feat(pageserver): use vectored_get in collect_keyspace

Signed-off-by: Alex Chi Z <chi@neon.tech>

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

@@ -17,6 +17,31 @@ inputs:
   compute_units:
     description: '[Min, Max] compute units'
     default: '[1, 1]'
+  # settings below only needed if you want the project to be sharded from the beginning
+  shard_split_project:
+    description: 'by default new projects are not shard-split, specify true to shard-split'
+    required: false
+    default: 'false'
+  admin_api_key:
+    description: 'Admin API Key needed for shard-splitting. Must be specified if shard_split_project is true'
+    required: false
+  shard_count:
+    description: 'Number of shards to split the project into, only applies if shard_split_project is true'
+    required: false
+    default: '8'
+  stripe_size:
+    description: 'Stripe size, optional, in 8kiB pages.  e.g. set 2048 for 16MB stripes. Default is 128 MiB, only applies if shard_split_project is true'
+    required: false
+    default: '32768'
+  psql_path:
+    description: 'Path to psql binary - it is caller responsibility to provision the psql binary'
+    required: false
+    default: '/tmp/neon/pg_install/v16/bin/psql'
+  libpq_lib_path:
+    description: 'Path to directory containing libpq library - it is caller responsibility to provision the libpq library'
+    required: false
+    default: '/tmp/neon/pg_install/v16/lib'
+  
 
 outputs:
   dsn:
@@ -63,6 +88,23 @@ runs:
         echo "project_id=${project_id}" >> $GITHUB_OUTPUT
 
         echo "Project ${project_id} has been created"
+
+        if [ "${SHARD_SPLIT_PROJECT}" = "true" ]; then
+          # determine tenant ID
+          TENANT_ID=`${PSQL} ${dsn} -t -A -c "SHOW neon.tenant_id"`
+          
+          echo "Splitting project ${project_id} with tenant_id ${TENANT_ID} into $((SHARD_COUNT)) shards with stripe size $((STRIPE_SIZE))"
+
+          echo "Sending PUT request to https://${API_HOST}/regions/${REGION_ID}/api/v1/admin/storage/proxy/control/v1/tenant/${TENANT_ID}/shard_split"
+          echo "with body {\"new_shard_count\": $((SHARD_COUNT)), \"new_stripe_size\": $((STRIPE_SIZE))}"
+          
+          # we need an ADMIN API KEY to invoke storage controller API for shard splitting (bash -u above checks that the variable is set)
+          curl -X PUT \
+            "https://${API_HOST}/regions/${REGION_ID}/api/v1/admin/storage/proxy/control/v1/tenant/${TENANT_ID}/shard_split" \
+            -H "Accept: application/json" -H "Content-Type: application/json" -H "Authorization: Bearer ${ADMIN_API_KEY}" \
+            -d "{\"new_shard_count\": $SHARD_COUNT, \"new_stripe_size\": $STRIPE_SIZE}"
+        fi
+
       env:
         API_HOST: ${{ inputs.api_host }}
         API_KEY: ${{ inputs.api_key }}
@@ -70,3 +112,9 @@ runs:
         POSTGRES_VERSION: ${{ inputs.postgres_version }}
         MIN_CU: ${{ fromJSON(inputs.compute_units)[0] }}
         MAX_CU: ${{ fromJSON(inputs.compute_units)[1] }}
+        SHARD_SPLIT_PROJECT: ${{ inputs.shard_split_project }}
+        ADMIN_API_KEY: ${{ inputs.admin_api_key }}
+        SHARD_COUNT: ${{ inputs.shard_count }}
+        STRIPE_SIZE: ${{ inputs.stripe_size }}
+        PSQL: ${{ inputs.psql_path }}
+        LD_LIBRARY_PATH: ${{ inputs.libpq_lib_path }}

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

@@ -158,8 +158,6 @@ jobs:
 
       - 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
@@ -217,8 +215,6 @@ jobs:
         env:
           NEXTEST_RETRIES: 3
         run: |
-          PQ_LIB_DIR=$(pwd)/pg_install/v16/lib
-          export PQ_LIB_DIR
           LD_LIBRARY_PATH=$(pwd)/pg_install/v17/lib
           export LD_LIBRARY_PATH
 
@@ -229,8 +225,13 @@ jobs:
           ${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
-            NEON_PAGESERVER_UNIT_TEST_VIRTUAL_FILE_IOENGINE=$io_engine ${cov_prefix} cargo nextest run $CARGO_FLAGS $CARGO_FEATURES  -E 'package(pageserver)'
+          for get_vectored_concurrent_io in sequential sidecar-task; do
+            for io_engine in std-fs tokio-epoll-uring ; do
+              NEON_PAGESERVER_UNIT_TEST_GET_VECTORED_CONCURRENT_IO=$get_vectored_concurrent_io \
+                NEON_PAGESERVER_UNIT_TEST_VIRTUAL_FILE_IOENGINE=$io_engine \
+                ${cov_prefix} \
+                cargo nextest run $CARGO_FLAGS $CARGO_FEATURES  -E 'package(pageserver)'
+            done
           done
 
           # Run separate tests for real S3
@@ -314,6 +315,7 @@ jobs:
           CHECK_ONDISK_DATA_COMPATIBILITY: nonempty
           BUILD_TAG: ${{ inputs.build-tag }}
           PAGESERVER_VIRTUAL_FILE_IO_ENGINE: tokio-epoll-uring
+          PAGESERVER_GET_VECTORED_CONCURRENT_IO: sidecar-task
           USE_LFC: ${{ matrix.lfc_state == 'with-lfc' && 'true' || 'false' }}
 
       # Temporary disable this step until we figure out why it's so flaky

.github/workflows/_check-codestyle-rust.yml

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

.github/workflows/build-macos.yml

@@ -235,7 +235,7 @@ jobs:
           echo 'CPPFLAGS=-I/usr/local/opt/openssl@3/include' >> $GITHUB_ENV
 
       - name: Run cargo build (only for v17)
-        run: PQ_LIB_DIR=$(pwd)/pg_install/v17/lib cargo build --all --release -j$(sysctl -n hw.ncpu)
+        run: cargo build --all --release -j$(sysctl -n hw.ncpu)
 
       - name: Check that no warnings are produced (only for v17)
         run: ./run_clippy.sh

.github/workflows/build_and_test.yml

@@ -164,77 +164,11 @@ jobs:
 
   check-codestyle-rust:
     needs: [ check-permissions, build-build-tools-image ]
-    strategy:
-      matrix:
-        arch: [ x64, arm64 ]
-    runs-on: ${{ fromJson(format('["self-hosted", "{0}"]', matrix.arch == 'arm64' && 'small-arm64' || 'small')) }}
-
-    container:
-      image: ${{ needs.build-build-tools-image.outputs.image }}-bookworm
-      credentials:
-        username: ${{ secrets.NEON_DOCKERHUB_USERNAME }}
-        password: ${{ secrets.NEON_DOCKERHUB_PASSWORD }}
-      options: --init
-
-    steps:
-      - name: Checkout
-        uses: actions/checkout@v4
-        with:
-          submodules: true
-
-      - name: Cache cargo deps
-        uses: actions/cache@v4
-        with:
-          path: |
-            ~/.cargo/registry
-            !~/.cargo/registry/src
-            ~/.cargo/git
-            target
-          key: v1-${{ runner.os }}-${{ runner.arch }}-cargo-${{ hashFiles('./Cargo.lock') }}-${{ hashFiles('./rust-toolchain.toml') }}-rust
-
-      # Some of our rust modules use FFI and need those to be checked
-      - name: Get postgres headers
-        run: make postgres-headers -j$(nproc)
-
-      # cargo hack runs the given cargo subcommand (clippy in this case) for all feature combinations.
-      # This will catch compiler & clippy warnings in all feature combinations.
-      # TODO: use cargo hack for build and test as well, but, that's quite expensive.
-      # NB: keep clippy args in sync with ./run_clippy.sh
-      #
-      # The only difference between "clippy --debug" and "clippy --release" is that in --release mode,
-      # #[cfg(debug_assertions)] blocks are not built. It's not worth building everything for second
-      # time just for that, so skip "clippy --release".
-      - run: |
-          CLIPPY_COMMON_ARGS="$( source .neon_clippy_args; echo "$CLIPPY_COMMON_ARGS")"
-          if [ "$CLIPPY_COMMON_ARGS" = "" ]; then
-            echo "No clippy args found in .neon_clippy_args"
-            exit 1
-          fi
-          echo "CLIPPY_COMMON_ARGS=${CLIPPY_COMMON_ARGS}" >> $GITHUB_ENV
-      - name: Run cargo clippy (debug)
-        run: cargo hack --features default --ignore-unknown-features --feature-powerset clippy $CLIPPY_COMMON_ARGS
-
-      - name: Check documentation generation
-        run: cargo doc --workspace --no-deps --document-private-items
-        env:
-            RUSTDOCFLAGS: "-Dwarnings -Arustdoc::private_intra_doc_links"
-
-      # Use `${{ !cancelled() }}` to run quck tests after the longer clippy run
-      - name: Check formatting
-        if: ${{ !cancelled() }}
-        run: cargo fmt --all -- --check
-
-      # https://github.com/facebookincubator/cargo-guppy/tree/bec4e0eb29dcd1faac70b1b5360267fc02bf830e/tools/cargo-hakari#2-keep-the-workspace-hack-up-to-date-in-ci
-      - name: Check rust dependencies
-        if: ${{ !cancelled() }}
-        run: |
-          cargo hakari generate --diff  # workspace-hack Cargo.toml is up-to-date
-          cargo hakari manage-deps --dry-run  # all workspace crates depend on workspace-hack
-
-      # https://github.com/EmbarkStudios/cargo-deny
-      - name: Check rust licenses/bans/advisories/sources
-        if: ${{ !cancelled() }}
-        run: cargo deny check --hide-inclusion-graph
+    uses: ./.github/workflows/_check-codestyle-rust.yml
+    with:
+      build-tools-image: ${{ needs.build-build-tools-image.outputs.image }}-bookworm
+      archs: '["x64", "arm64"]'
+    secrets: inherit
 
   build-and-test-locally:
     needs: [ tag, build-build-tools-image ]
@@ -886,11 +820,11 @@ jobs:
       - name: Print logs and clean up
         if: always()
         run: |
-          docker compose -f ./docker-compose/docker-compose.yml logs || 0
-          docker compose -f ./docker-compose/docker-compose.yml down
+          docker compose --profile test-extensions -f ./docker-compose/docker-compose.yml logs || true
+          docker compose --profile test-extensions -f ./docker-compose/docker-compose.yml down
 
   promote-images-dev:
-    needs: [ check-permissions, tag, vm-compute-node-image ]
+    needs: [ check-permissions, tag, vm-compute-node-image, neon-image ]
     runs-on: ubuntu-22.04
 
     permissions:
@@ -925,7 +859,7 @@ jobs:
           done
 
   promote-images-prod:
-    needs: [ check-permissions, tag, test-images, vm-compute-node-image ]
+    needs: [ check-permissions, tag, test-images, promote-images-dev ]
     runs-on: ubuntu-22.04
     if: github.ref_name == 'main' || github.ref_name == 'release' || github.ref_name == 'release-proxy' || github.ref_name == 'release-compute'
 
@@ -1144,12 +1078,6 @@ jobs:
               console.log(`Tag ${tag} created successfully.`);
             }
 
-            // TODO: check how GitHub releases looks for proxy/compute releases and enable them if they're ok
-            if (context.ref !== 'refs/heads/release') {
-              console.log(`GitHub release skipped for ${context.ref}.`);
-              return;
-            }
-
             try {
               const existingRelease = await github.rest.repos.getReleaseByTag({
                 owner: context.repo.owner,
@@ -1168,7 +1096,8 @@ jobs:
                 owner: context.repo.owner,
                 repo: context.repo.repo,
                 tag_name: tag,
-                generate_release_notes: true,
+                // TODO: Automate release notes properly
+                generate_release_notes: false,
               });
               console.log(`Release for tag ${tag} created successfully.`);
             }

.github/workflows/ingest_benchmark.yml

@@ -28,7 +28,24 @@ jobs:
     strategy:
       fail-fast: false # allow other variants to continue even if one fails
       matrix:
-        target_project: [new_empty_project, large_existing_project]
+        include:
+          - target_project: new_empty_project_stripe_size_2048 
+            stripe_size: 2048 # 16 MiB
+            postgres_version: 16
+          - target_project: new_empty_project_stripe_size_32768
+            stripe_size: 32768 # 256 MiB # note that this is different from null because using null will shard_split the project only if it reaches the threshold
+                               # while here it is sharded from the beginning with a shard size of 256 MiB
+            postgres_version: 16
+          - target_project: new_empty_project
+            stripe_size: null # run with neon defaults which will shard split only when reaching the threshold
+            postgres_version: 16
+          - target_project: new_empty_project
+            stripe_size: null # run with neon defaults which will shard split only when reaching the threshold
+            postgres_version: 17
+          - target_project: large_existing_project
+            stripe_size: null # cannot re-shared or choose different stripe size for existing, already sharded project
+            postgres_version: 16
+      max-parallel: 1 # we want to run each stripe size sequentially to be able to compare the results
     permissions:
       contents: write
       statuses: write
@@ -67,17 +84,21 @@ jobs:
         aws-oicd-role-arn: ${{ vars.DEV_AWS_OIDC_ROLE_ARN }}
 
     - name: Create Neon Project
-      if: ${{ matrix.target_project == 'new_empty_project' }}
+      if: ${{ startsWith(matrix.target_project, 'new_empty_project') }}
       id: create-neon-project-ingest-target
       uses: ./.github/actions/neon-project-create
       with:
         region_id: aws-us-east-2
-        postgres_version: 16
+        postgres_version: ${{ matrix.postgres_version }}
         compute_units: '[7, 7]' # we want to test large compute here to avoid compute-side bottleneck
         api_key: ${{ secrets.NEON_STAGING_API_KEY }}
+        shard_split_project: ${{ matrix.stripe_size != null && 'true' || 'false' }}
+        admin_api_key: ${{ secrets.NEON_STAGING_ADMIN_API_KEY }} 
+        shard_count: 8
+        stripe_size: ${{ matrix.stripe_size }}
 
     - name: Initialize Neon project
-      if: ${{ matrix.target_project == 'new_empty_project' }}
+      if: ${{ startsWith(matrix.target_project, 'new_empty_project') }}
       env:
           BENCHMARK_INGEST_TARGET_CONNSTR: ${{ steps.create-neon-project-ingest-target.outputs.dsn }}
           NEW_PROJECT_ID: ${{ steps.create-neon-project-ingest-target.outputs.project_id }}
@@ -130,7 +151,7 @@ jobs:
         test_selection: performance/test_perf_ingest_using_pgcopydb.py
         run_in_parallel: false
         extra_params: -s -m remote_cluster --timeout 86400 -k test_ingest_performance_using_pgcopydb
-        pg_version: v16
+        pg_version: v${{ matrix.postgres_version }}
         save_perf_report: true
         aws-oicd-role-arn: ${{ vars.DEV_AWS_OIDC_ROLE_ARN }}
       env:
@@ -146,7 +167,7 @@ jobs:
         ${PSQL} "${BENCHMARK_INGEST_TARGET_CONNSTR}" -c "\dt+"
 
     - name: Delete Neon Project
-      if: ${{ always() && matrix.target_project == 'new_empty_project' }}
+      if: ${{ always() && startsWith(matrix.target_project, 'new_empty_project') }}
       uses: ./.github/actions/neon-project-delete
       with:
         project_id: ${{ steps.create-neon-project-ingest-target.outputs.project_id }}

.github/workflows/neon_extra_builds.yml

@@ -114,7 +114,7 @@ jobs:
         run: make walproposer-lib -j$(nproc)
 
       - name: Produce the build stats
-        run: PQ_LIB_DIR=$(pwd)/pg_install/v17/lib cargo build --all --release --timings -j$(nproc)
+        run: cargo build --all --release --timings -j$(nproc)
 
       - name: Configure AWS credentials
         uses: aws-actions/configure-aws-credentials@v4

.github/workflows/pre-merge-checks.yml

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

.github/workflows/release.yml

@@ -3,8 +3,9 @@ name: Create Release Branch
 on:
   schedule:
     # It should be kept in sync with if-condition in jobs
-    - cron: '0 6 * * FRI' # Storage release
     - cron: '0 6 * * THU' # Proxy release
+    - cron: '0 6 * * FRI' # Storage release
+    - cron: '0 7 * * FRI' # Compute release
   workflow_dispatch:
     inputs:
       create-storage-release-branch:
@@ -55,7 +56,7 @@ jobs:
       ci-access-token: ${{ secrets.CI_ACCESS_TOKEN }}
 
   create-compute-release-branch:
-    if: inputs.create-compute-release-branch
+    if: ${{ github.event.schedule == '0 7 * * FRI' || inputs.create-compute-release-branch }}
 
     permissions:
       contents: write

Cargo.lock

@@ -179,7 +179,7 @@ dependencies = [
  "nom",
  "num-traits",
  "rusticata-macros",
- "thiserror",
+ "thiserror 1.0.69",
  "time",
 ]
 
@@ -718,14 +718,14 @@ dependencies = [
 
 [[package]]
 name = "axum"
-version = "0.7.9"
+version = "0.8.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "edca88bc138befd0323b20752846e6587272d3b03b0343c8ea28a6f819e6e71f"
+checksum = "6d6fd624c75e18b3b4c6b9caf42b1afe24437daaee904069137d8bab077be8b8"
 dependencies = [
- "async-trait",
  "axum-core",
  "base64 0.22.1",
  "bytes",
+ "form_urlencoded",
  "futures-util",
  "http 1.1.0",
  "http-body 1.0.0",
@@ -733,7 +733,7 @@ dependencies = [
  "hyper 1.4.1",
  "hyper-util",
  "itoa",
- "matchit 0.7.0",
+ "matchit",
  "memchr",
  "mime",
  "percent-encoding",
@@ -746,7 +746,7 @@ dependencies = [
  "sha1",
  "sync_wrapper 1.0.1",
  "tokio",
- "tokio-tungstenite 0.24.0",
+ "tokio-tungstenite 0.26.1",
  "tower 0.5.2",
  "tower-layer",
  "tower-service",
@@ -755,11 +755,10 @@ dependencies = [
 
 [[package]]
 name = "axum-core"
-version = "0.4.5"
+version = "0.5.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "09f2bd6146b97ae3359fa0cc6d6b376d9539582c7b4220f041a33ec24c226199"
+checksum = "df1362f362fd16024ae199c1970ce98f9661bf5ef94b9808fee734bc3698b733"
 dependencies = [
- "async-trait",
  "bytes",
  "futures-util",
  "http 1.1.0",
@@ -942,6 +941,18 @@ version = "1.6.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "8c3c1a368f70d6cf7302d78f8f7093da241fb8e8807c05cc9e51a125895a6d5b"
 
+[[package]]
+name = "bb8"
+version = "0.8.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d89aabfae550a5c44b43ab941844ffcd2e993cb6900b342debf59e9ea74acdb8"
+dependencies = [
+ "async-trait",
+ "futures-util",
+ "parking_lot 0.12.1",
+ "tokio",
+]
+
 [[package]]
 name = "bcder"
 version = "0.7.4"
@@ -1118,7 +1129,7 @@ dependencies = [
  "log",
  "nix 0.25.1",
  "regex",
- "thiserror",
+ "thiserror 1.0.69",
 ]
 
 [[package]]
@@ -1299,9 +1310,9 @@ dependencies = [
  "serde_with",
  "signal-hook",
  "tar",
- "thiserror",
+ "thiserror 1.0.69",
  "tokio",
- "tokio-postgres",
+ "tokio-postgres 0.7.7",
  "tokio-stream",
  "tokio-util",
  "tower 0.5.2",
@@ -1312,6 +1323,7 @@ dependencies = [
  "tracing-utils",
  "url",
  "utils",
+ "uuid",
  "vm_monitor",
  "workspace_hack",
  "zstd",
@@ -1407,9 +1419,9 @@ dependencies = [
  "serde",
  "serde_json",
  "storage_broker",
- "thiserror",
+ "thiserror 1.0.69",
  "tokio",
- "tokio-postgres",
+ "tokio-postgres 0.7.7",
  "tokio-util",
  "toml",
  "toml_edit",
@@ -1785,11 +1797,24 @@ dependencies = [
  "chrono",
  "diesel_derives",
  "itoa",
- "pq-sys",
- "r2d2",
  "serde_json",
 ]
 
+[[package]]
+name = "diesel-async"
+version = "0.5.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "51a307ac00f7c23f526a04a77761a0519b9f0eb2838ebf5b905a58580095bdcb"
+dependencies = [
+ "async-trait",
+ "bb8",
+ "diesel",
+ "futures-util",
+ "scoped-futures",
+ "tokio",
+ "tokio-postgres 0.7.12",
+]
+
 [[package]]
 name = "diesel_derives"
 version = "2.2.1"
@@ -2238,7 +2263,7 @@ dependencies = [
  "pin-project",
  "rand 0.8.5",
  "sha1",
- "thiserror",
+ "thiserror 1.0.69",
  "tokio",
  "tokio-util",
 ]
@@ -3364,12 +3389,6 @@ dependencies = [
  "regex-automata 0.1.10",
 ]
 
-[[package]]
-name = "matchit"
-version = "0.7.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b87248edafb776e59e6ee64a79086f65890d3510f2c656c000bf2a7e8a0aea40"
-
 [[package]]
 name = "matchit"
 version = "0.8.4"
@@ -3760,7 +3779,7 @@ dependencies = [
  "serde_json",
  "serde_path_to_error",
  "sha2",
- "thiserror",
+ "thiserror 1.0.69",
  "url",
 ]
 
@@ -3810,7 +3829,7 @@ dependencies = [
  "futures-sink",
  "js-sys",
  "pin-project-lite",
- "thiserror",
+ "thiserror 1.0.69",
  "tracing",
 ]
 
@@ -3842,7 +3861,7 @@ dependencies = [
  "opentelemetry_sdk",
  "prost",
  "reqwest",
- "thiserror",
+ "thiserror 1.0.69",
 ]
 
 [[package]]
@@ -3878,7 +3897,7 @@ dependencies = [
  "percent-encoding",
  "rand 0.8.5",
  "serde_json",
- "thiserror",
+ "thiserror 1.0.69",
  "tokio",
  "tokio-stream",
  "tracing",
@@ -3992,7 +4011,7 @@ dependencies = [
  "remote_storage",
  "serde_json",
  "svg_fmt",
- "thiserror",
+ "thiserror 1.0.69",
  "tokio",
  "tokio-util",
  "utils",
@@ -4041,8 +4060,8 @@ dependencies = [
  "pageserver_compaction",
  "pin-project-lite",
  "postgres",
- "postgres-protocol",
- "postgres-types",
+ "postgres-protocol 0.6.4",
+ "postgres-types 0.2.4",
  "postgres_backend",
  "postgres_connection",
  "postgres_ffi",
@@ -4068,12 +4087,12 @@ dependencies = [
  "strum_macros",
  "sysinfo",
  "tenant_size_model",
- "thiserror",
+ "thiserror 1.0.69",
  "tikv-jemallocator",
  "tokio",
  "tokio-epoll-uring",
  "tokio-io-timeout",
- "tokio-postgres",
+ "tokio-postgres 0.7.7",
  "tokio-stream",
  "tokio-tar",
  "tokio-util",
@@ -4114,7 +4133,7 @@ dependencies = [
  "storage_broker",
  "strum",
  "strum_macros",
- "thiserror",
+ "thiserror 1.0.69",
  "utils",
 ]
 
@@ -4129,9 +4148,9 @@ dependencies = [
  "postgres",
  "reqwest",
  "serde",
- "thiserror",
+ "thiserror 1.0.69",
  "tokio",
- "tokio-postgres",
+ "tokio-postgres 0.7.7",
  "tokio-stream",
  "tokio-util",
  "utils",
@@ -4437,7 +4456,7 @@ dependencies = [
  "futures-util",
  "log",
  "tokio",
- "tokio-postgres",
+ "tokio-postgres 0.7.7",
 ]
 
 [[package]]
@@ -4458,6 +4477,24 @@ dependencies = [
  "stringprep",
 ]
 
+[[package]]
+name = "postgres-protocol"
+version = "0.6.7"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "acda0ebdebc28befa84bee35e651e4c5f09073d668c7aed4cf7e23c3cda84b23"
+dependencies = [
+ "base64 0.22.1",
+ "byteorder",
+ "bytes",
+ "fallible-iterator",
+ "hmac",
+ "md-5",
+ "memchr",
+ "rand 0.8.5",
+ "sha2",
+ "stringprep",
+]
+
 [[package]]
 name = "postgres-protocol2"
 version = "0.1.0"
@@ -4481,7 +4518,18 @@ source = "git+https://github.com/neondatabase/rust-postgres.git?branch=neon#511f
 dependencies = [
  "bytes",
  "fallible-iterator",
- "postgres-protocol",
+ "postgres-protocol 0.6.4",
+]
+
+[[package]]
+name = "postgres-types"
+version = "0.2.8"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f66ea23a2d0e5734297357705193335e0a957696f34bed2f2faefacb2fec336f"
+dependencies = [
+ "bytes",
+ "fallible-iterator",
+ "postgres-protocol 0.6.7",
 ]
 
 [[package]]
@@ -4504,9 +4552,9 @@ dependencies = [
  "rustls 0.23.18",
  "rustls-pemfile 2.1.1",
  "serde",
- "thiserror",
+ "thiserror 1.0.69",
  "tokio",
- "tokio-postgres",
+ "tokio-postgres 0.7.7",
  "tokio-postgres-rustls",
  "tokio-rustls 0.26.0",
  "tokio-util",
@@ -4521,7 +4569,7 @@ dependencies = [
  "itertools 0.10.5",
  "once_cell",
  "postgres",
- "tokio-postgres",
+ "tokio-postgres 0.7.7",
  "url",
 ]
 
@@ -4542,7 +4590,7 @@ dependencies = [
  "pprof",
  "regex",
  "serde",
- "thiserror",
+ "thiserror 1.0.69",
  "tracing",
  "utils",
 ]
@@ -4553,7 +4601,7 @@ version = "0.1.0"
 dependencies = [
  "anyhow",
  "camino",
- "thiserror",
+ "thiserror 1.0.69",
  "tokio",
  "workspace_hack",
 ]
@@ -4586,7 +4634,7 @@ dependencies = [
  "smallvec",
  "symbolic-demangle",
  "tempfile",
- "thiserror",
+ "thiserror 1.0.69",
 ]
 
 [[package]]
@@ -4608,15 +4656,6 @@ version = "0.2.17"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "5b40af805b3121feab8a3c29f04d8ad262fa8e0561883e7653e024ae4479e6de"
 
-[[package]]
-name = "pq-sys"
-version = "0.6.3"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f6cc05d7ea95200187117196eee9edd0644424911821aeb28a18ce60ea0b8793"
-dependencies = [
- "vcpkg",
-]
-
 [[package]]
 name = "pq_proto"
 version = "0.1.0"
@@ -4624,10 +4663,10 @@ dependencies = [
  "byteorder",
  "bytes",
  "itertools 0.10.5",
- "postgres-protocol",
+ "postgres-protocol 0.6.4",
  "rand 0.8.5",
  "serde",
- "thiserror",
+ "thiserror 1.0.69",
  "tokio",
 ]
 
@@ -4698,7 +4737,7 @@ dependencies = [
  "memchr",
  "parking_lot 0.12.1",
  "procfs",
- "thiserror",
+ "thiserror 1.0.69",
 ]
 
 [[package]]
@@ -4868,11 +4907,11 @@ dependencies = [
  "strum",
  "strum_macros",
  "subtle",
- "thiserror",
+ "thiserror 1.0.69",
  "tikv-jemalloc-ctl",
  "tikv-jemallocator",
  "tokio",
- "tokio-postgres",
+ "tokio-postgres 0.7.7",
  "tokio-postgres2",
  "tokio-rustls 0.26.0",
  "tokio-tungstenite 0.21.0",
@@ -4929,17 +4968,6 @@ dependencies = [
  "proc-macro2",
 ]
 
-[[package]]
-name = "r2d2"
-version = "0.8.10"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "51de85fb3fb6524929c8a2eb85e6b6d363de4e8c48f9e2c2eac4944abc181c93"
-dependencies = [
- "log",
- "parking_lot 0.12.1",
- "scheduled-thread-pool",
-]
-
 [[package]]
 name = "rand"
 version = "0.7.3"
@@ -5276,7 +5304,7 @@ dependencies = [
  "http 1.1.0",
  "reqwest",
  "serde",
- "thiserror",
+ "thiserror 1.0.69",
  "tower-service",
 ]
 
@@ -5296,7 +5324,7 @@ dependencies = [
  "reqwest",
  "reqwest-middleware",
  "retry-policies",
- "thiserror",
+ "thiserror 1.0.69",
  "tokio",
  "tracing",
  "wasm-timer",
@@ -5312,7 +5340,7 @@ dependencies = [
  "async-trait",
  "getrandom 0.2.11",
  "http 1.1.0",
- "matchit 0.8.4",
+ "matchit",
  "opentelemetry",
  "reqwest",
  "reqwest-middleware",
@@ -5671,7 +5699,7 @@ dependencies = [
  "pageserver_api",
  "parking_lot 0.12.1",
  "postgres",
- "postgres-protocol",
+ "postgres-protocol 0.6.4",
  "postgres_backend",
  "postgres_ffi",
  "pprof",
@@ -5691,11 +5719,11 @@ dependencies = [
  "storage_broker",
  "strum",
  "strum_macros",
- "thiserror",
+ "thiserror 1.0.69",
  "tikv-jemallocator",
  "tokio",
  "tokio-io-timeout",
- "tokio-postgres",
+ "tokio-postgres 0.7.7",
  "tokio-stream",
  "tokio-tar",
  "tokio-util",
@@ -5730,7 +5758,7 @@ dependencies = [
  "reqwest",
  "safekeeper_api",
  "serde",
- "thiserror",
+ "thiserror 1.0.69",
  "utils",
  "workspace_hack",
 ]
@@ -5754,12 +5782,12 @@ dependencies = [
 ]
 
 [[package]]
-name = "scheduled-thread-pool"
-version = "0.2.7"
+name = "scoped-futures"
+version = "0.1.4"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "3cbc66816425a074528352f5789333ecff06ca41b36b0b0efdfbb29edc391a19"
+checksum = "1b24aae2d0636530f359e9d5ef0c04669d11c5e756699b27a6a6d845d8329091"
 dependencies = [
- "parking_lot 0.12.1",
+ "pin-project-lite",
 ]
 
 [[package]]
@@ -5939,7 +5967,7 @@ dependencies = [
  "rand 0.8.5",
  "serde",
  "serde_json",
- "thiserror",
+ "thiserror 1.0.69",
  "time",
  "url",
  "uuid",
@@ -6011,7 +6039,7 @@ checksum = "c7715380eec75f029a4ef7de39a9200e0a63823176b759d055b613f5a87df6a6"
 dependencies = [
  "percent-encoding",
  "serde",
- "thiserror",
+ "thiserror 1.0.69",
 ]
 
 [[package]]
@@ -6173,7 +6201,7 @@ checksum = "adc4e5204eb1910f40f9cfa375f6f05b68c3abac4b6fd879c8ff5e7ae8a0a085"
 dependencies = [
  "num-bigint",
  "num-traits",
- "thiserror",
+ "thiserror 1.0.69",
  "time",
 ]
 
@@ -6294,6 +6322,7 @@ dependencies = [
  "clap",
  "control_plane",
  "diesel",
+ "diesel-async",
  "diesel_migrations",
  "fail",
  "futures",
@@ -6308,16 +6337,16 @@ dependencies = [
  "pageserver_api",
  "pageserver_client",
  "postgres_connection",
- "r2d2",
  "rand 0.8.5",
  "reqwest",
  "routerify",
+ "scoped-futures",
  "scopeguard",
  "serde",
  "serde_json",
  "strum",
  "strum_macros",
- "thiserror",
+ "thiserror 1.0.69",
  "tokio",
  "tokio-util",
  "tracing",
@@ -6364,7 +6393,7 @@ dependencies = [
  "serde_json",
  "storage_controller_client",
  "tokio",
- "tokio-postgres",
+ "tokio-postgres 0.7.7",
  "tokio-postgres-rustls",
  "tokio-stream",
  "tokio-util",
@@ -6609,7 +6638,16 @@ version = "1.0.69"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "b6aaf5339b578ea85b50e080feb250a3e8ae8cfcdff9a461c9ec2904bc923f52"
 dependencies = [
- "thiserror-impl",
+ "thiserror-impl 1.0.69",
+]
+
+[[package]]
+name = "thiserror"
+version = "2.0.11"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d452f284b73e6d76dd36758a0c8684b1d5be31f92b89d07fd5822175732206fc"
+dependencies = [
+ "thiserror-impl 2.0.11",
 ]
 
 [[package]]
@@ -6623,6 +6661,17 @@ dependencies = [
  "syn 2.0.90",
 ]
 
+[[package]]
+name = "thiserror-impl"
+version = "2.0.11"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "26afc1baea8a989337eeb52b6e72a039780ce45c3edfcc9c5b9d112feeb173c2"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn 2.0.90",
+]
+
 [[package]]
 name = "thread_local"
 version = "1.1.7"
@@ -6773,13 +6822,13 @@ dependencies = [
 [[package]]
 name = "tokio-epoll-uring"
 version = "0.1.0"
-source = "git+https://github.com/neondatabase/tokio-epoll-uring.git?branch=main#33e00106a268644d02ba0461bbd64476073b0ee1"
+source = "git+https://github.com/neondatabase/tokio-epoll-uring.git?branch=main#781989bb540a1408b0b93daa1e9d1fa452195497"
 dependencies = [
  "futures",
  "nix 0.26.4",
  "once_cell",
  "scopeguard",
- "thiserror",
+ "thiserror 1.0.69",
  "tokio",
  "tokio-util",
  "tracing",
@@ -6823,13 +6872,39 @@ dependencies = [
  "percent-encoding",
  "phf",
  "pin-project-lite",
- "postgres-protocol",
- "postgres-types",
+ "postgres-protocol 0.6.4",
+ "postgres-types 0.2.4",
  "socket2",
  "tokio",
  "tokio-util",
 ]
 
+[[package]]
+name = "tokio-postgres"
+version = "0.7.12"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "3b5d3742945bc7d7f210693b0c58ae542c6fd47b17adbbda0885f3dcb34a6bdb"
+dependencies = [
+ "async-trait",
+ "byteorder",
+ "bytes",
+ "fallible-iterator",
+ "futures-channel",
+ "futures-util",
+ "log",
+ "parking_lot 0.12.1",
+ "percent-encoding",
+ "phf",
+ "pin-project-lite",
+ "postgres-protocol 0.6.7",
+ "postgres-types 0.2.8",
+ "rand 0.8.5",
+ "socket2",
+ "tokio",
+ "tokio-util",
+ "whoami",
+]
+
 [[package]]
 name = "tokio-postgres-rustls"
 version = "0.12.0"
@@ -6839,7 +6914,7 @@ dependencies = [
  "ring",
  "rustls 0.23.18",
  "tokio",
- "tokio-postgres",
+ "tokio-postgres 0.7.7",
  "tokio-rustls 0.26.0",
  "x509-certificate",
 ]
@@ -6936,14 +7011,14 @@ dependencies = [
 
 [[package]]
 name = "tokio-tungstenite"
-version = "0.24.0"
+version = "0.26.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "edc5f74e248dc973e0dbb7b74c7e0d6fcc301c694ff50049504004ef4d0cdcd9"
+checksum = "be4bf6fecd69fcdede0ec680aaf474cdab988f9de6bc73d3758f0160e3b7025a"
 dependencies = [
  "futures-util",
  "log",
  "tokio",
- "tungstenite 0.24.0",
+ "tungstenite 0.26.1",
 ]
 
 [[package]]
@@ -7003,12 +7078,9 @@ version = "0.12.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "877c5b330756d856ffcc4553ab34a5684481ade925ecc54bcd1bf02b1d0d4d52"
 dependencies = [
- "async-stream",
  "async-trait",
- "axum",
  "base64 0.22.1",
  "bytes",
- "h2 0.4.4",
  "http 1.1.0",
  "http-body 1.0.0",
  "http-body-util",
@@ -7020,7 +7092,6 @@ dependencies = [
  "prost",
  "rustls-native-certs 0.8.0",
  "rustls-pemfile 2.1.1",
- "socket2",
  "tokio",
  "tokio-rustls 0.26.0",
  "tokio-stream",
@@ -7257,16 +7328,16 @@ dependencies = [
  "log",
  "rand 0.8.5",
  "sha1",
- "thiserror",
+ "thiserror 1.0.69",
  "url",
  "utf-8",
 ]
 
 [[package]]
 name = "tungstenite"
-version = "0.24.0"
+version = "0.26.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "18e5b8366ee7a95b16d32197d0b2604b43a0be89dc5fac9f8e96ccafbaedda8a"
+checksum = "413083a99c579593656008130e29255e54dcaae495be556cc26888f211648c24"
 dependencies = [
  "byteorder",
  "bytes",
@@ -7276,7 +7347,7 @@ dependencies = [
  "log",
  "rand 0.8.5",
  "sha1",
- "thiserror",
+ "thiserror 2.0.11",
  "utf-8",
 ]
 
@@ -7372,7 +7443,7 @@ dependencies = [
 [[package]]
 name = "uring-common"
 version = "0.1.0"
-source = "git+https://github.com/neondatabase/tokio-epoll-uring.git?branch=main#33e00106a268644d02ba0461bbd64476073b0ee1"
+source = "git+https://github.com/neondatabase/tokio-epoll-uring.git?branch=main#781989bb540a1408b0b93daa1e9d1fa452195497"
 dependencies = [
  "bytes",
  "io-uring",
@@ -7471,7 +7542,7 @@ dependencies = [
  "signal-hook",
  "strum",
  "strum_macros",
- "thiserror",
+ "thiserror 1.0.69",
  "tokio",
  "tokio-stream",
  "tokio-tar",
@@ -7501,12 +7572,6 @@ version = "0.1.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "830b7e5d4d90034032940e4ace0d9a9a057e7a45cd94e6c007832e39edb82f6d"
 
-[[package]]
-name = "vcpkg"
-version = "0.2.15"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "accd4ea62f7bb7a82fe23066fb0957d48ef677f6eeb8215f372f52e48bb32426"
-
 [[package]]
 name = "version_check"
 version = "0.9.4"
@@ -7526,7 +7591,7 @@ dependencies = [
  "serde_json",
  "sysinfo",
  "tokio",
- "tokio-postgres",
+ "tokio-postgres 0.7.7",
  "tokio-util",
  "tracing",
  "tracing-subscriber",
@@ -7577,11 +7642,10 @@ dependencies = [
  "remote_storage",
  "serde",
  "serde_json",
- "thiserror",
+ "thiserror 1.0.69",
  "tikv-jemallocator",
  "tokio",
  "tokio-util",
- "tonic",
  "tonic-build",
  "tracing",
  "utils",
@@ -7990,8 +8054,6 @@ version = "0.1.0"
 dependencies = [
  "ahash",
  "anyhow",
- "axum",
- "axum-core",
  "base64 0.13.1",
  "base64 0.21.1",
  "base64ct",
@@ -8072,7 +8134,6 @@ dependencies = [
  "toml_edit",
  "tonic",
  "tower 0.4.13",
- "tower 0.5.2",
  "tracing",
  "tracing-core",
  "url",
@@ -8110,7 +8171,7 @@ dependencies = [
  "ring",
  "signature 2.2.0",
  "spki 0.7.3",
- "thiserror",
+ "thiserror 1.0.69",
  "zeroize",
 ]
 
@@ -8127,7 +8188,7 @@ dependencies = [
  "nom",
  "oid-registry",
  "rusticata-macros",
- "thiserror",
+ "thiserror 1.0.69",
  "time",
 ]
 

Cargo.toml

@@ -65,7 +65,7 @@ aws-smithy-types = "1.2"
 aws-credential-types = "1.2.0"
 aws-sigv4 = { version = "1.2", features = ["sign-http"] }
 aws-types = "1.3"
-axum = { version = "0.7.9", features = ["ws"] }
+axum = { version = "0.8.1", features = ["ws"] }
 base64 = "0.13.0"
 bincode = "1.3"
 bindgen = "0.70"
@@ -187,7 +187,7 @@ tokio-tar = "0.3"
 tokio-util = { version = "0.7.10", features = ["io", "rt"] }
 toml = "0.8"
 toml_edit = "0.22"
-tonic = {version = "0.12.3", features = ["tls", "tls-roots"]}
+tonic = {version = "0.12.3", default-features = false, features = ["channel", "tls", "tls-roots"]}
 tower = { version = "0.5.2", default-features = false }
 tower-http = { version = "0.6.2", features = ["request-id", "trace"] }
 tower-service = "0.3.3"

Dockerfile

@@ -45,7 +45,7 @@ COPY --chown=nonroot . .
 
 ARG ADDITIONAL_RUSTFLAGS
 RUN set -e \
-    && PQ_LIB_DIR=$(pwd)/pg_install/v${STABLE_PG_VERSION}/lib RUSTFLAGS="-Clinker=clang -Clink-arg=-fuse-ld=mold -Clink-arg=-Wl,--no-rosegment -Cforce-frame-pointers=yes ${ADDITIONAL_RUSTFLAGS}" cargo build \
+    && RUSTFLAGS="-Clinker=clang -Clink-arg=-fuse-ld=mold -Clink-arg=-Wl,--no-rosegment -Cforce-frame-pointers=yes ${ADDITIONAL_RUSTFLAGS}" cargo build \
       --bin pg_sni_router  \
       --bin pageserver  \
       --bin pagectl  \

Makefile

@@ -64,8 +64,6 @@ CARGO_BUILD_FLAGS += $(filter -j1,$(MAKEFLAGS))
 CARGO_CMD_PREFIX += $(if $(filter n,$(MAKEFLAGS)),,+)
 # Force cargo not to print progress bar
 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"
 

README.md

@@ -21,8 +21,10 @@ The Neon storage engine consists of two major components:
 
 See developer documentation in [SUMMARY.md](/docs/SUMMARY.md) for more information.
 
-## Running local installation
+## Running a local development environment
 
+Neon can be run on a workstation for small experiments and to test code changes, by
+following these instructions.
 
 #### Installing dependencies on Linux
 1. Install build dependencies and other applicable packages
@@ -238,7 +240,7 @@ postgres=# select * from t;
 > cargo neon stop
 ```
 
-More advanced usages can be found at [Control Plane and Neon Local](./control_plane/README.md).
+More advanced usages can be found at [Local Development Control Plane (`neon_local`))](./control_plane/README.md).
 
 #### Handling build failures
 

compute/compute-node.Dockerfile

@@ -67,6 +67,9 @@ RUN cd postgres && \
     # Enable some of contrib extensions
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/autoinc.control && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/dblink.control && \
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/postgres_fdw.control && \
+    file=/usr/local/pgsql/share/extension/postgres_fdw--1.0.sql && [ -e $file ] && \
+    echo 'GRANT USAGE ON FOREIGN DATA WRAPPER postgres_fdw TO neon_superuser;' >> $file && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/bloom.control && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/earthdistance.control && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/insert_username.control && \
@@ -360,6 +363,8 @@ COPY compute/patches/pgvector.patch /pgvector.patch
 RUN wget https://github.com/pgvector/pgvector/archive/refs/tags/v0.8.0.tar.gz -O pgvector.tar.gz && \
     echo "867a2c328d4928a5a9d6f052cd3bc78c7d60228a9b914ad32aa3db88e9de27b0 pgvector.tar.gz" | sha256sum --check && \
     mkdir pgvector-src && cd pgvector-src && tar xzf ../pgvector.tar.gz --strip-components=1 -C . && \
+    wget https://github.com/pgvector/pgvector/raw/refs/tags/v0.7.4/sql/vector.sql -O ./sql/vector--0.7.4.sql && \
+    echo "10218d05dc02299562252a9484775178b14a1d8edb92a2d1672ef488530f7778 ./sql/vector--0.7.4.sql" | sha256sum --check && \
     patch -p1 < /pgvector.patch && \
     make -j $(getconf _NPROCESSORS_ONLN) OPTFLAGS="" && \
     make -j $(getconf _NPROCESSORS_ONLN) OPTFLAGS="" install && \
@@ -995,24 +1000,50 @@ RUN wget https://github.com/kelvich/pg_tiktoken/archive/9118dd4549b7d8c0bbc98e04
 #########################################################################################
 #
 # Layer "pg-pgx-ulid-build"
-# Compile "pgx_ulid" extension
+# Compile "pgx_ulid" extension for v16 and below
 #
 #########################################################################################
 
 FROM rust-extensions-build AS pg-pgx-ulid-build
 ARG PG_VERSION
 
-# doesn't support v17 yet
-# https://github.com/pksunkara/pgx_ulid/pull/52
-RUN case "${PG_VERSION}" in "v17") \
-    echo "pgx_ulid does not support pg17 as of the latest version (0.1.5)" && exit 0;; \
+RUN case "${PG_VERSION}" in \
+    "v14" | "v15" | "v16") \
+        ;; \
+    *) \
+        echo "skipping the version of pgx_ulid for $PG_VERSION" && exit 0 \
+        ;; \
     esac && \
     wget https://github.com/pksunkara/pgx_ulid/archive/refs/tags/v0.1.5.tar.gz -O pgx_ulid.tar.gz && \
-    echo "9d1659a2da65af0133d5451c454de31b37364e3502087dadf579f790bc8bef17 pgx_ulid.tar.gz" | sha256sum --check && \
+    echo "9d1659a2da65af0133d5451c454de31b37364e3502087dadf579f790bc8bef17  pgx_ulid.tar.gz" | sha256sum --check && \
     mkdir pgx_ulid-src && cd pgx_ulid-src && tar xzf ../pgx_ulid.tar.gz --strip-components=1 -C . && \
-    sed -i 's/pgrx       = "^0.11.2"/pgrx = { version = "=0.11.3", features = [ "unsafe-postgres" ] }/g' Cargo.toml && \
+    sed -i 's/pgrx       = "^0.11.2"/pgrx       = { version = "0.11.3", features = [ "unsafe-postgres" ] }/g' Cargo.toml && \
     cargo pgrx install --release && \
-    echo "trusted = true" >> /usr/local/pgsql/share/extension/ulid.control
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/ulid.control
+
+#########################################################################################
+#
+# Layer "pg-pgx-ulid-pgrx12-build"
+# Compile "pgx_ulid" extension for v17 and up
+#
+#########################################################################################
+
+FROM rust-extensions-build-pgrx12 AS pg-pgx-ulid-pgrx12-build
+ARG PG_VERSION
+
+RUN case "${PG_VERSION}" in \
+    "v17") \
+        ;; \
+    *) \
+        echo "skipping the version of pgx_ulid for $PG_VERSION" && exit 0 \
+        ;; \
+    esac && \
+    wget https://github.com/pksunkara/pgx_ulid/archive/refs/tags/v0.2.0.tar.gz -O pgx_ulid.tar.gz && \
+    echo "cef6a9a2e5e7bd1a10a18989286586ee9e6c1c06005a4055cff190de41bf3e9f 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       = "^0.12.7"/pgrx       = { version = "0.12.9", features = [ "unsafe-postgres" ] }/g' Cargo.toml && \
+    cargo pgrx install --release && \
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/pgx_ulid.control
 
 #########################################################################################
 #
@@ -1157,6 +1188,7 @@ 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=pg-cron-pg-build /usr/local/pgsql/ /usr/local/pgsql/
 COPY --from=pg-pgx-ulid-build /usr/local/pgsql/ /usr/local/pgsql/
+COPY --from=pg-pgx-ulid-pgrx12-build /usr/local/pgsql/ /usr/local/pgsql/
 COPY --from=pg-session-jwt-build /usr/local/pgsql/ /usr/local/pgsql/
 COPY --from=rdkit-pg-build /usr/local/pgsql/ /usr/local/pgsql/
 COPY --from=pg-uuidv7-pg-build /usr/local/pgsql/ /usr/local/pgsql/
@@ -1234,11 +1266,12 @@ RUN set -e \
 
 #########################################################################################
 #
-# Layers "postgres-exporter" and "sql-exporter"
+# Layers "postgres-exporter", "pgbouncer-exporter", and "sql-exporter"
 #
 #########################################################################################
 
 FROM quay.io/prometheuscommunity/postgres-exporter:v0.16.0 AS postgres-exporter
+FROM quay.io/prometheuscommunity/pgbouncer-exporter:v0.10.2 AS pgbouncer-exporter
 
 # Keep the version the same as in build-tools.Dockerfile and
 # test_runner/regress/test_compute_metrics.py.
@@ -1320,9 +1353,6 @@ 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
-#pg_anon is not supported yet for pg v17 so, don't fail if nothing found
-COPY --from=pg-anon-pg-build /pg_anon.tar.g? /ext-src
-COPY compute/patches/pg_anon.patch /ext-src
 COPY --from=pg-ivm-build /pg_ivm.tar.gz /ext-src
 COPY --from=pg-partman-build /pg_partman.tar.gz /ext-src
 RUN cd /ext-src/ && for f in *.tar.gz; \
@@ -1333,9 +1363,6 @@ RUN cd /ext-src/rum-src && patch -p1 <../rum.patch
 RUN cd /ext-src/pgvector-src && patch -p1 <../pgvector.patch
 RUN cd /ext-src/pg_hint_plan-src && patch -p1 < /ext-src/pg_hint_plan_${PG_VERSION}.patch
 COPY --chmod=755 docker-compose/run-tests.sh /run-tests.sh
-RUN case "${PG_VERSION}" in "v17") \
-    echo "postgresql_anonymizer does not yet support PG17" && exit 0;; \
-    esac && 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
@@ -1376,6 +1403,7 @@ RUN mkdir -p /etc/local_proxy && chown postgres:postgres /etc/local_proxy
 
 # Metrics exporter binaries and  configuration files
 COPY --from=postgres-exporter /bin/postgres_exporter /bin/postgres_exporter
+COPY --from=pgbouncer-exporter /bin/pgbouncer_exporter /bin/pgbouncer_exporter
 COPY --from=sql-exporter      /bin/sql_exporter      /bin/sql_exporter
 
 COPY --chown=postgres compute/etc/postgres_exporter.yml /etc/postgres_exporter.yml

compute/etc/pgbouncer.ini

@@ -19,6 +19,8 @@ max_prepared_statements=0
 admin_users=postgres
 unix_socket_dir=/tmp/
 unix_socket_mode=0777
+; required for pgbouncer_exporter
+ignore_startup_parameters=extra_float_digits
 
 ;; Disable connection logging. It produces a lot of logs that no one looks at,
 ;; and we can get similar log entries from the proxy too. We had incidents in

compute/patches/pgvector.patch

@@ -1,8 +1,24 @@
+diff --git a/Makefile b/Makefile
+index 7a4b88c..56678af 100644
+--- a/Makefile
++++ b/Makefile
+@@ -3,7 +3,10 @@ EXTVERSION = 0.8.0
+ 
+ MODULE_big = vector
+ DATA = $(wildcard sql/*--*--*.sql)
+-DATA_built = sql/$(EXTENSION)--$(EXTVERSION).sql
++# This change is needed to install different per-version SQL files
++# like pgvector--0.8.0.sql and pgvector--0.7.4.sql
++# The corresponding file is downloaded during the Docker image build process
++DATA_built = sql/$(EXTENSION)--$(EXTVERSION).sql sql/vector--0.7.4.sql
+ OBJS = src/bitutils.o src/bitvec.o src/halfutils.o src/halfvec.o src/hnsw.o src/hnswbuild.o src/hnswinsert.o src/hnswscan.o src/hnswutils.o src/hnswvacuum.o src/ivfbuild.o src/ivfflat.o src/ivfinsert.o src/ivfkmeans.o src/ivfscan.o src/ivfutils.o src/ivfvacuum.o src/sparsevec.o src/vector.o
+ HEADERS = src/halfvec.h src/sparsevec.h src/vector.h
+ 
 diff --git a/src/hnswbuild.c b/src/hnswbuild.c
-index dcfb2bd..d5189ee 100644
+index b667478..fc1897c 100644
 --- a/src/hnswbuild.c
 +++ b/src/hnswbuild.c
-@@ -860,9 +860,17 @@ HnswParallelBuildMain(dsm_segment *seg, shm_toc *toc)
+@@ -843,9 +843,17 @@ HnswParallelBuildMain(dsm_segment *seg, shm_toc *toc)
  
  	hnswarea = shm_toc_lookup(toc, PARALLEL_KEY_HNSW_AREA, false);
  
@@ -20,7 +36,7 @@ index dcfb2bd..d5189ee 100644
  	/* Close relations within worker */
  	index_close(indexRel, indexLockmode);
  	table_close(heapRel, heapLockmode);
-@@ -1117,12 +1125,38 @@ BuildIndex(Relation heap, Relation index, IndexInfo *indexInfo,
+@@ -1100,12 +1108,38 @@ BuildIndex(Relation heap, Relation index, IndexInfo *indexInfo,
  	SeedRandom(42);
  #endif
  

compute/vm-image-spec-bookworm.yaml

@@ -27,6 +27,10 @@ commands:
     user: nobody
     sysvInitAction: respawn
     shell: 'DATA_SOURCE_NAME="user=cloud_admin sslmode=disable dbname=postgres application_name=postgres-exporter" /bin/postgres_exporter --config.file=/etc/postgres_exporter.yml'
+  - name: pgbouncer-exporter
+    user: postgres
+    sysvInitAction: respawn
+    shell: '/bin/pgbouncer_exporter --pgBouncer.connectionString="postgres:///pgbouncer?host=/tmp&port=6432&dbname=pgbouncer&user=pgbouncer"'
   - name: sql-exporter
     user: nobody
     sysvInitAction: respawn

compute/vm-image-spec-bullseye.yaml

@@ -27,6 +27,10 @@ commands:
     user: nobody
     sysvInitAction: respawn
     shell: 'DATA_SOURCE_NAME="user=cloud_admin sslmode=disable dbname=postgres application_name=postgres-exporter" /bin/postgres_exporter --config.file=/etc/postgres_exporter.yml'
+  - name: pgbouncer-exporter
+    user: postgres
+    sysvInitAction: respawn
+    shell: '/bin/pgbouncer_exporter --pgBouncer.connectionString="postgres:///pgbouncer?host=/tmp&port=6432&dbname=pgbouncer&user=pgbouncer"'
   - name: sql-exporter
     user: nobody
     sysvInitAction: respawn

compute_tools/Cargo.toml

@@ -51,6 +51,7 @@ tracing-subscriber.workspace = true
 tracing-utils.workspace = true
 thiserror.workspace = true
 url.workspace = true
+uuid.workspace = true
 prometheus.workspace = true
 
 postgres_initdb.workspace = true

compute_tools/src/bin/fast_import.rs

@@ -58,6 +58,8 @@ struct Args {
     pg_bin_dir: Utf8PathBuf,
     #[clap(long)]
     pg_lib_dir: Utf8PathBuf,
+    #[clap(long)]
+    pg_port: Option<u16>, // port to run postgres on, 5432 is default
 }
 
 #[serde_with::serde_as]
@@ -74,6 +76,13 @@ enum EncryptionSecret {
     KMS { key_id: String },
 }
 
+// copied from pageserver_api::config::defaults::DEFAULT_LOCALE to avoid dependency just for a constant
+const DEFAULT_LOCALE: &str = if cfg!(target_os = "macos") {
+    "C"
+} else {
+    "C.UTF-8"
+};
+
 #[tokio::main]
 pub(crate) async fn main() -> anyhow::Result<()> {
     utils::logging::init(
@@ -97,6 +106,10 @@ pub(crate) async fn main() -> anyhow::Result<()> {
     let working_directory = args.working_directory;
     let pg_bin_dir = args.pg_bin_dir;
     let pg_lib_dir = args.pg_lib_dir;
+    let pg_port = args.pg_port.unwrap_or_else(|| {
+        info!("pg_port not specified, using default 5432");
+        5432
+    });
 
     // Initialize AWS clients only if s3_prefix is specified
     let (aws_config, kms_client) = if args.s3_prefix.is_some() {
@@ -180,7 +193,7 @@ pub(crate) async fn main() -> anyhow::Result<()> {
     let superuser = "cloud_admin"; // XXX: this shouldn't be hard-coded
     postgres_initdb::do_run_initdb(postgres_initdb::RunInitdbArgs {
         superuser,
-        locale: "en_US.UTF-8", // XXX: this shouldn't be hard-coded,
+        locale: DEFAULT_LOCALE, // XXX: this shouldn't be hard-coded,
         pg_version,
         initdb_bin: pg_bin_dir.join("initdb").as_ref(),
         library_search_path: &pg_lib_dir, // TODO: is this right? Prob works in compute image, not sure about neon_local.
@@ -197,6 +210,7 @@ pub(crate) async fn main() -> anyhow::Result<()> {
     let mut postgres_proc = tokio::process::Command::new(pgbin)
         .arg("-D")
         .arg(&pgdata_dir)
+        .args(["-p", &format!("{pg_port}")])
         .args(["-c", "wal_level=minimal"])
         .args(["-c", "shared_buffers=10GB"])
         .args(["-c", "max_wal_senders=0"])
@@ -216,6 +230,7 @@ pub(crate) async fn main() -> anyhow::Result<()> {
             ),
         ])
         .env_clear()
+        .env("LD_LIBRARY_PATH", &pg_lib_dir)
         .stdout(std::process::Stdio::piped())
         .stderr(std::process::Stdio::piped())
         .spawn()
@@ -232,7 +247,7 @@ pub(crate) async fn main() -> anyhow::Result<()> {
 
     // Create neondb database in the running postgres
     let restore_pg_connstring =
-        format!("host=localhost port=5432 user={superuser} dbname=postgres");
+        format!("host=localhost port={pg_port} user={superuser} dbname=postgres");
 
     let start_time = std::time::Instant::now();
 
@@ -314,6 +329,7 @@ pub(crate) async fn main() -> anyhow::Result<()> {
             .arg(&source_connection_string)
             // how we run it
             .env_clear()
+            .env("LD_LIBRARY_PATH", &pg_lib_dir)
             .kill_on_drop(true)
             .stdout(std::process::Stdio::piped())
             .stderr(std::process::Stdio::piped())
@@ -347,6 +363,7 @@ pub(crate) async fn main() -> anyhow::Result<()> {
             .arg(&dumpdir)
             // how we run it
             .env_clear()
+            .env("LD_LIBRARY_PATH", &pg_lib_dir)
             .kill_on_drop(true)
             .stdout(std::process::Stdio::piped())
             .stderr(std::process::Stdio::piped())

compute_tools/src/compute.rs

@@ -41,14 +41,14 @@ use crate::local_proxy;
 use crate::pg_helpers::*;
 use crate::spec::*;
 use crate::spec_apply::ApplySpecPhase::{
-    CreateAndAlterDatabases, CreateAndAlterRoles, CreateAvailabilityCheck, CreateSuperUser,
-    DropInvalidDatabases, DropRoles, HandleNeonExtension, HandleOtherExtensions,
-    RenameAndDeleteDatabases, RenameRoles, RunInEachDatabase,
+    CreateAndAlterDatabases, CreateAndAlterRoles, CreateAvailabilityCheck, CreateSchemaNeon,
+    CreateSuperUser, DropInvalidDatabases, DropRoles, FinalizeDropLogicalSubscriptions,
+    HandleNeonExtension, HandleOtherExtensions, RenameAndDeleteDatabases, RenameRoles,
+    RunInEachDatabase,
 };
 use crate::spec_apply::PerDatabasePhase;
 use crate::spec_apply::PerDatabasePhase::{
-    ChangeSchemaPerms, DeleteDBRoleReferences, DropSubscriptionsForDeletedDatabases,
-    HandleAnonExtension,
+    ChangeSchemaPerms, DeleteDBRoleReferences, DropLogicalSubscriptions, HandleAnonExtension,
 };
 use crate::spec_apply::{apply_operations, MutableApplyContext, DB};
 use crate::sync_sk::{check_if_synced, ping_safekeeper};
@@ -340,6 +340,15 @@ impl ComputeNode {
         self.state.lock().unwrap().status
     }
 
+    pub fn get_timeline_id(&self) -> Option<TimelineId> {
+        self.state
+            .lock()
+            .unwrap()
+            .pspec
+            .as_ref()
+            .map(|s| s.timeline_id)
+    }
+
     // Remove `pgdata` directory and create it again with right permissions.
     fn create_pgdata(&self) -> Result<()> {
         // Ignore removal error, likely it is a 'No such file or directory (os error 2)'.
@@ -929,6 +938,48 @@ impl ComputeNode {
                 .map(|role| (role.name.clone(), role))
                 .collect::<HashMap<String, Role>>();
 
+            // Check if we need to drop subscriptions before starting the endpoint.
+            //
+            // It is important to do this operation exactly once when endpoint starts on a new branch.
+            // Otherwise, we may drop not inherited, but newly created subscriptions.
+            //
+            // We cannot rely only on spec.drop_subscriptions_before_start flag,
+            // because if for some reason compute restarts inside VM,
+            // it will start again with the same spec and flag value.
+            //
+            // To handle this, we save the fact of the operation in the database
+            // in the neon.drop_subscriptions_done table.
+            // If the table does not exist, we assume that the operation was never performed, so we must do it.
+            // If table exists, we check if the operation was performed on the current timelilne.
+            //
+            let mut drop_subscriptions_done = false;
+
+            if spec.drop_subscriptions_before_start {
+                let timeline_id = self.get_timeline_id().context("timeline_id must be set")?;
+                let query = format!("select 1 from neon.drop_subscriptions_done where timeline_id = '{}'", timeline_id);
+
+                info!("Checking if drop subscription operation was already performed for timeline_id: {}", timeline_id);
+
+                drop_subscriptions_done =  match
+                    client.simple_query(&query).await {
+                    Ok(result) => {
+                        matches!(&result[0], postgres::SimpleQueryMessage::Row(_))
+                    },
+                    Err(e) =>
+                    {
+                        match e.code() {
+                            Some(&SqlState::UNDEFINED_TABLE) => false,
+                            _ => {
+                                // We don't expect any other error here, except for the schema/table not existing
+                                error!("Error checking if drop subscription operation was already performed: {}", e);
+                                return Err(e.into());
+                            }
+                        }
+                    }
+                }
+            };
+
+
             let jwks_roles = Arc::new(
                 spec.as_ref()
                     .local_proxy_config
@@ -996,7 +1047,7 @@ impl ComputeNode {
                         jwks_roles.clone(),
                         concurrency_token.clone(),
                         db,
-                        [DropSubscriptionsForDeletedDatabases].to_vec(),
+                        [DropLogicalSubscriptions].to_vec(),
                     );
 
                     Ok(spawn(fut))
@@ -1024,6 +1075,7 @@ impl ComputeNode {
                 CreateAndAlterRoles,
                 RenameAndDeleteDatabases,
                 CreateAndAlterDatabases,
+                CreateSchemaNeon,
             ] {
                 info!("Applying phase {:?}", &phase);
                 apply_operations(
@@ -1064,6 +1116,17 @@ impl ComputeNode {
                     }
 
                     let conf = Arc::new(conf);
+                    let mut phases = vec![
+                        DeleteDBRoleReferences,
+                        ChangeSchemaPerms,
+                        HandleAnonExtension,
+                    ];
+
+                    if spec.drop_subscriptions_before_start && !drop_subscriptions_done {
+                        info!("Adding DropLogicalSubscriptions phase because drop_subscriptions_before_start is set");
+                        phases.push(DropLogicalSubscriptions);
+                    }
+
                     let fut = Self::apply_spec_sql_db(
                         spec.clone(),
                         conf,
@@ -1071,12 +1134,7 @@ impl ComputeNode {
                         jwks_roles.clone(),
                         concurrency_token.clone(),
                         db,
-                        [
-                            DeleteDBRoleReferences,
-                            ChangeSchemaPerms,
-                            HandleAnonExtension,
-                        ]
-                        .to_vec(),
+                        phases,
                     );
 
                     Ok(spawn(fut))
@@ -1088,12 +1146,20 @@ impl ComputeNode {
                 handle.await??;
             }
 
-            for phase in vec![
+            let mut phases = vec![
                 HandleOtherExtensions,
-                HandleNeonExtension,
+                HandleNeonExtension, // This step depends on CreateSchemaNeon
                 CreateAvailabilityCheck,
                 DropRoles,
-            ] {
+            ];
+
+            // This step depends on CreateSchemaNeon
+            if spec.drop_subscriptions_before_start && !drop_subscriptions_done {
+                info!("Adding FinalizeDropLogicalSubscriptions phase because drop_subscriptions_before_start is set");
+                phases.push(FinalizeDropLogicalSubscriptions);
+            }
+
+            for phase in phases {
                 debug!("Applying phase {:?}", &phase);
                 apply_operations(
                     spec.clone(),
@@ -1463,6 +1529,14 @@ impl ComputeNode {
                         Ok(())
                     },
                 )?;
+
+                let postgresql_conf_path = pgdata_path.join("postgresql.conf");
+                if config::line_in_file(
+                    &postgresql_conf_path,
+                    "neon.disable_logical_replication_subscribers=false",
+                )? {
+                    info!("updated postgresql.conf to set neon.disable_logical_replication_subscribers=false");
+                }
                 self.pg_reload_conf()?;
             }
             self.post_apply_config()?;

compute_tools/src/config.rs

@@ -129,6 +129,13 @@ pub fn write_postgres_conf(
 
     writeln!(file, "neon.extension_server_port={}", extension_server_port)?;
 
+    if spec.drop_subscriptions_before_start {
+        writeln!(file, "neon.disable_logical_replication_subscribers=true")?;
+    } else {
+        // be explicit about the default value
+        writeln!(file, "neon.disable_logical_replication_subscribers=false")?;
+    }
+
     // This is essential to keep this line at the end of the file,
     // because it is intended to override any settings above.
     writeln!(file, "include_if_exists = 'compute_ctl_temp_override.conf'")?;

compute_tools/src/http/extract/json.rs

@@ -1,9 +1,6 @@
 use std::ops::{Deref, DerefMut};
 
-use axum::{
-    async_trait,
-    extract::{rejection::JsonRejection, FromRequest, Request},
-};
+use axum::extract::{rejection::JsonRejection, FromRequest, Request};
 use compute_api::responses::GenericAPIError;
 use http::StatusCode;
 
@@ -12,7 +9,6 @@ use http::StatusCode;
 #[derive(Debug, Clone, Copy, Default)]
 pub(crate) struct Json<T>(pub T);
 
-#[async_trait]
 impl<S, T> FromRequest<S> for Json<T>
 where
     axum::Json<T>: FromRequest<S, Rejection = JsonRejection>,

compute_tools/src/http/extract/path.rs

@@ -1,9 +1,6 @@
 use std::ops::{Deref, DerefMut};
 
-use axum::{
-    async_trait,
-    extract::{rejection::PathRejection, FromRequestParts},
-};
+use axum::extract::{rejection::PathRejection, FromRequestParts};
 use compute_api::responses::GenericAPIError;
 use http::{request::Parts, StatusCode};
 
@@ -12,7 +9,6 @@ use http::{request::Parts, StatusCode};
 #[derive(Debug, Clone, Copy, Default)]
 pub(crate) struct Path<T>(pub T);
 
-#[async_trait]
 impl<S, T> FromRequestParts<S> for Path<T>
 where
     axum::extract::Path<T>: FromRequestParts<S, Rejection = PathRejection>,

compute_tools/src/http/extract/query.rs

@@ -1,9 +1,6 @@
 use std::ops::{Deref, DerefMut};
 
-use axum::{
-    async_trait,
-    extract::{rejection::QueryRejection, FromRequestParts},
-};
+use axum::extract::{rejection::QueryRejection, FromRequestParts};
 use compute_api::responses::GenericAPIError;
 use http::{request::Parts, StatusCode};
 
@@ -12,7 +9,6 @@ use http::{request::Parts, StatusCode};
 #[derive(Debug, Clone, Copy, Default)]
 pub(crate) struct Query<T>(pub T);
 
-#[async_trait]
 impl<S, T> FromRequestParts<S> for Query<T>
 where
     axum::extract::Query<T>: FromRequestParts<S, Rejection = QueryRejection>,

compute_tools/src/http/server.rs

@@ -1,15 +1,14 @@
 use std::{
     net::{IpAddr, Ipv6Addr, SocketAddr},
-    sync::{
-        atomic::{AtomicU64, Ordering},
-        Arc,
-    },
+    sync::Arc,
     thread,
     time::Duration,
 };
 
 use anyhow::Result;
 use axum::{
+    extract::Request,
+    middleware::{self, Next},
     response::{IntoResponse, Response},
     routing::{get, post},
     Router,
@@ -17,11 +16,9 @@ use axum::{
 use http::StatusCode;
 use tokio::net::TcpListener;
 use tower::ServiceBuilder;
-use tower_http::{
-    request_id::{MakeRequestId, PropagateRequestIdLayer, RequestId, SetRequestIdLayer},
-    trace::TraceLayer,
-};
+use tower_http::{request_id::PropagateRequestIdLayer, trace::TraceLayer};
 use tracing::{debug, error, info, Span};
+use uuid::Uuid;
 
 use super::routes::{
     check_writability, configure, database_schema, dbs_and_roles, extension_server, extensions,
@@ -34,37 +31,31 @@ async fn handle_404() -> Response {
     StatusCode::NOT_FOUND.into_response()
 }
 
-#[derive(Clone, Default)]
-struct ComputeMakeRequestId(Arc<AtomicU64>);
+const X_REQUEST_ID: &str = "x-request-id";
 
-impl MakeRequestId for ComputeMakeRequestId {
-    fn make_request_id<B>(
-        &mut self,
-        _request: &http::Request<B>,
-    ) -> Option<tower_http::request_id::RequestId> {
-        let request_id = self
-            .0
-            .fetch_add(1, Ordering::SeqCst)
-            .to_string()
-            .parse()
-            .unwrap();
+/// This middleware function allows compute_ctl to generate its own request ID
+/// if one isn't supplied. The control plane will always send one as a UUID. The
+/// neon Postgres extension on the other hand does not send one.
+async fn maybe_add_request_id_header(mut request: Request, next: Next) -> Response {
+    let headers = request.headers_mut();
 
-        Some(RequestId::new(request_id))
+    if headers.get(X_REQUEST_ID).is_none() {
+        headers.append(X_REQUEST_ID, Uuid::new_v4().to_string().parse().unwrap());
     }
+
+    next.run(request).await
 }
 
 /// Run the HTTP server and wait on it forever.
 #[tokio::main]
 async fn serve(port: u16, compute: Arc<ComputeNode>) {
-    const X_REQUEST_ID: &str = "x-request-id";
-
     let mut app = Router::new()
         .route("/check_writability", post(check_writability::is_writable))
         .route("/configure", post(configure::configure))
         .route("/database_schema", get(database_schema::get_schema_dump))
         .route("/dbs_and_roles", get(dbs_and_roles::get_catalog_objects))
         .route(
-            "/extension_server/*filename",
+            "/extension_server/{*filename}",
             post(extension_server::download_extension),
         )
         .route("/extensions", post(extensions::install_extension))
@@ -82,9 +73,8 @@ async fn serve(port: u16, compute: Arc<ComputeNode>) {
         .fallback(handle_404)
         .layer(
             ServiceBuilder::new()
-                .layer(SetRequestIdLayer::x_request_id(
-                    ComputeMakeRequestId::default(),
-                ))
+                // Add this middleware since we assume the request ID exists
+                .layer(middleware::from_fn(maybe_add_request_id_header))
                 .layer(
                     TraceLayer::new_for_http()
                         .on_request(|request: &http::Request<_>, _span: &Span| {

compute_tools/src/spec_apply.rs

@@ -47,7 +47,7 @@ pub enum PerDatabasePhase {
     DeleteDBRoleReferences,
     ChangeSchemaPerms,
     HandleAnonExtension,
-    DropSubscriptionsForDeletedDatabases,
+    DropLogicalSubscriptions,
 }
 
 #[derive(Clone, Debug)]
@@ -58,11 +58,13 @@ pub enum ApplySpecPhase {
     CreateAndAlterRoles,
     RenameAndDeleteDatabases,
     CreateAndAlterDatabases,
+    CreateSchemaNeon,
     RunInEachDatabase { db: DB, subphase: PerDatabasePhase },
     HandleOtherExtensions,
     HandleNeonExtension,
     CreateAvailabilityCheck,
     DropRoles,
+    FinalizeDropLogicalSubscriptions,
 }
 
 pub struct Operation {
@@ -331,7 +333,7 @@ async fn get_operations<'a>(
                             // NB: there could be other db states, which prevent us from dropping
                             // the database. For example, if db is used by any active subscription
                             // or replication slot.
-                            // Such cases are handled in the DropSubscriptionsForDeletedDatabases
+                            // Such cases are handled in the DropLogicalSubscriptions
                             // phase. We do all the cleanup before actually dropping the database.
                             let drop_db_query: String = format!(
                                 "DROP DATABASE IF EXISTS {} WITH (FORCE)",
@@ -442,13 +444,19 @@ async fn get_operations<'a>(
 
             Ok(Box::new(operations))
         }
+        ApplySpecPhase::CreateSchemaNeon => Ok(Box::new(once(Operation {
+            query: String::from("CREATE SCHEMA IF NOT EXISTS neon"),
+            comment: Some(String::from(
+                "create schema for neon extension and utils tables",
+            )),
+        }))),
         ApplySpecPhase::RunInEachDatabase { db, subphase } => {
             match subphase {
-                PerDatabasePhase::DropSubscriptionsForDeletedDatabases => {
+                PerDatabasePhase::DropLogicalSubscriptions => {
                     match &db {
                         DB::UserDB(db) => {
                             let drop_subscription_query: String = format!(
-                                include_str!("sql/drop_subscription_for_drop_dbs.sql"),
+                                include_str!("sql/drop_subscriptions.sql"),
                                 datname_str = escape_literal(&db.name),
                             );
 
@@ -666,10 +674,6 @@ async fn get_operations<'a>(
         }
         ApplySpecPhase::HandleNeonExtension => {
             let operations = vec![
-                Operation {
-                    query: String::from("CREATE SCHEMA IF NOT EXISTS neon"),
-                    comment: Some(String::from("init: add schema for extension")),
-                },
                 Operation {
                     query: String::from("CREATE EXTENSION IF NOT EXISTS neon WITH SCHEMA neon"),
                     comment: Some(String::from(
@@ -712,5 +716,9 @@ async fn get_operations<'a>(
 
             Ok(Box::new(operations))
         }
+        ApplySpecPhase::FinalizeDropLogicalSubscriptions => Ok(Box::new(once(Operation {
+            query: String::from(include_str!("sql/finalize_drop_subscriptions.sql")),
+            comment: None,
+        }))),
     }
 }

compute_tools/src/sql/finalize_drop_subscriptions.sql

@@ -0,0 +1,21 @@
+DO $$
+BEGIN
+    IF NOT EXISTS(
+        SELECT 1
+        FROM pg_catalog.pg_tables
+        WHERE tablename = 'drop_subscriptions_done'
+        AND schemaname = 'neon'
+    )
+    THEN
+        CREATE TABLE neon.drop_subscriptions_done
+        (id serial primary key, timeline_id text);
+    END IF;
+
+    -- preserve the timeline_id of the last drop_subscriptions run
+    -- to ensure that the cleanup of a timeline is executed only once.
+    -- use upsert to avoid the table bloat in case of cascade branching (branch of a branch)
+    INSERT INTO neon.drop_subscriptions_done VALUES (1, current_setting('neon.timeline_id'))
+    ON CONFLICT (id) DO UPDATE
+    SET timeline_id = current_setting('neon.timeline_id');
+END
+$$

control_plane/README.md

@@ -1,6 +1,10 @@
-# Control Plane and Neon Local
+# Local Development Control Plane (`neon_local`)
 
-This crate contains tools to start a Neon development environment locally. This utility can be used with the `cargo neon` command.
+This crate contains tools to start a Neon development environment locally. This utility can be used with the `cargo neon` command.  This is a convenience to invoke
+the `neon_local` binary.
+
+**Note**: this is a dev/test tool -- a minimal control plane suitable for testing
+code changes locally, but not suitable for running production systems.
 
 ## Example: Start with Postgres 16
 

control_plane/src/bin/neon_local.rs

@@ -1357,6 +1357,7 @@ async fn handle_endpoint(subcmd: &EndpointCmd, env: &local_env::LocalEnv) -> Res
                 args.pg_version,
                 mode,
                 !args.update_catalog,
+                false,
             )?;
         }
         EndpointCmd::Start(args) => {

control_plane/src/endpoint.rs

@@ -76,6 +76,7 @@ pub struct EndpointConf {
     http_port: u16,
     pg_version: u32,
     skip_pg_catalog_updates: bool,
+    drop_subscriptions_before_start: bool,
     features: Vec<ComputeFeature>,
 }
 
@@ -143,6 +144,7 @@ impl ComputeControlPlane {
         pg_version: u32,
         mode: ComputeMode,
         skip_pg_catalog_updates: bool,
+        drop_subscriptions_before_start: bool,
     ) -> Result<Arc<Endpoint>> {
         let pg_port = pg_port.unwrap_or_else(|| self.get_port());
         let http_port = http_port.unwrap_or_else(|| self.get_port() + 1);
@@ -162,6 +164,7 @@ impl ComputeControlPlane {
             // with this we basically test a case of waking up an idle compute, where
             // we also skip catalog updates in the cloud.
             skip_pg_catalog_updates,
+            drop_subscriptions_before_start,
             features: vec![],
         });
 
@@ -177,6 +180,7 @@ impl ComputeControlPlane {
                 pg_port,
                 pg_version,
                 skip_pg_catalog_updates,
+                drop_subscriptions_before_start,
                 features: vec![],
             })?,
         )?;
@@ -240,6 +244,7 @@ pub struct Endpoint {
     // Optimizations
     skip_pg_catalog_updates: bool,
 
+    drop_subscriptions_before_start: bool,
     // Feature flags
     features: Vec<ComputeFeature>,
 }
@@ -291,6 +296,7 @@ impl Endpoint {
             tenant_id: conf.tenant_id,
             pg_version: conf.pg_version,
             skip_pg_catalog_updates: conf.skip_pg_catalog_updates,
+            drop_subscriptions_before_start: conf.drop_subscriptions_before_start,
             features: conf.features,
         })
     }
@@ -625,6 +631,7 @@ impl Endpoint {
             shard_stripe_size: Some(shard_stripe_size),
             local_proxy_config: None,
             reconfigure_concurrency: 1,
+            drop_subscriptions_before_start: self.drop_subscriptions_before_start,
         };
         let spec_path = self.endpoint_path().join("spec.json");
         std::fs::write(spec_path, serde_json::to_string_pretty(&spec)?)?;

control_plane/src/pageserver.rs

@@ -352,6 +352,21 @@ impl PageServerNode {
                 .map(serde_json::from_str)
                 .transpose()
                 .context("Failed to parse 'compaction_algorithm' json")?,
+            l0_flush_delay_threshold: settings
+                .remove("l0_flush_delay_threshold")
+                .map(|x| x.parse::<usize>())
+                .transpose()
+                .context("Failed to parse 'l0_flush_delay_threshold' as an integer")?,
+            l0_flush_wait_upload: settings
+                .remove("l0_flush_wait_upload")
+                .map(|x| x.parse::<bool>())
+                .transpose()
+                .context("Failed to parse 'l0_flush_wait_upload' as a boolean")?,
+            l0_flush_stall_threshold: settings
+                .remove("l0_flush_stall_threshold")
+                .map(|x| x.parse::<usize>())
+                .transpose()
+                .context("Failed to parse 'l0_flush_stall_threshold' as an integer")?,
             gc_horizon: settings
                 .remove("gc_horizon")
                 .map(|x| x.parse::<u64>())
@@ -418,6 +433,26 @@ impl PageServerNode {
                 .map(serde_json::from_str)
                 .transpose()
                 .context("parse `wal_receiver_protocol_override` from json")?,
+            rel_size_v2_enabled: settings
+                .remove("rel_size_v2_enabled")
+                .map(|x| x.parse::<bool>())
+                .transpose()
+                .context("Failed to parse 'rel_size_v2_enabled' as bool")?,
+            gc_compaction_enabled: settings
+                .remove("gc_compaction_enabled")
+                .map(|x| x.parse::<bool>())
+                .transpose()
+                .context("Failed to parse 'gc_compaction_enabled' as bool")?,
+            gc_compaction_initial_threshold_kb: settings
+                .remove("gc_compaction_initial_threshold_kb")
+                .map(|x| x.parse::<u64>())
+                .transpose()
+                .context("Failed to parse 'gc_compaction_initial_threshold_kb' as integer")?,
+            gc_compaction_ratio_percent: settings
+                .remove("gc_compaction_ratio_percent")
+                .map(|x| x.parse::<u64>())
+                .transpose()
+                .context("Failed to parse 'gc_compaction_ratio_percent' as integer")?,
         };
         if !settings.is_empty() {
             bail!("Unrecognized tenant settings: {settings:?}")

control_plane/storcon_cli/src/main.rs

@@ -298,14 +298,7 @@ impl FromStr for SkSchedulingPolicyArg {
     type Err = anyhow::Error;
 
     fn from_str(s: &str) -> Result<Self, Self::Err> {
-        match s {
-            "active" => Ok(Self(SkSchedulingPolicy::Active)),
-            "disabled" => Ok(Self(SkSchedulingPolicy::Disabled)),
-            "decomissioned" => Ok(Self(SkSchedulingPolicy::Decomissioned)),
-            _ => Err(anyhow::anyhow!(
-                "Unknown scheduling policy '{s}', try active,disabled,decomissioned"
-            )),
-        }
+        SkSchedulingPolicy::from_str(s).map(Self)
     }
 }
 

deny.toml

@@ -41,8 +41,8 @@ allow = [
     "MIT",
     "MPL-2.0",
     "OpenSSL",
-    "Unicode-DFS-2016",
     "Unicode-3.0",
+    "Zlib",
 ]
 confidence-threshold = 0.8
 exceptions = [

docker-compose/compute_wrapper/Dockerfile

@@ -10,10 +10,7 @@ USER root
 RUN apt-get update &&       \
     apt-get install -y curl \
                        jq   \
-                       python3-pip \
                        netcat-openbsd
-#Faker is required for the pg_anon test
-RUN case $COMPUTE_IMAGE in compute-node-v17) OPT="--break-system-packages";; *) OPT= ;; esac && pip3 install $OPT Faker
 #This is required for the pg_hintplan test
 RUN mkdir -p /ext-src/pg_hint_plan-src && chown postgres /ext-src/pg_hint_plan-src 
 

docker-compose/docker-compose.yml

@@ -150,8 +150,8 @@ services:
         - REPOSITORY=${REPOSITORY:-neondatabase}
         - COMPUTE_IMAGE=compute-node-v${PG_VERSION:-16}
         - TAG=${TAG:-latest}
-        - http_proxy=$http_proxy
-        - https_proxy=$https_proxy
+        - http_proxy=${http_proxy:-}
+        - https_proxy=${https_proxy:-}
     environment:
       - PG_VERSION=${PG_VERSION:-16}
       #- RUST_BACKTRACE=1
@@ -185,6 +185,8 @@ services:
   neon-test-extensions:
     profiles: ["test-extensions"]
     image: ${REPOSITORY:-neondatabase}/neon-test-extensions-v${PG_TEST_VERSION:-16}:${TAG:-latest}
+    environment:
+      - PGPASSWORD=cloud_admin
     entrypoint:
       - "/bin/bash"
       - "-c"

docker-compose/docker_compose_test.sh

@@ -18,14 +18,10 @@ cd $(dirname $0)
 COMPUTE_CONTAINER_NAME=docker-compose-compute-1
 TEST_CONTAINER_NAME=docker-compose-neon-test-extensions-1
 PSQL_OPTION="-h localhost -U cloud_admin -p 55433 -d postgres"
-: ${http_proxy:=}
-: ${https_proxy:=}
-export http_proxy https_proxy
 
 cleanup() {
     echo "show container information"
     docker ps
-    docker compose --profile test-extensions -f $COMPOSE_FILE logs
     echo "stop containers..."
     docker compose --profile test-extensions -f $COMPOSE_FILE down
 }
@@ -35,12 +31,6 @@ for pg_version in ${TEST_VERSION_ONLY-14 15 16 17}; do
     echo "clean up containers if exists"
     cleanup
     PG_TEST_VERSION=$((pg_version < 16 ? 16 : pg_version))
-    # The support of pg_anon not yet added to PG17, so we have to add the corresponding option for other PG versions
-    if [ "${pg_version}" -ne 17 ]; then
-      SPEC_PATH="compute_wrapper/var/db/postgres/specs"
-      mv $SPEC_PATH/spec.json $SPEC_PATH/spec.bak
-      jq '.cluster.settings += [{"name": "session_preload_libraries","value": "anon","vartype": "string"}]' "${SPEC_PATH}/spec.bak" > "${SPEC_PATH}/spec.json"
-    fi
     PG_VERSION=$pg_version PG_TEST_VERSION=$PG_TEST_VERSION docker compose --profile test-extensions -f $COMPOSE_FILE up --build -d
 
     echo "wait until the compute is ready. timeout after 60s. "
@@ -50,7 +40,6 @@ for pg_version in ${TEST_VERSION_ONLY-14 15 16 17}; do
         cnt=`expr $cnt + 3`
         if [ $cnt -gt 60 ]; then
             echo "timeout before the compute is ready."
-            cleanup
             exit 1
         fi
         if docker compose --profile test-extensions -f $COMPOSE_FILE logs "compute_is_ready" | grep -q "accepting connections"; then
@@ -62,36 +51,19 @@ for pg_version in ${TEST_VERSION_ONLY-14 15 16 17}; do
     done
 
     if [ $pg_version -ge 16 ]; then
-        echo Enabling trust connection
-        docker exec $COMPUTE_CONTAINER_NAME bash -c "sed -i '\$d' /var/db/postgres/compute/pg_hba.conf && echo -e 'host\t all\t all\t all\t trust' >> /var/db/postgres/compute/pg_hba.conf && psql $PSQL_OPTION -c 'select pg_reload_conf()' "
-        echo Adding postgres role
-        docker exec $COMPUTE_CONTAINER_NAME psql $PSQL_OPTION -c "CREATE ROLE postgres SUPERUSER LOGIN"
         # This is required for the pg_hint_plan test, to prevent flaky log message causing the test to fail
         # It cannot be moved to Dockerfile now because the database directory is created after the start of the container
         echo Adding dummy config
         docker exec $COMPUTE_CONTAINER_NAME touch /var/db/postgres/compute/compute_ctl_temp_override.conf
-        # This block is required for the pg_anon extension test.
-        # The test assumes that it is running on the same host with the postgres engine.
-        # In our case it's not true, that's why we are copying files to the compute node
+        # The following block copies the files for the pg_hintplan test to the compute node for the extension test in an isolated docker-compose environment
         TMPDIR=$(mktemp -d)
-        # Add support for pg_anon for pg_v16
-        if [ $pg_version -ne 17 ]; then
-          docker cp $TEST_CONTAINER_NAME:/ext-src/pg_anon-src/data $TMPDIR/data
-          echo -e '1\t too \t many \t tabs' > $TMPDIR/data/bad.csv
-          docker cp $TMPDIR/data $COMPUTE_CONTAINER_NAME:/tmp/tmp_anon_alternate_data
-        rm -rf $TMPDIR
-        fi
-        TMPDIR=$(mktemp -d)
-        # The following block does the same for the pg_hintplan test
         docker cp $TEST_CONTAINER_NAME:/ext-src/pg_hint_plan-src/data $TMPDIR/data
         docker cp $TMPDIR/data $COMPUTE_CONTAINER_NAME:/ext-src/pg_hint_plan-src/
         rm -rf $TMPDIR
         # We are running tests now
-        if docker exec -e SKIP=timescaledb-src,rdkit-src,postgis-src,pgx_ulid-src,pgtap-src,pg_tiktoken-src,pg_jsonschema-src,pg_graphql-src,kq_imcx-src,wal2json_2_5-src \
+        if ! docker exec -e SKIP=timescaledb-src,rdkit-src,postgis-src,pgx_ulid-src,pgtap-src,pg_tiktoken-src,pg_jsonschema-src,pg_graphql-src,kq_imcx-src,wal2json_2_5-src \
             $TEST_CONTAINER_NAME /run-tests.sh | tee testout.txt
         then
-            cleanup
-        else
             FAILED=$(tail -1 testout.txt)
             for d in $FAILED
             do
@@ -101,13 +73,7 @@ for pg_version in ${TEST_VERSION_ONLY-14 15 16 17}; do
                 cat $d/regression.out $d/regression.diffs || true
             done
         rm -rf $FAILED
-        cleanup
         exit 1
         fi
     fi
-    cleanup
-    # Restore the original spec.json
-    if [ "$pg_version" -ne 17 ]; then
-      mv "$SPEC_PATH/spec.bak" "$SPEC_PATH/spec.json"
-    fi
 done

docker-compose/run-tests.sh

@@ -7,7 +7,10 @@ LIST=$( (echo -e "${SKIP//","/"\n"}"; ls -d -- *-src) | sort | uniq -u)
 for d in ${LIST}
 do
        [ -d "${d}" ] || continue
-    psql -c "select 1" >/dev/null || break
+       if ! psql -w -c "select 1" >/dev/null; then
+          FAILED="${d} ${FAILED}"
+          break
+       fi
        USE_PGXS=1 make -C "${d}" installcheck || FAILED="${d} ${FAILED}"
 done
 [ -z "${FAILED}" ] && exit 0

docs/rfcs/041-sharded-ingest.md

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

libs/compute_api/src/spec.rs

@@ -138,6 +138,13 @@ pub struct ComputeSpec {
     /// enough spare connections for reconfiguration process to succeed.
     #[serde(default = "default_reconfigure_concurrency")]
     pub reconfigure_concurrency: usize,
+
+    /// If set to true, the compute_ctl will drop all subscriptions before starting the
+    /// compute. This is needed when we start an endpoint on a branch, so that child
+    /// would not compete with parent branch subscriptions
+    /// over the same replication content from publisher.
+    #[serde(default)] // Default false
+    pub drop_subscriptions_before_start: bool,
 }
 
 /// Feature flag to signal `compute_ctl` to enable certain experimental functionality.

libs/pageserver_api/src/config.rs

@@ -120,6 +120,7 @@ pub struct ConfigToml {
     pub no_sync: Option<bool>,
     pub wal_receiver_protocol: PostgresClientProtocol,
     pub page_service_pipelining: PageServicePipeliningConfig,
+    pub get_vectored_concurrent_io: GetVectoredConcurrentIo,
 }
 
 #[derive(Debug, Clone, PartialEq, Eq, serde::Serialize, serde::Deserialize)]
@@ -158,6 +159,25 @@ pub enum PageServiceProtocolPipelinedExecutionStrategy {
     Tasks,
 }
 
+#[derive(Debug, Clone, PartialEq, Eq, serde::Serialize, serde::Deserialize)]
+#[serde(tag = "mode", rename_all = "kebab-case")]
+#[serde(deny_unknown_fields)]
+pub enum GetVectoredConcurrentIo {
+    /// The read path is fully sequential: layers are visited
+    /// one after the other and IOs are issued and waited upon
+    /// from the same task that traverses the layers.
+    Sequential,
+    /// The read path still traverses layers sequentially, and
+    /// index blocks will be read into the PS PageCache from
+    /// that task, with waiting.
+    /// But data IOs are dispatched and waited upon from a sidecar
+    /// task so that the traversing task can continue to traverse
+    /// layers while the IOs are in flight.
+    /// If the PS PageCache miss rate is low, this improves
+    /// throughput dramatically.
+    SidecarTask,
+}
+
 pub mod statvfs {
     pub mod mock {
         #[derive(Debug, Clone, PartialEq, Eq, serde::Serialize, serde::Deserialize)]
@@ -234,9 +254,21 @@ pub struct TenantConfigToml {
     // Duration::ZERO means automatic compaction is disabled.
     #[serde(with = "humantime_serde")]
     pub compaction_period: Duration,
-    // Level0 delta layer threshold for compaction.
+    /// Level0 delta layer threshold for compaction.
     pub compaction_threshold: usize,
     pub compaction_algorithm: crate::models::CompactionAlgorithmSettings,
+    /// Level0 delta layer threshold at which to delay layer flushes for compaction backpressure,
+    /// such that they take 2x as long, and start waiting for layer flushes during ephemeral layer
+    /// rolls. This helps compaction keep up with WAL ingestion, and avoids read amplification
+    /// blowing up. Should be >compaction_threshold. 0 to disable. Disabled by default.
+    pub l0_flush_delay_threshold: Option<usize>,
+    /// Level0 delta layer threshold at which to stall layer flushes. Must be >compaction_threshold
+    /// to avoid deadlock. 0 to disable. Disabled by default.
+    pub l0_flush_stall_threshold: Option<usize>,
+    /// If true, Level0 delta layer flushes will wait for S3 upload before flushing the next
+    /// layer. This is a temporary backpressure mechanism which should be removed once
+    /// l0_flush_{delay,stall}_threshold is fully enabled.
+    pub l0_flush_wait_upload: bool,
     // Determines how much history is retained, to allow
     // branching and read replicas at an older point in time.
     // The unit is #of bytes of WAL.
@@ -301,6 +333,20 @@ pub struct TenantConfigToml {
     pub timeline_offloading: bool,
 
     pub wal_receiver_protocol_override: Option<PostgresClientProtocol>,
+
+    /// Enable rel_size_v2 for this tenant. Once enabled, the tenant will persist this information into
+    /// `index_part.json`, and it cannot be reversed.
+    pub rel_size_v2_enabled: Option<bool>,
+
+    // gc-compaction related configs
+    /// Enable automatic gc-compaction trigger on this tenant.
+    pub gc_compaction_enabled: bool,
+    /// The initial threshold for gc-compaction in KB. Once the total size of layers below the gc-horizon is above this threshold,
+    /// gc-compaction will be triggered.
+    pub gc_compaction_initial_threshold_kb: u64,
+    /// The ratio that triggers the auto gc-compaction. If (the total size of layers between L2 LSN and gc-horizon) / (size below the L2 LSN)
+    /// is above this ratio, gc-compaction will be triggered.
+    pub gc_compaction_ratio_percent: u64,
 }
 
 pub mod defaults {
@@ -450,6 +496,11 @@ impl Default for ConfigToml {
                     execution: PageServiceProtocolPipelinedExecutionStrategy::ConcurrentFutures,
                 })
             },
+            get_vectored_concurrent_io: if !cfg!(test) {
+                GetVectoredConcurrentIo::Sequential
+            } else {
+                GetVectoredConcurrentIo::SidecarTask
+            },
         }
     }
 }
@@ -475,6 +526,8 @@ pub mod tenant_conf_defaults {
     pub const DEFAULT_COMPACTION_ALGORITHM: crate::models::CompactionAlgorithm =
         crate::models::CompactionAlgorithm::Legacy;
 
+    pub const DEFAULT_L0_FLUSH_WAIT_UPLOAD: bool = true;
+
     pub const DEFAULT_GC_HORIZON: u64 = 64 * 1024 * 1024;
 
     // Large DEFAULT_GC_PERIOD is fine as long as PITR_INTERVAL is larger.
@@ -494,6 +547,9 @@ pub mod tenant_conf_defaults {
     // By default ingest enough WAL for two new L0 layers before checking if new image
     // image layers should be created.
     pub const DEFAULT_IMAGE_LAYER_CREATION_CHECK_THRESHOLD: u8 = 2;
+    pub const DEFAULT_GC_COMPACTION_ENABLED: bool = false;
+    pub const DEFAULT_GC_COMPACTION_INITIAL_THRESHOLD_KB: u64 = 10240000;
+    pub const DEFAULT_GC_COMPACTION_RATIO_PERCENT: u64 = 100;
 }
 
 impl Default for TenantConfigToml {
@@ -510,6 +566,9 @@ impl Default for TenantConfigToml {
             compaction_algorithm: crate::models::CompactionAlgorithmSettings {
                 kind: DEFAULT_COMPACTION_ALGORITHM,
             },
+            l0_flush_delay_threshold: None,
+            l0_flush_stall_threshold: None,
+            l0_flush_wait_upload: DEFAULT_L0_FLUSH_WAIT_UPLOAD,
             gc_horizon: DEFAULT_GC_HORIZON,
             gc_period: humantime::parse_duration(DEFAULT_GC_PERIOD)
                 .expect("cannot parse default gc period"),
@@ -538,6 +597,10 @@ impl Default for TenantConfigToml {
             lsn_lease_length_for_ts: LsnLease::DEFAULT_LENGTH_FOR_TS,
             timeline_offloading: false,
             wal_receiver_protocol_override: None,
+            rel_size_v2_enabled: None,
+            gc_compaction_enabled: DEFAULT_GC_COMPACTION_ENABLED,
+            gc_compaction_initial_threshold_kb: DEFAULT_GC_COMPACTION_INITIAL_THRESHOLD_KB,
+            gc_compaction_ratio_percent: DEFAULT_GC_COMPACTION_RATIO_PERCENT,
         }
     }
 }

libs/pageserver_api/src/controller_api.rs

@@ -324,7 +324,7 @@ impl From<NodeSchedulingPolicy> for String {
 #[derive(Serialize, Deserialize, Clone, Copy, Eq, PartialEq, Debug)]
 pub enum SkSchedulingPolicy {
     Active,
-    Disabled,
+    Pause,
     Decomissioned,
 }
 
@@ -334,9 +334,13 @@ impl FromStr for SkSchedulingPolicy {
     fn from_str(s: &str) -> Result<Self, Self::Err> {
         Ok(match s {
             "active" => Self::Active,
-            "disabled" => Self::Disabled,
+            "pause" => Self::Pause,
             "decomissioned" => Self::Decomissioned,
-            _ => return Err(anyhow::anyhow!("Unknown scheduling state '{s}'")),
+            _ => {
+                return Err(anyhow::anyhow!(
+                    "Unknown scheduling policy '{s}', try active,pause,decomissioned"
+                ))
+            }
         })
     }
 }
@@ -346,7 +350,7 @@ impl From<SkSchedulingPolicy> for String {
         use SkSchedulingPolicy::*;
         match value {
             Active => "active",
-            Disabled => "disabled",
+            Pause => "pause",
             Decomissioned => "decomissioned",
         }
         .to_string()

libs/pageserver_api/src/key.rs

@@ -464,6 +464,18 @@ pub fn rel_size_to_key(rel: RelTag) -> Key {
     }
 }
 
+#[inline(always)]
+pub fn rel_size_key_to_rel(key: Key) -> RelTag {
+    assert_eq!(key.field1, 0x00);
+    assert_eq!(key.field6, 0xffff_ffff);
+    RelTag {
+        forknum: key.field5,
+        spcnode: key.field2,
+        dbnode: key.field3,
+        relnode: key.field4,
+    }
+}
+
 impl Key {
     #[inline(always)]
     pub fn is_rel_size_key(&self) -> bool {
@@ -559,6 +571,15 @@ pub fn slru_segment_size_to_key(kind: SlruKind, segno: u32) -> Key {
     }
 }
 
+#[inline(always)]
+pub fn slru_segment_size_key_to_segno(key: Key) -> u32 {
+    assert_eq!(key.field1, 0x01);
+    assert_eq!(key.field3, 1);
+    assert_eq!(key.field5, 0);
+    assert_eq!(key.field6, 0xffff_ffff);
+    key.field4
+}
+
 impl Key {
     pub fn is_slru_segment_size_key(&self) -> bool {
         self.field1 == 0x01

libs/pageserver_api/src/models.rs

@@ -33,7 +33,6 @@ use crate::{
     reltag::RelTag,
     shard::{ShardCount, ShardStripeSize, TenantShardId},
 };
-use anyhow::bail;
 use bytes::{Buf, BufMut, Bytes, BytesMut};
 
 /// The state of a tenant in this pageserver.
@@ -463,6 +462,12 @@ pub struct TenantConfigPatch {
     #[serde(skip_serializing_if = "FieldPatch::is_noop")]
     pub compaction_algorithm: FieldPatch<CompactionAlgorithmSettings>,
     #[serde(skip_serializing_if = "FieldPatch::is_noop")]
+    pub l0_flush_delay_threshold: FieldPatch<usize>,
+    #[serde(skip_serializing_if = "FieldPatch::is_noop")]
+    pub l0_flush_stall_threshold: FieldPatch<usize>,
+    #[serde(skip_serializing_if = "FieldPatch::is_noop")]
+    pub l0_flush_wait_upload: FieldPatch<bool>,
+    #[serde(skip_serializing_if = "FieldPatch::is_noop")]
     pub gc_horizon: FieldPatch<u64>,
     #[serde(skip_serializing_if = "FieldPatch::is_noop")]
     pub gc_period: FieldPatch<String>,
@@ -498,6 +503,14 @@ pub struct TenantConfigPatch {
     pub timeline_offloading: FieldPatch<bool>,
     #[serde(skip_serializing_if = "FieldPatch::is_noop")]
     pub wal_receiver_protocol_override: FieldPatch<PostgresClientProtocol>,
+    #[serde(skip_serializing_if = "FieldPatch::is_noop")]
+    pub rel_size_v2_enabled: FieldPatch<bool>,
+    #[serde(skip_serializing_if = "FieldPatch::is_noop")]
+    pub gc_compaction_enabled: FieldPatch<bool>,
+    #[serde(skip_serializing_if = "FieldPatch::is_noop")]
+    pub gc_compaction_initial_threshold_kb: FieldPatch<u64>,
+    #[serde(skip_serializing_if = "FieldPatch::is_noop")]
+    pub gc_compaction_ratio_percent: FieldPatch<u64>,
 }
 
 /// An alternative representation of `pageserver::tenant::TenantConf` with
@@ -511,6 +524,9 @@ pub struct TenantConfig {
     pub compaction_threshold: Option<usize>,
     // defer parsing compaction_algorithm, like eviction_policy
     pub compaction_algorithm: Option<CompactionAlgorithmSettings>,
+    pub l0_flush_delay_threshold: Option<usize>,
+    pub l0_flush_stall_threshold: Option<usize>,
+    pub l0_flush_wait_upload: Option<bool>,
     pub gc_horizon: Option<u64>,
     pub gc_period: Option<String>,
     pub image_creation_threshold: Option<usize>,
@@ -529,6 +545,10 @@ pub struct TenantConfig {
     pub lsn_lease_length_for_ts: Option<String>,
     pub timeline_offloading: Option<bool>,
     pub wal_receiver_protocol_override: Option<PostgresClientProtocol>,
+    pub rel_size_v2_enabled: Option<bool>,
+    pub gc_compaction_enabled: Option<bool>,
+    pub gc_compaction_initial_threshold_kb: Option<u64>,
+    pub gc_compaction_ratio_percent: Option<u64>,
 }
 
 impl TenantConfig {
@@ -540,6 +560,9 @@ impl TenantConfig {
             mut compaction_period,
             mut compaction_threshold,
             mut compaction_algorithm,
+            mut l0_flush_delay_threshold,
+            mut l0_flush_stall_threshold,
+            mut l0_flush_wait_upload,
             mut gc_horizon,
             mut gc_period,
             mut image_creation_threshold,
@@ -558,6 +581,10 @@ impl TenantConfig {
             mut lsn_lease_length_for_ts,
             mut timeline_offloading,
             mut wal_receiver_protocol_override,
+            mut rel_size_v2_enabled,
+            mut gc_compaction_enabled,
+            mut gc_compaction_initial_threshold_kb,
+            mut gc_compaction_ratio_percent,
         } = self;
 
         patch.checkpoint_distance.apply(&mut checkpoint_distance);
@@ -568,6 +595,13 @@ impl TenantConfig {
         patch.compaction_period.apply(&mut compaction_period);
         patch.compaction_threshold.apply(&mut compaction_threshold);
         patch.compaction_algorithm.apply(&mut compaction_algorithm);
+        patch
+            .l0_flush_delay_threshold
+            .apply(&mut l0_flush_delay_threshold);
+        patch
+            .l0_flush_stall_threshold
+            .apply(&mut l0_flush_stall_threshold);
+        patch.l0_flush_wait_upload.apply(&mut l0_flush_wait_upload);
         patch.gc_horizon.apply(&mut gc_horizon);
         patch.gc_period.apply(&mut gc_period);
         patch
@@ -602,6 +636,16 @@ impl TenantConfig {
         patch
             .wal_receiver_protocol_override
             .apply(&mut wal_receiver_protocol_override);
+        patch.rel_size_v2_enabled.apply(&mut rel_size_v2_enabled);
+        patch
+            .gc_compaction_enabled
+            .apply(&mut gc_compaction_enabled);
+        patch
+            .gc_compaction_initial_threshold_kb
+            .apply(&mut gc_compaction_initial_threshold_kb);
+        patch
+            .gc_compaction_ratio_percent
+            .apply(&mut gc_compaction_ratio_percent);
 
         Self {
             checkpoint_distance,
@@ -610,6 +654,9 @@ impl TenantConfig {
             compaction_period,
             compaction_threshold,
             compaction_algorithm,
+            l0_flush_delay_threshold,
+            l0_flush_stall_threshold,
+            l0_flush_wait_upload,
             gc_horizon,
             gc_period,
             image_creation_threshold,
@@ -628,6 +675,10 @@ impl TenantConfig {
             lsn_lease_length_for_ts,
             timeline_offloading,
             wal_receiver_protocol_override,
+            rel_size_v2_enabled,
+            gc_compaction_enabled,
+            gc_compaction_initial_threshold_kb,
+            gc_compaction_ratio_percent,
         }
     }
 }
@@ -970,6 +1021,13 @@ pub struct TenantConfigPatchRequest {
     pub config: TenantConfigPatch, // as we have a flattened field, we should reject all unknown fields in it
 }
 
+#[derive(Serialize, Deserialize, Debug)]
+pub struct TenantWaitLsnRequest {
+    #[serde(flatten)]
+    pub timelines: HashMap<TimelineId, Lsn>,
+    pub timeout: Duration,
+}
+
 /// See [`TenantState::attachment_status`] and the OpenAPI docs for context.
 #[derive(Serialize, Deserialize, Clone)]
 #[serde(tag = "slug", content = "data", rename_all = "snake_case")]
@@ -1400,6 +1458,8 @@ pub enum PagestreamFeMessage {
     GetPage(PagestreamGetPageRequest),
     DbSize(PagestreamDbSizeRequest),
     GetSlruSegment(PagestreamGetSlruSegmentRequest),
+    #[cfg(feature = "testing")]
+    Test(PagestreamTestRequest),
 }
 
 // Wrapped in libpq CopyData
@@ -1411,6 +1471,22 @@ pub enum PagestreamBeMessage {
     Error(PagestreamErrorResponse),
     DbSize(PagestreamDbSizeResponse),
     GetSlruSegment(PagestreamGetSlruSegmentResponse),
+    #[cfg(feature = "testing")]
+    Test(PagestreamTestResponse),
+}
+
+// Keep in sync with `pagestore_client.h`
+#[repr(u8)]
+enum PagestreamFeMessageTag {
+    Exists = 0,
+    Nblocks = 1,
+    GetPage = 2,
+    DbSize = 3,
+    GetSlruSegment = 4,
+    /* future tags above this line */
+    /// For testing purposes, not available in production.
+    #[cfg(feature = "testing")]
+    Test = 99,
 }
 
 // Keep in sync with `pagestore_client.h`
@@ -1422,7 +1498,28 @@ enum PagestreamBeMessageTag {
     Error = 103,
     DbSize = 104,
     GetSlruSegment = 105,
+    /* future tags above this line */
+    /// For testing purposes, not available in production.
+    #[cfg(feature = "testing")]
+    Test = 199,
 }
+
+impl TryFrom<u8> for PagestreamFeMessageTag {
+    type Error = u8;
+    fn try_from(value: u8) -> Result<Self, u8> {
+        match value {
+            0 => Ok(PagestreamFeMessageTag::Exists),
+            1 => Ok(PagestreamFeMessageTag::Nblocks),
+            2 => Ok(PagestreamFeMessageTag::GetPage),
+            3 => Ok(PagestreamFeMessageTag::DbSize),
+            4 => Ok(PagestreamFeMessageTag::GetSlruSegment),
+            #[cfg(feature = "testing")]
+            99 => Ok(PagestreamFeMessageTag::Test),
+            _ => Err(value),
+        }
+    }
+}
+
 impl TryFrom<u8> for PagestreamBeMessageTag {
     type Error = u8;
     fn try_from(value: u8) -> Result<Self, u8> {
@@ -1433,6 +1530,8 @@ impl TryFrom<u8> for PagestreamBeMessageTag {
             103 => Ok(PagestreamBeMessageTag::Error),
             104 => Ok(PagestreamBeMessageTag::DbSize),
             105 => Ok(PagestreamBeMessageTag::GetSlruSegment),
+            #[cfg(feature = "testing")]
+            199 => Ok(PagestreamBeMessageTag::Test),
             _ => Err(value),
         }
     }
@@ -1550,6 +1649,20 @@ pub struct PagestreamDbSizeResponse {
     pub db_size: i64,
 }
 
+#[cfg(feature = "testing")]
+#[derive(Debug, PartialEq, Eq, Clone)]
+pub struct PagestreamTestRequest {
+    pub hdr: PagestreamRequest,
+    pub batch_key: u64,
+    pub message: String,
+}
+
+#[cfg(feature = "testing")]
+#[derive(Debug)]
+pub struct PagestreamTestResponse {
+    pub req: PagestreamTestRequest,
+}
+
 // This is a cut-down version of TenantHistorySize from the pageserver crate, omitting fields
 // that require pageserver-internal types.  It is sufficient to get the total size.
 #[derive(Serialize, Deserialize, Debug)]
@@ -1569,7 +1682,7 @@ impl PagestreamFeMessage {
 
         match self {
             Self::Exists(req) => {
-                bytes.put_u8(0);
+                bytes.put_u8(PagestreamFeMessageTag::Exists as u8);
                 bytes.put_u64(req.hdr.reqid);
                 bytes.put_u64(req.hdr.request_lsn.0);
                 bytes.put_u64(req.hdr.not_modified_since.0);
@@ -1580,7 +1693,7 @@ impl PagestreamFeMessage {
             }
 
             Self::Nblocks(req) => {
-                bytes.put_u8(1);
+                bytes.put_u8(PagestreamFeMessageTag::Nblocks as u8);
                 bytes.put_u64(req.hdr.reqid);
                 bytes.put_u64(req.hdr.request_lsn.0);
                 bytes.put_u64(req.hdr.not_modified_since.0);
@@ -1591,7 +1704,7 @@ impl PagestreamFeMessage {
             }
 
             Self::GetPage(req) => {
-                bytes.put_u8(2);
+                bytes.put_u8(PagestreamFeMessageTag::GetPage as u8);
                 bytes.put_u64(req.hdr.reqid);
                 bytes.put_u64(req.hdr.request_lsn.0);
                 bytes.put_u64(req.hdr.not_modified_since.0);
@@ -1603,7 +1716,7 @@ impl PagestreamFeMessage {
             }
 
             Self::DbSize(req) => {
-                bytes.put_u8(3);
+                bytes.put_u8(PagestreamFeMessageTag::DbSize as u8);
                 bytes.put_u64(req.hdr.reqid);
                 bytes.put_u64(req.hdr.request_lsn.0);
                 bytes.put_u64(req.hdr.not_modified_since.0);
@@ -1611,13 +1724,24 @@ impl PagestreamFeMessage {
             }
 
             Self::GetSlruSegment(req) => {
-                bytes.put_u8(4);
+                bytes.put_u8(PagestreamFeMessageTag::GetSlruSegment as u8);
                 bytes.put_u64(req.hdr.reqid);
                 bytes.put_u64(req.hdr.request_lsn.0);
                 bytes.put_u64(req.hdr.not_modified_since.0);
                 bytes.put_u8(req.kind);
                 bytes.put_u32(req.segno);
             }
+            #[cfg(feature = "testing")]
+            Self::Test(req) => {
+                bytes.put_u8(PagestreamFeMessageTag::Test as u8);
+                bytes.put_u64(req.hdr.reqid);
+                bytes.put_u64(req.hdr.request_lsn.0);
+                bytes.put_u64(req.hdr.not_modified_since.0);
+                bytes.put_u64(req.batch_key);
+                let message = req.message.as_bytes();
+                bytes.put_u64(message.len() as u64);
+                bytes.put_slice(message);
+            }
         }
 
         bytes.into()
@@ -1645,56 +1769,66 @@ impl PagestreamFeMessage {
             ),
         };
 
-        match msg_tag {
-            0 => Ok(PagestreamFeMessage::Exists(PagestreamExistsRequest {
-                hdr: PagestreamRequest {
-                    reqid,
-                    request_lsn,
-                    not_modified_since,
-                },
-                rel: RelTag {
-                    spcnode: body.read_u32::<BigEndian>()?,
+        match PagestreamFeMessageTag::try_from(msg_tag)
+            .map_err(|tag: u8| anyhow::anyhow!("invalid tag {tag}"))?
+        {
+            PagestreamFeMessageTag::Exists => {
+                Ok(PagestreamFeMessage::Exists(PagestreamExistsRequest {
+                    hdr: PagestreamRequest {
+                        reqid,
+                        request_lsn,
+                        not_modified_since,
+                    },
+                    rel: RelTag {
+                        spcnode: body.read_u32::<BigEndian>()?,
+                        dbnode: body.read_u32::<BigEndian>()?,
+                        relnode: body.read_u32::<BigEndian>()?,
+                        forknum: body.read_u8()?,
+                    },
+                }))
+            }
+            PagestreamFeMessageTag::Nblocks => {
+                Ok(PagestreamFeMessage::Nblocks(PagestreamNblocksRequest {
+                    hdr: PagestreamRequest {
+                        reqid,
+                        request_lsn,
+                        not_modified_since,
+                    },
+                    rel: RelTag {
+                        spcnode: body.read_u32::<BigEndian>()?,
+                        dbnode: body.read_u32::<BigEndian>()?,
+                        relnode: body.read_u32::<BigEndian>()?,
+                        forknum: body.read_u8()?,
+                    },
+                }))
+            }
+            PagestreamFeMessageTag::GetPage => {
+                Ok(PagestreamFeMessage::GetPage(PagestreamGetPageRequest {
+                    hdr: PagestreamRequest {
+                        reqid,
+                        request_lsn,
+                        not_modified_since,
+                    },
+                    rel: RelTag {
+                        spcnode: body.read_u32::<BigEndian>()?,
+                        dbnode: body.read_u32::<BigEndian>()?,
+                        relnode: body.read_u32::<BigEndian>()?,
+                        forknum: body.read_u8()?,
+                    },
+                    blkno: body.read_u32::<BigEndian>()?,
+                }))
+            }
+            PagestreamFeMessageTag::DbSize => {
+                Ok(PagestreamFeMessage::DbSize(PagestreamDbSizeRequest {
+                    hdr: PagestreamRequest {
+                        reqid,
+                        request_lsn,
+                        not_modified_since,
+                    },
                     dbnode: body.read_u32::<BigEndian>()?,
-                    relnode: body.read_u32::<BigEndian>()?,
-                    forknum: body.read_u8()?,
-                },
-            })),
-            1 => Ok(PagestreamFeMessage::Nblocks(PagestreamNblocksRequest {
-                hdr: PagestreamRequest {
-                    reqid,
-                    request_lsn,
-                    not_modified_since,
-                },
-                rel: RelTag {
-                    spcnode: body.read_u32::<BigEndian>()?,
-                    dbnode: body.read_u32::<BigEndian>()?,
-                    relnode: body.read_u32::<BigEndian>()?,
-                    forknum: body.read_u8()?,
-                },
-            })),
-            2 => Ok(PagestreamFeMessage::GetPage(PagestreamGetPageRequest {
-                hdr: PagestreamRequest {
-                    reqid,
-                    request_lsn,
-                    not_modified_since,
-                },
-                rel: RelTag {
-                    spcnode: body.read_u32::<BigEndian>()?,
-                    dbnode: body.read_u32::<BigEndian>()?,
-                    relnode: body.read_u32::<BigEndian>()?,
-                    forknum: body.read_u8()?,
-                },
-                blkno: body.read_u32::<BigEndian>()?,
-            })),
-            3 => Ok(PagestreamFeMessage::DbSize(PagestreamDbSizeRequest {
-                hdr: PagestreamRequest {
-                    reqid,
-                    request_lsn,
-                    not_modified_since,
-                },
-                dbnode: body.read_u32::<BigEndian>()?,
-            })),
-            4 => Ok(PagestreamFeMessage::GetSlruSegment(
+                }))
+            }
+            PagestreamFeMessageTag::GetSlruSegment => Ok(PagestreamFeMessage::GetSlruSegment(
                 PagestreamGetSlruSegmentRequest {
                     hdr: PagestreamRequest {
                         reqid,
@@ -1705,7 +1839,21 @@ impl PagestreamFeMessage {
                     segno: body.read_u32::<BigEndian>()?,
                 },
             )),
-            _ => bail!("unknown smgr message tag: {:?}", msg_tag),
+            #[cfg(feature = "testing")]
+            PagestreamFeMessageTag::Test => Ok(PagestreamFeMessage::Test(PagestreamTestRequest {
+                hdr: PagestreamRequest {
+                    reqid,
+                    request_lsn,
+                    not_modified_since,
+                },
+                batch_key: body.read_u64::<BigEndian>()?,
+                message: {
+                    let len = body.read_u64::<BigEndian>()?;
+                    let mut buf = vec![0; len as usize];
+                    body.read_exact(&mut buf)?;
+                    String::from_utf8(buf)?
+                },
+            })),
         }
     }
 }
@@ -1748,6 +1896,15 @@ impl PagestreamBeMessage {
                         bytes.put_u32((resp.segment.len() / BLCKSZ as usize) as u32);
                         bytes.put(&resp.segment[..]);
                     }
+
+                    #[cfg(feature = "testing")]
+                    Self::Test(resp) => {
+                        bytes.put_u8(Tag::Test as u8);
+                        bytes.put_u64(resp.req.batch_key);
+                        let message = resp.req.message.as_bytes();
+                        bytes.put_u64(message.len() as u64);
+                        bytes.put_slice(message);
+                    }
                 }
             }
             PagestreamProtocolVersion::V3 => {
@@ -1816,6 +1973,18 @@ impl PagestreamBeMessage {
                         bytes.put_u32((resp.segment.len() / BLCKSZ as usize) as u32);
                         bytes.put(&resp.segment[..]);
                     }
+
+                    #[cfg(feature = "testing")]
+                    Self::Test(resp) => {
+                        bytes.put_u8(Tag::Test as u8);
+                        bytes.put_u64(resp.req.hdr.reqid);
+                        bytes.put_u64(resp.req.hdr.request_lsn.0);
+                        bytes.put_u64(resp.req.hdr.not_modified_since.0);
+                        bytes.put_u64(resp.req.batch_key);
+                        let message = resp.req.message.as_bytes();
+                        bytes.put_u64(message.len() as u64);
+                        bytes.put_slice(message);
+                    }
                 }
             }
         }
@@ -1958,6 +2127,28 @@ impl PagestreamBeMessage {
                         segment: segment.into(),
                     })
                 }
+                #[cfg(feature = "testing")]
+                Tag::Test => {
+                    let reqid = buf.read_u64::<BigEndian>()?;
+                    let request_lsn = Lsn(buf.read_u64::<BigEndian>()?);
+                    let not_modified_since = Lsn(buf.read_u64::<BigEndian>()?);
+                    let batch_key = buf.read_u64::<BigEndian>()?;
+                    let len = buf.read_u64::<BigEndian>()?;
+                    let mut msg = vec![0; len as usize];
+                    buf.read_exact(&mut msg)?;
+                    let message = String::from_utf8(msg)?;
+                    Self::Test(PagestreamTestResponse {
+                        req: PagestreamTestRequest {
+                            hdr: PagestreamRequest {
+                                reqid,
+                                request_lsn,
+                                not_modified_since,
+                            },
+                            batch_key,
+                            message,
+                        },
+                    })
+                }
             };
         let remaining = buf.into_inner();
         if !remaining.is_empty() {
@@ -1977,6 +2168,8 @@ impl PagestreamBeMessage {
             Self::Error(_) => "Error",
             Self::DbSize(_) => "DbSize",
             Self::GetSlruSegment(_) => "GetSlruSegment",
+            #[cfg(feature = "testing")]
+            Self::Test(_) => "Test",
         }
     }
 }

libs/remote_storage/src/azure_blob.rs

@@ -377,7 +377,8 @@ impl RemoteStorage for AzureBlobStorage {
 
                 let next_item = next_item?;
 
-                if timeout_try_cnt >= 2 {
+                // Log a warning if we saw two timeouts in a row before a successful request
+                if timeout_try_cnt > 2 {
                     tracing::warn!("Azure Blob Storage list timed out and succeeded after {} tries", timeout_try_cnt);
                 }
                 timeout_try_cnt = 1;

libs/safekeeper_api/src/models.rs

@@ -277,3 +277,8 @@ pub struct TimelineTermBumpResponse {
     pub previous_term: u64,
     pub current_term: u64,
 }
+
+#[derive(Debug, Clone, Deserialize, Serialize)]
+pub struct SafekeeperUtilization {
+    pub timeline_count: u64,
+}

libs/utils/src/env.rs

@@ -2,6 +2,7 @@
 
 use std::{fmt::Display, str::FromStr};
 
+/// For types `V` that implement [`FromStr`].
 pub fn var<V, E>(varname: &str) -> Option<V>
 where
     V: FromStr<Err = E>,
@@ -10,7 +11,9 @@ where
     match std::env::var(varname) {
         Ok(s) => Some(
             s.parse()
-                .map_err(|e| format!("failed to parse env var {varname}: {e:#}"))
+                .map_err(|e| {
+                    format!("failed to parse env var {varname} using FromStr::parse: {e:#}")
+                })
                 .unwrap(),
         ),
         Err(std::env::VarError::NotPresent) => None,
@@ -19,3 +22,24 @@ where
         }
     }
 }
+
+/// For types `V` that implement [`serde::de::DeserializeOwned`].
+pub fn var_serde_json_string<V>(varname: &str) -> Option<V>
+where
+    V: serde::de::DeserializeOwned,
+{
+    match std::env::var(varname) {
+        Ok(s) => Some({
+            let value = serde_json::Value::String(s);
+            serde_json::from_value(value)
+                .map_err(|e| {
+                    format!("failed to parse env var {varname} as a serde_json json string: {e:#}")
+                })
+                .unwrap()
+        }),
+        Err(std::env::VarError::NotPresent) => None,
+        Err(std::env::VarError::NotUnicode(_)) => {
+            panic!("env var {varname} is not unicode")
+        }
+    }
+}

libs/utils/src/failpoint_support.rs

@@ -11,31 +11,55 @@ use tracing::*;
 
 /// Declare a failpoint that can use to `pause` failpoint action.
 /// We don't want to block the executor thread, hence, spawn_blocking + await.
+///
+/// Optionally pass a cancellation token, and this failpoint will drop out of
+/// its pause when the cancellation token fires. This is useful for testing
+/// cases where we would like to block something, but test its clean shutdown behavior.
+/// The macro evaluates to a Result in that case, where Ok(()) is the case
+/// where the failpoint was not paused, and Err() is the case where cancellation
+/// token fired while evaluating the failpoint.
+///
+/// Remember to unpause the failpoint in the test; until that happens, one of the
+/// limited number of spawn_blocking thread pool threads is leaked.
 #[macro_export]
 macro_rules! pausable_failpoint {
-    ($name:literal) => {
+    ($name:literal) => {{
         if cfg!(feature = "testing") {
-            tokio::task::spawn_blocking({
-                let current = tracing::Span::current();
+            let cancel = ::tokio_util::sync::CancellationToken::new();
+            let _ = $crate::pausable_failpoint!($name, &cancel);
+        }
+    }};
+    ($name:literal, $cancel:expr) => {{
+        if cfg!(feature = "testing") {
+            let failpoint_fut = ::tokio::task::spawn_blocking({
+                let current = ::tracing::Span::current();
                 move || {
                     let _entered = current.entered();
-                    tracing::info!("at failpoint {}", $name);
-                    fail::fail_point!($name);
+                    ::tracing::info!("at failpoint {}", $name);
+                    ::fail::fail_point!($name);
+                }
+            });
+            let cancel_fut = async move {
+                $cancel.cancelled().await;
+            };
+            ::tokio::select! {
+                res = failpoint_fut => {
+                    res.expect("spawn_blocking");
+                    // continue with execution
+                    Ok(())
+                },
+                _ = cancel_fut => {
+                    Err(())
                 }
-            })
-            .await
-            .expect("spawn_blocking");
-        }
-    };
-    ($name:literal, $cond:expr) => {
-        if cfg!(feature = "testing") {
-            if $cond {
-                pausable_failpoint!($name)
             }
+        } else {
+            Ok(())
         }
-    };
+    }};
 }
 
+pub use pausable_failpoint;
+
 /// use with fail::cfg("$name", "return(2000)")
 ///
 /// The effect is similar to a "sleep(2000)" action, i.e. we sleep for the

libs/utils/src/guard_arc_swap.rs

@@ -0,0 +1,54 @@
+//! A wrapper around `ArcSwap` that ensures there is only one writer at a time and writes
+//! don't block reads.
+
+use arc_swap::ArcSwap;
+use std::sync::Arc;
+use tokio::sync::TryLockError;
+
+pub struct GuardArcSwap<T> {
+    inner: ArcSwap<T>,
+    guard: tokio::sync::Mutex<()>,
+}
+
+pub struct Guard<'a, T> {
+    _guard: tokio::sync::MutexGuard<'a, ()>,
+    inner: &'a ArcSwap<T>,
+}
+
+impl<T> GuardArcSwap<T> {
+    pub fn new(inner: T) -> Self {
+        Self {
+            inner: ArcSwap::new(Arc::new(inner)),
+            guard: tokio::sync::Mutex::new(()),
+        }
+    }
+
+    pub fn read(&self) -> Arc<T> {
+        self.inner.load_full()
+    }
+
+    pub async fn write_guard(&self) -> Guard<'_, T> {
+        Guard {
+            _guard: self.guard.lock().await,
+            inner: &self.inner,
+        }
+    }
+
+    pub fn try_write_guard(&self) -> Result<Guard<'_, T>, TryLockError> {
+        let guard = self.guard.try_lock()?;
+        Ok(Guard {
+            _guard: guard,
+            inner: &self.inner,
+        })
+    }
+}
+
+impl<T> Guard<'_, T> {
+    pub fn read(&self) -> Arc<T> {
+        self.inner.load_full()
+    }
+
+    pub fn write(&mut self, value: T) {
+        self.inner.store(Arc::new(value));
+    }
+}

libs/utils/src/lib.rs

@@ -98,6 +98,8 @@ pub mod try_rcu;
 
 pub mod pprof;
 
+pub mod guard_arc_swap;
+
 // Re-export used in macro. Avoids adding git-version as dep in target crates.
 #[doc(hidden)]
 pub use git_version;

libs/utils/src/sync/gate.rs

@@ -64,6 +64,12 @@ pub struct GateGuard {
     gate: Arc<GateInner>,
 }
 
+impl GateGuard {
+    pub fn try_clone(&self) -> Result<Self, GateError> {
+        Gate::enter_impl(self.gate.clone())
+    }
+}
+
 impl Drop for GateGuard {
     fn drop(&mut self) {
         if self.gate.closing.load(Ordering::Relaxed) {
@@ -107,11 +113,11 @@ impl Gate {
     /// to avoid blocking close() indefinitely: typically types that contain a Gate will
     /// also contain a CancellationToken.
     pub fn enter(&self) -> Result<GateGuard, GateError> {
-        let permit = self
-            .inner
-            .sem
-            .try_acquire()
-            .map_err(|_| GateError::GateClosed)?;
+        Self::enter_impl(self.inner.clone())
+    }
+
+    fn enter_impl(gate: Arc<GateInner>) -> Result<GateGuard, GateError> {
+        let permit = gate.sem.try_acquire().map_err(|_| GateError::GateClosed)?;
 
         // we now have the permit, let's disable the normal raii functionality and leave
         // "returning" the permit to our GateGuard::drop.
@@ -122,7 +128,7 @@ impl Gate {
 
         Ok(GateGuard {
             span_at_enter: tracing::Span::current(),
-            gate: self.inner.clone(),
+            gate,
         })
     }
 
@@ -252,4 +258,39 @@ mod tests {
         // Attempting to enter() is still forbidden
         gate.enter().expect_err("enter should fail finishing close");
     }
+
+    #[tokio::test(start_paused = true)]
+    async fn clone_gate_guard() {
+        let gate = Gate::default();
+        let forever = Duration::from_secs(24 * 7 * 365);
+
+        let guard1 = gate.enter().expect("gate isn't closed");
+
+        let guard2 = guard1.try_clone().expect("gate isn't clsoed");
+
+        let mut close_fut = std::pin::pin!(gate.close());
+
+        tokio::time::timeout(forever, &mut close_fut)
+            .await
+            .unwrap_err();
+
+        // we polled close_fut once, that should prevent all later enters and clones
+        gate.enter().unwrap_err();
+        guard1.try_clone().unwrap_err();
+        guard2.try_clone().unwrap_err();
+
+        // guard2 keeps gate open even if guard1 is closed
+        drop(guard1);
+        tokio::time::timeout(forever, &mut close_fut)
+            .await
+            .unwrap_err();
+
+        drop(guard2);
+
+        // now that the last guard is dropped, closing should complete
+        close_fut.await;
+
+        // entering is still forbidden
+        gate.enter().expect_err("enter should stilll fail");
+    }
 }

libs/vm_monitor/src/dispatcher.rs

@@ -7,7 +7,7 @@
 //! (notifying it of upscale).
 
 use anyhow::{bail, Context};
-use axum::extract::ws::{Message, WebSocket};
+use axum::extract::ws::{Message, Utf8Bytes, WebSocket};
 use futures::{
     stream::{SplitSink, SplitStream},
     SinkExt, StreamExt,
@@ -82,21 +82,21 @@ impl Dispatcher {
 
         let highest_shared_version = match monitor_range.highest_shared_version(&agent_range) {
             Ok(version) => {
-                sink.send(Message::Text(
+                sink.send(Message::Text(Utf8Bytes::from(
                     serde_json::to_string(&ProtocolResponse::Version(version)).unwrap(),
-                ))
+                )))
                 .await
                 .context("failed to notify agent of negotiated protocol version")?;
                 version
             }
             Err(e) => {
-                sink.send(Message::Text(
+                sink.send(Message::Text(Utf8Bytes::from(
                     serde_json::to_string(&ProtocolResponse::Error(format!(
                         "Received protocol version range {} which does not overlap with {}",
                         agent_range, monitor_range
                     )))
                     .unwrap(),
-                ))
+                )))
                 .await
                 .context("failed to notify agent of no overlap between protocol version ranges")?;
                 Err(e).context("error determining suitable protocol version range")?
@@ -126,7 +126,7 @@ impl Dispatcher {
 
         let json = serde_json::to_string(&message).context("failed to serialize message")?;
         self.sink
-            .send(Message::Text(json))
+            .send(Message::Text(Utf8Bytes::from(json)))
             .await
             .context("stream error sending message")
     }

libs/wal_decoder/Cargo.toml

@@ -17,7 +17,6 @@ postgres_ffi.workspace = true
 serde.workspace = true
 thiserror.workspace = true
 tokio = { workspace = true, features = ["io-util"] }
-tonic.workspace = true
 tracing.workspace = true
 utils.workspace = true
 workspace_hack = { version = "0.1", path = "../../workspace_hack" }

libs/wal_decoder/src/models.rs

@@ -45,7 +45,7 @@ pub mod proto {
     #![allow(clippy::derive_partial_eq_without_eq)]
     // The generated ValueMeta has a `len` method generate for its `len` field.
     #![allow(clippy::len_without_is_empty)]
-    tonic::include_proto!("interpreted_wal");
+    include!(concat!(env!("OUT_DIR"), concat!("/interpreted_wal.rs")));
 }
 
 #[derive(Copy, Clone, Serialize, Deserialize)]

pageserver/Cargo.toml

@@ -8,7 +8,7 @@ license.workspace = true
 default = []
 # Enables test-only APIs, incuding failpoints. In particular, enables the `fail_point!` macro,
 # which adds some runtime cost to run tests on outage conditions
-testing = ["fail/failpoints", "pageserver_api/testing", "wal_decoder/testing"]
+testing = ["fail/failpoints", "pageserver_api/testing", "wal_decoder/testing", "pageserver_client/testing"]
 
 [dependencies]
 anyhow.workspace = true
@@ -114,3 +114,7 @@ harness = false
 [[bench]]
 name = "upload_queue"
 harness = false
+
+[[bin]]
+name = "test_helper_slow_client_reads"
+required-features = [ "testing" ]

pageserver/client/Cargo.toml

@@ -4,6 +4,9 @@ version = "0.1.0"
 edition.workspace = true
 license.workspace = true
 
+[features]
+testing = [ "pageserver_api/testing" ]
+
 [dependencies]
 pageserver_api.workspace = true
 thiserror.workspace = true

pageserver/client/src/mgmt_api.rs

@@ -763,4 +763,19 @@ impl Client {
             .await
             .map_err(Error::ReceiveBody)
     }
+
+    pub async fn wait_lsn(
+        &self,
+        tenant_shard_id: TenantShardId,
+        request: TenantWaitLsnRequest,
+    ) -> Result<StatusCode> {
+        let uri = format!(
+            "{}/v1/tenant/{tenant_shard_id}/wait_lsn",
+            self.mgmt_api_endpoint,
+        );
+
+        self.request_noerror(Method::POST, uri, request)
+            .await
+            .map(|resp| resp.status())
+    }
 }

pageserver/client/src/page_service.rs

@@ -1,6 +1,9 @@
-use std::pin::Pin;
+use std::sync::{Arc, Mutex};
 
-use futures::SinkExt;
+use futures::{
+    stream::{SplitSink, SplitStream},
+    SinkExt, StreamExt,
+};
 use pageserver_api::{
     models::{
         PagestreamBeMessage, PagestreamFeMessage, PagestreamGetPageRequest,
@@ -10,7 +13,6 @@ use pageserver_api::{
 };
 use tokio::task::JoinHandle;
 use tokio_postgres::CopyOutStream;
-use tokio_stream::StreamExt;
 use tokio_util::sync::CancellationToken;
 use utils::{
     id::{TenantId, TimelineId},
@@ -62,15 +64,28 @@ impl Client {
             .client
             .copy_both_simple(&format!("pagestream_v3 {tenant_id} {timeline_id}"))
             .await?;
+        let (sink, stream) = copy_both.split(); // TODO: actually support splitting of the CopyBothDuplex so the lock inside this split adaptor goes away.
         let Client {
             cancel_on_client_drop,
             conn_task,
             client: _,
         } = self;
+        let shared = Arc::new(Mutex::new(PagestreamShared::ConnTaskRunning(
+            ConnTaskRunning {
+                cancel_on_client_drop,
+                conn_task,
+            },
+        )));
         Ok(PagestreamClient {
-            copy_both: Box::pin(copy_both),
-            conn_task,
-            cancel_on_client_drop,
+            sink: PagestreamSender {
+                shared: shared.clone(),
+                sink,
+            },
+            stream: PagestreamReceiver {
+                shared: shared.clone(),
+                stream,
+            },
+            shared,
         })
     }
 
@@ -97,7 +112,28 @@ impl Client {
 
 /// Create using [`Client::pagestream`].
 pub struct PagestreamClient {
-    copy_both: Pin<Box<tokio_postgres::CopyBothDuplex<bytes::Bytes>>>,
+    shared: Arc<Mutex<PagestreamShared>>,
+    sink: PagestreamSender,
+    stream: PagestreamReceiver,
+}
+
+pub struct PagestreamSender {
+    #[allow(dead_code)]
+    shared: Arc<Mutex<PagestreamShared>>,
+    sink: SplitSink<tokio_postgres::CopyBothDuplex<bytes::Bytes>, bytes::Bytes>,
+}
+
+pub struct PagestreamReceiver {
+    #[allow(dead_code)]
+    shared: Arc<Mutex<PagestreamShared>>,
+    stream: SplitStream<tokio_postgres::CopyBothDuplex<bytes::Bytes>>,
+}
+
+enum PagestreamShared {
+    ConnTaskRunning(ConnTaskRunning),
+    ConnTaskCancelledJoinHandleReturnedOrDropped,
+}
+struct ConnTaskRunning {
     cancel_on_client_drop: Option<tokio_util::sync::DropGuard>,
     conn_task: JoinHandle<()>,
 }
@@ -110,11 +146,11 @@ pub struct RelTagBlockNo {
 impl PagestreamClient {
     pub async fn shutdown(self) {
         let Self {
-            copy_both,
-            cancel_on_client_drop: cancel_conn_task,
-            conn_task,
-        } = self;
-        // The `copy_both` contains internal channel sender, the receiver of which is polled by `conn_task`.
+            shared,
+            sink,
+            stream,
+        } = { self };
+        // The `copy_both` split into `sink` and `stream` contains internal channel sender, the receiver of which is polled by `conn_task`.
         // When `conn_task` observes the sender has been dropped, it sends a `FeMessage::CopyFail` into the connection.
         // (see https://github.com/neondatabase/rust-postgres/blob/2005bf79573b8add5cf205b52a2b208e356cc8b0/tokio-postgres/src/copy_both.rs#L56).
         //
@@ -131,27 +167,77 @@ impl PagestreamClient {
         //
         // NB: page_service doesn't have a use case to exit the `pagestream` mode currently.
         // => https://github.com/neondatabase/neon/issues/6390
-        let _ = cancel_conn_task.unwrap();
+        let ConnTaskRunning {
+            cancel_on_client_drop,
+            conn_task,
+        } = {
+            let mut guard = shared.lock().unwrap();
+            match std::mem::replace(
+                &mut *guard,
+                PagestreamShared::ConnTaskCancelledJoinHandleReturnedOrDropped,
+            ) {
+                PagestreamShared::ConnTaskRunning(conn_task_running) => conn_task_running,
+                PagestreamShared::ConnTaskCancelledJoinHandleReturnedOrDropped => unreachable!(),
+            }
+        };
+        let _ = cancel_on_client_drop.unwrap();
         conn_task.await.unwrap();
-        drop(copy_both);
+
+        // Now drop the split copy_both.
+        drop(sink);
+        drop(stream);
+    }
+
+    pub fn split(self) -> (PagestreamSender, PagestreamReceiver) {
+        let Self {
+            shared: _,
+            sink,
+            stream,
+        } = self;
+        (sink, stream)
     }
 
     pub async fn getpage(
         &mut self,
         req: PagestreamGetPageRequest,
     ) -> anyhow::Result<PagestreamGetPageResponse> {
-        let req = PagestreamFeMessage::GetPage(req);
-        let req: bytes::Bytes = req.serialize();
-        // let mut req = tokio_util::io::ReaderStream::new(&req);
-        let mut req = tokio_stream::once(Ok(req));
+        self.getpage_send(req).await?;
+        self.getpage_recv().await
+    }
 
-        self.copy_both.send_all(&mut req).await?;
+    pub async fn getpage_send(&mut self, req: PagestreamGetPageRequest) -> anyhow::Result<()> {
+        self.sink.getpage_send(req).await
+    }
 
-        let next: Option<Result<bytes::Bytes, _>> = self.copy_both.next().await;
+    pub async fn getpage_recv(&mut self) -> anyhow::Result<PagestreamGetPageResponse> {
+        self.stream.getpage_recv().await
+    }
+}
+
+impl PagestreamSender {
+    // TODO: maybe make this impl Sink instead for better composability?
+    pub async fn send(&mut self, msg: PagestreamFeMessage) -> anyhow::Result<()> {
+        let msg = msg.serialize();
+        self.sink.send_all(&mut tokio_stream::once(Ok(msg))).await?;
+        Ok(())
+    }
+
+    pub async fn getpage_send(&mut self, req: PagestreamGetPageRequest) -> anyhow::Result<()> {
+        self.send(PagestreamFeMessage::GetPage(req)).await
+    }
+}
+
+impl PagestreamReceiver {
+    // TODO: maybe make this impl Stream instead for better composability?
+    pub async fn recv(&mut self) -> anyhow::Result<PagestreamBeMessage> {
+        let next: Option<Result<bytes::Bytes, _>> = self.stream.next().await;
         let next: bytes::Bytes = next.unwrap()?;
+        PagestreamBeMessage::deserialize(next)
+    }
 
-        let msg = PagestreamBeMessage::deserialize(next)?;
-        match msg {
+    pub async fn getpage_recv(&mut self) -> anyhow::Result<PagestreamGetPageResponse> {
+        let next: PagestreamBeMessage = self.recv().await?;
+        match next {
             PagestreamBeMessage::GetPage(p) => Ok(p),
             PagestreamBeMessage::Error(e) => anyhow::bail!("Error: {:?}", e),
             PagestreamBeMessage::Exists(_)
@@ -160,7 +246,14 @@ impl PagestreamClient {
             | PagestreamBeMessage::GetSlruSegment(_) => {
                 anyhow::bail!(
                     "unexpected be message kind in response to getpage request: {}",
-                    msg.kind()
+                    next.kind()
+                )
+            }
+            #[cfg(feature = "testing")]
+            PagestreamBeMessage::Test(_) => {
+                anyhow::bail!(
+                    "unexpected be message kind in response to getpage request: {}",
+                    next.kind()
                 )
             }
         }

pageserver/compaction/src/simulator/draw.rs

@@ -160,9 +160,12 @@ pub fn draw_history<W: std::io::Write>(history: &[LayerTraceEvent], mut output:
 
         // Fill in and thicken rectangle if it's an
         // image layer so that we can see it.
-        let mut style = Style::default();
-        style.fill = Fill::Color(rgb(0x80, 0x80, 0x80));
-        style.stroke = Stroke::Color(rgb(0, 0, 0), 0.5);
+        let mut style = Style {
+            fill: Fill::Color(rgb(0x80, 0x80, 0x80)),
+            stroke: Stroke::Color(rgb(0, 0, 0), 0.5),
+            opacity: 1.0,
+            stroke_opacity: 1.0,
+        };
 
         let y_start = lsn_max - lsn_start;
         let y_end = lsn_max - lsn_end;
@@ -214,10 +217,6 @@ pub fn draw_history<W: std::io::Write>(history: &[LayerTraceEvent], mut output:
         files_seen.insert(f);
     }
 
-    let mut record_style = Style::default();
-    record_style.fill = Fill::Color(rgb(0x80, 0x80, 0x80));
-    record_style.stroke = Stroke::None;
-
     writeln!(svg, "{}", EndSvg)?;
 
     let mut layer_events_str = String::new();

pageserver/pagebench/src/cmd/getpage_latest_lsn.rs

@@ -13,7 +13,7 @@ use rand::prelude::*;
 use tokio::task::JoinSet;
 use tracing::info;
 
-use std::collections::HashSet;
+use std::collections::{HashSet, VecDeque};
 use std::future::Future;
 use std::num::NonZeroUsize;
 use std::pin::Pin;
@@ -63,6 +63,10 @@ pub(crate) struct Args {
     #[clap(long)]
     set_io_mode: Option<pageserver_api::models::virtual_file::IoMode>,
 
+    /// Queue depth generated in each client.
+    #[clap(long, default_value = "1")]
+    queue_depth: NonZeroUsize,
+
     targets: Option<Vec<TenantTimelineId>>,
 }
 
@@ -298,6 +302,7 @@ async fn main_impl(
             start_work_barrier.wait().await;
             let client_start = Instant::now();
             let mut ticks_processed = 0;
+            let mut inflight = VecDeque::new();
             while !cancel.is_cancelled() {
                 // Detect if a request took longer than the RPS rate
                 if let Some(period) = &rps_period {
@@ -311,31 +316,37 @@ async fn main_impl(
                     ticks_processed = periods_passed_until_now;
                 }
 
-                let start = Instant::now();
-                let req = {
-                    let mut rng = rand::thread_rng();
-                    let r = &ranges[weights.sample(&mut rng)];
-                    let key: i128 = rng.gen_range(r.start..r.end);
-                    let key = Key::from_i128(key);
-                    assert!(key.is_rel_block_key());
-                    let (rel_tag, block_no) = key
-                        .to_rel_block()
-                        .expect("we filter non-rel-block keys out above");
-                    PagestreamGetPageRequest {
-                        hdr: PagestreamRequest {
-                            reqid: 0,
-                            request_lsn: if rng.gen_bool(args.req_latest_probability) {
-                                Lsn::MAX
-                            } else {
-                                r.timeline_lsn
+                while inflight.len() < args.queue_depth.get() {
+                    let start = Instant::now();
+                    let req = {
+                        let mut rng = rand::thread_rng();
+                        let r = &ranges[weights.sample(&mut rng)];
+                        let key: i128 = rng.gen_range(r.start..r.end);
+                        let key = Key::from_i128(key);
+                        assert!(key.is_rel_block_key());
+                        let (rel_tag, block_no) = key
+                            .to_rel_block()
+                            .expect("we filter non-rel-block keys out above");
+                        PagestreamGetPageRequest {
+                            hdr: PagestreamRequest {
+                                reqid: 0,
+                                request_lsn: if rng.gen_bool(args.req_latest_probability) {
+                                    Lsn::MAX
+                                } else {
+                                    r.timeline_lsn
+                                },
+                                not_modified_since: r.timeline_lsn,
                             },
-                            not_modified_since: r.timeline_lsn,
-                        },
-                        rel: rel_tag,
-                        blkno: block_no,
-                    }
-                };
-                client.getpage(req).await.unwrap();
+                            rel: rel_tag,
+                            blkno: block_no,
+                        }
+                    };
+                    client.getpage_send(req).await.unwrap();
+                    inflight.push_back(start);
+                }
+
+                let start = inflight.pop_front().unwrap();
+                client.getpage_recv().await.unwrap();
                 let end = Instant::now();
                 live_stats.request_done();
                 ticks_processed += 1;

pageserver/src/basebackup.rs

@@ -25,6 +25,7 @@ use tokio_tar::{Builder, EntryType, Header};
 
 use crate::context::RequestContext;
 use crate::pgdatadir_mapping::Version;
+use crate::tenant::storage_layer::IoConcurrency;
 use crate::tenant::Timeline;
 use pageserver_api::reltag::{RelTag, SlruKind};
 
@@ -123,6 +124,13 @@ where
         full_backup,
         replica,
         ctx,
+        io_concurrency: IoConcurrency::spawn_from_conf(
+            timeline.conf,
+            timeline
+                .gate
+                .enter()
+                .map_err(|e| BasebackupError::Server(e.into()))?,
+        ),
     };
     basebackup
         .send_tarball()
@@ -144,6 +152,7 @@ where
     full_backup: bool,
     replica: bool,
     ctx: &'a RequestContext,
+    io_concurrency: IoConcurrency,
 }
 
 /// A sink that accepts SLRU blocks ordered by key and forwards
@@ -303,7 +312,7 @@ where
             for part in slru_partitions.parts {
                 let blocks = self
                     .timeline
-                    .get_vectored(part, self.lsn, self.ctx)
+                    .get_vectored(part, self.lsn, self.io_concurrency.clone(), self.ctx)
                     .await
                     .map_err(|e| BasebackupError::Server(e.into()))?;
 
@@ -358,7 +367,7 @@ where
         let start_time = Instant::now();
         let aux_files = self
             .timeline
-            .list_aux_files(self.lsn, self.ctx)
+            .list_aux_files(self.lsn, self.ctx, self.io_concurrency.clone())
             .await
             .map_err(|e| BasebackupError::Server(e.into()))?;
         let aux_scan_time = start_time.elapsed();
@@ -422,7 +431,7 @@ where
         }
         let repl_origins = self
             .timeline
-            .get_replorigins(self.lsn, self.ctx)
+            .get_replorigins(self.lsn, self.ctx, self.io_concurrency.clone())
             .await
             .map_err(|e| BasebackupError::Server(e.into()))?;
         let n_origins = repl_origins.len();
@@ -489,7 +498,13 @@ where
             for blknum in startblk..endblk {
                 let img = self
                     .timeline
-                    .get_rel_page_at_lsn(src, blknum, Version::Lsn(self.lsn), self.ctx)
+                    .get_rel_page_at_lsn(
+                        src,
+                        blknum,
+                        Version::Lsn(self.lsn),
+                        self.ctx,
+                        self.io_concurrency.clone(),
+                    )
                     .await
                     .map_err(|e| BasebackupError::Server(e.into()))?;
                 segment_data.extend_from_slice(&img[..]);

pageserver/src/bin/pageserver.rs

@@ -135,6 +135,7 @@ fn main() -> anyhow::Result<()> {
     info!(?conf.virtual_file_io_mode, "starting with virtual_file IO mode");
     info!(?conf.wal_receiver_protocol, "starting with WAL receiver protocol");
     info!(?conf.page_service_pipelining, "starting with page service pipelining config");
+    info!(?conf.get_vectored_concurrent_io, "starting with get_vectored IO concurrency config");
 
     // The tenants directory contains all the pageserver local disk state.
     // Create if not exists and make sure all the contents are durable before proceeding.

pageserver/src/bin/test_helper_slow_client_reads.rs

@@ -0,0 +1,65 @@
+use std::{
+    io::{stdin, stdout, Read, Write},
+    time::Duration,
+};
+
+use clap::Parser;
+use pageserver_api::models::{PagestreamRequest, PagestreamTestRequest};
+use utils::{
+    id::{TenantId, TimelineId},
+    lsn::Lsn,
+};
+
+#[derive(clap::Parser)]
+struct Args {
+    connstr: String,
+    tenant_id: TenantId,
+    timeline_id: TimelineId,
+}
+
+#[tokio::main]
+async fn main() -> anyhow::Result<()> {
+    let Args {
+        connstr,
+        tenant_id,
+        timeline_id,
+    } = Args::parse();
+    let client = pageserver_client::page_service::Client::new(connstr).await?;
+    let client = client.pagestream(tenant_id, timeline_id).await?;
+    let (mut sender, _receiver) = client.split();
+
+    eprintln!("filling the pipe");
+    let mut msg = 0;
+    loop {
+        msg += 1;
+        let fut = sender.send(pageserver_api::models::PagestreamFeMessage::Test(
+            PagestreamTestRequest {
+                hdr: PagestreamRequest {
+                    reqid: 0,
+                    request_lsn: Lsn(23),
+                    not_modified_since: Lsn(23),
+                },
+                batch_key: 42,
+                message: format!("message {}", msg),
+            },
+        ));
+        let Ok(res) = tokio::time::timeout(Duration::from_secs(10), fut).await else {
+            eprintln!("pipe seems full");
+            break;
+        };
+        let _: () = res?;
+    }
+
+    let n = stdout().write(b"R")?;
+    assert_eq!(n, 1);
+    stdout().flush()?;
+
+    eprintln!("waiting for signal to tell us to exit");
+
+    let mut buf = [0u8; 1];
+    stdin().read_exact(&mut buf)?;
+
+    eprintln!("termination signal received, exiting");
+
+    anyhow::Ok(())
+}

pageserver/src/config.rs

@@ -191,6 +191,8 @@ pub struct PageServerConf {
     pub wal_receiver_protocol: PostgresClientProtocol,
 
     pub page_service_pipelining: pageserver_api::config::PageServicePipeliningConfig,
+
+    pub get_vectored_concurrent_io: pageserver_api::config::GetVectoredConcurrentIo,
 }
 
 /// Token for authentication to safekeepers
@@ -352,6 +354,7 @@ impl PageServerConf {
             no_sync,
             wal_receiver_protocol,
             page_service_pipelining,
+            get_vectored_concurrent_io,
         } = config_toml;
 
         let mut conf = PageServerConf {
@@ -396,6 +399,7 @@ impl PageServerConf {
             import_pgdata_aws_endpoint_url,
             wal_receiver_protocol,
             page_service_pipelining,
+            get_vectored_concurrent_io,
 
             // ------------------------------------------------------------
             // fields that require additional validation or custom handling

pageserver/src/http/routes.rs

@@ -10,6 +10,7 @@ use std::time::Duration;
 
 use anyhow::{anyhow, Context, Result};
 use enumset::EnumSet;
+use futures::future::join_all;
 use futures::StreamExt;
 use futures::TryFutureExt;
 use humantime::format_rfc3339;
@@ -40,6 +41,7 @@ use pageserver_api::models::TenantShardSplitRequest;
 use pageserver_api::models::TenantShardSplitResponse;
 use pageserver_api::models::TenantSorting;
 use pageserver_api::models::TenantState;
+use pageserver_api::models::TenantWaitLsnRequest;
 use pageserver_api::models::TimelineArchivalConfigRequest;
 use pageserver_api::models::TimelineCreateRequestMode;
 use pageserver_api::models::TimelineCreateRequestModeImportPgdata;
@@ -84,6 +86,7 @@ use crate::tenant::remote_timeline_client::list_remote_tenant_shards;
 use crate::tenant::remote_timeline_client::list_remote_timelines;
 use crate::tenant::secondary::SecondaryController;
 use crate::tenant::size::ModelInputs;
+use crate::tenant::storage_layer::IoConcurrency;
 use crate::tenant::storage_layer::LayerAccessStatsReset;
 use crate::tenant::storage_layer::LayerName;
 use crate::tenant::timeline::import_pgdata;
@@ -94,6 +97,8 @@ use crate::tenant::timeline::CompactOptions;
 use crate::tenant::timeline::CompactRequest;
 use crate::tenant::timeline::CompactionError;
 use crate::tenant::timeline::Timeline;
+use crate::tenant::timeline::WaitLsnTimeout;
+use crate::tenant::timeline::WaitLsnWaiter;
 use crate::tenant::GetTimelineError;
 use crate::tenant::OffloadedTimeline;
 use crate::tenant::{LogicalSizeCalculationCause, PageReconstructError};
@@ -2789,6 +2794,63 @@ async fn secondary_download_handler(
     json_response(status, progress)
 }
 
+async fn wait_lsn_handler(
+    mut request: Request<Body>,
+    cancel: CancellationToken,
+) -> Result<Response<Body>, ApiError> {
+    let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
+    let wait_lsn_request: TenantWaitLsnRequest = json_request(&mut request).await?;
+
+    let state = get_state(&request);
+    let tenant = state
+        .tenant_manager
+        .get_attached_tenant_shard(tenant_shard_id)?;
+
+    let mut wait_futures = Vec::default();
+    for timeline in tenant.list_timelines() {
+        let Some(lsn) = wait_lsn_request.timelines.get(&timeline.timeline_id) else {
+            continue;
+        };
+
+        let fut = {
+            let timeline = timeline.clone();
+            let ctx = RequestContext::new(TaskKind::MgmtRequest, DownloadBehavior::Error);
+            async move {
+                timeline
+                    .wait_lsn(
+                        *lsn,
+                        WaitLsnWaiter::HttpEndpoint,
+                        WaitLsnTimeout::Custom(wait_lsn_request.timeout),
+                        &ctx,
+                    )
+                    .await
+            }
+        };
+        wait_futures.push(fut);
+    }
+
+    if wait_futures.is_empty() {
+        return json_response(StatusCode::NOT_FOUND, ());
+    }
+
+    let all_done = tokio::select! {
+        results = join_all(wait_futures) => {
+            results.iter().all(|res| res.is_ok())
+        },
+        _ = cancel.cancelled() => {
+            return Err(ApiError::Cancelled);
+        }
+    };
+
+    let status = if all_done {
+        StatusCode::OK
+    } else {
+        StatusCode::ACCEPTED
+    };
+
+    json_response(status, ())
+}
+
 async fn secondary_status_handler(
     request: Request<Body>,
     _cancel: CancellationToken,
@@ -2938,8 +3000,15 @@ async fn list_aux_files(
         active_timeline_of_active_tenant(&state.tenant_manager, tenant_shard_id, timeline_id)
             .await?;
 
+    let io_concurrency = IoConcurrency::spawn_from_conf(
+        state.conf,
+        timeline.gate.enter().map_err(|_| ApiError::Cancelled)?,
+    );
+
     let ctx = RequestContext::new(TaskKind::MgmtRequest, DownloadBehavior::Download);
-    let files = timeline.list_aux_files(body.lsn, &ctx).await?;
+    let files = timeline
+        .list_aux_files(body.lsn, &ctx, io_concurrency)
+        .await?;
     json_response(StatusCode::OK, files)
 }
 
@@ -3569,6 +3638,9 @@ pub fn make_router(
         .post("/v1/tenant/:tenant_shard_id/secondary/download", |r| {
             api_handler(r, secondary_download_handler)
         })
+        .post("/v1/tenant/:tenant_shard_id/wait_lsn", |r| {
+            api_handler(r, wait_lsn_handler)
+        })
         .put("/v1/tenant/:tenant_shard_id/break", |r| {
             testing_api_handler("set tenant state to broken", r, handle_tenant_break)
         })

pageserver/src/metrics.rs

@@ -38,6 +38,9 @@ pub(crate) enum StorageTimeOperation {
     #[strum(serialize = "layer flush")]
     LayerFlush,
 
+    #[strum(serialize = "layer flush delay")]
+    LayerFlushDelay,
+
     #[strum(serialize = "compact")]
     Compact,
 
@@ -100,71 +103,30 @@ pub(crate) static VEC_READ_NUM_LAYERS_VISITED: Lazy<Histogram> = Lazy::new(|| {
     .expect("failed to define a metric")
 });
 
-// Metrics collected on operations on the storage repository.
-#[derive(
-    Clone, Copy, enum_map::Enum, strum_macros::EnumString, strum_macros::Display, IntoStaticStr,
-)]
-pub(crate) enum GetKind {
-    Singular,
-    Vectored,
-}
-
-pub(crate) struct ReconstructTimeMetrics {
-    singular: Histogram,
-    vectored: Histogram,
-}
-
-pub(crate) static RECONSTRUCT_TIME: Lazy<ReconstructTimeMetrics> = Lazy::new(|| {
-    let inner = register_histogram_vec!(
-        "pageserver_getpage_reconstruct_seconds",
-        "Time spent in reconstruct_value (reconstruct a page from deltas)",
-        &["get_kind"],
-        CRITICAL_OP_BUCKETS.into(),
+pub(crate) static CONCURRENT_INITDBS: Lazy<UIntGauge> = Lazy::new(|| {
+    register_uint_gauge!(
+        "pageserver_concurrent_initdb",
+        "Number of initdb processes running"
     )
-    .expect("failed to define a metric");
-
-    ReconstructTimeMetrics {
-        singular: inner.with_label_values(&[GetKind::Singular.into()]),
-        vectored: inner.with_label_values(&[GetKind::Vectored.into()]),
-    }
+    .expect("failed to define a metric")
 });
 
-impl ReconstructTimeMetrics {
-    pub(crate) fn for_get_kind(&self, get_kind: GetKind) -> &Histogram {
-        match get_kind {
-            GetKind::Singular => &self.singular,
-            GetKind::Vectored => &self.vectored,
-        }
-    }
-}
-
-pub(crate) struct ReconstructDataTimeMetrics {
-    singular: Histogram,
-    vectored: Histogram,
-}
-
-impl ReconstructDataTimeMetrics {
-    pub(crate) fn for_get_kind(&self, get_kind: GetKind) -> &Histogram {
-        match get_kind {
-            GetKind::Singular => &self.singular,
-            GetKind::Vectored => &self.vectored,
-        }
-    }
-}
-
-pub(crate) static GET_RECONSTRUCT_DATA_TIME: Lazy<ReconstructDataTimeMetrics> = Lazy::new(|| {
-    let inner = register_histogram_vec!(
-        "pageserver_getpage_get_reconstruct_data_seconds",
-        "Time spent in get_reconstruct_value_data",
-        &["get_kind"],
-        CRITICAL_OP_BUCKETS.into(),
+pub(crate) static INITDB_SEMAPHORE_ACQUISITION_TIME: Lazy<Histogram> = Lazy::new(|| {
+    register_histogram!(
+        "pageserver_initdb_semaphore_seconds_global",
+        "Time spent getting a permit from the global initdb semaphore",
+        STORAGE_OP_BUCKETS.into()
     )
-    .expect("failed to define a metric");
+    .expect("failed to define metric")
+});
 
-    ReconstructDataTimeMetrics {
-        singular: inner.with_label_values(&[GetKind::Singular.into()]),
-        vectored: inner.with_label_values(&[GetKind::Vectored.into()]),
-    }
+pub(crate) static INITDB_RUN_TIME: Lazy<Histogram> = Lazy::new(|| {
+    register_histogram!(
+        "pageserver_initdb_seconds_global",
+        "Time spent performing initdb",
+        STORAGE_OP_BUCKETS.into()
+    )
+    .expect("failed to define metric")
 });
 
 pub(crate) struct GetVectoredLatency {
@@ -1463,6 +1425,8 @@ pub enum SmgrQueryType {
     GetPageAtLsn,
     GetDbSize,
     GetSlruSegment,
+    #[cfg(feature = "testing")]
+    Test,
 }
 
 pub(crate) struct SmgrQueryTimePerTimeline {
@@ -2522,12 +2486,19 @@ impl StorageTimeMetricsTimer {
         }
     }
 
-    /// Record the time from creation to now.
-    pub fn stop_and_record(self) {
-        let duration = self.start.elapsed().as_secs_f64();
-        self.metrics.timeline_sum.inc_by(duration);
+    /// Returns the elapsed duration of the timer.
+    pub fn elapsed(&self) -> Duration {
+        self.start.elapsed()
+    }
+
+    /// Record the time from creation to now and return it.
+    pub fn stop_and_record(self) -> Duration {
+        let duration = self.elapsed();
+        let seconds = duration.as_secs_f64();
+        self.metrics.timeline_sum.inc_by(seconds);
         self.metrics.timeline_count.inc();
-        self.metrics.global_histogram.observe(duration);
+        self.metrics.global_histogram.observe(seconds);
+        duration
     }
 
     /// Turns this timer into a timer, which will always record -- usually this means recording
@@ -2547,6 +2518,13 @@ impl Drop for AlwaysRecordingStorageTimeMetricsTimer {
     }
 }
 
+impl AlwaysRecordingStorageTimeMetricsTimer {
+    /// Returns the elapsed duration of the timer.
+    pub fn elapsed(&self) -> Duration {
+        self.0.as_ref().expect("not dropped yet").elapsed()
+    }
+}
+
 /// Timing facilities for an globally histogrammed metric, which is supported by per tenant and
 /// timeline total sum and count.
 #[derive(Clone, Debug)]
@@ -2599,6 +2577,7 @@ pub(crate) struct TimelineMetrics {
     shard_id: String,
     timeline_id: String,
     pub flush_time_histo: StorageTimeMetrics,
+    pub flush_delay_histo: StorageTimeMetrics,
     pub flush_wait_upload_time_gauge: Gauge,
     pub compact_time_histo: StorageTimeMetrics,
     pub create_images_time_histo: StorageTimeMetrics,
@@ -2645,6 +2624,12 @@ impl TimelineMetrics {
             &shard_id,
             &timeline_id,
         );
+        let flush_delay_histo = StorageTimeMetrics::new(
+            StorageTimeOperation::LayerFlushDelay,
+            &tenant_id,
+            &shard_id,
+            &timeline_id,
+        );
         let flush_wait_upload_time_gauge = FLUSH_WAIT_UPLOAD_TIME
             .get_metric_with_label_values(&[&tenant_id, &shard_id, &timeline_id])
             .unwrap();
@@ -2793,6 +2778,7 @@ impl TimelineMetrics {
             shard_id,
             timeline_id,
             flush_time_histo,
+            flush_delay_histo,
             flush_wait_upload_time_gauge,
             compact_time_histo,
             create_images_time_histo,
@@ -3891,7 +3877,6 @@ pub fn preinitialize_metrics(conf: &'static PageServerConf) {
     });
 
     // Custom
-    Lazy::force(&RECONSTRUCT_TIME);
     Lazy::force(&BASEBACKUP_QUERY_TIME);
     Lazy::force(&COMPUTE_COMMANDS_COUNTERS);
     Lazy::force(&tokio_epoll_uring::THREAD_LOCAL_METRICS_STORAGE);

pageserver/src/page_service.rs

@@ -39,6 +39,7 @@ use tokio::io::{AsyncWriteExt, BufWriter};
 use tokio::task::JoinHandle;
 use tokio_util::sync::CancellationToken;
 use tracing::*;
+use utils::sync::gate::{Gate, GateGuard};
 use utils::sync::spsc_fold;
 use utils::{
     auth::{Claims, Scope, SwappableJwtAuth},
@@ -61,6 +62,7 @@ use crate::task_mgr::{self, COMPUTE_REQUEST_RUNTIME};
 use crate::tenant::mgr::ShardSelector;
 use crate::tenant::mgr::TenantManager;
 use crate::tenant::mgr::{GetActiveTenantError, GetTenantError, ShardResolveResult};
+use crate::tenant::storage_layer::IoConcurrency;
 use crate::tenant::timeline::{self, WaitLsnError};
 use crate::tenant::GetTimelineError;
 use crate::tenant::PageReconstructError;
@@ -90,6 +92,7 @@ pub struct Listener {
 pub struct Connections {
     cancel: CancellationToken,
     tasks: tokio::task::JoinSet<ConnectionHandlerResult>,
+    gate: Gate,
 }
 
 pub fn spawn(
@@ -110,6 +113,7 @@ pub fn spawn(
     let task = COMPUTE_REQUEST_RUNTIME.spawn(task_mgr::exit_on_panic_or_error(
         "libpq listener",
         libpq_listener_main(
+            conf,
             tenant_manager,
             pg_auth,
             tcp_listener,
@@ -134,11 +138,16 @@ impl Listener {
 }
 impl Connections {
     pub(crate) async fn shutdown(self) {
-        let Self { cancel, mut tasks } = self;
+        let Self {
+            cancel,
+            mut tasks,
+            gate,
+        } = self;
         cancel.cancel();
         while let Some(res) = tasks.join_next().await {
             Self::handle_connection_completion(res);
         }
+        gate.close().await;
     }
 
     fn handle_connection_completion(res: Result<anyhow::Result<()>, tokio::task::JoinError>) {
@@ -158,7 +167,9 @@ impl Connections {
 /// Returns Ok(()) upon cancellation via `cancel`, returning the set of
 /// open connections.
 ///
+#[allow(clippy::too_many_arguments)]
 pub async fn libpq_listener_main(
+    conf: &'static PageServerConf,
     tenant_manager: Arc<TenantManager>,
     auth: Option<Arc<SwappableJwtAuth>>,
     listener: tokio::net::TcpListener,
@@ -168,9 +179,15 @@ pub async fn libpq_listener_main(
     listener_cancel: CancellationToken,
 ) -> Connections {
     let connections_cancel = CancellationToken::new();
+    let connections_gate = Gate::default();
     let mut connection_handler_tasks = tokio::task::JoinSet::default();
 
     loop {
+        let gate_guard = match connections_gate.enter() {
+            Ok(guard) => guard,
+            Err(_) => break,
+        };
+
         let accepted = tokio::select! {
             biased;
             _ = listener_cancel.cancelled() => break,
@@ -190,6 +207,7 @@ pub async fn libpq_listener_main(
                 let connection_ctx = listener_ctx
                     .detached_child(TaskKind::PageRequestHandler, DownloadBehavior::Download);
                 connection_handler_tasks.spawn(page_service_conn_main(
+                    conf,
                     tenant_manager.clone(),
                     local_auth,
                     socket,
@@ -197,6 +215,7 @@ pub async fn libpq_listener_main(
                     pipelining_config.clone(),
                     connection_ctx,
                     connections_cancel.child_token(),
+                    gate_guard,
                 ));
             }
             Err(err) => {
@@ -211,13 +230,16 @@ pub async fn libpq_listener_main(
     Connections {
         cancel: connections_cancel,
         tasks: connection_handler_tasks,
+        gate: connections_gate,
     }
 }
 
 type ConnectionHandlerResult = anyhow::Result<()>;
 
 #[instrument(skip_all, fields(peer_addr))]
+#[allow(clippy::too_many_arguments)]
 async fn page_service_conn_main(
+    conf: &'static PageServerConf,
     tenant_manager: Arc<TenantManager>,
     auth: Option<Arc<SwappableJwtAuth>>,
     socket: tokio::net::TcpStream,
@@ -225,6 +247,7 @@ async fn page_service_conn_main(
     pipelining_config: PageServicePipeliningConfig,
     connection_ctx: RequestContext,
     cancel: CancellationToken,
+    gate_guard: GateGuard,
 ) -> ConnectionHandlerResult {
     let _guard = LIVE_CONNECTIONS
         .with_label_values(&["page_service"])
@@ -274,11 +297,13 @@ async fn page_service_conn_main(
     // But it's in a shared crate, so, we store connection_ctx inside PageServerHandler
     // and create the per-query context in process_query ourselves.
     let mut conn_handler = PageServerHandler::new(
+        conf,
         tenant_manager,
         auth,
         pipelining_config,
         connection_ctx,
         cancel.clone(),
+        gate_guard,
     );
     let pgbackend = PostgresBackend::new_from_io(socket, peer_addr, auth_type, None)?;
 
@@ -310,6 +335,7 @@ async fn page_service_conn_main(
 }
 
 struct PageServerHandler {
+    conf: &'static PageServerConf,
     auth: Option<Arc<SwappableJwtAuth>>,
     claims: Option<Claims>,
 
@@ -325,6 +351,8 @@ struct PageServerHandler {
     timeline_handles: Option<TimelineHandles>,
 
     pipelining_config: PageServicePipeliningConfig,
+
+    gate_guard: GateGuard,
 }
 
 struct TimelineHandles {
@@ -555,37 +583,52 @@ struct BatchedGetPageRequest {
     timer: SmgrOpTimer,
 }
 
+#[cfg(feature = "testing")]
+struct BatchedTestRequest {
+    req: models::PagestreamTestRequest,
+    timer: SmgrOpTimer,
+}
+
+/// NB: we only hold [`timeline::handle::WeakHandle`] inside this enum,
+/// so that we don't keep the [`Timeline::gate`] open while the batch
+/// is being built up inside the [`spsc_fold`] (pagestream pipelining).
 enum BatchedFeMessage {
     Exists {
         span: Span,
         timer: SmgrOpTimer,
-        shard: timeline::handle::Handle<TenantManagerTypes>,
+        shard: timeline::handle::WeakHandle<TenantManagerTypes>,
         req: models::PagestreamExistsRequest,
     },
     Nblocks {
         span: Span,
         timer: SmgrOpTimer,
-        shard: timeline::handle::Handle<TenantManagerTypes>,
+        shard: timeline::handle::WeakHandle<TenantManagerTypes>,
         req: models::PagestreamNblocksRequest,
     },
     GetPage {
         span: Span,
-        shard: timeline::handle::Handle<TenantManagerTypes>,
+        shard: timeline::handle::WeakHandle<TenantManagerTypes>,
         effective_request_lsn: Lsn,
         pages: smallvec::SmallVec<[BatchedGetPageRequest; 1]>,
     },
     DbSize {
         span: Span,
         timer: SmgrOpTimer,
-        shard: timeline::handle::Handle<TenantManagerTypes>,
+        shard: timeline::handle::WeakHandle<TenantManagerTypes>,
         req: models::PagestreamDbSizeRequest,
     },
     GetSlruSegment {
         span: Span,
         timer: SmgrOpTimer,
-        shard: timeline::handle::Handle<TenantManagerTypes>,
+        shard: timeline::handle::WeakHandle<TenantManagerTypes>,
         req: models::PagestreamGetSlruSegmentRequest,
     },
+    #[cfg(feature = "testing")]
+    Test {
+        span: Span,
+        shard: timeline::handle::WeakHandle<TenantManagerTypes>,
+        requests: Vec<BatchedTestRequest>,
+    },
     RespondError {
         span: Span,
         error: BatchedPageStreamError,
@@ -606,6 +649,12 @@ impl BatchedFeMessage {
                     page.timer.observe_execution_start(at);
                 }
             }
+            #[cfg(feature = "testing")]
+            BatchedFeMessage::Test { requests, .. } => {
+                for req in requests {
+                    req.timer.observe_execution_start(at);
+                }
+            }
             BatchedFeMessage::RespondError { .. } => {}
         }
     }
@@ -613,19 +662,23 @@ impl BatchedFeMessage {
 
 impl PageServerHandler {
     pub fn new(
+        conf: &'static PageServerConf,
         tenant_manager: Arc<TenantManager>,
         auth: Option<Arc<SwappableJwtAuth>>,
         pipelining_config: PageServicePipeliningConfig,
         connection_ctx: RequestContext,
         cancel: CancellationToken,
+        gate_guard: GateGuard,
     ) -> Self {
         PageServerHandler {
+            conf,
             auth,
             claims: None,
             connection_ctx,
             timeline_handles: Some(TimelineHandles::new(tenant_manager)),
             cancel,
             pipelining_config,
+            gate_guard,
         }
     }
 
@@ -735,7 +788,7 @@ impl PageServerHandler {
                 BatchedFeMessage::Exists {
                     span,
                     timer,
-                    shard,
+                    shard: shard.downgrade(),
                     req,
                 }
             }
@@ -754,7 +807,7 @@ impl PageServerHandler {
                 BatchedFeMessage::Nblocks {
                     span,
                     timer,
-                    shard,
+                    shard: shard.downgrade(),
                     req,
                 }
             }
@@ -773,7 +826,7 @@ impl PageServerHandler {
                 BatchedFeMessage::DbSize {
                     span,
                     timer,
-                    shard,
+                    shard: shard.downgrade(),
                     req,
                 }
             }
@@ -792,7 +845,7 @@ impl PageServerHandler {
                 BatchedFeMessage::GetSlruSegment {
                     span,
                     timer,
-                    shard,
+                    shard: shard.downgrade(),
                     req,
                 }
             }
@@ -844,6 +897,7 @@ impl PageServerHandler {
                 )
                 .await?;
 
+                // We're holding the Handle
                 let effective_request_lsn = match Self::wait_or_get_last_lsn(
                     &shard,
                     req.hdr.request_lsn,
@@ -861,11 +915,27 @@ impl PageServerHandler {
                 };
                 BatchedFeMessage::GetPage {
                     span,
-                    shard,
+                    shard: shard.downgrade(),
                     effective_request_lsn,
                     pages: smallvec::smallvec![BatchedGetPageRequest { req, timer }],
                 }
             }
+            #[cfg(feature = "testing")]
+            PagestreamFeMessage::Test(req) => {
+                let span = tracing::info_span!(parent: parent_span, "handle_test_request");
+                let shard = timeline_handles
+                    .get(tenant_id, timeline_id, ShardSelector::Zero)
+                    .instrument(span.clone()) // sets `shard_id` field
+                    .await?;
+                let timer =
+                    record_op_start_and_throttle(&shard, metrics::SmgrQueryType::Test, received_at)
+                        .await?;
+                BatchedFeMessage::Test {
+                    span,
+                    shard: shard.downgrade(),
+                    requests: vec![BatchedTestRequest { req, timer }],
+                }
+            }
         };
         Ok(Some(batched_msg))
     }
@@ -907,9 +977,7 @@ impl PageServerHandler {
                     assert_eq!(accum_pages.len(), max_batch_size.get());
                     return false;
                 }
-                if (accum_shard.tenant_shard_id, accum_shard.timeline_id)
-                    != (this_shard.tenant_shard_id, this_shard.timeline_id)
-                {
+                if !accum_shard.is_same_handle_as(&this_shard) {
                     trace!(%accum_lsn, %this_lsn, "stopping batching because timeline object mismatch");
                     // TODO: we _could_ batch & execute each shard seperately (and in parallel).
                     // But the current logic for keeping responses in order does not support that.
@@ -928,6 +996,44 @@ impl PageServerHandler {
                 accum_pages.extend(this_pages);
                 Ok(())
             }
+            #[cfg(feature = "testing")]
+            (
+                Ok(BatchedFeMessage::Test {
+                    shard: accum_shard,
+                    requests: accum_requests,
+                    ..
+                }),
+                BatchedFeMessage::Test {
+                    shard: this_shard,
+                    requests: this_requests,
+                    ..
+                },
+            ) if (|| {
+                assert!(this_requests.len() == 1);
+                if accum_requests.len() >= max_batch_size.get() {
+                    trace!(%max_batch_size, "stopping batching because of batch size");
+                    assert_eq!(accum_requests.len(), max_batch_size.get());
+                    return false;
+                }
+                if !accum_shard.is_same_handle_as(&this_shard) {
+                    trace!("stopping batching because timeline object mismatch");
+                    // TODO: we _could_ batch & execute each shard seperately (and in parallel).
+                    // But the current logic for keeping responses in order does not support that.
+                    return false;
+                }
+                let this_batch_key = this_requests[0].req.batch_key;
+                let accum_batch_key = accum_requests[0].req.batch_key;
+                if this_requests[0].req.batch_key != accum_requests[0].req.batch_key {
+                    trace!(%accum_batch_key, %this_batch_key, "stopping batching because batch key changed");
+                    return false;
+                }
+                true
+            })() =>
+            {
+                // ok to batch
+                accum_requests.extend(this_requests);
+                Ok(())
+            }
             // something batched already but this message is unbatchable
             (_, this_msg) => {
                 // by default, don't continue batching
@@ -941,6 +1047,7 @@ impl PageServerHandler {
         &mut self,
         pgb_writer: &mut PostgresBackend<IO>,
         batch: BatchedFeMessage,
+        io_concurrency: IoConcurrency,
         cancel: &CancellationToken,
         protocol_version: PagestreamProtocolVersion,
         ctx: &RequestContext,
@@ -969,7 +1076,7 @@ impl PageServerHandler {
                 fail::fail_point!("ps::handle-pagerequest-message::exists");
                 (
                     vec![self
-                        .handle_get_rel_exists_request(&shard, &req, ctx)
+                        .handle_get_rel_exists_request(&*shard.upgrade()?, &req, ctx)
                         .instrument(span.clone())
                         .await
                         .map(|msg| (msg, timer))
@@ -986,7 +1093,7 @@ impl PageServerHandler {
                 fail::fail_point!("ps::handle-pagerequest-message::nblocks");
                 (
                     vec![self
-                        .handle_get_nblocks_request(&shard, &req, ctx)
+                        .handle_get_nblocks_request(&*shard.upgrade()?, &req, ctx)
                         .instrument(span.clone())
                         .await
                         .map(|msg| (msg, timer))
@@ -1007,9 +1114,10 @@ impl PageServerHandler {
                         trace!(npages, "handling getpage request");
                         let res = self
                             .handle_get_page_at_lsn_request_batched(
-                                &shard,
+                                &*shard.upgrade()?,
                                 effective_request_lsn,
                                 pages,
+                                io_concurrency,
                                 ctx,
                             )
                             .instrument(span.clone())
@@ -1029,7 +1137,7 @@ impl PageServerHandler {
                 fail::fail_point!("ps::handle-pagerequest-message::dbsize");
                 (
                     vec![self
-                        .handle_db_size_request(&shard, &req, ctx)
+                        .handle_db_size_request(&*shard.upgrade()?, &req, ctx)
                         .instrument(span.clone())
                         .await
                         .map(|msg| (msg, timer))
@@ -1046,7 +1154,7 @@ impl PageServerHandler {
                 fail::fail_point!("ps::handle-pagerequest-message::slrusegment");
                 (
                     vec![self
-                        .handle_get_slru_segment_request(&shard, &req, ctx)
+                        .handle_get_slru_segment_request(&*shard.upgrade()?, &req, ctx)
                         .instrument(span.clone())
                         .await
                         .map(|msg| (msg, timer))
@@ -1054,6 +1162,27 @@ impl PageServerHandler {
                     span,
                 )
             }
+            #[cfg(feature = "testing")]
+            BatchedFeMessage::Test {
+                span,
+                shard,
+                requests,
+            } => {
+                fail::fail_point!("ps::handle-pagerequest-message::test");
+                (
+                    {
+                        let npages = requests.len();
+                        trace!(npages, "handling getpage request");
+                        let res = self
+                            .handle_test_request_batch(&*shard.upgrade()?, requests, ctx)
+                            .instrument(span.clone())
+                            .await;
+                        assert_eq!(res.len(), npages);
+                        res
+                    },
+                    span,
+                )
+            }
             BatchedFeMessage::RespondError { span, error } => {
                 // We've already decided to respond with an error, so we don't need to
                 // call the handler.
@@ -1193,6 +1322,17 @@ impl PageServerHandler {
             }
         }
 
+        let io_concurrency = IoConcurrency::spawn_from_conf(
+            self.conf,
+            match self.gate_guard.try_clone() {
+                Ok(guard) => guard,
+                Err(_) => {
+                    info!("shutdown request received in page handler");
+                    return Err(QueryError::Shutdown);
+                }
+            },
+        );
+
         let pgb_reader = pgb
             .split()
             .context("implementation error: split pgb into reader and writer")?;
@@ -1214,6 +1354,7 @@ impl PageServerHandler {
                     request_span,
                     pipelining_config,
                     protocol_version,
+                    io_concurrency,
                     &ctx,
                 )
                 .await
@@ -1227,6 +1368,7 @@ impl PageServerHandler {
                     timeline_handles,
                     request_span,
                     protocol_version,
+                    io_concurrency,
                     &ctx,
                 )
                 .await
@@ -1254,6 +1396,7 @@ impl PageServerHandler {
         mut timeline_handles: TimelineHandles,
         request_span: Span,
         protocol_version: PagestreamProtocolVersion,
+        io_concurrency: IoConcurrency,
         ctx: &RequestContext,
     ) -> (
         (PostgresBackendReader<IO>, TimelineHandles),
@@ -1288,7 +1431,14 @@ impl PageServerHandler {
             };
 
             let err = self
-                .pagesteam_handle_batched_message(pgb_writer, msg, &cancel, protocol_version, ctx)
+                .pagesteam_handle_batched_message(
+                    pgb_writer,
+                    msg,
+                    io_concurrency.clone(),
+                    &cancel,
+                    protocol_version,
+                    ctx,
+                )
                 .await;
             match err {
                 Ok(()) => {}
@@ -1312,6 +1462,7 @@ impl PageServerHandler {
         request_span: Span,
         pipelining_config: PageServicePipeliningConfigPipelined,
         protocol_version: PagestreamProtocolVersion,
+        io_concurrency: IoConcurrency,
         ctx: &RequestContext,
     ) -> (
         (PostgresBackendReader<IO>, TimelineHandles),
@@ -1455,6 +1606,7 @@ impl PageServerHandler {
                     self.pagesteam_handle_batched_message(
                         pgb_writer,
                         batch,
+                        io_concurrency.clone(),
                         &cancel,
                         protocol_version,
                         &ctx,
@@ -1556,6 +1708,7 @@ impl PageServerHandler {
                 .wait_lsn(
                     not_modified_since,
                     crate::tenant::timeline::WaitLsnWaiter::PageService,
+                    timeline::WaitLsnTimeout::Default,
                     ctx,
                 )
                 .await?;
@@ -1711,6 +1864,7 @@ impl PageServerHandler {
         timeline: &Timeline,
         effective_lsn: Lsn,
         requests: smallvec::SmallVec<[BatchedGetPageRequest; 1]>,
+        io_concurrency: IoConcurrency,
         ctx: &RequestContext,
     ) -> Vec<Result<(PagestreamBeMessage, SmgrOpTimer), BatchedPageStreamError>> {
         debug_assert_current_span_has_tenant_and_timeline_id();
@@ -1737,6 +1891,7 @@ impl PageServerHandler {
             .get_rel_page_at_lsn_batched(
                 requests.iter().map(|p| (&p.req.rel, &p.req.blkno)),
                 effective_lsn,
+                io_concurrency,
                 ctx,
             )
             .await;
@@ -1791,6 +1946,51 @@ impl PageServerHandler {
         ))
     }
 
+    // NB: this impl mimics what we do for batched getpage requests.
+    #[cfg(feature = "testing")]
+    #[instrument(skip_all, fields(shard_id))]
+    async fn handle_test_request_batch(
+        &mut self,
+        timeline: &Timeline,
+        requests: Vec<BatchedTestRequest>,
+        _ctx: &RequestContext,
+    ) -> Vec<Result<(PagestreamBeMessage, SmgrOpTimer), BatchedPageStreamError>> {
+        // real requests would do something with the timeline
+        let mut results = Vec::with_capacity(requests.len());
+        for _req in requests.iter() {
+            tokio::task::yield_now().await;
+
+            results.push({
+                if timeline.cancel.is_cancelled() {
+                    Err(PageReconstructError::Cancelled)
+                } else {
+                    Ok(())
+                }
+            });
+        }
+
+        // TODO: avoid creating the new Vec here
+        Vec::from_iter(
+            requests
+                .into_iter()
+                .zip(results.into_iter())
+                .map(|(req, res)| {
+                    res.map(|()| {
+                        (
+                            PagestreamBeMessage::Test(models::PagestreamTestResponse {
+                                req: req.req.clone(),
+                            }),
+                            req.timer,
+                        )
+                    })
+                    .map_err(|e| BatchedPageStreamError {
+                        err: PageStreamError::from(e),
+                        req: req.req.hdr,
+                    })
+                }),
+        )
+    }
+
     /// Note on "fullbackup":
     /// Full basebackups should only be used for debugging purposes.
     /// Originally, it was introduced to enable breaking storage format changes,
@@ -1845,6 +2045,7 @@ impl PageServerHandler {
                 .wait_lsn(
                     lsn,
                     crate::tenant::timeline::WaitLsnWaiter::PageService,
+                    crate::tenant::timeline::WaitLsnTimeout::Default,
                     ctx,
                 )
                 .await?;
@@ -2406,6 +2607,14 @@ impl From<GetActiveTimelineError> for QueryError {
     }
 }
 
+impl From<crate::tenant::timeline::handle::HandleUpgradeError> for QueryError {
+    fn from(e: crate::tenant::timeline::handle::HandleUpgradeError) -> Self {
+        match e {
+            crate::tenant::timeline::handle::HandleUpgradeError::ShutDown => QueryError::Shutdown,
+        }
+    }
+}
+
 fn set_tracing_field_shard_id(timeline: &Timeline) {
     debug_assert_current_span_has_tenant_and_timeline_id_no_shard_id();
     tracing::Span::current().record(

pageserver/src/pgdatadir_mapping.rs

@@ -17,19 +17,21 @@ use crate::span::{
     debug_assert_current_span_has_tenant_and_timeline_id,
     debug_assert_current_span_has_tenant_and_timeline_id_no_shard_id,
 };
+use crate::tenant::storage_layer::{IoConcurrency, ValuesReconstructState};
 use crate::tenant::timeline::GetVectoredError;
 use anyhow::{ensure, Context};
 use bytes::{Buf, Bytes, BytesMut};
 use enum_map::Enum;
 use itertools::Itertools;
-use pageserver_api::key::Key;
 use pageserver_api::key::{
     dbdir_key_range, rel_block_to_key, rel_dir_to_key, rel_key_range, rel_size_to_key,
     relmap_file_key, repl_origin_key, repl_origin_key_range, slru_block_to_key, slru_dir_to_key,
-    slru_segment_key_range, slru_segment_size_to_key, twophase_file_key, twophase_key_range,
-    CompactKey, AUX_FILES_KEY, CHECKPOINT_KEY, CONTROLFILE_KEY, DBDIR_KEY, TWOPHASEDIR_KEY,
+    slru_segment_key_range, slru_segment_size_key_to_segno, slru_segment_size_to_key,
+    twophase_file_key, twophase_key_range, CompactKey, AUX_FILES_KEY, CHECKPOINT_KEY,
+    CONTROLFILE_KEY, DBDIR_KEY, TWOPHASEDIR_KEY,
 };
-use pageserver_api::keyspace::SparseKeySpace;
+use pageserver_api::key::{rel_size_key_to_rel, Key};
+use pageserver_api::keyspace::{KeySpaceRandomAccum, SparseKeySpace};
 use pageserver_api::record::NeonWalRecord;
 use pageserver_api::reltag::{BlockNumber, RelTag, SlruKind};
 use pageserver_api::shard::ShardIdentity;
@@ -109,10 +111,21 @@ pub(crate) enum CollectKeySpaceError {
     Decode(#[from] DeserializeError),
     #[error(transparent)]
     PageRead(PageReconstructError),
+    #[error(transparent)]
+    GetVectored(GetVectoredError),
     #[error("cancelled")]
     Cancelled,
 }
 
+impl From<GetVectoredError> for CollectKeySpaceError {
+    fn from(err: GetVectoredError) -> Self {
+        match err {
+            GetVectoredError::Cancelled => Self::Cancelled,
+            err => Self::GetVectored(err),
+        }
+    }
+}
+
 impl From<PageReconstructError> for CollectKeySpaceError {
     fn from(err: PageReconstructError) -> Self {
         match err {
@@ -200,6 +213,7 @@ impl Timeline {
         blknum: BlockNumber,
         version: Version<'_>,
         ctx: &RequestContext,
+        io_concurrency: IoConcurrency,
     ) -> Result<Bytes, PageReconstructError> {
         match version {
             Version::Lsn(effective_lsn) => {
@@ -208,6 +222,7 @@ impl Timeline {
                     .get_rel_page_at_lsn_batched(
                         pages.iter().map(|(tag, blknum)| (tag, blknum)),
                         effective_lsn,
+                        io_concurrency.clone(),
                         ctx,
                     )
                     .await;
@@ -246,6 +261,7 @@ impl Timeline {
         &self,
         pages: impl ExactSizeIterator<Item = (&RelTag, &BlockNumber)>,
         effective_lsn: Lsn,
+        io_concurrency: IoConcurrency,
         ctx: &RequestContext,
     ) -> Vec<Result<Bytes, PageReconstructError>> {
         debug_assert_current_span_has_tenant_and_timeline_id();
@@ -309,7 +325,10 @@ impl Timeline {
             acc.to_keyspace()
         };
 
-        match self.get_vectored(keyspace, effective_lsn, ctx).await {
+        match self
+            .get_vectored(keyspace, effective_lsn, io_concurrency, ctx)
+            .await
+        {
             Ok(results) => {
                 for (key, res) in results {
                     let mut key_slots = keys_slots.remove(&key).unwrap().into_iter();
@@ -889,9 +908,15 @@ impl Timeline {
         &self,
         lsn: Lsn,
         ctx: &RequestContext,
+        io_concurrency: IoConcurrency,
     ) -> Result<HashMap<String, Bytes>, PageReconstructError> {
         let kv = self
-            .scan(KeySpace::single(Key::metadata_aux_key_range()), lsn, ctx)
+            .scan(
+                KeySpace::single(Key::metadata_aux_key_range()),
+                lsn,
+                ctx,
+                io_concurrency,
+            )
             .await?;
         let mut result = HashMap::new();
         let mut sz = 0;
@@ -914,8 +939,9 @@ impl Timeline {
         &self,
         lsn: Lsn,
         ctx: &RequestContext,
+        io_concurrency: IoConcurrency,
     ) -> Result<(), PageReconstructError> {
-        self.list_aux_files_v2(lsn, ctx).await?;
+        self.list_aux_files_v2(lsn, ctx, io_concurrency).await?;
         Ok(())
     }
 
@@ -923,17 +949,24 @@ impl Timeline {
         &self,
         lsn: Lsn,
         ctx: &RequestContext,
+        io_concurrency: IoConcurrency,
     ) -> Result<HashMap<String, Bytes>, PageReconstructError> {
-        self.list_aux_files_v2(lsn, ctx).await
+        self.list_aux_files_v2(lsn, ctx, io_concurrency).await
     }
 
     pub(crate) async fn get_replorigins(
         &self,
         lsn: Lsn,
         ctx: &RequestContext,
+        io_concurrency: IoConcurrency,
     ) -> Result<HashMap<RepOriginId, Lsn>, PageReconstructError> {
         let kv = self
-            .scan(KeySpace::single(repl_origin_key_range()), lsn, ctx)
+            .scan(
+                KeySpace::single(repl_origin_key_range()),
+                lsn,
+                ctx,
+                io_concurrency,
+            )
             .await?;
         let mut result = HashMap::new();
         for (k, v) in kv {
@@ -1050,11 +1083,30 @@ impl Timeline {
                 .into_iter()
                 .collect();
             rels.sort_unstable();
+            let mut relsize_keys_to_collect = KeySpaceRandomAccum::new();
             for rel in rels {
                 let relsize_key = rel_size_to_key(rel);
-                let mut buf = self.get(relsize_key, lsn, ctx).await?;
+                relsize_keys_to_collect.add_key(relsize_key);
+            }
+            // Skip the vectored-read max key check by using `get_vectored_impl`.
+            let io_concurrency = IoConcurrency::spawn_from_conf(
+                self.conf,
+                self.gate
+                    .enter()
+                    .map_err(|_| CollectKeySpaceError::Cancelled)?,
+            );
+            let res = self
+                .get_vectored_impl(
+                    relsize_keys_to_collect.consume_keyspace(),
+                    lsn,
+                    &mut ValuesReconstructState::new(io_concurrency),
+                    ctx,
+                )
+                .await?;
+            for (relsize_key, buf) in res {
+                let mut buf = buf?;
                 let relsize = buf.get_u32_le();
-
+                let rel = rel_size_key_to_rel(relsize_key);
                 result.add_range(rel_block_to_key(rel, 0)..rel_block_to_key(rel, relsize));
                 result.add_key(relsize_key);
             }
@@ -1073,11 +1125,30 @@ impl Timeline {
                 let dir = SlruSegmentDirectory::des(&buf)?;
                 let mut segments: Vec<u32> = dir.segments.iter().cloned().collect();
                 segments.sort_unstable();
+                let mut segsize_keys_to_collect = KeySpaceRandomAccum::new();
                 for segno in segments {
                     let segsize_key = slru_segment_size_to_key(kind, segno);
-                    let mut buf = self.get(segsize_key, lsn, ctx).await?;
+                    segsize_keys_to_collect.add_key(segsize_key);
+                }
+                // Skip the vectored-read max key check by using `get_vectored_impl`.
+                let io_concurrency = IoConcurrency::spawn_from_conf(
+                    self.conf,
+                    self.gate
+                        .enter()
+                        .map_err(|_| CollectKeySpaceError::Cancelled)?,
+                );
+                let res = self
+                    .get_vectored_impl(
+                        segsize_keys_to_collect.consume_keyspace(),
+                        lsn,
+                        &mut ValuesReconstructState::new(io_concurrency),
+                        ctx,
+                    )
+                    .await?;
+                for (segsize_key, buf) in res {
+                    let mut buf = buf?;
                     let segsize = buf.get_u32_le();
-
+                    let segno = slru_segment_size_key_to_segno(segsize_key);
                     result.add_range(
                         slru_block_to_key(kind, segno, 0)..slru_block_to_key(kind, segno, segsize),
                     );
@@ -2432,7 +2503,11 @@ mod tests {
             ("foo/bar2".to_string(), Bytes::from_static(b"content2")),
         ]);
 
-        let readback = tline.list_aux_files(Lsn(0x1008), &ctx).await?;
+        let io_concurrency = IoConcurrency::spawn_for_test();
+
+        let readback = tline
+            .list_aux_files(Lsn(0x1008), &ctx, io_concurrency.clone())
+            .await?;
         assert_eq!(readback, expect_1008);
 
         // Second modification: update one key, remove the other
@@ -2444,11 +2519,15 @@ mod tests {
         let expect_2008 =
             HashMap::from([("foo/bar1".to_string(), Bytes::from_static(b"content3"))]);
 
-        let readback = tline.list_aux_files(Lsn(0x2008), &ctx).await?;
+        let readback = tline
+            .list_aux_files(Lsn(0x2008), &ctx, io_concurrency.clone())
+            .await?;
         assert_eq!(readback, expect_2008);
 
         // Reading back in time works
-        let readback = tline.list_aux_files(Lsn(0x1008), &ctx).await?;
+        let readback = tline
+            .list_aux_files(Lsn(0x1008), &ctx, io_concurrency.clone())
+            .await?;
         assert_eq!(readback, expect_1008);
 
         Ok(())

pageserver/src/tenant.rs

@@ -37,6 +37,8 @@ use remote_timeline_client::manifest::{
     OffloadedTimelineManifest, TenantManifest, LATEST_TENANT_MANIFEST_VERSION,
 };
 use remote_timeline_client::UploadQueueNotReadyError;
+use remote_timeline_client::FAILED_REMOTE_OP_RETRIES;
+use remote_timeline_client::FAILED_UPLOAD_WARN_THRESHOLD;
 use std::collections::BTreeMap;
 use std::fmt;
 use std::future::Future;
@@ -95,6 +97,9 @@ use crate::deletion_queue::DeletionQueueError;
 use crate::import_datadir;
 use crate::is_uninit_mark;
 use crate::l0_flush::L0FlushGlobalState;
+use crate::metrics::CONCURRENT_INITDBS;
+use crate::metrics::INITDB_RUN_TIME;
+use crate::metrics::INITDB_SEMAPHORE_ACQUISITION_TIME;
 use crate::metrics::TENANT;
 use crate::metrics::{
     remove_tenant_metrics, BROKEN_TENANTS_SET, CIRCUIT_BREAKERS_BROKEN, CIRCUIT_BREAKERS_UNBROKEN,
@@ -2555,7 +2560,12 @@ impl Tenant {
                     // sizes etc. and that would get confused if the previous page versions
                     // are not in the repository yet.
                     ancestor_timeline
-                        .wait_lsn(*lsn, timeline::WaitLsnWaiter::Tenant, ctx)
+                        .wait_lsn(
+                            *lsn,
+                            timeline::WaitLsnWaiter::Tenant,
+                            timeline::WaitLsnTimeout::Default,
+                            ctx,
+                        )
                         .await
                         .map_err(|e| match e {
                             e @ (WaitLsnError::Timeout(_) | WaitLsnError::BadState { .. }) => {
@@ -5305,27 +5315,37 @@ impl Tenant {
             return Ok(());
         }
 
-        upload_tenant_manifest(
-            &self.remote_storage,
-            &self.tenant_shard_id,
-            self.generation,
-            &manifest,
+        // Remote storage does no retries internally, so wrap it
+        match backoff::retry(
+            || async {
+                upload_tenant_manifest(
+                    &self.remote_storage,
+                    &self.tenant_shard_id,
+                    self.generation,
+                    &manifest,
+                    &self.cancel,
+                )
+                .await
+            },
+            |_e| self.cancel.is_cancelled(),
+            FAILED_UPLOAD_WARN_THRESHOLD,
+            FAILED_REMOTE_OP_RETRIES,
+            "uploading tenant manifest",
             &self.cancel,
         )
         .await
-        .map_err(|e| {
-            if self.cancel.is_cancelled() {
-                TenantManifestError::Cancelled
-            } else {
-                TenantManifestError::RemoteStorage(e)
+        {
+            None => Err(TenantManifestError::Cancelled),
+            Some(Err(_)) if self.cancel.is_cancelled() => Err(TenantManifestError::Cancelled),
+            Some(Err(e)) => Err(TenantManifestError::RemoteStorage(e)),
+            Some(Ok(_)) => {
+                // Store the successfully uploaded manifest, so that future callers can avoid
+                // re-uploading the same thing.
+                *guard = Some(manifest);
+
+                Ok(())
             }
-        })?;
-
-        // Store the successfully uploaded manifest, so that future callers can avoid
-        // re-uploading the same thing.
-        *guard = Some(manifest);
-
-        Ok(())
+        }
     }
 }
 
@@ -5347,8 +5367,17 @@ async fn run_initdb(
         initdb_bin_path, initdb_target_dir, initdb_lib_dir,
     );
 
-    let _permit = INIT_DB_SEMAPHORE.acquire().await;
+    let _permit = {
+        let _timer = INITDB_SEMAPHORE_ACQUISITION_TIME.start_timer();
+        INIT_DB_SEMAPHORE.acquire().await
+    };
 
+    CONCURRENT_INITDBS.inc();
+    scopeguard::defer! {
+        CONCURRENT_INITDBS.dec();
+    }
+
+    let _timer = INITDB_RUN_TIME.start_timer();
     let res = postgres_initdb::do_run_initdb(postgres_initdb::RunInitdbArgs {
         superuser: &conf.superuser,
         locale: &conf.locale,
@@ -5441,6 +5470,9 @@ pub(crate) mod harness {
                 compaction_period: Some(tenant_conf.compaction_period),
                 compaction_threshold: Some(tenant_conf.compaction_threshold),
                 compaction_algorithm: Some(tenant_conf.compaction_algorithm),
+                l0_flush_delay_threshold: tenant_conf.l0_flush_delay_threshold,
+                l0_flush_stall_threshold: tenant_conf.l0_flush_stall_threshold,
+                l0_flush_wait_upload: Some(tenant_conf.l0_flush_wait_upload),
                 gc_horizon: Some(tenant_conf.gc_horizon),
                 gc_period: Some(tenant_conf.gc_period),
                 image_creation_threshold: Some(tenant_conf.image_creation_threshold),
@@ -5463,6 +5495,12 @@ pub(crate) mod harness {
                 lsn_lease_length_for_ts: Some(tenant_conf.lsn_lease_length_for_ts),
                 timeline_offloading: Some(tenant_conf.timeline_offloading),
                 wal_receiver_protocol_override: tenant_conf.wal_receiver_protocol_override,
+                rel_size_v2_enabled: tenant_conf.rel_size_v2_enabled,
+                gc_compaction_enabled: Some(tenant_conf.gc_compaction_enabled),
+                gc_compaction_initial_threshold_kb: Some(
+                    tenant_conf.gc_compaction_initial_threshold_kb,
+                ),
+                gc_compaction_ratio_percent: Some(tenant_conf.gc_compaction_ratio_percent),
             }
         }
     }
@@ -5696,7 +5734,7 @@ mod tests {
     use pageserver_api::value::Value;
     use pageserver_compaction::helpers::overlaps_with;
     use rand::{thread_rng, Rng};
-    use storage_layer::PersistentLayerKey;
+    use storage_layer::{IoConcurrency, PersistentLayerKey};
     use tests::storage_layer::ValuesReconstructState;
     use tests::timeline::{GetVectoredError, ShutdownMode};
     use timeline::{CompactOptions, DeltaLayerTestDesc};
@@ -6477,6 +6515,7 @@ mod tests {
     async fn test_get_vectored() -> anyhow::Result<()> {
         let harness = TenantHarness::create("test_get_vectored").await?;
         let (tenant, ctx) = harness.load().await;
+        let io_concurrency = IoConcurrency::spawn_for_test();
         let tline = tenant
             .create_test_timeline(TIMELINE_ID, Lsn(0x08), DEFAULT_PG_VERSION, &ctx)
             .await?;
@@ -6541,7 +6580,7 @@ mod tests {
                 .get_vectored_impl(
                     read.clone(),
                     reads_lsn,
-                    &mut ValuesReconstructState::new(),
+                    &mut ValuesReconstructState::new(io_concurrency.clone()),
                     &ctx,
                 )
                 .await;
@@ -6588,6 +6627,7 @@ mod tests {
         let harness = TenantHarness::create("test_get_vectored_aux_files").await?;
 
         let (tenant, ctx) = harness.load().await;
+        let io_concurrency = IoConcurrency::spawn_for_test();
         let tline = tenant
             .create_empty_timeline(TIMELINE_ID, Lsn(0), DEFAULT_PG_VERSION, &ctx)
             .await?;
@@ -6622,7 +6662,7 @@ mod tests {
             .get_vectored_impl(
                 aux_keyspace.clone(),
                 read_lsn,
-                &mut ValuesReconstructState::new(),
+                &mut ValuesReconstructState::new(io_concurrency.clone()),
                 &ctx,
             )
             .await;
@@ -6670,6 +6710,7 @@ mod tests {
         )
         .await?;
         let (tenant, ctx) = harness.load().await;
+        let io_concurrency = IoConcurrency::spawn_for_test();
 
         let mut current_key = Key::from_hex("010000000033333333444444445500000000").unwrap();
         let gap_at_key = current_key.add(100);
@@ -6770,7 +6811,7 @@ mod tests {
             .get_vectored_impl(
                 read.clone(),
                 current_lsn,
-                &mut ValuesReconstructState::new(),
+                &mut ValuesReconstructState::new(io_concurrency.clone()),
                 &ctx,
             )
             .await?;
@@ -6813,6 +6854,7 @@ mod tests {
     async fn test_get_vectored_ancestor_descent() -> anyhow::Result<()> {
         let harness = TenantHarness::create("test_get_vectored_on_lsn_axis").await?;
         let (tenant, ctx) = harness.load().await;
+        let io_concurrency = IoConcurrency::spawn_for_test();
 
         let start_key = Key::from_hex("010000000033333333444444445500000000").unwrap();
         let end_key = start_key.add(1000);
@@ -6905,7 +6947,7 @@ mod tests {
                         ranges: vec![child_gap_at_key..child_gap_at_key.next()],
                     },
                     query_lsn,
-                    &mut ValuesReconstructState::new(),
+                    &mut ValuesReconstructState::new(io_concurrency.clone()),
                     &ctx,
                 )
                 .await;
@@ -7351,6 +7393,7 @@ mod tests {
     async fn test_metadata_scan() -> anyhow::Result<()> {
         let harness = TenantHarness::create("test_metadata_scan").await?;
         let (tenant, ctx) = harness.load().await;
+        let io_concurrency = IoConcurrency::spawn_for_test();
         let tline = tenant
             .create_test_timeline(TIMELINE_ID, Lsn(0x10), DEFAULT_PG_VERSION, &ctx)
             .await?;
@@ -7404,7 +7447,7 @@ mod tests {
                 .get_vectored_impl(
                     keyspace.clone(),
                     lsn,
-                    &mut ValuesReconstructState::default(),
+                    &mut ValuesReconstructState::new(io_concurrency.clone()),
                     &ctx,
                 )
                 .await?
@@ -7519,6 +7562,7 @@ mod tests {
         let harness = TenantHarness::create("test_aux_file_e2e").await.unwrap();
 
         let (tenant, ctx) = harness.load().await;
+        let io_concurrency = IoConcurrency::spawn_for_test();
 
         let mut lsn = Lsn(0x08);
 
@@ -7538,7 +7582,10 @@ mod tests {
         }
 
         // we can read everything from the storage
-        let files = tline.list_aux_files(lsn, &ctx).await.unwrap();
+        let files = tline
+            .list_aux_files(lsn, &ctx, io_concurrency.clone())
+            .await
+            .unwrap();
         assert_eq!(
             files.get("pg_logical/mappings/test1"),
             Some(&bytes::Bytes::from_static(b"first"))
@@ -7554,7 +7601,10 @@ mod tests {
             modification.commit(&ctx).await.unwrap();
         }
 
-        let files = tline.list_aux_files(lsn, &ctx).await.unwrap();
+        let files = tline
+            .list_aux_files(lsn, &ctx, io_concurrency.clone())
+            .await
+            .unwrap();
         assert_eq!(
             files.get("pg_logical/mappings/test2"),
             Some(&bytes::Bytes::from_static(b"second"))
@@ -7565,7 +7615,10 @@ mod tests {
             .await
             .unwrap();
 
-        let files = child.list_aux_files(lsn, &ctx).await.unwrap();
+        let files = child
+            .list_aux_files(lsn, &ctx, io_concurrency.clone())
+            .await
+            .unwrap();
         assert_eq!(files.get("pg_logical/mappings/test1"), None);
         assert_eq!(files.get("pg_logical/mappings/test2"), None);
     }
@@ -7574,6 +7627,7 @@ mod tests {
     async fn test_metadata_image_creation() -> anyhow::Result<()> {
         let harness = TenantHarness::create("test_metadata_image_creation").await?;
         let (tenant, ctx) = harness.load().await;
+        let io_concurrency = IoConcurrency::spawn_for_test();
         let tline = tenant
             .create_test_timeline(TIMELINE_ID, Lsn(0x10), DEFAULT_PG_VERSION, &ctx)
             .await?;
@@ -7593,8 +7647,9 @@ mod tests {
             keyspace: &KeySpace,
             lsn: Lsn,
             ctx: &RequestContext,
+            io_concurrency: IoConcurrency,
         ) -> anyhow::Result<(BTreeMap<Key, Result<Bytes, PageReconstructError>>, usize)> {
-            let mut reconstruct_state = ValuesReconstructState::default();
+            let mut reconstruct_state = ValuesReconstructState::new(io_concurrency);
             let res = tline
                 .get_vectored_impl(keyspace.clone(), lsn, &mut reconstruct_state, ctx)
                 .await?;
@@ -7642,7 +7697,8 @@ mod tests {
 
             if iter % 5 == 0 {
                 let (_, before_delta_file_accessed) =
-                    scan_with_statistics(&tline, &keyspace, lsn, &ctx).await?;
+                    scan_with_statistics(&tline, &keyspace, lsn, &ctx, io_concurrency.clone())
+                        .await?;
                 tline
                     .compact(
                         &cancel,
@@ -7656,7 +7712,8 @@ mod tests {
                     )
                     .await?;
                 let (_, after_delta_file_accessed) =
-                    scan_with_statistics(&tline, &keyspace, lsn, &ctx).await?;
+                    scan_with_statistics(&tline, &keyspace, lsn, &ctx, io_concurrency.clone())
+                        .await?;
                 assert!(after_delta_file_accessed < before_delta_file_accessed, "after_delta_file_accessed={after_delta_file_accessed}, before_delta_file_accessed={before_delta_file_accessed}");
                 // Given that we already produced an image layer, there should be no delta layer needed for the scan, but still setting a low threshold there for unforeseen circumstances.
                 assert!(
@@ -7745,6 +7802,7 @@ mod tests {
     async fn test_vectored_missing_metadata_key_reads() -> anyhow::Result<()> {
         let harness = TenantHarness::create("test_vectored_missing_metadata_key_reads").await?;
         let (tenant, ctx) = harness.load().await;
+        let io_concurrency = IoConcurrency::spawn_for_test();
 
         let base_key = Key::from_hex("620000000033333333444444445500000000").unwrap();
         let base_key_child = Key::from_hex("620000000033333333444444445500000001").unwrap();
@@ -7883,7 +7941,7 @@ mod tests {
         );
 
         // test vectored scan on parent timeline
-        let mut reconstruct_state = ValuesReconstructState::new();
+        let mut reconstruct_state = ValuesReconstructState::new(io_concurrency.clone());
         let res = tline
             .get_vectored_impl(
                 KeySpace::single(Key::metadata_key_range()),
@@ -7909,7 +7967,7 @@ mod tests {
         );
 
         // test vectored scan on child timeline
-        let mut reconstruct_state = ValuesReconstructState::new();
+        let mut reconstruct_state = ValuesReconstructState::new(io_concurrency.clone());
         let res = child
             .get_vectored_impl(
                 KeySpace::single(Key::metadata_key_range()),
@@ -7947,7 +8005,9 @@ mod tests {
         lsn: Lsn,
         ctx: &RequestContext,
     ) -> Result<Option<Bytes>, GetVectoredError> {
-        let mut reconstruct_state = ValuesReconstructState::new();
+        let io_concurrency =
+            IoConcurrency::spawn_from_conf(tline.conf, tline.gate.enter().unwrap());
+        let mut reconstruct_state = ValuesReconstructState::new(io_concurrency);
         let mut res = tline
             .get_vectored_impl(
                 KeySpace::single(key..key.next()),
@@ -8048,6 +8108,7 @@ mod tests {
             .await
             .unwrap();
         let (tenant, ctx) = harness.load().await;
+        let io_concurrency = IoConcurrency::spawn_for_test();
 
         let key0 = Key::from_hex("620000000033333333444444445500000000").unwrap();
         let key1 = Key::from_hex("620000000033333333444444445500000001").unwrap();
@@ -8107,7 +8168,7 @@ mod tests {
 
         // Image layers are created at last_record_lsn
         let images = tline
-            .inspect_image_layers(Lsn(0x40), &ctx)
+            .inspect_image_layers(Lsn(0x40), &ctx, io_concurrency.clone())
             .await
             .unwrap()
             .into_iter()
@@ -8122,6 +8183,7 @@ mod tests {
             .await
             .unwrap();
         let (tenant, ctx) = harness.load().await;
+        let io_concurrency = IoConcurrency::spawn_for_test();
 
         let key1 = Key::from_hex("620000000033333333444444445500000001").unwrap();
         let key2 = Key::from_hex("620000000033333333444444445500000002").unwrap();
@@ -8172,7 +8234,7 @@ mod tests {
 
         // Image layers are created at last_record_lsn
         let images = tline
-            .inspect_image_layers(Lsn(0x30), &ctx)
+            .inspect_image_layers(Lsn(0x30), &ctx, io_concurrency.clone())
             .await
             .unwrap()
             .into_iter()
@@ -8185,6 +8247,7 @@ mod tests {
     async fn test_simple_bottom_most_compaction_images() -> anyhow::Result<()> {
         let harness = TenantHarness::create("test_simple_bottom_most_compaction_images").await?;
         let (tenant, ctx) = harness.load().await;
+        let io_concurrency = IoConcurrency::spawn_for_test();
 
         fn get_key(id: u32) -> Key {
             // using aux key here b/c they are guaranteed to be inside `collect_keyspace`.
@@ -8326,7 +8389,7 @@ mod tests {
 
         // Check if the image layer at the GC horizon contains exactly what we want
         let image_at_gc_horizon = tline
-            .inspect_image_layers(Lsn(0x30), &ctx)
+            .inspect_image_layers(Lsn(0x30), &ctx, io_concurrency.clone())
             .await
             .unwrap()
             .into_iter()
@@ -10039,7 +10102,12 @@ mod tests {
 
         let keyspace = KeySpace::single(get_key(0)..get_key(10));
         let results = tline
-            .get_vectored(keyspace, delta_layer_end_lsn, &ctx)
+            .get_vectored(
+                keyspace,
+                delta_layer_end_lsn,
+                IoConcurrency::sequential(),
+                &ctx,
+            )
             .await
             .expect("No vectored errors");
         for (key, res) in results {

pageserver/src/tenant/config.rs

@@ -281,6 +281,18 @@ pub struct TenantConfOpt {
     #[serde(default)]
     pub compaction_algorithm: Option<CompactionAlgorithmSettings>,
 
+    #[serde(skip_serializing_if = "Option::is_none")]
+    #[serde(default)]
+    pub l0_flush_delay_threshold: Option<usize>,
+
+    #[serde(skip_serializing_if = "Option::is_none")]
+    #[serde(default)]
+    pub l0_flush_stall_threshold: Option<usize>,
+
+    #[serde(skip_serializing_if = "Option::is_none")]
+    #[serde(default)]
+    pub l0_flush_wait_upload: Option<bool>,
+
     #[serde(skip_serializing_if = "Option::is_none")]
     #[serde(default)]
     pub gc_horizon: Option<u64>,
@@ -357,6 +369,18 @@ pub struct TenantConfOpt {
 
     #[serde(skip_serializing_if = "Option::is_none")]
     pub wal_receiver_protocol_override: Option<PostgresClientProtocol>,
+
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub rel_size_v2_enabled: Option<bool>,
+
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub gc_compaction_enabled: Option<bool>,
+
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub gc_compaction_initial_threshold_kb: Option<u64>,
+
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub gc_compaction_ratio_percent: Option<u64>,
 }
 
 impl TenantConfOpt {
@@ -382,6 +406,15 @@ impl TenantConfOpt {
                 .as_ref()
                 .unwrap_or(&global_conf.compaction_algorithm)
                 .clone(),
+            l0_flush_delay_threshold: self
+                .l0_flush_delay_threshold
+                .or(global_conf.l0_flush_delay_threshold),
+            l0_flush_stall_threshold: self
+                .l0_flush_stall_threshold
+                .or(global_conf.l0_flush_stall_threshold),
+            l0_flush_wait_upload: self
+                .l0_flush_wait_upload
+                .unwrap_or(global_conf.l0_flush_wait_upload),
             gc_horizon: self.gc_horizon.unwrap_or(global_conf.gc_horizon),
             gc_period: self.gc_period.unwrap_or(global_conf.gc_period),
             image_creation_threshold: self
@@ -425,6 +458,16 @@ impl TenantConfOpt {
             wal_receiver_protocol_override: self
                 .wal_receiver_protocol_override
                 .or(global_conf.wal_receiver_protocol_override),
+            rel_size_v2_enabled: self.rel_size_v2_enabled.or(global_conf.rel_size_v2_enabled),
+            gc_compaction_enabled: self
+                .gc_compaction_enabled
+                .unwrap_or(global_conf.gc_compaction_enabled),
+            gc_compaction_initial_threshold_kb: self
+                .gc_compaction_initial_threshold_kb
+                .unwrap_or(global_conf.gc_compaction_initial_threshold_kb),
+            gc_compaction_ratio_percent: self
+                .gc_compaction_ratio_percent
+                .unwrap_or(global_conf.gc_compaction_ratio_percent),
         }
     }
 
@@ -436,6 +479,9 @@ impl TenantConfOpt {
             mut compaction_period,
             mut compaction_threshold,
             mut compaction_algorithm,
+            mut l0_flush_delay_threshold,
+            mut l0_flush_stall_threshold,
+            mut l0_flush_wait_upload,
             mut gc_horizon,
             mut gc_period,
             mut image_creation_threshold,
@@ -454,6 +500,10 @@ impl TenantConfOpt {
             mut lsn_lease_length_for_ts,
             mut timeline_offloading,
             mut wal_receiver_protocol_override,
+            mut rel_size_v2_enabled,
+            mut gc_compaction_enabled,
+            mut gc_compaction_initial_threshold_kb,
+            mut gc_compaction_ratio_percent,
         } = self;
 
         patch.checkpoint_distance.apply(&mut checkpoint_distance);
@@ -470,6 +520,13 @@ impl TenantConfOpt {
             .apply(&mut compaction_period);
         patch.compaction_threshold.apply(&mut compaction_threshold);
         patch.compaction_algorithm.apply(&mut compaction_algorithm);
+        patch
+            .l0_flush_delay_threshold
+            .apply(&mut l0_flush_delay_threshold);
+        patch
+            .l0_flush_stall_threshold
+            .apply(&mut l0_flush_stall_threshold);
+        patch.l0_flush_wait_upload.apply(&mut l0_flush_wait_upload);
         patch.gc_horizon.apply(&mut gc_horizon);
         patch
             .gc_period
@@ -522,6 +579,16 @@ impl TenantConfOpt {
         patch
             .wal_receiver_protocol_override
             .apply(&mut wal_receiver_protocol_override);
+        patch.rel_size_v2_enabled.apply(&mut rel_size_v2_enabled);
+        patch
+            .gc_compaction_enabled
+            .apply(&mut gc_compaction_enabled);
+        patch
+            .gc_compaction_initial_threshold_kb
+            .apply(&mut gc_compaction_initial_threshold_kb);
+        patch
+            .gc_compaction_ratio_percent
+            .apply(&mut gc_compaction_ratio_percent);
 
         Ok(Self {
             checkpoint_distance,
@@ -530,6 +597,9 @@ impl TenantConfOpt {
             compaction_period,
             compaction_threshold,
             compaction_algorithm,
+            l0_flush_delay_threshold,
+            l0_flush_stall_threshold,
+            l0_flush_wait_upload,
             gc_horizon,
             gc_period,
             image_creation_threshold,
@@ -548,6 +618,10 @@ impl TenantConfOpt {
             lsn_lease_length_for_ts,
             timeline_offloading,
             wal_receiver_protocol_override,
+            rel_size_v2_enabled,
+            gc_compaction_enabled,
+            gc_compaction_initial_threshold_kb,
+            gc_compaction_ratio_percent,
         })
     }
 }
@@ -583,6 +657,9 @@ impl From<TenantConfOpt> for models::TenantConfig {
             compaction_target_size: value.compaction_target_size,
             compaction_period: value.compaction_period.map(humantime),
             compaction_threshold: value.compaction_threshold,
+            l0_flush_delay_threshold: value.l0_flush_delay_threshold,
+            l0_flush_stall_threshold: value.l0_flush_stall_threshold,
+            l0_flush_wait_upload: value.l0_flush_wait_upload,
             gc_horizon: value.gc_horizon,
             gc_period: value.gc_period.map(humantime),
             image_creation_threshold: value.image_creation_threshold,
@@ -603,6 +680,10 @@ impl From<TenantConfOpt> for models::TenantConfig {
             lsn_lease_length_for_ts: value.lsn_lease_length_for_ts.map(humantime),
             timeline_offloading: value.timeline_offloading,
             wal_receiver_protocol_override: value.wal_receiver_protocol_override,
+            rel_size_v2_enabled: value.rel_size_v2_enabled,
+            gc_compaction_enabled: value.gc_compaction_enabled,
+            gc_compaction_initial_threshold_kb: value.gc_compaction_initial_threshold_kb,
+            gc_compaction_ratio_percent: value.gc_compaction_ratio_percent,
         }
     }
 }

pageserver/src/tenant/layer_map.rs

@@ -57,6 +57,7 @@ use std::collections::{HashMap, VecDeque};
 use std::iter::Peekable;
 use std::ops::Range;
 use std::sync::Arc;
+use tokio::sync::watch;
 use utils::lsn::Lsn;
 
 use historic_layer_coverage::BufferedHistoricLayerCoverage;
@@ -67,7 +68,6 @@ use super::storage_layer::{LayerVisibilityHint, PersistentLayerDesc};
 ///
 /// LayerMap tracks what layers exist on a timeline.
 ///
-#[derive(Default)]
 pub struct LayerMap {
     //
     // 'open_layer' holds the current InMemoryLayer that is accepting new
@@ -93,7 +93,25 @@ pub struct LayerMap {
 
     /// L0 layers have key range Key::MIN..Key::MAX, and locating them using R-Tree search is very inefficient.
     /// So L0 layers are held in l0_delta_layers vector, in addition to the R-tree.
+    ///
+    /// NB: make sure to notify `watch_l0_deltas` on changes.
     l0_delta_layers: Vec<Arc<PersistentLayerDesc>>,
+
+    /// Notifies about L0 delta layer changes, sending the current number of L0 layers.
+    watch_l0_deltas: watch::Sender<usize>,
+}
+
+impl Default for LayerMap {
+    fn default() -> Self {
+        Self {
+            open_layer: Default::default(),
+            next_open_layer_at: Default::default(),
+            frozen_layers: Default::default(),
+            historic: Default::default(),
+            l0_delta_layers: Default::default(),
+            watch_l0_deltas: watch::channel(0).0,
+        }
+    }
 }
 
 /// The primary update API for the layer map.
@@ -466,6 +484,8 @@ impl LayerMap {
 
         if Self::is_l0(&layer_desc.key_range, layer_desc.is_delta) {
             self.l0_delta_layers.push(layer_desc.clone().into());
+            self.watch_l0_deltas
+                .send_replace(self.l0_delta_layers.len());
         }
 
         self.historic.insert(
@@ -488,6 +508,8 @@ impl LayerMap {
             let mut l0_delta_layers = std::mem::take(&mut self.l0_delta_layers);
             l0_delta_layers.retain(|other| other.key() != layer_key);
             self.l0_delta_layers = l0_delta_layers;
+            self.watch_l0_deltas
+                .send_replace(self.l0_delta_layers.len());
             // this assertion is related to use of Arc::ptr_eq in Self::compare_arced_layers,
             // there's a chance that the comparison fails at runtime due to it comparing (pointer,
             // vtable) pairs.
@@ -850,6 +872,11 @@ impl LayerMap {
         &self.l0_delta_layers
     }
 
+    /// Subscribes to L0 delta layer changes, sending the current number of L0 delta layers.
+    pub fn watch_level0_deltas(&self) -> watch::Receiver<usize> {
+        self.watch_l0_deltas.subscribe()
+    }
+
     /// debugging function to print out the contents of the layer map
     #[allow(unused)]
     pub async fn dump(&self, verbose: bool, ctx: &RequestContext) -> Result<()> {

pageserver/src/tenant/mgr.rs

@@ -1643,6 +1643,7 @@ impl TenantManager {
                         .wait_lsn(
                             *target_lsn,
                             crate::tenant::timeline::WaitLsnWaiter::Tenant,
+                            crate::tenant::timeline::WaitLsnTimeout::Default,
                             ctx,
                         )
                         .await

pageserver/src/tenant/remote_timeline_client.rs

@@ -382,6 +382,12 @@ pub(crate) struct RemoteTimelineClient {
     cancel: CancellationToken,
 }
 
+impl Drop for RemoteTimelineClient {
+    fn drop(&mut self) {
+        debug!("dropping RemoteTimelineClient");
+    }
+}
+
 impl RemoteTimelineClient {
     ///
     /// Create a remote storage client for given timeline
@@ -797,6 +803,12 @@ impl RemoteTimelineClient {
 
         upload_queue.dirty.metadata.apply(update);
 
+        // Defense in depth: if we somehow generated invalid metadata, do not persist it.
+        upload_queue
+            .dirty
+            .validate()
+            .map_err(|e| anyhow::anyhow!(e))?;
+
         self.schedule_index_upload(upload_queue);
 
         Ok(())

pageserver/src/tenant/remote_timeline_client/index.rs

@@ -77,8 +77,32 @@ pub struct IndexPart {
     ///
     /// None means no aux files have been written to the storage before the point
     /// when this flag is introduced.
+    ///
+    /// This flag is not used any more as all tenants have been transitioned to the new aux file policy.
     #[serde(skip_serializing_if = "Option::is_none", default)]
     pub(crate) last_aux_file_policy: Option<AuxFilePolicy>,
+
+    #[serde(skip_serializing_if = "Option::is_none", default)]
+    pub(crate) rel_size_migration: Option<RelSizeMigration>,
+
+    /// The LSN of gc-compaction horizon. Once gc-compaction is finished for all layer files below an LSN, this LSN will be updated.
+    #[serde(skip_serializing_if = "Option::is_none", default)]
+    pub(crate) l2_lsn: Option<Lsn>,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
+#[serde(rename_all = "camelCase")]
+pub enum RelSizeMigration {
+    /// The tenant is using the old rel_size format.
+    /// Note that this enum is persisted as `Option<RelSizeMigration>` in the index part, so
+    /// `None` is the same as `Some(RelSizeMigration::Legacy)`.
+    Legacy,
+    /// The tenant is migrating to the new rel_size format. Both old and new rel_size format are
+    /// persisted in the index part. The read path will read both formats and merge them.
+    Migrating,
+    /// The tenant has migrated to the new rel_size format. Only the new rel_size format is persisted
+    /// in the index part, and the read path will not read the old format.
+    Migrated,
 }
 
 impl IndexPart {
@@ -97,10 +121,12 @@ impl IndexPart {
     /// - 8: added `archived_at`
     /// - 9: +gc_blocking
     /// - 10: +import_pgdata
-    const LATEST_VERSION: usize = 10;
+    /// - 11: +rel_size_migration
+    /// - 12: +l2_lsn
+    const LATEST_VERSION: usize = 12;
 
     // Versions we may see when reading from a bucket.
-    pub const KNOWN_VERSIONS: &'static [usize] = &[1, 2, 3, 4, 5, 6, 7, 8, 9, 10];
+    pub const KNOWN_VERSIONS: &'static [usize] = &[1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12];
 
     pub const FILE_NAME: &'static str = "index_part.json";
 
@@ -116,6 +142,8 @@ impl IndexPart {
             gc_blocking: None,
             last_aux_file_policy: None,
             import_pgdata: None,
+            rel_size_migration: None,
+            l2_lsn: None,
         }
     }
 
@@ -152,6 +180,21 @@ impl IndexPart {
         };
         is_same_remote_layer_path(name, metadata, name, index_metadata)
     }
+
+    /// Check for invariants in the index: this is useful when uploading an index to ensure that if
+    /// we encounter a bug, we do not persist buggy metadata.
+    pub(crate) fn validate(&self) -> Result<(), String> {
+        if self.import_pgdata.is_none()
+            && self.metadata.ancestor_timeline().is_none()
+            && self.layer_metadata.is_empty()
+        {
+            // Unless we're in the middle of a raw pgdata import, or this is a child timeline,the index must
+            // always have at least one layer.
+            return Err("Index has no ancestor and no layers".to_string());
+        }
+
+        Ok(())
+    }
 }
 
 /// Metadata gathered for each of the layer files.
@@ -179,6 +222,10 @@ impl LayerFileMetadata {
             shard,
         }
     }
+    /// Helper to get both generation and file size in a tuple
+    pub fn generation_file_size(&self) -> (Generation, u64) {
+        (self.generation, self.file_size)
+    }
 }
 
 /// Limited history of earlier ancestors.
@@ -401,6 +448,8 @@ mod tests {
             gc_blocking: None,
             last_aux_file_policy: None,
             import_pgdata: None,
+            rel_size_migration: None,
+            l2_lsn: None,
         };
 
         let part = IndexPart::from_json_bytes(example.as_bytes()).unwrap();
@@ -446,6 +495,8 @@ mod tests {
             gc_blocking: None,
             last_aux_file_policy: None,
             import_pgdata: None,
+            rel_size_migration: None,
+            l2_lsn: None,
         };
 
         let part = IndexPart::from_json_bytes(example.as_bytes()).unwrap();
@@ -492,6 +543,8 @@ mod tests {
             gc_blocking: None,
             last_aux_file_policy: None,
             import_pgdata: None,
+            rel_size_migration: None,
+            l2_lsn: None,
         };
 
         let part = IndexPart::from_json_bytes(example.as_bytes()).unwrap();
@@ -541,6 +594,8 @@ mod tests {
             gc_blocking: None,
             last_aux_file_policy: None,
             import_pgdata: None,
+            rel_size_migration: None,
+            l2_lsn: None,
         };
 
         let empty_layers_parsed = IndexPart::from_json_bytes(empty_layers_json.as_bytes()).unwrap();
@@ -585,6 +640,8 @@ mod tests {
             gc_blocking: None,
             last_aux_file_policy: None,
             import_pgdata: None,
+            rel_size_migration: None,
+            l2_lsn: None,
         };
 
         let part = IndexPart::from_json_bytes(example.as_bytes()).unwrap();
@@ -632,6 +689,8 @@ mod tests {
             gc_blocking: None,
             last_aux_file_policy: None,
             import_pgdata: None,
+            rel_size_migration: None,
+            l2_lsn: None,
         };
 
         let part = IndexPart::from_json_bytes(example.as_bytes()).unwrap();
@@ -684,6 +743,8 @@ mod tests {
             gc_blocking: None,
             last_aux_file_policy: Some(AuxFilePolicy::V2),
             import_pgdata: None,
+            rel_size_migration: None,
+            l2_lsn: None,
         };
 
         let part = IndexPart::from_json_bytes(example.as_bytes()).unwrap();
@@ -741,6 +802,8 @@ mod tests {
             gc_blocking: None,
             last_aux_file_policy: Default::default(),
             import_pgdata: None,
+            rel_size_migration: None,
+            l2_lsn: None,
         };
 
         let part = IndexPart::from_json_bytes(example.as_bytes()).unwrap();
@@ -799,6 +862,8 @@ mod tests {
             gc_blocking: None,
             last_aux_file_policy: Default::default(),
             import_pgdata: None,
+            rel_size_migration: None,
+            l2_lsn: None,
         };
 
         let part = IndexPart::from_json_bytes(example.as_bytes()).unwrap();
@@ -862,6 +927,8 @@ mod tests {
             last_aux_file_policy: Default::default(),
             archived_at: None,
             import_pgdata: None,
+            rel_size_migration: None,
+            l2_lsn: None,
         };
 
         let part = IndexPart::from_json_bytes(example.as_bytes()).unwrap();
@@ -937,7 +1004,168 @@ mod tests {
                 started_at: parse_naive_datetime("2024-11-13T09:23:42.123000000"),
                 finished_at: parse_naive_datetime("2024-11-13T09:42:23.123000000"),
                 idempotency_key: import_pgdata::index_part_format::IdempotencyKey::new("specified-by-client-218a5213-5044-4562-a28d-d024c5f057f5".to_string()),
-            })))
+            }))),
+            rel_size_migration: None,
+            l2_lsn: None,
+        };
+
+        let part = IndexPart::from_json_bytes(example.as_bytes()).unwrap();
+        assert_eq!(part, expected);
+    }
+
+    #[test]
+    fn v11_rel_size_migration_is_parsed() {
+        let example = r#"{
+            "version": 11,
+            "layer_metadata":{
+                "000000000000000000000000000000000000-FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF__0000000001696070-00000000016960E9": { "file_size": 25600000 },
+                "000000000000000000000000000000000000-FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF__00000000016B59D8-00000000016B5A51": { "file_size": 9007199254741001 }
+            },
+            "disk_consistent_lsn":"0/16960E8",
+            "metadata": {
+                "disk_consistent_lsn": "0/16960E8",
+                "prev_record_lsn": "0/1696070",
+                "ancestor_timeline": "e45a7f37d3ee2ff17dc14bf4f4e3f52e",
+                "ancestor_lsn": "0/0",
+                "latest_gc_cutoff_lsn": "0/1696070",
+                "initdb_lsn": "0/1696070",
+                "pg_version": 14
+            },
+            "gc_blocking": {
+                "started_at": "2024-07-19T09:00:00.123",
+                "reasons": ["DetachAncestor"]
+            },
+            "import_pgdata": {
+                "V1": {
+                    "Done": {
+                        "idempotency_key": "specified-by-client-218a5213-5044-4562-a28d-d024c5f057f5",
+                        "started_at": "2024-11-13T09:23:42.123",
+                        "finished_at": "2024-11-13T09:42:23.123"
+                    }
+                }
+            },
+            "rel_size_migration": "legacy"
+        }"#;
+
+        let expected = IndexPart {
+            version: 11,
+            layer_metadata: HashMap::from([
+                ("000000000000000000000000000000000000-FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF__0000000001696070-00000000016960E9".parse().unwrap(), LayerFileMetadata {
+                    file_size: 25600000,
+                    generation: Generation::none(),
+                    shard: ShardIndex::unsharded()
+                }),
+                ("000000000000000000000000000000000000-FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF__00000000016B59D8-00000000016B5A51".parse().unwrap(), LayerFileMetadata {
+                    file_size: 9007199254741001,
+                    generation: Generation::none(),
+                    shard: ShardIndex::unsharded()
+                })
+            ]),
+            disk_consistent_lsn: "0/16960E8".parse::<Lsn>().unwrap(),
+            metadata: TimelineMetadata::new(
+                Lsn::from_str("0/16960E8").unwrap(),
+                Some(Lsn::from_str("0/1696070").unwrap()),
+                Some(TimelineId::from_str("e45a7f37d3ee2ff17dc14bf4f4e3f52e").unwrap()),
+                Lsn::INVALID,
+                Lsn::from_str("0/1696070").unwrap(),
+                Lsn::from_str("0/1696070").unwrap(),
+                14,
+            ).with_recalculated_checksum().unwrap(),
+            deleted_at: None,
+            lineage: Default::default(),
+            gc_blocking: Some(GcBlocking {
+                started_at: parse_naive_datetime("2024-07-19T09:00:00.123000000"),
+                reasons: enumset::EnumSet::from_iter([GcBlockingReason::DetachAncestor]),
+            }),
+            last_aux_file_policy: Default::default(),
+            archived_at: None,
+            import_pgdata: Some(import_pgdata::index_part_format::Root::V1(import_pgdata::index_part_format::V1::Done(import_pgdata::index_part_format::Done{
+                started_at: parse_naive_datetime("2024-11-13T09:23:42.123000000"),
+                finished_at: parse_naive_datetime("2024-11-13T09:42:23.123000000"),
+                idempotency_key: import_pgdata::index_part_format::IdempotencyKey::new("specified-by-client-218a5213-5044-4562-a28d-d024c5f057f5".to_string()),
+            }))),
+            rel_size_migration: Some(RelSizeMigration::Legacy),
+            l2_lsn: None,
+        };
+
+        let part = IndexPart::from_json_bytes(example.as_bytes()).unwrap();
+        assert_eq!(part, expected);
+    }
+
+    #[test]
+    fn v12_l2_lsn_is_parsed() {
+        let example = r#"{
+            "version": 12,
+            "layer_metadata":{
+                "000000000000000000000000000000000000-FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF__0000000001696070-00000000016960E9": { "file_size": 25600000 },
+                "000000000000000000000000000000000000-FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF__00000000016B59D8-00000000016B5A51": { "file_size": 9007199254741001 }
+            },
+            "disk_consistent_lsn":"0/16960E8",
+            "metadata": {
+                "disk_consistent_lsn": "0/16960E8",
+                "prev_record_lsn": "0/1696070",
+                "ancestor_timeline": "e45a7f37d3ee2ff17dc14bf4f4e3f52e",
+                "ancestor_lsn": "0/0",
+                "latest_gc_cutoff_lsn": "0/1696070",
+                "initdb_lsn": "0/1696070",
+                "pg_version": 14
+            },
+            "gc_blocking": {
+                "started_at": "2024-07-19T09:00:00.123",
+                "reasons": ["DetachAncestor"]
+            },
+            "import_pgdata": {
+                "V1": {
+                    "Done": {
+                        "idempotency_key": "specified-by-client-218a5213-5044-4562-a28d-d024c5f057f5",
+                        "started_at": "2024-11-13T09:23:42.123",
+                        "finished_at": "2024-11-13T09:42:23.123"
+                    }
+                }
+            },
+            "rel_size_migration": "legacy",
+            "l2_lsn": "0/16960E8"
+        }"#;
+
+        let expected = IndexPart {
+            version: 12,
+            layer_metadata: HashMap::from([
+                ("000000000000000000000000000000000000-FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF__0000000001696070-00000000016960E9".parse().unwrap(), LayerFileMetadata {
+                    file_size: 25600000,
+                    generation: Generation::none(),
+                    shard: ShardIndex::unsharded()
+                }),
+                ("000000000000000000000000000000000000-FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF__00000000016B59D8-00000000016B5A51".parse().unwrap(), LayerFileMetadata {
+                    file_size: 9007199254741001,
+                    generation: Generation::none(),
+                    shard: ShardIndex::unsharded()
+                })
+            ]),
+            disk_consistent_lsn: "0/16960E8".parse::<Lsn>().unwrap(),
+            metadata: TimelineMetadata::new(
+                Lsn::from_str("0/16960E8").unwrap(),
+                Some(Lsn::from_str("0/1696070").unwrap()),
+                Some(TimelineId::from_str("e45a7f37d3ee2ff17dc14bf4f4e3f52e").unwrap()),
+                Lsn::INVALID,
+                Lsn::from_str("0/1696070").unwrap(),
+                Lsn::from_str("0/1696070").unwrap(),
+                14,
+            ).with_recalculated_checksum().unwrap(),
+            deleted_at: None,
+            lineage: Default::default(),
+            gc_blocking: Some(GcBlocking {
+                started_at: parse_naive_datetime("2024-07-19T09:00:00.123000000"),
+                reasons: enumset::EnumSet::from_iter([GcBlockingReason::DetachAncestor]),
+            }),
+            last_aux_file_policy: Default::default(),
+            archived_at: None,
+            import_pgdata: Some(import_pgdata::index_part_format::Root::V1(import_pgdata::index_part_format::V1::Done(import_pgdata::index_part_format::Done{
+                started_at: parse_naive_datetime("2024-11-13T09:23:42.123000000"),
+                finished_at: parse_naive_datetime("2024-11-13T09:42:23.123000000"),
+                idempotency_key: import_pgdata::index_part_format::IdempotencyKey::new("specified-by-client-218a5213-5044-4562-a28d-d024c5f057f5".to_string()),
+            }))),
+            rel_size_migration: Some(RelSizeMigration::Legacy),
+            l2_lsn: Some("0/16960E8".parse::<Lsn>().unwrap()),
         };
 
         let part = IndexPart::from_json_bytes(example.as_bytes()).unwrap();

pageserver/src/tenant/remote_timeline_client/upload.rs

@@ -40,6 +40,10 @@ pub(crate) async fn upload_index_part(
     });
     pausable_failpoint!("before-upload-index-pausable");
 
+    // Safety: refuse to persist invalid index metadata, to mitigate the impact of any bug that produces this
+    // (this should never happen)
+    index_part.validate().map_err(|e| anyhow::anyhow!(e))?;
+
     // FIXME: this error comes too late
     let serialized = index_part.to_json_bytes()?;
     let serialized = Bytes::from(serialized);

pageserver/src/tenant/secondary/downloader.rs

@@ -559,6 +559,13 @@ impl JobGenerator<PendingDownload, RunningDownload, CompleteDownload, DownloadCo
     }
 }
 
+enum LayerAction {
+    Download,
+    NoAction,
+    Skip,
+    Touch,
+}
+
 /// This type is a convenience to group together the various functions involved in
 /// freshening a secondary tenant.
 struct TenantDownloader<'a> {
@@ -1008,69 +1015,17 @@ impl<'a> TenantDownloader<'a> {
                 return (Err(UpdateError::Restart), touched);
             }
 
-            // Existing on-disk layers: just update their access time.
-            if let Some(on_disk) = timeline_state.on_disk_layers.get(&layer.name) {
-                tracing::debug!("Layer {} is already on disk", layer.name);
-
-                if cfg!(debug_assertions) {
-                    // Debug for https://github.com/neondatabase/neon/issues/6966: check that the files we think
-                    // are already present on disk are really there.
-                    match tokio::fs::metadata(&on_disk.local_path).await {
-                        Ok(meta) => {
-                            tracing::debug!(
-                                "Layer {} present at {}, size {}",
-                                layer.name,
-                                on_disk.local_path,
-                                meta.len(),
-                            );
-                        }
-                        Err(e) => {
-                            tracing::warn!(
-                                "Layer {} not found at {} ({})",
-                                layer.name,
-                                on_disk.local_path,
-                                e
-                            );
-                            debug_assert!(false);
-                        }
-                    }
-                }
-
-                if on_disk.metadata != layer.metadata || on_disk.access_time != layer.access_time {
-                    // We already have this layer on disk.  Update its access time.
-                    tracing::debug!(
-                        "Access time updated for layer {}: {} -> {}",
-                        layer.name,
-                        strftime(&on_disk.access_time),
-                        strftime(&layer.access_time)
-                    );
-                    touched.push(layer);
-                }
-                continue;
-            } else {
-                tracing::debug!("Layer {} not present on disk yet", layer.name);
-            }
-
-            // Eviction: if we evicted a layer, then do not re-download it unless it was accessed more
-            // recently than it was evicted.
-            if let Some(evicted_at) = timeline_state.evicted_at.get(&layer.name) {
-                if &layer.access_time > evicted_at {
-                    tracing::info!(
-                        "Re-downloading evicted layer {}, accessed at {}, evicted at {}",
-                        layer.name,
-                        strftime(&layer.access_time),
-                        strftime(evicted_at)
-                    );
-                } else {
-                    tracing::trace!(
-                        "Not re-downloading evicted layer {}, accessed at {}, evicted at {}",
-                        layer.name,
-                        strftime(&layer.access_time),
-                        strftime(evicted_at)
-                    );
+            match self.layer_action(&timeline_state, &layer).await {
+                LayerAction::Download => (),
+                LayerAction::NoAction => continue,
+                LayerAction::Skip => {
                     self.skip_layer(layer);
                     continue;
                 }
+                LayerAction::Touch => {
+                    touched.push(layer);
+                    continue;
+                }
             }
 
             match self
@@ -1091,6 +1046,86 @@ impl<'a> TenantDownloader<'a> {
         (Ok(()), touched)
     }
 
+    async fn layer_action(
+        &self,
+        timeline_state: &SecondaryDetailTimeline,
+        layer: &HeatMapLayer,
+    ) -> LayerAction {
+        // Existing on-disk layers: just update their access time.
+        if let Some(on_disk) = timeline_state.on_disk_layers.get(&layer.name) {
+            tracing::debug!("Layer {} is already on disk", layer.name);
+
+            if cfg!(debug_assertions) {
+                // Debug for https://github.com/neondatabase/neon/issues/6966: check that the files we think
+                // are already present on disk are really there.
+                match tokio::fs::metadata(&on_disk.local_path).await {
+                    Ok(meta) => {
+                        tracing::debug!(
+                            "Layer {} present at {}, size {}",
+                            layer.name,
+                            on_disk.local_path,
+                            meta.len(),
+                        );
+                    }
+                    Err(e) => {
+                        tracing::warn!(
+                            "Layer {} not found at {} ({})",
+                            layer.name,
+                            on_disk.local_path,
+                            e
+                        );
+                        debug_assert!(false);
+                    }
+                }
+            }
+
+            if on_disk.metadata.generation_file_size() != on_disk.metadata.generation_file_size() {
+                tracing::info!(
+                    "Re-downloading layer {} with changed size or generation: {:?}->{:?}",
+                    layer.name,
+                    on_disk.metadata.generation_file_size(),
+                    on_disk.metadata.generation_file_size()
+                );
+                return LayerAction::Download;
+            }
+            if on_disk.metadata != layer.metadata || on_disk.access_time != layer.access_time {
+                // We already have this layer on disk.  Update its access time.
+                tracing::debug!(
+                    "Access time updated for layer {}: {} -> {}",
+                    layer.name,
+                    strftime(&on_disk.access_time),
+                    strftime(&layer.access_time)
+                );
+                return LayerAction::Touch;
+            }
+            return LayerAction::NoAction;
+        } else {
+            tracing::debug!("Layer {} not present on disk yet", layer.name);
+        }
+
+        // Eviction: if we evicted a layer, then do not re-download it unless it was accessed more
+        // recently than it was evicted.
+        if let Some(evicted_at) = timeline_state.evicted_at.get(&layer.name) {
+            if &layer.access_time > evicted_at {
+                tracing::info!(
+                    "Re-downloading evicted layer {}, accessed at {}, evicted at {}",
+                    layer.name,
+                    strftime(&layer.access_time),
+                    strftime(evicted_at)
+                );
+            } else {
+                tracing::trace!(
+                    "Not re-downloading evicted layer {}, accessed at {}, evicted at {}",
+                    layer.name,
+                    strftime(&layer.access_time),
+                    strftime(evicted_at)
+                );
+                return LayerAction::Skip;
+            }
+        }
+        LayerAction::Download
+    }
+
     async fn download_timeline(
         &self,
         timeline: HeatMapTimeline,

pageserver/src/tenant/storage_layer.rs

@@ -10,18 +10,26 @@ mod layer_desc;
 mod layer_name;
 pub mod merge_iterator;
 
+use crate::config::PageServerConf;
 use crate::context::{AccessStatsBehavior, RequestContext};
 use bytes::Bytes;
+use futures::stream::FuturesUnordered;
+use futures::StreamExt;
 use pageserver_api::key::Key;
 use pageserver_api::keyspace::{KeySpace, KeySpaceRandomAccum};
 use pageserver_api::record::NeonWalRecord;
 use pageserver_api::value::Value;
-use std::cmp::{Ordering, Reverse};
+use std::cmp::Ordering;
 use std::collections::hash_map::Entry;
 use std::collections::{BinaryHeap, HashMap};
+use std::future::Future;
 use std::ops::Range;
+use std::pin::Pin;
+use std::sync::atomic::AtomicUsize;
 use std::sync::Arc;
 use std::time::{Duration, SystemTime, UNIX_EPOCH};
+use tracing::{trace, Instrument};
+use utils::sync::gate::GateGuard;
 
 use utils::lsn::Lsn;
 
@@ -78,30 +86,151 @@ pub(crate) enum ValueReconstructSituation {
     Continue,
 }
 
-/// Reconstruct data accumulated for a single key during a vectored get
-#[derive(Debug, Default, Clone)]
-pub(crate) struct VectoredValueReconstructState {
-    pub(crate) records: Vec<(Lsn, NeonWalRecord)>,
-    pub(crate) img: Option<(Lsn, Bytes)>,
-
-    situation: ValueReconstructSituation,
+/// On disk representation of a value loaded in a buffer
+#[derive(Debug)]
+pub(crate) enum OnDiskValue {
+    /// Unencoded [`Value::Image`]
+    RawImage(Bytes),
+    /// Encoded [`Value`]. Can deserialize into an image or a WAL record
+    WalRecordOrImage(Bytes),
 }
 
-impl VectoredValueReconstructState {
-    fn get_cached_lsn(&self) -> Option<Lsn> {
-        self.img.as_ref().map(|img| img.0)
+/// Reconstruct data accumulated for a single key during a vectored get
+#[derive(Debug, Default)]
+pub(crate) struct VectoredValueReconstructState {
+    pub(crate) on_disk_values: Vec<(Lsn, OnDiskValueIoWaiter)>,
+
+    pub(crate) situation: ValueReconstructSituation,
+}
+
+#[derive(Debug)]
+pub(crate) struct OnDiskValueIoWaiter {
+    rx: tokio::sync::oneshot::Receiver<OnDiskValueIoResult>,
+}
+
+#[derive(Debug)]
+#[must_use]
+pub(crate) enum OnDiskValueIo {
+    /// Traversal identified this IO as required to complete the vectored get.
+    Required {
+        num_active_ios: Arc<AtomicUsize>,
+        tx: tokio::sync::oneshot::Sender<OnDiskValueIoResult>,
+    },
+    /// Sparse keyspace reads always read all the values for a given key,
+    /// even though only the first value is needed.
+    ///
+    /// This variant represents the unnecessary IOs for those values at lower LSNs
+    /// that aren't needed, but are currently still being done.
+    ///
+    /// The execution of unnecessary IOs was a pre-existing behavior before concurrent IO.
+    /// We added this explicit representation here so that we can drop
+    /// unnecessary IO results immediately, instead of buffering them in
+    /// `oneshot` channels inside [`VectoredValueReconstructState`] until
+    /// [`VectoredValueReconstructState::collect_pending_ios`] gets called.
+    Unnecessary,
+}
+
+type OnDiskValueIoResult = Result<OnDiskValue, std::io::Error>;
+
+impl OnDiskValueIo {
+    pub(crate) fn complete(self, res: OnDiskValueIoResult) {
+        match self {
+            OnDiskValueIo::Required { num_active_ios, tx } => {
+                num_active_ios.fetch_sub(1, std::sync::atomic::Ordering::Release);
+                let _ = tx.send(res);
+            }
+            OnDiskValueIo::Unnecessary => {
+                // Nobody cared, see variant doc comment.
+            }
+        }
     }
 }
 
-impl From<VectoredValueReconstructState> for ValueReconstructState {
-    fn from(mut state: VectoredValueReconstructState) -> Self {
-        // walredo expects the records to be descending in terms of Lsn
-        state.records.sort_by_key(|(lsn, _)| Reverse(*lsn));
+#[derive(Debug, thiserror::Error)]
+pub(crate) enum WaitCompletionError {
+    #[error("OnDiskValueIo was dropped without completing, likely the sidecar task panicked")]
+    IoDropped,
+}
 
-        ValueReconstructState {
-            records: state.records,
-            img: state.img,
+impl OnDiskValueIoWaiter {
+    pub(crate) async fn wait_completion(self) -> Result<OnDiskValueIoResult, WaitCompletionError> {
+        // NB: for Unnecessary IOs, this method never gets called because we don't add them to `on_disk_values`.
+        self.rx.await.map_err(|_| WaitCompletionError::IoDropped)
+    }
+}
+
+impl VectoredValueReconstructState {
+    /// # Cancel-Safety
+    ///
+    /// Technically fine to stop polling this future, but, the IOs will still
+    /// be executed to completion by the sidecar task and hold on to / consume resources.
+    /// Better not do it to make reasonsing about the system easier.
+    pub(crate) async fn collect_pending_ios(
+        self,
+    ) -> Result<ValueReconstructState, PageReconstructError> {
+        use utils::bin_ser::BeSer;
+
+        let mut res = Ok(ValueReconstructState::default());
+
+        // We should try hard not to bail early, so that by the time we return from this
+        // function, all IO for this value is done. It's not required -- we could totally
+        // stop polling the IO futures in the sidecar task, they need to support that,
+        // but just stopping to poll doesn't reduce the IO load on the disk. It's easier
+        // to reason about the system if we just wait for all IO to complete, even if
+        // we're no longer interested in the result.
+        //
+        // Revisit this when IO futures are replaced with a more sophisticated IO system
+        // and an IO scheduler, where we know which IOs were submitted and which ones
+        // just queued. Cf the comment on IoConcurrency::spawn_io.
+        for (lsn, waiter) in self.on_disk_values {
+            let value_recv_res = waiter
+                .wait_completion()
+                // we rely on the caller to poll us to completion, so this is not a bail point
+                .await;
+            // Force not bailing early by wrapping the code into a closure.
+            #[allow(clippy::redundant_closure_call)]
+            let _: () = (|| {
+                match (&mut res, value_recv_res) {
+                    (Err(_), _) => {
+                        // We've already failed, no need to process more.
+                    }
+                    (Ok(_), Err(wait_err)) => {
+                        // This shouldn't happen - likely the sidecar task panicked.
+                        res = Err(PageReconstructError::Other(wait_err.into()));
+                    }
+                    (Ok(_), Ok(Err(err))) => {
+                        let err: std::io::Error = err;
+                        // TODO: returning IO error here will fail a compute query.
+                        // Probably not what we want, we're not doing `maybe_fatal_err`
+                        // in the IO futures.
+                        // But it's been like that for a long time, not changing it
+                        // as part of concurrent IO.
+                        // => https://github.com/neondatabase/neon/issues/10454
+                        res = Err(PageReconstructError::Other(err.into()));
+                    }
+                    (Ok(ok), Ok(Ok(OnDiskValue::RawImage(img)))) => {
+                        assert!(ok.img.is_none());
+                        ok.img = Some((lsn, img));
+                    }
+                    (Ok(ok), Ok(Ok(OnDiskValue::WalRecordOrImage(buf)))) => {
+                        match Value::des(&buf) {
+                            Ok(Value::WalRecord(rec)) => {
+                                ok.records.push((lsn, rec));
+                            }
+                            Ok(Value::Image(img)) => {
+                                assert!(ok.img.is_none());
+                                ok.img = Some((lsn, img));
+                            }
+                            Err(err) => {
+                                res = Err(PageReconstructError::Other(err.into()));
+                            }
+                        }
+                    }
+                }
+            })();
         }
+
+        res
     }
 }
 
@@ -109,7 +238,7 @@ impl From<VectoredValueReconstructState> for ValueReconstructState {
 pub(crate) struct ValuesReconstructState {
     /// The keys will be removed after `get_vectored` completes. The caller outside `Timeline`
     /// should not expect to get anything from this hashmap.
-    pub(crate) keys: HashMap<Key, Result<VectoredValueReconstructState, PageReconstructError>>,
+    pub(crate) keys: HashMap<Key, VectoredValueReconstructState>,
     /// The keys which are already retrieved
     keys_done: KeySpaceRandomAccum,
 
@@ -119,27 +248,365 @@ pub(crate) struct ValuesReconstructState {
     // Statistics that are still accessible as a caller of `get_vectored_impl`.
     layers_visited: u32,
     delta_layers_visited: u32,
+
+    pub(crate) io_concurrency: IoConcurrency,
+    num_active_ios: Arc<AtomicUsize>,
+}
+
+/// The level of IO concurrency to be used on the read path
+///
+/// The desired end state is that we always do parallel IO.
+/// This struct and the dispatching in the impl will be removed once
+/// we've built enough confidence.
+pub(crate) enum IoConcurrency {
+    Sequential,
+    SidecarTask {
+        task_id: usize,
+        ios_tx: tokio::sync::mpsc::UnboundedSender<IoFuture>,
+    },
+}
+
+type IoFuture = Pin<Box<dyn Send + Future<Output = ()>>>;
+
+pub(crate) enum SelectedIoConcurrency {
+    Sequential,
+    SidecarTask(GateGuard),
+}
+
+impl std::fmt::Debug for IoConcurrency {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        match self {
+            IoConcurrency::Sequential => write!(f, "Sequential"),
+            IoConcurrency::SidecarTask { .. } => write!(f, "SidecarTask"),
+        }
+    }
+}
+
+impl std::fmt::Debug for SelectedIoConcurrency {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        match self {
+            SelectedIoConcurrency::Sequential => write!(f, "Sequential"),
+            SelectedIoConcurrency::SidecarTask(_) => write!(f, "SidecarTask"),
+        }
+    }
+}
+
+impl IoConcurrency {
+    /// Force sequential IO. This is a temporary workaround until we have
+    /// moved plumbing-through-the-call-stack
+    /// of IoConcurrency into `RequestContextq.
+    ///
+    /// DO NOT USE for new code.
+    ///
+    /// Tracking issue: <https://github.com/neondatabase/neon/issues/10460>.
+    pub(crate) fn sequential() -> Self {
+        Self::spawn(SelectedIoConcurrency::Sequential)
+    }
+
+    pub(crate) fn spawn_from_conf(
+        conf: &'static PageServerConf,
+        gate_guard: GateGuard,
+    ) -> IoConcurrency {
+        use pageserver_api::config::GetVectoredConcurrentIo;
+        let selected = match conf.get_vectored_concurrent_io {
+            GetVectoredConcurrentIo::Sequential => SelectedIoConcurrency::Sequential,
+            GetVectoredConcurrentIo::SidecarTask => SelectedIoConcurrency::SidecarTask(gate_guard),
+        };
+        Self::spawn(selected)
+    }
+
+    pub(crate) fn spawn(io_concurrency: SelectedIoConcurrency) -> Self {
+        match io_concurrency {
+            SelectedIoConcurrency::Sequential => IoConcurrency::Sequential,
+            SelectedIoConcurrency::SidecarTask(gate_guard) => {
+                let (ios_tx, ios_rx) = tokio::sync::mpsc::unbounded_channel();
+                static TASK_ID: AtomicUsize = AtomicUsize::new(0);
+                let task_id = TASK_ID.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
+                // TODO: enrich the span with more context (tenant,shard,timeline) + (basebackup|pagestream|...)
+                let span =
+                    tracing::info_span!(parent: None, "IoConcurrency_sidecar", task_id = task_id);
+                trace!(task_id, "spawning sidecar task");
+                tokio::spawn(async move {
+                    trace!("start");
+                    scopeguard::defer!{ trace!("end") };
+                    type IosRx = tokio::sync::mpsc::UnboundedReceiver<IoFuture>;
+                    enum State {
+                        Waiting {
+                            // invariant: is_empty(), but we recycle the allocation
+                            empty_futures: FuturesUnordered<IoFuture>,
+                            ios_rx: IosRx,
+                        },
+                        Executing {
+                            futures: FuturesUnordered<IoFuture>,
+                            ios_rx: IosRx,
+                        },
+                        ShuttingDown {
+                            futures: FuturesUnordered<IoFuture>,
+                        },
+                    }
+                    let mut state = State::Waiting {
+                        empty_futures: FuturesUnordered::new(),
+                        ios_rx,
+                    };
+                    loop {
+                        match state {
+                            State::Waiting {
+                                empty_futures,
+                                mut ios_rx,
+                            } => {
+                                assert!(empty_futures.is_empty());
+                                tokio::select! {
+                                    fut = ios_rx.recv() => {
+                                        if let Some(fut) = fut {
+                                            trace!("received new io future");
+                                            empty_futures.push(fut);
+                                            state = State::Executing { futures: empty_futures, ios_rx };
+                                        } else {
+                                            state = State::ShuttingDown { futures: empty_futures }
+                                        }
+                                    }
+                                }
+                            }
+                            State::Executing {
+                                mut futures,
+                                mut ios_rx,
+                            } => {
+                                tokio::select! {
+                                    res = futures.next() => {
+                                        trace!("io future completed");
+                                        assert!(res.is_some());
+                                        if futures.is_empty() {
+                                            state = State::Waiting { empty_futures: futures, ios_rx};
+                                        } else {
+                                            state = State::Executing { futures, ios_rx };
+                                        }
+                                    }
+                                    fut = ios_rx.recv() => {
+                                        if let Some(fut) = fut {
+                                            trace!("received new io future");
+                                            futures.push(fut);
+                                            state =  State::Executing { futures, ios_rx};
+                                        } else {
+                                            state = State::ShuttingDown { futures };
+                                        }
+                                    }
+                                }
+                            }
+                            State::ShuttingDown {
+                                mut futures,
+                            } => {
+                                trace!("shutting down");
+                                while let Some(()) = futures.next().await {
+                                    trace!("io future completed (shutdown)");
+                                    // drain
+                                }
+                                trace!("shutdown complete");
+                                break;
+                            }
+                        }
+                    }
+                    drop(gate_guard); // drop it right before we exit
+                }.instrument(span));
+                IoConcurrency::SidecarTask { task_id, ios_tx }
+            }
+        }
+    }
+
+    pub(crate) fn clone(&self) -> Self {
+        match self {
+            IoConcurrency::Sequential => IoConcurrency::Sequential,
+            IoConcurrency::SidecarTask { task_id, ios_tx } => IoConcurrency::SidecarTask {
+                task_id: *task_id,
+                ios_tx: ios_tx.clone(),
+            },
+        }
+    }
+
+    /// Submit an IO to be executed in the background. DEADLOCK RISK, read the full doc string.
+    ///
+    /// The IO is represented as an opaque future.
+    /// IO completion must be handled inside the future, e.g., through a oneshot channel.
+    ///
+    /// The API seems simple but there are multiple **pitfalls** involving
+    /// DEADLOCK RISK.
+    ///
+    /// First, there are no guarantees about the exexecution of the IO.
+    /// It may be `await`ed in-place before this function returns.
+    /// It may be polled partially by this task and handed off to another task to be finished.
+    /// It may be polled and then dropped before returning ready.
+    ///
+    /// This means that submitted IOs must not be interedependent.
+    /// Interdependence may be through shared limited resources, e.g.,
+    /// - VirtualFile file descriptor cache slot acquisition
+    /// - tokio-epoll-uring slot
+    ///
+    /// # Why current usage is safe from deadlocks
+    ///
+    /// Textbook condition for a deadlock is that _all_ of the following be given
+    /// - Mutual exclusion
+    /// - Hold and wait
+    /// - No preemption
+    /// - Circular wait
+    ///
+    /// The current usage is safe because:
+    /// - Mutual exclusion: IO futures definitely use mutexes, no way around that for now
+    /// - Hold and wait: IO futures currently hold two kinds of locks/resources while waiting
+    ///   for acquisition of other resources:
+    ///    - VirtualFile file descriptor cache slot tokio mutex
+    ///    - tokio-epoll-uring slot (uses tokio notify => wait queue, much like mutex)
+    /// - No preemption: there's no taking-away of acquired locks/resources => given
+    /// - Circular wait: this is the part of the condition that isn't met: all IO futures
+    ///   first acquire VirtualFile mutex, then tokio-epoll-uring slot.
+    ///   There is no IO future that acquires slot before VirtualFile.
+    ///   Hence there can be no circular waiting.
+    ///   Hence there cannot be a deadlock.
+    ///
+    /// This is a very fragile situation and must be revisited whenver any code called from
+    /// inside the IO futures is changed.
+    ///
+    /// We will move away from opaque IO futures towards well-defined IOs at some point in
+    /// the future when we have shipped this first version of concurrent IO to production
+    /// and are ready to retire the Sequential mode which runs the futures in place.
+    /// Right now, while brittle, the opaque IO approach allows us to ship the feature
+    /// with minimal changes to the code and minimal changes to existing behavior in Sequential mode.
+    ///
+    /// Also read the comment in `collect_pending_ios`.
+    pub(crate) async fn spawn_io<F>(&mut self, fut: F)
+    where
+        F: std::future::Future<Output = ()> + Send + 'static,
+    {
+        match self {
+            IoConcurrency::Sequential => fut.await,
+            IoConcurrency::SidecarTask { ios_tx, .. } => {
+                let fut = Box::pin(fut);
+                // NB: experiments showed that doing an opportunistic poll of `fut` here was bad for throughput
+                // while insignificant for latency.
+                // It would make sense to revisit the tokio-epoll-uring API in the future such that we can try
+                // a submission here, but never poll the future. That way, io_uring can make proccess while
+                // the future sits in the ios_tx queue.
+                match ios_tx.send(fut) {
+                    Ok(()) => {}
+                    Err(_) => {
+                        unreachable!("the io task must have exited, likely it panicked")
+                    }
+                }
+            }
+        }
+    }
+
+    #[cfg(test)]
+    pub(crate) fn spawn_for_test() -> impl std::ops::DerefMut<Target = Self> {
+        use std::ops::{Deref, DerefMut};
+        use tracing::info;
+        use utils::sync::gate::Gate;
+
+        // Spawn needs a Gate, give it one.
+        struct Wrapper {
+            inner: IoConcurrency,
+            #[allow(dead_code)]
+            gate: Box<Gate>,
+        }
+        impl Deref for Wrapper {
+            type Target = IoConcurrency;
+
+            fn deref(&self) -> &Self::Target {
+                &self.inner
+            }
+        }
+        impl DerefMut for Wrapper {
+            fn deref_mut(&mut self) -> &mut Self::Target {
+                &mut self.inner
+            }
+        }
+        let gate = Box::new(Gate::default());
+
+        // The default behavior when running Rust unit tests without any further
+        // flags is to use the new behavior.
+        // The CI uses the following environment variable to unit test both old
+        // and new behavior.
+        // NB: the Python regression & perf tests take the `else` branch
+        // below and have their own defaults management.
+        let selected = {
+            // The pageserver_api::config type is unsuitable because it's internally tagged.
+            #[derive(serde::Deserialize)]
+            #[serde(rename_all = "kebab-case")]
+            enum TestOverride {
+                Sequential,
+                SidecarTask,
+            }
+            use once_cell::sync::Lazy;
+            static TEST_OVERRIDE: Lazy<TestOverride> = Lazy::new(|| {
+                utils::env::var_serde_json_string(
+                    "NEON_PAGESERVER_UNIT_TEST_GET_VECTORED_CONCURRENT_IO",
+                )
+                .unwrap_or(TestOverride::SidecarTask)
+            });
+
+            match *TEST_OVERRIDE {
+                TestOverride::Sequential => SelectedIoConcurrency::Sequential,
+                TestOverride::SidecarTask => {
+                    SelectedIoConcurrency::SidecarTask(gate.enter().expect("just created it"))
+                }
+            }
+        };
+
+        info!(?selected, "get_vectored_concurrent_io test");
+
+        Wrapper {
+            inner: Self::spawn(selected),
+            gate,
+        }
+    }
+}
+
+/// Make noise in case the [`ValuesReconstructState`] gets dropped while
+/// there are still IOs in flight.
+/// Refer to `collect_pending_ios` for why we prefer not to do that.
+//
+/// We log from here instead of from the sidecar task because the [`ValuesReconstructState`]
+/// gets dropped in a tracing span with more context.
+/// We repeat the sidecar tasks's `task_id` so we can correlate what we emit here with
+/// the logs / panic handler logs from the sidecar task, which also logs the `task_id`.
+impl Drop for ValuesReconstructState {
+    fn drop(&mut self) {
+        let num_active_ios = self
+            .num_active_ios
+            .load(std::sync::atomic::Ordering::Acquire);
+        if num_active_ios == 0 {
+            return;
+        }
+        let sidecar_task_id = match &self.io_concurrency {
+            IoConcurrency::Sequential => None,
+            IoConcurrency::SidecarTask { task_id, .. } => Some(*task_id),
+        };
+        tracing::warn!(
+            num_active_ios,
+            ?sidecar_task_id,
+            backtrace=%std::backtrace::Backtrace::force_capture(),
+            "dropping ValuesReconstructState while some IOs have not been completed",
+        );
+    }
 }
 
 impl ValuesReconstructState {
-    pub(crate) fn new() -> Self {
+    pub(crate) fn new(io_concurrency: IoConcurrency) -> Self {
         Self {
             keys: HashMap::new(),
             keys_done: KeySpaceRandomAccum::new(),
             keys_with_image_coverage: None,
             layers_visited: 0,
             delta_layers_visited: 0,
+            io_concurrency,
+            num_active_ios: Arc::new(AtomicUsize::new(0)),
         }
     }
 
-    /// Associate a key with the error which it encountered and mark it as done
-    pub(crate) fn on_key_error(&mut self, key: Key, err: PageReconstructError) {
-        let previous = self.keys.insert(key, Err(err));
-        if let Some(Ok(state)) = previous {
-            if state.situation == ValueReconstructSituation::Continue {
-                self.keys_done.add_key(key);
-            }
-        }
+    /// Absolutely read [`IoConcurrency::spawn_io`] to learn about assumptions & pitfalls.
+    pub(crate) async fn spawn_io<F>(&mut self, fut: F)
+    where
+        F: std::future::Future<Output = ()> + Send + 'static,
+    {
+        self.io_concurrency.spawn_io(fut).await;
     }
 
     pub(crate) fn on_layer_visited(&mut self, layer: &ReadableLayer) {
@@ -159,29 +626,6 @@ impl ValuesReconstructState {
         self.layers_visited
     }
 
-    /// This function is called after reading a keyspace from a layer.
-    /// It checks if the read path has now moved past the cached Lsn for any keys.
-    ///
-    /// Implementation note: We intentionally iterate over the keys for which we've
-    /// already collected some reconstruct data. This avoids scaling complexity with
-    /// the size of the search space.
-    pub(crate) fn on_lsn_advanced(&mut self, keyspace: &KeySpace, advanced_to: Lsn) {
-        for (key, value) in self.keys.iter_mut() {
-            if !keyspace.contains(key) {
-                continue;
-            }
-
-            if let Ok(state) = value {
-                if state.situation != ValueReconstructSituation::Complete
-                    && state.get_cached_lsn() >= Some(advanced_to)
-                {
-                    state.situation = ValueReconstructSituation::Complete;
-                    self.keys_done.add_key(*key);
-                }
-            }
-        }
-    }
-
     /// On hitting image layer, we can mark all keys in this range as done, because
     /// if the image layer does not contain a key, it is deleted/never added.
     pub(crate) fn on_image_layer_visited(&mut self, key_range: &Range<Key>) {
@@ -199,70 +643,42 @@ impl ValuesReconstructState {
     ///
     /// If the key is in the sparse keyspace (i.e., aux files), we do not track them in
     /// `key_done`.
-    pub(crate) fn update_key(
-        &mut self,
-        key: &Key,
-        lsn: Lsn,
-        value: Value,
-    ) -> ValueReconstructSituation {
-        let state = self
-            .keys
-            .entry(*key)
-            .or_insert(Ok(VectoredValueReconstructState::default()));
+    // TODO: rename this method & update description.
+    pub(crate) fn update_key(&mut self, key: &Key, lsn: Lsn, completes: bool) -> OnDiskValueIo {
+        let state = self.keys.entry(*key).or_default();
+
         let is_sparse_key = key.is_sparse();
-        if let Ok(state) = state {
-            let key_done = match state.situation {
-                ValueReconstructSituation::Complete => {
-                    if is_sparse_key {
-                        // Sparse keyspace might be visited multiple times because
-                        // we don't track unmapped keyspaces.
-                        return ValueReconstructSituation::Complete;
-                    } else {
-                        unreachable!()
-                    }
-                }
-                ValueReconstructSituation::Continue => match value {
-                    Value::Image(img) => {
-                        state.img = Some((lsn, img));
-                        true
-                    }
-                    Value::WalRecord(rec) => {
-                        debug_assert!(
-                            Some(lsn) > state.get_cached_lsn(),
-                            "Attempt to collect a record below cached LSN for walredo: {} < {}",
-                            lsn,
-                            state
-                                .get_cached_lsn()
-                                .expect("Assertion can only fire if a cached lsn is present")
-                        );
 
-                        let will_init = rec.will_init();
-                        state.records.push((lsn, rec));
-                        will_init
-                    }
-                },
-            };
-
-            if key_done && state.situation == ValueReconstructSituation::Continue {
-                state.situation = ValueReconstructSituation::Complete;
-                if !is_sparse_key {
-                    self.keys_done.add_key(*key);
+        let required_io = match state.situation {
+            ValueReconstructSituation::Complete => {
+                if is_sparse_key {
+                    // Sparse keyspace might be visited multiple times because
+                    // we don't track unmapped keyspaces.
+                    return OnDiskValueIo::Unnecessary;
+                } else {
+                    unreachable!()
                 }
             }
+            ValueReconstructSituation::Continue => {
+                self.num_active_ios
+                    .fetch_add(1, std::sync::atomic::Ordering::Release);
+                let (tx, rx) = tokio::sync::oneshot::channel();
+                state.on_disk_values.push((lsn, OnDiskValueIoWaiter { rx }));
+                OnDiskValueIo::Required {
+                    tx,
+                    num_active_ios: Arc::clone(&self.num_active_ios),
+                }
+            }
+        };
 
-            state.situation
-        } else {
-            ValueReconstructSituation::Complete
+        if completes && state.situation == ValueReconstructSituation::Continue {
+            state.situation = ValueReconstructSituation::Complete;
+            if !is_sparse_key {
+                self.keys_done.add_key(*key);
+            }
         }
-    }
 
-    /// Returns the Lsn at which this key is cached if one exists.
-    /// The read path should go no further than this Lsn for the given key.
-    pub(crate) fn get_cached_lsn(&self, key: &Key) -> Option<Lsn> {
-        self.keys
-            .get(key)
-            .and_then(|k| k.as_ref().ok())
-            .and_then(|state| state.get_cached_lsn())
+        required_io
     }
 
     /// Returns the key space describing the keys that have
@@ -276,12 +692,6 @@ impl ValuesReconstructState {
     }
 }
 
-impl Default for ValuesReconstructState {
-    fn default() -> Self {
-        Self::new()
-    }
-}
-
 /// A key that uniquely identifies a layer in a timeline
 #[derive(Debug, PartialEq, Eq, Clone, Hash)]
 pub(crate) enum LayerId {
@@ -720,3 +1130,78 @@ impl<T: std::fmt::Display> std::fmt::Debug for RangeDisplayDebug<'_, T> {
         write!(f, "{}..{}", self.0.start, self.0.end)
     }
 }
+
+#[cfg(test)]
+mod tests2 {
+    use pageserver_api::key::DBDIR_KEY;
+    use tracing::info;
+
+    use super::*;
+    use crate::tenant::storage_layer::IoConcurrency;
+
+    /// TODO: currently this test relies on manual visual inspection of the --no-capture output.
+    /// Should look like so:
+    /// ```text
+    /// RUST_LOG=trace cargo nextest run  --features testing  --no-capture test_io_concurrency_noise
+    /// running 1 test
+    /// 2025-01-21T17:42:01.335679Z  INFO get_vectored_concurrent_io test selected=SidecarTask
+    /// 2025-01-21T17:42:01.335680Z TRACE spawning sidecar task task_id=0
+    /// 2025-01-21T17:42:01.335937Z TRACE IoConcurrency_sidecar{task_id=0}: start
+    /// 2025-01-21T17:42:01.335972Z TRACE IoConcurrency_sidecar{task_id=0}: received new io future
+    /// 2025-01-21T17:42:01.335999Z  INFO IoConcurrency_sidecar{task_id=0}: waiting for signal to complete IO
+    /// 2025-01-21T17:42:01.336229Z  WARN dropping ValuesReconstructState while some IOs have not been completed num_active_ios=1 sidecar_task_id=Some(0) backtrace=   0: <pageserver::tenant::storage_layer::ValuesReconstructState as core::ops::drop::Drop>::drop
+    ///              at ./src/tenant/storage_layer.rs:553:24
+    ///    1: core::ptr::drop_in_place<pageserver::tenant::storage_layer::ValuesReconstructState>
+    ///              at /home/christian/.rustup/toolchains/1.84.0-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/core/src/ptr/mod.rs:521:1
+    ///    2: core::mem::drop
+    ///              at /home/christian/.rustup/toolchains/1.84.0-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/core/src/mem/mod.rs:942:24
+    ///    3: pageserver::tenant::storage_layer::tests2::test_io_concurrency_noise::{{closure}}
+    ///              at ./src/tenant/storage_layer.rs:1159:9
+    ///   ...
+    ///   49: <unknown>
+    /// 2025-01-21T17:42:01.452293Z  INFO IoConcurrency_sidecar{task_id=0}: completing IO
+    /// 2025-01-21T17:42:01.452357Z TRACE IoConcurrency_sidecar{task_id=0}: io future completed
+    /// 2025-01-21T17:42:01.452473Z TRACE IoConcurrency_sidecar{task_id=0}: end
+    /// test tenant::storage_layer::tests2::test_io_concurrency_noise ... ok
+    ///
+    /// ```
+    #[tokio::test]
+    async fn test_io_concurrency_noise() {
+        crate::tenant::harness::setup_logging();
+
+        let io_concurrency = IoConcurrency::spawn_for_test();
+        match *io_concurrency {
+            IoConcurrency::Sequential => {
+                // This test asserts behavior in sidecar mode, doesn't make sense in sequential mode.
+                return;
+            }
+            IoConcurrency::SidecarTask { .. } => {}
+        }
+        let mut reconstruct_state = ValuesReconstructState::new(io_concurrency.clone());
+
+        let (io_fut_is_waiting_tx, io_fut_is_waiting) = tokio::sync::oneshot::channel();
+        let (do_complete_io, should_complete_io) = tokio::sync::oneshot::channel();
+        let (io_fut_exiting_tx, io_fut_exiting) = tokio::sync::oneshot::channel();
+
+        let io = reconstruct_state.update_key(&DBDIR_KEY, Lsn(8), true);
+        reconstruct_state
+            .spawn_io(async move {
+                info!("waiting for signal to complete IO");
+                io_fut_is_waiting_tx.send(()).unwrap();
+                should_complete_io.await.unwrap();
+                info!("completing IO");
+                io.complete(Ok(OnDiskValue::RawImage(Bytes::new())));
+                io_fut_exiting_tx.send(()).unwrap();
+            })
+            .await;
+
+        io_fut_is_waiting.await.unwrap();
+
+        // this is what makes the noise
+        drop(reconstruct_state);
+
+        do_complete_io.send(()).unwrap();
+
+        io_fut_exiting.await.unwrap();
+    }
+}

pageserver/src/tenant/storage_layer/delta_layer.rs

@@ -41,13 +41,12 @@ use crate::tenant::vectored_blob_io::{
     BlobFlag, BufView, StreamingVectoredReadPlanner, VectoredBlobReader, VectoredRead,
     VectoredReadPlanner,
 };
-use crate::tenant::PageReconstructError;
 use crate::virtual_file::owned_buffers_io::io_buf_ext::{FullSlice, IoBufExt};
 use crate::virtual_file::IoBufferMut;
 use crate::virtual_file::{self, MaybeFatalIo, VirtualFile};
 use crate::TEMP_FILE_SUFFIX;
 use crate::{DELTA_FILE_MAGIC, STORAGE_FORMAT_VERSION};
-use anyhow::{anyhow, bail, ensure, Context, Result};
+use anyhow::{bail, ensure, Context, Result};
 use camino::{Utf8Path, Utf8PathBuf};
 use futures::StreamExt;
 use itertools::Itertools;
@@ -60,7 +59,7 @@ use pageserver_api::shard::TenantShardId;
 use pageserver_api::value::Value;
 use rand::{distributions::Alphanumeric, Rng};
 use serde::{Deserialize, Serialize};
-use std::collections::VecDeque;
+use std::collections::{HashMap, VecDeque};
 use std::fs::File;
 use std::io::SeekFrom;
 use std::ops::Range;
@@ -77,7 +76,10 @@ use utils::{
     lsn::Lsn,
 };
 
-use super::{AsLayerDesc, LayerName, PersistentLayerDesc, ValuesReconstructState};
+use super::{
+    AsLayerDesc, LayerName, OnDiskValue, OnDiskValueIo, PersistentLayerDesc, ResidentLayer,
+    ValuesReconstructState,
+};
 
 ///
 /// Header stored in the beginning of the file
@@ -226,7 +228,7 @@ pub struct DeltaLayerInner {
     index_start_blk: u32,
     index_root_blk: u32,
 
-    file: VirtualFile,
+    file: Arc<VirtualFile>,
     file_id: FileId,
 
     layer_key_range: Range<Key>,
@@ -795,9 +797,11 @@ impl DeltaLayerInner {
         max_vectored_read_bytes: Option<MaxVectoredReadBytes>,
         ctx: &RequestContext,
     ) -> anyhow::Result<Self> {
-        let file = VirtualFile::open_v2(path, ctx)
-            .await
-            .context("open layer file")?;
+        let file = Arc::new(
+            VirtualFile::open_v2(path, ctx)
+                .await
+                .context("open layer file")?,
+        );
 
         let file_id = page_cache::next_file_id();
 
@@ -842,12 +846,11 @@ impl DeltaLayerInner {
     // Look up the keys in the provided keyspace and update
     // the reconstruct state with whatever is found.
     //
-    // If the key is cached, go no further than the cached Lsn.
-    //
     // Currently, the index is visited for each range, but this
     // can be further optimised to visit the index only once.
     pub(super) async fn get_values_reconstruct_data(
         &self,
+        this: ResidentLayer,
         keyspace: KeySpace,
         lsn_range: Range<Lsn>,
         reconstruct_state: &mut ValuesReconstructState,
@@ -875,17 +878,14 @@ impl DeltaLayerInner {
             data_end_offset,
             index_reader,
             planner,
-            reconstruct_state,
             ctx,
         )
         .await
         .map_err(GetVectoredError::Other)?;
 
-        self.do_reads_and_update_state(reads, reconstruct_state, ctx)
+        self.do_reads_and_update_state(this, reads, reconstruct_state, ctx)
             .await;
 
-        reconstruct_state.on_lsn_advanced(&keyspace, lsn_range.start);
-
         Ok(())
     }
 
@@ -895,7 +895,6 @@ impl DeltaLayerInner {
         data_end_offset: u64,
         index_reader: DiskBtreeReader<Reader, DELTA_KEY_SIZE>,
         mut planner: VectoredReadPlanner,
-        reconstruct_state: &mut ValuesReconstructState,
         ctx: &RequestContext,
     ) -> anyhow::Result<Vec<VectoredRead>>
     where
@@ -922,10 +921,9 @@ impl DeltaLayerInner {
                 assert!(key >= range.start);
 
                 let outside_lsn_range = !lsn_range.contains(&lsn);
-                let below_cached_lsn = reconstruct_state.get_cached_lsn(&key) >= Some(lsn);
 
                 let flag = {
-                    if outside_lsn_range || below_cached_lsn {
+                    if outside_lsn_range {
                         BlobFlag::Ignore
                     } else if blob_ref.will_init() {
                         BlobFlag::ReplaceAll
@@ -994,98 +992,78 @@ impl DeltaLayerInner {
 
     async fn do_reads_and_update_state(
         &self,
+        this: ResidentLayer,
         reads: Vec<VectoredRead>,
         reconstruct_state: &mut ValuesReconstructState,
         ctx: &RequestContext,
     ) {
-        let vectored_blob_reader = VectoredBlobReader::new(&self.file);
-        let mut ignore_key_with_err = None;
-
         let max_vectored_read_bytes = self
             .max_vectored_read_bytes
             .expect("Layer is loaded with max vectored bytes config")
             .0
             .into();
         let buf_size = Self::get_min_read_buffer_size(&reads, max_vectored_read_bytes);
-        let mut buf = Some(IoBufferMut::with_capacity(buf_size));
 
         // Note that reads are processed in reverse order (from highest key+lsn).
         // This is the order that `ReconstructState` requires such that it can
         // track when a key is done.
         for read in reads.into_iter().rev() {
-            let res = vectored_blob_reader
-                .read_blobs(&read, buf.take().expect("Should have a buffer"), ctx)
-                .await;
-
-            let blobs_buf = match res {
-                Ok(blobs_buf) => blobs_buf,
-                Err(err) => {
-                    let kind = err.kind();
-                    for (_, blob_meta) in read.blobs_at.as_slice() {
-                        reconstruct_state.on_key_error(
-                            blob_meta.key,
-                            PageReconstructError::Other(anyhow!(
-                                "Failed to read blobs from virtual file {}: {}",
-                                self.file.path(),
-                                kind
-                            )),
-                        );
-                    }
-
-                    // We have "lost" the buffer since the lower level IO api
-                    // doesn't return the buffer on error. Allocate a new one.
-                    buf = Some(IoBufferMut::with_capacity(buf_size));
-
-                    continue;
-                }
-            };
-            let view = BufView::new_slice(&blobs_buf.buf);
-            for meta in blobs_buf.blobs.iter().rev() {
-                if Some(meta.meta.key) == ignore_key_with_err {
-                    continue;
-                }
-                let blob_read = meta.read(&view).await;
-                let blob_read = match blob_read {
-                    Ok(buf) => buf,
-                    Err(e) => {
-                        reconstruct_state.on_key_error(
-                            meta.meta.key,
-                            PageReconstructError::Other(anyhow!(e).context(format!(
-                                "Failed to decompress blob from virtual file {}",
-                                self.file.path(),
-                            ))),
-                        );
-
-                        ignore_key_with_err = Some(meta.meta.key);
-                        continue;
-                    }
-                };
-
-                let value = Value::des(&blob_read);
-
-                let value = match value {
-                    Ok(v) => v,
-                    Err(e) => {
-                        reconstruct_state.on_key_error(
-                            meta.meta.key,
-                            PageReconstructError::Other(anyhow!(e).context(format!(
-                                "Failed to deserialize blob from virtual file {}",
-                                self.file.path(),
-                            ))),
-                        );
-
-                        ignore_key_with_err = Some(meta.meta.key);
-                        continue;
-                    }
-                };
-
-                // Invariant: once a key reaches [`ValueReconstructSituation::Complete`]
-                // state, no further updates shall be made to it. The call below will
-                // panic if the invariant is violated.
-                reconstruct_state.update_key(&meta.meta.key, meta.meta.lsn, value);
+            let mut ios: HashMap<(Key, Lsn), OnDiskValueIo> = Default::default();
+            for (_, blob_meta) in read.blobs_at.as_slice().iter().rev() {
+                let io = reconstruct_state.update_key(
+                    &blob_meta.key,
+                    blob_meta.lsn,
+                    blob_meta.will_init,
+                );
+                ios.insert((blob_meta.key, blob_meta.lsn), io);
             }
 
-            buf = Some(blobs_buf.buf);
+            let read_extend_residency = this.clone();
+            let read_from = self.file.clone();
+            let read_ctx = ctx.attached_child();
+            reconstruct_state
+                .spawn_io(async move {
+                    let vectored_blob_reader = VectoredBlobReader::new(&read_from);
+                    let buf = IoBufferMut::with_capacity(buf_size);
+
+                    let res = vectored_blob_reader.read_blobs(&read, buf, &read_ctx).await;
+                    match res {
+                        Ok(blobs_buf) => {
+                            let view = BufView::new_slice(&blobs_buf.buf);
+                            for meta in blobs_buf.blobs.iter().rev() {
+                                let io = ios.remove(&(meta.meta.key, meta.meta.lsn)).unwrap();
+
+                                let blob_read = meta.read(&view).await;
+                                let blob_read = match blob_read {
+                                    Ok(buf) => buf,
+                                    Err(e) => {
+                                        io.complete(Err(e));
+                                        continue;
+                                    }
+                                };
+
+                                io.complete(Ok(OnDiskValue::WalRecordOrImage(
+                                    blob_read.into_bytes(),
+                                )));
+                            }
+
+                            assert!(ios.is_empty());
+                        }
+                        Err(err) => {
+                            for (_, sender) in ios {
+                                sender.complete(Err(std::io::Error::new(
+                                    err.kind(),
+                                    "vec read failed",
+                                )));
+                            }
+                        }
+                    }
+
+                    // keep layer resident until this IO is done; this spawned IO future generally outlives the
+                    // call to `self` / the `Arc<DownloadedLayer>` / the `ResidentLayer` that guarantees residency
+                    drop(read_extend_residency);
+                })
+                .await;
         }
     }
 
@@ -1224,7 +1202,14 @@ impl DeltaLayerInner {
             let actionable = if let Some((key, lsn, start_offset)) = prev.take() {
                 let end_offset = offset;
 
-                Some((BlobMeta { key, lsn }, start_offset..end_offset))
+                Some((
+                    BlobMeta {
+                        key,
+                        lsn,
+                        will_init: false,
+                    },
+                    start_offset..end_offset,
+                ))
             } else {
                 None
             };
@@ -1560,7 +1545,9 @@ impl DeltaLayerIterator<'_> {
                 let lsn = DeltaKey::extract_lsn_from_buf(&raw_key);
                 let blob_ref = BlobRef(value);
                 let offset = blob_ref.pos();
-                if let Some(batch_plan) = self.planner.handle(key, lsn, offset) {
+                if let Some(batch_plan) =
+                    self.planner.handle(key, lsn, offset, blob_ref.will_init())
+                {
                     break batch_plan;
                 }
             } else {
@@ -1673,7 +1660,6 @@ pub(crate) mod test {
             .expect("In memory disk finish should never fail");
         let reader = DiskBtreeReader::<_, DELTA_KEY_SIZE>::new(0, root_offset, disk);
         let planner = VectoredReadPlanner::new(100);
-        let mut reconstruct_state = ValuesReconstructState::new();
         let ctx = RequestContext::new(TaskKind::UnitTest, DownloadBehavior::Error);
 
         let keyspace = KeySpace {
@@ -1691,7 +1677,6 @@ pub(crate) mod test {
             disk_offset,
             reader,
             planner,
-            &mut reconstruct_state,
             &ctx,
         )
         .await
@@ -1935,7 +1920,6 @@ pub(crate) mod test {
             );
 
             let planner = VectoredReadPlanner::new(constants::MAX_VECTORED_READ_BYTES);
-            let mut reconstruct_state = ValuesReconstructState::new();
             let keyspace = pick_random_keyspace(rng, &entries_meta.key_range);
             let data_end_offset = inner.index_start_blk as u64 * PAGE_SZ as u64;
 
@@ -1945,7 +1929,6 @@ pub(crate) mod test {
                 data_end_offset,
                 index_reader,
                 planner,
-                &mut reconstruct_state,
                 &ctx,
             )
             .await?;

pageserver/src/tenant/storage_layer/image_layer.rs

@@ -38,12 +38,11 @@ use crate::tenant::vectored_blob_io::{
     BlobFlag, BufView, StreamingVectoredReadPlanner, VectoredBlobReader, VectoredRead,
     VectoredReadPlanner,
 };
-use crate::tenant::PageReconstructError;
 use crate::virtual_file::owned_buffers_io::io_buf_ext::IoBufExt;
 use crate::virtual_file::IoBufferMut;
 use crate::virtual_file::{self, MaybeFatalIo, VirtualFile};
 use crate::{IMAGE_FILE_MAGIC, STORAGE_FORMAT_VERSION, TEMP_FILE_SUFFIX};
-use anyhow::{anyhow, bail, ensure, Context, Result};
+use anyhow::{bail, ensure, Context, Result};
 use bytes::Bytes;
 use camino::{Utf8Path, Utf8PathBuf};
 use hex;
@@ -56,12 +55,13 @@ use pageserver_api::shard::{ShardIdentity, TenantShardId};
 use pageserver_api::value::Value;
 use rand::{distributions::Alphanumeric, Rng};
 use serde::{Deserialize, Serialize};
-use std::collections::VecDeque;
+use std::collections::{HashMap, VecDeque};
 use std::fs::File;
 use std::io::SeekFrom;
 use std::ops::Range;
 use std::os::unix::prelude::FileExt;
 use std::str::FromStr;
+use std::sync::Arc;
 use tokio::sync::OnceCell;
 use tokio_stream::StreamExt;
 use tracing::*;
@@ -73,7 +73,10 @@ use utils::{
 };
 
 use super::layer_name::ImageLayerName;
-use super::{AsLayerDesc, LayerName, PersistentLayerDesc, ValuesReconstructState};
+use super::{
+    AsLayerDesc, LayerName, OnDiskValue, OnDiskValueIo, PersistentLayerDesc, ResidentLayer,
+    ValuesReconstructState,
+};
 
 ///
 /// Header stored in the beginning of the file
@@ -164,7 +167,7 @@ pub struct ImageLayerInner {
     key_range: Range<Key>,
     lsn: Lsn,
 
-    file: VirtualFile,
+    file: Arc<VirtualFile>,
     file_id: FileId,
 
     max_vectored_read_bytes: Option<MaxVectoredReadBytes>,
@@ -391,9 +394,11 @@ impl ImageLayerInner {
         max_vectored_read_bytes: Option<MaxVectoredReadBytes>,
         ctx: &RequestContext,
     ) -> anyhow::Result<Self> {
-        let file = VirtualFile::open_v2(path, ctx)
-            .await
-            .context("open layer file")?;
+        let file = Arc::new(
+            VirtualFile::open_v2(path, ctx)
+                .await
+                .context("open layer file")?,
+        );
         let file_id = page_cache::next_file_id();
         let block_reader = FileBlockReader::new(&file, file_id);
         let summary_blk = block_reader
@@ -439,6 +444,7 @@ impl ImageLayerInner {
     // the reconstruct state with whatever is found.
     pub(super) async fn get_values_reconstruct_data(
         &self,
+        this: ResidentLayer,
         keyspace: KeySpace,
         reconstruct_state: &mut ValuesReconstructState,
         ctx: &RequestContext,
@@ -448,7 +454,7 @@ impl ImageLayerInner {
             .await
             .map_err(GetVectoredError::Other)?;
 
-        self.do_reads_and_update_state(reads, reconstruct_state, ctx)
+        self.do_reads_and_update_state(this, reads, reconstruct_state, ctx)
             .await;
 
         reconstruct_state.on_image_layer_visited(&self.key_range);
@@ -570,6 +576,7 @@ impl ImageLayerInner {
 
     async fn do_reads_and_update_state(
         &self,
+        this: ResidentLayer,
         reads: Vec<VectoredRead>,
         reconstruct_state: &mut ValuesReconstructState,
         ctx: &RequestContext,
@@ -580,8 +587,13 @@ impl ImageLayerInner {
             .0
             .into();
 
-        let vectored_blob_reader = VectoredBlobReader::new(&self.file);
         for read in reads.into_iter() {
+            let mut ios: HashMap<(Key, Lsn), OnDiskValueIo> = Default::default();
+            for (_, blob_meta) in read.blobs_at.as_slice() {
+                let io = reconstruct_state.update_key(&blob_meta.key, blob_meta.lsn, true);
+                ios.insert((blob_meta.key, blob_meta.lsn), io);
+            }
+
             let buf_size = read.size();
 
             if buf_size > max_vectored_read_bytes {
@@ -611,50 +623,51 @@ impl ImageLayerInner {
                 }
             }
 
-            let buf = IoBufferMut::with_capacity(buf_size);
-            let res = vectored_blob_reader.read_blobs(&read, buf, ctx).await;
+            let read_extend_residency = this.clone();
+            let read_from = self.file.clone();
+            let read_ctx = ctx.attached_child();
+            reconstruct_state
+                .spawn_io(async move {
+                    let buf = IoBufferMut::with_capacity(buf_size);
+                    let vectored_blob_reader = VectoredBlobReader::new(&read_from);
+                    let res = vectored_blob_reader.read_blobs(&read, buf, &read_ctx).await;
 
-            match res {
-                Ok(blobs_buf) => {
-                    let view = BufView::new_slice(&blobs_buf.buf);
-                    for meta in blobs_buf.blobs.iter() {
-                        let img_buf = meta.read(&view).await;
+                    match res {
+                        Ok(blobs_buf) => {
+                            let view = BufView::new_slice(&blobs_buf.buf);
+                            for meta in blobs_buf.blobs.iter() {
+                                let io: OnDiskValueIo =
+                                    ios.remove(&(meta.meta.key, meta.meta.lsn)).unwrap();
+                                let img_buf = meta.read(&view).await;
 
-                        let img_buf = match img_buf {
-                            Ok(img_buf) => img_buf,
-                            Err(e) => {
-                                reconstruct_state.on_key_error(
-                                    meta.meta.key,
-                                    PageReconstructError::Other(anyhow!(e).context(format!(
-                                        "Failed to decompress blob from virtual file {}",
-                                        self.file.path(),
-                                    ))),
-                                );
+                                let img_buf = match img_buf {
+                                    Ok(img_buf) => img_buf,
+                                    Err(e) => {
+                                        io.complete(Err(e));
+                                        continue;
+                                    }
+                                };
 
-                                continue;
+                                io.complete(Ok(OnDiskValue::RawImage(img_buf.into_bytes())));
                             }
-                        };
-                        reconstruct_state.update_key(
-                            &meta.meta.key,
-                            self.lsn,
-                            Value::Image(img_buf.into_bytes()),
-                        );
+
+                            assert!(ios.is_empty());
+                        }
+                        Err(err) => {
+                            for (_, io) in ios {
+                                io.complete(Err(std::io::Error::new(
+                                    err.kind(),
+                                    "vec read failed",
+                                )));
+                            }
+                        }
                     }
-                }
-                Err(err) => {
-                    let kind = err.kind();
-                    for (_, blob_meta) in read.blobs_at.as_slice() {
-                        reconstruct_state.on_key_error(
-                            blob_meta.key,
-                            PageReconstructError::from(anyhow!(
-                                "Failed to read blobs from virtual file {}: {}",
-                                self.file.path(),
-                                kind
-                            )),
-                        );
-                    }
-                }
-            };
+
+                    // keep layer resident until this IO is done; this spawned IO future generally outlives the
+                    // call to `self` / the `Arc<DownloadedLayer>` / the `ResidentLayer` that guarantees residency
+                    drop(read_extend_residency);
+                })
+                .await;
         }
     }
 
@@ -1069,6 +1082,7 @@ impl ImageLayerIterator<'_> {
                     Key::from_slice(&raw_key[..KEY_SIZE]),
                     self.image_layer.lsn,
                     offset,
+                    true,
                 ) {
                     break batch_plan;
                 }

pageserver/src/tenant/storage_layer/inmemory_layer.rs

@@ -8,23 +8,22 @@ use crate::assert_u64_eq_usize::{u64_to_usize, U64IsUsize, UsizeIsU64};
 use crate::config::PageServerConf;
 use crate::context::{PageContentKind, RequestContext, RequestContextBuilder};
 use crate::tenant::ephemeral_file::EphemeralFile;
+use crate::tenant::storage_layer::{OnDiskValue, OnDiskValueIo};
 use crate::tenant::timeline::GetVectoredError;
-use crate::tenant::PageReconstructError;
 use crate::virtual_file::owned_buffers_io::io_buf_ext::IoBufExt;
 use crate::{l0_flush, page_cache};
-use anyhow::{anyhow, Result};
+use anyhow::Result;
 use camino::Utf8PathBuf;
 use pageserver_api::key::CompactKey;
 use pageserver_api::key::Key;
 use pageserver_api::keyspace::KeySpace;
 use pageserver_api::models::InMemoryLayerInfo;
 use pageserver_api::shard::TenantShardId;
-use pageserver_api::value::Value;
 use std::collections::{BTreeMap, HashMap};
 use std::sync::{Arc, OnceLock};
 use std::time::Instant;
 use tracing::*;
-use utils::{bin_ser::BeSer, id::TimelineId, lsn::Lsn, vec_map::VecMap};
+use utils::{id::TimelineId, lsn::Lsn, vec_map::VecMap};
 use wal_decoder::serialized_batch::{SerializedValueBatch, SerializedValueMeta, ValueMeta};
 // avoid binding to Write (conflicts with std::io::Write)
 // while being able to use std::fmt::Write's methods
@@ -36,9 +35,7 @@ use std::sync::atomic::Ordering as AtomicOrdering;
 use std::sync::atomic::{AtomicU64, AtomicUsize};
 use tokio::sync::RwLock;
 
-use super::{
-    DeltaLayerWriter, PersistentLayerDesc, ValueReconstructSituation, ValuesReconstructState,
-};
+use super::{DeltaLayerWriter, PersistentLayerDesc, ValuesReconstructState};
 
 pub(crate) mod vectored_dio_read;
 
@@ -415,10 +412,8 @@ impl InMemoryLayer {
 
     // Look up the keys in the provided keyspace and update
     // the reconstruct state with whatever is found.
-    //
-    // If the key is cached, go no further than the cached Lsn.
     pub(crate) async fn get_values_reconstruct_data(
-        &self,
+        self: &Arc<InMemoryLayer>,
         keyspace: KeySpace,
         end_lsn: Lsn,
         reconstruct_state: &mut ValuesReconstructState,
@@ -435,6 +430,9 @@ impl InMemoryLayer {
             read: vectored_dio_read::LogicalRead<Vec<u8>>,
         }
         let mut reads: HashMap<Key, Vec<ValueRead>> = HashMap::new();
+        let mut ios: HashMap<(Key, Lsn), OnDiskValueIo> = Default::default();
+
+        let lsn_range = self.start_lsn..end_lsn;
 
         for range in keyspace.ranges.iter() {
             for (key, vec_map) in inner
@@ -442,12 +440,7 @@ impl InMemoryLayer {
                 .range(range.start.to_compact()..range.end.to_compact())
             {
                 let key = Key::from_compact(*key);
-                let lsn_range = match reconstruct_state.get_cached_lsn(&key) {
-                    Some(cached_lsn) => (cached_lsn + 1)..end_lsn,
-                    None => self.start_lsn..end_lsn,
-                };
-
-                let slice = vec_map.slice_range(lsn_range);
+                let slice = vec_map.slice_range(lsn_range.clone());
 
                 for (entry_lsn, index_entry) in slice.iter().rev() {
                     let IndexEntryUnpacked {
@@ -463,55 +456,59 @@ impl InMemoryLayer {
                             Vec::with_capacity(len as usize),
                         ),
                     });
+
+                    let io = reconstruct_state.update_key(&key, *entry_lsn, will_init);
+                    ios.insert((key, *entry_lsn), io);
+
                     if will_init {
                         break;
                     }
                 }
             }
         }
+        drop(inner); // release the lock before we spawn the IO; if it's serial-mode IO we will deadlock on the read().await below
+        let read_from = Arc::clone(self);
+        let read_ctx = ctx.attached_child();
+        reconstruct_state
+            .spawn_io(async move {
+                let inner = read_from.inner.read().await;
+                let f = vectored_dio_read::execute(
+                    &inner.file,
+                    reads
+                        .iter()
+                        .flat_map(|(_, value_reads)| value_reads.iter().map(|v| &v.read)),
+                    &read_ctx,
+                );
+                send_future::SendFuture::send(f) // https://github.com/rust-lang/rust/issues/96865
+                    .await;
 
-        // Execute the reads.
-
-        let f = vectored_dio_read::execute(
-            &inner.file,
-            reads
-                .iter()
-                .flat_map(|(_, value_reads)| value_reads.iter().map(|v| &v.read)),
-            &ctx,
-        );
-        send_future::SendFuture::send(f) // https://github.com/rust-lang/rust/issues/96865
-            .await;
-
-        // Process results into the reconstruct state
-        'next_key: for (key, value_reads) in reads {
-            for ValueRead { entry_lsn, read } in value_reads {
-                match read.into_result().expect("we run execute() above") {
-                    Err(e) => {
-                        reconstruct_state.on_key_error(key, PageReconstructError::from(anyhow!(e)));
-                        continue 'next_key;
-                    }
-                    Ok(value_buf) => {
-                        let value = Value::des(&value_buf);
-                        if let Err(e) = value {
-                            reconstruct_state
-                                .on_key_error(key, PageReconstructError::from(anyhow!(e)));
-                            continue 'next_key;
+                for (key, value_reads) in reads {
+                    for ValueRead { entry_lsn, read } in value_reads {
+                        let io = ios.remove(&(key, entry_lsn)).expect("sender must exist");
+                        match read.into_result().expect("we run execute() above") {
+                            Err(e) => {
+                                io.complete(Err(std::io::Error::new(
+                                    e.kind(),
+                                    "dio vec read failed",
+                                )));
+                            }
+                            Ok(value_buf) => {
+                                io.complete(Ok(OnDiskValue::WalRecordOrImage(value_buf.into())));
+                            }
                         }
-
-                        let key_situation =
-                            reconstruct_state.update_key(&key, entry_lsn, value.unwrap());
-                        if key_situation == ValueReconstructSituation::Complete {
-                            // TODO: metric to see if we fetched more values than necessary
-                            continue 'next_key;
-                        }
-
-                        // process the next value in the next iteration of the loop
                     }
                 }
-            }
-        }
 
-        reconstruct_state.on_lsn_advanced(&keyspace, self.start_lsn);
+                assert!(ios.is_empty());
+
+                // Keep layer existent until this IO is done;
+                // This is kinda forced for InMemoryLayer because we need to inner.read() anyway,
+                // but it's less obvious for DeltaLayer and ImageLayer. So, keep this explicit
+                // drop for consistency among all three layer types.
+                drop(inner);
+                drop(read_from);
+            })
+            .await;
 
         Ok(())
     }
@@ -606,6 +603,7 @@ impl InMemoryLayer {
         // Write the batch to the file
         inner.file.write_raw(&raw, ctx).await?;
         let new_size = inner.file.len();
+
         let expected_new_len = base_offset
             .checked_add(raw.len().into_u64())
             // write_raw would error if we were to overflow u64.

pageserver/src/tenant/storage_layer/layer.rs

@@ -308,7 +308,7 @@ impl Layer {
         reconstruct_data: &mut ValuesReconstructState,
         ctx: &RequestContext,
     ) -> Result<(), GetVectoredError> {
-        let layer = self
+        let downloaded = self
             .0
             .get_or_maybe_download(true, Some(ctx))
             .await
@@ -318,11 +318,15 @@ impl Layer {
                 }
                 other => GetVectoredError::Other(anyhow::anyhow!(other)),
             })?;
+        let this = ResidentLayer {
+            downloaded: downloaded.clone(),
+            owner: self.clone(),
+        };
 
         self.record_access(ctx);
 
-        layer
-            .get_values_reconstruct_data(keyspace, lsn_range, reconstruct_data, &self.0, ctx)
+        downloaded
+            .get_values_reconstruct_data(this, keyspace, lsn_range, reconstruct_data, ctx)
             .instrument(tracing::debug_span!("get_values_reconstruct_data", layer=%self))
             .await
             .map_err(|err| match err {
@@ -1768,25 +1772,25 @@ impl DownloadedLayer {
 
     async fn get_values_reconstruct_data(
         &self,
+        this: ResidentLayer,
         keyspace: KeySpace,
         lsn_range: Range<Lsn>,
         reconstruct_data: &mut ValuesReconstructState,
-        owner: &Arc<LayerInner>,
         ctx: &RequestContext,
     ) -> Result<(), GetVectoredError> {
         use LayerKind::*;
 
         match self
-            .get(owner, ctx)
+            .get(&this.owner.0, ctx)
             .await
             .map_err(GetVectoredError::Other)?
         {
             Delta(d) => {
-                d.get_values_reconstruct_data(keyspace, lsn_range, reconstruct_data, ctx)
+                d.get_values_reconstruct_data(this, keyspace, lsn_range, reconstruct_data, ctx)
                     .await
             }
             Image(i) => {
-                i.get_values_reconstruct_data(keyspace, reconstruct_data, ctx)
+                i.get_values_reconstruct_data(this, keyspace, reconstruct_data, ctx)
                     .await
             }
         }

pageserver/src/tenant/storage_layer/layer/tests.rs

@@ -1,6 +1,6 @@
 use std::time::UNIX_EPOCH;
 
-use pageserver_api::key::CONTROLFILE_KEY;
+use pageserver_api::key::{Key, CONTROLFILE_KEY};
 use tokio::task::JoinSet;
 use utils::{
     completion::{self, Completion},
@@ -9,7 +9,13 @@ use utils::{
 
 use super::failpoints::{Failpoint, FailpointKind};
 use super::*;
-use crate::{context::DownloadBehavior, tenant::storage_layer::LayerVisibilityHint};
+use crate::{
+    context::DownloadBehavior,
+    tenant::{
+        harness::test_img,
+        storage_layer::{IoConcurrency, LayerVisibilityHint},
+    },
+};
 use crate::{task_mgr::TaskKind, tenant::harness::TenantHarness};
 
 /// Used in tests to advance a future to wanted await point, and not futher.
@@ -28,23 +34,55 @@ async fn smoke_test() {
     let span = h.span();
     let download_span = span.in_scope(|| tracing::info_span!("downloading", timeline_id = 1));
     let (tenant, _) = h.load().await;
+    let io_concurrency = IoConcurrency::spawn_for_test();
 
     let ctx = RequestContext::new(TaskKind::UnitTest, DownloadBehavior::Download);
 
+    let image_layers = vec![(
+        Lsn(0x40),
+        vec![(
+            Key::from_hex("620000000033333333444444445500000000").unwrap(),
+            test_img("foo"),
+        )],
+    )];
+
+    // Create a test timeline with one real layer, and one synthetic test layer.  The synthetic
+    // one is only there so that we can GC the real one without leaving the timeline's metadata
+    // empty, which is an illegal state (see [`IndexPart::validate`]).
     let timeline = tenant
-        .create_test_timeline(TimelineId::generate(), Lsn(0x10), 14, &ctx)
+        .create_test_timeline_with_layers(
+            TimelineId::generate(),
+            Lsn(0x10),
+            14,
+            &ctx,
+            Default::default(),
+            image_layers,
+            Lsn(0x100),
+        )
         .await
         .unwrap();
 
-    let layer = {
+    // Grab one of the timeline's layers to exercise in the test, and the other layer that is just
+    // there to avoid the timeline being illegally empty
+    let (layer, dummy_layer) = {
         let mut layers = {
             let layers = timeline.layers.read().await;
             layers.likely_resident_layers().cloned().collect::<Vec<_>>()
         };
 
-        assert_eq!(layers.len(), 1);
+        assert_eq!(layers.len(), 2);
 
-        layers.swap_remove(0)
+        layers.sort_by_key(|l| l.layer_desc().get_key_range().start);
+        let synthetic_layer = layers.pop().unwrap();
+        let real_layer = layers.pop().unwrap();
+        tracing::info!(
+            "real_layer={:?} ({}), synthetic_layer={:?} ({})",
+            real_layer,
+            real_layer.layer_desc().file_size,
+            synthetic_layer,
+            synthetic_layer.layer_desc().file_size
+        );
+        (real_layer, synthetic_layer)
     };
 
     // all layers created at pageserver are like `layer`, initialized with strong
@@ -55,7 +93,7 @@ async fn smoke_test() {
     };
 
     let img_before = {
-        let mut data = ValuesReconstructState::default();
+        let mut data = ValuesReconstructState::new(io_concurrency.clone());
         layer
             .get_values_reconstruct_data(
                 controlfile_keyspace.clone(),
@@ -65,10 +103,13 @@ async fn smoke_test() {
             )
             .await
             .unwrap();
+
         data.keys
             .remove(&CONTROLFILE_KEY)
             .expect("must be present")
-            .expect("should not error")
+            .collect_pending_ios()
+            .await
+            .expect("must not error")
             .img
             .take()
             .expect("tenant harness writes the control file")
@@ -87,7 +128,7 @@ async fn smoke_test() {
 
     // on accesses when the layer is evicted, it will automatically be downloaded.
     let img_after = {
-        let mut data = ValuesReconstructState::default();
+        let mut data = ValuesReconstructState::new(io_concurrency.clone());
         layer
             .get_values_reconstruct_data(
                 controlfile_keyspace.clone(),
@@ -101,7 +142,9 @@ async fn smoke_test() {
         data.keys
             .remove(&CONTROLFILE_KEY)
             .expect("must be present")
-            .expect("should not error")
+            .collect_pending_ios()
+            .await
+            .expect("must not error")
             .img
             .take()
             .expect("tenant harness writes the control file")
@@ -173,10 +216,13 @@ async fn smoke_test() {
 
     let rtc = &timeline.remote_client;
 
+    // Simulate GC removing our test layer.
     {
-        let layers = &[layer];
         let mut g = timeline.layers.write().await;
+
+        let layers = &[layer];
         g.open_mut().unwrap().finish_gc_timeline(layers);
+
         // this just updates the remote_physical_size for demonstration purposes
         rtc.schedule_gc_update(layers).unwrap();
     }
@@ -191,7 +237,10 @@ async fn smoke_test() {
 
     rtc.wait_completion().await.unwrap();
 
-    assert_eq!(rtc.get_remote_physical_size(), 0);
+    assert_eq!(
+        rtc.get_remote_physical_size(),
+        dummy_layer.metadata().file_size
+    );
     assert_eq!(0, LAYER_IMPL_METRICS.inits_cancelled.get())
 }
 

pageserver/src/tenant/tasks.rs

@@ -67,10 +67,9 @@ pub(crate) async fn concurrent_background_tasks_rate_limit_permit(
 ) -> tokio::sync::SemaphorePermit<'static> {
     let _guard = crate::metrics::BACKGROUND_LOOP_SEMAPHORE.measure_acquisition(loop_kind);
 
-    pausable_failpoint!(
-        "initial-size-calculation-permit-pause",
-        loop_kind == BackgroundLoopKind::InitialLogicalSizeCalculation
-    );
+    if loop_kind == BackgroundLoopKind::InitialLogicalSizeCalculation {
+        pausable_failpoint!("initial-size-calculation-permit-pause");
+    }
 
     // TODO: assert that we run on BACKGROUND_RUNTIME; requires tokio_unstable Handle::id();
     match CONCURRENT_BACKGROUND_TASKS.acquire().await {

pageserver/src/tenant/timeline.rs

@@ -20,12 +20,13 @@ use camino::Utf8Path;
 use chrono::{DateTime, Utc};
 use enumset::EnumSet;
 use fail::fail_point;
+use futures::{stream::FuturesUnordered, StreamExt};
 use handle::ShardTimelineId;
+use layer_manager::Shutdown;
 use offload::OffloadError;
 use once_cell::sync::Lazy;
 use pageserver_api::models::PageTraceEvent;
 use pageserver_api::{
-    config::tenant_conf_defaults::DEFAULT_COMPACTION_THRESHOLD,
     key::{
         KEY_SIZE, METADATA_KEY_BEGIN_PREFIX, METADATA_KEY_END_PREFIX, NON_INHERITED_RANGE,
         SPARSE_RANGE,
@@ -51,34 +52,35 @@ use tokio::{
 use tokio_util::sync::CancellationToken;
 use tracing::*;
 use utils::{
-    fs_ext, pausable_failpoint,
+    fs_ext,
+    guard_arc_swap::GuardArcSwap,
+    pausable_failpoint,
     postgres_client::PostgresClientProtocol,
     sync::gate::{Gate, GateGuard},
 };
 use wal_decoder::serialized_batch::{SerializedValueBatch, ValueMeta};
 
-use std::sync::atomic::Ordering as AtomicOrdering;
-use std::sync::{Arc, Mutex, RwLock, Weak};
+use std::array;
+use std::cmp::{max, min};
+use std::collections::btree_map::Entry;
+use std::collections::{BTreeMap, HashMap, HashSet};
+use std::ops::{ControlFlow, Deref, Range};
+use std::sync::atomic::{AtomicBool, AtomicU64, Ordering as AtomicOrdering};
+use std::sync::{Arc, Mutex, OnceLock, RwLock, Weak};
 use std::time::{Duration, Instant, SystemTime};
-use std::{
-    array,
-    collections::{BTreeMap, HashMap, HashSet},
-    sync::atomic::AtomicU64,
-};
-use std::{cmp::min, ops::ControlFlow};
-use std::{
-    collections::btree_map::Entry,
-    ops::{Deref, Range},
-};
-use std::{pin::pin, sync::OnceLock};
 
+use crate::l0_flush::{self, L0FlushGlobalState};
 use crate::{
     aux_file::AuxFileSizeEstimator,
+    page_service::TenantManagerTypes,
     tenant::{
         config::AttachmentMode,
         layer_map::{LayerMap, SearchResult},
         metadata::TimelineMetadata,
-        storage_layer::{inmemory_layer::IndexEntry, PersistentLayerDesc},
+        storage_layer::{
+            inmemory_layer::IndexEntry, IoConcurrency, PersistentLayerDesc,
+            ValueReconstructSituation,
+        },
     },
     walingest::WalLagCooldown,
     walredo,
@@ -99,10 +101,6 @@ use crate::{
 use crate::{
     disk_usage_eviction_task::EvictionCandidate, tenant::storage_layer::delta_layer::DeltaEntry,
 };
-use crate::{
-    l0_flush::{self, L0FlushGlobalState},
-    metrics::GetKind,
-};
 use crate::{
     metrics::ScanLatencyOngoingRecording, tenant::timeline::logical_size::CurrentLogicalSize,
 };
@@ -353,8 +351,8 @@ pub struct Timeline {
     // though let's keep them both for better error visibility.
     pub initdb_lsn: Lsn,
 
-    /// When did we last calculate the partitioning? Make it pub to test cases.
-    pub(super) partitioning: tokio::sync::Mutex<((KeyPartitioning, SparseKeyPartitioning), Lsn)>,
+    /// The repartitioning result. Allows a single writer and multiple readers.
+    pub(crate) partitioning: GuardArcSwap<((KeyPartitioning, SparseKeyPartitioning), Lsn)>,
 
     /// Configuration: how often should the partitioning be recalculated.
     repartition_threshold: u64,
@@ -404,6 +402,9 @@ pub struct Timeline {
     /// Timeline deletion will acquire both compaction and gc locks in whatever order.
     compaction_lock: tokio::sync::Mutex<()>,
 
+    /// If true, the last compaction failed.
+    compaction_failed: AtomicBool,
+
     /// Make sure we only have one running gc at a time.
     ///
     /// Must only be taken in two places:
@@ -429,7 +430,7 @@ pub struct Timeline {
 
     pub(crate) l0_flush_global_state: L0FlushGlobalState,
 
-    pub(crate) handles: handle::PerTimelineState<crate::page_service::TenantManagerTypes>,
+    pub(crate) handles: handle::PerTimelineState<TenantManagerTypes>,
 
     pub(crate) attach_wal_lag_cooldown: Arc<OnceLock<WalLagCooldown>>,
 
@@ -900,10 +901,17 @@ impl From<GetReadyAncestorError> for PageReconstructError {
     }
 }
 
+pub(crate) enum WaitLsnTimeout {
+    Custom(Duration),
+    // Use the [`PageServerConf::wait_lsn_timeout`] default
+    Default,
+}
+
 pub(crate) enum WaitLsnWaiter<'a> {
     Timeline(&'a Timeline),
     Tenant,
     PageService,
+    HttpEndpoint,
 }
 
 /// Argument to [`Timeline::shutdown`].
@@ -1002,9 +1010,7 @@ impl Timeline {
             ranges: vec![key..key.next()],
         };
 
-        // Initialise the reconstruct state for the key with the cache
-        // entry returned above.
-        let mut reconstruct_state = ValuesReconstructState::new();
+        let mut reconstruct_state = ValuesReconstructState::new(IoConcurrency::sequential());
 
         let vectored_res = self
             .get_vectored_impl(keyspace.clone(), lsn, &mut reconstruct_state, ctx)
@@ -1047,6 +1053,7 @@ impl Timeline {
         &self,
         keyspace: KeySpace,
         lsn: Lsn,
+        io_concurrency: super::storage_layer::IoConcurrency,
         ctx: &RequestContext,
     ) -> Result<BTreeMap<Key, Result<Bytes, PageReconstructError>>, GetVectoredError> {
         if !lsn.is_valid() {
@@ -1081,7 +1088,7 @@ impl Timeline {
             .get_vectored_impl(
                 keyspace.clone(),
                 lsn,
-                &mut ValuesReconstructState::new(),
+                &mut ValuesReconstructState::new(io_concurrency),
                 ctx,
             )
             .await;
@@ -1106,6 +1113,7 @@ impl Timeline {
         keyspace: KeySpace,
         lsn: Lsn,
         ctx: &RequestContext,
+        io_concurrency: super::storage_layer::IoConcurrency,
     ) -> Result<BTreeMap<Key, Result<Bytes, PageReconstructError>>, GetVectoredError> {
         if !lsn.is_valid() {
             return Err(GetVectoredError::InvalidLsn(lsn));
@@ -1137,7 +1145,7 @@ impl Timeline {
             .get_vectored_impl(
                 keyspace.clone(),
                 lsn,
-                &mut ValuesReconstructState::default(),
+                &mut ValuesReconstructState::new(io_concurrency),
                 ctx,
             )
             .await;
@@ -1149,46 +1157,63 @@ impl Timeline {
         vectored_res
     }
 
-    pub(super) async fn get_vectored_impl(
+    pub(crate) async fn get_vectored_impl(
         &self,
         keyspace: KeySpace,
         lsn: Lsn,
         reconstruct_state: &mut ValuesReconstructState,
         ctx: &RequestContext,
     ) -> Result<BTreeMap<Key, Result<Bytes, PageReconstructError>>, GetVectoredError> {
-        let get_kind = if keyspace.total_raw_size() == 1 {
-            GetKind::Singular
-        } else {
-            GetKind::Vectored
+        let traversal_res: Result<(), _> = self
+            .get_vectored_reconstruct_data(keyspace.clone(), lsn, reconstruct_state, ctx)
+            .await;
+        if let Err(err) = traversal_res {
+            // Wait for all the spawned IOs to complete.
+            // See comments on `spawn_io` inside `storage_layer` for more details.
+            let mut collect_futs = std::mem::take(&mut reconstruct_state.keys)
+                .into_values()
+                .map(|state| state.collect_pending_ios())
+                .collect::<FuturesUnordered<_>>();
+            while collect_futs.next().await.is_some() {}
+            return Err(err);
         };
 
-        let get_data_timer = crate::metrics::GET_RECONSTRUCT_DATA_TIME
-            .for_get_kind(get_kind)
-            .start_timer();
-        self.get_vectored_reconstruct_data(keyspace.clone(), lsn, reconstruct_state, ctx)
-            .await?;
-        get_data_timer.stop_and_record();
-
-        let reconstruct_timer = crate::metrics::RECONSTRUCT_TIME
-            .for_get_kind(get_kind)
-            .start_timer();
-        let mut results: BTreeMap<Key, Result<Bytes, PageReconstructError>> = BTreeMap::new();
         let layers_visited = reconstruct_state.get_layers_visited();
 
-        for (key, res) in std::mem::take(&mut reconstruct_state.keys) {
-            match res {
-                Err(err) => {
-                    results.insert(key, Err(err));
-                }
-                Ok(state) => {
-                    let state = ValueReconstructState::from(state);
+        let futs = FuturesUnordered::new();
+        for (key, state) in std::mem::take(&mut reconstruct_state.keys) {
+            futs.push({
+                let walredo_self = self.myself.upgrade().expect("&self method holds the arc");
+                async move {
+                    assert_eq!(state.situation, ValueReconstructSituation::Complete);
 
-                    let reconstruct_res = self.reconstruct_value(key, lsn, state).await;
-                    results.insert(key, reconstruct_res);
+                    let converted = match state.collect_pending_ios().await {
+                        Ok(ok) => ok,
+                        Err(err) => {
+                            return (key, Err(err));
+                        }
+                    };
+
+                    // The walredo module expects the records to be descending in terms of Lsn.
+                    // And we submit the IOs in that order, so, there shuold be no need to sort here.
+                    debug_assert!(
+                        converted
+                            .records
+                            .is_sorted_by_key(|(lsn, _)| std::cmp::Reverse(*lsn)),
+                        "{converted:?}"
+                    );
+
+                    (
+                        key,
+                        walredo_self.reconstruct_value(key, lsn, converted).await,
+                    )
                 }
-            }
+            });
         }
-        reconstruct_timer.stop_and_record();
+
+        let results = futs
+            .collect::<BTreeMap<Key, Result<Bytes, PageReconstructError>>>()
+            .await;
 
         // For aux file keys (v1 or v2) the vectored read path does not return an error
         // when they're missing. Instead they are omitted from the resulting btree
@@ -1283,6 +1308,7 @@ impl Timeline {
         &self,
         lsn: Lsn,
         who_is_waiting: WaitLsnWaiter<'_>,
+        timeout: WaitLsnTimeout,
         ctx: &RequestContext, /* Prepare for use by cancellation */
     ) -> Result<(), WaitLsnError> {
         let state = self.current_state();
@@ -1299,7 +1325,7 @@ impl Timeline {
                 | TaskKind::WalReceiverConnectionPoller => {
                     let is_myself = match who_is_waiting {
                         WaitLsnWaiter::Timeline(waiter) => Weak::ptr_eq(&waiter.myself, &self.myself),
-                        WaitLsnWaiter::Tenant | WaitLsnWaiter::PageService => unreachable!("tenant or page_service context are not expected to have task kind {:?}", ctx.task_kind()),
+                        WaitLsnWaiter::Tenant | WaitLsnWaiter::PageService | WaitLsnWaiter::HttpEndpoint => unreachable!("tenant or page_service context are not expected to have task kind {:?}", ctx.task_kind()),
                     };
                     if is_myself {
                         if let Err(current) = self.last_record_lsn.would_wait_for(lsn) {
@@ -1315,13 +1341,14 @@ impl Timeline {
             }
         }
 
+        let timeout = match timeout {
+            WaitLsnTimeout::Custom(t) => t,
+            WaitLsnTimeout::Default => self.conf.wait_lsn_timeout,
+        };
+
         let _timer = crate::metrics::WAIT_LSN_TIME.start_timer();
 
-        match self
-            .last_record_lsn
-            .wait_for_timeout(lsn, self.conf.wait_lsn_timeout)
-            .await
-        {
+        match self.last_record_lsn.wait_for_timeout(lsn, timeout).await {
             Ok(()) => Ok(()),
             Err(e) => {
                 use utils::seqwait::SeqWaitError::*;
@@ -1681,13 +1708,27 @@ impl Timeline {
             return Ok(false);
         }
 
-        match self.get_compaction_algorithm_settings().kind {
+        let result = match self.get_compaction_algorithm_settings().kind {
             CompactionAlgorithm::Tiered => {
                 self.compact_tiered(cancel, ctx).await?;
                 Ok(false)
             }
             CompactionAlgorithm::Legacy => self.compact_legacy(cancel, options, ctx).await,
-        }
+        };
+
+        // Signal compaction failure to avoid L0 flush stalls when it's broken.
+        match result {
+            Ok(_) => self.compaction_failed.store(false, AtomicOrdering::Relaxed),
+            Err(CompactionError::Other(_)) => {
+                self.compaction_failed.store(true, AtomicOrdering::Relaxed)
+            }
+            // Don't change the current value on offload failure or shutdown. We don't want to
+            // abruptly stall nor resume L0 flushes in these cases.
+            Err(CompactionError::Offload(_)) => {}
+            Err(CompactionError::ShuttingDown) => {}
+        };
+
+        result
     }
 
     /// Mutate the timeline with a [`TimelineWriter`].
@@ -2116,6 +2157,13 @@ impl Timeline {
             .unwrap_or(self.conf.default_tenant_conf.checkpoint_timeout)
     }
 
+    fn get_compaction_period(&self) -> Duration {
+        let tenant_conf = self.tenant_conf.load().tenant_conf.clone();
+        tenant_conf
+            .compaction_period
+            .unwrap_or(self.conf.default_tenant_conf.compaction_period)
+    }
+
     fn get_compaction_target_size(&self) -> u64 {
         let tenant_conf = self.tenant_conf.load();
         tenant_conf
@@ -2132,6 +2180,93 @@ impl Timeline {
             .unwrap_or(self.conf.default_tenant_conf.compaction_threshold)
     }
 
+    fn get_l0_flush_delay_threshold(&self) -> Option<usize> {
+        // Disable L0 flushes by default. This and compaction needs further tuning.
+        const DEFAULT_L0_FLUSH_DELAY_FACTOR: usize = 0; // TODO: default to e.g. 3
+
+        // If compaction is disabled, don't delay.
+        if self.get_compaction_period() == Duration::ZERO {
+            return None;
+        }
+
+        let compaction_threshold = self.get_compaction_threshold();
+        let tenant_conf = self.tenant_conf.load();
+        let l0_flush_delay_threshold = tenant_conf
+            .tenant_conf
+            .l0_flush_delay_threshold
+            .or(self.conf.default_tenant_conf.l0_flush_delay_threshold)
+            .unwrap_or(DEFAULT_L0_FLUSH_DELAY_FACTOR * compaction_threshold);
+
+        // 0 disables backpressure.
+        if l0_flush_delay_threshold == 0 {
+            return None;
+        }
+
+        // Clamp the flush delay threshold to the compaction threshold; it doesn't make sense to
+        // backpressure flushes below this.
+        // TODO: the tenant config should have validation to prevent this instead.
+        debug_assert!(l0_flush_delay_threshold >= compaction_threshold);
+        Some(max(l0_flush_delay_threshold, compaction_threshold))
+    }
+
+    fn get_l0_flush_stall_threshold(&self) -> Option<usize> {
+        // Disable L0 stalls by default. In ingest benchmarks, we see image compaction take >10
+        // minutes, blocking L0 compaction, and we can't stall L0 flushes for that long.
+        const DEFAULT_L0_FLUSH_STALL_FACTOR: usize = 0; // TODO: default to e.g. 5
+
+        // If compaction is disabled, don't stall.
+        if self.get_compaction_period() == Duration::ZERO {
+            return None;
+        }
+
+        // If compaction is failing, don't stall and try to keep the tenant alive. This may not be a
+        // good idea: read amp can grow unbounded, leading to terrible performance, and we may take
+        // on unbounded compaction debt that can take a long time to fix once compaction comes back
+        // online. At least we'll delay flushes, slowing down the growth and buying some time.
+        if self.compaction_failed.load(AtomicOrdering::Relaxed) {
+            return None;
+        }
+
+        let compaction_threshold = self.get_compaction_threshold();
+        let tenant_conf = self.tenant_conf.load();
+        let l0_flush_stall_threshold = tenant_conf
+            .tenant_conf
+            .l0_flush_stall_threshold
+            .or(self.conf.default_tenant_conf.l0_flush_stall_threshold);
+
+        // Tests sometimes set compaction_threshold=1 to generate lots of layer files, and don't
+        // handle the 20-second compaction delay. Some (e.g. `test_backward_compatibility`) can't
+        // easily adjust the L0 backpressure settings, so just disable stalls in this case.
+        if cfg!(feature = "testing")
+            && compaction_threshold == 1
+            && l0_flush_stall_threshold.is_none()
+        {
+            return None;
+        }
+
+        let l0_flush_stall_threshold = l0_flush_stall_threshold
+            .unwrap_or(DEFAULT_L0_FLUSH_STALL_FACTOR * compaction_threshold);
+
+        // 0 disables backpressure.
+        if l0_flush_stall_threshold == 0 {
+            return None;
+        }
+
+        // Clamp the flush stall threshold to the compaction threshold; it doesn't make sense to
+        // backpressure flushes below this.
+        // TODO: the tenant config should have validation to prevent this instead.
+        debug_assert!(l0_flush_stall_threshold >= compaction_threshold);
+        Some(max(l0_flush_stall_threshold, compaction_threshold))
+    }
+
+    fn get_l0_flush_wait_upload(&self) -> bool {
+        let tenant_conf = self.tenant_conf.load();
+        tenant_conf
+            .tenant_conf
+            .l0_flush_wait_upload
+            .unwrap_or(self.conf.default_tenant_conf.l0_flush_wait_upload)
+    }
+
     fn get_image_creation_threshold(&self) -> usize {
         let tenant_conf = self.tenant_conf.load();
         tenant_conf
@@ -2340,7 +2475,8 @@ impl Timeline {
                     // initial logical size is 0.
                     LogicalSize::empty_initial()
                 },
-                partitioning: tokio::sync::Mutex::new((
+
+                partitioning: GuardArcSwap::new((
                     (KeyPartitioning::new(), KeyPartitioning::new().into_sparse()),
                     Lsn(0),
                 )),
@@ -2367,6 +2503,7 @@ impl Timeline {
                 gate: Gate::default(),
 
                 compaction_lock: tokio::sync::Mutex::default(),
+                compaction_failed: AtomicBool::default(),
                 gc_lock: tokio::sync::Mutex::default(),
 
                 standby_horizon: AtomicLsn::new(0),
@@ -2800,12 +2937,10 @@ impl Timeline {
             "initial size calculation",
             // NB: don't log errors here, task_mgr will do that.
             async move {
-                let cancel = task_mgr::shutdown_token();
                 self_clone
                     .initial_logical_size_calculation_task(
                         initial_part_end,
                         cancel_wait_for_background_loop_concurrency_limit_semaphore,
-                        cancel,
                         background_ctx,
                     )
                     .await;
@@ -2815,11 +2950,21 @@ impl Timeline {
         );
     }
 
+    /// # Cancellation
+    ///
+    /// This method is sensitive to `Timeline::cancel`.
+    ///
+    /// It is _not_ sensitive to task_mgr::shutdown_token().
+    ///
+    /// # Cancel-Safety
+    ///
+    /// It does Timeline IO, hence this should be polled to completion because
+    /// we could be leaving in-flight IOs behind, which is safe, but annoying
+    /// to reason about.
     async fn initial_logical_size_calculation_task(
         self: Arc<Self>,
         initial_part_end: Lsn,
         skip_concurrency_limiter: CancellationToken,
-        cancel: CancellationToken,
         background_ctx: RequestContext,
     ) {
         scopeguard::defer! {
@@ -2832,7 +2977,6 @@ impl Timeline {
             let self_ref = &self;
             let skip_concurrency_limiter = &skip_concurrency_limiter;
             async move {
-                let cancel = task_mgr::shutdown_token();
                 let wait_for_permit = super::tasks::concurrent_background_tasks_rate_limit_permit(
                     BackgroundLoopKind::InitialLogicalSizeCalculation,
                     background_ctx,
@@ -2846,9 +2990,6 @@ impl Timeline {
                     _ = self_ref.cancel.cancelled() => {
                         return Err(CalculateLogicalSizeError::Cancelled);
                     }
-                    _ = cancel.cancelled() => {
-                        return Err(CalculateLogicalSizeError::Cancelled);
-                    },
                     () = skip_concurrency_limiter.cancelled() => {
                         // Some action that is part of a end user interaction requested logical size
                         // => break out of the rate limit
@@ -2865,6 +3006,14 @@ impl Timeline {
                     crate::metrics::initial_logical_size::START_CALCULATION.retry(circumstances)
                 };
 
+                let io_concurrency = IoConcurrency::spawn_from_conf(
+                    self_ref.conf,
+                    self_ref
+                        .gate
+                        .enter()
+                        .map_err(|_| CalculateLogicalSizeError::Cancelled)?,
+                );
+
                 let calculated_size = self_ref
                     .logical_size_calculation_task(
                         initial_part_end,
@@ -2874,7 +3023,11 @@ impl Timeline {
                     .await?;
 
                 self_ref
-                    .trigger_aux_file_size_computation(initial_part_end, background_ctx)
+                    .trigger_aux_file_size_computation(
+                        initial_part_end,
+                        background_ctx,
+                        io_concurrency,
+                    )
                     .await?;
 
                 // TODO: add aux file size to logical size
@@ -2907,22 +3060,18 @@ impl Timeline {
                             )
                             .expect("10min < 1hour"),
                         );
-                        tokio::time::sleep(sleep_duration).await;
+                        tokio::select! {
+                            _ = tokio::time::sleep(sleep_duration) => {}
+                            _ = self.cancel.cancelled() => return ControlFlow::Break(()),
+                        }
                     }
                 }
             }
         };
 
-        let (calculated_size, metrics_guard) = tokio::select! {
-            res = retrying  => {
-                match res {
-                    ControlFlow::Continue(calculated_size) => calculated_size,
-                    ControlFlow::Break(()) => return,
-                }
-            }
-            _ = cancel.cancelled() => {
-                return;
-            }
+        let (calculated_size, metrics_guard) = match retrying.await {
+            ControlFlow::Continue(calculated_size) => calculated_size,
+            ControlFlow::Break(()) => return,
         };
 
         // we cannot query current_logical_size.current_size() to know the current
@@ -2978,9 +3127,6 @@ impl Timeline {
         receiver
     }
 
-    /// # Cancel-Safety
-    ///
-    /// This method is cancellation-safe.
     #[instrument(skip_all)]
     async fn logical_size_calculation_task(
         self: &Arc<Self>,
@@ -2998,32 +3144,13 @@ impl Timeline {
             .enter()
             .map_err(|_| CalculateLogicalSizeError::Cancelled)?;
 
-        let self_calculation = Arc::clone(self);
-
-        let mut calculation = pin!(async {
-            let ctx = ctx.attached_child();
-            self_calculation
-                .calculate_logical_size(lsn, cause, &guard, &ctx)
-                .await
-        });
-
-        tokio::select! {
-            res = &mut calculation => { res }
-            _ = self.cancel.cancelled() => {
-                debug!("cancelling logical size calculation for timeline shutdown");
-                calculation.await
-            }
-        }
+        self.calculate_logical_size(lsn, cause, &guard, ctx).await
     }
 
     /// Calculate the logical size of the database at the latest LSN.
     ///
     /// NOTE: counted incrementally, includes ancestors. This can be a slow operation,
     /// especially if we need to download remote layers.
-    ///
-    /// # Cancel-Safety
-    ///
-    /// This method is cancellation-safe.
     async fn calculate_logical_size(
         &self,
         up_to_lsn: Lsn,
@@ -3036,7 +3163,10 @@ impl Timeline {
             self.timeline_id, up_to_lsn
         );
 
-        pausable_failpoint!("timeline-calculate-logical-size-pause");
+        if let Err(()) = pausable_failpoint!("timeline-calculate-logical-size-pause", &self.cancel)
+        {
+            return Err(CalculateLogicalSizeError::Cancelled);
+        }
 
         // See if we've already done the work for initial size calculation.
         // This is a short-cut for timelines that are mostly unused.
@@ -3469,7 +3599,12 @@ impl Timeline {
             }
         }
         ancestor
-            .wait_lsn(self.ancestor_lsn, WaitLsnWaiter::Timeline(self), ctx)
+            .wait_lsn(
+                self.ancestor_lsn,
+                WaitLsnWaiter::Timeline(self),
+                WaitLsnTimeout::Default,
+                ctx,
+            )
             .await
             .map_err(|e| match e {
                 e @ WaitLsnError::Timeout(_) => GetReadyAncestorError::AncestorLsnTimeout(e),
@@ -3589,6 +3724,12 @@ impl Timeline {
         mut layer_flush_start_rx: tokio::sync::watch::Receiver<(u64, Lsn)>,
         ctx: &RequestContext,
     ) {
+        // Subscribe to L0 delta layer updates, for compaction backpressure.
+        let mut watch_l0 = match self.layers.read().await.layer_map() {
+            Ok(lm) => lm.watch_level0_deltas(),
+            Err(Shutdown) => return,
+        };
+
         info!("started flush loop");
         loop {
             tokio::select! {
@@ -3613,43 +3754,68 @@ impl Timeline {
                     return;
                 }
 
-                let timer = self.metrics.flush_time_histo.start_timer();
+                // Break to notify potential waiters as soon as we've flushed the requested LSN. If
+                // more requests have arrived in the meanwhile, we'll resume flushing afterwards.
+                if flushed_to_lsn >= frozen_to_lsn {
+                    break Ok(());
+                }
 
-                let num_frozen_layers;
-                let frozen_layer_total_size;
-                let layer_to_flush = {
-                    let guard = self.layers.read().await;
-                    let Ok(lm) = guard.layer_map() else {
+                // Fetch the next layer to flush, if any.
+                let (layer, l0_count, frozen_count, frozen_size) = {
+                    let layers = self.layers.read().await;
+                    let Ok(lm) = layers.layer_map() else {
                         info!("dropping out of flush loop for timeline shutdown");
                         return;
                     };
-                    num_frozen_layers = lm.frozen_layers.len();
-                    frozen_layer_total_size = lm
+                    let l0_count = lm.level0_deltas().len();
+                    let frozen_count = lm.frozen_layers.len();
+                    let frozen_size: u64 = lm
                         .frozen_layers
                         .iter()
                         .map(|l| l.estimated_in_mem_size())
-                        .sum::<u64>();
-                    lm.frozen_layers.front().cloned()
-                    // drop 'layers' lock to allow concurrent reads and writes
+                        .sum();
+                    let layer = lm.frozen_layers.front().cloned();
+                    (layer, l0_count, frozen_count, frozen_size)
+                    // drop 'layers' lock
                 };
-                let Some(layer_to_flush) = layer_to_flush else {
+                let Some(layer) = layer else {
                     break Ok(());
                 };
-                if num_frozen_layers
-                    > std::cmp::max(
-                        self.get_compaction_threshold(),
-                        DEFAULT_COMPACTION_THRESHOLD,
-                    )
-                    && frozen_layer_total_size >= /* 128 MB */ 128000000
-                {
-                    tracing::warn!(
-                        "too many frozen layers: {num_frozen_layers} layers with estimated in-mem size of {frozen_layer_total_size} bytes",
-                    );
-                }
-                match self.flush_frozen_layer(layer_to_flush, ctx).await {
-                    Ok(this_layer_to_lsn) => {
-                        flushed_to_lsn = std::cmp::max(flushed_to_lsn, this_layer_to_lsn);
+
+                // Stall flushes to backpressure if compaction can't keep up. This is propagated up
+                // to WAL ingestion by having ephemeral layer rolls wait for flushes.
+                //
+                // NB: the compaction loop only checks `compaction_threshold` every 20 seconds, so
+                // we can end up stalling before compaction even starts. Consider making it more
+                // responsive (e.g. via `watch_level0_deltas`).
+                if let Some(stall_threshold) = self.get_l0_flush_stall_threshold() {
+                    if l0_count >= stall_threshold {
+                        warn!(
+                            "stalling layer flushes for compaction backpressure at {l0_count} \
+                            L0 layers ({frozen_count} frozen layers with {frozen_size} bytes)"
+                        );
+                        let stall_timer = self
+                            .metrics
+                            .flush_delay_histo
+                            .start_timer()
+                            .record_on_drop();
+                        tokio::select! {
+                            result = watch_l0.wait_for(|l0| *l0 < stall_threshold) => {
+                                if let Ok(l0) = result.as_deref() {
+                                    let delay = stall_timer.elapsed().as_secs_f64();
+                                    info!("resuming layer flushes at {l0} L0 layers after {delay:.3}s");
+                                }
+                            },
+                            _ = self.cancel.cancelled() => {},
+                        }
+                        continue; // check again
                     }
+                }
+
+                // Flush the layer.
+                let flush_timer = self.metrics.flush_time_histo.start_timer();
+                match self.flush_frozen_layer(layer, ctx).await {
+                    Ok(layer_lsn) => flushed_to_lsn = max(flushed_to_lsn, layer_lsn),
                     Err(FlushLayerError::Cancelled) => {
                         info!("dropping out of flush loop for timeline shutdown");
                         return;
@@ -3663,7 +3829,30 @@ impl Timeline {
                         break err.map(|_| ());
                     }
                 }
-                timer.stop_and_record();
+                let flush_duration = flush_timer.stop_and_record();
+
+                // Delay the next flush to backpressure if compaction can't keep up. We delay by the
+                // flush duration such that the flush takes 2x as long. This is propagated up to WAL
+                // ingestion by having ephemeral layer rolls wait for flushes.
+                if let Some(delay_threshold) = self.get_l0_flush_delay_threshold() {
+                    if l0_count >= delay_threshold {
+                        let delay = flush_duration.as_secs_f64();
+                        info!(
+                            "delaying layer flush by {delay:.3}s for compaction backpressure at \
+                            {l0_count} L0 layers ({frozen_count} frozen layers with {frozen_size} bytes)"
+                        );
+                        let _delay_timer = self
+                            .metrics
+                            .flush_delay_histo
+                            .start_timer()
+                            .record_on_drop();
+                        tokio::select! {
+                            _ = tokio::time::sleep(flush_duration) => {},
+                            _ = watch_l0.wait_for(|l0| *l0 < delay_threshold) => {},
+                            _ = self.cancel.cancelled() => {},
+                        }
+                    }
+                }
             };
 
             // Unsharded tenants should never advance their LSN beyond the end of the
@@ -3867,21 +4056,24 @@ impl Timeline {
 
         // Backpressure mechanism: wait with continuation of the flush loop until we have uploaded all layer files.
         // This makes us refuse ingest until the new layers have been persisted to the remote
-        let start = Instant::now();
-        self.remote_client
-            .wait_completion()
-            .await
-            .map_err(|e| match e {
-                WaitCompletionError::UploadQueueShutDownOrStopped
-                | WaitCompletionError::NotInitialized(
-                    NotInitialized::ShuttingDown | NotInitialized::Stopped,
-                ) => FlushLayerError::Cancelled,
-                WaitCompletionError::NotInitialized(NotInitialized::Uninitialized) => {
-                    FlushLayerError::Other(anyhow!(e).into())
-                }
-            })?;
-        let duration = start.elapsed().as_secs_f64();
-        self.metrics.flush_wait_upload_time_gauge_add(duration);
+        // TODO: remove this, and rely on l0_flush_{delay,stall}_threshold instead.
+        if self.get_l0_flush_wait_upload() {
+            let start = Instant::now();
+            self.remote_client
+                .wait_completion()
+                .await
+                .map_err(|e| match e {
+                    WaitCompletionError::UploadQueueShutDownOrStopped
+                    | WaitCompletionError::NotInitialized(
+                        NotInitialized::ShuttingDown | NotInitialized::Stopped,
+                    ) => FlushLayerError::Cancelled,
+                    WaitCompletionError::NotInitialized(NotInitialized::Uninitialized) => {
+                        FlushLayerError::Other(anyhow!(e).into())
+                    }
+                })?;
+            let duration = start.elapsed().as_secs_f64();
+            self.metrics.flush_wait_upload_time_gauge_add(duration);
+        }
 
         // FIXME: between create_delta_layer and the scheduling of the upload in `update_metadata_file`,
         // a compaction can delete the file and then it won't be available for uploads any more.
@@ -4028,18 +4220,15 @@ impl Timeline {
         flags: EnumSet<CompactFlags>,
         ctx: &RequestContext,
     ) -> Result<((KeyPartitioning, SparseKeyPartitioning), Lsn), CompactionError> {
-        let Ok(mut partitioning_guard) = self.partitioning.try_lock() else {
+        let Ok(mut guard) = self.partitioning.try_write_guard() else {
             // NB: there are two callers, one is the compaction task, of which there is only one per struct Tenant and hence Timeline.
             // The other is the initdb optimization in flush_frozen_layer, used by `boostrap_timeline`, which runs before `.activate()`
             // and hence before the compaction task starts.
-            // Note that there are a third "caller" that will take the `partitioning` lock. It is `gc_compaction_split_jobs` for
-            // gc-compaction where it uses the repartition data to determine the split jobs. In the future, it might use its own
-            // heuristics, but for now, we should allow concurrent access to it and let the caller retry compaction.
             return Err(CompactionError::Other(anyhow!(
-                "repartition() called concurrently, this is rare and a retry should be fine"
+                "repartition() called concurrently"
             )));
         };
-        let ((dense_partition, sparse_partition), partition_lsn) = &*partitioning_guard;
+        let ((dense_partition, sparse_partition), partition_lsn) = &*guard.read();
         if lsn < *partition_lsn {
             return Err(CompactionError::Other(anyhow!(
                 "repartition() called with LSN going backwards, this should not happen"
@@ -4067,9 +4256,9 @@ impl Timeline {
         let sparse_partitioning = SparseKeyPartitioning {
             parts: vec![sparse_ks],
         }; // no partitioning for metadata keys for now
-        *partitioning_guard = ((dense_partitioning, sparse_partitioning), lsn);
-
-        Ok((partitioning_guard.0.clone(), partitioning_guard.1))
+        let result = ((dense_partitioning, sparse_partitioning), lsn);
+        guard.write(result.clone());
+        Ok(result)
     }
 
     // Is it time to create a new image layer for the given partition?
@@ -4127,6 +4316,7 @@ impl Timeline {
 
     /// Create image layers for Postgres data. Assumes the caller passes a partition that is not too large,
     /// so that at most one image layer will be produced from this function.
+    #[allow(clippy::too_many_arguments)]
     async fn create_image_layer_for_rel_blocks(
         self: &Arc<Self>,
         partition: &KeySpace,
@@ -4135,6 +4325,7 @@ impl Timeline {
         ctx: &RequestContext,
         img_range: Range<Key>,
         start: Key,
+        io_concurrency: IoConcurrency,
     ) -> Result<ImageLayerCreationOutcome, CreateImageLayersError> {
         let mut wrote_keys = false;
 
@@ -4163,7 +4354,12 @@ impl Timeline {
                     || (last_key_in_range && key_request_accum.raw_size() > 0)
                 {
                     let results = self
-                        .get_vectored(key_request_accum.consume_keyspace(), lsn, ctx)
+                        .get_vectored(
+                            key_request_accum.consume_keyspace(),
+                            lsn,
+                            io_concurrency.clone(),
+                            ctx,
+                        )
                         .await?;
 
                     if self.cancel.is_cancelled() {
@@ -4242,9 +4438,10 @@ impl Timeline {
         img_range: Range<Key>,
         mode: ImageLayerCreationMode,
         start: Key,
+        io_concurrency: IoConcurrency,
     ) -> Result<ImageLayerCreationOutcome, CreateImageLayersError> {
         // Metadata keys image layer creation.
-        let mut reconstruct_state = ValuesReconstructState::default();
+        let mut reconstruct_state = ValuesReconstructState::new(io_concurrency);
         let begin = Instant::now();
         let data = self
             .get_vectored_impl(partition.clone(), lsn, &mut reconstruct_state, ctx)
@@ -4461,6 +4658,13 @@ impl Timeline {
                 )))
             });
 
+            let io_concurrency = IoConcurrency::spawn_from_conf(
+                self.conf,
+                self.gate
+                    .enter()
+                    .map_err(|_| CreateImageLayersError::Cancelled)?,
+            );
+
             if !compact_metadata {
                 let ImageLayerCreationOutcome {
                     image,
@@ -4473,6 +4677,7 @@ impl Timeline {
                         ctx,
                         img_range,
                         start,
+                        io_concurrency,
                     )
                     .await?;
 
@@ -4491,6 +4696,7 @@ impl Timeline {
                         img_range,
                         mode,
                         start,
+                        io_concurrency,
                     )
                     .await?;
                 start = next_start_key;
@@ -4625,6 +4831,10 @@ impl Drop for Timeline {
                 }
             }
         }
+        info!(
+            "Timeline {} for tenant {} is being dropped",
+            self.timeline_id, self.tenant_shard_id.tenant_id
+        );
     }
 }
 
@@ -5673,9 +5883,17 @@ impl Timeline {
         info!("force created image layer {}", image_layer.local_path());
         {
             let mut guard = self.layers.write().await;
-            guard.open_mut().unwrap().force_insert_layer(image_layer);
+            guard
+                .open_mut()
+                .unwrap()
+                .force_insert_layer(image_layer.clone());
         }
 
+        // Update remote_timeline_client state to reflect existence of this layer
+        self.remote_client
+            .schedule_layer_file_upload(image_layer)
+            .unwrap();
+
         Ok(())
     }
 
@@ -5726,9 +5944,17 @@ impl Timeline {
         info!("force created delta layer {}", delta_layer.local_path());
         {
             let mut guard = self.layers.write().await;
-            guard.open_mut().unwrap().force_insert_layer(delta_layer);
+            guard
+                .open_mut()
+                .unwrap()
+                .force_insert_layer(delta_layer.clone());
         }
 
+        // Update remote_timeline_client state to reflect existence of this layer
+        self.remote_client
+            .schedule_layer_file_upload(delta_layer)
+            .unwrap();
+
         Ok(())
     }
 
@@ -5738,13 +5964,14 @@ impl Timeline {
         self: &Arc<Timeline>,
         lsn: Lsn,
         ctx: &RequestContext,
+        io_concurrency: IoConcurrency,
     ) -> anyhow::Result<Vec<(Key, Bytes)>> {
         let mut all_data = Vec::new();
         let guard = self.layers.read().await;
         for layer in guard.layer_map()?.iter_historic_layers() {
             if !layer.is_delta() && layer.image_layer_lsn() == lsn {
                 let layer = guard.get_from_desc(&layer);
-                let mut reconstruct_data = ValuesReconstructState::default();
+                let mut reconstruct_data = ValuesReconstructState::new(io_concurrency.clone());
                 layer
                     .get_values_reconstruct_data(
                         KeySpace::single(Key::MIN..Key::MAX),
@@ -5753,8 +5980,9 @@ impl Timeline {
                         ctx,
                     )
                     .await?;
-                for (k, v) in reconstruct_data.keys {
-                    all_data.push((k, v?.img.unwrap().1));
+                for (k, v) in std::mem::take(&mut reconstruct_data.keys) {
+                    let v = v.collect_pending_ios().await?;
+                    all_data.push((k, v.img.unwrap().1));
                 }
             }
         }
@@ -5875,13 +6103,37 @@ impl TimelineWriter<'_> {
     async fn roll_layer(&mut self, freeze_at: Lsn) -> Result<(), FlushLayerError> {
         let current_size = self.write_guard.as_ref().unwrap().current_size;
 
+        // If layer flushes are backpressured due to compaction not keeping up, wait for the flush
+        // to propagate the backpressure up into WAL ingestion.
+        let l0_count = self
+            .tl
+            .layers
+            .read()
+            .await
+            .layer_map()?
+            .level0_deltas()
+            .len();
+        let wait_thresholds = [
+            self.get_l0_flush_delay_threshold(),
+            self.get_l0_flush_stall_threshold(),
+        ];
+        let wait_threshold = wait_thresholds.into_iter().flatten().min();
+
         // self.write_guard will be taken by the freezing
-        self.tl
+        let flush_id = self
+            .tl
             .freeze_inmem_layer_at(freeze_at, &mut self.write_guard)
             .await?;
 
         assert!(self.write_guard.is_none());
 
+        if let Some(wait_threshold) = wait_threshold {
+            if l0_count >= wait_threshold {
+                info!("layer roll waiting for flush due to compaction backpressure at {l0_count} L0 layers");
+                self.tl.wait_flush_completion(flush_id).await?;
+            }
+        }
+
         if current_size >= self.get_checkpoint_distance() * 2 {
             warn!("Flushed oversized open layer with size {}", current_size)
         }

pageserver/src/tenant/timeline/compaction.rs

@@ -42,8 +42,8 @@ use crate::tenant::storage_layer::merge_iterator::MergeIterator;
 use crate::tenant::storage_layer::{
     AsLayerDesc, PersistentLayerDesc, PersistentLayerKey, ValueReconstructState,
 };
-use crate::tenant::timeline::ImageLayerCreationOutcome;
 use crate::tenant::timeline::{drop_rlock, DeltaLayerWriter, ImageLayerWriter};
+use crate::tenant::timeline::{ImageLayerCreationOutcome, IoConcurrency};
 use crate::tenant::timeline::{Layer, ResidentLayer};
 use crate::tenant::{gc_block, DeltaLayer, MaybeOffloaded};
 use crate::virtual_file::{MaybeFatalIo, VirtualFile};
@@ -436,12 +436,14 @@ impl KeyHistoryRetention {
         if dry_run {
             return true;
         }
-        let guard = tline.layers.read().await;
-        if !guard.contains_key(key) {
-            return false;
+        let layer_generation;
+        {
+            let guard = tline.layers.read().await;
+            if !guard.contains_key(key) {
+                return false;
+            }
+            layer_generation = guard.get_from_key(key).metadata().generation;
         }
-        let layer_generation = guard.get_from_key(key).metadata().generation;
-        drop(guard);
         if layer_generation == tline.generation {
             info!(
                 key=%key,
@@ -2138,6 +2140,11 @@ impl Timeline {
             self.get_gc_compaction_watermark()
         };
 
+        if compact_below_lsn == Lsn::INVALID {
+            tracing::warn!("no layers to compact with gc: gc_cutoff not generated yet, skipping gc bottom-most compaction");
+            return Ok(vec![]);
+        }
+
         // Split compaction job to about 4GB each
         const GC_COMPACT_MAX_SIZE_MB: u64 = 4 * 1024;
         let sub_compaction_max_job_size_mb =
@@ -2146,12 +2153,7 @@ impl Timeline {
         let mut compact_jobs = Vec::new();
         // For now, we simply use the key partitioning information; we should do a more fine-grained partitioning
         // by estimating the amount of files read for a compaction job. We should also partition on LSN.
-        let ((dense_ks, sparse_ks), _) = {
-            let Ok(partition) = self.partitioning.try_lock() else {
-                bail!("failed to acquire partition lock during gc-compaction");
-            };
-            partition.clone()
-        };
+        let ((dense_ks, sparse_ks), _) = self.partitioning.read().as_ref().clone();
         // Truncate the key range to be within user specified compaction range.
         fn truncate_to(
             source_start: &Key,
@@ -2217,6 +2219,12 @@ impl Timeline {
                 } else {
                     end
                 };
+                let end = if ranges_num == idx + 1 {
+                    // extend the compaction range to the end of the key range if it's the last partition
+                    end.max(job.compact_key_range.end)
+                } else {
+                    end
+                };
                 info!(
                     "splitting compaction job: {}..{}, estimated_size={}",
                     start, end, total_size
@@ -2337,6 +2345,11 @@ impl Timeline {
                 // each of the retain_lsn. Therefore, if the user-provided `compact_lsn_range.end` is larger than the real gc cutoff, we will use
                 // the real cutoff.
                 let mut gc_cutoff = if compact_lsn_range.end == Lsn::MAX {
+                    if real_gc_cutoff == Lsn::INVALID {
+                        // If the gc_cutoff is not generated yet, we should not compact anything.
+                        tracing::warn!("no layers to compact with gc: gc_cutoff not generated yet, skipping gc bottom-most compaction");
+                        return Ok(());
+                    }
                     real_gc_cutoff
                 } else {
                     compact_lsn_range.end
@@ -2868,7 +2881,7 @@ impl Timeline {
             "produced {} delta layers and {} image layers, {} layers are kept",
             produced_delta_layers_len,
             produced_image_layers_len,
-            layer_selection.len()
+            keep_layers.len()
         );
 
         // Step 3: Place back to the layer map.
@@ -2914,8 +2927,28 @@ impl Timeline {
         // be batched into `schedule_compaction_update`.
         let disk_consistent_lsn = self.disk_consistent_lsn.load();
         self.schedule_uploads(disk_consistent_lsn, None)?;
+        // If a layer gets rewritten throughout gc-compaction, we need to keep that layer only in `compact_to` instead
+        // of `compact_from`.
+        let compact_from = {
+            let mut compact_from = Vec::new();
+            let mut compact_to_set = HashMap::new();
+            for layer in &compact_to {
+                compact_to_set.insert(layer.layer_desc().key(), layer);
+            }
+            for layer in &layer_selection {
+                if let Some(to) = compact_to_set.get(&layer.layer_desc().key()) {
+                    tracing::info!(
+                        "skipping delete {} because found same layer key at different generation {}",
+                        layer, to
+                    );
+                } else {
+                    compact_from.push(layer.clone());
+                }
+            }
+            compact_from
+        };
         self.remote_client
-            .schedule_compaction_update(&layer_selection, &compact_to)?;
+            .schedule_compaction_update(&compact_from, &compact_to)?;
 
         drop(gc_lock);
 
@@ -3175,6 +3208,7 @@ impl TimelineAdaptor {
                 ctx,
                 key_range.clone(),
                 start,
+                IoConcurrency::sequential(),
             )
             .await?;
 

pageserver/src/tenant/timeline/delete.rs

@@ -112,7 +112,7 @@ pub(super) async fn delete_local_timeline_directory(
 }
 
 /// It is important that this gets called when DeletionGuard is being held.
-/// For more context see comments in [`DeleteTimelineFlow::prepare`]
+/// For more context see comments in [`make_timeline_delete_guard`]
 async fn remove_maybe_offloaded_timeline_from_tenant(
     tenant: &Tenant,
     timeline: &TimelineOrOffloaded,
@@ -193,10 +193,8 @@ impl DeleteTimelineFlow {
     ) -> Result<(), DeleteTimelineError> {
         super::debug_assert_current_span_has_tenant_and_timeline_id();
 
-        let allow_offloaded_children = false;
-        let set_stopping = true;
         let (timeline, mut guard) =
-            Self::prepare(tenant, timeline_id, allow_offloaded_children, set_stopping)?;
+            make_timeline_delete_guard(tenant, timeline_id, TimelineDeleteGuardKind::Delete)?;
 
         guard.mark_in_progress()?;
 
@@ -333,75 +331,6 @@ impl DeleteTimelineFlow {
         Ok(())
     }
 
-    pub(super) fn prepare(
-        tenant: &Tenant,
-        timeline_id: TimelineId,
-        allow_offloaded_children: bool,
-        set_stopping: bool,
-    ) -> Result<(TimelineOrOffloaded, DeletionGuard), DeleteTimelineError> {
-        // Note the interaction between this guard and deletion guard.
-        // Here we attempt to lock deletion guard when we're holding a lock on timelines.
-        // This is important because when you take into account `remove_timeline_from_tenant`
-        // we remove timeline from memory when we still hold the deletion guard.
-        // So here when timeline deletion is finished timeline wont be present in timelines map at all
-        // which makes the following sequence impossible:
-        // T1: get preempted right before the try_lock on `Timeline::delete_progress`
-        // T2: do a full deletion, acquire and drop `Timeline::delete_progress`
-        // T1: acquire deletion lock, do another `DeleteTimelineFlow::run`
-        // For more context see this discussion: `https://github.com/neondatabase/neon/pull/4552#discussion_r1253437346`
-        let timelines = tenant.timelines.lock().unwrap();
-        let timelines_offloaded = tenant.timelines_offloaded.lock().unwrap();
-
-        let timeline = match timelines.get(&timeline_id) {
-            Some(t) => TimelineOrOffloaded::Timeline(Arc::clone(t)),
-            None => match timelines_offloaded.get(&timeline_id) {
-                Some(t) => TimelineOrOffloaded::Offloaded(Arc::clone(t)),
-                None => return Err(DeleteTimelineError::NotFound),
-            },
-        };
-
-        // Ensure that there are no child timelines, because we are about to remove files,
-        // which will break child branches
-        let mut children = Vec::new();
-        if !allow_offloaded_children {
-            children.extend(timelines_offloaded.iter().filter_map(|(id, entry)| {
-                (entry.ancestor_timeline_id == Some(timeline_id)).then_some(*id)
-            }));
-        }
-        children.extend(timelines.iter().filter_map(|(id, entry)| {
-            (entry.get_ancestor_timeline_id() == Some(timeline_id)).then_some(*id)
-        }));
-
-        if !children.is_empty() {
-            return Err(DeleteTimelineError::HasChildren(children));
-        }
-
-        // Note that using try_lock here is important to avoid a deadlock.
-        // Here we take lock on timelines and then the deletion guard.
-        // At the end of the operation we're holding the guard and need to lock timelines map
-        // to remove the timeline from it.
-        // Always if you have two locks that are taken in different order this can result in a deadlock.
-
-        let delete_progress = Arc::clone(timeline.delete_progress());
-        let delete_lock_guard = match delete_progress.try_lock_owned() {
-            Ok(guard) => DeletionGuard(guard),
-            Err(_) => {
-                // Unfortunately if lock fails arc is consumed.
-                return Err(DeleteTimelineError::AlreadyInProgress(Arc::clone(
-                    timeline.delete_progress(),
-                )));
-            }
-        };
-
-        if set_stopping {
-            if let TimelineOrOffloaded::Timeline(timeline) = &timeline {
-                timeline.set_state(TimelineState::Stopping);
-            }
-        }
-
-        Ok((timeline, delete_lock_guard))
-    }
-
     fn schedule_background(
         guard: DeletionGuard,
         conf: &'static PageServerConf,
@@ -483,6 +412,80 @@ impl DeleteTimelineFlow {
     }
 }
 
+#[derive(Copy, Clone, PartialEq, Eq)]
+pub(super) enum TimelineDeleteGuardKind {
+    Offload,
+    Delete,
+}
+
+pub(super) fn make_timeline_delete_guard(
+    tenant: &Tenant,
+    timeline_id: TimelineId,
+    guard_kind: TimelineDeleteGuardKind,
+) -> Result<(TimelineOrOffloaded, DeletionGuard), DeleteTimelineError> {
+    // Note the interaction between this guard and deletion guard.
+    // Here we attempt to lock deletion guard when we're holding a lock on timelines.
+    // This is important because when you take into account `remove_timeline_from_tenant`
+    // we remove timeline from memory when we still hold the deletion guard.
+    // So here when timeline deletion is finished timeline wont be present in timelines map at all
+    // which makes the following sequence impossible:
+    // T1: get preempted right before the try_lock on `Timeline::delete_progress`
+    // T2: do a full deletion, acquire and drop `Timeline::delete_progress`
+    // T1: acquire deletion lock, do another `DeleteTimelineFlow::run`
+    // For more context see this discussion: `https://github.com/neondatabase/neon/pull/4552#discussion_r1253437346`
+    let timelines = tenant.timelines.lock().unwrap();
+    let timelines_offloaded = tenant.timelines_offloaded.lock().unwrap();
+
+    let timeline = match timelines.get(&timeline_id) {
+        Some(t) => TimelineOrOffloaded::Timeline(Arc::clone(t)),
+        None => match timelines_offloaded.get(&timeline_id) {
+            Some(t) => TimelineOrOffloaded::Offloaded(Arc::clone(t)),
+            None => return Err(DeleteTimelineError::NotFound),
+        },
+    };
+
+    // Ensure that there are no child timelines, because we are about to remove files,
+    // which will break child branches
+    let mut children = Vec::new();
+    if guard_kind == TimelineDeleteGuardKind::Delete {
+        children.extend(timelines_offloaded.iter().filter_map(|(id, entry)| {
+            (entry.ancestor_timeline_id == Some(timeline_id)).then_some(*id)
+        }));
+    }
+    children.extend(timelines.iter().filter_map(|(id, entry)| {
+        (entry.get_ancestor_timeline_id() == Some(timeline_id)).then_some(*id)
+    }));
+
+    if !children.is_empty() {
+        return Err(DeleteTimelineError::HasChildren(children));
+    }
+
+    // Note that using try_lock here is important to avoid a deadlock.
+    // Here we take lock on timelines and then the deletion guard.
+    // At the end of the operation we're holding the guard and need to lock timelines map
+    // to remove the timeline from it.
+    // Always if you have two locks that are taken in different order this can result in a deadlock.
+
+    let delete_progress = Arc::clone(timeline.delete_progress());
+    let delete_lock_guard = match delete_progress.try_lock_owned() {
+        Ok(guard) => DeletionGuard(guard),
+        Err(_) => {
+            // Unfortunately if lock fails arc is consumed.
+            return Err(DeleteTimelineError::AlreadyInProgress(Arc::clone(
+                timeline.delete_progress(),
+            )));
+        }
+    };
+
+    if guard_kind == TimelineDeleteGuardKind::Delete {
+        if let TimelineOrOffloaded::Timeline(timeline) = &timeline {
+            timeline.set_state(TimelineState::Stopping);
+        }
+    }
+
+    Ok((timeline, delete_lock_guard))
+}
+
 pub(super) struct DeletionGuard(OwnedMutexGuard<DeleteTimelineFlow>);
 
 impl Deref for DeletionGuard {

pageserver/src/tenant/timeline/handle.rs

@@ -32,54 +32,151 @@
 //!
 //! # Design
 //!
+//! ## Data Structures
+//!
 //! There are three user-facing data structures:
 //! - `PerTimelineState`: a struct embedded into each Timeline struct. Lifetime == Timeline lifetime.
 //! - `Cache`: a struct private to each connection handler; Lifetime == connection lifetime.
 //! - `Handle`: a smart pointer that holds the Timeline gate open and derefs to `&Timeline`.
-//!   Lifetime: for a single request dispatch on the Timeline (i.e., one getpage request)
+//! - `WeakHandle`: downgrade of a `Handle` that does not keep the gate open, but allows
+//!   trying to ugprade back to a `Handle`, guaranteeing it's the same `Timeline` *object*.
 //!
-//! The `Handle` is just a wrapper around an `Arc<HandleInner>`.
+//! Internally, there is 0 or 1 `HandleInner` per `(Cache,Timeline)`.
+//! Since Cache:Connection is 1:1, there is 0 or 1 `HandleInner` per `(Connection,Timeline)`.
 //!
-//! There is one long-lived `Arc<HandleInner>`, which is stored in the `PerTimelineState`.
-//! The `Cache` stores a `Weak<HandleInner>` for each cached Timeline.
+//! The `HandleInner`  is allocated as a `Arc<Mutex<HandleInner>>` and
+//! referenced weakly and strongly from various places which we are now illustrating.
+//! For brevity, we will omit the `Arc<Mutex<>>` part in the following and instead
+//! use `strong ref` and `weak ref` when referring to the `Arc<Mutex<HandleInner>>`
+//! or `Weak<Mutex<HandleInner>>`, respectively.
+//!
+//! - The `Handle` is a strong ref.
+//! - The `WeakHandle` is a weak ref.
+//! - The `PerTimelineState` contains a `HashMap<CacheId, strong ref>`.
+//! - The `Cache` is a `HashMap<unique identifier for the shard, weak ref>`.
+//!
+//! Lifetimes:
+//! - `WeakHandle` and `Handle`: single pagestream request.
+//! - `Cache`: single page service connection.
+//! - `PerTimelineState`:  lifetime of the Timeline object (i.e., i.e., till `Timeline::shutdown`).
+//!
+//! ## Request Handling Flow (= filling and using the `Cache``)
 //!
 //! To dispatch a request, the page service connection calls `Cache::get`.
 //!
 //! A cache miss means we consult the tenant manager for shard routing,
-//! resulting in an `Arc<Timeline>`. We enter its gate _once_ and construct an
-//! `Arc<HandleInner>`. We store a `Weak<HandleInner>` in the cache
-//! and the `Arc<HandleInner>` in the `PerTimelineState`.
+//! resulting in an `Arc<Timeline>`. We enter its gate _once_ and store it in the the
+//! `Arc<Mutex<HandleInner>>>`. A weak ref is stored in the `Cache`
+//! and a strong ref in the `PerTimelineState`.
+//! A strong ref is returned wrapped in a `Handle`.
 //!
 //! For subsequent requests, `Cache::get` will perform a "fast path" shard routing
-//! and find the `Weak<HandleInner>` in the cache.
-//! We upgrade the `Weak<HandleInner>` to an `Arc<HandleInner>` and wrap it in the user-facing `Handle` type.
+//! and find the weak ref in the cache.
+//! We upgrade the weak ref to a strong ref and return it wrapped in a `Handle`.
 //!
-//! The request handler dispatches the request to the right `<Handle as Deref<Target = Timeline>>::$request_method`.
+//! The pagestream processing is pipelined and involves a batching step.
+//! While a request is batching, the `Handle` is downgraded to a `WeakHandle`.
+//! When the batch is ready to be executed, the `WeakHandle` is upgraded back to a `Handle`
+//! and the request handler dispatches the request to the right `<Handle as Deref<Target = Timeline>>::$request_method`.
 //! It then drops the `Handle`, which drops the `Arc<HandleInner>`.
 //!
-//! # Memory Management / How The Reference Cycle Is Broken
+//! # Performance
 //!
-//! The attentive reader may have noticed the strong reference cycle
-//! from `Arc<HandleInner>` to `PerTimelineState` to `Arc<Timeline>`.
+//! Remember from the introductory section:
 //!
-//! This cycle is intentional: while it exists, the `Cache` can upgrade its
-//! `Weak<HandleInner>` to an `Arc<HandleInner>` in a single atomic operation.
+//! > However, we want to avoid the overhead of entering the gate for every
+//! > method invocation.
+//!
+//! Why do we want to avoid that?
+//! Because the gate is a shared location in memory and entering it involves
+//! bumping refcounts, which leads to cache contention if done frequently
+//! from multiple cores in parallel.
+//!
+//! So, we only acquire the `GateGuard` once on `Cache` miss, and wrap it in an `Arc`.
+//! That `Arc` is private to the `HandleInner` and hence to the connection.
+//! (Review the "Data Structures" section if that is unclear to you.)
+//!
+//! A `WeakHandle` is a weak ref to the `HandleInner`.
+//! When upgrading a `WeakHandle`, we upgrade to a strong ref to the `HandleInner` and
+//! further acquire an additional strong ref to the `Arc<GateGuard>` inside it.
+//! Again, this manipulation of ref counts is is cheap because `Arc` is private to the connection.
+//!
+//! When downgrading a `Handle` to a `WeakHandle`, we drop the `Arc<GateGuard>`.
+//! Again, this is cheap because the `Arc` is private to the connection.
+//!
+//! In addition to the GateGuard, we need to provide `Deref<Target=Timeline>` impl.
+//! For this, both `Handle` need infallible access to an `Arc<Timeline>`.
+//! We could clone the `Arc<Timeline>` when upgrading a `WeakHandle`, but that would cause contention
+//! on the shared memory location that trakcs the refcount of the `Arc<Timeline>`.
+//! Instead, we wrap the `Arc<Timeline>` into another `Arc`.
+//! so that we can clone it cheaply when upgrading a `WeakHandle`.
+//!
+//! # Shutdown
+//!
+//! The attentive reader may have noticed the following reference cycle around the `Arc<Timeline>`:
+//!
+//! ```text
+//! Timeline --owns--> PerTimelineState --strong--> HandleInner --strong--> Timeline
+//! ```
+//!
+//! Further, there is this cycle:
+//!
+//! ```text
+//! Timeline --owns--> PerTimelineState --strong--> HandleInner --strong--> GateGuard --keepalive--> Timeline
+//! ```
+//!
+//! The former cycle is a memory leak if not broken.
+//! The latter cycle further prevents the Timeline from shutting down
+//! because we certainly won't drop the Timeline while the GateGuard is alive.
+//! Preventing shutdown is the whole point of this handle/cache system,
+//! but when the Timeline needs to shut down, we need to break the cycle.
 //!
 //! The cycle is broken by either
-//! - `PerTimelineState::shutdown` or
-//! - dropping the `Cache`.
+//! - Timeline shutdown (=> `PerTimelineState::shutdown`)
+//! - Connection shutdown (=> dropping the `Cache`).
 //!
-//! Concurrently existing `Handle`s will extend the existence of the cycle.
+//! Both transition the `HandleInner` from [`HandleInner::KeepingTimelineGateOpen`] to
+//! [`HandleInner::ShutDown`], which drops the only long-lived strong ref to the
+//! `Arc<GateGuard>`.
+//!
+//! `PerTimelineState::shutdown` drops all the `HandleInners` it contains,
+//! thereby breaking the cycle.
+//! It also initiates draining of already existing `Handle`s by
+//! poisoning things so that no new `HandleInner`'s can be added
+//! to the `PerTimelineState`, which will make subsequent `Cache::get` fail.
+//!
+//! Concurrently existing / already upgraded `Handle`s will extend the
+//! lifetime of the `Arc<Mutex<HandleInner>>` and hence cycles.
 //! However, since `Handle`s are short-lived and new `Handle`s are not
-//! handed out after either `PerTimelineState::shutdown` or `Cache` drop,
-//! that extension of the cycle is bounded.
+//! handed out from `Cache::get` or `WeakHandle::upgrade` after
+//! `PerTimelineState::shutdown`, that extension of the cycle is bounded.
+//!
+//! Concurrently existing `WeakHandle`s will fail to `upgrade()`:
+//! while they will succeed in upgrading `Weak<Mutex<HandleInner>>`,
+//! they will find the inner in state `HandleInner::ShutDown` state where the
+//! `Arc<GateGuard>` and Timeline has already been dropped.
+//!
+//! Dropping the `Cache` undoes the registration of this `Cache`'s
+//! `HandleInner`s from all the `PerTimelineState`s, i.e., it
+//! removes the strong ref to each of its `HandleInner`s
+//! from all the `PerTimelineState`.
+//!
+//! # Locking Rules
+//!
+//! To prevent deadlocks we:
+//!
+//! 1. Only ever hold one of the locks at a time.
+//! 2. Don't add more than one Drop impl that locks on the
+//!    cycles above.
+//!
+//! As per (2), that impl is in `Drop for Cache`.
 //!
 //! # Fast Path for Shard Routing
 //!
 //! The `Cache` has a fast path for shard routing to avoid calling into
 //! the tenant manager for every request.
 //!
-//! The `Cache` maintains a hash map of `ShardTimelineId` to `Weak<HandleInner>`.
+//! The `Cache` maintains a hash map of `ShardTimelineId` to `WeakHandle`s.
 //!
 //! The current implementation uses the first entry in the hash map
 //! to determine the `ShardParameters` and derive the correct
@@ -87,18 +184,18 @@
 //!
 //! It then looks up the hash map for that `ShardTimelineId := {ShardIndex,TimelineId}`.
 //!
-//! If the lookup is successful and the `Weak<HandleInner>` can be upgraded,
+//! If the lookup is successful and the `WeakHandle` can be upgraded,
 //! it's a hit.
 //!
 //! ## Cache invalidation
 //!
-//! The insight is that cache invalidation is sufficient and most efficiently done lazily.
+//! The insight is that cache invalidation is sufficient and most efficiently if done lazily.
 //! The only reasons why an entry in the cache can become stale are:
 //! 1. The `PerTimelineState` / Timeline is shutting down e.g. because the shard is
 //!    being detached, timeline or shard deleted, or pageserver is shutting down.
 //! 2. We're doing a shard split and new traffic should be routed to the child shards.
 //!
-//! Regarding (1), we will eventually fail to upgrade the `Weak<HandleInner>` once the
+//! Regarding (1), we will eventually fail to upgrade the `WeakHandle` once the
 //! timeline has shut down, and when that happens, we remove the entry from the cache.
 //!
 //! Regarding (2), the insight is that it is toally fine to keep dispatching requests
@@ -107,8 +204,6 @@
 
 use std::collections::hash_map;
 use std::collections::HashMap;
-use std::sync::atomic::AtomicBool;
-use std::sync::atomic::Ordering;
 use std::sync::Arc;
 use std::sync::Mutex;
 use std::sync::Weak;
@@ -152,7 +247,7 @@ pub(crate) struct Cache<T: Types> {
     map: Map<T>,
 }
 
-type Map<T> = HashMap<ShardTimelineId, Weak<HandleInner<T>>>;
+type Map<T> = HashMap<ShardTimelineId, WeakHandle<T>>;
 
 impl<T: Types> Default for Cache<T> {
     fn default() -> Self {
@@ -170,12 +265,22 @@ pub(crate) struct ShardTimelineId {
 }
 
 /// See module-level comment.
-pub(crate) struct Handle<T: Types>(Arc<HandleInner<T>>);
-struct HandleInner<T: Types> {
-    shut_down: AtomicBool,
-    timeline: T::Timeline,
-    // The timeline's gate held open.
-    _gate_guard: utils::sync::gate::GateGuard,
+pub(crate) struct Handle<T: Types> {
+    timeline: Arc<T::Timeline>,
+    #[allow(dead_code)] // the field exists to keep the gate open
+    gate_guard: Arc<utils::sync::gate::GateGuard>,
+    inner: Arc<Mutex<HandleInner<T>>>,
+}
+pub(crate) struct WeakHandle<T: Types> {
+    inner: Weak<Mutex<HandleInner<T>>>,
+}
+enum HandleInner<T: Types> {
+    KeepingTimelineGateOpen {
+        #[allow(dead_code)]
+        gate_guard: Arc<utils::sync::gate::GateGuard>,
+        timeline: Arc<T::Timeline>,
+    },
+    ShutDown,
 }
 
 /// Embedded in each [`Types::Timeline`] as the anchor for the only long-lived strong ref to `HandleInner`.
@@ -183,7 +288,8 @@ struct HandleInner<T: Types> {
 /// See module-level comment for details.
 pub struct PerTimelineState<T: Types> {
     // None = shutting down
-    handles: Mutex<Option<HashMap<CacheId, Arc<HandleInner<T>>>>>,
+    #[allow(clippy::type_complexity)]
+    handles: Mutex<Option<HashMap<CacheId, Arc<Mutex<HandleInner<T>>>>>>,
 }
 
 impl<T: Types> Default for PerTimelineState<T> {
@@ -243,49 +349,24 @@ impl<T: Types> Cache<T> {
         shard_selector: ShardSelector,
         tenant_manager: &T::TenantManager,
     ) -> Result<Handle<T>, GetError<T>> {
-        // terminates because each iteration removes an element from the map
-        loop {
-            let handle = self
-                .get_impl(timeline_id, shard_selector, tenant_manager)
-                .await?;
-            if handle.0.shut_down.load(Ordering::Relaxed) {
-                let removed = self
-                    .map
-                    .remove(&handle.0.timeline.shard_timeline_id())
-                    .expect("invariant of get_impl is that the returned handle is in the map");
-                assert!(
-                    Weak::ptr_eq(&removed, &Arc::downgrade(&handle.0)),
-                    "shard_timeline_id() incorrect?"
-                );
-            } else {
-                return Ok(handle);
-            }
-        }
-    }
-
-    #[instrument(level = "trace", skip_all)]
-    async fn get_impl(
-        &mut self,
-        timeline_id: TimelineId,
-        shard_selector: ShardSelector,
-        tenant_manager: &T::TenantManager,
-    ) -> Result<Handle<T>, GetError<T>> {
-        let miss: ShardSelector = {
+        // terminates because when every iteration we remove an element from the map
+        let miss: ShardSelector = loop {
             let routing_state = self.shard_routing(timeline_id, shard_selector);
             match routing_state {
                 RoutingResult::FastPath(handle) => return Ok(handle),
                 RoutingResult::SlowPath(key) => match self.map.get(&key) {
                     Some(cached) => match cached.upgrade() {
-                        Some(upgraded) => return Ok(Handle(upgraded)),
-                        None => {
+                        Ok(upgraded) => return Ok(upgraded),
+                        Err(HandleUpgradeError::ShutDown) => {
+                            // TODO: dedup with shard_routing()
                             trace!("handle cache stale");
                             self.map.remove(&key).unwrap();
-                            ShardSelector::Known(key.shard_index)
+                            continue;
                         }
                     },
-                    None => ShardSelector::Known(key.shard_index),
+                    None => break ShardSelector::Known(key.shard_index),
                 },
-                RoutingResult::NeedConsultTenantManager => shard_selector,
+                RoutingResult::NeedConsultTenantManager => break shard_selector,
             }
         };
         self.get_miss(timeline_id, miss, tenant_manager).await
@@ -302,7 +383,7 @@ impl<T: Types> Cache<T> {
             let Some((first_key, first_handle)) = self.map.iter().next() else {
                 return RoutingResult::NeedConsultTenantManager;
             };
-            let Some(first_handle) = first_handle.upgrade() else {
+            let Ok(first_handle) = first_handle.upgrade() else {
                 // TODO: dedup with get()
                 trace!("handle cache stale");
                 let first_key_owned = *first_key;
@@ -310,7 +391,7 @@ impl<T: Types> Cache<T> {
                 continue;
             };
 
-            let first_handle_shard_identity = first_handle.timeline.get_shard_identity();
+            let first_handle_shard_identity = first_handle.get_shard_identity();
             let make_shard_index = |shard_num: ShardNumber| ShardIndex {
                 shard_number: shard_num,
                 shard_count: first_handle_shard_identity.count,
@@ -329,11 +410,11 @@ impl<T: Types> Cache<T> {
             };
             let first_handle_shard_timeline_id = ShardTimelineId {
                 shard_index: first_handle_shard_identity.shard_index(),
-                timeline_id: first_handle.timeline.shard_timeline_id().timeline_id,
+                timeline_id: first_handle.shard_timeline_id().timeline_id,
             };
 
             if need_shard_timeline_id == first_handle_shard_timeline_id {
-                return RoutingResult::FastPath(Handle(first_handle));
+                return RoutingResult::FastPath(first_handle);
             } else {
                 return RoutingResult::SlowPath(need_shard_timeline_id);
             }
@@ -357,23 +438,30 @@ impl<T: Types> Cache<T> {
                     ShardSelector::Known(idx) => assert_eq!(idx, &key.shard_index),
                 }
 
-                let gate_guard = match timeline.gate().enter() {
-                    Ok(guard) => guard,
-                    Err(_) => {
-                        return Err(GetError::TimelineGateClosed);
-                    }
-                };
                 trace!("creating new HandleInner");
-                let handle = Arc::new(
-                    // TODO: global metric that keeps track of the number of live HandlerTimeline instances
-                    // so we can identify reference cycle bugs.
-                    HandleInner {
-                        shut_down: AtomicBool::new(false),
-                        _gate_guard: gate_guard,
-                        timeline: timeline.clone(),
-                    },
-                );
-                let handle = {
+                let handle_inner_arc = Arc::new(Mutex::new(HandleInner::KeepingTimelineGateOpen {
+                    gate_guard: Arc::new(
+                        // this enter() is expensive in production code because
+                        // it hits the global Arc<Timeline>::gate refcounts
+                        match timeline.gate().enter() {
+                            Ok(guard) => guard,
+                            Err(_) => {
+                                return Err(GetError::TimelineGateClosed);
+                            }
+                        },
+                    ),
+                    // this clone is expensive in production code because
+                    // it hits the global Arc<Timeline>::clone refcounts
+                    timeline: Arc::new(timeline.clone()),
+                }));
+                let handle_weak = WeakHandle {
+                    inner: Arc::downgrade(&handle_inner_arc),
+                };
+                let handle = handle_weak
+                    .upgrade()
+                    .ok()
+                    .expect("we just created it and it's not linked anywhere yet");
+                {
                     let mut lock_guard = timeline
                         .per_timeline_state()
                         .handles
@@ -381,7 +469,8 @@ impl<T: Types> Cache<T> {
                         .expect("mutex poisoned");
                     match &mut *lock_guard {
                         Some(per_timeline_state) => {
-                            let replaced = per_timeline_state.insert(self.id, Arc::clone(&handle));
+                            let replaced =
+                                per_timeline_state.insert(self.id, Arc::clone(&handle_inner_arc));
                             assert!(replaced.is_none(), "some earlier code left a stale handle");
                             match self.map.entry(key) {
                                 hash_map::Entry::Occupied(_o) => {
@@ -392,8 +481,7 @@ impl<T: Types> Cache<T> {
                                     unreachable!()
                                 }
                                 hash_map::Entry::Vacant(v) => {
-                                    v.insert(Arc::downgrade(&handle));
-                                    handle
+                                    v.insert(handle_weak);
                                 }
                             }
                         }
@@ -401,14 +489,62 @@ impl<T: Types> Cache<T> {
                             return Err(GetError::PerTimelineStateShutDown);
                         }
                     }
-                };
-                Ok(Handle(handle))
+                }
+                Ok(handle)
             }
             Err(e) => Err(GetError::TenantManager(e)),
         }
     }
 }
 
+pub(crate) enum HandleUpgradeError {
+    ShutDown,
+}
+
+impl<T: Types> WeakHandle<T> {
+    pub(crate) fn upgrade(&self) -> Result<Handle<T>, HandleUpgradeError> {
+        let Some(inner) = Weak::upgrade(&self.inner) else {
+            return Err(HandleUpgradeError::ShutDown);
+        };
+        let lock_guard = inner.lock().expect("poisoned");
+        match &*lock_guard {
+            HandleInner::KeepingTimelineGateOpen {
+                timeline,
+                gate_guard,
+            } => {
+                let gate_guard = Arc::clone(gate_guard);
+                let timeline = Arc::clone(timeline);
+                drop(lock_guard);
+                Ok(Handle {
+                    timeline,
+                    gate_guard,
+                    inner,
+                })
+            }
+            HandleInner::ShutDown => Err(HandleUpgradeError::ShutDown),
+        }
+    }
+
+    pub(crate) fn is_same_handle_as(&self, other: &WeakHandle<T>) -> bool {
+        Weak::ptr_eq(&self.inner, &other.inner)
+    }
+}
+
+impl<T: Types> std::ops::Deref for Handle<T> {
+    type Target = T::Timeline;
+    fn deref(&self) -> &Self::Target {
+        &self.timeline
+    }
+}
+
+impl<T: Types> Handle<T> {
+    pub(crate) fn downgrade(&self) -> WeakHandle<T> {
+        WeakHandle {
+            inner: Arc::downgrade(&self.inner),
+        }
+    }
+}
+
 impl<T: Types> PerTimelineState<T> {
     /// After this method returns, [`Cache::get`] will never again return a [`Handle`]
     /// to the [`Types::Timeline`] that embeds this per-timeline state.
@@ -430,43 +566,62 @@ impl<T: Types> PerTimelineState<T> {
             trace!("already shut down");
             return;
         };
-        for handle in handles.values() {
+        for handle_inner_arc in handles.values() {
             // Make hits fail.
-            handle.shut_down.store(true, Ordering::Relaxed);
+            let mut lock_guard = handle_inner_arc.lock().expect("poisoned");
+            lock_guard.shutdown();
         }
         drop(handles);
     }
 }
 
-impl<T: Types> std::ops::Deref for Handle<T> {
-    type Target = T::Timeline;
-    fn deref(&self) -> &Self::Target {
-        &self.0.timeline
-    }
-}
-
-#[cfg(test)]
-impl<T: Types> Drop for HandleInner<T> {
-    fn drop(&mut self) {
-        trace!("HandleInner dropped");
-    }
-}
-
 // When dropping a [`Cache`], prune its handles in the [`PerTimelineState`] to break the reference cycle.
 impl<T: Types> Drop for Cache<T> {
     fn drop(&mut self) {
-        for (_, weak) in self.map.drain() {
-            if let Some(strong) = weak.upgrade() {
-                // handle is still being kept alive in PerTimelineState
-                let timeline = strong.timeline.per_timeline_state();
-                let mut handles = timeline.handles.lock().expect("mutex poisoned");
-                if let Some(handles) = &mut *handles {
-                    let Some(removed) = handles.remove(&self.id) else {
-                        // There could have been a shutdown inbetween us upgrading the weak and locking the mutex.
-                        continue;
-                    };
-                    assert!(Arc::ptr_eq(&removed, &strong));
-                }
+        for (
+            _,
+            WeakHandle {
+                inner: handle_inner_weak,
+            },
+        ) in self.map.drain()
+        {
+            let Some(handle_inner_arc) = handle_inner_weak.upgrade() else {
+                continue;
+            };
+            let Some(handle_timeline) = handle_inner_arc
+                // locking rules: drop lock before acquiring other lock below
+                .lock()
+                .expect("poisoned")
+                .shutdown()
+            else {
+                // Concurrent PerTimelineState::shutdown.
+                continue;
+            };
+            // Clean up per_timeline_state so the HandleInner allocation can be dropped.
+            let per_timeline_state = handle_timeline.per_timeline_state();
+            let mut handles_lock_guard = per_timeline_state.handles.lock().expect("mutex poisoned");
+            let Some(handles) = &mut *handles_lock_guard else {
+                continue;
+            };
+            let Some(removed_handle_inner_arc) = handles.remove(&self.id) else {
+                // Concurrent PerTimelineState::shutdown.
+                continue;
+            };
+            drop(handles_lock_guard); // locking rules!
+            assert!(Arc::ptr_eq(&removed_handle_inner_arc, &handle_inner_arc));
+        }
+    }
+}
+
+impl<T: Types> HandleInner<T> {
+    fn shutdown(&mut self) -> Option<Arc<T::Timeline>> {
+        match std::mem::replace(self, HandleInner::ShutDown) {
+            HandleInner::KeepingTimelineGateOpen { timeline, .. } => Some(timeline),
+            HandleInner::ShutDown => {
+                // Duplicate shutdowns are possible because both Cache::drop and PerTimelineState::shutdown
+                // may do it concurrently, but locking rules disallow holding per-timeline-state lock and
+                // the handle lock at the same time.
+                None
             }
         }
     }
@@ -474,6 +629,8 @@ impl<T: Types> Drop for Cache<T> {
 
 #[cfg(test)]
 mod tests {
+    use std::sync::Weak;
+
     use pageserver_api::{
         key::{rel_block_to_key, Key, DBDIR_KEY},
         models::ShardParameters,
@@ -583,39 +740,13 @@ mod tests {
         //
         // fill the cache
         //
-        assert_eq!(
-            (Arc::strong_count(&shard0), Arc::weak_count(&shard0)),
-            (2, 1),
-            "strong: shard0, mgr; weak: myself"
-        );
-
         let handle: Handle<_> = cache
             .get(timeline_id, ShardSelector::Page(key), &mgr)
             .await
             .expect("we have the timeline");
-        let handle_inner_weak = Arc::downgrade(&handle.0);
         assert!(Weak::ptr_eq(&handle.myself, &shard0.myself));
-        assert_eq!(
-            (
-                Weak::strong_count(&handle_inner_weak),
-                Weak::weak_count(&handle_inner_weak)
-            ),
-            (2, 2),
-            "strong: handle, per_timeline_state, weak: handle_inner_weak, cache"
-        );
         assert_eq!(cache.map.len(), 1);
-
-        assert_eq!(
-            (Arc::strong_count(&shard0), Arc::weak_count(&shard0)),
-            (3, 1),
-            "strong: handleinner(per_timeline_state), shard0, mgr; weak: myself"
-        );
         drop(handle);
-        assert_eq!(
-            (Arc::strong_count(&shard0), Arc::weak_count(&shard0)),
-            (3, 1),
-            "strong: handleinner(per_timeline_state), shard0, mgr; weak: myself"
-        );
 
         //
         // demonstrate that Handle holds up gate closure
@@ -640,21 +771,11 @@ mod tests {
         // SHUTDOWN
         shard0.per_timeline_state.shutdown(); // keeping handle alive across shutdown
 
-        assert_eq!(
-            1,
-            Weak::strong_count(&handle_inner_weak),
-            "through local var handle"
-        );
         assert_eq!(
             cache.map.len(),
             1,
             "this is an implementation detail but worth pointing out: we can't clear the cache from shutdown(), it's cleared on first access after"
         );
-        assert_eq!(
-            (Arc::strong_count(&shard0), Arc::weak_count(&shard0)),
-            (3, 1),
-            "strong: handleinner(via handle), shard0, mgr; weak: myself"
-        );
 
         // this handle is perfectly usable
         handle.getpage();
@@ -678,16 +799,6 @@ mod tests {
         }
 
         drop(handle);
-        assert_eq!(
-            0,
-            Weak::strong_count(&handle_inner_weak),
-            "the HandleInner destructor already ran"
-        );
-        assert_eq!(
-            (Arc::strong_count(&shard0), Arc::weak_count(&shard0)),
-            (2, 1),
-            "strong: shard0, mgr; weak: myself"
-        );
 
         // closing gate succeeds after dropping handle
         tokio::select! {
@@ -706,10 +817,8 @@ mod tests {
         assert_eq!(cache.map.len(), 0);
 
         // ensure all refs to shard0 are gone and we're not leaking anything
-        let myself = Weak::clone(&shard0.myself);
         drop(shard0);
         drop(mgr);
-        assert_eq!(Weak::strong_count(&myself), 0);
     }
 
     #[tokio::test]
@@ -948,15 +1057,11 @@ mod tests {
                 handle
             };
             handle.getpage();
-            used_handles.push(Arc::downgrade(&handle.0));
+            used_handles.push(Arc::downgrade(&handle.timeline));
         }
 
-        // No handles exist, thus gates are closed and don't require shutdown
-        assert!(used_handles
-            .iter()
-            .all(|weak| Weak::strong_count(weak) == 0));
-
-        // ... thus the gate should close immediately, even without shutdown
+        // No handles exist, thus gates are closed and don't require shutdown.
+        // Thus the gate should close immediately, even without shutdown.
         tokio::select! {
             _ = shard0.gate.close() => { }
             _ = tokio::time::sleep(FOREVER) => {
@@ -964,4 +1069,172 @@ mod tests {
             }
         }
     }
+
+    #[tokio::test(start_paused = true)]
+    async fn test_weak_handles() {
+        crate::tenant::harness::setup_logging();
+        let timeline_id = TimelineId::generate();
+        let shard0 = Arc::new_cyclic(|myself| StubTimeline {
+            gate: Default::default(),
+            id: timeline_id,
+            shard: ShardIdentity::unsharded(),
+            per_timeline_state: PerTimelineState::default(),
+            myself: myself.clone(),
+        });
+        let mgr = StubManager {
+            shards: vec![shard0.clone()],
+        };
+
+        let refcount_start = Arc::strong_count(&shard0);
+
+        let key = DBDIR_KEY;
+
+        let mut cache = Cache::<TestTypes>::default();
+
+        let handle = cache
+            .get(timeline_id, ShardSelector::Page(key), &mgr)
+            .await
+            .expect("we have the timeline");
+        assert!(Weak::ptr_eq(&handle.myself, &shard0.myself));
+
+        let weak_handle = handle.downgrade();
+
+        drop(handle);
+
+        let upgraded_handle = weak_handle.upgrade().ok().expect("we can upgrade it");
+
+        // Start shutdown
+        shard0.per_timeline_state.shutdown();
+
+        // Upgrades during shutdown don't work, even if upgraded_handle exists.
+        weak_handle
+            .upgrade()
+            .err()
+            .expect("can't upgrade weak handle as soon as shutdown started");
+
+        // But upgraded_handle is still alive, so the gate won't close.
+        tokio::select! {
+            _ = shard0.gate.close() => {
+                panic!("handle is keeping gate open");
+            }
+            _ = tokio::time::sleep(FOREVER) => { }
+        }
+
+        // Drop the last handle.
+        drop(upgraded_handle);
+
+        // The gate should close now, despite there still being a weak_handle.
+        tokio::select! {
+            _ = shard0.gate.close() => { }
+            _ = tokio::time::sleep(FOREVER) => {
+                panic!("only strong handle is dropped and we shut down per-timeline-state")
+            }
+        }
+
+        // The weak handle still can't be upgraded.
+        weak_handle
+            .upgrade()
+            .err()
+            .expect("still shouldn't be able to upgrade the weak handle");
+
+        // There should be no strong references to the timeline object except the one on "stack".
+        assert_eq!(Arc::strong_count(&shard0), refcount_start);
+    }
+
+    #[tokio::test(start_paused = true)]
+    async fn test_reference_cycle_broken_when_cache_is_dropped() {
+        crate::tenant::harness::setup_logging();
+        let timeline_id = TimelineId::generate();
+        let shard0 = Arc::new_cyclic(|myself| StubTimeline {
+            gate: Default::default(),
+            id: timeline_id,
+            shard: ShardIdentity::unsharded(),
+            per_timeline_state: PerTimelineState::default(),
+            myself: myself.clone(),
+        });
+        let mgr = StubManager {
+            shards: vec![shard0.clone()],
+        };
+        let key = DBDIR_KEY;
+
+        let mut cache = Cache::<TestTypes>::default();
+
+        // helper to check if a handle is referenced by per_timeline_state
+        let per_timeline_state_refs_handle = |handle_weak: &Weak<Mutex<HandleInner<_>>>| {
+            let per_timeline_state = shard0.per_timeline_state.handles.lock().unwrap();
+            let per_timeline_state = per_timeline_state.as_ref().unwrap();
+            per_timeline_state
+                .values()
+                .any(|v| Weak::ptr_eq(&Arc::downgrade(v), handle_weak))
+        };
+
+        // Fill the cache.
+        let handle = cache
+            .get(timeline_id, ShardSelector::Page(key), &mgr)
+            .await
+            .expect("we have the timeline");
+        assert!(Weak::ptr_eq(&handle.myself, &shard0.myself));
+        let handle_inner_weak = Arc::downgrade(&handle.inner);
+        assert!(
+            per_timeline_state_refs_handle(&handle_inner_weak),
+            "we still hold `handle` _and_ haven't dropped `cache` yet"
+        );
+
+        // Drop the cache.
+        drop(cache);
+
+        assert!(
+            !(per_timeline_state_refs_handle(&handle_inner_weak)),
+            "nothing should reference the handle allocation anymore"
+        );
+        assert!(
+            Weak::upgrade(&handle_inner_weak).is_some(),
+            "the local `handle` still keeps the allocation alive"
+        );
+        // but obviously the cache is gone so no new allocations can be handed out.
+
+        // Drop handle.
+        drop(handle);
+        assert!(
+            Weak::upgrade(&handle_inner_weak).is_none(),
+            "the local `handle` is dropped, so the allocation should be dropped by now"
+        );
+    }
+
+    #[tokio::test(start_paused = true)]
+    async fn test_reference_cycle_broken_when_per_timeline_state_shutdown() {
+        crate::tenant::harness::setup_logging();
+        let timeline_id = TimelineId::generate();
+        let shard0 = Arc::new_cyclic(|myself| StubTimeline {
+            gate: Default::default(),
+            id: timeline_id,
+            shard: ShardIdentity::unsharded(),
+            per_timeline_state: PerTimelineState::default(),
+            myself: myself.clone(),
+        });
+        let mgr = StubManager {
+            shards: vec![shard0.clone()],
+        };
+        let key = DBDIR_KEY;
+
+        let mut cache = Cache::<TestTypes>::default();
+        let handle = cache
+            .get(timeline_id, ShardSelector::Page(key), &mgr)
+            .await
+            .expect("we have the timeline");
+        // grab a weak reference to the inner so can later try to Weak::upgrade it and assert that fails
+        let handle_inner_weak = Arc::downgrade(&handle.inner);
+
+        // drop the handle, obviously the lifetime of `inner` is at least as long as each strong reference to it
+        drop(handle);
+        assert!(Weak::upgrade(&handle_inner_weak).is_some(), "can still");
+
+        // Shutdown the per_timeline_state.
+        shard0.per_timeline_state.shutdown();
+        assert!(Weak::upgrade(&handle_inner_weak).is_none(), "can no longer");
+
+        // cache only contains Weak's, so, it can outlive the per_timeline_state without
+        // Drop explicitly solely to make this point.
+        drop(cache);
+    }
 }

pageserver/src/tenant/timeline/import_pgdata.rs

@@ -113,7 +113,7 @@ pub async fn doit(
             match res {
                 Ok(_) => break,
                 Err(err) => {
-                    info!(?err, "indefintely waiting for pgdata to finish");
+                    info!(?err, "indefinitely waiting for pgdata to finish");
                     if tokio::time::timeout(std::time::Duration::from_secs(10), cancel.cancelled())
                         .await
                         .is_ok()

pageserver/src/tenant/timeline/import_pgdata/importbucket_client.rs

@@ -308,7 +308,7 @@ impl ControlFile {
             202107181 => 14,
             202209061 => 15,
             202307071 => 16,
-            /* XXX pg17 */
+            202406281 => 17,
             catversion => {
                 anyhow::bail!("unrecognized catalog version {catversion}")
             }

pageserver/src/tenant/timeline/layer_manager.rs

@@ -337,16 +337,45 @@ impl OpenLayerManager {
         compact_to: &[ResidentLayer],
         metrics: &TimelineMetrics,
     ) {
-        // We can simply reuse compact l0 logic. Use a different function name to indicate a different type of layer map modification.
-        self.finish_compact_l0(compact_from, compact_to, metrics)
+        // gc-compaction could contain layer rewrites. We need to delete the old layers and insert the new ones.
+
+        // Match the old layers with the new layers
+        let mut add_layers = HashMap::new();
+        let mut rewrite_layers = HashMap::new();
+        let mut drop_layers = HashMap::new();
+        for layer in compact_from {
+            drop_layers.insert(layer.layer_desc().key(), layer.clone());
+        }
+        for layer in compact_to {
+            if let Some(old_layer) = drop_layers.remove(&layer.layer_desc().key()) {
+                rewrite_layers.insert(layer.layer_desc().key(), (old_layer.clone(), layer.clone()));
+            } else {
+                add_layers.insert(layer.layer_desc().key(), layer.clone());
+            }
+        }
+        let add_layers = add_layers.values().cloned().collect::<Vec<_>>();
+        let drop_layers = drop_layers.values().cloned().collect::<Vec<_>>();
+        let rewrite_layers = rewrite_layers.values().cloned().collect::<Vec<_>>();
+
+        self.rewrite_layers_inner(&rewrite_layers, &drop_layers, &add_layers, metrics);
     }
 
     /// Called post-compaction when some previous generation image layers were trimmed.
-    pub(crate) fn rewrite_layers(
+    pub fn rewrite_layers(
         &mut self,
         rewrite_layers: &[(Layer, ResidentLayer)],
         drop_layers: &[Layer],
         metrics: &TimelineMetrics,
+    ) {
+        self.rewrite_layers_inner(rewrite_layers, drop_layers, &[], metrics);
+    }
+
+    fn rewrite_layers_inner(
+        &mut self,
+        rewrite_layers: &[(Layer, ResidentLayer)],
+        drop_layers: &[Layer],
+        add_layers: &[ResidentLayer],
+        metrics: &TimelineMetrics,
     ) {
         let mut updates = self.layer_map.batch_update();
         for (old_layer, new_layer) in rewrite_layers {
@@ -382,6 +411,10 @@ impl OpenLayerManager {
         for l in drop_layers {
             Self::delete_historic_layer(l, &mut updates, &mut self.layer_fmgr);
         }
+        for l in add_layers {
+            Self::insert_historic_layer(l.as_ref().clone(), &mut updates, &mut self.layer_fmgr);
+            metrics.record_new_file_metrics(l.layer_desc().file_size);
+        }
         updates.flush();
     }
 

pageserver/src/tenant/timeline/offload.rs

@@ -2,10 +2,11 @@ use std::sync::Arc;
 
 use pageserver_api::models::{TenantState, TimelineState};
 
-use super::delete::{delete_local_timeline_directory, DeleteTimelineFlow, DeletionGuard};
+use super::delete::{delete_local_timeline_directory, DeletionGuard};
 use super::Timeline;
 use crate::span::debug_assert_current_span_has_tenant_and_timeline_id;
 use crate::tenant::remote_timeline_client::ShutdownIfArchivedError;
+use crate::tenant::timeline::delete::{make_timeline_delete_guard, TimelineDeleteGuardKind};
 use crate::tenant::{OffloadedTimeline, Tenant, TenantManifestError, TimelineOrOffloaded};
 
 #[derive(thiserror::Error, Debug)]
@@ -36,13 +37,10 @@ pub(crate) async fn offload_timeline(
     debug_assert_current_span_has_tenant_and_timeline_id();
     tracing::info!("offloading archived timeline");
 
-    let allow_offloaded_children = true;
-    let set_stopping = false;
-    let (timeline, guard) = DeleteTimelineFlow::prepare(
+    let (timeline, guard) = make_timeline_delete_guard(
         tenant,
         timeline.timeline_id,
-        allow_offloaded_children,
-        set_stopping,
+        TimelineDeleteGuardKind::Offload,
     )
     .map_err(|e| OffloadError::Other(anyhow::anyhow!(e)))?;
 
@@ -106,7 +104,7 @@ pub(crate) async fn offload_timeline(
 }
 
 /// It is important that this gets called when DeletionGuard is being held.
-/// For more context see comments in [`DeleteTimelineFlow::prepare`]
+/// For more context see comments in [`make_timeline_delete_guard`]
 ///
 /// Returns the strong count of the timeline `Arc`
 fn remove_timeline_from_tenant(

pageserver/src/tenant/timeline/walreceiver/connection_manager.rs

@@ -164,9 +164,10 @@ pub(super) async fn connection_manager_loop_step(
                     Ok(Some(broker_update)) => connection_manager_state.register_timeline_update(broker_update),
                     Err(status) => {
                         match status.code() {
-                            Code::Unknown if status.message().contains("stream closed because of a broken pipe") || status.message().contains("connection reset") => {
+                            Code::Unknown if status.message().contains("stream closed because of a broken pipe") || status.message().contains("connection reset") || status.message().contains("error reading a body from connection") => {
                                 // tonic's error handling doesn't provide a clear code for disconnections: we get
                                 // "h2 protocol error: error reading a body from connection: stream closed because of a broken pipe"
+                                // => https://github.com/neondatabase/neon/issues/9562
                                 info!("broker disconnected: {status}");
                             },
                             _ => {
@@ -273,7 +274,7 @@ pub(super) async fn connection_manager_loop_step(
                     };
 
                 last_discovery_ts = Some(std::time::Instant::now());
-                debug!("No active connection and no candidates, sending discovery request to the broker");
+                info!("No active connection and no candidates, sending discovery request to the broker");
 
                 // Cancellation safety: we want to send a message to the broker, but publish_one()
                 // function can get cancelled by the other select! arm. This is absolutely fine, because

pageserver/src/tenant/timeline/walreceiver/walreceiver_connection.rs

@@ -118,7 +118,7 @@ pub(super) async fn handle_walreceiver_connection(
     cancellation: CancellationToken,
     connect_timeout: Duration,
     ctx: RequestContext,
-    node: NodeId,
+    safekeeper_node: NodeId,
     ingest_batch_size: u64,
 ) -> Result<(), WalReceiverError> {
     debug_assert_current_span_has_tenant_and_timeline_id();
@@ -140,7 +140,7 @@ pub(super) async fn handle_walreceiver_connection(
 
     let (replication_client, connection) = {
         let mut config = wal_source_connconf.to_tokio_postgres_config();
-        config.application_name("pageserver");
+        config.application_name(format!("pageserver-{}", timeline.conf.id.0).as_str());
         config.replication_mode(tokio_postgres::config::ReplicationMode::Physical);
         match time::timeout(connect_timeout, config.connect(postgres::NoTls)).await {
             Ok(client_and_conn) => client_and_conn?,
@@ -162,7 +162,7 @@ pub(super) async fn handle_walreceiver_connection(
         latest_wal_update: Utc::now().naive_utc(),
         streaming_lsn: None,
         commit_lsn: None,
-        node,
+        node: safekeeper_node,
     };
     if let Err(e) = events_sender.send(TaskStateUpdate::Progress(connection_status)) {
         warn!("Wal connection event listener dropped right after connection init, aborting the connection: {e}");

pageserver/src/tenant/vectored_blob_io.rs

@@ -35,6 +35,7 @@ use crate::virtual_file::{self, VirtualFile};
 pub struct BlobMeta {
     pub key: Key,
     pub lsn: Lsn,
+    pub will_init: bool,
 }
 
 /// A view into the vectored blobs read buffer.
@@ -310,7 +311,15 @@ pub enum BlobFlag {
 /// * Iterate over the collected blobs and coalesce them into reads at the end
 pub struct VectoredReadPlanner {
     // Track all the blob offsets. Start offsets must be ordered.
-    blobs: BTreeMap<Key, Vec<(Lsn, u64, u64)>>,
+    // Values in the value tuples are:
+    // (
+    //   lsn of the blob,
+    //   start offset of the blob in the underlying file,
+    //   end offset of the blob in the underlying file,
+    //   whether the blob initializes the page image or not
+    //   see [`pageserver_api::record::NeonWalRecord::will_init`]
+    // )
+    blobs: BTreeMap<Key, Vec<(Lsn, u64, u64, bool)>>,
     // Arguments for previous blob passed into [`VectoredReadPlanner::handle`]
     prev: Option<(Key, Lsn, u64, BlobFlag)>,
 
@@ -371,12 +380,12 @@ impl VectoredReadPlanner {
         match flag {
             BlobFlag::None => {
                 let blobs_for_key = self.blobs.entry(key).or_default();
-                blobs_for_key.push((lsn, start_offset, end_offset));
+                blobs_for_key.push((lsn, start_offset, end_offset, false));
             }
             BlobFlag::ReplaceAll => {
                 let blobs_for_key = self.blobs.entry(key).or_default();
                 blobs_for_key.clear();
-                blobs_for_key.push((lsn, start_offset, end_offset));
+                blobs_for_key.push((lsn, start_offset, end_offset, true));
             }
             BlobFlag::Ignore => {}
         }
@@ -387,11 +396,17 @@ impl VectoredReadPlanner {
         let mut reads = Vec::new();
 
         for (key, blobs_for_key) in self.blobs {
-            for (lsn, start_offset, end_offset) in blobs_for_key {
+            for (lsn, start_offset, end_offset, will_init) in blobs_for_key {
                 let extended = match &mut current_read_builder {
-                    Some(read_builder) => {
-                        read_builder.extend(start_offset, end_offset, BlobMeta { key, lsn })
-                    }
+                    Some(read_builder) => read_builder.extend(
+                        start_offset,
+                        end_offset,
+                        BlobMeta {
+                            key,
+                            lsn,
+                            will_init,
+                        },
+                    ),
                     None => VectoredReadExtended::No,
                 };
 
@@ -399,7 +414,11 @@ impl VectoredReadPlanner {
                     let next_read_builder = ChunkedVectoredReadBuilder::new(
                         start_offset,
                         end_offset,
-                        BlobMeta { key, lsn },
+                        BlobMeta {
+                            key,
+                            lsn,
+                            will_init,
+                        },
                         self.max_read_size,
                     );
 
@@ -527,7 +546,7 @@ impl<'a> VectoredBlobReader<'a> {
 pub struct StreamingVectoredReadPlanner {
     read_builder: Option<ChunkedVectoredReadBuilder>,
     // Arguments for previous blob passed into [`StreamingVectoredReadPlanner::handle`]
-    prev: Option<(Key, Lsn, u64)>,
+    prev: Option<(Key, Lsn, u64, bool)>,
     /// Max read size per batch. This is not a strict limit. If there are [0, 100) and [100, 200), while the `max_read_size` is 150,
     /// we will produce a single batch instead of split them.
     max_read_size: u64,
@@ -550,27 +569,47 @@ impl StreamingVectoredReadPlanner {
         }
     }
 
-    pub fn handle(&mut self, key: Key, lsn: Lsn, offset: u64) -> Option<VectoredRead> {
+    pub fn handle(
+        &mut self,
+        key: Key,
+        lsn: Lsn,
+        offset: u64,
+        will_init: bool,
+    ) -> Option<VectoredRead> {
         // Implementation note: internally lag behind by one blob such that
         // we have a start and end offset when initialising [`VectoredRead`]
-        let (prev_key, prev_lsn, prev_offset) = match self.prev {
+        let (prev_key, prev_lsn, prev_offset, prev_will_init) = match self.prev {
             None => {
-                self.prev = Some((key, lsn, offset));
+                self.prev = Some((key, lsn, offset, will_init));
                 return None;
             }
             Some(prev) => prev,
         };
 
-        let res = self.add_blob(prev_key, prev_lsn, prev_offset, offset, false);
+        let res = self.add_blob(
+            prev_key,
+            prev_lsn,
+            prev_offset,
+            offset,
+            false,
+            prev_will_init,
+        );
 
-        self.prev = Some((key, lsn, offset));
+        self.prev = Some((key, lsn, offset, will_init));
 
         res
     }
 
     pub fn handle_range_end(&mut self, offset: u64) -> Option<VectoredRead> {
-        let res = if let Some((prev_key, prev_lsn, prev_offset)) = self.prev {
-            self.add_blob(prev_key, prev_lsn, prev_offset, offset, true)
+        let res = if let Some((prev_key, prev_lsn, prev_offset, prev_will_init)) = self.prev {
+            self.add_blob(
+                prev_key,
+                prev_lsn,
+                prev_offset,
+                offset,
+                true,
+                prev_will_init,
+            )
         } else {
             None
         };
@@ -587,10 +626,19 @@ impl StreamingVectoredReadPlanner {
         start_offset: u64,
         end_offset: u64,
         is_last_blob_in_read: bool,
+        will_init: bool,
     ) -> Option<VectoredRead> {
         match &mut self.read_builder {
             Some(read_builder) => {
-                let extended = read_builder.extend(start_offset, end_offset, BlobMeta { key, lsn });
+                let extended = read_builder.extend(
+                    start_offset,
+                    end_offset,
+                    BlobMeta {
+                        key,
+                        lsn,
+                        will_init,
+                    },
+                );
                 assert_eq!(extended, VectoredReadExtended::Yes);
             }
             None => {
@@ -598,7 +646,11 @@ impl StreamingVectoredReadPlanner {
                     Some(ChunkedVectoredReadBuilder::new_streaming(
                         start_offset,
                         end_offset,
-                        BlobMeta { key, lsn },
+                        BlobMeta {
+                            key,
+                            lsn,
+                            will_init,
+                        },
                     ))
                 };
             }
@@ -812,7 +864,7 @@ mod tests {
         let mut planner = StreamingVectoredReadPlanner::new(max_read_size, 1000);
         let mut reads = Vec::new();
         for (key, lsn, offset, _) in blob_descriptions.clone() {
-            reads.extend(planner.handle(key, lsn, offset));
+            reads.extend(planner.handle(key, lsn, offset, false));
         }
         reads.extend(planner.handle_range_end(652 * 1024));
 
@@ -850,7 +902,7 @@ mod tests {
         let mut planner = StreamingVectoredReadPlanner::new(max_read_size, 2);
         let mut reads = Vec::new();
         for (key, lsn, offset, _) in blob_descriptions.clone() {
-            reads.extend(planner.handle(key, lsn, offset));
+            reads.extend(planner.handle(key, lsn, offset, false));
         }
         reads.extend(planner.handle_range_end(652 * 1024));
 
@@ -875,7 +927,7 @@ mod tests {
         {
             let mut planner = StreamingVectoredReadPlanner::new(max_read_size, 1);
             let mut reads = Vec::new();
-            reads.extend(planner.handle(key, lsn, 0));
+            reads.extend(planner.handle(key, lsn, 0, false));
             reads.extend(planner.handle_range_end(652 * 1024));
             assert_eq!(reads.len(), 1);
             validate_read(&reads[0], &[(key, lsn, 0, BlobFlag::None)]);
@@ -883,8 +935,8 @@ mod tests {
         {
             let mut planner = StreamingVectoredReadPlanner::new(max_read_size, 1);
             let mut reads = Vec::new();
-            reads.extend(planner.handle(key, lsn, 0));
-            reads.extend(planner.handle(key, lsn, 128 * 1024));
+            reads.extend(planner.handle(key, lsn, 0, false));
+            reads.extend(planner.handle(key, lsn, 128 * 1024, false));
             reads.extend(planner.handle_range_end(652 * 1024));
             assert_eq!(reads.len(), 2);
             validate_read(&reads[0], &[(key, lsn, 0, BlobFlag::None)]);
@@ -893,8 +945,8 @@ mod tests {
         {
             let mut planner = StreamingVectoredReadPlanner::new(max_read_size, 2);
             let mut reads = Vec::new();
-            reads.extend(planner.handle(key, lsn, 0));
-            reads.extend(planner.handle(key, lsn, 128 * 1024));
+            reads.extend(planner.handle(key, lsn, 0, false));
+            reads.extend(planner.handle(key, lsn, 128 * 1024, false));
             reads.extend(planner.handle_range_end(652 * 1024));
             assert_eq!(reads.len(), 1);
             validate_read(
@@ -923,6 +975,7 @@ mod tests {
         let meta = BlobMeta {
             key: Key::MIN,
             lsn: Lsn(0),
+            will_init: false,
         };
 
         for (idx, (blob, offset)) in blobs.iter().zip(offsets.iter()).enumerate() {

pageserver/src/utilization.rs

@@ -49,7 +49,7 @@ pub(crate) fn regenerate(
     };
 
     // Express a static value for how many shards we may schedule on one node
-    const MAX_SHARDS: u32 = 20000;
+    const MAX_SHARDS: u32 = 5000;
 
     let mut doc = PageserverUtilization {
         disk_usage_bytes: used,

pageserver/src/walingest.rs

@@ -499,7 +499,13 @@ impl WalIngest {
 
                 let content = modification
                     .tline
-                    .get_rel_page_at_lsn(src_rel, blknum, Version::Modified(modification), ctx)
+                    .get_rel_page_at_lsn(
+                        src_rel,
+                        blknum,
+                        Version::Modified(modification),
+                        ctx,
+                        crate::tenant::storage_layer::IoConcurrency::sequential(),
+                    )
                     .await?;
                 modification.put_rel_page_image(dst_rel, blknum, content)?;
                 num_blocks_copied += 1;
@@ -1489,6 +1495,7 @@ mod tests {
     use super::*;
     use crate::tenant::harness::*;
     use crate::tenant::remote_timeline_client::{remote_initdb_archive_path, INITDB_PATH};
+    use crate::tenant::storage_layer::IoConcurrency;
     use postgres_ffi::RELSEG_SIZE;
 
     use crate::DEFAULT_PG_VERSION;
@@ -1532,6 +1539,7 @@ mod tests {
     #[tokio::test]
     async fn test_relsize() -> Result<()> {
         let (tenant, ctx) = TenantHarness::create("test_relsize").await?.load().await;
+        let io_concurrency = IoConcurrency::spawn_for_test();
         let tline = tenant
             .create_test_timeline(TIMELINE_ID, Lsn(8), DEFAULT_PG_VERSION, &ctx)
             .await?;
@@ -1599,7 +1607,13 @@ mod tests {
         // Check page contents at each LSN
         assert_eq!(
             tline
-                .get_rel_page_at_lsn(TESTREL_A, 0, Version::Lsn(Lsn(0x20)), &ctx)
+                .get_rel_page_at_lsn(
+                    TESTREL_A,
+                    0,
+                    Version::Lsn(Lsn(0x20)),
+                    &ctx,
+                    io_concurrency.clone()
+                )
                 .instrument(test_span.clone())
                 .await?,
             test_img("foo blk 0 at 2")
@@ -1607,7 +1621,13 @@ mod tests {
 
         assert_eq!(
             tline
-                .get_rel_page_at_lsn(TESTREL_A, 0, Version::Lsn(Lsn(0x30)), &ctx)
+                .get_rel_page_at_lsn(
+                    TESTREL_A,
+                    0,
+                    Version::Lsn(Lsn(0x30)),
+                    &ctx,
+                    io_concurrency.clone()
+                )
                 .instrument(test_span.clone())
                 .await?,
             test_img("foo blk 0 at 3")
@@ -1615,14 +1635,26 @@ mod tests {
 
         assert_eq!(
             tline
-                .get_rel_page_at_lsn(TESTREL_A, 0, Version::Lsn(Lsn(0x40)), &ctx)
+                .get_rel_page_at_lsn(
+                    TESTREL_A,
+                    0,
+                    Version::Lsn(Lsn(0x40)),
+                    &ctx,
+                    io_concurrency.clone()
+                )
                 .instrument(test_span.clone())
                 .await?,
             test_img("foo blk 0 at 3")
         );
         assert_eq!(
             tline
-                .get_rel_page_at_lsn(TESTREL_A, 1, Version::Lsn(Lsn(0x40)), &ctx)
+                .get_rel_page_at_lsn(
+                    TESTREL_A,
+                    1,
+                    Version::Lsn(Lsn(0x40)),
+                    &ctx,
+                    io_concurrency.clone()
+                )
                 .instrument(test_span.clone())
                 .await?,
             test_img("foo blk 1 at 4")
@@ -1630,21 +1662,39 @@ mod tests {
 
         assert_eq!(
             tline
-                .get_rel_page_at_lsn(TESTREL_A, 0, Version::Lsn(Lsn(0x50)), &ctx)
+                .get_rel_page_at_lsn(
+                    TESTREL_A,
+                    0,
+                    Version::Lsn(Lsn(0x50)),
+                    &ctx,
+                    io_concurrency.clone()
+                )
                 .instrument(test_span.clone())
                 .await?,
             test_img("foo blk 0 at 3")
         );
         assert_eq!(
             tline
-                .get_rel_page_at_lsn(TESTREL_A, 1, Version::Lsn(Lsn(0x50)), &ctx)
+                .get_rel_page_at_lsn(
+                    TESTREL_A,
+                    1,
+                    Version::Lsn(Lsn(0x50)),
+                    &ctx,
+                    io_concurrency.clone()
+                )
                 .instrument(test_span.clone())
                 .await?,
             test_img("foo blk 1 at 4")
         );
         assert_eq!(
             tline
-                .get_rel_page_at_lsn(TESTREL_A, 2, Version::Lsn(Lsn(0x50)), &ctx)
+                .get_rel_page_at_lsn(
+                    TESTREL_A,
+                    2,
+                    Version::Lsn(Lsn(0x50)),
+                    &ctx,
+                    io_concurrency.clone()
+                )
                 .instrument(test_span.clone())
                 .await?,
             test_img("foo blk 2 at 5")
@@ -1667,14 +1717,26 @@ mod tests {
         );
         assert_eq!(
             tline
-                .get_rel_page_at_lsn(TESTREL_A, 0, Version::Lsn(Lsn(0x60)), &ctx)
+                .get_rel_page_at_lsn(
+                    TESTREL_A,
+                    0,
+                    Version::Lsn(Lsn(0x60)),
+                    &ctx,
+                    io_concurrency.clone()
+                )
                 .instrument(test_span.clone())
                 .await?,
             test_img("foo blk 0 at 3")
         );
         assert_eq!(
             tline
-                .get_rel_page_at_lsn(TESTREL_A, 1, Version::Lsn(Lsn(0x60)), &ctx)
+                .get_rel_page_at_lsn(
+                    TESTREL_A,
+                    1,
+                    Version::Lsn(Lsn(0x60)),
+                    &ctx,
+                    io_concurrency.clone()
+                )
                 .instrument(test_span.clone())
                 .await?,
             test_img("foo blk 1 at 4")
@@ -1689,7 +1751,13 @@ mod tests {
         );
         assert_eq!(
             tline
-                .get_rel_page_at_lsn(TESTREL_A, 2, Version::Lsn(Lsn(0x50)), &ctx)
+                .get_rel_page_at_lsn(
+                    TESTREL_A,
+                    2,
+                    Version::Lsn(Lsn(0x50)),
+                    &ctx,
+                    io_concurrency.clone()
+                )
                 .instrument(test_span.clone())
                 .await?,
             test_img("foo blk 2 at 5")
@@ -1722,14 +1790,26 @@ mod tests {
         );
         assert_eq!(
             tline
-                .get_rel_page_at_lsn(TESTREL_A, 0, Version::Lsn(Lsn(0x70)), &ctx)
+                .get_rel_page_at_lsn(
+                    TESTREL_A,
+                    0,
+                    Version::Lsn(Lsn(0x70)),
+                    &ctx,
+                    io_concurrency.clone()
+                )
                 .instrument(test_span.clone())
                 .await?,
             ZERO_PAGE
         );
         assert_eq!(
             tline
-                .get_rel_page_at_lsn(TESTREL_A, 1, Version::Lsn(Lsn(0x70)), &ctx)
+                .get_rel_page_at_lsn(
+                    TESTREL_A,
+                    1,
+                    Version::Lsn(Lsn(0x70)),
+                    &ctx,
+                    io_concurrency.clone()
+                )
                 .instrument(test_span.clone())
                 .await?,
             test_img("foo blk 1")
@@ -1750,7 +1830,13 @@ mod tests {
         for blk in 2..1500 {
             assert_eq!(
                 tline
-                    .get_rel_page_at_lsn(TESTREL_A, blk, Version::Lsn(Lsn(0x80)), &ctx)
+                    .get_rel_page_at_lsn(
+                        TESTREL_A,
+                        blk,
+                        Version::Lsn(Lsn(0x80)),
+                        &ctx,
+                        io_concurrency.clone()
+                    )
                     .instrument(test_span.clone())
                     .await?,
                 ZERO_PAGE
@@ -1758,7 +1844,13 @@ mod tests {
         }
         assert_eq!(
             tline
-                .get_rel_page_at_lsn(TESTREL_A, 1500, Version::Lsn(Lsn(0x80)), &ctx)
+                .get_rel_page_at_lsn(
+                    TESTREL_A,
+                    1500,
+                    Version::Lsn(Lsn(0x80)),
+                    &ctx,
+                    io_concurrency.clone()
+                )
                 .instrument(test_span.clone())
                 .await?,
             test_img("foo blk 1500")
@@ -1851,6 +1943,7 @@ mod tests {
             .await?
             .load()
             .await;
+        let io_concurrency = IoConcurrency::spawn_for_test();
         let tline = tenant
             .create_test_timeline(TIMELINE_ID, Lsn(8), DEFAULT_PG_VERSION, &ctx)
             .await?;
@@ -1903,7 +1996,13 @@ mod tests {
             let data = format!("foo blk {} at {}", blkno, lsn);
             assert_eq!(
                 tline
-                    .get_rel_page_at_lsn(TESTREL_A, blkno, Version::Lsn(lsn), &ctx)
+                    .get_rel_page_at_lsn(
+                        TESTREL_A,
+                        blkno,
+                        Version::Lsn(lsn),
+                        &ctx,
+                        io_concurrency.clone()
+                    )
                     .instrument(test_span.clone())
                     .await?,
                 test_img(&data)
@@ -1931,7 +2030,13 @@ mod tests {
             let data = format!("foo blk {} at {}", blkno, lsn);
             assert_eq!(
                 tline
-                    .get_rel_page_at_lsn(TESTREL_A, blkno, Version::Lsn(Lsn(0x60)), &ctx)
+                    .get_rel_page_at_lsn(
+                        TESTREL_A,
+                        blkno,
+                        Version::Lsn(Lsn(0x60)),
+                        &ctx,
+                        io_concurrency.clone()
+                    )
                     .instrument(test_span.clone())
                     .await?,
                 test_img(&data)
@@ -1950,7 +2055,13 @@ mod tests {
             let data = format!("foo blk {} at {}", blkno, lsn);
             assert_eq!(
                 tline
-                    .get_rel_page_at_lsn(TESTREL_A, blkno, Version::Lsn(Lsn(0x50)), &ctx)
+                    .get_rel_page_at_lsn(
+                        TESTREL_A,
+                        blkno,
+                        Version::Lsn(Lsn(0x50)),
+                        &ctx,
+                        io_concurrency.clone()
+                    )
                     .instrument(test_span.clone())
                     .await?,
                 test_img(&data)
@@ -1987,7 +2098,13 @@ mod tests {
             let data = format!("foo blk {} at {}", blkno, lsn);
             assert_eq!(
                 tline
-                    .get_rel_page_at_lsn(TESTREL_A, blkno, Version::Lsn(Lsn(0x80)), &ctx)
+                    .get_rel_page_at_lsn(
+                        TESTREL_A,
+                        blkno,
+                        Version::Lsn(Lsn(0x80)),
+                        &ctx,
+                        io_concurrency.clone()
+                    )
                     .instrument(test_span.clone())
                     .await?,
                 test_img(&data)

pgxn/neon/file_cache.c

@@ -911,57 +911,85 @@ lfc_writev(NRelFileInfo rinfo, ForkNumber forkNum, BlockNumber blkno,
 			if (entry->access_count++ == 0)
 				dlist_delete(&entry->list_node);
 		}
-		else
+		/*-----------
+		 * If the chunk wasn't already in the LFC then we have these
+		 * options, in order of preference:
+		 *
+		 * Unless there is no space available, we can:
+		 *  1. Use an entry from the `holes` list, and
+		 *  2. Create a new entry.
+		 * We can always, regardless of space in the LFC:
+		 *  3. evict an entry from LRU, and
+		 *  4. ignore the write operation (the least favorite option)
+		 */
+		else if (lfc_ctl->used < lfc_ctl->limit)
 		{
-			/*
-			 * We have two choices if all cache pages are pinned (i.e. used in IO
-			 * operations):
-			 *
-			 * 1) Wait until some of this operation is completed and pages is
-			 * unpinned.
-			 *
-			 * 2) Allocate one more chunk, so that specified cache size is more
-			 * recommendation than hard limit.
-			 *
-			 * As far as probability of such event (that all pages are pinned) is
-			 * considered to be very very small: there are should be very large
-			 * number of concurrent IO operations and them are limited by
-			 * max_connections, we prefer not to complicate code and use second
-			 * approach.
-			 */
-			if (lfc_ctl->used >= lfc_ctl->limit && !dlist_is_empty(&lfc_ctl->lru))
-			{
-				/* Cache overflow: evict least recently used chunk */
-				FileCacheEntry *victim = dlist_container(FileCacheEntry, list_node, dlist_pop_head_node(&lfc_ctl->lru));
-	
-				for (int i = 0; i < BLOCKS_PER_CHUNK; i++)
-				{
-					lfc_ctl->used_pages -= (victim->bitmap[i >> 5] >> (i & 31)) & 1;
-				}
-				CriticalAssert(victim->access_count == 0);
-				entry->offset = victim->offset; /* grab victim's chunk */
-				hash_search_with_hash_value(lfc_hash, &victim->key, victim->hash, HASH_REMOVE, NULL);
-				neon_log(DEBUG2, "Swap file cache page");
-			}
-			else if (!dlist_is_empty(&lfc_ctl->holes))
+			if (!dlist_is_empty(&lfc_ctl->holes))
 			{
 				/* We can reuse a hole that was left behind when the LFC was shrunk previously */
-				FileCacheEntry *hole = dlist_container(FileCacheEntry, list_node, dlist_pop_head_node(&lfc_ctl->holes));
-				uint32		offset = hole->offset;
-				bool		hole_found;
-	
-				hash_search_with_hash_value(lfc_hash, &hole->key, hole->hash, HASH_REMOVE, &hole_found);
+				FileCacheEntry *hole = dlist_container(FileCacheEntry, list_node,
+													   dlist_pop_head_node(&lfc_ctl->holes));
+				uint32 offset = hole->offset;
+				bool hole_found;
+
+				hash_search_with_hash_value(lfc_hash, &hole->key,
+											hole->hash, HASH_REMOVE, &hole_found);
 				CriticalAssert(hole_found);
-	
+
 				lfc_ctl->used += 1;
-				entry->offset = offset;	/* reuse the hole */
+				entry->offset = offset;			/* reuse the hole */
 			}
 			else
 			{
 				lfc_ctl->used += 1;
-				entry->offset = lfc_ctl->size++;	/* allocate new chunk at end
-													 * of file */
+				entry->offset = lfc_ctl->size++;/* allocate new chunk at end
+												 * of file */
 			}
+		}
+		/*
+		 * We've already used up all allocated LFC entries.
+		 *
+		 * If we can clear an entry from the LRU, do that.
+		 * If we can't (e.g. because all other slots are being accessed)
+		 * then we will remove this entry from the hash and continue
+		 * on to the next chunk, as we may not exceed the limit.
+		 */
+		else if (!dlist_is_empty(&lfc_ctl->lru))
+		{
+			/* Cache overflow: evict least recently used chunk */
+			FileCacheEntry *victim = dlist_container(FileCacheEntry, list_node,
+													 dlist_pop_head_node(&lfc_ctl->lru));
+
+			for (int i = 0; i < BLOCKS_PER_CHUNK; i++)
+			{
+				lfc_ctl->used_pages -= (victim->bitmap[i >> 5] >> (i & 31)) & 1;
+			}
+
+			CriticalAssert(victim->access_count == 0);
+			entry->offset = victim->offset; /* grab victim's chunk */
+			hash_search_with_hash_value(lfc_hash, &victim->key,
+										victim->hash, HASH_REMOVE, NULL);
+			neon_log(DEBUG2, "Swap file cache page");
+		}
+		else
+		{
+			/* Can't add this chunk - we don't have the space for it */
+			hash_search_with_hash_value(lfc_hash, &entry->key, hash,
+										HASH_REMOVE, NULL);
+
+			/*
+			 * We can't process this chunk due to lack of space in LFC,
+			 * so skip to the next one
+			 */
+			LWLockRelease(lfc_lock);
+			blkno += blocks_in_chunk;
+			buf_offset += blocks_in_chunk;
+			nblocks -= blocks_in_chunk;
+			continue;
+		}
+
+		if (!found)
+		{
 			entry->access_count = 1;
 			entry->hash = hash;
 			memset(entry->bitmap, 0, sizeof entry->bitmap);

pgxn/neon/neon.c

@@ -19,6 +19,7 @@
 #include "access/xlogrecovery.h"
 #endif
 #include "replication/logical.h"
+#include "replication/logicallauncher.h"
 #include "replication/slot.h"
 #include "replication/walsender.h"
 #include "storage/proc.h"
@@ -434,6 +435,15 @@ _PG_init(void)
 
 	restore_running_xacts_callback = RestoreRunningXactsFromClog;
 
+	DefineCustomBoolVariable(
+							"neon.disable_logical_replication_subscribers",
+							"Disables incomming logical replication",
+							NULL,
+							&disable_logical_replication_subscribers,
+							false,
+							PGC_SIGHUP,
+							0,
+							NULL, NULL, NULL);
 
 	DefineCustomBoolVariable(
 							"neon.allow_replica_misconfig",