Try to find the best parallelisation

This reverts commit a0e923b70b.

.devcontainer/Dockerfile.devcontainer

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

.devcontainer/devcontainer.json

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

.github/actionlint.yml

@@ -7,13 +7,6 @@ self-hosted-runner:
     - small-arm64
     - us-east-2
 config-variables:
-  - AZURE_DEV_CLIENT_ID
-  - AZURE_DEV_REGISTRY_NAME
-  - AZURE_DEV_SUBSCRIPTION_ID
-  - AZURE_PROD_CLIENT_ID
-  - AZURE_PROD_REGISTRY_NAME
-  - AZURE_PROD_SUBSCRIPTION_ID
-  - AZURE_TENANT_ID
   - BENCHMARK_PROJECT_ID_PUB
   - BENCHMARK_PROJECT_ID_SUB
   - REMOTE_STORAGE_AZURE_CONTAINER

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

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

.github/workflows/build_and_test.yml

@@ -122,10 +122,12 @@ jobs:
 
   check-codestyle-rust:
     needs: [ check-permissions, build-build-tools-image ]
+    # There's no reason to expect clippy or code formatting to be different on different platforms,
+    # so it's enough to run these on x64 only.
     strategy:
       matrix:
         arch: [ x64, arm64 ]
-    runs-on: ${{ fromJson(format('["self-hosted", "{0}"]', matrix.arch == 'arm64' && 'small-arm64' || 'small')) }}
+    runs-on: ${{ fromJson(format('["self-hosted", "{0}"]', matrix.arch == 'arm64' && 'large-arm64' || 'large')) }}
 
     container:
       image: ${{ needs.build-build-tools-image.outputs.image }}
@@ -166,15 +168,25 @@ jobs:
             exit 1
           fi
           echo "CLIPPY_COMMON_ARGS=${CLIPPY_COMMON_ARGS}" >> $GITHUB_ENV
+
       - name: Run cargo clippy (debug)
-        run: cargo hack --feature-powerset clippy $CLIPPY_COMMON_ARGS
+        run: |
-      - name: Run cargo clippy (release)
+          parallel --jobs 8 "cargo hack --feature-powerset --partition {}/8 clippy --target-dir target/partition-{} $CLIPPY_COMMON_ARGS" ::: 1 2 3 4 5 6 7 8
-        run: cargo hack --feature-powerset clippy --release $CLIPPY_COMMON_ARGS
+
+      # instead of running the full release build, running debug build again,
+      # but with disabled `debug-assertions` to excersice release code paths
+      - name: Run cargo clippy (debug, with debug-assertions=false)
+        run: |
+          for N in 4 8 10 12 14 16 18 20; do
+            echo "Running clippy with debug-assertions=false for partition ${N}"
+            time parallel --jobs ${N} "cargo hack --feature-powerset --partition {}/${N} clippy --target-dir target/partition-{} $CLIPPY_COMMON_ARGS -C debug-assertions=off" ::: $(seq -s " " 1 ${N})
+            rm -rf target/partition-*
+          done
 
       - name: Check documentation generation
         run: cargo doc --workspace --no-deps --document-private-items
         env:
-            RUSTDOCFLAGS: "-Dwarnings -Arustdoc::private_intra_doc_links"
+          RUSTDOCFLAGS: "-Dwarnings -Arustdoc::private_intra_doc_links"
 
       # Use `${{ !cancelled() }}` to run quck tests after the longer clippy run
       - name: Check formatting
@@ -286,7 +298,6 @@ jobs:
           PERF_TEST_RESULT_CONNSTR: "${{ secrets.PERF_TEST_RESULT_CONNSTR }}"
           TEST_RESULT_CONNSTR: "${{ secrets.REGRESS_TEST_RESULT_CONNSTR_NEW }}"
           PAGESERVER_VIRTUAL_FILE_IO_ENGINE: tokio-epoll-uring
-          SYNC_AFTER_EACH_TEST: true
       # XXX: no coverage data handling here, since benchmarks are run on release builds,
       # while coverage is currently collected for the debug ones
 
@@ -794,6 +805,9 @@ jobs:
           docker compose -f ./docker-compose/docker-compose.yml down
 
   promote-images:
+    permissions:
+      contents: read  # This is required for actions/checkout
+      id-token: write # This is required for Azure Login to work.
     needs: [ check-permissions, tag, test-images, vm-compute-node-image ]
     runs-on: ubuntu-22.04
 
@@ -820,6 +834,28 @@ jobs:
                                                neondatabase/vm-compute-node-${version}:${{ needs.tag.outputs.build-tag }}
           done
 
+      - name: Azure login
+        if: github.ref_name == 'main'
+        uses: azure/login@6c251865b4e6290e7b78be643ea2d005bc51f69a  # @v2.1.1
+        with:
+          client-id: ${{ secrets.AZURE_DEV_CLIENT_ID }}
+          tenant-id: ${{ secrets.AZURE_TENANT_ID }}
+          subscription-id: ${{ secrets.AZURE_DEV_SUBSCRIPTION_ID }}
+
+      - name: Login to ACR
+        if: github.ref_name == 'main'
+        run: |
+          az acr login --name=neoneastus2
+
+      - name: Copy docker images to ACR-dev
+        if: github.ref_name == 'main'
+        run: |
+          for image in neon compute-tools {vm-,}compute-node-{v14,v15,v16}; do
+            docker buildx imagetools create \
+              -t neoneastus2.azurecr.io/neondatabase/${image}:${{ needs.tag.outputs.build-tag }} \
+                                        neondatabase/${image}:${{ needs.tag.outputs.build-tag }}
+          done
+
       - name: Add latest tag to images
         if: github.ref_name == 'main'
         run: |
@@ -857,30 +893,6 @@ jobs:
                                                369495373322.dkr.ecr.eu-central-1.amazonaws.com/${image}:${{ needs.tag.outputs.build-tag }}
           done
 
-  push-to-acr-dev:
-    if: github.ref_name == 'main'
-    needs: [ tag, promote-images ]
-    uses: ./.github/workflows/_push-to-acr.yml
-    with:
-      client_id: ${{ vars.AZURE_DEV_CLIENT_ID }}
-      image_tag: ${{ needs.tag.outputs.build-tag }}
-      images: neon compute-tools vm-compute-node-v14 vm-compute-node-v15 vm-compute-node-v16 compute-node-v14 compute-node-v15 compute-node-v16
-      registry_name: ${{ vars.AZURE_DEV_REGISTRY_NAME }}
-      subscription_id: ${{ vars.AZURE_DEV_SUBSCRIPTION_ID }}
-      tenant_id: ${{ vars.AZURE_TENANT_ID }}
-
-  push-to-acr-prod:
-    if: github.ref_name == 'release'|| github.ref_name == 'release-proxy'
-    needs: [ tag, promote-images ]
-    uses: ./.github/workflows/_push-to-acr.yml
-    with:
-      client_id: ${{ vars.AZURE_PROD_CLIENT_ID }}
-      image_tag: ${{ needs.tag.outputs.build-tag }}
-      images: neon compute-tools vm-compute-node-v14 vm-compute-node-v15 vm-compute-node-v16 compute-node-v14 compute-node-v15 compute-node-v16
-      registry_name: ${{ vars.AZURE_PROD_REGISTRY_NAME }}
-      subscription_id: ${{ vars.AZURE_PROD_SUBSCRIPTION_ID }}
-      tenant_id: ${{ vars.AZURE_TENANT_ID }}
-
   trigger-custom-extensions-build-and-wait:
     needs: [ check-permissions, tag ]
     runs-on: ubuntu-22.04
@@ -956,8 +968,8 @@ jobs:
           exit 1
 
   deploy:
-    needs: [ check-permissions, promote-images, tag, build-and-test-locally, trigger-custom-extensions-build-and-wait, push-to-acr-dev, push-to-acr-prod ]
+    needs: [ check-permissions, promote-images, tag, build-and-test-locally, trigger-custom-extensions-build-and-wait ]
-    if: (github.ref_name == 'main' || github.ref_name == 'release' || github.ref_name == 'release-proxy') && !failure() && !cancelled()
+    if: github.ref_name == 'main' || github.ref_name == 'release'|| github.ref_name == 'release-proxy'
 
     runs-on: [ self-hosted, small ]
     container: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/ansible:latest

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

@@ -7,11 +7,6 @@ on:
   pull_request_target:
     types:
       - opened
-  workflow_dispatch:
-    inputs:
-      github-actor:
-        description: 'GitHub username. If empty, the username of the current user will be used'
-        required: false
 
 # No permission for GITHUB_TOKEN by default; the **minimal required** set of permissions should be granted in each job.
 permissions: {}
@@ -31,31 +26,12 @@ jobs:
       id: check-user
       env:
         GH_TOKEN: ${{ secrets.CI_ACCESS_TOKEN }}
-        ACTOR: ${{ inputs.github-actor || github.actor }}
       run: |
-        expected_error="User does not exist or is not a member of the organization"
+        if gh api -H "Accept: application/vnd.github+json" -H "X-GitHub-Api-Version: 2022-11-28" "/orgs/${GITHUB_REPOSITORY_OWNER}/members/${GITHUB_ACTOR}"; then
-        output_file=output.txt
+          is_member=true
-
+        else
-        for i in $(seq 1 10); do
+          is_member=false
-          if gh api "/orgs/${GITHUB_REPOSITORY_OWNER}/members/${ACTOR}" \
+        fi
-              -H "Accept: application/vnd.github+json" \
-              -H "X-GitHub-Api-Version: 2022-11-28" > ${output_file}; then
-
-            is_member=true
-            break
-          elif grep -q "${expected_error}" ${output_file}; then
-            is_member=false
-            break
-          elif [ $i -eq 10 ]; then
-            title="Failed to get memmbership status for ${ACTOR}"
-            message="The latest GitHub API error message: '$(cat ${output_file})'"
-            echo "::error file=.github/workflows/label-for-external-users.yml,title=${title}::${message}"
-
-            exit 1
-          fi
-
-          sleep 1
-        done
 
         echo "is-member=${is_member}" | tee -a ${GITHUB_OUTPUT}
 

Cargo.lock

@@ -915,22 +915,25 @@ dependencies = [
 
 [[package]]
 name = "bindgen"
-version = "0.70.1"
+version = "0.65.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f49d8fed880d473ea71efb9bf597651e77201bdd4893efe54c9e5d65ae04ce6f"
+checksum = "cfdf7b466f9a4903edc73f95d6d2bcd5baf8ae620638762244d3f60143643cc5"
 dependencies = [
- "bitflags 2.4.1",
+ "bitflags 1.3.2",
  "cexpr",
  "clang-sys",
- "itertools 0.12.1",
+ "lazy_static",
+ "lazycell",
  "log",
- "prettyplease 0.2.17",
+ "peeking_take_while",
+ "prettyplease 0.2.6",
  "proc-macro2",
  "quote",
  "regex",
  "rustc-hash",
  "shlex",
  "syn 2.0.52",
+ "which",
 ]
 
 [[package]]
@@ -1189,9 +1192,9 @@ dependencies = [
 
 [[package]]
 name = "comfy-table"
-version = "7.1.1"
+version = "6.1.4"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b34115915337defe99b2aff5c2ce6771e5fbc4079f4b506301f5cf394c8452f7"
+checksum = "6e7b787b0dc42e8111badfdbe4c3059158ccb2db8780352fa1b01e8ccf45cc4d"
 dependencies = [
  "crossterm",
  "strum",
@@ -1246,7 +1249,7 @@ dependencies = [
  "tokio-postgres",
  "tokio-stream",
  "tokio-util",
- "toml_edit",
+ "toml_edit 0.19.10",
  "tracing",
  "tracing-opentelemetry",
  "tracing-subscriber",
@@ -1360,8 +1363,8 @@ dependencies = [
  "tokio",
  "tokio-postgres",
  "tokio-util",
- "toml",
+ "toml 0.7.4",
- "toml_edit",
+ "toml_edit 0.19.10",
  "tracing",
  "url",
  "utils",
@@ -1485,22 +1488,25 @@ checksum = "248e3bacc7dc6baa3b21e405ee045c3047101a49145e7e9eca583ab4c2ca5345"
 
 [[package]]
 name = "crossterm"
-version = "0.27.0"
+version = "0.25.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f476fe445d41c9e991fd07515a6f463074b782242ccf4a5b7b1d1012e70824df"
+checksum = "e64e6c0fbe2c17357405f7c758c1ef960fce08bdfb2c03d88d2a18d7e09c4b67"
 dependencies = [
- "bitflags 2.4.1",
+ "bitflags 1.3.2",
  "crossterm_winapi",
  "libc",
+ "mio",
  "parking_lot 0.12.1",
+ "signal-hook",
+ "signal-hook-mio",
  "winapi",
 ]
 
 [[package]]
 name = "crossterm_winapi"
-version = "0.9.1"
+version = "0.9.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "acdd7c62a3665c7f6830a51635d9ac9b23ed385797f70a83bb8bafe9c572ab2b"
+checksum = "2ae1b35a484aa10e07fe0638d02301c5ad24de82d310ccbd2f3693da5f09bf1c"
 dependencies = [
  "winapi",
 ]
@@ -2721,12 +2727,6 @@ dependencies = [
  "hashbrown 0.14.5",
 ]
 
-[[package]]
-name = "indoc"
-version = "2.0.5"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b248f5224d1d606005e02c97f5aa4e88eeb230488bcc03bc9ca4d7991399f2b5"
-
 [[package]]
 name = "infer"
 version = "0.2.3"
@@ -2943,6 +2943,12 @@ dependencies = [
  "spin 0.5.2",
 ]
 
+[[package]]
+name = "lazycell"
+version = "1.3.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "830d08ce1d1d941e6b30645f1a0eb5643013d835ce3779a5fc208261dbe10f55"
+
 [[package]]
 name = "libc"
 version = "0.2.150"
@@ -3141,7 +3147,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "fd01039851e82f8799046eabbb354056283fb265c8ec0996af940f4e85a380ff"
 dependencies = [
  "serde",
- "toml",
+ "toml 0.8.14",
 ]
 
 [[package]]
@@ -3657,7 +3663,7 @@ dependencies = [
  "thiserror",
  "tokio",
  "tokio-util",
- "toml_edit",
+ "toml_edit 0.19.10",
  "utils",
  "workspace_hack",
 ]
@@ -3695,7 +3701,6 @@ dependencies = [
  "humantime",
  "humantime-serde",
  "hyper 0.14.26",
- "indoc",
  "itertools 0.10.5",
  "md5",
  "metrics",
@@ -3744,7 +3749,7 @@ dependencies = [
  "tokio-stream",
  "tokio-tar",
  "tokio-util",
- "toml_edit",
+ "toml_edit 0.19.10",
  "tracing",
  "twox-hash",
  "url",
@@ -3761,7 +3766,6 @@ dependencies = [
  "bincode",
  "byteorder",
  "bytes",
- "camino",
  "chrono",
  "const_format",
  "enum-map",
@@ -3769,16 +3773,11 @@ dependencies = [
  "humantime",
  "humantime-serde",
  "itertools 0.10.5",
- "nix 0.27.1",
- "postgres_backend",
  "postgres_ffi",
  "rand 0.8.5",
- "remote_storage",
- "reqwest 0.12.4",
  "serde",
  "serde_json",
  "serde_with",
- "storage_broker",
  "strum",
  "strum_macros",
  "thiserror",
@@ -3907,9 +3906,8 @@ dependencies = [
 
 [[package]]
 name = "parquet"
-version = "53.0.0"
+version = "51.0.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
+source = "git+https://github.com/apache/arrow-rs?branch=master#2534976a564be3d2d56312dc88fb1b6ed4cef829"
-checksum = "f0fbf928021131daaa57d334ca8e3904fe9ae22f73c56244fc7db9b04eedc3d8"
 dependencies = [
  "ahash",
  "bytes",
@@ -3928,9 +3926,8 @@ dependencies = [
 
 [[package]]
 name = "parquet_derive"
-version = "53.0.0"
+version = "51.0.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
+source = "git+https://github.com/apache/arrow-rs?branch=master#2534976a564be3d2d56312dc88fb1b6ed4cef829"
-checksum = "86e9fcfae007533a06b580429a3f7e07cb833ec8aa37c041c16563e7918f057e"
 dependencies = [
  "parquet",
  "proc-macro2",
@@ -3967,6 +3964,12 @@ dependencies = [
  "sha2",
 ]
 
+[[package]]
+name = "peeking_take_while"
+version = "0.1.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "19b17cddbe7ec3f8bc800887bab5e717348c95ea2ca0b1bf0837fb964dc67099"
+
 [[package]]
 name = "pem"
 version = "3.0.3"
@@ -4120,7 +4123,7 @@ dependencies = [
 [[package]]
 name = "postgres"
 version = "0.19.4"
-source = "git+https://github.com/neondatabase/rust-postgres.git?rev=20031d7a9ee1addeae6e0968e3899ae6bf01cee2#20031d7a9ee1addeae6e0968e3899ae6bf01cee2"
+source = "git+https://github.com/neondatabase/rust-postgres.git?branch=neon#20031d7a9ee1addeae6e0968e3899ae6bf01cee2"
 dependencies = [
  "bytes",
  "fallible-iterator",
@@ -4133,7 +4136,7 @@ dependencies = [
 [[package]]
 name = "postgres-protocol"
 version = "0.6.4"
-source = "git+https://github.com/neondatabase/rust-postgres.git?rev=20031d7a9ee1addeae6e0968e3899ae6bf01cee2#20031d7a9ee1addeae6e0968e3899ae6bf01cee2"
+source = "git+https://github.com/neondatabase/rust-postgres.git?branch=neon#20031d7a9ee1addeae6e0968e3899ae6bf01cee2"
 dependencies = [
  "base64 0.20.0",
  "byteorder",
@@ -4152,7 +4155,7 @@ dependencies = [
 [[package]]
 name = "postgres-types"
 version = "0.2.4"
-source = "git+https://github.com/neondatabase/rust-postgres.git?rev=20031d7a9ee1addeae6e0968e3899ae6bf01cee2#20031d7a9ee1addeae6e0968e3899ae6bf01cee2"
+source = "git+https://github.com/neondatabase/rust-postgres.git?branch=neon#20031d7a9ee1addeae6e0968e3899ae6bf01cee2"
 dependencies = [
  "bytes",
  "fallible-iterator",
@@ -4264,9 +4267,9 @@ dependencies = [
 
 [[package]]
 name = "prettyplease"
-version = "0.2.17"
+version = "0.2.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "8d3928fb5db768cb86f891ff014f0144589297e3c6a1aba6ed7cecfdace270c7"
+checksum = "3b69d39aab54d069e7f2fe8cb970493e7834601ca2d8c65fd7bbd183578080d1"
 dependencies = [
  "proc-macro2",
  "syn 2.0.52",
@@ -4811,7 +4814,7 @@ dependencies = [
  "tokio",
  "tokio-stream",
  "tokio-util",
- "toml_edit",
+ "toml_edit 0.19.10",
  "tracing",
  "utils",
 ]
@@ -5321,7 +5324,7 @@ dependencies = [
  "tokio-stream",
  "tokio-tar",
  "tokio-util",
- "toml_edit",
+ "toml_edit 0.19.10",
  "tracing",
  "tracing-subscriber",
  "url",
@@ -5730,6 +5733,17 @@ dependencies = [
  "signal-hook-registry",
 ]
 
+[[package]]
+name = "signal-hook-mio"
+version = "0.2.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "29ad2e15f37ec9a6cc544097b78a1ec90001e9f71b81338ca39f430adaca99af"
+dependencies = [
+ "libc",
+ "mio",
+ "signal-hook",
+]
+
 [[package]]
 name = "signal-hook-registry"
 version = "1.4.1"
@@ -6042,21 +6056,21 @@ checksum = "73473c0e59e6d5812c5dfe2a064a6444949f089e20eec9a2e5506596494e4623"
 
 [[package]]
 name = "strum"
-version = "0.26.3"
+version = "0.24.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "8fec0f0aef304996cf250b31b5a10dee7980c85da9d759361292b8bca5a18f06"
+checksum = "063e6045c0e62079840579a7e47a355ae92f60eb74daaf156fb1e84ba164e63f"
 
 [[package]]
 name = "strum_macros"
-version = "0.26.4"
+version = "0.24.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "4c6bee85a5a24955dc440386795aa378cd9cf82acd5f764469152d2270e581be"
+checksum = "1e385be0d24f186b4ce2f9982191e7101bb737312ad61c1f2f984f34bcf85d59"
 dependencies = [
- "heck 0.5.0",
+ "heck 0.4.1",
  "proc-macro2",
  "quote",
  "rustversion",
- "syn 2.0.52",
+ "syn 1.0.109",
 ]
 
 [[package]]
@@ -6067,9 +6081,8 @@ checksum = "81cdd64d312baedb58e21336b31bc043b77e01cc99033ce76ef539f78e965ebc"
 
 [[package]]
 name = "svg_fmt"
-version = "0.4.3"
+version = "0.4.2"
-source = "registry+https://github.com/rust-lang/crates.io-index"
+source = "git+https://github.com/nical/rust_debug?rev=28a7d96eecff2f28e75b1ea09f2d499a60d0e3b4#28a7d96eecff2f28e75b1ea09f2d499a60d0e3b4"
-checksum = "20e16a0f46cf5fd675563ef54f26e83e20f2366bcf027bcb3cc3ed2b98aaf2ca"
 
 [[package]]
 name = "syn"
@@ -6397,7 +6410,7 @@ dependencies = [
 [[package]]
 name = "tokio-postgres"
 version = "0.7.7"
-source = "git+https://github.com/neondatabase/rust-postgres.git?rev=20031d7a9ee1addeae6e0968e3899ae6bf01cee2#20031d7a9ee1addeae6e0968e3899ae6bf01cee2"
+source = "git+https://github.com/neondatabase/rust-postgres.git?branch=neon#20031d7a9ee1addeae6e0968e3899ae6bf01cee2"
 dependencies = [
  "async-trait",
  "byteorder",
@@ -6508,6 +6521,18 @@ dependencies = [
  "tracing",
 ]
 
+[[package]]
+name = "toml"
+version = "0.7.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d6135d499e69981f9ff0ef2167955a5333c35e36f6937d382974566b3d5b94ec"
+dependencies = [
+ "serde",
+ "serde_spanned",
+ "toml_datetime",
+ "toml_edit 0.19.10",
+]
+
 [[package]]
 name = "toml"
 version = "0.8.14"
@@ -6517,7 +6542,7 @@ dependencies = [
  "serde",
  "serde_spanned",
  "toml_datetime",
- "toml_edit",
+ "toml_edit 0.22.14",
 ]
 
 [[package]]
@@ -6529,6 +6554,19 @@ dependencies = [
  "serde",
 ]
 
+[[package]]
+name = "toml_edit"
+version = "0.19.10"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "2380d56e8670370eee6566b0bfd4265f65b3f432e8c6d85623f728d4fa31f739"
+dependencies = [
+ "indexmap 1.9.3",
+ "serde",
+ "serde_spanned",
+ "toml_datetime",
+ "winnow 0.4.6",
+]
+
 [[package]]
 name = "toml_edit"
 version = "0.22.14"
@@ -6539,7 +6577,7 @@ dependencies = [
  "serde",
  "serde_spanned",
  "toml_datetime",
- "winnow",
+ "winnow 0.6.13",
 ]
 
 [[package]]
@@ -6952,7 +6990,7 @@ dependencies = [
  "tokio-stream",
  "tokio-tar",
  "tokio-util",
- "toml_edit",
+ "toml_edit 0.19.10",
  "tracing",
  "tracing-error",
  "tracing-subscriber",
@@ -7498,6 +7536,15 @@ version = "0.52.4"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "32b752e52a2da0ddfbdbcc6fceadfeede4c939ed16d13e648833a61dfb611ed8"
 
+[[package]]
+name = "winnow"
+version = "0.4.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "61de7bac303dc551fe038e2b3cef0f571087a47571ea6e79a87692ac99b99699"
+dependencies = [
+ "memchr",
+]
+
 [[package]]
 name = "winnow"
 version = "0.6.13"
@@ -7567,7 +7614,6 @@ dependencies = [
  "hyper 0.14.26",
  "indexmap 1.9.3",
  "itertools 0.10.5",
- "itertools 0.12.1",
  "lazy_static",
  "libc",
  "log",
@@ -7605,7 +7651,6 @@ dependencies = [
  "tokio",
  "tokio-rustls 0.24.0",
  "tokio-util",
- "toml_edit",
  "tonic",
  "tower",
  "tracing",

Cargo.toml

@@ -64,7 +64,7 @@ aws-types = "1.2.0"
 axum = { version = "0.6.20", features = ["ws"] }
 base64 = "0.13.0"
 bincode = "1.3"
-bindgen = "0.70"
+bindgen = "0.65"
 bit_field = "0.10.2"
 bstr = "1.0"
 byteorder = "1.4"
@@ -73,7 +73,7 @@ camino = "1.1.6"
 cfg-if = "1.0.0"
 chrono = { version = "0.4", default-features = false, features = ["clock"] }
 clap = { version = "4.0", features = ["derive"] }
-comfy-table = "7.1"
+comfy-table = "6.1"
 const_format = "0.2"
 crc32c = "0.6"
 crossbeam-deque = "0.8.5"
@@ -103,7 +103,6 @@ humantime-serde = "1.1.1"
 hyper = "0.14"
 tokio-tungstenite = "0.20.0"
 indexmap = "2"
-indoc = "2"
 inotify = "0.10.2"
 ipnet = "2.9.0"
 itertools = "0.10"
@@ -123,8 +122,8 @@ opentelemetry = "0.20.0"
 opentelemetry-otlp = { version = "0.13.0", default-features=false, features = ["http-proto", "trace", "http", "reqwest-client"] }
 opentelemetry-semantic-conventions = "0.12.0"
 parking_lot = "0.12"
-parquet = { version = "53", default-features = false, features = ["zstd"] }
+parquet = { version = "51.0.0", default-features = false, features = ["zstd"] }
-parquet_derive = "53"
+parquet_derive = "51.0.0"
 pbkdf2 = { version = "0.12.1", features = ["simple", "std"] }
 pin-project-lite = "0.2"
 procfs = "0.16"
@@ -158,10 +157,11 @@ signal-hook = "0.3"
 smallvec = "1.11"
 smol_str = { version = "0.2.0", features = ["serde"] }
 socket2 = "0.5"
-strum = "0.26"
+strum = "0.24"
-strum_macros = "0.26"
+strum_macros = "0.24"
 "subtle"  = "2.5.0"
-svg_fmt = "0.4.3"
+# Our PR https://github.com/nical/rust_debug/pull/4 has been merged but no new version released yet
+svg_fmt = { git = "https://github.com/nical/rust_debug", rev = "28a7d96eecff2f28e75b1ea09f2d499a60d0e3b4" }
 sync_wrapper = "0.1.2"
 tar = "0.4"
 task-local-extensions = "0.1.4"
@@ -177,8 +177,8 @@ tokio-rustls = "0.25"
 tokio-stream = "0.1"
 tokio-tar = "0.3"
 tokio-util = { version = "0.7.10", features = ["io", "rt"] }
-toml = "0.8"
+toml = "0.7"
-toml_edit = "0.22"
+toml_edit = "0.19"
 tonic = {version = "0.9", features = ["tls", "tls-roots"]}
 tower-service = "0.3.2"
 tracing = "0.1"
@@ -201,21 +201,10 @@ env_logger = "0.10"
 log = "0.4"
 
 ## Libraries from neondatabase/ git forks, ideally with changes to be upstreamed
-
+postgres = { git = "https://github.com/neondatabase/rust-postgres.git", branch="neon" }
-# We want to use the 'neon' branch for these, but there's currently one
+postgres-protocol = { git = "https://github.com/neondatabase/rust-postgres.git", branch="neon" }
-# incompatible change on the branch. See:
+postgres-types = { git = "https://github.com/neondatabase/rust-postgres.git", branch="neon" }
-#
+tokio-postgres = { git = "https://github.com/neondatabase/rust-postgres.git", branch="neon" }
-# - PR #8076 which contained changes that depended on the new changes in
-#   the rust-postgres crate, and
-# - PR #8654 which reverted those changes and made the code in proxy incompatible
-#   with the tip of the 'neon' branch again.
-#
-# When those proxy changes are re-applied (see PR #8747), we can switch using
-# the tip of the 'neon' branch again.
-postgres = { git = "https://github.com/neondatabase/rust-postgres.git", rev = "20031d7a9ee1addeae6e0968e3899ae6bf01cee2" }
-postgres-protocol = { git = "https://github.com/neondatabase/rust-postgres.git", rev = "20031d7a9ee1addeae6e0968e3899ae6bf01cee2" }
-postgres-types = { git = "https://github.com/neondatabase/rust-postgres.git", rev = "20031d7a9ee1addeae6e0968e3899ae6bf01cee2" }
-tokio-postgres = { git = "https://github.com/neondatabase/rust-postgres.git", rev = "20031d7a9ee1addeae6e0968e3899ae6bf01cee2" }
 
 ## Local libraries
 compute_api = { version = "0.1", path = "./libs/compute_api/" }
@@ -252,7 +241,11 @@ tonic-build = "0.9"
 [patch.crates-io]
 
 # Needed to get `tokio-postgres-rustls` to depend on our fork.
-tokio-postgres = { git = "https://github.com/neondatabase/rust-postgres.git", rev = "20031d7a9ee1addeae6e0968e3899ae6bf01cee2" }
+tokio-postgres = { git = "https://github.com/neondatabase/rust-postgres.git", branch="neon" }
+
+# bug fixes for UUID
+parquet = { git = "https://github.com/apache/arrow-rs", branch = "master" }
+parquet_derive = { git = "https://github.com/apache/arrow-rs", branch = "master" }
 
 ################# Binary contents sections
 

Dockerfile

@@ -87,7 +87,6 @@ RUN mkdir -p /data/.neon/ && \
        "pg_distrib_dir='/usr/local/'\n" \
        "listen_pg_addr='0.0.0.0:6400'\n" \
        "listen_http_addr='0.0.0.0:9898'\n" \
-       "availability_zone='local'\n" \
   > /data/.neon/pageserver.toml && \
   chown -R neon:neon /data/.neon
 

Dockerfile.build-tools

@@ -192,7 +192,7 @@ WORKDIR /home/nonroot
 
 # Rust
 # Please keep the version of llvm (installed above) in sync with rust llvm (`rustc --version --verbose | grep LLVM`)
-ENV RUSTC_VERSION=1.81.0
+ENV RUSTC_VERSION=1.80.1
 ENV RUSTUP_HOME="/home/nonroot/.rustup"
 ENV PATH="/home/nonroot/.cargo/bin:${PATH}"
 ARG RUSTFILT_VERSION=0.2.1
@@ -207,7 +207,7 @@ RUN curl -sSO https://static.rust-lang.org/rustup/dist/$(uname -m)-unknown-linux
     export PATH="$HOME/.cargo/bin:$PATH" && \
     . "$HOME/.cargo/env" && \
     cargo --version && rustup --version && \
-    rustup component add llvm-tools rustfmt clippy && \
+    rustup component add llvm-tools-preview rustfmt clippy && \
     cargo install rustfilt            --version ${RUSTFILT_VERSION} && \
     cargo install cargo-hakari        --version ${CARGO_HAKARI_VERSION} && \
     cargo install cargo-deny --locked --version ${CARGO_DENY_VERSION} && \

README.md

@@ -64,12 +64,6 @@ brew install protobuf openssl flex bison icu4c pkg-config
 echo 'export PATH="$(brew --prefix openssl)/bin:$PATH"' >> ~/.zshrc
 ```
 
-If you get errors about missing `m4` you may have to install it manually:
-```
-brew install m4
-brew link --force m4
-```
-
 2. [Install Rust](https://www.rust-lang.org/tools/install)
 ```
 # recommended approach from https://www.rust-lang.org/tools/install

compute_tools/src/pg_helpers.rs

@@ -22,10 +22,9 @@ use compute_api::spec::{Database, GenericOption, GenericOptions, PgIdent, Role};
 
 const POSTGRES_WAIT_TIMEOUT: Duration = Duration::from_millis(60 * 1000); // milliseconds
 
-/// Escape a string for including it in a SQL literal.
+/// Escape a string for including it in a SQL literal. Wrapping the result
-///
+/// with `E'{}'` or `'{}'` is not required, as it returns a ready-to-use
-/// Wrapping the result with `E'{}'` or `'{}'` is not required,
+/// SQL string literal, e.g. `'db'''` or `E'db\\'`.
-/// as it returns a ready-to-use SQL string literal, e.g. `'db'''` or `E'db\\'`.
 /// See <https://github.com/postgres/postgres/blob/da98d005cdbcd45af563d0c4ac86d0e9772cd15f/src/backend/utils/adt/quote.c#L47>
 /// for the original implementation.
 pub fn escape_literal(s: &str) -> String {

control_plane/src/bin/neon_local.rs

@@ -640,8 +640,6 @@ async fn handle_timeline(timeline_match: &ArgMatches, env: &mut local_env::Local
         }
         Some(("branch", branch_match)) => {
             let tenant_id = get_tenant_id(branch_match, env)?;
-            let new_timeline_id =
-                parse_timeline_id(branch_match)?.unwrap_or(TimelineId::generate());
             let new_branch_name = branch_match
                 .get_one::<String>("branch-name")
                 .ok_or_else(|| anyhow!("No branch name provided"))?;
@@ -660,6 +658,7 @@ async fn handle_timeline(timeline_match: &ArgMatches, env: &mut local_env::Local
                 .map(|lsn_str| Lsn::from_str(lsn_str))
                 .transpose()
                 .context("Failed to parse ancestor start Lsn from the request")?;
+            let new_timeline_id = TimelineId::generate();
             let storage_controller = StorageController::from_env(env);
             let create_req = TimelineCreateRequest {
                 new_timeline_id,
@@ -1571,6 +1570,7 @@ fn cli() -> Command {
                         .value_parser(value_parser!(PathBuf))
                         .value_name("config")
                 )
+                .arg(pg_version_arg.clone())
                 .arg(force_arg)
         )
         .subcommand(
@@ -1583,7 +1583,6 @@ fn cli() -> Command {
             .subcommand(Command::new("branch")
                 .about("Create a new timeline, using another timeline as a base, copying its data")
                 .arg(tenant_id_arg.clone())
-                .arg(timeline_id_arg.clone())
                 .arg(branch_name_arg.clone())
                 .arg(Arg::new("ancestor-branch-name").long("ancestor-branch-name")
                     .help("Use last Lsn of another timeline (and its data) as base when creating the new timeline. The timeline gets resolved by its branch name.").required(false))

control_plane/src/pageserver.rs

@@ -75,14 +75,14 @@ impl PageServerNode {
         }
     }
 
-    fn pageserver_make_identity_toml(&self, node_id: NodeId) -> toml_edit::DocumentMut {
+    fn pageserver_make_identity_toml(&self, node_id: NodeId) -> toml_edit::Document {
-        toml_edit::DocumentMut::from_str(&format!("id={node_id}")).unwrap()
+        toml_edit::Document::from_str(&format!("id={node_id}")).unwrap()
     }
 
     fn pageserver_init_make_toml(
         &self,
         conf: NeonLocalInitPageserverConf,
-    ) -> anyhow::Result<toml_edit::DocumentMut> {
+    ) -> anyhow::Result<toml_edit::Document> {
         assert_eq!(&PageServerConf::from(&conf), &self.conf, "during neon_local init, we derive the runtime state of ps conf (self.conf) from the --config flag fully");
 
         // TODO(christian): instead of what we do here, create a pageserver_api::config::ConfigToml (PR #7656)
@@ -137,9 +137,9 @@ impl PageServerNode {
 
         // Turn `overrides` into a toml document.
         // TODO: above code is legacy code, it should be refactored to use toml_edit directly.
-        let mut config_toml = toml_edit::DocumentMut::new();
+        let mut config_toml = toml_edit::Document::new();
         for fragment_str in overrides {
-            let fragment = toml_edit::DocumentMut::from_str(&fragment_str)
+            let fragment = toml_edit::Document::from_str(&fragment_str)
                 .expect("all fragments in `overrides` are valid toml documents, this function controls that");
             for (key, item) in fragment.iter() {
                 config_toml.insert(key, item.clone());

control_plane/storcon_cli/src/main.rs

@@ -4,8 +4,8 @@ use std::{str::FromStr, time::Duration};
 use clap::{Parser, Subcommand};
 use pageserver_api::{
     controller_api::{
-        NodeAvailabilityWrapper, NodeDescribeResponse, NodeShardResponse, ShardSchedulingPolicy,
+        NodeAvailabilityWrapper, NodeDescribeResponse, ShardSchedulingPolicy, TenantCreateRequest,
-        TenantCreateRequest, TenantDescribeResponse, TenantPolicyRequest,
+        TenantDescribeResponse, TenantPolicyRequest,
     },
     models::{
         EvictionPolicy, EvictionPolicyLayerAccessThreshold, LocationConfigSecondary,
@@ -80,10 +80,7 @@ enum Command {
     /// List nodes known to the storage controller
     Nodes {},
     /// List tenants known to the storage controller
-    Tenants {
+    Tenants {},
-        /// If this field is set, it will list the tenants on a specific node
-        node_id: Option<NodeId>,
-    },
     /// Create a new tenant in the storage controller, and by extension on pageservers.
     TenantCreate {
         #[arg(long)]
@@ -339,7 +336,7 @@ async fn main() -> anyhow::Result<()> {
                         listen_pg_port,
                         listen_http_addr,
                         listen_http_port,
-                        availability_zone_id,
+                        availability_zone_id: Some(availability_zone_id),
                     }),
                 )
                 .await?;
@@ -406,41 +403,7 @@ async fn main() -> anyhow::Result<()> {
                 )
                 .await?;
         }
-        Command::Tenants {
+        Command::Tenants {} => {
-            node_id: Some(node_id),
-        } => {
-            let describe_response = storcon_client
-                .dispatch::<(), NodeShardResponse>(
-                    Method::GET,
-                    format!("control/v1/node/{node_id}/shards"),
-                    None,
-                )
-                .await?;
-            let shards = describe_response.shards;
-            let mut table = comfy_table::Table::new();
-            table.set_header([
-                "Shard",
-                "Intended Primary/Secondary",
-                "Observed Primary/Secondary",
-            ]);
-            for shard in shards {
-                table.add_row([
-                    format!("{}", shard.tenant_shard_id),
-                    match shard.is_intended_secondary {
-                        None => "".to_string(),
-                        Some(true) => "Secondary".to_string(),
-                        Some(false) => "Primary".to_string(),
-                    },
-                    match shard.is_observed_secondary {
-                        None => "".to_string(),
-                        Some(true) => "Secondary".to_string(),
-                        Some(false) => "Primary".to_string(),
-                    },
-                ]);
-            }
-            println!("{table}");
-        }
-        Command::Tenants { node_id: None } => {
             let mut resp = storcon_client
                 .dispatch::<(), Vec<TenantDescribeResponse>>(
                     Method::GET,

libs/metrics/src/lib.rs

@@ -68,7 +68,6 @@ macro_rules! register_uint_gauge {
 static INTERNAL_REGISTRY: Lazy<Registry> = Lazy::new(Registry::new);
 
 /// Register a collector in the internal registry. MUST be called before the first call to `gather()`.
-///
 /// Otherwise, we can have a deadlock in the `gather()` call, trying to register a new collector
 /// while holding the lock.
 pub fn register_internal(c: Box<dyn Collector>) -> prometheus::Result<()> {

libs/pageserver_api/Cargo.toml

@@ -4,10 +4,6 @@ version = "0.1.0"
 edition.workspace = true
 license.workspace = true
 
-[features]
-# See pageserver/Cargo.toml
-testing = ["dep:nix"]
-
 [dependencies]
 serde.workspace = true
 serde_with.workspace = true
@@ -27,12 +23,6 @@ thiserror.workspace = true
 humantime-serde.workspace = true
 chrono = { workspace = true, features = ["serde"] }
 itertools.workspace = true
-storage_broker.workspace = true
-camino = {workspace = true, features = ["serde1"]}
-remote_storage.workspace = true
-postgres_backend.workspace = true
-nix = {workspace = true, optional = true}
-reqwest.workspace = true
 
 [dev-dependencies]
 bincode.workspace = true

libs/pageserver_api/src/config.rs

@@ -1,28 +1,15 @@
-use camino::Utf8PathBuf;
+use std::collections::HashMap;
+
+use const_format::formatcp;
 
 #[cfg(test)]
 mod tests;
 
-use const_format::formatcp;
 pub const DEFAULT_PG_LISTEN_PORT: u16 = 64000;
 pub const DEFAULT_PG_LISTEN_ADDR: &str = formatcp!("127.0.0.1:{DEFAULT_PG_LISTEN_PORT}");
 pub const DEFAULT_HTTP_LISTEN_PORT: u16 = 9898;
 pub const DEFAULT_HTTP_LISTEN_ADDR: &str = formatcp!("127.0.0.1:{DEFAULT_HTTP_LISTEN_PORT}");
 
-use postgres_backend::AuthType;
-use remote_storage::RemoteStorageConfig;
-use serde_with::serde_as;
-use std::{
-    collections::HashMap,
-    num::{NonZeroU64, NonZeroUsize},
-    str::FromStr,
-    time::Duration,
-};
-use utils::logging::LogFormat;
-
-use crate::models::ImageCompressionAlgorithm;
-use crate::models::LsnLease;
-
 // Certain metadata (e.g. externally-addressable name, AZ) is delivered
 // as a separate structure.  This information is not neeed by the pageserver
 // itself, it is only used for registering the pageserver with the control
@@ -42,476 +29,3 @@ pub struct NodeMetadata {
     #[serde(flatten)]
     pub other: HashMap<String, serde_json::Value>,
 }
-
-/// `pageserver.toml`
-///
-/// We use serde derive with `#[serde(default)]` to generate a deserializer
-/// that fills in the default values for each config field.
-///
-/// If there cannot be a static default value because we need to make runtime
-/// checks to determine the default, make it an `Option` (which defaults to None).
-/// The runtime check should be done in the consuming crate, i.e., `pageserver`.
-#[serde_as]
-#[derive(Clone, Debug, serde::Deserialize, serde::Serialize)]
-#[serde(default, deny_unknown_fields)]
-pub struct ConfigToml {
-    // types mapped 1:1 into the runtime PageServerConfig type
-    pub listen_pg_addr: String,
-    pub listen_http_addr: String,
-    pub availability_zone: Option<String>,
-    #[serde(with = "humantime_serde")]
-    pub wait_lsn_timeout: Duration,
-    #[serde(with = "humantime_serde")]
-    pub wal_redo_timeout: Duration,
-    pub superuser: String,
-    pub page_cache_size: usize,
-    pub max_file_descriptors: usize,
-    pub pg_distrib_dir: Option<Utf8PathBuf>,
-    #[serde_as(as = "serde_with::DisplayFromStr")]
-    pub http_auth_type: AuthType,
-    #[serde_as(as = "serde_with::DisplayFromStr")]
-    pub pg_auth_type: AuthType,
-    pub auth_validation_public_key_path: Option<Utf8PathBuf>,
-    pub remote_storage: Option<RemoteStorageConfig>,
-    pub tenant_config: TenantConfigToml,
-    #[serde_as(as = "serde_with::DisplayFromStr")]
-    pub broker_endpoint: storage_broker::Uri,
-    #[serde(with = "humantime_serde")]
-    pub broker_keepalive_interval: Duration,
-    #[serde_as(as = "serde_with::DisplayFromStr")]
-    pub log_format: LogFormat,
-    pub concurrent_tenant_warmup: NonZeroUsize,
-    pub concurrent_tenant_size_logical_size_queries: NonZeroUsize,
-    #[serde(with = "humantime_serde")]
-    pub metric_collection_interval: Duration,
-    pub metric_collection_endpoint: Option<reqwest::Url>,
-    pub metric_collection_bucket: Option<RemoteStorageConfig>,
-    #[serde(with = "humantime_serde")]
-    pub synthetic_size_calculation_interval: Duration,
-    pub disk_usage_based_eviction: Option<DiskUsageEvictionTaskConfig>,
-    pub test_remote_failures: u64,
-    pub ondemand_download_behavior_treat_error_as_warn: bool,
-    #[serde(with = "humantime_serde")]
-    pub background_task_maximum_delay: Duration,
-    pub control_plane_api: Option<reqwest::Url>,
-    pub control_plane_api_token: Option<String>,
-    pub control_plane_emergency_mode: bool,
-    pub heatmap_upload_concurrency: usize,
-    pub secondary_download_concurrency: usize,
-    pub virtual_file_io_engine: Option<crate::models::virtual_file::IoEngineKind>,
-    pub ingest_batch_size: u64,
-    pub max_vectored_read_bytes: MaxVectoredReadBytes,
-    pub image_compression: ImageCompressionAlgorithm,
-    pub ephemeral_bytes_per_memory_kb: usize,
-    pub l0_flush: Option<crate::models::L0FlushConfig>,
-    #[serde(skip_serializing)]
-    // TODO(https://github.com/neondatabase/neon/issues/8184): remove after this field is removed from all pageserver.toml's
-    pub compact_level0_phase1_value_access: serde::de::IgnoredAny,
-    pub virtual_file_direct_io: crate::models::virtual_file::DirectIoMode,
-    pub io_buffer_alignment: usize,
-}
-
-#[derive(Debug, Clone, PartialEq, Eq, serde::Serialize, serde::Deserialize)]
-#[serde(deny_unknown_fields)]
-pub struct DiskUsageEvictionTaskConfig {
-    pub max_usage_pct: utils::serde_percent::Percent,
-    pub min_avail_bytes: u64,
-    #[serde(with = "humantime_serde")]
-    pub period: Duration,
-    #[cfg(feature = "testing")]
-    pub mock_statvfs: Option<statvfs::mock::Behavior>,
-    /// Select sorting for evicted layers
-    #[serde(default)]
-    pub eviction_order: EvictionOrder,
-}
-
-pub mod statvfs {
-    pub mod mock {
-        #[derive(Debug, Clone, PartialEq, Eq, serde::Serialize, serde::Deserialize)]
-        #[serde(tag = "type")]
-        pub enum Behavior {
-            Success {
-                blocksize: u64,
-                total_blocks: u64,
-                name_filter: Option<utils::serde_regex::Regex>,
-            },
-            #[cfg(feature = "testing")]
-            Failure { mocked_error: MockedError },
-        }
-
-        #[cfg(feature = "testing")]
-        #[derive(Debug, Clone, Copy, PartialEq, Eq, serde::Serialize, serde::Deserialize)]
-        #[allow(clippy::upper_case_acronyms)]
-        pub enum MockedError {
-            EIO,
-        }
-
-        #[cfg(feature = "testing")]
-        impl From<MockedError> for nix::Error {
-            fn from(e: MockedError) -> Self {
-                match e {
-                    MockedError::EIO => nix::Error::EIO,
-                }
-            }
-        }
-    }
-}
-
-#[derive(Debug, Clone, Copy, PartialEq, Eq, serde::Serialize, serde::Deserialize)]
-#[serde(tag = "type", content = "args")]
-pub enum EvictionOrder {
-    RelativeAccessed {
-        highest_layer_count_loses_first: bool,
-    },
-}
-
-impl Default for EvictionOrder {
-    fn default() -> Self {
-        Self::RelativeAccessed {
-            highest_layer_count_loses_first: true,
-        }
-    }
-}
-
-#[derive(
-    Eq,
-    PartialEq,
-    Debug,
-    Copy,
-    Clone,
-    strum_macros::EnumString,
-    strum_macros::Display,
-    serde_with::DeserializeFromStr,
-    serde_with::SerializeDisplay,
-)]
-#[strum(serialize_all = "kebab-case")]
-pub enum GetVectoredImpl {
-    Sequential,
-    Vectored,
-}
-
-#[derive(
-    Eq,
-    PartialEq,
-    Debug,
-    Copy,
-    Clone,
-    strum_macros::EnumString,
-    strum_macros::Display,
-    serde_with::DeserializeFromStr,
-    serde_with::SerializeDisplay,
-)]
-#[strum(serialize_all = "kebab-case")]
-pub enum GetImpl {
-    Legacy,
-    Vectored,
-}
-
-#[derive(Copy, Clone, Debug, PartialEq, Eq, serde::Serialize, serde::Deserialize)]
-#[serde(transparent)]
-pub struct MaxVectoredReadBytes(pub NonZeroUsize);
-
-/// A tenant's calcuated configuration, which is the result of merging a
-/// tenant's TenantConfOpt with the global TenantConf from PageServerConf.
-///
-/// For storing and transmitting individual tenant's configuration, see
-/// TenantConfOpt.
-#[derive(Debug, Clone, PartialEq, Eq, serde::Serialize, serde::Deserialize)]
-#[serde(deny_unknown_fields, default)]
-pub struct TenantConfigToml {
-    // Flush out an inmemory layer, if it's holding WAL older than this
-    // This puts a backstop on how much WAL needs to be re-digested if the
-    // page server crashes.
-    // This parameter actually determines L0 layer file size.
-    pub checkpoint_distance: u64,
-    // Inmemory layer is also flushed at least once in checkpoint_timeout to
-    // eventually upload WAL after activity is stopped.
-    #[serde(with = "humantime_serde")]
-    pub checkpoint_timeout: Duration,
-    // Target file size, when creating image and delta layers.
-    // This parameter determines L1 layer file size.
-    pub compaction_target_size: u64,
-    // How often to check if there's compaction work to be done.
-    // Duration::ZERO means automatic compaction is disabled.
-    #[serde(with = "humantime_serde")]
-    pub compaction_period: Duration,
-    // Level0 delta layer threshold for compaction.
-    pub compaction_threshold: usize,
-    pub compaction_algorithm: crate::models::CompactionAlgorithmSettings,
-    // 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.
-    // Page versions older than this are garbage collected away.
-    pub gc_horizon: u64,
-    // Interval at which garbage collection is triggered.
-    // Duration::ZERO means automatic GC is disabled
-    #[serde(with = "humantime_serde")]
-    pub gc_period: Duration,
-    // Delta layer churn threshold to create L1 image layers.
-    pub image_creation_threshold: usize,
-    // Determines how much history is retained, to allow
-    // branching and read replicas at an older point in time.
-    // The unit is time.
-    // Page versions older than this are garbage collected away.
-    #[serde(with = "humantime_serde")]
-    pub pitr_interval: Duration,
-    /// Maximum amount of time to wait while opening a connection to receive wal, before erroring.
-    #[serde(with = "humantime_serde")]
-    pub walreceiver_connect_timeout: Duration,
-    /// Considers safekeepers stalled after no WAL updates were received longer than this threshold.
-    /// A stalled safekeeper will be changed to a newer one when it appears.
-    #[serde(with = "humantime_serde")]
-    pub lagging_wal_timeout: Duration,
-    /// Considers safekeepers lagging when their WAL is behind another safekeeper for more than this threshold.
-    /// A lagging safekeeper will be changed after `lagging_wal_timeout` time elapses since the last WAL update,
-    /// to avoid eager reconnects.
-    pub max_lsn_wal_lag: NonZeroU64,
-    pub eviction_policy: crate::models::EvictionPolicy,
-    pub min_resident_size_override: Option<u64>,
-    // See the corresponding metric's help string.
-    #[serde(with = "humantime_serde")]
-    pub evictions_low_residence_duration_metric_threshold: Duration,
-
-    /// If non-zero, the period between uploads of a heatmap from attached tenants.  This
-    /// may be disabled if a Tenant will not have secondary locations: only secondary
-    /// locations will use the heatmap uploaded by attached locations.
-    #[serde(with = "humantime_serde")]
-    pub heatmap_period: Duration,
-
-    /// If true then SLRU segments are dowloaded on demand, if false SLRU segments are included in basebackup
-    pub lazy_slru_download: bool,
-
-    pub timeline_get_throttle: crate::models::ThrottleConfig,
-
-    // How much WAL must be ingested before checking again whether a new image layer is required.
-    // Expresed in multiples of checkpoint distance.
-    pub image_layer_creation_check_threshold: u8,
-
-    /// Switch to a new aux file policy. Switching this flag requires the user has not written any aux file into
-    /// the storage before, and this flag cannot be switched back. Otherwise there will be data corruptions.
-    /// There is a `last_aux_file_policy` flag which gets persisted in `index_part.json` once the first aux
-    /// file is written.
-    pub switch_aux_file_policy: crate::models::AuxFilePolicy,
-
-    /// The length for an explicit LSN lease request.
-    /// Layers needed to reconstruct pages at LSN will not be GC-ed during this interval.
-    #[serde(with = "humantime_serde")]
-    pub lsn_lease_length: Duration,
-
-    /// The length for an implicit LSN lease granted as part of `get_lsn_by_timestamp` request.
-    /// Layers needed to reconstruct pages at LSN will not be GC-ed during this interval.
-    #[serde(with = "humantime_serde")]
-    pub lsn_lease_length_for_ts: Duration,
-}
-
-pub mod defaults {
-    use crate::models::ImageCompressionAlgorithm;
-
-    pub use storage_broker::DEFAULT_ENDPOINT as BROKER_DEFAULT_ENDPOINT;
-
-    pub const DEFAULT_WAIT_LSN_TIMEOUT: &str = "300 s";
-    pub const DEFAULT_WAL_REDO_TIMEOUT: &str = "60 s";
-
-    pub const DEFAULT_SUPERUSER: &str = "cloud_admin";
-
-    pub const DEFAULT_PAGE_CACHE_SIZE: usize = 8192;
-    pub const DEFAULT_MAX_FILE_DESCRIPTORS: usize = 100;
-
-    pub const DEFAULT_LOG_FORMAT: &str = "plain";
-
-    pub const DEFAULT_CONCURRENT_TENANT_WARMUP: usize = 8;
-
-    pub const DEFAULT_CONCURRENT_TENANT_SIZE_LOGICAL_SIZE_QUERIES: usize = 1;
-
-    pub const DEFAULT_METRIC_COLLECTION_INTERVAL: &str = "10 min";
-    pub const DEFAULT_METRIC_COLLECTION_ENDPOINT: Option<reqwest::Url> = None;
-    pub const DEFAULT_SYNTHETIC_SIZE_CALCULATION_INTERVAL: &str = "10 min";
-    pub const DEFAULT_BACKGROUND_TASK_MAXIMUM_DELAY: &str = "10s";
-
-    pub const DEFAULT_HEATMAP_UPLOAD_CONCURRENCY: usize = 8;
-    pub const DEFAULT_SECONDARY_DOWNLOAD_CONCURRENCY: usize = 1;
-
-    pub const DEFAULT_INGEST_BATCH_SIZE: u64 = 100;
-
-    pub const DEFAULT_MAX_VECTORED_READ_BYTES: usize = 128 * 1024; // 128 KiB
-
-    pub const DEFAULT_IMAGE_COMPRESSION: ImageCompressionAlgorithm =
-        ImageCompressionAlgorithm::Zstd { level: Some(1) };
-
-    pub const DEFAULT_VALIDATE_VECTORED_GET: bool = false;
-
-    pub const DEFAULT_EPHEMERAL_BYTES_PER_MEMORY_KB: usize = 0;
-
-    pub const DEFAULT_IO_BUFFER_ALIGNMENT: usize = 512;
-}
-
-impl Default for ConfigToml {
-    fn default() -> Self {
-        use defaults::*;
-
-        Self {
-            listen_pg_addr: (DEFAULT_PG_LISTEN_ADDR.to_string()),
-            listen_http_addr: (DEFAULT_HTTP_LISTEN_ADDR.to_string()),
-            availability_zone: (None),
-            wait_lsn_timeout: (humantime::parse_duration(DEFAULT_WAIT_LSN_TIMEOUT)
-                .expect("cannot parse default wait lsn timeout")),
-            wal_redo_timeout: (humantime::parse_duration(DEFAULT_WAL_REDO_TIMEOUT)
-                .expect("cannot parse default wal redo timeout")),
-            superuser: (DEFAULT_SUPERUSER.to_string()),
-            page_cache_size: (DEFAULT_PAGE_CACHE_SIZE),
-            max_file_descriptors: (DEFAULT_MAX_FILE_DESCRIPTORS),
-            pg_distrib_dir: None, // Utf8PathBuf::from("./pg_install"), // TODO: formely, this was std::env::current_dir()
-            http_auth_type: (AuthType::Trust),
-            pg_auth_type: (AuthType::Trust),
-            auth_validation_public_key_path: (None),
-            remote_storage: None,
-            broker_endpoint: (storage_broker::DEFAULT_ENDPOINT
-                .parse()
-                .expect("failed to parse default broker endpoint")),
-            broker_keepalive_interval: (humantime::parse_duration(
-                storage_broker::DEFAULT_KEEPALIVE_INTERVAL,
-            )
-            .expect("cannot parse default keepalive interval")),
-            log_format: (LogFormat::from_str(DEFAULT_LOG_FORMAT).unwrap()),
-
-            concurrent_tenant_warmup: (NonZeroUsize::new(DEFAULT_CONCURRENT_TENANT_WARMUP)
-                .expect("Invalid default constant")),
-            concurrent_tenant_size_logical_size_queries: NonZeroUsize::new(1).unwrap(),
-            metric_collection_interval: (humantime::parse_duration(
-                DEFAULT_METRIC_COLLECTION_INTERVAL,
-            )
-            .expect("cannot parse default metric collection interval")),
-            synthetic_size_calculation_interval: (humantime::parse_duration(
-                DEFAULT_SYNTHETIC_SIZE_CALCULATION_INTERVAL,
-            )
-            .expect("cannot parse default synthetic size calculation interval")),
-            metric_collection_endpoint: (DEFAULT_METRIC_COLLECTION_ENDPOINT),
-
-            metric_collection_bucket: (None),
-
-            disk_usage_based_eviction: (None),
-
-            test_remote_failures: (0),
-
-            ondemand_download_behavior_treat_error_as_warn: (false),
-
-            background_task_maximum_delay: (humantime::parse_duration(
-                DEFAULT_BACKGROUND_TASK_MAXIMUM_DELAY,
-            )
-            .unwrap()),
-
-            control_plane_api: (None),
-            control_plane_api_token: (None),
-            control_plane_emergency_mode: (false),
-
-            heatmap_upload_concurrency: (DEFAULT_HEATMAP_UPLOAD_CONCURRENCY),
-            secondary_download_concurrency: (DEFAULT_SECONDARY_DOWNLOAD_CONCURRENCY),
-
-            ingest_batch_size: (DEFAULT_INGEST_BATCH_SIZE),
-
-            virtual_file_io_engine: None,
-
-            max_vectored_read_bytes: (MaxVectoredReadBytes(
-                NonZeroUsize::new(DEFAULT_MAX_VECTORED_READ_BYTES).unwrap(),
-            )),
-            image_compression: (DEFAULT_IMAGE_COMPRESSION),
-            ephemeral_bytes_per_memory_kb: (DEFAULT_EPHEMERAL_BYTES_PER_MEMORY_KB),
-            l0_flush: None,
-            compact_level0_phase1_value_access: Default::default(),
-            virtual_file_direct_io: crate::models::virtual_file::DirectIoMode::default(),
-
-            io_buffer_alignment: DEFAULT_IO_BUFFER_ALIGNMENT,
-
-            tenant_config: TenantConfigToml::default(),
-        }
-    }
-}
-
-pub mod tenant_conf_defaults {
-
-    // FIXME: This current value is very low. I would imagine something like 1 GB or 10 GB
-    // would be more appropriate. But a low value forces the code to be exercised more,
-    // which is good for now to trigger bugs.
-    // This parameter actually determines L0 layer file size.
-    pub const DEFAULT_CHECKPOINT_DISTANCE: u64 = 256 * 1024 * 1024;
-    pub const DEFAULT_CHECKPOINT_TIMEOUT: &str = "10 m";
-
-    // FIXME the below configs are only used by legacy algorithm. The new algorithm
-    // has different parameters.
-
-    // Target file size, when creating image and delta layers.
-    // This parameter determines L1 layer file size.
-    pub const DEFAULT_COMPACTION_TARGET_SIZE: u64 = 128 * 1024 * 1024;
-
-    pub const DEFAULT_COMPACTION_PERIOD: &str = "20 s";
-    pub const DEFAULT_COMPACTION_THRESHOLD: usize = 10;
-    pub const DEFAULT_COMPACTION_ALGORITHM: crate::models::CompactionAlgorithm =
-        crate::models::CompactionAlgorithm::Legacy;
-
-    pub const DEFAULT_GC_HORIZON: u64 = 64 * 1024 * 1024;
-
-    // Large DEFAULT_GC_PERIOD is fine as long as PITR_INTERVAL is larger.
-    // If there's a need to decrease this value, first make sure that GC
-    // doesn't hold a layer map write lock for non-trivial operations.
-    // Relevant: https://github.com/neondatabase/neon/issues/3394
-    pub const DEFAULT_GC_PERIOD: &str = "1 hr";
-    pub const DEFAULT_IMAGE_CREATION_THRESHOLD: usize = 3;
-    pub const DEFAULT_PITR_INTERVAL: &str = "7 days";
-    pub const DEFAULT_WALRECEIVER_CONNECT_TIMEOUT: &str = "10 seconds";
-    pub const DEFAULT_WALRECEIVER_LAGGING_WAL_TIMEOUT: &str = "10 seconds";
-    // The default limit on WAL lag should be set to avoid causing disconnects under high throughput
-    // scenarios: since the broker stats are updated ~1/s, a value of 1GiB should be sufficient for
-    // throughputs up to 1GiB/s per timeline.
-    pub const DEFAULT_MAX_WALRECEIVER_LSN_WAL_LAG: u64 = 1024 * 1024 * 1024;
-    pub const DEFAULT_EVICTIONS_LOW_RESIDENCE_DURATION_METRIC_THRESHOLD: &str = "24 hour";
-    // By default ingest enough WAL for two new L0 layers before checking if new image
-    // image layers should be created.
-    pub const DEFAULT_IMAGE_LAYER_CREATION_CHECK_THRESHOLD: u8 = 2;
-
-    pub const DEFAULT_INGEST_BATCH_SIZE: u64 = 100;
-}
-
-impl Default for TenantConfigToml {
-    fn default() -> Self {
-        use tenant_conf_defaults::*;
-        Self {
-            checkpoint_distance: DEFAULT_CHECKPOINT_DISTANCE,
-            checkpoint_timeout: humantime::parse_duration(DEFAULT_CHECKPOINT_TIMEOUT)
-                .expect("cannot parse default checkpoint timeout"),
-            compaction_target_size: DEFAULT_COMPACTION_TARGET_SIZE,
-            compaction_period: humantime::parse_duration(DEFAULT_COMPACTION_PERIOD)
-                .expect("cannot parse default compaction period"),
-            compaction_threshold: DEFAULT_COMPACTION_THRESHOLD,
-            compaction_algorithm: crate::models::CompactionAlgorithmSettings {
-                kind: DEFAULT_COMPACTION_ALGORITHM,
-            },
-            gc_horizon: DEFAULT_GC_HORIZON,
-            gc_period: humantime::parse_duration(DEFAULT_GC_PERIOD)
-                .expect("cannot parse default gc period"),
-            image_creation_threshold: DEFAULT_IMAGE_CREATION_THRESHOLD,
-            pitr_interval: humantime::parse_duration(DEFAULT_PITR_INTERVAL)
-                .expect("cannot parse default PITR interval"),
-            walreceiver_connect_timeout: humantime::parse_duration(
-                DEFAULT_WALRECEIVER_CONNECT_TIMEOUT,
-            )
-            .expect("cannot parse default walreceiver connect timeout"),
-            lagging_wal_timeout: humantime::parse_duration(DEFAULT_WALRECEIVER_LAGGING_WAL_TIMEOUT)
-                .expect("cannot parse default walreceiver lagging wal timeout"),
-            max_lsn_wal_lag: NonZeroU64::new(DEFAULT_MAX_WALRECEIVER_LSN_WAL_LAG)
-                .expect("cannot parse default max walreceiver Lsn wal lag"),
-            eviction_policy: crate::models::EvictionPolicy::NoEviction,
-            min_resident_size_override: None,
-            evictions_low_residence_duration_metric_threshold: humantime::parse_duration(
-                DEFAULT_EVICTIONS_LOW_RESIDENCE_DURATION_METRIC_THRESHOLD,
-            )
-            .expect("cannot parse default evictions_low_residence_duration_metric_threshold"),
-            heatmap_period: Duration::ZERO,
-            lazy_slru_download: false,
-            timeline_get_throttle: crate::models::ThrottleConfig::disabled(),
-            image_layer_creation_check_threshold: DEFAULT_IMAGE_LAYER_CREATION_CHECK_THRESHOLD,
-            switch_aux_file_policy: crate::models::AuxFilePolicy::default_tenant_config(),
-            lsn_lease_length: LsnLease::DEFAULT_LENGTH,
-            lsn_lease_length_for_ts: LsnLease::DEFAULT_LENGTH_FOR_TS,
-        }
-    }
-}

libs/pageserver_api/src/controller_api.rs

@@ -1,4 +1,4 @@
-use std::collections::{HashMap, HashSet};
+use std::collections::HashSet;
 use std::str::FromStr;
 use std::time::{Duration, Instant};
 
@@ -57,7 +57,7 @@ pub struct NodeRegisterRequest {
     pub listen_http_addr: String,
     pub listen_http_port: u16,
 
-    pub availability_zone_id: String,
+    pub availability_zone_id: Option<String>,
 }
 
 #[derive(Serialize, Deserialize)]
@@ -74,17 +74,6 @@ pub struct TenantPolicyRequest {
     pub scheduling: Option<ShardSchedulingPolicy>,
 }
 
-#[derive(Serialize, Deserialize)]
-pub struct ShardsPreferredAzsRequest {
-    #[serde(flatten)]
-    pub preferred_az_ids: HashMap<TenantShardId, String>,
-}
-
-#[derive(Serialize, Deserialize)]
-pub struct ShardsPreferredAzsResponse {
-    pub updated: Vec<TenantShardId>,
-}
-
 #[derive(Serialize, Deserialize, Debug)]
 pub struct TenantLocateResponseShard {
     pub shard_id: TenantShardId,
@@ -112,21 +101,6 @@ pub struct TenantDescribeResponse {
     pub config: TenantConfig,
 }
 
-#[derive(Serialize, Deserialize, Debug)]
-pub struct NodeShardResponse {
-    pub node_id: NodeId,
-    pub shards: Vec<NodeShard>,
-}
-
-#[derive(Serialize, Deserialize, Debug)]
-pub struct NodeShard {
-    pub tenant_shard_id: TenantShardId,
-    /// Whether the shard is observed secondary on a specific node. True = yes, False = no, None = not on this node.
-    pub is_observed_secondary: Option<bool>,
-    /// Whether the shard is intended to be a secondary on a specific node. True = yes, False = no, None = not on this node.
-    pub is_intended_secondary: Option<bool>,
-}
-
 #[derive(Serialize, Deserialize)]
 pub struct NodeDescribeResponse {
     pub id: NodeId,
@@ -158,12 +132,8 @@ pub struct TenantDescribeResponseShard {
     pub is_splitting: bool,
 
     pub scheduling_policy: ShardSchedulingPolicy,
-
-    pub preferred_az_id: Option<String>,
 }
 
-/// Migration request for a given tenant shard to a given node.
-///
 /// Explicitly migrating a particular shard is a low level operation
 /// TODO: higher level "Reschedule tenant" operation where the request
 /// specifies some constraints, e.g. asking it to get off particular node(s)

libs/pageserver_api/src/key.rs

@@ -263,6 +263,15 @@ impl Key {
         field5: u8::MAX,
         field6: u32::MAX,
     };
+    /// A key slightly smaller than [`Key::MAX`] for use in layer key ranges to avoid them to be confused with L0 layers
+    pub const NON_L0_MAX: Key = Key {
+        field1: u8::MAX,
+        field2: u32::MAX,
+        field3: u32::MAX,
+        field4: u32::MAX,
+        field5: u8::MAX,
+        field6: u32::MAX - 1,
+    };
 
     pub fn from_hex(s: &str) -> Result<Self> {
         if s.len() != 36 {

libs/pageserver_api/src/models.rs

@@ -6,7 +6,6 @@ pub use utilization::PageserverUtilization;
 
 use std::{
     collections::HashMap,
-    fmt::Display,
     io::{BufRead, Read},
     num::{NonZeroU32, NonZeroU64, NonZeroUsize},
     str::FromStr,
@@ -62,7 +61,7 @@ use bytes::{Buf, BufMut, Bytes, BytesMut};
     serde::Serialize,
     serde::Deserialize,
     strum_macros::Display,
-    strum_macros::VariantNames,
+    strum_macros::EnumVariantNames,
     strum_macros::AsRefStr,
     strum_macros::IntoStaticStr,
 )]
@@ -305,10 +304,8 @@ pub struct TenantConfig {
     pub lsn_lease_length_for_ts: Option<String>,
 }
 
-/// The policy for the aux file storage.
+/// The policy for the aux file storage. It can be switched through `switch_aux_file_policy`
-///
+/// tenant config. When the first aux file written, the policy will be persisted in the
-/// It can be switched through `switch_aux_file_policy` tenant config.
-/// When the first aux file written, the policy will be persisted in the
 /// `index_part.json` file and has a limited migration path.
 ///
 /// Currently, we only allow the following migration path:
@@ -438,9 +435,7 @@ pub enum CompactionAlgorithm {
     Tiered,
 }
 
-#[derive(
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize)]
-    Debug, Clone, Copy, PartialEq, Eq, serde_with::DeserializeFromStr, serde_with::SerializeDisplay,
-)]
 pub enum ImageCompressionAlgorithm {
     // Disabled for writes, support decompressing during read path
     Disabled,
@@ -475,33 +470,11 @@ impl FromStr for ImageCompressionAlgorithm {
     }
 }
 
-impl Display for ImageCompressionAlgorithm {
-    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
-        match self {
-            ImageCompressionAlgorithm::Disabled => write!(f, "disabled"),
-            ImageCompressionAlgorithm::Zstd { level } => {
-                if let Some(level) = level {
-                    write!(f, "zstd({})", level)
-                } else {
-                    write!(f, "zstd")
-                }
-            }
-        }
-    }
-}
-
 #[derive(Eq, PartialEq, Debug, Clone, Serialize, Deserialize)]
 pub struct CompactionAlgorithmSettings {
     pub kind: CompactionAlgorithm,
 }
 
-#[derive(Debug, PartialEq, Eq, Clone, serde::Deserialize, serde::Serialize)]
-#[serde(tag = "mode", rename_all = "kebab-case", deny_unknown_fields)]
-pub enum L0FlushConfig {
-    #[serde(rename_all = "snake_case")]
-    Direct { max_concurrency: NonZeroUsize },
-}
-
 #[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize)]
 pub struct EvictionPolicyLayerAccessThreshold {
     #[serde(with = "humantime_serde")]
@@ -898,9 +871,7 @@ pub struct WalRedoManagerStatus {
     pub process: Option<WalRedoManagerProcessStatus>,
 }
 
-/// The progress of a secondary tenant.
+/// The progress of a secondary tenant is mostly useful when doing a long running download: e.g. initiating
-///
-/// It is mostly useful when doing a long running download: e.g. initiating
 /// a download job, timing out while waiting for it to run, and then inspecting this status to understand
 /// what's happening.
 #[derive(Default, Debug, Serialize, Deserialize, Clone)]
@@ -1685,33 +1656,21 @@ mod tests {
     #[test]
     fn test_image_compression_algorithm_parsing() {
         use ImageCompressionAlgorithm::*;
-        let cases = [
+        assert_eq!(
-            ("disabled", Disabled),
+            ImageCompressionAlgorithm::from_str("disabled").unwrap(),
-            ("zstd", Zstd { level: None }),
+            Disabled
-            ("zstd(18)", Zstd { level: Some(18) }),
+        );
-            ("zstd(-3)", Zstd { level: Some(-3) }),
+        assert_eq!(
-        ];
+            ImageCompressionAlgorithm::from_str("zstd").unwrap(),
-
+            Zstd { level: None }
-        for (display, expected) in cases {
+        );
-            assert_eq!(
+        assert_eq!(
-                ImageCompressionAlgorithm::from_str(display).unwrap(),
+            ImageCompressionAlgorithm::from_str("zstd(18)").unwrap(),
-                expected,
+            Zstd { level: Some(18) }
-                "parsing works"
+        );
-            );
+        assert_eq!(
-            assert_eq!(format!("{expected}"), display, "Display FromStr roundtrip");
+            ImageCompressionAlgorithm::from_str("zstd(-3)").unwrap(),
-
+            Zstd { level: Some(-3) }
-            let ser = serde_json::to_string(&expected).expect("serialization");
+        );
-            assert_eq!(
-                serde_json::from_str::<ImageCompressionAlgorithm>(&ser).unwrap(),
-                expected,
-                "serde roundtrip"
-            );
-
-            assert_eq!(
-                serde_json::Value::String(display.to_string()),
-                serde_json::to_value(expected).unwrap(),
-                "Display is the serde serialization"
-            );
-        }
     }
 }

libs/pageserver_api/src/models/utilization.rs

@@ -89,19 +89,8 @@ impl PageserverUtilization {
 
     /// If a node is currently hosting more work than it can comfortably handle.  This does not indicate that
     /// it will fail, but it is a strong signal that more work should not be added unless there is no alternative.
-    ///
-    /// When a node is overloaded, we may override soft affinity preferences and do things like scheduling
-    /// into a node in a less desirable AZ, if all the nodes in the preferred AZ are overloaded.
     pub fn is_overloaded(score: RawScore) -> bool {
-        // Why the factor of two?  This is unscientific but reflects behavior of real systems:
+        score >= Self::UTILIZATION_FULL
-        // - In terms of shard counts, a node's preferred max count is a soft limit intended to keep
-        //   startup and housekeeping jobs nice and responsive.  We can go to double this limit if needed
-        //   until some more nodes are deployed.
-        // - In terms of disk space, the node's utilization heuristic assumes every tenant needs to
-        //   hold its biggest timeline fully on disk, which is tends to be an over estimate when
-        //   some tenants are very idle and have dropped layers from disk.  In practice going up to
-        //   double is generally better than giving up and scheduling in a sub-optimal AZ.
-        score >= 2 * Self::UTILIZATION_FULL
     }
 
     pub fn adjust_shard_count_max(&mut self, shard_count: u32) {

libs/postgres_backend/src/lib.rs

@@ -69,10 +69,8 @@ impl QueryError {
 }
 
 /// Returns true if the given error is a normal consequence of a network issue,
-/// or the client closing the connection.
+/// or the client closing the connection. These errors can happen during normal
-///
+/// operations, and don't indicate a bug in our code.
-/// These errors can happen during normal operations,
-/// and don't indicate a bug in our code.
 pub fn is_expected_io_error(e: &io::Error) -> bool {
     use io::ErrorKind::*;
     matches!(
@@ -81,16 +79,17 @@ pub fn is_expected_io_error(e: &io::Error) -> bool {
     )
 }
 
+#[async_trait::async_trait]
 pub trait Handler<IO> {
     /// Handle single query.
     /// postgres_backend will issue ReadyForQuery after calling this (this
     /// might be not what we want after CopyData streaming, but currently we don't
     /// care). It will also flush out the output buffer.
-    fn process_query(
+    async fn process_query(
         &mut self,
         pgb: &mut PostgresBackend<IO>,
         query_string: &str,
-    ) -> impl Future<Output = Result<(), QueryError>>;
+    ) -> Result<(), QueryError>;
 
     /// Called on startup packet receival, allows to process params.
     ///

libs/postgres_backend/tests/simple_select.rs

@@ -23,6 +23,7 @@ async fn make_tcp_pair() -> (TcpStream, TcpStream) {
 
 struct TestHandler {}
 
+#[async_trait::async_trait]
 impl<IO: AsyncRead + AsyncWrite + Unpin + Send> Handler<IO> for TestHandler {
     // return single col 'hey' for any query
     async fn process_query(

libs/postgres_connection/src/lib.rs

@@ -7,7 +7,6 @@ use std::fmt;
 use url::Host;
 
 /// Parses a string of format either `host:port` or `host` into a corresponding pair.
-///
 /// The `host` part should be a correct `url::Host`, while `port` (if present) should be
 /// a valid decimal u16 of digits only.
 pub fn parse_host_port<S: AsRef<str>>(host_port: S) -> Result<(Host, Option<u16>), anyhow::Error> {

libs/postgres_ffi/build.rs

@@ -14,7 +14,7 @@ impl ParseCallbacks for PostgresFfiCallbacks {
     fn include_file(&self, filename: &str) {
         // This does the equivalent of passing bindgen::CargoCallbacks
         // to the builder .parse_callbacks() method.
-        let cargo_callbacks = bindgen::CargoCallbacks::new();
+        let cargo_callbacks = bindgen::CargoCallbacks;
         cargo_callbacks.include_file(filename)
     }
 
@@ -121,7 +121,6 @@ fn main() -> anyhow::Result<()> {
             .allowlist_type("XLogPageHeaderData")
             .allowlist_type("XLogLongPageHeaderData")
             .allowlist_var("XLOG_PAGE_MAGIC")
-            .allowlist_var("PG_MAJORVERSION_NUM")
             .allowlist_var("PG_CONTROL_FILE_SIZE")
             .allowlist_var("PG_CONTROLFILEDATA_OFFSETOF_CRC")
             .allowlist_type("PageHeaderData")

libs/postgres_ffi/src/lib.rs

@@ -44,9 +44,6 @@ macro_rules! postgres_ffi {
             // Re-export some symbols from bindings
             pub use bindings::DBState_DB_SHUTDOWNED;
             pub use bindings::{CheckPoint, ControlFileData, XLogRecord};
-
-            pub const ZERO_CHECKPOINT: bytes::Bytes =
-                bytes::Bytes::from_static(&[0u8; xlog_utils::SIZEOF_CHECKPOINT]);
         }
     };
 }
@@ -109,107 +106,6 @@ macro_rules! dispatch_pgversion {
     };
 }
 
-#[macro_export]
-macro_rules! enum_pgversion_dispatch {
-    ($name:expr, $typ:ident, $bind:ident, $code:block) => {
-        enum_pgversion_dispatch!(
-            name = $name,
-            bind = $bind,
-            typ = $typ,
-            code = $code,
-            pgversions = [
-                V14 : v14,
-                V15 : v15,
-                V16 : v16,
-            ]
-        )
-    };
-    (name = $name:expr,
-     bind = $bind:ident,
-     typ = $typ:ident,
-     code = $code:block,
-     pgversions = [$($variant:ident : $md:ident),+ $(,)?]) => {
-        match $name {
-            $(
-            self::$typ::$variant($bind) => {
-                use $crate::$md as pgv;
-                $code
-            }
-            ),+,
-        }
-    };
-}
-
-#[macro_export]
-macro_rules! enum_pgversion {
-    {$name:ident, pgv :: $t:ident} => {
-        enum_pgversion!{
-            name = $name,
-            typ = $t,
-            pgversions = [
-                V14 : v14,
-                V15 : v15,
-                V16 : v16,
-            ]
-        }
-    };
-    {$name:ident, pgv :: $p:ident :: $t:ident} => {
-        enum_pgversion!{
-            name = $name,
-            path = $p,
-            typ = $t,
-            pgversions = [
-                V14 : v14,
-                V15 : v15,
-                V16 : v16,
-            ]
-        }
-    };
-    {name = $name:ident,
-     typ = $t:ident,
-     pgversions = [$($variant:ident : $md:ident),+ $(,)?]} => {
-        pub enum $name {
-            $($variant ( $crate::$md::$t )),+
-        }
-        impl self::$name {
-            pub fn pg_version(&self) -> u32 {
-                enum_pgversion_dispatch!(self, $name, _ign, {
-                    pgv::bindings::PG_MAJORVERSION_NUM
-                })
-            }
-        }
-        $(
-        impl Into<self::$name> for $crate::$md::$t {
-            fn into(self) -> self::$name {
-                self::$name::$variant (self)
-            }
-        }
-        )+
-    };
-    {name = $name:ident,
-     path = $p:ident,
-     typ = $t:ident,
-     pgversions = [$($variant:ident : $md:ident),+ $(,)?]} => {
-        pub enum $name {
-            $($variant ($crate::$md::$p::$t)),+
-        }
-        impl $name {
-            pub fn pg_version(&self) -> u32 {
-                enum_pgversion_dispatch!(self, $name, _ign, {
-                    pgv::bindings::PG_MAJORVERSION_NUM
-                })
-            }
-        }
-        $(
-        impl Into<$name> for $crate::$md::$p::$t {
-            fn into(self) -> $name {
-                $name::$variant (self)
-            }
-        }
-        )+
-    };
-}
-
 pub mod pg_constants;
 pub mod relfile_utils;
 

libs/remote_storage/src/config.rs

@@ -185,7 +185,7 @@ mod tests {
     use super::*;
 
     fn parse(input: &str) -> anyhow::Result<RemoteStorageConfig> {
-        let toml = input.parse::<toml_edit::DocumentMut>().unwrap();
+        let toml = input.parse::<toml_edit::Document>().unwrap();
         RemoteStorageConfig::from_toml(toml.as_item())
     }
 
@@ -235,31 +235,6 @@ timeout = '5s'";
         );
     }
 
-    #[test]
-    fn test_storage_class_serde_roundtrip() {
-        let classes = [
-            None,
-            Some(StorageClass::Standard),
-            Some(StorageClass::IntelligentTiering),
-        ];
-        for class in classes {
-            #[derive(Serialize, Deserialize)]
-            struct Wrapper {
-                #[serde(
-                    deserialize_with = "deserialize_storage_class",
-                    serialize_with = "serialize_storage_class"
-                )]
-                class: Option<StorageClass>,
-            }
-            let wrapped = Wrapper {
-                class: class.clone(),
-            };
-            let serialized = serde_json::to_string(&wrapped).unwrap();
-            let deserialized: Wrapper = serde_json::from_str(&serialized).unwrap();
-            assert_eq!(class, deserialized.class);
-        }
-    }
-
     #[test]
     fn test_azure_parsing() {
         let toml = "\

libs/remote_storage/src/lib.rs

@@ -45,8 +45,6 @@ pub use azure_core::Etag;
 
 pub use error::{DownloadError, TimeTravelError, TimeoutOrCancel};
 
-/// Default concurrency limit for S3 operations
-///
 /// Currently, sync happens with AWS S3, that has two limits on requests per second:
 /// ~200 RPS for IAM services
 /// <https://docs.aws.amazon.com/AmazonRDS/latest/AuroraUserGuide/UsingWithRDS.IAMDBAuth.html>
@@ -302,9 +300,7 @@ pub trait RemoteStorage: Send + Sync + 'static {
     ) -> Result<(), TimeTravelError>;
 }
 
-/// Data part of an ongoing [`Download`].
+/// DownloadStream is sensitive to the timeout and cancellation used with the original
-///
-/// `DownloadStream` is sensitive to the timeout and cancellation used with the original
 /// [`RemoteStorage::download`] request. The type yields `std::io::Result<Bytes>` to be compatible
 /// with `tokio::io::copy_buf`.
 // This has 'static because safekeepers do not use cancellation tokens (yet)

libs/safekeeper_api/src/models.rs

@@ -60,16 +60,3 @@ pub struct TimelineCopyRequest {
     pub target_timeline_id: TimelineId,
     pub until_lsn: Lsn,
 }
-
-#[derive(Debug, Clone, Deserialize, Serialize)]
-pub struct TimelineTermBumpRequest {
-    /// bump to
-    pub term: Option<u64>,
-}
-
-#[derive(Debug, Clone, Deserialize, Serialize)]
-pub struct TimelineTermBumpResponse {
-    // before the request
-    pub previous_term: u64,
-    pub current_term: u64,
-}

libs/tenant_size_model/src/lib.rs

@@ -5,10 +5,9 @@
 mod calculation;
 pub mod svg;
 
-/// StorageModel is the input to the synthetic size calculation.
+/// StorageModel is the input to the synthetic size calculation. It represents
-///
+/// a tree of timelines, with just the information that's needed for the
-/// It represents a tree of timelines, with just the information that's needed
+/// calculation. This doesn't track timeline names or where each timeline
-/// for the calculation. This doesn't track timeline names or where each timeline
 /// begins and ends, for example. Instead, it consists of "points of interest"
 /// on the timelines. A point of interest could be the timeline start or end point,
 /// the oldest point on a timeline that needs to be retained because of PITR

libs/utils/src/circuit_breaker.rs

@@ -5,10 +5,8 @@ use std::{
 
 use metrics::IntCounter;
 
-/// Circuit breakers are for operations that are expensive and fallible.
+/// Circuit breakers are for operations that are expensive and fallible: if they fail repeatedly,
-///
+/// we will stop attempting them for some period of time, to avoid denial-of-service from retries, and
-/// If a circuit breaker fails repeatedly, we will stop attempting it for some
-/// period of time, to avoid denial-of-service from retries, and
 /// to mitigate the log spam from repeated failures.
 pub struct CircuitBreaker {
     /// An identifier that enables us to log useful errors when a circuit is broken

libs/utils/src/crashsafe.rs

@@ -1,4 +1,3 @@
-use std::os::fd::AsRawFd;
 use std::{
     borrow::Cow,
     fs::{self, File},
@@ -204,27 +203,6 @@ pub fn overwrite(
     Ok(())
 }
 
-/// Syncs the filesystem for the given file descriptor.
-#[cfg_attr(target_os = "macos", allow(unused_variables))]
-pub fn syncfs(fd: impl AsRawFd) -> anyhow::Result<()> {
-    // Linux guarantees durability for syncfs.
-    // POSIX doesn't have syncfs, and further does not actually guarantee durability of sync().
-    #[cfg(target_os = "linux")]
-    {
-        use anyhow::Context;
-        nix::unistd::syncfs(fd.as_raw_fd()).context("syncfs")?;
-    }
-    #[cfg(target_os = "macos")]
-    {
-        // macOS is not a production platform for Neon, don't even bother.
-    }
-    #[cfg(not(any(target_os = "linux", target_os = "macos")))]
-    {
-        compile_error!("Unsupported OS");
-    }
-    Ok(())
-}
-
 #[cfg(test)]
 mod tests {
 

libs/utils/src/id.rs

@@ -249,10 +249,8 @@ macro_rules! id_newtype {
     };
 }
 
-/// Neon timeline ID.
+/// Neon timeline IDs are different from PostgreSQL timeline
-///
+/// IDs. They serve a similar purpose though: they differentiate
-/// They are different from PostgreSQL timeline
-/// IDs, but serve a similar purpose: they differentiate
 /// between different "histories" of the same cluster.  However,
 /// PostgreSQL timeline IDs are a bit cumbersome, because they are only
 /// 32-bits wide, and they must be in ascending order in any given

libs/utils/src/lock_file.rs

@@ -100,9 +100,7 @@ pub enum LockFileRead {
 }
 
 /// Open & try to lock the lock file at the given `path`, returning a [handle][`LockFileRead`] to
-/// inspect its content.
+/// inspect its content. It is not an `Err(...)` if the file does not exist or is already locked.
-///
-/// It is not an `Err(...)` if the file does not exist or is already locked.
 /// Check the [`LockFileRead`] variants for details.
 pub fn read_and_hold_lock_file(path: &Utf8Path) -> anyhow::Result<LockFileRead> {
     let res = fs::OpenOptions::new().read(true).open(path);

libs/utils/src/logging.rs

@@ -3,9 +3,9 @@ use std::str::FromStr;
 use anyhow::Context;
 use metrics::{IntCounter, IntCounterVec};
 use once_cell::sync::Lazy;
-use strum_macros::{EnumString, VariantNames};
+use strum_macros::{EnumString, EnumVariantNames};
 
-#[derive(EnumString, strum_macros::Display, VariantNames, Eq, PartialEq, Debug, Clone, Copy)]
+#[derive(EnumString, EnumVariantNames, Eq, PartialEq, Debug, Clone, Copy)]
 #[strum(serialize_all = "snake_case")]
 pub enum LogFormat {
     Plain,
@@ -188,7 +188,7 @@ impl Drop for TracingPanicHookGuard {
 }
 
 /// Named symbol for our panic hook, which logs the panic.
-fn tracing_panic_hook(info: &std::panic::PanicHookInfo) {
+fn tracing_panic_hook(info: &std::panic::PanicInfo) {
     // following rust 1.66.1 std implementation:
     // https://github.com/rust-lang/rust/blob/90743e7298aca107ddaa0c202a4d3604e29bfeb6/library/std/src/panicking.rs#L235-L288
     let location = info.location();
@@ -274,14 +274,6 @@ impl From<String> for SecretString {
     }
 }
 
-impl FromStr for SecretString {
-    type Err = std::convert::Infallible;
-
-    fn from_str(s: &str) -> Result<Self, Self::Err> {
-        Ok(Self(s.to_string()))
-    }
-}
-
 impl std::fmt::Debug for SecretString {
     fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
         write!(f, "[SECRET]")

libs/utils/src/pageserver_feedback.rs

@@ -8,7 +8,6 @@ use tracing::{trace, warn};
 use crate::lsn::Lsn;
 
 /// Feedback pageserver sends to safekeeper and safekeeper resends to compute.
-///
 /// Serialized in custom flexible key/value format. In replication protocol, it
 /// is marked with NEON_STATUS_UPDATE_TAG_BYTE to differentiate from postgres
 /// Standby status update / Hot standby feedback messages.

libs/utils/src/poison.rs

@@ -65,8 +65,6 @@ impl<T> Poison<T> {
     }
 }
 
-/// Armed pointer to a [`Poison`].
-///
 /// Use [`Self::data`] and [`Self::data_mut`] to access the wrapped state.
 /// Once modifications are done, use [`Self::disarm`].
 /// If [`Guard`] gets dropped instead of calling [`Self::disarm`], the state is poisoned

libs/utils/src/shard.rs

@@ -13,11 +13,10 @@ pub struct ShardNumber(pub u8);
 #[derive(Ord, PartialOrd, Eq, PartialEq, Clone, Copy, Serialize, Deserialize, Debug, Hash)]
 pub struct ShardCount(pub u8);
 
-/// Combination of ShardNumber and ShardCount.
+/// Combination of ShardNumber and ShardCount.  For use within the context of a particular tenant,
-///
+/// when we need to know which shard we're dealing with, but do not need to know the full
-/// For use within the context of a particular tenant, when we need to know which shard we're
+/// ShardIdentity (because we won't be doing any page->shard mapping), and do not need to know
-/// dealing with, but do not need to know the full ShardIdentity (because we won't be doing
+/// the fully qualified TenantShardId.
-/// any page->shard mapping), and do not need to know the fully qualified TenantShardId.
 #[derive(Eq, PartialEq, PartialOrd, Ord, Clone, Copy, Hash)]
 pub struct ShardIndex {
     pub shard_number: ShardNumber,

libs/utils/src/simple_rcu.rs

@@ -49,11 +49,12 @@ use std::sync::{RwLock, RwLockWriteGuard};
 
 use tokio::sync::watch;
 
-/// Rcu allows multiple readers to read and hold onto a value without blocking
-/// (for very long).
 ///
-/// Storing to the Rcu updates the value, making new readers immediately see
+/// Rcu allows multiple readers to read and hold onto a value without blocking
-/// the new value, but it also waits for all current readers to finish.
+/// (for very long).  Storing to the Rcu updates the value, making new readers
+/// immediately see the new value, but it also waits for all current readers to
+/// finish.
+///
 pub struct Rcu<V> {
     inner: RwLock<RcuInner<V>>,
 }

libs/utils/src/sync/heavier_once_cell.rs

@@ -5,9 +5,7 @@ use std::sync::{
 use tokio::sync::Semaphore;
 
 /// Custom design like [`tokio::sync::OnceCell`] but using [`OwnedSemaphorePermit`] instead of
-/// `SemaphorePermit`.
+/// `SemaphorePermit`, allowing use of `take` which does not require holding an outer mutex guard
-///
-/// Allows use of `take` which does not require holding an outer mutex guard
 /// for the duration of initialization.
 ///
 /// Has no unsafe, builds upon [`tokio::sync::Semaphore`] and [`std::sync::Mutex`].

libs/utils/src/toml_edit_ext.rs

@@ -10,7 +10,7 @@ pub fn deserialize_item<T>(item: &toml_edit::Item) -> Result<T, Error>
 where
     T: serde::de::DeserializeOwned,
 {
-    let document: toml_edit::DocumentMut = match item {
+    let document: toml_edit::Document = match item {
         toml_edit::Item::Table(toml) => toml.clone().into(),
         toml_edit::Item::Value(toml_edit::Value::InlineTable(toml)) => {
             toml.clone().into_table().into()

libs/utils/src/vec_map.rs

@@ -7,7 +7,6 @@ pub enum VecMapOrdering {
 }
 
 /// Ordered map datastructure implemented in a Vec.
-///
 /// Append only - can only add keys that are larger than the
 /// current max key.
 /// Ordering can be adjusted using [`VecMapOrdering`]

libs/utils/src/yielding_loop.rs

@@ -6,10 +6,9 @@ pub enum YieldingLoopError {
     Cancelled,
 }
 
-/// Helper for long synchronous loops, e.g. over all tenants in the system.
+/// Helper for long synchronous loops, e.g. over all tenants in the system.  Periodically
-///
+/// yields to avoid blocking the executor, and after resuming checks the provided
-/// Periodically yields to avoid blocking the executor, and after resuming
+/// cancellation token to drop out promptly on shutdown.
-/// checks the provided cancellation token to drop out promptly on shutdown.
 #[inline(always)]
 pub async fn yielding_loop<I, T, F>(
     interval: usize,
@@ -24,7 +23,7 @@ where
     for (i, item) in iter.enumerate() {
         visitor(item);
 
-        if (i + 1) % interval == 0 {
+        if i + 1 % interval == 0 {
             tokio::task::yield_now().await;
             if cancel.is_cancelled() {
                 return Err(YieldingLoopError::Cancelled);

libs/walproposer/build.rs

@@ -4,6 +4,7 @@
 use std::{env, path::PathBuf, process::Command};
 
 use anyhow::{anyhow, Context};
+use bindgen::CargoCallbacks;
 
 fn main() -> anyhow::Result<()> {
     // Tell cargo to invalidate the built crate whenever the wrapper changes
@@ -63,25 +64,16 @@ fn main() -> anyhow::Result<()> {
             .map_err(|s| anyhow!("Bad postgres server path {s:?}"))?
     };
 
-    let unwind_abi_functions = [
-        "log_internal",
-        "recovery_download",
-        "start_streaming",
-        "finish_sync_safekeepers",
-        "wait_event_set",
-        "WalProposerStart",
-    ];
-
     // The bindgen::Builder is the main entry point
     // to bindgen, and lets you build up options for
     // the resulting bindings.
-    let mut builder = bindgen::Builder::default()
+    let bindings = bindgen::Builder::default()
         // The input header we would like to generate
         // bindings for.
         .header("bindgen_deps.h")
         // Tell cargo to invalidate the built crate whenever any of the
         // included header files changed.
-        .parse_callbacks(Box::new(bindgen::CargoCallbacks::new()))
+        .parse_callbacks(Box::new(CargoCallbacks))
         .allowlist_type("WalProposer")
         .allowlist_type("WalProposerConfig")
         .allowlist_type("walproposer_api")
@@ -113,12 +105,7 @@ fn main() -> anyhow::Result<()> {
         .allowlist_var("WL_SOCKET_MASK")
         .clang_arg("-DWALPROPOSER_LIB")
         .clang_arg(format!("-I{pgxn_neon}"))
-        .clang_arg(format!("-I{inc_server_path}"));
+        .clang_arg(format!("-I{inc_server_path}"))
-
-    for name in unwind_abi_functions {
-        builder = builder.override_abi(bindgen::Abi::CUnwind, name);
-    }
-    let bindings = builder
         // Finish the builder and generate the bindings.
         .generate()
         // Unwrap the Result and panic on failure.

libs/walproposer/src/api_bindings.rs

@@ -33,7 +33,7 @@ extern "C" fn get_shmem_state(wp: *mut WalProposer) -> *mut WalproposerShmemStat
     }
 }
 
-extern "C-unwind" fn start_streaming(wp: *mut WalProposer, startpos: XLogRecPtr) {
+extern "C" fn start_streaming(wp: *mut WalProposer, startpos: XLogRecPtr) {
     unsafe {
         let callback_data = (*(*wp).config).callback_data;
         let api = callback_data as *mut Box<dyn ApiImpl>;
@@ -187,7 +187,7 @@ extern "C" fn conn_blocking_write(
     }
 }
 
-extern "C-unwind" fn recovery_download(wp: *mut WalProposer, sk: *mut Safekeeper) -> bool {
+extern "C" fn recovery_download(wp: *mut WalProposer, sk: *mut Safekeeper) -> bool {
     unsafe {
         let callback_data = (*(*(*sk).wp).config).callback_data;
         let api = callback_data as *mut Box<dyn ApiImpl>;
@@ -272,7 +272,7 @@ extern "C" fn rm_safekeeper_event_set(sk: *mut Safekeeper) {
     }
 }
 
-extern "C-unwind" fn wait_event_set(
+extern "C" fn wait_event_set(
     wp: *mut WalProposer,
     timeout: ::std::os::raw::c_long,
     event_sk: *mut *mut Safekeeper,
@@ -324,7 +324,7 @@ extern "C" fn get_redo_start_lsn(wp: *mut WalProposer) -> XLogRecPtr {
     }
 }
 
-extern "C-unwind" fn finish_sync_safekeepers(wp: *mut WalProposer, lsn: XLogRecPtr) {
+extern "C" fn finish_sync_safekeepers(wp: *mut WalProposer, lsn: XLogRecPtr) {
     unsafe {
         let callback_data = (*(*wp).config).callback_data;
         let api = callback_data as *mut Box<dyn ApiImpl>;
@@ -340,7 +340,7 @@ extern "C" fn process_safekeeper_feedback(wp: *mut WalProposer, sk: *mut Safekee
     }
 }
 
-extern "C-unwind" fn log_internal(
+extern "C" fn log_internal(
     wp: *mut WalProposer,
     level: ::std::os::raw::c_int,
     line: *const ::std::os::raw::c_char,

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" ]
+testing = ["fail/failpoints"]
 
 [dependencies]
 anyhow.workspace = true
@@ -101,7 +101,6 @@ procfs.workspace = true
 criterion.workspace = true
 hex-literal.workspace = true
 tokio = { workspace = true, features = ["process", "sync", "fs", "rt", "io-util", "time", "test-util"] }
-indoc.workspace = true
 
 [[bench]]
 name = "bench_layer_map"

pageserver/benches/bench_ingest.rs

@@ -4,7 +4,7 @@ use bytes::Bytes;
 use camino::Utf8PathBuf;
 use criterion::{criterion_group, criterion_main, Criterion};
 use pageserver::{
-    config::PageServerConf,
+    config::{defaults::DEFAULT_IO_BUFFER_ALIGNMENT, PageServerConf},
     context::{DownloadBehavior, RequestContext},
     l0_flush::{L0FlushConfig, L0FlushGlobalState},
     page_cache,
@@ -167,7 +167,7 @@ fn criterion_benchmark(c: &mut Criterion) {
     virtual_file::init(
         16384,
         virtual_file::io_engine_for_bench(),
-        pageserver_api::config::defaults::DEFAULT_IO_BUFFER_ALIGNMENT,
+        DEFAULT_IO_BUFFER_ALIGNMENT,
     );
     page_cache::init(conf.page_cache_size);
 

pageserver/client/src/lib.rs

@@ -1,20 +1,2 @@
 pub mod mgmt_api;
 pub mod page_service;
-
-/// For timeline_block_unblock_gc, distinguish the two different operations. This could be a bool.
-// If file structure is per-kind not per-feature then where to put this?
-#[derive(Clone, Copy)]
-pub enum BlockUnblock {
-    Block,
-    Unblock,
-}
-
-impl std::fmt::Display for BlockUnblock {
-    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
-        let s = match self {
-            BlockUnblock::Block => "block",
-            BlockUnblock::Unblock => "unblock",
-        };
-        f.write_str(s)
-    }
-}

pageserver/client/src/mgmt_api.rs

@@ -12,8 +12,6 @@ use utils::{
 
 pub use reqwest::Body as ReqwestBody;
 
-use crate::BlockUnblock;
-
 pub mod util;
 
 #[derive(Debug, Clone)]
@@ -456,20 +454,6 @@ impl Client {
             .map_err(Error::ReceiveBody)
     }
 
-    pub async fn timeline_block_unblock_gc(
-        &self,
-        tenant_shard_id: TenantShardId,
-        timeline_id: TimelineId,
-        dir: BlockUnblock,
-    ) -> Result<()> {
-        let uri = format!(
-            "{}/v1/tenant/{tenant_shard_id}/timeline/{timeline_id}/{dir}_gc",
-            self.mgmt_api_endpoint,
-        );
-
-        self.request(Method::POST, &uri, ()).await.map(|_| ())
-    }
-
     pub async fn tenant_reset(&self, tenant_shard_id: TenantShardId) -> Result<()> {
         let uri = format!(
             "{}/v1/tenant/{}/reset",

pageserver/ctl/src/layer_map_analyzer.rs

@@ -4,6 +4,7 @@
 
 use anyhow::Result;
 use camino::{Utf8Path, Utf8PathBuf};
+use pageserver::config::defaults::DEFAULT_IO_BUFFER_ALIGNMENT;
 use pageserver::context::{DownloadBehavior, RequestContext};
 use pageserver::task_mgr::TaskKind;
 use pageserver::tenant::{TENANTS_SEGMENT_NAME, TIMELINES_SEGMENT_NAME};
@@ -147,7 +148,7 @@ pub(crate) async fn main(cmd: &AnalyzeLayerMapCmd) -> Result<()> {
     pageserver::virtual_file::init(
         10,
         virtual_file::api::IoEngineKind::StdFs,
-        pageserver_api::config::defaults::DEFAULT_IO_BUFFER_ALIGNMENT,
+        DEFAULT_IO_BUFFER_ALIGNMENT,
     );
     pageserver::page_cache::init(100);
 

pageserver/ctl/src/layers.rs

@@ -3,6 +3,7 @@ use std::path::{Path, PathBuf};
 use anyhow::Result;
 use camino::{Utf8Path, Utf8PathBuf};
 use clap::Subcommand;
+use pageserver::config::defaults::DEFAULT_IO_BUFFER_ALIGNMENT;
 use pageserver::context::{DownloadBehavior, RequestContext};
 use pageserver::task_mgr::TaskKind;
 use pageserver::tenant::block_io::BlockCursor;
@@ -193,7 +194,7 @@ pub(crate) async fn main(cmd: &LayerCmd) -> Result<()> {
             pageserver::virtual_file::init(
                 10,
                 virtual_file::api::IoEngineKind::StdFs,
-                pageserver_api::config::defaults::DEFAULT_IO_BUFFER_ALIGNMENT,
+                DEFAULT_IO_BUFFER_ALIGNMENT,
             );
             pageserver::page_cache::init(100);
 

pageserver/ctl/src/main.rs

@@ -20,13 +20,14 @@ use clap::{Parser, Subcommand};
 use index_part::IndexPartCmd;
 use layers::LayerCmd;
 use pageserver::{
+    config::defaults::DEFAULT_IO_BUFFER_ALIGNMENT,
     context::{DownloadBehavior, RequestContext},
     page_cache,
     task_mgr::TaskKind,
     tenant::{dump_layerfile_from_path, metadata::TimelineMetadata},
     virtual_file,
 };
-use pageserver_api::{config::defaults::DEFAULT_IO_BUFFER_ALIGNMENT, shard::TenantShardId};
+use pageserver_api::shard::TenantShardId;
 use postgres_ffi::ControlFileData;
 use remote_storage::{RemotePath, RemoteStorageConfig};
 use tokio_util::sync::CancellationToken;
@@ -174,7 +175,7 @@ async fn main() -> anyhow::Result<()> {
                 println!("specified prefix '{}' failed validation", cmd.prefix);
                 return Ok(());
             };
-            let toml_document = toml_edit::DocumentMut::from_str(&cmd.config_toml_str)?;
+            let toml_document = toml_edit::Document::from_str(&cmd.config_toml_str)?;
             let toml_item = toml_document
                 .get("remote_storage")
                 .expect("need remote_storage");

pageserver/src/bin/pageserver.rs

@@ -5,7 +5,6 @@
 use std::env;
 use std::env::{var, VarError};
 use std::io::Read;
-use std::str::FromStr;
 use std::sync::Arc;
 use std::time::Duration;
 
@@ -37,7 +36,6 @@ use pageserver::{
     virtual_file,
 };
 use postgres_backend::AuthType;
-use utils::crashsafe::syncfs;
 use utils::failpoint_support;
 use utils::logging::TracingErrorLayerEnablement;
 use utils::{
@@ -126,6 +124,7 @@ fn main() -> anyhow::Result<()> {
     // after setting up logging, log the effective IO engine choice and read path implementations
     info!(?conf.virtual_file_io_engine, "starting with virtual_file IO engine");
     info!(?conf.virtual_file_direct_io, "starting with virtual_file Direct IO settings");
+    info!(?conf.compact_level0_phase1_value_access, "starting with setting for compact_level0_phase1_value_access");
     info!(?conf.io_buffer_alignment, "starting with setting for IO buffer alignment");
 
     // The tenants directory contains all the pageserver local disk state.
@@ -156,7 +155,23 @@ fn main() -> anyhow::Result<()> {
         };
 
         let started = Instant::now();
-        syncfs(dirfd)?;
+        // Linux guarantees durability for syncfs.
+        // POSIX doesn't have syncfs, and further does not actually guarantee durability of sync().
+        #[cfg(target_os = "linux")]
+        {
+            use std::os::fd::AsRawFd;
+            nix::unistd::syncfs(dirfd.as_raw_fd()).context("syncfs")?;
+        }
+        #[cfg(target_os = "macos")]
+        {
+            // macOS is not a production platform for Neon, don't even bother.
+            drop(dirfd);
+        }
+        #[cfg(not(any(target_os = "linux", target_os = "macos")))]
+        {
+            compile_error!("Unsupported OS");
+        }
+
         let elapsed = started.elapsed();
         info!(
             elapsed_ms = elapsed.as_millis(),
@@ -208,15 +223,27 @@ fn initialize_config(
         }
     };
 
-    let config_file_contents =
+    let config: toml_edit::Document = match std::fs::File::open(cfg_file_path) {
-        std::fs::read_to_string(cfg_file_path).context("read config file from filesystem")?;
+        Ok(mut f) => {
-    let config_toml = serde_path_to_error::deserialize(
+            let md = f.metadata().context("stat config file")?;
-        toml_edit::de::Deserializer::from_str(&config_file_contents)
+            if md.is_file() {
-            .context("build toml deserializer")?,
+                let mut s = String::new();
-    )
+                f.read_to_string(&mut s).context("read config file")?;
-    .context("deserialize config toml")?;
+                s.parse().context("parse config file toml")?
-    let conf = PageServerConf::parse_and_validate(identity.id, config_toml, workdir)
+            } else {
-        .context("runtime-validation of config toml")?;
+                anyhow::bail!("directory entry exists but is not a file: {cfg_file_path}");
+            }
+        }
+        Err(e) => {
+            anyhow::bail!("open pageserver config: {e}: {cfg_file_path}");
+        }
+    };
+
+    debug!("Using pageserver toml: {config}");
+
+    // Construct the runtime representation
+    let conf = PageServerConf::parse_and_validate(identity.id, &config, workdir)
+        .context("Failed to parse pageserver configuration")?;
 
     Ok(Box::leak(Box::new(conf)))
 }

pageserver/src/context.rs

@@ -1,9 +1,7 @@
-//! Defines [`RequestContext`].
+//! This module defines `RequestContext`, a structure that we use throughout
-//!
+//! the pageserver to propagate high-level context from places
-//! It is a structure that we use throughout the pageserver to propagate
+//! that _originate_ activity down to the shared code paths at the
-//! high-level context from places that _originate_ activity down to the
+//! heart of the pageserver. It's inspired by Golang's `context.Context`.
-//! shared code paths at the heart of the pageserver. It's inspired by
-//! Golang's `context.Context`.
 //!
 //! For example, in `Timeline::get(page_nr, lsn)` we need to answer the following questions:
 //! 1. What high-level activity ([`TaskKind`]) needs this page?

pageserver/src/control_plane_client.rs

@@ -141,24 +141,10 @@ impl ControlPlaneGenerationsApi for ControlPlaneClient {
                         m.other
                     );
 
-                    let az_id = {
+                    let az_id = m
-                        let az_id_from_metadata = m
+                        .other
-                            .other
+                        .get("availability_zone_id")
-                            .get("availability_zone_id")
+                        .and_then(|jv| jv.as_str().map(|str| str.to_owned()));
-                            .and_then(|jv| jv.as_str().map(|str| str.to_owned()));
-
-                        match az_id_from_metadata {
-                            Some(az_id) => Some(az_id),
-                            None => {
-                                tracing::warn!("metadata.json does not contain an 'availability_zone_id' field");
-                                conf.availability_zone.clone()
-                            }
-                        }
-                    };
-
-                    if az_id.is_none() {
-                        panic!("Availablity zone id could not be inferred from metadata.json or pageserver config");
-                    }
 
                     Some(NodeRegisterRequest {
                         node_id: conf.id,
@@ -166,7 +152,7 @@ impl ControlPlaneGenerationsApi for ControlPlaneClient {
                         listen_pg_port: m.postgres_port,
                         listen_http_addr: m.http_host,
                         listen_http_port: m.http_port,
-                        availability_zone_id: az_id.expect("Checked above"),
+                        availability_zone_id: az_id,
                     })
                 }
                 Err(e) => {

pageserver/src/disk_usage_eviction_task.rs

@@ -41,15 +41,19 @@
 // - The `#[allow(dead_code)]` above various structs are to suppress warnings about only the Debug impl
 //   reading these fields. We use the Debug impl for semi-structured logging, though.
 
-use std::{sync::Arc, time::SystemTime};
+use std::{
+    sync::Arc,
+    time::{Duration, SystemTime},
+};
 
 use anyhow::Context;
-use pageserver_api::{config::DiskUsageEvictionTaskConfig, shard::TenantShardId};
+use pageserver_api::shard::TenantShardId;
 use remote_storage::GenericRemoteStorage;
-use serde::Serialize;
+use serde::{Deserialize, Serialize};
 use tokio::time::Instant;
 use tokio_util::sync::CancellationToken;
 use tracing::{debug, error, info, instrument, warn, Instrument};
+use utils::serde_percent::Percent;
 use utils::{completion, id::TimelineId};
 
 use crate::{
@@ -65,9 +69,23 @@ use crate::{
     CancellableTask, DiskUsageEvictionTask,
 };
 
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
+pub struct DiskUsageEvictionTaskConfig {
+    pub max_usage_pct: Percent,
+    pub min_avail_bytes: u64,
+    #[serde(with = "humantime_serde")]
+    pub period: Duration,
+    #[cfg(feature = "testing")]
+    pub mock_statvfs: Option<crate::statvfs::mock::Behavior>,
+    /// Select sorting for evicted layers
+    #[serde(default)]
+    pub eviction_order: EvictionOrder,
+}
+
 /// Selects the sort order for eviction candidates *after* per tenant `min_resident_size`
 /// partitioning.
-#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize)]
+#[serde(tag = "type", content = "args")]
 pub enum EvictionOrder {
     /// Order the layers to be evicted by how recently they have been accessed relatively within
     /// the set of resident layers of a tenant.
@@ -78,22 +96,23 @@ pub enum EvictionOrder {
         /// we read tenants is deterministic. If we find the need to use this as `false`, we need
         /// to ensure nondeterminism by adding in a random number to break the
         /// `relative_last_activity==0.0` ties.
+        #[serde(default = "default_highest_layer_count_loses_first")]
         highest_layer_count_loses_first: bool,
     },
 }
 
-impl From<pageserver_api::config::EvictionOrder> for EvictionOrder {
+impl Default for EvictionOrder {
-    fn from(value: pageserver_api::config::EvictionOrder) -> Self {
+    fn default() -> Self {
-        match value {
+        Self::RelativeAccessed {
-            pageserver_api::config::EvictionOrder::RelativeAccessed {
+            highest_layer_count_loses_first: true,
-                highest_layer_count_loses_first,
-            } => Self::RelativeAccessed {
-                highest_layer_count_loses_first,
-            },
         }
     }
 }
 
+fn default_highest_layer_count_loses_first() -> bool {
+    true
+}
+
 impl EvictionOrder {
     fn sort(&self, candidates: &mut [(EvictionPartition, EvictionCandidate)]) {
         use EvictionOrder::*;
@@ -276,7 +295,7 @@ async fn disk_usage_eviction_task_iteration(
         storage,
         usage_pre,
         tenant_manager,
-        task_config.eviction_order.into(),
+        task_config.eviction_order,
         cancel,
     )
     .await;
@@ -1238,6 +1257,7 @@ mod filesystem_level_usage {
 
     #[test]
     fn max_usage_pct_pressure() {
+        use super::EvictionOrder;
         use super::Usage as _;
         use std::time::Duration;
         use utils::serde_percent::Percent;
@@ -1249,7 +1269,7 @@ mod filesystem_level_usage {
                 period: Duration::MAX,
                 #[cfg(feature = "testing")]
                 mock_statvfs: None,
-                eviction_order: pageserver_api::config::EvictionOrder::default(),
+                eviction_order: EvictionOrder::default(),
             },
             total_bytes: 100_000,
             avail_bytes: 0,

pageserver/src/http/routes.rs

@@ -2076,7 +2076,7 @@ async fn disk_usage_eviction_run(
         evict_bytes: u64,
 
         #[serde(default)]
-        eviction_order: pageserver_api::config::EvictionOrder,
+        eviction_order: crate::disk_usage_eviction_task::EvictionOrder,
     }
 
     #[derive(Debug, Clone, Copy, serde::Serialize)]
@@ -2112,7 +2112,7 @@ async fn disk_usage_eviction_run(
         &state.remote_storage,
         usage,
         &state.tenant_manager,
-        config.eviction_order.into(),
+        config.eviction_order,
         &cancel,
     )
     .await;

pageserver/src/l0_flush.rs

@@ -1,7 +1,9 @@
 use std::{num::NonZeroUsize, sync::Arc};
 
-#[derive(Debug, PartialEq, Eq, Clone)]
+#[derive(Debug, PartialEq, Eq, Clone, serde::Deserialize)]
+#[serde(tag = "mode", rename_all = "kebab-case", deny_unknown_fields)]
 pub enum L0FlushConfig {
+    #[serde(rename_all = "snake_case")]
     Direct { max_concurrency: NonZeroUsize },
 }
 
@@ -14,16 +16,6 @@ impl Default for L0FlushConfig {
     }
 }
 
-impl From<pageserver_api::models::L0FlushConfig> for L0FlushConfig {
-    fn from(config: pageserver_api::models::L0FlushConfig) -> Self {
-        match config {
-            pageserver_api::models::L0FlushConfig::Direct { max_concurrency } => {
-                Self::Direct { max_concurrency }
-            }
-        }
-    }
-}
-
 #[derive(Clone)]
 pub struct L0FlushGlobalState(Arc<Inner>);
 

pageserver/src/metrics.rs

@@ -9,7 +9,7 @@ use metrics::{
 use once_cell::sync::Lazy;
 use pageserver_api::shard::TenantShardId;
 use strum::{EnumCount, VariantNames};
-use strum_macros::{IntoStaticStr, VariantNames};
+use strum_macros::{EnumVariantNames, IntoStaticStr};
 use tracing::warn;
 use utils::id::TimelineId;
 
@@ -27,7 +27,7 @@ const CRITICAL_OP_BUCKETS: &[f64] = &[
 ];
 
 // Metrics collected on operations on the storage repository.
-#[derive(Debug, VariantNames, IntoStaticStr)]
+#[derive(Debug, EnumVariantNames, IntoStaticStr)]
 #[strum(serialize_all = "kebab_case")]
 pub(crate) enum StorageTimeOperation {
     #[strum(serialize = "layer flush")]

pageserver/src/page_service.rs

@@ -1199,6 +1199,7 @@ impl PageServerHandler {
     }
 }
 
+#[async_trait::async_trait]
 impl<IO> postgres_backend::Handler<IO> for PageServerHandler
 where
     IO: AsyncRead + AsyncWrite + Send + Sync + Unpin,

pageserver/src/pgdatadir_mapping.rs

@@ -729,12 +729,8 @@ impl Timeline {
         let current_policy = self.last_aux_file_policy.load();
         match current_policy {
             Some(AuxFilePolicy::V1) => {
-                let res = self.list_aux_files_v1(lsn, ctx).await?;
+                warn!("this timeline is using deprecated aux file policy V1 (policy=V1)");
-                let empty_str = if res.is_empty() { ", empty" } else { "" };
+                self.list_aux_files_v1(lsn, ctx).await
-                warn!(
-                    "this timeline is using deprecated aux file policy V1 (policy=v1{empty_str})"
-                );
-                Ok(res)
             }
             None => {
                 let res = self.list_aux_files_v1(lsn, ctx).await?;
@@ -1021,10 +1017,9 @@ impl Timeline {
 }
 
 /// DatadirModification represents an operation to ingest an atomic set of
-/// updates to the repository.
+/// updates to the repository. It is created by the 'begin_record'
-///
+/// function. It is called for each WAL record, so that all the modifications
-/// It is created by the 'begin_record' function. It is called for each WAL
+/// by a one WAL record appear atomic.
-/// record, so that all the modifications by a one WAL record appear atomic.
 pub struct DatadirModification<'a> {
     /// The timeline this modification applies to. You can access this to
     /// read the state, but note that any pending updates are *not* reflected
@@ -1205,13 +1200,6 @@ impl<'a> DatadirModification<'a> {
         img: Bytes,
     ) -> anyhow::Result<()> {
         anyhow::ensure!(rel.relnode != 0, RelationError::InvalidRelnode);
-        let key = rel_block_to_key(rel, blknum);
-        if !key.is_valid_key_on_write_path() {
-            anyhow::bail!(
-                "the request contains data not supported by pageserver at {}",
-                key
-            );
-        }
         self.put(rel_block_to_key(rel, blknum), Value::Image(img));
         Ok(())
     }
@@ -1223,34 +1211,14 @@ impl<'a> DatadirModification<'a> {
         blknum: BlockNumber,
         img: Bytes,
     ) -> anyhow::Result<()> {
-        let key = slru_block_to_key(kind, segno, blknum);
+        self.put(slru_block_to_key(kind, segno, blknum), Value::Image(img));
-        if !key.is_valid_key_on_write_path() {
-            anyhow::bail!(
-                "the request contains data not supported by pageserver at {}",
-                key
-            );
-        }
-        self.put(key, Value::Image(img));
         Ok(())
     }
 
-    pub(crate) fn put_rel_page_image_zero(
+    pub(crate) fn put_rel_page_image_zero(&mut self, rel: RelTag, blknum: BlockNumber) {
-        &mut self,
+        self.pending_zero_data_pages
-        rel: RelTag,
+            .insert(rel_block_to_key(rel, blknum).to_compact());
-        blknum: BlockNumber,
-    ) -> anyhow::Result<()> {
-        anyhow::ensure!(rel.relnode != 0, RelationError::InvalidRelnode);
-        let key = rel_block_to_key(rel, blknum);
-        if !key.is_valid_key_on_write_path() {
-            anyhow::bail!(
-                "the request contains data not supported by pageserver: {} @ {}",
-                key,
-                self.lsn
-            );
-        }
-        self.pending_zero_data_pages.insert(key.to_compact());
         self.pending_bytes += ZERO_PAGE.len();
-        Ok(())
     }
 
     pub(crate) fn put_slru_page_image_zero(
@@ -1258,18 +1226,10 @@ impl<'a> DatadirModification<'a> {
         kind: SlruKind,
         segno: u32,
         blknum: BlockNumber,
-    ) -> anyhow::Result<()> {
+    ) {
-        let key = slru_block_to_key(kind, segno, blknum);
+        self.pending_zero_data_pages
-        if !key.is_valid_key_on_write_path() {
+            .insert(slru_block_to_key(kind, segno, blknum).to_compact());
-            anyhow::bail!(
-                "the request contains data not supported by pageserver: {} @ {}",
-                key,
-                self.lsn
-            );
-        }
-        self.pending_zero_data_pages.insert(key.to_compact());
         self.pending_bytes += ZERO_PAGE.len();
-        Ok(())
     }
 
     /// Call this at the end of each WAL record.
@@ -1697,7 +1657,7 @@ impl<'a> DatadirModification<'a> {
                 if aux_files_key_v1.is_empty() {
                     None
                 } else {
-                    warn!("this timeline is using deprecated aux file policy V1 (detected existing v1 files)");
+                    warn!("this timeline is using deprecated aux file policy V1");
                     self.tline.do_switch_aux_policy(AuxFilePolicy::V1)?;
                     Some(AuxFilePolicy::V1)
                 }
@@ -2084,7 +2044,6 @@ impl<'a> DatadirModification<'a> {
 
 /// This struct facilitates accessing either a committed key from the timeline at a
 /// specific LSN, or the latest uncommitted key from a pending modification.
-///
 /// During WAL ingestion, the records from multiple LSNs may be batched in the same
 /// modification before being flushed to the timeline. Hence, the routines in WalIngest
 /// need to look up the keys in the modification first before looking them up in the

pageserver/src/statvfs.rs

@@ -60,7 +60,32 @@ pub mod mock {
     use regex::Regex;
     use tracing::log::info;
 
-    pub use pageserver_api::config::statvfs::mock::Behavior;
+    #[derive(Debug, Clone, PartialEq, Eq, serde::Serialize, serde::Deserialize)]
+    #[serde(tag = "type")]
+    pub enum Behavior {
+        Success {
+            blocksize: u64,
+            total_blocks: u64,
+            name_filter: Option<utils::serde_regex::Regex>,
+        },
+        Failure {
+            mocked_error: MockedError,
+        },
+    }
+
+    #[derive(Debug, Clone, Copy, PartialEq, Eq, serde::Serialize, serde::Deserialize)]
+    #[allow(clippy::upper_case_acronyms)]
+    pub enum MockedError {
+        EIO,
+    }
+
+    impl From<MockedError> for nix::Error {
+        fn from(e: MockedError) -> Self {
+            match e {
+                MockedError::EIO => nix::Error::EIO,
+            }
+        }
+    }
 
     pub fn get(tenants_dir: &Utf8Path, behavior: &Behavior) -> nix::Result<Statvfs> {
         info!("running mocked statvfs");
@@ -91,7 +116,6 @@ pub mod mock {
                     block_size: *blocksize,
                 })
             }
-            #[cfg(feature = "testing")]
             Behavior::Failure { mocked_error } => Err((*mocked_error).into()),
         }
     }

pageserver/src/tenant.rs

@@ -1,9 +1,8 @@
-//! Timeline repository implementation that keeps old data in layer files, and
-//! the recent changes in ephemeral files.
 //!
-//! See tenant/*_layer.rs files. The functions here are responsible for locating
+//! Timeline repository implementation that keeps old data in files on disk, and
-//! the correct layer for the get/put call, walking back the timeline branching
+//! the recent changes in memory. See tenant/*_layer.rs files.
-//! history as needed.
+//! The functions here are responsible for locating the correct layer for the
+//! get/put call, walking back the timeline branching history as needed.
 //!
 //! The files are stored in the .neon/tenants/<tenant_id>/timelines/<timeline_id>
 //! directory. See docs/pageserver-storage.md for how the files are managed.
@@ -7091,13 +7090,13 @@ mod tests {
             vec![
                 // Image layer at GC horizon
                 PersistentLayerKey {
-                    key_range: Key::MIN..Key::MAX,
+                    key_range: Key::MIN..Key::NON_L0_MAX,
                     lsn_range: Lsn(0x30)..Lsn(0x31),
                     is_delta: false
                 },
-                // The delta layer below the horizon
+                // The delta layer covers the full range (with the layer key hack to avoid being recognized as L0)
                 PersistentLayerKey {
-                    key_range: get_key(3)..get_key(4),
+                    key_range: Key::MIN..Key::NON_L0_MAX,
                     lsn_range: Lsn(0x30)..Lsn(0x48),
                     is_delta: true
                 },

pageserver/src/tenant/config.rs

@@ -9,10 +9,11 @@
 //! may lead to a data loss.
 //!
 use anyhow::bail;
-pub(crate) use pageserver_api::config::TenantConfigToml as TenantConf;
 use pageserver_api::models::AuxFilePolicy;
+use pageserver_api::models::CompactionAlgorithm;
 use pageserver_api::models::CompactionAlgorithmSettings;
 use pageserver_api::models::EvictionPolicy;
+use pageserver_api::models::LsnLease;
 use pageserver_api::models::{self, ThrottleConfig};
 use pageserver_api::shard::{ShardCount, ShardIdentity, ShardNumber, ShardStripeSize};
 use serde::de::IntoDeserializer;
@@ -22,6 +23,50 @@ use std::num::NonZeroU64;
 use std::time::Duration;
 use utils::generation::Generation;
 
+pub mod defaults {
+
+    // FIXME: This current value is very low. I would imagine something like 1 GB or 10 GB
+    // would be more appropriate. But a low value forces the code to be exercised more,
+    // which is good for now to trigger bugs.
+    // This parameter actually determines L0 layer file size.
+    pub const DEFAULT_CHECKPOINT_DISTANCE: u64 = 256 * 1024 * 1024;
+    pub const DEFAULT_CHECKPOINT_TIMEOUT: &str = "10 m";
+
+    // FIXME the below configs are only used by legacy algorithm. The new algorithm
+    // has different parameters.
+
+    // Target file size, when creating image and delta layers.
+    // This parameter determines L1 layer file size.
+    pub const DEFAULT_COMPACTION_TARGET_SIZE: u64 = 128 * 1024 * 1024;
+
+    pub const DEFAULT_COMPACTION_PERIOD: &str = "20 s";
+    pub const DEFAULT_COMPACTION_THRESHOLD: usize = 10;
+    pub const DEFAULT_COMPACTION_ALGORITHM: super::CompactionAlgorithm =
+        super::CompactionAlgorithm::Legacy;
+
+    pub const DEFAULT_GC_HORIZON: u64 = 64 * 1024 * 1024;
+
+    // Large DEFAULT_GC_PERIOD is fine as long as PITR_INTERVAL is larger.
+    // If there's a need to decrease this value, first make sure that GC
+    // doesn't hold a layer map write lock for non-trivial operations.
+    // Relevant: https://github.com/neondatabase/neon/issues/3394
+    pub const DEFAULT_GC_PERIOD: &str = "1 hr";
+    pub const DEFAULT_IMAGE_CREATION_THRESHOLD: usize = 3;
+    pub const DEFAULT_PITR_INTERVAL: &str = "7 days";
+    pub const DEFAULT_WALRECEIVER_CONNECT_TIMEOUT: &str = "10 seconds";
+    pub const DEFAULT_WALRECEIVER_LAGGING_WAL_TIMEOUT: &str = "10 seconds";
+    // The default limit on WAL lag should be set to avoid causing disconnects under high throughput
+    // scenarios: since the broker stats are updated ~1/s, a value of 1GiB should be sufficient for
+    // throughputs up to 1GiB/s per timeline.
+    pub const DEFAULT_MAX_WALRECEIVER_LSN_WAL_LAG: u64 = 1024 * 1024 * 1024;
+    pub const DEFAULT_EVICTIONS_LOW_RESIDENCE_DURATION_METRIC_THRESHOLD: &str = "24 hour";
+    // By default ingest enough WAL for two new L0 layers before checking if new image
+    // image layers should be created.
+    pub const DEFAULT_IMAGE_LAYER_CREATION_CHECK_THRESHOLD: u8 = 2;
+
+    pub const DEFAULT_INGEST_BATCH_SIZE: u64 = 100;
+}
+
 #[derive(Debug, Copy, Clone, Serialize, Deserialize, PartialEq, Eq)]
 pub(crate) enum AttachmentMode {
     /// Our generation is current as far as we know, and as far as we know we are the only attached
@@ -236,20 +281,96 @@ impl LocationConf {
     }
 }
 
-impl Default for LocationConf {
+/// A tenant's calcuated configuration, which is the result of merging a
-    // TODO: this should be removed once tenant loading can guarantee that we are never
+/// tenant's TenantConfOpt with the global TenantConf from PageServerConf.
-    // loading from a directory without a configuration.
+///
-    // => tech debt since https://github.com/neondatabase/neon/issues/1555
+/// For storing and transmitting individual tenant's configuration, see
-    fn default() -> Self {
+/// TenantConfOpt.
-        Self {
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
-            mode: LocationMode::Attached(AttachedLocationConfig {
+pub struct TenantConf {
-                generation: Generation::none(),
+    // Flush out an inmemory layer, if it's holding WAL older than this
-                attach_mode: AttachmentMode::Single,
+    // This puts a backstop on how much WAL needs to be re-digested if the
-            }),
+    // page server crashes.
-            tenant_conf: TenantConfOpt::default(),
+    // This parameter actually determines L0 layer file size.
-            shard: ShardIdentity::unsharded(),
+    pub checkpoint_distance: u64,
-        }
+    // Inmemory layer is also flushed at least once in checkpoint_timeout to
-    }
+    // eventually upload WAL after activity is stopped.
+    #[serde(with = "humantime_serde")]
+    pub checkpoint_timeout: Duration,
+    // Target file size, when creating image and delta layers.
+    // This parameter determines L1 layer file size.
+    pub compaction_target_size: u64,
+    // How often to check if there's compaction work to be done.
+    // Duration::ZERO means automatic compaction is disabled.
+    #[serde(with = "humantime_serde")]
+    pub compaction_period: Duration,
+    // Level0 delta layer threshold for compaction.
+    pub compaction_threshold: usize,
+    pub compaction_algorithm: CompactionAlgorithmSettings,
+    // 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.
+    // Page versions older than this are garbage collected away.
+    pub gc_horizon: u64,
+    // Interval at which garbage collection is triggered.
+    // Duration::ZERO means automatic GC is disabled
+    #[serde(with = "humantime_serde")]
+    pub gc_period: Duration,
+    // Delta layer churn threshold to create L1 image layers.
+    pub image_creation_threshold: usize,
+    // Determines how much history is retained, to allow
+    // branching and read replicas at an older point in time.
+    // The unit is time.
+    // Page versions older than this are garbage collected away.
+    #[serde(with = "humantime_serde")]
+    pub pitr_interval: Duration,
+    /// Maximum amount of time to wait while opening a connection to receive wal, before erroring.
+    #[serde(with = "humantime_serde")]
+    pub walreceiver_connect_timeout: Duration,
+    /// Considers safekeepers stalled after no WAL updates were received longer than this threshold.
+    /// A stalled safekeeper will be changed to a newer one when it appears.
+    #[serde(with = "humantime_serde")]
+    pub lagging_wal_timeout: Duration,
+    /// Considers safekeepers lagging when their WAL is behind another safekeeper for more than this threshold.
+    /// A lagging safekeeper will be changed after `lagging_wal_timeout` time elapses since the last WAL update,
+    /// to avoid eager reconnects.
+    pub max_lsn_wal_lag: NonZeroU64,
+    pub eviction_policy: EvictionPolicy,
+    pub min_resident_size_override: Option<u64>,
+    // See the corresponding metric's help string.
+    #[serde(with = "humantime_serde")]
+    pub evictions_low_residence_duration_metric_threshold: Duration,
+
+    /// If non-zero, the period between uploads of a heatmap from attached tenants.  This
+    /// may be disabled if a Tenant will not have secondary locations: only secondary
+    /// locations will use the heatmap uploaded by attached locations.
+    #[serde(with = "humantime_serde")]
+    pub heatmap_period: Duration,
+
+    /// If true then SLRU segments are dowloaded on demand, if false SLRU segments are included in basebackup
+    pub lazy_slru_download: bool,
+
+    pub timeline_get_throttle: pageserver_api::models::ThrottleConfig,
+
+    // How much WAL must be ingested before checking again whether a new image layer is required.
+    // Expresed in multiples of checkpoint distance.
+    pub image_layer_creation_check_threshold: u8,
+
+    /// Switch to a new aux file policy. Switching this flag requires the user has not written any aux file into
+    /// the storage before, and this flag cannot be switched back. Otherwise there will be data corruptions.
+    /// There is a `last_aux_file_policy` flag which gets persisted in `index_part.json` once the first aux
+    /// file is written.
+    pub switch_aux_file_policy: AuxFilePolicy,
+
+    /// The length for an explicit LSN lease request.
+    /// Layers needed to reconstruct pages at LSN will not be GC-ed during this interval.
+    #[serde(with = "humantime_serde")]
+    pub lsn_lease_length: Duration,
+
+    /// The length for an implicit LSN lease granted as part of `get_lsn_by_timestamp` request.
+    /// Layers needed to reconstruct pages at LSN will not be GC-ed during this interval.
+    #[serde(with = "humantime_serde")]
+    pub lsn_lease_length_for_ts: Duration,
 }
 
 /// Same as TenantConf, but this struct preserves the information about
@@ -424,6 +545,51 @@ impl TenantConfOpt {
     }
 }
 
+impl Default for TenantConf {
+    fn default() -> Self {
+        use defaults::*;
+        Self {
+            checkpoint_distance: DEFAULT_CHECKPOINT_DISTANCE,
+            checkpoint_timeout: humantime::parse_duration(DEFAULT_CHECKPOINT_TIMEOUT)
+                .expect("cannot parse default checkpoint timeout"),
+            compaction_target_size: DEFAULT_COMPACTION_TARGET_SIZE,
+            compaction_period: humantime::parse_duration(DEFAULT_COMPACTION_PERIOD)
+                .expect("cannot parse default compaction period"),
+            compaction_threshold: DEFAULT_COMPACTION_THRESHOLD,
+            compaction_algorithm: CompactionAlgorithmSettings {
+                kind: DEFAULT_COMPACTION_ALGORITHM,
+            },
+            gc_horizon: DEFAULT_GC_HORIZON,
+            gc_period: humantime::parse_duration(DEFAULT_GC_PERIOD)
+                .expect("cannot parse default gc period"),
+            image_creation_threshold: DEFAULT_IMAGE_CREATION_THRESHOLD,
+            pitr_interval: humantime::parse_duration(DEFAULT_PITR_INTERVAL)
+                .expect("cannot parse default PITR interval"),
+            walreceiver_connect_timeout: humantime::parse_duration(
+                DEFAULT_WALRECEIVER_CONNECT_TIMEOUT,
+            )
+            .expect("cannot parse default walreceiver connect timeout"),
+            lagging_wal_timeout: humantime::parse_duration(DEFAULT_WALRECEIVER_LAGGING_WAL_TIMEOUT)
+                .expect("cannot parse default walreceiver lagging wal timeout"),
+            max_lsn_wal_lag: NonZeroU64::new(DEFAULT_MAX_WALRECEIVER_LSN_WAL_LAG)
+                .expect("cannot parse default max walreceiver Lsn wal lag"),
+            eviction_policy: EvictionPolicy::NoEviction,
+            min_resident_size_override: None,
+            evictions_low_residence_duration_metric_threshold: humantime::parse_duration(
+                DEFAULT_EVICTIONS_LOW_RESIDENCE_DURATION_METRIC_THRESHOLD,
+            )
+            .expect("cannot parse default evictions_low_residence_duration_metric_threshold"),
+            heatmap_period: Duration::ZERO,
+            lazy_slru_download: false,
+            timeline_get_throttle: crate::tenant::throttle::Config::disabled(),
+            image_layer_creation_check_threshold: DEFAULT_IMAGE_LAYER_CREATION_CHECK_THRESHOLD,
+            switch_aux_file_policy: AuxFilePolicy::default_tenant_config(),
+            lsn_lease_length: LsnLease::DEFAULT_LENGTH,
+            lsn_lease_length_for_ts: LsnLease::DEFAULT_LENGTH_FOR_TS,
+        }
+    }
+}
+
 impl TryFrom<&'_ models::TenantConfig> for TenantConfOpt {
     type Error = anyhow::Error;
 
@@ -452,8 +618,7 @@ impl TryFrom<toml_edit::Item> for TenantConfOpt {
                     .map_err(|e| anyhow::anyhow!("{}: {}", e.path(), e.inner().message()));
             }
             toml_edit::Item::Table(table) => {
-                let deserializer =
+                let deserializer = toml_edit::de::Deserializer::new(table.into());
-                    toml_edit::de::Deserializer::from(toml_edit::DocumentMut::from(table));
                 return serde_path_to_error::deserialize(deserializer)
                     .map_err(|e| anyhow::anyhow!("{}: {}", e.path(), e.inner().message()));
             }

pageserver/src/tenant/metadata.rs

@@ -1,8 +1,7 @@
-//! Describes the legacy now hopefully no longer modified per-timeline metadata.
+//! Describes the legacy now hopefully no longer modified per-timeline metadata stored in
-//!
+//! `index_part.json` managed by [`remote_timeline_client`]. For many tenants and their timelines,
-//! It is stored in `index_part.json` managed by [`remote_timeline_client`]. For many tenants and
+//! this struct and it's original serialization format is still needed because they were written a
-//! their timelines, this struct and its original serialization format is still needed because
+//! long time ago.
-//! they were written a long time ago.
 //!
 //! Instead of changing and adding versioning to this, just change [`IndexPart`] with soft json
 //! versioning.

pageserver/src/tenant/mgr.rs

@@ -282,10 +282,9 @@ impl BackgroundPurges {
 static TENANTS: Lazy<std::sync::RwLock<TenantsMap>> =
     Lazy::new(|| std::sync::RwLock::new(TenantsMap::Initializing));
 
-/// Responsible for storing and mutating the collection of all tenants
+/// The TenantManager is responsible for storing and mutating the collection of all tenants
-/// that this pageserver has state for.
+/// that this pageserver process has state for.  Every Tenant and SecondaryTenant instance
-///
+/// lives inside the TenantManager.
-/// Every Tenant and SecondaryTenant instance lives inside the TenantManager.
 ///
 /// The most important role of the TenantManager is to prevent conflicts: e.g. trying to attach
 /// the same tenant twice concurrently, or trying to configure the same tenant into secondary
@@ -2347,9 +2346,8 @@ pub enum TenantMapError {
     ShuttingDown,
 }
 
-/// Guards a particular tenant_id's content in the TenantsMap.
+/// Guards a particular tenant_id's content in the TenantsMap.  While this
-///
+/// structure exists, the TenantsMap will contain a [`TenantSlot::InProgress`]
-/// While this structure exists, the TenantsMap will contain a [`TenantSlot::InProgress`]
 /// for this tenant, which acts as a marker for any operations targeting
 /// this tenant to retry later, or wait for the InProgress state to end.
 ///

pageserver/src/tenant/remote_timeline_client.rs

@@ -2184,8 +2184,6 @@ pub fn remote_timeline_path(
     remote_timelines_path(tenant_shard_id).join(Utf8Path::new(&timeline_id.to_string()))
 }
 
-/// Obtains the path of the given Layer in the remote
-///
 /// Note that the shard component of a remote layer path is _not_ always the same
 /// as in the TenantShardId of the caller: tenants may reference layers from a different
 /// ShardIndex.  Use the ShardIndex from the layer's metadata.

pageserver/src/tenant/remote_timeline_client/download.rs

@@ -548,7 +548,7 @@ pub(crate) async fn download_initdb_tar_zst(
         cancel,
     )
     .await
-    .inspect_err(|_e| {
+    .map_err(|e| {
         // Do a best-effort attempt at deleting the temporary file upon encountering an error.
         // We don't have async here nor do we want to pile on any extra errors.
         if let Err(e) = std::fs::remove_file(&temp_path) {
@@ -556,6 +556,7 @@ pub(crate) async fn download_initdb_tar_zst(
                 warn!("error deleting temporary file {temp_path}: {e}");
             }
         }
+        e
     })?;
 
     Ok((temp_path, file))

pageserver/src/tenant/remote_timeline_client/index.rs

@@ -1,5 +1,4 @@
 //! In-memory index to track the tenant files on the remote storage.
-//!
 //! Able to restore itself from the storage index parts, that are located in every timeline's remote directory and contain all data about
 //! remote timeline layers and its metadata.
 

pageserver/src/tenant/storage_layer.rs

@@ -434,11 +434,10 @@ impl ReadableLayer {
     }
 }
 
-/// Layers contain a hint indicating whether they are likely to be used for reads.
+/// Layers contain a hint indicating whether they are likely to be used for reads.  This is a hint rather
-///
+/// than an authoritative value, so that we do not have to update it synchronously when changing the visibility
-/// This is a hint rather than an authoritative value, so that we do not have to update it synchronously
+/// of layers (for example when creating a branch that makes some previously covered layers visible).  It should
-/// when changing the visibility of layers (for example when creating a branch that makes some previously
+/// be used for cache management but not for correctness-critical checks.
-/// covered layers visible).  It should be used for cache management but not for correctness-critical checks.
 #[derive(Debug, Clone, PartialEq, Eq)]
 pub enum LayerVisibilityHint {
     /// A Visible layer might be read while serving a read, because there is not an image layer between it

pageserver/src/tenant/storage_layer/delta_layer.rs

@@ -39,7 +39,7 @@ use crate::tenant::disk_btree::{
 use crate::tenant::storage_layer::layer::S3_UPLOAD_LIMIT;
 use crate::tenant::timeline::GetVectoredError;
 use crate::tenant::vectored_blob_io::{
-    BlobFlag, StreamingVectoredReadPlanner, VectoredBlobReader, VectoredRead,
+    BlobFlag, MaxVectoredReadBytes, StreamingVectoredReadPlanner, VectoredBlobReader, VectoredRead,
     VectoredReadCoalesceMode, VectoredReadPlanner,
 };
 use crate::tenant::PageReconstructError;
@@ -52,7 +52,6 @@ use bytes::BytesMut;
 use camino::{Utf8Path, Utf8PathBuf};
 use futures::StreamExt;
 use itertools::Itertools;
-use pageserver_api::config::MaxVectoredReadBytes;
 use pageserver_api::keyspace::KeySpace;
 use pageserver_api::models::ImageCompressionAlgorithm;
 use pageserver_api::shard::TenantShardId;
@@ -136,11 +135,10 @@ impl Summary {
 // Flag indicating that this version initialize the page
 const WILL_INIT: u64 = 1;
 
-/// Struct representing reference to BLOB in layers.
+/// Struct representing reference to BLOB in layers. Reference contains BLOB
-///
+/// offset, and for WAL records it also contains `will_init` flag. The flag
-/// Reference contains BLOB offset, and for WAL records it also contains
+/// helps to determine the range of records that needs to be applied, without
-/// `will_init` flag. The flag helps to determine the range of records
+/// reading/deserializing records themselves.
-/// that needs to be applied, without reading/deserializing records themselves.
 #[derive(Debug, Serialize, Deserialize, Copy, Clone)]
 pub struct BlobRef(pub u64);
 

pageserver/src/tenant/storage_layer/image_layer.rs

@@ -1,9 +1,7 @@
 //! An ImageLayer represents an image or a snapshot of a key-range at
-//! one particular LSN.
+//! one particular LSN. It contains an image of all key-value pairs
-//!
+//! in its key-range. Any key that falls into the image layer's range
-//! It contains an image of all key-value pairs in its key-range. Any key
+//! but does not exist in the layer, does not exist.
-//! that falls into the image layer's range but does not exist in the layer,
-//! does not exist.
 //!
 //! An image layer is stored in a file on disk. The file is stored in
 //! timelines/<timeline_id> directory.  Currently, there are no
@@ -36,7 +34,8 @@ use crate::tenant::disk_btree::{
 };
 use crate::tenant::timeline::GetVectoredError;
 use crate::tenant::vectored_blob_io::{
-    BlobFlag, StreamingVectoredReadPlanner, VectoredBlobReader, VectoredRead, VectoredReadPlanner,
+    BlobFlag, MaxVectoredReadBytes, StreamingVectoredReadPlanner, VectoredBlobReader, VectoredRead,
+    VectoredReadPlanner,
 };
 use crate::tenant::{PageReconstructError, Timeline};
 use crate::virtual_file::owned_buffers_io::io_buf_ext::IoBufExt;
@@ -47,7 +46,6 @@ use bytes::{Bytes, BytesMut};
 use camino::{Utf8Path, Utf8PathBuf};
 use hex;
 use itertools::Itertools;
-use pageserver_api::config::MaxVectoredReadBytes;
 use pageserver_api::keyspace::KeySpace;
 use pageserver_api::shard::{ShardIdentity, TenantShardId};
 use rand::{distributions::Alphanumeric, Rng};

pageserver/src/tenant/storage_layer/inmemory_layer.rs

@@ -215,7 +215,7 @@ impl IndexEntry {
 
     const _ASSERT_DEFAULT_CHECKPOINT_DISTANCE_IS_VALID: () = {
         let res = Self::validate_checkpoint_distance(
-            pageserver_api::config::tenant_conf_defaults::DEFAULT_CHECKPOINT_DISTANCE,
+            crate::tenant::config::defaults::DEFAULT_CHECKPOINT_DISTANCE,
         );
         if res.is_err() {
             panic!("default checkpoint distance is valid")

pageserver/src/tenant/storage_layer/layer_desc.rs

@@ -12,10 +12,8 @@ use serde::{Deserialize, Serialize};
 #[cfg(test)]
 use utils::id::TenantId;
 
-/// A unique identifier of a persistent layer.
+/// A unique identifier of a persistent layer. This is different from `LayerDescriptor`, which is only used in the
-///
+/// benchmarks. This struct contains all necessary information to find the image / delta layer. It also provides
-/// This is different from `LayerDescriptor`, which is only used in the benchmarks.
-/// This struct contains all necessary information to find the image / delta layer. It also provides
 /// a unified way to generate layer information like file name.
 #[derive(Debug, PartialEq, Eq, Clone, Serialize, Deserialize, Hash)]
 pub struct PersistentLayerDesc {

pageserver/src/tenant/storage_layer/layer_name.rs

@@ -217,9 +217,8 @@ impl fmt::Display for ImageLayerName {
     }
 }
 
-/// LayerName is the logical identity of a layer within a LayerMap at a moment in time.
+/// LayerName is the logical identity of a layer within a LayerMap at a moment in time.  The
-///
+/// LayerName is not a unique filename, as the same LayerName may have multiple physical incarnations
-/// The LayerName is not a unique filename, as the same LayerName may have multiple physical incarnations
 /// over time (e.g. across shard splits or compression). The physical filenames of layers in local
 /// storage and object names in remote storage consist of the LayerName plus some extra qualifiers
 /// that uniquely identify the physical incarnation of a layer (see [crate::tenant::remote_timeline_client::remote_layer_path])

pageserver/src/tenant/storage_layer/merge_iterator.rs

@@ -226,11 +226,9 @@ impl<'a> IteratorWrapper<'a> {
     }
 }
 
-/// A merge iterator over delta/image layer iterators.
+/// A merge iterator over delta/image layer iterators. When duplicated records are
-///
+/// found, the iterator will not perform any deduplication, and the caller should handle
-/// When duplicated records are found, the iterator will not perform any
+/// these situation. By saying duplicated records, there are many possibilities:
-/// deduplication, and the caller should handle these situation. By saying
-/// duplicated records, there are many possibilities:
 ///
 /// * Two same delta at the same LSN.
 /// * Two same image at the same LSN.

pageserver/src/tenant/storage_layer/split_writer.rs

@@ -34,10 +34,9 @@ impl SplitWriterResult {
     }
 }
 
-/// An image writer that takes images and produces multiple image layers.
+/// An image writer that takes images and produces multiple image layers. The interface does not
-///
+/// guarantee atomicity (i.e., if the image layer generation fails, there might be leftover files
-/// The interface does not guarantee atomicity (i.e., if the image layer generation
+/// to be cleaned up)
-/// fails, there might be leftover files to be cleaned up)
 #[must_use]
 pub struct SplitImageLayerWriter {
     inner: ImageLayerWriter,
@@ -188,23 +187,22 @@ impl SplitImageLayerWriter {
             .await
     }
 
-    /// This function will be deprecated with #8841.
+    /// When split writer fails, the caller should call this function and handle partially generated layers.
     pub(crate) fn take(self) -> anyhow::Result<(Vec<SplitWriterResult>, ImageLayerWriter)> {
         Ok((self.generated_layers, self.inner))
     }
 }
 
-/// A delta writer that takes key-lsn-values and produces multiple delta layers.
+/// A delta writer that takes key-lsn-values and produces multiple delta layers. The interface does not
-///
+/// guarantee atomicity (i.e., if the delta layer generation fails, there might be leftover files
-/// The interface does not guarantee atomicity (i.e., if the delta layer generation fails,
+/// to be cleaned up).
-/// there might be leftover files to be cleaned up).
 ///
 /// Note that if updates of a single key exceed the target size limit, all of the updates will be batched
 /// into a single file. This behavior might change in the future. For reference, the legacy compaction algorithm
 /// will split them into multiple files based on size.
 #[must_use]
 pub struct SplitDeltaLayerWriter {
-    inner: Option<(Key, DeltaLayerWriter)>,
+    inner: DeltaLayerWriter,
     target_layer_size: u64,
     generated_layers: Vec<SplitWriterResult>,
     conf: &'static PageServerConf,
@@ -212,6 +210,7 @@ pub struct SplitDeltaLayerWriter {
     tenant_shard_id: TenantShardId,
     lsn_range: Range<Lsn>,
     last_key_written: Key,
+    start_key: Key,
 }
 
 impl SplitDeltaLayerWriter {
@@ -219,18 +218,29 @@ impl SplitDeltaLayerWriter {
         conf: &'static PageServerConf,
         timeline_id: TimelineId,
         tenant_shard_id: TenantShardId,
+        start_key: Key,
         lsn_range: Range<Lsn>,
         target_layer_size: u64,
+        ctx: &RequestContext,
     ) -> anyhow::Result<Self> {
         Ok(Self {
             target_layer_size,
-            inner: None,
+            inner: DeltaLayerWriter::new(
+                conf,
+                timeline_id,
+                tenant_shard_id,
+                start_key,
+                lsn_range.clone(),
+                ctx,
+            )
+            .await?,
             generated_layers: Vec::new(),
             conf,
             timeline_id,
             tenant_shard_id,
             lsn_range,
             last_key_written: Key::MIN,
+            start_key,
         })
     }
 
@@ -253,26 +263,9 @@ impl SplitDeltaLayerWriter {
         //
         // Also, keep all updates of a single key in a single file. TODO: split them using the legacy compaction
         // strategy. https://github.com/neondatabase/neon/issues/8837
-
-        if self.inner.is_none() {
-            self.inner = Some((
-                key,
-                DeltaLayerWriter::new(
-                    self.conf,
-                    self.timeline_id,
-                    self.tenant_shard_id,
-                    key,
-                    self.lsn_range.clone(),
-                    ctx,
-                )
-                .await?,
-            ));
-        }
-        let (_, inner) = self.inner.as_mut().unwrap();
-
         let addition_size_estimation = KEY_SIZE as u64 + 8 /* LSN u64 size */ + 80 /* value size estimation */;
-        if inner.num_keys() >= 1
+        if self.inner.num_keys() >= 1
-            && inner.estimated_size() + addition_size_estimation >= self.target_layer_size
+            && self.inner.estimated_size() + addition_size_estimation >= self.target_layer_size
         {
             if key != self.last_key_written {
                 let next_delta_writer = DeltaLayerWriter::new(
@@ -284,13 +277,13 @@ impl SplitDeltaLayerWriter {
                     ctx,
                 )
                 .await?;
-                let (start_key, prev_delta_writer) =
+                let prev_delta_writer = std::mem::replace(&mut self.inner, next_delta_writer);
-                    std::mem::replace(&mut self.inner, Some((key, next_delta_writer))).unwrap();
                 let layer_key = PersistentLayerKey {
-                    key_range: start_key..key,
+                    key_range: self.start_key..key,
                     lsn_range: self.lsn_range.clone(),
                     is_delta: true,
                 };
+                self.start_key = key;
                 if discard(&layer_key).await {
                     drop(prev_delta_writer);
                     self.generated_layers
@@ -301,18 +294,17 @@ impl SplitDeltaLayerWriter {
                     self.generated_layers
                         .push(SplitWriterResult::Produced(delta_layer));
                 }
-            } else if inner.estimated_size() >= S3_UPLOAD_LIMIT {
+            } else if self.inner.estimated_size() >= S3_UPLOAD_LIMIT {
                 // We have to produce a very large file b/c a key is updated too often.
                 anyhow::bail!(
                     "a single key is updated too often: key={}, estimated_size={}, and the layer file cannot be produced",
                     key,
-                    inner.estimated_size()
+                    self.inner.estimated_size()
                 );
             }
         }
         self.last_key_written = key;
-        let (_, inner) = self.inner.as_mut().unwrap();
+        self.inner.put_value(key, lsn, val, ctx).await
-        inner.put_value(key, lsn, val, ctx).await
     }
 
     pub async fn put_value(
@@ -331,6 +323,7 @@ impl SplitDeltaLayerWriter {
         self,
         tline: &Arc<Timeline>,
         ctx: &RequestContext,
+        end_key: Key,
         discard: D,
     ) -> anyhow::Result<Vec<SplitWriterResult>>
     where
@@ -342,15 +335,11 @@ impl SplitDeltaLayerWriter {
             inner,
             ..
         } = self;
-        let Some((start_key, inner)) = inner else {
-            return Ok(generated_layers);
-        };
         if inner.num_keys() == 0 {
             return Ok(generated_layers);
         }
-        let end_key = self.last_key_written.next();
         let layer_key = PersistentLayerKey {
-            key_range: start_key..end_key,
+            key_range: self.start_key..end_key,
             lsn_range: self.lsn_range.clone(),
             is_delta: true,
         };
@@ -369,14 +358,15 @@ impl SplitDeltaLayerWriter {
         self,
         tline: &Arc<Timeline>,
         ctx: &RequestContext,
+        end_key: Key,
     ) -> anyhow::Result<Vec<SplitWriterResult>> {
-        self.finish_with_discard_fn(tline, ctx, |_| async { false })
+        self.finish_with_discard_fn(tline, ctx, end_key, |_| async { false })
             .await
     }
 
-    /// This function will be deprecated with #8841.
+    /// When split writer fails, the caller should call this function and handle partially generated layers.
-    pub(crate) fn take(self) -> anyhow::Result<(Vec<SplitWriterResult>, Option<DeltaLayerWriter>)> {
+    pub(crate) fn take(self) -> anyhow::Result<(Vec<SplitWriterResult>, DeltaLayerWriter)> {
-        Ok((self.generated_layers, self.inner.map(|x| x.1)))
+        Ok((self.generated_layers, self.inner))
     }
 }
 
@@ -440,8 +430,10 @@ mod tests {
             tenant.conf,
             tline.timeline_id,
             tenant.tenant_shard_id,
+            get_key(0),
             Lsn(0x18)..Lsn(0x20),
             4 * 1024 * 1024,
+            &ctx,
         )
         .await
         .unwrap();
@@ -466,22 +458,11 @@ mod tests {
             )
             .await
             .unwrap();
-        let layers = delta_writer.finish(&tline, &ctx).await.unwrap();
+        let layers = delta_writer
+            .finish(&tline, &ctx, get_key(10))
+            .await
+            .unwrap();
         assert_eq!(layers.len(), 1);
-        assert_eq!(
-            layers
-                .into_iter()
-                .next()
-                .unwrap()
-                .into_resident_layer()
-                .layer_desc()
-                .key(),
-            PersistentLayerKey {
-                key_range: get_key(0)..get_key(1),
-                lsn_range: Lsn(0x18)..Lsn(0x20),
-                is_delta: true
-            }
-        );
     }
 
     #[tokio::test]
@@ -518,8 +499,10 @@ mod tests {
             tenant.conf,
             tline.timeline_id,
             tenant.tenant_shard_id,
+            get_key(0),
             Lsn(0x18)..Lsn(0x20),
             4 * 1024 * 1024,
+            &ctx,
         )
         .await
         .unwrap();
@@ -548,7 +531,10 @@ mod tests {
             .finish(&tline, &ctx, get_key(N as u32))
             .await
             .unwrap();
-        let delta_layers = delta_writer.finish(&tline, &ctx).await.unwrap();
+        let delta_layers = delta_writer
+            .finish(&tline, &ctx, get_key(N as u32))
+            .await
+            .unwrap();
         if discard {
             for layer in image_layers {
                 layer.into_discarded_layer();
@@ -567,14 +553,6 @@ mod tests {
                 .collect_vec();
             assert_eq!(image_layers.len(), N / 512 + 1);
             assert_eq!(delta_layers.len(), N / 512 + 1);
-            assert_eq!(
-                delta_layers.first().unwrap().layer_desc().key_range.start,
-                get_key(0)
-            );
-            assert_eq!(
-                delta_layers.last().unwrap().layer_desc().key_range.end,
-                get_key(N as u32)
-            );
             for idx in 0..image_layers.len() {
                 assert_ne!(image_layers[idx].layer_desc().key_range.start, Key::MIN);
                 assert_ne!(image_layers[idx].layer_desc().key_range.end, Key::MAX);
@@ -622,8 +600,10 @@ mod tests {
             tenant.conf,
             tline.timeline_id,
             tenant.tenant_shard_id,
+            get_key(0),
             Lsn(0x18)..Lsn(0x20),
             4 * 1024,
+            &ctx,
         )
         .await
         .unwrap();
@@ -662,35 +642,11 @@ mod tests {
             )
             .await
             .unwrap();
-        let layers = delta_writer.finish(&tline, &ctx).await.unwrap();
+        let layers = delta_writer
+            .finish(&tline, &ctx, get_key(10))
+            .await
+            .unwrap();
         assert_eq!(layers.len(), 2);
-        let mut layers_iter = layers.into_iter();
-        assert_eq!(
-            layers_iter
-                .next()
-                .unwrap()
-                .into_resident_layer()
-                .layer_desc()
-                .key(),
-            PersistentLayerKey {
-                key_range: get_key(0)..get_key(1),
-                lsn_range: Lsn(0x18)..Lsn(0x20),
-                is_delta: true
-            }
-        );
-        assert_eq!(
-            layers_iter
-                .next()
-                .unwrap()
-                .into_resident_layer()
-                .layer_desc()
-                .key(),
-            PersistentLayerKey {
-                key_range: get_key(1)..get_key(2),
-                lsn_range: Lsn(0x18)..Lsn(0x20),
-                is_delta: true
-            }
-        );
     }
 
     #[tokio::test]
@@ -710,8 +666,10 @@ mod tests {
             tenant.conf,
             tline.timeline_id,
             tenant.tenant_shard_id,
+            get_key(0),
             Lsn(0x10)..Lsn(N as u64 * 16 + 0x10),
             4 * 1024 * 1024,
+            &ctx,
         )
         .await
         .unwrap();
@@ -729,20 +687,10 @@ mod tests {
                 .await
                 .unwrap();
         }
-        let delta_layers = delta_writer.finish(&tline, &ctx).await.unwrap();
+        let delta_layers = delta_writer
+            .finish(&tline, &ctx, get_key(N as u32))
+            .await
+            .unwrap();
         assert_eq!(delta_layers.len(), 1);
-        let delta_layer = delta_layers
-            .into_iter()
-            .next()
-            .unwrap()
-            .into_resident_layer();
-        assert_eq!(
-            delta_layer.layer_desc().key(),
-            PersistentLayerKey {
-                key_range: get_key(0)..get_key(1),
-                lsn_range: Lsn(0x10)..Lsn(N as u64 * 16 + 0x10),
-                is_delta: true
-            }
-        );
     }
 }

pageserver/src/tenant/tasks.rs

@@ -10,6 +10,7 @@ use crate::context::{DownloadBehavior, RequestContext};
 use crate::metrics::TENANT_TASK_EVENTS;
 use crate::task_mgr;
 use crate::task_mgr::{TaskKind, BACKGROUND_RUNTIME};
+use crate::tenant::config::defaults::DEFAULT_COMPACTION_PERIOD;
 use crate::tenant::throttle::Stats;
 use crate::tenant::timeline::CompactionError;
 use crate::tenant::{Tenant, TenantState};
@@ -455,11 +456,9 @@ async fn ingest_housekeeping_loop(tenant: Arc<Tenant>, cancel: CancellationToken
 
             // If compaction period is set to zero (to disable it), then we will use a reasonable default
             let period = if period == Duration::ZERO {
-                humantime::Duration::from_str(
+                humantime::Duration::from_str(DEFAULT_COMPACTION_PERIOD)
-                    pageserver_api::config::tenant_conf_defaults::DEFAULT_COMPACTION_PERIOD,
+                    .unwrap()
-                )
+                    .into()
-                .unwrap()
-                .into()
             } else {
                 period
             };

pageserver/src/tenant/timeline.rs

@@ -66,6 +66,7 @@ use std::{
 use crate::{
     aux_file::AuxFileSizeEstimator,
     tenant::{
+        config::defaults::DEFAULT_PITR_INTERVAL,
         layer_map::{LayerMap, SearchResult},
         metadata::TimelineMetadata,
         storage_layer::{inmemory_layer::IndexEntry, PersistentLayerDesc},
@@ -101,7 +102,6 @@ use crate::{
     pgdatadir_mapping::{AuxFilesDirectory, DirectoryKind},
     virtual_file::{MaybeFatalIo, VirtualFile},
 };
-use pageserver_api::config::tenant_conf_defaults::DEFAULT_PITR_INTERVAL;
 
 use crate::config::PageServerConf;
 use crate::keyspace::{KeyPartitioning, KeySpace};
@@ -2243,7 +2243,7 @@ impl Timeline {
             };
 
             if aux_file_policy == Some(AuxFilePolicy::V1) {
-                warn!("this timeline is using deprecated aux file policy V1 (when loading the timeline)");
+                warn!("this timeline is using deprecated aux file policy V1");
             }
 
             result.repartition_threshold =

pageserver/src/tenant/timeline/compaction.rs

@@ -29,6 +29,7 @@ use utils::id::TimelineId;
 
 use crate::context::{AccessStatsBehavior, RequestContext, RequestContextBuilder};
 use crate::page_cache;
+use crate::tenant::config::defaults::{DEFAULT_CHECKPOINT_DISTANCE, DEFAULT_COMPACTION_THRESHOLD};
 use crate::tenant::remote_timeline_client::WaitCompletionError;
 use crate::tenant::storage_layer::merge_iterator::MergeIterator;
 use crate::tenant::storage_layer::split_writer::{
@@ -42,9 +43,6 @@ use crate::tenant::timeline::{drop_rlock, DeltaLayerWriter, ImageLayerWriter};
 use crate::tenant::timeline::{Layer, ResidentLayer};
 use crate::tenant::DeltaLayer;
 use crate::virtual_file::{MaybeFatalIo, VirtualFile};
-use pageserver_api::config::tenant_conf_defaults::{
-    DEFAULT_CHECKPOINT_DISTANCE, DEFAULT_COMPACTION_THRESHOLD,
-};
 
 use crate::keyspace::KeySpace;
 use crate::repository::{Key, Value};
@@ -911,13 +909,137 @@ impl Timeline {
         // we're compacting, in key, LSN order.
         // If there's both a Value::Image and Value::WalRecord for the same (key,lsn),
         // then the Value::Image is ordered before Value::WalRecord.
-        let mut all_values_iter = {
+        //
-            let mut deltas = Vec::with_capacity(deltas_to_compact.len());
+        // TODO(https://github.com/neondatabase/neon/issues/8184): remove the page cached blob_io
-            for l in deltas_to_compact.iter() {
+        // option and validation code once we've reached confidence.
-                let l = l.get_as_delta(ctx).await.map_err(CompactionError::Other)?;
+        enum AllValuesIter<'a> {
-                deltas.push(l);
+            PageCachedBlobIo {
+                all_keys_iter: VecIter<'a>,
+            },
+            StreamingKmergeBypassingPageCache {
+                merge_iter: MergeIterator<'a>,
+            },
+            ValidatingStreamingKmergeBypassingPageCache {
+                mode: CompactL0BypassPageCacheValidation,
+                merge_iter: MergeIterator<'a>,
+                all_keys_iter: VecIter<'a>,
+            },
+        }
+        type VecIter<'a> = std::slice::Iter<'a, DeltaEntry<'a>>; // TODO: distinguished lifetimes
+        impl AllValuesIter<'_> {
+            async fn next_all_keys_iter(
+                iter: &mut VecIter<'_>,
+                ctx: &RequestContext,
+            ) -> anyhow::Result<Option<(Key, Lsn, Value)>> {
+                let Some(DeltaEntry {
+                    key,
+                    lsn,
+                    val: value_ref,
+                    ..
+                }) = iter.next()
+                else {
+                    return Ok(None);
+                };
+                let value = value_ref.load(ctx).await?;
+                Ok(Some((*key, *lsn, value)))
+            }
+            async fn next(
+                &mut self,
+                ctx: &RequestContext,
+            ) -> anyhow::Result<Option<(Key, Lsn, Value)>> {
+                match self {
+                    AllValuesIter::PageCachedBlobIo { all_keys_iter: iter } => {
+                      Self::next_all_keys_iter(iter, ctx).await
+                    }
+                    AllValuesIter::StreamingKmergeBypassingPageCache { merge_iter } => merge_iter.next().await,
+                    AllValuesIter::ValidatingStreamingKmergeBypassingPageCache { mode, merge_iter, all_keys_iter } => async {
+                        // advance both iterators
+                        let all_keys_iter_item = Self::next_all_keys_iter(all_keys_iter, ctx).await;
+                        let merge_iter_item = merge_iter.next().await;
+                        // compare results & log warnings as needed
+                        macro_rules! rate_limited_warn {
+                            ($($arg:tt)*) => {{
+                                if cfg!(debug_assertions) || cfg!(feature = "testing") {
+                                    warn!($($arg)*);
+                                    panic!("CompactL0BypassPageCacheValidation failure, check logs");
+                                }
+                                use once_cell::sync::Lazy;
+                                use utils::rate_limit::RateLimit;
+                                use std::sync::Mutex;
+                                use std::time::Duration;
+                                static LOGGED: Lazy<Mutex<RateLimit>> =
+                                    Lazy::new(|| Mutex::new(RateLimit::new(Duration::from_secs(10))));
+                                let mut rate_limit = LOGGED.lock().unwrap();
+                                rate_limit.call(|| {
+                                    warn!($($arg)*);
+                                });
+                            }}
+                        }
+                        match (&all_keys_iter_item, &merge_iter_item) {
+                            (Err(_), Err(_)) => {
+                                // don't bother asserting equivality of the errors
+                            }
+                            (Err(all_keys), Ok(merge)) => {
+                                rate_limited_warn!(?merge, "all_keys_iter returned an error where merge did not: {all_keys:?}");
+                            },
+                            (Ok(all_keys), Err(merge)) => {
+                                rate_limited_warn!(?all_keys, "merge returned an error where all_keys_iter did not: {merge:?}");
+                            },
+                            (Ok(None), Ok(None)) => { }
+                            (Ok(Some(all_keys)), Ok(None)) => {
+                                rate_limited_warn!(?all_keys, "merge returned None where all_keys_iter returned Some");
+                            }
+                            (Ok(None), Ok(Some(merge))) => {
+                                rate_limited_warn!(?merge, "all_keys_iter returned None where merge returned Some");
+                            }
+                            (Ok(Some((all_keys_key, all_keys_lsn, all_keys_value))), Ok(Some((merge_key, merge_lsn, merge_value)))) => {
+                                match mode {
+                                    // TODO: in this mode, we still load the value from disk for both iterators, even though we only need the all_keys_iter one
+                                    CompactL0BypassPageCacheValidation::KeyLsn => {
+                                        let all_keys = (all_keys_key, all_keys_lsn);
+                                        let merge = (merge_key, merge_lsn);
+                                        if all_keys != merge {
+                                            rate_limited_warn!(?all_keys, ?merge, "merge returned a different (Key,LSN) than all_keys_iter");
+                                        }
+                                    }
+                                    CompactL0BypassPageCacheValidation::KeyLsnValue => {
+                                        let all_keys = (all_keys_key, all_keys_lsn, all_keys_value);
+                                        let merge = (merge_key, merge_lsn, merge_value);
+                                        if all_keys != merge {
+                                            rate_limited_warn!(?all_keys, ?merge, "merge returned a different (Key,LSN,Value) than all_keys_iter");
+                                        }
+                                    }
+                                }
+                            }
+                        }
+                        // in case of mismatch, trust the legacy all_keys_iter_item
+                        all_keys_iter_item
+                    }.instrument(info_span!("next")).await
+                }
+            }
+        }
+        let mut all_values_iter = match &self.conf.compact_level0_phase1_value_access {
+            CompactL0Phase1ValueAccess::PageCachedBlobIo => AllValuesIter::PageCachedBlobIo {
+                all_keys_iter: all_keys.iter(),
+            },
+            CompactL0Phase1ValueAccess::StreamingKmerge { validate } => {
+                let merge_iter = {
+                    let mut deltas = Vec::with_capacity(deltas_to_compact.len());
+                    for l in deltas_to_compact.iter() {
+                        let l = l.get_as_delta(ctx).await.map_err(CompactionError::Other)?;
+                        deltas.push(l);
+                    }
+                    MergeIterator::create(&deltas, &[], ctx)
+                };
+                match validate {
+                    None => AllValuesIter::StreamingKmergeBypassingPageCache { merge_iter },
+                    Some(validate) => AllValuesIter::ValidatingStreamingKmergeBypassingPageCache {
+                        mode: validate.clone(),
+                        merge_iter,
+                        all_keys_iter: all_keys.iter(),
+                    },
+                }
             }
-            MergeIterator::create(&deltas, &[], ctx)
         };
 
         // This iterator walks through all keys and is needed to calculate size used by each key
@@ -994,7 +1116,7 @@ impl Timeline {
         let mut keys = 0;
 
         while let Some((key, lsn, value)) = all_values_iter
-            .next()
+            .next(ctx)
             .await
             .map_err(CompactionError::Other)?
         {
@@ -1311,6 +1433,43 @@ impl TryFrom<CompactLevel0Phase1StatsBuilder> for CompactLevel0Phase1Stats {
     }
 }
 
+#[derive(Debug, PartialEq, Eq, Clone, serde::Deserialize, serde::Serialize)]
+#[serde(tag = "mode", rename_all = "kebab-case", deny_unknown_fields)]
+pub enum CompactL0Phase1ValueAccess {
+    /// The old way.
+    PageCachedBlobIo,
+    /// The new way.
+    StreamingKmerge {
+        /// If set, we run both the old way and the new way, validate that
+        /// they are identical (=> [`CompactL0BypassPageCacheValidation`]),
+        /// and if the validation fails,
+        /// - in tests: fail them with a panic or
+        /// - in prod, log a rate-limited warning and use the old way's results.
+        ///
+        /// If not set, we only run the new way and trust its results.
+        validate: Option<CompactL0BypassPageCacheValidation>,
+    },
+}
+
+/// See [`CompactL0Phase1ValueAccess::StreamingKmerge`].
+#[derive(Debug, PartialEq, Eq, Clone, serde::Deserialize, serde::Serialize)]
+#[serde(rename_all = "kebab-case")]
+pub enum CompactL0BypassPageCacheValidation {
+    /// Validate that the series of (key, lsn) pairs are the same.
+    KeyLsn,
+    /// Validate that the entire output of old and new way is identical.
+    KeyLsnValue,
+}
+
+impl Default for CompactL0Phase1ValueAccess {
+    fn default() -> Self {
+        CompactL0Phase1ValueAccess::StreamingKmerge {
+            // TODO(https://github.com/neondatabase/neon/issues/8184): change to None once confident
+            validate: Some(CompactL0BypassPageCacheValidation::KeyLsnValue),
+        }
+    }
+}
+
 impl Timeline {
     /// Entry point for new tiered compaction algorithm.
     ///
@@ -1809,6 +1968,7 @@ impl Timeline {
             .unwrap();
         // We don't want any of the produced layers to cover the full key range (i.e., MIN..MAX) b/c it will then be recognized
         // as an L0 layer.
+        let hack_end_key = Key::NON_L0_MAX;
         let mut delta_layers = Vec::new();
         let mut image_layers = Vec::new();
         let mut downloaded_layers = Vec::new();
@@ -1854,8 +2014,10 @@ impl Timeline {
             self.conf,
             self.timeline_id,
             self.tenant_shard_id,
+            Key::MIN,
             lowest_retain_lsn..end_lsn,
             self.get_compaction_target_size(),
+            ctx,
         )
         .await?;
 
@@ -1962,7 +2124,7 @@ impl Timeline {
         let produced_image_layers = if let Some(writer) = image_layer_writer {
             if !dry_run {
                 writer
-                    .finish_with_discard_fn(self, ctx, Key::MAX, discard)
+                    .finish_with_discard_fn(self, ctx, hack_end_key, discard)
                     .await?
             } else {
                 let (layers, _) = writer.take()?;
@@ -1975,7 +2137,7 @@ impl Timeline {
 
         let produced_delta_layers = if !dry_run {
             delta_layer_writer
-                .finish_with_discard_fn(self, ctx, discard)
+                .finish_with_discard_fn(self, ctx, hack_end_key, discard)
                 .await?
         } else {
             let (layers, _) = delta_layer_writer.take()?;

pageserver/src/tenant/vectored_blob_io.rs

@@ -16,6 +16,7 @@
 //! Note that the vectored blob api does *not* go through the page cache.
 
 use std::collections::BTreeMap;
+use std::num::NonZeroUsize;
 
 use bytes::BytesMut;
 use pageserver_api::key::Key;
@@ -28,6 +29,9 @@ use crate::context::RequestContext;
 use crate::tenant::blob_io::{BYTE_UNCOMPRESSED, BYTE_ZSTD, LEN_COMPRESSION_BIT_MASK};
 use crate::virtual_file::{self, VirtualFile};
 
+#[derive(Copy, Clone, Debug, PartialEq, Eq)]
+pub struct MaxVectoredReadBytes(pub NonZeroUsize);
+
 /// Metadata bundled with the start and end offset of a blob.
 #[derive(Copy, Clone, Debug)]
 pub struct BlobMeta {
@@ -593,10 +597,8 @@ impl<'a> VectoredBlobReader<'a> {
     }
 }
 
-/// Read planner used in [`crate::tenant::storage_layer::image_layer::ImageLayerIterator`].
+/// Read planner used in [`crate::tenant::storage_layer::image_layer::ImageLayerIterator`]. It provides a streaming API for
-///
+/// getting read blobs. It returns a batch when `handle` gets called and when the current key would just exceed the read_size and
-/// It provides a streaming API for getting read blobs. It returns a batch when
-/// `handle` gets called and when the current key would just exceed the read_size and
 /// max_cnt constraints.
 pub struct StreamingVectoredReadPlanner {
     read_builder: Option<VectoredReadBuilder>,

pageserver/src/virtual_file.rs

@@ -1,7 +1,6 @@
-//! VirtualFile is like a normal File, but it's not bound directly to
-//! a file descriptor.
 //!
-//! Instead, the file is opened when it's read from,
+//! VirtualFile is like a normal File, but it's not bound directly to
+//! a file descriptor. Instead, the file is opened when it's read from,
 //! and if too many files are open globally in the system, least-recently
 //! used ones are closed.
 //!
@@ -11,6 +10,7 @@
 //! This is similar to PostgreSQL's virtual file descriptor facility in
 //! src/backend/storage/file/fd.c
 //!
+use crate::config::defaults::DEFAULT_IO_BUFFER_ALIGNMENT;
 use crate::context::RequestContext;
 use crate::metrics::{StorageIoOperation, STORAGE_IO_SIZE, STORAGE_IO_TIME_METRIC};
 
@@ -19,7 +19,6 @@ use crate::tenant::TENANTS_SEGMENT_NAME;
 use camino::{Utf8Path, Utf8PathBuf};
 use once_cell::sync::OnceCell;
 use owned_buffers_io::io_buf_ext::FullSlice;
-use pageserver_api::config::defaults::DEFAULT_IO_BUFFER_ALIGNMENT;
 use pageserver_api::shard::TenantShardId;
 use std::fs::File;
 use std::io::{Error, ErrorKind, Seek, SeekFrom};

pageserver/src/virtual_file/io_engine.rs

@@ -84,14 +84,9 @@ pub(crate) fn get() -> IoEngine {
                         }
                     },
                     Err(std::env::VarError::NotPresent) => {
-                        #[cfg(target_os = "linux")]
+                        crate::config::defaults::DEFAULT_VIRTUAL_FILE_IO_ENGINE
-                        {
+                            .parse()
-                            IoEngineKind::TokioEpollUring
+                            .unwrap()
-                        }
-                        #[cfg(not(target_os = "linux"))]
-                        {
-                            IoEngineKind::StdFs
-                        }
                     }
                     Err(std::env::VarError::NotUnicode(_)) => {
                         panic!("env var {env_var_name} is not unicode");

pageserver/src/walingest.rs

@@ -25,7 +25,9 @@ use std::time::Duration;
 use std::time::SystemTime;
 
 use pageserver_api::shard::ShardIdentity;
-use postgres_ffi::{dispatch_pgversion, enum_pgversion, enum_pgversion_dispatch, TimestampTz};
+use postgres_ffi::v14::nonrelfile_utils::clogpage_precedes;
+use postgres_ffi::v14::nonrelfile_utils::slru_may_delete_clogsegment;
+use postgres_ffi::TimestampTz;
 use postgres_ffi::{fsm_logical_to_physical, page_is_new, page_set_lsn};
 
 use anyhow::{bail, Context, Result};
@@ -46,31 +48,16 @@ use pageserver_api::key::rel_block_to_key;
 use pageserver_api::reltag::{BlockNumber, RelTag, SlruKind};
 use postgres_ffi::pg_constants;
 use postgres_ffi::relfile_utils::{FSM_FORKNUM, INIT_FORKNUM, MAIN_FORKNUM, VISIBILITYMAP_FORKNUM};
+use postgres_ffi::v14::nonrelfile_utils::mx_offset_to_member_segment;
+use postgres_ffi::v14::xlog_utils::*;
+use postgres_ffi::v14::CheckPoint;
 use postgres_ffi::TransactionId;
 use postgres_ffi::BLCKSZ;
-use utils::bin_ser::SerializeError;
 use utils::lsn::Lsn;
 
-enum_pgversion! {CheckPoint, pgv::CheckPoint}
-
-impl CheckPoint {
-    fn encode(&self) -> Result<Bytes, SerializeError> {
-        enum_pgversion_dispatch!(self, CheckPoint, cp, { cp.encode() })
-    }
-
-    fn update_next_xid(&mut self, xid: u32) -> bool {
-        enum_pgversion_dispatch!(self, CheckPoint, cp, { cp.update_next_xid(xid) })
-    }
-
-    pub fn update_next_multixid(&mut self, multi_xid: u32, multi_offset: u32) -> bool {
-        enum_pgversion_dispatch!(self, CheckPoint, cp, {
-            cp.update_next_multixid(multi_xid, multi_offset)
-        })
-    }
-}
-
 pub struct WalIngest {
     shard: ShardIdentity,
+    pg_version: u32,
     checkpoint: CheckPoint,
     checkpoint_modified: bool,
     warn_ingest_lag: WarnIngestLag,
@@ -91,16 +78,12 @@ impl WalIngest {
         // Fetch the latest checkpoint into memory, so that we can compare with it
         // quickly in `ingest_record` and update it when it changes.
         let checkpoint_bytes = timeline.get_checkpoint(startpoint, ctx).await?;
-        let pgversion = timeline.pg_version;
+        let checkpoint = CheckPoint::decode(&checkpoint_bytes)?;
-
+        trace!("CheckPoint.nextXid = {}", checkpoint.nextXid.value);
-        let checkpoint = dispatch_pgversion!(pgversion, {
-            let checkpoint = pgv::CheckPoint::decode(&checkpoint_bytes)?;
-            trace!("CheckPoint.nextXid = {}", checkpoint.nextXid.value);
-            <pgv::CheckPoint as Into<CheckPoint>>::into(checkpoint)
-        });
 
         Ok(WalIngest {
             shard: *timeline.get_shard_identity(),
+            pg_version: timeline.pg_version,
             checkpoint,
             checkpoint_modified: false,
             warn_ingest_lag: WarnIngestLag {
@@ -134,7 +117,7 @@ impl WalIngest {
 
         modification.set_lsn(lsn)?;
 
-        if decoded.is_dbase_create_copy(pg_version) {
+        if decoded.is_dbase_create_copy(self.pg_version) {
             // Records of this type should always be preceded by a commit(), as they
             // rely on reading data pages back from the Timeline.
             assert!(!modification.has_dirty_data_pages());
@@ -354,67 +337,70 @@ impl WalIngest {
             pg_constants::RM_XLOG_ID => {
                 let info = decoded.xl_info & pg_constants::XLR_RMGR_INFO_MASK;
 
-                enum_pgversion_dispatch!(&mut self.checkpoint, CheckPoint, cp, {
+                if info == pg_constants::XLOG_NEXTOID {
-                    if info == pg_constants::XLOG_NEXTOID {
+                    let next_oid = buf.get_u32_le();
-                        let next_oid = buf.get_u32_le();
+                    if self.checkpoint.nextOid != next_oid {
-                        if cp.nextOid != next_oid {
+                        self.checkpoint.nextOid = next_oid;
-                            cp.nextOid = next_oid;
-                            self.checkpoint_modified = true;
-                        }
-                    } else if info == pg_constants::XLOG_CHECKPOINT_ONLINE
-                        || info == pg_constants::XLOG_CHECKPOINT_SHUTDOWN
-                    {
-                        let mut checkpoint_bytes = [0u8; pgv::xlog_utils::SIZEOF_CHECKPOINT];
-                        buf.copy_to_slice(&mut checkpoint_bytes);
-                        let xlog_checkpoint = pgv::CheckPoint::decode(&checkpoint_bytes)?;
-                        trace!(
-                            "xlog_checkpoint.oldestXid={}, checkpoint.oldestXid={}",
-                            xlog_checkpoint.oldestXid,
-                            cp.oldestXid
-                        );
-                        if (cp.oldestXid.wrapping_sub(xlog_checkpoint.oldestXid) as i32) < 0 {
-                            cp.oldestXid = xlog_checkpoint.oldestXid;
-                        }
-                        trace!(
-                            "xlog_checkpoint.oldestActiveXid={}, checkpoint.oldestActiveXid={}",
-                            xlog_checkpoint.oldestActiveXid,
-                            cp.oldestActiveXid
-                        );
-
-                        // A shutdown checkpoint has `oldestActiveXid == InvalidTransactionid`,
-                        // because at shutdown, all in-progress transactions will implicitly
-                        // end. Postgres startup code knows that, and allows hot standby to start
-                        // immediately from a shutdown checkpoint.
-                        //
-                        // In Neon, Postgres hot standby startup always behaves as if starting from
-                        // an online checkpoint. It needs a valid `oldestActiveXid` value, so
-                        // instead of overwriting self.checkpoint.oldestActiveXid with
-                        // InvalidTransactionid from the checkpoint WAL record, update it to a
-                        // proper value, knowing that there are no in-progress transactions at this
-                        // point, except for prepared transactions.
-                        //
-                        // See also the neon code changes in the InitWalRecovery() function.
-                        if xlog_checkpoint.oldestActiveXid == pg_constants::INVALID_TRANSACTION_ID
-                            && info == pg_constants::XLOG_CHECKPOINT_SHUTDOWN
-                        {
-                            let mut oldest_active_xid = cp.nextXid.value as u32;
-                            for xid in modification.tline.list_twophase_files(lsn, ctx).await? {
-                                if (xid.wrapping_sub(oldest_active_xid) as i32) < 0 {
-                                    oldest_active_xid = xid;
-                                }
-                            }
-                            cp.oldestActiveXid = oldest_active_xid;
-                        } else {
-                            cp.oldestActiveXid = xlog_checkpoint.oldestActiveXid;
-                        }
-
-                        // Write a new checkpoint key-value pair on every checkpoint record, even
-                        // if nothing really changed. Not strictly required, but it seems nice to
-                        // have some trace of the checkpoint records in the layer files at the same
-                        // LSNs.
                         self.checkpoint_modified = true;
                     }
-                });
+                } else if info == pg_constants::XLOG_CHECKPOINT_ONLINE
+                    || info == pg_constants::XLOG_CHECKPOINT_SHUTDOWN
+                {
+                    let mut checkpoint_bytes = [0u8; SIZEOF_CHECKPOINT];
+                    buf.copy_to_slice(&mut checkpoint_bytes);
+                    let xlog_checkpoint = CheckPoint::decode(&checkpoint_bytes)?;
+                    trace!(
+                        "xlog_checkpoint.oldestXid={}, checkpoint.oldestXid={}",
+                        xlog_checkpoint.oldestXid,
+                        self.checkpoint.oldestXid
+                    );
+                    if (self
+                        .checkpoint
+                        .oldestXid
+                        .wrapping_sub(xlog_checkpoint.oldestXid) as i32)
+                        < 0
+                    {
+                        self.checkpoint.oldestXid = xlog_checkpoint.oldestXid;
+                    }
+                    trace!(
+                        "xlog_checkpoint.oldestActiveXid={}, checkpoint.oldestActiveXid={}",
+                        xlog_checkpoint.oldestActiveXid,
+                        self.checkpoint.oldestActiveXid
+                    );
+
+                    // A shutdown checkpoint has `oldestActiveXid == InvalidTransactionid`,
+                    // because at shutdown, all in-progress transactions will implicitly
+                    // end. Postgres startup code knows that, and allows hot standby to start
+                    // immediately from a shutdown checkpoint.
+                    //
+                    // In Neon, Postgres hot standby startup always behaves as if starting from
+                    // an online checkpoint. It needs a valid `oldestActiveXid` value, so
+                    // instead of overwriting self.checkpoint.oldestActiveXid with
+                    // InvalidTransactionid from the checkpoint WAL record, update it to a
+                    // proper value, knowing that there are no in-progress transactions at this
+                    // point, except for prepared transactions.
+                    //
+                    // See also the neon code changes in the InitWalRecovery() function.
+                    if xlog_checkpoint.oldestActiveXid == pg_constants::INVALID_TRANSACTION_ID
+                        && info == pg_constants::XLOG_CHECKPOINT_SHUTDOWN
+                    {
+                        let mut oldest_active_xid = self.checkpoint.nextXid.value as u32;
+                        for xid in modification.tline.list_twophase_files(lsn, ctx).await? {
+                            if (xid.wrapping_sub(oldest_active_xid) as i32) < 0 {
+                                oldest_active_xid = xid;
+                            }
+                        }
+                        self.checkpoint.oldestActiveXid = oldest_active_xid;
+                    } else {
+                        self.checkpoint.oldestActiveXid = xlog_checkpoint.oldestActiveXid;
+                    }
+
+                    // Write a new checkpoint key-value pair on every checkpoint record, even
+                    // if nothing really changed. Not strictly required, but it seems nice to
+                    // have some trace of the checkpoint records in the layer files at the same
+                    // LSNs.
+                    self.checkpoint_modified = true;
+                }
             }
             pg_constants::RM_LOGICALMSG_ID => {
                 let info = decoded.xl_info & pg_constants::XLR_RMGR_INFO_MASK;
@@ -438,11 +424,7 @@ impl WalIngest {
                 let info = decoded.xl_info & pg_constants::XLR_RMGR_INFO_MASK;
                 if info == pg_constants::XLOG_RUNNING_XACTS {
                     let xlrec = crate::walrecord::XlRunningXacts::decode(&mut buf);
-
+                    self.checkpoint.oldestActiveXid = xlrec.oldest_running_xid;
-                    enum_pgversion_dispatch!(&mut self.checkpoint, CheckPoint, cp, {
-                        cp.oldestActiveXid = xlrec.oldest_running_xid;
-                    });
-
                     self.checkpoint_modified = true;
                 }
             }
@@ -557,7 +539,7 @@ impl WalIngest {
             && blk.has_image
             && decoded.xl_rmid == pg_constants::RM_XLOG_ID
             && (decoded.xl_info == pg_constants::XLOG_FPI
-            || decoded.xl_info == pg_constants::XLOG_FPI_FOR_HINT)
+                || decoded.xl_info == pg_constants::XLOG_FPI_FOR_HINT)
             // compression of WAL is not yet supported: fall back to storing the original WAL record
             && !postgres_ffi::bkpimage_is_compressed(blk.bimg_info, modification.tline.pg_version)
             // do not materialize null pages because them most likely be soon replaced with real data
@@ -1222,7 +1204,7 @@ impl WalIngest {
             if rec.blkno % pg_constants::SLOTS_PER_FSM_PAGE != 0 {
                 // Tail of last remaining FSM page has to be zeroed.
                 // We are not precise here and instead of digging in FSM bitmap format just clear the whole page.
-                modification.put_rel_page_image_zero(rel, fsm_physical_page_no)?;
+                modification.put_rel_page_image_zero(rel, fsm_physical_page_no);
                 fsm_physical_page_no += 1;
             }
             let nblocks = get_relsize(modification, rel, ctx).await?;
@@ -1244,7 +1226,7 @@ impl WalIngest {
             if rec.blkno % pg_constants::VM_HEAPBLOCKS_PER_PAGE != 0 {
                 // Tail of last remaining vm page has to be zeroed.
                 // We are not precise here and instead of digging in VM bitmap format just clear the whole page.
-                modification.put_rel_page_image_zero(rel, vm_page_no)?;
+                modification.put_rel_page_image_zero(rel, vm_page_no);
                 vm_page_no += 1;
             }
             let nblocks = get_relsize(modification, rel, ctx).await?;
@@ -1260,17 +1242,12 @@ impl WalIngest {
     fn warn_on_ingest_lag(
         &mut self,
         conf: &crate::config::PageServerConf,
-        wal_timestamp: TimestampTz,
+        wal_timestmap: TimestampTz,
     ) {
         debug_assert_current_span_has_tenant_and_timeline_id();
         let now = SystemTime::now();
         let rate_limits = &mut self.warn_ingest_lag;
-
+        match try_from_pg_timestamp(wal_timestmap) {
-        let ts = enum_pgversion_dispatch!(&self.checkpoint, CheckPoint, _cp, {
-            pgv::xlog_utils::try_from_pg_timestamp(wal_timestamp)
-        });
-
-        match ts {
             Ok(ts) => {
                 match now.duration_since(ts) {
                     Ok(lag) => {
@@ -1280,7 +1257,7 @@ impl WalIngest {
                                 warn!(%rate_limit_stats, %lag, "ingesting record with timestamp lagging more than wait_lsn_timeout");
                             })
                         }
-                    }
+                    },
                     Err(e) => {
                         let delta_t = e.duration();
                         // determined by prod victoriametrics query: 1000 * (timestamp(node_time_seconds{neon_service="pageserver"}) - node_time_seconds)
@@ -1294,6 +1271,7 @@ impl WalIngest {
                         }
                     }
                 };
+
             }
             Err(error) => {
                 rate_limits.timestamp_invalid_msg_ratelimit.call2(|rate_limit_stats| {
@@ -1401,17 +1379,14 @@ impl WalIngest {
         // truncated, but a checkpoint record with the updated values isn't written until
         // later. In Neon, a server can start at any LSN, not just on a checkpoint record,
         // so we keep the oldestXid and oldestXidDB up-to-date.
-        enum_pgversion_dispatch!(&mut self.checkpoint, CheckPoint, cp, {
+        self.checkpoint.oldestXid = xlrec.oldest_xid;
-            cp.oldestXid = xlrec.oldest_xid;
+        self.checkpoint.oldestXidDB = xlrec.oldest_xid_db;
-            cp.oldestXidDB = xlrec.oldest_xid_db;
-        });
         self.checkpoint_modified = true;
 
         // TODO Treat AdvanceOldestClogXid() or write a comment why we don't need it
 
         let latest_page_number =
-            enum_pgversion_dispatch!(self.checkpoint, CheckPoint, cp, { cp.nextXid.value }) as u32
+            self.checkpoint.nextXid.value as u32 / pg_constants::CLOG_XACTS_PER_PAGE;
-                / pg_constants::CLOG_XACTS_PER_PAGE;
 
         // Now delete all segments containing pages between xlrec.pageno
         // and latest_page_number.
@@ -1419,9 +1394,7 @@ impl WalIngest {
         // First, make an important safety check:
         // the current endpoint page must not be eligible for removal.
         // See SimpleLruTruncate() in slru.c
-        if dispatch_pgversion!(modification.tline.pg_version, {
+        if clogpage_precedes(latest_page_number, xlrec.pageno) {
-            pgv::nonrelfile_utils::clogpage_precedes(latest_page_number, xlrec.pageno)
-        }) {
             info!("could not truncate directory pg_xact apparent wraparound");
             return Ok(());
         }
@@ -1438,12 +1411,7 @@ impl WalIngest {
             .await?
         {
             let segpage = segno * pg_constants::SLRU_PAGES_PER_SEGMENT;
-
+            if slru_may_delete_clogsegment(segpage, xlrec.pageno) {
-            let may_delete = dispatch_pgversion!(modification.tline.pg_version, {
-                pgv::nonrelfile_utils::slru_may_delete_clogsegment(segpage, xlrec.pageno)
-            });
-
-            if may_delete {
                 modification
                     .drop_slru_segment(SlruKind::Clog, segno, ctx)
                     .await?;
@@ -1562,23 +1530,14 @@ impl WalIngest {
         xlrec: &XlMultiXactTruncate,
         ctx: &RequestContext,
     ) -> Result<()> {
-        let (maxsegment, startsegment, endsegment) =
+        self.checkpoint.oldestMulti = xlrec.end_trunc_off;
-            enum_pgversion_dispatch!(&mut self.checkpoint, CheckPoint, cp, {
+        self.checkpoint.oldestMultiDB = xlrec.oldest_multi_db;
-                cp.oldestMulti = xlrec.end_trunc_off;
-                cp.oldestMultiDB = xlrec.oldest_multi_db;
-                let maxsegment: i32 = pgv::nonrelfile_utils::mx_offset_to_member_segment(
-                    pg_constants::MAX_MULTIXACT_OFFSET,
-                );
-                let startsegment: i32 =
-                    pgv::nonrelfile_utils::mx_offset_to_member_segment(xlrec.start_trunc_memb);
-                let endsegment: i32 =
-                    pgv::nonrelfile_utils::mx_offset_to_member_segment(xlrec.end_trunc_memb);
-                (maxsegment, startsegment, endsegment)
-            });
-
         self.checkpoint_modified = true;
 
         // PerformMembersTruncation
+        let maxsegment: i32 = mx_offset_to_member_segment(pg_constants::MAX_MULTIXACT_OFFSET);
+        let startsegment: i32 = mx_offset_to_member_segment(xlrec.start_trunc_memb);
+        let endsegment: i32 = mx_offset_to_member_segment(xlrec.end_trunc_memb);
         let mut segment: i32 = startsegment;
 
         // Delete all the segments except the last one. The last segment can still
@@ -1737,7 +1696,7 @@ impl WalIngest {
                     continue;
                 }
 
-                modification.put_rel_page_image_zero(rel, gap_blknum)?;
+                modification.put_rel_page_image_zero(rel, gap_blknum);
             }
         }
         Ok(())
@@ -1803,7 +1762,7 @@ impl WalIngest {
 
             // fill the gap with zeros
             for gap_blknum in old_nblocks..blknum {
-                modification.put_slru_page_image_zero(kind, segno, gap_blknum)?;
+                modification.put_slru_page_image_zero(kind, segno, gap_blknum);
             }
         }
         Ok(())
@@ -1852,23 +1811,11 @@ mod tests {
         // TODO
     }
 
-    #[tokio::test]
+    static ZERO_CHECKPOINT: Bytes = Bytes::from_static(&[0u8; SIZEOF_CHECKPOINT]);
-    async fn test_zeroed_checkpoint_decodes_correctly() -> Result<()> {
-        for i in 14..=16 {
-            dispatch_pgversion!(i, {
-                pgv::CheckPoint::decode(&pgv::ZERO_CHECKPOINT)?;
-            });
-        }
-
-        Ok(())
-    }
 
     async fn init_walingest_test(tline: &Timeline, ctx: &RequestContext) -> Result<WalIngest> {
         let mut m = tline.begin_modification(Lsn(0x10));
-        m.put_checkpoint(dispatch_pgversion!(
+        m.put_checkpoint(ZERO_CHECKPOINT.clone())?;
-            tline.pg_version,
-            pgv::ZERO_CHECKPOINT.clone()
-        ))?;
         m.put_relmap_file(0, 111, Bytes::from(""), ctx).await?; // dummy relmapper file
         m.commit(ctx).await?;
         let walingest = WalIngest::new(tline, Lsn(0x10), ctx).await?;

pageserver/src/walredo.rs

@@ -43,12 +43,13 @@ use utils::lsn::Lsn;
 use utils::sync::gate::GateError;
 use utils::sync::heavier_once_cell;
 
-/// The real implementation that uses a Postgres process to
-/// perform WAL replay.
 ///
-/// Only one thread can use the process at a time, that is controlled by the
+/// This is the real implementation that uses a Postgres process to
-/// Mutex. In the future, we might want to launch a pool of processes to allow
+/// perform WAL replay. Only one thread can use the process at a time,
-/// concurrent replay of multiple records.
+/// that is controlled by the Mutex. In the future, we might want to
+/// launch a pool of processes to allow concurrent replay of multiple
+/// records.
+///
 pub struct PostgresRedoManager {
     tenant_shard_id: TenantShardId,
     conf: &'static PageServerConf,

pgxn/neon/walproposer.c

@@ -1038,12 +1038,9 @@ DetermineEpochStartLsn(WalProposer *wp)
 		if (SkipXLogPageHeader(wp, wp->propEpochStartLsn) != wp->api.get_redo_start_lsn(wp))
 		{
 			/*
-			 * However, allow to proceed if last_log_term on the node which gave
+			 * However, allow to proceed if previously elected leader was me;
-			 * the highest vote (i.e. point where we are going to start writing)
+			 * plain restart of walproposer not intervened by concurrent
-			 * actually had been won by me; plain restart of walproposer not
+			 * compute (who could generate WAL) is ok.
-			 * intervened by concurrent compute which wrote WAL is ok.
-			 *
-			 * This avoids compute crash after manual term_bump.
 			 */
 			if (!((dth->n_entries >= 1) && (dth->entries[dth->n_entries - 1].term ==
 											pg_atomic_read_u64(&walprop_shared->mineLastElectedTerm))))
@@ -1445,17 +1442,12 @@ RecvAppendResponses(Safekeeper *sk)
 		if (sk->appendResponse.term > wp->propTerm)
 		{
 			/*
-			 *
+			 * Another compute with higher term is running. Panic to restart
-			 * Term has changed to higher one, probably another compute is
+			 * PG as we likely need to retake basebackup. However, don't dump
-			 * running. If this is the case we could PANIC as well because
+			 * core as this is kinda expected scenario.
-			 * likely it inserted some data and our basebackup is unsuitable
-			 * anymore. However, we also bump term manually (term_bump endpoint)
-			 * on safekeepers for migration purposes, in this case we do want
-			 * compute to stay alive. So restart walproposer with FATAL instead
-			 * of panicking; if basebackup is spoiled next election will notice
-			 * this.
 			 */
-			wp_log(FATAL, "WAL acceptor %s:%s with term " INT64_FORMAT " rejected our request, our term " INT64_FORMAT ", meaning another compute is running at the same time, and it conflicts with us",
+			disable_core_dump();
+			wp_log(PANIC, "WAL acceptor %s:%s with term " INT64_FORMAT " rejected our request, our term " INT64_FORMAT ", meaning another compute is running at the same time, and it conflicts with us",
 				   sk->host, sk->port,
 				   sk->appendResponse.term, wp->propTerm);
 		}

poetry.lock

@@ -985,38 +985,43 @@ files = [
 
 [[package]]
 name = "cryptography"
-version = "43.0.1"
+version = "42.0.4"
 description = "cryptography is a package which provides cryptographic recipes and primitives to Python developers."
 optional = false
 python-versions = ">=3.7"
 files = [
-    {file = "cryptography-43.0.1-cp37-abi3-macosx_10_9_universal2.whl", hash = "sha256:8385d98f6a3bf8bb2d65a73e17ed87a3ba84f6991c155691c51112075f9ffc5d"},
+    {file = "cryptography-42.0.4-cp37-abi3-macosx_10_12_universal2.whl", hash = "sha256:ffc73996c4fca3d2b6c1c8c12bfd3ad00def8621da24f547626bf06441400449"},
-    {file = "cryptography-43.0.1-cp37-abi3-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:27e613d7077ac613e399270253259d9d53872aaf657471473ebfc9a52935c062"},
+    {file = "cryptography-42.0.4-cp37-abi3-macosx_10_12_x86_64.whl", hash = "sha256:db4b65b02f59035037fde0998974d84244a64c3265bdef32a827ab9b63d61b18"},
-    {file = "cryptography-43.0.1-cp37-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:68aaecc4178e90719e95298515979814bda0cbada1256a4485414860bd7ab962"},
+    {file = "cryptography-42.0.4-cp37-abi3-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:dad9c385ba8ee025bb0d856714f71d7840020fe176ae0229de618f14dae7a6e2"},
-    {file = "cryptography-43.0.1-cp37-abi3-manylinux_2_28_aarch64.whl", hash = "sha256:de41fd81a41e53267cb020bb3a7212861da53a7d39f863585d13ea11049cf277"},
+    {file = "cryptography-42.0.4-cp37-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:69b22ab6506a3fe483d67d1ed878e1602bdd5912a134e6202c1ec672233241c1"},
-    {file = "cryptography-43.0.1-cp37-abi3-manylinux_2_28_x86_64.whl", hash = "sha256:f98bf604c82c416bc829e490c700ca1553eafdf2912a91e23a79d97d9801372a"},
+    {file = "cryptography-42.0.4-cp37-abi3-manylinux_2_28_aarch64.whl", hash = "sha256:e09469a2cec88fb7b078e16d4adec594414397e8879a4341c6ace96013463d5b"},
-    {file = "cryptography-43.0.1-cp37-abi3-musllinux_1_2_aarch64.whl", hash = "sha256:61ec41068b7b74268fa86e3e9e12b9f0c21fcf65434571dbb13d954bceb08042"},
+    {file = "cryptography-42.0.4-cp37-abi3-manylinux_2_28_x86_64.whl", hash = "sha256:3e970a2119507d0b104f0a8e281521ad28fc26f2820687b3436b8c9a5fcf20d1"},
-    {file = "cryptography-43.0.1-cp37-abi3-musllinux_1_2_x86_64.whl", hash = "sha256:014f58110f53237ace6a408b5beb6c427b64e084eb451ef25a28308270086494"},
+    {file = "cryptography-42.0.4-cp37-abi3-musllinux_1_1_aarch64.whl", hash = "sha256:e53dc41cda40b248ebc40b83b31516487f7db95ab8ceac1f042626bc43a2f992"},
-    {file = "cryptography-43.0.1-cp37-abi3-win32.whl", hash = "sha256:2bd51274dcd59f09dd952afb696bf9c61a7a49dfc764c04dd33ef7a6b502a1e2"},
+    {file = "cryptography-42.0.4-cp37-abi3-musllinux_1_1_x86_64.whl", hash = "sha256:c3a5cbc620e1e17009f30dd34cb0d85c987afd21c41a74352d1719be33380885"},
-    {file = "cryptography-43.0.1-cp37-abi3-win_amd64.whl", hash = "sha256:666ae11966643886c2987b3b721899d250855718d6d9ce41b521252a17985f4d"},
+    {file = "cryptography-42.0.4-cp37-abi3-musllinux_1_2_aarch64.whl", hash = "sha256:6bfadd884e7280df24d26f2186e4e07556a05d37393b0f220a840b083dc6a824"},
-    {file = "cryptography-43.0.1-cp39-abi3-macosx_10_9_universal2.whl", hash = "sha256:ac119bb76b9faa00f48128b7f5679e1d8d437365c5d26f1c2c3f0da4ce1b553d"},
+    {file = "cryptography-42.0.4-cp37-abi3-musllinux_1_2_x86_64.whl", hash = "sha256:01911714117642a3f1792c7f376db572aadadbafcd8d75bb527166009c9f1d1b"},
-    {file = "cryptography-43.0.1-cp39-abi3-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:1bbcce1a551e262dfbafb6e6252f1ae36a248e615ca44ba302df077a846a8806"},
+    {file = "cryptography-42.0.4-cp37-abi3-win32.whl", hash = "sha256:fb0cef872d8193e487fc6bdb08559c3aa41b659a7d9be48b2e10747f47863925"},
-    {file = "cryptography-43.0.1-cp39-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:58d4e9129985185a06d849aa6df265bdd5a74ca6e1b736a77959b498e0505b85"},
+    {file = "cryptography-42.0.4-cp37-abi3-win_amd64.whl", hash = "sha256:c1f25b252d2c87088abc8bbc4f1ecbf7c919e05508a7e8628e6875c40bc70923"},
-    {file = "cryptography-43.0.1-cp39-abi3-manylinux_2_28_aarch64.whl", hash = "sha256:d03a475165f3134f773d1388aeb19c2d25ba88b6a9733c5c590b9ff7bbfa2e0c"},
+    {file = "cryptography-42.0.4-cp39-abi3-macosx_10_12_universal2.whl", hash = "sha256:15a1fb843c48b4a604663fa30af60818cd28f895572386e5f9b8a665874c26e7"},
-    {file = "cryptography-43.0.1-cp39-abi3-manylinux_2_28_x86_64.whl", hash = "sha256:511f4273808ab590912a93ddb4e3914dfd8a388fed883361b02dea3791f292e1"},
+    {file = "cryptography-42.0.4-cp39-abi3-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:a1327f280c824ff7885bdeef8578f74690e9079267c1c8bd7dc5cc5aa065ae52"},
-    {file = "cryptography-43.0.1-cp39-abi3-musllinux_1_2_aarch64.whl", hash = "sha256:80eda8b3e173f0f247f711eef62be51b599b5d425c429b5d4ca6a05e9e856baa"},
+    {file = "cryptography-42.0.4-cp39-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6ffb03d419edcab93b4b19c22ee80c007fb2d708429cecebf1dd3258956a563a"},
-    {file = "cryptography-43.0.1-cp39-abi3-musllinux_1_2_x86_64.whl", hash = "sha256:38926c50cff6f533f8a2dae3d7f19541432610d114a70808f0926d5aaa7121e4"},
+    {file = "cryptography-42.0.4-cp39-abi3-manylinux_2_28_aarch64.whl", hash = "sha256:1df6fcbf60560d2113b5ed90f072dc0b108d64750d4cbd46a21ec882c7aefce9"},
-    {file = "cryptography-43.0.1-cp39-abi3-win32.whl", hash = "sha256:a575913fb06e05e6b4b814d7f7468c2c660e8bb16d8d5a1faf9b33ccc569dd47"},
+    {file = "cryptography-42.0.4-cp39-abi3-manylinux_2_28_x86_64.whl", hash = "sha256:44a64043f743485925d3bcac548d05df0f9bb445c5fcca6681889c7c3ab12764"},
-    {file = "cryptography-43.0.1-cp39-abi3-win_amd64.whl", hash = "sha256:d75601ad10b059ec832e78823b348bfa1a59f6b8d545db3a24fd44362a1564cb"},
+    {file = "cryptography-42.0.4-cp39-abi3-musllinux_1_1_aarch64.whl", hash = "sha256:3c6048f217533d89f2f8f4f0fe3044bf0b2090453b7b73d0b77db47b80af8dff"},
-    {file = "cryptography-43.0.1-pp310-pypy310_pp73-macosx_10_9_x86_64.whl", hash = "sha256:ea25acb556320250756e53f9e20a4177515f012c9eaea17eb7587a8c4d8ae034"},
+    {file = "cryptography-42.0.4-cp39-abi3-musllinux_1_1_x86_64.whl", hash = "sha256:6d0fbe73728c44ca3a241eff9aefe6496ab2656d6e7a4ea2459865f2e8613257"},
-    {file = "cryptography-43.0.1-pp310-pypy310_pp73-manylinux_2_28_aarch64.whl", hash = "sha256:c1332724be35d23a854994ff0b66530119500b6053d0bd3363265f7e5e77288d"},
+    {file = "cryptography-42.0.4-cp39-abi3-musllinux_1_2_aarch64.whl", hash = "sha256:887623fe0d70f48ab3f5e4dbf234986b1329a64c066d719432d0698522749929"},
-    {file = "cryptography-43.0.1-pp310-pypy310_pp73-manylinux_2_28_x86_64.whl", hash = "sha256:fba1007b3ef89946dbbb515aeeb41e30203b004f0b4b00e5e16078b518563289"},
+    {file = "cryptography-42.0.4-cp39-abi3-musllinux_1_2_x86_64.whl", hash = "sha256:ce8613beaffc7c14f091497346ef117c1798c202b01153a8cc7b8e2ebaaf41c0"},
-    {file = "cryptography-43.0.1-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:5b43d1ea6b378b54a1dc99dd8a2b5be47658fe9a7ce0a58ff0b55f4b43ef2b84"},
+    {file = "cryptography-42.0.4-cp39-abi3-win32.whl", hash = "sha256:810bcf151caefc03e51a3d61e53335cd5c7316c0a105cc695f0959f2c638b129"},
-    {file = "cryptography-43.0.1-pp39-pypy39_pp73-macosx_10_9_x86_64.whl", hash = "sha256:88cce104c36870d70c49c7c8fd22885875d950d9ee6ab54df2745f83ba0dc365"},
+    {file = "cryptography-42.0.4-cp39-abi3-win_amd64.whl", hash = "sha256:a0298bdc6e98ca21382afe914c642620370ce0470a01e1bef6dd9b5354c36854"},
-    {file = "cryptography-43.0.1-pp39-pypy39_pp73-manylinux_2_28_aarch64.whl", hash = "sha256:9d3cdb25fa98afdd3d0892d132b8d7139e2c087da1712041f6b762e4f807cc96"},
+    {file = "cryptography-42.0.4-pp310-pypy310_pp73-macosx_10_12_x86_64.whl", hash = "sha256:5f8907fcf57392cd917892ae83708761c6ff3c37a8e835d7246ff0ad251d9298"},
-    {file = "cryptography-43.0.1-pp39-pypy39_pp73-manylinux_2_28_x86_64.whl", hash = "sha256:e710bf40870f4db63c3d7d929aa9e09e4e7ee219e703f949ec4073b4294f6172"},
+    {file = "cryptography-42.0.4-pp310-pypy310_pp73-manylinux_2_28_aarch64.whl", hash = "sha256:12d341bd42cdb7d4937b0cabbdf2a94f949413ac4504904d0cdbdce4a22cbf88"},
-    {file = "cryptography-43.0.1-pp39-pypy39_pp73-win_amd64.whl", hash = "sha256:7c05650fe8023c5ed0d46793d4b7d7e6cd9c04e68eabe5b0aeea836e37bdcec2"},
+    {file = "cryptography-42.0.4-pp310-pypy310_pp73-manylinux_2_28_x86_64.whl", hash = "sha256:1cdcdbd117681c88d717437ada72bdd5be9de117f96e3f4d50dab3f59fd9ab20"},
-    {file = "cryptography-43.0.1.tar.gz", hash = "sha256:203e92a75716d8cfb491dc47c79e17d0d9207ccffcbcb35f598fbe463ae3444d"},
+    {file = "cryptography-42.0.4-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:0e89f7b84f421c56e7ff69f11c441ebda73b8a8e6488d322ef71746224c20fce"},
+    {file = "cryptography-42.0.4-pp39-pypy39_pp73-macosx_10_12_x86_64.whl", hash = "sha256:f1e85a178384bf19e36779d91ff35c7617c885da487d689b05c1366f9933ad74"},
+    {file = "cryptography-42.0.4-pp39-pypy39_pp73-manylinux_2_28_aarch64.whl", hash = "sha256:d2a27aca5597c8a71abbe10209184e1a8e91c1fd470b5070a2ea60cafec35bcd"},
+    {file = "cryptography-42.0.4-pp39-pypy39_pp73-manylinux_2_28_x86_64.whl", hash = "sha256:4e36685cb634af55e0677d435d425043967ac2f3790ec652b2b88ad03b85c27b"},
+    {file = "cryptography-42.0.4-pp39-pypy39_pp73-win_amd64.whl", hash = "sha256:f47be41843200f7faec0683ad751e5ef11b9a56a220d57f300376cd8aba81660"},
+    {file = "cryptography-42.0.4.tar.gz", hash = "sha256:831a4b37accef30cccd34fcb916a5d7b5be3cbbe27268a02832c3e450aea39cb"},
 ]
 
 [package.dependencies]
@@ -1029,7 +1034,7 @@ nox = ["nox"]
 pep8test = ["check-sdist", "click", "mypy", "ruff"]
 sdist = ["build"]
 ssh = ["bcrypt (>=3.1.5)"]
-test = ["certifi", "cryptography-vectors (==43.0.1)", "pretend", "pytest (>=6.2.0)", "pytest-benchmark", "pytest-cov", "pytest-xdist"]
+test = ["certifi", "pretend", "pytest (>=6.2.0)", "pytest-benchmark", "pytest-cov", "pytest-xdist"]
 test-randomorder = ["pytest-randomly"]
 
 [[package]]

proxy/src/auth/backend.rs

@@ -311,9 +311,7 @@ async fn auth_quirks(
     let (allowed_ips, maybe_secret) = api.get_allowed_ips_and_secret(ctx, &info).await?;
 
     // check allowed list
-    if config.ip_allowlist_check_enabled
+    if !check_peer_addr_is_in_list(&ctx.peer_addr(), &allowed_ips) {
-        && !check_peer_addr_is_in_list(&ctx.peer_addr(), &allowed_ips)
-    {
         return Err(auth::AuthError::ip_address_not_allowed(ctx.peer_addr()));
     }
 
@@ -605,7 +603,6 @@ mod tests {
         rate_limiter_enabled: true,
         rate_limiter: AuthRateLimiter::new(&RateBucketInfo::DEFAULT_AUTH_SET),
         rate_limit_ip_subnet: 64,
-        ip_allowlist_check_enabled: true,
     });
 
     async fn read_message(r: &mut (impl AsyncRead + Unpin), b: &mut BytesMut) -> PgMessage {

proxy/src/auth/credentials.rs

@@ -538,17 +538,4 @@ mod tests {
         ));
         Ok(())
     }
-
-    #[test]
-    fn test_connection_blocker() {
-        fn check(v: serde_json::Value) -> bool {
-            let peer_addr = IpAddr::from([127, 0, 0, 1]);
-            let ip_list: Vec<IpPattern> = serde_json::from_value(v).unwrap();
-            check_peer_addr_is_in_list(&peer_addr, &ip_list)
-        }
-
-        assert!(check(json!([])));
-        assert!(check(json!(["127.0.0.1"])));
-        assert!(!check(json!(["255.255.255.255"])));
-    }
 }

proxy/src/bin/local_proxy.rs

@@ -224,7 +224,6 @@ fn build_config(args: &LocalProxyCliArgs) -> anyhow::Result<&'static ProxyConfig
             rate_limiter_enabled: false,
             rate_limiter: BucketRateLimiter::new(vec![]),
             rate_limit_ip_subnet: 64,
-            ip_allowlist_check_enabled: true,
         },
         require_client_ip: false,
         handshake_timeout: Duration::from_secs(10),

proxy/src/bin/proxy.rs

@@ -224,10 +224,6 @@ struct ProxyCliArgs {
     /// Whether to retry the wake_compute request
     #[clap(long, default_value = config::RetryConfig::WAKE_COMPUTE_DEFAULT_VALUES)]
     wake_compute_retry: String,
-
-    /// Configure if this is a private access proxy for the POC: In that case the proxy will ignore the IP allowlist
-    #[clap(long, default_value_t = false, value_parser = clap::builder::BoolishValueParser::new(), action = clap::ArgAction::Set)]
-    is_private_access_proxy: bool,
 }
 
 #[derive(clap::Args, Clone, Copy, Debug)]
@@ -686,7 +682,6 @@ fn build_config(args: &ProxyCliArgs) -> anyhow::Result<&'static ProxyConfig> {
         rate_limiter_enabled: args.auth_rate_limit_enabled,
         rate_limiter: AuthRateLimiter::new(args.auth_rate_limit.clone()),
         rate_limit_ip_subnet: args.auth_rate_limit_ip_subnet,
-        ip_allowlist_check_enabled: !args.is_private_access_proxy,
     };
 
     let config = Box::leak(Box::new(ProxyConfig {

proxy/src/cache/endpoints.rs

@@ -242,6 +242,6 @@ mod tests {
     #[test]
     fn test() {
         let s = "{\"branch_created\":null,\"endpoint_created\":{\"endpoint_id\":\"ep-rapid-thunder-w0qqw2q9\"},\"project_created\":null,\"type\":\"endpoint_created\"}";
-        serde_json::from_str::<ControlPlaneEventKey>(s).unwrap();
+        let _: ControlPlaneEventKey = serde_json::from_str(s).unwrap();
     }
 }

proxy/src/cache/timed_lru.rs

@@ -16,7 +16,7 @@ use tracing::debug;
 // On the other hand, `hashlink` has good download stats and appears to be maintained.
 use hashlink::{linked_hash_map::RawEntryMut, LruCache};
 
-use super::{common::Cached, timed_lru, Cache};
+use super::{common::Cached, *};
 
 /// An implementation of timed LRU cache with fixed capacity.
 /// Key properties:

proxy/src/config.rs

@@ -64,7 +64,6 @@ pub struct AuthenticationConfig {
     pub rate_limiter_enabled: bool,
     pub rate_limiter: AuthRateLimiter,
     pub rate_limit_ip_subnet: u8,
-    pub ip_allowlist_check_enabled: bool,
 }
 
 impl TlsConfig {

proxy/src/console/messages.rs

@@ -395,7 +395,7 @@ mod tests {
                 }
             }
         });
-        serde_json::from_str::<KickSession<'_>>(&json.to_string())?;
+        let _: KickSession<'_> = serde_json::from_str(&json.to_string())?;
 
         Ok(())
     }
@@ -403,7 +403,7 @@ mod tests {
     #[test]
     fn parse_db_info() -> anyhow::Result<()> {
         // with password
-        serde_json::from_value::<DatabaseInfo>(json!({
+        let _: DatabaseInfo = serde_json::from_value(json!({
             "host": "localhost",
             "port": 5432,
             "dbname": "postgres",
@@ -413,7 +413,7 @@ mod tests {
         }))?;
 
         // without password
-        serde_json::from_value::<DatabaseInfo>(json!({
+        let _: DatabaseInfo = serde_json::from_value(json!({
             "host": "localhost",
             "port": 5432,
             "dbname": "postgres",
@@ -422,7 +422,7 @@ mod tests {
         }))?;
 
         // new field (forward compatibility)
-        serde_json::from_value::<DatabaseInfo>(json!({
+        let _: DatabaseInfo = serde_json::from_value(json!({
             "host": "localhost",
             "port": 5432,
             "dbname": "postgres",
@@ -441,7 +441,7 @@ mod tests {
             "address": "0.0.0.0",
             "aux": dummy_aux(),
         });
-        serde_json::from_str::<WakeCompute>(&json.to_string())?;
+        let _: WakeCompute = serde_json::from_str(&json.to_string())?;
         Ok(())
     }
 
@@ -451,18 +451,18 @@ mod tests {
         let json = json!({
             "role_secret": "secret",
         });
-        serde_json::from_str::<GetRoleSecret>(&json.to_string())?;
+        let _: GetRoleSecret = serde_json::from_str(&json.to_string())?;
         let json = json!({
             "role_secret": "secret",
             "allowed_ips": ["8.8.8.8"],
         });
-        serde_json::from_str::<GetRoleSecret>(&json.to_string())?;
+        let _: GetRoleSecret = serde_json::from_str(&json.to_string())?;
         let json = json!({
             "role_secret": "secret",
             "allowed_ips": ["8.8.8.8"],
             "project_id": "project",
         });
-        serde_json::from_str::<GetRoleSecret>(&json.to_string())?;
+        let _: GetRoleSecret = serde_json::from_str(&json.to_string())?;
 
         Ok(())
     }

proxy/src/console/mgmt.rs

@@ -78,7 +78,7 @@ pub(crate) type ComputeReady = DatabaseInfo;
 
 // TODO: replace with an http-based protocol.
 struct MgmtHandler;
-
+#[async_trait::async_trait]
 impl postgres_backend::Handler<tokio::net::TcpStream> for MgmtHandler {
     async fn process_query(
         &mut self,

proxy/src/console/provider/neon.rs

@@ -38,7 +38,10 @@ impl Api {
         locks: &'static ApiLocks<EndpointCacheKey>,
         wake_compute_endpoint_rate_limiter: Arc<WakeComputeRateLimiter>,
     ) -> Self {
-        let jwt = std::env::var("NEON_PROXY_TO_CONTROLPLANE_TOKEN").unwrap_or_default();
+        let jwt = match std::env::var("NEON_PROXY_TO_CONTROLPLANE_TOKEN") {
+            Ok(v) => v,
+            Err(_) => String::new(),
+        };
         Self {
             endpoint,
             caches,

proxy/src/context.rs

@@ -6,7 +6,7 @@ use pq_proto::StartupMessageParams;
 use smol_str::SmolStr;
 use std::net::IpAddr;
 use tokio::sync::mpsc;
-use tracing::{debug, field::display, info, info_span, Span};
+use tracing::{field::display, info, info_span, Span};
 use try_lock::TryLock;
 use uuid::Uuid;
 
@@ -362,9 +362,7 @@ impl RequestMonitoringInner {
                 });
         }
         if let Some(tx) = self.sender.take() {
-            tx.send(RequestData::from(&*self))
+            let _: Result<(), _> = tx.send(RequestData::from(&*self));
-                .inspect_err(|e| debug!("tx send failed: {e}"))
-                .ok();
         }
     }
 
@@ -373,9 +371,7 @@ impl RequestMonitoringInner {
         // Here we log the length of the session.
         self.disconnect_timestamp = Some(Utc::now());
         if let Some(tx) = self.disconnect_sender.take() {
-            tx.send(RequestData::from(&*self))
+            let _: Result<(), _> = tx.send(RequestData::from(&*self));
-                .inspect_err(|e| debug!("tx send failed: {e}"))
-                .ok();
         }
     }
 }