofile 1048576' >>/etc/security/limits.conf \
+      && echo 'root - nofile 1048576' >>/etc/security/limits.conf \
+         )
+
+  # Allow postgres user (compute_ctl) to run swap resizer.
+  # Need to install sudo in order to allow this.
+  #
+  # Also, remove the 'read' permission from group/other on /neonvm/bin/resize-swap, just to be safe.
+  RUN set -e \
+      && apt update \
+      && apt install --no-install-recommends -y \
+             sudo \
+      && rm -rf /var/lib/apt/lists/* /tmp/* /var/tmp/*
+  COPY compute_ctl-resize-swap /etc/sudoers.d/compute_ctl-resize-swap
+
+  COPY cgconfig.conf /etc/cgconfig.conf
+
+  RUN set -e \
+      && chmod 0644 /etc/cgconfig.conf
+
+  COPY --from=libcgroup-builder /libcgroup-install/bin/*  /usr/bin/
+  COPY --from=libcgroup-builder /libcgroup-install/lib/*  /usr/lib/
+  COPY --from=libcgroup-builder /libcgroup-install/sbin/* /usr/sbin/

compute_tools/Cargo.toml

@@ -11,8 +11,6 @@ testing = []
 
 [dependencies]
 anyhow.workspace = true
-async-compression.workspace = true
-camino.workspace = true
 chrono.workspace = true
 cfg-if.workspace = true
 clap.workspace = true
@@ -25,8 +23,6 @@ num_cpus.workspace = true
 opentelemetry.workspace = true
 postgres.workspace = true
 regex.workspace = true
-scopeguard.workspace = true
-serde.workspace = true
 serde_json.workspace = true
 signal-hook.workspace = true
 tar.workspace = true
@@ -45,7 +41,6 @@ url.workspace = true
 compute_api.workspace = true
 utils.workspace = true
 workspace_hack.workspace = true
-toml_edit.workspace = true
 remote_storage = { version = "0.1", path = "../libs/remote_storage/" }
 vm_monitor = { version = "0.1", path = "../libs/vm_monitor/" }
 zstd = "0.13"

compute_tools/src/bin/compute_ctl.rs

@@ -43,7 +43,7 @@ use std::{thread, time::Duration};
 
 use anyhow::{Context, Result};
 use chrono::Utc;
-use clap::{Arg, ArgAction};
+use clap::Arg;
 use compute_tools::lsn_lease::launch_lsn_lease_bg_task_for_static;
 use signal_hook::consts::{SIGQUIT, SIGTERM};
 use signal_hook::{consts::SIGINT, iterator::Signals};
@@ -51,12 +51,13 @@ use tracing::{error, info, warn};
 use url::Url;
 
 use compute_api::responses::ComputeStatus;
-use compute_api::spec::{ComputeMode, ComputeSpec};
+use compute_api::spec::ComputeSpec;
 
 use compute_tools::compute::{
     forward_termination_signal, ComputeNode, ComputeState, ParsedSpec, PG_PID,
 };
 use compute_tools::configurator::launch_configurator;
+use compute_tools::extension_server::get_pg_version;
 use compute_tools::http::api::launch_http_server;
 use compute_tools::logger::*;
 use compute_tools::monitor::launch_monitor;
@@ -69,14 +70,13 @@ use rlimit::{setrlimit, Resource};
 // in-case of not-set environment var
 const BUILD_TAG_DEFAULT: &str = "latest";
 
-#[tokio::main]
-async fn main() -> Result<()> {
+fn main() -> Result<()> {
     let (build_tag, clap_args) = init()?;
 
     // enable core dumping for all child processes
     setrlimit(Resource::CORE, rlimit::INFINITY, rlimit::INFINITY)?;
 
-    let (http_handle, (pg_handle, start_pg_result)) = {
+    let (pg_handle, start_pg_result) = {
         // Enter startup tracing context
         let _startup_context_guard = startup_context_from_env();
 
@@ -84,61 +84,13 @@ async fn main() -> Result<()> {
 
         let cli_spec = try_spec_from_cli(&clap_args, &cli_args)?;
 
-        let compute = Arc::new(ComputeNode {
-            connstr: Url::parse(cli_args.connstr).context("cannot parse connstr as a URL")?,
-            pgdata: cli_args.pgdata.to_string(),
-            pgroot: cli_args.pgroot.to_string(),
-            pgversion: cli_args.pgversion.to_string(),
-            http_port: cli_args.http_port,
-            live_config_allowed: cli_spec.live_config_allowed,
-            state: Mutex::new({
-                let mut state = ComputeState::new();
+        let wait_spec_result = wait_spec(build_tag, cli_args, cli_spec)?;
 
-                if let Some(spec) = cli_spec.spec {
-                    let pspec = ParsedSpec::try_from(spec).map_err(|msg| anyhow::anyhow!(msg))?;
-                    info!("new pspec.spec: {:?}", pspec.spec);
-                    state.pspec = Some(pspec);
-                }
-
-                state
-            }),
-            state_changed: Condvar::new(),
-            ext_remote_storage: cli_args.ext_remote_storage.map(|s| s.to_string()),
-            ext_download_progress: RwLock::new(HashMap::new()),
-            build_tag: build_tag.clone(),
-        });
-
-        // If this is a pooled VM, prewarm before starting HTTP server and becoming
-        // available for binding. Prewarming helps Postgres start quicker later,
-        // because QEMU will already have its memory allocated from the host, and
-        // the necessary binaries will already be cached.
-        if compute.state.lock().unwrap().pspec.is_none() {
-            compute.prewarm_postgres()?;
-        }
-
-        // Launch http service first, so that we can serve control-plane requests
-        // while configuration is still in progress.
-        let http_handle = launch_http_server(cli_args.http_port, &compute)
-            .expect("cannot launch http endpoint thread");
-
-        wait_spec(&compute)?;
-
-        (
-            http_handle,
-            start_postgres(
-                compute,
-                #[cfg(target_os = "linux")]
-                &clap_args,
-                cli_args.resize_swap_on_bind,
-            )
-            .await?,
-        )
+        start_postgres(&clap_args, wait_spec_result)?
 
         // Startup is finished, exit the startup tracing span
     };
 
-    let _ = http_handle.join();
-
     // PostgreSQL is now running, if startup was successful. Wait until it exits.
     let wait_pg_result = wait_postgres(pg_handle)?;
 
@@ -168,14 +120,11 @@ fn init() -> Result<(String, clap::ArgMatches)> {
 }
 
 fn process_cli(matches: &clap::ArgMatches) -> Result<ProcessCliResult> {
-    let pgroot = matches
-        .get_one::<String>("pgroot")
+    let pgbin_default = "postgres";
+    let pgbin = matches
+        .get_one::<String>("pgbin")
         .map(|s| s.as_str())
-        .expect("pgroot is required");
-    let pgversion = matches
-        .get_one::<String>("pgversion")
-        .map(|s| s.as_str())
-        .expect("pgversion is required");
+        .unwrap_or(pgbin_default);
 
     let ext_remote_storage = matches
         .get_one::<String>("remote-ext-config")
@@ -206,8 +155,7 @@ fn process_cli(matches: &clap::ArgMatches) -> Result<ProcessCliResult> {
     Ok(ProcessCliResult {
         connstr,
         pgdata,
-        pgroot,
-        pgversion,
+        pgbin,
         ext_remote_storage,
         http_port,
         spec_json,
@@ -219,8 +167,7 @@ fn process_cli(matches: &clap::ArgMatches) -> Result<ProcessCliResult> {
 struct ProcessCliResult<'clap> {
     connstr: &'clap str,
     pgdata: &'clap str,
-    pgroot: &'clap str,
-    pgversion: &'clap str,
+    pgbin: &'clap str,
     ext_remote_storage: Option<&'clap str>,
     http_port: u16,
     spec_json: Option<&'clap String>,
@@ -338,8 +285,61 @@ struct CliSpecParams {
     live_config_allowed: bool,
 }
 
-fn wait_spec(compute: &Arc<ComputeNode>) -> Result<()> {
-    if compute.state.lock().unwrap().pspec.is_none() {
+fn wait_spec(
+    build_tag: String,
+    ProcessCliResult {
+        connstr,
+        pgdata,
+        pgbin,
+        ext_remote_storage,
+        resize_swap_on_bind,
+        http_port,
+        ..
+    }: ProcessCliResult,
+    CliSpecParams {
+        spec,
+        live_config_allowed,
+    }: CliSpecParams,
+) -> Result<WaitSpecResult> {
+    let mut new_state = ComputeState::new();
+    let spec_set;
+
+    if let Some(spec) = spec {
+        let pspec = ParsedSpec::try_from(spec).map_err(|msg| anyhow::anyhow!(msg))?;
+        info!("new pspec.spec: {:?}", pspec.spec);
+        new_state.pspec = Some(pspec);
+        spec_set = true;
+    } else {
+        spec_set = false;
+    }
+    let compute_node = ComputeNode {
+        connstr: Url::parse(connstr).context("cannot parse connstr as a URL")?,
+        pgdata: pgdata.to_string(),
+        pgbin: pgbin.to_string(),
+        pgversion: get_pg_version(pgbin),
+        live_config_allowed,
+        state: Mutex::new(new_state),
+        state_changed: Condvar::new(),
+        ext_remote_storage: ext_remote_storage.map(|s| s.to_string()),
+        ext_download_progress: RwLock::new(HashMap::new()),
+        build_tag,
+    };
+    let compute = Arc::new(compute_node);
+
+    // If this is a pooled VM, prewarm before starting HTTP server and becoming
+    // available for binding. Prewarming helps Postgres start quicker later,
+    // because QEMU will already have its memory allocated from the host, and
+    // the necessary binaries will already be cached.
+    if !spec_set {
+        compute.prewarm_postgres()?;
+    }
+
+    // Launch http service first, so that we can serve control-plane requests
+    // while configuration is still in progress.
+    let _http_handle =
+        launch_http_server(http_port, &compute).expect("cannot launch http endpoint thread");
+
+    if !spec_set {
         // No spec provided, hang waiting for it.
         info!("no compute spec provided, waiting");
 
@@ -369,13 +369,28 @@ fn wait_spec(compute: &Arc<ComputeNode>) -> Result<()> {
 
     launch_lsn_lease_bg_task_for_static(&compute);
 
-    Ok(())
+    Ok(WaitSpecResult {
+        compute,
+        http_port,
+        resize_swap_on_bind,
+    })
 }
 
-async fn start_postgres(
+struct WaitSpecResult {
     compute: Arc<ComputeNode>,
-    #[cfg(target_os = "linux")] matches: &clap::ArgMatches,
+    // passed through from ProcessCliResult
+    http_port: u16,
     resize_swap_on_bind: bool,
+}
+
+fn start_postgres(
+    // need to allow unused because `matches` is only used if target_os = "linux"
+    #[allow(unused_variables)] matches: &clap::ArgMatches,
+    WaitSpecResult {
+        compute,
+        http_port,
+        resize_swap_on_bind,
+    }: WaitSpecResult,
 ) -> Result<(Option<PostgresHandle>, StartPostgresResult)> {
     // We got all we need, update the state.
     let mut state = compute.state.lock().unwrap();
@@ -426,32 +441,27 @@ async fn start_postgres(
         }
     }
 
-    compute.prepare_compute().await?;
+    let extension_server_port: u16 = http_port;
 
+    // Start Postgres
     let mut pg = None;
     if !prestartup_failed {
-        match compute.get_mode() {
-            ComputeMode::Upgrade => {}
-            _ => {
-                // Start Postgres
-                pg = match compute.start_compute() {
-                    Ok(pg) => Some(pg),
-                    Err(err) => {
-                        error!("could not start the compute node: {:#}", err);
-                        let mut state = compute.state.lock().unwrap();
-                        state.error = Some(format!("{:?}", err));
-                        state.status = ComputeStatus::Failed;
-                        // Notify others that Postgres failed to start. In case of configuring the
-                        // empty compute, it's likely that API handler is still waiting for compute
-                        // state change. With this we will notify it that compute is in Failed state,
-                        // so control plane will know about it earlier and record proper error instead
-                        // of timeout.
-                        compute.state_changed.notify_all();
-                        drop(state); // unlock
-                        delay_exit = true;
-                        None
-                    }
-                }
+        pg = match compute.start_compute(extension_server_port) {
+            Ok(pg) => Some(pg),
+            Err(err) => {
+                error!("could not start the compute node: {:#}", err);
+                let mut state = compute.state.lock().unwrap();
+                state.error = Some(format!("{:?}", err));
+                state.status = ComputeStatus::Failed;
+                // Notify others that Postgres failed to start. In case of configuring the
+                // empty compute, it's likely that API handler is still waiting for compute
+                // state change. With this we will notify it that compute is in Failed state,
+                // so control plane will know about it earlier and record proper error instead
+                // of timeout.
+                compute.state_changed.notify_all();
+                drop(state); // unlock
+                delay_exit = true;
+                None
             }
         };
     } else {
@@ -676,17 +686,11 @@ fn cli() -> clap::Command {
                 .required(true),
         )
         .arg(
-            Arg::new("pgroot")
-                .short('R')
-                .long("pgroot")
-                .value_name("POSTGRES_ROOT")
-                .required(true),
-        )
-        .arg(
-            Arg::new("pgversion")
-                .long("pgversion")
-                .value_name("POSTGRES_VERSION")
-                .required(true),
+            Arg::new("pgbin")
+                .short('b')
+                .long("pgbin")
+                .default_value("postgres")
+                .value_name("POSTGRES_PATH"),
         )
         .arg(
             Arg::new("spec")
@@ -746,11 +750,6 @@ fn cli() -> clap::Command {
                 .long("resize-swap-on-bind")
                 .action(clap::ArgAction::SetTrue),
         )
-        .arg(
-            Arg::new("no-postgres")
-                .long("no-postgres")
-                .action(ArgAction::SetTrue),
-        )
 }
 
 /// When compute_ctl is killed, send also termination signal to sync-safekeepers

compute_tools/src/catalog.rs

@@ -4,7 +4,7 @@ use compute_api::{
 };
 use futures::Stream;
 use postgres::{Client, NoTls};
-use std::{process::Stdio, result::Result, sync::Arc};
+use std::{path::Path, process::Stdio, result::Result, sync::Arc};
 use tokio::{
     io::{AsyncBufReadExt, BufReader},
     process::Command,
@@ -55,7 +55,9 @@ pub async fn get_database_schema(
     compute: &Arc<ComputeNode>,
     dbname: &str,
 ) -> Result<impl Stream<Item = Result<bytes::Bytes, std::io::Error>>, SchemaDumpError> {
-    let pgdump = compute.get_my_pg_binary("pg_dump");
+    let pgbin = &compute.pgbin;
+    let basepath = Path::new(pgbin).parent().unwrap();
+    let pgdump = basepath.join("pg_dump");
     let mut connstr = compute.connstr.clone();
     connstr.set_path(dbname);
     let mut cmd = Command::new(pgdump)

compute_tools/src/compute.rs

@@ -1,9 +1,9 @@
 use std::collections::HashMap;
 use std::env;
 use std::fs;
+use std::io::BufRead;
 use std::os::unix::fs::{symlink, PermissionsExt};
 use std::path::Path;
-use std::path::PathBuf;
 use std::process::{Command, Stdio};
 use std::str::FromStr;
 use std::sync::atomic::AtomicU32;
@@ -13,8 +13,6 @@ use std::thread;
 use std::time::Instant;
 
 use anyhow::{Context, Result};
-use bytes::{Buf, BufMut};
-use camino::Utf8Path;
 use chrono::{DateTime, Utc};
 use futures::future::join_all;
 use futures::stream::FuturesUnordered;
@@ -22,15 +20,13 @@ use futures::StreamExt;
 use nix::unistd::Pid;
 use postgres::error::SqlState;
 use postgres::{Client, NoTls};
-use tokio;
-use tokio_postgres;
 use tracing::{debug, error, info, instrument, warn};
 use utils::id::{TenantId, TimelineId};
 use utils::lsn::Lsn;
-use utils::zstd::create_zst_tarball;
 
 use compute_api::responses::{ComputeMetrics, ComputeStatus};
 use compute_api::spec::{ComputeFeature, ComputeMode, ComputeSpec};
+use utils::measured_stream::MeasuredReader;
 
 use nix::sys::signal::{kill, Signal};
 
@@ -51,9 +47,8 @@ pub struct ComputeNode {
     // Url type maintains proper escaping
     pub connstr: url::Url,
     pub pgdata: String,
-    pub pgroot: String,
+    pub pgbin: String,
     pub pgversion: String,
-    pub http_port: u16,
     /// We should only allow live re- / configuration of the compute node if
     /// it uses 'pull model', i.e. it can go to control-plane and fetch
     /// the latest configuration. Otherwise, there could be a case:
@@ -314,13 +309,6 @@ impl ComputeNode {
         self.state.lock().unwrap().status
     }
 
-    /// Get the mode of this compute.
-    pub fn get_mode(&self) -> ComputeMode {
-        let state = self.state.lock().unwrap();
-
-        state.pspec.as_ref().unwrap().spec.mode
-    }
-
     // Remove `pgdata` directory and create it again with right permissions.
     fn create_pgdata(&self) -> Result<()> {
         // Ignore removal error, likely it is a 'No such file or directory (os error 2)'.
@@ -332,44 +320,15 @@ impl ComputeNode {
         Ok(())
     }
 
-    /// Get path pointing to requested binary directory.
-    pub fn get_pg_bindir(&self, version: &str) -> PathBuf {
-        Path::new(&self.pgroot)
-            .join(format!("v{}", version))
-            .join("bin")
-    }
-
-    /// Get path to requested Postgres binary.
-    pub fn get_pg_binary(&self, version: &str, binary: &str) -> String {
-        self.get_pg_bindir(version)
-            .join(binary)
-            .into_os_string()
-            .into_string()
-            .expect(&format!(
-                "path to {}-{} cannot be represented as UTF-8",
-                binary, version
-            ))
-    }
-
-    /// Get path to Postgres binary directory of this compute.
-    pub fn get_my_pg_bindir(&self) -> PathBuf {
-        self.get_pg_bindir(&self.pgversion)
-    }
-
-    /// Get path to specified Postgres binary of this compute.
-    pub fn get_my_pg_binary(&self, binary: &str) -> String {
-        self.get_pg_binary(&self.pgversion, binary)
-    }
-
     // Get basebackup from the libpq connection to pageserver using `connstr` and
     // unarchive it to `pgdata` directory overriding all its previous content.
     #[instrument(skip_all, fields(%lsn))]
-    async fn try_get_basebackup(&self, compute_state: &ComputeState, lsn: Lsn) -> Result<()> {
+    fn try_get_basebackup(&self, compute_state: &ComputeState, lsn: Lsn) -> Result<()> {
         let spec = compute_state.pspec.as_ref().expect("spec must be set");
         let start_time = Instant::now();
 
         let shard0_connstr = spec.pageserver_connstr.split(',').next().unwrap();
-        let mut config = tokio_postgres::Config::from_str(shard0_connstr)?;
+        let mut config = postgres::Config::from_str(shard0_connstr)?;
 
         // Use the storage auth token from the config file, if given.
         // Note: this overrides any password set in the connection string.
@@ -381,12 +340,7 @@ impl ComputeNode {
         }
 
         // Connect to pageserver
-        let (client, connection) = config.connect(tokio_postgres::NoTls).await?;
-        tokio::spawn(async move {
-            if let Err(e) = connection.await {
-                eprintln!("connection error: {}", e);
-            }
-        });
+        let mut client = config.connect(NoTls)?;
         let pageserver_connect_micros = start_time.elapsed().as_micros() as u64;
 
         let basebackup_cmd = match lsn {
@@ -398,18 +352,8 @@ impl ComputeNode {
             ),
         };
 
-        let mut copystream = std::pin::pin!(client.copy_out(basebackup_cmd.as_str()).await?);
-        let mut buf = bytes::BytesMut::with_capacity(1024);
-        while let Some(i) = copystream.next().await {
-            match i {
-                Ok(b) => {
-                    buf.put(b);
-                }
-                Err(e) => return Err(anyhow::anyhow!(e)),
-            }
-        }
-
-        let basebackup_bytes = buf.len();
+        let copyreader = client.copy_out(basebackup_cmd.as_str())?;
+        let mut measured_reader = MeasuredReader::new(copyreader);
 
         // Check the magic number to see if it's a gzip or not. Even though
         // we might explicitly ask for gzip, an old pageserver with no implementation
@@ -422,9 +366,11 @@ impl ComputeNode {
         // and 0x1f and 0x8b are unlikely first characters for any filename. Moreover,
         // we send the "global" directory first from the pageserver, so it definitely
         // won't be recognized as gzip.
-        let gzip = buf[0] == 0x1f && buf[1] == 0x8b;
-
-        let mut bufreader = buf.reader();
+        let mut bufreader = std::io::BufReader::new(&mut measured_reader);
+        let gzip = {
+            let peek = bufreader.fill_buf().unwrap();
+            peek[0] == 0x1f && peek[1] == 0x8b
+        };
 
         // Read the archive directly from the `CopyOutReader`
         //
@@ -444,14 +390,14 @@ impl ComputeNode {
         // Report metrics
         let mut state = self.state.lock().unwrap();
         state.metrics.pageserver_connect_micros = pageserver_connect_micros;
-        state.metrics.basebackup_bytes = basebackup_bytes as u64;
+        state.metrics.basebackup_bytes = measured_reader.get_byte_count() as u64;
         state.metrics.basebackup_ms = start_time.elapsed().as_millis() as u64;
         Ok(())
     }
 
     // Gets the basebackup in a retry loop
     #[instrument(skip_all, fields(%lsn))]
-    pub async fn get_basebackup(&self, compute_state: &ComputeState, lsn: Lsn) -> Result<()> {
+    pub fn get_basebackup(&self, compute_state: &ComputeState, lsn: Lsn) -> Result<()> {
         let mut retry_period_ms = 500.0;
         let mut attempts = 0;
         const DEFAULT_ATTEMPTS: u16 = 10;
@@ -464,7 +410,7 @@ impl ComputeNode {
         #[cfg(not(feature = "testing"))]
         let max_attempts = DEFAULT_ATTEMPTS;
         loop {
-            let result = self.try_get_basebackup(compute_state, lsn).await;
+            let result = self.try_get_basebackup(compute_state, lsn);
             match result {
                 Ok(_) => {
                     return result;
@@ -485,14 +431,10 @@ impl ComputeNode {
         }
     }
 
-    // Fast path for sync_safekeepers. If they're already synced we get the lsn
-    // in one roundtrip. If not, we should do a full sync_safekeepers.
-    pub async fn check_safekeepers_synced(
+    pub async fn check_safekeepers_synced_async(
         &self,
         compute_state: &ComputeState,
     ) -> Result<Option<Lsn>> {
-        let start_time = Utc::now();
-
         // Construct a connection config for each safekeeper
         let pspec: ParsedSpec = compute_state
             .pspec
@@ -561,7 +503,20 @@ impl ComputeNode {
             return Ok(None);
         }
 
-        let lsn = check_if_synced(responses);
+        Ok(check_if_synced(responses))
+    }
+
+    // Fast path for sync_safekeepers. If they're already synced we get the lsn
+    // in one roundtrip. If not, we should do a full sync_safekeepers.
+    pub fn check_safekeepers_synced(&self, compute_state: &ComputeState) -> Result<Option<Lsn>> {
+        let start_time = Utc::now();
+
+        // Run actual work with new tokio runtime
+        let rt = tokio::runtime::Builder::new_current_thread()
+            .enable_all()
+            .build()
+            .expect("failed to create rt");
+        let result = rt.block_on(self.check_safekeepers_synced_async(compute_state));
 
         // Record runtime
         self.state.lock().unwrap().metrics.sync_sk_check_ms = Utc::now()
@@ -569,8 +524,7 @@ impl ComputeNode {
             .to_std()
             .unwrap()
             .as_millis() as u64;
-
-        Ok(lsn)
+        result
     }
 
     // Run `postgres` in a special mode with `--sync-safekeepers` argument
@@ -579,8 +533,7 @@ impl ComputeNode {
     pub fn sync_safekeepers(&self, storage_auth_token: Option<String>) -> Result<Lsn> {
         let start_time = Utc::now();
 
-        let postgres = self.get_my_pg_binary("postgres");
-        let mut sync_handle = maybe_cgexec(&postgres)
+        let mut sync_handle = maybe_cgexec(&self.pgbin)
             .args(["--sync-safekeepers"])
             .env("PGDATA", &self.pgdata) // we cannot use -D in this mode
             .envs(if let Some(storage_auth_token) = &storage_auth_token {
@@ -636,7 +589,11 @@ impl ComputeNode {
     /// Do all the preparations like PGDATA directory creation, configuration,
     /// safekeepers sync, basebackup, etc.
     #[instrument(skip_all)]
-    pub async fn prepare_pgdata(&self, compute_state: &ComputeState) -> Result<()> {
+    pub fn prepare_pgdata(
+        &self,
+        compute_state: &ComputeState,
+        extension_server_port: u16,
+    ) -> Result<()> {
         let pspec = compute_state.pspec.as_ref().expect("spec must be set");
         let spec = &pspec.spec;
         let pgdata_path = Path::new(&self.pgdata);
@@ -646,7 +603,7 @@ impl ComputeNode {
         config::write_postgres_conf(
             &pgdata_path.join("postgresql.conf"),
             &pspec.spec,
-            Some(self.http_port),
+            Some(extension_server_port),
         )?;
 
         // Syncing safekeepers is only safe with primary nodes: if a primary
@@ -655,8 +612,7 @@ impl ComputeNode {
         let lsn = match spec.mode {
             ComputeMode::Primary => {
                 info!("checking if safekeepers are synced");
-                let lsn = if let Ok(Some(lsn)) = self.check_safekeepers_synced(compute_state).await
-                {
+                let lsn = if let Ok(Some(lsn)) = self.check_safekeepers_synced(compute_state) {
                     lsn
                 } else {
                     info!("starting safekeepers syncing");
@@ -674,24 +630,18 @@ impl ComputeNode {
                 info!("Initializing standby from latest Pageserver LSN");
                 Lsn(0)
             }
-            ComputeMode::Upgrade => {
-                info!("Starting upgrade node at latest Pageserver LSN");
-                Lsn(0)
-            }
         };
 
         info!(
             "getting basebackup@{} from pageserver {}",
             lsn, &pspec.pageserver_connstr
         );
-        self.get_basebackup(compute_state, lsn)
-            .await
-            .with_context(|| {
-                format!(
-                    "failed to get basebackup@{} from pageserver {}",
-                    lsn, &pspec.pageserver_connstr
-                )
-            })?;
+        self.get_basebackup(compute_state, lsn).with_context(|| {
+            format!(
+                "failed to get basebackup@{} from pageserver {}",
+                lsn, &pspec.pageserver_connstr
+            )
+        })?;
 
         // Update pg_hba.conf received with basebackup.
         update_pg_hba(pgdata_path)?;
@@ -739,7 +689,7 @@ impl ComputeNode {
         symlink("/dev/shm/", pgdata_path.join("pg_dynshmem"))?;
 
         match spec.mode {
-            ComputeMode::Primary | ComputeMode::Upgrade => {}
+            ComputeMode::Primary => {}
             ComputeMode::Replica | ComputeMode::Static(..) => {
                 add_standby_signal(pgdata_path)?;
             }
@@ -758,7 +708,7 @@ impl ComputeNode {
 
         // Run initdb to completion
         info!("running initdb");
-        let initdb_bin = self.get_my_pg_binary("initdb");
+        let initdb_bin = Path::new(&self.pgbin).parent().unwrap().join("initdb");
         Command::new(initdb_bin)
             .args(["-D", pgdata])
             .output()
@@ -774,8 +724,7 @@ impl ComputeNode {
 
         // Start postgres
         info!("starting postgres");
-        let postgres = self.get_my_pg_binary("postgres");
-        let mut pg = maybe_cgexec(&postgres)
+        let mut pg = maybe_cgexec(&self.pgbin)
             .args(["-D", pgdata])
             .spawn()
             .expect("cannot start postgres process");
@@ -808,8 +757,7 @@ impl ComputeNode {
         let pgdata_path = Path::new(&self.pgdata);
 
         // Run postgres as a child process.
-        let postgres = self.get_my_pg_binary("postgres");
-        let mut pg = maybe_cgexec(&postgres)
+        let mut pg = maybe_cgexec(&self.pgbin)
             .args(["-D", &self.pgdata])
             .envs(if let Some(storage_auth_token) = &storage_auth_token {
                 vec![("NEON_AUTH_TOKEN", storage_auth_token)]
@@ -830,152 +778,6 @@ impl ComputeNode {
         Ok((pg, logs_handle))
     }
 
-    pub async fn upgrade(&self, pg_version: &str) -> Result<()> {
-        let old_bindir = self.get_my_pg_bindir();
-        let new_bindir = self.get_pg_bindir(pg_version);
-
-        let old_datadir = Utf8Path::new(&self.pgdata);
-        let parent_dir = old_datadir.parent().unwrap();
-        let new_datadir = parent_dir.join("new-pgdata");
-
-        // Delete the new data directory before attempting, just in case it exists
-        let _ = std::fs::remove_dir_all(&new_datadir);
-
-        // Step 1: Create new cluster
-        info!(
-            "Running initdb to start a cluster upgrade from v{} to v{}",
-            self.pgversion, pg_version
-        );
-
-        let initdb_bin = self.get_pg_binary(pg_version, "initdb");
-        let mut initdb_cmd = Command::new(&initdb_bin);
-        initdb_cmd
-            .args(["--pgdata", new_datadir.as_str()])
-            .args(["--username", "cloud_admin"])
-            .args(["--encoding", "utf8"])
-            .args(["--auth-local", "trust"])
-            .env_clear()
-            .stdout(Stdio::inherit())
-            .stderr(Stdio::inherit());
-
-        match initdb_cmd.status() {
-            Ok(status) => {
-                if !status.success() {
-                    return Err(anyhow::anyhow!("failed to initialize the new database"));
-                }
-
-                info!("Initialized v{} database", pg_version);
-            }
-            Err(_) => {
-                return Err(anyhow::anyhow!(
-                    "failed to spawn initdb for the new database"
-                ))
-            }
-        };
-
-        // Step 2: Run pg_upgrade
-        info!(
-            "Running pg_upgrade to upgrade from v{} to v{}",
-            self.pgversion, pg_version
-        );
-
-        let pg_upgrade_bin = self.get_pg_binary(pg_version, "pg_upgrade");
-        let mut pg_upgrade_cmd = Command::new(&pg_upgrade_bin);
-        let mut child = pg_upgrade_cmd
-            .args([
-                "--old-bindir",
-                &old_bindir.into_os_string().into_string().unwrap(),
-            ])
-            .args(["--old-datadir", old_datadir.as_str()])
-            .args(["--old-options", "-c neon.safekeepers=''"])
-            .args([
-                "--new-bindir",
-                &new_bindir.into_os_string().into_string().unwrap(),
-            ])
-            .args(["--new-datadir", new_datadir.as_str()])
-            .args(["--new-options", "-c neon.safekeepers=''"])
-            .args(["--username", "cloud_admin"])
-            .arg("--no-transfer")
-            .stdout(Stdio::inherit())
-            .stderr(Stdio::inherit())
-            .spawn()?;
-
-        let status = child.wait()?;
-        if status.success() {
-            info!("pg_upgrade was successful");
-        } else {
-            return Err(anyhow::anyhow!(
-                "pg_upgrade failed with exit code {}",
-                status.code().unwrap()
-            ));
-        }
-
-        /* Step 3: Delete the script that pg_upgrade generates, which is created in the current
-         * working directory
-         */
-        // TODO: We should write a patch for upstream to not generate this file
-        if cfg!(windows) {
-            let _ = std::fs::remove_file("delete_old_cluster.bat");
-        } else {
-            let _ = std::fs::remove_file("delete_old_cluster.sh");
-        }
-
-        /* Step 4: Re-prepare the pgdata directory to work with the latest basebackup from the
-         * pageserver
-         */
-        {
-            let state = self.state.lock().unwrap().clone();
-
-            self.prepare_pgdata(&state).await?;
-        }
-
-        // Step 5: Create tarball minus things like pg_dynshm, etc.
-        info!("Creating tarball of upgraded initdb directory, minus some files");
-
-        let initdb_tar_path = parent_dir.join("initdb.tar.zst");
-        let _ = std::fs::remove_file(&initdb_tar_path);
-        create_zst_tarball(&new_datadir, &initdb_tar_path).await?;
-
-        // Step 6: Write new postgresql.conf file for upgraded initdb
-        std::fs::copy(
-            old_datadir.join("postgresql.conf"),
-            new_datadir.join("postgresql.conf"),
-        )?;
-
-        // Step 7: Write the tarball into the Postgres WAL
-        info!("Writing {} to WAL", initdb_tar_path.as_str());
-
-        let postgres_bin = self.get_my_pg_binary("postgres");
-        let mut wal_log_cmd = Command::new(&postgres_bin);
-        child = wal_log_cmd
-            .args(["--wal-log", initdb_tar_path.as_str()])
-            .env_clear()
-            .env("PGDATA", old_datadir.as_str())
-            .env("NEON_PURPOSE", "upgrade")
-            .stdout(Stdio::inherit())
-            .stderr(Stdio::inherit())
-            .spawn()
-            .expect("postgres --wal-log failed to start");
-        match child.wait() {
-            Ok(s) => {
-                if !s.success() {
-                    return Err(anyhow::anyhow!("Could not WAL log upgrade tarball"));
-                }
-            }
-            Err(e) => return Err(e.into()),
-        };
-
-        // Step 8: Sync the safekeepers to push WAL record to Neon
-        self.sync_safekeepers(None)?;
-
-        /* Note that whether any errors occur after this are unimportant. ALWAYS return success
-         * after this point. The compute will be terminated immediately after the upgrade. Remember
-         * that this is an upgrade-only compute, and it will not accept connections from users.
-         */
-
-        Ok(())
-    }
-
     /// Do post configuration of the already started Postgres. This function spawns a background thread to
     /// configure the database after applying the compute spec. Currently, it upgrades the neon extension
     /// version. In the future, it may upgrade all 3rd-party extensions.
@@ -1093,8 +895,8 @@ impl ComputeNode {
     // `pg_ctl` for start / stop.
     #[instrument(skip_all)]
     fn pg_reload_conf(&self) -> Result<()> {
-        let pgctl_bin = self.get_my_pg_binary("pg_ctl");
-        Command::new(&pgctl_bin)
+        let pgctl_bin = Path::new(&self.pgbin).parent().unwrap().join("pg_ctl");
+        Command::new(pgctl_bin)
             .args(["reload", "-D", &self.pgdata])
             .output()
             .expect("cannot run pg_ctl process");
@@ -1178,25 +980,43 @@ impl ComputeNode {
         Ok(())
     }
 
-    /// Prepares the compute for Postgres operations, including downloading
-    /// remote extensions and preparing the pgdata directory.
-    ///
-    /// The caller must hold the state mutex.
     #[instrument(skip_all)]
-    pub async fn prepare_compute(&self) -> Result<()> {
-        let state = self.state.lock().unwrap().clone();
-        let pspec = state.pspec.as_ref().expect("spec must be set");
-
+    pub fn start_compute(
+        &self,
+        extension_server_port: u16,
+    ) -> Result<(std::process::Child, std::thread::JoinHandle<()>)> {
+        let compute_state = self.state.lock().unwrap().clone();
+        let pspec = compute_state.pspec.as_ref().expect("spec must be set");
         info!(
-            "preparing compute for project {}, operation {}, tenant {}, timeline {}",
+            "starting compute for project {}, operation {}, tenant {}, timeline {}",
             pspec.spec.cluster.cluster_id.as_deref().unwrap_or("None"),
             pspec.spec.operation_uuid.as_deref().unwrap_or("None"),
             pspec.tenant_id,
             pspec.timeline_id,
         );
 
+        // tune pgbouncer
+        if let Some(pgbouncer_settings) = &pspec.spec.pgbouncer_settings {
+            info!("tuning pgbouncer");
+
+            let rt = tokio::runtime::Builder::new_current_thread()
+                .enable_all()
+                .build()
+                .expect("failed to create rt");
+
+            // Spawn a thread to do the tuning,
+            // so that we don't block the main thread that starts Postgres.
+            let pgbouncer_settings = pgbouncer_settings.clone();
+            let _handle = thread::spawn(move || {
+                let res = rt.block_on(tune_pgbouncer(pgbouncer_settings));
+                if let Err(err) = res {
+                    error!("error while tuning pgbouncer: {err:?}");
+                }
+            });
+        }
+
         info!(
-            "prepare_compute spec.remote_extensions {:?}",
+            "start_compute spec.remote_extensions {:?}",
             pspec.spec.remote_extensions
         );
 
@@ -1204,8 +1024,7 @@ impl ComputeNode {
         // remote shared_preload_libraries before postgres start (if any)
         if let Some(remote_extensions) = &pspec.spec.remote_extensions {
             // First, create control files for all availale extensions
-            let postgres_bin = self.get_my_pg_binary("postgres");
-            extension_server::create_control_files(remote_extensions, &postgres_bin);
+            extension_server::create_control_files(remote_extensions, &self.pgbin);
 
             let library_load_start_time = Utc::now();
             let remote_ext_metrics = self.prepare_preload_libraries(&pspec.spec)?;
@@ -1227,42 +1046,25 @@ impl ComputeNode {
             info!("{:?}", remote_ext_metrics);
         }
 
-        self.prepare_pgdata(&state).await?;
-
-        self.set_status(ComputeStatus::Prepared);
-
-        Ok(())
-    }
-
-    #[instrument(skip_all)]
-    pub fn start_compute(&self) -> Result<(std::process::Child, std::thread::JoinHandle<()>)> {
-        let compute_state = self.state.lock().unwrap().clone();
-        let pspec = compute_state.pspec.as_ref().expect("spec must be set");
+        self.prepare_pgdata(&compute_state, extension_server_port)?;
 
         let start_time = Utc::now();
         let pg_process = self.start_postgres(pspec.storage_auth_token.clone())?;
 
         let config_time = Utc::now();
-        if pspec.spec.mode == ComputeMode::Primary {
-            if !pspec.spec.skip_pg_catalog_updates {
-                let pgdata_path = Path::new(&self.pgdata);
-                // temporarily reset max_cluster_size in config
-                // to avoid the possibility of hitting the limit, while we are applying config:
-                // creating new extensions, roles, etc...
-                config::with_compute_ctl_tmp_override(
-                    pgdata_path,
-                    "neon.max_cluster_size=-1",
-                    || {
-                        self.pg_reload_conf()?;
-
-                        self.apply_config(&compute_state)?;
-
-                        Ok(())
-                    },
-                )?;
+        if pspec.spec.mode == ComputeMode::Primary && !pspec.spec.skip_pg_catalog_updates {
+            let pgdata_path = Path::new(&self.pgdata);
+            // temporarily reset max_cluster_size in config
+            // to avoid the possibility of hitting the limit, while we are applying config:
+            // creating new extensions, roles, etc...
+            config::with_compute_ctl_tmp_override(pgdata_path, "neon.max_cluster_size=-1", || {
                 self.pg_reload_conf()?;
-            }
-            self.post_apply_config()?;
+
+                self.apply_config(&compute_state)?;
+
+                Ok(())
+            })?;
+            self.pg_reload_conf()?;
         }
 
         let startup_end_time = Utc::now();
@@ -1343,11 +1145,9 @@ impl ComputeNode {
                 core_path.display()
             );
 
-            let postgres_bin = self.get_my_pg_binary("postgres");
-
             // Try first with gdb
             let backtrace = Command::new("gdb")
-                .args(["--batch", "-q", "-ex", "bt", &postgres_bin])
+                .args(["--batch", "-q", "-ex", "bt", &self.pgbin])
                 .arg(&core_path)
                 .output();
 
@@ -1480,12 +1280,11 @@ LIMIT 100",
         // then we try to download it here
         info!("downloading new extension {ext_archive_name}");
 
-        let postgres_bin = self.get_my_pg_binary("postgres");
         let download_size = extension_server::download_extension(
             &real_ext_name,
             &ext_path,
             ext_remote_storage,
-            &postgres_bin,
+            &self.pgbin,
         )
         .await
         .map_err(DownloadError::Other);

compute_tools/src/config.rs

@@ -74,7 +74,7 @@ pub fn write_postgres_conf(
     }
 
     match spec.mode {
-        ComputeMode::Primary | ComputeMode::Upgrade => {}
+        ComputeMode::Primary => {}
         ComputeMode::Static(lsn) => {
             // hot_standby is 'on' by default, but let's be explicit
             writeln!(file, "hot_standby=on")?;

compute_tools/src/extension_server.rs

@@ -75,6 +75,7 @@ use anyhow::Result;
 use anyhow::{bail, Context};
 use bytes::Bytes;
 use compute_api::spec::RemoteExtSpec;
+use regex::Regex;
 use remote_storage::*;
 use reqwest::StatusCode;
 use std::path::Path;
@@ -102,6 +103,35 @@ fn get_pg_config(argument: &str, pgbin: &str) -> String {
         .to_string()
 }
 
+pub fn get_pg_version(pgbin: &str) -> String {
+    // pg_config --version returns a (platform specific) human readable string
+    // such as "PostgreSQL 15.4". We parse this to v14/v15/v16 etc.
+    let human_version = get_pg_config("--version", pgbin);
+    return parse_pg_version(&human_version).to_string();
+}
+
+fn parse_pg_version(human_version: &str) -> &str {
+    // Normal releases have version strings like "PostgreSQL 15.4". But there
+    // are also pre-release versions like "PostgreSQL 17devel" or "PostgreSQL
+    // 16beta2" or "PostgreSQL 17rc1". And with the --with-extra-version
+    // configure option, you can tack any string to the version number,
+    // e.g. "PostgreSQL 15.4foobar".
+    match Regex::new(r"^PostgreSQL (?<major>\d+).+")
+        .unwrap()
+        .captures(human_version)
+    {
+        Some(captures) if captures.len() == 2 => match &captures["major"] {
+            "14" => return "v14",
+            "15" => return "v15",
+            "16" => return "v16",
+            "17" => return "v17",
+            _ => {}
+        },
+        _ => {}
+    }
+    panic!("Unsuported postgres version {human_version}");
+}
+
 // download the archive for a given extension,
 // unzip it, and place files in the appropriate locations (share/lib)
 pub async fn download_extension(
@@ -226,3 +256,42 @@ async fn download_extension_tar(ext_remote_storage: &str, ext_path: &str) -> Res
         ),
     }
 }
+
+#[cfg(test)]
+mod tests {
+    use super::parse_pg_version;
+
+    #[test]
+    fn test_parse_pg_version() {
+        assert_eq!(parse_pg_version("PostgreSQL 15.4"), "v15");
+        assert_eq!(parse_pg_version("PostgreSQL 15.14"), "v15");
+        assert_eq!(
+            parse_pg_version("PostgreSQL 15.4 (Ubuntu 15.4-0ubuntu0.23.04.1)"),
+            "v15"
+        );
+
+        assert_eq!(parse_pg_version("PostgreSQL 14.15"), "v14");
+        assert_eq!(parse_pg_version("PostgreSQL 14.0"), "v14");
+        assert_eq!(
+            parse_pg_version("PostgreSQL 14.9 (Debian 14.9-1.pgdg120+1"),
+            "v14"
+        );
+
+        assert_eq!(parse_pg_version("PostgreSQL 16devel"), "v16");
+        assert_eq!(parse_pg_version("PostgreSQL 16beta1"), "v16");
+        assert_eq!(parse_pg_version("PostgreSQL 16rc2"), "v16");
+        assert_eq!(parse_pg_version("PostgreSQL 16extra"), "v16");
+    }
+
+    #[test]
+    #[should_panic]
+    fn test_parse_pg_unsupported_version() {
+        parse_pg_version("PostgreSQL 13.14");
+    }
+
+    #[test]
+    #[should_panic]
+    fn test_parse_pg_incorrect_version_format() {
+        parse_pg_version("PostgreSQL 14");
+    }
+}

compute_tools/src/http/api.rs

@@ -10,7 +10,6 @@ use crate::catalog::{get_database_schema, get_dbs_and_roles};
 use crate::compute::forward_termination_signal;
 use crate::compute::{ComputeNode, ComputeState, ParsedSpec};
 use compute_api::requests::ConfigurationRequest;
-use compute_api::requests::UpgradeRequest;
 use compute_api::responses::{ComputeStatus, ComputeStatusResponse, GenericAPIError};
 
 use anyhow::Result;
@@ -138,20 +137,6 @@ async fn routes(req: Request<Body>, compute: &Arc<ComputeNode>) -> Response<Body
             }
         }
 
-        (&Method::POST, "/upgrade") => {
-            info!("serving /upgrade POST request");
-            match handle_upgrade_request(req, compute).await {
-                Ok(_) => Response::builder()
-                    .status(StatusCode::ACCEPTED)
-                    .body(Body::from("Starting upgrade"))
-                    .unwrap(),
-                Err((e, status)) => {
-                    error!("error handling /upgrade request: {e}");
-                    render_json_error(&format!("{}", e), status)
-                }
-            }
-        }
-
         (&Method::GET, "/dbs_and_roles") => {
             info!("serving /dbs_and_roles GET request",);
             match get_dbs_and_roles(compute).await {
@@ -412,59 +397,6 @@ async fn handle_terminate_request(compute: &Arc<ComputeNode>) -> Result<(), (Str
     Ok(())
 }
 
-async fn handle_upgrade_request(
-    req: Request<Body>,
-    compute: &Arc<ComputeNode>,
-) -> Result<(), (anyhow::Error, StatusCode)> {
-    let body_bytes = hyper::body::to_bytes(req.into_body()).await.unwrap();
-    let body_str = String::from_utf8(body_bytes.to_vec()).unwrap();
-
-    let body = match serde_json::from_str::<UpgradeRequest>(&body_str) {
-        Ok(r) => r,
-        Err(e) => return Err((Into::into(e), StatusCode::BAD_REQUEST)),
-    };
-
-    // No sense in trying to upgrade to the same version.
-    let curr_version = compute.pgversion.clone();
-    let new_version = body.pg_version;
-    if curr_version == new_version {
-        return Err((
-            anyhow::anyhow!("cannot upgrade endpoint to the same version"),
-            StatusCode::UNPROCESSABLE_ENTITY,
-        ));
-    }
-
-    // Check that we are in the running state before trying to upgrade.
-    match compute.get_status() {
-        ComputeStatus::Prepared => (),
-        ComputeStatus::Upgrading => {
-            return Err((
-                anyhow::anyhow!("upgrade already in progress"),
-                StatusCode::CONFLICT,
-            ));
-        }
-        _ => {
-            return Err((
-                anyhow::anyhow!("expected compute to be in the prepared state"),
-                StatusCode::CONFLICT,
-            ));
-        }
-    }
-
-    compute.set_status(ComputeStatus::Upgrading);
-
-    let c = compute.clone();
-    tokio::spawn(async move {
-        if let Err(e) = c.upgrade(&new_version).await {
-            error!("Failed to upgrade database: {}", e);
-        }
-
-        c.set_status(ComputeStatus::Running);
-    });
-
-    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>) {

compute_tools/src/http/openapi_spec.yaml

@@ -212,21 +212,6 @@ paths:
             application/json:
               schema:
                 $ref: "#/components/schemas/GenericError"
-  /upgrade:
-    post:
-      tags:
-        - Upgrade
-      summary: Upgrade project to another Postgres major version.
-      operationId: upgradePostgres
-      responses:
-        202:
-          description: Upgrade request in progress.
-        409:
-          description: Upgrade already in progress.
-        422:
-          description: Upgrade request could not be processed.
-        500:
-          description: Upgrade request failed.
 
   /terminate:
     post:

compute_tools/src/migrations/0011-grant_pg_show_replication_origin_status_to_neon_superuser.sql

@@ -0,0 +1 @@
+GRANT EXECUTE ON FUNCTION pg_show_replication_origin_status TO neon_superuser;

compute_tools/src/spec.rs

@@ -793,6 +793,9 @@ pub fn handle_migrations(client: &mut Client) -> Result<()> {
         include_str!(
             "./migrations/0010-grant_snapshot_synchronization_funcs_to_neon_superuser.sql"
         ),
+        include_str!(
+            "./migrations/0011-grant_pg_show_replication_origin_status_to_neon_superuser.sql"
+        ),
     ];
 
     MigrationRunner::new(client, &migrations).run_migrations()?;

control_plane/Cargo.toml

@@ -9,13 +9,10 @@ anyhow.workspace = true
 camino.workspace = true
 clap.workspace = true
 comfy-table.workspace = true
-futures.workspace = true
 git-version.workspace = true
 humantime.workspace = true
 nix.workspace = true
 once_cell.workspace = true
-postgres.workspace = true
-hex.workspace = true
 humantime-serde.workspace = true
 hyper.workspace = true
 regex.workspace = true
@@ -23,8 +20,6 @@ reqwest = { workspace = true, features = ["blocking", "json"] }
 scopeguard.workspace = true
 serde.workspace = true
 serde_json.workspace = true
-serde_with.workspace = true
-tar.workspace = true
 thiserror.workspace = true
 toml.workspace = true
 toml_edit.workspace = true

control_plane/src/background_process.rs

@@ -151,7 +151,7 @@ where
                     print!(".");
                     io::stdout().flush().unwrap();
                 }
-                thread::sleep(RETRY_INTERVAL);
+                tokio::time::sleep(RETRY_INTERVAL).await;
             }
             Err(e) => {
                 println!("error starting process {process_name:?}: {e:#}");

control_plane/src/bin/neon_local.rs

@@ -34,12 +34,14 @@ use safekeeper_api::{
     DEFAULT_HTTP_LISTEN_PORT as DEFAULT_SAFEKEEPER_HTTP_PORT,
     DEFAULT_PG_LISTEN_PORT as DEFAULT_SAFEKEEPER_PG_PORT,
 };
+use std::borrow::Cow;
 use std::collections::{BTreeSet, HashMap};
 use std::path::PathBuf;
 use std::process::exit;
 use std::str::FromStr;
 use std::time::Duration;
 use storage_broker::DEFAULT_LISTEN_ADDR as DEFAULT_BROKER_ADDR;
+use tokio::task::JoinSet;
 use url::Host;
 use utils::{
     auth::{Claims, Scope},
@@ -87,34 +89,35 @@ fn main() -> Result<()> {
 
     // Check for 'neon init' command first.
     let subcommand_result = if sub_name == "init" {
-        handle_init(sub_args).map(Some)
+        handle_init(sub_args).map(|env| Some(Cow::Owned(env)))
     } else {
         // all other commands need an existing config
-        let mut env =
-            LocalEnv::load_config(&local_env::base_path()).context("Error loading config")?;
-        let original_env = env.clone();
 
+        let env = LocalEnv::load_config(&local_env::base_path()).context("Error loading config")?;
+        let original_env = env.clone();
+        let env = Box::leak(Box::new(env));
         let rt = tokio::runtime::Builder::new_current_thread()
             .enable_all()
             .build()
             .unwrap();
 
         let subcommand_result = match sub_name {
-            "tenant" => rt.block_on(handle_tenant(sub_args, &mut env)),
-            "timeline" => rt.block_on(handle_timeline(sub_args, &mut env)),
-            "start" => rt.block_on(handle_start_all(&env, get_start_timeout(sub_args))),
-            "stop" => rt.block_on(handle_stop_all(sub_args, &env)),
-            "pageserver" => rt.block_on(handle_pageserver(sub_args, &env)),
-            "storage_controller" => rt.block_on(handle_storage_controller(sub_args, &env)),
-            "safekeeper" => rt.block_on(handle_safekeeper(sub_args, &env)),
-            "endpoint" => rt.block_on(handle_endpoint(sub_args, &env)),
-            "mappings" => handle_mappings(sub_args, &mut env),
+            "tenant" => rt.block_on(handle_tenant(sub_args, env)),
+            "timeline" => rt.block_on(handle_timeline(sub_args, env)),
+            "start" => rt.block_on(handle_start_all(env, get_start_timeout(sub_args))),
+            "stop" => rt.block_on(handle_stop_all(sub_args, env)),
+            "pageserver" => rt.block_on(handle_pageserver(sub_args, env)),
+            "storage_controller" => rt.block_on(handle_storage_controller(sub_args, env)),
+            "storage_broker" => rt.block_on(handle_storage_broker(sub_args, env)),
+            "safekeeper" => rt.block_on(handle_safekeeper(sub_args, env)),
+            "endpoint" => rt.block_on(handle_endpoint(sub_args, env)),
+            "mappings" => handle_mappings(sub_args, env),
             "pg" => bail!("'pg' subcommand has been renamed to 'endpoint'"),
             _ => bail!("unexpected subcommand {sub_name}"),
         };
 
-        if original_env != env {
-            subcommand_result.map(|()| Some(env))
+        if &original_env != env {
+            subcommand_result.map(|()| Some(Cow::Borrowed(env)))
         } else {
             subcommand_result.map(|()| None)
         }
@@ -798,19 +801,14 @@ async fn handle_endpoint(ep_match: &ArgMatches, env: &local_env::LocalEnv) -> Re
                 .get_one::<bool>("hot-standby")
                 .copied()
                 .unwrap_or(false);
-            let upgrade_only = sub_args
-                .get_one::<bool>("upgrade-only")
-                .copied()
-                .unwrap_or(false);
+
             let allow_multiple = sub_args.get_flag("allow-multiple");
 
-            let mode = match (lsn, hot_standby, upgrade_only) {
-                (Some(lsn), false, false) => ComputeMode::Static(lsn),
-                (None, true, false) => ComputeMode::Replica,
-                (None, false, false) => ComputeMode::Primary,
-                (None, false, true) => ComputeMode::Upgrade,
-                // Seeing this message means we aren't setting conflicts_with on clap arguments.
-                _ => anyhow::bail!("Invalid command line invocation"),
+            let mode = match (lsn, hot_standby) {
+                (Some(lsn), false) => ComputeMode::Static(lsn),
+                (None, true) => ComputeMode::Replica,
+                (None, false) => ComputeMode::Primary,
+                (Some(_), true) => anyhow::bail!("cannot specify both lsn and hot-standby"),
             };
 
             match (mode, hot_standby) {
@@ -1250,49 +1248,122 @@ async fn handle_safekeeper(sub_match: &ArgMatches, env: &local_env::LocalEnv) ->
     Ok(())
 }
 
+async fn handle_storage_broker(sub_match: &ArgMatches, env: &local_env::LocalEnv) -> Result<()> {
+    let (sub_name, sub_args) = match sub_match.subcommand() {
+        Some(broker_command_data) => broker_command_data,
+        None => bail!("no broker subcommand provided"),
+    };
+
+    match sub_name {
+        "start" => {
+            if let Err(e) = broker::start_broker_process(env, get_start_timeout(sub_args)).await {
+                eprintln!("broker start failed: {e}");
+                exit(1);
+            }
+        }
+
+        "stop" => {
+            if let Err(e) = broker::stop_broker_process(env) {
+                eprintln!("broker stop failed: {e}");
+                exit(1);
+            }
+        }
+
+        _ => bail!("Unexpected broker subcommand '{}'", sub_name),
+    }
+    Ok(())
+}
+
 async fn handle_start_all(
-    env: &local_env::LocalEnv,
+    env: &'static local_env::LocalEnv,
     retry_timeout: &Duration,
 ) -> anyhow::Result<()> {
+    let Err(errors) = handle_start_all_impl(env, *retry_timeout).await else {
+        neon_start_status_check(env, retry_timeout)
+            .await
+            .context("status check after successful startup of all services")?;
+        return Ok(());
+    };
+
+    eprintln!("startup failed because one or more services could not be started");
+
+    for e in errors {
+        eprintln!("{e}");
+        let debug_repr = format!("{e:?}");
+        for line in debug_repr.lines() {
+            eprintln!("  {line}");
+        }
+    }
+
+    try_stop_all(env, true).await;
+
+    exit(2);
+}
+
+/// Returns Ok() if and only if all services could be started successfully.
+/// Otherwise, returns the list of errors that occurred during startup.
+async fn handle_start_all_impl(
+    env: &'static local_env::LocalEnv,
+    retry_timeout: Duration,
+) -> Result<(), Vec<anyhow::Error>> {
     // Endpoints are not started automatically
 
-    broker::start_broker_process(env, retry_timeout).await?;
+    let mut js = JoinSet::new();
 
-    // Only start the storage controller if the pageserver is configured to need it
-    if env.control_plane_api.is_some() {
-        let storage_controller = StorageController::from_env(env);
-        if let Err(e) = storage_controller
-            .start(NeonStorageControllerStartArgs::with_default_instance_id(
-                (*retry_timeout).into(),
-            ))
-            .await
-        {
-            eprintln!("storage_controller start failed: {:#}", e);
-            try_stop_all(env, true).await;
-            exit(1);
+    // force infalliblity through closure
+    #[allow(clippy::redundant_closure_call)]
+    (|| {
+        js.spawn(async move {
+            let retry_timeout = retry_timeout;
+            broker::start_broker_process(env, &retry_timeout).await
+        });
+
+        // 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.into(),
+                    ))
+                    .await
+                    .map_err(|e| e.context("start storage_controller"))
+            });
+        }
+
+        for ps_conf in &env.pageservers {
+            js.spawn(async move {
+                let pageserver = PageServerNode::from_env(env, ps_conf);
+                pageserver
+                    .start(&retry_timeout)
+                    .await
+                    .map_err(|e| e.context(format!("start pageserver {}", ps_conf.id)))
+            });
+        }
+
+        for node in env.safekeepers.iter() {
+            js.spawn(async move {
+                let safekeeper = SafekeeperNode::from_env(env, node);
+                safekeeper
+                    .start(vec![], &retry_timeout)
+                    .await
+                    .map_err(|e| e.context(format!("start safekeeper {}", safekeeper.id)))
+            });
+        }
+    })();
+
+    let mut errors = Vec::new();
+    while let Some(result) = js.join_next().await {
+        let result = result.expect("we don't panic or cancel the tasks");
+        if let Err(e) = result {
+            errors.push(e);
         }
     }
 
-    for ps_conf in &env.pageservers {
-        let pageserver = PageServerNode::from_env(env, ps_conf);
-        if let Err(e) = pageserver.start(retry_timeout).await {
-            eprintln!("pageserver {} start failed: {:#}", ps_conf.id, e);
-            try_stop_all(env, true).await;
-            exit(1);
-        }
+    if !errors.is_empty() {
+        return Err(errors);
     }
 
-    for node in env.safekeepers.iter() {
-        let safekeeper = SafekeeperNode::from_env(env, node);
-        if let Err(e) = safekeeper.start(vec![], retry_timeout).await {
-            eprintln!("safekeeper {} start failed: {:#}", safekeeper.id, e);
-            try_stop_all(env, false).await;
-            exit(1);
-        }
-    }
-
-    neon_start_status_check(env, retry_timeout).await?;
-
     Ok(())
 }
 
@@ -1506,8 +1577,7 @@ fn cli() -> Command {
     let lsn_arg = Arg::new("lsn")
         .long("lsn")
         .help("Specify Lsn on the timeline to start from. By default, end of the timeline would be used.")
-        .required(false)
-        .conflicts_with("hot-standby");
+        .required(false);
 
     let hot_standby_arg = Arg::new("hot-standby")
         .value_parser(value_parser!(bool))
@@ -1678,6 +1748,19 @@ fn cli() -> Command {
                             .arg(stop_mode_arg.clone())
                             .arg(instance_id))
         )
+        .subcommand(
+            Command::new("storage_broker")
+                .arg_required_else_help(true)
+                .about("Manage broker")
+                .subcommand(Command::new("start")
+                            .about("Start broker")
+                            .arg(timeout_arg.clone())
+                )
+                .subcommand(Command::new("stop")
+                            .about("Stop broker")
+                            .arg(stop_mode_arg.clone())
+                )
+        )
         .subcommand(
             Command::new("safekeeper")
                 .arg_required_else_help(true)
@@ -1724,18 +1807,6 @@ fn cli() -> Command {
                     .arg(hot_standby_arg.clone())
                     .arg(update_catalog)
                     .arg(allow_multiple.clone())
-                    .arg(
-                        Arg::new("upgrade-only")
-                            .help("Mark this compute as an upgrade compute")
-                            .long("upgrade-only")
-                            .action(ArgAction::SetTrue)
-                            .conflicts_with_all(&[
-                                "config-only",
-                                "hot-standby",
-                                // Perhaps we could offer upgrades at a specific LSN in the future.
-                                "lsn",
-                            ])
-                    )
                 )
                 .subcommand(Command::new("start")
                     .about("Start postgres.\n If the endpoint doesn't exist yet, it is created.")

control_plane/src/endpoint.rs

@@ -182,7 +182,6 @@ impl ComputeControlPlane {
         tenant_id: TenantId,
         timeline_id: TimelineId,
     ) -> Result<()> {
-        // TODO: It really feels like I need to do some protection here
         if matches!(mode, ComputeMode::Primary) {
             // this check is not complete, as you could have a concurrent attempt at
             // creating another primary, both reading the state before checking it here,
@@ -394,7 +393,6 @@ impl Endpoint {
                     conf.append("recovery_prefetch", "off");
                 }
             }
-            ComputeMode::Upgrade => {}
         }
 
         Ok(conf)
@@ -504,7 +502,7 @@ impl Endpoint {
     /// Map safekeepers ids to the actual connection strings.
     fn build_safekeepers_connstrs(&self, sk_ids: Vec<NodeId>) -> Result<Vec<String>> {
         let mut safekeeper_connstrings = Vec::new();
-        if matches!(self.mode, ComputeMode::Primary | ComputeMode::Upgrade) {
+        if self.mode == ComputeMode::Primary {
             for sk_id in sk_ids {
                 let sk = self
                     .env
@@ -626,16 +624,13 @@ impl Endpoint {
                 self.endpoint_path().join("spec.json").to_str().unwrap(),
             ])
             .args([
-                "--pgroot",
-                &self
-                    .env
-                    .pg_distrib_dir
-                    .clone()
-                    .into_os_string()
-                    .into_string()
+                "--pgbin",
+                self.env
+                    .pg_bin_dir(self.pg_version)?
+                    .join("postgres")
+                    .to_str()
                     .unwrap(),
             ])
-            .args(["--pgversion", &self.pg_version.to_string()])
             .stdin(std::process::Stdio::null())
             .stderr(logfile.try_clone()?)
             .stdout(logfile);
@@ -679,7 +674,7 @@ impl Endpoint {
                             }
                             // keep retrying
                         }
-                        ComputeStatus::Running | ComputeStatus::Prepared => {
+                        ComputeStatus::Running => {
                             // All good!
                             break;
                         }
@@ -693,7 +688,6 @@ impl Endpoint {
                             );
                         }
                         ComputeStatus::Empty
-                        | ComputeStatus::Upgrading
                         | ComputeStatus::ConfigurationPending
                         | ComputeStatus::Configuration
                         | ComputeStatus::TerminationPending
@@ -708,7 +702,7 @@ impl Endpoint {
                     }
                 }
             }
-            std::thread::sleep(ATTEMPT_INTERVAL);
+            tokio::time::sleep(ATTEMPT_INTERVAL).await;
         }
 
         // disarm the scopeguard, let the child outlive this function (and neon_local invoction)

control_plane/src/local_env.rs

@@ -342,7 +342,7 @@ impl LocalEnv {
 
         #[allow(clippy::manual_range_patterns)]
         match pg_version {
-            14 | 15 | 16 => Ok(path.join(format!("v{pg_version}"))),
+            14 | 15 | 16 | 17 => Ok(path.join(format!("v{pg_version}"))),
             _ => bail!("Unsupported postgres version: {}", pg_version),
         }
     }

control_plane/src/pageserver.rs

@@ -17,9 +17,7 @@ use std::time::Duration;
 
 use anyhow::{bail, Context};
 use camino::Utf8PathBuf;
-use pageserver_api::models::{
-    self, AuxFilePolicy, LocationConfig, TenantHistorySize, TenantInfo, TimelineInfo,
-};
+use pageserver_api::models::{self, AuxFilePolicy, TenantInfo, TimelineInfo};
 use pageserver_api::shard::TenantShardId;
 use pageserver_client::mgmt_api;
 use postgres_backend::AuthType;
@@ -324,22 +322,6 @@ impl PageServerNode {
         background_process::stop_process(immediate, "pageserver", &self.pid_file())
     }
 
-    pub async fn page_server_psql_client(
-        &self,
-    ) -> anyhow::Result<(
-        tokio_postgres::Client,
-        tokio_postgres::Connection<tokio_postgres::Socket, tokio_postgres::tls::NoTlsStream>,
-    )> {
-        let mut config = self.pg_connection_config.clone();
-        if self.conf.pg_auth_type == AuthType::NeonJWT {
-            let token = self
-                .env
-                .generate_auth_token(&Claims::new(None, Scope::PageServerApi))?;
-            config = config.set_password(Some(token));
-        }
-        Ok(config.connect_no_tls().await?)
-    }
-
     pub async fn check_status(&self) -> mgmt_api::Result<()> {
         self.http_client.status().await
     }
@@ -540,19 +522,6 @@ impl PageServerNode {
         Ok(())
     }
 
-    pub async fn location_config(
-        &self,
-        tenant_shard_id: TenantShardId,
-        config: LocationConfig,
-        flush_ms: Option<Duration>,
-        lazy: bool,
-    ) -> anyhow::Result<()> {
-        Ok(self
-            .http_client
-            .location_config(tenant_shard_id, config, flush_ms, lazy)
-            .await?)
-    }
-
     pub async fn timeline_list(
         &self,
         tenant_shard_id: &TenantShardId,
@@ -636,14 +605,4 @@ impl PageServerNode {
 
         Ok(())
     }
-
-    pub async fn tenant_synthetic_size(
-        &self,
-        tenant_shard_id: TenantShardId,
-    ) -> anyhow::Result<TenantHistorySize> {
-        Ok(self
-            .http_client
-            .tenant_synthetic_size(tenant_shard_id)
-            .await?)
-    }
 }

control_plane/src/postgresql_conf.rs

@@ -4,13 +4,10 @@
 /// NOTE: This doesn't implement the full, correct postgresql.conf syntax. Just
 /// enough to extract a few settings we need in Neon, assuming you don't do
 /// funny stuff like include-directives or funny escaping.
-use anyhow::{bail, Context, Result};
 use once_cell::sync::Lazy;
 use regex::Regex;
 use std::collections::HashMap;
 use std::fmt;
-use std::io::BufRead;
-use std::str::FromStr;
 
 /// In-memory representation of a postgresql.conf file
 #[derive(Default, Debug)]
@@ -19,84 +16,16 @@ pub struct PostgresConf {
     hash: HashMap<String, String>,
 }
 
-static CONF_LINE_RE: Lazy<Regex> = Lazy::new(|| Regex::new(r"^((?:\w|\.)+)\s*=\s*(\S+)$").unwrap());
-
 impl PostgresConf {
     pub fn new() -> PostgresConf {
         PostgresConf::default()
     }
 
-    /// Read file into memory
-    pub fn read(read: impl std::io::Read) -> Result<PostgresConf> {
-        let mut result = Self::new();
-
-        for line in std::io::BufReader::new(read).lines() {
-            let line = line?;
-
-            // Store each line in a vector, in original format
-            result.lines.push(line.clone());
-
-            // Also parse each line and insert key=value lines into a hash map.
-            //
-            // FIXME: This doesn't match exactly the flex/bison grammar in PostgreSQL.
-            // But it's close enough for our usage.
-            let line = line.trim();
-            if line.starts_with('#') {
-                // comment, ignore
-                continue;
-            } else if let Some(caps) = CONF_LINE_RE.captures(line) {
-                let name = caps.get(1).unwrap().as_str();
-                let raw_val = caps.get(2).unwrap().as_str();
-
-                if let Ok(val) = deescape_str(raw_val) {
-                    // Note: if there's already an entry in the hash map for
-                    // this key, this will replace it. That's the behavior what
-                    // we want; when PostgreSQL reads the file, each line
-                    // overrides any previous value for the same setting.
-                    result.hash.insert(name.to_string(), val.to_string());
-                }
-            }
-        }
-        Ok(result)
-    }
-
     /// Return the current value of 'option'
     pub fn get(&self, option: &str) -> Option<&str> {
         self.hash.get(option).map(|x| x.as_ref())
     }
 
-    /// Return the current value of a field, parsed to the right datatype.
-    ///
-    /// This calls the FromStr::parse() function on the value of the field. If
-    /// the field does not exist, or parsing fails, returns an error.
-    ///
-    pub fn parse_field<T>(&self, field_name: &str, context: &str) -> Result<T>
-    where
-        T: FromStr,
-        <T as FromStr>::Err: std::error::Error + Send + Sync + 'static,
-    {
-        self.get(field_name)
-            .with_context(|| format!("could not find '{}' option {}", field_name, context))?
-            .parse::<T>()
-            .with_context(|| format!("could not parse '{}' option {}", field_name, context))
-    }
-
-    pub fn parse_field_optional<T>(&self, field_name: &str, context: &str) -> Result<Option<T>>
-    where
-        T: FromStr,
-        <T as FromStr>::Err: std::error::Error + Send + Sync + 'static,
-    {
-        if let Some(val) = self.get(field_name) {
-            let result = val
-                .parse::<T>()
-                .with_context(|| format!("could not parse '{}' option {}", field_name, context))?;
-
-            Ok(Some(result))
-        } else {
-            Ok(None)
-        }
-    }
-
     ///
     /// Note: if you call this multiple times for the same option, the config
     /// file will a line for each call. It would be nice to have a function
@@ -154,48 +83,8 @@ fn escape_str(s: &str) -> String {
     }
 }
 
-/// De-escape a possibly-quoted value.
-///
-/// See `DeescapeQuotedString` function in PostgreSQL sources for how PostgreSQL
-/// does this.
-fn deescape_str(s: &str) -> Result<String> {
-    // If the string has a quote at the beginning and end, strip them out.
-    if s.len() >= 2 && s.starts_with('\'') && s.ends_with('\'') {
-        let mut result = String::new();
-
-        let mut iter = s[1..(s.len() - 1)].chars().peekable();
-        while let Some(c) = iter.next() {
-            let newc = if c == '\\' {
-                match iter.next() {
-                    Some('b') => '\x08',
-                    Some('f') => '\x0c',
-                    Some('n') => '\n',
-                    Some('r') => '\r',
-                    Some('t') => '\t',
-                    Some('0'..='7') => {
-                        // TODO
-                        bail!("octal escapes not supported");
-                    }
-                    Some(n) => n,
-                    None => break,
-                }
-            } else if c == '\'' && iter.peek() == Some(&'\'') {
-                // doubled quote becomes just one quote
-                iter.next().unwrap()
-            } else {
-                c
-            };
-
-            result.push(newc);
-        }
-        Ok(result)
-    } else {
-        Ok(s.to_string())
-    }
-}
-
 #[test]
-fn test_postgresql_conf_escapes() -> Result<()> {
+fn test_postgresql_conf_escapes() -> anyhow::Result<()> {
     assert_eq!(escape_str("foo bar"), "'foo bar'");
     // these don't need to be quoted
     assert_eq!(escape_str("foo"), "foo");
@@ -214,13 +103,5 @@ fn test_postgresql_conf_escapes() -> Result<()> {
     assert_eq!(escape_str("fo\\o"), "'fo\\\\o'");
     assert_eq!(escape_str("10 cats"), "'10 cats'");
 
-    // Test de-escaping
-    assert_eq!(deescape_str(&escape_str("foo"))?, "foo");
-    assert_eq!(deescape_str(&escape_str("fo'o\nba\\r"))?, "fo'o\nba\\r");
-    assert_eq!(deescape_str("'\\b\\f\\n\\r\\t'")?, "\x08\x0c\n\r\t");
-
-    // octal-escapes are currently not supported
-    assert!(deescape_str("'foo\\7\\07\\007'").is_err());
-
     Ok(())
 }

control_plane/src/storage_controller.rs

@@ -28,6 +28,7 @@ use utils::{
     auth::{encode_from_key_file, Claims, Scope},
     id::{NodeId, TenantId},
 };
+use whoami::username;
 
 pub struct StorageController {
     env: LocalEnv,
@@ -183,7 +184,7 @@ impl StorageController {
     /// to other versions if that one isn't found.  Some automated tests create circumstances
     /// where only one version is available in pg_distrib_dir, such as `test_remote_extensions`.
     async fn get_pg_dir(&self, dir_name: &str) -> anyhow::Result<Utf8PathBuf> {
-        let prefer_versions = [STORAGE_CONTROLLER_POSTGRES_VERSION, 15, 14];
+        let prefer_versions = [STORAGE_CONTROLLER_POSTGRES_VERSION, 16, 15, 14];
 
         for v in prefer_versions {
             let path = Utf8PathBuf::from_path_buf(self.env.pg_dir(v, dir_name)?).unwrap();
@@ -211,7 +212,16 @@ impl StorageController {
     /// Readiness check for our postgres process
     async fn pg_isready(&self, pg_bin_dir: &Utf8Path, postgres_port: u16) -> anyhow::Result<bool> {
         let bin_path = pg_bin_dir.join("pg_isready");
-        let args = ["-h", "localhost", "-p", &format!("{}", postgres_port)];
+        let args = [
+            "-h",
+            "localhost",
+            "-U",
+            &username(),
+            "-d",
+            DB_NAME,
+            "-p",
+            &format!("{}", postgres_port),
+        ];
         let exitcode = Command::new(bin_path).args(args).spawn()?.wait().await?;
 
         Ok(exitcode.success())
@@ -225,7 +235,11 @@ impl StorageController {
     ///
     /// Returns the database url
     pub async fn setup_database(&self, postgres_port: u16) -> anyhow::Result<String> {
-        let database_url = format!("postgresql://localhost:{}/{DB_NAME}", postgres_port);
+        let database_url = format!(
+            "postgresql://{}@localhost:{}/{DB_NAME}",
+            &username(),
+            postgres_port
+        );
 
         let pg_bin_dir = self.get_pg_bin_dir().await?;
         let createdb_path = pg_bin_dir.join("createdb");
@@ -235,6 +249,10 @@ impl StorageController {
                 "localhost",
                 "-p",
                 &format!("{}", postgres_port),
+                "-U",
+                &username(),
+                "-O",
+                &username(),
                 DB_NAME,
             ])
             .output()
@@ -271,7 +289,7 @@ impl StorageController {
             // But tokio-postgres fork doesn't have this upstream commit:
             // https://github.com/sfackler/rust-postgres/commit/cb609be758f3fb5af537f04b584a2ee0cebd5e79
             // => we should rebase our fork => TODO https://github.com/neondatabase/neon/issues/8399
-            .user(&whoami::username())
+            .user(&username())
             .dbname(DB_NAME)
             .connect(tokio_postgres::NoTls)
             .await
@@ -328,6 +346,19 @@ impl StorageController {
             let pg_log_path = pg_data_path.join("postgres.log");
 
             if !tokio::fs::try_exists(&pg_data_path).await? {
+                let initdb_args = [
+                    "-D",
+                    pg_data_path.as_ref(),
+                    "--username",
+                    &username(),
+                    "--no-sync",
+                    "--no-instructions",
+                ];
+                tracing::info!(
+                    "Initializing storage controller database with args: {:?}",
+                    initdb_args
+                );
+
                 // Initialize empty database
                 let initdb_path = pg_bin_dir.join("initdb");
                 let mut child = Command::new(&initdb_path)
@@ -335,7 +366,7 @@ impl StorageController {
                         ("LD_LIBRARY_PATH".to_owned(), pg_lib_dir.to_string()),
                         ("DYLD_LIBRARY_PATH".to_owned(), pg_lib_dir.to_string()),
                     ])
-                    .args(["-D", pg_data_path.as_ref()])
+                    .args(initdb_args)
                     .spawn()
                     .expect("Failed to spawn initdb");
                 let status = child.wait().await?;
@@ -364,8 +395,14 @@ impl StorageController {
                 pg_data_path.as_ref(),
                 "-l",
                 pg_log_path.as_ref(),
+                "-U",
+                &username(),
                 "start",
             ];
+            tracing::info!(
+                "Starting storage controller database with args: {:?}",
+                db_start_args
+            );
 
             background_process::start_process(
                 "storage_controller_db",

control_plane/storcon_cli/Cargo.toml

@@ -11,14 +11,11 @@ clap.workspace = true
 comfy-table.workspace = true
 futures.workspace = true
 humantime.workspace = true
-hyper.workspace = true
 pageserver_api.workspace = true
 pageserver_client.workspace = true
 reqwest.workspace = true
-serde.workspace = true
 serde_json = { workspace = true, features = ["raw_value"] }
 storage_controller_client.workspace = true
-thiserror.workspace = true
 tokio.workspace = true
 tracing.workspace = true
 utils.workspace = true

docker-compose/README.md

@@ -2,8 +2,8 @@
 # Example docker compose configuration
 
 The configuration in this directory is used for testing Neon docker images: it is
-not intended for deploying a usable system.  To run a development environment where
-you can experiment with a minature Neon system, use `cargo neon` rather than container images.
+not intended for deploying a usable system. To run a development environment where
+you can experiment with a miniature Neon system, use `cargo neon` rather than container images.
 
 This configuration does not start the storage controller, because the controller
 needs a way to reconfigure running computes, and no such thing exists in this setup.

libs/compute_api/Cargo.toml

@@ -8,7 +8,6 @@ license.workspace = true
 anyhow.workspace = true
 chrono.workspace = true
 serde.workspace = true
-serde_with.workspace = true
 serde_json.workspace = true
 regex.workspace = true
 

libs/compute_api/src/requests.rs

@@ -12,9 +12,3 @@ use serde::Deserialize;
 pub struct ConfigurationRequest {
     pub spec: ComputeSpec,
 }
-
-/// Request body of the /upgrade API
-#[derive(Deserialize, Debug)]
-pub struct UpgradeRequest {
-    pub pg_version: String,
-}

libs/compute_api/src/responses.rs

@@ -44,15 +44,10 @@ pub enum ComputeStatus {
     // Compute node has spec and initial startup and
     // configuration is in progress.
     Init,
-    // Compute has been prepared, meaning that remote extensions have been
-    // downloaded and the data directory has been prepared.
-    Prepared,
     // Compute is configured and running.
     Running,
     // New spec is being applied.
     Configuration,
-    // Compute is upgrading Postgres.
-    Upgrading,
     // Either startup or configuration failed,
     // compute will exit soon or is waiting for
     // control-plane to terminate it.

libs/compute_api/src/spec.rs

@@ -199,11 +199,6 @@ pub enum ComputeMode {
     /// Future versions may want to distinguish between replicas with hot standby
     /// feedback and other kinds of replication configurations.
     Replica,
-    /// An upgrade-only node
-    ///
-    /// This node will not accept remote Postgres connections. It's only
-    /// purpose is to upgrade a timeline.
-    Upgrade,
 }
 
 #[derive(Clone, Debug, Default, Deserialize, Serialize)]

libs/consumption_metrics/Cargo.toml

@@ -5,9 +5,6 @@ edition = "2021"
 license = "Apache-2.0"
 
 [dependencies]
-anyhow.workspace = true
 chrono = { workspace = true, features = ["serde"] }
 rand.workspace = true
 serde.workspace = true
-serde_with.workspace = true
-utils.workspace = true

libs/consumption_metrics/src/lib.rs

@@ -5,7 +5,7 @@ use chrono::{DateTime, Utc};
 use rand::Rng;
 use serde::{Deserialize, Serialize};
 
-#[derive(Serialize, serde::Deserialize, Debug, Clone, Copy, Eq, PartialEq, Ord, PartialOrd)]
+#[derive(Serialize, Deserialize, Debug, Clone, Copy, Eq, PartialEq, Ord, PartialOrd)]
 #[serde(tag = "type")]
 pub enum EventType {
     #[serde(rename = "absolute")]
@@ -107,7 +107,7 @@ pub const CHUNK_SIZE: usize = 1000;
 
 // Just a wrapper around a slice of events
 // to serialize it as `{"events" : [ ] }
-#[derive(serde::Serialize, serde::Deserialize)]
+#[derive(serde::Serialize, Deserialize)]
 pub struct EventChunk<'a, T: Clone> {
     pub events: std::borrow::Cow<'a, [T]>,
 }

libs/desim/Cargo.toml

@@ -12,5 +12,4 @@ bytes.workspace = true
 utils.workspace = true
 parking_lot.workspace = true
 hex.workspace = true
-scopeguard.workspace = true
 smallvec = { workspace = true, features = ["write"] }

libs/pageserver_api/src/config.rs

@@ -104,9 +104,6 @@ pub struct ConfigToml {
     pub image_compression: ImageCompressionAlgorithm,
     pub ephemeral_bytes_per_memory_kb: usize,
     pub l0_flush: Option<crate::models::L0FlushConfig>,
-    #[serde(skip_serializing)]
-    // TODO(https://github.com/neondatabase/neon/issues/8184): remove after this field is removed from all pageserver.toml's
-    pub compact_level0_phase1_value_access: serde::de::IgnoredAny,
     pub virtual_file_direct_io: crate::models::virtual_file::DirectIoMode,
     pub io_buffer_alignment: usize,
 }
@@ -173,40 +170,6 @@ impl Default for EvictionOrder {
     }
 }
 
-#[derive(
-    Eq,
-    PartialEq,
-    Debug,
-    Copy,
-    Clone,
-    strum_macros::EnumString,
-    strum_macros::Display,
-    serde_with::DeserializeFromStr,
-    serde_with::SerializeDisplay,
-)]
-#[strum(serialize_all = "kebab-case")]
-pub enum GetVectoredImpl {
-    Sequential,
-    Vectored,
-}
-
-#[derive(
-    Eq,
-    PartialEq,
-    Debug,
-    Copy,
-    Clone,
-    strum_macros::EnumString,
-    strum_macros::Display,
-    serde_with::DeserializeFromStr,
-    serde_with::SerializeDisplay,
-)]
-#[strum(serialize_all = "kebab-case")]
-pub enum GetImpl {
-    Legacy,
-    Vectored,
-}
-
 #[derive(Copy, Clone, Debug, PartialEq, Eq, serde::Serialize, serde::Deserialize)]
 #[serde(transparent)]
 pub struct MaxVectoredReadBytes(pub NonZeroUsize);
@@ -338,8 +301,6 @@ pub mod defaults {
     pub const DEFAULT_IMAGE_COMPRESSION: ImageCompressionAlgorithm =
         ImageCompressionAlgorithm::Zstd { level: Some(1) };
 
-    pub const DEFAULT_VALIDATE_VECTORED_GET: bool = false;
-
     pub const DEFAULT_EPHEMERAL_BYTES_PER_MEMORY_KB: usize = 0;
 
     pub const DEFAULT_IO_BUFFER_ALIGNMENT: usize = 512;
@@ -376,7 +337,10 @@ impl Default for ConfigToml {
 
             concurrent_tenant_warmup: (NonZeroUsize::new(DEFAULT_CONCURRENT_TENANT_WARMUP)
                 .expect("Invalid default constant")),
-            concurrent_tenant_size_logical_size_queries: NonZeroUsize::new(1).unwrap(),
+            concurrent_tenant_size_logical_size_queries: NonZeroUsize::new(
+                DEFAULT_CONCURRENT_TENANT_SIZE_LOGICAL_SIZE_QUERIES,
+            )
+            .unwrap(),
             metric_collection_interval: (humantime::parse_duration(
                 DEFAULT_METRIC_COLLECTION_INTERVAL,
             )
@@ -417,7 +381,6 @@ impl Default for ConfigToml {
             image_compression: (DEFAULT_IMAGE_COMPRESSION),
             ephemeral_bytes_per_memory_kb: (DEFAULT_EPHEMERAL_BYTES_PER_MEMORY_KB),
             l0_flush: None,
-            compact_level0_phase1_value_access: Default::default(),
             virtual_file_direct_io: crate::models::virtual_file::DirectIoMode::default(),
 
             io_buffer_alignment: DEFAULT_IO_BUFFER_ALIGNMENT,
@@ -467,8 +430,6 @@ pub mod tenant_conf_defaults {
     // By default ingest enough WAL for two new L0 layers before checking if new image
     // image layers should be created.
     pub const DEFAULT_IMAGE_LAYER_CREATION_CHECK_THRESHOLD: u8 = 2;
-
-    pub const DEFAULT_INGEST_BATCH_SIZE: u64 = 100;
 }
 
 impl Default for TenantConfigToml {

libs/pageserver_api/src/key.rs

@@ -1,8 +1,8 @@
 use anyhow::{bail, Result};
 use byteorder::{ByteOrder, BE};
 use postgres_ffi::relfile_utils::{FSM_FORKNUM, VISIBILITYMAP_FORKNUM};
+use postgres_ffi::Oid;
 use postgres_ffi::RepOriginId;
-use postgres_ffi::{Oid, TransactionId};
 use serde::{Deserialize, Serialize};
 use std::{fmt, ops::Range};
 
@@ -350,7 +350,17 @@ impl Key {
 // 02 00000000 00000000 00000000 00   00000000
 //
 // TwoPhaseFile:
-// 02 00000000 00000000 00000000 00   XID
+//
+// 02 00000000 00000000 00XXXXXX XX   XXXXXXXX
+//
+//                        \______XID_________/
+//
+// The 64-bit XID is stored a little awkwardly in field6, field5 and
+// field4. PostgreSQL v16 and below only stored a 32-bit XID, which
+// fit completely in field6, but starting with PostgreSQL v17, a full
+// 64-bit XID is used. Most pageserver code that accesses
+// TwoPhaseFiles now deals with 64-bit XIDs even on v16, the high bits
+// are just unused.
 //
 // ControlFile:
 // 03 00000000 00000000 00000000 00   00000000
@@ -582,35 +592,36 @@ pub const TWOPHASEDIR_KEY: Key = Key {
 };
 
 #[inline(always)]
-pub fn twophase_file_key(xid: TransactionId) -> Key {
+pub fn twophase_file_key(xid: u64) -> Key {
     Key {
         field1: 0x02,
         field2: 0,
         field3: 0,
-        field4: 0,
-        field5: 0,
-        field6: xid,
+        field4: ((xid & 0xFFFFFF0000000000) >> 40) as u32,
+        field5: ((xid & 0x000000FF00000000) >> 32) as u8,
+        field6: (xid & 0x00000000FFFFFFFF) as u32,
     }
 }
 
 #[inline(always)]
-pub fn twophase_key_range(xid: TransactionId) -> Range<Key> {
+pub fn twophase_key_range(xid: u64) -> Range<Key> {
+    // 64-bit XIDs really should not overflow
     let (next_xid, overflowed) = xid.overflowing_add(1);
 
     Key {
         field1: 0x02,
         field2: 0,
         field3: 0,
-        field4: 0,
-        field5: 0,
-        field6: xid,
+        field4: ((xid & 0xFFFFFF0000000000) >> 40) as u32,
+        field5: ((xid & 0x000000FF00000000) >> 32) as u8,
+        field6: (xid & 0x00000000FFFFFFFF) as u32,
     }..Key {
         field1: 0x02,
         field2: 0,
-        field3: 0,
-        field4: 0,
-        field5: u8::from(overflowed),
-        field6: next_xid,
+        field3: u32::from(overflowed),
+        field4: ((next_xid & 0xFFFFFF0000000000) >> 40) as u32,
+        field5: ((next_xid & 0x000000FF00000000) >> 32) as u8,
+        field6: (next_xid & 0x00000000FFFFFFFF) as u32,
     }
 }
 

libs/pageserver_api/src/models.rs

@@ -37,14 +37,11 @@ use bytes::{Buf, BufMut, Bytes, BytesMut};
 /// ```mermaid
 /// stateDiagram-v2
 ///
-///     [*] --> Loading: spawn_load()
 ///     [*] --> Attaching: spawn_attach()
 ///
-///     Loading --> Activating: activate()
 ///     Attaching --> Activating: activate()
 ///     Activating --> Active: infallible
 ///
-///     Loading --> Broken: load() failure
 ///     Attaching --> Broken: attach() failure
 ///
 ///     Active --> Stopping: set_stopping(), part of shutdown & detach
@@ -68,10 +65,6 @@ use bytes::{Buf, BufMut, Bytes, BytesMut};
 )]
 #[serde(tag = "slug", content = "data")]
 pub enum TenantState {
-    /// This tenant is being loaded from local disk.
-    ///
-    /// `set_stopping()` and `set_broken()` do not work in this state and wait for it to pass.
-    Loading,
     /// This tenant is being attached to the pageserver.
     ///
     /// `set_stopping()` and `set_broken()` do not work in this state and wait for it to pass.
@@ -121,8 +114,6 @@ impl TenantState {
             // But, our attach task might still be fetching the remote timelines, etc.
             // So, return `Maybe` while Attaching, making Console wait for the attach task to finish.
             Self::Attaching | Self::Activating(ActivatingFrom::Attaching) => Maybe,
-            // tenant mgr startup distinguishes attaching from loading via marker file.
-            Self::Loading | Self::Activating(ActivatingFrom::Loading) => Attached,
             // We only reach Active after successful load / attach.
             // So, call atttachment status Attached.
             Self::Active => Attached,
@@ -191,10 +182,11 @@ impl LsnLease {
 }
 
 /// The only [`TenantState`] variants we could be `TenantState::Activating` from.
+///
+/// XXX: We used to have more variants here, but now it's just one, which makes this rather
+/// useless. Remove, once we've checked that there's no client code left that looks at this.
 #[derive(Clone, Copy, Debug, PartialEq, Eq, serde::Serialize, serde::Deserialize)]
 pub enum ActivatingFrom {
-    /// Arrived to [`TenantState::Activating`] from [`TenantState::Loading`]
-    Loading,
     /// Arrived to [`TenantState::Activating`] from [`TenantState::Attaching`]
     Attaching,
 }
@@ -495,7 +487,7 @@ pub struct CompactionAlgorithmSettings {
     pub kind: CompactionAlgorithm,
 }
 
-#[derive(Debug, PartialEq, Eq, Clone, serde::Deserialize, serde::Serialize)]
+#[derive(Debug, PartialEq, Eq, Clone, Deserialize, Serialize)]
 #[serde(tag = "mode", rename_all = "kebab-case", deny_unknown_fields)]
 pub enum L0FlushConfig {
     #[serde(rename_all = "snake_case")]
@@ -1562,11 +1554,8 @@ mod tests {
 
     #[test]
     fn tenantstatus_activating_serde() {
-        let states = [
-            TenantState::Activating(ActivatingFrom::Loading),
-            TenantState::Activating(ActivatingFrom::Attaching),
-        ];
-        let expected = "[{\"slug\":\"Activating\",\"data\":\"Loading\"},{\"slug\":\"Activating\",\"data\":\"Attaching\"}]";
+        let states = [TenantState::Activating(ActivatingFrom::Attaching)];
+        let expected = "[{\"slug\":\"Activating\",\"data\":\"Attaching\"}]";
 
         let actual = serde_json::to_string(&states).unwrap();
 
@@ -1581,13 +1570,7 @@ mod tests {
     fn tenantstatus_activating_strum() {
         // tests added, because we use these for metrics
         let examples = [
-            (line!(), TenantState::Loading, "Loading"),
             (line!(), TenantState::Attaching, "Attaching"),
-            (
-                line!(),
-                TenantState::Activating(ActivatingFrom::Loading),
-                "Activating",
-            ),
             (
                 line!(),
                 TenantState::Activating(ActivatingFrom::Attaching),

libs/postgres_backend/Cargo.toml

@@ -5,10 +5,8 @@ edition.workspace = true
 license.workspace = true
 
 [dependencies]
-async-trait.workspace = true
 anyhow.workspace = true
 bytes.workspace = true
-futures.workspace = true
 rustls.workspace = true
 serde.workspace = true
 thiserror.workspace = true

libs/postgres_backend/src/lib.rs

@@ -280,16 +280,6 @@ pub struct PostgresBackend<IO> {
 
 pub type PostgresBackendTCP = PostgresBackend<tokio::net::TcpStream>;
 
-pub fn query_from_cstring(query_string: Bytes) -> Vec<u8> {
-    let mut query_string = query_string.to_vec();
-    if let Some(ch) = query_string.last() {
-        if *ch == 0 {
-            query_string.pop();
-        }
-    }
-    query_string
-}
-
 /// Cast a byte slice to a string slice, dropping null terminator if there's one.
 fn cstr_to_str(bytes: &[u8]) -> anyhow::Result<&str> {
     let without_null = bytes.strip_suffix(&[0]).unwrap_or(bytes);

libs/postgres_ffi/Cargo.toml

@@ -5,13 +5,10 @@ edition.workspace = true
 license.workspace = true
 
 [dependencies]
-rand.workspace = true
 regex.workspace = true
 bytes.workspace = true
-byteorder.workspace = true
 anyhow.workspace = true
 crc32c.workspace = true
-hex.workspace = true
 once_cell.workspace = true
 log.workspace = true
 memoffset.workspace = true

libs/postgres_ffi/build.rs

@@ -56,7 +56,7 @@ fn main() -> anyhow::Result<()> {
         PathBuf::from("pg_install")
     };
 
-    for pg_version in &["v14", "v15", "v16"] {
+    for pg_version in &["v14", "v15", "v16", "v17"] {
         let mut pg_install_dir_versioned = pg_install_dir.join(pg_version);
         if pg_install_dir_versioned.is_relative() {
             let cwd = env::current_dir().context("Failed to get current_dir")?;

libs/postgres_ffi/src/lib.rs

@@ -57,6 +57,7 @@ macro_rules! for_all_postgres_versions {
         $macro!(v14);
         $macro!(v15);
         $macro!(v16);
+        $macro!(v17);
     };
 }
 
@@ -91,6 +92,7 @@ macro_rules! dispatch_pgversion {
                 14 : v14,
                 15 : v15,
                 16 : v16,
+                17 : v17,
             ]
         )
     };
@@ -121,6 +123,7 @@ macro_rules! enum_pgversion_dispatch {
                 V14 : v14,
                 V15 : v15,
                 V16 : v16,
+                V17 : v17,
             ]
         )
     };
@@ -150,6 +153,7 @@ macro_rules! enum_pgversion {
                 V14 : v14,
                 V15 : v15,
                 V16 : v16,
+                V17 : v17,
             ]
         }
     };
@@ -162,6 +166,7 @@ macro_rules! enum_pgversion {
                 V14 : v14,
                 V15 : v15,
                 V16 : v16,
+                V17 : v17,
             ]
         }
     };

libs/postgres_ffi/src/pg_constants.rs

@@ -9,8 +9,8 @@
 //! comments on them.
 //!
 
+use crate::PageHeaderData;
 use crate::BLCKSZ;
-use crate::{PageHeaderData, XLogRecord};
 
 //
 // From pg_tablespace_d.h
@@ -152,6 +152,9 @@ pub const XLH_UPDATE_OLD_ALL_VISIBLE_CLEARED: u8 = (1 << 0) as u8;
 pub const XLH_UPDATE_NEW_ALL_VISIBLE_CLEARED: u8 = (1 << 1) as u8;
 pub const XLH_DELETE_ALL_VISIBLE_CLEARED: u8 = (1 << 0) as u8;
 
+// From heapam_xlog.h
+pub const XLOG_HEAP2_REWRITE: u8 = 0x00;
+
 // From replication/message.h
 pub const XLOG_LOGICAL_MESSAGE: u8 = 0x00;
 
@@ -167,16 +170,7 @@ pub const RM_RELMAP_ID: u8 = 7;
 pub const RM_STANDBY_ID: u8 = 8;
 pub const RM_HEAP2_ID: u8 = 9;
 pub const RM_HEAP_ID: u8 = 10;
-pub const RM_BTREE_ID: u8 = 11;
-pub const RM_HASH_ID: u8 = 12;
-pub const RM_GIN_ID: u8 = 13;
-pub const RM_GIST_ID: u8 = 14;
-pub const RM_SEQ_ID: u8 = 15;
-pub const RM_SPGIST_ID: u8 = 16;
-pub const RM_BRIN_ID: u8 = 17;
-pub const RM_COMMIT_TS_ID: u8 = 18;
 pub const RM_REPLORIGIN_ID: u8 = 19;
-pub const RM_GENERIC_ID: u8 = 20;
 pub const RM_LOGICALMSG_ID: u8 = 21;
 
 // from neon_rmgr.h
@@ -190,27 +184,9 @@ pub const XLOG_NEON_HEAP_UPDATE: u8 = 0x20;
 pub const XLOG_NEON_HEAP_HOT_UPDATE: u8 = 0x30;
 pub const XLOG_NEON_HEAP_LOCK: u8 = 0x40;
 pub const XLOG_NEON_HEAP_MULTI_INSERT: u8 = 0x50;
-pub const XLOG_NEON_FILE: u8 = 0x60;
 
 pub const XLOG_NEON_HEAP_VISIBLE: u8 = 0x40;
 
-#[repr(C)]
-#[derive(Debug)]
-pub enum XlNeonFileFiletype {
-    UPGRADE_TARBALL,
-}
-
-impl TryFrom<u8> for XlNeonFileFiletype {
-    type Error = ();
-
-    fn try_from(value: u8) -> Result<Self, Self::Error> {
-        match value {
-            0 => Ok(XlNeonFileFiletype::UPGRADE_TARBALL),
-            _ => Err(()),
-        }
-    }
-}
-
 // from xlogreader.h
 pub const XLR_INFO_MASK: u8 = 0x0F;
 pub const XLR_RMGR_INFO_MASK: u8 = 0xF0;
@@ -218,8 +194,6 @@ pub const XLR_RMGR_INFO_MASK: u8 = 0xF0;
 pub const XLOG_TBLSPC_CREATE: u8 = 0x00;
 pub const XLOG_TBLSPC_DROP: u8 = 0x10;
 
-pub const SIZEOF_XLOGRECORD: u32 = size_of::<XLogRecord>() as u32;
-
 //
 // from xlogrecord.h
 //
@@ -243,18 +217,21 @@ pub const BKPIMAGE_HAS_HOLE: u8 = 0x01; /* page image has "hole" */
 /* From transam.h */
 pub const FIRST_NORMAL_TRANSACTION_ID: u32 = 3;
 pub const INVALID_TRANSACTION_ID: u32 = 0;
-pub const FIRST_BOOTSTRAP_OBJECT_ID: u32 = 12000;
-pub const FIRST_NORMAL_OBJECT_ID: u32 = 16384;
 
+/* pg_control.h */
 pub const XLOG_CHECKPOINT_SHUTDOWN: u8 = 0x00;
 pub const XLOG_CHECKPOINT_ONLINE: u8 = 0x10;
-pub const XLP_FIRST_IS_CONTRECORD: u16 = 0x0001;
-pub const XLP_LONG_HEADER: u16 = 0x0002;
+pub const XLOG_PARAMETER_CHANGE: u8 = 0x60;
+pub const XLOG_END_OF_RECOVERY: u8 = 0x90;
 
 /* From xlog.h */
 pub const XLOG_REPLORIGIN_SET: u8 = 0x00;
 pub const XLOG_REPLORIGIN_DROP: u8 = 0x10;
 
+/* xlog_internal.h */
+pub const XLP_FIRST_IS_CONTRECORD: u16 = 0x0001;
+pub const XLP_LONG_HEADER: u16 = 0x0002;
+
 /* From replication/slot.h */
 pub const REPL_SLOT_ON_DISK_OFFSETOF_RESTART_LSN: usize = 4*4  /* offset of `slotdata` in ReplicationSlotOnDisk  */
    + 64 /* NameData */  + 4*4;
@@ -272,33 +249,6 @@ pub const VM_HEAPBLOCKS_PER_PAGE: u32 =
 /* From origin.c */
 pub const REPLICATION_STATE_MAGIC: u32 = 0x1257DADE;
 
-// List of subdirectories inside pgdata.
-// Copied from src/bin/initdb/initdb.c
-pub const PGDATA_SUBDIRS: [&str; 22] = [
-    "global",
-    "pg_wal/archive_status",
-    "pg_commit_ts",
-    "pg_dynshmem",
-    "pg_notify",
-    "pg_serial",
-    "pg_snapshots",
-    "pg_subtrans",
-    "pg_twophase",
-    "pg_multixact",
-    "pg_multixact/members",
-    "pg_multixact/offsets",
-    "base",
-    "base/1",
-    "pg_replslot",
-    "pg_tblspc",
-    "pg_stat",
-    "pg_stat_tmp",
-    "pg_xact",
-    "pg_logical",
-    "pg_logical/snapshots",
-    "pg_logical/mappings",
-];
-
 // Don't include postgresql.conf as it is inconvenient on node start:
 // we need postgresql.conf before basebackup to synchronize safekeepers
 // so no point in overwriting it during backup restore. Rest of the files

libs/postgres_ffi/src/pg_constants_v14.rs

@@ -5,6 +5,33 @@ pub const BKPIMAGE_IS_COMPRESSED: u8 = 0x02; /* page image is compressed */
 pub const BKPIMAGE_APPLY: u8 = 0x04; /* page image should be restored during replay */
 pub const SIZEOF_RELMAPFILE: usize = 512; /* sizeof(RelMapFile) in relmapper.c */
 
+// List of subdirectories inside pgdata.
+// Copied from src/bin/initdb/initdb.c
+pub const PGDATA_SUBDIRS: [&str; 22] = [
+    "global",
+    "pg_wal/archive_status",
+    "pg_commit_ts",
+    "pg_dynshmem",
+    "pg_notify",
+    "pg_serial",
+    "pg_snapshots",
+    "pg_subtrans",
+    "pg_twophase",
+    "pg_multixact",
+    "pg_multixact/members",
+    "pg_multixact/offsets",
+    "base",
+    "base/1",
+    "pg_replslot",
+    "pg_tblspc",
+    "pg_stat",
+    "pg_stat_tmp",
+    "pg_xact",
+    "pg_logical",
+    "pg_logical/snapshots",
+    "pg_logical/mappings",
+];
+
 pub fn bkpimg_is_compressed(bimg_info: u8) -> bool {
     (bimg_info & BKPIMAGE_IS_COMPRESSED) != 0
 }

libs/postgres_ffi/src/pg_constants_v15.rs

@@ -11,6 +11,8 @@ pub const BKPIMAGE_COMPRESS_ZSTD: u8 = 0x10; /* page image is compressed */
 
 pub const SIZEOF_RELMAPFILE: usize = 512; /* sizeof(RelMapFile) in relmapper.c */
 
+pub use super::super::v14::bindings::PGDATA_SUBDIRS;
+
 pub fn bkpimg_is_compressed(bimg_info: u8) -> bool {
     const ANY_COMPRESS_FLAG: u8 = BKPIMAGE_COMPRESS_PGLZ | BKPIMAGE_COMPRESS_LZ4 | BKPIMAGE_COMPRESS_ZSTD;
 

libs/postgres_ffi/src/pg_constants_v16.rs

@@ -11,6 +11,8 @@ pub const BKPIMAGE_COMPRESS_ZSTD: u8 = 0x10; /* page image is compressed */
 
 pub const SIZEOF_RELMAPFILE: usize = 524; /* sizeof(RelMapFile) in relmapper.c */
 
+pub use super::super::v14::bindings::PGDATA_SUBDIRS;
+
 pub fn bkpimg_is_compressed(bimg_info: u8) -> bool {
     const ANY_COMPRESS_FLAG: u8 = BKPIMAGE_COMPRESS_PGLZ | BKPIMAGE_COMPRESS_LZ4 | BKPIMAGE_COMPRESS_ZSTD;
 

libs/postgres_ffi/src/pg_constants_v17.rs

@@ -0,0 +1,55 @@
+pub const XACT_XINFO_HAS_DROPPED_STATS: u32 = 1u32 << 8;
+
+pub const XLOG_DBASE_CREATE_FILE_COPY: u8 = 0x00;
+pub const XLOG_DBASE_CREATE_WAL_LOG: u8 = 0x10;
+pub const XLOG_DBASE_DROP: u8 = 0x20;
+
+pub const BKPIMAGE_APPLY: u8 = 0x02; /* page image should be restored during replay */
+pub const BKPIMAGE_COMPRESS_PGLZ: u8 = 0x04; /* page image is compressed */
+pub const BKPIMAGE_COMPRESS_LZ4: u8 = 0x08; /* page image is compressed */
+pub const BKPIMAGE_COMPRESS_ZSTD: u8 = 0x10; /* page image is compressed */
+
+pub const SIZEOF_RELMAPFILE: usize = 524; /* sizeof(RelMapFile) in relmapper.c */
+
+// List of subdirectories inside pgdata.
+// Copied from src/bin/initdb/initdb.c
+pub const PGDATA_SUBDIRS: [&str; 23] = [
+    "global",
+    "pg_wal/archive_status",
+    "pg_wal/summaries",
+    "pg_commit_ts",
+    "pg_dynshmem",
+    "pg_notify",
+    "pg_serial",
+    "pg_snapshots",
+    "pg_subtrans",
+    "pg_twophase",
+    "pg_multixact",
+    "pg_multixact/members",
+    "pg_multixact/offsets",
+    "base",
+    "base/1",
+    "pg_replslot",
+    "pg_tblspc",
+    "pg_stat",
+    "pg_stat_tmp",
+    "pg_xact",
+    "pg_logical",
+    "pg_logical/snapshots",
+    "pg_logical/mappings",
+];
+
+pub fn bkpimg_is_compressed(bimg_info: u8) -> bool {
+    const ANY_COMPRESS_FLAG: u8 = BKPIMAGE_COMPRESS_PGLZ | BKPIMAGE_COMPRESS_LZ4 | BKPIMAGE_COMPRESS_ZSTD;
+
+    (bimg_info & ANY_COMPRESS_FLAG) != 0
+}
+
+
+pub const XLOG_HEAP2_PRUNE_ON_ACCESS: u8 = 0x10;
+pub const XLOG_HEAP2_PRUNE_VACUUM_SCAN: u8 = 0x20;
+pub const XLOG_HEAP2_PRUNE_VACUUM_CLEANUP: u8 = 0x30;
+
+
+pub const XLOG_OVERWRITE_CONTRECORD: u8 = 0xD0;
+pub const XLOG_CHECKPOINT_REDO: u8 = 0xE0;

libs/postgres_ffi/src/xlog_utils.rs

@@ -26,11 +26,12 @@ use bytes::{Buf, Bytes};
 use log::*;
 
 use serde::Serialize;
+use std::ffi::OsStr;
 use std::fs::File;
 use std::io::prelude::*;
 use std::io::ErrorKind;
 use std::io::SeekFrom;
-use std::path::{Path, PathBuf};
+use std::path::Path;
 use std::time::SystemTime;
 use utils::bin_ser::DeserializeError;
 use utils::bin_ser::SerializeError;
@@ -78,19 +79,34 @@ pub fn XLogFileName(tli: TimeLineID, logSegNo: XLogSegNo, wal_segsz_bytes: usize
     )
 }
 
-pub fn XLogFromFileName(fname: &str, wal_seg_size: usize) -> (XLogSegNo, TimeLineID) {
-    let tli = u32::from_str_radix(&fname[0..8], 16).unwrap();
-    let log = u32::from_str_radix(&fname[8..16], 16).unwrap() as XLogSegNo;
-    let seg = u32::from_str_radix(&fname[16..24], 16).unwrap() as XLogSegNo;
-    (log * XLogSegmentsPerXLogId(wal_seg_size) + seg, tli)
+pub fn XLogFromFileName(
+    fname: &OsStr,
+    wal_seg_size: usize,
+) -> anyhow::Result<(XLogSegNo, TimeLineID)> {
+    if let Some(fname_str) = fname.to_str() {
+        let tli = u32::from_str_radix(&fname_str[0..8], 16)?;
+        let log = u32::from_str_radix(&fname_str[8..16], 16)? as XLogSegNo;
+        let seg = u32::from_str_radix(&fname_str[16..24], 16)? as XLogSegNo;
+        Ok((log * XLogSegmentsPerXLogId(wal_seg_size) + seg, tli))
+    } else {
+        anyhow::bail!("non-ut8 filename: {:?}", fname);
+    }
 }
 
-pub fn IsXLogFileName(fname: &str) -> bool {
-    return fname.len() == XLOG_FNAME_LEN && fname.chars().all(|c| c.is_ascii_hexdigit());
+pub fn IsXLogFileName(fname: &OsStr) -> bool {
+    if let Some(fname) = fname.to_str() {
+        fname.len() == XLOG_FNAME_LEN && fname.chars().all(|c| c.is_ascii_hexdigit())
+    } else {
+        false
+    }
 }
 
-pub fn IsPartialXLogFileName(fname: &str) -> bool {
-    fname.ends_with(".partial") && IsXLogFileName(&fname[0..fname.len() - 8])
+pub fn IsPartialXLogFileName(fname: &OsStr) -> bool {
+    if let Some(fname) = fname.to_str() {
+        fname.ends_with(".partial") && IsXLogFileName(OsStr::new(&fname[0..fname.len() - 8]))
+    } else {
+        false
+    }
 }
 
 /// If LSN points to the beginning of the page, then shift it to first record,
@@ -260,13 +276,6 @@ fn open_wal_segment(seg_file_path: &Path) -> anyhow::Result<Option<File>> {
     }
 }
 
-pub fn main() {
-    let mut data_dir = PathBuf::new();
-    data_dir.push(".");
-    let wal_end = find_end_of_wal(&data_dir, WAL_SEGMENT_SIZE, Lsn(0)).unwrap();
-    println!("wal_end={:?}", wal_end);
-}
-
 impl XLogRecord {
     pub fn from_slice(buf: &[u8]) -> Result<XLogRecord, DeserializeError> {
         use utils::bin_ser::LeSer;

libs/postgres_ffi/wal_craft/Cargo.toml

@@ -9,7 +9,6 @@ anyhow.workspace = true
 clap.workspace = true
 env_logger.workspace = true
 log.workspace = true
-once_cell.workspace = true
 postgres.workspace = true
 postgres_ffi.workspace = true
 camino-tempfile.workspace = true

libs/postgres_ffi/wal_craft/src/lib.rs

@@ -7,6 +7,7 @@ use postgres_ffi::{WAL_SEGMENT_SIZE, XLOG_BLCKSZ};
 use postgres_ffi::{
     XLOG_SIZE_OF_XLOG_LONG_PHD, XLOG_SIZE_OF_XLOG_RECORD, XLOG_SIZE_OF_XLOG_SHORT_PHD,
 };
+use std::ffi::OsStr;
 use std::path::{Path, PathBuf};
 use std::process::Command;
 use std::time::{Duration, Instant};
@@ -26,7 +27,6 @@ macro_rules! xlog_utils_test {
 
 postgres_ffi::for_all_postgres_versions! { xlog_utils_test }
 
-#[derive(Debug, Clone, PartialEq, Eq)]
 pub struct Conf {
     pub pg_version: u32,
     pub pg_distrib_dir: PathBuf,
@@ -53,7 +53,7 @@ impl Conf {
 
         #[allow(clippy::manual_range_patterns)]
         match self.pg_version {
-            14 | 15 | 16 => Ok(path.join(format!("v{}", self.pg_version))),
+            14 | 15 | 16 | 17 => Ok(path.join(format!("v{}", self.pg_version))),
             _ => bail!("Unsupported postgres version: {}", self.pg_version),
         }
     }
@@ -136,8 +136,8 @@ impl Conf {
 
     pub fn pg_waldump(
         &self,
-        first_segment_name: &str,
-        last_segment_name: &str,
+        first_segment_name: &OsStr,
+        last_segment_name: &OsStr,
     ) -> anyhow::Result<std::process::Output> {
         let first_segment_file = self.datadir.join(first_segment_name);
         let last_segment_file = self.datadir.join(last_segment_name);

libs/postgres_ffi/wal_craft/src/xlog_utils_test.rs

@@ -4,6 +4,7 @@ use super::*;
 use crate::{error, info};
 use regex::Regex;
 use std::cmp::min;
+use std::ffi::OsStr;
 use std::fs::{self, File};
 use std::io::Write;
 use std::{env, str::FromStr};
@@ -54,7 +55,7 @@ fn test_end_of_wal<C: crate::Crafter>(test_name: &str) {
         .wal_dir()
         .read_dir()
         .unwrap()
-        .map(|f| f.unwrap().file_name().into_string().unwrap())
+        .map(|f| f.unwrap().file_name())
         .filter(|fname| IsXLogFileName(fname))
         .max()
         .unwrap();
@@ -70,11 +71,11 @@ fn test_end_of_wal<C: crate::Crafter>(test_name: &str) {
             start_lsn
         );
         for file in fs::read_dir(cfg.wal_dir()).unwrap().flatten() {
-            let fname = file.file_name().into_string().unwrap();
+            let fname = file.file_name();
             if !IsXLogFileName(&fname) {
                 continue;
             }
-            let (segno, _) = XLogFromFileName(&fname, WAL_SEGMENT_SIZE);
+            let (segno, _) = XLogFromFileName(&fname, WAL_SEGMENT_SIZE).unwrap();
             let seg_start_lsn = XLogSegNoOffsetToRecPtr(segno, 0, WAL_SEGMENT_SIZE);
             if seg_start_lsn > u64::from(*start_lsn) {
                 continue;
@@ -93,10 +94,10 @@ fn test_end_of_wal<C: crate::Crafter>(test_name: &str) {
     }
 }
 
-fn find_pg_waldump_end_of_wal(cfg: &crate::Conf, last_segment: &str) -> Lsn {
+fn find_pg_waldump_end_of_wal(cfg: &crate::Conf, last_segment: &OsStr) -> Lsn {
     // Get the actual end of WAL by pg_waldump
     let waldump_output = cfg
-        .pg_waldump("000000010000000000000001", last_segment)
+        .pg_waldump(OsStr::new("000000010000000000000001"), last_segment)
         .unwrap()
         .stderr;
     let waldump_output = std::str::from_utf8(&waldump_output).unwrap();
@@ -117,7 +118,7 @@ fn find_pg_waldump_end_of_wal(cfg: &crate::Conf, last_segment: &str) -> Lsn {
 
 fn check_end_of_wal(
     cfg: &crate::Conf,
-    last_segment: &str,
+    last_segment: &OsStr,
     start_lsn: Lsn,
     expected_end_of_wal: Lsn,
 ) {
@@ -132,7 +133,8 @@ fn check_end_of_wal(
     // Rename file to partial to actually find last valid lsn, then rename it back.
     fs::rename(
         cfg.wal_dir().join(last_segment),
-        cfg.wal_dir().join(format!("{}.partial", last_segment)),
+        cfg.wal_dir()
+            .join(format!("{}.partial", last_segment.to_str().unwrap())),
     )
     .unwrap();
     let wal_end = find_end_of_wal(&cfg.wal_dir(), WAL_SEGMENT_SIZE, start_lsn).unwrap();
@@ -142,7 +144,8 @@ fn check_end_of_wal(
     );
     assert_eq!(wal_end, expected_end_of_wal);
     fs::rename(
-        cfg.wal_dir().join(format!("{}.partial", last_segment)),
+        cfg.wal_dir()
+            .join(format!("{}.partial", last_segment.to_str().unwrap())),
         cfg.wal_dir().join(last_segment),
     )
     .unwrap();

libs/pq_proto/Cargo.toml

@@ -8,10 +8,8 @@ license.workspace = true
 bytes.workspace = true
 byteorder.workspace = true
 itertools.workspace = true
-pin-project-lite.workspace = true
 postgres-protocol.workspace = true
 rand.workspace = true
 tokio = { workspace = true, features = ["io-util"] }
-tracing.workspace = true
 thiserror.workspace = true
 serde.workspace = true

libs/remote_storage/Cargo.toml

@@ -13,14 +13,11 @@ aws-smithy-async.workspace = true
 aws-smithy-types.workspace = true
 aws-config.workspace = true
 aws-sdk-s3.workspace = true
-aws-credential-types.workspace = true
 bytes.workspace = true
 camino = { workspace = true, features = ["serde1"] }
-humantime.workspace = true
 humantime-serde.workspace = true
 hyper = { workspace = true, features = ["stream"] }
 futures.workspace = true
-rand.workspace = true
 serde.workspace = true
 serde_json.workspace = true
 tokio = { workspace = true, features = ["sync", "fs", "io-util"] }

libs/remote_storage/src/lib.rs

@@ -127,10 +127,6 @@ impl RemotePath {
         &self.0
     }
 
-    pub fn extension(&self) -> Option<&str> {
-        self.0.extension()
-    }
-
     pub fn strip_prefix(&self, p: &RemotePath) -> Result<&Utf8Path, std::path::StripPrefixError> {
         self.0.strip_prefix(&p.0)
     }

libs/safekeeper_api/Cargo.toml

@@ -6,6 +6,5 @@ license.workspace = true
 
 [dependencies]
 serde.workspace = true
-serde_with.workspace = true
 const_format.workspace = true
 utils.workspace = true

libs/tracing-utils/Cargo.toml

@@ -9,8 +9,9 @@ hyper.workspace = true
 opentelemetry = { workspace = true, features=["rt-tokio"] }
 opentelemetry-otlp = { workspace = true, default-features=false, features = ["http-proto", "trace", "http", "reqwest-client"] }
 opentelemetry-semantic-conventions.workspace = true
-reqwest = { workspace = true, default-features = false, features = ["rustls-tls"] }
 tokio = { workspace = true, features = ["rt", "rt-multi-thread"] }
 tracing.workspace = true
 tracing-opentelemetry.workspace = true
-tracing-subscriber.workspace = true
+
+[dev-dependencies]
+tracing-subscriber.workspace = true    # For examples in docs

libs/utils/Cargo.toml

@@ -42,7 +42,6 @@ tracing.workspace = true
 tracing-error.workspace = true
 tracing-subscriber = { workspace = true, features = ["json", "registry"] }
 rand.workspace = true
-serde_with.workspace = true
 strum.workspace = true
 strum_macros.workspace = true
 url.workspace = true

libs/utils/src/accum.rs

@@ -1,33 +0,0 @@
-/// A helper to "accumulate" a value similar to `Iterator::reduce`, but lets you
-/// feed the accumulated values by calling the 'accum' function, instead of having an
-/// iterator.
-///
-/// For example, to calculate the smallest value among some integers:
-///
-/// ```
-/// use utils::accum::Accum;
-///
-/// let values = [1, 2, 3];
-///
-/// let mut min_value: Accum<u32> = Accum(None);
-/// for new_value in &values {
-///     min_value.accum(std::cmp::min, *new_value);
-/// }
-///
-/// assert_eq!(min_value.0.unwrap(), 1);
-/// ```
-pub struct Accum<T>(pub Option<T>);
-impl<T: Copy> Accum<T> {
-    pub fn accum<F>(&mut self, func: F, new_value: T)
-    where
-        F: FnOnce(T, T) -> T,
-    {
-        // If there is no previous value, just store the new value.
-        // Otherwise call the function to decide which one to keep.
-        self.0 = Some(if let Some(accum) = self.0 {
-            func(accum, new_value)
-        } else {
-            new_value
-        });
-    }
-}

libs/utils/src/http/error.rs

@@ -82,7 +82,7 @@ impl ApiError {
                 StatusCode::INTERNAL_SERVER_ERROR,
             ),
             ApiError::InternalServerError(err) => HttpErrorBody::response_from_msg_and_status(
-                err.to_string(),
+                format!("{err:#}"), // use alternative formatting so that we give the cause without backtrace
                 StatusCode::INTERNAL_SERVER_ERROR,
             ),
         }

libs/utils/src/id.rs

@@ -88,12 +88,6 @@ impl<'de> Deserialize<'de> for Id {
 }
 
 impl Id {
-    pub fn get_from_buf(buf: &mut impl bytes::Buf) -> Id {
-        let mut arr = [0u8; 16];
-        buf.copy_to_slice(&mut arr);
-        Id::from(arr)
-    }
-
     pub fn from_slice(src: &[u8]) -> Result<Id, IdError> {
         if src.len() != 16 {
             return Err(IdError::SliceParseError(src.len()));
@@ -179,10 +173,6 @@ impl fmt::Debug for Id {
 macro_rules! id_newtype {
     ($t:ident) => {
         impl $t {
-            pub fn get_from_buf(buf: &mut impl bytes::Buf) -> $t {
-                $t(Id::get_from_buf(buf))
-            }
-
             pub fn from_slice(src: &[u8]) -> Result<$t, IdError> {
                 Ok($t(Id::from_slice(src)?))
             }

libs/utils/src/leaky_bucket.rs

@@ -21,7 +21,13 @@
 //!
 //! Another explaination can be found here: <https://brandur.org/rate-limiting>
 
-use std::{sync::Mutex, time::Duration};
+use std::{
+    sync::{
+        atomic::{AtomicU64, Ordering},
+        Mutex,
+    },
+    time::Duration,
+};
 
 use tokio::{sync::Notify, time::Instant};
 
@@ -128,6 +134,7 @@ impl LeakyBucketState {
 
 pub struct RateLimiter {
     pub config: LeakyBucketConfig,
+    pub sleep_counter: AtomicU64,
     pub state: Mutex<LeakyBucketState>,
     /// a queue to provide this fair ordering.
     pub queue: Notify,
@@ -144,6 +151,7 @@ impl Drop for Requeue<'_> {
 impl RateLimiter {
     pub fn with_initial_tokens(config: LeakyBucketConfig, initial_tokens: f64) -> Self {
         RateLimiter {
+            sleep_counter: AtomicU64::new(0),
             state: Mutex::new(LeakyBucketState::with_initial_tokens(
                 &config,
                 initial_tokens,
@@ -163,15 +171,16 @@ impl RateLimiter {
 
     /// returns true if we did throttle
     pub async fn acquire(&self, count: usize) -> bool {
-        let mut throttled = false;
-
         let start = tokio::time::Instant::now();
 
+        let start_count = self.sleep_counter.load(Ordering::Acquire);
+        let mut end_count = start_count;
+
         // wait until we are the first in the queue
         let mut notified = std::pin::pin!(self.queue.notified());
         if !notified.as_mut().enable() {
-            throttled = true;
             notified.await;
+            end_count = self.sleep_counter.load(Ordering::Acquire);
         }
 
         // notify the next waiter in the queue when we are done.
@@ -184,9 +193,22 @@ impl RateLimiter {
                 .unwrap()
                 .add_tokens(&self.config, start, count as f64);
             match res {
-                Ok(()) => return throttled,
+                Ok(()) => return end_count > start_count,
                 Err(ready_at) => {
-                    throttled = true;
+                    struct Increment<'a>(&'a AtomicU64);
+
+                    impl Drop for Increment<'_> {
+                        fn drop(&mut self) {
+                            self.0.fetch_add(1, Ordering::AcqRel);
+                        }
+                    }
+
+                    // increment the counter after we finish sleeping (or cancel this task).
+                    // this ensures that tasks that have already started the acquire will observe
+                    // the new sleep count when they are allowed to resume on the notify.
+                    let _inc = Increment(&self.sleep_counter);
+                    end_count += 1;
+
                     tokio::time::sleep_until(ready_at).await;
                 }
             }

libs/utils/src/lib.rs

@@ -43,16 +43,9 @@ pub mod logging;
 pub mod lock_file;
 pub mod pid_file;
 
-// Misc
-pub mod accum;
-pub mod shutdown;
-
 // Utility for binding TcpListeners with proper socket options.
 pub mod tcp_listener;
 
-// Utility for putting a raw file descriptor into non-blocking mode
-pub mod nonblock;
-
 // Default signal handling
 pub mod sentry_init;
 pub mod signals;

libs/utils/src/lsn.rs

@@ -1,6 +1,5 @@
 #![warn(missing_docs)]
 
-use camino::Utf8Path;
 use serde::{de::Visitor, Deserialize, Serialize};
 use std::fmt;
 use std::ops::{Add, AddAssign};
@@ -145,14 +144,6 @@ impl Lsn {
         i128::from(self.0) - i128::from(other)
     }
 
-    /// Parse an LSN from a filename in the form `0000000000000000`
-    pub fn from_filename<F>(filename: F) -> Result<Self, LsnParseError>
-    where
-        F: AsRef<Utf8Path>,
-    {
-        Lsn::from_hex(filename.as_ref().as_str())
-    }
-
     /// Parse an LSN from a string in the form `0000000000000000`
     pub fn from_hex<S>(s: S) -> Result<Self, LsnParseError>
     where

libs/utils/src/nonblock.rs

@@ -1,17 +0,0 @@
-use nix::fcntl::{fcntl, OFlag, F_GETFL, F_SETFL};
-use std::os::unix::io::RawFd;
-
-/// Put a file descriptor into non-blocking mode
-pub fn set_nonblock(fd: RawFd) -> Result<(), std::io::Error> {
-    let bits = fcntl(fd, F_GETFL)?;
-
-    // If F_GETFL returns some unknown bits, they should be valid
-    // for passing back to F_SETFL, too. If we left them out, the F_SETFL
-    // would effectively clear them, which is not what we want.
-    let mut flags = OFlag::from_bits_retain(bits);
-    flags |= OFlag::O_NONBLOCK;
-
-    fcntl(fd, F_SETFL(flags))?;
-
-    Ok(())
-}

libs/utils/src/shutdown.rs

@@ -1,7 +0,0 @@
-/// Immediately terminate the calling process without calling
-/// atexit callbacks, C runtime destructors etc. We mainly use
-/// this to protect coverage data from concurrent writes.
-pub fn exit_now(code: u8) -> ! {
-    // SAFETY: exiting is safe, the ffi is not safe
-    unsafe { nix::libc::_exit(code as _) };
-}

libs/utils/src/vec_map.rs

@@ -120,32 +120,6 @@ impl<K: Ord, V> VecMap<K, V> {
         Ok((None, delta_size))
     }
 
-    /// Split the map into two.
-    ///
-    /// The left map contains everything before `cutoff` (exclusive).
-    /// Right map contains `cutoff` and everything after (inclusive).
-    pub fn split_at(&self, cutoff: &K) -> (Self, Self)
-    where
-        K: Clone,
-        V: Clone,
-    {
-        let split_idx = self
-            .data
-            .binary_search_by_key(&cutoff, extract_key)
-            .unwrap_or_else(std::convert::identity);
-
-        (
-            VecMap {
-                data: self.data[..split_idx].to_vec(),
-                ordering: self.ordering,
-            },
-            VecMap {
-                data: self.data[split_idx..].to_vec(),
-                ordering: self.ordering,
-            },
-        )
-    }
-
     /// Move items from `other` to the end of `self`, leaving `other` empty.
     /// If the `other` ordering is different from `self` ordering
     /// `ExtendOrderingError` error will be returned.

libs/vm_monitor/Cargo.toml

@@ -15,13 +15,11 @@ anyhow.workspace = true
 axum.workspace = true
 clap.workspace = true
 futures.workspace = true
-inotify.workspace = true
 serde.workspace = true
 serde_json.workspace = true
 sysinfo.workspace = true
 tokio = { workspace = true, features = ["rt-multi-thread"] }
 tokio-postgres.workspace = true
-tokio-stream.workspace = true
 tokio-util.workspace = true
 tracing.workspace = true
 tracing-subscriber.workspace = true

libs/walproposer/build.rs

@@ -5,6 +5,8 @@ use std::{env, path::PathBuf, process::Command};
 
 use anyhow::{anyhow, Context};
 
+const WALPROPOSER_PG_VERSION: &str = "v17";
+
 fn main() -> anyhow::Result<()> {
     // Tell cargo to invalidate the built crate whenever the wrapper changes
     println!("cargo:rerun-if-changed=bindgen_deps.h");
@@ -36,7 +38,10 @@ fn main() -> anyhow::Result<()> {
     // Rebuild crate when libwalproposer.a changes
     println!("cargo:rerun-if-changed={walproposer_lib_search_str}/libwalproposer.a");
 
-    let pg_config_bin = pg_install_abs.join("v16").join("bin").join("pg_config");
+    let pg_config_bin = pg_install_abs
+        .join(WALPROPOSER_PG_VERSION)
+        .join("bin")
+        .join("pg_config");
     let inc_server_path: String = if pg_config_bin.exists() {
         let output = Command::new(pg_config_bin)
             .arg("--includedir-server")
@@ -53,7 +58,7 @@ fn main() -> anyhow::Result<()> {
             .into()
     } else {
         let server_path = pg_install_abs
-            .join("v16")
+            .join(WALPROPOSER_PG_VERSION)
             .join("include")
             .join("postgresql")
             .join("server")

pageserver/Cargo.toml

@@ -15,7 +15,6 @@ anyhow.workspace = true
 arc-swap.workspace = true
 async-compression.workspace = true
 async-stream.workspace = true
-async-trait.workspace = true
 bit_field.workspace = true
 byteorder.workspace = true
 bytes.workspace = true
@@ -23,12 +22,9 @@ camino.workspace = true
 camino-tempfile.workspace = true
 chrono = { workspace = true, features = ["serde"] }
 clap = { workspace = true, features = ["string"] }
-const_format.workspace = true
 consumption_metrics.workspace = true
 crc32c.workspace = true
-crossbeam-utils.workspace = true
 either.workspace = true
-flate2.workspace = true
 fail.workspace = true
 futures.workspace = true
 git-version.workspace = true
@@ -57,10 +53,6 @@ serde.workspace = true
 serde_json = { workspace = true, features = ["raw_value"] }
 serde_path_to_error.workspace = true
 serde_with.workspace = true
-signal-hook.workspace = true
-smallvec = { workspace = true, features = ["write"] }
-svg_fmt.workspace = true
-sync_wrapper.workspace = true
 sysinfo.workspace = true
 tokio-tar.workspace = true
 thiserror.workspace = true
@@ -73,7 +65,6 @@ tokio-stream.workspace = true
 tokio-util.workspace = true
 toml_edit = { workspace = true, features = [ "serde" ] }
 tracing.workspace = true
-twox-hash.workspace = true
 url.workspace = true
 walkdir.workspace = true
 metrics.workspace = true

pageserver/benches/bench_walredo.rs

@@ -1,7 +1,7 @@
 //! Quantify a single walredo manager's throughput under N concurrent callers.
 //!
 //! The benchmark implementation ([`bench_impl`]) is parametrized by
-//! - `redo_work` => [`Request::short_request`] or [`Request::medium_request`]
+//! - `redo_work` => an async closure that takes a `PostgresRedoManager` and performs one redo
 //! - `n_redos` => number of times the benchmark shell execute the `redo_work`
 //! - `nclients` => number of clients (more on this shortly).
 //!
@@ -10,7 +10,7 @@
 //! Each task executes the `redo_work` `n_redos/nclients` times.
 //!
 //! We exercise the following combinations:
-//! - `redo_work = short / medium``
+//! - `redo_work = ping / short / medium``
 //! - `nclients = [1, 2, 4, 8, 16, 32, 64, 128]`
 //!
 //! We let `criterion` determine the `n_redos` using `iter_custom`.
@@ -27,33 +27,43 @@
 //!
 //! # Reference Numbers
 //!
-//! 2024-04-15 on i3en.3xlarge
+//! 2024-09-18 on im4gn.2xlarge
 //!
 //! ```text
-//! short/1           time:   [24.584 µs 24.737 µs 24.922 µs]
-//! short/2           time:   [33.479 µs 33.660 µs 33.888 µs]
-//! short/4           time:   [42.713 µs 43.046 µs 43.440 µs]
-//! short/8           time:   [71.814 µs 72.478 µs 73.240 µs]
-//! short/16          time:   [132.73 µs 134.45 µs 136.22 µs]
-//! short/32          time:   [258.31 µs 260.73 µs 263.27 µs]
-//! short/64          time:   [511.61 µs 514.44 µs 517.51 µs]
-//! short/128         time:   [992.64 µs 998.23 µs 1.0042 ms]
-//! medium/1          time:   [110.11 µs 110.50 µs 110.96 µs]
-//! medium/2          time:   [153.06 µs 153.85 µs 154.99 µs]
-//! medium/4          time:   [317.51 µs 319.92 µs 322.85 µs]
-//! medium/8          time:   [638.30 µs 644.68 µs 652.12 µs]
-//! medium/16         time:   [1.2651 ms 1.2773 ms 1.2914 ms]
-//! medium/32         time:   [2.5117 ms 2.5410 ms 2.5720 ms]
-//! medium/64         time:   [4.8088 ms 4.8555 ms 4.9047 ms]
-//! medium/128        time:   [8.8311 ms 8.9849 ms 9.1263 ms]
+//! ping/1                  time:   [21.789 µs 21.918 µs 22.078 µs]
+//! ping/2                  time:   [27.686 µs 27.812 µs 27.970 µs]
+//! ping/4                  time:   [35.468 µs 35.671 µs 35.926 µs]
+//! ping/8                  time:   [59.682 µs 59.987 µs 60.363 µs]
+//! ping/16                 time:   [101.79 µs 102.37 µs 103.08 µs]
+//! ping/32                 time:   [184.18 µs 185.15 µs 186.36 µs]
+//! ping/64                 time:   [349.86 µs 351.45 µs 353.47 µs]
+//! ping/128                time:   [684.53 µs 687.98 µs 692.17 µs]
+//! short/1                 time:   [31.833 µs 32.126 µs 32.428 µs]
+//! short/2                 time:   [35.558 µs 35.756 µs 35.992 µs]
+//! short/4                 time:   [44.850 µs 45.138 µs 45.484 µs]
+//! short/8                 time:   [65.985 µs 66.379 µs 66.853 µs]
+//! short/16                time:   [127.06 µs 127.90 µs 128.87 µs]
+//! short/32                time:   [252.98 µs 254.70 µs 256.73 µs]
+//! short/64                time:   [497.13 µs 499.86 µs 503.26 µs]
+//! short/128               time:   [987.46 µs 993.45 µs 1.0004 ms]
+//! medium/1                time:   [137.91 µs 138.55 µs 139.35 µs]
+//! medium/2                time:   [192.00 µs 192.91 µs 194.07 µs]
+//! medium/4                time:   [389.62 µs 391.55 µs 394.01 µs]
+//! medium/8                time:   [776.80 µs 780.33 µs 784.77 µs]
+//! medium/16               time:   [1.5323 ms 1.5383 ms 1.5459 ms]
+//! medium/32               time:   [3.0120 ms 3.0226 ms 3.0350 ms]
+//! medium/64               time:   [5.7405 ms 5.7787 ms 5.8166 ms]
+//! medium/128              time:   [10.412 ms 10.574 ms 10.718 ms]
 //! ```
 
 use anyhow::Context;
 use bytes::{Buf, Bytes};
 use criterion::{BenchmarkId, Criterion};
+use once_cell::sync::Lazy;
 use pageserver::{config::PageServerConf, walrecord::NeonWalRecord, walredo::PostgresRedoManager};
 use pageserver_api::{key::Key, shard::TenantShardId};
 use std::{
+    future::Future,
     sync::Arc,
     time::{Duration, Instant},
 };
@@ -61,40 +71,59 @@ use tokio::{sync::Barrier, task::JoinSet};
 use utils::{id::TenantId, lsn::Lsn};
 
 fn bench(c: &mut Criterion) {
-    {
-        let nclients = [1, 2, 4, 8, 16, 32, 64, 128];
-        for nclients in nclients {
-            let mut group = c.benchmark_group("short");
-            group.bench_with_input(
-                BenchmarkId::from_parameter(nclients),
-                &nclients,
-                |b, nclients| {
-                    let redo_work = Arc::new(Request::short_input());
-                    b.iter_custom(|iters| bench_impl(Arc::clone(&redo_work), iters, *nclients));
-                },
-            );
-        }
-    }
-    {
-        let nclients = [1, 2, 4, 8, 16, 32, 64, 128];
-        for nclients in nclients {
-            let mut group = c.benchmark_group("medium");
-            group.bench_with_input(
-                BenchmarkId::from_parameter(nclients),
-                &nclients,
-                |b, nclients| {
-                    let redo_work = Arc::new(Request::medium_input());
-                    b.iter_custom(|iters| bench_impl(Arc::clone(&redo_work), iters, *nclients));
-                },
-            );
-        }
+    macro_rules! bench_group {
+        ($name:expr, $redo_work:expr) => {{
+            let name: &str = $name;
+            let nclients = [1, 2, 4, 8, 16, 32, 64, 128];
+            for nclients in nclients {
+                let mut group = c.benchmark_group(name);
+                group.bench_with_input(
+                    BenchmarkId::from_parameter(nclients),
+                    &nclients,
+                    |b, nclients| {
+                        b.iter_custom(|iters| bench_impl($redo_work, iters, *nclients));
+                    },
+                );
+            }
+        }};
     }
+    //
+    // benchmark the protocol implementation
+    //
+    let pg_version = 14;
+    bench_group!(
+        "ping",
+        Arc::new(move |mgr: Arc<PostgresRedoManager>| async move {
+            let _: () = mgr.ping(pg_version).await.unwrap();
+        })
+    );
+    //
+    // benchmarks with actual record redo
+    //
+    let make_redo_work = |req: &'static Request| {
+        Arc::new(move |mgr: Arc<PostgresRedoManager>| async move {
+            let page = req.execute(&mgr).await.unwrap();
+            assert_eq!(page.remaining(), 8192);
+        })
+    };
+    bench_group!("short", {
+        static REQUEST: Lazy<Request> = Lazy::new(Request::short_input);
+        make_redo_work(&REQUEST)
+    });
+    bench_group!("medium", {
+        static REQUEST: Lazy<Request> = Lazy::new(Request::medium_input);
+        make_redo_work(&REQUEST)
+    });
 }
 criterion::criterion_group!(benches, bench);
 criterion::criterion_main!(benches);
 
 // Returns the sum of each client's wall-clock time spent executing their share of the n_redos.
-fn bench_impl(redo_work: Arc<Request>, n_redos: u64, nclients: u64) -> Duration {
+fn bench_impl<F, Fut>(redo_work: Arc<F>, n_redos: u64, nclients: u64) -> Duration
+where
+    F: Fn(Arc<PostgresRedoManager>) -> Fut + Send + Sync + 'static,
+    Fut: Future<Output = ()> + Send + 'static,
+{
     let repo_dir = camino_tempfile::tempdir_in(env!("CARGO_TARGET_TMPDIR")).unwrap();
 
     let conf = PageServerConf::dummy_conf(repo_dir.path().to_path_buf());
@@ -135,17 +164,20 @@ fn bench_impl(redo_work: Arc<Request>, n_redos: u64, nclients: u64) -> Duration
     })
 }
 
-async fn client(
+async fn client<F, Fut>(
     mgr: Arc<PostgresRedoManager>,
     start: Arc<Barrier>,
-    redo_work: Arc<Request>,
+    redo_work: Arc<F>,
     n_redos: u64,
-) -> Duration {
+) -> Duration
+where
+    F: Fn(Arc<PostgresRedoManager>) -> Fut + Send + Sync + 'static,
+    Fut: Future<Output = ()> + Send + 'static,
+{
     start.wait().await;
     let start = Instant::now();
     for _ in 0..n_redos {
-        let page = redo_work.execute(&mgr).await.unwrap();
-        assert_eq!(page.remaining(), 8192);
+        redo_work(Arc::clone(&mgr)).await;
         // The real pageserver will rarely if ever do 2 walredos in a row without
         // yielding to the executor.
         tokio::task::yield_now().await;

pageserver/client/src/mgmt_api.rs

@@ -432,7 +432,7 @@ impl Client {
             self.mgmt_api_endpoint
         );
 
-        self.request(Method::POST, &uri, req)
+        self.request(Method::PUT, &uri, req)
             .await?
             .json()
             .await

pageserver/compaction/Cargo.toml

@@ -9,41 +9,19 @@ default = []
 
 [dependencies]
 anyhow.workspace = true
-async-compression.workspace = true
 async-stream.workspace = true
-byteorder.workspace = true
-bytes.workspace = true
-chrono = { workspace = true, features = ["serde"] }
 clap = { workspace = true, features = ["string"] }
-const_format.workspace = true
-consumption_metrics.workspace = true
-crossbeam-utils.workspace = true
-either.workspace = true
-flate2.workspace = true
-fail.workspace = true
 futures.workspace = true
 git-version.workspace = true
-hex.workspace = true
-humantime.workspace = true
-humantime-serde.workspace = true
 itertools.workspace = true
 once_cell.workspace = true
 pageserver_api.workspace = true
 pin-project-lite.workspace = true
 rand.workspace = true
-smallvec = { workspace = true, features = ["write"] }
 svg_fmt.workspace = true
-sync_wrapper.workspace = true
-thiserror.workspace = true
 tokio = { workspace = true, features = ["process", "sync", "fs", "rt", "io-util", "time"] }
-tokio-io-timeout.workspace = true
-tokio-util.workspace = true
 tracing.workspace = true
-tracing-error.workspace = true
 tracing-subscriber.workspace = true
-url.workspace = true
-walkdir.workspace = true
-metrics.workspace = true
 utils.workspace = true
 workspace_hack.workspace = true
 

pageserver/ctl/Cargo.toml

@@ -8,7 +8,6 @@ license.workspace = true
 
 [dependencies]
 anyhow.workspace = true
-bytes.workspace = true
 camino.workspace = true
 clap = { workspace = true, features = ["string"] }
 git-version.workspace = true
@@ -24,5 +23,4 @@ toml_edit.workspace = true
 utils.workspace = true
 svg_fmt.workspace = true
 workspace_hack.workspace = true
-serde.workspace = true
 serde_json.workspace = true

pageserver/ctl/src/layer_map_analyzer.rs

@@ -79,16 +79,24 @@ pub(crate) fn parse_filename(name: &str) -> Option<LayerFile> {
         return None;
     }
     let keys: Vec<&str> = split[0].split('-').collect();
-    let mut lsns: Vec<&str> = split[1].split('-').collect();
-    let is_delta = if lsns.len() == 1 {
-        lsns.push(lsns[0]);
+    let lsn_and_opt_generation: Vec<&str> = split[1].split('v').collect();
+    let lsns: Vec<&str> = lsn_and_opt_generation[0].split('-').collect();
+    let the_lsns: [&str; 2];
+
+    /*
+     * Generations add a -vX-XXXXXX postfix, which causes issues when we try to
+     * parse 'vX' as an LSN.
+     */
+    let is_delta = if lsns.len() == 1 || lsns[1].is_empty() {
+        the_lsns = [lsns[0], lsns[0]];
         false
     } else {
+        the_lsns = [lsns[0], lsns[1]];
         true
     };
 
     let key_range = Key::from_hex(keys[0]).unwrap()..Key::from_hex(keys[1]).unwrap();
-    let lsn_range = Lsn::from_hex(lsns[0]).unwrap()..Lsn::from_hex(lsns[1]).unwrap();
+    let lsn_range = Lsn::from_hex(the_lsns[0]).unwrap()..Lsn::from_hex(the_lsns[1]).unwrap();
     let holes = Vec::new();
     Some(LayerFile {
         key_range,

pageserver/src/basebackup.rs

@@ -30,9 +30,8 @@ use pageserver_api::reltag::{RelTag, SlruKind};
 
 use postgres_ffi::dispatch_pgversion;
 use postgres_ffi::pg_constants::{DEFAULTTABLESPACE_OID, GLOBALTABLESPACE_OID};
-use postgres_ffi::pg_constants::{PGDATA_SPECIAL_FILES, PGDATA_SUBDIRS, PG_HBA};
+use postgres_ffi::pg_constants::{PGDATA_SPECIAL_FILES, PG_HBA};
 use postgres_ffi::relfile_utils::{INIT_FORKNUM, MAIN_FORKNUM};
-use postgres_ffi::TransactionId;
 use postgres_ffi::XLogFileName;
 use postgres_ffi::PG_TLI;
 use postgres_ffi::{BLCKSZ, RELSEG_SIZE, WAL_SEGMENT_SIZE};
@@ -255,8 +254,11 @@ where
 
         let lazy_slru_download = self.timeline.get_lazy_slru_download() && !self.full_backup;
 
+        let pgversion = self.timeline.pg_version;
+        let subdirs = dispatch_pgversion!(pgversion, &pgv::bindings::PGDATA_SUBDIRS[..]);
+
         // Create pgdata subdirs structure
-        for dir in PGDATA_SUBDIRS.iter() {
+        for dir in subdirs.iter() {
             let header = new_tar_header_dir(dir)?;
             self.ar
                 .append(&header, &mut io::empty())
@@ -606,7 +608,7 @@ where
     //
     // Extract twophase state files
     //
-    async fn add_twophase_file(&mut self, xid: TransactionId) -> Result<(), BasebackupError> {
+    async fn add_twophase_file(&mut self, xid: u64) -> Result<(), BasebackupError> {
         let img = self
             .timeline
             .get_twophase_file(xid, self.lsn, self.ctx)
@@ -617,7 +619,11 @@ where
         buf.extend_from_slice(&img[..]);
         let crc = crc32c::crc32c(&img[..]);
         buf.put_u32_le(crc);
-        let path = format!("pg_twophase/{:>08X}", xid);
+        let path = if self.timeline.pg_version < 17 {
+            format!("pg_twophase/{:>08X}", xid)
+        } else {
+            format!("pg_twophase/{:>016X}", xid)
+        };
         let header = new_tar_header(&path, buf.len() as u64)?;
         self.ar
             .append(&header, &buf[..])

pageserver/src/config.rs

@@ -13,7 +13,6 @@ use pageserver_api::{
 use remote_storage::{RemotePath, RemoteStorageConfig};
 use std::env;
 use storage_broker::Uri;
-use utils::crashsafe::path_with_suffix_extension;
 use utils::logging::SecretString;
 
 use once_cell::sync::OnceCell;
@@ -33,7 +32,7 @@ use crate::tenant::storage_layer::inmemory_layer::IndexEntry;
 use crate::tenant::{TENANTS_SEGMENT_NAME, TIMELINES_SEGMENT_NAME};
 use crate::virtual_file;
 use crate::virtual_file::io_engine;
-use crate::{TENANT_HEATMAP_BASENAME, TENANT_LOCATION_CONFIG_NAME, TIMELINE_DELETE_MARK_SUFFIX};
+use crate::{TENANT_HEATMAP_BASENAME, TENANT_LOCATION_CONFIG_NAME};
 
 /// Global state of pageserver.
 ///
@@ -257,17 +256,6 @@ impl PageServerConf {
             .join(timeline_id.to_string())
     }
 
-    pub(crate) fn timeline_delete_mark_file_path(
-        &self,
-        tenant_shard_id: TenantShardId,
-        timeline_id: TimelineId,
-    ) -> Utf8PathBuf {
-        path_with_suffix_extension(
-            self.timeline_path(&tenant_shard_id, &timeline_id),
-            TIMELINE_DELETE_MARK_SUFFIX,
-        )
-    }
-
     /// Turns storage remote path of a file into its local path.
     pub fn local_path(&self, remote_path: &RemotePath) -> Utf8PathBuf {
         remote_path.with_base(&self.workdir)
@@ -281,7 +269,7 @@ impl PageServerConf {
 
         #[allow(clippy::manual_range_patterns)]
         match pg_version {
-            14 | 15 | 16 => Ok(path.join(format!("v{pg_version}"))),
+            14 | 15 | 16 | 17 => Ok(path.join(format!("v{pg_version}"))),
             _ => bail!("Unsupported postgres version: {}", pg_version),
         }
     }
@@ -336,7 +324,6 @@ impl PageServerConf {
             max_vectored_read_bytes,
             image_compression,
             ephemeral_bytes_per_memory_kb,
-            compact_level0_phase1_value_access: _,
             l0_flush,
             virtual_file_direct_io,
             concurrent_tenant_warmup,
@@ -491,11 +478,6 @@ pub struct ConfigurableSemaphore {
 }
 
 impl ConfigurableSemaphore {
-    pub const DEFAULT_INITIAL: NonZeroUsize = match NonZeroUsize::new(1) {
-        Some(x) => x,
-        None => panic!("const unwrap is not yet stable"),
-    };
-
     /// Initializse using a non-zero amount of permits.
     ///
     /// Require a non-zero initial permits, because using permits == 0 is a crude way to disable a
@@ -516,12 +498,6 @@ impl ConfigurableSemaphore {
     }
 }
 
-impl Default for ConfigurableSemaphore {
-    fn default() -> Self {
-        Self::new(Self::DEFAULT_INITIAL)
-    }
-}
-
 impl PartialEq for ConfigurableSemaphore {
     fn eq(&self, other: &Self) -> bool {
         // the number of permits can be increased at runtime, so we cannot really fulfill the
@@ -558,16 +534,6 @@ mod tests {
             .expect("parse_and_validate");
     }
 
-    #[test]
-    fn test_compactl0_phase1_access_mode_is_ignored_silently() {
-        let input = indoc::indoc! {r#"
-            [compact_level0_phase1_value_access]
-            mode = "streaming-kmerge"
-            validate = "key-lsn-value"
-        "#};
-        toml_edit::de::from_str::<pageserver_api::config::ConfigToml>(input).unwrap();
-    }
-
     /// If there's a typo in the pageserver config, we'd rather catch that typo
     /// and fail pageserver startup than silently ignoring the typo, leaving whoever
     /// made it in the believe that their config change is effective.

pageserver/src/consumption_metrics.rs

@@ -178,7 +178,7 @@ async fn collect_metrics(
                 )
                 .await;
                 if let Err(e) = res {
-                    tracing::error!("failed to upload to S3: {e:#}");
+                    tracing::error!("failed to upload to remote storage: {e:#}");
                 }
             }
         };

pageserver/src/http/routes.rs

@@ -2955,7 +2955,7 @@ pub fn make_router(
             "/v1/tenant/:tenant_shard_id/timeline/:timeline_id/preserve_initdb_archive",
             |r| api_handler(r, timeline_preserve_initdb_handler),
         )
-        .post(
+        .put(
             "/v1/tenant/:tenant_shard_id/timeline/:timeline_id/archival_config",
             |r| api_handler(r, timeline_archival_config_handler),
         )

pageserver/src/import_datadir.rs

@@ -506,10 +506,6 @@ async fn import_file(
         return Ok(None);
     }
 
-    if file_name == "pg_internal.init" {
-        return Ok(None);
-    }
-
     if file_path.starts_with("global") {
         let spcnode = postgres_ffi::pg_constants::GLOBALTABLESPACE_OID;
         let dbnode = 0;
@@ -584,9 +580,11 @@ async fn import_file(
         import_slru(modification, slru, file_path, reader, len, ctx).await?;
         debug!("imported multixact members slru");
     } else if file_path.starts_with("pg_twophase") {
-        let xid = u32::from_str_radix(file_name.as_ref(), 16)?;
-
         let bytes = read_all_bytes(reader).await?;
+
+        // In PostgreSQL v17, this is a 64-bit FullTransactionid. In previous versions,
+        // it's a 32-bit TransactionId, which fits in u64 anyway.
+        let xid = u64::from_str_radix(file_name.as_ref(), 16)?;
         modification
             .put_twophase_file(xid, Bytes::copy_from_slice(&bytes[..]), ctx)
             .await?;