Test

.config/hakari.toml

@@ -23,30 +23,10 @@ platforms = [
 ]
 
 [final-excludes]
-workspace-members = [
-    # vm_monitor benefits from the same Cargo.lock as the rest of our artifacts, but
-    # it is built primarly in separate repo neondatabase/autoscaling and thus is excluded
-    # from depending on workspace-hack because most of the dependencies are not used.
-    "vm_monitor",
-    # All of these exist in libs and are not usually built independently.
-    # Putting workspace hack there adds a bottleneck for cargo builds.
-    "compute_api",
-    "consumption_metrics",
-    "desim",
-    "metrics",
-    "pageserver_api",
-    "postgres_backend",
-    "postgres_connection",
-    "postgres_ffi",
-    "pq_proto",
-    "remote_storage",
-    "safekeeper_api",
-    "tenant_size_model",
-    "tracing-utils",
-    "utils",
-    "wal_craft",
-    "walproposer",
-]
+# vm_monitor benefits from the same Cargo.lock as the rest of our artifacts, but
+# it is built primarly in separate repo neondatabase/autoscaling and thus is excluded
+# from depending on workspace-hack because most of the dependencies are not used.
+workspace-members = ["vm_monitor"]
 
 # Write out exact versions rather than a semver range. (Defaults to false.)
 # exact-versions = true

.config/nextest.toml

@@ -1,2 +1,2 @@
 [profile.default]
-slow-timeout = { period = "60s", terminate-after = 3 }
+slow-timeout = { period = "20s", terminate-after = 3 }

.devcontainer/Dockerfile.devcontainer

@@ -1 +0,0 @@
-FROM neondatabase/build-tools:pinned

.devcontainer/devcontainer.json

@@ -1,23 +0,0 @@
-// https://containers.dev/implementors/json_reference/
-{
-  "name": "Neon",
-  "build": {
-    "context": "..",
-    "dockerfile": "Dockerfile.devcontainer"
-  },
-
-  "postCreateCommand": {
-    "build neon": "BUILD_TYPE=debug CARGO_BUILD_FLAGS='--features=testing' mold -run make -s -j`nproc`",
-    "install python deps": "./scripts/pysync"
-  },
-
-  "customizations": {
-    "vscode": {
-      "extensions": [
-        "charliermarsh.ruff",
-        "github.vscode-github-actions",
-        "rust-lang.rust-analyzer"
-      ]
-    }
-  }
-}

.dockerignore

@@ -8,7 +8,6 @@
 !scripts/combine_control_files.py
 !scripts/ninstall.sh
 !vm-cgconfig.conf
-!docker-compose/run-tests.sh
 
 # Directories
 !.cargo/
@@ -18,13 +17,11 @@
 !libs/
 !neon_local/
 !pageserver/
-!patches/
 !pgxn/
 !proxy/
-!storage_scrubber/
+!s3_scrubber/
 !safekeeper/
 !storage_broker/
-!storage_controller/
 !trace/
 !vendor/postgres-*/
 !workspace_hack/

.gitattributes

@@ -1,2 +0,0 @@
-# allows for nicer hunk headers with git show
-*.rs diff=rust

.github/ISSUE_TEMPLATE/config.yml

@@ -1,6 +0,0 @@
-
-blank_issues_enabled: true
-contact_links:
-  - name: Feature request
-    url: https://console.neon.tech/app/projects?modal=feedback
-    about: For feature requests in the Neon product, please submit via the feedback form on `https://console.neon.tech`

.github/actionlint.yml

@@ -1,22 +1,14 @@
 self-hosted-runner:
   labels:
     - arm64
+    - dev
+    - gen3
     - large
-    - large-arm64
+    # Remove `macos-14` from the list after https://github.com/rhysd/actionlint/pull/392 is merged.
+    - macos-14
     - small
-    - small-arm64
     - us-east-2
 config-variables:
-  - AZURE_DEV_CLIENT_ID
-  - AZURE_DEV_REGISTRY_NAME
-  - AZURE_DEV_SUBSCRIPTION_ID
-  - AZURE_PROD_CLIENT_ID
-  - AZURE_PROD_REGISTRY_NAME
-  - AZURE_PROD_SUBSCRIPTION_ID
-  - AZURE_TENANT_ID
-  - BENCHMARK_PROJECT_ID_PUB
-  - BENCHMARK_PROJECT_ID_SUB
   - REMOTE_STORAGE_AZURE_CONTAINER
   - REMOTE_STORAGE_AZURE_REGION
   - SLACK_UPCOMING_RELEASE_CHANNEL_ID
-  - DEV_AWS_OIDC_ROLE_ARN

.github/actions/allure-report-generate/action.yml

@@ -150,7 +150,7 @@ runs:
 
         # Use aws s3 cp (instead of aws s3 sync) to keep files from previous runs to make old URLs work,
         # and to keep files on the host to upload them to the database
-        time s5cmd --log error cp "${WORKDIR}/report/*" "s3://${BUCKET}/${REPORT_PREFIX}/${GITHUB_RUN_ID}/"
+        time aws s3 cp --recursive --only-show-errors "${WORKDIR}/report" "s3://${BUCKET}/${REPORT_PREFIX}/${GITHUB_RUN_ID}"
 
         # Generate redirect
         cat <<EOF > ${WORKDIR}/index.html
@@ -183,7 +183,7 @@ runs:
       uses: actions/cache@v4
       with:
         path: ~/.cache/pypoetry/virtualenvs
-        key: v2-${{ runner.os }}-${{ runner.arch }}-python-deps-${{ hashFiles('poetry.lock') }}
+        key: v2-${{ runner.os }}-python-deps-${{ hashFiles('poetry.lock') }}
 
     - name: Store Allure test stat in the DB (new)
       if: ${{ !cancelled() && inputs.store-test-results-into-db == 'true' }}

.github/actions/download/action.yml

@@ -26,7 +26,7 @@ runs:
         TARGET: ${{ inputs.path }}
         ARCHIVE: /tmp/downloads/${{ inputs.name }}.tar.zst
         SKIP_IF_DOES_NOT_EXIST: ${{ inputs.skip-if-does-not-exist }}
-        PREFIX: artifacts/${{ inputs.prefix || format('{0}/{1}/{2}', github.event.pull_request.head.sha || github.sha, github.run_id, github.run_attempt) }}
+        PREFIX: artifacts/${{ inputs.prefix || format('{0}/{1}', github.run_id, github.run_attempt) }}
       run: |
         BUCKET=neon-github-public-dev
         FILENAME=$(basename $ARCHIVE)

.github/actions/neon-branch-create/action.yml

@@ -3,14 +3,14 @@ description: 'Create Branch using API'
 
 inputs:
   api_key:
-    description: 'Neon API key'
+    desctiption: 'Neon API key'
     required: true
   project_id:
-    description: 'ID of the Project to create Branch in'
+    desctiption: 'ID of the Project to create Branch in'
     required: true
   api_host:
-    description: 'Neon API host'
-    default: console-stage.neon.build
+    desctiption: 'Neon API host'
+    default: console.stage.neon.tech
 outputs:
   dsn:
     description: 'Created Branch DSN (for main database)'

.github/actions/neon-branch-delete/action.yml

@@ -3,17 +3,17 @@ description: 'Delete Branch using API'
 
 inputs:
   api_key:
-    description: 'Neon API key'
+    desctiption: 'Neon API key'
     required: true
   project_id:
-    description: 'ID of the Project which should be deleted'
+    desctiption: 'ID of the Project which should be deleted'
     required: true
   branch_id:
-    description: 'ID of the branch to delete'
+    desctiption: 'ID of the branch to delete'
     required: true
   api_host:
-    description: 'Neon API host'
-    default: console-stage.neon.build
+    desctiption: 'Neon API host'
+    default: console.stage.neon.tech
 
 runs:
   using: "composite"

.github/actions/neon-project-create/action.yml

@@ -3,19 +3,22 @@ description: 'Create Neon Project using API'
 
 inputs:
   api_key:
-    description: 'Neon API key'
+    desctiption: 'Neon API key'
     required: true
   region_id:
-    description: 'Region ID, if not set the project will be created in the default region'
+    desctiption: 'Region ID, if not set the project will be created in the default region'
     default: aws-us-east-2
   postgres_version:
-    description: 'Postgres version; default is 16'
-    default: '16'
+    desctiption: 'Postgres version; default is 15'
+    default: 15
   api_host:
-    description: 'Neon API host'
-    default: console-stage.neon.build
+    desctiption: 'Neon API host'
+    default: console.stage.neon.tech
+  provisioner:
+    desctiption: 'k8s-pod or k8s-neonvm'
+    default: 'k8s-pod'
   compute_units:
-    description: '[Min, Max] compute units'
+    desctiption: '[Min, Max] compute units; Min and Max are used for k8s-neonvm with autoscaling, for k8s-pod values Min and Max should be equal'
     default: '[1, 1]'
 
 outputs:
@@ -34,6 +37,10 @@ runs:
       # A shell without `set -x` to not to expose password/dsn in logs
       shell: bash -euo pipefail {0}
       run: |
+        if [ "${PROVISIONER}" == "k8s-pod" ] && [ "${MIN_CU}" != "${MAX_CU}" ]; then
+          echo >&2 "For k8s-pod provisioner MIN_CU should be equal to MAX_CU"
+        fi
+
         project=$(curl \
           "https://${API_HOST}/api/v2/projects" \
           --fail \
@@ -45,7 +52,7 @@ runs:
               \"name\": \"Created by actions/neon-project-create; GITHUB_RUN_ID=${GITHUB_RUN_ID}\",
               \"pg_version\": ${POSTGRES_VERSION},
               \"region_id\": \"${REGION_ID}\",
-              \"provisioner\": \"k8s-neonvm\",
+              \"provisioner\": \"${PROVISIONER}\",
               \"autoscaling_limit_min_cu\": ${MIN_CU},
               \"autoscaling_limit_max_cu\": ${MAX_CU},
               \"settings\": { }
@@ -68,5 +75,6 @@ runs:
         API_KEY: ${{ inputs.api_key }}
         REGION_ID: ${{ inputs.region_id }}
         POSTGRES_VERSION: ${{ inputs.postgres_version }}
+        PROVISIONER: ${{ inputs.provisioner }}
         MIN_CU: ${{ fromJSON(inputs.compute_units)[0] }}
         MAX_CU: ${{ fromJSON(inputs.compute_units)[1] }}

.github/actions/neon-project-delete/action.yml

@@ -3,14 +3,14 @@ description: 'Delete Neon Project using API'
 
 inputs:
   api_key:
-    description: 'Neon API key'
+    desctiption: 'Neon API key'
     required: true
   project_id:
-    description: 'ID of the Project to delete'
+    desctiption: 'ID of the Project to delete'
     required: true
   api_host:
-    description: 'Neon API host'
-    default: console-stage.neon.build
+    desctiption: 'Neon API host'
+    default: console.stage.neon.tech
 
 runs:
   using: "composite"

.github/actions/run-python-test-set/action.yml

@@ -43,7 +43,7 @@ inputs:
   pg_version:
     description: 'Postgres version to use for tests'
     required: false
-    default: 'v16'
+    default: 'v14'
   benchmark_durations:
     description: 'benchmark durations JSON'
     required: false
@@ -56,14 +56,14 @@ runs:
       if: inputs.build_type != 'remote'
       uses: ./.github/actions/download
       with:
-        name: neon-${{ runner.os }}-${{ runner.arch }}-${{ inputs.build_type }}-artifact
+        name: neon-${{ runner.os }}-${{ inputs.build_type }}-artifact
         path: /tmp/neon
 
     - name: Download Neon binaries for the previous release
       if: inputs.build_type != 'remote'
       uses: ./.github/actions/download
       with:
-        name: neon-${{ runner.os }}-${{ runner.arch }}-${{ inputs.build_type }}-artifact
+        name: neon-${{ runner.os }}-${{ inputs.build_type }}-artifact
         path: /tmp/neon-previous
         prefix: latest
 
@@ -71,7 +71,7 @@ runs:
       if: inputs.build_type != 'remote'
       uses: ./.github/actions/download
       with:
-        name: compatibility-snapshot-${{ runner.arch }}-${{ inputs.build_type }}-pg${{ inputs.pg_version }}
+        name: compatibility-snapshot-${{ inputs.build_type }}-pg${{ inputs.pg_version }}
         path: /tmp/compatibility_snapshot_pg${{ inputs.pg_version }}
         prefix: latest
         # The lack of compatibility snapshot (for example, for the new Postgres version)
@@ -83,12 +83,13 @@ runs:
       uses: actions/checkout@v4
       with:
         submodules: true
+        fetch-depth: 1
 
     - name: Cache poetry deps
       uses: actions/cache@v4
       with:
         path: ~/.cache/pypoetry/virtualenvs
-        key: v2-${{ runner.os }}-${{ runner.arch }}-python-deps-${{ hashFiles('poetry.lock') }}
+        key: v2-${{ runner.os }}-python-deps-${{ hashFiles('poetry.lock') }}
 
     - name: Install Python deps
       shell: bash -euxo pipefail {0}
@@ -113,8 +114,6 @@ runs:
         export PLATFORM=${PLATFORM:-github-actions-selfhosted}
         export POSTGRES_DISTRIB_DIR=${POSTGRES_DISTRIB_DIR:-/tmp/neon/pg_install}
         export DEFAULT_PG_VERSION=${PG_VERSION#v}
-        export LD_LIBRARY_PATH=${POSTGRES_DISTRIB_DIR}/v${DEFAULT_PG_VERSION}/lib
-        export BENCHMARK_CONNSTR=${BENCHMARK_CONNSTR:-}
 
         if [ "${BUILD_TYPE}" = "remote" ]; then
           export REMOTE_ENV=1
@@ -130,8 +129,8 @@ runs:
           exit 1
         fi
         if [[ "${{ inputs.run_in_parallel }}" == "true" ]]; then
-          # -n sets the number of parallel processes that pytest-xdist will run
-          EXTRA_PARAMS="-n12 $EXTRA_PARAMS"
+          # -n16 uses sixteen processes to run tests via pytest-xdist
+          EXTRA_PARAMS="-n16 $EXTRA_PARAMS"
 
           # --dist=loadgroup points tests marked with @pytest.mark.xdist_group
           # to the same worker to make @pytest.mark.order work with xdist
@@ -169,28 +168,23 @@ runs:
           EXTRA_PARAMS="--durations-path $TEST_OUTPUT/benchmark_durations.json $EXTRA_PARAMS"
         fi
 
-        if [[ $BUILD_TYPE == "debug" && $RUNNER_ARCH == 'X64' ]]; then
+        if [[ "${{ inputs.build_type }}" == "debug" ]]; then
           cov_prefix=(scripts/coverage "--profraw-prefix=$GITHUB_JOB" --dir=/tmp/coverage run)
+        elif [[ "${{ inputs.build_type }}" == "release" ]]; then
+          cov_prefix=()
         else
           cov_prefix=()
         fi
 
         # Wake up the cluster if we use remote neon instance
         if [ "${{ inputs.build_type }}" = "remote" ] && [ -n "${BENCHMARK_CONNSTR}" ]; then
-          QUERIES=("SELECT version()")
-          if [[ "${PLATFORM}" = "neon"* ]]; then
-            QUERIES+=("SHOW neon.tenant_id")
-            QUERIES+=("SHOW neon.timeline_id")
-          fi
-
-          for q in "${QUERIES[@]}"; do
-            ${POSTGRES_DISTRIB_DIR}/v${DEFAULT_PG_VERSION}/bin/psql ${BENCHMARK_CONNSTR} -c "${q}"
-          done
+          ${POSTGRES_DISTRIB_DIR}/v${DEFAULT_PG_VERSION}/bin/psql ${BENCHMARK_CONNSTR} -c "SELECT version();"
         fi
 
         # Run the tests.
         #
-        # --alluredir saves test results in Allure format (in a specified directory)
+        # The junit.xml file allows CI tools to display more fine-grained test information
+        # in its "Tests" tab in the results page.
         # --verbose prints name of each test (helpful when there are
         # multiple tests in one file)
         # -rA prints summary in the end
@@ -199,6 +193,7 @@ runs:
         #
         mkdir -p $TEST_OUTPUT/allure/results
         "${cov_prefix[@]}" ./scripts/pytest \
+          --junitxml=$TEST_OUTPUT/junit.xml \
           --alluredir=$TEST_OUTPUT/allure/results \
           --tb=short \
           --verbose \
@@ -211,13 +206,13 @@ runs:
         fi
 
     - name: Upload compatibility snapshot
-      # Note, that we use `github.base_ref` which is a target branch for a PR
-      if: github.event_name == 'pull_request' && github.base_ref == 'release'
+      if: github.ref_name == 'release'
       uses: ./.github/actions/upload
       with:
-        name: compatibility-snapshot-${{ runner.arch }}-${{ inputs.build_type }}-pg${{ inputs.pg_version }}
+        name: compatibility-snapshot-${{ inputs.build_type }}-pg${{ inputs.pg_version }}-${{ github.run_id }}
         # Directory is created by test_compatibility.py::test_create_snapshot, keep the path in sync with the test
         path: /tmp/test_output/compatibility_snapshot_pg${{ inputs.pg_version }}/
+        prefix: latest
 
     - name: Upload test results
       if: ${{ !cancelled() }}

.github/actions/set-docker-config-dir/action.yml

@@ -1,36 +0,0 @@
-name: "Set custom docker config directory"
-description: "Create a directory for docker config and set DOCKER_CONFIG"
-
-# Use custom DOCKER_CONFIG directory to avoid conflicts with default settings
-runs:
-  using: "composite"
-  steps:
-  - name: Show warning on GitHub-hosted runners
-    if: runner.environment == 'github-hosted'
-    shell: bash -euo pipefail {0}
-    run: |
-      # Using the following environment variables to find a path to the workflow file
-      # ${GITHUB_WORKFLOW_REF} - octocat/hello-world/.github/workflows/my-workflow.yml@refs/heads/my_branch
-      # ${GITHUB_REPOSITORY}   - octocat/hello-world
-      # ${GITHUB_REF}          - refs/heads/my_branch
-      # From https://docs.github.com/en/actions/writing-workflows/choosing-what-your-workflow-does/variables
-
-      filename_with_ref=${GITHUB_WORKFLOW_REF#"$GITHUB_REPOSITORY/"}
-      filename=${filename_with_ref%"@$GITHUB_REF"}
-
-      # https://docs.github.com/en/actions/writing-workflows/choosing-what-your-workflow-does/workflow-commands-for-github-actions#setting-a-warning-message
-      title='Unnecessary usage of `.github/actions/set-docker-config-dir`'
-      message='No need to use `.github/actions/set-docker-config-dir` action on GitHub-hosted runners'
-      echo "::warning file=${filename},title=${title}::${message}"
-
-  - uses: pyTooling/Actions/with-post-step@74afc5a42a17a046c90c68cb5cfa627e5c6c5b6b # v1.0.7
-    env:
-      DOCKER_CONFIG: .docker-custom-${{ github.run_id }}-${{ github.run_attempt }}
-    with:
-      main: |
-        mkdir -p "${DOCKER_CONFIG}"
-        echo DOCKER_CONFIG=${DOCKER_CONFIG} | tee -a $GITHUB_ENV
-      post: |
-        if [ -d "${DOCKER_CONFIG}" ]; then
-          rm -r "${DOCKER_CONFIG}"
-        fi

.github/actions/upload/action.yml

@@ -8,7 +8,7 @@ inputs:
     description: "A directory or file to upload"
     required: true
   prefix:
-    description: "S3 prefix. Default is '${GITHUB_SHA}/${GITHUB_RUN_ID}/${GITHUB_RUN_ATTEMPT}'"
+    description: "S3 prefix. Default is '${GITHUB_RUN_ID}/${GITHUB_RUN_ATTEMPT}'"
     required: false
 
 runs:
@@ -45,7 +45,7 @@ runs:
       env:
         SOURCE: ${{ inputs.path }}
         ARCHIVE: /tmp/uploads/${{ inputs.name }}.tar.zst
-        PREFIX: artifacts/${{ inputs.prefix || format('{0}/{1}/{2}', github.event.pull_request.head.sha || github.sha, github.run_id , github.run_attempt) }}
+        PREFIX: artifacts/${{ inputs.prefix || format('{0}/{1}', github.run_id, github.run_attempt) }}
       run: |
         BUCKET=neon-github-public-dev
         FILENAME=$(basename $ARCHIVE)

.github/workflows/_benchmarking_preparation.yml

@@ -1,154 +0,0 @@
-name: Prepare benchmarking databases by restoring dumps
-
-on:
-  workflow_call:
-    # no inputs needed
-    
-defaults:
-  run:
-    shell: bash -euxo pipefail {0}
-
-jobs:
-  setup-databases:
-    strategy:
-      fail-fast: false
-      matrix:
-        platform: [ aws-rds-postgres, aws-aurora-serverless-v2-postgres, neon ] 
-        database: [ clickbench, tpch, userexample ]
-  
-    env:
-      LD_LIBRARY_PATH: /tmp/neon/pg_install/v16/lib
-      PLATFORM: ${{ matrix.platform }}
-      PG_BINARIES: /tmp/neon/pg_install/v16/bin
-
-    runs-on: [ self-hosted, us-east-2, x64 ]
-    container:
-      image: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/build-tools:pinned
-      options: --init
-
-    steps:
-    - name: Set up Connection String
-      id: set-up-prep-connstr
-      run: |
-        case "${PLATFORM}" in
-          neon)
-            CONNSTR=${{ secrets.BENCHMARK_CAPTEST_CONNSTR }} 
-            ;;
-          aws-rds-postgres)
-            CONNSTR=${{ secrets.BENCHMARK_RDS_POSTGRES_CONNSTR }} 
-            ;;
-          aws-aurora-serverless-v2-postgres)
-            CONNSTR=${{ secrets.BENCHMARK_RDS_AURORA_CONNSTR }} 
-            ;;
-          *)
-            echo >&2 "Unknown PLATFORM=${PLATFORM}"
-            exit 1
-            ;;
-        esac
-
-        echo "connstr=${CONNSTR}" >> $GITHUB_OUTPUT  
-
-    - uses: actions/checkout@v4
-
-    - name: Download Neon artifact
-      uses: ./.github/actions/download
-      with:
-        name: neon-${{ runner.os }}-${{ runner.arch }}-release-artifact
-        path: /tmp/neon/
-        prefix: latest
-
-    # we create a table that has one row for each database that we want to restore with the status whether the restore is done    
-    - name: Create benchmark_restore_status table if it does not exist
-      env:
-        BENCHMARK_CONNSTR: ${{ steps.set-up-prep-connstr.outputs.connstr }}
-        DATABASE_NAME: ${{ matrix.database }}
-      # to avoid a race condition of multiple jobs trying to create the table at the same time, 
-      # we use an advisory lock
-      run: |
-        ${PG_BINARIES}/psql "${{ env.BENCHMARK_CONNSTR }}" -c "
-        SELECT pg_advisory_lock(4711);  
-        CREATE TABLE IF NOT EXISTS benchmark_restore_status (
-        databasename text primary key,
-        restore_done boolean
-        );
-        SELECT pg_advisory_unlock(4711);
-        "
-    
-    - name: Check if restore is already done
-      id: check-restore-done
-      env:
-        BENCHMARK_CONNSTR: ${{ steps.set-up-prep-connstr.outputs.connstr }}
-        DATABASE_NAME: ${{ matrix.database }}
-      run: |
-        skip=false
-        if ${PG_BINARIES}/psql "${{ env.BENCHMARK_CONNSTR }}" -tAc "SELECT 1 FROM benchmark_restore_status WHERE databasename='${{ env.DATABASE_NAME }}' AND restore_done=true;" | grep -q 1; then
-          echo "Restore already done for database ${{ env.DATABASE_NAME }} on platform ${{ env.PLATFORM }}. Skipping this database."
-          skip=true
-        fi
-        echo "skip=${skip}" | tee -a $GITHUB_OUTPUT
-
-    - name: Check and create database if it does not exist
-      if: steps.check-restore-done.outputs.skip != 'true'
-      env:
-        BENCHMARK_CONNSTR: ${{ steps.set-up-prep-connstr.outputs.connstr }}
-        DATABASE_NAME: ${{ matrix.database }}
-      run: |
-        DB_EXISTS=$(${PG_BINARIES}/psql "${{ env.BENCHMARK_CONNSTR }}" -tAc "SELECT 1 FROM pg_database WHERE datname='${{ env.DATABASE_NAME }}'")
-        if [ "$DB_EXISTS" != "1" ]; then
-          echo "Database ${{ env.DATABASE_NAME }} does not exist. Creating it..."
-          ${PG_BINARIES}/psql "${{ env.BENCHMARK_CONNSTR }}" -c "CREATE DATABASE \"${{ env.DATABASE_NAME }}\";"
-        else
-          echo "Database ${{ env.DATABASE_NAME }} already exists."
-        fi
-
-    - name: Download dump from S3 to /tmp/dumps
-      if: steps.check-restore-done.outputs.skip != 'true'
-      env:
-        DATABASE_NAME: ${{ matrix.database }}
-      run: |
-        mkdir -p /tmp/dumps
-        aws s3 cp s3://neon-github-dev/performance/pgdumps/$DATABASE_NAME/$DATABASE_NAME.pg_dump /tmp/dumps/ 
-
-    - name: Replace database name in connection string
-      if: steps.check-restore-done.outputs.skip != 'true'
-      id: replace-dbname
-      env:
-        DATABASE_NAME: ${{ matrix.database }}
-        BENCHMARK_CONNSTR: ${{ steps.set-up-prep-connstr.outputs.connstr }}
-      run: |
-        # Extract the part before the database name
-        base_connstr="${BENCHMARK_CONNSTR%/*}"
-        # Extract the query parameters (if any) after the database name
-        query_params="${BENCHMARK_CONNSTR#*\?}"
-        # Reconstruct the new connection string
-        if [ "$query_params" != "$BENCHMARK_CONNSTR" ]; then
-          new_connstr="${base_connstr}/${DATABASE_NAME}?${query_params}"
-        else
-          new_connstr="${base_connstr}/${DATABASE_NAME}"
-        fi
-        echo "database_connstr=${new_connstr}" >> $GITHUB_OUTPUT  
-
-    - name: Restore dump
-      if: steps.check-restore-done.outputs.skip != 'true'
-      env:
-        DATABASE_NAME: ${{ matrix.database }}
-        DATABASE_CONNSTR: ${{ steps.replace-dbname.outputs.database_connstr }}
-        # the following works only with larger computes: 
-        # PGOPTIONS: "-c maintenance_work_mem=8388608 -c max_parallel_maintenance_workers=7"
-        # we add the || true because:
-        # the dumps were created with Neon and contain neon extensions that are not 
-        # available in RDS, so we will always report an error, but we can ignore it
-      run: |
-        ${PG_BINARIES}/pg_restore --clean --if-exists --no-owner --jobs=4 \
-        -d "${DATABASE_CONNSTR}" /tmp/dumps/${DATABASE_NAME}.pg_dump || true
-
-    - name: Update benchmark_restore_status table
-      if: steps.check-restore-done.outputs.skip != 'true'
-      env:
-        BENCHMARK_CONNSTR: ${{ steps.set-up-prep-connstr.outputs.connstr }}
-        DATABASE_NAME: ${{ matrix.database }}
-      run: |
-        ${PG_BINARIES}/psql "${{ env.BENCHMARK_CONNSTR }}" -c "
-        INSERT INTO benchmark_restore_status (databasename, restore_done) VALUES ('${{ env.DATABASE_NAME }}', true)
-        ON CONFLICT (databasename) DO UPDATE SET restore_done = true;
-        "

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

@@ -1,303 +0,0 @@
-name: Build and Test Locally
-
-on:
-  workflow_call:
-    inputs:
-      arch:
-        description: 'x64 or arm64'
-        required: true
-        type: string
-      build-tag:
-        description: 'build tag'
-        required: true
-        type: string
-      build-tools-image:
-        description: 'build-tools image'
-        required: true
-        type: string
-      build-type:
-        description: 'debug or release'
-        required: true
-        type: string
-      pg-versions:
-        description: 'a json array of postgres versions to run regression tests on'
-        required: true
-        type: string
-
-defaults:
-  run:
-    shell: bash -euxo pipefail {0}
-
-env:
-  RUST_BACKTRACE: 1
-  COPT: '-Werror'
-  AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_DEV }}
-  AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_KEY_DEV }}
-
-jobs:
-  build-neon:
-    runs-on: ${{ fromJson(format('["self-hosted", "{0}"]', inputs.arch == 'arm64' && 'large-arm64' || 'large')) }}
-    container:
-      image: ${{ inputs.build-tools-image }}
-      credentials:
-        username: ${{ secrets.NEON_DOCKERHUB_USERNAME }}
-        password: ${{ secrets.NEON_DOCKERHUB_PASSWORD }}
-      # Raise locked memory limit for tokio-epoll-uring.
-      # On 5.10 LTS kernels < 5.10.162 (and generally mainline kernels < 5.12),
-      # io_uring will account the memory of the CQ and SQ as locked.
-      # More details: https://github.com/neondatabase/neon/issues/6373#issuecomment-1905814391
-      options: --init --shm-size=512mb --ulimit memlock=67108864:67108864
-    env:
-      BUILD_TYPE: ${{ inputs.build-type }}
-      GIT_VERSION: ${{ github.event.pull_request.head.sha || github.sha }}
-      BUILD_TAG: ${{ inputs.build-tag }}
-
-    steps:
-      - name: Fix git ownership
-        run: |
-          # Workaround for `fatal: detected dubious ownership in repository at ...`
-          #
-          # Use both ${{ github.workspace }} and ${GITHUB_WORKSPACE} because they're different on host and in containers
-          #   Ref https://github.com/actions/checkout/issues/785
-          #
-          git config --global --add safe.directory ${{ github.workspace }}
-          git config --global --add safe.directory ${GITHUB_WORKSPACE}
-          for r in 14 15 16; do
-            git config --global --add safe.directory "${{ github.workspace }}/vendor/postgres-v$r"
-            git config --global --add safe.directory "${GITHUB_WORKSPACE}/vendor/postgres-v$r"
-          done
-
-      - uses: actions/checkout@v4
-        with:
-          submodules: true
-
-      - name: Set pg 14 revision for caching
-        id: pg_v14_rev
-        run: echo pg_rev=$(git rev-parse HEAD:vendor/postgres-v14) >> $GITHUB_OUTPUT
-
-      - name: Set pg 15 revision for caching
-        id: pg_v15_rev
-        run: echo pg_rev=$(git rev-parse HEAD:vendor/postgres-v15) >> $GITHUB_OUTPUT
-
-      - name: Set pg 16 revision for caching
-        id: pg_v16_rev
-        run: echo pg_rev=$(git rev-parse HEAD:vendor/postgres-v16) >> $GITHUB_OUTPUT
-
-      # Set some environment variables used by all the steps.
-      #
-      # CARGO_FLAGS is extra options to pass to "cargo build", "cargo test" etc.
-      #   It also includes --features, if any
-      #
-      # CARGO_FEATURES is passed to "cargo metadata". It is separate from CARGO_FLAGS,
-      #   because "cargo metadata" doesn't accept --release or --debug options
-      #
-      # We run tests with addtional features, that are turned off by default (e.g. in release builds), see
-      # corresponding Cargo.toml files for their descriptions.
-      - name: Set env variables
-        env:
-          ARCH: ${{ inputs.arch }}
-        run: |
-          CARGO_FEATURES="--features testing"
-          if [[ $BUILD_TYPE == "debug" && $ARCH == 'x64' ]]; then
-            cov_prefix="scripts/coverage --profraw-prefix=$GITHUB_JOB --dir=/tmp/coverage run"
-            CARGO_FLAGS="--locked"
-          elif [[ $BUILD_TYPE == "debug" ]]; then
-            cov_prefix=""
-            CARGO_FLAGS="--locked"
-          elif [[ $BUILD_TYPE == "release" ]]; then
-            cov_prefix=""
-            CARGO_FLAGS="--locked --release"
-          fi
-          {
-            echo "cov_prefix=${cov_prefix}"
-            echo "CARGO_FEATURES=${CARGO_FEATURES}"
-            echo "CARGO_FLAGS=${CARGO_FLAGS}"
-            echo "CARGO_HOME=${GITHUB_WORKSPACE}/.cargo"
-          } >> $GITHUB_ENV
-
-      - name: Cache postgres v14 build
-        id: cache_pg_14
-        uses: actions/cache@v4
-        with:
-          path: pg_install/v14
-          key: v1-${{ runner.os }}-${{ runner.arch }}-${{ inputs.build-type }}-pg-${{ steps.pg_v14_rev.outputs.pg_rev }}-${{ hashFiles('Makefile', 'Dockerfile.build-tools') }}
-
-      - name: Cache postgres v15 build
-        id: cache_pg_15
-        uses: actions/cache@v4
-        with:
-          path: pg_install/v15
-          key: v1-${{ runner.os }}-${{ runner.arch }}-${{ inputs.build-type }}-pg-${{ steps.pg_v15_rev.outputs.pg_rev }}-${{ hashFiles('Makefile', 'Dockerfile.build-tools') }}
-
-      - name: Cache postgres v16 build
-        id: cache_pg_16
-        uses: actions/cache@v4
-        with:
-          path: pg_install/v16
-          key: v1-${{ runner.os }}-${{ runner.arch }}-${{ inputs.build-type }}-pg-${{ steps.pg_v16_rev.outputs.pg_rev }}-${{ hashFiles('Makefile', 'Dockerfile.build-tools') }}
-
-      - name: Build postgres v14
-        if: steps.cache_pg_14.outputs.cache-hit != 'true'
-        run: mold -run make postgres-v14 -j$(nproc)
-
-      - name: Build postgres v15
-        if: steps.cache_pg_15.outputs.cache-hit != 'true'
-        run: mold -run make postgres-v15 -j$(nproc)
-
-      - name: Build postgres v16
-        if: steps.cache_pg_16.outputs.cache-hit != 'true'
-        run: mold -run make postgres-v16 -j$(nproc)
-
-      - name: Build neon extensions
-        run: mold -run make neon-pg-ext -j$(nproc)
-
-      - name: Build walproposer-lib
-        run: mold -run make walproposer-lib -j$(nproc)
-
-      - name: Run cargo build
-        run: |
-          PQ_LIB_DIR=$(pwd)/pg_install/v16/lib
-          export PQ_LIB_DIR
-          ${cov_prefix} mold -run cargo build $CARGO_FLAGS $CARGO_FEATURES --bins --tests
-
-      # Do install *before* running rust tests because they might recompile the
-      # binaries with different features/flags.
-      - name: Install rust binaries
-        env:
-          ARCH: ${{ inputs.arch }}
-        run: |
-          # Install target binaries
-          mkdir -p /tmp/neon/bin/
-          binaries=$(
-            ${cov_prefix} cargo metadata $CARGO_FEATURES --format-version=1 --no-deps |
-            jq -r '.packages[].targets[] | select(.kind | index("bin")) | .name'
-          )
-          for bin in $binaries; do
-            SRC=target/$BUILD_TYPE/$bin
-            DST=/tmp/neon/bin/$bin
-            cp "$SRC" "$DST"
-          done
-
-          # Install test executables and write list of all binaries (for code coverage)
-          if [[ $BUILD_TYPE == "debug" && $ARCH == 'x64' ]]; then
-            # Keep bloated coverage data files away from the rest of the artifact
-            mkdir -p /tmp/coverage/
-
-            mkdir -p /tmp/neon/test_bin/
-
-            test_exe_paths=$(
-              ${cov_prefix} cargo test $CARGO_FLAGS $CARGO_FEATURES --message-format=json --no-run |
-              jq -r '.executable | select(. != null)'
-            )
-            for bin in $test_exe_paths; do
-              SRC=$bin
-              DST=/tmp/neon/test_bin/$(basename $bin)
-
-              # We don't need debug symbols for code coverage, so strip them out to make
-              # the artifact smaller.
-              strip "$SRC" -o "$DST"
-              echo "$DST" >> /tmp/coverage/binaries.list
-            done
-
-            for bin in $binaries; do
-              echo "/tmp/neon/bin/$bin" >> /tmp/coverage/binaries.list
-            done
-          fi
-
-      - name: Run rust tests
-        env:
-          NEXTEST_RETRIES: 3
-        run: |
-          PQ_LIB_DIR=$(pwd)/pg_install/v16/lib
-          export PQ_LIB_DIR
-          LD_LIBRARY_PATH=$(pwd)/pg_install/v16/lib
-          export LD_LIBRARY_PATH
-
-          #nextest does not yet support running doctests
-          ${cov_prefix} cargo test --doc $CARGO_FLAGS $CARGO_FEATURES
-
-          # run all non-pageserver tests
-          ${cov_prefix} cargo nextest run $CARGO_FLAGS $CARGO_FEATURES -E '!package(pageserver)'
-
-          # run pageserver tests with different settings
-          for io_engine in std-fs tokio-epoll-uring ; do
-            for io_buffer_alignment in 0 1 512 ; do
-              NEON_PAGESERVER_UNIT_TEST_VIRTUAL_FILE_IOENGINE=$io_engine NEON_PAGESERVER_UNIT_TEST_IO_BUFFER_ALIGNMENT=$io_buffer_alignment ${cov_prefix} cargo nextest run $CARGO_FLAGS $CARGO_FEATURES  -E 'package(pageserver)'
-            done
-          done
-
-          # Run separate tests for real S3
-          export ENABLE_REAL_S3_REMOTE_STORAGE=nonempty
-          export REMOTE_STORAGE_S3_BUCKET=neon-github-ci-tests
-          export REMOTE_STORAGE_S3_REGION=eu-central-1
-          ${cov_prefix} cargo nextest run $CARGO_FLAGS $CARGO_FEATURES -E 'package(remote_storage)' -E 'test(test_real_s3)'
-
-          # Run separate tests for real Azure Blob Storage
-          # XXX: replace region with `eu-central-1`-like region
-          export ENABLE_REAL_AZURE_REMOTE_STORAGE=y
-          export AZURE_STORAGE_ACCOUNT="${{ secrets.AZURE_STORAGE_ACCOUNT_DEV }}"
-          export AZURE_STORAGE_ACCESS_KEY="${{ secrets.AZURE_STORAGE_ACCESS_KEY_DEV }}"
-          export REMOTE_STORAGE_AZURE_CONTAINER="${{ vars.REMOTE_STORAGE_AZURE_CONTAINER }}"
-          export REMOTE_STORAGE_AZURE_REGION="${{ vars.REMOTE_STORAGE_AZURE_REGION }}"
-          ${cov_prefix} cargo nextest run $CARGO_FLAGS $CARGO_FEATURES -E 'package(remote_storage)' -E 'test(test_real_azure)'
-
-      - name: Install postgres binaries
-        run: cp -a pg_install /tmp/neon/pg_install
-
-      - name: Upload Neon artifact
-        uses: ./.github/actions/upload
-        with:
-          name: neon-${{ runner.os }}-${{ runner.arch }}-${{ inputs.build-type }}-artifact
-          path: /tmp/neon
-
-      # XXX: keep this after the binaries.list is formed, so the coverage can properly work later
-      - name: Merge and upload coverage data
-        if: inputs.build-type == 'debug'
-        uses: ./.github/actions/save-coverage-data
-
-  regress-tests:
-    # Don't run regression tests on debug arm64 builds
-    if: inputs.build-type != 'debug' || inputs.arch != 'arm64'
-    needs: [ build-neon ]
-    runs-on: ${{ fromJson(format('["self-hosted", "{0}"]', inputs.arch == 'arm64' && 'large-arm64' || 'large')) }}
-    container:
-      image: ${{ inputs.build-tools-image }}
-      credentials:
-        username: ${{ secrets.NEON_DOCKERHUB_USERNAME }}
-        password: ${{ secrets.NEON_DOCKERHUB_PASSWORD }}
-      # for changed limits, see comments on `options:` earlier in this file
-      options: --init --shm-size=512mb --ulimit memlock=67108864:67108864
-    strategy:
-      fail-fast: false
-      matrix:
-        pg_version: ${{ fromJson(inputs.pg-versions) }}
-    steps:
-      - uses: actions/checkout@v4
-        with:
-          submodules: true
-
-      - name: Pytest regression tests
-        uses: ./.github/actions/run-python-test-set
-        timeout-minutes: 60
-        with:
-          build_type: ${{ inputs.build-type }}
-          test_selection: regress
-          needs_postgres_source: true
-          run_with_real_s3: true
-          real_s3_bucket: neon-github-ci-tests
-          real_s3_region: eu-central-1
-          rerun_flaky: true
-          pg_version: ${{ matrix.pg_version }}
-        env:
-          TEST_RESULT_CONNSTR: ${{ secrets.REGRESS_TEST_RESULT_CONNSTR_NEW }}
-          CHECK_ONDISK_DATA_COMPATIBILITY: nonempty
-          BUILD_TAG: ${{ inputs.build-tag }}
-          PAGESERVER_VIRTUAL_FILE_IO_ENGINE: tokio-epoll-uring
-
-      # Temporary disable this step until we figure out why it's so flaky
-      # Ref https://github.com/neondatabase/neon/issues/4540
-      - name: Merge and upload coverage data
-        if: |
-          false &&
-          inputs.build-type == 'debug' && matrix.pg_version == 'v16'
-        uses: ./.github/actions/save-coverage-data

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

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

.github/workflows/actionlint.yml

@@ -24,7 +24,7 @@ jobs:
 
   actionlint:
     needs: [ check-permissions ]
-    runs-on: ubuntu-22.04
+    runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v4
       - uses: reviewdog/action-actionlint@v1
@@ -36,16 +36,3 @@ jobs:
           fail_on_error: true
           filter_mode: nofilter
           level: error
-
-      - name: Disallow 'ubuntu-latest' runners
-        run: |
-          PAT='^\s*runs-on:.*-latest'
-          if grep -ERq $PAT .github/workflows; then
-            grep -ERl $PAT .github/workflows |\
-            while read -r f
-            do
-              l=$(grep -nE $PAT $f | awk -F: '{print $1}' | head -1)
-              echo "::error file=$f,line=$l::Please use 'ubuntu-22.04' instead of 'ubuntu-latest'"
-            done
-            exit 1
-          fi

.github/workflows/approved-for-ci-run.yml

@@ -18,7 +18,6 @@ on:
 
 concurrency:
   group: ${{ github.workflow }}-${{ github.event.pull_request.number }}
-  cancel-in-progress: false
 
 env:
   GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
@@ -44,7 +43,7 @@ jobs:
       contains(fromJSON('["opened", "synchronize", "reopened", "closed"]'), github.event.action) &&
       contains(github.event.pull_request.labels.*.name, 'approved-for-ci-run')
 
-    runs-on: ubuntu-22.04
+    runs-on: ubuntu-latest
 
     steps:
       - run: gh pr --repo "${GITHUB_REPOSITORY}" edit "${PR_NUMBER}" --remove-label "approved-for-ci-run"
@@ -60,7 +59,7 @@ jobs:
       github.event.action == 'labeled' &&
       contains(github.event.pull_request.labels.*.name, 'approved-for-ci-run')
 
-    runs-on: ubuntu-22.04
+    runs-on: ubuntu-latest
 
     steps:
       - run: gh pr --repo "${GITHUB_REPOSITORY}" edit "${PR_NUMBER}" --remove-label "approved-for-ci-run"
@@ -69,41 +68,15 @@ jobs:
         with:
           ref: main
           token: ${{ secrets.CI_ACCESS_TOKEN }}
-      
-      - name: Look for existing PR
-        id: get-pr
-        env:
-          GH_TOKEN: ${{ secrets.CI_ACCESS_TOKEN }}
-        run: |
-          ALREADY_CREATED="$(gh pr --repo ${GITHUB_REPOSITORY} list --head ${BRANCH} --base main --json number --jq '.[].number')"
-          echo "ALREADY_CREATED=${ALREADY_CREATED}" >> ${GITHUB_OUTPUT}
-      
-      - name: Get changed labels
-        id: get-labels
-        if: steps.get-pr.outputs.ALREADY_CREATED != ''
-        env:
-          ALREADY_CREATED: ${{ steps.get-pr.outputs.ALREADY_CREATED }}
-          GH_TOKEN: ${{ secrets.CI_ACCESS_TOKEN }}
-        run: |
-          LABELS_TO_REMOVE=$(comm -23 <(gh pr --repo ${GITHUB_REPOSITORY} view ${ALREADY_CREATED} --json labels --jq '.labels.[].name'| ( grep -E '^run' || true ) | sort) \
-          <(gh pr --repo ${GITHUB_REPOSITORY} view ${PR_NUMBER} --json labels --jq '.labels.[].name' | ( grep -E '^run' || true ) | sort ) |\
-          ( grep -v run-e2e-tests-in-draft || true ) | paste -sd , -)
-          LABELS_TO_ADD=$(comm -13 <(gh pr --repo ${GITHUB_REPOSITORY} view ${ALREADY_CREATED} --json labels --jq '.labels.[].name'| ( grep -E '^run' || true ) |sort) \
-          <(gh pr --repo ${GITHUB_REPOSITORY} view ${PR_NUMBER} --json labels --jq '.labels.[].name' |  ( grep -E '^run' || true ) | sort ) |\
-          paste -sd , -)
-          echo "LABELS_TO_ADD=${LABELS_TO_ADD}" >> ${GITHUB_OUTPUT}
-          echo "LABELS_TO_REMOVE=${LABELS_TO_REMOVE}" >> ${GITHUB_OUTPUT}
 
       - run: gh pr checkout "${PR_NUMBER}"
 
       - run: git checkout -b "${BRANCH}"
 
       - run: git push --force origin "${BRANCH}"
-        if: steps.get-pr.outputs.ALREADY_CREATED == ''
 
       - name: Create a Pull Request for CI run (if required)
-        if: steps.get-pr.outputs.ALREADY_CREATED == ''
-        env: 
+        env:
           GH_TOKEN: ${{ secrets.CI_ACCESS_TOKEN }}
         run: |
           cat << EOF > body.md
@@ -114,33 +87,16 @@ jobs:
             Feel free to review/comment/discuss the original PR #${PR_NUMBER}.
           EOF
 
-          LABELS=$( (gh pr --repo "${GITHUB_REPOSITORY}" view ${PR_NUMBER}  --json labels --jq '.labels.[].name'; echo run-e2e-tests-in-draft  )| \
-          grep -E '^run' | paste -sd , -)
-          gh pr --repo "${GITHUB_REPOSITORY}" create --title "CI run for PR #${PR_NUMBER}" \
+          ALREADY_CREATED="$(gh pr --repo ${GITHUB_REPOSITORY} list --head ${BRANCH} --base main --json number --jq '.[].number')"
+          if [ -z "${ALREADY_CREATED}" ]; then
+            gh pr --repo "${GITHUB_REPOSITORY}" create --title "CI run for PR #${PR_NUMBER}" \
                                                        --body-file "body.md" \
                                                        --head "${BRANCH}" \
                                                        --base "main" \
-                                                       --label ${LABELS} \
+                                                       --label "run-e2e-tests-in-draft" \
                                                        --draft
-      - name: Modify the existing pull request (if required)
-        if: steps.get-pr.outputs.ALREADY_CREATED != ''
-        env:
-          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-          LABELS_TO_ADD: ${{ steps.get-labels.outputs.LABELS_TO_ADD }}
-          LABELS_TO_REMOVE: ${{ steps.get-labels.outputs.LABELS_TO_REMOVE }}
-          ALREADY_CREATED: ${{ steps.get-pr.outputs.ALREADY_CREATED }}
-        run: |
-          ADD_CMD=
-          REMOVE_CMD=
-          [ -z "${LABELS_TO_ADD}" ] || ADD_CMD="--add-label ${LABELS_TO_ADD}"
-          [ -z "${LABELS_TO_REMOVE}" ] || REMOVE_CMD="--remove-label ${LABELS_TO_REMOVE}"
-          if [ -n "${ADD_CMD}" ] || [ -n "${REMOVE_CMD}" ]; then
-            gh pr --repo "${GITHUB_REPOSITORY}" edit ${ALREADY_CREATED} ${ADD_CMD} ${REMOVE_CMD}
           fi
 
-      - run: git push --force origin "${BRANCH}"
-        if: steps.get-pr.outputs.ALREADY_CREATED != ''
-             
   cleanup:
     # Close PRs and delete branchs if the original PR is closed.
 
@@ -152,7 +108,7 @@ jobs:
       github.event.action == 'closed' &&
       github.event.pull_request.head.repo.full_name != github.repository
 
-    runs-on: ubuntu-22.04
+    runs-on: ubuntu-latest
 
     steps:
       - name: Close PR and delete `ci-run/pr-${{ env.PR_NUMBER }}` branch

.github/workflows/benchmarking.yml

@@ -38,11 +38,6 @@ on:
         description: 'AWS-RDS and AWS-AURORA normally only run on Saturday. Set this to true to run them on every workflow_dispatch'
         required: false
         default: false
-      run_only_pgvector_tests:
-        type: boolean
-        description: 'Run pgvector tests but no other tests. If not set, all tests including pgvector tests will be run'
-        required: false
-        default: false
 
 defaults:
   run:
@@ -55,118 +50,11 @@ concurrency:
 
 jobs:
   bench:
-    if: ${{ github.event.inputs.run_only_pgvector_tests == 'false' || github.event.inputs.run_only_pgvector_tests == null }}
-    permissions:
-      contents: write
-      statuses: write
-      id-token: write # Required for OIDC authentication in azure runners
-    strategy:
-      fail-fast: false
-      matrix:
-        include:
-          - DEFAULT_PG_VERSION: 16
-            PLATFORM: "neon-staging"
-            region_id: ${{ github.event.inputs.region_id || 'aws-us-east-2' }}
-            RUNNER: [ self-hosted, us-east-2, x64 ]
-            IMAGE: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/build-tools:pinned
-          - DEFAULT_PG_VERSION: 16
-            PLATFORM: "azure-staging"
-            region_id: 'azure-eastus2'
-            RUNNER: [ self-hosted, eastus2, x64 ]
-            IMAGE: neondatabase/build-tools:pinned
     env:
       TEST_PG_BENCH_DURATIONS_MATRIX: "300"
       TEST_PG_BENCH_SCALES_MATRIX: "10,100"
       POSTGRES_DISTRIB_DIR: /tmp/neon/pg_install
-      DEFAULT_PG_VERSION: ${{ matrix.DEFAULT_PG_VERSION }}
-      TEST_OUTPUT: /tmp/test_output
-      BUILD_TYPE: remote
-      SAVE_PERF_REPORT: ${{ github.event.inputs.save_perf_report || ( github.ref_name == 'main' ) }}
-      PLATFORM: ${{ matrix.PLATFORM }}
-
-    runs-on: ${{ matrix.RUNNER }}
-    container:
-      image: ${{ matrix.IMAGE }}
-      options: --init
-
-    steps:
-    - uses: actions/checkout@v4
-
-    - name: Configure AWS credentials # necessary on Azure runners
-      uses: aws-actions/configure-aws-credentials@v4
-      with:
-        aws-region: eu-central-1
-        role-to-assume: ${{ vars.DEV_AWS_OIDC_ROLE_ARN }}
-        role-duration-seconds: 18000 # 5 hours
-
-    - name: Download Neon artifact
-      uses: ./.github/actions/download
-      with:
-        name: neon-${{ runner.os }}-${{ runner.arch }}-release-artifact
-        path: /tmp/neon/
-        prefix: latest
-
-    - name: Create Neon Project
-      id: create-neon-project
-      uses: ./.github/actions/neon-project-create
-      with:
-        region_id: ${{ matrix.region_id }}
-        postgres_version: ${{ env.DEFAULT_PG_VERSION }}
-        api_key: ${{ secrets.NEON_STAGING_API_KEY }}
-
-    - name: Run benchmark
-      uses: ./.github/actions/run-python-test-set
-      with:
-        build_type: ${{ env.BUILD_TYPE }}
-        test_selection: performance
-        run_in_parallel: false
-        save_perf_report: ${{ env.SAVE_PERF_REPORT }}
-        pg_version: ${{ env.DEFAULT_PG_VERSION }}
-        # Set --sparse-ordering option of pytest-order plugin
-        # to ensure tests are running in order of appears in the file.
-        # It's important for test_perf_pgbench.py::test_pgbench_remote_* tests
-        extra_params:
-          -m remote_cluster
-          --sparse-ordering
-          --timeout 14400
-          --ignore test_runner/performance/test_perf_olap.py
-          --ignore test_runner/performance/test_perf_pgvector_queries.py
-          --ignore test_runner/performance/test_logical_replication.py
-          --ignore test_runner/performance/test_physical_replication.py
-      env:
-        BENCHMARK_CONNSTR: ${{ steps.create-neon-project.outputs.dsn }}
-        VIP_VAP_ACCESS_TOKEN: "${{ secrets.VIP_VAP_ACCESS_TOKEN }}"
-        PERF_TEST_RESULT_CONNSTR: "${{ secrets.PERF_TEST_RESULT_CONNSTR }}"
-
-    - name: Delete Neon Project
-      if: ${{ always() }}
-      uses: ./.github/actions/neon-project-delete
-      with:
-        project_id: ${{ steps.create-neon-project.outputs.project_id }}
-        api_key: ${{ secrets.NEON_STAGING_API_KEY }}
-
-    - 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" # dev-staging-stream
-        slack-message: |
-          Periodic perf testing: ${{ job.status }}
-          <${{ 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 }}
-
-  replication-tests:
-    if: ${{ github.event.inputs.run_only_pgvector_tests == 'false' || github.event.inputs.run_only_pgvector_tests == null }}
-    env:
-      POSTGRES_DISTRIB_DIR: /tmp/neon/pg_install
-      DEFAULT_PG_VERSION: 16
+      DEFAULT_PG_VERSION: 14
       TEST_OUTPUT: /tmp/test_output
       BUILD_TYPE: remote
       SAVE_PERF_REPORT: ${{ github.event.inputs.save_perf_report || ( github.ref_name == 'main' ) }}
@@ -180,81 +68,69 @@ jobs:
     steps:
     - uses: actions/checkout@v4
 
-
     - name: Download Neon artifact
       uses: ./.github/actions/download
       with:
-        name: neon-${{ runner.os }}-${{ runner.arch }}-release-artifact
+        name: neon-${{ runner.os }}-release-artifact
         path: /tmp/neon/
         prefix: latest
 
-    - name: Run Logical Replication benchmarks
-      uses: ./.github/actions/run-python-test-set
+    - name: Create Neon Project
+      id: create-neon-project
+      uses: ./.github/actions/neon-project-create
       with:
-        build_type: ${{ env.BUILD_TYPE }}
-        test_selection: performance/test_logical_replication.py
-        run_in_parallel: false
-        save_perf_report: ${{ env.SAVE_PERF_REPORT }}
-        extra_params: -m remote_cluster --timeout 5400
-        pg_version: ${{ env.DEFAULT_PG_VERSION }}
-      env:
-        VIP_VAP_ACCESS_TOKEN: "${{ secrets.VIP_VAP_ACCESS_TOKEN }}"
-        PERF_TEST_RESULT_CONNSTR: "${{ secrets.PERF_TEST_RESULT_CONNSTR }}"
-        NEON_API_KEY: ${{ secrets.NEON_STAGING_API_KEY }}
-        BENCHMARK_PROJECT_ID_PUB: ${{ vars.BENCHMARK_PROJECT_ID_PUB }}
-        BENCHMARK_PROJECT_ID_SUB: ${{ vars.BENCHMARK_PROJECT_ID_SUB }}
+        region_id: ${{ github.event.inputs.region_id || 'aws-us-east-2' }}
+        postgres_version: ${{ env.DEFAULT_PG_VERSION }}
+        api_key: ${{ secrets.NEON_STAGING_API_KEY }}
 
-    - name: Run Physical Replication benchmarks
+    - name: Run benchmark
       uses: ./.github/actions/run-python-test-set
       with:
         build_type: ${{ env.BUILD_TYPE }}
-        test_selection: performance/test_physical_replication.py
+        test_selection: performance
         run_in_parallel: false
         save_perf_report: ${{ env.SAVE_PERF_REPORT }}
-        extra_params: -m remote_cluster --timeout 5400
-        pg_version: ${{ env.DEFAULT_PG_VERSION }}
+        # Set --sparse-ordering option of pytest-order plugin
+        # to ensure tests are running in order of appears in the file.
+        # It's important for test_perf_pgbench.py::test_pgbench_remote_* tests
+        extra_params: -m remote_cluster --sparse-ordering --timeout 5400 --ignore test_runner/performance/test_perf_olap.py
       env:
+        BENCHMARK_CONNSTR: ${{ steps.create-neon-project.outputs.dsn }}
         VIP_VAP_ACCESS_TOKEN: "${{ secrets.VIP_VAP_ACCESS_TOKEN }}"
         PERF_TEST_RESULT_CONNSTR: "${{ secrets.PERF_TEST_RESULT_CONNSTR }}"
-        NEON_API_KEY: ${{ secrets.NEON_STAGING_API_KEY }}
+
+    - name: Delete Neon Project
+      if: ${{ always() }}
+      uses: ./.github/actions/neon-project-delete
+      with:
+        project_id: ${{ steps.create-neon-project.outputs.project_id }}
+        api_key: ${{ secrets.NEON_STAGING_API_KEY }}
 
     - name: Create Allure report
-      id: create-allure-report
       if: ${{ !cancelled() }}
       uses: ./.github/actions/allure-report-generate
-      with:
-        store-test-results-into-db: true
-      env:
-        REGRESS_TEST_RESULT_CONNSTR_NEW: ${{ secrets.REGRESS_TEST_RESULT_CONNSTR_NEW }}
 
     - name: Post to a Slack channel
       if: ${{ github.event.schedule && failure() }}
       uses: slackapi/slack-github-action@v1
       with:
-        channel-id: "C06T9AMNDQQ" # on-call-compute-staging-stream
-        slack-message: |
-          Periodic replication testing: ${{ job.status }}
-          <${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}|GitHub Run>
-          <${{ steps.create-allure-report.outputs.report-url }}|Allure report>
+        channel-id: "C033QLM5P7D" # dev-staging-stream
+        slack-message: "Periodic perf testing: ${{ job.status }}\n${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}"
       env:
         SLACK_BOT_TOKEN: ${{ secrets.SLACK_BOT_TOKEN }}
 
   generate-matrices:
-    if: ${{ github.event.inputs.run_only_pgvector_tests == 'false' || github.event.inputs.run_only_pgvector_tests == null }}
     # Create matrices for the benchmarking jobs, so we run benchmarks on rds only once a week (on Saturday)
     #
     # Available platforms:
-    # - neonvm-captest-new: Freshly created project (1 CU)
-    # - neonvm-captest-freetier: Use freetier-sized compute (0.25 CU)
-    # - neonvm-captest-azure-new: Freshly created project (1 CU) in azure region
-    # - neonvm-captest-azure-freetier: Use freetier-sized compute (0.25 CU) in azure region
-    # - neonvm-captest-reuse: Reusing existing project
+    # - neon-captest-new: Freshly created project (1 CU)
+    # - neon-captest-freetier: Use freetier-sized compute (0.25 CU)
+    # - neon-captest-reuse: Reusing existing project
     # - rds-aurora: Aurora Postgres Serverless v2 with autoscaling from 0.5 to 2 ACUs
     # - rds-postgres: RDS Postgres db.m5.large instance (2 vCPU, 8 GiB) with gp3 EBS storage
     env:
       RUN_AWS_RDS_AND_AURORA: ${{ github.event.inputs.run_AWS_RDS_AND_AURORA || 'false' }}
-      DEFAULT_REGION_ID: ${{ github.event.inputs.region_id || 'aws-us-east-2' }}
-    runs-on: ubuntu-22.04
+    runs-on: ubuntu-latest
     outputs:
       pgbench-compare-matrix: ${{ steps.pgbench-compare-matrix.outputs.matrix }}
       olap-compare-matrix: ${{ steps.olap-compare-matrix.outputs.matrix }}
@@ -264,37 +140,22 @@ jobs:
     - name: Generate matrix for pgbench benchmark
       id: pgbench-compare-matrix
       run: |
-        region_id_default=${{ env.DEFAULT_REGION_ID }}
-        runner_default='["self-hosted", "us-east-2", "x64"]'
-        runner_azure='["self-hosted", "eastus2", "x64"]'
-        image_default="369495373322.dkr.ecr.eu-central-1.amazonaws.com/build-tools:pinned"
         matrix='{
-          "pg_version" : [
-            16
-          ],
-          "region_id" : [
-            "'"$region_id_default"'"
-            ],
           "platform": [
-            "neonvm-captest-new",
-            "neonvm-captest-reuse",
+            "neon-captest-new",
+            "neon-captest-reuse",
             "neonvm-captest-new"
           ],
           "db_size": [ "10gb" ],
-          "runner": ['"$runner_default"'],
-          "image": [ "'"$image_default"'" ],
-          "include": [{ "pg_version": 16, "region_id": "'"$region_id_default"'", "platform": "neonvm-captest-freetier",       "db_size": "3gb" ,"runner": '"$runner_default"', "image": "'"$image_default"'" },
-                      { "pg_version": 16, "region_id": "'"$region_id_default"'", "platform": "neonvm-captest-new",            "db_size": "10gb","runner": '"$runner_default"', "image": "'"$image_default"'" },
-                      { "pg_version": 16, "region_id": "'"$region_id_default"'", "platform": "neonvm-captest-new",            "db_size": "50gb","runner": '"$runner_default"', "image": "'"$image_default"'" },
-                      { "pg_version": 16, "region_id": "azure-eastus2",          "platform": "neonvm-azure-captest-freetier", "db_size": "3gb" ,"runner": '"$runner_azure"',   "image": "neondatabase/build-tools:pinned" },
-                      { "pg_version": 16, "region_id": "azure-eastus2",          "platform": "neonvm-azure-captest-new",      "db_size": "10gb","runner": '"$runner_azure"',   "image": "neondatabase/build-tools:pinned" },
-                      { "pg_version": 16, "region_id": "azure-eastus2",          "platform": "neonvm-azure-captest-new",      "db_size": "50gb","runner": '"$runner_azure"',   "image": "neondatabase/build-tools:pinned" },
-                      { "pg_version": 16, "region_id": "'"$region_id_default"'", "platform": "neonvm-captest-sharding-reuse", "db_size": "50gb","runner": '"$runner_default"', "image": "'"$image_default"'" }]
+          "include": [{ "platform": "neon-captest-freetier",   "db_size": "3gb"  },
+                      { "platform": "neon-captest-new",        "db_size": "50gb" },
+                      { "platform": "neonvm-captest-freetier", "db_size": "3gb"  },
+                      { "platform": "neonvm-captest-new",      "db_size": "50gb" }]
         }'
 
-        if [ "$(date +%A)" = "Saturday" ] || [ ${RUN_AWS_RDS_AND_AURORA} = "true" ]; then
-          matrix=$(echo "$matrix" | jq '.include += [{ "pg_version": 16, "region_id": "'"$region_id_default"'", "platform": "rds-postgres", "db_size": "10gb","runner": '"$runner_default"', "image": "'"$image_default"'" },
-                                                     { "pg_version": 16, "region_id": "'"$region_id_default"'", "platform": "rds-aurora", "db_size": "10gb","runner": '"$runner_default"', "image": "'"$image_default"'" }]')
+        if [ "$(date +%A)" = "Saturday" ]; then
+          matrix=$(echo "$matrix" | jq '.include += [{ "platform": "rds-postgres", "db_size": "10gb"},
+                                                   { "platform": "rds-aurora",   "db_size": "50gb"}]')
         fi
 
         echo "matrix=$(echo "$matrix" | jq --compact-output '.')" >> $GITHUB_OUTPUT
@@ -304,13 +165,13 @@ jobs:
       run: |
         matrix='{
           "platform": [
-            "neonvm-captest-reuse"
+            "neon-captest-reuse"
           ]
         }'
 
         if [ "$(date +%A)" = "Saturday" ] || [ ${RUN_AWS_RDS_AND_AURORA} = "true" ]; then
           matrix=$(echo "$matrix" | jq '.include += [{ "platform": "rds-postgres" },
-                                                     { "platform": "rds-aurora"   }]')
+                                                   { "platform": "rds-aurora"   }]')
         fi
 
         echo "matrix=$(echo "$matrix" | jq --compact-output '.')" >> $GITHUB_OUTPUT
@@ -320,7 +181,7 @@ jobs:
       run: |
         matrix='{
           "platform": [
-            "neonvm-captest-reuse"
+            "neon-captest-reuse"
           ],
           "scale": [
             "10"
@@ -329,22 +190,13 @@ jobs:
 
         if [ "$(date +%A)" = "Saturday" ] || [ ${RUN_AWS_RDS_AND_AURORA} = "true" ]; then
           matrix=$(echo "$matrix" | jq '.include += [{ "platform": "rds-postgres", "scale": "10" },
-                                                     { "platform": "rds-aurora",   "scale": "10" }]')
+                                                    { "platform": "rds-aurora",   "scale": "10" }]')
         fi
 
         echo "matrix=$(echo "$matrix" | jq --compact-output '.')" >> $GITHUB_OUTPUT
 
-  prepare_AWS_RDS_databases:
-    uses: ./.github/workflows/_benchmarking_preparation.yml
-    secrets: inherit
-
   pgbench-compare:
-    if: ${{ github.event.inputs.run_only_pgvector_tests == 'false' || github.event.inputs.run_only_pgvector_tests == null }}
-    needs: [ generate-matrices, prepare_AWS_RDS_databases ]
-    permissions:
-      contents: write
-      statuses: write
-      id-token: write # Required for OIDC authentication in azure runners
+    needs: [ generate-matrices ]
 
     strategy:
       fail-fast: false
@@ -354,15 +206,15 @@ jobs:
       TEST_PG_BENCH_DURATIONS_MATRIX: "60m"
       TEST_PG_BENCH_SCALES_MATRIX: ${{ matrix.db_size }}
       POSTGRES_DISTRIB_DIR: /tmp/neon/pg_install
-      DEFAULT_PG_VERSION: ${{ matrix.pg_version }}
+      DEFAULT_PG_VERSION: 14
       TEST_OUTPUT: /tmp/test_output
       BUILD_TYPE: remote
       SAVE_PERF_REPORT: ${{ github.event.inputs.save_perf_report || ( github.ref_name == 'main' ) }}
       PLATFORM: ${{ matrix.platform }}
 
-    runs-on: ${{ matrix.runner }}
+    runs-on: [ self-hosted, us-east-2, x64 ]
     container:
-      image: ${{ matrix.image }}
+      image: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/build-tools:pinned
       options: --init
 
     # Increase timeout to 8h, default timeout is 6h
@@ -371,41 +223,37 @@ jobs:
     steps:
     - uses: actions/checkout@v4
 
-    - name: Configure AWS credentials # necessary on Azure runners
-      uses: aws-actions/configure-aws-credentials@v4
-      with:
-        aws-region: eu-central-1
-        role-to-assume: ${{ vars.DEV_AWS_OIDC_ROLE_ARN }}
-        role-duration-seconds: 18000 # 5 hours
-
     - name: Download Neon artifact
       uses: ./.github/actions/download
       with:
-        name: neon-${{ runner.os }}-${{ runner.arch }}-release-artifact
+        name: neon-${{ runner.os }}-release-artifact
         path: /tmp/neon/
         prefix: latest
 
+    - name: Add Postgres binaries to PATH
+      run: |
+        ${POSTGRES_DISTRIB_DIR}/v${DEFAULT_PG_VERSION}/bin/pgbench --version
+        echo "${POSTGRES_DISTRIB_DIR}/v${DEFAULT_PG_VERSION}/bin" >> $GITHUB_PATH
+
     - name: Create Neon Project
-      if: contains(fromJson('["neonvm-captest-new", "neonvm-captest-freetier", "neonvm-azure-captest-freetier", "neonvm-azure-captest-new"]'), matrix.platform)
+      if: contains(fromJson('["neon-captest-new", "neon-captest-freetier", "neonvm-captest-new", "neonvm-captest-freetier"]'), matrix.platform)
       id: create-neon-project
       uses: ./.github/actions/neon-project-create
       with:
-        region_id: ${{ matrix.region_id }}
+        region_id: ${{ github.event.inputs.region_id || 'aws-us-east-2' }}
         postgres_version: ${{ env.DEFAULT_PG_VERSION }}
         api_key: ${{ secrets.NEON_STAGING_API_KEY }}
-        compute_units: ${{ (contains(matrix.platform, 'captest-freetier') && '[0.25, 0.25]') || '[1, 1]' }}
+        compute_units: ${{ (matrix.platform == 'neon-captest-freetier' && '[0.25, 0.25]') || '[1, 1]' }}
+        provisioner: ${{ (contains(matrix.platform, 'neonvm-') && 'k8s-neonvm') || 'k8s-pod' }}
 
     - name: Set up Connection String
       id: set-up-connstr
       run: |
         case "${PLATFORM}" in
-          neonvm-captest-reuse)
+          neon-captest-reuse)
             CONNSTR=${{ secrets.BENCHMARK_CAPTEST_CONNSTR }}
             ;;
-          neonvm-captest-sharding-reuse)
-            CONNSTR=${{ secrets.BENCHMARK_CAPTEST_SHARDING_CONNSTR }}
-            ;;
-          neonvm-captest-new | neonvm-captest-freetier | neonvm-azure-captest-new | neonvm-azure-captest-freetier)
+          neon-captest-new | neon-captest-freetier | neonvm-captest-new | neonvm-captest-freetier)
             CONNSTR=${{ steps.create-neon-project.outputs.dsn }}
             ;;
           rds-aurora)
@@ -422,6 +270,12 @@ jobs:
 
         echo "connstr=${CONNSTR}" >> $GITHUB_OUTPUT
 
+        QUERY="SELECT version();"
+        if [[ "${PLATFORM}" = "neon"* ]]; then
+          QUERY="${QUERY} SHOW neon.tenant_id; SHOW neon.timeline_id;"
+        fi
+        psql ${CONNSTR} -c "${QUERY}"
+
     - name: Benchmark init
       uses: ./.github/actions/run-python-test-set
       with:
@@ -430,7 +284,6 @@ jobs:
         run_in_parallel: false
         save_perf_report: ${{ env.SAVE_PERF_REPORT }}
         extra_params: -m remote_cluster --timeout 21600 -k test_pgbench_remote_init
-        pg_version: ${{ env.DEFAULT_PG_VERSION }}
       env:
         BENCHMARK_CONNSTR: ${{ steps.set-up-connstr.outputs.connstr }}
         VIP_VAP_ACCESS_TOKEN: "${{ secrets.VIP_VAP_ACCESS_TOKEN }}"
@@ -444,7 +297,6 @@ jobs:
         run_in_parallel: false
         save_perf_report: ${{ env.SAVE_PERF_REPORT }}
         extra_params: -m remote_cluster --timeout 21600 -k test_pgbench_remote_simple_update
-        pg_version: ${{ env.DEFAULT_PG_VERSION }}
       env:
         BENCHMARK_CONNSTR: ${{ steps.set-up-connstr.outputs.connstr }}
         VIP_VAP_ACCESS_TOKEN: "${{ secrets.VIP_VAP_ACCESS_TOKEN }}"
@@ -458,7 +310,6 @@ jobs:
         run_in_parallel: false
         save_perf_report: ${{ env.SAVE_PERF_REPORT }}
         extra_params: -m remote_cluster --timeout 21600 -k test_pgbench_remote_select_only
-        pg_version: ${{ env.DEFAULT_PG_VERSION }}
       env:
         BENCHMARK_CONNSTR: ${{ steps.set-up-connstr.outputs.connstr }}
         VIP_VAP_ACCESS_TOKEN: "${{ secrets.VIP_VAP_ACCESS_TOKEN }}"
@@ -472,7 +323,6 @@ jobs:
         api_key: ${{ secrets.NEON_STAGING_API_KEY }}
 
     - name: Create Allure report
-      id: create-allure-report
       if: ${{ !cancelled() }}
       uses: ./.github/actions/allure-report-generate
 
@@ -481,133 +331,7 @@ jobs:
       uses: slackapi/slack-github-action@v1
       with:
         channel-id: "C033QLM5P7D" # dev-staging-stream
-        slack-message: |
-          Periodic perf testing on ${{ matrix.platform }}: ${{ job.status }}
-          <${{ 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 }}
-
-  pgbench-pgvector:
-    permissions:
-      contents: write
-      statuses: write
-      id-token: write # Required for OIDC authentication in azure runners
-    strategy:
-      fail-fast: false
-      matrix:
-        include:
-          - PLATFORM: "neonvm-captest-pgvector"
-            RUNNER: [ self-hosted, us-east-2, x64 ]
-            IMAGE: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/build-tools:pinned
-          - PLATFORM: "azure-captest-pgvector"
-            RUNNER: [ self-hosted, eastus2, x64 ]
-            IMAGE: neondatabase/build-tools:pinned
-
-    env:
-      TEST_PG_BENCH_DURATIONS_MATRIX: "15m"
-      TEST_PG_BENCH_SCALES_MATRIX: "1"
-      POSTGRES_DISTRIB_DIR: /tmp/neon/pg_install
-      DEFAULT_PG_VERSION: 16
-      TEST_OUTPUT: /tmp/test_output
-      BUILD_TYPE: remote
-      LD_LIBRARY_PATH: /home/nonroot/pg/usr/lib/x86_64-linux-gnu
-      SAVE_PERF_REPORT: ${{ github.event.inputs.save_perf_report || ( github.ref_name == 'main' ) }}
-      PLATFORM: ${{ matrix.PLATFORM }}
-
-    runs-on: ${{ matrix.RUNNER }}
-    container:
-      image: ${{ matrix.IMAGE }}
-      options: --init
-
-    steps:
-    - uses: actions/checkout@v4
-
-    # until https://github.com/neondatabase/neon/issues/8275 is fixed we temporarily install postgresql-16
-    # instead of using Neon artifacts containing pgbench
-    - name: Install postgresql-16 where pytest expects it
-      run: |
-        cd /home/nonroot
-        wget -q https://apt.postgresql.org/pub/repos/apt/pool/main/p/postgresql-16/libpq5_16.4-1.pgdg110%2B1_amd64.deb
-        wget -q https://apt.postgresql.org/pub/repos/apt/pool/main/p/postgresql-16/postgresql-client-16_16.4-1.pgdg110%2B1_amd64.deb
-        wget -q https://apt.postgresql.org/pub/repos/apt/pool/main/p/postgresql-16/postgresql-16_16.4-1.pgdg110%2B1_amd64.deb 
-        dpkg -x libpq5_16.4-1.pgdg110+1_amd64.deb pg
-        dpkg -x postgresql-client-16_16.4-1.pgdg110+1_amd64.deb pg
-        dpkg -x postgresql-16_16.4-1.pgdg110+1_amd64.deb pg
-        mkdir -p /tmp/neon/pg_install/v16/bin
-        ln -s /home/nonroot/pg/usr/lib/postgresql/16/bin/pgbench /tmp/neon/pg_install/v16/bin/pgbench  
-        ln -s /home/nonroot/pg/usr/lib/postgresql/16/bin/psql /tmp/neon/pg_install/v16/bin/psql  
-        ln -s /home/nonroot/pg/usr/lib/x86_64-linux-gnu /tmp/neon/pg_install/v16/lib 
-        /tmp/neon/pg_install/v16/bin/pgbench --version
-        /tmp/neon/pg_install/v16/bin/psql --version
-
-    - name: Set up Connection String
-      id: set-up-connstr
-      run: |
-        case "${PLATFORM}" in
-          neonvm-captest-pgvector)
-            CONNSTR=${{ secrets.BENCHMARK_PGVECTOR_CONNSTR }}
-            ;;
-          azure-captest-pgvector)
-            CONNSTR=${{ secrets.BENCHMARK_PGVECTOR_CONNSTR_AZURE }}
-            ;;
-          *)
-            echo >&2 "Unknown PLATFORM=${PLATFORM}"
-            exit 1
-            ;;
-        esac
-
-        echo "connstr=${CONNSTR}" >> $GITHUB_OUTPUT
-
-    - name: Configure AWS credentials # necessary on Azure runners to read/write from/to S3
-      uses: aws-actions/configure-aws-credentials@v4
-      with:
-        aws-region: eu-central-1
-        role-to-assume: ${{ vars.DEV_AWS_OIDC_ROLE_ARN }}
-        role-duration-seconds: 18000 # 5 hours
-
-    - name: Benchmark pgvector hnsw indexing
-      uses: ./.github/actions/run-python-test-set
-      with:
-        build_type: ${{ env.BUILD_TYPE }}
-        test_selection: performance/test_perf_olap.py
-        run_in_parallel: false
-        save_perf_report: ${{ env.SAVE_PERF_REPORT }}
-        extra_params: -m remote_cluster --timeout 21600 -k test_pgvector_indexing
-        pg_version: ${{ env.DEFAULT_PG_VERSION }}
-      env:
-        VIP_VAP_ACCESS_TOKEN: "${{ secrets.VIP_VAP_ACCESS_TOKEN }}"
-        PERF_TEST_RESULT_CONNSTR: "${{ secrets.PERF_TEST_RESULT_CONNSTR }}"
-        BENCHMARK_CONNSTR: ${{ steps.set-up-connstr.outputs.connstr }}
-
-    - name: Benchmark pgvector queries
-      uses: ./.github/actions/run-python-test-set
-      with:
-        build_type: ${{ env.BUILD_TYPE }}
-        test_selection: performance/test_perf_pgvector_queries.py
-        run_in_parallel: false
-        save_perf_report: ${{ env.SAVE_PERF_REPORT }}
-        extra_params: -m remote_cluster --timeout 21600
-        pg_version: ${{ env.DEFAULT_PG_VERSION }}
-      env:
-        BENCHMARK_CONNSTR: ${{ steps.set-up-connstr.outputs.connstr }}
-        VIP_VAP_ACCESS_TOKEN: "${{ secrets.VIP_VAP_ACCESS_TOKEN }}"
-        PERF_TEST_RESULT_CONNSTR: "${{ secrets.PERF_TEST_RESULT_CONNSTR }}"
-
-    - 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" # dev-staging-stream
-        slack-message: |
-          Periodic perf testing on ${{ env.PLATFORM }}: ${{ job.status }}
-          <${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}|GitHub Run>
-          <${{ steps.create-allure-report.outputs.report-url }}|Allure report>
+        slack-message: "Periodic perf testing ${{ matrix.platform }}: ${{ job.status }}\n${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}"
       env:
         SLACK_BOT_TOKEN: ${{ secrets.SLACK_BOT_TOKEN }}
 
@@ -619,8 +343,8 @@ jobs:
     #
     # *_CLICKBENCH_CONNSTR: Genuine ClickBench DB with ~100M rows
     # *_CLICKBENCH_10M_CONNSTR: DB with the first 10M rows of ClickBench DB
-    if: ${{ !cancelled() && (github.event.inputs.run_only_pgvector_tests == 'false' || github.event.inputs.run_only_pgvector_tests == null) }}
-    needs: [ generate-matrices, pgbench-compare, prepare_AWS_RDS_databases ]
+    if: ${{ !cancelled() }}
+    needs: [ generate-matrices, pgbench-compare ]
 
     strategy:
       fail-fast: false
@@ -628,7 +352,7 @@ jobs:
 
     env:
       POSTGRES_DISTRIB_DIR: /tmp/neon/pg_install
-      DEFAULT_PG_VERSION: 16
+      DEFAULT_PG_VERSION: 14
       TEST_OUTPUT: /tmp/test_output
       TEST_OLAP_COLLECT_EXPLAIN: ${{ github.event.inputs.collect_olap_explain }}
       TEST_OLAP_COLLECT_PG_STAT_STATEMENTS: ${{ github.event.inputs.collect_pg_stat_statements }}
@@ -647,15 +371,20 @@ jobs:
     - name: Download Neon artifact
       uses: ./.github/actions/download
       with:
-        name: neon-${{ runner.os }}-${{ runner.arch }}-release-artifact
+        name: neon-${{ runner.os }}-release-artifact
         path: /tmp/neon/
         prefix: latest
 
+    - name: Add Postgres binaries to PATH
+      run: |
+        ${POSTGRES_DISTRIB_DIR}/v${DEFAULT_PG_VERSION}/bin/pgbench --version
+        echo "${POSTGRES_DISTRIB_DIR}/v${DEFAULT_PG_VERSION}/bin" >> $GITHUB_PATH
+
     - name: Set up Connection String
       id: set-up-connstr
       run: |
         case "${PLATFORM}" in
-          neonvm-captest-reuse)
+          neon-captest-reuse)
             CONNSTR=${{ secrets.BENCHMARK_CAPTEST_CLICKBENCH_10M_CONNSTR }}
             ;;
           rds-aurora)
@@ -665,13 +394,19 @@ jobs:
             CONNSTR=${{ secrets.BENCHMARK_RDS_POSTGRES_CLICKBENCH_10M_CONNSTR }}
             ;;
           *)
-            echo >&2 "Unknown PLATFORM=${PLATFORM}. Allowed only 'neonvm-captest-reuse', 'rds-aurora', or 'rds-postgres'"
+            echo >&2 "Unknown PLATFORM=${PLATFORM}. Allowed only 'neon-captest-reuse', 'rds-aurora', or 'rds-postgres'"
             exit 1
             ;;
         esac
 
         echo "connstr=${CONNSTR}" >> $GITHUB_OUTPUT
 
+        QUERY="SELECT version();"
+        if [[ "${PLATFORM}" = "neon"* ]]; then
+          QUERY="${QUERY} SHOW neon.tenant_id; SHOW neon.timeline_id;"
+        fi
+        psql ${CONNSTR} -c "${QUERY}"
+
     - name: ClickBench benchmark
       uses: ./.github/actions/run-python-test-set
       with:
@@ -680,7 +415,6 @@ jobs:
         run_in_parallel: false
         save_perf_report: ${{ env.SAVE_PERF_REPORT }}
         extra_params: -m remote_cluster --timeout 21600 -k test_clickbench
-        pg_version: ${{ env.DEFAULT_PG_VERSION }}
       env:
         VIP_VAP_ACCESS_TOKEN: "${{ secrets.VIP_VAP_ACCESS_TOKEN }}"
         PERF_TEST_RESULT_CONNSTR: "${{ secrets.PERF_TEST_RESULT_CONNSTR }}"
@@ -690,7 +424,6 @@ jobs:
         TEST_OLAP_SCALE: 10
 
     - name: Create Allure report
-      id: create-allure-report
       if: ${{ !cancelled() }}
       uses: ./.github/actions/allure-report-generate
 
@@ -699,10 +432,7 @@ jobs:
       uses: slackapi/slack-github-action@v1
       with:
         channel-id: "C033QLM5P7D" # dev-staging-stream
-        slack-message: |
-          Periodic OLAP perf testing on ${{ matrix.platform }}: ${{ job.status }}
-          <${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}|GitHub Run>
-          <${{ steps.create-allure-report.outputs.report-url }}|Allure report>
+        slack-message: "Periodic OLAP perf testing ${{ matrix.platform }}: ${{ job.status }}\n${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}"
       env:
         SLACK_BOT_TOKEN: ${{ secrets.SLACK_BOT_TOKEN }}
 
@@ -713,8 +443,8 @@ jobs:
     # We might change it after https://github.com/neondatabase/neon/issues/2900.
     #
     # *_TPCH_S10_CONNSTR: DB generated with scale factor 10 (~10 GB)
-    if: ${{ !cancelled() && (github.event.inputs.run_only_pgvector_tests == 'false' || github.event.inputs.run_only_pgvector_tests == null) }}
-    needs: [ generate-matrices, clickbench-compare, prepare_AWS_RDS_databases ]
+    if: ${{ !cancelled() }}
+    needs: [ generate-matrices, clickbench-compare ]
 
     strategy:
       fail-fast: false
@@ -722,7 +452,7 @@ jobs:
 
     env:
       POSTGRES_DISTRIB_DIR: /tmp/neon/pg_install
-      DEFAULT_PG_VERSION: 16
+      DEFAULT_PG_VERSION: 14
       TEST_OUTPUT: /tmp/test_output
       BUILD_TYPE: remote
       SAVE_PERF_REPORT: ${{ github.event.inputs.save_perf_report || ( github.ref_name == 'main' ) }}
@@ -740,24 +470,29 @@ jobs:
     - name: Download Neon artifact
       uses: ./.github/actions/download
       with:
-        name: neon-${{ runner.os }}-${{ runner.arch }}-release-artifact
+        name: neon-${{ runner.os }}-release-artifact
         path: /tmp/neon/
         prefix: latest
 
+    - name: Add Postgres binaries to PATH
+      run: |
+        ${POSTGRES_DISTRIB_DIR}/v${DEFAULT_PG_VERSION}/bin/pgbench --version
+        echo "${POSTGRES_DISTRIB_DIR}/v${DEFAULT_PG_VERSION}/bin" >> $GITHUB_PATH
+
     - name: Get Connstring Secret Name
       run: |
         case "${PLATFORM}" in
-          neonvm-captest-reuse)
+          neon-captest-reuse)
             ENV_PLATFORM=CAPTEST_TPCH
             ;;
           rds-aurora)
             ENV_PLATFORM=RDS_AURORA_TPCH
             ;;
           rds-postgres)
-            ENV_PLATFORM=RDS_POSTGRES_TPCH
+            ENV_PLATFORM=RDS_AURORA_TPCH
             ;;
           *)
-            echo >&2 "Unknown PLATFORM=${PLATFORM}. Allowed only 'neonvm-captest-reuse', 'rds-aurora', or 'rds-postgres'"
+            echo >&2 "Unknown PLATFORM=${PLATFORM}. Allowed only 'neon-captest-reuse', 'rds-aurora', or 'rds-postgres'"
             exit 1
             ;;
         esac
@@ -772,6 +507,12 @@ jobs:
 
         echo "connstr=${CONNSTR}" >> $GITHUB_OUTPUT
 
+        QUERY="SELECT version();"
+        if [[ "${PLATFORM}" = "neon"* ]]; then
+          QUERY="${QUERY} SHOW neon.tenant_id; SHOW neon.timeline_id;"
+        fi
+        psql ${CONNSTR} -c "${QUERY}"
+
     - name: Run TPC-H benchmark
       uses: ./.github/actions/run-python-test-set
       with:
@@ -780,7 +521,6 @@ jobs:
         run_in_parallel: false
         save_perf_report: ${{ env.SAVE_PERF_REPORT }}
         extra_params: -m remote_cluster --timeout 21600 -k test_tpch
-        pg_version: ${{ env.DEFAULT_PG_VERSION }}
       env:
         VIP_VAP_ACCESS_TOKEN: "${{ secrets.VIP_VAP_ACCESS_TOKEN }}"
         PERF_TEST_RESULT_CONNSTR: "${{ secrets.PERF_TEST_RESULT_CONNSTR }}"
@@ -788,7 +528,6 @@ jobs:
         TEST_OLAP_SCALE: ${{ matrix.scale }}
 
     - name: Create Allure report
-      id: create-allure-report
       if: ${{ !cancelled() }}
       uses: ./.github/actions/allure-report-generate
 
@@ -797,16 +536,13 @@ jobs:
       uses: slackapi/slack-github-action@v1
       with:
         channel-id: "C033QLM5P7D" # dev-staging-stream
-        slack-message: |
-          Periodic TPC-H perf testing on ${{ matrix.platform }}: ${{ job.status }}
-          <${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}|GitHub Run>
-          <${{ steps.create-allure-report.outputs.report-url }}|Allure report>
+        slack-message: "Periodic TPC-H perf testing ${{ matrix.platform }}: ${{ job.status }}\n${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}"
       env:
         SLACK_BOT_TOKEN: ${{ secrets.SLACK_BOT_TOKEN }}
 
   user-examples-compare:
-    if: ${{ !cancelled() && (github.event.inputs.run_only_pgvector_tests == 'false' || github.event.inputs.run_only_pgvector_tests == null) }}
-    needs: [ generate-matrices, tpch-compare, prepare_AWS_RDS_databases ]
+    if: ${{ !cancelled() }}
+    needs: [ generate-matrices, tpch-compare ]
 
     strategy:
       fail-fast: false
@@ -814,7 +550,7 @@ jobs:
 
     env:
       POSTGRES_DISTRIB_DIR: /tmp/neon/pg_install
-      DEFAULT_PG_VERSION: 16
+      DEFAULT_PG_VERSION: 14
       TEST_OUTPUT: /tmp/test_output
       BUILD_TYPE: remote
       SAVE_PERF_REPORT: ${{ github.event.inputs.save_perf_report || ( github.ref_name == 'main' ) }}
@@ -831,15 +567,20 @@ jobs:
     - name: Download Neon artifact
       uses: ./.github/actions/download
       with:
-        name: neon-${{ runner.os }}-${{ runner.arch }}-release-artifact
+        name: neon-${{ runner.os }}-release-artifact
         path: /tmp/neon/
         prefix: latest
 
+    - name: Add Postgres binaries to PATH
+      run: |
+        ${POSTGRES_DISTRIB_DIR}/v${DEFAULT_PG_VERSION}/bin/pgbench --version
+        echo "${POSTGRES_DISTRIB_DIR}/v${DEFAULT_PG_VERSION}/bin" >> $GITHUB_PATH
+
     - name: Set up Connection String
       id: set-up-connstr
       run: |
         case "${PLATFORM}" in
-          neonvm-captest-reuse)
+          neon-captest-reuse)
             CONNSTR=${{ secrets.BENCHMARK_USER_EXAMPLE_CAPTEST_CONNSTR }}
             ;;
           rds-aurora)
@@ -849,13 +590,19 @@ jobs:
             CONNSTR=${{ secrets.BENCHMARK_USER_EXAMPLE_RDS_POSTGRES_CONNSTR }}
             ;;
           *)
-            echo >&2 "Unknown PLATFORM=${PLATFORM}. Allowed only 'neonvm-captest-reuse', 'rds-aurora', or 'rds-postgres'"
+            echo >&2 "Unknown PLATFORM=${PLATFORM}. Allowed only 'neon-captest-reuse', 'rds-aurora', or 'rds-postgres'"
             exit 1
             ;;
         esac
 
         echo "connstr=${CONNSTR}" >> $GITHUB_OUTPUT
 
+        QUERY="SELECT version();"
+        if [[ "${PLATFORM}" = "neon"* ]]; then
+          QUERY="${QUERY} SHOW neon.tenant_id; SHOW neon.timeline_id;"
+        fi
+        psql ${CONNSTR} -c "${QUERY}"
+
     - name: Run user examples
       uses: ./.github/actions/run-python-test-set
       with:
@@ -864,14 +611,12 @@ jobs:
         run_in_parallel: false
         save_perf_report: ${{ env.SAVE_PERF_REPORT }}
         extra_params: -m remote_cluster --timeout 21600 -k test_user_examples
-        pg_version: ${{ env.DEFAULT_PG_VERSION }}
       env:
         VIP_VAP_ACCESS_TOKEN: "${{ secrets.VIP_VAP_ACCESS_TOKEN }}"
         PERF_TEST_RESULT_CONNSTR: "${{ secrets.PERF_TEST_RESULT_CONNSTR }}"
         BENCHMARK_CONNSTR: ${{ steps.set-up-connstr.outputs.connstr }}
 
     - name: Create Allure report
-      id: create-allure-report
       if: ${{ !cancelled() }}
       uses: ./.github/actions/allure-report-generate
 
@@ -880,10 +625,6 @@ jobs:
       uses: slackapi/slack-github-action@v1
       with:
         channel-id: "C033QLM5P7D" # dev-staging-stream
-        slack-message: |
-          Periodic TPC-H perf testing on ${{ matrix.platform }}: ${{ job.status }}
-          <${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}|GitHub Run>
-          <${{ steps.create-allure-report.outputs.report-url }}|Allure report>
-
+        slack-message: "Periodic User example perf testing ${{ matrix.platform }}: ${{ job.status }}\n${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}"
       env:
         SLACK_BOT_TOKEN: ${{ secrets.SLACK_BOT_TOKEN }}

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

@@ -21,7 +21,6 @@ defaults:
 
 concurrency:
   group: build-build-tools-image-${{ inputs.image-tag }}
-  cancel-in-progress: false
 
 # No permission for GITHUB_TOKEN by default; the **minimal required** set of permissions should be granted in each job.
 permissions: {}
@@ -30,6 +29,7 @@ jobs:
   check-image:
     uses: ./.github/workflows/check-build-tools-image.yml
 
+  # This job uses older version of GitHub Actions because it's run on gen2 runners, which don't support node 20 (for newer versions)
   build-image:
     needs: [ check-image ]
     if: needs.check-image.outputs.found == 'false'
@@ -38,7 +38,7 @@ jobs:
       matrix:
         arch: [ x64, arm64 ]
 
-    runs-on: ${{ fromJson(format('["self-hosted", "{0}"]', matrix.arch == 'arm64' && 'large-arm64' || 'large')) }}
+    runs-on: ${{ fromJson(format('["self-hosted", "dev", "{0}"]', matrix.arch)) }}
 
     env:
       IMAGE_TAG: ${{ inputs.image-tag }}
@@ -54,38 +54,40 @@ jobs:
             exit 1
           fi
 
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v3
 
-      - uses: ./.github/actions/set-docker-config-dir
-      - uses: docker/setup-buildx-action@v3
-        with:
-          cache-binary: false
+      # Use custom DOCKER_CONFIG directory to avoid conflicts with default settings
+      # The default value is ~/.docker
+      - name: Set custom docker config directory
+        run: |
+          mkdir -p /tmp/.docker-custom
+          echo DOCKER_CONFIG=/tmp/.docker-custom >> $GITHUB_ENV
 
-      - uses: docker/login-action@v3
+      - uses: docker/setup-buildx-action@v2
+
+      - uses: docker/login-action@v2
         with:
           username: ${{ secrets.NEON_DOCKERHUB_USERNAME }}
           password: ${{ secrets.NEON_DOCKERHUB_PASSWORD }}
 
-      - uses: docker/login-action@v3
-        with:
-          registry: cache.neon.build
-          username: ${{ secrets.NEON_CI_DOCKERCACHE_USERNAME }}
-          password: ${{ secrets.NEON_CI_DOCKERCACHE_PASSWORD }}
-
-      - uses: docker/build-push-action@v6
+      - uses: docker/build-push-action@v4
         with:
           context: .
           provenance: false
           push: true
           pull: true
           file: Dockerfile.build-tools
-          cache-from: type=registry,ref=cache.neon.build/build-tools:cache-${{ matrix.arch }}
-          cache-to: ${{ github.ref_name == 'main' && format('type=registry,ref=cache.neon.build/build-tools:cache-{0},mode=max', matrix.arch) || '' }}
+          cache-from: type=registry,ref=neondatabase/build-tools:cache-${{ matrix.arch }}
+          cache-to: type=registry,ref=neondatabase/build-tools:cache-${{ matrix.arch }},mode=max
           tags: neondatabase/build-tools:${{ inputs.image-tag }}-${{ matrix.arch }}
 
+      - name: Remove custom docker config directory
+        run: |
+          rm -rf /tmp/.docker-custom
+
   merge-images:
     needs: [ build-image ]
-    runs-on: ubuntu-22.04
+    runs-on: ubuntu-latest
 
     env:
       IMAGE_TAG: ${{ inputs.image-tag }}

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

@@ -19,23 +19,30 @@ permissions: {}
 
 jobs:
   check-image:
-    runs-on: ubuntu-22.04
+    runs-on: ubuntu-latest
     outputs:
       tag: ${{ steps.get-build-tools-tag.outputs.image-tag }}
       found: ${{ steps.check-image.outputs.found }}
 
     steps:
-      - uses: actions/checkout@v4
-
       - name: Get build-tools image tag for the current commit
         id: get-build-tools-tag
         env:
-          IMAGE_TAG: |
-            ${{ hashFiles('Dockerfile.build-tools',
-                          '.github/workflows/check-build-tools-image.yml',
-                          '.github/workflows/build-build-tools-image.yml') }}
+          COMMIT_SHA: ${{ github.event.pull_request.head.sha || github.sha }}
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
         run: |
-          echo "image-tag=${IMAGE_TAG}" | tee -a $GITHUB_OUTPUT
+          LAST_BUILD_TOOLS_SHA=$(
+            gh api \
+              -H "Accept: application/vnd.github+json" \
+              -H "X-GitHub-Api-Version: 2022-11-28" \
+              --method GET \
+              --field path=Dockerfile.build-tools \
+              --field sha=${COMMIT_SHA} \
+              --field per_page=1 \
+              --jq ".[0].sha" \
+              "/repos/${GITHUB_REPOSITORY}/commits"
+          )
+          echo "image-tag=${LAST_BUILD_TOOLS_SHA}" | tee -a $GITHUB_OUTPUT
 
       - name: Check if such tag found in the registry
         id: check-image

.github/workflows/check-permissions.yml

@@ -16,7 +16,7 @@ permissions: {}
 
 jobs:
   check-permissions:
-    runs-on: ubuntu-22.04
+    runs-on: ubuntu-latest
     steps:
     - name: Disallow CI runs on PRs from forks
       if: |

.github/workflows/cleanup-caches-by-a-branch.yml

@@ -9,7 +9,7 @@ on:
 
 jobs:
   cleanup:
-    runs-on: ubuntu-22.04
+    runs-on: ubuntu-latest
     steps:
       - name: Cleanup
         run: |

.github/workflows/label-for-external-users.yml

@@ -1,78 +0,0 @@
-name: Add `external` label to issues and PRs created by external users
-
-on:
-  issues:
-    types:
-      - opened
-  pull_request_target:
-    types:
-      - opened
-  workflow_dispatch:
-    inputs:
-      github-actor:
-        description: 'GitHub username. If empty, the username of the current user will be used'
-        required: false
-
-# No permission for GITHUB_TOKEN by default; the **minimal required** set of permissions should be granted in each job.
-permissions: {}
-
-env:
-  LABEL: external
-
-jobs:
-  check-user:
-    runs-on: ubuntu-22.04
-
-    outputs:
-      is-member: ${{ steps.check-user.outputs.is-member }}
-
-    steps:
-    - name: Check whether `${{ github.actor }}` is a member of `${{ github.repository_owner }}`
-      id: check-user
-      env:
-        GH_TOKEN: ${{ secrets.CI_ACCESS_TOKEN }}
-        ACTOR: ${{ inputs.github-actor || github.actor }}
-      run: |
-        expected_error="User does not exist or is not a member of the organization"
-        output_file=output.txt
-
-        for i in $(seq 1 10); do
-          if gh api "/orgs/${GITHUB_REPOSITORY_OWNER}/members/${ACTOR}" \
-              -H "Accept: application/vnd.github+json" \
-              -H "X-GitHub-Api-Version: 2022-11-28" > ${output_file}; then
-
-            is_member=true
-            break
-          elif grep -q "${expected_error}" ${output_file}; then
-            is_member=false
-            break
-          elif [ $i -eq 10 ]; then
-            title="Failed to get memmbership status for ${ACTOR}"
-            message="The latest GitHub API error message: '$(cat ${output_file})'"
-            echo "::error file=.github/workflows/label-for-external-users.yml,title=${title}::${message}"
-
-            exit 1
-          fi
-
-          sleep 1
-        done
-
-        echo "is-member=${is_member}" | tee -a ${GITHUB_OUTPUT}
-
-  add-label:
-    if: needs.check-user.outputs.is-member == 'false'
-    needs: [ check-user ]
-
-    runs-on: ubuntu-22.04
-    permissions:
-      pull-requests: write # for `gh pr edit`
-      issues: write        # for `gh issue edit`
-
-    steps:
-    - name: Add `${{ env.LABEL }}` label
-      env:
-        GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-        ITEM_NUMBER: ${{ github.event[github.event_name == 'pull_request_target' && 'pull_request' || 'issue'].number }}
-        GH_CLI_COMMAND: ${{ github.event_name == 'pull_request_target' && 'pr' || 'issue' }}
-      run: |
-        gh ${GH_CLI_COMMAND} --repo ${GITHUB_REPOSITORY} edit --add-label=${LABEL} ${ITEM_NUMBER}

.github/workflows/neon_extra_builds.yml

@@ -56,6 +56,7 @@ jobs:
         uses: actions/checkout@v4
         with:
           submodules: true
+          fetch-depth: 1
 
       - name: Install macOS postgres dependencies
         run: brew install flex bison openssl protobuf icu4c pkg-config
@@ -132,13 +133,212 @@ jobs:
       - name: Check that no warnings are produced
         run: ./run_clippy.sh
 
+  check-linux-arm-build:
+    needs: [ check-permissions, build-build-tools-image ]
+    timeout-minutes: 90
+    runs-on: [ self-hosted, dev, arm64 ]
+
+    env:
+      # Use release build only, to have less debug info around
+      # Hence keeping target/ (and general cache size) smaller
+      BUILD_TYPE: release
+      CARGO_FEATURES: --features testing
+      CARGO_FLAGS: --release
+      AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_DEV }}
+      AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_KEY_DEV }}
+
+    container:
+      image: ${{ needs.build-build-tools-image.outputs.image }}
+      credentials:
+        username: ${{ secrets.NEON_DOCKERHUB_USERNAME }}
+        password: ${{ secrets.NEON_DOCKERHUB_PASSWORD }}
+      options: --init
+
+    steps:
+      - name: Fix git ownership
+        run: |
+          # Workaround for `fatal: detected dubious ownership in repository at ...`
+          #
+          # Use both ${{ github.workspace }} and ${GITHUB_WORKSPACE} because they're different on host and in containers
+          #   Ref https://github.com/actions/checkout/issues/785
+          #
+          git config --global --add safe.directory ${{ github.workspace }}
+          git config --global --add safe.directory ${GITHUB_WORKSPACE}
+          for r in 14 15 16; do
+            git config --global --add safe.directory "${{ github.workspace }}/vendor/postgres-v$r"
+            git config --global --add safe.directory "${GITHUB_WORKSPACE}/vendor/postgres-v$r"
+          done
+
+      - name: Checkout
+        uses: actions/checkout@v4
+        with:
+          submodules: true
+          fetch-depth: 1
+
+      - name: Set pg 14 revision for caching
+        id: pg_v14_rev
+        run: echo pg_rev=$(git rev-parse HEAD:vendor/postgres-v14) >> $GITHUB_OUTPUT
+
+      - name: Set pg 15 revision for caching
+        id: pg_v15_rev
+        run: echo pg_rev=$(git rev-parse HEAD:vendor/postgres-v15) >> $GITHUB_OUTPUT
+
+      - name: Set pg 16 revision for caching
+        id: pg_v16_rev
+        run: echo pg_rev=$(git rev-parse HEAD:vendor/postgres-v16) >> $GITHUB_OUTPUT
+
+      - name: Set env variables
+        run: |
+          echo "CARGO_HOME=${GITHUB_WORKSPACE}/.cargo" >> $GITHUB_ENV
+
+      - name: Cache postgres v14 build
+        id: cache_pg_14
+        uses: actions/cache@v4
+        with:
+          path: pg_install/v14
+          key: v1-${{ runner.os }}-${{ runner.arch }}-${{ env.BUILD_TYPE }}-pg-${{ steps.pg_v14_rev.outputs.pg_rev }}-${{ hashFiles('Makefile') }}
+
+      - name: Cache postgres v15 build
+        id: cache_pg_15
+        uses: actions/cache@v4
+        with:
+          path: pg_install/v15
+          key: v1-${{ runner.os }}-${{ runner.arch }}-${{ env.BUILD_TYPE }}-pg-${{ steps.pg_v15_rev.outputs.pg_rev }}-${{ hashFiles('Makefile') }}
+
+      - name: Cache postgres v16 build
+        id: cache_pg_16
+        uses: actions/cache@v4
+        with:
+          path: pg_install/v16
+          key: v1-${{ runner.os }}-${{ runner.arch }}-${{ env.BUILD_TYPE }}-pg-${{ steps.pg_v16_rev.outputs.pg_rev }}-${{ hashFiles('Makefile') }}
+
+      - name: Build postgres v14
+        if: steps.cache_pg_14.outputs.cache-hit != 'true'
+        run: mold -run make postgres-v14 -j$(nproc)
+
+      - name: Build postgres v15
+        if: steps.cache_pg_15.outputs.cache-hit != 'true'
+        run: mold -run make postgres-v15 -j$(nproc)
+
+      - name: Build postgres v16
+        if: steps.cache_pg_16.outputs.cache-hit != 'true'
+        run: mold -run make postgres-v16 -j$(nproc)
+
+      - name: Build neon extensions
+        run: mold -run make neon-pg-ext -j$(nproc)
+
+      - name: Build walproposer-lib
+        run: mold -run make walproposer-lib -j$(nproc)
+
+      - name: Run cargo build
+        run: |
+          mold -run cargo build --locked $CARGO_FLAGS $CARGO_FEATURES --bins --tests
+
+      - name: Run cargo test
+        env:
+          NEXTEST_RETRIES: 3
+        run: |
+          cargo nextest run $CARGO_FEATURES
+
+          # Run separate tests for real S3
+          export ENABLE_REAL_S3_REMOTE_STORAGE=nonempty
+          export REMOTE_STORAGE_S3_BUCKET=neon-github-ci-tests
+          export REMOTE_STORAGE_S3_REGION=eu-central-1
+          # Avoid `$CARGO_FEATURES` since there's no `testing` feature in the e2e tests now
+          cargo nextest run --package remote_storage --test test_real_s3
+
+          # Run separate tests for real Azure Blob Storage
+          # XXX: replace region with `eu-central-1`-like region
+          export ENABLE_REAL_AZURE_REMOTE_STORAGE=y
+          export AZURE_STORAGE_ACCOUNT="${{ secrets.AZURE_STORAGE_ACCOUNT_DEV }}"
+          export AZURE_STORAGE_ACCESS_KEY="${{ secrets.AZURE_STORAGE_ACCESS_KEY_DEV }}"
+          export REMOTE_STORAGE_AZURE_CONTAINER="${{ vars.REMOTE_STORAGE_AZURE_CONTAINER }}"
+          export REMOTE_STORAGE_AZURE_REGION="${{ vars.REMOTE_STORAGE_AZURE_REGION }}"
+          # Avoid `$CARGO_FEATURES` since there's no `testing` feature in the e2e tests now
+          cargo nextest run --package remote_storage --test test_real_azure
+
+  check-codestyle-rust-arm:
+    needs: [ check-permissions, build-build-tools-image ]
+    timeout-minutes: 90
+    runs-on: [ self-hosted, dev, arm64 ]
+
+    container:
+      image: ${{ needs.build-build-tools-image.outputs.image }}
+      credentials:
+        username: ${{ secrets.NEON_DOCKERHUB_USERNAME }}
+        password: ${{ secrets.NEON_DOCKERHUB_PASSWORD }}
+      options: --init
+
+    steps:
+      - name: Fix git ownership
+        run: |
+          # Workaround for `fatal: detected dubious ownership in repository at ...`
+          #
+          # Use both ${{ github.workspace }} and ${GITHUB_WORKSPACE} because they're different on host and in containers
+          #   Ref https://github.com/actions/checkout/issues/785
+          #
+          git config --global --add safe.directory ${{ github.workspace }}
+          git config --global --add safe.directory ${GITHUB_WORKSPACE}
+          for r in 14 15 16; do
+            git config --global --add safe.directory "${{ github.workspace }}/vendor/postgres-v$r"
+            git config --global --add safe.directory "${GITHUB_WORKSPACE}/vendor/postgres-v$r"
+          done
+
+      - name: Checkout
+        uses: actions/checkout@v4
+        with:
+          submodules: true
+          fetch-depth: 1
+
+      # Some of our rust modules use FFI and need those to be checked
+      - name: Get postgres headers
+        run: make postgres-headers -j$(nproc)
+
+      # cargo hack runs the given cargo subcommand (clippy in this case) for all feature combinations.
+      # This will catch compiler & clippy warnings in all feature combinations.
+      # TODO: use cargo hack for build and test as well, but, that's quite expensive.
+      # NB: keep clippy args in sync with ./run_clippy.sh
+      - run: |
+          CLIPPY_COMMON_ARGS="$( source .neon_clippy_args; echo "$CLIPPY_COMMON_ARGS")"
+          if [ "$CLIPPY_COMMON_ARGS" = "" ]; then
+            echo "No clippy args found in .neon_clippy_args"
+            exit 1
+          fi
+          echo "CLIPPY_COMMON_ARGS=${CLIPPY_COMMON_ARGS}" >> $GITHUB_ENV
+      - name: Run cargo clippy (debug)
+        run: cargo hack --feature-powerset clippy $CLIPPY_COMMON_ARGS
+      - name: Run cargo clippy (release)
+        run: cargo hack --feature-powerset clippy --release $CLIPPY_COMMON_ARGS
+
+      - name: Check documentation generation
+        run: cargo doc --workspace --no-deps --document-private-items
+        env:
+            RUSTDOCFLAGS: "-Dwarnings -Arustdoc::private_intra_doc_links"
+
+      # Use `${{ !cancelled() }}` to run quck tests after the longer clippy run
+      - name: Check formatting
+        if: ${{ !cancelled() }}
+        run: cargo fmt --all -- --check
+
+      # https://github.com/facebookincubator/cargo-guppy/tree/bec4e0eb29dcd1faac70b1b5360267fc02bf830e/tools/cargo-hakari#2-keep-the-workspace-hack-up-to-date-in-ci
+      - name: Check rust dependencies
+        if: ${{ !cancelled() }}
+        run: |
+          cargo hakari generate --diff  # workspace-hack Cargo.toml is up-to-date
+          cargo hakari manage-deps --dry-run  # all workspace crates depend on workspace-hack
+
+      # https://github.com/EmbarkStudios/cargo-deny
+      - name: Check rust licenses/bans/advisories/sources
+        if: ${{ !cancelled() }}
+        run: cargo deny check
+
   gather-rust-build-stats:
     needs: [ check-permissions, build-build-tools-image ]
     if: |
       contains(github.event.pull_request.labels.*.name, 'run-extra-build-stats') ||
       contains(github.event.pull_request.labels.*.name, 'run-extra-build-*') ||
       github.ref_name == 'main'
-    runs-on: [ self-hosted, large ]
+    runs-on: [ self-hosted, gen3, large ]
     container:
       image: ${{ needs.build-build-tools-image.outputs.image }}
       credentials:
@@ -148,6 +348,8 @@ jobs:
 
     env:
       BUILD_TYPE: release
+      # remove the cachepot wrapper and build without crate caches
+      RUSTC_WRAPPER: ""
       # build with incremental compilation produce partial results
       # so do not attempt to cache this build, also disable the incremental compilation
       CARGO_INCREMENTAL: 0
@@ -157,6 +359,7 @@ jobs:
         uses: actions/checkout@v4
         with:
           submodules: true
+          fetch-depth: 1
 
       # Some of our rust modules use FFI and need those to be checked
       - name: Get postgres headers
@@ -166,7 +369,7 @@ 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: cargo build --all --release --timings
 
       - name: Upload the build stats
         id: upload-stats

.github/workflows/periodic_pagebench.yml

@@ -1,155 +0,0 @@
-name: Periodic pagebench performance test on dedicated EC2 machine in eu-central-1 region
-
-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:  '0 18 * * *' # Runs at 6 PM UTC every day
-  workflow_dispatch: # Allows manual triggering of the workflow
-    inputs:
-      commit_hash:
-        type: string
-        description: 'The long neon repo commit hash for the system under test (pageserver) to be tested.'
-        required: false
-        default: ''
-
-defaults:
-  run:
-    shell: bash -euo pipefail {0}
-
-concurrency:
-  group: ${{ github.workflow }}
-  cancel-in-progress: false
-
-jobs:
-  trigger_bench_on_ec2_machine_in_eu_central_1:
-    runs-on: [ self-hosted, small ]
-    container:
-      image: neondatabase/build-tools:pinned
-      credentials:
-        username: ${{ secrets.NEON_DOCKERHUB_USERNAME }}
-        password: ${{ secrets.NEON_DOCKERHUB_PASSWORD }}
-      options: --init
-    timeout-minutes: 360  # Set the timeout to 6 hours
-    env:
-      API_KEY: ${{ secrets.PERIODIC_PAGEBENCH_EC2_RUNNER_API_KEY }}
-      RUN_ID: ${{ github.run_id }}
-      AWS_ACCESS_KEY_ID: ${{ secrets.AWS_EC2_US_TEST_RUNNER_ACCESS_KEY_ID }}
-      AWS_SECRET_ACCESS_KEY : ${{ secrets.AWS_EC2_US_TEST_RUNNER_ACCESS_KEY_SECRET }}
-      AWS_DEFAULT_REGION : "eu-central-1"
-      AWS_INSTANCE_ID : "i-02a59a3bf86bc7e74"
-    steps:
-    # we don't need the neon source code because we run everything remotely
-    # however we still need the local github actions to run the allure step below
-    - uses: actions/checkout@v4
-
-    - name: Show my own (github runner) external IP address - usefull for IP allowlisting
-      run: curl https://ifconfig.me
-
-    - name: Start EC2 instance and wait for the instance to boot up
-      run: |
-        aws ec2 start-instances --instance-ids $AWS_INSTANCE_ID
-        aws ec2 wait instance-running --instance-ids $AWS_INSTANCE_ID
-        sleep 60 # sleep some time to allow cloudinit and our API server to start up
-
-    - name: Determine public IP of the EC2 instance and set env variable EC2_MACHINE_URL_US
-      run: |
-        public_ip=$(aws ec2 describe-instances --instance-ids $AWS_INSTANCE_ID --query 'Reservations[*].Instances[*].PublicIpAddress' --output text)
-        echo "Public IP of the EC2 instance: $public_ip"
-        echo "EC2_MACHINE_URL_US=https://${public_ip}:8443" >> $GITHUB_ENV
-
-    - name: Determine commit hash
-      env:
-        INPUT_COMMIT_HASH: ${{ github.event.inputs.commit_hash }}
-      run: |
-        if [ -z "$INPUT_COMMIT_HASH" ]; then
-          echo "COMMIT_HASH=$(curl -s https://api.github.com/repos/neondatabase/neon/commits/main | jq -r '.sha')" >> $GITHUB_ENV
-        else
-          echo "COMMIT_HASH=$INPUT_COMMIT_HASH" >> $GITHUB_ENV
-        fi
-
-    - name: Start Bench with run_id   
-      run: |
-        curl -k -X 'POST' \
-        "${EC2_MACHINE_URL_US}/start_test/${GITHUB_RUN_ID}" \
-        -H 'accept: application/json' \
-        -H 'Content-Type: application/json' \
-        -H "Authorization: Bearer $API_KEY" \
-        -d "{\"neonRepoCommitHash\": \"${COMMIT_HASH}\"}"
-
-    - name: Poll Test Status
-      id: poll_step
-      run: |
-        status=""
-        while [[ "$status" != "failure" && "$status" != "success" ]]; do
-          response=$(curl -k -X 'GET' \
-          "${EC2_MACHINE_URL_US}/test_status/${GITHUB_RUN_ID}" \
-          -H 'accept: application/json' \
-          -H "Authorization: Bearer $API_KEY")
-          echo "Response: $response"
-          set +x
-          status=$(echo $response | jq -r '.status')
-          echo "Test status: $status"
-          if [[ "$status" == "failure" ]]; then
-            echo "Test failed"
-            exit 1 # Fail the job step if status is failure
-          elif [[ "$status" == "success" || "$status" == "null" ]]; then
-            break
-          elif [[ "$status" == "too_many_runs" ]]; then
-            echo "Too many runs already running"
-            echo "too_many_runs=true" >> "$GITHUB_OUTPUT"
-            exit 1
-          fi
-
-          sleep 60 # Poll every 60 seconds
-        done
-
-    - name: Retrieve Test Logs
-      if: always() && steps.poll_step.outputs.too_many_runs != 'true'
-      run: |
-        curl -k -X 'GET' \
-        "${EC2_MACHINE_URL_US}/test_log/${GITHUB_RUN_ID}" \
-        -H 'accept: application/gzip' \
-        -H "Authorization: Bearer $API_KEY" \
-        --output "test_log_${GITHUB_RUN_ID}.gz"
-    
-    - name: Unzip Test Log and Print it into this job's log
-      if: always() && steps.poll_step.outputs.too_many_runs != 'true'
-      run: |
-        gzip -d "test_log_${GITHUB_RUN_ID}.gz"
-        cat "test_log_${GITHUB_RUN_ID}"
-
-    - name: Create Allure report
-      env:
-        AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_DEV }}
-        AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_KEY_DEV }}
-      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" # dev-staging-stream
-        slack-message: "Periodic pagebench testing on dedicated hardware: ${{ job.status }}\n${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}"
-      env:
-        SLACK_BOT_TOKEN: ${{ secrets.SLACK_BOT_TOKEN }}
-
-    - name: Cleanup Test Resources
-      if: always() 
-      run: |
-        curl -k -X 'POST' \
-        "${EC2_MACHINE_URL_US}/cleanup_test/${GITHUB_RUN_ID}" \
-        -H 'accept: application/json' \
-        -H "Authorization: Bearer $API_KEY" \
-        -d ''
-
-    - name: Stop EC2 instance and wait for the instance to be stopped
-      if: always() && steps.poll_step.outputs.too_many_runs != 'true'
-      run: |
-        aws ec2 stop-instances --instance-ids $AWS_INSTANCE_ID
-        aws ec2 wait instance-stopped --instance-ids $AWS_INSTANCE_ID

.github/workflows/pg-clients.yml

@@ -1,211 +0,0 @@
-name: Test Postgres client libraries
-
-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:  '23 02 * * *' # run once a day, timezone is utc
-  pull_request:
-    paths:
-      - '.github/workflows/pg-clients.yml'
-      - 'test_runner/pg_clients/**'
-      - 'test_runner/logical_repl/**'
-      - 'poetry.lock'
-  workflow_dispatch:
-
-concurrency:
-  group: ${{ github.workflow }}-${{ github.ref_name }}
-  cancel-in-progress: ${{ github.event_name == 'pull_request' }}
-
-defaults:
-  run:
-    shell: bash -euxo pipefail {0}
-
-env:
-  DEFAULT_PG_VERSION: 16
-  PLATFORM: neon-captest-new
-  AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_DEV }}
-  AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_KEY_DEV }}
-  AWS_DEFAULT_REGION: eu-central-1
-
-jobs:
-  check-permissions:
-    if: ${{ !contains(github.event.pull_request.labels.*.name, 'run-no-ci') }}
-    uses: ./.github/workflows/check-permissions.yml
-    with:
-      github-event-name: ${{ github.event_name }}
-
-  check-build-tools-image:
-    needs: [ check-permissions ]
-    uses: ./.github/workflows/check-build-tools-image.yml
-
-  build-build-tools-image:
-    needs: [ check-build-tools-image ]
-    uses: ./.github/workflows/build-build-tools-image.yml
-    with:
-      image-tag: ${{ needs.check-build-tools-image.outputs.image-tag }}
-    secrets: inherit
-
-  test-logical-replication:
-    needs: [ build-build-tools-image ]
-    runs-on: ubuntu-22.04
-
-    container:
-      image: ${{ needs.build-build-tools-image.outputs.image }}
-      credentials:
-        username: ${{ secrets.NEON_DOCKERHUB_USERNAME }}
-        password: ${{ secrets.NEON_DOCKERHUB_PASSWORD }}
-      options: --init --user root
-    services:
-      clickhouse:
-        image: clickhouse/clickhouse-server:24.6.3.64
-        ports:
-          - 9000:9000
-          - 8123:8123
-      zookeeper:
-        image: quay.io/debezium/zookeeper:2.7
-        ports:
-          - 2181:2181
-      kafka:
-        image: quay.io/debezium/kafka:2.7
-        env:
-          ZOOKEEPER_CONNECT: "zookeeper:2181"
-          KAFKA_ADVERTISED_LISTENERS: PLAINTEXT://kafka:9092
-          KAFKA_BROKER_ID: 1
-          KAFKA_OFFSETS_TOPIC_REPLICATION_FACTOR: 1
-          KAFKA_JMX_PORT: 9991
-        ports:
-          - 9092:9092
-      debezium:
-        image: quay.io/debezium/connect:2.7
-        env:
-          BOOTSTRAP_SERVERS: kafka:9092
-          GROUP_ID: 1
-          CONFIG_STORAGE_TOPIC: debezium-config
-          OFFSET_STORAGE_TOPIC: debezium-offset
-          STATUS_STORAGE_TOPIC: debezium-status
-          DEBEZIUM_CONFIG_CONNECTOR_CLASS: io.debezium.connector.postgresql.PostgresConnector
-        ports:
-          - 8083:8083
-    steps:
-      - uses: actions/checkout@v4
-
-      - name: Download Neon artifact
-        uses: ./.github/actions/download
-        with:
-          name: neon-${{ runner.os }}-${{ runner.arch }}-release-artifact
-          path: /tmp/neon/
-          prefix: latest
-
-      - name: Create Neon Project
-        id: create-neon-project
-        uses: ./.github/actions/neon-project-create
-        with:
-          api_key: ${{ secrets.NEON_STAGING_API_KEY }}
-          postgres_version: ${{ env.DEFAULT_PG_VERSION }}
-
-      - name: Run tests
-        uses: ./.github/actions/run-python-test-set
-        with:
-          build_type: remote
-          test_selection: logical_repl
-          run_in_parallel: false
-          extra_params: -m remote_cluster
-          pg_version: ${{ env.DEFAULT_PG_VERSION }}
-        env:
-          BENCHMARK_CONNSTR: ${{ steps.create-neon-project.outputs.dsn }}
-
-      - name: Delete Neon Project
-        if: always()
-        uses: ./.github/actions/neon-project-delete
-        with:
-          project_id: ${{ steps.create-neon-project.outputs.project_id }}
-          api_key: ${{ secrets.NEON_STAGING_API_KEY }}
-
-      - name: Create Allure report
-        if: ${{ !cancelled() }}
-        id: create-allure-report
-        uses: ./.github/actions/allure-report-generate
-        with:
-          store-test-results-into-db: true
-        env:
-          REGRESS_TEST_RESULT_CONNSTR_NEW: ${{ secrets.REGRESS_TEST_RESULT_CONNSTR_NEW }}
-
-      - name: Post to a Slack channel
-        if: github.event.schedule && failure()
-        uses: slackapi/slack-github-action@v1
-        with:
-          channel-id: "C06KHQVQ7U3" # on-call-qa-staging-stream
-          slack-message: |
-            Testing the logical replication: <${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}|${{ job.status }}> (<${{ steps.create-allure-report.outputs.report-url }}|test report>)
-        env:
-          SLACK_BOT_TOKEN: ${{ secrets.SLACK_BOT_TOKEN }}
-
-  test-postgres-client-libs:
-    needs: [ build-build-tools-image ]
-    runs-on: ubuntu-22.04
-
-    container:
-      image: ${{ needs.build-build-tools-image.outputs.image }}
-      credentials:
-        username: ${{ secrets.NEON_DOCKERHUB_USERNAME }}
-        password: ${{ secrets.NEON_DOCKERHUB_PASSWORD }}
-      options: --init --user root
-
-    steps:
-    - uses: actions/checkout@v4
-
-    - name: Download Neon artifact
-      uses: ./.github/actions/download
-      with:
-        name: neon-${{ runner.os }}-${{ runner.arch }}-release-artifact
-        path: /tmp/neon/
-        prefix: latest
-
-    - name: Create Neon Project
-      id: create-neon-project
-      uses: ./.github/actions/neon-project-create
-      with:
-        api_key: ${{ secrets.NEON_STAGING_API_KEY }}
-        postgres_version: ${{ env.DEFAULT_PG_VERSION }}
-
-    - name: Run tests
-      uses: ./.github/actions/run-python-test-set
-      with:
-        build_type: remote
-        test_selection: pg_clients
-        run_in_parallel: false
-        extra_params: -m remote_cluster
-        pg_version: ${{ env.DEFAULT_PG_VERSION }}
-      env:
-        BENCHMARK_CONNSTR: ${{ steps.create-neon-project.outputs.dsn }}
-
-    - name: Delete Neon Project
-      if: always()
-      uses: ./.github/actions/neon-project-delete
-      with:
-        project_id: ${{ steps.create-neon-project.outputs.project_id }}
-        api_key: ${{ secrets.NEON_STAGING_API_KEY }}
-
-    - name: Create Allure report
-      if: ${{ !cancelled() }}
-      id: create-allure-report
-      uses: ./.github/actions/allure-report-generate
-      with:
-        store-test-results-into-db: true
-      env:
-        REGRESS_TEST_RESULT_CONNSTR_NEW: ${{ secrets.REGRESS_TEST_RESULT_CONNSTR_NEW }}
-
-    - name: Post to a Slack channel
-      if: github.event.schedule && failure()
-      uses: slackapi/slack-github-action@v1
-      with:
-        channel-id: "C06KHQVQ7U3" # on-call-qa-staging-stream
-        slack-message: |
-          Testing Postgres clients: <${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}|${{ job.status }}> (<${{ steps.create-allure-report.outputs.report-url }}|test report>)
-      env:
-        SLACK_BOT_TOKEN: ${{ secrets.SLACK_BOT_TOKEN }}

.github/workflows/pg_clients.yml

@@ -0,0 +1,98 @@
+name: Test Postgres client libraries
+
+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:  '23 02 * * *' # run once a day, timezone is utc
+
+  workflow_dispatch:
+
+concurrency:
+  # Allow only one workflow per any non-`main` branch.
+  group: ${{ github.workflow }}-${{ github.ref_name }}-${{ github.ref_name == 'main' && github.sha || 'anysha' }}
+  cancel-in-progress: true
+
+jobs:
+  test-postgres-client-libs:
+    # TODO: switch to gen2 runner, requires docker
+    runs-on: [ ubuntu-latest ]
+
+    env:
+      DEFAULT_PG_VERSION: 14
+      TEST_OUTPUT: /tmp/test_output
+
+    steps:
+    - name: Checkout
+      uses: actions/checkout@v4
+
+    - uses: actions/setup-python@v4
+      with:
+        python-version: 3.9
+
+    - name: Install Poetry
+      uses: snok/install-poetry@v1
+
+    - name: Cache poetry deps
+      uses: actions/cache@v4
+      with:
+        path: ~/.cache/pypoetry/virtualenvs
+        key: v2-${{ runner.os }}-python-deps-ubunutu-latest-${{ hashFiles('poetry.lock') }}
+
+    - name: Install Python deps
+      shell: bash -euxo pipefail {0}
+      run: ./scripts/pysync
+
+    - name: Create Neon Project
+      id: create-neon-project
+      uses: ./.github/actions/neon-project-create
+      with:
+        api_key: ${{ secrets.NEON_STAGING_API_KEY }}
+        postgres_version: ${{ env.DEFAULT_PG_VERSION }}
+
+    - name: Run pytest
+      env:
+        REMOTE_ENV: 1
+        BENCHMARK_CONNSTR: ${{ steps.create-neon-project.outputs.dsn }}
+        POSTGRES_DISTRIB_DIR: /tmp/neon/pg_install
+      shell: bash -euxo pipefail {0}
+      run: |
+        # Test framework expects we have psql binary;
+        # but since we don't really need it in this test, let's mock it
+        mkdir -p "$POSTGRES_DISTRIB_DIR/v${DEFAULT_PG_VERSION}/bin" && touch "$POSTGRES_DISTRIB_DIR/v${DEFAULT_PG_VERSION}/bin/psql";
+        ./scripts/pytest \
+          --junitxml=$TEST_OUTPUT/junit.xml \
+          --tb=short \
+          --verbose \
+          -m "remote_cluster" \
+          -rA "test_runner/pg_clients"
+
+    - name: Delete Neon Project
+      if: ${{ always() }}
+      uses: ./.github/actions/neon-project-delete
+      with:
+        project_id: ${{ steps.create-neon-project.outputs.project_id }}
+        api_key: ${{ secrets.NEON_STAGING_API_KEY }}
+
+    # We use GitHub's action upload-artifact because `ubuntu-latest` doesn't have configured AWS CLI.
+    # It will be fixed after switching to gen2 runner
+    - name: Upload python test logs
+      if: always()
+      uses: actions/upload-artifact@v4
+      with:
+        retention-days: 7
+        name: python-test-pg_clients-${{ runner.os }}-stage-logs
+        path: ${{ env.TEST_OUTPUT }}
+
+    - name: Post to a Slack channel
+      if: ${{ github.event.schedule && failure() }}
+      uses: slackapi/slack-github-action@v1
+      with:
+        channel-id: "C033QLM5P7D" # dev-staging-stream
+        slack-message: "Testing Postgres clients: ${{ job.status }}\n${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}"
+      env:
+        SLACK_BOT_TOKEN: ${{ secrets.SLACK_BOT_TOKEN }}

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

@@ -7,20 +7,12 @@ on:
         description: 'Source tag'
         required: true
         type: string
-      force:
-        description: 'Force the image to be pinned'
-        default: false
-        type: boolean
   workflow_call:
     inputs:
       from-tag:
         description: 'Source tag'
         required: true
         type: string
-      force:
-        description: 'Force the image to be pinned'
-        default: false
-        type: boolean
 
 defaults:
   run:
@@ -28,20 +20,16 @@ defaults:
 
 concurrency:
   group: pin-build-tools-image-${{ inputs.from-tag }}
-  cancel-in-progress: false
 
-# No permission for GITHUB_TOKEN by default; the **minimal required** set of permissions should be granted in each job.
 permissions: {}
 
-env:
-  FROM_TAG: ${{ inputs.from-tag }}
-  TO_TAG: pinned
-
 jobs:
-  check-manifests:
-    runs-on: ubuntu-22.04
-    outputs:
-      skip: ${{ steps.check-manifests.outputs.skip }}
+  tag-image:
+    runs-on: ubuntu-latest
+
+    env:
+      FROM_TAG: ${{ inputs.from-tag }}
+      TO_TAG: pinned
 
     steps:
       - name: Check if we really need to pin the image
@@ -58,44 +46,27 @@ jobs:
 
           echo "skip=${skip}" | tee -a $GITHUB_OUTPUT
 
-  tag-image:
-    needs: check-manifests
-
-    # use format(..) to catch both inputs.force = true AND inputs.force = 'true'
-    if: needs.check-manifests.outputs.skip == 'false' || format('{0}', inputs.force) == 'true'
-
-    runs-on: ubuntu-22.04
-
-    permissions:
-      id-token: write # for `azure/login`
-
-    steps:
       - uses: docker/login-action@v3
-
+        if: steps.check-manifests.outputs.skip == 'false'
         with:
           username: ${{ secrets.NEON_DOCKERHUB_USERNAME }}
           password: ${{ secrets.NEON_DOCKERHUB_PASSWORD }}
 
+      - name: Tag build-tools with `${{ env.TO_TAG }}` in Docker Hub
+        if: steps.check-manifests.outputs.skip == 'false'
+        run: |
+          docker buildx imagetools create -t neondatabase/build-tools:${TO_TAG} \
+                                             neondatabase/build-tools:${FROM_TAG}
+
       - uses: docker/login-action@v3
+        if: steps.check-manifests.outputs.skip == 'false'
         with:
           registry: 369495373322.dkr.ecr.eu-central-1.amazonaws.com
           username: ${{ secrets.AWS_ACCESS_KEY_DEV }}
           password: ${{ secrets.AWS_SECRET_KEY_DEV }}
 
-      - name: Azure login
-        uses: azure/login@6c251865b4e6290e7b78be643ea2d005bc51f69a  # @v2.1.1
-        with:
-          client-id: ${{ secrets.AZURE_DEV_CLIENT_ID }}
-          tenant-id: ${{ secrets.AZURE_TENANT_ID }}
-          subscription-id: ${{ secrets.AZURE_DEV_SUBSCRIPTION_ID }}
-
-      - name: Login to ACR
-        run: |
-          az acr login --name=neoneastus2
-
-      - name: Tag build-tools with `${{ env.TO_TAG }}` in Docker Hub, ECR, and ACR
+      - name: Tag build-tools with `${{ env.TO_TAG }}` in ECR
+        if: steps.check-manifests.outputs.skip == 'false'
         run: |
           docker buildx imagetools create -t 369495373322.dkr.ecr.eu-central-1.amazonaws.com/build-tools:${TO_TAG} \
-                                          -t neoneastus2.azurecr.io/neondatabase/build-tools:${TO_TAG} \
-                                          -t neondatabase/build-tools:${TO_TAG} \
                                              neondatabase/build-tools:${FROM_TAG}

.github/workflows/release-notify.yml

@@ -19,7 +19,7 @@ on:
 
 jobs:
   notify:
-    runs-on: ubuntu-22.04
+    runs-on: [ ubuntu-latest ]
 
     steps:
       - uses: neondatabase/dev-actions/release-pr-notify@main

.github/workflows/release.yml

@@ -26,7 +26,7 @@ defaults:
 jobs:
   create-storage-release-branch:
     if: ${{ github.event.schedule == '0 6 * * MON' || format('{0}', inputs.create-storage-release-branch) == 'true' }}
-    runs-on: ubuntu-22.04
+    runs-on: ubuntu-latest
 
     permissions:
       contents: write # for `git push`
@@ -52,22 +52,20 @@ jobs:
       env:
         GH_TOKEN: ${{ secrets.CI_ACCESS_TOKEN }}
       run: |
-        TITLE="Storage & Compute release ${RELEASE_DATE}"
-
         cat << EOF > body.md
-          ## ${TITLE}
+          ## Release ${RELEASE_DATE}
 
           **Please merge this Pull Request using 'Create a merge commit' button**
         EOF
 
-        gh pr create --title "${TITLE}" \
+        gh pr create --title "Release ${RELEASE_DATE}" \
                      --body-file "body.md" \
                      --head "${RELEASE_BRANCH}" \
                      --base "release"
 
   create-proxy-release-branch:
     if: ${{ github.event.schedule == '0 6 * * THU' || format('{0}', inputs.create-proxy-release-branch) == 'true' }}
-    runs-on: ubuntu-22.04
+    runs-on: ubuntu-latest
 
     permissions:
       contents: write # for `git push`
@@ -93,15 +91,13 @@ jobs:
       env:
         GH_TOKEN: ${{ secrets.CI_ACCESS_TOKEN }}
       run: |
-        TITLE="Proxy release ${RELEASE_DATE}"
-
         cat << EOF > body.md
-          ## ${TITLE}
+          ## Proxy release ${RELEASE_DATE}
 
           **Please merge this Pull Request using 'Create a merge commit' button**
         EOF
 
-        gh pr create --title "${TITLE}" \
+        gh pr create --title "Proxy release ${RELEASE_DATE}" \
                      --body-file "body.md" \
                      --head "${RELEASE_BRANCH}" \
                      --base "release-proxy"

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

@@ -13,11 +13,13 @@ defaults:
 env:
   # A concurrency group that we use for e2e-tests runs, matches `concurrency.group` above with `github.repository` as a prefix
   E2E_CONCURRENCY_GROUP: ${{ github.repository }}-e2e-tests-${{ github.ref_name }}-${{ github.ref_name == 'main' && github.sha || 'anysha' }}
+  AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_DEV }}
+  AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_KEY_DEV }}
 
 jobs:
   cancel-previous-e2e-tests:
     if: github.event_name == 'pull_request'
-    runs-on: ubuntu-22.04
+    runs-on: ubuntu-latest
 
     steps:
       - name: Cancel previous e2e-tests runs for this PR
@@ -29,7 +31,7 @@ jobs:
               --field concurrency_group="${{ env.E2E_CONCURRENCY_GROUP }}"
 
   tag:
-    runs-on: ubuntu-22.04
+    runs-on: [ ubuntu-latest ]
     outputs:
       build-tag: ${{ steps.build-tag.outputs.tag }}
 
@@ -60,88 +62,58 @@ jobs:
 
   trigger-e2e-tests:
     needs: [ tag ]
-    runs-on: ubuntu-22.04
+    runs-on: [ self-hosted, gen3, small ]
     env:
-      EVENT_ACTION: ${{ github.event.action }}
-      GH_TOKEN: ${{ secrets.CI_ACCESS_TOKEN }}
       TAG: ${{ needs.tag.outputs.build-tag }}
+    container:
+      image: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/base:pinned
+      options: --init
     steps:
-      - name: Wait for `promote-images` job to finish
-        # It's important to have a timeout here, the script in the step can run infinitely
-        timeout-minutes: 60
+      - name: check if ecr image are present
         run: |
-          if [ "${GITHUB_EVENT_NAME}" != "pull_request" ] || [ "${EVENT_ACTION}" != "ready_for_review" ]; then
-            exit 0
-          fi
-
-          # For PRs we use the run id as the tag
-          BUILD_AND_TEST_RUN_ID=${TAG}
-          while true; do
-            conclusion=$(gh run --repo ${GITHUB_REPOSITORY} view ${BUILD_AND_TEST_RUN_ID} --json jobs --jq '.jobs[] | select(.name == "promote-images") | .conclusion')
-            case "$conclusion" in
-              success)
-                break
-                ;;
-              failure | cancelled | skipped)
-                echo "The 'promote-images' job didn't succeed: '${conclusion}'. Exiting..."
-                exit 1
-                ;;
-              *)
-                echo "The 'promote-images' hasn't succeed yet. Waiting..."
-                sleep 60
-                ;;
-            esac
+          for REPO in neon compute-tools compute-node-v14 vm-compute-node-v14 compute-node-v15 vm-compute-node-v15 compute-node-v16 vm-compute-node-v16; do
+            OUTPUT=$(aws ecr describe-images --repository-name ${REPO} --region eu-central-1 --query "imageDetails[?imageTags[?contains(@, '${TAG}')]]" --output text)
+            if [ "$OUTPUT" == "" ]; then
+              echo "$REPO with image tag $TAG not found" >> $GITHUB_OUTPUT
+              exit 1
+            fi
           done
 
-      - name: Set e2e-platforms
-        id: e2e-platforms
-        env:
-          PR_NUMBER: ${{ github.event.pull_request.number }}
-          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-        run: |
-          # Default set of platforms to run e2e tests on
-          platforms='["docker", "k8s"]'
-
-          # If the PR changes vendor/, pgxn/ or libs/vm_monitor/ directories, or Dockerfile.compute-node, add k8s-neonvm to the list of platforms.
-          # If the workflow run is not a pull request, add k8s-neonvm to the list.
-          if [ "$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)
-                  platforms=$(echo "${platforms}" | jq --compact-output '. += ["k8s-neonvm"] | unique')
-                  ;;
-                *)
-                  # no-op
-                  ;;
-              esac
-            done
-          else
-            platforms=$(echo "${platforms}" | jq --compact-output '. += ["k8s-neonvm"] | unique')
-          fi
-
-          echo "e2e-platforms=${platforms}" | tee -a $GITHUB_OUTPUT
-
       - name: Set PR's status to pending and request a remote CI test
-        env:
-          E2E_PLATFORMS: ${{ steps.e2e-platforms.outputs.e2e-platforms }}
-          COMMIT_SHA: ${{ github.event.pull_request.head.sha || github.sha }}
-          GH_TOKEN: ${{ secrets.CI_ACCESS_TOKEN }}
         run: |
-          REMOTE_REPO="${GITHUB_REPOSITORY_OWNER}/cloud"
+          # For pull requests, GH Actions set "github.sha" variable to point at a fake merge commit
+          # but we need to use a real sha of a latest commit in the PR's branch for the e2e job,
+          # to place a job run status update later.
+          COMMIT_SHA=${{ github.event.pull_request.head.sha }}
+          # For non-PR kinds of runs, the above will produce an empty variable, pick the original sha value for those
+          COMMIT_SHA=${COMMIT_SHA:-${{ github.sha }}}
 
-          gh api "/repos/${GITHUB_REPOSITORY}/statuses/${COMMIT_SHA}" \
-            --method POST \
-            --raw-field "state=pending" \
-            --raw-field "description=[$REMOTE_REPO] Remote CI job is about to start" \
-            --raw-field "context=neon-cloud-e2e"
+          REMOTE_REPO="${{ github.repository_owner }}/cloud"
 
-          gh workflow --repo ${REMOTE_REPO} \
-            run testing.yml \
-              --ref "main" \
-              --raw-field "ci_job_name=neon-cloud-e2e" \
-              --raw-field "commit_hash=$COMMIT_SHA" \
-              --raw-field "remote_repo=${GITHUB_REPOSITORY}" \
-              --raw-field "storage_image_tag=${TAG}" \
-              --raw-field "compute_image_tag=${TAG}" \
-              --raw-field "concurrency_group=${E2E_CONCURRENCY_GROUP}" \
-              --raw-field "e2e-platforms=${E2E_PLATFORMS}"
+          curl -f -X POST \
+          https://api.github.com/repos/${{ github.repository }}/statuses/$COMMIT_SHA \
+          -H "Accept: application/vnd.github.v3+json" \
+          --user "${{ secrets.CI_ACCESS_TOKEN }}" \
+          --data \
+            "{
+              \"state\": \"pending\",
+              \"context\": \"neon-cloud-e2e\",
+              \"description\": \"[$REMOTE_REPO] Remote CI job is about to start\"
+            }"
+
+          curl -f -X POST \
+          https://api.github.com/repos/$REMOTE_REPO/actions/workflows/testing.yml/dispatches \
+          -H "Accept: application/vnd.github.v3+json" \
+          --user "${{ secrets.CI_ACCESS_TOKEN }}" \
+          --data \
+            "{
+              \"ref\": \"main\",
+              \"inputs\": {
+                \"ci_job_name\": \"neon-cloud-e2e\",
+                \"commit_hash\": \"$COMMIT_SHA\",
+                \"remote_repo\": \"${{ github.repository }}\",
+                \"storage_image_tag\": \"${TAG}\",
+                \"compute_image_tag\": \"${TAG}\",
+                \"concurrency_group\": \"${{ env.E2E_CONCURRENCY_GROUP }}\"
+              }
+            }"

.neon_clippy_args

@@ -1,5 +1,4 @@
 # * `-A unknown_lints` – do not warn about unknown lint suppressions
 #                        that people with newer toolchains might use
 # * `-D warnings`      - fail on any warnings (`cargo` returns non-zero exit status)
-# * `-D clippy::todo`  - don't let `todo!()` slip into `main`
-export CLIPPY_COMMON_ARGS="--locked --workspace --all-targets -- -A unknown_lints -D warnings -D clippy::todo"
+export CLIPPY_COMMON_ARGS="--locked --workspace --all-targets -- -A unknown_lints -D warnings"

CODEOWNERS

@@ -1,13 +1,13 @@
 /compute_tools/ @neondatabase/control-plane @neondatabase/compute
-/storage_controller @neondatabase/storage
+/control_plane/attachment_service @neondatabase/storage
 /libs/pageserver_api/ @neondatabase/storage
-/libs/postgres_ffi/ @neondatabase/compute @neondatabase/storage
+/libs/postgres_ffi/ @neondatabase/compute @neondatabase/safekeepers
 /libs/remote_storage/ @neondatabase/storage
-/libs/safekeeper_api/ @neondatabase/storage
+/libs/safekeeper_api/ @neondatabase/safekeepers
 /libs/vm_monitor/ @neondatabase/autoscaling
 /pageserver/ @neondatabase/storage
 /pgxn/ @neondatabase/compute
-/pgxn/neon/ @neondatabase/compute @neondatabase/storage
+/pgxn/neon/ @neondatabase/compute @neondatabase/safekeepers
 /proxy/ @neondatabase/proxy
-/safekeeper/ @neondatabase/storage
+/safekeeper/ @neondatabase/safekeepers
 /vendor/ @neondatabase/compute

Cargo.toml

@@ -3,7 +3,7 @@ resolver = "2"
 members = [
     "compute_tools",
     "control_plane",
-    "control_plane/storcon_cli",
+    "control_plane/attachment_service",
     "pageserver",
     "pageserver/compaction",
     "pageserver/ctl",
@@ -12,10 +12,9 @@ members = [
     "proxy",
     "safekeeper",
     "storage_broker",
-    "storage_controller",
-    "storage_controller/client",
-    "storage_scrubber",
+    "s3_scrubber",
     "workspace_hack",
+    "trace",
     "libs/compute_api",
     "libs/pageserver_api",
     "libs/postgres_ffi",
@@ -41,31 +40,28 @@ license = "Apache-2.0"
 
 ## All dependency versions, used in the project
 [workspace.dependencies]
-ahash = "0.8"
 anyhow = { version = "1.0", features = ["backtrace"] }
 arc-swap = "1.6"
 async-compression = { version = "0.4.0", features = ["tokio", "gzip", "zstd"] }
-atomic-take = "1.1.0"
-azure_core = { version = "0.19", default-features = false, features = ["enable_reqwest_rustls", "hmac_rust"] }
-azure_identity = { version = "0.19", default-features = false, features = ["enable_reqwest_rustls"] }
-azure_storage = { version = "0.19", default-features = false, features = ["enable_reqwest_rustls"] }
-azure_storage_blobs = { version = "0.19", default-features = false, features = ["enable_reqwest_rustls"] }
+azure_core = "0.18"
+azure_identity = "0.18"
+azure_storage = "0.18"
+azure_storage_blobs = "0.18"
 flate2 = "1.0.26"
 async-stream = "0.3"
 async-trait = "0.1"
-aws-config = { version = "1.3", default-features = false, features=["rustls"] }
-aws-sdk-s3 = "1.26"
+aws-config = { version = "1.1.4", default-features = false, features=["rustls"] }
+aws-sdk-s3 = "1.14"
 aws-sdk-iam = "1.15.0"
-aws-smithy-async = { version = "1.2.1", default-features = false, features=["rt-tokio"] }
-aws-smithy-types = "1.1.9"
-aws-credential-types = "1.2.0"
-aws-sigv4 = { version = "1.2.1", features = ["sign-http"] }
-aws-types = "1.2.0"
+aws-smithy-async = { version = "1.1.4", default-features = false, features=["rt-tokio"] }
+aws-smithy-types = "1.1.4"
+aws-credential-types = "1.1.4"
+aws-sigv4 = { version = "1.2.0", features = ["sign-http"] }
+aws-types = "1.1.7"
 axum = { version = "0.6.20", features = ["ws"] }
 base64 = "0.13.0"
 bincode = "1.3"
-bindgen = "0.70"
-bit_field = "0.10.2"
+bindgen = "0.65"
 bstr = "1.0"
 byteorder = "1.4"
 bytes = "1.0"
@@ -73,10 +69,9 @@ camino = "1.1.6"
 cfg-if = "1.0.0"
 chrono = { version = "0.4", default-features = false, features = ["clock"] }
 clap = { version = "4.0", features = ["derive"] }
-comfy-table = "7.1"
+comfy-table = "6.1"
 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"
@@ -84,13 +79,13 @@ enum-map = "2.4.2"
 enumset = "1.0.12"
 fail = "0.5.0"
 fallible-iterator = "0.2"
-framed-websockets = { version = "0.1.0", git = "https://github.com/neondatabase/framed-websockets" }
+fs2 = "0.4.3"
 futures = "0.3"
 futures-core = "0.3"
 futures-util = "0.3"
 git-version = "0.3"
-hashbrown = "0.14"
-hashlink = "0.9.1"
+hashbrown = "0.13"
+hashlink = "0.8.4"
 hdrhistogram = "7.5.2"
 hex = "0.4"
 hex-literal = "0.4"
@@ -101,42 +96,41 @@ http-types = { version = "2", default-features = false }
 humantime = "2.1"
 humantime-serde = "1.1.1"
 hyper = "0.14"
-tokio-tungstenite = "0.20.0"
-indexmap = "2"
-indoc = "2"
+hyper-tungstenite = "0.11"
 inotify = "0.10.2"
 ipnet = "2.9.0"
 itertools = "0.10"
 jsonwebtoken = "9"
 lasso = "0.7"
+leaky-bucket = "1.0.1"
 libc = "0.2"
 md5 = "0.7.0"
-measured = { version = "0.0.22", features=["lasso"] }
-measured-process = { version = "0.0.22" }
+measured = { version = "0.0.13", features=["default", "lasso"] }
 memoffset = "0.8"
-nix = { version = "0.27", features = ["dir", "fs", "process", "socket", "signal", "poll"] }
+native-tls = "0.2"
+nix = { version = "0.27", features = ["fs", "process", "socket", "signal", "poll"] }
 notify = "6.0.0"
 num_cpus = "1.15"
 num-traits = "0.2.15"
 once_cell = "1.13"
 opentelemetry = "0.20.0"
-opentelemetry-otlp = { version = "0.13.0", default-features=false, features = ["http-proto", "trace", "http", "reqwest-client"] }
+opentelemetry-otlp = { version = "0.13.0", default_features=false, features = ["http-proto", "trace", "http", "reqwest-client"] }
 opentelemetry-semantic-conventions = "0.12.0"
 parking_lot = "0.12"
-parquet = { version = "53", default-features = false, features = ["zstd"] }
-parquet_derive = "53"
+parquet = { version = "49.0.0", default-features = false, features = ["zstd"] }
+parquet_derive = "49.0.0"
 pbkdf2 = { version = "0.12.1", features = ["simple", "std"] }
 pin-project-lite = "0.2"
-procfs = "0.16"
-prometheus = {version = "0.13", default-features=false, features = ["process"]} # removes protobuf dependency
+procfs = "0.14"
+prometheus = {version = "0.13", default_features=false, features = ["process"]} # removes protobuf dependency
 prost = "0.11"
 rand = "0.8"
 redis = { version = "0.25.2", features = ["tokio-rustls-comp", "keep-alive"] }
 regex = "1.10.2"
-reqwest = { version = "0.12", default-features = false, features = ["rustls-tls"] }
-reqwest-tracing = { version = "0.5", features = ["opentelemetry_0_20"] }
-reqwest-middleware = "0.3.0"
-reqwest-retry = "0.5"
+reqwest = { version = "0.11", default-features = false, features = ["rustls-tls"] }
+reqwest-tracing = { version = "0.4.7", features = ["opentelemetry_0_20"] }
+reqwest-middleware = "0.2.0"
+reqwest-retry = "0.2.2"
 routerify = "3"
 rpds = "0.13"
 rustc-hash = "1.1.0"
@@ -146,8 +140,7 @@ rustls-split = "0.3"
 scopeguard = "1.1"
 sysinfo = "0.29.2"
 sd-notify = "0.4.1"
-send-future = "0.1.0"
-sentry = { version = "0.32", default-features = false, features = ["backtrace", "contexts", "panic", "rustls", "reqwest" ] }
+sentry = { version = "0.31", default-features = false, features = ["backtrace", "contexts", "panic", "rustls", "reqwest" ] }
 serde = { version = "1.0", features = ["derive"] }
 serde_json = "1"
 serde_path_to_error = "0.1"
@@ -158,14 +151,14 @@ signal-hook = "0.3"
 smallvec = "1.11"
 smol_str = { version = "0.2.0", features = ["serde"] }
 socket2 = "0.5"
-strum = "0.26"
-strum_macros = "0.26"
+strum = "0.24"
+strum_macros = "0.24"
 "subtle"  = "2.5.0"
-svg_fmt = "0.4.3"
+svg_fmt = "0.4.1"
 sync_wrapper = "0.1.2"
 tar = "0.4"
 task-local-extensions = "0.1.4"
-test-context = "0.3"
+test-context = "0.1"
 thiserror = "1.0"
 tikv-jemallocator = "0.5"
 tikv-jemalloc-ctl = "0.5"
@@ -177,45 +170,34 @@ tokio-rustls = "0.25"
 tokio-stream = "0.1"
 tokio-tar = "0.3"
 tokio-util = { version = "0.7.10", features = ["io", "rt"] }
-toml = "0.8"
-toml_edit = "0.22"
+toml = "0.7"
+toml_edit = "0.19"
 tonic = {version = "0.9", features = ["tls", "tls-roots"]}
-tower-service = "0.3.2"
 tracing = "0.1"
 tracing-error = "0.2.0"
-tracing-opentelemetry = "0.21.0"
-tracing-subscriber = { version = "0.3", default-features = false, features = ["smallvec", "fmt", "tracing-log", "std", "env-filter", "json"] }
-try-lock = "0.2.5"
+tracing-opentelemetry = "0.20.0"
+tracing-subscriber = { version = "0.3", default_features = false, features = ["smallvec", "fmt", "tracing-log", "std", "env-filter", "json"] }
 twox-hash = { version = "1.6.3", default-features = false }
-typed-json = "0.1"
 url = "2.2"
 urlencoding = "2.1"
 uuid = { version = "1.6.1", features = ["v4", "v7", "serde"] }
 walkdir = "2.3.2"
-rustls-native-certs = "0.7"
+webpki-roots = "0.25"
 x509-parser = "0.15"
-whoami = "1.5.1"
 
 ## TODO replace this with tracing
 env_logger = "0.10"
 log = "0.4"
 
 ## Libraries from neondatabase/ git forks, ideally with changes to be upstreamed
+postgres = { git = "https://github.com/neondatabase/rust-postgres.git", branch="neon" }
+postgres-native-tls = { git = "https://github.com/neondatabase/rust-postgres.git", branch="neon" }
+postgres-protocol = { git = "https://github.com/neondatabase/rust-postgres.git", branch="neon" }
+postgres-types = { git = "https://github.com/neondatabase/rust-postgres.git", branch="neon" }
+tokio-postgres = { git = "https://github.com/neondatabase/rust-postgres.git", branch="neon" }
 
-# We want to use the 'neon' branch for these, but there's currently one
-# incompatible change on the branch. See:
-#
-# - PR #8076 which contained changes that depended on the new changes in
-#   the rust-postgres crate, and
-# - PR #8654 which reverted those changes and made the code in proxy incompatible
-#   with the tip of the 'neon' branch again.
-#
-# When those proxy changes are re-applied (see PR #8747), we can switch using
-# the tip of the 'neon' branch again.
-postgres = { git = "https://github.com/neondatabase/rust-postgres.git", rev = "20031d7a9ee1addeae6e0968e3899ae6bf01cee2" }
-postgres-protocol = { git = "https://github.com/neondatabase/rust-postgres.git", rev = "20031d7a9ee1addeae6e0968e3899ae6bf01cee2" }
-postgres-types = { git = "https://github.com/neondatabase/rust-postgres.git", rev = "20031d7a9ee1addeae6e0968e3899ae6bf01cee2" }
-tokio-postgres = { git = "https://github.com/neondatabase/rust-postgres.git", rev = "20031d7a9ee1addeae6e0968e3899ae6bf01cee2" }
+## Other git libraries
+heapless = { default-features=false, features=[], git = "https://github.com/japaric/heapless.git", rev = "644653bf3b831c6bb4963be2de24804acf5e5001" } # upstream release pending
 
 ## Local libraries
 compute_api = { version = "0.1", path = "./libs/compute_api/" }
@@ -232,7 +214,6 @@ remote_storage = { version = "0.1", path = "./libs/remote_storage/" }
 safekeeper_api = { version = "0.1", path = "./libs/safekeeper_api" }
 desim = { version = "0.1", path = "./libs/desim" }
 storage_broker = { version = "0.1", path = "./storage_broker/" } # Note: main broker code is inside the binary crate, so linking with the library shouldn't be heavy.
-storage_controller_client = { path = "./storage_controller/client" }
 tenant_size_model = { version = "0.1", path = "./libs/tenant_size_model/" }
 tracing-utils = { version = "0.1", path = "./libs/tracing-utils/" }
 utils = { version = "0.1", path = "./libs/utils/" }
@@ -251,8 +232,13 @@ tonic-build = "0.9"
 
 [patch.crates-io]
 
-# Needed to get `tokio-postgres-rustls` to depend on our fork.
-tokio-postgres = { git = "https://github.com/neondatabase/rust-postgres.git", rev = "20031d7a9ee1addeae6e0968e3899ae6bf01cee2" }
+# This is only needed for proxy's tests.
+# TODO: we should probably fork `tokio-postgres-rustls` instead.
+tokio-postgres = { git = "https://github.com/neondatabase/rust-postgres.git", branch="neon" }
+
+# bug fixes for UUID
+parquet = { git = "https://github.com/neondatabase/arrow-rs", branch = "neon-fix-bugs" }
+parquet_derive = { git = "https://github.com/neondatabase/arrow-rs", branch = "neon-fix-bugs" }
 
 ################# Binary contents sections
 

Dockerfile

@@ -17,7 +17,7 @@ COPY --chown=nonroot pgxn pgxn
 COPY --chown=nonroot Makefile Makefile
 COPY --chown=nonroot scripts/ninstall.sh scripts/ninstall.sh
 
-ENV BUILD_TYPE=release
+ENV BUILD_TYPE release
 RUN set -e \
     && mold -run make -j $(nproc) -s neon-pg-ext \
     && rm -rf pg_install/build \
@@ -29,15 +29,25 @@ WORKDIR /home/nonroot
 ARG GIT_VERSION=local
 ARG BUILD_TAG
 
+# Enable https://github.com/paritytech/cachepot to cache Rust crates' compilation results in Docker builds.
+# Set up cachepot to use an AWS S3 bucket for cache results, to reuse it between `docker build` invocations.
+# cachepot falls back to local filesystem if S3 is misconfigured, not failing the build
+ARG RUSTC_WRAPPER=cachepot
+ENV AWS_REGION=eu-central-1
+ENV CACHEPOT_S3_KEY_PREFIX=cachepot
+ARG CACHEPOT_BUCKET=neon-github-dev
+#ARG AWS_ACCESS_KEY_ID
+#ARG AWS_SECRET_ACCESS_KEY
+
 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/v16/lib                       pg_install/v16/lib
 COPY --chown=nonroot . .
 
-ARG ADDITIONAL_RUSTFLAGS
+# Show build caching stats to check if it was used in the end.
+# Has to be the part of the same RUN since cachepot daemon is killed in the end of this RUN, losing the compilation stats.
 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 \
+    && RUSTFLAGS="-Clinker=clang -Clink-arg=-fuse-ld=mold -Clink-arg=-Wl,--no-rosegment" cargo build  \
       --bin pg_sni_router  \
       --bin pageserver  \
       --bin pagectl  \
@@ -46,8 +56,8 @@ RUN set -e \
       --bin storage_controller  \
       --bin proxy  \
       --bin neon_local \
-      --bin storage_scrubber \
-      --locked --release
+      --locked --release \
+    && cachepot -s
 
 # Build final image
 #
@@ -59,6 +69,8 @@ RUN set -e \
     && apt install -y \
         libreadline-dev \
         libseccomp-dev \
+        libicu67 \
+        openssl \
         ca-certificates \
     && rm -rf /var/lib/apt/lists/* /tmp/* /var/tmp/* \
     && useradd -d /data neon \
@@ -72,7 +84,6 @@ COPY --from=build --chown=neon:neon /home/nonroot/target/release/storage_broker
 COPY --from=build --chown=neon:neon /home/nonroot/target/release/storage_controller  /usr/local/bin
 COPY --from=build --chown=neon:neon /home/nonroot/target/release/proxy               /usr/local/bin
 COPY --from=build --chown=neon:neon /home/nonroot/target/release/neon_local          /usr/local/bin
-COPY --from=build --chown=neon:neon /home/nonroot/target/release/storage_scrubber    /usr/local/bin
 
 COPY --from=pg-build /home/nonroot/pg_install/v14 /usr/local/v14/
 COPY --from=pg-build /home/nonroot/pg_install/v15 /usr/local/v15/
@@ -81,25 +92,20 @@ 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.
 # Now, when `docker run ... pageserver` is run, it can start without errors, yet will have some default dummy values.
-RUN mkdir -p /data/.neon/ && \
-  echo "id=1234" > "/data/.neon/identity.toml" && \
-  echo "broker_endpoint='http://storage_broker:50051'\n" \
-       "pg_distrib_dir='/usr/local/'\n" \
-       "listen_pg_addr='0.0.0.0:6400'\n" \
-       "listen_http_addr='0.0.0.0:9898'\n" \
-       "availability_zone='local'\n" \
-  > /data/.neon/pageserver.toml && \
-  chown -R neon:neon /data/.neon
+RUN mkdir -p /data/.neon/ && chown -R neon:neon /data/.neon/ \
+    && /usr/local/bin/pageserver -D /data/.neon/ --init \
+       -c "id=1234" \
+       -c "broker_endpoint='http://storage_broker:50051'" \
+       -c "pg_distrib_dir='/usr/local/'" \
+       -c "listen_pg_addr='0.0.0.0:6400'" \
+       -c "listen_http_addr='0.0.0.0:9898'"
 
 # 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/v16/lib
 
 
 VOLUME ["/data"]
 USER neon
 EXPOSE 6400
 EXPOSE 9898
-
-CMD ["/usr/local/bin/pageserver", "-D", "/data/.neon"]
-

Dockerfile.build-tools

@@ -1,13 +1,5 @@
 FROM debian:bullseye-slim
 
-# Use ARG as a build-time environment variable here to allow.
-# It's not supposed to be set outside.
-# Alternatively it can be obtained using the following command
-# ```
-# . /etc/os-release && echo "${VERSION_CODENAME}"
-# ```
-ARG DEBIAN_VERSION_CODENAME=bullseye
-
 # Add nonroot user
 RUN useradd -ms /bin/bash nonroot -b /home
 SHELL ["/bin/bash", "-c"]
@@ -34,6 +26,7 @@ RUN set -e \
         liblzma-dev \
         libncurses5-dev \
         libncursesw5-dev \
+        libpq-dev \
         libreadline-dev \
         libseccomp-dev \
         libsqlite3-dev \
@@ -58,40 +51,29 @@ RUN set -e \
     && rm -rf /var/lib/apt/lists/* /tmp/* /var/tmp/*
 
 # protobuf-compiler (protoc)
-ENV PROTOC_VERSION=25.1
+ENV PROTOC_VERSION 25.1
 RUN curl -fsSL "https://github.com/protocolbuffers/protobuf/releases/download/v${PROTOC_VERSION}/protoc-${PROTOC_VERSION}-linux-$(uname -m | sed 's/aarch64/aarch_64/g').zip" -o "protoc.zip" \
     && unzip -q protoc.zip -d protoc \
     && mv protoc/bin/protoc /usr/local/bin/protoc \
     && mv protoc/include/google /usr/local/include/google \
     && rm -rf protoc.zip protoc
 
-# s5cmd
-ENV S5CMD_VERSION=2.2.2
-RUN curl -sL "https://github.com/peak/s5cmd/releases/download/v${S5CMD_VERSION}/s5cmd_${S5CMD_VERSION}_Linux-$(uname -m | sed 's/x86_64/64bit/g' | sed 's/aarch64/arm64/g').tar.gz" | tar zxvf - s5cmd \
-    && chmod +x s5cmd \
-    && mv s5cmd /usr/local/bin/s5cmd
-
 # LLVM
-ENV LLVM_VERSION=18
+ENV LLVM_VERSION=17
 RUN curl -fsSL 'https://apt.llvm.org/llvm-snapshot.gpg.key' | apt-key add - \
-    && echo "deb http://apt.llvm.org/${DEBIAN_VERSION_CODENAME}/ llvm-toolchain-${DEBIAN_VERSION_CODENAME}-${LLVM_VERSION} main" > /etc/apt/sources.list.d/llvm.stable.list \
+    && echo "deb http://apt.llvm.org/bullseye/ llvm-toolchain-bullseye-${LLVM_VERSION} main" > /etc/apt/sources.list.d/llvm.stable.list \
     && apt update \
     && apt install -y clang-${LLVM_VERSION} llvm-${LLVM_VERSION} \
     && bash -c 'for f in /usr/bin/clang*-${LLVM_VERSION} /usr/bin/llvm*-${LLVM_VERSION}; do ln -s "${f}" "${f%-${LLVM_VERSION}}"; done' \
     && rm -rf /var/lib/apt/lists/* /tmp/* /var/tmp/*
 
-# Install docker
-RUN curl -fsSL https://download.docker.com/linux/ubuntu/gpg | gpg --dearmor -o /usr/share/keyrings/docker-archive-keyring.gpg \
-    && echo "deb [arch=$(dpkg --print-architecture) signed-by=/usr/share/keyrings/docker-archive-keyring.gpg] https://download.docker.com/linux/debian ${DEBIAN_VERSION_CODENAME} stable" > /etc/apt/sources.list.d/docker.list \
+# PostgreSQL 14
+RUN curl -fsSL 'https://www.postgresql.org/media/keys/ACCC4CF8.asc' | apt-key add - \
+    && echo 'deb http://apt.postgresql.org/pub/repos/apt bullseye-pgdg main' > /etc/apt/sources.list.d/pgdg.list \
     && apt update \
-    && apt install -y docker-ce docker-ce-cli \
+    && apt install -y postgresql-client-14 \
     && rm -rf /var/lib/apt/lists/* /tmp/* /var/tmp/*
 
-# Configure sudo & docker
-RUN usermod -aG sudo nonroot && \
-    echo '%sudo ALL=(ALL) NOPASSWD:ALL' >> /etc/sudoers && \
-    usermod -aG docker nonroot
-
 # AWS CLI
 RUN curl "https://awscli.amazonaws.com/awscli-exe-linux-$(uname -m).zip" -o "awscliv2.zip" \
     && unzip -q awscliv2.zip \
@@ -99,7 +81,7 @@ RUN curl "https://awscli.amazonaws.com/awscli-exe-linux-$(uname -m).zip" -o "aws
     && rm awscliv2.zip
 
 # Mold: A Modern Linker
-ENV MOLD_VERSION=v2.33.0
+ENV MOLD_VERSION v2.4.0
 RUN set -e \
     && git clone https://github.com/rui314/mold.git \
     && mkdir mold/build \
@@ -124,51 +106,12 @@ RUN for package in Capture::Tiny DateTime Devel::Cover Digest::MD5 File::Spec JS
     && make install \
     && rm -rf ../lcov.tar.gz
 
-# Compile and install the static OpenSSL library
-ENV OPENSSL_VERSION=1.1.1w
-ENV OPENSSL_PREFIX=/usr/local/openssl
-RUN wget -O /tmp/openssl-${OPENSSL_VERSION}.tar.gz https://www.openssl.org/source/openssl-${OPENSSL_VERSION}.tar.gz && \
-    echo "cf3098950cb4d853ad95c0841f1f9c6d3dc102dccfcacd521d93925208b76ac8 /tmp/openssl-${OPENSSL_VERSION}.tar.gz" | sha256sum --check && \
-    cd /tmp && \
-    tar xzvf /tmp/openssl-${OPENSSL_VERSION}.tar.gz && \
-    rm /tmp/openssl-${OPENSSL_VERSION}.tar.gz && \
-    cd /tmp/openssl-${OPENSSL_VERSION} && \
-    ./config --prefix=${OPENSSL_PREFIX}  -static --static no-shared -fPIC && \
-    make -j "$(nproc)" && \
-    make install && \
-    cd /tmp && \
-    rm -rf /tmp/openssl-${OPENSSL_VERSION}
-
-# Use the same version of libicu as the compute nodes so that
-# clusters created using inidb on pageserver can be used by computes.
-#
-# TODO: at this time, Dockerfile.compute-node uses the debian bullseye libicu
-# package, which is 67.1. We're duplicating that knowledge here, and also, technically,
-# Debian has a few patches on top of 67.1 that we're not adding here.
-ENV ICU_VERSION=67.1
-ENV ICU_PREFIX=/usr/local/icu
-
-# Download and build static ICU
-RUN wget -O /tmp/libicu-${ICU_VERSION}.tgz https://github.com/unicode-org/icu/releases/download/release-${ICU_VERSION//./-}/icu4c-${ICU_VERSION//./_}-src.tgz && \
-    echo "94a80cd6f251a53bd2a997f6f1b5ac6653fe791dfab66e1eb0227740fb86d5dc /tmp/libicu-${ICU_VERSION}.tgz" | sha256sum --check && \
-    mkdir /tmp/icu && \
-    pushd /tmp/icu && \
-    tar -xzf /tmp/libicu-${ICU_VERSION}.tgz && \
-    pushd icu/source && \
-    ./configure --prefix=${ICU_PREFIX}  --enable-static --enable-shared=no CXXFLAGS="-fPIC" CFLAGS="-fPIC" && \
-    make -j "$(nproc)" && \
-    make install && \
-    popd && \
-    rm -rf icu && \
-    rm -f /tmp/libicu-${ICU_VERSION}.tgz && \
-    popd
-
 # Switch to nonroot user
 USER nonroot:nonroot
 WORKDIR /home/nonroot
 
 # Python
-ENV PYTHON_VERSION=3.9.19 \
+ENV PYTHON_VERSION=3.9.18 \
     PYENV_ROOT=/home/nonroot/.pyenv \
     PATH=/home/nonroot/.pyenv/shims:/home/nonroot/.pyenv/bin:/home/nonroot/.poetry/bin:$PATH
 RUN set -e \
@@ -192,14 +135,9 @@ WORKDIR /home/nonroot
 
 # Rust
 # Please keep the version of llvm (installed above) in sync with rust llvm (`rustc --version --verbose | grep LLVM`)
-ENV RUSTC_VERSION=1.81.0
+ENV RUSTC_VERSION=1.77.0
 ENV RUSTUP_HOME="/home/nonroot/.rustup"
 ENV PATH="/home/nonroot/.cargo/bin:${PATH}"
-ARG RUSTFILT_VERSION=0.2.1
-ARG CARGO_HAKARI_VERSION=0.9.30
-ARG CARGO_DENY_VERSION=0.16.1
-ARG CARGO_HACK_VERSION=0.6.31
-ARG CARGO_NEXTEST_VERSION=0.9.72
 RUN curl -sSO https://static.rust-lang.org/rustup/dist/$(uname -m)-unknown-linux-gnu/rustup-init && whoami && \
 	chmod +x rustup-init && \
 	./rustup-init -y --default-toolchain ${RUSTC_VERSION} && \
@@ -207,14 +145,16 @@ RUN curl -sSO https://static.rust-lang.org/rustup/dist/$(uname -m)-unknown-linux
     export PATH="$HOME/.cargo/bin:$PATH" && \
     . "$HOME/.cargo/env" && \
     cargo --version && rustup --version && \
-    rustup component add llvm-tools rustfmt clippy && \
-    cargo install rustfilt            --version ${RUSTFILT_VERSION} && \
-    cargo install cargo-hakari        --version ${CARGO_HAKARI_VERSION} && \
-    cargo install cargo-deny --locked --version ${CARGO_DENY_VERSION} && \
-    cargo install cargo-hack          --version ${CARGO_HACK_VERSION} && \
-    cargo install cargo-nextest       --version ${CARGO_NEXTEST_VERSION} && \
+    rustup component add llvm-tools-preview rustfmt clippy && \
+    cargo install --git https://github.com/paritytech/cachepot && \
+    cargo install rustfilt && \
+    cargo install cargo-hakari && \
+    cargo install cargo-deny --locked && \
+    cargo install cargo-hack && \
+    cargo install cargo-nextest && \
     rm -rf /home/nonroot/.cargo/registry && \
     rm -rf /home/nonroot/.cargo/git
+ENV RUSTC_WRAPPER=cachepot
 
 # Show versions
 RUN whoami \
@@ -224,6 +164,3 @@ RUN whoami \
     && rustup --version --verbose \
     && rustc --version --verbose \
     && clang --version
-
-# Set following flag to check in Makefile if its running in Docker
-RUN touch /home/nonroot/.docker_build

Dockerfile.compute-node

@@ -89,16 +89,16 @@ RUN apt update && \
 # 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 && \
     echo "4e39b3b2adada6254a7bdba6d297bb28e1a9835a9f879b74f37e2dab70203232 SFCGAL.tar.gz" | sha256sum --check && \
-    mkdir sfcgal-src && cd sfcgal-src && tar xzf ../SFCGAL.tar.gz --strip-components=1 -C . && \
+    mkdir sfcgal-src && cd sfcgal-src && tar xvzf ../SFCGAL.tar.gz --strip-components=1 -C . && \
     cmake -DCMAKE_BUILD_TYPE=Release . && make -j $(getconf _NPROCESSORS_ONLN) && \
     DESTDIR=/sfcgal make install -j $(getconf _NPROCESSORS_ONLN) && \
     make clean && cp -R /sfcgal/* /
 
-ENV PATH="/usr/local/pgsql/bin:$PATH"
+ENV PATH "/usr/local/pgsql/bin:$PATH"
 
 RUN wget https://download.osgeo.org/postgis/source/postgis-3.3.3.tar.gz -O postgis.tar.gz && \
     echo "74eb356e3f85f14233791013360881b6748f78081cc688ff9d6f0f673a762d13 postgis.tar.gz" | sha256sum --check && \
-    mkdir postgis-src && cd postgis-src && tar xzf ../postgis.tar.gz --strip-components=1 -C . && \
+    mkdir postgis-src && cd postgis-src && tar xvzf ../postgis.tar.gz --strip-components=1 -C . && \
     find /usr/local/pgsql -type f | sed 's|^/usr/local/pgsql/||' > /before.txt &&\
     ./autogen.sh && \
     ./configure --with-sfcgal=/usr/local/bin/sfcgal-config && \
@@ -124,7 +124,7 @@ RUN wget https://download.osgeo.org/postgis/source/postgis-3.3.3.tar.gz -O postg
 
 RUN 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 pgrouting-src && cd pgrouting-src && tar xvzf ../pgrouting.tar.gz --strip-components=1 -C . && \
     mkdir build && cd build && \
     cmake -DCMAKE_BUILD_TYPE=Release .. && \
     make -j $(getconf _NPROCESSORS_ONLN) && \
@@ -149,7 +149,7 @@ RUN apt update && \
 
 RUN 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 . && \
+    mkdir plv8-src && cd plv8-src && tar xvzf ../plv8.tar.gz --strip-components=1 -C . && \
     # generate and copy upgrade scripts
     mkdir -p upgrade && ./generate_upgrade.sh 3.1.10 && \
     cp upgrade/* /usr/local/pgsql/share/extension/ && \
@@ -194,7 +194,7 @@ RUN case "$(uname -m)" in \
 
 RUN 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 h3-src && cd h3-src && tar xvzf ../h3.tar.gz --strip-components=1 -C . && \
     mkdir build && cd build && \
     cmake .. -DCMAKE_BUILD_TYPE=Release && \
     make -j $(getconf _NPROCESSORS_ONLN) && \
@@ -204,7 +204,7 @@ RUN wget https://github.com/uber/h3/archive/refs/tags/v4.1.0.tar.gz -O h3.tar.gz
 
 RUN 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 . && \
+    mkdir h3-pg-src && cd h3-pg-src && tar xvzf ../h3-pg.tar.gz --strip-components=1 -C . && \
     export PATH="/usr/local/pgsql/bin:$PATH" && \
     make -j $(getconf _NPROCESSORS_ONLN) && \
     make -j $(getconf _NPROCESSORS_ONLN) install && \
@@ -222,7 +222,7 @@ 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 && \
     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 . && \
+    mkdir postgresql-unit-src && cd postgresql-unit-src && tar xvzf ../postgresql-unit.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
     make -j $(getconf _NPROCESSORS_ONLN) install PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
     # unit extension's "create extension" script relies on absolute install path to fill some reference tables.
@@ -241,17 +241,11 @@ RUN wget https://github.com/df7cb/postgresql-unit/archive/refs/tags/7.7.tar.gz -
 FROM build-deps AS vector-pg-build
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
-COPY 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 && \
-    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 && \
-    make -j $(getconf _NPROCESSORS_ONLN) OPTFLAGS="" PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
-    make -j $(getconf _NPROCESSORS_ONLN) OPTFLAGS="" install PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
+RUN wget https://github.com/pgvector/pgvector/archive/refs/tags/v0.5.1.tar.gz -O pgvector.tar.gz && \
+    echo "cc7a8e034a96e30a819911ac79d32f6bc47bdd1aa2de4d7d4904e26b83209dc8 pgvector.tar.gz" | sha256sum --check && \
+    mkdir pgvector-src && cd pgvector-src && tar xvzf ../pgvector.tar.gz --strip-components=1 -C . && \
+    make -j $(getconf _NPROCESSORS_ONLN) PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
+    make -j $(getconf _NPROCESSORS_ONLN) install PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/vector.control
 
 #########################################################################################
@@ -266,7 +260,7 @@ 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 && \
     echo "cfdefb15007286f67d3d45510f04a6a7a495004be5b3aecb12cda667e774203f pgjwt.tar.gz" | sha256sum --check && \
-    mkdir pgjwt-src && cd pgjwt-src && tar xzf ../pgjwt.tar.gz --strip-components=1 -C . && \
+    mkdir pgjwt-src && cd pgjwt-src && tar xvzf ../pgjwt.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) install PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/pgjwt.control
 
@@ -281,7 +275,7 @@ 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 && \
     echo "0821011743083226fc9b813c1f2ef5897a91901b57b6bea85a78e466187c6819 hypopg.tar.gz" | sha256sum --check && \
-    mkdir hypopg-src && cd hypopg-src && tar xzf ../hypopg.tar.gz --strip-components=1 -C . && \
+    mkdir hypopg-src && cd hypopg-src && tar xvzf ../hypopg.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
     make -j $(getconf _NPROCESSORS_ONLN) install PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/hypopg.control
@@ -297,7 +291,7 @@ 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 && \
     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 . && \
+    mkdir pg_hashids-src && cd pg_hashids-src && tar xvzf ../pg_hashids.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) PG_CONFIG=/usr/local/pgsql/bin/pg_config USE_PGXS=1 && \
     make -j $(getconf _NPROCESSORS_ONLN) install PG_CONFIG=/usr/local/pgsql/bin/pg_config USE_PGXS=1 && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/pg_hashids.control
@@ -311,12 +305,9 @@ RUN wget https://github.com/iCyberon/pg_hashids/archive/refs/tags/v1.2.1.tar.gz
 FROM build-deps AS rum-pg-build
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
-COPY patches/rum.patch /rum.patch
-
 RUN 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 && \
+    mkdir rum-src && cd rum-src && tar xvzf ../rum.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) PG_CONFIG=/usr/local/pgsql/bin/pg_config USE_PGXS=1 && \
     make -j $(getconf _NPROCESSORS_ONLN) install PG_CONFIG=/usr/local/pgsql/bin/pg_config USE_PGXS=1 && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/rum.control
@@ -332,7 +323,7 @@ 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 && \
     echo "9c7c3de67ea41638e14f06da5da57bac6f5bd03fea05c165a0ec862205a5c052 pgtap.tar.gz" | sha256sum --check && \
-    mkdir pgtap-src && cd pgtap-src && tar xzf ../pgtap.tar.gz --strip-components=1 -C . && \
+    mkdir pgtap-src && cd pgtap-src && tar xvzf ../pgtap.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
     make -j $(getconf _NPROCESSORS_ONLN) install PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/pgtap.control
@@ -348,7 +339,7 @@ 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 && \
     echo "0f7b1f159974f49a47842a8ab6751aecca1ed1142b6d5e38d81b064b2ead1b4b ip4r.tar.gz" | sha256sum --check && \
-    mkdir ip4r-src && cd ip4r-src && tar xzf ../ip4r.tar.gz --strip-components=1 -C . && \
+    mkdir ip4r-src && cd ip4r-src && tar xvzf ../ip4r.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
     make -j $(getconf _NPROCESSORS_ONLN) install PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/ip4r.control
@@ -364,7 +355,7 @@ 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 && \
     echo "4342f251432a5f6fb05b8597139d3ccde8dcf87e8ca1498e7ee931ca057a8575 prefix.tar.gz" | sha256sum --check && \
-    mkdir prefix-src && cd prefix-src && tar xzf ../prefix.tar.gz --strip-components=1 -C . && \
+    mkdir prefix-src && cd prefix-src && tar xvzf ../prefix.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
     make -j $(getconf _NPROCESSORS_ONLN) install PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/prefix.control
@@ -380,7 +371,7 @@ 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 && \
     echo "e2f55a6f4c4ab95ee4f1b4a2b73280258c5136b161fe9d059559556079694f0e hll.tar.gz" | sha256sum --check && \
-    mkdir hll-src && cd hll-src && tar xzf ../hll.tar.gz --strip-components=1 -C . && \
+    mkdir hll-src && cd hll-src && tar xvzf ../hll.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
     make -j $(getconf _NPROCESSORS_ONLN) install PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/hll.control
@@ -396,7 +387,7 @@ 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 && \
     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 . && \
+    mkdir plpgsql_check-src && cd plpgsql_check-src && tar xvzf ../plpgsql_check.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) PG_CONFIG=/usr/local/pgsql/bin/pg_config USE_PGXS=1 && \
     make -j $(getconf _NPROCESSORS_ONLN) install PG_CONFIG=/usr/local/pgsql/bin/pg_config USE_PGXS=1 && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/plpgsql_check.control
@@ -411,7 +402,7 @@ FROM build-deps AS timescaledb-pg-build
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
 ARG PG_VERSION
-ENV PATH="/usr/local/pgsql/bin:$PATH"
+ENV PATH "/usr/local/pgsql/bin:$PATH"
 
 RUN case "${PG_VERSION}" in \
       "v14" | "v15") \
@@ -427,7 +418,7 @@ RUN case "${PG_VERSION}" in \
     apt-get install -y cmake && \
     wget https://github.com/timescale/timescaledb/archive/refs/tags/${TIMESCALEDB_VERSION}.tar.gz -O timescaledb.tar.gz && \
     echo "${TIMESCALEDB_CHECKSUM} timescaledb.tar.gz" | sha256sum --check && \
-    mkdir timescaledb-src && cd timescaledb-src && tar xzf ../timescaledb.tar.gz --strip-components=1 -C . && \
+    mkdir timescaledb-src && cd timescaledb-src && tar xvzf ../timescaledb.tar.gz --strip-components=1 -C . && \
     ./bootstrap -DSEND_TELEMETRY_DEFAULT:BOOL=OFF -DUSE_TELEMETRY:BOOL=OFF -DAPACHE_ONLY:BOOL=ON -DCMAKE_BUILD_TYPE=Release && \
     cd build && \
     make -j $(getconf _NPROCESSORS_ONLN) && \
@@ -444,7 +435,7 @@ FROM build-deps AS pg-hint-plan-pg-build
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
 ARG PG_VERSION
-ENV PATH="/usr/local/pgsql/bin:$PATH"
+ENV PATH "/usr/local/pgsql/bin:$PATH"
 
 RUN case "${PG_VERSION}" in \
       "v14") \
@@ -465,11 +456,36 @@ RUN case "${PG_VERSION}" in \
     esac && \
     wget https://github.com/ossc-db/pg_hint_plan/archive/refs/tags/REL${PG_HINT_PLAN_VERSION}.tar.gz -O pg_hint_plan.tar.gz && \
     echo "${PG_HINT_PLAN_CHECKSUM} pg_hint_plan.tar.gz" | sha256sum --check && \
-    mkdir pg_hint_plan-src && cd pg_hint_plan-src && tar xzf ../pg_hint_plan.tar.gz --strip-components=1 -C . && \
+    mkdir pg_hint_plan-src && cd pg_hint_plan-src && tar xvzf ../pg_hint_plan.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) && \
     make install -j $(getconf _NPROCESSORS_ONLN) && \
     echo "trusted = true" >> /usr/local/pgsql/share/extension/pg_hint_plan.control
 
+#########################################################################################
+#
+# Layer "kq-imcx-pg-build"
+# compile kq_imcx extension
+#
+#########################################################################################
+FROM build-deps AS kq-imcx-pg-build
+COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
+
+ENV PATH "/usr/local/pgsql/bin/:$PATH"
+RUN apt-get update && \
+    apt-get install -y git libgtk2.0-dev libpq-dev libpam-dev libxslt-dev libkrb5-dev cmake && \
+    wget https://github.com/ketteq-neon/postgres-exts/archive/e0bd1a9d9313d7120c1b9c7bb15c48c0dede4c4e.tar.gz -O kq_imcx.tar.gz && \
+    echo "dc93a97ff32d152d32737ba7e196d9687041cda15e58ab31344c2f2de8855336 kq_imcx.tar.gz" | sha256sum --check && \
+    mkdir kq_imcx-src && cd kq_imcx-src && tar xvzf ../kq_imcx.tar.gz --strip-components=1 -C . && \
+    find /usr/local/pgsql -type f | sed 's|^/usr/local/pgsql/||' > /before.txt &&\
+    mkdir build && cd build && \
+    cmake -DCMAKE_BUILD_TYPE=Release .. && \
+    make -j $(getconf _NPROCESSORS_ONLN) && \
+    make -j $(getconf _NPROCESSORS_ONLN) install && \
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/kq_imcx.control && \
+    find /usr/local/pgsql -type f | sed 's|^/usr/local/pgsql/||' > /after.txt &&\
+    mkdir -p /extensions/kq_imcx && cp /usr/local/pgsql/share/extension/kq_imcx.control /extensions/kq_imcx && \
+    sort -o /before.txt /before.txt && sort -o /after.txt /after.txt && \
+    comm -13 /before.txt /after.txt | tar --directory=/usr/local/pgsql --zstd -cf /extensions/kq_imcx.tar.zst -T -
 
 #########################################################################################
 #
@@ -480,10 +496,10 @@ RUN case "${PG_VERSION}" in \
 FROM build-deps AS pg-cron-pg-build
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
-ENV PATH="/usr/local/pgsql/bin/:$PATH"
+ENV PATH "/usr/local/pgsql/bin/:$PATH"
 RUN wget https://github.com/citusdata/pg_cron/archive/refs/tags/v1.6.0.tar.gz -O pg_cron.tar.gz && \
     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 . && \
+    mkdir pg_cron-src && cd pg_cron-src && tar xvzf ../pg_cron.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) && \
     make -j $(getconf _NPROCESSORS_ONLN) install && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/pg_cron.control
@@ -506,10 +522,10 @@ RUN apt-get update && \
         libboost-system1.74-dev \
         libeigen3-dev
 
-ENV PATH="/usr/local/pgsql/bin/:/usr/local/pgsql/:$PATH"
+ENV PATH "/usr/local/pgsql/bin/:/usr/local/pgsql/:$PATH"
 RUN wget https://github.com/rdkit/rdkit/archive/refs/tags/Release_2023_03_3.tar.gz -O rdkit.tar.gz && \
     echo "bdbf9a2e6988526bfeb8c56ce3cdfe2998d60ac289078e2215374288185e8c8d rdkit.tar.gz" | sha256sum --check && \
-    mkdir rdkit-src && cd rdkit-src && tar xzf ../rdkit.tar.gz --strip-components=1 -C . && \
+    mkdir rdkit-src && cd rdkit-src && tar xvzf ../rdkit.tar.gz --strip-components=1 -C . && \
     cmake \
         -D RDK_BUILD_CAIRO_SUPPORT=OFF \
         -D RDK_BUILD_INCHI_SUPPORT=ON \
@@ -546,10 +562,10 @@ RUN wget https://github.com/rdkit/rdkit/archive/refs/tags/Release_2023_03_3.tar.
 FROM build-deps AS pg-uuidv7-pg-build
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
-ENV PATH="/usr/local/pgsql/bin/:$PATH"
+ENV PATH "/usr/local/pgsql/bin/:$PATH"
 RUN wget https://github.com/fboulnois/pg_uuidv7/archive/refs/tags/v1.0.1.tar.gz -O pg_uuidv7.tar.gz && \
     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 . && \
+    mkdir pg_uuidv7-src && cd pg_uuidv7-src && tar xvzf ../pg_uuidv7.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) && \
     make -j $(getconf _NPROCESSORS_ONLN) install && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/pg_uuidv7.control
@@ -563,10 +579,10 @@ RUN wget https://github.com/fboulnois/pg_uuidv7/archive/refs/tags/v1.0.1.tar.gz
 FROM build-deps AS pg-roaringbitmap-pg-build
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
-ENV PATH="/usr/local/pgsql/bin/:$PATH"
+ENV PATH "/usr/local/pgsql/bin/:$PATH"
 RUN wget https://github.com/ChenHuajun/pg_roaringbitmap/archive/refs/tags/v0.5.4.tar.gz -O pg_roaringbitmap.tar.gz && \
     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 . && \
+    mkdir pg_roaringbitmap-src && cd pg_roaringbitmap-src && tar xvzf ../pg_roaringbitmap.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) && \
     make -j $(getconf _NPROCESSORS_ONLN) install && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/roaringbitmap.control
@@ -580,10 +596,10 @@ RUN wget https://github.com/ChenHuajun/pg_roaringbitmap/archive/refs/tags/v0.5.4
 FROM build-deps AS pg-semver-pg-build
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
-ENV PATH="/usr/local/pgsql/bin/:$PATH"
+ENV PATH "/usr/local/pgsql/bin/:$PATH"
 RUN wget https://github.com/theory/pg-semver/archive/refs/tags/v0.32.1.tar.gz -O pg_semver.tar.gz && \
     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 . && \
+    mkdir pg_semver-src && cd pg_semver-src && tar xvzf ../pg_semver.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) && \
     make -j $(getconf _NPROCESSORS_ONLN) install && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/semver.control
@@ -598,7 +614,7 @@ FROM build-deps AS pg-embedding-pg-build
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
 ARG PG_VERSION
-ENV PATH="/usr/local/pgsql/bin/:$PATH"
+ENV PATH "/usr/local/pgsql/bin/:$PATH"
 RUN case "${PG_VERSION}" in \
       "v14" | "v15") \
         export PG_EMBEDDING_VERSION=0.3.5 \
@@ -609,7 +625,7 @@ RUN case "${PG_VERSION}" in \
     esac && \
     wget https://github.com/neondatabase/pg_embedding/archive/refs/tags/${PG_EMBEDDING_VERSION}.tar.gz -O pg_embedding.tar.gz && \
     echo "${PG_EMBEDDING_CHECKSUM} pg_embedding.tar.gz" | sha256sum --check && \
-    mkdir pg_embedding-src && cd pg_embedding-src && tar xzf ../pg_embedding.tar.gz --strip-components=1 -C . && \
+    mkdir pg_embedding-src && cd pg_embedding-src && tar xvzf ../pg_embedding.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) && \
     make -j $(getconf _NPROCESSORS_ONLN) install
 
@@ -622,10 +638,10 @@ RUN case "${PG_VERSION}" in \
 FROM build-deps AS pg-anon-pg-build
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
-ENV PATH="/usr/local/pgsql/bin/:$PATH"
+ENV PATH "/usr/local/pgsql/bin/:$PATH"
 RUN wget  https://github.com/neondatabase/postgresql_anonymizer/archive/refs/tags/neon_1.1.1.tar.gz -O pg_anon.tar.gz && \
     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 . && \
+    mkdir pg_anon-src && cd pg_anon-src && tar xvzf ../pg_anon.tar.gz --strip-components=1 -C . && \
     find /usr/local/pgsql -type f | sed 's|^/usr/local/pgsql/||' > /before.txt &&\
     make -j $(getconf _NPROCESSORS_ONLN) install PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/anon.control && \
@@ -657,7 +673,7 @@ RUN curl -sSO https://static.rust-lang.org/rustup/dist/$(uname -m)-unknown-linux
     chmod +x rustup-init && \
     ./rustup-init -y --no-modify-path --profile minimal --default-toolchain stable && \
     rm rustup-init && \
-    cargo install --locked --version 0.11.3 cargo-pgrx && \
+    cargo install --locked --version 0.10.2 cargo-pgrx && \
     /bin/bash -c 'cargo pgrx init --pg${PG_VERSION:1}=/usr/local/pgsql/bin/pg_config'
 
 USER root
@@ -672,15 +688,10 @@ 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 && \
-    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
-    # `unsafe-postgres` feature allows to build pgx extensions
-    # against postgres forks that decided to change their ABI name (like us).
-    # With that we can build extensions without forking them and using stock
-    # pgx. As this feature is new few manual version bumps were required.
-    sed -i 's/pgrx = "0.11.3"/pgrx = { version = "0.11.3", features = [ "unsafe-postgres" ] }/g' Cargo.toml && \
+RUN wget https://github.com/supabase/pg_jsonschema/archive/refs/tags/v0.2.0.tar.gz -O pg_jsonschema.tar.gz && \
+    echo "9118fc508a6e231e7a39acaa6f066fcd79af17a5db757b47d2eefbe14f7794f0 pg_jsonschema.tar.gz" | sha256sum --check && \
+    mkdir pg_jsonschema-src && cd pg_jsonschema-src && tar xvzf ../pg_jsonschema.tar.gz --strip-components=1 -C . && \
+    sed -i 's/pgrx = "0.10.2"/pgrx = { version = "0.10.2", features = [ "unsafe-postgres" ] }/g' Cargo.toml && \
     cargo pgrx install --release && \
     echo "trusted = true" >> /usr/local/pgsql/share/extension/pg_jsonschema.control
 
@@ -694,10 +705,10 @@ 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 && \
-    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 && \
+RUN wget https://github.com/supabase/pg_graphql/archive/refs/tags/v1.4.0.tar.gz -O pg_graphql.tar.gz && \
+    echo "bd8dc7230282b3efa9ae5baf053a54151ed0e66881c7c53750e2d0c765776edc pg_graphql.tar.gz" | sha256sum --check && \
+    mkdir pg_graphql-src && cd pg_graphql-src && tar xvzf ../pg_graphql.tar.gz --strip-components=1 -C . && \
+    sed -i 's/pgrx = "=0.10.2"/pgrx = { version = "0.10.2", features = [ "unsafe-postgres" ] }/g' Cargo.toml && \
     cargo pgrx install --release && \
     # it's needed to enable extension because it uses untrusted C language
     sed -i 's/superuser = false/superuser = true/g' /usr/local/pgsql/share/extension/pg_graphql.control && \
@@ -716,10 +727,7 @@ 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 && \
     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
-    sed -i 's/pgrx = { version = "=0.10.2",/pgrx = { version = "0.11.3",/g' Cargo.toml && \
-    sed -i 's/pgrx-tests = "=0.10.2"/pgrx-tests = "0.11.3"/g' Cargo.toml && \
+    mkdir pg_tiktoken-src && cd pg_tiktoken-src && tar xvzf ../pg_tiktoken.tar.gz --strip-components=1 -C . && \
     cargo pgrx install --release && \
     echo "trusted = true" >> /usr/local/pgsql/share/extension/pg_tiktoken.control
 
@@ -733,10 +741,14 @@ 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 && \
-    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 && \
+RUN wget https://github.com/pksunkara/pgx_ulid/archive/refs/tags/v0.1.3.tar.gz -O pgx_ulid.tar.gz && \
+    echo "ee5db82945d2d9f2d15597a80cf32de9dca67b897f605beb830561705f12683c pgx_ulid.tar.gz" | sha256sum --check && \
+    mkdir pgx_ulid-src && cd pgx_ulid-src && tar xvzf ../pgx_ulid.tar.gz --strip-components=1 -C . && \
+    echo "******************* Apply a patch for Postgres 16 support; delete in the next release ******************" && \
+    wget https://github.com/pksunkara/pgx_ulid/commit/f84954cf63fc8c80d964ac970d9eceed3c791196.patch && \
+    patch -p1 < f84954cf63fc8c80d964ac970d9eceed3c791196.patch && \
+    echo "********************************************************************************************************" && \
+    sed -i 's/pgrx       = "=0.10.2"/pgrx = { version = "=0.10.2", features = [ "unsafe-postgres" ] }/g' Cargo.toml && \
     cargo pgrx install --release && \
     echo "trusted = true" >> /usr/local/pgsql/share/extension/ulid.control
 
@@ -750,10 +762,10 @@ RUN wget https://github.com/pksunkara/pgx_ulid/archive/refs/tags/v0.1.5.tar.gz -
 FROM build-deps AS wal2json-pg-build
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
-ENV PATH="/usr/local/pgsql/bin/:$PATH"
+ENV PATH "/usr/local/pgsql/bin/:$PATH"
 RUN 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 . && \
+    mkdir wal2json-src && cd wal2json-src && tar xvzf ../wal2json_2_5.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) && \
     make -j $(getconf _NPROCESSORS_ONLN) install
 
@@ -766,10 +778,10 @@ RUN wget https://github.com/eulerto/wal2json/archive/refs/tags/wal2json_2_5.tar.
 FROM build-deps AS pg-ivm-build
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
-ENV PATH="/usr/local/pgsql/bin/:$PATH"
+ENV PATH "/usr/local/pgsql/bin/:$PATH"
 RUN wget https://github.com/sraoss/pg_ivm/archive/refs/tags/v1.7.tar.gz -O pg_ivm.tar.gz && \
     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 . && \
+    mkdir pg_ivm-src && cd pg_ivm-src && tar xvzf ../pg_ivm.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) && \
     make -j $(getconf _NPROCESSORS_ONLN) install && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/pg_ivm.control
@@ -783,10 +795,10 @@ 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
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
-ENV PATH="/usr/local/pgsql/bin/:$PATH"
+ENV PATH "/usr/local/pgsql/bin/:$PATH"
 RUN wget https://github.com/pgpartman/pg_partman/archive/refs/tags/v5.0.1.tar.gz -O pg_partman.tar.gz && \
     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 . && \
+    mkdir pg_partman-src && cd pg_partman-src && tar xvzf ../pg_partman.tar.gz --strip-components=1 -C . && \
     make -j $(getconf _NPROCESSORS_ONLN) && \
     make -j $(getconf _NPROCESSORS_ONLN) install && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/pg_partman.control
@@ -822,6 +834,7 @@ COPY --from=hll-pg-build /usr/local/pgsql/ /usr/local/pgsql/
 COPY --from=plpgsql-check-pg-build /usr/local/pgsql/ /usr/local/pgsql/
 COPY --from=timescaledb-pg-build /usr/local/pgsql/ /usr/local/pgsql/
 COPY --from=pg-hint-plan-pg-build /usr/local/pgsql/ /usr/local/pgsql/
+COPY --from=kq-imcx-pg-build /usr/local/pgsql/ /usr/local/pgsql/
 COPY --from=pg-cron-pg-build /usr/local/pgsql/ /usr/local/pgsql/
 COPY --from=pg-pgx-ulid-build /usr/local/pgsql/ /usr/local/pgsql/
 COPY --from=rdkit-pg-build /usr/local/pgsql/ /usr/local/pgsql/
@@ -909,70 +922,6 @@ RUN rm -r /usr/local/pgsql/include
 # if they were to be used by other libraries.
 RUN rm /usr/local/pgsql/lib/lib*.a
 
-
-#########################################################################################
-#
-# Layer neon-pg-ext-test
-#
-#########################################################################################
-
-FROM neon-pg-ext-build AS neon-pg-ext-test
-ARG PG_VERSION
-RUN mkdir /ext-src
-
-#COPY --from=postgis-build /postgis.tar.gz /ext-src/
-#COPY --from=postgis-build /sfcgal/* /usr
-COPY --from=plv8-build /plv8.tar.gz /ext-src/
-COPY --from=h3-pg-build /h3-pg.tar.gz /ext-src/
-COPY --from=unit-pg-build /postgresql-unit.tar.gz /ext-src/
-COPY --from=vector-pg-build /pgvector.tar.gz /ext-src/
-COPY --from=vector-pg-build /pgvector.patch /ext-src/
-COPY --from=pgjwt-pg-build /pgjwt.tar.gz /ext-src
-#COPY --from=pg-jsonschema-pg-build /home/nonroot/pg_jsonschema.tar.gz /ext-src
-#COPY --from=pg-graphql-pg-build /home/nonroot/pg_graphql.tar.gz /ext-src
-#COPY --from=pg-tiktoken-pg-build /home/nonroot/pg_tiktoken.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 --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
-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 --from=pg-cron-pg-build /pg_cron.tar.gz /ext-src
-COPY 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
-COPY --from=pg-roaringbitmap-pg-build /pg_roaringbitmap.tar.gz /ext-src
-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 --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; \
-    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 $f; done
-RUN cd /ext-src/pgvector-src && patch -p1 <../pgvector.patch
-RUN cd /ext-src/rum-src && patch -p1 <../rum.patch
-# cmake is required for the h3 test
-RUN apt-get update && apt-get install -y cmake
-RUN cd /ext-src/pg_hint_plan-src && patch -p1 < /ext-src/pg_hint_plan.patch
-COPY --chmod=755 docker-compose/run-tests.sh /run-tests.sh
-RUN patch -p1 </ext-src/pg_anon.patch
-RUN patch -p1 </ext-src/pg_cron.patch
-ENV PATH=/usr/local/pgsql/bin:$PATH
-ENV PGHOST=compute
-ENV PGPORT=55433
-ENV PGUSER=cloud_admin
-ENV PGDATABASE=postgres
 #########################################################################################
 #
 # Final layer
@@ -995,9 +944,6 @@ 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
 
-# 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)
@@ -1034,6 +980,6 @@ RUN apt update &&  \
     rm -rf /var/lib/apt/lists/* /tmp/* /var/tmp/* && \
     localedef -i en_US -c -f UTF-8 -A /usr/share/locale/locale.alias en_US.UTF-8
 
-ENV LANG=en_US.utf8
+ENV LANG en_US.utf8
 USER postgres
 ENTRYPOINT ["/usr/local/bin/compute_ctl"]

Makefile

@@ -3,9 +3,6 @@ ROOT_PROJECT_DIR := $(dir $(abspath $(lastword $(MAKEFILE_LIST))))
 # Where to install Postgres, default is ./pg_install, maybe useful for package managers
 POSTGRES_INSTALL_DIR ?= $(ROOT_PROJECT_DIR)/pg_install/
 
-OPENSSL_PREFIX_DIR := /usr/local/openssl
-ICU_PREFIX_DIR := /usr/local/icu
-
 #
 # We differentiate between release / debug build types using the BUILD_TYPE
 # environment variable.
@@ -23,31 +20,19 @@ else
 	$(error Bad build type '$(BUILD_TYPE)', see Makefile for options)
 endif
 
-ifeq ($(shell test -e /home/nonroot/.docker_build && echo -n yes),yes)
-	# Exclude static build openssl, icu for local build (MacOS, Linux)
-	# Only keep for build type release and debug
-	PG_CFLAGS += -I$(OPENSSL_PREFIX_DIR)/include
-	PG_CONFIGURE_OPTS += --with-icu
-	PG_CONFIGURE_OPTS += ICU_CFLAGS='-I/$(ICU_PREFIX_DIR)/include -DU_STATIC_IMPLEMENTATION'
-	PG_CONFIGURE_OPTS += ICU_LIBS='-L$(ICU_PREFIX_DIR)/lib -L$(ICU_PREFIX_DIR)/lib64 -licui18n -licuuc -licudata -lstdc++ -Wl,-Bdynamic -lm'
-	PG_CONFIGURE_OPTS += LDFLAGS='-L$(OPENSSL_PREFIX_DIR)/lib -L$(OPENSSL_PREFIX_DIR)/lib64 -L$(ICU_PREFIX_DIR)/lib -L$(ICU_PREFIX_DIR)/lib64 -Wl,-Bstatic -lssl -lcrypto -Wl,-Bdynamic -lrt -lm -ldl -lpthread'
-endif
-
 UNAME_S := $(shell uname -s)
 ifeq ($(UNAME_S),Linux)
 	# Seccomp BPF is only available for Linux
 	PG_CONFIGURE_OPTS += --with-libseccomp
 else ifeq ($(UNAME_S),Darwin)
-	ifndef DISABLE_HOMEBREW
-		# macOS with brew-installed openssl requires explicit paths
-		# It can be configured with OPENSSL_PREFIX variable
-		OPENSSL_PREFIX := $(shell brew --prefix openssl@3)
-		PG_CONFIGURE_OPTS += --with-includes=$(OPENSSL_PREFIX)/include --with-libraries=$(OPENSSL_PREFIX)/lib
-		PG_CONFIGURE_OPTS += PKG_CONFIG_PATH=$(shell brew --prefix icu4c)/lib/pkgconfig
-		# macOS already has bison and flex in the system, but they are old and result in postgres-v14 target failure
-		# brew formulae are keg-only and not symlinked into HOMEBREW_PREFIX, force their usage
-		EXTRA_PATH_OVERRIDES += $(shell brew --prefix bison)/bin/:$(shell brew --prefix flex)/bin/:
-	endif
+	# macOS with brew-installed openssl requires explicit paths
+	# It can be configured with OPENSSL_PREFIX variable
+	OPENSSL_PREFIX ?= $(shell brew --prefix openssl@3)
+	PG_CONFIGURE_OPTS += --with-includes=$(OPENSSL_PREFIX)/include --with-libraries=$(OPENSSL_PREFIX)/lib
+	PG_CONFIGURE_OPTS += PKG_CONFIG_PATH=$(shell brew --prefix icu4c)/lib/pkgconfig
+	# macOS already has bison and flex in the system, but they are old and result in postgres-v14 target failure
+	# brew formulae are keg-only and not symlinked into HOMEBREW_PREFIX, force their usage
+	EXTRA_PATH_OVERRIDES += $(shell brew --prefix bison)/bin/:$(shell brew --prefix flex)/bin/:
 endif
 
 # Use -C option so that when PostgreSQL "make install" installs the
@@ -69,8 +54,6 @@ CARGO_CMD_PREFIX += CARGO_TERM_PROGRESS_WHEN=never CI=1
 # Set PQ_LIB_DIR to make sure `storage_controller` get linked with bundled libpq (through diesel)
 CARGO_CMD_PREFIX += PQ_LIB_DIR=$(POSTGRES_INSTALL_DIR)/v16/lib
 
-CACHEDIR_TAG_CONTENTS := "Signature: 8a477f597d28d172789f06886806bc55"
-
 #
 # Top level Makefile to build Neon and PostgreSQL
 #
@@ -81,38 +64,26 @@ all: neon postgres neon-pg-ext
 #
 # The 'postgres_ffi' depends on the Postgres headers.
 .PHONY: neon
-neon: postgres-headers walproposer-lib cargo-target-dir
+neon: postgres-headers walproposer-lib
 	+@echo "Compiling Neon"
 	$(CARGO_CMD_PREFIX) cargo build $(CARGO_BUILD_FLAGS)
-.PHONY: cargo-target-dir
-cargo-target-dir:
-	# https://github.com/rust-lang/cargo/issues/14281
-	mkdir -p target
-	test -e target/CACHEDIR.TAG || echo "$(CACHEDIR_TAG_CONTENTS)" > target/CACHEDIR.TAG
 
 ### PostgreSQL parts
 # Some rules are duplicated for Postgres v14 and 15. We may want to refactor
 # to avoid the duplication in the future, but it's tolerable for now.
 #
 $(POSTGRES_INSTALL_DIR)/build/%/config.status:
-
-	mkdir -p $(POSTGRES_INSTALL_DIR)
-	test -e $(POSTGRES_INSTALL_DIR)/CACHEDIR.TAG || echo "$(CACHEDIR_TAG_CONTENTS)" > $(POSTGRES_INSTALL_DIR)/CACHEDIR.TAG
-
 	+@echo "Configuring Postgres $* build"
 	@test -s $(ROOT_PROJECT_DIR)/vendor/postgres-$*/configure || { \
 		echo "\nPostgres submodule not found in $(ROOT_PROJECT_DIR)/vendor/postgres-$*/, execute "; \
 		echo "'git submodule update --init --recursive --depth 2 --progress .' in project root.\n"; \
 		exit 1; }
 	mkdir -p $(POSTGRES_INSTALL_DIR)/build/$*
-
-	VERSION=$*; \
-	EXTRA_VERSION=$$(cd $(ROOT_PROJECT_DIR)/vendor/postgres-$$VERSION && git rev-parse HEAD); \
-	(cd $(POSTGRES_INSTALL_DIR)/build/$$VERSION && \
-	env PATH="$(EXTRA_PATH_OVERRIDES):$$PATH" $(ROOT_PROJECT_DIR)/vendor/postgres-$$VERSION/configure \
+	(cd $(POSTGRES_INSTALL_DIR)/build/$* && \
+	env PATH="$(EXTRA_PATH_OVERRIDES):$$PATH" $(ROOT_PROJECT_DIR)/vendor/postgres-$*/configure \
 		CFLAGS='$(PG_CFLAGS)' \
-		$(PG_CONFIGURE_OPTS) --with-extra-version=" ($$EXTRA_VERSION)" \
-		--prefix=$(abspath $(POSTGRES_INSTALL_DIR))/$$VERSION > configure.log)
+		$(PG_CONFIGURE_OPTS) \
+		--prefix=$(abspath $(POSTGRES_INSTALL_DIR))/$* > configure.log)
 
 # nicer alias to run 'configure'
 # Note: I've been unable to use templates for this part of our configuration.
@@ -148,8 +119,6 @@ postgres-%: postgres-configure-% \
 	$(MAKE) -C $(POSTGRES_INSTALL_DIR)/build/$*/contrib/pageinspect install
 	+@echo "Compiling amcheck $*"
 	$(MAKE) -C $(POSTGRES_INSTALL_DIR)/build/$*/contrib/amcheck install
-	+@echo "Compiling test_decoding $*"
-	$(MAKE) -C $(POSTGRES_INSTALL_DIR)/build/$*/contrib/test_decoding install
 
 .PHONY: postgres-clean-%
 postgres-clean-%:

README.md

@@ -1,6 +1,4 @@
-[![Neon](https://github.com/neondatabase/neon/assets/11527560/f15a17f0-836e-40c5-b35d-030606a6b660)](https://neon.tech)
-
-
+[![Neon](https://user-images.githubusercontent.com/13738772/236813940-dcfdcb5b-69d3-449b-a686-013febe834d4.png)](https://neon.tech)
 
 # Neon
 
@@ -64,12 +62,6 @@ brew install protobuf openssl flex bison icu4c pkg-config
 echo 'export PATH="$(brew --prefix openssl)/bin:$PATH"' >> ~/.zshrc
 ```
 
-If you get errors about missing `m4` you may have to install it manually:
-```
-brew install m4
-brew link --force m4
-```
-
 2. [Install Rust](https://www.rust-lang.org/tools/install)
 ```
 # recommended approach from https://www.rust-lang.org/tools/install
@@ -132,7 +124,7 @@ make -j`sysctl -n hw.logicalcpu` -s
 To run the `psql` client, install the `postgresql-client` package or modify `PATH` and `LD_LIBRARY_PATH` to include `pg_install/bin` and `pg_install/lib`, respectively.
 
 To run the integration tests or Python scripts (not required to use the code), install
-Python (3.9 or higher), and install the python3 packages using `./scripts/pysync` (requires [poetry>=1.8](https://python-poetry.org/)) in the project directory.
+Python (3.9 or higher), and install the python3 packages using `./scripts/pysync` (requires [poetry>=1.3](https://python-poetry.org/)) in the project directory.
 
 
 #### Running neon database
@@ -268,7 +260,7 @@ By default, this runs both debug and release modes, and all supported postgres v
 testing locally, it is convenient to run just one set of permutations, like this:
 
 ```sh
-DEFAULT_PG_VERSION=16 BUILD_TYPE=release ./scripts/pytest
+DEFAULT_PG_VERSION=15 BUILD_TYPE=release ./scripts/pytest
 ```
 
 ## Flamegraphs

compute_tools/Cargo.toml

@@ -4,11 +4,6 @@ version = "0.1.0"
 edition.workspace = true
 license.workspace = true
 
-[features]
-default = []
-# Enables test specific features.
-testing = []
-
 [dependencies]
 anyhow.workspace = true
 async-compression.workspace = true
@@ -32,12 +27,10 @@ reqwest = { workspace = true, features = ["json"] }
 tokio = { workspace = true, features = ["rt", "rt-multi-thread"] }
 tokio-postgres.workspace = true
 tokio-util.workspace = true
-tokio-stream.workspace = true
 tracing.workspace = true
 tracing-opentelemetry.workspace = true
 tracing-subscriber.workspace = true
 tracing-utils.workspace = true
-thiserror.workspace = true
 url.workspace = true
 
 compute_api.workspace = true
@@ -49,4 +42,3 @@ vm_monitor = { version = "0.1", path = "../libs/vm_monitor/" }
 zstd = "0.13"
 bytes = "1.0"
 rust-ini = "0.20.0"
-rlimit = "0.10.1"

compute_tools/src/bin/compute_ctl.rs

@@ -6,7 +6,7 @@
 //! - Every start is a fresh start, so the data directory is removed and
 //!   initialized again on each run.
 //! - If remote_extension_config is provided, it will be used to fetch extensions list
-//!   and download `shared_preload_libraries` from the remote storage.
+//!  and download `shared_preload_libraries` from the remote storage.
 //! - Next it will put configuration files into the `PGDATA` directory.
 //! - Sync safekeepers and get commit LSN.
 //! - Get `basebackup` from pageserver using the returned on the previous step LSN.
@@ -33,6 +33,7 @@
 //!             -b /usr/local/bin/postgres \
 //!             -r http://pg-ext-s3-gateway \
 //! ```
+//!
 use std::collections::HashMap;
 use std::fs::File;
 use std::path::Path;
@@ -44,14 +45,12 @@ use std::{thread, time::Duration};
 use anyhow::{Context, Result};
 use chrono::Utc;
 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};
-use tracing::{error, info, warn};
+use tracing::{error, info};
 use url::Url;
 
 use compute_api::responses::ComputeStatus;
-use compute_api::spec::ComputeSpec;
 
 use compute_tools::compute::{
     forward_termination_signal, ComputeNode, ComputeState, ParsedSpec, PG_PID,
@@ -63,45 +62,12 @@ use compute_tools::logger::*;
 use compute_tools::monitor::launch_monitor;
 use compute_tools::params::*;
 use compute_tools::spec::*;
-use compute_tools::swap::resize_swap;
-use rlimit::{setrlimit, Resource};
 
 // this is an arbitrary build tag. Fine as a default / for testing purposes
 // in-case of not-set environment var
 const BUILD_TAG_DEFAULT: &str = "latest";
 
 fn main() -> Result<()> {
-    let (build_tag, clap_args) = init()?;
-
-    // enable core dumping for all child processes
-    setrlimit(Resource::CORE, rlimit::INFINITY, rlimit::INFINITY)?;
-
-    let (pg_handle, start_pg_result) = {
-        // Enter startup tracing context
-        let _startup_context_guard = startup_context_from_env();
-
-        let cli_args = process_cli(&clap_args)?;
-
-        let cli_spec = try_spec_from_cli(&clap_args, &cli_args)?;
-
-        let wait_spec_result = wait_spec(build_tag, cli_args, cli_spec)?;
-
-        start_postgres(&clap_args, wait_spec_result)?
-
-        // Startup is finished, exit the startup tracing span
-    };
-
-    // PostgreSQL is now running, if startup was successful. Wait until it exits.
-    let wait_pg_result = wait_postgres(pg_handle)?;
-
-    let delay_exit = cleanup_after_postgres_exit(start_pg_result)?;
-
-    maybe_delay_exit(delay_exit);
-
-    deinit_and_exit(wait_pg_result);
-}
-
-fn init() -> Result<(String, clap::ArgMatches)> {
     init_tracing_and_logging(DEFAULT_LOG_LEVEL)?;
 
     let mut signals = Signals::new([SIGINT, SIGTERM, SIGQUIT])?;
@@ -116,15 +82,9 @@ fn init() -> Result<(String, clap::ArgMatches)> {
         .to_string();
     info!("build_tag: {build_tag}");
 
-    Ok((build_tag, cli().get_matches()))
-}
-
-fn process_cli(matches: &clap::ArgMatches) -> Result<ProcessCliResult> {
-    let pgbin_default = "postgres";
-    let pgbin = matches
-        .get_one::<String>("pgbin")
-        .map(|s| s.as_str())
-        .unwrap_or(pgbin_default);
+    let matches = cli().get_matches();
+    let pgbin_default = String::from("postgres");
+    let pgbin = matches.get_one::<String>("pgbin").unwrap_or(&pgbin_default);
 
     let ext_remote_storage = matches
         .get_one::<String>("remote-ext-config")
@@ -150,32 +110,7 @@ fn process_cli(matches: &clap::ArgMatches) -> Result<ProcessCliResult> {
         .expect("Postgres connection string is required");
     let spec_json = matches.get_one::<String>("spec");
     let spec_path = matches.get_one::<String>("spec-path");
-    let resize_swap_on_bind = matches.get_flag("resize-swap-on-bind");
 
-    Ok(ProcessCliResult {
-        connstr,
-        pgdata,
-        pgbin,
-        ext_remote_storage,
-        http_port,
-        spec_json,
-        spec_path,
-        resize_swap_on_bind,
-    })
-}
-
-struct ProcessCliResult<'clap> {
-    connstr: &'clap str,
-    pgdata: &'clap str,
-    pgbin: &'clap str,
-    ext_remote_storage: Option<&'clap str>,
-    http_port: u16,
-    spec_json: Option<&'clap String>,
-    spec_path: Option<&'clap String>,
-    resize_swap_on_bind: bool,
-}
-
-fn startup_context_from_env() -> Option<opentelemetry::ContextGuard> {
     // Extract OpenTelemetry context for the startup actions from the
     // TRACEPARENT and TRACESTATE env variables, and attach it to the current
     // tracing context.
@@ -212,7 +147,7 @@ fn startup_context_from_env() -> Option<opentelemetry::ContextGuard> {
     if let Ok(val) = std::env::var("TRACESTATE") {
         startup_tracing_carrier.insert("tracestate".to_string(), val);
     }
-    if !startup_tracing_carrier.is_empty() {
+    let startup_context_guard = if !startup_tracing_carrier.is_empty() {
         use opentelemetry::propagation::TextMapPropagator;
         use opentelemetry::sdk::propagation::TraceContextPropagator;
         let guard = TraceContextPropagator::new()
@@ -222,17 +157,8 @@ fn startup_context_from_env() -> Option<opentelemetry::ContextGuard> {
         Some(guard)
     } else {
         None
-    }
-}
+    };
 
-fn try_spec_from_cli(
-    matches: &clap::ArgMatches,
-    ProcessCliResult {
-        spec_json,
-        spec_path,
-        ..
-    }: &ProcessCliResult,
-) -> Result<CliSpecParams> {
     let compute_id = matches.get_one::<String>("compute-id");
     let control_plane_uri = matches.get_one::<String>("control-plane-uri");
 
@@ -273,34 +199,6 @@ fn try_spec_from_cli(
         }
     };
 
-    Ok(CliSpecParams {
-        spec,
-        live_config_allowed,
-    })
-}
-
-struct CliSpecParams {
-    /// If a spec was provided via CLI or file, the [`ComputeSpec`]
-    spec: Option<ComputeSpec>,
-    live_config_allowed: bool,
-}
-
-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;
 
@@ -328,17 +226,19 @@ fn wait_spec(
 
     // 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
+    // because QEMU will already have it's 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.
+    // Launch http service first, so we were able to 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");
 
+    let extension_server_port: u16 = http_port;
+
     if !spec_set {
         // No spec provided, hang waiting for it.
         info!("no compute spec provided, waiting");
@@ -353,47 +253,21 @@ fn wait_spec(
                 break;
             }
         }
-
-        // Record for how long we slept waiting for the spec.
-        let now = Utc::now();
-        state.metrics.wait_for_spec_ms = now
-            .signed_duration_since(state.start_time)
-            .to_std()
-            .unwrap()
-            .as_millis() as u64;
-
-        // Reset start time, so that the total startup time that is calculated later will
-        // not include the time that we waited for the spec.
-        state.start_time = now;
     }
 
-    launch_lsn_lease_bg_task_for_static(&compute);
-
-    Ok(WaitSpecResult {
-        compute,
-        http_port,
-        resize_swap_on_bind,
-    })
-}
-
-struct WaitSpecResult {
-    compute: Arc<ComputeNode>,
-    // 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();
+
+    // Record for how long we slept waiting for the spec.
+    state.metrics.wait_for_spec_ms = Utc::now()
+        .signed_duration_since(state.start_time)
+        .to_std()
+        .unwrap()
+        .as_millis() as u64;
+    // Reset start time to the actual start of the configuration, so that
+    // total startup time was properly measured at the end.
+    state.start_time = Utc::now();
+
     state.status = ComputeStatus::Init;
     compute.state_changed.notify_all();
 
@@ -401,72 +275,33 @@ fn start_postgres(
         "running compute with features: {:?}",
         state.pspec.as_ref().unwrap().spec.features
     );
-    // before we release the mutex, fetch the swap size (if any) for later.
-    let swap_size_bytes = state.pspec.as_ref().unwrap().spec.swap_size_bytes;
     drop(state);
 
     // Launch remaining service threads
     let _monitor_handle = launch_monitor(&compute);
     let _configurator_handle = launch_configurator(&compute);
 
-    let mut prestartup_failed = false;
-    let mut delay_exit = false;
-
-    // Resize swap to the desired size if the compute spec says so
-    if let (Some(size_bytes), true) = (swap_size_bytes, resize_swap_on_bind) {
-        // To avoid 'swapoff' hitting postgres startup, we need to run resize-swap to completion
-        // *before* starting postgres.
-        //
-        // In theory, we could do this asynchronously if SkipSwapon was enabled for VMs, but this
-        // carries a risk of introducing hard-to-debug issues - e.g. if postgres sometimes gets
-        // OOM-killed during startup because swap wasn't available yet.
-        match resize_swap(size_bytes) {
-            Ok(()) => {
-                let size_gib = size_bytes as f32 / (1 << 20) as f32; // just for more coherent display.
-                info!(%size_bytes, %size_gib, "resized swap");
-            }
-            Err(err) => {
-                let err = err.context("failed to resize swap");
-                error!("{err:#}");
-
-                // Mark compute startup as failed; don't try to start postgres, and report this
-                // error to the control plane when it next asks.
-                prestartup_failed = true;
-                let mut state = compute.state.lock().unwrap();
-                state.error = Some(format!("{err:?}"));
-                state.status = ComputeStatus::Failed;
-                compute.state_changed.notify_all();
-                delay_exit = true;
-            }
-        }
-    }
-
-    let extension_server_port: u16 = http_port;
-
     // Start Postgres
-    let mut pg = None;
-    if !prestartup_failed {
-        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 {
-        warn!("skipping postgres startup because pre-startup step failed");
-    }
+    let mut delay_exit = false;
+    let mut exit_code = None;
+    let 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
+        }
+    };
 
     // Start the vm-monitor if directed to. The vm-monitor only runs on linux
     // because it requires cgroups.
@@ -499,7 +334,7 @@ fn start_postgres(
             // This token is used internally by the monitor to clean up all threads
             let token = CancellationToken::new();
 
-            let vm_monitor = rt.as_ref().map(|rt| {
+            let vm_monitor = &rt.as_ref().map(|rt| {
                 rt.spawn(vm_monitor::start(
                     Box::leak(Box::new(vm_monitor::Args {
                         cgroup: cgroup.cloned(),
@@ -512,41 +347,12 @@ fn start_postgres(
         }
     }
 
-    Ok((
-        pg,
-        StartPostgresResult {
-            delay_exit,
-            compute,
-            #[cfg(target_os = "linux")]
-            rt,
-            #[cfg(target_os = "linux")]
-            token,
-            #[cfg(target_os = "linux")]
-            vm_monitor,
-        },
-    ))
-}
-
-type PostgresHandle = (std::process::Child, std::thread::JoinHandle<()>);
-
-struct StartPostgresResult {
-    delay_exit: bool,
-    // passed through from WaitSpecResult
-    compute: Arc<ComputeNode>,
-
-    #[cfg(target_os = "linux")]
-    rt: Option<tokio::runtime::Runtime>,
-    #[cfg(target_os = "linux")]
-    token: tokio_util::sync::CancellationToken,
-    #[cfg(target_os = "linux")]
-    vm_monitor: Option<tokio::task::JoinHandle<Result<()>>>,
-}
-
-fn wait_postgres(pg: Option<PostgresHandle>) -> Result<WaitPostgresResult> {
     // Wait for the child Postgres process forever. In this state Ctrl+C will
     // propagate to Postgres and it will be shut down as well.
-    let mut exit_code = None;
     if let Some((mut pg, logs_handle)) = pg {
+        // Startup is finished, exit the startup tracing span
+        drop(startup_context_guard);
+
         let ecode = pg
             .wait()
             .expect("failed to start waiting on Postgres process");
@@ -561,25 +367,6 @@ fn wait_postgres(pg: Option<PostgresHandle>) -> Result<WaitPostgresResult> {
         exit_code = ecode.code()
     }
 
-    Ok(WaitPostgresResult { exit_code })
-}
-
-struct WaitPostgresResult {
-    exit_code: Option<i32>,
-}
-
-fn cleanup_after_postgres_exit(
-    StartPostgresResult {
-        mut delay_exit,
-        compute,
-        #[cfg(target_os = "linux")]
-        vm_monitor,
-        #[cfg(target_os = "linux")]
-        token,
-        #[cfg(target_os = "linux")]
-        rt,
-    }: StartPostgresResult,
-) -> Result<bool> {
     // Terminate the vm_monitor so it releases the file watcher on
     // /sys/fs/cgroup/neon-postgres.
     // Note: the vm-monitor only runs on linux because it requires cgroups.
@@ -621,19 +408,13 @@ fn cleanup_after_postgres_exit(
         error!("error while checking for core dumps: {err:?}");
     }
 
-    Ok(delay_exit)
-}
-
-fn maybe_delay_exit(delay_exit: bool) {
     // If launch failed, keep serving HTTP requests for a while, so the cloud
     // control plane can get the actual error.
     if delay_exit {
         info!("giving control plane 30s to collect the error before shutdown");
         thread::sleep(Duration::from_secs(30));
     }
-}
 
-fn deinit_and_exit(WaitPostgresResult { exit_code }: WaitPostgresResult) -> ! {
     // Shutdown trace pipeline gracefully, so that it has a chance to send any
     // pending traces before we exit. Shutting down OTEL tracing provider may
     // hang for quite some time, see, for example:
@@ -741,15 +522,10 @@ fn cli() -> clap::Command {
             Arg::new("filecache-connstr")
                 .long("filecache-connstr")
                 .default_value(
-                    "host=localhost port=5432 dbname=postgres user=cloud_admin sslmode=disable application_name=vm-monitor",
+                    "host=localhost port=5432 dbname=postgres user=cloud_admin sslmode=disable",
                 )
                 .value_name("FILECACHE_CONNSTR"),
         )
-        .arg(
-            Arg::new("resize-swap-on-bind")
-                .long("resize-swap-on-bind")
-                .action(clap::ArgAction::SetTrue),
-        )
 }
 
 /// When compute_ctl is killed, send also termination signal to sync-safekeepers

compute_tools/src/catalog.rs

@@ -1,116 +0,0 @@
-use compute_api::{
-    responses::CatalogObjects,
-    spec::{Database, Role},
-};
-use futures::Stream;
-use postgres::{Client, NoTls};
-use std::{path::Path, process::Stdio, result::Result, sync::Arc};
-use tokio::{
-    io::{AsyncBufReadExt, BufReader},
-    process::Command,
-    task,
-};
-use tokio_stream::{self as stream, StreamExt};
-use tokio_util::codec::{BytesCodec, FramedRead};
-use tracing::warn;
-
-use crate::{
-    compute::ComputeNode,
-    pg_helpers::{get_existing_dbs, get_existing_roles},
-};
-
-pub async fn get_dbs_and_roles(compute: &Arc<ComputeNode>) -> anyhow::Result<CatalogObjects> {
-    let connstr = compute.connstr.clone();
-    task::spawn_blocking(move || {
-        let mut client = Client::connect(connstr.as_str(), NoTls)?;
-        let roles: Vec<Role>;
-        {
-            let mut xact = client.transaction()?;
-            roles = get_existing_roles(&mut xact)?;
-        }
-        let databases: Vec<Database> = get_existing_dbs(&mut client)?.values().cloned().collect();
-
-        Ok(CatalogObjects { roles, databases })
-    })
-    .await?
-}
-
-#[derive(Debug, thiserror::Error)]
-pub enum SchemaDumpError {
-    #[error("Database does not exist.")]
-    DatabaseDoesNotExist,
-    #[error("Failed to execute pg_dump.")]
-    IO(#[from] std::io::Error),
-}
-
-// It uses the pg_dump utility to dump the schema of the specified database.
-// The output is streamed back to the caller and supposed to be streamed via HTTP.
-//
-// Before return the result with the output, it checks that pg_dump produced any output.
-// If not, it tries to parse the stderr output to determine if the database does not exist
-// and special error is returned.
-//
-// To make sure that the process is killed when the caller drops the stream, we use tokio kill_on_drop feature.
-pub async fn get_database_schema(
-    compute: &Arc<ComputeNode>,
-    dbname: &str,
-) -> Result<impl Stream<Item = Result<bytes::Bytes, std::io::Error>>, SchemaDumpError> {
-    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)
-        .arg("--schema-only")
-        .arg(connstr.as_str())
-        .stdout(Stdio::piped())
-        .stderr(Stdio::piped())
-        .kill_on_drop(true)
-        .spawn()?;
-
-    let stdout = cmd.stdout.take().ok_or_else(|| {
-        std::io::Error::new(std::io::ErrorKind::Other, "Failed to capture stdout.")
-    })?;
-
-    let stderr = cmd.stderr.take().ok_or_else(|| {
-        std::io::Error::new(std::io::ErrorKind::Other, "Failed to capture stderr.")
-    })?;
-
-    let mut stdout_reader = FramedRead::new(stdout, BytesCodec::new());
-    let stderr_reader = BufReader::new(stderr);
-
-    let first_chunk = match stdout_reader.next().await {
-        Some(Ok(bytes)) if !bytes.is_empty() => bytes,
-        Some(Err(e)) => {
-            return Err(SchemaDumpError::IO(e));
-        }
-        _ => {
-            let mut lines = stderr_reader.lines();
-            if let Some(line) = lines.next_line().await? {
-                if line.contains(&format!("FATAL:  database \"{}\" does not exist", dbname)) {
-                    return Err(SchemaDumpError::DatabaseDoesNotExist);
-                }
-                warn!("pg_dump stderr: {}", line)
-            }
-            tokio::spawn(async move {
-                while let Ok(Some(line)) = lines.next_line().await {
-                    warn!("pg_dump stderr: {}", line)
-                }
-            });
-
-            return Err(SchemaDumpError::IO(std::io::Error::new(
-                std::io::ErrorKind::Other,
-                "failed to start pg_dump",
-            )));
-        }
-    };
-    let initial_stream = stream::once(Ok(first_chunk.freeze()));
-    // Consume stderr and log warnings
-    tokio::spawn(async move {
-        let mut lines = stderr_reader.lines();
-        while let Ok(Some(line)) = lines.next_line().await {
-            warn!("pg_dump stderr: {}", line)
-        }
-    });
-    Ok(initial_stream.chain(stdout_reader.map(|res| res.map(|b| b.freeze()))))
-}

compute_tools/src/compute.rs

@@ -56,7 +56,6 @@ pub struct ComputeNode {
     /// - we push new spec and it does reconfiguration
     /// - but then something happens and compute pod / VM is destroyed,
     ///   so k8s controller starts it again with the **old** spec
-    ///
     /// and the same for empty computes:
     /// - we started compute without any spec
     /// - we push spec and it does configuration
@@ -400,15 +399,7 @@ impl ComputeNode {
     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;
-        #[cfg(feature = "testing")]
-        let max_attempts = if let Ok(v) = env::var("NEON_COMPUTE_TESTING_BASEBACKUP_RETRIES") {
-            u16::from_str(&v).unwrap()
-        } else {
-            DEFAULT_ATTEMPTS
-        };
-        #[cfg(not(feature = "testing"))]
-        let max_attempts = DEFAULT_ATTEMPTS;
+        let max_attempts = 10;
         loop {
             let result = self.try_get_basebackup(compute_state, lsn);
             match result {
@@ -807,11 +798,7 @@ impl ComputeNode {
         // In this case we need to connect with old `zenith_admin` name
         // and create new user. We cannot simply rename connected user,
         // but we can create a new one and grant it all privileges.
-        let mut connstr = self.connstr.clone();
-        connstr
-            .query_pairs_mut()
-            .append_pair("application_name", "apply_config");
-
+        let connstr = self.connstr.clone();
         let mut client = match Client::connect(connstr.as_str(), NoTls) {
             Err(e) => match e.code() {
                 Some(&SqlState::INVALID_PASSWORD)
@@ -831,15 +818,9 @@ impl ComputeNode {
                         Client::connect(zenith_admin_connstr.as_str(), NoTls)
                             .context("broken cloud_admin credential: tried connecting with cloud_admin but could not authenticate, and zenith_admin does not work either")?;
                     // Disable forwarding so that users don't get a cloud_admin role
-
-                    let mut func = || {
-                        client.simple_query("SET neon.forward_ddl = false")?;
-                        client.simple_query("CREATE USER cloud_admin WITH SUPERUSER")?;
-                        client.simple_query("GRANT zenith_admin TO cloud_admin")?;
-                        Ok::<_, anyhow::Error>(())
-                    };
-                    func().context("apply_config setup cloud_admin")?;
-
+                    client.simple_query("SET neon.forward_ddl = false")?;
+                    client.simple_query("CREATE USER cloud_admin WITH SUPERUSER")?;
+                    client.simple_query("GRANT zenith_admin TO cloud_admin")?;
                     drop(client);
 
                     // reconnect with connstring with expected name
@@ -851,48 +832,39 @@ impl ComputeNode {
         };
 
         // Disable DDL forwarding because control plane already knows about these roles/databases.
-        client
-            .simple_query("SET neon.forward_ddl = false")
-            .context("apply_config SET neon.forward_ddl = false")?;
+        client.simple_query("SET neon.forward_ddl = false")?;
 
         // Proceed with post-startup configuration. Note, that order of operations is important.
         let spec = &compute_state.pspec.as_ref().expect("spec must be set").spec;
-        create_neon_superuser(spec, &mut client).context("apply_config create_neon_superuser")?;
-        cleanup_instance(&mut client).context("apply_config cleanup_instance")?;
-        handle_roles(spec, &mut client).context("apply_config handle_roles")?;
-        handle_databases(spec, &mut client).context("apply_config handle_databases")?;
-        handle_role_deletions(spec, connstr.as_str(), &mut client)
-            .context("apply_config handle_role_deletions")?;
+        create_neon_superuser(spec, &mut client)?;
+        cleanup_instance(&mut client)?;
+        handle_roles(spec, &mut client)?;
+        handle_databases(spec, &mut client)?;
+        handle_role_deletions(spec, connstr.as_str(), &mut client)?;
         handle_grants(
             spec,
             &mut client,
             connstr.as_str(),
             self.has_feature(ComputeFeature::AnonExtension),
-        )
-        .context("apply_config handle_grants")?;
-        handle_extensions(spec, &mut client).context("apply_config handle_extensions")?;
-        handle_extension_neon(&mut client).context("apply_config handle_extension_neon")?;
-        create_availability_check_data(&mut client)
-            .context("apply_config create_availability_check_data")?;
+        )?;
+        handle_extensions(spec, &mut client)?;
+        handle_extension_neon(&mut client)?;
+        create_availability_check_data(&mut client)?;
 
         // 'Close' connection
         drop(client);
 
         // Run migrations separately to not hold up cold starts
         thread::spawn(move || {
-            let mut connstr = connstr.clone();
-            connstr
-                .query_pairs_mut()
-                .append_pair("application_name", "migrations");
-
             let mut client = Client::connect(connstr.as_str(), NoTls)?;
-            handle_migrations(&mut client).context("apply_config handle_migrations")
+            handle_migrations(&mut client)
         });
         Ok(())
     }
 
-    // Wrapped this around `pg_ctl reload`, but right now we don't use
-    // `pg_ctl` for start / stop.
+    // We could've wrapped this around `pg_ctl reload`, but right now we don't use
+    // `pg_ctl` for start / stop, so this just seems much easier to do as we already
+    // have opened connection to Postgres and superuser access.
     #[instrument(skip_all)]
     fn pg_reload_conf(&self) -> Result<()> {
         let pgctl_bin = Path::new(&self.pgbin).parent().unwrap().join("pg_ctl");
@@ -935,39 +907,38 @@ impl ComputeNode {
         // temporarily reset max_cluster_size in config
         // to avoid the possibility of hitting the limit, while we are reconfiguring:
         // creating new extensions, roles, etc...
-        config::with_compute_ctl_tmp_override(pgdata_path, "neon.max_cluster_size=-1", || {
-            self.pg_reload_conf()?;
+        config::compute_ctl_temp_override_create(pgdata_path, "neon.max_cluster_size=-1")?;
+        self.pg_reload_conf()?;
 
-            let mut client = Client::connect(self.connstr.as_str(), NoTls)?;
+        let mut client = Client::connect(self.connstr.as_str(), NoTls)?;
 
-            // Proceed with post-startup configuration. Note, that order of operations is important.
-            // Disable DDL forwarding because control plane already knows about these roles/databases.
-            if spec.mode == ComputeMode::Primary {
-                client.simple_query("SET neon.forward_ddl = false")?;
-                cleanup_instance(&mut client)?;
-                handle_roles(&spec, &mut client)?;
-                handle_databases(&spec, &mut client)?;
-                handle_role_deletions(&spec, self.connstr.as_str(), &mut client)?;
-                handle_grants(
-                    &spec,
-                    &mut client,
-                    self.connstr.as_str(),
-                    self.has_feature(ComputeFeature::AnonExtension),
-                )?;
-                handle_extensions(&spec, &mut client)?;
-                handle_extension_neon(&mut client)?;
-                // We can skip handle_migrations here because a new migration can only appear
-                // if we have a new version of the compute_ctl binary, which can only happen
-                // if compute got restarted, in which case we'll end up inside of apply_config
-                // instead of reconfigure.
-            }
+        // Proceed with post-startup configuration. Note, that order of operations is important.
+        // Disable DDL forwarding because control plane already knows about these roles/databases.
+        if spec.mode == ComputeMode::Primary {
+            client.simple_query("SET neon.forward_ddl = false")?;
+            cleanup_instance(&mut client)?;
+            handle_roles(&spec, &mut client)?;
+            handle_databases(&spec, &mut client)?;
+            handle_role_deletions(&spec, self.connstr.as_str(), &mut client)?;
+            handle_grants(
+                &spec,
+                &mut client,
+                self.connstr.as_str(),
+                self.has_feature(ComputeFeature::AnonExtension),
+            )?;
+            handle_extensions(&spec, &mut client)?;
+            handle_extension_neon(&mut client)?;
+            // We can skip handle_migrations here because a new migration can only appear
+            // if we have a new version of the compute_ctl binary, which can only happen
+            // if compute got restarted, in which case we'll end up inside of apply_config
+            // instead of reconfigure.
+        }
 
-            // 'Close' connection
-            drop(client);
-
-            Ok(())
-        })?;
+        // 'Close' connection
+        drop(client);
 
+        // reset max_cluster_size in config back to original value and reload config
+        config::compute_ctl_temp_override_remove(pgdata_path)?;
         self.pg_reload_conf()?;
 
         let unknown_op = "unknown".to_string();
@@ -1058,17 +1029,12 @@ impl ComputeNode {
                 // 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()?;
+                config::compute_ctl_temp_override_create(pgdata_path, "neon.max_cluster_size=-1")?;
+                self.pg_reload_conf()?;
 
-                        self.apply_config(&compute_state)?;
+                self.apply_config(&compute_state)?;
 
-                        Ok(())
-                    },
-                )?;
+                config::compute_ctl_temp_override_remove(pgdata_path)?;
                 self.pg_reload_conf()?;
             }
             self.post_apply_config()?;
@@ -1125,7 +1091,7 @@ impl ComputeNode {
     // EKS worker nodes have following core dump settings:
     //   /proc/sys/kernel/core_pattern -> core
     //   /proc/sys/kernel/core_uses_pid -> 1
-    //   ulimit -c -> unlimited
+    //   ulimint -c -> unlimited
     // which results in core dumps being written to postgres data directory as core.<pid>.
     //
     // Use that as a default location and pattern, except macos where core dumps are written
@@ -1296,12 +1262,10 @@ LIMIT 100",
         .await
         .map_err(DownloadError::Other);
 
-        if download_size.is_ok() {
-            self.ext_download_progress
-                .write()
-                .expect("bad lock")
-                .insert(ext_archive_name.to_string(), (download_start, true));
-        }
+        self.ext_download_progress
+            .write()
+            .expect("bad lock")
+            .insert(ext_archive_name.to_string(), (download_start, true));
 
         download_size
     }
@@ -1404,9 +1368,7 @@ pub fn forward_termination_signal() {
     let pg_pid = PG_PID.load(Ordering::SeqCst);
     if pg_pid != 0 {
         let pg_pid = nix::unistd::Pid::from_raw(pg_pid as i32);
-        // Use 'fast' shutdown (SIGINT) because it also creates a shutdown checkpoint, which is important for
-        // ROs to get a list of running xacts faster instead of going through the CLOG.
-        // See https://www.postgresql.org/docs/current/server-shutdown.html for the list of modes and signals.
-        kill(pg_pid, Signal::SIGINT).ok();
+        // use 'immediate' shutdown (SIGQUIT): https://www.postgresql.org/docs/current/server-shutdown.html
+        kill(pg_pid, Signal::SIGQUIT).ok();
     }
 }

compute_tools/src/config.rs

@@ -6,8 +6,8 @@ use std::path::Path;
 use anyhow::Result;
 
 use crate::pg_helpers::escape_conf_value;
-use crate::pg_helpers::{GenericOptionExt, PgOptionsSerialize};
-use compute_api::spec::{ComputeMode, ComputeSpec, GenericOption};
+use crate::pg_helpers::PgOptionsSerialize;
+use compute_api::spec::{ComputeMode, ComputeSpec};
 
 /// Check that `line` is inside a text file and put it there if it is not.
 /// Create file if it doesn't exist.
@@ -83,27 +83,12 @@ pub fn write_postgres_conf(
         ComputeMode::Replica => {
             // hot_standby is 'on' by default, but let's be explicit
             writeln!(file, "hot_standby=on")?;
-        }
-    }
 
-    if cfg!(target_os = "linux") {
-        // Check /proc/sys/vm/overcommit_memory -- if it equals 2 (i.e. linux memory overcommit is
-        // disabled), then the control plane has enabled swap and we should set
-        // dynamic_shared_memory_type = 'mmap'.
-        //
-        // This is (maybe?) temporary - for more, see https://github.com/neondatabase/cloud/issues/12047.
-        let overcommit_memory_contents = std::fs::read_to_string("/proc/sys/vm/overcommit_memory")
-            // ignore any errors - they may be expected to occur under certain situations (e.g. when
-            // not running in Linux).
-            .unwrap_or_else(|_| String::new());
-        if overcommit_memory_contents.trim() == "2" {
-            let opt = GenericOption {
-                name: "dynamic_shared_memory_type".to_owned(),
-                value: Some("mmap".to_owned()),
-                vartype: "enum".to_owned(),
-            };
-
-            write!(file, "{}", opt.to_pg_setting())?;
+            // Inform the replica about the primary state
+            // Default is 'false'
+            if let Some(primary_is_running) = spec.primary_is_running {
+                writeln!(file, "neon.primary_is_running={}", primary_is_running)?;
+            }
         }
     }
 
@@ -125,17 +110,18 @@ pub fn write_postgres_conf(
     Ok(())
 }
 
-pub fn with_compute_ctl_tmp_override<F>(pgdata_path: &Path, options: &str, exec: F) -> Result<()>
-where
-    F: FnOnce() -> Result<()>,
-{
+/// create file compute_ctl_temp_override.conf in pgdata_dir
+/// add provided options to this file
+pub fn compute_ctl_temp_override_create(pgdata_path: &Path, options: &str) -> Result<()> {
     let path = pgdata_path.join("compute_ctl_temp_override.conf");
     let mut file = File::create(path)?;
     write!(file, "{}", options)?;
-
-    let res = exec();
-
-    file.set_len(0)?;
-
-    res
+    Ok(())
+}
+
+/// remove file compute_ctl_temp_override.conf in pgdata_dir
+pub fn compute_ctl_temp_override_remove(pgdata_path: &Path) -> Result<()> {
+    let path = pgdata_path.join("compute_ctl_temp_override.conf");
+    std::fs::remove_file(path)?;
+    Ok(())
 }

compute_tools/src/http/api.rs

@@ -5,21 +5,17 @@ use std::net::SocketAddr;
 use std::sync::Arc;
 use std::thread;
 
-use crate::catalog::SchemaDumpError;
-use crate::catalog::{get_database_schema, get_dbs_and_roles};
 use crate::compute::forward_termination_signal;
 use crate::compute::{ComputeNode, ComputeState, ParsedSpec};
 use compute_api::requests::ConfigurationRequest;
 use compute_api::responses::{ComputeStatus, ComputeStatusResponse, GenericAPIError};
 
 use anyhow::Result;
-use hyper::header::CONTENT_TYPE;
 use hyper::service::{make_service_fn, service_fn};
 use hyper::{Body, Method, Request, Response, Server, StatusCode};
 use tokio::task;
-use tracing::{debug, error, info, warn};
+use tracing::{error, info, warn};
 use tracing_utils::http::OtelName;
-use utils::http::request::must_get_query_param;
 
 fn status_response_from_state(state: &ComputeState) -> ComputeStatusResponse {
     ComputeStatusResponse {
@@ -48,7 +44,7 @@ async fn routes(req: Request<Body>, compute: &Arc<ComputeNode>) -> Response<Body
     match (req.method(), req.uri().path()) {
         // Serialized compute state.
         (&Method::GET, "/status") => {
-            debug!("serving /status GET request");
+            info!("serving /status GET request");
             let state = compute.state.lock().unwrap();
             let status_response = status_response_from_state(&state);
             Response::new(Body::from(serde_json::to_string(&status_response).unwrap()))
@@ -137,34 +133,6 @@ async fn routes(req: Request<Body>, compute: &Arc<ComputeNode>) -> Response<Body
             }
         }
 
-        (&Method::GET, "/dbs_and_roles") => {
-            info!("serving /dbs_and_roles GET request",);
-            match get_dbs_and_roles(compute).await {
-                Ok(res) => render_json(Body::from(serde_json::to_string(&res).unwrap())),
-                Err(_) => {
-                    render_json_error("can't get dbs and roles", StatusCode::INTERNAL_SERVER_ERROR)
-                }
-            }
-        }
-
-        (&Method::GET, "/database_schema") => {
-            let database = match must_get_query_param(&req, "database") {
-                Err(e) => return e.into_response(),
-                Ok(database) => database,
-            };
-            info!("serving /database_schema GET request with database: {database}",);
-            match get_database_schema(compute, &database).await {
-                Ok(res) => render_plain(Body::wrap_stream(res)),
-                Err(SchemaDumpError::DatabaseDoesNotExist) => {
-                    render_json_error("database does not exist", StatusCode::NOT_FOUND)
-                }
-                Err(e) => {
-                    error!("can't get schema dump: {}", e);
-                    render_json_error("can't get schema dump", StatusCode::INTERNAL_SERVER_ERROR)
-                }
-            }
-        }
-
         // download extension files from remote extension storage on demand
         (&Method::POST, route) if route.starts_with("/extension_server/") => {
             info!("serving {:?} POST request", route);
@@ -335,25 +303,10 @@ fn render_json_error(e: &str, status: StatusCode) -> Response<Body> {
     };
     Response::builder()
         .status(status)
-        .header(CONTENT_TYPE, "application/json")
         .body(Body::from(serde_json::to_string(&error).unwrap()))
         .unwrap()
 }
 
-fn render_json(body: Body) -> Response<Body> {
-    Response::builder()
-        .header(CONTENT_TYPE, "application/json")
-        .body(body)
-        .unwrap()
-}
-
-fn render_plain(body: Body) -> Response<Body> {
-    Response::builder()
-        .header(CONTENT_TYPE, "text/plain")
-        .body(body)
-        .unwrap()
-}
-
 async fn handle_terminate_request(compute: &Arc<ComputeNode>) -> Result<(), (String, StatusCode)> {
     {
         let mut state = compute.state.lock().unwrap();

compute_tools/src/http/openapi_spec.yaml

@@ -68,51 +68,6 @@ paths:
               schema:
                 $ref: "#/components/schemas/Info"
 
-  /dbs_and_roles:
-    get:
-      tags:
-        - Info
-      summary: Get databases and roles in the catalog.
-      description: ""
-      operationId: getDbsAndRoles
-      responses:
-        200:
-          description: Compute schema objects
-          content:
-            application/json:
-              schema:
-                $ref: "#/components/schemas/DbsAndRoles"
-
-  /database_schema:
-    get:
-      tags:
-        - Info
-      summary: Get schema dump
-      parameters:
-        - name: database
-          in: query
-          description: Database name to dump.
-          required: true
-          schema:
-            type: string
-          example: "postgres"
-      description: Get schema dump in SQL format.
-      operationId: getDatabaseSchema
-      responses:
-        200:
-          description: Schema dump
-          content:
-            text/plain:
-              schema:
-                type: string
-                description: Schema dump in SQL format.
-        404:
-          description: Non existing database.
-          content:
-            application/json:
-              schema:
-                $ref: "#/components/schemas/GenericError"
-
   /check_writability:
     post:
       tags:
@@ -274,73 +229,6 @@ components:
         num_cpus:
           type: integer
 
-    DbsAndRoles:
-      type: object
-      description: Databases and Roles
-      required:
-        - roles
-        - databases
-      properties:
-        roles:
-          type: array
-          items:
-            $ref: "#/components/schemas/Role"
-        databases:
-          type: array
-          items:
-            $ref: "#/components/schemas/Database"
-
-    Database:
-      type: object
-      description: Database
-      required:
-        - name
-        - owner
-        - restrict_conn
-        - invalid
-      properties:
-        name:
-          type: string
-        owner:
-          type: string
-        options:
-          type: array
-          items:
-            $ref: "#/components/schemas/GenericOption"
-        restrict_conn:
-          type: boolean
-        invalid:
-          type: boolean
-
-    Role:
-      type: object
-      description: Role
-      required:
-        - name
-      properties:
-        name:
-          type: string
-        encrypted_password:
-          type: string
-        options:
-          type: array
-          items:
-            $ref: "#/components/schemas/GenericOption"
-
-    GenericOption:
-      type: object
-      description: Schema Generic option
-      required:
-        - name
-        - vartype
-      properties:
-        name:
-          type: string
-        value:
-          type: string
-        vartype:
-          type: string
-
     ComputeState:
       type: object
       required:

compute_tools/src/lib.rs

@@ -8,14 +8,10 @@ pub mod configurator;
 pub mod http;
 #[macro_use]
 pub mod logger;
-pub mod catalog;
 pub mod compute;
 pub mod extension_server;
-pub mod lsn_lease;
-mod migration;
 pub mod monitor;
 pub mod params;
 pub mod pg_helpers;
 pub mod spec;
-pub mod swap;
 pub mod sync_sk;

compute_tools/src/lsn_lease.rs

@@ -1,186 +0,0 @@
-use anyhow::bail;
-use anyhow::Result;
-use postgres::{NoTls, SimpleQueryMessage};
-use std::time::SystemTime;
-use std::{str::FromStr, sync::Arc, thread, time::Duration};
-use utils::id::TenantId;
-use utils::id::TimelineId;
-
-use compute_api::spec::ComputeMode;
-use tracing::{info, warn};
-use utils::{
-    lsn::Lsn,
-    shard::{ShardCount, ShardNumber, TenantShardId},
-};
-
-use crate::compute::ComputeNode;
-
-/// Spawns a background thread to periodically renew LSN leases for static compute.
-/// Do nothing if the compute is not in static mode.
-pub fn launch_lsn_lease_bg_task_for_static(compute: &Arc<ComputeNode>) {
-    let (tenant_id, timeline_id, lsn) = {
-        let state = compute.state.lock().unwrap();
-        let spec = state.pspec.as_ref().expect("Spec must be set");
-        match spec.spec.mode {
-            ComputeMode::Static(lsn) => (spec.tenant_id, spec.timeline_id, lsn),
-            _ => return,
-        }
-    };
-    let compute = compute.clone();
-
-    let span = tracing::info_span!("lsn_lease_bg_task", %tenant_id, %timeline_id, %lsn);
-    thread::spawn(move || {
-        let _entered = span.entered();
-        if let Err(e) = lsn_lease_bg_task(compute, tenant_id, timeline_id, lsn) {
-            // TODO: might need stronger error feedback than logging an warning.
-            warn!("Exited with error: {e}");
-        }
-    });
-}
-
-/// Renews lsn lease periodically so static compute are not affected by GC.
-fn lsn_lease_bg_task(
-    compute: Arc<ComputeNode>,
-    tenant_id: TenantId,
-    timeline_id: TimelineId,
-    lsn: Lsn,
-) -> Result<()> {
-    loop {
-        let valid_until = acquire_lsn_lease_with_retry(&compute, tenant_id, timeline_id, lsn)?;
-        let valid_duration = valid_until
-            .duration_since(SystemTime::now())
-            .unwrap_or(Duration::ZERO);
-
-        // Sleep for 60 seconds less than the valid duration but no more than half of the valid duration.
-        let sleep_duration = valid_duration
-            .saturating_sub(Duration::from_secs(60))
-            .max(valid_duration / 2);
-
-        info!(
-            "Succeeded, sleeping for {} seconds",
-            sleep_duration.as_secs()
-        );
-        thread::sleep(sleep_duration);
-    }
-}
-
-/// Acquires lsn lease in a retry loop. Returns the expiration time if a lease is granted.
-/// Returns an error if a lease is explicitly not granted. Otherwise, we keep sending requests.
-fn acquire_lsn_lease_with_retry(
-    compute: &Arc<ComputeNode>,
-    tenant_id: TenantId,
-    timeline_id: TimelineId,
-    lsn: Lsn,
-) -> Result<SystemTime> {
-    let mut attempts = 0usize;
-    let mut retry_period_ms: f64 = 500.0;
-    const MAX_RETRY_PERIOD_MS: f64 = 60.0 * 1000.0;
-
-    loop {
-        // Note: List of pageservers is dynamic, need to re-read configs before each attempt.
-        let configs = {
-            let state = compute.state.lock().unwrap();
-
-            let spec = state.pspec.as_ref().expect("spec must be set");
-
-            let conn_strings = spec.pageserver_connstr.split(',');
-
-            conn_strings
-                .map(|connstr| {
-                    let mut config = postgres::Config::from_str(connstr).expect("Invalid connstr");
-                    if let Some(storage_auth_token) = &spec.storage_auth_token {
-                        info!("Got storage auth token from spec file");
-                        config.password(storage_auth_token.clone());
-                    } else {
-                        info!("Storage auth token not set");
-                    }
-                    config
-                })
-                .collect::<Vec<_>>()
-        };
-
-        let result = try_acquire_lsn_lease(tenant_id, timeline_id, lsn, &configs);
-        match result {
-            Ok(Some(res)) => {
-                return Ok(res);
-            }
-            Ok(None) => {
-                bail!("Permanent error: lease could not be obtained, LSN is behind the GC cutoff");
-            }
-            Err(e) => {
-                warn!("Failed to acquire lsn lease: {e} (attempt {attempts}");
-
-                thread::sleep(Duration::from_millis(retry_period_ms as u64));
-                retry_period_ms *= 1.5;
-                retry_period_ms = retry_period_ms.min(MAX_RETRY_PERIOD_MS);
-            }
-        }
-        attempts += 1;
-    }
-}
-
-/// Tries to acquire an LSN lease through PS page_service API.
-fn try_acquire_lsn_lease(
-    tenant_id: TenantId,
-    timeline_id: TimelineId,
-    lsn: Lsn,
-    configs: &[postgres::Config],
-) -> Result<Option<SystemTime>> {
-    fn get_valid_until(
-        config: &postgres::Config,
-        tenant_shard_id: TenantShardId,
-        timeline_id: TimelineId,
-        lsn: Lsn,
-    ) -> Result<Option<SystemTime>> {
-        let mut client = config.connect(NoTls)?;
-        let cmd = format!("lease lsn {} {} {} ", tenant_shard_id, timeline_id, lsn);
-        let res = client.simple_query(&cmd)?;
-        let msg = match res.first() {
-            Some(msg) => msg,
-            None => bail!("empty response"),
-        };
-        let row = match msg {
-            SimpleQueryMessage::Row(row) => row,
-            _ => bail!("error parsing lsn lease response"),
-        };
-
-        // Note: this will be None if a lease is explicitly not granted.
-        let valid_until_str = row.get("valid_until");
-
-        let valid_until = valid_until_str.map(|s| {
-            SystemTime::UNIX_EPOCH
-                .checked_add(Duration::from_millis(u128::from_str(s).unwrap() as u64))
-                .expect("Time larger than max SystemTime could handle")
-        });
-        Ok(valid_until)
-    }
-
-    let shard_count = configs.len();
-
-    let valid_until = if shard_count > 1 {
-        configs
-            .iter()
-            .enumerate()
-            .map(|(shard_number, config)| {
-                let tenant_shard_id = TenantShardId {
-                    tenant_id,
-                    shard_count: ShardCount::new(shard_count as u8),
-                    shard_number: ShardNumber(shard_number as u8),
-                };
-                get_valid_until(config, tenant_shard_id, timeline_id, lsn)
-            })
-            .collect::<Result<Vec<Option<SystemTime>>>>()?
-            .into_iter()
-            .min()
-            .unwrap()
-    } else {
-        get_valid_until(
-            &configs[0],
-            TenantShardId::unsharded(tenant_id),
-            timeline_id,
-            lsn,
-        )?
-    };
-
-    Ok(valid_until)
-}

compute_tools/src/migration.rs

@@ -1,105 +0,0 @@
-use anyhow::{Context, Result};
-use postgres::Client;
-use tracing::info;
-
-pub(crate) struct MigrationRunner<'m> {
-    client: &'m mut Client,
-    migrations: &'m [&'m str],
-}
-
-impl<'m> MigrationRunner<'m> {
-    pub fn new(client: &'m mut Client, migrations: &'m [&'m str]) -> Self {
-        // The neon_migration.migration_id::id column is a bigint, which is equivalent to an i64
-        assert!(migrations.len() + 1 < i64::MAX as usize);
-
-        Self { client, migrations }
-    }
-
-    fn get_migration_id(&mut self) -> Result<i64> {
-        let query = "SELECT id FROM neon_migration.migration_id";
-        let row = self
-            .client
-            .query_one(query, &[])
-            .context("run_migrations get migration_id")?;
-
-        Ok(row.get::<&str, i64>("id"))
-    }
-
-    fn update_migration_id(&mut self, migration_id: i64) -> Result<()> {
-        let setval = format!("UPDATE neon_migration.migration_id SET id={}", migration_id);
-
-        self.client
-            .simple_query(&setval)
-            .context("run_migrations update id")?;
-
-        Ok(())
-    }
-
-    fn prepare_migrations(&mut self) -> Result<()> {
-        let query = "CREATE SCHEMA IF NOT EXISTS neon_migration";
-        self.client.simple_query(query)?;
-
-        let query = "CREATE TABLE IF NOT EXISTS neon_migration.migration_id (key INT NOT NULL PRIMARY KEY, id bigint NOT NULL DEFAULT 0)";
-        self.client.simple_query(query)?;
-
-        let query = "INSERT INTO neon_migration.migration_id VALUES (0, 0) ON CONFLICT DO NOTHING";
-        self.client.simple_query(query)?;
-
-        let query = "ALTER SCHEMA neon_migration OWNER TO cloud_admin";
-        self.client.simple_query(query)?;
-
-        let query = "REVOKE ALL ON SCHEMA neon_migration FROM PUBLIC";
-        self.client.simple_query(query)?;
-
-        Ok(())
-    }
-
-    pub fn run_migrations(mut self) -> Result<()> {
-        self.prepare_migrations()?;
-
-        let mut current_migration = self.get_migration_id()? as usize;
-        while current_migration < self.migrations.len() {
-            macro_rules! migration_id {
-                ($cm:expr) => {
-                    ($cm + 1) as i64
-                };
-            }
-
-            let migration = self.migrations[current_migration];
-
-            if migration.starts_with("-- SKIP") {
-                info!("Skipping migration id={}", migration_id!(current_migration));
-            } else {
-                info!(
-                    "Running migration id={}:\n{}\n",
-                    migration_id!(current_migration),
-                    migration
-                );
-
-                self.client
-                    .simple_query("BEGIN")
-                    .context("begin migration")?;
-
-                self.client.simple_query(migration).with_context(|| {
-                    format!(
-                        "run_migrations migration id={}",
-                        migration_id!(current_migration)
-                    )
-                })?;
-
-                // Migration IDs start at 1
-                self.update_migration_id(migration_id!(current_migration))?;
-
-                self.client
-                    .simple_query("COMMIT")
-                    .context("commit migration")?;
-
-                info!("Finished migration id={}", migration_id!(current_migration));
-            }
-
-            current_migration += 1;
-        }
-
-        Ok(())
-    }
-}

compute_tools/src/migrations/0001-neon_superuser_bypass_rls.sql

@@ -1 +0,0 @@
-ALTER ROLE neon_superuser BYPASSRLS;

compute_tools/src/migrations/0002-alter_roles.sql

@@ -1,18 +0,0 @@
-DO $$
-DECLARE
-    role_name text;
-BEGIN
-    FOR role_name IN SELECT rolname FROM pg_roles WHERE pg_has_role(rolname, 'neon_superuser', 'member')
-    LOOP
-        RAISE NOTICE 'EXECUTING ALTER ROLE % INHERIT', quote_ident(role_name);
-        EXECUTE 'ALTER ROLE ' || quote_ident(role_name) || ' INHERIT';
-    END LOOP;
-
-    FOR role_name IN SELECT rolname FROM pg_roles
-        WHERE
-            NOT pg_has_role(rolname, 'neon_superuser', 'member') AND NOT starts_with(rolname, 'pg_')
-    LOOP
-        RAISE NOTICE 'EXECUTING ALTER ROLE % NOBYPASSRLS', quote_ident(role_name);
-        EXECUTE 'ALTER ROLE ' || quote_ident(role_name) || ' NOBYPASSRLS';
-    END LOOP;
-END $$;

compute_tools/src/migrations/0003-grant_pg_create_subscription_to_neon_superuser.sql

@@ -1,6 +0,0 @@
-DO $$
-BEGIN
-    IF (SELECT setting::numeric >= 160000 FROM pg_settings WHERE name = 'server_version_num') THEN
-        EXECUTE 'GRANT pg_create_subscription TO neon_superuser';
-    END IF;
-END $$;

compute_tools/src/migrations/0004-grant_pg_monitor_to_neon_superuser.sql

@@ -1 +0,0 @@
-GRANT pg_monitor TO neon_superuser WITH ADMIN OPTION;

compute_tools/src/migrations/0005-grant_all_on_tables_to_neon_superuser.sql

@@ -1,4 +0,0 @@
--- SKIP: Deemed insufficient for allowing relations created by extensions to be
---       interacted with by neon_superuser without permission issues.
-
-ALTER DEFAULT PRIVILEGES IN SCHEMA public GRANT ALL ON TABLES TO neon_superuser;

compute_tools/src/migrations/0006-grant_all_on_sequences_to_neon_superuser.sql

@@ -1,4 +0,0 @@
--- SKIP: Deemed insufficient for allowing relations created by extensions to be
---       interacted with by neon_superuser without permission issues.
-
-ALTER DEFAULT PRIVILEGES IN SCHEMA public GRANT ALL ON SEQUENCES TO neon_superuser;

compute_tools/src/migrations/0007-grant_all_on_tables_to_neon_superuser_with_grant_option.sql

@@ -1,3 +0,0 @@
--- SKIP: Moved inline to the handle_grants() functions.
-
-ALTER DEFAULT PRIVILEGES IN SCHEMA public GRANT ALL ON TABLES TO neon_superuser WITH GRANT OPTION;

compute_tools/src/migrations/0008-grant_all_on_sequences_to_neon_superuser_with_grant_option.sql

@@ -1,3 +0,0 @@
--- SKIP: Moved inline to the handle_grants() functions.
-
-ALTER DEFAULT PRIVILEGES IN SCHEMA public GRANT ALL ON SEQUENCES TO neon_superuser WITH GRANT OPTION;

compute_tools/src/migrations/0009-revoke_replication_for_previously_allowed_roles.sql

@@ -1,13 +0,0 @@
--- SKIP: The original goal of this migration was to prevent creating
---       subscriptions, but this migration was insufficient.
-
-DO $$
-DECLARE
-    role_name TEXT;
-BEGIN
-    FOR role_name IN SELECT rolname FROM pg_roles WHERE rolreplication IS TRUE
-    LOOP
-        RAISE NOTICE 'EXECUTING ALTER ROLE % NOREPLICATION', quote_ident(role_name);
-        EXECUTE 'ALTER ROLE ' || quote_ident(role_name) || ' NOREPLICATION';
-    END LOOP;
-END $$;

compute_tools/src/migrations/0010-grant_snapshot_synchronization_funcs_to_neon_superuser.sql

@@ -1,7 +0,0 @@
-DO $$
-BEGIN
-    IF (SELECT setting::numeric >= 160000 FROM pg_settings WHERE name = 'server_version_num') THEN
-       EXECUTE 'GRANT EXECUTE ON FUNCTION pg_export_snapshot TO neon_superuser';
-       EXECUTE 'GRANT EXECUTE ON FUNCTION pg_log_standby_snapshot TO neon_superuser';
-    END IF;
-END $$;

compute_tools/src/monitor.rs

@@ -17,11 +17,7 @@ const MONITOR_CHECK_INTERVAL: Duration = Duration::from_millis(500);
 // should be handled gracefully.
 fn watch_compute_activity(compute: &ComputeNode) {
     // Suppose that `connstr` doesn't change
-    let mut connstr = compute.connstr.clone();
-    connstr
-        .query_pairs_mut()
-        .append_pair("application_name", "compute_activity_monitor");
-    let connstr = connstr.as_str();
+    let connstr = compute.connstr.as_str();
 
     // During startup and configuration we connect to every Postgres database,
     // but we don't want to count this as some user activity. So wait until

compute_tools/src/pg_helpers.rs

@@ -22,10 +22,9 @@ use compute_api::spec::{Database, GenericOption, GenericOptions, PgIdent, Role};
 
 const POSTGRES_WAIT_TIMEOUT: Duration = Duration::from_millis(60 * 1000); // milliseconds
 
-/// Escape a string for including it in a SQL literal.
-///
-/// Wrapping the result with `E'{}'` or `'{}'` is not required,
-/// as it returns a ready-to-use SQL string literal, e.g. `'db'''` or `E'db\\'`.
+/// Escape a string for including it in a SQL literal. Wrapping the result
+/// with `E'{}'` or `'{}'` is not required, as it returns a ready-to-use
+/// SQL string literal, e.g. `'db'''` or `E'db\\'`.
 /// See <https://github.com/postgres/postgres/blob/da98d005cdbcd45af563d0c4ac86d0e9772cd15f/src/backend/utils/adt/quote.c#L47>
 /// for the original implementation.
 pub fn escape_literal(s: &str) -> String {
@@ -45,7 +44,7 @@ pub fn escape_conf_value(s: &str) -> String {
     format!("'{}'", res)
 }
 
-pub trait GenericOptionExt {
+trait GenericOptionExt {
     fn to_pg_option(&self) -> String;
     fn to_pg_setting(&self) -> String;
 }
@@ -490,7 +489,7 @@ pub fn handle_postgres_logs(stderr: std::process::ChildStderr) -> JoinHandle<()>
 /// Read Postgres logs from `stderr` until EOF. Buffer is flushed on one of the following conditions:
 /// - next line starts with timestamp
 /// - EOF
-/// - no new lines were written for the last 100 milliseconds
+/// - no new lines were written for the last second
 async fn handle_postgres_logs_async(stderr: tokio::process::ChildStderr) -> Result<()> {
     let mut lines = tokio::io::BufReader::new(stderr).lines();
     let timeout_duration = Duration::from_millis(100);

compute_tools/src/spec.rs

@@ -2,7 +2,7 @@ use std::fs::File;
 use std::path::Path;
 use std::str::FromStr;
 
-use anyhow::{anyhow, bail, Context, Result};
+use anyhow::{anyhow, bail, Result};
 use postgres::config::Config;
 use postgres::{Client, NoTls};
 use reqwest::StatusCode;
@@ -10,7 +10,6 @@ use tracing::{error, info, info_span, instrument, span_enabled, warn, Level};
 
 use crate::config;
 use crate::logger::inlinify;
-use crate::migration::MigrationRunner;
 use crate::params::PG_HBA_ALL_MD5;
 use crate::pg_helpers::*;
 
@@ -303,9 +302,9 @@ pub fn handle_roles(spec: &ComputeSpec, client: &mut Client) -> Result<()> {
             RoleAction::Create => {
                 // This branch only runs when roles are created through the console, so it is
                 // safe to add more permissions here. BYPASSRLS and REPLICATION are inherited
-                // from neon_superuser.
+                // from neon_superuser. (NOTE: REPLICATION has been removed from here for now).
                 let mut query: String = format!(
-                    "CREATE ROLE {} INHERIT CREATEROLE CREATEDB BYPASSRLS REPLICATION IN ROLE neon_superuser",
+                    "CREATE ROLE {} INHERIT CREATEROLE CREATEDB BYPASSRLS IN ROLE neon_superuser",
                     name.pg_quote()
                 );
                 info!("running role create query: '{}'", &query);
@@ -491,7 +490,7 @@ pub fn handle_databases(spec: &ComputeSpec, client: &mut Client) -> Result<()> {
                 "rename_db" => {
                     let new_name = op.new_name.as_ref().unwrap();
 
-                    if existing_dbs.contains_key(&op.name) {
+                    if existing_dbs.get(&op.name).is_some() {
                         let query: String = format!(
                             "ALTER DATABASE {} RENAME TO {}",
                             op.name.pg_quote(),
@@ -699,8 +698,7 @@ pub fn handle_grants(
 
         // it is important to run this after all grants
         if enable_anon_extension {
-            handle_extension_anon(spec, &db.owner, &mut db_client, false)
-                .context("handle_grants handle_extension_anon")?;
+            handle_extension_anon(spec, &db.owner, &mut db_client, false)?;
         }
     }
 
@@ -745,24 +743,21 @@ pub fn handle_extension_neon(client: &mut Client) -> Result<()> {
     // which may happen in two cases:
     // - extension was just installed
     // - extension was already installed and is up to date
-    let query = "ALTER EXTENSION neon UPDATE";
-    info!("update neon extension version with query: {}", query);
-    if let Err(e) = client.simple_query(query) {
-        error!(
-            "failed to upgrade neon extension during `handle_extension_neon`: {}",
-            e
-        );
-    }
+    // DISABLED due to compute node unpinning epic
+    // let query = "ALTER EXTENSION neon UPDATE";
+    // info!("update neon extension version with query: {}", query);
+    // client.simple_query(query)?;
 
     Ok(())
 }
 
 #[instrument(skip_all)]
-pub fn handle_neon_extension_upgrade(client: &mut Client) -> Result<()> {
-    info!("handle neon extension upgrade");
-    let query = "ALTER EXTENSION neon UPDATE";
-    info!("update neon extension version with query: {}", query);
-    client.simple_query(query)?;
+pub fn handle_neon_extension_upgrade(_client: &mut Client) -> Result<()> {
+    info!("handle neon extension upgrade (not really)");
+    // DISABLED due to compute node unpinning epic
+    // let query = "ALTER EXTENSION neon UPDATE";
+    // info!("update neon extension version with query: {}", query);
+    // client.simple_query(query)?;
 
     Ok(())
 }
@@ -775,27 +770,103 @@ pub fn handle_migrations(client: &mut Client) -> Result<()> {
     // !BE SURE TO ONLY ADD MIGRATIONS TO THE END OF THIS ARRAY. IF YOU DO NOT, VERY VERY BAD THINGS MAY HAPPEN!
     // !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
 
-    // Add new migrations in numerical order.
     let migrations = [
-        include_str!("./migrations/0001-neon_superuser_bypass_rls.sql"),
-        include_str!("./migrations/0002-alter_roles.sql"),
-        include_str!("./migrations/0003-grant_pg_create_subscription_to_neon_superuser.sql"),
-        include_str!("./migrations/0004-grant_pg_monitor_to_neon_superuser.sql"),
-        include_str!("./migrations/0005-grant_all_on_tables_to_neon_superuser.sql"),
-        include_str!("./migrations/0006-grant_all_on_sequences_to_neon_superuser.sql"),
-        include_str!(
-            "./migrations/0007-grant_all_on_tables_to_neon_superuser_with_grant_option.sql"
-        ),
-        include_str!(
-            "./migrations/0008-grant_all_on_sequences_to_neon_superuser_with_grant_option.sql"
-        ),
-        include_str!("./migrations/0009-revoke_replication_for_previously_allowed_roles.sql"),
-        include_str!(
-            "./migrations/0010-grant_snapshot_synchronization_funcs_to_neon_superuser.sql"
-        ),
+        "ALTER ROLE neon_superuser BYPASSRLS",
+        r#"
+DO $$
+DECLARE
+    role_name text;
+BEGIN
+    FOR role_name IN SELECT rolname FROM pg_roles WHERE pg_has_role(rolname, 'neon_superuser', 'member')
+    LOOP
+        RAISE NOTICE 'EXECUTING ALTER ROLE % INHERIT', quote_ident(role_name);
+        EXECUTE 'ALTER ROLE ' || quote_ident(role_name) || ' INHERIT';
+    END LOOP;
+
+    FOR role_name IN SELECT rolname FROM pg_roles
+        WHERE
+            NOT pg_has_role(rolname, 'neon_superuser', 'member') AND NOT starts_with(rolname, 'pg_')
+    LOOP
+        RAISE NOTICE 'EXECUTING ALTER ROLE % NOBYPASSRLS', quote_ident(role_name);
+        EXECUTE 'ALTER ROLE ' || quote_ident(role_name) || ' NOBYPASSRLS';
+    END LOOP;
+END $$;
+"#,
+        r#"
+DO $$
+BEGIN
+    IF (SELECT setting::numeric >= 160000 FROM pg_settings WHERE name = 'server_version_num') THEN
+        EXECUTE 'GRANT pg_create_subscription TO neon_superuser';
+    END IF;
+END
+$$;"#,
+        "GRANT pg_monitor TO neon_superuser WITH ADMIN OPTION",
+        // Don't remove: these are some SQLs that we originally applied in migrations but turned out to execute somewhere else.
+        "",
+        "",
+        "",
+        "",
+        // Add new migrations below.
+        r#"
+DO $$
+DECLARE
+    role_name TEXT;
+BEGIN
+    FOR role_name IN SELECT rolname FROM pg_roles WHERE rolreplication IS TRUE
+    LOOP
+        RAISE NOTICE 'EXECUTING ALTER ROLE % NOREPLICATION', quote_ident(role_name);
+        EXECUTE 'ALTER ROLE ' || quote_ident(role_name) || ' NOREPLICATION';
+    END LOOP;
+END
+$$;"#,
     ];
 
-    MigrationRunner::new(client, &migrations).run_migrations()?;
+    let mut query = "CREATE SCHEMA IF NOT EXISTS neon_migration";
+    client.simple_query(query)?;
+
+    query = "CREATE TABLE IF NOT EXISTS neon_migration.migration_id (key INT NOT NULL PRIMARY KEY, id bigint NOT NULL DEFAULT 0)";
+    client.simple_query(query)?;
+
+    query = "INSERT INTO neon_migration.migration_id VALUES (0, 0) ON CONFLICT DO NOTHING";
+    client.simple_query(query)?;
+
+    query = "ALTER SCHEMA neon_migration OWNER TO cloud_admin";
+    client.simple_query(query)?;
+
+    query = "REVOKE ALL ON SCHEMA neon_migration FROM PUBLIC";
+    client.simple_query(query)?;
+
+    query = "SELECT id FROM neon_migration.migration_id";
+    let row = client.query_one(query, &[])?;
+    let mut current_migration: usize = row.get::<&str, i64>("id") as usize;
+    let starting_migration_id = current_migration;
+
+    query = "BEGIN";
+    client.simple_query(query)?;
+
+    while current_migration < migrations.len() {
+        let migration = &migrations[current_migration];
+        if migration.is_empty() {
+            info!("Skip migration id={}", current_migration);
+        } else {
+            info!("Running migration:\n{}\n", migration);
+            client.simple_query(migration)?;
+        }
+        current_migration += 1;
+    }
+    let setval = format!(
+        "UPDATE neon_migration.migration_id SET id={}",
+        migrations.len()
+    );
+    client.simple_query(&setval)?;
+
+    query = "COMMIT";
+    client.simple_query(query)?;
+
+    info!(
+        "Ran {} migrations",
+        (migrations.len() - starting_migration_id)
+    );
 
     Ok(())
 }

compute_tools/src/swap.rs

@@ -1,45 +0,0 @@
-use std::path::Path;
-
-use anyhow::{anyhow, Context};
-use tracing::warn;
-
-pub const RESIZE_SWAP_BIN: &str = "/neonvm/bin/resize-swap";
-
-pub fn resize_swap(size_bytes: u64) -> anyhow::Result<()> {
-    // run `/neonvm/bin/resize-swap --once {size_bytes}`
-    //
-    // Passing '--once' causes resize-swap to delete itself after successful completion, which
-    // means that if compute_ctl restarts later, we won't end up calling 'swapoff' while
-    // postgres is running.
-    //
-    // NOTE: resize-swap is not very clever. If present, --once MUST be the first arg.
-    let child_result = std::process::Command::new("/usr/bin/sudo")
-        .arg(RESIZE_SWAP_BIN)
-        .arg("--once")
-        .arg(size_bytes.to_string())
-        .spawn();
-
-    child_result
-        .context("spawn() failed")
-        .and_then(|mut child| child.wait().context("wait() failed"))
-        .and_then(|status| match status.success() {
-            true => Ok(()),
-            false => {
-                // The command failed. Maybe it was because the resize-swap file doesn't exist?
-                // The --once flag causes it to delete itself on success so we don't disable swap
-                // while postgres is running; maybe this is fine.
-                match Path::new(RESIZE_SWAP_BIN).try_exists() {
-                    Err(_) | Ok(true) => Err(anyhow!("process exited with {status}")),
-                    // The path doesn't exist; we're actually ok 
-                    Ok(false) => {
-                        warn!("ignoring \"not found\" error from resize-swap to avoid swapoff while compute is running");
-                        Ok(())
-                    },
-                }
-            }
-        })
-        // wrap any prior error with the overall context that we couldn't run the command
-        .with_context(|| {
-            format!("could not run `/usr/bin/sudo {RESIZE_SWAP_BIN} --once {size_bytes}`")
-        })
-}

control_plane/Cargo.toml

@@ -6,6 +6,7 @@ license.workspace = true
 
 [dependencies]
 anyhow.workspace = true
+async-trait.workspace = true
 camino.workspace = true
 clap.workspace = true
 comfy-table.workspace = true
@@ -16,7 +17,6 @@ nix.workspace = true
 once_cell.workspace = true
 postgres.workspace = true
 hex.workspace = true
-humantime-serde.workspace = true
 hyper.workspace = true
 regex.workspace = true
 reqwest = { workspace = true, features = ["blocking", "json"] }
@@ -27,7 +27,6 @@ serde_with.workspace = true
 tar.workspace = true
 thiserror.workspace = true
 toml.workspace = true
-toml_edit.workspace = true
 tokio.workspace = true
 tokio-postgres.workspace = true
 tokio-util.workspace = true
@@ -39,7 +38,6 @@ safekeeper_api.workspace = true
 postgres_connection.workspace = true
 storage_broker.workspace = true
 utils.workspace = true
-whoami.workspace = true
 
 compute_api.workspace = true
 workspace_hack.workspace = true

control_plane/attachment_service/Cargo.toml

@@ -1,5 +1,5 @@
 [package]
-name = "storage_controller"
+name = "attachment_service"
 version = "0.1.0"
 edition.workspace = true
 license.workspace = true
@@ -18,7 +18,6 @@ anyhow.workspace = true
 aws-config.workspace = true
 bytes.workspace = true
 camino.workspace = true
-chrono.workspace = true
 clap.workspace = true
 fail.workspace = true
 futures.workspace = true
@@ -26,14 +25,12 @@ git-version.workspace = true
 hex.workspace = true
 hyper.workspace = true
 humantime.workspace = true
-itertools.workspace = true
 lasso.workspace = true
 once_cell.workspace = true
 pageserver_api.workspace = true
 pageserver_client.workspace = true
 postgres_connection.workspace = true
-rand.workspace = true
-reqwest = { workspace = true, features = ["stream"] }
+reqwest.workspace = true
 routerify.workspace = true
 serde.workspace = true
 serde_json.workspace = true
@@ -42,20 +39,13 @@ tokio.workspace = true
 tokio-util.workspace = true
 tracing.workspace = true
 measured.workspace = true
-scopeguard.workspace = true
-strum.workspace = true
-strum_macros.workspace = true
 
-diesel = { version = "2.1.4", features = [
-    "serde_json",
-    "postgres",
-    "r2d2",
-    "chrono",
-] }
+diesel = { version = "2.1.4", features = ["serde_json", "postgres", "r2d2"] }
 diesel_migrations = { version = "2.1.0" }
 r2d2 = { version = "0.8.10" }
 
-utils = { path = "../libs/utils/" }
-metrics = { path = "../libs/metrics/" }
-control_plane = { path = "../control_plane" }
-workspace_hack = { version = "0.1", path = "../workspace_hack" }
+utils = { path = "../../libs/utils/" }
+metrics = { path = "../../libs/metrics/" }
+control_plane = { path = ".." }
+workspace_hack = { version = "0.1", path = "../../workspace_hack" }
+

control_plane/attachment_service/src/compute_hook.rs

@@ -0,0 +1,462 @@
+use std::{collections::HashMap, time::Duration};
+
+use control_plane::endpoint::{ComputeControlPlane, EndpointStatus};
+use control_plane::local_env::LocalEnv;
+use hyper::{Method, StatusCode};
+use pageserver_api::shard::{ShardCount, ShardNumber, ShardStripeSize, TenantShardId};
+use postgres_connection::parse_host_port;
+use serde::{Deserialize, Serialize};
+use tokio_util::sync::CancellationToken;
+use utils::{
+    backoff::{self},
+    id::{NodeId, TenantId},
+};
+
+use crate::service::Config;
+
+const BUSY_DELAY: Duration = Duration::from_secs(1);
+const SLOWDOWN_DELAY: Duration = Duration::from_secs(5);
+
+pub(crate) const API_CONCURRENCY: usize = 32;
+
+struct ShardedComputeHookTenant {
+    stripe_size: ShardStripeSize,
+    shard_count: ShardCount,
+    shards: Vec<(ShardNumber, NodeId)>,
+}
+
+enum ComputeHookTenant {
+    Unsharded(NodeId),
+    Sharded(ShardedComputeHookTenant),
+}
+
+impl ComputeHookTenant {
+    /// Construct with at least one shard's information
+    fn new(tenant_shard_id: TenantShardId, stripe_size: ShardStripeSize, node_id: NodeId) -> Self {
+        if tenant_shard_id.shard_count.count() > 1 {
+            Self::Sharded(ShardedComputeHookTenant {
+                shards: vec![(tenant_shard_id.shard_number, node_id)],
+                stripe_size,
+                shard_count: tenant_shard_id.shard_count,
+            })
+        } else {
+            Self::Unsharded(node_id)
+        }
+    }
+
+    /// Set one shard's location.  If stripe size or shard count have changed, Self is reset
+    /// and drops existing content.
+    fn update(
+        &mut self,
+        tenant_shard_id: TenantShardId,
+        stripe_size: ShardStripeSize,
+        node_id: NodeId,
+    ) {
+        match self {
+            Self::Unsharded(existing_node_id) if tenant_shard_id.shard_count.count() == 1 => {
+                *existing_node_id = node_id
+            }
+            Self::Sharded(sharded_tenant)
+                if sharded_tenant.stripe_size == stripe_size
+                    && sharded_tenant.shard_count == tenant_shard_id.shard_count =>
+            {
+                if let Some(existing) = sharded_tenant
+                    .shards
+                    .iter()
+                    .position(|s| s.0 == tenant_shard_id.shard_number)
+                {
+                    sharded_tenant.shards.get_mut(existing).unwrap().1 = node_id;
+                } else {
+                    sharded_tenant
+                        .shards
+                        .push((tenant_shard_id.shard_number, node_id));
+                    sharded_tenant.shards.sort_by_key(|s| s.0)
+                }
+            }
+            _ => {
+                // Shard count changed: reset struct.
+                *self = Self::new(tenant_shard_id, stripe_size, node_id);
+            }
+        }
+    }
+}
+
+#[derive(Serialize, Deserialize, Debug)]
+struct ComputeHookNotifyRequestShard {
+    node_id: NodeId,
+    shard_number: ShardNumber,
+}
+
+/// Request body that we send to the control plane to notify it of where a tenant is attached
+#[derive(Serialize, Deserialize, Debug)]
+struct ComputeHookNotifyRequest {
+    tenant_id: TenantId,
+    stripe_size: Option<ShardStripeSize>,
+    shards: Vec<ComputeHookNotifyRequestShard>,
+}
+
+/// Error type for attempts to call into the control plane compute notification hook
+#[derive(thiserror::Error, Debug)]
+pub(crate) enum NotifyError {
+    // Request was not send successfully, e.g. transport error
+    #[error("Sending request: {0}")]
+    Request(#[from] reqwest::Error),
+    // Request could not be serviced right now due to ongoing Operation in control plane, but should be possible soon.
+    #[error("Control plane tenant busy")]
+    Busy,
+    // Explicit 429 response asking us to retry less frequently
+    #[error("Control plane overloaded")]
+    SlowDown,
+    // A 503 response indicates the control plane can't handle the request right now
+    #[error("Control plane unavailable (status {0})")]
+    Unavailable(StatusCode),
+    // API returned unexpected non-success status.  We will retry, but log a warning.
+    #[error("Control plane returned unexpected status {0}")]
+    Unexpected(StatusCode),
+    // We shutdown while sending
+    #[error("Shutting down")]
+    ShuttingDown,
+    // A response indicates we will never succeed, such as 400 or 404
+    #[error("Non-retryable error {0}")]
+    Fatal(StatusCode),
+}
+
+impl ComputeHookTenant {
+    fn maybe_reconfigure(&self, tenant_id: TenantId) -> Option<ComputeHookNotifyRequest> {
+        match self {
+            Self::Unsharded(node_id) => Some(ComputeHookNotifyRequest {
+                tenant_id,
+                shards: vec![ComputeHookNotifyRequestShard {
+                    shard_number: ShardNumber(0),
+                    node_id: *node_id,
+                }],
+                stripe_size: None,
+            }),
+            Self::Sharded(sharded_tenant)
+                if sharded_tenant.shards.len() == sharded_tenant.shard_count.count() as usize =>
+            {
+                Some(ComputeHookNotifyRequest {
+                    tenant_id,
+                    shards: sharded_tenant
+                        .shards
+                        .iter()
+                        .map(|(shard_number, node_id)| ComputeHookNotifyRequestShard {
+                            shard_number: *shard_number,
+                            node_id: *node_id,
+                        })
+                        .collect(),
+                    stripe_size: Some(sharded_tenant.stripe_size),
+                })
+            }
+            Self::Sharded(sharded_tenant) => {
+                // Sharded tenant doesn't yet have information for all its shards
+
+                tracing::info!(
+                    "ComputeHookTenant::maybe_reconfigure: not enough shards ({}/{})",
+                    sharded_tenant.shards.len(),
+                    sharded_tenant.shard_count.count()
+                );
+                None
+            }
+        }
+    }
+}
+
+/// The compute hook is a destination for notifications about changes to tenant:pageserver
+/// mapping.  It aggregates updates for the shards in a tenant, and when appropriate reconfigures
+/// the compute connection string.
+pub(super) struct ComputeHook {
+    config: Config,
+    state: tokio::sync::Mutex<HashMap<TenantId, ComputeHookTenant>>,
+    authorization_header: Option<String>,
+}
+
+impl ComputeHook {
+    pub(super) fn new(config: Config) -> Self {
+        let authorization_header = config
+            .control_plane_jwt_token
+            .clone()
+            .map(|jwt| format!("Bearer {}", jwt));
+
+        Self {
+            state: Default::default(),
+            config,
+            authorization_header,
+        }
+    }
+
+    /// For test environments: use neon_local's LocalEnv to update compute
+    async fn do_notify_local(
+        &self,
+        reconfigure_request: ComputeHookNotifyRequest,
+    ) -> anyhow::Result<()> {
+        let env = match LocalEnv::load_config() {
+            Ok(e) => e,
+            Err(e) => {
+                tracing::warn!("Couldn't load neon_local config, skipping compute update ({e})");
+                return Ok(());
+            }
+        };
+        let cplane =
+            ComputeControlPlane::load(env.clone()).expect("Error loading compute control plane");
+        let ComputeHookNotifyRequest {
+            tenant_id,
+            shards,
+            stripe_size,
+        } = reconfigure_request;
+
+        let compute_pageservers = shards
+            .into_iter()
+            .map(|shard| {
+                let ps_conf = env
+                    .get_pageserver_conf(shard.node_id)
+                    .expect("Unknown pageserver");
+                let (pg_host, pg_port) = parse_host_port(&ps_conf.listen_pg_addr)
+                    .expect("Unable to parse listen_pg_addr");
+                (pg_host, pg_port.unwrap_or(5432))
+            })
+            .collect::<Vec<_>>();
+
+        for (endpoint_name, endpoint) in &cplane.endpoints {
+            if endpoint.tenant_id == tenant_id && endpoint.status() == EndpointStatus::Running {
+                tracing::info!("Reconfiguring endpoint {}", endpoint_name,);
+                endpoint
+                    .reconfigure(compute_pageservers.clone(), stripe_size)
+                    .await?;
+            }
+        }
+
+        Ok(())
+    }
+
+    async fn do_notify_iteration(
+        &self,
+        client: &reqwest::Client,
+        url: &String,
+        reconfigure_request: &ComputeHookNotifyRequest,
+        cancel: &CancellationToken,
+    ) -> Result<(), NotifyError> {
+        let req = client.request(Method::PUT, url);
+        let req = if let Some(value) = &self.authorization_header {
+            req.header(reqwest::header::AUTHORIZATION, value)
+        } else {
+            req
+        };
+
+        tracing::info!(
+            "Sending notify request to {} ({:?})",
+            url,
+            reconfigure_request
+        );
+        let send_result = req.json(&reconfigure_request).send().await;
+        let response = match send_result {
+            Ok(r) => r,
+            Err(e) => return Err(e.into()),
+        };
+
+        // Treat all 2xx responses as success
+        if response.status() >= StatusCode::OK && response.status() < StatusCode::MULTIPLE_CHOICES {
+            if response.status() != StatusCode::OK {
+                // Non-200 2xx response: it doesn't make sense to retry, but this is unexpected, so
+                // log a warning.
+                tracing::warn!(
+                    "Unexpected 2xx response code {} from control plane",
+                    response.status()
+                );
+            }
+
+            return Ok(());
+        }
+
+        // Error response codes
+        match response.status() {
+            StatusCode::TOO_MANY_REQUESTS => {
+                // TODO: 429 handling should be global: set some state visible to other requests
+                // so that they will delay before starting, rather than all notifications trying
+                // once before backing off.
+                tokio::time::timeout(SLOWDOWN_DELAY, cancel.cancelled())
+                    .await
+                    .ok();
+                Err(NotifyError::SlowDown)
+            }
+            StatusCode::LOCKED => {
+                // Delay our retry if busy: the usual fast exponential backoff in backoff::retry
+                // is not appropriate
+                tokio::time::timeout(BUSY_DELAY, cancel.cancelled())
+                    .await
+                    .ok();
+                Err(NotifyError::Busy)
+            }
+            StatusCode::SERVICE_UNAVAILABLE
+            | StatusCode::GATEWAY_TIMEOUT
+            | StatusCode::BAD_GATEWAY => Err(NotifyError::Unavailable(response.status())),
+            StatusCode::BAD_REQUEST | StatusCode::UNAUTHORIZED | StatusCode::FORBIDDEN => {
+                Err(NotifyError::Fatal(response.status()))
+            }
+            _ => Err(NotifyError::Unexpected(response.status())),
+        }
+    }
+
+    async fn do_notify(
+        &self,
+        url: &String,
+        reconfigure_request: ComputeHookNotifyRequest,
+        cancel: &CancellationToken,
+    ) -> Result<(), NotifyError> {
+        let client = reqwest::Client::new();
+        backoff::retry(
+            || self.do_notify_iteration(&client, url, &reconfigure_request, cancel),
+            |e| matches!(e, NotifyError::Fatal(_) | NotifyError::Unexpected(_)),
+            3,
+            10,
+            "Send compute notification",
+            cancel,
+        )
+        .await
+        .ok_or_else(|| NotifyError::ShuttingDown)
+        .and_then(|x| x)
+    }
+
+    /// Call this to notify the compute (postgres) tier of new pageservers to use
+    /// for a tenant.  notify() is called by each shard individually, and this function
+    /// will decide whether an update to the tenant is sent.  An update is sent on the
+    /// condition that:
+    /// - We know a pageserver for every shard.
+    /// - All the shards have the same shard_count (i.e. we are not mid-split)
+    ///
+    /// Cancellation token enables callers to drop out, e.g. if calling from a Reconciler
+    /// that is cancelled.
+    ///
+    /// This function is fallible, including in the case that the control plane is transiently
+    /// unavailable.  A limited number of retries are done internally to efficiently hide short unavailability
+    /// periods, but we don't retry forever.  The **caller** is responsible for handling failures and
+    /// ensuring that they eventually call again to ensure that the compute is eventually notified of
+    /// the proper pageserver nodes for a tenant.
+    #[tracing::instrument(skip_all, fields(tenant_id=%tenant_shard_id.tenant_id, shard_id=%tenant_shard_id.shard_slug(), node_id))]
+    pub(super) async fn notify(
+        &self,
+        tenant_shard_id: TenantShardId,
+        node_id: NodeId,
+        stripe_size: ShardStripeSize,
+        cancel: &CancellationToken,
+    ) -> Result<(), NotifyError> {
+        let mut locked = self.state.lock().await;
+
+        use std::collections::hash_map::Entry;
+        let tenant = match locked.entry(tenant_shard_id.tenant_id) {
+            Entry::Vacant(e) => e.insert(ComputeHookTenant::new(
+                tenant_shard_id,
+                stripe_size,
+                node_id,
+            )),
+            Entry::Occupied(e) => {
+                let tenant = e.into_mut();
+                tenant.update(tenant_shard_id, stripe_size, node_id);
+                tenant
+            }
+        };
+
+        let reconfigure_request = tenant.maybe_reconfigure(tenant_shard_id.tenant_id);
+        let Some(reconfigure_request) = reconfigure_request else {
+            // The tenant doesn't yet have pageservers for all its shards: we won't notify anything
+            // until it does.
+            tracing::info!("Tenant isn't yet ready to emit a notification");
+            return Ok(());
+        };
+
+        if let Some(notify_url) = &self.config.compute_hook_url {
+            self.do_notify(notify_url, reconfigure_request, cancel)
+                .await
+        } else {
+            self.do_notify_local(reconfigure_request)
+                .await
+                .map_err(|e| {
+                    // This path is for testing only, so munge the error into our prod-style error type.
+                    tracing::error!("Local notification hook failed: {e}");
+                    NotifyError::Fatal(StatusCode::INTERNAL_SERVER_ERROR)
+                })
+        }
+    }
+}
+
+#[cfg(test)]
+pub(crate) mod tests {
+    use pageserver_api::shard::{ShardCount, ShardNumber};
+    use utils::id::TenantId;
+
+    use super::*;
+
+    #[test]
+    fn tenant_updates() -> anyhow::Result<()> {
+        let tenant_id = TenantId::generate();
+        let mut tenant_state = ComputeHookTenant::new(
+            TenantShardId {
+                tenant_id,
+                shard_count: ShardCount::new(0),
+                shard_number: ShardNumber(0),
+            },
+            ShardStripeSize(12345),
+            NodeId(1),
+        );
+
+        // An unsharded tenant is always ready to emit a notification
+        assert!(tenant_state.maybe_reconfigure(tenant_id).is_some());
+        assert_eq!(
+            tenant_state
+                .maybe_reconfigure(tenant_id)
+                .unwrap()
+                .shards
+                .len(),
+            1
+        );
+        assert!(tenant_state
+            .maybe_reconfigure(tenant_id)
+            .unwrap()
+            .stripe_size
+            .is_none());
+
+        // Writing the first shard of a multi-sharded situation (i.e. in a split)
+        // resets the tenant state and puts it in an non-notifying state (need to
+        // see all shards)
+        tenant_state.update(
+            TenantShardId {
+                tenant_id,
+                shard_count: ShardCount::new(2),
+                shard_number: ShardNumber(1),
+            },
+            ShardStripeSize(32768),
+            NodeId(1),
+        );
+        assert!(tenant_state.maybe_reconfigure(tenant_id).is_none());
+
+        // Writing the second shard makes it ready to notify
+        tenant_state.update(
+            TenantShardId {
+                tenant_id,
+                shard_count: ShardCount::new(2),
+                shard_number: ShardNumber(0),
+            },
+            ShardStripeSize(32768),
+            NodeId(1),
+        );
+
+        assert!(tenant_state.maybe_reconfigure(tenant_id).is_some());
+        assert_eq!(
+            tenant_state
+                .maybe_reconfigure(tenant_id)
+                .unwrap()
+                .shards
+                .len(),
+            2
+        );
+        assert_eq!(
+            tenant_state
+                .maybe_reconfigure(tenant_id)
+                .unwrap()
+                .stripe_size,
+            Some(ShardStripeSize(32768))
+        );
+
+        Ok(())
+    }
+}

control_plane/attachment_service/src/heartbeater.rs

@@ -6,7 +6,10 @@ use std::{
 };
 use tokio_util::sync::CancellationToken;
 
-use pageserver_api::{controller_api::NodeAvailability, models::PageserverUtilization};
+use pageserver_api::{
+    controller_api::{NodeAvailability, UtilizationScore},
+    models::PageserverUtilization,
+};
 
 use thiserror::Error;
 use utils::id::NodeId;
@@ -19,8 +22,7 @@ struct HeartbeaterTask {
 
     state: HashMap<NodeId, PageserverState>,
 
-    max_offline_interval: Duration,
-    max_warming_up_interval: Duration,
+    max_unavailable_interval: Duration,
     jwt_token: Option<String>,
 }
 
@@ -30,9 +32,6 @@ pub(crate) enum PageserverState {
         last_seen_at: Instant,
         utilization: PageserverUtilization,
     },
-    WarmingUp {
-        started_at: Instant,
-    },
     Offline,
 }
 
@@ -57,18 +56,12 @@ pub(crate) struct Heartbeater {
 impl Heartbeater {
     pub(crate) fn new(
         jwt_token: Option<String>,
-        max_offline_interval: Duration,
-        max_warming_up_interval: Duration,
+        max_unavailable_interval: Duration,
         cancel: CancellationToken,
     ) -> Self {
         let (sender, receiver) = tokio::sync::mpsc::unbounded_channel::<HeartbeatRequest>();
-        let mut heartbeater = HeartbeaterTask::new(
-            receiver,
-            jwt_token,
-            max_offline_interval,
-            max_warming_up_interval,
-            cancel,
-        );
+        let mut heartbeater =
+            HeartbeaterTask::new(receiver, jwt_token, max_unavailable_interval, cancel);
         tokio::task::spawn(async move { heartbeater.run().await });
 
         Self { sender }
@@ -84,12 +77,9 @@ impl Heartbeater {
                 pageservers,
                 reply: sender,
             })
-            .map_err(|_| HeartbeaterError::Cancel)?;
+            .unwrap();
 
-        receiver
-            .await
-            .map_err(|_| HeartbeaterError::Cancel)
-            .and_then(|x| x)
+        receiver.await.unwrap()
     }
 }
 
@@ -97,16 +87,14 @@ impl HeartbeaterTask {
     fn new(
         receiver: tokio::sync::mpsc::UnboundedReceiver<HeartbeatRequest>,
         jwt_token: Option<String>,
-        max_offline_interval: Duration,
-        max_warming_up_interval: Duration,
+        max_unavailable_interval: Duration,
         cancel: CancellationToken,
     ) -> Self {
         Self {
             receiver,
             cancel,
             state: HashMap::new(),
-            max_offline_interval,
-            max_warming_up_interval,
+            max_unavailable_interval,
             jwt_token,
         }
     }
@@ -143,12 +131,11 @@ impl HeartbeaterTask {
                 // Clone the node and mark it as available such that the request
                 // goes through to the pageserver even when the node is marked offline.
                 // This doesn't impact the availability observed by [`crate::service::Service`].
-                let mut node_clone = node.clone();
-                node_clone
-                    .set_availability(NodeAvailability::Active(PageserverUtilization::full()));
+                let mut node = node.clone();
+                node.set_availability(NodeAvailability::Active(UtilizationScore::worst()));
 
                 async move {
-                    let response = node_clone
+                    let response = node
                         .with_client_retries(
                             |client| async move { client.get_utilization().await },
                             &jwt_token,
@@ -173,12 +160,6 @@ impl HeartbeaterTask {
                             last_seen_at: Instant::now(),
                             utilization,
                         }
-                    } else if let NodeAvailability::WarmingUp(last_seen_at) =
-                        node.get_availability()
-                    {
-                        PageserverState::WarmingUp {
-                            started_at: *last_seen_at,
-                        }
                     } else {
                         PageserverState::Offline
                     };
@@ -204,66 +185,38 @@ impl HeartbeaterTask {
             }
         }
 
-        let mut warming_up = 0;
-        let mut offline = 0;
-        for state in new_state.values() {
-            match state {
-                PageserverState::WarmingUp { .. } => {
-                    warming_up += 1;
-                }
-                PageserverState::Offline { .. } => offline += 1,
-                PageserverState::Available { .. } => {}
-            }
-        }
-
-        tracing::info!(
-            "Heartbeat round complete for {} nodes, {} warming-up, {} offline",
-            new_state.len(),
-            warming_up,
-            offline
-        );
-
         let mut deltas = Vec::new();
         let now = Instant::now();
-        for (node_id, ps_state) in new_state.iter_mut() {
+        for (node_id, ps_state) in new_state {
             use std::collections::hash_map::Entry::*;
-            let entry = self.state.entry(*node_id);
+            let entry = self.state.entry(node_id);
 
             let mut needs_update = false;
             match entry {
                 Occupied(ref occ) => match (occ.get(), &ps_state) {
                     (PageserverState::Offline, PageserverState::Offline) => {}
                     (PageserverState::Available { last_seen_at, .. }, PageserverState::Offline) => {
-                        if now - *last_seen_at >= self.max_offline_interval {
-                            deltas.push((*node_id, ps_state.clone()));
+                        if now - *last_seen_at >= self.max_unavailable_interval {
+                            deltas.push((node_id, ps_state.clone()));
                             needs_update = true;
                         }
                     }
-                    (_, PageserverState::WarmingUp { started_at }) => {
-                        if now - *started_at >= self.max_warming_up_interval {
-                            *ps_state = PageserverState::Offline;
-                        }
-
-                        deltas.push((*node_id, ps_state.clone()));
-                        needs_update = true;
-                    }
                     _ => {
-                        deltas.push((*node_id, ps_state.clone()));
+                        deltas.push((node_id, ps_state.clone()));
                         needs_update = true;
                     }
                 },
                 Vacant(_) => {
-                    // This is a new node. Don't generate a delta for it.
-                    deltas.push((*node_id, ps_state.clone()));
+                    deltas.push((node_id, ps_state.clone()));
                 }
             }
 
             match entry {
                 Occupied(mut occ) if needs_update => {
-                    (*occ.get_mut()) = ps_state.clone();
+                    (*occ.get_mut()) = ps_state;
                 }
                 Vacant(vac) => {
-                    vac.insert(ps_state.clone());
+                    vac.insert(ps_state);
                 }
                 _ => {}
             }

control_plane/attachment_service/src/http.rs

@@ -2,26 +2,18 @@ use crate::metrics::{
     HttpRequestLatencyLabelGroup, HttpRequestStatusLabelGroup, PageserverRequestLabelGroup,
     METRICS_REGISTRY,
 };
-use crate::persistence::SafekeeperPersistence;
 use crate::reconciler::ReconcileError;
-use crate::service::{LeadershipStatus, Service, STARTUP_RECONCILE_TIMEOUT};
-use anyhow::Context;
+use crate::service::{Service, STARTUP_RECONCILE_TIMEOUT};
 use futures::Future;
 use hyper::header::CONTENT_TYPE;
 use hyper::{Body, Request, Response};
 use hyper::{StatusCode, Uri};
-use metrics::{BuildInfo, NeonMetrics};
-use pageserver_api::controller_api::{
-    MetadataHealthListOutdatedRequest, MetadataHealthListOutdatedResponse,
-    MetadataHealthListUnhealthyResponse, MetadataHealthUpdateRequest, MetadataHealthUpdateResponse,
-    ShardsPreferredAzsRequest, TenantCreateRequest,
-};
 use pageserver_api::models::{
-    TenantConfigRequest, TenantLocationConfigRequest, TenantShardSplitRequest,
-    TenantTimeTravelRequest, TimelineArchivalConfigRequest, TimelineCreateRequest,
+    TenantConfigRequest, TenantCreateRequest, TenantLocationConfigRequest, TenantShardSplitRequest,
+    TenantTimeTravelRequest, TimelineCreateRequest,
 };
 use pageserver_api::shard::TenantShardId;
-use pageserver_client::{mgmt_api, BlockUnblock};
+use pageserver_client::mgmt_api;
 use std::sync::Arc;
 use std::time::{Duration, Instant};
 use tokio_util::sync::CancellationToken;
@@ -42,8 +34,7 @@ use utils::{
 };
 
 use pageserver_api::controller_api::{
-    NodeAvailability, NodeConfigureRequest, NodeRegisterRequest, TenantPolicyRequest,
-    TenantShardMigrateRequest,
+    NodeAvailability, NodeConfigureRequest, NodeRegisterRequest, TenantShardMigrateRequest,
 };
 use pageserver_api::upcall_api::{ReAttachRequest, ValidateRequest};
 
@@ -52,19 +43,15 @@ use control_plane::storage_controller::{AttachHookRequest, InspectRequest};
 use routerify::Middleware;
 
 /// State available to HTTP request handlers
+#[derive(Clone)]
 pub struct HttpState {
     service: Arc<crate::service::Service>,
     auth: Option<Arc<SwappableJwtAuth>>,
-    neon_metrics: NeonMetrics,
     allowlist_routes: Vec<Uri>,
 }
 
 impl HttpState {
-    pub fn new(
-        service: Arc<crate::service::Service>,
-        auth: Option<Arc<SwappableJwtAuth>>,
-        build_info: BuildInfo,
-    ) -> Self {
+    pub fn new(service: Arc<crate::service::Service>, auth: Option<Arc<SwappableJwtAuth>>) -> Self {
         let allowlist_routes = ["/status", "/ready", "/metrics"]
             .iter()
             .map(|v| v.parse().unwrap())
@@ -72,7 +59,6 @@ impl HttpState {
         Self {
             service,
             auth,
-            neon_metrics: NeonMetrics::new(build_info),
             allowlist_routes,
         }
     }
@@ -102,7 +88,7 @@ async fn handle_validate(mut req: Request<Body>) -> Result<Response<Body>, ApiEr
 
     let validate_req = json_request::<ValidateRequest>(&mut req).await?;
     let state = get_state(&req);
-    json_response(StatusCode::OK, state.service.validate(validate_req).await?)
+    json_response(StatusCode::OK, state.service.validate(validate_req))
 }
 
 /// Call into this before attaching a tenant to a pageserver, to acquire a generation number
@@ -148,6 +134,52 @@ async fn handle_tenant_create(
     )
 }
 
+// For tenant and timeline deletions, which both implement an "initially return 202, then 404 once
+// we're done" semantic, we wrap with a retry loop to expose a simpler API upstream.  This avoids
+// needing to track a "deleting" state for tenants.
+async fn deletion_wrapper<R, F>(service: Arc<Service>, f: F) -> Result<Response<Body>, ApiError>
+where
+    R: std::future::Future<Output = Result<StatusCode, ApiError>> + Send + 'static,
+    F: Fn(Arc<Service>) -> R + Send + Sync + 'static,
+{
+    let started_at = Instant::now();
+    // To keep deletion reasonably snappy for small tenants, initially check after 1 second if deletion
+    // completed.
+    let mut retry_period = Duration::from_secs(1);
+    // On subsequent retries, wait longer.
+    let max_retry_period = Duration::from_secs(5);
+    // Enable callers with a 30 second request timeout to reliably get a response
+    let max_wait = Duration::from_secs(25);
+
+    loop {
+        let status = f(service.clone()).await?;
+        match status {
+            StatusCode::ACCEPTED => {
+                tracing::info!("Deletion accepted, waiting to try again...");
+                tokio::time::sleep(retry_period).await;
+                retry_period = max_retry_period;
+            }
+            StatusCode::NOT_FOUND => {
+                tracing::info!("Deletion complete");
+                return json_response(StatusCode::OK, ());
+            }
+            _ => {
+                tracing::warn!("Unexpected status {status}");
+                return json_response(status, ());
+            }
+        }
+
+        let now = Instant::now();
+        if now + retry_period > started_at + max_wait {
+            tracing::info!("Deletion timed out waiting for 404");
+            // REQUEST_TIMEOUT would be more appropriate, but CONFLICT is already part of
+            // the pageserver's swagger definition for this endpoint, and has the same desired
+            // effect of causing the control plane to retry later.
+            return json_response(StatusCode::CONFLICT, ());
+        }
+    }
+}
+
 async fn handle_tenant_location_config(
     service: Arc<Service>,
     mut req: Request<Body>,
@@ -219,12 +251,6 @@ async fn handle_tenant_time_travel_remote_storage(
     json_response(StatusCode::OK, ())
 }
 
-fn map_reqwest_hyper_status(status: reqwest::StatusCode) -> Result<hyper::StatusCode, ApiError> {
-    hyper::StatusCode::from_u16(status.as_u16())
-        .context("invalid status code")
-        .map_err(ApiError::InternalServerError)
-}
-
 async fn handle_tenant_secondary_download(
     service: Arc<Service>,
     req: Request<Body>,
@@ -233,7 +259,7 @@ async fn handle_tenant_secondary_download(
     let wait = parse_query_param(&req, "wait_ms")?.map(Duration::from_millis);
 
     let (status, progress) = service.tenant_secondary_download(tenant_id, wait).await?;
-    json_response(map_reqwest_hyper_status(status)?, progress)
+    json_response(status, progress)
 }
 
 async fn handle_tenant_delete(
@@ -243,17 +269,10 @@ async fn handle_tenant_delete(
     let tenant_id: TenantId = parse_request_param(&req, "tenant_id")?;
     check_permissions(&req, Scope::PageServerApi)?;
 
-    let status_code = service
-        .tenant_delete(tenant_id)
-        .await
-        .and_then(map_reqwest_hyper_status)?;
-
-    if status_code == StatusCode::NOT_FOUND {
-        // The pageserver uses 404 for successful deletion, but we use 200
-        json_response(StatusCode::OK, ())
-    } else {
-        json_response(status_code, ())
-    }
+    deletion_wrapper(service, move |service| async move {
+        service.tenant_delete(tenant_id).await
+    })
+    .await
 }
 
 async fn handle_tenant_timeline_create(
@@ -281,111 +300,12 @@ async fn handle_tenant_timeline_delete(
 
     let timeline_id: TimelineId = parse_request_param(&req, "timeline_id")?;
 
-    // For timeline deletions, which both implement an "initially return 202, then 404 once
-    // we're done" semantic, we wrap with a retry loop to expose a simpler API upstream.
-    async fn deletion_wrapper<R, F>(service: Arc<Service>, f: F) -> Result<Response<Body>, ApiError>
-    where
-        R: std::future::Future<Output = Result<StatusCode, ApiError>> + Send + 'static,
-        F: Fn(Arc<Service>) -> R + Send + Sync + 'static,
-    {
-        let started_at = Instant::now();
-        // To keep deletion reasonably snappy for small tenants, initially check after 1 second if deletion
-        // completed.
-        let mut retry_period = Duration::from_secs(1);
-        // On subsequent retries, wait longer.
-        let max_retry_period = Duration::from_secs(5);
-        // Enable callers with a 30 second request timeout to reliably get a response
-        let max_wait = Duration::from_secs(25);
-
-        loop {
-            let status = f(service.clone()).await?;
-            match status {
-                StatusCode::ACCEPTED => {
-                    tracing::info!("Deletion accepted, waiting to try again...");
-                    tokio::time::sleep(retry_period).await;
-                    retry_period = max_retry_period;
-                }
-                StatusCode::NOT_FOUND => {
-                    tracing::info!("Deletion complete");
-                    return json_response(StatusCode::OK, ());
-                }
-                _ => {
-                    tracing::warn!("Unexpected status {status}");
-                    return json_response(status, ());
-                }
-            }
-
-            let now = Instant::now();
-            if now + retry_period > started_at + max_wait {
-                tracing::info!("Deletion timed out waiting for 404");
-                // REQUEST_TIMEOUT would be more appropriate, but CONFLICT is already part of
-                // the pageserver's swagger definition for this endpoint, and has the same desired
-                // effect of causing the control plane to retry later.
-                return json_response(StatusCode::CONFLICT, ());
-            }
-        }
-    }
-
     deletion_wrapper(service, move |service| async move {
-        service
-            .tenant_timeline_delete(tenant_id, timeline_id)
-            .await
-            .and_then(map_reqwest_hyper_status)
+        service.tenant_timeline_delete(tenant_id, timeline_id).await
     })
     .await
 }
 
-async fn handle_tenant_timeline_archival_config(
-    service: Arc<Service>,
-    mut req: Request<Body>,
-) -> Result<Response<Body>, ApiError> {
-    let tenant_id: TenantId = parse_request_param(&req, "tenant_id")?;
-    check_permissions(&req, Scope::PageServerApi)?;
-
-    let timeline_id: TimelineId = parse_request_param(&req, "timeline_id")?;
-
-    let create_req = json_request::<TimelineArchivalConfigRequest>(&mut req).await?;
-
-    service
-        .tenant_timeline_archival_config(tenant_id, timeline_id, create_req)
-        .await?;
-
-    json_response(StatusCode::OK, ())
-}
-
-async fn handle_tenant_timeline_detach_ancestor(
-    service: Arc<Service>,
-    req: Request<Body>,
-) -> Result<Response<Body>, ApiError> {
-    let tenant_id: TenantId = parse_request_param(&req, "tenant_id")?;
-    check_permissions(&req, Scope::PageServerApi)?;
-
-    let timeline_id: TimelineId = parse_request_param(&req, "timeline_id")?;
-
-    let res = service
-        .tenant_timeline_detach_ancestor(tenant_id, timeline_id)
-        .await?;
-
-    json_response(StatusCode::OK, res)
-}
-
-async fn handle_tenant_timeline_block_unblock_gc(
-    service: Arc<Service>,
-    req: Request<Body>,
-    dir: BlockUnblock,
-) -> Result<Response<Body>, ApiError> {
-    let tenant_id: TenantId = parse_request_param(&req, "tenant_id")?;
-    check_permissions(&req, Scope::PageServerApi)?;
-
-    let timeline_id: TimelineId = parse_request_param(&req, "timeline_id")?;
-
-    service
-        .tenant_timeline_block_unblock_gc(tenant_id, timeline_id, dir)
-        .await?;
-
-    json_response(StatusCode::OK, ())
-}
-
 async fn handle_tenant_timeline_passthrough(
     service: Arc<Service>,
     req: Request<Body>,
@@ -444,9 +364,11 @@ async fn handle_tenant_timeline_passthrough(
     }
 
     // We have a reqest::Response, would like a http::Response
-    let mut builder = hyper::Response::builder().status(map_reqwest_hyper_status(resp.status())?);
+    let mut builder = hyper::Response::builder()
+        .status(resp.status())
+        .version(resp.version());
     for (k, v) in resp.headers() {
-        builder = builder.header(k.as_str(), v.as_bytes());
+        builder = builder.header(k, v);
     }
 
     let response = builder
@@ -470,21 +392,12 @@ async fn handle_tenant_describe(
     service: Arc<Service>,
     req: Request<Body>,
 ) -> Result<Response<Body>, ApiError> {
-    check_permissions(&req, Scope::Scrubber)?;
+    check_permissions(&req, Scope::Admin)?;
 
     let tenant_id: TenantId = parse_request_param(&req, "tenant_id")?;
     json_response(StatusCode::OK, service.tenant_describe(tenant_id)?)
 }
 
-async fn handle_tenant_list(
-    service: Arc<Service>,
-    req: Request<Body>,
-) -> Result<Response<Body>, ApiError> {
-    check_permissions(&req, Scope::Admin)?;
-
-    json_response(StatusCode::OK, service.tenant_list())
-}
-
 async fn handle_node_register(mut req: Request<Body>) -> Result<Response<Body>, ApiError> {
     check_permissions(&req, Scope::Admin)?;
 
@@ -498,10 +411,7 @@ async fn handle_node_list(req: Request<Body>) -> Result<Response<Body>, ApiError
     check_permissions(&req, Scope::Admin)?;
 
     let state = get_state(&req);
-    let nodes = state.service.node_list().await?;
-    let api_nodes = nodes.into_iter().map(|n| n.describe()).collect::<Vec<_>>();
-
-    json_response(StatusCode::OK, api_nodes)
+    json_response(StatusCode::OK, state.service.node_list().await?)
 }
 
 async fn handle_node_drop(req: Request<Body>) -> Result<Response<Body>, ApiError> {
@@ -512,14 +422,6 @@ async fn handle_node_drop(req: Request<Body>) -> Result<Response<Body>, ApiError
     json_response(StatusCode::OK, state.service.node_drop(node_id).await?)
 }
 
-async fn handle_node_delete(req: Request<Body>) -> Result<Response<Body>, ApiError> {
-    check_permissions(&req, Scope::Admin)?;
-
-    let state = get_state(&req);
-    let node_id: NodeId = parse_request_param(&req, "node_id")?;
-    json_response(StatusCode::OK, state.service.node_delete(node_id).await?)
-}
-
 async fn handle_node_configure(mut req: Request<Body>) -> Result<Response<Body>, ApiError> {
     check_permissions(&req, Scope::Admin)?;
 
@@ -536,7 +438,7 @@ async fn handle_node_configure(mut req: Request<Body>) -> Result<Response<Body>,
         StatusCode::OK,
         state
             .service
-            .external_node_configure(
+            .node_configure(
                 config_req.node_id,
                 config_req.availability.map(NodeAvailability::from),
                 config_req.scheduling,
@@ -545,130 +447,6 @@ async fn handle_node_configure(mut req: Request<Body>) -> Result<Response<Body>,
     )
 }
 
-async fn handle_node_status(req: Request<Body>) -> Result<Response<Body>, ApiError> {
-    check_permissions(&req, Scope::Admin)?;
-
-    let state = get_state(&req);
-    let node_id: NodeId = parse_request_param(&req, "node_id")?;
-
-    let node_status = state.service.get_node(node_id).await?;
-
-    json_response(StatusCode::OK, node_status)
-}
-
-async fn handle_node_shards(req: Request<Body>) -> Result<Response<Body>, ApiError> {
-    check_permissions(&req, Scope::Admin)?;
-
-    let state = get_state(&req);
-    let node_id: NodeId = parse_request_param(&req, "node_id")?;
-
-    let node_status = state.service.get_node_shards(node_id).await?;
-
-    json_response(StatusCode::OK, node_status)
-}
-
-async fn handle_get_leader(req: Request<Body>) -> Result<Response<Body>, ApiError> {
-    check_permissions(&req, Scope::Admin)?;
-
-    let state = get_state(&req);
-    let leader = state.service.get_leader().await.map_err(|err| {
-        ApiError::InternalServerError(anyhow::anyhow!(
-            "Failed to read leader from database: {err}"
-        ))
-    })?;
-
-    json_response(StatusCode::OK, leader)
-}
-
-async fn handle_node_drain(req: Request<Body>) -> Result<Response<Body>, ApiError> {
-    check_permissions(&req, Scope::Admin)?;
-
-    let state = get_state(&req);
-    let node_id: NodeId = parse_request_param(&req, "node_id")?;
-
-    state.service.start_node_drain(node_id).await?;
-
-    json_response(StatusCode::ACCEPTED, ())
-}
-
-async fn handle_cancel_node_drain(req: Request<Body>) -> Result<Response<Body>, ApiError> {
-    check_permissions(&req, Scope::Admin)?;
-
-    let state = get_state(&req);
-    let node_id: NodeId = parse_request_param(&req, "node_id")?;
-
-    state.service.cancel_node_drain(node_id).await?;
-
-    json_response(StatusCode::ACCEPTED, ())
-}
-
-async fn handle_node_fill(req: Request<Body>) -> Result<Response<Body>, ApiError> {
-    check_permissions(&req, Scope::Admin)?;
-
-    let state = get_state(&req);
-    let node_id: NodeId = parse_request_param(&req, "node_id")?;
-
-    state.service.start_node_fill(node_id).await?;
-
-    json_response(StatusCode::ACCEPTED, ())
-}
-
-async fn handle_cancel_node_fill(req: Request<Body>) -> Result<Response<Body>, ApiError> {
-    check_permissions(&req, Scope::Admin)?;
-
-    let state = get_state(&req);
-    let node_id: NodeId = parse_request_param(&req, "node_id")?;
-
-    state.service.cancel_node_fill(node_id).await?;
-
-    json_response(StatusCode::ACCEPTED, ())
-}
-
-async fn handle_metadata_health_update(mut req: Request<Body>) -> Result<Response<Body>, ApiError> {
-    check_permissions(&req, Scope::Scrubber)?;
-
-    let update_req = json_request::<MetadataHealthUpdateRequest>(&mut req).await?;
-    let state = get_state(&req);
-
-    state.service.metadata_health_update(update_req).await?;
-
-    json_response(StatusCode::OK, MetadataHealthUpdateResponse {})
-}
-
-async fn handle_metadata_health_list_unhealthy(
-    req: Request<Body>,
-) -> Result<Response<Body>, ApiError> {
-    check_permissions(&req, Scope::Admin)?;
-
-    let state = get_state(&req);
-    let unhealthy_tenant_shards = state.service.metadata_health_list_unhealthy().await?;
-
-    json_response(
-        StatusCode::OK,
-        MetadataHealthListUnhealthyResponse {
-            unhealthy_tenant_shards,
-        },
-    )
-}
-
-async fn handle_metadata_health_list_outdated(
-    mut req: Request<Body>,
-) -> Result<Response<Body>, ApiError> {
-    check_permissions(&req, Scope::Admin)?;
-
-    let list_outdated_req = json_request::<MetadataHealthListOutdatedRequest>(&mut req).await?;
-    let state = get_state(&req);
-    let health_records = state
-        .service
-        .metadata_health_list_outdated(list_outdated_req.not_scrubbed_for)
-        .await?;
-
-    json_response(
-        StatusCode::OK,
-        MetadataHealthListOutdatedResponse { health_records },
-    )
-}
-
 async fn handle_tenant_shard_split(
     service: Arc<Service>,
     mut req: Request<Body>,
@@ -700,41 +478,6 @@ async fn handle_tenant_shard_migrate(
     )
 }
 
-async fn handle_tenant_update_policy(mut req: Request<Body>) -> Result<Response<Body>, ApiError> {
-    check_permissions(&req, Scope::Admin)?;
-
-    let tenant_id: TenantId = parse_request_param(&req, "tenant_id")?;
-    let update_req = json_request::<TenantPolicyRequest>(&mut req).await?;
-    let state = get_state(&req);
-
-    json_response(
-        StatusCode::OK,
-        state
-            .service
-            .tenant_update_policy(tenant_id, update_req)
-            .await?,
-    )
-}
-
-async fn handle_update_preferred_azs(mut req: Request<Body>) -> Result<Response<Body>, ApiError> {
-    check_permissions(&req, Scope::Admin)?;
-
-    let azs_req = json_request::<ShardsPreferredAzsRequest>(&mut req).await?;
-    let state = get_state(&req);
-
-    json_response(
-        StatusCode::OK,
-        state.service.update_shards_preferred_azs(azs_req).await?,
-    )
-}
-
-async fn handle_step_down(req: Request<Body>) -> Result<Response<Body>, ApiError> {
-    check_permissions(&req, Scope::Admin)?;
-
-    let state = get_state(&req);
-    json_response(StatusCode::OK, state.service.step_down().await)
-}
-
 async fn handle_tenant_drop(req: Request<Body>) -> Result<Response<Body>, ApiError> {
     let tenant_id: TenantId = parse_request_param(&req, "tenant_id")?;
     check_permissions(&req, Scope::PageServerApi)?;
@@ -744,18 +487,6 @@ async fn handle_tenant_drop(req: Request<Body>) -> Result<Response<Body>, ApiErr
     json_response(StatusCode::OK, state.service.tenant_drop(tenant_id).await?)
 }
 
-async fn handle_tenant_import(req: Request<Body>) -> Result<Response<Body>, ApiError> {
-    let tenant_id: TenantId = parse_request_param(&req, "tenant_id")?;
-    check_permissions(&req, Scope::PageServerApi)?;
-
-    let state = get_state(&req);
-
-    json_response(
-        StatusCode::OK,
-        state.service.tenant_import(tenant_id).await?,
-    )
-}
-
 async fn handle_tenants_dump(req: Request<Body>) -> Result<Response<Body>, ApiError> {
     check_permissions(&req, Scope::Admin)?;
 
@@ -778,14 +509,6 @@ async fn handle_consistency_check(req: Request<Body>) -> Result<Response<Body>,
     json_response(StatusCode::OK, state.service.consistency_check().await?)
 }
 
-async fn handle_reconcile_all(req: Request<Body>) -> Result<Response<Body>, ApiError> {
-    check_permissions(&req, Scope::Admin)?;
-
-    let state = get_state(&req);
-
-    json_response(StatusCode::OK, state.service.reconcile_all_now().await?)
-}
-
 /// Status endpoint is just used for checking that our HTTP listener is up
 async fn handle_status(_req: Request<Body>) -> Result<Response<Body>, ApiError> {
     json_response(StatusCode::OK, ())
@@ -808,55 +531,6 @@ impl From<ReconcileError> for ApiError {
     }
 }
 
-/// Return the safekeeper record by instance id, or 404.
-///
-/// Not used by anything except manual testing.
-async fn handle_get_safekeeper(req: Request<Body>) -> Result<Response<Body>, ApiError> {
-    check_permissions(&req, Scope::Admin)?;
-
-    let id = parse_request_param::<i64>(&req, "id")?;
-
-    let state = get_state(&req);
-
-    let res = state.service.get_safekeeper(id).await;
-
-    match res {
-        Ok(b) => json_response(StatusCode::OK, b),
-        Err(crate::persistence::DatabaseError::Query(diesel::result::Error::NotFound)) => {
-            Err(ApiError::NotFound("unknown instance_id".into()))
-        }
-        Err(other) => Err(other.into()),
-    }
-}
-
-/// Used as part of deployment scripts.
-///
-/// Assumes information is only relayed to storage controller after first selecting an unique id on
-/// control plane database, which means we have an id field in the request and payload.
-async fn handle_upsert_safekeeper(mut req: Request<Body>) -> Result<Response<Body>, ApiError> {
-    check_permissions(&req, Scope::Admin)?;
-
-    let body = json_request::<SafekeeperPersistence>(&mut req).await?;
-    let id = parse_request_param::<i64>(&req, "id")?;
-
-    if id != body.id {
-        // it should be repeated
-        return Err(ApiError::BadRequest(anyhow::anyhow!(
-            "id mismatch: url={id:?}, body={:?}",
-            body.id
-        )));
-    }
-
-    let state = get_state(&req);
-
-    state.service.upsert_safekeeper(body).await?;
-
-    Ok(Response::builder()
-        .status(StatusCode::NO_CONTENT)
-        .body(Body::empty())
-        .unwrap())
-}
-
 /// Common wrapper for request handlers that call into Service and will operate on tenants: they must only
 /// be allowed to run if Service has finished its initial reconciliation.
 async fn tenant_service_handler<R, H>(
@@ -891,17 +565,9 @@ where
     .await
 }
 
-/// Check if the required scope is held in the request's token, or if the request has
-/// a token with 'admin' scope then always permit it.
 fn check_permissions(request: &Request<Body>, required_scope: Scope) -> Result<(), ApiError> {
     check_permission_with(request, |claims| {
-        match crate::auth::check_permission(claims, required_scope) {
-            Err(e) => match crate::auth::check_permission(claims, Scope::Admin) {
-                Ok(()) => Ok(()),
-                Err(_) => Err(e),
-            },
-            Ok(()) => Ok(()),
-        }
+        crate::auth::check_permission(claims, required_scope)
     })
 }
 
@@ -911,47 +577,6 @@ struct RequestMeta {
     at: Instant,
 }
 
-pub fn prologue_leadership_status_check_middleware<
-    B: hyper::body::HttpBody + Send + Sync + 'static,
->() -> Middleware<B, ApiError> {
-    Middleware::pre(move |req| async move {
-        let state = get_state(&req);
-        let leadership_status = state.service.get_leadership_status();
-
-        enum AllowedRoutes<'a> {
-            All,
-            Some(Vec<&'a str>),
-        }
-
-        let allowed_routes = match leadership_status {
-            LeadershipStatus::Leader => AllowedRoutes::All,
-            LeadershipStatus::SteppedDown => {
-                // TODO: does it make sense to allow /status here?
-                AllowedRoutes::Some(["/control/v1/step_down", "/status", "/metrics"].to_vec())
-            }
-            LeadershipStatus::Candidate => {
-                AllowedRoutes::Some(["/ready", "/status", "/metrics"].to_vec())
-            }
-        };
-
-        let uri = req.uri().to_string();
-        match allowed_routes {
-            AllowedRoutes::All => Ok(req),
-            AllowedRoutes::Some(allowed) if allowed.contains(&uri.as_str()) => Ok(req),
-            _ => {
-                tracing::info!(
-                    "Request {} not allowed due to current leadership state",
-                    req.uri()
-                );
-
-                Err(ApiError::ResourceUnavailable(
-                    format!("Current leadership status is {leadership_status}").into(),
-                ))
-            }
-        }
-    })
-}
-
 fn prologue_metrics_middleware<B: hyper::body::HttpBody + Send + Sync + 'static>(
 ) -> Middleware<B, ApiError> {
     Middleware::pre(move |req| async move {
@@ -1002,11 +627,10 @@ fn epilogue_metrics_middleware<B: hyper::body::HttpBody + Send + Sync + 'static>
     })
 }
 
-pub async fn measured_metrics_handler(req: Request<Body>) -> Result<Response<Body>, ApiError> {
+pub async fn measured_metrics_handler(_req: Request<Body>) -> Result<Response<Body>, ApiError> {
     pub const TEXT_FORMAT: &str = "text/plain; version=0.0.4";
 
-    let state = get_state(&req);
-    let payload = crate::metrics::METRICS_REGISTRY.encode(&state.neon_metrics);
+    let payload = crate::metrics::METRICS_REGISTRY.encode();
     let response = Response::builder()
         .status(200)
         .header(CONTENT_TYPE, TEXT_FORMAT)
@@ -1035,10 +659,8 @@ where
 pub fn make_router(
     service: Arc<Service>,
     auth: Option<Arc<SwappableJwtAuth>>,
-    build_info: BuildInfo,
 ) -> RouterBuilder<hyper::Body, ApiError> {
     let mut router = endpoint::make_router()
-        .middleware(prologue_leadership_status_check_middleware())
         .middleware(prologue_metrics_middleware())
         .middleware(epilogue_metrics_middleware());
     if auth.is_some() {
@@ -1053,7 +675,7 @@ pub fn make_router(
     }
 
     router
-        .data(Arc::new(HttpState::new(service, auth, build_info)))
+        .data(Arc::new(HttpState::new(service, auth)))
         .get("/metrics", |r| {
             named_request_span(r, measured_metrics_handler, RequestName("metrics"))
         })
@@ -1084,13 +706,6 @@ pub fn make_router(
         .post("/debug/v1/node/:node_id/drop", |r| {
             named_request_span(r, handle_node_drop, RequestName("debug_v1_node_drop"))
         })
-        .post("/debug/v1/tenant/:tenant_id/import", |r| {
-            named_request_span(
-                r,
-                handle_tenant_import,
-                RequestName("debug_v1_tenant_import"),
-            )
-        })
         .get("/debug/v1/tenant", |r| {
             named_request_span(r, handle_tenants_dump, RequestName("debug_v1_tenant"))
         })
@@ -1111,9 +726,6 @@ pub fn make_router(
                 RequestName("debug_v1_consistency_check"),
             )
         })
-        .post("/debug/v1/reconcile_all", |r| {
-            request_span(r, handle_reconcile_all)
-        })
         .put("/debug/v1/failpoints", |r| {
             request_span(r, |r| failpoints_handler(r, CancellationToken::new()))
         })
@@ -1121,9 +733,6 @@ pub fn make_router(
         .post("/control/v1/node", |r| {
             named_request_span(r, handle_node_register, RequestName("control_v1_node"))
         })
-        .delete("/control/v1/node/:node_id", |r| {
-            named_request_span(r, handle_node_delete, RequestName("control_v1_node_delete"))
-        })
         .get("/control/v1/node", |r| {
             named_request_span(r, handle_node_list, RequestName("control_v1_node"))
         })
@@ -1134,61 +743,6 @@ pub fn make_router(
                 RequestName("control_v1_node_config"),
             )
         })
-        .get("/control/v1/node/:node_id", |r| {
-            named_request_span(r, handle_node_status, RequestName("control_v1_node_status"))
-        })
-        .get("/control/v1/node/:node_id/shards", |r| {
-            named_request_span(
-                r,
-                handle_node_shards,
-                RequestName("control_v1_node_describe"),
-            )
-        })
-        .get("/control/v1/leader", |r| {
-            named_request_span(r, handle_get_leader, RequestName("control_v1_get_leader"))
-        })
-        .put("/control/v1/node/:node_id/drain", |r| {
-            named_request_span(r, handle_node_drain, RequestName("control_v1_node_drain"))
-        })
-        .delete("/control/v1/node/:node_id/drain", |r| {
-            named_request_span(
-                r,
-                handle_cancel_node_drain,
-                RequestName("control_v1_cancel_node_drain"),
-            )
-        })
-        .put("/control/v1/node/:node_id/fill", |r| {
-            named_request_span(r, handle_node_fill, RequestName("control_v1_node_fill"))
-        })
-        .delete("/control/v1/node/:node_id/fill", |r| {
-            named_request_span(
-                r,
-                handle_cancel_node_fill,
-                RequestName("control_v1_cancel_node_fill"),
-            )
-        })
-        // Metadata health operations
-        .post("/control/v1/metadata_health/update", |r| {
-            named_request_span(
-                r,
-                handle_metadata_health_update,
-                RequestName("control_v1_metadata_health_update"),
-            )
-        })
-        .get("/control/v1/metadata_health/unhealthy", |r| {
-            named_request_span(
-                r,
-                handle_metadata_health_list_unhealthy,
-                RequestName("control_v1_metadata_health_list_unhealthy"),
-            )
-        })
-        .post("/control/v1/metadata_health/outdated", |r| {
-            named_request_span(
-                r,
-                handle_metadata_health_list_outdated,
-                RequestName("control_v1_metadata_health_list_outdated"),
-            )
-        })
         // Tenant Shard operations
         .put("/control/v1/tenant/:tenant_shard_id/migrate", |r| {
             tenant_service_handler(
@@ -1211,33 +765,6 @@ pub fn make_router(
                 RequestName("control_v1_tenant_describe"),
             )
         })
-        .get("/control/v1/tenant", |r| {
-            tenant_service_handler(r, handle_tenant_list, RequestName("control_v1_tenant_list"))
-        })
-        .put("/control/v1/tenant/:tenant_id/policy", |r| {
-            named_request_span(
-                r,
-                handle_tenant_update_policy,
-                RequestName("control_v1_tenant_policy"),
-            )
-        })
-        .put("/control/v1/preferred_azs", |r| {
-            named_request_span(
-                r,
-                handle_update_preferred_azs,
-                RequestName("control_v1_preferred_azs"),
-            )
-        })
-        .put("/control/v1/step_down", |r| {
-            named_request_span(r, handle_step_down, RequestName("control_v1_step_down"))
-        })
-        .get("/control/v1/safekeeper/:id", |r| {
-            named_request_span(r, handle_get_safekeeper, RequestName("v1_safekeeper"))
-        })
-        .post("/control/v1/safekeeper/:id", |r| {
-            // id is in the body
-            named_request_span(r, handle_upsert_safekeeper, RequestName("v1_safekeeper"))
-        })
         // Tenant operations
         // The ^/v1/ endpoints act as a "Virtual Pageserver", enabling shard-naive clients to call into
         // this service to manage tenants that actually consist of many tenant shards, as if they are a single entity.
@@ -1289,47 +816,7 @@ pub fn make_router(
                 RequestName("v1_tenant_timeline"),
             )
         })
-        .post(
-            "/v1/tenant/:tenant_id/timeline/:timeline_id/archival_config",
-            |r| {
-                tenant_service_handler(
-                    r,
-                    handle_tenant_timeline_archival_config,
-                    RequestName("v1_tenant_timeline_archival_config"),
-                )
-            },
-        )
-        .put(
-            "/v1/tenant/:tenant_id/timeline/:timeline_id/detach_ancestor",
-            |r| {
-                tenant_service_handler(
-                    r,
-                    handle_tenant_timeline_detach_ancestor,
-                    RequestName("v1_tenant_timeline_detach_ancestor"),
-                )
-            },
-        )
-        .post(
-            "/v1/tenant/:tenant_id/timeline/:timeline_id/block_gc",
-            |r| {
-                tenant_service_handler(
-                    r,
-                    |s, r| handle_tenant_timeline_block_unblock_gc(s, r, BlockUnblock::Block),
-                    RequestName("v1_tenant_timeline_block_unblock_gc"),
-                )
-            },
-        )
-        .post(
-            "/v1/tenant/:tenant_id/timeline/:timeline_id/unblock_gc",
-            |r| {
-                tenant_service_handler(
-                    r,
-                    |s, r| handle_tenant_timeline_block_unblock_gc(s, r, BlockUnblock::Unblock),
-                    RequestName("v1_tenant_timeline_block_unblock_gc"),
-                )
-            },
-        )
-        // Tenant detail GET passthrough to shard zero:
+        // Tenant detail GET passthrough to shard zero
         .get("/v1/tenant/:tenant_id", |r| {
             tenant_service_handler(
                 r,
@@ -1337,14 +824,13 @@ pub fn make_router(
                 RequestName("v1_tenant_passthrough"),
             )
         })
-        // The `*` in the  URL is a wildcard: any tenant/timeline GET APIs on the pageserver
-        // are implicitly exposed here.  This must be last in the list to avoid
-        // taking precedence over other GET methods we might implement by hand.
-        .get("/v1/tenant/:tenant_id/*", |r| {
+        // Timeline GET passthrough to shard zero.  Note that the `*` in the URL is a wildcard: any future
+        // timeline GET APIs will be implicitly included.
+        .get("/v1/tenant/:tenant_id/timeline*", |r| {
             tenant_service_handler(
                 r,
                 handle_tenant_timeline_passthrough,
-                RequestName("v1_tenant_passthrough"),
+                RequestName("v1_tenant_timeline_passthrough"),
             )
         })
 }

control_plane/attachment_service/src/id_lock_map.rs

@@ -0,0 +1,54 @@
+use std::{collections::HashMap, sync::Arc};
+
+/// A map of locks covering some arbitrary identifiers. Useful if you have a collection of objects but don't
+/// want to embed a lock in each one, or if your locking granularity is different to your object granularity.
+/// For example, used in the storage controller where the objects are tenant shards, but sometimes locking
+/// is needed at a tenant-wide granularity.
+pub(crate) struct IdLockMap<T>
+where
+    T: Eq + PartialEq + std::hash::Hash,
+{
+    /// A synchronous lock for getting/setting the async locks that our callers will wait on.
+    entities: std::sync::Mutex<std::collections::HashMap<T, Arc<tokio::sync::RwLock<()>>>>,
+}
+
+impl<T> IdLockMap<T>
+where
+    T: Eq + PartialEq + std::hash::Hash,
+{
+    pub(crate) fn shared(
+        &self,
+        key: T,
+    ) -> impl std::future::Future<Output = tokio::sync::OwnedRwLockReadGuard<()>> {
+        let mut locked = self.entities.lock().unwrap();
+        let entry = locked.entry(key).or_default();
+        entry.clone().read_owned()
+    }
+
+    pub(crate) fn exclusive(
+        &self,
+        key: T,
+    ) -> impl std::future::Future<Output = tokio::sync::OwnedRwLockWriteGuard<()>> {
+        let mut locked = self.entities.lock().unwrap();
+        let entry = locked.entry(key).or_default();
+        entry.clone().write_owned()
+    }
+
+    /// Rather than building a lock guard that re-takes the [`Self::entities`] lock, we just do
+    /// periodic housekeeping to avoid the map growing indefinitely
+    pub(crate) fn housekeeping(&self) {
+        let mut locked = self.entities.lock().unwrap();
+        locked.retain(|_k, lock| lock.try_write().is_err())
+    }
+}
+
+impl<T> Default for IdLockMap<T>
+where
+    T: Eq + PartialEq + std::hash::Hash,
+{
+    fn default() -> Self {
+        Self {
+            entities: std::sync::Mutex::new(HashMap::new()),
+        }
+    }
+}

control_plane/attachment_service/src/lib.rs

@@ -2,23 +2,19 @@ use serde::Serialize;
 use utils::seqwait::MonotonicCounter;
 
 mod auth;
-mod background_node_operations;
 mod compute_hook;
-mod drain_utils;
 mod heartbeater;
 pub mod http;
 mod id_lock_map;
-mod leadership;
 pub mod metrics;
 mod node;
 mod pageserver_client;
-mod peer_client;
 pub mod persistence;
 mod reconciler;
 mod scheduler;
 mod schema;
 pub mod service;
-mod tenant_shard;
+mod tenant_state;
 
 #[derive(Ord, PartialOrd, Eq, PartialEq, Copy, Clone, Serialize)]
 struct Sequence(u64);

control_plane/attachment_service/src/main.rs

@@ -1,31 +1,26 @@
 use anyhow::{anyhow, Context};
+use attachment_service::http::make_router;
+use attachment_service::metrics::preinitialize_metrics;
+use attachment_service::persistence::Persistence;
+use attachment_service::service::{Config, Service, MAX_UNAVAILABLE_INTERVAL_DEFAULT};
+use camino::Utf8PathBuf;
 use clap::Parser;
-use hyper::Uri;
+use diesel::Connection;
 use metrics::launch_timestamp::LaunchTimestamp;
-use metrics::BuildInfo;
-use std::path::PathBuf;
 use std::sync::Arc;
-use std::time::Duration;
-use storage_controller::http::make_router;
-use storage_controller::metrics::preinitialize_metrics;
-use storage_controller::persistence::Persistence;
-use storage_controller::service::chaos_injector::ChaosInjector;
-use storage_controller::service::{
-    Config, Service, HEARTBEAT_INTERVAL_DEFAULT, MAX_OFFLINE_INTERVAL_DEFAULT,
-    MAX_WARMING_UP_INTERVAL_DEFAULT, RECONCILER_CONCURRENCY_DEFAULT,
-};
 use tokio::signal::unix::SignalKind;
 use tokio_util::sync::CancellationToken;
-use tracing::Instrument;
 use utils::auth::{JwtAuth, SwappableJwtAuth};
 use utils::logging::{self, LogFormat};
 
-use utils::sentry_init::init_sentry;
 use utils::{project_build_tag, project_git_version, tcp_listener};
 
 project_git_version!(GIT_VERSION);
 project_build_tag!(BUILD_TAG);
 
+use diesel_migrations::{embed_migrations, EmbeddedMigrations};
+pub const MIGRATIONS: EmbeddedMigrations = embed_migrations!("./migrations");
+
 #[derive(Parser)]
 #[command(author, version, about, long_about = None)]
 #[command(arg_required_else_help(true))]
@@ -47,14 +42,15 @@ struct Cli {
     #[arg(long)]
     control_plane_jwt_token: Option<String>,
 
-    #[arg(long)]
-    peer_jwt_token: Option<String>,
-
     /// URL to control plane compute notification endpoint
     #[arg(long)]
     compute_hook_url: Option<String>,
 
-    /// URL to connect to postgres, like postgresql://localhost:1234/storage_controller
+    /// Path to the .json file to store state (will be created if it doesn't exist)
+    #[arg(short, long)]
+    path: Option<Utf8PathBuf>,
+
+    /// URL to connect to postgres, like postgresql://localhost:1234/attachment_service
     #[arg(long)]
     database_url: Option<String>,
 
@@ -64,50 +60,7 @@ struct Cli {
 
     /// Grace period before marking unresponsive pageserver offline
     #[arg(long)]
-    max_offline_interval: Option<humantime::Duration>,
-
-    /// More tolerant grace period before marking unresponsive pagserver offline used
-    /// around pageserver restarts
-    #[arg(long)]
-    max_warming_up_interval: Option<humantime::Duration>,
-
-    /// Size threshold for automatically splitting shards (disabled by default)
-    #[arg(long)]
-    split_threshold: Option<u64>,
-
-    /// Maximum number of reconcilers that may run in parallel
-    #[arg(long)]
-    reconciler_concurrency: Option<usize>,
-
-    /// How long to wait for the initial database connection to be available.
-    #[arg(long, default_value = "5s")]
-    db_connect_timeout: humantime::Duration,
-
-    #[arg(long, default_value = "false")]
-    start_as_candidate: bool,
-
-    // TODO: make this mandatory once the helm chart gets updated
-    #[arg(long)]
-    address_for_peers: Option<Uri>,
-
-    /// `neon_local` sets this to the path of the neon_local repo dir.
-    /// Only relevant for testing.
-    // TODO: make `cfg(feature = "testing")`
-    #[arg(long)]
-    neon_local_repo_dir: Option<PathBuf>,
-
-    /// Chaos testing
-    #[arg(long)]
-    chaos_interval: Option<humantime::Duration>,
-
-    // Maximum acceptable lag for the secondary location while draining
-    // a pageserver
-    #[arg(long)]
-    max_secondary_lag_bytes: Option<u64>,
-
-    // Period with which to send heartbeats to registered nodes
-    #[arg(long)]
-    heartbeat_interval: Option<humantime::Duration>,
+    max_unavailable_interval: Option<humantime::Duration>,
 }
 
 enum StrictMode {
@@ -133,28 +86,28 @@ struct Secrets {
     public_key: Option<JwtAuth>,
     jwt_token: Option<String>,
     control_plane_jwt_token: Option<String>,
-    peer_jwt_token: Option<String>,
 }
 
 impl Secrets {
     const DATABASE_URL_ENV: &'static str = "DATABASE_URL";
     const PAGESERVER_JWT_TOKEN_ENV: &'static str = "PAGESERVER_JWT_TOKEN";
     const CONTROL_PLANE_JWT_TOKEN_ENV: &'static str = "CONTROL_PLANE_JWT_TOKEN";
-    const PEER_JWT_TOKEN_ENV: &'static str = "PEER_JWT_TOKEN";
     const PUBLIC_KEY_ENV: &'static str = "PUBLIC_KEY";
 
     /// Load secrets from, in order of preference:
     /// - CLI args if database URL is provided on the CLI
     /// - Environment variables if DATABASE_URL is set.
+    /// - AWS Secrets Manager secrets
     async fn load(args: &Cli) -> anyhow::Result<Self> {
-        let Some(database_url) = Self::load_secret(&args.database_url, Self::DATABASE_URL_ENV)
+        let Some(database_url) =
+            Self::load_secret(&args.database_url, Self::DATABASE_URL_ENV).await
         else {
             anyhow::bail!(
                 "Database URL is not set (set `--database-url`, or `DATABASE_URL` environment)"
             )
         };
 
-        let public_key = match Self::load_secret(&args.public_key, Self::PUBLIC_KEY_ENV) {
+        let public_key = match Self::load_secret(&args.public_key, Self::PUBLIC_KEY_ENV).await {
             Some(v) => Some(JwtAuth::from_key(v).context("Loading public key")?),
             None => None,
         };
@@ -162,18 +115,18 @@ impl Secrets {
         let this = Self {
             database_url,
             public_key,
-            jwt_token: Self::load_secret(&args.jwt_token, Self::PAGESERVER_JWT_TOKEN_ENV),
+            jwt_token: Self::load_secret(&args.jwt_token, Self::PAGESERVER_JWT_TOKEN_ENV).await,
             control_plane_jwt_token: Self::load_secret(
                 &args.control_plane_jwt_token,
                 Self::CONTROL_PLANE_JWT_TOKEN_ENV,
-            ),
-            peer_jwt_token: Self::load_secret(&args.peer_jwt_token, Self::PEER_JWT_TOKEN_ENV),
+            )
+            .await,
         };
 
         Ok(this)
     }
 
-    fn load_secret(cli: &Option<String>, env_name: &str) -> Option<String> {
+    async fn load_secret(cli: &Option<String>, env_name: &str) -> Option<String> {
         if let Some(v) = cli {
             Some(v.clone())
         } else if let Ok(v) = std::env::var(env_name) {
@@ -184,24 +137,24 @@ impl Secrets {
     }
 }
 
+/// Execute the diesel migrations that are built into this binary
+async fn migration_run(database_url: &str) -> anyhow::Result<()> {
+    use diesel::PgConnection;
+    use diesel_migrations::{HarnessWithOutput, MigrationHarness};
+    let mut conn = PgConnection::establish(database_url)?;
+
+    HarnessWithOutput::write_to_stdout(&mut conn)
+        .run_pending_migrations(MIGRATIONS)
+        .map(|_| ())
+        .map_err(|e| anyhow::anyhow!(e))?;
+
+    Ok(())
+}
+
 fn main() -> anyhow::Result<()> {
-    logging::init(
-        LogFormat::Plain,
-        logging::TracingErrorLayerEnablement::Disabled,
-        logging::Output::Stdout,
-    )?;
-
-    // log using tracing so we don't get confused output by default hook writing to stderr
-    utils::logging::replace_panic_hook_with_tracing_panic_hook().forget();
-
-    let _sentry_guard = init_sentry(Some(GIT_VERSION.into()), &[]);
-
-    let hook = std::panic::take_hook();
+    let default_panic = std::panic::take_hook();
     std::panic::set_hook(Box::new(move |info| {
-        // let sentry send a message (and flush)
-        // and trace the error
-        hook(info);
-
+        default_panic(info);
         std::process::exit(1);
     }));
 
@@ -218,22 +171,24 @@ fn main() -> anyhow::Result<()> {
 async fn async_main() -> anyhow::Result<()> {
     let launch_ts = Box::leak(Box::new(LaunchTimestamp::generate()));
 
+    logging::init(
+        LogFormat::Plain,
+        logging::TracingErrorLayerEnablement::Disabled,
+        logging::Output::Stdout,
+    )?;
+
     preinitialize_metrics();
 
     let args = Cli::parse();
     tracing::info!(
-        "version: {}, launch_timestamp: {}, build_tag {}, listening on {}",
+        "version: {}, launch_timestamp: {}, build_tag {}, state at {}, listening on {}",
         GIT_VERSION,
         launch_ts.to_string(),
         BUILD_TAG,
+        args.path.as_ref().unwrap_or(&Utf8PathBuf::from("<none>")),
         args.listen
     );
 
-    let build_info = BuildInfo {
-        revision: GIT_VERSION,
-        build_tag: BUILD_TAG,
-    };
-
     let strict_mode = if args.dev {
         StrictMode::Dev
     } else {
@@ -273,35 +228,20 @@ async fn async_main() -> anyhow::Result<()> {
     let config = Config {
         jwt_token: secrets.jwt_token,
         control_plane_jwt_token: secrets.control_plane_jwt_token,
-        peer_jwt_token: secrets.peer_jwt_token,
         compute_hook_url: args.compute_hook_url,
-        max_offline_interval: args
-            .max_offline_interval
+        max_unavailable_interval: args
+            .max_unavailable_interval
             .map(humantime::Duration::into)
-            .unwrap_or(MAX_OFFLINE_INTERVAL_DEFAULT),
-        max_warming_up_interval: args
-            .max_warming_up_interval
-            .map(humantime::Duration::into)
-            .unwrap_or(MAX_WARMING_UP_INTERVAL_DEFAULT),
-        reconciler_concurrency: args
-            .reconciler_concurrency
-            .unwrap_or(RECONCILER_CONCURRENCY_DEFAULT),
-        split_threshold: args.split_threshold,
-        neon_local_repo_dir: args.neon_local_repo_dir,
-        max_secondary_lag_bytes: args.max_secondary_lag_bytes,
-        heartbeat_interval: args
-            .heartbeat_interval
-            .map(humantime::Duration::into)
-            .unwrap_or(HEARTBEAT_INTERVAL_DEFAULT),
-        address_for_peers: args.address_for_peers,
-        start_as_candidate: args.start_as_candidate,
-        http_service_port: args.listen.port() as i32,
+            .unwrap_or(MAX_UNAVAILABLE_INTERVAL_DEFAULT),
     };
 
-    // Validate that we can connect to the database
-    Persistence::await_connection(&secrets.database_url, args.db_connect_timeout.into()).await?;
+    // After loading secrets & config, but before starting anything else, apply database migrations
+    migration_run(&secrets.database_url)
+        .await
+        .context("Running database migrations")?;
 
-    let persistence = Arc::new(Persistence::new(secrets.database_url));
+    let json_path = args.path;
+    let persistence = Arc::new(Persistence::new(secrets.database_url, json_path.clone()));
 
     let service = Service::spawn(config, persistence.clone()).await?;
 
@@ -310,7 +250,7 @@ async fn async_main() -> anyhow::Result<()> {
     let auth = secrets
         .public_key
         .map(|jwt_auth| Arc::new(SwappableJwtAuth::new(jwt_auth)));
-    let router = make_router(service.clone(), auth, build_info)
+    let router = make_router(service.clone(), auth)
         .build()
         .map_err(|err| anyhow!(err))?;
     let router_service = utils::http::RouterService::new(router).unwrap();
@@ -328,22 +268,6 @@ async fn async_main() -> anyhow::Result<()> {
     tracing::info!("Serving on {0}", args.listen);
     let server_task = tokio::task::spawn(server);
 
-    let chaos_task = args.chaos_interval.map(|interval| {
-        let service = service.clone();
-        let cancel = CancellationToken::new();
-        let cancel_bg = cancel.clone();
-        (
-            tokio::task::spawn(
-                async move {
-                    let mut chaos_injector = ChaosInjector::new(service, interval.into());
-                    chaos_injector.run(cancel_bg).await
-                }
-                .instrument(tracing::info_span!("chaos_injector")),
-            ),
-            cancel,
-        )
-    });
-
     // Wait until we receive a signal
     let mut sigint = tokio::signal::unix::signal(SignalKind::interrupt())?;
     let mut sigquit = tokio::signal::unix::signal(SignalKind::quit())?;
@@ -355,28 +279,21 @@ async fn async_main() -> anyhow::Result<()> {
     }
     tracing::info!("Terminating on signal");
 
-    // Stop HTTP server first, so that we don't have to service requests
-    // while shutting down Service.
-    server_shutdown.cancel();
-    match tokio::time::timeout(Duration::from_secs(5), server_task).await {
-        Ok(Ok(_)) => {
-            tracing::info!("Joined HTTP server task");
-        }
-        Ok(Err(e)) => {
-            tracing::error!("Error joining HTTP server task: {e}")
-        }
-        Err(_) => {
-            tracing::warn!("Timed out joining HTTP server task");
-            // We will fall through and shut down the service anyway, any request handlers
-            // in flight will experience cancellation & their clients will see a torn connection.
+    if json_path.is_some() {
+        // Write out a JSON dump on shutdown: this is used in compat tests to avoid passing
+        // full postgres dumps around.
+        if let Err(e) = persistence.write_tenants_json().await {
+            tracing::error!("Failed to write JSON on shutdown: {e}")
         }
     }
 
-    // If we were injecting chaos, stop that so that we're not calling into Service while it shuts down
-    if let Some((chaos_jh, chaos_cancel)) = chaos_task {
-        chaos_cancel.cancel();
-        chaos_jh.await.ok();
+    // Stop HTTP server first, so that we don't have to service requests
+    // while shutting down Service
+    server_shutdown.cancel();
+    if let Err(e) = server_task.await {
+        tracing::error!("Error joining HTTP server task: {e}")
     }
+    tracing::info!("Joined HTTP server task");
 
     service.shutdown().await;
     tracing::info!("Service shutdown complete");

control_plane/attachment_service/src/metrics.rs

@@ -8,16 +8,14 @@
 //! The rest of the code defines label group types and deals with converting outer types to labels.
 //!
 use bytes::Bytes;
-use measured::{label::LabelValue, metric::histogram, FixedCardinalityLabel, MetricGroup};
-use metrics::NeonMetrics;
+use measured::{
+    label::{LabelValue, StaticLabelSet},
+    FixedCardinalityLabel, MetricGroup,
+};
 use once_cell::sync::Lazy;
 use std::sync::Mutex;
-use strum::IntoEnumIterator;
 
-use crate::{
-    persistence::{DatabaseError, DatabaseOperation},
-    service::LeadershipStatus,
-};
+use crate::persistence::{DatabaseError, DatabaseOperation};
 
 pub(crate) static METRICS_REGISTRY: Lazy<StorageControllerMetrics> =
     Lazy::new(StorageControllerMetrics::default);
@@ -28,28 +26,21 @@ pub fn preinitialize_metrics() {
 
 pub(crate) struct StorageControllerMetrics {
     pub(crate) metrics_group: StorageControllerMetricGroup,
-    encoder: Mutex<measured::text::BufferedTextEncoder>,
+    encoder: Mutex<measured::text::TextEncoder>,
 }
 
 #[derive(measured::MetricGroup)]
-#[metric(new())]
 pub(crate) struct StorageControllerMetricGroup {
     /// Count of how many times we spawn a reconcile task
     pub(crate) storage_controller_reconcile_spawn: measured::Counter,
-
     /// Reconciler tasks completed, broken down by success/failure/cancelled
     pub(crate) storage_controller_reconcile_complete:
         measured::CounterVec<ReconcileCompleteLabelGroupSet>,
 
-    /// Count of how many times we make an optimization change to a tenant's scheduling
-    pub(crate) storage_controller_schedule_optimization: measured::Counter,
-
     /// HTTP request status counters for handled requests
     pub(crate) storage_controller_http_request_status:
         measured::CounterVec<HttpRequestStatusLabelGroupSet>,
-
     /// HTTP request handler latency across all status codes
-    #[metric(metadata = histogram::Thresholds::exponential_buckets(0.1, 2.0))]
     pub(crate) storage_controller_http_request_latency:
         measured::HistogramVec<HttpRequestLatencyLabelGroupSet, 5>,
 
@@ -61,7 +52,6 @@ pub(crate) struct StorageControllerMetricGroup {
     /// Latency of HTTP requests to the pageserver, broken down by pageserver
     /// node id, request name and method. This include both successful and unsuccessful
     /// requests.
-    #[metric(metadata = histogram::Thresholds::exponential_buckets(0.1, 2.0))]
     pub(crate) storage_controller_pageserver_request_latency:
         measured::HistogramVec<PageserverRequestLabelGroupSet, 5>,
 
@@ -73,7 +63,6 @@ pub(crate) struct StorageControllerMetricGroup {
     /// Latency of pass-through HTTP requests to the pageserver, broken down by pageserver
     /// node id, request name and method. This include both successful and unsuccessful
     /// requests.
-    #[metric(metadata = histogram::Thresholds::exponential_buckets(0.1, 2.0))]
     pub(crate) storage_controller_passthrough_request_latency:
         measured::HistogramVec<PageserverRequestLabelGroupSet, 5>,
 
@@ -82,36 +71,75 @@ pub(crate) struct StorageControllerMetricGroup {
         measured::CounterVec<DatabaseQueryErrorLabelGroupSet>,
 
     /// Latency of database queries, broken down by operation.
-    #[metric(metadata = histogram::Thresholds::exponential_buckets(0.1, 2.0))]
     pub(crate) storage_controller_database_query_latency:
         measured::HistogramVec<DatabaseQueryLatencyLabelGroupSet, 5>,
-
-    pub(crate) storage_controller_leadership_status: measured::GaugeVec<LeadershipStatusGroupSet>,
 }
 
 impl StorageControllerMetrics {
-    pub(crate) fn encode(&self, neon_metrics: &NeonMetrics) -> Bytes {
+    pub(crate) fn encode(&self) -> Bytes {
         let mut encoder = self.encoder.lock().unwrap();
-        neon_metrics
-            .collect_group_into(&mut *encoder)
-            .unwrap_or_else(|infallible| match infallible {});
-        self.metrics_group
-            .collect_group_into(&mut *encoder)
-            .unwrap_or_else(|infallible| match infallible {});
+        self.metrics_group.collect_into(&mut *encoder);
         encoder.finish()
     }
 }
 
 impl Default for StorageControllerMetrics {
     fn default() -> Self {
-        let mut metrics_group = StorageControllerMetricGroup::new();
-        metrics_group
-            .storage_controller_reconcile_complete
-            .init_all_dense();
-
         Self {
-            metrics_group,
-            encoder: Mutex::new(measured::text::BufferedTextEncoder::new()),
+            metrics_group: StorageControllerMetricGroup::new(),
+            encoder: Mutex::new(measured::text::TextEncoder::new()),
+        }
+    }
+}
+
+impl StorageControllerMetricGroup {
+    pub(crate) fn new() -> Self {
+        Self {
+            storage_controller_reconcile_spawn: measured::Counter::new(),
+            storage_controller_reconcile_complete: measured::CounterVec::new(
+                ReconcileCompleteLabelGroupSet {
+                    status: StaticLabelSet::new(),
+                },
+            ),
+            storage_controller_http_request_status: measured::CounterVec::new(
+                HttpRequestStatusLabelGroupSet {
+                    path: lasso::ThreadedRodeo::new(),
+                    method: StaticLabelSet::new(),
+                    status: StaticLabelSet::new(),
+                },
+            ),
+            storage_controller_http_request_latency: measured::HistogramVec::new(
+                measured::metric::histogram::Thresholds::exponential_buckets(0.1, 2.0),
+            ),
+            storage_controller_pageserver_request_error: measured::CounterVec::new(
+                PageserverRequestLabelGroupSet {
+                    pageserver_id: lasso::ThreadedRodeo::new(),
+                    path: lasso::ThreadedRodeo::new(),
+                    method: StaticLabelSet::new(),
+                },
+            ),
+            storage_controller_pageserver_request_latency: measured::HistogramVec::new(
+                measured::metric::histogram::Thresholds::exponential_buckets(0.1, 2.0),
+            ),
+            storage_controller_passthrough_request_error: measured::CounterVec::new(
+                PageserverRequestLabelGroupSet {
+                    pageserver_id: lasso::ThreadedRodeo::new(),
+                    path: lasso::ThreadedRodeo::new(),
+                    method: StaticLabelSet::new(),
+                },
+            ),
+            storage_controller_passthrough_request_latency: measured::HistogramVec::new(
+                measured::metric::histogram::Thresholds::exponential_buckets(0.1, 2.0),
+            ),
+            storage_controller_database_query_error: measured::CounterVec::new(
+                DatabaseQueryErrorLabelGroupSet {
+                    operation: StaticLabelSet::new(),
+                    error_type: StaticLabelSet::new(),
+                },
+            ),
+            storage_controller_database_query_latency: measured::HistogramVec::new(
+                measured::metric::histogram::Thresholds::exponential_buckets(0.1, 2.0),
+            ),
         }
     }
 }
@@ -125,7 +153,7 @@ pub(crate) struct ReconcileCompleteLabelGroup {
 #[derive(measured::LabelGroup)]
 #[label(set = HttpRequestStatusLabelGroupSet)]
 pub(crate) struct HttpRequestStatusLabelGroup<'a> {
-    #[label(dynamic_with = lasso::ThreadedRodeo, default)]
+    #[label(dynamic_with = lasso::ThreadedRodeo)]
     pub(crate) path: &'a str,
     pub(crate) method: Method,
     pub(crate) status: StatusCode,
@@ -134,21 +162,40 @@ pub(crate) struct HttpRequestStatusLabelGroup<'a> {
 #[derive(measured::LabelGroup)]
 #[label(set = HttpRequestLatencyLabelGroupSet)]
 pub(crate) struct HttpRequestLatencyLabelGroup<'a> {
-    #[label(dynamic_with = lasso::ThreadedRodeo, default)]
+    #[label(dynamic_with = lasso::ThreadedRodeo)]
     pub(crate) path: &'a str,
     pub(crate) method: Method,
 }
 
+impl Default for HttpRequestLatencyLabelGroupSet {
+    fn default() -> Self {
+        Self {
+            path: lasso::ThreadedRodeo::new(),
+            method: StaticLabelSet::new(),
+        }
+    }
+}
+
 #[derive(measured::LabelGroup, Clone)]
 #[label(set = PageserverRequestLabelGroupSet)]
 pub(crate) struct PageserverRequestLabelGroup<'a> {
-    #[label(dynamic_with = lasso::ThreadedRodeo, default)]
+    #[label(dynamic_with = lasso::ThreadedRodeo)]
     pub(crate) pageserver_id: &'a str,
-    #[label(dynamic_with = lasso::ThreadedRodeo, default)]
+    #[label(dynamic_with = lasso::ThreadedRodeo)]
     pub(crate) path: &'a str,
     pub(crate) method: Method,
 }
 
+impl Default for PageserverRequestLabelGroupSet {
+    fn default() -> Self {
+        Self {
+            pageserver_id: lasso::ThreadedRodeo::new(),
+            path: lasso::ThreadedRodeo::new(),
+            method: StaticLabelSet::new(),
+        }
+    }
+}
+
 #[derive(measured::LabelGroup)]
 #[label(set = DatabaseQueryErrorLabelGroupSet)]
 pub(crate) struct DatabaseQueryErrorLabelGroup {
@@ -162,13 +209,7 @@ pub(crate) struct DatabaseQueryLatencyLabelGroup {
     pub(crate) operation: DatabaseOperation,
 }
 
-#[derive(measured::LabelGroup)]
-#[label(set = LeadershipStatusGroupSet)]
-pub(crate) struct LeadershipStatusGroup {
-    pub(crate) status: LeadershipStatus,
-}
-
-#[derive(FixedCardinalityLabel, Clone, Copy)]
+#[derive(FixedCardinalityLabel)]
 pub(crate) enum ReconcileOutcome {
     #[label(rename = "ok")]
     Success,
@@ -176,7 +217,7 @@ pub(crate) enum ReconcileOutcome {
     Cancel,
 }
 
-#[derive(FixedCardinalityLabel, Copy, Clone)]
+#[derive(FixedCardinalityLabel, Clone)]
 pub(crate) enum Method {
     Get,
     Put,
@@ -201,12 +242,11 @@ impl From<hyper::Method> for Method {
     }
 }
 
-#[derive(Clone, Copy)]
 pub(crate) struct StatusCode(pub(crate) hyper::http::StatusCode);
 
 impl LabelValue for StatusCode {
     fn visit<V: measured::label::LabelVisitor>(&self, v: V) -> V::Output {
-        v.write_int(self.0.as_u16() as i64)
+        v.write_int(self.0.as_u16() as u64)
     }
 }
 
@@ -224,13 +264,12 @@ impl FixedCardinalityLabel for StatusCode {
     }
 }
 
-#[derive(FixedCardinalityLabel, Clone, Copy)]
+#[derive(FixedCardinalityLabel)]
 pub(crate) enum DatabaseErrorLabel {
     Query,
     Connection,
     ConnectionPool,
     Logical,
-    Migration,
 }
 
 impl DatabaseError {
@@ -240,22 +279,6 @@ impl DatabaseError {
             Self::Connection(_) => DatabaseErrorLabel::Connection,
             Self::ConnectionPool(_) => DatabaseErrorLabel::ConnectionPool,
             Self::Logical(_) => DatabaseErrorLabel::Logical,
-            Self::Migration(_) => DatabaseErrorLabel::Migration,
-        }
-    }
-}
-
-/// Update the leadership status metric gauges to reflect the requested status
-pub(crate) fn update_leadership_status(status: LeadershipStatus) {
-    let status_metric = &METRICS_REGISTRY
-        .metrics_group
-        .storage_controller_leadership_status;
-
-    for s in LeadershipStatus::iter() {
-        if s == status {
-            status_metric.set(LeadershipStatusGroup { status: s }, 1);
-        } else {
-            status_metric.set(LeadershipStatusGroup { status: s }, 0);
         }
     }
 }

control_plane/attachment_service/src/node.rs

@@ -1,14 +1,13 @@
 use std::{str::FromStr, time::Duration};
 
+use hyper::StatusCode;
 use pageserver_api::{
     controller_api::{
-        NodeAvailability, NodeDescribeResponse, NodeRegisterRequest, NodeSchedulingPolicy,
-        TenantLocateResponseShard,
+        NodeAvailability, NodeRegisterRequest, NodeSchedulingPolicy, TenantLocateResponseShard,
     },
     shard::TenantShardId,
 };
 use pageserver_client::mgmt_api;
-use reqwest::StatusCode;
 use serde::Serialize;
 use tokio_util::sync::CancellationToken;
 use utils::{backoff, id::NodeId};
@@ -36,8 +35,6 @@ pub(crate) struct Node {
     listen_pg_addr: String,
     listen_pg_port: u16,
 
-    availability_zone_id: String,
-
     // This cancellation token means "stop any RPCs in flight to this node, and don't start
     // any more". It is not related to process shutdown.
     #[serde(skip)]
@@ -48,8 +45,6 @@ pub(crate) struct Node {
 /// whether/how they changed it.
 pub(crate) enum AvailabilityTransition {
     ToActive,
-    ToWarmingUpFromActive,
-    ToWarmingUpFromOffline,
     ToOffline,
     Unchanged,
 }
@@ -63,15 +58,6 @@ impl Node {
         self.id
     }
 
-    #[allow(unused)]
-    pub(crate) fn get_availability_zone_id(&self) -> &str {
-        self.availability_zone_id.as_str()
-    }
-
-    pub(crate) fn get_scheduling(&self) -> NodeSchedulingPolicy {
-        self.scheduling
-    }
-
     pub(crate) fn set_scheduling(&mut self, scheduling: NodeSchedulingPolicy) {
         self.scheduling = scheduling
     }
@@ -84,7 +70,6 @@ impl Node {
             && self.listen_http_port == register_req.listen_http_port
             && self.listen_pg_addr == register_req.listen_pg_addr
             && self.listen_pg_port == register_req.listen_pg_port
-            && self.availability_zone_id == register_req.availability_zone_id
     }
 
     /// For a shard located on this node, populate a response object
@@ -100,52 +85,36 @@ impl Node {
         }
     }
 
-    pub(crate) fn get_availability(&self) -> &NodeAvailability {
-        &self.availability
-    }
-
     pub(crate) fn set_availability(&mut self, availability: NodeAvailability) {
-        use AvailabilityTransition::*;
-        use NodeAvailability::WarmingUp;
-
-        match self.get_availability_transition(&availability) {
-            ToActive => {
+        match self.get_availability_transition(availability) {
+            AvailabilityTransition::ToActive => {
                 // Give the node a new cancellation token, effectively resetting it to un-cancelled.  Any
                 // users of previously-cloned copies of the node will still see the old cancellation
                 // state.  For example, Reconcilers in flight will have to complete and be spawned
                 // again to realize that the node has become available.
                 self.cancel = CancellationToken::new();
             }
-            ToOffline | ToWarmingUpFromActive => {
+            AvailabilityTransition::ToOffline => {
                 // Fire the node's cancellation token to cancel any in-flight API requests to it
                 self.cancel.cancel();
             }
-            Unchanged | ToWarmingUpFromOffline => {}
-        }
-
-        if let (WarmingUp(crnt), WarmingUp(proposed)) = (&self.availability, &availability) {
-            self.availability = WarmingUp(std::cmp::max(*crnt, *proposed));
-        } else {
-            self.availability = availability;
+            AvailabilityTransition::Unchanged => {}
         }
+        self.availability = availability;
     }
 
     /// Without modifying the availability of the node, convert the intended availability
     /// into a description of the transition.
     pub(crate) fn get_availability_transition(
         &self,
-        availability: &NodeAvailability,
+        availability: NodeAvailability,
     ) -> AvailabilityTransition {
         use AvailabilityTransition::*;
         use NodeAvailability::*;
 
-        match (&self.availability, availability) {
+        match (self.availability, availability) {
             (Offline, Active(_)) => ToActive,
             (Active(_), Offline) => ToOffline,
-            (Active(_), WarmingUp(_)) => ToWarmingUpFromActive,
-            (WarmingUp(_), Offline) => ToOffline,
-            (WarmingUp(_), Active(_)) => ToActive,
-            (Offline, WarmingUp(_)) => ToWarmingUpFromOffline,
             _ => Unchanged,
         }
     }
@@ -161,17 +130,16 @@ impl Node {
 
     /// Is this node elegible to have work scheduled onto it?
     pub(crate) fn may_schedule(&self) -> MaySchedule {
-        let utilization = match &self.availability {
-            NodeAvailability::Active(u) => u.clone(),
-            NodeAvailability::Offline | NodeAvailability::WarmingUp(_) => return MaySchedule::No,
+        let score = match self.availability {
+            NodeAvailability::Active(score) => score,
+            NodeAvailability::Offline => return MaySchedule::No,
         };
 
         match self.scheduling {
-            NodeSchedulingPolicy::Active => MaySchedule::Yes(utilization),
+            NodeSchedulingPolicy::Active => MaySchedule::Yes(score),
             NodeSchedulingPolicy::Draining => MaySchedule::No,
-            NodeSchedulingPolicy::Filling => MaySchedule::Yes(utilization),
+            NodeSchedulingPolicy::Filling => MaySchedule::Yes(score),
             NodeSchedulingPolicy::Pause => MaySchedule::No,
-            NodeSchedulingPolicy::PauseForRestart => MaySchedule::No,
         }
     }
 
@@ -181,7 +149,6 @@ impl Node {
         listen_http_port: u16,
         listen_pg_addr: String,
         listen_pg_port: u16,
-        availability_zone_id: String,
     ) -> Self {
         Self {
             id,
@@ -189,9 +156,8 @@ impl Node {
             listen_http_port,
             listen_pg_addr,
             listen_pg_port,
-            scheduling: NodeSchedulingPolicy::Active,
+            scheduling: NodeSchedulingPolicy::Filling,
             availability: NodeAvailability::Offline,
-            availability_zone_id,
             cancel: CancellationToken::new(),
         }
     }
@@ -204,7 +170,6 @@ impl Node {
             listen_http_port: self.listen_http_port as i32,
             listen_pg_addr: self.listen_pg_addr.clone(),
             listen_pg_port: self.listen_pg_port as i32,
-            availability_zone_id: self.availability_zone_id.clone(),
         }
     }
 
@@ -219,7 +184,6 @@ impl Node {
             listen_http_port: np.listen_http_port as u16,
             listen_pg_addr: np.listen_pg_addr,
             listen_pg_port: np.listen_pg_port as u16,
-            availability_zone_id: np.availability_zone_id,
             cancel: CancellationToken::new(),
         }
     }
@@ -246,7 +210,7 @@ impl Node {
         fn is_fatal(e: &mgmt_api::Error) -> bool {
             use mgmt_api::Error::*;
             match e {
-                SendRequest(_) | ReceiveBody(_) | ReceiveErrorBody(_) => false,
+                ReceiveBody(_) | ReceiveErrorBody(_) => false,
                 ApiError(StatusCode::SERVICE_UNAVAILABLE, _)
                 | ApiError(StatusCode::GATEWAY_TIMEOUT, _)
                 | ApiError(StatusCode::REQUEST_TIMEOUT, _) => false,
@@ -292,19 +256,6 @@ impl Node {
         )
         .await
     }
-
-    /// Generate the simplified API-friendly description of a node's state
-    pub(crate) fn describe(&self) -> NodeDescribeResponse {
-        NodeDescribeResponse {
-            id: self.id,
-            availability: self.availability.clone().into(),
-            scheduling: self.scheduling,
-            listen_http_addr: self.listen_http_addr.clone(),
-            listen_http_port: self.listen_http_port,
-            listen_pg_addr: self.listen_pg_addr.clone(),
-            listen_pg_port: self.listen_pg_port,
-        }
-    }
 }
 
 impl std::fmt::Display for Node {

control_plane/attachment_service/src/pageserver_client.rs

@@ -1,18 +1,13 @@
 use pageserver_api::{
     models::{
-        detach_ancestor::AncestorDetached, LocationConfig, LocationConfigListResponse,
-        PageserverUtilization, SecondaryProgress, TenantScanRemoteStorageResponse,
-        TenantShardSplitRequest, TenantShardSplitResponse, TimelineArchivalConfigRequest,
-        TimelineCreateRequest, TimelineInfo, TopTenantShardsRequest, TopTenantShardsResponse,
+        LocationConfig, LocationConfigListResponse, PageserverUtilization, SecondaryProgress,
+        TenantShardSplitRequest, TenantShardSplitResponse, TimelineCreateRequest, TimelineInfo,
     },
     shard::TenantShardId,
 };
-use pageserver_client::{
-    mgmt_api::{Client, Result},
-    BlockUnblock,
-};
+use pageserver_client::mgmt_api::{Client, Result};
 use reqwest::StatusCode;
-use utils::id::{NodeId, TenantId, TimelineId};
+use utils::id::{NodeId, TimelineId};
 
 /// Thin wrapper around [`pageserver_client::mgmt_api::Client`]. It allows the storage
 /// controller to collect metrics in a non-intrusive manner.
@@ -93,18 +88,6 @@ impl PageserverClient {
         )
     }
 
-    pub(crate) async fn tenant_scan_remote_storage(
-        &self,
-        tenant_id: TenantId,
-    ) -> Result<TenantScanRemoteStorageResponse> {
-        measured_request!(
-            "tenant_scan_remote_storage",
-            crate::metrics::Method::Get,
-            &self.node_id_label,
-            self.inner.tenant_scan_remote_storage(tenant_id).await
-        )
-    }
-
     pub(crate) async fn tenant_secondary_download(
         &self,
         tenant_id: TenantShardId,
@@ -118,27 +101,6 @@ impl PageserverClient {
         )
     }
 
-    pub(crate) async fn tenant_secondary_status(
-        &self,
-        tenant_shard_id: TenantShardId,
-    ) -> Result<SecondaryProgress> {
-        measured_request!(
-            "tenant_secondary_status",
-            crate::metrics::Method::Get,
-            &self.node_id_label,
-            self.inner.tenant_secondary_status(tenant_shard_id).await
-        )
-    }
-
-    pub(crate) async fn tenant_heatmap_upload(&self, tenant_id: TenantShardId) -> Result<()> {
-        measured_request!(
-            "tenant_heatmap_upload",
-            crate::metrics::Method::Post,
-            &self.node_id_label,
-            self.inner.tenant_heatmap_upload(tenant_id).await
-        )
-    }
-
     pub(crate) async fn location_config(
         &self,
         tenant_shard_id: TenantShardId,
@@ -230,55 +192,6 @@ impl PageserverClient {
         )
     }
 
-    pub(crate) async fn timeline_archival_config(
-        &self,
-        tenant_shard_id: TenantShardId,
-        timeline_id: TimelineId,
-        req: &TimelineArchivalConfigRequest,
-    ) -> Result<()> {
-        measured_request!(
-            "timeline_archival_config",
-            crate::metrics::Method::Post,
-            &self.node_id_label,
-            self.inner
-                .timeline_archival_config(tenant_shard_id, timeline_id, req)
-                .await
-        )
-    }
-
-    pub(crate) async fn timeline_detach_ancestor(
-        &self,
-        tenant_shard_id: TenantShardId,
-        timeline_id: TimelineId,
-    ) -> Result<AncestorDetached> {
-        measured_request!(
-            "timeline_detach_ancestor",
-            crate::metrics::Method::Put,
-            &self.node_id_label,
-            self.inner
-                .timeline_detach_ancestor(tenant_shard_id, timeline_id)
-                .await
-        )
-    }
-
-    pub(crate) async fn timeline_block_unblock_gc(
-        &self,
-        tenant_shard_id: TenantShardId,
-        timeline_id: TimelineId,
-        dir: BlockUnblock,
-    ) -> Result<()> {
-        // measuring these makes no sense because we synchronize with the gc loop and remote
-        // storage on block_gc so there should be huge outliers
-        measured_request!(
-            "timeline_block_unblock_gc",
-            crate::metrics::Method::Post,
-            &self.node_id_label,
-            self.inner
-                .timeline_block_unblock_gc(tenant_shard_id, timeline_id, dir)
-                .await
-        )
-    }
-
     pub(crate) async fn get_utilization(&self) -> Result<PageserverUtilization> {
         measured_request!(
             "utilization",
@@ -287,16 +200,4 @@ impl PageserverClient {
             self.inner.get_utilization().await
         )
     }
-
-    pub(crate) async fn top_tenant_shards(
-        &self,
-        request: TopTenantShardsRequest,
-    ) -> Result<TopTenantShardsResponse> {
-        measured_request!(
-            "top_tenants",
-            crate::metrics::Method::Post,
-            &self.node_id_label,
-            self.inner.top_tenant_shards(request).await
-        )
-    }
 }

control_plane/attachment_service/src/persistence.rs

@@ -0,0 +1,765 @@
+pub(crate) mod split_state;
+use std::collections::HashMap;
+use std::str::FromStr;
+use std::time::Duration;
+
+use self::split_state::SplitState;
+use camino::Utf8Path;
+use camino::Utf8PathBuf;
+use diesel::pg::PgConnection;
+use diesel::prelude::*;
+use diesel::Connection;
+use pageserver_api::controller_api::{NodeSchedulingPolicy, PlacementPolicy};
+use pageserver_api::models::TenantConfig;
+use pageserver_api::shard::ShardConfigError;
+use pageserver_api::shard::ShardIdentity;
+use pageserver_api::shard::ShardStripeSize;
+use pageserver_api::shard::{ShardCount, ShardNumber, TenantShardId};
+use serde::{Deserialize, Serialize};
+use utils::generation::Generation;
+use utils::id::{NodeId, TenantId};
+
+use crate::metrics::{
+    DatabaseQueryErrorLabelGroup, DatabaseQueryLatencyLabelGroup, METRICS_REGISTRY,
+};
+use crate::node::Node;
+
+/// ## What do we store?
+///
+/// The storage controller service does not store most of its state durably.
+///
+/// The essential things to store durably are:
+/// - generation numbers, as these must always advance monotonically to ensure data safety.
+/// - Tenant's PlacementPolicy and TenantConfig, as the source of truth for these is something external.
+/// - Node's scheduling policies, as the source of truth for these is something external.
+///
+/// Other things we store durably as an implementation detail:
+/// - Node's host/port: this could be avoided it we made nodes emit a self-registering heartbeat,
+///   but it is operationally simpler to make this service the authority for which nodes
+///   it talks to.
+///
+/// ## Performance/efficiency
+///
+/// The storage controller service does not go via the database for most things: there are
+/// a couple of places where we must, and where efficiency matters:
+/// - Incrementing generation numbers: the Reconciler has to wait for this to complete
+///   before it can attach a tenant, so this acts as a bound on how fast things like
+///   failover can happen.
+/// - Pageserver re-attach: we will increment many shards' generations when this happens,
+///   so it is important to avoid e.g. issuing O(N) queries.
+///
+/// Database calls relating to nodes have low performance requirements, as they are very rarely
+/// updated, and reads of nodes are always from memory, not the database.  We only require that
+/// we can UPDATE a node's scheduling mode reasonably quickly to mark a bad node offline.
+pub struct Persistence {
+    connection_pool: diesel::r2d2::Pool<diesel::r2d2::ConnectionManager<PgConnection>>,
+
+    // In test environments, we support loading+saving a JSON file.  This is temporary, for the benefit of
+    // test_compatibility.py, so that we don't have to commit to making the database contents fully backward/forward
+    // compatible just yet.
+    json_path: Option<Utf8PathBuf>,
+}
+
+/// Legacy format, for use in JSON compat objects in test environment
+#[derive(Serialize, Deserialize)]
+struct JsonPersistence {
+    tenants: HashMap<TenantShardId, TenantShardPersistence>,
+}
+
+#[derive(thiserror::Error, Debug)]
+pub(crate) enum DatabaseError {
+    #[error(transparent)]
+    Query(#[from] diesel::result::Error),
+    #[error(transparent)]
+    Connection(#[from] diesel::result::ConnectionError),
+    #[error(transparent)]
+    ConnectionPool(#[from] r2d2::Error),
+    #[error("Logical error: {0}")]
+    Logical(String),
+}
+
+#[derive(measured::FixedCardinalityLabel, Clone)]
+pub(crate) enum DatabaseOperation {
+    InsertNode,
+    UpdateNode,
+    DeleteNode,
+    ListNodes,
+    BeginShardSplit,
+    CompleteShardSplit,
+    AbortShardSplit,
+    Detach,
+    ReAttach,
+    IncrementGeneration,
+    ListTenantShards,
+    InsertTenantShards,
+    UpdateTenantShard,
+    DeleteTenant,
+    UpdateTenantConfig,
+}
+
+#[must_use]
+pub(crate) enum AbortShardSplitStatus {
+    /// We aborted the split in the database by reverting to the parent shards
+    Aborted,
+    /// The split had already been persisted.
+    Complete,
+}
+
+pub(crate) type DatabaseResult<T> = Result<T, DatabaseError>;
+
+impl Persistence {
+    // The default postgres connection limit is 100.  We use up to 99, to leave one free for a human admin under
+    // normal circumstances.  This assumes we have exclusive use of the database cluster to which we connect.
+    pub const MAX_CONNECTIONS: u32 = 99;
+
+    // We don't want to keep a lot of connections alive: close them down promptly if they aren't being used.
+    const IDLE_CONNECTION_TIMEOUT: Duration = Duration::from_secs(10);
+    const MAX_CONNECTION_LIFETIME: Duration = Duration::from_secs(60);
+
+    pub fn new(database_url: String, json_path: Option<Utf8PathBuf>) -> Self {
+        let manager = diesel::r2d2::ConnectionManager::<PgConnection>::new(database_url);
+
+        // We will use a connection pool: this is primarily to _limit_ our connection count, rather than to optimize time
+        // to execute queries (database queries are not generally on latency-sensitive paths).
+        let connection_pool = diesel::r2d2::Pool::builder()
+            .max_size(Self::MAX_CONNECTIONS)
+            .max_lifetime(Some(Self::MAX_CONNECTION_LIFETIME))
+            .idle_timeout(Some(Self::IDLE_CONNECTION_TIMEOUT))
+            // Always keep at least one connection ready to go
+            .min_idle(Some(1))
+            .test_on_check_out(true)
+            .build(manager)
+            .expect("Could not build connection pool");
+
+        Self {
+            connection_pool,
+            json_path,
+        }
+    }
+
+    /// Wraps `with_conn` in order to collect latency and error metrics
+    async fn with_measured_conn<F, R>(&self, op: DatabaseOperation, func: F) -> DatabaseResult<R>
+    where
+        F: Fn(&mut PgConnection) -> DatabaseResult<R> + Send + 'static,
+        R: Send + 'static,
+    {
+        let latency = &METRICS_REGISTRY
+            .metrics_group
+            .storage_controller_database_query_latency;
+        let _timer = latency.start_timer(DatabaseQueryLatencyLabelGroup {
+            operation: op.clone(),
+        });
+
+        let res = self.with_conn(func).await;
+
+        if let Err(err) = &res {
+            let error_counter = &METRICS_REGISTRY
+                .metrics_group
+                .storage_controller_database_query_error;
+            error_counter.inc(DatabaseQueryErrorLabelGroup {
+                error_type: err.error_label(),
+                operation: op,
+            })
+        }
+
+        res
+    }
+
+    /// Call the provided function in a tokio blocking thread, with a Diesel database connection.
+    async fn with_conn<F, R>(&self, func: F) -> DatabaseResult<R>
+    where
+        F: Fn(&mut PgConnection) -> DatabaseResult<R> + Send + 'static,
+        R: Send + 'static,
+    {
+        let mut conn = self.connection_pool.get()?;
+        tokio::task::spawn_blocking(move || -> DatabaseResult<R> { func(&mut conn) })
+            .await
+            .expect("Task panic")
+    }
+
+    /// When a node is first registered, persist it before using it for anything
+    pub(crate) async fn insert_node(&self, node: &Node) -> DatabaseResult<()> {
+        let np = node.to_persistent();
+        self.with_measured_conn(
+            DatabaseOperation::InsertNode,
+            move |conn| -> DatabaseResult<()> {
+                diesel::insert_into(crate::schema::nodes::table)
+                    .values(&np)
+                    .execute(conn)?;
+                Ok(())
+            },
+        )
+        .await
+    }
+
+    /// At startup, populate the list of nodes which our shards may be placed on
+    pub(crate) async fn list_nodes(&self) -> DatabaseResult<Vec<NodePersistence>> {
+        let nodes: Vec<NodePersistence> = self
+            .with_measured_conn(
+                DatabaseOperation::ListNodes,
+                move |conn| -> DatabaseResult<_> {
+                    Ok(crate::schema::nodes::table.load::<NodePersistence>(conn)?)
+                },
+            )
+            .await?;
+
+        tracing::info!("list_nodes: loaded {} nodes", nodes.len());
+
+        Ok(nodes)
+    }
+
+    pub(crate) async fn update_node(
+        &self,
+        input_node_id: NodeId,
+        input_scheduling: NodeSchedulingPolicy,
+    ) -> DatabaseResult<()> {
+        use crate::schema::nodes::dsl::*;
+        let updated = self
+            .with_measured_conn(DatabaseOperation::UpdateNode, move |conn| {
+                let updated = diesel::update(nodes)
+                    .filter(node_id.eq(input_node_id.0 as i64))
+                    .set((scheduling_policy.eq(String::from(input_scheduling)),))
+                    .execute(conn)?;
+                Ok(updated)
+            })
+            .await?;
+
+        if updated != 1 {
+            Err(DatabaseError::Logical(format!(
+                "Node {node_id:?} not found for update",
+            )))
+        } else {
+            Ok(())
+        }
+    }
+
+    /// At startup, load the high level state for shards, such as their config + policy.  This will
+    /// be enriched at runtime with state discovered on pageservers.
+    pub(crate) async fn list_tenant_shards(&self) -> DatabaseResult<Vec<TenantShardPersistence>> {
+        let loaded = self
+            .with_measured_conn(
+                DatabaseOperation::ListTenantShards,
+                move |conn| -> DatabaseResult<_> {
+                    Ok(crate::schema::tenant_shards::table.load::<TenantShardPersistence>(conn)?)
+                },
+            )
+            .await?;
+
+        if loaded.is_empty() {
+            if let Some(path) = &self.json_path {
+                if tokio::fs::try_exists(path)
+                    .await
+                    .map_err(|e| DatabaseError::Logical(format!("Error stat'ing JSON file: {e}")))?
+                {
+                    tracing::info!("Importing from legacy JSON format at {path}");
+                    return self.list_tenant_shards_json(path).await;
+                }
+            }
+        }
+        Ok(loaded)
+    }
+
+    /// Shim for automated compatibility tests: load tenants from a JSON file instead of database
+    pub(crate) async fn list_tenant_shards_json(
+        &self,
+        path: &Utf8Path,
+    ) -> DatabaseResult<Vec<TenantShardPersistence>> {
+        let bytes = tokio::fs::read(path)
+            .await
+            .map_err(|e| DatabaseError::Logical(format!("Failed to load JSON: {e}")))?;
+
+        let mut decoded = serde_json::from_slice::<JsonPersistence>(&bytes)
+            .map_err(|e| DatabaseError::Logical(format!("Deserialization error: {e}")))?;
+        for shard in decoded.tenants.values_mut() {
+            if shard.placement_policy == "\"Single\"" {
+                // Backward compat for test data after PR https://github.com/neondatabase/neon/pull/7165
+                shard.placement_policy = "{\"Attached\":0}".to_string();
+            }
+        }
+
+        let tenants: Vec<TenantShardPersistence> = decoded.tenants.into_values().collect();
+
+        // Synchronize database with what is in the JSON file
+        self.insert_tenant_shards(tenants.clone()).await?;
+
+        Ok(tenants)
+    }
+
+    /// For use in testing environments, where we dump out JSON on shutdown.
+    pub async fn write_tenants_json(&self) -> anyhow::Result<()> {
+        let Some(path) = &self.json_path else {
+            anyhow::bail!("Cannot write JSON if path isn't set (test environment bug)");
+        };
+        tracing::info!("Writing state to {path}...");
+        let tenants = self.list_tenant_shards().await?;
+        let mut tenants_map = HashMap::new();
+        for tsp in tenants {
+            let tenant_shard_id = TenantShardId {
+                tenant_id: TenantId::from_str(tsp.tenant_id.as_str())?,
+                shard_number: ShardNumber(tsp.shard_number as u8),
+                shard_count: ShardCount::new(tsp.shard_count as u8),
+            };
+
+            tenants_map.insert(tenant_shard_id, tsp);
+        }
+        let json = serde_json::to_string(&JsonPersistence {
+            tenants: tenants_map,
+        })?;
+
+        tokio::fs::write(path, &json).await?;
+        tracing::info!("Wrote {} bytes to {path}...", json.len());
+
+        Ok(())
+    }
+
+    /// Tenants must be persisted before we schedule them for the first time.  This enables us
+    /// to correctly retain generation monotonicity, and the externally provided placement policy & config.
+    pub(crate) async fn insert_tenant_shards(
+        &self,
+        shards: Vec<TenantShardPersistence>,
+    ) -> DatabaseResult<()> {
+        use crate::schema::tenant_shards::dsl::*;
+        self.with_measured_conn(
+            DatabaseOperation::InsertTenantShards,
+            move |conn| -> DatabaseResult<()> {
+                conn.transaction(|conn| -> QueryResult<()> {
+                    for tenant in &shards {
+                        diesel::insert_into(tenant_shards)
+                            .values(tenant)
+                            .execute(conn)?;
+                    }
+                    Ok(())
+                })?;
+                Ok(())
+            },
+        )
+        .await
+    }
+
+    /// Ordering: call this _after_ deleting the tenant on pageservers, but _before_ dropping state for
+    /// the tenant from memory on this server.
+    pub(crate) async fn delete_tenant(&self, del_tenant_id: TenantId) -> DatabaseResult<()> {
+        use crate::schema::tenant_shards::dsl::*;
+        self.with_measured_conn(
+            DatabaseOperation::DeleteTenant,
+            move |conn| -> DatabaseResult<()> {
+                diesel::delete(tenant_shards)
+                    .filter(tenant_id.eq(del_tenant_id.to_string()))
+                    .execute(conn)?;
+
+                Ok(())
+            },
+        )
+        .await
+    }
+
+    pub(crate) async fn delete_node(&self, del_node_id: NodeId) -> DatabaseResult<()> {
+        use crate::schema::nodes::dsl::*;
+        self.with_measured_conn(
+            DatabaseOperation::DeleteNode,
+            move |conn| -> DatabaseResult<()> {
+                diesel::delete(nodes)
+                    .filter(node_id.eq(del_node_id.0 as i64))
+                    .execute(conn)?;
+
+                Ok(())
+            },
+        )
+        .await
+    }
+
+    /// When a tenant invokes the /re-attach API, this function is responsible for doing an efficient
+    /// batched increment of the generations of all tenants whose generation_pageserver is equal to
+    /// the node that called /re-attach.
+    #[tracing::instrument(skip_all, fields(node_id))]
+    pub(crate) async fn re_attach(
+        &self,
+        node_id: NodeId,
+    ) -> DatabaseResult<HashMap<TenantShardId, Generation>> {
+        use crate::schema::tenant_shards::dsl::*;
+        let updated = self
+            .with_measured_conn(DatabaseOperation::ReAttach, move |conn| {
+                let rows_updated = diesel::update(tenant_shards)
+                    .filter(generation_pageserver.eq(node_id.0 as i64))
+                    .set(generation.eq(generation + 1))
+                    .execute(conn)?;
+
+                tracing::info!("Incremented {} tenants' generations", rows_updated);
+
+                // TODO: UPDATE+SELECT in one query
+
+                let updated = tenant_shards
+                    .filter(generation_pageserver.eq(node_id.0 as i64))
+                    .select(TenantShardPersistence::as_select())
+                    .load(conn)?;
+                Ok(updated)
+            })
+            .await?;
+
+        let mut result = HashMap::new();
+        for tsp in updated {
+            let tenant_shard_id = TenantShardId {
+                tenant_id: TenantId::from_str(tsp.tenant_id.as_str())
+                    .map_err(|e| DatabaseError::Logical(format!("Malformed tenant id: {e}")))?,
+                shard_number: ShardNumber(tsp.shard_number as u8),
+                shard_count: ShardCount::new(tsp.shard_count as u8),
+            };
+
+            let Some(g) = tsp.generation else {
+                // If the generation_pageserver column was non-NULL, then the generation column should also be non-NULL:
+                // we only set generation_pageserver when setting generation.
+                return Err(DatabaseError::Logical(
+                    "Generation should always be set after incrementing".to_string(),
+                ));
+            };
+            result.insert(tenant_shard_id, Generation::new(g as u32));
+        }
+
+        Ok(result)
+    }
+
+    /// Reconciler calls this immediately before attaching to a new pageserver, to acquire a unique, monotonically
+    /// advancing generation number.  We also store the NodeId for which the generation was issued, so that in
+    /// [`Self::re_attach`] we can do a bulk UPDATE on the generations for that node.
+    pub(crate) async fn increment_generation(
+        &self,
+        tenant_shard_id: TenantShardId,
+        node_id: NodeId,
+    ) -> anyhow::Result<Generation> {
+        use crate::schema::tenant_shards::dsl::*;
+        let updated = self
+            .with_measured_conn(DatabaseOperation::IncrementGeneration, move |conn| {
+                let updated = diesel::update(tenant_shards)
+                    .filter(tenant_id.eq(tenant_shard_id.tenant_id.to_string()))
+                    .filter(shard_number.eq(tenant_shard_id.shard_number.0 as i32))
+                    .filter(shard_count.eq(tenant_shard_id.shard_count.literal() as i32))
+                    .set((
+                        generation.eq(generation + 1),
+                        generation_pageserver.eq(node_id.0 as i64),
+                    ))
+                    // TODO: only returning() the generation column
+                    .returning(TenantShardPersistence::as_returning())
+                    .get_result(conn)?;
+
+                Ok(updated)
+            })
+            .await?;
+
+        // Generation is always non-null in the rseult: if the generation column had been NULL, then we
+        // should have experienced an SQL Confilict error while executing a query that tries to increment it.
+        debug_assert!(updated.generation.is_some());
+        let Some(g) = updated.generation else {
+            return Err(DatabaseError::Logical(
+                "Generation should always be set after incrementing".to_string(),
+            )
+            .into());
+        };
+
+        Ok(Generation::new(g as u32))
+    }
+
+    /// For use when updating a persistent property of a tenant, such as its config or placement_policy.
+    ///
+    /// Do not use this for settting generation, unless in the special onboarding code path (/location_config)
+    /// API: use [`Self::increment_generation`] instead.  Setting the generation via this route is a one-time thing
+    /// that we only do the first time a tenant is set to an attached policy via /location_config.
+    pub(crate) async fn update_tenant_shard(
+        &self,
+        tenant_shard_id: TenantShardId,
+        input_placement_policy: PlacementPolicy,
+        input_config: TenantConfig,
+        input_generation: Option<Generation>,
+    ) -> DatabaseResult<()> {
+        use crate::schema::tenant_shards::dsl::*;
+
+        self.with_measured_conn(DatabaseOperation::UpdateTenantShard, move |conn| {
+            let query = diesel::update(tenant_shards)
+                .filter(tenant_id.eq(tenant_shard_id.tenant_id.to_string()))
+                .filter(shard_number.eq(tenant_shard_id.shard_number.0 as i32))
+                .filter(shard_count.eq(tenant_shard_id.shard_count.literal() as i32));
+
+            if let Some(input_generation) = input_generation {
+                // Update includes generation column
+                query
+                    .set((
+                        generation.eq(Some(input_generation.into().unwrap() as i32)),
+                        placement_policy
+                            .eq(serde_json::to_string(&input_placement_policy).unwrap()),
+                        config.eq(serde_json::to_string(&input_config).unwrap()),
+                    ))
+                    .execute(conn)?;
+            } else {
+                // Update does not include generation column
+                query
+                    .set((
+                        placement_policy
+                            .eq(serde_json::to_string(&input_placement_policy).unwrap()),
+                        config.eq(serde_json::to_string(&input_config).unwrap()),
+                    ))
+                    .execute(conn)?;
+            }
+
+            Ok(())
+        })
+        .await?;
+
+        Ok(())
+    }
+
+    pub(crate) async fn update_tenant_config(
+        &self,
+        input_tenant_id: TenantId,
+        input_config: TenantConfig,
+    ) -> DatabaseResult<()> {
+        use crate::schema::tenant_shards::dsl::*;
+
+        self.with_measured_conn(DatabaseOperation::UpdateTenantConfig, move |conn| {
+            diesel::update(tenant_shards)
+                .filter(tenant_id.eq(input_tenant_id.to_string()))
+                .set((config.eq(serde_json::to_string(&input_config).unwrap()),))
+                .execute(conn)?;
+
+            Ok(())
+        })
+        .await?;
+
+        Ok(())
+    }
+
+    pub(crate) async fn detach(&self, tenant_shard_id: TenantShardId) -> anyhow::Result<()> {
+        use crate::schema::tenant_shards::dsl::*;
+        self.with_measured_conn(DatabaseOperation::Detach, move |conn| {
+            let updated = diesel::update(tenant_shards)
+                .filter(tenant_id.eq(tenant_shard_id.tenant_id.to_string()))
+                .filter(shard_number.eq(tenant_shard_id.shard_number.0 as i32))
+                .filter(shard_count.eq(tenant_shard_id.shard_count.literal() as i32))
+                .set((
+                    generation_pageserver.eq(Option::<i64>::None),
+                    placement_policy.eq(serde_json::to_string(&PlacementPolicy::Detached).unwrap()),
+                ))
+                .execute(conn)?;
+
+            Ok(updated)
+        })
+        .await?;
+
+        Ok(())
+    }
+
+    // When we start shard splitting, we must durably mark the tenant so that
+    // on restart, we know that we must go through recovery.
+    //
+    // We create the child shards here, so that they will be available for increment_generation calls
+    // if some pageserver holding a child shard needs to restart before the overall tenant split is complete.
+    pub(crate) async fn begin_shard_split(
+        &self,
+        old_shard_count: ShardCount,
+        split_tenant_id: TenantId,
+        parent_to_children: Vec<(TenantShardId, Vec<TenantShardPersistence>)>,
+    ) -> DatabaseResult<()> {
+        use crate::schema::tenant_shards::dsl::*;
+        self.with_measured_conn(DatabaseOperation::BeginShardSplit, move |conn| -> DatabaseResult<()> {
+            conn.transaction(|conn| -> DatabaseResult<()> {
+                // Mark parent shards as splitting
+
+                let updated = diesel::update(tenant_shards)
+                    .filter(tenant_id.eq(split_tenant_id.to_string()))
+                    .filter(shard_count.eq(old_shard_count.literal() as i32))
+                    .set((splitting.eq(1),))
+                    .execute(conn)?;
+                if u8::try_from(updated)
+                    .map_err(|_| DatabaseError::Logical(
+                        format!("Overflow existing shard count {} while splitting", updated))
+                    )? != old_shard_count.count() {
+                    // Perhaps a deletion or another split raced with this attempt to split, mutating
+                    // the parent shards that we intend to split. In this case the split request should fail.
+                    return Err(DatabaseError::Logical(
+                        format!("Unexpected existing shard count {updated} when preparing tenant for split (expected {})", old_shard_count.count())
+                    ));
+                }
+
+                // FIXME: spurious clone to sidestep closure move rules
+                let parent_to_children = parent_to_children.clone();
+
+                // Insert child shards
+                for (parent_shard_id, children) in parent_to_children {
+                    let mut parent = crate::schema::tenant_shards::table
+                        .filter(tenant_id.eq(parent_shard_id.tenant_id.to_string()))
+                        .filter(shard_number.eq(parent_shard_id.shard_number.0 as i32))
+                        .filter(shard_count.eq(parent_shard_id.shard_count.literal() as i32))
+                        .load::<TenantShardPersistence>(conn)?;
+                    let parent = if parent.len() != 1 {
+                        return Err(DatabaseError::Logical(format!(
+                            "Parent shard {parent_shard_id} not found"
+                        )));
+                    } else {
+                        parent.pop().unwrap()
+                    };
+                    for mut shard in children {
+                        // Carry the parent's generation into the child
+                        shard.generation = parent.generation;
+
+                        debug_assert!(shard.splitting == SplitState::Splitting);
+                        diesel::insert_into(tenant_shards)
+                            .values(shard)
+                            .execute(conn)?;
+                    }
+                }
+
+                Ok(())
+            })?;
+
+            Ok(())
+        })
+        .await
+    }
+
+    // When we finish shard splitting, we must atomically clean up the old shards
+    // and insert the new shards, and clear the splitting marker.
+    pub(crate) async fn complete_shard_split(
+        &self,
+        split_tenant_id: TenantId,
+        old_shard_count: ShardCount,
+    ) -> DatabaseResult<()> {
+        use crate::schema::tenant_shards::dsl::*;
+        self.with_measured_conn(
+            DatabaseOperation::CompleteShardSplit,
+            move |conn| -> DatabaseResult<()> {
+                conn.transaction(|conn| -> QueryResult<()> {
+                    // Drop parent shards
+                    diesel::delete(tenant_shards)
+                        .filter(tenant_id.eq(split_tenant_id.to_string()))
+                        .filter(shard_count.eq(old_shard_count.literal() as i32))
+                        .execute(conn)?;
+
+                    // Clear sharding flag
+                    let updated = diesel::update(tenant_shards)
+                        .filter(tenant_id.eq(split_tenant_id.to_string()))
+                        .set((splitting.eq(0),))
+                        .execute(conn)?;
+                    debug_assert!(updated > 0);
+
+                    Ok(())
+                })?;
+
+                Ok(())
+            },
+        )
+        .await
+    }
+
+    /// Used when the remote part of a shard split failed: we will revert the database state to have only
+    /// the parent shards, with SplitState::Idle.
+    pub(crate) async fn abort_shard_split(
+        &self,
+        split_tenant_id: TenantId,
+        new_shard_count: ShardCount,
+    ) -> DatabaseResult<AbortShardSplitStatus> {
+        use crate::schema::tenant_shards::dsl::*;
+        self.with_measured_conn(
+            DatabaseOperation::AbortShardSplit,
+            move |conn| -> DatabaseResult<AbortShardSplitStatus> {
+                let aborted =
+                    conn.transaction(|conn| -> DatabaseResult<AbortShardSplitStatus> {
+                        // Clear the splitting state on parent shards
+                        let updated = diesel::update(tenant_shards)
+                            .filter(tenant_id.eq(split_tenant_id.to_string()))
+                            .filter(shard_count.ne(new_shard_count.literal() as i32))
+                            .set((splitting.eq(0),))
+                            .execute(conn)?;
+
+                        // Parent shards are already gone: we cannot abort.
+                        if updated == 0 {
+                            return Ok(AbortShardSplitStatus::Complete);
+                        }
+
+                        // Sanity check: if parent shards were present, their cardinality should
+                        // be less than the number of child shards.
+                        if updated >= new_shard_count.count() as usize {
+                            return Err(DatabaseError::Logical(format!(
+                                "Unexpected parent shard count {updated} while aborting split to \
+                            count {new_shard_count:?} on tenant {split_tenant_id}"
+                            )));
+                        }
+
+                        // Erase child shards
+                        diesel::delete(tenant_shards)
+                            .filter(tenant_id.eq(split_tenant_id.to_string()))
+                            .filter(shard_count.eq(new_shard_count.literal() as i32))
+                            .execute(conn)?;
+
+                        Ok(AbortShardSplitStatus::Aborted)
+                    })?;
+
+                Ok(aborted)
+            },
+        )
+        .await
+    }
+}
+
+/// Parts of [`crate::tenant_state::TenantState`] that are stored durably
+#[derive(Queryable, Selectable, Insertable, Serialize, Deserialize, Clone, Eq, PartialEq)]
+#[diesel(table_name = crate::schema::tenant_shards)]
+pub(crate) struct TenantShardPersistence {
+    #[serde(default)]
+    pub(crate) tenant_id: String,
+    #[serde(default)]
+    pub(crate) shard_number: i32,
+    #[serde(default)]
+    pub(crate) shard_count: i32,
+    #[serde(default)]
+    pub(crate) shard_stripe_size: i32,
+
+    // Latest generation number: next time we attach, increment this
+    // and use the incremented number when attaching.
+    //
+    // Generation is only None when first onboarding a tenant, where it may
+    // be in PlacementPolicy::Secondary and therefore have no valid generation state.
+    pub(crate) generation: Option<i32>,
+
+    // Currently attached pageserver
+    #[serde(rename = "pageserver")]
+    pub(crate) generation_pageserver: Option<i64>,
+
+    #[serde(default)]
+    pub(crate) placement_policy: String,
+    #[serde(default)]
+    pub(crate) splitting: SplitState,
+    #[serde(default)]
+    pub(crate) config: String,
+}
+
+impl TenantShardPersistence {
+    pub(crate) fn get_shard_identity(&self) -> Result<ShardIdentity, ShardConfigError> {
+        if self.shard_count == 0 {
+            Ok(ShardIdentity::unsharded())
+        } else {
+            Ok(ShardIdentity::new(
+                ShardNumber(self.shard_number as u8),
+                ShardCount::new(self.shard_count as u8),
+                ShardStripeSize(self.shard_stripe_size as u32),
+            )?)
+        }
+    }
+
+    pub(crate) fn get_tenant_shard_id(&self) -> Result<TenantShardId, hex::FromHexError> {
+        Ok(TenantShardId {
+            tenant_id: TenantId::from_str(self.tenant_id.as_str())?,
+            shard_number: ShardNumber(self.shard_number as u8),
+            shard_count: ShardCount::new(self.shard_count as u8),
+        })
+    }
+}
+
+/// Parts of [`crate::node::Node`] that are stored durably
+#[derive(Serialize, Deserialize, Queryable, Selectable, Insertable, Eq, PartialEq)]
+#[diesel(table_name = crate::schema::nodes)]
+pub(crate) struct NodePersistence {
+    pub(crate) node_id: i64,
+    pub(crate) scheduling_policy: String,
+    pub(crate) listen_http_addr: String,
+    pub(crate) listen_http_port: i32,
+    pub(crate) listen_pg_addr: String,
+    pub(crate) listen_pg_port: i32,
+}

control_plane/attachment_service/src/reconciler.rs

@@ -1,39 +1,34 @@
 use crate::pageserver_client::PageserverClient;
 use crate::persistence::Persistence;
 use crate::service;
-use pageserver_api::controller_api::PlacementPolicy;
+use hyper::StatusCode;
 use pageserver_api::models::{
     LocationConfig, LocationConfigMode, LocationConfigSecondary, TenantConfig,
 };
 use pageserver_api::shard::{ShardIdentity, TenantShardId};
 use pageserver_client::mgmt_api;
-use reqwest::StatusCode;
 use std::collections::HashMap;
 use std::sync::Arc;
 use std::time::{Duration, Instant};
 use tokio_util::sync::CancellationToken;
-use utils::backoff::exponential_backoff;
-use utils::failpoint_support;
 use utils::generation::Generation;
 use utils::id::{NodeId, TimelineId};
 use utils::lsn::Lsn;
-use utils::pausable_failpoint;
 use utils::sync::gate::GateGuard;
 
 use crate::compute_hook::{ComputeHook, NotifyError};
 use crate::node::Node;
-use crate::tenant_shard::{IntentState, ObservedState, ObservedStateLocation};
+use crate::tenant_state::{IntentState, ObservedState, ObservedStateLocation};
 
 const DEFAULT_HEATMAP_PERIOD: &str = "60s";
 
 /// Object with the lifetime of the background reconcile task that is created
 /// for tenants which have a difference between their intent and observed states.
 pub(super) struct Reconciler {
-    /// See [`crate::tenant_shard::TenantShard`] for the meanings of these fields: they are a snapshot
+    /// See [`crate::tenant_state::TenantState`] for the meanings of these fields: they are a snapshot
     /// of a tenant's state from when we spawned a reconcile task.
     pub(super) tenant_shard_id: TenantShardId,
     pub(crate) shard: ShardIdentity,
-    pub(crate) placement_policy: PlacementPolicy,
     pub(crate) generation: Option<Generation>,
     pub(crate) intent: TargetState,
 
@@ -41,9 +36,6 @@ pub(super) struct Reconciler {
     /// to detach this tenant shard.
     pub(crate) detach: Vec<Node>,
 
-    /// Configuration specific to this reconciler
-    pub(crate) reconciler_config: ReconcilerConfig,
-
     pub(crate) config: TenantConfig,
     pub(crate) observed: ObservedState,
 
@@ -56,15 +48,11 @@ pub(super) struct Reconciler {
 
     /// To avoid stalling if the cloud control plane is unavailable, we may proceed
     /// past failures in [`ComputeHook::notify`], but we _must_ remember that we failed
-    /// so that we can set [`crate::tenant_shard::TenantShard::pending_compute_notification`] to ensure a later retry.
+    /// so that we can set [`crate::tenant_state::TenantState::pending_compute_notification`] to ensure a later retry.
     pub(crate) compute_notify_failure: bool,
 
-    /// Reconciler is responsible for keeping alive semaphore units that limit concurrency on how many
-    /// we will spawn.
-    pub(crate) _resource_units: ReconcileUnits,
-
     /// A means to abort background reconciliation: it is essential to
-    /// call this when something changes in the original TenantShard that
+    /// call this when something changes in the original TenantState that
     /// will make this reconciliation impossible or unnecessary, for
     /// example when a pageserver node goes offline, or the PlacementPolicy for
     /// the tenant is changed.
@@ -78,79 +66,7 @@ pub(super) struct Reconciler {
     pub(crate) persistence: Arc<Persistence>,
 }
 
-pub(crate) struct ReconcilerConfigBuilder {
-    config: ReconcilerConfig,
-}
-
-impl ReconcilerConfigBuilder {
-    pub(crate) fn new() -> Self {
-        Self {
-            config: ReconcilerConfig::default(),
-        }
-    }
-
-    pub(crate) fn secondary_warmup_timeout(self, value: Duration) -> Self {
-        Self {
-            config: ReconcilerConfig {
-                secondary_warmup_timeout: Some(value),
-                ..self.config
-            },
-        }
-    }
-
-    pub(crate) fn secondary_download_request_timeout(self, value: Duration) -> Self {
-        Self {
-            config: ReconcilerConfig {
-                secondary_download_request_timeout: Some(value),
-                ..self.config
-            },
-        }
-    }
-
-    pub(crate) fn build(self) -> ReconcilerConfig {
-        self.config
-    }
-}
-
-#[derive(Default, Debug, Copy, Clone)]
-pub(crate) struct ReconcilerConfig {
-    // During live migration give up on warming-up the secondary
-    // after this timeout.
-    secondary_warmup_timeout: Option<Duration>,
-
-    // During live migrations this is the amount of time that
-    // the pagserver will hold our poll.
-    secondary_download_request_timeout: Option<Duration>,
-}
-
-impl ReconcilerConfig {
-    pub(crate) fn get_secondary_warmup_timeout(&self) -> Duration {
-        const SECONDARY_WARMUP_TIMEOUT_DEFAULT: Duration = Duration::from_secs(300);
-        self.secondary_warmup_timeout
-            .unwrap_or(SECONDARY_WARMUP_TIMEOUT_DEFAULT)
-    }
-
-    pub(crate) fn get_secondary_download_request_timeout(&self) -> Duration {
-        const SECONDARY_DOWNLOAD_REQUEST_TIMEOUT_DEFAULT: Duration = Duration::from_secs(20);
-        self.secondary_download_request_timeout
-            .unwrap_or(SECONDARY_DOWNLOAD_REQUEST_TIMEOUT_DEFAULT)
-    }
-}
-
-/// RAII resource units granted to a Reconciler, which it should keep alive until it finishes doing I/O
-pub(crate) struct ReconcileUnits {
-    _sem_units: tokio::sync::OwnedSemaphorePermit,
-}
-
-impl ReconcileUnits {
-    pub(crate) fn new(sem_units: tokio::sync::OwnedSemaphorePermit) -> Self {
-        Self {
-            _sem_units: sem_units,
-        }
-    }
-}
-
-/// This is a snapshot of [`crate::tenant_shard::IntentState`], but it does not do any
+/// This is a snapshot of [`crate::tenant_state::IntentState`], but it does not do any
 /// reference counting for Scheduler.  The IntentState is what the scheduler works with,
 /// and the TargetState is just the instruction for a particular Reconciler run.
 #[derive(Debug)]
@@ -364,13 +280,11 @@ impl Reconciler {
     ) -> Result<(), ReconcileError> {
         // This is not the timeout for a request, but the total amount of time we're willing to wait
         // for a secondary location to get up to date before
-        let total_download_timeout = self.reconciler_config.get_secondary_warmup_timeout();
+        const TOTAL_DOWNLOAD_TIMEOUT: Duration = Duration::from_secs(300);
 
         // This the long-polling interval for the secondary download requests we send to destination pageserver
         // during a migration.
-        let request_download_timeout = self
-            .reconciler_config
-            .get_secondary_download_request_timeout();
+        const REQUEST_DOWNLOAD_TIMEOUT: Duration = Duration::from_secs(20);
 
         let started_at = Instant::now();
 
@@ -381,14 +295,14 @@ impl Reconciler {
                         client
                             .tenant_secondary_download(
                                 tenant_shard_id,
-                                Some(request_download_timeout),
+                                Some(REQUEST_DOWNLOAD_TIMEOUT),
                             )
                             .await
                     },
                     &self.service_config.jwt_token,
                     1,
                     3,
-                    request_download_timeout * 2,
+                    REQUEST_DOWNLOAD_TIMEOUT * 2,
                     &self.cancel,
                 )
                 .await
@@ -416,7 +330,7 @@ impl Reconciler {
                 return Ok(());
             } else if status == StatusCode::ACCEPTED {
                 let total_runtime = started_at.elapsed();
-                if total_runtime > total_download_timeout {
+                if total_runtime > TOTAL_DOWNLOAD_TIMEOUT {
                     tracing::warn!("Timed out after {}ms downloading layers to {node}.  Progress so far: {}/{} layers, {}/{} bytes",
                         total_runtime.as_millis(),
                         progress.layers_downloaded,
@@ -570,32 +484,17 @@ impl Reconciler {
 
         // During a live migration it is unhelpful to proceed if we couldn't notify compute: if we detach
         // the origin without notifying compute, we will render the tenant unavailable.
-        let mut notify_attempts = 0;
         while let Err(e) = self.compute_notify().await {
             match e {
                 NotifyError::Fatal(_) => return Err(ReconcileError::Notify(e)),
-                NotifyError::ShuttingDown => return Err(ReconcileError::Cancel),
                 _ => {
                     tracing::warn!(
                         "Live migration blocked by compute notification error, retrying: {e}"
                     );
                 }
             }
-
-            exponential_backoff(
-                notify_attempts,
-                // Generous waits: control plane operations which might be blocking us usually complete on the order
-                // of hundreds to thousands of milliseconds, so no point busy polling.
-                1.0,
-                10.0,
-                &self.cancel,
-            )
-            .await;
-            notify_attempts += 1;
         }
 
-        pausable_failpoint!("reconciler-live-migrate-post-notify");
-
         // Downgrade the origin to secondary.  If the tenant's policy is PlacementPolicy::Attached(0), then
         // this location will be deleted in the general case reconciliation that runs after this.
         let origin_secondary_conf = build_location_config(
@@ -724,7 +623,7 @@ impl Reconciler {
                 generation,
                 &self.shard,
                 &self.config,
-                &self.placement_policy,
+                !self.intent.secondary.is_empty(),
             );
             match self.observed.locations.get(&node.get_id()) {
                 Some(conf) if conf.conf.as_ref() == Some(&wanted_conf) => {
@@ -736,8 +635,11 @@ impl Reconciler {
                     // reconcile this location.  This includes locations with different configurations, as well
                     // as locations with unknown (None) observed state.
 
-                    // Incrementing generation is the safe general case, but is inefficient for changes that only
-                    // modify some details (e.g. the tenant's config).
+                    // The general case is to increment the generation.  However, there are cases
+                    // where this is not necessary:
+                    // - if we are only updating the TenantConf part of the location
+                    // - if we are only changing the attachment mode (e.g. going to attachedmulti or attachedstale)
+                    //   and the location was already in the correct generation
                     let increment_generation = match observed {
                         None => true,
                         Some(ObservedStateLocation { conf: None }) => true,
@@ -746,11 +648,18 @@ impl Reconciler {
                         }) => {
                             let generations_match = observed.generation == wanted_conf.generation;
 
-                            // We may skip incrementing the generation if the location is already in the expected mode and
-                            // generation.  In principle it would also be safe to skip from certain other modes (e.g. AttachedStale),
-                            // but such states are handled inside `live_migrate`, and if we see that state here we're cleaning up
-                            // after a restart/crash, so fall back to the universally safe path of incrementing generation.
-                            !generations_match || (observed.mode != wanted_conf.mode)
+                            use LocationConfigMode::*;
+                            let mode_transition_requires_gen_inc =
+                                match (observed.mode, wanted_conf.mode) {
+                                    // Usually the short-lived attachment modes (multi and stale) are only used
+                                    // in the case of [`Self::live_migrate`], but it is simple to handle them correctly
+                                    // here too.  Locations are allowed to go Single->Stale and Multi->Single within the same generation.
+                                    (AttachedSingle, AttachedStale) => false,
+                                    (AttachedMulti, AttachedSingle) => false,
+                                    (lhs, rhs) => lhs != rhs,
+                                };
+
+                            !generations_match || mode_transition_requires_gen_inc
                         }
                     };
 
@@ -820,8 +729,6 @@ impl Reconciler {
             self.location_config(&node, conf, None, false).await?;
         }
 
-        failpoint_support::sleep_millis_async!("sleep-on-reconcile-epilogue");
-
         Ok(())
     }
 
@@ -842,10 +749,7 @@ impl Reconciler {
                 // It is up to the caller whether they want to drop out on this error, but they don't have to:
                 // in general we should avoid letting unavailability of the cloud control plane stop us from
                 // making progress.
-                if !matches!(e, NotifyError::ShuttingDown) {
-                    tracing::warn!("Failed to notify compute of attached pageserver {node}: {e}");
-                }
-
+                tracing::warn!("Failed to notify compute of attached pageserver {node}: {e}");
                 // Set this flag so that in our ReconcileResult we will set the flag on the shard that it
                 // needs to retry at some point.
                 self.compute_notify_failure = true;
@@ -876,15 +780,8 @@ pub(crate) fn attached_location_conf(
     generation: Generation,
     shard: &ShardIdentity,
     config: &TenantConfig,
-    policy: &PlacementPolicy,
+    has_secondaries: bool,
 ) -> LocationConfig {
-    let has_secondaries = match policy {
-        PlacementPolicy::Attached(0) | PlacementPolicy::Detached | PlacementPolicy::Secondary => {
-            false
-        }
-        PlacementPolicy::Attached(_) => true,
-    };
-
     LocationConfig {
         mode: LocationConfigMode::AttachedSingle,
         generation: generation.into(),

control_plane/attachment_service/src/scheduler.rs

@@ -0,0 +1,352 @@
+use crate::{node::Node, tenant_state::TenantState};
+use pageserver_api::controller_api::UtilizationScore;
+use serde::Serialize;
+use std::collections::HashMap;
+use utils::{http::error::ApiError, id::NodeId};
+
+/// Scenarios in which we cannot find a suitable location for a tenant shard
+#[derive(thiserror::Error, Debug)]
+pub enum ScheduleError {
+    #[error("No pageservers found")]
+    NoPageservers,
+    #[error("No pageserver found matching constraint")]
+    ImpossibleConstraint,
+}
+
+impl From<ScheduleError> for ApiError {
+    fn from(value: ScheduleError) -> Self {
+        ApiError::Conflict(format!("Scheduling error: {}", value))
+    }
+}
+
+#[derive(Serialize, Eq, PartialEq)]
+pub enum MaySchedule {
+    Yes(UtilizationScore),
+    No,
+}
+
+#[derive(Serialize)]
+struct SchedulerNode {
+    /// How many shards are currently scheduled on this node, via their [`crate::tenant_state::IntentState`].
+    shard_count: usize,
+
+    /// Whether this node is currently elegible to have new shards scheduled (this is derived
+    /// from a node's availability state and scheduling policy).
+    may_schedule: MaySchedule,
+}
+
+impl PartialEq for SchedulerNode {
+    fn eq(&self, other: &Self) -> bool {
+        let may_schedule_matches = matches!(
+            (&self.may_schedule, &other.may_schedule),
+            (MaySchedule::Yes(_), MaySchedule::Yes(_)) | (MaySchedule::No, MaySchedule::No)
+        );
+
+        may_schedule_matches && self.shard_count == other.shard_count
+    }
+}
+
+impl Eq for SchedulerNode {}
+
+/// This type is responsible for selecting which node is used when a tenant shard needs to choose a pageserver
+/// on which to run.
+///
+/// The type has no persistent state of its own: this is all populated at startup.  The Serialize
+/// impl is only for debug dumps.
+#[derive(Serialize)]
+pub(crate) struct Scheduler {
+    nodes: HashMap<NodeId, SchedulerNode>,
+}
+
+impl Scheduler {
+    pub(crate) fn new<'a>(nodes: impl Iterator<Item = &'a Node>) -> Self {
+        let mut scheduler_nodes = HashMap::new();
+        for node in nodes {
+            scheduler_nodes.insert(
+                node.get_id(),
+                SchedulerNode {
+                    shard_count: 0,
+                    may_schedule: node.may_schedule(),
+                },
+            );
+        }
+
+        Self {
+            nodes: scheduler_nodes,
+        }
+    }
+
+    /// For debug/support: check that our internal statistics are in sync with the state of
+    /// the nodes & tenant shards.
+    ///
+    /// If anything is inconsistent, log details and return an error.
+    pub(crate) fn consistency_check<'a>(
+        &self,
+        nodes: impl Iterator<Item = &'a Node>,
+        shards: impl Iterator<Item = &'a TenantState>,
+    ) -> anyhow::Result<()> {
+        let mut expect_nodes: HashMap<NodeId, SchedulerNode> = HashMap::new();
+        for node in nodes {
+            expect_nodes.insert(
+                node.get_id(),
+                SchedulerNode {
+                    shard_count: 0,
+                    may_schedule: node.may_schedule(),
+                },
+            );
+        }
+
+        for shard in shards {
+            if let Some(node_id) = shard.intent.get_attached() {
+                match expect_nodes.get_mut(node_id) {
+                    Some(node) => node.shard_count += 1,
+                    None => anyhow::bail!(
+                        "Tenant {} references nonexistent node {}",
+                        shard.tenant_shard_id,
+                        node_id
+                    ),
+                }
+            }
+
+            for node_id in shard.intent.get_secondary() {
+                match expect_nodes.get_mut(node_id) {
+                    Some(node) => node.shard_count += 1,
+                    None => anyhow::bail!(
+                        "Tenant {} references nonexistent node {}",
+                        shard.tenant_shard_id,
+                        node_id
+                    ),
+                }
+            }
+        }
+
+        for (node_id, expect_node) in &expect_nodes {
+            let Some(self_node) = self.nodes.get(node_id) else {
+                anyhow::bail!("Node {node_id} not found in Self")
+            };
+
+            if self_node != expect_node {
+                tracing::error!("Inconsistency detected in scheduling state for node {node_id}");
+                tracing::error!("Expected state: {}", serde_json::to_string(expect_node)?);
+                tracing::error!("Self state: {}", serde_json::to_string(self_node)?);
+
+                anyhow::bail!("Inconsistent state on {node_id}");
+            }
+        }
+
+        if expect_nodes.len() != self.nodes.len() {
+            // We just checked that all the expected nodes are present.  If the lengths don't match,
+            // it means that we have nodes in Self that are unexpected.
+            for node_id in self.nodes.keys() {
+                if !expect_nodes.contains_key(node_id) {
+                    anyhow::bail!("Node {node_id} found in Self but not in expected nodes");
+                }
+            }
+        }
+
+        Ok(())
+    }
+
+    /// Increment the reference count of a node.  This reference count is used to guide scheduling
+    /// decisions, not for memory management: it represents one tenant shard whose IntentState targets
+    /// this node.
+    ///
+    /// It is an error to call this for a node that is not known to the scheduler (i.e. passed into
+    /// [`Self::new`] or [`Self::node_upsert`])
+    pub(crate) fn node_inc_ref(&mut self, node_id: NodeId) {
+        let Some(node) = self.nodes.get_mut(&node_id) else {
+            tracing::error!("Scheduler missing node {node_id}");
+            debug_assert!(false);
+            return;
+        };
+
+        node.shard_count += 1;
+    }
+
+    /// Decrement a node's reference count.  Inverse of [`Self::node_inc_ref`].
+    pub(crate) fn node_dec_ref(&mut self, node_id: NodeId) {
+        let Some(node) = self.nodes.get_mut(&node_id) else {
+            debug_assert!(false);
+            tracing::error!("Scheduler missing node {node_id}");
+            return;
+        };
+
+        node.shard_count -= 1;
+    }
+
+    pub(crate) fn node_upsert(&mut self, node: &Node) {
+        use std::collections::hash_map::Entry::*;
+        match self.nodes.entry(node.get_id()) {
+            Occupied(mut entry) => {
+                entry.get_mut().may_schedule = node.may_schedule();
+            }
+            Vacant(entry) => {
+                entry.insert(SchedulerNode {
+                    shard_count: 0,
+                    may_schedule: node.may_schedule(),
+                });
+            }
+        }
+    }
+
+    pub(crate) fn node_remove(&mut self, node_id: NodeId) {
+        if self.nodes.remove(&node_id).is_none() {
+            tracing::warn!(node_id=%node_id, "Removed non-existent node from scheduler");
+        }
+    }
+
+    /// Where we have several nodes to choose from, for example when picking a secondary location
+    /// to promote to an attached location, this method may be used to pick the best choice based
+    /// on the scheduler's knowledge of utilization and availability.
+    ///
+    /// If the input is empty, or all the nodes are not elegible for scheduling, return None: the
+    /// caller can pick a node some other way.
+    pub(crate) fn node_preferred(&self, nodes: &[NodeId]) -> Option<NodeId> {
+        if nodes.is_empty() {
+            return None;
+        }
+
+        // TODO: When the utilization score returned by the pageserver becomes meaningful,
+        // schedule based on that instead of the shard count.
+        let node = nodes
+            .iter()
+            .map(|node_id| {
+                let may_schedule = self
+                    .nodes
+                    .get(node_id)
+                    .map(|n| n.may_schedule != MaySchedule::No)
+                    .unwrap_or(false);
+                (*node_id, may_schedule)
+            })
+            .max_by_key(|(_n, may_schedule)| *may_schedule);
+
+        // If even the preferred node has may_schedule==false, return None
+        node.and_then(|(node_id, may_schedule)| if may_schedule { Some(node_id) } else { None })
+    }
+
+    pub(crate) fn schedule_shard(&self, hard_exclude: &[NodeId]) -> Result<NodeId, ScheduleError> {
+        if self.nodes.is_empty() {
+            return Err(ScheduleError::NoPageservers);
+        }
+
+        let mut tenant_counts: Vec<(NodeId, usize)> = self
+            .nodes
+            .iter()
+            .filter_map(|(k, v)| {
+                if hard_exclude.contains(k) || v.may_schedule == MaySchedule::No {
+                    None
+                } else {
+                    Some((*k, v.shard_count))
+                }
+            })
+            .collect();
+
+        // Sort by tenant count.  Nodes with the same tenant count are sorted by ID.
+        tenant_counts.sort_by_key(|i| (i.1, i.0));
+
+        if tenant_counts.is_empty() {
+            // After applying constraints, no pageservers were left.  We log some detail about
+            // the state of nodes to help understand why this happened.  This is not logged as an error because
+            // it is legitimately possible for enough nodes to be Offline to prevent scheduling a shard.
+            tracing::info!("Scheduling failure, while excluding {hard_exclude:?}, node states:");
+            for (node_id, node) in &self.nodes {
+                tracing::info!(
+                    "Node {node_id}: may_schedule={} shards={}",
+                    node.may_schedule != MaySchedule::No,
+                    node.shard_count
+                );
+            }
+
+            return Err(ScheduleError::ImpossibleConstraint);
+        }
+
+        let node_id = tenant_counts.first().unwrap().0;
+        tracing::info!(
+            "scheduler selected node {node_id} (elegible nodes {:?}, exclude: {hard_exclude:?})",
+            tenant_counts.iter().map(|i| i.0 .0).collect::<Vec<_>>()
+        );
+
+        // Note that we do not update shard count here to reflect the scheduling: that
+        // is IntentState's job when the scheduled location is used.
+
+        Ok(node_id)
+    }
+}
+
+#[cfg(test)]
+pub(crate) mod test_utils {
+
+    use crate::node::Node;
+    use pageserver_api::controller_api::{NodeAvailability, UtilizationScore};
+    use std::collections::HashMap;
+    use utils::id::NodeId;
+    /// Test helper: synthesize the requested number of nodes, all in active state.
+    ///
+    /// Node IDs start at one.
+    pub(crate) fn make_test_nodes(n: u64) -> HashMap<NodeId, Node> {
+        (1..n + 1)
+            .map(|i| {
+                (NodeId(i), {
+                    let mut node = Node::new(
+                        NodeId(i),
+                        format!("httphost-{i}"),
+                        80 + i as u16,
+                        format!("pghost-{i}"),
+                        5432 + i as u16,
+                    );
+                    node.set_availability(NodeAvailability::Active(UtilizationScore::worst()));
+                    assert!(node.is_available());
+                    node
+                })
+            })
+            .collect()
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    use crate::tenant_state::IntentState;
+    #[test]
+    fn scheduler_basic() -> anyhow::Result<()> {
+        let nodes = test_utils::make_test_nodes(2);
+
+        let mut scheduler = Scheduler::new(nodes.values());
+        let mut t1_intent = IntentState::new();
+        let mut t2_intent = IntentState::new();
+
+        let scheduled = scheduler.schedule_shard(&[])?;
+        t1_intent.set_attached(&mut scheduler, Some(scheduled));
+        let scheduled = scheduler.schedule_shard(&[])?;
+        t2_intent.set_attached(&mut scheduler, Some(scheduled));
+
+        assert_eq!(scheduler.nodes.get(&NodeId(1)).unwrap().shard_count, 1);
+        assert_eq!(scheduler.nodes.get(&NodeId(2)).unwrap().shard_count, 1);
+
+        let scheduled = scheduler.schedule_shard(&t1_intent.all_pageservers())?;
+        t1_intent.push_secondary(&mut scheduler, scheduled);
+
+        assert_eq!(scheduler.nodes.get(&NodeId(1)).unwrap().shard_count, 1);
+        assert_eq!(scheduler.nodes.get(&NodeId(2)).unwrap().shard_count, 2);
+
+        t1_intent.clear(&mut scheduler);
+        assert_eq!(scheduler.nodes.get(&NodeId(1)).unwrap().shard_count, 0);
+        assert_eq!(scheduler.nodes.get(&NodeId(2)).unwrap().shard_count, 1);
+
+        if cfg!(debug_assertions) {
+            // Dropping an IntentState without clearing it causes a panic in debug mode,
+            // because we have failed to properly update scheduler shard counts.
+            let result = std::panic::catch_unwind(move || {
+                drop(t2_intent);
+            });
+            assert!(result.is_err());
+        } else {
+            t2_intent.clear(&mut scheduler);
+            assert_eq!(scheduler.nodes.get(&NodeId(1)).unwrap().shard_count, 0);
+            assert_eq!(scheduler.nodes.get(&NodeId(2)).unwrap().shard_count, 0);
+        }
+
+        Ok(())
+    }
+}

control_plane/attachment_service/src/schema.rs

@@ -0,0 +1,28 @@
+// @generated automatically by Diesel CLI.
+
+diesel::table! {
+    nodes (node_id) {
+        node_id -> Int8,
+        scheduling_policy -> Varchar,
+        listen_http_addr -> Varchar,
+        listen_http_port -> Int4,
+        listen_pg_addr -> Varchar,
+        listen_pg_port -> Int4,
+    }
+}
+
+diesel::table! {
+    tenant_shards (tenant_id, shard_number, shard_count) {
+        tenant_id -> Varchar,
+        shard_number -> Int4,
+        shard_count -> Int4,
+        shard_stripe_size -> Int4,
+        generation -> Nullable<Int4>,
+        generation_pageserver -> Nullable<Int8>,
+        placement_policy -> Varchar,
+        splitting -> Int2,
+        config -> Text,
+    }
+}
+
+diesel::allow_tables_to_appear_in_same_query!(nodes, tenant_shards,);

control_plane/attachment_service/src/tenant_state.rs

@@ -0,0 +1,983 @@
+use std::{
+    collections::{HashMap, HashSet},
+    sync::Arc,
+    time::Duration,
+};
+
+use crate::{
+    metrics::{self, ReconcileCompleteLabelGroup, ReconcileOutcome},
+    persistence::TenantShardPersistence,
+};
+use pageserver_api::controller_api::PlacementPolicy;
+use pageserver_api::{
+    models::{LocationConfig, LocationConfigMode, TenantConfig},
+    shard::{ShardIdentity, TenantShardId},
+};
+use serde::Serialize;
+use tokio::task::JoinHandle;
+use tokio_util::sync::CancellationToken;
+use tracing::{instrument, Instrument};
+use utils::{
+    generation::Generation,
+    id::NodeId,
+    seqwait::{SeqWait, SeqWaitError},
+    sync::gate::Gate,
+};
+
+use crate::{
+    compute_hook::ComputeHook,
+    node::Node,
+    persistence::{split_state::SplitState, Persistence},
+    reconciler::{
+        attached_location_conf, secondary_location_conf, ReconcileError, Reconciler, TargetState,
+    },
+    scheduler::{ScheduleError, Scheduler},
+    service, Sequence,
+};
+
+/// Serialization helper
+fn read_mutex_content<S, T>(v: &std::sync::Mutex<T>, serializer: S) -> Result<S::Ok, S::Error>
+where
+    S: serde::ser::Serializer,
+    T: Clone + std::fmt::Display,
+{
+    serializer.collect_str(&v.lock().unwrap())
+}
+
+/// In-memory state for a particular tenant shard.
+///
+/// This struct implement Serialize for debugging purposes, but is _not_ persisted
+/// itself: see [`crate::persistence`] for the subset of tenant shard state that is persisted.
+#[derive(Serialize)]
+pub(crate) struct TenantState {
+    pub(crate) tenant_shard_id: TenantShardId,
+
+    pub(crate) shard: ShardIdentity,
+
+    // Runtime only: sequence used to coordinate when updating this object while
+    // with background reconcilers may be running.  A reconciler runs to a particular
+    // sequence.
+    pub(crate) sequence: Sequence,
+
+    // Latest generation number: next time we attach, increment this
+    // and use the incremented number when attaching.
+    //
+    // None represents an incompletely onboarded tenant via the [`Service::location_config`]
+    // API, where this tenant may only run in PlacementPolicy::Secondary.
+    pub(crate) generation: Option<Generation>,
+
+    // High level description of how the tenant should be set up.  Provided
+    // externally.
+    pub(crate) policy: PlacementPolicy,
+
+    // Low level description of exactly which pageservers should fulfil
+    // which role.  Generated by `Self::schedule`.
+    pub(crate) intent: IntentState,
+
+    // Low level description of how the tenant is configured on pageservers:
+    // if this does not match `Self::intent` then the tenant needs reconciliation
+    // with `Self::reconcile`.
+    pub(crate) observed: ObservedState,
+
+    // Tenant configuration, passed through opaquely to the pageserver.  Identical
+    // for all shards in a tenant.
+    pub(crate) config: TenantConfig,
+
+    /// If a reconcile task is currently in flight, it may be joined here (it is
+    /// only safe to join if either the result has been received or the reconciler's
+    /// cancellation token has been fired)
+    #[serde(skip)]
+    pub(crate) reconciler: Option<ReconcilerHandle>,
+
+    /// If a tenant is being split, then all shards with that TenantId will have a
+    /// SplitState set, this acts as a guard against other operations such as background
+    /// reconciliation, and timeline creation.
+    pub(crate) splitting: SplitState,
+
+    /// Optionally wait for reconciliation to complete up to a particular
+    /// sequence number.
+    #[serde(skip)]
+    pub(crate) waiter: std::sync::Arc<SeqWait<Sequence, Sequence>>,
+
+    /// Indicates sequence number for which we have encountered an error reconciling.  If
+    /// this advances ahead of [`Self::waiter`] then a reconciliation error has occurred,
+    /// and callers should stop waiting for `waiter` and propagate the error.
+    #[serde(skip)]
+    pub(crate) error_waiter: std::sync::Arc<SeqWait<Sequence, Sequence>>,
+
+    /// The most recent error from a reconcile on this tenant
+    /// TODO: generalize to an array of recent events
+    /// TOOD: use a ArcSwap instead of mutex for faster reads?
+    #[serde(serialize_with = "read_mutex_content")]
+    pub(crate) last_error: std::sync::Arc<std::sync::Mutex<String>>,
+
+    /// If we have a pending compute notification that for some reason we weren't able to send,
+    /// set this to true. If this is set, calls to [`Self::maybe_reconcile`] will run a task to retry
+    /// sending it.  This is the mechanism by which compute notifications are included in the scope
+    /// of state that we publish externally in an eventually consistent way.
+    pub(crate) pending_compute_notification: bool,
+}
+
+#[derive(Default, Clone, Debug, Serialize)]
+pub(crate) struct IntentState {
+    attached: Option<NodeId>,
+    secondary: Vec<NodeId>,
+}
+
+impl IntentState {
+    pub(crate) fn new() -> Self {
+        Self {
+            attached: None,
+            secondary: vec![],
+        }
+    }
+    pub(crate) fn single(scheduler: &mut Scheduler, node_id: Option<NodeId>) -> Self {
+        if let Some(node_id) = node_id {
+            scheduler.node_inc_ref(node_id);
+        }
+        Self {
+            attached: node_id,
+            secondary: vec![],
+        }
+    }
+
+    pub(crate) fn set_attached(&mut self, scheduler: &mut Scheduler, new_attached: Option<NodeId>) {
+        if self.attached != new_attached {
+            if let Some(old_attached) = self.attached.take() {
+                scheduler.node_dec_ref(old_attached);
+            }
+            if let Some(new_attached) = &new_attached {
+                scheduler.node_inc_ref(*new_attached);
+            }
+            self.attached = new_attached;
+        }
+    }
+
+    /// Like set_attached, but the node is from [`Self::secondary`].  This swaps the node from
+    /// secondary to attached while maintaining the scheduler's reference counts.
+    pub(crate) fn promote_attached(
+        &mut self,
+        _scheduler: &mut Scheduler,
+        promote_secondary: NodeId,
+    ) {
+        // If we call this with a node that isn't in secondary, it would cause incorrect
+        // scheduler reference counting, since we assume the node is already referenced as a secondary.
+        debug_assert!(self.secondary.contains(&promote_secondary));
+
+        // TODO: when scheduler starts tracking attached + secondary counts separately, we will
+        // need to call into it here.
+        self.secondary.retain(|n| n != &promote_secondary);
+        self.attached = Some(promote_secondary);
+    }
+
+    pub(crate) fn push_secondary(&mut self, scheduler: &mut Scheduler, new_secondary: NodeId) {
+        debug_assert!(!self.secondary.contains(&new_secondary));
+        scheduler.node_inc_ref(new_secondary);
+        self.secondary.push(new_secondary);
+    }
+
+    /// It is legal to call this with a node that is not currently a secondary: that is a no-op
+    pub(crate) fn remove_secondary(&mut self, scheduler: &mut Scheduler, node_id: NodeId) {
+        let index = self.secondary.iter().position(|n| *n == node_id);
+        if let Some(index) = index {
+            scheduler.node_dec_ref(node_id);
+            self.secondary.remove(index);
+        }
+    }
+
+    pub(crate) fn clear_secondary(&mut self, scheduler: &mut Scheduler) {
+        for secondary in self.secondary.drain(..) {
+            scheduler.node_dec_ref(secondary);
+        }
+    }
+
+    /// Remove the last secondary node from the list of secondaries
+    pub(crate) fn pop_secondary(&mut self, scheduler: &mut Scheduler) {
+        if let Some(node_id) = self.secondary.pop() {
+            scheduler.node_dec_ref(node_id);
+        }
+    }
+
+    pub(crate) fn clear(&mut self, scheduler: &mut Scheduler) {
+        if let Some(old_attached) = self.attached.take() {
+            scheduler.node_dec_ref(old_attached);
+        }
+
+        self.clear_secondary(scheduler);
+    }
+
+    pub(crate) fn all_pageservers(&self) -> Vec<NodeId> {
+        let mut result = Vec::new();
+        if let Some(p) = self.attached {
+            result.push(p)
+        }
+
+        result.extend(self.secondary.iter().copied());
+
+        result
+    }
+
+    pub(crate) fn get_attached(&self) -> &Option<NodeId> {
+        &self.attached
+    }
+
+    pub(crate) fn get_secondary(&self) -> &Vec<NodeId> {
+        &self.secondary
+    }
+
+    /// If the node is in use as the attached location, demote it into
+    /// the list of secondary locations.  This is used when a node goes offline,
+    /// and we want to use a different node for attachment, but not permanently
+    /// forget the location on the offline node.
+    ///
+    /// Returns true if a change was made
+    pub(crate) fn demote_attached(&mut self, node_id: NodeId) -> bool {
+        if self.attached == Some(node_id) {
+            // TODO: when scheduler starts tracking attached + secondary counts separately, we will
+            // need to call into it here.
+            self.attached = None;
+            self.secondary.push(node_id);
+            true
+        } else {
+            false
+        }
+    }
+}
+
+impl Drop for IntentState {
+    fn drop(&mut self) {
+        // Must clear before dropping, to avoid leaving stale refcounts in the Scheduler
+        debug_assert!(self.attached.is_none() && self.secondary.is_empty());
+    }
+}
+
+#[derive(Default, Clone, Serialize)]
+pub(crate) struct ObservedState {
+    pub(crate) locations: HashMap<NodeId, ObservedStateLocation>,
+}
+
+/// Our latest knowledge of how this tenant is configured in the outside world.
+///
+/// Meaning:
+///     * No instance of this type exists for a node: we are certain that we have nothing configured on that
+///       node for this shard.
+///     * Instance exists with conf==None: we *might* have some state on that node, but we don't know
+///       what it is (e.g. we failed partway through configuring it)
+///     * Instance exists with conf==Some: this tells us what we last successfully configured on this node,
+///       and that configuration will still be present unless something external interfered.
+#[derive(Clone, Serialize)]
+pub(crate) struct ObservedStateLocation {
+    /// If None, it means we do not know the status of this shard's location on this node, but
+    /// we know that we might have some state on this node.
+    pub(crate) conf: Option<LocationConfig>,
+}
+pub(crate) struct ReconcilerWaiter {
+    // For observability purposes, remember the ID of the shard we're
+    // waiting for.
+    pub(crate) tenant_shard_id: TenantShardId,
+
+    seq_wait: std::sync::Arc<SeqWait<Sequence, Sequence>>,
+    error_seq_wait: std::sync::Arc<SeqWait<Sequence, Sequence>>,
+    error: std::sync::Arc<std::sync::Mutex<String>>,
+    seq: Sequence,
+}
+
+#[derive(thiserror::Error, Debug)]
+pub enum ReconcileWaitError {
+    #[error("Timeout waiting for shard {0}")]
+    Timeout(TenantShardId),
+    #[error("shutting down")]
+    Shutdown,
+    #[error("Reconcile error on shard {0}: {1}")]
+    Failed(TenantShardId, String),
+}
+
+impl ReconcilerWaiter {
+    pub(crate) async fn wait_timeout(&self, timeout: Duration) -> Result<(), ReconcileWaitError> {
+        tokio::select! {
+            result = self.seq_wait.wait_for_timeout(self.seq, timeout)=> {
+                result.map_err(|e| match e {
+                    SeqWaitError::Timeout => ReconcileWaitError::Timeout(self.tenant_shard_id),
+                    SeqWaitError::Shutdown => ReconcileWaitError::Shutdown
+                })?;
+            },
+            result = self.error_seq_wait.wait_for(self.seq) => {
+                result.map_err(|e| match e {
+                    SeqWaitError::Shutdown => ReconcileWaitError::Shutdown,
+                    SeqWaitError::Timeout => unreachable!()
+                })?;
+
+                return Err(ReconcileWaitError::Failed(self.tenant_shard_id, self.error.lock().unwrap().clone()))
+            }
+        }
+
+        Ok(())
+    }
+}
+
+/// Having spawned a reconciler task, the tenant shard's state will carry enough
+/// information to optionally cancel & await it later.
+pub(crate) struct ReconcilerHandle {
+    sequence: Sequence,
+    handle: JoinHandle<()>,
+    cancel: CancellationToken,
+}
+
+/// When a reconcile task completes, it sends this result object
+/// to be applied to the primary TenantState.
+pub(crate) struct ReconcileResult {
+    pub(crate) sequence: Sequence,
+    /// On errors, `observed` should be treated as an incompleted description
+    /// of state (i.e. any nodes present in the result should override nodes
+    /// present in the parent tenant state, but any unmentioned nodes should
+    /// not be removed from parent tenant state)
+    pub(crate) result: Result<(), ReconcileError>,
+
+    pub(crate) tenant_shard_id: TenantShardId,
+    pub(crate) generation: Option<Generation>,
+    pub(crate) observed: ObservedState,
+
+    /// Set [`TenantState::pending_compute_notification`] from this flag
+    pub(crate) pending_compute_notification: bool,
+}
+
+impl ObservedState {
+    pub(crate) fn new() -> Self {
+        Self {
+            locations: HashMap::new(),
+        }
+    }
+}
+
+impl TenantState {
+    pub(crate) fn new(
+        tenant_shard_id: TenantShardId,
+        shard: ShardIdentity,
+        policy: PlacementPolicy,
+    ) -> Self {
+        Self {
+            tenant_shard_id,
+            policy,
+            intent: IntentState::default(),
+            generation: Some(Generation::new(0)),
+            shard,
+            observed: ObservedState::default(),
+            config: TenantConfig::default(),
+            reconciler: None,
+            splitting: SplitState::Idle,
+            sequence: Sequence(1),
+            waiter: Arc::new(SeqWait::new(Sequence(0))),
+            error_waiter: Arc::new(SeqWait::new(Sequence(0))),
+            last_error: Arc::default(),
+            pending_compute_notification: false,
+        }
+    }
+
+    /// For use on startup when learning state from pageservers: generate my [`IntentState`] from my
+    /// [`ObservedState`], even if it violates my [`PlacementPolicy`].  Call [`Self::schedule`] next,
+    /// to get an intent state that complies with placement policy.  The overall goal is to do scheduling
+    /// in a way that makes use of any configured locations that already exist in the outside world.
+    pub(crate) fn intent_from_observed(&mut self, scheduler: &mut Scheduler) {
+        // Choose an attached location by filtering observed locations, and then sorting to get the highest
+        // generation
+        let mut attached_locs = self
+            .observed
+            .locations
+            .iter()
+            .filter_map(|(node_id, l)| {
+                if let Some(conf) = &l.conf {
+                    if conf.mode == LocationConfigMode::AttachedMulti
+                        || conf.mode == LocationConfigMode::AttachedSingle
+                        || conf.mode == LocationConfigMode::AttachedStale
+                    {
+                        Some((node_id, conf.generation))
+                    } else {
+                        None
+                    }
+                } else {
+                    None
+                }
+            })
+            .collect::<Vec<_>>();
+
+        attached_locs.sort_by_key(|i| i.1);
+        if let Some((node_id, _gen)) = attached_locs.into_iter().last() {
+            self.intent.set_attached(scheduler, Some(*node_id));
+        }
+
+        // All remaining observed locations generate secondary intents.  This includes None
+        // observations, as these may well have some local content on disk that is usable (this
+        // is an edge case that might occur if we restarted during a migration or other change)
+        //
+        // We may leave intent.attached empty if we didn't find any attached locations: [`Self::schedule`]
+        // will take care of promoting one of these secondaries to be attached.
+        self.observed.locations.keys().for_each(|node_id| {
+            if Some(*node_id) != self.intent.attached {
+                self.intent.push_secondary(scheduler, *node_id);
+            }
+        });
+    }
+
+    /// Part of [`Self::schedule`] that is used to choose exactly one node to act as the
+    /// attached pageserver for a shard.
+    ///
+    /// Returns whether we modified it, and the NodeId selected.
+    fn schedule_attached(
+        &mut self,
+        scheduler: &mut Scheduler,
+    ) -> Result<(bool, NodeId), ScheduleError> {
+        // No work to do if we already have an attached tenant
+        if let Some(node_id) = self.intent.attached {
+            return Ok((false, node_id));
+        }
+
+        if let Some(promote_secondary) = scheduler.node_preferred(&self.intent.secondary) {
+            // Promote a secondary
+            tracing::debug!("Promoted secondary {} to attached", promote_secondary);
+            self.intent.promote_attached(scheduler, promote_secondary);
+            Ok((true, promote_secondary))
+        } else {
+            // Pick a fresh node: either we had no secondaries or none were schedulable
+            let node_id = scheduler.schedule_shard(&self.intent.secondary)?;
+            tracing::debug!("Selected {} as attached", node_id);
+            self.intent.set_attached(scheduler, Some(node_id));
+            Ok((true, node_id))
+        }
+    }
+
+    pub(crate) fn schedule(&mut self, scheduler: &mut Scheduler) -> Result<(), ScheduleError> {
+        // TODO: before scheduling new nodes, check if any existing content in
+        // self.intent refers to pageservers that are offline, and pick other
+        // pageservers if so.
+
+        // TODO: respect the splitting bit on tenants: if they are currently splitting then we may not
+        // change their attach location.
+
+        // Build the set of pageservers already in use by this tenant, to avoid scheduling
+        // more work on the same pageservers we're already using.
+        let mut modified = false;
+
+        // Add/remove nodes to fulfil policy
+        use PlacementPolicy::*;
+        match self.policy {
+            Attached(secondary_count) => {
+                let retain_secondaries = if self.intent.attached.is_none()
+                    && scheduler.node_preferred(&self.intent.secondary).is_some()
+                {
+                    // If we have no attached, and one of the secondaries is elegible to be promoted, retain
+                    // one more secondary than we usually would, as one of them will become attached futher down this function.
+                    secondary_count + 1
+                } else {
+                    secondary_count
+                };
+
+                while self.intent.secondary.len() > retain_secondaries {
+                    // We have no particular preference for one secondary location over another: just
+                    // arbitrarily drop from the end
+                    self.intent.pop_secondary(scheduler);
+                    modified = true;
+                }
+
+                // Should have exactly one attached, and N secondaries
+                let (modified_attached, attached_node_id) = self.schedule_attached(scheduler)?;
+                modified |= modified_attached;
+
+                let mut used_pageservers = vec![attached_node_id];
+                while self.intent.secondary.len() < secondary_count {
+                    let node_id = scheduler.schedule_shard(&used_pageservers)?;
+                    self.intent.push_secondary(scheduler, node_id);
+                    used_pageservers.push(node_id);
+                    modified = true;
+                }
+            }
+            Secondary => {
+                if let Some(node_id) = self.intent.get_attached() {
+                    // Populate secondary by demoting the attached node
+                    self.intent.demote_attached(*node_id);
+                    modified = true;
+                } else if self.intent.secondary.is_empty() {
+                    // Populate secondary by scheduling a fresh node
+                    let node_id = scheduler.schedule_shard(&[])?;
+                    self.intent.push_secondary(scheduler, node_id);
+                    modified = true;
+                }
+                while self.intent.secondary.len() > 1 {
+                    // We have no particular preference for one secondary location over another: just
+                    // arbitrarily drop from the end
+                    self.intent.pop_secondary(scheduler);
+                    modified = true;
+                }
+            }
+            Detached => {
+                // Never add locations in this mode
+                if self.intent.get_attached().is_some() || !self.intent.get_secondary().is_empty() {
+                    self.intent.clear(scheduler);
+                    modified = true;
+                }
+            }
+        }
+
+        if modified {
+            self.sequence.0 += 1;
+        }
+
+        Ok(())
+    }
+
+    /// Query whether the tenant's observed state for attached node matches its intent state, and if so,
+    /// yield the node ID.  This is appropriate for emitting compute hook notifications: we are checking that
+    /// the node in question is not only where we intend to attach, but that the tenant is indeed already attached there.
+    ///
+    /// Reconciliation may still be needed for other aspects of state such as secondaries (see [`Self::dirty`]): this
+    /// funciton should not be used to decide whether to reconcile.
+    pub(crate) fn stably_attached(&self) -> Option<NodeId> {
+        if let Some(attach_intent) = self.intent.attached {
+            match self.observed.locations.get(&attach_intent) {
+                Some(loc) => match &loc.conf {
+                    Some(conf) => match conf.mode {
+                        LocationConfigMode::AttachedMulti
+                        | LocationConfigMode::AttachedSingle
+                        | LocationConfigMode::AttachedStale => {
+                            // Our intent and observed state agree that this node is in an attached state.
+                            Some(attach_intent)
+                        }
+                        // Our observed config is not an attached state
+                        _ => None,
+                    },
+                    // Our observed state is None, i.e. in flux
+                    None => None,
+                },
+                // We have no observed state for this node
+                None => None,
+            }
+        } else {
+            // Our intent is not to attach
+            None
+        }
+    }
+
+    fn dirty(&self, nodes: &Arc<HashMap<NodeId, Node>>) -> bool {
+        let mut dirty_nodes = HashSet::new();
+
+        if let Some(node_id) = self.intent.attached {
+            // Maybe panic: it is a severe bug if we try to attach while generation is null.
+            let generation = self
+                .generation
+                .expect("Attempted to enter attached state without a generation");
+
+            let wanted_conf = attached_location_conf(
+                generation,
+                &self.shard,
+                &self.config,
+                !self.intent.secondary.is_empty(),
+            );
+            match self.observed.locations.get(&node_id) {
+                Some(conf) if conf.conf.as_ref() == Some(&wanted_conf) => {}
+                Some(_) | None => {
+                    dirty_nodes.insert(node_id);
+                }
+            }
+        }
+
+        for node_id in &self.intent.secondary {
+            let wanted_conf = secondary_location_conf(&self.shard, &self.config);
+            match self.observed.locations.get(node_id) {
+                Some(conf) if conf.conf.as_ref() == Some(&wanted_conf) => {}
+                Some(_) | None => {
+                    dirty_nodes.insert(*node_id);
+                }
+            }
+        }
+
+        for node_id in self.observed.locations.keys() {
+            if self.intent.attached != Some(*node_id) && !self.intent.secondary.contains(node_id) {
+                // We have observed state that isn't part of our intent: need to clean it up.
+                dirty_nodes.insert(*node_id);
+            }
+        }
+
+        dirty_nodes.retain(|node_id| {
+            nodes
+                .get(node_id)
+                .map(|n| n.is_available())
+                .unwrap_or(false)
+        });
+
+        !dirty_nodes.is_empty()
+    }
+
+    #[allow(clippy::too_many_arguments)]
+    #[instrument(skip_all, fields(tenant_id=%self.tenant_shard_id.tenant_id, shard_id=%self.tenant_shard_id.shard_slug()))]
+    pub(crate) fn maybe_reconcile(
+        &mut self,
+        result_tx: &tokio::sync::mpsc::UnboundedSender<ReconcileResult>,
+        pageservers: &Arc<HashMap<NodeId, Node>>,
+        compute_hook: &Arc<ComputeHook>,
+        service_config: &service::Config,
+        persistence: &Arc<Persistence>,
+        gate: &Gate,
+        cancel: &CancellationToken,
+    ) -> Option<ReconcilerWaiter> {
+        // If there are any ambiguous observed states, and the nodes they refer to are available,
+        // we should reconcile to clean them up.
+        let mut dirty_observed = false;
+        for (node_id, observed_loc) in &self.observed.locations {
+            let node = pageservers
+                .get(node_id)
+                .expect("Nodes may not be removed while referenced");
+            if observed_loc.conf.is_none() && node.is_available() {
+                dirty_observed = true;
+                break;
+            }
+        }
+
+        let active_nodes_dirty = self.dirty(pageservers);
+
+        // Even if there is no pageserver work to be done, if we have a pending notification to computes,
+        // wake up a reconciler to send it.
+        let do_reconcile =
+            active_nodes_dirty || dirty_observed || self.pending_compute_notification;
+
+        if !do_reconcile {
+            tracing::info!("Not dirty, no reconciliation needed.");
+            return None;
+        }
+
+        // If we are currently splitting, then never start a reconciler task: the splitting logic
+        // requires that shards are not interfered with while it runs. Do this check here rather than
+        // up top, so that we only log this message if we would otherwise have done a reconciliation.
+        if !matches!(self.splitting, SplitState::Idle) {
+            tracing::info!("Refusing to reconcile, splitting in progress");
+            return None;
+        }
+
+        // Reconcile already in flight for the current sequence?
+        if let Some(handle) = &self.reconciler {
+            if handle.sequence == self.sequence {
+                tracing::info!(
+                    "Reconciliation already in progress for sequence {:?}",
+                    self.sequence,
+                );
+                return Some(ReconcilerWaiter {
+                    tenant_shard_id: self.tenant_shard_id,
+                    seq_wait: self.waiter.clone(),
+                    error_seq_wait: self.error_waiter.clone(),
+                    error: self.last_error.clone(),
+                    seq: self.sequence,
+                });
+            }
+        }
+
+        // Build list of nodes from which the reconciler should detach
+        let mut detach = Vec::new();
+        for node_id in self.observed.locations.keys() {
+            if self.intent.get_attached() != &Some(*node_id)
+                && !self.intent.secondary.contains(node_id)
+            {
+                detach.push(
+                    pageservers
+                        .get(node_id)
+                        .expect("Intent references non-existent pageserver")
+                        .clone(),
+                )
+            }
+        }
+
+        // Reconcile in flight for a stale sequence?  Our sequence's task will wait for it before
+        // doing our sequence's work.
+        let old_handle = self.reconciler.take();
+
+        let Ok(gate_guard) = gate.enter() else {
+            // Shutting down, don't start a reconciler
+            return None;
+        };
+
+        // Advance the sequence before spawning a reconciler, so that sequence waiters
+        // can distinguish between before+after the reconcile completes.
+        self.sequence = self.sequence.next();
+
+        let reconciler_cancel = cancel.child_token();
+        let reconciler_intent = TargetState::from_intent(pageservers, &self.intent);
+        let mut reconciler = Reconciler {
+            tenant_shard_id: self.tenant_shard_id,
+            shard: self.shard,
+            generation: self.generation,
+            intent: reconciler_intent,
+            detach,
+            config: self.config.clone(),
+            observed: self.observed.clone(),
+            compute_hook: compute_hook.clone(),
+            service_config: service_config.clone(),
+            _gate_guard: gate_guard,
+            cancel: reconciler_cancel.clone(),
+            persistence: persistence.clone(),
+            compute_notify_failure: false,
+        };
+
+        let reconcile_seq = self.sequence;
+
+        tracing::info!(seq=%reconcile_seq, "Spawning Reconciler for sequence {}", self.sequence);
+        let must_notify = self.pending_compute_notification;
+        let reconciler_span = tracing::info_span!(parent: None, "reconciler", seq=%reconcile_seq,
+                                                        tenant_id=%reconciler.tenant_shard_id.tenant_id,
+                                                        shard_id=%reconciler.tenant_shard_id.shard_slug());
+        metrics::METRICS_REGISTRY
+            .metrics_group
+            .storage_controller_reconcile_spawn
+            .inc();
+        let result_tx = result_tx.clone();
+        let join_handle = tokio::task::spawn(
+            async move {
+                // Wait for any previous reconcile task to complete before we start
+                if let Some(old_handle) = old_handle {
+                    old_handle.cancel.cancel();
+                    if let Err(e) = old_handle.handle.await {
+                        // We can't do much with this other than log it: the task is done, so
+                        // we may proceed with our work.
+                        tracing::error!("Unexpected join error waiting for reconcile task: {e}");
+                    }
+                }
+
+                // Early check for cancellation before doing any work
+                // TODO: wrap all remote API operations in cancellation check
+                // as well.
+                if reconciler.cancel.is_cancelled() {
+                    metrics::METRICS_REGISTRY
+                        .metrics_group
+                        .storage_controller_reconcile_complete
+                        .inc(ReconcileCompleteLabelGroup {
+                            status: ReconcileOutcome::Cancel,
+                        });
+                    return;
+                }
+
+                // Attempt to make observed state match intent state
+                let result = reconciler.reconcile().await;
+
+                // If we know we had a pending compute notification from some previous action, send a notification irrespective
+                // of whether the above reconcile() did any work
+                if result.is_ok() && must_notify {
+                    // If this fails we will send the need to retry in [`ReconcileResult::pending_compute_notification`]
+                    reconciler.compute_notify().await.ok();
+                }
+
+                // Update result counter
+                let outcome_label = match &result {
+                    Ok(_) => ReconcileOutcome::Success,
+                    Err(ReconcileError::Cancel) => ReconcileOutcome::Cancel,
+                    Err(_) => ReconcileOutcome::Error,
+                };
+
+                metrics::METRICS_REGISTRY
+                    .metrics_group
+                    .storage_controller_reconcile_complete
+                    .inc(ReconcileCompleteLabelGroup {
+                        status: outcome_label,
+                    });
+
+                result_tx
+                    .send(ReconcileResult {
+                        sequence: reconcile_seq,
+                        result,
+                        tenant_shard_id: reconciler.tenant_shard_id,
+                        generation: reconciler.generation,
+                        observed: reconciler.observed,
+                        pending_compute_notification: reconciler.compute_notify_failure,
+                    })
+                    .ok();
+            }
+            .instrument(reconciler_span),
+        );
+
+        self.reconciler = Some(ReconcilerHandle {
+            sequence: self.sequence,
+            handle: join_handle,
+            cancel: reconciler_cancel,
+        });
+
+        Some(ReconcilerWaiter {
+            tenant_shard_id: self.tenant_shard_id,
+            seq_wait: self.waiter.clone(),
+            error_seq_wait: self.error_waiter.clone(),
+            error: self.last_error.clone(),
+            seq: self.sequence,
+        })
+    }
+
+    /// Called when a ReconcileResult has been emitted and the service is updating
+    /// our state: if the result is from a sequence >= my ReconcileHandle, then drop
+    /// the handle to indicate there is no longer a reconciliation in progress.
+    pub(crate) fn reconcile_complete(&mut self, sequence: Sequence) {
+        if let Some(reconcile_handle) = &self.reconciler {
+            if reconcile_handle.sequence <= sequence {
+                self.reconciler = None;
+            }
+        }
+    }
+
+    // If we had any state at all referring to this node ID, drop it.  Does not
+    // attempt to reschedule.
+    pub(crate) fn deref_node(&mut self, node_id: NodeId) {
+        if self.intent.attached == Some(node_id) {
+            self.intent.attached = None;
+        }
+
+        self.intent.secondary.retain(|n| n != &node_id);
+
+        self.observed.locations.remove(&node_id);
+
+        debug_assert!(!self.intent.all_pageservers().contains(&node_id));
+    }
+
+    pub(crate) fn to_persistent(&self) -> TenantShardPersistence {
+        TenantShardPersistence {
+            tenant_id: self.tenant_shard_id.tenant_id.to_string(),
+            shard_number: self.tenant_shard_id.shard_number.0 as i32,
+            shard_count: self.tenant_shard_id.shard_count.literal() as i32,
+            shard_stripe_size: self.shard.stripe_size.0 as i32,
+            generation: self.generation.map(|g| g.into().unwrap_or(0) as i32),
+            generation_pageserver: self.intent.get_attached().map(|n| n.0 as i64),
+            placement_policy: serde_json::to_string(&self.policy).unwrap(),
+            config: serde_json::to_string(&self.config).unwrap(),
+            splitting: SplitState::default(),
+        }
+    }
+}
+
+#[cfg(test)]
+pub(crate) mod tests {
+    use pageserver_api::{
+        controller_api::NodeAvailability,
+        shard::{ShardCount, ShardNumber},
+    };
+    use utils::id::TenantId;
+
+    use crate::scheduler::test_utils::make_test_nodes;
+
+    use super::*;
+
+    fn make_test_tenant_shard(policy: PlacementPolicy) -> TenantState {
+        let tenant_id = TenantId::generate();
+        let shard_number = ShardNumber(0);
+        let shard_count = ShardCount::new(1);
+
+        let tenant_shard_id = TenantShardId {
+            tenant_id,
+            shard_number,
+            shard_count,
+        };
+        TenantState::new(
+            tenant_shard_id,
+            ShardIdentity::new(
+                shard_number,
+                shard_count,
+                pageserver_api::shard::ShardStripeSize(32768),
+            )
+            .unwrap(),
+            policy,
+        )
+    }
+
+    /// Test the scheduling behaviors used when a tenant configured for HA is subject
+    /// to nodes being marked offline.
+    #[test]
+    fn tenant_ha_scheduling() -> anyhow::Result<()> {
+        // Start with three nodes.  Our tenant will only use two.  The third one is
+        // expected to remain unused.
+        let mut nodes = make_test_nodes(3);
+
+        let mut scheduler = Scheduler::new(nodes.values());
+
+        let mut tenant_state = make_test_tenant_shard(PlacementPolicy::Attached(1));
+        tenant_state
+            .schedule(&mut scheduler)
+            .expect("we have enough nodes, scheduling should work");
+
+        // Expect to initially be schedule on to different nodes
+        assert_eq!(tenant_state.intent.secondary.len(), 1);
+        assert!(tenant_state.intent.attached.is_some());
+
+        let attached_node_id = tenant_state.intent.attached.unwrap();
+        let secondary_node_id = *tenant_state.intent.secondary.iter().last().unwrap();
+        assert_ne!(attached_node_id, secondary_node_id);
+
+        // Notifying the attached node is offline should demote it to a secondary
+        let changed = tenant_state.intent.demote_attached(attached_node_id);
+        assert!(changed);
+        assert!(tenant_state.intent.attached.is_none());
+        assert_eq!(tenant_state.intent.secondary.len(), 2);
+
+        // Update the scheduler state to indicate the node is offline
+        nodes
+            .get_mut(&attached_node_id)
+            .unwrap()
+            .set_availability(NodeAvailability::Offline);
+        scheduler.node_upsert(nodes.get(&attached_node_id).unwrap());
+
+        // Scheduling the node should promote the still-available secondary node to attached
+        tenant_state
+            .schedule(&mut scheduler)
+            .expect("active nodes are available");
+        assert_eq!(tenant_state.intent.attached.unwrap(), secondary_node_id);
+
+        // The original attached node should have been retained as a secondary
+        assert_eq!(
+            *tenant_state.intent.secondary.iter().last().unwrap(),
+            attached_node_id
+        );
+
+        tenant_state.intent.clear(&mut scheduler);
+
+        Ok(())
+    }
+
+    #[test]
+    fn intent_from_observed() -> anyhow::Result<()> {
+        let nodes = make_test_nodes(3);
+        let mut scheduler = Scheduler::new(nodes.values());
+
+        let mut tenant_state = make_test_tenant_shard(PlacementPolicy::Attached(1));
+
+        tenant_state.observed.locations.insert(
+            NodeId(3),
+            ObservedStateLocation {
+                conf: Some(LocationConfig {
+                    mode: LocationConfigMode::AttachedMulti,
+                    generation: Some(2),
+                    secondary_conf: None,
+                    shard_number: tenant_state.shard.number.0,
+                    shard_count: tenant_state.shard.count.literal(),
+                    shard_stripe_size: tenant_state.shard.stripe_size.0,
+                    tenant_conf: TenantConfig::default(),
+                }),
+            },
+        );
+
+        tenant_state.observed.locations.insert(
+            NodeId(2),
+            ObservedStateLocation {
+                conf: Some(LocationConfig {
+                    mode: LocationConfigMode::AttachedStale,
+                    generation: Some(1),
+                    secondary_conf: None,
+                    shard_number: tenant_state.shard.number.0,
+                    shard_count: tenant_state.shard.count.literal(),
+                    shard_stripe_size: tenant_state.shard.stripe_size.0,
+                    tenant_conf: TenantConfig::default(),
+                }),
+            },
+        );
+
+        tenant_state.intent_from_observed(&mut scheduler);
+
+        // The highest generationed attached location gets used as attached
+        assert_eq!(tenant_state.intent.attached, Some(NodeId(3)));
+        // Other locations get used as secondary
+        assert_eq!(tenant_state.intent.secondary, vec![NodeId(2)]);
+
+        scheduler.consistency_check(nodes.values(), [&tenant_state].into_iter())?;
+
+        tenant_state.intent.clear(&mut scheduler);
+        Ok(())
+    }
+}