Do not set last written LSN for target page prefetch request

Co-authored-by: Heikki Linnakangas <heikki@neon.tech>

.dockerignore

@@ -14,6 +14,7 @@
 !pgxn/
 !proxy/
 !safekeeper/
+!storage_broker/
 !vendor/postgres-v14/
 !vendor/postgres-v15/
 !workspace_hack/

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

@@ -32,8 +32,8 @@ runs:
           exit 2
         fi
 
-    - name: Calculate key
-      id: calculate-key
+    - name: Calculate variables
+      id: calculate-vars
       shell: bash -euxo pipefail {0}
       run: |
         # TODO: for manually triggered workflows (via workflow_dispatch) we need to have a separate key
@@ -41,14 +41,22 @@ runs:
         pr_number=$(jq --raw-output .pull_request.number "$GITHUB_EVENT_PATH" || true)
         if [ "${pr_number}" != "null" ]; then
           key=pr-${pr_number}
-        elif [ "${GITHUB_REF}" = "refs/heads/main" ]; then
+        elif [ "${GITHUB_REF_NAME}" = "main" ]; then
           # Shortcut for a special branch
           key=main
+        elif [ "${GITHUB_REF_NAME}" = "release" ]; then
+          # Shortcut for a special branch
+          key=release
         else
-          key=branch-$(echo ${GITHUB_REF#refs/heads/} | tr -c "[:alnum:]._-" "-")
+          key=branch-$(printf "${GITHUB_REF_NAME}" | tr -c "[:alnum:]._-" "-")
         fi
         echo "KEY=${key}" >> $GITHUB_OUTPUT
 
+        # Sanitize test selection to remove `/` and any other special characters
+        # Use printf instead of echo to avoid having `\n` at the end of the string
+        test_selection=$(printf "${{ inputs.test_selection }}" | tr -c "[:alnum:]._-" "-" )
+        echo "TEST_SELECTION=${test_selection}" >> $GITHUB_OUTPUT
+
     - uses: actions/setup-java@v3
       if: ${{ inputs.action == 'generate' }}
       with:
@@ -74,10 +82,11 @@ runs:
     - name: Upload Allure results
       if: ${{ inputs.action == 'store' }}
       env:
-        REPORT_PREFIX: reports/${{ steps.calculate-key.outputs.KEY }}/${{ inputs.build_type }}
-        RAW_PREFIX: reports-raw/${{ steps.calculate-key.outputs.KEY }}/${{ inputs.build_type }}
+        REPORT_PREFIX: reports/${{ steps.calculate-vars.outputs.KEY }}/${{ inputs.build_type }}
+        RAW_PREFIX: reports-raw/${{ steps.calculate-vars.outputs.KEY }}/${{ inputs.build_type }}
         TEST_OUTPUT: /tmp/test_output
         BUCKET: neon-github-public-dev
+        TEST_SELECTION: ${{ steps.calculate-vars.outputs.TEST_SELECTION }}
       shell: bash -euxo pipefail {0}
       run: |
         # Add metadata
@@ -98,7 +107,7 @@ runs:
           BUILD_TYPE=${{ inputs.build_type }}
         EOF
 
-        ARCHIVE="${GITHUB_RUN_ID}-${{ inputs.test_selection }}-${GITHUB_RUN_ATTEMPT}-$(date +%s).tar.zst"
+        ARCHIVE="${GITHUB_RUN_ID}-${TEST_SELECTION}-${GITHUB_RUN_ATTEMPT}-$(date +%s).tar.zst"
         ZSTD_NBTHREADS=0
 
         tar -C ${TEST_OUTPUT}/allure/results -cf ${ARCHIVE} --zstd .
@@ -109,8 +118,9 @@ runs:
       if: ${{ inputs.action == 'generate' }}
       shell: bash -euxo pipefail {0}
       env:
-        LOCK_FILE: reports/${{ steps.calculate-key.outputs.KEY }}/lock.txt
+        LOCK_FILE: reports/${{ steps.calculate-vars.outputs.KEY }}/lock.txt
         BUCKET: neon-github-public-dev
+        TEST_SELECTION: ${{ steps.calculate-vars.outputs.TEST_SELECTION }}
       run: |
         LOCK_TIMEOUT=300 # seconds
 
@@ -123,12 +133,12 @@ runs:
             fi
             sleep 1
           done
-          echo "${GITHUB_RUN_ID}-${GITHUB_RUN_ATTEMPT}-${{ inputs.test_selection }}" > lock.txt
+          echo "${GITHUB_RUN_ID}-${GITHUB_RUN_ATTEMPT}-${TEST_SELECTION}" > lock.txt
           aws s3 mv --only-show-errors lock.txt "s3://${BUCKET}/${LOCK_FILE}"
 
           # A double-check that exactly WE have acquired the lock
           aws s3 cp --only-show-errors "s3://${BUCKET}/${LOCK_FILE}" ./lock.txt
-          if [ "$(cat lock.txt)" = "${GITHUB_RUN_ID}-${GITHUB_RUN_ATTEMPT}-${{ inputs.test_selection }}" ]; then
+          if [ "$(cat lock.txt)" = "${GITHUB_RUN_ID}-${GITHUB_RUN_ATTEMPT}-${TEST_SELECTION}" ]; then
             break
           fi
         done
@@ -137,8 +147,8 @@ runs:
       if: ${{ inputs.action == 'generate' }}
       id: generate-report
       env:
-        REPORT_PREFIX: reports/${{ steps.calculate-key.outputs.KEY }}/${{ inputs.build_type }}
-        RAW_PREFIX: reports-raw/${{ steps.calculate-key.outputs.KEY }}/${{ inputs.build_type }}
+        REPORT_PREFIX: reports/${{ steps.calculate-vars.outputs.KEY }}/${{ inputs.build_type }}
+        RAW_PREFIX: reports-raw/${{ steps.calculate-vars.outputs.KEY }}/${{ inputs.build_type }}
         TEST_OUTPUT: /tmp/test_output
         BUCKET: neon-github-public-dev
       shell: bash -euxo pipefail {0}
@@ -192,12 +202,13 @@ runs:
       if: ${{ inputs.action == 'generate' && always() }}
       shell: bash -euxo pipefail {0}
       env:
-        LOCK_FILE: reports/${{ steps.calculate-key.outputs.KEY }}/lock.txt
+        LOCK_FILE: reports/${{ steps.calculate-vars.outputs.KEY }}/lock.txt
         BUCKET: neon-github-public-dev
+        TEST_SELECTION: ${{ steps.calculate-vars.outputs.TEST_SELECTION }}
       run: |
         aws s3 cp --only-show-errors "s3://${BUCKET}/${LOCK_FILE}" ./lock.txt || exit 0
 
-        if [ "$(cat lock.txt)" = "${GITHUB_RUN_ID}-${GITHUB_RUN_ATTEMPT}-${{ inputs.test_selection }}" ]; then
+        if [ "$(cat lock.txt)" = "${GITHUB_RUN_ID}-${GITHUB_RUN_ATTEMPT}-${TEST_SELECTION}" ]; then
           aws s3 rm "s3://${BUCKET}/${LOCK_FILE}"
         fi
 

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

@@ -0,0 +1,138 @@
+name: 'Create Branch'
+description: 'Create Branch using API'
+
+inputs:
+  api_key:
+    desctiption: 'Neon API key'
+    required: true
+  project_id:
+    desctiption: 'ID of the Project to create Branch in'
+    required: true
+  api_host:
+    desctiption: 'Neon API host'
+    default: console.stage.neon.tech
+outputs:
+  dsn:
+    description: 'Created Branch DSN (for main database)'
+    value: ${{ steps.change-password.outputs.dsn }}
+  branch_id:
+    description: 'Created Branch ID'
+    value: ${{ steps.create-branch.outputs.branch_id }}
+
+runs:
+  using: "composite"
+  steps:
+    - name: Create New Branch
+      id: create-branch
+      shell: bash -euxo pipefail {0}
+      run: |
+        for i in $(seq 1 10); do
+         branch=$(curl \
+            "https://${API_HOST}/api/v2/projects/${PROJECT_ID}/branches" \
+            --header "Accept: application/json" \
+            --header "Content-Type: application/json" \
+            --header "Authorization: Bearer ${API_KEY}" \
+            --data "{
+              \"branch\": {
+                \"name\": \"Created by actions/neon-branch-create; GITHUB_RUN_ID=${GITHUB_RUN_ID} at $(date +%s)\"
+              },
+              \"endpoints\": [
+                {
+                  \"type\": \"read_write\"
+                }
+              ]
+            }")
+
+          if [ -z "${branch}" ]; then
+            sleep 1
+            continue
+          fi
+
+          branch_id=$(echo $branch | jq --raw-output '.branch.id')
+          if [ "${branch_id}" == "null" ]; then
+            sleep 1
+            continue
+          fi
+
+          break
+        done
+
+        if [ -z "${branch_id}" ] || [ "${branch_id}" == "null" ]; then
+          echo 2>&1 "Failed to create branch after 10 attempts, the latest response was: ${branch}"
+          exit 1
+        fi
+
+        branch_id=$(echo $branch | jq --raw-output '.branch.id')
+        echo "branch_id=${branch_id}" >> $GITHUB_OUTPUT
+
+        host=$(echo $branch | jq --raw-output '.endpoints[0].host')
+        echo "host=${host}" >> $GITHUB_OUTPUT
+      env:
+        API_HOST: ${{ inputs.api_host }}
+        API_KEY: ${{ inputs.api_key }}
+        PROJECT_ID: ${{ inputs.project_id }}
+
+    - name: Get Role name
+      id: role-name
+      shell: bash -euxo pipefail {0}
+      run: |
+        roles=$(curl \
+          "https://${API_HOST}/api/v2/projects/${PROJECT_ID}/branches/${BRANCH_ID}/roles" \
+          --fail \
+          --header "Accept: application/json" \
+          --header "Content-Type: application/json" \
+          --header "Authorization: Bearer ${API_KEY}"
+          )
+
+        role_name=$(echo $roles | jq --raw-output '.roles[] | select(.protected == false) | .name')
+        echo "role_name=${role_name}" >> $GITHUB_OUTPUT
+      env:
+        API_HOST: ${{ inputs.api_host }}
+        API_KEY: ${{ inputs.api_key }}
+        PROJECT_ID: ${{ inputs.project_id }}
+        BRANCH_ID: ${{ steps.create-branch.outputs.branch_id }}
+
+    - name: Change Password
+      id: change-password
+      # A shell without `set -x` to not to expose password/dsn in logs
+      shell: bash -euo pipefail {0}
+      run: |
+        for i in $(seq 1 10); do
+          reset_password=$(curl \
+            "https://${API_HOST}/api/v2/projects/${PROJECT_ID}/branches/${BRANCH_ID}/roles/${ROLE_NAME}/reset_password" \
+            --request POST \
+            --header "Accept: application/json" \
+            --header "Content-Type: application/json" \
+            --header "Authorization: Bearer ${API_KEY}"
+            )
+
+          if [ -z "${reset_password}" ]; then
+            sleep 1
+            continue
+          fi
+
+          password=$(echo $reset_password | jq --raw-output '.role.password')
+          if [ "${password}" == "null" ]; then
+            sleep 1
+            continue
+          fi
+
+          echo "::add-mask::${password}"
+          break
+        done
+
+        if [ -z "${password}" ] || [ "${password}" == "null" ]; then
+          echo 2>&1 "Failed to reset password after 10 attempts, the latest response was: ${reset_password}"
+          exit 1
+        fi
+
+        dsn="postgres://${ROLE_NAME}:${password}@${HOST}/neondb"
+        echo "::add-mask::${dsn}"
+        echo "dsn=${dsn}" >> $GITHUB_OUTPUT
+      env:
+        API_HOST: ${{ inputs.api_host }}
+        API_KEY: ${{ inputs.api_key }}
+        PROJECT_ID: ${{ inputs.project_id }}
+        BRANCH_ID: ${{ steps.create-branch.outputs.branch_id }}
+        ROLE_NAME: ${{ steps.role-name.outputs.role_name }}
+        HOST: ${{ steps.create-branch.outputs.host }}

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

@@ -0,0 +1,58 @@
+name: 'Delete Branch'
+description: 'Delete Branch using API'
+
+inputs:
+  api_key:
+    desctiption: 'Neon API key'
+    required: true
+  project_id:
+    desctiption: 'ID of the Project which should be deleted'
+    required: true
+  branch_id:
+    desctiption: 'ID of the branch to delete'
+    required: true
+  api_host:
+    desctiption: 'Neon API host'
+    default: console.stage.neon.tech
+
+runs:
+  using: "composite"
+  steps:
+    - name: Delete Branch
+      # Do not try to delete a branch if .github/actions/neon-project-create
+      # or .github/actions/neon-branch-create failed before
+      if: ${{ inputs.project_id != '' && inputs.branch_id != '' }}
+      shell: bash -euxo pipefail {0}
+      run: |
+        for i in $(seq 1 10); do
+          deleted_branch=$(curl \
+            "https://${API_HOST}/api/v2/projects/${PROJECT_ID}/branches/${BRANCH_ID}" \
+            --request DELETE \
+            --header "Accept: application/json" \
+            --header "Content-Type: application/json" \
+            --header "Authorization: Bearer ${API_KEY}"
+            )
+
+          if [ -z "${deleted_branch}" ]; then
+            sleep 1
+            continue
+          fi
+
+          branch_id=$(echo $deleted_branch | jq --raw-output '.branch.id')
+          if [ "${branch_id}" == "null" ]; then
+            sleep 1
+            continue
+          fi
+
+          break
+        done
+
+        if [ -z "${branch_id}" ] || [ "${branch_id}" == "null" ]; then
+          echo 2>&1 "Failed to delete branch after 10 attempts, the latest response was: ${deleted_branch}"
+          exit 1
+        fi
+      env:
+        API_HOST: ${{ inputs.api_host }}
+        API_KEY: ${{ inputs.api_key }}
+        PROJECT_ID: ${{ inputs.project_id }}
+        BRANCH_ID: ${{ inputs.branch_id }}

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

@@ -5,12 +5,16 @@ inputs:
   api_key:
     desctiption: 'Neon API key'
     required: true
-  environment:
-    desctiption: 'dev (aka captest) or stage'
-    required: true
   region_id:
     desctiption: 'Region ID, if not set the project will be created in the default region'
-    required: false
+    default: aws-us-east-2
+  postgres_version:
+    desctiption: 'Postgres version; default is 15'
+    default: 15
+  api_host:
+    desctiption: 'Neon API host'
+    default: console.stage.neon.tech
+
 outputs:
   dsn:
     description: 'Created Project DSN (for main database)'
@@ -22,38 +26,13 @@ outputs:
 runs:
   using: "composite"
   steps:
-    - name: Parse Input
-      id: parse-input
-      shell: bash -euxo pipefail {0}
-      run: |
-        case "${ENVIRONMENT}" in
-          dev)
-            API_HOST=console.dev.neon.tech
-            REGION_ID=${REGION_ID:-eu-west-1}
-            ;;
-          staging)
-            API_HOST=console.stage.neon.tech
-            REGION_ID=${REGION_ID:-us-east-1}
-            ;;
-          *)
-            echo 2>&1 "Unknown environment=${ENVIRONMENT}. Allowed 'dev' or 'staging' only"
-            exit 1
-            ;;
-        esac
-
-        echo "api_host=${API_HOST}" >> $GITHUB_OUTPUT
-        echo "region_id=${REGION_ID}" >> $GITHUB_OUTPUT
-      env:
-        ENVIRONMENT: ${{ inputs.environment }}
-        REGION_ID: ${{ inputs.region_id }}
-
     - name: Create Neon Project
       id: create-neon-project
       # A shell without `set -x` to not to expose password/dsn in logs
       shell: bash -euo pipefail {0}
       run: |
         project=$(curl \
-          "https://${API_HOST}/api/v1/projects" \
+          "https://${API_HOST}/api/v2/projects" \
           --fail \
           --header "Accept: application/json" \
           --header "Content-Type: application/json" \
@@ -61,7 +40,7 @@ runs:
           --data "{
             \"project\": {
               \"name\": \"Created by actions/neon-project-create; GITHUB_RUN_ID=${GITHUB_RUN_ID}\",
-              \"platform_id\": \"aws\",
+              \"pg_version\": ${POSTGRES_VERSION},
               \"region_id\": \"${REGION_ID}\",
               \"settings\": { }
             }
@@ -70,13 +49,14 @@ runs:
         # Mask password
         echo "::add-mask::$(echo $project | jq --raw-output '.roles[] | select(.name != "web_access") | .password')"
 
-        dsn=$(echo $project | jq --raw-output '.roles[] | select(.name != "web_access") | .dsn')/main
+        dsn=$(echo $project | jq --raw-output '.connection_uris[0].connection_uri')
         echo "::add-mask::${dsn}"
         echo "dsn=${dsn}" >> $GITHUB_OUTPUT
 
-        project_id=$(echo $project | jq --raw-output '.id')
+        project_id=$(echo $project | jq --raw-output '.project.id')
         echo "project_id=${project_id}" >> $GITHUB_OUTPUT
       env:
+        API_HOST: ${{ inputs.api_host }}
         API_KEY: ${{ inputs.api_key }}
-        API_HOST: ${{ steps.parse-input.outputs.api_host }}
-        REGION_ID: ${{ steps.parse-input.outputs.region_id }}
+        REGION_ID: ${{ inputs.region_id }}
+        POSTGRES_VERSION: ${{ inputs.postgres_version }}

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

@@ -5,50 +5,29 @@ inputs:
   api_key:
     desctiption: 'Neon API key'
     required: true
-  environment:
-    desctiption: 'dev (aka captest) or stage'
-    required: true
   project_id:
     desctiption: 'ID of the Project to delete'
     required: true
+  api_host:
+    desctiption: 'Neon API host'
+    default: console.stage.neon.tech
 
 runs:
   using: "composite"
   steps:
-    - name: Parse Input
-      id: parse-input
-      shell: bash -euxo pipefail {0}
-      run: |
-        case "${ENVIRONMENT}" in
-          dev)
-            API_HOST=console.dev.neon.tech
-            ;;
-          staging)
-            API_HOST=console.stage.neon.tech
-            ;;
-          *)
-            echo 2>&1 "Unknown environment=${ENVIRONMENT}. Allowed 'dev' or 'staging' only"
-            exit 1
-            ;;
-        esac
-
-        echo "api_host=${API_HOST}" >> $GITHUB_OUTPUT
-      env:
-        ENVIRONMENT: ${{ inputs.environment }}
-
     - name: Delete Neon Project
+      # Do not try to delete a project if .github/actions/neon-project-create failed before
+      if: ${{ inputs.project_id != '' }}
       shell: bash -euxo pipefail {0}
       run: |
-        # Allow PROJECT_ID to be empty/null for cases when .github/actions/neon-project-create failed
-        if [ -n "${PROJECT_ID}" ]; then
-          curl -X "POST" \
-            "https://${API_HOST}/api/v1/projects/${PROJECT_ID}/delete" \
-            --fail \
-            --header "Accept: application/json" \
-            --header "Content-Type: application/json" \
-            --header "Authorization: Bearer ${API_KEY}"
-        fi
+        curl \
+          "https://${API_HOST}/api/v2/projects/${PROJECT_ID}" \
+          --fail \
+          --request DELETE \
+          --header "Accept: application/json" \
+          --header "Content-Type: application/json" \
+          --header "Authorization: Bearer ${API_KEY}"
       env:
+        API_HOST: ${{ inputs.api_host }}
         API_KEY: ${{ inputs.api_key }}
         PROJECT_ID: ${{ inputs.project_id }}
-        API_HOST: ${{ steps.parse-input.outputs.api_host }}

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

@@ -55,6 +55,22 @@ runs:
         name: neon-${{ runner.os }}-${{ inputs.build_type }}-artifact
         path: /tmp/neon
 
+    - name: Download Neon binaries for the previous release
+      if: inputs.build_type != 'remote'
+      uses: ./.github/actions/download
+      with:
+        name: neon-${{ runner.os }}-${{ inputs.build_type }}-artifact
+        path: /tmp/neon-previous
+        prefix: latest
+
+    - name: Download compatibility snapshot for Postgres 14
+      if: inputs.build_type != 'remote'
+      uses: ./.github/actions/download
+      with:
+        name: compatibility-snapshot-${{ inputs.build_type }}-pg14
+        path: /tmp/compatibility_snapshot_pg14
+        prefix: latest
+
     - name: Checkout
       if: inputs.needs_postgres_source == 'true'
       uses: actions/checkout@v3
@@ -73,23 +89,18 @@ runs:
       shell: bash -euxo pipefail {0}
       run: ./scripts/pysync
 
-    - name: Download compatibility snapshot for Postgres 14
-      if: inputs.build_type != 'remote'
-      uses: ./.github/actions/download
-      with:
-        name: compatibility-snapshot-${{ inputs.build_type }}-pg14
-        path: /tmp/compatibility_snapshot_pg14
-        prefix: latest
-
     - name: Run pytest
       env:
         NEON_BIN: /tmp/neon/bin
+        COMPATIBILITY_NEON_BIN: /tmp/neon-previous/bin
+        COMPATIBILITY_POSTGRES_DISTRIB_DIR: /tmp/neon-previous/pg_install
         TEST_OUTPUT: /tmp/test_output
         BUILD_TYPE: ${{ inputs.build_type }}
         AWS_ACCESS_KEY_ID: ${{ inputs.real_s3_access_key_id }}
         AWS_SECRET_ACCESS_KEY: ${{ inputs.real_s3_secret_access_key }}
         COMPATIBILITY_SNAPSHOT_DIR: /tmp/compatibility_snapshot_pg14
-        ALLOW_BREAKING_CHANGES: contains(github.event.pull_request.labels.*.name, 'breaking changes')
+        ALLOW_BACKWARD_COMPATIBILITY_BREAKAGE: contains(github.event.pull_request.labels.*.name, 'backward compatibility breakage')
+        ALLOW_FORWARD_COMPATIBILITY_BREAKAGE: contains(github.event.pull_request.labels.*.name, 'forward compatibility breakage')
       shell: bash -euxo pipefail {0}
       run: |
         # PLATFORM will be embedded in the perf test report
@@ -112,7 +123,12 @@ runs:
           exit 1
         fi
         if [[ "${{ inputs.run_in_parallel }}" == "true" ]]; then
+          # -n4 uses four processes to run tests via pytest-xdist
           EXTRA_PARAMS="-n4 $EXTRA_PARAMS"
+
+          # --dist=loadgroup points tests marked with @pytest.mark.xdist_group
+          # to the same worker to make @pytest.mark.order work with xdist
+          EXTRA_PARAMS="--dist=loadgroup $EXTRA_PARAMS"
         fi
 
         if [[ "${{ inputs.run_with_real_s3 }}" == "true" ]]; then
@@ -147,9 +163,9 @@ runs:
         # --verbose prints name of each test (helpful when there are
         # multiple tests in one file)
         # -rA prints summary in the end
-        # -n4 uses four processes to run tests via pytest-xdist
         # -s is not used to prevent pytest from capturing output, because tests are running
         # in parallel and logs are mixed between different tests
+        #
         mkdir -p $TEST_OUTPUT/allure/results
         "${cov_prefix[@]}" ./scripts/pytest \
           --junitxml=$TEST_OUTPUT/junit.xml \
@@ -169,12 +185,12 @@ runs:
       uses: ./.github/actions/upload
       with:
         name: compatibility-snapshot-${{ inputs.build_type }}-pg14-${{ github.run_id }}
-        # The path includes a test name (test_prepare_snapshot) and directory that the test creates (compatibility_snapshot_pg14), keep the path in sync with the test
-        path: /tmp/test_output/test_prepare_snapshot/compatibility_snapshot_pg14/
+        # The path includes a test name (test_create_snapshot) and directory that the test creates (compatibility_snapshot_pg14), keep the path in sync with the test
+        path: /tmp/test_output/test_create_snapshot/compatibility_snapshot_pg14/
         prefix: latest
 
     - name: Create Allure report
-      if: always()
+      if: success() || failure()
       uses: ./.github/actions/allure-report
       with:
         action: store

.github/ansible/.gitignore

@@ -1,5 +1,3 @@
-zenith_install.tar.gz
-.zenith_current_version
 neon_install.tar.gz
 .neon_current_version
 

.github/ansible/get_binaries.sh

@@ -25,6 +25,7 @@ mkdir neon_install/bin/
 docker cp ${ID}:/usr/local/bin/pageserver neon_install/bin/
 docker cp ${ID}:/usr/local/bin/pageserver_binutils neon_install/bin/
 docker cp ${ID}:/usr/local/bin/safekeeper neon_install/bin/
+docker cp ${ID}:/usr/local/bin/storage_broker neon_install/bin/
 docker cp ${ID}:/usr/local/bin/proxy neon_install/bin/
 docker cp ${ID}:/usr/local/v14/bin/ neon_install/v14/bin/
 docker cp ${ID}:/usr/local/v15/bin/ neon_install/v15/bin/

.github/ansible/production.hosts.yaml

@@ -22,6 +22,10 @@ storage:
           console_region_id: aws-us-west-2
         zenith-1-ps-3:
           console_region_id: aws-us-west-2
+        zenith-1-ps-4:
+          console_region_id: aws-us-west-2
+        zenith-1-ps-5:
+          console_region_id: aws-us-west-2
 
     safekeepers:
       hosts:

.github/ansible/scripts/init_pageserver.sh

@@ -1,7 +1,8 @@
 #!/bin/sh
 
-# get instance id from meta-data service
+# fetch params from meta-data service
 INSTANCE_ID=$(curl -s http://169.254.169.254/latest/meta-data/instance-id)
+AZ_ID=$(curl -s http://169.254.169.254/latest/meta-data/placement/availability-zone)
 
 # store fqdn hostname in var
 HOST=$(hostname -f)
@@ -16,7 +17,8 @@ cat <<EOF | tee /tmp/payload
   "instance_id": "${INSTANCE_ID}",
   "http_host": "${HOST}",
   "http_port": 9898,
-  "active": false
+  "active": false,
+  "availability_zone_id": "${AZ_ID}"
 }
 EOF
 

.github/ansible/staging.eu-west-1.hosts.yaml

@@ -0,0 +1,33 @@
+storage:
+  vars:
+    bucket_name: neon-dev-storage-eu-west-1
+    bucket_region: eu-west-1
+    console_mgmt_base_url: http://console-staging.local
+    etcd_endpoints: etcd-0.eu-west-1.aws.neon.build:2379
+    pageserver_config_stub:
+      pg_distrib_dir: /usr/local
+      remote_storage:
+        bucket_name: "{{ bucket_name }}"
+        bucket_region: "{{ bucket_region }}"
+        prefix_in_bucket: "pageserver/v1"
+    safekeeper_s3_prefix: safekeeper/v1/wal
+    hostname_suffix: ""
+    remote_user: ssm-user
+    ansible_aws_ssm_region: eu-west-1
+    ansible_aws_ssm_bucket_name: neon-dev-storage-eu-west-1
+    console_region_id: aws-eu-west-1
+
+  children:
+    pageservers:
+      hosts:
+        pageserver-0.eu-west-1.aws.neon.build:
+          ansible_host: i-01d496c5041c7f34c
+
+    safekeepers:
+      hosts:
+        safekeeper-0.eu-west-1.aws.neon.build:
+          ansible_host: i-05226ef85722831bf
+        safekeeper-1.eu-west-1.aws.neon.build:
+          ansible_host: i-06969ee1bf2958bfc
+        safekeeper-2.eu-west-1.aws.neon.build:
+          ansible_host: i-087892e9625984a0b

.github/ansible/staging.hosts.yaml

@@ -3,7 +3,7 @@ storage:
     bucket_name: zenith-staging-storage-us-east-1
     bucket_region: us-east-1
     console_mgmt_base_url: http://console-staging.local
-    etcd_endpoints: zenith-us-stage-etcd.local:2379
+    etcd_endpoints: etcd-0.us-east-2.aws.neon.build:2379
     pageserver_config_stub:
       pg_distrib_dir: /usr/local
       remote_storage:

.github/ansible/staging.us-east-2.hosts.yaml

@@ -22,6 +22,8 @@ storage:
       hosts:
         pageserver-0.us-east-2.aws.neon.build:
           ansible_host: i-0c3e70929edb5d691
+        pageserver-1.us-east-2.aws.neon.build:
+          ansible_host: i-0565a8b4008aa3f40
 
     safekeepers:
       hosts:

.github/ansible/systemd/pageserver.service

@@ -5,7 +5,7 @@ After=network.target auditd.service
 [Service]
 Type=simple
 User=pageserver
-Environment=RUST_BACKTRACE=1 NEON_REPO_DIR=/storage/pageserver LD_LIBRARY_PATH=/usr/local/v14/lib
+Environment=RUST_BACKTRACE=1 NEON_REPO_DIR=/storage/pageserver LD_LIBRARY_PATH=/usr/local/v14/lib SENTRY_DSN={{ SENTRY_URL_PAGESERVER }}
 ExecStart=/usr/local/bin/pageserver -c "pg_distrib_dir='/usr/local'" -c "listen_pg_addr='0.0.0.0:6400'" -c "listen_http_addr='0.0.0.0:9898'" -c "broker_endpoints=['{{ etcd_endpoints }}']" -D /storage/pageserver/data
 ExecReload=/bin/kill -HUP $MAINPID
 KillMode=mixed

.github/ansible/systemd/safekeeper.service

@@ -5,7 +5,7 @@ After=network.target auditd.service
 [Service]
 Type=simple
 User=safekeeper
-Environment=RUST_BACKTRACE=1 NEON_REPO_DIR=/storage/safekeeper/data LD_LIBRARY_PATH=/usr/local/v14/lib
+Environment=RUST_BACKTRACE=1 NEON_REPO_DIR=/storage/safekeeper/data LD_LIBRARY_PATH=/usr/local/v14/lib SENTRY_DSN={{ SENTRY_URL_SAFEKEEPER }}
 ExecStart=/usr/local/bin/safekeeper -l {{ inventory_hostname }}{{ hostname_suffix }}:6500 --listen-http {{ inventory_hostname }}{{ hostname_suffix }}:7676 -D /storage/safekeeper/data --broker-endpoints={{ etcd_endpoints }} --remote-storage='{bucket_name="{{bucket_name}}", bucket_region="{{bucket_region}}", prefix_in_bucket="{{ safekeeper_s3_prefix }}"}'
 ExecReload=/bin/kill -HUP $MAINPID
 KillMode=mixed

.github/helm-values/dev-eu-west-1-zeta.neon-proxy-scram.yaml

@@ -0,0 +1,31 @@
+# Helm chart values for neon-proxy-scram.
+# This is a YAML-formatted file.
+
+image:
+  repository: neondatabase/neon
+
+settings:
+  authBackend: "console"
+  authEndpoint: "http://console-staging.local/management/api/v2"
+  domain: "*.eu-west-1.aws.neon.build"
+
+# -- Additional labels for neon-proxy pods
+podLabels:
+  zenith_service: proxy-scram
+  zenith_env: dev
+  zenith_region: eu-west-1
+  zenith_region_slug: eu-west-1
+
+exposedService:
+  annotations:
+    service.beta.kubernetes.io/aws-load-balancer-type: external
+    service.beta.kubernetes.io/aws-load-balancer-nlb-target-type: ip
+    service.beta.kubernetes.io/aws-load-balancer-scheme: internet-facing
+    external-dns.alpha.kubernetes.io/hostname: eu-west-1.aws.neon.build
+
+#metrics:
+#  enabled: true
+#  serviceMonitor:
+#    enabled: true
+#    selector:
+#      release: kube-prometheus-stack

.github/helm-values/dev-eu-west-1-zeta.neon-storage-broker.yaml

@@ -0,0 +1,53 @@
+# Helm chart values for neon-storage-broker
+podLabels:
+  neon_env: staging
+  neon_service: storage-broker
+
+ingress:
+  enabled: true
+  annotations:
+    kubernetes.io/ingress.class: nginx-internal
+    nginx.ingress.kubernetes.io/backend-protocol: "GRPC"
+    nginx.ingress.kubernetes.io/ssl-redirect: "true"
+    nginx.ingress.kubernetes.io/force-ssl-redirect: "true"
+    cert-manager.io/cluster-issuer: "cert-manager-clusterissuer"
+
+  hosts:
+    - host: storage-broker.zeta.eu-west-1.internal.aws.neon.build
+      paths:
+        - path: /
+          pathType: Prefix
+  tls:
+    - hosts:
+        - storage-broker.zeta.eu-west-1.internal.aws.neon.build
+      secretName: storage-broker-tls
+
+
+metrics:
+  enabled: false
+
+extraManifests:
+  - apiVersion: operator.victoriametrics.com/v1beta1
+    kind: VMServiceScrape
+    metadata:
+      name: "{{ include \"neon-storage-broker.fullname\" . }}"
+      labels:
+        helm.sh/chart: neon-storage-broker-{{ .Chart.Version }}
+        app.kubernetes.io/name: neon-storage-broker
+        app.kubernetes.io/instance: neon-storage-broker
+        app.kubernetes.io/version: "{{ .Chart.AppVersion }}"
+        app.kubernetes.io/managed-by: Helm
+      namespace: "{{ .Release.Namespace }}"
+    spec:
+      selector:
+        matchLabels:
+          app.kubernetes.io/name: "neon-storage-broker"
+      endpoints:
+        - port: broker
+          path: /metrics
+          interval: 10s
+          scrapeTimeout: 10s
+      namespaceSelector:
+        matchNames:
+          - "{{ .Release.Namespace }}"
+

.github/helm-values/dev-us-east-2-beta.neon-proxy-link.yaml

@@ -0,0 +1,39 @@
+# Helm chart values for neon-proxy-link.
+# This is a YAML-formatted file.
+
+image:
+  repository: neondatabase/neon
+
+settings:
+  authBackend: "link"
+  authEndpoint: "https://console.stage.neon.tech/authenticate_proxy_request/"
+  uri: "https://console.stage.neon.tech/psql_session/"
+
+# -- Additional labels for neon-proxy-link pods
+podLabels:
+  zenith_service: proxy
+  zenith_env: dev
+  zenith_region: us-east-2
+  zenith_region_slug: us-east-2
+
+service:
+  type: LoadBalancer
+  annotations:
+    service.beta.kubernetes.io/aws-load-balancer-type: external
+    service.beta.kubernetes.io/aws-load-balancer-nlb-target-type: ip
+    service.beta.kubernetes.io/aws-load-balancer-scheme: internal
+    external-dns.alpha.kubernetes.io/hostname: neon-proxy-link-mgmt.beta.us-east-2.aws.neon.build
+
+exposedService:
+  annotations:
+    service.beta.kubernetes.io/aws-load-balancer-type: external
+    service.beta.kubernetes.io/aws-load-balancer-nlb-target-type: ip
+    service.beta.kubernetes.io/aws-load-balancer-scheme: internet-facing
+    external-dns.alpha.kubernetes.io/hostname: neon-proxy-link.beta.us-east-2.aws.neon.build
+
+#metrics:
+#  enabled: true
+#  serviceMonitor:
+#    enabled: true
+#    selector:
+#      release: kube-prometheus-stack

.github/helm-values/dev-us-east-2-beta.neon-proxy-scram-legacy.yaml

@@ -0,0 +1,31 @@
+# Helm chart values for neon-proxy-scram.
+# This is a YAML-formatted file.
+
+image:
+  repository: neondatabase/neon
+
+settings:
+  authBackend: "console"
+  authEndpoint: "http://console-staging.local/management/api/v2"
+  domain: "*.cloud.stage.neon.tech"
+
+# -- Additional labels for neon-proxy pods
+podLabels:
+  zenith_service: proxy-scram-legacy
+  zenith_env: dev
+  zenith_region: us-east-2
+  zenith_region_slug: us-east-2
+
+exposedService:
+  annotations:
+    service.beta.kubernetes.io/aws-load-balancer-type: external
+    service.beta.kubernetes.io/aws-load-balancer-nlb-target-type: ip
+    service.beta.kubernetes.io/aws-load-balancer-scheme: internet-facing
+    external-dns.alpha.kubernetes.io/hostname: neon-proxy-scram-legacy.beta.us-east-2.aws.neon.build
+
+#metrics:
+#  enabled: true
+#  serviceMonitor:
+#    enabled: true
+#    selector:
+#      release: kube-prometheus-stack

.github/helm-values/dev-us-east-2-beta.neon-storage-broker.yaml

@@ -0,0 +1,53 @@
+# Helm chart values for neon-storage-broker
+podLabels:
+  neon_env: staging
+  neon_service: storage-broker
+
+ingress:
+  enabled: true
+  annotations:
+    kubernetes.io/ingress.class: nginx-internal
+    nginx.ingress.kubernetes.io/backend-protocol: "GRPC"
+    nginx.ingress.kubernetes.io/ssl-redirect: "true"
+    nginx.ingress.kubernetes.io/force-ssl-redirect: "true"
+    cert-manager.io/cluster-issuer: "cert-manager-clusterissuer"
+
+  hosts:
+    - host: storage-broker.beta.us-east-2.internal.aws.neon.build
+      paths:
+        - path: /
+          pathType: Prefix
+  tls:
+    - hosts:
+        - storage-broker.beta.us-east-2.internal.aws.neon.build
+      secretName: storage-broker-tls
+
+
+metrics:
+  enabled: false
+
+extraManifests:
+  - apiVersion: operator.victoriametrics.com/v1beta1
+    kind: VMServiceScrape
+    metadata:
+      name: "{{ include \"neon-storage-broker.fullname\" . }}"
+      labels:
+        helm.sh/chart: neon-storage-broker-{{ .Chart.Version }}
+        app.kubernetes.io/name: neon-storage-broker
+        app.kubernetes.io/instance: neon-storage-broker
+        app.kubernetes.io/version: "{{ .Chart.AppVersion }}"
+        app.kubernetes.io/managed-by: Helm
+      namespace: "{{ .Release.Namespace }}"
+    spec:
+      selector:
+        matchLabels:
+          app.kubernetes.io/name: "neon-storage-broker"
+      endpoints:
+        - port: broker
+          path: /metrics
+          interval: 10s
+          scrapeTimeout: 10s
+      namespaceSelector:
+        matchNames:
+          - "{{ .Release.Namespace }}"
+

.github/helm-values/neon-stress.neon-storage-broker.yaml

@@ -0,0 +1,54 @@
+# Helm chart values for neon-storage-broker
+podLabels:
+  neon_env: neon-stress
+  neon_service: storage-broker
+
+ingress:
+  enabled: true
+  annotations:
+    kubernetes.io/ingress.class: alb
+    alb.ingress.kubernetes.io/healthcheck-path: /status
+    alb.ingress.kubernetes.io/listen-ports: '[{"HTTPS":443}]'
+    alb.ingress.kubernetes.io/scheme: "internal"
+    alb.ingress.kubernetes.io/target-type: "ip"
+    alb.ingress.kubernetes.io/ssl-redirect: "443"
+    alb.ingress.kubernetes.io/backend-protocol-version: "GRPC"
+
+  hosts:
+    - host: storage-broker-stress.stage.neon.tech
+      paths:
+        - path: /
+          pathType: Prefix
+
+metrics:
+  enabled: true
+  serviceMonitor:
+    enabled: true
+    selector:
+      release: kube-prometheus-stack
+
+extraManifests:
+  - apiVersion: operator.victoriametrics.com/v1beta1
+    kind: VMServiceScrape
+    metadata:
+      name: "{{ include \"neon-storage-broker.fullname\" . }}"
+      labels:
+        helm.sh/chart: neon-storage-broker-{{ .Chart.Version }}
+        app.kubernetes.io/name: neon-storage-broker
+        app.kubernetes.io/instance: neon-storage-broker
+        app.kubernetes.io/version: "{{ .Chart.AppVersion }}"
+        app.kubernetes.io/managed-by: Helm
+      namespace: "{{ .Release.Namespace }}"
+    spec:
+      selector:
+        matchLabels:
+          app.kubernetes.io/name: "neon-storage-broker"
+      endpoints:
+        - port: broker
+          path: /metrics
+          interval: 10s
+          scrapeTimeout: 10s
+      namespaceSelector:
+        matchNames:
+          - "{{ .Release.Namespace }}"
+

.github/helm-values/prod-ap-southeast-1-epsilon.neon-storage-broker.yaml

@@ -0,0 +1,53 @@
+# Helm chart values for neon-storage-broker
+podLabels:
+  neon_env: production
+  neon_service: storage-broker
+
+ingress:
+  enabled: true
+  annotations:
+    kubernetes.io/ingress.class: nginx-internal
+    nginx.ingress.kubernetes.io/backend-protocol: "GRPC"
+    nginx.ingress.kubernetes.io/ssl-redirect: "true"
+    nginx.ingress.kubernetes.io/force-ssl-redirect: "true"
+    cert-manager.io/cluster-issuer: "cert-manager-clusterissuer"
+
+  hosts:
+    - host: storage-broker.epsilon.ap-southeast-1.internal.aws.neon.tech
+      paths:
+        - path: /
+          pathType: Prefix
+  tls:
+    - hosts:
+        - storage-broker.epsilon.ap-southeast-1.internal.aws.neon.tech
+      secretName: storage-broker-tls
+
+
+metrics:
+  enabled: false
+
+extraManifests:
+  - apiVersion: operator.victoriametrics.com/v1beta1
+    kind: VMServiceScrape
+    metadata:
+      name: "{{ include \"neon-storage-broker.fullname\" . }}"
+      labels:
+        helm.sh/chart: neon-storage-broker-{{ .Chart.Version }}
+        app.kubernetes.io/name: neon-storage-broker
+        app.kubernetes.io/instance: neon-storage-broker
+        app.kubernetes.io/version: "{{ .Chart.AppVersion }}"
+        app.kubernetes.io/managed-by: Helm
+      namespace: "{{ .Release.Namespace }}"
+    spec:
+      selector:
+        matchLabels:
+          app.kubernetes.io/name: "neon-storage-broker"
+      endpoints:
+        - port: broker
+          path: /metrics
+          interval: 10s
+          scrapeTimeout: 10s
+      namespaceSelector:
+        matchNames:
+          - "{{ .Release.Namespace }}"
+

.github/helm-values/prod-eu-central-1-gamma.neon-storage-broker.yaml

@@ -0,0 +1,53 @@
+# Helm chart values for neon-storage-broker
+podLabels:
+  neon_env: production
+  neon_service: storage-broker
+
+ingress:
+  enabled: true
+  annotations:
+    kubernetes.io/ingress.class: nginx-internal
+    nginx.ingress.kubernetes.io/backend-protocol: "GRPC"
+    nginx.ingress.kubernetes.io/ssl-redirect: "true"
+    nginx.ingress.kubernetes.io/force-ssl-redirect: "true"
+    cert-manager.io/cluster-issuer: "cert-manager-clusterissuer"
+
+  hosts:
+    - host: storage-broker.gamma.eu-central-1.internal.aws.neon.tech
+      paths:
+        - path: /
+          pathType: Prefix
+  tls:
+    - hosts:
+        - storage-broker.gamma.eu-central-1.internal.aws.neon.tech
+      secretName: storage-broker-tls
+
+
+metrics:
+  enabled: false
+
+extraManifests:
+  - apiVersion: operator.victoriametrics.com/v1beta1
+    kind: VMServiceScrape
+    metadata:
+      name: "{{ include \"neon-storage-broker.fullname\" . }}"
+      labels:
+        helm.sh/chart: neon-storage-broker-{{ .Chart.Version }}
+        app.kubernetes.io/name: neon-storage-broker
+        app.kubernetes.io/instance: neon-storage-broker
+        app.kubernetes.io/version: "{{ .Chart.AppVersion }}"
+        app.kubernetes.io/managed-by: Helm
+      namespace: "{{ .Release.Namespace }}"
+    spec:
+      selector:
+        matchLabels:
+          app.kubernetes.io/name: "neon-storage-broker"
+      endpoints:
+        - port: broker
+          path: /metrics
+          interval: 10s
+          scrapeTimeout: 10s
+      namespaceSelector:
+        matchNames:
+          - "{{ .Release.Namespace }}"
+

.github/helm-values/prod-us-east-2-delta.neon-storage-broker.yaml

@@ -0,0 +1,53 @@
+# Helm chart values for neon-storage-broker
+podLabels:
+  neon_env: production
+  neon_service: storage-broker
+
+ingress:
+  enabled: true
+  annotations:
+    kubernetes.io/ingress.class: nginx-internal
+    nginx.ingress.kubernetes.io/backend-protocol: "GRPC"
+    nginx.ingress.kubernetes.io/ssl-redirect: "true"
+    nginx.ingress.kubernetes.io/force-ssl-redirect: "true"
+    cert-manager.io/cluster-issuer: "cert-manager-clusterissuer"
+
+  hosts:
+    - host: storage-broker.delta.us-east-2.internal.aws.neon.tech
+      paths:
+        - path: /
+          pathType: Prefix
+  tls:
+    - hosts:
+        - storage-broker.delta.us-east-2.internal.aws.neon.tech
+      secretName: storage-broker-tls
+
+
+metrics:
+  enabled: false
+
+extraManifests:
+  - apiVersion: operator.victoriametrics.com/v1beta1
+    kind: VMServiceScrape
+    metadata:
+      name: "{{ include \"neon-storage-broker.fullname\" . }}"
+      labels:
+        helm.sh/chart: neon-storage-broker-{{ .Chart.Version }}
+        app.kubernetes.io/name: neon-storage-broker
+        app.kubernetes.io/instance: neon-storage-broker
+        app.kubernetes.io/version: "{{ .Chart.AppVersion }}"
+        app.kubernetes.io/managed-by: Helm
+      namespace: "{{ .Release.Namespace }}"
+    spec:
+      selector:
+        matchLabels:
+          app.kubernetes.io/name: "neon-storage-broker"
+      endpoints:
+        - port: broker
+          path: /metrics
+          interval: 10s
+          scrapeTimeout: 10s
+      namespaceSelector:
+        matchNames:
+          - "{{ .Release.Namespace }}"
+

.github/helm-values/prod-us-west-2-eta.neon-proxy-scram.yaml

@@ -0,0 +1,31 @@
+# Helm chart values for neon-proxy-scram.
+# This is a YAML-formatted file.
+
+image:
+  repository: neondatabase/neon
+
+settings:
+  authBackend: "console"
+  authEndpoint: "http://console-release.local/management/api/v2"
+  domain: "*.us-west-2.aws.neon.tech"
+
+# -- Additional labels for neon-proxy pods
+podLabels:
+  zenith_service: proxy-scram
+  zenith_env: prod
+  zenith_region: us-west-2
+  zenith_region_slug: us-west-2
+
+exposedService:
+  annotations:
+    service.beta.kubernetes.io/aws-load-balancer-type: external
+    service.beta.kubernetes.io/aws-load-balancer-nlb-target-type: ip
+    service.beta.kubernetes.io/aws-load-balancer-scheme: internet-facing
+    external-dns.alpha.kubernetes.io/hostname: us-west-2.aws.neon.tech
+
+#metrics:
+#  enabled: true
+#  serviceMonitor:
+#    enabled: true
+#    selector:
+#      release: kube-prometheus-stack

.github/helm-values/prod-us-west-2-eta.neon-storage-broker.yaml

@@ -0,0 +1,53 @@
+# Helm chart values for neon-storage-broker
+podLabels:
+  neon_env: production
+  neon_service: storage-broker
+
+ingress:
+  enabled: true
+  annotations:
+    kubernetes.io/ingress.class: nginx-internal
+    nginx.ingress.kubernetes.io/backend-protocol: "GRPC"
+    nginx.ingress.kubernetes.io/ssl-redirect: "true"
+    nginx.ingress.kubernetes.io/force-ssl-redirect: "true"
+    cert-manager.io/cluster-issuer: "cert-manager-clusterissuer"
+
+  hosts:
+    - host: storage-broker.eta.us-west-2.internal.aws.neon.tech
+      paths:
+        - path: /
+          pathType: Prefix
+  tls:
+    - hosts:
+        - storage-broker.eta.us-west-2.internal.aws.neon.tech
+      secretName: storage-broker-tls
+
+
+metrics:
+  enabled: false
+
+extraManifests:
+  - apiVersion: operator.victoriametrics.com/v1beta1
+    kind: VMServiceScrape
+    metadata:
+      name: "{{ include \"neon-storage-broker.fullname\" . }}"
+      labels:
+        helm.sh/chart: neon-storage-broker-{{ .Chart.Version }}
+        app.kubernetes.io/name: neon-storage-broker
+        app.kubernetes.io/instance: neon-storage-broker
+        app.kubernetes.io/version: "{{ .Chart.AppVersion }}"
+        app.kubernetes.io/managed-by: Helm
+      namespace: "{{ .Release.Namespace }}"
+    spec:
+      selector:
+        matchLabels:
+          app.kubernetes.io/name: "neon-storage-broker"
+      endpoints:
+        - port: broker
+          path: /metrics
+          interval: 10s
+          scrapeTimeout: 10s
+      namespaceSelector:
+        matchNames:
+          - "{{ .Release.Namespace }}"
+

.github/helm-values/production.neon-storage-broker.yaml

@@ -0,0 +1,54 @@
+# Helm chart values for neon-storage-broker
+podLabels:
+  neon_env: production
+  neon_service: storage-broker
+
+ingress:
+  enabled: true
+  annotations:
+    kubernetes.io/ingress.class: alb
+    alb.ingress.kubernetes.io/healthcheck-path: /status
+    alb.ingress.kubernetes.io/listen-ports: '[{"HTTPS":443}]'
+    alb.ingress.kubernetes.io/scheme: "internal"
+    alb.ingress.kubernetes.io/target-type: "ip"
+    alb.ingress.kubernetes.io/ssl-redirect: "443"
+    alb.ingress.kubernetes.io/backend-protocol-version: "GRPC"
+
+  hosts:
+    - host: storage-broker.neon.tech
+      paths:
+        - path: /
+          pathType: Prefix
+
+metrics:
+  enabled: true
+  serviceMonitor:
+    enabled: true
+    selector:
+      release: kube-prometheus-stack
+
+extraManifests:
+  - apiVersion: operator.victoriametrics.com/v1beta1
+    kind: VMServiceScrape
+    metadata:
+      name: "{{ include \"neon-storage-broker.fullname\" . }}"
+      labels:
+        helm.sh/chart: neon-storage-broker-{{ .Chart.Version }}
+        app.kubernetes.io/name: neon-storage-broker
+        app.kubernetes.io/instance: neon-storage-broker
+        app.kubernetes.io/version: "{{ .Chart.AppVersion }}"
+        app.kubernetes.io/managed-by: Helm
+      namespace: "{{ .Release.Namespace }}"
+    spec:
+      selector:
+        matchLabels:
+          app.kubernetes.io/name: "neon-storage-broker"
+      endpoints:
+        - port: broker
+          path: /metrics
+          interval: 10s
+          scrapeTimeout: 10s
+      namespaceSelector:
+        matchNames:
+          - "{{ .Release.Namespace }}"
+

.github/helm-values/staging.neon-storage-broker.yaml

@@ -0,0 +1,54 @@
+# Helm chart values for neon-storage-broker
+podLabels:
+  neon_env: staging
+  neon_service: storage-broker
+
+ingress:
+  enabled: true
+  annotations:
+    kubernetes.io/ingress.class: alb
+    alb.ingress.kubernetes.io/healthcheck-path: /status
+    alb.ingress.kubernetes.io/listen-ports: '[{"HTTPS":443}]'
+    alb.ingress.kubernetes.io/scheme: "internal"
+    alb.ingress.kubernetes.io/target-type: "ip"
+    alb.ingress.kubernetes.io/ssl-redirect: "443"
+    alb.ingress.kubernetes.io/backend-protocol-version: "GRPC"
+
+  hosts:
+    - host: storage-broker.stage.neon.tech
+      paths:
+        - path: /
+          pathType: Prefix
+
+metrics:
+  enabled: true
+  serviceMonitor:
+    enabled: true
+    selector:
+      release: kube-prometheus-stack
+
+extraManifests:
+  - apiVersion: operator.victoriametrics.com/v1beta1
+    kind: VMServiceScrape
+    metadata:
+      name: "{{ include \"neon-storage-broker.fullname\" . }}"
+      labels:
+        helm.sh/chart: neon-storage-broker-{{ .Chart.Version }}
+        app.kubernetes.io/name: neon-storage-broker
+        app.kubernetes.io/instance: neon-storage-broker
+        app.kubernetes.io/version: "{{ .Chart.AppVersion }}"
+        app.kubernetes.io/managed-by: Helm
+      namespace: "{{ .Release.Namespace }}"
+    spec:
+      selector:
+        matchLabels:
+          app.kubernetes.io/name: "neon-storage-broker"
+      endpoints:
+        - port: broker
+          path: /metrics
+          interval: 10s
+          scrapeTimeout: 10s
+      namespaceSelector:
+        matchNames:
+          - "{{ .Release.Namespace }}"
+

.github/workflows/benchmarking.yml

@@ -15,9 +15,6 @@ on:
 
   workflow_dispatch: # adds ability to run this manually
     inputs:
-      environment:
-        description: 'Environment to run remote tests on (dev or staging)'
-        required: false
       region_id:
         description: 'Use a particular region. If not set the default region will be used'
         required: false
@@ -37,97 +34,69 @@ concurrency:
 
 jobs:
   bench:
-    # this workflow runs on self hosteed runner
-    # it's environment is quite different from usual guthub runner
-    # probably the most important difference is that it doesn't start from clean workspace each time
-    # e g if you install system packages they are not cleaned up since you install them directly in host machine
-    # not a container or something
-    # See documentation for more info: https://docs.github.com/en/actions/hosting-your-own-runners/about-self-hosted-runners
-    runs-on: [self-hosted, zenith-benchmarker]
-
     env:
-      POSTGRES_DISTRIB_DIR: /usr/pgsql
+      TEST_PG_BENCH_DURATIONS_MATRIX: "300"
+      TEST_PG_BENCH_SCALES_MATRIX: "10,100"
+      POSTGRES_DISTRIB_DIR: /tmp/neon/pg_install
       DEFAULT_PG_VERSION: 14
+      TEST_OUTPUT: /tmp/test_output
+      BUILD_TYPE: remote
+      SAVE_PERF_REPORT: ${{ github.event.inputs.save_perf_report || ( github.ref == 'refs/heads/main' ) }}
+      PLATFORM: "neon-staging"
+
+    runs-on: [ self-hosted, us-east-2, x64 ]
+    container:
+      image: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/rust:pinned
+      options: --init
 
     steps:
-    - name: Checkout zenith repo
-      uses: actions/checkout@v3
+    - uses: actions/checkout@v3
 
-    # actions/setup-python@v2 is not working correctly on self-hosted runners
-    # see https://github.com/actions/setup-python/issues/162
-    # and probably https://github.com/actions/setup-python/issues/162#issuecomment-865387976 in particular
-    # so the simplest solution to me is to use already installed system python and spin virtualenvs for job runs.
-    # there is Python 3.7.10 already installed on the machine so use it to install poetry and then use poetry's virtuealenvs
-    - name: Install poetry & deps
-      run: |
-        python3 -m pip install --upgrade poetry wheel
-        # since pip/poetry caches are reused there shouldn't be any troubles with install every time
-        ./scripts/pysync
-
-    - name: Show versions
-      run: |
-        echo Python
-        python3 --version
-        poetry run python3 --version
-        echo Poetry
-        poetry --version
-        echo Pgbench
-        ${POSTGRES_DISTRIB_DIR}/v${DEFAULT_PG_VERSION}/bin/pgbench --version
+    - name: Download Neon artifact
+      uses: ./.github/actions/download
+      with:
+        name: neon-${{ runner.os }}-release-artifact
+        path: /tmp/neon/
+        prefix: latest
 
     - name: Create Neon Project
       id: create-neon-project
       uses: ./.github/actions/neon-project-create
       with:
-        environment: ${{ github.event.inputs.environment || 'staging' }}
-        api_key: ${{ ( github.event.inputs.environment || 'staging' ) == 'staging' && secrets.NEON_STAGING_API_KEY  || secrets.NEON_CAPTEST_API_KEY }}
+        region_id: ${{ github.event.inputs.region_id || 'aws-us-east-2' }}
+        postgres_version: ${{ env.DEFAULT_PG_VERSION }}
+        api_key: ${{ secrets.NEON_STAGING_API_KEY }}
 
     - name: Run benchmark
-      # pgbench is installed system wide from official repo
-      # https://download.postgresql.org/pub/repos/yum/13/redhat/rhel-7-x86_64/
-      # via
-      # sudo tee /etc/yum.repos.d/pgdg.repo<<EOF
-      # [pgdg13]
-      # name=PostgreSQL 13 for RHEL/CentOS 7 - x86_64
-      # baseurl=https://download.postgresql.org/pub/repos/yum/13/redhat/rhel-7-x86_64/
-      # enabled=1
-      # gpgcheck=0
-      # EOF
-      # sudo yum makecache
-      # sudo yum install postgresql13-contrib
-      # actual binaries are located in /usr/pgsql-13/bin/
+      uses: ./.github/actions/run-python-test-set
+      with:
+        build_type: ${{ env.BUILD_TYPE }}
+        test_selection: performance
+        run_in_parallel: false
+        save_perf_report: ${{ env.SAVE_PERF_REPORT }}
+        # Set --sparse-ordering option of pytest-order plugin
+        # to ensure tests are running in order of appears in the file.
+        # It's important for test_perf_pgbench.py::test_pgbench_remote_* tests
+        extra_params: -m remote_cluster --sparse-ordering --timeout 5400 --ignore test_runner/performance/test_perf_olap.py
       env:
-        # The pgbench test runs two tests of given duration against each scale.
-        # So the total runtime with these parameters is 2 * 2 * 300 = 1200, or 20 minutes.
-        # Plus time needed to initialize the test databases.
-        TEST_PG_BENCH_DURATIONS_MATRIX: "300"
-        TEST_PG_BENCH_SCALES_MATRIX: "10,100"
-        PLATFORM: "neon-staging"
         BENCHMARK_CONNSTR: ${{ steps.create-neon-project.outputs.dsn }}
-        REMOTE_ENV: "1" # indicate to test harness that we do not have zenith binaries locally
-      run: |
-        # just to be sure that no data was cached on self hosted runner
-        # since it might generate duplicates when calling ingest_perf_test_result.py
-        rm -rf perf-report-staging
-        mkdir -p perf-report-staging
-        # Set --sparse-ordering option of pytest-order plugin to ensure tests are running in order of appears in the file,
-        # it's important for test_perf_pgbench.py::test_pgbench_remote_* tests
-        ./scripts/pytest test_runner/performance/ -v -m "remote_cluster" --sparse-ordering --out-dir perf-report-staging --timeout 5400
-
-    - name: Submit result
-      env:
         VIP_VAP_ACCESS_TOKEN: "${{ secrets.VIP_VAP_ACCESS_TOKEN }}"
         PERF_TEST_RESULT_CONNSTR: "${{ secrets.PERF_TEST_RESULT_CONNSTR }}"
-      run: |
-        REPORT_FROM=$(realpath perf-report-staging) REPORT_TO=staging scripts/generate_and_push_perf_report.sh
 
     - name: Delete Neon Project
       if: ${{ always() }}
       uses: ./.github/actions/neon-project-delete
       with:
-        environment: staging
         project_id: ${{ steps.create-neon-project.outputs.project_id }}
         api_key: ${{ secrets.NEON_STAGING_API_KEY }}
 
+    - name: Create Allure report
+      if: success() || failure()
+      uses: ./.github/actions/allure-report
+      with:
+        action: generate
+        build_type: ${{ env.BUILD_TYPE }}
+
     - name: Post to a Slack channel
       if: ${{ github.event.schedule && failure() }}
       uses: slackapi/slack-github-action@v1
@@ -144,15 +113,24 @@ jobs:
         # neon-captest-new: Run pgbench in a freshly created project
         # neon-captest-reuse: Same, but reusing existing project
         # neon-captest-prefetch: Same, with prefetching enabled (new project)
-        platform: [ neon-captest-new, neon-captest-reuse, neon-captest-prefetch ]
+        # rds-aurora: Aurora Postgres Serverless v2 with autoscaling from 0.5 to 2 ACUs
+        # rds-postgres: RDS Postgres db.m5.large instance (2 vCPU, 8 GiB) with gp3 EBS storage
+        platform: [ neon-captest-new, neon-captest-prefetch, rds-postgres ]
         db_size: [ 10gb ]
+        runner: [ us-east-2 ]
         include:
+          - platform: neon-captest-reuse
+            db_size: 10gb
+            runner: dev  # TODO: Switch to us-east-2 after dry-bonus-223539 migration to staging
           - platform: neon-captest-new
             db_size: 50gb
+            runner: us-east-2
           - platform: neon-captest-prefetch
             db_size: 50gb
+            runner: us-east-2
           - platform: rds-aurora
             db_size: 50gb
+            runner: us-east-2
 
     env:
       TEST_PG_BENCH_DURATIONS_MATRIX: "60m"
@@ -164,9 +142,9 @@ jobs:
       SAVE_PERF_REPORT: ${{ github.event.inputs.save_perf_report || ( github.ref == 'refs/heads/main' ) }}
       PLATFORM: ${{ matrix.platform }}
 
-    runs-on: dev
+    runs-on: [ self-hosted, "${{ matrix.runner }}", x64 ]
     container:
-      image: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/rustlegacy:pinned
+      image: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/rust:pinned
       options: --init
 
     timeout-minutes: 360 # 6h
@@ -191,8 +169,9 @@ jobs:
       id: create-neon-project
       uses: ./.github/actions/neon-project-create
       with:
-        environment: ${{ github.event.inputs.environment || 'dev' }}
-        api_key: ${{ ( github.event.inputs.environment || 'dev' ) == 'staging' && secrets.NEON_STAGING_API_KEY  || secrets.NEON_CAPTEST_API_KEY }}
+        region_id: ${{ github.event.inputs.region_id || 'aws-us-east-2' }}
+        postgres_version: ${{ env.DEFAULT_PG_VERSION }}
+        api_key: ${{ secrets.NEON_STAGING_API_KEY }}
 
     - name: Set up Connection String
       id: set-up-connstr
@@ -205,10 +184,13 @@ jobs:
             CONNSTR=${{ steps.create-neon-project.outputs.dsn }}
             ;;
           rds-aurora)
-            CONNSTR=${{ secrets.BENCHMARK_RDS_CONNSTR }}
+            CONNSTR=${{ secrets.BENCHMARK_RDS_AURORA_CONNSTR }}
+            ;;
+          rds-postgres)
+            CONNSTR=${{ secrets.BENCHMARK_RDS_POSTGRES_CONNSTR }}
             ;;
           *)
-            echo 2>&1 "Unknown PLATFORM=${PLATFORM}. Allowed only 'neon-captest-reuse', 'neon-captest-new', 'neon-captest-prefetch' or 'rds-aurora'"
+            echo 2>&1 "Unknown PLATFORM=${PLATFORM}. Allowed only 'neon-captest-reuse', 'neon-captest-new', 'neon-captest-prefetch', 'rds-aurora', or 'rds-postgres'"
             exit 1
             ;;
         esac
@@ -220,8 +202,11 @@ jobs:
     - name: Set database options
       if: matrix.platform == 'neon-captest-prefetch'
       run: |
-        psql ${BENCHMARK_CONNSTR} -c "ALTER DATABASE main SET enable_seqscan_prefetch=on"
-        psql ${BENCHMARK_CONNSTR} -c "ALTER DATABASE main SET seqscan_prefetch_buffers=10"
+        DB_NAME=$(psql ${BENCHMARK_CONNSTR} --no-align --quiet -t -c "SELECT current_database()")
+
+        psql ${BENCHMARK_CONNSTR} -c "ALTER DATABASE ${DB_NAME} SET enable_seqscan_prefetch=on"
+        psql ${BENCHMARK_CONNSTR} -c "ALTER DATABASE ${DB_NAME} SET effective_io_concurrency=32"
+        psql ${BENCHMARK_CONNSTR} -c "ALTER DATABASE ${DB_NAME} SET maintenance_io_concurrency=32"
       env:
         BENCHMARK_CONNSTR: ${{ steps.set-up-connstr.outputs.connstr }}
 
@@ -264,20 +249,19 @@ jobs:
         VIP_VAP_ACCESS_TOKEN: "${{ secrets.VIP_VAP_ACCESS_TOKEN }}"
         PERF_TEST_RESULT_CONNSTR: "${{ secrets.PERF_TEST_RESULT_CONNSTR }}"
 
-    - name: Create Allure report
-      if: always()
-      uses: ./.github/actions/allure-report
-      with:
-        action: generate
-        build_type: ${{ env.BUILD_TYPE }}
-
     - name: Delete Neon Project
       if: ${{ steps.create-neon-project.outputs.project_id && always() }}
       uses: ./.github/actions/neon-project-delete
       with:
-        environment: dev
         project_id: ${{ steps.create-neon-project.outputs.project_id }}
-        api_key: ${{ secrets.NEON_CAPTEST_API_KEY }}
+        api_key: ${{ secrets.NEON_STAGING_API_KEY }}
+
+    - name: Create Allure report
+      if: success() || failure()
+      uses: ./.github/actions/allure-report
+      with:
+        action: generate
+        build_type: ${{ env.BUILD_TYPE }}
 
     - name: Post to a Slack channel
       if: ${{ github.event.schedule && failure() }}
@@ -287,3 +271,226 @@ jobs:
         slack-message: "Periodic perf testing ${{ matrix.platform }}: ${{ job.status }}\n${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}"
       env:
         SLACK_BOT_TOKEN: ${{ secrets.SLACK_BOT_TOKEN }}
+
+  clickbench-compare:
+    # ClichBench DB for rds-aurora and rds-Postgres deployed to the same clusters
+    # we use for performance testing in pgbench-compare.
+    # Run this job only when pgbench-compare is finished to avoid the intersection.
+    # We might change it after https://github.com/neondatabase/neon/issues/2900.
+    #
+    # *_CLICKBENCH_CONNSTR: Genuine ClickBench DB with ~100M rows
+    # *_CLICKBENCH_10M_CONNSTR: DB with the first 10M rows of ClickBench DB
+    if: success() || failure()
+    needs: [ pgbench-compare ]
+
+    strategy:
+      fail-fast: false
+      matrix:
+        # neon-captest-prefetch: We have pre-created projects with prefetch enabled
+        # rds-aurora: Aurora Postgres Serverless v2 with autoscaling from 0.5 to 2 ACUs
+        # rds-postgres: RDS Postgres db.m5.large instance (2 vCPU, 8 GiB) with gp3 EBS storage
+        platform: [ neon-captest-prefetch, rds-postgres, rds-aurora ]
+
+    env:
+      POSTGRES_DISTRIB_DIR: /tmp/neon/pg_install
+      DEFAULT_PG_VERSION: 14
+      TEST_OUTPUT: /tmp/test_output
+      BUILD_TYPE: remote
+      SAVE_PERF_REPORT: ${{ github.event.inputs.save_perf_report || ( github.ref == 'refs/heads/main' ) }}
+      PLATFORM: ${{ matrix.platform }}
+
+    runs-on: [ self-hosted, us-east-2, x64 ]
+    container:
+      image: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/rust:pinned
+      options: --init
+
+    timeout-minutes: 360 # 6h
+
+    steps:
+    - uses: actions/checkout@v3
+
+    - name: Download Neon artifact
+      uses: ./.github/actions/download
+      with:
+        name: neon-${{ runner.os }}-release-artifact
+        path: /tmp/neon/
+        prefix: latest
+
+    - name: Add Postgres binaries to PATH
+      run: |
+        ${POSTGRES_DISTRIB_DIR}/v${DEFAULT_PG_VERSION}/bin/pgbench --version
+        echo "${POSTGRES_DISTRIB_DIR}/v${DEFAULT_PG_VERSION}/bin" >> $GITHUB_PATH
+
+    - name: Set up Connection String
+      id: set-up-connstr
+      run: |
+        case "${PLATFORM}" in
+          neon-captest-prefetch)
+            CONNSTR=${{ secrets.BENCHMARK_CAPTEST_CLICKBENCH_10M_CONNSTR }}
+            ;;
+          rds-aurora)
+            CONNSTR=${{ secrets.BENCHMARK_RDS_AURORA_CLICKBENCH_10M_CONNSTR }}
+            ;;
+          rds-postgres)
+            CONNSTR=${{ secrets.BENCHMARK_RDS_POSTGRES_CLICKBENCH_10M_CONNSTR }}
+            ;;
+          *)
+            echo 2>&1 "Unknown PLATFORM=${PLATFORM}. Allowed only 'neon-captest-prefetch', 'rds-aurora', or 'rds-postgres'"
+            exit 1
+            ;;
+        esac
+
+        echo "connstr=${CONNSTR}" >> $GITHUB_OUTPUT
+
+        psql ${CONNSTR} -c "SELECT version();"
+
+    - name: Set database options
+      if: matrix.platform == 'neon-captest-prefetch'
+      run: |
+        DB_NAME=$(psql ${BENCHMARK_CONNSTR} --no-align --quiet -t -c "SELECT current_database()")
+
+        psql ${BENCHMARK_CONNSTR} -c "ALTER DATABASE ${DB_NAME} SET enable_seqscan_prefetch=on"
+        psql ${BENCHMARK_CONNSTR} -c "ALTER DATABASE ${DB_NAME} SET effective_io_concurrency=32"
+        psql ${BENCHMARK_CONNSTR} -c "ALTER DATABASE ${DB_NAME} SET maintenance_io_concurrency=32"
+      env:
+        BENCHMARK_CONNSTR: ${{ steps.set-up-connstr.outputs.connstr }}
+
+    - name: ClickBench benchmark
+      uses: ./.github/actions/run-python-test-set
+      with:
+        build_type: ${{ env.BUILD_TYPE }}
+        test_selection: performance/test_perf_olap.py
+        run_in_parallel: false
+        save_perf_report: ${{ env.SAVE_PERF_REPORT }}
+        extra_params: -m remote_cluster --timeout 21600 -k test_clickbench
+      env:
+        VIP_VAP_ACCESS_TOKEN: "${{ secrets.VIP_VAP_ACCESS_TOKEN }}"
+        PERF_TEST_RESULT_CONNSTR: "${{ secrets.PERF_TEST_RESULT_CONNSTR }}"
+        BENCHMARK_CONNSTR: ${{ steps.set-up-connstr.outputs.connstr }}
+
+    - name: Create Allure report
+      if: success() || failure()
+      uses: ./.github/actions/allure-report
+      with:
+        action: generate
+        build_type: ${{ env.BUILD_TYPE }}
+
+    - name: Post to a Slack channel
+      if: ${{ github.event.schedule && failure() }}
+      uses: slackapi/slack-github-action@v1
+      with:
+        channel-id: "C033QLM5P7D" # dev-staging-stream
+        slack-message: "Periodic OLAP perf testing ${{ matrix.platform }}: ${{ job.status }}\n${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}"
+      env:
+        SLACK_BOT_TOKEN: ${{ secrets.SLACK_BOT_TOKEN }}
+
+  tpch-compare:
+    # TCP-H DB for rds-aurora and rds-Postgres deployed to the same clusters
+    # we use for performance testing in pgbench-compare & clickbench-compare.
+    # Run this job only when clickbench-compare is finished to avoid the intersection.
+    # We might change it after https://github.com/neondatabase/neon/issues/2900.
+    #
+    # *_TPCH_S10_CONNSTR: DB generated with scale factor 10 (~10 GB)
+    if: success() || failure()
+    needs: [ clickbench-compare ]
+
+    strategy:
+      fail-fast: false
+      matrix:
+        # neon-captest-prefetch: We have pre-created projects with prefetch enabled
+        # rds-aurora: Aurora Postgres Serverless v2 with autoscaling from 0.5 to 2 ACUs
+        # rds-postgres: RDS Postgres db.m5.large instance (2 vCPU, 8 GiB) with gp3 EBS storage
+        platform: [ neon-captest-prefetch, rds-postgres, rds-aurora ]
+
+    env:
+      POSTGRES_DISTRIB_DIR: /tmp/neon/pg_install
+      DEFAULT_PG_VERSION: 14
+      TEST_OUTPUT: /tmp/test_output
+      BUILD_TYPE: remote
+      SAVE_PERF_REPORT: ${{ github.event.inputs.save_perf_report || ( github.ref == 'refs/heads/main' ) }}
+      PLATFORM: ${{ matrix.platform }}
+
+    runs-on: [ self-hosted, us-east-2, x64 ]
+    container:
+      image: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/rustlegacy:pinned
+      options: --init
+
+    timeout-minutes: 360 # 6h
+
+    steps:
+    - uses: actions/checkout@v3
+
+    - name: Download Neon artifact
+      uses: ./.github/actions/download
+      with:
+        name: neon-${{ runner.os }}-release-artifact
+        path: /tmp/neon/
+        prefix: latest
+
+    - name: Add Postgres binaries to PATH
+      run: |
+        ${POSTGRES_DISTRIB_DIR}/v${DEFAULT_PG_VERSION}/bin/pgbench --version
+        echo "${POSTGRES_DISTRIB_DIR}/v${DEFAULT_PG_VERSION}/bin" >> $GITHUB_PATH
+
+    - name: Set up Connection String
+      id: set-up-connstr
+      run: |
+        case "${PLATFORM}" in
+          neon-captest-prefetch)
+            CONNSTR=${{ secrets.BENCHMARK_CAPTEST_TPCH_S10_CONNSTR }}
+            ;;
+          rds-aurora)
+            CONNSTR=${{ secrets.BENCHMARK_RDS_AURORA_TPCH_S10_CONNSTR }}
+            ;;
+          rds-postgres)
+            CONNSTR=${{ secrets.BENCHMARK_RDS_POSTGRES_TPCH_S10_CONNSTR }}
+            ;;
+          *)
+            echo 2>&1 "Unknown PLATFORM=${PLATFORM}. Allowed only 'neon-captest-prefetch', 'rds-aurora', or 'rds-postgres'"
+            exit 1
+            ;;
+        esac
+
+        echo "connstr=${CONNSTR}" >> $GITHUB_OUTPUT
+
+        psql ${CONNSTR} -c "SELECT version();"
+
+    - name: Set database options
+      if: matrix.platform == 'neon-captest-prefetch'
+      run: |
+        DB_NAME=$(psql ${BENCHMARK_CONNSTR} --no-align --quiet -t -c "SELECT current_database()")
+
+        psql ${BENCHMARK_CONNSTR} -c "ALTER DATABASE ${DB_NAME} SET enable_seqscan_prefetch=on"
+        psql ${BENCHMARK_CONNSTR} -c "ALTER DATABASE ${DB_NAME} SET effective_io_concurrency=32"
+        psql ${BENCHMARK_CONNSTR} -c "ALTER DATABASE ${DB_NAME} SET maintenance_io_concurrency=32"
+      env:
+        BENCHMARK_CONNSTR: ${{ steps.set-up-connstr.outputs.connstr }}
+
+    - name: Run TPC-H benchmark
+      uses: ./.github/actions/run-python-test-set
+      with:
+        build_type: ${{ env.BUILD_TYPE }}
+        test_selection: performance/test_perf_olap.py
+        run_in_parallel: false
+        save_perf_report: ${{ env.SAVE_PERF_REPORT }}
+        extra_params: -m remote_cluster --timeout 21600 -k test_tpch
+      env:
+        VIP_VAP_ACCESS_TOKEN: "${{ secrets.VIP_VAP_ACCESS_TOKEN }}"
+        PERF_TEST_RESULT_CONNSTR: "${{ secrets.PERF_TEST_RESULT_CONNSTR }}"
+        BENCHMARK_CONNSTR: ${{ steps.set-up-connstr.outputs.connstr }}
+
+    - name: Create Allure report
+      if: success() || failure()
+      uses: ./.github/actions/allure-report
+      with:
+        action: generate
+        build_type: ${{ env.BUILD_TYPE }}
+
+    - name: Post to a Slack channel
+      if: ${{ github.event.schedule && failure() }}
+      uses: slackapi/slack-github-action@v1
+      with:
+        channel-id: "C033QLM5P7D" # dev-staging-stream
+        slack-message: "Periodic TPC-H perf testing ${{ matrix.platform }}: ${{ job.status }}\n${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}"
+      env:
+        SLACK_BOT_TOKEN: ${{ secrets.SLACK_BOT_TOKEN }}

.github/workflows/build_and_test.yml

@@ -18,8 +18,8 @@ env:
 
 jobs:
   tag:
-    runs-on: dev
-    container: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/base:latest
+    runs-on: [ self-hosted, dev, x64 ]
+    container: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/base:pinned
     outputs:
       build-tag: ${{steps.build-tag.outputs.tag}}
 
@@ -46,7 +46,7 @@ jobs:
         id: build-tag
 
   build-neon:
-    runs-on: dev
+    runs-on: [ self-hosted, dev, x64 ]
     container:
       image: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/rust:pinned
       options: --init
@@ -236,7 +236,7 @@ jobs:
         uses: ./.github/actions/save-coverage-data
 
   regress-tests:
-    runs-on: dev
+    runs-on: [ self-hosted, dev, x64 ]
     container:
       image: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/rust:pinned
       options: --init
@@ -268,34 +268,8 @@ jobs:
         if: matrix.build_type == 'debug'
         uses: ./.github/actions/save-coverage-data
 
-  upload-latest-artifacts:
-    runs-on: dev
-    container:
-      image: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/rust:pinned
-      options: --init
-    needs: [ regress-tests ]
-    if: github.ref_name == 'main'
-    steps:
-      - name: Copy Neon artifact to the latest directory
-        shell: bash -euxo pipefail {0}
-        env:
-          BUCKET: neon-github-public-dev
-          PREFIX: artifacts/${{ github.run_id }}
-        run: |
-          for build_type in debug release; do
-            FILENAME=neon-${{ runner.os }}-${build_type}-artifact.tar.zst
-
-            S3_KEY=$(aws s3api list-objects-v2 --bucket ${BUCKET} --prefix ${PREFIX} | jq -r '.Contents[].Key' | grep ${FILENAME} | sort --version-sort | tail -1 || true)
-            if [ -z "${S3_KEY}" ]; then
-              echo 2>&1 "Neither s3://${BUCKET}/${PREFIX}/${FILENAME} nor its version from previous attempts exist"
-              exit 1
-            fi
-
-            time aws s3 cp --only-show-errors s3://${BUCKET}/${S3_KEY} s3://${BUCKET}/artifacts/latest/${FILENAME}
-          done
-
   benchmarks:
-    runs-on: dev
+    runs-on: [ self-hosted, dev, x64 ]
     container:
       image: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/rust:pinned
       options: --init
@@ -326,12 +300,12 @@ jobs:
       # while coverage is currently collected for the debug ones
 
   merge-allure-report:
-    runs-on: dev
+    runs-on: [ self-hosted, dev, x64 ]
     container:
       image: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/rust:pinned
       options: --init
     needs: [ regress-tests, benchmarks ]
-    if: always()
+    if: ${{ !cancelled() }}
     strategy:
       fail-fast: false
       matrix:
@@ -364,7 +338,7 @@ jobs:
           DATABASE_URL="$TEST_RESULT_CONNSTR" poetry run python3 scripts/ingest_regress_test_result.py --revision ${SHA} --reference ${GITHUB_REF} --build-type ${BUILD_TYPE} --ingest suites.json
 
   coverage-report:
-    runs-on: dev
+    runs-on: [ self-hosted, dev, x64 ]
     container:
       image: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/rust:pinned
       options: --init
@@ -441,15 +415,19 @@ jobs:
         shell: bash -euxo pipefail {0}
 
   trigger-e2e-tests:
-    runs-on: dev
+    runs-on: [ self-hosted, dev, x64 ]
     container:
       image: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/base:pinned
       options: --init
-    needs: [ build-neon ]
+    needs: [ push-docker-hub, tag ]
     steps:
       - name: Set PR's status to pending and request a remote CI test
         run: |
+          # For pull requests, GH Actions set "github.sha" variable to point at a fake merge commit
+          # but we need to use a real sha of a latest commit in the PR's branch for the e2e job,
+          # to place a job run status update later.
           COMMIT_SHA=${{ github.event.pull_request.head.sha }}
+          # For non-PR kinds of runs, the above will produce an empty variable, pick the original sha value for those
           COMMIT_SHA=${COMMIT_SHA:-${{ github.sha }}}
 
           REMOTE_REPO="${{ github.repository_owner }}/cloud"
@@ -475,12 +453,14 @@ jobs:
               \"inputs\": {
                 \"ci_job_name\": \"neon-cloud-e2e\",
                 \"commit_hash\": \"$COMMIT_SHA\",
-                \"remote_repo\": \"${{ github.repository }}\"
+                \"remote_repo\": \"${{ github.repository }}\",
+                \"storage_image_tag\": \"${{ needs.tag.outputs.build-tag }}\",
+                \"compute_image_tag\": \"${{ needs.tag.outputs.build-tag }}\"
               }
             }"
 
   neon-image:
-    runs-on: dev
+    runs-on: [ self-hosted, dev, x64 ]
     needs: [ tag ]
     container: gcr.io/kaniko-project/executor:v1.9.0-debug
 
@@ -498,7 +478,7 @@ jobs:
         run: /kaniko/executor --snapshotMode=redo --cache=true --cache-repo 369495373322.dkr.ecr.eu-central-1.amazonaws.com/cache --snapshotMode=redo --context . --build-arg GIT_VERSION=${{ github.sha }} --destination 369495373322.dkr.ecr.eu-central-1.amazonaws.com/neon:${{needs.tag.outputs.build-tag}}
 
   compute-tools-image:
-    runs-on: dev
+    runs-on: [ self-hosted, dev, x64 ]
     needs: [ tag ]
     container: gcr.io/kaniko-project/executor:v1.9.0-debug
 
@@ -512,28 +492,8 @@ jobs:
       - name: Kaniko build compute tools
         run: /kaniko/executor --snapshotMode=redo --cache=true --cache-repo 369495373322.dkr.ecr.eu-central-1.amazonaws.com/cache --snapshotMode=redo --context . --build-arg GIT_VERSION=${{ github.sha }} --dockerfile Dockerfile.compute-tools --destination 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-tools:${{needs.tag.outputs.build-tag}}
 
-  compute-node-image:
-    runs-on: dev
-    container: gcr.io/kaniko-project/executor:v1.9.0-debug
-    needs: [ tag ]
-    steps:
-      - name: Checkout
-        uses: actions/checkout@v1 # v3 won't work with kaniko
-        with:
-          submodules: true
-          fetch-depth: 0
-
-      - name: Configure ECR login
-        run: echo "{\"credsStore\":\"ecr-login\"}" > /kaniko/.docker/config.json
-
-        # compute-node uses postgres 14, which is default now
-        # cloud repo depends on this image name, thus duplicating it
-        # remove compute-node when cloud repo is updated
-      - name: Kaniko build compute node with extensions v14 (compatibility)
-        run: /kaniko/executor --skip-unused-stages --snapshotMode=redo --cache=true --cache-repo 369495373322.dkr.ecr.eu-central-1.amazonaws.com/cache --snapshotMode=redo --context . --build-arg GIT_VERSION=${{ github.sha }} --dockerfile Dockerfile.compute-node-v14 --destination 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node:${{needs.tag.outputs.build-tag}}
-
   compute-node-image-v14:
-    runs-on: dev
+    runs-on: [ self-hosted, dev, x64 ]
     container: gcr.io/kaniko-project/executor:v1.9.0-debug
     needs: [ tag ]
     steps:
@@ -549,9 +509,8 @@ jobs:
       - name: Kaniko build compute node with extensions v14
         run: /kaniko/executor --skip-unused-stages  --snapshotMode=redo --cache=true --cache-repo 369495373322.dkr.ecr.eu-central-1.amazonaws.com/cache  --context . --build-arg GIT_VERSION=${{ github.sha }} --dockerfile Dockerfile.compute-node-v14 --destination 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node-v14:${{needs.tag.outputs.build-tag}}
 
-
   compute-node-image-v15:
-    runs-on: dev
+    runs-on: [ self-hosted, dev, x64 ]
     container: gcr.io/kaniko-project/executor:v1.9.0-debug
     needs: [ tag ]
     steps:
@@ -567,18 +526,58 @@ jobs:
       - name: Kaniko build compute node with extensions v15
         run: /kaniko/executor --skip-unused-stages --snapshotMode=redo --cache=true --cache-repo 369495373322.dkr.ecr.eu-central-1.amazonaws.com/cache --context . --build-arg GIT_VERSION=${{ github.sha }} --dockerfile Dockerfile.compute-node-v15 --destination 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node-v15:${{needs.tag.outputs.build-tag}}
 
+  test-images:
+    needs: [ tag, neon-image, compute-node-image-v14, compute-node-image-v15, compute-tools-image ]
+    runs-on: [ self-hosted, dev, x64 ]
+
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v3
+        with:
+          fetch-depth: 0
+
+      # `neondatabase/neon` contains multiple binaries, all of them use the same input for the version into the same version formatting library.
+      # Pick pageserver as currently the only binary with extra "version" features printed in the string to verify.
+      # Regular pageserver version string looks like
+      #   Neon page server git-env:32d14403bd6ab4f4520a94cbfd81a6acef7a526c failpoints: true, features: []
+      # Bad versions might loop like:
+      #   Neon page server git-env:local failpoints: true, features: ["testing"]
+      # Ensure that we don't have bad versions.
+      - name: Verify image versions
+        shell: bash # ensure no set -e for better error messages
+        run: |
+          pageserver_version=$(docker run --rm 369495373322.dkr.ecr.eu-central-1.amazonaws.com/neon:${{needs.tag.outputs.build-tag}} "/bin/sh" "-c" "/usr/local/bin/pageserver --version")
+
+          echo "Pageserver version string: $pageserver_version"
+
+          if ! echo "$pageserver_version" | grep -qv 'git-env:local' ; then
+            echo "Pageserver version should not be the default Dockerfile one"
+            exit 1
+          fi
+
+          if ! echo "$pageserver_version" | grep -qv '"testing"' ; then
+            echo "Pageserver version should have no testing feature enabled"
+            exit 1
+          fi
+
+      - name: Verify docker-compose example
+        run: env REPOSITORY=369495373322.dkr.ecr.eu-central-1.amazonaws.com TAG=${{needs.tag.outputs.build-tag}} ./docker-compose/docker_compose_test.sh
+
+      - name: Print logs and clean up
+        if: always()
+        run: |
+          docker compose -f ./docker-compose/docker-compose.yml logs || 0
+          docker compose -f ./docker-compose/docker-compose.yml down
+
   promote-images:
-    runs-on: dev
-    needs: [ tag, neon-image, compute-node-image, compute-node-image-v14, compute-node-image-v15, compute-tools-image ]
+    runs-on: [ self-hosted, dev, x64 ]
+    needs: [ tag, test-images ]
     if: github.event_name != 'workflow_dispatch'
     container: amazon/aws-cli
     strategy:
       fail-fast: false
       matrix:
-        # compute-node uses postgres 14, which is default now
-        # cloud repo depends on this image name, thus duplicating it
-        # remove compute-node when cloud repo is updated
-        name: [ neon, compute-node, compute-node-v14, compute-node-v15, compute-tools ]
+        name: [ neon, compute-node-v14, compute-node-v15, compute-tools ]
 
     steps:
       - name: Promote image to latest
@@ -587,7 +586,7 @@ jobs:
           aws ecr put-image --repository-name ${{ matrix.name }} --image-tag latest --image-manifest "$MANIFEST"
 
   push-docker-hub:
-    runs-on: dev
+    runs-on: [ self-hosted, dev, x64 ]
     needs: [ promote-images, tag ]
     container: golang:1.19-bullseye
 
@@ -608,9 +607,6 @@ jobs:
       - name: Pull compute tools image from ECR
         run: crane pull 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-tools:${{needs.tag.outputs.build-tag}} compute-tools
 
-      - name: Pull compute node image from ECR
-        run: crane pull 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node:${{needs.tag.outputs.build-tag}} compute-node
-
       - name: Pull compute node v14 image from ECR
         run: crane pull 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node-v14:${{needs.tag.outputs.build-tag}} compute-node-v14
 
@@ -627,7 +623,6 @@ jobs:
         run: |
           crane copy 369495373322.dkr.ecr.eu-central-1.amazonaws.com/neon:${{needs.tag.outputs.build-tag}} 093970136003.dkr.ecr.eu-central-1.amazonaws.com/neon:latest
           crane copy 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-tools:${{needs.tag.outputs.build-tag}} 093970136003.dkr.ecr.eu-central-1.amazonaws.com/compute-tools:latest
-          crane copy 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node:${{needs.tag.outputs.build-tag}} 093970136003.dkr.ecr.eu-central-1.amazonaws.com/compute-node:latest
           crane copy 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node-v14:${{needs.tag.outputs.build-tag}} 093970136003.dkr.ecr.eu-central-1.amazonaws.com/compute-node-v14:latest
           crane copy 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node-v15:${{needs.tag.outputs.build-tag}} 093970136003.dkr.ecr.eu-central-1.amazonaws.com/compute-node-v15:latest
 
@@ -643,9 +638,6 @@ jobs:
       - name: Push compute tools image to Docker Hub
         run: crane push compute-tools neondatabase/compute-tools:${{needs.tag.outputs.build-tag}}
 
-      - name: Push compute node image to Docker Hub
-        run: crane push compute-node neondatabase/compute-node:${{needs.tag.outputs.build-tag}}
-
       - name: Push compute node v14 image to Docker Hub
         run: crane push compute-node-v14 neondatabase/compute-node-v14:${{needs.tag.outputs.build-tag}}
 
@@ -662,7 +654,6 @@ jobs:
         run: |
           crane tag neondatabase/neon:${{needs.tag.outputs.build-tag}} latest
           crane tag neondatabase/compute-tools:${{needs.tag.outputs.build-tag}} latest
-          crane tag neondatabase/compute-node:${{needs.tag.outputs.build-tag}} latest
           crane tag neondatabase/compute-node-v14:${{needs.tag.outputs.build-tag}} latest
           crane tag neondatabase/compute-node-v15:${{needs.tag.outputs.build-tag}} latest
 
@@ -677,11 +668,11 @@ jobs:
       - id: set-matrix
         run: |
           if [[ "$GITHUB_REF_NAME" == "main" ]]; then
-            STAGING='{"env_name": "staging", "proxy_job": "neon-proxy", "proxy_config": "staging.proxy", "kubeconfig_secret": "STAGING_KUBECONFIG_DATA", "console_api_key_secret": "NEON_STAGING_API_KEY"}'
-            NEON_STRESS='{"env_name": "neon-stress", "proxy_job": "neon-stress-proxy", "proxy_config": "neon-stress.proxy", "kubeconfig_secret": "NEON_STRESS_KUBECONFIG_DATA", "console_api_key_secret": "NEON_CAPTEST_API_KEY"}'
+            STAGING='{"env_name": "staging", "proxy_job": "neon-proxy", "proxy_config": "staging.proxy", "storage_broker_ns": "neon-storage-broker", "storage_broker_config": "staging.neon-storage-broker", "kubeconfig_secret": "STAGING_KUBECONFIG_DATA", "console_api_key_secret": "NEON_STAGING_API_KEY"}'
+            NEON_STRESS='{"env_name": "neon-stress", "proxy_job": "neon-stress-proxy", "proxy_config": "neon-stress.proxy", "storage_broker_ns": "neon-stress-storage-broker", "storage_broker_config": "neon-stress.neon-storage-broker", "kubeconfig_secret": "NEON_STRESS_KUBECONFIG_DATA", "console_api_key_secret": "NEON_CAPTEST_API_KEY", storage_broker_config: }'
             echo "include=[$STAGING, $NEON_STRESS]" >> $GITHUB_OUTPUT
           elif [[ "$GITHUB_REF_NAME" == "release" ]]; then
-            PRODUCTION='{"env_name": "production", "proxy_job": "neon-proxy", "proxy_config": "production.proxy", "kubeconfig_secret": "PRODUCTION_KUBECONFIG_DATA", "console_api_key_secret": "NEON_PRODUCTION_API_KEY"}'
+            PRODUCTION='{"env_name": "production", "proxy_job": "neon-proxy", "proxy_config": "production.proxy", "storage_broker_ns": "neon-storage-broker", "storage_broker_config": "production.neon-storage-broker", "kubeconfig_secret": "PRODUCTION_KUBECONFIG_DATA", "console_api_key_secret": "NEON_PRODUCTION_API_KEY"}'
             echo "include=[$PRODUCTION]" >> $GITHUB_OUTPUT
           else
             echo "GITHUB_REF_NAME (value '$GITHUB_REF_NAME') is not set to either 'main' or 'release'"
@@ -741,11 +732,11 @@ jobs:
           ssh-add ssh-key
           rm -f ssh-key ssh-key-cert.pub
           ansible-galaxy collection install sivel.toiletwater
-          ansible-playbook deploy.yaml -i ${{ matrix.env_name }}.hosts.yaml -e CONSOLE_API_TOKEN=${{ secrets[matrix.console_api_key_secret] }}
+          ansible-playbook deploy.yaml -i ${{ matrix.env_name }}.hosts.yaml -e CONSOLE_API_TOKEN=${{ secrets[matrix.console_api_key_secret] }} -e SENTRY_URL_PAGESERVER=${{ secrets.SENTRY_URL_PAGESERVER }} -e SENTRY_URL_SAFEKEEPER=${{ secrets.SENTRY_URL_SAFEKEEPER }}
           rm -f neon_install.tar.gz .neon_current_version
 
   deploy-new:
-    runs-on: dev
+    runs-on: [ self-hosted, dev, x64 ]
     container: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/ansible:pinned
     # We need both storage **and** compute images for deploy, because control plane picks the compute version based on the storage version.
     # If it notices a fresh storage it may bump the compute version. And if compute image failed to build it may break things badly
@@ -770,7 +761,6 @@ jobs:
         run: |
           export DOCKER_TAG=${{needs.tag.outputs.build-tag}}
           cd "$(pwd)/.github/ansible"
-
           if [[ "$GITHUB_REF_NAME" == "main" ]]; then
             ./get_binaries.sh
           elif [[ "$GITHUB_REF_NAME" == "release" ]]; then
@@ -779,9 +769,41 @@ jobs:
             echo "GITHUB_REF_NAME (value '$GITHUB_REF_NAME') is not set to either 'main' or 'release'"
             exit 1
           fi
+          ansible-galaxy collection install sivel.toiletwater
+          ansible-playbook deploy.yaml -i staging.${{ matrix.target_region }}.hosts.yaml -e @ssm_config -e CONSOLE_API_TOKEN=${{ secrets.NEON_STAGING_API_KEY }} -e SENTRY_URL_PAGESERVER=${{ secrets.SENTRY_URL_PAGESERVER }} -e SENTRY_URL_SAFEKEEPER=${{ secrets.SENTRY_URL_SAFEKEEPER }}
+          rm -f neon_install.tar.gz .neon_current_version
+
+  deploy-pr-test-new:
+    runs-on: [ self-hosted, dev, x64 ]
+    container: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/ansible:pinned
+    # We need both storage **and** compute images for deploy, because control plane picks the compute version based on the storage version.
+    # If it notices a fresh storage it may bump the compute version. And if compute image failed to build it may break things badly
+    needs: [ push-docker-hub, tag, regress-tests ]
+    if: |
+      contains(github.event.pull_request.labels.*.name, 'deploy-test-storage') && 
+      github.event_name != 'workflow_dispatch'
+    defaults:
+      run:
+        shell: bash
+    strategy:
+      matrix:
+        target_region: [ eu-west-1 ]
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v3
+        with:
+          submodules: true
+          fetch-depth: 0
+
+      - name: Redeploy
+        run: |
+          export DOCKER_TAG=${{needs.tag.outputs.build-tag}}
+          cd "$(pwd)/.github/ansible"
+
+          ./get_binaries.sh
 
           ansible-galaxy collection install sivel.toiletwater
-          ansible-playbook deploy.yaml -i staging.${{ matrix.target_region }}.hosts.yaml -e @ssm_config -e CONSOLE_API_TOKEN=${{secrets.NEON_STAGING_API_KEY}}
+          ansible-playbook deploy.yaml -i staging.${{ matrix.target_region }}.hosts.yaml -e @ssm_config -e CONSOLE_API_TOKEN=${{ secrets.NEON_STAGING_API_KEY }} -e SENTRY_URL_PAGESERVER=${{ secrets.SENTRY_URL_PAGESERVER }} -e SENTRY_URL_SAFEKEEPER=${{ secrets.SENTRY_URL_SAFEKEEPER }}
           rm -f neon_install.tar.gz .neon_current_version
 
   deploy-prod-new:
@@ -789,7 +811,7 @@ jobs:
     container: 093970136003.dkr.ecr.eu-central-1.amazonaws.com/ansible:latest
     # We need both storage **and** compute images for deploy, because control plane picks the compute version based on the storage version.
     # If it notices a fresh storage it may bump the compute version. And if compute image failed to build it may break things badly
-    needs: [ push-docker-hub, calculate-deploy-targets, tag, regress-tests ]
+    needs: [ push-docker-hub, tag, regress-tests ]
     if: |
       (github.ref_name == 'release') &&
       github.event_name != 'workflow_dispatch'
@@ -821,11 +843,11 @@ jobs:
           fi
 
           ansible-galaxy collection install sivel.toiletwater
-          ansible-playbook deploy.yaml -i prod.${{ matrix.target_region }}.hosts.yaml -e @ssm_config -e CONSOLE_API_TOKEN=${{secrets.NEON_PRODUCTION_API_KEY}}
+          ansible-playbook deploy.yaml -i prod.${{ matrix.target_region }}.hosts.yaml -e @ssm_config -e CONSOLE_API_TOKEN=${{ secrets.NEON_PRODUCTION_API_KEY }} -e SENTRY_URL_PAGESERVER=${{ secrets.SENTRY_URL_PAGESERVER }} -e SENTRY_URL_SAFEKEEPER=${{ secrets.SENTRY_URL_SAFEKEEPER }}
           rm -f neon_install.tar.gz .neon_current_version
 
   deploy-proxy:
-    runs-on: dev
+    runs-on: [ self-hosted, dev, x64 ]
     container: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/base:latest
     # Compute image isn't strictly required for proxy deploy, but let's still wait for it to run all deploy jobs consistently.
     needs: [ push-docker-hub, calculate-deploy-targets, tag, regress-tests ]
@@ -863,14 +885,54 @@ jobs:
       - name: Re-deploy proxy
         run: |
           DOCKER_TAG=${{needs.tag.outputs.build-tag}}
-          helm upgrade ${{ matrix.proxy_job }}       neondatabase/neon-proxy --namespace neon-proxy --install -f .github/helm-values/${{ matrix.proxy_config }}.yaml --set image.tag=${DOCKER_TAG} --wait --timeout 15m0s
-          helm upgrade ${{ matrix.proxy_job }}-scram neondatabase/neon-proxy --namespace neon-proxy --install -f .github/helm-values/${{ matrix.proxy_config }}-scram.yaml --set image.tag=${DOCKER_TAG} --wait --timeout 15m0s
+          helm upgrade ${{ matrix.proxy_job }}       neondatabase/neon-proxy --namespace neon-proxy --install -f .github/helm-values/${{ matrix.proxy_config }}.yaml       --set image.tag=${DOCKER_TAG} --set settings.sentryUrl=${{ secrets.SENTRY_URL_PROXY }} --wait --timeout 15m0s
+          helm upgrade ${{ matrix.proxy_job }}-scram neondatabase/neon-proxy --namespace neon-proxy --install -f .github/helm-values/${{ matrix.proxy_config }}-scram.yaml --set image.tag=${DOCKER_TAG} --set settings.sentryUrl=${{ secrets.SENTRY_URL_PROXY }} --wait --timeout 15m0s
 
-  deploy-proxy-new:
-    runs-on: dev
-    container: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/ansible:pinned
+  deploy-storage-broker-staging:
+    runs-on: [ self-hosted, dev, x64 ]
+    container: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/base:latest
     # Compute image isn't strictly required for proxy deploy, but let's still wait for it to run all deploy jobs consistently.
     needs: [ push-docker-hub, calculate-deploy-targets, tag, regress-tests ]
+    if: |
+      (github.ref_name == 'main' || github.ref_name == 'release') &&
+      github.event_name != 'workflow_dispatch'
+    defaults:
+      run:
+        shell: bash
+    strategy:
+      matrix:
+        include: ${{fromJSON(needs.calculate-deploy-targets.outputs.matrix-include)}}
+    env:
+      KUBECONFIG: .kubeconfig
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v3
+        with:
+          submodules: true
+          fetch-depth: 0
+
+      - name: Add curl
+        run: apt update && apt install curl -y
+
+      - name: Store kubeconfig file
+        run: |
+          echo "${{ secrets[matrix.kubeconfig_secret] }}" | base64 --decode > ${KUBECONFIG}
+          chmod 0600 ${KUBECONFIG}
+
+      - name: Setup helm v3
+        run: |
+          curl -s https://raw.githubusercontent.com/helm/helm/main/scripts/get-helm-3 | bash
+          helm repo add neondatabase https://neondatabase.github.io/helm-charts
+
+      - name: Deploy storage-broker
+        run:
+          helm upgrade neon-storage-broker neondatabase/neon-storage-broker --namespace ${{ matrix.storage_broker_ns }} --create-namespace --install --atomic -f .github/helm-values/${{ matrix.storage_broker_config }}.yaml --set image.tag=${{ needs.tag.outputs.build-tag }} --wait --timeout 5m0s
+
+  deploy-proxy-new:
+    runs-on: [ self-hosted, dev, x64 ]
+    container: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/ansible:pinned
+    # Compute image isn't strictly required for proxy deploy, but let's still wait for it to run all deploy jobs consistently.
+    needs: [ push-docker-hub, tag, regress-tests ]
     if: |
       (github.ref_name == 'main') &&
       github.event_name != 'workflow_dispatch'
@@ -882,6 +944,12 @@ jobs:
         include:
           - target_region:  us-east-2
             target_cluster: dev-us-east-2-beta
+            deploy_link_proxy: true
+            deploy_legacy_scram_proxy: true
+          - target_region:  eu-west-1
+            target_cluster: dev-eu-west-1-zeta
+            deploy_link_proxy: false
+            deploy_legacy_scram_proxy: false
     steps:
       - name: Checkout
         uses: actions/checkout@v3
@@ -894,16 +962,62 @@ jobs:
           helm repo add neondatabase https://neondatabase.github.io/helm-charts
           aws --region ${{ matrix.target_region }} eks update-kubeconfig --name  ${{ matrix.target_cluster }}
 
-      - name: Re-deploy proxy
+      - name: Re-deploy scram proxy
         run: |
           DOCKER_TAG=${{needs.tag.outputs.build-tag}}
-          helm upgrade neon-proxy-scram neondatabase/neon-proxy --namespace neon-proxy --create-namespace --install -f .github/helm-values/${{ matrix.target_cluster }}.neon-proxy-scram.yaml --set image.tag=${DOCKER_TAG} --wait --timeout 15m0s
+          helm upgrade neon-proxy-scram neondatabase/neon-proxy --namespace neon-proxy --create-namespace --install -f .github/helm-values/${{ matrix.target_cluster }}.neon-proxy-scram.yaml --set image.tag=${DOCKER_TAG} --set settings.sentryUrl=${{ secrets.SENTRY_URL_PROXY }} --wait --timeout 15m0s
+
+      - name: Re-deploy link proxy
+        if: matrix.deploy_link_proxy
+        run: |
+          DOCKER_TAG=${{needs.tag.outputs.build-tag}}
+          helm upgrade neon-proxy-link neondatabase/neon-proxy --namespace neon-proxy --create-namespace --install -f .github/helm-values/${{ matrix.target_cluster }}.neon-proxy-link.yaml --set image.tag=${DOCKER_TAG} --set settings.sentryUrl=${{ secrets.SENTRY_URL_PROXY }} --wait --timeout 15m0s
+
+      - name: Re-deploy legacy scram proxy
+        if: matrix.deploy_legacy_scram_proxy
+        run: |
+          DOCKER_TAG=${{needs.tag.outputs.build-tag}}
+          helm upgrade neon-proxy-scram-legacy neondatabase/neon-proxy --namespace neon-proxy --create-namespace --install -f .github/helm-values/${{ matrix.target_cluster }}.neon-proxy-scram-legacy.yaml --set image.tag=${DOCKER_TAG} --set settings.sentryUrl=${{ secrets.SENTRY_URL_PROXY }} --wait --timeout 15m0s
+
+  deploy-storage-broker-dev-new:
+    runs-on: [ self-hosted, dev, x64 ]
+    container: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/ansible:pinned
+    # Compute image isn't strictly required for proxy deploy, but let's still wait for it to run all deploy jobs consistently.
+    needs: [ push-docker-hub, tag, regress-tests ]
+    if: |
+      (github.ref_name == 'main') &&
+      github.event_name != 'workflow_dispatch'
+    defaults:
+      run:
+        shell: bash
+    strategy:
+      matrix:
+        include:
+          - target_region:  us-east-2
+            target_cluster: dev-us-east-2-beta
+          - target_region:  eu-west-1
+            target_cluster: dev-eu-west-1-zeta
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v3
+        with:
+          submodules: true
+          fetch-depth: 0
+
+      - name: Configure environment
+        run: |
+          helm repo add neondatabase https://neondatabase.github.io/helm-charts
+          aws --region ${{ matrix.target_region }} eks update-kubeconfig --name  ${{ matrix.target_cluster }}
+
+      - name: Deploy storage-broker
+        run:
+          helm upgrade neon-storage-broker neondatabase/neon-storage-broker --namespace neon-storage-broker --create-namespace --install --atomic -f .github/helm-values/${{ matrix.target_cluster }}.neon-storage-broker.yaml --set image.tag=${{ needs.tag.outputs.build-tag }} --wait --timeout 5m0s
 
   deploy-proxy-prod-new:
     runs-on: prod
     container: 093970136003.dkr.ecr.eu-central-1.amazonaws.com/ansible:latest
     # Compute image isn't strictly required for proxy deploy, but let's still wait for it to run all deploy jobs consistently.
-    needs: [ push-docker-hub, calculate-deploy-targets, tag, regress-tests ]
+    needs: [ push-docker-hub, tag, regress-tests ]
     if: |
       (github.ref_name == 'release') &&
       github.event_name != 'workflow_dispatch'
@@ -915,6 +1029,8 @@ jobs:
         include:
           - target_region:  us-east-2
             target_cluster: prod-us-east-2-delta
+          - target_region:  us-west-2
+            target_cluster: prod-us-west-2-eta
           - target_region: eu-central-1
             target_cluster: prod-eu-central-1-gamma
           - target_region: ap-southeast-1
@@ -934,10 +1050,48 @@ jobs:
       - name: Re-deploy proxy
         run: |
           DOCKER_TAG=${{needs.tag.outputs.build-tag}}
-          helm upgrade neon-proxy-scram neondatabase/neon-proxy --namespace neon-proxy --create-namespace --install -f .github/helm-values/${{ matrix.target_cluster }}.neon-proxy-scram.yaml --set image.tag=${DOCKER_TAG} --wait --timeout 15m0s
+          helm upgrade neon-proxy-scram neondatabase/neon-proxy --namespace neon-proxy --create-namespace --install -f .github/helm-values/${{ matrix.target_cluster }}.neon-proxy-scram.yaml --set image.tag=${DOCKER_TAG} --set settings.sentryUrl=${{ secrets.SENTRY_URL_PROXY }} --wait --timeout 15m0s
 
-  promote-compatibility-test-snapshot:
-    runs-on: dev
+  deploy-storage-broker-prod-new:
+    runs-on: prod
+    container: 093970136003.dkr.ecr.eu-central-1.amazonaws.com/ansible:latest
+    # Compute image isn't strictly required for proxy deploy, but let's still wait for it to run all deploy jobs consistently.
+    needs: [ push-docker-hub, tag, regress-tests ]
+    if: |
+      (github.ref_name == 'release') &&
+      github.event_name != 'workflow_dispatch'
+    defaults:
+      run:
+        shell: bash
+    strategy:
+      matrix:
+        include:
+          - target_region:  us-east-2
+            target_cluster: prod-us-east-2-delta
+          - target_region:  us-west-2
+            target_cluster: prod-us-west-2-eta
+          - target_region: eu-central-1
+            target_cluster: prod-eu-central-1-gamma
+          - target_region: ap-southeast-1
+            target_cluster: prod-ap-southeast-1-epsilon
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v3
+        with:
+          submodules: true
+          fetch-depth: 0
+
+      - name: Configure environment
+        run: |
+          helm repo add neondatabase https://neondatabase.github.io/helm-charts
+          aws --region ${{ matrix.target_region }} eks update-kubeconfig --name  ${{ matrix.target_cluster }}
+
+      - name: Deploy storage-broker
+        run:
+          helm upgrade neon-storage-broker neondatabase/neon-storage-broker --namespace neon-storage-broker --create-namespace --install --atomic -f .github/helm-values/${{ matrix.target_cluster }}.neon-storage-broker.yaml --set image.tag=${{ needs.tag.outputs.build-tag }} --wait --timeout 5m0s
+
+  promote-compatibility-data:
+    runs-on: [ self-hosted, dev, x64 ]
     container:
       image: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/rust:pinned
       options: --init
@@ -950,9 +1104,24 @@ jobs:
           BUCKET: neon-github-public-dev
           PREFIX: artifacts/latest
         run: |
+          # Update compatibility snapshot for the release
           for build_type in debug release; do
             OLD_FILENAME=compatibility-snapshot-${build_type}-pg14-${GITHUB_RUN_ID}.tar.zst
             NEW_FILENAME=compatibility-snapshot-${build_type}-pg14.tar.zst
 
             time aws s3 mv --only-show-errors s3://${BUCKET}/${PREFIX}/${OLD_FILENAME} s3://${BUCKET}/${PREFIX}/${NEW_FILENAME}
           done
+
+          # Update Neon artifact for the release (reuse already uploaded artifact)
+          for build_type in debug release; do
+            OLD_PREFIX=artifacts/${GITHUB_RUN_ID}
+            FILENAME=neon-${{ runner.os }}-${build_type}-artifact.tar.zst
+
+            S3_KEY=$(aws s3api list-objects-v2 --bucket ${BUCKET} --prefix ${OLD_PREFIX} | jq -r '.Contents[].Key' | grep ${FILENAME} | sort --version-sort | tail -1 || true)
+            if [ -z "${S3_KEY}" ]; then
+              echo 2>&1 "Neither s3://${BUCKET}/${OLD_PREFIX}/${FILENAME} nor its version from previous attempts exist"
+              exit 1
+            fi
+
+            time aws s3 cp --only-show-errors s3://${BUCKET}/${S3_KEY} s3://${BUCKET}/${PREFIX}/${FILENAME}
+          done

.github/workflows/codestyle.yml

@@ -48,11 +48,11 @@ jobs:
         if: matrix.os == 'ubuntu-latest'
         run: |
           sudo apt update
-          sudo apt install build-essential libreadline-dev zlib1g-dev flex bison libseccomp-dev libssl-dev
+          sudo apt install build-essential libreadline-dev zlib1g-dev flex bison libseccomp-dev libssl-dev protobuf-compiler
 
       - name: Install macOS postgres dependencies
         if: matrix.os == 'macos-latest'
-        run: brew install flex bison openssl
+        run: brew install flex bison openssl protobuf
 
       - name: Set pg 14 revision for caching
         id: pg_v14_rev
@@ -115,7 +115,7 @@ jobs:
         run: cargo build --locked --all --all-targets
 
   check-rust-dependencies:
-    runs-on: dev
+    runs-on: [ self-hosted, dev, x64 ]
     container:
       image: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/rust:pinned
       options: --init

.github/workflows/pg_clients.yml

@@ -23,6 +23,7 @@ jobs:
     runs-on: [ ubuntu-latest ]
 
     env:
+      DEFAULT_PG_VERSION: 14
       TEST_OUTPUT: /tmp/test_output
 
     steps:
@@ -51,8 +52,8 @@ jobs:
       id: create-neon-project
       uses: ./.github/actions/neon-project-create
       with:
-        environment: staging
         api_key: ${{ secrets.NEON_STAGING_API_KEY }}
+        postgres_version: ${{ env.DEFAULT_PG_VERSION }}
 
     - name: Run pytest
       env:
@@ -63,7 +64,7 @@ jobs:
       run: |
         # Test framework expects we have psql binary;
         # but since we don't really need it in this test, let's mock it
-        mkdir -p "$POSTGRES_DISTRIB_DIR/v14/bin" && touch "$POSTGRES_DISTRIB_DIR/v14/bin/psql";
+        mkdir -p "$POSTGRES_DISTRIB_DIR/v${DEFAULT_PG_VERSION}/bin" && touch "$POSTGRES_DISTRIB_DIR/v${DEFAULT_PG_VERSION}/bin/psql";
         ./scripts/pytest \
           --junitxml=$TEST_OUTPUT/junit.xml \
           --tb=short \
@@ -75,7 +76,6 @@ jobs:
       if: ${{ always() }}
       uses: ./.github/actions/neon-project-delete
       with:
-        environment: staging
         project_id: ${{ steps.create-neon-project.outputs.project_id }}
         api_key: ${{ secrets.NEON_STAGING_API_KEY }}
 

.gitmodules

@@ -1,7 +1,7 @@
 [submodule "vendor/postgres-v14"]
 	path = vendor/postgres-v14
 	url = https://github.com/neondatabase/postgres.git
-	branch = main
+	branch = REL_14_STABLE_neon
 [submodule "vendor/postgres-v15"]
 	path = vendor/postgres-v15
 	url = https://github.com/neondatabase/postgres.git

CODEOWNERS

@@ -0,0 +1,11 @@
+/compute_tools/ @neondatabase/control-plane
+/control_plane/ @neondatabase/compute @neondatabase/storage
+/libs/pageserver_api/ @neondatabase/compute @neondatabase/storage
+/libs/postgres_ffi/ @neondatabase/compute 
+/libs/remote_storage/ @neondatabase/storage 
+/libs/safekeeper_api/ @neondatabase/safekeepers  
+/pageserver/ @neondatabase/compute @neondatabase/storage 
+/pgxn/ @neondatabase/compute
+/proxy/ @neondatabase/control-plane 
+/safekeeper/ @neondatabase/safekeepers
+/vendor/ @neondatabase/compute

Cargo.toml

@@ -16,6 +16,7 @@ members = [
     "pageserver",
     "proxy",
     "safekeeper",
+    "storage_broker",
     "workspace_hack",
     "libs/*",
 ]
@@ -25,6 +26,10 @@ members = [
 # Besides, debug info should not affect the performance.
 debug = true
 
+# disable debug symbols for all packages except this one to decrease binaries size
+[profile.release.package."*"]
+debug = false
+
 [profile.release-line-debug]
 inherits = "release"
 debug = 1 # true = 2 = all symbols, 1 = line only

Dockerfile

@@ -44,7 +44,7 @@ COPY . .
 # Show build caching stats to check if it was used in the end.
 # Has to be the part of the same RUN since cachepot daemon is killed in the end of this RUN, losing the compilation stats.
 RUN set -e \
-&& mold -run cargo build --bin pageserver --bin pageserver_binutils --bin draw_timeline_dir --bin safekeeper --bin proxy --locked --release \
+&& mold -run cargo build --bin pageserver --bin pageserver_binutils --bin draw_timeline_dir --bin safekeeper --bin storage_broker --bin proxy --locked --release \
     && cachepot -s
 
 # Build final image
@@ -67,6 +67,7 @@ COPY --from=build --chown=neon:neon /home/nonroot/target/release/pageserver
 COPY --from=build --chown=neon:neon /home/nonroot/target/release/pageserver_binutils /usr/local/bin
 COPY --from=build --chown=neon:neon /home/nonroot/target/release/draw_timeline_dir   /usr/local/bin
 COPY --from=build --chown=neon:neon /home/nonroot/target/release/safekeeper          /usr/local/bin
+COPY --from=build --chown=neon:neon /home/nonroot/target/release/storage_broker         /usr/local/bin
 COPY --from=build --chown=neon:neon /home/nonroot/target/release/proxy               /usr/local/bin
 
 COPY --from=pg-build /home/nonroot/pg_install/v14 /usr/local/v14/

Dockerfile.compute-node-v14

@@ -200,9 +200,6 @@ COPY --from=compute-tools --chown=postgres /home/nonroot/target/release-line-deb
 # libreadline8 for psql
 # libossp-uuid16 for extension ossp-uuid
 # libgeos, libgdal, libproj and libprotobuf-c1 for PostGIS
-#
-# Lastly, link compute_ctl into zenith_ctl while we're at it,
-# so that we don't need to put this in another layer.
 RUN apt update &&  \
     apt install --no-install-recommends -y \
         libreadline8 \
@@ -211,8 +208,7 @@ RUN apt update &&  \
         libgdal28 \
         libproj19 \
         libprotobuf-c1 && \
-    rm -rf /var/lib/apt/lists/* /tmp/* /var/tmp/* && \
-    ln /usr/local/bin/compute_ctl /usr/local/bin/zenith_ctl
+    rm -rf /var/lib/apt/lists/* /tmp/* /var/tmp/*
 
 USER postgres
 ENTRYPOINT ["/usr/local/bin/compute_ctl"]

Dockerfile.compute-node-v15

@@ -200,9 +200,6 @@ COPY --from=compute-tools --chown=postgres /home/nonroot/target/release-line-deb
 # libreadline8 for psql
 # libossp-uuid16 for extension ossp-uuid
 # libgeos, libgdal, libproj and libprotobuf-c1 for PostGIS
-#
-# Lastly, link compute_ctl into zenith_ctl while we're at it,
-# so that we don't need to put this in another layer.
 RUN apt update &&  \
     apt install --no-install-recommends -y \
         libreadline8 \
@@ -211,8 +208,7 @@ RUN apt update &&  \
         libgdal28 \
         libproj19 \
         libprotobuf-c1 && \
-    rm -rf /var/lib/apt/lists/* /tmp/* /var/tmp/* && \
-    ln /usr/local/bin/compute_ctl /usr/local/bin/zenith_ctl
+    rm -rf /var/lib/apt/lists/* /tmp/* /var/tmp/*
 
 USER postgres
 ENTRYPOINT ["/usr/local/bin/compute_ctl"]

Dockerfile.compute-node.legacy

@@ -1,88 +0,0 @@
-#
-# Legacy version of the Dockerfile for the compute node.
-# Used by e2e CI. Building Dockerfile.compute-node will take
-# unreasonable ammount of time without v2 runners.
-#
-# TODO: remove once cloud repo CI is moved to v2 runners.
-#
-
-
-# Allow specifiyng different compute-tools tag and image repo, so we are
-# able to use different images
-ARG REPOSITORY=369495373322.dkr.ecr.eu-central-1.amazonaws.com
-ARG IMAGE=compute-tools
-ARG TAG=latest
-
-#
-# Image with pre-built tools
-#
-FROM $REPOSITORY/$IMAGE:$TAG AS compute-deps
-# Only to get ready compute_ctl binary as deppendency
-
-#
-# Image with Postgres build deps
-#
-FROM debian:bullseye-slim AS build-deps
-
-RUN apt-get update && apt-get -yq install automake libtool build-essential bison flex libreadline-dev zlib1g-dev libxml2-dev \
-                                          libcurl4-openssl-dev libossp-uuid-dev
-
-#
-# Image with built Postgres
-#
-FROM build-deps AS pg-build
-
-# Add user postgres
-RUN adduser postgres
-RUN mkdir /pg && chown postgres:postgres /pg
-
-# Copy source files
-# version 14 is default for now
-COPY ./vendor/postgres-v14 /pg/
-COPY ./pgxn /pg/
-
-# Build and install Postgres locally
-RUN mkdir /pg/compute_build && cd /pg/compute_build && \
-    ../configure CFLAGS='-O2 -g3' --prefix=$(pwd)/postgres_bin --enable-debug --with-uuid=ossp && \
-    # Install main binaries and contribs
-    make MAKELEVEL=0 -j $(getconf _NPROCESSORS_ONLN) -s install && \
-    make MAKELEVEL=0 -j $(getconf _NPROCESSORS_ONLN) -s -C contrib/ install && \
-    # Install headers
-    make MAKELEVEL=0 -j $(getconf _NPROCESSORS_ONLN) -s -C src/include install
-
-# Install neon contrib
-RUN make MAKELEVEL=0 PG_CONFIG=/pg/compute_build/postgres_bin/bin/pg_config -j $(getconf _NPROCESSORS_ONLN) -C /pg/neon install
-
-USER postgres
-WORKDIR /pg
-
-#
-# Final compute node image to be exported
-#
-FROM debian:bullseye-slim
-
-# libreadline-dev is required to run psql
-RUN apt-get update && apt-get -yq install libreadline-dev libossp-uuid-dev
-
-# Add user postgres
-RUN mkdir /var/db && useradd -m -d /var/db/postgres postgres && \
-    echo "postgres:test_console_pass" | chpasswd && \
-    mkdir /var/db/postgres/compute && mkdir /var/db/postgres/specs && \
-    chown -R postgres:postgres /var/db/postgres && \
-    chmod 0750 /var/db/postgres/compute
-
-# Copy ready Postgres binaries
-COPY --from=pg-build /pg/compute_build/postgres_bin /usr/local
-
-# Copy binaries from compute-tools
-COPY --from=compute-deps /usr/local/bin/compute_ctl /usr/local/bin/compute_ctl
-
-# XXX: temporary symlink for compatibility with old control-plane
-RUN ln -s /usr/local/bin/compute_ctl /usr/local/bin/zenith_ctl
-
-# Add postgres shared objects to the search path
-RUN echo '/usr/local/lib' >> /etc/ld.so.conf && /sbin/ldconfig
-
-USER postgres
-
-ENTRYPOINT ["/usr/local/bin/compute_ctl"]

Makefile

@@ -20,18 +20,18 @@ else
 	$(error Bad build type '$(BUILD_TYPE)', see Makefile for options)
 endif
 
-# Seccomp BPF is only available for Linux
 UNAME_S := $(shell uname -s)
 ifeq ($(UNAME_S),Linux)
+	# Seccomp BPF is only available for Linux
 	PG_CONFIGURE_OPTS += --with-libseccomp
-endif
-
-# macOS with brew-installed openssl requires explicit paths
-# It can be configured with OPENSSL_PREFIX variable
-UNAME_S := $(shell uname -s)
-ifeq ($(UNAME_S),Darwin)
-    OPENSSL_PREFIX ?= $(shell brew --prefix openssl@3)
-    PG_CONFIGURE_OPTS += --with-includes=$(OPENSSL_PREFIX)/include --with-libraries=$(OPENSSL_PREFIX)/lib
+else ifeq ($(UNAME_S),Darwin)
+	# macOS with brew-installed openssl requires explicit paths
+	# It can be configured with OPENSSL_PREFIX variable
+	OPENSSL_PREFIX ?= $(shell brew --prefix openssl@3)
+	PG_CONFIGURE_OPTS += --with-includes=$(OPENSSL_PREFIX)/include --with-libraries=$(OPENSSL_PREFIX)/lib
+	# macOS already has bison and flex in the system, but they are old and result in postgres-v14 target failure
+	# brew formulae are keg-only and not symlinked into HOMEBREW_PREFIX, force their usage
+	EXTRA_PATH_OVERRIDES += $(shell brew --prefix bison)/bin/:$(shell brew --prefix flex)/bin/:
 endif
 
 # Use -C option so that when PostgreSQL "make install" installs the
@@ -73,7 +73,8 @@ $(POSTGRES_INSTALL_DIR)/build/v14/config.status:
 	+@echo "Configuring Postgres v14 build"
 	mkdir -p $(POSTGRES_INSTALL_DIR)/build/v14
 	(cd $(POSTGRES_INSTALL_DIR)/build/v14 && \
-	$(ROOT_PROJECT_DIR)/vendor/postgres-v14/configure CFLAGS='$(PG_CFLAGS)' \
+	env PATH="$(EXTRA_PATH_OVERRIDES):$$PATH" $(ROOT_PROJECT_DIR)/vendor/postgres-v14/configure \
+		CFLAGS='$(PG_CFLAGS)' \
 		$(PG_CONFIGURE_OPTS) \
 		--prefix=$(abspath $(POSTGRES_INSTALL_DIR))/v14 > configure.log)
 
@@ -81,7 +82,8 @@ $(POSTGRES_INSTALL_DIR)/build/v15/config.status:
 	+@echo "Configuring Postgres v15 build"
 	mkdir -p $(POSTGRES_INSTALL_DIR)/build/v15
 	(cd $(POSTGRES_INSTALL_DIR)/build/v15 && \
-	$(ROOT_PROJECT_DIR)/vendor/postgres-v15/configure CFLAGS='$(PG_CFLAGS)' \
+	env PATH="$(EXTRA_PATH_OVERRIDES):$$PATH" $(ROOT_PROJECT_DIR)/vendor/postgres-v15/configure \
+		CFLAGS='$(PG_CFLAGS)' \
 		$(PG_CONFIGURE_OPTS) \
 		--prefix=$(abspath $(POSTGRES_INSTALL_DIR))/v15 > configure.log)
 
@@ -111,6 +113,8 @@ postgres-v14: postgres-v14-configure \
 	$(MAKE) -C $(POSTGRES_INSTALL_DIR)/build/v14 MAKELEVEL=0 install
 	+@echo "Compiling libpq v14"
 	$(MAKE) -C $(POSTGRES_INSTALL_DIR)/build/v14/src/interfaces/libpq install
+	+@echo "Compiling pg_prewarm v14"
+	$(MAKE) -C $(POSTGRES_INSTALL_DIR)/build/v14/contrib/pg_prewarm install
 	+@echo "Compiling pg_buffercache v14"
 	$(MAKE) -C $(POSTGRES_INSTALL_DIR)/build/v14/contrib/pg_buffercache install
 	+@echo "Compiling pageinspect v14"
@@ -123,6 +127,8 @@ postgres-v15: postgres-v15-configure \
 	$(MAKE) -C $(POSTGRES_INSTALL_DIR)/build/v15 MAKELEVEL=0 install
 	+@echo "Compiling libpq v15"
 	$(MAKE) -C $(POSTGRES_INSTALL_DIR)/build/v15/src/interfaces/libpq install
+	+@echo "Compiling pg_prewarm v15"
+	$(MAKE) -C $(POSTGRES_INSTALL_DIR)/build/v15/contrib/pg_prewarm install
 	+@echo "Compiling pg_buffercache v15"
 	$(MAKE) -C $(POSTGRES_INSTALL_DIR)/build/v15/contrib/pg_buffercache install
 	+@echo "Compiling pageinspect v15"

README.md

@@ -2,29 +2,20 @@
 
 Neon is a serverless open-source alternative to AWS Aurora Postgres. It separates storage and compute and substitutes the PostgreSQL storage layer by redistributing data across a cluster of nodes.
 
-The project used to be called "Zenith". Many of the commands and code comments
-still refer to "zenith", but we are in the process of renaming things.
-
 ## Quick start
-[Join the waitlist](https://neon.tech/) for our free tier to receive your serverless postgres instance. Then connect to it with your preferred postgres client (psql, dbeaver, etc) or use the online SQL editor.
+Try the [Neon Free Tier](https://neon.tech/docs/introduction/technical-preview-free-tier/) to create a serverless Postgres instance. Then connect to it with your preferred Postgres client (psql, dbeaver, etc) or use the online [SQL Editor](https://neon.tech/docs/get-started-with-neon/query-with-neon-sql-editor/). See [Connect from any application](https://neon.tech/docs/connect/connect-from-any-app/) for connection instructions.
 
 Alternatively, compile and run the project [locally](#running-local-installation).
 
 ## Architecture overview
 
-A Neon installation consists of compute nodes and a Neon storage engine.
-
-Compute nodes are stateless PostgreSQL nodes backed by the Neon storage engine.
+A Neon installation consists of compute nodes and the Neon storage engine. Compute nodes are stateless PostgreSQL nodes backed by the Neon storage engine.
 
 The Neon storage engine consists of two major components:
 - Pageserver. Scalable storage backend for the compute nodes.
-- WAL service. The service receives WAL from the compute node and ensures that it is stored durably.
+- Safekeepers. The safekeepers form a redundant WAL service that received WAL from the compute node, and stores it durably until it has been processed by the pageserver and uploaded to cloud storage.
 
-Pageserver consists of:
-- Repository - Neon storage implementation.
-- WAL receiver - service that receives WAL from WAL service and stores it in the repository.
-- Page service - service that communicates with compute nodes and responds with pages from the repository.
-- WAL redo - service that builds pages from base images and WAL records on Page service request
+See developer documentation in [/docs/SUMMARY.md](/docs/SUMMARY.md) for more information.
 
 ## Running local installation
 
@@ -35,12 +26,12 @@ Pageserver consists of:
 * On Ubuntu or Debian, this set of packages should be sufficient to build the code:
 ```bash
 apt install build-essential libtool libreadline-dev zlib1g-dev flex bison libseccomp-dev \
-libssl-dev clang pkg-config libpq-dev etcd cmake postgresql-client
+libssl-dev clang pkg-config libpq-dev etcd cmake postgresql-client protobuf-compiler
 ```
 * On Fedora, these packages are needed:
 ```bash
 dnf install flex bison readline-devel zlib-devel openssl-devel \
-  libseccomp-devel perl clang cmake etcd postgresql postgresql-contrib
+  libseccomp-devel perl clang cmake etcd postgresql postgresql-contrib protobuf-compiler
 ```
 
 2. [Install Rust](https://www.rust-lang.org/tools/install)
@@ -53,7 +44,7 @@ curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh
 1. Install XCode and dependencies
 ```
 xcode-select --install
-brew install protobuf etcd openssl
+brew install protobuf etcd openssl flex bison
 ```
 
 2. [Install Rust](https://www.rust-lang.org/tools/install)
@@ -125,24 +116,23 @@ Python (3.9 or higher), and install python3 packages using `./scripts/pysync` (r
 # Create repository in .neon with proper paths to binaries and data
 # Later that would be responsibility of a package install script
 > ./target/debug/neon_local init
-Starting pageserver at '127.0.0.1:64000' in '.neon'
-
-Pageserver started
-Successfully initialized timeline 7dd0907914ac399ff3be45fb252bfdb7
-Stopping pageserver gracefully...done!
+Starting pageserver at '127.0.0.1:64000' in '.neon'.
+pageserver started, pid: 2545906
+Successfully initialized timeline de200bd42b49cc1814412c7e592dd6e9
+Stopped pageserver 1 process with pid 2545906
 
 # start pageserver and safekeeper
 > ./target/debug/neon_local start
-Starting etcd broker using /usr/bin/etcd
-Starting pageserver at '127.0.0.1:64000' in '.neon'
-
-Pageserver started
-Starting safekeeper at '127.0.0.1:5454' in '.neon/safekeepers/sk1'
-Safekeeper started
+Starting etcd broker using "/usr/bin/etcd"
+etcd started, pid: 2545996
+Starting pageserver at '127.0.0.1:64000' in '.neon'.
+pageserver started, pid: 2546005
+Starting safekeeper at '127.0.0.1:5454' in '.neon/safekeepers/sk1'.
+safekeeper 1 started, pid: 2546041
 
 # start postgres compute node
 > ./target/debug/neon_local pg start main
-Starting new postgres main on timeline de200bd42b49cc1814412c7e592dd6e9 ...
+Starting new postgres (v14) main on timeline de200bd42b49cc1814412c7e592dd6e9 ...
 Extracting base backup to create postgres instance: path=.neon/pgdatadirs/tenants/9ef87a5bf0d92544f6fafeeb3239695c/main port=55432
 Starting postgres node at 'host=127.0.0.1 port=55432 user=cloud_admin dbname=postgres'
 
@@ -230,12 +220,20 @@ CARGO_BUILD_FLAGS="--features=testing" make
 
 ## Documentation
 
-Now we use README files to cover design ideas and overall architecture for each module and `rustdoc` style documentation comments. See also [/docs/](/docs/) a top-level overview of all available markdown documentation.
+[/docs/](/docs/) Contains a top-level overview of all available markdown documentation.
 
 - [/docs/sourcetree.md](/docs/sourcetree.md) contains overview of source tree layout.
 
 To view your `rustdoc` documentation in a browser, try running `cargo doc --no-deps --open`
 
+See also README files in some source directories, and `rustdoc` style documentation comments.
+
+Other resources:
+
+- [SELECT 'Hello, World'](https://neon.tech/blog/hello-world/): Blog post by Nikita Shamgunov on the high level architecture
+- [Architecture decisions in Neon](https://neon.tech/blog/architecture-decisions-in-neon/): Blog post by Heikki Linnakangas
+- [Neon: Serverless PostgreSQL!](https://www.youtube.com/watch?v=rES0yzeERns): Presentation on storage system by Heikki Linnakangas in the CMU Database Group seminar series
+
 ### Postgres-specific terms
 
 Due to Neon's very close relation with PostgreSQL internals, numerous specific terms are used.

cli-v2-story.md

@@ -1,188 +0,0 @@
-Create a new Zenith repository in the current directory:
-
-    ~/git-sandbox/zenith (cli-v2)$ ./target/debug/cli init
-    The files belonging to this database system will be owned by user "heikki".
-    This user must also own the server process.
-    
-    The database cluster will be initialized with locale "en_GB.UTF-8".
-    The default database encoding has accordingly been set to "UTF8".
-    The default text search configuration will be set to "english".
-    
-    Data page checksums are disabled.
-    
-    creating directory tmp ... ok
-    creating subdirectories ... ok
-    selecting dynamic shared memory implementation ... posix
-    selecting default max_connections ... 100
-    selecting default shared_buffers ... 128MB
-    selecting default time zone ... Europe/Helsinki
-    creating configuration files ... ok
-    running bootstrap script ... ok
-    performing post-bootstrap initialization ... ok
-    syncing data to disk ... ok
-    
-    initdb: warning: enabling "trust" authentication for local connections
-    You can change this by editing pg_hba.conf or using the option -A, or
-    --auth-local and --auth-host, the next time you run initdb.
-    new zenith repository was created in .zenith
-
-Initially, there is only one branch:
-
-    ~/git-sandbox/zenith (cli-v2)$ ./target/debug/cli branch
-      main
-
-Start a local Postgres instance on the branch:
-
-    ~/git-sandbox/zenith (cli-v2)$ ./target/debug/cli start main
-    Creating data directory from snapshot at 0/15FFB08...
-    waiting for server to start....2021-04-13 09:27:43.919 EEST [984664] LOG:  starting PostgreSQL 14devel on x86_64-pc-linux-gnu, compiled by gcc (Debian 10.2.1-6) 10.2.1 20210110, 64-bit
-    2021-04-13 09:27:43.920 EEST [984664] LOG:  listening on IPv6 address "::1", port 5432
-    2021-04-13 09:27:43.920 EEST [984664] LOG:  listening on IPv4 address "127.0.0.1", port 5432
-    2021-04-13 09:27:43.927 EEST [984664] LOG:  listening on Unix socket "/tmp/.s.PGSQL.5432"
-    2021-04-13 09:27:43.939 EEST [984665] LOG:  database system was interrupted; last known up at 2021-04-13 09:27:33 EEST
-    2021-04-13 09:27:43.939 EEST [984665] LOG:  creating missing WAL directory "pg_wal/archive_status"
-    2021-04-13 09:27:44.189 EEST [984665] LOG:  database system was not properly shut down; automatic recovery in progress
-    2021-04-13 09:27:44.195 EEST [984665] LOG:  invalid record length at 0/15FFB80: wanted 24, got 0
-    2021-04-13 09:27:44.195 EEST [984665] LOG:  redo is not required
-    2021-04-13 09:27:44.225 EEST [984664] LOG:  database system is ready to accept connections
-     done
-    server started
-
-Run some commands against it:
-
-    ~/git-sandbox/zenith (cli-v2)$ psql postgres -c "create table foo (t text);" 
-    CREATE TABLE
-    ~/git-sandbox/zenith (cli-v2)$ psql postgres -c "insert into foo values ('inserted on the main branch');" 
-    INSERT 0 1
-    ~/git-sandbox/zenith (cli-v2)$ psql postgres -c "select * from foo" 
-                  t              
-    -----------------------------
-     inserted on the main branch
-    (1 row)
-
-Create a new branch called 'experimental'. We create it from the
-current end of the 'main' branch, but you could specify a different
-LSN as the start point instead.
-
-    ~/git-sandbox/zenith (cli-v2)$ ./target/debug/cli branch experimental main
-    branching at end of WAL: 0/161F478
-    
-    ~/git-sandbox/zenith (cli-v2)$ ./target/debug/cli branch 
-      experimental
-      main
-
-Start another Postgres instance off the 'experimental' branch:
-
-    ~/git-sandbox/zenith (cli-v2)$ ./target/debug/cli start experimental -- -o -p5433
-    Creating data directory from snapshot at 0/15FFB08...
-    waiting for server to start....2021-04-13 09:28:41.874 EEST [984766] LOG:  starting PostgreSQL 14devel on x86_64-pc-linux-gnu, compiled by gcc (Debian 10.2.1-6) 10.2.1 20210110, 64-bit
-    2021-04-13 09:28:41.875 EEST [984766] LOG:  listening on IPv6 address "::1", port 5433
-    2021-04-13 09:28:41.875 EEST [984766] LOG:  listening on IPv4 address "127.0.0.1", port 5433
-    2021-04-13 09:28:41.883 EEST [984766] LOG:  listening on Unix socket "/tmp/.s.PGSQL.5433"
-    2021-04-13 09:28:41.896 EEST [984767] LOG:  database system was interrupted; last known up at 2021-04-13 09:27:33 EEST
-    2021-04-13 09:28:42.265 EEST [984767] LOG:  database system was not properly shut down; automatic recovery in progress
-    2021-04-13 09:28:42.269 EEST [984767] LOG:  redo starts at 0/15FFB80
-    2021-04-13 09:28:42.272 EEST [984767] LOG:  invalid record length at 0/161F4B0: wanted 24, got 0
-    2021-04-13 09:28:42.272 EEST [984767] LOG:  redo done at 0/161F478 system usage: CPU: user: 0.00 s, system: 0.00 s, elapsed: 0.00 s
-    2021-04-13 09:28:42.321 EEST [984766] LOG:  database system is ready to accept connections
-     done
-    server started
-
-Insert some a row on the 'experimental' branch:
-
-    ~/git-sandbox/zenith (cli-v2)$ psql postgres -p5433 -c "select * from foo" 
-                  t              
-    -----------------------------
-     inserted on the main branch
-    (1 row)
-    
-    ~/git-sandbox/zenith (cli-v2)$ psql postgres -p5433 -c "insert into foo values ('inserted on experimental')" 
-    INSERT 0 1
-    ~/git-sandbox/zenith (cli-v2)$ psql postgres -p5433 -c "select * from foo" 
-                  t              
-    -----------------------------
-     inserted on the main branch
-     inserted on experimental
-    (2 rows)
-    
-See that the other Postgres instance is still running on 'main' branch on port 5432:
-
-
-    ~/git-sandbox/zenith (cli-v2)$ psql postgres -p5432 -c "select * from foo" 
-                  t              
-    -----------------------------
-     inserted on the main branch
-    (1 row)
-
-
-
-
-Everything is stored in the .zenith directory:
-
-    ~/git-sandbox/zenith (cli-v2)$ ls -l .zenith/
-    total 12
-    drwxr-xr-x 4 heikki heikki 4096 Apr 13 09:28 datadirs
-    drwxr-xr-x 4 heikki heikki 4096 Apr 13 09:27 refs
-    drwxr-xr-x 4 heikki heikki 4096 Apr 13 09:28 timelines
-
-The 'datadirs' directory contains the datadirs of the running instances:
-
-    ~/git-sandbox/zenith (cli-v2)$ ls -l .zenith/datadirs/
-    total 8
-    drwx------ 18 heikki heikki 4096 Apr 13 09:27 3c0c634c1674079b2c6d4edf7c91523e
-    drwx------ 18 heikki heikki 4096 Apr 13 09:28 697e3c103d4b1763cd6e82e4ff361d76
-    ~/git-sandbox/zenith (cli-v2)$ ls -l .zenith/datadirs/3c0c634c1674079b2c6d4edf7c91523e/
-    total 124
-    drwxr-xr-x 5 heikki heikki  4096 Apr 13 09:27 base
-    drwxr-xr-x 2 heikki heikki  4096 Apr 13 09:27 global
-    drwxr-xr-x 2 heikki heikki  4096 Apr 13 09:27 pg_commit_ts
-    drwxr-xr-x 2 heikki heikki  4096 Apr 13 09:27 pg_dynshmem
-    -rw------- 1 heikki heikki  4760 Apr 13 09:27 pg_hba.conf
-    -rw------- 1 heikki heikki  1636 Apr 13 09:27 pg_ident.conf
-    drwxr-xr-x 4 heikki heikki  4096 Apr 13 09:32 pg_logical
-    drwxr-xr-x 4 heikki heikki  4096 Apr 13 09:27 pg_multixact
-    drwxr-xr-x 2 heikki heikki  4096 Apr 13 09:27 pg_notify
-    drwxr-xr-x 2 heikki heikki  4096 Apr 13 09:27 pg_replslot
-    drwxr-xr-x 2 heikki heikki  4096 Apr 13 09:27 pg_serial
-    drwxr-xr-x 2 heikki heikki  4096 Apr 13 09:27 pg_snapshots
-    drwxr-xr-x 2 heikki heikki  4096 Apr 13 09:27 pg_stat
-    drwxr-xr-x 2 heikki heikki  4096 Apr 13 09:34 pg_stat_tmp
-    drwxr-xr-x 2 heikki heikki  4096 Apr 13 09:27 pg_subtrans
-    drwxr-xr-x 2 heikki heikki  4096 Apr 13 09:27 pg_tblspc
-    drwxr-xr-x 2 heikki heikki  4096 Apr 13 09:27 pg_twophase
-    -rw------- 1 heikki heikki     3 Apr 13 09:27 PG_VERSION
-    lrwxrwxrwx 1 heikki heikki    52 Apr 13 09:27 pg_wal -> ../../timelines/3c0c634c1674079b2c6d4edf7c91523e/wal
-    drwxr-xr-x 2 heikki heikki  4096 Apr 13 09:27 pg_xact
-    -rw------- 1 heikki heikki    88 Apr 13 09:27 postgresql.auto.conf
-    -rw------- 1 heikki heikki 28688 Apr 13 09:27 postgresql.conf
-    -rw------- 1 heikki heikki    96 Apr 13 09:27 postmaster.opts
-    -rw------- 1 heikki heikki   149 Apr 13 09:27 postmaster.pid
-
-Note how 'pg_wal' is just a symlink to the 'timelines' directory. The
-datadir is ephemeral, you can delete it at any time, and it can be reconstructed
-from the snapshots and WAL stored in the 'timelines' directory. So if you push/pull
-the repository, the 'datadirs' are not included. (They are like git working trees)
-
-    ~/git-sandbox/zenith (cli-v2)$ killall -9 postgres
-    ~/git-sandbox/zenith (cli-v2)$ rm -rf .zenith/datadirs/*
-    ~/git-sandbox/zenith (cli-v2)$ ./target/debug/cli start experimental -- -o -p5433
-    Creating data directory from snapshot at 0/15FFB08...
-    waiting for server to start....2021-04-13 09:37:05.476 EEST [985340] LOG:  starting PostgreSQL 14devel on x86_64-pc-linux-gnu, compiled by gcc (Debian 10.2.1-6) 10.2.1 20210110, 64-bit
-    2021-04-13 09:37:05.477 EEST [985340] LOG:  listening on IPv6 address "::1", port 5433
-    2021-04-13 09:37:05.477 EEST [985340] LOG:  listening on IPv4 address "127.0.0.1", port 5433
-    2021-04-13 09:37:05.487 EEST [985340] LOG:  listening on Unix socket "/tmp/.s.PGSQL.5433"
-    2021-04-13 09:37:05.498 EEST [985341] LOG:  database system was interrupted; last known up at 2021-04-13 09:27:33 EEST
-    2021-04-13 09:37:05.808 EEST [985341] LOG:  database system was not properly shut down; automatic recovery in progress
-    2021-04-13 09:37:05.813 EEST [985341] LOG:  redo starts at 0/15FFB80
-    2021-04-13 09:37:05.815 EEST [985341] LOG:  invalid record length at 0/161F770: wanted 24, got 0
-    2021-04-13 09:37:05.815 EEST [985341] LOG:  redo done at 0/161F738 system usage: CPU: user: 0.00 s, system: 0.00 s, elapsed: 0.00 s
-    2021-04-13 09:37:05.866 EEST [985340] LOG:  database system is ready to accept connections
-     done
-    server started
-    ~/git-sandbox/zenith (cli-v2)$ psql postgres -p5433 -c "select * from foo" 
-                  t              
-    -----------------------------
-     inserted on the main branch
-     inserted on experimental
-    (2 rows)
-

compute_tools/Cargo.toml

@@ -5,7 +5,7 @@ edition = "2021"
 
 [dependencies]
 anyhow = "1.0"
-chrono = "0.4"
+chrono = { version = "0.4", default-features = false, features = ["clock"] }
 clap = "4.0"
 env_logger = "0.9"
 futures = "0.3.13"

compute_tools/src/pg_helpers.rs

@@ -65,7 +65,7 @@ impl GenericOption {
             let name = match self.name.as_str() {
                 "safekeepers" => "neon.safekeepers",
                 "wal_acceptor_reconnect" => "neon.safekeeper_reconnect_timeout",
-                "wal_acceptor_connect_timeout" => "neon.safekeeper_connect_timeout",
+                "wal_acceptor_connection_timeout" => "neon.safekeeper_connection_timeout",
                 it => it,
             };
 

control_plane/Cargo.toml

@@ -23,6 +23,7 @@ url = "2.2.2"
 # Note: Do not directly depend on pageserver or safekeeper; use pageserver_api or safekeeper_api
 # instead, so that recompile times are better.
 pageserver_api = { path = "../libs/pageserver_api" }
+postgres_connection = { path = "../libs/postgres_connection" }
 safekeeper_api = { path = "../libs/safekeeper_api" }
 utils = { path = "../libs/utils" }
 workspace_hack = { version = "0.1", path = "../workspace_hack" }

control_plane/src/background_process.rs

@@ -14,20 +14,32 @@
 
 use std::ffi::OsStr;
 use std::io::Write;
-use std::path::Path;
+use std::os::unix::prelude::AsRawFd;
+use std::os::unix::process::CommandExt;
+use std::path::{Path, PathBuf};
 use std::process::{Child, Command};
 use std::time::Duration;
 use std::{fs, io, thread};
 
-use anyhow::{anyhow, bail, Context, Result};
+use anyhow::Context;
 use nix::errno::Errno;
+use nix::fcntl::{FcntlArg, FdFlag};
 use nix::sys::signal::{kill, Signal};
 use nix::unistd::Pid;
+use utils::pid_file::{self, PidFileRead};
 
-use utils::lock_file;
-
-const RETRIES: u32 = 15;
-const RETRY_TIMEOUT_MILLIS: u64 = 500;
+// These constants control the loop used to poll for process start / stop.
+//
+// The loop waits for at most 10 seconds, polling every 100 ms.
+// Once a second, it prints a dot ("."), to give the user an indication that
+// it's waiting. If the process hasn't started/stopped after 5 seconds,
+// it prints a notice that it's taking long, but keeps waiting.
+//
+const RETRY_UNTIL_SECS: u64 = 10;
+const RETRIES: u64 = (RETRY_UNTIL_SECS * 1000) / RETRY_INTERVAL_MILLIS;
+const RETRY_INTERVAL_MILLIS: u64 = 100;
+const DOT_EVERY_RETRIES: u64 = 10;
+const NOTICE_AFTER_RETRIES: u64 = 50;
 
 /// Argument to `start_process`, to indicate whether it should create pidfile or if the process creates
 /// it itself.
@@ -39,11 +51,16 @@ pub enum InitialPidFile<'t> {
 }
 
 /// Start a background child process using the parameters given.
-pub fn start_process<F, S: AsRef<OsStr>>(
+pub fn start_process<
+    F,
+    S: AsRef<OsStr>,
+    EI: IntoIterator<Item = (String, String)>, // Not generic AsRef<OsStr>, otherwise empty `envs` prevents type inference
+>(
     process_name: &str,
     datadir: &Path,
     command: &Path,
     args: &[S],
+    envs: EI,
     initial_pid_file: InitialPidFile,
     process_status_check: F,
 ) -> anyhow::Result<Child>
@@ -69,6 +86,15 @@ where
         .stderr(same_file_for_stderr)
         .args(args);
     let filled_cmd = fill_aws_secrets_vars(fill_rust_env_vars(background_command));
+    filled_cmd.envs(envs);
+
+    let pid_file_to_check = match initial_pid_file {
+        InitialPidFile::Create(path) => {
+            pre_exec_create_pidfile(filled_cmd, path);
+            path
+        }
+        InitialPidFile::Expect(path) => path,
+    };
 
     let mut spawned_process = filled_cmd.spawn().with_context(|| {
         format!("Could not spawn {process_name}, see console output and log files for details.")
@@ -79,44 +105,23 @@ where
             .with_context(|| format!("Subprocess {process_name} has invalid pid {pid}"))?,
     );
 
-    let pid_file_to_check = match initial_pid_file {
-        InitialPidFile::Create(target_pid_file_path) => {
-            match lock_file::create_lock_file(target_pid_file_path, pid.to_string()) {
-                lock_file::LockCreationResult::Created { .. } => {
-                    // We use "lock" file here only to create the pid file. The lock on the pidfile will be dropped as soon
-                    // as this CLI invocation exits, so it's a bit useless, but doesn't any harm either.
-                }
-                lock_file::LockCreationResult::AlreadyLocked { .. } => {
-                    anyhow::bail!("Cannot write pid file for {process_name} at path {target_pid_file_path:?}: file is already locked by another process")
-                }
-                lock_file::LockCreationResult::CreationFailed(e) => {
-                    return Err(e.context(format!(
-                    "Failed to create pid file for {process_name} at path {target_pid_file_path:?}"
-                )))
-                }
-            }
-            None
-        }
-        InitialPidFile::Expect(pid_file_path) => Some(pid_file_path),
-    };
-
     for retries in 0..RETRIES {
-        match process_started(pid, pid_file_to_check, &process_status_check) {
+        match process_started(pid, Some(pid_file_to_check), &process_status_check) {
             Ok(true) => {
                 println!("\n{process_name} started, pid: {pid}");
                 return Ok(spawned_process);
             }
             Ok(false) => {
-                if retries < 5 {
+                if retries == NOTICE_AFTER_RETRIES {
+                    // The process is taking a long time to start up. Keep waiting, but
+                    // print a message
+                    print!("\n{process_name} has not started yet, continuing to wait");
+                }
+                if retries % DOT_EVERY_RETRIES == 0 {
                     print!(".");
                     io::stdout().flush().unwrap();
-                } else {
-                    if retries == 5 {
-                        println!() // put a line break after dots for second message
-                    }
-                    println!("{process_name} has not started yet, retrying ({retries})...");
                 }
-                thread::sleep(Duration::from_millis(RETRY_TIMEOUT_MILLIS));
+                thread::sleep(Duration::from_millis(RETRY_INTERVAL_MILLIS));
             }
             Err(e) => {
                 println!("{process_name} failed to start: {e:#}");
@@ -127,17 +132,49 @@ where
             }
         }
     }
-    anyhow::bail!("{process_name} could not start in {RETRIES} attempts");
+    println!();
+    anyhow::bail!("{process_name} did not start in {RETRY_UNTIL_SECS} seconds");
+}
+
+/// Send SIGTERM to child process
+pub fn send_stop_child_process(child: &std::process::Child) -> anyhow::Result<()> {
+    let pid = child.id();
+    match kill(
+        nix::unistd::Pid::from_raw(pid.try_into().unwrap()),
+        Signal::SIGTERM,
+    ) {
+        Ok(()) => Ok(()),
+        Err(Errno::ESRCH) => {
+            println!("child process with pid {pid} does not exist");
+            Ok(())
+        }
+        Err(e) => anyhow::bail!("Failed to send signal to child process with pid {pid}: {e}"),
+    }
 }
 
 /// Stops the process, using the pid file given. Returns Ok also if the process is already not running.
 pub fn stop_process(immediate: bool, process_name: &str, pid_file: &Path) -> anyhow::Result<()> {
-    if !pid_file.exists() {
-        println!("{process_name} is already stopped: no pid file {pid_file:?} is present");
-        return Ok(());
-    }
-    let pid = read_pidfile(pid_file)?;
+    let pid = match pid_file::read(pid_file)
+        .with_context(|| format!("read pid_file {pid_file:?}"))?
+    {
+        PidFileRead::NotExist => {
+            println!("{process_name} is already stopped: no pid file present at {pid_file:?}");
+            return Ok(());
+        }
+        PidFileRead::NotHeldByAnyProcess(_) => {
+            // Don't try to kill according to file contents beacuse the pid might have been re-used by another process.
+            // Don't delete the file either, it can race with new pid file creation.
+            // Read `pid_file` module comment for details.
+            println!(
+                "No process is holding the pidfile. The process must have already exited. Leave in place to avoid race conditions: {pid_file:?}"
+            );
+            return Ok(());
+        }
+        PidFileRead::LockedByOtherProcess(pid) => pid,
+    };
+    // XXX the pid could become invalid (and recycled) at any time before the kill() below.
 
+    // send signal
     let sig = if immediate {
         print!("Stopping {process_name} with pid {pid} immediately..");
         Signal::SIGQUIT
@@ -149,8 +186,9 @@ pub fn stop_process(immediate: bool, process_name: &str, pid_file: &Path) -> any
     match kill(pid, sig) {
         Ok(()) => (),
         Err(Errno::ESRCH) => {
+            // Again, don't delete the pid file. The unlink can race with a new pid file being created.
             println!(
-                "{process_name} with pid {pid} does not exist, but a pid file {pid_file:?} was found"
+                "{process_name} with pid {pid} does not exist, but a pid file {pid_file:?} was found. Likely the pid got recycled. Lucky we didn't harm anyone."
             );
             return Ok(());
         }
@@ -158,21 +196,23 @@ pub fn stop_process(immediate: bool, process_name: &str, pid_file: &Path) -> any
     }
 
     // Wait until process is gone
-    for _ in 0..RETRIES {
+    for retries in 0..RETRIES {
         match process_has_stopped(pid) {
             Ok(true) => {
                 println!("\n{process_name} stopped");
-                if let Err(e) = fs::remove_file(pid_file) {
-                    if e.kind() != io::ErrorKind::NotFound {
-                        eprintln!("Failed to remove pid file {pid_file:?} after stopping the process: {e:#}");
-                    }
-                }
                 return Ok(());
             }
             Ok(false) => {
-                print!(".");
-                io::stdout().flush().unwrap();
-                thread::sleep(Duration::from_secs(1))
+                if retries == NOTICE_AFTER_RETRIES {
+                    // The process is taking a long time to start up. Keep waiting, but
+                    // print a message
+                    print!("\n{process_name} has not stopped yet, continuing to wait");
+                }
+                if retries % DOT_EVERY_RETRIES == 0 {
+                    print!(".");
+                    io::stdout().flush().unwrap();
+                }
+                thread::sleep(Duration::from_millis(RETRY_INTERVAL_MILLIS));
             }
             Err(e) => {
                 println!("{process_name} with pid {pid} failed to stop: {e:#}");
@@ -180,24 +220,28 @@ pub fn stop_process(immediate: bool, process_name: &str, pid_file: &Path) -> any
             }
         }
     }
-
-    anyhow::bail!("{process_name} with pid {pid} failed to stop in {RETRIES} attempts");
+    println!();
+    anyhow::bail!("{process_name} with pid {pid} did not stop in {RETRY_UNTIL_SECS} seconds");
 }
 
 fn fill_rust_env_vars(cmd: &mut Command) -> &mut Command {
-    let mut filled_cmd = cmd.env_clear().env("RUST_BACKTRACE", "1");
+    // If RUST_BACKTRACE is set, pass it through. But if it's not set, default
+    // to RUST_BACKTRACE=1.
+    let backtrace_setting = std::env::var_os("RUST_BACKTRACE");
+    let backtrace_setting = backtrace_setting
+        .as_deref()
+        .unwrap_or_else(|| OsStr::new("1"));
 
-    let var = "LLVM_PROFILE_FILE";
-    if let Some(val) = std::env::var_os(var) {
-        filled_cmd = filled_cmd.env(var, val);
+    let mut filled_cmd = cmd.env_clear().env("RUST_BACKTRACE", backtrace_setting);
+
+    // Pass through these environment variables to the command
+    for var in ["LLVM_PROFILE_FILE", "FAILPOINTS", "RUST_LOG"] {
+        if let Some(val) = std::env::var_os(var) {
+            filled_cmd = filled_cmd.env(var, val);
+        }
     }
 
-    const RUST_LOG_KEY: &str = "RUST_LOG";
-    if let Ok(rust_log_value) = std::env::var(RUST_LOG_KEY) {
-        filled_cmd.env(RUST_LOG_KEY, rust_log_value)
-    } else {
-        filled_cmd
-    }
+    filled_cmd
 }
 
 fn fill_aws_secrets_vars(mut cmd: &mut Command) -> &mut Command {
@@ -213,6 +257,69 @@ fn fill_aws_secrets_vars(mut cmd: &mut Command) -> &mut Command {
     cmd
 }
 
+/// Add a `pre_exec` to the cmd that, inbetween fork() and exec(),
+/// 1. Claims a pidfile with a fcntl lock on it and
+/// 2. Sets up the pidfile's file descriptor so that it (and the lock)
+///    will remain held until the cmd exits.
+fn pre_exec_create_pidfile<P>(cmd: &mut Command, path: P) -> &mut Command
+where
+    P: Into<PathBuf>,
+{
+    let path: PathBuf = path.into();
+    // SAFETY
+    // pre_exec is marked unsafe because it runs between fork and exec.
+    // Why is that dangerous in various ways?
+    // Long answer:  https://github.com/rust-lang/rust/issues/39575
+    // Short answer: in a multi-threaded program, other threads may have
+    // been inside of critical sections at the time of fork. In the
+    // original process, that was allright, assuming they protected
+    // the critical sections appropriately, e.g., through locks.
+    // Fork adds another process to the mix that
+    //   1. Has a single thread T
+    //   2. In an exact copy of the address space at the time of fork.
+    // A variety of problems scan occur now:
+    //   1. T tries to grab a lock that was locked at the time of fork.
+    //      It will wait forever since in its address space, the lock
+    //      is in state 'taken' but the thread that would unlock it is
+    //      not there.
+    //   2. A rust object that represented some external resource in the
+    //      parent now got implicitly copied by the the fork, even though
+    //      the object's type is not `Copy`. The parent program may use
+    //      non-copyability as way to enforce unique ownership of an
+    //      external resource in the typesystem. The fork breaks that
+    //      assumption, as now both parent and child process have an
+    //      owned instance of the object that represents the same
+    //      underlying resource.
+    // While these seem like niche problems, (1) in particular is
+    // highly relevant. For example, `malloc()` may grab a mutex internally,
+    // and so, if we forked while another thread was mallocing' and our
+    // pre_exec closure allocates as well, it will block on the malloc
+    // mutex forever
+    //
+    // The proper solution is to only use C library functions that are marked
+    // "async-signal-safe": https://man7.org/linux/man-pages/man7/signal-safety.7.html
+    //
+    // With this specific pre_exec() closure, the non-error path doesn't allocate.
+    // The error path uses `anyhow`, and hence does allocate.
+    // We take our chances there, hoping that any potential disaster is constrained
+    // to the child process (e.g., malloc has no state ourside of the child process).
+    // Last, `expect` prints to stderr, and stdio is not async-signal-safe.
+    // Again, we take our chances, making the same assumptions as for malloc.
+    unsafe {
+        cmd.pre_exec(move || {
+            let file = pid_file::claim_for_current_process(&path).expect("claim pid file");
+            // Remove the FD_CLOEXEC flag on the pidfile descriptor so that the pidfile
+            // remains locked after exec.
+            nix::fcntl::fcntl(file.as_raw_fd(), FcntlArg::F_SETFD(FdFlag::empty()))
+                .expect("remove FD_CLOEXEC");
+            // Don't run drop(file), it would close the file before we actually exec.
+            std::mem::forget(file);
+            Ok(())
+        });
+    }
+    cmd
+}
+
 fn process_started<F>(
     pid: Pid,
     pid_file_to_check: Option<&Path>,
@@ -223,14 +330,11 @@ where
 {
     match status_check() {
         Ok(true) => match pid_file_to_check {
-            Some(pid_file_path) => {
-                if pid_file_path.exists() {
-                    let pid_in_file = read_pidfile(pid_file_path)?;
-                    Ok(pid_in_file == pid)
-                } else {
-                    Ok(false)
-                }
-            }
+            Some(pid_file_path) => match pid_file::read(pid_file_path)? {
+                PidFileRead::NotExist => Ok(false),
+                PidFileRead::LockedByOtherProcess(pid_in_file) => Ok(pid_in_file == pid),
+                PidFileRead::NotHeldByAnyProcess(_) => Ok(false),
+            },
             None => Ok(true),
         },
         Ok(false) => Ok(false),
@@ -238,21 +342,6 @@ where
     }
 }
 
-/// Read a PID file
-///
-/// We expect a file that contains a single integer.
-fn read_pidfile(pidfile: &Path) -> Result<Pid> {
-    let pid_str = fs::read_to_string(pidfile)
-        .with_context(|| format!("failed to read pidfile {pidfile:?}"))?;
-    let pid: i32 = pid_str
-        .parse()
-        .map_err(|_| anyhow!("failed to parse pidfile {pidfile:?}"))?;
-    if pid < 1 {
-        bail!("pidfile {pidfile:?} contained bad value '{pid}'");
-    }
-    Ok(Pid::from_raw(pid))
-}
-
 fn process_has_stopped(pid: Pid) -> anyhow::Result<bool> {
     match kill(pid, None) {
         // Process exists, keep waiting

control_plane/src/bin/neon_local.rs

@@ -324,7 +324,7 @@ fn handle_init(init_match: &ArgMatches) -> anyhow::Result<LocalEnv> {
             pg_version,
         )
         .unwrap_or_else(|e| {
-            eprintln!("pageserver init failed: {e}");
+            eprintln!("pageserver init failed: {e:?}");
             exit(1);
         });
 

control_plane/src/compute.rs

@@ -322,6 +322,9 @@ impl PostgresNode {
         conf.append("shared_preload_libraries", "neon");
         conf.append_line("");
         conf.append("neon.pageserver_connstring", &pageserver_connstr);
+        if let AuthType::NeonJWT = auth_type {
+            conf.append("neon.safekeeper_token_env", "$ZENITH_AUTH_TOKEN");
+        }
         conf.append("neon.tenant_id", &self.tenant_id.to_string());
         conf.append("neon.timeline_id", &self.timeline_id.to_string());
         if let Some(lsn) = self.lsn {
@@ -343,7 +346,7 @@ impl PostgresNode {
         //   To be able to restore database in case of pageserver node crash, safekeeper should not
         //   remove WAL beyond this point. Too large lag can cause space exhaustion in safekeepers
         //   (if they are not able to upload WAL to S3).
-        conf.append("max_replication_write_lag", "500MB");
+        conf.append("max_replication_write_lag", "15MB");
         conf.append("max_replication_flush_lag", "10GB");
 
         if !self.env.safekeepers.is_empty() {

control_plane/src/connection.rs

@@ -1,57 +0,0 @@
-use url::Url;
-
-#[derive(Debug)]
-pub struct PgConnectionConfig {
-    url: Url,
-}
-
-impl PgConnectionConfig {
-    pub fn host(&self) -> &str {
-        self.url.host_str().expect("BUG: no host")
-    }
-
-    pub fn port(&self) -> u16 {
-        self.url.port().expect("BUG: no port")
-    }
-
-    /// Return a `<host>:<port>` string.
-    pub fn raw_address(&self) -> String {
-        format!("{}:{}", self.host(), self.port())
-    }
-
-    /// Connect using postgres protocol with TLS disabled.
-    pub fn connect_no_tls(&self) -> Result<postgres::Client, postgres::Error> {
-        postgres::Client::connect(self.url.as_str(), postgres::NoTls)
-    }
-}
-
-impl std::str::FromStr for PgConnectionConfig {
-    type Err = anyhow::Error;
-
-    fn from_str(s: &str) -> Result<Self, Self::Err> {
-        let mut url: Url = s.parse()?;
-
-        match url.scheme() {
-            "postgres" | "postgresql" => {}
-            other => anyhow::bail!("invalid scheme: {other}"),
-        }
-
-        // It's not a valid connection url if host is unavailable.
-        if url.host().is_none() {
-            anyhow::bail!(url::ParseError::EmptyHost);
-        }
-
-        // E.g. `postgres:bar`.
-        if url.cannot_be_a_base() {
-            anyhow::bail!("URL cannot be a base");
-        }
-
-        // Set the default PG port if it's missing.
-        if url.port().is_none() {
-            url.set_port(Some(5432))
-                .expect("BUG: couldn't set the default port");
-        }
-
-        Ok(Self { url })
-    }
-}

control_plane/src/etcd.rs

@@ -6,7 +6,7 @@ use crate::{background_process, local_env};
 
 pub fn start_etcd_process(env: &local_env::LocalEnv) -> anyhow::Result<()> {
     let etcd_broker = &env.etcd_broker;
-    println!(
+    print!(
         "Starting etcd broker using {:?}",
         etcd_broker.etcd_binary_path
     );
@@ -39,6 +39,7 @@ pub fn start_etcd_process(env: &local_env::LocalEnv) -> anyhow::Result<()> {
         &etcd_data_dir,
         &etcd_broker.etcd_binary_path,
         &args,
+        [],
         background_process::InitialPidFile::Create(&pid_file_path),
         || {
             for broker_endpoint in &etcd_broker.broker_endpoints {

control_plane/src/lib.rs

@@ -9,7 +9,6 @@
 
 mod background_process;
 pub mod compute;
-pub mod connection;
 pub mod etcd;
 pub mod local_env;
 pub mod pageserver;

control_plane/src/pageserver.rs

@@ -1,19 +1,20 @@
 use std::collections::HashMap;
-use std::fs::{self, File};
+use std::fs::File;
 use std::io::{BufReader, Write};
 use std::num::NonZeroU64;
 use std::path::{Path, PathBuf};
 use std::process::Child;
 use std::{io, result};
 
-use crate::connection::PgConnectionConfig;
-use anyhow::{bail, Context};
+use anyhow::{bail, ensure, Context};
 use pageserver_api::models::{
     TenantConfigRequest, TenantCreateRequest, TenantInfo, TimelineCreateRequest, TimelineInfo,
 };
+use postgres_connection::{parse_host_port, PgConnectionConfig};
 use reqwest::blocking::{Client, RequestBuilder, Response};
 use reqwest::{IntoUrl, Method};
 use thiserror::Error;
+use utils::auth::{Claims, Scope};
 use utils::{
     http::error::HttpErrorBody,
     id::{TenantId, TimelineId},
@@ -77,30 +78,24 @@ pub struct PageServerNode {
 
 impl PageServerNode {
     pub fn from_env(env: &LocalEnv) -> PageServerNode {
+        let (host, port) = parse_host_port(&env.pageserver.listen_pg_addr)
+            .expect("Unable to parse listen_pg_addr");
+        let port = port.unwrap_or(5432);
         let password = if env.pageserver.auth_type == AuthType::NeonJWT {
-            &env.pageserver.auth_token
+            Some(env.pageserver.auth_token.clone())
         } else {
-            ""
+            None
         };
 
         Self {
-            pg_connection_config: Self::pageserver_connection_config(
-                password,
-                &env.pageserver.listen_pg_addr,
-            ),
+            pg_connection_config: PgConnectionConfig::new_host_port(host, port)
+                .set_password(password),
             env: env.clone(),
             http_client: Client::new(),
             http_base_url: format!("http://{}/v1", env.pageserver.listen_http_addr),
         }
     }
 
-    /// Construct libpq connection string for connecting to the pageserver.
-    fn pageserver_connection_config(password: &str, listen_addr: &str) -> PgConnectionConfig {
-        format!("postgresql://no_user:{password}@{listen_addr}/no_db")
-            .parse()
-            .unwrap()
-    }
-
     pub fn initialize(
         &self,
         create_tenant: Option<TenantId>,
@@ -173,29 +168,21 @@ impl PageServerNode {
             }
             Err(e) => eprintln!("{e:#}"),
         }
-        match pageserver_process.kill() {
-            Err(e) => {
-                eprintln!(
-                    "Failed to stop pageserver {} process with pid {}: {e:#}",
-                    self.env.pageserver.id,
-                    pageserver_process.id(),
-                )
-            }
-            Ok(()) => {
-                println!(
-                    "Stopped pageserver {} process with pid {}",
-                    self.env.pageserver.id,
-                    pageserver_process.id(),
-                );
-                // cleanup after pageserver startup, since we do not call regular `stop_process` during init
-                let pid_file = self.pid_file();
-                if let Err(e) = fs::remove_file(&pid_file) {
-                    if e.kind() != io::ErrorKind::NotFound {
-                        eprintln!("Failed to remove pid file {pid_file:?} after stopping the process: {e:#}");
-                    }
-                }
-            }
-        }
+        background_process::send_stop_child_process(&pageserver_process)?;
+
+        let exit_code = pageserver_process.wait()?;
+        ensure!(
+            exit_code.success(),
+            format!(
+                "pageserver init failed with exit code {:?}",
+                exit_code.code()
+            )
+        );
+        println!(
+            "Stopped pageserver {} process with pid {}",
+            self.env.pageserver.id,
+            pageserver_process.id(),
+        );
         init_result
     }
 
@@ -237,7 +224,7 @@ impl PageServerNode {
         datadir: &Path,
         update_config: bool,
     ) -> anyhow::Result<Child> {
-        println!(
+        print!(
             "Starting pageserver at '{}' in '{}'",
             self.pg_connection_config.raw_address(),
             datadir.display()
@@ -259,11 +246,21 @@ impl PageServerNode {
             args.extend(["-c", config_override]);
         }
 
+        let envs = if self.env.pageserver.auth_type != AuthType::Trust {
+            // Generate a token to connect from the pageserver to a safekeeper
+            let token = self
+                .env
+                .generate_auth_token(&Claims::new(None, Scope::SafekeeperData))?;
+            vec![("ZENITH_AUTH_TOKEN".to_owned(), token)]
+        } else {
+            vec![]
+        };
         background_process::start_process(
             "pageserver",
             datadir,
             &self.env.pageserver_bin(),
             &args,
+            envs,
             background_process::InitialPidFile::Expect(&self.pid_file()),
             || match self.check_status() {
                 Ok(()) => Ok(true),
@@ -362,6 +359,11 @@ impl PageServerNode {
                 .map(|x| x.parse::<NonZeroU64>())
                 .transpose()
                 .context("Failed to parse 'max_lsn_wal_lag' as non zero integer")?,
+            trace_read_requests: settings
+                .remove("trace_read_requests")
+                .map(|x| x.parse::<bool>())
+                .transpose()
+                .context("Failed to parse 'trace_read_requests' as bool")?,
         };
         if !settings.is_empty() {
             bail!("Unrecognized tenant settings: {settings:?}")
@@ -424,6 +426,11 @@ impl PageServerNode {
                     .map(|x| x.parse::<NonZeroU64>())
                     .transpose()
                     .context("Failed to parse 'max_lsn_wal_lag' as non zero integer")?,
+                trace_read_requests: settings
+                    .get("trace_read_requests")
+                    .map(|x| x.parse::<bool>())
+                    .transpose()
+                    .context("Failed to parse 'trace_read_requests' as bool")?,
             })
             .send()?
             .error_from_body()?;

control_plane/src/safekeeper.rs

@@ -5,12 +5,12 @@ use std::sync::Arc;
 use std::{io, result};
 
 use anyhow::Context;
+use postgres_connection::PgConnectionConfig;
 use reqwest::blocking::{Client, RequestBuilder, Response};
 use reqwest::{IntoUrl, Method};
 use thiserror::Error;
 use utils::{http::error::HttpErrorBody, id::NodeId};
 
-use crate::connection::PgConnectionConfig;
 use crate::pageserver::PageServerNode;
 use crate::{
     background_process,
@@ -86,10 +86,7 @@ impl SafekeeperNode {
 
     /// Construct libpq connection string for connecting to this safekeeper.
     fn safekeeper_connection_config(port: u16) -> PgConnectionConfig {
-        // TODO safekeeper authentication not implemented yet
-        format!("postgresql://no_user@127.0.0.1:{port}/no_db")
-            .parse()
-            .unwrap()
+        PgConnectionConfig::new_host_port(url::Host::parse("127.0.0.1").unwrap(), port)
     }
 
     pub fn datadir_path_by_id(env: &LocalEnv, sk_id: NodeId) -> PathBuf {
@@ -169,6 +166,7 @@ impl SafekeeperNode {
             &datadir,
             &self.env.safekeeper_bin(),
             &args,
+            [],
             background_process::InitialPidFile::Expect(&self.pid_file()),
             || match self.check_status() {
                 Ok(()) => Ok(true),

docker-compose/compute_wrapper/Dockerfile

@@ -0,0 +1,13 @@
+ARG REPOSITORY=369495373322.dkr.ecr.eu-central-1.amazonaws.com
+ARG COMPUTE_IMAGE=compute-node-v14
+ARG TAG=latest
+
+FROM $REPOSITORY/${COMPUTE_IMAGE}:$TAG
+
+USER root
+RUN apt-get update &&       \
+    apt-get install -y curl \
+                       jq   \
+                       netcat
+
+USER postgres

docker-compose/docker-compose.yml

@@ -2,6 +2,7 @@ version: '3'
 
 services:
   etcd:
+    restart: always
     image: quay.io/coreos/etcd:v3.5.4
     ports:
       - 2379:2379
@@ -9,7 +10,7 @@ services:
     environment:
       # This signifficantly speeds up etcd and we anyway don't data persistency there.
       ETCD_UNSAFE_NO_FSYNC: "1"
-    command: 
+    command:
       - "etcd"
       - "--auto-compaction-mode=revision"
       - "--auto-compaction-retention=1"
@@ -24,6 +25,7 @@ services:
       - "--quota-backend-bytes=134217728" # 128 MB
 
   minio:
+    restart: always
     image: quay.io/minio/minio:RELEASE.2022-10-20T00-55-09Z
     ports:
       - 9000:9000
@@ -41,7 +43,7 @@ services:
     entrypoint:
       - "/bin/sh"
       - "-c"
-    command: 
+    command:
       - "until (/usr/bin/mc alias set minio http://minio:9000 $$MINIO_ROOT_USER $$MINIO_ROOT_PASSWORD) do
              echo 'Waiting to start minio...' && sleep 1;
          done;
@@ -51,7 +53,8 @@ services:
       - minio
 
   pageserver:
-    image: neondatabase/neon:${TAG:-latest}
+    restart: always
+    image: ${REPOSITORY:-neondatabase}/neon:${TAG:-latest}
     environment:
       - BROKER_ENDPOINT='http://etcd:2379'
       - AWS_ACCESS_KEY_ID=minio
@@ -77,7 +80,8 @@ services:
       - minio_create_buckets
 
   safekeeper1:
-    image: neondatabase/neon:${TAG:-latest}
+    restart: always
+    image: ${REPOSITORY:-neondatabase}/neon:${TAG:-latest}
     environment:
       - SAFEKEEPER_ADVERTISE_URL=safekeeper1:5454
       - SAFEKEEPER_ID=1
@@ -106,7 +110,8 @@ services:
       - minio_create_buckets
 
   safekeeper2:
-    image: neondatabase/neon:${TAG:-latest}
+    restart: always
+    image: ${REPOSITORY:-neondatabase}/neon:${TAG:-latest}
     environment:
       - SAFEKEEPER_ADVERTISE_URL=safekeeper2:5454
       - SAFEKEEPER_ID=2
@@ -135,7 +140,8 @@ services:
       - minio_create_buckets
 
   safekeeper3:
-    image: neondatabase/neon:${TAG:-latest}
+    restart: always
+    image: ${REPOSITORY:-neondatabase}/neon:${TAG:-latest}
     environment:
       - SAFEKEEPER_ADVERTISE_URL=safekeeper3:5454
       - SAFEKEEPER_ID=3
@@ -164,18 +170,21 @@ services:
       - minio_create_buckets
 
   compute:
+    restart: always
     build:
-      context: ./image/compute
+      context: ./compute_wrapper/
       args:
-        - COMPUTE_IMAGE=compute-node-v${PG_VERSION:-14}:${TAG:-latest}
+        - COMPUTE_IMAGE=compute-node-v${PG_VERSION:-14}
+        - TAG=${TAG:-latest}
         - http_proxy=$http_proxy
         - https_proxy=$https_proxy
     environment:
       - PG_VERSION=${PG_VERSION:-14}
       #- RUST_BACKTRACE=1
+    # Mount the test files directly, for faster editing cycle.
     volumes:
-      - ./compute/var/db/postgres/specs/:/var/db/postgres/specs/
-      - ./compute/shell/:/shell/
+      - ./compute_wrapper/var/db/postgres/specs/:/var/db/postgres/specs/
+      - ./compute_wrapper/shell/:/shell/
     ports:
       - 55433:55433 # pg protocol handler
       - 3080:3080 # http endpoints

docker-compose/docker_compose_test.sh

@@ -0,0 +1,60 @@
+#!/bin/bash
+
+# A basic test to ensure Docker images are built correctly.
+# Build a wrapper around the compute, start all services and runs a simple SQL query.
+# Repeats the process for all currenly supported Postgres versions.
+
+# Implicitly accepts `REPOSITORY` and `TAG` env vars that are passed into the compose file
+# Their defaults point at DockerHub `neondatabase/neon:latest` image.`,
+# to verify custom image builds (e.g pre-published ones).
+
+# XXX: Current does not work on M1 macs due to x86_64 Docker images compiled only, and no seccomp support in M1 Docker emulation layer.
+
+set -eux -o pipefail
+
+SCRIPT_DIR="$( cd -- "$( dirname -- "${BASH_SOURCE[0]}" )" &> /dev/null && pwd )"
+COMPOSE_FILE=$SCRIPT_DIR/docker-compose.yml
+
+COMPUTE_CONTAINER_NAME=docker-compose-compute-1
+SQL="CREATE TABLE t(key int primary key, value text); insert into t values(1,1); select * from t;"
+PSQL_OPTION="-h localhost -U cloud_admin -p 55433 -c '$SQL' postgres"
+
+cleanup() {
+    echo "show container information"
+    docker ps
+    docker compose -f $COMPOSE_FILE logs
+    echo "stop containers..."
+    docker compose -f $COMPOSE_FILE down
+}
+
+echo "clean up containers if exists"
+cleanup
+
+for pg_version in 14 15; do
+    echo "start containers (pg_version=$pg_version)."
+    PG_VERSION=$pg_version docker compose -f $COMPOSE_FILE up --build -d
+
+    echo "wait until the compute is ready. timeout after 60s. "
+    cnt=0
+    while sleep 1; do
+        # check timeout
+        cnt=`expr $cnt + 1`
+        if [ $cnt -gt 60 ]; then
+            echo "timeout before the compute is ready."
+            cleanup
+            exit 1
+        fi
+
+        # check if the compute is ready
+        set +o pipefail
+        result=`docker compose -f $COMPOSE_FILE logs "compute_is_ready" | grep "accepting connections" | wc -l`
+        set -o pipefail
+        if [ $result -eq 1 ]; then
+            echo "OK. The compute is ready to connect."
+            echo "execute simple queries."
+            docker exec $COMPUTE_CONTAINER_NAME /bin/bash -c "psql $PSQL_OPTION"
+            cleanup
+            break
+        fi
+    done
+done

docker-compose/image/compute/Dockerfile

@@ -1,10 +0,0 @@
-ARG COMPUTE_IMAGE=compute-node-v14:latest
-FROM neondatabase/${COMPUTE_IMAGE}
-
-USER root
-RUN apt-get update &&       \
-    apt-get install -y curl \
-                       jq   \
-                       netcat
-
-USER postgres

docs/SUMMARY.md

@@ -37,7 +37,7 @@
 
 - [Source view](./sourcetree.md)
   - [docker.md](./docker.md) — Docker images and building pipeline.
-  - [Error handling and logging]()
+  - [Error handling and logging](./error-handling.md)
   - [Testing]()
     - [Unit testing]()
     - [Integration testing]()

docs/authentication.md

@@ -1,30 +1,154 @@
 ## Authentication
 
 ### Overview
+We use JWT tokens in communication between almost all components (compute, pageserver, safekeeper, CLI) regardless of the protocol used (HTTP/PostgreSQL).
+Etcd currently has no authentication.
+Authentication is optional and is disabled by default for easier debugging.
+It is used in some tests, though.
+Note that we do not cover authentication with `pg.neon.tech` here.
 
-Current state of authentication includes usage of JWT tokens in communication between compute and pageserver and between CLI and pageserver. JWT token is signed using RSA keys. CLI generates a key pair during call to `neon_local init`. Using following openssl commands:
+For HTTP connections we use the Bearer authentication scheme.
+For PostgreSQL connections we expect the token to be passed as a password.
+There is a caveat for `psql`: it silently truncates passwords to 100 symbols, so to correctly pass JWT via `psql` you have to either use `PGPASSWORD` environment variable, or store password in `psql`'s config file.
+
+Current token scopes are described in `utils::auth::Scope`.
+There are no expiration or rotation schemes.
+
+_TODO_: some scopes allow both access to server management API and to the data.
+These probably should be split into multiple scopes.
+
+Tokens should not occur in logs.
+They may sometimes occur in configuration files, although this is discouraged
+because configs may be parsed and dumped into logs.
+
+#### Tokens generation and validation
+JWT tokens are signed using a private key.
+Compute/pageserver/safekeeper use the private key's public counterpart to validate JWT tokens.
+These components should not have access to the private key and may only get tokens from their configuration or external clients. 
+
+The key pair is generated once for an installation of compute/pageserver/safekeeper, e.g. by `neon_local init`.
+There is currently no way to rotate the key without bringing down all components.
+
+### CLI
+CLI generates a key pair during call to `neon_local init` with the following commands:
 
 ```bash
-openssl genrsa -out private_key.pem 2048
-openssl rsa -in private_key.pem -pubout -outform PEM -out public_key.pem
+openssl genrsa -out auth_private_key.pem 2048
+openssl rsa -in auth_private_key.pem -pubout -outform PEM -out auth_public_key.pem
 ```
 
-CLI also generates signed token and saves it in the config for later access to pageserver. Now authentication is optional. Pageserver has two variables in config: `auth_validation_public_key_path` and `auth_type`, so when auth type present and set to `NeonJWT` pageserver will require authentication for connections. Actual JWT is passed in password field of connection string. There is a caveat for psql, it silently truncates passwords to 100 symbols, so to correctly pass JWT via psql you have to either use PGPASSWORD environment variable, or store password in psql config file.
+Configuration files for all components point to `public_key.pem` for JWT validation.
+However, authentication is disabled by default.
+There is no way to automatically enable it everywhere, you have to configure each component individually.
 
-Currently there is no authentication between compute and safekeepers, because this communication layer is under heavy refactoring. After this refactoring support for authentication will be added there too. Now safekeeper supports "hardcoded" token passed via environment variable to be able to use callmemaybe command in pageserver.
+CLI also generates signed token (full access to Pageserver) and saves it in
+the CLI's `config` file under `pageserver.auth_token`.
+Note that pageserver's config does not have any similar parameter.
+CLI is the only component which accesses that token.
+Technically it could generate it from the private key on each run,
+but it does not do that for some reason (_TODO_).
 
-Compute uses token passed via environment variable to communicate to pageserver and in the future to the safekeeper too.
+### Compute
+#### Overview
+Compute is a per-timeline PostgreSQL instance, so it should not have
+any access to data of other tenants.
+All tokens used by a compute are restricted to a specific tenant.
+There is no auth isolation from other timelines of the same tenant,
+but a non-rogue client never accesses another timeline even by an accident:
+timeline IDs are random and hard to guess.
 
-JWT authentication now supports two scopes: tenant and pageserverapi. Tenant scope is intended for use in tenant related api calls, e.g. create_branch. Compute launched for particular tenant also uses this scope. Scope pageserver api is intended to be used by console to manage pageserver. For now we have only one management operation - create tenant.
+#### Incoming connections
+All incoming connections are from PostgreSQL clients.
+Their authentication is just plain PostgreSQL authentication and out of scope for this document.
 
-Examples for token generation in python:
+There is no administrative API except those provided by PostgreSQL.
+
+#### Outgoing connections
+Compute connects to Pageserver for getting pages.
+The connection string is configured by the `neon.pageserver_connstring` PostgreSQL GUC, e.g. `postgresql://no_user:$ZENITH_AUTH_TOKEN@localhost:15028`.
+The environment variable inside the connection string is substituted with
+the JWT token.
+
+Compute connects to Safekeepers to write and commit data.
+The token is the same for all safekeepers.
+It's stored in an environment variable, whose name is configured
+by the `neon.safekeeper_token_env` PostgreSQL GUC.
+If the GUC is unset, no token is passed.
+
+Note that both tokens can be (and typically are) the same;
+the scope is the tenant and the token is usually passed through the
+`$ZENITH_AUTH_TOKEN` environment variable.
+
+### Pageserver
+#### Overview
+Pageserver keeps track of multiple tenants, each having multiple timelines.
+For each timeline, it connects to the corresponding Safekeeper.
+Information about "corresponding Safekeeper" is published by Safekeepers
+in the Etcd, but they do not publish access tokens, otherwise what is
+the point of authentication.
+
+Pageserver keeps a connection to some set of Safekeepers, which
+may or may not correspond to active Computes.
+Hence, we cannot obtain a per-timeline access token from a Compute.
+E.g. if the timeline's Compute terminates before all WAL is
+consumed by the Pageserver, the Pageserver continues consuming WAL.
+
+Pageserver replicas' authentication is the same as the main's.
+
+#### Incoming connections
+Pageserver listens for connections from computes.
+Each compute should present a token valid for the timeline's tenant.
+
+Pageserver also has HTTP API: some parts are per-tenant,
+some parts are server-wide, these are different scopes.
+
+The `auth_type` configuration variable in Pageserver's config may have
+either of three values:
+
+* `Trust` removes all authentication. The outdated `MD5` value does likewise
+* `NeonJWT` enables JWT validation.
+   Tokens are validated using the public key which lies in a PEM file
+   specified in the `auth_validation_public_key_path` config.
+
+#### Outgoing connections
+Pageserver makes a connection to a Safekeeper for each active timeline.
+As Pageserver may want to access any timeline it has on the disk,
+it is given a blanket JWT token to access any data on any Safekeeper.
+This token is passed through an environment variable called `ZENITH_AUTH_TOKEN`
+(non-configurable as of writing this text).
+
+A better way _may be_ to store JWT token for each timeline next to it,
+but may be not.
+
+### Safekeeper
+#### Overview
+Safekeeper keeps track of multiple tenants, each having multiple timelines.
+
+#### Incoming connections
+Safekeeper accepts connections from Compute/Pageserver, each
+connection corresponds to a specific timeline and requires
+a corresponding JWT token.
+
+Safekeeper also has HTTP API: some parts are per-tenant,
+some parts are server-wide, these are different scopes.
+
+The `auth-validation-public-key-path` command line options controls
+the authentication mode:
+
+* If the option is missing, there is no authentication or JWT token validation.
+* If the option is present, it should be a path to the public key PEM file used for JWT token validation.
+
+#### Outgoing connections
+No connections are initiated by a Safekeeper.
+
+### In the source code
+Tests do not use authentication by default.
+If you need it, you can enable it by configuring the test's environment:
 
 ```python
-# generate pageserverapi token
-management_token = jwt.encode({"scope": "pageserverapi"}, auth_keys.priv, algorithm="RS256")
-
-# generate tenant token
-tenant_token = jwt.encode({"scope": "tenant", "tenant_id": ps.initial_tenant}, auth_keys.priv, algorithm="RS256")
+neon_env_builder.auth_enabled = True
 ```
 
-Utility functions to work with jwts in rust are located in libs/utils/src/auth.rs
+You will have to generate tokens if you want to access components inside the test directly,
+use `AuthKeys.generate_*_token` methods for that.
+If you create a new scope, please create a new method to prevent mistypes in scope's name.

docs/error-handling.md

@@ -0,0 +1,198 @@
+# Error handling and logging
+
+## Logging errors
+
+The principle is that errors are logged when they are handled. If you
+just propagate an error to the caller in a function, you don't need to
+log it; the caller will. But if you consume an error in a function,
+you *must* log it (if it needs to be logged at all).
+
+For example:
+
+```rust
+fn read_motd_file() -> std::io::Result<String> {
+    let mut f = File::open("/etc/motd")?;
+    let mut result = String::new();
+    f.read_to_string(&mut result)?;
+    result
+}
+```
+
+Opening or reading the file could fail, but there is no need to log
+the error here. The function merely propagates the error to the
+caller, and it is up to the caller to log the error or propagate it
+further, if the failure is not expected. But if, for example, it is
+normal that the "/etc/motd" file doesn't exist, the caller can choose
+to silently ignore the error, or log it as an INFO or DEBUG level
+message:
+
+```rust
+fn get_message_of_the_day() -> String {
+    // Get the motd from /etc/motd, or return the default proverb
+    match read_motd_file() {
+        Ok(motd) => motd,
+        Err(err)  => {
+            // It's normal that /etc/motd doesn't exist, but if we fail to
+            // read it for some other reason, that's unexpected. The message
+            // of the day isn't very important though, so we just WARN and
+            // continue with the default in any case.
+            if err.kind() != std::io::ErrorKind::NotFound {
+                 tracing::warn!("could not read \"/etc/motd\": {err:?}");
+            }
+            "An old error is always more popular than a new truth. - German proverb"
+        }
+    }
+}
+```
+
+## Error types
+
+We use the `anyhow` crate widely. It contains many convenient macros
+like `bail!` and `ensure!` to construct and return errors, and to
+propagate many kinds of low-level errors, wrapped in `anyhow::Error`.
+
+A downside of `anyhow::Error` is that the caller cannot distinguish
+between different error cases. Most errors are propagated all the way
+to the mgmt API handler function, or the main loop that handles a
+connection with the compute node, and they are all handled the same
+way: the error is logged and returned to the client as an HTTP or
+libpq error.
+
+But in some cases, we need to distinguish between errors and handle
+them differently. For example, attaching a tenant to the pageserver
+could fail either because the tenant has already been attached, or
+because we could not load its metadata from cloud storage. The first
+case is more or less expected. The console sends the Attach request to
+the pageserver, and the pageserver completes the operation, but the
+network connection might be lost before the console receives the
+response. The console will retry the operation in that case, but the
+tenant has already been attached. It is important that the pagserver
+responds with the HTTP 403 Already Exists error in that case, rather
+than a generic HTTP 500 Internal Server Error.
+
+If you need to distinguish between different kinds of errors, create a
+new `Error` type. The `thiserror` crate is useful for that. But in
+most cases `anyhow::Error` is good enough.
+
+## Panics
+
+Depending on where a panic happens, it can cause the whole pageserver
+or safekeeper to restart, or just a single tenant. In either case,
+that is pretty bad and causes an outage. Avoid panics. Never use
+`unwrap()` or other calls that might panic, to verify inputs from the
+network or from disk.
+
+It is acceptable to use functions that might panic, like `unwrap()`, if
+it is obvious that it cannot panic. For example, if you have just
+checked that a variable is not None, it is OK to call `unwrap()` on it,
+but it is still preferable to use `expect("reason")` instead to explain
+why the function cannot fail.
+
+`assert!` and `panic!` are reserved for checking clear invariants and
+very obvious "can't happen" cases. When in doubt, use anyhow `ensure!`
+or `bail!` instead.
+
+## Error levels
+
+`tracing::Level` doesn't provide very clear guidelines on what the
+different levels mean, or when to use which level. Here is how we use
+them:
+
+### Error
+
+Examples:
+- could not open file "foobar"
+- invalid tenant id
+
+Errors are not expected to happen during normal operation. Incorrect
+inputs from client can cause ERRORs. For example, if a client tries to
+call a mgmt API that doesn't exist, or if a compute node sends passes
+an LSN that has already been garbage collected away.
+
+These should *not* happen during normal operations. "Normal
+operations" is not a very precise concept. But for example, disk
+errors are not expected to happen when the system is working, so those
+count as Errors. However, if a TCP connection to a compute node is
+lost, that is not considered an Error, because it doesn't affect the
+pageserver's or safekeeper's operation in any way, and happens fairly
+frequently when compute nodes are shut down, or are killed abruptly
+because of errors in the compute.
+
+**Errors are monitored, and always need human investigation to determine
+the cause.**
+
+Whether something should be logged at ERROR, WARNING or INFO level can
+depend on the callers and clients. For example, it might be unexpected
+and a sign of a serious issue if the console calls the
+"timeline_detail" mgmt API for a timeline that doesn't exist. ERROR
+would be appropriate in that case. But if the console routinely calls
+the API after deleting a timeline, to check if the deletion has
+completed, then it would be totally normal and an INFO or DEBUG level
+message would be more appropriate. If a message is logged as an ERROR,
+but it in fact happens frequently in production and never requires any
+action, it should probably be demoted to an INFO level message.
+
+### Warn
+
+Examples:
+- could not remove temporary file "foobar.temp"
+- unrecognized file "foobar" in timeline directory
+
+Warnings are similar to Errors, in that they should not happen
+when the system is operating normally. The difference between Error and
+Warning is that an Error means that the operation failed, whereas Warning
+means that something unexpected happened, but the operation continued anyway.
+For example, if deleting a file fails because the file already didn't exist,
+it should be logged as Warning.
+
+> **Note:** The python regression tests, under `test_regress`, check the
+> pageserver log after each test for any ERROR and WARN lines. If there are
+> any ERRORs or WARNs that have not been explicitly listed in the test as
+> allowed, the test is marked a failed. This is to catch unexpected errors
+> e.g. in background operations, that don't cause immediate misbehaviour in
+> the tested functionality.
+
+### Info
+
+Info level is used to log useful information when the system is
+operating normally. Info level is appropriate e.g. for logging state
+changes, background operations, and network connections.
+
+Examples:
+- "system is shutting down"
+- "tenant was created"
+- "retrying S3 upload"
+
+### Debug & Trace
+
+Debug and Trace level messages are not printed to the log in our normal
+production configuration, but could be enabled for a specific server or
+tenant, to aid debugging. (Although we don't actually have that
+capability as of this writing).
+
+## Context
+
+We use logging "spans" to hold context information about the current
+operation. Almost every operation happens on a particular tenant and
+timeline, so we enter a span with the "tenant_id" and "timeline_id"
+very early when processing an incoming API request, for example. All
+background operations should also run in a span containing at least
+those two fields, and any other parameters or information that might
+be useful when debugging an error that might happen when performing
+the operation.
+
+TODO: Spans are not captured in the Error when it is created, but when
+the error is logged. It would be more useful to capture them at Error
+creation. We should consider using `tracing_error::SpanTrace` to do
+that.
+
+## Error message style
+
+PostgreSQL has a style guide for writing error messages:
+
+https://www.postgresql.org/docs/current/error-style-guide.html
+
+Follow that guide when writing error messages in the PostgreSQL
+extension. We don't follow it strictly in the pageserver and
+safekeeper, but the advice in the PostgreSQL style guide is generally
+good, and you can't go wrong by following it.

docs/sourcetree.md

@@ -2,6 +2,11 @@
 
 Below you will find a brief overview of each subdir in the source tree in alphabetical order.
 
+`storage_broker`:
+
+Neon storage broker, providing messaging between safekeepers and pageservers.
+[storage_broker.md](./storage_broker.md)
+
 `/control_plane`:
 
 Local control plane.
@@ -40,9 +45,9 @@ and create new databases and accounts (control plane API in our case).
 
 Integration tests, written in Python using the `pytest` framework.
 
-`/vendor/postgres-v14`:
+`/vendor/postgres-v14` and `/vendor/postgres-v15`:
 
-PostgreSQL source tree, with the modifications needed for Neon.
+PostgreSQL source tree per version, with the modifications needed for Neon.
 
 `/pgxn/neon`:
 
@@ -83,6 +88,16 @@ A subject for future modularization.
 `/libs/metrics`:
 Helpers for exposing Prometheus metrics from the server.
 
+### Adding dependencies
+When you add a Cargo dependency, you should update hakari manifest by running commands below and committing the updated `Cargo.lock` and `workspace_hack/`. There may be no changes, that's fine.
+
+```bash
+cargo hakari generate
+cargo hakari manage-deps
+```
+
+If you don't have hakari installed (`error: no such subcommand: hakari`), install it by running `cargo install cargo-hakari`.
+
 ## Using Python
 Note that Debian/Ubuntu Python packages are stale, as it commonly happens,
 so manual installation of dependencies is not recommended.

docs/storage_broker.md

@@ -0,0 +1,27 @@
+# Storage broker
+
+Storage broker targets two issues:
+- Allowing safekeepers and pageservers learn which nodes also hold their
+  timelines, and timeline statuses there.
+- Avoiding O(n^2) connections between storage nodes while doing so.
+
+This is used
+- By pageservers to determine the most advanced and alive safekeeper to pull WAL from.
+- By safekeepers to synchronize on the timeline: advance
+  `remote_consistent_lsn`, `backup_lsn`, choose who offloads WAL to s3.
+
+Technically, it is a simple stateless pub-sub message broker based on tonic
+(grpc) making multiplexing easy. Since it is stateless, fault tolerance can be
+provided by k8s; there is no built in replication support, though it is not hard
+to add.
+
+Currently, the only message is `SafekeeperTimelineInfo`. Each safekeeper, for
+each active timeline, once in a while pushes timeline status to the broker.
+Other nodes subscribe and receive this info, using it per above.
+
+Broker serves /metrics on the same port as grpc service. 
+
+grpcurl can be used to check which values are currently being pushed:
+```
+grpcurl -proto broker/proto/broker.proto -d '{"all":{}}' -plaintext localhost:50051 storage_broker.BrokerService/SubscribeSafekeeperInfo
+```

libs/pageserver_api/src/models.rs

@@ -15,19 +15,37 @@ use bytes::{BufMut, Bytes, BytesMut};
 /// A state of a tenant in pageserver's memory.
 #[derive(Debug, Clone, Copy, PartialEq, Eq, serde::Serialize, serde::Deserialize)]
 pub enum TenantState {
-    /// Tenant is fully operational, its background jobs might be running or not.
-    Active { background_jobs_running: bool },
-    /// A tenant is recognized by pageserver, but not yet ready to operate:
-    /// e.g. not present locally and being downloaded or being read into memory from the file system.
-    Paused,
-    /// A tenant is recognized by the pageserver, but no longer used for any operations, as failed to get activated.
+    // This tenant is being loaded from local disk
+    Loading,
+    // This tenant is being downloaded from cloud storage.
+    Attaching,
+    /// Tenant is fully operational
+    Active,
+    /// A tenant is recognized by pageserver, but it is being detached or the
+    /// system is being shut down.
+    Stopping,
+    /// A tenant is recognized by the pageserver, but can no longer be used for
+    /// any operations, because it failed to be activated.
     Broken,
 }
 
+impl TenantState {
+    pub fn has_in_progress_downloads(&self) -> bool {
+        match self {
+            Self::Loading => true,
+            Self::Attaching => true,
+            Self::Active => false,
+            Self::Stopping => false,
+            Self::Broken => false,
+        }
+    }
+}
+
 /// A state of a timeline in pageserver's memory.
 #[derive(Debug, Clone, Copy, PartialEq, Eq, serde::Serialize, serde::Deserialize)]
 pub enum TimelineState {
-    /// Timeline is fully operational, its background jobs are running.
+    /// Timeline is fully operational. If the containing Tenant is Active, the timeline's
+    /// background jobs are running otherwise they will be launched when the tenant is activated.
     Active,
     /// A timeline is recognized by pageserver, but not yet ready to operate.
     /// The status indicates, that the timeline could eventually go back to Active automatically:
@@ -35,8 +53,9 @@ pub enum TimelineState {
     Suspended,
     /// A timeline is recognized by pageserver, but not yet ready to operate and not allowed to
     /// automatically become Active after certain events: only a management call can change this status.
-    Paused,
-    /// A timeline is recognized by the pageserver, but no longer used for any operations, as failed to get activated.
+    Stopping,
+    /// A timeline is recognized by the pageserver, but can no longer be used for
+    /// any operations, because it failed to be activated.
     Broken,
 }
 
@@ -73,6 +92,7 @@ pub struct TenantCreateRequest {
     pub walreceiver_connect_timeout: Option<String>,
     pub lagging_wal_timeout: Option<String>,
     pub max_lsn_wal_lag: Option<NonZeroU64>,
+    pub trace_read_requests: Option<bool>,
 }
 
 #[serde_as]
@@ -112,6 +132,7 @@ pub struct TenantConfigRequest {
     pub walreceiver_connect_timeout: Option<String>,
     pub lagging_wal_timeout: Option<String>,
     pub max_lsn_wal_lag: Option<NonZeroU64>,
+    pub trace_read_requests: Option<bool>,
 }
 
 impl TenantConfigRequest {
@@ -130,6 +151,7 @@ impl TenantConfigRequest {
             walreceiver_connect_timeout: None,
             lagging_wal_timeout: None,
             max_lsn_wal_lag: None,
+            trace_read_requests: None,
         }
     }
 }
@@ -165,6 +187,8 @@ pub struct TimelineInfo {
     pub latest_gc_cutoff_lsn: Lsn,
     #[serde_as(as = "DisplayFromStr")]
     pub disk_consistent_lsn: Lsn,
+    #[serde_as(as = "DisplayFromStr")]
+    pub remote_consistent_lsn: Lsn,
     pub current_logical_size: Option<u64>, // is None when timeline is Unloaded
     pub current_physical_size: Option<u64>, // is None when timeline is Unloaded
     pub current_logical_size_non_incremental: Option<u64>,
@@ -177,10 +201,6 @@ pub struct TimelineInfo {
     pub last_received_msg_ts: Option<u128>,
     pub pg_version: u32,
 
-    #[serde_as(as = "Option<DisplayFromStr>")]
-    pub remote_consistent_lsn: Option<Lsn>,
-    pub awaits_download: bool,
-
     pub state: TimelineState,
 
     // Some of the above fields are duplicated in 'local' and 'remote', for backwards-

libs/postgres_connection/Cargo.toml

@@ -0,0 +1,17 @@
+[package]
+name = "postgres_connection"
+version = "0.1.0"
+edition = "2021"
+
+# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
+
+[dependencies]
+anyhow = "1.0"
+itertools = "0.10.3"
+postgres = { git = "https://github.com/neondatabase/rust-postgres.git", rev = "d052ee8b86fff9897c77b0fe89ea9daba0e1fa38" }
+tokio-postgres = { git = "https://github.com/neondatabase/rust-postgres.git", rev="d052ee8b86fff9897c77b0fe89ea9daba0e1fa38" }
+url = "2.2.2"
+workspace_hack = { version = "0.1", path = "../../workspace_hack" }
+
+[dev-dependencies]
+once_cell = "1.13.0"

libs/postgres_connection/src/lib.rs

@@ -0,0 +1,253 @@
+use anyhow::{bail, Context};
+use itertools::Itertools;
+use std::borrow::Cow;
+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> {
+    let (host, port) = match host_port.as_ref().rsplit_once(':') {
+        Some((host, port)) => (
+            host,
+            // +80 is a valid u16, but not a valid port
+            if port.chars().all(|c| c.is_ascii_digit()) {
+                Some(port.parse::<u16>().context("Unable to parse port")?)
+            } else {
+                bail!("Port contains a non-ascii-digit")
+            },
+        ),
+        None => (host_port.as_ref(), None), // No colons, no port specified
+    };
+    let host = Host::parse(host).context("Unable to parse host")?;
+    Ok((host, port))
+}
+
+#[cfg(test)]
+mod tests_parse_host_port {
+    use crate::parse_host_port;
+    use url::Host;
+
+    #[test]
+    fn test_normal() {
+        let (host, port) = parse_host_port("hello:123").unwrap();
+        assert_eq!(host, Host::Domain("hello".to_owned()));
+        assert_eq!(port, Some(123));
+    }
+
+    #[test]
+    fn test_no_port() {
+        let (host, port) = parse_host_port("hello").unwrap();
+        assert_eq!(host, Host::Domain("hello".to_owned()));
+        assert_eq!(port, None);
+    }
+
+    #[test]
+    fn test_ipv6() {
+        let (host, port) = parse_host_port("[::1]:123").unwrap();
+        assert_eq!(host, Host::<String>::Ipv6(std::net::Ipv6Addr::LOCALHOST));
+        assert_eq!(port, Some(123));
+    }
+
+    #[test]
+    fn test_invalid_host() {
+        assert!(parse_host_port("hello world").is_err());
+    }
+
+    #[test]
+    fn test_invalid_port() {
+        assert!(parse_host_port("hello:+80").is_err());
+    }
+}
+
+#[derive(Clone)]
+pub struct PgConnectionConfig {
+    host: Host,
+    port: u16,
+    password: Option<String>,
+    options: Vec<String>,
+}
+
+/// A simplified PostgreSQL connection configuration. Supports only a subset of possible
+/// settings for simplicity. A password getter or `to_connection_string` methods are not
+/// added by design to avoid accidentally leaking password through logging, command line
+/// arguments to a child process, or likewise.
+impl PgConnectionConfig {
+    pub fn new_host_port(host: Host, port: u16) -> Self {
+        PgConnectionConfig {
+            host,
+            port,
+            password: None,
+            options: vec![],
+        }
+    }
+
+    pub fn host(&self) -> &Host {
+        &self.host
+    }
+
+    pub fn port(&self) -> u16 {
+        self.port
+    }
+
+    pub fn set_host(mut self, h: Host) -> Self {
+        self.host = h;
+        self
+    }
+
+    pub fn set_port(mut self, p: u16) -> Self {
+        self.port = p;
+        self
+    }
+
+    pub fn set_password(mut self, s: Option<String>) -> Self {
+        self.password = s;
+        self
+    }
+
+    pub fn extend_options<I: IntoIterator<Item = S>, S: Into<String>>(mut self, i: I) -> Self {
+        self.options.extend(i.into_iter().map(|s| s.into()));
+        self
+    }
+
+    /// Return a `<host>:<port>` string.
+    pub fn raw_address(&self) -> String {
+        format!("{}:{}", self.host(), self.port())
+    }
+
+    /// Build a client library-specific connection configuration.
+    /// Used for testing and when we need to add some obscure configuration
+    /// elements at the last moment.
+    pub fn to_tokio_postgres_config(&self) -> tokio_postgres::Config {
+        // Use `tokio_postgres::Config` instead of `postgres::Config` because
+        // the former supports more options to fiddle with later.
+        let mut config = tokio_postgres::Config::new();
+        config.host(&self.host().to_string()).port(self.port);
+        if let Some(password) = &self.password {
+            config.password(password);
+        }
+        if !self.options.is_empty() {
+            // These options are command-line options and should be escaped before being passed
+            // as an 'options' connection string parameter, see
+            // https://www.postgresql.org/docs/15/libpq-connect.html#LIBPQ-CONNECT-OPTIONS
+            //
+            // They will be space-separated, so each space inside an option should be escaped,
+            // and all backslashes should be escaped before that. Although we don't expect options
+            // with spaces at the moment, they're supported by PostgreSQL. Hence we support them
+            // in this typesafe interface.
+            //
+            // We use `Cow` to avoid allocations in the best case (no escaping). A fully imperative
+            // solution would require 1-2 allocations in the worst case as well, but it's harder to
+            // implement and this function is hardly a bottleneck. The function is only called around
+            // establishing a new connection.
+            #[allow(unstable_name_collisions)]
+            config.options(
+                &self
+                    .options
+                    .iter()
+                    .map(|s| {
+                        if s.contains(['\\', ' ']) {
+                            Cow::Owned(s.replace('\\', "\\\\").replace(' ', "\\ "))
+                        } else {
+                            Cow::Borrowed(s.as_str())
+                        }
+                    })
+                    .intersperse(Cow::Borrowed(" ")) // TODO: use impl from std once it's stabilized
+                    .collect::<String>(),
+            );
+        }
+        config
+    }
+
+    /// Connect using postgres protocol with TLS disabled.
+    pub fn connect_no_tls(&self) -> Result<postgres::Client, postgres::Error> {
+        postgres::Config::from(self.to_tokio_postgres_config()).connect(postgres::NoTls)
+    }
+}
+
+impl fmt::Debug for PgConnectionConfig {
+    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+        // We want `password: Some(REDACTED-STRING)`, not `password: Some("REDACTED-STRING")`
+        // so even if the password is `REDACTED-STRING` (quite unlikely) there is no confusion.
+        // Hence `format_args!()`, it returns a "safe" string which is not escaped by `Debug`.
+        f.debug_struct("PgConnectionConfig")
+            .field("host", &self.host)
+            .field("port", &self.port)
+            .field(
+                "password",
+                &self
+                    .password
+                    .as_ref()
+                    .map(|_| format_args!("REDACTED-STRING")),
+            )
+            .finish()
+    }
+}
+
+#[cfg(test)]
+mod tests_pg_connection_config {
+    use crate::PgConnectionConfig;
+    use once_cell::sync::Lazy;
+    use url::Host;
+
+    static STUB_HOST: Lazy<Host> = Lazy::new(|| Host::Domain("stub.host.example".to_owned()));
+
+    #[test]
+    fn test_no_password() {
+        let cfg = PgConnectionConfig::new_host_port(STUB_HOST.clone(), 123);
+        assert_eq!(cfg.host(), &*STUB_HOST);
+        assert_eq!(cfg.port(), 123);
+        assert_eq!(cfg.raw_address(), "stub.host.example:123");
+        assert_eq!(
+            format!("{:?}", cfg),
+            "PgConnectionConfig { host: Domain(\"stub.host.example\"), port: 123, password: None }"
+        );
+    }
+
+    #[test]
+    fn test_ipv6() {
+        // May be a special case because hostname contains a colon.
+        let cfg = PgConnectionConfig::new_host_port(Host::parse("[::1]").unwrap(), 123);
+        assert_eq!(
+            cfg.host(),
+            &Host::<String>::Ipv6(std::net::Ipv6Addr::LOCALHOST)
+        );
+        assert_eq!(cfg.port(), 123);
+        assert_eq!(cfg.raw_address(), "[::1]:123");
+        assert_eq!(
+            format!("{:?}", cfg),
+            "PgConnectionConfig { host: Ipv6(::1), port: 123, password: None }"
+        );
+    }
+
+    #[test]
+    fn test_with_password() {
+        let cfg = PgConnectionConfig::new_host_port(STUB_HOST.clone(), 123)
+            .set_password(Some("password".to_owned()));
+        assert_eq!(cfg.host(), &*STUB_HOST);
+        assert_eq!(cfg.port(), 123);
+        assert_eq!(cfg.raw_address(), "stub.host.example:123");
+        assert_eq!(
+            format!("{:?}", cfg),
+            "PgConnectionConfig { host: Domain(\"stub.host.example\"), port: 123, password: Some(REDACTED-STRING) }"
+        );
+    }
+
+    #[test]
+    fn test_with_options() {
+        let cfg = PgConnectionConfig::new_host_port(STUB_HOST.clone(), 123).extend_options([
+            "hello",
+            "world",
+            "with space",
+            "and \\ backslashes",
+        ]);
+        assert_eq!(cfg.host(), &*STUB_HOST);
+        assert_eq!(cfg.port(), 123);
+        assert_eq!(cfg.raw_address(), "stub.host.example:123");
+        assert_eq!(
+            cfg.to_tokio_postgres_config().get_options(),
+            Some("hello world with\\ space and\\ \\\\\\ backslashes")
+        );
+    }
+}

libs/postgres_ffi/src/lib.rs

@@ -163,6 +163,27 @@ pub fn page_set_lsn(pg: &mut [u8], lsn: Lsn) {
     pg[4..8].copy_from_slice(&(lsn.0 as u32).to_le_bytes());
 }
 
+// This is port of function with the same name from freespace.c.
+// The only difference is that it does not have "level" parameter because XLogRecordPageWithFreeSpace
+// always call it with level=FSM_BOTTOM_LEVEL
+pub fn fsm_logical_to_physical(addr: BlockNumber) -> BlockNumber {
+    let mut leafno = addr;
+    const FSM_TREE_DEPTH: u32 = if pg_constants::SLOTS_PER_FSM_PAGE >= 1626 {
+        3
+    } else {
+        4
+    };
+
+    /* Count upper level nodes required to address the leaf page */
+    let mut pages: BlockNumber = 0;
+    for _l in 0..FSM_TREE_DEPTH {
+        pages += leafno + 1;
+        leafno /= pg_constants::SLOTS_PER_FSM_PAGE;
+    }
+    /* Turn the page count into 0-based block number */
+    pages - 1
+}
+
 pub mod waldecoder {
 
     use crate::{v14, v15};

libs/postgres_ffi/src/pg_constants.rs

@@ -197,6 +197,16 @@ pub const XLOG_CHECKPOINT_SHUTDOWN: u8 = 0x00;
 pub const XLOG_CHECKPOINT_ONLINE: u8 = 0x10;
 pub const XLP_LONG_HEADER: u16 = 0x0002;
 
+/* From fsm_internals.h */
+const FSM_NODES_PER_PAGE: usize = BLCKSZ as usize - SIZEOF_PAGE_HEADER_DATA - 4;
+const FSM_NON_LEAF_NODES_PER_PAGE: usize = BLCKSZ as usize / 2 - 1;
+const FSM_LEAF_NODES_PER_PAGE: usize = FSM_NODES_PER_PAGE - FSM_NON_LEAF_NODES_PER_PAGE;
+pub const SLOTS_PER_FSM_PAGE: u32 = FSM_LEAF_NODES_PER_PAGE as u32;
+
+/* From visibilitymap.c */
+pub const VM_HEAPBLOCKS_PER_PAGE: u32 =
+    (BLCKSZ as usize - SIZEOF_PAGE_HEADER_DATA) as u32 * (8 / 2); // MAPSIZE * (BITS_PER_BYTE / BITS_PER_HEAPBLOCK)
+
 // List of subdirectories inside pgdata.
 // Copied from src/bin/initdb/initdb.c
 pub const PGDATA_SUBDIRS: [&str; 22] = [

libs/postgres_ffi/wal_craft/src/lib.rs

@@ -1,7 +1,6 @@
 use anyhow::*;
 use core::time::Duration;
 use log::*;
-use once_cell::sync::Lazy;
 use postgres::types::PgLsn;
 use postgres::Client;
 use postgres_ffi::{WAL_SEGMENT_SIZE, XLOG_BLCKSZ};
@@ -26,15 +25,13 @@ pub struct PostgresServer {
     client_config: postgres::Config,
 }
 
-pub static REQUIRED_POSTGRES_CONFIG: Lazy<Vec<&'static str>> = Lazy::new(|| {
-    vec![
-        "wal_keep_size=50MB",            // Ensure old WAL is not removed
-        "shared_preload_libraries=neon", // can only be loaded at startup
-        // Disable background processes as much as possible
-        "wal_writer_delay=10s",
-        "autovacuum=off",
-    ]
-});
+pub static REQUIRED_POSTGRES_CONFIG: [&str; 4] = [
+    "wal_keep_size=50MB",            // Ensure old WAL is not removed
+    "shared_preload_libraries=neon", // can only be loaded at startup
+    // Disable background processes as much as possible
+    "wal_writer_delay=10s",
+    "autovacuum=off",
+];
 
 impl Conf {
     pub fn pg_distrib_dir(&self) -> anyhow::Result<PathBuf> {

libs/remote_storage/Cargo.toml

@@ -9,8 +9,11 @@ async-trait = "0.1"
 metrics = { version = "0.1", path = "../metrics" }
 utils = { version = "0.1", path = "../utils" }
 once_cell = "1.13.0"
-rusoto_core = "0.48"
-rusoto_s3 = "0.48"
+aws-smithy-http = "0.51.0"
+aws-types = "0.51.0"
+aws-config = { version = "0.51.0", default-features = false, features=["rustls"] }
+aws-sdk-s3 = "0.21.0"
+hyper = { version = "0.14", features = ["stream"] }
 serde = { version = "1.0", features = ["derive"] }
 serde_json = "1"
 tokio = { version = "1.17", features = ["sync", "macros", "fs", "io-util"] }

libs/remote_storage/src/lib.rs

@@ -10,7 +10,7 @@ mod s3_bucket;
 
 use std::{
     collections::HashMap,
-    fmt::{Debug, Display},
+    fmt::Debug,
     num::{NonZeroU32, NonZeroUsize},
     ops::Deref,
     path::{Path, PathBuf},
@@ -41,44 +41,27 @@ pub const DEFAULT_REMOTE_STORAGE_S3_CONCURRENCY_LIMIT: usize = 100;
 
 const REMOTE_STORAGE_PREFIX_SEPARATOR: char = '/';
 
-#[derive(Clone, PartialEq, Eq)]
-pub struct RemoteObjectId(String);
+/// Path on the remote storage, relative to some inner prefix.
+/// The prefix is an implementation detail, that allows representing local paths
+/// as the remote ones, stripping the local storage prefix away.
+#[derive(Debug, Clone, PartialEq, Eq, PartialOrd, Ord, Hash)]
+pub struct RemotePath(PathBuf);
+
+impl RemotePath {
+    pub fn new(relative_path: &Path) -> anyhow::Result<Self> {
+        anyhow::ensure!(
+            relative_path.is_relative(),
+            "Path {relative_path:?} is not relative"
+        );
+        Ok(Self(relative_path.to_path_buf()))
+    }
+
+    pub fn with_base(&self, base_path: &Path) -> PathBuf {
+        base_path.join(&self.0)
+    }
 
-///
-/// A key that refers to an object in remote storage. It works much like a Path,
-/// but it's a separate datatype so that you don't accidentally mix local paths
-/// and remote keys.
-///
-impl RemoteObjectId {
-    // Needed to retrieve last component for RemoteObjectId.
-    // In other words a file name
-    /// Turn a/b/c or a/b/c/ into c
     pub fn object_name(&self) -> Option<&str> {
-        // corner case, char::to_string is not const, thats why this is more verbose than it needs to be
-        // see https://github.com/rust-lang/rust/issues/88674
-        if self.0.len() == 1 && self.0.chars().next().unwrap() == REMOTE_STORAGE_PREFIX_SEPARATOR {
-            return None;
-        }
-
-        if self.0.ends_with(REMOTE_STORAGE_PREFIX_SEPARATOR) {
-            self.0.rsplit(REMOTE_STORAGE_PREFIX_SEPARATOR).nth(1)
-        } else {
-            self.0
-                .rsplit_once(REMOTE_STORAGE_PREFIX_SEPARATOR)
-                .map(|(_, last)| last)
-        }
-    }
-}
-
-impl Debug for RemoteObjectId {
-    fn fmt(&self, fmt: &mut std::fmt::Formatter<'_>) -> Result<(), std::fmt::Error> {
-        Debug::fmt(&self.0, fmt)
-    }
-}
-
-impl Display for RemoteObjectId {
-    fn fmt(&self, fmt: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
-        Display::fmt(&self.0, fmt)
+        self.0.file_name().and_then(|os_str| os_str.to_str())
     }
 }
 
@@ -87,49 +70,40 @@ impl Display for RemoteObjectId {
 /// providing basic CRUD operations for storage files.
 #[async_trait::async_trait]
 pub trait RemoteStorage: Send + Sync + 'static {
-    /// Attempts to derive the storage path out of the local path, if the latter is correct.
-    fn remote_object_id(&self, local_path: &Path) -> anyhow::Result<RemoteObjectId>;
-
-    /// Gets the download path of the given storage file.
-    fn local_path(&self, remote_object_id: &RemoteObjectId) -> anyhow::Result<PathBuf>;
-
     /// Lists all items the storage has right now.
-    async fn list(&self) -> anyhow::Result<Vec<RemoteObjectId>>;
+    async fn list(&self) -> anyhow::Result<Vec<RemotePath>>;
 
     /// Lists all top level subdirectories for a given prefix
     /// Note: here we assume that if the prefix is passed it was obtained via remote_object_id
     /// which already takes into account any kind of global prefix (prefix_in_bucket for S3 or storage_root for LocalFS)
     /// so this method doesnt need to.
-    async fn list_prefixes(
-        &self,
-        prefix: Option<&RemoteObjectId>,
-    ) -> anyhow::Result<Vec<RemoteObjectId>>;
+    async fn list_prefixes(&self, prefix: Option<&RemotePath>) -> anyhow::Result<Vec<RemotePath>>;
 
     /// Streams the local file contents into remote into the remote storage entry.
     async fn upload(
         &self,
-        from: Box<(dyn io::AsyncRead + Unpin + Send + Sync + 'static)>,
+        data: Box<(dyn io::AsyncRead + Unpin + Send + Sync + 'static)>,
         // S3 PUT request requires the content length to be specified,
         // otherwise it starts to fail with the concurrent connection count increasing.
-        from_size_bytes: usize,
-        to: &RemoteObjectId,
+        data_size_bytes: usize,
+        to: &RemotePath,
         metadata: Option<StorageMetadata>,
     ) -> anyhow::Result<()>;
 
     /// Streams the remote storage entry contents into the buffered writer given, returns the filled writer.
     /// Returns the metadata, if any was stored with the file previously.
-    async fn download(&self, from: &RemoteObjectId) -> Result<Download, DownloadError>;
+    async fn download(&self, from: &RemotePath) -> Result<Download, DownloadError>;
 
     /// Streams a given byte range of the remote storage entry contents into the buffered writer given, returns the filled writer.
     /// Returns the metadata, if any was stored with the file previously.
     async fn download_byte_range(
         &self,
-        from: &RemoteObjectId,
+        from: &RemotePath,
         start_inclusive: u64,
         end_exclusive: Option<u64>,
     ) -> Result<Download, DownloadError>;
 
-    async fn delete(&self, path: &RemoteObjectId) -> anyhow::Result<()>;
+    async fn delete(&self, path: &RemotePath) -> anyhow::Result<()>;
 
     /// Downcast to LocalFs implementation. For tests.
     fn as_local(&self) -> Option<&LocalFs> {
@@ -178,34 +152,35 @@ impl std::error::Error for DownloadError {}
 /// Every storage, currently supported.
 /// Serves as a simple way to pass around the [`RemoteStorage`] without dealing with generics.
 #[derive(Clone)]
-pub struct GenericRemoteStorage(Arc<dyn RemoteStorage>);
+pub enum GenericRemoteStorage {
+    LocalFs(LocalFs),
+    AwsS3(Arc<S3Bucket>),
+}
 
 impl Deref for GenericRemoteStorage {
     type Target = dyn RemoteStorage;
 
     fn deref(&self) -> &Self::Target {
-        self.0.as_ref()
+        match self {
+            GenericRemoteStorage::LocalFs(local_fs) => local_fs,
+            GenericRemoteStorage::AwsS3(s3_bucket) => s3_bucket.as_ref(),
+        }
     }
 }
 
 impl GenericRemoteStorage {
-    pub fn new(storage: impl RemoteStorage) -> Self {
-        Self(Arc::new(storage))
-    }
-
     pub fn from_config(
-        working_directory: PathBuf,
         storage_config: &RemoteStorageConfig,
     ) -> anyhow::Result<GenericRemoteStorage> {
         Ok(match &storage_config.storage {
             RemoteStorageKind::LocalFs(root) => {
                 info!("Using fs root '{}' as a remote storage", root.display());
-                GenericRemoteStorage::new(LocalFs::new(root.clone(), working_directory)?)
+                GenericRemoteStorage::LocalFs(LocalFs::new(root.clone())?)
             }
             RemoteStorageKind::AwsS3(s3_config) => {
                 info!("Using s3 bucket '{}' in region '{}' as a remote storage, prefix in bucket: '{:?}', bucket endpoint: '{:?}'",
                       s3_config.bucket_name, s3_config.bucket_region, s3_config.prefix_in_bucket, s3_config.endpoint);
-                GenericRemoteStorage::new(S3Bucket::new(s3_config, working_directory)?)
+                GenericRemoteStorage::AwsS3(Arc::new(S3Bucket::new(s3_config)?))
             }
         })
     }
@@ -219,23 +194,12 @@ impl GenericRemoteStorage {
         &self,
         from: Box<dyn tokio::io::AsyncRead + Unpin + Send + Sync + 'static>,
         from_size_bytes: usize,
-        from_path: &Path,
+        to: &RemotePath,
     ) -> anyhow::Result<()> {
-        let target_storage_path = self.remote_object_id(from_path).with_context(|| {
-            format!(
-                "Failed to get the storage path for source local path '{}'",
-                from_path.display()
-            )
-        })?;
-
-        self.upload(from, from_size_bytes, &target_storage_path, None)
+        self.upload(from, from_size_bytes, to, None)
             .await
             .with_context(|| {
-                format!(
-                    "Failed to upload from '{}' to storage path '{:?}'",
-                    from_path.display(),
-                    target_storage_path
-                )
+                format!("Failed to upload data of length {from_size_bytes} to storage path {to:?}")
             })
     }
 
@@ -244,24 +208,11 @@ impl GenericRemoteStorage {
     pub async fn download_storage_object(
         &self,
         byte_range: Option<(u64, Option<u64>)>,
-        to_path: &Path,
+        from: &RemotePath,
     ) -> Result<Download, DownloadError> {
-        let remote_object_path = self
-            .remote_object_id(to_path)
-            .with_context(|| {
-                format!(
-                    "Failed to get the storage path for target local path '{}'",
-                    to_path.display()
-                )
-            })
-            .map_err(DownloadError::BadInput)?;
-
         match byte_range {
-            Some((start, end)) => {
-                self.download_byte_range(&remote_object_path, start, end)
-                    .await
-            }
-            None => self.download(&remote_object_path).await,
+            Some((start, end)) => self.download_byte_range(from, start, end).await,
+            None => self.download(from).await,
         }
     }
 }
@@ -271,23 +222,6 @@ impl GenericRemoteStorage {
 #[derive(Debug, Clone, PartialEq, Eq)]
 pub struct StorageMetadata(HashMap<String, String>);
 
-fn strip_path_prefix<'a>(prefix: &'a Path, path: &'a Path) -> anyhow::Result<&'a Path> {
-    if prefix == path {
-        anyhow::bail!(
-            "Prefix and the path are equal, cannot strip: '{}'",
-            prefix.display()
-        )
-    } else {
-        path.strip_prefix(prefix).with_context(|| {
-            format!(
-                "Path '{}' is not prefixed with '{}'",
-                path.display(),
-                prefix.display(),
-            )
-        })
-    }
-}
-
 /// External backup storage configuration, enough for creating a client for that storage.
 #[derive(Debug, Clone, PartialEq, Eq)]
 pub struct RemoteStorageConfig {
@@ -431,21 +365,24 @@ mod tests {
     use super::*;
 
     #[test]
-    fn object_name() {
-        let k = RemoteObjectId("a/b/c".to_owned());
+    fn test_object_name() {
+        let k = RemotePath::new(Path::new("a/b/c")).unwrap();
         assert_eq!(k.object_name(), Some("c"));
 
-        let k = RemoteObjectId("a/b/c/".to_owned());
+        let k = RemotePath::new(Path::new("a/b/c/")).unwrap();
         assert_eq!(k.object_name(), Some("c"));
 
-        let k = RemoteObjectId("a/".to_owned());
+        let k = RemotePath::new(Path::new("a/")).unwrap();
         assert_eq!(k.object_name(), Some("a"));
 
         // XXX is it impossible to have an empty key?
-        let k = RemoteObjectId("".to_owned());
-        assert_eq!(k.object_name(), None);
-
-        let k = RemoteObjectId("/".to_owned());
+        let k = RemotePath::new(Path::new("")).unwrap();
         assert_eq!(k.object_name(), None);
     }
+
+    #[test]
+    fn rempte_path_cannot_be_created_from_absolute_ones() {
+        let err = RemotePath::new(Path::new("/")).expect_err("Should fail on absolute paths");
+        assert_eq!(err.to_string(), "Path \"/\" is not relative");
+    }
 }

libs/remote_storage/src/local_fs.rs

@@ -5,6 +5,7 @@
 //! volume is mounted to the local FS.
 
 use std::{
+    borrow::Cow,
     future::Future,
     path::{Path, PathBuf},
     pin::Pin,
@@ -18,60 +19,33 @@ use tokio::{
 use tracing::*;
 use utils::crashsafe::path_with_suffix_extension;
 
-use crate::{Download, DownloadError, RemoteObjectId};
+use crate::{Download, DownloadError, RemotePath};
 
-use super::{strip_path_prefix, RemoteStorage, StorageMetadata};
+use super::{RemoteStorage, StorageMetadata};
 
 const LOCAL_FS_TEMP_FILE_SUFFIX: &str = "___temp";
 
-/// Convert a Path in the remote storage into a RemoteObjectId
-fn remote_object_id_from_path(path: &Path) -> anyhow::Result<RemoteObjectId> {
-    Ok(RemoteObjectId(
-        path.to_str()
-            .ok_or_else(|| anyhow::anyhow!("unexpected characters found in path"))?
-            .to_string(),
-    ))
-}
-
+#[derive(Debug, Clone)]
 pub struct LocalFs {
-    working_directory: PathBuf,
     storage_root: PathBuf,
 }
 
 impl LocalFs {
     /// Attempts to create local FS storage, along with its root directory.
-    pub fn new(root: PathBuf, working_directory: PathBuf) -> anyhow::Result<Self> {
-        if !root.exists() {
-            std::fs::create_dir_all(&root).with_context(|| {
-                format!(
-                    "Failed to create all directories in the given root path '{}'",
-                    root.display(),
-                )
+    /// Storage root will be created (if does not exist) and transformed into an absolute path (if passed as relative).
+    pub fn new(mut storage_root: PathBuf) -> anyhow::Result<Self> {
+        if !storage_root.exists() {
+            std::fs::create_dir_all(&storage_root).with_context(|| {
+                format!("Failed to create all directories in the given root path {storage_root:?}")
             })?;
         }
-        Ok(Self {
-            working_directory,
-            storage_root: root,
-        })
-    }
-
-    ///
-    /// Get the absolute path in the local filesystem to given remote object.
-    ///
-    /// This is public so that it can be used in tests. Should not be used elsewhere.
-    ///
-    pub fn resolve_in_storage(&self, remote_object_id: &RemoteObjectId) -> anyhow::Result<PathBuf> {
-        let path = PathBuf::from(&remote_object_id.0);
-        if path.is_relative() {
-            Ok(self.storage_root.join(path))
-        } else if path.starts_with(&self.storage_root) {
-            Ok(path)
-        } else {
-            bail!(
-                "Path '{}' does not belong to the current storage",
-                path.display()
-            )
+        if !storage_root.is_absolute() {
+            storage_root = storage_root.canonicalize().with_context(|| {
+                format!("Failed to represent path {storage_root:?} as an absolute path")
+            })?;
         }
+
+        Ok(Self { storage_root })
     }
 
     async fn read_storage_metadata(
@@ -103,45 +77,48 @@ impl LocalFs {
 
 #[async_trait::async_trait]
 impl RemoteStorage for LocalFs {
-    /// Convert a "local" path into a "remote path"
-    fn remote_object_id(&self, local_path: &Path) -> anyhow::Result<RemoteObjectId> {
-        let path = self.storage_root.join(
-            strip_path_prefix(&self.working_directory, local_path)
-                .context("local path does not belong to this storage")?,
-        );
-        remote_object_id_from_path(&path)
+    async fn list(&self) -> anyhow::Result<Vec<RemotePath>> {
+        Ok(get_all_files(&self.storage_root, true)
+            .await?
+            .into_iter()
+            .map(|path| {
+                path.strip_prefix(&self.storage_root)
+                    .context("Failed to strip storage root prefix")
+                    .and_then(RemotePath::new)
+                    .expect(
+                        "We list files for storage root, hence should be able to remote the prefix",
+                    )
+            })
+            .collect())
     }
 
-    fn local_path(&self, remote_object_id: &RemoteObjectId) -> anyhow::Result<PathBuf> {
-        let storage_path = PathBuf::from(&remote_object_id.0);
-        let relative_path = strip_path_prefix(&self.storage_root, &storage_path)
-            .context("local path does not belong to this storage")?;
-        Ok(self.working_directory.join(relative_path))
-    }
-
-    async fn list(&self) -> anyhow::Result<Vec<RemoteObjectId>> {
-        get_all_files(&self.storage_root, true).await
-    }
-
-    async fn list_prefixes(
-        &self,
-        prefix: Option<&RemoteObjectId>,
-    ) -> anyhow::Result<Vec<RemoteObjectId>> {
+    async fn list_prefixes(&self, prefix: Option<&RemotePath>) -> anyhow::Result<Vec<RemotePath>> {
         let path = match prefix {
-            Some(prefix) => Path::new(&prefix.0),
-            None => &self.storage_root,
+            Some(prefix) => Cow::Owned(prefix.with_base(&self.storage_root)),
+            None => Cow::Borrowed(&self.storage_root),
         };
-        get_all_files(path, false).await
+        Ok(get_all_files(path.as_ref(), false)
+            .await?
+            .into_iter()
+            .map(|path| {
+                path.strip_prefix(&self.storage_root)
+                    .context("Failed to strip preifix")
+                    .and_then(RemotePath::new)
+                    .expect(
+                        "We list files for storage root, hence should be able to remote the prefix",
+                    )
+            })
+            .collect())
     }
 
     async fn upload(
         &self,
-        from: Box<(dyn io::AsyncRead + Unpin + Send + Sync + 'static)>,
-        from_size_bytes: usize,
-        to: &RemoteObjectId,
+        data: Box<(dyn io::AsyncRead + Unpin + Send + Sync + 'static)>,
+        data_size_bytes: usize,
+        to: &RemotePath,
         metadata: Option<StorageMetadata>,
     ) -> anyhow::Result<()> {
-        let target_file_path = self.resolve_in_storage(to)?;
+        let target_file_path = to.with_base(&self.storage_root);
         create_target_directory(&target_file_path).await?;
         // We need this dance with sort of durable rename (without fsyncs)
         // to prevent partial uploads. This was really hit when pageserver shutdown
@@ -162,8 +139,8 @@ impl RemoteStorage for LocalFs {
                 })?,
         );
 
-        let from_size_bytes = from_size_bytes as u64;
-        let mut buffer_to_read = from.take(from_size_bytes);
+        let from_size_bytes = data_size_bytes as u64;
+        let mut buffer_to_read = data.take(from_size_bytes);
 
         let bytes_read = io::copy(&mut buffer_to_read, &mut destination)
             .await
@@ -220,27 +197,22 @@ impl RemoteStorage for LocalFs {
         Ok(())
     }
 
-    async fn download(&self, from: &RemoteObjectId) -> Result<Download, DownloadError> {
-        let file_path = self
-            .resolve_in_storage(from)
-            .map_err(DownloadError::BadInput)?;
-        if file_exists(&file_path).map_err(DownloadError::BadInput)? {
+    async fn download(&self, from: &RemotePath) -> Result<Download, DownloadError> {
+        let target_path = from.with_base(&self.storage_root);
+        if file_exists(&target_path).map_err(DownloadError::BadInput)? {
             let source = io::BufReader::new(
                 fs::OpenOptions::new()
                     .read(true)
-                    .open(&file_path)
+                    .open(&target_path)
                     .await
                     .with_context(|| {
-                        format!(
-                            "Failed to open source file '{}' to use in the download",
-                            file_path.display()
-                        )
+                        format!("Failed to open source file {target_path:?} to use in the download")
                     })
                     .map_err(DownloadError::Other)?,
             );
 
             let metadata = self
-                .read_storage_metadata(&file_path)
+                .read_storage_metadata(&target_path)
                 .await
                 .map_err(DownloadError::Other)?;
             Ok(Download {
@@ -254,7 +226,7 @@ impl RemoteStorage for LocalFs {
 
     async fn download_byte_range(
         &self,
-        from: &RemoteObjectId,
+        from: &RemotePath,
         start_inclusive: u64,
         end_exclusive: Option<u64>,
     ) -> Result<Download, DownloadError> {
@@ -266,20 +238,15 @@ impl RemoteStorage for LocalFs {
                 return Err(DownloadError::Other(anyhow::anyhow!("Invalid range, start ({start_inclusive}) and end_exclusive ({end_exclusive:?}) difference is zero bytes")));
             }
         }
-        let file_path = self
-            .resolve_in_storage(from)
-            .map_err(DownloadError::BadInput)?;
-        if file_exists(&file_path).map_err(DownloadError::BadInput)? {
+        let target_path = from.with_base(&self.storage_root);
+        if file_exists(&target_path).map_err(DownloadError::BadInput)? {
             let mut source = io::BufReader::new(
                 fs::OpenOptions::new()
                     .read(true)
-                    .open(&file_path)
+                    .open(&target_path)
                     .await
                     .with_context(|| {
-                        format!(
-                            "Failed to open source file '{}' to use in the download",
-                            file_path.display()
-                        )
+                        format!("Failed to open source file {target_path:?} to use in the download")
                     })
                     .map_err(DownloadError::Other)?,
             );
@@ -289,7 +256,7 @@ impl RemoteStorage for LocalFs {
                 .context("Failed to seek to the range start in a local storage file")
                 .map_err(DownloadError::Other)?;
             let metadata = self
-                .read_storage_metadata(&file_path)
+                .read_storage_metadata(&target_path)
                 .await
                 .map_err(DownloadError::Other)?;
 
@@ -308,15 +275,12 @@ impl RemoteStorage for LocalFs {
         }
     }
 
-    async fn delete(&self, path: &RemoteObjectId) -> anyhow::Result<()> {
-        let file_path = self.resolve_in_storage(path)?;
+    async fn delete(&self, path: &RemotePath) -> anyhow::Result<()> {
+        let file_path = path.with_base(&self.storage_root);
         if file_path.exists() && file_path.is_file() {
             Ok(fs::remove_file(file_path).await?)
         } else {
-            bail!(
-                "File '{}' either does not exist or is not a file",
-                file_path.display()
-            )
+            bail!("File {file_path:?} either does not exist or is not a file")
         }
     }
 
@@ -332,7 +296,7 @@ fn storage_metadata_path(original_path: &Path) -> PathBuf {
 fn get_all_files<'a, P>(
     directory_path: P,
     recursive: bool,
-) -> Pin<Box<dyn Future<Output = anyhow::Result<Vec<RemoteObjectId>>> + Send + Sync + 'a>>
+) -> Pin<Box<dyn Future<Output = anyhow::Result<Vec<PathBuf>>> + Send + Sync + 'a>>
 where
     P: AsRef<Path> + Send + Sync + 'a,
 {
@@ -346,20 +310,20 @@ where
                     let file_type = dir_entry.file_type().await?;
                     let entry_path = dir_entry.path();
                     if file_type.is_symlink() {
-                        debug!("{:?} us a symlink, skipping", entry_path)
+                        debug!("{entry_path:?} us a symlink, skipping")
                     } else if file_type.is_dir() {
                         if recursive {
                             paths.extend(get_all_files(&entry_path, true).await?.into_iter())
                         } else {
-                            paths.push(remote_object_id_from_path(&dir_entry.path())?)
+                            paths.push(entry_path)
                         }
                     } else {
-                        paths.push(remote_object_id_from_path(&dir_entry.path())?);
+                        paths.push(entry_path);
                     }
                 }
                 Ok(paths)
             } else {
-                bail!("Path '{}' is not a directory", directory_path.display())
+                bail!("Path {directory_path:?} is not a directory")
             }
         } else {
             Ok(Vec::new())
@@ -394,173 +358,6 @@ fn file_exists(file_path: &Path) -> anyhow::Result<bool> {
     }
 }
 
-#[cfg(test)]
-mod pure_tests {
-    use tempfile::tempdir;
-
-    use super::*;
-
-    #[test]
-    fn storage_path_positive() -> anyhow::Result<()> {
-        let workdir = tempdir()?.path().to_owned();
-
-        let storage_root = PathBuf::from("somewhere").join("else");
-        let storage = LocalFs {
-            working_directory: workdir.clone(),
-            storage_root: storage_root.clone(),
-        };
-
-        let local_path = workdir
-            .join("timelines")
-            .join("some_timeline")
-            .join("file_name");
-        let expected_path = storage_root.join(local_path.strip_prefix(&workdir)?);
-
-        let actual_path = PathBuf::from(
-            storage
-                .remote_object_id(&local_path)
-                .expect("Matching path should map to storage path normally")
-                .0,
-        );
-        assert_eq!(
-            expected_path,
-            actual_path,
-            "File paths from workdir should be stored in local fs storage with the same path they have relative to the workdir"
-        );
-
-        Ok(())
-    }
-
-    #[test]
-    fn storage_path_negatives() -> anyhow::Result<()> {
-        #[track_caller]
-        fn storage_path_error(storage: &LocalFs, mismatching_path: &Path) -> String {
-            match storage.remote_object_id(mismatching_path) {
-                Ok(wrong_path) => panic!(
-                    "Expected path '{}' to error, but got storage path: {:?}",
-                    mismatching_path.display(),
-                    wrong_path,
-                ),
-                Err(e) => format!("{:?}", e),
-            }
-        }
-
-        let workdir = tempdir()?.path().to_owned();
-        let storage_root = PathBuf::from("somewhere").join("else");
-        let storage = LocalFs {
-            working_directory: workdir.clone(),
-            storage_root,
-        };
-
-        let error_string = storage_path_error(&storage, &workdir);
-        assert!(error_string.contains("does not belong to this storage"));
-        assert!(error_string.contains(workdir.to_str().unwrap()));
-
-        let mismatching_path_str = "/something/else";
-        let error_message = storage_path_error(&storage, Path::new(mismatching_path_str));
-        assert!(
-            error_message.contains(mismatching_path_str),
-            "Error should mention wrong path"
-        );
-        assert!(
-            error_message.contains(workdir.to_str().unwrap()),
-            "Error should mention server workdir"
-        );
-        assert!(error_message.contains("does not belong to this storage"));
-
-        Ok(())
-    }
-
-    #[test]
-    fn local_path_positive() -> anyhow::Result<()> {
-        let workdir = tempdir()?.path().to_owned();
-        let storage_root = PathBuf::from("somewhere").join("else");
-        let storage = LocalFs {
-            working_directory: workdir.clone(),
-            storage_root: storage_root.clone(),
-        };
-
-        let name = "not a metadata";
-        let local_path = workdir.join("timelines").join("some_timeline").join(name);
-        assert_eq!(
-            local_path,
-            storage
-                .local_path(&remote_object_id_from_path(
-                    &storage_root.join(local_path.strip_prefix(&workdir)?)
-                )?)
-                .expect("For a valid input, valid local path should be parsed"),
-            "Should be able to parse metadata out of the correctly named remote delta file"
-        );
-
-        let local_metadata_path = workdir
-            .join("timelines")
-            .join("some_timeline")
-            .join("metadata");
-        let remote_metadata_path = storage.remote_object_id(&local_metadata_path)?;
-        assert_eq!(
-            local_metadata_path,
-            storage
-                .local_path(&remote_metadata_path)
-                .expect("For a valid input, valid local path should be parsed"),
-            "Should be able to parse metadata out of the correctly named remote metadata file"
-        );
-
-        Ok(())
-    }
-
-    #[test]
-    fn local_path_negatives() -> anyhow::Result<()> {
-        #[track_caller]
-        fn local_path_error(storage: &LocalFs, storage_path: &RemoteObjectId) -> String {
-            match storage.local_path(storage_path) {
-                Ok(wrong_path) => panic!(
-                    "Expected local path input {:?} to cause an error, but got file path: {:?}",
-                    storage_path, wrong_path,
-                ),
-                Err(e) => format!("{:?}", e),
-            }
-        }
-
-        let storage_root = PathBuf::from("somewhere").join("else");
-        let storage = LocalFs {
-            working_directory: tempdir()?.path().to_owned(),
-            storage_root,
-        };
-
-        let totally_wrong_path = "wrong_wrong_wrong";
-        let error_message =
-            local_path_error(&storage, &RemoteObjectId(totally_wrong_path.to_string()));
-        assert!(error_message.contains(totally_wrong_path));
-
-        Ok(())
-    }
-
-    #[test]
-    fn download_destination_matches_original_path() -> anyhow::Result<()> {
-        let workdir = tempdir()?.path().to_owned();
-        let original_path = workdir
-            .join("timelines")
-            .join("some_timeline")
-            .join("some name");
-
-        let storage_root = PathBuf::from("somewhere").join("else");
-        let dummy_storage = LocalFs {
-            working_directory: workdir,
-            storage_root,
-        };
-
-        let storage_path = dummy_storage.remote_object_id(&original_path)?;
-        let download_destination = dummy_storage.local_path(&storage_path)?;
-
-        assert_eq!(
-            original_path, download_destination,
-            "'original path -> storage path -> matching fs path' transformation should produce the same path as the input one for the correct path"
-        );
-
-        Ok(())
-    }
-}
-
 #[cfg(test)]
 mod fs_tests {
     use super::*;
@@ -572,7 +369,7 @@ mod fs_tests {
         storage: &LocalFs,
         #[allow(clippy::ptr_arg)]
         // have to use &PathBuf due to `storage.local_path` parameter requirements
-        remote_storage_path: &RemoteObjectId,
+        remote_storage_path: &RemotePath,
         expected_metadata: Option<&StorageMetadata>,
     ) -> anyhow::Result<String> {
         let mut download = storage
@@ -595,41 +392,16 @@ mod fs_tests {
 
     #[tokio::test]
     async fn upload_file() -> anyhow::Result<()> {
-        let workdir = tempdir()?.path().to_owned();
         let storage = create_storage()?;
 
-        let (file, size) = create_file_for_upload(
-            &storage.working_directory.join("whatever"),
-            "whatever_contents",
-        )
-        .await?;
-        let target_path = "/somewhere/else";
-        match storage
-            .upload(
-                Box::new(file),
-                size,
-                &RemoteObjectId(target_path.to_string()),
-                None,
-            )
-            .await
-        {
-            Ok(()) => panic!("Should not allow storing files with wrong target path"),
-            Err(e) => {
-                let message = format!("{:?}", e);
-                assert!(message.contains(target_path));
-                assert!(message.contains("does not belong to the current storage"));
-            }
-        }
-        assert!(storage.list().await?.is_empty());
-
-        let target_path_1 = upload_dummy_file(&workdir, &storage, "upload_1", None).await?;
+        let target_path_1 = upload_dummy_file(&storage, "upload_1", None).await?;
         assert_eq!(
             storage.list().await?,
             vec![target_path_1.clone()],
             "Should list a single file after first upload"
         );
 
-        let target_path_2 = upload_dummy_file(&workdir, &storage, "upload_2", None).await?;
+        let target_path_2 = upload_dummy_file(&storage, "upload_2", None).await?;
         assert_eq!(
             list_files_sorted(&storage).await?,
             vec![target_path_1.clone(), target_path_2.clone()],
@@ -643,7 +415,7 @@ mod fs_tests {
     async fn upload_file_negatives() -> anyhow::Result<()> {
         let storage = create_storage()?;
 
-        let id = storage.remote_object_id(&storage.working_directory.join("dummy"))?;
+        let id = RemotePath::new(Path::new("dummy"))?;
         let content = std::io::Cursor::new(b"12345");
 
         // Check that you get an error if the size parameter doesn't match the actual
@@ -668,16 +440,14 @@ mod fs_tests {
     }
 
     fn create_storage() -> anyhow::Result<LocalFs> {
-        LocalFs::new(tempdir()?.path().to_owned(), tempdir()?.path().to_owned())
+        LocalFs::new(tempdir()?.path().to_owned())
     }
 
     #[tokio::test]
     async fn download_file() -> anyhow::Result<()> {
-        let workdir = tempdir()?.path().to_owned();
-
         let storage = create_storage()?;
         let upload_name = "upload_1";
-        let upload_target = upload_dummy_file(&workdir, &storage, upload_name, None).await?;
+        let upload_target = upload_dummy_file(&storage, upload_name, None).await?;
 
         let contents = read_and_assert_remote_file_contents(&storage, &upload_target, None).await?;
         assert_eq!(
@@ -687,7 +457,7 @@ mod fs_tests {
         );
 
         let non_existing_path = "somewhere/else";
-        match storage.download(&RemoteObjectId(non_existing_path.to_string())).await {
+        match storage.download(&RemotePath::new(Path::new(non_existing_path))?).await {
             Err(DownloadError::NotFound) => {} // Should get NotFound for non existing keys
             other => panic!("Should get a NotFound error when downloading non-existing storage files, but got: {other:?}"),
         }
@@ -696,11 +466,9 @@ mod fs_tests {
 
     #[tokio::test]
     async fn download_file_range_positive() -> anyhow::Result<()> {
-        let workdir = tempdir()?.path().to_owned();
-
         let storage = create_storage()?;
         let upload_name = "upload_1";
-        let upload_target = upload_dummy_file(&workdir, &storage, upload_name, None).await?;
+        let upload_target = upload_dummy_file(&storage, upload_name, None).await?;
 
         let full_range_download_contents =
             read_and_assert_remote_file_contents(&storage, &upload_target, None).await?;
@@ -766,11 +534,9 @@ mod fs_tests {
 
     #[tokio::test]
     async fn download_file_range_negative() -> anyhow::Result<()> {
-        let workdir = tempdir()?.path().to_owned();
-
         let storage = create_storage()?;
         let upload_name = "upload_1";
-        let upload_target = upload_dummy_file(&workdir, &storage, upload_name, None).await?;
+        let upload_target = upload_dummy_file(&storage, upload_name, None).await?;
 
         let start = 1_000_000_000;
         let end = start + 1;
@@ -812,11 +578,9 @@ mod fs_tests {
 
     #[tokio::test]
     async fn delete_file() -> anyhow::Result<()> {
-        let workdir = tempdir()?.path().to_owned();
-
         let storage = create_storage()?;
         let upload_name = "upload_1";
-        let upload_target = upload_dummy_file(&workdir, &storage, upload_name, None).await?;
+        let upload_target = upload_dummy_file(&storage, upload_name, None).await?;
 
         storage.delete(&upload_target).await?;
         assert!(storage.list().await?.is_empty());
@@ -826,7 +590,8 @@ mod fs_tests {
             Err(e) => {
                 let error_string = e.to_string();
                 assert!(error_string.contains("does not exist"));
-                assert!(error_string.contains(&upload_target.0));
+                let expected_path = upload_target.with_base(&storage.storage_root);
+                assert!(error_string.contains(expected_path.to_str().unwrap()));
             }
         }
         Ok(())
@@ -834,8 +599,6 @@ mod fs_tests {
 
     #[tokio::test]
     async fn file_with_metadata() -> anyhow::Result<()> {
-        let workdir = tempdir()?.path().to_owned();
-
         let storage = create_storage()?;
         let upload_name = "upload_1";
         let metadata = StorageMetadata(HashMap::from([
@@ -843,7 +606,7 @@ mod fs_tests {
             ("two".to_string(), "2".to_string()),
         ]));
         let upload_target =
-            upload_dummy_file(&workdir, &storage, upload_name, Some(metadata.clone())).await?;
+            upload_dummy_file(&storage, upload_name, Some(metadata.clone())).await?;
 
         let full_range_download_contents =
             read_and_assert_remote_file_contents(&storage, &upload_target, Some(&metadata)).await?;
@@ -883,23 +646,32 @@ mod fs_tests {
     }
 
     async fn upload_dummy_file(
-        workdir: &Path,
         storage: &LocalFs,
         name: &str,
         metadata: Option<StorageMetadata>,
-    ) -> anyhow::Result<RemoteObjectId> {
-        let timeline_path = workdir.join("timelines").join("some_timeline");
-        let relative_timeline_path = timeline_path.strip_prefix(&workdir)?;
-        let storage_path = storage.storage_root.join(relative_timeline_path).join(name);
-        let remote_object_id = RemoteObjectId(storage_path.to_str().unwrap().to_string());
-
-        let from_path = storage.working_directory.join(name);
+    ) -> anyhow::Result<RemotePath> {
+        let from_path = storage
+            .storage_root
+            .join("timelines")
+            .join("some_timeline")
+            .join(name);
         let (file, size) = create_file_for_upload(&from_path, &dummy_contents(name)).await?;
 
+        let relative_path = from_path
+            .strip_prefix(&storage.storage_root)
+            .context("Failed to strip storage root prefix")
+            .and_then(RemotePath::new)
+            .with_context(|| {
+                format!(
+                    "Failed to resolve remote part of path {:?} for base {:?}",
+                    from_path, storage.storage_root
+                )
+            })?;
+
         storage
-            .upload(Box::new(file), size, &remote_object_id, metadata)
+            .upload(Box::new(file), size, &relative_path, metadata)
             .await?;
-        remote_object_id_from_path(&storage_path)
+        Ok(relative_path)
     }
 
     async fn create_file_for_upload(
@@ -924,7 +696,7 @@ mod fs_tests {
         format!("contents for {name}")
     }
 
-    async fn list_files_sorted(storage: &LocalFs) -> anyhow::Result<Vec<RemoteObjectId>> {
+    async fn list_files_sorted(storage: &LocalFs) -> anyhow::Result<Vec<RemotePath>> {
         let mut files = storage.list().await?;
         files.sort_by(|a, b| a.0.cmp(&b.0));
         Ok(files)

libs/remote_storage/src/s3_bucket.rs

@@ -4,27 +4,34 @@
 //! allowing multiple api users to independently work with the same S3 bucket, if
 //! their bucket prefixes are both specified and different.
 
-use std::path::{Path, PathBuf};
+use std::env::var;
+use std::sync::Arc;
+use std::time::Duration;
 
 use anyhow::Context;
-use rusoto_core::{
-    credential::{InstanceMetadataProvider, StaticProvider},
-    HttpClient, Region, RusotoError,
+use aws_config::{
+    environment::credentials::EnvironmentVariableCredentialsProvider, imds,
+    imds::credentials::ImdsCredentialsProvider, meta::credentials::provide_credentials_fn,
 };
-use rusoto_s3::{
-    DeleteObjectRequest, GetObjectError, GetObjectRequest, ListObjectsV2Request, PutObjectRequest,
-    S3Client, StreamingBody, S3,
+use aws_sdk_s3::{
+    config::Config,
+    error::{GetObjectError, GetObjectErrorKind},
+    types::{ByteStream, SdkError},
+    Client, Endpoint, Region,
 };
+use aws_smithy_http::body::SdkBody;
+use aws_types::credentials::{CredentialsError, ProvideCredentials};
+use hyper::Body;
 use tokio::{io, sync::Semaphore};
 use tokio_util::io::ReaderStream;
 use tracing::debug;
 
+use super::StorageMetadata;
 use crate::{
-    strip_path_prefix, Download, DownloadError, RemoteObjectId, RemoteStorage, S3Config,
-    REMOTE_STORAGE_PREFIX_SEPARATOR,
+    Download, DownloadError, RemotePath, RemoteStorage, S3Config, REMOTE_STORAGE_PREFIX_SEPARATOR,
 };
 
-use super::StorageMetadata;
+const DEFAULT_IMDS_TIMEOUT: Duration = Duration::from_secs(10);
 
 pub(super) mod metrics {
     use metrics::{register_int_counter_vec, IntCounterVec};
@@ -91,32 +98,9 @@ pub(super) mod metrics {
     }
 }
 
-fn download_destination(
-    id: &RemoteObjectId,
-    workdir: &Path,
-    prefix_to_strip: Option<&str>,
-) -> PathBuf {
-    let path_without_prefix = match prefix_to_strip {
-        Some(prefix) => id.0.strip_prefix(prefix).unwrap_or_else(|| {
-            panic!(
-                "Could not strip prefix '{}' from S3 object key '{}'",
-                prefix, id.0
-            )
-        }),
-        None => &id.0,
-    };
-
-    workdir.join(
-        path_without_prefix
-            .split(REMOTE_STORAGE_PREFIX_SEPARATOR)
-            .collect::<PathBuf>(),
-    )
-}
-
 /// AWS S3 storage.
 pub struct S3Bucket {
-    workdir: PathBuf,
-    client: S3Client,
+    client: Client,
     bucket_name: String,
     prefix_in_bucket: Option<String>,
     // Every request to S3 can be throttled or cancelled, if a certain number of requests per second is exceeded.
@@ -125,50 +109,53 @@ pub struct S3Bucket {
     concurrency_limiter: Semaphore,
 }
 
+#[derive(Default)]
+struct GetObjectRequest {
+    bucket: String,
+    key: String,
+    range: Option<String>,
+}
 impl S3Bucket {
     /// Creates the S3 storage, errors if incorrect AWS S3 configuration provided.
-    pub fn new(aws_config: &S3Config, workdir: PathBuf) -> anyhow::Result<Self> {
+    pub fn new(aws_config: &S3Config) -> anyhow::Result<Self> {
         debug!(
             "Creating s3 remote storage for S3 bucket {}",
             aws_config.bucket_name
         );
-        let region = match aws_config.endpoint.clone() {
-            Some(custom_endpoint) => Region::Custom {
-                name: aws_config.bucket_region.clone(),
-                endpoint: custom_endpoint,
-            },
-            None => aws_config
-                .bucket_region
-                .parse::<Region>()
-                .context("Failed to parse the s3 region from config")?,
-        };
-        let request_dispatcher = HttpClient::new().context("Failed to create S3 http client")?;
+        let mut config_builder = Config::builder()
+            .region(Region::new(aws_config.bucket_region.clone()))
+            .credentials_provider(provide_credentials_fn(|| async {
+                match var("AWS_ACCESS_KEY_ID").is_ok() && var("AWS_SECRET_ACCESS_KEY").is_ok() {
+                    true => {
+                        EnvironmentVariableCredentialsProvider::new()
+                            .provide_credentials()
+                            .await
+                    }
+                    false => {
+                        let imds_client = imds::Client::builder()
+                            .connect_timeout(DEFAULT_IMDS_TIMEOUT)
+                            .read_timeout(DEFAULT_IMDS_TIMEOUT)
+                            .build()
+                            .await
+                            .map_err(CredentialsError::unhandled)?;
+                        ImdsCredentialsProvider::builder()
+                            .imds_client(imds_client)
+                            .build()
+                            .provide_credentials()
+                            .await
+                    }
+                }
+            }));
 
-        let access_key_id = std::env::var("AWS_ACCESS_KEY_ID").ok();
-        let secret_access_key = std::env::var("AWS_SECRET_ACCESS_KEY").ok();
-        // session token is used when authorizing through sso
-        // which is typically the case when testing locally on developer machine
-        let session_token = std::env::var("AWS_SESSION_TOKEN").ok();
-
-        let client = if access_key_id.is_none() && secret_access_key.is_none() {
-            debug!("Using IAM-based AWS access");
-            S3Client::new_with(request_dispatcher, InstanceMetadataProvider::new(), region)
-        } else {
-            debug!(
-                "Using credentials-based AWS access. Session token is set: {}",
-                session_token.is_some()
+        if let Some(custom_endpoint) = aws_config.endpoint.clone() {
+            let endpoint = Endpoint::immutable(
+                custom_endpoint
+                    .parse()
+                    .expect("Failed to parse S3 custom endpoint"),
             );
-            S3Client::new_with(
-                request_dispatcher,
-                StaticProvider::new(
-                    access_key_id.unwrap_or_default(),
-                    secret_access_key.unwrap_or_default(),
-                    session_token,
-                    None,
-                ),
-                region,
-            )
-        };
+            config_builder.set_endpoint_resolver(Some(Arc::new(endpoint)));
+        }
+        let client = Client::from_conf(config_builder.build());
 
         let prefix_in_bucket = aws_config.prefix_in_bucket.as_deref().map(|prefix| {
             let mut prefix = prefix;
@@ -182,16 +169,41 @@ impl S3Bucket {
             }
             prefix
         });
-
         Ok(Self {
             client,
-            workdir,
             bucket_name: aws_config.bucket_name.clone(),
             prefix_in_bucket,
             concurrency_limiter: Semaphore::new(aws_config.concurrency_limit.get()),
         })
     }
 
+    fn s3_object_to_relative_path(&self, key: &str) -> RemotePath {
+        let relative_path =
+            match key.strip_prefix(self.prefix_in_bucket.as_deref().unwrap_or_default()) {
+                Some(stripped) => stripped,
+                // we rely on AWS to return properly prefixed paths
+                // for requests with a certain prefix
+                None => panic!(
+                    "Key {} does not start with bucket prefix {:?}",
+                    key, self.prefix_in_bucket
+                ),
+            };
+        RemotePath(
+            relative_path
+                .split(REMOTE_STORAGE_PREFIX_SEPARATOR)
+                .collect(),
+        )
+    }
+
+    fn relative_path_to_s3_object(&self, path: &RemotePath) -> String {
+        let mut full_path = self.prefix_in_bucket.clone().unwrap_or_default();
+        for segment in path.0.iter() {
+            full_path.push(REMOTE_STORAGE_PREFIX_SEPARATOR);
+            full_path.push_str(segment.to_str().unwrap_or_default());
+        }
+        full_path
+    }
+
     async fn download_object(&self, request: GetObjectRequest) -> Result<Download, DownloadError> {
         let _guard = self
             .concurrency_limiter
@@ -202,20 +214,33 @@ impl S3Bucket {
 
         metrics::inc_get_object();
 
-        match self.client.get_object(request).await {
-            Ok(object_output) => match object_output.body {
-                None => {
-                    metrics::inc_get_object_fail();
-                    Err(DownloadError::Other(anyhow::anyhow!(
-                        "Got no body for the S3 object given"
-                    )))
-                }
-                Some(body) => Ok(Download {
-                    metadata: object_output.metadata.map(StorageMetadata),
-                    download_stream: Box::pin(io::BufReader::new(body.into_async_read())),
-                }),
-            },
-            Err(RusotoError::Service(GetObjectError::NoSuchKey(_))) => Err(DownloadError::NotFound),
+        let get_object = self
+            .client
+            .get_object()
+            .bucket(request.bucket)
+            .key(request.key)
+            .set_range(request.range)
+            .send()
+            .await;
+
+        match get_object {
+            Ok(object_output) => {
+                let metadata = object_output.metadata().cloned().map(StorageMetadata);
+                Ok(Download {
+                    metadata,
+                    download_stream: Box::pin(io::BufReader::new(
+                        object_output.body.into_async_read(),
+                    )),
+                })
+            }
+            Err(SdkError::ServiceError {
+                err:
+                    GetObjectError {
+                        kind: GetObjectErrorKind::NoSuchKey(..),
+                        ..
+                    },
+                ..
+            }) => Err(DownloadError::NotFound),
             Err(e) => {
                 metrics::inc_get_object_fail();
                 Err(DownloadError::Other(anyhow::anyhow!(
@@ -228,25 +253,7 @@ impl S3Bucket {
 
 #[async_trait::async_trait]
 impl RemoteStorage for S3Bucket {
-    fn remote_object_id(&self, local_path: &Path) -> anyhow::Result<RemoteObjectId> {
-        let relative_path = strip_path_prefix(&self.workdir, local_path)?;
-        let mut key = self.prefix_in_bucket.clone().unwrap_or_default();
-        for segment in relative_path {
-            key.push(REMOTE_STORAGE_PREFIX_SEPARATOR);
-            key.push_str(&segment.to_string_lossy());
-        }
-        Ok(RemoteObjectId(key))
-    }
-
-    fn local_path(&self, storage_path: &RemoteObjectId) -> anyhow::Result<PathBuf> {
-        Ok(download_destination(
-            storage_path,
-            &self.workdir,
-            self.prefix_in_bucket.as_deref(),
-        ))
-    }
-
-    async fn list(&self) -> anyhow::Result<Vec<RemoteObjectId>> {
+    async fn list(&self) -> anyhow::Result<Vec<RemotePath>> {
         let mut document_keys = Vec::new();
 
         let mut continuation_token = None;
@@ -261,12 +268,11 @@ impl RemoteStorage for S3Bucket {
 
             let fetch_response = self
                 .client
-                .list_objects_v2(ListObjectsV2Request {
-                    bucket: self.bucket_name.clone(),
-                    prefix: self.prefix_in_bucket.clone(),
-                    continuation_token,
-                    ..ListObjectsV2Request::default()
-                })
+                .list_objects_v2()
+                .bucket(self.bucket_name.clone())
+                .set_prefix(self.prefix_in_bucket.clone())
+                .set_continuation_token(continuation_token)
+                .send()
                 .await
                 .map_err(|e| {
                     metrics::inc_list_objects_fail();
@@ -277,7 +283,7 @@ impl RemoteStorage for S3Bucket {
                     .contents
                     .unwrap_or_default()
                     .into_iter()
-                    .filter_map(|o| Some(RemoteObjectId(o.key?))),
+                    .filter_map(|o| Some(self.s3_object_to_relative_path(o.key()?))),
             );
 
             match fetch_response.continuation_token {
@@ -291,13 +297,10 @@ impl RemoteStorage for S3Bucket {
 
     /// See the doc for `RemoteStorage::list_prefixes`
     /// Note: it wont include empty "directories"
-    async fn list_prefixes(
-        &self,
-        prefix: Option<&RemoteObjectId>,
-    ) -> anyhow::Result<Vec<RemoteObjectId>> {
+    async fn list_prefixes(&self, prefix: Option<&RemotePath>) -> anyhow::Result<Vec<RemotePath>> {
         // get the passed prefix or if it is not set use prefix_in_bucket value
         let list_prefix = prefix
-            .map(|p| p.0.clone())
+            .map(|p| self.relative_path_to_s3_object(p))
             .or_else(|| self.prefix_in_bucket.clone())
             .map(|mut p| {
                 // required to end with a separator
@@ -322,13 +325,12 @@ impl RemoteStorage for S3Bucket {
 
             let fetch_response = self
                 .client
-                .list_objects_v2(ListObjectsV2Request {
-                    bucket: self.bucket_name.clone(),
-                    prefix: list_prefix.clone(),
-                    continuation_token,
-                    delimiter: Some(REMOTE_STORAGE_PREFIX_SEPARATOR.to_string()),
-                    ..ListObjectsV2Request::default()
-                })
+                .list_objects_v2()
+                .bucket(self.bucket_name.clone())
+                .set_prefix(list_prefix.clone())
+                .set_continuation_token(continuation_token)
+                .delimiter(REMOTE_STORAGE_PREFIX_SEPARATOR.to_string())
+                .send()
                 .await
                 .map_err(|e| {
                     metrics::inc_list_objects_fail();
@@ -340,7 +342,7 @@ impl RemoteStorage for S3Bucket {
                     .common_prefixes
                     .unwrap_or_default()
                     .into_iter()
-                    .filter_map(|o| Some(RemoteObjectId(o.prefix?))),
+                    .filter_map(|o| Some(self.s3_object_to_relative_path(o.prefix()?))),
             );
 
             match fetch_response.continuation_token {
@@ -356,7 +358,7 @@ impl RemoteStorage for S3Bucket {
         &self,
         from: Box<(dyn io::AsyncRead + Unpin + Send + Sync + 'static)>,
         from_size_bytes: usize,
-        to: &RemoteObjectId,
+        to: &RemotePath,
         metadata: Option<StorageMetadata>,
     ) -> anyhow::Result<()> {
         let _guard = self
@@ -366,17 +368,18 @@ impl RemoteStorage for S3Bucket {
             .context("Concurrency limiter semaphore got closed during S3 upload")?;
 
         metrics::inc_put_object();
+
+        let body = Body::wrap_stream(ReaderStream::new(from));
+        let bytes_stream = ByteStream::new(SdkBody::from(body));
+
         self.client
-            .put_object(PutObjectRequest {
-                body: Some(StreamingBody::new_with_size(
-                    ReaderStream::new(from),
-                    from_size_bytes,
-                )),
-                bucket: self.bucket_name.clone(),
-                key: to.0.to_owned(),
-                metadata: metadata.map(|m| m.0),
-                ..PutObjectRequest::default()
-            })
+            .put_object()
+            .bucket(self.bucket_name.clone())
+            .key(self.relative_path_to_s3_object(to))
+            .set_metadata(metadata.map(|m| m.0))
+            .content_length(from_size_bytes.try_into()?)
+            .body(bytes_stream)
+            .send()
             .await
             .map_err(|e| {
                 metrics::inc_put_object_fail();
@@ -385,10 +388,10 @@ impl RemoteStorage for S3Bucket {
         Ok(())
     }
 
-    async fn download(&self, from: &RemoteObjectId) -> Result<Download, DownloadError> {
+    async fn download(&self, from: &RemotePath) -> Result<Download, DownloadError> {
         self.download_object(GetObjectRequest {
             bucket: self.bucket_name.clone(),
-            key: from.0.to_owned(),
+            key: self.relative_path_to_s3_object(from),
             ..GetObjectRequest::default()
         })
         .await
@@ -396,7 +399,7 @@ impl RemoteStorage for S3Bucket {
 
     async fn download_byte_range(
         &self,
-        from: &RemoteObjectId,
+        from: &RemotePath,
         start_inclusive: u64,
         end_exclusive: Option<u64>,
     ) -> Result<Download, DownloadError> {
@@ -404,20 +407,19 @@ impl RemoteStorage for S3Bucket {
         // and needs both ends to be exclusive
         let end_inclusive = end_exclusive.map(|end| end.saturating_sub(1));
         let range = Some(match end_inclusive {
-            Some(end_inclusive) => format!("bytes={}-{}", start_inclusive, end_inclusive),
-            None => format!("bytes={}-", start_inclusive),
+            Some(end_inclusive) => format!("bytes={start_inclusive}-{end_inclusive}"),
+            None => format!("bytes={start_inclusive}-"),
         });
 
         self.download_object(GetObjectRequest {
             bucket: self.bucket_name.clone(),
-            key: from.0.to_owned(),
+            key: self.relative_path_to_s3_object(from),
             range,
-            ..GetObjectRequest::default()
         })
         .await
     }
 
-    async fn delete(&self, remote_object_id: &RemoteObjectId) -> anyhow::Result<()> {
+    async fn delete(&self, path: &RemotePath) -> anyhow::Result<()> {
         let _guard = self
             .concurrency_limiter
             .acquire()
@@ -427,11 +429,10 @@ impl RemoteStorage for S3Bucket {
         metrics::inc_delete_object();
 
         self.client
-            .delete_object(DeleteObjectRequest {
-                bucket: self.bucket_name.clone(),
-                key: remote_object_id.0.to_owned(),
-                ..DeleteObjectRequest::default()
-            })
+            .delete_object()
+            .bucket(self.bucket_name.clone())
+            .key(self.relative_path_to_s3_object(path))
+            .send()
             .await
             .map_err(|e| {
                 metrics::inc_delete_object_fail();
@@ -440,181 +441,3 @@ impl RemoteStorage for S3Bucket {
         Ok(())
     }
 }
-
-#[cfg(test)]
-mod tests {
-    use tempfile::tempdir;
-
-    use super::*;
-
-    #[test]
-    fn test_download_destination() -> anyhow::Result<()> {
-        let workdir = tempdir()?.path().to_owned();
-        let local_path = workdir.join("one").join("two").join("test_name");
-        let relative_path = local_path.strip_prefix(&workdir)?;
-
-        let key = RemoteObjectId(format!(
-            "{}{}",
-            REMOTE_STORAGE_PREFIX_SEPARATOR,
-            relative_path
-                .iter()
-                .map(|segment| segment.to_str().unwrap())
-                .collect::<Vec<_>>()
-                .join(&REMOTE_STORAGE_PREFIX_SEPARATOR.to_string()),
-        ));
-
-        assert_eq!(
-            local_path,
-            download_destination(&key, &workdir, None),
-            "Download destination should consist of s3 path joined with the workdir prefix"
-        );
-
-        Ok(())
-    }
-
-    #[test]
-    fn storage_path_positive() -> anyhow::Result<()> {
-        let workdir = tempdir()?.path().to_owned();
-
-        let segment_1 = "matching";
-        let segment_2 = "file";
-        let local_path = &workdir.join(segment_1).join(segment_2);
-
-        let storage = dummy_storage(workdir);
-
-        let expected_key = RemoteObjectId(format!(
-            "{}{REMOTE_STORAGE_PREFIX_SEPARATOR}{segment_1}{REMOTE_STORAGE_PREFIX_SEPARATOR}{segment_2}",
-            storage.prefix_in_bucket.as_deref().unwrap_or_default(),
-        ));
-
-        let actual_key = storage
-            .remote_object_id(local_path)
-            .expect("Matching path should map to S3 path normally");
-        assert_eq!(
-            expected_key,
-            actual_key,
-            "S3 key from the matching path should contain all segments after the workspace prefix, separated with S3 separator"
-        );
-
-        Ok(())
-    }
-
-    #[test]
-    fn storage_path_negatives() -> anyhow::Result<()> {
-        #[track_caller]
-        fn storage_path_error(storage: &S3Bucket, mismatching_path: &Path) -> String {
-            match storage.remote_object_id(mismatching_path) {
-                Ok(wrong_key) => panic!(
-                    "Expected path '{}' to error, but got S3 key: {:?}",
-                    mismatching_path.display(),
-                    wrong_key,
-                ),
-                Err(e) => e.to_string(),
-            }
-        }
-
-        let workdir = tempdir()?.path().to_owned();
-        let storage = dummy_storage(workdir.clone());
-
-        let error_message = storage_path_error(&storage, &workdir);
-        assert!(
-            error_message.contains("Prefix and the path are equal"),
-            "Message '{}' does not contain the required string",
-            error_message
-        );
-
-        let mismatching_path = PathBuf::from("somewhere").join("else");
-        let error_message = storage_path_error(&storage, &mismatching_path);
-        assert!(
-            error_message.contains(mismatching_path.to_str().unwrap()),
-            "Error should mention wrong path"
-        );
-        assert!(
-            error_message.contains(workdir.to_str().unwrap()),
-            "Error should mention server workdir"
-        );
-        assert!(
-            error_message.contains("is not prefixed with"),
-            "Message '{}' does not contain a required string",
-            error_message
-        );
-
-        Ok(())
-    }
-
-    #[test]
-    fn local_path_positive() -> anyhow::Result<()> {
-        let workdir = tempdir()?.path().to_owned();
-        let storage = dummy_storage(workdir.clone());
-        let timeline_dir = workdir.join("timelines").join("test_timeline");
-        let relative_timeline_path = timeline_dir.strip_prefix(&workdir)?;
-
-        let s3_key = create_s3_key(
-            &relative_timeline_path.join("not a metadata"),
-            storage.prefix_in_bucket.as_deref(),
-        );
-        assert_eq!(
-            download_destination(&s3_key, &workdir, storage.prefix_in_bucket.as_deref()),
-            storage
-                .local_path(&s3_key)
-                .expect("For a valid input, valid S3 info should be parsed"),
-            "Should be able to parse metadata out of the correctly named remote delta file"
-        );
-
-        let s3_key = create_s3_key(
-            &relative_timeline_path.join("metadata"),
-            storage.prefix_in_bucket.as_deref(),
-        );
-        assert_eq!(
-            download_destination(&s3_key, &workdir, storage.prefix_in_bucket.as_deref()),
-            storage
-                .local_path(&s3_key)
-                .expect("For a valid input, valid S3 info should be parsed"),
-            "Should be able to parse metadata out of the correctly named remote metadata file"
-        );
-
-        Ok(())
-    }
-
-    #[test]
-    fn download_destination_matches_original_path() -> anyhow::Result<()> {
-        let workdir = tempdir()?.path().to_owned();
-        let original_path = workdir
-            .join("timelines")
-            .join("some_timeline")
-            .join("some name");
-
-        let dummy_storage = dummy_storage(workdir);
-
-        let key = dummy_storage.remote_object_id(&original_path)?;
-        let download_destination = dummy_storage.local_path(&key)?;
-
-        assert_eq!(
-            original_path, download_destination,
-            "'original path -> storage key -> matching fs path' transformation should produce the same path as the input one for the correct path"
-        );
-
-        Ok(())
-    }
-
-    fn dummy_storage(workdir: PathBuf) -> S3Bucket {
-        S3Bucket {
-            workdir,
-            client: S3Client::new("us-east-1".parse().unwrap()),
-            bucket_name: "dummy-bucket".to_string(),
-            prefix_in_bucket: Some("dummy_prefix/".to_string()),
-            concurrency_limiter: Semaphore::new(1),
-        }
-    }
-
-    fn create_s3_key(relative_file_path: &Path, prefix: Option<&str>) -> RemoteObjectId {
-        RemoteObjectId(relative_file_path.iter().fold(
-            prefix.unwrap_or_default().to_string(),
-            |mut path_string, segment| {
-                path_string.push(REMOTE_STORAGE_PREFIX_SEPARATOR);
-                path_string.push_str(segment.to_str().unwrap());
-                path_string
-            },
-        ))
-    }
-}

libs/tenant_size_model/src/lib.rs

@@ -33,8 +33,8 @@ pub struct Segment {
     /// Logical size before this state
     start_size: u64,
 
-    /// Logical size at this state
-    pub end_size: u64,
+    /// Logical size at this state. Can be None in the last Segment of a branch.
+    pub end_size: Option<u64>,
 
     /// Indices to [`Storage::segments`]
     ///
@@ -115,7 +115,7 @@ impl<K: std::hash::Hash + Eq + 'static> Storage<K> {
             start_lsn: 0,
             end_lsn: 0,
             start_size: 0,
-            end_size: 0,
+            end_size: Some(0),
             children_after: Vec::new(),
         };
 
@@ -125,6 +125,39 @@ impl<K: std::hash::Hash + Eq + 'static> Storage<K> {
         }
     }
 
+    /// Advances the branch with a new point, at given LSN.
+    pub fn insert_point<Q: ?Sized>(
+        &mut self,
+        branch: &Q,
+        op: Cow<'static, str>,
+        lsn: u64,
+        size: Option<u64>,
+    ) where
+        K: std::borrow::Borrow<Q>,
+        Q: std::hash::Hash + Eq,
+    {
+        let lastseg_id = *self.branches.get(branch).unwrap();
+        let newseg_id = self.segments.len();
+        let lastseg = &mut self.segments[lastseg_id];
+
+        assert!(lsn > lastseg.end_lsn);
+
+        let newseg = Segment {
+            op,
+            parent: Some(lastseg_id),
+            start_lsn: lastseg.end_lsn,
+            end_lsn: lsn,
+            start_size: lastseg.end_size.unwrap(),
+            end_size: size,
+            children_after: Vec::new(),
+            needed: false,
+        };
+        lastseg.children_after.push(newseg_id);
+
+        self.segments.push(newseg);
+        *self.branches.get_mut(branch).expect("read already") = newseg_id;
+    }
+
     /// Advances the branch with the named operation, by the relative LSN and logical size bytes.
     pub fn modify_branch<Q: ?Sized>(
         &mut self,
@@ -145,8 +178,8 @@ impl<K: std::hash::Hash + Eq + 'static> Storage<K> {
             parent: Some(lastseg_id),
             start_lsn: lastseg.end_lsn,
             end_lsn: lastseg.end_lsn + lsn_bytes,
-            start_size: lastseg.end_size,
-            end_size: (lastseg.end_size as i64 + size_bytes) as u64,
+            start_size: lastseg.end_size.unwrap(),
+            end_size: Some((lastseg.end_size.unwrap() as i64 + size_bytes) as u64),
             children_after: Vec::new(),
             needed: false,
         };
@@ -321,7 +354,7 @@ impl<K: std::hash::Hash + Eq + 'static> Storage<K> {
                 Some(SegmentSize {
                     seg_id,
                     method: SnapshotAfter,
-                    this_size: seg.end_size,
+                    this_size: seg.end_size.unwrap(),
                     children,
                 })
             } else {

libs/tenant_size_model/src/main.rs

@@ -174,7 +174,7 @@ fn graphviz_recurse(segments: &[Segment], node: &SegmentSize) {
     let seg_id = node.seg_id;
     let seg = segments.get(seg_id).unwrap();
     let lsn = seg.end_lsn;
-    let size = seg.end_size;
+    let size = seg.end_size.unwrap_or(0);
     let method = node.method;
 
     println!("  {{");
@@ -226,7 +226,7 @@ fn graphviz_recurse(segments: &[Segment], node: &SegmentSize) {
             print!(
                 " label=\"{} / {}\"",
                 next.end_lsn - seg.end_lsn,
-                (next.end_size as i128 - seg.end_size as i128)
+                (next.end_size.unwrap_or(0) as i128 - seg.end_size.unwrap_or(0) as i128)
             );
         } else {
             print!(" label=\"{}: {}\"", next.op, next.end_lsn - seg.end_lsn);

libs/utils/Cargo.toml

@@ -4,6 +4,7 @@ version = "0.1.0"
 edition = "2021"
 
 [dependencies]
+sentry = "0.29.0"
 async-trait = "0.1"
 anyhow = "1.0"
 bincode = "1.3"

libs/utils/src/auth.rs

@@ -7,7 +7,7 @@ use serde;
 use std::fs;
 use std::path::Path;
 
-use anyhow::{bail, Result};
+use anyhow::Result;
 use jsonwebtoken::{
     decode, encode, Algorithm, DecodingKey, EncodingKey, Header, TokenData, Validation,
 };
@@ -21,8 +21,16 @@ const JWT_ALGORITHM: Algorithm = Algorithm::RS256;
 #[derive(Debug, Serialize, Deserialize, Clone)]
 #[serde(rename_all = "lowercase")]
 pub enum Scope {
+    // Provides access to all data for a specific tenant (specified in `struct Claims` below)
+    // TODO: join these two?
     Tenant,
+    // Provides blanket access to all tenants on the pageserver plus pageserver-wide APIs.
+    // Should only be used e.g. for status check/tenant creation/list.
     PageServerApi,
+    // Provides blanket access to all data on the safekeeper plus safekeeper-wide APIs.
+    // Should only be used e.g. for status check.
+    // Currently also used for connection from any pageserver to any safekeeper.
+    SafekeeperData,
 }
 
 #[serde_as]
@@ -40,22 +48,6 @@ impl Claims {
     }
 }
 
-pub fn check_permission(claims: &Claims, tenant_id: Option<TenantId>) -> Result<()> {
-    match (&claims.scope, tenant_id) {
-        (Scope::Tenant, None) => {
-            bail!("Attempt to access management api with tenant scope. Permission denied")
-        }
-        (Scope::Tenant, Some(tenant_id)) => {
-            if claims.tenant_id.unwrap() != tenant_id {
-                bail!("Tenant id mismatch. Permission denied")
-            }
-            Ok(())
-        }
-        (Scope::PageServerApi, None) => Ok(()), // access to management api for PageServerApi scope
-        (Scope::PageServerApi, Some(_)) => Ok(()), // access to tenant api using PageServerApi scope
-    }
-}
-
 pub struct JwtAuth {
     decoding_key: DecodingKey,
     validation: Validation,

libs/utils/src/fs_ext.rs

@@ -0,0 +1,45 @@
+/// Extensions to `std::fs` types.
+use std::{fs, io, path::Path};
+
+pub trait PathExt {
+    /// Returns an error if `self` is not a directory.
+    fn is_empty_dir(&self) -> io::Result<bool>;
+}
+
+impl<P> PathExt for P
+where
+    P: AsRef<Path>,
+{
+    fn is_empty_dir(&self) -> io::Result<bool> {
+        Ok(fs::read_dir(self)?.into_iter().next().is_none())
+    }
+}
+
+#[cfg(test)]
+mod test {
+    use std::path::PathBuf;
+
+    #[test]
+    fn is_empty_dir() {
+        use super::PathExt;
+
+        let dir = tempfile::tempdir().unwrap();
+        let dir_path = dir.path();
+
+        // test positive case
+        assert!(
+            dir_path.is_empty_dir().expect("test failure"),
+            "new tempdir should be empty"
+        );
+
+        // invoke on a file to ensure it returns an error
+        let file_path: PathBuf = dir_path.join("testfile");
+        let f = std::fs::File::create(&file_path).unwrap();
+        drop(f);
+        assert!(file_path.is_empty_dir().is_err());
+
+        // do it again on a path, we know to be nonexistent
+        std::fs::remove_file(&file_path).unwrap();
+        assert!(file_path.is_empty_dir().is_err());
+    }
+}

libs/utils/src/http/endpoint.rs

@@ -1,6 +1,5 @@
-use crate::auth::{self, Claims, JwtAuth};
+use crate::auth::{Claims, JwtAuth};
 use crate::http::error;
-use crate::id::TenantId;
 use anyhow::anyhow;
 use hyper::header::AUTHORIZATION;
 use hyper::{header::CONTENT_TYPE, Body, Request, Response, Server};
@@ -144,10 +143,14 @@ pub fn auth_middleware<B: hyper::body::HttpBody + Send + Sync + 'static>(
     })
 }
 
-pub fn check_permission(req: &Request<Body>, tenant_id: Option<TenantId>) -> Result<(), ApiError> {
+pub fn check_permission_with(
+    req: &Request<Body>,
+    check_permission: impl Fn(&Claims) -> Result<(), anyhow::Error>,
+) -> Result<(), ApiError> {
     match req.context::<Claims>() {
-        Some(claims) => Ok(auth::check_permission(&claims, tenant_id)
-            .map_err(|err| ApiError::Forbidden(err.to_string()))?),
+        Some(claims) => {
+            Ok(check_permission(&claims).map_err(|err| ApiError::Forbidden(err.to_string()))?)
+        }
         None => Ok(()), // claims is None because auth is disabled
     }
 }

libs/utils/src/id.rs

@@ -3,6 +3,13 @@ use std::{fmt, str::FromStr};
 use hex::FromHex;
 use rand::Rng;
 use serde::{Deserialize, Serialize};
+use thiserror::Error;
+
+#[derive(Error, Debug)]
+pub enum IdError {
+    #[error("invalid id length {0}")]
+    SliceParseError(usize),
+}
 
 /// Neon ID is a 128-bit random ID.
 /// Used to represent various identifiers. Provides handy utility methods and impls.
@@ -22,6 +29,15 @@ impl Id {
         Id::from(arr)
     }
 
+    pub fn from_slice(src: &[u8]) -> Result<Id, IdError> {
+        if src.len() != 16 {
+            return Err(IdError::SliceParseError(src.len()));
+        }
+        let mut id_array = [0u8; 16];
+        id_array.copy_from_slice(src);
+        Ok(id_array.into())
+    }
+
     pub fn as_arr(&self) -> [u8; 16] {
         self.0
     }
@@ -100,6 +116,10 @@ macro_rules! id_newtype {
                 $t(Id::get_from_buf(buf))
             }
 
+            pub fn from_slice(src: &[u8]) -> Result<$t, IdError> {
+                Ok($t(Id::from_slice(src)?))
+            }
+
             pub fn as_arr(&self) -> [u8; 16] {
                 self.0.as_arr()
             }
@@ -204,6 +224,17 @@ pub struct TenantId(Id);
 
 id_newtype!(TenantId);
 
+/// Neon Connection Id identifies long-lived connections (for example a pagestream
+/// connection with the page_service). Is used for better logging and tracing
+///
+/// NOTE: It (de)serializes as an array of hex bytes, so the string representation would look
+/// like `[173,80,132,115,129,226,72,254,170,201,135,108,199,26,228,24]`.
+/// See [`Id`] for alternative ways to serialize it.
+#[derive(Clone, Copy, PartialEq, Eq, Hash, Serialize, Deserialize, PartialOrd, Ord)]
+pub struct ConnectionId(Id);
+
+id_newtype!(ConnectionId);
+
 // A pair uniquely identifying Neon instance.
 #[derive(Debug, Clone, Copy, PartialOrd, Ord, PartialEq, Eq, Hash, Serialize, Deserialize)]
 pub struct TenantTimelineId {

libs/utils/src/lib.rs

@@ -34,6 +34,7 @@ pub mod sock_split;
 pub mod logging;
 
 pub mod lock_file;
+pub mod pid_file;
 
 // Misc
 pub mod accum;
@@ -46,8 +47,30 @@ pub mod tcp_listener;
 pub mod nonblock;
 
 // Default signal handling
+pub mod sentry_init;
 pub mod signals;
 
+pub mod fs_ext;
+
+/// use with fail::cfg("$name", "return(2000)")
+#[macro_export]
+macro_rules! failpoint_sleep_millis_async {
+    ($name:literal) => {{
+        let should_sleep: Option<std::time::Duration> = (|| {
+            fail::fail_point!($name, |v: Option<_>| {
+                let millis = v.unwrap().parse::<u64>().unwrap();
+                Some(Duration::from_millis(millis))
+            });
+            None
+        })();
+        if let Some(d) = should_sleep {
+            tracing::info!("failpoint {:?}: sleeping for {:?}", $name, d);
+            tokio::time::sleep(d).await;
+            tracing::info!("failpoint {:?}: sleep done", $name);
+        }
+    }};
+}
+
 /// This is a shortcut to embed git sha into binaries and avoid copying the same build script to all packages
 ///
 /// we have several cases:

libs/utils/src/lock_file.rs

@@ -1,81 +1,133 @@
-//! A module to create and read lock files. A lock file ensures that only one
-//! process is running at a time, in a particular directory.
+//! A module to create and read lock files.
 //!
-//! File locking is done using [`fcntl::flock`], which means that holding the
-//! lock on file only prevents acquiring another lock on it; all other
-//! operations are still possible on files. Other process can still open, read,
-//! write, or remove the file, for example.
-//! If the file is removed while a process is holding a lock on it,
-//! the process that holds the lock does not get any error or notification.
-//! Furthermore, you can create a new file with the same name and lock the new file,
-//! while the old process is still running.
-//! Deleting the lock file while the locking process is still running is a bad idea!
+//! File locking is done using [`fcntl::flock`] exclusive locks.
+//! The only consumer of this module is currently [`pid_file`].
+//! See the module-level comment there for potential pitfalls
+//! with lock files that are used to store PIDs (pidfiles).
 
-use std::{fs, os::unix::prelude::AsRawFd, path::Path};
+use std::{
+    fs,
+    io::{Read, Write},
+    ops::Deref,
+    os::unix::prelude::AsRawFd,
+    path::{Path, PathBuf},
+};
 
 use anyhow::Context;
-use nix::fcntl;
+use nix::{errno::Errno::EAGAIN, fcntl};
 
 use crate::crashsafe;
 
-pub enum LockCreationResult {
-    Created {
-        new_lock_contents: String,
-        file: fs::File,
-    },
-    AlreadyLocked {
-        existing_lock_contents: String,
-    },
-    CreationFailed(anyhow::Error),
+/// A handle to an open and unlocked, but not-yet-written lock file.
+/// Returned by [`create_exclusive`].
+#[must_use]
+pub struct UnwrittenLockFile {
+    path: PathBuf,
+    file: fs::File,
 }
 
-/// Creates a lock file in the path given and writes the given contents into the file.
-/// Note: The lock is automatically released when the file closed. You might want to use Box::leak to make sure it lives until the end of the program.
-pub fn create_lock_file(lock_file_path: &Path, contents: String) -> LockCreationResult {
-    let lock_file = match fs::OpenOptions::new()
+/// Returned by [`UnwrittenLockFile::write_content`].
+#[must_use]
+pub struct LockFileGuard(fs::File);
+
+impl Deref for LockFileGuard {
+    type Target = fs::File;
+
+    fn deref(&self) -> &Self::Target {
+        &self.0
+    }
+}
+
+impl UnwrittenLockFile {
+    /// Replace the content of this lock file with the byte representation of `contents`.
+    pub fn write_content(mut self, contents: String) -> anyhow::Result<LockFileGuard> {
+        self.file
+            .set_len(0)
+            .context("Failed to truncate lockfile")?;
+        self.file
+            .write_all(contents.as_bytes())
+            .with_context(|| format!("Failed to write '{contents}' contents into lockfile"))?;
+        crashsafe::fsync_file_and_parent(&self.path).context("fsync lockfile")?;
+        Ok(LockFileGuard(self.file))
+    }
+}
+
+/// Creates and opens a lock file in the path, grabs an exclusive flock on it, and returns
+/// a handle that allows overwriting the locked file's content.
+///
+/// The exclusive lock is released when dropping the returned handle.
+///
+/// It is not an error if the file already exists.
+/// It is an error if the file is already locked.
+pub fn create_exclusive(lock_file_path: &Path) -> anyhow::Result<UnwrittenLockFile> {
+    let lock_file = fs::OpenOptions::new()
         .create(true) // O_CREAT
         .write(true)
         .open(lock_file_path)
-        .context("Failed to open lock file")
-    {
-        Ok(file) => file,
-        Err(e) => return LockCreationResult::CreationFailed(e),
-    };
+        .context("open lock file")?;
 
-    match fcntl::flock(
+    let res = fcntl::flock(
         lock_file.as_raw_fd(),
         fcntl::FlockArg::LockExclusiveNonblock,
-    ) {
-        Ok(()) => {
-            match lock_file
-                .set_len(0)
-                .context("Failed to truncate lockfile")
-                .and_then(|()| {
-                    fs::write(lock_file_path, &contents).with_context(|| {
-                        format!("Failed to write '{contents}' contents into lockfile")
-                    })
-                })
-                .and_then(|()| {
-                    crashsafe::fsync_file_and_parent(lock_file_path)
-                        .context("Failed to fsync lockfile")
-                }) {
-                Ok(()) => LockCreationResult::Created {
-                    new_lock_contents: contents,
-                    file: lock_file,
-                },
-                Err(e) => LockCreationResult::CreationFailed(e),
-            }
-        }
-        Err(nix::errno::Errno::EAGAIN) => {
-            match fs::read_to_string(lock_file_path).context("Failed to read lockfile contents") {
-                Ok(existing_lock_contents) => LockCreationResult::AlreadyLocked {
-                    existing_lock_contents,
-                },
-                Err(e) => LockCreationResult::CreationFailed(e),
-            }
-        }
-        Err(e) => {
-            LockCreationResult::CreationFailed(anyhow::anyhow!("Failed to lock lockfile: {e}"))
-        }
+    );
+    match res {
+        Ok(()) => Ok(UnwrittenLockFile {
+            path: lock_file_path.to_owned(),
+            file: lock_file,
+        }),
+        Err(EAGAIN) => anyhow::bail!("file is already locked"),
+        Err(e) => Err(e).context("flock error"),
+    }
+}
+
+/// Returned by [`read_and_hold_lock_file`].
+/// Check out the [`pid_file`] module for what the variants mean
+/// and potential caveats if the lock files that are used to store PIDs.
+pub enum LockFileRead {
+    /// No file exists at the given path.
+    NotExist,
+    /// No other process held the lock file, so we grabbed an flock
+    /// on it and read its contents.
+    /// Release the flock by dropping the [`LockFileGuard`].
+    NotHeldByAnyProcess(LockFileGuard, String),
+    /// The file exists but another process was holding an flock on it.
+    LockedByOtherProcess {
+        not_locked_file: fs::File,
+        content: String,
+    },
+}
+
+/// Open & try to lock the lock file at the given `path`, returning a [handle][`LockFileRead`] to
+/// inspect its content. 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: &Path) -> anyhow::Result<LockFileRead> {
+    let res = fs::OpenOptions::new().read(true).open(path);
+    let mut lock_file = match res {
+        Ok(f) => f,
+        Err(e) => match e.kind() {
+            std::io::ErrorKind::NotFound => return Ok(LockFileRead::NotExist),
+            _ => return Err(e).context("open lock file"),
+        },
+    };
+    let res = fcntl::flock(
+        lock_file.as_raw_fd(),
+        fcntl::FlockArg::LockExclusiveNonblock,
+    );
+    // We need the content regardless of lock success / failure.
+    // But, read it after flock so that, if it succeeded, the content is consistent.
+    let mut content = String::new();
+    lock_file
+        .read_to_string(&mut content)
+        .context("read lock file")?;
+    match res {
+        Ok(()) => Ok(LockFileRead::NotHeldByAnyProcess(
+            LockFileGuard(lock_file),
+            content,
+        )),
+        Err(EAGAIN) => Ok(LockFileRead::LockedByOtherProcess {
+            not_locked_file: lock_file,
+            content,
+        }),
+        Err(e) => Err(e).context("flock error"),
     }
 }

libs/utils/src/lsn.rs

@@ -138,7 +138,7 @@ impl FromStr for Lsn {
     ///
     /// If the input string is missing the '/' character, then use `Lsn::from_hex`
     fn from_str(s: &str) -> Result<Self, Self::Err> {
-        let mut splitter = s.split('/');
+        let mut splitter = s.trim().split('/');
         if let (Some(left), Some(right), None) = (splitter.next(), splitter.next(), splitter.next())
         {
             let left_num = u32::from_str_radix(left, 16).map_err(|_| LsnParseError)?;
@@ -270,6 +270,11 @@ mod tests {
         );
         assert_eq!(Lsn::from_hex("0"), Ok(Lsn(0)));
         assert_eq!(Lsn::from_hex("F12345678AAAA5555"), Err(LsnParseError));
+
+        let expected_lsn = Lsn(0x3C490F8);
+        assert_eq!(" 0/3C490F8".parse(), Ok(expected_lsn));
+        assert_eq!("0/3C490F8 ".parse(), Ok(expected_lsn));
+        assert_eq!(" 0/3C490F8 ".parse(), Ok(expected_lsn));
     }
 
     #[test]

libs/utils/src/pid_file.rs

@@ -0,0 +1,165 @@
+//! Abstraction to create & read pidfiles.
+//!
+//! A pidfile is a file in the filesystem that stores a process's PID.
+//! Its purpose is to implement a singleton behavior where only
+//! one process of some "kind" is supposed to be running at a given time.
+//! The "kind" is identified by the pidfile.
+//!
+//! During process startup, the process that is supposed to be a singleton
+//! must [claim][`claim_for_current_process`] the pidfile first.
+//! If that is unsuccessful, the process must not act as the singleton, i.e.,
+//! it must not access any of the resources that only the singleton may access.
+//!
+//! A common need is to signal a running singleton process, e.g., to make
+//! it shut down and exit.
+//! For that, we have to [`read`] the pidfile. The result of the `read` operation
+//! tells us if there is any singleton process, and if so, what PID it has.
+//! We can then proceed to signal it, although some caveats still apply.
+//! Read the function-level documentation of [`read`] for that.
+//!
+//! ## Never Remove Pidfiles
+//!
+//! It would be natural to assume that the process who claimed the pidfile
+//! should remove it upon exit to avoid leaving a stale pidfile in place.
+//! However, we already have a reliable way to detect staleness of the pidfile,
+//! i.e., the `flock` that [claiming][`claim_for_current_process`] puts on it.
+//!
+//! And further, removing pidfiles would introduce a **catastrophic race condition**
+//! where two processes are running that are supposed to be singletons.
+//! Suppose we were to remove our pidfile during process shutdown.
+//! Here is how the race plays out:
+//! - Suppose we have a service called `myservice` with pidfile `myservice.pidfile`.
+//! - Process `A` starts to shut down.
+//! - Process `B` is just starting up
+//!     - It `open("myservice.pid", O_WRONLY|O_CREAT)` the file
+//!     - It blocks on `flock`
+//! - Process `A` removes the pidfile as the last step of its shutdown procedure
+//!     - `unlink("myservice.pid")
+//! - Process `A` exits
+//!     - This releases its `flock` and unblocks `B`
+//! - Process `B` still has the file descriptor for `myservice.pid` open
+//! - Process `B` writes its PID into `myservice.pid`.
+//! - But the `myservice.pid` file has been unlinked, so, there is `myservice.pid`
+//!   in the directory.
+//! - Process `C` starts
+//!     - It `open("myservice.pid", O_WRONLY|O_CREAT)` which creates a new file (new inode)
+//!     - It `flock`s the file, which, since it's a different file, does not block
+//!     - It writes its PID into the file
+//!
+//! At this point, `B` and `C` are running, which is hazardous.
+//! Morale of the story: don't unlink pidfiles, ever.
+
+use std::{ops::Deref, path::Path};
+
+use anyhow::Context;
+use nix::unistd::Pid;
+
+use crate::lock_file::{self, LockFileRead};
+
+/// Keeps a claim on a pidfile alive until it is dropped.
+/// Returned by [`claim_for_current_process`].
+#[must_use]
+pub struct PidFileGuard(lock_file::LockFileGuard);
+
+impl Deref for PidFileGuard {
+    type Target = lock_file::LockFileGuard;
+
+    fn deref(&self) -> &Self::Target {
+        &self.0
+    }
+}
+
+/// Try to claim `path` as a pidfile for the current process.
+///
+/// If another process has already claimed the pidfile, and it is still running,
+/// this function returns ane error.
+/// Otherwise, the function `flock`s the file and updates its contents to the
+/// current process's PID.
+/// If the update fails, the flock is released and an error returned.
+/// On success, the function returns a [`PidFileGuard`] to keep the flock alive.
+///
+/// ### Maintaining A Claim
+///
+/// It is the caller's responsibility to maintain the claim.
+/// The claim ends as soon as the returned guard object is dropped.
+/// To maintain the claim for the remaining lifetime of the current process,
+/// use [`std::mem::forget`] or similar.
+pub fn claim_for_current_process(path: &Path) -> anyhow::Result<PidFileGuard> {
+    let unwritten_lock_file = lock_file::create_exclusive(path).context("lock file")?;
+    // if any of the next steps fail, we drop the file descriptor and thereby release the lock
+    let guard = unwritten_lock_file
+        .write_content(Pid::this().to_string())
+        .context("write pid to lock file")?;
+    Ok(PidFileGuard(guard))
+}
+
+/// Returned by [`read`].
+pub enum PidFileRead {
+    /// No file exists at the given path.
+    NotExist,
+    /// The given pidfile is currently not claimed by any process.
+    /// To determine this, the [`read`] operation acquired
+    /// an exclusive flock on the file. The lock is still held and responsibility
+    /// to release it is returned through the guard object.
+    /// Before releasing it, other [`claim_for_current_process`] or [`read`] calls
+    /// will fail.
+    ///
+    /// ### Caveats
+    ///
+    /// Do not unlink the pidfile from the filesystem. See module-comment for why.
+    NotHeldByAnyProcess(PidFileGuard),
+    /// The given pidfile is still claimed by another process whose PID is given
+    /// as part of this variant.
+    ///
+    /// ### Caveats
+    ///
+    /// 1. The other process might exit at any time, turning the given PID stale.
+    /// 2. There is a small window in which `claim_for_current_process` has already
+    ///    locked the file but not yet updates its contents. [`read`] will return
+    ///    this variant here, but with the old file contents, i.e., a stale PID.
+    ///
+    /// The kernel is free to recycle PID once it has been `wait(2)`ed upon by
+    /// its creator. Thus, acting upon a stale PID, e.g., by issuing a `kill`
+    /// system call on it, bears the risk of killing an unrelated process.
+    /// This is an inherent limitation of using pidfiles.
+    /// The only race-free solution is to have a supervisor-process with a lifetime
+    /// that exceeds that of all of its child-processes (e.g., `runit`, `supervisord`).
+    LockedByOtherProcess(Pid),
+}
+
+/// Try to read the file at the given path as a pidfile that was previously created
+/// through [`claim_for_current_process`].
+///
+/// On success, this function returns a [`PidFileRead`].
+/// Check its docs for a description of the meaning of its different variants.
+pub fn read(pidfile: &Path) -> anyhow::Result<PidFileRead> {
+    let res = lock_file::read_and_hold_lock_file(pidfile).context("read and hold pid file")?;
+    let ret = match res {
+        LockFileRead::NotExist => PidFileRead::NotExist,
+        LockFileRead::NotHeldByAnyProcess(guard, _) => {
+            PidFileRead::NotHeldByAnyProcess(PidFileGuard(guard))
+        }
+        LockFileRead::LockedByOtherProcess {
+            not_locked_file: _not_locked_file,
+            content,
+        } => {
+            // XXX the read races with the write in claim_pid_file_for_pid().
+            // But pids are smaller than a page, so the kernel page cache will lock for us.
+            // The only problem is that we might get the old contents here.
+            // Can only fix that by implementing some scheme that downgrades the
+            // exclusive lock to shared lock in claim_pid_file_for_pid().
+            PidFileRead::LockedByOtherProcess(parse_pidfile_content(&content)?)
+        }
+    };
+    Ok(ret)
+}
+
+fn parse_pidfile_content(content: &str) -> anyhow::Result<Pid> {
+    let pid: i32 = content
+        .parse()
+        .map_err(|_| anyhow::anyhow!("parse pidfile content to PID"))?;
+    if pid < 1 {
+        anyhow::bail!("bad value in pidfile '{pid}'");
+    }
+    Ok(Pid::from_raw(pid))
+}

libs/utils/src/sentry_init.rs

@@ -0,0 +1,27 @@
+use sentry::ClientInitGuard;
+use std::borrow::Cow;
+use std::env;
+
+pub use sentry::release_name;
+
+#[must_use]
+pub fn init_sentry(
+    release_name: Option<Cow<'static, str>>,
+    extra_options: &[(&str, &str)],
+) -> Option<ClientInitGuard> {
+    let dsn = env::var("SENTRY_DSN").ok()?;
+
+    let guard = sentry::init((
+        dsn,
+        sentry::ClientOptions {
+            release: release_name,
+            ..Default::default()
+        },
+    ));
+    sentry::configure_scope(|scope| {
+        for &(key, value) in extra_options {
+            scope.set_extra(key, value.into());
+        }
+    });
+    Some(guard)
+}

pageserver/Cargo.toml

@@ -18,7 +18,7 @@ async-stream = "0.3"
 async-trait = "0.1"
 byteorder = "1.4.3"
 bytes = "1.0.1"
-chrono = "0.4.19"
+chrono = { version = "0.4.23", default-features = false, features = ["clock"] }
 clap = { version = "4.0", features = ["string"] }
 close_fds = "0.3.2"
 const_format = "0.2.21"
@@ -35,6 +35,7 @@ itertools = "0.10.3"
 nix = "0.25"
 num-traits = "0.2.15"
 once_cell = "1.13.0"
+pin-project-lite = "0.2.7"
 postgres = { git = "https://github.com/neondatabase/rust-postgres.git", rev="d052ee8b86fff9897c77b0fe89ea9daba0e1fa38" }
 postgres-protocol = { git = "https://github.com/neondatabase/rust-postgres.git", rev="d052ee8b86fff9897c77b0fe89ea9daba0e1fa38" }
 postgres-types = { git = "https://github.com/neondatabase/rust-postgres.git", rev="d052ee8b86fff9897c77b0fe89ea9daba0e1fa38" }
@@ -61,6 +62,7 @@ walkdir = "2.3.2"
 etcd_broker = { path = "../libs/etcd_broker" }
 metrics = { path = "../libs/metrics" }
 pageserver_api = { path = "../libs/pageserver_api" }
+postgres_connection = { path = "../libs/postgres_connection" }
 postgres_ffi = { path = "../libs/postgres_ffi" }
 pq_proto = { path = "../libs/pq_proto" }
 remote_storage = { path = "../libs/remote_storage" }
@@ -76,3 +78,7 @@ tempfile = "3.2"
 [[bench]]
 name = "bench_layer_map"
 harness = false
+
+[[bench]]
+name = "bench_walredo"
+harness = false

pageserver/benches/README.md

@@ -0,0 +1,12 @@
+## Pageserver Benchmarks
+
+# How to run
+
+To run all benchmarks:
+`cargo bench`
+
+To run a specific file:
+`cargo bench --bench bench_layer_map`
+
+To run a specific function:
+`cargo bench --bench bench_layer_map -- real_map_uniform_queries`

pageserver/src/auth.rs

@@ -0,0 +1,22 @@
+use anyhow::{bail, Result};
+use utils::auth::{Claims, Scope};
+use utils::id::TenantId;
+
+pub fn check_permission(claims: &Claims, tenant_id: Option<TenantId>) -> Result<()> {
+    match (&claims.scope, tenant_id) {
+        (Scope::Tenant, None) => {
+            bail!("Attempt to access management api with tenant scope. Permission denied")
+        }
+        (Scope::Tenant, Some(tenant_id)) => {
+            if claims.tenant_id.unwrap() != tenant_id {
+                bail!("Tenant id mismatch. Permission denied")
+            }
+            Ok(())
+        }
+        (Scope::PageServerApi, None) => Ok(()), // access to management api for PageServerApi scope
+        (Scope::PageServerApi, Some(_)) => Ok(()), // access to tenant api using PageServerApi scope
+        (Scope::SafekeeperData, _) => {
+            bail!("SafekeeperData scope makes no sense for Pageserver")
+        }
+    }
+}

pageserver/src/bin/draw_timeline_dir.rs

@@ -11,8 +11,8 @@
 //!
 //! Example use:
 //! ```
-//! $ cd test_output/test_pgbench\[neon-45-684\]/repo/tenants/$TENANT/timelines/$TIMELINE
-//! $ ls | grep "__" | cargo run --release --bin draw_timeline_dir > out.svg
+//! $ ls test_output/test_pgbench\[neon-45-684\]/repo/tenants/$TENANT/timelines/$TIMELINE | \
+//! $   grep "__" | cargo run --release --bin draw_timeline_dir > out.svg
 //! $ firefox out.svg
 //! ```
 //!
@@ -25,6 +25,8 @@ use anyhow::Result;
 use pageserver::repository::Key;
 use std::cmp::Ordering;
 use std::io::{self, BufRead};
+use std::path::PathBuf;
+use std::str::FromStr;
 use std::{
     collections::{BTreeMap, BTreeSet},
     ops::Range,
@@ -65,7 +67,11 @@ fn main() -> Result<()> {
     let mut ranges: Vec<(Range<Key>, Range<Lsn>)> = vec![];
     let stdin = io::stdin();
     for line in stdin.lock().lines() {
-        let range = parse_filename(&line.unwrap());
+        let line = line.unwrap();
+        let line = PathBuf::from_str(&line).unwrap();
+        let filename = line.file_name().unwrap();
+        let filename = filename.to_str().unwrap();
+        let range = parse_filename(filename);
         ranges.push(range);
     }
 

pageserver/src/bin/pageserver.rs

@@ -1,11 +1,12 @@
 //! Main entry point for the Page Server executable.
 
+use std::env::{var, VarError};
+use std::sync::Arc;
 use std::{env, ops::ControlFlow, path::Path, str::FromStr};
 
 use anyhow::{anyhow, Context};
 use clap::{Arg, ArgAction, Command};
 use fail::FailScenario;
-use nix::unistd::Pid;
 use tracing::*;
 
 use metrics::set_build_info_metric;
@@ -21,9 +22,10 @@ use pageserver::{
 use remote_storage::GenericRemoteStorage;
 use utils::{
     auth::JwtAuth,
-    lock_file, logging,
+    logging,
     postgres_backend::AuthType,
     project_git_version,
+    sentry_init::{init_sentry, release_name},
     signals::{self, Signal},
     tcp_listener,
 };
@@ -83,6 +85,9 @@ fn main() -> anyhow::Result<()> {
         }
     };
 
+    // initialize sentry if SENTRY_DSN is provided
+    let _sentry_guard = init_sentry(release_name!(), &[("node_id", &conf.id.to_string())]);
+
     let tenants_path = conf.tenants_path();
     if !tenants_path.exists() {
         utils::crashsafe::create_dir_all(conf.tenants_path()).with_context(|| {
@@ -199,29 +204,28 @@ fn start_pageserver(conf: &'static PageServerConf) -> anyhow::Result<()> {
     logging::init(conf.log_format)?;
     info!("version: {}", version());
 
+    // If any failpoints were set from FAILPOINTS environment variable,
+    // print them to the log for debugging purposes
+    let failpoints = fail::list();
+    if !failpoints.is_empty() {
+        info!(
+            "started with failpoints: {}",
+            failpoints
+                .iter()
+                .map(|(name, actions)| format!("{name}={actions}"))
+                .collect::<Vec<String>>()
+                .join(";")
+        )
+    }
+
     let lock_file_path = conf.workdir.join(PID_FILE_NAME);
-    let lock_file = match lock_file::create_lock_file(&lock_file_path, Pid::this().to_string()) {
-        lock_file::LockCreationResult::Created {
-            new_lock_contents,
-            file,
-        } => {
-            info!("Created lock file at {lock_file_path:?} with contenst {new_lock_contents}");
-            file
-        }
-        lock_file::LockCreationResult::AlreadyLocked {
-            existing_lock_contents,
-        } => anyhow::bail!(
-            "Could not lock pid file; pageserver is already running in {:?} with PID {}",
-            conf.workdir,
-            existing_lock_contents
-        ),
-        lock_file::LockCreationResult::CreationFailed(e) => {
-            return Err(e.context(format!("Failed to create lock file at {lock_file_path:?}")))
-        }
-    };
+    let lock_file =
+        utils::pid_file::claim_for_current_process(&lock_file_path).context("claim pid file")?;
+    info!("Claimed pid file at {lock_file_path:?}");
+
     // ensure that the lock file is held even if the main thread of the process is panics
     // we need to release the lock file only when the current process is gone
-    let _ = Box::leak(Box::new(lock_file));
+    std::mem::forget(lock_file);
 
     // TODO: Check that it looks like a valid repository before going further
 
@@ -256,18 +260,43 @@ fn start_pageserver(conf: &'static PageServerConf) -> anyhow::Result<()> {
     };
     info!("Using auth: {:#?}", conf.auth_type);
 
+    match var("ZENITH_AUTH_TOKEN") {
+        Ok(v) => {
+            info!("Loaded JWT token for authentication with Safekeeper");
+            pageserver::config::SAFEKEEPER_AUTH_TOKEN
+                .set(Arc::new(v))
+                .map_err(|_| anyhow!("Could not initialize SAFEKEEPER_AUTH_TOKEN"))?;
+        }
+        Err(VarError::NotPresent) => {
+            info!("No JWT token for authentication with Safekeeper detected");
+        }
+        Err(e) => {
+            return Err(e).with_context(|| {
+                "Failed to either load to detect non-present ZENITH_AUTH_TOKEN environment variable"
+            })
+        }
+    };
+
     let remote_storage = conf
         .remote_storage_config
         .as_ref()
-        .map(|storage_config| {
-            GenericRemoteStorage::from_config(conf.workdir.clone(), storage_config)
-        })
+        .map(GenericRemoteStorage::from_config)
         .transpose()
         .context("Failed to init generic remote storage")?;
-    let remote_index = {
-        let _rt_guard = BACKGROUND_RUNTIME.enter();
-        tenant_mgr::init_tenant_mgr(conf, remote_storage.clone())?
-    };
+
+    let (init_result_sender, init_result_receiver) =
+        std::sync::mpsc::channel::<anyhow::Result<()>>();
+    let storage_for_spawn = remote_storage.clone();
+    let _handler = BACKGROUND_RUNTIME.spawn(async move {
+        let result = tenant_mgr::init_tenant_mgr(conf, storage_for_spawn).await;
+        init_result_sender.send(result)
+    });
+    match init_result_receiver.recv() {
+        Ok(init_result) => init_result.context("Failed to init tenant_mgr")?,
+        Err(_sender_dropped_err) => {
+            anyhow::bail!("Failed to init tenant_mgr: no init status was returned");
+        }
+    }
 
     // Spawn all HTTP related tasks in the MGMT_REQUEST_RUNTIME.
     // bind before launching separate thread so the error reported before startup exits
@@ -276,7 +305,7 @@ fn start_pageserver(conf: &'static PageServerConf) -> anyhow::Result<()> {
     {
         let _rt_guard = MGMT_REQUEST_RUNTIME.enter();
 
-        let router = http::make_router(conf, auth.clone(), remote_index, remote_storage)?;
+        let router = http::make_router(conf, auth.clone(), remote_storage)?;
         let service =
             utils::http::RouterService::new(router.build().map_err(|err| anyhow!(err))?).unwrap();
         let server = hyper::Server::from_tcp(http_listener)?

pageserver/src/config.rs

@@ -5,13 +5,16 @@
 //! See also `settings.md` for better description on every parameter.
 
 use anyhow::{anyhow, bail, ensure, Context, Result};
-use remote_storage::RemoteStorageConfig;
+use remote_storage::{RemotePath, RemoteStorageConfig};
 use std::env;
 use utils::crashsafe::path_with_suffix_extension;
+use utils::id::ConnectionId;
 
+use once_cell::sync::OnceCell;
 use std::num::NonZeroUsize;
 use std::path::{Path, PathBuf};
 use std::str::FromStr;
+use std::sync::Arc;
 use std::time::Duration;
 use toml_edit;
 use toml_edit::{Document, Item};
@@ -22,13 +25,11 @@ use utils::{
     postgres_backend::AuthType,
 };
 
-use crate::tenant::TIMELINES_SEGMENT_NAME;
+use crate::tenant::{TENANT_ATTACHING_MARKER_FILENAME, TIMELINES_SEGMENT_NAME};
 use crate::tenant_config::{TenantConf, TenantConfOpt};
-
-/// The name of the metadata file pageserver creates per timeline.
-pub const METADATA_FILE_NAME: &str = "metadata";
-pub const TIMELINE_UNINIT_MARK_SUFFIX: &str = "___uninit";
-const TENANT_CONFIG_NAME: &str = "config";
+use crate::{
+    IGNORED_TENANT_FILE_NAME, METADATA_FILE_NAME, TENANT_CONFIG_NAME, TIMELINE_UNINIT_MARK_SUFFIX,
+};
 
 pub mod defaults {
     use crate::tenant_config::defaults::*;
@@ -144,6 +145,15 @@ pub struct PageServerConf {
     pub concurrent_tenant_size_logical_size_queries: ConfigurableSemaphore,
 }
 
+/// We do not want to store this in a PageServerConf because the latter may be logged
+/// and/or serialized at a whim, while the token is secret. Currently this token is the
+/// same for accessing all tenants/timelines, but may become per-tenant/per-timeline in
+/// the future, more tokens and auth may arrive for etcd and/or its rewrite (see
+/// https://github.com/neondatabase/neon/issues/2394), completely changing the logic.
+/// Hence, we resort to a global variable for now instead of passing the token from the
+/// startup code to the connection code through a dozen layers.
+pub static SAFEKEEPER_AUTH_TOKEN: OnceCell<Arc<String>> = OnceCell::new();
+
 #[derive(Debug, Clone, PartialEq, Eq)]
 pub enum ProfilingConfig {
     Disabled,
@@ -323,10 +333,6 @@ impl PageServerConfigBuilder {
     }
 
     pub fn build(self) -> anyhow::Result<PageServerConf> {
-        let broker_endpoints = self
-            .broker_endpoints
-            .ok_or(anyhow!("No broker endpoints provided"))?;
-
         Ok(PageServerConf {
             listen_pg_addr: self
                 .listen_pg_addr
@@ -362,7 +368,9 @@ impl PageServerConfigBuilder {
             profiling: self.profiling.ok_or(anyhow!("missing profiling"))?,
             // TenantConf is handled separately
             default_tenant_conf: TenantConf::default(),
-            broker_endpoints,
+            broker_endpoints: self
+                .broker_endpoints
+                .ok_or(anyhow!("No broker endpoints provided"))?,
             broker_etcd_prefix: self
                 .broker_etcd_prefix
                 .ok_or(anyhow!("missing broker_etcd_prefix"))?,
@@ -389,6 +397,15 @@ impl PageServerConf {
         self.tenants_path().join(tenant_id.to_string())
     }
 
+    pub fn tenant_attaching_mark_file_path(&self, tenant_id: &TenantId) -> PathBuf {
+        self.tenant_path(tenant_id)
+            .join(TENANT_ATTACHING_MARKER_FILENAME)
+    }
+
+    pub fn tenant_ignore_mark_file_path(&self, tenant_id: TenantId) -> PathBuf {
+        self.tenant_path(&tenant_id).join(IGNORED_TENANT_FILE_NAME)
+    }
+
     /// Points to a place in pageserver's local directory,
     /// where certain tenant's tenantconf file should be located.
     pub fn tenant_config_path(&self, tenant_id: TenantId) -> PathBuf {
@@ -414,6 +431,22 @@ impl PageServerConf {
         )
     }
 
+    pub fn traces_path(&self) -> PathBuf {
+        self.workdir.join("traces")
+    }
+
+    pub fn trace_path(
+        &self,
+        tenant_id: &TenantId,
+        timeline_id: &TimelineId,
+        connection_id: &ConnectionId,
+    ) -> PathBuf {
+        self.traces_path()
+            .join(tenant_id.to_string())
+            .join(timeline_id.to_string())
+            .join(connection_id.to_string())
+    }
+
     /// Points to a place in pageserver's local directory,
     /// where certain timeline's metadata file should be located.
     pub fn metadata_path(&self, timeline_id: TimelineId, tenant_id: TenantId) -> PathBuf {
@@ -421,6 +454,28 @@ impl PageServerConf {
             .join(METADATA_FILE_NAME)
     }
 
+    /// Files on the remote storage are stored with paths, relative to the workdir.
+    /// That path includes in itself both tenant and timeline ids, allowing to have a unique remote storage path.
+    ///
+    /// Errors if the path provided does not start from pageserver's workdir.
+    pub fn remote_path(&self, local_path: &Path) -> anyhow::Result<RemotePath> {
+        local_path
+            .strip_prefix(&self.workdir)
+            .context("Failed to strip workdir prefix")
+            .and_then(RemotePath::new)
+            .with_context(|| {
+                format!(
+                    "Failed to resolve remote part of path {:?} for base {:?}",
+                    local_path, self.workdir
+                )
+            })
+    }
+
+    /// Turns storage remote path of a file into its local path.
+    pub fn local_path(&self, remote_path: &RemotePath) -> PathBuf {
+        remote_path.with_base(&self.workdir)
+    }
+
     //
     // Postgres distribution paths
     //
@@ -457,7 +512,7 @@ impl PageServerConf {
         let mut builder = PageServerConfigBuilder::default();
         builder.workdir(workdir.to_owned());
 
-        let mut t_conf: TenantConfOpt = Default::default();
+        let mut t_conf = TenantConfOpt::default();
 
         for (key, item) in toml.iter() {
             match key {
@@ -588,6 +643,12 @@ impl PageServerConf {
         if let Some(max_lsn_wal_lag) = item.get("max_lsn_wal_lag") {
             t_conf.max_lsn_wal_lag = Some(parse_toml_from_str("max_lsn_wal_lag", max_lsn_wal_lag)?);
         }
+        if let Some(trace_read_requests) = item.get("trace_read_requests") {
+            t_conf.trace_read_requests =
+                Some(trace_read_requests.as_bool().with_context(|| {
+                    "configure option trace_read_requests is not a bool".to_string()
+                })?);
+        }
 
         Ok(t_conf)
     }
@@ -597,8 +658,9 @@ impl PageServerConf {
         PathBuf::from(format!("../tmp_check/test_{test_name}"))
     }
 
-    #[cfg(test)]
     pub fn dummy_conf(repo_dir: PathBuf) -> Self {
+        let pg_distrib_dir = PathBuf::from(env!("CARGO_MANIFEST_DIR")).join("../pg_install");
+
         PageServerConf {
             id: NodeId(0),
             wait_lsn_timeout: Duration::from_secs(60),
@@ -609,7 +671,7 @@ impl PageServerConf {
             listen_http_addr: defaults::DEFAULT_HTTP_LISTEN_ADDR.to_string(),
             superuser: "cloud_admin".to_string(),
             workdir: repo_dir,
-            pg_distrib_dir: PathBuf::new(),
+            pg_distrib_dir,
             auth_type: AuthType::Trust,
             auth_validation_public_key_path: None,
             remote_storage_config: None,
@@ -986,6 +1048,35 @@ broker_endpoints = ['{broker_endpoint}']
         Ok(())
     }
 
+    #[test]
+    fn parse_tenant_config() -> anyhow::Result<()> {
+        let tempdir = tempdir()?;
+        let (workdir, pg_distrib_dir) = prepare_fs(&tempdir)?;
+
+        let broker_endpoint = "http://127.0.0.1:7777";
+        let trace_read_requests = true;
+
+        let config_string = format!(
+            r#"{ALL_BASE_VALUES_TOML}
+pg_distrib_dir='{}'
+broker_endpoints = ['{broker_endpoint}']
+
+[tenant_config]
+trace_read_requests = {trace_read_requests}"#,
+            pg_distrib_dir.display(),
+        );
+
+        let toml = config_string.parse()?;
+
+        let conf = PageServerConf::parse_and_validate(&toml, &workdir)?;
+        assert_eq!(
+            conf.default_tenant_conf.trace_read_requests, trace_read_requests,
+            "Tenant config from pageserver config file should be parsed and udpated values used as defaults for all tenants",
+        );
+
+        Ok(())
+    }
+
     fn prepare_fs(tempdir: &TempDir) -> anyhow::Result<(PathBuf, PathBuf)> {
         let tempdir_path = tempdir.path();
 

pageserver/src/http/openapi_spec.yml

@@ -274,6 +274,7 @@ paths:
         schema:
           type: string
           format: hex
+
     post:
       description: Schedules attach operation to happen in the background for given tenant
       responses:
@@ -325,7 +326,9 @@ paths:
           type: string
           format: hex
     post:
-      description: Detach local tenant
+      description: |
+        Remove tenant data (including all corresponding timelines) from pageserver's memory and file system.
+        Files on the remote storage are not affected.
       responses:
         "200":
           description: Tenant detached
@@ -354,6 +357,92 @@ paths:
               schema:
                 $ref: "#/components/schemas/Error"
 
+  /v1/tenant/{tenant_id}/ignore:
+    parameters:
+      - name: tenant_id
+        in: path
+        required: true
+        schema:
+          type: string
+          format: hex
+    post:
+      description: |
+        Remove tenant data (including all corresponding timelines) from pageserver's memory.
+        Files on local disk and remote storage are not affected.
+
+        Future pageserver restarts won't load the data back until `load` is called on such tenant.
+      responses:
+        "200":
+          description: Tenant ignored
+        "400":
+          description: Error when no tenant id found in path parameters
+          content:
+            application/json:
+              schema:
+                $ref: "#/components/schemas/Error"
+        "401":
+          description: Unauthorized Error
+          content:
+            application/json:
+              schema:
+                $ref: "#/components/schemas/UnauthorizedError"
+        "403":
+          description: Forbidden Error
+          content:
+            application/json:
+              schema:
+                $ref: "#/components/schemas/ForbiddenError"
+        "500":
+          description: Generic operation error
+          content:
+            application/json:
+              schema:
+                $ref: "#/components/schemas/Error"
+
+  /v1/tenant/{tenant_id}/load:
+    parameters:
+      - name: tenant_id
+        in: path
+        required: true
+        schema:
+          type: string
+          format: hex
+    post:
+      description: |
+        Schedules an operation that attempts to load a tenant from the local disk and
+        synchronise it with the remote storage (if enabled), repeating pageserver's restart logic for tenant load.
+        If the tenant was ignored before, removes the ignore mark and continues with load scheduling.
+
+        Errors if the tenant is absent on disk, already present in memory or fails to schedule its load.
+        Scheduling a load does not mean that the tenant would load successfully, check tenant status to ensure load correctness.
+      responses:
+        "202":
+          description: Tenant scheduled to load successfully
+        "400":
+          description: Error when no tenant id found in path parameters
+          content:
+            application/json:
+              schema:
+                $ref: "#/components/schemas/Error"
+        "401":
+          description: Unauthorized Error
+          content:
+            application/json:
+              schema:
+                $ref: "#/components/schemas/UnauthorizedError"
+        "403":
+          description: Forbidden Error
+          content:
+            application/json:
+              schema:
+                $ref: "#/components/schemas/ForbiddenError"
+        "500":
+          description: Generic operation error
+          content:
+            application/json:
+              schema:
+                $ref: "#/components/schemas/Error"
+
   /v1/tenant/{tenant_id}/size:
     parameters:
       - name: tenant_id
@@ -604,13 +693,7 @@ components:
         id:
           type: string
         state:
-            oneOf:
-              - type: string
-              - type: object
-                properties:
-                  background_jobs_running:
-                    type: boolean
-
+          type: string
         current_physical_size:
           type: integer
         has_in_progress_downloads:
@@ -665,8 +748,8 @@ components:
         - tenant_id
         - last_record_lsn
         - disk_consistent_lsn
-        - awaits_download
         - state
+        - latest_gc_cutoff_lsn
       properties:
         timeline_id:
           type: string
@@ -707,10 +790,11 @@ components:
           format: hex
         last_received_msg_ts:
           type: integer
-        awaits_download:
-          type: boolean
         state:
           type: string
+        latest_gc_cutoff_lsn:
+          type: string
+          format: hex
 
         # These 'local' and 'remote' fields just duplicate some of the fields
         # above. They are kept for backwards-compatibility. They can be removed,

pageserver/src/http/routes.rs

@@ -4,30 +4,26 @@ use anyhow::{anyhow, Context, Result};
 use hyper::StatusCode;
 use hyper::{Body, Request, Response, Uri};
 use remote_storage::GenericRemoteStorage;
-use tokio::task::JoinError;
 use tracing::*;
 
-use super::models::{LocalTimelineInfo, RemoteTimelineInfo, TimelineInfo};
 use super::models::{
-    StatusResponse, TenantConfigRequest, TenantCreateRequest, TenantCreateResponse, TenantInfo,
-    TimelineCreateRequest,
+    LocalTimelineInfo, RemoteTimelineInfo, StatusResponse, TenantConfigRequest,
+    TenantCreateRequest, TenantCreateResponse, TenantInfo, TimelineCreateRequest, TimelineInfo,
 };
 use crate::pgdatadir_mapping::LsnForTimestamp;
-use crate::storage_sync;
-use crate::storage_sync::index::{RemoteIndex, RemoteTimeline};
-use crate::tenant::{TenantState, Timeline};
+use crate::tenant::Timeline;
 use crate::tenant_config::TenantConfOpt;
 use crate::{config::PageServerConf, tenant_mgr};
 use utils::{
     auth::JwtAuth,
     http::{
-        endpoint::{self, attach_openapi_ui, auth_middleware, check_permission},
+        endpoint::{self, attach_openapi_ui, auth_middleware, check_permission_with},
         error::{ApiError, HttpErrorBody},
         json::{json_request, json_response},
         request::parse_request_param,
         RequestExt, RouterBuilder,
     },
-    id::{TenantId, TenantTimelineId, TimelineId},
+    id::{TenantId, TimelineId},
     lsn::Lsn,
 };
 
@@ -40,7 +36,6 @@ use crate::CheckpointConfig;
 struct State {
     conf: &'static PageServerConf,
     auth: Option<Arc<JwtAuth>>,
-    remote_index: RemoteIndex,
     allowlist_routes: Vec<Uri>,
     remote_storage: Option<GenericRemoteStorage>,
 }
@@ -49,7 +44,6 @@ impl State {
     fn new(
         conf: &'static PageServerConf,
         auth: Option<Arc<JwtAuth>>,
-        remote_index: RemoteIndex,
         remote_storage: Option<GenericRemoteStorage>,
     ) -> anyhow::Result<Self> {
         let allowlist_routes = ["/v1/status", "/v1/doc", "/swagger.yml"]
@@ -60,7 +54,6 @@ impl State {
             conf,
             auth,
             allowlist_routes,
-            remote_index,
             remote_storage,
         })
     }
@@ -79,19 +72,37 @@ fn get_config(request: &Request<Body>) -> &'static PageServerConf {
     get_state(request).conf
 }
 
+fn check_permission(request: &Request<Body>, tenant_id: Option<TenantId>) -> Result<(), ApiError> {
+    check_permission_with(request, |claims| {
+        crate::auth::check_permission(claims, tenant_id)
+    })
+}
+
 // Helper function to construct a TimelineInfo struct for a timeline
-async fn build_timeline_info(
-    state: &State,
+fn build_timeline_info(
     timeline: &Arc<Timeline>,
     include_non_incremental_logical_size: bool,
     include_non_incremental_physical_size: bool,
 ) -> anyhow::Result<TimelineInfo> {
+    let mut info = build_timeline_info_common(timeline)?;
+    if include_non_incremental_logical_size {
+        info.current_logical_size_non_incremental =
+            Some(timeline.get_current_logical_size_non_incremental(info.last_record_lsn)?);
+    }
+    if include_non_incremental_physical_size {
+        info.current_physical_size_non_incremental =
+            Some(timeline.get_physical_size_non_incremental()?)
+    }
+    Ok(info)
+}
+
+fn build_timeline_info_common(timeline: &Arc<Timeline>) -> anyhow::Result<TimelineInfo> {
     let last_record_lsn = timeline.get_last_record_lsn();
     let (wal_source_connstr, last_received_msg_lsn, last_received_msg_ts) = {
         let guard = timeline.last_received_wal.lock().unwrap();
         if let Some(info) = guard.as_ref() {
             (
-                Some(info.wal_source_connstr.clone()),
+                Some(format!("{:?}", info.wal_source_connconf)), // Password is hidden, but it's for statistics only.
                 Some(info.last_received_msg_lsn),
                 Some(info.last_received_msg_ts),
             )
@@ -100,22 +111,6 @@ async fn build_timeline_info(
         }
     };
 
-    let (remote_consistent_lsn, awaits_download) = if let Some(remote_entry) = state
-        .remote_index
-        .read()
-        .await
-        .timeline_entry(&TenantTimelineId {
-            tenant_id: timeline.tenant_id,
-            timeline_id: timeline.timeline_id,
-        }) {
-        (
-            Some(remote_entry.metadata.disk_consistent_lsn()),
-            remote_entry.awaits_download,
-        )
-    } else {
-        (None, false)
-    };
-
     let ancestor_timeline_id = timeline.get_ancestor_timeline_id();
     let ancestor_lsn = match timeline.get_ancestor_lsn() {
         Lsn(0) => None,
@@ -130,6 +125,7 @@ async fn build_timeline_info(
     };
     let current_physical_size = Some(timeline.get_physical_size());
     let state = timeline.current_state();
+    let remote_consistent_lsn = timeline.get_remote_consistent_lsn().unwrap_or(Lsn(0));
 
     let info = TimelineInfo {
         tenant_id: timeline.tenant_id,
@@ -137,28 +133,19 @@ async fn build_timeline_info(
         ancestor_timeline_id,
         ancestor_lsn,
         disk_consistent_lsn: timeline.get_disk_consistent_lsn(),
+        remote_consistent_lsn,
         last_record_lsn,
         prev_record_lsn: Some(timeline.get_prev_record_lsn()),
         latest_gc_cutoff_lsn: *timeline.get_latest_gc_cutoff_lsn(),
         current_logical_size,
         current_physical_size,
-        current_logical_size_non_incremental: if include_non_incremental_logical_size {
-            Some(timeline.get_current_logical_size_non_incremental(last_record_lsn)?)
-        } else {
-            None
-        },
-        current_physical_size_non_incremental: if include_non_incremental_physical_size {
-            Some(timeline.get_physical_size_non_incremental()?)
-        } else {
-            None
-        },
+        current_logical_size_non_incremental: None,
+        current_physical_size_non_incremental: None,
         wal_source_connstr,
         last_received_msg_lsn,
         last_received_msg_ts,
         pg_version: timeline.pg_version,
 
-        remote_consistent_lsn,
-        awaits_download,
         state,
 
         // Duplicate some fields in 'local' and 'remote' fields, for backwards-compatility
@@ -170,7 +157,7 @@ async fn build_timeline_info(
             current_physical_size,
         },
         remote: RemoteTimelineInfo {
-            remote_consistent_lsn,
+            remote_consistent_lsn: Some(remote_consistent_lsn),
         },
     };
     Ok(info)
@@ -178,6 +165,7 @@ async fn build_timeline_info(
 
 // healthcheck handler
 async fn status_handler(request: Request<Body>) -> Result<Response<Body>, ApiError> {
+    check_permission(&request, None)?;
     let config = get_config(&request);
     json_response(StatusCode::OK, StatusResponse { id: config.id })
 }
@@ -187,34 +175,30 @@ async fn timeline_create_handler(mut request: Request<Body>) -> Result<Response<
     let request_data: TimelineCreateRequest = json_request(&mut request).await?;
     check_permission(&request, Some(tenant_id))?;
 
-    let state = get_state(&request);
+    let new_timeline_id = request_data
+        .new_timeline_id
+        .unwrap_or_else(TimelineId::generate);
 
-    let tenant = tenant_mgr::get_tenant(tenant_id, true).map_err(ApiError::NotFound)?;
-    let new_timeline_info = async {
-        match tenant.create_timeline(
-            request_data.new_timeline_id.map(TimelineId::from),
-            request_data.ancestor_timeline_id.map(TimelineId::from),
-            request_data.ancestor_start_lsn,
-            request_data.pg_version.unwrap_or(crate::DEFAULT_PG_VERSION)
-        ).await {
-            Ok(Some(new_timeline)) => {
-                // Created. Construct a TimelineInfo for it.
-                let timeline_info = build_timeline_info(state, &new_timeline, false, false)
-                    .await
-                    .map_err(ApiError::InternalServerError)?;
-                Ok(Some(timeline_info))
-            }
-            Ok(None) => Ok(None), // timeline already exists
-            Err(err) => Err(ApiError::InternalServerError(err)),
+    let tenant = tenant_mgr::get_tenant(tenant_id, true)
+        .await
+        .map_err(ApiError::NotFound)?;
+    match tenant.create_timeline(
+        new_timeline_id,
+        request_data.ancestor_timeline_id.map(TimelineId::from),
+        request_data.ancestor_start_lsn,
+        request_data.pg_version.unwrap_or(crate::DEFAULT_PG_VERSION)
+    )
+    .instrument(info_span!("timeline_create", tenant = %tenant_id, new_timeline = ?request_data.new_timeline_id, timeline_id = %new_timeline_id, lsn=?request_data.ancestor_start_lsn, pg_version=?request_data.pg_version))
+    .await {
+        Ok(Some(new_timeline)) => {
+            // Created. Construct a TimelineInfo for it.
+            let timeline_info = build_timeline_info_common(&new_timeline)
+                .map_err(ApiError::InternalServerError)?;
+            json_response(StatusCode::CREATED, timeline_info)
         }
+        Ok(None) => json_response(StatusCode::CONFLICT, ()), // timeline already exists
+        Err(err) => Err(ApiError::InternalServerError(err)),
     }
-    .instrument(info_span!("timeline_create", tenant = %tenant_id, new_timeline = ?request_data.new_timeline_id, lsn=?request_data.ancestor_start_lsn, pg_version=?request_data.pg_version))
-        .await?;
-
-    Ok(match new_timeline_info {
-        Some(info) => json_response(StatusCode::CREATED, info)?,
-        None => json_response(StatusCode::CONFLICT, ())?,
-    })
 }
 
 async fn timeline_list_handler(request: Request<Body>) -> Result<Response<Body>, ApiError> {
@@ -225,27 +209,29 @@ async fn timeline_list_handler(request: Request<Body>) -> Result<Response<Body>,
         query_param_present(&request, "include-non-incremental-physical-size");
     check_permission(&request, Some(tenant_id))?;
 
-    let state = get_state(&request);
+    let response_data = async {
+        let tenant = tenant_mgr::get_tenant(tenant_id, true)
+            .await
+            .map_err(ApiError::NotFound)?;
+        let timelines = tenant.list_timelines();
 
-    let timelines = info_span!("timeline_list", tenant = %tenant_id).in_scope(|| {
-        let tenant = tenant_mgr::get_tenant(tenant_id, true).map_err(ApiError::NotFound)?;
-        Ok(tenant.list_timelines())
-    })?;
+        let mut response_data = Vec::with_capacity(timelines.len());
+        for timeline in timelines {
+            let timeline_info = build_timeline_info(
+                &timeline,
+                include_non_incremental_logical_size,
+                include_non_incremental_physical_size,
+            )
+            .context("Failed to convert tenant timeline {timeline_id} into the local one: {e:?}")
+            .map_err(ApiError::InternalServerError)?;
 
-    let mut response_data = Vec::with_capacity(timelines.len());
-    for timeline in timelines {
-        let timeline_info = build_timeline_info(
-            state,
-            &timeline,
-            include_non_incremental_logical_size,
-            include_non_incremental_physical_size,
-        )
-        .await
-        .context("Failed to convert tenant timeline {timeline_id} into the local one: {e:?}")
-        .map_err(ApiError::InternalServerError)?;
+            response_data.push(timeline_info);
+        }
 
-        response_data.push(timeline_info);
+        Ok(response_data)
     }
+    .instrument(info_span!("timeline_list", tenant = %tenant_id))
+    .await?;
 
     json_response(StatusCode::OK, response_data)
 }
@@ -289,24 +275,20 @@ async fn timeline_detail_handler(request: Request<Body>) -> Result<Response<Body
         query_param_present(&request, "include-non-incremental-physical-size");
     check_permission(&request, Some(tenant_id))?;
 
-    let state = get_state(&request);
-
     let timeline_info = async {
-        let timeline = tokio::task::spawn_blocking(move || {
-            tenant_mgr::get_tenant(tenant_id, true)?.get_timeline(timeline_id, false)
-        })
-        .await
-        .map_err(|e: JoinError| ApiError::InternalServerError(e.into()))?;
+        let tenant = tenant_mgr::get_tenant(tenant_id, true)
+            .await
+            .map_err(ApiError::NotFound)?;
 
-        let timeline = timeline.map_err(ApiError::NotFound)?;
+        let timeline = tenant
+            .get_timeline(timeline_id, false)
+            .map_err(ApiError::NotFound)?;
 
         let timeline_info = build_timeline_info(
-            state,
             &timeline,
             include_non_incremental_logical_size,
             include_non_incremental_physical_size,
         )
-        .await
         .context("Failed to get local timeline info: {e:#}")
         .map_err(ApiError::InternalServerError)?;
 
@@ -330,6 +312,7 @@ async fn get_lsn_by_timestamp_handler(request: Request<Body>) -> Result<Response
     let timestamp_pg = postgres_ffi::to_pg_timestamp(timestamp);
 
     let timeline = tenant_mgr::get_tenant(tenant_id, true)
+        .await
         .and_then(|tenant| tenant.get_timeline(timeline_id, true))
         .map_err(ApiError::NotFound)?;
     let result = match timeline
@@ -351,117 +334,28 @@ async fn tenant_attach_handler(request: Request<Body>) -> Result<Response<Body>,
 
     info!("Handling tenant attach {tenant_id}");
 
-    tokio::task::spawn_blocking(move || match tenant_mgr::get_tenant(tenant_id, false) {
-        Ok(tenant) => {
-            if tenant.list_timelines().is_empty() {
-                info!("Attaching to tenant {tenant_id} with zero timelines");
-                Ok(())
-            } else {
-                Err(ApiError::Conflict(
-                    "Tenant is already present locally".to_owned(),
-                ))
-            }
-        }
-        Err(_) => Ok(()),
-    })
-    .await
-    .map_err(|e: JoinError| ApiError::InternalServerError(e.into()))??;
-
     let state = get_state(&request);
-    let remote_index = &state.remote_index;
 
-    let mut index_accessor = remote_index.write().await;
-    if let Some(tenant_entry) = index_accessor.tenant_entry_mut(&tenant_id) {
-        if tenant_entry.has_in_progress_downloads() {
-            return Err(ApiError::Conflict(
-                "Tenant download is already in progress".to_string(),
-            ));
-        }
-
-        for (timeline_id, remote_timeline) in tenant_entry.iter_mut() {
-            storage_sync::schedule_layer_download(tenant_id, *timeline_id);
-            remote_timeline.awaits_download = true;
-        }
-        return json_response(StatusCode::ACCEPTED, ());
-    }
-    // no tenant in the index, release the lock to make the potentially lengthy download operation
-    drop(index_accessor);
-
-    // download index parts for every tenant timeline
-    let remote_timelines = match gather_tenant_timelines_index_parts(state, tenant_id).await {
-        Ok(Some(remote_timelines)) => remote_timelines,
-        Ok(None) => return Err(ApiError::NotFound(anyhow!("Unknown remote tenant"))),
-        Err(e) => {
-            error!("Failed to retrieve remote tenant data: {:?}", e);
-            return Err(ApiError::NotFound(anyhow!(
-                "Failed to retrieve remote tenant"
-            )));
-        }
-    };
-
-    // recheck that download is not in progress because
-    // we've released the lock to avoid holding it during the download
-    let mut index_accessor = remote_index.write().await;
-    let tenant_entry = match index_accessor.tenant_entry_mut(&tenant_id) {
-        Some(tenant_entry) => {
-            if tenant_entry.has_in_progress_downloads() {
-                return Err(ApiError::Conflict(
-                    "Tenant download is already in progress".to_string(),
-                ));
-            }
-            tenant_entry
-        }
-        None => index_accessor.add_tenant_entry(tenant_id),
-    };
-
-    // populate remote index with the data from index part and create directories on the local filesystem
-    for (timeline_id, mut remote_timeline) in remote_timelines {
-        tokio::fs::create_dir_all(state.conf.timeline_path(&timeline_id, &tenant_id))
+    if let Some(remote_storage) = &state.remote_storage {
+        // FIXME: distinguish between "Tenant already exists" and other errors
+        tenant_mgr::attach_tenant(state.conf, tenant_id, remote_storage.clone())
+            .instrument(info_span!("tenant_attach", tenant = %tenant_id))
             .await
-            .context("Failed to create new timeline directory")
             .map_err(ApiError::InternalServerError)?;
-
-        remote_timeline.awaits_download = true;
-        tenant_entry.insert(timeline_id, remote_timeline);
-        // schedule actual download
-        storage_sync::schedule_layer_download(tenant_id, timeline_id);
+    } else {
+        return Err(ApiError::BadRequest(anyhow!(
+            "attach_tenant is not possible because pageserver was configured without remote storage"
+        )));
     }
 
     json_response(StatusCode::ACCEPTED, ())
 }
 
-/// Note: is expensive from s3 access perspective,
-/// for details see comment to `storage_sync::gather_tenant_timelines_index_parts`
-async fn gather_tenant_timelines_index_parts(
-    state: &State,
-    tenant_id: TenantId,
-) -> anyhow::Result<Option<Vec<(TimelineId, RemoteTimeline)>>> {
-    let index_parts = match state.remote_storage.as_ref() {
-        Some(storage) => {
-            storage_sync::gather_tenant_timelines_index_parts(state.conf, storage, tenant_id).await
-        }
-        None => return Ok(None),
-    }
-    .with_context(|| format!("Failed to download index parts for tenant {tenant_id}"))?;
-
-    let mut remote_timelines = Vec::with_capacity(index_parts.len());
-    for (timeline_id, index_part) in index_parts {
-        let timeline_path = state.conf.timeline_path(&timeline_id, &tenant_id);
-        let remote_timeline = RemoteTimeline::from_index_part(&timeline_path, index_part)
-            .with_context(|| {
-                format!("Failed to convert index part into remote timeline for timeline {tenant_id}/{timeline_id}")
-            })?;
-        remote_timelines.push((timeline_id, remote_timeline));
-    }
-    Ok(Some(remote_timelines))
-}
-
 async fn timeline_delete_handler(request: Request<Body>) -> Result<Response<Body>, ApiError> {
     let tenant_id: TenantId = parse_request_param(&request, "tenant_id")?;
     let timeline_id: TimelineId = parse_request_param(&request, "timeline_id")?;
     check_permission(&request, Some(tenant_id))?;
 
-    let state = get_state(&request);
     tenant_mgr::delete_timeline(tenant_id, timeline_id)
         .instrument(info_span!("timeline_delete", tenant = %tenant_id, timeline = %timeline_id))
         .await
@@ -470,12 +364,6 @@ async fn timeline_delete_handler(request: Request<Body>) -> Result<Response<Body
         // it.
         .map_err(ApiError::InternalServerError)?;
 
-    let mut remote_index = state.remote_index.write().await;
-    remote_index.remove_timeline_entry(TenantTimelineId {
-        tenant_id,
-        timeline_id,
-    });
-
     json_response(StatusCode::OK, ())
 }
 
@@ -492,8 +380,34 @@ async fn tenant_detach_handler(request: Request<Body>) -> Result<Response<Body>,
         // Replace this with better handling once the error type permits it.
         .map_err(ApiError::InternalServerError)?;
 
-    let mut remote_index = state.remote_index.write().await;
-    remote_index.remove_tenant_entry(&tenant_id);
+    json_response(StatusCode::OK, ())
+}
+
+async fn tenant_load_handler(request: Request<Body>) -> Result<Response<Body>, ApiError> {
+    let tenant_id: TenantId = parse_request_param(&request, "tenant_id")?;
+    check_permission(&request, Some(tenant_id))?;
+
+    let state = get_state(&request);
+    tenant_mgr::load_tenant(state.conf, tenant_id, state.remote_storage.clone())
+        .instrument(info_span!("load", tenant = %tenant_id))
+        .await
+        .map_err(ApiError::InternalServerError)?;
+
+    json_response(StatusCode::ACCEPTED, ())
+}
+
+async fn tenant_ignore_handler(request: Request<Body>) -> Result<Response<Body>, ApiError> {
+    let tenant_id: TenantId = parse_request_param(&request, "tenant_id")?;
+    check_permission(&request, Some(tenant_id))?;
+
+    let state = get_state(&request);
+    let conf = state.conf;
+    tenant_mgr::ignore_tenant(conf, tenant_id)
+        .instrument(info_span!("ignore_tenant", tenant = %tenant_id))
+        .await
+        // FIXME: Errors from `ignore_tenant` can be caused by both both user and internal errors.
+        // Replace this with better handling once the error type permits it.
+        .map_err(ApiError::InternalServerError)?;
 
     json_response(StatusCode::OK, ())
 }
@@ -501,16 +415,17 @@ async fn tenant_detach_handler(request: Request<Body>) -> Result<Response<Body>,
 async fn tenant_list_handler(request: Request<Body>) -> Result<Response<Body>, ApiError> {
     check_permission(&request, None)?;
 
-    let state = get_state(&request);
-    // clone to avoid holding the lock while awaiting for blocking task
-    let remote_index = state.remote_index.read().await.clone();
-
-    let response_data = tokio::task::spawn_blocking(move || {
-        let _enter = info_span!("tenant_list").entered();
-        crate::tenant_mgr::list_tenant_info(&remote_index)
-    })
-    .await
-    .map_err(|e: JoinError| ApiError::InternalServerError(e.into()))?;
+    let response_data = tenant_mgr::list_tenants()
+        .instrument(info_span!("tenant_list"))
+        .await
+        .iter()
+        .map(|(id, state)| TenantInfo {
+            id: *id,
+            state: *state,
+            current_physical_size: None,
+            has_in_progress_downloads: Some(state.has_in_progress_downloads()),
+        })
+        .collect::<Vec<TenantInfo>>();
 
     json_response(StatusCode::OK, response_data)
 }
@@ -519,58 +434,37 @@ async fn tenant_status(request: Request<Body>) -> Result<Response<Body>, ApiErro
     let tenant_id: TenantId = parse_request_param(&request, "tenant_id")?;
     check_permission(&request, Some(tenant_id))?;
 
-    // if tenant is in progress of downloading it can be absent in global tenant map
-    let tenant = tenant_mgr::get_tenant(tenant_id, false);
+    let tenant_info = async {
+        let tenant = tenant_mgr::get_tenant(tenant_id, false).await?;
 
-    let state = get_state(&request);
-    let remote_index = &state.remote_index;
-
-    let index_accessor = remote_index.read().await;
-    let has_in_progress_downloads = index_accessor
-        .tenant_entry(&tenant_id)
-        .map(|t| t.has_in_progress_downloads())
-        .unwrap_or_else(|| {
-            info!("Tenant {tenant_id} not found in remote index");
-            false
-        });
-
-    let (tenant_state, current_physical_size) = match tenant {
-        Ok(tenant) => {
-            let timelines = tenant.list_timelines();
-            // Calculate total physical size of all timelines
-            let mut current_physical_size = 0;
-            for timeline in timelines {
-                current_physical_size += timeline.get_physical_size();
-            }
-
-            (tenant.current_state(), Some(current_physical_size))
+        // Calculate total physical size of all timelines
+        let mut current_physical_size = 0;
+        for timeline in tenant.list_timelines().iter() {
+            current_physical_size += timeline.get_physical_size();
         }
-        Err(e) => {
-            error!("Failed to get local tenant state: {e:#}");
-            if has_in_progress_downloads {
-                (TenantState::Paused, None)
-            } else {
-                (TenantState::Broken, None)
-            }
-        }
-    };
 
-    json_response(
-        StatusCode::OK,
-        TenantInfo {
+        let state = tenant.current_state();
+        Ok(TenantInfo {
             id: tenant_id,
-            state: tenant_state,
-            current_physical_size,
-            has_in_progress_downloads: Some(has_in_progress_downloads),
-        },
-    )
+            state,
+            current_physical_size: Some(current_physical_size),
+            has_in_progress_downloads: Some(state.has_in_progress_downloads()),
+        })
+    }
+    .instrument(info_span!("tenant_status_handler", tenant = %tenant_id))
+    .await
+    .map_err(ApiError::InternalServerError)?;
+
+    json_response(StatusCode::OK, tenant_info)
 }
 
 async fn tenant_size_handler(request: Request<Body>) -> Result<Response<Body>, ApiError> {
     let tenant_id: TenantId = parse_request_param(&request, "tenant_id")?;
     check_permission(&request, Some(tenant_id))?;
 
-    let tenant = tenant_mgr::get_tenant(tenant_id, false).map_err(ApiError::InternalServerError)?;
+    let tenant = tenant_mgr::get_tenant(tenant_id, true)
+        .await
+        .map_err(ApiError::InternalServerError)?;
 
     // this can be long operation, it currently is not backed by any request coalescing or similar
     let inputs = tenant
@@ -618,7 +512,6 @@ async fn tenant_create_handler(mut request: Request<Body>) -> Result<Response<Bo
     check_permission(&request, None)?;
 
     let request_data: TenantCreateRequest = json_request(&mut request).await?;
-    let remote_index = get_state(&request).remote_index.clone();
 
     let mut tenant_conf = TenantConfOpt::default();
     if let Some(gc_period) = request_data.gc_period {
@@ -659,6 +552,9 @@ async fn tenant_create_handler(mut request: Request<Body>) -> Result<Response<Bo
     if let Some(max_lsn_wal_lag) = request_data.max_lsn_wal_lag {
         tenant_conf.max_lsn_wal_lag = Some(max_lsn_wal_lag);
     }
+    if let Some(trace_read_requests) = request_data.trace_read_requests {
+        tenant_conf.trace_read_requests = Some(trace_read_requests);
+    }
 
     tenant_conf.checkpoint_distance = request_data.checkpoint_distance;
     if let Some(checkpoint_timeout) = request_data.checkpoint_timeout {
@@ -685,20 +581,39 @@ async fn tenant_create_handler(mut request: Request<Body>) -> Result<Response<Bo
         .map(TenantId::from)
         .unwrap_or_else(TenantId::generate);
 
-    let new_tenant_id = tokio::task::spawn_blocking(move || {
-        let _enter = info_span!("tenant_create", tenant = ?target_tenant_id).entered();
-        let conf = get_config(&request);
+    let state = get_state(&request);
 
-        tenant_mgr::create_tenant(conf, tenant_conf, target_tenant_id, remote_index)
-            // FIXME: `create_tenant` can fail from both user and internal errors. Replace this
-            // with better error handling once the type permits it
-            .map_err(ApiError::InternalServerError)
-    })
+    let new_tenant = tenant_mgr::create_tenant(
+        state.conf,
+        tenant_conf,
+        target_tenant_id,
+        state.remote_storage.clone(),
+    )
+    .instrument(info_span!("tenant_create", tenant = ?target_tenant_id))
     .await
-    .map_err(|e: JoinError| ApiError::InternalServerError(e.into()))??;
+    // FIXME: `create_tenant` can fail from both user and internal errors. Replace this
+    // with better error handling once the type permits it
+    .map_err(ApiError::InternalServerError)?;
 
-    Ok(match new_tenant_id {
-        Some(id) => json_response(StatusCode::CREATED, TenantCreateResponse(id))?,
+    Ok(match new_tenant {
+        Some(tenant) => {
+            // We created the tenant. Existing API semantics are that the tenant
+            // is Active when this function returns.
+            if let res @ Err(_) = tenant.wait_to_become_active().await {
+                // This shouldn't happen because we just created the tenant directory
+                // in tenant_mgr::create_tenant, and there aren't any remote timelines
+                // to load, so, nothing can really fail during load.
+                // Don't do cleanup because we don't know how we got here.
+                // The tenant will likely be in `Broken` state and subsequent
+                // calls will fail.
+                res.context("created tenant failed to become active")
+                    .map_err(ApiError::InternalServerError)?;
+            }
+            json_response(
+                StatusCode::CREATED,
+                TenantCreateResponse(tenant.tenant_id()),
+            )?
+        }
         None => json_response(StatusCode::CONFLICT, ())?,
     })
 }
@@ -746,6 +661,9 @@ async fn tenant_config_handler(mut request: Request<Body>) -> Result<Response<Bo
     if let Some(max_lsn_wal_lag) = request_data.max_lsn_wal_lag {
         tenant_conf.max_lsn_wal_lag = Some(max_lsn_wal_lag);
     }
+    if let Some(trace_read_requests) = request_data.trace_read_requests {
+        tenant_conf.trace_read_requests = Some(trace_read_requests);
+    }
 
     tenant_conf.checkpoint_distance = request_data.checkpoint_distance;
     if let Some(checkpoint_timeout) = request_data.checkpoint_timeout {
@@ -766,17 +684,13 @@ async fn tenant_config_handler(mut request: Request<Body>) -> Result<Response<Bo
         );
     }
 
-    tokio::task::spawn_blocking(move || {
-        let _enter = info_span!("tenant_config", tenant = ?tenant_id).entered();
-
-        let state = get_state(&request);
-        tenant_mgr::update_tenant_config(state.conf, tenant_conf, tenant_id)
-            // FIXME: `update_tenant_config` can fail because of both user and internal errors.
-            // Replace this `map_err` with better error handling once the type permits it
-            .map_err(ApiError::InternalServerError)
-    })
-    .await
-    .map_err(|e: JoinError| ApiError::InternalServerError(e.into()))??;
+    let state = get_state(&request);
+    tenant_mgr::update_tenant_config(state.conf, tenant_conf, tenant_id)
+        .instrument(info_span!("tenant_config", tenant = ?tenant_id))
+        .await
+        // FIXME: `update_tenant_config` can fail because of both user and internal errors.
+        // Replace this `map_err` with better error handling once the type permits it
+        .map_err(ApiError::InternalServerError)?;
 
     json_response(StatusCode::OK, ())
 }
@@ -821,22 +735,16 @@ async fn timeline_gc_handler(mut request: Request<Body>) -> Result<Response<Body
     let timeline_id: TimelineId = parse_request_param(&request, "timeline_id")?;
     check_permission(&request, Some(tenant_id))?;
 
-    // FIXME: currently this will return a 500 error on bad tenant id; it should be 4XX
-    let tenant = tenant_mgr::get_tenant(tenant_id, false).map_err(ApiError::NotFound)?;
     let gc_req: TimelineGcRequest = json_request(&mut request).await?;
 
-    let gc_horizon = gc_req.gc_horizon.unwrap_or_else(|| tenant.get_gc_horizon());
-
-    // Use tenant's pitr setting
-    let pitr = tenant.get_pitr_interval();
-    let result = tenant
-        .gc_iteration(Some(timeline_id), gc_horizon, pitr, true)
-        .instrument(info_span!("manual_gc", tenant = %tenant_id, timeline = %timeline_id))
+    let wait_task_done = tenant_mgr::immediate_gc(tenant_id, timeline_id, gc_req).await?;
+    let gc_result = wait_task_done
         .await
-        // FIXME: `gc_iteration` can return an error for multiple reasons; we should handle it
-        // better once the types support it.
+        .context("wait for gc task")
+        .map_err(ApiError::InternalServerError)?
         .map_err(ApiError::InternalServerError)?;
-    json_response(StatusCode::OK, result)
+
+    json_response(StatusCode::OK, gc_result)
 }
 
 // Run compaction immediately on given timeline.
@@ -846,11 +754,16 @@ async fn timeline_compact_handler(request: Request<Body>) -> Result<Response<Bod
     let timeline_id: TimelineId = parse_request_param(&request, "timeline_id")?;
     check_permission(&request, Some(tenant_id))?;
 
-    let tenant = tenant_mgr::get_tenant(tenant_id, true).map_err(ApiError::NotFound)?;
+    let tenant = tenant_mgr::get_tenant(tenant_id, true)
+        .await
+        .map_err(ApiError::NotFound)?;
     let timeline = tenant
         .get_timeline(timeline_id, true)
         .map_err(ApiError::NotFound)?;
-    timeline.compact().map_err(ApiError::InternalServerError)?;
+    timeline
+        .compact()
+        .await
+        .map_err(ApiError::InternalServerError)?;
 
     json_response(StatusCode::OK, ())
 }
@@ -862,7 +775,9 @@ async fn timeline_checkpoint_handler(request: Request<Body>) -> Result<Response<
     let timeline_id: TimelineId = parse_request_param(&request, "timeline_id")?;
     check_permission(&request, Some(tenant_id))?;
 
-    let tenant = tenant_mgr::get_tenant(tenant_id, true).map_err(ApiError::NotFound)?;
+    let tenant = tenant_mgr::get_tenant(tenant_id, true)
+        .await
+        .map_err(ApiError::NotFound)?;
     let timeline = tenant
         .get_timeline(timeline_id, true)
         .map_err(ApiError::NotFound)?;
@@ -884,7 +799,6 @@ async fn handler_404(_: Request<Body>) -> Result<Response<Body>, ApiError> {
 pub fn make_router(
     conf: &'static PageServerConf,
     auth: Option<Arc<JwtAuth>>,
-    remote_index: RemoteIndex,
     remote_storage: Option<GenericRemoteStorage>,
 ) -> anyhow::Result<RouterBuilder<hyper::Body, ApiError>> {
     let spec = include_bytes!("openapi_spec.yml");
@@ -921,8 +835,7 @@ pub fn make_router(
 
     Ok(router
         .data(Arc::new(
-            State::new(conf, auth, remote_index, remote_storage)
-                .context("Failed to initialize router state")?,
+            State::new(conf, auth, remote_storage).context("Failed to initialize router state")?,
         ))
         .get("/v1/status", status_handler)
         .put(
@@ -938,6 +851,8 @@ pub fn make_router(
         .post("/v1/tenant/:tenant_id/timeline", timeline_create_handler)
         .post("/v1/tenant/:tenant_id/attach", tenant_attach_handler)
         .post("/v1/tenant/:tenant_id/detach", tenant_detach_handler)
+        .post("/v1/tenant/:tenant_id/load", tenant_load_handler)
+        .post("/v1/tenant/:tenant_id/ignore", tenant_ignore_handler)
         .get(
             "/v1/tenant/:tenant_id/timeline/:timeline_id",
             timeline_detail_handler,